THERMODYNAMICS IN ALUMINUM-PRODCCIXG ELECTROLYTES
Nov., 1861
composition.-Harvey, Edmison, Jones, Seybert and Catto3 found the activation energy for the decomposition of pure potassium perchlorate to be 70 kcal., which coincides with the energy needed to break a chlorine-oxygen bond with formation of potassium chlorate and a gaseous oxygen atom. For the sodium salts a t 298"K., the corresponding energies are7 NaC104(s) = NaClOs(s) NaC103(s) = NaCIOz(s)
+ O(g) + O(g)
AH = 65.6 kcal. AH = 72.2 kcal.
The experimental activation energies for perchlorate and chlorate decomposition in sodium hydroxide are 47.3 and 53.2 kcal., less by 18.3and 19.0 kcal, respectively. This suggests that the influence of the fused salt medium is essentially the same for perchlorate (7) W.&I.Latimer, "Oxidation Potentials," Prentice-Hall Book Co.9 New York, N. Y., 1952.
208 1
and for chlorate decomposition. That the hydroxyl ion is the important factor is indicated by the reduction in rate attendant on diluting the hydroxyl with nitrate. A mechanism which would be consistent with the observations is C10,(210s-
+ OH- --+ ClOa- + HOz+ OH- +C102- + HOn-
followed by a rapid conversion of (3102- to Cl-. It not likely that any significant concentration of Ha2 could accumulate as equilibrium should lie toward the right in HOZ-
+ OH-
=
Hi0
+
02-
and the presumably rapid reaction 202'
+ 2H20 = 02 + 40H-
would cause the eventual disappearance of peroxide.
THERMODYNAMIC CONSIDERATIONS IN THE ALUMINUM-PRODUCISG ELECTROLYTE BY W. B. FRANK Aluminum Company of America, Alcoa Research Laboratories, Physical Cheniistry Division, h'ew Kensington, Pennsylvania Received June 19. I081
The thermodynamic values for NaaL41F6,NaF and AlFs appearing in the literature are corrected for an apparent error in AlF3 temperature measurement. The free energy change for the postulated dissociation mechanism NasAIFe e 3NaF is calculated a t 1300'K. Dissociation according to this scheme is absent or very slight. Thermochemical properties are developed for undissociated li uid cryolite and molten sodium tetrafluoroaluminate. The use of these functions substantiates the dissociation meclanism of molten cryolite as Na3AlFB + 2NaF NaAIF,. The possibility of the formation of Eiodium aluminate by dissolution of aluminum oxide in fused cryolite is considered. To a nominal alumina content of about 4 weight % this reaction is thermodynamically possible a t 1300'K.: 2NasA1Fo 2&03 = 3Na;1102 3NaAlF4.
+
+
+
Introduction In the industrial production of aluminum, an electrolyte consisting primarily of aluminum oxide dissolved in molten cryolite is used. Despite the intensive experimental and theoretical interest in this system, there is disagreement in the literature concerning the ionic constitution of the solvent and the interaction of the solvent with the solute. It has been established that cryolite dissociates appreciably at high temperatures. However, the nature and degree of dissociation remain controversial. IVhile most investigators accept chemical interaction between alumina and molten cryolite, in contrast t o physical solution, numerous aluminates and oxyfluoroaluminatcs have been proposed as reaction products. Fairly complete and reliable thermochemical data are now available for the pertinent constituents of the electrolyte to undertake a thermodynamic study of the dissociation mechanism and a trcatmciit of one of the proposed schemes of solution. The high temperature heat content measurements on cryolite by O'Brien and Kelley' do not agree with the measurements of Albrighk2 The discrepancy in the two data sets results in a significant difference in derived thermochemical functions a t high temperatures. For example, the use of the data of O'Brien and Kelley results in a value (1) C. J. O'Brien and
