1198
P. V. POPAT AND N. HACKERMAN
solution with respect to the most amorphous layer. 6. The activation energy for the solution of Sp. B. silica in dilute caustic is 18 kcal./mole and 17.8 kcal./mole for S. L. silica in water. 7. This study shows that it is possible to
Vol. 62
characterize silicas by solubility and kinetic data for solution. Acknowledgment.-Thanks are due the Research Department of the Diamond Alkali Company for their financial support of this work.
CAPACITY OF THE ELECTRICAL DOUBLE LAYER AND ADSORPTION AT POLARIZED PLATINUM ELECTRODES. I. ADSORPTlON OF ANIONS BY PRANJIVAN VELJIPOPAT AND NORMAN HACKERMAN Department of Chemistry, University of Texas, Austin, Texas Received March S, 1968
The differential capacity of the electrical double layer a t polarized platinum electrodes has been investigated as a function of the type and concentration of anions. The method of charging curves, utilizing a square-wave signal was employed. The “hump” in the potential-capacity curve is characteristic of the anion involved and is explained in terms of adsorption or desorption of the anion. The tenacity of adsorption is found to increase with the covalent character of the adsorbed anion as NOa- and F- < SOI-- < C1- < Br- < I-. Evidence is presented to indicate that the adsorbed iodide, bromide and chloride ions are dehydrated before their anodic discharge. Fluoride and nitrate ions do not appear to be dehydrated even on the anodic side of the zero point of charge. A similarity is established between polarized platinum and mercury electrodes and the location of the zero point of charge for platinum is indicated. It is observed that hydrogen over-voltage on platinum is a function of the type of anions present.
Introduction There have been several investigations of the differential capacity of the electrical double layer (e.d.1.) a t polarized platinum electrodes. Ershlerl and Dolin and Ershler2 studied the capacity of smooth platinum electrodes in acidic and alkaline media as a function of the frequency of alternating current in order to determine the slow step in the discharge of hydrogen ions on platinum. Schuldiner3 used capacity data to estimate true surface area of platinum electrodes. Robertson4 investigated the effect of a.c. frequency and HC1 concentration on the capacity of platinum electrodes. Kheifets and Krasikov6 used capacity measurements to study the effect of surface active substances on the hydrogen overvoltage on platinum. A review6 of these studies showed little similarity between the several results. One reason for this may be found in the fact that no two investigators used the same experimental conditions. A recent paper reported7 on the capacity of polarized platinum electrodes in KBr and KI solutions. All of these measurements were made with impedance bridges. This method is the most accurate but it is limited to microelectrodes which, with solid metals, are difficult to prepare in a reproducible manner. Moreover, it is tedious and time-consuming.8 Recently, Brodd and Hackermang reported a variation of the charging curve method. This is (1) B. Ershler, Acta Physicochim. U.R.S.S., 7 , 327 (1937). (2) P. Dolin and B. Ershler, ibid., 13, 747 (1940). (3) 5. Schuldiner, J . Electrochem. Soc., 99, 4 8 8 (1952). (4) W. D. Robertson. ibid., 100, 194 (1953). (5) V. L. Kheifets and B. S. Krasikov, Doklady Akad. Nauk S.S.S.R., 94, 101 (1954). ( 6 ) D. C . Grahame, Ann. Reu. Phys. Chem., 6 , 337 (1955). (7) J. N. Sariuoosakis and M . J. Prager, J . EZectrochem. Soc., 104,454 (1957). (8) B. Breyer, Revs. Puw and A p p . Chem., 6 , 249 (1956). (9) R . J. Brodd and N. Hackerman, J . EZeclrochsm. Soc., 104, 704 (1957).
suitable for projected surface areas of about 2 and is relatively fast. The method utilizes a square-wave signal of known frequency to obtain time potential traces on a cathode ray oscilloscope. McMullen and Hackermanlo modified the method and obtained reproducible results with mercury and with several solid metal electrodes. The results with mercury were in good agreement with those obtained by the impedance bridge method. These authors showed that for sufficiently small charging times (ie., when the change in the potential of the test electrode is a linear function of time), the capacity of a polarized electrode can be obtained from the relationship
where E , is the potential of the test electrode, t is the time, Ei is the magnitude of the input squarewave signal, R, is a standard resistance through which the square-wave signal passes before entering the test electrode and C is the differential capacity of the electrode. This method is currently being used here for studies of the electrical double layer a t platinum and other solid metal electrodes. The problem of immediate interest was to determine whether there is a similarity between polarized platiiium and polarized mercury electrodes. Robertson4 and Sarmousakis and Prager’ found no similarity between the two electrodes. Another important aspect of the work was to try t o locate the zero point of charge (z.P.c.) for platinum by this method. Other problems of interest concerning the electrical double layer in general have been discussed by BreyerS8 One of these is whether ions in the layer are. solvated. Anions in the layer are believed to be much less solvated than cations and it is assumed that, with the possible exception of (10) J . J. MoMulIen and N. Hackerman (to be published).
