摘要:

1975 1729 Electron Spin Resonance Studies on Bis(dimethylglyoximato)cobalt( 11) and its Complexes with Pyridine By Antal Rockenbauer,' Eva Budb-ZBhonyi and LBst16 I. Simsndi Central Research Institute for Chemistry, Hungarian Academy of Sciences Budapest Hungary The e m . spectra of bis(dimethylglyoximato)cobalt(ll) and its pyridine complexes have been studied in methanol. The spectra obtained in both liquid and frozen methanol show the existence of three paramagnetic centres viz. [Co(Hdmg),] [Co(Hdmg),(py)] and [Co(Hdmg),(py),] (Hdmg is the monoanion of dimethylglyoxime) and that of a diamagnetic adduct [{Co(Hdmg),(py)),]. The magnetic parameters have been determined for each paramagnetic species ; a significant temperature dependence has been observed in the case of [Co(Hdmg),].The parameters indicate a low-spin 2Al ground state. The Fermi-contact term for [Co(Hdmg),] varies between -20.0 x 1 0-4 and -1 0.5 x 1 O4 cm-l on decreasing the temperature from 62 "C to the freezing point of methanol ; the values for [Co(Hdmg),(py)] and [Co(Hdmg),(py),] are -3.9 x cm-l respectively at liquid-nitrogen temperature. The observed trend is consistent with the variation in the axial ligand field which affects the Fermi term through 3d-4s mixing. The linewidth variation of the hyperfine multiplet in the e.s.r. spectrum of [Co(Hdmg),] has been studied in methanol as a function of temperature. The widths of the strongly overlapping lines have been determined by an automatic-fitting procedure. Good simulation and rapid con-vergence have been obtained by assuming Lorentzian lineshapes.The predominant factors contributing to the linewidths are spin-rotational relaxation a t high temperatures and anisotropic relaxation at low temperatures. and 9.8 -BIS( DIMETHYLGLYOXIMATO)COBALT(II) [Co(Hdmg),] has been extensively studied as a vitamin Bizr model compound.1 On the basis of the e.s.r. spectra it has been established that vitamin B12r and [Co(Hdmg),] have also similar magnetic properties the cobalt ion is in the low-spin ,Al state and the unpaired electron is localized mainly on the d,a The axial localization of the unpaired electron permits the form-ation of various mixed-ligand complexes to be followed using the e.s.r. spectra. In an earlier paper the e.s.r. spectra of liquid methanol solutions containing [Co-(Hdmg),] and varying amounts of pyridine (py) were studied.The results clearly pointed to the existence of 1 1 and 1 2 adductsof composition [Co(Hdmg),(py)J (n = 1 or 2). The corresponding stability. constants were determined. However the e.s.r. spectra recorded in frozen solution led to the conclusion 2 9 3 that various nitrogen bases form only 1 2 adducts; it was not possible to detect the 1 1 adducts because of extensive dimerization. There is some weak axial co-ordination even in non-co-ordinating solvents as shown by the solvent dependence of the magnetic parameter^.^^^ The objective of this paper is the detection of the 1 1 complex [Co(Hdmg),(py)] in frozen solution and to determine the dimerization constant from the liquid-phase e.s.r.spectra. In addition to a study and inter-pretation of the magnetic parameters we shall deal with the variation in linewidths in the liquid phase. RESULTS E.S.R. Sfiectra.-In frozen methanol solution [Co(Hdnig),] had a complex spectrum which changed slightly with the concentration and with the rate of cooling; therefore we shall only deal with the interpretation of spectra obtained at low concentrations and rapid freezing. The spectra exhibited a nearly axial symmetry and a S9C0 hyperfine structure. Attempts a t simulating the perpendicular part G. N. Schrauzer Accounts Chem. Res. 1968 1 97; G. Costa, Pure Appl. Chem. 1972 30 335; A. Bigotto G. Costa G. Mestroni G. Pellizer A. Puxeddu E. Reisenhofer L. Stefani and G. Tauzher Inorg. Chim. Acta Rev. 1970,4 41 ; and refs.therein. G. N. Schrauzer and Lian-Pin Lee J . Amer. Chem. Soc., 1968 90 6541. of the spectrum were unsuccessful when taking into account a rhombic di~tortion,~ quadrupole effect,* or extra ' absorption. An adequate fit was obtained however, between simulated and experimental spectra when variation in linewidths with the magnetic quantum number and orientation was permitted (Figure 1). FIGURE 1 E.s.r. spectrum of [Co(Hdmg),] in frozen methanol solution at 77 K. (-) Experimental; (- - -) calculated On adding py to [Co(Hdmg),] the form of the spectrum did not change until the concentration of py reached ca. 80% of that of the complex; however the amplitude decreased with increasing py concentration. When the py [Co(Hdmg),] ratio exceeded 0.8 1 a 1 1 1 triplet superhyperfine structure appeared on the parallel hyperfine lines.At a ratio of 1 .O 1 the triplet predominated although a 1 2 3 2 1 quintet