S. GILMAN
70
The Mechanism of Electrochemical Oxidation of Carbon Monoxide and Methanol on Platinum. 11, The (‘Reactant-Pair” Mechanism for
Electrochemical Oxidation of Carbon Monoxide and Methanol’
by S. Gilman Geneial Electric Research L n h a l o r y , Schenectady, hTew York
(Receiaed J u l y 16, 1963)
Data were prcviously reported for surface coverage of platinum with carbon monoxide and “oxygen” after application of a fixed potential to a surface covered with (approximately) a CO monolayer. The data are consistent with a “reactant-pair” mechanism for the electrochemical oxidation of CO. This mechanism assumes that the activated complex in the loss of the first electron from CO involves an adsorbed CO molecule adjacent to an adsorbed water molecule. The reactant-pair mechanism is also consistent with available data for electrochcrnical oxidation of methanol in acid solution. Some observations suggest that a similar mechanism may be involved in the oxidation of a number of simple organic molecules at the lower range of overvoltages.
Introduction
V ~ O U S ~publications -~
Considerable information has already been amassed concerning CO arid “oxygen” adsorption on platinum under electrochemical conditions. Because of relative structural simplicity and chemical stability, this information was attainable to a higher degree of precision than can likely be ohtained for more complicated organic molccules. This makes CO attractive as a model compound in spite of the fact that the clectrochemical oxidation occurs a t a convenient rate only at potentials a t which surface oxidation also occurs. In this study, an attempt is made to deal with this latter complication, and a mechanism is offered in explanation of the resulting data. I t is also demonstrated that the mechanism offered for C N is consistent with the data availablc for methanol. Finally, it is suggested that the oxidation of other simple organic molecules may proceed through an activated complex similar to that proposed for CO. *t3
Experimental Eyuipmcrit and reagents used have been described previously.2 Where indicated, data taken from preT h e Journal of Physical Chemistry
have been utilized. S e w experimental results were obtained by means of the previously discussed techniques, 2 , 3 the potential sequence appearing in Fig. 1, and the procedures listed in Table I.
Results and Discussion I . The Electrochemical Oxidation of CO on Platinum in 1 N Perchloric Acid. A . Current-Time Traces at Potmtial U . Surface Initially Covered wilh CO. When the surface is equilibrated with a solution of I N perchloric acid saturated with a 1% CO, 99% argon gas mixture or with 100% CO, only about 20% of the sites available for hydrogen deposition remain unobscured by C 0 2 and we may assume the surface almost completely saturated with CO in each case. IJnder these conditions thc current-time trace a t constant potential This work was rnade possihle by thc support of tho Advancrd Itrscarch Projects Agency (Ordrr N o . 247-61) through t h e United Statps Army Engiriecr Research and Devcloprnent Lab0ratoric.s undcr Contract Nuinbcr DA-44-009-E:NG-4853. S.Gilman. J . Phys. Chem.. 66, 2657 (1962); (b) 67, 78 (1963). S.Gilman, ibid., 67, 1898 (1963). S. Gilman and M. W. Brciter, J . Electrochem. Soc., 109, 1099 (1962).
MECHANISM OF ELECTROCHEMICAL OXIDATION ~-
OF
co AND METHANOL ON Pt
‘71 -
Table I Sequence Figure no. no.
I
I1
Step PO.
Procedure
Purpose
Final result
la
A
Bubble 99% argon-l% CO mix- ( 1 ) T o keep the solution saturated ture with paddle-stirring (200 with gas while removing oxir.p.m.) for 15 sec. dizable adsorbed impurities from the electrode surface. Also t o deposit passive oxygen film
la
B
Bubbling and stirring continued (2) To sweep away and dilute mofor 20 sec. Stirring without bublecular oxygen produced durbling continued for an additional ing last step. The passive film 10 sec. is retained
la
B
The stirring is discontinued for an additional time T B ’ a t step B
la
C
la
13
Solution is allowed t o come t o ( 4 ) The electrode is quickly re- (4) The surface is partially covered with CO and the solution is duced at U = 0.4 v. and CO rest for 1.5 min. quiet. The concentration of ie brought t o the surface by CO adjacent to surface is less the agitation of the solution (from the previnus step). than that in the bulk. Additional adsorption of CO during Since the rate of agitation decreases rapidly the surface step D is negligible (ref. 2) need not reach equilibrium coverage by the end of the step The oscilloscope is triggered a t ( 5 ) To record the current-time ( 5 ) Current-time trace a t potential the beginning of the potential trace corresponding t o a surU for partially covered surface face partially covered with CO step
lb
A
Solution of 1 M methanol, 1 A; (1) To remove oxidizable adsorbed impurities and methanol from perchloric acid; stir for 15 sec. the electrode surface with argon bubbling
lb
B
Continue stirring with argon for ( 2 ) To reduce surface and sweep away 0 2 and oxidation prodI min. ucts and allow adsorption of methanol Allow solution t o cDme t o rest for 1.5 min.
( 3 ) Topermit the solution t o reach a rate of agitation greater than aero but less than that in the actively stirred solution
(3) To allow for mass transport by diffusion only
lb
C
(4)Trigger oscilloscope
(4) To record current-time trace
(4) Current-time trace for oxidlation of methanol a t U = 0.7 v.
lb
I)
(5) Continue oscilloscope trace
(5) Record current-time trace
(5) Trace for oxidation of methanol a t U = 0.5 v. after “activation” at U = 0.7 v.
