7669
On the Electronic Ground States of Manganocene and 1,1'-Dimethylmanganocene Mary Ellen Switzer,' Richard Wang, Michael F. Rettig,* and August H. Maki Contribution f r o m the Department of Chemistry, The University of California, Riverside, California 92502. Received April 29, 1974
Abstract: Alkyl substitution on metallocene rings generally has little effect on the chemical and magnetic properties of the metallocene. However. unlike the other first transition series metallocenes, manganocene ((V-CsHs)?Mn, or MnCp2) and I . 1'-dimethylmanganocene (Mn(Cp-CH3)2) have quite different properties. I n dilute solution in hydrocarbon solvents the magnetic susceptibility of Mn(Cp-CH3). maximizes near 310°K, while the susceptibility of magnetically dilute MnCpz follows the Curie law. The electron spin resonance spectrum of Mn(Cp-CH3)2 in hydrocarbon glasses indicates that the ground state of this d5 system a t 4.2'K is 2E2g, the same ground state as observed for isoelectronic ferricenium-like ions. In contrast. , the electron spin resonance spectrum of MnCp2 doped into MgCp2 at 4.2OK is consistent with the high-spin h A ~ ground state. An equilibrium involving doublet and sextet states is proposed to account for the anomalous magnetic susceptibility of Mn(Cp-CH&. This is the first example of high spin-low spin equilibrium in d 5 Mn(I1). Nuclear magnetic resonance results ( ' H ) for MnCp2 and Mn(Cp-CHj)z are also presented and discussed.
Manganocene ((~-CjH5)2Mn or MnCp2) is unique among the metallocene class of organometallic compounds in two respects: ( I ) the ground electronic configuration is high spin (6Alg),2while all other known MCp2 compounds are low spin; and (2) crystalline MnCp2 and its 1 ,l'-dimethyl derivative, Mn(Cp-CH3)2, behave antiferromagnetically.2.3 Our interest in MnCp2 and Mn(Cp-CH3)l was heightened when we made the unexpected observation that Mn(Cp-CH3)2 has some properties which are quite different from MnCpr. The most significant difference in the two compounds is in magnetic behavior; we have found that Mn(Cp-CH3)2 has a subnormal (for S = and temperature dependent magnetic moment even in dilute hydrocarbon solution, in contrast to the normal magnetic behavior of MnCpz under similar conditions. We report here spectral and magnetic observations on these compounds which lead to the conclusion that Mn(Cp-CH3)2 is a a-sandwich molecule existing as an equilibrium mixture of doublet and sextet electronic states. After completion of this work Orchard and coworkers obtained independent evidence for high spinlow spin equilibration in gaseous Mn(Cp-CH3)2 by means of photoelectron s p e c t r o ~ c o p yThis . ~ is the only known example of a high spin-low spin equilibrium for Mn(I1).
Results The magnetic susceptibility of Mn(Cp-CH3)2 was determined in toluene solution as a function of temperature by the Evans' nmr method. The .Mn(Cp-CH3): results for solutions 5.22 X IO-'. 2.42 X and 1.62 X IO-' M in toluene are presented in Table T. A plot of these data is presented in Figure 1. Magnetic moments of 4.8 to 5.0 BM were also observed at 30-50" for "Vin(Cp-CH3)z solutions i n benzene, cyclohexane, and mixtures of benzene and cyclohexane. Q-band (-35 G H z ) epr spectra of Mn(Cp-CHj)? were obtained at 4.2OK. The samples were studied as toluene or methylcyclohexane glasses a t concentrations near IO-' M . In either medium, the observed spectrum is characteristic of randomly oriented S = I/? molecules having an axially symmetric g tensor. A typical spectrum is shown in Figure 2. The observed g values are as follows: in methylcyclohexane g , = 2.909 and g L = 1.893; in toluene gl; = 2.887 and g L = 1.900. Q-band epr spectra were also obtained for MnCp-7 in var'
