Application of Magnetic Circular Dichroism Spectroscopy to the Identification of Small Matrix-isolated Metal Clusters and the Assignment of their Electronic Spectra BY ROGERGRINTER ARMSTRONG, STEPHEN UPALIA. JAYASOORIYA JUNE MCCOMBIE AND JON P. SPRINGALL DAVID NORRIS School of Chemical Sciences University of East Anglia Norwich NR4 7TJ Received 1st August 1979 Those principles of magnetic circular dichroism which are particularly relevant to the assignment of the spectra of matrix-isolated metal atoms and small molecules formed from such atoms are briefly reviewed. These principles are illustrated through a discussion of the m.c.d. spectra of Ag and Cu isolated in argon and methane matrices. The analysis of the spectra of the silver species presents few problems and for Ag2 and Ag the results strongly confirm the assignments of species and electronic states which have been made hitherto.For the copper spectra the situation is much less satisfactory and further work is indicated. The study of matrix-isolated metal atoms and small aggregates of such atoms is a subject of much current re~earch.'-~ For many such interesting species the matrix phase would appear to be the only one in which they can be prepared and studied. The assignment of the electronic spectra of these molecules is an important task for the attribution of a particular spectral band to a specific molecule M, will only be really certain when detailed assignments of the bands have been made. Furthermore the process of assignment will also help to reveal the details of the electronic and geo- metrical structure of the molecules.Assignment of the electronic spectra of matrix-isolated molecules particularly those which have not been observed in other media presents difficulties over and above those normally encountered in the assignment of the spectra of molecules in solution or the gas phase. For example if the symmetry of the matrix site in which the guest species finds itself is low then the degeneracy of electronic states may be lifted and two or more bands observed where only one is expected. A well-known example of this phenomenon is found in the spectra of matrix-isolated copper silver and gold atoms for which the absorption band corresponding to the 2S1/2 -+2P3/z transition is invariably split into at least two It should be pointed out that there is not universal agreement on this subject and the formation of exciplexes has been suggested as the origin of the splitting.' Also in some cases silver in xenon for = example further splittings which cannot be explained by site-symmetry effects are also observed.6 A further cause of difficulty in assignment arises from the possibility of different matrix sites.If multiple sites exist in the host then the absorption bands of the guest may be correspondingly multiplied. Such effects are commonly seen in the spectra of metal atoms in matrices and can frequently be identified as such by the fact that R. GRINTER et al. they disappear on annealing.',' However there is no guarantee that all such bands will be removed by an annealing process.The purpose of the present paper is to illustrate the use of magnetic circular di- chroism (m.c.d.) spectroscopy in the assignment of the electronic spectral bands of matrix-isolated species. This subject has received rather little attention to date8v9 though the potential of m.c.d. in other fields has been amply demonstrated." EXPERI M E NTAL Matrices were prepared on a sapphire window within a matrix-isolation/m.c.d. cryostat which has been described elsewhere." Where they were not formed during deposition molecular species were produced within the matrix by thermal annealing or by irradiation at the wavelengths of the strong atomic absorption bands. Absorption spectra were measured with a Cary 14 spectrophotometer and m.c.d.spectra with a Cary 61 spectro-polarimeter. RESULTS AND DISCUSSION In using m.c.d. spectroscopy to assign electronic spectral bands and identify species two aspects of the technique play particularly important roles. This can best be illustrated by the following simple energy-level diagrams fig. 1 and 2. B=O I II I II FIG.1.-A diagrammatic illustration of the origin of a Faraday A-term. In fig. 1 we consider an atomic transition from a 'S state to a 'P which in the absence of a magnetic field would appear as a single line or band. In the presence of a magnetic field the triple degeneracy of the excited state is lifted and transitions to the states having ML = -1 and Mt= +1 are allowed for right circularly polarised (r.c.p.) light and left circularly polarised (1.c.p.) light respectively if the direction ofthe M.C.D.OF METAL CLUSTERS I FIG.2.