Systematic evaluation of acetone and acetonitrile for use in hydrophilic interaction liquid...

Research Article

Received: 19 July 2011 Revised: 20 September 2011 Accepted: 21 September 2011 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2011, 25, 3666–3674

3666

Systematic evaluation of acetone and acetonitrile for use inhydrophilic interaction liquid chromatography coupled withelectrospray ionization mass spectrometry of basicsmall molecules

James Heaton1, Michael D. Jones1,2, Cristina Legido-Quigley1, Robert S. Plumb1,2 andNorman W. Smith1*1Pharmaceutical Science Division, School of Biomedical and Health Sciences, King’s College London, 150 Stamford Street,London, SE1 9NH, UK2Waters Corporation, 34 Maple Street, Milford, MA 01757, USA

Sub-2-mm particle size hydrophilic interaction liquid chromatography [HILIC] combined with mass spectrometryhas been increasing in popularity as a complementary technique to reversed-phase LC for the analysis of polar ana-lytes. The organic-rich mobile phase associated with HILIC techniques provides increases in compound ionization,due to increased desolvation efficiency during electrospray ionisation mass spectrometric (ESI-MS) analysis.Although recent publications illustrated selectivity and response comparisons between reversed-phase LC/MS andHILIC LC/MS, there are limited discussions evaluating the optimisation of the mass spectrometry parameters regard-ing analytes and alternative mobile phases. The use of acetone as an alternative organic modifier in HILIC has beeninvestigated with respect to signal-to-noise in ESI-MS for a variety of polar analytes. Analyte reponses were mea-sured based on a variety of cone and capillary voltages at low and high pH in both acetone and acetonitrile. In orderto visualise compound behaviour in the ESI source, surface plots were constructed to assist in interpreting theobserved results. The use of acetone in ESI is complicated at low m/z due to the formation of condensation products.Favourable responses were observed for certain analytes and we envisage offering an insight into the use of acetoneas an alternative to acetonitrile under certain analytical conditions for particular compound classifications for smallmolecule analysis. We also highlight the importance of optimising source voltages in order to obtain the maximumsignal stability and sensitivity, which are invariably, highly solvent composition dependent parameters. Copyright ©2011 John Wiley & Sons, Ltd.

(wileyonlinelibrary.com) DOI: 10.1002/rcm.5271

Reversed-phase chromatography is, in many cases, anunsuitable mode of retention for polar, hydrophilic com-pounds and notably undesirable peak shapes are observedwith highly basic solutes under acidic conditions makingpeak integration more difficult whilst leading to low columnefficiencies. Furthermore, this mode of chromatography maynot be suitable in certain circumstances, in particular whenhyphenation to mass spectrometry is required. For instance,the use of triethylamine competing base[1] in the mobile phaseor an ion-pairing[2] approach may invariably render thehyphenation of separations to mass spectrometry redundantor undesirable due to ion suppression. For analytical scaleliquid chromatographic separations of hydrophilic polarcompounds few alternatives arise, where normal-phase isnot considered to be environmentally friendly due to thetoxicity of the eluents used and at present supercritical fluidchromatography (SFC) is not widely accessible.

* Correspondence to: N. W. Smith, Pharmaceutical ScienceDivision, School of Biomedical and Health Sciences, King’sCollege London, 150 Stamford Street, London SE1 9NH, UK.E-mail: [email protected]

Rapid Commun. Mass Spectrom. 2011, 25, 3666–3674

The term hydrophilic interaction chromatography (HILIC)was first proposed by Alpert et al.[3] for the separation ofpolar biological solutes, and was followed up with furtherwork on the separation of complex carbohydrates[4] and thedevelopment of an assay for tyrosine protein kinase activity[5]

in the early 1990s. Retention was initially thought to be attrib-uted to the partitioning of a solute between an immobilisedlayer of water on the stationary phase and the bulk organic-rich mobile phase; however, recent studies suggest a morecomplex scenario is taking place comprising of multiplesolute-stationary phase interactions. As proposed by Peseket al.[6] and McCalley,[7] HILIC should arguably be termedaqueous normal-phase chromatography (ANP) due to thepolar nature of the stationary phase and the relative lowpolarity of the corresponding eluent. The utility of aqueous-organic mobile phases on bare silica was exploited in an ear-lier quantitative LC/MS/MS application for the separation ofmorphine and its related glucuronides[8] and was shown tobe a robust and sensitive approach. A review by Hemströmand Irgum[9] has summarised many of these interactions con-cluding that coulombic, hydrogen-bonding and liquid-liquidpartitioning are all taking place to different degrees. Asillustrated recently by McCalley,[10] the HILIC retention

Copyright © 2011 John Wiley & Sons, Ltd.

