5760
Stereochemical Rigidity in ML, Complexes. 111. Detailed Line Shape Analysis and Rearrangement Barriers in Pentakisphosphite Complexes of Cobalt( I), Rhodium( I), Iridium( I), Nickel( II), Palladiuin( II), and Platinum( 11) J. P. Jesson" and P. Meakin Contribution No. 2076 f r o m the Central Research Department, E. I. du Pont de Nemours & Compaiiy, Experimental Station, Wilmington, Delaware 19898. Receiued December 8, 1973 Abstract: An nmr investigation has been carried out for a class of cationic MLScomplexes (M = Co, Rh, Ir, Ni, Pd, and Pt, and L = phosphite). More than 20 complexes were studied, all of which are stereochemically nonrigid
at ambient temperatures and stereochemically rigid on the nmr time scale at low temperatures. The rearrangements are shown to be intramolecular by observation of retention of metal-phosphorus coupling in the high temperature limit for the AsB3X spin systems and by observing the invariance of the spectra to excess ligand for the A2B3spin systems. A series of complete density matrix calculations for intramolecular exchange have been carried out, and the permutational nature of the exchange has been established for specific complexes of Pd(Il), Ni(II), Co(I), and Rh(1); in all cases the rearrangements involve simultaneous exchange of the axial ligands with a pair of equatorial ligands in the trigonal bipyramidal ground state structure and are therefore consistent with the Berry process. Barriers lie within the range 5-12 kcal mol-' and increase with increasing steric bulk of the phosphite ligand up to the point where the bulk is too great for MLs complexes to be formed (i.e., where the ligand dissociation equilibrium ML5e MLI L lies well to the right). For a given ligand, barriers are relatively insensitive to variation of the central metal although the ordering Co > Ir 2 Ni > Rh > Pt > Pd can be established; thus sequences Co > Rh < Ir and Ni > Pd < Pt apply and the members of the first triad have uniformly higher barriers than those of the corresponding members of the second triad.
+
I
n earlier papers in this series, we have presented the first evidence for the stereochemical rigidity of In the present paper, we ML, species in describe a n nmr study of the intramolecular exchange processes in ML, cations for the six metals (Co, Rh, Ir, Ni, Pd, Pt). At low temperatures they all show A2B3 or A2B3X 31P{1H/nmr patterns. As the temperature is raised, the spectra broaden and coalesce into a single line (A, spectrum) or a doublet (A part of A5X spectrum). Maintenance of A-X coupling (A5X cases) and invariance to added ligand (A, cases) show that the rearrangements are intramolecular. The variation of chemical shift with temperature noted in some species1'? and thought possibly to arise from ion pairing effects is found not to be a general feature of the spectra of these complexes. Concentration, solvent, and anion variation studies eliminate ion pairing as an important factor in the dynamic behavior of the cations. Activation parameters for rearrangements have been determined for the majority of the complexes and mechanistic studies have been carried out for Co[P(OCH2)3CCH3]5+and Ni[P(OCH,),CC2H,];,2+ using a complete density matrix treatment; the results show the same basic permutational behavior established for Rh[P(OCH3)3],L earlier? and exclude any physical mechanisms which involve single axial-equatorial exchanges. They are consistent with the Berry process5 as well as some other proposed mechanisms such as the turnstile rotation.6 ( 1 ) J. P. JessonandP. Meakin,J. Amer. Chem. Soc., 95,1344(1973). (2) Paper I in this series: P. Meakin and J. P. Jesson, J. Amer. Chem. Soc., 95, 7272 (1973). (3) J. P. Jesson and P. Meakin, J . Iizorg. Nucl. Chem. Lett., 9, 1221 (1973). (4) Paper I1 in this series: P. Meakin and J. P. Jesson, J . Amer. Chent. Soc., 96,5751 (1974). ( 5 ) R. S . Berry, J. Chem. Phys., 32, 933 (1960).