-A diagrammatic illustration of the origin of a Faraday C-term. field is the same as that of the propagation of the light. Thus a plot of AA (absorbance of 1.c.p. light -absorbance of r.c.p. light) against wavelength gives rise to a sigmoid curve the properties of which depend critically upon the quantitative details of the excited state. This type of spectral feature is known as a Faraday A-term.I2 In fig. 2 the reversed situation a 'P+ 'S transition is illustrated. Again im- position of the magnetic field lifts the degeneracy of the 'Pstate and a sigmoid curve results from a plot of AA against the wavelength or frequency of the light.How-ever in this case the two lobes of the curve will not be equal in magnitude. At low temperatures in particular the Boltzmann population of the lowest level will exceed that of the highest and the population difference will grow as the temperature decreases until at very low temperatures (below 5 K in practice) the single band remaining will cease to grow and the signal is said to be saturated. Before the onset of saturation there is a temperature region in which the area of the band is linearly dependent upon the inverse temperature. The observation of this type of band which is known as a Faraday C-term,12 is an unambiguous indication of ground-state degeneracy for the absorbing species. For atomic systems and to a lesser extent for molecules analysis of the effects described above is a powerful way of obtaining quantitative estimates of ground or excited state g-factors spin-orbit coupling etc.In the present paper we shall be more concerned to show that qualitative observation of A and C term behaviour is quite sufficient for assignment purposes in many cases. We begin our discussion with some aspects of the spectra of Ag in argon matrices. Ag IN ARGON MATRICES Several laboratories have reported absorption spectra attributed to Ag (n < 10) in noble-gas matrice~.~-~*~ To date attribution of the bands observed has been based largely upon kinetic considerations as far as the species responsible for the band is R. GRINTER etal. concerned and upon theoretical work with respect to assignment.The work of Ozin and Huber' is the most relevant for comparison with our results since they have also used argon matrices. Table 1 is taken from their paper.2 TABLEBAND ASSIGNMENTS FOR Ag and Ag ISOLATED IN ARGON MATRICES~ d/nm Ag2 assignment Alnm Ag3 assignment 227 ? 245 ? 261 1264 1z; 3mu (244)" 387/412 (257.5/262)" 12; 3Iz; 440 HOMO -+LUMO a The figures in parentheses are the wavelengths recorded in the present work where they differ from those of Ozin and Huber.2 Fig. 3 shows our m.c.d. and absorption spectra of an argon matrix containing molecular silver species in the region between 235 and 270 nm. Though our wave- length data for the major atomic transitions agree exactly with those of Ozin and Huber we find differences of up to 3.5 nm for the peak positions of the bands in the spectral region shown in fig.3. These discrepancies which we do not regard as im- 2L0 250 260 wave Ie ng t h / nm FIG.3.-Absorption and m.c.d. spectra of silver in an argon matrix. The m.c.d. spectra were measured at a field of 6 T. Absorption spectrum at 10 K (.-.--.); m.c.d. spectrum at 10.0 (-) and 20.0 K (--). portant in the present context may arise from the lower temperatures of our measure- ments. The most notable feature of the m.c.d. spectra in fig. 3 is the temperature dependence of AA in the 246 nm region as opposed to its virtual absence between 254 and 270 nm. The temperature dependence does not have the most simple C-term form a subject to which we return but it is strong evidence for the assignment of the 244 nm absorption band to an electronic transition of the paramagnetic Ag molecule.M.C.D. OF METAL CLUSTERS Conversely the effective absence of temperature dependence suggests that the bands at 258 and 262 nm be assigned to a diamagnetic species in particular Ag,. There is a wavelength shift with changing temperature in the 254-270 nm region and this we believe is responsible for the slight change in the amplitude of the negative lobe of the m.c.d. at ~258 nm. However this is not a C-term phenomenon and we can further substantiate Ozin and Huber's assignment of the two bands in this spectral region. First we observe that each of the two partially-resolved absorption bands gives rise to an m.c.d.signal of sigmoid form crossing the AA = 0 line very close to the position of the absorption maximum. This is clear evidence of excited state de- generacy for both absorption bands. The two sigmoid features are of extremely similar form and their relative magnitudes parallel those of the corresponding ab- sorption bands. It appears therefore that we are dealing here with two transitions of a very similar nature but at slightly different energies. The suggestion2 that these two bands be assigned to 'C -f 'nutransitions at different matrix sites is therefore in excellent agreement with our m.c.d. evidence. We can further confirm the above conclusion by considering the signs and mag- nitudes of the m.c.d. features in fig. 3.The sign of the m.c.d. A-terms