COMPLEXING OF TAC&BY CHLORIDE ION IN FUSED MEDIA
Feb. 5, 1959
Acknowledgments.-The authors wish to thank the Office of Naval Research for financial aid under contract No. N6 onr 26913. They also are indebted to Messrs. A. E. Moore, A. H. Guenther,
[CONTRIBUTION FROM E. I. DU
535
B. T. Guran and W. L. Gumby for assistance with the extensive calculations involved. UNIVERSITY
P O N T DE
PARK, PENNA.
NEMOURS & CO., INC.]
Complexing of Tantalum Pentachloride by Chloride Ion in Fused Salt Media' BY CHARLES M. COOK,JR. RECEIVED JULY 21, 1958 The interaction of TaCls with KaCl in their solutions in molten NaFeCll was studied a t 300 and 400'. The T a c h activity, as measured by its vapor pressure, was found to vary with TaC15and NaCl concentrations in a manner consistent with the equilibrium reaction TaCls f C1- F? TaCle-. The corresponding equilibrium constants were found to be log XTaCL#-/ ~ T ~ C I ~= X 2.4 C I -=t0.3 a t 300" and 1.7 i 0.3 a t 400".
The first observation of the TaCls- anion appears to be that of Gutmanq2 who found the compound [Pyr.POClz+][TaC&-] during an investigation of the Poc13 solvent system. More recently, Morozov and his colleagues have studied the phase diagrams of the NaC1-NbC16 and the NaC1-TaC15 system3b and have reported the solid compounds NaNbC16 and NaTaCls. These workers have estimated the free energies of formation of these compounds from the NbCL and TaC16 vapor pressures above solid, powdered mixtures of NaCl -t NaNbCle and NaCl NaTaC16, respectively. I n the work described below, the reaction between tantalum pentachloride and sodium chloride in a molten salt solution
+
TaC15
+ C1-
= TaCle-
(1)
has been studied by measurement of the equilibrium TaC15 activity as a function of gross TaC16 and NaCl concentrations. The salt NaFeCld, was chosen as the solvent for this system because it is low melting, essentially inert to the reactants and dissolves both TaC15 and NaC1. Experimental The vapor pressure of tantalum pentachloride above a mixture containing known quantities of NaCI, FeCl, and TaC16 was measured as a function of temperature. The salt mixture was contained in a sealed, evacuated, Pyrex reaction bulb to which was attached a sickle gauge for measurement of the internal pressure. Both bulb and gauge were positioned inside a tube furnace, which was constructed of copper t o minimize temperature gradients. The melt was stirred until equilibrium was reached, a t which time the bulb temperature and pressure were observed. Ingredients.-Sodium chloride (Mallinkrodt, Analytical Reagent) was weighed into the Pyrex reaction bulb and dried overnight a t 480" under a stream of HCl. Iron wire (J. T. Baker Analyzed Reagent) was placed in a silica glass side arm attached by a graded seal t o the Pyrex reaction bulb. The iron was burned in a stream of chlorine (Matheson Company, Oxygen Free grade), and the ferric chloride product was quantitatively sublimed into the bulb. Tantalum powder (Fansteel Metallurgical Company, 120-325 mesh) was placed in the silica side arm with the (1) Presented a t the 11th Summer Symposium of the Division of Analytical Chemistry of t h e American Chemical Society, Schenectady,
N. Y., 6/20/58. ( 2 ) V. Gutmann, Monalsh., 86, 1077 (1954). (3) (a) I. S. Morozov and B. G. Korshunov, Zhurnal neouganicheskoi Khimii, 1, 145 (1956); (b) I. S. Morozov, B. G. Korshunov and A. T. Simonich, ibid., 1 , 1646 (1956). (4) I. S.Morozov and A . T. Simonich, i b i d . , 2, 1907 (1957).
iron wire, chlorinated, and the pentachloride sublimed into the reaction bulb. Traces of iron or tantalum oxides remaining after the sublimation of the chlorides were chlorinated into the bulb by treatment with phosgene. The chlorination gases were dried with Mg( ClO&. After having been charged with the NaC1, FeCla and TaC15, the reaction bulb was evacuated, backfilled with Cl2 a t approximately 60 mm. pressure, and both side arms fused shut and removed. The bulb and sickle gauge assembly was then placed in the tube furnace and connected t o a pressure measuring manifold. Temperature and Pressure.-The temperature of the molten salt mixture was measured with two chromelalumel thermocouples held in contact with the surface of the reaction bulb by strips of glass tape. The salt melt was stirred while being held a t a constant temperature for approximately 1 hr . before a final pressure-temperature reading was taken. Temperature fluctuations within the furnace due to line voltage changes were eliminated by use of a constant voltage transformer in the power source. The use of duplicate thermocouples, the stirring of the salt melt and the elimination of sharp temperature gradients in the furnace should reduce error in the measured salt temperatures. This error has been estimated to be less than f 3 O .