ious media, In either toluene or methylcyclohexane ( I O - ' M ), or as the neat solid, MnCpl exhibits only a single resonance near g = 2, whose line width varies with temperature. Line widths between SO and 1500 G were observed, the broader lines occurring a t lower temperatures. The epr spectra of MnCpz in glassy media were weak and featureless, consisting only of a single broad signal near g = 2. We believe that MnCp2 crystallizes from these solvents and, because of its known antiferromagnetic character. it exhibits a weak epr absorption at I .5-4.2"K. MnCpz doped (2%) into MgCpz gave very complex spectra at Q band. At 4.2'K there were four principal areas of absorption extending from 550 G (g 43) to 14 kG ( g 1.7). The 77'K MgCpz/MnCp? spectrum was similar to that at 4.2'K. Proton nmr spectra of the manganese metallocenes were obtained in benzene-dh and in cyclohexane-di2, a t 100 M H z and 310°K. In benzene-dh, MnCp2 exhibits a single resonance, with an isotropic shift of -21.8 f 0.5 ppm (to low field, referenced to FeCpz) and a line width of ca. I500 Hz. Also in benzene-ds the spectrum of Mn(Cp-CH3)2 consists of two broad overlapping peaks with isotropic shifts of -137 f 14 and -77 f 4 ppm (referenced to Fe(CpCH3)2); in cyclohexane-d I 2 these isotropic shifts are - 1 13 f 8 and -66 f 3 ppm. The uncertainties in the peak positions are due to the extreme widths of about 7 and 6 kHz, respectively. In benzene-ds, the lower field absorption for Mn(Cp-CH3)2 appeared partially resolved into two broad peaks, and the combined area of the partially resolved lower field absorptions was slightly greater than that at -77 f 4 ppm. These observations are consistent with the assignment of the lowest field absorption(s) to the C p ring hydrogens, with the attached methyl occurring as the narrow resonance at ca. - 70 ppm.
-
-
Discussion It has been known for many bears that the magnetic properties o f both MnCp2 and Mn(Cp-CH3)l are distinctly a n o m a l o ~ s . MnCp2 ~,~ is antiferromagnetic as a crystalline solid but behaves as a normal S = 5/2 d5 complex when magnetically dilute (doped into MgCp2 or dissolved in benzene).2 Mn(Cp-CH3)z also behaves antiferromagnetically both i n the crystalline solid and i n the melt, while its solutions in tetra hydrofuran have normal, temperature-independent moments appropriate for the S = /2' state.3 Wilkinson Rettig, et a/.
/ Manganocene and
1, I '-Dimethylmanganocene
7670 8600 r
a
6
10
12
14
16 KGAUS S
62CO-
Figure 2. Q-band epr spectrum of Mn(Cp-CH3)Z in glassy methylcyclohexane at 4.2'K. 200
220
240
1 260
I
200
300
TEMPERATURE
I
320
I 340
I
I
360
300
(OK)
CH3)2 is found in the electronic spectrum, where the visible absorptions, which are presumed to be d-d bands, are sixto tenfold stronger than those observed for MnCp2. All d-d transitions for MnCp2 are spin forbidden, a restriction which would not apply to a doublet state. Table I. Temperature Dependence of the Magnetic The bulk susceptibility of dilute MnCp2 in MgCpz has Susceptibilityof Mn(Cp-CH& in Toluene been determined down to 77OK,* and it is apparent from the [Mn(Cp-CHM T. lieif, bulk susceptibility of MnCp2/MgCp2 that a t 77'K there is M =K 106X11~orr a,b BMc no antiferromagnetism and the only populated state is 6 A ~ , . 5 . 2 2 X lo-* 214 5990 3.23 Our 77OK esr spectrum for MnCp2 in MgCp2 therefore 235 6680 3.55 must be the spectrum characteristic of the 6A1, state of 253 6980 3.77 MnCp2. Since there are transitions near zero field in the 285 781C 4.23 Q-band spectrum (-35 GHz), the value of D,the zero-field 303 7980 4.41 splitting parameter, must be substantially larger than those 2.42 X 310 7920 4.44 3 1C 7920 4.44 1 . 6 2 x lo-* usually observed for manganous salts, perhaps as large as 331 783C 4.58 5 . 