2880
S. GILMAN
Multipulse Potentiodynamic Studies of the Competitive Adsorption of Neutral Organic Molecules and Anions on Platinum Electrodes. I.
Competitive
Adsorption of Carbon Monoxide and Chloride Ions1
by S. Gilman General Electric Research d Development Center, Schenectady, New York
(ReceCed March 7 , 2966)
I n the absence of C1- and in the potential range -0.1 to 0.6 v, the structure of the adlayer, the maximum coverage, and the experimental rate of adsorption (initially diffusion controlled) remain constant for the adsorption of CO from 1 N HC104 solution. I n the presence of any extent of initial surface coverage with C1- in this potential range, the characteristics of the CO adsorption remain unchanged. Therefore, the system tends toward full coverage with CO and very small residual coverage with C1-. For any transient coverage with CO, there is an equilibrium coverage with C1- which is rapidly (probably transport limited) established by either desorption or adsorption of the ions. There is a simple linear relationship between equilibrium coverage with C1- and transient coverage with CO. It is possible to explain the inhibitive effect of dissolved chloride ions on the “polarization curve” on the basis of the findings of the competitive adsorption studies.
Introduction It has long been recognized that the adsorption of anions can have profound effects on the adsorption and reaction of neutral molecule^.^-^ Quantitative determinations of mixed surface coverages would seem a necessary prerequisite toward the understanding of such effects. Such determinations appear to be lacking, although there have been considerable studies of individual adsorbates.2 Since (on platinum) adsorbed organic and anionic species have distinctly different voltammetric properties, it is possible to devise a multipulse potentiodynamic (MPP) sequence which permits quantitative study of competitive adsorption. For the first study, CO and C1- were chosen because of availability of considerable previous information for these adsorbate~.~-”
Experimental Section The glassware, electronic equipment, etc. have been described previously.6 The electrolyte was 1 N perchloric acid, prepared from the AR grade acid and quartz-distilled water. The hydrochloric acid was AR grade. Gas mixtures of CO and argon were preThe Journal
of
Physical Chemistry
pared, bottled, and analyzed by the Matheson Co. using CP grade CO and “prepurified” grade argon. The working electrode was a length of CP grade platinum wire which was annealed in a hydrogen flame and sealed in a soft-glass tube so that a geometric area of 0.071 cm2 was exposed to the electrolyte. The “saturation hydrogen coverage,” sQH, measured using a linear cathodic sweep5 was 0.29 mcoulomb/ (1) This paper was presented at the spring meeting of the Electrochemical Society, Cleveland, Ohio, May 1-6, 1966. (2) (a) A. N. Frumkin in “Modern Aspects of Electrochemistry,” Vol. 3, J. O’M.Bockris and B. E. Conway, Ed., Butterworth Inc., Washington. D. C., 1964, Chapter 3; (b) P. Delahay, “Double Layer and Electrode Kinetics,” Interscience Publishers, Inc., New Tork, E.Y., 1965. (3) M,W. Breiter, Electrochim. Acta, 9, 827 (1964). (4) B. I. Podlovchenko and Z. A. Iofa, Zh. Fiz. Khim., 38, 211 (1964). (5) S. Gilman, J . Phys. Chem., 66, 2657 (1962). (6) S. Gilman, ibid., 67, 78 (1963). (7)S. Gilman, ibid., 68, 2098 (1964). (8) S. Gilman, ibid., 68, 2112 (1964). (9) T. B. Warner and S. Schuldiner, J. Electrochem. SOC., 111, 992 (1964). (10) R. A. Munson, J. EEectroamZ. C h m . , 5, 292 (1963). (11) S. B. Brummer and J. I. Ford, J . Phys. Chem., 69, 1355 (1965).
2881
C O J I P E T I T I V E L4DSORPTIOX O F C A R B O N M O N O X I D E AND C H L O R I D E I O N S
cm2. This implies a surface roughness factor (RF) of 1.4 based on RF = 1 for S&H = 0.21 mcoulomb/ cm2. All experimental quantities are reported on the basis of the geonzetric area unless otherwise noted. Measurements were made at 30 0.1". All potentials are reported against a reversible hydrogen electrode immersed in the adsorbate-free solution.