Pressure measurements were made over a zero to two atmosphere range by the sickle gauge between the reaction bulb and the pressure manifold. The observed pressure was that manifold pressure required to return the gauge pointer to the null position. In the majority of the runs, the gauge sensitivity was greater than f 2 mm.; for the runs made a t high TaClS concentrations, where the pressures were higher, sturdier gauges of about A8 mm. sensitivity were used. Before data were taken, the melt was held a t 400" (or the maximum temperature allowed by the pressure limitations of the system) for several hours to bring about complete solution of the NaCl in the solvent. When this precaution was taken, the observed pressures equilibrated rapidly, showed no temperature-pressure hysteresis and underwent no drift with time. Calculation of TaC15 Pressure.-The experimental data obtained from any one run consist of a series of observations of total gas pressure inside the reaction bulb as a function of temperature. The T a c k pressure a t a given temperature is the difference between the total pressure observed a t this temperature and the pressure exerted by the Cl2 initially ~ I The Cl2 present in the bulb, ;.e., P T a C l a = P T ~- ~PcI,. pressure can be observed directly a t low temperatures where TaC15 pressure is negligible, and Pel, a t higher temperatures was calculated from the ideal gas law. Figure 1 shows the data from a typical run and illustrates this treatment. The solid line passes through the experimental points; the dotted line represents the chlorine gas pressure as calculated from a measurement a t 25'; the dashed line represents the difference between these two pressures and accordingly is taken as the TaC15 pressure. For those runs made a t low TaCls concentrations, the uncertainty in Pelt is a major source of uncertainty in the C 300", I ~ since under these condivalues calculated for P T ~at
536
CHARLES M. COOK,JR.
300c -TOTAL
listed in Table I with the corresponding values of
PRESSURE
X L s and Xkl-. These data show the tantalum
INERT GAS PRESSURE
-
1000
-_
Vol. 81
pentachloride vapor pressure to increase sharply with increasing TaC16 and with decreasing free chloride concentrations. The magnitude and the direction of these changes in TaC16 pressure with solution composition imply that TaC16 dissolves in the melt in part as a non-volatile chloride complex and in part as free TaCI5.
TdCI, PRESSURE
r
5 300 ; u
&
TABLE I T a c h PRESSURE ABOVE I\;aFeCl,-NaCl
100
/ 30
I
200
-
Fig. 1.-Total
I
Run I
I
I
I
I
I
300
1 2 3 4 6 6
/
400
Temp., 'C. pressure v5. temperature for run 6.
tions, Pa, PTsCls. The chlorine pressure correction rapidly decreases in importance with increasing temperatures and/or increasing Tact concentrations. For the above procedure for calculation of the TaClb pressure from the total pressure to be valid, the C1, pressure must follow the gas law and the FezCls pressure above the melt must be negligible. To demonstrate that the system fulfills these requirements, a T a c k f r e e mixture of initial composition X F ~ C=I ~ 0.47, X N ~ C=I 0.53 was prepared, and the pressure above this mixture was measured from room temperature to 410'. The observed pressure was found to deviate by less than 10% from the gas law from room temperature to 410'. A t the highest temperature the measured pressure was in excess of the gas law value by 10 f 5 mm. To establish that at the temperatures of measurement TaCls does not react with NaFeClc to liberate significant pressures of FezCls mixtures of the same initial composition as in the above paragraph were equilibrated at 300' and at 400' with a TaC16 vapor and argon mixture that was bubbled through them. The argon exiting from the saturated melts was found to contain TaCb that was, judging from the small amount of red coloration in the condensate, contaminated with only trace quantities of FezCls. Analysis of the NaFeCt-TaClr solutions indicated the presence of TaC4 in the melts. 5-10 mole
0.016 .037 .040 .043 f 1 .065 f 3 .067 .118 ,114 ,144 .22 .048
7
(I
X),Cl,
8 9 10 11 12 .048 13 .048 14 ,047 Extrapolated values.
