Reference Electrodes with Salt Bridges Contained in NanoporousGlass: An Underappreciated Source of ErrorMaral P. S. Mousavi and Philippe Buhlmann*

Department of Chemistry, University of Minnesota, 207 Pleasant Street South East, Minneapolis, Minnesota 55455, United States

*S Supporting Information

ABSTRACT: In electroanalytical measurements, nanoporous glassplugs are widely used to contain the electrolyte solution that forms asalt bridge between the sample and the reference electrode. Eventhough reference electrodes with plugs made of nanoporous glass(such as Vycor or CoralPor glass) are commercially available and arefrequently used, the limits of their use have not been exploredthoroughly. It is shown here that at ionic strengths lower than 100mM, the half-cell potentials of reference electrodes with nanoporousglass plugs are not sample independent, as it would be desirable, butthey depend on the ionic sample composition. Sample dependentshifts of more than 50 mV in the half cell potential were encountered.Reference potentials were found to be affected in aqueous solutionsby HCl, NaCl, KCl, and CaCl2 and in acetonitrile by CF3COOH and the supporting electrolyte tetrabutylammoniumperchlorate. These observations cannot be explained by the liquid junction potential between two mutually miscible electrolytesolutions, as commonly described with the Henderson equation. Instead, they result from the surface charge density on the glasssurface and the resulting electrostatic screening of ion transfer into the glass pores when the latter have dimensions comparableto or smaller than the Debye length. Users of reference electrodes with nanoporous glass plugs need to be aware of theselimitations to avoid substantial measurement errors.

Reference electrodes are an inevitable part of almost everyelectrochemical measurement, be it potentiometry,

amperometry, voltammetry, impedance spectroscopy, oranother electroanalytical method. Most reference electrodescontact the reference solution and the sample through a saltbridge filled with an electrolyte solution.1 Limiting factors forthe use of these salt bridges are the loss of electrolyte into thesample and contamination of the salt bridge by samplecomponents.2 To minimize these effects and to obtain lowand stable junction potentials, porous plugs filled withelectrolyte solution are often used, separating the referencefrom the sample solution while maintaining electrical contactbetween the two. These porous plugs must have a low electricalresistance while avoiding excessive transfer of electrolyte andsample components through the junction. In order to providethese characteristics, several materials with a high porosity andrelatively small pores have been used both by academicresearchers and commercial suppliers.3

Many studies investigated the properties of referenceelectrodes with microporous ceramic plugs,1,4,5 which werereported to have pore diameters in the 0.1−3 μm range.4,6

Nanoporous glass plugs commercially available under the brandnames Vycor and (very recently) CoralPor have also been verypopular for the fabrication of reference electrodes and offer theadvantage of further restricting liquid flow.3,7−9 The prepara-tion of these glasses is based on the observation that ternarymixtures of silicon dioxide, boric acid, and alkali oxides mayphase-separate on proper heat treatment, resulting in a porous

skeleton rich in silica and surrounded by alkali and boricoxide.10 The latter can be leached out with acids, leaving back ananoporous and strongly hydrophilic glass, sometimes referredto as “thirsty glass”. The pore diameter can be controlled tosome extent and has been reported to lie in the range of 4−20nm.11

Reference electrodes with nanoporous glass junctions areused widely for electrochemical measurements in both aqueousand organic media. In the brief period from January 2012 toMay 2013 alone, more than 70 publications in various journalsexplicitly reported using reference electrodes equipped withVycor glass plugs (see the Supporting Information for a list ofthese references). Since many publications from the same timeperiod described the use of reference electrodes of anunspecified nature, the actual number of studies for whichreference electrodes with nanoporous glass plugs were used islikely much larger. However, even though the use of referenceelectrodes with nanoporous glass plugs was first proposedalready in 1955,12 the effect of nanoporous glass junctions onthe performance of reference electrodes in water has to ourknowledge not been reported in the literature and limitations oftheir performance in organic solvents have been only associatedwith insolubility effects and mobility differences.9

Received: July 16, 2013Accepted: September 4, 2013Published: September 4, 2013

Editors' Highlight

pubs.acs.org/ac

© 2013 American Chemical Society 8895 dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−8901

