Two - to three - dimensional transition intermediate : growth of ž / ž / benzene on Ru 001 and Mo 100

Ž . Ž . The complex phase transitions of physisorbed benzene on Ru 001 and Mo 100 have been investigated using Ž . Ž . temperature programmed desorption TPD , metastable impact electron spectroscopy MIES and work function measurements. The phase transition from benzene molecules whose planes are parallel to the surface to molecules with their planes approximately perpendicular to the surface is governed by a reversible two-dimensional gas–solid transition in the first physisorbed benzene layer. Furthermore, an additional phase of loosely bound benzene molecules, which can be considered as a twoto three-dimensional transition intermediate, is identified. q 1999 Elsevier Science B.V. All rights reserved. The condensation of benzene as a model for the formation of physisorbed multilayers has attracted considerable interest for more than a decade. Benzene is of particular interest because of its anisotropic but simple structure, i.e. the disc shape molecules are nearly isotropic in one plane but strongly anisotropic with respect to the plane normal. Therefore the condensation of benzene as a simple structured representative of a class of soft, condensed matter can provide information about processes such as selfŽ . organization, two-dimensional 2D phase transitions and the interplay between twoand three-dimenŽ . w x sional 3D phases 1 . The present study will take advantage of these characteristics to provide a more in-depth view of the surface dynamics during the initial stages of ) Corresponding author. Fax: q1 409 845 6822; e-mail: goodman@chemvx.tamu.edu benzene condensation. Particular attention to the transition from a 2D to a 3D system will provide a more detailed understanding regarding the interplay of the surface–adsorbate and adsorbate–adsorbate interaction during the nucleation of 3D, bulk-like structures. w x In this context, Polta and Thiel 2 and Jakob and w x Menzel 3 have published extensive thermal proŽ . grammed desorption TPD measurements of benŽ . zene adsorbed on Ru 001 and established an adsorption model consisting of three surface phases, which has been found to be valid for many other transition w w x x metal surfaces see Refs. 4–6 for a review . At Ž coverages up to ;1.0 monolayer 1 ML, which is the amount needed to saturate the first chemisorbed . benzene layer , the molecules chemisorb with their ring planes essentially parallel to the substrate surface and are tightly bonded. For that reason, this first chemisorbed layer saturates the active bonds of the surface, but is not directly involved in the surface 0009-2614r99r$ see front matter q 1999 Elsevier Science B.V. All rights reserved. Ž . PII: S0009-2614 99 00476-5 ( ) J. Gunster et al.rChemical Physics Letters 306 1999 335–340 ̈ 336 dynamics of the subsequently physisorbed benzene layers. On top of the closely packed and uniform Ž . chemisorbed layer a physisorbed layer a con1 denses with the same orientation and with approximately the same molecular density. Further adsorpŽ . tion leads to a layer a with a more dense satura2 tion coverage and with the benzene molecules standŽ ing edge-on ring planes approximately perpendicu. lar to the surface . Finally, additional adsorption leads to the formation of bulk-like, randomly oriŽ . ented benzene a . 3 Following the pioneering work of Jakob and Menw x zel 3,7 many investigations have been devoted to this subject. A variety of methods have been used to identify the proposed phase transitions and to provide information about the role of the substrate–adŽ sorbate interaction in the phase transitions see Ref. w x . 8 for a review . Jakob and Menzel have recently offered a refined model which takes into consideration a reorganization of the first physisorbed layer a1 and a disordered amorphous layer a X at the ben2 zene–vacuum interface. However, there still exists considerable confusion regarding the appearance of the a desorption feature in TPD. As discussed in 2 w x Ref. 6 the 1.8:1 ratio of the a ra TPD peak 2 1 areas, which suggests the formation of the a phase 2 on top of the saturated a phase, is not consistent 1 Ž with recent IRAS infrared reflection adsorption . spectroscopy data. Those data suggest significant reorientation of the molecules in the saturated a1 phase with additional benzene adsorption. In order to provide more detailed information about the dynamics of the phase transitions a TMa 1 2 and a TMa , we have re-examined benzene adsorp2 3 Ž . Ž . tion on Ru 001 and Mo 100 . The present study provides evidence that the appearance of the edge-on a phase is governed by a reversible 2D gas–solid 2 transition in the first physisorbed benzene layer and that the amount of benzene necessary to form the a2 edge-on phase does not correspond to the fully deŽ . veloped a desorption feature in TPD. On Ru 001 2 the saturated a phase yields an a peak intensity 2 2 essentially identical to a in TPD, i.e., a ra f1. 