Recognition Of Anions 13ThePoet05

http://www.nd-warez.info/

http://www.bookstores.com/books/-/-/9783540790914.html?kbid=1010

129Structure and Bonding

Series Editor: D. M. P. Mingos

Editorial Board:

P. Day · X. Duan · L. H. Gade · T. J. MeyerG. Parkin · J.-P. Sauvage

Structure and BondingSeries Editor: D. M. P. Mingos

Recently Published and Forthcoming Volumes

Contemporary Metal Boron Chemistry IVolume Editors: Marder, T. B., Lin, Z.Vol. 130, 2008

Recognition of AnionsVolume Editor: Vilar, R.Vol. 129, 2008

Liquid Crystalline Functional Assembliesand Their Supramolecular StructuresVolume Editor: Kato, T.Vol. 128, 2008

Organometallic and Coordination Chemistryof the ActinidesVolume Editor: Albrecht-Schmitt, T. E.Vol. 127, 2008

Halogen BondingFundamentals and ApplicationsVolume Editors: Metrangolo, P., Resnati, G.Vol. 126, 2008

High Energy Density MaterialsVolume Editor: Klapötke, T. H.Vol. 125, 2007

Ferro- and AntiferroelectricityVolume Editors: Dalal, N. S.,Bussmann-Holder, A.Vol. 124, 2007

Photofunctional Transition Metal ComplexesVolume Editor: V. W. W. YamVol. 123, 2007

Single-Molecule Magnetsand Related PhenomenaVolume Editor: Winpenny, R.Vol. 122, 2006

Non-Covalent Multi-Porphyrin AssembliesSynthesis and PropertiesVolume Editor: Alessio, E.Vol. 121, 2006

Recent Developments in Mercury ScienceVolume Editor: Atwood, David A.Vol. 120, 2006

Layered Double HydroxidesVolume Editors: Duan, X., Evans, D. G.Vol. 119, 2005

Semiconductor Nanocrystalsand Silicate NanoparticlesVolume Editors: Peng, X., Mingos, D. M. P.Vol. 118, 2005

Magnetic Functions Beyondthe Spin-HamiltonianVolume Editor: Mingos, D. M. P.Vol. 117, 2005

Intermolecular Forces and Clusters IIVolume Editor: Wales, D. J.Vol. 116, 2005

Intermolecular Forces and Clusters IVolume Editor: Wales, D. J.Vol. 115, 2005

Superconductivity in Complex SystemsVolume Editors: Müller, K. A.,Bussmann-Holder, A.Vol. 114, 2005

Principles and Applicationsof Density Functional Theoryin Inorganic Chemistry IIVolume Editors:Kaltsoyannis, N., McGrady, J. E.Vol. 113, 2004

Recognition of Anions

Volume Editor: Ramón Vilar

With contributions by

P. Ballester · G. W. Bates · S. R. Bayly · P. D. Beer · S. L. EwenP. A. Gale · I. Hamachi · J. H. G. Steinke · S. Tamaru · R. Vilar

123

The series Structure and Bonding publishes critical reviews on topics of research concerned withchemical structure and bonding. The scope of the series spans the entire Periodic Table. It focusesattention on new and developing areas of modern structural and theoretical chemistry such as na-nostructures, molecular electronics, designed molecular solids, surfaces, metal clusters and supra-molecular structures. Physical and spectroscopic techniques used to determine, examine and modelstructures fall within the purview of Structure and Bonding to the extent that the focus is on thescientific results obtained and not on specialist information concerning the techniques themselves.Issues associated with the development of bonding models and generalizations that illuminate thereactivity pathways and rates of chemical processes are also relevant.As a rule, contributions are specially commissioned. The editors and publishers will, however, alwaysbe pleased to receive suggestions and supplementary information. Papers are accepted for Structureand Bonding in English.In references Structure and Bonding is abbeviated Struct Bond and is cited as a journal.

Springer WWW home page: springer.comVisit the Struct Bond content at springerlink.com

ISBN 978-3-540-79091-4 e-ISBN 978-3-540-79092-1DOI 10.1007/978-3-540-79092-1

Structure and Bonding ISSN 0081-5993

Library of Congress Control Number: 2008923960

c© 2008 Springer-Verlag Berlin Heidelberg

This work is subject to copyright. All rights are reserved, whether the whole or part of the materialis concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broad-casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication ofthis publication or parts thereof is permitted only under the provisions of the German Copyright Lawof September 9, 1965, in its current version, and permission for use must always be obtained fromSpringer. Violations are liable to prosecution under the German Copyright Law.

The use of general descriptive names, registered names, trademarks, etc. in this publication does notimply, even in the absence of a specific statement, that such names are exempt from the relevantprotective laws and regulations and therefore free for general use.

Cover design: WMXDesign GmbH, HeidelbergTypesetting and Production: le-tex publishing services oHG, Leipzig

Printed on acid-free paper

9 8 7 6 5 4 3 2 1 0

springer.com

Series EditorProf. D. Michael P. Mingos

PrincipalSt. Edmund HallOxford OX1 4AR, [email protected]

Volume Editor

Dr. Ramón Vilar

Imperial College LondonDepartment of ChemistrySouth KensingtonLondon, SW7 2AZ, [email protected]

Editorial Board

Prof. Peter Day

Director and Fullerian Professorof ChemistryThe Royal Institution of Great Britain21 Albermarle StreetLondon W1X 4BS, [email protected]

Prof. Xue Duan

DirectorState Key Laboratoryof Chemical Resource EngineeringBeijing University of Chemical Technology15 Bei San Huan Dong LuBeijing 100029, P.R. [email protected]

Prof. Lutz H. Gade

Anorganisch-Chemisches InstitutUniversität HeidelbergIm Neuenheimer Feld 27069120 Heidelberg, [email protected]

Prof. Thomas J. Meyer

Department of ChemistryCampus Box 3290Venable and Kenan LaboratoriesThe University of North Carolinaand Chapel HillChapel Hill, NC 27599-3290, [email protected]

Prof. Gerard Parkin

Department of Chemistry (Box 3115)Columbia University3000 BroadwayNew York, New York 10027, [email protected]

Prof. Jean-Pierre Sauvage

Faculté de ChimieLaboratoires de ChimieOrgano-MinéraleUniversité Louis Pasteur4, rue Blaise Pascal67070 Strasbourg Cedex, [email protected]

Structure and BondingAlso Available Electronically

For all customers who have a standing order to Structure and Bonding, weoffer the electronic version via SpringerLink free of charge. Please contactyour librarian who can receive a password or free access to the full articles byregistering at:

springerlink.com

If you do not have a subscription, you can still view the tables of contents of thevolumes and the abstract of each article by going to the SpringerLink Home-page, clicking on “Browse by Online Libraries”, then “Chemical Sciences”, andfinally choose Structure and Bonding.

You will find information about the

– Editorial Board– Aims and Scope– Instructions for Authors– Sample Contribution

at springer.com using the search function.

Color figures are published in full color within the electronic version onSpringerLink.

Preface

A large number of biologically relevant species are negatively charged, there-fore it is not surprising that nature has developed sophisticated receptors torecognise, detect and transform anions. For example, complex receptors suchas phosphate- and sulphate-binding proteins are employed by living cells toselectively recognise these two geometrically analogous anions. In additionto their roles in biological systems, some anions also have important envi-ronmental impacts. For example, cyanide, pertechnetate and chromates poseserious health problems if present in water supplies.

Because of their important biological roles and potential environmentalimpact there is great current interest in developing molecular receptors toselectively recognise anions and in doing so be able to sequester, transform orsense them. The six chapters presented in this volume provide an overview ofanion recognition and the most recent advances in this fast-growing area ofsupramolecular chemistry are highlighted.

The first chapter by Bates and Gale provides an overview of the coordinationof anions by synthetic organic hosts. The different organic functional groupsused to bind anions are presented and this provides an introduction to thestructural and electronic properties that hosts must have to recognise anionicguests. On the other hand, Bayly and Beer give a detailed account of the use ofmetal complexes as anion receptors. Besides the important structural featuresthat metals can confer to receptors, their optical and redox properties makethem attractive for the development of anion sensors.

Metal-based receptors have found particularly interesting applications inthe recognition of phosphorylated species of biological interest (e.g. phospho-rylated amino acids and peptides). This area is reviewed in depth by Tamaruand Hamachi with particular emphasis on a series of receptors based onzinc(II) centres which have been shown to bind phosphates with very highbinding constants in aqueous media. The applications of this type of receptorfor the detection of samples of biological interest are also presented.

Ballester provides an interesting account of anion · · · π interactions andtheir impact in host design. Over the past few years there has been mountingevidence that this type of interaction plays an important role in anion recog-nition. The chapter starts with a detailed overview of the theoretical aspectsof anion · · ·π interactions which is followed by a discussion of the existing

X Preface

experimental evidence for this type of interaction. Both, solution and crystal-lographic studies are analysed showing the potential impact that this type ofinteraction could have in the design of new anion receptors.

The use of anions as templating agents is discussed by Vilar. The chapterstarts with a general overview of the area and a discussion of the applicationsof anion templates in organic and coordination chemistry. The second partof the chapter deals with examples where anions are employed as templatesin dynamic combinatorial libraries. This approach promises to provide anefficient route for the synthesis of better and more selective anion receptors.The last chapter by Ewen and Steinke also deals with the use of anions astemplates but in this case in the context of molecular imprinted polymers. Thefirst half of the chapter provides an introduction into molecularly imprintedpolymers and this is followed by a detailed discussion of examples whereanionic species have been used to imprint this class of polymeric materials.

The topics discussed in this volume provide an exciting and stimulatingoverview of the most recent studies within anion recognition and templation.Although the supramolecular chemistry of anions took a long time to develop,it is now a mature area that provides solutions to challenging problems. There isno doubt that its growth will continue yielding more sophisticated and efficientreceptors for the recognition of a wide range of negatively charged species.

London, February 2008 Ramón Vilar

Contents

An Introduction to Anion Receptors Based on Organic FrameworksG. W. Bates · P. A. Gale . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Metal-Based Anion Receptor SystemsS. R. Bayly · P. D. Beer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Recent Progress of Phosphate Derivatives RecognitionUtilizing Artificial Small Molecular Receptors in Aqueous MediaS. Tamaru · I. Hamachi . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Anions and π-Aromatic Systems. Do They Interact Attractively?P. Ballester . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Anion Templates in Synthesis and Dynamic Combinatorial LibrariesR. Vilar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Molecularly Imprinted Polymers Using Anions as TemplatesS. L. Ewen · J. H. G. Steinke . . . . . . . . . . . . . . . . . . . . . . . . . 207

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

Struct Bond (2008) 129: 1–44DOI 10.1007/430_2007_069© Springer-Verlag Berlin HeidelbergPublished online: 10 November 2007

An Introduction to Anion ReceptorsBased on Organic Frameworks

Gareth W. Bates · Philip A. Gale (�)

School of Chemistry, University of Southampton, Southampton SO17 1BJ, [email protected]

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Neutral Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1 Acyclic Amide and Sulfonamide-Based Receptors . . . . . . . . . . . . . . 22.2 Macrocyclic Amide Receptors . . . . . . . . . . . . . . . . . . . . . . . . . 62.3 Urea and Thiourea-Based Receptors . . . . . . . . . . . . . . . . . . . . . . 12

3 Aromatic NH Donor Containing Neutral Receptors . . . . . . . . . . . . . 213.1 Pyrrole-Based Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213.2 Carbazole and Indole-Based Receptors . . . . . . . . . . . . . . . . . . . . 29

4 Hydroxy (OH) Donors in Neutral Receptors . . . . . . . . . . . . . . . . . 32

5 Charged Receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335.1 Imidazolium and Pyridinium-Based Receptors . . . . . . . . . . . . . . . . 345.2 Guanidinium-Based Receptors . . . . . . . . . . . . . . . . . . . . . . . . . 375.3 Ammonium-Containing Receptors . . . . . . . . . . . . . . . . . . . . . . . 39

6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Abstract This review article provides a broad overview to the area of anion coordinationby synthetic organic receptors and includes examples of different functional groups usedto bind anions. The first section examines neutral anion receptors containing amide-,sulfonamide-, urea- and thiourea-based receptors. Then aromatics such as pyrrole, car-bazole and indole are discussed before concluding the discussion of neutral systems withexamples of hydroxy OH donors. A brief overview of charged systems is also provided.

Keywords Anion recognition · Complexation · Crystal structures · Hydrogen bonding ·Supramolecular chemistry

1Introduction

The development of new anion receptors based on organic frameworks con-tinues to attract considerable research effort [1–4]. A wide variety of systemshave been published in the last 15 years with both macrocyclic and acyclic

2 G.W. Bates · P.A. Gale

systems functioning as effective and selective anion receptors. This reviewis not comprehensive but provides examples of important classes of anionreceptor systems based on organic frameworks.

2Neutral Receptors

2.1Acyclic Amide and Sulfonamide-Based Receptors

Secondary amides are versatile and highly accessible hydrogen bond donorsthat have been used in numerous synthetic receptors. In the biological arena,there are many examples of proteins that employ amide NH· · ·anion interac-tions to bind negatively charged guests [5–9]. The first example of a syntheticamide containing receptor, published in 1986 by Pascal and co-workers, wasa crytpand-like tris-amide that was shown to interact with fluoride in DMSO-d6 [10].

In 1993, Reinhoudt and co-workers described the synthesis and bind-ing properties of a series of tris-amides and tris-sulfonamides based uponthe tren skeleton [11]. These receptors proved to be selective for phosphatein acetonitrile solution and demonstrated, arguably for the first time, thatanion receptor systems need not be difficult to make but rather that sim-ple organic compounds could function as very effective receptors. Stabilityconstants were calculated by conductivity experiments and showed that re-ceptor 1f bound dihydrogenphosphate with the highest affinity (14 200 M–1)in acetonitrile presumably due to the preorganization of the receptor via π–π

interactions between the naphthyl groups.

Four years later, in 1997, Crabtree and co-workers reported that simpleisophthalamide receptors e.g. 2 can bind anions in organic solution [12].These receptors, even simpler than Reinhoudt’s tren-based anion binders,were found to bind smaller halides selectively in dichloromethane solution.

The X-ray crystal structure of the bromide complex of 2a shows the receptoradopting the syn–syn conformation with the bromide anion coordinated to theamide NH’s with N· · ·Br distances of 3.44 and 3.64 A (Fig. 1). The crystal struc-

An Introduction to Anion Receptors Based on Organic Frameworks 3

ture also reveals that the bromide anion is positioned above the plane of thecentral aryl ring. Solution studies revealed that receptors 2a and 2b have highaffinity for halide anions and form complexes with exclusively 1 : 1 host/gueststoichiometry. Stability constants for 2b were determined by 1H NMR titrationstudies and found to be 6.1×104 M–1 for chloride, 7.1×103 M–1 for bromideand 4.6×102 M–1 for iodide in dichloromethane-d2.

Fig. 1 X-ray crystal structure of a bromide complex of 2a

Almost contemporaneously, B.D. Smith and co-workers reported the use offunctionalized isophthalamide receptors for the coordination of anions [13].Smith appended boronate groups to the peripheral aryl groups in order toform interactions between the Lewis acidic boron and the carbonyl oxygensof the amides therefore “pre-organizing” the receptor into the syn–syn con-formation (preferable for anion coordination) and presumably increasing theacidity of the NH group. Proton COSY and NOE difference experiments indi-cated that the receptor did indeed adopt the desired syn–syn conformation inDMSO-d6. NMR titration experiments in DMSO-d6 at 295 K showed that re-


ceptor 3 bound acetate with a stability constant of 2.1×103 M–1 as comparedto 1.1×102 M–1 found with the non-preorganized receptor 2a and acetate.

Recently J.T. Davis, Gale, Quesada and co-workers have shown that ap-pended hydroxy groups on the central aryl ring of an isophthalamide canpre-organize the receptor into the syn–syn conformation and again presum-ably increase the acidity of the amide NH groups, which increases the re-ceptor’s ability to bind chloride (Fig. 2) [14]. The preorganization occurs dueto intramolecular hydrogen bonds between the hydroxy groups and carbonyloxygen of the amides.

Proton NMR titration experiments in CD3CN at 298 K revealed that re-ceptor 4 bound chloride most strongly with a stability constant of 5230 M–1,whereas a stability constant of 195 M–1 was obtained for the unfunctionalizedcleft 5. Model compound 6 contains methoxy groups in the 4- and 6-positionsand functions as a control. In this system the intramolecular hydrogen bondsform between the amide NH groups and the methoxy oxygens and conse-quently the compound does not interact with chloride (Fig. 2). Most interest-ingly, it was shown that compound 4 functions as a highly efficient chloridetransport agent across EYPC lipid bilayers whilst the analogous isophtha-lamide 5, and model compound 6 show no transport ability. Compound 4seems to be the simplest synthetic lipid bilayer transport agent for chloridestudied so far. The origin of this ability is currently being investigated.

Thordarson et al. have studied the aggregation of pyromellitamide 7 andits response to anions. It was found that compound 7 aggregates in non-polar solvent by the formation of one-dimensional intermolecular hydrogen-bonding networks. Upon the introduction of anions the aggregation of 7 isdisturbed [15]. Proton NMR titration experiments in d6-acetone at 300 K re-

Fig. 2 X-ray crystal structures of 4 (left) and 6 (right)


vealed that compound 7 bound a range of anions with 2 : 1 anion/receptorstoichiometry. Although compound 7 has two discrete binding sites the an-ions were found to bind with negative cooperativity with the strength ofanion binding to 7 decreasing in the order Cl– < CH3CO2

– < Br– < NO3– ≈ I–.

Prohens and co-workers have synthesized compounds 8a and 8b, simplesquaramido-based receptors and investigated their ability to coordinate car-boxylate anion in competitive solvents [16]. The amide NH groups of thesquaramide form a more open cleft (similar to ureas) than the isophthala-mides. Receptors 8a and 8b therefore adopt a more suitable geometry forthe coordination of bidentate anions, such as carboxylate anions, throughtwo approximately linear hydrogen bonds. Proton NMR titration experimentsrevealed association constants of 217 M–1 and 1980 M–1 for the binding ofacetate by 8a and 8b respectively in DMSO-d6 at 295 K.

A.P. Davis and co-workers have designed a number of acyclic receptorsusing the steroid cholic acid as a scaffold upon which they appended sulfon-amide and carbamate amide groups [17]. The inflexibility of the fused ringsystem and the axial conformation of the functional groups in the 7α and 12αpositions results in the formation of a convergent hydrogen-bonding array,ideal for anion binding.

Both the structural rigidity and the shape of the receptor result in the for-mation of a relatively small, well-defined binding site that shows selectivityfor halide anions with particularly high affinities observed for fluoride withreceptor 9a (15 400 M–1 association constants were determined by 1H NMRtitration experiments in CDCl3 at 298 K). In the case of compound 9b theassociation constant for fluoride was too high to be determined however, as-


sociation constants for chloride and bromide were found to be much higherthan for compound 9a [17].

2.2Macrocyclic Amide Receptors

Macrocyclic receptors often possess a higher degree of selectivity than acyclicsystems. Hamilton and Choi have described the synthesis and anion bind-ing properties of a family of cyclic triamides 10a and 10b [18]. These C3symmetric receptors were found to be selective for oxo-anions such as tosy-late with association constants of 2.6×105 M–1 and 2.1×105 M–1 obtainedfor compounds 10a and 10b respectively in CDCl3/2% dimethylsulfoxide at296 K. Hamilton also studied the binding ability of an acyclic analogue (11)and found significantly lower stability constants for the anion complexesformed. For example, in the case of nitrate, a stability constant of 620 M–1

was calculated for compound 11 whereas a stability constant of 4.6×105 M–1

(K2 = 2.1×103 M–1) was found with compound 10a.


Interestingly, the NMR data led the authors to suggest that receptor 10bforms a 2 : 1 host/anion “sandwich” complex at low concentrations of iodide(as evidenced by an initial up-field shift of the NH resonances until ca. 0.5equivalents of iodide) then switching to a 1 : 1 complex at higher concentra-tions of iodide (down-field shift of NH resonances after ca. 0.5 equivalents ofiodide) (Fig. 3). Titrations with 10a and 10b in CDCl3/2% DMSO-d6 mixturedisplayed complex binding behavior, thus titration experiments were con-ducted in 100% DMSO-d6 at 296 K to simplify the equilibria occurring insolution. All data for receptor 10b were fitted to a 1 : 1 binding model andagain the macrocycle was found to be oxo-anion selective with the higheststability constants being found with dihydrogen phosphate and hydrogen sul-fate (1.5×104 M–1 and 1.7×103 M–1, respectively).

Fig. 3 Changes in the amide 1H NMR proton resonance of 10b with increasing [I–]concentrations. Reprinted in part with permission from [18]

Chmielewski and Jurczak have reported a series of extended tetra-amidemacrocycles containing two pyridine-2,6-dicarboxamide “caps” linked viashort aliphatic chains. The macrocycles possess a well-defined cavity with allthe amide groups directed inwards [19, 20].

Proton NMR titrations in DMSO-d6 at 298 K were conducted in order todetermine the stability constants of receptors 12a, 12b and 12c with a rangeof anions, added as their tetrabutylammonium salts. The strongest associa-tion constants were obtained for the 20-membered macrocycle 12b with themost significant increase in affinity between the macrocycles being observedin the binding of chloride. Enlargement from the 18-membered macrocycle


12a to the 20-membered macrocycle 12b results in a 30-fold increase in theassociation constant for chloride (65 M–1 for 12a against 1930 M–1) whereasfurther enlargement to the 24-membered macrocycle 12c results in the reduc-tion of the association constant by two orders of magnitude (1930 M–1 against18 M–1). This suggests that the 20-membered macrocycle 12b has good sizecomplementarity with the chloride anion. Although the 24-membered macro-cycle 12c was designed with a large enough cavity to accommodate two oxy-gen atoms from oxo-anionic guests, the stability constants obtained were thelowest of the receptors tested. This is evidence that the additional flexibilityintroduced into the macrocycle via the longer aliphatic chain has a detrimen-tal effect on the anion binding ability of the receptor.

More recently the same authors have studied the anion binding ability ofsimilar macrocyclic systems based on isophthalamides [21]. The isophthala-mide moieties were introduced as previous studies have shown isophthala-mide derivatives bind anions more strongly than the analogous pyridine-2,6-dicarboxamides [22].

Stability constants were obtained for receptors 13a, 13b and 13c using anal-ogous conditions to those employed for receptors 12a, 12b and 12c. As withthe pyridine-2,6-dicarboxamide macrocycles, the stability constants for theisophthalamide macrocycles appear to be influenced by the size and flexi-bility of the system with the higher constants observed in the 20-memberedreceptor 13a with notable decreases in the association constants with the 22-and 24-membered receptors 13b and 13c. The greatest decreases where ob-served in the stability constants obtained with the carboxylate anions. In thecase of acetate the constants decreased from 3130 M–1 for 13a to 552 M–1

and 205 M–1 for 13b and 13c, respectively. In the case of benzoate a constantof 601 M–1 was calculated for 13a decreasing to 302 M–1 and 82 M–1 for 13band 13c, respectively. Unexpectedly lower binding constants were obtainedfor the isophthalamide macrocycles (13a–13c) compared to the pyridine-


2,6-dicarboxamide macrocycles (12a–12c), a result rationalized in terms ofthe competition between the formation of intramolecular hydrogen bonds(arising from the preferred syn–anti conformation of the isophthalamide insolution) and complexation of the anion (Scheme 1).

Scheme 1 Intramolecular hydrogen-bonding vs. anion binding in compound 13a

Bowman-James and co-workers have designed polyamide cryptand-typesystems based on triamines, such as tren (e.g. 14) and trpn (e.g. 15), andshown that they bind anions [23]. The crystal structure of the hydrochloricacid and fluoride complexes of 14 reveal that the anions are encapsulatedwithin the cavity of the amidocryptand and bound to the six-amide NHgroups. In contrast the hydrochloric acid structure of the expanded trpn-based amidocryptand 15 shows the encapsulation of two chloride anionswithin the cryptand, bridged by a water molecule. Each chloride is bound tothe water molecule as well as a protonated bridgehead amine and two hydro-gen bonds from the amides groups.

Stability constants for 14 and 15 with different anions (added as their tetra-butylammonium salts) were obtained by 1H NMR titrations in DMSO-d6. In


both cases, a slow equilibrium was observed in the titrations with fluoridewith stability constants >105 M–1. The expansion of the cavity from recep-tor 14 to 15 results in a significant change in the binding and selectivity foranions. In the smaller receptor, 14, chloride is bound much more strongly(3000 M–1) as compared to 15 (180 M–1) whereas the receptor 15 has a muchhigher affinity for hydrogen sulfate with a stability constant of 2700 M–1 ascompared to 68 M–1 for 14. These findings may be due to the size comple-mentarity between the receptors and guests with 14 being an ideal size toencapsulate chloride and 15 being ideal for hydrogen sulfate, as illustrated bythe crystal structures (Fig. 4).

Fig. 4 X-ray crystal structures of the chloride complex of 14 and the sulfate complex of 15

The authors have also synthesized 16, a tricyclic cryptand-like receptor,and have studied its ability to bind anions. Proton NMR titration experimentsin DMSO-d6 at 23 ◦C revealed that compound 16 was selective for bifluoride(FHF–) with an association constant of 5500 M–1 being calculated. Dihydro-genphosphate, azide and acetate were also found to bind to 16 with stabilityconstants of 740 M–1, 340 M–1 and 100 M–1, respectively [24].


Kubik and co-workers have developed a series of highly effective anionreceptors based upon cyclic peptides. Cyclic hexapeptide receptors such as17 consist of alternately linked l-proline and 6-aminopicolinic acid sub-units [25]. A 1 : 1 binding stoichiometry for 17 and the sodium salt of ben-zenesulfonate was confirmed by a Job plot but in the case of the halide andsulfate sodium salts 2 : 1 host/guest complexes were found. This was con-firmed by electrospray mass spectrometry and in the case of iodide a crystalstructure of the 2 : 1 complex was obtained where the iodide was “sand-wiched” between two cyclic hexapeptide receptors.

The formation of the 2 : 1 complexes observed in 17 led the authors todesign and synthesize compound 18 where two cyclic hexapeptides are co-valently linked. The new receptor binds anions in a 1 : 1 stoichiometry inmethanol/water mixtures efficiently, with high affinity and selectivity for sul-fate being observed. Both 1H NMR titrations and ITC experiments wereconducted in 50% methanol/water at 298 K and stability constants (log Ka) of


5.54 and 4.55±0.23 were found by 1H NMR and ITC respectively for 18 withsulfate (added as Na2SO4) [26].

2.3Urea and Thiourea-Based Receptors

Ureas and thioureas possess two parallel NH hydrogen bond donor groupsand have been shown in a wide variety of receptors to function as highly ef-ficient binding sites for “Y-shaped” anions such as carboxylates. Thioureasare more acidic than analogous ureas and on this basis might be expected toform stronger complexes with anions. However, other effects can often maskor reverse this expected trend.

There have been a number of reports of anions triggering the deproto-nation of neutral NH groups in anion receptor systems. This is often due,in the case of fluoride, to the formation of the stable HF2

– anion drivingthe deprotonation process [27–30]. Fabbrizzi and co-workers have shownthat this process can occur in urea systems containing electron-withdrawinggroups. The interactions between a number of anions and the simple 1,2-bis(4-nitrophenyl) urea 19 were investigated. Oxo-anions were found to bindto the receptor with a 1 : 1 host/guest stoichiometry with the strength of theinteraction depending on the partial negative charge located on each oxygenatom of the anion [31].

Stability constants were determined for compound 19 by UV-Vis spec-trophotometric titrations in acetonitrile at 25 ◦C and revealed that theassociation constants increased with the increasing basicity of the anion(CH3COO– > C6H5COO– > H2PO4

– > NO2– > HSO4

– > NO3–). Addition of

fluoride appears to stabilize a strong 1 : 1 complex at low anion concentration,however at higher anion concentrations deprotonation of the urea subunit oc-curs resulting in the formation of HF2

– (confirmed by 1H NMR). This processwas also characterized by the formation of a new band at 475 nm in the UV-Vis spectrum upon the additions of the fluoride anion and was clearly presentafter the addition of two equivalents of fluoride [32].

Gunnlaugsson and co-workers have studied several receptors containinga thiourea group attached to an anthracene moiety [33]. These compoundswere designed to behave as fluorescent PET (photo-induced electron trans-fer) sensors for the detection of anionic species. Proton NMR titration ex-periments, conducted in DMSO-d6, confirmed that the anions bind to the


receptors through the two NH protons of the thiourea group and form a 1 : 1complex. The authors demonstrated that 20a–d act as ideal PET sensors (onlyfluorescent quantum yield affected upon additions of anions) with quench-ing of the fluorescence being observed with the addition of fluoride, acetateand dihydrogen phosphate anions. Chloride and bromide did not induce anychanges in the fluorescence spectra.

Yoon and co-workers have reported a series of mono- and bis-func-tionalized anthracenes and described their colorimetric and fluorescentproperties for the sensing of both fluoride and pyrophosphate anions. Theauthors appended either phenylurea or p-nitrophenylurea groups through the1-position (for the mono-functionalized derivatives 21a and 21b) and the1- and 8-position (for the bis-functionalized derivative 22a and 22b) of theanthracene [34].

Fluorescent titration experiments with receptors 21b and 22b were carriedout in DMSO with a variety of anions, added as the tetrabutylammoniumsalts, in order to compare the stability constants of the mono- and bis-functionalized receptors. The strongest anion binding was observed with thebis-functionalized receptor (22b) with stability constants of 108 000, 9700


and 6000 M–1 calculated for fluoride, bromide and pyrophosphate, respec-tively. For compound 21b a much lower binding constant of 4000 M–1 wasfound with fluoride as compared to the strong anion complexation observedwith 22b, illustrating that there is a cooperative binding effect in operationwith the two urea groups in receptors 22a and 22b. Temperature-dependant1H NMR experiments in DMF-d7 also revealed that the anion complex stabil-ity was enhanced by the formation of a hydrogen bond between the hydrogenatom in the 9-position and both the fluoride and pyrophosphate guests inreceptors 22a and 22b.

Gale and co-workers have designed 23a, a bis-urea cleft based on o-phenylenediamine, to selectively bind carboxylate anions [35]. The geometryof the receptors provides a convergent cleft appropriate for the binding ofcarboxylate anion through four hydrogen bonds. Stability constants of 3210,1330 and 732 M–1 were calculated for acetate, benzoate and dihydrogen phos-phate, respectively, after analysis of data from 1H NMR titration experimentsconducted in DMSO-d6/0.5% water at 298 K. The crystal structure of the ben-zoate complex of compound 23a is shown in Fig. 5 revealing that the receptorbinds this carboxylate via four hydrogen bonds in the solid state. The sta-bility constants for compound 23a were found to be greater than constants

Fig. 5 X-ray crystal structure of 23a with benzoate. Chem Commun, p 4696 (2005) repro-duced by permission of The Royal Society of Chemistry


obtained with N,N′-diphenylurea with a 2.5-fold increase observed for 23awith acetate compared to the diphenylurea (3210 M–1 for 23a and 1261 M–1

for diphenylurea [36]).The same authors appended electron-withdrawing groups onto both the

central aryl ring and peripheral aryl rings of the receptor in order to increasethe acidity of the urea NH groups. Titration studies were conducted under thesame conditions as for 23a and enhanced stability constants were observedfor both receptors 23b and 23c compared to 23a [37]. In the case of acetate thestability constant increased from 3210 M–1 for 23a, to 4020 M–1 and 8080 M–1

for 23b and 23c, respectively. In the case of dihydrogenphosphate there isa decrease in affinity from 732 M–1 for 23a, to 666 M–1 for 23b but a largeincrease to 4720 M–1 for 23c. The authors proposed that the dihydrogenphos-phate anion interacted most strongly with the central NH groups thus withthe increased acidity of these NH groups in 23c, due to the presence of the twochloro-groups, stronger complexation is observed.

Naphthalene and binaphthalene appended with thiourea groups (24 and25, respectively) have been synthesized by Kondo and co-workers in orderto investigate potential cooperative binding between two thiourea groups in25 [38]. This group found that 1 : 1 complexes were formed between 25 andfluoride, acetate and dihydrogen phosphate anion, which was confirmed byJob plots in acetonitrile and ESI-MS.

UV-Vis spectroscopy titration experiments in acetonitrile solution werecarried out to ascertain the anion binding properties of the receptors withacetate, dihydrogen phosphate, fluoride and chloride. The binding constantsrevealed that the presence of the second thiourea group in 25 significantly im-proves the receptor’s affinity for anions with respect to the mono-thiourea 24.The most significant differences were obtained for titration with fluoride andacetate where binding constants of 1.1×105 M–1 and 2.1×106 M–1 were elu-cidated for 25 and 3.7×103 and 7.7×103 for compound 24 with acetate andfluoride, respectively.

Pfeffer and co-workers have described the use of a highly rigid[3]poly-norbornane as a scaffold on which to append electron-deficient thioureasubunits [39].


The anion binding abilities of 26a and 26b were evaluated by 1H NMRtitration techniques in DMSO-d6 with CH3COO–, F–, H2PO4

– and H2P2O72–

(added as their tetrabutylammonium salts). Additions of fluoride to the re-ceptor resulted in a distinctive color change attributed to deprotonation. Thisprocess was characterized by the loss of the thiourea NH proton resonancesand the appearance of the HF2

– resonance in the 1H NMR during the titra-tion. Analysis of the binding isotherms of receptors 26a and 26b with acetaterevealed that the anions were strongly bound by both 26a and 26b in a 1 : 2receptor-to-anion complex with each of the thiourea units binding to a sin-gle acetate anion. Binding constants (log β1 and log β2 values) of 3.5 (±0.1)and 2.4 (±0.1) were calculated for 26a and 3.5 (±0.1) and 3.0 (±0.1) for 26bwith acetate. Titrations with H2PO4

– were fitted to a 1 : 1 binding model andconstants of 3.9 (±0.1) and 3.6 (±0.1) (log β values) were calculated for re-ceptors 26a and 26b, respectively. Pyrophosphate was then investigated toevaluate the binding ability of 26a and 26b with a dianion. Analysis of thetitration curves for both 26a and 26b with pyrophosphate revealed the forma-tion of a 2 : 1 receptor-to-anion stoichiometry in which each anion terminusis accommodated by two urea groups of a single receptor.

Extending their work on “cholapods”, A.P. Davis and co-workers have ap-pended urea and thiourea groups from the 7 and 12 positions of the steroidscaffold and evaluated the ability of these receptors to bind chloride and bro-mide (added as their tetraethylammonium salts) [40].

NMR data was found to be consistent with the formation of predomin-antly 1 : 1 complexes of the receptors and anions. Stability constants weredetermined by Cram’s extraction method in water-saturated chloroform at30 ◦C [40]. Affinities for both chloride and bromide anions increased throughthe series 27a–d, reflecting the increase in acidity of the NH groups dueto the electron-withdrawing aryl substituents and the change from urea tothiourea in 27d. In the case of chloride the association constant for the“unsubstituted” derivative 27a was calculated to be 1.62×107 M–1, with the


nitrophenyl substituted 27c the constant increased to 4.77×108 M–1 and theconstant increased further to 1.05 × 109 M–1 with the introduction of thethiourea (27d). The addition of the nitrosulfonamide group in 28a–28c alsoenhances the anion-binding affinities with the largest constants being ob-served in the thiourea derivative 28c with constants of 1.03×1011 M–1 and2.59×1010 M–1 calculated for chloride and bromide, respectively.

A further detailed study of these “cholapod” anion receptors was con-ducted where the anion binding ability of several receptors with increas-ing numbers of hydrogen bond donor groups was investigated [41]. It wasfound that a combination of increasing numbers of hydrogen bonding groupsand increasing acidity of the NH groups via electron-withdrawing sub-stituents had a significant effect on the anion stability constants. Receptor29 was found to have the highest affinities for all the anions investigatedexcept acetate where the previously studied 28b and 28c had higher affini-ties (2.6×1011 M–1 and 2.0×1011 M–1, respectively) when compared to 29(1.3×1011 M–1). These steroid-based receptors have also been studied astransport agents for anions across vesicle and cell membranes. Electrochem-


ical, NMR and fluorescence techniques were employed and revealed that the“cholapod” receptors act as mobile carriers and facilitate the transport ofchloride ions across vesicle membranes [42].

Reinhoudt and co-workers have synthesized both acyclic and cyclic recep-tors containing multiple urea-binding sites (e.g. 30 and 31). Anion-bindingstudies were conducted with these systems and a variety of putative anionicguests (added as their tetrabutylammonium salts) using 1H NMR titration ex-periments in DMSO-d6 [43]. In the case of the cleft-like receptors dihydrogenphosphate caused the largest shift in the NH group resonances of all the re-ceptors however an association constant could not be obtained for 30a dueto the complexity of the binding processes in solution. Job plot analysis ofreceptor 30b showed the formation of an exclusive 1 : 2 host/guest complexwith dihydrogen phosphate and an association constant of 5×107 M–2 wascalculated. The thiourea functionalized 30c cleft was also shown to bind di-hydrogen phosphate with a 1 : 2 host/guest stoichiometry and chloride with1 : 1 host/guest stoichiometry.

Macrocyclic receptors 31a and 31b were found to bind both dihydrogenphosphate and chloride in exclusively 1 : 1 host/guest stoichiometries. Bind-ing constants were calculated for 31a and 31b with dihydrogen phosphate andchloride and revealed that dihydrogenphosphate was bound more strongly(2.5×103 M–1 for 31a and 4.0×103 M–1 for 31b) than chloride (500 M–1 for31a and <50 M–1 for 31b).

Gale and co-workers have combined urea and amide groups into a newmacrocyclic motif and studied the anion complexation properties of 32 [44].Stability constants with a variety of anionic guests were elucidated by1H NMR titration techniques in both DMSO-d6/0.5% water and DMSO-d6/5% water at 298 K. The macrocyclic receptor shows significant selectivityfor carboxylate anions over dihydrogen phosphate and chloride. Titrations


in DMSO-d6/0.5% water resulted in high stability constants for acetate(>104 M–1) and benzoate (6430 M–1) therefore the titrations were conductedin 5% water, a much more competitive media, and stability constants of5170 M–1 and 1830 M–1 were calculated for acetate and benzoate, respectively.Interestingly, a crystal structure of a carbonate complex was obtained froma crystallization with tetrabutylammonium fluoride (Fig. 6). It was presumedthat carbonate was gained via the fixation of atmospheric CO2 by the fluoridesalt-macrocycle solution.

Fig. 6 X-ray crystal structure of a carbonate complex of 32 reproduced by permission ofThe Royal Society of Chemistry [43]

In 2000 Lee and Hong synthesized tris-thiourea macrocycles 33a and33b and studied their anion recognition properties by 1H NMR titrationexperiments in DMSO-d6 at 25 ◦C. It was found that macrocycle 33a wasselective for dihydrogen phosphate (800 M–1) over acetate (320 M–1) andchloride (40 M–1). In contrast macrocycle 33b was found to be selective


for acetate (5300 M–1) over dihydrogen phosphate (1600 M–1) and chloride(95 M–1) [45].

Recently, Tobe and co-workers have designed cryptand-like macrocyclesbased on homobenzylic tripodal thiourea and compared their anion-bindingproperties to a series of acyclic tripod-type receptors [46].

The proton resonances in the 1H NMR spectra of cryptand-type recep-tor 34b in various solvents were found to be very broad, possibly due toconformational changes that are slow on the NMR timescale. Therefore, thecomplexation of 34b with anionic species was evaluated by 1H NMR titra-tion experiments in CDCl2CDCl2 at 373 K. Association constants of 116 M–1

and 112 M–1 were calculated for acetate and chloride, respectively, and werefound to be much lower than the tripodal receptor 35a under the same condi-tion (3030 M–1 for acetate and 3700 M–1 for chloride). This low binding abilityof 34b was attributed to strong intramolecular hydrogen bonds between thethiourea groups. Receptors 35a and 35c were then compared and the stabilityconstants (obtained from 1H NMR titrations in DMSO-d6 at 303 K) revealedthat 35a has poor affinity towards all anionic species in DMSO solutionswhereas 35c has high affinity for dihydrogen phosphate and acetate.


3Aromatic NH Donor Containing Neutral Receptors

3.1Pyrrole-Based Receptors

Sessler and co-workers have pioneered the use of the pyrrole NH hydrogenbond donor group in both charged and neutral anion receptor systems [47].In 1992 they reported the anion-binding abilities and fluoride selectivityof sapphyrin 36a, a pentapyrrolic macrocycle [48]. Fluorescence titrationexperiments carried out in methanol revealed that 1 : 1 complexes formedbetween the diprotonated sapphyrin 36a and halide anions and associationconstants of 2.8×105, ∼102 and <102 M–1 were calculated with fluoride, chlo-ride and bromide, respectively. Four-years later Sessler reported the effectivebinding of phosphate by receptor 36b. Phosphorus NMR titration experi-ments were carried out in methanol-d4 at ambient temperature and revealedthat compound 36b bound both phosphoric acid and phenylphosphonic acidwith affinity (1.8×104 and 1.3×104 M–1 for H3PO4 and C6H7PO3, respec-tively) [49].

Gale and co-workers have developed a number of receptors based onthe 2,5-dicarboxamidopyrrole skeleton where the combination of a pyrroleand amide groups form convergent hydrogen-bonding arrays (e.g. 37a and37b) [50].

Differences in the solubility of receptors 37a and 37b meant that their abil-ity to bind anions was assessed in different solvents (DMSO-d6/0.5% waterfor 37a and CD3CN for 37b at 298 K). Both receptors proved to be selec-tive for oxo-anions, however 37a bound dihydrogen phosphate most strongly(1450 M–1) whereas 37b bound benzoate most strongly (2500 M–1) (Fig. 7).

Continuing this work, Gale and co-workers have synthesized more acidicdiamidopyrrole derivatives 37c and 37d by including electron-withdrawing


Fig. 7 X-ray crystal structure of 37b benzoate complex

nitro groups on the peripheral phenyl rings and assessed their anion-bindingability compared to the unfunctionalized 37a [30]. Proton NMR titrationswere carried out in DMSO-d6/0.5% water solutions at 298 K and revealed thatthe presence of the electron-withdrawing groups in receptors 37c and 37dimproved the systems affinity for anionic guests. In the case of benzoate theassociation constants increased significantly from 560 M–1 (obtained for 37aunder identical conditions) to 4150 M–1 for 37c and 4200 M–1 for 37d. Uponthe addition of one equivalent fluoride to 37d the anion appears to coordinateto the receptor. However, upon further additions of fluoride deprotonationoccurs (as indicated by the evolution of a blue color in solution due to the de-protonated pyrrole). In the case of compound 37a fluoride was found to bindto the receptor with a stability constant of 1245 M–1.


Gale, Smith and co-workers have developed prodigiosin mimics based onamidopyrroles in order to co-transport hydrochloride acid across a vesiclemembrane. The inclusion of the protonatable imidazole group allows thereceptor to carry the proton of the acid and in addition, once protonated, pro-vides an extra hydrogen bond donor group to bind the chloride within thecleft [51].

Sessler, Gale and co-workers have further developed receptors based on the2,5-diamidopyrrole skeleton by appending 2-aminopyrrole groups (e.g. com-pound 39) thus increasing the number of hydrogen-bonding groups, whichwas hoped would increase the selectivity for oxo-anions compared to the pre-viously studied 37a and 37b [52].

Proton NMR titration experiments were conducted in DMSO-d6/0.5% wa-ter at 298 K to investigate the solution phase anion complexation propertiesof 39 compared to 37a. The stability constants revealed that 39 showed en-


hanced selectivity for both dihydrogen phosphate and benzoate compared to37a however 39 was found to bind dihydrogen phosphate with higher affin-ity (10 300 M–1) than benzoate (5500 M–1), the reverse selectivity observed for37a.

Recently, Sessler and co-workers have appended the 2-aminopyrroles sub-unit onto a pyridine-2,6-dicarboxamide spacer group to afford two acyclicreceptors (compounds 40a and 40b) [53].

Elucidation of the solution phase anion complexation properties of recep-tors 40a and 40b (determined by UV-Vis spectrophotometric titrations indichloromethane at 298 K) revealed that 40b bound only acetate with a stabil-ity constant of 13 900 M–1 whereas strong binding was observed for benzoate,acetate, NO2

– and CN– with receptor 40a (43 000, 19 000, 13 000 and 5600 M–1,respectively).

Gale and co-workers have also investigated 5,5′-dicarboxamido-dipyrrolyl-methanes as anion receptors [54]. These systems demonstrated a remarkableaffinity and selectivity for dihydrogen phosphate in highly competitive sol-vent media with stability constants of 234 M–1 and 20 M–1 being calculated(by 1H NMR titration experiments at 298 K) for 41a and 41b respectively inDMSO-d6/25% water. Receptors 41a and 41b were found to be unstable insolution therefore analogous compounds containing two methyl groups at-tached to the meso-carbon were synthesized (42a and 42b) in the hope theywould display increased stability in solution [55].


The meso-substituted derivatives showed lower affinities for anion com-pared to 41a and 41b however 42a did display selectivity for dihydrogen phos-phate over other anions with constants of 1092 M–1 for dihydrogen phosphate,124 M–1 for fluoride and 41 M–1 for benzoate calculated in DMSOd6/5% waterat 298 K.

Sessler, Ustynyuk and co-workers have incorporated dipyrromethanesubunits into 2,6-diamidopyridinedipyrromethane hybrid macrocyclic sys-tems [56]. Initially 43 was synthesized and the anion-binding ability waselucidated by UV-Vis spectroscopic titrations in CH3CN at 23 ◦C. Weak bind-ing was observed for chloride, bromide, cyanide and nitrate which wasrationalized by the receptor forming a deep cavity that favors the formationof well-oriented, directional NH-anion hydrogen bonds. Hydrogensulfate wasfound to bind strongly to the receptor in a 1 : 1 host/anion fashion in ace-tonitrile (Ka = 64 000 ± 2600 M–1) presumably due to the orientation of thehydrogen-bonding groups within the macrocycle being ideal for tetrahedralanions.

The same authors went onto “fine tune” the anion-binding properties ofthe pyridine-2,6-dicarboxamide-dipyrromethane-hybrid macrocyclic systemand designed receptor 44 after DFT calculations suggested it to be more suit-able for hydrogensulfate complexation [57]. The receptor’s affinity for a num-ber of anions was determined by UV-Vis spectroscopic titrations in CH3CN at23 ◦C and revealed that bromide, nitrate and chloride are not bound by com-pound 44. Acetate and dihydrogen phosphate bound with a 1 : 1 stoichiometrywith constants of 12 600 ± 450 M–1 and 29 000 ± 1900 M–1, respectively. Anenhanced affinity and selectivity was observed for 44 with hydrogen sulfate(108 000±17 000 M–1) compared to 43.

Recently Katayev, Sessler and co-workers have further developed the hy-brid macrocycle systems by replacing the pyridine-2,6-dicarboxamide moi-eties used in receptors 43 and 44 with bipyrrole and dipyrromethane subunits


(compounds 45 and 46) [58]. UV-Vis spectroscopic titrations in acetoni-trile at 23 ◦C showed that the smaller macrocycle 45 bound hydrogensulfatewith high affinity (2.7×106 M–1) similar to the previously studied receptors43 and 44 whereas the larger bis-dipyrromethane macrocycle 46 exhibiteda different selectivity for anions with chloride being selectively bound withhigh affinity (281 000 M–1). Interestingly, titrations with hydrogensulfate re-sulted in the lowest association constant of all the anions investigated with 46(2100 M–1).

In 1996, Sessler and co-workers reported the anion complexation proper-ties of calix[4]pyrroles e.g. 47. Compound 47 (meso-octamethylcalix[4]pyrr-ole) was originally synthesized by Baeyer in 1886 and is arguably the simplestanion receptor to synthesize as it is formed in one step via the acid-catalyzedcondensation of pyrrole and acetone [59]. The macrocycle forms four hy-drogen bonds to anionic guests and binds fluoride and chloride strongly.Recently, it was discovered that this receptor actually functions as an ion-pair receptor with large charge diffuse cations sitting in the cup-shaped cavityformed by the pyrrole rings when complexed to an anion [60–62]. A wide var-iety of functionalized calixpyrroles have been synthesized and are reviewedextensively elsewhere [63, 64].

One approach to dramatically increase the affinity of calixpyrroles foranions is to introduce a “strap” across the macrocycle containing addi-


tional hydrogen bond donor groups. For example, Lee, Sessler and co-workers have recently reported the synthesis of isophthalamide strappedcalix[4]pyrroles [65]. The authors studied the effect of varying the length ofthe strap on the anion complexation properties of the macrocycle and isother-mal titration calorimetry was employed to investigate the anion-bindingabilities of receptors 48a–c (in CH3CN at 30 ◦C). Receptors 48a–c were foundto have high binding affinities toward halide anions however they failed toshow an appreciable size-dependence selectivity based on the increase ofstrap length. This is illustrated in the case of chloride (the most stronglybound anion) where association constants of 3.89×106 M–1, 3.35×106 M–1

and 3.24×106 M–1 were calculated for 48a, 48b and 48c, respectively.

Recently, Lee and co-workers have shown that a binol-strapped calix[4]pyr-role (49) can be used in the enantioselective recognition of carboxylateanions [66]. Both the R- and S-enantiomers of the strapped calixpyrrolewere isolated and characterized. Detailed studies of the enantioselectivityof the S enantiomer were carried out by isothermal titration calorime-try experiments in acetonitrile with the chiral anions (R)-2-phenylbutyrateand (S)-2-phenylbutyrate. Stability constants were determined and revealed


that receptor (S)-49 shows selectivity for (S)-2-phenylbutyrate over (R)-2-phenylbutyrate with an order of magnitude difference in the constants(Ka(R) = 9.8×103 M–1 and Ka(S) = 1.0×105 M–1).

In 2000 Kohnke and co-workers described the synthesis of meso-octame-thylcalix[6]pyrrole via the conversion of calix[6]furan into dodecaketone,which was then treated with ammonium acetate to obtain calix[6]pyrrole50 [67].

Association constants were determined by Cram extraction methods,which revealed that the macrocycle 50 formed a strong complex with chlo-ride (12 800 ± 1300 M–1). Proton NMR titration experiments conducted inCD2Cl2 also revealed the size dependence selectivity of the calix[n]pyrroleswhere bromide was found to bind approximately seven times stronger tothe larger calix[6]pyrrole compared to calix[4]pyrrole (710 ± 25 M–1 for 50against 10 M–1 for 47) [68].

Recently Cafeo, Kohnke and co-workers have continued work on ex-panded calixpyrroles and have reported the anion-binding properties of twocalix[2]benzo[4]pyrroles 51a and 51b [69].

Elucidation of the stability constants (determined by 1H NMR titrations inCD2Cl2 at 20 ◦C) showed that although chemically similar the two macrocy-cles have significantly different anion-binding properties. Receptor 51b onlybound fluoride and acetate to an appreciable level (2246± 132 M–1 for flu-oride and 597 ± 236 M–1 for acetate) whereas receptor 51a bound a number


of anions with higher affinities. 51a was found to have selectivity for fluo-ride with a constant of approximately 20 000 M–1 estimated from competitivebinding studies in the presence of 51b. This selectivity is presumably due togood size complementarity between 51b and the fluoride anion.

3.2Carbazole and Indole-Based Receptors

Jurczak and co-workers have described 1,8-diamino-3,6-dichlorocarbazole asa building block for anion receptor construction. Stability constants for re-ceptors 52a and 52b with various anions, added as tetrabutylammonium salts,were calculated by 1H NMR titration experiments in DMSO-d6/0.5% wa-ter and it was found that compound 52b bound anions more strongly withconstants of 115 M–1 and 8340 M–1 for chloride and benzoate, respectively,compared to 13 M–1 and 1230 M–1 with compound 52a with chloride andacetate, respectively [70].

Sessler and co-workers have incorporated two carbazole subunits intoexpanded calixpyrrole-type macrocycle 53 [71]. Fluorescence titration ex-periments in dichloromethane at 0.5 µM concentration of host revealed thatcompound 53 shows selectivity for acetate (Ka = 229 000 M–1) over a numberof other carboxylate-type anions (benzoate, oxalate and succinate).

Beer and co-workers have reported the anion-binding ability of a num-ber of indolocarbazoles sensors (54a–c) [72]. The highest association con-stants were obtained with receptor 54c presumably due to the presence ofthe electron-withdrawing bromide groups increasing the acidity of the NHgroups. UV-Vis spectroscopic titrations in acetone at 25 ◦C revealed that


compound 54c bound benzoate most strongly (log Ka = 5.9) followed by phos-phate (log Ka = 5.3) then fluoride (log Ka = 5.0) and chloride (log Ka = 4.9),a trend observed in both 54a and 54b.

Recently, indole-based receptors have attracted increasing attention.Sessler and co-workers have described the use of indole subunits for an-ion recognition within diindolylquinoxalines receptors (55a and 55b) [73].Stability constants were determined by UV-Vis spectroscopic titrations indichloromethane at 22 ◦C and showed that both receptors have apprecia-ble selectivity for dihydrogenphosphate with constants of 6800 M–1 and20 000 M–1 calculated for compounds 55a and 55b, respectively. Receptor 55bwas found to be highly colored and upon the addition of dihydrogenphos-phate a visible change in color was observed.

A number of biindoyl-based systems have been developed by Jeong andco-workers for the binding of anionic species. This group have prepared mo-lecular clefts based on the 2,2′-biindoyl scaffold with amide groups attachedvia alkyne linkers and compared the anion-binding ability of 56 to the morerigid ethyno-bridged receptor 57 [74].

Stability constants were determined by UV-Vis spectroscopy titration ex-periments in CH3CN at 22 ◦C and revealed that receptor 57 did indeed bindthe anionic species with higher affinities than 56, the less rigid receptor.Chloride was bound 22-times more strongly by receptor 57 compared to56 (1.1×105 M–1 for 57 against 5.1×103 M–1 for 56) whereas bromide wasbound 41-times more strongly by 57 compared to 56 (8.7×103 M–1 for 57against 2.1×102 M–1 for 56) illustrating the importance of preorganization inthe binding of small anionic guests.


Jeong and co-workers have extended the research into biindolyl scaffoldsas a building block for anion receptors by synthesizing new macrocyclic sys-tems 58 and 59 [75]. Both compounds 58 and 59 were found to stronglybind various anions (stability constants determined by UV-Vis spectroscopytitrations in CH3CN at 295 K). In the case of the halide anions a size comple-mentarity was observed for both macrocycles where the smaller fluoride andchloride anions were found to bind more strongly to 58 and 59 than the largerbromide and iodide anions.

Recently, Gale and co-workers have synthesized simple clefts with indolesubunits appended to isophthalamide and pyridine-2,6-dicarboxamide spac-ers and described their anion-binding properties [76].

Proton NMR titrations in DMSO-d6/0.5 water at 298 K were conducted toelucidate stability constants for a number of anions, added as their terabuty-lammonium salts. Fluoride affects the largest change on the proton reson-ance however an association constant was only calculated for 60a (>104 M–1)therefore titrations were repeated in more competitive media (DMSO-d6/5%water). Both receptors showed selectivity for fluoride however 60a boundwith a 1 : 1 binding stoichiometry (1360 M–1) whereas the data for 60b couldonly be fitted to a 1 : 2 receptor/anion model (K1 = 940 M–1 and K2 = 21 M–1).The crystal structures of 60a with chloride and fluoride are shown in Figs. 8and 9 respectively.


Fig. 8 X-ray crystal structure of a chloride complex of 60a reproduced by permission ofThe Royal Society of Chemistry [76]

Fig. 9 X-ray crystal structure of a fluoride complex of 60a reproduced by permission ofThe Royal Society of Chemistry [76]

4Hydroxy (OH) Donors in Neutral Receptors

In 2003, D.K. Smith showed that simple aromatic hydroxides can complexchloride anions. Smith compared stability constants (obtained from NMRcompetition experiments in CD3CN) of phenol, 61, resorcinol, 62 and cat-echol, 63, with chloride (added as its tetrabutylammonium salt) and foundthat 63 bound chloride with greater affinity than 61 and 62 (1015 M–1 for 63against 125 M–1 for 61 and 145 for 62) [77].