i.e. long-wavelength lobe negative short-wavelength lobe positive is quite in accord with the 'nuassignment for the excited state. The magnitude of the A-terms as measured by their first moment divided by the zero'th moment of the corresponding absorp- tion band 9,-,,13 can be shown to be equal to minus one half of the g-factor for the orbital magnetic moment of the excited state. Because the bands are not fully re- solved only approximate g-values can be obtained but the values of 0.14 for the long- wavelength band and 0.17 for the other provide good supporting evidence for the Illu assignment. When we examine the longer-wavelength regions of the spectrum we find clear absorption bands at 440 and 388 nm and some evidence of a weak shoulder around 412 nm.All these figures are in good agreement with Ozin and Huber's observations.2 However we can find no detectable m.c.d. under these bands even with a magnetic field of 6 T. For Ag, at least this negative result is to be expected if the 388/412 nm bands are assigned as 'ZC,+-f 'Xi. Theoretical analysis shows that for such a transi- tion the only m.c.d. to be expected is the so-called B-term12 which arises from the mix- ing of the %,+ excited state with other states notably the In, by the magnetic field. Since the latter lies some 13 000 cm-l away from the former this mixing will be small and a very weak B-term will result in agreement with our observations. In the case of the 440 nm band attributed to Ag the questions of molecular geo- metry and electronic structure both arise.Both theoretical14" and experimental 14' work indicates that a linear or nearly linear geometry is preferred for this molecule and if this is indeed the case then theoretical considerations suggest that the m.c.d. of the HOMO 3 LUMO band is likely to be weak and featureless as we observe. In that case the band at 245 nm might well correspond to a transition to an excited state of II symmetry. A state under spin-orbit coupling would give rise to ,113/,and components. Temperature dependence is expected on account of the 2X ground state degeneracy and a characteristic sigmoid m.c.d. signal should be found under the ,II3/ band but not under the 2111/2. The possibility of B-terms is always present. Though the temperature dependence is not as simple as that shown in fig.5 probably on account of an underlying background of temperature-dependent plasmon ab~orption,~ the m.c.d. spectra are quite consistent with the assignment of the 244 nm absorption band to a 211state of the Ag molecule. The m.c.d. in this spectral region is clearly not R. GRINTER etal. simple and a more detailed investigation may well reveal the positions of both the 2n3,2 components of the 211state. and 2111,2 CU IN ARGON AND METHANE MATRICES The position with regard to the spectra of copper species is by no means as clear cut as with silver. Nevertheless a number of interesting points may be made. One of the problems which makes the certain identification of copper aggregates difficult is the large number of atomic bands which occui in the ultraviolet between 200 and 300 nm.' Furthermore a large number of formally forbidden atomic transitions lie in this regionI5 and it appears to us quite possible that perturbations due to matrix- site asymmetry or adjacent copper atoms might make some of these transitions weakly allowed.We therefore begin our analysis of copper-cluster spectra by comparing the atomic transitions which occur in the ultraviolet (particularly those which arise from transitions to the configuration 3d9 4s' 4p') in methane and argon matrices. Fig. 4 and table 2 summarize our results on this subject. The m.c.d. spectra 220 230 240 250 wavelength 1 nm FIG.4.-M.c.d. spectra of copper atoms in argon (--) and methane (-) matrices at 4.5 K TABLE2.-wAVELENGTHS AND ASSIGNMENTS FOR THE ABSORPTION AND M.C.D.SPECTRA OF COPPER ATOMS ISOLATEDINARGON AND METHANE MATRICES IN THE SPECTRALREGION 21 5-250 nm assignmentUpb A(argon)/nm abs.b m.c.d.C I(methane)/nm abs.b m.c.d.C 4p3,2 238.3 239.5 242 243 4p112 (233.6 229.3 234.5 227 ? ? 238 222.5 4D1,~ 220.7 221.5 ? 21 8.5 2p1/2 218.0 218.5 21 6 216.5 a Electron configuration 3d94s' 4p' in all cases. Wavelengths of absorption maxima and assign- ments as given by Moskovits and Hulse.' = Wavelengths of the m.c.d. maxima see fig. 4. M.C.D. OF METAL CLUSTERS shown were readily obtained from matrices whose absorption spectra showed only very weak features which could be ascribed to the atomic transitions detailed in table 2; all are T-dependent.The power of a differential technique is clear in these spectra and taking the assignments of Moskovits and Hulse' as our basis we suggest the following additional assignments relating the spectra in argon to those in methane. The strong sigmoid features in the region of 240 nm are interpreted as two oppo- sitely signed C-terms on account of their marked temperature dependence (not