.I. Chem. SOC., Faraday Trans. 1, 1987, 83, 1355-1361 Study of the Electronic Structure of UBr, using X-Ray Photoelectron Spectroscopy Geoffrey C. Allen"? Laboratoire de Radiochimie, Institut de Physique Nuclkaire, BPI, 91406 Orsay, France Jonathan W. Tyler Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucestershire GL13 9PB X-Ray photoelectron spectra of uranium and bromine core levels and valence bands have been recorded from UBr, using a Kratos ES300 electron spectrometer. The spectra are compared to their counterparts recorded from UF,, UCl, and UO,. The results obtained are discussed as a function of metal ionicity, covalency of the metal-ligand bond and the participation of 5f electrons in bonding. The satellite structure associated with the principal core levels in the early lanthanide and actinide halides are compared to the equivalent satellites observed in transition-metal halides and their origins are discussed.The uranium halides are of commercial importance for nuclear fuel fabrication and reprocessing cycles. In recent years the electronic structure of these compounds as elucidated by ultraviolet and X-ray photoelectron spectroscopy has attracted consid- erable attention.l-lO Interest is centred on the role of the U 5felectrons in compound formation and the nature of the uranium-halogen bond. Binding energies recorded for core and valence-band photoelectron peaks vary over the range k0.5 eV. This may be attributed to the sensitivity of the solids to atmospheric attack and the variety of techniques used in handling samples and calibrating the energy levels measured. Here we report core and valence-band spectra recorded from UBr, and compare the results with those obtained from UF,, UCl, and U0,.8v lo, l1 Experiment a1 X-Ray photoelectron spectra were obtained from a Kratos ES300 spectrometer using A1 Ka radiation at 300 W.All spectra were recorded in the fixed analyser transmission mode with a pass energy of 65 eV and source and collector slit widths set to 1.8 and 3.0 mm, respectively. Measurements were made using a Digital micro PDPl 1 computer and Kratos DS800 software. The UBr, sample was prepared at the Laboratoire de Radiochimie, Institut de Physique Nucleaire, Universite Paris Sud, Orsay, France. The polycrystalline sample was cut to generate a fresh surface suitable for analysis and was mounted on a specimen probe using double-sided adhesive tape.Samples were handled and prepared under argon in a polythene glove box attached to the preparation chamber of the spectrometer in order to prevent atmospheric contamination and to provide radiolytic containment. The spectra recorded showed no significant oxygen contamination. Pure copper foil and the known 233.0 eV difference between A1 Ka and Mg Ka radiation was used to calibrate the binding-energy scale.l2* l3 During the measurements the sample acquired a surface Permanent address: Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucester GL I3 9PB. 13551356 Electronic Structure of UBr, Table 1. Binding energies (eV) of core and valence electrons in UBr," level binding energy/eV 782.2 739.5 391.5 380.7 105.2 97.4 29.1 19.2 4.7 2.7 256.2 189.4 282.8 69.5 16.2 a Values are calibrated against the C 1s peak at 285.0 eV (E Au 4f7,2 at 84.0 eV).charge of 6-7 eV and therefore the energy scale was referenced to traces of adsorbed carbon. The binding energies presented in table 1 are averaged best estimates from several experiments using a C 1s value of 285.0 eV. The problem of selecting a suitable reference energy has always been a difficult one. Two methods are commonly used.', The first involves the use of adventitious carbon contamination within the spectrometer which we have chosen here. The second involves the evaporation of a thin layer of gold onto the sample so that binding energies may be referenced to the Au 4f712 line.Both techniques have disadvantages. In the case of carbon, the peak is often small and its position may vary with the substrate in question. Values have been