The thermodynamic properties of sodium and potassium dissolved in their molten chlorides, bromides, and iodides

THE THERMODYNAMIC PROPERTIES OF SODIUM AND POTASSIUM DISSOLVED IN THEIR MOLTEN CHLORIDES, BROMIDES, AND IODIDES M. V. SMIRNOV, V. V. CHEBYKIN and L. A. TSIOVKINA Institute of Electrochemistry, (Received

Sverdlovsk,

7 October

U.S.S.R.

1980)

Abstract-The saturated vapour pressures are measured as a function of temperature and alkali metal concentration for the liquid solutions Na + NaCI, Na + NaBr, Na + NaI, K + KCI, K + KBr. and K + KI. The sodium and pot8s&m partial pressures are determined. The expressions for their temperature dependence are. obtained. Vaporization heats, activity coethcient, partial excess enthalpies and entropies are calculated from the experimental data for sodium and potassium in these solutions. Equations are derived for the redox potential 8s 8 function of temperature and the alkali metal concentnrtion in the melts. The cathodic polarization of an iron electrode in molten NaCI. KCl, and NaBr is investigated. The steady-state alkali metal concentration in the cathode layer of the electrolyte is shown to be tincar in the current density, varying from - 300 to Id A m -r. The effective alkali metal diffusion coefficients estimated from the linear relation are much greater than the diffusion coeficients of the cations Na+ and K+ in the same molten salts. The cathode potentials are found to become nearly independent of the current density when it exceeds * l@Am-*. the stead+state metal concentration in the cathode layer of the electrolyte being still far from saturated:

I.

INTRODUCTION

Investigating the thermodynamic properties of the alkali metals dissolved in their molten halides is of great importance for the electrochemistry of molten salts. They are generated inevitably on the cathode in electrolysing

the electrolytes

containing

molten

alkali

halides. Their formation can inflttenee electrode processes. Thus, reducing cations takes place not only on the cathode surface but also in the bulk of the electrolyte, with the deposited metal being dispersed in it. The current efficiency falls as a result of the alkali metal evaporating, its consumption in some concurrent chemical reactions, or of oxidizing on the anode. The metals and alloys, the steady electrode potentials of which in molten alkali halides are more than 2 V electronegative towards the corresponding halogen electrodes, are oxidized by alkali cations to yield alkali metal solutions in the molten salts. The alkali metals leaving the reaction zone through the gaseous phase or otherwise, metal corrosion can proceed rather intensively. Such formation of the alkali metal solutions contributes greatly to the noncurrent transfer of electronegative metals (eg Be, Ti, Zr, etc.) onto more electropositive ones (eg Cu, Fe, MO, W, etc.) through the molten halides[l]. At present the cells with alkali metal electrodes and halide melts attract the attention of many investigators. Naturally, there is the problem of metal solubility in the electrolytes. One could refer to other cases where one deals with alkali metal solutions in their molten halides. These solutions are also of great theoretical interest; they are the systems where one can follow the alteration of their properties in the gradual transition from the most typical heteropolar compounds to the most typical metals. The thermodynamic properties for instance produce interesting information concerning the ionic and metallic bond. 1275

Indeed, the alkali metal solutions in their molten halides can be regarded as nonstoichiometric compounds[2] where halogenide anions and electrons are substituted for each other. In the available literature there is little on the thermodynamic properties of such solutions. In two only were the mixing enthalpies of liquid c&urn with its molten fluoride[3] and chIoride[4] measured by the direct calorimetric method. Most papers deal with the phase diagrams for alkali metaUalides[5]. The rest[6-ll] give the excess partial thermodynamic function, valued by approximate methods from the experimental phase diagrams. Direct experimental temperature and concentration relations being unknown, the authors proceeded in their calculations from model assumptions. Experimental works on the thermodynamics of alkali metal solutions in their molten halides are fairly scarce, which may be explained to a certain extent by difficulties in experimenting with such highly reactive objects and confine themselves mainly to measuring the redox potentials. This method, however, proved to be rather unreliable because fixing alkali metal concentration in molten salts at elevated temperatures is a matter of great difficulty. Moreover, the alkali metal solutions possess appreciable electron conductivity which can distort the results of the redox potential measurements. In this respect, the tensiometric methods seem to be more reliable. Thus, Bredig et al-[23 used the boilingpoint method for determining the partial pressure of potassium in the saturated vapours of the solutions K + KF. The observed data are suspect because the metal concentration could change considerably during measuring. By saturating the molten salts from the vapourous phase the partial pressures of sodium and potassium in the saturated vapours of the solutions Na +NaC1[13] and K+KCl[14, 15-J were found as a

1276

M. V.

SMIRNOV,

V.

v.

CHEBYKIN

function of temperature and concentration to an accuracy of 25 per cent and 13 per cent correspondingly. No mixed compounds of sodium or potassium with their chlorides were shown to be present in the vapours. The partial pressure of cesium in the saturated vapours of the liquid mixtures Cs + CsF containing 360 mol o/e was measured by the static method[16]. The construction of the apparatus used did not allow metal concentration changes through evaporation or casual losses to be taken into account. This circumstance reduces the reliability of the measurements. We also made use of the static method for measuring saturation vapour pressures of the sodium and potassium solutions in their molten chlorides, bromides, and iodides, our device being improved to eliminate the above defect.

2. EXPERIMENTAL 2.1. Partial pressures of wdium and potassium in saturated vapours of the solutions Na + NaCl, Na +NaBr, Na+Nai, K+KCl, K+KBr, and K+KI The saturated vapour pressures of the sodium and potassium solutions in their molten chlorides, bromides, and iodides were mekured as a function of

I 07

AND

L.

