Langmuir 2008, 24, 3289-3293
3289
Adsorption-Induced Desorption of Benzene on Si(111)-7 × 7 by Substrate-Mediated Electronic Interactions Kian Soon Yong,†,‡ Shuo-Wang Yang,‡ Yong Ping Zhang,† Ping Wu,‡ and Guo Qin Xu*,† Department of Chemistry, National UniVersity of Singapore, 3 Science DriVe 3, Singapore 117543 and Institute of High Performance Computing, Singapore Science Park II, Singapore 117528 ReceiVed NoVember 5, 2007. In Final Form: January 3, 2008 The process of benzene adsorption on an adjacent adatom-rest atom pair on Si(111)-7 × 7 at room temperature was studied using in-situ scanning tunneling microscopy (STM). Both adsorption and desorption of benzene were observed to take place mostly at adjacent sites during the process. DFT calculation results show that the bond length between the rest atom and the carbon atom in a pre-adsorbed benzene molecule increases due to the charge transfer from a neighboring rest atom in response to an approaching benzene molecule. Such increase in the bond length, when coupled resonantly to the C-Si thermal vibration, could result in bond breakage and desorption of the adsorbate. The studies provide evidence for the desorption of a chemisorbed benzene caused by an adsorbing benzene at a neighboring site through a substrate-mediated electronic interaction.
I. Introduction Adsorbates on a surface can interact with one another either indirectly through the substrate or directly through electrostatic interactions or orbital overlapping. An atomistic knowledge of these interactions contributes to our understanding of technologically important processes such as catalysis,1 film growth,2 and nanostructure formation.3 Numerous studies of adsorbateadsorbate interactions have been carried out on the metal surfaces using both experimental4,5 and theoretical techniques.4,6 In contrast, similar studies on the silicon surfaces are less available in the literature. Substrate-mediated interactions or excitations via silicon are largely related to photoinduced reactions7-10 or investigated via theoretical calculations.11-13 On the basis of their calculation results, Tiersten et al. concluded the significance of the substrate-mediated phonon interactions that act between pairs of As-As dimers11 and O atoms12 on the Si(100)-2 × 1 surface. Schranz and co-workers also showed, through theoretical studies, the presence of substrate-mediated indirect interactions between two H adatoms on Si(100) and Si(111).13 We report here the study of the process of benzene adsorption on Si(111)-7 × 7 at room temperature using scanning tunneling microscopy (STM). The in-situ STM monitoring of the benzene adsorption process indicates that a significant amount of benzene * To whom correspondence should be addressed. Fax: (65) 6779 1691. E-mail:
[email protected]. † National University of Singapore. ‡ Institute of High Performance Computing. (1) Lang, N. D.; Holloway, S.; Nørskov, J. K. Surf. Sci. 1985, 150, 24. (2) Fichthorn, K. A.; Scheffler, M. Phys. ReV. Lett. 2000, 84, 5371. (3) Fichthorn, K. A.; Merrick, M. L.; Scheffler, M. Phys. ReV. B 2003, 68, 041404(R). (4) Trost, J.; Zambelli, T.; Wintterlin, J.; Ertl, G. Phys. ReV. B 1996, 54, 17850. (5) Brosseau, R.; Ellis, T. H.; Morin, M.; Wang, H. J. Electron Spectrosc. Relat. Phenom. 1990, 54/55, 659. (6) (a) Muscat, J.-P. Surf. Sci. 1984, 139, 491. Muscat, J.-P. Surf. Sci. 1981, 110, 85. (b) Muscat, J. P.; Newns, D. M. Surf. Sci. 1981, 105, 570. (c) Einstein, T. L. Surf. Sci. 1979, 84, L497. (d) Burke, N. R. Surf. Sci. 1976, 58, 349. (7) Carbone, M.; Piancastelli, M. N.; Casaletto, M. P.; Zanoni, R.; Comtet, G.; Dujardin, G.; Hellner, L. Chem. Phys. 2003, 289, 93. (8) Watanabe, K.; Matsumoto, Y. J. Chem. Phys. 2001, 115, 4259. (9) Wen, C.-R.; Chou, L.-C. J. Chem. Phys. 2000, 112, 9068. (10) Ying, Z. C.; Ho, W. J. Chem. Phys. 1990, 93, 9089. (11) Tiersten, S. C.; Reinecke, T. L.; Ying, S. C. Phys. ReV. B 1989, 39, 12575. (12) Tiersten, S. C.; Reinecke, T. L.; Ying, S. C. Phys. ReV. B 1991, 43, 12045. (13) Schranz, D. W.; Davison, S. G. J. Mol. Struct. (Theochem) 2000, 501502, 465.