Procedures and Results I . CO Adsorption in the Absence of Chloride Ions. The adsorption on smooth platinum of CO has been studied recently by several investigator^.^,^,^-" Generally, the adsorption studies have been confined to a rather narrow range of potentials. For the purpose of this study, it was necessary to gather additional information on the effect of potential and partial pressure on the rate of adsorption and on the final surface coverage. Procedure. The sequence of Figure 1 was used in making these measurements. The electrode was normally held at a potential of 0.4 v. Step A was introduced to eliminate possible adsorbed anionic impurities. During step B adsorbed CO is oxidized to COz and desorbed. At the same time, a passive oxygen film is dep0sitc.d on the surface, preventing readsorption of CO. I n step C, the passive film is retained, allowing (in the absence of significant reaction at the surface) equilibration of the solution adjacent to the electrode ITith the bulk of the solution. For li > 0.6 v, qtep D lvas introduced (Tu = 10 msec) for prereduction of the surface. After reduction of the surface either during siep D or (for I' < 0.6 v) during the first few milliseconds of step E, the adsorption was allowed to proceed for time duration, T E . For C < 0.2 v, step F (5°F = 10 msec) was introduced to decrease the hydrogen-oxidation current during subsequent sweep G. The total duration of steps D, F, and G was aln-nys sufficient ly brief so that appreciable adsorption occurs only during step E. Traces corresponding to particular values of I' and T E appear in Figure 1. T-dues of Q r o (charge corresponding to quantitative oxidation of CO t o COJ may be obtained from such traces by subtracting the area under the trace for the clean surface from that of the CO-covered surface.j Results. For 0.1 atni partial pressure of CO, values of Qco were obtained for several values of adsorption time and potential. The results appear in Table I. For each adsorption time, the braces indicate a range of potentials over which Qco is constant with the indicated per cent average deviation. Measurement ( C - = 0.4 v and T E = 100 sec) at 0.01-1.0 atm partial pressure of CO, gave the constant value of 0.38 nicoulomb/cm2 for QCO.
-
2.0 r
0.5
I.o
2.0 Potential, volts
u J!!.ol oo.,sec 0.0 v 10Stc2Sec 30Sec A B C
1, D
1,
E
TF 1, F G
Figure 1. Linear anodic sweep traces corresponding to the adsorption of CO in the absence of C1-. The solution was 1 N HClO4 saturated with a gas mixture of 10% CO and 90% argon (30") and was paddle stirred (360 rpm). Step E is the adsorption step. Steps D and F are surface reduction and hydrogen removal steps, respectively, Trace msec: Tn = TP = 0. surface): I: = o,4 v: Tr- = Trace 2: U = -0.1 t o 0.7 v ; T E = 10 sec: T o = o for U 6 0.6 v ; T D = 10 msec for 'I = 0.7 v : T F = 0 for U > 0.3 v ; T F = 10 msec for C < 0.2 v. Trace 3: U = -0.2 v; T E = 10sec; TD = 0, T F = 10msec. I
For 0.1 atm partial pressure and TE = 10 see, traces corresponding to values of Lr from -0.1 to 0.7 v may be superimposed (trace 2 of Figure l), while the trace corresponding to -0.2 v is shifted to less anodic potentials (trace 3, Figure 1). For 0.1 atm partial pressure and T E = 100 see, the traces from C' = 0.4-0.8 v are identical; however, for potentials lower than 0.0 v, the trace for the CO-covered surface does not merge with that of the clean surface but is otherwise identical with the traces obtained at higher potentials. I I . Alixed Adsorption of CO and C1- Starting icith Initial CO Coverage of Zero. Procedure. d solution of 1 N perchloric acid, containing X HCl and with a partial pressure of 0.01 atm of CO, was employed. The pulse sequence used appears in Figure 2 . The electrode mas normally held at 0.4 v. Step d causes desorption of C1-6 but not of CO, which adsorbs at even lower potentials (Table I). I n step B, only 2% of a monolayer of C1- is adsorbed,' all previously adsorbed CO is oxidized to C o n and desorbed, and a passive oxygen film forms, preventing further adsorption of C1- or of CO. I n step C, the passive film is retained while the solution adjacent to the surface is equilibrated with the bulk (including elimination of the products COZ,Cln, and 0 2 , from step B). In step D, the passive Volume 7 0 , Sztmber 9
September 1966
S. GILMAN
2882
Table I: CO Adsorption in the Absence of C1TE = 1--ces
TE = 10 sec
I
Qco,
u, v
u, v
mcoulomb/cma
-0.20 -0.1 0.0 0.4 0.6 0.7 0.8
0.234 0.286 0.285, 0.301 3% av 0.284 dev 0.2681 0.179 0.111
mcoulomb/cml
-0.10
0.00 0.40 0.70 0.80
N
I
b
Lot 0.5
/
I
0.5
To :1.3 W.