This study critically probes the weaknesses of referenceelectrodes with nanoporous glass plugs and discusses theconditions that can lead to large measurement errorsintroduced by shifts in the reference potential. Even thoughlimited variations in the potential of reference electrodes maynot be a concern in some electrochemical measurements, manyothers require a highly reproducible reference electrodepotential. For example, an error of 5 mV in the measurementof a monovalent ion with an ion-selective electrode results in anerror of 18% in the determined ion activity.13 In this study, weencountered discrepancies in reference potentials fromreference electrodes with nanoporous glass plugs of morethan 50 mV and found a substantial dependence of referencepotentials on the composition of the sample solutions.Importantly, the largest errors that we observed cannot be

explained by the well-known liquid junction potentials as theyoccur at the interface of two miscible electrolyte solutions.Instead, large errors were caused by electrostatic screening ofion transport through the nanopores of the porous glass plugsand the concomitant formation of a phase boundary potentialat the interface of sample and plug. This phenomenon is notonly analogous to the formation of a phase boundary potentialat the surface of ion exchange resins14 but has also beenobserved recently for a number of nanomaterials. White and co-workers demonstrated permselectivity for ion transport throughchemically modified nanoporous opal films15 and verified thatpermselectivity arises from restricted ion diffusion through theopal nanopores in contact with solutions with low pH and ionicstrength. Similarly, ionic strength and pore size were shown tocontrol ion transport through single nanopores,16−18 metalnanotubule membranes,19 and chemically modified goldnanotubes.20

Surprisingly, it has not yet been reported in the literature thatelectrostatic screening also affects the popular referenceelectrodes with nanoporous glass plugs. This study showshow large such errors can be. It illustrates the effects of theionic strength and the composition of aqueous and organicsample solutions and describes how errors due to electrostati-cally induced permselectivity can be distinguished from thosecaused by liquid junction potentials at the interface of freelymiscible electrolyte solutions. The understanding of theseeffects leads directly to procedures that avoid errors caused byinappropriate use of reference electrodes with nanoporous glassplugs.

■ EXPERIMENTAL SECTION

Cylindrical nanoporous glass plugs (porous Vycor tips, MF-2064; 3 mm diameter, 3 mm length) were purchased fromBioanalytical Systems (Mount Vernon, IN). A Vycor plug and aglass tube (5 cm long, 3 mm diameter) were placed in a Teflonheat shrink tube, followed by heating with a heat gun for 3 minto attach the plug to the glass tube. Aqueous referenceelectrodes with nanoporous glass plugs were then prepared byinserting an AgCl/Ag electrode into the glass tube equippedwith the Vycor glass plug at one end and filling of the tube fromthe other end with a solution that contained 1.0 mM KCl and1.0 M lithium acetate (LiOAc). Note that KCl and LiOAc areboth equitransferrent salts, i.e., salts composed of a cation andan anion with equal mobility. In order to completely fill thenanoporous glass plugs with 1.0 M LiOAc, the electrodes werethen stored in stirred 1.0 M LiOAc solutions for at least 1 week.Calibration curves were measured by successive dilution or

addition of aliquots of concentrated solutions of NaCl, KCl,CaCl2, or HCl.

Measurements in Aqueous Media. All potentialmeasurements were performed in stirred aqueous solution atroom temperature (25 °C) with an EMF 16 channelpotentiometer (Lawson Laboratories, Malvern, PA) controlledwith EMF Suite 1.02 software (Lawson Laboratories). Forcomparison, external reference electrodes (DX-200; MettlerToledo, Columbus, OH) with a free flow double-junctionAgCl/Ag electrode, a 1.0 M lithium acetate (LiOAc) bridgeelectrolyte (movable ground glass sleeve junction)1 and a 3.0 MKCl reference electrolyte were used. All the solutions wereprepared with deionized purified water (18.2 MΩ cm specificresistance, EMD Millipore, Philadelphia, PA), and all measure-ments were performed in triplicate. The pH measurementswere performed with a pH half-cell electrode from MettlerToledo.