2 2 X lo-* 0.5 cm-I. The various peaks observed in the spectrum are 334 7690 4.55 allowed transitions among the M, sublevels of 6Alg, but a 366 7580 4.73 complete assignment appears impossible without single371 762@ 4.77 crystal data. The spectrum is essentially of the same form a X A I is~ X~ ~ corrected ~I ~ for ligand diamagnetism (for this correcdown to 1.5OK, with some band shifts believed to arise from tion we used the diamagnetic susceptibility of FeCp,6 as an approxia temperature dependence of D. mation: X\r(FeCp2) = -125 X 10-6 cgs). * Estimated precision Voitlander and coworker^^^^.^ have previously investi1 2 % . p e f f = 2.83 (xxCorr T)'::;estimated, 1.0.05-0.1 BM. gated the epr of MnCp2 a t X band. The published experimental was limited to a temperature study of undiluted single crystals with the briefest mention that the g and coworkers suggested that the antiferromagnetism for both MnCp2 and Mn(Cp-CH3)z arises i n t e r m o l e ~ u l a r l y , ~ , ~values (2.01 and 1.99) are unchanged above 156" when MnCp2 is doped into MgCp2. A later theoretical interpretaand in the case of molten Mn(Cp-CH3)2 retention of some dealt only with the single-crystal results. It was concrystalline ordering even in the liquid was believed responsicluded that the observation of only one resonance near g = ble for the liquid antiferromagneti~m.~ 2 was the result of either (a) exchange averaging of the exOur observation of "antiferromagnetic" behavior for pected fine and hyperfine structure or (b) observation of Mn(Cp-CH3)2 in dilute hydrocarbon solutions could possionly the -'/z + ] / 2 transition due to a large zero-field splitbly be accounted for by oligomerization equilibria. Oligomting combined with the limitations of X-band operation. erization is ruled out, however, by the absence of concentraOur magnetically dilute Q-band results for MnCpz/MgCp2 tion dependence of the observed subnormal (for S = %) are in accord with expectation^'^^ of a large zero-field splitmoments, and the corresponding anomalous behavior of the ting. Thus Krieger and V ~ i t l a n d e robtained '~~ a theoretical susceptibility must therefore arise from intramolecular value of D = 0.24 cm-'. The transition observed .in the equilibration of electronic doublet and sextet states of present work near zero field (actually -550 G ) could either M ~ ( C P - C H ~ ) ~ .The ' - ~ high spin state is 6 A ~ ,derived'O be assigned to the allowed -3/2 (-20) or to the alfrom the configuration (e2,)2(a1,)'(e;,)2, and the potential -3 /z ( - 4 0 ) transition. This would place /Dl low spin doublet states are 2E~g((e2g)3(al,)2)or 2Alg lowed in the range 0.25-0.50 cm-'. ((e2g)4(a~s)'). Direct observation of doublet Mn(Cp-CH3)2 was accomAt room temperature, our results indicate the presence of plished with Q-band epr measurements a t 4.2'K. The specnearly equal amounts of the doublet and sextet forms of tra obtained in methylcyclohexane and toluene glasses (FigMn(Cp-CH3)2. The room-temperature solution infrared ure 2) are very similar to the spectra of 2 E ~ ferricenium, spectrum of Mn(Cp-CH3)2 is very similar to that of Fe(Cplike further, the observed spectra are not a t all CH3)2; therefore major structural changes accompanying consistent with expectations for a sextet state, as are the spin state changes are not indicated. Both high and low spin MnCpz/MgCpz results. The g anisotropy observed for states of Mn(Cp-CH3)2 are thus believed to be "sandwich" Mn(Cp-CH3)z is consistent only with the ground configuraspecies with pentahapto rings. This conclusion is reinforced tion . . . (e2J3 ( a l J 2 . . . , or 2E2g, and we may use the theoby the observation of an epr spectrum of Mn(Cp-CH3)z a t ry developed by Maki and BerryI2 to obtain bonding infor4.2OK which is very similar to that observed for isoelectronmation from the g values. It was shownI2 that if the crystalic 7r sandwich' I ferricenium-like i0ns.~~.13 Further indicaline field affecting the metal in sandwich cations departs tion of the intervention of a nonsextet state for Mn(CpFigure 1. Temperature dependence of X M ~ O " for Mn(Cp-CH3)z: 0 5.22 X A i ; 0 2.42 X M ; 0 1.62 X M.