XL,-
MELTSAT 300' pTsCli
0.051
14 f 8 31 f 5 .058 f 1 30 f 5 .058 40 f 6 .058 f 1 260 f 30 .050 340 f 30 .054 1430 f 200' .048 1500 f 200' .045 2050 f 300' .045 P T s C i l a equals P!aclr .071 68 f 20 .05l 43 f 7 .030 355 f 25 .010 695 f 40 = 3200 mm. a t 300'.
.OB
The data of run 10 show that TaC16 is not completely miscible with the solvent salt but that a t some TaCls concentration less than 22 mole yo the TaC16-NaFeC14 system splits into two phases, one of which is essentially pure TaCls. The solubility limit can be estimated more accurately from the data of runs 1-9 by plotting P T a C l , VS. X&ls and extrapolating to that gross tantalum pentachloride concentration a t which the vapor pressure becomes equal to that of pure TaC16. Such a plot = P&, a t X&acls 0.19; indicates that PT.Cls a t this gross tantalum pentachloride concentration Discussion most of the available chloride is in the form of The compound formed by reaction of FeCla with TaCle-, since X&- = 0.045 the free TaC16 conNaCl has been demonstrated to be NaFeClo by centration a t the solubility limit accordingly is 0.145. Dum5 who found the FezCls vapor pressure above X T a C I s The uncomplexed TaC16 present in the ionized mixtures of NaCl and FeC&to drop sharply to zero at NaCl/FeC& ratios equal to or greater than NaFeCla melt should exhibit a considerable posiunity. The similar compound NaAlCL is well es- tive deviation from Raoult's law behavior. At the solubility limit of TaC16 in NaFeC4 the activity tablished.6 In this work, the solvent salt melts have all been coefficient of the uncomplexed TaC16, 7,must equal on the NaC1-rich side and are considered to be the reciprocal of the solubility. Thus from the (0.145)-l = 6.9. The best overNaFeC14 containing dissolved NaC1. In the sub- above data y sequent presentation of the data the quantity all fit of the data a t 300' is in fact provided by the Xk,- signifies the mole fraction of the initial "free" value y = 6.5, and accordingly the concentration is taken to chloride ion in the melt, i.e., the number of mole- of free tantalum pentachloride, XTsCls, a t this temperature. cules of NaCl minus the number of molecules of be equal to PTaCls(6.5P&aC1! Figure 2 is an illustration of the behavior of inFeC&initially added to the sample bulb divided by the total number of ions, Na+, FeCld-, etc., pres- creasing amounts of tantalum pentachloride disent in the final melt. The gross mole fraction of solved in a melt containing a constant initial conTaCl5 in the solution is similarly written X k s ~ i r ; centration of available chloride. The ordinate of while the concentration of unreacted TaC16 is this figure is a function of tantalum pentachloride pressure and concentration that is equivalent to the XTIC16* The vapor pressures of tantalum pentachloride fraction of the total tantalum that is not complexed a t 300' above its solutions in NaFeC14-NaC1 are by chloride, i.e., XTscls/Ak s c l , . The abscissa represents the concentration of total TaC16 that is ( 5 ) W. E. Dunn, presented at A.C S. Del. Valley Regional Meeting, in excess of the free chloride ion. The plotted Feb. 16, 1956. points represent those runs where the XkI- 0.05. (6) E. W. Dewing, THIS JOURNAL, 77, 2G39 (1958).
-
-
N
-
1.0
1.0
-... 0.i5
;0.75
s
0
s
i:
5
{
537
COMPLEXING OF TaC& BY CHLORIDE IONIN FUSED MEDIA
Feb. 5, 1959
"., c &In 0.50
0.50
rd >
>
8 0.25
'
W
i)
U
5
0.25
-0.05
0 x*TaCI,
+0.05
-0.05
+O.lO
- X'Cl-.
-0.025 X'Cl-
Fig. 2.-Fraction of TaCls not complexed a t 300' vs. TaC16 concentration in excess of chloride. Chloride concentration in the range X'ci- = 0.045 to 0.058.
0
+0.025
- X'TaClr.
