1975 969Solid State Studies. Part VII. A Single-crystal Raman Study of theVibration of the Cr(CO), Unit in Hexa- and Penta-methylbenzenetri-carbonylchromiumBy H. J. Buttery and Sidney F. A. Kettle,' School of Chemical Sciences, University of East Anglia, NorwichNR4 7TJ1. Paul, Department of Chemistry, Queen Elizabeth College, Campden Hill Road, London W.8Single-crystal Raman studies of the isomorphous title compounds have shown that, whilst a vibrational factor-group method offers the simplest explanation of the 2000 cm-l region, some features remain unexplained by it.IN a recent study of the solid-state vibrational spectraof the isomorphous compounds hexamethyl- andpent ame t h yl-benzene t ricarbon ylchromium it was sug-gested that, although the correct factor group is Di,",an analysis of the spectra in the 2000 cm-l region ismore appropriately carried out in Dit, the size of theunit cell being ha1ved.l The latter is preferable, as afirst approximation, because it recognises that in thehexamethyl compound (and, presumably, the penta-methyl also) each Cr(CO), lies on a local, but not (quite)crystallographic, mirror plane, all these local mirrorplanes being parallel.Even this approach does notrecognise the full pseudo-symmetry of the Cr(CO),units. Each of these retains, to a good approximation,the threefold rotation axis of the isolated molecule.In the crystal structure, although z = 8, the Cr(CO),units are packed so that these pseudo-three-fold axesare almost parallel. Although it does not seem possibleto exploit this observation by a simple transpositionto a factor group appropriate to a trigonal system, thearrangement undoubtedly has an important effecton the vibrational spectra.For instance, it providesan explanation for the observation of a single intenseband derived from the (molecular) totally symmetricCO stretching vibration whilst the full factor grouppredicts three bands in this region and the modifiedfactor group predicts two.The crystal structure of the hexamethyl compoundhas been determined and we have shown that thepentamethyl is isomorphous. This observation impliesa disorder in the crystal packing of the latter compoundto accommodate the CH,-group difference between thetwo compounds.It is to be expected that, as resolution increases, theanalysis of the 2000 cm-l region in terms of a vibrationalfactor group will break down since its use implies de-generacy between modes which are distinct in the fullfactor group.Only if appropriate symmetry non-related interaction constants are precisely equal can sucha degeneracy persist. A test of the applicabilityof the vibrational factor-group model can be made intwo ways. First, the spectral resolution may beincreased. Secondly, the symmetry characteristicsassociated with individual spectral features may bedetermined. The room-temperature i.r. spectra arerelatively broad and, although some narrowing occurson cooling to -196 "C, no new features emerge. TheRaman peaks in a polycrystalline sample are alsonarrower but, again, no new features appear on cooling.In all these cases the resolution in the spectra was band-width limited (the instrumental resolution being 1 cm-lor better).We have therefore directed our attentionto single-crystal Raman studies. Ideally, these shouldbe carried out at -196 "C. Although such studies havebeen attempted the decreased collection angle, resultingfrom the large cryostat needed, resulted in a poor signal-to-noise ratio. Further, the difficulty of crystal align-ment resulted in increased breakthrough. The low-temperature spectra are, consequent ly, clearly inferiorto those obtained at room temperature. We thereforereport only the latter.RESULTS AND DISCUSSIONTricarbon ylpen t ame t h ylbenzenechromium cry st allisesin the orthorhombic Pbca space group with eight mole-cules in a unit cell of dimensions a = 13.65, b = 13.45,and c = 15.24 A.Crystals were easily obtained from1 : 1 diethyl ether : di-isopropyl ether solutions as rect-angular blocks of dimensions ca. 10 x 3 x 1 mm boundedby (100) and with well developed (010) faces. Thecrystal shape is therefore ideal for a single-crystalRaman study. In particular, breakthrough would beexpected to be small. This proved to be the case for allthe several crystals studied. The