also appeared. The above changes were most pronounced in the fourth parallel line (Figure 2). When the py concentration was greater than that of [Co(Hdmg),] the 1 2 3 2 1 quintet became the pre-dominant superhyperfine pattern. The overall intensity of the spectrum showed a characteristic dependence on E. K. Ivanova I. N. Marov A. T. Panfilov 0. M. Petruhin, andV. V. Zhukov Zhur. neovg. Khim. 1973,18 1298. J . R. Pilbrow and M. E. Winfield MoZ. Phys. 1973,2!5 1073. A. Rockenbauer E. Bud6-Z&honyi and L. I. SimOndi J . Co-ordination Chem. 1972 2 53. (a) I?. D. Tsay H. B. Gray and J .Danon J . Chem. Phys., 1971 54 3760; ( b ) L. D. Rollman and S. I. Chan ibid. 1969 50, 3416 1730 J.C.S. Dalton [py]. When the py [Co(Hdmg),] ratio was either zero or very large the overall signal intensity was the same but at a ratio of 1.0 1 the intensity was at a minimum. At [Co(Hdmg),]~ = lo- mol dm-3 the mimimum intensity was ca. 20 times smaller than at the maximum. The appearance of 1 1 1 and 1 2 3 2 1 superhyperfine structures confirms the existence of adducts with com-positions [Co(Hdmg),(py)] and [Co(Hdmg),(py),] and the observed intensity changes indicate dimerization of the five-co-ordinate complex in agreement with an earlier 10 G FIGURE 2 Part of the e.s.r. spectrum of the [Co(Hdmg),]-py system in frozen methanol solution at 77 K.Fourth line of the cobalt hyperfine structure in paralIel orientation ( a ) , [Co(Hdmg)il; ( b ) [Co(Hdmg),(py)I ; (4 [Co(Hdmg),@y),I report.? the spectra of frozen solutions are listed in Table 1. Values of the magnetic parameters obtained from TABLE 1 E.s.r. parameters for the [Co(Hdmg),]-pyridine system in methanol at 77 K Aph A l C O A p cm-l Adduct gr gl [Co(Hdmg)J 2.0106 2.236 102.0 28.6 [Co(Hdmg),(py)] 2.0137 2.24 * 86.6 16.0 * 16.6 [Co(Hdmg),(py),] 2.0161 2.206 * 78.0 9.0 * 15.8 * Approximate estimation. The solution spectra also support this conclusion. The e.s.r. spectrum of [Co(Hdmg),] in methanol was an eight-line multiplet. The width of the hyperfine lines the line separation and the centre of the spectrum were strongly temperature dependent.Since the hyperline lines over-lapped considerably a simulation procedure was used to determine the isotropic g values (gim) the isotropic hyper-fine constant (Aim) and the different widths of the eight lines. Spectra of samples containing various amounts of py consisted of three components indicating the relatively long lifetimes of the mixed-ligand species [Co (Hdmg) ,(py)] and [Co(Hdmg),(py) ,I. The iterative procedure described earlier for the determination of the stepwise stability constants from e.s.r. data has now been modified so that in addition to K and K, the dimerization constant (if the extent of dimerization is significant) and the spectrum of each paramagnetic component can also be obtained using e.s.r. spectra of six different solutions of suitable com-position.The dependence of signal intensity on cobalt concentration shows that dimerization occurs only in the case of [Co(Hdmg),(py)] according to the equation For the definition of the stability constants see Table 2 in which the corresponding values are listed. Table 3 con-tains the isotropic magnetic parameters determined from the resolved spectra. For theAiso value of [Co(Hdmg),(py),] only an upper limit is given since its spectrum was a structureless singlet. TABLE 2 Stability constants (dm3 mol-l) of the [Co(Hdmg) ,I-py systems in methanol 2[CO (Hdmg) 2(PY)l + [(PY)(Hdmg),CoCo(Hdmg),(PY)l. Method O,/"C Kl Kzl Kd E.s.r. 41 131 & 20 0.58 f 0.08 E.s.r. 20 175f 30 0.83 f 0.08 E.s.r. 0 300% 60 1.62 f 0.30 E.s.r.-30 863 f 100 3.00 f 0.4 2.4 f 4 E.s.r. -50 1 610 + 200 7.34 f 0.8 77.0 f 10 Optical r.t. 128 0.93 (py)21/[~(Hdmg)~@y)l by]* and Kd = [{Co(Hdmg)2@y))~l/ Kl = [Co(Hdmg),(PY)l/[Co(Hdmg)21rPYl K2 = [Co(Hdmg)2-[Co (Hdmg) dPY)I2. TABLE 3 E.s.r. parameters and near-i.r. bands for the [Co(Hdmg),]-py system in methanol at room temperature Adduct g, 104A~s,/cm-1 1.r. band (cm-') 8 600 11 200 2.193 68.2 10 800 38.9 [Co (Hdmg) ,@Y) J 2.142 < 20 13 600 [CO (Hdmg) J [ WHdmg) 2(PY)I 2.194 Electronic Absorption Spectra.-Electronic absorption spectra were recorded a t room temperature. The complex [Co(Hdmg),] in methanol showed a weak asymmetric band in the near4.r. region a medium-intensity band and two shoulders in the visible and strong charge-transfer (c.t.) bands in the U.V.region. The i.r. band appeared at 10 800 cm-l and the visible band at 21 200 cm-l. On adding py to the samples the positions of these bands varied. With increasing [py] the asymmetric i.r. band first split into two and the visible one showed a blue shift; at high [py] there was again one i.r. band (13 500 cm-1) and the visible band returned t o 21 200 cm-l. By means of the electronic absorption spectra the stepwise stability constants (dimerization is negligible a t room temperature) and the spectra of [Co(Hdmg),(py)] and [Co(Hdmg),(py),] were determined with the same computer program as used in the analysis of the e.s.r. spectra. The stability constants obtained by the two methods were in fairly good agreement (cf. Table 2). Linewidth Measurements.