-
-
--+
Journal of the American Chemical Society / 96:25
/ December 1 1 ,
1974
767 1 from axial symmetry (measured by a parameter 6), then the g values for the 2E2gstate are given by
g,, = 2 + 4k’(l - t’))/(l+ t’) = 45/(1
(1)
+ 5’)
(2) In eq 1 and 2, { = (6/6)/[1 11 (6/()2)’/2] with 6 = k’40, where k‘ is the orbital reduction factor for the metal e2gf orbitals and 60 is the spin-orbit coupling constant for the gaseous bare metal ion. Using the experimental g values for Mn(Cp-CH3)2 in methylcyclohexane (gil = 2.909; g i = 1.893) and setting 60 = 305 cm-’ ( a p p r ~ p r i a t efor ’ ~ Mn+, 3d54sl), we find that k’ = 0.71 and 161 = 630 cm-I. For comparison, P r i n ~ finds ’ ~ ~k’ = 0.83 and 161 = 500 cm-I for Fe(Cp-CH3)2+13- in glassy N , N - dimethylformamide; however, Prins’ may have suffered from decomposition of the cations and have been (however, see ref 13g). Rowe and M ~ C a f f e r y found ’ ~ ~ that 161 = 1043 cm-l and k’ = 0.75 for [FeCpz]BF4 in methacrylic acid. Maki and BerryI2 found ranges of 364-750 cm-’ (6) and 0.76-0.90 ( k ’ ) for carborane analogs of FeCp*+. The relatively large observed 161 for Mn(Cp-CH3)2 in the nonpolar medium suggests that the ring substituents can make a substantial contribution to the lowering of the symmetry of the ligand field. It is interesting that the orbital reduction factor for Mn(Cp-CI-13)2 is smaller than any observed for iron sandwich s p e c i e s , ’ ’ ~which ~~ could indicate that the e2gf metal orbitals of Mn are more involved in “back donation” to the C p rings than is the case in the iron compounds. Since Mn is in the zero (formal) oxidation state compared to the 1 state for the iron .compounds, greater back donation for Mn is understandable. According to the theory, the ’E2, state is actually split into two Kramers’ doublets in the iron compoundsI2 and therefore also in Mn(Cp-CH3)z. For Mn(Cp-CH3)2, the energy difference between these doublets is 2(12 6 2 ) 1 / 2 or 1340 cm-’. This may be compared to a splitting of 1200 cm-I observed for F e ( C ~ - C € i 3 ) 2 + , and ~ ~ ” 1043 cm-’ for F e C p ~ + . The l ~ ~ 1340-cm-I splitting of the two Kramers’ doublets in Mn(Cp-CH3)2 is too large to permit significant population of the higher energy doublet, even a t our highest temperature for measurement of bulk susceptibility (371 O K ) . At that temperature the population of the upper state would be only ca. 0.5% of the lower state. If we assume that a t all temperatures we are dealing with thermal equilibration of only two electronic states, namely 6 A ~and g 2E2g, then the data in Table I may be used for calculation of A H o and A S o for the reaction ~i
+ +
-’
+
IO3 x I / T
(OK)
Figure 3. Plots of log K vs. 1 / T for Mn(Cp-CH3)l in toluene (0)and for neat liquid Mn(Cp-CH3)l (0).
