472
ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978
Solubility Products of Bis( 0 , 0’-diethyldithiophosphato)copper( 11) and 0 , 0 ’-d imethy Idit hiophosphat ocopper (I) Walter Rudzinski and Quintus Fernando* Department of Chemistry, University of Arizona, Tucson, Arizona
8572 1
The ammonium salt of 0 ,O’diethyldithiophosphonic acid reacts with copper(I1) in aqueous solution to give a mixture of the copper( I ) and copper( 11) complexes of 0 ,0’-diethyldithiophosphonic acid. The solubility product of the copper(I1) complex was found to be 10-15.22 in a KNOBmedium of ionic strength 0.1. The reaction of copper (11) with the ammonium salt of 0,O’-dimethyldithiophosphonic acid gave a mixture of several complexes. The dimeric copper(1) complex of 0,0’-dimethyldithiophosphonic acid was readily synthesized and an approximate value for its solubility product was obtained.
A large number of sulfur-containing compounds are employed as flotation agents and among these are the dithiophosphonic acids and their sodium, potassium, or ammonium salts (aerofloats). Unlike most flotation agents, the dithiophosphonic acids are stable in the presence of strong acids and are potentially useful in certain hydrometallurgical processes that make use of strong acids for the partial disruption of an ore matrix. A disadvantage of the dithiophosphonic acids is t h e relatively high solubility of their transition metal complexes. Of special interest to us are the values of the solubility products of a series of bis(O,O’-dialkyldithiophosphato)copper(II)compounds that were first reported by Kakovsky ( I ) and subsequently by Tulyupa ( 2 ) . Both sets of solubility products decreased as the number of carbon atoms in the alkoxy substituents on the phosphorus atoms increased. There were considerable discrepancies. however, in the magnitude of the solubility products reported by the two workers. For example, the values reported for the solubility product of bis(0,O’-diethy1dithiophosphato)copper(II), Cu(dtp),, differed by a factor of lo4. This large discrepancy raises a question concerning the identity of the copper complex that was synthesized by the two investigators. Wasson has reported that attempts to prepare (0,O’-dialkyldithiophosphato)copper(II) complexes as solids were unsuccessful probably because the dithiophosphate ligand acted as a reducing agent to form copper(1) complexes (3). The formation of bis(0,O’-dimethy1dithiophosphato)copperiII). Cu(dmp),, is of particular importance because it is t h e basis for a proposed titrimetric method for the determination of malathion by Hill and co-workers ( 4 ) . In a subsequent report, Hill presented evidence which indicated t h a t the copper(I1) complex existed in equilibrium with the corresponding copper(1) complex and the disulfide formed by t h e oxidation of t h e dithiophosphate ligand ( 5 ) . The work that is described below was carried out to resolve the widely divergent values of the solubility products that have been reported for Cu(dtp), and Cu(dmp),. In the course of this work, it was essential to determine the oxidation state of the copper in the complexes that were synthesized and to verify that the complexes had the predicted composition. EXPERIMENTAL Synthesis of NH4[(S)SP(OC,H,),], NH4(dtp). The preparation was adapted from the method of Coldberry. Fernelius, and Shamma (6). One hundred mL of ethanol was added slowly over a period of 1.5 h to 110 g of finely powdered P2Ss(Eastman 0003-2700/78/0350-0472$01 O O / O