Measurements in Organic Solutions. Reference electro-des with nanoporous glass plugs for use in organic solutionswere prepared by inserting an Ag wire into a glass tubeequipped with a Vycor glass plug at one end, filling of the tubefrom the other end with an acetonitrile solution that contained10.0 mM AgNO3 and 100 mM tetrabutylammoniumperchlorate (NBu4ClO4), and inserting the thus preparedelectrodes into a stirred solution of identical composition atleast for 1 week prior to measurements. Cyclic voltammogramsof bis(pentamethylcyclopentadienyl)iron, Me10Fc, (97%; Al-drich, Milwaukee, WI) were measured with a CHI600Cpotentiostat (CH Instruments, Austin, TX). All electrochemicalmeasurements were carried out with a scan rate of 100 mV/s,using a three electrode setup with a 3.0 mm-diameter gold diskas the working electrode (BAS, West Lafayette, IN), a 0.25 mmPt wire coil (99.998%, Alfa Aesar, Ward Hill, MA) as theauxiliary electrode (total surface area ≈ 0.5 cm2), and thereference electrodes with nanoporous glass plugs prepared inhouse. The working electrode was polished on Microclothpolishing pads using 5.0 μm Micropolish II deagglomeratedalumina, both from Buehler (Lake Bluff, IL). After polishing,the working electrode was rinsed thoroughly with deionizedwater and then ethanol, followed by drying under a stream ofargon. Prior to measurements, all solutions were purged withargon for 15 min while stirring vigorously to remove dissolvedoxygen. All potentials are reported with respect to 10 mM Ag+/Ag reference electrodes with nanoporous glass plugs.All electromotive force (emf) values for measurements with a

free-flow liquid-junction were corrected for liquid junctionpotentials using the mobility of the ions21,22 and the Hendersonequation,23 except for the measurements with trifluoroaceticacid, CF3COOH (Figure 7). Ionic activities were calculatedusing a two-parameter Debye−Huckel approximation.24 Toobtain free flow salt bridge junctions, glass pipettes were pulledin a flame to give tip diameters of ≈20 μm and filled with eitheraqueous LiOAc solutions (1.0 M) or acetonitrile solutions ofNBu4ClO4 (100 mM). In all free flow experiments, referenceelectrodes with nanoporous glass plugs were placed into thesepipets filled with electrolyte solution and a small flow ofsolution out of the pipet into the sample was confirmed. Thesemeasurements were then repeated in the same sample solutionsbut without the electrolyte solution filled pipet, i.e., with directcontact of the nanoporous glass plugs and the sample solution.A 1255B frequency response analyzer and a SI 1287

electrochemical interface from Solartron (Farnborough, Hamp-shire, U.K.) with a three-electrode system were used to

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018896

determine the electrical resistance of the nanoporous glassplugs. A Pt wire and a free flow double-junction AgCl/Agelectrode with a 1.0 M lithium acetate (LiOAc) bridgeelectrolyte were used as counter and reference electrodes,respectively. Resistances were obtained by fitting of impedancespectra with ZPlot software (Scribner Associates, SouthernPines, NC) in the 104 to 105 Hz range.

■ RESULTS AND DISCUSSIONIn an ideal electroanalytical measurement, the salt bridge thatconnects the sample and reference half cells does notcontribute to the measured cell potential. To achieve nearideality, salt bridges typically contain an equitransferrent salt inhigh concentration, minimizing the liquid junction potential atthe interface of the salt bridge and the sample. This work showsthat, when used in samples of moderate or low ionic strength,reference electrodes with nanoporous glass plugs exhibit farfrom ideal behavior. Quite to the contrary, these referenceelectrodes exhibited near Nernstian responses to several ions,both in aqueous media and organic solvents.As shown in Figure 1, reference electrodes with nanoporous

glass plugs and aqueous 1.0 M LiOAc/1.0 mM KCl inner

electrolyte solutions respond potentiometrically to aqueousHCl solutions. To investigate whether the measured emfresponses were due to entrapment of previously measuredsamples in the plug pores, as well-known for referenceelectrodes with conventional porous ceramic plugs whenthere is small or negligible flow of salt bridge electrolyte intothe sample,25 the response to HCl was measured in foursequential steps. At first, the HCl solution was diluted to raisethe pH, then HCl was added into the diluted solution to lowerthe pH again, then a second dilution was performed, and finallyHCl was added a second time. If indeed occurring, low diffusiveor convective mass transfer into and out of the porous plugwould be expected to result in a substantial memory effect.However, Figure 1 shows only a very small hysteresis effect.The four pH response curves from the two dilutions and twoHCl additions deviate from one another by only a few

