Thin Films Adsorbed on Au(111) and on Self ... - ACS Publications

Aug 20, 2005 - Christelle Hauchard, Christian Pépin, and Paul Rowntree*. De´partement de chimie, Universite´ de Sherbrooke, Sherbrooke, Centre for Sel...
0 downloads 0 Views 181KB Size
Download PDF
9166

Langmuir 2005, 21, 9166-9175

Fe(CO)5 Thin Films Adsorbed on Au(111) and on Self-Assembled Organic Monolayers: II. Thermal Transformations Christelle Hauchard, Christian Pe´pin, and Paul Rowntree* De´ partement de chimie, Universite´ de Sherbrooke, Sherbrooke, Centre for Self-Assembled Chemical Systems, Sherbrooke, Que´ bec J1K 2R1, Canada Received March 13, 2005. In Final Form: June 23, 2005 The thermal transformations of as-deposited Fe(CO)5 films adsorbed on Au(111)/mica and C4, C8, C12, and C16 self-assembled methyl-terminated monolayer organic surfaces have been studied using infrared spectroscopy to probe how the physical restructuring influences the sensitivity of these systems to lowenergy electron beams. A companion publication shows that the as-deposited monolayers are composed of molecules physisorbed with one axial and two equatorial carbonyl groups directed toward the substrate; subsequent layers are preferentially oriented with the C3 molecular axis aligned perpendicular to the substrate (i.e., one axial carbonyl group directed toward the substrate). In this work, we show that the as-deposited films are structurally unstable above 125 K on Au(111)/mica surfaces and above 100 K on the organic self-assembled monolayers. Above these thresholds, the layered structures transform into three-dimensional aggregates, implying strongly nonwetting behavior for Fe(CO)5 on each of these substrates; molecular desorption from this aggregate structure takes place between 140 and 160 K. The irreversibility of this temperature-induced transformation demonstrates that the as-deposited layered films do not represent a thermodynamically well-defined phase; this key feature of the as-deposited films is believed to be the cause of the discrepancies in previous attempts to understand Fe(CO)5/surface structures based on infrared results. Moreover, the thermally induced transformation to 3D aggregate structures is shown to decrease the apparent sensitivity of the adsorbed Fe(CO)5 to low-energy electron-induced decarbonylation (0-10 eV) by over 3 orders of magnitude.

Introduction Physisorbed molecular films are widely used as model systems to understand fundamental molecule-surface chemistry, to study interphase phenomena in systems of reduced dimension and dimensionality, and to explore the interplay between molecule-surface and moleculemolecule interactions.1-5 In practice, they can also provide valuable insight into diverse applications such as lubricating films6-9 and reactive precursor films.6,10,11 The structural characteristics of these films, the mechanisms of thin-film growth, and the phase diagrams for these systems provide information that is required to optimize the adhesive properties of films, to ensure uniform “wetting” of the substrate, and to rationally design the film constituents for a given application. We are currently interested in using thin adsorbed films of metal-carbonyls [e.g., Fe(CO)5, Ni(CO)4] as precursors for electron-induced * To whom correspondence should be addressed. E-mail: [email protected]. (1) Brusch, L. W.; Cole, M. W.; Zaremba, E. Physical Adsorption: Forces and Phenomena; Clarendon Press: Oxford, U.K., 1997. (2) Shrimpton, N. D.; Cole, M. W.; Steele, W. A.; Chan, M. H. W. Surface Properties of Layered Structures; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1992. (3) Zeppenfeld, P. Physics of Covered Surfaces; New Series Group III; Springer: Berlin, Germany, 2001; Vol. 42. (4) Hautman, J.; Klein, M. L. Phys. Rev. Lett. 1991, 67, 1763. (5) Gatica, S. M.; Johnson, J. K.; Zhao, X. C.; Cole, M. W. J. Phys. Chem. B 2004, 108, 11704. (6) Ren, D.; Sung, D.; Gellman, A. J. Tribol. Lett. 2001, 10, 179. (7) Nicholls, M. A.; Than Do, T.; Norton, P. R.; Kasrai, M.; Bancroft, G. M. Tribol. Int. 2005, 38, 15. (8) Barrena, E.; Kopta, S.; Ogletree, D. F.; Charych, D. H.; Salmeron, M. Phys. Rev. Lett. 1999, 82, 2880. (9) Salmeron, M. Tribol. Lett. 2001, 10, 69. (10) Thibaudau, F.; Masson, L.; Chemam, A.; Roche, J. R.; Salvan, F. J. Vac. Sci. Technol., A 1998, 16, 2967. (11) Swanson, J. R.; Friend, C. M.; Chabal, Y. J. J. Chem. Phys. 1987, 87, 5028.

metallization processes; the precursor materials of interest are principally organometallics, because of the experimental ease with which these target species can be vapordeposited,10-23 and subsequently dissociated using electron beams.11,22,24,25 The structural properties of these asdeposited multilayer films have important consequences for the success of this strategy, and the determination of the film properties using nondestructive probes has become increasingly important in this regard. Following Bauer’s classification,26 thin films can deposit under equilibrium conditions following (i) Frank-van der Merwe (i.e., layer-by-layer) growth in the case of strong (and perhaps specific) adsorbate-substrate interactions, (ii) Volmer-Weber (i.e., 3D clustering) growth in the case of strong interadsorbate attractions, or (iii) the interme(12) Parnis, J. M.; Thompson, M. G. K.; Ashenhurst, L. M. J. Phys. Chem. A 2003, 107, 7390. (13) Zanoni, R.; Piancastelli, M. N.; Marsi, M.; Margaritondo, G. Solid State Commun. 1994, 89, 673. (14) Sato, S.; Ukisu, Y.; Ogawa, H.; Takasu, Y. J. Chem. Soc., Faraday Trans. 1993, 89, 4387. (15) Sato, S.; Ukisu, Y.; Ogawa, H.; Takasu, Y. Appl. Surf. Sci. 1994, 79/80, 428. (16) Sato, S.; Minoura, S.; Urisu, T.; Takasu, Y. Appl. Surf. Sci. 1995, 90, 29. (17) Sato, S.; Suzuki, T. J. Phys. Chem. 1996, 100, 14769. (18) Sato, S.; Suzuki, T. J. Electron Spectrosc. Relat. Phenom. 1997, 83, 85. (19) Sato, S.; Tanaka, S. Appl. Surf. Sci. 1998, 135, 83. (20) Sun, L.; McCash, E. M. Surf. Sci. 1999, 422, 77. (21) Hauchard, C.; Pepin, C.; Rowntree, P. A. Langmuir, 2005, 21, 9154-9165. (22) Henderson, M. A.; Ramsier, R. D.; Yates, J. T., Jr. Surf. Sci. 1991, 259, 173. (23) Henderson, M. A.; Ramsier, R. D.; Yates, J. T., Jr. Surf. Sci. 1992, 275, 297. (24) Foord, J. S.; Jackman, R. B. Chem. Phys. Lett. 1984, 112, 190. (25) Hauchard, C.; Rowntree, P. A. J. Phys. Chem. Manuscript submitted. (26) Bauer, E. Z. Kristallogr. 110, 372, 1958.