Chemical Co.). The reaction mixture was heated under reflux for 3 h and the HPS that was evolved was passed through scrubbers containing H202and NaOH. The reaction product which consisted of a black oily liquid was filtered without suction and the filtrate was extracted with three 50-mL portions of water. Gaseous ammonia was bubbled through the extract containing the 0,0’-diethyldithiophosphonic acid. Two hundred mL of acetone was added and the mixture was concentrated to one half of its original volume. White crystals of the ammonium salt which formed on standing were filtered, dried, and recrystallized from ethyl acetate. Titration of the ammonium salt, dissolved in 90% ethanol, with NaOH showed that the purity of the salt was 99.3%. Synthesis of Cu[S(S)P(OC2H6),12,Cu(dtp), and Cu[(S)SP(OCH3),I2,Cu(dmp),. Attempts were made to prepare Cu(dtp), and Cu(dmp)2by the slow addition of NH4(dtp) or NH4(dmp) (Aldrich Chemical Co.) to an aqueous solution of copper(I1) nitrate. In each case, a yellowish brown oil was formed when the reaction was complete. The components of the oily brown liquid that was obtained in the synthesis of Cu(dtp), were separated on silica gel plates with a solvent mixture that consisted of 92% v/v toluene and 8% v/v ethyl acetate. Synthesis of Cu[(S)SP(OCH3)2], Cu(dmp). An aqueous solution of copper(I1) nitrate was added to a solution containing an excess of NH4(dmp). When a pale yellow precipitate was obtained, SO2 was passed through the solution to reduce any Cu(drnp)?that may have formed. The pale yellow precipitate was washed repeatedly with water and dried in a vacuum desiccator. (Cu: 28.8% calcd, 28.9% found; P: 14.04% calcd, 14.15% found.) Infrared Spectra. A Perkin-Elmer Infracord Spectrophotometer was used to record all infrared spectra in the range 4000-700 cm-’. A Beckman IR 12 double beam spectrophotometer was used to record the spectra in the region 700-200 cm-’. The infrared spectra of Cu(dtp), and Cu(dmp), which were obtained as oily liquids were run as Nujol mulls and as neat oils on AgCl disks. The infrared spectrum of the pale yellow compound, Cu(dmp),was run as a KBr pellet in the region 4000-200 cm-‘. Mass Spectrum of Cu(dmp). The mass spectrum of Cu(dmp) was obtained with a Hewlett-Packard Model 5930A mass spectrometer with an electron energy of 8 eV. The compound was introduced into the sample chamber which was maint,ained at 200 “ C by the direct insertion method. Photoelectron Spectroscopy. X-ray photoelectron spectra were obtained with a McPherson ESCA 36 Photoelectron spectrometer equipped with a Sargent-Welch turbomolecular pumping system (lo-‘ t o Torr). The sample was irradiated with A1 Kn x-rays (1482.6 eV). A finely powdered sample of the compound was spread on double-stick adhesive tape (3M) which was attached to an aluminum planchet. The sample holder was cooled with liquid nitrogen. The binding energies of the photoelectrons were determined by assuming that the carbon electrons from the adhesive tape had a binding energy of 285.0 f 0.24 eV. Atomic Absorption Spectrophotometry. The total copper concentrations in solutions that were in equilibrium with solid Cu(dtp), were determined with a Varian Model AA-5 atomic absorption spectrophotometer. The copper concentrations of solutions that were in equilibrium with Cu(dmp) were determined with a Heath Model 703 atomic absorption spectrophotometer. Single slot burners designed for use with an air-acetylene flame were employed. A standard copper hollow cathode lamp was used as the source and the copper resonance line at 3248.8 A was used for all the measurements. Fluorescence Measurements. The concentrations of the ligand, dmp, in solutions that were in equilibrium with the pale 32 1978 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978
Table I. Infrared Absorption Bands of Ni(dtp), and Related Compounds (cm-') Yellowish brown oil Ni(dtp),a Cu(dtp)b 1302 ... 1235 1162 1159 1160 1098 1095 1103 1033 1049 1040 1015 1008 1004 965 964 97 0 ... 955 955 ... a22 823 ... 806 a07 ... 785 786 ... 774 775 ... ... 690 643 640 630 544 538 518 ... 524 502 ... ... 482 ... 401 396 .
L -L-&--~--' 2~- -
2
4
6
8
0
1
2
1
4
-
1
m-
Figure 1.
