910
CECILV. KINGAND BORISLEVY
Vol. 59
ADSORPTION AND SILVER-SILVER ION EXCHANGE’ BY CECILV. KINGAND BORISLEVY Department of Chemistry, New York University, New York S,N . Y . Received February 66,I966
The adsorption of silver sulfate on silver has been measured with Sa6as tracer and been shown to be dependent on both Sod- and Ag+ concentrations. It was found that sodium sulfate is also adsorbed on silver. Part of each adsorption is irreversible in nature (chemisorption). Adsorption is rapid on initial immersion of the metal in solution, and the amount remaining on the surface decreases with time. Exchange was studied with Agllo as tracer. Exchange is rapid on initial immersion and is greatly retarded as adsorption interferes. If reversibly adsorbed salt is washed off frequently much higher levels of exchange are obtained in a given total time of immersion. Evidence is presented that exchange between metal and adsorbed salt is faster than exchange between adsorbed salt and solution.
The adsorption of salts (or their ions) on metals The results of a few experiments which illustrate is of obvious interest in several connections, no- this method have been published.6 Adhering liquid tably the kinetics of corrosion and dissolution, the was removed by blotting or wiping the coupons kinetics and mechanism of electrode processes, the with absorbent tissue. Since adsorption is exinterpretation of overvoltage, even the measure- pected to result in monolayer coverage or less, ment of reversible potentials. Classical methods about 20 X 10-lo g. atoms/cm.,2B6a procedure is of measuring adsorption require large surface areas; needed which leaves not more than 10% of this methods adaptable t o a few are desirable. amount of unadsorbed salt, or the equivalent of a The interpretation of electrical double layer ca- thin but reproducible liquid film. Others have pacity, or its equivalent, for this purpose, has been dealt with this problem in various ways. Erbacher’ explored, and radioactive tracers are being widely allowed metal coupons to drain, cut off and disemployed. carded the lower half. Cook and Hackerman* A knowledge of the kinetics and extent of ex- estimated the liquid remaining on steel powder by change of a metal with its own ions is equally weighing while moist with solvent. Smith and important, in studying the nature of the metal Alleng estimated the liquid necessary to wet their and especially its surface, and in the interpretation samples directly. To test the blotting method of removing liquid, of electrode kinetics and potentials. Exchange current a t an electrode can be deduced from over- sodium sulfate with S35as tracer was used, since it potential studies, but direct measurement is was thought that this salt would be little or not a t welcome. Too small exchange current makes the all adsorbed. The same tracer was then used to term “reversible potential’’ meaningless; mercury measure both reversible and irreversible adsorption makes a poor hydrogen electrode, and a passive of silver sulfate, since this avoids the complication metal does not respond, in potential, to its own of simultaneous exchange. ions. Experimental For these reasons it was attempted, in this Radioactive sulfur and silver were obtained from the Oak research, to obtain quantitative measurements of Ridge National Laboratory. The Sa was in dilute hydrothe adsorption of silver salts on silver by tracer chloric acid, which was expelled by adding a known amount techniques, and t o study further the exchange rate of sulfuric acid and evaporating repeatedly. The solution of silver with its ions. Classical adsorption meas- was then adjusted to 0.001 M HzSO,, and portions were added to known solutions of sodium and silver sulfates. urements cannot easily distinguish between physi- Silver nitrate solutions were made by weighing both active cal and chemisorption; exchange experiments, as and inactive salts. Silver sulfate solutions containing Ag”0 customarily carried out, cannot distinguish between were made by mixing active silver nitrate with slightly acid actual exchange and chemisorption. The simul- silver sulfate, warming to dryness several times to expel acid, dissolving and diluting to volume. taneous study of adsorption and exchange has nitric Commercial rolled silver sheet (99.8% Ag), 0.2 mm. thick proved helpful. (Baker and Co., Inc.) was cut into 1 X 1” squares. The At first it was hoped to make both measurements coupons were suspended in solution from glass hooks, by simultaneously with Agl‘O as tracer. If a silver means of ‘/le’’ drilled holes, in covered vessels. The solutions were not deaerated and were not stirred after it was coupon is immersed in an active silver nitrate found that stirring had little or no effect. Most of the exsolution, the amount of exchange quickly reaches a periments were a t room temperature (27-32’), with one “ternporary plateau’’ and thereafter increases very series at approximately 0” (ice- and water-bath). I t was shown that mild abrasion of the silver surface res10wly.~ It has been suspected that a portion of sulted in more retention of liquid on blotting and about, twice the activity pick-up is due to irreversible adsorp- as much adsorption as on the smooth surface. Most cout i ~ n . We ~ hoped to include an estimate of reversi- pons were subjected to only mild cleansing as described beble adsorption as well, by freeing the specimens low. Radioactivity was measured with conventional shielding from liquid after immersion, counting the activity, and counting equipment. With Sa6,a weak @-emitter, the washing thoroughly and counting again. coupons were mounted as close as practical to a thin window (1) Based on a Ph.D. thesis submitted by Boris Levy t o the Graduate Faculty of New York University. Work done under Contract DA-30-069-ORD-1113 between the Office of Ordnance Research and New York University. (2) R . S. Hansen and B. H. Clampitt, THISJOURNAL, 58, 908 (1954). (3) H. Gerischer and W. Vielstich, 2. Elektrochem., 66, 380 (1952). (4) A. Baerg and C. A. Winkler, Can. J . Chem., 81, 319, 521 (1953).