preparation andanalysis of the compound have been detailed e1sewhere.lTricarbonylhexamethylbenzenechromium also crystal-lises in the orthorhombic space group Pbca (Dii) with2 = 8 and has a unit cell of dimensions a = 13.67,b = 13.53, and c = 15.27 A.2 In contrast to the penta-methyl compound, it proved very difficult to obtaingood quality crystals of tricarbonylhexamethylbenzene-chromium. The compound is only moderately solublein di-isopropyl ether (a solvent we have found it ad-vantageous to use for crystal-growing purposes). It isslightly more soluble in benzene, but the crystalsobtained were inferior in quality to those from di-isopropyl ether.Crystals were eventually obtainedfrom di-isopropyl ether as essentially rectangularplates, of varying dimensions. The largest, ca. 3 x1 x 0.2 mm, was used for the spectra reported inthis paper. The crystals were commonly of hexagonalcross-section, a few with developed (102) faces, butthe majority with well developed (001) faces.Asfor the pentamethyl compound, the crystal studiedH. J. Buttery, G. Keeling, S. F. A. Kettle. I. Paul, and P. J .Stamper, J . Chenz. SOC. ( A ) , 1969, 2224.a M. F. Bailey and L. F. Dahl, Inorg. Chem., 1965, 4, 1298970 J.C.S. Daltonwas oriented from its conoscopic interference figuresand the axes identified by X-ray techniques (see Figure1). The small size of the crystal prevented cuttingb100102(JOOIFIGURE 1 The morphology of tricarbonylhexamethylbenzene-chromiumto a rigorous rectangular shape, so some breakthrougheffects are to be expected in the spectra.We have found that the expectation of less break-through in the case of the pentamethyl compoundisomorphic point group. The correlation is given inTable 1, from which the spectral predictions followdirectly (note that A , vibrations are the only inactivespecies).D;; DEA,+ A , -1 BsuB1, + B1, + BzuB2g- B29 + B1uBag __+ B3s i- A,L4u- At4 + B3yB1U + B1u + B2gBzu+ B2t' + B1,B3u- B3u + A,The simplest distinction between the vibrationaland full factor-group analyses is that the latter predictstwice as many Raman peaks as the former. It wason this basis that the applicability of a vibrationalTABLE 1Correlation table for tricarbonylhexamethyIbenzenechromium1;ull €actor \'ibrationalMoIecule Site group factor group l'scu do -cite Moleculec3v Cl DiE ( 8 ) jz; (4) c, c3r.( i r e , R) \(Lr., R)A \ (ix., R) \ \1; (i.r.,was realised.This species, then, provides the moreincisive test of the applicability of the alternativefactor groups.Because the symmetry elements whoseexistence is implied by a Dii factor group are notcrystallographic the breakdown of the vibrationalfactor-group model would most probably be rnani-fested by spectral observations which are inter-mediate between those appropriate to DZ and Dii.That is, the degeneracies implied by 02: would berelieved.In changing from Dkf to Dii there is, in standardsetting, a permutation of co-ordinate axes: x(Di:) -+y _t x. We avoid the problems arising from this byusing the inverse permutation to refer Dg factor-grouppredictions to the same co-ordinate system as DZ.Using this notation, the two groups are related bywhat is, in D& the factor-group coset isomorphic to~ ( y x ) in Dm becoming the y axial glide coset in DE.Physically, the distinction between the two spacegroups is that the two ' half ' unit cells in DiR arerequired to vibrate in-phase in DG (being relatedby a pure translation) but in DE the out-of-phasevibration is also allowed.Evidently, the two casesmay be distinguished by the oh(yx) operation in thefactor-group approach was first suggested. The transi-tion Dii Dii would be expected to be characterisedby weak peaks occurring in the Raman at positionsapproximating to those of i.r. absorptions (Table 1).Further, the symmetry species of these weak peaksis clear. Thus, in those vibrations arising from mole-cular A1(&) modes, Di: predicts Raman featuresof A , and B3g symmetry, whilst the applicability ofDii will be indicated by the presence of additionalpeaks associated with vibrations of B1, and B2, species.Analogous arguments apply to the i.r.spectrum butthe peaks are too broad to permit the observation ofadditional features unless these are themselves ratherintense.The arrangement of the Cr(CO), units in the unit cellin ab projection is shown in Figure 2 for the hexamethylcompound.2 The orientation of the pseudo-mirrorplanes and pseudo-three-fold axes discussed earlier isevident from this Figure.The correlations between the molecular, vibrational,and full factor groups are shown in Table 2 togetherwith the predicted Raman activities. The single-crystal Raman spectra of the pentamethyl derivativeare shown in Figure 3 and those of the hexamethy1975 971in Figure 4, the corresponding data being detailed inTables 3 and 4.TABLE 2Factor and point group correlationsFull Vibrationalfactor group Molecular mode factor group(DZ) (Cat4 (D2)A , Modes [z(xx)y, z(yy)x, y(zx)x].