Table 11. Temperature Dependence of the Spin Equilibrium of Mn(Cp-CH& 214 235 25 3 285 303 331 3 34 366 371
0.791 0.721 0.669 0.551 0.501 0.452 0.461 0.407 0.395
0.209 0.279 0.331 0.449 0.499 0.548 C . 537 0.593 0.605
3.78 2.58 2.02 1.23 1 .CQ 0.825 0.855 0.686 0.653
+
Table 11, and a plot of log K vs. 1/T is shown in Figure 3. The linearity of the log K vs. 1 / T plot supports our assumption of a two-state equilibration, as in eq 3. From the plot we find AH” = -1800 f 100 cal/mol and A S “ = -5.8 f 0.6 eu. It is of interest to compare our dilute solution results to those obtained by Reynolds and Wilkinson3 for neat molten Mn(Cp-CH3)2. Applying the same assumptions to that set of data as to our own, we find for the melt that A H o = -1900 f 100 cal/mol and ASo = -6.5 f 0.6 eu (see Figure 3). Since the results for the melt are practically the same as for the dilute solution, we conclude that in both cases equilibration of doublet and sextet states is responsible for the anomalous magnetic behavior. It was found in Mn(Cp-CH,)?(‘A,,) ==== M n ( C p - CH,),(’E2J (3) the earlier work3 that Mn(Cp-CH3)’ behaves as a normal S The experimental magnetic moments (pcxp)are relatedI6 to = s/2 complex on dissolving in tetrahydrofuran. This in no the high and low spin moments by eq 4 where way conflicts with our solution results, since it is known that (4) MnCp2 (and presumably also Mn(Cp-CH3)2) is strongly Pex, = PIS?XIS + (1 - XlS)jLhS2 solvated by tetrahydrofuran;2 this solvation probably displS and k h s are the effective moments for the 2E2g and 6 A ~ g rupts the *-sandwich structure.[’ states, respectively, while X I , and ( 1 - X I , ) are the mole The observed nmr spectra of MnCp2 and Mn(Cp-CH3)l fractions of low and high spin states, respectively. Equation a t 310’K are quite different from one another, which be5 relates the experimental g values for the ZE’g state to PI,, havior is unprecedented, since methyl substitution has not previously been observed to significantly alter the net spin density a t C p hydrogens. However, as we have shown, a t 310’ there is a nearly equal population of doublet and sextet Mn(Cp-CH3)2 in solution; thus the unusually large and using methylcyclohexane g values, we find plS = 1.98 isotropic shifts of Mn(Cp-CH3)2 compared to MnCp2 are BM. Finally, we assume that p h s = 5.92 BM, which is apinitially not surprising. propriate for the 6Al, state and close to the value obtained The relatively small isotropic shift of MnCp2 converts to for MnCp2 when magnetically dilute.’ Equation 4 was solved for X I , and therefore K,, = A’lJ(1 - X I , ) a t each a hyperfine coupling constant, A H , of +0.073 MHz.~’ temperature from 214-371 OK, using the data in Table I for Since there is no reason to expect a significant dipolar (pseudocontact) shift for the 6Alg state of MnCp2, the the 5.22 X M solution. The results are presented in
’
Rettig, et al.
/ Manganocene and 1 If-Dimethylmanganocene ~
7672 0.073-MHz coupling constant must be mainly Fermi conmetry axis). For some such systems, the observed coupling tact in origin. The other paramagnetic first-row metalloconstants are as follows: VCp2, 2.33 MHz;'* CrCp2, 3.92 cenes have been f o ~ n d ' to~ have . ~ ~ electron-ring hydrogen MHz;18 bis(a-benzene)chromium(I), 9.69 M H Z , ; ' ~bis(q~ A H values as follows: VCp2 (2.33 MHz), CrCp2 (3.92 benzene)vanadium(O), 1 1.2 M H Z ; bis(7-benzene)vanadi~ ~ ~ MHz), CoCp2 (-2.39 MHz), and NiCpz (-3.53 MHz). It um( I), 3.16 M H Z ; ' ~(a-cycloheptatrienyl)(q-cyclopenta~ is apparent that the MnCp2 rest& is anomalously small, dienyl)vanadium(O), 5.0 M H z (CsHs) and 12.0 M H z only 3% of the smallest A H for the other metallocenes. Per(C7H7);26a (a-benzene)(11-cyclopentadienyl)chromium(O), haps such a finding is not surprising, in view of the long 6.58 M H z ( C j H j ) and 13.0 M H z (C6H6);26b (a-cyclopenheld notion that the bonding in MnCp2 is essentially "ionic" tadienyl)( 11-cyclohexatrienyl)chromium(I), 6.05 M H z in n a t ~ r e . However, ~.