desorption takes place adjacent to the adatom site where chemisorption occurs. Our density functional theory (DFT) calculation results suggest that the adsorption of a benzene molecule induces desorption of a neighboring chemisorbed benzene due to transfer of charges as mediated through the Si substrate. This work demonstrates the possible breakage of chemical bonds due to the substrate-mediated electronic interactions between adsorbates. II. Experimental and Computational Details The experiments were performed in an ultrahigh vacuum (UHV) chamber that is equipped with an Omicron variable-temperature STM and has a base pressure better than 1 × 10-10 Torr. The sample used in the experiment was cut from a P-doped mirror-polished (on one side) Si(111) wafer with a resistivity of 1-2 Ωcm and a size of 12 mm × 2 mm × 0.5 mm. It was prepared in situ using an annealing-only method, which yields a clean and well-reconstructed Si(111)-7 × 7 surface, as confirmed by STM and shown in Figure 1. Benzene (Aldrich) was introduced into the chamber via a variable leak valve after purification through several freeze-pump-thaw cycles, and its purity was verified using a mass spectrometer. The Si sample was maintained at room temperature during the benzene exposure, which is reported in langmuir (L) (where 1 L ) 1 × 10-6 Torr‚s). The STM scanning of the Si surface, which made use of an electrochemically etched W tip, was carried out during as well as after benzene exposure. As the STM is located in a separate chamber from where the benzene was being dosed, the amount of benzene being exposed to the surface while it was being scanned by the STM tip, at 1 × 10-8 Torr as recorded by the ion gauge, would be much less as compared to when the sample was directly dosed before being transported to the STM chamber for analysis. Therefore, a much longer time was required for the sample to achieve saturation coverage during the scan-while-dose experiment. The STM images were acquired in a constant-current mode with a tunneling current of 0.1 nA, and all voltages (Vs) reported in this paper were biased to the sample. The DFT calculations were performed using the DMol3 code14 in Materials Studio (version 4.0.0.0) of Accelrys. For these calculations, clusters comprised of benzene molecules bonded to the corresponding adsorption sites were cut from a Si(111)-7 × 7 supercell built using the geometries as obtained from experimental (14) Delley, B. J. Chem. Phys. 1990, 92, 508. Delley, B. J. Chem. Phys. 2000, 113, 7756. Delley, B. Comp. Mater. Sci. 2000, 17, 122.
10.1021/la7034483 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/20/2008
3290 Langmuir, Vol. 24, No. 7, 2008
Yong et al.
Figure 1. STM image (50 × 50 nm2) of a clean Si(111)-7 × 7 surface, with a defect density of less than 1%, obtained at Vs ) 1.5 V. results.15 Cluster reduction was performed in such a way so as to enhance the characteristics of the reacting center and corner adatoms by having, respectively, their neighboring two and one rest atom retained in the cluster. Apart from the adatoms and rest atoms, these four Si-layers clusters were terminated at the boundaries by H atoms. In our calculations, the double-numeric quality basis set with polarization functions (DNP) and the generalized gradient approximation (GGA) functional Perdew-Burke-Ernzerhof (PBE) developed by Perdew and co-workers16 were employed. A realspace cutoff of 4.6 Å was applied, and a FINE quality mesh size of the program was used for the computation.