Potentiometric titrations of copper(I1)nitrate (2.166 X
M) with NH,dtp (2.232 X lo-' M) and copper(I1)nitrate (2.211 X M) with NH,dmp (2.241 X lo-' M) with a cupric ion selective electrode
yellow solid, Cu(dmp), were determined fluorimetrically. A Perkin-Elmer Model 204.4 fluorescence spectrophotometer was used for the fluorescence measurements; the excitation wavelength was 384 nm and the fluorescence intensities were measured in a 1-cm quartz cell a t 385 nm. Potentiometric Measurements. All potentiometric determinations were carried out in a water-jacketed vessel that was maintained at 25 "C. The potentiometric measurements were made with an Orion Model 701 digital pH/mV meter with a cupric ion selective electrode (Model 94-29) and a double-junction reference electrode (Model 90-02-00). Standard solutions of copper(I1) nitrate were made with sufficient K N 0 3 to maintain a constant ionic strength of 0.1. The calibration and determination of the response of the ciipric ion selective electrode have been described before ( 7 ) . Twenty-five mI, of a 2.166 X M solution of Cu(I1) nitrate was titrated with a 2.232 X lo-' M solution of NH,(dtp). The solution was stirred at a constant rate throughout the titration. The concentration of Cu'+ was calculated from the measured potential difference between the Orion cupric ion selective electrode and the double-junction reference electrode at each point in the course of the titration. It was found essential to keep the cupric ion electrode surface free of precipitated Cu(dtp),. This was accomplished by removing the electrode from the titration vessel and wiping the electrode surface to free it from any adhering precipitate. Failure to do this at least three or four times in the course of a titration resulted in nonreproducible potential differences. All titrations were carried out in triplicate to ensure that the potential differences were reproducible. The above titrimetric procedure was repeated with NH,(dmp) as the titrant. Twenty-five mL of a 2.211 x M solution of copper(I1) nitrate was titrated with a 2.241 X lo-' M solution of NH4(dmp). Examples of the experimental curves are shown in Figure 1.
RESULTS AND DISCUSSION Attempts to synthesize Cu(dtp), resulted in the formation of a yellowish brown oil which was separated into two components by thin-layer chromatography with a mixture of toluene and ethyl acetate (92:8 v/v) as the developing solvent. The yellowish brown oil separated into two zones. a colorless zone which contained the dimeric copper(1) complex (Ri = 0) and a pale yellow zone which contained the Cu(dtp), complex ( R , = 0.55). The dimeric copper(1) complex was identified as the 1:l complex of copper(1) and d t p by the characteristic bands in its infrared spectrum (9). The pale yellow zone contained (Cu(dtp), which has an absorption band in the visible region a t 420 nm; neither the disulfide oxidation product nor the 1:l complex of copper(1) absorbed in this
473
a Ref. (8).
3wc
.
I
Ref. ( 9 ) .
Z(h*'
500
I330
%X
el>,
crr 1
Figure 2.
Infrared spectrum of the copper(1)complex of d m p in KBr
1
I ~-.L
5x
~
~~~~.
32:
:4 : : T ~ l
Infrared spectrum of the copper(1)complex of dmp in KBr in the low frequency region Figure 3.
region of the spectrum ( 5 ) . The colorless zone a t R, = 0 was extracted with chloroform. The solution gradually darkened and after about 2 h it had an absorption band at 420 nm. This indicated that Cu(dtp), was formed in the chloroform solution of the copper(1) complex and existed in equilibrium with the copper(1) complex and the disulfide ( 5 ) . R,P(S)S-S(S)PR,
+ Cu,[S(S)PR,],
2 Cu[S(S)PR,]?
(1)
where R = OC2Hj. A comparison of the infrared frequencies that have been reported for Ni(dtp), (8) and the copper(1) complex of d t p (91,with the frequencies that were observed for the yellowish brown oil (Table I) confirmed the finding that it is a mixture of the copper(1) and copper(I1) complexes of dtp. The infrared spectrum of the copper(1) complex of dmp, which is a pale yellow solid, has two prominent bands a t 840 cm-l and 489 cm-l (Figures 2 and 3) which are not found in the infrared spectra of either the copper(1) complex of d t p
474
0
ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978
Table 11. Mass Spectrum of the Copper(1) Complex of dmp Relative mle intensity mie 314 76 5.5 315 93 100 316 125 65.0 370 157 10.5 377 158 9.1 379 188 11.3 440 220 6.0 442 0.9 222 250 7.1
Relative intensity 39.3 4.2 7.6 2.5 2.9 1.8
0.3 0.4
