0.2 Z
0
-
0.1
I-
O Z
3
0.0-
LL I-
=
W
-0.1
[r
E
3
0 -0.2
- 0.3 I
I
160
I
I
80
0
I
-80
I
I
-160
P O T E N T I A L , mV Figure 5. Comparison of theoretical third-order stationary electrode polarograms with dissolution of mercury into a 6mM cyanide solution buffered to pH = 7.0 with 0.2M Na2HP04NaH2P04 buffer Solid line in theory with a potential axis ( E - Eli~)n. Points are experiment with the first point -0.315 V US. SCE,scan rate is 0.120 V/sec for which the linear diffusion assumptions are justified. Because relatively rapid loss of cyanide was observed in the near neutral solutions where the principal cyanide form was HCN, constant concentration was tested by taking the stationary
electrode polarograms both before and after the constantpotential experiments. If peak currents agreed within 2 %, constant concentration was assumed. The linear diffusion assumption appeared valid in the scan rate range used, 0.1 to 1.0 V/sec, where i,/d;was found constant. The procedure was carried out on two solutions, one that was pH = 7.5 and a cyanide concentration of approximately 0.4 m M where Reaction 18 was considered operative and the other that was pH = 7.0 and a cyanide concentration of approximately 6 m M where Reaction 19 was obeyed. Both solutions gave a constant slope of i 1;s. l l d t f o r times greater than about 5 seconds (Figure 3) indicating diffusion control at longer times. Deviations at shorter times were due to a chemical reaction, the dissociation of HCN (14). The experimental equivalent of the forward scan peak current function for the first solution was 0.35 f 0.01 and was 0.30 i 0.01 for the second solution, agreeing with theory within experimental error. A comparison between experimental and theoretical stationary electrode polarograms for these two solutions is presented in Figures 4 and 5 . Agreement is good indicating that at least for the scan rates employed, mercury dissolution into cyanide solutions appears to involve reversible charge transfers with reaction orders greater than first order. Similar comparisons applied to other systems should assist in characterizing electrode mechanisms assumed to possess nonunity reaction orders. RECEIVED for review June 27, 1968. Accepted October 17, 1968. Work supported by funds from the Texas Christian University Research Foundation under Grant Nos. C6675 and C6785.
Electrochemical Oxidation of Nitrite and Oxide on Platinum Surfaces in Fused Potassium Nitrate-Sodium Nitrate Eutectic Melts at 250 "C P. G.McCormickl and H. S. Swofford, Jr. Department of Chemistry, University of Minnesota, Minneapolis, Minn. Experimental work is described concerning the electrochemical behavior of nitrite and oxide ion in sodium nitrate-potassium nitrate eutectic melts at 250 O C . Data are presented to support the postulate that a reversible couple is formed between nitrite and NO2 gas in the melt, evidence of which can be observed by current-voltage and chronopotentiometric techniques. In the case of oxide, it is suggested that oxidation of the ion proceeds via a one-electron process, and that the product, peroxide, is decomposed in a post chemical step. Also discussed is the behavior of oxalate in the melt, which is shown to remain essentially undissociated until catalytically decomposed to yield oxide.
IN A PREVIOUS PAPER ( I ) , experimental work was described concerning the electrochemical behavior of nitrite and oxide ion in sodium nitrate-potassium nitrate eutectic melts at 250 "C. Also described in this paper ( I ) was preliminary work re1 Present address, Department of Chemistry, Marquette University, Milwaukee, Wis.
lated to the use of oxalate as a source of oxide in the same eutectic melt. Since this publication, work has progressed, and further evidence now exists which greatly illuminates the fundamental chemical and electrochemical processes occurring. This paper describes the conclusions resulting from this work. EXPERIMENTAL
Equipment. All equipment used has been described previously ( I ) , with the following additions : Current-voltage curves were recorded using either a Sargent Model XV or Model XXI polarograph. For chronopotentiometric work, a controlled current source, based on a design by Reilley and Scribner (2), was constructed. The equipment was calibrated with standard resistors and a Leeds and Northrup student potentiometer. This potentiometer was also used for all direct potential measurements. Potential-time measurements were made using either a
~~
(1) H. S. Swofford, Jr., and P. G. McCormick, ANAL.CHEM., 37,
970 (1965).
146
ANALYTICAL CHEMISTRY
(2) C . N. Reilley and W. G. Scribner, ANAL.CHEM.,27, 1210 (1955).
