Adsorption
Behavior
of ro-Hexanol
onAg(lll)
from
Aqueous
0.05
M
KCIO4
Maria Luisa
Foresti*
and Massimo
Innocenti
Dipartimento di Chimica, Universita’di Firenze, Via G. Capponi 9, 50121 Firenze,
Italy
Antoinette Hamelin
Laboratoire d’ElectrochimieInterfaciale, CNRS, 1 PlaceA. Briand,
92195MeudonCedex,France
ReceivedMay9, 1994. In Final Form:September 27, 1994®
Acomputerized chronocoulometric techniquewas employedtoinvestigate the adsorption behavior of n-hexanol(NHEX)on the (111) faceofa silversinglecrystalfromaqueous 0.05M KCIO*. Thecharge density, <tm, as a function ofthe applied potential, E, andofthe NHEX concentration was analyzed
thermodynamically,yieldingthe Gibbs surfaceexcess andthe standardGibbsenergy ofadsorption,AG°ads,
of NHEX as afunction of Eand ctm· Acomparisonofthe experimental adsorption parameterswith a
molecular model points toaflat orientation oftheadsorbedNHEXmolecules. Thelow valueof—AG°ads
atthe potentialofmaximum adsorption,18.4kJmol™1,isindicativeofthestronglyhydrophilicityofsilver. Introduction
Investigationsoforganicadsorptionon solid metalsare
importantfrom bothatechnological anda fundamental
point ofview. Thus,these researchesmay allowabetter understanding ofthemechanism ofelectrode processes
in electrosynthesis and electrocatalysis,as wellasabetter control in theuse oforganicsas corrosioninhibitorsand polishing agents. Moreover, the adsorption ofneutral
organic compounds at electrodes provides valuable in-formation on the structure oftheinterfacialregion.
Quantitative investigations ofthe adsorption of
un-chargedorganicmoleculeson singlecrystalelectrodesare
scanty andthemajorityof themwere carried outon metals
withlowmeltingpointssuchas Bi,1Sn,*12
Zn,3andPb,4on thebasisofdifferential capacity measurements andwith
an extrathermodynamic analysis oftheexperimentaldata. Differential capacity measurements often fail to yield
realistic quantitative data5 because ofthelack of equi-librium at the interphase: this isthe main reason why
these measurements are hardlyamenableto a thermo-dynamic analysis, which is quite demanding as to the
accuracy andreproducibilityto the data tobe analyzed.
Results accurate enough to permit a satisfactory thermodynamicanalysis havebeenmore recentlyobtained
bychronocoulometry, which allowsreproducible
adsorp-tion measurements even if adsorption is very slow, providedthat asmallpotentialrangeis available where desorption is fast and complete. The differences in behavioron differentsinglecrystalfacesare small for the low melting metals, probably due to a relatively high
surfacemobility.6 Conversely,substantialdifferencesare
observedwith Ag and Au. Chronocoulometryhas been employed in the study ofthe adsorption oftert-pentyl
alcohol,pyridine, diethyl ether, andbenzonitrile on Au
®
Abstract published inAdvanceACS Abstracts,December15,
1994.
(1)Paltusova,N. A.;Alumaa,A. R.;Palm,U. V.Elektrokhimiya1979, 15, 1870.
(2)Kuprin, . P.;Grigoryev, N. B.Elektrokhimiya1980, 16, 383. (3)Ipatov, Yu.P.;Batrakov, V. V.Elektrokhimiya1975, 11, 1282. (4)Khmelevaya,L. P.;Chizhov,A.V.;Damaskin,B. B.;Vainblat,T.
I.Elektrokhimiya1980, 16, 257.
(5)Nguyen VanHuong,C.;Hinnen,C.;Dalbera,J. P.;Parsons,R.
J.Electroanal.Chem. 1981, 125, 177.
(6)Parsons,R.J.Electroanal. Chem. 1983, 150, 51.
single crystalfaces andin that ofpyridineon Ag single crystalfaces.7
Quantitativeinvestigationsoftheadsorption ofsimple aliphatic compoundsare quitelimitedin numberon Au single crystal faces.8 No definite conclusions have yet
beendrawn7astothe effectofthedifferent crystallographic
orientations ofAu on teri-pentyl alcohol adsorption.910
Morescarce resultswere publishedforAg single crystal
faces.
For adsorption of aliphatic compounds on Ag single crystal faces, no experimental data accurate enough to allowathermodynamic analysisare presently available
in the literature. Hinnenetal.11comparedthedifferential capacity(C)vs potentialcurves of0.02MNaFaqueous solutions saturated with ethyl ether on the three low-indexfacesofAgwiththesame curves in theabsenceof
ether. Ethyl ether does not affect theCvs E curve on
Ag(110),thusindicating that itsadsorptionon thisface is either vanishinglysmallor verysloweven atthe low frequency adopted (15 Hz). However, ethyl ether de-presses thedifferentialcapacityon the
Ag(lll)
and Ag-(100)facesin theproximityofthepotential ofzerocharge,this depressionbeingslightlygreateron
Ag(lll)
thanonAg(100). The lackofa desorption peak on the positive
side ofthecapacityminimum,which contrasts withthe
presence ofadesorption peakon the negative side, was
ascribed by the authors to slow desorptionkinetics,11which is particularlypronounced at positive potentials. This
prevented a quantitative estimate of diethyl ether
ad-sorptionon Ag fromdifferentialcapacity measurements.
VitanovandPopovdetermined the adsorption isotherms of -hexanol,12isobutylalcohol,13andro-pentanol14on
Ag-(111)andAg(100) atthe potential of maximum adsorption from differential capacity measurements using an
ex-(7)Lipkowski,J.;Stolberg, L. InAdsorptionofMolecules at Metal
Electrodes·,Lipkowski,J.,Ross,P.N.,Eds.;PlenumPress: New York,
1992;p 171.
(8)Lipkowski,J.;Nguyen Van Huong,C.;Hinnen,C.;Parsons, R.
J. Electroanal.Chem. 1983, 143, 375.
(9)Richer, J.; Stolberg, L.;Lipkowski,J.Langmuir1986, 2, 630. (10)Richer,J.;Lipkowski,J.J.Electroanal.Chem. 1988, 251, 217. (11)Hinnen,C.; Nguyen Van Huong, C.;Dalbera,J. P. J. Chim.
Phys.(France) 1982, 79, 37.