millivolts, i.e., much less than the nearly 100 mV of emfresponse between pH 2 and 5. This clearly shows that sampleentrapment in the porous plug and ensuing hysteresis have onlya minor effect on the observed potentiometric response.Interestingly, the pH response is not limited to the region oflowest ionic strength but is largest from pH 4 all the way downto pH 2, where the ionic strength is 10 mM. Indeed, in thisregion the pH response has a slope of approximately 30 mV/decade, which is roughly half of the Nernstian response slope of58 mV/decade as exhibited by a pH glass electrode.Figure 2 illustrates how the potentiometric response of the

reference electrode with the nanoporous plug depends on the

ionic strength of the sample. In HCl solutions with abackground of high ionic strength (50 mM CaCl2), theobserved emf changes less than 2 mV upon a thousandfoldchange in the concentration of HCl (filled triangles), whereassimilar measurements without the CaCl2 background yieldedemf changes of more than 100 mV (filled circles). Note that thedifference in the potentiometric responses cannot be explainedby liquid junction potentials between the sample and theadjacent bridge electrolyte solution, as illustrated by the lack ofa potentiometric response to HCl in solutions without theCaCl2 background when a free-flow double-junction referenceelectrode was used (open circles).These potentiometric responses are all consistent with

electrostatic screening of ion access into the pores of thenanoporous glass plugs, as it is similarly known, e.g., for ionexchange resins,14 nanoporous opal films,15 nanopores,26 andnanotubes.20 Access of ions into and diffusive transport of ionsthrough porous materials may be affected by electrostatic forcesif the surface of these materials exhibit a surface charge due toelectrically charged surface functional groups or adsorbedions.16 If the sum of the Debye screening length of the electricfield within the pores and the ion radius is much smaller thanthe pore size, ionic diffusion through the porous material is notsignificantly affected by electrostatic forces. However, when thepore size approaches the nanometer range, electrostaticinteractions with the surface charge start to affect the presenceof cations and anions with opposite effects. This results in

Figure 1. Potentiometric response to HCl of a reference electrodewith a nanoporous glass plug, relative to a conventional free-flowdouble-junction reference electrode. To assess possible hysteresis, theemf was measured in sequential steps, i.e., (1) starting by stepwisedilution of a HCl solution of low pH (empty black circles), then (2)adding aliquots of concentrated HCl to lower the pH (filled greencircles), followed by (3) a second sequence of dilutions (empty blacktriangles), and finally (4) adding once more concentrated HCl toreturn to the original pH (filled red triangles). The triangle in the topright corner illustrates for comparison the ideal Nernstian responseslope of an ion selective electrode to a monocation.

Figure 2. Potentiometric response to HCl of a reference electrodewith a nanoporous glass plug, relative to a conventional free-flowdouble-junction reference electrode. Filled circles and triangles:responses of the nanoporous plug reference electrode in directcontact with HCl samples, sample with a background of 0 and 50 mMCaCl2, respectively. Empty circles: response of a free-flow double-junction reference electrode consisting of the above referenceelectrode with a nanoporous glass plug inserted into a pipet filledwith 1.0 M LiOAc. A small flow of the LiOAc solution from the pipetinto the sample solution was ensured.

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018897

partial or even complete exclusion of ions of one sign from theporous material, an effect that is referred to as permselectivity.Since the electrochemical potential of any type of ion in thepores has to equal its electrochemical potential in the bulk of asolution in equilibrium with this porous material, it can beshown readily that an electrical phase boundary potential mustdevelop at the interface between the bulk solution and theporous material.14

A lower pH reduces the surface charge and weakenselectrostatic screening, while at higher pH, the negative surfacecharge on the glass surface due to deprotonation of silanolgroups27,28 results in partial exclusion of anions from the glassnanopores. This is consistent with the effect of the ionicstrength on the potentiometric response. Since the Debyescreening length decreases with the ionic strength,29,30 thepermselectivity of nanoporous materials is lost at high ionicstrengths.15,16,20,26,31,32 This explains the different potentio-metric pH responses in absence and presence of a CaCl2background, as shown in Figure 2.To further test the consistency of this model with

experimental observations, the potentiometric responses ofreference electrodes with nanoporous glass plugs to solutions ofdifferent metal chlorides were determined. Figure 3 shows that

