Interactions of Carbonyl Groups in Compounds ContainingMetal-Metal BondsBY M. L. N. REDDY AND D. S. URCHChemistry Department, Queen Mary College, Mile End, London E.l.Received 20th February, 1969Infra-red spectra of compounds of the type ((CO), . Fe[X . CsYs])2, (X = S, Se or Te; Y = H orF), have been studied in the carbonyl stretching region. Four bands were observed. The inter-actions between the carbonyl groups were assumed to take place via iron 3d orbitals. A model todescribe the spectra was developed in which a three-fold axis was retained at each Fe(CO)3 groupand between the two iron atoms. It was concluded that the carbonyl-carbonyl interaction routebetween the iron atoms was via direct 3d-3d overlap and not via the d or p orbitals of X. Interactionforce constants were calculated and their variation with X and Y discussed. The model was extendedto explain the spectra of related compounds.Compounds of general formula (CO),Fe(X.c6Y5)(x'. c6Y5). Fe . (CO),,(X,X = S , Se or Te; Y = H or F) can be prepared by reacting xZ(c6Y5)2 with ironcarbonyls.' The basic structure is shown in fig. 1. The diamagnetism of these com-pounds may be explained by postulating a " bent " iron-bond. The t2s orbitals (assum-ing, for descriptive purposes Oh symmetry at each iron) on iron atoms are completelyFIG. 1.filled and presumably " back-bond " to the empty anti-bonding orbitals on the carbonylgroups. Since these 3d orbitals will also interact with d andp orbitals on the X atomsof the bridge and also with each other, two possible routes are available for the car-bony1 groups at the ends of the molecule to interact with each other.Carbonyl stretching infra-red spectra are shown diagrammatically in fig.2 foreleven compounds. Quantitative data for these and some related compounds areshown in table 1. Certain interesting features may be noted. The substitution ofphenyl by perfluorophenyl groups causes a general shift to longer wave numbers; thedifference between the average frequency of the two high frequency and the two lowfrequency bands (column A in table 1) is almost constant (-55 cm-I); the highestfrequency band is the weakest, the remainder are all strong; the two lowest frequencybands are often close together; also, four bands are observed.The moleculesbasically have C,, symmetry, in which case five bands might have been expected. On5TABLE 1principal carbonyl stretching frequencies (cm-1)V1 (moderate)2073207320782079208720752073208 1206920752082206620722080206620562070206220602055(CO), . co- ‘kb(co)3low temp. formy2 (v. strong)2036203820412045205820382037204220322041205520292038205 12033201920412026v 3 (strong)200320032009200820232003ZOO0200519962004202420002000201619891978200719962020 1998201 8 19942071 2004v 4 (strong)1995200120042010199219911990200220081988200419691997198819791966L---2042vl+vz2054205520592062207220562055206 12050205820682047205520652049203720552044204020362209 1y3+yq1999200320052006201619971995200519932003201 61994200020101989197320021992198819802or v 32043A555254565659605657555253555560645352525648*this work: A = c1;v2)-c3:v4) M.L. N. REDDY AND D. S. URCH 55the other hand, the molecule might retain much three-fold symmetry so that for thecarbonyl-carbonyl interactions the effective point group would be C3" (if no inter-action across the iron-iron bond) or DSk. In the former two and in the latter, threebands should be observed.I I I II I II2 0 8 0 2 0 6 0 2040 2 0 2 0 2000 1960cm-1FIG. 2.-Carbonyl stretching frequencies.MODELThe main features of the observed spectra may be explained if it is postulated thatthe CO stretching force-constant for carbonyl groups trans to the X atoms is different(k,) from the carbonyl groups trans to the iron-iron bond (k,) and also that local C3"symmetry is retained at each Fe(CO), site.Assuming that it is possible to treat thecarbonyl stretching modes independently of any other vibrations in the molecule, thenthe following normal coordinate functions can be formulated:al = (3)-0.5.c12 = ( 3 ) - 0 ' 5 .a3 = (2)-Oa5.a4 = (21-0-5.a5 = (6)-O*'.a6 = ( 6 ) - O S 5 .