(12)Vitanov,T.; Popov, A.Elektrokhimiya1974, 10, 1373. (13)Vitanov,T.; Popov, A.Elektrokhimiya1976, 12, 319. (14) Popov,A.;Velev, O.;Vitanov,T. J.Electroanal.Chem. 1988, 256, 405.
0743-7463/95/2411-0498$09.00/0 © 1995American Chemical Society
Downloaded via UNIV DEGLI STUDI DI FIRENZE on October 6, 2020 at 14:41:46 (UTC).
Adsorption Behaviorof
Ag(lll)
Langmuir, Vol. 11, 2, trathermodynamic approach based on thetwo-parallelcapacitormodel. Theadsorptionofn-hexanolfrom0.05 MNa2S04was investigatedon Ag single crystals grown
electrolyticallyinaglasscapillary; thesinglecrystalfaces
exhibited a number of screw dislocations which were
partially eliminatedbyelectrodepositingAg+ ionsata 1
mVoverpotential.12 Accordingto the authors, thetime
requiredtoattain adsorptionequilibriumatthe potential ofmaximum adsorptionwas about4hforthe mostdilute alcohol solutions and30 min for the saturated solution.
After attaining these equilibrium conditions, the dif-ferentialcapacitywas measuredover thewholepotential rangewithouthavingtoestablishanyfurther equilibrium conditions. Inthe adsorption measurementsofisobutyl
alcoholfrom0.05MNa2S04, the Ag single crystal faces
were electropolishedanodicallybeforeuse:13thisprocedure
yielded reproducible positive desorption peaks and ap-parentlyreducedthetimerequiredforadsorption equi-librium to 30min forthemostdilutealcoholsolutionsand to 10min for themost concentrated ones. Comparable timeswere requiredforro-pentanol adsorption from0.1
MKFon Ag singlecrystals grownelectrolytically in Teflon
capillaries.14 TheabsolutevalueoftheGibbsadsorption freeenergy ofn-hexanol12was foundtobeslightlygreater
on
Ag(l
11)thanon Ag(100)(AG°m= —24.6kJmol-1andAG°ioo= —23.0kJmol-1) whereasan
oppositetrendwas reportedforn-pentanol14 (AG°m = -15.0 kJmol-1 and AG°ioo= -15.8 kJmol-1). Theauthors explainedthe slow attainment ofadsorptionequilibriumwhenshiftingthe potentialfrom negative valuestothepotential ofmaximum
adsorption by a slow adsorption on the growth stepsof the silver surface. However, such a slow adsorptionis
apparently inconsistentwiththe sharpandhigh adsorp-tion-desorptionpeaks reportedby the authors on both
sidesofthe flat differential capacityminimum.
Experimental
SectionThe water usedwas obtained frombottledlightmineralwater bydistillingit once andthendistillingthe water so obtained
from acidic permanganate, while the heads were constantly
discarded. MerckKC104was recrystallized twice frombidistilled
waterandthendried. Fluka analyticalreagent grade n-hexanol
was distilled inan atmosphereofnitrogenbeforeuse. The alcohol concentrationswas varied only from3.46x 10-3to3.46x 10-2
Mbecausefor lowerconcentrationsthe capacitance—potential
curves ofthe base electrolyte are not altered by addition of n-hexanol(NHEX),andno higherconcentrationswere studied
becauseof solubility limits in0.05MKC104. Thesolutionswere
freshly prepared just before the beginning ofeach series of measurements.
The working electrodesused in this work were cylindrical
silver single crystals grown accordingtotheBridgeman technique inagraphitecrucible, oriented by X-rays andcut15either inthe Laboratoire d’Electrochimie Interfacialedu ONESin Meudon-Bellevue (France)or inFlorence, Italy. Theseelectrodeswere
polishedwith successivelyfinergradesof aluminapower down to 0.05µ and thenannealed in a muffle furnace at normal
atmospheric pressurefor30minat650 °C. Allelectrochemical measurements were carried out in Florence. Before each
experiment the electrode was polished chemically with CrOa
according to the procedure describedinref16. Afterpolishing, the electrode surfacewas soakedin concentratedsulfuricacid
forabout20minandthen rinsed thoroughly withwater. The hangingsolutionmethod17was employed. Thecylindricalsingle
crystalelectrodewas held byaplatinum wiresealedintoaglass
tube, whichwas secured toamovable stand; thelatterwas moved
(15)Hamelin, A. In ModernAspectsofElectrochemistry; Conway,B. E.,White,R. E.,Bockris,J.O’M.,Eds.;PlenumPress: New York,1985;
Vol. 16,p 1.
(16)Hamelin, A.; Stoicoviciu, L.; Doubova, L; Trasatti, S. J. Electroanal. Chem. 1988,244,133.
(17)Dickertmann,D.;Schultze, J. W.; Koppitz,F. D.Electrochim. Acta, 1976, 21, 967.
upor downatan adjustable rate bymeans ofan oleopneumatic
pistonwhich ensured the complete absence of vibrations. A pressuregauge connected to theoleopneumatic systempermitted
us to reproduce theoptimum rate (1.5 mm/s). The standwas
connected to a digital position sensor, which permitted us to estimatea0.1mm verticalshift ofthestand. Bythissystem the vertical position oftheelectrode, andhencethe meniscusofthe
hangingsolution, couldbereproducedwith a highaccuracy.
Thecellwas water-jacketedandthermostatedat25 ± 0.2 °C.