(9) or Ni(dmp), (8). The presence of these two bands may be attributed either to the contamination of the copper(1) complex of dmp with a small amount of Cu(dmp), or to the presence of the disulfide oxidation product. The former is unlikely since these two bands are not found in the infrared spectra of either Ni(dtp)z or Ni(dmp),. Moreover, the presence of a copper(I1) complex could not be detected in the x-ray photoelectron spectrum of the pale yellow copper(1) complex of dmp. The 2p1p peak was found a t 953 eV (FWHM = 3 eV) and 2p3p peak a t 933 eV (FWHM = 2 eV). The satellite or secondary peaks that are usually associated with the x-ray photoelectron spectra of copper(I1) compounds were absent. It is conceivable that the copper(1) complex of dmp can have additional infrared absorption bands that arise as a consequence of ring puckering and alteration of the symmetry of the molecule. It was decided, therefore, to obtain the mass spectrum of the compound with the expectation that any copper-containing peaks would be readily identifiable since the 63Cu:65Cuisotope ratio is 7:3. The values of m / e with the corresponding relative intensities for the mass spectrum of the copper(1) complex of dmp are given in Table 11. The low molecular weight fragments at m / e values of 93 and 125 are characteristic of all dithiophosphate compounds (10) and may be assigned to the fragments P(OCHJ2 and P(OCHJZ(S), respectively. By analogy to the fragmentation patterns that have been observed for Ni[(S)SP(CH,),],, ( 1 1 , 121, the fragments observed a t m / e values of 314, 250, 157. and 188 may be attributed to the following disulfide fragments: (CH3O),P(S)-S-S-(S)P(OCHJz; (CH30)2P-S-S-P(OCH3)?; (CH30)zP-S-S-P02; (CHJ0),PS2 and (CH30),P-S-S-P. Unfortunately, no copper containing peaks could be identified. The peaks a t m / e values of 377 and 379 may be attributed to Cu(dmp), even though the ratio of the peak intensities do not correspond to a value of 7/3. It is possible that Cu(dmp), is formed when Cu,(dmp), is subjected to electron impact. The peaks at 440 and 442 may be indicative of the presence of a fragment of the dimer, but the peaks are too weak to be interpretable. The only conclusion that can be drawn is that the 1:l complex of copper(1) with the ligand, 0,O’-dimethyldithiophosphate, is nonvolatile and probably dimeric or polymeric. It has been established from the foregoing experiments that the yellowish brown oil formed by the interaction of copper(I1) nitrate with NH4(dtp) in an aqueous solution is a mixture of the copper(1) and copper(I1) complexes of dtp. The reducing action of the ligand, dtp, results in the formation of the copper(1) complex as well as the disulfide, and the following equilibria are established in the aqueous solution. Cu2++ 2[S(S)PR2] Z Cu[S(S)PR,l,
(2)
(3)
where R = OC2H5. It can be inferred from Equations 2 and 3 that copper(I1) reacts with the ligand, dtp, in the stoichiometric ratio of 1:2 to form not only the copper(I1) complex of dtp but also the copper(1) complex of dtp and the disulfide. It should be possible to verify the stoichiometry of the reaction and t o follow the course of the titration of a solution of copper(I1) nitrate with NH4(dtp) with a cupric ion selective electrode. If there are no other complexes of importance in solution, it should be possible to calculate the solubility product of Cu(dtp), from the titrimetric data in the manner outlined below. The solubility product of Cu(dtp), at a constant ionic strength of 0.1 is defined by: Ksp = [ C ~ ~ + ] [ d t p - ] ~
(4)
where the terms in square brackets are molar concentrations. If the equilibrium concentrations of Cu2+ and dtp are determined, K,, can be calculated. This can be accomplished by the addition of an excess of dtp to a solution containing copper(I1); the uncomplexed Cu2+can be measured with the cupric ion selective electrode and the unreacted dtp can be calculated from the stoichiometric excess of NH,(dtp) added, Le., after the equivalence point in the titration of copper(I1) nitrate with r’;H4(dtp), [dtp-] =
c, v,
- 2c,
vo +
vo
(5)
VL
where C, and CL are the initial molar concentrations of copper(I1) nitrate and NH,(dtp), respectively; V, represents the initial volume, in mL, of copper(I1) nitrate and V, the volume of NH,(dtp) added. At the equivalence point, V, mL of the titrant has been added and, 2c,
v, = CLV,
(6)
Substitution of Equations 5 and 6 in 4 and rearrangement gives, ( V o + VL)
[CU2+]1
-.