10.1021/la050678y CCC: $30.25 © 2005 American Chemical Society Published on Web 08/20/2005

Thermal Transformations of Fe(CO)5

diate Stranski-Krastanov growth, whereby a finite number of 2D layers is capped by clusters that “ripen” with increased exposure. Systems can undergo transitions among these growth modes as the equilibrium conditions are changed; in some systems, Volmer-Weber films can be replaced by Frank-van der Merwe films via a firstorder phase transition, as the wetting temperature is exceeded.27,28 Despite the utility of this thermodynamic classification, the prerequisite thermodynamic equilibrium is difficult to attain under many deposition conditions; this is especially true for thin-film growth under vacuum and for substrate temperatures that are far below the triple point of the film constituents. Films deposited under these conditions may lack the long-range order that characterizes equilibrated phases, as well as the thermodynamic stability that is normally associated with a well-defined phase. The current report is an extension of the companion publication21 (herein referred to as I) and describes the thermally induced structural transformations of thin films of iron pentacarbonyl [Fe(CO)5, referred to herein as IPC]. Together, these works were undertaken to develop a protocol to deposit metallic films on fragile organic surfaces without resorting to electrochemical processes29,30 (and the associated solvents required) or thermal metal-atom evaporation,31-33 which can induce irreversible chemical and physical damage to the organic substrate. In principle, IPC provides a nearly ideal precursor model because of its thermal stability, ease of manipulation, and sensitivity to electron-induced ligand elimination in the gas phase.34,35 The two distinct carbonyl ligand types of IPC (i.e., the three equivalent equatorial CO groups and the two equivalent axial CO groups, herein referred to as EQ and AX, respectively) have strong signature absorptions in the 1900-2100 cm-1 infrared region, thus enabling the use of nondestructive infrared reflection-absorption spectroscopy (IRRAS) to probe the surface. Notwithstanding these properties, the use of adsorbed IPC is far from trivial. Spontaneous molecular decomposition has been reported on reactive substrates,20,36-41 while adsorption appears to be nondissociative at or below ambient temperatures on Au(111),18,21 graphite,42-44 and Ag(111)22 surfaces. To study the electron-induced dissociation dynamics of these species, we have chosen to use chemically inert substrates such as Au(111)/mica and self(27) Cahn, J. W. J. Chem. Phys. 1977, 66, 3667. (28) Ebner, C.; Saam, W. F. Phys. Rev. Lett. 1977, 38, 1486. (29) Kind, H.; Bittner, A. M.; Cavalleri, O.; Kern, K.; Greber, T. J. Phys. Chem. B 1998, 102, 7582. (30) Nishizawa, M.; Sunagawa, T.; Yoneyama, H. Langmuir 1997, 13, 5215. (31) Herdt, G. C.; Czanderna, A. W. J. Vac. Sci. Technol., A 1994, 12, 2410. (32) Jung, D. R.; Czanderna, A. W. J. Vac. Sci. Technol., A 1995, 13, 1337. (33) Herdt, G. C.; Czanderna, A. W. J. Vac. Sci. Technol., A 1999, 17, 3415. (34) Pignataro, S.; Foffani, A.; Grasso, F.; Cantone, B. Z. Phys. Chim. Neue Folge 1965, 47, 106. (35) Compton, R. N.; Stockdale, J. A. D. Int. J. Mass Spectrom. Ion Phys. 1976, 22, 47. (36) Gluck, N. S.; Ying, Z.; Bartosch, C. E.; Ho, W. J. Chem. Phys. 1987, 86, 4957. (37) Zaera, F. Surf. Sci. 1991, 255, 280. (38) Xu, M.; Zaera, F. Surf. Sci. 1994, 315, 40. (39) Mu¨lbauer, S.; Petkova, A.; Froitzheim, H. Surf. Sci. 2004, 562, 195. (40) Rocklein, M. N.; Land, D. P. Surf. Sci. 1999, 436, L702. (41) Pires, J.; Brotas de Carvalho, M.; Ribeiro, F. R.; Derouane, E. G. Microporous Mesoporous Mater. 1995, 3, 573. (42) Wang, R.; Taub, H.; Schechter, Brener, R.; Suzanne, J.; Hansen, F. Y. Phys. Rev. B 1983, 27, 5864. (43) Dennison, J. R.; Taub, H. Hansen, F. Y.; Shechter, H.; Brener, R. Phys. Rev. B 1988, 37, 2266. (44) Brener, R.; Shechter, H. Phys. Rev. B 1989, 39, 2803.

Langmuir, Vol. 21, No. 20, 2005 9167

assembled monolayers (SAMs) composed of alkanethiols bound to Au(111) substrates via the terminal sulfur atom. Even on unreactive surfaces, however, the dissociation dynamics of IPC are complex and only poorly understood. Previous workers have measured the electron-induced cross-sections for CO elimination from IPC/Ag(111)22 and found values ranging from 1 to 15 Å2, depending on the incident energy. These relatively low cross-sections, even at higher incident energies, would require extensive irradiation to effect complete CO elimination from the film, with the associated risk of extensive damage to the substrate material. Our initial measurements confirmed the low sensitivity to incident electron beams for IPC/ Au(111) under the conditions previously employed22 for IPC/Ag(111) and prompted a series of studies aimed at understanding why the sensitivity of IPC is so dramatically reduced when adsorbed using standard protocols onto inert substrates. The companion paper I describes our structural studies of the as-deposited films of IPC on Au(111)/mica and alkanethiol SAMs; the electron-induced processes for IPC/Au(111) and IPC/SAM surfaces are described separately elsewhere.25,45 This work focuses on an important aspect of the adsorbed IPC films that is, we believe, the underlying cause for the reduced sensitivity of the films to low-energy electroninduced processes reported previously22 as compared to the isolated molecular species under gas-phase conditions.34,35 We show that IPC films undergo a structural transformation when heated above ∼100-140 K; this transformation is observed for all chemically inert substrates studied with only subtle variations. The structural transformation of these films constitutes an interesting subject insofar as it represents the interplay of moleculemolecule and molecule-surface interactions and, for our purposes, because this structural transformation has a profound influence on the electron-induced dissociation dynamics. In the context of the companion paper I, the results presented herein are important because they confirm that the as-deposited films (at ∼50 K) do not represent a thermodynamically equilibrated phase, despite the apparent spectral simplicity and narrow spectral bandwidths that these films exhibit. Experimental Procedures All aspects of the IRRAS-related experimental apparatus and protocols are described in the companion paper I. In the case of IRRAS results obtained during temperature ramps, a heating rate of 3 K/min was used and the reported temperatures are the average temperature taken during the acquisition of a given spectra and thus the average across a ∼3 K range. In addition, this system is equipped with a custom-built electron source to irradiate the surface with a 1-10 µA, 0-20 eV electron beam. The beam axis is perpendicular to the surface plane, such that the IRRAS signature of the sample can be measured before, during, and after the irradiation without physically moving the sample. Electron emission is from a Ta disk supported on a heated W filament (Kimball Physics), providing a relatively narrow energy distribution of ∼0.3-0.5 eV full width at half maximum (fwhm). The emitted electrons are directed toward the surface using three cylindrical lenses whose potentials are fixed with respect to the filament. Two orthogonal sets of deflection plates are used to sweep the electron-beam across the surface at ∼10 Hz to ensure uniform exposure. Electron energies are determined by the difference in potential between the center of the filament (i.e., the potential of the Ta disk) and the sample, which is maintained at -2 eV below the chamber ground to reject lowenergy electrons scattered from the inside of the electrostatic lens stack. The instantaneous electron-current absorbed by the sample is measured by a custom-built electrometer that floats (45) Pepin, C.; Rowntree, P. A. Manuscript in preparation.