A. TSIOYKINA

temperature and metal concentration. An original apparatus was constructed[UJ. All parts in contact with mixtures of the alkali metals and salts under study were made of stainless-steel resistant to their action at high temperature. The pressure developed by the saturated vapous in the hermetically sealed inner part of the device was 6xed to within f 250 Pa with a bellows manometei which had an automatic regulator for balancing the pressure of argon in the outer, also hermetically sealed part of the apparatus. The experimental installation was tested by measuring the pressure of metallic potassium and c&urn saturated vapours over wide temperature ranges. The measured values were in good agreement with the most reliable reference data. Metallic sodium and potassium contained no more than Ct.010A contamination. Particular attention was paid to preparing anhydrous salts. Sodium and potassium bromides and iodides of grade ‘Particularly Pure” were recrystallized and dried under vacuum, the temperature being raised slowly up to 450°C. Heating was continued until water condensation in the liquid-nitrogen trap ceased. The alkalinity of salts prepared in such a way did not exceed 0.005 to 0.01 mol %. The mixtures of the metals and salts under investigation were composed directly in the apparatus by

I 08

I 09 I/T

x IO3

Fig. 1. Saturatal npour pressureof liquid mixtureaNa +NaC1 containingp0.00(l), 0.21(2), 0.30(3), 0.51(4), 0.52(5), 0.66(6). O.B4(7),O.%(S), 2.42(9). 4.06(10), and 100 mol % Na( 11).

Thermodynamic properties of Na and K dissolved in their molten chlorides, bromides and iodides distilling sodium or potassium into the wntainers filled with their halides. The real alkali metal amount in the solution was found alkalimetrically at the close of an experiment, the metal having heen distilled off the salt into the cooled collector to be unsoldered. This enabled the metal concentration to be fived with good accuracy, with the metal-vapour content in the gas space of the device being taken into account (it made 4-10 per cent of the metal amount charged]. The accumulated relative error in determining the metal concentration did not exceed + 3 %. Temperature and pressure were measured to a precision of f I” and zk2%The observed data are shown in Figs 1-6, with the decadic logarithms of the pressure as ordinates and the inverse Kelvin temperatures as abscissas. On lowering the temperature, one observes the bends on the polytherms, corresponding to the formation of the second liquid or solid phase (the saturated solution of the salt in the molten metal or the metal in the solid salt). Their coordinates relative to temperature and metal concentration are in close agreement with the

1277

phase diagrams, especially in the case of forming solutions of salts in liquid metals. The alkali metal partial pressures were computed as the differences between the measured saturated vapour pressures of the solutions under study and the partial salt nressures. The latter were estimated bv the Ra0;ltk law: pull = (I- N,) P& where N, Smole fraction of the alkali metal, the most reliable reference data being taken for the kturated vapour pressures PE,, of NaCl, NaBr, NaI, KCl, KBr, and KI[ 181, The partial pressures of the alkali halides being much lower than those of sodium and potassium in the temperature ranges investigated, the feasible mistakes brought about by such approximation are within experimental error. The temperature dependence of the sodium and potassium partial pressures in the saturated vapours of all the solutions studied can be expressed by the empirical equations: lnp-=A-;.

Fis_ 2. Saturated vapour ~IXSSIUC of liquid mixturee Na + NaBr containingzOrno( 0.22(z), 0.47(3), 0.49(4). 0.66(5), 0.76(6). 0.90(7), 1.25(S), 1.97(9), 2.22(10), 253(11), 3.20(12), 3.8103). 10.4(14), 20_4(15),and 100 mol % Na( 16).

1278

M. V. SMIRNOV, V. V. CHEBYKIN AND

I

I

08

I

L. A. Tsrovrtrhtr

I 10

09

I/T

x IO3

Fig. 3. Saturatedvapeur pressureof liquid mixturesNa + Nal containingz 0.00(l), 0.30(2),tI37(3), O-49(4), 0.64(5),0.95(6),1X(7), 2.64(B),A17(9), 5.17(10),8.28(11),838(12), 14.2(13),and 100mol % Na(14).

The values of A and B calculated from the experimental data are tabulated, the pressures being given in pascals. The standard deviations S and the percentage confidence interval d on the collfidence kvel of 0.95 are also represented in Tables 1 and 2. 2.2. The vaporization heat of sodium and potassium from the solutions Na + NaCI, Na + NaBr, No + Nal,

K+KCI,

K+KBr,ad

K+KI

The temperature dependence of the partial pressures in the saturated vapours having been determined, one can find the vaporimtion heat of sodium and potassium from their solutions in the molten chlorides, bromides, and iodides. The alkali metal vapours include monatomic and diatomic molecules and do not obey the ideal gas law strictly at rather low pressures. The volume occupied by one gram-atom of an alkali metal in vaporous state is equal to

where 01is the dissociation degree of the metal dimers and B, is the second virial coefficient. Hence, the vaporization heat turns out to be equal to: A%,

= 1.1$1~(*+pr)(RT+B,p,3,

Jmol-t.

Here I3 is the temperature coefficient in the empirical equation for the temperature dependence of the partial pressure prna (in Pa). The vaporization heat was computed as a function of temperature and concentration for all the solutions under study, with the values of a and B, evaluated in the SL Units being taken from the book[19]. It grows as temperature rises and the halides anions radii increase. Its dependence on the metal concentration is more complicated (see Fig 7). The sodium vaporization heat first decmases and thereupon increases as the metal concentration is raised. The minimum on the isotherms of A,H a is deeper the lower the temperature and the A are the halide anion radii. The potassium vaporixation heat grows steadily as the metal concentration is incmased. Such b&&our of

Thermodynamic propcrtics of Na and K dissolved in their molten chlorides, bromides and iodides