-The temperature dependence of the e.s.r.spectra of [Co(Hdmg),] in methanol was studied in the interval between 62 and -60 "C. At 62 "C the spectrum was a broad symmetric singlet (Figure 3). On decreasing the temperature the eight-line hyperfine structure gradually appeared. The best resolution was obtained at ca. -20 "C (Figures 4 and 5). The linewidth increased on going to higher fields. At ca. -60 "C the lines collapsed into an asymmetric singlet (Figure 6). 7 G. N. Schrauzer and R. J. Windgassen Chern. Ber. 1966,99, 602 1975 1731 For determination of the linewidths an automatic simulation procedure was developed. The spectrum was assumed to be the superposition of eight individual derivative curves as in equation (l) where I is the baseline parameter A the amplitude of the individual curves and M the magnetic quantum number for the W o nucleus with values from -3 to 1.2 . The lineshape function (F) is FIGURE 3 E.s.r. spectrum of [Co(Hdmg),] in methanol solution at 60 "C. (-) Experimental (--) calculated Lorentzian curves FIGURE 4 E.s.r. spectrum of [Co(Hdmg)& in methanol solution at - 24 "C. (-) Experimental (- - -) calculated Lorentzian curves FIGURE 5 at 2 "C. curves E.s.r. spectrum of [Co(Hdmg),] in methanol solution (-) Experimental (- - -) calculated Gaussian either a derivative Lorentzian [equation (2)] or a derivative FL(B,Bdl,Cbi) = - ( B - B&f)obi[odf2 + (B - %f)21-2 (2) Fa-(B,Bbl,obl) Gaussian function [equation (3)] where BM is the position = -(B - BU)oM-3exp [-(B - B H ) ~ l n 2 o ~ - ~ ] (3) of the line centre and 2 1 s ~ the linewidth at one half the maximum intensity.For an isotropic-spin Hamiltonian the position of the hyperfine lines in second order is given by equation (4). The parameters gi and Aim can be determined from the simulation parameters B and B,, according to equation (6) where h is Planck's constant Po is the Bohr magneton and a the microwave frequency (its value in our experiments was 6.76 x 1O1O s-l). In the case B x = B - BmpM - Bw2(2B0)-1 (9 - M2) (4) (6) giso = h~o(2~PoBo)-l Aim = aoBsep/Bo of symmetrical lineshapes the number of parameters to be optimized in the curve-fitting procedure is 12; these are A I, B, B and the eight ax values. We also carried out calculations for asymmetrical lineshapes in which the two wings of each line are to be characterized by different CM values.This increases the number of parameters to be I 100G , L FIGURE 6 E.s.r. spectrum of [Co(Hdmg),] in methanol solution at -60 "C. (-) Experimental (--) calculated Lorentzian curves optimized to 20. The fitting was made by a non-linear least-squares procedure with the iteration steps based on Meiron's formula.* The above automatic-fitting procedure ensured con-vergence in ca. 10 steps for strongly overlapping lines. For lines with greater separation the convergence was more rapid. While the fit was very good for the Lorentzian lineshape poor results were obtained with Gaussian line-shape (cf. Figures 4 and 5). The asymmetrical lineshape permits slightly better simulation than the symmetrical one; however it can lead to loss of some of the strongly overlapping lines in the high-field part of the spectra.Therefore we used only the data obtained from fitting by symmetrical Lorentzian curves (Table 4). DISCUSSION Interpretation of the Magnelic Parameters.-The local symmetry of [Co(Hdmg),] and its mixed py derivatives is C2,; however the nearly axial g tensor corresponds to an effective C4tl tetragonal symmetry. The local ligand field splits the orbitally degenerate 3d levels into three singlets viz. b&&~-~a) bl(dzy) al(dz*) and a doublet e(dzz,dyz). the lack of superhyperfine interaction with the equatorial nitrogen atoms and the appearance of a superhyperfine structure due to the axial py nitrogens show that the unpaired electron is localized on the dzs orbital and the According to Schrauzer and Lian-Pin Lee J.Meiron J . Opt. SOC. Amer. 1965,55 1105 1732 J.C.S. Dalton electronic ground state can be given by the 2Al(dz~-yt2,dza) liquid-phase spectra permit the isotropic g(gi,) value hole-electron configuration. Ligand-field calculations and the isotropic hyperfine constant (Aiso) to be deter-of the g factors by Engelhardt and Green support this mined. As there is significant orbital quenching in the conclusion. spectra of low-spin cobalt(I1) complexes here the higher-The formulas of the spin-Hamiltonian parameters order