The Electrochemistry of Gold:I The Redox Behaviour of the Metal in

Aqueous Media

L D Burke and P F NugentChemistry Department, University College Cork, Ireland

Gold is frequently regarded as the ideal metal for the investigation of solid electrode behaviour, which inaqueous media is often considered in very simplistic terms as being that of a metal which is highly resistantto dissolution. Gold possesses very weak chemisorbing properties and an extensive double layer region thatin the presence of most pure electrolytes is often assumed to be totally free of Faradaic behaviour, andexhibits a monolayer (or AUZ03) oxide formation/removal reaction at quite positive potentials. However,recent investigations have revealed that the electrochemistry of polycrystalline gold in aqueous solution isconsiderably more complex. Two significantly different types of oxide deposits, monolayer (or a) andhydrous (or ~), may be produced on the metal and the behaviour of the ~-deposit is quite unusual. It issuggested that not only the behaviour of the ~-oxide but, far more important from a practical viewpoint, thecatalytic and electrocatalytic behaviour of gold (which will be discussed in more detail in Part II)' may berationalized in terms of the active state (or states) of gold. This active state (frequently present only at verylow coverage) reacts in a manner that is quite different from that of stable gold. The nature of the active stateof gold deserves far more attention than it has received to date.

The importance of surface reactions was highlighted ina recent article by Sachtler (l) who pointed our that ca17% of GNP in the US and ca 25% in the case ofGermany are derived from materials produced bycatalytic processes. Most of these processes involveheterogeneous catalysis of gas phase reactions on metalor oxide surfaces. The catalysis of electrode reactions isalso important, egca 13 million tons (valued at ca 2.2billion dollars) of chlorine gas (2) is produced annuallyin the US, largely with the aid of electro catalyticallyactive dimensionally stable (Ru02/Ti02)/Ti) anodes.

The potential in the electro catalysis field is alsoquite significant; for example, the development ofeffective electrocatalysts for the direct methanol/air fuelcell would revolutionise the transport industry.Compared with the internal combustion engine,widely used at present, fuel cells in general offer theprospect of more efficient, cleaner and less noisyenergy conversion. Methanol is a much more easilystored fuel than hydrogen for use in mobile energyconversion systems and may, with the development of

CS:9' GoldBulletin 1997, 30(2)

more efficient methanol oxidation electrocatalysts, bethe system of choice for operation at ambient, orslightly elevated, temperatures in vehicles.

It is generally accepted that reactions at surfaces, egheterogeneous catalysis, electrocatalysis and corrosion,occur predominantly at active sites (3) and in view ofthe importance of such reactions one could regard suchsites as among the most important entities in chemistry.Yet they are rarely mentioned in Physical Chemistrytextbooks, and electrocatalysis is rarely discussed indepth, even in many of the recently publishedspecialized textbooks in the field of electrochemistry.The basic problem in the area seems to be that a clearunderstanding of the nature, and especially the mode ofoperation, of such sites is not available (4).

There is an additional problem in theelectrocaralysis area in that different modes ofoperation are assumed to be involved in different typesof electrocatalytic processes, viz:(i) Noble metal electrocatalysis is assumed to involve

activated chemisorption (5), viz:

43

HCOOH------+H"""COOH = 2H+ + CO2 + Ze (1)

It is generally assumed that in this case the surfacemetal atoms (e) do not undergo a redox transition.

(ii) Electrocatalysis by solution species is generallyassumed to involve a cyclic redox process (6), viz:

(2)

P is the oxidation product of solution species R(a reductant) and the dissolved oxidation mediator,Mq-n)+, is repeatedly regenerated at the anodesurface. A similar scheme may be written for thereduction of a dissolved oxidant (0) in whichcase the mediator is the reduced form, M:; ,of the redox couple.

(iii) Electrocaralysis by surface-bonded mediators, egredox groups or enzymes, is assumed to occur in asimilar manner to that outlined in (ii) except thatin this case the redox couple remains bonded tothe electrode surface (7). An in teresting variationof this mode of e1ectrocatalysis is found in thecase of RuOrbased anodes (8) where themediator is a ruthenate or perruthenate speciesderived from the surface of the electrode material.

Recent work on noble metal electrochemistryprovides the basis of a more unified approach toelectrocatalysis. It was demonstrated (9-11) that inmany instances the cyclic redox or mediator mode ofclectrocatalysis applies also to noble metal electrodeprocesses. The importance of adsorption is notdiscounted; it is the major (indeed sole) factor in someimportant processes, eg H 2 evolution or ionization,weakening of C-H bonds in methanol, etc. However,the mediator route for electrocatalysis seems to bemore widespread; apart from providing a commonbasis for the different types of electrocatalysis itcorrelates activity for oxidation and reduction for thesame electrode system, it applies to both metals andoxides, and rationalizes behaviour in certain importantprocesses, eg removal of inhibiting COads species,which is a process of considerable practical importance- especially in the fuel cell anode area.

The new approach to noble metal electrocatalysis,sometimes referred to as the IHOAM (IncipientHydrous Oxide/Adatom Mediator) model, IScontroversial as:(i) It stresses the importance of the defect state of

surfaces and assumes that, for the same surface,two significantly different types of electrochemicalresponses may be observed - one for high

44

coordination and the other for low coordinationsurface metal atoms.

(ii) Considerable stress is placed in the new approachon two largely ignored features of noble metalelectrochemistry, namely hydrous oxides andactive states of the metal.