GM tube. The counting error was usually less than 5%, but in cases of very weak adsorption as high as 12%. With
(5) C. V. King, Ann. N . Y . Acad. Sci., 68, 910 (1954). (6) C. V. King and R. K. Schochet, THISJOURNAL, 67, 895 (1953). (7) 0. Erbaoher, 2. physik. Chem., 8163, 196,215, 231 (1933), and elsewhere. (8) E. L. Cook and N. Hackerman, THIBJOURNAL, 66, 549 (1951). (9) H. A. Smith and K. A. Allen, ibid., 68, 449 (1954).
a
Sept., 1955
ADSORPTION AND SILVER-SILVER ION EXCHANGE
Ag"0 the counting error was much smaller. The over-all re roducibility is shown below. gadioactive counts were converted to amount of salt on the coupons by comparison with evaporated samples of each solution (0.5 ml. of a suitably diluted portion, spread over a silver coupon). With Agl'O, counts from both sides were added, or solution was evaporated on both sides and the counts averaged. Decay corrections were made by calculation or by recounting the calibration coupons. The coupons used with sodium and silver sulfates were first washed with ether, continuously distilled for about one hour in a Soxhlet extractor. Several subsequent surface treatments were tried, on groups of six coupons. Since the differences in amount and reproducibility of adsorption were within the rather large experimental error, the following treatment was chosen: one minute in 0.1 itf KCN, wash, one minute in 4.5% "01, wash; 24 hours in inactive 0.1 N AgNOs (to reduce local cell action), wash one hour. Six coupons so treated were used in preliminary measurements, in which it became evident that both sodium and silver sulfates show irreversible adsorption, not removable by several hours of washing. Since a one-minute dip in 0.1 M KCN removes practically all of the activity, subsequent experiments included this treatment if much activity remained on the coupon. I t was noticed incidentally that even a slight visible sulfide tarnish on the coupons did not measurably affect the adsorption.
911
sulfate ion concentrations are given, since it was found that the amount of adsorption depends on both. The activity found after blotting was corrected for adhering liquid as described, and chemisorption refers to activity remaining after washing one hour. If the chemisorption values are plotted as reciprocals, aside from a few scattered points a reasonably linear relation is found; there is no appreciable temperature effect. The physical adsorption is more erratic, partly at least because of the disparity in silver and sulfate concentrations. The values M Ag+ in Table I1 include the under 1.91 X effect of adding excess sulfuric acid, which increases adsorption greatly. This is not merely additive adsorption of silver and sulfate ions. The percentage average deviation column in Table I1 shows the degree of concordance in each group of four coupons; these deviations are typical. Duplicate runs sometimes showed greater differences than the maximum deviation in one group. The amount of physical adsorption found changes remarkably with time of immersion of the specimen. Results Adsorption of Sodium Sulfate and Adhering A number of coupons were immersed in a solution M Ag+, 4.9 X N Liquid Correction.-Four of the coupons were containing 1.91 X used in the experiments shown in Table I. The Sod* for periods from one second to 20 hours. second and fourth columns give the average Up to 3 minutes the physical adsorption averaged amount of sulfate remaining after washing the 0.85 X 10-lo equiv./cm.2; after 90 minutes to 20 coupons in distilled water for one hour. The third hours this value had fallen to about 0.30 X 10-lo and fifth columns give the activity which was equiv./cm.2. Chemisorption varied between 0.2 removed by washing, after blotting dry, in terms and 0.11 X 10-lo equiv./cm.2 during this time and of thickness of an equivalent layer of solution. it is not sure that any change is real. Chemisorbed Other sodium sulfate concentrations up to nearly sodium or silver sulfates exchanged onIy very slowly 1 M gave similar values, indicating that most of the with inactive sulfates. Exchange and Adsorption with Silver Nitrate washable salt actually comes from adhering liquid rather than being physically adsorbed. Con- (Agllo).