Oct., 1958
ADSORPTION OF ANIONSAT
THE
fluoride,ll anions are largely bonded to the electrode by covalent bonds. However, no attempt had been made to correlate the degree of covalent character of an adsorbed anion with its polarizability. Whether or not cations a t the interface are hydrated is even less clear.8 It therefore was decided to investigate the capacity of polarized platinum electrodes as a function of anions (and a few cations) over a sufficiently wide potential range.
POLARIZED PLATINUM ELECTRODE 70
60 A
e
I
I
I\,
I
I
I
1199 I
The circuit diagram used in this work was similar to that already described.lO The frequency of the input squarewave signal was 500 C.P.S. and the standard resistance R, was 15,000 ohms in all cases. The change in the electrode potential due to the square-wave signal was about 10 mv. The all-glass, Pyrex cell used in all experiments was similar to that described earlier.g A large area platinized platinum cylindrical gauze was used as a non-polarizable auxiliary electrode. All potential measurements reported in this study were made against a saturated calomel electrode (SCE) and all potentials discussed are calculated to SCE where necessary. Solutions were prepared with triple distilled, conductivity water (1 X lo-’ mho/cm.). Analytical reagent grade chemicals were further purified by at least one recrystallization from conductivity water. All solutions (except those containing bromide or iodide) were purified by pre-electrolysis a t 3 ma./cm.a for about 20 hours under a helium atmosphere. Bureau of Mines grade A helium reported as 99.99770 pure (with less than 10-6per cent. of either oxygen or water vapor) was bubbled through the pre-electrolysis cell during pre-electrolysis and through the experimental cell throughout the capacity measurements. The rate of helium bubbling during capacity measurements was maintained uniformly low. The helium passed through a presaturator containing conductivity water, and passed out through a similar trap to prevent back diffusion of air. Contamination by grease or other organic materials was carefully avoided. Platinum wire (0.016’’ dia.) of reagent grade, supplied by Baker Co., New Jersey, was used throughout. Test pieces of the wire, about 9 cm. long, were sealed in 6 mm. Pyrex tubes. The length of the test electrode actually exposed to the solution was 7.5 cm., which gave an apparent area of 1 The tube holding the test electrode was sealed to a Pyrex glass holder which fitted the cell. Contact between the test electrode and the rest of the circuit was made by a column of mercury. Each electrode was used only once. Of the several methods of treating the electrode that were tried, the following, which gave reproducible and consistent results, was used throughout. The electrode was thoroughly washed with conductivity water and then slowly heated to bright red color for several minutes. It was allowed to cool to room temperature in a clean Pyrex tube holder and then introduced into the cell. All experiments were carried out a t room temperature (25 to 30’) unless otherwise stated.
Experimental Results Variation of Capacity with Time.-Preliminary experiments indicated that the capacity of polarized platinum electrodes at a given electrode potential varies with time. Curves of Fig. 1 are representative of this behavior. Curve (a) was observed typically when there was little change in capacity with a change in polarizing potential. Curves like (b) and (c) resulted when there was considerable change in capacity with a change in polarizing potential. Similar behavior was observed also by Sarmousakis and Prager.’ Curves (d) and (e) were for the solutions containing a surface active (11) D. C. Grahame and B. A. Soderberg, J . Chem. Phya., 22, 449 (1954).