K. I10 kcal., the reaction is highly improbable and occurs only under “unusual” circumstances. Dodge, in presenting these criteria for the feasibility of a given reaction for an industrial process, emphasized that they are only guides useful in exploratory work. He points out that commercial processes often are based on reactions having positive values for the free energy change. There is no definite value for the free energy change which can be specified as indicating that a reaction will or will not take place. The statement that a reaction “is thermodynamically inconceivable” has no meaning without qualification. Starting with reactants uncontaminated by products, any reaction will have a tendency to proceed to some extent even though the degree may be infinitesimal. The value for the free energy change and the resultant equilibrium constant (AFn = -RT In K ) merely specify a limiting relationship for activities of the products and reactants. While a large positive value for AF, (Le., a small value for the equilibrium constant) generally suggests that a t equilibrium the activities of the products are small compared to the activities of the reactants, the expression for the equilibrium constant must be examined carefully before drawing conclusions as to the possible extent of the reaction. Reaction 30 has in fact been demonstrated to proceed quantitatively a t about 13OOOK. Albert and B~-eit,~’ by rapidly cooling mixtures containing sodium fluoride and aluminum oxide held at 1050’. observed the stoichiometric yield of sodium aluminate by titration of the aqueous extract of the
(25) L. Ivanova, Izuest Vysshakh Ueheb. Zovedenzz Tsvet. M e t , 2 , 67 (1959). (26) J. 1’. Coughlin, J . Am. Chcm. Soo., 80, 1802 (1958). (27) A. V. Christensen, K. C. Conway and K. K. Kelley, Bureau bf Minea Report of Investiaations 5565 (1960).
(28) E. G. King, J . A m . Chem. Soc., 7 7 , 3189 (1955). (29) B. F. Dodge, “Chemical Engineering Thermodynamics,” McGraw-Hill Book Co., New York, N, Y., 1944. (30) 0. Albert and H. Breit, Aluminum Ranshoffen, No. S, 3 (1955).
D. The Formation of Sodium Aluminate in Cryolite-Alumina Fusions.-A recent Russian publication25 states that “the formation of sodium aluminate during the solution of aluminum oxide in molten cryolite is thermodynamically inconceivable. With the addition of surplus sodium fluoride, the possibility of the formation of NaA102 increases but still remains only slightly probable thermodynamically.” These conclusions were based on large positive values derived for the free energy changes a t 1300’K. for the reactions
+ + +
+
xasA1F~ 2A1203 = 3NaA102 2AlFs (29) 6NaF 2.21203 = 3NaA102 NasAIFe (30) 3NaF 2A1203= 3NaA1O2 AlF3 (31)
+ +
The values obtained by Ivanova for the free energy changes a t 1300OK.were 126,527,30,167 and 78,347 cal. for reactions 29,30 and 31. New pertinent thermochemical data are available since the investigation of Ivanova. The values for the heats of formation of some of the compounds involved have been drastically revised. The values used in the original study are compared with the presently accepted values in Table 11. TABLE I1 HEATOF FORMATION AT 25O, AHOf 298 (KcAL.) Compound
Na A 1F A1203 NaAlOz AlF3 XaF
Value used
by lvanova
-759 -399 -270 -311 -136
60 09 84 .OO 30
Presently accepted value
-784 8” -400 -270 -356 -136
-798 9514 412 8426 3l2
312
KOmeasurements of the heat capacity of sodium
THERMODYNAMICS IN ALUMINUM-PRODUCING ELECTROLYTES
Nov.,1961
quenched specimen. Cryolite was identified by an X-ray powder diffraction pattern as the second product of the reaction. It has been established that the sodium aluminate formed is almost completely insoluble3*32 in the resulting liquid phase a t temperatures of about 1000°. Specifying the phases for reaction 30 at 1300'K. GNaF(1)
+ 2A12031s)= 3NaA410z(s)+ NaahlF8(l) (32)
Because the reactant alumina and the product sodium aluminate are solid phases a t unity activity, the equilibrium constant for reaction 32 can be written K = -aNasAlF6 ~'N~F
2087