-The coupons for exchange experiments sequently the average thickness values of Table I were washed with acetone, acidified thiourea and were used as a correction factor in subsequent water, immersed 45 hours in inactive 0.1 N silver nitrate to "equilibrate," then washed for one hour. measurements. Figure 1 shows the results of exchange experiIf the chemisorption of sodium sulfate is plotted as reciprocal of amount adsorbed vs. reciprocal of ments in 0.5 N silver nitrate. Seven coupons were concentration, it is evident that there is no notice- immersed for various periods of time, the first, able temperature effect, and linearity of the plot immersion periods being 10, 30 see., 1, 5, 10, 30, 60 may indicate Langmuir-type isotherms. Unfor- min. After each immersion the coupon was blotted tunately the curves cannot be extrapolated ac- dry, counted, washed one hour, recounted and immersed again. curately to l / c = 0. The first few immersions of coupons 1 to 5 are TABLE I not shown in Fig. 1. The amount of exchange is CHEMISORPTION OF SODIUMSULFATEO N SILVER,A N D far more dependent on the number of immersions AMOUNTOF THE SALTREMAINING AFTER BLOTTING than on the time; "exchange plateaus;' are found Four coupons immersed 1 hr. in each solution. only because of the way the immersion periods are Room temp. Ice-bath temp. chosen. Most of the exchange of each immersion ChemiLayer ChemiLayer takes place in the first few seconds, although slow sorp., thicksorp., thickmoles/ ness, moles/ ness, exchange continues indefinitely. The first dip CNn2804, cm.l X cm. X om.%X om. X M 10'0 106 10'0 106 results in exchange corresponding to equilibrium 1.01 X 0.418 1.77 0.450 1.13 with 4.5 to 5 atomic layers, based on apparent 2.06 X 0.98 .091 1.50 .I05 area and considering 19.9 X 10-lo g. atoms/cm.Z 1.21 x 10-3 .075 2.08 ... ... as one layer.5 No doubt there is an equal amount 1.11 9.81 x 10-4 ,050 1.74 .053 of exchange on subsequent immersion, but this 1.65 7.12 x 10-4 .055 2.75 055 amounts to only 0.5 to 1 additional layer of 1.71 .024 1.96 ,025 4.43 x 10-4 activity pick-up. av. 1.97 1.32 Chemisorption can of course not be detected in av. dev., % 15 22 these experiments, and is very small compared t o Adsorption of Silver Sulfate.-The same four the amount of exchange. The activity which could silver coupons were now used in silver sulfate solu- be washed off, after blotting and counting, detions, as summarized in Table 11. Both silver and creased with the number of immersions, later
CECILV. KINGAND BORISLEVY
912
Vol. 59
TABLE I1 THEADSORPTION OF SILVER SULFATE(Sa) One hour immersion. CA,+. N 2.16 X lo-*
C804-, N 2.17 X lo-*
3.39 x 10-3 2.54 X 10-8 2.00 x 10-8 1.26 X
3.49 x 2.64 x 2.01 x 1.35 x
6.28 X lo-'
6.77
eqJcrn.2 x 1010
x 10-4
8.03 x 3.64 x 2.91 x 4.91 x
10-4
1.91 x 10-4 8.04 X 2.10 x 10-6 2.13 X 10-6
1.00 x 2.00 x 1.00 x 8.2 X
10-8 10-8 10-4 10-
Room temp. Total ads. av. dev.,
%
10-4 10-4 10-4
increased again. The correction for adhering liquid, as used previously, would be 98 X 10-lo equiv./cm.2, a value larger than some of the numbers actually found. This indicates that the adsorbed salt contains less active silver than the dissolved salt, a t the time of washing. The decrease in washable activity and later increase with time was found in all experiments and has been noted in previous experiments.6
g
25
-
20
-
-4
2.8 3.9 2.09 1.02 0.64 1.40 1.14 0.37 .18 .24 .10
.I8 .05 .09 .20 .38 .23
I
I
I
200
-
0
0
4.2
4.43 2.62 1.63 3.33
1.32 0.59 0.86 0.51
1.71
0.35
1.41 1.43 0.44 1.23
0.23 0.31 0.10 0.15
1.15
0.25
.
.Ol
A single coupon was given repeated 30 second dips, with only 5 minutes washing; after each five dips it was washed one hour and counted. These results are also plotted in Fig. 2. Apparently the 5 minute wash is not as effective in restoring exchange reactivity as longer washing. Mere removal from the active solution and blotting was quite ineffective in promoting exchange.