Evaluation of acetone and acetonitrile for use in HILIC/ESI-MS

366

mechanism is highly dependent on the presence of a bondedgroup on the stationary phase surface as the introduction ofpolar groups results in dramatic changes in selectivity formethods development.The other benefit of HILIC is the by-product of using

organic-rich mobile phases, in that eluent desolvation isachievedmore efficiently due to the lower viscosity and surfacetension properties enhancing sensitivity for electrospray ionisa-tion mass spectrometry (ESI-MS).[11–13] Many of the reporteduses of HILIC are accompanied significantly by hyphenationto mass spectrometry in comparison to ultraviolet (UV) detec-tion and/or other detection systems such as ELSD or CAD.Obviously, the use of acetone in analytical separations isrestricted by the strong adsorption of the carbonyl group whenutilising UV detection at wavelengths <330 nm; however, fewevaluations have been carried out regarding this polar aproticsolvent as an alternative to acetonitrile when using amass spec-trometer as a means of detection for small molecule analysis.Therefore, the use of acetone as the weak organic modifier forHILIC separations requires further evaluation, being that it isan aprotic polar solvent with similar properties to acetonitrile;it is necessary that the immobilised water layer on the surfaceof the stationary is not disrupted by protic solvents such asmethanol for the ‘pure HILIC’ contribution of the retentionmechanism to become established. We define the term ‘pureHILIC’ in the sense that an immobilised layer of water onthe surface of the stationary is present which is otherwisenot part of an extensively hydrogen-bonded solvent system,i.e. water/methanol mixtures.There have already been reports on the use of acetone for

reversed-phase LC/MS applications. The first such investiga-tion being conducted by Fritz and co-workers[14] for peptideanalysis by directly comparing acetone, acetonitrile andmethanol. They reported sharper peaks, most likely due toenhanced diffusivity as a function of reduced viscosity inusing acetone, shorter run times and overall superior separa-tions with this organic modifier. More recently, Keppel andco-workers[15] further evaluated acetone for RPLC/MS for asimilar application involving peptide analysis. The authorsdiscussed chemical noise and ionisation efficiency as well ascolumn performance features of acetone, concluding in somecases this solvent to be inferior to acetonitrile. Duderstadt andFischer[16] investigated the utility of acetone as a post-columnadditive for the analysis of polyalkenes using APCI and APPIfor promoting gas-phase proton transfer as a means of proto-nation using photoionisation. Their findings suggested thatthe use of acetone as an ionisation dopant showed limited sig-nal enhancements to only a few of their evaluated analytes.Fountain et al.[17] discussed the use of acetone with HILIC/MS noting poor signal intensities for basic compounds incomparison to acetonitrile suggesting the presence of solventcondensation products and adduct formation. The aim of thework presented here was to investigate further the discrepan-cies in signal intensity observed previously when comparingacetone and acetonitrile for HILIC/ESI-MS by comprehensiveoptimisation of source voltages in order to maximise thesignal-to-noise values for a set of hydrophilic and basic ana-lytes. The two parameters deemed most important for sys-tematic adjustment being the electrospray capillary and conevoltages once the appropriate desolvation temperature andgas flows had been optimised to accommodate the interfacedchromatography. We highlight the importance in applying

Copyright © 2011Rapid Commun. Mass Spectrom. 2011, 25, 3666–3674

the optimal source voltages for achieving both the moststable, reproducible electrospray conditions and also inachieving the highest analyte signal-to-noise ratio. The pur-pose of our work herein is to emphasise on the dependenceof ESI source conditions in order to achieve optimum sensi-tivity and stability as a function of solvent type and composi-tion for typical pharmaceutically relevant basic solutes. Wealso feel HILIC is limited in solvent selection due to its depen-dence on aprotic solvents in comparison to its reversed-phasecounterpart when using bare silica stationary phases. Weshow that acetone is perhaps an unfavourable alternative toacetonitrile in the context of hydrophilic and basic soluteanalysis using HILIC with ESI-MS.

EXPERIMENTAL

Chemicals

The organic solvents acetonitrile (CH3CN) and 2-propanol(IPA) were optima grade whilst acetone ((CH3)2CO) wasHPLC grade and were supplied by ThermoFisher Scientific(Fair Lawn, NJ, USA). Water (18.2 MΩcm, 5 ppb TOC) waspurified using a Milli-Q Advantage A10. The followinganalyte test probes were purchased from Sigma-Aldrich(St. Louis, MO, USA): cytosine (99%), adenine (99%), caffeine(meets USP specifications, anhydrous), procainamide hydro-chloride, nicotine (hydrogen tartrate salt), nortriptylinehydrochloride (>98%), and diphenhydramine hydrochloride(analytical standard). The physicochemical properties of thetest probes and the solvents are given in Tables 1 and 2,respectively. Buffer reagents ammonium acetate (puriss),ammonium formate (puriss), ammonium hydroxide (>25%solution inwater), and formic acid (eluent additive for LC/MS)were supplied by Fluka (distributed by Sigma-Aldrich,Allentown, PA, USA).