A gold coilwas used as a counter electrode and an external
saturatedcalomel electrode (SCE) servedasthe reference. The
four-electrode potentiostaticsystembyHerrmannetal.18was
employed to minimize the noise. The wholly computerized instrumentation fordifferentialcapacity and charge
measure-mentsisdescribedinthe next section; positive feedbackcircuitry
was utilizedto compensateforthe electrolyte resistance. Before carryingout adsorption measurements, the state of the electrode surfaceincontactwithan aqueoussolution of0.05
MKCIO4,previouslypurgedwith argonforabout30min,was checkedbyrecording thecurve ofthedifferentialcapacityvsthe
appliedpotentialat20 Hzofacmodulation. Measurementswere
carriedoutonlyuponascertainingthequantitativeagreement ofthesecurves withthese reported in the literature.19 Each set
ofdifferential capacity-potential curves and charge
measure-mentsatdifferentalcoholconcentrationswas carriedoutwhile
the singlecrystalfacewas maintained inconstant contactwith the solution through the meniscus. To this end the alcohol concentrationin thecellwas increased progressively by adding
deaerated alcohol froma Hamilton microlitersyringeintothe previously deaerated aqueoussolutionof0.05 MKC104. The plunger ofthesyringewas fastenedtightlytothe rodofa
digital-display micrometerscrew of0.005mm pitch. Themicrometer
screw was held byamovablestandwhich permittedthe syringe needle tobeloweredintothesolution duringtheadditionand raisedabovethesolutionjust aftertheaddition. Beforeentering thecell,the argonusedforthedeaerationwas bubbledinavessel
containingasolution ofthesame compositionasthat inthe cell;
thisprevented bothachangeinthe alcoholconcentration in the
cell and a decreasein thelevel of the solution surfacewhich wouldhavealtered theshapeofthemeniscus. Aftereachaddition ofthe alcohol, thesolutionwas stirredmildlywith amagnetic
stirrer foratimelong enough toallow its completedissolution
(~20 minfor the concentration atsaturation,~4 x 10-2M). For
an accurate estimate of the stirring time necessary for the dissolution of the alcohol, charge vs potential curves were
previously recorded on a pressurized static mercury drop electrode20at increasingstirringtimesuntilaconstancy in these curves was attained. Argonwas bubbledintothesolutionunder studyforafewminutesaftereachadditionofthealcohol,whereas it was flowedover the solution during itscompletedissolution
in thesolutionandthe subsequent measurements. Aftereach
addition,thedifferentialcapacity and the capacitive chargewere
recordedin successionas a functionofthe appliedpotential.
Results
Differential
Capacity.
Thecurves ofthedifferential capacity(C)againstthe appliedpotential(E)were recordedassuminga simple series RCequivalentcircuit and are
showninFigure 1;an ac signalof20Hz and 5 mVrms was superimposedon thedcvoltageramp. Inagreement
withthe literature19 thecurve ofthesupporting electrolyte
showsa minimum duetothecontributionofthediffuse layerat-0.746 V. Atthe lowchosen KC104concentration
specific adsorption of this 1,1-valent electrolyte in the
proximity ofthe minimum is negligible, and hence the
potentialoftheminimum correspondsto thepotentialof zero charge (pzc).
The lowest concentrations of NHEX cause a modest
depressionofthedifferentialcapacity inthe neighborhood oftheminimum; thisdepressionisclearlynot compensated
(18)Herrmann,C. C.;Perrault,G. G.;Konrad,D.;Pilla,A.A.Bull.
Soc.Chim. Fr. 1972, 12, 4468.
(19)Valette,G.J.Electroanal.Chem. 1989,269,191.
E/V(SCE)
Figure1. Thefirstcurve, as countedfromtop tobottomover
the intermediate potential region, is theCvs Eplot for the
interfacebetween
Ag(lll)
and aqueous 0.05MKCIO4in theabsence of NHEX; the successive curves are relative to
progressiveadditionsof NHEXfrom3.46x 10'3 to3.46x 10“2M by log(c/M)incrementsof0.1.
for by an equivalent increase with respect to the dif-ferentialcapacityofthesupporting electrolyteon thetwo
sidesoftheminimum,although theCvs Ecurves for all the NHEX concentrations exploredmerge with that of
the supporting electrolyteat potentials negativeof -1.3
V and positive of -0.3 V. Although thermodynamic analysis couldbedonefromcapacitance dataextrapolating experimental results to 0Hz, some qualitative findings are drawn fromcapacitancedataobservedfrom5to150 Hz. If we integrate the Cvs E curves (20Hz) starting
from any ofthesetwo extreme potentials, whereNHEX
desorptionis complete,andwe forcetheresultingcurves
ofthechargedensityom vs Etocoincide atone ofthese
twoextremes,theywillnotcoincideon theother extreme. This behaviordenotesan irreversibleadsorption and/or an adsorption—desorptionstepwhichis slowin thetime
scale of(20 Hz)'1. In what follows, evidence will be
providedbychronocoulometrythatirreversible adsorption
ofNHEXmoleculestakesplaceonly at the lowestNHEX
concentrations,whileaslowdesorption ofthesemolecules at positive potentialsoccurs atall NHEXconcentrations
explored.
As the NHEX concentration (c) is progressively
in-creased,thedifferentialcapacityminimumbecomeslower
and broader, while well-defined desorption peaks of
increasingheightdevelopon thetwosidesoftheminimum.
The differential capacity minimum in the presence of NHEXisslightlyshiftedon the negativesidewithrespect
to the pzc in the absence of NHEX. The frequency dispersion is small in the proximity ofthe differential
capacityminimum,whereas it increases notably at the
desorption peaks, as expected. Therefore, the capacity data cannot be usedfor a quantitative investigation of the adsorption of NHEX.
ChargeMeasurements. Capacitivecharge
measure-ments by the chronocoulometric techniquewere carried outbyholding the appliedpotentialatan initial value(E)
atwhich the soluteis adsorbedforatime longenough to attain adsorptionequilibrium;the appliedpotentialwas thenstepped toafinalvalue(Ef)atwhich desorptiontakes placerapidlyand completely,i.e.thecurrent [¿(i)]following
this potential step is zero after a few milliseconds. Electronic integration of this transient yields the cor-respondingcharge [Q(i)]. The experimental strategy in charge measurements is describedthoroughly in refs21
(21)Foresti,M.L;Moncelli,M.R.;Aloisi,G.;Carla’,M. J.Electroanal.
Chem. 1988,255,267.
and22for measurements atamercury drop andinref23
formeasurements on solidelectrodes.
Themethodcan be successfullyappliedif no faradaic
process takes place at theEf,whichisgenerallyan extreme negativeor positive potentialchosenoutsidethe potential rangewherethesubstanceunder studyisadsorbed. Then thecontribution tochargeis purelycapacitive.
It should be noted that no correction for a faradaic
contributioncan beregardedas entirelysatisfactoryif a
disturbing electrode process yields a product which is adsorbedon theelectrode surface, thusinterfering with the adsorption of the substance under investigation.