08

09 I/T

i279

IO x

IO3

Fig, 4. Saturated vapour pressure of liquid mixtures K + KC1 containing: 0.00(I), 0.14(2), 0.25(3), 0.54(4), 0.85(S), 1.42(6), 3.80(7), 11.6(S). and 100 mol % K(9). the isotherms gjves evidence of greater reinforcement for the metal particle bond in vapours in comparison with solutions, the temperature being reduced. 2.3. The thernwdynumk fmtiom of sodium and porassium dissolued in their molten chlorides, bromides, and iodides The state of a pure alkali metal being designated as its standard, the rational relative aaivity coefbient of the alkali & in the mixtures with their molten halidm can .be degned by the relation Y-=N,f:*

f,

The symbols used here are the same as in the previous section of this paper. The activity coefIkients computed in this way agree in order of magnitude with those availabk in the literature, which were obtained by meaauring the redox potentials[ZO-22J and by saturation from the vapour phase[13,15] for the solutions Na+NaCl and K + KC& or derivation by approximate methods from the phase diagrams[10,11]: for all studied solutions they are well above unity and increase as the metal concentration falls (see Fig. 8). The temperature dependence closely follows the empirical equations lny,=

where N, is the mole fraction of the metal in the solutkx~,~~ and f r arefug&ties of the monoatomic metal in the saturated vapcnu-s of solutions and pure liquid metals correspondingly at the same temperature. The metal monomer fugacity is related to the metal vapour pressure by the equation

lny,-

A+:

inwhichthevaluesofAandBcanbecalculatedfrom the experimental data We have derived the inhktion formulae for tbe m&ion of the activity co&ficients to thermodyDnmic temperature and the mole fraction of the alLnli metals in the solutions. 40

1-t ho + btP’W,Il

-

+dT

Nme,)+ (b,, +b,,T)(N,/l

-N,)'

.

M. V. SMIRNOV,V. V. CHEWIUN

1280

I

I

AND

L. A. TSIO~KINA

I

I

09

08

I/T

IO

x Id

Fig. 5. Saturated vapour pressure of liquid mixtures K + KBr containing: 0.00(l), 0.26(2), 0.34(3), O-53(4), 0.74(s), 0.78(6), 0.89(7), 1.16(E), 1.68(Q), 1.75(10), 5.75(11), 9.36(12), 35.1(13),and 100 mol % (K(14).

The constants a,, cr. b,,, b, I, b,,, and b,, computed by the least-squares method from the experimental results are given in Table 3. Temperature and cmncentration ranges, standard deviation S, number of variants n, and the accumulated percentage error d estimated on the con6dcndi level of 0.95 are also quoted therein. The alkali metal solutions in their molten halides display considerable positive deviations from the ideal mixtures. The deviations grow as sodium is substituted for potassium and their iodides are replaced by bromides or chlcwides. In the respect of the concentration dependence these solutions are characteristic of the systems where the particle interaction in the individual solutes is stronger than that in the mixtures. The activity coefbcient of the components in the solutions with stronger particle interaction in comparkon with the individual solutes, on the contrary, are less than unity. They dwrcase and reactr their mi-

nimum values as the coneantration is reduced to zeroThe activity co&Gent is connected with the partial excess Gibbs energy by the simple relation Ai? The

temperature

= RTlny

= 2.303RTlny.

dependence

of the

sodium

and

potassium activity coefficients being expressed by the equations In y = A + B/T within the accuracy of observation, their partial excess enthalpies and entropies can be calculated by the formulae Awz,

= 19.144 B, J

mol-

’ and AT&, = 19.144 A,

JK-lmol-’

The computed values are given in Tables 4 and 5,the standard deviation S and the perantagc error a estimated from experimental reauhs on the confidence level of 0.95 being also represented. Taking account the accuracy of the experimental data reported by other authors, of which we made use in our computations, the percentage error rises half or twice as much as shown in the Tables. The partial excess enthalpies and entropies of sodium and potassium dissolved in their chlorides, broe and iodides have positive values, being constant within the experimental error over the investigated temperature ranges; they diminish as the metal concentration is raised. Replacing the heavier alkali halides by the lighter ones results in increasing AR% and AX%. Substitution ofsodium solutions for

Thermodynamic properties of Na and K dissolved in their molten chLork&s,bromides and iodides

I

I

08

I

I

09 l/T

1281

IO

x

td

Fig. 6. Saturated vapcnu pressure of liquid mixtures K + KI containing 0.00(l). 0.29(2), O&(3) 0.53(4) 0.71(5), 0.95(6), 1-38(7X1.5W). 2.03(9), 2.52(10),3.54(1l), 5.60(12), 6.63(13x20.8(14),and 100 mol i K(15)1

70,cm

.

T-. E h-l

B 2l

40,GQo

Fig. 7. Vaporization heat of one sodium (a) and potassium (b) ggtm from the dutions in th& molten chlorides [A), and iodidcs [ O] at 800 (open symbols) and 1ooo”C (closed symbols).

[o],

bromides

1282

M.

Table

V. SMIINOV,

V.

V. CHEBYKIN

AND

L. A. TSIOVKINA

1. The partial plersure (in Pa) dF$odiuti in the ratairkt&vapours liquid solutions Na + N&l. Na + NaBr. and Na + NaI

of the

Temperature NN~. lb

mnge (Kl

A

B

SA

Si3

130 54 28

Na + NaCl 1075-1307 108*1294 1077-1323 10761289 1074-1313 107&1277 1076-1317 loB9-1204 1144-1243

5.775 2309 5.819 2218 6.105 2170 6.250 2247 6.225 2133 2144 6.365 6.43 1 2248 6.X35 2319 7.039 2419 Na + NaBr