Fig. 3.-Fraction of T a c k not complexed a t 300' us. chloride concentration in excess of Tack. Tantalum concen= 0.048. tration constant a t X'T~CIS
The points are seen to fall along a neutralization None of these alternative reactions fit the experitype of curve, rising sharply when the excess NaCl mental data as well as reaction 1. has been consumed by added TaC16. The three The rapid increase in tantalum pentachloride dotted points, taken a t high X T a C l p where the pres- pressure with temperature and the limited bursting sure limitations of the apparatus did not allow strength of the apparatus restricted the study of direct measurement of the pressures, represent the TaCl6-Cl- interaction a t 400' to relatively PTsc1,values that were estimated by extrapolation. dilute TaC& solutions. As a result the TaC16 Two curves, representing the XTaC~6/X$aClr solubility limit in the salt melt could not be measbehavior to be expected if TaC16 reacts with chlo- ured a t this temperature and thus y was not acride to yield TaCle- according to equation 1, are curately determined. presented in Fig. 2 for comparison with the plotted To determine whether the complexing reaction points. The fraction of the TaCIS that is uncomTaCls C1- = TaClB(1) plexed follows the solid curve if the concentration equilibrium constant for reaction 1 is K = XTaCls-/established from the experimental data observed X T ~ CXcl~ , = 103.b,the dotted curve corresponds to a t 300' also represents the data a t 400°, the equilibK = lo3. With the exception of the three dotted rium constants K / y = XTsCl1-/yXTaClsXC1points, the experimental data are seen to fall between have been calculated for the data listed in Table 11. the two curves over a ninefold change in XkaCla, with I n the calculations of these constants the tansome tendency to favor the log K = 3 curve a t low talum pentachloride activities were determined tantalum pentachloride concentrations. The ex- from the observed pressures by the relationship ~ CPIT ~ s ~ i a / P O T a ~ i a ; the X T a C l a and Xclperimental results can be considered to be consist- ~ Y X T = concentrations were estimated using two assumed ent with equation 1 where log K = 3.2 + 0.3. I n Fig. 3 the fraction of the tantalum penta- values of the tantalum pentachloride activity cochloride that is uncomplexed is plotted, for data efficient, y = 2 and y = 5. The data listed in where XkaClr 0.05, against the available chloride Table I1 have been restricted to those runs where ion concentration in the form X&- - X$acl,. TABLE IThe amount of free TaCb is seen to drop steadily TaCl6 PRESSURE^ ABOVE XaFeClr-SaC1 MELTSAT 400' with increasing chloride ion concentration. The log K / , r log K1-l points are again in agreement, over a sevenfold range Run PTaCld ifr=2 i f y = a of Xkl-, with the behavior predicted by log K = 1 97 1.51 1.67 3.2 f 0.3, the solid line and dotted line corre2 210 1.73 1.88 sponding to the XTaCII/X:sCI, behavior consistent 3 390 1.73 1.62 with K = 103.5and K = IO3, respectively. 2.10 4 165 1.97 An attempt was made t o account for the ob5 820 1.23 1.06 served variation of XTsCla/X~acls with X$aCls and 11 250 1.72 1.87 X",- by assuming the possible complexing reac12 425 1.63 2.08 __ __ tions Av. 1.65 1.73 TaCI, 2CI- +TaCIr' (2 1
+
-
or
or
+
2TaCb --+ TaC14+ Tach
+ TaCla-
+ FeCL- +(TaClvFeC14)-
P&lS = 12,000 mxn. a t 400'.
(3) (4)
XT&Ia and Xkl- so that XT&],-and xclcan be estimated with reason-
yXTsrlr is small compared to
538
C. J. BALLHAUSEN AND ANDREW D. LIEHR
able certainty despite uncertainty in y. The values of the K/y constants, reported in Table 11, are seen to be only slightly sensitive to the choice of y, to remain constant over the XkaCls or X & - concentrations studied and to lie in the range log K / y = 1.7 i. 0.3. Thus the vapor pressures of tantalum pentachloride above its solutions in NaFeClb-NaCl mixtures appear to be consistent a t 300 and 400' with the complexing reaction 1. The association constants K1 = XTaCis-//aTaclsXr1-, are equal to log K1 = 2.4 rt 0.3 a t 300' and to log K1 = 1.7 i. 0.3 a t 400'. It is of interest to compare the free energy values calculated for the reaction NaCl(c) 4- TaCl&atm).
=
NaTaClG(c)
Vel. 81
Substitution into formula 6 of the measured values of K/y, the known values of P$scla, and of values of y X & T a C I 6 / Y N a C ] estimated by formula 7 yields the calculated values A F j = -2.6 rt 0.7 kcal./mole a t 300'; AFj = +0.5 i. 0.7 kcal./mole a t 400'. Norozov's group reports AFj = - 1.9 and $1.0 kcal ./"ole a t these temperatures. Acknowledgments.-The author wishes to thank Mr. W. M. Johnston for analytical data, the Pigments Department of E. I. du Pont de Nemours & Company for permission to release these results and Dr. J. O'M. Bockris and Dr. G. W. Watt for helpful criticism.