contributions are neglected. have been derived by taking into account the effect of the 2E(d~-al~2,dzz,yz) excited state through spin-orbit coupling because among the excited states only this gives a first-order contribution and only this state may be of such low energy that contributions of higher order are not negligible.In the molecular-orbital (m.0.) approach both the spin-orbit and hyperfine coupling constants are changed relative to their ionic values, because of the difference in shielding of the d electrons and the only partial localization of the unpaired electron on the dza orbital. The latter effect is taken into account by introducing the reduction parameters k and k' so that k2 and kr2 are approximately characteristic ezc m712 Csl2 0112 b - l l d O512 O-312 O-512 (J- 712 a P R.m.s. error i On the basis of the hyperfine parameters estimates can be obtained for the reduction parameter kt2 and the Fermi constant K [cf.equations (8) (9) and (ll)]. For this however the signs of the coupling constants must be known. The large values of the cobalt hyper-fine coupling constants and the large orbital contribution to gl permit the conclusion that the unpaired electron is primarily localized on the cobalt 3d orbital. Therefore, the assumption that the value of the reduction 62 76.4 79.4 79.9 81.5 82.6 86.1 90.5 96.7 159.2 - 3.06 0.745 - 0.21 1 0.537 of the localization of the orbital. Here k and k' delocalization of electrons 45 60.5 61.4 62.5 65.0 67.5 71.6 76.5 77.9 129.1 - 6.41 0.472 0.124 1.095 37 50.2 61.5 53.6 55.1 57.9 60.1 64.0 68.6 109.3 -4.37 - 0.059 0.462 0.376 18 42.3 44.3 47.1 50.9 54.4 60.0 61.4 68.9 101.8 - 8.08 0.455 0.06 0.499 the hyperfine coupling constant ( A and P = goP,gNPxT-3) to different extents.Here go = 2.0023 the factor of the free electron PN the nuclear magneton g N is the nuclear factor for 59C0 and? the average inverse cube of the distance between the unpaired electron and the cobalt nucleus. In order that k2 and kT2 should actually characterize delocalization the shielding effect is taken into account separately by choosing A and P to be smaller than their ionic values. Considering a number of complexes we have found that the values of A = 515 cm-l and P = 220 x cm-l taken from Dunn lo and Abragam and Pryce,ll always provide acceptable k and k' values.For these conditions equations (6)-(9) gl = [l - #(k2A/AE)'] (6) g l = go [l + 3k2(A/AE) - 3(k2A/AE)*] (7) All = -K + ;[4kI2 - (g1 - go) - 5(RI - g,)l (8) %Lkll - go) (9) A = -K + ;[-2kt2 + v(gl - go) -are obtained where K is the Fermi constant. The ' L. M. Engelhardt and M. Green J.C.S. Dalton 1972 724. lo T. M. Dunn Trans. Faraday SOC. 1961 57 144. TABLE 4 Linewidths (G) and relaxation parameters (10' rad s-l) for [Co(Hdmg),] in methanol unpaired electron on the d,a may be different since the influences the spin-orbit and 2 33.3 35.6 39.3 43.6 48.9 54.6 60.5 66.5 89.2 - 9.98 -0.555 0.06 0.125 - 20 26.1 29.9 35.5 42.1 48.9 57.5 67.8 85.6 87.1 14.22 1.16 0.435 - 42 29.4 34.8 41.5 48.2 56.5 66.6 97.2 134.0 100.5 - 15.72 1.47 0.155 - 50 32.4 40.0 46.8 53.4 62.0 76.9 150.3 275.6 110.1 - 17.70 3.70 0.486 parameter kt2 is between 0.5 and 1.0 seems to be justified.Using this condition as a criterion one first determines the signs of the coupling constants and then the refined values of parameters kr2 and K can be given. When estimating the value of parameter kI2 in the five-co-ordinate case the Aiso and A, values were used because in the frozen solution superposition of the spectra for various co-ordination numbers does not allow un-ambiguous determination of the perpendicular com-ponents. The relative signs of A, and A l for the six-co-ordinate case have been selected so as to be consistent with the estimated upper limit for Aiso obtained from the liquid-phase spectrum.Information on k2 from the g values can only be obtained if the excitation energy is known [see equations (6) (7) and (lo)]. As k2 cannot differ significantly from k'2 the value of AE should be ca. 8000-14000 cm-l. Therefore the asymmetric band observed in this range can probably be ascribed to this transition. The asym-metry in the four- and six-co-ordinate cases and the splitting of the line in the five-co-ordinate case can be interpreted either in terms of rhombic distortion or of the occurrence of the competing 2Z3,-2A transition. Therefore the parameters k2 were evaluated by taking A206 173. 