J . Phys. Chem. 1990. 94, 6864-6875
6864
found between the adsorbate binding geometry as deduced from alterations in band shapes and band frequencies upon adsorption and from the relative intensities by utilizing the selection rules. As inferred also from earlier ~ ~ r k , the ~ examination ~ - ~ ~ - of~ C-H . ~ stretching modes is particularly insightful in this regard. The selection rule for C , adsorbates based on comparison of ring modes of a,, b,, and b, symmetry, as introduced by Creighton for pyridine a d ~ o r p t i o nhas , ~ also established its value for the present class of adsorbates. A practical limitation to its applicability, however, is brought about by the relatively low intensity of ring modes in these symmetry classes. The clear-cut qualitative obeyance to the SERS selection rules for the present systems, especially the very marked dependence of the aromatic C-H stretching intensities to the adsorbate orientation, might be considered surprising. This is because such a sharp demarcation is anticipated only for Raman excitation and scattering wavelengths that are distinctly to the red of the surface plasmon resonance frequencies.'.2a (This is because only then is the electric field vector normal to the surface anticipated to be much greater than that parallel to the interface.) For the present case, however, the surface plasmon resonances for roughened gold are expected to lie at frequencies comparable to that of the excitation wavelength h (647 nm).21 It would clearly be interesting to examine the extent to which selection rules are manifested as X is altered,22although this will be restricted to X Z 600 nm for gold due to the presence of an interband transition at shorter wavelengths. Of course, the electromagnetic theory upon which (21) For example, see: Wang, D. S.; Kerker, M. Phys. Rec. B 1981. 24,
1777; 1982, 25, 2433.
(22) Birke. R . L.; Lombardi. J. R. In Soectroelectrochemisrrv: Gale, R. J., Ed.: Plenum: New York. 1988; Chapter 6. (23) For example, see: (a) Gao, P.; Patterson, M. L.; Tadayyoni, M. A,; Weaver, M. J. Langmuir 1985, I , 173. (b) Weaver, M . J.; Corrigan, D. S.; Gao, P.; Gosztola, D.; Leung, L.-W. H. J . Electron Spectrosc. Relat. Phenom 1987. 45, 291,
the selection-rule predictions are based predicts that the plasmon resonance is dependent upon the geometry of the surface roughness.ib,21The theoretical predictions noted herein assume that the surface roughness takes the form of equal-sized Even though the predictions are independent of the spherical radius r (if r 1 Torr), and an appreciable (lo-" Torr) level of background gases such as water vapor. In order to draw a connection between surface thermal and photochemical reactions and film composition, the influence of all aspects of film growth conditions must be assessed. In this paper are reported the results of a systematic study of thin films photochemically deposited using a CW, lowpower 257-nm laser beam and the Cr, Mo, and W hexacarbonyls. The effect of light intensity, background gas pressure, and use of a buffer gas on film composition and structure are investigated, using scanning Auger and electron microscopies (SAM and SEM). The data show clearly that only the light intensity is unimportant over the experimental range used and that rapid surface and bulk chemical reactions exert a profound influence over film compositions. The compositions of films grown in ultrahigh vacuum and transferred under vacuum for analysis provide a data set which can be directly compared to studies of surface photolysis of the parent hexacarbonyls. By working with a focused laser beam and analyzing films grown in illuminated regions and away from them, it has been possible to distinguish surface photochemical reactions from spontaneous or thermal reactions for all three metal systems. Experimental Section
All films are deposited on Si( 1 11 ) or Si( loo), doped n- or p-type (1 0 Q-cm) cleaned in an acid oxidant bath and left covered by its native oxide. Cr(C0)6, Mo(CO),, and W(CO)6 are obtained from Aldrich: they are degassed by three freeze-pump-thaw cycles before each experiment. The light source is a frequencydoubled Ar+ ion laser (Coherent) which provides up to 4 mW at 257 nm. The 257-nm beam is expanded and focused to a 7-pm spot. Films are grown with an initial incident power density ranging from 40 to 3700 W/cm2 and exposure times of 5 s to 15 min. The laser powers used in this work are too low to lead to appreciable substrate heating (estimated to be