1
I
50
k
3 40
P s230
.d
Experimental
I
u
20
40 60 80 100 120 140 Time (minutes). Fig. 1.-Variation of capacity with time: (a) 0.1 M Na2S04 (0.4 v.); (b) 0.1 M Na,SOc (0.1 v.); (c) 0.1 M 7 X M/1. nK2Cr207 (0.75 v.); (d) 0.1 M NaPSOl caproic acid (1.0 v.); (e) 0.1 M NazSOl 7 X 10-o M/1. n-caproic acid (0.9 v.). 0
20
++
substance. Delahay and Trachtenberg12 observed similar behavior with a mercury electrode in presence of n-hexanol and showed that adsorption of a surface active substance a t mercury electrode is diffusion controlled. The variation of capacity with time in the absence of a surface active substance, however, cannot be explained on the basis of diffusion alone because the concentration of ions in these solutions is so large that their diffusion is not likely to be of major importance. It is more likely that the process of ionic adsorption-desorption itself is slow as indicated later. In any event, for the differential capacity curves, enough time was allowed for the capacities to reach steady values. The Effect of Anions.-The results obtained with neutral 1 N solutions of NaF, KCI, KBr, KI, KN03 and K2S04,respectively, are given in Figs. 2 and 3. These solutions were selected so that a comparison could be made with similar work on mercury and, wherever possible, with platinum. Substitution of NaF for K F was not expected to interfere with the effect of fluoride ions. Curves for NaF and KI solutions represent the data of two independent experiments which gave practically identical results. Each of the rest of the curves of Figs. 2 and 3 represents the average of at least three independent sets of data which were reproducible to within less than 5%. The direction of change of polarizing potential in all cases (except for KI) was from the anodic to the cathodic side. With KI, reproducibility was poor in going from the anodic to the cathodic potential probably because of interference by traces of iodine liberated a t the more anodic potentials. The results for 1N IK!l(Fig. 2) are in satisfactory agreement with those for 1 N HC1 obtained by R o b e r t ~ o n . ~The minimum capacity of 19.5 pF./ cmn2at 0.8 v. obtained in the present work agrees well with his value of 20 kF./cm.2 a t 0.79 v. However, his curve does not show the characteristic (12) P. Delahay and I. Trachtenberg, J . Am. Chem. Soc., TO, 2355 (1957).
1200
P. V. POPAT AND N. HACKERMAN
Vol. 62
“hump” obtained in the present studies. This may 2-7 is the “hump” that appears in the intermediate be due to the fact that he used solutions of low pH range of potentials in each case. Similar humpB in and therefore reached the potential of hydrogen the potential-capacity curves have been reported evolution before the hump could appear. for mercury electrodes in various electrolyte^,^^*'^ Similarly the results with 1 N KBr and K I solu- and Grahame16has proposed an “ice-layer” theory tions (Fig. 2) agree well with those of Sarmousakis to explain the observed effects of electrolyte conand Prager’ who obtained a minimum capacity of centration and temperature on the capacity humps about 18 pF./cm2. at 0.6 v. with 1 N KBr and of for mercury electrodes in NaF and KNOI solutions. 18 pF./cm.2 at about 0.1 v. with 1 N KI. How- According to this theory, the surface of the elecever, these authors did not go far enough in the trode may be covered with a “pseudo-crystalline cathodic direction and therefore missed the hump ice-like layer” of solvent through which anions pass shown in the present work. The shapes of the only with difficulty, but without activation energy. curves obtained by them do indicate the beginning This “semi-rigid” layer is supposed to melt only of such a hump, but they did not discuss this. The gradually since the hump disappears gradually with value of the maximum capacity of about 40 rise in temperature. pF./cm.2 for K I solution obtained in the present A different and perhaps more satisfactory inwork does not agree with their value (extrapolated) terpretation of similarh umps found with platinum of about 55 pF./cm.2. This may be due to some electrodes is proposed here, Before doing so, the hysteresis effect in the present work. However, the significance of other parts of curves of Fig. 2 is potential at which the capacity maximum occurs is discussed briefly. believed to be more important and here the agreeThe steep rise of capacity at the extreme left of ment appears good (on extrapolating their curve). each curve (except for NaF) represents the pseudo The shape of the curve with 1 N K2S04is similar capacity due to the discharge of the halide ion under to that obtained by Kheifets and Krasikovs with consideration. The potential at which this occurs 0.5 M Na2S04. Their value of minimum capacity is characteristic of the anion involved. The steep (30 pF./cm.2) in the anodic region does not agree rise at the extreme right of each curve (Fig. 2) with the present value of about 18 pF./cm.2 The represents the pseudo capacity due to the discharge internal self-consistency of the present results, to- of hydrogen ions in each case. The potential at gether with the agreement of results for KBr, which this occurs also appears to be a function of KC1 and K I solutions with those of other workers the anion involved. suggests that the present data are more reliable. Between these two limits, the electrode behaves The Effect of Electrolyte Concentration.