[CO( 1) + CO(3) + CO(5)],[ C W ) - COC5)],ICO(4) - C0(6)1,[C0(2) + CO(4) + CO(6)],[2 . CO(1) - CO(3) - CO(5)],[ 2 . C0(2)-C0(4)-C0(6)].al and a2 will both belong to representations a, in CJV symmetry and (a3, a,) and(a4, a6) will have representations e.In order to consider the interactions between the carbonyl groups in a morequantitative way, the approximate method of Cotton and Kraihanzel may be used.First, let us take the interactions of the carbonyl groups attached to the same ironatom and let the perturbation force constant for two carbonyl groups cis to eachother be k,.The appropriate force constants for the normal coordinate functions arexi, a2 : +(k, + 2kx + 6kc) = +ky + +kx + 2kc~ 3 , a4 : H2kX-2kc) = kx-k,U S , a6 : &(4ky+2k,-6kc) = 3k,++kx-kc.Next, carbonyl-carbonyl interactions across the Fe(X.C 6Y 5)2Fe region can be con-sidered. The simplest assumption is that the interaction between any two carbonylgroups across the bridge region gives rise to a perturbation force constant of kd,irrespective of the relative positions of the carbonyl groups. This is equivalent t56 INTERACTIONS OF CARBONYL GROUPSassuming that those orbitals that carry the interactions have threefold symmetry andthat, to a first approximation, for the carbonyl groups, the system has DJh symmetry.Normal coordinate functions for all six carbonyl groups can now be constructedand their force constants written down.The only effect of this '' bridge " interaction is to cause a big splitting of forceconstants for (a, +a2) and (a, -a2), whereas all other force constants are unaltered.Thus, four distinct force constants are found.If the symmetries of the correspondingfunctions are determined it will be possible to determine which will give rise to inf'ra-red active frequencies.Although D3h symmetry was assumed, the actual symmetryof the molecules is CZ0 or less. Table 2 therefore shows which functions would beinfra-red active (indicated by *) in both these symmetry groups. If it is assumed thatthe distortion from DSh to cZu symmetry is small, then a frequency forbidden in D3hbut permitted in Cz, would be expected to be weak whilst frequencies permitted underboth symmetries would be strong. The various stages in the development of thismodel axe summarized diagrammatically in fig. 3.TABLE 2functions irreducible representationsD3h c2uDISCUSSIONThis model explains the observation of four bands in the carbonyl-stretching regionand also why the highest frequency band should be weaker than the others. Theother special features of the spectra may also be rationalized.The difference betweenthe average of the two high and the two low frequency lines is related to the cisperturbation force constant only (3k,). Changing the nature of X or Y should notchange this factor greatly and so it is reasonable that the difference should be more orless constant. The relation between force-constants (mdynes A-') and frequency(cm-l) is k = v2 4-0383 x Thus, k, is 0.31 mdynes A-l, comparable to thevalues given by Cotton and Kraihanzel.' The observation of only four lines in theinfra-red spectra is a direct consequence of the coefficients used for the functions a3---.6.Since these coefficients result from the assumption of local C,,symmetryat the Fe(C0)3sites, the deviation from this symmetry caused by postulating different force constantsk, and k, is slight. This, in turn, suggests that the two low frequency Lines should beclose together as is observedM . L . N. REDDY AND D . S . URCH2 000-2020-57X-Y-rl10 E2 0 4 0 -2 0 6 0 -I \ \\,-\ \ 11- - \ "i,B C D C B- X- YAFIG. 