1.0
1.5
2.5
,
1.0
I
1.0
1.5
10% A
2.0
2~ B
1.5
20
2.5
wrcmdL. volts
d
2.5
3 0 ~ To C
0
TcTrTe E F G
Figure 2. Linear anodic sweep traces corresponding to mixed adsorption of CO and C1- a t U = 0.6 v. The solution M HC1 and saturated with was 1 N HClOd containing a gas mixture of 1% CO and 99% argon (30",paddle stirred a t 360 rpm). Step D is the adsorption step. Steps E and F are used to eliminate adsorbed CO for separate determination of C1- coverage. Trace 1 (clean surface, measured in absence of CO and Cl-): TE = TF = 0. Trace 2 (surface covered with both CO and C1- from step D ) : T E = TF = 0. Trace 3 (surface covered with only Cl- from step D): TE = 2 msec; TF = 8 msec; the dashed traces correspond to CO adsorbed in the absence of C1under conditions approximating those for trace 2.
film is largely reduced within the first few milliseconds, allowing adsorption to commence. In the absence of the adsorbates, when T E = T F = 0, the trace obtained upon applying sweep G (traces 1 of Figure 2 ) is that The JOUTTMZ~ of Phvsical Chemistry
TE
,
-0.20
1
-s
7
Qco,
0.331 0.348, 0.354\ 0.375 4% av 0,340 dev 0.338) 0.182
t
=
.100 -----ces
Qco,
u, v
mcoulomb/cmZ
0.4 0.7 0.8
0.378 0.375, 0.378 l % a v
1
0.369)dev
for the clean surface. In the presence of the adsorbates the adlayer contains both CO and C1- after adsorption for time interval, T D . Hence, when T E = T F = 0, the application of sweep G results in traces (those marked 2 in Figure 2 ) which are characteristic of the mixed adsorption. If step E is introduced ( T E = 2 msec), the CO portion of the mixed adlayer is desorbed, the C1- is retained,8 and (after reduction of the surface for T F = 8 msec) the trace obtained (traces 3 of Figure 2) after application of sweep G is characteristic of the C1- portion of the mixed adlayer obtained after adsorption time, T D . It should be noted that the total duration of steps E, F, and G is sufficiently brief so that adsorption during these steps is negligible. The dashed traces correspond to CO adsorbed in the absence of C1- and were included for comparison with traces 2 . Results. The extent of surface coverage with CO and C1- may be derived from traces such as those of Figure 2 as follows. Using Figure 2a as an example, all charge lying to the left of the dashed tangent to trace 1 is disregarded (assumed largely capacitive). The charge AQ1-a is obtained by subtracting the area under trace 3 from the area under trace 1. The charges AQZ-l and A&*-3 are obtained in similar fashion. The charge AQ1-3 is equivalent to AQo, which in turn is the charge equivalent of "oxygen adsorption" blocked by C1- adsorption7 AQ1-3
=
AQo
(1)
Since 1 equiv of C1- blocks 2 equiv of oxygen AQo
=
2FI'cl-
(2)
where r C l - is the absolute surface coverage with C1in moles per square centimeter. The charge AQ2-1 comprises A& and also QCO (charge equivalent of quantitative oxidation of adsorbed CO to COZ, ref 5 ) . Hence AQ2-1
= Qco
- AQo
(3)
COMPETITIVE ADSORPTION OF CARBON MONOXIDE AND CHLORIDE IONS
D
2883
E F G
To, Sec Figure 3. Mixed adsorption of CO and C1-. TDis the adsorption time for both CO and C1- from 1 iV HCIOl containing 10-4 M HC1 and saturated with a gas mixture of 1% CO and 99% argon (30°, paddle stirred a t 360 rpm). A& is related to absolute and relative coverages with C1- by eq 2 and 7, respectively. Qco is related to absolute and relative coverages with CO by eq 4 and 6, respectively. AQo and &co were evaluated from traceb similar to those of Figure 2 .