11 A Abragam and M. H. L. Pryce Proc. Roy. SOC. 1951 1975 1733 the centre of gravity of the lines as a basis.The approximate agreement between k2 and k'2 data seems to support the assignment of the near4.r. band (cf. Table 5). TABLE 5 [Co(Hdmg),]-py systems Fermi terms and orbital-reduction factors for the 104K/ Adduct cm-1 k 2 k t 2 T2 -18.8 0.978 0.721 0.0656 -3.9 0.876 0.716 0.0531 [Co (Hdmg) 21 [CO (Hdmg)2 (PY)l [Co(Hdmg) 2 (PY) 2l The effect of co-ordination on the magnetic para-meters m.0. parameters and on the Fermi constant can be discussed on the basis of Tables 1 3 and 5. For isonitrilecobalt (11) complexes Maher l2 found that gL and gi,o decrease and the Fermi term increases with increasing co-ordination number. We have observed a similar tendency in the case of the [Co(Hdmg),(py),,] type mixed-ligand complexes.The five-co-ordinate species is however an exception as its giso value is slightly greater than that of the four-co-ordinate com-plex. This anomaly may be attributed to the effect of lower symmetry manifested through changes in the energy levels and mixing of the orbitals. At the same 9.8 0.874 0.749 0.0442 CoII ion and Kds is the isotropic hyperfine constant of a 4s electron in the CoII ion. The two terms in (12) have opposite signs Kf.i. is positive (according to RlcGarvey l3 its value is 85 x cm-l) and Kgs is negative (SCF calculations for neutral cobalt yield l4 a value of -1 220 x 10-4 cm-l). For the CoII complexes under consideration KgS is greater than this value and thus the actual extents of 3 d 4 s mixing are smaller than those calculated in this work.However this does not affect the trends obtained for the values of q. Although k" may differ from the orbital-reduction factor k' we use the approximation k' = k" which presumably does not influence significantly the estimation of 7. The calcu-lated q values are listed in Table 5. The decrease in q with increasing co-ordination number is in accord with the expectation that higher co-ordination numbers imply weaker distortions of the octahedral ligand field. In the study of the liquid-phase spectra we found that for the four-co-ordinate case the temperature exerts a strong influence on the spectrum whereas for the five-and six-co-ordinate cases the spectra suffer only in-significant changes. The temperature dependence of TABLE 6 Temperature dependence of magnetic parameters in the [Co(Hdmg) J-py system A it30 K ezc giao - go cm-l AU - A 1 * gu - g l * _-62 0.2066 65.4 - 20 67.6 0.3099 45 0.2042 61.3 - 16.4 68.5 0.3063 0.2996 37 0.1997 59.7 - 15.8 70.3 18 0.1913 58.3 - 16.1 73.6 0.2870 2 0.1887 54.5 - 12.9 74.7 0.2831 - 20 0.1877 51.3 - 10.0 75.1 0.2816 - 42 0.1824 50.6 - 10.5 77.2 0.2736 * Value calculated by comparing parameters obtained in frozen solution and in the liquid phase.time the Fermi term changes in the ' normal ' fashion, indicating that the value of K is primarily determined by the axial ligand field and is practically unaffected by the lowering of symmetry. The dependence of the Fermi term on the co-ordination number is due to direct 3 d 4 s mixing.13 As no such mixing occurs in the case of octahedral symmetry the extent of mixing depends mainly on the tetragonal distortion being less sensitive to further distortion.Tetragonal distortion of octa-hedral symmetry is extensive if the axial ligand field is weak compared with the equatorial field significant 3 d 4 s mixing being observed only in the case of weak axial co-ordination. To express the correlation between 3 d 4 s mixing (7) and the Fermi term ( K ) in a quanti-tative way equation (12) has been used where K" is a reduction parameter Ki-i. is the Fermi term for the free l2 J. P. Msher J . Chem. SOC. ( A ) 1968 2918. l3 B. R. NIcGarvey J . Phys. Chenz. 1967 71 51. l4 E. Clementi J . Chem. Phys. 1964 41 295. l6 J. J. Alexander and H. B. Gray J .Amev. Chem. SOC. 1967, l6 M. E. Kimball D. W. Pratt and W. G. Kaska Inorg. Chem., N. Kataoka and 13. Kon J . Amer. Chem. SOC. 1968 90, 89 3366. 1968 7 2006. 2978. giso Aiso and K for the four-co-ordinate case is shown in Table 6. The temperature dependence of the mag-netic parameters may be due either to changes in the position of the equatorial Hdmg ligands or some changes in the axial positions. As the d, orbital is insensitive to changes in the equatorial plane it seems more likely that the observed phenomena are caused by variations in the ligand-field strength of the axially co-ordinated solvent molecules. This receives support from the fact that on decreasing the temperature the magnetic para-meters gs0 A i s o and K change in the same direction as that associated with increasing co-ordination number.In order to study the correlation between the strength of the axial ligand and the Fermi constant we analyzed the magnetic parameters of some representative CoII complexes with tetragonal symmetry,4p6*12915-22 where J . M. Assour and W. K. Kahn J . Amer. Chem. SOC. 1966,87, 207. l9 J. M. Assour J . Amer. Chem. SOC. 1965 87 4701. *O J. M. Assour J . Chem. Phys. 1965 43 2477. 21 H. A. 0. Hill P. J. Sadler and R. J. P. Williams J.C.S. 22 J. Danon P. P. Muniz A. 0. Caride and I. Wolfson J . Dalton 1973 1663. Mol. Struct. 