9168

Langmuir, Vol. 21, No. 20, 2005

Hauchard et al.

on the sample potential, and the integration of this current versus exposure time provides the total electron dose.

Summary of Companion Paper (I) In I, we characterized the as-deposited IPC films using the same substrates employed in this work. When deposited on Au(111) or alkanethiol SAMs, the IRRAS spectra of IPC are dominated by strong AX bands absorbance at 2050-2065 cm-1, depending on the coverage and substrate. Thicker deposits favor narrower and higher frequency AX bands. The absorbance of this band grows approximately linearly with coverage beyond the first monolayer. In the 0-1 monolayer regime a weak band because of the doubly degenerate EQ vibrational modes is found at 2010-2012 cm-1. Unlike the AX band, the intensity of the EQ band does not change significantly beyond one monolayer, and its position varies only slightly with coverage or substrate. These results demonstrated that the IPC molecules in the first monolayer are adsorbed with one AX and two EQ carbonyl groups directed toward the surface; this geometry implies that the molecular C3 axes are inclined by ∼63° from the surface normal, as has been found for IPC/C(0001).42-44 However, even if all C3 axes were rigorously parallel and the IPC was arranged on a well-defined 2D lattice (i.e., in the absence of azimuthal and positional disorder, respectively), this monolayer structure does not correspond to a crystal plane of the semi-infinite solid as determined by X-ray diffraction.46 Spectroscopic modeling of the interactions among the transition dipole moments suggested that there is significant azimuthal disorder in the first layer structure under the current deposition conditions.21 The C3 axes for molecules in subsequent layers are oriented more parallel to the surface normal; this geometry leads to strong dipolar coupling of the AX-mode transition moment dipoles, leading to the observed profound blue shifts of the principal axial mode with coverage.47-49 The observation that the first and subsequent layers do not share a common molecular orientation is important and confirms that the as-deposited films are not structurally equivalent to the solid-state structure of IPC, in which all AX carbonyls are parallel. Although the C3 axes of the bulk solid phase also share a single orientation, there are significant differences between the structure of the second (and subsequent layers) and the bulk solid. For example, IPC in the bulk solid is found with only two possible rotations about this C3 axis; there is no reason to expect that such a highly ordered system would develop during low-temperature deposition on an azimuthally disordered first layer. More importantly, the bulk IPC layers perpendicular to the C3 axes are arranged in a complex ABCDEFGHIJ-ABCDEFGHIJ stacking sequence, with a repeat distance along the C3 axis of approximately 30 Å; this degree of longrange ordering of asymmetrical molecules for deposition at such low temperatures is not expected. Indeed, the spectral modeling presented in I shows that a simple fcc stacking sequence can explain most of the spectral properties of the as-deposited IPC films. As such, there is no a priori reason to conclude that the as-deposited physisorbed species represent a thermodynamically stable phase; this conclusion is reinforced by the results of the present study. A summary of the pertinent spectroscopic information for IPC and the Cn alkanethiol SAMs is included in the companion paper I. (46) Hanson, A. W. Acta Crystallogr. 1962, 15, 930. (47) Persson, B. N. J.; Ryberg, R. Phys. Rev. B 1981, 24 6954. (48) Persson, B. N. J.; Liebsch, A. Surf. Sci. 1981, 110, 356. (49) Ryberg, R. Adv. Chem. Phys. 1989, 1; Lawley, K. P., Ed.

Figure 1. IRRAS spectra of 40 L of IPC adsorbed on Au(111)/ mica, deposited at 45 K. The top curve shows the absorption spectrum of the as-deposited film, while subsequent spectra show the evolution of the absorption spectrum as a function of the temperature with a heating rate of 3 K/min. The arrows denote reproducible features identified in Table 1.

Results IPC was deposited and characterized on five types of substrates: Au(111), C4, C8, C12, and C16 SAMs, all at 50 K. All of the Cn SAM substrates have well-defined 3D structures with a collective molecular tilt of 25-30° from the surface normal50-52 and present highly uniform methyl terminations to the incoming IPC, the principal effect of the varying Cn SAM chain length is to vary the adsorbateAu(111) separation from 0.6 to 2.0 nm. In all cases, the results obtained using C4, C8, and C12 substrates (i.e., band positions, relative and absolute shifts, intensities, etc.) were intermediate between those of Au(111)/mica and those of C16 SAMs. To clarify the presentation, not all data is presented herein but will be discussed as required. Thermal Behavior. The temperature-dependent behavior of IPC layers was studied on the five substrates with a typical heating rate of 3 K/min; this slow rate allowed us to monitor the processes in real time using IRRAS. The topmost curves of Figures 1 and 2 present the IRRAS absorption spectra for 40 L (∼8 ML) of IPC adsorbed on Au(111)/mica and a C16 SAM, respectively. The details of these spectra are described in I. Extensive changes in the C-O stretching region are seen as the samples are heated; except for slight differences noted below, similar results are obtained for all substrates studied. The first consequence of the higher temperature is a decrease of the AX band intensity (∼2060 cm-1), accompanied by a slight broadening and a shift to lower frequency. For all samples, new infrared bands appear in the spectra in the 2000-2045 cm-1 region for T > 125 K; these bands rapidly increase in absolute intensity by as much as a factor of 5 (relative to the relatively featureless absorbance in this region at 50 K) as the temperature is increased to ∼140 K. Above 140 K, desorption causes the infrared signal to rapidly decrease with no significant additional changes in the band shapes; no significant optical absorption is detected above ∼160 K in the CO stretching region. In all cases, IRRAS spectra of the C-H stretching bands (2800-3000 cm-1) Cn SAMs above 160 (50) Truong, K. D.; Rowntree, P. A. J. Phys. Chem. 1996, 100, 19917. (51) Parikh, A. N.; Allara, D. L. J. Chem. Phys. 1992, 96, 927. (52) Zhang, L.; Goddard, W. A., III; Jiang, S. J. Chem. Phys. 2002, 117, 7342.

Thermal Transformations of Fe(CO)5

Langmuir, Vol. 21, No. 20, 2005 9169

Figure 3. Integrated intensity of the 1950-2100 cm-1 spectral region for IPC deposits on Au(111)/mica (9) and on C16 SAM substrates (b), expressed a function of sample temperature during a 3 K/min temperature ramp. A spline has been drawn through the data to guide the eye.

