637
for such a comparison. One result of the analysis is the emphasis on the interaction constant which reflects the interaction between the stretching of bonds that are trans to each other. It is this interaction that precludes satisfactory agreement between observed and calculated frequencies using a modified Urey-Bradley force field, and one may muse on the possible implications of this result in regard to phenomena which seemingly involve a trans effect.39 Although the remaining interaction constants ha, f i r , and faa are nontrivial to the degree that they are required in order to obtain the good fit observed in Table VI, they are small, and physical interpretation of them would be unwise. Values for the constants fra, fry, faat, and faa from the chromium calculation are transferred to the cobalt analysis. Only the magnitudes of fy, fa, and &, to which the analysis is more sensitive, are varied. Significantly larger values are required for C O ( N H & ~ +than for C T ( N H ~ ) ~and ~ + this , may be interpreted as reflecting a somewhat firmer coordination bond in the former complex.21 The stretching force constants should be singled out for comment, since reasonable estimates of their magnitudes are required in the Marcus-Hush theory of electron-transfer rates. 39-42 The constanp (fi = 1.60 mdyn/& and fi = 1.72 mdyn/A, C O ( N H , ) ~ ~ diverge +) widely from the values reported for K(M-N) f r o p the Urey-Bradley treatment (K(Sr-N) = 0.84 mdyn/A and K(Co-N) = 1.05 mdyn/A).20p21 However, they are remarkably similar to the values of the Urey-Bradley diagonal matrix element of the F,, degenetate stretching coordinate (Fdia($rN) = 1.55 mdyn/A and Fdia(CoN) = 1.75 mdyn/A), whose relation to the strength of the coordinate bond has been a r g ~ e d . ~ ~ Also, l~~J~ calculated a cobalt-nitrofyyt,
(41) R. A. Marcus, J . Chem. Phys., 24, 966, 979 (1956); 26, 867 (1957); Discuss. Faraday Soc., 29, 129 (1960). (42) N. S. Hush, Z. Elekfrochem.,61,734 (1957); J. Chem. Phys., 48, 962 (1968); Trans. Faraday Soc., 57, 557 (1961). (43) H. Block, ibid., 55,867 (1959).
gen stretching force constant of 2.00 mdyn/A based on a modified valence-force-field treatment of incomplete vibrational data, but including interaction between skeletal and ligand modes. The values of Sy derived from our analyses do not appear to be aberrational, therefore, and their incorporation in kinetic discussions would be feasible. Photodecomposition. As discussed above, the laserstimulated decomposition of CT(NH&~+in aqueous solution is very likely identical with the stepwise photoaquation previously investigated. For the reactions of the crystalline materials, it is not possible to distinguish between a photochemical substitution process and thermal decomposition, since local heating from the focused laser beam could occur. However, there is little reason to believe that the solid-state process is not directly analogous to the aqueous solution reaction with photosubstitution by the outer-sphere anion for an ammine ligand. The occurrence of this substitution process in thermal decompositions of chromium(II1) ammines has been well d o ~ u m e n t e d , ~ and ~ ! a~ ~direct photochemical parallel is available in the reaction of crystalline trisethylenediaminechromium(II1) salts. For Cr(NH3)6(N03)3,the infrared spectrum of the decomposition product is consistent with the formation of a complex in which a nitrate resides in the first coordination sphere of the Cr(II1). Once again, the involvement of the 2E, state in photosubstitution reactions of Cr(II1) is i n d i ~ a t e d , ~ * ~ lsince - l ~ * this ~ * is the state that is resonance populated by the He-Ne laser radiation. Acknowledgments. The authors thank Professors J. A. Stanko, G. M. Rosenblatt, and S . R. Polo of this university and Professor T. M. Dunn of the University of Michigan for constructive conversations related to the discussions presented here. (44) W. W. Wendlandt and C. Y. Chou, J. Inorg. Nucl. Chem., 26, 943 (1964). (45) N. Tanaka and K. Nagase, Bull. Chem. SOC.Jap., 42,2854 (1969).