U (lye CO solution) proceeds from a small initial value to a maximum current and then declines to 0.3 At the same time, the surface coverage with CO declines gradually from its initial equilibrium value to 0. Thie downward trend of current at constant potential is easily explicable on the basis of decreasing surface concentration and hence activity of CO. The rising portion of the trace implies that the over-all rate of the
electrochemical reaction increases as a direct function of (a) increasing time, (b) decreasing surface coverage with CO, or (c) a combination of (a) and (b). A simple situation corresponding t o situation (a), for example, is that the reactant or surface undergoes change in reactivity (structure) with time a t the applied potential. To test the relationship between current, time, and &o, it is possible to start the experiment with the surface Volume 68,Number 1 January, 1964
S.GILMAN
72
B
r~
0.4v
u
-
-J
5 Iz
w
I O -
a
(bt 1.6~ A
0.5~ B TIME
-
e
0.5~
D
Figure 1. Potential sequences applied to the working electrode.
only partially covered with CO and to observe the current near zero t i n e . This experiment was performed employing sequence I, Table I. U = 0.80 v. was chosen since the current a t this potential includes no mea.surable contribution from surface oxidation.6v6 Figure 2 presents the results. Trace 1 corresponds to a
IllTb=O
121Tb=40rec.
1.0
2.0
T ~ sec. ,
Figure 2 . Currenktime traces obtained for an electrode partially covered with CO. Trace 1 corresponds t o a surface initially equilibrated with CO. Traces 2 and 3 correspond to partially equilibrated surfaces.
surface initially covered to the equilibrium value with CO. After decay of the capacit,y current, the initial current starts a t the familiar low level and rises to a maximum. Traces 2 and 3 correspond to only partial surface coverage with CO. These traces exhibit initial CO-oxidation currents approaching the maximum current of trace 1. This establishes that the increase of current before the maximum of trace 1 depends directly on the surface coverage with CO, and only indirectly on the time, TD, elapsed after application of potential U . If we were to shift traces 1, 2 , and 3 of Fig. 2 along the time axis so as to superimpose the points a t which the current drops to zero, we would observe that traces The Journal of Physical Chemistry
2 and 3 do not simply constitute portions of trace 1. We would not expect such behavior if a single fixed relationship existed between current and OCO. An explanation for this complication will be offered below on the basis of different initial distributions of admolecules on the surface. B. The Adsorption and Desorption of Carbon Dioxide. I t is possible that product COz might act as a “poison” by affecting the surface free energy or desorbing with difficulty, etc. The tendency of COz to adsorb from solution under our conditions was therefore examined to gage the probability of such kinetic considerations. The procedure employed was identical with that previously used to study saturation adsorption of CO.*& The electrolyte was saturated with pure COZ and its adsorption was followed with time by application of a linear cathodic pulse and determination of the charge due to hydrogen codeposition. KO reduction in the charge due to hydrogen deposition and hence no COz adsorption was observed over 1000 sec. The evidence suggests that COZ does not chemisorb under our conditions. This is in accord with the observation that COZ does not adsorb from the gas phase a t this temperature.’ It can thus be safely assumed that dissolved product COz plays no role in the kinetics of CO oxidation. This still allows the possibility that COP formed on the surface desorbs with difficulty. This possibility is in turn rendered unlikely by the observed fact that hydrogen codeposition increases regularly as CO is ~ x i d i z e d . ~This argues against a buildup of adsorbed COz on the surface. C. The “Reactant-Pair” Mechanism f o r CO Oxidation. Starting with the observation that the over-all rate of electrochemical reaction increases with decrease in Oca, we are confronted with a situation which may be described as self-poisoning of the reaction with the reactant, One simple physical situation corresponding to such an observation is that the electrochemical reaction involves adjacent surface sites. For the time being, we will assume that an adsorbed CO molecule and a “free site” (not covered with CO but presumably covered with other solvent species, oxide, etc.) are involved. Corresponding to the situation in which the current is determined by rate of electron transfer from such a “reactant-pair,” we may write ICoB = nKcoBFS, exp(cun’UF/RT) (1) where ZcoB is the partial current for the oxidation of bridged CO; KCoBis the rate constant for oxidation of (5) H. A. Laitinen and C. G. Enke, J . Electrochem. SOC.,107, 773 (1960). (6) S. Gilman, to be published. (7) A. C. Collins and B. M. W. Trapnell, Trans. Faraday SOC.,53, 1476 (1957).