Figure 2. IRRAS spectra of 40 L of IPC adsorbed on a C16 SAM, deposited at 50 K. The top curve shows the absorption spectrum of the as-deposited film, while subsequent spectra show the evolution of the absorption spectrum as a function of the temperature with a heating rate of 3 K/min. The arrows denote reproducible features identified in Table 1.

K are identical to those obtained at the same temperature prior to IPC deposition, showing that desorption is complete. The RGA system detects desorbing IPC above ∼140 K, although the extensive fragmentation of the parent molecule in the ion source does not allow us to directly determine if the desorption is molecular in nature. Henderson et al.22 have determined that desorption of IPC from mono- and multilayers on Ag(111) is molecular, and we expect similar behavior for the chemically inert substrates employed in this work. Additional changes are observed in the weak completely symmetric vibrational mode that we believe is made infrared active by adsorption site perturbations (I). Whereas the as-deposited band at 2118 cm-1 has a fwhm of ∼5 cm-1, thermal transformation above 130 K causes a shift to 2115.5 cm-1 and a reduced fwhm of 2-3 cm-1. For spectra measured at or below the onset of rapid desorption, an isobestic point is observed at ∼2117 cm-1, indicating that the transformation between these two molecular states is quantitative and confirms that the desorption of the IPC molecules that give rise to this specific optical absorption is relatively unimportant below 140 K. For heating rates greater than the 3 K/min, the maximum intensity of the post-transformation spectra at ∼2045 cm-1 increases to approximately that of the original AX band at 2062 cm-1, with no apparent changes in the final band positions or the relative intensities of the transformed peaks with respect to the more slowly heated samples; the absolute intensities of the post-transformation features appear to be determined by the competition between the rate of the physical transformation and the slower kinetics of molecular desorption. As the initial IPC coverage decreases, the relative intensity of the 2045 cm-1 feature decreases more rapidly than that of the ∼2030 cm-1 features. At 5 L, the 2010, 2030, and 2044 cm-1 structures are of similar intensity following identical heating cycles; the principal change induced by the heating in this coverage regime is the loss of the developing high frequency AX feature. The transformation can also be observed in the integrated intensity of the 1950-2100 cm-1 region of the spectra, expressed as a function of temperature (Figure 3) for ∼40 L of IPC adsorbed on Au/mica (9) and C16 (b).

The integrated intensity at 50 K is similar for all substrates, and ∼5-10% decreases are always observed as the samples are heated to ∼100 K; these decreases of the integrated intensities parallel the drop in intensity of the principal AX band shown in Figures 1 and 2 are likely due to desorption of IPC from the surface. When deposited on Au(111)/mica, the integrated C-O stretching intensity presents an increase from ∼125-140 K; this increased intensity is also clearly visible when heating IPC/Cn samples above ∼125 K. In all cases, the increase of the integrated intensity from ∼125-140 K coincides with the extensive restructuring of the C-O stretching spectra. The magnitude of this increase is approximately proportional to the quantity of IPC deposited. The temperature for which this maximum integrated intensity is observed increases with coverage, with 5 L deposits showing a maximum near 132 K. In the case of Cn substrates, however, the restructuring of the IPC spectra is preceded by a smaller increase in integrated intensity from ∼100130 K (i.e., 25 K below the above-mentioned “main” restructuring temperature seen on all substrates); it is difficult to identify specific spectral changes that correspond to the first low-temperature intensity increase in the IPC/Cn samples, indicating that it is due to changes of relatively broad spectral features. The relative intensity of the two features in the integrated intensity profiles for Cn SAMs is dependent upon the heating rate and the quantity of adsorbed IPC, with faster rates and thicker deposits favoring the visibility of the second increase. Figure 3 indicates that the “pretransformation” observed on SAM surfaces induces the transformed IPC state at lower temperatures than found on Au(111) surfaces, thus reducing the desorption losses of weakly bound IPC molecules; this may explain the higher intensities at 130145 K found on the organic supports. Up to seven distinct bands in the 1950-2100 cm-1 spectral window can be assigned in the “transformed spectra” observed from ∼125-140 K; the relative intensities of these bands depend on the substrate and the initial film thickness. Because of the competition between structural rearrangement and desorption, the absolute intensities should be considered as qualitative but the trend is that the post-transformation spectrum has the lowest intensity on the Au(111) and on the longer chain Cn substrates. The frequencies and intensities of the bands observed on the five substrates are shown in Table 1 for an initial deposit of 40 L. It is clear that with minor deviations, the same series of peaks are obtained upon heating to ∼140 K for all substrates, suggesting that the mechanism and final state of the transformation are similar in all cases.

9170

Langmuir, Vol. 21, No. 20, 2005

Hauchard et al.

Table 1. Infrared Frequencies and Absorption Intensities Observed in the “Transformed Spectra” of 40 L of Fe(CO)5 on Five Different Substrates, Using a 3 K/min Heating Ratea Au(111)

C4

C8

C12

C16

2044 cm-1 33 mAbs

2049 cm-1 20 mAbs

2047 cm-1 44 mAbs

2042 cm-1 35 mAbs

2044 cm-1 28 mAbs

2036 cm-1 24 mAbs

2037 cm-1 49 mAbs

2037 cm-1 50 mAbs

2036 cm-1 34 mAbs

2036 cm-1 26 mAbs

2032 cm-1 21 mAbs

unresolved

2030 cm-1 42 mAbs

2030 cm-1 31 mAbs

unresolved

2027 cm-1 21 mAbs

2025 cm-1 53 mAbs

2026 cm-1 39 mAbs

unresolved

2029 cm-1 24 mAbs

2014 cm-1 8 mAbs

2014 cm-1 34 mAbs

2016 cm-1 30 mAbs

2017 cm-1 15 mAbs

2008 cm-1 6 mAbs

1991 cm-1 3 mAbs

1995 cm-1 5 mAbs

1997 cm-1 4 mAbs

1996 cm-1 3 mAbs

1996 cm-1 3 mAbs

1979 cm-1 2 mAbs

1979 cm-1 4 mAbs

1979 cm-1 3 mAbs

1979 cm-1 2 mAbs

1980 cm-1 2 mAbs

a The pretransformation absorption of the dominant axial band was 70-80 mAbs for these studies.

Figure 4. C-H stretching spectra of 50 K C16 substrates during IPC deposition. (A) 50 K pre- and postdeposition absorption spectra are shown as the solid and dashed lines, respectively. The reference spectrum for both results is that of the “direct beam” passing in front of the surface. (B) ∆Absorption spectra during the deposition process for 6, 10, 30, and 40 L of total coverage. The reference for these spectra was the reflectivity of the pristine 50 K C16 SAM prior to IPC deposition.