-This more or less as an ideal polarized electrode. At is illustrated by Figs. 4 and 5 where each curve the minimum to the left of the hump, the anions represents the average of at least three independent populate the inner region of the e.d.l., whereas at sets of data which were reproducible to within ex- the minimum t o the right of the hump the cations perimental error. The precision of measurements populate the inner region of the e.d.1. in 0.01 N solution was slightly lower because of the According to the interpretation proposed here, large I R drop. Also the range of capacity within each hump in the potential-capacity curve for which the method in its present form gives precise platinum electrodes in various anions represents a data is limited to 10-60 pF./cm.2. Beyond 60 rearrangement of the e.d.1. The main process durpF./cm.2 the slope of the time potential trace was ing this rearrangement is the desorption of anions too small to measure accurately and below 10 in going from left to right. Similar humps (or pF./cm.2, the time potential pattern extended peaks) due to desorption of surface active subbeyond the range of the oscilloscope. stances have been reported. l 4 Particularly, the Dilution of the chloride solution (Fig. 4) has desorption of a surface active anion gives a single considerable effect on the capacity of platinum peak.6 electrodes in the intermediate range of potentials. In the present case, each hump is characteristic The capacity maximum with 1, 0.1 and 0.01 N of the anion involved. The most important feasolutions of KC1 appears at -0.40, -0.30 and -0.25 ture of the hump in this connection is the potential v., respectively. A similar dilution effect was at which the capacity maximum occurs in each case. observed with two KBr solutions. The capacity These valuesl6 are -0.80, -0.55, -0.40 and 0.00 v. maximum appeared at about -0.55 v. with 1 N for I-, Br-, C1- and F- ions, respectively. AnKBr and at - 0 . 5 0 ~ .with0.1NKBr. Thedilution other important feature of each hump is the poeffect of KI and NaF solutions was not studied. tential range within which each hump occurs. This Dilution of nitrate solution or of sulfate solution decreases in the order: I-> Br-> CI->F-. (Fig. 5 ) showed very little effect on the capacity of On the basis of these observations, it can be platinum electrodes.l 3 concluded that the humps with various anions Some results on the effect of temperature and of under investigation must be due to their desorption change in cations are given in Figs. 6 and 7, re- as the potential is made more cathodic. The rise spectively. in capacity during desorption probably is due to a change in the ratio of the dielectric constant to the Discussion thickness of the double layer. most interesting Adsorption of Anions.-The (14) D.C.Grahame, Chem. Reus..41, 441 (1947). portion of the potential-capacity curves of Figs. (15) D. C. Grahame, J . Am. Chem. Soc., 79, 2093 (1957). (13) L. Rice and N. Hackerman (unpublished results) found that the zeta potentials of platinum, of gold and of silver ahowed very little dilution effect for nitrate and sulfate solutions.
(16) These values have not been corrected for junction potentiale. (17) B. E. Conway, “Electrochemical Data,” Elsevier Pub. CO., N. Y.,1952,p. 26.
I
Oct., 1958
ADSORPTIONO F ANIONSAT THE
It follows that the degree of covalent character of the halide ions adsorbed on the electrode decreases in the order: I-> Br-> C1- >F-. It is interesting that a plot of the polarizability of each halide against the potential of the capacity maximum for each hump suggests a linear relationship (Fig. 8). From the above considerations and from the relative size of the adsorbed ions, it also follows that the anions, during adsorption, must be dehydrated in the following order: I- > Br-> C1> F-, and that the process of dehydration (going from right to left) must be occurring within the potential range of each hump. Additional support for this assumption is found in the fact that only a slight increase of potential beyond the minimum on the left of the hump produces a steep rise in capacity (Fig. 2) due to the discharge of the halide ions (except for F- ions) under consideration. Evidently, these anions must be dehydrated before discharge. The curves for fluoride ions (Fig. 2) and for nitrate ions (Fig. 3) indicate that these ions are not dehydrated to any appreciable extent even on the anodic side of the humps. This conclusion is supported by the marked similarity of curves for the nitrate and fluoride systems and by the fact that the anodic process a t the extreme left in both cases is oxygen evolution. Here it may be noted that the polarizability of OH- ions is 1.89Is and that of water dipoles is 1.4417 compared to 0.81 for fluoride ions. Therefore, in the presence of fluoride and nitrate ions the inner region of the e.d.1. must be populated largely by OH- ions and oriented water dipoles. If this is true, then the humps with fluoride and nitrate solutions may be attributed to the desorption of OH- ions and reorientation of water dipoles. In this regard, sulfate ions (Fig. 3) resemble fluoride and nitrate ions, although the former appear to be slightly more adsorbed than the latter. Moreover, sulfate ions show a slight “dip” at about 0.4 v. Whether this is due to a decrease in the capacity of the diffuse part of the e.d.1. a t or near the Z.P.C.of the electrode,lgor is associated with the divalent character of the anionll is not known. The effects of dilution (Figs. 4,5 ) and temperature (Fig. 6) further support the foregoing conclusions. Thus, dilution of chloride and bromide solutions reduces the specific adsorption as expected. With nitrate (and fluoride) ions which show least specific adsorption, dilution has no effect on the capacity (Fig. 5 ) within the concentration range investigated. Behavior of sulfate ions is similar. Also a rise in temperature should decrease the specific adsorption as illustrated by the curves for chloride solutions (Fig. 6) a t the two temperatures. Adsorption of Cations.-Figure 7 indicates that potassium ions in the e.d.1. are probably less hydrated than sodium ions. Considering the relative sizes of the two ions, this is not unexpected. Such differences in the behavior of cations at polarized mercury electrodes have not been observed. If (18) J. A. A. Ketelaar, “Chemical Constitution,” Elsevier Pub. Co., N. Y., 1953, p. 90. (19) M. A. V. Devanathen, Trans. Faraday Soc., 6 0 , 373 (1954).