MECHANISM OF ELECTROCHEMICAL OXIDATION OF CO AND METHANOL ON Pt
bridged (XI; n is the total number of electrons in overall oxidation = 2 ; F is the Faraday constant; a is the transfer coefioient; n’ is the number of electrons in the rate-determining step; and S, is the density of reactant pairs on the surface. 12 strictly analogous argument applies to lincar (XI. I t is next necessary to express S,i n tcrms of t,hc cxpt?rimentallymeasurable quantities, &OR, and (211. ‘Yo this end we may regard our reactant pairs as bcing located between clusters of CO ad-moleculcs arid “frec? sites.” Physical models for this situation will be offered below. Case 1. Assumptions. Oxidation of CO not adjacctnt to a “frce site” is completely hindered. Increase in the number of reactant pairs may occur only by the outward extension of the line of contact between clusters of adsorbed CO molecules and clusters of “free sites.” At high values of 8co we may regard the clusters of free sites as openings in a n otherwise continuously covcrcd surface arid
sa‘
=
l?’CCF,, i
(2)
where CF is the circumference of a n individual cluster of “free sites” and 1’ is a constant. At low values of eco, we may regard the CO clusters as openings in an otherwise “free” surface and
S,”
l“CCc0
=
B
,i
i
(3)
where CcoB, i is the circumference of an individual cluster of bridged CO ad-molecules and 1” is a constant. Over the medium range of surface coverage we may there fore write
sa = zCCF,i c c o B , i
(4)
1
where circumferences with the same value of i are adjacent . Initial Conditions at Potential U . The adsorption of a gas on an ideal surface with sites of uniform free energy might be expected initially t o result in single isolated ad-molecules (cluster size = 1 molecule/cluster) a t low surface coverage. Similarly, at high surface coverages we might expect thc size (area) of clusters of “free sites” to be minimal. This would lead t o the maximum value for the circumference terms in eq. 4. On the other hand, for a real surface we might expect a distribution of free energies of adsorption on the surface and a corresponding tendency for cluster sizes to be large and for the circumference terms in eq. 4 to approach a minimum for a fixed value of eco. Further, we must expect the circumference term in (4) to vary with the path t)y which the particular surface coverage was established, e.g., adsorption on a completely or
73
incompletely reduced surface; saturation adsorption followed by some stripping-off by oxidation, ctc. I t is thus conceivable that the value of S, and hence the current at constant potential may vary by orders of magnitude a t the beginning of an clcct,rochcmical oxidation step. Case l a . Assumptions. These are the same as for case 1 plus the assumption that (free) site clusters may expand or (CO ad-sites) diminish only isotropically during CO-oxidation. Let us assume the clusters are initially circular in shape and also let us accept the approximation of equal size for similar clusters. Then
(5) where NcoB is the number of bridged CO clusters and RcoB is the radius of bridged CO clusters. Also Cco’
OCoB
=
2~Nco’~Rco~
= Q~o~/(&co~), = mNcoB*(RcoR)2 ( 6 )
where is the charge density equivalent to “saturation coverage” with CO, taken as 0.44mcoulomb/cm.2, and m is an appropriate constant. Examination of eq. 5 and 6 reveals that CcoB is proportional to (ecoB)li’. Similar arguments appy to CF. Hence over the medium range of surface coverages
ICO” = ~ ~ c o ~ ~ ’ F ( ~ c o ” ) ‘ ”exp( ( ~ Fa)n’ ’’F~/ R T )
(7)
where kcoB’ is a formal rate constant. Case ib. Assumptions. These are the same as for case 1, plus the assumption that site clusters may expand only in a decidedly nonisotropic manner. As for case la, let us further simplify by assuming all similar clusters equal in size. Assuming rectangular geometry Cco” = NcoB2(c
+ d)
(8)
where c and d are the dimensions of the average site cluster. Also ecoB = mfNcoBcd
(9)
where m’ is the appropriate constant. Inspection of eq. 8 and 9 reveals direct proportionality between CcoB and ecoB if we assume c >> d. Similar arguments apply to CF and O F . Hence for the medium range of surface coverages we may write
Z C O ~= ~ F ~ c o ’ ” ’ B c o ~exp(aUn’F/RT) B~ (lo) where kco”” is the formal rate constant. Case 1 c. Assumptions-Generalization of Cases l a and ib. We may generalize upon eq. 7 and 10 by writing
Volume 68,Number 1 Januaru, 1964
S. GILMAN
74
ICO’
=
~ZUCCO’(Bco’)
”(6,)
’ exp(
LY
Un.‘I“/Iz7’)
(11)
where p and p are exponents having values between and 1. Equation I 1 now carries with i t the assumption of cases l a and I b that all clusters arc equal in area and hence in circumference. I t is reasonable, therefore, to term unusually large clust,ers “inactive” since the circumference per unit area is small. We may make correction for thti total inactive surface coverage as
ICO”= 2Fkco”(O~o”- a)”(6F
-
arbitrarily assume PtO stoichiometry2* and take BF = (QII - Q0/2)/QS1r. Equation 13 is tested for both bridged and linear CO in Fig. 3. In Fig. 3a, values of
b)* exp(aUn’F/RT) (1%
where n and h are fractional “inactive” surface coverages. Case 2. Assumptions. Thcre is some oxidation of CO ad-molecules not adjacent to a “free” site. This situation would lead to nucleation of new clusters of “free sites.” ‘l’hese new clusters would have larger circumference to diameter ratios than the average older clusters and the efyect may be to increase the current greatly although BF increases only slightly. For any nucleation rate, the accelerated dependence of current on OF may he cxpccted to increase in importance as the number of original clusters of free sites decreases and their average radius increases. The simplest situation would arise when the average nucleated cluster soon attains the same circumference as the average original cluster, when the deperidcricc of the current upon surface coverage could still be represented by eq. 12. D . Application of the “Rcactnnt-Pair’’ Mechanism to Constant Potmtia! I l a h Obtained in a Solution of 1 % CO. Interprctatiori of the data in terms of the individual models discussed docs not seem feasible without detailed knowledge of surface morphology and the structure of thc ad-layer. It will only he attempted to show that, thc gencralizcd equation (12) is consistent with the data obtained. Integration of eq. 12 from zero time to duration of time, T I ) , at, constant potential L7 results in a charge corresponding to oxidation of bridged CO. For the surface ori&ially equilibrated with a saturated solution of I yo CO, this charge is i n turn equal to the difference between the initial equilibrium value of the charge corresponding to adsorbed CO arid its instlantaneous value (1co’)dm 2Flico”
=
B
[cQcoB),.- Qco
JOT’)
(Oco” -
I
=
a)”(Bv - h)q drD
(13)
In the absence of oxidation of the surface (li 6 0.80 v.) , we will assume that hydrogen deposition is a measure of the “free surface”2a and take BF = Q~I/&’H. In the presence of surface oxidation ( U > 0.80 v.) we will T h e Journal of Phfpical Chemislry
t
I
I
I
I
I
I
I
0.3
a4 To sec.