Spectral Response of Cn Substrates to Fe(CO)5 Deposition. The deposition and thermal transformation of IPC on Cn SAMs have observable consequences for the IRRAS spectrum of the C-H stretching region (28003000 cm-1). The solid line in Figure 4A shows the IRRAS spectrum of a ∼50 K C16 SAM, obtained using the “direct beam” reference, showing the five peaks (d(, r(, r+ FR) characteristic of long chain organic monolayers.21,53,54 The dotted curve in Figure 4 shows the spectrum obtained after a 40 L exposure of IPC onto the methyl-terminated film. The absorption bands corresponding to the methylene groups (d+, 2850 cm-1; d-, 2919 cm-1) within the organic substrate show no evidence of perturbations by the (53) Nuzzo, R. G.; Korenic, E. M.; Dubois, L. H. J. Chem. Phys. 1990, 93, 767. (54) Garand, E., Picard, J.-F.; Rowntree, P. J. Phys. Chem. B 2004, 108, 8182.

molecular adsorption, while each of the methyl-derived -1 -1 bands (r+, 2875 cm-1; r+ FR, 2936 cm ; ra,b, 2965 cm ) show systematic decreases in intensity when in intimate contact with the growing IPC layer. In addition, the strong symmetric methyl vibration band (r+) has shifted by ∼ -3 cm-1 with respect to the pristine SAM. The decrease in the r+ band intensity with IPC coverage follows a logarithmic trend with coverage, typical of a saturation phenomenon.55,56 To more clearly observe the changes induced by the deposition of IPC, the results are replotted in a ∆Absorbance representation (Figure 4B), using the predeposition reflectivity of the SAM as the reference signal; positive signals correspond to increased optical absorption induced by the physisorption of the IPC, and negative signals are due to decreased optical absorption. Decreasing intensities of the r+ (2876 cm-1), r+ FR (2939 -1) bands are clearly / r (2955 and 2964 cm cm-1), and ra b observed; the decreases seen in the differential presentation (Figure 4B) occur precisely at the original peak positions of the pristine organic substrate in the absorption spectra (Figure 4A), and a slight increase of absorption is observed at 2871 cm-1, slightly below the original r+ peak position. The frequencies of these maxima and minima in the differential presentation do not vary with increasing IPC coverage, as would be expected if they were due to a simple progressive shift of the r+ band to lower frequencies; the observed displacement of the r+ band to lower frequencies in the absorption spectra of Figure 4A is therefore the net result of an simultaneous in place decrease of the original peak at 2876 cm-1 and the growth of a new feature at 2871 cm-1. Depositions of IPC on C4, C8, and C12 SAMs exhibit identical trends of the methyl bands as presented here for C16 substrates. This phenomena is discussed elsewhere,55-57 but for the purpose of this work, we note that the changes of the methylderived band intensities when in intimate contact with adsorbates appear to be ubiquitous; Engquist and Liedberg57 observed similar spectral transformations using H2O and D2O deposition on methyl-terminated SAMs, and we have found qualitatively similar results with CO2, SF6, Xe, acetonitrile, acetone, benzene, and other adsorbates.55,56 Engquist and Liedberg57 proposed that the dipolar interaction between H2O adsorbates and the methyl chain terminations is responsible for these spectral modifications; our results55,56 with a wide range of adsorbates suggest that a more universal process is operant in many, if not all, cases. Regardless of the origin of this phenomenon, however, it is highly reproducible and provides a sensitive test of the environment of the buried IPC/Cn interface and, in particular, the perturbation of this interface during structural changes of the adsorbates. We use the spectral response of the methyl termination to the presence of adsorbates as a probe that nondestructively interrogates the buried IPC/Cn interface. In the temperature regime that causes changes to the C-O stretching spectra (i.e., 100-140 K), there are also significant changes in the methyl stretching spectra for the IPC/Cn systems, suggesting that the buried IPC/Cn interface is evolving as well. Figure 5 shows the evolution of the absorbance and the frequency of the methyl-derived r+ band (∼2875 cm-1) as a function of the temperature during the initial cooling of the pristine C16 SAM and the heating of the adsorbed IPC layer. All data in this figure were referenced to the “direct beam” to measure absolute (55) Pe´pin, C. M.Sc. Thesis, Universite´ de Sherbrooke, Que´bec, Canada, 2001. (56) Pe´pin, C.; Rowntree, P. A. Manuscript in preparation. (57) Engquist, I.; Liedberg, B. J. Phys. Chem. 1996, 100, 20089.

Thermal Transformations of Fe(CO)5

Figure 5. Behavior of the r+ absorption band of the C16 SAM during a deposition-heating-desorption cycle. (A) Intensity of the r+ band. (B) Position of this peak. Arrows identify (1) the variations observed for the pristine C16 SAM as the temperature is reduced from 300 to 50 K, (2) the 50 K adsorption of IPC onto the C16 SAM, (3) the pretransformation heating of the IPC/C16 system, (4) the transformation region of the IPC film, and (5) the heating and desorption of the fully transformed IPC/SAM system. The vertical line at 140 K identifies the temperature for which the integrated intensity of the IPC deposit is at a maximum.

intensities of the substrate and to avoid temperaturedependent reference spectra. Again, no adsorbate-induced changes were observed for the methylene-derived spectral bands (d(), because these bands were not sensitive to the presence of the adsorbate (e.g., parts A and B of Figure 4). The r+ intensity and position vary in a quasilinear manner as the sample is cooled from 300 to 50 K (0, arrow 1); the net shift from ∼2877 cm-1 to ∼2875 cm-1 and the ∼50% increase in intensity are due to reversible intramolecule effects within the SAM constituents.54 Deposition of 20 L of IPC onto the C16 substrate (arrow 2) at 50 K induces the loss of intensity and further red shift of the r+ peak, as shown previously in Figure 4 and described above. As the sample is heated from 50 to ∼100 K (arrow 3), no changes are observed in the r+ stretching band; recall that the only changes observed in this temperature regime for the C-O bands were a slight shift of the AX peak and a decrease of intensity because of desorption of weakly bound IPC. In the absence of the IPC adsorbate, a blue shift and a decrease of intensity would be observed across this temperature range; the temperature invariance of the r+ band in this regime indicates that the as-deposited IPC strongly perturbs the intramolecular couplings that are key features of the r+ spectral signature in C16 SAMs.54 From 100 to 140 K (arrow 4), the r+ band shows a rapid increase in intensity and a shift to higher frequency; recall that heating pristine SAMs causes lower band intensities. At 140 K, the temperature at which the transformation in the C-O stretching region is complete and the integrated intensity of the transformed IPC film is at a maximum, the absorbance and frequency of the r+ band have been completely restored to the values measured at this temperature in the absence of the IPC film. This indicates that the perturbation of the methyl-terminated organic substrate by the transformed IPC has been reduced to negligible levels, despite the continued presence

Langmuir, Vol. 21, No. 20, 2005 9171

Figure 6. IRRAS absorption spectra of a 20 L IPC deposit during a heating-transformation-cooling cycle, starting at the instrumental base temperature (top curve) up to the transformation temperature of ∼140 K and then recooling the sample. The temperature identified for each spectrum is the average across a 2-3 K window during data acquisition. The heating rate was 3 K/min, while the cooling rate was instrumentally limited to ∼1 K/min.