the KCl and NaCl responses are identical within error, givingno evidence for specific interactions of the monovalent cationsK+ and Na+ with the glass surface. In an intermediate rangefrom 1.0 mM to 30 mM, the KCl and NaCl responses are closeto the theoretical (Nernstian) response that is expected for thepermselective population of the nanopores with monocations.Also, as expected for an increasingly shorter Debye screeninglength at high ionic strength, the measured emf reaches aplateau at high KCl and NaCl concentrations. Similarobservations were made for the CaCl2, but the responseslope in the 1.0 mM to 30 mM range was only half as large asfor NaCl and KCl. This is indeed expected in the case ofpermselective cation transfer into the glass nanoporesconsidering the inverse relationship of the phase boundarypotential change and the charge number of the cation.As for the HCl response shown in Figure 2 too, the high

ionic strength provided by a 1.0 M LiOAc backgroundeliminated the responses to NaCl, KCl, and CaCl2 (Figure4A), which is again consistent with the electrostatic screeningmodel. Use of a more dilute ionic background of 1.0 mMLiOAc to provide a well-defined phase boundary potentialrestores the response to NaCl, KCl, and CaCl2 but againconfirms the overall low level of ion selectivity (Figure 4B).

However, the H+ response stands apart from the responses tothe two monovalent cations Na+ and K+, which is not surprisinggiven the specific interaction of H+ with deprotonated surfacegroups on glass.Electrostatic screening effects at the interface between

sample solutions and the nanoporous salt bridge plugs areconsistent with the dimensions of the nanopores. The Debyelengths in 0.001 and 1.0 M LiOAc are 9.6 and 0.30 nm,29,30

respectively. The former is much larger than the ionic radii ofH+, Na+, K+, Ca2+, and Cl−, which are all less than 1.0 nm.33

Moreover, the Debye length in the 0.001 M LiOAc solutions isalso substantially larger than the average pore size of the porousglass plugs used in this study, which was 4.0 nm.11

Consequently, electrostatic screening effects are to be expected.Further insight into the effect of electrostatic screening on

the ion transport through the nanoporous glass plugs wasgained from electrical resistance measurements. For thispurpose, a reference electrode with a nanoporous glass plugfilled with 1.0 M LiOAc/1.0 mM LiCl and an inner fillingsolution of the same composition was inserted into a stirredsolution with a much lower ionic concentration (1.0 mMLiOAc/1.0 mM LiCl). As Figure 5 shows, the electricalresistance of the nanoporous glass plug remained steady over24 h at 5.8 ± 0.8 kΩ. Note that if free ion transport from theplug into the 1.0 mM LiOAc solution had gradually depletedthe LiOAc in the nanoporous plug, the electrical resistanceshould have gradually increased over the time frame of severalhours that is necessary for diffusion through a plug of 3 mmthickness. Moreover, the measured value of the electricalresistance in this experiment is much closer to the 2.1 ± 0.1 kΩresistance of a nanoporous glass plug filled with 1.0 M LiOAc/1.0 mM LiCl solution when in contact on both sides with 1.0 MLiOAc/1.0 mM LiCl than to the 166 ± 6 kΩ resistance of ananoporous glass plug filled with 1.0 mM LiOAc/1.0 mM LiClin contact on both sides with 1.0 mM LiOAc/1.0 mM LiCl.This suggests that even after 24 h of exposure on one side to1.0 mM LiOAc/1.0 mM LiCl, the plug which was filledoriginally 1.0 M LiOAc/1.0 mM LiCl, is still filled

Figure 3. Effect of CaCl2, NaCl, and KCl on the potentiometricresponse of reference electrodes with nanoporous glass plugs as afunction of the ionic strength.

Figure 4. Ion selectivity of the HCl, CaCl2, KCl, and NaCl responsesof reference electrodes with a nanoporous glass plug in a backgroundof (A) 1.0 M LiOAc and (B) 1.0 mM LiOAc.