Preparation of solutions

The aqueous mobile phase component consisted of either100 mM ammonium acetate or 100 mM ammonium formateprepared in H2O and adjusted to either pH 9 or 3 withammonium hydroxide or formic acid, respectively. The stocksolutions were blended via the on-line fluidics to achieve90:10 v/v organicmodifier/100mMbuffer in order tomaintaina constant ionic strength of 10 mM buffer during theexperimental process. The needle wash solvent and seal washsolvent were prepared with a ratio 80:20 of 2-propanol andwater, respectively.

Sample preparations

The probe solutes adenine, cytosine, caffeine, procainamide,nicotine, nortriptyline and diphenhydramine were preparedat a stock concentration of 1 mg/mL in 75:25 CH3CN/MeOH.Dilutions were performed to yield working standard concen-trations of 1 mg/mL in 95:5 CH3CN/H2O for injection into theLC/MS instrument.

Chromatographic conditions

Separations were performed on a Waters ACQUITY H-Classultra-performance liquid chromatography (UPLC) system(Waters Corp., MA, USA) with a quaternary solvent deliverysystem. The ACQUITY H-Class system was modified by

wileyonlinelibrary.com/journal/rcmJohn Wiley & Sons, Ltd.

7

Table 2. Solvent physiochemical properties modified fromCharles et al.[31]

H2O CH3CN (CH3)2CO

e 80.4 36.1 20.1Z (105 .poises) 1004 383 327Vapor pressure (Torr) at 25�C 24 90 203Proton affinity (kcal/mol) 165 186.2 194

J. Heaton et al.

3668

installing the hexane/THF compatibility kit. All data was col-lected using the Empower 2 chromatography data system(Waters Corp., MA, USA). The chromatography was per-formed on a Waters BEH (ethylene bridged hybrid) bare silicaHILIC column (2.1 mm� 100 mm, 1.7 mm) operated at a flowrate of 800 mL/min and the column temperature maintainedat 30 �C for all analysis. No post-column splitting of eluentswas carried out with direct interfacing of the column to theESI source. All analyte solutions were injected separately toavoid a competing ion effect of co-eluted species.

Mass spectrometry conditions

In order to investigate the effect of ESI conditions on the sig-nal-to-noise (s/n) ratio of the selected test probes under dif-ferent eluotropic conditions, capillary voltages were variedfrom 0.5 to 3.5 kV in 0.5 kV steps. At each capillary voltagea concomitant cone voltage (declustering potential on certaininstruments) from 10 to 30 eV was applied and increasedstepwise by 5 eV. The mass spectrometer used for this workwas a Waters SQD detector (Waters Corp., Manchester, UK)equipped with an ESI source operated in positive ion mode.The respective cone and desolvation temperatures were150 �C and 550 �C. The cone and desolvation gas flows were10 and 900 L/min, respectively. Source temperatures andflow rates were maintained during all experiments unlessotherwise stated. Mass spectral data was determined andextracted using thr Empower 2 chromatography data system(Waters Corp., MA, USA). The mass range investigated was100–600 Da operated at a scan speed of 5000 Da per second.Noise was determined using the average peak-to-peak pro-cessing algorithm within Empower 2. Measurements weremade over 1 min using a segment width of 20 s averagedfrom the EIC at times when the analyte of interest was noteluting. Data treatment was performed using Microsoft Excel2007 and SimcaP (Umea, Sweden).

RESULTS AND DISCUSSION

Optimisation of source parameters

To investigate the use of acetone for HILIC/MS of basic com-pounds experiments were performed whereby a responsesurface of s/n was constructed for each of the probe solutesin order to determine the optimum source conditions. Thepositive ion ESI conditions were optimised as a function ofcapillary and cone voltages in full scan mode and the

Table 1. Physiochemical properties of hydrophilic testprobes[30]

CompoundMW

(g mol�1) pKa logP

Adenine 135.1 4.2, 9.9 -Cytosine 111.1 4.2 -Caffeine 194.2 14 -0.07Procainamide 271.8 9.2 0.9Nicotine 162.3 3.2, 7.9 1.2Nortriptyline 263.4 9.7 1.7Diphenhydramine 255.1 9 3.3