shown for the sake of clarity) and in analogy with the m.c.d. spectra of other atomic transi- tions in copper silver and gold.16 The similarity in the form of these bands for the two different matrices suggests that they be assigned to the same transitions; the 4P3,2 according to Moskovits and Hu1se.l Thus the second component of this transi- tion is found to lie at 238 nm in a methane matrix.The remarkable resemblance of the two m.c.d. spectra in this region in particular the very similar separation of the bands (800 cm-l in argon 850 cm-l in methane) raises the important question of the origin of this splitting. It is rather difficult to believe that a crystal field effect of two different matrices could give rise to so similar a splitting and the exciplex explanation presents a similar problem. Further detailed analysis of the m.c.d. spectra of these and other atomic transitions should throw some light on these problems and is in progress.16 As we move towards higher energies further comparisons between the two m.c.d. spectra may be made and assignments suggested as in table 2.Apart from the dis- crete bands two inflexions are visible in the spectra at 235 nm in methane and 242 nm in argon. These features might be due to atomic transitions or to molecules of which we otherwise see no sign in this spectral region. The absence of bands definitely ascribable to molecules is surprising since Moskovits and Hulse find absorption bands which they attribute to Cu2 and Cu in this region for both argon and methane matrices. However we do find very strong evidence for both paramagnetic and diamagnetic molecules in the spectral region 340-600 nm fig. 5. The distorted sigmoid curve I I I I I I 2 2t c Q L I I I I I I 350 400 450 500 550 600 wavelength / nm FIG.5.-Absorption and m.c.d. spectra of copper in a methane matrix.Upper absorption spectrum at 4.5 K; lower m.c.d. spectra at 4.5 (-) 6.6 (--) 10.0 (.-*--.-. ) and 15.0 K (-.*--..-). Note that the wavelength scale changes at 430 nm. R. GRINTER et al. centred z 390 nm clearly belongs to a paramagnetic species with a degenerate excited state and a linear Cu molecule seems most likely although a triangular shape is also possible. The m.c.d. beneath the absorption bands peaking at 510 and 550 nm is quite weak and independent of temperature clearly suggesting the presence of diamagnetic species. The two bands could be due to different transitions of the same species a matrix site splitting or to two different species Cu2 and Cu for example. There is no evidence for a band due to Cu in this region but if such a molecule is linear and the absorption corresponds to a 2C-+2C transition then the m.c.d.might be very weak though mixing with the nearly 211state should be quite significant and moderately strong B-terms could result. More surprising is the absence of a strong signal due to Cu3 in methane at 227 nm (fig. 4) in view of Mosko- vits and Hulse’s assignment. In summary the assignment of the spectra of polymeric copper species presents considerable problems and further work is required on almost all aspects of the subject. In this task more theoretical work would be of great assistance to the experimentalist. Predictions concerning molecular shapes energies of electronic states and the effects of spin-orbit coupling would be most useful not only for copper but also for silver molecules and radicals.We are grateful to the S.R.C. for their support of this work by provision of apparatus liquid helium and Research Fellowships and Studentships. The m.c.d./m.i. cryostat was constructed with the aid of a grant from the Paul Instrument Fund of the Royal Society. M. Moskovits and J. E. Hulse J. Chem. Phys. 1977 67 4271. ’ G. A. Ozin and H. Huber Znorg. Chem. 1978 17 155. S. A. Mitchell and G. A. Ozin J. Amer. Chem. SOC., 1978 100 6776. W. Schulze H. U. Becker and H. Abe Chem. Phys. 1978 35 177. F. Forstmann D. M. Kolb D. Leutloff and W. Schulze J. Chem. Phys. 1977 66,2806. F. Forstmann D. M. Kolb and W. Schulze J. Chem. Phys. 1976 64 2552. L. Andrews and G. C. Pimentel J. Chem. Phys. 1967 47 2905.* I. N. Douglas R. Grinter and A. J. Thomson Mol. Phys. 1974 28 1377. R. L. Mowery J. C. Miller E. R. Krausz P. N. Schatz S. M. Jacobs and L. Andrews J. Chem. Phys. 1979,70,3920. lo P. J. Stephens Ann. Rev. Phys. Chem. 1974 25 201. T. J. Barton R. Grinter and A. J. Thomson J.C.S. Dalton 1978 608. l2 P. N. Schatz and A. J. McCaffery Quart. Reu. Chem. SOC., 1969,23,552; 1970,24,329. l3 P. J. Stephens J. Chem. Phys. 1970,52,3489. l4 (a) R. C. Baetzold J. Chenz. Phys. 1971 55 4363; R. C. Baetzold and E. Mack J. Chem. Phys. 1975 62 1513; (6) W. Schulze H. U. Becker R. Minkwitz and K. Manzel Chem. Phys. Letters 1978 55 59. l5 C. E. Moore Atomic Energy Levels (United States National Bureau of Standards 1958). l6 R. Grinter S. Armstrong U. A. Jayasooriya J. McCombie D. Norris and J. P. Springall unpublished observations.