Journal oj'the American Chemical Society
/
96:18
The barriers to rearrangement range from 6 to 12 kcal mol-' and increase with increasing ligand steric bulk for a given metal; the relationships Co > R h < Ir and Ni > Pd < Pt are established for variation of the central metal for a given ligand, with the members of the cobalt triad having uniformly higher barriers than the corresponding members of the nickel triad. Experimental Section The preparation and initial studies of the intermolecular exchange behavior of the complexes investigated in this study have been described.' In a number of cases it proved convenient to prepare complexes in situ in nmr tubes. The 31P( 'HI and I3C(IH] nmr spectra were obtained using a Bruker HFX 90-Digilab FTS/NMR-3 system as previously described.2 The line shape calculations use density matrix methods',* coupled with group theoretical methods of permutational analysis and computer techniques for symmetry factoring.*,9,'0 Results Low temperature limit nmr spectral parameters for all compounds considered in this section are given in the preceding publication. A. Rhodium. The low temperature limit spectra are A2BaX patterns showing that the complexes have Dah symmetry (configuration 1) on the nmr time scale, 2a4 Detailed analyses of the temperature dependence of the 31P( 1H) spectra for a range of RhLj+ complexes have been presented earlier. Data for one (6) I. Ugi, D. Marguarding, H. Klusacek, P. Gillespie, and F. Ramirez, Accounts Chem. Res., 4,288 (1971). (7) J. I. Kaplan, J . Chem. Phys., 28,278 (1958); 29,462 (1958). (8) S. Alexander, J . Chem. Phys., 37, 967, 974 (1962); 38. 1787 (1963); 40,2741 (1964). (9) P. Meakin, E. L. Muetterties. F. N. Tcbbe, and P. J. Jesson, J . Amer. Chem. Soc., 93,4701 (1971). (10) J. P. Jesson and P. Meakin, Accounts Chem. Res., 6,269 (1973).
1 September 4, 1974
5761
PA
Co[P(OCpH,b]lCLOBSERVED
CALCUUTED
PA
1
additional RhLj+ compound have been obtained since the original publication and these are described below. (i) Rh(phosph~radamantane)~+B(C~H~)~-. This complex (phosphoradamantane = 2,8,9-trioxaphosphaadamantane (configuration 2)) was prepared in situ P
-93'
1(
1.4PPM
E
h 2
in the nmr tube.4 The temperature dependence of the chemical shift separation is quite small but larger than for complexes with the more rigid sterically less bulky ligands of the type P(OCH2),CR. For Rh(phosphoradamantane)a+B(C6HS)4-the spin system is more tightly coupled (IJABI/I~A - 8 ~ is1 larger) than for other RhLj+ cations which we have investigated* giving rise to a less easily recognized A2B3Xspectrum. * From an nmr line shape analysis a rate of 75 sec-' is obtained at -64" corresponding to a free energy of activation of 10.25 kcal mol-' at this temperature. An exchange process involving simultaneous exchange of the two axial ligands with two of the equatorial ligands was assumed. B. Cobalt. (i) CO[P(OC~H~)~]~+CI-. l1 The temperature dependence of 31P{1H}nmr spectra of Co[P(OC2Hj)3]j+Cl- in CHClF2 solution is shown in Figure 1, together with a set of calculated spectra. The calculated spectra were obtained using an A2B3 model in which the only parameter varied was the chemical shift separation between the A and B nuclei. It can be seen that the calculations give excellent agreement with the experimental data. Superficially the observed temperature-dependent spectra appear to correspond to a classical line shape variation normally associated with chemical exchange (broadening of a sharp, complex, low temperature spectrum with increasing temperature and finally, at higher temperatures, collapse into a single line) but the simulation is obtained without assuming exchange. This is a fortuitous case in which the chemical shift variation with temperature is large and the magnitude of the shift goes from an easily measurable value to a very small value over the temperature range studied. The example indicates the danger of assuming the presence of an exchange process on the basis of qualitative line shape effects; the good agreement with calculation indicates that the cation has D3h symmetry on the nmr time scale over essentially the whole temperature range shown in Figure 1. Exchange almost certainly occurs when the temperature is raised still further (it would be difficult t o detect with the almost zero chemical shift separation) but there is no clear evidence for exchange on the basis of the data in Figure 1. In obtaining the calculated spectra the AB (11) We are indebted to Dr. L. W. Gosser for a sample of this com-
plex.