Recently, Smith and Winstanley have further explored aromatic hydroxidesas anion receptors by studying the effect of ortho-substituents on the chloridebinding affinity of catechols 64a, 64b and 65 [78]. Binding constants were elu-cidated by proton NMR titrations in CD3CN : DMSO-d6 (9 : 1) solutions andshowed that 65 bound chloride with the highest affinity (235 M–1) presum-ably due to the additional hydrogen bonding provided by the amide groups.Although amide groups are present in 64a and 64b it appeared that they werenot involved in binding the anion and as a result compounds 64a and 64bbound chloride with lower affinities (110 M–1 for 64a and 115 M–1 for 64b).

Row, Maitra and co-workers have linked two steroid subunits to synthe-size macrocycle 66. The fluoride-binding properties of compound 66 werethen investigated by a 1H NMR titration experiment in CDCl3 at 22 ◦C, whichfound that the receptor bound fluoride with a 1 : 2 receptor/anion stoichi-ometry (a result confirmed by Job plot analysis) and stability constants ofK1 = 1.8 (±0.1) ×103 M–1 and K2 = 2.5 (±0.35) ×102 M–1 were found [79].

5Charged Receptors

The incorporation of charged groups into receptors designed for anion recog-nition allows for the receptors to bind the anion with both electrostatic in-teraction and additional interactions dependent on the group and receptor


design. In the case of imidazolium groups, the cation can stabilize the anioncomplex with additional CH· · ·A– type hydrogen bonds.

5.1Imidazolium and Pyridinium-Based Receptors

Kang and Kim have synthesized the fluorescent anion receptor compound67 where two methylene bridged bis-imidazolium subunits are attached toa naphthalene backbone through the 1- and 8-position [80].

Molecular modelling showed that the receptor forms a convergent con-cave cavity with all the imidazolium C(2)-H’s pointing inwards. Modellingstudies led the authors to suggest that the shape of the cavity was predis-posed for the binding of halide anions. Fluorescence titration experiments in90 : 10 CH3CN : DMSO solutions were carried out with chloride, bromide andiodide anions (added as their tetrabutylammonium salts) and stability con-stants were calculated that showed that compound 54 had highest affinity forI– (5000±470 M–1) followed by Br– (243±15 M–1) then Cl– (185±13 M–1).

Yoon, Kim and co-workers have reported a highly effective fluorescent sen-sor for dihydrogen phosphate based on a 1,8-disubstituted-anthracene-dimermacrocycle bridged by two imidazolium subunits (68) [81].

Fluorescence titration experiments were conducted in 9 : 1 acetoni-trile : DMSO solutions in order to elucidate association constants for 68 withdihydrogenphosphate, fluoride, chloride and bromide. The results confirmed


that compound 68 selectively binds to dihydrogenphosphate over the otheranions tested with a stability constant >1 300 000 M–1. Fluoride also boundto compound 68 with high affinity (340 000 M–1) and competition studiesof dihydrogenphosphate and fluoride with respect to compound 68 clearlyshowed that no interference to the dihydrogenphosphate binding occurred inthe presence of fluoride.

Beer and co-workers have shown how a number of tetrakis(imidazolium)macrocyclic receptors, 69a–d, can be used for anion binding [82]. ProtonNMR titration investigations revealed that the macrocycles bind halide anionsstrongly with fluoride being most strongly bound by 69b and 69c (>104 M–1

for both 69b and 69c). Good size complementarity is seen for iodide with 69das it gave the highest stability constant (900 M–1) compared to the other re-ceptors (370 M, 560 and 470 M–1 for 69a, 69b and 69c, respectively). Benzoateanions were found to bind to the receptor in a 1 : 2 host/anion stoichiometry,a result rationalized by the relative size of the benzoate anion compared withthe spherical halides, thus the anion is only partially bound within the cavityallowing a second anion to interact with the cavity.

Alcalde and co-workers have reported that imidazolium-based hetero-phanes, such as 70, are capable of anion recognition [83].

Proton NMR spectroscopy was employed in order to examine the anion-binding behavior of receptor 70. Upon the addition of a number of anionic


guests (added as their tetrabutylammonium salts) to compound 70, signifi-cant changes in the C2 proton resonance of the imidazolium ring were ob-served in both CD3CN and DMSO-d6 solutions. Proton NMR titration experi-ments in DMSO-d6 were then carried out and revealed that 70 binds acetatemost strongly and with a 1 : 1 binding stoichiometry (Ka = 359±42 M–1).

Steed and co-workers have also utilized a tripodal backbone to constructa number of tri-pyridinium “venus flytrap” receptors (71a–c) and investi-gated their anion binding and sensing properties [84, 85].

Receptors 71a and 71b showed similar anion-binding behavior with bothreceptors binding chloride most strongly (constants of >100 000 M–1 calcu-lated for both receptors). In the case of compound 71b reduced affinitieswere observed for both bromide and acetate (3953 M–1 and 2511 M–1, re-spectively) compared to 71a (13 800 M–1 and 10 500 M–1, respectively) whichwas attributed to the increased steric bulk provided by the benzyl groups.For compound 71c chloride is bound stronger than bromide (similar to 71aand 71b) however the affinities for halides are greatly reduced compared to71a and 71b (5370 M–1 for chloride and 486 M–1 for bromide). Receptor 71cwas highly selective for acetate with the stability constant being almost anorder of magnitude higher than the chloride constant (49 000 M–1 against5370 M–1). Variable-temperature 1H NMR experiments were carried out and


showed that compound 71c selectively binds acetate over the spherical halideanions due to mixtures of conformers being adopted in solution throughanthracene-anthracene mutual interactions. Further evidence of the confor-mational behavior of compound 71c and its selectivity for acetate over otheranions was provided by UV spectroscopy and fluorescence studies.

Shinoda and co-workers have reported the one-step synthesis and anionbinding properties of macrocycle 72. Proton NMR titration experiments (inD2O) were used to determine the binding properties of 72 for tricarboxylateanions and revealed that the tricarboxylate 72b was bound with the highestaffinity (log Ka = 5.1) [86].

5.2Guanidinium-Based Receptors

Guanidinium groups may be regarded as charged analogues of ureas inthat they have two parallel NH groups and as a consequence often showhigh affinities for oxyanionic species such as carboxylates binding these an-ions by a combination of hydrogen bonding and electrostatic interactions.Guanidinium-carboxylate and phosphate interactions occur in many biolog-ical systems as a guanidinium group is present in the amino acid arginine [4]

Schmidtchen and co-workers have described the binding of benzoate tothe guanidinium-based receptors 73a and 73b [87]. ITC titrations were car-ried with the iodide salts of 73a and 73b in acetonitrile at 30 ◦C with benzoate(added as its tetraethylammonium salt) and binding constants of 280 000 M–1

and 203 000 M–1 were calculated for 73a and 73b, respectively. Receptor 73awas investigated further where ITC titration experiments (under identicalconditions to the previous titration) were carried with 73a and a variety ofcounter anions. It was found that the change in counter anion had signifi-cant effects upon the binding constant of benzoate observed, for example with


the tetrafluoroborate anion a binding constant of 414 000 M–1 was calculatedcompared to a binding of 38 000 M–1 with chloride.

For several years Schmuck has investigated the binding affinities of guani-dinium salts appended to hydrogen bonding pyrrole-containing motifs andhas shown how carboxylate anion binding is enhanced by the hybrid re-ceptors. In 1999 Schmuck reported the binding ability of 74a with variouscarboxylate anions in highly competitive media [88]. Proton NMR titrationin DMSO-d6/40% H2O at 25 ◦C revealed that 74a formed stronger complexeswith acetate and Ac-L-Phe anions with binding constants of 2790 M–1 and1700 M–1, respectively.

Schmuck then investigated a series of guanidinium-appended pyrrole re-ceptors and found that 74b bound Ac-l-Ala-O– more strongly than 74a(1610 M–1 and 770 M–1, respectively) [89]. The binding ability of 74b wasthen assessed with a range of carboxylate anions by 1H NMR titrations inDMSO-d6/40% H2O at 25 ◦C and showed that compound 74b formed strongercomplexes with the anions than 74a. High affinities were observed for 2-pyrrole-COO– and acetate (5275 M–1 and 3380 M–1). A notable result wasthat compound 74b displayed enantioselectivity in the case of Ac-Ala-O– an-ions where a higher affinity was observed for the l-enantiomer over thed-enantiomer (1610 M–1 vs. 930 M–1).

In 2005 Schmuck and Schwegmann reported the study of a tripodal “mo-lecular flytrap” 75 where the pyrrole-guanidinium moieties were appendedto a triamide backbone. The receptor was designed to bind tricarboxylate an-ions and UV and fluorescence titration experiments in water showed that 75bound citrate and trimesoate with association constants >105 M–1 [90].

Recently, de Mendoza and co-workers have reported the complexationof nitrate to an acyclic cleft and a series of macrocycles based on guani-dinium [91].

Association constants were calculated by ITC titrations in acetonitrile at303 K and it was found that the macrocyclic receptors bound the nitrate an-


ion more strongly than the acyclic system. The constants obtained for themacrocycles 77a–c revealed a size dependence on the binding of nitrate withthe largest macrocycle 77c giving rise to the highest association constant of73.7×103 M–1, an order of magnitude greater than the smallest macrocycle77a (7.26×103 M–1).

5.3Ammonium-Containing Receptors

The work of Park and Simmons [92] has inspired many efforts into the re-search of ammonium- and polyammonium-based anion receptors and arethe subject of numerous reviews [93, 94] Here we will only look at recep-tors containing quaternary ammonium centers. These groups bind anions viaelectrostatic interactions only.

An early pioneer in the area of ammonium-based anion receptors,Schmidtchen synthesized the quaternary ammonium-based macrocyclic re-ceptors 78a–c in 1977 [95].


NMR spectroscopy revealed that upon the addition of 1 equiv. of alkalimetal halide salts to receptors 78a–c, 1 : 1 complexes were formed. Stabilityconstants were measured using halide electrodes in water solutions, which re-vealed that the receptors bound bromide anions (log Ka = 1.8, 2.45 and 2.45for 78a, 78b and 78c, respectively) and iodide anions (log Ka = 2.2 and 2.4 for78b and 78c, respectively).

Frontera, Anslyn and co-workers have described the binding of tricarboxy-late salts with the tris-ammonium-squaramide-appended tripodal receptor79 [96]. Isothermal titration calorimetry was used to study the association ofa number of tricarboxylate salts and 79 in 1 : 3 water:ethanol solutions at 294 K.It was found that 79 bound the less rigid tricarboxylates citrate and tricaballatemore strongly than the rigid benzene-1,3,5-tricarboxylate (1.1±0.1×105 M–1,1.5 ± 0.2×105 M–1 and 4.5 ± 0.5×104 M–1, respectively). Compound 79 wasalso investigated as a receptor for the biscarboxylates gluterate and succinate.The stability constants revealed that gluterate was bound with significantlyhigher affinity than succinate (2.2±0.2×104 vs. ∼ 2.8×102, respectively).

Costa and co-workers have reported 80, a fluorescent squaramide-containing macrocyclic receptor for monitoring sulfate in water [97]. Isother-mal titration calorimetry was employed to characterize the host–guestassociation of 80 with SO4

2–, PhOPO32– and C2O4

2– dianions (titrationcarried out in methanol at 294 K). The data was fitted to a 1 : 1 bindingmodel and it was found that 80 bound SO4

2– (4.6 ± 1.0×106 M–1) with the


strongest affinity followed by C2O42– (3.2 ± 0.3×105 M–1) then PhOPO3

2–

(1.5 ± 0.2×104 M–1). A fluorescein-80 complex was then synthesized andcompetitive fluorescent titration experiments in 9:1 methanol:water solutionswere conducted with the complex against sodium sulfate and an associationconstant of 5.2±1.2×106 M–1 was calculated.

Bowman-James and co-workers have synthesized a series of amide-basedmacrocycles containing either tertiary amine spacer groups or quaternizedammonium functionalities and have assessed their ability to bind a numberof anions in solution [98].

Stability constants were calculated by proton NMR titration experiments inDMSO-d6 solutions and revealed that the quaternized macrocycles 82a and82b showed higher affinities for anions compared to the neutral analogues81a and 81b attributed to the additional electrostatic attraction of the qua-ternary ammonium group. The pyridine analogues 81b and 82b were alsoshown to have higher binding constants than the isophthaloyl derivatives 81aand 82a attributed to the pyridine-assisted preorganization of the macrocy-cle. All the anions binding data was fitted to 1 : 1 isotherms and receptor 82bwas found to have the highest affinity for anions with dihydrogenphosphatebeing most strongly bound (log K = 5.32). Receptor 82b was found to bindhalide anions in the order Cl– > Br– > I– > F– whilst receptor 82a was alsofound to bind chloride more strongly than other halide anions (log K = 3.23and 2.14 for 82a with Cl– and Br–, respectively), illustrating the good sizecomplementarity of 82a and 82b to chloride.


6Conclusions

The examples discussed in this review provide a broad overview of the varietyof synthetic organic receptors used for binding and sensing of anionic species.As the understanding of the processes and factors that influence the effect-ive binding of anions improves there is an increasing impetus to apply thisknowledge to solve real-world problems. Areas that are likely to benefit fromthis knowledge are in the transport of anions, in separation processes and inbiological systems leading towards new treatments for disease and cancer.

Acknowledgements We would like to thank the EPSRC/Crystal Faraday for a project stu-dentship (GWB).

References

1. Gale PA, Quesada R (2006) Coord Chem Rev 250:32192. Gale PA, García-Garrido SE, Garric J (2008) Chem Soc Rev. DOI 10.1039/b715825d3. Gale PA (2006) Acc Chem Res 39:4654. Sessler JL, Gale PA, Cho W-S (2006) Anion Receptor Chemistry. RSC Publishing,

Cambridge5. Xu H, Strater N, Schroder W, Bottcher C, Ludwind K, Saenger W (2003) Acta Cryst

D59:8156. Pflugrath JW, Quiocho FA (1985) Nature 314:2577. Pflugrath JW, Quiocho FA (1988) J Mol Biol 200:1638. He JJ, Quiocho FA (1991) Science 251:14979. Luecke H, Quiocho FA (1990) Nature 347:402

10. Pascal RA, Spergel J, van Engen D (1986) Tetrahedron Lett 27:409911. Valiyaveettil S, Engbersen JFJ, Verboom W, Reinhoudt D (1993) Angew Chem Int Ed

32:90012. Kavallieratos K, de Gala SR, Austin DJ, Crabtree RH (1997) J Am Chem Soc 119:232513. Hughes MP, Smith BD (1997) J Org Chem 62:449214. Santacroce PV, Davis JT, Light ME, Gale PA, Iglesias-Sanchez JC, Prados P, Quesada R

(2007) J Am Chem Soc 129:188615. Webb JEA, Crossley MJ, Turner P, Thordarson P (2007) J Am Chem Soc 129:715516. Prohens R, Tomas S, Morey J, Deya PM, Ballester P, Costa A (1998) Tetrahedron Lett

39:106317. Davis AP, Perry JJ, Williams RP (1997) J Am Chem Soc 119:179318. Choi K, Hamilton AD (2001) J Am Chem Soc 123:245619. Szumna A, Jurczak J (2001) Eur J Org Chem, p 403120. Chmielewski MJ, Jurczak J (2005) Chem Eur J 11:608021. Chmielewski MJ, Jurczak J (2006) Chem Eur J 12:765222. Kavallieratos K, Bertao CM, Crabtree RH (1999) J Org Chem 64:167523. Kang SO, Powell D, Bowman-James K (2005) J Am Chem Soc 127:1347824. Kang SO, Powell D, Day VW, Bowman-James K (2006) Angew Chem Int Ed 45:192125. Kubik S, Goddard R, Kirchner R, Nolting D, Seidel J (2001) Angew Chem Int Ed

40:2648


26. Kubik S, Kirchner R, Nolting D, Seidel J (2002) J Am Chem Soc 124:1275227. Gunnlaugsson T, Kruger PE, Jensen P, Pfeffer FM, Hussey GM (2003) Tetrahedron Lett

44:890928. Camiolo S, Gale PA, Hursthouse MB, Light ME, Shi AJ (2002) Chem Commun, p 75829. Gale PA, Navakhun K, Camiolo S, Light ME, Hursthouse MB (2002) J Am Chem Soc

124:1122830. Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Org Biomol Chem 1:74131. Boiocchi M, Boca LD, Gomez DE, Fabbrizzi L, Licchelli M, Monzani E (2004) J Am

Chem Soc 126:1650732. Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res

39:34333. Gunnlaugsson T, Davis AP, Hussey GM, Tierney J, Glynn M (2004) Org Biomol Chem

2:185634. Kwon JY, Jang YJ, Kim SK, Lee K-H, Kim JS, Yoon J (2004) J Org Chem 69:515535. Brooks SJ, Gale PA, Light ME (2005) Chem Commun, p 469636. Brooks SJ, Garcia-Garrido SE, Light ME, Cole PA, Gale PA (2007) Chem Eur J 13:332037. Brooks SJ, Edwards PR, Gale PA, Light ME (2006) New J Chem 30:6538. Kondo S-I, Nagamine M, Yano Y (2003) Tetrahedron Lett 44:880139. Pfeffer FM, Gunnlaugsson T, Jensen P, Kruger PE (2005) Org Lett 7:535740. Ayling AJ, Perez-Payan N, Davis AP (2001) J Am Chem Soc 123:1271641. Clare JP, Ayling AJ, Joos J-B, Sisson AL, Magro G, Perez-Payan MN, Lambert TN,

Shukla R, Smith BD, Davis AP (2005) J Am Chem Soc 127:1073942. Koulov AV, Lambert TN, Shukla R, Jain M, Boon JM, Smith BD, Li H, Sheppard DN,

Joos J-B, Clare JP, Davis AP (2003) Angew Chem Int Ed 42:493143. Snellink-Ruel BHM, Antonisse MMG, Engbersen JFJ, Timmerman P, Reinhoudt DN

(2000) Eur J Org Chem, p 16544. Brooks SJ, Gale PA, Light ME (2006) Chem Commun, p 434445. Lee KH, Hong J (2000) Tetrahedron Lett 41:608346. Hisaki I, Sasaki S-I, Hirose K, Tobe Y (2007) Eur J Org Chem, p 60747. Sessler JL, Camiolo S, Gale PA (2003) Coord Chem Rev 240:1748. Shionoya M, Furuta H, Lynch V, Harriman A, Sessler JL (1992) J Am Chem Soc

114:571449. Kral V, Furuta H, Shreder K, Lynch V, Sessler JL (1996) J Am Chem Soc 118:159550. Gale PA, Camiolo S, Tizzard GJ, Chapman CP, Light ME, Coles SJ, Hursthouse MB

(2001) J Org Chem 66:784951. Gale PA, Light ME, McNally B, Navakhun K, Sliwinski KE, Smith BD (2005) Chem

Commun, p 377352. Sessler JL, Pantos GD, Gale PA, Light ME (2006) Org Lett 8:159353. Sessler JL, Barkey NM, Pantos GD, Lynch VM (2007) New J Chem 31:64654. Vega IED, Camiolo S, Gale PA, Hursthouse MB, Light ME (2003) Chem Commun,

p 168655. Vega IED, Gale PA, Hursthouse MB, Light ME (2004) Org Biomol Chem 2:293556. Sessler JL, Katayev E, Pantos GD, Ustynyuk YA (2004) Chem Commun, p 27657. Sessler JL, Katayev E, Pantos GD, Scherbakov P, Reshetova MD, Khurstalev VN,

Lynch VM, Ustynyuk YA (2005) J Am Chem Soc 127:1144258. Katayev E, Boev N, Khurstalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007) J Org

Chem 72:288659. Gale PA, Sessler JL, Kral V, Lynch VM (1996) J Am Chem Soc 118:514060. Custelcean R, Delmau LH, Moyer BA, Sessler JL, Cho WS, Gross D, Bates GW,

Brooks SJ, Light ME, Gale PA (2005) Angew Chem Int Ed 44:2537


61. Sessler JL, Gross D, Cho WS, Lynch VM, Schmidtchen FP, Bates GW, Light ME,Gale PA (2006) J Am Chem Soc 128:12281

62. Bates GW, Gale PA, Light ME (2006) Cryst Eng Comm 8:30063. Gale PA, Sessler JL, Kral V (1998) Chem Commun, p 164. Gale PA, Anzenbacher P Jr, Sessler JL (2001) Coord Chem Rev 222:5765. Lee C-H, Lee J-S, Na H-K, Yoon D-W, Miyaji H, Cho W-S, Sessler JL (2005) J Org

Chem 70:206766. Miyaji H, Hong S-J, Jeong S-D, Yoon D-W, Na H-K, Hong J, Ham S, Sessler JL, Lee C-H

(2007) Angew Chem Int Ed 46:250867. Cafeo G, Kohnke FH, La Torre GL, White AJP, Williams DJ (2000) Angew Chem Int

Ed 39:149668. Cafeo G, Kohnke FH, La Torre GL, Parisi MF, Nascone RP, White AJP, Williams DJ

(2002) Chem Eur J 8:314869. Cafeo G, Kohnke FH, White AJP, Garozzo D, Messina A (2007) Chem Eur J 13:64970. Chmielewski MJ, Charon M, Jurczak J (2004) Org Lett 6:350171. Piatek P, Lynch VM, Sessler JL (2004) J Am Chem Soc 126:1607372. Curiel D, Cowley A, Beer PD (2005) Chem Commun, p 23673. Sessler JL, Cho D-G, Lynch V (2006) J Am Chem Soc 128:1651874. Chang K-J, Chae MK, Lee C-H, Lee J-Y, Jeong K-S (2006) Tetrahedron Lett 47:638575. Chang K-J, Moon D, Lah MS, Jeong K-S (2005) Angew Chem Int Ed 44:792676. Bates GW, Gale PA, Light ME (2007) Chem Commun, p 212177. Smith DK (2003) Org Biomol Chem 1:387478. Winstanley KJ, Smith DK (2007) J Org Chem 72:280379. Ghosh S, Choudhury AR, Row TN, Maitra U (2005) Org Lett 7:144180. Kim H, Kang J (2005) Tetrahedron Lett 46:544381. Yoon J, Kim SK, Singh J, Lee JW, Yang YJ, Chellappan K, Kim KS (2004) J Org Chem

69:58182. Wong WWH, Vickers MS, Cowley AR, Paul RL, Beer PD (2005) Org Biomol Chem

3:420183. Alcalde E, Mesquida N, Perez-Garcia L (2006) Eur J Org Chem, p 398884. Abouderbala LO, Belcher WJ, Boutelle MG, Cragg PJ, Dhaliwal J, Fabre M, Steed JW,

Turner DR, Wallace KJ (2002) Chem Commun, p 35885. Wallace KJ, Belcher WJ, Turner DR, Syed KF, Steed JW (2003) J Am Chem Soc

125:969986. Shinoda S, Tadokoro M, Tsukube H, Arakawa R (1998) Chem Commun, p 18187. Haj-Zaroubi M, Mitzel NW, Schmidtchen FP (2002) Angew Chem Int Ed 41:10488. Schmuck C (1999) Chem Commun, p 84389. Schmuck C (2000) Chem Eur J 6:70990. Schmuck C, Schwegmann M (2005) J Am Chem Soc 127:337391. Blondeau P, Benet-Buchholz J, de Mendoza J (2007) New J Chem 31:73692. Park CH, Simmons HE (1968) J Am Chem Soc 90:243193. Garcia-Espana E, Diaz P, Llinares JM, Bianchi A (2006) Coord Chem Rev 250:295294. Llinares JM, Powell D, Bowman-James K (2003) Coord Chem Rev 240:5795. Schmidtchen FP (1977) Angew Chem Int Ed Engl 16:72096. Frontera A, Morey J, Oliver A, Piña MN, Quiñonero D, Costa A, Ballester P, Deyà PM,

Anslyn EV (2006) J Org Chem 71:718597. Prohens R, Martorell G, Ballester P, Costa A (2001) Chem Commun, p 145698. Hossain MA, Kang SO, Powell D, Bowman-James K (2003) Inorg Chem 42:1397

Struct Bond (2008) 129: 45–94DOI 10.1007/430_2007_073© Springer-Verlag Berlin HeidelbergPublished online: 19 January 2008

Metal-Based Anion Receptor Systems

Simon R. Bayly (�) · Paul D. Beer (�)

Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford,South Parks Road, Oxford OX1 3QR, [email protected], [email protected]

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461.1 Metal Complexes in Anion Sensing . . . . . . . . . . . . . . . . . . . . . . 461.2 Basis of Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2 Redox-Active Transition Metal-Based Receptorsfor Electrochemical Anion Sensing . . . . . . . . . . . . . . . . . . . . . . 47

2.1 Theory of Electrochemical Sensing . . . . . . . . . . . . . . . . . . . . . . 472.2 Metallocene Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . . . . 482.3 Mixed Metal Metallocene-Lewis Acid Anion Receptors . . . . . . . . . . . . 532.4 Transition Metal Polypyridyl Anion Receptors . . . . . . . . . . . . . . . . 562.5 Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing . 592.6 Dendrimers as Redox Anion Sensors . . . . . . . . . . . . . . . . . . . . . 602.7 Surface Confined Redox Anion-Sensing Systems . . . . . . . . . . . . . . . 63

3 Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.1 Theory of Optical Anion Sensing . . . . . . . . . . . . . . . . . . . . . . . . 683.2 Metallocene Optical Anion Sensors . . . . . . . . . . . . . . . . . . . . . . 683.3 Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors . . . . . . 723.4 Optical Anion Sensing Using Reporter Groups

Based on Other Transition-Metal Complexes . . . . . . . . . . . . . . . . . 753.5 Transition Metals as Anion-Binding Groups and/or Structural Components 793.6 Optical Anions Sensing by Lanthanide(III) Complexes . . . . . . . . . . . . 863.7 Surface Confined Systems for Optical Anion Sensing . . . . . . . . . . . . . 88

4 Metal-Based Anion Receptors Without Reporter Groups . . . . . . . . . . 89

5 Conclusion/Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

Abstract Metal complexes play an important role in anion receptor chemistry. In themajority of examples metal centres are used as optical and/or electrochemical re-porter groups in anion-sensing applications. Metal centres can also act as Lewis acidicanion-binding sites in their own right, and/or as structural components allowing theself-assembly of anion-binding domains. This review describes the development of metal-based receptors with regard to their anion-sensing properties, and is therefore dividedinto sections on electrochemical and optical anion sensing. Within these sections cover-age has been given to the diverse range of metals and anion-binding groups that havebeen studied. Emphasis has been placed on recently described novel supramolecular,nanoscale and surface confined anion receptor systems that give added functionality. Thelast section describes metal-based anion receptors without reporter groups.

46 S.R. Bayly · P.D. Beer

Keywords Anion sensing · Anion recognition · Anion coordination · Anion receptors ·Organometallic receptors

1Introduction

1.1Metal Complexes in Anion Sensing

Metal complexes have played an important role in anion receptor chemistrysince its earliest examples. The presence of a metal ion can introduce a rangeof advantageous physicochemical properties to this class of receptor. In themajority of examples the metal complex is incorporated as a reporter group,whose photochemical or redox response is changed upon proximal binding ofan anion. Furthermore, the metal can contribute directly to anion binding, ei-ther by using its positive charge to electrostatically attract the anion and/orby acting as a Lewis acidic binding site. The metal complex motif can alsobe utilised as a structural component in anion receptors, where its coordina-tion geometric properties allow the self-assembly of receptor sites with a widerange of topologies not possible with simple tetrahedral covalent bonds. Byexploiting these different properties, often in combination, metal complexanion receptors achieve a range of functionality beyond the scope of purelyorganic structures.

Previous reviews on this topic have included many aspects of anion recog-nition by metal-based receptors [1–6]. This review does not seek to be com-prehensive; instead it is designed to provide an introduction to the areaby highlighting notable examples and to bring the reader up to date withsignificant recent results, especially in the application of metal-based anionreceptors in surface fabricated nanoscale sensor systems.

1.2Basis of Anion Sensing

Molecular sensing refers to a remotely detectable change in the proper-ties of a receptor molecule on binding of an analyte. The generic designfor an anion sensor utilises a spacer group to covalently link an anion re-ceptor site to a signalling or reporter group. Provided the spacer allowssome degree of coupling between the components (through-space, through-bond, or by conformational change), binding of an anion at the receptorsite perturbs the electronic properties of the signalling group in a way thatcan be detected spectroscopically or electrochemically. Thus, when the sig-nalling group is a suitable redox-active metal centre, binding can be probedelectrochemically (e.g. by voltammetry). If a suitable chromophore or flu-

Metal-Based Anion Receptor Systems 47

orophore is used, sensing can be accomplished via optical spectroscopy(UV/Vis absorption or luminescence, respectively). The spectroscopic prop-erties of many metal complexes make them particularly amenable for this.NMR spectroscopy can also be used to detect anion binding to diamag-netic metal-based receptors. However, this is not generally regarded asa remote sensing technique since it obviously requires the sample to beplaced in an external magnetic field. This review is conveniently dividedinto sections on metal-based receptors for electrochemical anion sensing,receptors for optical anion sensing and anion receptors without reportergroups.

2Redox-Active Transition Metal-Based Receptorsfor Electrochemical Anion Sensing

2.1Theory of Electrochemical Sensing

When a redox-active transition metal is used as the signalling unit of a recep-tor, anion binding is coupled to electron transfer, i.e. anion binding changesthe redox potential (couple) of the transition metal. This electrochemicalshift can be represented as ∆E0, the difference in redox potentials betweenthe receptor : anion complex and the receptor alone. Concomitantly, elec-tron transfer at the redox centre also changes the affinity of the receptor forthe guest species. These coupled processes are linked thermodynamically byEq. 1, where Kred and Kox are the stability constants of the reduced and oxi-dised forms of the receptor:anion complex respectively [7].

nF(∆E0) = RT ln(Kox/Kred) (1)

From a thermodynamic standpoint, the value of the shift in redox potentialis determined by the ratio of Kox/Kred, instead of the absolute value of eitherKox or Kred. As a consequence a receptor need not necessarily have a veryhigh binding strength for the anion to be sensed. If electron transfer leadsto a sufficiently large change in the stability of the receptor : anion complex,a measurable change in redox potential can be observed.

Anion binding stabilises the oxidised form of the receptor, hence Kox/

Kred > 1 and the redox potential of the reporter group is shifted to a morenegative value (cathodic shift). The Kox/Kred ratio is also a measure of howefficient the coupling is between the metal-based reporter group and theanion-binding site.


2.2Metallocene Redox Anion Sensors

Anion receptors incorporating the redox-active cobaltocenium group havebeen studied extensively due to the combination of an accessible redoxcouple and favourable electrostatic interactions of the cationic organometallicmetallocene motif with anions. The first anion receptors based on thisspecies were reported by Beer and co-workers in 1989 [8]. The macro-cyclic bis-cobaltocenium receptor 1 was shown to bind bromide in ace-tonitrile solution (due to electrostatic interaction). Electrochemical anionsensing was also demonstrated, where bromide caused the potential of thecobaltocenium/cobaltocene to undergo a cathodic shift. Augmentation ofcobaltocenium-based receptors with hydrogen-bond donor groups, such asamides in receptors 2 and 3, generates both stronger and more selective an-ion binding [9]. Proton NMR anion titration studies in CD3CN reveal 2 and3 to have selectivity for dihydrogenphosphate over chloride by approximatelyan order of magnitude [10]. This is attributed not only to the greater basic-ity of the dihydrogenphosphate anion, but also to complementary hydrogenbonding between the receptor and the anion. In these receptors the elec-trochemical sensing of anions is also enhanced, with chloride giving riseto cathodic shifts in the cobaltocenium/cobaltocene redox couple of 30 and85 mV for 2 and 3, respectively. The dihydrogenphosphate anion generatescathodic shifts of 200 and 240 mV respectively, confirming that in this classof amide hydrogen-bonding receptors the magnitude of the electrochemicalresponse directly mirrors the strength of the receptor-anion interaction.

Ferrocene has also been extensively exploited in redox responsive anionreceptor design. One advantage of ferrocene is that its synthetic chemistry


is highly developed, in particular in terms of its conjugation to organicmolecules. From the point of view of anion sensing the most relevant dif-ference between the metallocenes is that ferrocene is neutral in charge andtherefore its derivatives have no inherent electrostatic interaction with anions(until oxidised to ferrocenium) and therefore their complexes with anionsexhibit relatively lower stability constants.

Molecules 4–8 are a selection of ferrocene-based receptors which incorpo-rate amide groups for the hydrogen-bonding of anions [11, 12]. In acetonitrilesolution dihydrogenphosphate induced cathodic shifts of up to 240 mV inthe ferrocene/ferrocenium couples of these receptors. Competition experi-ments demonstrated the same shift even in the presence of a 10-fold excessof chloride or hydrogensulfate. In these receptors it is largely the stability ofthe electrostatically enhanced anion : ferrocenium complex which determinesthe magnitude of the redox shift. Receptor 8 has the opposite selectivity, dis-playing a hydrogensulfate induced shift of 220 mV which does not changein the presence of excess dihydrogenphosphate. It is thought that binding of


HSO4– leads to protonation of the amine group of the receptor. The result-

ing complex is cationic and this has a very high affinity for the residual SO42–

anion. Electrostatic interactions are particularly important in redox anionsensing, and even very simple anion-binding motifs such as the ammoniumcation provide an increased redox response. This has been demonstratedby Moutet and co-workers using 9 [13]. This molecule is able to sense di-hydrogenphosphate and ATP2– in a range of solvents, displaying a shift of470 mV in CH2Cl2 with dihydrogenphosphate, solely due to a strong ion-pairing interaction.

Sensing anions in aqueous conditions is a particular challenge whichmust be met for molecular anion receptors to become a useful technologyin biological or environmental analysis. The high dielectric and compet-itive hydrogen-bond donor capacity of water diminishes anion-receptorinteractions. In general strong electrostatic interactions are required toovercome this. Beer et al. have developed a series of ferrocene-based re-ceptors appended with various open chain and cyclic amine functionalgroups, e.g. molecules 10 and 11, that bind ATP2– and dihydrogenphos-phate in water [14–16]. The selectivity of this class of receptors is pHdriven. At pH 6.5 at least two of the amines are protonated and the 1 : 1anion : receptor complexes formed show cathodic shifts of 60–80 mV in theferrocene/ferrocenium redox couple. Quantitative determination of phos-phate and sulfate in the presence of competitor anions was demonstrated bymetallacyclic receptors 12 and 13. The electrochemistry of these receptorswas studied in 70 : 30 THF : H2O over a range of pHs. A maximum selectiveredox shift of 54 mV for phosphate over sulfate was observed at pH 4 for 12,whereas 13 gave a maximum selective redox shift of 50 mV for phosphate oversulfate at pH 7.


Recognition of fluoride in aqueous media is particularly difficult due to thestrongly hydrated nature of the anion. Shinkai and co-workers have demon-strated that ferrocene-boronic acid 14 acts as a selective redox sensor for flu-oride which operates in H2O [17]. The favourable interaction between boronand fluoride (a hard-acid and hard-base, respectively) generates a stabilityconstant of 700 M–1 for the fluoride-ferrocenium complex. Stability constantsfor both the bromide and chloride complexes are < 2 M–1.

In receptor molecules that contain multiple metal centres and anion bind-ing groups the redox sensing properties are dictated by the precise spatialarrangement of these groups. In receptors where the pendant reporter groupsare in close proximity to each other an increased redox response to anionbinding is often observed. For example, the cyclotriveratrylene amides 15 and16 include closely spaced multiple amide anion-binding groups with pen-dant ferrocene reporter units [18]. In CH2Cl2 or acetonitrile solution cathodicshifts of up to 260 mV were observed in the presence dihydrogenphosphate orATP2–. On oxidation 15 and 16 gain a triple positive charge. The magnitude ofthe redox shift can be attributed to the increased electrostatic affinity of thismultiply charged oxidised species for a single anion compared to monoferro-cenyl receptors which are monopositive on oxidation.

Receptor 17 has a similar topology—it comprises four cobaltoceniumamide groups attached to a porphyrin backbone as the cis-α,α,α,α-atrop-isomer [19]. Proton NMR titrations in CD3CN showed chloride and bro-mide to be bound in 1 : 1 stoichiometry with stability constants of 860and 820 M–1, respectively, whereas nitrate exhibited weaker binding withK = 190 M–1. Electrochemical studies displayed cathodic shifts in the cobal-tocene/cobaltocenium redox couple of 35–75 mV on addition of chloride orhydrogensulfate, and 225 mV for dihydrogenphosphate in acetonitrile solu-


tion. Smaller shifts were seen in the porphyrin oxidation wave. The overallselectivity trend was Cl– > Br– � NO3

–.The anion recognition properties of cobaltocenium calix[4]arene recep-

tors 18–20 were found to be dependent on the structure of the upper-rim ofthe calix[4]arene [20]. In 1H NMR studies in DMSO-d6 solution 18 showsa greater affinity for acetate than for dihydrogenphosphate whereas its iso-mer 19 displays the opposite trend. In 20 there is only a single cobaltoceniumgroup which bridges the upper rim of the calix[4]arene. This receptor displaysa significantly greater affinity for the above anions despite possessing onlya single positive charge. For example, the cobaltocene/cobaltocenium redoxcouple of 20 was found to undergo a cathodic shift of 155 mV in the presenceof acetate. It is proposed that the surprising strength of the interaction is dueto the topology of the anion-binding cavity, in which the arrangement of thetwo amide hydrogen bond donors is complementary to bidentate anions suchas carboxylates. The same selectivity is seen in the ferrocene-1,1′-bisamideanalogues 21–23 [21]. Results of 1H NMR studies in CD3CN show that thesereceptors also preferentially bind carboxylate anions (acetate and benzoate)over dihydrogenphosphate and chloride.

Other carboxylate selective redox sensors based on ferrocene include neu-tral molecule 24, which utilises hydrogen bonding to bind mono and dicar-


boxylate anions with 2 : 1 and 1 : 1 guest : host stoichiometry [22]. However,it cannot distinguish between these types of anion electrochemically, givinga maximum cathodic shift of 150 mV for both acetate and phthalate. A morerecent example shows that selectivity for dicarboxylate anions over monocar-boxylates and other simple anions can be achieved. Tetra-ammonium macro-cycle 25 binds phthalate, isopthalate and dipicolinate with a 2 : 1 guest : hoststoichiometry, giving maximum cathodic shifts in the redox potential of275, 193 and 168 mV, respectively [23]. In comparison the monoacid 4-nitrobenzoate produced a maximum cathodic shift of only 49 mV.

The incorporation of crown ether units into a cobaltocenium receptor hasbeen shown to allow the switchable redox sensing of anions. Proton NMRtitrations of receptor 26 in CD3CN solution gave log K values of 3.1 for chlo-ride and 3.0 for bromide [24]. Electrochemical titrations showed cathodicshifts of the cobaltocene/cobaltocenium redox couple of 60 and 30 mV for thetwo anions, respectively. However, when either the NMR or electrochemicaltitrations were carried out in the presence of K+ no significant anion inducedshifts were observed. It is proposed that the K+ ions form a 1 : 1 intramolec-ular sandwich complex with the two benzocrown ether units of the receptorcausing a concomitant change in conformation of the amide groups whichreduces their availability for anion binding.

Molina and co-workers have investigated the urea-crown ether function-alised ferrocene 27 [25]. This receptor-produced anion induced cathodic shiftsin the ferrocene/ferrocenium redox couple of 52 mV with fluoride and 190 mVwith dihydrogenphosphate. In acetonitrile solution on addition of 2 equiva-lents of K+ ions a dramatic attenuation in the anion-induced cathodic shift wasobserved, with dihydrogenphosphate giving rise to a shift of only 50 mV.


2.3Mixed Metal Metallocene-Lewis Acid Anion Receptors

Lewis acidic metal centres can be utilized as anion-binding groups, which incombination with ferrocenyl reporter groups provide enhanced redox anionsensing. Ferrocene-amide receptor 28 utilises pendant phosphine groups toallow the coordination of various transition metals to generate mixed-metalcomplexes 29–32 [26]. 1H NMR titrations carried out in CD2Cl2 solutionrevealed that the neutral molecules 29–31 bind chloride approximately anorder of magnitude more strongly than the parent phosphine. The affin-ity of these complexes for bromide, iodide and hydrogenphosphate was alsofound to be increased, but the effect was smaller. Cationic complex 32 wasfound to bind the same anions an order of magnitude more strongly again,due to the added influence of the electrostatic attraction. With chloride,bromide and hydrogensulfate in acetonitrile/dichloromethane solution a sig-nificant anion-induced cathodic redox shift was observed in both the fer-rocene/ferrocenium couple and the irreversible oxidation of the secondarymetal centre.

Receptor 33 also incorporates a secondary Lewis acid anion bindingsite. This molecule is the zinc metallated analogue of 17 with the cobal-tocenium reporter groups replaced with ferrocenes [27]. The freebase pre-cursor to 33 in dichloromethane solution shows no significant anion in-duced shifts in the 1H NMR signals of the amide protons, whereas themetalloporphyrin binds bromide (K = 6200 M–1), nitrate (K = 2300 M–1)


and hydrogenphosphate (K = 2100 M–1). Electrochemical studies in 3 : 2dichloromethane/acetonitrile revealed anion-induced cathodic shifts inboth the porphyrin (∆E = 85–115 mV) and tetraferrocene oxidation (∆E =20–60 mV) waves. The trend in magnitude of ∆E for the porphyrin oxidationwave is hydrogensulfate > chloride > bromide > nitrate, reflecting the chargedensity (charge to radius ratio) of the anionic guest species. Atropisomers of33 (other than the α,α,α,α-atropisomer) were also studied and showed dif-


ferent anion-sensing properties. For instance the α,β,α,β-atropisomer wasfound to be selective for nitrate (∆Eporphyrin = 175, ∆Eferrocene = 110).

Jurkschat and co-workers have investigated the redox anion-sensingproperties of metallomacrocycle 34 [28]. This comprises two ferrocene re-porter units linked together covalently by two Lewis acidic organotin spac-ers. Electrochemical measurements in dichloromethane solution showedanion-induced cathodic shifts in the ferrocene/ferrocenium redox couple of130 mV for chloride, 210 mV for fluoride and 480 mV for dihydrogenphos-phate.

Another class of mixed-metal anion receptors has been investigated whichpossess redox reporter groups based on two different metal complexes. Thisenables the qualitative comparison of their comparative anion-sensing abili-ties. Macrocycles 35 and 36 combine the {RuII(bpy)3}moiety with a bridgingferrocene or cobaltocenium unit [29]. Electrochemical experiments in ace-tonitrile solution revealed that the RuII/RuIII redox potential was insensitiveto anion binding, whereas the ferrocene/ferrocenium (in 35) and cobal-tocene/cobaltocenium (in 36) redox couples were shifted cathodically (by60 mV and 110 mV respectively with chloride). However, the first reductionof {RuII(bpy)3}, a ligand-centred process based on the amide substitutedbipyridyl, was also found to undergo an anion induced cathodic shift (40 mVand 90 mV with chloride for 35 and 36, respectively).

2.4Transition Metal Polypyridyl Anion Receptors

Redox sensing of anions using {RuII(bpy)3}-amides as a combined recep-tor/reporter system has also been studied using complexes 37–40 [30–32].The single crystal X-ray structure of the chloride complex of 37 clearly in-dicates that the anion is bound tightly within the bipy amide ligand by sixhydrogen bonds. It forms hydrogen bonds not only to the two N–H groupsbut also to four aromatic C–H groups with H–Cl distances ranging from 2.51to 2.71 A. 1H NMR titrations in DMSO-d6 revealed strong binding of chlorideand dihydrogenphosphate. Four reversible redox couples (one metal-centredoxidation and three ligand centred reductions are expected for {RuII(bpy)3}species) were observed in electrochemical studies. Of these only the least ca-thodic ligand reduction was significantly shifted in the presence of anionicguests. The assignment of this redox process to the relatively electron-pooramide-substituted bipyridyl reaffirms the X-ray structure evidence that an-ion recognition takes place at this site. It is interesting to note that thecalix[4]arene modified molecule 40 which can be compared with the cobal-tocenium and ferrocene complexes 7 and 18–23, shows particular selectivityfor dihydrogenphosphate and is able to electrochemically sense this anion(giving a cathodic shift of 175 mV) in the presence of a 10-fold excess of chlo-ride or hydrogensulfate [30].


{RuII(bpy)3} macrocycles 41–44 demonstrate the influence of the top-ology of the anion-binding site on redox sensing [29]. The macrocycles withthe larger cavities 42–44 were shown by 1H NMR studies in DMSO-d6 tobind chloride preferentially to acetate and dihydrogenphosphate. This is theopposite trend to 41, which has the smallest macrocyclic cavity and bindsacetate more strongly than either chloride or dihydrogenphosphate. Recep-tor 43 displays outstanding selectivity; with a stability constant for chlorideof 40 000 M–1 and no measurable affinity for dihydrogenphosphate. The pres-ence of a second positively charged metal centre in 43 and 44 leads to in-creased overall anion affinity. The substitution of one Ru(II) centre for Os(II)in 44 results in both an increased stability constant for the chloride com-plex and a diminished affinity for acetate—in effect an increase in selectivitybetween the two anions. Results of electrochemical measurements on thecomplexes in acetonitrile solution confirmed the pattern of anion selectivity.In 41 the cathodic shift in the first bpy redox wave was 30 mV for chlorideand 55 mV for acetate. In 43 and 44 cathodic shifts of 110 and 125 mV, re-spectively, were recorded in the presence of chloride, whereas for acetate thechange was 15 and 20 mV, respectively. This class of anion receptor also ex-hibits optical anion-sensing properties (see Sect. 3.2).{RuII(bpy)3} has also been used as the basis of a redox sensor for fluoride.

Receptors 45 and 46 were studied in acetonitrile solution using differential


pulse voltammetry. Addition of two equivalents of fluoride was found to pro-duce a peak at 0.73 V vs. Ag/AgCl, ascribed to a ligand-centred redox process.Other anions, including halides, nitrate and hydrogensulfate caused no sig-nificant change in the electrochemistry [33].

A receptor based on {CoIII(bpy)3} has been used by Sessler and co-workers for redox fluoride sensing [34]. In the cyclic voltammetry of 47 inDMSO solution addition of fluoride led to a complete disappearance of theCo(II)/(III) reduction wave. Addition of water to this solution restored this re-dox process, suggesting that the presence of a strongly bound fluoride anionrenders the complex redox inactive. Chloride and dihydrogenphosphate pro-duced cathodic shifts of 160 mV and 70 mV, respectively. It is proposed that


the pyrrole NH protons of the quinoxaline phenanthroline ligand are mademore acidic by the electron-withdrawing effect of the coordinated metal cen-tre, thereby giving the complex an increased affinity for fluoride compared tothe free ligand.

2.5Transition Metal Dithiocarbamates as Receptors for Redox Anion Sensing

Another interesting development has been the self-assembly of metallo-dithiocarbamate macrocyclic receptors for electrochemical anion sensing.The naphthyl-based Cu(II) macrocycle 48 displays a cathodic shift of 85 mVin the Cu(II)/(III) redox couple in the presence of dihydrogenphosphate orperrhenate, but gives no response to halides in acetonitrile solution [35]. Itis proposed that this selectivity is controlled by cavity size. A related recep-tor incorporating thiourea and hydrogen-bond donor groups, 49, revealed


cathodic shifts in the Cu(II)/(III) couple of up to 160 mV with hydrogenphos-phate [36]. Cobalt (III) dithiocarbamate cryptands 50 and 51 also function asredox active anion sensors [37]. In dichloromethane solution the irreversibleCo(IV)/Co(III) redox couple of the complexes was found to undergo signifi-cant anion-induced cathodic perturbation; up to a maximum of 125 mV for51 with dihydrogenphosphate.

2.6Dendrimers as Redox Anion Sensors

Dendrimers have been investigated as a platform for enhanced anion sensing.Astruc and co-workers have synthesised dendrimers 52–54, incorporatingup to 18 amido-ferrocene units. These multi-metallic multi-binding site re-ceptors are able to electrochemically sense anions in dichloromethane so-lution [38]. In 54 the selectivity trend is dihydrogenphosphate > hydrogen-


sulfate > chloride > nitrate. Evidence of a “dendritic effect” was observedin the redox response of the consecutive dendrimer generations in thepresence of dihydrogenphosphate or hydrogensulfate. As the number ofamido-ferrocene units is increased, the magnitude of the cathodic shift in


the ferrocene/ferrocenium couple also increases. Stability constants for thehydrogensulfate complexes of 53 and 54 were reported to be 9390 and216 900 M–1, respectively.

Kaifer and co-workers have studied an analogous series of dendrimersbased on a commercial DSM polyamine core with 4, 8, 16 and 32 periph-eral ferrocenyl urea groups as the anion-sensing component [39]. No stabilityconstants are reported, but cathodic shifts in the redox response of the fer-rocene/ferrocenium couple with various anions in DMSO show a similarselectivity trend to 52–54. In this case the data suggests two ferrocene ureaarms are involved in binding a single dihydrogenphosphate anion. The den-dritic effect was observed in the change from the first (4 ferrocene units) tosecond (8 ferrocene units) generation dendrimers. No further increase in re-sponse was seen in the third generation and the fourth generation dendrimershowed a decreased response, presumably due to steric crowding.


Astruc and co-workers have also investigated five generations of penta-methyl-amidoferrocene dendrimers using the DSM polyamine core [40]. Thepentamethyl-substituted ferrocene was chosen to overcome the irreversibleelectrochemistry and electrode adsorption observed with 52–54. In this se-ries the dendritic effect seen in the electrochemistry DMF solution variedaccording to the anion studied. In changing from lower to higher den-drimer generations modest increases in the anion-induced cathodic shift ofthe ferrocene/ferrocenium couple were observed with dihydrogenphosphate,whereas with ATP2– anion binding progressed from weak to relatively strong.This is perhaps due to ATP2– adopting a 1 : 2 anion/ferrocene unit bindingstoichiometry, whereas dihydrogenphosphate binds 1 : 1.

2.7Surface Confined Redox Anion-Sensing Systems

In a step towards the fabrication of prototype sensory devices organisationof redox-active anion receptors on to electrode surfaces is being exploited.Importantly, self-assembled monolayers or thin polymer films of metal-basedreceptors can generate an amplified response to anion binding akin to thedendritic effect and could potentially become the basis of robust anion-sensing devices.

Beer and co-workers have investigated this concept using self-assembledmonolayers of the 1,1′-bis(alkyl-N-amido)ferrocene 55 on gold electrodes [41].The pendant disulfide groups serve to covalently anchor the receptorto the gold surface. In electrochemical experiments on 55 in acetoni-trile/dichloromethane solution anion-induced cathodic shifts of the fer-rocene/ferrocenium redox couple were observed for chloride (40 mV),


bromide (20 mV) and dihydrogenphosphate (210 mV). When confined toa monolayer the anion-induced shifts measured were 100 mV for chloride,30 mV for bromide and 300 mV for dihydrogenphosphate in the same solventsystem—consistently greater than for the solution-phase receptor. This rep-resents a significant “surface sensing amplification”. The modified electrodeswere also able to selectively detect dihydrogenphosphate in the presence ofa 100-fold excess of halide. In aqueous solution the selectivity of the systemwas altered, enabling the detection of the poorly hydrated anion perrhenatein the presence of dihydrogenphosphate.

A number of groups have been exploring the anion-sensing propertiesof thin polymer films which incorporate metal-based receptors. Monomer56 consists of a cobaltocenium amide redox signalling group with a poly-merisable pyrrole unit. Thin films of the receptor were prepared by elec-tropolymerisation on a platinum or carbon electrode [42]. In electrochem-ical experiments on 56 in acetonitrile solution significant anion-induced ca-thodic shifts of the cobaltocene/cobaltocenium redox couple were observedfor dihydrogenphosphate (45 mV) and hydrogensulfate (20 mV) only. Whenconfined to a polymer film the cathodic shifts were amplified: 210 mV fordihydrogenphosphate and 250 mV for hydrogensulfate. Chloride and bro-mide could also be detected, both giving shifts of 20 mV. Polymerisationof 56 as well as giving rise to surface sensing amplification also resultedin a change in selectivity from dihydrogenphosphate to hydrogensulfate. Itwas also found that film thickness influences the sensitivity of the sensor.Thin films (Γ = 1.8×10–9 mol cm–2) exhibited higher sensitivity to dihydro-genphosphate at low concentrations (< 50 µM), whereas thick films (Γ =2.7×10–8 mol cm–2) extend the measurable concentration range to higherlevels (up to 2 mM).

Films of the analogous ferrocene monomer 57 have been studied byMoutet and co-workers [43]. Anion-induced shifts of the ferrocene/ferro-cenium couple were measured in acetonitrile for hydrogensulfate (30 mV),ATP2– (180 mV) and dihydrogenphosphate (220 mV). Again this representsa surface sensing amplification. The same group has also explored theanion-sensing properties of viologen 58 in thin polymer films [44]. In aque-ous solution poly-58 registered small anion-induced cathodic shifts of theferrocene/ferrocenium redox couple with hydrogensulfate (20 mV), S2O4

2–

(10 mV), and ATP2– (35 mV).Metal-based receptors that are able to form self-assembled monolayers on

planar electrodes can also be used to functionalise the surface of nanoparti-cles, leading to a surface sensing amplification effect. The very large surfacearea of nanoparticles may also allow greater overall sensitivity to anions.

Astruc and co-workers have prepared the amidoferrocenylalkylthiol(AFAT)-gold nanoparticle system 59 [45]. The proportion of AFAT to do-decanethiol obtained by ligand substitution on different batches of dode-canethiol stabilised nanoparticles ranged from 7–38%, corresponding to an


average of 8–39 AFAT units per nanoparticle. Electrochemical measurementsin dichloromethane solution show a single reversible redox wave for the fer-rocene/ferrocenium couple at identical potential in each case. Addition ofdihydrogenphosphate led to the appearance of a new redox wave (220 mVcathodically shifted) with the attenuation of the initial wave, which was com-pletely replaced at 1 equivalent of anion per AFAT branch, indicating of 1 : 1anion/branch binding. The cathodic shift is the same irrespective of the AFATloading and is considerably larger than observed for the comparable amido-ferrocene monomer FcCONHCH2CH2OPh (45 mV) or even a representativeferrocene tripod PhC(CH2CH2CH2NHCOFc)3 (110 mV).

The same group has also investigated the anion-sensing properties of goldnanoparticles 60 and 61 substituted with dendrons comprising three amid-oferrocene or silyl ferrocene branches [40]. The surface loadings of 60 and61 were 3% and 4.8% respectively, corresponding to an average 3 and 5dendrons per nanoparticle. Nanoparticles of type 60 show very similar prop-erties to the AFAT-modified nanoparticles 59, with a dihydrogenphosphate-induced cathodic shift of 210 mV in dichloromethane solution. Despite lack-ing any hydrogen-bonding groups the nanoparticles dendronised with silyl


ferrocenes, 61, gave a dihydrogenphosphate-induced cathodic shift of 110 mVin the same solvent.