CL
VL -
(Ksp)’
CL ve ___ (Ksp)’’
(7)
A plot of values of (Vo + V L ) / [ C U ~ + ]as” ~ordinates and the corresponding values of VL as abscissa should give a straight line of slope CL/(K,,)’j2. Values of [Cu”] were obtained from the equation [CuZ+]= 1O(E-E’)/z9.58 (at 25 ’C)
(8)
where E is the potential difference, in millivolts, between the cupric ion selective electrode and the double junction reference electrode and E’is a constant, the value of which depends on the standard electrode potential of the cupric ion selective electrode, the liquid junction potential, the potential of the reference electrode, and the activity coefficient of Cu2+ in solution. The constant E‘was evaluated by substitution in Equation 8 of the calculated values of [Cu2+]before the equivalence point and the corresponding values of E that were obtained experimentally. Before the equivalence point, the concentration of Cu2+ions can be calculated on the assumption that the only equilibria that govern the free Cu2+concentration are represented by Equations 2 and 3. Hence, [CU2+]=
C,Vo-0.5C~V,
vo +
(9)
VL
Measured values of E and the calculated constant E‘ were employed in obtaining values of [Cu*+]after the equivalence point. The average value of the slope of the straight line given by Equation 7 was found t o be 9.088 X lo5 for three replicate
ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978
titrations. The initial concentration of the ligand, CL,was 2.232 X lo-* M and the calculated value of K,,, the solubility product of Cu(dtp),, is 10-15.22 The basis for the above calculation is, (a) that Cu2+reacts stoichiometrically with the ligand, dtp, in the ratio 1:2 and (b) t h a t in the presence of a n excess of the ligand, the only copper-containing species in solution is the uncomplexed Cu2+. The stoichiometry of the reaction was verified by determination of the x intercept of the straight line given by Equation 7 . The maximum difference between the value of V , that was determined from the x intercept and the calculated value on the basis of the assumed stoichiometry for the three replicate titrations was 0.08 mL which represented an error of +1.6?&. The total copper concentration in a solution containing an excess of the ligand was determined by atomic absorption spectrophotometry. A 25-mL aliquot of a standard solution M) was titrated with a standard of copper(I1) nitrate solution of NH,(dtp) (-lo-' M). The titrant was added until the d t p was in slight excess. T h e yellow-brown oil was removed by filtration, first through a 5 p m Millipore filter and then through a 0.2-pm Millipore filter. Three successive filtrations removed the oil as well as any suspended particles that were present in solution. The total concentration of copper in solution was determined by atomic absorption spectrophotometry and the free ligand concentration, [dtp-I, was calculated from Equation 5 . In a typical experiment, the total copper concentration det,ermined by atomic absorption spectrophotometry and assumed to be the value of [Cu"] in solution was 1.42 X M. The concentration of the free ligand, [dtp-I, that was calculated from Equation 5 was 1.86 X Hence the calculated solubility product of Cu(dtp),, from Equation 4 is 10-15.3. This value is in reasonable agreement with the value of K,, determined potentiometrically with the cupric ion selective electrode, and the assumptions on which the calculation is based are valid. The thermodynamic solubility product of Cu(dtp), reported by Kakovsky ( I ) is 10""''; if an approximate correction is made for ionic strength effects, the solubility product in 0.1 M KN03 is 10-'5.13which is in fair agreement with the value of 10-15.22 that was measured potentiometrically with the cupric ion selective electrode. The solubility product of 10-"-" reported by Tulyupa (2) is in error. Replacement of the ethoxy groups on the phosphorus atom in dithiophosphonic acid by methoxy groups results in a marked change in the properties of the ligand. The reaction products of 0,O'-dimethyldithiophosphonic acid, (drnp), or its ammonium salt, NH,(dmp), and copper(I1) nitrate are quite different from those obtained with 0,O'-diethyldithiophosphonic acid. The predominant complex is a copper(1) complex of dmp instead of the expected copper(I1) complex, Cu(dmp),. In addition, experimental evidence was obtained for the successive formation of a series of copper-containing complexes. A solution of copper(I1) nitrate was titrated with NH,(dmp) and the concentration of Cu'+ was monitored throughout the course of the titration with a cupric ion selective electrode. A plot of the potential differences between the cupric ion selective electrode and the reference electrode as ordinates and the volume of the titrant added as abscissa did not give the expected sigmoid-shaped curve with a well-defined vertical segment a t the end point. Instead, a titration curve with a long drawn out end point was obtained (Figure 1) which indicated that the concentration of Cu" gradually decreased over a wide concentration range of added ligand. I t may be deduced from the absence of a sharp drop in Cu2+concentration in solution that the successive formation of several copper-containing complexes contributes to the gradual decrease in the concentration of Cu2+ in solution.