1967 1 127 1734 J.C.S. Dalton this was permitted by the available data. The data were evaluated on the basis of the same principles used in the case of dimethylglyoximatocobalt complexes.The re-sults obtained for pentacyanocobaltate(r1) and iso-nitrilecobalt (11) complexes are listed in Table 7 whereas the data for B12r [Co(Hdmg)2] Na,[Co(pts)] [Co(mp)], [Co(pc)] and [Co(tpp)] are collected in Table 8 (pts, mp pc and tpp = tetrasulphonated phthalocyanine, mesoporphyrin phthalocyanine and tetraphenylpor-phyrin). In Figure 7 the 11Ag = l/(gim - go) values are plotted against 3 for all complexes considered. The complexes in Tables 7 and 8 are arranged roughly in order of gL and K. As can be seen from these Tables and from Figure 7 gl All A l and K vary in the same direction. According to Figure 7 a linear correlation exists between the extent of 3 d 4 s mixing (q) and l/Ag for the complexes studied all points fall on a straight line with the only exception of a [Co(pc)] [Co(tpp)] and [Co(mp)J centre.The observed linearity may be attributed to an approximately linear dependence of both parameters on the axial distortion of the local octahedral ligand field. The extent of 3 d 4 s mixing (q) depends on (3dIVtetragonal14s) and on the energy separ-ation between the 3d64s and the 3d7 configurations. For the free Corl ion this energy separation is 46 000 cm-l; 23 its value for the complexes in question may differ from this but must be approximately constant within the series. Therefore q is proportional to the tetragonal distortion of the octahedral ligand field. On the other TABLE 7 E.s.r. parameters of low-spin CoII complexes with carbon donors [(l) and (13) from ref. 22 (2)-(11) from ref.12 and (12) from ref. 151 A II Al I< gl gu kt2 -4 cm-l Complex Solvent (1) [co(cN),I~- K~[CO(CN),] 2.096 2.006 50.6 -68.4 42.3 0.763 0.136 (2) [Co(CNMe),C1,1 CH,Cl 2.090 2.006 61.4 -73.0 41.1 0.837 0.157 (3) [Co(CNMe),] 2+ MeNC 2.092 2.026 61.4 -73.2 42.4 0.838 0.157 2.0946 58.4 -66.8 37.8 0.791 0.158 2.083 b -72.0 39.4 0.802 0.154 2.0866 2.00 68.1 -68.2 35.2 0.841 0.172 2.1227 2.000 72.0 -58.5 32.7 0.862 0.182 CH,Cl 2.1276 2.0064 76.0 -49.9 26.8 0.840 0.191 (lo) [Co(CNC6H11) 51 CH,C12 2.116 2.0061 73.0 -53.3 27.9 0.831 0.187 (1 1) [CO(CNP~),]~+ CH,Cl 2.118 2.004 80.5 -49.4 23.1 0.852 0.202 (1 2) [CO (CN) 513- MeOH 2.156 2.006 81.8 -28.8 14.5 0.804 0.220 (13) [co(cN)~I~- K3[CO(CN)61 2.174 2.004 83.3 -25.4 14.4 0.820 0.213 (4) [Co(CNEt),j2+ EtNC 2.089 a -70.2 37.4 0.849 0.169 (6) [Co(CNPh),12+ Xk: (5) [Co(CNC,H,*),I2+ (7) [Co(CNPh),12+ HZO (9) [Co(CNEt),I CH,CI (8) [Co(CNMe),l,~ ‘Ai, = -23.1.‘Ais- = -26.9. Complex TABLE 8 E.s.r. parameters of low-spin Con complexes with nitrogen donors K k t 2 A A l 2.2976 2.009 103.0 8.6 3.3 0.919 9.8 0.749 MeOH 2.206 2.0161 78.0 -13.0 MeOH 2.240 2.0137 86.5 16.0 -3.9 0.716 MeOH 2.236 2.0106 102.0 28.6 -18.8 0.721 Me2S0 2.267 2.006 98.0 21.0 -9.7 0.766 MeCl 2.610 2.047 97.0 81.0 -11.9 0.806 2.646 2.029 85.0 96.0 -12.5 0.712 2.422 2.007 116.0 66.0 -21.2 0.860 160.0 265.0 -85.0 0.764 3.322 1.798 197.0 396.0 -106.0 0.917 C,H,(NO,) 3.611 1.891 179.0 426.0 -91.7 0.884 - ~ ~ _ _ Solvent gl gn cm-1 H2SO.I a-Pn (PC)l H&PP HdPP 2.900 1.91 P-[zn@c)’ 2.606 2.034 116.0 92.0 -26.0 0.834 There is a correlation in Figure 7 between the position of the individual complexes and the axial ligand field: the six-co-ordinate isonitrile complexes are in the upper left corner the five-co-ordinate species being below them.It is noteworthy that in the middle of the Figure [Co(Hdmg),] and B12r are very closely located, supporting the view that [Co(Hdmg)J may be regarded as a vitamin Bizr model compound. The lower right corner includes solid complexes with very weak axial ligand field. As the axial field in the solid state is determined by the crystal lattice the appreciable differences between various modifications can be readily interpreted. T 0.248 0.210 0.230 0.266 0.248 0.259 0.246 0.278 0.361 0.282 0.388 0.370 Ref.4 This work This work This work 21 19 18 18 20 20 21 6b hand l/(giso - go) is proportional to the 2E-2A1 energy separation. According to the ligand-field calculations of Engelhardt and Green,g the 2E-zA1 separation is large for strong axial ligand fields and vice versa i.e. the energy separation is roughly proportional to the strength of the axial ligand field. This proportionality implies a linear relation between the distortion of the octahedral ligand field and l/(giso - go). In the case of very weak axial ligand fields this probably ceases to be true, explaining the anomalous behaviour of one of the [Co(pc)] [Co(tpp)] and [Co(mp)] centres. 