Poul B. Petersen and Richard J. Saykally1 Mar 2006 15:28 AR ANRV272-PC57-12.tex...

1 Mar 2006 15:28 AR ANRV272-PC57-12.tex XMLPublishSM(2004/02/24) P1: IKH

10.1146/annurev.physchem.57.032905.104609

Annu. Rev. Phys. Chem. 2006. 57:333–64doi: 10.1146/annurev.physchem.57.032905.104609

Copyright c© 2006 by Annual Reviews. All rights reservedFirst published online as a Review in Advance on December 7, 2005

ON THE NATURE OF IONS AT THE LIQUID WATER

SURFACE

Poul B. Petersen and Richard J. SaykallyDepartment of Chemistry, University of California, Berkeley, California 94720;email: [email protected], [email protected]

Key Words surface enhancement, electrolytes, interfacial ions, second harmonicgeneration, Jones-Ray effect, Hofmeister series

■ Abstract A qualitatively new understanding of the nature of ions at the liquid wa-ter surface is emerging. Traditionally, the characterization of liquid surfaces has beenlimited to macroscopic experimental techniques such as surface tension and electro-static potential measurements, wherein the microscopic picture then has to be inferredby applying theoretical models. Because the surface tension of electrolyte solutionsgenerally increases with ion concentration, all inorganic ions have been thought tobe repelled from the air-water interface, leaving the outermost surface layer essen-tially devoid of ions. This oversimplified picture has recently been challenged: first bychemical kinetics measurements, then by theoretical molecular dynamics simulationsusing polarizable models, and most recently by new surface sensitive experimentalobservations. Here we present an overview of the nature of the interfacial structure ofelectrolyte solutions and give a detailed description of the new picture that is emerging.

1. INTRODUCTION

It is well-known that the addition of most inorganic salts to water raises the surfacetension approximately linearly with the salt concentration above 0.01 M (1–10).The Gibbs adsorption equation associates an increase in surface tension with anegative surface excess, i.e., with the ions being expelled from the interface (8–10). A molecular interpretation of this surface depletion, first presented by Onsager& Samaras (11), was based on modeling the interface as a sharp discontinuitybetween two continuous dielectric media and the ions as point charges, whereinthe ions are repelled from the interface by the image charge repulsion, as firstdescribed by Wagner (12). This simple model has since been elaborated to includemost interfacial forces (3, 13–22); however, except for a few recent studies ateither dilute concentrations (19, 20) or with a smoothly varying surface (21, 22),continuum models still predict a monotonic distribution of the ions at the surfaceas well as an essentially ion-free region at the surface.

Recent findings have challenged this traditional picture (23). The presenceof ions at the outermost molecular layers of the water surface has recently been

0066-426X/06/0505-0333$20.00 333

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.


334 PETERSEN � SAYKALLY

suggested by the atmospheric community in order to rationalize chemical reactionsfound to occur on aqueous sea-salt particles, ocean surfaces, and in laboratoryaerosol experiments (24–27). In these experiments, surface reactions involvingions are invoked to explain the observed reactions, and a detailed description ofthe ion adsorption to the surface is needed to improve the atmospheric models.This led to molecular dynamics (MD) simulations using polarizable potentials (28–36), which indicate that highly polarizable anions can actually be preferentiallyadsorbed to the outermost liquid layers. In this description, the ions are polarizedby the anisotropy of the interface, creating a dipole that is not present in the bulk.The interaction between the polarized ions and the surrounding water moleculescan, depending on the magnitude of the polarizability, compensate for the reducedsolvation available at the interface. A delicate balance between the interfacial forcesdetermines whether the ions are ultimately repelled or attracted to the surface;i.e., the larger (and more polarizable) halides (I− and Br−) are predicted to bepreferentially adsorbed to the outermost molecular layers, whereas the smallesthalide (F−) is predicted to be repelled from the surface, with Cl− being neutral(28, 31). When anions are preferentially adsorbed, subsequent liquid layers ofthe interface are depleted of the anions, generating a highly structured surfacedistribution, such that the total surface concentration, when integrated over theentire interfacial region, is depleted relative to the bulk value, in accord withthe Gibbs adsorption equation. We note that these predictions were made withclassical MD simulations using simple effective potentials and are subject to theaccompanying approximations and uncertainties.

Recently, this proposed surface enhancement of polarizable anions has beenverified experimentally. Sum-frequency generation (SFG) has been used to studythe surface vibrational spectrum in the OH-stretch region of water for electrolytesolutions (35, 37–47). Although some disagreement in the interpretation of thedata exists, the most recent studies exhibit changes in the SFG spectra that tend toagree with the MD simulations (35, 47, 48). The vacuum technique of photoelectronspectroscopy has recently been employed to aqueous electrolyte solutions (49, 50).Again with some confusion of the interpretation of the data (49), clear enhancementof anions over cations is observed for saturated aqueous films on crystalline saltsurfaces (50).

We have recently employed the surface-specific and molecule-selective tech-nique of resonantly enhanced femtosecond second harmonic generation (SHG)in the UV to directly examine the adsorption behavior of anions at the liquidwater surface (36, 51–53). These studies have unambiguously confirmed an en-hanced surface adsorption of polarizable anions in two distinct concentration re-gions. At low (approximately millimolar) concentrations, iodide and ferrocyanideanions bind strongly to the liquid surface, in accordance with the controversialJones-Ray effect, to generate a dilute but strongly enhanced surface layer. Cur-rently, the molecular origin of this strong binding remains unexplained. At high (ap-proximately molar) concentrations, azide and thiocyanate anions are enhanced atthe outermost liquid layer, as predicted by the MD simulations. Very interestingly,

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.


IONS AT THE WATER SURFACE 335

quantum simulations have predicted that the hydronium cation is also enhanced atthe liquid water surface (54), and this prediction is supported by a combined SFGand classical MD study (35) and by comparing the iodide surface adsorption ofhydriodic acid and alkali iodide solutions in our SHG experiments, wherein theobserved iodide enhancement is much larger for the acid (55).

In this review, we present an overview of both traditional and modern exper-iments and theoretical models, developed to rationalize the surface properties ofelectrolyte solutions. We show that the classical and modern investigations canbe united in a new, consistent picture, wherein salts comprising small hard ionsbehave classically with monotonic surface distributions, and salts comprising asufficiently polarizable anion exhibit a nonmonotonic surface distribution. In thatcase, the anions are enhanced at the outermost liquid layer but depleted in the nextsublayer, where the cations, in turn, are enhanced. Before discussing the proper-ties of electrolyte solutions, we review the description of second-order nonlinearoptical probes, which have greatly enhanced the molecular-level understanding ofliquid surfaces, as well as the properties of the pure water surface.

2. NONLINEAR OPTICAL PROBES OF LIQUID SURFACES

The nonlinear optical spectroscopic techniques of SHG and SFG have been de-veloped into powerful surface probes. As even-order processes, SHG and SFGare forbidden in bulk centrosymmetric media within the dipole approximation andare thus surface-specific probes for liquids (56, 57). In these processes, two pho-tons of either same (SHG) or different (SFG) energies are incident on the sample,and a third photon with the sum of the incident energies is emitted, leaving thechromophore in the initial energy level. Any chromophore satisfying the sym-metry requirement will give a nonresonant contribution, but when a moleculartransition is resonant with either one of the photons or their sum, the processesare resonantly enhanced (by factors as large as 106). Scanning the photon ener-gies while collecting the nonlinear intensity will thus result in a surface spectrumof the chromophore. Usually SHG is employed in the visible range to probe thesurface UV-visible spectrum, while SFG with a visible and IR photon probes thesurface vibrational spectrum. Several reviews have already described these pro-cesses and their applications to liquid surfaces in detail (43–45, 58–64) and onlya brief overview of their respective capabilities is given here.

An electron in a perfect harmonic well will not generate an even-order non-linear response. The second-order nonlinearity comes from the first-order an-harmonicity of the potential. In order to exhibit a first-order anharmonicity, theatomic/molecular orbital must be noncentrosymmetric. This individual atomic ormolecular nonlinear response is called the hyperpolarizability (α

(2)lmn or βlmn) and

comprises a third-rank tensor. The general expression for the hyperpolarizability,as obtained from perturbation theory, is a rather complicated sum-over-states ex-pression (56, 57). Under resonant conditions, however, one term in the sum will

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



dominate, and the expression reduces to a strongly frequency-dependent resonantterm and an approximately constant nonresonant term. The resonant term can beviewed as a product of simpler processes. This has recently been worked out forany nonlinear optical process in a beautiful set of papers by Moad & Simpson (65,66). For IR-vis SFG with the IR frequency (ωIR) resonant, the resonant expressionbecomes (65)

βRlmn = −1

2h

∑

ν

〈g|αlm|ν〉〈ν|μn|g〉ων − ωIR − i�ν

, 1.

where 〈g|αlm|ν〉 is the Raman, and 〈ν|μn|g〉 is the IR transition moment betweenthe initial g and excited state ν, and ων and �ν are the frequency and linewidth ofthe transition. The nonresonant terms sum to a constant nonresonant contribution(βNR

lmn). The resonant contribution can then be viewed as a product of an IR andRaman transition with nonlinear susceptibility βR

lmn . Similarly, for a two-photonresonant SHG experiment we obtain (65)

βRlmn = −1

4h

∑

ν

〈g|μl |ν〉〈ν|αmn| g〉ων − 2ω − i�ν

, 2.

where 〈g|μl|ν〉 is the one-photon and 〈ν|αmn|g〉 is the two-photon absorption crosssection. The resonant contribution is now viewed as a product of a one-photon(dipole) and a two-photon (Raman) absorption process.

The total second-order response for a given system is the susceptibility (χ(2)IJK),

which comprises the sum of the individual hyperpolarizabilities from each atom/molecule (i):

χ(2)IJK =

∑

lmn,i

S(IJK, lmn)βlmn,i = N × 〈S(IJK, lmn)βlmn,i〉Orientation. 3.

Here S(IJK, lmn) is the Euler rotational matrix transforming molecular into lab-oratory coordinates and N is the number of atoms/molecules in the orientationalaverage, denoted by 〈〉Orientation. Due to the coherent nature of the second-orderresponse, two contributions with opposite phase (i.e., from opposite molecular ori-entation) will interfere destructively and cancel. A centrosymmetric distributionof noncentrosymmetric molecules will thus not generate an overall second-orderresponse. Therefore, the symmetry requirement of broken inversion symmetryfor the second-order processes has to be fulfilled on both the microscopic (theatomic/molecular source on the size scale of a molecular orbital) and macroscopic(the distribution of sources within a region on the order of the wavelength of thelight) levels.

For a liquid, the average centrosymmetric orientation of molecules in the bulkwill thus not generate a second-order response, leaving only a surface contributionfrom the net orientation of molecules due to the interfacial asymmetry. Further-more, the asymmetry of the interface can itself induce an asymmetry, and thusa hyperpolarizability, in a otherwise symmetric molecule. This description is all

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



within the dipole approximation. The higher-order quadrupole contribution is usu-ally negligible, but does not exhibit the same symmetry requirement as the dipolecontribution, and could be important for certain systems due to the much highernumber of molecules contributing from the bulk (see Section 3.2).