0.5
0.6
0.7
I
0.1
a2
I
I
I
I
I
0.8
Figure 3. ICxperimental and theorctical CO-oxidation chargo = 0.85 v . : (ti) bridged CO; ( b ) linear CO.
at U
[ ( Q c o ” )~ Qco”] are plotted from previously3 obtained
data. The circles on the figure are values of $,7D(BcoB - b)’ d m obtained by graphic integration with n = b = 0 and p = q = 1. Compared to the smooth curve, these points have values which are too large a t small values of TD. This could correspond to the situation where most of the initially available free sites are “inactive” in the sense discussed above. This may be compensated for by choosing b = 0.092, and leaving a = 0, p = p = 1. The squares of Fig. 3a are the resulting points obtained which (with normalization a t TD = 0.3 see.) are a fair fitj to the cxpcrimental curve. In E‘ig. 3b a similar analysis was made for linear CO with the circles now corresponding to 6 = 0.26, a = 0, p = q = 1. An excellent fit of theory to expcrimcntal data is obtained. The possible physical significanca of different values of h for bridged and linear CO is that the active free sites which are generated by the oxidation of bridged CO ad-molecules initially do not occur in the immediate vicinity of linear CO ad-molecules. Hence t.he new free sites are “active” for (adjacent to) bridged ad -molecules hiit “inactive” for (not adjacent to) linear CO ad-molecules. The argument allows us to establish only that a fit of
MECHAXISM OF ELECTROCHEMICAL OXIDATION OF CO AND METHANOL ON Pt
t.heory to experimental data may be obt,airied and does not in itself constitute proof of the t,hcory of any of the dctailcd situations discussed in cases 1 and 2. S o doubt, ot,hcr satisfactlory fit’s may also be obtained with other choices for thc coristants of eq. 1 3 . If, for any fixod potential c‘, the rate of oxidation of CO to C’02were determined by an electron transfer, we would expect from ey. 1 that a t constant potential li
Ico”/S,
=
nFKco’’
=
constant
(14)
l‘hc values of KCO”for different values of C‘ might be expected to fall on a Tafel plot. A similar argument applies t,o the formal rate eorist,ant, kcoR, obtained by means of eq. 13. Since the values of the constants of ey. 1 3 may be different for each potential, and since the choice of constants is in any case fairly arbitrary, it was felt more general and convenient to work with eq. 12aridw = b = 0; p = q = 1. The use of eq. 12 requires mcasurcment of a current due to oxidation of a particular CO species, whereas the tlot~al currc?nt measured includes several other contribution^.^ ICO’ is thus best obtained by differentiation of the appropriate charge (see ey. 13)
75
-
IO
1.0 N
E
o
h
0
I
:k
I
4
(01
/
eLCo;0.075
-
,150 = ,225
B
= ,300
0
=
a
376 .“.
1
L I
.