23 C. E. Moore ‘Atomic Energy Levels,’ Nat.Bureau Stand., 1958 1975 The data in Tables 7 and 8 show that the orbital-reduction parameter k f 2 varies in a rather narrow interval (0.712-0.862). This lends support to the correctness of the selection of the signs of the hyperfine constant and to the correct magnitude of parameters P and 1. On the other hand the near constancy of k' L (10) FIGURE 7 Correlation between 1/Ag and the extent of 3d-4s mixing for low-spin CO" complexes (the letters and numbers are those given in Tables 7 and 8) indicates that delocalization of the unpaired electron is but slightly affected by changes in the nature of the equatorial ligands. A more direct estimate for the delocalization can be obtained from the superhyperfine constants. Using the results of Maki and McGarvey 24 we obtain delocalization of €22 = 0.1 for [Co(Hdmg),(py)] and E~~ = 0.2 for [Co(Hdmg),(py),].This contradicts our observation that the reduction parameter k shows no significant differences between these two complexes. This dis-crepancy may be due either to the arbitrary assignment of the 2E-2A transition to the rather complex corre-lation 25 between the delocalization and the reduction parameter k or to enhanced delocalization of the un-paired electron onto the methanol molecule occupying the sixth co-ordination site in [Co(Hdmg),(py) 3. If one assumes that the narrowest hyperfine line in the spectrum of [Co(Hdmg),] in frozen solution is an unresolved superposition of the superhyperfine structure due to the four equatorial nitrogens an interesting correlation can be obtained between the axial and equatorial delocalization of the unpaired electron.A slight equatorial delocalization is conceivable as the d,s orbital has a small lobe in the xy plane. This assumption is supported by the fact that the shape of the fourth line in the parallel hyperfine structure can be remarkably well fitted with the envelope of a 24 A. H. Mabi and B. R. McGarvey J . Chem. Phys. 1968 29, 25 M. Gerloch and J. R. Miller Progr. Inorg. Chem. 1968,10 1. 36. 1735 1 4 10 16 19 16 10 4 1 multiplet corresponding to four equivalent nitrogen atoms (Figure 8 ) . The superhyperfine constants obtained by this fitting pro-cedure are 2.5 x lo4 cm-l for [Co(Hdmg)J and 1.75 x Consequently a higher axial delocalization is associated with a lower equatorial delocalization of the unpaired electron.The same phenomenon has been observed by Falk et aZ.26 in the case of [Cu(Hdmg)J. Interpretation of Linewidth Data.-The parameters of equation (13) proposed by Wilson and Kivelson 27 were cm-l for [Co(Hdmg),(py),J. QM = C( + PM + yM2 + 8M3 (13) fitted to the linewidth data by a least-squares procedure. Since a t low temperatures the outermost lines in the high-field part are very broad at -20 "C only seven, whereas at and below -42 "C only six linewidths were taken into account (the linewidths are neglected when OM is greater than twice the hyperfine separation). Parameters u p y and 6 and the linewidth data obtained by the simulation procedure are listed in Table 4.The evaluation is based on the formulae of Wilson and Kivelson,27 with the assumption of axial symmetry. AS gis and Aim depend on the temperature the tem-perature dependence of the anisotropic parameters f I 1 I 20 G FIGURE 8 Part of the [Co(Hdmg),] e.s.r. spectrum on frozen methanol fourth line of the cobalt hyperfine structure in parallel orientation ; the assumed but unresolved super-hyperfine structure of in-plane nitrogen atoms is also shown should also be taken into account as a correction (Table 4). Assuming that the temperature dependence of the parameters is due to changes in the excitation energy, k L - gjl>T = 1-5(gis~ - go)T (14) (Ai - A J I . = ( A - AJLN + 1-82(giso.T - giso,LN) (15) 26 R. E. Falk E. Ivanova B. Roos and T.Vanngard Inorg. 27 R. Wilson and D. Kivelson J . Chern. Phys. 1966 44 154. Chem. 1970 9 656 1736 J.C.S. Dalton equations (14) and (15) can be obtained where the sub-scripts T and LN refer to the actual and to liquid-nitrogen temperature. We assume that the Debye theory of rotational relaxation is applicable in the present case and the correlation time 713 is given by equation (16),28 where r) is the coefficient of viscosity and r the radius of the equivalent rotating sphere. The q values for methanol at different temperatures have been taken from ref. 29. It follows from equation (16) TB = 4xqr3/3kT (16) that the relaxation parameters should be proportional to q/T. This is true in the case of p and y but a reveals a minimum at ca. -20 "C. The best fit for the p values was obtained with Y = 3.6 A which also gave a satis-factory fit for the y values (Table 9).Although the fit is poorer at the highest and lowest temperatures it is still acceptable as under these conditions the hyperfine structure practically collapses into a