(iii) In the conventional view of the metal/solutionboundary or double layer the surface or boundaryis frequently represented as an inert flat planewhereas in the IHOAM approach the metalatoms at the surface are regarded as having adegree of mobility and a range of chemicalreactivities; some of these atoms are assumed toundergo electro-oxidation reactions in aqueousmedia at potentials that are significantly morenegative than that normally associated with theonset of regular monolayer oxide formation.

In this, the first of two articles (the second will dealspecifically with electrocatalytic processes on gold), anaccount will be given of the electrochemistry ofpolycrystalline gold in aqueous media. Since theconventional chemical behaviour of this electrodesystem has been described earlier by various authors(12-14), attention will be focused here largely on morerecent work in which anomalous behaviour wasobserved. In the present context the behaviour of ametal or oxide is regarded as anomalous if it deviatessignificantly from that expected from conventionalthermodynamic data as summarized, for instance, byPourbaix (15).

THE CONVENTIONAL BEHAVIOUROF GOLD

Gold is the noblest of metals. According to thethermodynamic data for this element (15) gold shouldbe oxidized to the trivalent state, according to thereaction:

(3)

at 1.457V if the product is the hydrated oxide (whichmay be represented as Au(OHh or AU203.3H20), orat 1.511 V if the product is the anhydrous oxide. Thepotentials quoted here are standard reversible (EO)values, calculated using standard chemical potential(u0) data, and the aqueous solution is assumed to befree of complexing agents. Most potential values arequoted in this review in terms of the RHE scale, iewith respect to a reversible hydrogen electrode (PH2 =

@ Gold Bulletin 1997, 30(2)

1.0atm) in the same solution. This is a convenient scalefor use with metal/metal oxide electrodes as, in termsof the conventional behaviour of such systems, theelectrode potential value for a given transition shouldbe pH-independent (the pH-dependence of theworking, or oxide, electrode being the same as that ofthe reference electrode). The fact that significantlydifferent E/pH variation is observed, in terms of thisRHE scale, for certain oxide/metal (and indeedoxide/oxide) transitions was discussed earlier (10, 16).

A typical cyclic voltammogram* recorded for goldin acid is shown in Figure 1. The most notable featurein the positive sweep is the increase in anodic current,

0.30

commencing at ca 1.36V, which is due to monolayer(or o-) oxide formation at the electrode surface. Thisprocess commences at a slightly lower potential thanexpected on the basis of the EO values quoted hereearlier. However, it must be borne in mind that thelatter were estimated for the stable bulk metal atomsand it is only for the latter that f-LO(Au) = O. Surfaceatoms are inevitably more active; the absence of similarmetal neighbours on one side means that they arelacking some of the lattice stabilization energy of theirbulk equivalents and, hence, they tend to oxidize at alower potential.

The early stage of oxidation of noble metal surfacesis usually discussed in terms of adsorbed hydroxyradical formation (14), viz:

Figure 1 Tjpical cyclic uoltammogram (0.0 to 1.80 V, 50mV 5'1) recorded for a polycrystalline gold discelectrode in 1.0 mol dm,3 H2S0 4 at 25C

'Cyclic voltammetry is a widely used electrochemical technique hasco onthe usc of a potcntiostat to control the potential (by automatic adjustmentof current flow) of the electrode under investigation. In this technique thepotential impressed on the electrode is varied in a triangular manner at apreselected sweep rate, eg50 mY SI, between preselected upper and lowerlimits, The resulting response is a cyclic volramrnogram, ic a plot of current

versus potential for the forward and reverse sweeps. A large positive

potential corresponds to highly oxidizing conditions at theelectrode/solution interface and this, depending on the nature of the metaland solution pH, may result in dissolution of, ot oxide forrnarion at, theelectrode surface. The technique is normally used 10 study the redoxbehaviour of either solution species or rhe electrode surface; hut the sameprocedure has been used extensively ar Cork 10 produce and invcsrigareunusual oxide deposits on electrode surfaces (see reference 16 for derails)and to modify, or activate, rhe outer layers of metal electrodes. The study ofrhe acrive stares of surfaces is a very challenging area of research but theresults obtained to dare from cyclic volrarnrnerry work are of considerableinterest in areas such as heterogeneous catalysis.

(4)Such an approach may be inadequate in the case

of gold as the hydroxy radical is a very high energyspecies; the EO value for the H 20/OH couple (17) is2.85V(SHE) and gold has an extremely weakchemisorbing capability, eg the coverage of Hads ongold at O.OV is only ca0.03% of a monolayer (12). Ifthe involvement of a hydroxy species is to beretained, it may be more appropriate to consider theinitial oxidation product as some type of metal-hydroxy compound, eg Auo+... OHo-, the productbeing a polar covalently-bonded species. The chargeassociated with further monolayer oxidation does notgenerally give rise to a single sharp peak above 1.36Y.Instead, the charge in question, for gold in acid,tends to be distributed along a plateau (12) (usuallywith some minor features evident on the latter) withno major change until oxygen gas evolutioncommences at ca 2.0V (this latter feature is notshown in Figure 1).

The conditions for complete monolayer oxideformation on gold surfaces are important from a practicalviewpoint as it forms the basis of a useful method fordetermining the true surface area (or roughness value) ofthe metal surface. For this purpose, Woods (12) hasrecommended that the potential of an initially clean goldelectrode in 1.0 mol drrr-' H 2S04 be held initially at1.8V for 100 sec to achieve monolayer oxide coverage.The charge for monolayer oxide reduction, estimatedfollowing the application of a subsequent negative sweep,may be converted to real surface area using therecommended conversion factor of 386 f-LC crn? (realarea). This is a very convenient, though perhaps not veryprecise, method for the determination of a relativelyimportant electrode parameter.

