With a = : 2.96 A,, the nearest Xi-Xi separation in i'ii0, we find a t 1500°K. A H * = 42.6 kcal. and AS" = -- 7.4 cal. deg.-l. As previously pointed out,8 these quantities are composites of the thermodynamic AS0 and AHQfor the formation of the defect' responsible for diffusion and the kinetic AHi* and A&* for the mobility of the defects. There i s good evidence that, in thcse oxides with the rocksalt structure cations diffuse by a vacancy mechanism.9 Excess oxygen dissolves in N O by the reaction
l/ZOp,
+ Ni+2--+
O=
+ VNi+
+
(8)
[ M i t ~ f f gassumes ~ that the 'vacancy is doubly ionized, but his data are in better accord wit'h the more reailsonable assumption of eq. 8.1 From eq. 8
[vsi+]= ~ ~ / x p ~ , ~ / & We then havc AH*
=
AHi*
AS* = ASi"
+ "H8 + / 2
'/2AX8
where Ahrs and AX8 are the enthalpy a,nd entropy of reaction 8. The data of Nitoff and of Carter and RichardsonlO yield AH8 ilc 30 kcal. and AS8 'v - 12 cal. deg.-l. Therefore, for the step in which a (8) W.J. Xoore and J. K. Lee, Trans. Faraday Soc.. 48, 916 (1952). (9) (a) C. E. Birchenall, Met. Rev., 8 , 266 (1958), reviews the eubject; (b) 8. P. Mitoff, J . Chem. Phys., S6, 882 (1961). (10) R. E. Carter and F. D. Richardson, Trans. A I M E , 194, 1244 (1984).
nickel jumps into a ncighboring vacancy AHi" = 27.6 and AS;* = -1.4. I n the case of pure C00,'o AH,* = 30 andfAH8 = 10. It is no1 surprising to find the value of AH,* as much as 2.4 kcal lower in the case of NiO. The concentration of trivalent] impurities in the General Electric S i 0 was about 3.8 X which is comparAble with the total concentration of vacanciea. It might at first seem likely, therefore, that the AH8 in the S i 0 samples had been lowered by the traces of impurities. Yet it mould not assist the D of to have a vacancy tied to a Cr+3 site. Therefore the impurities may not have any marked efiect on the D. This view is supported by the fact that the diffusion coefficients in the crystals containing 60 X cobalt were not significantly different from those in the purer crystal. The ionic radii according to Pauling , 0.69. Thus we are Fef2, 0.75; C O + ~0.72; mould expect to have a somewhat lower AHi* for this reason. The most consistent interpretation therefore appears to be the one given, which neglects the effect of trivalent impurities. It is quite clear from all the data on FeO, COO, and NiO that the activation energy for 1) in Xi0 should be close to the value of 45 kcal. The higher value of 56 kcal. given by I,indner,3 even with AH," = 30, mould require AH8 = 52. Such a value would be inconsistent with the observed excess oxygen concentration in KiO, which a t 1350' is only about 1/20 as large as the value in COO. Such a factor corresponds to a difference in AHs of 18 kcal., so that with Richardson'slo accurate valueof AH* (COO) = 10, the AH8 (NiO) should be about 28 kcal.
OXIDATIOS OF PIIETALS I N MOLTEN SALTS. 81LJ7ER IS SODIUM CHLORIDE1 BYKURTIT. STERX Electrochemistry Sectzon, iVational Bureau of Standasds, IVashangtota, U . C'. Received J a m a r y 13, 1962
The rate of oxidation of silver in sodium chloride near BOO" has been determined under a variety of conditions. When thc rate of O2 transport to the Ag-I\r'aCl interface is low, the reaction is zero order; when it is high, the rate varies as t/= indicating a rate controlled by the diffusion of products away from the reaction site. Undrr all conditions the Ag+/O' ratio is greater than ran be :iccounted for by the formation of Ag,O. The results can be accounted for bv hypothesizing two siiiiultaneous reactions. ( a ) 2Ag I /2Ol 7 2Agi ()', (b) Ago K a + 4Ag+ +.Sao, which produce"ionic silver, the relative contributions to total Agf of the reactions depending 011 conditions. The equilibriuiii constant for(a) is 8.3 x 10-7 on the mole fraction scale.