Autoxidation of a Coordinated Trialkylphosphine' Donald D. Schmidt2 and John T. Yoke* a
Contributionfrom the Department of Chemistry, Oregon State University, Corvallis, Oregon 97331. Received July 17, 1970 Abstract: That the sole product of the slow but quantitative autoxidation of CoCl2(P(C2H5)JZ in several organic solvents is the phosphine oxide complex CoC1z(OP(C2H~)3)z is proven by mass balance and complete characterization4 of the product. In its initial stages, the reaction is first order in oxygen and first order in complex. Azobisisobutyronitrile and hydroquinone have no effect. A dissociation mechanism is precluded since autoxidation of uncoordinated phosphines is a radical process giving mixed RnP(0)(OR)3-,,products.5 The mechanism may involve the formation and rearrangement of an O2adduct of the cobalt complex. During the reaction in solution the redistribution equilibrium C O C ~ ~ ( P ( C , H+ ~)~ C)O~ C ~ ~ ( O P ( C ~e H ~~COC~~(P(C~H~)~)(OP(C~H~)~) )~)~ is observed, with an equilibrium constant of about ten. Resolution of the components of the transiton 4A2 (T1(P)(Td) is partial in the complexes CoCLL2 (CzJ and complete in the complex CoCLLL' (CJ. -f
n the autoxidation of free trialkylphosphines,6 intermediate phosphoranyl radicals ROPR3 can decempose to give phosphine oxides or to give trivalent phos-
I
(1) Presented in part at the 158th National Meeting of the A n - " Chemlcal Society, New York, N. Y., Sept 1969, Abstract INOR 203; taken from the P h D . Dissertation of Donald D. Schmidt, Oregon State University, 1970. (2) Woodrow Wilson Fellow, 1965-1966; NDEA Fellow, 1966-1969. (3) To whom inquiries should be addressed.
phorus esters ROPR2, susceptible to further oxidatioa6 In this way, a mixture of all the R,PO(OR)3-, products is obtained. In a coordinated phosphine ligand, phosphorus is four coordinate and its lone pair is involved in the ,, portion of the coordinate bond. In the autoxida(4) D. D. Schmidt and J. T. Yoke, Inorg. Chem., 9, 1176 (1970). S.A. Buckler, J. Amer. Chem. SOC.,84, 3093 (1962); M. B. Floyd and C. B. Boozer, ibid., 85,984 (1963). (6) C. Walling, Aduan. Chem. Ser., No. 75, 170 (1968). (5)
Schmidt, Yoke
Autoxidation of a Coordinated Trialkylphosphine
638
-
.... .,/
0
.20
0 6d e
.I5
-
0
P
I _ -,g/-, ..I..
13
,. , p . 14
I
15
16
17
18
FREOUENCY, kK
Figure 1. Visible spectra of cobalt complexes in benzene: COC~~(P(C~0 H,~COC~~(OP(C~H~)~)~; )~)~; -, 1 :1 mixture.
- - -,
.05
t zi I
OP
4,
4
0
d, I
I
I
I
2
4
6
8
*i 1
X
Figure 2. Continuous variations studies; X = [CoClZ(OP(CzH&)zI/([CoC1z(P(CzH~)a)z] [COC~~(OP(CZH~)~)Z]}, [CO]= 1.OO X Benzene, 24”: 0,700 nm; 0,669 nm. tert-Butylbenzene, 700nm: A, 36”; A,6 5 ” .
+
tion of transition metal-phosphine complexes, examples have been reported of oxidation of the metal or of the phosphine or both, and of the formation of both reversible and irreversible oxygen add~cts.~-~O Jensen and visible spectra were obtained of samples removed a t coworkers” report that C O C ~ , ( P ( C H ~is) ~oxidized )~ to various times during the course of the autoxidation a cobalt(II1) complex, while solid C O C ~ ~ ( P ( C ~ His& ) ~reaction in benzene. These spectra consist of three said to be “easily oxidized on lying in the air . . . attains major peaks, the positions of which d o not correspond a lighter blue color and becomes insoluble in pentane, to any of the peaks in the spectra of the starting material being transformed into a complex compound of the or of the final product. This anomaly was explained phosphine oxide.” No experimental evidence is given. by the finding that mixtures of benzene solutions of the Some phosphine complexes, such as Co(NCS)*(P(C2pure phosphine and phosphine oxide complexes do not H&),, react instantly with air,12 while others, such as obey Beer’s law and show the same three-peak spectrum. NiBr2(P(CH3)&, are scarcely affected.13 The decompoIdentical spectral behavior was found in tert-butylsition of dichlorobis(tri-n-butylphosphine)cobalt(II) in benzene, but not in o-dichlorobenzene (vide infra). The an oxygen-helium atmosphere at elevated temperatures spectra of benzene solutions of CoC12(P(C2H5)3)2, has been described.l4 C O C ~ ~ ( O P ( C ~ Hand ~ ) ~a) ~1, : l mixture of them are A study of the autoxidation of dichlorobis(triethy1given in Figure 1. Continuous-variations plotsI5 for phosphine)cobalt(II) is now