I
015-
6.6
.Ol
I I
Ice-bath temp. Chemisorp., Total ads., eq./cm.a eq./cm.¶ x lolo x 1010
Chemisorp., eq./cm.' x 10'0
18 21 5 3 8 1 5 3 13 25 4 18 15 7 6 6 11 37 15
10.5 11.4 3.98 2.03 1.56 1.77 1.92 2.03 1.51 0.76 .86 .40 .28 .61 .86 1.28 0.06 .04 .04
10-3 10-3 10-8 10-3
x 10-4 x 10-4 x 10-4 x 10-4
6.03 2.64 1.91 1.91
Total adsorption corrected for adhering liquid.
I
I
I
I
'
-
1
N-
k o -
b
50 5 -
-
I
I
I
I
I
4 6 8 10 No, of immersions. Fig. 2.-Exchange from 0.02 M silver sulfate: 0 , avera e of 4 coupons, 30 second dips, one hour wash; 0 , sing?e coupon, each point represents five 30 second dips with 5 minute wash each then One hour 2
1
I
I
1
I
60 120 180 240 300 Total immersion time, min. Fig. 1.-Exchange from 0.5 N silver nitrate, the effect of different immersion periods. Low number coupons had most immersions, number 7 only three.
Exchange with 0.02 M Silver Sulfate.-Since the extent of exchange depends more on the number of immersions in active solution (with washing between dips) than on the total time of immersion, coupons were given several 30 second dips, each followed by an hour's washing. Four coupons were used and the results averaged; the amount of exchange (plus chemisorption) is shown in Fig. 2.
In these experiments stirring the solution had no effect on the exchange rate. Discussion The outstanding features of this research are (1) initial adsorption of silver salts is rapid and much greater in amount than equilibrium adsorption, and equilibrium is attained only slowly; (2) the rate of silver-silver ion exchange (and consequently the exchange current density a t a silver
. ,
Sept., 1955
ADSORPTION AND SILVER-SILVER ION EXCHANGE
electrode) is highly dependent on the stage of the adsorption process. Adsorption.-Rapid initial adsorption followed by slow decay to a final state has been found in other cases, namely, the adsorption of organic molecules on steello and of sodium sulfate on iron.’l However, chemisorption was referred to in these cases, and in the present system chemisorption is established quickly. I n many measurements of adsorption on metals it has been customary to allow one or two hours for equilibrium to be reached78l2 and it has been shown13that current flows for about two hours between a large silver electrode which has been equilibrated in silver nitrate solution and another similar electrode freshly immersed in the same solution. The anion probably has an effect; it was found that adsorption on silver powder from silver perchlorate solutions changes slowly up to 72 hourse6 Hackerman’s explanation of the decrease with time from a maximumlOJ1is that there is a large initial pick-up of loosely held ions (or molecules), followed by a slow rearrangement to the equilibrium condition, with proper orientation and formation of chemical bonds a t sites capable of forming them. A similar explanation with regard to washable adsorption is more difficult, but this is a matter qf degree, since there is no doubt every stage from very loose to very firm binding. Silver ion is probably primarily adsorbed to a much greater extent than sulfate, forming a continuation of the metal lattice. The electric charges of the adsorbed ions would not remain localized, and anions would be attracted to the entire surface. If the surface is heterogeneous with respect to adsorption sites, a smaller number of more firmly held anions could eventually take the place of a larger number more loosely held. The increased adsorption with excess sulfuric acid (Table 11) resembles a solubility product effect; firm adsorption of more sulfate ions leads to more sites for the adsorption of more silver ions, and vice versa. Autoradiographs of some of the coupons with adsorbed silver sulfate showed that more activity accumulates at obvious imperfections than elsewhere-along the cut edges, around the holes and along minor scratch marks. This may be a secondary effect of continuing local cell action in spite of previous “equilibration”; or it may indicate that such places are especially suitable for specific sulfate adsorption. Silver has often been reported to behave as a “null electrode” in solutions containing about 10-6 N silver ion; Proskurnin and Frumkin13 found no adsorption a t this concentration, and either silver dissolution or excess anion adsorption a t lower concentrations. The last two experiments of Table I1 show a small activity pick-up in such solutions, but of course do not show which ion is primarily adsorbed. (10) N. Hackerman and E. E. Glenn, THISJOURNAL, 64, 497 (1950). (11) N. Hackerman and 9. J. Stephens, ibid., 68, 904 (1954). (12) H.von Euler and A. Hedelius, Arkdu Kerni, Mineraloyi, Qeologi, 7, No.31 (1920). (13) M. Proskurnin and A. Frumkin, Z. p h y s i k . Chew., A166, 29 (1931).