-E
c
1.0
= ,150 -
’-
I
/
= ,225 = ,300
0
I d 0.8
0.9
I .o
I. I
I.2
1.3
I 1.4
U (VOLTS1 Figure 41) is a plot of Z C O ” / ~ C O ~against ~F U . Figure 4a is a similar analysis for linear CO. The points for any Figure 4. Potential-dependence of the oxidation currents const,ant value of U correspond to values of 8con and for CO: (a) linear CO; ( b ) bridged CO. 8co” increasing i n iritcrvals of 0.075. Ideally (cq. 14) the points a t constant potential would superimpose. The actual tendency of the constant-potential points t.o oxidation of lincar and hridgcd CO arc 2 x and 6 X amp./cm 2, respectively The rate for the clust~rrmay be called “fair” i n light of the complieatcd oxidation of methanol (assurniiig a slow electron transfcr manner of obtaining the data and the much simplified step first order in methanol and cither first or zero order modcl cmployod in its analysis. This diHiculty is minimizcd by the fact, that the data may be analyzed . ~ the rate in “free” surface) IS 2 X 10- a ~ n p . / c marid further on a logarithmic scale. The best l h e s through of the surface oxidation a t 80 = 0.1 is also 2 X the points from U = 0.85 to 1 . 1 v. result in apparent, amp /cm.2.6 The similar values of rate constant for values of an’ of 0.82 and 0.25 for linear and bridged CO, C‘O, methanol, and 1%-surfaceoxidation already suggcst similar rate-controlling stcps. respectively, and suggest a slow one-electron transfer in each case. Although the values of the ordinate The scatter of the points a t constant potential for C‘O throughout t,he potential range studied are not greatly about the mean (Fig. 4) is random with rcspcct to different for the two forms of adsorbed CO, the small surfacc coverage with CO, time, and hence also with differcncc in l’afel slopes results in the values of the extent of surface ~ x i d a t i o n . ~I t does not seein rcasonintercept, a t U = 0 v. dif‘fering considcrably. The ablc, thereforr, to hold surfacc oxidation out as the reason for the decided departure of the points for U values are 2Fkco” = 6 X l0-’3 and 2FkcoL = 5 x 10-9 amp./’cm.2for the ratc at, I: = 0 expressed as current. = 1.2 v. from the Tafel lines. One possible explanaThese may be compared with the valucs of 10-” tion for the effect is that the fitld becomes rionlincar with increasing interfacial potential (and changing amp.//cm. previously reported for mcthanol oxidation during the asceiiding sweep of the “polarization c u r ~ c . ” ~ structure of the ionic doublr layer). W’hile the rate shows a tcrtdcncy to dcpart from thc Tsfel line a t high These valucs all involve extrapolation back t’o I/‘ = 0 using uncertain Tafel slopes. A better comparison is pottntials, it docs not exhibit a maximum in the range probably obtained a t U = 0.8 v. where the rates for the examined. Hence the gradual decreasc, with incrcas-
Volume 68,Number 1 January, 1964
S. GILMAN
76
ing potential, of the current measured for the “polarization curve”2a may not be attributed to a decrease in the rate of thc elcct,ron-transfer step. This matter will be treated further subsequently. E . Comparison o ~ ”Constant-I’otential Voltammetric Data Obtained with Solutions nf 1% and 100% CO. At U = 1.0 v. and for a solution of 1% CO thc C‘Ooxidation charge passed before the current decays to virtually (compared with maximum currcnt) zero corresponds almost entirely with CO initially adsorbed on the surfacc, the addit,ional CO diffusing i n from the solution being riegligihle.a I?or the corresponding experiment in a satiirat,ed solution of 100% W, rcsupply of CO from the solution btcomes appreciable. IJndcr these latter conditions, CO supplied from the solution contributes approximately half tho total C N oxidation charge passed before the surface concentration drops to When, in this lat,ter case, the surface coricentration does fall to zero, it is because the diffusion layer thickness has Iricreased to the extent where it may only support the rate of oxidation on t>he bare surface arid no excess of CO flux to the electrode exists which might allow the surface covcragc to rise above zero. IIencc, at, that point the oxidat’ion is diffusion-controlled and occurs with Oc0 = 0. Before this situation of diff usion-controllcd oxidation currcnt is established, we may expect the CO-oxidation currrnt, to he related to OCO by an expression such as cy. 12, as was found for the experiment which cniployed a solution of 1yo CO. Further, the surface-oxidation current is also limited by CO surfacc covcragc. H(?nce, if thc distribution of (10 011 the surface were similar for the experiments employing differcrit concentrations of dissolved gas, we might expect the total currents to be similar functions of OCO. In Fig. 5 total current’ is plotted as a function of total surface coverage with CO, OCO = Q C O ~ ! ( & C O ~ ) ~ .We see that the general appearance of the two curves is quite similar but that the currents of plot b are 0.6 (at thc maximum) t,hat of
plot a. The initial values of @cot were 0.41 and 0.43 mcoulomb icm.2, rcspectively. By the arguments already presented, the number of reactant pairs for plot b must initially be smaller than for plot a, arid the number of such pairs will remain smaller if t h t rate of nuclcatiori of new clustrrs of free sites is small or zero. This experiment hence lrnds support to casc 1 of thc gcmeral mechanism offered above. I t was prcviously suggested that the sudden rise in anodic current with increasing potential of the “polarization curve1’2amight be due to higher activity of 11011adsorbed (newly diffused to the surfacc from the solution) CO. Figure 5 establishes that in fact lower currents arc supported uridcr the conditions of rcsupply of (’0 from the solution Hence, thc sharply rising current in question must be due only to the increase in rate which accompanies decreasing surface coverage (see below). F . Further Interpretation of thc “Polarization Curtie. ” The current-potential traces obtaincd by application of a slow pcriodic triangular sweep to a workitig electrode immersed in 100% CO wcre previously reported.2a h correlation was demonstrated to exist betwcen OF and the measurrd rurrents. ITurthrr examination of the “polarization rurve” in thc light of the new material presented here sccms warranted. Figurc 6 is a trace taken with a swccp speed of 0.04 v. ’sec., CO bubbling through thc electrolyte, and with
U
IVOLTSI
0
E’igurc 6. “Polarization curve” for CO in rapidly stirred stttnrated soliition of CO. Triangular sweep itpplied with potential varying bctwccn 0.4 and 1.8 v. (solid trace) or 1.4 v. (dashed trace); c = 0.04 v./sec.