A highly ambitious multicomponent surface-anchored anion-sensing ro-taxane assembly has recently been described [46]. This comprises two re-ceptor molecules, 62—an isophthalamide macrocycle with an exocyclic fer-rocene reporter group, and 63 —a cationic pyridinium amide thread bearinga disulfide tether for SAM formation at one terminus, and a pentaphenylfer-rocene at the other as a combined redox reporter and bulky stopper group.In low polarity solvents such as dichloromethane chloride is bound simul-taneously by both receptors, causing the threading of 63 into the annulusof 62 to form a pseudorotaxane (Fig. 1). Adsorption of the pseudorotaxaneonto a clean gold surface to form a SAM causes the components to be locked


together as a rotaxane. This rotaxane SAM shows a remarkable selective elec-trochemical response to anions compared to 62 or 63 alone. The addition ofmolar excesses of chloride, dihydrogenphosphate or hydrogensulfate to 63 inacetonitrile solution or as simple SAMs resulted in only a small cathodic shift(∆E < 10 mV) of the pentaphenylferrocene redox couple. The macrocycle 62was more responsive to anions, undergoing a cathodic shift in the ferroceneredox couple of 45 mV with dihydrogenphosphate, 15 mV with hydrogen-sulfate and < 10 mV with chloride. In the rotaxane SAM the pentaphenyl-ferrocene centre of 63 exhibits redox responses to chloride and oxoanionsbroadly similar to SAMs of this receptor on its own. In contrast, withinthe surface assembled rotaxane the ferrocene of the macrocyclic compon-ent exhibits a markedly greater electrochemical cathodic response to chloride

Fig. 1 Self-assembly of anion templated rotaxane SAM (RC = redox centre). From [46],reproduced by permission of The Royal Society of Chemistry


(40±5 mV), but gives no significant response to the oxoanions tested. Thissuggests that chloride binding inside the interlocked cavity of the surfaceconfined rotaxane results in a conformation where the macrocycle’s pendantferrocene group is in proximity to the complexed halide anion, whereas theoxoanions are too large to penetrate the rotaxane binding pocket. Preliminaryelectrochemical competition experiments in acetonitrile solutions revealedthat these rotaxane SAMs are capable of selectively detecting chloride in thepresence of 100-fold excess amounts of dihydrogenphosphate and exhibit anappreciably greater detection sensitivity than that shown by the free macro-cycle. The superior electrochemical response of rotaxane SAMs to chlorideover dihydrogenphosphate mirrors the high degree of chloride anion selectiv-ity of previous rotaxanes prepared via chloride anion templation.

3Optical Anion Sensors

3.1Theory of Optical Anion Sensing

Optical reporter groups signal anion binding through a change in their elec-tronic absorption or emission spectra. The precise nature of the response inthe UV/vis absorption spectrum will largely depend on the energy differencesbetween the molecular orbitals of the receptor before and after anion bind-ing. For changes of any magnitude to be observed the anion-binding site mustbe strongly coupled to the metal centre. In the case of metal-based transitionssuch as d-d transitions this typically requires the anion to bind directly to themetal centre. In the case of MLCT or LMCT transitions it is advantageous ifthe anion-binding site is π-conjugated to the ligand involved. Luminescencespectroscopy can be a more sensitive technique for probing anion bindingto metal-based receptors. In addition to altering the energy of the emissionmaxima, anion binding often causes significant changes in their intensity.The intrinsic luminescence of a particular reporter group can be “switchedoff” if the anion : receptor complex provides a more efficient pathway fornon-radiative energy loss. Similarly, in systems where the luminophore isquenched by a nearby functional group, anion binding can “switch on” theluminescent emission by blocking this non-radiative decay process.

3.2Metallocene Optical Anion Sensors

The common reporter groups cobaltocenium and ferrocene have not fre-quently been used in optical anion sensing, since these chromophores aregenerally insensitive to anion binding. However, metallocene-based receptors


that incorporate a suitable organic chromophore or luminophore have beenshown to operate as combined optical and electrochemical anion sensors.For instance the tetra-cobaltocenium porphyrin 17, exhibits the same selec-tivity trend (Cl– > Br– � NO3

–) in UV-vis anion-binding experiments thatwas observed by electrochemistry [19]. In acetonitrile solution the Soret band(λmax = 425 nm, due to the porphyrin) of 17 was significantly bathochromi-


cally shifted on addition of dihydrogenphosphate (∆λmax = 15 nm), hyp-sochromically shifted with C1– (∆λmax = 10 nm) and split into two maxima(λ = 430,440 nm) with HSO4

–.The novel series of ferrocene receptors 64–68, which incorporate thiobar-

biturate anion binding/chromophore groups, have been shown to operate asUV-vis anion sensors in acetonitrile solution. Addition of basic anions suchas cyanide, acetate and benzoate causes the attenuation of the absorptionmaximum at around 440 nm (due to a charge transfer transition between theamine and the thiobarbiturate group), with the simultaneous occurrence ofa new band at 370 nm. In cases where the titration data could be fitted, 1 : 2receptor : anion binding stoichiometries were found [47].

Another interesting example of a ferrocene-based optical sensor is 69,which acts as a chromogenic molecular switch [48]. Appended to one cy-clopentadienyl ring of the ferrocene of molecule 69 is a p-nitrophenyl ureaunit which acts as a combined anion-binding site and chromophore. A crownether is attached to the other cyclopentadienyl ring for cation binding. Tuckerand co-workers reported that on addition of fluoride to a solution of 69 inacetonitrile a significant perturbation of the UV-vis spectrum was observedincluding the appearance of a new absorption at 472 nm. The Ka for the1 : 1 anion-receptor complex was determined as 9340 M–1. The addition of10 equivalents of KPF6 to the solution of 69 containing 10 equivalents of flu-oride caused the complete disappearance of the 475 nm absorption. Ka ofthe receptor with K+ in the presence of fluoride was calculated as 1460 M–1.Surprisingly model receptor 70, which lacks the crown ether moiety, exhib-ited similar switching properties. Ka of this receptor for fluoride is equivalent(9660 M–1) but the Ka for K+ in the presence of fluoride is far lower (230 M–1).Inhibition by K+ of the response of these receptors to fluoride is thereforethought to be due to the ion-pairing interaction between fluoride and K+.

Aldridge et al. have demonstrated that boryl-ferrocene 71 can be used asa selective colourimetric sensor for fluoride [49]. When fluoride was addedto a CH2Cl2 solution of 71 under aerobic conditions a colour change fromorange to pale green was observed. This did not occur with any other aniontested. Spectroscopic and electrochemical measurements suggest that com-


plexation of fluoride causes the spontaneous formation of a ferroceniumspecies, i.e. the 150 mV anodic shift in the ferrocene/ferrocenium redox po-tential caused by fluoride complexation reduces the redox potential enoughfor the 71 : 2F– complex to be oxidised by atmospheric O2. The pale greencolour is due to the characteristic absorption of ferrocenium. In this case theoptical response is the direct result of an anion-induced redox process anddoes not require an additional chromogenic group.

Luminescence sensing of anions has also been achieved using ferrocene re-ceptors. Example 72 uses amide groups for anion binding in conjunction withnaphthalene groups to provide the fluorescence signal [50]. Addition of fluo-ride to a DMSO solution of 72 led to a 3-fold enhancement (at 5 equivalents)of the intramolecular naphthalene-naphthalene excimer emission at 492 nm.Dihydrogenphosphate also generated a significant response, causing a 2-foldenhancement at 5 equivalents. In electrochemical studies in DMF electrolytesolution fluoride generated a 120 mV cathodic shift in the redox potential.

The receptor 73, based on an azaferrocenophane structure bearing twourea groups as linkers between the redox active (ferrocene) and fluorescent(naphthalene) signalling subunits, also shows both fluorescent and electro-chemical sensing of fluoride [51]. On addition of excess fluoride it displaysan enhancement factor of 13 in the naphthalene emission bands at 362 and380 nm in DMF solution and a cathodic shift of the ferrocene/ferroceniumcouple of 190 mV in DMSO electrolyte solution.


The unusual new [3,3]ferrocenophane 74·H+ also acts as a selective fluor-escent anion sensor—in this case for nitrate [52]. The protonated receptor isweakly fluorescent (Φ = 0.043) in CH2Cl2 solution and on addition of nitratethe naphthalene-based emission at 354 nm is quenched (to Φ = 0.020). Add-ition of acetate, hydrogensulfate and dihydrogenphosphate merely induceddeprotonation of the receptor. This receptor is also able to act as a redoxsensor for other anions and as a fluorescent sensor for group II cations.

3.3Ruthenium(II) Polypyridyl Complexes as Optical Anion Sensors

The spectroscopic and redox properties of {RuII(bpy)3} have allowed thismetal complex to be used for combined optical and electrochemical sensingof anions without the need for additional chromophores or luminophores.

For example Sessler’s complex 75 gives a UV-vis response to fluoride inDMSO solution, with a stability constant for the receptor : fluoride complexof 12 000 M–1 [34]. The new {RuII(bpy)3}-pyrrole 76 has also been found toselectively sense fluoride in DMSO solution by UV/vis (Ka = 7000 M–1) [53].


The luminescent emission of {RuII(bpy)3} is also very useful for signallinganion binding. In the emission spectra of 37–40 both a blue shift (of up to16 nm for 40) and an increase in intensity of the λmax of the MLCT emis-sion band was observed on addition of dihydrogenphosphate. It has beenproposed that the conformational flexibility of the receptors is decreased bycomplexation of the anion guest thus reducing the rate of non-radiative decaythrough vibrational and rotational relaxation. Similarly, macrocyclic com-plexes 41–44 and 77–79 were also found to sense chloride by luminescenceenhancement.

In other examples anion binding can cause quenching of the {RuII(bpy)3}MLCT emission. In aqueous solution polyaza receptors 80-82 bind phosphateand ATP anions, producing up to a 15% reduction in the emission intensity ofλmax at 605 nm [16]. Similarly, 76 shows up to a 40% reduction in the intensityof the luminescent emission at 630 nm in the presence of dihydrogenphos-phate in DMSO solution.

The RuIIbipyridylcalix[4]diquinone receptor 83 selectively binds andsenses acetate anions (from 1H NMR titrations in DMSO-d6 solutionK = 9990 M–1) [54]. This receptor is only weakly luminescent because the{RuII(bpy)3} MLCT emission is partially quenched by oxidative electrontransfer to the electron-poor calix[4]diquinone. Addition of acetate to ace-tonitrile solutions of 83 resulted in a five-fold increase in luminescenceintensity (60% for chloride) concomitant with a slight blue shift of the emis-sion maximum. Anion binding causes this increase in emission intensityby interrupting the electron transfer pathway from the {RuII(bpy)3} to thecalix[4]diquinone, thus reducing its quenching effect.

A similar effect is seen in macrocycle 36 which incorporates the{RuII(bpy)3} moiety with a bridging cobaltocenium unit [29]. In acetoni-


trile solution the quantum yield of the {RuII(bpy)3} emission of this complexis relatively low. However, in the presence of chloride a 100% increase inemission intensity is observed.

Complexes 45 and 46 are capable of selectively sensing fluoride in ace-tonitrile solution both by UV-vis and luminescence spectroscopy [33]. Thisanion causes a dramatic reduction in the intensity of the MLCT absorption inthe 350–450 nm range with a new absorption appearing in the 500–650 nmrange ascribed to a ligand-based CT process. The emission spectra of thecomplexes (exciting at 465 nm) show no significant peaks, indicating that thecharacteristic Ru-centred luminescence is quenched by the dinitrophenylhy-drazone group. Upon addition of fluoride a strong peak at 625 nm develops,with quantum yields for the 45 : F– and 46 : F– adducts of 8.0×10–5 and4.0×10–4, respectively. Again it is apparent that anion binding interrupts thenon-radiative decay pathway.

Deetz and Smith have prepared a heteroditopic {RuII(bpy)3} receptor 84incorporating both amide and boronic acid groups which selectively sensescertain phosphorylated sugars in aqueous solution [55]. Boronic acids areknown to form covalent complexes with the diol groups of saccharides,whereas the adjacent amides are positioned to complex the anionic phosphatecomponent. Sensing was accomplished by measuring luminescence enhance-ment, with fructose-6-phosphate generating the highest stability constant(log Ka = 3.1). Non-phosphorylated saccharides gave much smaller changesin emission intensity (log Ka < 1.2), showing that the anionic component ofthe guest is essential for strong binding. A covalent attachment between theanion and the saccharide is not required. In the presence of sodium phos-phate buffer non-phosphorylated saccharides are bound with similar strengthto their phosphorylated counterparts. It is reported that this apparent coop-erativity is a result of favourable hydrogen bonding between the phosphateanion and the saccharide. Watanabe and co-workers have also shown thatanionic and neutral phosphodiesters can be sensed in acetone by the imi-dazole functionalised receptor 85 by both UV-vis and luminescence spectro-scopies [56].


3.4Optical Anion Sensing Using Reporter GroupsBased on Other Transition-Metal Complexes

Zn(II) porphyrins are another class of complex which can operate both asUV/vis and luminescent sensors for anions. Example 86 is a picket-fence por-phyrin with four imidazolium anion-binding groups [57]. In DMSO solutionthis receptor undergoes slight (5 nm) bathochromic shifts in the Q-bands ofthe absorption spectrum on addition of anions—indicative of the anion bind-ing directly to the Zn centre. Using UV/vis titrations in DMSO solution 86 wasfound to be selective for hydrogensulfate over chloride and dihydrogenphos-phate. In water : DMSO (5 : 95) solution the receptor was found to be selectivefor sulfate and hydrogensulfate over ATP2– and dihydrogenphosphate. Undersimilar conditions the Q′ bands in the luminescence emission spectrum of 86(λex = 424 nm) are also bathochromically shifted with a concomitant decreasein intensity.

It should be noted that 86 also functions as an electrochemical anionsensor, where the Zn(II) porphyrin-based oxidation potential is sensitive to


anion binding. In acetonitrile solution a greater cathodic shift was observedwith chloride (175 mV) than hydrogensulfate (140 mV) or nitrate (95 mV).This demonstrates that the anion selectivity of sensor systems is dependentupon the mode of detection used and underlines the fact that optical andelectrochemical anion sensing operate by different mechanisms.

Bis-terpyridine Iridium(III) has been used as an optical reporter group inanion sensing. In aqueous solution isomers 87 and 88 both exhibit halide-induced luminescence quenching with selectivity for chloride [58]. Receptor87, although less sensitive than 88, is reported to possess good characteristicsfor sensing chloride at physiologically relevant concentrations.

A related series of novel cyclometallated iridium(III) polypyridine thioureacomplexes also display anion-induced luminescence quenching. The repre-sentative complex 89 was tested in acetonitrile solution with fluoride, acetateand dihydrogenphosphate, and gave log K values for the 1 : 1 complex of 3.38,


4.03 and 3.14, respectively. This selectivity trend is ascribed to the combinedeffect of the basicity and geometry of the guest anions [59].

Complexes of rhenium(I)tricarbonylchloride with pyridyl ligands are lu-minophores and hence in anion sensing have principally been used as re-porter groups. For instance calix[4]diquinone receptor 90 selectively bindsand senses acetate in DMSO solution [54] (from 1H NMR titrations K =1790 M–1 in DMSO-d6 solution). The receptor exhibits relatively weak lumi-nescence because calix[4]diquinone is an electron acceptor, quenching theRe(I) bipyridyl emission by oxidative electron transfer. Addition of anions toDMSO solutions of 90 resulted in a significant increase in luminescence inten-sity. It is clear that the presence of the anion in the binding pocket betweenthe {ReI(bpy)} moiety and the quencher interrupts the oxidative electron-transfer process.

The mixed Re(I)/Pd(II) molecular square 91 has been found to senseperchlorate in acetone (giving K = 900 M–1) by enhancement of the Re(I) lu-minescent emission [60]. In this case luminescence quenching by oxidativetransfer to the Pd(II) ion is inhibited by the bound anion. The Pd(II) ionalso plays a role as a structural element and charge carrier. Squares 92–96are very similar, but incorporate a bis-phosphinylferrocene supporting lig-and [61]. Again the normally strong luminescence of the Re(I) component ispartially quenched by the bimetallic corners. Binding studies of the squareswith different inorganic anions were carried out by luminescence titrationsin acetone solution. Of the anions investigated, only hexafluorophosphate andtetrafluoroborate induced significant changes in luminescence. As these an-ions were added an initial decrease in emission intensity was followed byan increase to a plateau. This is taken to indicate the presence of two com-peting quenching pathways which are inhibited to different extents by anionbinding.

Lees and co-workers have investigated Re(I) bipyridyl anion hosts basedon aryl bisamide skeletons 97–99 [62]. Measurement of anion-induced lumi-nescence quenching in CH2Cl2 showed 97 to have strong binding affinities forhalides, acetate and cyanide, weaker affinity for dihydrogenphosphate, andeven less affinity for nitrate and perchlorate. The iso- and terephthalamide re-ceptors 98 and 99 possess smaller stability constants for all the anions tested.It is proposed that the anion-sensing efficiency of 98 is due to intramolecularhydrogen-bonding of the amido NH proton to the pyridyl nitrogen holdingthe receptor in a “cleft” conformation.

A metal-templated approach has been used by Thomas and co-workers toproduce the Re(I) metallomacrocycle 100 [63]. In acetonitrile solution thisreceptor displayed luminescence enhancement on addition of anions. Stabil-ity constants for the 1 : 1 adducts with BF4

– (1575 M–1), SO42– (7135 M–1)

and BPh4– (2895 M–1) were determined. Since SO4

2– is of similar size toBF4

– the comparatively high affinity of this anion for 100 is thought to bedue to its additional charge. The slightly higher stability of the 100 : BPh4

–


complex compared to 100 : BF4– is attributed to π–π interactions with the

receptor.Interlocked supramolecular assemblies such as rotaxanes and catenanes

have the potential to provide tailor-made binding cavities for guest species.An example of this is 101, a rotaxane formed using anion templated synthe-sis which incorporates the Re(I) bipyridyl fragment [64]. Addition of chlo-ride, hydrogensulfate or nitrate to the receptor in acetone solution causedan enhancement of the fluorescence emission. Curiously, although the ro-taxane was formed using chloride as the template, it was found to be se-lective for hydrogensulfate with which a stability constant of > 106 M–1 wasdetermined.


3.5Transition Metals as Anion-Binding Groups and/or Structural Components

Complexes of the late transition metals are well studied in anion sensing,much of the work has been pioneered by Fabbrizzi and co-workers. As wellas providing an optical signalling function the Lewis acid metal ion can actas a binding site for the anion. In addition the metal ion often forms an or-ganisational unit designed to create a receptor of a specific shape. In order toharness the metal-anion interaction for anion sensing the binding propertiesof the metal must be modulated by an ancillary ligand. In this way one or twovacant coordination sites can be made available for anion binding, and otherelements appended to allow signalling or modified selectivity.

In complexes with simple tripodal amines such as 102 and 103 zinc(II)forms five-coordinate metal complexes of trigonal bipyramidal geometry,leaving one of the axial coordination sites of the metal available for an-ion binding. The Zn(II) complex of 102 was found to undergo quench-ing (by photoinduced electron transfer) of the anthracene fluorescentemission in the presence of aromatic carboxylate anions such as 4-N,N-dimethylaminebenzoate in ethanol solution [65]. Complete quenching wasobserved at the 1 : 1 anion to receptor ratio (log K = 5.45). Likewise the Zn(II)complex of 103 was found to form 1 : 1 adducts with carboxylate anions inmethanol solution, with log K values ranging from 4 to 5 [66]. Only aromaticcarboxylates induced quenching of the ligand luminescence.


The bis(boradiazaindacene) substituted bipyridine ligand 104 is highlyfluorescent in organic solvents whereas its Zn(II) complex is not [67]. Itwas found that the complex progressively regained its fluorescent emissionwhen it was titrated with various anions in acetonitrile. Stability constants


for the anion:receptor complexes were calculated for fluoride (4160 M–1),chloride (3230 M–1), bromide (2500 M–1), acetate (4760 M–1), and phosphate(4000 M–1). Decomplexation of the chelated Zn(II) ion from 104 by the weaklycoordinating anions was ruled out as the sensing mechanism. It is proposedthat the quenching of the ligand luminescence by electron transfer to theZn(II) centre is inhibited by anion coordination.

Fabbrizzi and co-workers have demonstrated the use of bis-copper(II)cryptates to sense ambidentate anions [68]. On titrating molecule 105 withNaN3 in aqueous solution the colour changed from pale blue to bright greenand an anion-metal LMCT absorption appeared at 400 nm. X-ray diffractionstudies have shown that the azide anion is held colinear with the two Cu(II)centres, coordinated through the two terminal sp2 hybridised nitrogen atoms.Stability constants for 105 with a variety of anions in aqueous solution werecalculated and the selectivity of this anion sensor for the azide anion wasfound to be determined by the bite distance between the two copper atoms.

Cryptate 106, in which the aryl spacer of 107 is replaced with a furanylunit, acts as a colourimetric sensor for anions. UV/vis titrations in aqueoussolution gave log K values for the 1 : 1 halide/receptor adducts of 3.98 for chlo-ride, 3.01 for bromide and 2.39 for iodide. X-ray diffraction studies confirmthat bromide is held between the two copper atoms. Under the same con-ditions 106 also interacts strongly with azide (log K = 4.7) and thiocyanate(log K = 4.28) anions. This receptor is interesting because of its lack of se-lectivity compared to 105. The complex appears to be able to expand andcontract its bite-length in order to accommodate anions of various dimen-sions.

The use of this class of receptors in practical applications is limited bythe small changes in UV-vis absorption which indicate anion binding. Toovercome this problem of sensitivity a chemosensing ensemble approach hasbeen applied. The fluorescent indicator coumarine 343 carries a carboxylategroup which allows it to be bound by 105 in a 1 : 1 complex (log K = 4.8) withcomplete quenching of the luminescent emission [69]. Titration of a solutioncontaining 0.2 mM 105 and 0.1 µM coumarine 343 with hydrogencarbonate,azide or cyanate anions resulted in complete recovery of the indicator lu-minescence. Anions with a lower affinity for the receptor were unable todisplace the coumarine 343 and produced only a slight luminescence en-hancement. The usefulness of this chemosensing ensemble was demonstratedby the quantitative determination of carbonate in mineral water. Using thesame principle Han and Kim have recently reported a chemosensing ensem-ble made up of the dizinc complex 107 and pyrocatechol violet which isselective for phosphate [70].

Anslyn and co-workers have developed a series of tripodal Cu(II) com-plexes 108 and 109 in which the metal ion and three cationic organic groupsform a tetrahedral cavity designed to host phosphate [71]. Receptor 108 isbuilt from the tris(2-ethylamino)amine skeleton with appended benzylamine


groups. UV/vis anion titrations were carried out in aqueous solution at pH 7.4where it can be assumed the terminal amines are all protonated. Hydrogen-phosphate and its congener hydrogenarsenate were found to bind stronglyin a 1 : 1 anion/host ratio, both with a log K value of 4.40. Perrhenate wasbound an order of magnitude less strongly and the affinity for chloride wastoo small to measure. Model complex 110, which has no ammonium groups,gave a log K value of 2.95 with hydrogenphosphate, indicating that the Cu(II)-anion interaction contributes significantly to anion binding. Receptor 109 fol-lows the same design principle, this time incorporating guanidinium bindingunits with a tris-[(2-pyridyl)methyl]amine skeleton. log K values for hydro-genphosphate and hydrogenarsenate were found to be 4.18 and 4.23, respec-tively. Other anions, including perrhenate had no significant affinity for 109.It is apparent that the guanidinium groups are responsible for the improvedselectivity of this receptor for phosphate. In a separate study the driving forcefor hydrogenphosphate binding was found to be entropic for receptor 108, butenthalpic for receptor 109 [72]. Partnered with the colourimetric indicator5-(and 6)-carboxyfluorescein receptor 109 provides an effective chemosens-ing ensemble for the determination of inorganic phosphate in serum andsaliva [73].

Zinc has been used as a binding site for the detection of pyrophosphatein aqueous solution by fluorescence. Complex 111 couples two tridentate Zncentres to a fluorescent naphthalenediimide core. In HEPES buffer the ap-pearance of a new emission band at 490 nm was observed on addition ofpyrophosphate, attributed to naphthalenediimide excimer formation. Thisdid not occur with other anions including halides, acetate and ADP. It is pro-posed that 111 binds pyrophosphate in a 2 + 2 complex, which brings twonaphthalenediimides in close enough proximity to give the excimer emis-sion [74].

The recent tripodal Cu(II) complex 112 has intriguing optical anion-sensing properties [75]. This receptor has a cavity with two distinct anion-binding sites—the vacant site on the Lewis acidic Cu(II) centre, and thethree favourably arranged nitrophenylurea fragments. On titration with up


to one equivalent of azide or dihydrogen phosphate in DMSO solution theCu(II) d–d bands in the region 600–900 nm increased markedly in intensity.This indicates that the anion is binding at the metal centre. Upon additionof a second equivalent of the anion the absorption associated with the nitro-phenylurea groups (below 500 nm) increased in intensity, showing that these


hydrogen-bonding units are involved in binding the second anion. On theirown halide anions were found to give only 1 : 1 complexes (with the halidebound at the Cu(II) centre). However, on titration of dihydrogenphosphateinto a solution of the receptor pre-saturated with chloride (i.e. dissolved ina solution containing a 150-fold excess of chloride) formation of a 1 : 1 adductof the [112 : Cl] complex with dihydrogenphosphate was observed. The stabil-ity constant of this species was found to be approximately 700 times higherthan the stability constant of the analogous [112 : H2PO4] complex with a sec-ond equivalent of dihydrogenphosphate.

Stepwise anion coordination equilibria are also observed in the Cu(II)complexes of ligands 113 and 114 [76]. UV/vis titrations in acetonitrile so-lution show that each Cu(II) complex binds two anions (chloride, bromide,iodide, nitrate or thiocyanate), the first at the Cu(II) centre and the secondin the bis-imidazolium compartment. The Cu(I) complexes of these ligandsare able to host only one nitrate anion (in the bis-imidazolium cavity), whileother anions induce demetallation. Cyclic voltammetry and spectroelectro-chemical experiments showed that in the presence of one equivalent of nitratethe Cu(II)/Cu(I) redox change causes the anion to translocate quickly andreversibly from the metal-based binding site in the Cu(II) complex to the im-idazolium binding site in the Cu(I) system.

Another noteworthy example in which Cu(I) forms the basis of an opti-cal anion sensor is 115, in which the metal complex acts both as a UV/vissignalling group and as a structural component dictating the topology of theurea anion-binding site [77]. The MLCT band within the CuI(phenanthroline)complex at 282 nm is sensitive to halide ions, acetate and dihydrogenphos-phate in 4 : 1 v/v THF/MeCN (a relatively low polarity solvent). However, inDMSO solution, only acetate and dihydrogenphosphate produced a UV/vis re-


sponse. Crucially, dihydrogenphosphate was found to bind the receptor with1 : 1 stoichiometry, whereas acetate bound 2 : 1 anion : receptor. It is pro-posed that dihydrogenphosphate is able to bind simultaneously to the twourea groups to form an overall helical structure. This is not the case withthe free ligand—the CuI centre is required to template the formation of thebinding cavity, and augments dihydrogenphosphate binding. Acetate can onlyhydrogen-bond to one urea subunit at a time, and the affinity of 115 for thisanion is only slightly higher than that of the free ligand (presumably due toelectrostatic effects).

Cadmium(II), the heavier congener of zinc(II) can also act as a coor-dination site for anion binding. Mizukami et al. employ a novel approach

Fig. 2 X-ray crystal structure of bromide encapsulated in the Fe : 117 complex. Repro-duced with permission from [79]. © Wiley-VCH Verlag GmbH & Co. KGaA


with receptor 116, by covalently attaching the coumarin indicator group ofa chemosensing ensemble to a macrocyclic Cd(II) anion receptor [78]. Itis designed such that anions will displace the nitrogen of the aromatic flu-orophore from the coordination sphere of the Cd(II) centre. In aqueoussolution as the receptor was titrated with pyrophosphate the luminescentemission of the molecule was observed to shift gradually from 342 nm to383 nm. Receptor 116 shows a high degree of selectivity for pyrophosphate,citrate, ATP and ADP with log K’s in the range 4–5. It is worthy of note thatthe Zn(II) analogue of 116 was found to be ineffective for anion sensing.

Iron(II) has been used as a supramolecular template for the formationof a tris-imidazolium receptor from ligand 117 [79]. 1H NMR studies andX-ray crystal structure determination were used to demonstrate the encap-sulation of bromide in the cavity of the receptor, with the anion coordi-nated by six C–H fragments (Fig. 2). Spectrophotometric titrations in ace-tonitrile solution revealed that this receptor binds halides with selectivityfor chloride > bromide > iodide, but has no affinity for dihydrogenphosphateor hydrogensulfate. Presumably the restricted size of the receptor cavity ex-cludes the binding of these larger tetrahedral anions. The linear anions azide,cyanate and thiocyanate also produced a response in the UV/vis spectrum,and azide was found to bind preferentially to 117 in comparison to the non-symmetrical linear anions.

3.6Optical Anions Sensing by Lanthanide(III) Complexes

Given their favourable luminescence properties and propensity to coordinateoxy-anions it is perhaps surprising that complexes of lanthanide(III) ionshave received comparatively little attention as anion sensors.

Parker and co-workers have reported a novel method for the selective de-tection of carboxy anions by time-delayed luminescence using the Eu(III)and Tb(III) complexes of 118 [80]. Luminescence measurements on the co-ordinatively unsaturated complexes in aqueous solution showed significantincreases in lifetime and emission intensity in the presence of anions. This be-haviour is consistent with the anions displacing water from the non-ligatedcoordination sites at the metal centre. Studies allowed the number of watermolecules (q) remaining coordinated in the presence of added anions to beestimated. For the triflate salt of the Eu(III) receptor q = 2.14 whereas withhydrogencarbonate q = 0.34 and with hydrogenphosphate q = 0.74. The selec-tivity for hydrogencarbonate reflects the ability of this anion to chelate theEu(III) centre whereas hydrogenphosphate prefers to bind in a monodentatefashion. This behaviour is also pH-dependent (pKa HCO3

–/H2CO3– = 6.38) so

that for a given starting bicarbonate concentration, a pH-dependent lifetimeand emission intensity was observed [81].


Other groups have subsequently reported anion receptors that work on thesame principle. For instance an Eu(III) complex of the bis-bipyridinephenyl-phosphine oxide ligand 119 made by Ziessel and co-workers is able to senseanions by luminescence enhancement in acetonitrile with stability constantswhich follow the trend fluoride > acetate > chloride > nitrate [82]. Tsukubeand co-workers have investigated the properties of the Eu(III) and Tb(III)complexes of the chiral ligand 120 [83]. Anion binding was assessed by profil-ing luminescence enhancement in acetonitrile and it was found that the dif-ferent metal centres provided different selectivities. The emission at 548 nm


of the Tb(III) complex was increased by 5.5 times in the presence of 3 equiva-lents of chloride compared to 2.2 for nitrate and 1.1 for acetate. Conversely theemission at 618 nm of the Eu(III) complex was increased 8.3 times by 3 equiv-alents of nitrate, 2.5 times for chloride and 1.0 times for acetate. Stabilityconstants were not reported.

The same group most recently reported the use of neutral lanthanide(III)tris-diketonates of type 121 for the determination of chloride [84]. Theresponse in luminescence of the Eu(III) complex for chloride in acetoni-trile solution was large enough to be seen by the naked eye. Incorpora-tion of the complexes in PVC membrane electrodes allowed measurementof potentiometric selectivity coefficients. These showed the Eu(III) complexto be the more selective for chloride than the Pr(III), Dy(III) or Yb(III)analogues.

Gunnlaugsson and co-workers have developed an anion sensor based ona ternary europium complex [85]. In 122 the naphthalene β-diketonato ligandacts as a sensitising group for Eu(III) emission. Displacement of this antennagroup from 122 by competitor anionic species would therefore be expectedto decrease the intensity of this emission. In aqueous solution at pH 7.4 itwas found that iodide and dihydrogenphosphate reduced the intensity of theemission at 616 nm by 20–40%. More pronounced changes, which could beseen with the naked eye, were observed with highly competitive anions suchas tartarate and fluoride.

Gunnlaugsson and co-workers have also studied di-europium(III) com-plex 123, which incorporates two different macrocyclic ligands for Eu(III) inaddition to a covalently bound antenna group [86]. Upon titrating 123 withacetate, aspartate and succinate at pH 6.5, each of the Eu(III) emission bandswas quenched by up to 50% for acetate and aspartate. Malonate, however, pro-duced a nearly two-fold enhancement in the emission intensity. It is thoughtthat this particular anion is able to displace coordinated water molecules fromboth the Eu(III) centres thereby reducing the rate of non-radiative energytransfer.

3.7Surface Confined Systems for Optical Anion Sensing

Beer and co-workers have developed a surface-enhanced optical anion sensorbased on gold nanoparticles [87]. Dodecanethiol stabilised gold nanoparti-cles were modified by ligand substitution with a disulfide-substituted zincporphyrin 124 to provide 30 and 80 receptors per nanoparticle. Titration ofboth the free receptor and the modified nanoparticles with various anionsin dichloromethane or DMSO solution revealed significant changes in the in-tense porphyrin absorption bands. Calculated stability constants are given inTable 1 and reveal highly enhanced anion-binding affinities (up to two ordersof magnitude with chloride and dihydrogenphosphate in DMSO solution) for


Table 1 Association constant (log K) data of 124 and 124 modified nanoparticles in DMSOsolution determined at 293 K, errors ±0.1

Anion 124 124-nanoparticles a

Cl– < 2 4.3H2PO4

– 2.5 4.1

a Association constant values for the 1 : 1 porphyrin–anion complex on the nanoparticlesurface

the surface-bound porphyrin receptor with respect to the free metallopor-phyrin.

4Metal-Based Anion Receptors Without Reporter Groups

Described below are a number of interesting examples from the literature ofmetal-based anion receptors for which anion binding has only been studiedby NMR.

If one of the cyclopentadienyl rings of ferrocene is replaced with an arenemoiety a cationic complex is generated, thereby gaining the potential for elec-trostatic interaction with anions. Beer and co-workers have exploited thisprinciple in the simple receptor 125 [88]. Similarly, Atwood and co-workershave derivatised cyclotriveratrylene to generate receptor 126 [89]. Qualitative1H NMR studies demonstrate that 125 binds chloride and bromide in CH3CN,and 126 binds halides in acetone. It is proposed that the strength of the in-teraction in 126, which does not possess amides or other hydrogen-bonding


groups, is due to the arrangement of positive charges around the upper-rimof the hydrophobic cavity.

The anion-binding properties of a series of highly cationic metal-lated calix[4]arenes have also been investigated. Host 127, in which thecalix[4]arene is coordinated to four Ru(η6-p-cymene) units was found by1H NMR titration to bind chloride, bromide, iodide and nitrate in aqueoussolution [90]. Stability constants were determined with Ka = 551, 133, 51, and49 M–1, respectively.

Cr(CO)3 is an effective electron-withdrawing group [91]. The coordinationof this organometallic unit to the arene substituents of the isophthalamide in128 increases the acidity of the NH protons and hence their affinity for an-ions. 1H NMR titrations in CD3CN solution provided stability constants of thesame order of magnitude as those for the non-metallated receptor in the farless competitive solvent CD2Cl2.

Loeb, Gale and co-workers have used Pt(II) as a structural template for theself-assembly of a series of anion receptors. Receptors 129–132 were foundto bind a variety of anions in DMSO-d6 solution [92, 93]. Even the simpletetrapyridine receptor 129 has an affinity for anions due to its electrostaticcharge. The addition of pyrrole hydrogen-bond donors increases the stabil-ity of the receptor : anion complexes by more than five-fold in 130, but in 131


they are poorly orientated for anion coordination and the stability constantsare only marginally higher than in 129. Receptor 132 which uses urea groupsas hydrogen-bond donors binds anions with 1 : 2 receptor : anion stoichiom-etry, with stability constants an order of magnitude higher than 131.

5Conclusion/Outlook

The geometric, Lewis acidic, optical and electrochemical properties of metalsmake them ideal for use as multifunctional components in the constructionof anion receptors. They are able to combine roles such as reporter groups,anion-binding sites and structural components to form receptor systems withsophisticated anion-sensing properties. In particular metal-containing recep-tors form the basis of systems in which anion-binding can be switched on andoff, or in which the anion guest can even be moved from one binding site toanother. Another significant advance has been the development of nanoscalestructures such as dendrimers, nanoparticles, thin polymer films and self-assembled monolayers which incorporate redox- or photoactive metal centresas an anion-sensing component. The wide variety of added functionality andsheer adaptability that metals bring to anion receptor chemistry means thatmetal-based systems are set to continue at the forefront of research in thisarea.

References

1. Beer PD, Cadman J (2000) Coord Chem Rev 205:1312. Beer PD, Gale PA (2001) Angew Chem Int Ed 40:4863. Beer PD, Hayes EJ (2003) Coord Chem Rev 240:1674. Sun S-S, Lees AJ (2002) Coord Chem Rev 230:1715. Amendola V, Fabbrizzi L, Mangano C, Pallavicini P, Poggi A, Taglietti A (2001) Coord

Chem Rev 219:821


6. Bayly SR, Beer PD (2005) Anion Sensing. Topics Curr Chem 255:1257. Beer PD, Gale PA, Chen GZ (1999) J Chem Soc Dalton Trans, p 18978. Beer PD, Keefe AD, Drew MGB (1989) J Organomet Chem 378:4379. Beer PD (1996) Chem Commun, p 689

10. Beer PD, Hesek D, Kingston JE, Smith DK, Stokes SE, Drew MGB (1995) Organo-metallics 14:3288

11. Beer PD, Graydon AR, Johnson AOM, Smith DK (1997) Inorg Chem 36:211212. Beer PD, Chen Z, Goulden AJ, Graydon A, Stokes SE, Wear T (1993) J Chem Soc Chem

Commun, p 183413. Reynes O, Moutet J-C, Pecaut J, Royal G, Saint-Aman E (2002) New J Chem 26:914. Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc

Chem Commun, p 104615. Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim

Acta 246:14316. Beer PD, Cadman J, Lloris JM, Martinez-Manez R, Padilla ME, Pardo T, Smith DK,

Soto J (1999) J Chem Soc Dalton Trans, p 12717. Dusemund C, Sandanayake KRAS, Shinkai S (1995) J Chem Soc Chem Commun,

p 33318. Reynes O, Maillard F, Moutet J-C, Royal G, Saint-Aman E, Stanciu G, Dutasta J-P,

Gosse I, Mulatier J-C (2001) J Organomet Chem 637-639:35619. Beer PD, Drew MGB, Hesek D, Jagessar R (1995) J Chem Soc Chem Commun, p 118720. Beer PD, Hesek D, Nam KC, Drew MGB (1999) Organometallics 18:393321. Tomapatanaget B, Tuntulani T, Chailapakul O (2003) Org Lett 5:153922. Kim DS, Miyaji H, Chang BY, Park SM, Ahn KH (2006) Chem Commun, p 331423. Cui XL, Delgado R, Carapuca HM, Drew MGB, Felix V (2005) Dalton Trans, p 329724. Beer PD, Stokes SE (1995) Polyhedron 14:87325. Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) Dalton Trans, p 368526. Kingston JE, Ashford L, Beer PD, Drew MGB (1999) J Chem Soc Dalton Trans, p 25127. Beer PD, Drew MGB, Jagessar R (1997) J Chem Soc Dalton Trans, p 88128. Altmann R, Gausset O, Horn D, Jurkschat K, Schuermann M, Fontani M, Zanello P

(2000) Organometallics 19:43029. Beer PD, Szemes F, Balzani V, Sala CM, Drew MGB, Dent SW, Maestri M (1997) J Am

Chem Soc 119:1186430. Szemes F, Hesek D, Chen Z, Dent SW, Drew MGB, Goulden AJ, Graydon AR, Grieve A,

Mortimer RJ, Wear T, Weightman JS, Beer PD (1996) Inorg Chem 35:586831. Beer PD (1998) Acc Chem Res 31:7132. Beer PD, Dent SW, Wear TJ (1996) J Chem Soc Dalton Trans, p 234133. Lin ZH, Zhao YG, Duan CY, Zhang BG, Bai ZP (2006) Dalton Trans, p 367834. Mizuno T, Wei WH, Eller LR, Sessler JL (2002) J Am Chem Soc 124:113435. Beer PD, Berry N, Drew MGB, Fox OD, Padilla-Tosta ME, Patell S (2001) Chem Com-

mun, p 19936. Berry NG, Pratt MD, Fox OD, Beer PD (2001) Supramol Chem 13:67737. Beer PD, Berry NG, Cowley AR, Hayes EJ, Oates EC, Wong WWH (2003) Chem Com-

mun, p 240838. Valerio C, Fillaut J-L, Ruiz J, Guittard J, Blais J-C, Astruc D (1997) J Am Chem Soc

119:258839. Alonso B, Casado CM, Cuadrado I, Moran M, Kaifer AE (2002) Chem Commun,

p 177840. Daniel M-C, Ruiz J, Blais J-C, Daro N, Astruc D (2003) Chem Eur J 9:4371


41. Beer PD, Davis Jason J, Drillsma-Milgrom DA, Szemes F (2002) Chem Com-mun, p 1716

42. del Peso I, Alonso B, Lobete F, Casado CM, Cuadrado I, Losada del Barrio J (2002)Inorg Chem Commun 5:288

43. Reynes O, Gulon T, Moutet J-C, Royal G, Saint-Aman E (2002) J Organomet Chem656:116

44. Reynes O, Bucher C, Moutet J-C, Royal G, Saint-Aman E (2004) Chem Commun, p 42845. Labande A, Astruc D (2000) Chem Commun, p 100746. Bayly SR, Gray TM, Chmielewski MJ, Davis JJ, Beer PD (2007) Chem Commun, p 223447. Basurto S, Riant O, Moreno D, Rojo J, Torroba T (2007) J Org Chem 72:467348. Miyaji H, Collinson SR, Prokes I, Tucker JHR (2003) Chem Commun, p 6449. Aldridge S, Bresner C, Fallis IA, Coles SJ, Hursthouse MB (2002) Chem Commun,

p 74050. Zhang BG, Xu J, Zhao YG, Duan CY, Cao X, Meng QJ (2006) Dalton Trans, p 127151. Oton F, Tarraga A, Espinosa A, Velasco MD, Molina P (2006) J Org Chem 71:459052. Oton F, Tarraga A, Molina P (2006) Org Lett 8:210753. Plitt P, Gross DE, Lynch VM, Sessler JL (2007) Chem Eur J 13:137454. Beer PD, Timoshenko V, Maestri M, Passaniti P, Balzani V (1999) Chem Commun,

p 175555. Deetz MJ, Smith BD (1998) Tetrahedron Lett 39:684156. Watanabe S, Onogawa O, Komatsu Y, Yoshida K (1998) J Am Chem Soc 120:22957. Cormode DP, Murray SS, Cowley AR, Beer PD (2006) Dalton Trans, p 513558. Goodall W, Williams JAG (2000) J Chem Soc Dalton Trans, p 289359. Lo KKW, Lau JSY, Lo DKK, Lo LTL (2006) Eur J Inorg Chem, p 405460. Slone RV, Yoon DI, Calhoun RM, Hupp JT (1995) J Am Chem Soc 117:1181361. Sun S-S, Anspach JA, Lees AJ, Zavalij PY (2002) Organometallics 21:68562. Sun S-S, Lees AJ, Zavalij PY (2003) Inorg Chem 42:344563. de Wolf P, Waywell P, Hanson M, Heath SL, Meijer A, Teat SJ, Thomas JA (2006) Chem

Eur J 12:218864. Curiel D, Beer PD (2005) Chem Commun, p 190965. DeSantis G, Fabbrizzi L, Licchelli M, Poggi A, Taglietti A (1996) Angew Chem Int Ed

Engl 35:20266. Fabbrizzi L, Licchelli M, Parodi L, Poggi A, Taglietti A (1999) Eur J Inorg Chem, p 3567. Coskun A, Baytekin BT, Akkaya EU (2003) Tetrahedron Lett 44:564968. Fabbrizzi L, Licchelli M, Rabaioli G, Taglietti A (2000) Coord Chem Rev 205:8569. Fabbrizzi L, Leone A, Taglietti A (2001) Angew Chem Int Ed 40:306670. Han MS, Kim DH (2002) Angew Chem Int Ed 41:380971. Tobey SL, Jones BD, Anslyn EV (2003) J Am Chem Soc 125:402672. Tobey SL, Anslyn EV (2003) J Am Chem Soc 125:1480773. Tobey SL, Anslyn EV (2003) Org Lett 5:202974. Lee HN (2007) J Am Chem Soc 129:382875. Allevi M, Bonizzoni M, Fabbrizzi L (2007) Chem Eur J 13:378776. Amendola V, Colasson B, Fabbrizzi L, Douton MJR (2007) Chem Eur J 13:498877. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L (2006) Inorg Chem 45:613878. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:392079. Amendola V, Boiocchi M, Colasson B, Fabbrizzi L, Douton MJR, Ugozzoli F (2006)

Angew Chem Int Ed 45:692080. Dickins RS, Gunnlaugsson T, Parker D, Peacock RD (1998) Chem Commun, p 164381. Bruce JI, Dickins RS, Govenlock LJ, Gunnlaugsson T, Lopinski S, Lowe MP, Parker D,

Peacock RD, Perry JJB, Aime S, Botta M (2000) J Am Chem Soc 122:9674


82. Montalti M, Prodi L, Zaccheroni N, Charbonniere L, Douce L, Ziessel R (2001) J AmChem Soc 123:12694

83. Yamada T, Shinoda S, Tsukube H (2002) Chem Commun, p 121884. Mahajan RK, Kaur I, Kaur R, Uchida S, Onimaru A, Shinoda S, Tsukube H (2003)

Chem Commun, p 223885. Leonard JP, dos Santos CMG, Plush SE, McCabe T, Gunnlaugsson T (2007) Chem

Commun, p 12986. Plush SE, Gunnlaugsson T (2007) Org Lett 9:191987. Beer PD, Cormode DP, Davis JJ (2004) Chem Commun, p 41488. Beer PD, Dickson CAP, Fletcher N, Goulden AJ, Grieve A, Hodacova J, Wear T (1993)

J Chem Soc Chem Commun, p 82889. Holman KT, Orr GW, Atwood JL, Steed JW (1998) Chem Commun, p 210990. Staffilani M, Hancock KSB, Steed JW, Holman KT, Atwood JL, Juneja RK, Burkhal-

ter RS (1997) J Am Chem Soc 119:632491. Camiolo S, Coles SJ, Gale PA, Hursthouse MB, Mayer TA, Paver MA (2000) Chem

Commun, p 27592. Bondy CR, Gale PA, Loeb SJ (2004) J Am Chem Soc 126:503093. Vega IE, Gale PA, Light ME, Loeb SJ (2005) Chem Commun, p 4913


Recent Progress of Phosphate Derivatives RecognitionUtilizing Artificial Small Molecular Receptorsin Aqueous Media

Shun-ichi Tamaru1 (�) · Itaru Hamachi2

1Department of Nano-science, Sojo University,4-22-1 Ikeda, 860-0082 Kumamoto, [email protected]

2Department of Synthetic Chemistry and Biological Chemistry, Kyoto University,Kyotodaigaku Katsura, Nishikyo-ku, 615-8510 Kyoto, Japan

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

2 Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds 982.1 Cyclic/Acyclic Polyamine Type Artificial Chemosensors . . . . . . . . . . . 982.2 Artificial Chemosensors Containing Other Cationic Groups . . . . . . . . . 102

3 Phosphate Anion Recognition Utilizing Coordination Chemistry . . . . . 1043.1 Zn-cyclen Type Artificial Chemosensors . . . . . . . . . . . . . . . . . . . . 1053.2 Zn-Dpa Type Artificial Chemosensors . . . . . . . . . . . . . . . . . . . . . 1063.3 Ratiometric Detection of Phosphate Derivatives . . . . . . . . . . . . . . . 109

4 Recognition of Phosphorylated Protein Surfaces . . . . . . . . . . . . . . . 112

5 High Throughput Sensing Systems for Phosphate DerivativesUsing Semi-wet Sensor Array . . . . . . . . . . . . . . . . . . . . . . . . . 119

5.1 Molecular Recognition of Chemosensors in a Supramolecular Hydrogel . . 1195.2 Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibers

for Discrimination of Phosphate Derivatives . . . . . . . . . . . . . . . . . 120

6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

Abstract Phosphate derivatives are ubiquitous in living organisms playing several import-ant roles. Therefore, the development of sophisticated artificial phosphate receptors andchemosensors that can work in aqueous conditions is currently an area of great inter-est. There is a need to develop new methodologies for detection, separation or transportof biologically relevant phosphate derivatives. In the past two decades, many artificialchemosensors have been reported; these can be broadly classified into (1) chemosensorsutilizing electrostatic interactions and hydrogen bonds and (2) chemosensors utilizing co-ordination chemistry. The development of these receptors is often inspired by the designof substrate-binding centers in natural enzymes such as protein kinases, phosphatasesand phospholipases. More recently, the targets of such chemosensors have been broad-ened to include the recognition of phosphorylated protein surfaces. Here we review therecent progress in the development of molecular receptors and chemosensors that canselectively detect phosphate derivatives in aqueous media.

96 S.-i. Tamaru · I. Hamachi

Keywords Phosphate · Molecular recognition · Phosphoprotein · Host-guest ·Fluorescent sensing

AbbreviationsADP adenosine 5′-diphosphateAMP adenosine 5′-monophosphateATP adenosine 5′-triphosphateAXP adenosine nucleotidecAMP adenosine 3′,5′-cyclic monophosphateCD circular dichroismCTP cytosine 5′-triphosphateCoA coenzyme A5-CF 5-carboxylfluoresceindiMeP dimethyl phosphateGTP guanosine 5′-triphosphateIP3 d-myoinositol 1,4,5-triphosphateMAPK mitogen-activated protein kinaseMeP methyl phosphateMLCT metal-ligand charge transferNPP 4-nitrophenyl phosphatePDGFR-β platelet-derived grows factor receptor-βPET photo-induced electron transferPhP phenyl phosphatePPi pyrophosphatePV pyrocatechol violetp-Tyr O-phospho-L-tyrosineUDP-Gal uridine 5′-diphospho-α-d-galactoseZn-Dpa zinc(II)-dipicolylamine

1Introduction

In nature, a variety of phosphate anion derivatives are ubiquitously dis-tributed [1]. In addition to inorganic phosphate, there are many phosphatesfound in biological molecules. For instance, the backbones of DNA and RNAconsist of phosphate diesters connecting 3′- and 5′-oxygens of ribose. Nu-cleoside phosphates (nucleotides) are not only the structural units of RNAand DNA, but also the essential parts of several cofactors such as coen-zyme A (CoA), flavin adenine dinucleotide (FAD), and nicotinamide ade-nine dinucleotide (NAD). They play important roles in metabolic cascadesof biological substances and in cellular signaling. Adenosine triphosphate(ATP) is a primary energy currency molecule in all living organisms, whichis produced from ADP and inorganic phosphate by ATPase in cellular res-piration or photosynthesis. The covalent bond connecting the phosphategroups stores a high energy that is easily broken down by hydrolysis to re-

Phosphate Recognition Utilizing Artificial Small Molecular Receptors 97

lease an amount of energy sufficient to drive many biochemical reactions.For example, the nucleoside pyrophosphate derivatives of saccharides suchas uridine 5′-diphospho-α-D-galactose (UDP-Gal) are universal substrates ofglycosyltransferase necessary to form a glycosyl bond for the synthesis ofglycoprotein and oligo-saccharides. Unlike the molecular elements in bioen-ergetics, adenosine 3′,5′-cyclic monophosphate (cAMP) and D-myoinositol1,4,5-triphosphate (IP3) represent important second messengers playing cen-tral roles in intra-cellular signal transduction [2, 3].

Phosphorylated proteins are other important phosphate derivatives in bio-logical phenomena, which are produced by protein kinases, a family of en-zymes catalyzing phosphorylation. Protein phosphorylation is one of themost common post-translational protein modifications in eucaryotic cells.The vast majority of phosphorylation occurs as a mechanism to regulate thebiological activity of a protein [4]. From the genomic sequence, the presenceof over 500 kinds of protein kinase in a human body is forecasted [5]. Thisfact suggests that such post-translational CENOTE phosphorylation is widelyused to control diverse protein functions. In particular, phosphorylation onserine, threonine or tyrosine residues by protein kinases is fundamental tocell signaling networks. In these networks protein functions are controlledvia the reversible phosphorylation/dephosphorylation processes catalyzed bycorresponding kinase/phosphatase coupling with spatio-temporal fashions inresponse to changes of the cellular environment [6]. In practice, the phos-phorylation acts as a molecular switch to turn on the protein activity. A typ-ical example of such allosteric control of protein functions is observed inmitogen-activated protein kinase (MAPK) ERK2. In the deactivated state, twodomains of ERK2 (extracellular signal-regulated kinase 2) are separated fromeach other by the loop strand. The structural consequences of phosphoryla-tion at threonine (Thr 183) and tyrosine (Tyr 185) residues on the loop strandinclude active site closure, alignment of key catalytic residues that interactwith ATP, and remodeling of the activation loop, resulting in recovery of thekinase activity [7]. Protein phosphorylation also plays a crucial role in con-trolling the reversible assembly of a cellular signaling complex [8, 9]. Phos-phorylation of the appropriate peptide sequence generates new interactionsites exposed on a protein surface which are recognized by the downstreamproteins containing phosphoprotein-binding domains, forming multibindingdomains such as SH2, PTB, FHA, and WW domains involved in many signal-ing proteins [10].

Due to such significant roles of phosphate derivatives in biological events,it is generally anticipated that development of sophisticated artificial recep-tors toward phosphates would greatly contribute to the understanding ofbiological phenomenon involving phosphate derivatives. Such advancementin understanding should afford new methodologies for detection, separa-tion, or transport of biologically important phosphates. In the following wereview the recent progress of molecular recognition of phosphate deriva-


tives utilizing artificial small molecular receptors that can work in aqueousmedia.

2Phosphate Recognition by Electrostatic Interactions and Hydrogen Bonds

2.1Cyclic/Acyclic Polyamine Type Artificial Chemosensors

Polyamines that exist as polycationic species in water under appropriate pHconditions are suitable scaffolds to capture phosphate derivatives by the for-mation of multiple-hydrogen bonding and electrostatic interaction. It haspreviously been shown that naturally occurring polyamines such as spermineand spermidine bind phosphates or DNA [11]. Nakai et al. reported that theapparent affinity constants, logKapp for the 1 : 1 complexation of sperminewith AMP, ADP, and ATP were found to be 2.6, 3.1 and 4.0, respectively, in50 mM tris buffer (pH 7.5) [12]. Based on this moderate affinity, several ar-tificial phosphate receptors containing polyamines were developed. Becauseof the difficulty of capturing phosphate anions by non-covalent interactionin aqueous media, however, successful examples of artificial small molecularreceptors or sensors for phosphates are still relatively scarce.

In a pioneering work, Czarnik and co-workers synthesized the branch-ing polyamine-appended anthracene derivative as a fluorescent chemosensorfor phosphate anions. Compound 1, having a branching polyamine chain onthe 9-position of the anthracene moiety, formed a 1 : 1 complex with inor-ganic phosphate or ATP in aqueous solution at pH 6 (logKeq = 0.82 for PO4

2–,4.2 for ATP) [13]. The complexation caused a fluorescence enhancement thatcan be ascribed to the cancellation of quenching due to photo-induced elec-tron transfer (PET), as shown in Scheme 1. In the absence of phosphate ions,1 showed weak fluorescence due to the PET quenching by the free aminegroup. The complexation of the anionic phosphate oxygens with ammoniumions on 1 localized the remaining phosphate OH group in close proximityto the free amine, and then favorable intracomplex proton transfer occurred.As a result, the intramolecular PET quenching was eliminated and strongeremission was observed.


Scheme 1 Proposed binding mechanism of Chemosensor 1 with inorganic phosphate.The PET quenching is canceled upon the phosphate binding with 1 causing fluorescenceenhancement

The Czarnik group then prepared compound 2, which has two polyaminechains at the 1- and 8-positions of anthracene, as a fluorescence chemosen-sor toward pyrophosphate (PPi) [14]. To effectively bind both phosphatesof PPi simultaneously, the polyamine chains were geometrically preorga-nized. The dissociation constant (Kd) for the 2-PPi complex was found to be2.9 µM, whereas that for 2-phosphate was 6.3 mM. In particular, chemosen-sor 2 displayed a 2200-fold pyrophosphate/phosphate discrimination at pH 7(Scheme 2). Such high ion selectivity allows a real-time assay of pyrophos-phate hydrolysis catalyzed by inorganic phosphatase.

