S. GILMAK
1898
Vol. 67
THE MECHANISM OF ELECTROCHEMICAL OXIDATION OF CARBON MONOXIDE AND METHANOL ON PLATINUM. I. CARBON MONOXIDE ADSORPTION AND DESORPTIOS AND SIMULTASEOUS OXIDATIO-U OF THE PLATINUM SURFACE AT CONSTANT POTENTIAL1 BY S. GILMAN General Electric Research Laboratory, Schenectady, New York Received M a y 10, 1965
A platinum electrode may be covered reproducibly with adsorbed CO from 1 N perchloric acid saturated with a 100 or 1%CO gas mixture. Upon raising the electrode potential from U = 0.4 v. to U 0.8 v., a reproducible transient current-time trace may be obtained. The total corresponding charge includes contributions from CO originally adsorbed on the surface, from CO which readsorbs from solution, from surface oxidation, and from charging of the electrical double layer. It is possible to separate the total charge into these various components by application of anodic and cathodic pulses. The results indicate that oxidation of the platinum surface is rate-controlled by removal of adsorbed CO. It is also possible to break the CO-oxidation charge down into “bridged” and “linear” CO components. The results reveal no highly preferred order for oxidation of the two surface isomers.
>
Introduction I n previous papers2 methods were described for the rapid determination of CO surface coverage on platinum. These techniques were employed in study of the mechanism and kinetics of CO adsorptionzband in the qualitative explanation of the “polarization curve.”za Detailed information on the mechanism of electrooxidation of CO could not be obtained from the “polarization curve” since the anodic current rises very rapidly with increasing potential in such a way as to cause the system to undergo a transition from virtual inactivity to mass transport-control over a very narrow range of potentials. I n this study, a well defined and reproducible surface state was first established. The adsorption of CO and “oxygen” was then followed with time after the application of a potential step, and before the electrode reaction became transport-limited. These studies permit us to reach interesting conclusions not only on CO adsorption, but also on platinum surface oxidation in the presence of an adsorbate. Analysis of the adsorption data, and the currenttime relationship measured a t constant potential, allow us to come to some conclusions on the mechanism and kinetics of electrochemical oxidation of CO. This work will be reported in the second paper of this series. Evidence will also be presented to support the conclusion that methanol and other simple organic molecules form similar activated complexes in their anodic oxidation. Experimental The equipment and reagents used have been described previously.2 The electrode was a length of annealed C.P. platinum wire of 0.020-in. diameter, having a geometric surface area of 0.062 cm.2. The saturation coverage of theelectrode with hydrogen, &BE,determined as described previously,*bwas 0.310 mcoul./ cm.2, suggesting a roughness factor of 1.5. All measurements were made in a bath thermostated a t 30”. All potentials are referred to a reversible hydrogen electrode in 1 N HC10,. The potential functions employed are diagrammed in Fig. 1. The procedure during each step of each sequence is summarized in Table I, along with the significance of the procedure. The (1) This work was made possible by the support of the Advanced Research Projects Agency (Order No. 247-61) through the United States Army Engineer Research and Development Laboratories under Contract Number DA-44-009-ENG-4853. (2) (a) S. Gilman, J. Phgs. Chem., 66, 2657 (1962); (b) S. Gilman, i b i d . , 67,78 (1963).
general philosophy of the pre-treatment steps already has been discussed.2b
Results
I. The Constant-Potential Oxidation of CO on the Electrode Equilibrated with a Saturated Solution of 1% CO in 1 N HC10, A. Current-Time Traces at Constant Potential.-The rate of surface-coverage with CO under the experimental conditions employed here was found previouslyzbto be transport-controlled until high coverages are achieved. Further, the equilibrium surface coverage a t this partial pressure of CO (0.01 atm.) is close to that obtained a t 1 atm. It is therefore possible to cover the surface with CO in a reproducible manner by stirring the solution and allowing excess time for equilibration. This was accomplished by means of sequence I, Table I. By allowing the solution to become quiet before impressing the test potential, U , it was possible (sequence I) to restrict the flow of CO to the surface so as to make the equivalent “diffusion charge” negligible with respect to a monolayer of CO. This conclusion will be supported in the next paragraph. Equilibrium surface coverage with CO under these experimental conditions is equivalent to 0.40 mcoul./ of charge, (C&O)~, which must be supplied to convert CO to Con. In the unstirred solution, and for short periods of time, maximum additional CO is brought to the surface by semi-infinite linear diffusion. The equivalent maximum theoretical charge density, ( Q d ~ ~ ) m amay x , be expressed as a function of time by integration of the familiar equation expressing current as a function of time,za resulting in the relationship (Qdco),,x = 1.10~’/~ mcoul./cm.2 for a soln.