wileyonlinelibrary.com/journal/rcm Copyright © 2011 John Wile

extracted ion was smoothed with a 9 point mean filter andbaseline corrected to filter noise ions for each analyte and thisdata was used for the construction of the response surfaces.Figure 1 shows the influence of varying cone and capillaryvoltages for two different analytes diphenhydramine, ahydrophobic (logP 3.3) tertiary amine (Fig. 1(a)), and thehydrophilic analyte cytosine (Fig. 1(b)). For diphenhydra-mine, high capillary and cone voltages were required foreffective protonation under the applied eluotropic conditions((CH3)2CO, low pH) with almost no signal detected at lowervoltage potentials. Cytosine showed different source behav-iour in comparison and could be successfully protonated atlower capillary voltages; however, higher cone voltages wererequired for achieving optimal s/n values. Optimum sourceconditions were found to be similar for the other nucleobaseadenine (Table 3). The difficulties in protonating nucleobaseswith a pKa>3 through gas-phase proton transfer reactionswithammonium cations in electrospray were evaluated by Charlesand co-workers.[18] In the context of this work, adenine andcytosine have similar pKa values of 4.2 and that any contribu-tions to protonation via gas-phase ammonium cations isperhaps less likely. Furthermore, the presence of acetone(194 kcal/mol) with a higher proton affinity than acetonitrile(186.2 kcal/mol) suggests it may be competingmore effectivelyfor protonating hydrogen atoms, hence the poor s/n observedfor the nucleobases and indeed also for caffeine in this study.

Adenine and cytosine have low pKa values (Table 1) andare considered non-ionised under the conditions evaluatedherein. In comparison to the other evaluated basic com-pounds, the removal of aqueous solvation in the desolvationprocess make compounds such as nucleobases more difficultto protonate. Caffeine is very weakly basic and overall showsdifficulty in achieving effective protonation apart from whenusing acetonitrile at high pH, as shown in Tables 4 and 5. Thisobservation correlates well with Zhou and Cook[19] in thatsolution-phase protonation is unlikely and instead either agas-phase ion-molecule interaction or collision-induced dis-sociation of the ammonium cations and the analyte is takingplace. Clearly, without any obvious basic centre for proton-ation adenine, cytosine and caffeine illustrate the complexionisation process involved with improving the response ofthese effectively neutral analytes, and that serious considera-tion in attenuating the interplay between eluent and sourceconditions is essential for improving sensitivity.

Table 3 indicates the optimally determined source voltages foreach analyte based on the systematic variation of capillary andcone voltages. With the exception of the nucleobases and caf-feine, the same optimum capillary voltages using acetone atpH 3 were determined for all other probe compounds. In con-trast, higher voltages were required when using acetonitrile forall analytes under acidic conditions. Interestingly, in several

y & Sons, Ltd. Rapid Commun. Mass Spectrom. 2011, 25, 3666–3674

10

20

30

0

100

200

300

400

500

600

700

0.5 1.0 1.5 2.0 2.5 3.0 3.5

s/n

X: Capillary Voltage (kV) Z: Cone Voltage (eV)

600-700

500-600

400-500

300-400

200-300

100-200

0-100

10

15

2025

300

100200

300

400

500

600

700

0.51.0

1.52.0

2.53.0

3.5

s/n

X: Capillary Voltage (kV) Z: Cone Voltage (eV)

600-700

500-600

400-500

300-400

200-300

100-200

0-100

(b)(a)

Figure 1. Signal-to-noise surface plots of cone vs. capillary voltage for (a) diphenhydramine determined with 90:10(CH3)2CO/10 mM ammonium formate (pH 3.0) and (b) cytosine determined with 90:10 CH3CN/10 mM ammoniumformate (pH 9.0). The x-, y- and z-axes represent capillary voltage (kV), cone voltage (eV) and signal-to-noise, respectively.


instances optimal s/n was obtained for some of the analytes atlow capillary voltages suggesting that perhaps thermallyassisted ionisation is the dominant mechanism. The earlier workof Vestal and co-workers[20,21] on thermospray ionisation notedthat under certain conditions molecular ions were producedwhen the hot filament used to ionise compounds emerging fromthe vapour jet was turned off, reporting substantially lowerdetection limits as a result. We observed previously[22] low opti-mal capillary voltages (0.8 kV) for the UPLC/MS/MS analysisof the basic solutes quinine and its major metabolite 3(S)-3-hydroxyquinine, using an instrument with a similar sourcedesign to the one used in the present study. This observationmay also be related to the work of Langley and co-workers[23]

noting increased signal in the absence of high voltage usingSFC/MS, termed novo-spray. However, in their study the defi-nitive ionisationmechanism responsible for affording increasedsignal was not elucidated. Nevertheless, the precision valueswe obtained at these lower voltages were remarkably stableand not indicative of thermospray alone, which is typically

Table 3. Optimal source voltages determined for each compouion data. Data based on extracted [M+H] ions for each compou

(CH3)2CO

CompoundCapillary voltage

(kV)Cone

pH 3.0Cytosine 3.5Adenine 3.5Caffeine 2.0Procainamide 0.5Nicotine 0.5Nortriptyline 0.5Diphenhydramine 0.5

pH 9.0Cytosine 0.5Adenine 2.0Caffeine n.d.Procainamide 0.5Nicotine 1.0Nortriptyline 0.5Diphenhydramine 1.0


not as reproducible as ESI. It is likely then that the dominantionisation mechanism is thermospray, with protonation occur-ring mainly through abstraction of protons from ammoniumcations in the gas phase. The data reported here may thereforebe termed electrically stabilised thermospray. Clearly, thesefindings warrant further investigation elsewhere, particularlyas most modern ESI mass spectrometers are now able tooperate with high desolvation temperatures and gas flow rates,just as we decided to use in order to accommodate the higherchromatographic inlet flows in this study.