However, the faradaic complicationcan beovercome not only in thetrivial case in which thedisturbingfaradaic
process is slowin thetimescale adoptedbutalsoin the
more significantcase inwhich theprocess is notdiffusion
controlled and hence gives rise to a constant current density. In thiscase theresultingQ(t) increaseslinearly in time and a linear extrapolation to t = 0 corrects
satisfactorilyforthe faradaic contribution, providedthat
thelatter contributionisnegligibleduringthe very short time
(1-2
ms)requiredfortheimposedpotentialtoswitch from theinitial tothefinalvalue.ForeachNHEXconcentrationinvestigatedtwodifferent chronocoulometric measurements were carried out in
sequence: withboth measurements theinitial potential
(E)was varied between-1.3 and-0.3 Vby20mVsteps,
butthefinal potential(Ef)was setalternativelyequalto
-1.2 and -0.1 V. The time required for attaining
adsorptionequilibrium at£ undermild stirringwas found
tobelessthan10sover the whole rangeofpotentialsand NHEX concentrations investigated. Hence each mea-surementwas carriedoutby
stirring
the solutionat each E value for 10 s, blocking thestirring
2 s before thepotentialstep£ —
Ef,holding thepotentialatEffor100 ms, andthen steppingthepotentialback to thenew initial valueof£. Thiscyclewas repeatedundercomputer control
untilthewhole rangeofEvalueswas covered. The charge
Q(E,t)was sampled at 1 ms intervals startingfrom the
instant t = 0ofeach
potentialstep(E —
Ef). The time
requiredforattainingadsorptionequilibriumismuchless thanthatreportedbyVitanov andPopov12for thesame
system.
By properadjustment ofthepositivefeedbackcircuitry,
the charge Q(E,t) following the potential step E —
Ef
attained a practicallytime-independent value in a few millisecondswhen Efwasset equal to
-0.1V;
thisindicates that no detectable faradaicprocessesoccur at this finalpotential in the 100 ms time window (see Figure 2). Conversely,the Q(E,t)vs tcurves recordedbysettingEf
= —1.2 V exhibit a
slight constant slope due to a mild hydrogen evolution whichoccurs at-1.2 V. Atany rate,
the correction for this faradaic contribution by linear
extrapolation ofQ(E,t) to t = 0is
quite accurateand is entirelyjustified in view ofthelackof hydrogenadsorption on silver. Theextrapolation tot= 0oftheQ(E,t)values
sampledover thetime interval fromt= 50to100ms after
thepotential stepE—·
Efwill be denoted by Q(E—
Ef).
Figure3showsthe Q(E—* Ef)vs Ecurves obtained in
0.05 M KCIO4 by setting Ef equalto —0.1 and —1.2 V,
respectively. Both curves were converted into ctm vs £ curves by settingQ(E— Ef)
equalto zero at the pzc as estimated from the minimum ofthe corresponding
dif-ferentialcapacitycurve. It isapparentthatthetwocurves overlapexactlyover abroadintermediate potentialrange,
whiletheydiffersomewhatattheextreme potentials. In
(22)Foresti,M.L.;Pezzatini,G.;Monteagudo,J.C. G.J.Electroanal.
Chem. 1990,295,251.
Adsorption Behaviorof
Ag(lll)
Figure2. PlotsofAQvs tfor0.02MNHEX in0.05MKC104.
Thecurves refertodifferent initial potentials,E, andone final
potential,Ef= -0.1 V. Dashedcurves referto
potentialsat which hydrogen evolution interferes andwere therefore dis-cardedin our analysis.
E/V(SCE)
Figure3. Plotsofctmvs Efor0.05MKC104intheabsenceof
NHEXobtainedas describedin thetext atan
Ag(lll)
singlecrystal electrodewhichhad beenworking fora long time by
settingEf equal to-0.1 V(dashedcurve) and to—1.2V(solid curve).
particular,the absolute valuesofom are greater for Ef= -0.1
Vthanfor£f=
-1.2V at themore negativepotentials, whereas theyare smaller at themore positivepotentials. This behaviorcan beexplained by considering the wayinwhichtheseOmvalueswere obtained. Whenthe electrode
isheld at themore negativesvaluesfor10s,adetectable amount ofhydrogen is formedat its surface: if we step thepotentialto afinal value Ef= —1.2V,thishydrogen
will remainunaltered,but if we stepthepotentialtoEf
= —0.1 V, it will be
instantaneously reoxidized, thus increasing slightly the positive charge Q(E— -0.1 V)
involved this step with respect to its pure capacitive
contribution. In addition, the high OH™ concentration createdattheelectrodesurfaceentailsan increasein the
measured charge, due toOH™adsorption.24 Inthisrespect theay¡values atthe more negativepotentials are more reliableif theyare obtainedbysettingEf= —1.2V. On
the otherhand,whenthe electrode iskept atmore positive E values for10s,aslightamount of silveroxideisformed on itssurface: if westepthe potential toE¡= -0.1 V,this
oxidewill remain unaltered; in addition,more oxidewill
be formedduringthe holding time at —0.1V (100 ms).
However, if we step thepotential to Ef = —1.2 V, this
oxide will be instantaneouslyreduced,with a resulting
slight increase in the negative charge Q(E— —1.2
V) involved in this stepwith respect to its pure capacitive contribution. In this respectthe ctmvalues at the more
(24)Hamelin,A.;Morin,S.;Lipkowski,J.J.Electroanal.Chem.1991, 304, 195.
Langmuir, 11,No. 2,
positivepotentialsare determined convenientlybysetting
Et=
-0.1 V.Whenstartingfroma freshlyprepared single crystal
electrode or from an electrode entirely restored by
mechanical polishing and annealingofthesinglecrystal face,we haveconstantlyobservedthat thediscrepancies pointedout in Figure3increasewith the repeateduse of theelectrodein adsorptionmeasurements ofthealcohol.
Thus, a freshly prepared electrode or an electrode
em-ployedin onlya fewseriesofadsorptionmeasurements
wouldyieldom vs Ecurves ofthesupporting electrolyte as obtained by thetwodifferentprocedureswhich practi-callycoincidedover the wholepotentialrange from—1.3
to-0.3 V: thetwocurves in Figure3 were indeedobtained atan electrodewhich had been used for several setsof
measurements. The above experimental observations
stronglysuggestthatthe repeateduse ofasilversingle crystalelectrodein measurementsofspecificadsorption increases the density ofsurface defects which are not completelyremovedby the chemicalpolishingwith CrOg and which constitute preferential sites for hydrogen
evolution and surface oxide formation. Therefore, the experimental data reported herein where obtained on
electrodeswhichwere either freshly preparedor hadbeen employedinno more thanthreeor four seriesofadsorption
measurements.