satd. with 100% CO (1) or ( ~ ~ c o ) , ,= 0.0117’/* mcoul./cm.Z for 1% CO,
assuming applicability of Henry’s law
(2)
For 7 = 1 see. (a longer interval than encountered in any of the reported experiments), (&d~o)max = 0.01 mcoul. or only 2.5y0 of the charge equivalent for full coverage with CO, and hence may be neglected. Representative current-time traces (corresponding
Sept., 1963
AlECHA4NISM O F
ELECTROCHEMICAL OXIDATION
OF
co AND AIETHANOL OK Pt
1899
TABLE I Purpose
Procedure
Fig. no.
Step no.
la
A
la
B
la
C
la
C
la
D
Ib
A-D
lb
E
lb
F
IC
A-D
IC
E
IC
F
117
1a
A-D
(1) to(5). Sameassequence I above with 100% CO substituted for 1%CO.
V
lb
A-F
VI
1d
A-D
Same as sequence 11, but lOOyo Same as sequence 11. CO substituted for li yo CO, T E = 1msec.; speed of linear anodic pulse = 887 v./sec. (1)-(5). Same as sequence I\’. (1)-(5). Same as for sequence
Sequence no.
I
I1
IT1
Final result
(1) To keep s o h . satd. with gas (1) Bubble 99% argon, 1%CO while removing oxidizable mixture with paddleadsorbed impurities from stirring (200 r.p.m.) for 15 electrode surface. Also see. to deposit passive “oxygen” film. ( 2 ) To sweep away and dilute ( 2 ) Bubbling and stirring conmolecular oxygen protinued for 20 sec. and duced during last step. stirring without bubbling The passive film is refor additional 10 sec. tained. (3) To equilibrate surface with (3) Stirring is continued for 20 CO reproducibly. sec. Surface is reproducibly equili(4) To limit mass transport t o (4) The soln. is allowed to bebrated with CO. Concn. of semi-infinite linear diffucome quiet for 1.5 min. CO near surface = bulk concn. sion during next step. Soln. is quiet. Current-time trace a t potential (5) Oscilloscope is triggered a t (5) Record current-time trace. U. the beginning of the potential step. Electrode brought to potential U (1) t o ( 5 ) . S a m e a s ( l ) t , o ( 5 ) o f (1) to(5). Sameas(l)-(!j)of under same conditions as in sequence I. Step D of sequence I. sequence I. variable duration TD. (6) Maintain potential a t 0.40 (6) To reduce surface “oxygen” Surface coverage with CO at the end of step D, Fig. 1B is mainso that no correction need (in quiet s o h ) fcr T E = tained, and surface coverage be made for partial sur10 msec. with LLoxygen” is negligible. face oxidation. (7) Measurement of Qtco, the Qtco may be determined. (7) Apply linear anodic sweep charge corresponding to (v = 362v./sec.) and surface coverage with CO measure corresponding a t end of step D, Fig. 1B. current-time trace. Electrode brought to potential U ( l ) t o ( 5 ) . S a m e a s ( l ) t o ( 5 ) o f ( 1 ) to(5). Sameas(l)-(Ei)of under same conditions as sesequence I. sequence I, step D of quence I. variable duration m. (6) Maintain potential ab 0.4 v. (6) To reduce surface “oxygen.” Surface coverage with CO a t the end of step D, Fig. ICis mainfor T E = 10 msec. tained and surface coverage with “oxygen” is made negligible. QH may be determined. (7) Apply linear cathodic sweep ( 7 ) Measurement of QH, the (v = 60 v./sec.) and charge corresponding to measure current-time H-codeposition on psrtitrace. ally CO-covered surface a t the end of step D, Fig. IC.
