3080 J. Phys. Chem. 1992, 96, 3080-3086 Use of Factor Analysis and XPS To Study Defective Nickel Oxide Agustin R. Gonziilez-Elipe,*Juan P. Holgado, Rafael Alvarez, and Cuillermo Munuera Instituto de Ciencia de Materiales de Sevilla, CSIC- Universidad Sevilla, and Departamento de Qdmica Inorgcinica, P.O. Box 1115, 41080 Sevilla, Spain (Received: June 11, 1991) The mathematical method of factor analysis has been used to analyze the Ni 2p3/2 and 0 1s photoelectron spectra of a nonstoichiometric nickel oxide sample with a high concentration of Ni3+defects. The spectra of these two levels are characterized by main peaks (at 854.6 and 529.6 eV for the Ni 2p3/2 and 0 1s levels) and satellites at ca. 2 eV higher binding energy, whose intensity depends on the degree of nonstoichiometry. Heating in vacuum at T > 723 K produces a decrease in the intensity of these two satellites and the appearance of a new peak in the Ni 2p spectra at 852.6 eV due to Nio. Factor analysis shows that these two phenomena are coupled. The existence of a new form of oxygen at 531.0 eV, unstable at 373 K under vacuum or after reaction with CO at 298 K, is also shown by this method. This new form, tentatively attributed to a peroxo-like species, is also generated after the adsorption of oxygen on the nickel oxide sample deeply reduced by Ar+ bombarding, a fact that points to its stabilization on surface unsaturated nickel sites, produced in a high concentration by sputtering. Introduction The degree of nonstoichiometry of NiO has been considered for many years as a key factor controlling the adsorptive and catalytic properties of this material.' In the earlier works on this subject, correlations have been tried between the changes in the electronic and transport properties of this material, both dependent on its degree of nonstoichiometry, with the amount of molecules adsorbed on its surface.2 More recently, the use of different spectroscopic techniques such as UV-V~S,~ XPS,4-9 and other photoemission techniquesg-' ' has enabled a direct investigation of the electronic structure of NiO or, as in the case of the IR spectroscopy, provided a way to characterize the species adsorbed on its surface after the adsorption of different gases.12J3 In the studies by XPS, an important feature at the 0 1s and Ni 2p core level spectra is the appearance of satellites at ca. 2 eV higher binding energy than the main peaks (in the following HBEP), which are commonly associated with the nonstoichiometry of the sample (i.e. with the presence of Ni3+ species). These peaks have been recently described by TomelliniI4 as multielectron final states induced by defective sites. According to this author, the energy shift between main peaks and satellites, in both the 0 1s and Ni 2p XPS spectra, is due to a change in the ionic charge and in the oxygen coordination induced by the cation vacancies, while their intensity depends on the degree of nonstoichiometry of the sample. However, in this and in other recent works by XPS on NiO,I5J6 the quantification of the main peaks and satellites is not intended, despite the fact that such calculations could provide some cor- (I) Gravelle, P. C.; Teichner, S. J. Adu. Catal. 1969, 20, 167. (2) (a) Larkins, F. P.; Frensham, P. J.; Sanders, J. V.; Trans. Faraday Soc. 1970,66, 1748. (b) Larkins, F. P.; Fansham, P. J.; Trans. Faraday SOC. 1970, 66, 1755. (3) (a) Le Calvar, M.; Lenglet, M. Stud. Surf. Sci. Catal. 1989, 48, 576. (b) Klier, K. Carol. Reo. 1967, I, 207. (4) Norton, P. R.; Tapping, R. L.; Goodale, J. W. Surf. Sci. 1977, 65, 13. (5) Kim, K. S.; Davis, R. E. J. Electron Spectrosc. Relat. Phenom. (6) Kim, K. S.; Winograd, N. Surf. Sci. 1974, 43, 625. (7) Roberts, M. W.; Smart, R. St. C. Surf. Sci. 1980, 100, 590. (8) Roberts, M. W.; Smart, R. St. C. J. Chem. SOC., Faraday Trans. I (9) McKay, J. M.; Henrich, V. E. Phys. Reu. B 1985, 32, 6764. (10) Hufner, S.; Hulliger, F.; Osterwalder, J.; Riesterer, T. Solid State 1972-73, I, 251. 