recorded ranging from 284.7eV on noble metals to 285.6eV on oxidised magnesium.15 For gold the binding energy for the deposited layer may vary with coverage owing to extra-atomic relaxation effects between the layer and the substrate16-18 and chemical interaction with substrates such as halides and ~yanides.l~-~l Here we favour the use of adventitious carbon and have referenced the C 1s binding energy to the value 285.0 0.15 eV recorded from a clean UO, surface which had been left in the spectrometer for periods up to several days. The reproducibility of the present results is exemplified by the fact that even after 4 days exposure to the vacuum within the spectrometer, the spectrum recorded was virtually identical to that obtained immediately after insertion into the apparatus, the only difference being a very small increase in the very weak peaks associated with the 0 1s and C 1s levels. Results and Discussion Photoelectron spectra for the U 4fand valence-band regions from UBr, are shown in fig.1 and 2, respectively. These regions are compared with the corresponding spectra recorded from UF,, UCl, and UO, in fig. 3 and 4. Binding energies for the uranium and bromine photoelectron peaks are presented in table 1. Through a comparison with previous X-ray photoelectron studies of uranium halides1° and oxides,ll it is evident from fig.1 and 3 that a single 'shake-up' satellite is present situated at 6.1 eV to the high-binding-energy side of each uranium core level peak. The intensity of this satelliteG. C. Allen and J . W. Tyler 1357 380.7 391.7 1 c U L - 3 6.1 eV 6.leV 1 I I I 41 0 400 390 380 binding energy/eV Fig. 1. The U 4fphotoelectron spectrum of UBr, recorded using A1 Ka X-rays. F.w.h.m. U 4f5,2, 7,2 = 2.3 eV. u- B r 4 p I I I I I 40 30 20 10 0 binding energy/eV Fig. 2. The valence band photoelectron spectrum of UBr, recorded using A1 Ka X-rays. in the UBr, spectrum is ca. 45% of the parent peak. In contrast, however, the very much weaker but broader satellite situated ca. 15 eV to the high-binding-energy side of the U 4f;,, core peak is barely discernible in the UBr, spectrum, but is more easily observed in the corresponding fluoride compound.These higher-energy satellites were thought to be due largely to energy loss of the photoelectron on its way through the solid.3> Five main features are identified in the valence-band spectra of UBr, (fig. 2). The most intense band, centred at 4.7 eV, may be assigned to photoemission from orbitals which1358 Electronic Structure of UBr, " O L I I I I I I 420 410 400 390 380 37 binding energy/eV Fig. 3. The U 4fphotoelectron region of UF,, UCl,, UBr, and UO, recorded using A1 Ka X-rays. are mainly Br 4p in character (with some U 5 f , 6d and 7s admixture). We will refer to this entire structure as the 'bonding band' but identify separately the well defined shoulder to the low-energy side at ca.2.7 eV which we attribute to ionisation from the U 5flevel. At higher energy in the region of 20 eV, the broad band is in fact considered to be a poorly resolved doublet comprising photoionisations at 19.2 and 16.2 eV from the U 6p3,2 and Br 4s levels, respectively. The weak peak at 29.1 eV is associated with the U 6p1,2 level. Previously it was argued that X-ray photoelectron binding energies and peak intensities for uranium halides and oxides could be divided into two main categories.1° The first included binding-energy measurements from the U 4f and U 5f levels, which may be influenced by the effect of both covalency and Madelung potential. The second included data derived from peak separations and intensities such as the satellite-metal 4f peak separation and the corresponding ratio of their intensities, the U 5f 'bonding band' separation and intensity ratio and the intensity of the U Sfpeak itself.These should be independent of lattice effects and might be expected to reflect the covalency of the metal-ligand bond directly. These values for UBr, are presented in table 2 . For comparison, data from UF,, UCI, and UO, are also included.G. C. Allen and J . W. Tyler 1359 1 1 I I 15 10 5 0 ‘binding energy/eV Fig. 4. The valence band region of UF,, UCI,, UBr, and UO, recorded using A1 Ka X-rays. It is evident that the results for UBr, follow the trends observed previ~usly.