(1) to(5). Sameas(l)-(+!j)of sequence I.
IV. Id
E
(6) See footnote a.
Electrode equilibrated with soln. of 100% CO in 1iV HC104 RThen experiment begun a t potential U . Current-time trace obtained a t potential U . Qtoo may be meas. in presence of soln. satd. with 100% CO.
Electrode equilibrated with satd. ssln. of 100% CO before application of potential U , step D, Fig. Id. QR is determined by integration of traces obtained.
(6) To measure traces from which Q R may be determined. a Surface is reduced electrochemically and the current-time trace is recorded. Sensitivity 1 i3 adjusted so that the initial trace is obtained a t maximum current sensitivity over one decade of time. The current sensitivity is increased and the trace measured over succeeding decades of time until the current falls to an immeasurable small value (
5 0.4
$
6 0.3 0.2 0.1 J
0.5
1.0
1.5
2.0
1.0
0.5
1.5
0.5
2.0
1.0
1.5
2.0
T D / R ,msec.
Fig. %--Derivation of the charge due to surface oxidation in the presence of a saturated solution of 1%CO. The results are compared with similar results in the absence of CO: ( a ) U = 1.0 v., R = 10; ( b ) U = 1.1v., R = 5 ; (e) U = 1.2 v., R = 1. Risalinearscale factor.
adsorbed CO may be separated into the bridged and linear components by means of the equations (ref. 2 ) . @CO
= ~Q’H
t
QLco = Q c o
~ Q H -
Qtco
- Q’co
(7) (8)
QBcoand QLc0were determined graphically in Fig. 4. From Fig. 4 we see that there is no general preferred order for the decrease in either the bridged or linear structures with oxidation at potential C. At 0.85 v., the bridged form begins to be consumed earlier than the linear form, with considerable range of overlap. From U = 1.0 to U = 1.2 v., the initial order of preference is reversed, and again there is considerable overlap. This is an initial indication that the electrochemical reactivity of tlhe two forms is similar. This situation is in sharp contrast to the situation during adsorption of CO, where saturation coverage with the bridged form occurs before linear structures appear 011 the surface.2b The results suggest no conversion of one adsorption form to another, as might be suspected if one form appeared to be oxidized to the exclusion of the other. 11. Simultaneous Electrochemical Surface Oxidation During Constant-Potential Oxidation of CO at Potential U. A. Solution Saturated with 1% C0.As already explained, potential U is applied after the surface has been equilibrated with the solution. The equilibrium surface coverage with CO under these conditions has the charge equivalent (Qtco), = 0.40 mcoul./cm.2. As already demonstrated, the current a t potential y/ includes no appreciable contribution from diffusion of CO to the surface during the experiment, but does include contributions from oxidation of CO already on the surface, from oxidation of the surface, and from charging of the double layer. The same reasoning applies to the area (charge) under the currenttime trace. We may express the charge a t any time T D a t potential U as QT
[(Qtco)e - Q’co]
+ + Qo
QDL”
(9)
where total anodic charge (area under the current-time trace) after time TD a t potential U (Qtco). = charge corresponding to equilibrium coverage of QT
=
the surface with CO = 0.40 mcoul./cm.2 Qtco = instantaneous charge equivalent of adsorbed CO Qo = charge consumed in surface oxidation to any tinie
TD
a t potential U QDL” = charge added to the double layer after potential U has been established
The first two (bracketed) terms on the right-hand side of eq. 9 account for the charge due to oxidation of originally adsorbed CO. Qc’ = Qo QDL” = 40 may be obtained since QT may be obtained by graphic integration of the current-time trace. and Qtco have been measured directly as described above. The various experimental charges are plotted, and eq. 9 applied to obtain Qo’ graphically in Fig. 8. To make direct comparison between Q R , obtained in the absence of 60 with Qo’, we must first examine QDL”. This charge results from converting from state E in which the electrode is fully covered with CO to state F in which the surface is partially covered with CO and partially oxidized, all at constant potential U,. Then
+
QDL”
=
A”’ Cf dU - lua C, dU
For QR and Qo’ to be comparable QDL - QDL” must tend to 0. Combining eq. 5 and 10
All of the capacities in eq. 11 correspond to states in Tvhich the surface is partially or wholly covered with CO and/or oxygen, both adsorbates teiidiiig to drive the capacity d o ~ v n . 3 The ~ ~ ~presence of terms of opposite sign also tend to minimize the difference between QDL a d QDL”. Referring to Fig. 8, we see that 00’(surface oxidation charge in the presence of CO) tends to approach QR (surface oxidation charge in the absence of CO) toward large values of ‘1) when eco approaches zero. This signifies that final surface coverage with “oxygen” tends to the same value both iii the presence and ab(5) (6)
>I. W. Breiter, J . Electrochem. Soc., 109,42 (1962). AI. W. Breiter, Electrochzm. Acta, 7 , 601 (1962).