-105'
),
n
il ,'L
(
1.9 PPM
n
1
11
I;;r
Figure 1. The temperature-dependent Fourier mode 31P('H}nmr ~ +CHC1F2. C~The only spectra for a solution of C O [ P ( O C ~ H ~ ) ~ ] in parameter varied to obtain the corresponding calculated spectra was the chemical shift difference between the axial and equatorial phosphorus sites. This chemical shift separation is given with the simulated spectra.
coupling constant was kept fixed at 148 Hz (the value obtained by fitting the -138" spectrum) and the AB chemical shift separation was varied to give a good fit to the observed spectra. A case similar to the one discussed above, where qualitatively there is ambiguity as to whether chemical shift variation or exchange behavior accounts for the spectral variation with temperature, is discussed in the section dealing with iridium complexes; for the iridium species Ir[P(OCH2)3CCH3]j+ the calculations show that in this case exchange processes account for the line shape behavior. Before considering other cobalt complexes, one additional feature in Figure 1 deserves mention; in the lower temperdture range there is some disagreement between the observed and calculated spectra. This arises because in the experimental spectrum the lines largely associated with the equatorial phosphorus nuclei (low field section of spectrum) are broader than those largely associated with the axial phosphorus nuclei. Effects of this type were observed only for the CoL6+complexes and are attributed to partial decoupling of the 31P-6gC0 spin-spin coupling resulting from the rapid quadrupole relaxation of the j9C0nucleus with spin Z = 7 / 2 . At low temperature (- 138") the quadrupole relaxation is Jesson, Meakin
Stereochemical Rigidity in ML5 Complexes
5762
Figure 2. Low temperature limit Fourier mode 31P('H) nmrspectrum for a solution of Co[P(0CH*)3CCHJ5+B(C6Hj)(in chlorodifluoromethane. The calculated spectrum was obtained using an A2B3model.
very effective (the rotational correlation times are long because of the relatively high viscosity and low temperature); as a result, the decoupling is almost complete and the line width variations across the spectrum are small. As the temperature is raised the line shape effects due to quadrupole relaxation become more pronounced. The lines to low field are broader because J p n c o > J p l c 0 (configuration 1). Above the temperature range shown in Figure 1, the spectrum broadens again as the viscosity of the solution decreases and the quadrupole decoupling of the 'P-59Co spin-spin coupling becomes less effective. (ii) CO[P(OCH~)~]~+B(C~H~)~-. The temperaturedependent 31P{'H) nmr spectra of this complex in chlorodifluoromethane over the temperature range -140 to -70" are very similar to those shown in Figure 1 except that now the chemical shift associated with the equatorial phosphorus nuclei is upfield. The 31P('H) spectrum at - 138" is almost a mirror image of the low temperature limit spectrum for Co[P(OC2H&I5+(Figure 1). Again, the chemical shift difference decreases in magnitude as the temperature is increased and the spectrum coalesces into a single line; there is no evidence for an intramolecular exchange process in these spectra. The lines to the upfield side of the spectrum are broader than those to the downfield side indicating (see section B (i)) that, as for Co[P(OC2Hj)3]j+,Jruc0> JpAco(configuration 1). (iii) Co[P( OCH2)3CCH3]j +B(C,3H5)4-. The low temperature limit (- 122") 3'P{ 'H) nmr spectrum for a solution of this complex in chlorodifluoromethane is shown in Figure 2 together with a spectrum simulated using an A2B3 model. The chemical shift of the equatorial phosphorus nuclei is to high field and the chemical shift difference is almost temperature independent. The relatively large chemical shift difference is an additional advantage in establishing intramolecular exchange. As the temperature is raised, two types of line shape effects can be observed (Figure 3): (a) the quadrupole relaxation of the S9C0 nucleus becomes less effective and the upfield side of the spectrum begins to broaden more than the downfield side (again J p i r C o> J P . ~ c M~ = ; Co, configuration 1); (b) simul-
Figure 3. Temperature-dependent Fourier mode 31P{'H)nmr in chlorospectra for a solution of CO[P(OCH~)~CCH~]~+B(CBH~)~difluoromethane. The calculated spectra were obtained using permutational mechanism A .