At high pH using acetone, the overall s/n values were muchlower and suppression of the responses obtained for thestrongly basic solutes was impaired dramatically comparedwith those determined using the same solvent under acidic con-ditions. This can perhaps be related to the reactivity of acetonewhich exists as an enolate in aqueous basic conditions whichcan then result in the formation of analyte adducts and or diace-tone, triacetone and higher condensation products, as observedby Keppel et al.[15] and Fountain et al.[17] It is also known that in

nd based on surface response plots determined by extractednd

CH3CN

voltage(eV)

Capillary voltage(kV)

Cone voltage(eV)

30 3.5 3025 3.5 3010 2.0 1525 2.0 2510 2.5 2025 2.5 2520 3.5 20

20 0.5 2530 0.5 25n.d. 3.0 2030 1.5 1510 3.0 2530 2.0 3020 3.0 10


3669

Table 4. Comparison between acetonitrile and acetone at pH 3.0 determined at optimum source voltages (n = 10)

Compound(CH3)2CO (s/n)

Mean

% R.S.D CH3CN (s/n)Mean

% R.S.D

s/n Signal Intensity Peak Area s/n Signal Intensity Peak Area

Cytosine 34 20.5 5.6 6.1 82 31.0 11.5 11.7Adenine 33 13.7 5.9 7.5 174 42.8 12.4 11.3Caffeine n.d. n.d. n.d. n.d. 9 33.3 20.2 16.9Procainamide 483 13.1 1.5 4.2 664 13.5 2.5 3.0Nicotine 125 12.1 3.9 4.0 342 19.8 4.8 3.7Nortriptyline 293 12.5 1.9 1.9 686 13.1 3.1 3.0Diphenhydramine 1386 14.3 1.2 1.2 2242 23.3 1.2 1.4

Table 5. Comparison between acetonitrile and acetone at pH 9.0 determined at optimum source voltages (n = 10)

Compound(CH3)2CO (s/n)

Mean

% R.S.D CH3CN (s/n)Mean

% R.S.D

s/n Signal Intensity Peak Area s/n Signal Intensity Peak Area

Cytosine 217 9.0 4.4 5.3 602 11.2 4.6 5.7Adenine 77 19.8 15.4 15.2 613 14.6 5.7 7.9Caffeine n.d. n.d. n.d. n.d. 62 58.7 13.7 16.1Procainamide 325 17.9 3.9 3.7 673 20.7 5.7 4.2Nicotine 75 11.7 7.9 6.7 373 23.7 16.3 16.5Nortriptyline 379 20.3 2.9 3.1 1582 31.3 12.3 12.3Diphenhydramine 303 20.5 4.3 4.4 1920 16.7 2.1 2.4

J. Heaton et al.

3670

solution chemistry tertiary amines do not react with carbonylgroups suggesting that only weakly dissociative analyte-solventinteractions are taking place. Figure 2 shows the presence ofhigh intensity diacetone and triacetone condensation products

(a)

(b)

(c)

Diacetone

Cytosine [M+H]

0

2x107

4x107

6x107

0.0

5.0x106

1.0x107

1.5x107

0

2x107

4x107

6x107

100.00 110.00 120.00 130.00 140.00 150.00 1

Inte

nsity

Inte

nsity

Inte

nsity

Figure 2. (a) Combined overlay of acetone backgroundover 2minCDSundergoingbaselinecorrection;and(c)cytosinespectracombin


in the extracted ion spectra from an injection of cytosine. The sig-nal intensity for cytosine is in the order of 1.5� 107 whereas thecorresponding signals for diacetone and triacetone are 2� 107

and 6.5� 107, respectively. The observed adduct formation can

m/z60.00 170.00 180.00 190.00 200.00 210.00 220.00

Cytosine [M+H+(CH3)2CO]

Triacetone

Cytosine: combined, no subtraction

Cytosine: integrated peak (baseline corrected)

Baseline 2-4 minutes combined, no subtraction

; (b) cytosine spectra as result of peak integration in Empower 2edat thebaselineofthepeakwithoutbaselinenoisesubtraction.


Figure 3. Diacetone ion at 116 m/z monitored using various cone gas flow rates with amobile phase at pH 9.0 (200 L/min was tested twice at random to verify observations).