1.60.8 1.2FlV (RHE)

004

0)-0.30

-0.60

0

($9' GoldBulletin 1997, 30(2) 45

(5)

A very common feature of monolayer oxideIormation/rcduction behaviour in the case of the noblemetals is hysteresis, ie the difference in the potentialrange for monolayer oxide formation (positive sweep)as compared with that for monolayer oxide reduction(negative sweep) - see Figure 1. It is generally assumedthat hysteresis in this case is due to post-electrochemical processes, ie the product of a givenelectrochemical reaction undergoes a change after theelectron transfer step such that the species, energiesand potential range involved in going in one directionare appreciably different to those involved when theprocess is reversed.

The origin of hysteresis in the monolayer oxideformation/removal processes is usually attributed (14)to gradual changes in the nature of the oxide film.During the growth process dipolar (Auo+.OHO,) speciesare produced which, at appreciable coverage, repel oneanother. Such repulsion raises the energy required togenerate additional dipoles; hence, there is an increasein potential with increasing coverage. This is evidentlythe reason why the monolayer oxide formationresponse is an extended plateau rather than a sharppeak. The post electrochemical process involved in thiscase is outlined schematically in Figure 2. Rotation ofsome of the surface dipoles, a process also referred to asplace-exchange, relieves much of the lateral repulsionor stress in the surface layer, resulting in a more stablesurface deposit. In the negative sweep the dipolecoverage (and any residual stress) is progressivelyreduced. No electrostatic repulsion barrier is developedin this case and a relative sharp cathodic peak isobserved.

Figure 2 Schematic representation ofthepost-electrochemical, place-exchange reactioninvolved in a monolayer oxideformation. Somesurface metal atoms (unshaded) switch positionswith adsorbedoxygen species (shadedcircles).Obviously ifthese species are charged, asindicated here, the place-exchange reaction leadsto a considerable reduction in electrostaticrepulsion energy

46

There is an additional factor, related to changes inthe activity of the surface metal atoms, that may alsocontribute to the hysteresis effect. This is the change inactivity of surface metal atoms during the course ofreaction. Three types (or states of activity) of goldatoms may be considered, namely:(a) Bulk lattice atoms (AuO) - these have little

relevance to the reaction at the surface;(b) Regular surface atoms (Au") which are more

active than those in the bulk.(c) Displaced surface atoms (Au**) which are in a

state of still higher activity as these are generatedinitially in a state of unusually low latticecoordination number on reduction of thepartially place-exchanged surface oxide.

Atoms of type (c) are assumed to undergospontaneous rapid alteration to the state represen tedhere by (b). The different states in question hereshould not be viewed in a very rigorous manner as, inthe first instance, solid surfaces tend to beheterogeneous (it is more a question of the spread andoccupancy of different energy states).

According to this approach hysteresis may beconsidered in terms of the following scheme, viz:

Au* + H 20 ----- Auo+.OHo, + H+ + e'

1HO'.AuO+

1+ 2H20 - 2H+ - 2e-+3H+ + 3e'Au** + 3H20 .-----Au(OH)"

The two post-electrochemical steps are representedhere by the vertical arrows. The scheme also highlightsanother complication in the case of the positive sweep,namely, the conversion of surface dipoles, Auo+.OHo"to a more regular oxide species, ie formation of speciessuch as Au(OHh, or AU20".

Examples of cyclic voltammograms recorded forgold in base are shown in Figure 3. The onset potentialfor the initiation of monolayer oxide growth in thiscase, ca 1.25Y, is significantly lower than in acid (18).An unusual feature in the response for gold in base isthe large anodic peak in the positive sweep (which hasno cathodic counterpart on the subsequent negativesweep) extending over the range of ca 1.6 to 2.GY - seeFigure 3(b). The process giving rise to this response isassumed to be oxygen gas evolution catalysed in atransient manner by some type of nascent hydrousgold oxide species formed at the monolayeroxide/aqueous solution interface (19). The transitory

@ GoldBulletin 1997,30(2)

nature of the activity in this region was attributed toconversion of the active gold species to a less activedeposit which may well be conventional gold hydrousoxide material. Regular oxygen gas evolution on goldin base (as in acid) occurs vigorously only above ca2.0Y. The behaviour shown here on the negative sweepover the range 2.1 to lo6Y (see Figure 3(b)) may wellbe a better reflection of the steady state oxygen gasevolution activity of gold in base than that shown inthe corresponding region of the positive sweep wherethe influence of transient activity is quite marked.

(6)

namely, (i) the hydrous oxide electrochemistry of gold,and (ii) the premonolayer oxidation of gold.