0.110 0.w7 0.024 0.023 0.029 0.018 0.032 0.007 o.w.8

0.22 0.47 0.49 0.66 0.76 0.90 I .25 I .97 2.22 2.53 3.20 3.81 10.4 20.4

l-1329 1027-l 329 1010-1274 1032-1328 1020-1316 106B-1281 1014=1319 101 l-1299 1009-1297 101 l-1263 1039-1291 1061-1281 1183-1292 1268-1352

5.723 6.074 5.954 6.153 6.089 6.344 6.597 6.715 6.856 6.865 7.019 7.115 7.813 8.424

0.032 0.027 0.013 0.018 0.026 0.014 0.016 0.016 0.017 0.016 0.017 0.027 0.013 0.03 I

0.30 0.37 0.50 0.64 0.95 1.22 2.64 4.17 5.17 8.28 8.38 14.2

950-1253 968-1272 935-1238 955-1262 1025-1185 945-1232 970-l 286 1034-1268 l-1291 1095-1292 1127-1296 117&1307

6.019 6.016 6.361 6.503 6.517 6.775 7.070 7.367 7.433 7.734 7.715 8.125

0.21 0.30 0.51 0.52 0.66 0.84 E 4:06

2416 2401 2120 2303 2222 2401 2556 2525 2599 2558 2675 2720 3238 3875 Na + NaI 2679 2532 2785 2802 2655 2834 2891 3074 3057 3279 3233 3589

0.069 0.054 0.016 0.045 0.018 0.022 0.026 0.023 0.014 0.032 0.033 0.036

27 34 21 37 5;

E :‘: 30 16 19 18 19 18 20 31 16 41 75 59 17 49 f: ;5 16 2 44

0.0195 0.0088 0.0051 0.0048 0.0061 0.0033 ROWS 0.0007 0.0036

4.4 1.6 0.85 0.88 E9 1.0 0.12 0.72

0.0065 0.0079 0.0036 0.0047 0.0072 0.0017 0.0042 0.0050 0.0022 0.0045 0.0029 0.0051 0.0010 0.0023

1.1 1.1 0.51 0.74 1.1 0.55 0.56 0.65 0.40 0.66 0.46 0.93 0.21 0.49

0.0163 0.0152 0.0050 0.0143 0.0018 0.0063

3.6 2.4 0.79 2.1 3.1 1.1

o”E 0.0030 0.0046 0.0044 0.0035

::: 0.47 1.0 0.8 0.69

[b) 20

b

Fig. 8. Activity

coc&-icnts of sodium (a) and potassium (b) dissolved in their molten chlorides [o], bromides [A]+ and iodidcs [n-Jat900°c.

1283

Thermodynamic properties of Na and K dissolved in their molten chlorides, bromides and iodides Table 2. The partial pressure(in Pa) of pota&un in tbc &urptcd vapours of the liquid solutions K + KCJI,K + KBr, and K + KI N,.

IO’

Temperature B=w (K)

A

SA

B

SW

K+KCI 0.14 0.25 0.54 0.85 1.42 3.80 11.6

1041-1234 1054-1235 106-1266 1053-1325 10X-1317 1058-1297 1034-1289

5.431 5.702 6.038 6.175 6.687 7.402

0.26 0.33 0.53 0.74

10141374 1090-1287 1014-1353 1020-1374 1009-1351 1015-1283 1008-1369 1003-1372 1018-1300 1009-1367 983-1346 10X&1296

5.630 5.402 5.621 5.803 5.713 5.770 5.961 6.209 6.132 6.930 7.194 a.273

956-1132 97s1233 9661207 %8-1200 974-1325 957-1325 99-1251 lOlcL1316 947-1267 959-1340 98&l 322 951-1270 973-1236

5.202 5.422 5.701 5.879 5.981 6.122 6.218 6.314 6.567 6.712 6.972 7.277 7.942

1913 1639 1654 1826 1748 2011 2552

0.026 0.023 0.033 0.018 0.030 0.023 0.007

29 26 37 20 35 27 8

O*WS4

0.0034 0.0064 0.W4-4 0.0073 osnM5 0.0019

0.76 0.88 1.0 0.7 1.7 0.71 0.23

1899 1866 1832 1853 1950 1988 1970 2345 2490 3453

0.041 0.047 0.034 0.020 0.026 0.020 0.023 0.019 0.020 0.009 0.010 0.027

48 56 40

0.0158 0.0064 0.0097 0.0072 0.0078 aW43 0.0082 0.0072 O.WSl 0.0031 0.0039 0.0043

2.0 2.1 1.6 0.89 1.3 0.97 1.2 0.89 0.92 0.43 0.50 0.79

K+KI 1780 1809 2029 2090 2077 2065 2122 2119 2313 2355 2466 2641 3129

0.032 0.073 0.046 0.058 0.030 0.025 0.w 0.030 0.013 0.015 0.013 0.030 0.013

0.0123 0.0160 0.0249 0.0348 0.0219 0.0176 0.0223 0.0156 O.W86 0.0110

0.92 1.3 1.4

5.438

K+KBr

0.78 0.89

1.16 1.68 1.75 5.75 9.36 35.1 0.29 0.44 0.53 0.71 0.9s 1.38 1.58 2.03 2.52 3.54 5.60 8.63

20.8

2169 1836

the potassium ones is accompanied by the increase of the enthalpy and the decrease of the entropy.

The redox potential of an alkali metal solution in its molten h&de can bc defined as the relation

where EL+,,,,. is a standard redox potential, aM+ and aM arc the activities assigned to the oxidation and reduction states of the alkali metal in the melts. If these solutions are eonsidcred as systems of alkali cations, halide anions and electrons. and if the alkali cation activity is suppuacd ix@ to unity over the whole

= Et.,,.-yIna,-

because on this assumption aMM’ I aM+ -a,- and aMx

0.0207 0.0076

::: 1.3 1.8 1.4 0.70 0.83 0.64 1.5 0.57

E r&xx = -l.984.10-4Tlny,.N,.,

Y.