+
+
Introduction The oxidation of metals in and by molten salts is of interest from several points of view. In a
molten salt, electron transfers not possible in aqueous solution may occur, the subject has practical applications in the field of corrosion, and the particular system discussed here is rele~7antto the Ag-AgC1 reference electrodes commo~dyused for electrochemical measurements in fused salts. ( 1 ) Presented in part a t the 137th National RIeetlnv of the American C l i c m i ~ a lSociety, Clrvelnnd April. 1960
.+
That silver is oxidized in molten sodium chloride has been known for a long time.2 Suggestions that
t'he reaction proceeds, in the presence of air, by t'he formation of metallic sodium which then is oxidized to KazO, are likewise old.3 Recently, this idea has been formulated more precisely4.5 by suggesting ( 2 ) K. A. Winkler, "EuropLische Amalgation der Silbererze und silberhaltigen Htittenprodukte," Freiberg, 1846. (3) H. Rose, P o Q ~Ann., . 68, 286 (1846). (4) K. Bmrjek, I. Sekerka. and V. Seifert, Chem. L i s t y , 6 0 , 721 (lU.50). ( 6 ) IC. H. Stern, J . l'hys. Chem., 62, 385 (1958).
KURTH. STERN
1312
that, although the equilibrium for the reaction Ag NaCl = AgCl Na lies very far to the left ( K = 1.07 X 10-lo a t 900') the reaction may be driven by the vaporization of metallic sodium from the system. One interesting feature of this mechanism5 is that it does not require the presence of oxygen, which would presumably lead to the formation of oxide ion. Thus, an experiment in the absence of oxygen would clearly distinguish between two mechanisms possible in the presence of oxygen; the sequence
+
+
+
+
Ago Na+ = Ag+ KaO 2 NaO $. 1 / 2 0 2 = Na20
(1) (2)
and the reaction sequence which is probably analogous to that in the aqueous systemJ6vix.
+
2Ag '/zQa = AgzO Ag20 3. 2NaC1 = ZAgC1 NazO 2Ag
+
1/202
+
-
+ 2NaC1 = 2AgC1 + NazO
which is equivalent to the net reaction 2Ag(s)
+ '/zOz(g)
(2Ag"
+
O-)inNaCl
(3)
Although, in the presence of oxygen, both mechanisms lead to the same products, the second mechanism might be expected to lead to an equilibrium since all products remain in the system, whereas the first need not if Nao(g) continually distils out of the system. In that case no definite stoichiometric relation between Ag+ and 0- need exist, except that more Ag+ should be in the system than is expected from the formula Ag20. From another viewpoint the presence of Ag20 in the molten salt would be interesting since the pure compound itself is unstable at this temperature. In order to suppress reaction 3 altogether it, would be necessary t o reduce the oxygen partial pressure in this system to zero. Since this is virtually impossible we have restricted ourselves to comparing the rate of oxidation a t 1 atm., in air, and in argon in which the O2 pressure is atm. I n addition, an attempt was made to measure synthetically the quasi-solubility product [Ag+ ] 2[O-] which might be expected to apply to reaction 3, i.e. K
=
[Ag']2[O']/po21'2
(4)
by additions of AgCl and NazO to NaCl melts.
Experimental I n order to avoid possible reactions with containers, none of which is truly inert over long periods of time, the silver to be reacted was made into a container for most of the kinetic studies. Other .reactions, described below, were carried out in alumina and mullite tubes. Materials.-Silver crucibles, 6 in. long, 1 in. o.d., approximately in. thick, and weighing -400 g. were prepared by electrodeposition onto a stainless steel mandril. For runs in oxyqen the tubes mere washed in NH40H and H20 and dried. For runs in argon they were additionally pumped out as described below. For some experiments (see below) 1 in. long capsules with caps were made from */4 in. silver rods. Spectroscopic analysis of the silver crucibles gave the following impurities (weight 7 0 ) : AI 0.0005, Cu 0.0005, Fe 0.005, Mg 0.001, Pb0.0005, SiO.001. For miscellaneous experiments silver wire and sheet of rmnt quality were used. Merck Reagent NaCl was dried by heating under continuous pumping at 500" for several days, (6) In aqueous NaCl silver is oxidized only in the presence of oxygen: SOC. chim. France, 454 (1952).
J. C. Pariaud and P. Arohinard, BdZ.
Vol. 66
using a diffusion pump. Some batches also were treated with HC1 gas after drying and again pumped down to -0.1 ,u. NOsignificant difference in the reactions (with Ag) produced by the HC1 treatment was noted. Results of spectroscopic analysis of a typical batch of NaC1 before and after reactions with silver are shown in Table I.