Multilayer hydrous oxide deposits are readilyproduced on gold in aqueous solution by either dcpolarization (21) at potentials within the oxygen gasevolution region (E 2>. 2.0Y) or by repetitive potentialcycling (22) or pulsing (23). Such growth occurs mostreadily in acid solution and the product obtained is arather amorphous, hydrated Au(III) oxide. As pointedout here earlier, thermodynamic data suggest that thismaterial should be produced at ca 1.46Y; however, atthe latter value most of the gold surface is alreadycoated with a monolayer oxide () film whichevidently passivates the surface with regard to hydrous([3) oxide growth. The a. deposit is assumed to be anadherent layer of compact material through whichAu3+ ion migration is quite slow. Both oxygen gasevolution and hydrous oxide growth may involveformation of an unstable (15) Au(IY) oxide species asoutlined in the following scheme, viz:

1.60.8EN (RI-IE)

o

0.2(3)

0-0.2

':5 0.6 (b)

~0.2

0

-0.2

Figure 3 7ypical cyclic uoltammograms fir apolycrystalline goldelectrode in 1.0 mol dm-3NaOH at 25C: (a) 0.0 to 1.60 Vat 50mVs'j; (b) -0.2 to 2.1 Vat 10 mVs'j

An interesting feature of the negative sweep inFigure 3(b) is the appearance of a second cathodic peakat ca. 0.8SY. The monolayer oxide reduction peak wasobserved in this case at ca 1.07V and the subsequentpeak is assumed to be due to reduction of hydrous goldoxide species formed on the gold surface at the upperend of the cycle. The growth and subsequent reductionof such hydrous oxide deposits were discussed in somedepth in earlier reviews (16, 20).

ANO~OUSBEHA~OUROFGOLD

1 Hydrous OxideIn terms of gross departure from behaviour expectedon the basis of thermodynamic data, two aspects of theelectrochemistry of gold merit special consideration,

According to this scheme some of the monolayeroxide material is converted, at E 2>. 2.0V to goldperoxide species which decomposes to yield oxygengas. This process continues in a cyclic manner,resulting also (possibly as a side reaction) in thegeneration of an increasing thickness of the [3-material.There are other possibilities, eg oxygen evolution onthe o -oxide may result in the generation of flaws in thelatter through which gold cations emerge to form the[3-oxide, ie the growth of the latter may proceedwithout the intervention of any other oxide species.

The use of repetitive potential cycling as atechnique for producing hydrous oxide deposits hasalready been discussed (16). At the upper limit of eachcycle some metal atoms are displaced in accordancewith the scheme shown here in Figure 2. Reduction ofmuch of the o.-film at the lower limit of the negativesweep results in the generation of active (low latticecoordination) surface metal atoms which reactspontaneously with water molecules to yield some [3-material. Important requirements in this case are that(i) the [3-material must be of a low density or porous innature to allow access of water to the surface of the a.-oxide where the hydrous oxide formation is assumed to

C69' Gold Bulletin 1997, 30(2) 47

occur, and (ii) the lower limit of the oxide growth cyclemust be such that much of the -, but little if any ofthe 13-, oxide is reduced.

The responses recorded during the course of anegative sweep for a hydrous oxide-coated goldelectrode are complicated by the fact that thebehaviour observed is strongly influenced by factorssuch as the thickness of the deposit and the pH of thesolution. In most cases the interface in questioncontains both the a- (monolayer) and 13- (multilayer)materials arranged in the following manner: gold/a-oxide/Bvoxide/aqueous solution. Generally the firstcathodic response is that due to reduction of the a-oxide and this usually results in a peak with a potentialmaximum (Ep) in the region of ca 1.0V (the observedvalue varies (24) with factors such as oxide formationpotential, oxide coverage, sweep rate, etc). With thinoxide films in acid solution (21) the reduction of thel3-oxide deposit gives rise to a single peak whichoverlaps with the monolayer oxide reduction peak.Similar behaviour was observed for the reduction ofrelatively thin oxide films on gold in base, except thatin this case a clear separation, in terms of potential, isobserved between the two peaks. The a-oxidereduction occurs again at ca 1.0V (with a minor shiftto lower values of Ep with increasing pH) but themaximum of the l3-oxide reduction peak occurs (21) atca 0.65Y. Such a drop in potential, ca -0.5 (2.3RT/F)V per pH unit, on changing from acid to base, is quitecommon for hydrous oxide systems. Similar behaviour,sometimes referred to as a super-Nernstian shift, hasbeen observed even in the case of highly reversibleoxide/oxide transitions (16), eg those involving Ir, Rh,Fe and Ni ions.

There are complications in the interpretation ofsuper-Nernsrian E/pH shifts in the case of gold in that(i) the structure, composition and state of charge of theoxyspecies are unknown, and (ii) the overall reductionstep occurs in an irreversible manner. It has beensuggested (24) that the behaviour in question may beinterpreted in terms of the following scheme (thisdescribes the reaction in terms of a monomer unit ofthe oxide polymer), viz:

AUz(OH)3-9 + 9H+ + 6e- = 2Au** + 9HzO (fast) (7)

2Au** -- 2Au* (8)

The first step is assumed to be rapid and reversible andthis process may be treated in a Nernsrian manner.Application of the Nernst equation to Equation 7yields:

48

o 2.3Rl' 2.3RT (2.3Rl')E=E +~log aor 3F log aAu,,-1.5 -p- pH

(9)(aox represents the activity of the oxyspecies).According to the latter equation the equilibriumpotential for the process shown in Equation 7 shoulddecrease by ca -1.5 (2.3RT/F) V/pH unit (pH-independent scale) or ca -0.5(2.3RT/F) V/pH unit(RHE scale). An obvious objection to this approach isthat the reduced form of the couple (Au**) is not atconstant activity. However, the same E/pH trend isobserved for intermediate (oxide/oxide) transitions inthe case of other, redox active, hydrous oxide systems(16) where the decay of metal atom activity is clearlynot involved, ie supcr-Nernstian E/pH shifts are awidespread feature with redox-active hydrous oxidedeposits.