Sargent Model SR recorder or a Tektronix Model 532 cathoderay oscilloscope, employing a Type D plug-in D C preamplifier. Electrodes. The reference and indicator electrodes used for current-voltage investigations have been described previously ( I ) . For chronopotentiometry, a cylindrical platinum electrode was prepared by sealing a length of platinum wire (27 gauge, B & S) in soft glass tubing, and cutting the exposed end to a length which gave an area of approximately 0.1 cm2. The platinum wire was silver-soldered inside the glass tubing to copper wire which provided an electrical contact and lead. Reagents. All reagents were as previously described ( I ) . Techniques. Besides those already cited ( I ) , the following techniques were employed: At the times when qualitative indication of COz and CO in the exit gas was desired, their presence was detected in two ways. With the first method effluent gas was passed through a washing tower containing lime water which sufficed to detect COz. The sensitivity of this detection method is about 0.05 % (3). A basic ammoniacal silver solution containing excess silver was used to detect the presence of CO, and worked effectively; the gas reduced the silver complex to a dark flocculent precipitate, probably metallic silver. A second washing tower containing this solution was used to indicate the presence of CO in the exit gas stream. The sensitivity of this system is approximately 0.005 Z (3). At various stages in the work, it was also necessary to determine quantitatively the amount of oxide and/or oxalate present in the melt, and of carbon dioxide present in exit gases. These materials were determined titrimetrically. Oxide was determined by dissolving a weighed sample of the melt in water, and titrating the hydroxide produced by the reaction 0’-
+ HzO = 20H-
with 0.01N HC1 using m-cresol purple as indicator. This indicator was chosen because of its low salt error as well as its vivid color change (purple to yellow) compared with that of phenolphthalein (red to colorless). The acid was standardized with previously standardized NaOH in a solution made 20% in pure eutectic salt (to approximate actual experimental ionic strength) using the same indicator. The usual permanganate titration ( 4 ) for oxalate did not give accurate or reproducible results in these studies. This was probably because of the high ionic strength of the titrated solution. Based on previously reported work (5-7), a second method was developed which uses nitrate as a background electrolyte, thus circumventing possible interference. The titrant employed was 0.10N cerric ammonium nitrate in 2N HNOI (more dilute solutions did not give sharp end points). The indicator used was 0.025M Ferroin (ferrous o-phenanthroline) which gave a better end point than Nitroferroin in that the color change (red to blue) was more pronounced, and the final blue color did not fade so rapidly as that given with Nitroferroin. The method developed was as follows: Dissolve sample in about 30 ml 2N HN03. Add 1 ml of 10% sulfamic acid (to destroy nitrite) and 1 ml of 20% manganese sulfate; warm over a low flame for 30 to 45 sec. Add 2 drops of 0.025M Ferroin and titrate with constant stirring directly to a blue end point color (correction equals 0.03 ml of 0.10N titrant). (3) P. G. McCormick, Ph.D. thesis, University of Minnesota, 1967. (4) J. Rosin, “Reagent Chemicals and Standards,” 4th ed., Van Nostrand, Princeton, N. J., 1961, p 343. (5) G. F. Smith and C. A. Getz, IND. ENG. CHEM.,ANAL.ED., 10, 191 (1938). (6) V. R. Wheatley, Analyst, 69,207 (1944). ( 7 ) J. P. Watson, ibid., 76,177 (1951).