TABLE I ANALYSISOF NaCl (WEIGHT %) AI Cu Fe Mg Xi Si -4g Before 0.0001 0.001 0.0001 0.00005 0,0001 0.0001 0,0001 After 0.0005 0.001 0.002 0.0001 0.0001 0.005 >1
Fisher C.P. Ag20, Baker Reagent Na202, and Mackay Na2O were also used in some experiments. The latter is not pure but, contains some NazOz. Apparatus.-The apparatus used in most of the experiments carried out under controlled atmosphere is shown in Fig. 1. The lower part, of Vycor, containing the silver tube, had to be renewed after nearly every run because of extensive attack, which results in transition t o crystalline trymidite as the container is cooled. The upper part, having the male standard taper joint, was equipped v i t h three stopcocks m shown to permit a variety of operations. Since the thermal expansion of Pyrex is greater than that of Vycor, the system becomes self-sealing at high temperatures. Stopcocks A and €3 serve for entrance and exit of gases or for attachment to a vacuum pump while the brass assembly rests on O-rings on the glass tubing projecting above the stopcock C. The stirrer is introduced into the u per end of the tube through O-rings before C is opened. imilarly, on withdrawing the stirrer, C is closed after the lower e$ of the stirrer has just passed it. This arrangement permits stirring of the melt without having air leak past the stirrer. The melt can be sampled by dipping a cold porcelain rod into it and quickly withdrawing it. Such porcelain rods also served as stirrers. Furnace and Temperature Regulation.--RIost experiments were carried out in a 10 in. Marshall furnace, appropriately shunted to minimize temperature gradients, regulated by a Marshall control panel. Temperaturcs were generally measured by dipping a Chromel-Alumel thermocouple into the melt a t the end of a run. After it was found that changes of 50" had only a slight effect on the reaction rate, subsequent runs were only made a t a single temperature, goo", which lies conveniently between the melting points of NaCl(800") and Ag (960'). Chemical Analysis.-Most of the analysis for silver in XaC1 was carried out colorimctrically.~ The method gives good results for ionic silver down to 0.05 mole 70.For concentrations near or below this a spcctroscopic method was used. Good agreement between the methods near 0.05% makes it clear that the silver in NaCl melts is ionic and places the upper limit for atomically dissolved Ag near this value. More decisive evidence is provided by the clectron spin resonance spectrum of the quenched melt (see below). Some melts having high silver concentrations were anal zed by electrodeposition. d l t s were analyaed for oxide ions by titration with 0.01 N HC1 to pH 4.5, blanks being run for the NaCl and for the H20 used to dissolve the samples. Rate Studies. A. In OnAtmospheres.-A large number of experiments using the Ag crucibles and Vycor apparatus was carried out. The effects of 0%pressure, rate of on transport to the metal, and temperature were studied. In these experiments 40-60 g. of dried NaC1 wm used and samples of a few tenths of a gram were taken a t intervals by dipping a cold porcelain rod into the melt and quickly withdrawing it. The 02,previously dried by passing over CaClg, MgC104,and freed of C02by Ascwite, was slowly leaked through the stopcock into the vessel. A number of runs also were carried out with undried NaCl and in crucibles open to the atmosphere. In still others, 02 was bubbled through the melt, to provide maximum 0 2 concentration throughout the melt. Runs in "Oxygen-Free'' Environment.-A number of experiments were carried out to determine if oxygen is required for the oxidation of silver. None of these was entirely successful as the system is inordinately sensitive to 0 2 and concentrations of 0 2 below 0.01% are difficult to measure in a (thermodynamically) open system. I n a (7) 8. Siggia, Anal. Chem., 19, 922 (1947).