taneously, line shape effects can be observed that are a result of a mutual exchange process. Figure 3 com. pares the observed temperature dependence of the nmr spectrum with that simulated using a complete density matrix line shape analysis. In the simulations a permutational mechanism, A , which simultaneously switches the two axial ligands with two of the equatorial ligands was used. This gives a better fit to the observed spectra than the other possible permutational mechanism, B, which involves single axial equatorial switches (see the Discussion section). The spectra in Figure 3 (6/J = 5.1) are very similar to those observed for Ni[P(OCH2)3CC2H5]s2+(BFr-)2 (S/J = 4.74); for the latter molecule a detailed comparison has been made between the observed spectra and those simulated for both A and B permutational exchanges (vide infra). A comparison of these simulations with the spectra of Co[P(OCH&CCH3]5+B(C6H5)4shown in Figure 3, and with other experimental spectra not shown, clearly indicates that the simultaneous exchange mechanism must predominate for both the Ni2+and the Co+ complexes. Due to the quadrupolar relaxation effects the fit obtained between the observed and calculated spectra shown in Figure 3 is not as good as for the other ML6+ systems we have analyzed. The rate data are presented in the form of an Arrhenius plot in Figure 4; the straight line corresponding to the rate expression rate(T) = 10 15.28e-l2,60O/RT The activation parameters at 200°K are given in Table 11. The apparent entropy of activation is rather large
Journal qf'the American Chemical Society / 96:18 / September 4 , I974
5763 5.2
-
5.0
-
4.8
-
I 03 T
CALCULATED
1
4'41 4.2
\ 2.0 3.0 4.0 I.o
400
LOGIOIRATE)
Figure 4. Arrhenius plot for CO[P(OCH~)~CCH&,+B(CP,H&-
for a simple intramolecular process. This is probably a result of experimental error arising from the limited temperature range over which the measurements were made (the solvent boils at -40") and the errors introduced by the quadrupole relaxation of the 59C0 nucleus. In a separate experiment the 31P{ 'H) spectra of a solution of C O [ P ( O C H ~ ) ~ C C H ~ ] ~ + B ( C with ~ Hligand ~)~added were recorded over the temperature range - 140 to -20" in chlorodifluoromethane. The line shapes of the spectrum assigned to the complex were invariant to added ligand, and the ligand resonance remained a sharp single line at all temperatures showing that the exchange process is intramolecular in the -140 to -20" range. C. Iridium. (i) Ir[P(OCH3)3]6+B(C6H6)4-.The temperature-dependent 3lP{lH ] nmr spectrum for a solution of Ir[P(OCH3)3]5+B(C6H6)4-in CHCIFz is shown in Figure 5. The simulated spectra shown in this figure were obtained using an A& model assuming mechanism A . Since the spectra are almost first order, no distinction can be made in practice between mechanisms A and B. (This distinction could be made by observing the spectra at a lower field such that higher order effects are present in the slow exchange spectrum.) The chemical shift difference is quite strongly temperature dependent (approximately - 5.0 Hz deg-l or -0.15 ppm deg-l over the temperature range - 100 to - 150"). Consequently an extrapolation of the temperature dependence of the shift difference into the temperature range where the shift cannot be measured directly was used to obtain some of the simulated spectra shown in Figure 5. The large shift difference makes the percentage error introduced by the extrapolation small, and accurate exchange rates can still be obtained even with a rather poor extrapolation. The spectra are invariant to added ligand over the temperature range shown in Figure 5; the ligand resonance remains a sharp single line. This observation is taken to indicate that dissociative processes are slow at these temperatures and that the observed line shape effects can be attributed to an intramolecular process. From an Arrhenius plot of the data shown in Figure 5 the rate expression rate(T) = 1013.2e--8500/RTset-1
Figure 5. The temperature-dependent Fourier mode 31P{ IH)nmr spectra for a solution of Ir[P(OCH,),]6+B(CsH,)4- in CHCIFz together with spectra simulated using an AzB3model and the density matrix formalism.