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018898

predominantly with this solution of much higher concentration.This is consistent with an interfacial layer at the water/porousglass interface that inhibits anion transfer to a large extent byelectrostatic screening.A look at the recently published literature shows that

reference electrodes with nanoporous glass plugs are not onlyused frequently for measurements with aqueous samples butare also very popular for measurements in organic solvents suchas acetonitrile,34 propylenecarbonate,35 methanol,36 dichloro-methane,37 and ionic liquids.38 This prompted us to investigatethe performance of reference electrodes with nanoporous glassplugs also with acetonitrile as a representative organic solvent.To assess the potential contribution of the nanoporous glassplugs, the peak potential in the oxidation of decamethylferro-cene, Me10Fc, in a cyclic voltammogram (CV) was measuredwith respect to a reference electrode with a nanoporous plugand an inner filling solution containing 10.0 mM AgNO3 and100.0 mM tetrabutylammonium perchlorate (NBu4ClO4) inacetonitrile. The Me10Fc

+/Me10Fc couple was chosen because itis known to be minimally affected by the sample composi-tion.39,40 For comparison and analogous to the aqueousexperiments performed for Figure 2, the peak potential of theMe10Fc oxidation was also determined with respect to a free-flow double junction electrode, for which only a conventionalliquid junction potential is expected at the interface to thesample. For the latter purpose, a reference electrode with ananoporous plug was placed into a pipet filled with anacetonitrile solution containing 100 mM NBu4ClO4 as theelectrolyte solution. When the thus prepared double junctionelectrode was placed into samples, it was ensured that a verysmall flow of the 100 mM NBu4ClO4 into the sample occurred.As expected, in the free flow setup, the potential of the

reference electrode was only minimally affected by theconcentration of the supporting electrolyte (Figure 6, emptycircles). However, when the nanoporous plugs were directly incontact with the sample, a change in the peak potential of theMe10Fc oxidation of 20 mV was observed upon stepwisedilution of the supporting electrolyte from 100 to 6.25 mM.This finding confirms electrostatic screening also in acetonitrilesolution, despite the fact that the pKa values are well-known tobe much higher in acetonitrile than in water,41 making it harderto ionize the glass surface in this solvent.

The effect of CF3COOH, on the potential of referenceelectrodes was investigated as well. CF3COOH is commonlyused to change the pH of organic solutions in order to probethe pH dependence of redox potentials and elucidate themechanism of electron transfer reactions.42−44 Such studieswould be seriously compromised if the potential contributionof the reference electrode were affected by addition ofCF3COOH but changes observed in the CVs were mistakenlyinterpreted as changes in the redox potentials of anelectroactive species caused by the involvement of protons inthe electron transfer step.As can be seen from Figure 7, the reference electrodes with

nanoporous glass plugs show permselectivity for CF3COOH atlow ionic strength. Addition of CF3COOH to a solution thatdoes not have a sufficiently high concentration of supportingelectrolyte (6.25 mM) causes significant undesirable changes inthe reference potential (Figure 7A). However, by increasing theconcentration of supporting electrolyte or avoiding directcontact between the porous plug and the sample, the changesin the reference potential during the measurement can beminimized (Figure 7B). As in the case of the experiments withaqueous solutions, the potential changes at low ionic strengthscan be explained by anion exclusion and formation of a chargeseparation layer at the interface of the sample and thenanoporous glass plug.

■ CONCLUSIONSIn this work, we demonstrated that the porous glass plugsfrequently used to provide an interface between referenceelectrodes and samples can cause unwanted variations inreference potentials. Repulsion forces between freely dissolvedanions and the negative charges of the nanopore surfaces havebeen recognized as the source of error. Not only at low buteven at moderately high ionic strengths, the Debye lengthexceeds the radius of the pores, resulting in anion exclusionfrom the porous plugs and formation of a charge separationlayer at the interface of the porous glass and the sample. Thiseffect occurs both with aqueous and organic solutions. Thepotential of the reference electrodes with nanoporous glassplugs is sample independent only in solutions with com-paratively high ionic strength, and the use of a highlyconcentrated salt bridge electrolyte alone does not prevent

Figure 5. Electrical resistance of a nanoporous glass plug that wasinitially filled with 1.0 M LiOAc/1.0 mM LiCl and stored in contact onboth sides with 1.0 M LiOAc/1.0 mM LiCl for at least 12 h and thenbrought at t = 0 h on one side in contact with 1.0 mM LiOAc/1.0 mMLiCl while keeping the same 1.0 M LiOAc/1.0 mM LiCl solution onthe other side. Error bars represent results from three separatemeasurements.

Figure 6. Effect of concentration of NBu4ClO4 supporting electrolyte(6.25−100 mM) on the potential of reference electrodes. Shown is thepeak potential of the Me10Fc (0.1 mM) oxidation as determined fromCVs. Filled circles: reference electrode with a nanoporous glass plug indirect contact with the sample. Empty circles: free-flow double-junction reference electrode with a 100 mM NBu4ClO4 solutionflowing into the sample.