E (volts)vs A g / A g O +0.0
+O.C
+0.4
+0.2
1
I
Figure 1. Composite current-voltage curve for NOz
+ NOz-
If the sulfamic acid is not added, the sum of nitrite and oxalate can be determined. This titration can be carried out on a sample previously titrated with HC1 in the determination of oxide and still containing the cresol purple indicator. The acid-base indicator is red at this pH and does not interfere with the detection of the end point, though a new correction must be determined (approximately 0.045 ml of 0.10N titrant). The Ce(1V) titrant is standardized against primary standard sodium oxalate. The titrimetric procedure adopted by IUPAC (8) using barium saccharate solution and back-titrating with oxalic acid was used in the determination of carbon dioxide. It gave most satisfactory results. All other assay methods used were standard procedures, and can be found in the reference by Rosin ( 4 ) . RESULTS AND DISCUSSION
Nitrite. In our previous paper ( I ) , it was implied that the nitrite wave arose from a reversible one-electron oxidation of nitrite to NO*. That the process transferred one electron was demonstrated by coulometric experiments, but it was not adequately demonstrated that the process was reversible. Indeed, our inability to demonstrate electrochemical activity due to NOZgas would seem to indicate either that the process was not reversible, or that NO2 was nearly insoluble in the melt. Evidence already presented ( I ) seems to point to reversibility ; the fact that Topol, Osteryoung, and Christie (9) have reported obtaining solutions of NO2 in this eutectic in the range of 5mM would indicate the latter possibility as also being unlikely. It is probable that our problem was one of solution kinetics-Le., NO2 is soluble, but dissolves slowly. It is also possible that in passing a mixture of NO2 and nitrogen through the melt, the latter swept out the former. (8) “Methods for Determination of Toxic Substances in Air,” IUPAC, 1959, Section 9.1 (adopted 1951). (9) L. E. Topol, R. A. Osteryoung, and J. H. Christie, J. Plzys. Chem., 70, 2857 (1966). VOL. 41, NO. 1, JANUARY 1969
0
147
-I
Current Reversal
Table I. Comparison of Characteristic Potential Values for Forward and Reverse Nitrite Wavesa i, mA/cm2 Ella forward EO.z~ reverse Concn, M 0.0145 0.025 0.035 0.045 0.045
3.45 4.60 11.50 15.33 1.67
+0.46 v 0.49 0.46 0.49 0.49 Av 0.418
+0.49 v 0.48 0.48 0.48 0.41 0.480
Current-voltage Ei/z = 0.48 V ( I ) .
The gaseous NOZsaturation experiment was repeated, this time passing the gas mixture through a fritted glass disk to increase the effective surface area of the gas exposed to the melt by reducing the bubble size. The flow rate was also reduced to provide a gas mixture rich in NOs. A dense atmosphere of NO2 was produced over the melt surface and the flask was stoppered and left overnight. After being sealed in this fashion for 18 hr, a reference electrode (prepared externally) was introduced and current-voltage curves were run. Well formed composite current-voltage curves were produced (with Not- added), as typified by Figure 1. Wave analysis of the composite wave showed a slope of 0.103 V (theoretical value = 0.104 Vat 250 “C)thus confirming its reversibility. Topol et al. ( 9 ) propose a reaction between NO, and nitrite, as follows:
N02(g,
+ NOz-
NOS-
+ NO
TIME ( s e d
Figure 2. Potential-time curve for nitrite A = 0.131 cm2, i = 0.45 mA, C = 0.0145M
This reaction was confirmed in that both anodic and cathodic branches of the wave in Figure 1 decreased in magnitude upon sitting for even short periods of time (less than 1 hr). This reaction was further confirmed in a qualitative manner by adding nitrite to a melt and noting the reduction in limiting current values for the nitrite wave as a function of NOz passed through the melt. Finally the solubility of NOz was checked in the melt at 250 “C by passing pure NO2 from a cylinder through the fritted disk. After 2 hr, a limiting current of 158 pA was recorded for the cathodic wave, representing a concentration of NOz in the melt of about 2mM, if it is assumed that its diffusion coefficient is the same as that for nitrite. The value reported by Topol et al. (9) for the solubility was 5 i 3 X 10-3M at 300 O C. This value, then, is in good agreement with their work. It was decided to examine the problem further using chronopotentiometry with current reversal. If the product NOZ, formed from the forward anodic oxidation, does indeed form a reversible couple with nitrite, it should be observable in the cathodic branch of a potential-time curve following current reversal. For this study, the cylindrical platinum electrode already described was used. Others who have used an electrode of this geometry, notably Lingane ( I O ) in aqueous media, and Narayan and Inman (11) in this melt, noted that its application introduces no significant data handling problems. At low current densities residual potential-time curves of the melt showed the presence of residual nitrite in the forward anodic branch, and current reversal did indicate a pause in the cathodic branch, possibly due to the product, NOz. A complete chronopotentiometric study of the nitrite oxi-
This implies that, for a semi-infinite linear diffusion-controlled process in the absence of kinetic complications, the value of i r l /‘IC should remain constant and independent of the experimental parameters.
(10) J. J. Lingane, J. Elecfroanal. Chern., 1, 379 (1960). (11) R. Narayan and D. Inman, J. Polurog. Soc., 11,27 (1965).