The l3-oxide material exists outside the metalsurface, ie it is clearly not an adsorbed state. Anexample of materials that may be similar to the l3-oxideare certain soil colloids (25) such as layered silica claysand hydrous oxides of iron and aluminium. In most ofthese colloids the oxycation framework bears a negativecharge with ion-exchangeable cations, eg H 30+, Na",Caz+, present as counterions in the intercalated water.The assumption here is that the gold oxide andaqueous components are intermingled in the 13-layer.The accumulation of negative charge on the oxidestrands or layers in the l3-material may be viewed interms of a hydrolysis effect (26), ie as a combination ofexcess hydroxide ion coordination and loss of protonsfrom coordinated water molecules, viz:

+OR -H+HO-Au3+-OHz ---Au3+-OHz -- -Au 3+-OH-

(10)at highly charged cation (Au3+) sites. The involvementof super-Nernsrian E/pH shift effects in theelecrrocatalytic behaviour of gold has already beendemonstrated (27).

Recent reinvestigation of hydrous oxide growth ongold under repetitive potential cycling conditions (28)have shown that the behaviour of these oxides issignificantly more complex than assumed earlier. Withthicker oxide deposits in acid solution four oxidereduction peaks were noted, Figure 4(a). The Ep valuesand oxide assignments are as follows:- Cj , 1.1V(monolayer oxide); C'z and C"z, 0.97 and 0.94V (theoxide involved, HOI, is assumed to contain twoslightly different components); C 3, 0.73V (H02); C4,0.58V (H03). It may be noted that evidence of a

@ GoldBulletin 1997, 30(2)

10000

H03

TOT

2500 5000 7500NO. OF CYCLES

1.2r--- - - - - - - - --------.

Figure 5 Variation of the charge fOr hydrous oxidereduction as a function ofthe number ofoxidegrowth cycles. The conditions used (apart fromthe number ofcycles) are as specified in Figure4(a). The latter diagram also shows the oxidepeak assignments:- C] + C'], H01; c:~,H02; C4, H03

peak current (ip) values increased linearly, withincreasing oxide reduction sweep rate. Such behaviourindicates that the oxide reduction processes (especiallyfor HO 1 and H02 - the peak for H03 was quitebroad at fast, >5mV s', sweep rates) occur rapidly, tein a quasi-reversible manner (ifEquations 7 and 8).

'1

520 U H020'

.,

~ HOIe 0

:::,

(b)

004 0.8

FJV (RHE)

-5

similar nature was obtained recently for the presence ofthree distinct components in hydrous oxide filmsgrown on Pt in acid (29).

Figure 4 Reduction profiles, 1.20 to 0.0 V at 2 m V s:',fOr multilayer hydrous oxide films grown on amildly abraded gold wire electrode by repetitivepotential cycling (oxide growth conditions: -1.0xl ()1 cycles between 0.90 and 2.40 Vat 50VsI) in 1.0 mol dm3 H2S0 4 at (a) 25Cand (b) 35C

One of the interesting features that emerged fromthe recent investigation (and which explains why thecomplexity of the system was not appreciated earlier)was that some of the components involved, especiallyH02, are produced only after an extended number (ca2500) of oxide growth cycles, Figurc 5. For films ofconstant thickness the charge for reduction of thevarious oxide components was virtually independent ofsweep rate (5-100 mV s'). The peak potential (Ep)values decreased, though not dramatically, while the

The reduction behaviour of thick hydrous oxidefilms on gold in base is much more complex than thatobserved for gold in acid. Conditions for producingsuch films in base, using the repetitive potential cyclingtechnique, were investigated by Burke and Hopkins(22) who observed in some instances oxide reductionpeaks with a maximum (Ep) at ca 0.2\1: More recentwork (30), an example from which is outlined here inFigure 6, has yielded a behaviour pattern that is evenmore unusual as a major portion of the charge requiredfor oxide reduction at a very slow sweep rate wasobserved in the region between ca -0.1 to -0.3\1: Withan organic base electrolyte the extent of reduction of agold hydrous oxide film in the negative sweep atE >O.OV was quite small. The main oxide reductionresponse occurred in this case (at 10mV s"; Figure 7)only at E < -0.3\1:

~ GoldBulletin 1997,30(2) 49

E/V(RHE)

Figure 6 Reduction profile, 1.2 to -0.4 Vat 2. 0 m V r lfor a thick oxide film grown on a pre-abradedgold wire electrodein 1.0 mol dm-3 NaOH at2YC (the dashed line shows the subsequentpositive sweep. The oxide deposit was producedin-situ ~y potential cycling, 0.75 to 2.40 Vat50 V s', 6364 cycles

(11)2M-yr ----C - nF'YlP

Equation 11 is well known from electrochemicalnucleation theory (rc is the critical radius of thecluster, M is the molar mass of the metal, P is the

Apart from establishing the basic electrochemistryof these gold hydrous oxide systems in aqueous media,very little other work has been devoted to this topic;hence only basic ideas can be outlined here. Thereactions involved, even in acid solution, obviously donot occur under ideal thermodynamically reversibleconditions, ie there is significant cathodic overpotentialinvolved and there are various possible contributions tothe latter. It seems unlikely that the monolayer oxidedeposit acts as a barrier to f3-oxide reduction, especiallyin base where (in the case of thin films) the twocathodic peaks in the negative sweep are quite farapart. The f3-layer may be regarded as a semi-rigid,amorphous deposit and since reduction probablyoccurs at the oxide/metal interface there may well bedifficulty in maintaining contact between the metalsurface and much of the nearby oxide deposit when thelatter is undergoing reduction. Oxides and hydroxidesare generally more reactive (eg more prone todissolution) in acid solution. It was pointed out earlier(21,24) that the activity of Au3+ cations, existing in agel network of OH- and OHz species in the f3-layer, isinfluenced by the OH-ion activity in the bulk solution.This approach, developed originally to rationalize thesuper-Nernstian E/pH effect mentioned earlier,highlights one important contribution to the markedoverpotential observed in the case of the f3-oxidereduction in base, as compared to acid.