0.0E+00

1.0E+07

2.0E+07

3.0E+07

4.0E+07

5.0E+07

6.0E+07

7.0E+07

Sig

nal

Inte

nsi

ty

0.0E+00

1.0E+07

2.0E+07

3.0E+07

4.0E+07

5.0E+07

6.0E+07

7.0E+07

Sig

nal

Inte

nsi

ty

(a)

(b)

Figure 4. Mean signal intensity data based on n= 10 injec-tions: (a) at pH 3 and (b) at pH 9. Blue and red are acetonitrileand acetone, respectively.


367

be related to the work of O’Hair and Reid[24] on glycine residuereactivity towards acetone in the gas phase; however, wedid not see any ions of +40 m/z which would be consistentwith the [M+H+ (CH3)2CO –H2O]+ product.In an attempt to improve the undesirable s/n values

observed when using acetone, the effect of increasing conegas flow (Fig. 3) was evaluated as depicted by monitoringthe 116 m/z ion (diacetone). By applying values up to 200 L/minand at intermediate flow rates ranging from zero cone gas,minor signal decreases were observed on the disruption of theacetone condensation products. A slight decrease in the signalfor diacetone was observed compared to the response foundwithout application of cone gas; however, this was not appre-ciably effective in reducing the contribution to noise whenusing this solvent. Disruption of the acetone-analyte interactionwas not achievable usinghigher conevoltages or increases in conegas (Fig. 3) to improve s/n values. This is not surprising sincediacetone and triacetone are condensationproducts of acetone andincreased gas flows would not disrupt these covalent ions. Thepurpose of increasing cone gas flow was to try and inhibit theentrance of these acetone product ions into the ion source in orderto improve s/n values using cytosine as a model test compound.From the determined optimum source voltages, analyte

reproducibility was determined for s/n, signal intensity andpeak area, as shown in Tables 4 and 5 for each of the solventsystems. Using these conditions ionisation efficiency shouldbe optimal as indicated by the precision determined for peakareas measured for ten consecutive injections of each testsolute; overall <20% relative standard deviation (RSD) foreach analyte determined in all solvent systems. Notably,much more stable measurements for the acetone experimentsat both pH 3 and pH 9 in comparison to acetonitrile for peakarea (<10%, (CH3)2CO) were observed. Peak area reproduc-ibility for adenine (15.2%) was poor using acetone at pH 9and is likely to be attributed to the difficulty in protonationof this compound. On the contrary, under the same conditionsnicotine showed lower s/n except the peak area precision(6.7%) was superior. Overall, peak area and s/n precision wasfound to be superior for acetone versus acetonitrile under bothpH conditions investigated. This may be a consequence ofsolvent cleanliness as fewer extraneous ions were observed


when using acetone, although overall higher background chemi-cal noisewas observeddue to the previouslydiscussed condensa-tion products. Figure 4 shows the signal intensity values for eachof the evaluated solvent systems at low and high pH, the % RSDvalues for ten replicate injections are shown in Tables 4 and 5.For practical purposes it is of more interest to report s/n valuessince this provides a more quantitative measure for determin-ing the relative sensitivity, or limit of detection, for each analyte.


1

J. Heaton et al.

3672

The vapour pressure for acetone (Table 2) is higher thanacetonitrile and this should facilitate more efficient desolva-tion due to reduced surface tension of the droplets. However,the presence of acetone condensation products and adductformation results in inferior use of this solvent for positiveion ESI in the lower mass range (<300 amu). For analyteswith the lowest s/n values (i.e. adenine, cytosine andcaffeine) observed in the full scan extracted ion experiments,a comparison was made with selected ion monitoring (SIM)so as to ascertain the extent of the background noise,acetone-related chemical noise and adduct formation. Herethe quadrupole was set to each ion separately, representingthe protonated molecule of interest (e.g. 136.1 [M+H] foradenine) and, as a result, interference was somewhat filteredout resulting in increased observed s/n values. The resultsare shown in Fig. 5 from ten injections of each analyte underthe stated chromatographic conditions. This observation is inagreement with any selected ion mass spectrometry and indi-cates that when using acetone as an eluent for HILIC major

0 2000 4000 6000 8000

Cytosine

Adenine

Caffeine

s/n

(a)

(b)

Figure 5. Comparison between (a) (CH3)2CO and (b) CH3CNat pH 9.0 using SIM for test solutes (n = 10).

-3

-2

-1

0

1

2

3

t[1]

CH3CN, pH 3.0 (CH3)2CO, pH 3.0 CH3CN, pH 9.0 (CH3)2CO, pH 9.0

CytosineAdenine

Caffeine

Procainamide

Nicotine

Nortriptyline

Diphenhydramine

2 SD

2 SD

3 SD

3 SD

1

Figure 6. Scatter plot encompassing mean signal-to-noise values for overall sample set.


improvements in s/n can be achieved in comparison to whendata is acquired in a full scan mode. There is an improvementwhen using acetone under these conditions; however, the useof acetonitrile is far superior in comparison and should be thepreferred solvent when maximum sensitivity is required.