about 14%,34the sodium aluminate, if present, must be a constituent of the melt a t low alumina contents. The equilibrium constant for reaction 34 can be written K =
a3NsAIOp a3NaA1Fa a'NaaA1Fe
(35)
To test the possibility of reaction 34 proceeding quantitatively for reasonable alumina contents, the activities of all species mere calculated assuming stoichiometric formation of sodium aluminate, ideality of the melt, and equilibrium between NaaAIFs, NaF and NaA1F4 as specified by the equilibrium constant a t 1300°K.21
(33)
The value for this function is calculated for various nominal alumina contents assuming quantitative formation and complete insolubility of sodium aluminate and no dissociation of the cryolite formed. The value of this function increases with increased initial alumina content and reaches the value 3 X [value calculated for K from free energy change for reaction 321 a t a weighed-in alumina content of about 10%. Thus the quantitative formation of sodium aluminate by reaction 32 a t 1300'K. is not surprising with a large excess of sodium fluoride. In light of the present knowledge of the constitution of molten cryolite, a more realistic mechanism for the formation of sodium aluminate in cryolitealumina melts is
+ 2Al2O3(s)= 3NaA102(1)+ 3TaAIF4(1)
2Na3illF0(l)
(34)
The free energies of formation for the products and reactants of this reaction are calculated at 1300'K. The free energies of formation a t 1300'K. for NaA410zand &os are -199.6433 and -301.67 kcal., the same values cited previously. A value of -399.51 kcal. is obtained for the free energy of formation of IYaiilR a t 1300OK. by interpolation of equations 26 and 27. A value of -610.03 kcal. is obtained for the free energy of formation of undissociated cryolite from the expression for ( F T H298)u.l. From the free energies of formation of the products and reactants, the free energy change for reaction 34 becomes 25.9 kcal. This corresponds to an equilibrium constant of 4.5 x Since the solubility of alumina a t 1300'K. is (31) E. Bonnier, Ann. P h y s , 8 , (12), 258 (1953). (32) A. Seyyedi and G. Petit, J. phys. radzum, 20, 832 (1959). (33) The free energy of formation of supercooled liquid sodium aluminate should be used In calculating the free energy change for reaction 34. Since no data are available for molten sodium aluminate the free energy of formation of the solid is used as an approximation.
The resulting activities and the value of the expression representing the equilibrium constant are given in Table 111. TABLE I11
0 2.7 5.4 8.1
0.360 ,329 ,294 .2GO
0.427 ,385 .34G .310
0.213 .239 .266 .293
0 ,047 ,093 .138
0 1.3 X 1 . 8 X 10-4 1 . 0 X 10-3
It is seen that the function representing the equilibrium constant reaches the value 4.5 X 10-5 a t an alumina content of about 4%. This means that i t is possible for alumina to react quantitatively with cryolite to form sodium aluminate up to this composition. The activities shown for the two highest alumina contents are meaningless since quantitative reaction is thermodynamically impossible. If alumina reacts according to reaction 34 in the dilute range it must dissolve by a second mechanism a t higher alumina contents. It is interesting to note that an independent study, freezing point lowering in the cryolitealumina ~ystem,3~ suggested reaction 34 as the mechanism for the solution of alumina in cryolite to an alumina content of about 5%. Above this composition, a rather sharp deviation from this reaction mechanism was observed. Thus both the cryoscopic study and thermodynamic considerations suggest that alumina can react stoichiometrically with cryolite to form sodium aluminate and sodium tetrafluoroaluminate to a nominal alumina content of 4-5%. Additional aluminum oxide apparently dissolves by a second mechanism. (34) P. -1.Foster, Jr., J. A n . Ceram. Soc., 2, 66 (1960). (35) P. A. Foster, Jr., and W. B. Frank, J . E'lectrochem. Soc., 107, No. 12, 997 (1960).