1
I
I
I
0 8ca‘0
‘
Hence absolute surface coverages may be derived from the charges AQ1-3 and AQ2-3 using eq 1, 2 , 4 , and 5 . Results. Values of QCO and AQo were obtained for values of U from 0.2 to 0.6 v and as a function of adsorption time, TD. These values are plotted in Figure 3. -4 second representation of the data appears in Figure 4, in terms of fractional surface coverages. These are defined as
I
0.3 .
P
0.2.
0.1 ’
ob
I
I
1
1
0.2
0.4
0.6
0.8
I
I .o
where ( Q c o ) ~= ~0.38 ~ ~mcoulomb/crn2 and
U, VOLTS Figure 4. The dependence of C1- surface coverage on potential for different constant CO surface coverages (derived from data of Fignre 3 ) .
also Qco = 2Frco
(4)
where I’co is the absolute surface coverage with CO. Combining eq 1 and 3 AQ2-3
= QCO
(5)
where SQH = 0.29 mcoulomb/cm2. These definitions of fractional surface coverage are somewhat arbitrary but may be converted to absolute coverage by means of eq 2 and 4. I n particular, 8ci- is based on the assumption that one adsorbed chloride ion formally occupies one hydrogen adsorption site. I I I . Mixed Adsorption of CO and C’l- Starting with Initial CO Coverage Greater Than Zero. I n section 11, Volume 70, .?‘umber 9 Septamber 1966
S.GILMAN
2884
I
I
1
0.3+
h
I
3.0
C
5
2.0
\ 0
E U
I .o
0
I
1
5
IO
15
( T b t l ) , Sec
Figure 5 . Sequential mixed adsorption of CO and C1-. The 1 S €IC104 contained Ji‘ HCI, was saturated with a gas mixture of 1‘; CO and 99C6 argon (30’)- and Fa: paddle stirred (360 rpm). Adsorption of CO occurs during both steps D and E for T D 1 see. Adsorption of C1- occurs only during step E (for 1 see). Qco is related to absolute and relative coverages with CO by eq 4 and 6, respectively. SQO is related to absolute and relative coverages with Cl- by eq 2 a i d 7 , respectively. A& and QCOwere evaliiaied from traces similar to those of Figure 2 .
+
above, adsorption n-as conducted a t potential U so that CO and C1- could adsorb siiizultaneously. I n the experiments described below, CO was first adsorbed a t :L potential (0.06 v) a t which only CO could adsorb. After achieving :Ldesired CO coverage, the potential was then raised to 0.6 v, a t which potential both CO and C1- could adsorb. Procedure. The pulse sequence used appears in Figure 3. Steps A4-C of the sequence are identical with those employed in section I1 above. I n step D, CO only was adoorbed for time interval T D . I n step E, further adzorption of CO could occur a t 0.6 v, along with adsorption of C1-. The rest of the sequence served t o obtain A& and QCO, asdescribed in the previous section. Results. T-nlueb of charge are plotted against the 1) allowed for CO adsorption in total time (7‘1, Figure 5 . Thc solid curves passing through the data points are identical with those measured in the previous section, for adsorption at 0.6 v. I V . The (‘Polniixalion Curve” for CO. The linear anodic swecp (las) traces of Figures 1 and 2 correspond to CO adsorbed beJore the application of the rapid sweep. There is thus no appreciable “turnover” (additional adsorption and reaction) during the sweep. For exami-
+
The Journal of Physical Chemistry
0.8
1.0
1.2 1.4 Potential, v o l t s
1.6
1.8
Figure 6. Anodic “polarization curves” for CO. The traces were recorded during the positive-going portion of a periodic triangular sweep (sweep speed, u = 0.04 v/sec) operating between 0.4 and 1.8 v. Trace 1 was obtained in the absence and trace 2 in the presence of ,If HCl. The 1 N HC104 electrolyte was saturated with pure CO (30”) for both traces and was paddle stirred (360 rpm). The hatched areas correspond t o regions of oscillation of the c i u ~ e n t .
nation of the qualitative effects of C1- adsorption on steady-state anode performance, a slow periodic triangular sweep mas applied alternatively in the presence and absence of dissolved C1-. The results appear in Figure 6 .