reported. The characteribenzene and tert-butylbenzene solutions are shown in zation of the complete series of potential autoxidation Figure 2 ; these clearly indicate a 1 : 1 interaction of the has been reproducts, COC~,[(C,H~),PO(OC~H~)~-,~~, two complexes, for which a redistribution equilibrium ported previously. reaction may be written. COC~Z(P(CZHS)~)Z COC~Z(OP(CZH~)~)Z e Results ~COCIZ(P(CZH~)~)(OP(CZH~)~) Dichlorobis(triethylphosphine)cobalt(II) was found An accurate evaluation of the equilibrium constant to be stable in dry air, in contrast to the previous refrom spectrophotometric data was not practical in this port.” When the complex was stored under an atmosystem owing to continuous overlap of the spectra of the sphere of pure oxygen at room temperature for 1 day, components throughout the region of interest and to the there was no change in pressure or in the weight, color, low concentration range available because of solubility or pentane solubility of the complex. The complex is limitations. Approximate values were obtainedI5 slightly hygroscopic, and presumably Jensen’s observausing Schaeppi and Treadwell’s tangent method and tions were due to moisture. Autoxidation does take Schwarzenbach’s method modified for use with a nonplace in various organic solvents, slowly at room temlinear least-squares curve-fitting program. l6 In both perature, as is indicated by oxygen consumption, lightsolvents and at the several temperatures studied, values ening of the blue color, and changes ir, the infrared of the order of magnitudes of ten were obtained for the spectrum. The sole product of quantitative autoxidaequilibrium constant. tion in benzene, o-dichlorobenzene, and tert-butylKinetic Studies. Volumetric measurements were benzene is dichlorobis(triethy1phosphine oxide)comade of the rate of absorption of oxygen gas by tertbalt(I1) as is proven by the gain in weight (correbutylbenzene solutions of dichlorobis(triethy1phossponding to exactly 1.OO mol of oxygen/mol of complex) phine)cobalt(II) at 36”. The oxygen pressure was held and the elemental analysis, melting point, and infrared constant during each run. The agitation was sufficient spectrum of the product,4 as well as the isolation of trithat diffusion across the gas-liquid interface was nonethylphosphine oxide in pure form following its dislimiting. Autoxidation of the solvent was negligible. placement from the cobalt complex by pyridine. Plots of gas absorption us. time are given in Figure 3. Visible Spectra. A surprising result was found when At constant oxygen pressure, doubling or tripling the (7) A. Sacco, M. Rossi, and C. F. Nobile, Chem. Commun., 589 (1966). concentration of complex increases the rate proportion(8) L.Vaska, Science, 140, 809 (1963). (9) L. Vaska and D . L. Catone, J . Amer. Chem. Soc., 88,5324 (1966). ally, while at constant complex concentration, doubling (IO) G. Wilke, H. Schott, and P. Heimbach, Angew. Chem., Int. Ed. the oxygen pressure doubles the rate. Hence in the Engl., 6 , 92 (1967). (11) K. A. Jensen, P. H. Nielsen, and C. T. Pedersen, Acta Chem. initial stages of the reaction the rate law is Scand., 17, 1115 (1963). (12) A. Turco, C. Pecile, M. Nicolini, and M. Martelli, J . Amer. Chem. d(mo1 of Oz)/dt = k[CoClz(P(CzHs)3)zlPo~
+
Soc., 85, 3510 (1963); M . Nicolini, C. Pecile, and A. Turco, ibid., 87,
2379 (1965). (13) B. B. Chastain, D. W. Meek, E. Billig, J. E. Hix, and H. B. Gray, Inorg. Chem., 7, 2412 (1968). (14) K. Moedritzer and R. E. Miller, J . Therm. Anal., 1, 151 (1969).
Journal of the American Chemical Society
1 93:3 /
(15) F. J. C. Rossotti and H. Rossotti, “The Determination of Stability Constants,” McGraw-Hill, New York, N. y., 1961. (16) P. B. DeGroot, Ph.D. Dissertation, Oregon State UniverSity, 1970.
February 10, 1971
639 Table I. Solvent Effects on the Visible Spectrum of CoClz(P(CzHa)a)z Absorption maxima, nm Solvent (extinction coefficienty Benzene tert-Butylbenzene n-Pentane Chloroform Acetone o-Dichlorobenzene o-Dichlorobenzeneb Carbon tetrachloride .Abbreviations: * A t 50".
s
730 (460) s, 623 (580), 606 (590) 730 (380) s, 622 (520), 603 (530) 730 br, 610 sh, 590 sh, 580 br, 570 sh 716 (450) s, 613 (650) 715 (350) s, 620 (430) sh, 600 (470) 690 (430), 654 (410), 640 (410), 608 (510) 683 (500), 650 (480), 606 (480) 668 (>lo3), 555 ( > 5 X lo3), 470 (>5 X 102) sh =
sharp, br = broad, sh = shoulder.