Since macrocyclic polyamines have a much greater charge density withinthe molecular skeletons in the multi-protonated state as compared to the lin-ear polyamines, they should have an entropic advantage for the complexationof phosphates. On the basis of this advantage of macrocyclic polyamines overthe corresponding acyclic polyamines, Kimura et al. reported that a macro-cyclic hexaamine 3 showed stronger affinity to AMP, ADP and ATP at pH 8.0

Scheme 2 Proposed binding mechanism of Chemosensor 2 with pyrophosphate. The PETquenching is canceled upon the phosphate binding with 1 causing fluorescence enhance-ment


(logKs = 3.2, 5.6 and 6.4, respectively) than those for corresponding sperminecomplexes (mentioned above) [15].

Lehn and his co-workers reported that the oxygen-containing macrocyclicpolyamine 4 is a good receptor of nucleotides such as ATP and ADP [16, 17].Due to the formation of multiple ionic interactions, 4 strongly bound ATP, es-pecially under very acidic conditions (logKs = 4.8 at pH 7.6, 11.0 at pH 4.0).Interestingly, they found that compounds 4 and 5 were capable of catalyzingthe hydrolysis of ATP to ADP (Scheme 3). In the complex of 4-ATP, a nitrogenatom attacked the terminal phosphate to yield ADP and thus phosphorylated4. The phosphorylated 4 was immediately hydrolyzed to regenerate the ori-ginal 4 in the catalytic process. Compound 5, in which a fluorogenic acridinering was introduced to the polyamine ring, showed greater selectivity be-tween ATP and ADP than the parent compound 4. Compound 5 bound suchnucleotides using two distinct forces, electrostatic interactions between themacrocyclic polycationic moiety and the polyphosphate and π-π stackinginteractions between the acridine ring and nucleic base. The latter inter-actions induced the fluorescence change of the acridine moiety of 5 uponcomplexation.

Scheme 3 Schematic representation of the catalytic cycle for ATP hydrolysis by the chemo-sensor 5 following the nucleophilic pathway


Other types of polycationic macrocyclic compound like 6 ∼ 8 were alsoreported as water-soluble receptors for nucleotides [18–20]. The apparentcomplexation constants, logKapp, toward ATP for 6 and 8 at neutral pH werefound to be 5.2 and 8.4, respectively, while the logKs for 7 was 4.8.

Coupling a suitable signal transducer to the binding scaffold is generallycrucial for designing efficient chemosensors. For example, redox-active metalcomplexes such as metallocenes have been incorporated into various guest-binding scaffolds to yield chemosensors that detect the corresponding guestselectrochemically. Beer et al. synthesized many artificial receptors contain-ing redox active metal centers and evaluated their usefulness[21–23] In thesestudies, they prepared water-soluble phosphate chemosensors. Chemosen-sors 9 ∼ 11, tethering cyclic polyamines on a ferrocene moiety, bound ATPor phosphate to form a 1 : 1 complex at pH 6.5. By cyclic voltammetric ex-periments, it was revealed that the receptors were able to detect these phos-phate anions using cathodic shifts of the redox potential of ferrocene by60–80 mV [24, 25].

Ruthenium-(II)-tris-(2,2′-bipyridiyl), Ru(bpy)3, has proven to be a usefulsignal transducer by virtue of its ability to change its luminescent emis-sion upon guest binding. Water soluble Ru(bpy)3-appended polyamine re-ceptors 12 ∼ 14 were prepared by Beer’s group. These receptors detectedATP and phosphate in aqueous media by changing the metal-ligand chargetransfer (MLCT) luminescent emission [26]. The intensity of MLCT emission(600–630 nm) was sensitive to pH and phosphate anions. The emission titra-tion studies with phosphate and ATP showed that the MLCT emission of thesechemosensors was quenched by 15% upon complexation of the guest.


2.2Artificial Chemosensors Containing Other Cationic Groups

In addition to polyamines, guanidinium groups have also shown to be use-ful binding groups for phosphate anions. It is well-known that the activesite of some phosphatases such as Staphylococcal nuclease [27] and alkalinephosphatase [28] contain guanidinium groups (from Arg residues) that areactively involved in binding phosphate derivatives. Inspired by this approach,several groups have designed guanidinium-based phosphate receptors.

Anslyn and co-workers have published several interesting papers on mo-lecular recognition of phosphate by cleft-like receptor molecules containingpoly-guanidinium or -ammonium rims [29–31]. Phosphate receptors 15, 16were designed for detection of D-myoinositol 1,4,5-triphosphate (IP3) [30]. Toevaluate the binding affinity, they devised a fluorescent competitive assay. Inthis system, 5-carboxylfluorescein (5-CF) was initially bound to the receptors(15, 16) to form a 1 : 1 complex. The fluorescence of 5-CF increased as a con-sequence of the entrapment in the cleft of the receptor from aqueous media.The addition of IP3 to an aqueous solution containing 15-(5-CF) complexled the guest exchange to form the 15-IP3 complex and caused the concur-rent release of 5-CF from the cleft to aqueous media, which in turn caused adramatic decrease of 5-CF emission. From the change in fluorescence it waspossible to calculate a logK value of 5.7 for the formation of 15-IP3.

On the basis of this cleft-type scaffold, Anslyn’s group prepared a chemo-sensor library to discover a sophisticated nucleotide receptor possessing highselectivity toward ATP [32]. The attachment of the scaffold on Wang resinfor solid phase synthesis, followed by tethering two tripeptide chains branch-


ing off the guanidium groups by the split-and-pool method yielded a smalllibrary of potential candidates for chemosensors. Several “hit” moleculeswere selected and picked up by the fluorescence assay with ATP derivative(N-methylanthraniloyl-ATP: MANT-ATP). Among them, chemosensor 17 con-taining Ser-Tyr-Ser sequences, was finally discovered as an ATP selectivechemosensor. The binding constant of 17 for ATP was found to be 3.4×103 M–1,approximately 10 times greater than that of the reference compound 18.

Alternatively, Rebek et al. rationally designed and synthesized a water-soluble nucleotide receptor 19 that is comprised of a carbazole backbone,a cyclic guanidinium cation, and two Kemp’s triacidic imides [33]. Compound19 bound cyclic AMP (cAMP) to form a 1 : 1 complex (logKs = 2.8) in aqueousmedia by cooperatively accumulated interactions, specifically electrostatic in-teraction between its guanidinium cation and the anionic phosphate diester,π-π stacking interaction between its carbazole backbone and adenine ring,and hydrogen bonding interactions (Fig. 1).

Fig. 1 Structure of 19 and the predicted lowest energy conformation of the complexbetween 19 and cAMP

Amidinium cations are used as an alternative to guanidinium cation andas they are more easily accessible in organic synthesis. Gramlich and co-


workers reported the complexation behavior of cationic “cavitands” havingfour amidinium cations on the peripheral of resorcin[4]arene, toward nu-cleotides [34]. Cavitand 20 possessed four ethylene glycol chains on the lowerrim of resorcin arene that enhance the water-solubility. The binding constantsof 20 to nucleotides were estimated to be 6.6×105 M–1 for ATP, 4.9×104 M–1

for ADP, 1.0×104 M–1 for AMP, and 1.4×103 M–1 for cAMP. 1H NMR studiessuggested that the adenine ring was encapsulated in the hydrophobic cavity ofresorcin[4]arene (Fig. 2). The affinity of 20 to AXP was only slightly greaterthan that to the corresponding nucleotides having other nucleobases, how-ever, suggesting that electrostatic interactions between amidinium groupsand phosphates are predominant in the complexation between the cavitandand nucleotides.

Fig. 2 Structure of 20 and the energy-minimized structure of the complex formed be-tween chemosensor 20 and AMP

Similar to guanidinium, imidazole, a basic group of histidine side chains,might be used as a binding unit for phosphate in artificial receptors. As far aswe know, however, there are no reported examples of such receptors that cansuccessfully bind phosphates in aqueous media. Known examples are limitedto receptors used in organic media. This again indicates the difficulty in thedevelopment of sophisticated phosphate receptors usable in aqueous solutions.

3Phosphate Anion Recognition Utilizing Coordination Chemistry

As mentioned above, many type of phosphate receptors containing polyaminesas binding sites have been developed. Because polyamines also have high


affinity toward various metal ions, however, the metal complexation with suchbinding sites competes with the desired phosphate capturing under physio-logical conditions. Therefore, the utilization of the polyamine type phosphatereceptors in biological applications may be seriously restricted. In addition,the moderate binding affinity of polyamines toward phosphates under bi-ological conditions is another drawback. On the other hand, coordinationchemistry provided by metal complexes is often employed to bind a phos-phate unit of substrates in the enzyme active site [35, 36]. This gives onean important clue for the design of phosphate-binding motifs that show thestronger affinity in aqueous media.

3.1Zn-cyclen Type Artificial Chemosensors

Kimura et al. reported a series of zinc(II) complexes of cyclic polyamineligands as biomimetic models of zinc-containing enzymes such as carbonicanhydrase and alkaline phosphatase A [37]. In these biomimetic model sys-tems, the phosphate group from the substrates coordinate to the zinc(II)centers [38]. Initially, it became clear that the mono-nuclear Zn-cyclen com-plex 21 formed a 1 : 1 complex with monodentate phosphate dianions such asphosphate, phenyl phosphate (PhP, logKs = 3.5), and 4-nitrophenyl phosphate(NPP, logKs = 3.1), whereas the metal-free cyclen showed considerably weakeraffinity with these phosphate dianions under the same conditions. This in-dicated that the coordination bond between phosphate and metal center wasstrong enough in aqueous media, thus such the non-covalent interactionshould be powerful candidate to capture phosphate anions in water. Ditopiccompound 22 bound PhP or NPP as a sandwich-shaped 1 : 1 complex withmoderate affinity (logKs = 4.6 for PhP, 4.0 for NPP) [40]. Trinuclear zinc com-plex 23, a mimic of the active site of zinc enzymes such as phospholipase C,showed the strongest affinity toward these phosphates (logKs = 5.8 for PP, 6.6for NPP) among the Zn-cyclen type receptors, probably due to capturing thephosphate moiety in its pseudo-cleft like binding site [41]. Three moleculesof a bipyridine derivative equipped with two Zn-cyclen complexes 24 were


Scheme 4 Cartoon of the structure of chemosensor (24)3-Ru

assembled with ruthenium(II) to form another type of phosphate receptor(24)3-Ru (Scheme 4) [42]. Having two sets of the trinuclear binding site at thetop and bottom of the assembly, the receptor selectively bound IP3 or CTP3in a 1 : 2 stoichiometry.

Interestingly, Kimura and coworkers found that receptor 21 selectivelyformed a 1 : 1 complex (logKs =3.4) with deoxythymidine (dT) or uracilanalogues at physiological pH. In these systems, the deprotonated imide an-ion of dT coordinated to the zinc (II) ion center and two sets of hydrogenbonds were formed between carbonyl oxygens of dT and the amine protons of21 [43]. This is the first dT (or U)-selective artificial receptor to work in aque-ous solution. They subsequently accomplished the selective recognition ofthymidine nucleotide by using 22 with relatively high affinity (logKs = 5 ∼ 6).In these complexes, the thymine moiety was bound to Zn-cyclen and thephosphate anion moiety was concurrently coordinated to the other Zn center.

3.2Zn-Dpa Type Artificial Chemosensors

Hamachi and co-workers developed a fluorescent ATP chemosensor con-sisting of dinuclear zinc(II)-dipicolylamine (Zn-Dpa) appended to 2-acetyl


anthracene 25 [44]. The apparent complexation constants of 25, Kapp, towardATP, ADP, and AMP were found to be 2.2×106, 2.2×105, and 6.6×104 M–1,respectively, at neutral aqueous pH, while the considerably weaker bindingof mononuclear analogue 26 to AXP was observed. This suggests that thetwo Zn-Dpa moieties cooperatively captured the phosphate units of the nu-cleotides. The pattern of the fluorescence change of 25 upon complexationwas dependent on the kind of nucleic bases of nucleotides. The binding ofATP to 25 caused a dramatic fluorescence enhancement (I/I0≈3), whereasa small increase of fluorescence was observed in the case of CTP. In con-trast, the emission was diminished with the addition of GTP, probably due tothe electron transfer quenching by the guanine group. Similar nucleic base-dependent fluorescence change was also reported in the nucleotides sensingstudy utilizing a cyclic polyamine-type of receptor 5.

Hong et al. produced binuclear Zn complex type receptors 27, 28 as fluoro-genic and chromogenic phosphate sensors, having high selectivity toward py-rophosphate [45, 46]. Kim and his co-workers reported that a complex of 29with pyrocatechol violet (PV) can colorimetrically detect phosphate deriva-tives such as inorganic phosphate and AMP through the ligand exchange-induced color change [47, 48]. In this system, the bound PV was effectively


Scheme 5 Schematic representation of the mechanism of chromatic detection by chemo-sensor 29-PV toward inorganic phosphate. The effective displacement of PV with phos-phate anion at the binding site results in color change of the solution from blue to yellow

displaced with phosphate anion at the binding site so as to cause the remark-able color change of the solution from blue to yellow (Scheme 5).

Excimers, excited dimers of two planar fluorophores, often display sig-nificantly red-shifted fluorescence relative to the corresponding monomer.Hence, suitable molecules that can form excimers resulting from complexa-tion with the target phosphate would become unique chemosensors that showdrastic fluorescence change during phosphate recognition. A chemosensor 30attaching two Zn-Dpa units to both ends of naphtalenediimide formed a 2 : 2complex with pyrophosphate resulting in an excimer formation (logKapp =4.1×105 M–1), whereas the addition of other anions including AXP did notshow any meaningful fluorescence change [49]. This sensor can selectivelydetect pyrophosphate in the presence of ATP or inorganic phosphate by thecharacteristic red-shifted emission. In addition to the four Zn binding sitesfor pyrophosphate, the favorable π-π interaction of the two flat aromatic

Scheme 6 Proposed binding mechanism of chemosensor 30 with PPi. The formation ofa 2 : 2 type binding between 30 and pyrophosphate leads to excimer formation


moieties allowed the formation of a 2 : 2 type binding between 30 and py-rophosphate (Scheme 6).

3.3Ratiometric Detection of Phosphate Derivatives

The detection of a specific anion by using the emission or excitation changeat two different wavelengths provides a significant advantage over conven-tional measurements using a single wavelength change. Such a dual exci-tation/emission system enables a ratiometric detection of an analyte, thusallowing precise and quantitative analysis and imaging even in complex en-vironments such as that inside living cells. Hamachi et al. successfully de-veloped a bis(Zn-Dpa) appended xanthone type receptor 31 that is capableof ratiometric detection of phosphate derivatives under physiological condi-tions using excitation fluorescence [50]. As shown in Fig. 3, the addition ofATP to an aqueous solution of 31 resulted in a see-saw type excitation spectralchange. The mechanism of such a excitation spectral change was proposed asfollows. In the absence of phosphate, the carbonyl oxygen atom of the xan-thone fluorophore was coordinated to both Zn centers, as shown in Scheme 7.When a phosphate derivative was bound, the fickle coordination bond be-tween one of the Zn centers and the carbonyl oxygen atom was cleaved bythe strong interaction of the metal ion with the phosphate, which caused theexcitation change of the xanthone fluorophore.

Acridine type chemosensors 32, 33 were also designed by Hamachi andcoworkers as dual-emission chemosensors of nucleoside pyrophosphate

Fig. 3 Change in the excitation spectrum of 31 induced by ATP addition


Scheme 7 Phosphate anion-induced coordination rearrangement of 31 upon binding toa phosphate derivative

(Fig. 4) [51]. In the resting state, these sensors are in equilibrium betweenmono- and binuclear Zn complexes. In this state the sensors predominantlyexist as mononuclear Zn complexes, with binuclear Zn complexes as theminor species, as illustrated in Scheme 8. The binding of anionic nucleo-

Fig. 4 Change in the fluorescence emission of 32 (a) and 33 (b) induced by ATP addition


Scheme 8 Schematic representation of the dual-emission sensing mechanism of the acri-dine chemosensor for nucleoside PPs

side pyrophosphate with the sensor molecules induced complexation of thesecond Zn, concurrently cleaving the originally formed coordination bondbetween Zn and the acridine nitrogen atom. Such a guest-induced rearrange-ment of the coordination chemistry resulted in the dual-emission change.The chemosensors showed high selectivity for nucleotide di- or tri-phosphateover nucleoside monophosphate, inorganic phosphate, and nucleotide deriva-tives of monosaccharides such as uridine 5′-diphospho-α-D-galactose (UDP-Gal). On the basis of such selectivity, these chemosensors were successfullyutilized for real-time monitoring of glycosyl transfer catalyzed by β-1,4-galactosyltransferase. Since UDP-Gal is converted into UDP in the process ofthe glycosyltransfer reaction, the detection of UDP by chemosensor 33 im-plied a monitoring of the reaction progress. It should be noted that UDP-Galhas a lower affinity toward 33 than UDP. This can be done in a dual-emissionchange manner without any modification of the reactants such as UDP-Galand the glycosyl acceptors (Scheme 9).

Kikuchi et al. reported a cyclen-Cd complex containing 7-amino-4-trifluorocumarin 34 as a fluorogenic chemosensor [52]. During the complex-ation of 34 with phosphates, the coordination bond of the aromatic aminogroup of the coumarin to Cd was cleaved, causing a dual-change of the ex-citation spectrum (Scheme 10). The dissociation constant of 34-nucleotide


Scheme 9 Fluorescence sensing of the glycosyl-transfer reaction based on ratiometricdetection by 33

Scheme 10 Design concept of sensing phosphates utilizing cyclen-Cd complex 34

complexes was approximately 10–5 M. This chemosensor was utilized for thereal-time monitoring of the activity of a phosphodiesterase which cleavesan undetectable cyclic nucleotide (e.g. cAMP) to produce the correspondingdetectable nucleotide (e.g. AMP).

4Recognition of Phosphorylated Protein Surfaces

In addition to small molecular phosphate derivatives, phosphate anionshaving large molecular weight recently emerged as important new sub-jects in anion recognition research. Hamachi and his co-workers developed


the first artificial chemosensors that can recognize phosphorylated pro-teins/peptides [53, 54]. These receptors 35, 36 in which the anthracene deriva-tives were linked with two Zn-Dpa complexes, bound single phosphategroups of peptides or proteins by the cooperative action of two Zn-Dpasites, so that the fluorescence intensity increased due to the anthracene fluo-rophore. From the fluorescence change, the apparent affinity constants for the1 : 1 complexation with inorganic phosphate or phosphate monoesters such aso-phosphotyrosine (p-Tyr) and relevant species were found to be in the rangeof 104∼105 M–1 in aqueous solution (Table 1). These chemosensors tightlycomplexed with ATP or ADP with stronger affinity (Kapp > 107 M–1), whereasthe phosphate diesters such as dimethyl phosphate (diMeP) and cyclic AMPdid not cause any fluorescence change up to the millimolar concentrationrange. No evidence for binding of the chemosensors was obtained when otheranions such as sulfate, nitrate, acetate, fluoride, or carboxylate were added, in-dicating that the chemosensors possessed high selectivity toward phosphate

Table 1 Summary of the apparent binding constants (Kapp, M–1) of 35 and 36 to phos-phate species by fluorescence change

Phosphate Chemosensor Phosphate Chemosensorderivative a 35 36 derivative a 35 36

NaH2PO4b 4.2×105 2.9×105 ATP c > 107 4.0×105

PhP b 2.1×105 5.1×104 ADP c > 107 1.6×105

p-Tyr b 3.1×105 6.1×105 AMP c 2.3×105 9.1×103

MeP b 1.1×105 7.9×103 cAMP c – d – d

DiMeP b – d – d

a PhP = phenyl phosphate, p-Tyr = O-phospho-L-tyrosine, MeP = methyl phosphate,diMeP = dimethyl phosphate, ATP = adenosine 5′-triphosphate, ADP = adenosine 5′-di-phosphate, AMP = adenosine 5′-monophosphate, cAMP = adenosine 3′,5′-cyclic mono-phosphate.b Measurement conditions: 10 mM HEPES, pH 7.2, 20 ◦C.c 50 mM HEPES, 50 mM NaCl, pH 7.2, 20 ◦C.d Since the fluorescence change was scarcely observed, the association constant cannot beobtained.


species among various anions, and can distinguish phosphate and phosphatemonoesters from phosphate diesters.

Determination of the binding ability of the chemosensors 35, 36 to phos-phorylated peptides was conducted as a model study of phosphorylated pro-tein surface recognition. Similar to phosphate sensing, the fluorescence inten-sity of 35 increased upon addition of a phosphorylated peptide in aqueoussolution. In the titration study with peptide-a (EEEI-pY-EEFD), a consensussequence phosphorylated by a protein kinase v-Src, the fluorescence of 35was enhanced, with the intensity finally reaching a 2.5 fold increase relativeto that in the absence of the peptide. In contrast, the corresponding non-phosphorylated peptide-g (EEEI-Y-EEFD) did not cause any emission change,showing that the chemosensor can distinguish a phosphorylated peptidefrom a non-phosphorylated one. Interestingly, the affinity of the chemosen-sors depended on the number of negative charges located on the phospho-rylated peptide. Among the tested peptides, both chemosensors showed thestrongest binding affinity (Kapp of 106–107 M–1) for peptide-a, which has alarger negative charge (– 8) (Table 2).

The significant emission enhancement of these chemosensors and the highselectivity towards phosphorylated peptides enabled the detection of phos-phorylated peptides by naked inspection of the emission change. This isillustrated in the photograph shown in Fig. 5. Such fluorescence intensifica-tion of the chemosensors is clearly ascribed to the phosphate-assisted bindingof the second Zn cation. A schematic illustration of the sensing mechanismtoward the phosphorylated peptide is depicted in Scheme 11. In the absenceof a phosphorylated peptide, the second Dpa site of the chemosensor is

Table 2 Amino acid sequences of peptides containing optimal consensus sequences phos-phorylated by different protein kinases and apparent binding constants (Kapp, M–1) of 35and 36 to the peptides as determined by fluorescence change

Consensus substrate Kinase Net 35 36sequence charge

peptide-a Glu-Glu-Glu-Ile-pTyr-Glu-Glu-Phe-Asp v-Src – 8 8.9×106 9.5×105

peptide-b Asp-Glu-Glu-Ile-pTyr-Gly-Glu-Phe-Phe c-Src – 6 1.5×106 3.6×105

peptide-c Ala-Glu-Glu-Ile-pTyr-Gly-Val-Leu-Phe Lck1 – 4 8.2×105 1.5×105

peptide-d aLys-Ser-Gly-pTyr-Leu-Ser-Ser-Glu EGFR – 2 5.8×104 1.2×104

peptide-e Ala-Arg-Arg-Gly-pSer-Ile-Ala-Ala-Phe PKA 0 – b – b

peptide-f Arg-Arg-Phe-Gly-pSer-Ile-Arg-Arg-Phe Bck1 + 2 – b – b

peptide-g Glu-Glu-Glu-Ile-Tyr-Glu-Glu-Phe-Asp v-Src – b – b

a Amino acid sequence of ezrin (142–149) phosphorylated by EGFR.b Since the fluorescence change was scarcely observed, the association constant cannot beobtained.


Fig. 5 Photograph of the increased emission of 36 in the presence of phosphorylatedpeptide-a (middle) compared to 36 only (left) and 36 with non-phosphorylated peptide-g(right)

Scheme 11 Schematic representation of the sensing mechanism of the chemosensors 35,36 toward the phosphorylated peptide


partially free from the Zn-complexation, so that the PET quenching by thebenzylic amine of Dpa lessened the fluorescence intensity of anthracene. Inthe presence of phosphorylated peptide, on the other hand, the binding ofthe second Zn cation to the free Dpa site was facilitated, and as a result thePET quenching was canceled so as to recover the fluorescence intensity. Thecareful thermodynamic study of this molecular recognition using isothermaltitration calorimetry (ITC) undoubtedly demonstrated that the binding wasan endothermic and entropy-driven event in the aqueous buffer solution.

These chemosensors were applied to two biological assays. Firstly, the real-time fluorescence monitoring of phosphatase-catalyzed dephosphorylationreactions was demonstrated. A phosphopeptide DADE-pY-LIPNNG (a frag-ment (988–998) of EGFR) was dephosphorylated by phosphatase PTP1Bto yield the non-phosphorylated peptide (Fig. 6) [54]. The pre-binding ofthe substrate peptide with 35 enhanced the fluorescence intensity of thechemosensor. After the addition of PTP1B, the fluorescence declined in atime-dependent manner during the progress of enzymatic dephosphoryla-tion of the peptide. This method is much simpler than the conventionalone using a radio-active phosphorylated peptide as the enzyme substrate.Secondly, the selective staining of phosphoprotein in SDS-PAGE was car-ried out [55]. The chemosensor 36 was shown to work well as a fluorescentstaining reagent in the conventional gel electrophoresis of the protein mix-tures. As shown in Fig. 7, only two distinct bands were fluorescently observedunder a UV trans-illuminator, those corresponding to phospho-ovalbumin(MW = 45.0 kDa) and phospho-α-casein (MW = 23.6 kDa). In contrast, veryslight or no emission was observed in other bands corresponding to the

Fig. 6 Time trace of PTP1B-catalyzed dephosphorylation monitored by the emission of 35with (circle) or without (square) PTP1B using a fluorescence plate reader


Fig. 7 Selective phosphoprotein detection in SDS-PAGE using 36. Each lane includestwo phosphoproteins (ovalbumin (45.0 kDa) and α-casein (24.0 kDa)), and four non-phosphorylated proteins (β-galactosidase) (MW = 116.0 kDa), bovine serum albumin(MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme (MW = 14.4 kDa). Lane 1: CBBstaining of the six proteins. Lane 2, 3: Detection of the phosphoproteins with UV transil-luminator after staining with 36. The amount of each protein in lane 2 and lane 3 is 5.0and 2.5 µg, respectively

four non-phosphorylated proteins β-galactosidase (MW = 116.0 kDa), bovineserum albumin (MW = 66.2 kDa), avidin (MW = 18.0 kDa), and lysozyme(MW = 14.4 kDa). Interestingly, brighter fluorescence was observed at theband of phospho-α-casein which has eight to nine phosphorylated aminoacid residues, compared to that of less phosphorylated phospho-ovalbumincontaining one to two phosphate units. This suggested that the chemosensorscan distinguish the degree of protein phosphorylation.

Recent advances in understanding of post-translational modification re-vealed that multisite phosphorylation, so-called hyper-phosphorylation, isa common mechanism for regulating protein functions in cell signaling path-ways. For example, platelet-derived growth factor receptor-β (PDGFR-β),a membrane-bound cytokine receptor, exposes multiple tyrosine residues inthe cytoplasmic domain. The binding of PDGF to the extracellular domaininduces auto-phosphorylation at these tyrosine residues, followed by recruit-ing of specific signaling proteins containing SH-2 domains, consequentlytriggering multiple signaling pathways. Thus, the development of artificialreceptors that can sense hyperphosphorylated proteins is currently anotherimportant topic in biology and biochemistry. Toward this end Hamachi et al.subsequently designed chemosensors 37, 38, in which two Zn-Dpa unitsfunctioning as phosphate binding sites were connected with bipyridine spac-ers [56, 57]. These chemosensors were able to bind a multiply-phosphorylatedprotein surface. The Zn-Dpa units were juxtaposed at an appropriately dis-tal position to enable the cross-linking interaction with two distinct phos-phate groups on a protein surface (Scheme 12). The binding abilities of thechemosensors with a series of bis-phosphorylated model peptides were evalu-


Scheme 12 Strategy for phosphorylated protein/peptide recognition by chemosensors 37,38 applying a cross-linking interaction with two phosphate groups on a protein surface

ated by circular dichroism (CD) spectral studies (Fig. 8), in which the α-helixcontent of the peptide was measured upon addition of the chemosensors.Sensors 37, 38 induced the α-helix formation of peptides having two phos-phoserine residues (pS-5,16, – 9,16, and – 12.16), whereas they did not affectthe secondary structure of the mono-phosphorylated peptide (pS-16). Thecomplexation process was also monitored by changes in the emission inten-sity. The fluorescence titration of 38 with pS-9,16 gave the affinity constant(Kapp = 2.0×106 M–1), the value of which is over 40-fold higher than thatof 38 with the mono-phosphorylated pS-16 peptide (Kapp = 4.8×104 M–1).This indicated that 38 can discriminate the number of phosphorylation sites


Fig. 8 CD (θ) value change of phosphopeptides upon addition of bipyridiyl-type chemo-sensor 37 ∼ 41

of peptides. More recently, it was found that similar binuclear type artificialreceptors could strongly bind phosphorylated CTD peptide by tight two-point interactions between Zn-Dpa sites and phosphates on the peptide. Sucha complexation disrupted phosphoprotein/protein interactions in a phospho-rylated CTD peptide and the Pin1 WW domain, a phosphoprotein-bindingdomain [57]. The strategy based on a small molecular disruptor that directlyinteracts with phosphoprotein is unique and should be promising in develop-ing a designer inhibitor for phosphoprotein-protein interaction.

5High Throughput Sensing System for Phosphate DerivativesUsing Semi-wet Sensor Arrays

5.1Molecular Recognition of Chemosensors in a Supramolecular Hydrogel

In order to achieve a complete understanding of the functions and roles ofphosphate derivatives in biological systems, it is necessary to collect datasimultaneously for a large range of different phosphorylation events by ex-haustive analysis. Rapid and high throughput methods based on chemosen-sors are expected to significantly contribute to determining the phosphoryla-tion level of biological samples. Similar to DNA arrays [58], a chemosensorarray immobilizing a number of chemosensors on a surface of solid sub-strates is a promising candidate to carry out convenient and high-throughputsensing. To establish the chemosensor array, the effective immobilizationof sensor molecules on a solid support is regarded to be of primary im-


portance. Hamachi et al. recently proposed a unique chemosensor arrayutilizing a supramolecular hydrogel as a matrix in which the chemosen-sors were non-covalently immobilized in the water-rich environment ofthe gel, while practically retaining the original molecular recognition func-tions [59, 60].

An artificial glycolipid mimic GalNAc-suc-glu-(O-methyl-cyc-hexyl)2 spon-taneously formed into supramolecular fibers based on a bimolecular layerstructure in water (Fig. 9) [61–64]. The entanglement of the fibers gavea transparent hydrogel consisting largely of aqueous media and well-developed hydrophobic domains. Chemosensor 36 was non-covalently em-bedded in the hydrogel and showed binding affinity/selectivity toward phos-phate derivatives and consequent fluorescence responses comparable to thoseobserved in aqueous solution [59].

Fig. 9 Schematic representation of the hierarchal molecular assembly used to forma supramolecular hydrogel. The artificial glycolipid mimic GalNAc-suc-glu-(O-metyl-cyc-hexyl)2 forms incipient nano-fibers based on a bimolecular layer structure. Such fiberscontain extensive hydrophobic domains in their cores with oriented saccharide arrays ex-posed at the interfaces. The incipient nano-fibers are bundled to give thicker fibrils whoseentangling results in the formation of a hydrogel

5.2Use of Hydrophobic Micro-domains of Supramolecular Hydrogel Fibersfor Discrimination of Phosphate Derivatives

In addition to the uses above, the elaborate utilization of hydrophobic nano-fiber domains in the supramolecular hydrogel allowed more selective dis-crimination of the Zn-Dpa-based chemosensors among various phosphate


derivatives [60]. The chemosensor 42, newly synthesized by Hamachi andcoworkers, had two Zn-Dpa binding sites and an environmentally sensitivefluorophore (dansyl). After the chemosensor was embedded in the supra-molecular hydrogel, solutions including various anions were placed on it. Asshown in Fig. 10, roughly three patterns of the fluorescence spectral changewere observed: (1) the emission intensity of 42 (at 512 nm) increased witha blue shift in the case of the relatively hydrophobic PhP, (2) the inten-sity decreased with a red shift of the emission for the relatively hydrophilicATP, phosphate and phospho-Tyr, (3) addition of phospho-diesters and non-phosphate anions caused no fluorescence changes. Fluorescence titration withATP (Fig. 10c) in the gel spot showed that the emission gradually decreasedwith the concurrent red-shift of the emission maximum. In contrast, thereverse type of spectral change was observed when PhP was added to thehydrogel spot containing 42 (Fig. 10d). From these spectral changes, the bind-ing constants were calculated to be 1.8×105 M–1 for ATP and 7.2×103 M–1

for PhP. Interestingly, such fluorescence change never occurred in homo-

Fig. 10 a Schematic illustration of chemosensor redistribution upon binding to a hy-drophobic or hydrophilic phosphate derivative between the hydrophobic hydrogelnanofiber and the hydrophilic cavity. The three patterns of change observed in thefluorescence spectra of the hydrogel containing the dansyl-appended receptor 42.b Fluorescence spectral change of 42 (60 µM, red line) embedded in the hydrogel uponaddition of PhP, ATP, phosphate, and phospho-Tyr. The emission intensity of 42 in-creases with the blue shift for PhP, whereas the intensity decreases with the red shift forATP, phosphate, or phospho-Tyr. (c–d) Fluorescence spectral change and the fluorescencetitration plots (inset) of 42 embedded in the hydrogel upon addition of ATP or PhP, re-spectively: [42] = 2 µM, [ATP] = 0–60 µM for (e), [42] = 6 µM, [PhP] = 0–600 µM for(f) at λex = 322 nm


geneous aqueous solutions, indicating that the hydrophobic domain of thepresent supramolecular hydrogel is crucial for the phosphate-induced change.Most significantly, the pattern of fluorescence response depended on the hy-drophilicity of the phosphate derivatives. Specifically, the strongly hydrophilicATP induced a red-shift in the emission of 42 with reduced intensity, whereasthe rather hydrophobic PhP caused an emission increase with a blue shift.These spectral changes imply that the microenvironment of the 42-ATP com-plex is more hydrophilic than that of 42 itself, whereas the 42-PhP complexlocates in the more hydrophobic microenvironment relative to the original 42.Using the unique semi-wet hydrogel array, a variety of phosphate anion speciesmay be rapidly and conveniently discriminated from each other by both theemission intensity change and the wavelength shift as shown in Fig. 11.

Fig. 11 Photo images of the sensing patterns of semi-wet molecular recognition (MR)chips of the hydrogel of GalNAc-suc-glu(O-metyl-cyc-hexyl)2 containing (a) 36 (40 µM),(b) 42 (60 µM)

6Summary

Phosphates and their derivatives are abundant in biological systems includinginorganic phosphates, small molecular organic phosphates such as ATP, phos-pholipids and phosphorylated organic intermediates, and phosphate deriva-tives having large molecular weight such as DNA and phospho-peptides andproteins. It is now clear that these derivatives play distinct roles in biologi-


Fig. 12 a Construction of the hybrid biosensor and fluorescence sensing for the doublyphosphorylated peptide. b Structure of the chemosensor unit on the hybrid biosensor.c Amino acid sequences of the WW domain and its mutant

cal phenomena under aqueous conditions, depending on their correspondingforms. Therefore, development of sophisticated artificial chemosensors thatpossess high binding affinity and selectivity toward the target phosphate iskeenly desired in order to significantly contribute the understanding of theroles of phosphorylation in complex biological systems. Although many ar-tificial chemosensors have been developed in the past two decades, severalproblems still remain to be overcome. In particular, clear discriminationamong the phosphate anion families, such as phosphates, phosphate esters,and ATP derivatives, etc., is thus far quite difficult and remains challenging.In addition, molecular recognition in aqueous systems is generally difficultto manage. Fundamental research in many areas is required for the rationaldesign of efficient artificial chemosensors. Most recently, Hamachi et al. pro-posed a hybrid system of artificial sensors with a biological receptor scaffoldto achieve improved selectivity among various phosphates. They successfullyconstructed a unique biosensor in which phosphoprotein-binding peptideWW domains were attached to Zn-Dpa units on the side chains (Fig. 12) [65].Such pattern recognition and detection may also be promising for the pre-cise discrimination between various phosphates as proposed by Anslyn [66]and Hamachi [67]. More creative ideas and concepts are required for progressin this growing area because the design and synthesis of novel chemosen-sors is envisioned to have a great impact on both basic research and practical,diverse applications in the biological, diagnostic and medicinal fields.

References

1. Fraústo da Silva JJR, Williams RJP (1991) The Biological Chemistry of the Elements.Clarendon Press, Oxford

2. Reitz AB (1990) Inositol Phosphates and Derivatives. American Chemical Society,Washington DC


3. Woodgett J (ed) (2000) Protein Kinase Function. Oxford University Press, New York4. Sefton BM, Hunter T (1998) Protein Phosphorylation. Academic Press, New York5. Lander SE et al. (2001) Nature 409:8606. Johnson LN, Lewis RJ (2001) Chem Rev 101:22097. Canagarajah BL, Khokhlatchev A, Cobb MH, Goldsmith EJ (1997) Cell 90:8598. Yaffe MB (2002) Nat Rev 3:1779. Yaffe MB, Elia AEH (2001) Curr Opin Cell Biol 13:131

10. Holmberg CJ, Tran SEF, Eriksson JE, Sistonen L (2002) Trends Biochem Sci 27:61911. Labadi I, Jenei E, Lahti R, Lönnberg H (1991) Acta Chem Scand 45:105512. Nakai C, Glinsmann W (1979) Biochemistry 16:563613. Huston ME, Akkaya EU, Czarnik AW (1989) J Am Chem Soc 111:873514. Vance DH, Czarnik AW (1994) J Am Chem Soc 116:939715. Kimura E (1985) Top Curr Chem 128:11316. Hosseini MW, Lehn JM (1987) Helv Chim Acta 70:131217. Hosseini MW, Blacker AJ, Lehn JM (1990) J Am Chem Soc 112:389618. Aguilar JA, García-España E, Guerrero JA, Luis SV, Llinares JM, Miravet JF, Rami-

rez JA, Soriano C (1995) J Chem Soc Chem Commun 21:223719. Bazzicalupi C, Bencini A, Bianchi A, Fusi V, Giorgi C, Granchi A, Paoletti P, Valtan-

coli B (1997) J Chem Soc Perkin Trans 2 4:77520. Baudoin O, Gbonnet F, Teulade-Fichou MP Vigneron JP, Tabet JC, Lehn JM (1999)

Chem Eur J 5:276221. Beer PD, Gale PA, Chen Z (1999) Coord Chem Rev 185–186:322. Beer PD, Cadman J (2000) Coord Chem Rev 205:13123. Beer PD, Gale PA (2001) Angew Chem Int Ed 40:48624. Beer PD, Chen Z, Drew MGB, Kingston J, Ogden M, Spencer P (1993) J Chem Soc

Chem Commun 24:104625. Beer PD, Chen Z, Drew MGB, Johnson AOM, Smith DK, Spencer P (1996) Inorg Chim

Acta 246:14326. Beer PD, Cadman J (1999) New J Chem 23:34727. Cotton FA, Hazen EE Jr, Legg MJ (1979) Proc Natl Acad Sci USA 76:255128. Kim EE, Wyckoff HW (1991) Clin Chim Acta 186:17529. Metzger A, Anslyn EV (1998) Angew Chem Int Ed 37:64930. Niikura K, Metzger A, Anslyn EV (1998) J Am Chem Soc 120:853331. Suzanne LT, Anslyn EV (2003) J Am Chem Soc 125:1480732. Schneider SE, O’Neil SN, Anslyn EV (2000) J Am Chem Soc 122:54233. Kato Y, Conn MM, Rebek J Jr (1994) J Am Chem Soc 116:327934. Sabo L, Diederich F, Gramlich V (2000) Helv Chim Acta 83:9335. Wilcox DE (1996) Chem Rev 96:243536. Lipscomb WN, Sträter N (1996) Chem Rev 96:237537. Kimura E (2000) Curr Opin Chem Biol 4:20738. Aoki S, Kimura E (2002) Rev Mol Biotechnol 90:12939. Koike T, Kimura E (1991) J Am Chem Soc 113:893540. Fujioka H, Koike T, Yamada N, Kimura E (1996) Heterocycles 42:77541. Kimura E, Aoki S, Koike T, Shiro M (1997) J Am Chem Soc 119:306842. Aoki S, Zulkefeli M, Shiro M, Kohsako M, Takeda K, Kimura E (2005) J Am Chem Soc

127:912943. Aoki S, Kimura E (2000) J Am Chem Soc 122:454244. Ojida A, Park S, Mito-oka Y, Hamachi I (2002) Tetrahedron Lett 43:619345. Lee DH, Kim SY, Hong JI (2004) Angew Chem Int Ed 43:477746. Lee DH, Im JH, Son SU, Chung YK, Hong JI (2003) J Am Chem Soc 125:7752


47. Han MS, Kim DH (2003) Bioorg Med Chem Lett 13:107948. Han MS, Kim DH (2002) Angew Chem Int Ed 41:380949. Lee HN, Xu Z, Kim SK, Swamy KMK, Kim Y, Kim SJ, Yoon J (2007) J Am Chem Soc

129:382850. Ojida A, Nonaka H, Miyahata Y, Tamaru S, Sada K, Hamachi I (2006) Angew Chem

Int Ed 45:551851. Ojida A, Miyahara Y, Wongkongkatep J, Tamaru S, Sada K, Hamachi I (2006) Chem

Asian J 1:55552. Mizukami S, Nagano T, Urano Y, Odani A, Kikuchi K (2002) J Am Chem Soc 124:392053. Ojida A, Mito-oka Y, Inoue M, Hamachi I (2002) J Am Chem Soc 124:625654. Ojida A, Mito-oka Y, Sada K, Hamachi I (2004) J Am Chem Soc 126:245455. Ojida A, Kohira T, Hamachi I (2004) Chem Lett 33:102456. Ojida A, Inoue M, Mito-oka Y, Hamachi I (2003) J Am Chem Soc 125:1018457. Ojida A, Inoue M, Mito-oka Y, Tsutsumi H, Sada K, Hamachi I (2006) J Am Chem Soc

128:205258. Shena M, Shalon D, Davis RW, Brown PO (1995) Science 270:46759. Yoshimura I, Miyahara Y, Kasagi N, Yamane H, Ojida A, Hamachi I (2004) J Am Chem

Soc 126:1220460. Yamaguchi S, Yoshimura I, Kohira T, Tamaru S, Hamachi I (2005) J Am Chem Soc

127:1183561. Kiyonaka S, Snikai S, Hamahci I (2003) Chem Eur J 9:97662. Kiyonaka S, Sugiyasu K, Shinkai S, Hamachi I (2002) J Am Chem Soc 124:1095463. Tamaru S, Kiyonaka S, Hamachi I (2005) Chem Eur J 11:729464. Kiyonaka S, Sada K, Yoshimura I, Shinkai S, Kato N, Hamachi I (2004) Nat Mater 3:5865. Anai T, Nakata E, Koshi Y, Ojida A, Hamachi I (2007) J Am Chem Soc 129:623266. Lavigne JL, Anslyn EV (2001) Angew Chem Int Ed 40:311867. Koshi Y, Nakata E, Yamane H, Hamachi I (2006) J Am Chem Soc 128:10413


Anions and π-Aromatic Systems.Do They Interact Attractively?

Pablo Ballester

ICREA, Pg. Lluís Companys 23, 08010 Barcelona, [email protected]

Present address:Institute of Chemical Research of Catalonia (ICIQ), Avda. Països Catalans 16,43007 Tarragona, Spain

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

2 Theoretical Investigations of the Interaction of Anions (Halides)with π Aromatic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

3 Experimental Evidence of the Anion-π Interaction . . . . . . . . . . . . . 1523.1 Solution and Related Crystallographic Studies . . . . . . . . . . . . . . . . 1523.2 Further Crystallographic Evidence of Anion-π Interactions . . . . . . . . . 163

4 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

Abstract Hydrogen bonds and charge–charge interactions have been widely used (eitheralone or in combination) in the design of efficient and selective synthetic receptors foranions. Intuitively, the interaction between anions and π-aromatic systems is associatedwith a repulsive force. Consequently, for many years, anion-π interactions have beencompletely neglected as favorable non-covalent interactions for the construction of ef-ficient anion receptors. Recently, however, theoretical studies indicate the existence ofan attractive interaction between anions and certain type of π-acidic aromatic systems.These theoretical studies together with the observation of supramolecular complexesin the solid state in which anions are included in deep aromatic cavities have encour-aged an in depth study of the anion-π interaction. Nowadays, the anion-π interactionis considered by several researchers a potential non-covalent interaction for the designof anion receptors. This chapter will provide an overview of recent theoretical investiga-tions, performed since 2002, on anion binding involving six-membered aromatic rings.A series of experimental studies, carried out since 2004, also evidencing the existence ofa possible attractive interactions betweens anions and six-membered aromatic moietiesof host-guest systems in the solid state and in solution is also discussed.

Keywords Anion-π interactions · Anion coordination · Anion receptors · Anions ·Host–guest systems · Supramolecular chemistry

AbbreviationsHB Hydrogen bondRHF Restricted Hartree-Fock

128 P. Ballester

BSSE Basis set superposition errorMEP Molecular electrostatic potentialMP2 Møller PlessetDFT Density functional theoryZPE Zero point energySI International system of unitsPCM Polarized continuum solvent modelAIM Atoms in moleculesGMIPp General molecular interaction potential with polarizationMIPp Molecular interaction potential with polarizationEPS Electrostatic potential surfacesSAPT Symmetry-adapted perturbation theoryNICS Nucleus-independent chemical shiftRI-MP2 Resolution identity MP2DFT Density functional theoryTCB 1,2,4,5-TetracyanobenzeneCSD Cambridge Structural Databasebptz 3,6-bis(2′-pyridyl)-1,2,4,5-teatrazinebppn 3,6-bis(2′-pyridyl)-1,2-pyridazine

1Introduction

The driving forces controlling the formation of non-covalent complexes be-tween molecular or ionic species are quite varied and include the so-calledhydrogen-bonding, stacking, hydrophobic, charge-transfer, van der Waalsand ionic interactions. There is good experimental evidence of the domin-ant role of electrostatic in many of those intermolecular interactions [1]. Anelectrostatic or Coulombic interaction by definition requires no adjustment ofthe electronic properties around the ion or the molecule. Consequently, ion–ion, ion–dipole, dipole–dipole, ion–quadrupole and quadrupole–quadrupoleinteractions can be considered as mainly electrostatic non-covalent in-teractions. A “conventional” hydrogen bond (HB) may be represented asD – H· · ·A whereby D (donor) and A (acceptor) are both electronegativeatoms (usually N and O). Hydrogen bonds are a prominent and typical rep-resentation of a dipole–dipole interaction. However, HBs are not necessarilyrestricted to N and O, but may involve other electronegative donor and ac-ceptor functionalities. For example, a hydrogen bond of the type N – H· · ·X–,where X– is an anion, is termed an ionic hydrogen bond. In fact, moleculescontaining polarized N – H functionalities which behave as H-bond donorstowards anions are widely used as receptors for recognition and sensing pur-poses in aprotic solvents [2]. The coordination of the anion by the N – Hfunctionality is an ion–dipole interaction clearly dominated by electrostatics.The covalent bonding contribution to the interaction can be considered as in-significant. Equation 1 represents the electrostatic potential created by a point

Anions and π-Aromatic Systems. Do They Interact Attractively? 129

charge q, at a distance r from the charge. Equation 2 corresponds to the elec-trostatic potential at a distance r from the center of a dipole of q chargeshaving a 2d separation in space. θ is the value of the angle defined by thelines that connect the center of the dipole with the point at r distance and thetwo point charges. As deduced from Eqs. 1 and 2, the ion–dipole interactionhas the same dependence on the dielectric environment as the interaction be-tween fully charged partners but is appreciably weaker on an absolute scalefalling off with distance more steeply.

V(r) =1

4πξ0

qr

(1)

V(r, θ) =1

4πξ0

2dqr2 cos θ (2)

The main virtue of the ion–dipole compared to the ion–ion interaction, fromthe viewpoint of host–guest chemistry, is its directionality (i.e., the energy ofthe interaction of an ion and the dipole depends on their mutual orientationdefined by the θ angle). This translates into a fundamental property to achieveselective recognition of anions using hydrogen-bonding receptors, that is, thetopology of the targeted anion should be considered when deciding the place-ment of the hydrogen bond donors in the receptor’s structure. Other virtuesof hydrogen bonding in the design of abiotic receptors for anions derive fromits electroneutrality, from its capacity to form simultaneously several bondinginteractions with the anion and from the rich chemistry available to incorpo-rate this functionality into the molecular scaffold.

Scheme 1 Preparation of the diazabicycloalkanes and ion pairing in acidic aqueous media

130 P. Ballester

A clear analogy exists in the supramolecular chemistry of anions and thecoordination chemistry of main group cations and transition metals [3, 4]1.Fundamental design principles of cationic receptors (i.e., macrocyclic effectand preorganization) have been directly applied to the supramolecular chem-istry of anions. Many hosts capable of forming ion-pairs with anions havebeen prepared by simply inverting the electronic nature of macrocyclic andmacrobicyclic receptors by protonation of a suitable Lewis-basic center. Inprinciple at least, the non-protonated receptor could be suitable for cationbinding. This was the case of the serendipitous discovery of the first recep-tor of a guest chloride emerged during the preparation of a macrobicycliccompound having two converging tertiary amines (Lewis-basic centers).

The protonation of the Lewis-basic centers afforded two protonated am-monium groups (Lewis-acid centers) capable of maintaining the sphericalchloride anion within the receptor’s cavity through the formation of twoion-pair-reinforced hydrogen-bonding interactions (salt-bridge) [5–7]. In thecase at hand, the protonation of the amine groups has a twofold effect: 1) addspositive charge to the receptor and 2) converts a tertiary amine into an excel-lent hydrogen bond donor (ammonium). Nevertheless, the addition of posi-tive charge to an anionic receptor, even in the absence of additional hydrogen-bonding influences, enhances anion binding. The great majority of positivelycharged organic receptors for anions are based on nitrogen compounds. Theincorporation of cationic metal centers into the molecular scaffold of the re-ceptor represents an interesting alternative to the introduction of positivecharge by protonation or quaternization of the amine functionality. Further-more, when the coordination features of the metal are partially fulfilled theycan be exploited to assist the electrostatic binding with metal-anion covalentbonding. The major drawback of using positively charged anion receptors isthe internal competition established between the anionic target and the coun-teranion of the cationic host. This is an unavoidable disadvantage of positivelycharged receptors when compared with electroneutral hosts for anions.

The “anticrown chemistry” term has been coined to describe the way inwhich many electroneutral hosts were built to recognize anions. That is the in-corporation of multiple and convergent Lewis-acid centers into preorganizedcyclic molecular scaffolds. The Lewis-acid centers, usually transition metals orelements like B, Si, Sn or Hg, should expose their electron-deficient sites forinteraction with the lone electron pairs of the anions. The hydrogen-bondinginteraction of an anion with an amide N – H mentioned above has also beenwidely used in the preparation of neutral receptors for anionic guests [8].

While the above-mentioned non-covalent interactions (hydrogen bonds,charge–charge) have been widely used in the design of efficient and selectivesynthetic receptors for anions either alone or combined, the use of attrac-

1 The researchers working in the area of anion recognition having an inorganic chemistry back-ground usually refer to the field as anion coordination chemistry


tive interactions between negative charges and aryl rings towards the sameend has remained dormant in the literature since they were first noticed byScheneider in 1993 [9].

Recently, the studies into the anion-π interaction have intensified andgained interest among the scientific community. This is due, on the onehand, to the publication in 2002 of several papers dedicated to exploring thephysical nature of the interaction of anions with aromatic compounds usingelectronic structure methods and, on the other hand, to the observation since2004 of supramolecular complexes in the solid state in which anions are in-cluded in deep aromatic cavities. Furthermore, the recently noticed fact thatthe orientation of negatively charged groups [10] and lone pairs [11] abovethe plane of an aromatic is a frequently occurring structural motif in biopoly-mers has added additional interest to the subject.

At first sight, a stabilizing effect for the interaction between a pair ofelectrons or a negative charge and the face of a π system seems to be coun-terintuitive. Not surprisingly, simple modelling studies indicated a repulsiveinteraction between anions and a benzene ring. It is now clear that non-covalent interaction between aromatics and positively charged cations, thecation-π interaction, are prominent in a wide variety of systems and shouldbe considered as an important and general binding force [12]. The cation-π interactions are expected simply from electrostatic arguments. In fact,a simple electrostatic model based on the visual inspection of electrostaticpotential surface of the aromatic ring has been used to rationalize the majorbinding trends of the cation-π interaction. In line with the principles outlinedabove for the preparation of anion receptors through the inversion of theelectronic character of suitable Lewis-basic centers (amine to ammonium),a possible explanation for the observed geometry that points lone pairs orplaces negative charges into the face of the π system has to do with an inver-sion of the electronic character of the aromatic ring [13].

Theoretical investigations have clearly established the existence of bind-ing interactions between a negatively charged species and an electron-poorπ system. The theoretical results have encouraged the experimental obser-

Fig. 1 Inversion of the electronic character of Lewis-basic centers: a the lone pair ofan amine is a good binding site for a cation while the protonated ammonium group iscomplementary to anions; b the π-system of an electronically rich aromatic compoundestablishes attractive interactions with cations (cation-π interaction) the introductionof electro-withdrawing substituents produces the depletion of electronic density of theπ-system giving rise to attractive interaction with anions

132 P. Ballester

vation and evaluation of this new non-covalent interaction. Several authorshave already claimed the potential use of anion-π interactions in the prep-aration of selective receptors for anions. Solid-state structural examples ofanions included in deep aromatic cavities of supramolecular complexes havebeen used as evidences for the existence of a substantial anion-π attractiveinteraction. Few studies, however, have attempted to quantify the strength ofthe interaction in solution. Since the anion-π interaction seems to be a direc-tional non-covalent interaction it has potential and unexplored applicationsin the design of selective neutral receptors for anion recognition. This chap-ter will provide a short overview of recent theoretical investigations on anionbinding involving six-membered aromatic rings followed by a summary ofexperimental studies evidencing or not the existence of attractive interactionsbetween anions and six-membered aromatic moieties of host–guest systemsin the solid state and in solution2.

2Theoretical Investigations of the Interaction of Anions (Halides)with π Aromatic Systems

Since 2002 several groups have studied in detail the interaction of anionswith electron deficient aromatic rings using theoretical methods. These stud-ies can be considered to derive from earlier fundamental work carried outby Dougherty [14], Besnard [15] and Alkorta [16] on the interaction betweenhexafluorobenzene and the heteroatom in molecules such as H2O, HCN andHF wherein the negative end of the dipole is directed toward the π-systemand aligned with the C6 axis of the ring.

In particular, Alkorta et al. [16] studied the interaction of C6F6 withHF using both MP2 and hybrid DFT methods (B3LYP) with 6-31-G∗∗ and6-311++G∗∗ basis sets. The authors considered that these methods are moreaccurate for weak complexes than methods without electron correlation.A shortening effect in the calculated distance value between the interactingatom and the centroid of the aromatic ring is clearly observed by the inclusionof electron correlation (MP2 and B3LYP methods). Thus, for the minimumstructure of the C6F6· · ·FH complex the computed distance is 3.076 A whenusing the RHF method and the 6-31G∗∗ basis set and 2.858 A and 2.814 A,respectively, when the B3LYP/6-31G∗∗ and MP2/6-31G∗∗ methodologies areapplied. The use of a more extended basis set B3LYP/6-311++G∗∗ producesa larger elongation of the same distance to 3.127 A while the calculated bind-ing energy at this level of theory and basis set for the C6F6· · ·FH complex is

2 Complexes in which the anion is one component of an aromatic ring sandwich complex are out-side of the scope of this chapter. Likewise, examples for the interaction of anions with five and sevenmembered aromatic rings will not be presented


Fig. 2 Schematic representation of the geometry of the complexes of water, HCN, and HFwith C6H6

Fig. 3 Calculated MP2/6-31+G∗ MEP surfaces for a 1,3,5-triazine, b trifluoro-1,3,5-tri-azine, and c hexafluorobenzene. Electrostatic potential surfaces energies range from –19(red) to +19 (blue) kcal/mol for 1,3,5-triazine, –39 (red) to +39 (blue) kcal/mol fortrifluoro-1,3,5-triazine, and –24 (red) to +24 (blue) kcal/mol for hexafluorobenzene

∆EBSSE = 1.23 kcal mol–13. The interaction energy of the C6F6· · ·FH complexis similar to the one observed in the formation of weak hydrogen bonds.