Another major factor in the oxide film reductionprocess is a combination of lattice stabilization energyand particle size effect at the gold surface. There isevidence from Scanning Tunnelling Microscopy(STM) (31) that discharge of metal ions from solutionoccurs most readily at ledge sites on the surface. Metalions in the f3-layer (especially in base where the layer ismore stable) are less free to move to such sites; theremay, for instance, be some preliminary discharge atsuch sites, followed by localized loss of contact. Then,over much of the surface, oxide reduction may have tooccur under conditions where the metal atoms areproduced in a poorly lattice stabilized, active form.Small clusters of atoms formed at the interface alsoconstitute a high energy state of the metal - the inverserelationship between cluster radius and overpotential(32).

2.0

1.20.80.4E/V(RHE)

o-0.4

(b)0

0.2 (a)Ei -10 0

-0.2

Figure 7 Reduction profile, 1.2 to -0.5 V at 10mV s:!fir a thick oxide (produced in-situ by dcpolarization at 2.3 Vfor 90 min) in 1.0 moldm-3 {C2H5)qNOH at 25C. The inset showsa cyclicuoltarnmogram, 0.0 to 1.6 Vat 50m V 5- 1, for the same electrodein the absence ofthe hydrous oxide deposit

50 C69' GoldBulletin 1997, 30(2)

density of the cluster which has a surface tension orsurface energy 'Y). Since the cluster is unstable (orcannot form) until r > rc the driving force oroverpotential ('Y]) must be increased (this also has theeffect of reducing the nucleation barrier energy and rcvalue) until oxide reduction occurs at a rapid rate. Theunusually high overpotential observed in someinstances in alkaline solution, Figures 6 and 7, mayreflect a combination of very low Au 3+ activity (theoxide matrix being more stable in base) and high goldatom activity. The unusually low reduction potentialsobserved in the presence of the terralkylamrnoniurnhydroxide electrolyte, Figure 7, may reflect the effect ofadsorption of Nit;+ ions on either the gold surface orthe initially produced adarorns (32) - this may result ininhibition of cluster growth, an unusually high metalactivity and a very low reduction potential. Anotherpossibility is that accumulation of hydrophobic cationsat the interface affects either the oxide polymer or somepre-electrochemical step (perhaps depolyrnerization).

The presence of several peaks in the oxidereduction process, sec Figure 4, may be due to theexistence of more than one type of hydrous oxidespecies or more than one state of activity of the freshlyproduced gold atoms - the influence of activity, inNernstian terms, is evident from Equation 9. Whilethe precise origin of such behaviour is unknown, thephenomenon is of significant importance as theelectrocatalytic behaviour of gold (33) (and indeedplatinum (1 1)) is also assumed to involve the reactionof two distinct surface oxide mediator species.

2 Pre-Monolayer OxidationGold is frequently regarded as the ideal solid electrodesystem for fundamental investigations inelectrochemistry as, in the absence of redox activespecies in the aqueous phase, the system apparentlyexhibits only non-Faradaic (or double layer) behaviourover the range of about 0 to 1.3 V in acid or 0 to 1.2 Vin base. However, as pointed out by Woods (12), therehave been repeated assertions that Faradaic behaviourdue to the formation of oxyspecies at the gold surfaceoccur well within the double layer region. Such reportswere usually dismissed, the effects in question beingattributed to the presence of impurities either on themetal surface or in the bulk solution.

More convincing evidence for the formation ofoxide species in the double layer region of gold, iepremonolayer oxidation of the latter, has beenobtained more recently with the aid ofoptical/spectroscopic techniques. Nguyen Van Huongand co-workers (34), on the basis of electrochemical

($i9' Gold Bulletin 1997, 30(2)

and spectroscopic data, postulated in 1980 that sometype of oxyspecies was formed in the double layerregion (to a coverage of 10-20% of the surface) on goldin base. In the following year Watanabe and Gerischer(35) postulated, on the basis of photochemical data,that gold in acid exhibited premonolayer oxidationextending over the potential range 0.85 to 1.35 V.Desilvestro and Weaver (36), using Surface EnhancedRaman Spectroscopy (SERS), established that for goldin base the product of prernonolayer oxidation was ahydroxy species, and, quite importantly, one that wasof different character to the species involved in theregular monolayer oxidation reaction. Hutton andWilliams (37), on the basis of scanning lasermicroscopy data, also claimed that some oxidation ofthe gold surface occurred prior to regular monolayeroxide formation; furthermore, they found that theincipient oxide involved was exceptionally stable - onlybeing removed by prolonged evolution of hydrogengas.