Substituting the expressions found for the activity coefficients, we have obtained the formulas for the relation of the rcdox potentials to tcmpcrature and the mole fraction of the metals in the solutions. When NM&Q lo-‘, one can simplify to the equations: E r&ox =

l.63.10-CT-0.575 1 + 6.1 N,,

-l.984.10-4TInN,,

+O.OlOV in Na+NaCI; E

concentration range from the dilute solutions up to the liquid metals, the above equation is simplified to E redox- E~-,,.-~lnoMO

z: 50 62 34 27 51 34 14 17 14 33 IS

= a&f* . a, - = ax -. The alkali metal electrode hcing taken as a reference, E ‘&+,M- = 0 and

2.4. The redox potentials of the sodium and potassium

solutions in their molten chlorides, bromides, and i&ides

z 23 26 22 23 11 11 32

rcdor= 1.92. 1O-4 T-OS47

1 + 4.ON,

- 1.984. 10e4TlnN,,

f 0.009 V in Na + NaBr; E

~~= _ 1.50.10-4T-0.489 1 + 5.1 N,

- 1.984. 10-4TlnNN.

f 0.009 V in Na + NaI;

1284

SMIKNOV.V. V. CHEBYKIN AND L. A. TSIO~~INII

ht. V. Table

3. The activity coclikknts of sodium and potassium in the solutions Na +NaBr, Na + NaI, K + KCL, K + KBr, and K + KI

Value

Na+NaCl

o-o.05 106&l 330 0.823 2899 6.06 0 0 0 0.065 61 9.5

Solution K+KCl Na + NaI

Na+NaBr o-O.25 10131350 0.967 2757 7.85 0.00330 3.02 0.00119 0.045 137 8.6

K+KBr

O-O.15 10241330 0.883 2453 5.48 - 0.00198 4.15 -0.01019 0.040 67 8.6

(to.15 929-1310 0.755 2465 6.99 0.00166 8.45 0 0.031 96 8.4

Na+

K+KI w-25 954-1340 0.979 2126 7.62 0.00323 7.50 0.00298 0.030 109 8.1

o-o.40 1007-1380 1.043 2248 5.57 OXKJ176 0 0 0.058 106 8.7

Table 4. The partial excess enthalpy and entropy of sodium in the solutions NafNaBr. and Na+NaI (in Jmol-’ and JK-lmol-‘) Na + NaCl

Na + NaBr

N N.. lot

AR%

ATE.

0.21 0.30 0.51 0.52 0.66 0.84 0.96 2.42 4.06

53.1 54.7 54.6 52.9 54.1 53.6 51.6 47.2 44.6

16.4 18.4 16.7 13.9 15.7 14.9 14.7 12.7 13.0

s

1.0

1.1

6, %

0.9

1.0

Table

N,.

102 0.22 0.47 0.49 0.66 0.76 0.90 1.25 1.97 2.22 2.53 3.20 3.81 10.4 20.4

AR;I,

NaCI,

Na + NaCl,

Na + NaI

A3g;.

N Ns. 10’

51.0 51.1 56.1 52.4 54.0 50.3 47.0 46.6 44.8 45.0 42.8 41.5 30.8 19.8

17.9 17.2 19.6 18.2 20.6 17.0 14.7 15.6 13.7 14.2 13.3 12.6 7.8 3.1

s

1.4

1.1

6. %

1.0

0.8

0.30 0.37 0.49 0.64 0.95 1.22 2.64 4.17 5.17 8.28 8.38 14.2

s 6. %

AmE_

ATE*

47.0 49.4 44.7 43.9 46.1 42.5 40.0 36.0 35.7 31.2 31.7 25.1

15.4 16.9 12.8 121 14.7 11.8 11.7 9.6 9.8 8.0 8.2 5.3

1.4

1.3

1.0

1.0

5. The partial excess cnthalpy and entropy of potassium in the solutions K + KCl, K +KBr,andK+KI(inJmol-‘andJK-‘mol-’) K+KCI

K+KBr

N,.lO’

ARE

A3g

N,.

10’

0.14 0.25 0.54 0.85 1.42 3.80 11.6

41.8 46.8 46.1 42.5 43.6 37.9 27.8

11.9 16.6 17.6 14.8 16.2 14.2 10.4

0.26 034 0.53 0.74 0.78 0.89 1.16 1.48 1.75 5.75 9.36 35.1

K+KI

A@

d3~

N,.

36.5 42.7 41.5 41.5 42.4 42.7 40.5 39.4 40.1 32.1 29.4 12.3

13.0 19.4 19.0 17.8 20.2 20.8 19.1 17.2 19.3 13.5 12.8 4.7

0.29 0.44 0.53 0.71 0.95 1.38 1.58 2.03 252 3.54 5.60 8.63 20_8

S

S

0.9

1.3

69 %

8. %

0.7

0.9

10’

S 8, %

AI$

A3=ux

45.4 44.1 40.1 38.7 38.5 38.7 37.6 37.2 z< 30:4 27.4 18.5

23.5 22.1 18.4 17.3 17.3 17.8 17.1 17.0 14.6 14.2 12.9 11.2 6.6

1.1

0.9

0.8

0.7

Thermodynamic properties of Na and K dissolved in their molten chlorides, bromides and iodide4

E redoX=

1.75*10-4T-o.487 1 + 7.8 N,

-1.984.10-+TlnN,

*O.OWVinK+KCI; E

_

= 2.07.10-4T-o.446

1 + 3.4N,

- 1.984.10-*Tin

N,

+0.009 V in K+KBr; E redox=

1.94.10-“T-O.422 1 + 4.ON,

- 1.984. 10-4Tln

N,

&OX@9 V in K+KI.