For an achiral surface with atoms/molecules distributed symmetrically withinthe plane of the interface (symmetric with respect to the angle around the sur-face normal), the 27 susceptibility elements reduce to four independent elements:χ

(2)Z Z Z , χ

(2)X X Z = χ

(2)Y Y Z , χ

(2)X Z X = χ

(2)Y ZY and χ

(2)Z X X = χ

(2)ZY Y . In addition, for SHG, the

last four terms are equal (e.g., χ(2)X Z X = χ

(2)Z X X ). A chiral surface will introduce

additional independent elements (65, 67–69). The independent susceptibility el-ements are probed using different polarization schemes of the incident light. Foran arbitrary incident polarization angle (γ ), the S and P polarized SHG intensitiesare given by (58, 70)

I S2ω(γ ) = C |s1 sin(2γ )χXXZ|2(Iω)2 4.

I P2ω(γ ) = C |s5χZ X X + (s2χXXZ + s3χZXX

+ s4χZZZ − s5χZXX) cos2(γ )|2(Iω)2. 5.

In this equation, the polarization angle of the fundamental beam is defined withγ = 0 being P-polarized light. The constant C depends on the experimentalconditions, while the fitting parameters s1 − s5 depend on both the experimentalgeometry and Fresnel factors (58, 70). By measuring the SHG intensity at differentpolarizations, the fitting parameters can be determined and the different χ

(2)IJK-

elements evaluated. Once the different χ(2)IJK-elements have been evaluated, the

molecular orientation of the surface species can in principle be derived givencertain assumptions. However, great care must be taken in this process, as recentlydescribed (70–72).

Under a given polarization combination and experimental geometry, the differ-ent χIJK-elements add to yield an effective susceptibility (χeff). The SFG intensityis then given by

Iω1+ω2∝ ∣∣χ (2)

eff

∣∣2 × Iω1Iω2

. 6.

For a more compact notation, the resonance behavior is often incorporated directlyinto the effective susceptibility:

χ(2)eff = χ

(2)NR + χ

(2)R = χ

(2)NR +

∑

ν

Aν

ων − ω − i�ν

. 7.

Nonresonant susceptibilities (χ(2)NR) are real, but the resonant amplitudes Aν contain

a complex phase. This can cause difficulties when multiple transitions overlap andthus interfere constructively or destructively, as in the case of SFG in the OH-stretchregion of the water surface.

For the SHG experiments presented later, the effective susceptibility can bedivided into a nonresonant contribution from the water background and a resonant

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



contribution from surface anions, with the nonresonant contribution from the an-ions being negligible or indistinguishable from the change in the water backgrounddue to the presence of the ions in the interfacial layer. We then have

χ(2)eff = χWater

NR + N AnionS × 〈βAnion〉Orientation, 8.

where N AnionS is the surface concentration of the anions. Because the nonlinear re-

sponse of the anions is weak, its interference with the water background signal be-comes important. However, by measuring the SHG intensity on- and off-resonancewith the anions, the contributions from the water background and the anions canbe separated.

3. THE PURE WATER SURFACE

3.1. Classical Methods

When a condensed state is the stable form of any matter, it will exhibit a posi-tive surface tension because the cohesion energy, due to the intermolecular inter-actions, dominates the thermal energy, and the surface atoms/molecules exhibitbroken intermolecular bonds. Surface tension is then the excess Gibbs free en-ergy per unit area needed to create a surface and results from the difference in theforces/pressures acting parallel and perpendicular to the surface in the interfacialregion. The surface tension of pure water is ∼72 mN m−1 at 25◦C (8). This largevalue reflects the strong hydrogen bonding between water molecules.

The electrical surface potential of pure water is due to the net alignment of thedipole and quadrupole moments of the water molecules at the interface. Becauseonly changes in the surface potential can be measured experimentally, the smallsurface potential of pure water is not easily determined. In fact, until recently, eventhe sign of the potential was debated. The general sign convention for the surfacepotential is taken as measured from the air into the solution. Measurements of thetemperature dependence of the surface potential have shown a decreasing potentialwith increasing temperature. As a temperature increase will make the surfacepotential approach zero by randomization of the water dipoles, the surface potentialof pure water has to be positive. The electrical surface potential is estimated tobe between 0.1 and 0.2 V, corresponding to having the water hydrogens pointingslightly toward the bulk (2, 3, 73, 74).

3.2. Surface Probes

SHG does not yield vibrational information, and so far has only been employed forstudies of water at nonresonant wavelengths, but important information has beengained from this technique. The first SHG studies of the pure water surface wereperformed by the Eisenthal group (75, 76). By analyzing the different componentsof the second-order susceptibility tensor and their temperature dependence, theabsolute orientation of water molecules at the water interface was determined,

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



showing the water hydrogens to be pointing into the liquid, in agreement withthe electric surface potential. Furthermore, Kleinman symmetry (equality of theχ

(2)ZXX and χ

(2)XZX elements) was observed to be violated, which was taken as an

indication of a substantial quadrupole contribution to the total SHG intensity. Thiswas substantiated by the nonvanishing nonlinear susceptibility measured at hightemperatures. The dipole contribution is due to ordering at the interface and shouldthus vanish at high temperatures, whereas the quadrupole contribution from theisotropic bulk should persist.

The polarization-dependent experiments on the water surface were later re-peated by Girault and coworkers (77), who compared liquid-liquid interfaces withliquid-air interfaces, and later by Frey and coworkers (78), who also examined thetemperature dependence. Again, based on Kleinman symmetry arguments, Giraultconcluded that the SHG response from the liquid-air interface contains a significantbulk quadrupole contribution, where as liquid-liquid interfaces are dominated bythe surface dipole response. Frey, on the other hand, attributed the breakdown ofKleinman symmetry to dispersion and uncertainty in the extraction of the differentχ

(2)IJK-elements and focused on their temperature dependence. The relatively weak

temperature dependence was in agreement with both the theoretical MD simula-tion by Sokhan & Tildesley (79) and the SFG spectrum (80), described below, andFrey concluded that the SHG intensity of the pure water-air interface is dominatedby the surface dipole contribution, with the bulk quadrupole contribution beingnegligible. Furthermore, it has recently been proven that the Kleinman symme-try rule holds only very far from resonance, which makes it inapplicable to mostnonlinear optical applications (81). It is now widely accepted that both SFG andSHG of liquid surfaces are dominated by the dipole contribution and thus indeedsurface-specific techniques (78, 79, 82–84).

The SFG spectrum of the liquid water surface was first measured by the Shengroup (80). Since then, several other groups have also studied this (45, 47 63). Thedifferences between the reported SFG spectra are due to different experimentalconditions employed, but they all show the same general features. Figure 1 showsthe latest SFG spectrum as taken by the Richmond group. The spectrum consistsof the characteristic sharp feature at 3700 cm−1 and a broad feature between 3000and 3600 cm−1, comprising two major humps. The sharp feature is attributed to afree-OH or “dangling OH”-bond, directed out of the surface and the broad featureto hydrogen-bonded OH-bonds. The number of free-OH bonds at the surface havebeen estimated to be about 20% of a monolayer (80), which is almost the samenumber that one would estimate from a truncated ice structure (25%). The broadhydrogen-bonded feature is generally compared with the IR absorption and Ramanspectra of bulk water. As the red part of the spectrum resembles the spectrum of iceand the blue part resembles that of the liquid, the two parts are usually referred toas the ice-like, or symmetric, and water-like, or asymmetric features, respectively.This general assignment is further supported by the temperature dependence ofthe SFG spectrum, wherein the blue part of the spectrum increases as the red partdecreases with increasing temperature. Curiously, the free-OH feature does not

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 1 SFG spectrum of the pure water surface taken by the Richmond Group

(46). The spectrum show the characteristic free-OH feature at 3700 cm−1 and the

hydrogen-bonded feature between 3000–3600 cm−1 comprised of two major parts.

Reprinted with permission from the Journal of Physical Chemistry (46), copyright

2004, American Chemical Society.

decrease significantly with temperatures up to 80◦C, indicating that the surfaceretains structure even at high temperatures (80). This has lead to the notion thatthe water surface is more ordered, or ice-like, than is the bulk, which is curiousbecause the ice surface is supposed to be covered with a quasi-liquid layer (85).

Attempts have been made to assign substructures within the surface hydrogen-bonded feature to specific molecular environments or structures, but this remainscontroversial. The Richmond group has performed isotope dilution experiments inorder to untangle the contributions to the hydrogen-bonded region (86–88) whilethe others have assigned the substructures within the hydrogen-bonded featureto substructures observed in bulk IR and Raman studies (43, 45, 47). Part ofthe controversy results from problems with deconvoluting the hydrogen-bondedfeature in the SFG spectrum into different sub-bands (Equation 7).

Recently, the very notion of assigning apparent substructures in the hydrogen-bonded feature of both bulk IR and Raman studies to specific molecular structuresor environments has been questioned (89). The so-called multi-state picture ofdividing the hydrogen-bonding environment into specific structures correspondingto separate minima on a potential energy curve has been shown to be invalid, and aso-called continuum description, wherein all the molecules fluctuate rapidly withina single minimum in the potential of mean force, is more appropriate. The apparentsubstructures in the OH-spectra result from short-lived fluctuations within the samepotential minima (89). Reanalysis of the SFG spectrum in these terms thus seemsappropriate.

3.3. Theoretical Models

Numerous MD simulations of the pure water-air interface have appeared (79, 90–99). Due to the need for periodic boundary conditions in such simulations, the

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



so-called slab geometry is used, wherein the water molecules are organized in asheet with two surfaces. In order to properly simulate the water surface, the slab hasto be wide enough to preclude interactions between the two surfaces and properlysimulate bulk water in the middle.

Molecular dynamics simulations are capable of providing a detailed moleculardescription of the interface and can greatly help in assigning and rationalizingthe nonlinear experiments. Generally, good agreement with experiments is found,revealing an interfacial thickness of ∼4 A. The water surface is also observed toexhibit some order. On the vapor side of the interface, most water molecules areoriented with their hydrogens pointing toward the vapor, but on the liquid side ofthe interface, the water molecules have the hydrogens orientated slightly towardthe bulk (79, 95). This is thus consistent with both the SHG and SFG experiments.

Benjamin first calculated the vibrational spectra of interfacial water molecules(93). The SFG spectrum was later calculated by Morita & Hynes (97, 100) andrecently by Perry et al. (99). The SHG response has been calculated by Sokhan &Tildesley (79). The calculations are in good agreement with the experiments. Thefree-OH is shown to originate exclusively from the outermost liquid layer, with20% of the monolayer contributing. The majority (>90%) of the nonlinear responsecomes from the outermost two liquid layers, confirming the surface-specific natureof the processes (79, 97).

Although, the quadrupole contribution to the second-order response cannot becompletely excluded, it is negligible for liquid surfaces, particularly when reflec-tion geometry is used (78, 79, 82–84). However, recent theoretical advances haveenabled the direct calculation of quadrupole contributions from MD simulations(101).

4. AQUEOUS ELECTROLYTE SOLUTIONS

Whereas the traditional surface tension and electric potential measurements onlyindirectly sample the surface structure, the detailed microscopic structure of theinterface can now be probed directly with the development of surface-specificexperimental techniques, such as the SHG and SFG, and extension of X-ray tech-niques to aqueous solutions.