(12) P. Delahay, “New Instrumental Methods in Electrochemistry,” Interscience, New York, N. Y.,1954, Chapter 8.
148
ANALYTICAL CHEMISTRY
dation problem was undertaken, varying both concentration and current density as much as practicable. Potential-time curves were displayed on an oscilloscope or recorder, depending on the times involved. Transition time values were measured graphically by the method of Delahay (12). A typical potential-time curve is presented in Figure 2. The cathodic branch ( T ~does ) indeed show a pause in the vicinity of the quarter-wave potential at which the oxidation appears. Poising the electrode at the anodic limit for a short time before current reversal always extended this cathodic branch. This would seem to indicate that the cathodic wave is due to NO2 because this gas is vigorously produced at the anodic limit of potential. To demonstrate this further, the melt was again saturated with NOZ gas and potential-time curves were recorded ;the wave was again enhanced somewhat. During the course of this study insufficient NOz was dissolved in the melt to shift the rest potential of the indicator electrode far enough in the anodic direction so that cathodic chronopotentiograms for the reduction of NO2 could be recorded. Topol et al. (9), however, have succeeded in obtaining potential-time curves for NO*, and report satisfactory results. The well known Sand equation (12) may be rearranged to produce the so-called “chronopotentiometric constant :” j7 I / 2
_ _ _C
D I/ 2
I/
2
Data from chronopotentiometric experiments were corrected according to the method suggested by Bard (13). Values obtained for the constant were plotted against current density and shown to be invariant for more than a fivefold change in current, thus providing assurance that the process is diff usion-controlled with no kinetic complications. The chronopotentiometric constant is proportional to the diffusion coefficient. If the value of n and the area of the electrode are known, the constant can be used to determine the value of D. With n having been established as one electron by the coulometric experiments and if the geometric area of the electrode (0.131 cm2)adequately represents the electrochemical area, a value for D is obtained: D = 2.75 X 10-5 cm2/sec. This value is reasonable, being of the same order of magnitude as values obtained by others for diffusion coefficients in fused salts (14-16). A further test of reversibility is the equality of the Ell4value observed for observed for the forward anodic wave and Eo.nL the cathodic branch following current reversal (12). The characteristic values for the forward and reverse processes were checked at various concentrations and current densities. These results are shown in Table I. The agreement between these two values, as well as their similarity to the value of El/* obtained in the current-voltage work, lends still further credence to the reversibility argument. This now gives rise to the problem of trying to characterize the process actually occurring at the electrode surface. Lyalikov and Novik (17), though they do not discuss the problem explicitly, seem to accept implicitly the process as the reversible one-electron oxidation of nitrite. Inman and Braunstein (18) have suggested that the product NO2 is adsorbed on the electrode surface and, at higher nitrite concentrations and more positive potentials, undergoes another one-electron transfer, forming the species NOz+ which quickly reacts with oxide (in the melt or on the Pt surface) or nitrate ion. Their communication is short and gives few details, but our results do not seem to support their findings. The second oxidation wave mentioned by them was completely absent in all the present studies, and the ratio of the forward-to-reverse transition times indicates a further problem with Inman and Braunstein's proposed scheme. Though a reverse transition was seen during the chronopotentiometric studies, indicating the presence of NOn, the expected ratio of forward-to-reverse transition times (near 3:l) (3, 12) was never realized. The ratio varied from 6:l at low current densities to near 4:l at the higher current densities, but came no closer to theory (Table 11). When current reversal took place during the forward transition, the ratios were even higher (Table 111). If the NOs had been adsorbed by the electrode, the ratio should have approached unity. Topol, Osteryoung, and Christie (9) have reported that N O L in the melt could be reduced to NOz-, but could not be oxidized. They also noted an inability to achieve the expected 3:l ratio of transition times during chronopotentiometric investigation of nitrite. The evidence collected during this research leads to an interpretation of the process as a reversible one-electron oxi(13) A. J. Bard, ANAL.CHEM., 35, 340 (1963). (14) Yu. K. Delimarskii and B. Markov, Zh. Fiz. Khim., 27, 1848 (1953). (15) N. G. Chovnyk and V. V. Vashenco, ibid., 35, 580 (1961). (16) C. E. Thalmayer, S. Bruckenstein, and C. M. Gruen, J . Inorg. Nucl. Chem., 26, 347 (1964). (17) Yu. S. Lyalikov and R. M. Novik, Uch. Zap. Kishinecsk. Gos. Uniu., 27, 61 (1957); C . A . , 54, 22101d (1960). (18) D. Inman and J. Braunstein, Chem. Commun. 1966,148.