The limiting accuracy indicated in these equations was estimated on the confidence level ot’O.95, with the error of our measurements and computations, and also with the data of other authors cited being taken into account. There are only few papers concerning the immediate determination of redox potentials for the solutions Na + NaClC20,22] and K + KCl[Zl]. The values computed by our equations are 0.10 to 0.12V more electronegative than those measured at lower temperatures. They approach each other as the temperature is raised for solutions of Na + NaCl. Better coincidence is observed for the solutions K+ KC1 where the disagreement at lower temperature does not exceed mainly with d-06 V. These discrepancies are co~ected mistakes in determining the metal concentration under the conditions of measuring the redox potentials_ The metal concentration being the:same, the redox potentials become more electropositive as temperature is raised, the lighter halide anions are replaced by the heavier ones, and potassium solutions are substituted for sodium ones. For the dilute solutions the prelogarithmic coefficient in the isotherm equations is approximately equal to the magnitude 2_303RT/F at a given temperature. When N,. > 0.01, the isotherms deviate from the linear run towards less electronegative values. The alkali metal electrodes strictly reversible towards molten halides cannot be realized in practice

owing to the mutual solubility of the metalsand salts. In this respect, halogen e&trodes are more suitable. Halogens dissolve in ldidc melts !iu less in cornparison with alkali metals. The molten salts have no Muence upon the activity of the gaseous halogen on the carbonaceous bases of the electrodes. Therefoq the halogen electrodes are used widely in potentiometric measurements. The gaseous halogen electrodes being chosen for in the redox potential equations reference, E$+,,. become equal to the decomposition voltages taken with reversed sign for the comspondllg molten halides, which can be calculated from the ref&ence thermodynamic data for alkali chlorides[23], bromides, and iodides[24], the alkali metal being in liquid state. VNacI = 3.896-5.94. 1W4T, V N~Br=3.583-66.04.10-4T,V V N., = 3.008 - 5.63. 10-4T, VK,=4.189-6.66.10-4T,V v KBt = 3.934 - 6.74. 10-4T, V,, = 3.422 - 5.80.10-+T,

V V V V

(1073- 14OOK) (1023-1323K) ( 973 - 1323 K) (1O43-14OOK) (1023 - 1323 K) ( 973 - 1323 K).

Sometimes it is of interest to know the alkali metal concentration in the halide melts at certain redox potentials. This can be done by using the relation reported in this paper. As an example in Table 6 the mole fractions of sodium and potassium in the solutions Na + NaCI, Na + NaBr, Na + NaI. K + KCl, K + KBr, and K + KJ are given for 800,900, and 1000°C at the quoted rcdox potentials. The decomposition voltages of the corresponding alkali halides are also adduced here. 2.5. The formation of the solutions Na + NaCl, K + KCI, and Na + NaBr on the cathode by electrolysing molten NaCl, KC& and NuBr Qne can obtain alkali metal solutions in their molten halides not only by dissolving the metals in the salts, but also on the cathode by electrolysing the latter. The electrode process consists essentially in the transition

Table 6. The sodium and potassium mole fractions in their molten halida at certain redox potentials Solution

-Efi+,,o

T,K

EraI, v

-1.8

-2.0

- 2.2

- 2.4

- 2.6

- 2.8

Na + NaCl

3.259 3.199 3.140

1073 1173 1273

1.9.10-9 2;2.lo-8 1.7.10-’

1.6.10-’ 1.6.10- 7 1.1.10-6

1.4.10-7 1.1.10-6 6.7.10-’

l.2.1o-e 8.3.1W6 4.1.10-5

1.1.10-5 6,0.11I-~ 26.10-‘1

9.3.10- 5 4.4.10-a 1.6.10-’

Na + NaBr

2935 2.875 2.814

1073 1173 1273

1.2.10-’ 1.0.10-6 6.1.1O-6

1.0.10-6 7.2.10-6 3.8.10-’

8.8.10-6 5.2.lO-5 23.10-’

7.7.10- 5 3.8.10-’ 1.5.10-a

6.7.10-4 28.10-3 9.9.10-5

6.4.10- 3 27.10-’

Na+NaI

2.404 2.348 2.291

1073 1173 1273

4.210-’ 20.10-4 7.5.10-4

3.7.10-4 1.5.10-3 4.9.10-3

3.4.10-S 1.3.10-1 4.7.10-Z

6.&10-a _ _

K+KCL

3.474 3.408 3.341

1073 1173 1273

5.4.10-‘” 7.6.10-’ 7.1.1o-8

4.7.10-9 5.5.lo-8 4.4.10-’

4.1.10-* d_O_lO-’ 2.7.lO-6

3.6.10- ’ 2.9.10-’ 1.7.10-’

3.1.10-6 2.1_10-’ l.o_lo-c

2.7.lO-5 1.5.10~a 6.4_10-’

K+KBr

3.211 3.143 3.076

1073 1173 1273

2.1.10-8 2.3.10-’ 1.7.10-4

1.8.10-’ l.6.10-6 1.0.10-’

1.6.10-6 1.210-S 6.4.10- 5

l.d.lO- s 8.6.10- 5 d.CklO-*

1.2.10-* 6.2.10-+ 2.5.10- ’

1.1.10- 3 W.lO- 3 1.7.10-2

K+KI

;:z

1073 1173 1273

t$X’”

1.7.10- 5 9.5.10- 5 4_O_lo-~

1.5.10-” 6.9.10-’ 2.5.10-3

l.3.1o-3 5.2.10-” 1.7.10-Z

1.3.10-’ 4.9.10- 2 1.7.10-’

2.684

5 6.X10- ’