4.1. Surface Tension

Measurements of the surface tension of electrolyte solutions have been repeatedintermittently during the last century, showing qualitatively consistent results, butwith some quantitative variance (1–10). At concentrations above 0.01 M, the sur-face tension increases approximately linearly with concentration, with the mag-nitude of the increase following the charge density of the anions, i.e., the smallhard anions cause a larger surface tension increase than do the large soft anions.The cations only affect the surface tension to a minor degree, however. The sur-face tension increase is interpreted via the Gibbs adsorption equation to yield a net

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



surface depletion of ions with respect to the bulk, when integrated along the surfacenormal through the entire interfacial region.

4.1.1. THE GIBBS ADSORPTION EQUATION AND THE SURFACE EXCESS The decrease/increase in surface tension of aqueous solutions with respect to pure water isrelated to a surface excess/deficit (negative surface excess) by the Gibbs equation,a macroscopic thermodynamic relation (8–10)

dγ =∑

i

�i dμi , 9.

wherein γ is the surface tension (units: dyn cm−1 = mN m−1), �i is the surfaceexcess (moles cm−2) relative to the Gibbs dividing surface of the solvent, μi is thechemical potential (J mole−1), and the summation is taken over all solute species.The surface excess, with respect to a particular position at the interface, is thedifference between the actual concentration integrated over the whole interfacialregion and the integration of the bulk concentration extended to and truncated atthat particular position. The Gibbs dividing surface is defined as the point wherethe water (or more general solvent) surface excess is zero, i.e., the midpoint of thewater density profile, which is approximately where the water density is 50% ofthe bulk.

Because dissolved salt dissociates into cations and anions, the sum will alwaysbe over the multiple components of electrolyte solutions. However, because surfaceneutrality must be obeyed over macroscopic distances (although not at moleculardistances), the surface excess of cations and anions for symmetric electrolytesmust be the same:

dγ = �dμ±, 10.

where � is the salt surface excess and μ± is the average chemical potential of theions. As a macroscopic relation, however, the Gibbs adsorption equation does nottake into account the microscopic distribution of the ions. It is thus possible to havean excess of solute species at the outmost liquid layer of the interface followed bylarger deficiency in subsequent layers in order to effect a net depleted concentration,as compared with the bulk, when integrated through the entire interface.

It is worth noting that direct experimental techniques, such as SHG, SFG andphotoelectron spectroscopy, measure the surface concentration, not the surfaceexcess that is measured by the indirect surface tension experiments. The surfaceexcess is the difference between the surface and the bulk concentration, as recentlydescribed in detail (53). Furthermore, it is very important to note that differentsurface-sensitive probes have different probing depths into the liquid and eachwill accordingly measure different values of various surface properties.

Recently, the surface-sensitive technique Neutral Impact Collision Ion Scat-tering Spectroscopy (NICISS) has been used for a rather interesting applicationof the Gibbs equation (102). NICISS scattering allows for high spatial resolutiondepth profiling of the surface but, due to the vacuum requirements of the technique,has thus far only been applied to nonaqueous liquids with lower vapor pressure.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Normally, the Gibbs equation is used to extract the surface excess of a substancefrom surface tension experiments. However, in doing so, an assumption about theactivity of surface species must be invoked in order to calculate the chemical po-tential in Equation 9. Morgner and coworkers used NICISS to measure the surfaceprofiles directly, which can then be integrated to obtain the surface excess. Thesurface excess is then compared with the surface tension of the solution to obtainthe activities of the solute by the chemical potential.

4.1.2. THE JONES-RAY EFFECT Below 0.01 M, the values of the surface tensions ofelectrolyte solutions are still debated. By using the capillary rise method, Jonesand Ray measured a minimum in the surface tension around 1 mM for 13 differentelectrolytes (103–107). The decreasing surface tension, in the concentration regionbefore the minimum, would imply a positive surface excess and thus a surfaceenhancement of the ions. This contradicted the theoretical models at the timeand the measurements were deemed by Langmuir to be influenced by artifacts.Langmuir argued that changes in the thickness of the wetting layer inside thecapillaries would give rise to an apparent minimum (108, 109). This led to adiscussion of the correct theoretical treatment of the problem (110–113), but theresults were later reproduced by an advanced version of the ring-method that isnot affected by the possible artifacts (114). However, the Jones-Ray effect remainstoday a relatively obscure controversy. A detailed discussion of the Jones-Rayeffect is found in Reference (53).

4.2. Surface Electrical Potential

The surface potential of electrolyte solutions is a measure of both the chargedistribution of the ions as well as the dipole and quadrupole orientation of thewater molecules at the interface. While the actual value of the surface potentialof the pure water surface remains controversial, the change in the potential withelectrolyte concentration is readily measured, as shown by Jarvis & Scheiman(note the opposite sign convention) (1). The topic has recently been reviewed (2,3, 73) and only a brief summary of the results is given here.

Simple 1:1 salts engender a positive change in surface potential with increasingelectrolyte concentration, indicating that the anions approach closer to the surfacethan do the cations. In a few cases, i.e., for larger or divalent ions, a negativechange in the surface potential is observed, indicating that cations can sometimesapproach closer to the interface than the anions. However, as the water dipoles alsocontribute to the surface potential, any changes in their orientation would obscurethe contribution from the ions. The average structure of the aqueous interface,with OH bonds directed into the liquid, could be partly responsible for the anionsapproaching closer to the interface than cations. Again, the anions seem to dictatethe sign of the surface potential change and largely the magnitude, with the cationsonly contributing a minor effect. However, opposite the surface tension changes,larger, softer anions engender a larger change in the surface potential than dosmaller, harder anions. Furthermore, the surface potential does not vary linearlywith concentration, unlike the surface tension.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



The physical properties of the surface structure that give rise to the surfacetension and potential are clearly connected, but the exact relation between them isnot yet understood. The surface tension depends only on the surface excess of theions, which is a measure of the magnitude of the intermolecular attractive forces.This means that the surface tension critically depends on where the ions are locatedwith respect to the water surface, as defined by the Gibbs dividing surface. Thesurface potential, on the other hand, depends on asymmetric electric distributionof anions and cations at the interface, but not with respect to the Gibbs dividingsurface. However, the electric potential changes are convoluted with the changesin the orientation of the water dipoles and quadrupoles due to the presence of ionsin the interfacial layer.

4.3. Theoretical Models

The theoretical treatment of surfaces has traditionally been limited to electrostaticcontinuum models. Continuum models have the advantage of providing generalanalytical solutions to the problem but inherently lack the molecular nature of thesolvent and the asymmetric solvation of positive and negative charges in water.With ever increasing computer power, molecular dynamics simulations have re-cently become feasible for modeling surfaces. MD simulations directly incorporatethe molecular interactions at the expense of a general solution to the problem; i.e.,a simulation has to be run for each system.

4.3.1. CONTINUUM MODELS The first theoretical model of the surface tension ofelectrolyte solution was published by Onsager & Samaras in 1934 (11), wherein theinterface was represented as a sharp, planar discontinuity between two continuousdielectric media and the ions as point charges, and only the image charge force,first suggested by Wagner (12), was taken into account. The ion distribution atthe interface was then obtained by integrating the Poisson-Boltzmann equation.Their model agreed qualitatively with experimental data at low concentrations butcould not describe specific-ion effects. This approach is generally referred to asthe Wagner-Onsager-Samaras (WOS) or the modified Poissant-Boltzmann (MPB)model.

Several improvements have been made to the primitive model, taking intoaccount most molecular effects and forces, such as finite ion sizes, dispersion,ion hydration, and polarizability (3, 13–22). Interestingly, these ion-specific, orHofmeister effects, are the same as those that govern a multitude of biological pro-cesses (115–119). An extensive effort has been made by Ninham and coworkersto evaluate these intermolecular forces and their influence on the surface tensionof electrolyte solutions (118, 120–124). A few recent papers describe the currentstate-of-the-art continuum models (19, 20). The various improvements to the mod-els extend the range of agreement with experiment with the inclusion of empiricalfitted parameters, the physical meaning of which can be questioned. However,all these models involving a discontinuous interface predict monotonic surface

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



distributions, wherein the ions are repelled from the interface, leaving the outer-most surface layer of the interface essentially devoid of ions. Potential deficienciesin the models include the imperfect description of intermolecular forces (118) aswell as the inherent neglect of explicit water-ion interactions, which account forthe different solvation of negative and positive charge by the asymmetric watermolecules, as well as the discrete nature of the solvent.

Interestingly, the models of Karraker & Radke (19) and Manciu & Ruckenstein(20) both incorporate a dilute surface layer of hydroxide at the interface thatgradually becomes exchanged with other anions. These models can explain theelusive Jones-Ray effect (see Section 4.1.2) at low concentrations, as well the highconcentration data, including the specific ion effect, by an appropriate choice ofparameters. It would be interesting to provide an experimental test for the presenceof a hydroxide layer at the pure water interface.

Recent studies have incorporated a smooth, rather than discontinuous, interfacein polarizable continuum models by letting the dielectric constant change con-tinuously from the bulk water value to vacuum over a few angstroms (21, 22).The distribution of halogen anions at the interface was modeled and produced aminimum in the Gibbs free energy just beneath the Gibbs dividing surface. Theminimum was largest for iodide and decreased with decreasing polarizability:GAds = −1.4 kcal mol−1, −0.9 kcal mol−1, −0.7 kcal mol−1, and −0.1 kcalmol−1 for iodide, bromide, chloride, and fluoride, respectively. These values areremarkably close to those obtained in the MD simulations described in the next sec-tion: GAds = −1.5 kcal mol−1, −0.9 kcal mol−1 for iodide and bromide fromDang (30), and GAds = −0.8 kcal mol−1, −0.5 kcal mol−1 from Jungwirth &Tobias (31).

Continuum models have proven to be quite useful in modeling the electro-static interactions over distances that are larger than few molecular distances. Atshorter distances, details of molecular structure and short-range interactions be-come important. Likewise, at distances within a few molecular layers from theinterface, the molecular structure of the solvent interface becomes important andthe approximation of a continuous medium with a discontinuous interface breaksdown. Nevertheless, the approach by Cammi and coworkers of incorporating asmooth interface in their polarizable continuum model (21, 22) seems to capturethe essential physics of the interface and make the results comparable with MDsimulations.