III. Results and Discussions A. STM Data. Previous experimental17-21 and theoretical19,22-24 studies provide evidence for a nondissociative, di-σ adsorption configuration of benzene on Si(111)-7 × 7. This configuration involves the binding of two opposite carbons on benzene to an adjacent adatom-rest atom pair on the substrate to form a 1,4cyclohexadiene-like structure. Such adatom-rest atom cooperative binding of benzene can be detected by STM study. The STM images obtained after exposing Si(111)-7 × 7 to 2 L of benzene are shown in Figure 2. Comparing Figure 2 to the STM image of a clean Si(111)-7 × 7 surface as shown in Figure 1, it is apparent that benzene adsorption creates the observed adatom vacancies on the surface, with each vacancy indicating single chemisorbed benzene. The disappearance of the adatoms in the STM images was attributed to elimination of the adatom dangling bonds induced by the surface-adsorbates bond formation as well as the lack of low-lying energy states of the adsorbed benzene that are close to the Fermi level. From the STM results, each half (15) Tong, S. Y.; Huang, H.; Wei, C. M.; Packard, W. E.; Men, F. K.; Glander, G.; Webb, M. B. J. Vac. Sci. Technol. A 1988, 6, 615. (16) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865. (17) Cao, Y.; Wei, X. M.; Chin, W. S.; Lai, Y. H.; Deng, J. F.; Bernasek, S. L.; Xu, G. Q. J. Phys. Chem. B 1999, 103, 5698. (18) Carbone, M.; Piancastelli, M. N.; Zanoni, R.; Comtet, G.; Dujardin, G.; Hellner, L. Surf. Sci. 1998, 407, 275. (19) Tomimoto, H.; Sekitani, T.; Sumii, R.; Sako, E. O.; Wada, S.-i.; Tanaka, K. Surf. Sci. 2004, 566-568, 664. (20) Tomimoto, H.; Takehara, T.; Fukawa, K.; Sumii, R.; Sekitani, T.; Tanaka, K. Surf. Sci. 2003, 526, 341. (21) Kawasaki, T.; Sakai, D.; Kishimoto, H.; Akbar, A. A.; Ogawa, T.; Oshima, C. Surf. Interface Anal. 2003, 31, 126. (22) Lu, X.; Wang, X.; Yuan, Q.; Zhang, Q. J. Am. Chem. Soc. 2003, 125, 7923. (23) Wang, Z. H.; Cao, Y.; Xu, G. Q. Chem. Phys. Lett. 2001, 338, 7. (24) Petsalakis, I. D.; Polanyi, J. C.; Theodorakopoulos, G. Surf. Sci. 2003, 544, 162.
Figure 2. STM images (Vs ) 1.5 V) of the same area of benzeneexposed Si(111)-7 × 7 with b scanned about 4 min after a. The disappearing and reappearing adatoms in b are marked by the circles and crosses, respectively. F and U represent the faulted and unfaulted half unit cells, respectively.
unit cell contains at most three disappeared adatoms, which do not align in a straight line. Such observations are in accordance to previous STM studies19-21 and provide evidence for the binding of benzene to an adjacent pair of adatom and rest atom on Si(111)-7 × 7. Furthermore, the preferences of benzene to adsorb at the center over the corner adatoms and faulted over the unfaulted halves as observed in our STM images are also consistent with previous STM results.19-21,25 Figure 2 also compares two consecutive STM images scanned with a time difference of about 4 min. Individual triangular half unit cells are marked out by the white lines superimposed onto these images, with ‘U’ and ‘F’ representing the unfaulted and faulted halves, respectively. Through close comparison between Figure 2a and b, adatoms that disappeared (marked by the circles) and reappeared (marked by the crosses) can be identified in the latter image. The reappeared adatoms indicate the departure of chemisorbed benzene, which allows the vacated adatom to restore its original electronic states and be detected by STM. The departed (25) Wolkow, R. A.; Moffatt, D. J. J. Chem. Phys. 1995, 103, 10696.