Claims based on electrochemical data forpremonolayer oxidation in the double layer regioncontinue to appear. Of particular interest is theprocedure used by Kirk and co-workers (38) who useda relatively low upper sweep limit. By avoiding thelarge response for the monolayer oxideformation/ removal process, and using fast sweep ratesand high recorder sensitivities, they observed markedFaradaic responses in the double-layer region for goldin base. Other authors who have reported unusualresponses attributable to the formation of oxyspecieson gold in aqueous media include Conway and co-workers (for both acid (39-4 I) and base (42)), and oneof the present authors (19, 43). Further data obtainedusing non-electrochemical techniques that support thepremonolayer oxidation approach include theradioactive tracer work of Horanyi and Rizmayer (44),and the Quartz Crystal Microbalance (QCM)investigations of Gordon and Johnson (45) (these twogroups worked with gold in base and acid,respectively). More recent QCM data for gold in baseby Kaurek and co-workers (46) were also interpreted interms of prernonolayer oxidation behaviour.

The essential point seems to be that nearly allmetal surfaces contain defects which from a catalytic orelectrocatalytic viewpoint may be considered as thesites at which catalytic and electrocatalytic reactionsoccur (11). The metal atoms at such sites are lowcoordination species - this was established recently byErtl and co-workers (47), using STM, in the case ofnitric oxide decomposition on ruthenium single crystalplane surfaces. The low lattice stabilization energy of

51

the active site metal atoms means that these atoms aresignificantly more electronegative than their bulklattice equivalents, ie these protruding metal atomsoxidize at unusually low potentials.

In general, very little is known about active sitebehaviour (4). Apart from the fact that the surfacecoverage of such sites is low, the material or atomspresent at such sites constitute an active state of themetal and the atoms involved are (i) far from being inthermodynamic equilibrium with the bulk phase, and(ii) apparently exhibit an unusual mode ofelectrochemical behaviour (11). It is assumed thatadarorns and small clusters of same at the surface areoxidized in aqueous media to form incipient hydrousoxide (rather than OHads-type) species. It was pointedout earlier (48) that whereas the electrocatalyticactivity of metals is usually quite marked in the doublelayer region (due to the involvement of low coverageincipient oxides), the onset of monolayer oxideformation frequently results in inhibition.

It is known, eg from the work of Henglein (49),that with extremely small metal particles the redoxpotential of the metal shifts in the negative directionwith decreasing particle size; this effect is quite dramaticwhen the number of metal atoms involved is extremelysmall. The same concept is employed in the IHOAJAapproach (11) except that the active metal atoms,instead of being in a free particle, are attached to thesurface of the electrode. Evidence supporting the idea ofactive states of metals is provided by the work ofParmigiani and co-workers (50) who found thatsupported platinum microclusters also undergoanomalous oxidation, ie they react with oxygen underfar milder conditions than the bulk metal. Suggs andBard (51) have emphasized that microscopic potentialsof metal atoms at different surface sites are not thesame; however, they also indicated that there is nomethod of determining the surface energetics ofdifferent sites (the standard P value for a metalrepresents an averaging of values over atoms at differentsites). Pre-monolayer responses, as observed in cyclicvoltammetry experiments (43), may be regarded as apreliminary attempt to determine the redox behaviourof active surface atoms which playa vital role in thecatalysis of many surface and interfacial reactions.

CONCLUSIONSFrom the work described in this, the first of twoarticles reviewing the electrochemistry of gold, we havedescribed recent work on the behaviour of

52

polycrystalline gold in acidic and basic aqueous media.Two significantly different types of oxide deposits maybe produced on the metal and the conditions underwhich these are formed and their properties have beenconsidered in Part 1. In Part II the catalytic, andparticularly the electrocatalytic, behaviour of gold willbe considered in relation to the results described above.The special properties of gold, as revealed in thesearticles, will undoubtedly lead to further investigationsand could lead to new applications.

ABOUT THE AUTHORSLawrence Declan Burke received his BSc and MScfrom University College Cork in 1959 and 1961,respectively, and a PhD in 1964 from Queen'sUniversity Belfast where he worked with FA. Lewis onthe palladium hydride system. He spent a year as anAlexander von Humbolt Fellow in KarlsruheUniversity, Germany, where he worked on solid stateelectrochemistry with Professor Hans Rickert. Sincereturning to Cork he has been involved in an extensiveinvestigation of the electrochemistry of metals, oxidesand especially hydrous oxides. This work, which isconcerned largely with the unusual or non-idealbehaviour of these systems, is relevant to such areas aselectrocatalysis, electrochromic systems, electrolessdeposition of metals, etc. Professor Burke was awardedFellowshi p of the Electrochemical Society in 1995.

Patrick Francis Nugent received his BSc degree inchemistry from University College Cork in 1992. He iscurrently completing his PhD degree on theelectrochemistry of gold in aqueous media.

REFERENCESWM.H. Sachtler, Ber. Bunsenges. Phys. Chem.,1995,99,1295

2 G. Pillay and cr. Chen, j. Electrochem. Soc.,1996,143,3410

3 G.M. Schwab in 'Catalysis - Science andTechnology', Vol. 2, ed. I .R. Anderson andM. Boudart, Springer-Verlag, Berlin, 1981, p.4

4 1: Wolfram and S. Ellialtioglu in 'Theory ofChemisorption', ed. j.R. Smith, Springer-Verlag,Berlin, 1980, p.150

5 D. Pletcher,}. Appl. Electrochem., 1984, 14,4036 j.C Farmer, in 'Environmental Oriented

Electrochemistry', ed. CA.C Sequeira, Elsevier,Amsterdam, 1994, p.565

@ GoldBulletin1997, 30(2)

7 R.D. Rochlin and R.W Murray, j. Phys. Chem.,1981,85,2104

8 L.D. Burke and ].E Healy, j. Electroanal. Chem.,1981,124,327

9 L.D. Burke and T.G. Ryan, j. Appl. Electrocbem.,1990,20,1053

10 L.D. Burke, Electrochim. Acta, 1994,39,184111 L.D. Burke and P.E Nugent, Electrochim. Acta,

1997,42,39912 R. Woods, in 'Electroanalytical Chemistry', Vol.