S. GILMAN
1904
---I% co
-100% co
0
(O’cde, I%CO
‘
\
0.10
0
Fig. 9.-Extent
0.20 0.30 Otc,, MCOULIcrn.2,
0.40
of surface oxidation as a function of amount of adsorbed GO; U = 1.0v.
sence of adsorbed CO, even though the surface oxidation is hindered in the former case. The reason for this tendency is that the surface coverage in the absence of CO tends to increase approximately with the log of timea and is hence sensitive to the order of magnitude of T D and not to its absolute value. Therefore once CO has been completely stripped off the surface, its oxidation may “catch up” (approximately) with that of the surface not originally covered with CO. It is this principle which makes it possible to make simple correction for extent of surface oxidation when iiieasuring Qtcoby means of a linear anodic sweep.2 By plotting the appropriate values of QH in Fig. 8, we see that the rate a t which Qo’increases closely parallels the rate a t which QH increases and hence the rate at which “free surface” is generated by the removal of CO. The simplest assumption is that CO simply physically masks surface oxidation sites. The charges QH and Qoare related to adsorption sites as QH
=
mFXn
(12)
QO
=
noFSo
(13)
where F = Faraday constant
Sg = site density available for codeposition of hydrogen So = site density available for oxidation of the surface = no. of electrons/site required for deposition of hydrogen =
ng
1 no = no. of electrons/site required for oxidation of the surface
If we make the assumption that So = SH,then no/nH =
Q‘/QH
=
-1 over a wide range of
TD
(14)
If the assumptions were correct, this would serve as evidence that the stoichiometry of the surface “oxygen” corresponds to PtzOor PtOH. It has already been suggested that the latter species might be the first and rapid step in the surface oxidation.3J6 By this interpretation the rate of this step is determined by the removal of CO and is hence slow enough to observe before the formation of higher oxides.3 It is not unlikely however that under our conditions structures of several different stoichiometries result along with uiioxidized sites SO that QO/QH 1. B. Solution Saturated with 100% C0.-The measurement of QR ‘v Qo in the presence of a saturated solution of 1OOyoCO was described in a previous section. These values are compared with Qo’ nieasured in
-
Vol. 67
the solution of 1% CO in Fig. 9. Q t C o (proportional to OCO) is used as the abscissa in Fig. 9. The experiment was started with the surface covered to the extent marked by the appropriate arrow and proceeded from right to left along the abscissa. At high QtPo the 100% CO system leads in extent of surface oxidation, but equalization occurs toward the center of the range, and the two systems give essentially identical results from half to zero coverage with CO. This tends to reinforce the suggestion that extent of surface oxidation is dependent in a simple may on extent of surface coverage with CO, since the time axes are considerably different for the two experiments (see Fig. 5). 111. Kinetics of CO Adsorption at U = 1.0 v.Charge data have been obtained and already discussed for 100% CO a t U = 1.0 v. By analysis of these data it is possible, in principle, to determine the rate of CO readsorption a t this potential. The total charge at any time T D iii this case may be expressed by an equation similar to (9) with the addition of a term accounting for CO-readsorption, &,, QT =
(QtcoIe - Q‘co