-It is clearfrom both Figures 3 and 4 that only two strongthe spectra.Although no such peak is shown associ-ated with the 1850 cm-l A , mode in Table 4 it seemsprobable that one occurs, unresolved from the 1845cm-l B1, peak (the separation between the two peaksin the pentamethyl spectra is only 2 cm-l). In thecase of the higher energy A , mode the additional featurecannot be associated with a breakdown in the vibrationalfactor-group model since the full factor group does notpredict an additional A , mode.We incline to theview that this peak is a hot band involving a low-frequency lattice mode. It would not be expectedthat cooling to -196 "C would significantly alter theintensity of this peak and, indeed, we were unable todetect any change (although the poor signal-to-noiseratio of our low temperature spectrum might well haveobscured any variation). The status of the A , peaksTABLE 3Infrared and single-crystal Raman frequencies (cm-1) for the carbonyl stretching vibrations oftricarbonylpentarnethylbenzenechromium> Raman intensities $InfraredSolutionAY Z ( X X ) Y Z( Y Y ) X Y(2.Z); Z( Y X ) Y Z ( X 2 ) Y Z ( Y Z ) Y Assignme2--ziGz--4 Bsgca.1939sh cn. 1940sh 1933 25 48 20 A01929 * 1929 * ca. 1928 9 22 9 ( A 8 )47 B2ol l ? (B2A1876 1877 1879 (4)ca. 1858 1859 1883 1868 10 P s g )1853 1849 1859 100 B391822 * 1821 * 1857 39 11 19 (11) (3) A ,1851 5 6 7 (Ag)1849 ( 3) 69 B1,I I1 (CC14)1947 1948 1957 ca. 1934ca. 1866sh 1867 ca. 1873* I3CO. t I, Room temperature: 11, low temperature. Peak heights (arbitrary units).TABLE 4Infrared and single-crystal Raman frequencies (cm-l) for the carbonyl stretching vibrations oftricarbonylhexamethylbenzenechromiumInfrared--Solid state1945ca. 1938sh1926 *1869ca. 1854sh18491819 *v Intensities 1Solution r A (CC14) Av Z ( X X ) Y X ( Y Y ) Z X ( Z 2 ) Y X ( Y X ) Z X ( Z X ) Z X ( 2 Y ) Z1953 1924sh 12 20 61928 24 46 16 (3)35121879 1851 1001872 (6) (16)1850 46 8 30 (4)1845 t (14) (4) (5) (2) (4)1864* '3CO.t There is possibly a coincident peak. 1 Peak heights (arbitrary units).peaks are common to all the above spectra (becausex2, y2, and z2 transform independently as A,, identicalrelative band intensities are not expected for (xx),(yy), and (zz) observations, only a correspondencebetween frequencies]. An additional feature in the(xz) spectra is undoubtedly breakthrough of a Bz, modewhich is seen strongly in (xz) polarisation. This peakis also seen in the (xx) spectrum of the hexamethylcompound. Although not evident from Figures 3and 4, it is clear from Tables 3 and 4 that there is alsosome breakthrough from a B1, mode. An unexpectedobservation is the appearance of a second A , .peaksome 5 cm-l to low energy of the main A , peaks in allat 1845(9) cm-1 is also unclear.Their presence couldbe taken as evidence against the applicability of thevibrational factor-group approach. However, beforeaccepting this viewpoint we would seek similar evidencerelating to modes of different symmetry. In the absenceof such confirmatory data we incline to the view thatthese peaks have a similar origin to those at 1924(8)cm-l.B1, and B5 Modes [x(yx)z, z(yx)y, x(xz)y), x(zx)x].-These modes are considered together because in eachcase the vibrational factor group predicts a single peakwhilst the full factor group predicts two. As is evidentfrom Figures 3 and 4, to a first approximation th972vibrational factor-group method is applicable.Inno spectrum is there any evidence for a second peakof BI, symmetry. In the pentamethyl compound thereFIGURE 2 The projection of the Cr(CO), units, symbolicallyrepresented, in the unit cell of tricarbonylhexamethylbenzene -chromium onto the ab plane1900 1820 -1 - cmFIGURE 3 Single-crystal Raman spectra oftricarbonylpentamethylbenzenechromium-1 - cmFIGURE 4 Single-crystal Raman spectra oftricarbonylhexameth ylbenzenechromiumis probably a second BQ peak, some 6 cm-1 lower inenergy than the main peak. This additional peak isreminiscent of the additional peaks found in the A ,spectra and could have a similar origin. Equally, itJ.C.S. Daltoncould indicate the approximate nature of the vibrationalf ac tor-group approach.B3g i'lfodes [x(zy)z, ~(yz)yJ.