876
DAVID W. JAMES
The Rate of Oxidation of Nitrite Ions in Dilute Solutions of Sodium Nitrite in Molten Lithium Perchloratel by David W. James Department of Chemistry, University of Queenaland, Briabane, Australia
(Received July 34, 1967)
I n dilute solutions of sodium nitrite in molten lithium perchlorate at 245-301", nitrite ions are oxidized to nitrate ions while perchlorate ions are reduced to chloride ions in a pseudo-first-order homogeneous reaction. The first-order rate constant is 3.51 X sec-I at 274", and the activation energy is 31.5 kcal/mol. There is no buildup of chlorate ions during the course of the reaction. The reaction is unaffected by gaseous nitrogen or substantial additions of water, LiOH ( 5 mol %), or LiNO3 (10 mol %) and by gaseous oxygen. The evidence suggests the possibility that the rate-determining step involves a direct oxygen transfer between a nitrite and a perchlorate ion.
Introduction There is very little quantitative information about reaction rates in ionic liquids and even less information about reaction mechanisms. The singular presence of ionic species, the interaction of which is not diluted by a nonionic solvent, gives these systems their unique character. Thus it has been found that when molten nitrates are used as solvents in acid-base reactions, the nitrate ion enters actively into the reaction mechanism with the formation of YO2+ and N0z.2 On the other hand, in systems containing bromate ions, the participation of nitrate ions was suppressed with the formation of BrOz+ and Br2.3 It is not uncommon for such reactions to be quite fast. In many instances, the rate is probably determined by the diffusion of reacting species. Thus the chemical behavior of many molten salt systems is successfully rationalized by the assumption that thermodynamic equilibrium is attained very rapidly. On the other hand, it is also well known that molten salt reactions which involve oxo ions may be rather slow, as, for example, the decomposition of molten alkali metal nitrates a few hundred degrees above their melting points. In the present paper, we report quantitative data on a slow reaction between oxo anions in an ionic melt. Small amounts of KOz- in the form of NaNOz mere added to molten LiC104 and the N02- anions were oxidized to NO3-, while the solvent C104- anions were simultaneously reduced to C1-. The reaction isotherm was measured spectrophotometrically by determining the change with time of the absorption intensity of NOz- and Nos- bands.
Experimental Section Lithium perchlorate of 99.5% purity (reagent grade) was recrystallized twice from water and dried under a The Journal of Phvsical Chemistry
current of dry nitrogen, while the temperature was slowly raised to 270". The molten salt was maintained at this temperature for several hours with dry nitrogen bubbling through it. Then the salt was filtered through a fine Pyrex filter and sealed in a Pyrex tube under a reduced pressure of dry nitrogen. The dry, solidified material was subsequently handled in a controlledatmosphere box. The chlorate content of this material was 0.004 mol %. Reagent grade sodium nitrite was recrystallized from water, dried under reduced pressure at 100" for 48 hr, and stored in a desiccator. Lithium nitrite4 was recrystallized twice from water and dried at 50" under reduced pressure. This produced the monohydrate. Lithium hydroxide, lithium nitrate, and sodium nitrate were recrystallized from water and dried by heating to their melting points under vacuum. Reaction isotherms were determined by measuring the ultraviolet spectra of the solution as functions of time. Lithium perchlorate is essentially transparent to 190 mp, while nitrite, nitrate, and chloride ions have absorption bands at longer wavelengths in the ultraviolet. The nitrite ion dissolved in molten lithium perchlorate has an absorption maximum at 330 mp, while the nitrate ion has a maximum of 286 mp. Either of these bands may be monitored to follow the course of the reaction. The nitrate peak occurs at a wavelength where nitrite (1) Part of this research was sponsored by the U. S. Atomic Energy Commission under contract with the Union Carbide Cow. (2) F. R. Duke and S. Yamamoto, J , Am. Chem. SOC.,81, 6378 (1959); F. R. Duke and M. L. Iveraon, ibid., 80, 5061 (1958); Anal. Chem., 31, 1283 (1959). (3) E". R. Duke and J. Schlegal, J . Phys. Chem., 67, 2487 (1963). (4) Dr. D. Gruen, Argonne National Laboratory, is thanked for the gdt of LiNOz*H20.