Figure 5 . Total anodic current at li = 1.0 v.
The Jourital o j Phusical Chemistry
paddlc-stirring a t 200 r.p.m. For the ascending portion of thc swccp, virtually no current is observed from 0.4 to 0.91 v. ‘i’liis is i n contrast with the appreciable currents which may bc obtained for U as low as 0.8 v. in an uristirred solution of 1% CO. The explanation
offered is that the flux of CO to the electrode under these conditions is sufficient to prevent d ~ c r c m cin &o, which in turn causes the current to remain low. At L’ > 0,()1v., the ratc of ouidatioii or1 even the highly covered surface twgiris to t)ccorne appi*cciat)l(.with respect to the rate of readsorption of ( ” 0from solution a t high eco (where tht. rate of adsorption is srnaller than the rate of diffusion3). This causes eco to drop, which in turn causes tho rate of ( ’ 0 oxidation to rise. As 8c0 dcC T ~ B S C S apprecia1)ly the ratc of readsorptioii iiicreases and becomes diffusioti-c.otitrolled As a result of the demands made on the adjactlnt solution by the enhanced rates of oxidation arid rwdsorption, the diffuto an equi1it)rium value. sion-layer thickness increa ?‘he specific rate with wh (’0 may oxidize on the surface rises to such an eutent that all (”0ou the surface is consumed and any CO whivli strikes the surface is imrnediatcly consumed. M-e therefore have a type of avalanche effect where the system a t c‘ = 0.91 v. abruptly moves from a situation a t which eco is maximum and current is minimum to a situation in which 8co is zero and currcnt is diffusion-controlled. I n Fig 6 W P see a gradual decline in current from the peak (diffusion-lirriitcd) value rwar 0.91 v. to th(t mininium value a t 1.6 v. I t was previously suggested that t,his dcclinr could 1~ due either to (I) physical masking with surface “oxide” or ( 2 ) reduction i n the rate constant for t~lectrochemicaloxidation of ( ” 0 . Possibility ( I ) may still riot he ruled out, esptvially in the light of evidence that the total surface coverage with oxygen at low potentials is more extensive than previously believed (equal to 0.5 monolayer of J’tO or 0 23 monolayer of PtOz at C’ = 1.0 v. after 1 s e ~ . ~ )I’ossihility . (2) may he elitninatc~d,in view of the finding that the rate of oxidation ilicrcasrs gradually throughout the potcntial range studied Actually, if the rate of oxidation were to drcrease, readsorption must occur unless this is also Iiindercd. No readsorption of thci depolarizer at high potentials (past t h e currclit maximum) was found for eit1ic.r (’0 or tnethano1.b rl’lLis leaves t he possihility that it is the. rate of adsorption that is niarkedly afTectrd by “adsorbed oxygen” for both of tlicse systt.ms. Since we are speaking of adsorption at a surface coverage of zero, we might alternatc~lythink of the proctw as a “wrfaw activation” and suspect that it is tliisactivation (pr ding tlic elertroii transfer) which is h i ndr red I k r i n g t h r l tltwwitling s.c\-eepfrom 1 4 v. of I’ig. ti, we scc that thcl currclnt level reached at tlic (>ridof tlic ascenditig sivwp (lowrr than limiting current) is maintained, (~stat)lisliiiigthat it is the sui.face oxide directly and not thr iiicwaw in potential which produced the original currwit dcclinr Finally, we observe the
“hysteresis effect” at low potentials. This cffect mtty now be ascribed to the fact tliat t h c siirf:wc~ coverage with CO is now smaller than it was tfiiririg tht>:went, hence the number of reactant pairs is l a r g c ~ and , the rate of oxidation is greater. G 7’hc Structure. o j thP A c t i m t r d (’ovip1r.s -Summary oJ Ihe illcchaizism LVc niay no\v c*onsidcr t he possible nature of the “ f r w sites” involved 111 thc postulated reactant-pair. Thc “free site” adjaccmt to an adsorbed (“0rnolecnle may s w \ e o i i of ~ t w hasic functions. (1) It may permit the a d s o r t d (‘0to form an additional tlond to t hv surfacc. and tw “activated.” (2) I t may contain ai1 adsorl)ed spctcics which enters into the activated complex of thr t.att.-dctc~t,miriingstep Let us assume that possibility ( 1 ) applies. Then, since the currcnt a t cmstant potcnt ial teiids toward a maximum, we would also cxpect the number of bonds per adsorbed C’O moleculc to proceed toward a maximum. Alternatively, we may (’xpress thc argumcnt in terms of surface concentrations of linear and bridged (“0. “Activated” h e a r CO rnay be c>xpt’ctcldto resemble bridged CO physically. Tlence, wc might expect only bridged (%-site) (“0to be active, and the ratr of oxidation to be initially proportional to its surface concentration. Jmear (’0would Iiavv to convert to the 1)ridgc’d form to oxidize. Higher currents at, constant potential would iniply larger surface concentratioris of bridged CO. .Ictually. t h e linear form tends to decrease at approximately the same rate as the bridged form over the potential range studied, and in any casc, no increase in bridged ( ” 0o w r t h r init~al surface concentration is found during oxidation a t constant potential. I t therefore swms rrasonablr to discard possibility (1) abo1.e. F’or possibility ( 2 ) we might suppose that an adsorbed water molecule (or protonated water molecule) anion, or “oxygcn” might be ii~volvcd. Sinw ptwhloruto is involved in this casc, the possibility of an adsorbed ariiori does not seein attractive. ‘l’he “oxygen” might bo present as jstoichiotrir~trirully)thr hydroxide, oxide, etc Since CO will oxidize a t 0.8 1’ where the surface coverage r\ ith ‘hxygm’’ is imm(wurably small, ttir possibility of “oxygen” as a coreactant serms unattractive I:ui+her, for rncthaiiol and forniir acid, oxidation may be obsorvcd at as low as 0.6 v., niaking “oxygen” an even less :ittractire canditlatc. i l n adsorlied water species is left a s the prescwt most promising choico. ’l’he following schrnie is thcv-efore offered for thc elcctrochemic.al oxidation of (-0.