1 2 1 The remainder of the a desorption feature is formed 2 Ž . by an additional phase b of loosely bonded benzene molecules. In this context it is important to mention that the new b phase should not be conX w x Ž fused with the a 2 phase mentioned in Ref. 6 see . discussion section . In accordance with the irreversibility of the a to a phase transition, we 2 3 describe this phase as a 2D to 3D transition intermediate. w x In a recent study 9 we have shown that the extremely surface sensitive electron spectroscopic technique, metastable impact electron spectroscopy Ž . MIES , is able to identify in situ, the formation of Ž . the upright a edge-on phase on the outermost 2 Ž surface during benzene adsorption. For an introduction to MIES and its various applications in molecular and surface spectroscopy see the recent review w x. from Harada et al. 10 . By combining MIES with TPD we are able to correlate the development of the Ž . edge-on a phase on the surface and its desorption 2 feature in TPD. In order to investigate the influence of the substrate–benzene interaction on the adsorption dynamics we have deliberately chosen to comŽ . pare the highly reactive Mo 100 surface to the less Ž . Ž . reactive Ru 001 surface. In addition, the Ru 001 substrate serves as a reference, since detailed benzene adsorption experiments have been performed by w x Jakob and Menzel 3,6,7 on this surface. w x The apparatus has been described previously 9 . Briefly, our ultrahigh vacuum system consists of two interconnected chambers, one for sample treatment, TPD, and LEED and the other for electron spectroscopy. In the latter there are facilities for X-ray Ž . photoelectron spectroscopy XPS , Auger electron Ž . spectroscopy AES , ultraviolet photoelectron specŽ . troscopy UPS and MIES. TPD experiments were carried out using a differentially pumped UTI type Ž . quadrupole mass spectrometer QMS . TPD spectra were collected at a linear heating rate of 2 Krs with the sample in line-of-sight to the QMS. MIES spectra were measured using a cold-cathode discharge w x U source 11,12 which provides metastable He 3 1 Ž ) . 2 Sr2 S E s19.8r20.6 eV atoms with thermal kinetic energy and a SrS ratio of 7:1. The electron spectra were acquired with incident projectile beams 458 with respect to the surface normal in a constant pass energy mode using a double pass cylindrical mirror analyzer, PHI Model 15-255G. Since the metallic substrate and the analyzer are in electrical contact, the Fermi energy appears at a constant position and permits the work function change of the surface to be measured directly from the high energy w x Ž . cutoff of the electron spectra 13 . Benzene C H 6 6 ( ) J. Gunster et al.rChemical Physics Letters 306 1999 335–340 ̈ 337 Ž . Spectro Grade, Caledon Laboratories Ltd. was dosed by backfilling the vacuum chamber after further purification in the vacuum manifold via several freeze–pump–thaw cycles. In order to facilitate an easier integration of our data in the framework of the present results, we will use a nomenclature based on the one first introduced Ž . by Jakob and Menzel see introduction section for the following presentation and discussion of our TPD data. Fig. 1 presents benzene desorption spectra Ž . taken from the Ru 001 substrate as a function of the benzene exposure at 100 K. The spectra are collected w x for mres78. According to a previous study 3 , three distinct desorption features, labeled as a , a , 1 2 and a , are apparent, corresponding to the desorp3 tion of benzene at 167, 143 and 153 K, respectively. Fig. 2 presents benzene desorption spectra taken Ž . from the Mo 100 surface as a function of the benzene exposure at 100 K. Clearly visible are three Ž . Ž . desorption features: at 6 langmuirs L a feature a2 at 143 K, for increasing benzene coverage a feature Ž . b at 140 K and finally for high benzene expoŽ . sures, a third prominent feature a at 153 K. Since 3 Ž . Fig. 1. Temperature programmed desorption spectra 2 Krs of Ž . Ž . C H mres78 taken from the Ru 001 surface after different 6 6 Ž . benzene dosages at 100 K substrate temperature. The a labeled 2 peak area corresponds to the extra amount of benzene necessary to complete the transition a TM a . 