of virtually all of the IPC on the surface. Further heating above 140 K shows no additional structure in the r+ band except for that expected on the basis of the reversible changes of the C16 film without interference by the IPC, as evidenced by the close agreement of the two data sets in the 140-300 K regime (arrow 5); even the complete desorption of the adsorbed multilayer quantities of IPC at ∼160 K has no discernible effects on the r+ band intensity or position. Irreversibility of Thermal Transformation. To determine the physical nature of the process that accompanies the spectral transformations at ∼125-140 K (100-140 K for Cn substrates), Figure 6 presents a series of results obtained by heating a 20 L IPC/Au(111)/mica sample to induce the spectral transformations and then cooling the system before significant desorption takes place. The topmost four spectra (50, 90, 102, and 135 K, respectively) taken during the heating cycle are very similar aside from the slight decrease in intensity for the AX band with heating as described above. The fifth curve shows the spectrum of the completely transformed sample at 139 K, with four principal bands at 2044, 2032, 2027, and 2013 cm-1. Recooling the sample to 50 K leaves the C-O stretching region in its transformed state. Extremely slow cooling rates did not change this behavior, suggesting that the reverse transformation to restore the as-deposited spectrum is not kinetically limited. The transformation that takes place in the IPC infrared spectrum during the heating process is clearly irreversible with temperature. Identical results are obtained using Cn substrates; the only changes observed in the C-H stretching region during this recooling process are the reversible increases to the C16 SAM bands caused by the decrease in temperature. The absorbance of the Cn film in the presence of the fully transformed IPC film is indistinguishable from that of a pristine Cn film at the same temperature. The optical absorbance of the transformed film is independent of the temperature in the 50-140 K regime, indicating that the changes reported in Figures 1 and 2 are only due to the physical transformation and desorption processes of the film. These measurements demonstrate that the thermally transformed IPC films are extremely stable and impose

9172

Langmuir, Vol. 21, No. 20, 2005

Figure 7. IRRAS ∆absorbance spectra of sequential 20 L deposits of IPC/C16. The lower curve shows the spectrum of the first deposit at 50 K prior to sample heating and transformation; the reference for this spectrum is the reflectivity of the pristine C16 surface. The upper curve shows the spectrum obtained following the deposition of a further 20 L of IPC onto the thermally transformed 20 L film; the reference for this spectrum is the reflectivity of the thermally transformed first deposit of the 20 L IPC sample.

negligible perturbations on the methyl terminations of the SAM substrates; all evidence indicates that the details of the substrate structure and composition have little effect on the adsorbed films or their subsequent thermal transformations, and we therefore conclude that the interaction of the transformed films with the Au(111) substrates are sharply reduced from the as-deposited state. To study the structure of the transformed IPC layer, an additional 20 L of IPC was deposited on a thermally transformed 20 L IPC/C16 film at 50 K. Figure 7 shows the C-O stretching spectrum obtained after the first 20 L exposure onto the clean C16 substrate (bottom curve) and that of the second 20 L deposit on the thermally transformed IPC/C16 film (top curve); for ease of comparison, the references for these data were the reflectivity of the substrate just prior to the deposits, and the spectra therefore exclusively contain information regarding the changes induced by the most recent IPC deposit. These spectra are almost identical, with only minor differences observed from 2020 to 2055 cm-1 (the region of the maxima in the transformed IPC spectra). As shown in I, IPC films are strongly affected by intermolecular couplings of the transition dipole moments, and the oscillations from 2020 to 2050 cm-1 show the coupling of the vibrations of the newly deposited IPC with that of the underlying aggregates. The evolution of the C-O spectral region during this second deposition (not shown) shows precisely the same initial spectra, band shifts, peak narrowings, etc. that were observed during the first 20 L deposition on the clean organic film (e.g., as shown in Figures 1-3 of I), with precisely the same sharp decrease in the r+ band of the C16 substrate as shown in Figures 4 and 5. We have repeated the process of deposition/transformation/redeposition 3 times on one sample, and for each cycle, the state of the organic substrate is restored to its apparently unperturbed state by the thermal transformation step, despite the ever-increasing quantity of adsorbed IPC. Identical results have been obtained for the redeposition of 20 L of IPC onto a transformed IPC/Au(111) substrate, again showing that the choice of the substrate does not dramatically influence the nature of the transformed state. Effect of Thermal Transformation on Dissociation Processes. The previous discussion has focused on the physical modifications of the IPC film that accompany heating to ∼140 K; equally important are the changes to the dynamical properties of the system. As mentioned in the Introduction, one of the underlying motivations for this work was to understand the correlations between the physical state of the surface and the molecular dynamics

Hauchard et al.

Figure 8. Effect of 5 eV electron irradiation on 20 L IPC/C16 deposits. (A) Depletion of the sharp AX absorption band at 2060 cm-1, following an electron dose of 0.05 mC/cm2. (B) 100fold increased irradiation dose has little perceptible effect on an equivalent quantity of IPC/C16 following the thermal transformation.

that can be induced by low-energy electron scattering. The sensibility to electronic irradiation was measured on Au(111) and C16 for an as-deposited and a thermally transformed IPC film at 5 eV. At this energy, it has been shown by Olsen et al.58 that the organic substrate is unaffected by low-energy electron irradiation. The results obtained for 20 L of IPC/C16 are presented in Figure 8. The as-deposited IPC film (before transformation) is extremely sensitive to electron irradiation, as would be expected on the basis of the extreme sensitivity of gasphase IPC to low-energy electrons.34,35 After a 0.05 mC/ cm2 dose, the AX band has lost ∼50% of its intensity and a shoulder has appeared above the main band at ∼2070 cm-1. This decrease in the C-O stretching band intensity is caused by the dissociation of Fe-CO bonds via two-step dissociative electron attachment (DEA) to the carbonyl

Fe(CO)5(ads) + e- f Fe(CO)5-(ads) Fe(CO)5-(ads) f Fe(CO)5-n-(ads) + nCO(g) and subsequent desorption of CO (CO is not bound to Au or organic surfaces at 50 K); the quantity of CO eliminated by the electron attachment process (i.e., the value of n) increases with increasing energy in the gas phase, albeit with decreasing dissociation probabilities.34,35 Details of this process in IPC films and its spectral consequences are discussed elsewhere.25 The thermally transformed IPC surface is much less sensitive to electronic irradiation, as shown in Figure 8B; the infrared signal of the CO stretching region shows only minor changes and less than a 1% loss of overall intensity following irradiation of a 50 K “transformed” sample with a 5.0 mC/cm2 dose (i.e., 100 times greater than that of the identical protocol on the as-deposited IPC layers). Similar results were obtained using electron energies from 1 to 50 eV. We estimate that the apparent sensitivity of the IPC film to the incident electron flux has been reduced by at least a factor of 103 during the thermal transformation, over a range of only ∼30 K. Discussion As discussed in I, the similarity of the spectra for lowcoverage IPC adsorbed on the metal and organic SAM (58) Olsen, C.,; Rowntree, P. A. J. Chem. Phys. 1998, 108, 3750.