~ an alternative and more likely expla(CsHs) and 10.1 M H z (C7H7).26b In all these systems, the nation for the very small MnCp2 coupling constant would positive hyperfine coupling indicates substantial direct deloinvolve cancellation of spin density a t the C p hydrogens, calization of unpaired d22 electron density to ring hydrodue to the combined effects of unpaired D and r electron g e n ~ . ~Our ~ observation , ~ ~ . ~of ~a large ~ ~ positive ~ hyperfine delocalization. The ground configuration of MnCpz is coupling in Mn(Cp-CH& therefore suggests that a t higher . . . ( e ~ (alg)1(elg*)2 ~ ) ~ . . . , with five unpaired spins, based temperatures there is actually thermal equilibration among on earlier magnetic susceptibility data2 and our own epr rethree states: 6 A ~ , (from (e2g)2(alg)1(e;g)2);2 E ~ g(from sults. On the basis of previous work with VCp2 and (ezg)3(alg)2), and 2 A ~ g(from (e2g)4(als)'). It is the last CrCp218*19 as well as with FeCp2+,13a,2',22 the unpaired ezg configuration which is expected to lead to a large positive (d,?., dx2-?.2) electrons are expected to have only a small efA H , by analogy with the other sandwich systems above. fect on observed proton conta'ct shifts. However, the alg None of the three configurations is expected to have large (dzz) and elg* (d,,, dyz) electrons may well have compara- dipolar shifts, either because of favorable geometric factors ble contributions to A H, but of opposite sign.10~1s~19 In prin(2E2g)or because of lack of g anisotropy ((jA1,,I4 2 A ~ , 1 2 ) . ciple, these contributions may be sorted out by means of Thermal population of the ? A l gstate in ferricenium type methyl substitution, but in so doing, as has been shown ions has been suggested as one factor to consider in interabove, the problem intensifies with the appearance of doupreting the temperature dependence of pCfrfor ferricenium blet as-well as sextet Mn(Cp-CH3)2. ~ a t i o n s . ~ However, ~ ~ ~ ~ ' . Cowan ~~ and coworkers discount The nmr spectrum of Mn(Cp-CH3)z is particularly curithermal population of 'A), in their biferrocene cations, ous in view of the large observed isotropic shift of the C p since no epr absorption attributable to the ? A I g state was and CH2 hydrogens. The observed spectrum a t 310'K is observed a: high temperature, nor was there any significant presumably time averaged over nearly equal populations of deviation of the observed peff from expectations based on doublet and sextet states, and methyl substitution is not exthe low temperature (2E2g) g values.13cBoth Cowan and pected to significantly alter the spin-density distribution in coworkers13Cand Prinsl'n used photoelectron results29 for the sextet state. Further, other things being equal, the douFeCp2 to estimate that the state in FeCp2+ probably blet shifts should be only ca. 8.5% of the sextet shifts belies -2800 cm-I above the 2E2gground state, too high to cause of the much weaker magnetization of doublet affect strongly the magnetic results. These earlier estimates M ~ ( C P - C H ~Clearly, ) ~ . ~ ~the mode of electron delocalizaof S('E2, - *AI,) for FeCp2+ are in agreement with more tion cannot be the same in doublet and sextet Mn(Cprecent photoelectron spectroscopy results for F e C p ~ . ~ ~ , ~ l CH3)2; the point is that whatever the mode of delocaliiaHendrickson and coworkers28 computed similar separations tion in doublet Mn(Cp-CH3)2, it must be highly efficient to between the :AI, and 2E2glevels from the magnetic suscepproduce such large isotropic shifts in an S = '/2 system. If tibility data for [FeCp2]13 (2317 cm-') and we assume (a) that our interpretation of the bulk suscepti[FeCp2IiDDQH- (3380 cm-') (DDQH = 2,3-dichlorobility correctly implies that a t 310'K the relative popula5,6-dicyanobenzohydroquinone). Perhaps more closely retions of sextet and doublet M ~ ( C P - C H are ~ ) ~nearly equal lated to the present findings is the proposal30 that the elecand (b) that methyl substitution does not significantly alter tronic ground state of gaseous MnCp? is 2Alg.32All these the spin density a t C p hydrogens in the sextet state, then we observations suggest the possibility that the 2Al, state i n find that the C p hydrogen coupling constant in doublet Mn(Cp-CHi)r may well lie quite near the ground ?E?, Mn(Cp-CH3)2 is ca. 9 MHz. This coupling constant is two state, and our high-temperature pmr data may therefore be to four times larger than those observed for any other neureasonably explained. tral m e t a l l o ~ e n e , ~and ~ . ' ~it is about ten times larger than The intervention of a second doublet state (JA2g) at highthe largest possible Fermi contact coupling to C p hydrogen er temperatures should not seriously affect our earlier conin *Ezg ferricenium ion^.