Desorption of Benzene on Si(111)-7 × 7
Langmuir, Vol. 24, No. 7, 2008 3291
Figure 4. Plots of the reappeared faulted adatoms, with and without adjacent adatoms that disappeared, as a function of time during exposure to benzene. The trend line shows the coverage of benzene on the surface in monolayer.
Figure 3. STM images (Vs ) 1.0 V) of three separate regions (i, ii, and iii) of Si(111)-7 × 7 with b scanned about 3 min after a during exposure to benzene. The reappeared adatoms (marked by crosses) are accompanied by adjacent disappeared adatoms (marked by circles) in b. F and U represent the faulted and unfaulted half unit cells, respectively.
benzene either desorbs from the surface or diffuses to other vacant sites for readsorption, which were captured by STM scanning as a newly disappeared adatom. In accordance with previous STM results,21,25 higher activities of readsorption and desorption of benzene were found at the faulted half unit cells, especially at the center adatoms. Additionally, it was observed that about 94% of the disappearance and reappearance events occurred at adatom sites that are nonadjacent to each other, suggesting that the desorbed benzene diffuses on the surface before it chemisorbs on an adatom-rest atom pair again. Carbone and co-workers suggested successive occurrence of the breakage of C-Si bond at the adatom site and rotation of the other C-Si bond about the rest atom site as the mechanism adapted by benzene for diffusing on the Si(111)-7 × 7 surface before settling on a site that is nonadjacent to the initial chemisorbed site.18 It should be noted that a low STM bias of 1.0-1.5 V was used for the imaging so as to minimize interference of the adsorbates due to the scanning tip.26 As such, the desorption and diffusion of benzene as observed via STM imaging is independent of the tip scanning, as pointed out by Wolkow.25 In-situ STM monitoring of the benzene adsorption process was also performed during the benzene exposure. Figure 3 shows three sets of STM images, with Figure 3b obtained at a time (26) Mayne, A. J.; Dujardin, G.; Comtet, G.; Riedel, D. Chem. ReV. 2006, 106, 4355.
lapse of about 3 min after Figure 3a, of different regions scanned during the exposure of benzene. The benzene coverage during this period of time is about 0.05 ML (where 1 monolayer (ML) is defined as the coverage for 1 benzene on a 1 × 1-unreconstructed unit cell of the Si(111) surface). By comparing Figure 3b to the previous images in Figure 3a, the adatoms that disappeared (marked by the circles) and those that reappeared (marked by the crosses) can be identified. Apparently, the reappearance events took place at adatoms that are adjacent to sites with disappearance events, in contrast to the case when STM imaging was carried out when the benzene dosing had been terminated (Figure 2). Statistically, such neighboring events account for about 80% of the reappeared adatoms that were detected. Furthermore, it can be observed from Figure 3b that high activities of the benzene adsorption and desorption processes occurred at the faulted halves of the 7 × 7 cells. The results of quantitative analysis are presented in Figure 4, which displays the number of reappeared faulted adatoms as a function of time during the exposure to benzene. Each of the columns is divided into the number of adatoms that have and do not have adjacent disappeared adatoms. The small number of reappeared adatoms that do not have neighboring disappeared adatoms indicates that the majority of the benzene desorptions are accompanied by benzene adsorption at an adjacent site. A trend line is also inserted in Figure 4 that shows the approach of the benzene coverage toward saturation of 0.12 ML (6/49). The corresponding plot for the unfaulted adatoms is not shown due to the significantly lower number of reappeared adatoms detected at the unfaulted halves. B. DFT Calculations. For the case when the STM experiment was conducted after the benzene exposure had been terminated, the benzene desorption and readsorption sites are generally nonadjacent to each other. This is attributed to the desorption, diffusion, and readsorption processes of benzene on Si(111)-7 × 7.21,25 However, by performing the STM experiment at the same time when the surface was being exposed to benzene, most desorption events of the chemisorbed benzene were found to be adjacent to the newly adsorbed molecules. One of the possible reasons to explain the observed adsorption and desorption of benzene at adjacent adatom sites is that a chemisorbed benzene desorbs and then readsorbs on a neighboring adatom site. However, there is no reason that such events could not take place in the experiment whereby STM imaging was performed after the benzene exposure had been terminated. It is noted that the conditions for the two experiments, including the tip bias, are similar except for the presence of gaseous benzene above the surface in the in-situ STM measurement. Therefore, the difference between the two results is likely due to the incoming benzene