Thermal Transformations of Fe(CO)5

substrates indicates that the environment of the asdeposited films is remarkably similar for all substrates and there is no evidence for molecular fragmentation or structural modifications upon adsorption. Neither of these observations is surprising for chemically inert supports, but they provide evidence that the substrates are almost passive spectators to the adsorption processes. Heating of the IPC adsorbed layer induces a shift of the C-O stretching bands to lower frequencies as well as an increase in the integrated intensity of the C-O stretching region. Tanabe et al. have also reported anomalous spectral shifts under these conditions for IPC adsorbed onto evaporated Fe films;59 the nature and details of the transformation were not explored because their interest was the photochemistry of the as-deposited layers. Although the changes in the spectra prior to desorption are most evident for multilayer deposits, changes are detected even for monolayer quantities. Insofar as the structural model presented in I suggests that the molecular axes of all but the first layer are preferentially oriented along the surface normal (which maximizes the AX band absorption), an increased IRRAS intensity can only arise by increased adsorption by the EQ mode at lower frequencies. This requires significant molecular reorientation, induced by the enhanced translational and orientational mobility prior to complete desorption at 160 K. This reorientation and translation would modify the repulsive interactions of the transition dipole moments of IPC and thus reduce the blue shifts caused by this dynamic coupling.21,47-49 Simple reorientation relative to the surface normal will also affect the relative intensities of the AX and EQ bands, according to the metal surface selection rule;60,61 unfortunately, there is no direct information available to determine if the C3 axes are aligned with some preferential angle with respect to the surface normal following aggregation. Thus, the spectral intensity changes of the IPC upon transformation are difficult to interpret with precision. Similar results were observed on the organic SAMs showing that the transformation is not induced by the metal substrate or specific molecule-surface interactions. Furthermore, the thermal transformation induced in the adsorbed IPC layer is irreversible (Figure 6), indicating that the observed transformation does not correspond to an equilibrium phase transition. Together, these results demonstrate that the as-deposited layered structure described above and characterized in I is not the thermodynamically stable structure at low temperatures, even though the short-range structure of these layers is closely related to the known crystal structure.21 We presume that the inadequate lateral diffusion coefficient of the IPC molecules at 50 K during deposition leads to this out-ofequilibrium structure that lacks long-range order. The role of surface mobility in forming stable structures/phases is evidenced by the integrated intensity of the 1950-2100 cm-1 region during heating, as seen in Figure 3. The first evidence of the transformation on organic substrates is at 100 K; this coincides with the onset of the recovery of the free-methyl spectrum of the substrate (Figure 5). This initial process is absent for IPC adsorbed on the Au(111) surface and suggests that the barrier to molecular transport at the IPC-C16 interface is lower than in the bulk IPC film or at the IPC/Au interface. It is interesting to note that these lower barriers facilitate the spectral transformations and thus reduce the desorption losses of (59) Tanabe, T.; Morisato, T.; Suzuki, Y.; Matsumoto, Y.; Wadayama, T.; Hatta, A. Vib. Spectrosc. 1998, 18, 141. (60) Chabal, Y. J. Surf. Sci. Rep. 1988, 8. (61) Greenler, R. J. Chem. Phys. 1969, 50, 1963.

Langmuir, Vol. 21, No. 20, 2005 9173

weakly bound IPC. The observations that (1) the thermal transformation is activated ∼20-40 K below the desorption temperature and (2) the rate of transformation is more strongly affected by the heating rate than the rate of desorption indicates that the energetic barrier to transformation is lower than that of molecular desorption, as expected for a diffusion-limited process. The similarity of the as-deposited layer structure and density to the 3D crystal structure indicates that the energetic driving force for the transformation is not due to subtle molecular reorientations. The studies carried out on organic substrates show that the interactions between the terminal methyl groups and the IPC molecules sharply decrease as the transformation of the IPC spectrum proceeds (Figure 5) and indicate that a large fraction of the methyl-terminated surface is no longer covered by the IPC film. These results are consistent with the formation of solid 3D IPC clusters on the surface when the surface temperature activates lateral diffusion, just prior to the desorption temperature. Similar aggregation processes have been reported by Engquist and Liedberg57 for H2O adsorbed on SAM surfaces principally based on the spectral response of the terminal methyl groups to the changing CH3-H2O interactions that accompany nucleation and crystallization. This hypothesis is supported by the observation that, when additional IPC is deposited on the structurally transformed film (Figure 7), the spectral evolution of the C-O stretches for the newly deposited molecules (and of the CH3 modes for SAM substrates) follow exactly that of the initial deposition on the pristine substrates; prior to the second deposition, the vast majority of methyl groups are in fact unperturbed and thus present identical IRRAS signatures as the pristine SAM. A crude estimate of the size and surface density of these aggregates can be made by considering the recovery of the r+ band as a measure of the exposed organic surface that is formed by the structural transformation; in the following, we assume a monodispersed size distribution of hemispherical clusters for convenience. Assuming (1) that desorption has not significantly reduced the quantity of IPC adsorbed on the surface (complete desorption of a 30 L transformed IPC film requires approximately 9 h at 136 K) and (2) that the density of the as-deposited film is similar to that of the aggregates, the conservation of mass requires that VHS, the volume (per cm2 of substrate surface area) of n clusters, each of which has a radius of RHS, is equal to Vfilm, the volume (per cm2) of an l monolayer as-deposited film, each layer of which has a thickness t. AHS, the area occupied by these clusters (per cm2), is trivially defined and corresponds to the fraction of the substrate that is in contact with the transformed clusters

14 VHS ) n πR3HS 23

(

)

Vfilm ) lt 3

∴RHS(n) )

3lt x2nπ

AHS(n) ) nπR2HS

(3lt2 ) /

∴AHS(n) ) n1/3π1/3

2

3

This simple model shows that, for a given surface coverage, AHS, the areal density of clusters decreases for the transformation of increasingly thick films. For the purpose of the present estimation, the thickness of one

9174

Langmuir, Vol. 21, No. 20, 2005

Figure 9. Estimates of the radius (negative slope lines, expressed in nanometers) and fraction of surface occupied by hemispherical clusters of IPC (positive slope lines), expressed as a function of the number of clusters per cm2. The top and bottom panels show the characteristics for monodispersed hemispherical clusters formed by heating 1 and 5 layers of IPC, respectively.