1285

M. V. Shua~ov, V. V.

1286

CHEBYKIN

of electrons from the cathode into the molten salt electrolyte, where they behave just like those introduced when the metal is added. To see how the electrons pass from the cathode surface into the bulk of the electrolyte,we have studied the cathodic polarixation of the inditferent electrode by electrolysing molten sodium and potassium chlorides, and sodium bromide. The experiments were carried out with the apparatus shown in Fig. 9. The investigated solutions were obtained in the cylindrical steel crucible the inner diameter of which was 36 mm. The cathode was a 6 mm dia. thin armco-iron disk. This was attached to the molybdenum rod serving as a current lead. The cathode potential was measured against the halogen (chlorine or bromine) electrode which was placed in the test-tube made of sintered beryllium oxide. Its porous wall served as a diaphragm separating the electrolyte of the reference electrode from the solution in question. Temperature was measured with the Rt-pt/Rh thermocouple enclosed in a stainless-steel case which was immersed in the melt. The cell was accomodated in the hermetically closed stainless-steel container setting the differential pressure up to 6 atm. It was heated in a crucible oven with a thermostat which could be maintained constant at given values within f 3”. The measurements with NaCl were made

P

12

HP,

H2O

Fig. 9. The electrolytic cell.

AND

L. A. Tsmvimh

at 830 and 9OO”C,and with NaBr and KC1 at 800 and 900°C. The salts were of high purity. To remove possible traces of oxidizing agents and to fix initial alkali metal concentration, the molten salts in the device were preelectrolysed, the inner wall of the crucible being a cathode and the reference halogen electrode serving as an anode. By preliminary tests the stationary potentials were shown to be reached in 4-6 s after switching on the electrolysis current. The cathodic current density being raised, the time interval shortened. In all our experiments the cathode was polarized with a constant current density during 12 s. Its potential was registered on the film of the rotating mirror osclllograph to an accuracy of f 10 mV at the instant of interrupting the polarizing current, which was made by a special relay for no longer than l-5 ms. A high-resistance cathodic repeater was inserted beiween the celI and oscillograph to eliminate the depolarization of the cathode owing to the current drained from the cell by the instrument (it did not exceed 10e6 A). The cathodic current density was increase d from 1 to lo5 Am-’ stepwise at intervals of6s. The volume of electrolyte being embraced by the large-anode surface, the anodic oxidation of the a&ah metal balanced its generation on the cathode. Indeed, in all experiments the cathode potential was restored rather quickly to its initial value after the polarizing current had been interrupted. That fact can also testify to the perfect stability of the medium inside our device. The results of measuring the polarization are represented in Fig. 10 as the curves In i-4. The cathodic current density being raised, the cathode potential becomes more electronegative as a result of increasing the alkali metal concentration in the cathode layer of the electrolyte. Naturally, the more is its initial concentration in the pre-electrolysed melt, the higher is the current density at which the cathodic potential begins to shift towards the less electropositive values. In the range of current density from 3.10’ to 104 Ame it is proportional to the logarithm of current density, the prelogarithmic coefficient being approximately equal to the magnitude of 2.303RT/F at a given temperature. Above N 104 Am-* the potential approaches values that are considerably less than the decomposition voltage of the molten salt. It will be noted that at high current densities [ (2-7). 10’ Am- *] the cathode potential became unstable. The time of polarizing being prolonged to 20 or 30 s, it grew more electropositive probably owing to reducing the real current density as a result of extending the electrode-reaction front. The cathodic and anodic polarization of the iron electrode in molten sodium chloride containing dissolved metal was found[25] to obey the diffusion control. It is a safe assumption that such should be the case, too, for the solution K + KC1 and Na + NaBr. That enabled the relation of the steady metal-concentration difference between the cathode layer (CL) and the bulk (CO,) of the electrolyte to the current density (i) to be obtained from the equations for the redox potential as a function of the metal concentration, the latter being expressed in mol m- a. The results of the calculations are represented in Fig. 11 in logarithmic scale. On the intervals from N 300 to ltiArnmZ the

Thermodynamic

propertiesof Na

nnd K disaoked

in their mdtw

chlorides, bromides and iodidcs

1287

(b)

Fig. 10. Cathodic polarizdion of arma~-iron electrode in the melts: (a) Na + NaCl containing Na (mol YJ 2.3.1O-4 (A) and 1.3.10~s(o)at830”C,and22.l0~4(A)and29.10~*(o)at900”C;(b)K+KClcontainingK(mol~~1.3.10~‘(A)~d1.~1O~3 (o)at 8Q@‘C,aad 1.4.10-6 (A)and 5.4.10-’ (o)at 900°C; (c)Na + NaBrcontaining Na (mol %) 5.0.10-’ (A)and 7.6.10-” (0) at 8CUPC, and 1_2.10-’ (A) and 2.1.10-’ (0) at 900°C.

Fig, 11.Relation of current densityof difference between steady-statemetal concentrationsin the cathode layer and the balk of the elcctrolytcfor the melts: (a) Na + N&l contrining Na (mol”/ 23.10-’ I- -) and 1.3.1O-z (p) at 830”C, and 2.2.10-’ (-.-) & 2.9.10-4 (---) at WC; (b) K + KCI containing K (mol y> 1.3.10-’ (- -) and 1.0.10e3(-) at 8WC, aad 1.4.10-” (--) and 5.4.10-’ (----) at WC; (c) Na + NaBr containing Na (molyJ S.O.lO-” (- -) and 7.6.10-s (-), at 8MPC, and 1.2.10m4(- . -) and 2.1.10-3 (----) at 900°C.

concentration is linear in the current density. Above ff lo4 Am-‘, it approaches the constant values which are far from saturated. Thus, the steady-state corxentration reaches only 0.6 mol % Na in NaCl, 2.1 mol % K in KCI, and 1.1 mol% Na in NaBr at lo5 Am-’ and 900°C. It follows that in the investigated range of the current density metallic sodium and potassium are not isolated on the cathode, but unsaturated solutions in the electrolyte are formed. Under the conditions of the onodimcnaional steady-state diffusion (which were observed on the cathode in our experiments) the liuear sections of the curves lo i-ln (C’, - CO,) can be described by the equation