Several years later, in 2002, Mascal et al. [17], Alkorta et al. [18] andFrontera, Deyà et al. [19] reported almost simultaneously the theoreticaldemonstration of the existence of an attractive interaction between a formallycharged negative species and the π-system of an aromatic ring.

Mascal et al. [17] performed ab initio orbital calculations at the MP2level of theory with the 6-31+G∗ basis set, including counterpoise correc-tions for the basis set superposition error (BSSE), for the interaction ofboth 1,3,5-triazine and trifluoro-1,3,5-triazine with chloride and fluoride. Themolecular electrostatic potential (MEP)4 maps of 1,3,5-triazine and trifluoro-1,3,5-triazine clearly indicate an area of positive density concentrated on theC3 rotational axis passing through the center of the hexagonal ring and beingperpendicular to the plane of the aromatic ring, similar to that observed forC6H6 on the C6 rotational axis.

Optimization from geometries that place the chloride near the area ofpositive density found on the C3 rotational axis of 1,3,5-triazine demon-

3 BSSE stands for Basis Set Superposition Error and the interaction energies are corrected for thisinherent error4 The MEP maps have been used for long time as a tool to identify both nucleophilic and elec-trophilic regions in a molecule

134 P. Ballester

strated the existence of an attractive interaction between the chloride and theπ-system. The optimized structure obtained for the chloride-1,3,5-triazinecomplex positioned the chloride anion exactly on the C3 rotational axis of the1,3,5-triazine (Φ = 90) and at a distance of 3.2 A of the aryl centroid. The factthat the structure of the complex was a minimum was confirmed by calculat-ing the corresponding frequencies (no negative vibrational frequencies werefound). The MP2/6-31+G∗ energy calculated for the geometry of the chloride-aryl-centroid complex optimized using the same level of theory and basis setwas ∆EBSSE

0 K =– 4.8 kcal mol–1.Mascal et al. [17] also located another geometry for the complexation of

the chloride and bromide anion with 1,3,5-triazine. This geometry is alsoa minimum and involves a C – H· · ·Cl– hydrogen bond. The formed hydro-gen bond shows good geometrical characteristics, having a C· · ·Cl– distanceof 3.4 A and a C – H· · ·Cl– angle of 180◦. The calculated MP2/6-31+G∗ en-ergy of this interaction geometry optimized at the same level of theory andbasis set was ∆EBSSE

0 K =– 7.4 kcal mol–1, that is 2.6 kcal mol–1 more stable thanchloride-aryl centroid complex discussed above. The authors clearly state thatalthough the participation of ions in gas phase chemistry yields interactionenergies which seem to be exaggerated when compared to solution values,the reported energies of these interactions may still be relevant in the inte-rior of a receptor. The authors used the trifluoro-1,3,5-triazine as a modelto evaluate the extent to which further electron withdrawal in the triazine

Fig. 4 Side and top views of the minimized structures at the MP2/6-31+G∗ level of a thetriazine chloride aryl centroid complex and b the triazine chloride hydrogen-bondingcomplex. The triazine and the chloride are shown in a scaled ball-and-stick representation


Fig. 5 Side view and top view of the “attack” structure for the fluoride 1,3,5-triazinecomplex

Fig. 6 Molecular structures of the perfluoroaromatic compounds studied by Alkortaet al. [18]

system would enhance complexation. Higher interaction energy and evenstronger chloride-aryl centroid complexes were located using the trifluoro-1,3,5-triazine (∆EBSSE

0 K =– 14.8 kcal mol–1, r = 3.0 A and θ = 90◦). In fact, thishigh energy and close interaction distance computed for the Cl–-trifluoro-1,3,5-triazine complex is comparable to those of the potassium cation-π com-plexes of benzene calculated at the same level of theory [20, 21]5.

The corresponding fluoride-aryl centroid complexes with 1,3,5-triazineand trifluoro-1,3,5-triazine are characterized by the presence of one negativevibrational frequency and represent a shallow inflection point on a surfaceconnecting to a second geometry, the so-called “attack” structure which isa real minimum structure. This structure was suggestive of a reactant com-plex with the nucleophilic fluoride anion “attacking” one of the carbon atomswith close to the Bürgi-Dunitz trajectory [22, 23]. The considerable stabiliza-tion energy of the “attack” structure (∆EBSSE

0 K =– 18 kcal mol–1, r = 1.5 A andθ = 106.6◦) together with the distance values for the C· · ·F interaction and thelengthening of the adjacent C – N bonds points to a strong σ -complex.

5 Na+· · ·C6H6 interaction E0 = –15.0 kcal mol–1, r=2.84 A. K+· · ·C6H6 interaction E0=–18.3 kcal mol–1

136 P. Ballester

The calculated energies for these interactions fall off sharply with increas-ing polarity of the medium. The simple comparison of uncorrected MP2single point energies of chloride-aryl centroid complex with triazine, usingTomasi’s Polarized Continuum solvent Model (PCM), gas phase (–8.5), hep-tane (–0.4), chloroform (2.1), ethanol (2.8), and finally water (3.2 kcal mol–1)suggest that the practical manifestation of these forces will be most likely inthe context of anion containment.

Alkorta et al. [18] studied the complexes formed by a variety of anionswith perfluoro derivatives of benzene, naphthalene, tiophene and furan usingDFT (B3LYP/6-31++G∗∗) and MP2 (MP2/6-31++G∗ and MP2/6-311++G∗∗)ab initio methods.

The minimum structures of hexafluorobenzene with chloride and bromideshow the anion interacting with the π-system with an anion-aryl centroidgeometry. In both cases, a C6v symmetry was assumed during the optimiza-tion procedure. However, the C6H6· · ·F– complex structure that locates theanion on the C6 symmetry axis and on top of the aromatic ring (r = 2.5 A)shows two degenerate imaginary frequencies. In this case, the minimumstructure corresponds to an “attack” geometry similarly to the interaction offluoride with 1,3,5-triazine discussed above.

The calculated interaction energy values for the anion-aryl centroid geom-etry range between –18.7 and –12.19 kcal mol–1, which is comparable, asmentioned above, to the interaction energy of some benzene-cation com-plexes [20, 21].

The centroid-to-anion distance varies from 2.554 to 3.230 A (see Table 1).The effects of the complexation of the anion to the C6F6 molecule are short-ening of the C – C bonds (0.004 A) and a lengthening of the C – F bonds(0.006 A). On complexation, the fluorine atoms move slightly towards theanion and the plane formed by the carbon atoms is about 0.02 A furtheraway from the anion than that formed by the fluorine atoms. The hexafluoro-benzene molecule adopts a “cup” conformation.

Non-covalent interactions have been characterized using Bader’s theory of“Atoms In Molecules” (AIM) [24] which has been used successfully to under-

Table 1 Corrected interaction energies (kcal mol–1) and C6F6 centroid-anion distance(r, A) calculated at the MP2/6-31++G∗ level together with the electron densities and theirLaplacian (au) for the calculated bond critical point

Complex ∆EBSSE r ρ ∇2ρ

MP2/6-31++G∗

C6F6 · · · F– –18.63 2.554 0.0118 0.0470C6F6 · · · Cl– –12.76 3.159 0.0080 0.0238C6F6 · · · Br– –12.19 3.230 0.0087 0.0240


Fig. 7 Schematic representation of the location of a bond (red), a ring (blue) and the cagecritical point (black) originated from the interaction of hexafluorobenzene with fluoride

stand conventional [25] hydrogen bonds and cation-π interactions [26]. TheAIM analysis of the electron density of these complexes (C6F6· · ·X–, X= F, Cl,Br) carried out by Alkorta et al. [18] indicated the formation of six degeneratebond critical points (bcp) between the anion and each of the carbon atoms ofthe C6F6 molecule in a similar way to what was observed for the C6F6· · ·X – Yneutral complexes [16] and the C6F6· · ·Na+ complex [20]. In addition, six-ring critical points (rcp) and one cage critical point (ccp) are generated. Thercp connect the anion with the middle of the C – C bond. The ccp is locatedover the hexafluorobenzene molecules along the C6 axis, connecting the an-ion with the center of the ring. The quantitative values for ρ and ∇2ρ at thecps give a hint on the character and strength of the interaction. The values ofelectron density of the new bcp and its corresponding rcp are almost the samevalues. The positive and small value of the Laplacian of the bcps indicatesa depletion of the electron density, as is common in closed shell interactionslike those found in hydrogen bonds, ionic and cation-π interactions.

Table 2 Corrected interaction energy at the MP2/6-31++G∗ level (kcal mol–1) and electro-static, polarization and van der Waals contribution (kcal mol–1) to the interaction energycalculated using GMIPp method

Complex ∆EBSSE Eele Epol EvdW Et

C6F6 · · · Cl– –12.7 –11.8 –6.6 5.2 –13.2

138 P. Ballester

It is worth mentioning that the interaction energies calculated at theMP2/6-31++G∗∗ level of the perfluoronaphtalene complexes with anions arethe largest of all the study (∆EBSSE =– 17.31 (F–), –16.70 (Cl–) and –16.51(Br–) kcal mol–1) approximately 4.5 kcal mol–1 more than the energies of thecorresponding C6F6· · ·X– complexes (see Table 1). However, the anions are lo-cated over the C(4a)–C(8a) bond, that is, the anion interacts with one bond,not with one aromatic ring. In the optimized structure, the naphthalene ringbent away from the anion for the complexes with chloride and bromide.In this case, instead of the “cup” observed for hexafluorobenzene, the per-fluoronaphtalene adopts a “book” conformation pointing toward the anion.Alkorta et al. also analyzed the interaction energies using the general molecu-lar interaction potential with polarization partition (GMIPp) [27, 28] of allthe complexes of the aromatic compounds with chloride in order to calculatethe contribution of the electrostatic, polarization and van der Waals terms tothe overall interaction energy. The obtained results indicate that the polariza-tion energy is 50 to 100% of the electrostatic energy. The sum of the threeterms calculated with the GMIPp method provides an energetic value veryclose to the corrected interaction energy obtained at the MP2 level. Again, theobtained results for the contribution of the polarization and the electrostaticterm to the overall interaction are analogous to those found for the cation-πinteraction.

Frontera, Deyà et al. [19] also studied the interactions of several anionswith C6F6 using HF/6-31++G∗∗ and MP2/6-31++G∗∗ ab initio methods. Inall the complexes the anion is positioned over the aromatic ring along the C6rotational axis passing through the center of the hexagonal ring and beingperpendicular to the plane of the aromatic ring, that is, having an anion-aryl centroid geometry. Initially the geometries of all complexes were fullyoptimized at the HF/6-31++G∗∗ level. The authors reported the correspond-ing binding energies with and without basis set superposition error (BSSE)and zero-point vibrational energy (ZPE) corrections. Frequency calculationsat the same level confirmed that the structures are at their energy minimum.When they extended the calculations to the MP2/6-31++G∗∗ level, assum-ing C6v symmetry for the complexes C6F6· · ·X–, X= F, Cl, Br, the frequencycalculations gave either one or more imaginary frequencies. This problemwas solved by performing the geometry optimizations without imposingany symmetry constrains. In complete agreement with Alkorta’s report, theminimum energy complex found for the interaction C6F6· · ·F– corresponds tothe nucleophilic attack of the anion at one carbon atom. In general, they con-clude that the MP2-computed binding energies are more negative than the HFones and the equilibrium distances are shorter.

Frontera, Deyà et al. [19] also performed a topological analysis of thecharge-density distribution and properties of critical points in the complexesusing the AIM method providing an unambiguous definition of chemicalbonding between the anion and the π-system. These calculations were per-


Table 3 Interaction energies with (∆EBSSE, ∆EBSSE0 K kcal mol–1) and without (∆E,

kcal mol–1) the basis set superposition error (BSSE) and zero-point vibrational energycorrection (ZPE, 0 K) and equilibrium distances (r/A) at HF/6-31G++G∗∗ and MP2/6-31++G∗∗ (italics) levels of theory and selected electron-density topological properties forthe complexes of hexafluorobenzene with anions from reference [19]

Complex ∆E ∆EBSSE ∆EBSSE0 K r CPa n ρ ∇2ρ

C6F6 · · · F– –18.8 –18.1 –17.8 2.669 bond 6 0.01001 0.04200ring 6 0.00996 0.04226cage 1 0.00702 0.04106

C6F6 · · · Cl– –11.0 –10.8 –10.6 3.404 bond 6 0.00572 0.01637–18.0 –13.2 –12.9 3.155 ring 6 0.00571 0.01637

cage 1 0.00450 0.01820C6F6 · · · Br– –13.2 –9.3 –9.5 3.479 bond 6 0.00618 0.01612

–20.7 –12.4 –11.9 3.214 ring 6 0.00616 0.01613cage 1 0.00480 0.01887

a The electron density (ρ) and its Laplacian (∇2ρ) in atomic units at the critical points(CP) originated upon complexation are given, as well as, the number (n) of each type ofcritical points

formed by means of the program AIMPAC using HF/6-31++G∗∗ wavefunc-tion and the obtained results are completely coincident with the ones we havepreviously discussed based on Alkorta’s report. The physical nature of theanion-π interaction and the importance of the polarization were analyzed bycomputing its contribution to the total interaction energy using the molecularinteraction potential with polarization (MIPp) [29]. The MIPp is an improvedgeneralization of the molecular electrostatic potential (MEP) and was alsoused by Alkorta in the energetic analysis of the C6F6· · ·Cl–. In the MIPpcalculation three terms contribute to the total interaction energy: 1) an elec-trostatic term identical to MEP, 2) a classical dispersion-repulsion term, and3) a polarization term derived from perturbation theory. In the calculationF– ion was considered as a classical nonpolarizable particle. The electro-

Table 4 Contribution to the total interaction energy (kcal mol) calculated with MIPp forhexafluorobenzene interacting with F– at several distances (A) from the center of the ring

Distance Eele Epol Evdw Et

1.5 –36.59 –43.84 1047.82 967.382.0 –22.40 –24.89 119.05 71.722.5 –16.39 –13.93 13.44 –16.893.0 –12.66 –7.96 –0.50 –20.123.5 –9.90 –4.74 –0.83 –15.48

140 P. Ballester

Fig. 8 Schematic representation (top and side view) of the positive molecular quadrupolemoment of C6F6 as cylinders with positive charges on the ends and negative charges inthe center

static (Eele), polarization (Epol), van der Waals (Evdw), and total interactingenergies were calculated when a fluorine ion approaches the hexafluoroben-zene molecule perpendicular to the center of the aromatic ring. The obtainedresults point out the importance of the polarization component, which issimilar to the electrostatic term in the 2.0 to 3.0 A range, the equilibrium dis-tance for the fluoride aryl centroid complex is 2.6 A. The authors mention theimportance of the quadrupole moment for understanding intermolecular in-teractions of aromatic system but they do not elaborate on this issue in thiswork.

To date, most of the work reported on anion-π interactions using theor-etical methods is between anions and electron deficient aromatics rings, i.e.,hexaflurorobenzene, 1,3,5-trinitrobenzene, and 1,3,5-triazine. The electron-deficient aromatic rings are also called π-acidic aromatic rings and are char-acterized by having a permanent positive quadrupole moment value Qzz. Thevalue of the quadrupole moment is a measure of the distribution of chargewithin a molecule, relative to a particular axis. In six-membered aromaticrings, the Qzz quadrupole measures the distribution of charge relative to theC6 rotational axis passing through the center of the hexagonal ring and be-ing perpendicular to the plane of the aromatic ring. When the ring presentsa high degree of symmetry, one may relate the distribution of charge with re-spect to the axis perpendicular to the aromatic plain to that along the mainrotational axis and gain a quick characterization of the charge distribution inthe molecule [30]. The SI value of the electric quadrupole moment of hex-afluorobenzene is 31.7×10–40 C m2 (Qzz =+ 9.5 B); 1 B (Buckingham)6. Theschematic representation of such a molecule as two like positive charges,through which the main rotational axis (C6), passes, separated by the oppo-site, balancing negative charges lying perpendicular to the main rotationalaxis gives a clear picture of its molecular positive quadrupole value.

Topologically, the quadrupoles can be considered equivalent to d orbitals, asdipoles are to p orbital [12]. In particular, the Qzz quadrupole of six-membered

6 Debye suggested in the 1960s that the quantity 1 unit of charge distributed over 1 A2 be termedthe Buckingham


aromatic rings can be considered equivalent to a d2z orbital, having a non-

spherical charge distribution with regions of relative negative and positivecharges. Molecules like C6F6 having no net dipole have a defined charge dis-tribution, that is, the molecular quadrupole described above. Consequently,the Qzz molecular quadrupole of aromatics has the adequate spatial orienta-tion to be involved in the formation of non-covalent anion-π complexes inwhich the halide is located above the arene centroid (anion-aryl centroid geom-etry). The great majority of the theoretical studies on the anion-π interactiondeal with complexes displaying this type of geometry in the gas phase. Ithas been shown conclusively and elegantly by Alkorta [18], and Fronteraand Deyà [19] using the molecular interaction potential with polarization(MIPp) energetic partition scheme as discussed above that the two funda-mental components contributing to the stabilization energy of non-covalentanion-π complexes in the gas phase are electrostatic and anion-induced po-larization. The electrostatic component correlates well with the magnitudeof the Qzz of the aromatic ring [31]. In molecules having a very positiveQzz, the electrostatic contribution dominates the anion-π interaction but thepolarization energy is not negligible. For example, 27% of the total energy(Et =– 16.5 kcal mol–1) calculated using MIPp at the MP2/6-31++G∗ for theinteraction of trifluoro-1,3,5-triazine (Qzz =+ 8.23 B) with Cl– when the an-ion approaches perpendicular to the center of the aromatic ring is due topolarization energy (Epol =– 4.5 kcal mol–1). The electrostatic term for this in-teraction is the major component Eele =– 12.9 kcal mol–1 while the van derWaals term is almost negligible Evdw = 0.9 kcal mol–1. The computed molecu-lar polarizability of the trifluoro-1,3,5-triazine aromatic ring is α|| = 30.26 a.u.As the Qzz value diminishes, the term of the electrostatic energy (Eele)becomes less important. Thus, the contributions to the total interaction en-ergy (Et =– 6.6 kcal mol–1) of 1,3,5-triazine (Qzz =+ 0.9 B) with chloride are:Eele =– 2.2 kcal mol–1, Epol =– 4.1 kcal mol–1 and Evdw =– 0.3 kcal mol–1. It isworth to note that the computed molecular polarizability of the 1,3,5-triazinearomatic ring is α|| = 30.34 a.u. very similar to the fluoride derivative. This re-lationship corroborates the central role of the quadrupole moment value inanion-π interactions when the anion approaches the aromatic compound per-pendicular to the center of the aromatic ring. Since the molecular quadrupolemoment describes the electron density of the aromatics, and because a directcorrelation between the interaction energy of the anion binding of positive Qzzvalue and the aromatic Qzz value has been observed, it is reasonable to suggestthat the anion-π interaction is governed by electrostatics. Frontera and Deyàhave also demonstrated that the contribution due to polarization (Epol) to thetotal energy calculated with MIPp increases linearly with the computed mo-lecular polarizabilities of the compounds [31]. This becomes the predominantterm of the total interaction energy for compounds having Qzz values lowerthan 1 B. Overall, however, the contribution due to polarization (Epol) can beconsidered as almost constant with a value of 4–6 kcal/mol.

142 P. Ballester

Based on the results of the theoretical calculations, the binding energy ofthe anion-π interaction is not 100% electrostatic. In fact, as shown in Fig. 9the fraction of the total binding energy that is electrostatic varies consider-ably depending on the aromatic. However, the variation of the anion-π bind-ing energies is faithfully mirrored by the electrostatic term (plots a and d).To predict the trend in an anion-π interaction across a series of similar aro-matics, all with the same anion, i.e., chloride, it is enough to consider theelectrostatic term. It is for this reason that the visual inspection of electro-static potential surfaces (EPS), which are a good way to visualize the chargedistribution of aromatics and consequently the quadrupole moment, providesa simple and reliable guide to the relative strength of anion-π interactionacross a series of aromatics interacting with the same anion. The EPSs arealso useful to locate the computed minimum geometry of anion-aryl centroidcomplex stabilized by anion-π interaction, although this geometry is not al-ways observed in the experimental studies (vide infra). The simple consider-ation of the electrostatic term explains many trends derived from the theoret-

Fig. 9 Plot of the regressions between the quadrupole moments and molecular polariz-abilities to: a the electrostatic contribution, b polarization contribution and c van derWaals contribution of the total interaction energy calculated with MIPp for the com-pounds interacting with chloride at the minimum. d Plot of the regression between thequadrupole moments to the interaction energy at the MP2/6-31++G∗∗ level of theory forthe compounds interacting with chloride at the minimum. From reference [31]


ical studies although sometimes the major component of the binding energycould be “non-electrostatic”. It is worth to note that across a series of aromat-ics compound having quadrupole moment values from Qzz = 8.23 to 0.57 Bthe “non-electrostatic” component of the interaction energy (Epol + EvdW)makes a contribution of –3 to –7 kcal mol–1. The binding energy of the anion-π interaction decreases slightly with the increase of the ionic radius of theanions which is also consistent with an electrostatic model.

Kim et al. [32] also studied the nature of the anion-π interactions usingthe symmetry-adapted perturbation theory (SAPT) [33] to obtain a physi-cal interpretation of the interaction energy. In this method, the interactionenergy is expressed as a sum of perturbative corrections in which eachcorrection results from a different physical effect. The different intermole-cular terms obtained from this method can be summarized in electrostatic,exchange-repulsion, induction, and dispersion contribution. The authorsshow that for different halogen complexes with several π systems the elec-trostatic term follows the same trend as the total interaction energy. Thistendency can be explained by taking in consideration the fact that both dis-persion and induction energies can be ascribed, to a large extent, to theinteraction of the molecular orbitals of the anion and the π system. The at-tractive dispersion and induction energies increase as the diffuse electroncloud of the anion interacts with the substrate. The repulsive exchange inter-action also depends on the molecular orbitals overlap. This has the effect ofestablishing a balance between dispersion-induction and exchange-repulsionenergies resulting in a good correlation between the electrostatic energy andthe total interaction energy. One of the main conclusions of Kim’s work isthat the total interaction energies calculated for the anion-π complexes arecomparable to those obtained for the cation-π complexes. This is a funda-mental statement also mentioned in other theoretical studies, if one wantsto hypothesise on the strength and importance of the anion-π interactionin solution. It has been possible to estimate an energy value in the range of–2 to –0.5 kcal mol–1 for a single cation-π interaction using supramolecularmodel systems [34–36] and protein engineering studies [37]. As complementto these experimental studies, other experimental gas-phase measurementsand high-level theoretical studies have been used to assign a binding energyof approximately –10 kcal mol–1 for the interaction of tetramethylammoniumwith benzene in the absence of solvent [12].

Frontera and Deyà [38] warn that although it is true that the interac-tion energies of benzene with cations and hexafluorobenzene with anionsare similar, it is not possible to generalize that the interaction energies cal-culated for the anion-π complexes are comparable to those obtained for thecation-π complexes. These authors also indicate that the same is applicable toKim’s conclusion stating that the largest contribution in anion-π complexesare electrostatic and induction, because as we have seen before, these con-tributions sharply depend of the Qzz and α|| values of the aromatic system.

144 P. Ballester

Molecules with negligible Qzz values are expected to interact favorably witheither anions or cations, and the strength of the interaction would be com-parable, especially if the ionic van der Waals radii of the charged species aresimilar. Other conclusions from the work by Frontera and Deyà [38] com-paring different aspects of the cation-π and anion-π interactions are: a) thecontribution of dispersion and correlation terms to the total interaction en-ergy are small, but they are more important in anion-π complexes, b) thedensity at the cage critical point generated upon complexation of the ionis a useful parameter for measuring the strength of the interaction, evenwhen comparing anion-π to cation-π complexes, and c) a gain in aromatic-ity of the ring is observed, based on the nucleus-independent chemical shift(NICS) [39] criterion at the center of the ring, upon complexation of the an-ion, and the contrary is observed for the cation. Many authors agree that thecontribution from the dispersion energy is more important in the anion-πcomplexes than in the cation-π interaction. The induction energy emergesfrom the interaction of the occupied p orbital of the halide anion and theLUMO of the π-system. The inductive type of the MO interaction can alsobe correlated to the extent of charge transfer from the anion to the π sys-tem. In fact, several experimental studies have established a charge-transfercharacter to anion-π complexes (see below). The degree of charge transferin several anion-π complexes has been evaluated using different quantumchemical approaches. The computed charge transfer reported in the work ofFrontera and Deyà from the anion (F–) to the π-system is in the range of 0.1 to0.2|e| and 0.005 to 0.14|e| in Kim’s study. The results based on AIM method-ology indicate that the charge transfer is almost negligible for all complexesstudied.

Table 5 Binding energies MP2(full)/6-31++G∗∗//RI-MP2(full)/6-31++G∗∗ (kcal mol–1)with (∆EBSSE) and without (∆E) basis set superposition error correction and anion-arylcentroid distance (r, A)

Complex ∆E ∆EBSSE r

TFZ· · ·Cl– –20.3 –15.0 3.008(TFZ)2· · ·Cl– –38.5 –28.5 3.006(TFZ)3· · ·Cl– –65.6 –41.0 3.019 a

TFZ· · ·Br– –21.8 –14.2 3.176(TFZ)2· · ·Br– –41.7 –26.8 3.170(TFZ)3· · ·Br– –75.3 –38.6 3.172 a

TAZ· · ·Cl– –9.0 –5.2 3.220(TAZ)2· · ·Cl– –17.4 –10.4 3.213(TAZ)3· · ·Cl– –39.6 –22.2 3.015 a

a Mean distance


Fig. 10 Schematic geometries of the anion-π complexes studied by Frontera and Deyà [38]

Fig. 11 RI-MP2(full)/6-31++G∗∗ fully optimized structure of the trimeric complex ofTAZ with chloride. The distances between the N atom and the C – H bonds of the 1,3,5-triazine, as well as, between the anion and the centroid of the aromatic ring are shown

The additivity of the anion-π interactions has also been explored by Fron-tera and Deyà using high-level ab initio calculations [38]. They optimizedchloride and bromide complexes with one, two, and three aromatic units,such as trifluoro-1,3,5-triazine (TFZ) and 1,3,5-triazine (TAZ) – and analyzedthe interaction using the AIM theory and studied the charge transfer usingseveral methods for deriving atomic charges. The results revealed additivitiesin the binding energies and complex geometries that are almost insensitive

Table 6 Computed binding energies (∆E, kcal mol–1) for TFZ complexes with chloridesimulating two solvents and in the gas phase at the MP2(full)/6-31++G∗∗/RI-MP2(full)/6-31++G∗∗ level of theory

Complex ∆E (CHCl3) ∆E (H2O) ∆EBSSE (gas phase)

TFZ· · ·Cl– –5.9 –3.5 –15.0TFZ2· · ·Cl– –11.1 –7.0 –28.5TFZ3· · ·Cl– –31.7 –24.2 –41.0receptor –15.8 –11.9 –31.0

146 P. Ballester

Fig. 12 CAChe optimized structure of the chloride complex based on the receptorproposed by Frontera and Deyà [38]. The receptor binds the anion via four anion-πinteractions

to its stoichiometry (geometry additivity). To speed up the calculations pro-cess, the authors used the resolution identity MP2 method (RI-MP2) [40, 41].They state that the interaction energies and equilibrium distances obtainedwith the RI-MP2 method in the study of anion-π and cation-π interac-tions are almost identical to those obtained with the time consuming MP2calculations.

Calculations simulating two solvents systems (CHCl3 and H2O) within theself-consisting reaction field PCM model using the RI-MP2(full)/6-31++G∗∗geometries were also performed. The additivity for a set of complexes (TFZwith chloride) is maintained in both solvents, i.e., the binding energy of the1 : 2 complex is twice the value of the 1 : 1 complex. There is, however, a sig-nificant reduction of energy in comparison with the gas phase although thisis less important in the 1 : 3 complex.

As we discussed in the introduction, the anion-π interaction has poten-tial application in the field of molecular recognition of anions. In this sense,several theoretical studies have proposed structures for novel receptors based

Fig. 13 Molecular structures of the anion receptors proposed by Mascal [13]. The anion isbound by a combination of three ion-pair reinforced hydrogen-bonding and two anion-πinteractions


on multiple anion-π interactions. Thus, Frontera and Deyà propose a neutralreceptor for the binding of chloride that is composed of a neutral platformof 1,3,5-triazine substituted by three 2,4-difluoro-6-methylene triazine groupsresulting in a tripodal architecture. The receptor adopts a cup-like cavity thatincludes and binds the chloride via four anion-π interactions. The theoret-ical study simulating water and chloroform has been extended to the receptorindicating that this type of receptor could be adequate for the binding of chlo-ride in organic solvents.

Mascal [13] has also proposed the synthesis of three novel receptors for thebinding of anions that take advantage of the high conformational stability ofcylindrophane, which can effectively discriminate guests based on size. Thereceptors are designed to bind the anion through a combination of ion-pairreinforced hydrogen bonds and two anion-π interactions.

Mascal performed a detailed theoretical study of the association of the tri-azine cage and its cyanuric acid and boroxine analogues with the fluorideand chloride anions. Since the principal motivation of the work is the se-lective complexation of anions within the cages, the author compared thebisector distance (Ar – Ar distance/2 = 2.30 A average) for the empty re-ceptors with the anion-aryl centroid distance of halide sandwich complexes(raverage Ar···Cl– = 3.21 A and raverage Ar···F– = 2.45 A), which are analogous tothe 1 : 2 complex studied by Frontera and Deyà. The energies calculated forthe sandwich complexes are approximately additive, which is in agreementwith the work of Frontera and Deyà. The comparison of the bisector distanceswith the anion-aryl centroid distance of the sandwich complexes suggest thatboth chloride and fluoride are too large for all cage-like receptors. In the threereceptors, however, the optimal fluoride distance is better accommodated forthe empty cavity.

Table 7 Interaction energies (kcal mol–1) for the complexes of the triazine cages (TAZ)and its analogues cyanuric acid (CNA) and boroxine (BOX) with fluoride and chlorideion

Complex ∆EBSSE a∆EF–Cl

b ∆EBSSEH2O

a∆EF–Cl(H2O)

b

TAZcage· · ·Cl– –237.5 –1.1CNAcage· · ·Cl– –245.0 –13.4BOXcage· · ·Cl– –245.8 –21.8TAZcage· · ·F– –292.6 –55.1 –30.0 –28.9CNAcage· · ·F– –296.3 –51.4 –36.9 –23.5BOXcage· · ·F– –295.6 –40.7 –35.7 –13.9

a BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE) andin an aqueous solvent model (∆EBSSE

H2O )b Differences between the energy values of the fluoride and chloride complexes

148 P. Ballester

The binding properties of the halide complexes (F–, Cl–) were evaluatedin the gas phase and solution. Energies were also determined using theconductor-like polarisable continuum model for water, with the molecularcavity specified by the United Atom Topological Model applied on radii op-timized for the PBE0/6-31g(d) level of theory. The complexation of fluoridewith the three receptors is calculated to be more energetically favorable thanchloride by >40 kcal mol–1 in the gas phase and >13 kcal mol–1 in water. It isworth noting that the complexation energies calculated for F– ion are aboutthree times the experimental enthalpy of F– hydration. The author indicatesthat the raw comparison of these values is of limited predictive significancein regard to the potential of these receptors to extract F– from water, giventhe absence of activation and entropy values, as well as, the assumptions in-herent to nonexplicit, polarized continuum models of water. The comparisonof relative energetics of complexation, however, clearly indicates a preferencefor fluoride binding over chloride, from the gas phase to water, in all threereceptors (see Table 7).

The principal interaction force in these complexes is the ion-pair rein-forced hydrogen bond of the ammonium groups with the anions. It may besupposed that the intrinsic stronger ammonium· · ·F– interaction could actu-ally form the basis for the difference in binding energy. The isolation of therelative binding contributions (ion-pair reinforced hydrogen bond and anion-π interactions) was achieved by modelling the association of three NH4

+ ionsin a trigonal plane around a chloride and fluoride ions. The obtained resultsare shown in Table 8.

The high differences in binding energies calculated for the cage complexesand the modelled NH4

+ trigonal systems are clearly due to the fact that thethree ammonium groups are enforced in close proximity in the free recep-tor. The point to note is that although the stabilization of three NH· · ·F–

bonds in the trigonal (NH4+)3· · ·F– complex is about 11 kcal mol–1 greater

than the corresponding (NH4+)3· · ·Cl– complex, in the triazine and cyanuric

acid cage complexes there is still >10 kcal mol–1 “additional” stability in thefluoride complexes in aqueous solution. This observation suggests that thediscrimination has to do with a better fit of the anion at least in these two

Table 8 Interaction energies (kcal mol–1) for NH4+· · ·X– complexes

Aggregate ∆EBSSE a∆EF–Cl

b ∆EBSSEH2O

a∆EF–Cl(H2O)

(NH4+)3 · · · Cl– –130.0 –10.8

(NH4+)3 · · · F– –161.9 –31.9 –21.6 –10.8

a BSSE corrected B3LYP/6-31+G(d, p) interaction energies in the gas phase (∆EBSSE) andin an aqueous solvent model (∆EBSSE

H2O )b Difference between the energy values of the fluoride and chloride complexes


Table 9 Interaction energies (kcal mol–1) and noncovalent bonded distances for selectedcomplexes of 1,3,5-triazine (TAZ) and boroxine (BOX) with chloride and fluoride anions

Complex ∆E a ∆EBSSE a∆EF–Cl

b r c

TAZ· · ·Cl– –4.2 –4.0 3.48TAZ· · ·F– –10.4 –9.0 –5.0 2.66BOX· · ·Cl– –12.9 –12.5 2.94BOX· · ·F– –31.9 –29.6 –17.1 1.93TAZ· · ·Cl–· · ·TAZ –8.1 –7.6 3.49TAZ· · ·F–· · ·TAZ –18.9 –16.5 –8.9 2.79BOX· · ·Cl–· · ·BOX –23.1 –22.1 3.02BOX· · ·F–· · ·BOX –47.2 –43.9 –21.8 2.23TAZcage –55.1BOXcage –40.7

a B3LYP/6-31+G(d, p) interaction energies without (∆E) and with (∆EBSSE) BSSE correc-tionb Difference between the ∆EBSSE values of the fluoride and chloride complexesc Noncovalent bond distance (r) for simple (Ar· · ·X–), sandwich (Ar· · ·X–· · ·Ar). Corres-ponding data for the cage complex can be found in Table 7

Fig. 14 Structure of the tweezers studied in [42]

Table 10 Calculated interaction energies (kcal mol–1) of the investigated tweezer-ion com-plexes

Complex ∆Ecomplexa ∆Einter-MP2

b

6FT 5.20 –4.0610FT –0.28 –9.7214FT –5.95 –15.82

a Obtained by using B3LYP functional ∆Ecomplex = ∆Einter + ∆Edeformb Obtained by performing single point MP2 calculations on the optimized geometries

150 P. Ballester

cages. Surprisingly, the order of complementarity or selectivity for F– vs Cl– isTZNcage > CNAcage > BOXcage, which is completely reverse to the discrim-ination of the halide ions observed in simple binding complexes.

Hermida-Ramón and Estévez [42] have recently conducted a theoreticalstudy showing that fluorinated tweezers, that derive from those synthesizedby Kläner [43] were able to bind anions. Kläner tweezers are generally usedfor the molecular recognition of eletrodeficient aromatic and aliphatic sub-strates as well as organic cations. Several complexes formed between thefluorinated tweezers derivatives and an iodide anion were characterized. Thenature of the interaction was analyzed using the SAPT method while the en-ergetics for the complexation process were computed using calculations at theMP2 level on complex geometries that were optimized using density func-tional theory (DFT) and the B3LYP functional. The differences between MP2and B3LYP interacting energies can be considered as a rough approximationof the importance of the dispersive interactions in the stability of the tweezer-anion complex.

The values of the molecular electrostatic potential in the center of the cav-ity are –10.25 (6 FT), –5.79 (10 FT) and –0.64 (14 FT) kcal mol–1. Clearly, thebest receptor to accommodate iodine in its cavity is 14 FT, in which the MEPis almost neutral.

An increase in the fluoride substitution produces a stabilizing trend in theenergetics of the complexes. A comparison between the MEP value and thebinding energies indicates that the stability comes from the depletion of theelectrostatic repulsion between the anion and the π-clouds of the aromaticrings, which have a lower electron density owing to the influence of fluorine.The decrease in electrostatic repulsion is accompanied by a decrease in thedistance between anion and the tweezer together with an increase in attrac-tive energies (induction and dispersion), which gives more stability to thecomplexes. Calculations that include a polarizable continuum model of thesolvent (H2O) indicate that the complex is not stable. It must be noted thatfor the I–@14 FT complex, a large attractive binding energy is calculated eventhough the electrostatic potential inside the cavity is slightly negative.

Lewis and Clements [44] have performed quantum mechanical computa-tions that show that negative Qzz aromatics bind anions in the gas phase,dispelling the idea that the π electron density of negative Qzz aromatics isonly appropriate for cation binding. No correlation was observed, however,between anion binding enthalpies and the Qzz values for negative Qzz aro-matics. This observation is in striking contrast to what has been reportedand mentioned above for positive Qzz aromatics binding anions. The authorsdo indicate that the Morokuma-Kitaura decomposition calculations show thatthe major contribution to binding in the case of negative Qzz aromatics withanions is the energy due to polarizability of the aromatic ring. They alsomention that it may be not the polarizability of the aromatic ring densitythat is responsible for the binding but rather it may be the polarizability of


Fig. 15 Schematic representations of the geometries for Cl– complexes with TCBm

the aromatics substituents. The work concludes indicating that even electronrich, negative Qzz aromatics should be considered experimentally tenable foranion-π interactions. We believe that this statement should be taken withserious caution.

Johnson, Hay et al. [45] have refined the nature of the interactions be-tween electron deficient arenes and halide anions. In particular, they haveperformed calculations at the MP2/aug-cc-pVDZ level of theory of 1 : 1 com-plex formed between 1,2,4,5-tetracyanobenzene (TCB) and F–, Cl–, Br– an-ions. Four geometries were evaluated for each halide. The well-known anion-aryl centroid (A) (termed non-covalent anion-π complex in this study), twocharge transfer (CT) complexes in which the halide is positioned abovea C – H bond (B) or above a C – CN bond (C), and a C – H hydrogen bondcomplex (D) similar to the one already discussed by Mascal in his seminalstudy of the interaction of triazine with halide anions [17]. The theoreticalanalysis of TCB halide complexes revealed the first example in which non-covalent anion-aryl centroid complexes are not stable for Cl– and Br–. Thesehalides do interact with the π system, but the interaction involves CT, char-acterized by a high second-order stabilization energy ∆E(2). The resultingoptimized geometries of the CT complexes locate the anion over the periph-ery ring rather than over the center of the ring.

Since as we have presented above prior computational studies on Cl– andBr– have focused almost exclusively on the anion-aryl centroid complex, therehas been no investigation to determine the existence of such CT complexesfor other arenes. Consequently, the authors expand the electronic structurecalculations to halide complexes with triazine, hexafluorobenzene and 1,3,5-

152 P. Ballester

Fig. 16 Calculated EPS at the HF-6-31G level for TCB. The range of energy values is–39 (red) to +39 (blue) kcal mol–1

tricyanobenzene. The results with F– and triazine are fully consistent withearlier work [17]. Although the anion-aryl centroid and hydrogen-bondinggeometries have been previously identified as minima for Cl– and Br– [17],none of the previous studies report the existence of an “attack” geometryfor either Cl– or Br– only for F–. Two geometries were located for hexafluo-robenzene, the anion-aryl centroid which is not a stable point for F– and the“attack” geometry that is the global minimum for F– but was not located asa stable point for Cl– or Br–. These results are in complete agreement withprior theoretical studies [18, 19]. Results obtained for 1,3,5-tricyanobenzeneare similar to those obtained for TCB. The anion-aryl centroid geometry isonly stable for Br–. The CT complex that locates the anion above a C – H bondis the global minimum for the three halides. The hydrogen bond geometry isalso a stable structure for all the halides. All anion-aryl centroid complexeshave a low extend of charge transfer. Alternate geometries do have a highextend of charge transfer. It is worth noting that for the interaction of F–

with triazine in the “attack” geometry, the computed values for the param-eters used to gauge the extent of charge transfer, indicate the formation ofa strong σ bond. In conclusion, triazine, TCB and 1,3,5-tricyanobenzene formstable off-center CT complexes with Cl– and Br–. Except for the CT complexof Br– with triazine, the CT complexes are more stable than the anion-arylcentroid complexes. The differences in energies, however, are very small(<1 kcal mol–1), indicating a relative flat potential surface for positioning thehalide above the arene plane. In fact, the EPS of TCB shows a widespreaddistribution of positive electrostatic potential values all over the π-system.

3Experimental Evidence of the Anion-π Interaction

3.1Solution and Related Crystallographic Studies

To the best of our knowledge, the first report suggesting the experimen-tal existence of attractive interactions between anionic negative charges andπ-electron aromatic systems was published by Schneider et al. in 1993 [9].The 1H NMR measurement of the association constant between a simple


Fig. 17 Schematic representation of the complex formed with the cleft-like para-sulfonatoreceptor and diphenylamine placing the ionic part above the aromatic guest center

diphenylmethane organic host bearing negatively charged para-sulfonatogroups and an electroneutral diphenylamine show in aqueous solutionweak but attractive interactions that were quantified to reach approximately0.5 kcal mol–1 per X–/arene unit7.

The small complexation induced shift observed for the proton Hp(0.09 ppm) is interpreted by the authors as the existence of a displacement ofthe aryl rings in the complex. The authors state “such a slight displacementwill replace repulsions between permanent negative charges of the π-cloudand of the sulfonato substituent by attractions, as the π-cloud can still be po-larized by the charge of the sulfonato even if the charge is not exactly abovethe arene center”. In other words, although anion-π interactions are claimedto be responsible for the stability of the complex, the proposed geometry doesnot coincide with the one mainly investigated by the theoretical calculationsthat locates the anion above the π-aromatic systems. The nature of the attrac-tion is primarily assigned to the interaction between the anion and the dipolethat it induces in the arene (polarization). While the theoretical studies haverevealed that this “nonelectrostatic” term maybe important, it seems that it’snot the defining feature of the anion-π interactions. In this and another re-port [46], Schneider claims that the anion-π interaction can have almost thesame size as the cation-π interaction.

In 2004, an experimental study on the solid-state and solution behaviorof halide complexes with a series of highly electron-deficient arenes (Fig. 18)also suggests that the complex geometry investigated theoretically, whereinthe anion is located along the principal axes of the π-system, may not beoperative in these highly electron deficient arene systems [47].

In this study, the recognition of halide anions X– (X = Cl, Br, I) is clearlyestablished by the isolation and X-ray structure determination of a series ofwell-defined complexes containing the halide salt and the admixed aromatic

7 This study will suggest anion-arene interactions involving aromatics with negative Qzz values.A note in reference [44] states based: “on the systems studied and figures presented in the publica-tion it appears that the binding energy was probably due to arene-arene face to face interactions”.

154 P. Ballester

Fig. 18 Molecular structures of the neutral organic π-acceptors investigated in thestudy [47]

compounds in different stoichiometries, as well as by the spectral assign-ment of diagnostic charge transfer absorption bands to the formation of theanion-π complex. Intense colorations were observed upon the addition ofthe anions as the corresponding tetraalkylammonium salts to acetonitrilesolutions containing the aromatic π acceptor, i.e., tetracyanopyrazine TCP(calculated Qzz = 18.53 B) [48]. A new absorption band appears and growswith increasing halide concentration. Close inspection of the new absorptionband reveals that it consists of two Gaussian components. The stability con-

Fig. 19 Mulliken dependence of the energy of the low-energy absorption band (νCT) inthe TCP/Hal– complexes and the oxidation potential of the anion


Table 11 Solid-state characteristics of the halide complexes obtained by slow diffusion ofhexanes into 1 : 1 mixtures of neutral organic π-acceptors with the corresponding halidesas tetraalkylammonium salts in CH3CN/CH2Cl2

Molar ration Counterion X– · · · C [A] a

TCP/Br– 3 : 2 Et4N+ 3.164 : 1 Pr4N+ 3.15

TCP/I– 2 : 1 Et4N+ 3.521 : 1 Bu4N+ 3.49 b

TCP/Cl– 4 : 1 Bu4N+ 3.07o-CA/Br– 1 : 1 Pr4N+ 2.93

a The X– · · · C distance in closest contacts; note that the sums of the van der Waals radiiare 3.45 A (Cl– · · · C), 3.55 A (Br– · · · C), and 3.65 A (I– · · · C)b The average of the distances to the two or three neighbouring acceptors is given

stant values calculated for the complexes formed in acetonitrile with TCP andBr– as different tetraalkylammonium salts are in the range of 7–9 M–1. Fur-thermore, a job plot indicates a 1 : 1 stoichiometry for the TCP/Br– complexformed in solution. A clear correlation between the energy of the low-energyband and the oxidation potential of the anion was also observed and used toestablish a charge-transfer character for the complexes.

Furthermore, the study of the interaction of Br– with the series of neu-tral organic π-acceptor reveals that the increase of acceptor strength of thearomatic (characterized by a positive shift of the reduction potential) is ac-companied by a bathochromic shift of the new absorption band (from 355 to465 nm). This observation further confirms the charge-transfer (CT) charac-ter of these complexes.

The isolation and X-ray characterization of single crystals containing thehalide salt and the aromatic compound allows the identification in the solid-state of the non-covalent anion-π complexes. The anion is located in the space

Fig. 20 Local packing of the solid-state structures of the Br– complexes with TCP. Left:Bromide used as tetraethylammonium salt. Right: Bromide used as tetrabutylammoniumsalt. Only molecules that are in short contact (< sum of vdw radii) with respect to the Br–

atom are shown. In the case of the [(TCP)3(Br–)2](Et4N+)2 complex, the two different Br–

ions located within the asymmetric unit are represented

156 P. Ballester

close to the electron deficient aromatic ring explaining the observed elec-tronic (charge-transfer) transitions. The bromide ion lies approximately 3 Aover the periphery of the aromatic ring. The observed complex geometry dif-fers from those derived from quantum-mechanical calculations that usuallyshow the anion placed along an axis perpendicular to the aromatic ring andpassing through its centroid. The existence of multiple halide-π interactionsbecomes evident either directly in the asymmetric unit or after the packingof the lattice. The complex stoichiometries that can be derived from the X-raymolecular formula of the asymmetric unit i.e., [(TCP)3(Br–)2](Et4N+)2 (1.5 :1), [(TCP)4Br–]Pr4N+ (4 : 1) and [(TCP)I–]Bu4N+ (1 : 1), [(TCP)2I–]Et4N+

(2 : 1), as well as, the number of contacts established between the anion andthe aromatic rings seems to be controlled not only by the anion itself but bythe size of its countercation, which should influence the three-dimensionalsolid structure.

Johnson, Hay et al. [45] have used 1,2,4,5-tetracyanobenzene (TCB) togain further structural information of the anion-π interaction. TCP, 18-crown-6 and alkali halide were dissolved in acetonitrile (KBr) or 9 : 1dichloromethane:acetonitrile (KI, NaI) mixture. The purpose of the crownether is to enhance the solubility of the salt. It is important to note that thesolution of the crown ether and TCB is colorless, however, the addition of thehalide salt produces a color change consistent with the formation of CT com-plexes and in complete agreement with Kochi’s previous observation usingTCP instead. Slow evaporation at room temperature yielded single crystalssuitable for X-ray analysis. Despite packing differences, the local environmentabout each halide is remarkably similar in all crystals. As shown in Fig. 21,four TCB molecules surround the anion at interaction distances. Three dif-ferent orientations can be distinguished for the anion: above the arene planenearest to a C atom bearing a CN group (a and b), above the arene plane near-est to a C atom bearing a H atom (c), and nearly within the plane of the arene,forming a C – H hydrogen bond (d). This latter orientation was not observedin the study of Kochi. In conclusion, Kochi’s and Johnson’s results clearly

Fig. 21 Local packing of four TCB molecules around the anion. The closest contact to eacharene is indicated by a dotted black line


show that the anions are not located over the center of the arene (anion-arylcentroid geometry) in the solid state structures of these complexes. As dis-cussed above, theoretical studies indicate that TCB and 1,3,5-tricyanobenzeneare capable of forming stable off-center CT complexes with Cl– and Br–.

In 2003, Hoffmann et al. [49] introduced an uncharged host for the selec-tive binding of Cl– as the tetrabutylammonium salt in CDCl3 : DMSO 95 : 5(K–

Cl/K–Br = 105). This host features a triazine-trione platform with three short

arms that are conformationally preorganized through the introduction ofmethyl groups. The three arms are equipped with p-nitrophenylsulfonamidegroups capable of hydrogen-bonding to the anion. In this study, they also usedan analogous receptor with slightly longer side chains in order to evaluatehow the relative position of the anionic host with respect to the triazine-trioneplatform (anion-π interaction) could affect the stability of the complex.

The binding parameters of both hosts with chloride were determinedusing isothermal titration calorimetry. The microcalorimetry experiments re-

Fig. 22 Molecular structures of the receptors based on the triazine-trione platform hav-ing three arms of different length terminated with p-nitrosulfonamide hydrogen-bondinggroups

Table 12 Thermodynamic binding parameters determined using ITC for the complexationof tetrabutylammonium chloride by the tris-sulfonamides in CHCl3 at 298 K

Host long arm Host short arm

K/L mol–1 81 000±10 000 155 000±11 000∆G/kcal mol–1 –6.7±0.1 –7.1±0.1∆H/kcal mol–1 –4.7±0.1 –2.2±0.1– T∆S/kcal mol–1 –2.1±0.1 –4.8±0.1

158 P. Ballester

vealed that the host with the longer arms binds chloride slightly stronger thatthe one with the shorter arms (∆∆G ≈ 0.4 kcal mol–1).

There are, however, more important differences in the binding processesof the hosts to the tetrabutylammonium chloride. Although both complex-ation processes of chloride by the hosts are entropically favored, due to therelease upon binding of ordered solvent molecules from the surface of thechloride and the host to the bulk, the entropic change (– T∆S) for the hostwith longer arms is far less negative. This result may be the consequence ofa higher reduction in the rotational entropy of the host with the long arms oncomplex formation. In terms of enthalpy, the binding of chloride by the hostwith short arms is clearly less exothermic. In order to explain the observedresult, the authors speculate about the relationship between complex geom-etry and enthalpy. The more flexible host could adopt a conformation that ismore optimal for coordination of chloride giving rise to higher exothermic-ity. Furthermore, they conclude that from the measured enthalpy values ofcomplexation a potential electrostatic attraction of chloride by the triazine-trione platform in the receptor with the short arms, an anion-π interaction,can be discarded. We believe that the thermodynamic parameters measuredfor the chloride complexation with these systems are highly contaminatedwith solvent effects. If the number of solvent molecules released to the bulkon complexation is different for each host, then the comparison of the data iscompletely inappropriate. In terms of free energy of binding, the experimen-tal results indicate that both chloride complexes formed with the two triden-tate p-nitrophenylsulfonamide hosts have approximately the same bindingenergy and the potential electrostatic attraction of chloride by the triazine-trione (cyanuric) platform is not reflected.

In 2005, Frontera, Saczewski, Deyà et al. [50] performed a combinedcrystallographic and computational study of the anion-π interactions with

Fig. 23 Minimized structures (Maestro, MMFFF) of the chloride complexes formed withthe tridentate p-nitrophenylsulfonamide having long and short arms. The distance of thechloride atoms to the centroid of the cyanuric acid ring is shown in each case


Fig. 24 Synthesis of the halide salts of the 2-ethyleneamine derivatives of thio- and dithio-cyanuric acid

cyanuric acids [50]. These authors interpreted the experimental results pre-viously obtained by Hoffmann et al. making use of MIPp calculations. TheMIPp value for the interaction of cyanuric acid with a chloride anion placed2.5 A apart from the ring centroid is almost the same to that computed ata distance of 3.5 A (≈–16 kcal mol–1). Consequently, a likely explanation forthe similarity of free energies of complexation measured experimentally withlong or short arms tridentate p-nitrophenylsulfonamide hosts is that the ef-fect of the cyanuric platform on the anion binding is similar and does notallow to differentiate one with respect to the other.

To obtain experimental evidences of the ability of cyanuric acid to interactfavorable with anions, Frontera, Saczewski, Deyà et al. [50] synthesizedseveral derivatives of thiocyanuric and dithiocyanuric acid with a flexible2-ethyleneamine arm attached to the s-triazine ring. The one-step synthesisof these compounds requires a 1–4 h reflux in aqueous 12% HX (X = Cl, Br,

Fig. 25 Top view and side view of the local packing of the chloride in the crystal structureof the hydrochloride salt of the 2-ethyleneamine thiocyanuric acid derivative

160 P. Ballester

Fig. 26 Top view and side view of the local packing of the chloride in the crystal structureof the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

Fig. 27 Top view and side view of the local packing of the bromide in the crystal structureof the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

Fig. 28 Top view and side view of the local packing of the iodide in the crystal structureof the hydrochloride salt of the 2-ethyleneamine dithiocyanuric acid derivative

I) acid. After cooling the reaction mixture to ambient temperature, colorlesscrystals suitable for X-ray analysis of the corresponding hydrohalide salts ofthe amine were obtained.


The X-ray crystal structures of all the hydrohalide salts except the hydro-bromide of the thiocyanuric acid derivative revealed the existence of anion-πinteractions. The organic cation, which has several hydrogen-bond donor andacceptor functionalities, displays an electrostatic interaction between the Natom of the ammonium moiety and an anion (X–) located nearly above thecenter of the electron-deficient ring. The packing of the lattice reveals that theanion (X–) placed on top of the aromatic rings also interacts electrostaticallywith at least one ammonium group of an adjacent cyanuric acid derivative.Additionally, one of the NH groups of the s-triazine ring of another adjacentorganic cation binds the anion (X–) through a hydrogen-bond interaction.Finally, two other molecules of the organic cation are also surrounding theanion. One of them has the 2-ethyleneamine arm positioned at short contactof the anion (distance < sum of vdW radii).

The distance between the ring centroid and the chloride anion is 0.1 Alonger in the dithione than in the monothione derivative. In conclusion, theX-ray structures of the halide salts of mono- and dithiocyanuric acid deriva-tives having an ethylenenammonium arm attached to one of the nitrogenatoms of the s-triazine do show the existence of anion-π accompanied by salt-bridge and hydrogen-bonding interactions as a robust structural motif of thesolid-state packing.

In an attempt to probe the efficacy of the non-covalent anion-π interac-tion, Johnson et al. [51] have prepared two sulfonamide-derived receptors(A and B in Fig. 29). The design of the receptor is based on a convergenttwo-point recognition motif utilizing both a hydrogen bond and an aromaticring that can be involved in the formation of anion-π interactions. The pre-pared receptors differ in the electronic properties of one of the rings of thebiphenylamino moiety due to the substitution with five fluorine atoms. Thequadrupole moment of such an aromatic ring should be highly positive whilethe simple phenyl should have a negative quadrupole moment.

Receptor B has a pentafluoro substituted aromatic ring that is electron-deficient and more appropriate to engage in an attractive anion-π interaction.