3292 Langmuir, Vol. 24, No. 7, 2008
Figure 5. (a) Optimized structure of a benzene adsorbate (B1) on a center adatom-rest atom (Ce1-Re1) pair on Si(111)-7 × 7. (b) Final structure of a second benzene molecule (B2) adsorbed on another center adatom-rest atom (Ce2-Re2) pair obtained through optimization in the presence of B1. Each cluster has the following notation: ball-and-stick, benzene; yellow lines, unfaulted half; green lines, faulted half; red spheres, adatoms; orange spheres, rest atoms. The schematic diagram on the left of each cluster shows the attachment position of the benzene molecule within a 7 × 7 unit cell, where the shaded and unshaded triangles represent the faulted and unfaulted halves, respectively.
in the latter experiment, which adsorbs on an adatom-rest atom pair and induces the desorption of a chemisorbed benzene on a neighboring adatom site. In order to evaluate the feasibility of such a mechanism, a cluster (Figure 5) consisting of two adjacent faulted adatom-
Yong et al.
rest atom sites was constructed and a benzene molecule (B1) was first positioned close to one of the center adatom (Ce1) and rest atom (Re1) pairs. The cluster with the single benzene was allowed to relax to the optimized structure as shown in Figure 5a, with both C-Ce1 and C-Re1 having bond distances of 2.02 Å. A second benzene molecule (B2) was subsequently placed close to the neighboring vacant center adatom (Ce2) and rest atom (Re2) site and allowed to relax in steps consisting of three iterations. The C-Si bond lengths of B1 and B2 were measured after each step, and the cluster was allowed to relax again from the previous optimized structure. This process was repeated until the energy change was converged to 10-5 Ha and the structure shown in Figure 5b was obtained. The relaxed C-Si bond lengths for both the adsorbates are about 2.02 Å. During the optimizations, the two benzene molecules, the interacting adatoms and rest atoms and their immediate neighboring Si atoms, as well as some of the substrate atoms connecting the two reacting sites were allowed to relax. The rest of the atoms were frozen to simulate a bulk-like environment. Figure 6a and b shows the variations of the C-Si bond distances of B1 and B2 as a function of the optimization step. It is interesting to note that benzene forms a bond with the adatom first before the rest atom during the adsorption process of B2 on the surface, as depicted in Figure 6a. This is in accordance with the result from Lu’s study of the [4 + 2] cycloaddition of benzene on Si(111)-7 × 7, which detects a transition state with benzene attached to the adatom.22 Additionally, when B2 was distant from the substrate at the initial steps, very little changes can be detected in the C-Si separations in B1. However, when B2 was within the bond formation distance from the substrate at steps 4 and 5, a sudden increase of 0.025 Å in the C-Re1 bond distance in B1 can be observed (Figure 6b). Such increment in the C-Re1
Figure 6. (a) Variations of the C-Ce1 (C-Ce2) and C-Re1 (C-Re2) bond distances in B1 (B2). The enlarged scale of the region indicated by the oval in a is shown in b, which shows a sudden change in the C-Re1 bond length in B1 as B2 approaches the surface. (c) The Mulliken charge differences at Ce1, Re1, Ce2, and Re2 as obtained by subtracting the charges on the Si atoms from those on the corresponding atoms in the next optimized structure. (d) The total adsorption energy of B1 and B2 as a function of the optimization steps of B2 adsorption on the cluster. The values were obtained by subtracting the total energy of two gas-phase benzenes and the free substrate cluster from the energy of the cluster comprised of the adsorbate molecules and the substrate.