Table 11. Comparison of Forward and Reverse Transition Timesa Concn, M i, mA T ! , sec T ~ sec , 0.0145 0.0250 0.0350 0.0350 0.0350 0.0450 0.0450 4
0.45 0.768 1.00 2.00 3.00 3.00 4.00
Electrode area
3.82 3.02 3.76 0.96 0.27 0.41 0.20
= 0.131
0.67 0.53 0.67 0.20 0.07 0.07 0.04
T//T?
5.7 5.7 5.6 4.8 3.9 5.9 5.0
cm2.
Table 111. Interrupted Chronopotentiometry of Nitrite i = 0.60mA A = 0.117 cm2 C = 65mM r = 2.36 sec
sec
rr, sec
tJ/ry
1.18 1.57 1.97 2.16 2.36
0.079 0.12 0.16 0.20 0.40
14.9 13.1 12.3 10.8 5.9
tf,
dation of nitrite to NOZ. The reversible composite currentvoltage curve for NO2 NOZ-, the theoretically correct slope of the wave analysis for that curve, and the reproducibility of characteristic potentials for the forward and reverse processes during chronopotedtiometric investigations all support the argument of reversibility, while the coulometric experiments firmly establish the process as transferring one electron. The inability to achieve the theoretical forward-to-reverse transition time ratio during chronopotentiometric experiments can be explained in terms of the difficulty encountered in dissolving NOnin the melt. Though the gas does dissolve in moderate quantities, the apparently slow kinetics of solution greatly limit the ability of gas formed at the electrode surface to dissolve in the surrounding melt, and thus encourage loss of NOz from the bulk by convective diffusion and volatilization. Oxide. As noted in our previous paper ( I ) , a wave appeared between -0.20 and +0.20 V L'S. Ag/Ag(I) following cathodic reduction of the melt. This wave was attributed to oxide ion oxidation. The value of the half wave potential has been established for this wave as equal to -0.078 V. The difficulty of adding oxide ion to the melt in a solid form has been mentioned briefly ( 1 ) . Since that time it has been found that sodium peroxide has some use as a source of oxide. Shams El Din (19) and Reddy (20) have both suggested that sodium peroxide decomposes in the melt to yield oxide via the following reaction:
+
Na202= N a 2 0
+ '/?On
Sodium peroxide and barium oxide were added to the melt and, in both cases, a wave appeared at the same position on the potential axis as that described above; the height of the wave was linearly related to sodium peroxide and barium oxide addition. (19) A. M. Shams El Din and A. A. A. Gerges, Proc. Australian Conf. on Electrochemistry, 1963, Pergamon Press, London, England, 1965, p 562. (20) T. B. Reddy, Electrochem. Technol., 1, 325 (1963). VOL. 41, NO. 1, JANUARY 1969
149
If one assumes that oxide is reversibly oxidized to oxygen via the electrochemical reaction 02-
=
1
+
/ZOZ 2e
a plot of log (il - i)/i1’2 cs. E should be linear with a slope of 0.052 V indicating a two-electron transfer. The slope of such a plot is much too high to be indicative of a two-electron change, having a value slightly higher than 2.3RTIF. Such a slope could be interpreted as more indicative of a somewhat irreversible one-electron transfer. Though unexpected, this behavior has been previously noted in the literature. Littlewood and Argent (21) studied the behavior of an oxide-oxygen electrode fabricated by immersing a platinum electrode in their chloride melt and flushing it with oxygen gas. They found that this electrode responded to changes in oxide concentration added as sodium oxide. However, their slope of the potential us. concentration plot indicated n as being equal to one rather than the expected two electrons. They attempted to explain this observed behavior by postulating changes in a water equilibrium which, they say, would buffer further additions of oxide. Shams El Din (19) has noted this same behavior in fused KN03. Additions of sodium peroxide, potassium oxide, and hydroxide to “unbuffered” melts produced potential-concentration curves for an oxygen electrode which had slopes indicative of exactly one electron. His explanation was similar to that of Littlewood and Argent. Laitinen and Bhatia (22) have noted that an oxygen electrode employing carbon as the electrode material does not behave reversibly in fused LiClKC1. Despite its seemingly well documented irreversible behavior, the oxygen electrode continues to find rather widespread use in fused salt electrochemistry, probably because it does respond reproducibly. Nonetheless, the appearance of a one-electron electrode process, rather than the expected two, has never been satisfactorily explained. The decomposition of peroxide discussed previously gave rise to the postulate that a product other than oxygen could be formed at the electrode surface as a result of the oxidation, peroxide being the most obvious choice via the following electrode reaction :
+ 2e
2 0 2 - = 02-