-In contrast to the pre-diction of two and three peaks, respectively, from thevibrational and full factor-group methods, it is clearfrom Figures 3 and 4 that only a single strong B3gpeak occurs (some breakthrough of the ca.1930 cm-lA, peak is seen). The ' missing ' peak should alsoappear in the ca. 1930 cm-l region. However the reasonfor its absence seems clear; it lies in the near-parallelnature of the local threefold axes of the individualmolecules discussed at the beginning of this paper.This alignment means that in forming the crystallo-graphic derived-polarisability tensors appropriate tothe various Raman-active species derived from thelocal A, (in C,,) modes, only the totally symmetriccombination (+ + + +) of molecular derived-polaris-ability tensors will not vanish (using a Wolkensteinapproach). A zero, or, at least, very low intensityfor the Raman peak associated with the Bag mode istherefore predicted.Similar arguments (in 0::) applyto the B,, and B, peaks in this region and indicate theimportance of the CQ. 1850 cm-l region in distinguishingbetween the alternative factor groups.The B3g species are of particular importance in thisrespect because, as Tables 3 and 4 indicate, there is apeak of this species CQ. 10 cm-1 to high energy of thestrong B% peak in this region in both penta- and hexa-methyl spectra. Although a two-quantum explanationcannot be excluded, the most probable explanation forthis observation seems to be the approximate natureof the vibrational factor-group method.The Low Frequency Region.-The interpretation ofthe low frequency spectra of metal carbonyl speciesin the solid state is, like the corresponding molecularspectra, made difficult by the extensive mixing ofM-C stretching vibrations with M-C-0 deformations.Additionally, further complications are introduced bythe appearance of lattice modes together with corre-sponding overtone and combination bands.That thereis some considerable lack of concordance between thetwo sets of data below CQ. 300 cm-l may well be associatedwith the disorder which is presumably present in thepentamethyl compound. The data are similar to thosefound by Adams and Squire in their work on poly-crystalline forms.A common observation is that a particular peakoccurs in more than one polarisation.The magnitudeof the effect is such that it cannot be attributed tobreakthrough. Rather, we believe, it originates in thesmall extent of factor groups splitting in the low-fre-quency internal modes. In this situation, the inform-ation which can be obtained from single-crystal studiesis small. We shall therefore not discuss our datafurther. Detailed data have been treated as a Supple-mentary publication (No. SUP 21293, 5 pp.).** For details of the Supplementary publication scheme, seeNotice t o Authors No. 7, in J.C.S. Dalton, 1974, Index issue.3 D. M. Adams and A. Squire, J . Chem. SOC. ( A ) . 1970,8141975 973ConcZusiost.-Although the vibrational factor groupoffers the simplest explanation for the main featuresobserved in both the polycrystalline and single-crystalRaman spectra of penta- and hexa-methylbenzenetri-carbonylchromium in the 2000 cm-l region, it cannotaccount for all the additional features revealed by single-crystal studies. It is clear that the explanation of someof these features lies outside any simple harmonicoscillator/factor-group approach. However, an ad-ditional peak observed in (yz) polaxisation can mostreadily be explained by the full-factor group. Some ofthe present ambiguities could only be resolved by asingle-crystal study at 4 K.EXPERIMENTALSpectra were recorded on a Spex 1401 instrument using632.8 nm excitation and photon counting. Incidentpower was usually in the range 10-20 mW. Spectralslit width was 1 cm-l or better.[4/264 Received, 11th February, 1974