1 2 Ž . Fig. 2. Temperature programmed desorption spectra 2 Krs of Ž . Ž . C H mres78 taken from the Mo 100 surface after different 6 6 benzene dosages at 100 K substrate temperature. The a labeled 2 peak area corresponds to the extra amount of benzene necessary to complete the transition a TM a . 1 2 the substrate area analyzed for both samples and the scaling in Figs. 1 and 2 are the same, a direct comparison of intensities is possible. At first glance, a comparison between Fig. 1 and Fig. 2 shows that Ž . the a feature at 167 K on the Ru 001 surface is 1 not apparent in Fig. 2. However, a broad desorption Ž . feature in the Mo 100 TPD spectra is apparent at Ž . ;230 K not visible in Fig. 2 , and appears at low benzene exposures. It is most likely that this feature has the same origin as the a peak in Fig. 1, i.e., 1 desorption of benzene molecules in the a phase, 1 but due to the higher reactivity of the molybdenum substrate, it appears at a higher desorption temperaŽ w x. ture see also Ref. 14 . This leads to a question regarding the nature of the molecule–substrate interaction, i.e., physisorption versus chemisorption, of the benzene molecules in the a phase. In this 1 context, the additional work function decrease on both substrates during the formation of the edge-on a phase shown in Fig. 3 indicates that the screening 2 of the substrate by the first chemisorbed benzene layer is not complete. The adsorption, then, of benzene after the first chemisorbed layer cannot be ( ) J. Gunster et al.rChemical Physics Letters 306 1999 335–340 ̈ 338 Fig. 3. Comparison of the relative MIES peak intensities of the 1e and 1a r3e benzene bands with the substrate work 1g 2u 2g function as a function of the benzene exposure. considered as purely physisorption. This is a new Ž . result, since the work function change of the Ru 001 w x substrate presented in a previous study 7 was monitored only to a benzene coverage at which the work function initially plateaus. It is noteworthy that the Ž . relative work function change of the Ru 001 surface with benzene exposure in Fig. 3 is smaller than in w x Ref. 7 . This difference we can attribute only to measurement methodology or surface preparation. The appearance of an additional work function change is, however, a phenomena which has been Ž . observed on both substrates, i.e. Ru 001 and Ž . Mo 100 , at the same benzene exposures. This coincidence considering the significantly different reactivities of these surfaces argues against the possibility that the additional work function decrease is due to impurities. Another clear feature of the TPD data is the appearance of the a labeled peak at 153 K on both 3 substrates for high benzene coverages. Since additional benzene adsorption leads basically to an intensity increase of the a peak, we attribute this feature 3 to the sublimation of benzene molecules adsorbed in w x a bulk-like arrangement 3 . The identification of the two additional low temperature peaks in Fig. 2, labeled a and b , appears more complicated and for 2 that reason we have performed coverage dependant MIES measurements on both substrates. As shown in w x a previous study 9 , a characteristic modulation of the spectral intensities of the 1e and 1b r3e 1g 2u 1u benzene bands, with essentially p and s character, respectively, in MIES due to a collective reorientation of the adsorbed molecules enables a clear identification of the upright a phase. In accordance with 2 this study we attribute the minimum intensity of the 1e band relative to the 1b r3e band at about 6.5 1g 2u 1u L exposure in Fig. 3 to the fully developed a phase 2 Ž w x . on both surfaces see Ref. 9 for more details . Comparing the MIES data of Fig. 3 with the TPD Ž . data, shows that the edge-on phase a is fully 2 developed in MIES. However, the low temperature Ž . TPD feature labeled as a is incomplete in Figs. 1 2 and 2. At benzene exposures over 6 L, Fig. 2 shows Ž . the formation of a new desorption feature b at 140 K. Assuming that this b species represents a new benzene adsorption state, which corresponds neither to a nor to a , it appears plausible that the inten2 3 sity increase of a in Fig. 1 over 6 L is due to the 2 formation of the b phase as well. However, even without comparing the two