9, ed. A]. Bard, Marcel Dekker, New York, 1976,p.1

13 G. Belanger and A.K. Vijh, in 'Oxides and OxideFilms', Vol. 5, ed. A.K. Vijh, Marcel Dekker, NewYork, 1977, p.l

14 B.E. Conway, in 'Progress in Surface Science',Vol. 49, ed. S.G. Davidson, Pergamon, NewYork, 1995, p.331

15 M. Pourbaix, 10 'Atlas of ElectrochemicalEquilibria 10 Aqueous Solutions', Pergamon,Oxford, 1966, p.399

16 L.D. Burke and M.E.G. Lyons, 10 'ModernAspects of Electrochemistry' No.18, ed. R.E.White, ].O'M. Bockris and B.E. Conway,Plenum Press, New York,1986, p.169

17 G. Milazzo and S. Caroli, "Tables of StandardElectrode Potentials', John Wiley, Chichester,1978, p. 229

18 L.D. Burke, M.M. MacCarthy and M.B.C.Roche,]. Electroanal. Chem., 1984,167,291

19 L.D. Burke, Y.]. Cunnane and B.H. Lee,j. Electrocbem. Soc., 1992, 139, 399

20 L.D. Burke, D.T. Buckley and A]. Morrissey,Analyst, 1994,119,841

21 L.D. Burke and M. McRann, j. Electroanal.Chem., 1981, 125,387

22 L.D. Burke and G.P. Hopkins, j. Appl.Electrochem., 1984, 14, 679

23 AC. Chialvo, WE. Triaca and A.]. Arvia,j. Electroanal. Chem., 1984,171,303

24 L.D. Burke, M.E. Lyons and D.P. Whelan,j. Electroanal.Chem., 1982,139,131

25 N.C. Brady, 'The Nature and Properties of Soils',10th Edition, Macmillan, New York,1990, p. 177

26 L.D. Burke, M.E. Lyons, E.].M. O'Sullivan andD.P. Whelan, j. Electroanal. Chem.,1981,122,403

27 L.D. Burke and K.]. O'Dwyer, Electrochim. Acta,1989,34,1659

($S' GoldBulletin 1997, 30(2)

28 L.D. Burke and P.E Nugent, to be published29 L.D. Burke and A]. Morrissey, j. Appl.

Electrochem., 1996, 26, 59330 L.D. Burke and ].M. Moran, to be published31 R.]. Nichols, D.M. Kolb and R.]. Behm,

j. Electroanal. Chem., 1991,313,10932 M.T. Reetz, W Helbig and SA Quaiser, 10

'Active Metals - Preparation, Characterization,Application', ed. A Fiirstner, VCH, Weinheim,1995,p.279

33 L.D. Burke and B.H. Lee, j. Electroanal. Chem.,1992,330,637

34 G. Nguyen Van Huong, C. Hinnen and]. Lecoeur,]. Electroanal. Chern.,1980, 106,185

35 T. Watanabe and H. Gerischer, j.Electroanal.Chem., 1981, 122,73

36 ]. Desilvestro and M.]. Weavers, j.Electroanal.Chem., 1986,209,377

37 R.S. Hutton and D.E. Williams, j. Am. Chern.So~, 1994,116,3453

38 D.W Kirk, ER. Foulkes and W.E Graydon,j. Electrocbem. Soc., 1980, 127, 1069

39 H. Angerstein-Kozlowska, B.E. Conway andA Hamelin, Electrocbim. Acta, 1990,277,233

40 H. Angerstein-Kozlowska, B.E. Conway,A Hamelin and L. Stoicoviciu, Electrochim. Acta,1986,31, 1051

41 H. Angerstein-Kozlowska, B.E. Conway,A Hamelin and L. Stoicoviciu, j. Electroanal.Chem., 1987,228,429

42 H. Angerstein-Kozlowska, B.E. Conway,B. Barnett and]. Mozora, j. Electroanal. Chem.,1979,100,417

43 L.D. Burke and W.A. O'Leary, j. Appl.Electrocbem., 1989,19,758

44 G. Horanyi and E.M. Rizrnayer, j. Electroanal.Chern., 1984,165,279

45 ].S. Gordon and D.C. Johnson, j. Electroanal.Chem., 1994,365,267

46 W. Kautek, M. Sahre and D.M. Soarnes,Ber Bunsenges. Pf~ys. Chem., 1995,99,667

47 T. Zambelli, ]. Winterlin, ]. Trost and G. Errl,Science, 1996, 273, 1688

48 L.D. Burke, Platinum Metals Reu., 1994,38, 16649 AHenglein, Ber. Bunsenyes. Phys. Chem., 1995,

99, 90350 E Parmigiani, E. Kay and P.S. Bagus, j. Electron.

Spectrosc. Relat. Phenom., 1990, 50, 3951 D.W. Suggs and A.]. Bard, j. Am. Chern. Soc.,

1994,116, 10725

53