+ Qo +
QDL”
+
Qra
+
(15)
As previously, me will assume that QR = Qo QDL” = -Qo (see eq. 3 and 11 and accompanying text). Since all of the other charge quantities in eq. 15 have been determined, it is possible to determine Q,, graphically. This has been accomplished in Fig. 6. Q,, may iiow be compared with the maximum charge Qdcowhich can be supplied by semi-infinite linear diffusion by use of eq. 2. Actually, values of Q d c o so calculated appear 20% too low a t higher values of T D which is a reasonable variation due to uncertainty in the constants used. Calculated values of Qdco were therefore reduced by 20% to normalize, and the results plotted on Fig. 6 as Qri. From the figure it is seen that the readsorption follows the semi-infinite diffusion law well from 20 msec. on, and is thus clearly diffusion-limited in this range. If we express the readsorption as a current
I,, = nFk,,Cco (16) then assuming semi-infinite linear diffusion and constant value of k,,, I r a / I d must be less than 1.0 for sufficiently small values of 71, and equal to 1.0 a t large values of T D . 7 Observing the slopes of the Qdcoand Qra plots, we see that I r a / J d < 1 for T D < -5 msec. and Ira/Ia> 1 for -5 < 71) < -20 msec. This must be due to the fact that IC, is in fact not constant but a function of OCo = f(8co)
(17)
with k,, increasing as eco decreases. It has previously been reportedzb that the adsorption of CO at U = 0.4 v. starting with eco = 0 is already diffusion-limited after an adsorptioii time = 5 nisec. and departs from diffusion control only when the surface is almost entirely covered. In this experiment. we have started with the surface covered and proceeded in the opposite direction at U = 1.0 v. with qualitatively similar results. Since the appropriate differential equation subject to the conditions of eq. 17 cannot be solved, it is not possible rigorously to derive numerical rates for T D < 20 msec. At = 5 msec. the value of IF, and the use of eq. 16 (7) P. Dolahay, “New Instrumental hZethod6 in Electroohemlstry,” Intersclence Pubhslier6, Kew York, N. Y., 1954, p. 76.
BERR CONSTANT OF WATER
Sept., 1963
gives us Lra -- 0.07 cm./sec. (neglecting depletion of the diffusion layer) corresponding to 3% depletion of equilibrium coverage with CO. This allows us only to conclude that when the surface coverage has been re-
190s
duced to any significant extent from the equilibrium coverage, the rate of readsorption corresponds to perhaps as low as 0.1 cm./sec. and rapidly rises with decrease in CO surface coverage.
THE KERR CONSTANT OF WATER1s2 BY Department
w.H. ORTTUNG3 AKD J. A. MEYERS3,4 01Chemistry, Stanford University, Stanford, California Received M a v If, 1963
A light scattering apparatus was modified to allow measurement of the Kerr constant in conducting solutions of small molecules. Test measurements on carbon disulfide gave the accepted value within 3%. Results obtained for pure water a t 25" were B X IO7 = 3.72 rt 0.14, 2.89 f 0.04, and 2.72 f 0.11 for 436, 546, and 578 mp, respectively. An extension of Scholte's theory for ellipsoidal molecules, in combination with Pople's estimate of the dipole correlation factor, led to optical anisotropies, [al - l / Z ( ~ CY))] Z x 1024, of $0.019 for the three wave lengths. The possibility that part of the anisotropy arose from a distortion effect seemed likely, but could not be decided without more precise knowledge of the temperature dependence.