C&=C&+&-i M

where F is Faraday constant, D, is the el%cti~e value of the alkali metal c!o&icicnt diffusion in the molten salts, and b is the thickness of diffusion layer on the cathode. Here the ratio 6/D, is constant. The thickness of the diffusion layer in the solutions Na + NaCl was shown to be equal to 2.10s4 m[26]. Such being the case for the solutions Na+NaBr and K + KCl, the effective diffusion coefficients of sodium and potassium prove to be much greater than those of the cations Na+ and K+ in the same molten salts. For example, at 900°C the diffusion coefficients for sodium in Nail and NaBr, and for potassium in KCl are equal to 1.8.10-‘,S.O.lO-‘and 1.3.10-7m2s-1accordingly, whereas those of the cations Na+ and K+ in NaCl and KC1 are equal to 1.15.10-a and 9.41.10e9 mz s-l. The higher mobility of metallic sodium and potassium in their molten halides in comparison with the

M. V.

1288 cations

may be connected

SMIBNDV.

v.

v.

with electron transitions

CH3WIUN

of

the type e-

Na” ?Na+

eNa+-

e-

+

Na”,

the electrons being localized not only on one particle, but also among several cations. The alkali metal concentration in the electrolyte being raised, the electronic+onductivity contribution grows quickly. Our observations suggest thaat it may become predominant when the concentration reaches the values at which the current increases without changing the cathode potential. Here the electrode reaction comes off the cathode surface and proceeds in the bulk of the electrolyte where the concentrated solutions tiith prevalent electronic conductivity make contact with more dilute ones the conduction of which is still predominantly electrolytic. REFERENCES 1. M. V. Smirnov, Elakrrodnye powntsfaly v raspluulennykh khloridakh, Nauka, Moskva (1973). 2. M. V. Smimov. V. N. Chebotin, V. Ja. Kudjakov and N. A. Loginov, Efektrokhtiyu 13, 754 (1977). 3. E E. Shoilrain and D. N. Kagan, _ Teplofiz. Vys. Temp. 7. 787 (l&Q). 4. H. Yokokawa, 0. J. Kleppa and N. H. Nachtrieb, J. Chem. Phys. 71.4099 (1979). (Edited by 5. M. A. B&g In Molten Salt ChemWry. Blander) p. 367. Interscience Publishers (1964). New York. 6. K. S. Pitter, J. Am. Churn. Sot. 114, 2025 (1962). 7. A. Neckel, Mow&h. chemie 96, 1617 (1965). 8. L. E. Topol, J. Phys. Chem 69, 11 (1965).

AND

L.

A.

TS~OYKINA

9. R. Vilcu and C Misdolea, Reu. Rown. Ch*n 13, 281 (I!%@; ibid 14, 1353 (1969). 10. A. G. Morachevsky and F. I. Lvovich, Zhur. priklad. K&II. 46, 2640 (1973). 11. A. Krupkowaki and K. Fitzner, Arch. Hutn. 15, 359 (1970). 12. M. A. Bredig, H. R. Bronstein and W. T. Smith, J. An. cheul. sot. 77. 1454 119551. 13. M. V. Smimov; I. Ja iyub&taeva, L. A. Tsiovkina and V. V. Chebykin, Deposited in VINITI 1973,646373 (1973). 14. J. W. Johnson &d M. A. Bred& J. Phys. Chem. 52, 604 (1958). 15. M. V. Smimov. I. Ja. Lyubimtseva, V. V. Cbebykin and L. A. Tsiovkina. Zhur. p&lad. Khim. 50, 1145 (1977). 16. E. E. Shuihain. E. E. Totskv and Ju. V. Karmvshin. Trudv * Moskov: En&et. Instir&, 75, 62 (1970). 1 17. M. V. Smirnov, V. V. Chebykin and L. A. Tsiovkina, Ju. N. Krasnov. Zhur. fiz. Khim. 51. 1848 (19771. 18. L. Topor, J: Chern- Therm&&s 4, ‘739 (1972). 19. E. E. Shpilrain, K. A. Jakimovich, E. E. Totsky, D. L. Timrot abd V. A. Fomin, TeploJizicheskie svoist~a scheiochnykh mefaliov. Moskra, Isdatelstvo standartov (1970). 20. M. V. Smirnov. L. A. Tsiovkina. I. Ja. Lvubimtseva and V. V. Chebykin, &posited in VIkITI, 19?3,6462-73 (1973). 21. L. A. Tsiovkina, 1. Ja. Lyubimtseva and V. V. Chebykin, T&y Inst. Elekrokhim. U.N.Z. AN SSSR, 25,22 (1376). 22. A Boozeny, Traw. met. Sot. AIME, 224, 950 (1962). svoistva individualnykh veschestv 23. Termodinamieheskie (Redaktor V. P. Glushko) Isdatelstvo AN SSSR (1962). 24. C. J, Janz and G. M. Sijkhuis, Molten Saks, Vol. 2, p. 1, U.S. Dep. of commerce NSRDS-NBS 26 (1969). 25. M. V. Smirnov, L. A. Tsiovkina, V. V. Chebykin, I. Ja. Lyubimtscva and Ju. N. Krasnov, Elektrokhimiyo 10, 1389 (1974). 26. M. V. Smimov, L. A. Tsiovkina. V. V. Chebykin, I. Ja. Lyubimtseva and Ju. N. Krasnov, Trudy S Vses. Knof: Fiz. Khim. Elektrokhim. Rasplavl. Soiei Tuerd. Elektrolit..(Sverdlousk 2, p. 3 (1973).