877
THERATEOF OKIDATIONOF KITRITEIONS absorbs somewhat, while the nitrite maximum has little or no significant contribution from the nitrate band at the concentrations we studied. The chloride ion has an intense absorption in the region 180-200 mp, but our measurements were made beyond the effective zero of the tail of this band. The chlorate ion spectrum was determined and found to have a weak band (-0.1) a t a wavelength of 286 mp.6 However, since this ion was present only in trace concentrations, this band did not contribute to the measured absorbance. The reaction vessel was a standard l-cm square absorption cell made of fused silica. The absorption was continuous1,y monitored at a fixed wavelength with a Cary Model 11 spectrophotometer. The furnace and absorption cells used have been described previously.6 Abso:rption measurements mere made against air as a reference. Losses due to solvent absorption and internal reflections were measured before addition of nitrite and subtracted from subsequent measurements. A sample of liiC104 was weighed into the absorption cell in the drybox and transferred, under an atmosphere of dry nitrogen, to the preheated spectrophotometer cell compartment. The salt was allowed to melt and come to equilibrium at the temperature of the measurement. Then all air bubbles that formed during melting were removed by vigorous stirring. The absorption of cell with solvent was determined over a small wavelength interval in the neighborhood of 330 mp. Then an addition of dry NaKOz to give a concentration between 0.05 and 0.3 mlol % was made, and the mixture was stirred for several minutes. Stirring was carried out manually and also by bubbling with dry nitrogen gas, both methods giving identical results. It was not possible to stir continuously during spectrophotometric measurements. Therefore the reaction mixture was stirred for 30 sec at 15-min intervals and allowed to equilibrate. The absorption of 330 mp was followed continuously (including mixing periods). The temperature fluctuation during the addition and mixing procedure was estimated to be no more than 2-3". The increase in concentration of chloride and chlorate ions was determined for a selection of samples after the reaction was complete. The furnace temperature was controlled by a Leeds and Northrup Speedomax G/DAT controller, and the optical cell inside the furnace was located within a silver block which acted as a heat reservoir. The temperature of this block did not vary by more than 0.5" during 24 hr. A second series of measurements was made to examine the effect on the reaction of changing some of the chemical variables. These reactions were carried out in Pyrex tubes heated in an aluminium-block tube furnace controlled by a Thermoelectric Model 80410 proportional controller. The reaction was allowed to proceed for a fixed time, after which the tube was
chilled to solidify the reaction medium. The increase in chloride ion content of the medium was determined by one of three methods. In most cases, a titration with mercuric nitrate using diphenylcarbazone as indicator was used.' However, samples that contained an excess of nitrite, and also some other samples, were analyzed either by Volhard titrations with silver nitrate or by gravimetric precipitation as silver chloride.8
Results and Discussion There was a brief initial period during which the reaction rate showed irregular behavior. This normally lasted from 10 to 20 min, and for the whole of the subsequent time, the reaction isotherm was simple and regular. The reaction was studied at five temperatures from 246 to 301". At each temperature, several determinatioiis of the reaction isotherm were made to obtain the final kinetic parameters. Results for duplicate runs agreed to within experimental uncertainty. At all temperatures, the nitrite concentration followed a first-order rate of decrease that was independent of the initial nitrite concentration for the dilute solutions studied (0.05-0.30 mol %). The collected data from all determinations are given in Table I. When an Arrhenius plot is made of these data, it is found that all, except the lowest temperature datum point, fit a straight line. The lowest temperature determination was ignored in the Arrhenius plot, because difficulty was Table I: First-Order Rate Constants
245 260 274 290 30 1
0.28 0.1-0.3 0.15-0.3 0.2-0.3 0.2-0.3
0.404 i 0 . 1 1.79 rt 0.05 3.51 =t0.05 9.61 =k 0.05 13.1Oj=O0.05
encountered in measurements within 10" of the melting point of LiC104, so it was considered that large errors might be associated with the data a t 245". The activation energy and frequency factor derived from the remainder of the data are 31.5 kcal and 8.91 X 108 sec-l, respectively. Since the reaction involves an oxygen transfer of some sort from perchlorate to nitrite, a material balance provides some information on the reaction. It is known (5) Dr, W. G. Williams is thanked for the gift of a sample of anhydrous LiC108. (6) C. R.Boston and G. P. Smith, J. Phys. Chem., 62, 409 (1958). (7) F. E. Clark, Anal. Chem., 22, 553 (1950); I. Roberts, Ind. Eng. Chem. Anal. Ed., 8,365 (1936);G. €3. Smith, Anal. Chim. Acta, 7, 330 (1952). (8) A. I. Vogel, "Quantitative Inorganic Analysis," 2nd ed, Longmans, Green and Go., London, 1957,pp 251, 258.