was obtained. The activation parameters are given in Table 11. (ii) Ir[P(OC2H5)3]5+B(C6H6)4-.From a complete line shape analysis, we obtained an exchange rate of 75 sec-l at -60". Substituted into the Eyring equation this result gives a free energy of activation (AG*)of 10.4 kcal mol-'. (iii) Ir[P(O-n-C4H9)3]5+B(C6H5)4-. The exchange rate at -42" is 60 sec-l. This corresponds to a free energy of activation of 11.5 kcal mol-' at this temperature. (iv) Ir[P(OCH2)aCCH3]6+B(C6H5)4-.The temperature-dependent 31P( 'H) nmr spectrum for a solution of this complex in chlorodifluoromethane is shown in Figure 6 . In this case, the chemical shift difference at low temperatures (slow exchange rates) is essentially temperature independent. In Figure 6 the observed spectra (center of figure) are compared with spectra Jesson, Meakin 1 Stereochemical Rigidity in MLs Complexes
5764 I r [P(OCH,),CCH3]; CHEMICAL SHIFT
OBSERVED
Ni [P(OCH31J] + + 2SbF6-
B(C,H&
5
EXCHANGE RATE
OBSERVED - 120"
CALCULATED
Figure 7. Observed and calculated low temperature limit proton decoupled 31P nmr spectrum for a solution of Ni[P(OCH3)3]5*+(SbFe-)%in chlorodifluoromethane. An AzB3 model was used to compute the simulated spectrum.
correspond to the Arrhenius expression rate(T) = 1013.5e-92001RT
Figure 6. Temperature-dependent Fourier mode 31P(lH 1 nmr spectrum for a solution of Ir[P(OCH2)3CCH3]
>
equal within experimental error. The AG* values are accurate to ca. 5 0 . 2 kcal mol-' so that the following meaningful trends can be established from the data. (i) There is a steady increase in the barrier with increasing steric bulk of the ligand. This is best exemplified by the Rh(1) and Ir(1) series but is consistent also with the data on Pd(I1) and Pt(I1). The result suggests steric crowding in the transition state relative to the ground state as a principal factor in determining the barrier height. Support for this concept of steric crowding is obtained from the synthetic work4 which shows that for sufficiently bulky ligands the Dahground state can become so unstable that ligand dissociation according to eq 7 occurs to the point where the equilibMLj
MLI
+L
(7)
rium is far to the right. Using the ligand cone angle concept l 7 as a measure of steric bulk, it appears to be difficult to make ML, complexes with sizes of the order of P(OCsH5)3(cone angle 128') or bigger. (ii) For fixed ligand, the central member of a vertical triad has a lower barrier than the upper or lower members and Co(1) > Ir(1) > Rh(1); Ni(I1) > Pt(I1) > Pd(I1) (see Table 111). The data in Table I11 represent the only complete set, but the trend is supported by other cross comparisons which can be made in Table I1 and is different from that observed in hydrides of the form HM(PF3)?where'there is a steady increase of barrier on going down a triad.I4 The trend in barriers down a triad may be tentatively interpreted in terms of a stronger tendency to four us. five coordination for the central member; the lower barrier would then reflect a weaker metal to phosphorus bond either in terms of a lower bending force constant or a longer M-P bond or both; these factors would render the intramolecular rearrangement more facile. This basic picture is supported by preliminary data on intermolecular exchange which suggests that the AGi for intramolecular exchange may correlate directly with AGt* for intermolecular exchange. It is tempting to speculate that for fixed ligand AGt* is smaller for smaller AGi and that for fixed metal AGt is smaller The intermolecular exchange studies for larger AGi are not straightforward, and more work is required to definitively establish these suggested trends. (iii) There is a decrease in barrier on going from a metal of the cobalt triad to the corresponding member of the nickel triad for a given ligand (see Table 111). Again, this is opposite to the trend observed for HM(PF& species on passing from the iron to the cobalt triad (in fact our initial studies of ML5 species were partially directed by the assumption, based on the HM(PF3)A results, that increasing positive charge on the central metal would increase the barrier). These results for ML, species may, however, be rationalized in a similar manner to the results in (ii).
*
*
*.
*
(17) C. A. Tolman, J . Amer. Chem. Soc., 92,2956 (1970).
Ir[P(OCH2)3CCH315+ 8.4