1984,80, 2957. Commun. 1984, 50, 83. (11) Hufner, S.; Wertheim, G. K. Phys. Reu. B 1973, 8, 4857. (12) Escalona Platero, E.; Spoto, G.; Coluccia, S.; Zecchina, A. Lonzmuir 1987, 3, 291. (13) Escalona Platero, E.; Coluccia, S.; Zecchina, A. Surf Sci. 1986, 171, 465. (14) Tomellini, M. J. Chem. Soc., Faraday Trans. I 1988, 84, 3501. (15) Oku, M.; Tokuda, H.; Hirokawa, K. J. Electron Spectrosc. Relat. (16) Badyal, J. P. S.; Zhang, X.; Lambert, R. M. Surf. Sci. Lett. 1990, Phenom. 1990, 50, 61. 225, LIS 0022-3654/92/2096-3080$03.00/0 relations that might help in the understanding of the origin of the different features of the spectra. Quantification of the components existing in a series of spectra can be done by the factor analysis (FA) method.I7 This math- ematical technique is a multivariate statistical procedure which can be used for data handling and interpretation in many fields of the physical and analytical chemistry.I7-l9 It was introduced by Gaarenstroom20-22 for data evaluation in Auger electron spectroscopy, where the quantification in the derivative mode via peak-to-peak heights is usually distorted by peak shifts and shape alterations due to chemical effects. Besides Auger spectrosco- py,20-25 it has been also used for quantification of other surface analytical techniques such as In the present work we have used FA and fitting to analyze the 0 1s and Ni 2p XP spectra of a high surface area, nonstoi- chiometric NiO sample submitted to different thermal and Ar+ sputtering treatments to modify its degree of nonstoichiometry at the surface. The evolution of the different spectral features during these treatments as well as after the adsorption of oxygen under different conditions has provided evidence about the rela- tionship existing among the concentration of Ni3+ sites in NiO, its reducibility to NiO, and the formation of new forms of oxygen species adsorbed on its surface. Experimental Section The NiO sample used in the present work was prepared ac- cording to the method of Gravelle et al.,' by decomposition in vacuum at 510 K of a Ni(OH)* prepared according to the method of van Dillen et by homogeneous precipitation with urea from a Ni(N03)2 solution. The sample had a green yellowish color after it was heated in vacuum, but it became black by exposure to the air for storage. A particle size of ca. 30 A was estimated by XRD following the method of Sherrer,30 while a surface area of 159.2 m2 g-' was determined by the BET method. Figure 1 shows a TEM micrograph of a typical aggregate observed in this sample. The average sizes of the particles in the aggregates are similar to the value estimated by XRD. XPS,27 or SIMS.28 (17) Malinowski, E. R.; Howery, D. G. Factor Analysis in Chemistry; (18) Weiner, P. H.; Malinowski, E. R.; Levinstone, A. R. J. Phys. Chem. (19) Hugus, Z. 2.; El-Awady, A. A. J. Phys. Chem. 1971, 75, 2954. (20) Gaarenstrwm, S. W. J. Vac. Sci. Technol. 1979, 16, 600. (21) Gaarenstroom, S. W. J. Vac. Sci. Technol. 1982, 20, 458. (22) Gaarenstroom, S. W. Appl. Surf. Sci. 1986, 26, 561. (23) Steffen, J.; Hofmann, S. SIA, Surf. Interface Anal. 1988, 11, 617. (24) Hofmann, S.; Steffen, J. SIA, Surf. Interface Anal. 1989, 14, 59. (25) Atzrcdt, V.; Lange, H. Phys. Status Solidi A 1983, 79, 489. (26) Solomon, J. S. SIA, Surf. Interface Anal. 1987, 10, 75. (27) Koenig, M. F.; Grant, J. T. J. EIectron Spectrosc. Relat. Phenom. (28) Solomon, J. S. SIA, Surf. Interface Anal. 1987, IO, 216. (29) van Dillen, J. A.; Geus, J. W. Proc. Int. Congr. Catal., 6th 1976, 677. (30) Seherrer, P. Gott. Nachr. 1918, 2, 98. Wiley: New York, 1980. 1970, 74, 4531. 1986, 41, 145. 0 1992 American Chemical Society