4.3.2. MOLECULAR DYNAMICS SIMULATIONS An alternative approach is to employMD simulations, which directly takes into account explicit water-ion interactions.The surface structure is then strongly dependent on the specific potential usedfor the involved interactions. Nonpolarizable models generally show the ions asbeing repelled from the surface (125–130) whereas the polarizable models showpositive anion adsorption at the interface for larger, more polarizable ions (28–36).The surface density profiles of the sodium halides, as obtained in such a MD sim-ulation, are shown in Figure 2. The large anions are polarized by the anisotropy

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 2 MD simulation of surface density profiles of electrolyte solutions obtained

by Jungwirth & Tobias (31). The dotted lines are the oxygen densities showing the

water surface around z = 12–16 A . The surface enhancement of the anions follow

the polarizability, with fluoride being repelled and exhibiting a monotonic density

profile and iodide being strongly enhanced at the outermost liquid layer but depleted

in the sublayer in a highly nonmonotonic distribution. Reprinted with permission from

the Journal of Physical Chemistry (31), copyright 2002, American Chemical Society.

of the interface and the attractive polarized ion-water interaction compensates forthe loss of solvation. When the polarizability in these simulations is turned off, theions retract into the solution, highlighting that it is explicitly the polarizability ofthe ions and surface water molecules that are responsible for the enhanced surfaceconcentration of the anions. An induced surface dipole of polarizable anions is alsoobserved in the continuum models with a smooth surface (22). Whether or not ananion exhibits surface activity depends on a detailed balance between the electro-static forces, repelling the charge from the surface, and the polarizability, acting toattract the anion to the surface. The surface enhancements have now been demon-strated for several singly charged anions such as the larger halides (28–31), nitrate(32), azide (33), and thiocyanate (36). For multi-charged anions, e.g., sulfate (131,132), the electrostatic forces dominate and the anions are repelled from the surface.

It then remains to rationalize the increased surface tension corresponding tothe negative surface excess, observed experimentally. How can both surface ionenhancement and a negative surface excess be achieved at the same time? The

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



concentration profiles in the MD simulations show an enhanced anion concentra-tion in the outermost liquid layer, but a depletion in the sublayer, where the cations,in turn, are enhanced. This means that the total surface excess, when integratedalong the surface normal through the whole interfacial region, is negative, eventhough the outermost surface layer is enhanced. Hence, there is no contradictionwith thermodynamics (34). The reason for the apparent controversy is the implicitassumption that the concentration profiles are monotonic, such that a negativesurface excess can be related to a corresponding thickness of pure water at thesurface. For the highly polarizable anions in question, the surface concentrationprofiles are highly nonmonotonic and a more detailed picture is needed. On theother hand, small hard ions behave classically, exhibiting monotonic concentrationprofiles and an outermost liquid layer essentially devoid of ions.

A recent study (129) compares the explicit water model with the continuummodel, concluding that the continuum model predicts a qualitatively wrong specificion effect, in that NaCl engendered a higher surface tension increase than did NaF.This is due to the dispersion forces, manifested through B-coefficients, that favorsmall hard ions (small B-coefficient) at the interface over larger ions (larger B).In the continuum model, due to the symmetry between positive and negative ions,this further predicts that cations should approach closer to the surface than anions,in conflict with the surface electrical potential measurements.

Generally, cations are repelled from the surface due to their higher chargedensity and smaller polarizability. However, it has recently been proposed thathydronium approaches the outermost liquid layer (133) and is actually enhancedat the surface (35, 54). The proposed mechanism that brings hydronium to the sur-face is quite different than that operating for anions. Due to the reduced negativecharge on the oxygen atom, hydronium is only capable of forming three hydro-gen bonds (all donor), as opposed to the two acceptor and two donor hydrogenbonds formed among solvent molecules. Hydronium thus perturbs the bulk hy-drogen bonding network and is expelled to the surface, where it is more easilyaccommodated (54), quite like lattice defects in solid-state physics. The existenceof enhanced surface hydronium concentrations could have important effects inbiology, separation science, and other fields.

4.4. Photoemission Experiments

The high vapor pressure of water complicates the application of standard surface-sensitive vacuum techniques, such as photoemission spectroscopy, to aqueous so-lutions. Faubel and coworkers have circumvented this difficulty by the use of liquidmicrojets to study the electronic structure of liquid surfaces via UV photoelectronspectroscopy (134, 135). Recently, Weber et al. (49) used tunable synchrotronradiation to study the surface of sodium iodide solutions. Measuring the bulk con-centration dependence of the photoelectron intensity, they found identical behaviorfor sodium and iodide ions. Their surface concentration profiles show a sub-lineardependence on the bulk concentration, resembling a Langmuir isotherm with aGibbs free energy of approximately 1 kcal mol−1. However, due to the sub-linear

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



concentration profile, they conclude that the ions are depleted, rather than en-hanced, at the interface, in accordance with the Gibbs adsorption equation.

We believe that the apparent discrepancy between their and our own experi-ments can be reconciled by considering the different respective probing depths andthe fact that both experiments measure the surface concentration, not the surfaceexcess. The cations exhibit a concentration maximum just below the water surfaceand thus a corresponding minimum in the Gibbs free energy. Given that photoelec-tron experiments probe a depth of a few nanometers, and thus sample the entireinterfacial depth, cations should then indeed show a Langmuir behavior, which,due to charge neutrality, must be the same as for iodide. Since the position of theminimum in the Gibbs free energy with respect to the Gibbs dividing surface is notknown, the Langmuir behavior does not automatically imply a surface enhance-ment in the outermost liquid layer or a net surface excess. However, the Langmuirbehavior does establish a nonmonotonic ion distribution at the interface, and theions must accordingly be enhanced at some part of the interface. Whether or notthe position of this concentration maximum is at the outermost surface layer mustbe determined by theoretical calculations.

Salmeron and coworkers circumvented the difficulty of the high vacuum pres-sure of water by studying the thin water film on salt crystals above the deliquescencepoint (50). In this case they were limited to examining the saturated solution ontop of the crystal. Their results clearly show an enhanced anion/cation ratio atthe interface that gradually decays as the probing depth is increased, as shown inFigure 3. Furthermore, the anion/cation ratio is larger for iodide than for bromide,

Figure 3 Photoemission experiments of Salmeron and coworkers (50). Panel a shows

the surface composition of the saturated solutions obtained at 200 eV. The dotted lines

indicate the bulk composition. The surface is observed to be enhanced in the anions

and depleted by water. Panel b shows the anion to cation ratio in the surface layer

as a function of the photoelectron kinetic energy, which is related to the probing

depth. The surface enhancement is clearly observed to decay at higher kinetic energies

corresponding to longer probing depths. Reprinted with permission from Science (50),

copyright 2005, AAAS.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



as predicted by the computer simulations, but the extent of the surface enhancementis significantly larger than predicted (50).

4.5. SFG Experiments

The first SFG experiments probing the surface OH-stretch region of inorganic saltand acid solutions were done by the Shen (37) and Shultz (38–42) groups, assummarized in their recent review papers (43–45). While the Shen group stud-ied sulfuric acid-water mixtures, the Shultz group studied both sulfuric, nitric,and hydrochloric acid and their salts. Generally, acids were found to affect thehydrogen-bonding network at the surface more than salts do. At low concentra-tions, an increase in the intensity of the hydrogen-bonded part of the spectrum isobserved, while the free-OH feature remains mostly unaffected. At high concen-tration, all the spectral features disappear from the SFG spectrum.

The interpretation of the spectral changes, however, is different between the twogroups. While the Shen group attributes the low concentration changes to a moreordered hydrogen-bonding network induced by bonding between acid and watermolecules and the disappearance of the spectral features at high concentrationsto surface crystallization, the Shultz group interprets their results in terms of thedouble-layer model. Because anions approach closer to the surface than do cations,an electric double layer forms. At low concentrations, the ions approach close tobut do not penetrate the outermost surface layer, as indicated by the undiminishedfree-OH feature. The hydrogen-bonded features increase as the water moleculesare rotated out of their nearly parallel alignment with the surface plane due tothe electric field generated by the double layer. At high concentrations, significantion pairing occurs at the interface, neutralizing the double layer and removing thehydrogen-bonded part of the spectrum, while the free-OH disappears as the neutralion pairs penetrate the outermost surface layer and bind to the dangling OH bonds.

Recently, the Richmond (46) and Allen (47) groups have used SFG to probethe water surface structure of sodium halide solutions. Both groups examined saltsolutions in the 0.5–2 M range. The Richmond group studied sodium halide solu-tions at a mole fraction of 0.030 (around 1.6 M), varying the isotopic distributionsof the water, and attributed the small changes observed in the water spectrum tothe perturbation from ions, but concluded that the small changes observed indi-cated that the ions were not present in the outermost liquid layer. The Allen groupexamined sodium halide solutions at mole fractions of 0.015 and 0.036 (around0.8 and 1.9 M, respectively). As the SFG transition can be viewed as the productof a IR and Raman transition (Equation 1), they compared the spectral changes inthe hydrogen-bonded part of the SFG spectrum with the product of the changes inthe bulk IR and Raman spectra. The changes were much larger for iodide than bro-mide, while the chloride and fluoride solutions showed only minor changes. Theyinterpreted their results as an indication of increased interfacial width effected bythe addition of the ions.

In a combined theoretical and experimental effort, the SFG spectrum and MDsimulations of halide acid, hydroxide and alkali halide salt solutions were compared

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



(35). While the simulation shows a clear preference of both the larger halides andhydronium for the surface, hydroxide is repelled from the surface. The predictionswere found to be in agreement with the measured SFG spectra (35).

The main difficulty in interpreting the SFG experiments of electrolyte solutionsis that they probe how the solvent structure at the surface is perturbed by theions. The interpretation of the spectra then has to rely on detailed assignmentsof the various spectral features and their changes to perturbations by the ions. Arecent MD simulation investigates the effects of salts on the surface water structureand concludes that the salts only have minor effects on the contribution from theoutermost surface layer, even though the ions are actually present therein, butsignificantly perturb the water structure in the adjacent sublayer (48). This agreeswith the experimental observations and the interpretation by Allen and coworkers,but disagrees with Richmond and coworkers, due to their different assignmentsand interpretations of the spectral features. This new analysis of the changes in thesurface water structure is also consistent with the earlier observations by the Shenand Shultz groups.

4.6. SHG Experiments

Our contribution to the field has been to use SHG to directly measure the adsorp-tion of several small ions on the surfaces of electrolyte solutions. By choosing thetwo-photon resonant SHG scheme, we are able to directly probe the anions viatheir charge-transfer-to-solvent (CTTS) transitions in the UV. CTTS transitionsare known to exhibit very large nonlinear cross sections due to the large chargeseparation associated with the transition. To date, we have studied four differentanions: iodide (I−) (51), ferrocyanide [Fe(CN)4−

6 ] (53), azide (N−3 ) (52), and thio-

cyanate (SCN−) (36). Their different CTTS bands are shown in Figure 4. In thesame spectral region, the water hyperpolarizability is nonresonant and thus does

Figure 4 Bulk UV-vis spectra of examined salts. The strong CTTS transitions are

probed by two-photon resonant SHG corresponding to incident light at twice the CTTS

wavelength.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 5 SHG intensity of NaI at 225 nm as function of bulk concentration. The SHG

response can be divided into three different concentration ranges. A rapid decrease at

dilute concentrations (blue) due to salt adsorption, an intermediate range (green) due

to a change in the water background from the presence of the ions in the interfacial

layer, and a high concentration range (red) due to surface enhancement of anions at

the outermost liquid layer.

not change with wavelength (Equation 8). By measuring the SHG intensity on andoff resonance with the anions, the total SHG intensity can be separated into thecontributions from the anions and the water background, and the surface concen-tration of the anions can be extracted. In order to evaluate the surface enhancementof the anions, the surface concentration of the anions is measured as a function ofbulk concentration and then fitted to the Langmuir adsorption model, from whichthe Gibbs free energy of adsorption of the anions is determined.