Summarised statistical data treatment of the source voltageoptimisation process and the resulting optimum values foreach solvent at low and high pH are shown in Fig. 6. Theanalytes are grouped into sets of conditions and the mostfavourable solvent system is highlighted as using acetonitrileat high pH determined as the highest s/n value. This is con-sistent with observations by Peng and Farkas[25] and Delatourand Leclercq[26] on the uses of basic mobile phases with posi-tive ion ESI where higher sensitivity was reported in com-parison to acidic conditions. The optimal s/n for adenineand cytosine was also found using acetonitrile at highpH. The least favourable data was the pH 9.0 buffer withacetone, also highlighted, yielding the overall poorestresults for all analytes.

Influence of acetone organic modifier on retention

As already indicated by several workers[27–29] underivatisedsilica has been shown to be beneficial for the analysis of cat-ionic drugs by ion-exchange retention mechanisms. In termsof HILIC, there is a limited range of solvents that can be incor-porated into the methods development process compared toreversed-phase LC. Spiraling costs due to shortages of aceto-nitrile supplies can also be troublesome as recent history hashighlighted. Consequently, running costs for HILIC-basedseparations have spiraled resulting in poor economical returnin implementing this mode of chromatography. Figure 7illustrates the selectivity differences between acetonitrile andacetone at pH 3. Clearly, there are dramatic shifts in retentionfor certain analytes and in particular for nicotine, which hadapproximately twice the retention time in acetonitrile comparedto acetone. Slight decreases in retention were observed for allanalytes using acetone which is considered as a weaker solventthan acetonitrile for HILIC. Interestingly, although our evalua-tion was carried out only using a model test mix, superior selec-tivity between analytes was achieved using acetone compared


1.5x107

7

Minutes

Sig

nal I

nten

sity

Sig

nal I

nten

sity

(a)

0.0

3.0x107

4.5x107

76.0x10

2.00

0.0

1.2x10

2.4x107

3.6x107

4.8x107

0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80

(b)

(ii)

(vi)

(vii)

(i)

(iv)

(v)

(iii)

(ii)

(vi)

(vii)

(i)

(iv)

(v)

Figure 7. (a) (CH3)2CO, pH 3.0 and (b) CH3CN, pH 3.0.Composition of solvent/buffer (90:10, v/v). Analytes: (i)cytosine, (ii) adenine, (iii) caffeine, (iv) procainamide, (v)nicotine, (vi) nortriptyline, (vii) diphenhydramine.


with acetonitrile for the other compounds in our study. Clearly,more work needs to be invested into investigating the selectivitymerits of acetone for HILIC as overall we found this solvent toresult in reduced retention of our analytes, apart from nicotine.This is likely due to acetone behaving as a hydrogen-bondacceptor interacting with ionised sites on the probe analytes;this may also explain the improved selectivity observed com-pared with acetonitrile. Therefore, it can be suggested formethod development purposes, and where sensitive positiveion ESI work is not required, that acetone could have utilityfor HILIC interfaced to ESI-MS or other alternative detectorsto ultraviolet.

367

Column robustness metrics

The experimental process in generating the results presentedhere utilised three columns accounting for approximately8500 injections. Each column lasted on average around 3000injections before deteriorating, seen typically in the form ofpeak broadening and tailing. The maximum pressure thecolumn experienced was less than 7000 psi and a maximumtemperature of 30 �C throughout all analyses. All columnswere subjected to high and low pH switching with waterrinses prior to switching, organic solvent switching, constantionic strength blending, and non-constant use over the spanof the column life. ‘Non-constant’ use is defined in thiscontext as the column in use for 24 hours for multiple daysbut placed in shutdown with zero flow until next used.Columns were not stored when not in use and remained onthe system until additional experiments were deemed neces-sary. We feel the remarkable stability of these bare BEH-silicacolumns should be mentioned consequentially as a derivativeto our investigation, adding confidence that this support canbe routinely used.


CONCLUSIONS

Although clearly problematic for positive ion electrosprayionisation (ESI), acetone offers complementary selectivity toacetonitrile which is important for method developmentusing HILIC separations. Our objective was not to embellishor question the mechanisms of ESI but rather offer an objec-tive measure of solvent compatibility with this mode of chro-matography. The chemical noise observed when usingacetone instead of acetonitrile was evaluated. This is impor-tant to measure in order to reveal more clearly the LOD forparticular analytes, which is essential when sensitivity isrequired to be maximised. We have presented and contribu-ted further evidence on the use of basic mobile phases inorder to achieve maximum signal-to-noise of basic soluteswith positive ion ESI-MS. Acetone condensation product for-mation and high average peak-to-peak noise were observedin the low molecular weight region, which may inhibit lowerlimit of quantification (LLOQ) of particular classes of ana-lytes, especially those which do not ionise efficiently as isthe case for nucleobases. We believe our systematic optimisa-tion approach has highlighted some important variables formethod development in the HILIC/MS arena. Further workis underway with regards to the use of acetone as a strongsolvent for use with RPLC and for HILIC interfaced withAPCI-MS for small molecule work.