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018899

measurement errors. In situations where modifying the samplecomposition to achieve a high ionic strength is not possible, thefree-flow double-junction setup is preferable. Electrostatictheory suggests that, alternatively, micro- rather than nano-porous plugs may be used to avoid unwanted sampledependencies of the reference potential.While we emphasize here the problems that can arise when

using reference electrodes with nanoporous glass plugs, it isevident that reference electrodes with nanoporous plugsprepared from other materials may also be affected by thesame problem if they have surfaces with ionizable groups. Forexample, some of the problems observed by Illingworth5 with aselection of porous ceramic plugs may have had similar origins,although this has apparently never been investigated in detail.With a broader point of view, our results also add a newexample to the still relatively small number of examples forelectrostatic screening at nanoporous materials.15,20,26 Implica-tions of this effect go beyond electroanalytical chemistry andshould also be important to the development of, e.g., batterymaterials and nanoporous supercapacitors.

■ ASSOCIATED CONTENT*S Supporting InformationAdditional information as noted in text. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: Buhlmann@umn.edu.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This project was partially supported by Grant RL-0012-12 fromthe Initiative for Renewable Energy and the Environment of theUniversity of Minnesota. We acknowledge Jon Thompson forinteresting discussions about porous glasses.

■ REFERENCES(1) Dohner, R. E.; Wegmann, D.; Morf, W. E.; Simon, W. Anal.Chem. 1986, 58, 2585−2589.(2) Zhang, T.; Lai, C. Z.; Fierke, M. A.; Stein, A.; Buhlmann, P. Anal.Chem. 2012, 84, 7771−7778.(3) BASi Reference Electrodes. http://www.basinc.com/products/ec/ref.php (accessed May 3, 2013).(4) Bosch, R. W.; Straetmans, S.; Dyck, S. V. J. Mater. Sci. Lett. 2002,37, 3973−3979.(5) Illingworth, J. A. Biochem. J. 1981, 195, 259−262.(6) Ito, S.; Kobayashi, F.; Baba, K.; Asano, Y.; Wada, H. Talanta1996, 43, 135−142.(7) Bott, A. W. Curr. Sep. 1995, 14, 64−68.(8) BioLogic Science Instruments Reference Electrodes. http://www.bio-logic.info/potentiostat/accessories/ReferenceElectrodes.html (ac-cessed May 25, 2013).(9) Carano, M.; Colonna, B.; Echegoyen, L.; Le Derf, F.; Levillain, E.;Salle, M. J. Supramol. Chem. 2003, 15, 83−85.(10) Hood, H. P.; Nordberg, M. E. Treated Borosilicate Glass. U.S.Patent 2,106,744, February 1, 1938.(11) VYCOR Brand Porous Glass 7930. http://www.corning.com/docs/specialtymaterials/pisheets/Vycor%207930.pdf (accessed May 5,2013).(12) Carson, W. N.; Michelson, C. E.; Koyama, K. Anal. Chem. 1955,27, 472−473.(13) Zou, X. U.; Buhlmann, P. Anal. Chem. 2013, 85, 3817−3821.(14) Helfferich, F. G. Ion Exchange; McGraw-Hill: New York, 1962.(15) Newton, M. R.; Bohaty, A. K.; White, H. S.; Zharov, I. J. Am.Chem. Soc. 2005, 127, 7268−7269.(16) Wang, G.; Zhang, B.; Wayment, J. R.; Harris, J. M.; White, H. S.J. Am. Chem. Soc. 2006, 128, 7679−7686.(17) Zhang, Y.; Zhang, B.; White, H. S. J. Phys. Chem. Soc. 2006, 110,1768−1774.(18) Lan, W.; Holden, D. A.; White, H. S. J. Am. Chem. Soc. 2011,133, 13300−13303.(19) Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268,700−702.(20) Jagerszki, G.; Gyurcsanyi, R. E.; Hofler, L.; Pretsch, E. Nano Lett.2007, 7, 1609−1612.(21) Morf, W. E. The Principles of Ion-Selective Electrodes and ofMembrane Transport; Elsevier: New York, 1981.(22) Springer, C. H.; Coetzee, J. F.; Kay, R. L. J. Phys. Chem. 1969,73, 471−476.(23) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:Fundamentals and Applications, 2nd ed.; John Wiley and Sons: NewYork, 2001.(24) Meier, P. C. Anal. Chim. Acta 1982, 136, 363−368.(25) Brezinski, D. P. Analyst 1983, 108, 425−442.(26) Gyurcsanyi, R. E. Trends Anal. Chem. 2008, 27, 627−639.(27) Behrens, S. H.; Grier, D. G. J. Phys. Chem. 2001, 115 (14),6716−6721.(28) Li, P.; Zhang, F. J. Non-Cryst. Solids 1990, 119, 112−118.(29) Russel, W. B.; Saville, D. A.; Schowalter, W. R. ColloidalDispersion; Cambridge University Press: Cambridge, U.K., 1989.