The concentration dependence of the SHG intensity can be divided into threedistinct concentration regions as shown in Figure 5 for NaI. At dilute concentra-tions, the SHG intensity initially decreases rapidly due to destructive interferencebetween resonant and nonresonant contributions resulting from the formation of asurface layer of anions in the millimolar range. This surface enhancement is con-sistent with and attributed to the Jones-Ray Effect (Section 4.1.2). At intermediateconcentrations, the SHG intensity increases due to the change in water structureinduced by the presence of anions in the interfacial layer. This change is nonreso-nant; i.e., it occurs at all wavelengths. At high concentrations, the SHG intensityincreases further due to the resonant contribution from the anions. This surface en-hancement at molar concentrations is attributed to the high anionic polarizability,as observed in the computer simulations.

After describing the Langmuir model, which is used to evaluate the surfaceenhancement of the anions and derive their Gibbs free energy of adsorption, wesummarize the dilute and concentrated surface enhancements observed. For fur-ther details we refer readers to the individual papers (36, 51–53, 55). The resultsobtained in the intermediate concentration range and the details of the laser systemand experimental setup will be described in a future publication (136).

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



4.6.1. THE LANGMUIR ADSORPTION MODEL The simplest and most commonlyused surface adsorption model is the Langmuir model, which assumes a fixednumber of identical active sites or positions at the surface, which the surfactantcan adsorb to, and negligible surfactant interactions. A discussion of the validityof the Langmuir model to electrolyte solutions is found in Reference 36. In thiscase, the surface concentration is given by

NS = N maxS

KC

KC + Cw

≈ N maxS × C

C + 55.5M × exp(GAds/RT ), 11.

where NS is the surface concentration, N maxS is the number of active surface sites,

K is the equilibrium constant for surface adsorption, C and Cw are the bulk so-lute and water concentration, respectively, and GAds is the Gibbs free energyof adsorption. When measuring the surface concentration as a function of bulkcomposition, the shape of the resulting curve (reflecting the concentration rangewherein the surface adsorption saturates) gives the Gibbs free energy of adsorption,and the asymptotic surface concentration gives the number of surface sites.

For a real system, the Gibbs free energy of the solute is a function of the distanceof the solute from the surface. A minimum in this Gibbs energy profile thus definesa surface layer, and the experimentally measured Gibbs free energy of adsorption isthe difference between the bulk and surface energy, averaged over the probing depthof the experiment. However, the absolute position of the minimum, or the solutesurface, with respect to the Gibbs dividing surface, or water surface, is not known.

4.6.2. THE DILUTE CONCENTRATION ENHANCEMENT The effective nonlinear sus-ceptibilities of iodide and ferrocyanide as a function of the bulk concentration areshown in Figure 6. Assuming that the orientation of the anions is independent of theconcentration, the effective nonlinear susceptibilities of the anions are proportional

Figure 6 Surface enhancement at low concentrations. The effective second-order

susceptibility at various wavelength at dilute concentrations for NaI (panel a) and

K4Fe(CN)6 (panel b). All wavelengths are fitted to the same Langmuir isotherm with

a magnitude of the SHG response reflective of the surface spectrum. Due to the strong

adsorption the surface is saturated around 10 mM.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



to the surface concentration with a magnitude that depends on the second-ordercross section at the given wavelength, i.e., the surface spectrum. The curves arefits to the Langmuir adsorption model with Gibbs free energies of −6.2 ± 0.2kcal mol−1 and −6.8 ± 0.3 kcal mol−1 for iodide (51) and ferrocyanide (53),respectively. This strong binding reflects the surface saturation occurring in themilli-molar concentration range, as observed in the surface tension data of Jones& Ray (103–107).

To directly compare the SHG results with the surface tension, the surface ex-cess has to be calculated from the surface concentration. This is done by assuminga simple model for the surface, consisting of a surface layer of a given thick-ness, wherein the surface concentration follows the Langmuir model with a givenGibbs free energy of adsorption and maximum surface coverage. The details ofthe model are described in Reference 53. The model thus has three parameters andeffectively reproduces the minimum in the surface tension with reasonable param-eter values: surface thicknesses of 6–9 A, maximum surface coverages of 1010–1011 cm−2 corresponding to 20.000–300.000 water molecules per anion or inter-anionic distances of 100–500 A, and Gibbs free energies of –7–10 kcal mol−1

(53). This is an even stronger surface binding energy than measured in the SHGexperiments and could be due to the simple model used. The fits to all the 13Jones-Ray salts are shown in Figure 7.

The molecular mechanism that engenders the strong surface binding of theJones-Ray effect remains to be explained. The effect could be induced by thecations, which are, in general, repelled from the surface and accumulate belowthe surface. The anions, which are less repelled from or even attracted to thesurface, adsorb to the outermost surface layer to compensate for the positive chargefrom the cations. This mechanism would explain the observed increased Gibbs freeenergy of adsorption for multi-charged cations and the disappearance of the Jones-Ray effect for acidic solutions, as the hydronium ions penetrate to the outermostsurface layer and repel the cations (53).

4.6.3. THE HIGH CONCENTRATION ENHANCEMENT The origin of the surface en-hancement observed at high concentrations is qualitatively different that at lowconcentrations. Figure 8 shows the effective nonlinear susceptibility of azide andthiocyanide as a function of the bulk concentration. Again, the different magnitudesobserved at different wavelengths reflects the surface spectrum of the anion, andthe lines are fits to the Langmuir model with Gibbs free energies of −2.4 ± 0.1 kcalmol−1 and −1.80 ± 0.03 kcal mol−1 for azide and thiocyanate, respectively. In thiscase, the binding energies are not strong enough to cause complete surface satura-tion within the solubility limit of the salt, and only gentle deviations from straightlines are observed in the graphs. The measurements are in good agreement with theMD simulations (33, 36) and the measured energies of the less polarizable anionsfrom both continuum models and MD simulations, which are discussed in Section4.3.1. The SHG experiments thus provide direct proof of the surface enhancementof polarizable anions at the molar concentration relevant to atmospheric chem-istry. Unfortunately, the second-order response from iodide is not strong enough to

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 7 The Jones-Ray surface tension data. The change in the surface tension of

all 13 Jones-Ray salts are fitted to a simple model, as described in the text. Good

agreement is found with even larger surface adsorption energies as determined in

the SHG experiments. Reproduced with permission from Journal of the AmericanChemical Society (53). Copyright 2005 American Chemical Society.

observe a significant deviation from a straight line and thus derive the Gibbs freeenergy of adsorption for this region. However, a Gibbs free energy of around−1 kcal mol−1 cannot be excluded and thus provides a sensitivity limit for theextraction of a Gibbs free energy from the SHG measurements via the Langmuirmodel.

4.6.4. ENHANCED HYDRONIUM CONCENTRATION AT THE SURFACE As described inSection 4.3.2, it has recently been proposed that hydronium is actually enhancedat the water surface, as opposed to the behavior of ordinary cations. To test this, weperformed SHG experiments of hydriodic acid solutions and found a 55 ± 10%and 34 ± 7% higher surface concentration of iodide as compared with sodiumand potassium iodide, respectively, at 4 M bulk concentration (55). The SHGintensities observed for hydriodic acid, sodium and potassium iodide solutions isshown in Figure 9. The higher surface iodide concentration for hydriodic acid canbe understood by considering the inter-ionic forces acting in the surface region.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 8 Surface enhancement at high concentrations. The effective second-order

susceptibility at various wavelengths at high concentrations for NaN3 (panel a) and

NaSCN (panel b) is shown. All wavelengths are fitted to the same Langmuir isotherm

with a magnitude of the SHG response reflective of the surface spectrum. The

∼2 kcal mol−1 adsorption energy in the high concentration range is not adequate

to completely saturate the surface within the solubility limit of the salt. Panel b re-

produced with permission from Journal of Physical Chemistry (36). Copyright 2005

American Chemical Society.

Figure 9 Surface hydronium enhancement. Panel a shows the SHG intensity for hy-

driodic acid as compared with sodium and potassium iodide. A 55 ± 10% and 34 ±7% higher surface concentration of iodide is found for hydriodic acid compared with

sodium and potassium iodide, respectively, and is indirect evidence for an enhanced

hydronium concentration at the interface. Panel b shows the molecular mechanism pro-

posed by Voth and coworkers (54) for the surface adsorption of hydronium. Hydronium

is only capable of forming three hydrogen bonds in the bulk (circle c) as compared

with the four for water (circle b). The hydronium ion can then be considered a lattice

defect in the hydrogen bonding network and is pushed to the surface (circle a) where

it is accomodated more easily. Reproduced with permission from Journal of PhysicalChemistry (55). Copyright 2005 American Chemical Society.

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



As polarizable anions are attracted to the surface, a negative surface layer buildsup. The cations, which otherwise are repelled from the surface, are now attracted tothis negative layer of the anions, rendering the surface electrically neutral, althoughan electrical double layer structure may persist. By Newton’s third law, the cation inturn imposes a force on the anions, pulling them away from the surface and into thebulk. The cations thus limit the extent of the anion enhancement at the outermostsurface layer. This can also be understood in terms of the requirement of surfaceneutrality. Because the cation and anion concentration as integrated over the entiresurface region has to be the same, their properties become entangled and theirsurface excess is given by their average propensity for the surface. A cation that isless strongly repelled from, or even enhanced at, the surface will allow for higheranion concentration at the surface. The increased surface iodide concentration forhydriodic acid compared with sodium and potassium thus constitutes evidence thathydronium is less repelled from the surface than either sodium and potassium, andtaking the magnitude into account, it may even be enhanced at the surface (55),relative to the bulk.

5. SURFACE STRUCTURE AND EXPERIMENTALCONSIDERATIONS

The new results emerging from both theoretical calculations and experimentalinvestigations indicate a revised picture of the surface structure of electrolyte solu-tions. This emerging picture refines, rather than contradicts, the current textbookdescriptions. Given the previous lack of suitable surface-specific techniques todirectly probe interfacial structure, the concentration of electrolyte in the surfacelayer was assumed to vary monotonically with the distance from the surface. Asthe surface tension increases for all salt solutions (1–10), all ions were thought tobe repelled from the surface, with a concentration increasing monotonically fromzero to the bulk value, leaving the outermost surface layer virtually devoid of ions.The simple continuum models support this traditional picture (3, 13–20). How-ever, inorganic acids constituted a problem for this description. Acids lower thesurface tension and should thus be enhanced at the surface, but the classic theorywould not accomodate for this and speculation of ion pairing at the surface aroseas a possible source of surface enhancement. Moreover, the classical picture couldnot explain the reaction kinetics found in the recent atmospheric investigations(24–27).