+
Introduction Recent considerationsj both empirical and theoretical, have indicated that measurement of the Kerr effect of water would be of great interest. I n a paper on the anisotropy of the 0-H bond in alcohols, Le Fevre, et aZ.,5 discussed possible values of the optical anisotropy of water. A positive anisotropy (pohrizability parallel to the dipole larger than the mean value) was suggested by their analysis, but the possibility of a negative value was also discussed. Since the anisotropy of the water molecule is small, the temperature-independent distortion effect included in the Born-Langevin theory by Buckingham and Pople6might be of importance. Klages' and Steppuhn8 considered the analysis of the Kerr effect in polar liquids from the point, of view of t,he model theories of the dielectric constant suggested by Onsager9 and Scholte.Io Buckingham and Raab" ga,ve a statistical mechanical treatment for polar liquids t,hat predicted, somewhat surprisingly, a negative orientation factor for water. The data obtained in the present work have been analyzed by an extension of the ellipsoidal cavity model theory of Scholte,lo in combination with Pople's estimat,e of Kirkwood's correlation factor.l2 It is shown that this theory gives a reasonable estimate of the dielectric constant and predicts that both the orientation and optical factors are positive in the Kerr effect of water. In addition, it is shown that the distortion of the molecule by the field probably is not negligible. (1) This work was supported b y Grant No. G 15555 from the ,National Science Foundation and b y a n institutional research grant of the American Cancer Society. (2) Presented in part a t the 144th National Meeting of the American Chemical Society, 120s Angeles, Calif., March 31-April 5 , 1963. (3) Department of ChemEatry, Tiniversity of California, Riverside, California. (4) This investigation was supported in part b y a Public Health Service fellowship, GPM-18,279, from the Division of General Xedical Sciences, Public Health Service. (6) C. G. Le Fevre, R. J. W. Le Fevre, B. P. Rao, and A . J. Williams, J . Chem. Soc., 123 (1960). (6) A. D. Buckingham and J. A . Pople, Proc. P h y s . SOC.(London), 8 6 8 , 905 (1955). (7) G. Klages, 2. Naturjovsck., l a , 669 (1952). (8) A. Steppuhn, ibid., lla, 912 (195ti). (9) L. Onsager, J . Am. Chena. Soe., 68, 1486 (1936). (10) T. G. Scholte, Physiea, 15, 437 (1949). (11) A . D. Ruoking!iam and R. E. Raab, J . Chem. Soc., 2341 (1957). (12) J. A . Pople, Proc. E o v . Soc. (London), 8 2 0 5 , 163 (1951).
Experimental Method.-The Kerr cell was arranged in the usual way, with the applied electric field a t an angle of 45" to the crossed polarizer and analyzer. The appreciable conductivity of water required the use of single-pulse fields. The retardation of the component of the light beam polarized parallel to the field with respect to the retardation of the component perpendicular to the field is
where 6 is in radians, 1 is the effective length of the sample subjected to the field, A is the wave length in air of the light beam, and nil and n i are the refractive indices of the sample parallel and perpendicular to the applied field. As usually defined, the Kerr constant is where E is the field strength in e.s.u. If I ois the change in light intensity upon rotating the analyzer from the crossed to the parallel position in the absence of birefringence, and if Is is the intensity passing through the crossed analyzer arising from birefringence, S may be calculated from
IdIo =
l/2(1
- cos S)cos r N 1/~6*cos r (3)
where r is the deviation of the analyzer from the crossed position. The approximation for small 6 values represented by the second equality could be used for all signals obtained in the present work. Since 10was too large to measure on a photomultiplier, a calibrated neutral filter was inserted in the beam. This method was used because the analyzer could not be set precisely to a small deviation from the cirossed position. The precision of setting the analyzer to the crossed position by observing the minimum intensity photoelectrically was &0.15", resulting in a negligible error, as may be seen from eq. 3. Apparatus.-A schematic of the apparatus is shown in Fig. 1. An absolute light scattering photometer13$14was modified for the Kerr effect measurements. The light source was repraced by a 200-watt mercury arc lamp and a low ripple d.c. power supply.16 A beam splitter consisting of a microscope cover glass was inserted after the aperture stop. The light reflected from the beam splitter passed through a stop identical to the field stop (3 X 6 mm.) and then through attenuators of darkened 35-mm. film to a monitoring RCA 1P21 photomultiplier whose purpose was t o allow compensation for light fluctuations. A Cenco No. 86140 adjustable slit was eubstituted for the aperture stop. The optical density 2, 3, and 4 neutral filters supplied with the instrument were calibrated to =t170on the Aminco photometer for each wave length (436, 546, and 578 mp). The Polaroid HY22 polarizer (13) No. 4-6210, Bulletin 2329, American Instrument Go., Silver Spring,
Md. (14) G. Oster, in "Physical Methods of Organic Chemistry," Vol. I , Part 3, 3rd Ed., A. Weissberger, Ed., Interscience Publishers, Inc., New York, N. Y., 1960, pp. 2128-2131. (15) P E K Laba, Inc., Palo Alto, California.