~. lo As the electronegativity of the ligand is diminished the U 5fpeak height decreases with respect to the U 4f core levels and the U 5f ‘bonding band’ separation also decreases as does the binding energy of the U 4f core levels.In addition the satellites associated with the U 4f levels approach the parent peaks more closely in UBr, and their relative intensity is increased. All of these phenomena may be taken as an indication of increasing covalent character in the metal-ligand bond along the series UF,-UBr, and a progressively greater participation of the metal 5f level in bonding orbitals. The results are in good agreement with those of Thibaut et ~ l . , ~ who studied many different uranium halide and oxyhalide compounds. However, considerable variation exists in the U 4f,,2 binding energy values reported for the uranium tetrahalides. These are summarised in table 3.As noted earlier, this variation reflects the different sample handling and energy reference methods employed. During the past decade much attention has been paid to the observation of satellite structure in photoelectron spectra, particularly in compounds of the transition metals. However, the actinides are equally fascinating from this point of view. One important manifestation of the incorporation of oxygen in the fluoride lattice during the oxidation of UO, was the change in satellite structure and we have therefore paid special attention to the origin of these features in compounds of the actinide metals. Satellite behaviour appears to be very similar in the early lanthanide and actinide elements. Just as we have noted a progressive increase in the metal 4fcore-level satellite1360 Electronic Structur,? of UBr, Table 2.Summary of main peak binding energies and intensities in UF,, UCl,, UBr, and UO, compound UF, UCl, UBr, UO, U 4f7/2 binding energy/eV 382.8 381.5 380.7 380.2 U 5f binding energy/eV 3.2 2.7 2.7 1.5 ‘bonding band’ to U 5f peak 5.3 2.9 2.0 4.3 U 4f satellite-to-main peak 0.24 0.26 0.45 0.25 U 4f satellite separation 6.8 5.9 6.1 6.8 U 5f to U 4f intensity ratio 0.017 0.015 0.010 0.014 separationlev intensity ratio from main peak/eV Table 3. Binding energies (eV) of U 4f77/2 core electrons in UF,, UC1, and UBr, 382.8 684.5 381.5 380.7 10, this work 382.2 684.7 380.2 379.9 7b 1 382.7 684.8 3b 382.2 684.7 5c 382.5 (5) 684.6 - gd 382.3 684.6 2 - - - - - - - - - 381.5 - aValues are calibrated against the C 1s peak at 285.0eV (=Au 4f77/2 at 84.0 eV).Referenced to Au 4fiI2 = 83.8 eV, values increased by 0.2 eV. Referenced to C 1s = 284.8 eV, values increased by 0.2 eV. Referenced to F 1s = 684.6 eV. intensity along the series UF, to UBr,, so Weber et aZ.22 have reported the same effect in 3d photoelectron spectra from the compounds LaF, and LaBr,. Their molecular orbital calculations using the multiple scattering Xa method also indicate a greater covalency between the ligand and central metal ion for LaBr,. This is especially interesting when the recent results of de Boer et ~ 1 . ~ ~ are taken into account. For a similar series of scandium compounds these authors noted a general decrease in intensity from ScF, to ScI, and a similar trend for TiF, to TiI,. These satellites were assigned to exciton satellites which arise when the vacancy in the core level following photoelectron emission is screened by polarisation of the ligands. Such polarisation in turn induces an interaction between ligand p and s levels and satellites may arise from intra-atomic excitations on the ligand from a ligand p to an s orbital.In contrast, the opposite behaviour observed in our study may be attributed to screening of the core hole by the transfer of charge towards the metal atom. Thus the charge-transfer satellites observed may be described by