The new results suggest a refined picture, consistent with both theory and ex-periments. Solutions of small hard ions, as well as multiply charged ions, behaveclassically, and the ions are repelled from the interface with essentially monotonicsurface profiles. For electrolyte solutions containing highly polarizable ions, theanion concentration profiles exhibit a highly nonmonotonic dependence on thedistance from the surface. The MD simulations (28–36) and continuum modelswith a smooth interface (21, 22) suggest that a dipole is induced in the anion at the

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



Figure 10 Surface structure of iodide. Due to the high polarizability, a dipole is

induced in the otherwise spherical iodide anion at the surface. The attractive interaction

between the induced surface dipole and the surrounding water molecules compensates

for the electrostatic repulsion and the loss of solvation energy.

surface, as shown in Figure 10, compensating for the electrostatic repulsion andthe anions are enhanced at the outermost liquid layer. The cations are drawn towardthe interface by the electrostatic interactions with the anions and are enhanced justbelow the outermost liquid layer, where the anions are then depleted. Accordingly,there will be a minimum in the Gibbs free energy for the anions at the outermostliquid layer but a maximum just below it, while for the cations; there will be aminimum just below the outermost layer. Hydronium constitutes an exception tothis behavior and is also enhanced at the interface, allowing for a large ion concen-tration and thus a positive surface excess, with a corresponding decreased surfacetension. Furthermore, and very importantly, the polarizable anions present in theoutermost surface layer are available for chemical reactions, in accordance withthe observed reaction kinetics (24–27). Perhaps future investigations will revealsimilar surface activity for hydronium ions. The new picture thus explains thefeatures of the traditional methods, of a smaller change in the surface tension buta larger change in the electric potential, of salts comprised of highly polarizableanions compared with small hard anions, and the negative surface tension increaseof acids.

Several surface sensitive experiments provide support for this new picture ofion solvation at the water surface (35, 36, 47, 49–53, 55). The differences betweenthe various observations can be understood by considering how each experimentprobes the surface structure. The greatest detail is obtained in the simulations,but due to uncertainties in the potentials, the results have to be verified throughexperiments. A particular surface-sensitive technique will measure a weightedaverage concentration over the probing depth of the experiment, and may thus

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



yield quite different results, depending explicitly on this probing depth. If anexperiment probes the entire surface region evenly (i.e., a probing depth muchlarger than 10 A), then the cations will show a behavior that is identical to thatof the anions due to surface neutrality. However, if the probing depth is shallowerthan the entire surface region, or if the outermost layers dominate the average takenover the surface region, then the distinct behavior of the cations and anions can beobserved.

Furthermore, the various surface-sensitive experimental techniques probe thesurface concentration and not the surface excess. An experiment that shows surfaceenhancement indicates a corresponding minimum in the Gibbs free energy profileat the surface. However, the position of this minimum with respect to the Gibbsdividing surface is normally not known and thus complicates the calculation of thesurface excess. This is further exacerbated if the probing depth is unknown. Thusgreat care must be taken in determining a surface enhancement or depletion witha particular experiment.

An experiment may indicate a minimum in the Gibbs free energy profile, thusproving the existence of a nonmonotonic surface distribution. However, surfaceenhancement in the outermost liquid layer or true surface enhancement, as indi-cated by a positive surface excess, can normally not be established experimentallywithout additional knowledge about the position of the Gibbs dividing surface orinferred surface structure from a theoretical calculation or simulation.

ACKNOWLEDGMENTS

This work is supported by the Experimental Physical Chemistry Division of theNational Science Foundation. P.B.P. was also funded by the Danish ResearchTraining Council.

The Annual Review of Physical Chemistry is online athttp://physchem.annualreviews.org

LITERATURE CITED

1. Jarvis NL, Scheiman MA. 1968. Sur-

face potentials of aqueous electrolyte so-

lutions. J. Phys. Chem. 72:74–78

2. Randles JEB. 1977. Structure at the free

surface of water and aqueous electrolyte

solutions. Phys. Chem. Liq. 7:107–79

3. Conway BE. 1977. The state of water and

hydrated ions at surfaces. Adv. Colloid In-terface Sci. 8:91–211

4. Hey MJ, Shield DW, Speight JM, Will

MC. 1981. Surface tensions of aque-

ous solutions of some 1:1 electrolytes. J.

Chem. Soc. Faraday Trans. 1(77):123–

28

5. Weissenborn PK, Pugh RJ. 1995. Surface

tension and bubble coalescence phenom-

ena of aqueous solutions of electrolytes.

Langmuir 11:1422–26

6. Weissenborn PK, Pugh RJ. 1996.

Surface tension of aqueous solutions

of electrolytes: Relationship with ion

hydration, oxygen solubility, and bubble

coalescence. J. Colloid Interface Sci.184:550–63

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



7. Matubayasi N, Tsunetomo K, Sato I, Ak-

izuki R, Morishita T, et al. 2001. Thermo-

dynamic quantities of surface formation

of aqueous electrolyte solutions. J. Col-loid Interface Sci. 243:444–56

8. Adamson AW, Gast AP. 1997. PhysicalChemistry of Surfaces. New York: Wiley.

6th ed.

9. Durand-Vidal S, Simonin JP, Turq P.

2000. Electrolytes at Interfaces. Dor-

drecht: Kluwer Acad.

10. Fawcett WR. 2004. Liquids, Solutions,and Interfaces. Oxford: Oxford Univ.

Press

11. Onsager L, Samaras NNT. 1934. The

surface tension of Debye-Huckel elec-

trolytes. J. Chem. Phys. 2:528–36

12. Wagner C. 1924. Die Oberflachens-

pannung verdunnter Elektrolytlosungen.

Phys. Z. 25:474–77

13. Bhuiyan LB, Bratko D, Outhwaite CW.

1991. Electrolyte surface tension in the

modified Poisson–Boltzmann approxima-

tion. J. Phys. Chem. 95:336–40

14. Stairs RA. 1995. Calculation of surface

tension of salt solutions: Effective polar-

izability of solvated ions. Can. J. Chem.73:781–87

15. Markin VS, Volkov AG. 2002. Quantita-

tive theory of surface tension and surface

potential of aqueous solutions of elec-

trolytes. J. Phys. Chem. B 106:11810–17

16. Maheshwari R, Sreeram KJ, Dhathath-

reyan A. 2003. Surface energy of aque-

ous solutions of Hofmeister electrolytes

at air/liquid and solid/liquid interface.

Chem. Phys. Lett. 375:157–61

17. Ohshima H, Matsubara H. 2004. Surface

tension of electrolyte solutions. ColloidPolym. Sci. 282:1044–48

18. Woelki S, Kohler HH. 2004. Effect of

dispersion forces on the potential of

charged interfaces. Chem. Phys. 306:209–

17

19. Karraker KA, Radke CJ. 2002. Dis-

joining pressures, zeta potentials and

surface tensions of aqueous non-ionic

surfactant/electrolyte solutions: theory

and comparison to experiment. Adv. Col-loid Interface Sci. 96:231–64

20. Manciu M, Ruckenstein E. 2003. Specific

ion effect via ion hydration: I. Surface ten-

sion. Adv. Colloid Interface Sci. 105:63–

101

21. Frediani L, Cammi R, Corni S, Tomasi J.

2004. A polarizable continuum model for

molecules at diffuse interfaces. J. Chem.Phys. 120:3893–907

22. Frediani L, Mennucci B, Cammi R. 2004.

Quantum-mechanical continuum solva-

tion study of the polarizability of halides

at the water/air interface. J. Phys. Chem.B 108:13796–806

23. Garrett BC. 2004. Ions at the air/water in-

terface. Science 303:1146–47

24. Hu JH, Shi Q, Davidovits P, Worsnop

DR, Zahniser MS, Kolb CE. 1995. Reac-

tive uptake of Cl2(g) and Br2(g) by aque-

ous surfaces as a function of Br− and I−

ion concentration: The effect of chemical

reaction at the interface. J. Phys. Chem.99:8768–76

25. Knipping EM, Lakin MJ, Foster KL,

Jungwirth P, Tobias DJ, et al. 2000. Exper-

iments and simulations of ion-enhanced

interfacial chemistry on aqueous NaCl

aerosols. Science 288:301–6

26. Laskin A, Gaspar DJ, Wang WH, Hunt

SW, Cowin JP, et al. 2003. Reactions at

interfaces as a source of sulfate formation

in sea-salt particles. Science 301:340–44

27. Hunt SW, Roeselova M, Wang W, Wingen

LM, Knipping EM, et al. 2004. Formation

of molecular bromide from the reaction of

ozone with deliquesced NaBr aerosol: Ev-

idence for interfacial chemistry. J. Phys.Chem. A 108:11559–72

28. Jungwirth P, Tobias DJ. 2001. Molecular

structure of salt solutions: A new view of

the interface with implications for hetero-

geneous atmospheric chemistry. J. Phys.Chem. B 105:10468–72

29. Dang LX, Chang TM. 2002. Molecu-

lar mechanism of ion binding to the liq-

uid/vapor interface of water. J. Phys.Chem. B 106:235–38

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



30. Dang LX. 2002. Computational study of

ion binding to the liquid interface of water.

J. Phys. Chem. B 106:10388–94

31. Jungwirth P, Tobias DJ. 2002. Ions at

the air/water interface. J. Phys. Chem. B106:6361–73

32. Salvador P, Curtis JE, Tobias DJ, Jung-

wirth P. 2003. Polarizability of the nitrate

anion and its solvation at the air/water

interface. Phys. Chem. Chem. Phys. 5:

3752–57

33. Yang X, Kiran B, Wang XB, Wang LS,

Mucha M, Jungwirth P. 2004. Solva-

tion of the azide anion (N−3 ) in wa-

ter clusters and aqueous interfaces: A

combined investigation by photoelectron

spectroscopy, density functional calcula-

tions, and molecular dynamics simula-

tions. J. Phys. Chem. A 108:7820–26

34. Vrbka L, Mucha M, Minofar B, Jungwirth

P, Brown EC, Tobias DJ. 2004. Propen-

sity of soft ions for the air/water interface.

Curr. Opin. Colloid Interface Sci. 9:67–

73

35. Mucha M, Frigato T, Levering LM, Allen

HC, Tobias DJ, et al. 2005. Unified molec-

ular picture of the surfaces of aqueous

acid, base, and salt solutions. J. Phys.Chem. B 109:7617–23

36. Petersen PB, Saykally RJ, Mucha M,

Jungwirth P. 2005. Enhanced concen-

tration of polarizable anions at the liq-

uid water surface: SHG spectroscopy and

MD simulations of sodium thiocyanate. J.Phys. Chem. B 109:10915–21

37. Raduge C, Pflumio V, Shen YR. 1997.

Surface vibrational spectroscopy of sul-

furic acid-water mixtures at the liquid–

vapor interface. Chem. Phys. Lett. 274:

140–44

38. Baldelli S, Schnitzer C, Shultz MJ, Camp-

bell DJ. 1997. Sum frequency genera-

tion investigation of water at the surface

of H2O/H2SO4 binary systems. J. Phys.Chem. B 101:10435–41

39. Baldelli S, Schnitzer C, Shultz MJ, Camp-

bell DJ. 1998. Sum frequency generation

investigation of water at the surface of

H2O/H2SO4 and H2O/Cs2SO4 binary sys-

tems. Chem. Phys. Lett. 287:143–47

40. Baldelli S, Schnitzer C, Shultz MJ. 1999.

The structure of water on HCl solutions

studied with sum frequency generation.