Figure 7. Effect of CF3COOH on the potential of reference electrodeswith nanoporous glass plugs. (A) CVs of 0.1 mM Me10Fc in 6.25 mMNBu4ClO4 in the absence and presence of CF3COOH. (B)Dependence of potential of Me10Fc (0.1 mM) oxidation on theCF3COOH concentration. Filled circles and triangles show results fora reference electrode with a porous plug in direct contact to sampleswith an electrolyte concentration of 6.25 and 100 mM NBu4ClO4,respectively. Empty circles: sample with a 6.25 mM NBu4ClO4electrolyte and a free-flow double-junction reference electrode (100mM NBu4ClO4 solution flowing into the sample).

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018900

(30) Israelachvili, J. Intermolecular and Surface Forces, 1st ed.;Academic Press Inc.: London, 1985.(31) Newton, M. R.; Bohaty, A. K.; Zhang, Y.; White, H. S.; Zharov,I. Langmuir 2006, 22, 4429−4432.(32) Jagerszki, G.; Takacs, A.; Bitter, I.; Gyurcsanyi, R. E. Angew.Chem., Int. Ed. 2011, 123, 1694−1697.(33) Harris, D. C. Exploring Chemical Analysis, 3rd ed.; W. H.Freeman and Company: New York, 2005.(34) Rickert, P. G.; Antonio, M. R.; Firestone, M. A.; Kubatko, K. A.;Szreder, T.; Wishart, J. F.; Dietz, M. L. Dalton Trans. 2007, 5, 529−531.(35) Olson, E. J.; Buhlmann, P. J. Electrochem. Soc. 2013, 160, A320−A323.(36) Asangil, D.; Tasdemir, I. H.; Kılıc, E. J. Pharm. Anal. 2012, 2,193−199.(37) Do, K.; Kim, C.; Song, K.; Yun, S. J.; Lee, J. K.; Ko, J. Sol. EnergyMater. Sol. Cells 2013, 115, 52−57.(38) Xu, M.; Ivey, D. G.; Xie, Z.; Qu, W. Electrochim. Acta 2013, 89,756−762.(39) Ruiz, J.; Astruc, D. C. R. Acad. Sci., Ser. II Univers. 1998, 1, 21−27.(40) Noviandri, I.; Brown, K. N.; Fleming, D. S.; Gulyas, P. T.; Lay,P. A.; Masters, A. F.; Phillips, L. J. Phys. Chem. B 1999, 103, 6713−6722.(41) Raamat, E.; Kaupmees, K.; Ovsjannikov, G.; Trummal, A.; Kutt,A.; Saame, J.; Koppel, I.; Kaljurand, I.; Lipping, L.; Rodima, T.; Pihl,V.; Koppel, I. A.; Leito, I. J. Phys. Org. Chem. 2013, 26, 162−170.(42) Le, T. H.; Nafady, A.; Qu, X.; Martin, L. L.; Bond, A. M. Anal.Chem. 2011, 83, 6731−6737.(43) Ou, Z.; Khoury, T.; Fang, Y.; Zhu, W.; Sintic, P. J.; Crossley, M.J.; Kadish, K. M. Inorg. Chem. 2013, 52, 2474−2483.(44) Wang, D.; Zhang, M.; Buhlmann, P.; Que, L. J. Am. Chem. Soc.2010, 132, 7638−7644.

Analytical Chemistry Editors' Highlight

dx.doi.org/10.1021/ac402170u | Anal. Chem. 2013, 85, 8895−89018901