AnincreaseintheNHEX concentrationcauses amodest butappreciableincreasein therate ofhyrogenevolution at E values negativeofca. —1.15V, as indicatedbythe
increaseintheslopeofQ(t)vs t curves obtained by stepping
to thesepotentials. Agradualincreasein c also causes an incipientincrease andthena decreasein the rate of oxideformation atEvalues positiveofca. —0.4V. Thus, thechargedensitiesQ(E—· -12 V) obtainedatthelatter
E values first increase slightly over the corresponding
charge values for the supporting electrolyte alone and
thendecrease: thisbehaviordenotesan increaseandthen adecreasein the faradaiccontributionto Q(E— —1.2
V)
due to surface oxide formation. Since at potentials
negativeof—1.15Vandpositiveof—0.4Vthedifferential
capacitycurves of Figure 1pointtoan almost complete desorptionofNHEX, theaboveeffectsare probablytobe ascribed to the catalytic action of alcohol molecules irreversibly adsorbedon the surface defects existingon the electrode surface. This hypothesis is supported by
the observation that the lowest NHEX concentrations
depressthe differentialcapacitycurves in the region of theminimum (seeFigure 1),while theydonot increase
themon thetwo sidesoftheminimum: this suggeststhat
the lowest NHEX concentrations tend to saturate the
surface defects almost completely, while adsorption on the terracesisstill quitelow. Furtherevidencein favor
ofthisinterpretationisprovided by theeffectofthefirst
NHEXadditions, performed by regularlogarithmic incre-ments logc ^ 0.1,on thecurves oftheabsolute charge
density(om)against E andon the correspondingcurves
oftheinterfacialtension(y)againstE: thefirst addition of NHEX causes a change in um (see Figure 4) and a
depressionin ywhichare greaterthanthose producedby thesubsequentaddition. Thishasledus to discardthe
ctm vs E curve of the supporting electrolyte in the thermodynamic analysis of the experimental data.
We already stated that the charge densities
Q(E— —1.2V)are not
complicatedbyhydrogenevolution
at extreme negative potentials, whereas the charge
densitiesQ(E— —0.1
V)are not complicatedby surface
oxideformation at extreme positivepotentials. In view ofthese observatons,the following strategywas adopted
to obtain reliable vs Ecurves at the differentNHEX concentrations. The two curves ofQ(E— —0.1
Figure4. In passingfrom the steepest totheflattest curve over theintermediate potentialregion,thefirstcurve istheau vs E plot for the interfacebetween
Ag(lll)
and aqueous 0.05 MKCIO4in theabsenceof NHEX; the successivecurves arerelativeto progressiveadditions of NHEX from4.35 x 10-3to 3.46 x 10~2Mby log(c/M)increments of0.1.
and ofQ(E—* -1.2 V)vs E relativeto thesame c value
were first shifted along theverticalaxiswith respectto
eachotherso astoput them incoincidenceover thebroad
intermediatepotentialrange fromca. —1.0toca. —0.6V.
Atpotentials negativeofthis range,where thetwocurves
diverge,only theQ(£—· —1.2
V)vs Ecurve was retained;
conversely, at potentials positive of this rangeonly the
Q(E— —0.3
V)vs Ecurve was retained. The Q(E)vs E curve for the supporting electrolytesoobtainedwas then convertedinto the corresponding curve ofthe absolute
chargedensityom vs EbysettingQ(E)equaltozero atthe
pzc,asestimated from theminimum inthe corresponding
differential capacitycurve. All other Q(£) vs E curves
relativeto thevariousNHEXconcentrationswere finally
made to coincidewith that ofthe supporting electrolyte at themore negativepotentials,whereNHEXiscompletely
desorbed. Figure4shows awholesetofum vs E curves obtainedbythisprocedure. Allthesecurves intersectat
asingle point,whosecoordinateslocatethecharge,om=
—1.65
µ
cm-2, and the potential, Em = —0.788V, of maximum adsorption. Unfortunately,even byfollowing
this procedure, thecurves do not merge at —0.3 V, but rathertheyrun almostparallelover theextremepositive potentialrange explored. Thisagreeswiththedifferential
capacity curves of Figure 1 coinciding over the same
potentialrange. This behaviorisprobably ascribable to a slowdesorption in the time scaleof100 ms as we step
to the positivefinal potential Ef= —0.1V.
Incidentally,
aslow desorptionatpotentials positivetothe adsorption regionhas beenreportedforethyl etheron
Ag(lll)
andAg(100)byHinnenet al.11 andfor ferf-pentylalcoholon Au(111)byRicheretal.9 Sincethe vs £curves obtained
with E{= —1.2
V, at which potential desorptionis fast, are almost exactly superimposableon thecurves obtained
with
£f
= -0.1 Vover the potential range fromca. -1.0to ca. —0.6 V, we are led to conclude that the rate of desorptionfollowingapotentialstep£— —0.1Vdecreases
progressively andrapidly with a decreaseinthewidthof
this step. For the above reasons, results obtained at
potentials positiveto-0.6 V willnotbetaken intoaccount
in the further datatreatment.
Thermodynamic Analysis.
The thermodynamicanalysis oftheexperimentalcurves ofomvs £was carried
out by an entirely numerical procedure25 using molar concentrations in place of activities; such a tactic is expected to produce a practicallynegligibleerror at the (25)Carla’,M;Alois!,G.;Foresti, M.L;Guidelli,R.J.Electroanal.
Chem. 1986, 197, 123.
relativelylowNHEXconcentrationsinvestigated.26 The
surface excesses F(£,c), were obtained by numerical differentiationwithrespect to Incoftheinterfacialtension
valuesy(£,c). Thesecurves were obtained byintegration
withrespect to£ ofthedirectlymeasuredom(£,c) values. Incalculating the isothermsat constant charge, Parsons’
function
( , )
=+ ay¡Ewas first obtained. The surface excesses T(um,c) were then calculated by numerical
differentiation of|(ctm,c)with respect toInc.