^^^,^^,^^ Both molecular 0 r b i t a l ~ ~ 3 ' ~clusions from the X M vs. T data, since the expected mag( M O ) and e ~ p e r i m e n t a l l ~ ~results ~ I ~ -suggest ~ ~ ~ ~ ~ that a netic moment of the '&% state should be close to the spin ?Elg configuration is unlikely to produce a C p hydrogen only value of 1.73 BM, which is coincidentally similar to the value of 1.98 BM derived for the 2E2, state.33 coupling constant of anything like +9 MHz. In fact, the MO results predict negative coupling via ezg d,-p, back Conclusion donation.10.19The possibility of a large pseudocontact contribution to the 9-MHz coupling constant is unlikely. The The magnetic susceptibility and epr results reported here Mn ring centroid distance in MnCp2 has been determined are fully consistent with equilibration of high spin (S = 5/2) to be 2.05 8, by electron diffracti01-1.~~ If we assume a slight and low spin (5' = I/?) forms of the r-sandwich molecule shortening to 2.0 A in doublet Mn(Cp-CH3)l and use our Mn(Cp-CH3)2. There is no indication of any such equiliexperimental g values along with Prins' geometric facbration of MnCpr either dissolved in organic solvents2 or tors,13" we find that the expected dipolar coupling would be doped into MgCpz, even down to 1.5'K. However, it seems approximately 0.30 and -0.16 M H z for the C p ring hydrolikely that in the condensed phase MnCp2 (the only "highgen and the methyl hydrogen, respectively. spin" metallocene) is itself close to a high spin-low spin Previous nmr and epr studies of sandwich molecules and crossover, since small perturbations such as volatilizaions have shown that those systems having (at least) an untion30.32or addition of two methyl groups lead to doublet paired electron in the alg orbital exhibit large positive hyground states. iMethyl groups are ordinarily considered to be electron releasing compared to hydrogen, and this eviperfine coupling to the ring hydrogens (the alg orbital is derived from d,?, which is directed along the principal symdently is important enough in Mn(Cp-CH3)r to significantJournal of the American Chemical Society
/ 96.25 / December
11, 1974
1673
ly increase the donor character of the ligands and cause spin pairing. Evidence of increased ring electron density in Fe(Cp-CH3)2 compared to FeCp2 is found in their respective behavior toward oxidation. Fe(Cp-CH3)? is more readily oxidized, presumably because of the electron-releasing character of the substituent methyl g r o ~ p s . ~ , ~ ~ The pairing energies for d4, dS, and dh are ~ i m i l a r and ,~ both the d4 (CrCpz) and dh (FeCp2) metallocenes are low spin. Since MnCpz seems very near a magnetic crossover point, we conclude that the overall d-orbital splittings in MnCp2 are similar to its neighbors, Thus, although MnCp? behaves ionically in that it reacts rapidly with ferrous salts,2 dissociates to give conducting solutions,? and exchanges rings instantaneously with L ~ C S D our ~ , ~results ~ suggest that the overall degree of M-Cp orbital mixing is similar for Mn and for Fe, a conclusion which is not surprising since the electronegativities of Mn and Fe are practically the same, regardless of the scale used. In this connection we note that Orchard and coworker^,^ who have provided independent evidence supporting the high spin-low spin equilibration in Mn(Cp-CH3)Z (from the photoelectron spectrum), have observed that the ligand field splitting parameters deduced for FeCpz should lead to a low-spin ground state for MnCp2, if such parameters were to be directly transferred. Finally, our results, together with the photoelectron results for gaseous MnCp?,30s32raise the possibility that the crystalline antiferromagnetism of both MnCpz and Mn(Cp-CH3), long believed to be intermolecular in origin, actually may arise intramolecularly. Experimental Section