475
Although the copper(I1) complex of dmp could not be prepared in aqueous solution as a stable species, the dimeric copper(1) complex of dmp, Cu,(dmp),. was readily sqmthesized as a pale yellow solid. An approximate value for the solubility product of Cu,(dmp), was obtained by two methods: (a) The solid Cu,(dmp), was equilibrated with an aqueous solution containing copper(I1) nitrate at an ionic strengt.h of 0.1. The unreacted Cu?(dmp),,was separated by filtration and the total copper in solution was measured by atomic absorption spectrophotometry. (b) The solid Cuz(dmp)2was equilibrated in an aqueous solution containing NH,(dmp). The unreacted Cu2(dmp), was separated by filtration and the total dmp in solution was determined fluorimetrically. In this experiment, no inert electrolyte was added to control the ionic strength because of the quenching effect exerted by most common anions. The solubility product of Cu2(dmp), is given by K , = [Cu+I[dmpI
(10)
When Cu2(dmp),is equilibrated with aqueous solutions of M to 5.53 X lo-' M) the total copper(I1) nitrate (1.11X copper in solution varied from 2.21 X 10.' M to 7.42 X M. The excess copper in solution was produced by the dissociation of Cu2(dmp),and was assumed to be present only as Cut. On the basis of this assumption, the calculated solubility product of Cu2(dmp), varied between and
10 9.35
When Cu2(dmp)2is equilibrated with a known excess of the ligand, the solubility product of the complex may be calculated from Equation 10 in which [dmp-] is the sum of the concentrations of added NH,(dmp) and the dmp produced by the dissociation of Cu,(dmp),. The added ligand varied from 1.12 x M t o 5.61 X lo4 M, and the total dmp found in solution by the fluorimetric method Iraried from 3.60 X lo-' M to 4.09 X 10-' M. Hence, the concentration of dmp introduced into solution by the dissociation of Cu2(dmp),ranged from 3.49 x 10 M to 3.53 X 10-5M. The solubility product therefore. of Cu,(dmp), calculated from Equation 10 varied between and 10".". In the absence of any added NH,(dmp), the concentration of dmp in solut,ion measured fluorimetrically in solution was 3.17 X lo-' M and the solubility product of Cu,(dmp), calculated from Equation 10 was lo-'.". There are many uncertainties in the experimental techniques and in the assumptions involving the molecular and ionic species that are present in solution. The above measurements therefore, give only an order of magnitude of the solubility product of Cu,(dmp),.
'
LITERATURE CITED I . A. Kakovsky, "Proceedings of the Second International Congress of Surface Activity, London", Vol. IV. Academic Press, New York, N.Y., 1957, p 225-237. F. M. Tutyupa, Khim. Tekhnol., 15, 115 (1969); Chem. Abstr., 74, 68389s 119691. J. R . Wasson, G. M. Woltermann, and H J. Stoklosa, Forfschr. Chem. Forsch.. 35, 65 (1973). A. C. Hili, M. Akhtar. M. Mumtaz, and J. A. Osmani. Ana/ysf(London). 92, 496 (1967). A. C. Hill, J . Sci. Food Agric., 20, 4 (1969). 0.E. Coldberry, W. C. Fernelius, and M. Shanma, Inorg. Synth., 6, 142 (1960). H. Wada and Q. Fernando, Anal. Chem., 43, 751 (1971). W. Rudzinski, G . T. Behnke, and Q. Fernando, Inorg. Chem., 16, 1206 (1977). K. Sakata and M. Nanjo, Tohoku Daigaku Senko Seiren Kenkyusho Iho. 26, ( I ) . 1 (1970): Chem. Abstr., 74, 930381 (1971). J. N. Damico, J . Assoc. Offic. Anal. Chem.. 49, 1027 (1966). R. G. Cavell, W. Eyers, and E.D. Day, Inorg. Chem.. 10, 2110 (1971). S. E Livingstone and A. E. Mihkelson. Inorg. Chem , 9, 2545 (1970).
RECEIVED for review November 9, 1977. Accepted December 27, 1977.