Theparameters forNHEXadsorptionon
Ag(lll)
fromaqueous 0.05MKC104are summarized in Table1,where
they are compared with the parameters for NHEX adsorptionon Hg fromaqueous0.1MNaF.27 The charge,
am, andthe potential,£m,ofmaximum adsorptionwere determinedfrom thecoordinatesofthecommon intersec-tion point ofthe om vs £ plots. The maximum surface concentration, Fm, was determinedfrom the adsorption isotherm at£m, whichshowsadistinct
limiting
value(seeFigure 5). The differential capacity Ci at maximum
coveragewas obtainedfrom the interceptofplotsofthe differentialcapacityCat£ = £m
against on the = Fm
axis, as describedin ref28. Thepotential ofzero charge
at maximum coveragewas obtainedbydetermining the interceptson thectm= 0 axisofthelinearsectionsofthe
vs £ plots about £ = £m at alcohol concentrations corresponding to =
I7Tm values greater than 0.5, by plottingtheseintercepts against andbymeasuring the
interceptoftheresultingploton the 0=1 axis. Theshift
£n ofthepotentialofzero chargewhen passing from =
0 to 1 was then obtained by subtracting the potential
charge at maximum coverage so obtained from the
potential ofzero charge,-0.746 V/SCE,in theabsenceof the surfactant.
The standard Gibbs energy of adsorption at zero
coverage,AG°ads,as a functionoftheappliedpotential£
was obtained by
fitting
the experimental isotherms at constant potentialto the Frumkin isotherm:553 expl~
-RT)=
expfoil1 (1)whereaistheFrumkininteractionfactor. ValuesofAG°ads anda were thus obtained from the intercepts andslopes
oftheleast-squaresfittingstoa straight lineof
ln[c(l
—0)/0]vs plots at constant £, over the range from0.2 to0.7. Figure6showsaplotofAG°adS vs £,whereasthe
AG°adsanda valuesat £ = £mare
reported in Table 1.
This table also reports values ofthe coefficients fcE =
3(AG°ads/RT)/d(E
-£m)2 for the quadraticdependenceof
the standardGibbsenergyofadsorption upon(£
-£m),
the superscripts + and —
denoting whether the
least-squares
fitting
toastraight linewas made atnegativeor positive valuesof(£—£m). This
fitting
was confinedto thelower(£-£m)2values,whenever appreciable
devia-tions fromlinearbehaviorwere observed athighervalues.
Theadsorption isothermat constant charge obtained
at = am is
practicallyidentical tothe corresponding
isothermat constantpotentialobtainedat£ = £m. This
is the obvious consequence of the independence ofthe
intersection pointoftheomvs £ curves upon thealcohol concentration, which causes the potential £ to remain
practicallyconstant at£mwhenauiskeptconstant atom. (26)Moncelli, M.R.;Foresti, M.L;Guidelli,R.J.Electroanal.Chem. 1990,295, 225.
(27)Foresti, M.L;Moncelli, M.R.J.Electroanal.Chem.1993,348,
429.
(28)Moncelli, M.R.;Foresti, M.L.J.Electroanal.Chem. 1993, 359, 293.
Adsorption Behaviorof
Ag(lll)
Table 1. Parameters for NHEXAdsorptionon
Langmuir, Vol. 11, No. 2,
Ag(lll)
from Aqueous0.05MKCIO4and on Hg from Aqueous0.1M NaF parameterAg(lll)
(this work) Hg27 parameterAg(lll)
(this work) Hg2710"10rm/mol cm"2 2 3.4 (AG°ada)min/kJmol"1 -18.4° -23.8“
A/nm2 0.49 GE(E—
Em) ~ -2.9 -2.6
am/pCcm"2 -1.65 -1.9 ¿>E+/V"2 35 10.8
EJV(SCE) -0.788 -0.520 Z>e"/V"2 28 8.1
Ci/µ cm"2 14 6.2 6t7+/cm4pC~2 0.037 0.029
Fn/V 0.060 0.185 ba~!cm4 pC~2 0.023 0.021
0Thestandardstate for the solventisthe pure solvent, and that for the soluteisthe hypothetical ideal puresubstancein bothbulk
andadsorbedstates.
Figure5. Plotsof vs logcatconstant ErelativetoNHEX adsorptiononAg(111)fromaqueous 0.05MKCIO4 forEvalues equal to-0.780 (a),-0.860 (b),-0.700 (c),-0.660 (d),-0.940
(e), -0.620 (f), -0.980 (g),and -1.020 V(h).
Figure7. Plotsof vs logcat constant ctmrelativetoNHEX
adsorptionon Ag( 111)from aqueous 0.05MKCIO4 foromvalues
equal to-2 (a),0(b), -4 (c),1 (d),2 (e), -6 (f), 3(g),-7 (h),
—8(i), and -10
µ
cm 2 (j). 1 -15 -i -19 ¡-1-1- -!-1--1 -0.8 -0.6 E/V(SCE)Figure6. PlotofAG°adsvs EforNHEX adsorptionon
Ag(lll)
fromaqueous 0.05MKCIO4.
TheAG°adsanda values atthe constantcharge =
om as obtained from the interceptand slope ofthe
ln[c(l
)/ ]
vs plotat om are thereforepractically identicaltothe corresponding values at the constant potentialE =
Em. However,as we depart from thechargeofmaximum
adsorption, am, the vs logc plots at constant charge densitytendtoan apparentlimitingvaluewhichdecreases progressivelywithincreasing|ctm
—
om|,as shownin Figure
7. This behavioroftheadsorption isotherms at constant chargeas compared tothat oftheisotherms at constant
potential is observedforpractically all simple aliphatic compounds (see, for instance, the vs log c curves at
constantsandamfor n-pentanol adsorption fromaqueous
0.5 M Na2SC>4 in Figures 3 and 4 of ref 26). As a
consequenceof this behavior, the
ln[c(l
—)/ ]
vs plotsat constantchargeas obtained bysetting = r/Tm,with
Fmderived from the adsorption isothermatam= am,show increasing deviations from linear behavioranda
progres-sive decreaseinthe averagevalue of their slopewithan
increasein |ctm —
am|. Due to theselarge deviations,no
attemptwas madetoestimateFrumkininteractionfactors
aatconstant chargeforchargesam ^ am. Moreover,AG°ads
values at constantamwere obtained bylinear extrapolation
Figure8. PlotofAG°adsvs amforNHEX adsorption on
Ag-(111) from aqueous 0.05 M KCIO4. The dashed curve was
calculatedas describedinthe text. to = 0ofthelower portionof ln[c(l
)/ ]
vs plotsatconstantcharge, i.e.the portion from = 0.1to0.4. Figure
8 showsaplotofAG°adsvs crM. Table1 reports valuesof
the coefficients ba = d(AG°a¿s/RT)/d(am
-am)2 for the quadratic dependence ofthe standard Gibbs energy of adsorptionupon(am~
om),where the superscripts+ and
— denote
whether the least-squares
fitting
toa straightlinewas madeat negativeor positive valuesof
(
-am).