Fig. 29 Molecular structures of the receptors used by Johnson et al. to evaluate the anion-π interaction in solution

162 P. Ballester

Fig. 30 CPK representation of the single-crystal X-ray structure of the pentafluorophenylreceptor. The receptor adopts in the solid-state a conformation that is adequate to interactsimultaneously through hydrogen-bonding (sulfonamide NH) and anion-π interactionwith an halide located on top of the pentafluorophenyl ring

Table 13 Ka (M–1) for the sulfonamide receptors with Cl–, Br– and I–

Anion Phenyl receptor (A) Pentafluorophenyl receptor (B)

Cl– <1 30±3Br– <1 20±2I– <1 34±6

The tetraethylammonium salts of each anion were used

The stability constants of the complex formed with receptor B and a seriesof anions (Cl–, Br–, I–) were evaluated in CDCl3 using NMR titration tech-niques. Receptor B binds all the screened halides with a measurable but mod-est association constant. The association constants measured were 20 M–1 for

Fig. 31 X-ray crystal structure of the complex formed between the pentafluorophenylreceptor B and tetra-n-butylammonium bromide. The sulfonamide receptor and the or-ganic cation are shown in stick representation while the bromide is represented with vander Waals surface


Br–, 30 M–1 for Cl– and 34 M–1 for I–. On the other hand, the associationconstant for receptor A lacking the electron-deficient aromatic ring requiredfor the anion-π interaction, were too small to be determined using the sametitration methodology. The enhanced affinity exhibited by receptor B over re-ceptor A with the array of halides is attributed to the existence of anion-πinteraction in the anion complex.

The experiments presented by Johnson et al. support further studies onthe possible use of the anion-π interaction as an emerging no covalent inter-action of the selective targeting of anions in solution. The difference in thereported stability constants (∆∆G) can be used to estimate a value for theanion-π interaction in the range of approximately –2 kcal mol–1. This valuealmost doubles the value estimated for the cation-π interaction using pro-tein engineering [37] (–2.75 kcal mol–1 for a cavity lined by three π-systems– 2.75/3 = 0.9 kcal mol–1) and supramolecular chemical model systems [35](–2.4 kcal mol–1 for a receptor with four π-systems – 2.4/4 = 0.6 kcal mol–1).Probably, the anion-π interaction is over-evaluated when using the chem-ical model based on sulfonamides. Johnson et al. performed the 1H NMRtitrations experiments in CHCl3, and the literature indicates that tetraalky-lammonium salts form tight ion pairs with small anions, i.e., Cl– in CHCl3solutions [34, 52]. Consequently, the sulfonamide receptors bind the tightion-pair and the observed enhancement of the association constants for thepentafluorophenyl receptors could be partially caused due to the existence ofelectrostatic interactions between the fluorine atoms and the tetraalkylam-monium salt. The X-ray crystal structure of the pentafluorophenyl receptorand tetra-n-butylammonium bromide, in which the ion-pairing in the solidstate is held responsible to force the sulfonamide NH away from the preferredconformation, shows close contacts between the fluorine atoms and the or-ganic cation (Fig. 21).

3.2Further Crystallographic Evidence of Anion-π Interactions

Examples of crystal structures of supramolecular complexes in which an-ions are nested in the interior of aromatic cavities have been used as ev-idence for the existence of an attractive anion-π interaction. In 2004, thefirst crystallographic evidence of anion recognition by aromatic receptors wasdescribed by Meyer et al. [53], and Gamez, Reedijk et al. [54]. While inves-tigating the structural and magnetic properties of metal complexes with thehexakis(pyridine-2-yl)-[1,3,5]triazine-2,4,6-triamine ligand (L), Meyer syn-thesized a copper (II) chloride complex that showed chloride-triazine bind-ing with geometrical parameters almost exactly to those calculated com-putationally for the 1,3,5-triazine· · ·Cl– complex. The cationic moiety of[L2(CuCl)3)]3+ consists of two ligands L that are stacked in a parallel fashionand held together by three copper(II) ions. The four pyridine-N groups and

164 P. Ballester

Fig. 32 Left: molecular structure of the ligand (L). Right: side view and top view of thecopper(II) chloride complex, few solvent molecules, some hydrogen and chloride atomsare omitted for clarity. The atoms Cl(8) and Cl(5) are shown in CPK representation. Spe-cial emphasis is placed on the location of these two atoms in relation to the triazine ringof the cationic moiety of the complex [L2(CuCl)3]3+

an apical chloride constitute a square pyramidal coordination of each metalcenter. Cl atoms of Cl2CH2 are placed between the paddles of the structure.Both triazine rings are arranged in almost perfect face-to-face arrangement.The most interesting structural feature related to the anion-π interaction isthe position of the chloride atoms 8 and 5 belonging to [CuCl4]2– ions. Cl(8)is located above one of the triazine rings in an anion-aryl centroid geometry.The distance between the centroid of the ring and the anion is 3.17 A whilethe angle Cl–· · ·centroid axis to the plane of the ring is 87◦, that is, the chlo-ride is almost located on the C3 axis above the ring. The Cl(5) chloride showsa similar geometry.

Fig. 33 Side view and top view of a fragment of the crystal structure of the Cl– complexreported by Meyer. The six hydrogen-bonding interactions with the aryl C – H groups ofthe Cu(II) coordinated pyridine rings are indicated. C – H· · ·Cl– angles ≥150◦ and C toCl– distances range from 3.9 to 4.3 A


Fig. 34 Left: Molecular structure of azadendritz. Right: side views and top view ofthe crystal structure of the Cu(II) complex of azadendritz reported by Gamez, Reedijket al. [54]. Solvent molecules and hydrogen atoms are omitted for clarity. The structureshows the four pentacoordinate Cu(II) ions coordinated two apical chloride atoms (scaledball-and-stick representation) and two encapsulated chloride anions (CPK representa-tion)

The authors state that a “Cl–· · ·triazine complex due to electrostatic anion-π interaction is present in the complex”. Johnson, Hay et al. [45] have recentlynoticed that the role of the C – H groups of deficient arenes as potent hydro-gen bonds should not be overlooked, in fact the structure of Meyer nicelyillustrates this observation. The Cl(8) anion positioned above the center ofthe melamine is interacting with the C – H groups of the Cu(II)-coordinatedpyridines. The hydrogen bonds formed may play a dominant role in deter-mining the position of the anion within this cavity.

Almost at the same time, Gamez, Reedijk et al. [54] described the first co-ordination compound of the ligand azadendritz. The supramolecular Cu(II)complex shows intramolecular π–π interactions between two 1,3,5-triazinerings.

Similar to the structure reported by Meyer, four pyridine-N groups andan apical chloride constitute a square pyramidal coordination of each metalcenter. In this case, however, the chloride atoms are not located on topof the 1,3,5-triazine rings, which are perfectly stacked, instead the anionis nestled against four aromatic pyridyl residues. In Meyer’s structure, thissame position was occupied by Cl atoms of Cl2CH2 molecules. Each en-capsulated chloride is in close contact with the four pyridine rings withcentroid· · ·Cl– distances ranging from 3.5 to 3.7 A. It has to be noted thatin both examples the electron-poor character of the pyridine moieties isprobably enhanced by their coordination to the Cu(II) metals. In additionto the π interaction with the pyridine rings the authors noted that thechloride ions are in close proximity to the triazine rings, suggesting theexistence of some additional electrostatic interactions with the electron de-ficient triazine moiety. The angles of the Cl–· · ·centroid axis to the planeof the different pyridine rings ranges from 74◦ to 82◦. This result en-couraged the authors to investigate further the anion · · · triazine interaction

166 P. Ballester

using other triazine-based ligands and anions [55, 56]. Thus, the reactionof Zn(NO3) ·6H2O or Cu(NO3) ·3H2O with the ligand dipicatriz in acetoni-trile afforded mono- or trinuclear coordination complexes depending on themetal-to-ligand ratio used during the crystallization process [56]. In par-ticular, the trinuclear Zn(II) complex shows anion-π interactions betweentwo nitrate anions coordinated to two different zinc atoms and the same tri-azine ring, one in each face of the aromatic ring. On the other hand, thecorresponding trinuclear Cu(II) complex exhibits an even more remarkabledouble anion-π interaction between the triazine ring and two nitrate anions.In this case, the two nitrate anions are now uncoordinated. One of the ni-trates interacts via a somewhat longer distances, which is most likely dueto the fact that the triazine is already involved in a nitrate-centroid contact.These examples constitute one of the first reports on anion-π-anion inter-action. It is also worth noting that the observed spatial arrangement of thenitrate anion with respect to the triazine ring is distinct to the one pro-posed by Kim et al. [32] based on theoretical calculations. Gamez, Reedijket al. [56] performed ab initio calculations demonstrating that the interactionenergies calculated for this peculiar positioning of the nitrate is compara-ble to the energy of the binding geometry proposed by Kim et al. in histheoretical studies of the interaction of triazine with nitrate [32]. In themodel proposed by Kim et al., the anion is parallel to the triazine ring andthe oxygen atoms are located on top of the electropositive carbon atomsof the ring.

Related to the previous example Gamez, Reedijk et al. [55] also foundcrystallographic evidences of another geometry for the nitrate-π interac-tion in the tetranuclear complex [Cu4(dpatta)(NO3)4](NO3)4 obtained underhydrothermal conditions from the reaction of copper(II) nitrate and theligand dpatta [55]. Each two triazine units are perfectly π–π stacked butstaggered with a centroid-to-centroid distance of 3.45 A. Both triazine rings

Fig. 35 Left: Molecular structure of dipicatriz. Right: side views of the crystal structuresof the Zn(II) and the Cu(II) trinuclear complexes of dipicatriz. The structures show thethree coordinate Zn(II) and Cu(II) ions in scaled ball-and-stick representation and thedistances between the centroids of the triazine ring and the closest oxygen atoms of theinteracting nitrate anions


are alternatively involved in an interaction with a lattice nitrate but witha different geometry. A DFT analysis was performed with a model complextriazine · · · nitrate to analyze the anion-π interaction. In particular, the pref-erence for the coordination of the nitrate ion observed in the X-ray structurewith one oxygen pointing to one nitrogen of the triazine ring. A DFT-B3LYPgeometry optimization indicates that a local minimum exists with a geometryvery similar to that observed in the X-ray crystal structure. The plane of thenitrate is not completely parallel to the triazine ring and one of the oxygenatoms points in the direction of one of the N atoms of the ring. This calcu-lated structure is 0.7 kcal mol–1 more stable that Kim’s proposed geometry.Furthermore, the dpatta Cu(II) complex represents one of the first crystallo-graphic evidences of anion-π–π interactions (nitrate-triazine-triazine). Thecomputed energy for the optimized nitrate-triazine-triazine complex is only0.46 kcal mol–1 higher than the complex including one ring. Due to this lowenergy difference, it is difficult to draw any conclusions about the stabiliza-tion of the ternary complex induced by the extra π–π interaction. Limitationsin the functionals available to describe the π–π interactions together withunconsidered coordination forces may account for the impossibility to repro-duce this experimental observation.

Fig. 36 Molecular structure of dpatta. Stick representation of the crystal structure ofthe tetranuclear complex [Cu4(dpatta)(NO3)4](NO3)4 complex. Hydrogen atoms, non-coordinated nitrate anions and lattice molecules are omitted for clarity. In box: partialstick view of the structure showing the anion-π–π interactions

168 P. Ballester

Other beautiful examples of anion-π interactions in which the aromaticsystem is a N-containing arene have been described in recent years [57–59].For example, Dunbar et al. [60] undertook a tandem crystallographic andcomputational investigation of the effect of the anion-π interaction on thepreferred structural motifs of Ag(I) complexes with tetrazine and pyridazine-based ligands. The investigated anions are polynuclear like PF6

–, AsF6–,

SbF6–, and BF4

–. The ligands used are 3,6-bis(2′-pyridyl)-1,2,4,5-tetrazine(bptz) and 3,6-bis(2′-pyridyl)-1,2-pyridazine (bppn). The MEP of the freeligands indicated that the bptz tetrazine ring has a higher π-acidic char-acter compared to the bppn pyridazine ring. Consequently, the bptz ringis more likely to participate in anion-π interactions. This preference is incomplete agreement with the observed structural motifs of the complexesof Ag(I)-bptz, regardless of the anion. In the Ag(I)-bptz compounds, multi-ple and shorter anion-π interactions are established between the anion andthe central tetrazine rings, whereas in the Ag(I)-bppn complexes intermole-cular interactions are maximized at the expenses of anion-π interactions. Thesolid-state evidences are also supported by DFT calculations.

The preceding theoretical evaluation of the anion-π interaction, especiallythe results of Johnson, Hay et al. [45], together with the experimental ev-idences derived from the work of Kochi et al. [47], seem to suggest thatthe anion-aryl centroid complex geometry may not be the more commongeometry in the interaction of halides with π-deficient electron systems. TheCambridge Structural Database (CSD) is a convenient and reliable store-house for geometrical information. Furthermore, it is well established theutility of the solid-state structure of small molecules in analyzing the geo-metric parameters for intermolecular interactions. In order to confirm theexperimental existence of the anion-π interaction in a general way, a sur-

Fig. 37 Molecular structure of the ligands bptz and bppn. Middle: portion of the pack-ing diagram of [Ag2(bptz)3][SbF6]2 depicting the anion and cation arrangement. Right:portion of the grid-type structure of [Ag2(bppn)4][PF6]4 depicting π–π and anion-πinteractions


vey of the CSD has been carried out by several authors [10, 19, 45]. It isclear that the result of a search of the CSD will be highly dependent onthe search criteria used to store a hit. Johnson, Hay et al. [45] carriedout a search for halide anions located within 4.0 A of the centroid of six-membered ring π systems yielded 600 examples. In most cases, however,the π-system was either positively charged or bonded directly to a posi-tively charged atom. If only charge-neutral π systems are considered, a muchsmaller set of 19 different structures is retrieved. Within these structures,30 halide arene complexes were found. They performed an analysis of sev-eral distances of the complexes, i.e., distance anion-to-centroid, distanceanion-to-plane, etc., concluding that 84% of the anions in the data set arecloser to the ring carbons than to the centroid. Thus, the available struc-tural data of the CSD indicates that the CT-binding motif of a halide in-teracting with a π systems is more prevalent that the anion-aryl centroidmotif. The authors explained that the incongruence of their results withprior analysis of the CSD—which reported evidences of a marked prefer-ence of charge-neutral atoms bearing lone pairs to position themselves overthe center of pentafluoroarenes [19] and trinitrobenzene derivatives [61]—isdue to the use of a misleading search criterion. Finally, based on the sta-tistical analysis of the hits obtained for a search criteria, which is that theelectronegative atoms must be 4 A of the pentafluorophenyl centroid, theyestablished a complete absence of any preferred location over the π sys-tem. The use of solid-state structures to analyze the geometric parametersof weak intermolecular forces such as anion-π interactions, has a generaldrawback, that is, other intermolecular interactions that are stronger, i.e.,charge–charge interactions may determine the final crystal packing arrange-ment. Consequently, the geometry observed in the crystal for the weak inter-action could be not the preferred one but the one resulting of the balance ofintermolecular forces that controls the packing of the lattice. Probably, thisis one of the causes that produces the complete absence of a preferred loca-tion of the anion over the π system of perfluorobenzenes. However, severalsearches carried out in the CSD do show that many of the geometries the-oretically calculated for the interaction of anions with π systems are in factpresent in the solid state. Consequently, this type of non-covalent interactionis worth to be seriously investigated by the practitioners of supramolecularchemistry.

A recent account by Matile et al. [62] can be considered as the first ex-perimental example of the use of anion-π interactions for the design ofa synthetic anion channel. Matile and co-workers report the design, syn-thesis, and evaluation of π-acidic, shape-persistent oligo-(p-phenylene)-N,N-naphthalenediimide (O-NDI) rods that can transport anions across lipidbilayer membranes with a rare selectivity Cl– > F– > Br– > I– and a sub-stantial anomalous mole fraction effect. DFT calculations revealed a globalquadrupole moment Qzz =+ 19.4 B for a model NDI. By comparison with

170 P. Ballester

Fig. 38 a The concept of the anion-π slide in lipid bilayer; b MEP for model NDI; red:electron-rich, blue: electron-poor

rigid p-oligophenyl rods, the authors deduced that the alignment of threeNDI acceptors separated by phenyl spacers would afford rods with the ap-propriate length for hydrophobic matching with common lipid bilayer mem-branes. The results obtained in this study are in agreement with opera-tional dynamic anion-π interactions and the existence of multiple anion-πsites for transmembrane anion hopping, that is, anion-π slide as shown inFig. 38. A final caveat of the authors indicates that further studies are ne-cessary to corroborate insights on the novel and complex system introducedin the study.

4Summary and Outlook

Theoretical studies indicate the existence of a counter-intuitive attractive in-teraction between anions and π-systems. The strength of the interaction


(binding energies) seems to be drastically reduced by the inclusion of sol-vents effects in the calculations, although in some cases the interaction is stillfavorable. Calculations of F–, Cl–, Br– and NO3

– complexes with electron-deficient aromatic systems (quadrupole moment Qzz > 0) like, triazine, hex-afluorobenzene and polycyanobenzenes establish the existence of differentbinding geometries. When the halide lies above the plane of the π-systemstrongly covalent sigma complexes, weakly covalent donor-π acceptor (CT)complexes and anion-aryl centroid complexes may be located as stable ge-ometries. The halides can also form stable hydrogen-bonded complexes withπ-aromatic systems having C – H donors groups. In fact, the presence ofelectro-withdrawing substituents increases the acidity of the C – H donorand strengthens this type of interaction. The most stable binding geom-etry (global minimum) depends on the anion and the π-system. Also, somebinding geometries which are not even stable points (minimum) for certainanions could be global minimum for others. In many cases, the off-centerCT complexes are more stable that the anion-aryl centroid binding geometry.The physical nature of the anion-π interaction has also been investigated ingreat detail from a theoretical point of view. The Qzz molecular quadrupoleof aromatics has the adequate spatial orientation to be involved in the for-mation of non-covalent anion-π complexes in which the halide is locatedabove the arene centroid (anion-aryl centroid geometry). The two funda-mental components contributing to the stabilization energy of non-covalentanion-π or anion-aryl centroid complexes in the gas phase are electrostaticand anion-induced polarization. The electrostatic component correlates wellwith the magnitude of the Qzz of the aromatic ring and the contributiondue to polarization can be considered as almost constant. The contribu-tions of the dispersion and correlation terms to the total interaction en-ergy are small, but they are more important in anion-π complexes, thanin cation-π complexes.

In recent years, numerous experimental studies have shown crystallo-graphic evidences of anions encapsulated in aromatic cavities formed byelectron-deficient arenes. These observations clearly hint to the existence ofan attractive interaction between anions and π-systems. Nevertheless, manyof the reported crystal structures do not show the anion located above thearene centroid (anion-aryl centroid geometry). On the contrary, the greatmajority of the theoretical studies on the anion-π interaction deal with com-plexes displaying the anion-aryl centroid geometry in the gas phase.

Although several theoretical studies have proposed structures for novel re-ceptors based on multiple anion-π interactions, to date, and to the best ofour knowledge, the experimental realization of such type of receptors andtheir anion-binding properties have not been reported. The few experimentalstudies that have attempted the experimental quantification of the anion-π interaction in solution have produced very different and even oppositeresults.

172 P. Ballester

The last paragraph of a chapter like this one asks the writer to predictif the non-covalent interaction of anions with π-system will be as import-ant and useful as the nowadays well-established cation-π interaction. We arecautious in answering this question affirmatively, but we have to confess thatwe are currently working in the construction of synthetic receptors for an-ions that do incorporate electron deficient π-aromatic systems. We hope thatour designs will be valuable for the experimental evaluation of the anion-π interaction in solution and we will report shortly our findings. Once thestrength of the anion-π interaction in solution is determined, it will be easierto evaluate its possible use in the construction of selective receptors for an-ions making use of the directionality properties that we have discussed in thechapter.

Acknowledgements I want to thank Dr. Antonio Frontera, Dr. David Quiñonero, and Prof.Pere M. Deyà from the University of the Balearic Islands for sharing with me his interestand results of their studies on the anion-π interaction. As a consequence, my group is nowinvolved in trying to quantify experimentally the strength of this new, counterintuitive andrather unnoticed non-covalent interaction, and I find myself writing a book chapter on thetopic. Generous financial support from MEC (CTQ2005-08989-C01-02/BQU and CSD2006-0003), ICIQ Foundation, ICREA Foundation and Generalitat de Catalunya (2005SGR00108)is gratefully acknowledged.

References

1. Hunter CA (2004) Angew Chem Int Ed 43:53102. Amendola V, Esteban-Gomez D, Fabbrizzi L, Licchelli M (2006) Acc Chem Res 39:3433. Bowman-James K (2005) Acc Chem Res 38:6714. Beer PD, Schmitt P (1997) Curr Opin Chem Biol 1:4755. Simmons HE, Park CH (1968) J Am Chem Soc 90:24286. Park CH, Simmons HE (1968) J Am Chem Soc 90:24297. Park CH, Simmons HE (1968) J Am Chem Soc 90:24318. Schmidtchen FP, Berger M (1997) Chem Rev 97:16099. Schneider HJ, Werner F, Blatter T (1993) J Phys Org Chem 6:590

10. Gamez P, Mooibroek T, Teat S, Reedijk J (2007) Acc Chem Res 40:43511. Egli M, Sarkhel S (2007) Acc Chem Res 40:19712. Ma JC, Dougherty DA (1997) Chem Rev 97:130313. Mascal M (2006) Angew Chem Int Ed 45:289014. Gallivan JP, Dougherty DA (1999) Org Lett 1:10315. Danten Y, Tassaing T, Besnard M (1999) J Phys Chem A 103:353016. Alkorta I, Rozas I, Elguero J (1997) J Org Chem 62:468717. Mascal M, Armstrong A, Bartberger MD (2002) J Am Chem Soc 124:627418. Alkorta I, Rozas I, Elguero J (2002) J Am Chem Soc 124:859319. Quiñonero D, Garau C, Rotger C, Frontera A, Ballester P, Costa A, Deyà PM (2002)

Angew Chem Int Ed 41:338920. Caldwell JW, Kollman PA (1995) J Am Chem Soc 117:417721. Sunner J, Nishizawa K, Kebarle P (1981) J Phys Chem 85:181422. Burgi HB, Dunitz JD, Shefter E (1973) J Am Chem Soc 95:5065


23. Bürgi HB, Lehn JM, Wipff G (1974) J Am Chem Soc 96:195624. Bader RFW (1991) Chem Rev 91:89325. Cheeseman JR, Carroll MT, Bader RFW (1988) Chem Phys Lett 143:45026. Koch U, Popelier PLA (1995) J Phys Chem 99:974727. Cubero E, Luque FJ, Orozco M (1998) Proc Natl Acad Sci USA 95:597628. Hernandez B, Luque FJ, Orozco M (1999) J Comput Chem 20:93729. Luque FJ, Orozco M (1998) J Comput Chem 19:86630. Williams JH (1993) Acc Chem Res 26:59331. Garau C, Frontera A, Quinonero D, Ballester P, Costa A, Deya PM (2003) Chem Phys

Chem 4:134432. Kim D, Tarakeshwar P, Kim KS (2004) J Phys Chem A 108:125033. Jeziorski B, Moszynski R, Szalewicz K (1994) Chem Rev 94:188734. Hunter CA, Low CMR, Rotger C, Vinter JG, Zonta C (2002) Proc Natl Acad Sci USA

99:487335. Kearney PC, Mizoue LS, Kumpf RA, Forman JE, McCurdy A, Dougherty DA (1993)

J Am Chem Soc 115:990736. Hans-Jörg Schneider TB, Patrick Zimmermann (1990) Angew Chem Int Ed Engl

29:116137. Ting AY, Shin I, Lucero C, Schultz PG (1998) J Am Chem Soc 120:713538. Garau C, Quinonero D, Frontera A, Ballester P, Costa A, Deya PM (2005) J Phys

Chem A 109:934139. Schleyer PvR, Maerker C, Dransfeld A, Jiao H, Hommes NJRvE (1996) J Am Chem Soc

118:631740. Vahtras O, Almloef J, Feyereisen MW (1993) Chem Phys Lett 213:51441. Feyereisen M, Fitzgerald G, Komornicki A (1993) Chem Phys Lett 208:35942. Hermida-Ramon JM, Estevez CM (2007) Chem Eur J 13:474343. Klarner F-G, Kahlert B (2003) Acc Chem Res 36:91944. Clements A, Lewis M (2006) J Phys Chem A 110:1270545. Berryman OB, Bryantsev VS, Stay DP, Johnson DW, Hay BP (2007) J Am Chem Soc

129:4846. Schneider H-J, Yatsimirsky AK (2000) Principles and Methods in Supramolecular

Chemistry. Wiley, Chichester47. Rosokha YS, Lindeman SV, Rosokha SV, Kochi JK (2004) Angew Chem Int Ed 43:

465048. Gaussian 03W Version 6.1. DFT B3LYP/6-31G(d,p)49. Hettche F, Hoffmann RW (2003) New J Chem 27:17250. Frontera A, Saczewski F, Gdaniec M, Dziemidowicz-Borys E, Kurland A, Deyà PM,

Quiñonero D, Garau C (2005) Chem Eur J 11:656051. Berryman OB, Hof F, Hynes MJ, Johnson DW (2006) Chem Commun, p 50652. Mo H, Wang A, Wilkinson PS, Pochapsky TC (1997) J Am Chem Soc 119:1166653. Demeshko S, Dechert S, Meyer F (2004) J Am Chem Soc 126:450854. de Hoog P, Gamez P, Mutikainen I, Turpeinen U, Reedijk J (2004) Angew Chem Int Ed

43:581555. Casellas H, Massera C, Buda F, Gamez P, Reedijk J (2006) New J Chem 30:156156. Maheswari PU, Modec B, Pevec A, Kozlevcar B, Massera C, Gamez P, Reedijk J (2006)

Inorg Chem 45:663757. Dorn T, Janiak C, Abu-Shandi K (2005) Cryst Eng Comm 7:63358. Gural’skiy IyA, Solntsev PV, Krautscheid H, Domasevitch KV (2006) Chem Commun,

p 480859. Zhou X-P, Zhang X, Lin S-H, Li D (2007) Crystal Growth & Design 7:485

174 P. Ballester

60. Schottel BL, Chifotides HT, Shatruk M, Chouai A, Perez LM, Bacsa J, Dunbar KR(2006) J Am Chem Soc 128:5895

61. Quinonero D, Garau C, Frontera A, Ballester P, Costa A, Deya PM (2002) Chem PhysLett 359:486

62. Gorteau V, Bollot G, Mareda J, Perez-Velasco A, Matile S (2006) J Am Chem Soc128:14788

Struct Bond (2008) 129: 175–206DOI 10.1007/430_2008_083© Springer-Verlag Berlin HeidelbergPublished online: 12 March 2008

Anion Templates in Synthesisand Dynamic Combinatorial Libraries

Ramon Vilar

Department of Chemistry, Imperial College London, London SW7 2AZ, [email protected]

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

2 Templating Effects in Chemistry . . . . . . . . . . . . . . . . . . . . . . . . 177

3 Recent Examples of Anion-Templated Processes . . . . . . . . . . . . . . . 1783.1 Macrocycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1783.2 Cages and Capsules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1823.3 Interlocked Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

4 Anions as Templates in Dynamic Combinatorial Chemistry . . . . . . . . 1914.1 Using Metal–Ligand Coordination Bonds . . . . . . . . . . . . . . . . . . . 1924.2 Using Reversible Covalent Bonds . . . . . . . . . . . . . . . . . . . . . . . . 201

5 Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Abstract This review presents an overview of the area of anion-templated synthesis ofmolecules and supramolecular assemblies. The review is divided into two main sections:the first part deals with anion-templated systems where the final products are linked bybonds that are not reversible under the conditions of the experiment. Several recent ex-amples of macrocycles, cages and interlocked species are presented in this section. Thesecond part of the chapter, presents a discussion of anion-templation in systems contain-ing reversible bonds that give rise to dynamic combinatorial libraries (either by formationof coordination metal–ligand bonds or by reversible covalent bonds).

Keywords Anion template · Dynamic combinatorial library · Molecular recognition ·Self assembly

1Introduction

The use of templates in chemistry is nowadays a widespread strategy forthe synthesis of complex molecules and supramolecular assemblies [1–4].A plethora of species ranging from macrocycles and cages to interlockedmolecules and imprinted polymers have been efficiently synthesised usingthis approach. In a reaction where several potential products can be formed,

176 R. Vilar

chemical templates pre-organize the molecular building blocks to favour theformation of products with specific nuclearity, geometry and overall struc-ture. The interactions between the building blocks and the correspondingtemplate are generally of a non-covalent nature (e.g., hydrogen bonding,electrostatic interactions, coordination bonds or π-π stacking). In order fora templated process to successfully yield the targeted product, the structuraland electrostatic properties of the template need to be carefully selected.From a structural point of view, both the size and geometry of the templatehave to be considered, while electrostatically the choice is restricted to neu-tral, positive or negatively charged species. While cations have been widelyused as templates in synthetic chemistry, the role of anion templates did notstart to be exploited until relatively recently. In spite of the initial reservationsregarding the high solvation energies, sizes and pH dependency of anions, thepast few years have shown the great potential these species have as templatesin a wide range of synthetic routes [5–10].

In a templated reaction, once the building blocks have been pre-organisedby the template, they can be linked together by irreversible covalent bonds orby reversible interactions (either covalent or supramolecular). Initially, mosttemplated reactions aimed at forming irreversible covalent bonds betweenthe building blocks being brought together by the template. However, morerecently, a different approach to templated processes has been developed inwhich the building blocks are linked together by reversible bonds which canrapidly form and break. If in this reaction mixture there are several differ-ent building blocks that can react with each other in a reversible fashion,then a dynamic equilibrium between all the possible combinations of the mo-lecular components can be established. This approach, known as dynamiccombinatorial chemistry (DCC), can then lead to the formation of virtuallibraries of compounds – named dynamic combinatorial libraries (DCL) –that are thermodynamically controlled [11–15]. As with any other chem-ical equilibrium, different external stimula (e.g., temperature, concentrations,addition of a template) can modify the equilibrium between the componentsof a DCL. Amongst the most efficient ways of modifying the compositionof these virtual combinatorial libraries, is by adding a template which canshift the equilibrium towards the formation of one particular assembly ormolecule over others. This process is known as amplification.

The concept of dynamic combinatorial chemistry was pioneered bySanders [16, 17] and Lehn [15, 18, 19] in the 1990s. Since then, a wide rangeof reports have appeared in which different templates have been used to am-plify specific components of a dynamic library. In most cases, the templatesused are either cationic or neutral species leading to the amplification of thecorresponding receptors for these types of species. Considering that aniontemplates have only started to be used recently in synthesis, it is not surpris-ing that there are only a handful of examples where DCLs are amplified byanionic species.

Anion Templates in Synthesis and Dynamic Combinatorial Libraries 177

This chapter aims to provide an update on the role of anions as templates.The review is divided in two main sections: (a) anion-templated synthesisof assemblies linked together by irreversible bonds (or bonds that are inertunder mild experimental conditions); (b) anion templates in systems wherethe bonds linking the components are reversible and lead to anion-controlleddynamic combinatorial libraries. Since some comprehensive reviews in thearea of anion templation have appeared over the past few years [5–7], thischapter will mainly focus on papers published recently and will aim to showthe principles of anion templation rather than being a comprehensive accountof the literature. In addition, the scope of the chapter will be restricted tofinite assemblies (molecular or supramolecular) and not polymeric (for a re-view on molecularly imprinted polymers using anions see Steinke’s chapterin this volume).

2Templating Effects in Chemistry

In the 1960s, Busch carried out pioneering work on metal-directed synthe-ses of macrocycles establishing the concept of chemical template. As definedby him “a chemical template organizes an assembly of atoms, with respectto one or more geometric loci, in order to achieve a particular linking ofatoms” [20–22]. This provides an efficient route to prepare a specific molecu-lar assembly when several others can be potentially formed. Ideally templatesshould be removed from the final product once the reaction has reached com-pletion; however, templates often form an integral part of the final product;hence, they cannot always be removed from it.

A templated process can be driven thermodynamically or kinetically [23].In the former case, the template binds more strongly to one of the prod-ucts present in an equilibrium (i.e., a mixture under thermodynamic control)shifting the reaction towards the formation of this specific product which isthen obtained in higher yields. As will be discussed later in this review, this isparticularly important in the generation of dynamic combinatorial libraries.On the other hand, kinetic templates operate under irreversible conditions bystabilising the transition state leading to the final product.

In a templated process the interactions between the directing group andthe building blocks can be either covalent or non-covalent. The latter canmake use of a wide range of supramolecular interactions, such as hydrogenbonding, π-π stacking, electrostatic interactions and hydrophobic effects.From these, hydrogen bonding interactions are particularly useful since theyare relatively strong and, more importantly, are directional. In fact, hydrogenbonding interactions are employed as the main driving force in many of Na-ture’s templated processes. One of the most elegant examples of this is thereplication and transcription of nucleic acids.

178 R. Vilar

3Recent Examples of Anion-Templated Processes

Since the first examples of anion-directed assemblies were reported in the1990s, a wide range of molecular and supramolecular systems have been suc-cessfully prepared using anions as templates. This section has been divided bythe type of molecular or supramolecular species formed and, as mentionedbefore, it will mainly focus on systems that have been recently reported inthe literature (making reference to previous examples when important for thesake of clarity and completion).

3.1Macrocycles

Böhmer has recently shown that the presence of chloride can dictate thesize of macrocyclic poly-urea systems. The reaction between diamine 1 anddiisocyanate 2 yield the macrocyclic species 3 and 4 in a 5:1 ratio (seeScheme 1) [24].

When this reaction was repeated in the presence of two equivalents oftetrabutylammonium chloride, the formation of the larger macrocycle 4 wasfavoured over the trimeric species 3. In fact, the ratio reverted to 1:5 infavour of the larger macrocycle. Crystals of macrocycle 4 were obtained fromthe crude of the first reaction when grown in the presence of tetrabutyl-ammonium chloride. The structure (Fig. 1) revealed an interesting host-guestcomplex in which two chlorides are bound by the macrocycle (which explainsthe need of two equivalents of chloride to favour the formation of the hexam-eric compound).

Alfonso and Luis have recently reported an example of anion-templatedsynthesis of a pseudopeptidic macrocycle. The reaction between diamine 5and dialdehyde 6 was carried out (see Scheme 2) [25]. 1H NMR spectroscopyshowed a complex patter of signals together with protons associated withaldehyde and methoxyamine suggesting the presence of a range of differentmacrocyclic and acyclic products. Addition of different anions to the reactionmixture had little effect to the distribution of products shown by 1H NMRspectroscopy. However, when the reaction was repeated in the presence tereph-thalate, the almost quantitative formation of a single product as revealed by1H NMR spectroscopy was observed. This intermediate product has beenpostulated to be host-guest complex 7· Terephthalate). This imine-containingmacrocycle was then reduced to the corresponding amine to yield macrocycle8, which was isolated and fully characterised. In the first step of this reaction,terephthalate is expected to form hydrogen bonding interactions between itsanionic carboxylates and the amides of the macrocycle. In addition, π-π in-teractions between the aromatic ring of the terephthalate and that one of thedi-aldehyde are likely to also play a role in the templating process.


Scheme 1 Synthetic procedure for the preparation of macrocycles 3 and 4

Fig. 1 X-ray crystal structure of macrocycle 4 highlighting the two chloride anions (solidspheres) bound to the macrocycle

180 R. Vilar

Scheme 2 Terephthalate-templated synthesis of macrocycle 8

It is worth noting that this process could in principle lead to dynamic com-binatorial libraries of products since the type of bond that brings togetherthe different components in the reaction is a reversible one (namely imineformation). In fact, the initial mixture of products observed by the authorshas been postulated to be a mixture of different-sized macrocycles and linearspecies (from which one of them is amplified upon addition of terephthalate).More detailed studies would be needed to determine whether this system in-deed leads to the formation of a DCL of receptors (see Sect. 4 for examples ofanion-directed DCLs).

A widely used approach to the synthesis of large macrocycles and cages (seeSect. 3.2) is by self-assembly of metal centres and polydentate ligands. Carefulchoice of the geometry around the metal and the number and relative positionof the coordinating groups on the bridging ligands can generate large 2D and3D assemblies from a one-pot reaction. Often, the addition of a template to thereaction mixture provides a more efficient path for the formation of one spe-cific assembly. Some recent examples of anion-templated metallo-macrocyclesand -cages are discussed in this and the following sections.

Lippert has shown that the assembly between platinum(II) centres andpurine bases can be controlled by specific anions. In analogy to the hydrogen-bonded tetrads that guanine bases can form, this group synthesised a metallo-square in which the purine bases are interconnected by coordination to plat-inum(II) centres rather than hydrogen bonding interactions (see Fig. 2) [26].


Fig. 2 Schematic representation of: a guanine quartet with a cationic guest and b metallo-square 9 based on platinum(II) and methylpurine with an anionic guest

Square 9 formed over a period of 5 days by self-assembly process of fourunits of [(NH3)2Pt(Pur)(H2O)]2+ (where Pur = 9-methylpurine). Interest-ingly, in this process the formation of a second species with a triangulargeometry was observed (see Scheme 3). NMR experiments showed that thetriangular metallo-macrocycle 10 is favoured (in a 0.6:1 ratio) to square 9.However, if the self-assembly process is carried out in the presence of SO4

2–,the proportion of the two species changes and the square is favoured over thetriangle in a 2.5:1 ratio. This change in the preference for the square over thetriangle has been attributed to the templating properties of the sulfate anion.

Scheme 3 Reaction scheme for the formation of metallo-triangle 10

Maekawa and Kitagawa have recently reported an interesting anion-directed approach to the synthesis of bowl-shaped metallo-macrocycles [27].Initially, they observed that the reaction between [Cu(MeCN)4](PF6) and 4-(2-pyridyl)pyrimidine (pprd) under an atmosphere of C2H4 and in acetoneyielded the coordination polymer {[Cu(pprd)(C2H4)](PF6)}n (see Scheme 4).

182 R. Vilar

Scheme 4 Formation of a coordination polymer from the reaction between [Cu(MeCN)4](PF6) and 4-(2-pyridyl)pyrimidine (pprd)

When the same reaction was carried out in methanol the macrocyclic species[Cu4(pprd)4(C2H4)4](PF6)4 (11) was formed instead. The crystal structure ofthis metallo-assembly revealed a bowl-shaped structure with the PF6

– anionpositioned at its centre (see Fig. 3).

Fig. 3 X-ray crystal structures of a the tetra-copper assembly [Cu4(pprd)4(C2H4)4](PF6)4(11) and b the tri-copper assembly {[Cu3(pprd)3(C2H4)3](ClO4)3}3 (12)

Interestingly, the reaction between [Cu(C2H4)n]ClO4 and pprd in Me2COunder a C2H4 atmosphere yielded the trinuclear metallo-macrocycle {[Cu3(pprd)3(C2H4)3](ClO4)3}3 (12). Although the reactions to obtain the twodifferently-sized macrocycles are not identical, it is plausible to suggest thatthe volume of the anions dictates the size of the macrocycle. With PF6, thetetranuclear metallo-macrocycle forms while the smaller tetrahedral ClO4 an-ion directs the formation of the tri-copper macrocycle.

3.2Cages and Capsules

In the late 1990s we reported one of the first examples of anion-templatedmetallo-cage [28–30]. The system was based on the chloride-directed assem-


bly of eight units of amidinothiourea (H-atu, see Scheme 5) and six nickel(II)centres to yield [Ni6(atu)8⊂Cl]Cl3 (13). The resulting assembly showed tocontain a completely encapsulated chloride interacting via hydrogen bondingand weak metal-anion interactions with the cage.

Scheme 5 Reaction scheme for the synthesis of hexanickel cage 13

The approach described above was expanded afterwards to incorporatea second type of metal within the framework yielding the mixed Ni-Pd andNi-Pt complexes [Ni4M2(atu)8⊂Cl]Cl3 (M = Pd, 15; Pt, 16) (see Fig. 4). Thiswas carried out by first preparing [Ni(atu)2] (14) and then reacting it with thecorresponding palladium(II) or platinum(II) salt. As in the case of the hexa-nickel cage, it was found that the mixed-metal cage would only form in thepresence of the appropriate halide, namely chloride or bromide.

The assembly of the nickel(II)/H-atu cage 13, is accompanied by a dramaticcolour change, from orange to green. Considering that in methanol only chlo-ride acts as a template for the formation of cage 13, we have recently employedthis anion-templated process to develop a colorimetric chemical sensor forchlorides [31].

Bidentate pyrazolyl-based ligands have shown to be versatile buildingblocks for the synthesis of a plethora of coordination assemblies. Initial workby McCleverty and Ward in the late 1990s showed that the synthesis of cages[Co4(Ln)6⊂X](X)7 (where L1 and L2 = bidentate pyrazolyl-pyridine ligands;

184 R. Vilar

Fig. 4 X-ray crystal structure of mixed-metal cage [Ni4Pt2(atu)8⊂Cl]Cl3 showing the en-capsulated chloride anion

X = BF4–, ClO4

–; see Figs. 5 and 6) was dependant on the presence of specificanions [32]. Detailed investigations (by NMR spectroscopy) of the assem-bly process in solution demonstrated that the tetrahedral BF4

– and ClO4–

anions indeed act as templating agents for the formation of these metallo-assemblies [33].

More recently, Ward has explored the effect that the length of bidentatepyrazolyl-based ligands has on the formation of the metallo-cages [34]. When

Fig. 5 A selection of bidentate pyrazolyl-based ligands employed for the synthesis ofmetallo assemblies


Fig. 6 X-ray crystal structure of [Co4(L1)6⊂(ClO4)](ClO4)7 highlighting the encapsulatedtetrahedral anion (in space-fill representation)

L3 was used as ligand, the expected [Co4(L3)6⊂X](X)7 (X = BF4–, ClO4

–,PF6

–, I–) cages were formed, but in this case the anion did not seem to playa determinant role in defining the geometry of the final assembly. As indi-cated by the authors, the resulting cage is sufficiently large to leave gaps in thecentre of the faces through which the encapsulated anion can easily exchangewith the external anions.

186 R. Vilar

Amouri has reported the anion-directed synthesis of a series of coordi-natively unsaturated metallo-cages with general formula [Co2(L4)4(RCN)2⊂(BF4)](BF4)3 (R = Me, Et, Ph) [35]. The X-ray crystal structures of some ofthese assemblies (see Fig. 7 for an example) have revealed the presence of anencapsulated BF4

– anion which interacts with the coordinatively unsaturatedcobalt(II) centres. Interestingly when analogous reactions were performed inthe presence of other anions (such as Cl and NO3) different metal-organicassemblies were formed [36].

Fig. 7 X-ray crystal structure of [Co2(L4)4(MeCN)2⊂(BF4)](BF4)3 showing the encapsu-lated anion and its interactions with the metal centres

3.3Interlocked Species

Molecular interlocked systems such as catenanes and rotaxanes can also beprepared using anion-directed approaches. Although anionic templates didnot make their way into this area until relatively recently, nowadays there areseveral examples that demonstrate the utility of this approach for the synthe-sis of this topologically interesting species.

The first examples of interlocked species synthesized by anion tem-plated approaches were the pseudorotaxanes and rotaxanes reported in thelate 1990s by Stoddart and Vögtle. Stoddart reported that by mixing fourequivalents of [NH2(CH2Ph)2][PF6] with one equivalent of the macrocycletetrakis-p-phenylene[68]crown-20, the quadruply-stranded pseudorotaxane(17) formed (see Fig. 8) [37]. This assembly was structurally characterized re-vealing the presence of a PF6

– anion at its center forming multiple C–H· · ·Fhydrogen bonds with the hydroquinone methine and the benzylic methylenehydrogen atoms.


Fig. 8 Schematic representation of quadruply-stranded pseudorotaxane 17

Using a different approach, Vögtle successfully showed that organic an-ions can induce the formation of rotaxanes [38–41]. In this approach a stronghost-guest complex between a tetralactam macrocycle and a phenolate anionis first formed (see Scheme 6). In this assembly the phenolate anion is posi-tioned at the center of the ring to further react with a second component (e.g.,an alkyl bromide or acyl chloride) yielding a rotaxane. The negatively chargedphenolic functionality can be located either at the stopper component or atthe axle precursor providing a range of different possibilities for the synthesisof the interlocked species.

More recently, Beer has developed a series of synthetic procedures for thehalide-templated syntheses of pseudorotaxanes, rotaxanes and catenanes. Thismethodology is based on combining the recognition of halides by a hydrogen-bonding host (e.g., a macrocycle) with ion-pairing between the correspondinghalide and a cationic species (see Fig. 9). These two interactions allow for theinterpenetration of the two molecular components to yield the interlockedmolecules. Some examples of the macrocycles and axels employed to generatethe corresponding pseudorotaxanes are shown in Fig. 10 [38–41].

A similar approach was later employed to prepare rotaxanes by “clipping”(via ruthenium-catalysed olefin metathesis) the acyclic hydrogen bonding

188 R. Vilar

Scheme 6 Schematic representation of the anion templated synthesis of rotaxanes basedon tetralactam macrocycles and a phenolate anions

Fig. 9 Schematic representation of the anion-templated synthesis of interlocked speciesbased on combining halides recognition by a hydrogen-bonding host with ion-pairing

species 18 with the corresponding axel 19 in the presence of chloride (seeScheme 7) [42]. The resulting rotaxane (20) has been structurally characterisedshowing the strong binding between the interlocked species and the templatingchloride (see Fig. 11).

Using the same general strategy described above, Beer reported the firstanion-templated synthesis of catenanes [43, 44]. Mixing macrocycle 21 with


Fig. 10 Selection of macrocycles and axels employed to synthesise pseudorotaxanes androtaxanes

Scheme 7 Reaction scheme for the chloride-templated synthesis of rotaxane 20

190 R. Vilar

Fig. 11 X-ray crystal structure of rotaxane 20 showing the chloride anion bound to theinterlocked species

Scheme 8 Reaction scheme for the chloride-templated synthesis of catenane 23

Fig. 12 X-ray crystal structure of catenane 23 showing the chloride anion bound to theinterlocked species


the acyclic hydrogen-bonding molecule 22 in the presence of chloride asa template, led to the formation of [2]catenane 23 (see Scheme 8). As for ro-taxane 20, the crystal structure of this interlocked species revealed the stronginteraction between the templating chloride and the two macrocyclic compo-nents of the catenane (Fig. 12).

For a comprehensive review of this area, see the recent reviews publishedby Beer [7, 9, 10].

4Anions as Templates in Dynamic Combinatorial Chemistry

One of the seminal papers in defining the concept of dynamic combinatorialchemistry was published by Lehn in the 1990s [45] In this work it was shown

Scheme 9 Chloride-templated synthesis of circular helicate 24

192 R. Vilar

that the assembly of iron(II) salts and a tris-bipy ligand (L5) is a dynamicprocess that can yield a range of different metal helicates. Which assemblyis formed, is highly dependent on the nature of the counter-anions presentin solution. With FeCl2, the pentanuclear circular helicate [Fe5(L5)5Cl]9+ (24)was formed in high yields (see Scheme 9), while a mixture of the penta-and hexa-nuclear helicates were obtained in the presence of bromide (andonly the hexa-nuclear helicate [Fe6(L5)6(SO4)]10+ (25) was formed with sul-fate). Further studies by the same authors demonstrated that in the reactionwith FeCl2 (and also in the analogous one with NiCl2) a linear helicate isformed first (i.e., the kinetic product) which progressively converts into thethermodynamic circular helicate product [Fe5(L5)5Cl]9+ [46] Structural char-acterization of 25 confirmed the assembly to be a circular double helix witha chloride ion located in the central cavity.

Although this anion-directed system was one of the first examples of DCCsreported in the literature, there are in fact very few known systems to datewhere negatively charged species are used to amplify the formation of a spe-cific assembly from a dynamic combinatorial library. Such examples will bereviewed in Sects. 4.1 and 4.2.

4.1Using Metal–Ligand Coordination Bonds

Dunbar has elegantly demonstrated the use of anionic templates for the syn-theses of a range of nickel(II) and zinc(II) metalla-cyclophanes using 3,6-bis(2-pyridyl)-1,2,4,5-tetrazine (bptz) as bridging ligand (see Scheme 10) [47,48]. Anions such as BF4

– and ClO4– induce the formation of the tetra-metallic

square assemblies [{M4(bptz)4(CH3CN)8}X](X)7, (M = ZnII, NiII; X = BF4–,

ClO4–), while the larger octahedral anion SbF6

– templates the formationof the molecular pentagon [{Ni5(bptz)5(CH3CN)10}SbF6](SbF6)9. The X-raycrystal structures of these species (see Figs. 13 and 14) have shown that inboth the squares and pentagon one anion is encapsulated at the centre of thecorresponding metalla-cyclophane (displaying anion-π interactions betweenthe O and F atoms of the anions and the tetrazine rings of bptz).

Further studies by the same authors showed that the molecular pentagoncan be easily converted into the corresponding molecular square in the pres-ence of excess BF4

– and ClO4–. The conversion of the molecular pentagon to

the square is also observed upon addition of iodide (which due to its large sizeand polarizability it can adopt the directionality of a tetrahedral anion).

In contrast to the above, the conversion of the nickel square to the corres-ponding pentagon upon addition of excess SbF6

– is not readily observed (onlypartial conversion takes place). This apparent higher stability of the molecu-lar square in comparison to the molecular pentagon has been attributed tomore strain in the pentagon. This is supported by the X-ray crystal structureof the metallo-pentagon in which the btpz ligands are considerably bent.


Scheme 10 Reaction scheme showing the anion-directed synthesis of metallo-macrocyclesand their interconversion

Fig. 13 X-ray crystal structure of metallo-square [{Ni4(bptz)4(CH3CN)8}BF4](BF4)7

More recently, Fujita has reported an interesting example of anion-controlled dynamic equilibrium of palladium-containing assemblies [49].The reaction of ligands L6 and L7 with palladium(II) was investigated underdifferent experimental conditions, namely in different solvents, at differ-ent concentrations of the ligand and in the presence of different counter-anions.

194 R. Vilar

Fig. 14 X-ray crystal structure of metallo-square [{Ni5(bptz)5(CH3CN)10}SbF6](SbF6)9

In this study it was shown that the nature of the resulting metallo-assemblies was highly dependant on the conditions employed. For example,the reaction between L6 and [Pd(en)(NO3)2] in DMSO was found to yield twomacrocycles (26 and 27) in roughly 60:40 proportions when the concentra-tion of the ligand was 5 mM. By reducing the concentration of the ligand to1 mM, the equilibrium was shifted to the simpler [2+2] assembly 26 whichwas present in nearly 90%.

Increasing the concentration of ligand to 20 mM or above, yielded yet an-other product, metallo-assembly 28 which at 500 mM concentration of ligandwas practically the only product observed by 1H NMR spectroscopy.

In the same paper, the reactions between these two ligands and “naked”palladium(II) cations were also discussed. Interestingly, the structure and nu-clearity of the resulting assemblies was shown to depend on the counteranionof the palladium salt used. Thus, the reaction between Pd(NO3)2 and ligand

196 R. Vilar

Fig. 15 X-ray crystal structure of metallo-tetrahedron 29

L6 yielded mainly the tetrahedral assembly 29 (see Fig. 15). An analogous as-sembly was obtained in the presence of BF4

–. However, when the reactionwas carried out in the presence of triflate a double-walled triangle (30) wasobtained as the major product (Scheme 11).

Scheme 11 Schematic representation of double-walled metallo-triangle

When carrying out a similar reaction with ligand L7 (which is longer thanL6) the structures of the resulting metallo-assemblies were again found to beanion-dependant. In this case, dynamic equilibrium between the two assem-blies 31 and 32 was observed. With nitrate as the counter-anion, a roughly 1:1mixture of the two assemblies was observed; however, with triflate the main


Scheme 12 Schematic representation of the equilibrium between the double-walledmetallo-triangle 31 and metallo-tetrahedron 32

product formed was the double-walled triangle. In contrast in the presence ofthe aromatic p-tosylate anion, the formation of the M4L8 assembly was favored.

A similar approach to that reported by Fujita for the generation of dynamiccombinatorial libraries of metallo-assemblies has been developed in our owngroup [50]. In contrast to most of the bipyridyl-based ligands employed forthe synthesis of metallo-macrocycles reported so far, we were interested inusing bipyridyl ligands containing spacers with hydrogen bonding function-alities such as ligands L8 and L9. It was rationalized that having hydrogenbonding donor groups would aid in the interaction with potential anionicguests/templates.

The reactions between each of these two ligands and [Pd(dppp)(OTf)2](dppp = 1,3-bis(diphenylphosphino)propane) were studied aiming at pro-

198 R. Vilar

Scheme 13 Reaction scheme for the synthesis of metallo-macrocycles 33 and 34

ducing cyclic metallo-assemblies (see Scheme 13). The combination of theseligands and the metal complex (in 1:1 ratios) could in principle give a rangeof different cyclic and acyclic materials. Since palladium-pyridine bonds arerelatively labile, it was expected that a dynamic equilibrium of different as-semblies would be established.

Crystals were grown from the corresponding reaction mixtures. The X-raycrystal structures obtained revealed that the crystallised products correspondto [2+2] assemblies with general formula [Pd(dppp)(L)]2(OTf)4 (L = L8, 33;L = L9, 34) and a “bowl-type” structure (see Fig. 16).

Fig. 16 X-ray crystal structure of assembly 34 showing the triflate anion at the centre ofthe bowl-type assembly


Although the solid-state structures obtained for these systems indicate theformation of the [2+2] assembly, 1H NMR studies showed that more than onespecies co-existed in solution. As indicated above, in principle a 1:1 mixtureof the bis-pyridyl ligands and cis-[Pd(dppp)]2+ centres could yield macrocy-cles of different sizes or a range of different acyclic products. Several 1H NMRspectroscopic and ESI-mass spectrometric studies were carried out (using[Pd(dppp)(OTf)2] and L8) to establish the behaviour of the system in solu-tion. These studies suggested that there is an equilibrium between two specieswhich have been assigned to the [2+2] and a [3+3] metallo-assemblies (seeScheme 14).

Scheme 14 Reaction scheme showing the equilibrium between the [2+2] and [3+3]metallo-macrocycles

Interestingly, this equilibrium can be shifted by modify the experimentalconditions such as the solvent, concentration of ligand and temperature. Fur-thermore, the equilibrium between the different metallo-assemblies presentin solution can also be shifted by the presence of different anions. For ex-ample, addition of several equivalents of H2PO4

– to a DMSO solution con-taining a mixture of [2+2] and [3+3] assemblies, shifted the equilibrium tothe formation of only the [2+2] assembly. Surprisingly, addition of HSO4

–

to the 1:1 mixture in DMSO of [Pd(dppp)(OTf)2] and L8 did not modify theequilibrium between the [2+2] and [3+3].

Williams has reported another type of system in which a dynamic combi-natorial library of metal complexes is generated by metal-ligand interactionsand controlled by anionic templates [51]. In this work it was shown thatchloride can modify the distribution of products in an equilibrated solutioncontaining cobalt(II) salts and 2,2′-bipyridyl ligands (bipy or the chiral ligandL10 – see Scheme 15). More specifically, mixing Co(NO3)2 with bipy and L10

generated a library of complexes with general formula [Co(bipy)X(L10)3–X]2+.This was shown by electrospray mass spectrometry which revealed thatall the possible combinations of products were indeed present in solution:

200 R. Vilar

Scheme 15 Reaction scheme showing the different possible complexes that can be formedby mixing Co(NO3)2 with L10 and bipy

[Co(L10)3]2+ (80%), [Co(bipy)(L10)2]2+ (100%), [Co(bipy)2(L10)]2+ (91%)and [Co(bipy)3]2+ (11%). Furthermore, 1H NMR spectroscopy indicated thatfor each of the complexes – except for [Co(bipy)3]2+ which gave an enan-tiomeric pair – the corresponding ∆- and Λ-diastereomers were present.