S. GILMAX
78
* ad jacrn t sites - . _ _ _ _ _ . _ A
*
_ - (HZO)
*
(linear adsorbcd CO)
(adsorbed K O )
+ 2* + (1120)--+
(B) CO (dissolved)
* (HZO)
(CO)
*I *!
*
(bridged adsorbed CO) (adsorbed H,O)
11. (CO) (1120)
*I
I
1
*
+ (CO)
*I
- - - (Il2O) - - - e
*I
(activated complex) --f
(CO) - - - (011) iI i1
*
+ e- + I-I+
*
111. (CO) - -. - (OH) -+ COz
I*
molcculcs of CO and oxygen and the reaction rate therefore also displays a maximum with increase in CO surface covcragr as it docs in our caw. I I . A pplicntzon of the LiRPactant-i’azr” illechanzsm to the Electrochemzcal Oxzclalion of Afelhanol i?i Perchloric Acid Solutzon. ‘I’he electrochcniical oxidation of methanol was previously interpreted4 on the basis of a rate-controlling one-clrctron transfer, first order in adsorbed methanol. Two anomalies arosc in this interprrtation. (1) l’alues of I for 0.3 2 OM 2 0.9 (concentration of methanol = 1 M) do not fall on the Tafcl line. (2) A "hysteresis effect” (currents at low potcntials are higher during descending than asccnding sweeps) more pronounced than that observed for CO occurs. I t was previously suggrstcd4 that these anomalies might bc explained cithrr by a highcr ratr constant for “unadsorhcd CO” or by some inysterious “activation” of the surface. Both of the s, howcwr, suggrst similarity with the CO system Further similarity is suggested hy the current-time tracc obtained at constant potcntial. Sequencc I1 of Tahlc I was employcd in obtaining Fig. 7. At U = 0.3 IT., the initial current is virtually zero
*I
Steps I A and I13 arc adsorption steps which will be very rapid for any surfacr coveragc slightly less than the equilibrium value and will be rate-controlled by diffusion uridcr ariy simple cxpcrimental situation. Step I determincs thc initial surfacc conccntration of rcactant pairs. l‘he progress of adsorbed linear CO is followed in stcps TI and 111, t h r argurncnt for the bridged form being identical. Strp IT is tlie clectron-transfer strp, which for any fixed surface concentration of reactant pairs is ratc-dctmmining. The loss of the first elcctron goes through the indicated activated complex. Step I11 is t h r final clrctrori trarisfrr wtiicli is assumed rapid comparcd with (IT). I n a scnsc, st,rps I1 and I11 rcseniblc rcaction of CO with “dissociated itater ” We may think of CO as lowring the cnorgy of activation for the dissociatioii of water (surface oxidation). I+’urther, at potrritiuls of br < 0.80 v. the product of surface oxidation \vould tcrid to b e rc-rcduced, whereas the corrcspondirig back rractions for stjeps TI and 111 are assumed to havc no appreciable rate. Thc proposed mechanism for thc electroehcmical oxidation of CO has some similarity to that proposed for the catalyzcd gas-phase oxidation of CO by ovygcn Thc latter oxidation is said to involvc adjarcnt adsorbed Thr .lournnl of Phgsknl Chemistry
I
+ e- + H + + 2(*)
a
T , sec
Figure 7 . Current-time trace for oxidation of 1 ,If methanol at constltnt potential: (a) Z‘ = 0.7 v.; ( b ) Z‘ = 0.5 v .