Volume 78, Number 9 March 1968
DAVIDW. JAMES
878 that the decomposition of sodium perchlorate in fused hydroxide takes place in two steps; the first step leads to formation of sodium chlorate and is about 18 times as fast as the second step which is the deconiposition of chlorate to ~ h l o r i d e . ~If a similar reaction path were followed in the present reaction, chlorate ion would be the dominant reduced species with small amounts of chloride ion. Analysis of reaction mixtures from several kinetic determinations showed no increase in chlorate concentration over that initially present in the lithium perchlorate (0.004 mol %). If an over-all reaction obeying the equation C104-
+ 4NOz-
4 C1-
+ 4NOs-
(1)
is postulated, then a final concentration of chloride ion equal to one-fourth of the initial nitrite concentration should be formed. The results of several determinations of terminal chloride concentration are presented in Table 11, which shows that complete reduction of the perchlorate ion takes place. The terminal concentration of nitrate ion was checked spectrophotometrically (interference from other ions was negligible) and found to agree closely with the concentration of sodium nitrite originally added. Table I1 : Stoichiomet,ry of Over-all Reaction W t of
Concn of
NaN09,
NaNOz,
g
0.0282 0.0206 0.0157
mol
70
0.08 0.07 0.04
OC
measd, g-ions X 1OJ
Wt of C1- oalcd, g-ions x 10,
280 280 275
3.58 =t0.15 2,44 =t0.15 1.77 =t0 . 1
3.58 2.59 1.97
wt of c1Temp,
Seward and Otto9 found considerable dependence of the perchlorate decomposition rate on the nature of the cation. The cations in our melts are predominantly Li+ with a small concentration of Xa+ from the I\JaN02 added. Kevertheless, an attempt was made to carry out the reaction with lithium as the only cationic species. iinhydrous lithium nitrite is unstable and so the monohydrated salt was used. It was not possible to obtain consistent reaction isotherms at any temperature in the range studied (245-301'). The reaction isotherms were the same when either silica or Pyrex vessels were used as containers. Furthermore, when finely ground silica or Pyrex was added in amounts sufficient to increase the surface area by at least an order of magnitude, the change in the rate constant measured at 270" was less than experimental uncertainty. These measurements indicate that the reaction is homogeneous. The effect of adding a number of chemicals in varying concentrations was carried out in order to elucidate the reaction mechanism with the results given in Table 111. The Journal of Phvsical Chemistry
The concentration of water added was quite uncertain as the lithium perchlorate dehydrates at the temperatures used. The addition of lithium hydroxide and nitrate demonstrates that the reaction is independent of minor concentrations of hydroxyl and nitrate ions (in contrast to the findings of Seward and Otto for related systemsg). Presence of dissolved oxygen gas or larger cations however had a pronounced retarding effect,the effect of oxygen persisting even after the melt was well scrubbed by bubbling with dry nitrogen. It may be noted that an increase in glass surface did not alter the measured rate in the presence of dissolved oxygen, showing that the retardation was independent of surface area. The pronounced retardation of the reaction noted when significant concentrations of NaXOa were added must be attributed to the presence of S a + ions in the melt, as no retardation was noted with LiN03. We therefore conclude that the strong polarizing power of the lithium ion (compared with the sodium ion) probably plays an important role in the reaction mechanism. Although the data presented here do not identify the reaction path or permit an unambiguous specification of the rate-controlling step, they are in better accord with certain possible rate-controlling steps than others. The reaction, as shown in eq 1, goes all the way to chloride, so that four oxygens are transferred for each perchlorate consumed, The reaction is very efficient; virtually all oxygens end up in nitrate ions rather than oxygen gas. Possible intermediates include CIOa-, CIOz-, and C10-. There is no buildup of C103-, so that if this particular intermediate is formed, it is consumed very rapidly compared with the rate-controlling step, Furthermore, the analysis of quenched-reaction mixtures indicates that C1- builds up about as fast as KO,- reacts, so that it would appear that the reaction of other intermediates is not rate controlling either. We are led to suppose that the rate-determining step is an oxygen transfer between Clod- and NOz-. This rate-determining oxygen transfer may take place by direct nitrite-perchlorate contact or by diffusion. The latter alternative corresponds to decomposition of the perchlorate, There are two considerations which suggest that perchlorate decomposition is not rate controlling. First, the oxidation of the dissolved nitrite was definitely faster than the decomposition of lithium perchlorate in the absence of nitrite. Second, the rate of decomposition of perchlorate in molten NaOH is reportedg to be lo4 times as fast as the same reaction in pure NaC104 or KC104, whereas we found that additions of 5 mol % LiOH to LiC104 had no appreciable effect on the rate of oxidation of dissolved nitrite. When lithium is the only cationic species, anion contact is inevitable, and, thus, it is considered (9) R. P. Seward and
H,W.