Desorption of Benzene on Si(111)-7 × 7
separation could be the result of charge transfer from Re2 due to the approach of B2. Figure 6c shows the Mulliken charge differences at Ce1, Ce2, Re1, and Re2 during the optimization steps. Evidently, a charge transfer away from Re2 takes place in response to the gradual bond formation with a C atom on B2. This is agreeable with the result from the photoemission study of benzene adsorption on Si(111)-7 × 7 by Taguchi, which suggests that charge transfer from the rest atom plays an important role in the chemisorption process.27 Simultaneously, charge transfers of 0.031 and 0.012 e to Ce2 and Re1, respectively, are detected from Figure 6c at step 4. As a result of these transfers of charges from Re2, the C-Ce2 and C-Re1 bond distances increase as shown in Figure 6b. Hence, the above results provide evidence for an electronic interaction between an adsorbing and a chemisorbed benzene as mediated through the Si(111)-7 × 7 substrate. In order to access the energy change of the cluster during the adsorption process of B2, the total adsorption energy of the two benzene molecules on the substrate was plotted as shown in Figure 6d. The energy of the system decreases gradually as B2 approaches and eventually forms covalent bonds with the substrate. The final energy of the system is 2.07 eV, which agrees well with the ∼1 eV adsorption energy of a benzene molecule on Si(111)-7 × 7.19,22,23 However, at step 4 when the charge transfer to Re1 causes the lengthening of the C-Re1 bond distance, an increase of 0.541 eV is observed in the energy plot. The energy barrier due to the transition state of benzene in the [4 + 2] cycloaddition-like pathway was calculated to be 0.334 eV.22 Hence, the configuration of B1 is likely to have contributed to the large energy instability observed at step 4 of Figure 6d as the interaction of B2 with the surface cannot solely account for the large energy increment. The instability of B1 that contributed to the energy rise may trigger its desorption from the (27) Taguchi, Y.; Ohta, Y.; Katsumi, T.; Ichikawa, K.; Aita, O. J. Electron Spectrosc. Relat. Phenom. 1998, 88-91, 671.
Langmuir, Vol. 24, No. 7, 2008 3293
surface through resonant coupling with the thermal vibrations of the adsorbate. However, it should be noted that not all the benzene adsorption would induce desorption of chemisorbed benzene from adjacent sites. Otherwise, the overall increase in the quantity of benzene adsorbate during the dosing process till saturation coverage, with three benzenes occupying one-half unit cell, would not be observed. On the other hand, some of the benzene desorption is not caused by adsorption of benzene at an adjacent adatom-rest atom site, as evidenced by the presence of reappeared adatoms with no adjacent disappeared adatoms in Figure 4. Additionally, the latter part of the plot in Figure 4 corresponds to high coverage of benzene that occupies three adatoms per half unit cell. Under such conditions, the observation of neighboring events of reappearance and disappearance of adatoms is likely due to desorption of a chemisorbed benzene prior to adsorption of another benzene at the newly vacated adatom site.
IV. Conclusions STM was used in situ to monitor the room temperature adsorption process of benzene on Si(111)-7 × 7. A significant amount of benzene adsorption and desorption, as indicated by the disappearance and reappearance of Si adatoms, respectively, was observed to occur at adjacent sites. The DFT calculation clearly indicates that the approach of a benzene molecule toward an adatom-rest atom pair causes charge transfer away from the rest atom to a neighboring benzene-bonded rest atom site. This in turn results in an increase in the bond length between the rest atom and the carbon atom of the chemisorbed benzene. When coupled resonantly to the C-Si thermal vibration, this increase in the bond length may cause bond breakage and desorption of the adsorbate. Thus, the current studies provide evidence for a substrate-mediated electronic interaction between an adsorbing and a chemisorbed benzene as well as the adsorption-induced desorption of the chemisorbed benzene on Si(111)-7 × 7. LA7034483