(5)
by Morozov and his co-workers with the free energy changes for the same reaction that can be estimated from the measured equilibrium constants for reaction 1. The free energy change of reaction 5 will be related to the equilibrium constant by the equation 4F5 = - R T ln(aKaTaCla/aKaCiPTsCls) = - R T I ~ [ ( K / ~ P ~ , c ~ ~ ) ( ~ N , T (, 6c), ~ /
where yKsTaC16 and yKaclare the activity coefficients of dissolved NaTaCIG and NaCl in molten NaFeCld, with the standard states of both dissolved salts being the pure crystal. The activity coeffiY N be ~ C ~approximated by cient ratio Y ~ ~ T ~ C ~ ~ /can
(7) I n t h e use of this relation the assumption is made t h a t solutions of KaC1 and of NaTaCls in liquid NaFeCla. with t h e standard states of the solutes taken a s t h e hypothetical supercooled liquids a t temperature T , deviate from ideality approximately t o a n equal extent. This assumption would appear t o be reasonable since i t appears t h a t many simple salt mixtures do not, in t h e absence of complexins by t h e solvent, deviate markedly from ideal behavior. Values of Lf(xaC1),the heat of fusion of XaCl,as a function of tem~perature ~ ~ c Iwere ) I calculated from d a t a in AT. B. S . Bulletin 500. T h e melting point of NaTaCl6 has been reported a s 470°.3b T h e heat of fusion of NaTaCla is not known, and, for t h e purpose of these calculations, has been estimated from Lf(Naci) and the ratio of the melting points. T h u s
-
Lf(NaTacls) Lf(xacI)[743°/10810] WILMINGTON, DELAWARE
=
4.7 kcal./inole
-
[CONTRIBUTION FROM
THE
BELLTELEPHOXE LABORATORIES, INC.]
Some Comments on the Anomalous Magnetic Behavior of Certain Ni(I1) Complexes BY
c. J. BALLHAUSEN* AND h D R E W D. LIEHR RECEIVED AUGUST4, 1958
The magnetic and spectral properties of tetra-coordinated Ni(I1) complexes are discussed in the light of the modern theory of ligand fields. It is pointed out that for such Ni(I1) complexes there exist both strong experimental evidence and convincing theoretical arguments against ( a ) the existence of dsp2 bonding, and ( b ) the existence in solutions and melts of planartetrahedral conformational equilibria. Equations governing the magnetic susceptibility of partially paramagnetic planar Xi( 11) systems are derived and their usefulness in determining the singlet-triplet energy separation is underlined. Some conjectures concerning the nature of the famous Lifschitz salts of nickel are presented and an interesting analogy between these compounds and the recently characterized copper alkanoates is noted.
Introduction The tetra-coordinated complexes of Ni(I1) have occupied a prominent position in the development of the theory of inorganic complex ions. They have been frequently cited as prime examples for the "magnetic criterion of bond type" and have hence been classified as planar or tetrahedral accordingly as they are dia- or paramagnetic. Indeed, the anomalous magnetic susceptibilities exhibited by solutions and melts of certain tetracoordinated Ni(I1) complexes have been explained as due to the establishment of a dynamic equilibrium between these two conformations. I n this note we wish to point out that the modern theory of inorganic complexes, the ligand field theory, offers an alternative interpretation of the anomalous magnetic behavior of tetra-coordinated Ni (11) complexes and that (1) Visiting Scholar 1957-1998. On leave from Chemical Laboratory A , Technical University of Denmark, Copenhagen, Denmark.
extremely useful information concerning the low lying energy levels of these complexes may be obtained from studies of the variation of the magnetic susceptibility with temperature.
Theory As is well known2 the imposition of a ligand field splits the fivefold degeneracy of the transition metal 3d electronic orbitals in a manner characteristic of the spatial symmetry of the attached ligands. In Fig. 1 is depicted this splitting for the cases here of interest. If one neglects configuration interaction, the ground electronic state of a weakly tetragonal Ni(I1) complex may be written as either (e,) 4(b2g) 2(alg) * (diamagnetic) or (e,) (2) See for example, t h e following recent review articles: (a) J. S . Griffith and L. E. Orgel, Ouort. Rev., 11, 381 (1957) (non-mathematical); W. E. Moffitt and C. J. Ballhausen, An%. Rev. Phys. Chem., 7 , 107 (1956) (mathematical).