Pseudo-first-order plots were constructed as follows, of the function In [CoC1z(P(C~H~)~)~l/[CoC1~(P(C~H~)~)~]~ 4 8 12 16 us. time. The number of moles of oxygen absorbed a t Time, min. each time was taken to represent conversion of phosFigure 3. Reaction of 0% with CoClz(P(CzHj)& (40.0 ml of tertphine complex to phosphine oxide complex, using the M Co, 678 butylbenzene solution). Bottom, 36": 0 , 1.95 X Torr; 0, 1.31 X 10-2 M Co, 678 Torr; 0,0.65 X loA2M C o , 678 equation Torr; A, 1.31 X M Co, 339 Torr. Top, 65", 1.31 X M CoCldP(CzH&)z 0 2 --+ CoCMOP(CzH.da)z Co, 679 Torr: A, 0.76 X M AIBN; 0, 1.00 X M hy. The concentration of original phosphine complex redroquinone; 0,no additive (duplicate runs). maining was then calculated using approximate values vent effects on the visible spectrum of dichlorobis(triof the equilibrium constant for the redistribution reacethylphosphine)cobalt(II) are demonstrated by the data tion. Data from experiments with three different of Table I. The spectra in the first five solvents listed starting concentrations of cobalt complex fell on the are similar. Significant differences in o-dichlorobensame pseudo-first-order curve, but this was linear only zene may be attributed to partial ionic dissociation in for the first quarter-life. It then bent upward, indithis solvent; at room temperature the molar conductcating retardation of the later stages of the reaction. ance of a 10-3 M solution was 0.27 ohm-' cm2 mol-', Values of the redistribution equilibrium constant from which may be compared to the value 1.43 ohm-' cmz 10 to 40 were tried; a higher order of magnitude would mol-' found for tetra-n-butylphosphonium chloride in be needed to improve the linearity of the kinetic plot. M soluthis solvent. In carbon tetrachloride, a Also shown in Figure 3 are measurements of the rate tion could not be prepared at room temperature; the of oxygen consumption of tert-butylbenzene solutions solubility appeared to be somewhat less than half of of dichlorobis(triethylphosphine)cobalt(II) at 65 O conthis. The very intense band at 555 nm of the deep taining azobisisobutyronitrile (AIBN), hydroquinone, violet solution suggests a rapid chemical change, perand no additive. Data from the three experiments fell haps a radical oxidation-reduction. 2o on the same curve. With an order of magnitude of ten for the redistribution equilibrium constant, the mixed complex CoC12Discussion (P(CzHj)3)(0P(CzH5)3) is the predominant solute Spectra of the Cobalt(I1) Complexes. The visible M solutions of species present in a 1 : 1 mixture of absorption is due to the transition v 3 = 4Az+ (T1(P) in C O C ~ ~ ( P ( C ~and H ~ )CoC1z(OP(CzH~)3)z. ~)~ The resolutetrahedral symmetry, which is decomposed to 4Az + tion of the three components of the visible transition is 4Az 4B1 4B2in microsymmetry Czv(CoX,L2) and to seen from Figure 1 to be complete for this mixed com24A" in microsymmetry C, (CoXzLLr). plex of microsymmetry C,. The approximate value of 4 A t t+- 4A' From Figure 1, it is seen that the approximation of an the equilibrium constant is reasonable statistically for a average ligand environment applies better to CoClZredistribution reaction in which coordinate bond energy (OP(CzHj)3)~ than it does to C O C ~ ~ ( P ( C ~ H That ~ ) ~ ) ~ . changes are small. there is much less splitting of the transition in the phosThe Autoxidation Reaction. A dissociative mechaphine oxide complex is reasonable in terms of the nism can be eliminated, since the autoxidation of ungreater similarity of chloride to oxygen than to phoscoordinated tertiary phosphines gives a mixture of prodphorus in coordinate bond type and position in the specucts, R,P(O) n = 0-3, and is subject to radical trochemical series. initiation and inhibition.5 For the reaction The spectrum of the phosphine complex has been reCOCL(PR~)Z 0 2 CoClz(OPR3)z ported previously in benzene solution17 and in reflectit seems reasonable energetically that the formation of ance.'* In the solution spectrum, resolution of the the two new P=O bonds should be closely associated three components of the transition into a peak at 730 with the rupture of the 0-0 bond. The formation and nm and a doublet at 623 and 606 nm is apparent. A COCL(PR~)~ + 0 2 02COCldPRah (1) very similar splitting, and its loss on dissolution in niOzCoClg(PR& +CoClz(OPR3)z (2) tromethane, has been reported for the corresponding subsequent rearrangement of an 0 2 adductz1would be triphenylphosphine complex. l9 Some significant solconsistent with the results of the initial-rate studies.
+
+
+
+
+
(17) I