layer is taken to be 0.3 nm.21,46 Figure 9 shows the behavior of RHS and AHS, expressed as a function of n, corresponding to the transformed clusters for 1 and 5 IPC layers. We estimate (on the basis of the available signal-to-noise ratios and the reproducibility of the data) that the IRRAS r+ signal has recovered at least 95% of its predeposition intensity during the formation of the solid clusters, leaving less than 5% of the surface in contact with the IPC clusters; this imposes an upper limit of AHS ) 0.05 for each system studied. In the case of the 1 layer model, this corresponds to an upper limit of ∼2 × 1010 clusters/cm2 and a lower limit cluster radius of ∼9 nm; transformation of 5 layers of IPC would produce a maximum of 7 × 108 clusters/cm2, with a minimum cluster radius of ∼45 nm. The tendency to form fewer (and larger) clusters for the thicker films may be the underlying reason that the thicker films show a maximum of integrated intensity at higher temperatures than thin films, because more extensive mass transport is required for the thick film systems. It is clear, however, that the conclusions of this approach are highly sensitive to the unknown morphological properties of the aggregates; in situ observation of the clusters would be of great use in this regard. Physisorbed systems may undergo a dewetting transition as the temperature is reduced, typically at or near the triple-point temperature of the adsorbate.1-5,27,28 However, this type of equilibrium wetting transition is clearly different from the transformation reported herein, where heating above ∼132-140 K causes the system to form 3D nonwetting clusters. The triple-point temperature of IPC is 252 K,43 and it is readily seen that a droplet of liquid IPC completely wets a clean Au(111)/mica surface at ∼300 K. It is not possible in the present UHV system to study the IPC adsorbate structure near the triple-point temperature. Interestingly, liquid IPC does not wet a C16 SAM at all at 300 K, showing that a more detailed exploration of all interactions is required to understand such phenomena, even for simple molecular adsorbates such as IPC.

Hauchard et al.

The spectral modeling of the IPC/Au(111) system by Sato et al.18 derived the refractive index profile from a “reference spectrum” of multilayer IPC adsorbed on sapphire and SiO2/sapphire surfaces at 130-140 K. Our transmission results show that the EQ band of the IPC/ sapphire at 126 K has shifted by +7 cm-1 and the AX band intensity is strongly reduced, relative to the transmission spectrum at 50 K; the EQ band shift and AX intensity reduction continue until ∼131 K, at which point desorption begins. Thus, our results indicate that heated IPC/sapphire aggregates in the same manner as the IPC/ Au and IPC/Cn surfaces and the peaks observed following the heating cycle are significantly red-shifted with respect to those of the as-deposited films. This thermally induced restructuring appears to be a universal characteristic of IPC on chemically inert surfaces such as Au, Cn SAMs, or sapphire. As such, the inability of the spectral modeling18 to reproduce the large blue shift and narrow peaks associated with the as-deposited IPC films may be partially due to the nature of the reference spectrum employed. In addition, as these authors point out, their spectroscopic modeling is predicated upon the assumption that molecules in multilayer structures maintain the orientations of the monolayer constituents; this is inconsistent with the results of I. The thermal transformation described above induces a very large decrease in the apparent sensitivity to lowenergy electrons. Two possible structure-related mechanisms can be envisioned. If the incident electron flux can penetrate into transformed, aggregated material, caging in the compact solid structure may prevent CO ejection and facilitate recombinative regeneration of the IPC with no significant spectral consequences. Trushelm and Jackson62 have reported this process in their study of the photochemistry of IPC/silica, where the crystallization of the adsorbed IPC accompanied a decrease in the sensitivity to photons. Alternatively, if electron transport into or within the solid structure is reduced because of surface charging or high-probability reactive or inelastic events, the majority of the IPC would not be exposed to the electron flux and little dissociation would be observed. We favor the second interpretation because the measured crosssections for as-deposited layers are spectacularly large (∼10-300 Å2, depending on incident energy) and because the recombination reaction implicit in the first mechanism would have to involve the elimination of the bound charge on the Fe(CO)5-n- fragment, because IPC does not support a stable anion state.33,35 It is interested to compare the high sensitivity to the incident electron beam shown in Figure 8 to the TPDbased results of Henderson et al.;22 they found that the cross-sections for the decarbonylation of IPC monolayers adsorbed on Ag(111) surfaces were 1-2 Å2 for 3-8 eV electrons. This is much lower than our results25 and the results that the gas-phase sensitivity34,35 would indicate. Their monolayers were prepared by flash-heating a multilayer deposit to 157 K to desorb all but the first layer. As has been shown above, however, such a protocol may lead to unwanted molecular aggregation on the surface, thus reducing the dissociation cross-sections; although we have not studied the IPC/Ag(111) system directly, the present results indicate that the thermodynamic instability of the as-deposited physisorbed state may be a widespread characteristic. Mu¨lbauer et al.39 have recently used electron energy loss spectroscopy to study the decomposition of IPC on Si(111) for sample temperatures between 100 and 400 K. (62) Trushelm, M. R.; Jackson, R. L. J. Phys. Chem. 1983, 87, 1910.

Thermal Transformations of Fe(CO)5

As expected,11,36 the initial exposure to IPC at 100 K induces molecular fragmentation and the adsorption of vertically aligned Fe-CO, presumably with the Fe bonded to the substrate. A drastic decrease in the scattering loss intensities of single-loss events was observed following ∼15 L exposure (corresponding to ∼3 ML according to our estimates); multiple loss events showed a less significant reduction, indicating that multiple-scattering processes were more probable after the intensity loss. Similar effects were observed when the sample temperature exceeded ∼160 K. They concluded that a surface reorganization process was involved in this dramatic redistribution of spectral densities, but the exact nature of the process was not determined. We can speculate, on the basis of the present results, that, as the exposure increases above ∼3 ML at 100 K, any IPC that is physisorbed onto the FeCO passivated Si(111) surface could structurally transform in much the same manner as we see for multilayers on Cn SAM surfaces at T ∼ 100 K; electron scattering in the 3D aggregates may explain the relatively strong multiple scattering events observed in the work of Mu¨lbauer et al.39 Heating above the 100 K base temperature would accelerate this effect, as shown in Figures 1 and 2. As shown in Figure 8, electron interactions with the transformed aggregates are significantly weaker than those with the layered systems, which may be related to the dramatic collapse of the scattering intensities that these workers reported. Conclusions The results presented herein demonstrate that asdeposited IPC layers are subject to significant structural

Langmuir, Vol. 21, No. 20, 2005 9175

changes when heated above 100-125 K. The thermally induced transformation is witnessed by changes to the IPC spectrum, changes to the spectrum of the substrate in the case of Cn substrates, and by the marked decrease in the sensitivity to low-energy electron irradiation. These changes are irreversible and are not kinetically limited, indicating that the transformation is not a phase transition. We interpret the post-transformation spectra as characteristics of large 3D aggregates that reduce the interactions with the substrate while maximizing the interadsorbate interactions; the transformed system does not wet the Au(111) or the Cn SAM substrates at low temperatures. We suggest that the metastable structures adopted by the as-deposited layers are the result of rapid energetic accommodation of the incoming IPC particles to the temperature of the surface, prior to global energy minimization that would normally produce the thermodynamically stable ground state.

Acknowledgment. This work has been funded by NSERC (Canada) and FQRNT (Que´bec). The authors thank Prof. H. Taub (University of MissourisColumbia) for providing details of his unpublished work on the IPC/ C(0001) system. The expert assistance of D. Poulin and J.-L. Be´dard (mechanical workshops, Universite´ de Sherbrooke) in the development of this instrumentation is gratefully acknowledged. LA050678Y