41. Schnitzer C, Baldelli S, Shultz MJ. 1999.

Sum frequency generation by water on su-

percooled H2SO4/H2O liquid solutions at

stratospheric temperatures. Chem. Phys.Lett. 313:416–20

42. Schnitzer C, Baldelli S, Shultz MJ. 2000.

Sum frequency generation of water on

NaCl, NaNO3, KHSO4, HCl, HNO3, and

H2SO4 aqueous solutions. J. Phys. Chem.B 104:585–90

43. Miranda PB, Shen YR. 1999. Liquid in-

terfaces: A study by sum-frequency vi-

brational spectroscopy. J. Phys. Chem. B103:3292–307

44. Shultz MJ, Schnitzer C, Simonelli D,

Baldelli S. 2000. Sum frequency gener-

ation spectroscopy of the aqueous inter-

face: ionic and soluble molecular solu-

tions. Int. Rev. Phys. Chem. 19:123–53

45. Shultz MJ, Baldelli S, Schnitzer C,

Simonelli D. 2002. Aqueous solution/air

interfaces probed with sum frequency

generation spectroscopy. J. Phys. Chem.B 106:5313–24

46. Raymond EA, Richmond GL. 2004. Prob-

ing the molecular structure and bonding

of the surface of aqueous salt solutions. J.Phys. Chem. B 108:5051–59

47. Liu D, Ma G, Levering LM, Allen HC.

2004. Vibrational spectroscopy of aque-

ous sodium halide solutions and air–liquid

interfaces: Observation of increased inter-

facial depth. J. Phys. Chem. B 108:2252–

60

48. Brown EC, Mucha M, Jungwirth P, To-

bias DJ. 2005. Structure and vibrational

spectroscopy of salt water/air interfaces:

Predictions from classical molecular dy-

namics simulations. J. Phys. Chem. B 109:

7934–40

49. Weber R, Winter B, Schmidt PM, Widdra

W, Hertel IV, et al. 2004. Photoemission

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



from aqueous alkali-metal–iodide salt so-

lutions using EUV synchrotron radiation.

J. Phys. Chem. B 108:4729–36

50. Ghosal S, Hemminger JC, Bluhm H, Mun

BS, Hebenstreit ELD, et al. 2005. Elec-

tron spectroscopy of aqueous solution in-

terfaces reveals surface enhancement of

halides. Science 307:563–66

51. Petersen PB, Johnson JC, Knutsen KP,

Saykally RJ. 2004. Direct experimental

validation of the Jones-Ray effect. Chem.Phys. Lett. 397:46–50

52. Petersen PB, Saykally RJ. 2004. Confir-

mation of enhanced anion concentration

at the liquid water surface. Chem. Phys.Lett. 397:51–55

53. Petersen PB, Saykally RJ. 2005. Adsorp-

tion of ions to the surface of dilute elec-

trolyte solutions: The Jones-Ray effect

revisited. J. Am. Chem. Soc. 127:15446–

52

54. Petersen MK, Iyengar SS, Day TJF, Voth

GA. 2004. The hydrated proton at the wa-

ter liquid/vapor interface. J. Phys. Chem.B 108:14804–6

55. Petersen PB, Saykally RJ. 2005. Evidence

for an enhanced hydronium concentration

at the liquid water surface. J. Phys. Chem.B 109:7976–80

56. Shen YR. 1984. The Principles of Nonlin-ear Optics. New York: Wiley

57. Boyd RW. 2003. Nonlinear Optics. Ams-

terdam: Elsevier. 2nd ed.

58. Corn RM, Higgins DA. 1994. Optical sec-

ond harmonic generation as a probe of sur-

face chemistry. Chem. Rev. 94:107–25

59. Eisenthal KB. 1993. Liquid interfaces.

Acc. Chem. Res. 26:636–43

60. Eisenthal KB. 1996. Liquid interfaces

probed by second-harmonic and sum-

frequency spectroscopy. Chem. Rev. 96:

1343–60

61. Gragson DE, Richmond GL. 1998. Inves-

tigations of the structure and hydrogen

bonding of water molecules at liquid sur-

faces by vibrational sum frequency spec-

troscopy. J. Phys. Chem. B 102:3847–61

62. Richmond GL. 2001. Structure and bond-

ing of molecules at aqueous surfaces.

Annu. Rev. Phys. Chem. 52:357–89

63. Richmond GL. 2002. Molecular bond-

ing and interactions at aqueous surfaces

as probed by vibrational sum frequency

spectroscopy. Chem. Rev. 102:2693–724

64. Buck M, Himmelhaus M. 2001. Vi-

brational spectroscopy of interfaces by

infrared-visible sum frequency genera-

tion. J. Vac. Sci. Technol. A 19:2717–36

65. Moad AJ, Simpson GJ. 2004. A unified

treatment of selection rules and symmetry

relations for sum-frequency and second

harmonic spectroscopies. J. Phys. Chem.B 108:3548–62

66. Moad AJ, Simpson GJ. 2005. Self-

consistent approach for simplifying the

molecular interpretation of nonlinear opti-

cal and multiphoton phenomena. J. Phys.Chem. A 109:1316–23

67. Kauranen M, Verbiest T, Persoons A.

1998. Second-order nonlinear optical sig-

natures of surface chirality. J. Mod. Opt.45:403–23

68. Hayashi M, Lin SH, Shen YR. 2004.

Applications of molecular theory of

sum-frequency generations to study

molecular chirality. J. Phys. Chem. A 108:

8058–76

69. Simpson GJ. 2004. Molecular origins

of the remarkable chiral sensitivity of

second-order nonlinear optics. Chem.Phys. Chem. 5:1301–10

70. Simpson GJ. 2001. New tools for surface

second-harmonic generation. Appl. Spec-trosc. 55:16–32

71. Simpson GJ, Westerbuhr SG, Rowlen KL.

2000. Molecular orientation and angular

distribution probed by angle-resolved ab-

sorbance and second harmonic genera-

tion. Anal. Chem. 72:887–98

72. Simpson GJ, Rowlen KL. 2000.

Orientation-insensitive methodology

for second harmonic generation. 2.

Application to adsorption isotherm and

kinetics measurements. Anal. Chem. 72:

3407–11

73. Koczorowski Z. 1997. Surface potentials

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



of liquid solutions and solvents. Bull. Pol.Acad. Sci. Chem. 45:225–42

74. Paluch M. 2000. Electric properties of free

surfaces of water and aqueous solutions.

Adv. Colloid Interface Sci. 84:27–45

75. Goh MC, Hicks JM, Kemnitz K, Pinto

GR, Heinz TF, et al. 1988. Absolute ori-

entation of water molecules at the neat

water surface. J. Phys. Chem. 92:5074–

75

76. Goh MC, Eisenthal KB. 1989. The ener-

getics of orientation at the liquid–vapor

interface of water. Chem. Phys. Lett.157:101–4

77. Luca AAT, Hebert P, Brevet PF, Girault

HH. 1995. Surface second-harmonic gen-

eration at air/solvent and solvent/solvent

interfaces. J. Chem. Soc. Faraday Trans.91:1763–68

78. Fordyce AJ, Bullock WJ, Timson AJ,

Haslam S, Spencer-Smith RD, et al. 2001.

The temperature dependence of surface

second-harmonic generation from the

air-water interface. Mol. Phys. 99:677–

87

79. Sokhan VP, Tildesley DJ. 1997. The free

surface of water: molecular orientation,

surface potential and nonlinear suscepti-

bility. Mol. Phys. 92:625–40

80. Du Q, Superfine R, Freysz E, Shen YR.

1993. Vibrational spectroscopy of water at

the vapor/water interface. Phys. Rev. Lett.70:2313–16

81. Dailey CA, Burke BJ, Simpson GJ. 2004.

The general failure of Kleinman symme-

try in practical nonlinear optical applica-

tions. Chem. Phys. Lett. 390:8–13

82. Shen YR. 1999. Surface contribution ver-

sus bulk contribution in surface nonlin-

ear optical spectroscopy. Appl. Phys. B68:295–300

83. Wei X, Hong SC, Lvovsky AI, Held H,

Shen YR. 2000. Evaluation of surface

vs bulk contributions in sum-frequency

vibrational spectroscopy using reflection

and transmission geometries. J. Phys.Chem. B 104:3349–54

84. Held H, Lvovsky AI, Wei X, Shen YR.

2002. Bulk contribution from isotropic

media in surface sum-frequency genera-

tion. Phys. Rev. B 66:205110

85. Wei X, Miranda PB, Shen YR. 2001. Sur-

face vibrational spectroscopic study of

surface melting of ice. Phys. Rev. Lett.86:1554–57

86. Brown MG, Raymond EA, Allen HC,

Scatena LF, Richmond GL. 2000. The

analysis of interference effects in the sum

frequency spectra of water interfaces. J.Phys. Chem. A 104:10220–26

87. Raymond EA, Tarbuck TL, Richmond

GL. 2002. Isotopic dilution studies of the

vapor/water interface as investigated by

vibrational sum-frequency spectroscopy.

J. Phys. Chem. B 106:2817–20

88. Raymond EA, Tarbuck TL, Brown

MG, Richmond GL. 2003. Hydrogen-

bonding interactions at the vapor/water

interface investigated by vibrational

sum-frequency spectroscopy of HOD/

H2O/D2O mixtures and molecular dy-

namics simulations. J. Phys. Chem. B 107:

546–56

89. Smith JD, Cappa CD, Wilson KR, Cohen

RC, Geissler PL, Saykally JR. 2005. Uni-

fied description of temperature-dependent

hydrogen bond rearrangements in liq-

uid water. Proc. Natl. Acad. Sci. USA.102:14171–74

90. Pohorille A, Wilson MA. 1992. Molec-

ular structure of aqueous interfaces. J.Mol. Struct. (THEOCHEM) 284:271–

98

91. Yang B, Sullivan DE, Gray CG. 1994.

Density-functional theory of the water

liquid–vapour interface: II. J. Phys.: Con-dens. Matter 6:4823–42

92. Alejandre J, Tildesley DJ, Chapela GA.

1995. Molecular dynamics simulation of

the orthobaric densities and surface ten-

sion of water. J. Chem. Phys. 102:4574–

83

93. Benjamin I. 1994. Vibrational spectrum of

water at the liquid/vapor interface. Phys.Rev. Lett. 73:2083–86

94. Benjamin I. 1996. Chemical reactions and

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



solvation at liquid interfaces: A micro-

scopic perspective. Chem. Rev. 96:1449–

76

95. Taylor RS, Dang LX, Garrett BC. 1996.

Molecular dynamics simulations of the

liquid/vapor interface of SPC/E water. J.Phys. Chem. 100:11720–25

96. Dang LX, Chang TM. 1997. Molecular

dynamics study of water clusters, liquid,

and liquid–vapor interface of water with

many-body potentials. J. Chem. Phys.106:8149–59

97. Morita A, Hynes JT. 2000. A theoreti-

cal analysis of the sum frequency genera-

tion spectrum of the water surface. Chem.Phys. 258:371–90

98. Kuo IFW, Mundy CJ. 2004. An ab initio

molecular dynamics study of the aqueous

liquid–vapor interface. Science 303:658–

60

99. Perry AP, Neipert C, Kasprzyk CR, Green

T, Space B, Moore PB. 2005. A theoreti-

cal description of the polarization depen-

dence of the sum frequency generation

spectroscopy of the water/vapor interface.