Discussion I . Adsorption of CO in the Absence of Cl-. Structure of the Adsorbed Layer. Previous studies of the CO adlayer at 0.4 v6 suggested that the adsorbed layer retained the coniposition of the adsorbate in the gas phase. The results mere also consijtent with the conelusion that the adlayer comprised CO bonded to the surface in both bridged and liriear configurations. The las traces measured for -0.1 v C 0.7 v after 10-sec adsorption time (trace 2, Figure 1) are identical. This suggests that the adlayer formed over this range is constant in composition and structure. On the other hand, the trace obtained at -0.2 x- i* noticeably shifted to the left on the potential axis (trace 3, Figure 1). At medium potentials (e.g., 0.4 I-, ref G ) , such shifts sometimes correspond to decreasirig coverage for an adsorbed phase of constant However, since the oxidative charge corresponding to trace 3 is almost identical with that for trace 2, the observed potential shift is not simply a surface coverage effect but must likely represent a change in composition or
<
0.04, BCO < 0.8. Equation 11 describes the empirical behavior of the system over most of the range of conditions encountered. Referring to eq 11, we may now summarize the behavior of the system. I n the presence of excess CO, OCO will eventually become 1.O, with the adsorption transport controlled until high coverage is achieved. As CO adsorbs, C1- will rapidly adsorb or desorb to yield the equilibrium values of 0 ~ 1 -predicted by eq 11. Equation 11 mill fail to hold once Ocl- approaches 0.04 or BCO exceeds 0.8. I n the absence of CO (eq 10) the results resemble those previously found for other systems. A linear dependence of coverage (in the higher range of coverages) on potential was found for the adsorption on mercury of a number of anions, including C1- (calculations made by Parsons on the experimental work of Grahame,”), and for the adsorption of I- on platinum.I8 Dependence of tiurface coverage on the logarithm of activity of the ion in solution has been found for Iadsorption on mercury.Ig We may now consider the origin of the dependence of C1- surface coverage upon CO surface coverage, as given by eq 11. I t is clear that CO does not simply “crowd out” C1- since the combined coverage with both adsorbates is always less than a monolayer. The coverage with specifically adsorbed anions may be expected t o depend on the heat of adsorption of the ion (in the absence of a field)” and on the charge stored in the ionic double layer.*O We might expect variations in the heat of adsorption of the anion (in the presence
2887
CHLORIDE I O N S
of CO) to affect the y intercept, rather than the slope of the Ocl- us. potential plot. Since the slope is affected, an explanation based on the charge is preferred. The adsorption of a neutral molecule drives down the double .layer capacityzb and hence the charge stored in the double-layer a t fixed potential. Thus the linear dependence of ecl- upon OCO (at fixed potential) may reflect a linear dependence of charge upon CO coverage and a linear dependence of C1- coverage upon the charge. I V . The Efect of Chloride Ion Adsorption on the Electrochemical Oxidation of CO. A proper quantitative treatment of the effect of adsorbed chloride ions on CO oxidation kinetics would require considerable information of the type previously1z presented for the C1-free system. Such a treatment will be postponed to some future date. However, the information presently a t hand makes possible a qualitative discussion of the topic and, particularly, of the CO “polarization curve.” The dashed traces of Figure 2 correspond to the oxidation of adsorbed CO and of the platinum surface, in the absence of adsorbed chloride ions. The characteristics (shape, position) of the trace are determined by a variety of factors (sweep speed, coverage, etc.). However, the general tendency of the initial rise in current to shift to more anodic potentials with increase jn surface coverage (or adsorption time) is probably a manifestation of the second-order kinetics law followed for the oxidation of adsorbed C0.I2 The traces marked 2 in Figure 2 correspond to oxidation of CO and of the surface, in the presence of adsorbed chloride ions. For any fixed coverage with CO (or fixed adsorption time), the adsorbed C1- results in increased overvoltage for the oxidation of the adsorbed CO, corresponding to the shift from the dashed trace to trace 2. This corresponds approximately to the shift for the surface oxidation trace measured in the absence of C1- (trace 1) to more anodic potentials when C1is adsorbed (trace 3). Under these particular experimental conditions, a plot of the potential shift os. Ocl- mas found to be linear with a slope of 0.75. The potential shift represents an additional overvoltage due to ion adsorption; hence, the current corresponding to the oxidation of adsorbed CO may be represented by I = Ii exp( -m6cI-) where I and I i are the currents measured for CO oxidation in the presence and absence, respectively, of (17) N.F. Mott and R. J. Watts-Tobin, Electrochim. Acta, 4,79 (1961). (18) K.Schwabe and W. Schwenke, ibid., 9, 1003 (1964). (19) D.C. Grahame, J . A m . Chem. Soc., 8 0 , 4201 (1958). (20) R. Parsons, Trans.Faraday Soc., 51, 1518 (1955).