This process would transfer one electron per mole of oxide For a reversible process, the Nernst expression in terms of currents would be 0.104 2
E,,, = EOo2-/0-~2 - -log (i, -
iP/i
and a plot of the log term L‘S. EaBp should produce a straight line with a slope of 0.052. Typical data plotted using this expression showed a slope which was still higher than theory (0.079 V) but could be interpreted as being indicative of an irreversible one-electron oxidation. The solution to this problem was approached using two separate methods in an attempt to establish firmly whether one or two electrons is transferred during the oxidation of oxide. First, chronopotentiometric potential-time curves were obtained over the same potential range as in the current-voltage work ( I ) . The general appearance of these curves is typified (21) R. Littlewood and E. J. Argent, Electrochim. Acta, 4, 114 (1961). ( 2 2 ) H. A. Laitinen and B. B. Bhatia, . I Electrochem. . Soc., 107, 705 (1960).
150
ANALYTICAL CHEMISTRY
Cur r e n t Reversal
1.0
2.0
3.0
4.0
TIME ( s e d
Figure 3. Potential-timecurve for oxide A = 0.131 cm2, i
=
0.35 mA, C = llmM
by Figure 3. These curves were analyzed in a manner analogous to that used in the current-voltage work. Typical results showed a slope of 0.055 V. As shown in the discussion concerning the chronopotentiometry of nitrite, if the value of i7”*/C can be shown to remain constant while i and C are varied, it may be assumed that the conditions of the Sand equation are being obeyed and useful calculations made based on this relationship, As seen from the rearranged equation, the chronopotentiometric conTherefore, stant is directly proportional to both n and D1’2. if the constant is divided by a reasonable value of 0 1 1 2 , the value of n, the number of electrons being transferred, can be estimated . The value of ~ T “ ~was / Cconstant over a sixfold change in i. Thus, the Sand equation was being obeyed, and a value was calculated for n using the value for D obtained for nitrite in the above discussion (2.75 X 10-6 cm2/sec). The value obtained for n by this procedure was 0.91, indicating strongly that one electron is being transferred in the electrode reaction. The second method used to establish n was coulometric analysis. Oxide can be removed electrochemically in the same manner as nitrite, the potential of the working electrode being controlled at a value on the limiting current of the oxide wave prior to the onset of the nitrite oxidation. A silver coulometer was incorporated into the electrolysis circuit during removal. The concentration of oxide in the melt was determined by removing samples from the bulk at various stages in the electrolysis, weighing them to determine volume, dissolving them in water, and titrating the liberated hydroxide. A number of experiments were run in this manner, the results of which are summarized in Table IV. Because of the long times involved and the technique of removing samples of melt, the peroxide produced in the electrode reaction decomposed chemically (vide infra) to oxide and oxygen. Thus, the net reaction would be
+ 2e
02-= O2
indicating an apparent transfer of two electrons per oxide ion. As shown in the table, the results are consistent with the above interpretation which presupposes a one-electron transfer in the electrochemical step. As a final check, the above experiment was repeated, removing oxide electrolytically, simultaneously following the concentration of remaining oxide electrochemically. With the relationship between current at the indicator electrode, and oxide concentration, and the volume of the melt known, it was possible to calculate the amount of oxide removed by comparing the limiting current in the oxide waves run before and after electrolysis. This value compared to the quantity of silver transferred in the coulometer provided a value of n = 1, again confirming the previous results. These results, employing chemical as well as electrochemical evidence, point to the conclusion that the oxidation of oxide at a platinum electrode proceeds via an irreversible one-electron process. Irreversibility is suggested by the fact that, in chronopotentiometry, current reversal produces no pause in the cathodic cycle in the potential region where oxide is oxidized, indicating that the product produced on oxidation is not reducible in the same potential range. To demonstrate that this was not an artifact of the platinum electrode material, the work was repeated on a gold-plated platinum electrode. The waves observed were no different in appearance from those described above. The electrode, however, was not usable for extended lengths of time because the gold plated layer was physically stripped from the platinum surface. It is suggested that the nature of the melt, containing