1201
POLARIZED I’LATINUM ELECTRODE
I
I
I
1
I
I
60
.* P
s230
V
20 10 I 1.2
I I I I I 0.8 0.4 0 -0.4 -0.8 -1.2 Potential (v. us. SCE). Fig. 2.-The effect of halide ions on the capacity of polarized platinum electrodes: (a) 1.0 N NaF; (b) 1.0 N KCl; (c) 1.0 N KBr; (d) 1.0 N KI. I ’
I
1
I
I
0.8 0.4 0 -0.4 -0.8 Potential (v. us. SCE). Fig. 3.-The effect of nitrate and sulfate ions on the capacity of polarized platinum electrodes: (a) 1.0 N KNOI; ( b ) 1.0 N KeS04. 1.2
70
I
I
I
I
I
1
0.8 0.4 0 -0.4 -0.8 -1.2 Potential (v. us. SCE). Fig. 4.-The effect .of chloride ion concentration on the capacity of olarized platinum electrodes: (a) 1.0 N KCl; (b) 0.1 N K8l; (c) 0.01 N KCl. 1.2
confirmed, this means that cations are also dehydrated before being cathodically discharged. Also, Fig. 2 and Fig. 7 indicate that the contribution of cations to the hump is relatively small. This suggests that Kf and Na+ ions are held at or near the platinum electrode mainly by the electrostatic forces. More work is required before final conclusions can be drawn. Similarity between Platinum and Mercury Electrodes.-Comparison of the various curves reported here with those for mercury e l e ~ t r o d e s ‘ ~ ~ ~ 4
P. V. POPAT AND N. HACKERMAN
1202
Vol. 62
reveals a basic, qualitative similarity between the two electrodes within the corresponding range of potentials where no electrochemical reaction occurs. 6 20 Robertson4 did not observe such similarity, prob@ ably because he used acid solutions and therefore . missed the point of comparison, ie., the hump. lo 1.2 0.8 0.4 0 -0.4 -0.8 On the basis of this similarity, it may be assumed that the foregoing interpretation of the humps with platinum electrodes may also apply qualitatively to those with mercury electrodes. If so, this would u 20 provide a perhaps more satisfactory explanation of the humps than that given by Grahame’s “ice10 layer” theory. 1.2 0.8 0.4 0 -0.4 -0.8 Capacity Minimum and the Zero Point of Charge Potential (v. us. SCE). (z.p.c.).-The Z.P.C.usually is taken as the potential Fig. 5.-The effect of nitrate and sulfate ion concentration a t which a well-defined minimum occurs in the on the capacity of polarized platinum electrodes: (a) 1.0 N potential-capacity curves with dilute ( Br- > C1- > F-. It must be pointed out here that the interpretation given by Sarmousakis and Prager’ for the steep rise of capacity in the cathodic region of their curves for the bromide and iodide solutions (as due to the discharge of hydrogen ions, etc.) seems incorrect on the following grounds: (i) If this rise in capacity represented hydrogen evolution, then the capacity should continuously rise with more cathodic potentials. This is contrary to experimental observation (Fig. 2). Evidently, they did not cover sufficient potential range and therefore missed the hump as well as the potential of hydrogen evolution. (ii) If the rise in capacity under question represented hydrogen evolution, then the electrode potential should become relatively insensitive to further changes (in more cathodic direction) of the applied potential. This was not observed in the present work. (iii) Sarmousakis and Prager followed Dolin and Ershler2 in their interpretation. However, the latter workers used acid solutions whereas the former used neutral solutions of KBr and KI. Evidently the potential of hydrogen evolution must be different in the two cases. It is interesting to note that Fig. 2 suggests a convenient method of estimating the magnitudes of hydrogen overvoltages in neutral solutions of various anions. Acknowledgment.-The authors are pleased to express their appreciation to the Office of Naval Research for the financial support of this work.