Volume 70,Number 9
September 1966
2888
adsorbed chloride, but under otherwise identical conditions. The term mOcl- represents the increased overvoltage due to the ion adsorption and m may vary with the conditions of the experiment (but was found to have the value 0.75 for the experiments of Figure 2, as mentioned above). The ion may exert the type of effect given by eq 12 through a “double-layer” eff ect.2 Current-voltage curves obtained at low sweep speed for stirred, CO-saturated solutions appear in Figure 6. The general characteristics of the “polarization curve” for the Cl--free system (trace 1, Figure 6) have been discussed p r e ~ i o u s l y . ~During ~’~ the anodic portion of trace 1, the current remains negligibly small until the potential rises above 0.9 v. At 0.92 v, there is an abrupt increase in current to a value corresponding to the CO-transport limit and then a gradual decrease in anodic current until the onset of appreciable oxygen evolution (1.6 v). The decreasing current accompanies buildup of surface “oxygen.” The effect of addition of C1- (trace 2, Figure 6) is to shift the initial vertical rise in current to 0.95 v (30-mv shift) and then to decrease the current (up to 40%) over the range extending from 0.95 to 1 . G v. From section 111, above, we know that the surface is covered with CO to the same extent (approximately a monolayer) below 0.9 v, both in the presence and absence of C1-. I n the presence of dissolved C1-. there is additionally 4% of a chloride ion monolayer on the surface. This small C1- coverage decreases the reactivity of the previously adsorbed CO (according to eq 12) and causes the small (30-mv) shift in potential. It is to be emphasized that the shift is small because of the tendency for CO to drive the
The Journal of Physical Chemistry
S. GILMAN
anion off the surface. Conversely, the more pronounced effect of the specifically adsorbed ion on the kinetics of oxidation of methanol3 and of ethanol and acetaldehyde4 may correspond to a decreased tendency for the anion to be desorbed by the organic adsorbate. After the initial surge in current, OCO = 0 for both traces of Figure 6. Hence, the decreasing current of trace 1 does not reflect a decrease in reactivity of the adlayer but possibly the rate of some “surface activation” step (essentially the rate of adsorption a t OCO = 0, ref 12). For trace 2, once the coverage with CO drops to zero (at 0.95 v), the coverage with C1- may rise close to its maximum value of OCI- = 0.5.3 The first sharp drop in current corresponds to approximately the time required for C1- transport from solution. This relatively high coverage with C1- simply produces an effect similar to that produced by increased coverage with “oxygen.” Toward 1.6 v, where the combined coverage with C1- and “oxygen” probably approaches a monolayer for both systems, traces 1 and 2 of Figure 6 tend to merge.
Acknotcledgiiaent. The author wishes to acltnowledge helpful discussions with 11. W. Breiter and F. Will. This work is a part of the program under Contracts DA-44-039-AMC-479(T) and DA-44-009ENG-4909, ARPA Order No. 247, with the U. S. Army Engineer Research &- Development Laboratories, Ft. Belvoir, Va., to develop a technology which will facilitate the design and fabrication of practical military fuel cell power plants for .operation on ambient air and hydrocarbon fuels.