913
It was shown in previous work6 that adsorption on silver is dependent on the anion (Ag2SO4 > AgC104 > AgN03). This no doubt reflects both the polarizability and the specific bond-forming tendencies of the ions. Total adsorption and its distribution must be in accordance with both the electrode potential and the double layer capacity, and the amounts actually found indicate a complex double layer. The extent of silver sulfate adsorption found is in reasonable agreement with that found on finely divided silver16for example about 0.5 monolayer, based on apparent area, in 0.02 N silver sulfate. Exchange.--Exchange is no doubt extremely rapid a t the surface of the bare metal. After a few seconds of immersion, the limiting factor in exchange kinetics is transmission of the ion through the adsorbed layer rather than diffusion in solution or diffusion within the metal. After adsorption equilibrium is reached, both desorption and new adsorption must be slow; since complete coverage is not required t o make exchange very slow, adsorption may be a necessary intermediate step in exchange. The decrease in washable activity and subsequent increase with time mentioned above indicates that exchange depletes the adsorbed layer of activity, which is not replenished from solution for several hours. Gerischer and Vielstich3 found the time required to reach a given fraction of their “exchange plateau” to vary inversely with the solution concentration. While they do not state exactly how the experiments were performed, the behavior indicates that a definite time pattern was followed with each solution, with increasing time periods for successive immersions. If several coupons are immersed and one removed from time to time for counting, all the results are on a single (slowly rising) plateau. This was evident in previous work6; and in the present experiments with 0.5 N silver nitrate, the first immersion, whether for 30 seconds or one hour, always resulted in an exchange of 4.5 to 5 apparent atomic layers. If the experiments of Fig. 2 were continued long enough, probably a much higher plateau of slowly increasing exchange would be reached, with the rate truly limited by diffusion within the metal. Exchange equivalent to several thousand atomic layers in an hour or less has been found between copper and its ions.14 The difference may be only that adsorbed copper salt does not retard exchange. Such rapid exchange is usually ascribed to local cell reaction, but there is probably also the equivalent of a high diffusion coefficient t o an unknown depth within the metal. The effect of removing adsorbed ions was also noted by Baerg and Winkler,4 in a somewhat different way. These authors immersed a coupon for one minute, rinsed, and transferred to dilute sulfuric acid. It was then made cathodic with a modified Bowden-Rideal technique to measure the number of coulombs required to reach the hydrogen overpotential. On repeating, an additional 0.5-1 monolayer of radioactivity was acquired per (14) M.Halasinaky, M.Cottin and B. Varjabedian, J . chiw. p h y s . , M. Quintin, P. Sue and M. Biaouard, C o m p f . reud., 226, 1723 (1048). 45, 212 (1948);
9 14
CECILV. KINGAND BORISLEVY
immersion, and the number of coulombs passed in the charging process was in fair agreement with the activity pick-up. The authors interpret this to mean that only chemisorption takes place after the first immersion, and that the adsorbed ions are deposited as metal on electrolysis, leaving the surface again available for chemisorption. Our experiments show, first, that far too little chemisorption occurs to account for these results, and, second, that washing is sufficient to restore the ability to exchange. To avoid any accidental electrolysis some of the coupons were handled with care t o avoid any contact with other metals, etc. Further, intentional electrolysis with the silver as cathode had no observable effect. The activity pick-up found by Baerg and Winkler is similar in magnitude to exchange found in the present work after thorough washing, although the solution concentrations were quite different. I n previous work on the rate of dissolution of silver in ferric sulfate solutions,15it was found that (15) H. Salzberg and C. V. King, J . Eleclrochem. Soc., 97, 290 (1950).
VOl. 59
slow desorption, especially of silver ions, must be a controlling factor in the kinetics. The present findings substantiate this possibility. Further, it was found that the potential of a silver electrode in ferric-ferrous perchlorate is polarized to the potential of a platinum electrode in the same solution, a t a smaller silver ion concentration than would be necessary in the absence of a corrosion reaction. M This indicates that slow desorption of silver ion formed in the corrosion process causes the potential t o correspond to a higher concentration than is present in the solution.
Discussion Gerischer and Tischerl’ have published additional experiments on silver-silver ion exchange. Single-crystal silver gives results similar to polycrystalline silver. The difference between samples washed and rcimmersed several times, compared to samples immersed for longer times, has been noted. The previous view concerning the importance of internal diffusion has been modified. (16) C.
V. King and F. Lang, ibid.,
99, 295 (1952).