The equilibrium of the above mixture was shown to change upon add-ition of CF3COOH. Addition of the acid led to protonation of the aminesgroups on L8 inducing diasteroselectivity and, as a consequence, some of thecomplexes initially present in the mixture disappeared. Interestingly, whenDCl rather than CF3COOH was added to the reaction mixture, only the twohomoleptic complexes {Cl2⊂∆-[Co(L10H2)3]6+} and [Co(bipy)3]2+ could bedetected suggesting that chloride acts as a template amplifying the formationof these species. These observations have been rationalised on the groundsthat coulombic repulsion between the protonated amines is minimized by thepresence of chloride.

A similar approach to the one discussed above has been recently reportedby Rice [52, 53]. In these investigations it was shown that nitrate can mod-ify the distribution of products in a mixture of cobalt(II) and two differentN,N′ chelating ligands. First, the reaction between Co(ClO4)2 and ligand L11

was studied showing that a triple helicate with formula [Co2(L11)3](ClO4)4

Scheme 16 Reaction scheme showing the different possible complexes that can be formedby mixing Co(ClO4)2 with L11 and L12


formed (see Scheme 16). The ligands around the cobalt(II) centres generate“pockets” of the right size to bind perchlorate anions via hydrogen bonding(see Fig. 17).

Fig. 17 X-ray crystal structure of the triple helicate [Co2(L11)3](ClO4)4 showing the per-chlorate anions bound to the “pockets” formed by the three ligands

The authors then studied the possibility of anion exchange in this complex.Thus, upon addition of two equivalents of [Bu4N][NO3] to [Co2(L11)3](ClO4)4a new host-guest complex with formula [Co2(L11)3](ClO4)2(NO3)2 was ob-tained. The X-ray crystal structure of this mixed-anion helicate showed thatthe perchlorates initially bound to the binding pockets of the helicate, hadbeen replaced by nitrates.

Having established the basic host-guest chemistry between the helicateand the two anions, the authors then investigated the reaction between[Co(ClO4)2]·6H2O, L11 and L12 (see Scheme 16 for the chemical structure ofthe ligands) in a 2 to 1.5 to 1.5 ratio. This reaction resulted in the forma-tion of four complexes: [Co2(L11)3]4+, [Co2(L11)2(L12)]4+, [Co2(L11) (L12)2]4+

and [Co2(L12)3]4+, with a 1:3:3:1 statistical distribution. This was confirmedby both 1H NMR spectroscopy and ES mass spectrometry. Interestingly,upon addition of KNO3 to this mixture a dramatic change in product dis-tribution was observed. The two homoleptic complexes [Co2(L11)3]4+ and[Co2(L12)3]4+ were found to be the main components of the mixture (withonly 5% of the heteroleptic compounds being present). This behaviour hasbeen attributed to the strong binding of nitrate (which acts as a template) tothe anion-binding pockets present in [Co2(L11)3]4+.

4.2Using Reversible Covalent Bonds

A large number of dynamic combinatorial libraries reported in the literaturemake use of reversible covalent chemistry to generate the required dynamicequilibria [12]. In spite of this, there are very few examples of anion-directed

202 R. Vilar

DCLs where this type of reversible bond-formation is used to generate thevirtual library of receptors. One of these examples has been reported byOtto and Kubic who developed a DCL of anion receptors [54]. The libraryis based on disulfide exchange reactions between a dimeric cyclic peptide(35) – where the two peptidic cycles are linked by a disulfide group – anda range of different thiol-substituted spacers a–f (see Scheme 17). Mixing ofall these components yields a range of dimeric cyclic peptides in differentproportions.

Scheme 17 Components of a DCL of dimeric cyclic peptides linked by disulfide bonds

Interestingly, addition of K2SO4 or KI to the DCL amplified the formationof three receptors (35a, 35b and 35c), while addition of other anions suchas chloride or fluoride did not shift the equilibrium. Two of these receptors(35b and 35c) were amplified more than 35a; these were then chosen for fur-ther studies and therefore prepared and isolated using a second generationbiased library. Isothermal titration calorimetry studies revealed the bindingconstants between the selected dimeric-receptors and sulfate or iodide to bearound 106 (in a mixture of MeCN/H2O). This example nicely demonstrateshow dynamic combinatorial chemistry can be employed to improve the selec-tivity and binding properties of a specific receptor; in this case this has beendone by simply modifying the length and flexibility of the spacer that joinsthe two macrocycles.

A different type of reversible covalent bond has been employed by Sesslerto generate anion-directed DCLs. More specifically, the synthesis of new


bipyrrole-based macrocyclic receptors has been achieved by reacting diamine36 with diformylbipyrrole (37) in acidic media (see Scheme 18) [55]. Interest-ingly, depending on the acid used (namely HCl, HBr, CH3CO2H, CF3CO2H,H3PO4, H2SO4 or HNO3) different distributions of oligomeric species andmacrocycles were obtained. While several products were formed with mostacids, in the presence of sulfuric acid the [2+2] macrocycle 38·2H2SO4formed in nearly quantitative yield.

Scheme 18 Synthesis of macrocycle 38·2H2SO4 by reacting 36 and 37 in the presence ofH2SO4

The anion-free macrocycle 38 was isolated upon addition of triethylamineto 38·2H2SO4. This macrocycle was structurally characterised and its anion-binding properties studied, revealing high association constants between thereceptor and HSO4

– (1:1; Ka = 63 500±3000 M–1) and H2PO4– (2:1; Ka1 =

191 000±15 400 M–1 and Ka2 = 60 200±6000 M–1). Interestingly, when 38 wasallowed to stand in acetonitrile for 5 days in the presence of HSO4

– or H2PO4–

(as tetrabutylammonium salts) a rearrangement of the poly-imine compoundtook place. More specifically, the [2+2] macrocycle 38 expanded into the[3+3] macrocycle 39 (see Fig. 18) quantitatively in the presence of H2PO4

–

(and in 47% yield in the presence of HSO4–). More recently, the same authors

have published a comprehensive study in which they show a similar be-haviour for other related di-amino and di-aldehyde building blocks [56, 57].The nature of the anions (from the corresponding acid) present in solution,dictates the size of the macrocycle formed. Under specific circumstances,some of the macrocycles undergo anion-induced ring-expansion.

204 R. Vilar

Fig. 18 Schematic representation of the [3+3] macrocycle 39

The results discussed above, strongly suggest the presence of a dynamicequilibrium between various macrocyclic species in solution. Upon additionof the appropriate anionic template, amplification of one of them is thenobserved.

5Conclusions and Outlook

The first examples of anion-templated processes were reported in the early1990s. Since then, this area of supramolecular chemistry has grown steadilyshowing the important role that anionic templates can play in directingthe synthesis of specific molecules and supramolecular assemblies. A widerange of organic and metal-organic macrocycles have now been prepared byanion-templated processes. Similarly, the syntheses of molecules with com-plex topologies (such as those of pseudorotaxanes, rotaxanes and catenanes)can now be achieved by anion-directed assembly of simple building blocks.Another area where anion templates have had an important impact is in thesynthesis of coordination cages and capsules. More recently, the use of anionsas templates in dynamic combinatorial libraries has been realized showing


that structurally challenging anion-receptors can be developed using thisdynamic approach. This promises to be an area of important future develop-ments with the potential to generate receptors that can not be easily achievedusing more “classical” synthetic methodologies.

References

1. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry II. Top CurrChem, vol 249

2. Schalley CA, Voegtle F, Doetz KH (eds) (2005) Templates in Chemistry I. Top CurrChem, vol 248

3. Vilar R (2004) Struct Bond 111:854. Diederich F, Stang PJ (eds) (2000) Templated Organic Synthesis. Wiley VCH, Wein-

heim5. Gimeno N, Vilar R (2006) Coord Chem Rev 250:31616. Vilar R (2003) Angew Chem Int Ed 42:14607. Beer PD, Sambrook MR, Curiel D (2006) Chem Commun, p 21058. Beer PD, Wong WWH (2005) Macrocyclic Chem, p 1059. Lankshear MD, Beer PD (2006) Coord Chem Rev 250:3142

10. Lankshear MD, Beer PD (2007) Acc Chem Res 40:65711. Corbett PT, Leclaire J, Vial L, West KR, Wietor J-L, Sanders JKM, Otto S (2006) Chem

Rev 106:365212. Rowan SJ, Cantrill SJ, Cousins GRL, Sanders JKM, Stoddart JF (2002) Angew Chem

Int Ed 41:89913. Otto S (2003) Curr Opin Drug Discov Dev 6:50914. Lehn J-M (2007) Chem Soc Rev 36:15115. Lehn J-M, Eliseev AV (2001) Science 291:233116. Otto S, Furlan RLE, Sanders JKM (2000) J Am Chem Soc 122:1206317. Rowan SJ, Lukeman PS, Reynolds DJ, Sanders JKM (1998) New J Chem 22:101518. Eliseev AV, Lehn JM (1999) Curr Topics Microbiol Immunol 243:15919. Lehn J-M (1999) Chem Eur J 5:245520. Melson GA, Busch DH (1964) J Am Chem Soc 86:483421. Thompson MC, Busch DH (1964) J Am Chem Soc 86:365122. Busch DH (1992) J Inclusion Phenom 12:38923. Anderson S, Anderson HL, Sanders JKM (1993) Acc Chem Res 26:46924. Meshcheryakov D, Boehmer V, Bolte M, Hubscher-Bruder V, Arnaud-Neu F, Hersch-

bach H, Van Dorsselaer A, Thondorf I, Moegelin W (2006) Angew Chem Int Ed45:1648

25. Bru M, Alfonso I, Burguete MI, Luis SV (2006) Angew Chem Int Ed 45:615526. Roitzsch M, Lippert B (2006) Angew Chem Int Ed 45:14727. Maekawa M, Konaka H, Minematsu T, Kuroda-Sowa T, Munakata M, Kitagawa S

(2007) Chem Commun, p 517928. Vilar R, Mingos DMP, White AJP, Williams DJ (1998) Angew Chem Int Ed 37:125829. Vilar R, Mingos DMP, White AJP, Williams DJ (1999) Chem Commun, p 22930. Cheng S-T, Doxiadi E, Vilar R, White AJP, Williams DJ (2001) J Chem Soc Dalton

Trans, p 223931. Diaz P, Mingos DMP, Vilar R, White AJP, Williams DJ (2004) Inorg Chem 43:7597

206 R. Vilar

32. Fleming JS, Mann KLV, Carraz C-A, Psillakis E, Jeffery JC, McCleverty JA, Ward MD(1998) Angew Chem Int Ed 37:1279

33. Paul RL, Bell ZR, Jeffery JC, McCleverty JA, Ward MD (2002) Proc Nat Acad Sci USA99:4883

34. Paul RL, Argent SP, Jeffery JC, Harding LP, Lynam JM, Ward MD (2004) Dalton Trans,p 3453

35. Amouri H, Mimassi L, Rager MN, Mann BE, Guyard-Duhayon C, Raehm L (2005)Angew Chem Int Ed 44:4543

36. Amouri H, Desmarets C, Bettoschi A, Rager MN, Boubekeur K, Rabu P, Drillon M(2007) Chem Eur J 13:5401

37. Fyfe MCT, Glink PT, Menzer S, Stoddart JF, White AJP, Williams DJ (1997) AngewChem Int Ed Engl 36:2068

38. Hubner GM, Glaser J, Seel C, Vogtle F (1999) Angew Chem Int Ed 38:38339. Reuter C, Schmieder R, Vogtle F (2000) Pure Appl Chem 72:223340. Reuter C, Wienand W, Hubner GM, Seel C, Vogtle F (1999) Chem Eur J 5:269241. Seel C, Vogtle F (2000) Chem Eur J 6:2142. Wisner JA, Beer PD, Drew MGB, Sambrook MR (2002) J Am Chem Soc 124:1246943. Sambrook MR, Beer PD, Wisner JA, Paul RL, Cowley AR (2004) J Am Chem Soc

126:1536444. Ng K-Y, Cowley AR, Beer PD (2006) Chem Commun, p 367645. Hasenknopf B, Lehn J-M, Kneisel BO, Baum G, Fenske D (1996) Angew Chem Int Ed

Engl 35:183846. Hasenknopf B, Lehn J-M, Boumediene N, Dupont-Gervais A, Van Dorsselaer A,

Kneisel B, Fenske D (1997) J Am Chem Soc 119:1095647. Campos-Fernandez CS, Schottel BL, Chifotides HT, Bera JK, Bacsa J, Koomen JM, Rus-

sell DH, Dunbar KR (2005) J Am Chem Soc 127:1290948. Campos-Fernandez CS, Clerac R, Dunbar KR (1999) Angew Chem Int Ed 38:347749. Chand DK, Biradha K, Kawano M, Sakamoto S, Yamaguchi K, Fujita M (2006) Chem

Asian J 1:8250. Diaz P, Tovilla JA, Ballester P, Benet-Buchholz J, Vilar R (2007) Dalton Trans, p 351651. Telfer SG, Yang X-J, Williams AF (2004) Dalton Trans, p 69952. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Dalton Trans,

p 239653. Harding LP, Jeffery JC, Riis-Johannessen T, Rice CR, Zeng Z (2004) Chem Commun,

p 65454. Otto S, Kubik S (2003) J Am Chem Soc 125:780455. Katayev EA, Pantos GD, Reshetova MD, Khrustalev VN, Lynch VM, Ustynyuk YA,

Sessler JL (2005) Angew Chem Int Ed 44:738656. Katayev EA, Boev NV, Khrustalev VN, Ustynyuk YA, Tananaev IG, Sessler JL (2007)

J Org Chem 72:288657. Katayev EA, Sessler JL, Khrustalev VN, Ustynyuk YA (2007) J Org Chem 72:7244

Struct Bond (2008) 129: 207–248DOI 10.1007/430_2008_084© Springer-Verlag Berlin HeidelbergPublished online: 25 April 2008

Molecularly Imprinted PolymersUsing Anions as Templates

Sally L. Ewen · Joachim H. G. Steinke (�)

Department of Chemistry, Imperial College London, South Kensington Campus,London SW7 2AZ, [email protected]

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2081.1 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

2 Molecularly Imprinted Polymers . . . . . . . . . . . . . . . . . . . . . . . 2092.1 The Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2092.2 Inspiration, Incorporation and Assimilation . . . . . . . . . . . . . . . . . 2102.3 A Brief “Developmental” History . . . . . . . . . . . . . . . . . . . . . . . . 2112.4 Scope and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2152.5 Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172.6 Design Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

3 Molecular Imprinting Using Anionic Templates . . . . . . . . . . . . . . . 2223.1 Anionic Phosphate Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . 2253.2 Carboxylates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2323.3 Anionic Sulfate Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 2383.4 Other Anions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2403.5 Concluding Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244

Abstract Molecularly imprinted polymers (MIPs) are a class of solid-phase artificial re-ceptors that are prepared using templates during the polymer network forming step; theseare subsequently removed to generate the selective receptor sites. The ease of synthesis,the possibility of nanomolar binding constants and high levels of molecular discrimina-tion, as well as environmental stability and ability to be reused, have led to a dramaticboost in research interest in MIPs. One particularly promising area of study is the useof anionic templates in the synthesis of MIPs and the targeting of substrates that carrybiologically important anionic functionalities. Benefiting from concurrent developmentsin supramolecular receptor design and synthesis, it has become clear that MIPs for anionrecognition will impact biological screening, diagnosis, point-of-care devices (includingonline sensing) and read-out.

Keywords Anionic templates · Molecularly imprinted polymers · Self-assembly ·Biomimicry · Synthetic receptors

AbbreviationsFIP Functional group imprinted polymerISFET Ion-sensitive field effect transistor

208 S.L. Ewen · J.H.G. Steinke

ITC Isothermal titration calorimetryITO Indium tin oxideMIP Molecularly imprinted polymerMPA Methylphosphonic acidNIP Non-molecularly imprinted polymerNMR Nuclear magnetic resonanceODS OctadecylsiloxanePMP Pinacolyl methylphosphonateSDS Sodium dodecyl sulfateSPE Solid phase extractionTSA Transition state analogue

1Introduction

Molecular templates are involved universally in relaying structure and func-tion into both biological and synthetic molecular assemblies. Information onshape, size and electronic spatial configuration of a template, i.e. the molecu-lar state of a molecule, is used to guide and control the formation, structureand function of new self-assembled molecular entities. The template eitherbecomes part of the new structure or is separated from it and reused, or in-volved in other templated processes. Messenger RNA in the polymerase chainreaction functions as a reusable template for new complementary polynu-cleotide strands [1]. Diblock copolymers can be used as templates for min-eralisation of non-trivial inorganic phosphate structures [2, 3]. Patterned sur-faces of phosphate groups template the mineralisation and resulting complexmorphology of inorganic salts such as calcium carbonate [4]. In liquid crys-tals the addition of dopants can cause a switch to a different liquid crystallinephase [5]. The formation of a molecular capsule can be triggered through thepresence of the correct solvent which occupies the interior [6]. Micelles havebeen used for many years as templating agents in the formation of zeolites [7].In this review we will be discussing synthetic polymer receptors formed ascrosslinked networks in the presence of molecular templates. The focus willbe on anionic templates and the performance of the resulting solid-phase re-ceptor structures in terms of chemical selectivity and molecular recognition.

1.1Context

Anion templation as a means of producing molecularly imprinted polymershas to deal with the issues that one encounters designing and synthesisingmolecular receptors for anions in general [8–10]. The increasing understand-ing of the factors involved in anion recognition, such as how to design recep-tor sites and ligand arrays that are appropriate for the larger and more diffuse

Molecularly Imprinted Polymers Using Anions as Templates 209

charge distribution in anions, is one reason for the recent surge in interest inpreparing MIPs for anion recognition. Another reason is the impact of pro-teomics and the challenges and opportunities that accompany it with regardto sensing and detection of charged species (simple anions as well as carboxy-late, phosphate and sulfate derivatives) in the interrogation and analysis ofcomplex biological systems. A third reason relates to advances in the ability ofchemists to apply supramolecular and self-assembly approaches to the designof complex structures and materials, from which MIP synthesis has alreadybegun to benefit.

2Molecularly Imprinted Polymers

2.1The Concept

The synthesis of a molecularly imprinted polymer is at first sight a straight-forward affair. A template molecule, which can essentially be freely chosen, ismixed together with one or several polymerisable receptor molecules (“func-tional monomers”) [11–14] (Fig. 1). The latter, so-called binding sites, areselected on the basis of preferably strong and directed interactions withthe template molecule to maximise the molecular fidelity of the imprint-ing process (covalent and non-covalent interactions are possible). This step

Fig. 1 Pictorial representation of molecular imprinting, showing a formation of pre-polymerisation complex; b polymerisation; c removal of template molecule to generatemolecularly imprinted recognition site


takes place once template and binding sites have self-assembled, typicallyin solution, and crosslinking molecules have been added which producea crosslinked polymer network once polymerisation has been initiated, inmost instances thermally or photochemically although electrochemical or re-dox initiation has also been demonstrated successfully. The template is thenremoved from the polymer, most commonly through liquid extraction, leav-ing behind a polymer matrix made up of supramolecular receptor sites whichhave been formed throughout the polymer network as a result of the pre-organisation of the binding site/template complex [15]. These receptor sitesare often referred to as molecularly imprinted cavities as a polymer matrixhas formed around the template, becoming part of the receptor structure.The complementarity of shape and electron density distribution transferredinto the polymer matrix has obvious analogies with the substrate recognitionsites of enzymes or antibodies where the spatial organisation of the func-tional groups of amino acid side chains define catalytic sites and substrateselectivity [12]. As the ability of conformational adaptation to the substrateis a key feature of enzyme recognition and catalysis, so is a certain levelof conformational freedom in a MIP [16]. Regarding the molecular recogni-tion mechanism operational in MIPs, a number of recent studies illustratethe decade-old controversy between the imprint mechanism (cavity gener-ated by template) and the association mechanism (trapped templates act asbinding/nucleation sites) [17, 18].

2.2Inspiration, Incorporation and Assimilation

The concept of molecularly imprinted polymers is strikingly simple andgeneric. Very often MIPs are referred to as synthetic enzyme or antibodymimics due to the analogy of having spatially defined functional groupspositioned inside a molecular pocket. Once this simplistic view took hold,most MIP research revolved around identifying the level to which MIPs canmimic the structure and function of protein receptors and catalyst activesites [11–14]. However, just as inspirational as the likening of MIPs to an-tibodies and enzymes has been, as incomplete is the appreciation of theirpotential under conditions where proteins are mostly unsuitable or inade-quate; examples are their use in organic solvents and in variable and ex-treme temperature environments. In recent years, MIPs have been discussedmore generally as a special type and format of receptor design and struc-ture, within the wider chemical world of supramolecular chemistry [19–21].Consequently more of the advances in self-assembly and molecular recog-nition have been explored and incorporated into MIP design and synthe-sis. This trend will be even more beneficial to the development of MIPsin years to come, during which many concepts and approaches will havebeen assimilated, causing a step change in MIP performance. Some instruc-


tive examples of this progress include stoichiometric non-covalent bindingsites (vide infra Sects. 2.3 and 3), high-throughput and parallel synthe-sis experimentation [22–24], integration of sensing modules into MIPs [25,26], new polymerisation methods [27], modelling approaches of templat-ing complexes [28, 29] and single site MIPs as dendrimer [30] or micro-gel [31]. Future activities will include: improvements of recognition in wa-ter [32]; the use of (dynamic) combinatorial chemistry [10, 33] and thermo-dynamically controlled approaches for MIP synthesis; the use of MIPs asthe sensing component for medical diagnostics [34] and environmental an-alysis [35], also as polymeric drugs, but, more importantly, as drug deliv-ery vehicles (including molecular tags and labels) [36]; the application ofMIPs as identification and separation tools for biological screening [37];the preparation and processing of MIPs with smaller feature sizes andmore complex shapes [38, 39]; the successful use of more challenging tem-plate molecules (proteins, complex drug molecules). On the whole, MIPsare relatively easily prepared multidentate receptors, in contrast with smallmolecule receptor analogues, and have shown nanomolar binding affini-ties [40]. Their solid-state nature is in itself a useful format for screeningstrategies, with further benefits arising from long-term stability, recyclabilityand modularity.

2.3A Brief “Developmental” History

Polyakov et al. were the first to report the effect of generating selectivity ina polymer matrix through addition of a template molecule (for a more de-tailed account see [14]). In 1931 they prepared silica particles from sodiumsilicate and (NH4)2CO3 in water, adding benzene, toluene or xylene in theprocess. After prolonged drying and exhaustive extraction with hot water, thepolymer showed higher capacity for the additive (template) than for struc-turally related compounds. A contemporary debate ensued about the originof antibody selectivity in the immune system with contributions from Breinland Haurowitz [41], Mudd [42] and Linus Pauling [43]. The latter favouredthe view that antibodies are generated in the presence of an intruding anti-gen which would determine the antibody conformation (as he attempted todemonstrate in 1942 [14]). Similar work to that of Polyakov was reported in1949 by Dickey, this time using alkyl orange dyes as the templates. The im-printed silica gel showed pronounced selectivity for the template, which waspresent throughout the silica network-forming step [44]. Curti et al. showedthe first examples of enantioselectivity with both mandelic acid and camphorsulfonic acid (also in silica) and their application as stationary phases forchromatography (Fig. 2) [45, 46].

A particularly imaginative example was contributed by Patrikeev et al. in1960, in which they resorted to bacteria as templates. Curiously the imprinted


Fig. 2 A selection of important anionic template molecules and their corresponding bind-ing sites reflecting the development of MIPs


silica promoted certain bacterial growth over that of control samples im-printed with other varieties [47]. In a related study either the levo or dextroform of the same bacillus (Bacillus cereus var. mycoides) was incorporatedinto a silica gel matrix showing discrimination for L- and D-linanool vapourrespectively [48]. The same group provided perhaps the earliest example ofa catalytic MIP templating amino acid condensation reactions, though rateswere enhanced modestly by a factor of two and experimental details arescarce [49]. Two decades later, in 1972, Wulff et al. [50] and Klotz et al. wereindependently pioneering the use of organic polymers as tailor-made recep-tors. Wulff et al. were the first to harness the versatility of radical vinyl chem-istry as polymerisation methodology and exemplified the covalent imprintingapproach (Fig. 2), whereas Klotz and coworkers harnessed the benefits ofring-opening polymerisation combined with reversible disulfide crosslinkingemploying non-covalent interactions [51]. In the following years Wulff et al.elaborated the covalent approach with the addition of non-covalent (predom-inantly electrostatic) interactions [52, 53]. A method to imprint on a surface(“surface imprinting”) was first developed by Sagiv et al. in 1979 using sil-ica particles as surfaces with polymerisable siloxane as surface modifiers [54].This methodology is transferable to other surfaces as long as the templateabsorbs onto the chosen surface [55]. The work in the group of Mosbachin 1981 revolutionised molecular imprinting through the non-covalent ap-proach [56] (Fig. 2). The impact derived from the much simpler syntheticroute for making MIPs. Rather than having to prepare templates connectedto polymerisable binding sites via reversible covalent bonds, non-covalent in-teractions allow the self-assembly of template and binding sites in solution,affording a pre-polymerisation complex prior to vinyl monomer network for-mation. In further development, Mosbach et al. demonstrated that MIPs withhigh selectivity could be obtained even without the necessity to invoke cova-lent or ionic bonds [57]. With time, more and more groups have taken up thenon-covalent approach and today it is the most widely used methodology toprepare MIPs [14].

However, comparative studies of covalent and non-covalent imprinting arerare, and even those reported lack the confidence in which optimised condi-tions have been employed for both approaches, so that clear conclusions fromwhich to select the better approach could not be drawn [58, 59].

Although the notion of combining the advantages of covalent bonds duringthe imprinting step with those of non-covalent interactions for rebinding waspresent in earlier work by Wulff et al. [50, 52, 60, 61], the first example of usingexclusively covalent interactions for the imprinting step and noncovalent onesfor rebinding was executed by Sellergren and Andersson in 1990 [62]. Poly-merisable groups were linked by ester bonds, though the target molecule,p-aminophenylalanine ethyl ester, was necessarily different to the template.This changed completely in 1994, when Whitcombe et al. significantly refinedthe semi-covalent approach, outlining a concept based on a small covalent


fragment as space holder (“sacrificial spacer approach”) [63]. With this strat-egy, template and a binding site are covalently linked with a cleavable spacerwhich is designed to reveal complementary functional groups upon removal,closely following the geometry of the covalent juncture for improved com-plementarity [64]. The semi-covalent approach is an attempt to synergise theadvantages of the covalent methodology (strict control of functional grouplocation, more uniform distribution) with that of the non-covalent one (re-duced kinetic restriction upon rebinding).

As a “natural” coalescence of the covalent and the original non-covalentapproach, a stoichiometric non-covalent approach was developed [65]. If thenon-covalent interactions are strong enough to produce association constantsof at least 103 M–1 (or preferably higher) the equilibrium will lie well on theside of the template-functional monomer complex. On the way to stoichio-metric non-covalent interactions (discussed in detail in Sect. 3) a representa-tive example of carboxylic acid templates and various N-base binding sites isfound in Kempe et al. (1993) [66]. Polymerisable amidines were first used forstoichiometric imprinting of a transition-state analogue (TSA) by Wulff et al.in 1997 (Fig. 2) [67]. Other noteworthy developments were valine-derivedbinding sites for dipeptide imprinting by Yano et al. in the same year [68]and bidentate dioxoborolanyl recognition sites for carboxylates exploited byLübke et al. in 2000 [69]. In 1994 Sellergren was already successful in comple-menting the standard MAA binding site monomer with pentamidine, offeringassociation constants close to those typically considered to be required forstoichiometric imprinting strategies (Fig. 2) [70].

With time and improved synthetic protocols, larger templates (fullerenes,dendrimers, nanoparticles, colloids, micelles, lipid bilayers, self-assembledblock copolymers, oligonucleotides, DNA and proteins) have been im-printed [14] and the choice of matrices has expanded to liquid crystalpolysiloxanes, carbon networks, zeolites, layered aluminophosphates andcolloidal crystals, though organic polymer networks remain the dominantimprint casting medium [14].

Further important developments related to templates emerged in the mid-and late 1990’s. In 1999, Sreenivasan et al. demonstrated that it is possibleto imprint with two different template molecules simultaneously [71]. A yearlater, Rachkov and Minoura introduced the epitope imprinting methodologyin which a structurally unique 3-amino acid fragment of a peptide chain wasshown to be a sufficient template to produce MIPs that are selective for theentire nonapeptide [72]. Another year later, Sellergren et al. used a templateanalogue rather than the target molecule itself for generating MIPs [73], fur-ther suggesting that complex molecules can be imprinted successfully on thebasis of a molecular information-rich substructure or derivative. Also, ma-nipulation of the binding site chemistry has widened the opportunities forMIP applications. Useful synthetic post-polymerisation transformations wereelaborated on disulfide templates by Mukawa et al. in 2002, as their reduction


was found to yield thiols as binding sites (semi-covalent approach), whichcan also be oxidised up to sulfonic acids, changing the chemical nature ofthe non-covalent binding site in a controlled manner [74]. The first MIPsdesigned for influencing molecular reactivity came from the laboratories ofShea et al. in 1978 [75], closely followed by Damen and Neckers in 1980 [76].Cycloadditions leading to cyclopropane and cyclobutane dicarboxylic acidsrespectively were performed, with significant regio- and diastereoselectivityin the latter case. Also in 1980, Belokon et al. organised amino acid tem-plates via Schiff base formation inside a MIP cavity, and showed that upondeprotonation the amino acid carbanion retained its original configurationunlike the same reaction in solution [77]. A few years later, in 1987, Sarhanet al. showed the possibility of stereospecific reversal of configuration ratherthan retention with mandelic acid [78]. Andersson et al. achieved the cata-lytic deprotonation of bound amino acids (vide supra Belokon et al.) electinga pyridoxal-coenzyme analogue as a template in a non-covalent imprintingsystem [79]. In this particularly eventful year Wulff et al. were the first tocarry out an asymmetric reaction in a chiral cavity designed to generate op-tically active amino acids, which was achieved with an ee of 36% and usedglycine as a template [80]. The observed chiral induction is solely the re-sult of enantioselective induction by the chiral cavity. The above examplesgive a taste of the potential of MIP cavities as nanoreactor sites. MIPs thatare catalytically active (apart from some early work in silica matrices [81])and which thereby influence reaction outcomes, were only embarked upon in1987, by Leonhardt et al. [82]. Initially the product of a reaction was pursued,and not the better mimicry of using a transition state analogue. It producedesterolytic activity with a 2–3 fold rate enhancement and substrate selectivityof MIPs. Two years later however Robinson et al. prepared a MIP imprintedwith a TSA. Poly(vinylimidazole) and p-nitrophenylphosphonate TSAs werecoordinated to a CoII ion and the resulting template assembly was crosslinkedwith dibromobutane. Despite the more rational design of the template onlya small enhancement of 1.6-fold was observed [83]. As the development ofcatalytic MIPs is strongly intertwined with the evolution of stoichiometricnon-covalent binding sites as oxyanionic receptor sites, we will return to thisconnection in one of the later sections (Sect. 3).

2.4Scope and Limitations

Discussions about the current and general scope and limitations associatedwith MIPs are ongoing. For a while, a good number of publications were oftwo minds in their assessment of the potential of MIPs, where small advancesin performance were hailed as stepping stones to rival, if not excel, anti-body and enzyme methods. This was followed by a healthy sobering period.Over the last ten years a much more realistic assessment of the potential of


MIPs has emerged, accompanied by a substantial increase in the number ofresearchers involved in MIP science and technology, creating a rise in publi-cation numbers. Strengths of MIPs include:

• Robustness/longevity

– MIPs have been shown to keep their molecular recognition perform-ance over months and even years; this is a stark contrast to enzymesand antibodies. Depending on the chosen polymer matrix, MIPs arechemically and physically robust and not limited to an aqueous envi-ronment or well-controlled physiological conditions (pH, temperature,solvent). MIPs can be recycled and reused [84].

• Solid-state format

– Most MIPs are solid-phase which makes them amenable to automatedprocesses requiring only extraction and reloading cycles, with sub-strate isolation being a simple filtration step [85].

• Relative ease of synthesis

– The concept of molecular imprinting is simple and elegant, which isalso reflected in the relatively straightforward manner in which theycan be synthesised [14]. This is certainly true if one compares MIP syn-thesis with other strategies available to produce synthetic receptors.Many of the latter require a large number of synthetic steps, rather thanself-assembly, to achieve the spatial positioning of functional groupsidentified as being key for mimicking enzyme catalysis or antibodyperformance.

• Relative ease of design

– MIPs rely entirely on self-assembly (cf. Pauling’s theory of antibodyformation) and present a modular system, which also allows MIPsto be designed more easily in contrast to approaches in which thegeometric requirements of the receptor molecule have to be carefullycalculated or estimated during the design phase [86] (though the orig-inally chosen template may not be the best “template” for the desiredperformance) [87, 88].

• Unrestricted choice of substrate

– As long as it is possible, either in solution or in bulk, to self-assembletemplate with binding sites, and as long as the polymerisation chem-istry is compatible with such a complex, the choice of templatemolecule would appear to be unrestricted [14].

• Can be made to give a response (e.g. QCM, fluorescence).

– Through the addition of multi-functional binding sites, additionalfunction for sensing, or some other form of responsive behaviour,can be incorporated without having to redesign template or receptorsites [25, 26, 34, 35].


Limitations of MIPs:

• There are some common limitations associated with MIPs regardless ofthe chosen template, binding site, solvent, etc.:

– Binding sites are heterogeneous (polyclonal rather than monoclonalas in enzymes and monoclonal antibodies) as a consequence of thestatistical nature of the polymer network forming process, leading toundesirable band broadening in chromatographic separations and toa distribution of different activities of catalytic sites [89].

– Similarly not all template molecules may be accessible within the poly-mer matrix, or will require very long extraction times leading to prob-lems of template leaching especially when employing MIPs for traceanalysis [35].

– Diffusion to and from the recognition sites within the MIP matrix istypically slow, depending on the solid-phase format and phase struc-ture. High pore volume and interconnectivity can reduce the diffusionproblem but generating such a hierarchical pore structure may notbe compatible with the requirement of forming a strongly associatedtemplate complex. Thin film, surface imprinting and membrane ap-proaches are alternative means of addressing diffusion problems butare in many cases accompanied by lower selectivity due to interfaceeffects or changes in the crosslinking stoichiometry, leading to a reduc-tion of the molecular fidelity of the imprint [14].

– For applications such as enantiopolishing or chiral separations, MIPsare said to offer low capacity. This is true when comparing MIPs withsorbents that rely on an interaction with a surface or a surface modi-fied with a chiral selector (e.g. Pirkle phases). On the other hand, MIPsare at least competitive if one compares their atom economy with thatof enzymes or antibodies adding up the molar mass of crosslinker andreceptor sites per template (between ∼5–100 kDa depending on syn-thesis recipe) [14].

– Dealing with polar and particularly water soluble template moleculesand recognition in aqueous milieu have been major issues for MIPsfor many years. Progress across the field however indicates that genericstrategies are available to overcome this limitation [34].

• Major issues regarding deficits in technology for anion MIPs will be dis-cussed in more detail in Sects. 3.1–3.4.

2.5Formats

Molecularly imprinted polymers are implicated for a wide range of applica-tions. Their solid-phase nature necessitates that the synthesis step includesthe processing step. As is the case with any type of covalently crosslinked ma-


terial, once polymerisation has taken place the molecular structure is fixedand cannot be reconfigured. The most common formats encountered forMIPs are:

• Irregular particles

– The most popular format prepared by grinding the bulk-polymerisedMIP into smaller particles. For chromatographic applications this isusually followed by sizing through sieving [90].

• Monoliths

– Upon crosslinking, the polymerisation mixture retains the shape of thereaction vessel. Separating the inner wall of the reaction vessel fromthe polymer reveals a single piece of MIP, a monolith. The monolithscould be used for MIP applications that have been shown to be use-ful as a direct means of packing chromatographic columns (HPLC,CEC) [90, 91]. If the monolith is exposed to large variations in sol-vent polarity without space constraints it will slowly disintegrate intosmaller particles due to the high density of crosslinks in connectionwith a glass transition temperature (Tg) that is usually well above roomtemperature.

• Beads

– This third format is particularly useful to pack chromatography columnsand generally to manipulate MIPs in automated processes [90]. As it re-quires a two- or multi-phase solvent system during polymerisation toimpart the spherical shape, until recently it has proven quite challeng-ing to obtain narrow disperse bead sizes with high yield at the desiredsize. Protocols are now available that offer generic solutions for beadinga wide range of MIP formulations [92]. Beads are particularly desir-able for catalytic MIP applications as selectivity improves (compared tomonolith synthesis) and catalytic activity increases due to faster masstransfer [93].

• Films

– Especially for sensing (UV-Vis, luminescence, QCM), film formats, i.e.surfaces coated with a MIP layer, are attractive as a simple means ofdevice fabrication. This does not differ greatly from the monolith for-mat, and film thicknesses can be controlled through an appropriatelyshaped substrate or through spin coating. The changes in solvent con-centration caused by evaporation of solvent and precipitation of theMIP during the film forming processes will alter the stoichiometry ofthe template/binding site complex set in the starting solution [94].

• Membranes

– Free standing films of MIPs have been prepared in various ways.Mechanical robustness has been imparted through lower crosslink


densities or monomers generating lower Tg polymer backbones com-pared to monoliths. Reducing crosslink density or rigidity generallyreduces selectivity and long-term stability of MIPs. Membrane com-posites offer a means of accommodating brittle MIP particles into anoverall flexible matrix. Phase inversion techniques with polar high-performance polymers have also been successfully applied to formMIP membranes [95].

• (Hydro)gels

– Instead of high crosslink levels and thus polymer networks with highTg, MIP gels and hydrogels can be produced in the same way as mono-liths, by selecting monomers leading to lower Tg polymer backbonesand/or by choosing polar monomers with high affinity to water andprotic solvents, so that the solvent causes the polymer network to swelland to become more flexible, i.e. gel-like. Interesting applications areantibody mimics with MIPs becoming single site catalysts [31] and,more recently, controlled drug delivery [96].

• Single molecule species

– Zimmerman et al. were the first to have synthesised a dendrimerimprinted with a porphyrin derivative using the sacrificial spacermethodology: a single cavity within a single polymer confinementstructure [30].

• Spun fibres

– Electrospinning has been applied to MIP synthesis employing mem-brane formation techniques with the help of a support polymer,promising higher surface area, faster diffusion and more efficientwashing out of the template molecule [97].

2.6Design Criteria

MIPs are complex synthetic receptors prepared by self-assembly, followedby network formation. They are by their very nature amorphous solids, andthose are inherently difficult to analyse at the molecular level. Most ana-lyses of MIPs are indirect, assessing the fidelity of the molecular recognitionevents through binding assays of various kinds. Nuclear magnetic resonance(NMR) studies on the self-assembled complex formed prior to polymerisa-tion as well as molecular modelling have been shown to be useful in arrivingat a more rational design of MIPs [98], though pre-polymerisation analysisand modelling [99–102] are still in stages too early to extrapolate trend pat-terns and reach more general conclusions on reaction conditions, bindingstrength and template/binding site complex formation. The vastness of the


parameter space involved when synthesising MIPs has prompted the use ofparallel synthesis combined with statistical data analysis to more rapidly ana-lyse data sets and identify trends. These activities will mature and, with time,will have wider impact on our ability to design MIPs for any given templatemolecule with tunable thermodynamic and kinetic performance parameters.Although there are always exceptions, the following design criteria have beenselected as being useful and reliable guides that help one arrive at a reason-able starting point for developing a new MIP:

• Polymerisation temperature– Lower temperatures improve selectivity and capacity. The exothermic

nature of radical vinyl polymerisation compounded by the gelling ofthe imprint mixture is likely to cause higher internal temperatures thanthose used to control the environment externally.

• Binding site design– Statistical copolymerisation kinetics with the crosslinker lessens the

heterogeneity of receptor sites.– Too much conformational freedom between binding site and polymer

backbone reduces selectivity.– Strong, ideally close to stoichiometric ratios of binding interactions

improve the quality of the imprint.– Faster equilibration kinetics increase performance in chromatography-

based application.• Crosslinker

– There is an optimum level of crosslinking with typical values to bearound 70–95 mol %. Conformational flexibility of the crosslinker hasto be balanced with the overall level of crosslinks. The chemical na-ture of the crosslinker determines the useful solvent range for a MIP.The size of the crosslinker ideally matches the dimensions of the tem-plate, to avoid conformational frustrations within the polymer matrixas a result of a structural misfit, which can lead to a loss in capacityand/or selectivity.

• Solvent– The best choice is a solvent that maximises template binding site in-

teractions while producing a highly porous MIP matrix to minimisediffusion limitations.

• Template– Imprinted cavities can be viewed as multi-point receptor sites plus

the added shape of the cavity as an additional recognition feature.A template offering many points for binding and where stoichiomet-ric binding sites for the exhibited functional groups exist offers thebest conditions for high selectivity. On the other hand, large templates


with many binding positions become more difficult to imprint with asthe differences between related molecules become relatively smaller.This is also the case for templates with only one functional group,where the shape of the cavity becomes a more important contribu-tor. Larger templates pose the problem of being difficult to extractfrom the polymer as they more readily become entrapped. Degrad-able templates or template fragments (epitopes) have been shown to behelpful in overcoming this limitation. Conformationally dynamic anddemanding templates such as proteins are a particular case in point.Presenting proteins in native form on a degradable nanocrystal sur-face, for example, is an ingenious way to address both shape stabilityof the template and access to the imprinted cavity.

• Template to binding site stoichiometry

– A good starting point is a ratio that reflects the number of sites avail-able on the template for interaction with the chosen polymerisablereceptor. For covalently bound templates or stoichiometrically coordi-nating binding sites the ratio is obvious.

• Crosslinker to self-assembled complex stoichiometry

– The minimum number of “atoms” that are required to form the desiredbinding pocket and typical stoichiometries that give high selectivityare 20 : 1 in equimolar terms. Larger ratios can lead to enhanced selec-tivity but this is traded against a loss in capacity.

• Covalent versus non-covalent imprinting

– The covalent approach is synthetically more labour-intensive but pro-duces MIPs that have better defined affinity distributions. It is stillundecided as to which situation one of these two strategies will lead tothe higher performing MIP.

The polymerisation chemistry and other network forming processes appliedto MIP generation (e.g. membrane phase inversion) lack a thermodynamicmeans of controlling the formation of receptor sites to render them less poly-clonal. Some means of increased control such as living [103] and thermody-namically controllable polymerisation chemistry [27, 104, 105] (in analogy todynamic combinatorial library synthesis [106]), are at a very early stage. Im-proved binding site design, especially in water or buffered system, would ben-efit from hydrophobic effects for recognition cf. enzyme pocket; for this, how-ever, one needs to incorporate the necessary conformational/environmentalchange associated with it. Also, the area of modelling for selecting optimisedstoichiometries for MIP manufacture would become a more powerful meansof guiding the synthesis, with more precisely defined interactions that gener-ate an imprint with increased 3D/shape fidelity.


3Molecular Imprinting Using Anionic Templates

One of the most attractive features of molecular imprinting is its applicabil-ity to such a diversity of analytes. Thus the literature boasts MIPs made usinga striking array of templates, which range from small molecules (<100 Da) tobiological macromolecules (several kDa), and include sugars, steroids, dyes,herbicides, pharmaceuticals and biochemicals amongst others (Table 1). Nev-ertheless, upon inspection it is apparent that anionic templates have seldombeen considered throughout the history of imprinting. Indeed, up until thelast 7 or 8 years, there has been a surprising paucity of MIPs that are designedeither to recognise “naked” anions (such as oxyanions, halides or cyanide)or to exploit the predominating anionic character of a particular analyte. In-directly, however, anionic templates have played a role as early as the 1940sup to the present time. For example, Dickey prepared silica gels in the pres-ence of methyl orange (p-dimethylamino-azobenzenesulfonic acid) and thenlater on using a sulfonamide analogue as template [44, 107]. In a subsequentcommunication it was disclosed that the two MIPs exhibited very similar se-lectivity for their target molecule, indicating that the negative charge on thep-substituent was nonessential to the recognition mechanism [107]. Curtiet al. imprinted in silica using enantiomers of mandelic acid and camphor-sulfonic acid. While the conditions for their MIP preparation (pH 4) suggestthat the sulfonic acid, and to some extent the carboxylic acid, were disso-ciated throughout the imprinting process, the aforesaid functional groupswere not deliberately targeted in the MIP synthesis [45]. In a more recentexample, Rajkumar et al. generated a MIP for fructosyl valine recognition,using boronic acid-functionalised monomers to covalently interact with thediol moieties [108]. The carboxyl group, which would have been fully dissoci-ated during the rebinding event (pH 11.4), was not targeted by any functionalmonomer.

The relative scarcity of anion-imprinted polymers is especially remark-able given the prevalence of anionic molecules of importance. It is widelyrecognised that the design of anion hosts is a challenging task, especially incontrast to cation receptor design. The main reason for this is the diffusenature of anions. Since Anions are significantly larger than the equivalentisoelectronic cations, they have a lower charge to radius ratio; consequently,electrostatic binding interactions are markedly diminished. For the same rea-son, anions are generally more polarisable than cations, and it is thereforemore difficult to develop specific binding sites for their recognition.

With regard to molecular imprinting, a significant hindrance to the devel-opment of anion-recognition MIPs has been the fact that anionic species arevery often incompatible with apolar media. The molecular imprinting processinvolves the formation of a pre-polymerisation complex between the templatemolecule and functional monomers. This is typically based upon H-bonding


Table 1 Examples of molecules that have been used as templates for molecular imprinting

Template molecule Selected Refs.

Cyclobarbital [162, 163]

Naproxen [164–166]

(S)-Propranolol [167]

L-Menthol [27]

Glucose [168, 169]

Cholesterol [64, 170, 171]

Nicotine [172, 173]

Bisphenol A [174]


and electrostatic interactions (though it might also include dipole–dipoleattractions or metal ion coordination). In general, such interactions are max-imised when a non-competitive, apolar solvent is employed during the im-printing step, thereby affording a well-defined binding site. Similarly, the useof such a solvent for subsequent rebinding of the target molecule tends to pro-mote specific binding as well as minimise levels of undesirable nonspecificinteractions. The last 20 years have witnessed notable advances pertaining tothe compatibility of molecular imprinting with polar/aqueous media. Whilemuch of that is outside of the scope of this review, one particular issue –the influence of functional monomers used – is of prime importance to ourdiscussion.

Traditionally, the majority of MIPs have been made using a limited reper-toire of simple, commercially available functional monomers (for example,methacrylic acid, methacrylamide, 1-vinylimidazole), in accordance with thenon-covalent imprinting strategy. In this regard, Simon and Spivak recentlyinvestigated a selection of commercially available functional monomersdeemed to be particularly suitable for oxyanionic templates (in particular,for carboxylates and phosphonates) [109]. They prepared a series of MIPs,incorporating various functional monomers including pyridine-derivatives,aliphatic amines, and 1-vinylimidazole. A chiral template molecule, t-Boc-L-phenylalanine, was employed to study enantioselective discrimination. TheMIP particles were slurry-packed into HPLC columns, and capacity factorsand separation factors of the L- and D-enantiomers were calculated to providevalues for binding affinity and enantioselectivity. Interestingly, the authorsobserved that, while the highest binding affinity was exhibited by the morebasic aliphatic amine derivatives, the best selectivity was obtained by func-tional monomers bearing aromatic amine groups. Directional binding, asafforded by these latter functional monomers, is evidently vital for the pro-duction of selective binding sites.

However, while the use of simple, commercially available functionalmonomers is relatively undemanding from a synthetic chemistry perspective,the drawback is that such monomers generally tend to afford only relativelyweak binding interactions. This is all the more evident in aqueous/polar me-dia. As a consequence, it is usually necessary to imprint using a large excessof functional monomer relative to the amount of template molecule, in orderto push the equilibrium and ensure a high degree of complexation. Unfor-tunately, this approach results in undesirable binding site heterogeneity andsubsequent non-specific rebinding of the analyte.

The introduction of a novel imprinting methodology, which has attractedconsiderable attention within the field, serves to explain much of the re-cent progress with anion-imprinted polymers. Termed “stoichiometric non-covalent imprinting”, this approach uses carefully selected (often custom-made) functional monomers which bind with relatively high affinity to par-ticular motifs on the template molecule [110]. Since the vast majority of


functional monomer is associated with the template molecule, it is possibleto use an equimolar mixture of template and functional monomer during im-printing. In theory, this not only reduces non-specific interactions but alsoaffords binding sites that are more defined, since the functional monomersare more precisely arranged by the template molecule in a definite spatialorientation within the imprint cavity. Furthermore, the strong binding inter-actions afforded by these functional monomers means they are often strongenough to work in solvent systems more suited to anionic molecules, as wellas apolar systems. Wulff et al. were the first to propose the concept of stoi-chiometric non-covalent imprinting [111] (this can be thought of as a logicalextension to the covalent imprinting strategy, also pioneered by Wulff). Theyreported the use of custom-made polymerisable amidinium species for tar-geting oxyanions, in line with the prevalence of guanidinium-based arginineresidues in oxyanion binding sites in biological systems. More recently, otherauthors have promoted functional monomers based upon urea [112, 113],thiourea [114] and guanidinium groups [115, 116].

Regarding the scope of this review, in the following subsections, the use ofmolecular imprinting for the recognition of selected classes of anions will bediscussed. In defining “anions”, we refer not only to small “typical” anions,such as nitrate, phosphate, sulfate, etc., but also to larger, increasingly morecomplex substrates, which bear an anionic moiety. Moreover, we shall limitour discussion to MIPs in which the anionic part of the template molecule isdeliberately targeted or used to advantage. In this way, we disregard any re-ports of molecular imprinting where the template molecule might happen tofeature an anionic component (for example, the C-terminus of peptides) butwhere this is not deliberately accounted for in the design of the functionalbinding site.

3.1Anionic Phosphate Derivatives

Phosphate derivatives are ubiquitous in biological systems, rendering thema very attractive target for anion recognition chemistry. For example, manycoenzymes and metabolic intermediates are esters of phosphoric and pyro-phosphoric acid; the nucleic acids DNA and RNA are based upon a phosphatediester backbone; and phospholipids, the major component of all biologi-cal membranes, contain a phosphate diester moiety. Protein phosphorylationis a reversible post-translational modification involving the kinase-catalysedtransfer of a phosphate group from ATP to a protein (typically to serine, thre-onine or tyrosine residues). This highly important process can be thought ofas a means of “switching on” proteins, and underlies the regulation of virtu-ally all cellular events. In addition, many man-made molecules of interest arephosphate derivatives. This includes environmental pollutants and degrada-tion products of chemical warfare agents.


One of the earliest examples of a polymer imprinted with a phosphatederivative came from Turkewitsch et al., who constructed a MIP for theselective recognition of cAMP in aqueous media (Fig. 3) [117]. The custom-designed functional monomer 1 was used to target the nucleotide via botharyl stacking and electrostatic interactions between the positively-chargednitrogen and the phosphate diester anion (Fig. 4). Rather cleverly, this func-

Fig. 3 Schematic representation of cAMP inside MIP, showing possible binding interac-tions

Fig. 4 Functional monomer custom-designed for cAMP-imprinted polymer


tional monomer also served as a transducer element: binding of cAMP ef-fected fluorescence quenching within the imprinted polymer, enabling it to beused as a chemosensor. Further recognition points within the imprinted cav-ity were afforded by the use of a large excess of HEMA, which was assumedto bind to the purine moiety via a network of H-bonds. The fluorescent na-ture of the MIP was used to calculate an association constant, Ka, for bindingof cAMP by the MIP in aqueous media: this was determined to be in the orderof 105 M–1. With regard to selectivity, the MIP was shown to discriminate be-tween the sodium salts of cAMP and the potentially competitive analogue,cGMP, thus highlighting the role of the adenosine moiety in the recognitionprocess. From the standpoint of our discussion, an investigation into selectiv-ity over other oxyanions would also have been of interest.

A steadily increasing number of publications report the use of phosphate-derived template molecules to prepare MIPs for solid-phase extraction (SPE).SPE involves the selective pre-concentration of trace-level compounds fromcomplex mixtures. The high selectivities and affinities obtainable by molecu-lar imprinting render these polymers ideal for use as SPE sorbents.

For example, Möller et al. prepared a MIP for the selective isolation ofdiphenyl phosphate (a flame retardant hydrolysis product) from urine [118].The MIP was prepared using an excess of 2-vinylpyridine as the functionalmonomer, with ditolyl phosphate as the template molecule (imprinting witha structural analogue of the target molecule eliminates complications causedby possible leaching of the template during subsequent rebinding analysis).In this study, the influence of matrix components normally present in hu-man urine (urea, NaCl) was investigated. While urea did not compete withdiphenyl phosphate for the basic binding sites, it was observed that NaCl af-fected both the recovery and repeatability of diphenyl phosphate binding.This was explained by the assumed complexation between Na+ ions and theanionic analyte, which would render the molecule more neutral and therebyhinder ionic interaction with the polymer.

The last decade has seen considerable interest in the application of mo-lecular imprinting to nerve agent detection. The general objective is to de-velop a sensitive and specific portable sensor device that boasts faster re-sponse times than existing nerve agent analysis techniques (gas chromatog-raphy, HPLC, immunoassay). In this regard, Jenkins et al. used pinacolylmethylphosphonate (PMP), the hydrolysis product of Soman, as a templatein the preparation of an imprinted polymeric luminescent sensor [119, 120].PMP reversibly bound to ligand-coordinated Eu3+, which was incorporatedinto the recognition site, and this binding event corresponded to the ap-pearance of a narrow luminescence band in the 610 nm region of the Eu3+

spectrum. The presence of organophosphorous pesticides and herbicides wasdetermined not to interfere with PMP recognition (although none of the com-pounds screened bore an anionic moiety).


Fig. 5 Schematic representation of MPA-imprinted polymer

Zhou et al. developed a potentiometric chemosensor for the detectionof methylphosphonic acid (MPA), another degradation product of nerveagents [121]. In the presence of MPA, an octadecylsiloxane (ODS) thinlayer was covalently bound to the surface of an indium tin oxide (ITO)transducer (Fig. 5). Subsequent removal of the physisorbed MPA affordedwhat the authors claim to be imprinted recognition sites, with functionaland steric complementarity. It was demonstrated that rebinding of MPAby the “imprinted” film generated a greater potentiometric response thanbinding of homologous substrates (ethylphosphonic acid, propylphosphonicacid, tert-butylphosphonic acid). However, binding of MPA by the con-trol, non-imprinted film, afforded a greater potentiometric response thanbinding of MPA homologues by the “imprinted” film. It is thus prob-able that recognition in this case is governed by a size-exclusion phe-nomenon. Further experiments should be carried out to eliminate thisambiguity.

Prathish et al. also generated a potentiometric sensor for MPA detection,although using an altogether different methodology [122]. MIP particles, pre-pared via the copolymerisation of EDMA and either methacrylic acid or4-vinylpyridine in the presence of MPA (which was afterwards removed using1:1 (v/v) methanol:acetic acid) were dispersed in 2-nitrophenyloctyl ether(plasticizer) and polyvinyl chloride, and this solution was used to cast a mem-brane of thickness ∼0.45 mm. Potentiometric analysis of ground water sam-ples, which were spiked with MPA and adjusted to pH 10 in the presence ofTris buffer, yielded a greater response by the MIPs than by the correspondingNIPs, thus indicating an imprinting effect. However, the high level of bind-ing by the NIPs suggests that a substantial degree of nonspecific, presumablyhydrophobic interactions is involved in, or even dominates, the recognitionevent. It should also be noted that since the amount of template recoveredfrom the imprinted polymers was not reported, the possibility of residualtemplate “bleeding” from the MIPs and affecting the responses cannot bewholly discounted.