3. -Carbony1 interactions for [ Fe( CO) S( C H 41 2.A, fundamental frequencies vx, vy; B, normal coordinate function frequencies derived from A; C ,effect on B of " cis " carbonyl interaction in an Fe(CO)3 group; D, effect on C of including carbonylinteractions across the Fe(X .C6Y&Fe bridge.We now calculate values for k, and k,,. The corresponding:frequencies have beencalculated in table 1 in columns I and 11. Two possible situations can be envisaged, v,is either greater or less than v,, the former case is given in I, the latter in 11.v x = v 3 + j { ( + ( T ) ] 1 v,+v,v3+v4vy = v, +(v3 - v4),The general formulation of the problem is the same for both I and I1 and we nowdetermine which case obtains for the compounds. It seems reasonable to connect thedifferences between CO groups 1 and 2 and the other CO groups with their positionsrelative to the (XC,Y,) groups. d-Orbitals on the X atoms will interact with the 3dorbitals of the iron atoms and the n-orbitals of the phenyl or perfluorophenyl groups;this will provide a conjugated route for the transmission of effects due to changes ineither X or Y .A simple consideration of orbital overlap suggests that such change58 INTERACTIONS OF CARBONYL GROUPSshould have a more profound effect on the CO groups trans to X than on those transto the iron-iron bond. Since fluorine is an electronegative atom the effect of replacingC6H5 by C6F5 will be to withdraw electrons through the conjugated system, i.e.,backbonding to the carbonyl groups will be reduced and so k, will be increased. Thisresults in the general shift to higher frequencies observed originally. Similarly, whenS is replaced by Se or Te, the 3d orbitals are replaced by 4d and 5d orbitals which willbe less efficient at n-bonding.8 The iron-X resonance integral will decrease and soback-bonding to carbonyl groups will be increased. This will cause k, to decrease andexplains the shifts to lower frequencies observed when the chalcogen increases inatomic number.Variations in X and Y will therefore cause larger changes in v, than v,,.Uponexamining columns I and Il it is often possible to find a trio of compounds in which asY is varied (for constant X), v, is more or less constant but in which v, increases asfirst one and then two phenyl groups are replaced by perfl uorophenyl groups. Thevalues of v, and v, that follow from this simple rule are in bold type in either I or 11.Table 3 shows the corresponding values of k, and k,, for (Fe(C0)3.S(C6Y,)),.TABLE 3bridge componentsFe(C0) ~[bridge]Fe(CO) 3force constants mdynes A-1kx kYWhen two phenyl groups are present v,>v,, the situation is reversed for two perfluoro-phenyl groups and when the compound contains one group of each kind v,-v, SOthat the separation between v 3 and v4 is very small.As evident from table 1 the generalpattern of bonding suggested in this model can be extended to compounds of relatedstructure containing bridging elements other than members of group V1 and also tothe low temperature form of CO,(CO)~.In this model the interaction across the metal-metal bond is measured by thesplitting between the two high frequency lines. Using the equation above, k, for((CO)&S(C6H5))2 is 0.09 mdynes Hi-' and is weaker than the cis interaction as mightbe expected for a long range effect. Changing phenyl for perfluorophenyl has theeffect of reducing kd somewhat. This is in accord with the postulated increase inFe-X x-bonding attendant upon this change that permits an increase in the carbonylstretching frequencies. In the same way as back-bonding from the iron to carbonylgroups is reduced, so also will the bonding between the iron atoms due to 3d orbitalsbe reduced, thus kd will be slightly reduced.The authors acknowledge the cooperation of Dr. A. G. Massey in the preparationof this paper.E. Kostiner, M. L. N. Reddy, D. S. Urch and A. G. Massey, J. Organometal. Chem., 1968,15, 383.H. Hieber and W. Beck, 2. anorg. Chem., 1960, 305, 265.S. Kettle and L. Orgel, J. Chem. Soc., 1960, 3890.H. Hieber and T. Kruck, Ber., 1960,95,2027.B. E. Job, R. A. N. McLean and D. T. Thompson, Chem. Comm., 1966, 895.G. Bor, Spectrochim. Acta 1963, 19, 1209; K. Noack, ibid., 1925.F. A. Cottonand C. S. Kraihanzel, J. Amer. Chem. Sac., 1962, 84,4432.D. S. Urch, J. Inorg. Nucl. Chem., 1963, 25,771