Instrumentation. L'v-visible spectra were recorded on a Cary 14 spectrophotometer. Ir spectra were recorded on a Perkin-Elmer iModel 621 grating spectrophotometer. A Varian HA-IO0 nmr spectrometer with an external Hewlett-Packard X Y recorder was used to record the nmr spectra of the metallocenes; a Varian A60D nmr spectrometer was used for the determination of the magnetic moments. Epr spectra were recorded at Q band. The field was calibrated using an nmr gaussmeter employing a D2O sample. The frequency was measured with a Hewlett-Packard 5 2 4 5 1 frequency counter. The sample and cavity were immersed directly in liquid He during measurements. Air-sensitive compounds were generally handled and stored in a Vacuum-Atmospheres Company Dri-Train dry box. Solvents. For syntheses and magnetic moment measurements, reagent-grade solvents were degassed with prepurified nitrogen and dried with Linde 3A or 5A molecular sieves. For infrared and nnir spectra, spectral grade solvents were treated similarly. Methylcyclohexane for the epr and uv-visible studies was stirred for several hours with sulfuric acid. washed with distilled water, and distilled from BaO under nitrogen; toluene for the epr study was degassed with prepurified nitrogen and dried with Linde 3A molecular sieves. Synthesis of Compounds. MnCp? was prepared from MnBrz and N a C p by the published method.* MnCp2 was isolated by sublimation at 200' (0.5 mm) from the reaction vessel and freshly resublimed at 80' (0.5 mm) just before use. M n ( C p - C H 3 ) 2was prepared from N a ( C p - C H I ) and MnBr2 in T H F essentially by the method of Reynolds and Wilkinson.3 After removal of the solvent, the .Mn(Cp-CH3)2 was extracted into pentane. and the pentane was evaporated. I f the product did not crystalize spontaneously, i t was vacuum distilled a t C-Br > C CI. These species a r e formed in the absence of normal stabilizing ligands and their stabilities vary greatly depending on the organic group attached. For example ChFjPdBr is stable a t room temperature in the air whereas CH3Pdl decomposes a t extremely low temperature under vacuum. The stability order is C6Fs > C F 3 C F 3 C F 2 CF3CF2CF2 > C h H j >> CF2Br > (CF3)zCF CH3 CH3CH2. The ArPdX species decompose thermally to Ar-Ar, Pd,. and PdBrz. Some of the RPdX species decompose free radically at low temperature. T h e R C O P d X species decompose by decarbonylation and have stabilities in the order of R = CF3CF2CF2 > CF3 > C s F j > CsHs > CH3 C H ~ C H ~ C HThese Z . intermediates, ArPdX, RPdX, and RCOPdX, a r e highly coordinatively unsaturated and thus extremely reactive with phosphines. I n fact many or these materials can be readily trapped with tertiary phosphines to form the bisphosphine adducts (RAP)>Pd(Ar)X, =Pd(R)X, and in some cases =Pd(RCO)X. This serves as a rapid method of preparation of these complexes and in some cases quite high yields of these materials can be prepared by simultaneous deposition of palladium atoms, ArX, and Et3P a t -1 96'.
-
- -
-
Simple organopalladium salts ArPdX and RCOPdX (no stabilizing ligands) have been proposed as intermediates in a number of important reactions. Heck,) Henry,4 and Mait l i ~ formulate ,~ "phenylpalladium chloride (C6HSPdCl)" as the reactive intermediate formed in solutions of palladium dichloride with an arylating agent such as phenylmercuric chloride. These solutions have been used for olefin arylations, alkylations, carboalkylations, etc. Similarly, Henry4 and Tsuji and coworkersh discuss RCOPdCl as a likely short-lived species in palladium catalyzed CO insertion Journal of the American Chemical Society
-
reactions, olefin carbonylation reactions. and reaction mechanisms relating to the Rosenmund reductiom6 Furthermore, Maitlis and coworkers' have reported on interesting chemistry of "lightly stabilized" organopalladium compounds (very weak donating ligands). I n light of these works, we thought it would be quite interesting to produce and study "nonligand stabilized" organopalladium compounds (no donating ligands a t all). In this paper we describe the synthesis, isolation, trapping and other studies of a series of nonligand stabilized
/ 96.25 / December 1 I ,
1974