nitrate and oxide, both of which attack gold, coupled with the anodic current contributed to this phenomenon. Now, assured that n = 1, and further that the Sand’s relation holds, the chronopotentiometric constant can be used, as in the nitrite case, to calculate a value for the diffusioncoefficient for oxide; the value obtained was D = 2.24 X cm-2/sec. An electrode poised at a potential on the limiting current portion of the oxide wave in oxide-containing melts is observed to evolve gas bubbles, This gas evolution, however, does not commence immediately, but begins 20 to 30 sec following the initiation of the electrolysis. This observation suggests that the peroxide formed in the electrochemical step is slowly decomposing, probably to oxide and oxygen. To test this postulate, weighed portions of sodium peroxide were added to the melt. The peroxide dissolved slowly, the dissolution being accompanied by a slow evolution of gas bubbles, After dissolution was complete, the contents of the fritted compartment in which the experiment had been conducted were dissolved in water and analyzed for oxide. The oxide found indicated in all cases that the peroxide had produced stoichiometric amounts of oxide. Further evidence for the production of oxide from peroxide is supplied by the fact that current-voltage curves run on solutions of dissolved sodium peroxide have appearances identical to those observed following electrochemical generation of oxide. Apparently, because the chemical decomposition is slow, the appearance of a catalytic oxide wave is prevented. The body of the evidence all tends to support the original postulate of an irreversible one-electron oxidation of oxide to peroxide, followed by very slow chemical decomposition of the product to oxide and oxygen. This being the case, the seeming “reversibility” of the “oxygen electrode” can be explained: The electrode would respond to oxide in the melt
Table IV. Coulometric Removal of Oxide Trial I I1 111 IV Ag transferred, mg 35.01 28.56 15.76 78.34 Ag transferred (I), mmoles 0.324 0.265 0.146 0.725 Oxide present before, mmoles 0.623 0.467 0.326 0.572 Oxide present after, mmoles 0.467 0.326 0.252 0.230 Oxide removed (2), mmoles 0.156 0.141 0.074 0.342 Ratio 1:2 = n 2.08 1.88 1.97 2.12 Av of 5 values = 2.01
v
4.57 0.042 0.230 0.209 0.021 2.00
Table V. Constancy of Nitrite Current with Oxalate Additiona Oxalate concn, (mM) Residual
Total current, pA at 0.70 V
Oxide current, p A at 0.20 V
105 154 162 186 200
1 50 58 82 96
1.04 1.25 1.77 1.92
Electrode area
=
Nitrite alone, PA 104 104 104 104 104
0.108 cm2.
via the electrode process itself, while the response to oxygen partial pressure observed by Kust and Duke (23) is explained via its effect on the second or decomposition process. This work would seem to indicate, however, that an electrode operating via an oxide-oxygen couple does not exist in unbuffered nitrate melts. There is evidence (19) that such an electrode does operate in well buffered melts (melts containing substantial quantities of both oxidized and reduced forms of oxyanions such as dichromate-chromate) and melts of oxyanions of low dissociation, such as carbonate melts [cf work of Dubois ( 2 4 . Oxalate. It was reported previously (1) that, in searching for a suitable compound which would yield stoichiometric oxide in the melt upon decomposition, potassium oxalate was tried; it produced current-voltage curves identical to those for oxide added via either cathodic reduction of nitrate or peroxide addition. These waves, when analyzed in the same manner as those for oxide (vide supra), again gave similar results. To show that the wave was being produced entirely by addition of oxalate, the limiting current for the oxidation of nitrite was shown to remain unchanged throughout the addition of oxalate (Table V). It was expected that, if oxide were the species in solution, addition of silver ion would produce insoluble silver oxide. Instead, addition of silver nitrate to oxalate-containing melts produced strong gas evolution and a grey precipitate. The gas was shown to be COS by passing it through lime water. The grey precipitate was dissolved in nitric acid and analyzed for silver. It was pure silver metal. Thus, it appears that oxalate remains substantially undissociated until addition of silver ion which reacts as follows: 2Ag+
+
C204’- =
2C02(g)
+ 2Ag(.,
(23) R. N. Kust and F. R. Duke, J. Amer. Clrem. SOC.,85, 3338 (1963). (24) J. Dubois, Ann. Chim. (Paris), 10, 145 (1965). VOL. 41, NO. 1 , JANUARY 1969
151
Table VI. Catalytic Oxalate Decomposition by Silvera Hr Oxide concn, mM 0 1 2 3 4
0.0 0.0 0.0 0.0 0.0
Introduced silver wire 6 7 8 9 33 72 a
0.315 0.495 0.630 0.970 4.70 10.82
All determinations based on titrimetric data.