July, 1962
OXIDATIONOF METALSIN MOLTEN SALTS
typical experiment a silver crucible was outgassed in the reaction vessel at 800" using an oil diffusion pump and liquid N2 trap (10-6 mm.) for a week and then cooled under vacuum. This procedure removes 0 2 dissolved in the metal.**$ The crucible then was filled with previously dried NaCl and again pumped down for a week, the temperature being raised gradually to 600'. After cooling, the entire assembly wm transferred to an argon filled drybox'" in which the 0 2 partial pressure could be maintained between and 10-6 atm. indefinitely. (The dew point is near - g o 0 , corresponding to 10-4 atm. € 1 2 0 . ) The apparatus was placed in a furnace and the temperature raised to 900". Stopcocks were opened and the porcelain stirrer introduced. At intervals, the melt was sampled M described previously. Mass transfer in this case was nearly as extensive, though slower, as in pure 0 2 . As a variation on this procedure, the temperature of the assembly, after a week of salt drying a t 500°, was raised to 900" under continuous pumping. I n this case, since the total pressure is only ~ 1 0 - 6atm., the The method does not perpartial pressure of 0 2 is -10-7. mit Sam ling the melt at intervals, however. The salt vaporize: within a day and deposited on the cool parts of the apparatus. It contained appreciable quantities of Ag' and 0- ions. Deterdination of Ae2O Solubilitv Product.-Solutions of AgCl in NaCl were prepared by mdting the components together in an alundum tube in the drybox at 900". The melt was poured out to cool into an enamel tray. Solutions of NazO in NaCl were prepared in similar tubes, but in air, since the NazOz contained in the starting material decomposes with 0 2 evolution. A test for peroxide in the resulting cooled melt showed that -1% of the oxide concentration was present as peroxide. Various portions of the AgCl and NazO solutions were melted together until precipitation of metallic silver occurred -(qualitative description below). Samples of the supernatant liquid were taken for analysis. These experiments were carried out both in air and in the drybox. Evaporation of AgCl and NazO from Melts.-Since the composition of reaction melts may be altered by preferential evaporation of some components, this was determined sep' NazO-NaCl solution a t arately. Heating of a 2 mole % 900" in air showed that the composition of the melt remained unchanged for 6 days. However, a similar study of an AgC1-NaC1 melt showed that the preferential evaporation of AgCl is considerable; e.g., a melt initially of 1.6 mole % AgCl decreased to 1.374 in 21 hr. "Closed-System" Ex eriment.-In an attempt to obtain equilibrium data small x g capsules 1 in. long and machined from b / g in. diameter rods were filled with molten NaC1, closed by welding on silver covers, and each tube sealed into a separate Vycor envelope under vacuum. The tubes were placed in a muffle furnace a t 880" and removed a t intervals for analysis. After a day or so molten NaCl could be seen outside the capsules (but inside the Vycor). Careful examination revealed no imperfections in the capsules. It must be concluded that the NaCl diffused through the metal. For most of the run, then, the capsules lay in a pool of molten salt. The concentrations of Ag+ in the salt increased from 0.19 mole % a t 51 hr. to 0.47% a t 284 hr., considerably slower than in systems open to 0 2 . Analysis of Silver for Sodium.-In order to investigate the possibility that any metallic sodium might, instead of vaporizing, alloy with the silver, some pieces of silver sheet were heated in NaCl a t 900" in an alumina crucible under drybox conditions. After one day the melt and metal were poured out, onto a tray. As soon as the hot silver became exposed to the atmosphere, although the latter is very low in 0 2 , yellow flames burst from the metal and continued burning fox: several minutes. Several pieces of the cooled metal were subsequently freed of adherent salt mechanically, washed briefly in dilute ",OH, and separately dissolved in HNOa. The resulting solutions were analyzed for sodium b flame photometry, and for chloride ion. Both Na+ and ions were found, indicating diffusion of salt into the metal, but the results on different pieces fluctuated considerably. All were near mole fraction in the ions,
d-
(8) E. W. R. Steacie and P. M. G. Johnson, Proc. Roy. SOC.(London), A l l S , 642 (1926). (9) D. N. Craig, J. I. Hoffman, C. A. Law, and W. J. Hamer, J . Res. Null. Bur. Std., 6 4 8 , 381 (1960). (10) J. M. Sherfey, Ind. Eno. Chern., 46, 435 (1954).
1313
BALL BEARINGS-
C
A
FURNACE LEVEL
1
-CERAMIC STIRRER
MELTLEVEL
-SILVER CRUCIBLE
38 m m VYCOR
Fig. l.-Apparatus
for kinetic runs.
however. For example, one piece washed in dilute ",OH to remove adherent AgCl gave [Na'] = 2.4 X lom4,1'21-1 = 0.9 X 10-4, or [Na+]/[Cl-] = 2.6, whereas another piece washed only in HzO gave [Na'] = 1.5 X [Cl-] = 1.1 X lo-*, [Na+]/[Cl-] = 1.4.
Results A. Qualitative Description.-All experiments in which NaCl was fused in Ag containers as described above exhibited certain features, irrespective of quantitative differences. These are briefly described. After about an hour the melt becomes pale yellow, while quenched samples are light gray. With time the quenched melt looks almost black although the same material when molten is quite homogeneous and yellow. A microscopic examination of the black material shows that the (macroscopic) color is produced by a very small number of black, opaque crystals in a matrix of clear transparent ones. If oxygen transport to the melt is restricted, the first color in the quenched melt is a faint brown. Under the microscope this is seen to rcsult from a very few (