substrates, the completion of the a phase, as measured by MIES, long 2 Ž . before the a desorption feature on the Ru 001 2 surface saturates indicates the formation of a new benzene phase. This result is in contradiction to previous studies, since it implies that only 40% Ž . area of the a labeled desorption feature in Fig. 1 2 can be attributed to the desorption of the edge-on Ž . phase. Thus, in previous studies on Ru 100 , the a2 labeled desorption feature is formed by two, unresolved benzene adsorption states, which are resolved Ž . on Mo 100 . This leads to the new peak assignment shown in Fig. 1 in parentheses. The origin of b can be understood in the following way: after completion of the edge-on a phase, additionally adsorbed ben2 zene is unable, initially, to form the bulk-like a3 Ž . structure see Fig. 4b-2 . The reason for this is that the coverage of b-benzene is insufficient to initialize a space consuming reorganization of the a phase. 2 At a critical b-benzene coverage, reorganization of the a phase into the bulk-like a phase is energeti2 3 Ž . cally preferred see Fig. 4b . The ‘disordered’ b-be3 nzene then serves as an intermediate in the conversion from a 2D to a 3D phase. At this point it is unclear to what extent our b phase appears in Ref. ( ) J. Gunster et al.rChemical Physics Letters 306 1999 335–340 ̈ 339 Fig. 4. Schematic sketch of the various phases involved in the Ž physisorption of benzene on transition metals see also Refs. w x. Ž . 3,6 . The benzene coverage increases from the top right 1 to Ž . w x the bottom left 5 . According to a recent study 6 a disordered X Ž . a phase terminates the benzene surface in 5 . 2 w x X 6 as the amorphous a phase. However, Jakob and 2 Menzel have described the a X phase as a stable 2 buffer layer at the crystalline benzene vacuum interface, which minimizes the free surface energy. In contrast, our TPD data identify b as a phase, which is consumed in the transition a TMa . 2 3 Ž Since the areas of the new a in parentheses in 2 . Fig. 1 and a peaks are essentially identical, we 1 assume that the more dense a phase, which has 2 approximately twice the lateral density of benzene compared with the a phase, is formed by com1 pressing the benzene in the first physisorbed layer rather than by adsorption into a second physisorbed layer. Thus, the a phase is localized directly on the 2 first layer of chemisorbed benzene molecules. The additional work function decrease during the formation of a confirms its vicinity to the substrate. 2 Jakob and Menzel have suggested in their refined adsorption model that the first physisorbed layer of benzene, initially with the planes of its molecules parallel to the surface, in the a phase reorient after 1 w x additional benzene adsorption 6 . However, the details of the location of the a phase and the quantifi2 cation of the involved phases were not well-defined. Given that the benzene molecules have sufficient w x mobility in the a phase 14,15 , the formation of 1 the edge-on a phase can be described as a 2D 2 w x condensation 16 of a , due to an increase of the 1 lateral pressure with the adsorption of additional benzene. MIES measurements acquired as a function of the substrate temperature show this phase transition to be reversible. This model, which is based on a high lateral mobility of the molecules in the first physisorbed layer, explains in a straight forward manner the fluctuations found in previous studies between the a and a phases, even at cryogenic 1 2 w x temperatures 2,3 . The model presented in this Letter for benzene adsorption on transition metal surfaces is based on the appearance of the fully developed a phase of 2 Ž . upright oriented benzene molecules on Ru 001 long before the TPD a feature saturates. A comparison 2 Ž . Ž . of data obtained from the Ru 001 and Mo 100 substrates also supports this model. TPD from the two substrates shows no significant change in the adsorption dynamics of benzene but does allow the identification of a new b species. The coincidental appearance of the edge-on phase with further decrease in the work function of both substrates indicates that the screening of the substrate by the first chemisorbed benzene layer is not complete. A more detailed examination of the differences in the TPD Ž . Ž . data obtained from the Mo 100 and the Ru 001 substrates provide a more complete understanding of the nucleation mechanism and phase transitions in soft, condensed matter and will be published elsew x where 17 .