Otto,
J. Phys. C h m . , 6 5 , 2078 (1961).
879
THERATEOF OXIDATION OF NITRITEIONS Table I11 : Effect of Added Reagents on Rate Constant” Added chemical
Concn, mol %
Temp,
OC
Rate constant X 104, 8ec -1
Comment
HzO
8-40
265,275
2.3, 3 . 5
Water added before fusion
Nz
...
265-300
2.3-13.1
Nitrogen gas bubbled through melt
265 275
0.7 \ 0.85/
Oxygen bubbled for 10 min before and after nitrite addition
..
275
0.85
Oxygen as above followed by nitrogen bubbling for 10 min
5 10 10 20
270 270 270 270
2.9
...
+ Nz
01
LiOH LiNOa NaNOa
Salts added to the LiClOa before fusion 0.08
Rate constant estimated by determination of chloride concentration in quenched reaction.
that oxygen transfer probably involves an anion-anion contact pair. No details of‘ subsequent steps in the reaction can be verified. It is known however that LiC103is extremely unstable above 170°,10and, hence, a rapid decomposition of the chlorate ion is assumed. The resultant atomic oxygen could either form molecular oxygen or diffuse through the melt and react with nitrite ions. Stoichiometric considerations and the absence of evolved oxygen support the latter proposal. Although the above mechanism is plausible and describes the experimental isotherms reasonably well, it does not explain why the reaction is retarded by larger cations, is unaffected by anions, water, or nitrogen, and is reiarded by molecular oxygen. On a size basis, the larger cations would be expected to hinder anion-anion contact more effectively than the lithium ion. It is unlikely however that the sodium ion is large enough to produce the observed retardation on this basis. The small size of the lithium ion confers a high polarizing power which may act on either or both of the anions. It is known that the perchlorate ion is difficult to polarize and the lithium ion will tend to %olvate” itself with polarizable species either ionic or dipo1ar.l’ The nitrite ion being very unsymmetrical is very readily polarized. Hence whereas the lithium ion has a relatively minor effect on the perchlorate ion, the nitrite ion is strongly polarized, and, as shown by
the instability of lithium nitrite, this polarization decreases the potential energy of the ion. The active complex may then be seen as the lithium-nitrite ion pair and the problem of the kinetics resolves itself into a consideration of the probability of a suitably orientated collision. The determination of the symmetry of this ion pair and the nature of the Li+-02N interaction are far from simple and will be the subject of a separate communication. On the basis of the active participation of Li+ in the reaction mechanism, the retarding effect of sodium ions is understood as predominantly due to a reduction in the average polarization of the nitrite ions. The effect of dissolved oxygen gas is of obvious importance to the reaction mechanism. The effect is evidently quite complex, however, since the oxygen is not removed by prolonged scrubbing with nitrogen and more work is necessary before the action of dissolved oxygen can be adequately described.
Acknowledgments. Dr. G. P. Smith and Dr. J. H. O’Donnell are thanked for helpful discussion during the preparation of this paper. Mr. J. Burnett and Mr. D. Clegg are thanked for assistance in some of the analytical determinations. (10) D. G. Williams, private communioation, (11) D. W. James, W. H. Leong, and R. C. Marshall, unpublished
work.
Volume 72, Number 3 March 1968