Theratios (6±e/5±0)1/2are in the range from 31to35µ¥ cm-2andare thereforerelativelyclosetothedifferential
capacity Co ofthe Ag(111)/electrolyte interface in the absenceofsurfactants, inaccordancewith predictions.26
Discussion
TheTmvalueforNHEXon
Ag(lll)
isappreciablylessthanthe value, «3.95 x 10~10molcm"2,correspondingto full coverageofNHEX, calculatedformoleculesin a flat
orientation, area = 0.42 nm2, on the basis ofa
space-filling
model.26 In contrast, the experimentalTm value forNHEXadsorptionon mercury as reported in Table1Table2. Standard GibbsEnergyofAdsorption (kj mol-1) ofNHEX at Metal/Water Interfaces at the
Potential of Maximum Adsorption
Hg Bi Sn In(Ga) Cd
Ag(lll)
Zn Ga-23.8 -21.2 -21.1 -20.2 -19.4 -18.4 -18.3 -17.6
water moleculeswhichremainadsorbedon silverwhen
reachesits
limiting
value points toahigherhydrophi-licity
of this metal with respect to mercury. Furtherevidence in favor of this conclusion is provided by the muchmore negative standardGibbsenergy ofadsorption
ofNHEXon Hg, -23.8 kJ mol-1.27 As a matter offact, the standard Gibbs energy ofadsorption, AG°adS, of a
neutral compoundwhich interacts weakly with metals can beusedas ameasure oftheirhydrophilicity,sincethe main difference in AG°adswhen passing fromone metal
toanothercan thenbe ascribedto the difference in the energyrequiredtodesorbthe watermoleculeswhich must makeroom for the adsorbingcompound. Table2 sum-marizesAG°adSvaluesforNHEXadsorptionfrom aqueous
solutions on various sp metals as well as on
Ag(lll).
Taking intoaccountthatanincrementinthehydrocarbon chain length ofre-aliphatic alcoholsbyone-CH2-
groupis accompaniedbya roughlyconstant increment in the negative valueofAG°ads,theAG°adsvaluesforNHEXon
spmetalsinTable2were obtainedbylinearextrapolation
ofthecorresponding values for re-propanol, re-butanol, and re-pentanolavailable intheliterature (seeref 29forGa, ref27 for Hg, andref30forall othersp metals). It can
be seen that, accordingto this hydrophilicity scale,
Ag-il
11)isnearlyashydrophilicas gallium,andhenceturns out to be one ofthe most hydrophilic metals. A more detailed discussion is given in ref31 for the (100) and (110)silver faces.Thedifference inbehavioroftheadsorption isotherms at constant chargein Figure 7 with respectto those at
constant potential in Figure5agreeswiththe predictions of a three-dimensional lattice model of TIP4P water molecules andofpolar dimeric solutemolecules against
a charged wall, recently developed in our laboratory.32
The model, which does not make use of molecules, represents solute molecules as dimers consisting ofa nonpolarsegment,which simulates avery short hydro-carbon chain, and ofa polar segment,which simulates the polar head. This model predicts that the surface
concentration,
,
at constant charge increaseswithlogc, tendingtoan ill-defined apparentlimiting
valuelessthanTm, whichislower the more the chargedeparts from its value ofmaximum adsorption, am. This behavior32 is explainedby the factthatthe localelectricfield anchoring the water molecules totheelectrode surface consists of
theexternal field 4 plusthe additional polarization field
created by the neighboring water molecules,whichacts
in the directionopposite totheexternal electric field.As
the surface coverage increases, the polarization field decreasesaccordingly,andhencethe localfieldatconstant
charge increases in absolute value, approaching 4 . Hence,theresidualadsorbed water moleculeswillbeheld in the adsorbed state by the localfieldmore
tightly
the higheris6;this explains theattainment ofan apparentlimiting
valueofthe surface coverage,albeit ill-defined,whichdecreaseswith increasing| |. No suchabehavior
ispredicted when it isthe applied potential E whichis
(29)Pezzatini, G.; Moncelli, M. R.; Innocent!, M.; Guidelli, R. J. Electroanal.Chem. 1990,295,275.
(30)Trasatti,S.InModernAspectsofElectrochemistry, Conway, B.
E.;Bockris,J.O’M.,Eds.;PlenumPress: New York,1979;Vol13,p81.
(31)Foresti, M. L.; Innocenti, M.;Guidelli,R.J. Electroanal.Chem. 1994, 376, 85.
(32)Guidelli,R.;Aloisi,G.J.Electroanal.Chem. 1992, 329, 39.
heldconstant. Infact,theimposedconstancy ofErequires agradual decreasein theabsolutevalueoftheexternal electric field4 withincreasing ,inordertocompensate for the gradual decrease in the absolute value ofthe counterfieldproducedby theadsorbedwater molecules. Therefore, at constant E the progressive increase in
doesnotcause the residualadsorbedwater molecules to
bemore stronglyanchored to the surface as in the case
ofconstantom, andhenceFmturns outtobepractically
independentofE. Thepenalty whichone hasto pay in using theaboverefined three-dimensionallatticemodel,
whichdoes notmakeuse ofadjustableparameters,32 is the difficulty of predicting the adsorption behavior of
re-aliphaticcompoundshavingalong hydrocarbon chain,
due to the notable increase in the complexity ofthe
statistical mechanical treatment ofthe model with an
increaseinthenumberofmonomeric segments composing
the solute molecule.