AcknowledgementsJames Heaton and Dr. Norman W. Smith would like to thankWaters Corp. (MA, USA) for funding.

REFERENCES

[1] R. W. Roos, C. A. Lau-Cam. J. Chromatogr. A 1986, 370, 403.[2] S. Å. Gustavsson, J. Samskog, K. E. Markides, B. Långström.

J. Chromatogr. A 2001, 937, 41.[3] A. J. Alpert. J. Chromatogr. 1990, 499, 177.[4] A. J. Alpert, M. Shukla, A. K. Shukla, L. R. Zieske, S.W. Yuen,

M. A. Ferguson, A. Mehlert, M. Pauly, R. Orlando. J. Chroma-togr. A 1994, 676, 191.

[5] J. A. Boutin, A. P. Ernould, G. Ferry, A. Genton, A. J. Alpert.J. Chromatogr. 1992, 583, 137.

[6] J. J. Pesek, M. T. Matyska, S. Larrabee. J. Sep. Sci. 2007, 30, 637.[7] D. V. McCalley. J. Chromatogr. A 2007, 1171, 46.[8] W. Naidong, J. W. Lee, X. Jiang, M. Wehling, J. D. Hulse, P. P.

Lin. J. Chromatogr. B: Anal. Technol. Biomed. Sci. 1999, 735, 255.[9] P. Hemstrom, K. Irgum. J. Sep. Sci. 2006, 29, 1784.

[10] D. V. McCalley. J. Chromatogr. A 2010, 1217, 3408.[11] H. P. Nguyen, K. A. Schug. J. Sep. Sci. 2008, 31, 1465.[12] S. R. Needham, P. R. Brown, K. Duff, D. Bell. J. Chromatogr. A

2000, 869, 159.[13] E. S. Grumbach, D. M. Diehl, U. D. Neue. J. Sep. Sci.

2008, 31, 1511.[14] R. Fritz, W. Ruth, U. Kragl. Rapid Commun. Mass Spectrom.

2009, 23, 2139.[15] T. R. Keppel, M. E. Jacques, D. D. Weis. Rapid Commun. Mass

Spectrom. 2010, 24, 6.[16] R. E. Duderstadt, S. M. Fischer. J. Chromatogr. A 2008, 1193, 70.[17] K. J. Fountain, J. Xu, D. M. Diehl, D. Morrison. J. Sep. Sci.

2010, 33, 740.[18] T.-Y. Yen, M. Judith Charles, R. D. Voyksner. J. Am. Soc. Mass

Spectrom. 1996, 7, 1106.


3

J. Heaton et al.

3674

[19] S. Zhou, K. D. Cook. J. Am. Soc. Mass Spectrom. 2000, 11, 961.[20] C. R. Blakley, M. L. Vestal. Anal. Chem. 1983, 55, 750.[21] C. R. Blakley, J. J. Carmody, M. L. Vestal. J. Am. Chem. Soc.

1980, 102, 5931.[22] J. Heaton, N. Rahmioglu, K. R. Ahmadi, C. Legido-Quigley,

N. W. Smith. J. Pharm. Biomed. Anal. 2011, 55, 494.[23] M.A. Thite, R. Boughtflower, J. Caldwell, L.Hitzel, C.Holyoak,

S. J. Lane, P. Oakley, F. S. Pullen, S. Richardson, G. J. Langley.Rapid Commun. Mass Spectrom. 2008, 22, 3673.

[24] R. A. O’Hair, G. E. Reid. J. Am. Soc. Mass Spectrom. 2000, 11,244.


[25] L. Peng, T. Farkas. J. Chromatogr. A 2008, 1179, 131.[26] C. Delatour, L. Leclercq. Rapid Commun. Mass Spectrom.

2005, 19, 1359.[27] R. J. Flanagan, I. Jane. J. Chromatogr. A 1985, 323, 173.[28] R. M. Smith, J. P. Westlake, R. Gill, M. David Osselton. J.

Chromatogr. A 1990, 514, 97.[29] R. J. Flanagan, G. C. A. Storey, R. K. Bhamra, I. Jane. J.

Chromatogr. A 1982, 247, 15.[30] Available: www.chemspider.com.[31] L. Charles, D. Pépin, F. Gonnet, J.-C. Tabet. J. Am. Soc. Mass

Spectrom. 2001, 12, 1077.


http://www.chemspider.com

Systematic evaluation of acetone and acetonitrile for use in hydrophilic interaction liquid...

Documents

Transcript of Systematic evaluation of acetone and acetonitrile for use in hydrophilic interaction liquid...