Table VII. Recovery of Carbon Dioxide from Silver Carbonate Added to Melt Ag2COBnadded, mg COZTheor, mg COz found, mg 95.6
15.24
13.94
174.2
27.77
27.24
Silver carbonate prepared from reagent grade materials, dried, and analyzed by titration with KCNS, 99.12z pure. a
TIME (sed
Figure 4. Potential-time curve for oxalate A = 0.077 cm2, i = 30 PA, C = 4mM
Solutions containing weighed quantities of oxalate were analyzed titrimetrically for oxide and oxalate separately. Oxalate could be recovered quantitatively for several hours at 250 O C. After more than 10 hr, however, the total oxalate concentration decreased in a linear fashion, presumably because of sublimation from the bulk melt. The oxalate could, however, be made to decompose according to the following reaction: c2042-
=
con + co + 0 2 -
if a catalyst were provided. This stoichiometry was verified by trapping both Conand CO as noted above. The catalysts found useful were higher temperature, 280 to 300 O C; a platinum surface; and a solid silver surface. A solid silver surface proved most effective, the amount of oxide produced being linear with time at constant temperature (Table VI). It was suggested that the stable product of silver ion addition to oxalate-containing melts was silver carbonate. This was tested by preparing pure silver carbonate and adding weighed portions to the melt. It decomposed rapidly into silver oxide and carbon dioxide. The gas given off was determined and found to be quantitatively produced, as shown in Table VII. That oxalate was indeed the stable species in the melt was verified chemically. Calcium nitrate was added to a sample of melt in which oxalate had been dissolved and the resulting solution was cooled and dissolved in water. A white precipitate was deposited which was analyzed titrimetrically. By virtue of its equivalent weight, the precipitate proved to be calcium oxalate. It would be difficult to draw any positive conclusions from this evidence. The fact that oxalate remains, of itself, unchanged in the melt leads to the conclusion that the similarity in the current-voltage curves could be coincidental. This is supported somewhat by chronopotentiometry of oxalate-containing melts. Potential-time curves, as typified by Figure 4, 152
ANALYTICAL CHEMISTRY
were similar to those obtained for oxide and gave similar wave analyses. Despite these similarities, however, the chronopotentiometric constant did not remain constant as in the oxide case. Though the value remained constant at any particular concentration, it changed over the concentration range investigated. Also, the average value obtained for the diffusion coefficient is roughly half that obtained for oxide, although it is still of the same order of magnitude, and, as before, still a reasonable value (D = 1.03 X 10-jcm2/sec). The observations concerning the inconstancy of iT1/I/C strongly suggests a kinetic complication. A reasonable possibility would be that oxalate is decomposing at the electrode surface and the electroactive species is indeed oxide. This process could involve the Pt electrode surface as a catalyst because it has already been shown that Pt is capable of catalyzing the decomposition of oxalate. Supporting this suggestion are the many strong similarities between both currentvoltage and chronopotentiometric data for oxalate and oxide, including curve shape, position on the voltage axis, and wave analysis values. Finally, supporting the postulate that oxalate provides oxide at the electrode surface is the fact that current-voltage curves recorded at various intervals during a catalytic conversion process were in no way altered in shape or position on the voltage axis. Finding oxalate per se in samples of melt does not preclude the loss of an oxide ion in the melt because this could easily be returned by hydrolysis on dissolving in water. The only necessary condition is that the carbon-carbon bond remain intact. RECEIVED for review July 7, 1967. Resubmitted February 8, 1968. Accepted August 28, 1968. Work supported by the School of Chemistry, University of Minnesota, and by the Proctor and Gamble Co., the Socony-Mobil Co., and the DuPont Chemical Co. (summer fellowships to P.G.M).