For a quantitativecomparisonbetweenexperimental behavior and modelistic predictions, we shall adopt anothermodel ofadsorbed water monolayer and solute
moleculesin whichthe water moleculesinteract witheach
othervia H-bondsand dipole—dipoleforces,whereas solute molecules interact with each other and with water
molecules via dipole—dipole forces only.33 This model makesuse oftwoadjustableparameterstofitexperimental
results, namely the number(re) ofadsorbedwater
mol-ecules displaced by one adsorbing solute molecule and
thenormal component(mn) ofthe dipole momentofthe
adsorbed solute molecules. Theunknownvaluesofreand
µ are obtainedfrom the experimental valuesofamand
ba~bymakinguse ofacalculatedplotoftheµ /reratiovs omtoobtain
µ /
andthenofcalculated plotsofµ /revs ba~ at differentrevaluestoobtainre.26-34 Inthe presentcase (NHEX) the estimated valueofthe
µ^/
ratioisclose to -0.3 D. Confining ourselves to re values which aremultiples of0.5,theresultingrevalueequals4,andhence
coincideswiththe valueestimatedforNHEXadsorption
on Hg26 on the basis ofthe same model. This re value
compliessatisfactorily with the surface area, 0.42 nm2,
estimatedforaNHEXmoleculeinaflat orientationwhen we considerthatthearea projected byone adsorbedwater
moleculeis closeto 0.10 nm2. Thisresult supports the
point ofviewthat theNHEXmolecules are adsorbedin a flat orientationon
Ag(lll)
aswellas on Hg,atleast at low surface coverages; moreover, their dipole momentnormalcomponent(mn)isapproximatelythesame on both
electrodematerials. Thenegativevalue
ofµ
impliesthattheNHEXmoleculeis adsorbedwiththe positiveendof
itsdipoletowardthe metal;thisjustifies thepositive value of60mV fortheshiftEnofthe pzcwhen passing from = 0to1,alsoinviewofthefactthatthewater dipolesare
expected toturn theirnegative endtoward the metal in the proximity ofthepzc.
Themodel adopted herein forthe estimate of theµ andrevalues33predictsan adsorption behavior at constant
chargewhichisinfairlygoodagreementwith theFrumkin isothermofeq1,atleast for < 0.7. Hence,
fitting ln[c(l
—
)/ ]
vs plots calculated from the modelto a straightlineover the range from0to0.7yieldsan intercepton the6= 0axiswhichmeasures the standardGibbsenergy
ofadsorption at zero coverage, AG°adS, apart from an additive constant;moreover, theslopeofthisstraightline
providesavalue for theinteractionfactorotobecompared
with the experimentalone. The dashedcurve in Figure
8 isaplot ofAG°adsagainsta calculated from themodel
forµ /re= —0.3Dandre= 4;this curve was shiftedalong (33)Guidelli,R.;Foresti, M. L.J.Electroanal.Chem.1986,197,103.
(34)Moncelli, M.R.;Pezzatini,G.;Guidelli,R.J.Electroanal.Chem. 1989, 272, 217.
Adsorption Behaviorof
Ag(lll)
Langmuir, 11, 2, 1995theverticalaxisso as toobtainthebestoverlapwiththe correspondingexperimentalcurve. Theavaluecalculated
from the model,
-3.7,
is somewhat greater than the corresponding experimental value.Thedifferentialcapacity Ci at maximumcoverage on the
Ag(lll)
face, 14pF cm™2, is relatively high whencomparedwith the corresponding valueon mercury, 6.2 pF cm™2.27 This higher C\ may be due partly to the
presenceofsurface defectswhichincreasethe realarea
oftheelectrode surfacewithrespect to thegeometricarea. However, theratio ofthesetwo areas is expected tobe hardlygreater than l.l.35 Afurthercause oftheincrease
ofCi in passing from Hg to Agis due tothefactthatthe
water content intheadsorbedfilmatmaximumcoverage isgreateron Agthanon Hg. Thus, thepresenceofwater
in the adsorbedmonomolecular film may increase the
distortional dielectric constant (e) ofthefilm36·37and/or
decreaseitsaveragethickness(ó):22·38botheffectstendto
increase Ci in view ofthe well-known expression C =
e/(4jrd) for the capacity of a parallel-plate capacitor.
AnotherfactorforCi being greateron Ag thanon Hg may
beascribed to the factthatthe negativecontribution, CefS from electron spillover to the reciprocal (Co™1) of the
differentialcapacityat zero coverage isprobably greater for Ag than for Hg. Roughly speaking, this negative
contributionisgreater thelargerthepolarizabilityofthe
“electronictail”,whichinturn isexpected to increasewith
an increase in the bulk density offree electrons. The
problem of the effective density of free electrons is
complicated in the case of sd metals, because the d
electrons in these metals can neither be considered as
freenor asbound. SchmicklerandHenderson39estimated (35)Foresti, M.L;Guidelli,R.;Hamelin, A.J.Electroanal.Chem. 1993, 346, 73.
(36)Damaskin,B. B.;Survila, A. A.; Rybalka, L.E.Elektrokhimiya,
1967, 3, 146.
(37)Schumann,D.J.Electroanal.Chem. 1986,201,247. (38)Carla’, M.;Aloisi,G.;Moncelli, M.R.;Foresti, M. L.J.Colloid InterfaceSci. 1989, 132, 72.
the electron densityofAgtobemuchcloser to thehigh
value for Ga than to the relatively low value for Hg.
However,it must beconsideredthattherelativeweight oftheCei”1contributiontoCr1iscertainlylessthanthat
to Co™1,since
Cr1
> Co”1.Sofardiscussionwas basedon themodel describedby
Guidelli andco-workers.32·33
Recentexperimentalfindings about orderingofwater
at the
Ag(lll)/NaF
interface40byin-situ SXRS (surfaceX-rays scattering)suggestnot only reorientation ofwater dipoleswhen the signofthechargeon the metalchanges
butalsoachangeofareal densityinthefirst layer from
1.1 at -10 pC cm”2 to 1.8 at 25 pC cm™2. Our base
electrolytewas KCIO4, notNaF, thereforethesefindings cannotbedirectlyintroduced inour results;furthermore,
it is difficult to foresee what would be the influence of NHEXadsorptionon theselargefirst-layerwater
densi-ties. These SXRSobservationsdemonstratethat thereis
no changein the atomicsurface structure
ofAg(lll)
from—10to25 pCcm”2, i.e.no reconstruction,butonlyasmall contraction between the top and second layer of silver atoms at the more positive charges. Therefore, the
potentialstepsare performedallalongour experimenton the same superficial atomic structure, it isnot the case for faces which reconstruct, low-index gold faces for instance.
Acknowledgment.
The authors wish to thank Mr. FerdinandoCapolupoand Mr.AndreaPozziforvaluabletechnicalassistance. Thefinancial support oftheMURST
and ofthe
Italian
CNRisgratefullyacknowledged. LA9403789(39)Schmickler, W.; Henderson,D.J.Chem. Phys. 1986, 85, 1650. (40) Toney,M.F.;Howard, J. N.; Richer,J.; Borges, G.L;Gordon,
J. G.;Melroy,O. R.;Wiesler, D. G.; Yee, D.; Sorensen,L.B.Nature