Imprinting with phosphorylated/phosphonylated template molecules hasbeen particularly valuable in the synthesis of catalytically-active enzymemimicking MIPs. In this approach, the tetrahedral geometry of the phos-phate/phosphonate ester functional group is likened to the tetrahedral transi-tion state of carbonate or ester hydrolysis reactions and the template moleculeis thus referred to as a TSA. The recognition sites in the resulting im-printed polymer are designed to stabilise the high energy transition stateof the hydrolysis reaction, thereby leading to enhanced rates of reaction.This strategy, pioneered by Mosbach [83], emulates a method of produc-ing catalytically-active polyclonal antibodies, viz. immunisation with stableTSAs [123]; however, in contrast to polyclonal antibodies, MIPs are consider-ably more robust and cheaper to produce.

The work of Kawanami et al., who produced a MIP to catalyse p-nitro-phenyl acetate hydrolysis, serves to illustrate the methodology (Fig. 6) [124].p-Nitrophenyl phosphate, employed as a template molecule, was observedto form an insoluble complex when mixed in solution with a 10-fold excessof the chosen functional monomer, 1-vinylimidazole; this was then copoly-merised with a 20-fold excess of divinylbenzene. Following polymerisation,

Fig. 6 Schematic representation of: a molecular imprinting with template (TSA);b imprinted recognition site following template removal; c,d catalysis


the template molecule was removed by means of continuous extraction withacetonitrile, phosphate buffer and methanol, affording an insoluble macro-porous polymer. MIP-catalysed hydrolysis of p-nitrophenyl acetate was ob-served to be 85 times faster than uncatalysed hydrolysis. This was two timesfaster than the catalytic activity of a NIP, thus indicating that catalysis oc-curs within the imprinted sites. Further evidence for this was provided bythe fact that addition of the template molecule, p-nitrophenyl phosphate,clearly inhibits hydrolytic activity. However, the authors conceded that thecatalytic performance of the MIP does not yet compare with that of cata-lytic antibodies for the same hydrolysis reaction. Although no explanationhas been given for this shortcoming, the use of an excess of functionalmonomer (cf. stoichiometric non-covalent imprinting strategy), which wouldhave afforded a heterogeneity of binding sites encompassing a broad spec-trum of affinities and selectivities, is likely to have been one contributingfactor.

Also using a phosphate-derived TSA template, Wulff et al. have demon-strated significant progress in the synthesis of catalytic MIPs modelled oncarboxypeptidase A [115]. The active site of this hydrolytic enzyme featurestwo guanidinium groups (arginine residues) and a Zn2+ cation. One of theguanidinium moieties is thought to bind the oxyanion formed during hydro-lysis, while the metal ion promotes catalysis via polarisation of the substrate;the second guanidinium and a hydrophobic pocket influence substrate speci-ficity. Following this, the authors prepared a MIP with a phosphate diestertemplate molecule employed as a TSA of carbonate hydrolysis (Fig. 7). A func-tional monomer was custom-designed, incorporating an amidinium groupas an H-bond donor (cf. guanidinium groups) and a triamine to coordinateZn2+. This monomer bound to the template molecule through stoichiometricnon-covalent interaction.

Fig. 7 Schematic representation of: a molecular imprinting with template (TSA);b catalysis of diphenylcarbonate hydrolysis


The catalytic performance of the MIP and appropriate controls were in-vestigated with the hydrolysis of diphenylcarbonate. Reaction kinetics, fol-lowed by HPLC analysis of aliquots, were calculated as pseudo first-orderrate constants. Rather encouragingly, the imprinted polymer showed typicalMichaelis-Menten kinetics, in line with natural enzymes. The catalytic activ-ity of the MIP (expressed as kcat/kuncat, the ratio of turnover number of thecatalysed reaction to turnover number in the absence of catalyst) was calcu-lated to be 6900. This is markedly higher than has been reported for catalyticantibodies for carbonate hydrolysis (kcat/kuncat = 810), clearly demonstratingthe great potential of MIPs for use as artificial enzymes.

All of the imprinted polymers discussed above involve phosphory-lated/phosphonylated organic molecules, where the recognition process reliesequally upon binding to, and accommodation of, both the organic backboneand the anionic functional group of the template. In fact, we are aware of onlyone report of a MIP designed to target inorganic phosphate, H2PO4

–. In thisone example, Kugimiya et al. present a series of MIPs prepared using thiourea-derived functional monomers [114]. The thiourea group is known to be a goodhost for anions, particularly for the phosphate anion, to which the N – H moi-eties bind through a set of coplanar H-bonds. The authors’ initial experimentsindicated that 1-allyl-2-thiourea is particularly suitable for use in aqueousmedia: this was in direct contrast to a related functional monomer, N-methyl-N′-(4-vinylphenyl)-thiourea, which showed almost no binding ability underthe conditions studied. Such results were explained by the greater hydrophilic-ity of 1-allyl-2-thiourea. Thus a polymer was imprinted with phenylphosphonicacid as the template molecule, using EGDMA as the crosslinker, acetonitrile asthe porogen and four equivalents (to template molecule) of 1-allyl-2-thioureaas the functional monomer. Incidentally, Kugimiya et al. omit explaination asto why a phosphonic acid template molecule was chosen for a MIP designed forNaH2PO4 recognition. Presumably they had anticipated that the difference ingeometry between these two functional groups would be negligible.

To evaluate the ability of the MIP to selectively bind NaH2PO4, MIP andNIP particles were stirred in distilled water with a mixture of various anions(NaH2PO4, KNO3, NaCl, Na2SO4, CH3COONa). After 1 h, the supernatantwas removed and analysed by ion chromatography. Under these conditions,both the MIP and the NIP preferentially bound phosphate over the other an-ions measured, and the MIP bound 58% more template than the NIP, thusdemonstrating a clear imprinting effect and selectivity for the target anion.Unfortunately, due to the insufficient level of information provided, it is diffi-cult to fully appreciate the data. For example, while the amount of each anionbound is presented as % binding activity, it is unclear how this relates to theestimated number of available binding sites. Furthermore, it was belatedlydiscovered that CH3COO– could not actually be detected via ion chromatog-raphy, and the influence of this potentially-competitive anion was thereforenot determined.


3.2Carboxylates

The prevalence of the carboxylate moiety in both biogenic and man-mademolecules of interest makes this functional group a popular target for an-ion host chemistry. Needless to say, carboxylates are a major constituent ofproteins, peptides and amino acids, and the expansion of proteomics begetsincreasing requirements for means of specific detection of such biomolecules.Other relevant examples of carboxylates include fatty acids, while many smallmolecule di- and tricarboxylates are implicated in key metabolic pathwayssuch as the citric acid cycle (e.g. citrate, succinate, fumarate and malonate).Carboxylated anthropogenic molecules include trichloroacetic acids, anionicsurfactants and β-lactam antibiotics.

While a small number of early examples of molecular imprinting withcarboxylate-bearing templates exist (e.g. Curti et al. imprinting with mandelicacid [45]) in many of these cases the carboxylate moiety was not deliber-ately accounted for. In 1982, Sarhan and Wulff imprinted with D-glyceric acid,using monomers bearing a boronic acid (for covalent interactions) and anamino functionality [125]. In 1990, Andersson and Mosbach prepared MIPsusing N-protected amino acid derivatives as templates with a 4-fold excessof methacrylic acid as the functional monomer, demonstrating that imprint-ing (and enantiomeric resolution) is possible in the absence of both covalentand ionic interactions (binding was predominantly due to H-bond interac-tions) [126]. The imprinting of sialic acid reported in 1996 by Kugimiya et al.is worthy of mention as it used a combination of covalent (boronate esterifi-cation) and ionic interactions for MIP formation, illustrating that this is pos-sible in aqueous media [127]. This brief synopsis illustrates the initial under-representation of carboxylate-targeted MIPs. The situation began to changewith a few groups exploring the use of polymerisable nitrogen bases as po-tential functional monomers for (non-stoichiometric) non-covalent bindingof carboxylic acids [66, 128, 129]. Today, a number of groups focus their ef-forts on constructing MIPs which deliberately target the carboxylate moiety.In particular, various tailor-made functional monomers have been reported,many adapted from naturally-occurring binding motifs, which are designedto bind to the template with affinity sufficient to incorporate them stoichio-metrically into the polymer matrix.

In one of the earliest examples of stoichiometric non-covalent imprinting,Lübke et al. used a combination of custom-designed functional monomersto imprint with ampicillin, a β-lactam antibiotic which bears both anamine and a carboxylate functional group [69]. Rather elegantly, a poly-merisable electron-deficient quinone 2 was used to interact with the amine(via formation of an n–π complex), while the carboxylate was targetedusing a bis(boronate-amide)-functionalised monomer 3 (Figs. 8–10). Thislatter species, adapted from synthetic receptors described by Hughes and


Fig. 8 Functional monomer custom-designed to bind the carboxylate of ampicillin

Fig. 9 Functional monomer custom-designed to bind the amine of ampicillin

Smith [130], exploits internal Lewis acid coordination between the boron andthe carbonyl oxygen, which serves to polarise the amide, thereby increasingaffinity for the oxyanion. Imprinting was carried out by allowing stoichiomet-ric amounts of the monomers and the tetrabutylammonium salt of ampicillinto complex in DMSO over 24 h, prior to incorporation into a highly crosslinkedpolymer. Subsequent extraction with acetonitrile afforded a 92% recovery ofthe template. The resulting MIP showed good potential, performing well inaqueous media with optimal binding at pH 8 (where the carboxylate is presentas the anion) and with selectivity over structural analogues.

Zhang et al. reported the design and four-step synthesis of a novelguanidinium-functionalised monomer which was built upon a vinylan-thracene scaffold to effect fluorescence [116]. NMR titration experimentsin deuteriomethanol indicated the formation of a 1 : 1 complex with ac-etate, with estimated association constants of the order of 105 M–1. Althoughsynthesis of this particular monomer does demand more expertise thanmonomers typically used for imprinting – with somewhat uninspiring re-coveries reported by the authors – it nonetheless appears to be a promisingcandidate for stoichiometric non-covalent imprinting with carboxylates.

Hall et al. are amongst the most prolific authors in the field of car-boxylate imprinting. Making use of expertise in the field of supramolecular


Fig. 10 Schematic representation of ampicillin inside MIP, showing possible binding in-teractions

chemistry, viz. Hamilton et al. [131, 132], they have developed a series ofnovel monotopic and ditopic functional monomers based upon a urea core(Fig. 11) [112, 133]. These synthetically accessible host species can usually beprepared in a one-step procedure, either from 1-isopropenyl-4-isopropyl-2-isocyanate and the corresponding amine or by reaction of 4-vinylaniline withthe corresponding isocyanate. Ureas bind to oxyanions via donation of twoH-bonds, and it was shown that varying the substitution of the monomers– and hence modifying the acidity of the urea core – can result in quitedramatic changes in affinity for the model substrate, tetrabutylammoniumbenzoate. For example, NMR titration experiments in DMSO-d6 afforded anassociation constant of 1322 ± 48 M–1 for 4 where R1 = R2 = H, while thecorresponding value for 4 where R1 = R2 = CF3 was 8820 ± 1600 M–1. Theauthors have explored a number of potential applications for these urea-basedmonomers. These include a class-selective MIP for β-lactam antibiotics, forwhich it was hypothesised that the mono-urea functionalised binding sitedirectly targets the β-lactam carboxylate moiety while providing sufficient


Fig. 11 Functional monomers based upon urea core, designed by Hall et al.

space to accommodate pendant substituents [88, 134]. In another example,a nitrophenyl-functionalised analogue was used to prepare a chromogenicMIP for the enantioselective recognition of Z-glutamate and the related drug,methotrexate; this monolithic polymer exhibited a change in colour intensity,visible to the naked eye, upon binding the target analytes in basic, aque-ous media [135]. Along similar lines, Schmitzer and Gomy have used NMRtitrations, isothermal titration calorimetry (ITC) and molecular modelling todesign, synthesise and screen a library of functional monomers that are alsobased upon the urea motif [113]. Out of 16 candidates, the two monomers 5and 6 (Fig. 12) demonstrated particularly high affinity for the model sub-strate, bis(tetrabutylammonium)-N-Z-L-glutamate. Association constants be-tween this latter species and both 5 and 6 were calculated to be 3370 ±260 M–1 and 4000 ± 400 M–1 respectively (in DMSO-d6). These monomersshould be promising receptors for future stoichiometric incorporation into animprinted polymer.

While most of the aforementioned MIPs were constructed using func-tionalised, highly crosslinked organic polymers, other workers have demon-strated that inorganic matrices, assembled by a surface sol–gel technique,are also valid candidates for anion receptor chemistry [136, 137]. With this


Fig. 12 Functional monomers based upon a urea core, designed by Schmitzer and Gomy

approach, molecularly imprinted ultrathin TiO2 films were prepared by firstallowing titanium butoxide to covalently complex (as confirmed by FT-IRmeasurements) with a carboxylated template molecule in a mixture of tolueneand ethanol (Fig. 13). Water was subsequently added, and the complex wasallowed to chemisorb onto a hydroxylated solid substrate, such as the gatesurface of an ion-sensitive field effect transistor (ISFET) device [138]. Follow-

Fig. 13 Hydrolysis of Ti(IV)-carboxylate template and subsequent rebinding


ing complete hydrolysis, aqueous ammonia was used to remove the template,affording a sol–gel film with imprinted recognition sites. It was suggestedthat rebinding of the template molecule occurs via a combination of covalentbonding, H-bonding, metal coordination, and hydrophobic interactions withTi–O moieties. Such molecularly imprinted films are particularly suited to useas transducer elements in chemosensor devices.

This method was demonstrated to afford molecularly imprinted TiO2films selective for azobenzene carboxylic acids [136, 139], chloroaromaticacids [138], chiral carboxylic acids including amino acid derivatives [140, 141]and anthracenecarboxylic acids [139]. The methodology was also adaptedfor use with water-soluble di- and tri-peptides [142]. In this latter case, eventhough NaOH solution was used as an alternative to aqueous ammonia, com-plete template removal proved impracticable, with 30–40% of the templatetrapped in the film: nonetheless, the TiO2 films were said to show repro-ducible binding for guest molecules. More recently, TiO2 films were also usedfor imprinting thiolate- and phosphonate-functionalised templates [143].

The inherent simplicity of metal oxide sol–gel films, in particular thelack of requirement for carefully chosen functional monomers, makes thisapproach an attractive alternative to the use of highly crosslinked organicpolymers. The technique is optimised for – indeed, necessitates – an anionicfunctional group on the template, and results in imprinted films with a no-table degree of regioselectivity, structural selectivity and enantioselectivity.To date no studies have been published demonstrating the degree of selectiv-ity between substrates with differing anionic functional groups.

Finally, we wish to mention an interesting and somewhat more unusualexample of molecular imprinting, albeit it is debatable as to whether thisparticular application should rightly be included within a discussion onanionic templates. D’Souza et al. copolymerised 6-methacrylamidohexanoicacid and divinylbenzene in the presence of calcite (calcium carbonate poly-morph) crystals and, after exhaustive removal of the template, showed thatthe imprinted polymer promoted nucleation of calcite from an aqueous su-persaturated calcium carbonate solution [144, 145]. Using several methods ofanalysis, it was demonstrated that no template remained in the MIP after thewash steps; subsequent crystal-formation was therefore attributed to hetero-geneous nucleation as opposed to seeding by residual crystals. Regarding thefunctional monomer used, though the authors do not explicitly state theirreasoning behind choosing 6-methacrylamidohexanoic acid, they report thatthe use of other functional monomers (acrylic or methacrylic acids) in placeof 6-methacrylamidohexanoic acid afforded MIPs with inferior nucleatingabilities. It was suggested that the presence of the flexible spacer allowed theacidic head group of the monomer to more easily match the spacing of ionson the surface of the crystal. More recently, Egan et al. applied this samemethodology to imprinting with calcium oxalate crystals, using the resultingMIP to model nucleation of renal calculi from artificial and real urine [146].


3.3Anionic Sulfate Derivatives

While perhaps not as ubiquitous as phosphate derivatives and carboxylates,sulfate derivatives are nonetheless a frequently encountered class of oxyan-ions and are therefore another attractive target for anion receptor chemistry.Important examples of such molecules include sulfated sugars and anionicsurfactants.

Intriguingly, there are actually relatively few accounts of imprinting withsulfates/sulfonates and, with the exception of aforementioned historical re-ports where the anionic component was either not directly targeted [45] orfound to be unnecessary for the recognition event [107], most examples comefrom the last 3–4 years. That said, looking at the binding motifs that tar-get sulfate derivatives in biological systems and in small molecule receptorchemistry, it is apparent that many of the functional monomers designed forphosphate and carboxylate recognition could eventually be adapted for usewith sulfates/sulfonates.

In 2004, Caro et al. imprinted a polymer with 1-naphthalene sulfonic acid,using a 4-fold excess of 4-vinylpyridine as functional monomer (cf. stoichio-metric non-covalent imprinting), EGDMA as crosslinker and AIBN as freeradical initiator [147]. A 4 : 1 v/v mixture of methanol and water was usedas the porogen, due to the high polarity of the template, and it was assumedthat molecular recognition would occur via a combination of hydrophobiceffect and ionic interactions. Following extraction of the template, the poly-mer particles were packed into a polyethylene cartridge, and a mixture ofdifferent polar compounds (including eight naphthalene sulfonates) was per-colated through at pH 2.3. All compounds, except oxamyl and methomyl,were retained on the MIP in the loading step. The MIP was then washedwith MeOH to remove non-selectively bound species. Somewhat unexpect-edly, all eight of the naphthalene sulfonates were retained by the imprintedpolymer in spite of the wash step; these included various mono- and di-sulfonated species, some of which were hydroxyl- or amine-substituted, aswell as the original template, 1-naphthalene sulfonic acid. However, all otherpolar compounds (phenol, nitrophenols, bentazone and 4-chloro-2-methyl-phenoxy acetic acid) were eluted.

The explanation offered by the authors is that the MIP was cross-reactivefor naphthalene sulfonates. However, at pH 2.3, only the naphthalene sul-fonates were anionic (pKa 0.5–0.6). It is quite plausible that in this example,molecular recognition was predominantly an anion-exchange mechanism,and that this overshadowed the influence of any molecularly imprinted cav-ities. In this way, the MeOH wash step could be seen to have disruptednon-covalent, non-ionic interactions, leaving behind the anionic naphthalenesulfonates, which were retained much more strongly via ionic interactionswith the protonated pyridyl moieties. Unfortunately, this is always a poten-


tial hazard with constructing such hosts: there is often a fine balance betweenobtaining high affinity binding interactions while still retaining the selectivityafforded by the act of imprinting.

The recent work of Albano et al. illustrates a particularly innovative ap-plication of molecular imprinting [148]. In the presence of sodium dodecylsulfate (SDS), a thin layer of polypyrrole was electrodeposited onto the sur-face of a quartz crystal. Removal of the SDS by washing with water affordeda molecularly imprinted sensor element which could be used to monitor lev-els of pollutant surfactants in river water. Thus, the piezoelectric quartz sensorwas contained within a flow cell so that the MIP-coated side of the crystal re-mained in contact with analyte solution while the electrodes of the sensor wereconnected to an oscillator circuit. A significant drop in the quartz crystal os-cillation frequency occurred when the sensor was in contact with SDS. This isconsistent with an increase in mass on the surface of the sensor, suggesting thatSDS molecules had bound to the layer of imprinted polypyrrole. A referencepolypyrrole-coated sensor, prepared in parallel to the MIP-sensor but in theabsence of SDS, showed only a minimal response to the target analyte.

The influence of pH on sensor performance was investigated: MIP-coatedsensors exhibiting the highest response and sensitivity were those bufferedat pH 9 during the polymerisation step; likewise, an alkaline measurementsolution (pH 8) was found to afford the best results. It is evident that theanionic form of SDS, rather than the free acid, is required for optimal bind-ing to the pyrrole moieties. With regard to selectivity, the sensor was alsoapplied to solutions of both sodium dodecanoate (structurally identical toSDS except bearing a carboxylate group in place of the sulfate ester) andsodium dodecylbenzenesulfonate. While the sensor showed low sensitivity tothe carboxylate-functionalised analyte, a response equal to that afforded bySDS was attained with the benzenesulfonate analogue. Nonetheless, the degreeto which molecular recognition by the MIP was governed by ionic strength, asopposed to a definite molecular imprinting effect, is as yet unclear.

Recently, Sineriz et al. reported the first example of a MIP imprinted witha sulfated sugar [149]. Sulfated sugars, such as heparan sulfate and chon-droitin sulfate, are a class of highly abundant biological molecules with im-portant intercellular roles. Their activities are directly related to patterns anddegree of sulfation. Understanding their biological functions therefore neces-sitates structure elucidation, yet antibodies developed for this purpose havegenerally been found to be insufficiently selective.

In an initial study, various amine-functionalised monomers (Fig. 14) wereincorporated into polymers templated around glucose-6-O-sulfate. Relativebinding strengths of the resulting MIPs were determined by equilibrationwith glucose-6-O-sulfate solution followed by HPLC analysis of the super-natant. In DMSO, it was found that the MIP made using quaternary amine-functionalised monomer 7 exhibited high affinity towards the analyte but noselectivity (the MIP and corresponding NIP performing equally). In contrast,


Fig. 14 Functional monomers screened by Sineriz et al.

the MIP made using a primary amine (8) was observed to bind up to 80% ofa given concentration of glucose-6-O-sulfate in DMSO, while the correspond-ing NIP showed negligible binding. Consistent with the aforementioned workof Simon and Spivak [109], it was apparent that the directional nature of H-bonds from the primary amine, coupled with the inherent pre-organisationof the imprinted binding site, were crucial for selectivity. Further experi-ments showed that the MIP prepared using the primary amine-functionalisedmonomer preferentially bound glucose-6-O-sulfate over other sulfated sac-charides (galactose-6-sulfate, glucose-3-sulfate and N-acetyl-glucosamine-6-sulfate). The degree of selectivity over other anionic functional groups wasnot investigated.

3.4Other Anions

While relatively few MIPs have been made to recognise the oxyanionicfunctional groups phosphate/phosphonate, carboxylate and sulfate/sulfonate,MIPs designed to target inorganic anions are even more scarce. The followingparagraphs list examples of this mode of imprinting.

Fujiwara and co-workers attempted to surface imprint pyridine-bearingmicrospheres using tetravalent ferrocyanide anions as the template [150]. Ad-sorption studies indicated that, while the imprinted microspheres did showgreater affinity for the template anion in comparison to non-imprinted mi-crospheres, they lacked selectivity and bound all other polyvalent anionstested. This suggests that, although the pyridine groups were apparently pre-organised by the imprinting process, negligible selectivity for the target an-ions may be due to the absence of imprinted cavities on the surface of themicrospheres.

In 1988, Dong et al. reported a chloride-ion selective electrode, preparedvia electropolymerisation of pyrrole in the presence of LiCl [151]. The resul-tant polymer film was observed to give a Nernstian response to chloride, witha limit of detection of 3.5×10–5 M (said to be comparable to that of mostother contemporary chloride-selective electrodes). Later, the same methodol-ogy was adapted to produce a chloride-selective microsensor, used to detectchloride in serum [152]. Both the ion-selective electrode and the microsensor


showed poor selectivity for chloride over other anions. For example, poten-tial selectivity coefficients of the electrode were of the order 10–1 for otherhalides, IO4

– and ClO4–, and >1 for NO3

–, HCOO– [151]. It was postulatedthat selectivity was related to ionic radius.

A rare example of electropolymerisation employed for the formation ofan anion selective crosslinked polymer matrix was introduced by Kamataet al. [153, 154]. Reductive coupling between trifunctional p-cyanopyridiniumcrosslinkers (Fig. 15) in water containing the required counteranion led tostable polymer networks which showed thermodynamic and kinetic anionselectivity. A size exclusion effect was observed whereby only the imprinthalogen (and any smaller halogen anion) was electrochemically recognised.Diffusion of the counterion was reduced in the case of the same template an-ion MIP, but increased with the size of the templating anion. Experimentsclearly demonstrating anion selectivity have yet to be carried out.

Fig. 15 Trifunctional p-cyanopyridinium crosslinker used to prepare halide-imprintedpolymer

In the mid-1990s, the molecular imprinting principle was adopted for thepreparation of a nitrate-selective electrode [155]. Ion-selective electrodes,i.e. sensors which convert the activity of a specific ion into an electricalpotential, have widespread application in biochemical and biophysical an-alysis; however, their major limitation is poor selectivity leading to interfer-ence from other ions. Hutchins and Bachas electropolymerised pyrrole ontoa glass-carbon electrode in the presence of an aqueous solution of NaNO3.The polymer-coated electrode was demonstrated to detect aqueous nitrate,giving a near-Nernstian response, with response times ranging from 24 to<6 s. Recognition appeared to be based on size-exclusion phenomena. There-fore, while the polymer was able to discriminate over traditional interferentswhose radii is larger than that of NO3

–, such as ClO4– and I–, the much

smaller SCN– still effected a response. It was hypothesised that the hydropho-bicity of the polypyrrole films induced the superimposition of a Hofmeister-type selectivity (i.e. based upon lipophilicity) upon the nitrate imprintingselectivity. Thus lipophilic anions not large enough to be sterically hindered


from the film, such as thiocyanate, still interfered with nitrate recognition.Much more recently, this technology has been translated into the productionof nanolitre-scale electrochemical cells [156].

3.5Concluding Comments

Tailoring molecular imprinting for anion recognition is a recent developmentin a relatively young art. To date, the large majority of anion-imprinted poly-mers have been designed to recognise organic molecules which bear one ormore oxyanionic functional group, and we see few examples where the tar-get analyte is a simple, inorganic anion. A plausible explanation for this biasmay be that there is simply less demand for such receptors: after all, simpleanions can be resolved with relative ease using ion chromatography. Yet thiscontrasts sharply with the rest of the synthetic receptor scene, where the focusis very much upon developing host molecules for anions such as phosphate,sulfate, acetate and the halides [10, 157, 158]. As an alternative explanation,it has been suggested that small and simple species are much poorer “im-printogens” than larger, multi-functionalised molecules [159], and that it isgenerally quite difficult to create a well-defined recognition site when lesschemical information is available during the imprinting step. For example,Li and Wu reported an inability to imprint with formate, acetate and pro-pionate [160]. Following computer simulation studies, they suggested thatthe crosslinker (EDMA) had been unable to form adequate cavities for suchsmall, noncomplex molecules. The extent to which this applies to the use ofother crosslinkers has not been investigated.

It is apparent that the current emphasis is towards MIPs that can dis-criminate by both the anionic moiety and the organic residue of a substrate.Progress in this direction is likely to be particularly beneficial to the recentefforts to imprint with proteins [20, 161], or indeed to any application wherethe target species is adorned with multiple anionic functionalities. To sepa-rately target each functional group on a complex template molecule shouldnaturally give rise to better defined binding sites with greater affinity andselectivity for the target species. Equally beneficial will be progress towardscompatibility with aqueous media and pH independent binding.

An interesting topic of discussion concerns the different levels of selectiv-ity required by anion-recognition MIPs; such requirements are dictated by theintended final application of the imprinted polymer. To date, most MIPs havebeen designed to be specific for one particular molecule, and in this regardthe aim has been to optimise imprinting technology so as to make bindingsites as uniform and “monoclonal” as possible. Still, there are a number of in-stances where imprinted polymers benefit from a degree of cross-selectivity.This is illustrated by the aforementioned MIP from Hall et al., which recog-nises compounds bearing the anionic structural motif particular to β-lactam


antibiotics [134]. However, while recognition of a particular (anionic) sub-structure is one notion of cross-selectivity, a possible extension of this wouldbe to develop imprinted polymers that are specific for one particular (an-ionic) functional group, irrespective of the type, class, substructure or size ofthe rest of the molecule (Fig. 16). In a way, this could be thought of as “turn-ing the technology on its head”. Whereas with classic molecular imprintingthe intention is to eradicate all substrate promiscuity, this novel mode ofimprinting would afford binding sites that selectively recognise a particularfunctional group present on the substrate, while discrimination by compoundtype, class, substructure or size is deliberately suppressed. One might theorisethat such “functional group imprinted polymers” (“FIPs”) could be preparedusing an adaptation of the “epitope approach” to imprinting, whereby a sub-structure of the target molecule(s) – in this case the functional group –is used to imprint. Certainly the very act of imprinting, where functionalbinding sites are pre-organised into an exact spatial and geometric arrange-ment, lends itself (conceptually, at least) to discrimination between evennear-isosteric functional groups.

While molecular imprinting, both in general and with regard to anionictemplates, has seen significant development since the first, exploratory ex-periments of Polyakov and Dickey, there is still vast room for improve-ment. The technology will continue to benefit from concomitant advancesin a breadth of disciplines, from supramolecular chemistry to polymer sci-ence to organic synthesis. And yet, the possibilities of molecular imprintingare manifest. We hope that this review will serve to promote the potential of

Fig. 16 Pictorial representation of functional group imprinted polymer (FIP) concept


MIPs, as an extremely versatile class of potent synthetic receptors, where thebenefits for anion recognitions are beginning to emerge.

Acknowledgements Support for S.L.E. through a Chemical Biology Centre studentship atImperial College London is gratefully acknowledged.

References

1. Mullis KB (1994) Angew Chem Int Edit 33:12092. Cheng YJ, Gutmann JS (2006) J Am Chem Soc 128:46583. Smarsly B, Antonietti M (2006) Eur J Inorg Chem: 11114. Antonietti M, Breulmann M, Goltner CG, Colfen H, Wong KKW, Walsh D, Mann S

(1998) Chem Eur J 4:24935. Shibaev V, Bobrovsky A, Boiko N (2003) Prog Polym Sci 28:7296. Conn MM, Rebek J (1997) Chem Rev 97:16477. Tanev PT, Pinnavaia TJ (1995) Science 267:8658. Blondeau P, Segura M, Perez-Fernandez R, de Mendoza J (2007) Chem Soc Rev 36:1989. Hay BP, Firman TK, Moyer BA (2005) J Am Chem Soc 127:1810

10. Kubik S, Reyheller C, Stuwe S (2005) J Incl Phenom Macrocycl Chem 52:13711. Steinke J, Sherrington DC, Dunkin IR (1995) Adv Polym Sci 123:8112. Wulff G (1995) Angew Chem Int Edit 34:181213. Zhang HQ, Ye L, Mosbach K (2006) J Mol Recognit 19:24814. Alexander C, Andersson HS, Andersson LI, Ansell RJ, Kirsch N, Nicholls IA, O’Ma-

hony J, Whitcombe MJ (2006) J Mol Recognit 19:10615. Andersson HS, Nicholls IA (1997) Bioorganic Chem 25:20316. Wulff G (2002) Chem Rev 102:117. Maier WF, Benmustapha W (1997) Catal Lett 46:13718. Davis ME, Katz A, Ahmad WR (1996) Chem Mat 8:182019. Liu ZS, Zheng C, Yan C, Ga RY (2007) Electrophoresis 28:12720. Bossi A, Bonini F, Turner APF, Piletsky SA (2007) Biosens Bioelectron 22:113121. Qiao FX, Sun HW, Yan HY, Row KH (2006) Chromatographia 64:62522. Dirion B, Cobb Z, Schillinger E, Andersson LI, Sellergren B (2003) J Am Chem Soc

125:1510123. Lanza F, Sellergren B (2004) Macromol Rapid Commun 25:5924. Martin-Esteban A, Tadeo JL (2006) Comb Chem High Throughput Screen 9:74725. Southard GE, Van Houten KA, Ott EW, Murray GM (2007) Macromolecules 581:20226. Rachkov A, McNiven S, El’skaya A, Yano K, Karube I (2000) Anal Chim Acta 405:2327. Patel A, Fouace S, Steinke JHG (2004) Anal Chim Acta 504:5328. Karim K, Breton F, Rouillon R, Piletska EV, Guerreiro A, Chianella I, Piletsky SA

(2005) Adv Drug Deliv Rev 57:179529. Wei ST, Jakusch M, Mizaikoff B (2006) Anal Chim Acta 578:5030. Zimmerman SC, Wendland MS, Rakow NA, Zharov I, Suslick KS (2002) Nature

418:39931. Wulff G, Chong BO, Kolb U (2006) Angew Chem Int Edit 45:295532. Wang XJ, Xu ZL, Yang ZG, Bing NC (2007) Prog Chem 19:80533. Batra D, Shea KJ (2003) Curr Opin Chem Biol 7:43434. Piletsky SA, Turner NW, Laitenberger P (2006) Med Eng Phys 28:97135. Caro E, Marce RM, Borrull F, Cormack PAG, Sherrington DC (2006) Trends Anal

Chem 25:143


36. Sellergren B, Allender CJ (2005) Adv Drug Deliv Rev 57:173337. Theodoridis GA, Papadoyannis LN (2006) Curr Pharm Anal 2:38538. Voicu R, Faid K, Farah AA, Bensebaa F, Barjovanu R, Py C, Tao Y (2007) Langmuir

23:545239. Belmont AS, Sokuler M, Haupt K, Gheber LA (2007) Appl Phys Lett 90:340. Berglund J, Nicholls IA, Lindbladh C, Mosbach K (1996) Bioorg Med Chem Lett

6:223741. Breinl F, Haurowitz F (1930) Z Physiol Chem 192:4542. Mudd S (1932) J Immunol 23:42343. Pauling L (1940) J Am Chem Soc 62:264344. Dickey FH (1949) Proc Nat Acad Sci USA 35:22745. Curti R, Colombo U (1952) J Amer Chem Soc 74:396146. Curti R, Colombo U, Clerici F (1952) Gazz Chim Ital 82:49147. Patrikeev VV, Balandin AA, Klabunovskii EI, Mardashev YS, Maksimova GI (1960)

Dokl Akad Nauk SSSR 132:85048. Patrikeev VV, Smirnova ZS, Maksimova GI (1962) Dokl Akad Nauk SSSR 146:70749. Patrikeev VV, Kozarenko TD, Balandin AA (1962) Izvest Akad Nauk SSSR, Ser

Khimi, p 17050. Wulff G, Sarhan A (1972) Angew Chem Int Edit 11:34151. Takagishi T, Klotz IM (1972) Biopolymers 11:48352. Wulff G, Sarhan A, Zabrocki K (1973) Tetrahedron Lett, p 432953. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:336454. Sagiv J (1979) Isr J Chem 18:34655. Rubinstein I, Steinberg S, Tor Y, Shanzer A, Sagiv J (1988) Nature 332:42656. Arshady R, Mosbach K (1981) Macromol Chem Phys 182:68757. Andersson L, Sellergren B, Mosbach K (1984) Tetrahedron Lett 25:521158. Kugimiya A, Matsui J, Takeuchi T (1997) Mater Sci Eng C 4:26359. Piletsky SA, Piletskaya EV, Panasyuk TL, El’skaya AV, Levi R, Karube I, Wulff G

(1998) Macromolecules 31:213760. Wulff G, Sarhan A, Gimpel J, Lohmar E (1974) Chem Ber 107:336461. Lauer M, Boehnke H, Grotstollen R, Salehnia M, Wulff G (1985) Chem Ber 118:24662. Sellergren B, Andersson L (1990) J Org Chem 55:338163. Whitcombe MJ, Rodriguez ME, Vulfson EN (1994) Sep Biotech III 158:56564. Whitcombe MJ, Rodriguez ME, Villar P, Vulfson EN (1995) J Am Chem Soc 117:710565. Wulff G, Knorr K (2001) Bioseparations 10:25766. Kempe M, Fischer L, Mosbach K (1993) J Mol Recogn 6:2567. Wulff G, Gross T, Schonfeld R (1997) Angew Chem Int Edit 36:196268. Yano K, Nakagiri T, Takeuchi T, Matsui J, Ikebukuro K, Karube I (1997) Anal Chim

Acta 357:9169. Luebke C, Luebke M, Whitcombe MJ, Vulfson EN (2000) Macromolecules 33:509870. Sellergren B (1994) Anal Chem 66:157871. Sreenivasan K, Sivakumar R (1999) J Appl Polym Sci 71:182372. Rachkov A, Minoura N (2000) J Chromatogr A 889:11173. Quaglia M, Chenon K, Hall AJ, De Lorenzi E, Sellergren B (2001) J Am Chem Soc

123:214674. Mukawa T, Goto T, Takeuchi T (2002) Analyst 127:140775. Shea KJ, Thompson E (1978) J Org Chem 43:425376. Damen J, Neckers DC (1980) J Am Chem Soc 102:326577. Belokon YN, Tararov VI, Savel’eva TF, Vitt SV, Bakhmutov VI, Belikov VM (1980)

Makromol Chem 181:89


78. Sarhan A, El-Zahab MA (1987) Makromol Chem Rapid Commun 8:55579. Andersson LI, Mosbach K (1989) Makromol Chem Rapid Commun 10:49180. Wulff G, Vietmeier J (1989) Makromol Chem 190:172781. Wulff G (2002) Nanoporous Mater III 141:3582. Leonhardt A, Mosbach K (1987) React Polym 6:28583. Robinson DK, Mosbach K (1989) J Chem Soc Chem Commun, p 96984. Owens PK, Karlsson L, Lutz ESM, Andersson LI (1999) Trends Anal Chem 18:14685. Turiel E, Tadeo JL, Martin-Esteban A (2007) Anal Chem 79:309986. Mayes AG, Whitcombe MJ (2005) Adv Drug Deliv Rev 57:174287. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2005) Anal Chim Acta

552:8188. Urraca JL, Moreno-Bondi MAC, Hall AJ, Sellergren B (2007) Anal Chem 79:69589. Toth B, Pap T, Horvath V, Horvai G (2007) Anal Chim Acta 591:1790. Liu HY, Row KH, Yan GL (2005) Chromatographia 61:42991. Lanza F, Hall AJ, Sellergren B, Bereczki A, Horvai G, Bayoudh S, Cormack PAG,

Sherrington DC (2001) Anal Chim Acta 435:9192. Yoshimatsu K, Reimhult K, Krozer A, Mosbach K, Sode K, Ye L (2007) Anal Chim

Acta 584:11293. Strikovsky AG, Kasper D, Grun M, Green BS, Hradil J, Wulff G (2000) J Am Chem

Soc 122:629594. Schmidt RH, Mosbach K, Haupt K (2004) Adv Mater 16:71995. Ulbricht M (2004) J Chrom B 804:11396. Siemoneit U, Schmitt C, Alvarez-Lorenzo C, Luzardo A, Otero-Espinar F, Concheiro A,

Blanco-Mendez J (2006) Int J Pharm 312:6697. Chronakis IS, Milosevic B, Frenot A, Ye L (2006) Macromolecules 39:35798. Svenson J, Ning Z, Fohrman U, Nicholls IA (2005) Anal Lett 38:5799. Liu Y, Wang F, Tan TW, Lei M (2007) Anal Chim Acta 581:137

100. Dineiro Y, Menendez MI, Blanco-Lopez MC, Lobo-Castanon MJ, Miranda-Ordieres AJ,Tunon-Blanco P (2006) Biosens Bioelectron 22:364

101. Nantasenamat C, Naenna T, Ayudhya CIN, Prachayasittikul V (2005) J Comput AidedMol Des 19:509

102. Dong WG, Yan M, Zhang ML, Liu Z, Li YM (2005) Anal Chim Acta 542:186103. Vaughan AD, Sizemore SP, Byrne ME (2007) Polymer 48:74104. Buchmeiser MR (2001) J Chrom A 918:233105. Enholm EJ, Allais F, Martin RT, Mohamed R (2006) Macromolecules 39:7859106. Corbett PT, Leclaire J, Vial L, West KR, Wietor JL, Sanders JKM, Otto S (2006) Chem

Rev 106:3652107. Bernhard SA (1952) J Amer Chem Soc 74:4946108. Rajkumar R, Warsinke A, Moehwald H, Scheller FW, Katterle M (2007) Biosens Bio-

electr 22:3318109. Simon RL, Spivak DA (2004) J Chrom B 804:203110. Wulff G, Knorr K (2002) Bioseparation 10:257111. Wulff G, Schonfeld R (1998) Adv Mater 10:957112. Hall AJ, Manesiotis P, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2005)

J Org Chem 70:1732113. Gomy C, Schmitzer AR (2006) J Org Chem 71:3121114. Kugimiya A, Takei H (2006) Anal Chim Acta 564:179115. Liu JQ, Wulff G (2004) Angew Chem Int Edit 43:1287116. Zhang H, Verboom W, Reinhoudt DN (2001) Tetrahedron Lett 42:4413117. Turkewitsch P, Wandelt B, Darling GD, Powell WS (1998) Anal Chem 70:2025


118. Moller K, Nilsson U, Crescenzi C (2004) J Chromatogr B 811:171119. Jenkins AL, Uy OM, Murray GM (1997) Anal Commun 34:221120. Jenkins AL, Uy OM, Murray GM (1999) Anal Chem 71:373121. Zhou YX, Yu B, Shiu E, Levon K (2004) Anal Chem 76:2689122. Prathish KP, Prasad K, Rao TP, Suryanarayana MVS (2007) Talanta 71:1976123. Lerner RA, Benkovic SJ, Schultz PG (1991) Science 252:659124. Kawanami Y, Yunoki T, Nakamura A, Fujii K, Umano K, Yamauchi H, Masuda K

(1999) J Mol Catal A 145:107125. Sarhan A, Wulff G (1982) Makromol Chem 183:85126. Andersson LI, Mosbach K (1990) J Chrom 516:313127. Kugimiya A, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1996) Anal Lett

29:1099128. Steinke JHG, Dunkin IR, Sherrington DC (1999) Trends Anal Chem 18:159129. Spivak D, Shea KJ (1999) J Org Chem 64:4627130. Hughes MP, Smith BD (1997) J Org Chem 62:4492131. Fan E, Van Arman SA, Kincaid S, Hamilton AD (1993) J Am Chem Soc 115:369132. Linton BR, Goodman MS, Fan E, Van Arman SA, Hamilton AD (2001) J Org Chem

66:7313133. Hall AJ, Achilli L, Manesiotis P, Quaglia M, De Lorenzi E, Sellergren B (2003) J Org

Chem 68:9132134. Urraca JL, Hall AJ, Moreno-Bondi MC, Sellergren B (2006) Angew Chem Int Edit

45:5158135. Manesiotis P, Hall AJ, Emgenbroich M, Quaglia M, De Lorenzi E, Sellergren B (2004)

Chem Commun, p 2278136. Lee SW, Ichinose I, Kunitake T (1998) Langmuir 14:2857137. Kunitake T, Lee SW (2004) Anal Chim Acta 504:1138. Lahav M, Kharitonov AB, Katz O, Kunitake T, Willner I (2001) Anal Chem 73:720139. Lee SW, Yang DH, Kunitake T (2005) Sens Actuators B 104:35140. Lahav M, Kharitonov AB, Willner I (2001) Chem Eur J 7:3992141. Lee SW, Ichinose I, Kunitake T (2002) Chem Lett, p 678142. Ichinose I, Kikuchi T, Lee SW, Kunitake T (2002) Chem Lett, p 104143. Pogorelova SP, Kharitonov AB, Willner I, Sukenik CN, Pizem H, Bayer T (2004) Anal

Chim Acta 504:113144. D’Souza SM, Alexander C, Carr SW, Waller AM, Whitcombe MJ, Vulfson EN (1999)

Nature 398:312145. D’Souza SM, Alexander C, Whitcombe MJ, Waller AM, Vulfson EN (2001) Polym Int

50:429146. Egan TJ, Rodgers AL, Siele T (2004) J Biol Inorg Chem 9:195147. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom A 1047:175148. Albano DR, Sevilla F (2007) Sens Actuators B 121:129149. Sineriz F, Ikeda Y, Petit E, Bultel L, Haupt K, Kovensky J, Papy-Garcia D (2007)

Tetrahedron 63:1857150. Fujiwara I, Maeda M, Takagi M (2003) Anal Sci 19:617151. Dong S, Sun Z, Lu Z (1988) Analyst 113:1525152. Dong SJ, Che GL (1991) Talanta 38:111153. Kamata K, Suzuki T, Kawai T, Iyoda T (1999) J Electroanal Chem 473:145154. Kamata K, Kawai T, Iyoda T (2001) Langmuir 17:155155. Hutchins RS, Bachas LG (1995) Anal Chem 67:1654156. Lenihan JS, Ball JC, Gavalas VG, Lumpp JK, Hines J, Daunert S, Bachas LG (2007)

Anal Bioanal Chem 387:259


157. Schmidtchen FP (2005) Anion Sensing (Topics in Current Chemistry) 255:1158. Katayev EA, Ustynyuk YA, Sessler JL (2006) Coord Chem Rev 250:3004159. Sun BW, Yang ML, Li YZ, Chang WB (2003) Acta Chim Sin 61:878160. Wu LQ, Li YZ (2004) Anal Chim Acta 517:145161. Turner NW, Jeans CW, Brain KR, Allender CJ, Hlady V, Britt DW (2006) Biotech Prog

22:1474162. Tanabe K, Takeuchi T, Matsui J, Ikebukuro K, Yano K, Karube I (1995) J Chem Soc

Chem Commun, p 2303163. Kubo H, Yoshioka N, Takeuchi T (2005) Org Lett 7:359164. Kempe M, Mosbach K (1994) J Chrom A 664:276165. Caro E, Marce RM, Cormack PAG, Sherrington DC, Borrull F (2004) J Chrom B

813:137166. Haginaka J, Sanbe H (2001) J Chrom A 913:141167. Andersson LI (1996) Anal Chem 68:111168. Oral E, Peppas NA (2006) J Biomed Mater Res A 78A:205169. Seong H, Lee HB, Park K (2002) J Biomater Sci – Polym Edit 13:637170. Asanuma H, Kakazu M, Shibata M, Hishiya T, Komiyama M (1997) Chem Commun,

p 1971171. Hwang CC, Lee WC (2002) J Chrom A 962:69172. Matsui J, Takeuchi T (1997) Anal Commun 34:199173. Zander A, Findlay P, Renner T, Sellergren B, Swietlow A (1998) Anal Chem 70:3304174. Sanbe H, Hosoya K, Haginaka J (2003) Anal Sci 19:715

Subject Index

Adenosine triphosphate (ATP) 96Alkaline phosphatase 102Aluminophosphates 214Amides 2Amidinothiourea 183Amidoferrocenylalkylthiol (AFAT)-gold

647-Amino-4-trifluorocumarin 1116-Aminopicolinic acid 11Ammonium-containing receptors 39Ampicillin 232Anionic templates 175, 207Anion-π interactions 127, 152Anthracene 13, 99Anthracenecarboxylic acids 237Anticrown chemistry 130Aryl bisamide 77ATPase 96Azadendritz 165Azaferrocenophane 71Azobenzene carboxylic acids 237

Bidentate pyrazolyl-based ligands 183Binaphthalene 15Biomimicry 2073,6-Bis(2′-pyridyl)-1,2-pyridazine (bppn)

1683,6-Bis(2-pyridyl)-1,2,4,5-tetrazine (bptz)

1921,2-Bis(4-nitrophenyl) urea 121,1′-Bis(alkyl-N-amido)ferrocene, gold

electrodes 63Bis-bipyridinephenylphosphine oxide 87Bis-cobaltocenium receptor 48Bis-copper(II) cryptates 81Bis-phosphinylferrocene 77Bis-terpyridine Iridium(III) 76Boroxine (BOX) 147Bowl-type structure 198

Cages 182Calix[4]arenes 90–, receptors 52Calix[4]diquinone receptor 77Calix[6]furan 28Calix[n]pyrroles 26–28cAMP 97, 103, 226Capsules 182Carbazole-based receptors 29Carboxylates 232Carboxypeptidase A 230Catechol 32Catenanes 78, 186Cd(II) 85CENOTE phosphorylation 97Charged receptors 33Chemosensors, artificial 98Chloride-selective microsensor 240Cholapod anion receptors 17Chondroitin sulfate 239Cobalt (III) dithiocarbamate cryptands

60Cobaltocenium/cobaltocene 48Coenzyme A (CoA) 96Complexation 1Coordination polymer, bowl-shaped

metallo-macrocycles 182Covalent bonds, reversible 201Cryptates 81Crystal structures 1Crystallographic studies 152p-Cyanopyridinium crosslinker 241Cyanuric acid (CNA) 147Cyclen-Cd complex,

7-amino-4-trifluorocumarin 111Cyclic peptides 202Cyclobarbital 223Cyclotriveratrylene amides 51Cytokine receptor 117

250 Subject Index

Dendrimers, redox anion sensors 60Deoxythymidine 1062,5-Diamidopyrrole skeleton 231,8-Diamino-3,6-dichlorocarbazole 29Diazabicycloalkanes 1295,5′-Dicarboxamido-dipyrrolylmethanes

242,5-Dicarboxamidopyrrole skeleton 21Diindolylquinoxalines receptors 30Diphenyl phosphate 227Diphenylcarbonate hydrolysis 230Diphenylurea 15Dipicatriz 166Dithiocarbamates, redox anion sensing

59Dithiocyanuric acid, 2-ethyleneamine

derivatives 159DNA 96, 98, 225Dodecanethiol 88Dynamic combinatorial chemistry (DCC)

176, 191Dynamic combinatorial library 175

Electrochemical sensing 47ERK2 (extracellular signal-regulated

kinase 2) 97Eu(III) 86Excimers 108EYPC lipid bilayers 4

Ferrocene 48Ferrocene-boronic acid 51[3,3]Ferrocenophane 72Flavin adenine dinucleotide (FAD) 96Fluorescent sensing 95Fluoride-ferrocenium 51

Glucose-6-O-sulfate 239Guanidinium-based receptors 37, 102,

225

H donor, aromatic, neutral receptors 21Halides 2–, π aromatic systems 132Helicates 191Heparan sulfate 239Hexafluorobenzene 132Hexakis(pyridine-2-yl)-[1,3,5]triazine-

2,4,6-triamine 163Hexanickel cage 183

Hexapeptide receptors 11Host/anion “sandwich” 7Host–guest systems 95, 127Hydrogel, supramolecular 119Hydrogel fibers, hydrophobic micro-

domains, phosphate derivatives 120Hydrogen bond donors 2Hydrogen bonding 1Hydroxy (OH) donors, neutral receptors

32Hyper-phosphorylation 117

Imidazolium-based receptors 34Indium tin oxide (ITO) 228Indole-based receptors 29Indolocarbazoles sensors 29Interlocked species 186Ion-sensitive field effect transistor (ISFET)

device 236IP3 97, 102Iridium(III) polypyridine thiourea,

cyclometallated 76Isophthalamide receptors 21-Isopropenyl-4-isopropyl-2-isocyanate

234

Kemp’s triacidic imides 103Kinase-catalysed transfer 225

Lanthanide(III) complexes 86

Macrocycles 178Macrocyclic amide receptors 6Metal complexes, anion sensing 46Metal–ligand coordination bonds 192Metalla-cyclophanes 192Metallocene optical anion sensors 68Metallocene redox anion sensors 48Metallocene-Lewis acid anion receptors

53Metallodithiocarbamate macrocyclic

receptors 59Metallo-squares 193Metallo-tetrahedron 196Metallo-triangle 1966-Methacrylamidohexanoic acid 237Methotrexate 235Methyl orange (p-dimethylamino-

azobenzenesulfonic acid) 222Methylphosphonic acid (MPA) 228

Subject Index 251

Methylpurine 181Micelles 208MIPs 207Mitogen-activated protein kinase (MAPK)

97Molecular interaction potential with

polarization (MIPp) 139Molecular recognition 95, 175Molecularly imprinted polymers (MIPs)

207, 209Multisite phosphorylation 117Myoinositol-1,4,5-triphosphate 97, 102

Naphthalene 15Naphthalene sulfonates 238Nerve agent detection, MIPs 227Neutral receptors 2Nicotinamide adenine dinucleotide (NAD)

96p-Nitrophenyl acetate hydrolysis 229p-Nitrophenylphosphonate 215Nitrophenylurea 82Nucleoside phosphates (nucleotides) 96

Octadecylsiloxane (ODS) 228Octamethylcalix[4]pyrrole 26Oligo-(p-phenylene)-N,N-naphthalene-

diimide (O-NDI) 169Online sensing 207Optical anion sensors 68Organometallic receptors 45Organophosphorous pesticides/herbicides

227

PDGFR-β 117Pentafluorophenyl receptor 162Pentamethyl-amidoferrocene dendrimers

63Perfluoroaromatic compounds 135Pesticides/herbicides, organophosphorous

227PET sensors 12Phenylbutyrate 27Phosphate 95Phosphate anion recognition, coordination

chemistry 104Phosphate derivatives 225–, ratiometric detection 109Phosphate recognition, electrostatic

interactions/hydrogen bonds 98

Phosphoprotein 95Photo-induced electron transfer (PET) 98Pinacolyl methylphosphonate (PMP),

Soman 227Platinum(II), methylpurine 181Point-of-care devices 207Poly(vinylimidazole) 215Polyamine, cyclic/acyclic 98Polyammonium-based anion receptors 39Polyaza receptors 73Poly-guanidinium 102Polynucleotides 208Polypyridyl anion receptors 56Polypyrrole 239Poly-urea 178Prodigiosin mimics 23Protein phosphorylation 225Protein surfaces, phosphorylated 112Pseudopeptidic macrocycle 178Pseudorotaxanes 66, 186Pt(II) 90Pyridine-2,6-dicarboxamide “caps” 7Pyridinium-based receptors 34Pyromellitamide 4Pyrophosphate 99Pyrrole-based receptors 21

Quinoxaline phenanthroline 59

Read-out 207Redox anion-sensing systems, surface

confined 63Resorcin arene 104Resorcinol 32Rhenium(I)tricarbonylchloride 77RNA 96, 225Rotaxanes 67, 78, 186Ruthenium(II)bipyridylcalix[4]diquinone

receptor 73Ruthenium(II) polypyridyl 72Ruthenium-(II)-tris-(2,2′-bipyridiyl),

Ru(bpy)3 101

Sacrificial spacer approach 214Sapphyrin 21Self-assembly 175, 207Semi-wet sensor array, phosphate

derivatives, high throughput sensing119

Sialic acid 232

252 Subject Index

Solid-phase extraction (SPE), MIPs 227Soman 227Spermine/spermidine 98Squaramide-containing macrocyclic

receptor, fluorescent 40Squaramido-based receptors 5Staphylococcus nuclease 102Steroids 33Sulfate derivatives 238Sulfate in water 40Sulfonamide receptors 162Sulfonamide-based receptors 2Supramolecular chemistry 1, 127Surface confined systems, optical anion

sensing 88Surface imprinting 213Symmetry-adapted perturbation theory

(SAPT) 143syn–syn conformation 3Synthetic receptors 207

Tb(III) 86Templating effects 177Terephthalate 178Tetracyanobenzene (TCB) 151, 156Tetrakis(imidazolium) macrocyclic

receptors 35Tetrakis-p-phenylene[68]crown-20 186Tetralactam macrocycles 188Thiocyanuric acid, 2-ethyleneamine

derivatives 159Thiourea 225Thiourea-based receptors 12

Transition metal dithiocarbamates, redoxanion sensing, receptors 59

Transition metal polypyridyl anionreceptors 56

Transition metal-based receptors,redox-active 47

Transition metals, anion-binding groups79

Tren skeleton 2Triazine 133Tricarboxylate anions 37Trichloroacetic acids 232Trifluoro-1,3,5-triazine 133Tris-[(2-pyridyl)methyl]amine 82Tris-amides 2Tris-sulfonamides 2Tweezers 149

UDP-Gal 97, 111Urea 225, 234Urea-based receptors 12Urea-crown ether functionalised ferrocene

53

Vesicle membranes, transport of chlorideions 18

Vinylanthracene 233

Zeolites 208, 214Zinc(II)-dipicolylamine (Zn-Dpa) type

artificial chemosensors 106Zn-cyclen type artificial chemosensors

105

Recognition Of Anions 13ThePoet05

Documents

Transcript of Recognition Of Anions 13ThePoet05