J. Chem. Phys. 123:144705

100. Morita A, Hynes JT. 2002. A theoretical

analysis of the sum frequency generation

spectrum of the water surface. II. Time-

dependent approach. J. Phys. Chem. B106:673–85

101. Morita A. 2004. Toward computation of

bulk quadrupolar signals in vibrational

sum frequency generation spectroscopy.


102. Andersson G, Krebs T, Morgner H. 2005.

Activity of surface active substances de-

termined from their surface excess. Phys.Chem. Chem. Phys. 7:136–42

103. Jones G, Ray WA. 1935. The surface

tension of solutions. J. Am. Chem. Soc.57:957–58

104. Jones G, Ray WA. 1937. The surface ten-

sion of solutions of electrolytes as a func-

tion of the concentration. I. A differential

method for measuring relative surface ten-

sion. J. Am. Chem. Soc. 59:187–98



tion of the concentration. II. J. Am. Chem.Soc. 63:288–94

106. Jones G, Ray WA. 1941. The surface

tension of solutions of electrolytes as a

function of the concentration. III. Sodium

chloride. J. Am. Chem. Soc. 63:3262–63



tion of the concentration. IV. Magnesium

sulfate. J. Am. Chem. Soc. 64:2744–45

108. Langmuir I. 1938. Repulsive forces be-

tween charged surfaces in water, and the

cause of the Jones-Ray effect. Science88:430–32

109. Langmuir I. 1938. The role of attractive

and repulsive forces in the formation of

tactoids, thixotropic gels, protein crystals,

and coacervates. J. Chem. Phys. 6:873–96

110. Jones G, Frizzell LD. 1940. A theoretical

and experimental analysis of the capillary

rise method for measuring the surface ten-

sion of solutions of electrolytes. J. Chem.Phys. 8:986–97

111. Jones G, Wood LA. 1945. The measure-

ment of potentials at the interface between

vitreous silica and solutions of potas-

sium chloride by the streaming potential

method. J. Chem. Phys. 13:106–21

112. Wood LA. 1945. The effect of thorium

chloride on the potential at the interface

between vitreous silica and solutions of

potassium chloride. J. Chem. Phys. 13:

429–39

113. Wood LA, Robinson LB. 1946. A test of

Langmuir’s interpretation of the Jones-

Ray effect. J. Chem. Phys. 14:258–62

114. Dole M, Swartout JA. 1940. A twin-ring

surface tensiometer. I. The apparent sur-

face tension of potassium chloride solu-

tions. J. Am. Chem. Soc. 62:3039–45

115. Cacace MG, Landau EM, Ramsden JJ.

1997. The Hofmeister series: salt and sol-

vent effects on interfacial phenomena. Q.Rev. Biophy. 30:241–77

116. Bostrom M, Williams DRM, Ninham

BW. 2002. Influence of Hofmeister ef-

fects on surface pH and binding of

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.



peptides to membranes. Langmuir 18:

8609–15


BW. 2004. Why the properties of pro-

teins in salt solutions follow a Hofmeister

series. Curr. Opin. Colloid Interface Sci.9:48–52

118. Kunz W, Nostro PL, Ninham BW. 2004.

The present state of affairs with Hofmeis-

ter effects. Curr. Opin. Colloid InterfaceSci. 9:1–18

119. Kunz W, Henle J, Ninham BW. 2004.

‘Zur Lehre von der Wirkung der Salze’

(about the science of the effect of

salts): Franz Hofmeister’s historical pa-

pers. Curr. Opin. Colloid Interface Sci.9:19–37

120. Ninham BW, Yaminsky V. 1997. Ion

binding and ion specificity: The Hofmeis-

ter effect and Onsager and Lifshitz theo-

ries. Langmuir 13:2097–108


BW. 2001. Surface tension of electrolytes:

specific ion effect explained by dispersion

forces. Langmuir 17:4475–78

122. Bostrom M, Ninham BW. 2004. Disper-

sion self-free energies and interaction free

energies of finite-sized ions in salt solu-

tions. Langmuir 20:7569–74

123. Kunz W, Belloni L, Bernard O, Ninham

BW. 2004. Osmotic coefficients and sur-

face tensions of aqueous electrolyte solu-

tions: Role of dispersion forces. J. Phys.Chem. B 108:2398–2404

124. Bostrom M, Kunz W, Ninham BW. 2005.

Hofmeister effects in surface tension of

aqueous electrolyte solution. Langmuir21:2619–23

125. Benjamin I. 1991. Theoretical study of ion

solvation at the water liquid–vapor inter-

face. J. Chem. Phys. 95:3698–709

126. Wilson MA, Pohorille A. 1991. Interac-

tion of monovalent ions with the water

liquid–vapor interface: A molecular dy-

namics study. J. Chem. Phys. 95:6005–13

127. Benjamin I. 1995. Theory and computer

simulations of solvation and chemical re-

actions at liquid interfaces. Acc. Chem.Res. 28:233–39

128. Paul S, Chandra A. 2003. Dynamics of

water molecules at liquid–vapour inter-

faces of aqueous ionic solutions: effects

of ion concentration. Chem. Phys. Lett.373:87–93

129. Bhatt D, Newman J, Radke CJ. 2004.

Molecular dynamics simulations of sur-

face tensions of aqueous electrolytic so-

lutions. J. Phys. Chem. B 108:9077–84

130. Bhatt D, Chee R, Newman J, Radke CJ.

2004. Molecular simulation of the surface

tension of simple aqueous electrolytes

and the Gibbs adsorption equation. Curr.Opin. Colloid Interface Sci. 9:145–48

131. Jungwirth P, Curtis JE, Tobias DJ. 2003.

Polarizability and aqueous solvation of

the sulfate dianion. Chem. Phys. Lett.367:704–10

132. Gopalakrishnan S, Jungwirth P, Tobias

DJ, Allen HC. 2005. Air–liquid inter-

faces of aqueous solutions containing am-

monium and sulfate: spectroscopic and

molecular dynamics studies. J. Phys.Chem. B 109:8861–72

133. Dang LX. 2003. Solvation of the hydro-

nium ion at the water liquid/vapor inter-

face. J. Chem. Phys. 119:6351–53

134. Faubel M, Steiner B, Toennies JP. 1997.

Photoelectron spectroscopy of liquid wa-

ter, some alcohols, and pure nonane in free

micro jets. J. Chem. Phys. 106:9013–31

135. Winter B, Weber R, Widdra W, Dittmar

M, Faubel M, Hertel IV. 2004. Full va-

lence band photoemission from liquid wa-

ter using EUV synchrotron radiation. J.Phys. Chem. A 108:2625–32

136. Petersen PB, Saykally RJ. 2006. Using

femtosecond second harmonic generation

to probe the interfacial structure of aque-

ous electrolyte solutions. J. Phys. Chem.B. Submitted

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.

P1: JRX/KUV

February 23, 2006 1:40 Annual Reviews AR272-FM

Annual Review of Physical ChemistryVolume 57, 2006

CONTENTS

REFLECTIONS ON PHYSICAL CHEMISTRY: SCIENCE AND SCIENTISTS,

Joshua Jortner 1

ON A RESEARCH ROLLERCOASTER WITH FRIENDS,

Robin M. Hochstrasser 37

4D ULTRAFAST ELECTRON DIFFRACTION, CRYSTALLOGRAPHY,

AND MICROSCOPY, Ahmed H. Zewail 65

HETEROGENEOUS CHEMISTRY OF CARBON AEROSOLS,

Amanda M. Nienow and Jeffrey T. Roberts 105

PROGRESS IN THE THEORY OF MIXED QUANTUM-CLASSICAL

DYNAMICS, Raymond Kapral 129

STARK DECELERATION AND TRAPPING OF OH RADICALS,

Sebastiaan Y.T. van de Meerakker, Nicolas Vanhaecke,and Gerard Meijer 159

ATMOSPHERIC FIELD MEASUREMENTS OF THE HYDROXYL RADICAL

USING LASER-INDUCED FLUORESCENCE SPECTROSCOPY,

Dwayne E. Heard 191

EXCITONS IN CONJUGATED OLIGOMER AGGREGATES, FILMS,

AND CRYSTALS, Frank C. Spano 217

LASER PROBING OF SINGLE-AEROSOL DROPLET DYNAMICS,

Jonathan P. Reid and Laura Mitchem 245

CONNECTING CHEMICAL DYNAMICS IN GASES AND LIQUIDS,

Christopher G. Elles and F. Fleming Crim 273

NEAR-FIELD OPTICAL MICROSCOPY AND SPECTROSCOPY WITH

POINTED PROBES, Lukas Novotny and Stephan J. Stranick 303

ON THE NATURE OF IONS AT THE LIQUID WATER SURFACE,

Poul B. Petersen and Richard J. Saykally 333

CORRELATED ELECTRONIC STRUCTURE NONLINEAR RESPONSE

METHODS FOR STRUCTURED ENVIRONMENTS, Kurt V. Mikkelsen 365

COHERENT EXCITATION OF VIBRATIONAL MODES IN METALLIC

NANOPARTICLES, Gregory V. Hartland 403

ix

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.

P1: JRX/KUV

February 23, 2006 1:40 Annual Reviews AR272-FM

x CONTENTS

ION PAIR DISSOCIATION: SPECTROSCOPY AND DYNAMICS,

Arthur G. Suits and John W. Hepburn 431

REACTIVITY OF THE GERMANIUM SURFACE: CHEMICAL PASSIVATION

AND FUNCTIONALIZATION, Paul W. Loscutoff and Stacey F. Bent 467

SCANNING TUNNELING MICROSCOPY MANIPULATION OF COMPLEX

ORGANIC MOLECULES ON SOLID SURFACES, Roberto Otero,Federico Rosei, and Flemming Besenbacher 497

RAMAN CRYSTALLOGRAPHY AND OTHER BIOCHEMICAL

APPLICATIONS OF RAMAN MICROSCOPY, Paul R. Carey 527

FEMTOSECOND TIME-RESOLVED PHOTOELECTRON IMAGING,

Toshinori Suzuki 555

SINGLE-MOLECULE ELECTRICAL JUNCTIONS, Yoram Selzerand David L. Allara 593

DYNAMICAL STUDIES OF THE OZONE ISOTOPE EFFECT: A STATUS

REPORT, R. Schinke, S. Yu. Grebenshchikov, M.V. Ivanov,and P. Fleurat-Lessard 625

INDEXES

Subject Index 663

Cumulative Index of Contributing Authors, Volumes 53–57 683

Cumulative Index of Chapter Titles, Volumes 53–57 685

ERRATA

An online log of corrections to Annual Review of Physical Chemistrychapters may be found at http://physchem.annualreviews.org/errata.shtml

Ann

u. R

ev. P

hys.

Che

m. 2

006.

57:3

33-3

64. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Cal

ifor

nia

- B

erke

ley

on 0

5/04

/06.

For

per

sona

l use

onl

y.

Poul B. Petersen and Richard J. Saykally1 Mar 2006 15:28 AR ANRV272-PC57-12.tex...

Documents

Transcript of Poul B. Petersen and Richard J. Saykally1 Mar 2006 15:28 AR ANRV272-PC57-12.tex...