single broad line. eorc 62 46 37 18 2 - 20 - 42 - 60 a' 14.6 17.1 18.3 21.4 26.6 39.0 64.2 83.1 molecular radius estimated from X-ray data in the case of [VO(pd)] (pd = pentane-2,4-dionate) although the Debye radius was somewhat smaller than expected. The better agreement in the present system is due probably to the fact that the conditions required for the Debye theory to be valid are satisfied to a better approximation for [Co(Hdmg),] in methanol than for [VO(pd)] in benzene.More specifically the size of the methanol molecule relative to that of [Co(Hdmg),] ensures that the solvent can be regarded as a viscous continuum whereas this is not true to the same extent for the benzene solution of [VO(pd)].32 To check the validity of the assumption underlying equation (17) we estimated the rotational-correlation 28 time from (18) where I is the molecular moment of T a = I/Sxr3r) (18) inertia. X-Ray data for [Ni(Hdmg),] 31 were used to calculate the moment of inertia for the tetragonal axis, TABLE 9 Relaxation parameters and correlation times for [Co(Hdmg),] in methanol * U" 144.7 112.0 91.0 80.4 62.6 48.1 36.3 27.0 10-7 rad s-l 153.0 116.0 98.4 65.7 46.0 28.2 15.1 11.0 P - 4.46 -5.30 -5.86 -7.11 - 9.09 - 13.5 - 23.0 - 30.4 Y 0.235 0.288 0.332 0.449 0.601 0.930 1.703 2.31 qT-1p K-1 I01%& 1.04 0.1476 1.35 0.1917 1.52 0.2213 2.08 0.2955 2.88 0.409 4.67 0.654 8.23 1.17 11.0 1.56 101%/s 9.75 7.90 7.22 5.60 4.28 2.88 1.79 1.39 * For anisotropic relaxation Y = 3.6 A and for spin-rotational relaxation Y = 3.5 %i.The calculated 6 values are smaller than lo5 s-l; conse-quently this contribution to the linewidth can be neglected. By calculating a' from the relaxation formulas z7 and subtracting it from the measured value of a one obtains the ' residual linewidth,' a". Since a" shows good linearity with T/q the major contribution to the residual linewidth is from spin-rotational relaxation.According t o Atkins and K i v e l ~ o n ~ ~ if the correlation time for re-orientation is much greater than for rotation the spin-rotational relaxation time is given approximately by equation (17) where AgIl and AgL are the differences between the corresponding g values and that for the free electron. The best fit between T2-l and a" was obtained with Y = 3.5 A (Table 9). The excellent agree-ment between the radii calculated from the two different relaxation mechanisms and the molecular radius of [Ni(Hdmg),] based on X-ray data31 supports the correctness of the Debye model in the present system. Wilson and Kivelson 27 also found satisfactory agree-ment between the effective Debye radius and the 28 N. Blombergen E. W. Purcell and R.V. Pound Phys. Rev. 1948 73 679. 2Q ' Handbook of Chemistry and Physics,' ed. R. C. Weast, The Chemical Rubber Co. Cleveland 1970-1971. 80 P. W. Atkins and D. Kivelson J . Chem. Phys. 1966,44,169. the result being I = 3.5 x g cm2. The correlation time q, obtained with this value is three to four orders of magnitude smaller than 7%. The calculated relax-ation parameters the correlation times and q/T values are listed in Table 9. EXPERIMENTAL The e.s.r. spectra were recorded on a JES-ME-3X type spectrometer in X band with a field modulation of 100 kHz. The magnetic field was measured with a JES-FC-1 proton-field calibrator unit. For the measurement of g values a Mn-MgO probe was employed. The spectra were digitalized using a JRS-5 analogue-digital converter and the calculations were made on a CDC 3300 computer. Solution e.s.r. spectra were recorded between -60 and +SO "C. The temperature was controlled by a copper-constantan thermocouple. Spectra of frozen solutions were recorded a t liquid-nitrogen temperature. As the solvent was methanol solution spectra were recorded in calibrated capillaries. The concentration of the complex was varied between 0.5 and 10 mmol ~ l m - ~ and that of pyridine in the interval 0-8.0 mol dm-3. In order to maintain a nearly identical sensitivity a t a given tem-perature the same radio-frequency power of ca. 10 mW 31 K. Nakamoto and P. J. McCarthy ' Spectroscopy and Structure of Metal Chelate Compounds,' John Wiley and Sons, Inc. New York 1968. 32 A. Spernol and K. Wirtz 2. Naturforsch. 1963 AS 522. Samples were prepared in an atmosphere of argon 1975 1737 incident on the cavity and the same modulation amplitude of 5 G was used throughout except in the case of frozen solutions.* A Mn-MgO standard was applied to correct for the sensitivity differences. Electronic absorption spectra were recorded at room temperature on a Unicam SP 700 instrument in the near-i.r. and on a Hitachi-Perkin-Elmer 124 spectrometer in the visible and U.V. regions. [3/881 Received 25th April 19731 * 1 G = 10-4 T

ISSN:1477-9226    

DOI:10.1039/DT9750001729

出版商:RSC    

年代:1975

数据来源: RSC