on the scale cmploycd. Vpon raising t h t potriitial to U = 0.7 v . , the current riscs from a minimum to a maximum valuc (tracc a) whilc thc surface coverage with methanol decreases somrwhat. 1-pon stcpping the putcntial back to 0.5 v., a considerable transient currcnt is obtained which finally decays to the initial zero value (trace b). Tracr a of I’ig 7 is prcciscly what is to hc expected for the reactant-pair mechanism (current rearhcs a maximum as surface coverage dccreases) arid tliffcrs qualitatively from CO in two main aspects--the minimum current is considerably larger comparrd with the maximurrl current and the currcnt does not drop to zero. The relatively large initial ~~
(9)
~
G. C. Bond, “Catalysis by Metals,” Academic Press, Nra York, N.Y.. 1962. p. 460.
MECHANISM OF ELECTROCHEMICAL OXIDATION OF CO AND METHANOL ON Pt
current may be ascribed to the presence of considerable free adjacent sites (compared with CO) at the beginning of the experiment. The failure of the current to drop off to zero is simply due to the much higher flux of methanol in this concentrated solution (compared with a saturated solution of CO). This tends to keep the surface coverage with methanol a t some “equilibrium value” larger than zero, and it is presumably such values which are sampled a t each potential by the “polarization curve” previously reported.* Trace b of the figure would then correspond to the higher currents obtained until the surface coverage again rises toward a monolayer, since the oxidation rate has dropped sufficiently to make this possible. Because of the structural complexity of methanol (compared with CO) it is also reasonable to attempt an explanation of Pig. 7 on the basis of “intermediates.” For trace a of the figure, we might be tempted to argue that the oxidation proceeds only to formaldehyde a t the beginning of the trace, but proceeds to COn toward the end of the trace. This allows for a ratio of maximum to minimum currents of only three. This value is clearly exceeded. Two additional explanations for Fig. 7 involve additional current due to nonadsorbed methanol and “surface activation.” We have seen that enhanced currents for CO oxidation are certainly not due to nonadsorbed CO and therefore tend to discount this argument by analogy. “Surface activation” due to “oxygen adsorption” is rendered highly unlikely here because of the low potential involved. This leaves us only with a very vague idea of “surface activation” for which the reactant-pair mechanism presently appears the only physical explanation. In Fig. 8 thc data previously reported for 1 M methanol, obtained during a 0.03 v./sec. triangular sweep,‘ arc plotted on semilogarithmic paper employing three different functions for the current ordinate
where OF = QII/@H and = 1 - OF. The first function is the one previously plotted4 in support of the simpler electron-transfer mechanism. Ihnctions b and c are for the reactant-pair mechanism, employing generalized eq. 12 with a = b = 0 and p = q = 1 for function b and p = q = ‘/z for function c. For the descending sweep, function b leads to the best linear plot. For the ascending sweep, function a results in the best fit in the medium and high range of potentials, but results in values which are too low In the lower
79
I e;
er
ASCENDING SWEEP
I
I, ASCENDING
SWEEP
8”
i q q k *ASCENDING
I e;
DESCENDING SWEEP
OF
8:
A
SWEEP
I
DESCENDING SWEEP
(e; eFPe I ’ DESCENDING SWEEP
”i Figure 8. Analysis of methanol “polarization data.”
potential range. Function b provides a good linear relationship at the high and low ends of the range, but departs from linearity in the medium range of methanol surface coverages. Function c is satisfactory only in the medium range. I t seems reasonable to conclude that the unique validity of either of the proposed mechanisms may not be established by t)his analysis. Part of the difficulty lies in the nature of the functions themselves. E’or any fixed value of I and 0, = 0.1 (high potcntial range), functions a--c have the values 101, 111, and 3.31. Hence (a) and (b) give similar results. For fixed I and e,, = 0.5 (medium potential range) functions a-c have the values 21, 41, and 21. Hence (a) and (c) give similar results. l h d l y for fixed I and e, = 0.9 (low potential range), functions a-c have values of 1.11, 111, and 3.31. Hence no similarity of values exist in this one range. Referring again to Fig, 8 we see that, the y-intercepts and hence apparent rate constants for the ascending and descending sweeps differ by over an order of magnitude. Clearly, there is nothing in the modal of a simple electron-transfer step first order in adsorbed mothanol which will explain this difference. The reactant-pair mechanism explains t,liis effcct, on thr basis of diffrrcnt initial surface concentrat,ions of reactant pairs. During the ascending sweep e,,, is high, OF is low, and the Volume 68,A’umher I
January, 1904
80
initial density of reactant pairs is small. These may only increase in number as OF increases. During the descending sweep, the surface is initially bare and adsorbed methanol molecules tend to deposit in clusters of minimum sizc; hence, the number of reactant pairs will tend to be a maximum for the particular surface coverage. One of the most striking differences between the “polarization curves” for methanol4 and CO is that the former exhibits the Tafcl region, starting at U > 0.7 v., whereas the latter has only an abrupt rise at 0.91 v. This difference may be qualitatively explained by the reactant-pair mechanism. For CO, as already proposed in a previous section, the active free surface is initially small at thc beginning of the ascending swcep. The current below U = 0.91 v. is so small that masstransport keeps the surface coverage high. When
The Journal of Phusical Chemistry
S. GILMAN
appreciable decrease in OCO may be accomplished, the potential is already so large that a sudden rise in current results. For methanol, we propose that the active free surface is initially appreciable. Hence some current flows starting a t U = 0.7 v. The tendency for the surface coverage to drop abruptly is opposed by the flow of methanol to the electrode which is large (compared to CO) at the concentrations studied. There is firially a n abrupt decrease in surface coverage a t U > -0.8 v. and a corresponding abrupt incrcase in current. “Activation” and “hysteresis” effects observed in preliminary experiments with formlc acid and formaldehyde suggest that the reactant-pair mechanism might be of general application. The activated complex would involve an adsorbed water molecule (or perhaps OH- in basic solutions) in all cases.