transfer of valence electrons from a neighbouring atom to low-lying localised levels on the cation. Thus it would seem that though the exciton model may be appropriate for the early 3d metal compounds the strong satellites in many other transition-metal compounds, the compounds of the rare earths and those of the actinides, where a higher degree of covalency is present in the metal-ligand bond, are best described by the charge-transfer model.G.C. Allen and J . W. Tyler 1361 Conclusions For the X-ray photoelectron spectrum of UBr, a detailed study has been made of the U 5fpeak intensity, U 5f bonding band separation, U 4f binding energies and satellite- metal 4f peak separations and intensities. The measurements confirm that these parameters serve as a sensitive indicator of the nature of the metal-ligand bond in uranium compounds. They indicate a progressively greater participation of the outermost U 5J’levels in bonding orbitals along the series: UF, > UO, > UCl, > UBr,. The ‘ shake-up’ satellite structure associated with the U 4f photoelectron levels in uranium compounds has proved a valuable aid in the characterisation of oxide fuel.The present study identifies the importance of this feature in halide compounds. For the actinide elements such satellites are best described using a model in which screening of the core hole occurs by the transfer of anion electrons to the metal ion. This work was carried out at the Laboratoire de Radiochimie, Institut de Physique Nucleaire, Orsay, France, and at the Berkeley Nuclear Laboratories of the Technology Planning and Research Division and the paper is published with permission of the Central Electricity Generating Board. One of us (G. C. A.) thanks Professors Genet and Guillaumont and their colleagues at Orsay for their hospitality and the privilege of discussion with them during the preparation of this paper.References 1 D. Chadwick, Chem. Phys. Lett., 1973, 21, 291. 2 C. Miyake, H. Sakurai and S. Imoto, Chem. Phys. Lett., 1975, 36, 158. 3 J. J. Pireaux, N. Mirtensson, R. Didriksson, K. Siegbahn, J. Riga and J . J. Verbist, Chem. Phys. Lett., 4 J. J. Pireaux, J. Riga, E. Thibaut, D. Tenret-Noel, R. Caudano and J. J. Verbist, Chem. Phys., 1977, 5 W-Y. Howng and R. J. Thorn, Chem. Phys. Lett., 1979, 62, 57. 6 G. Thornton, N. Edelstein, N. Rosch, R. G. Egdell and D. R. Woodwark, J. Chem. Phys.. 1979, 70, 7 E. Thibaut, J-P. Boutique, J . J. Verbist, J-C. Levet and H. Noel, J. Am. Chem. Soc., 1982, 104, 5266. 8 J. M.Dyke, G. D. Josland, A. Morris, P. M. Tucker and J. W. Tyler, J. Chem. Soc.. Faraday Trans. 9 L. D. Trowbridge and H. L. Richards, Surf. Interface Anal., 1982, 4, 89. 10 G . C. Allen, P. M. Tucker and J . W. Tyler, Phifos. Mag. B, 1983,48, 63. I 1 G. C. Allen, I. R. Trickle and P. M. Tucker, Philos. Mag. B, 1981, 43, 689. 12 S. Evans, in Handbook of X-ray and Ultrariolet Photoelectron Spectroscopy, ed. D. Briggs (Heyden, 13 M. T. Anthony and M. P. Seah, Surf. Interface Anal., 1984, 6, 95. 14 P. Swift, D. Shuttleworth and M. P. Seah, in Practical Surface Analysis by Auger and X-Ra? Photo- 15 P. Swift, Surf. Interface Anal., 1982. 4, 47. 16 D. S. Urch and M. Weber, J . Electron Spectrosc. Relat. Phenom., 1974, 5, 791. 17 Y. Uwamino and T. Ishizuka, J. Electron Spectrosc. Relat. Phenom., 1981, 23, 55. 18 S. Hohiki and K. Oki, J . Electron Spectrosc. Relat. Phenom., 1985, 36, 105. 19 D. Betteridge, J. C. Carver and D. M. Hercules, J. Electron Spectrosc. Relar. Phenom.. 1973, 2, 327. 20 L. J. Matienzo and S. 0. Grim, Anal. Chem., 1974, 46, 2052. 21 V. 1. Nefedov, Ya. V. Salyn, G. Leonhardt and R. Scheibe, J. Electron Spectrosc. Relat. Phenom., 1977, 10, 121. 22 J. Weber, H. Berthou and C. K. Jargensen, Chem. Phys., 1977, 26, 69. 23 D K. G. de Boer, C. Haas and G. A. Sawatzky, Phys. Rev. B, 1984, 29,4401. 1977, 46, 2 15. 22, 113. 52 18. 2, 1981, 77, 1273. London, 1977), p. 130. electron Spectroscopy, ed. D. Briggs and M. P. Seah (John Wiley, Chichester, 1983), p. 437. Paper 6/ 1440; Received 18th July, I986