Methylthiolate induced vacancy formation on Au(1 1 1): a density functional theoretical study

Methylthiolate induced vacancy formation on Au(1 1 1): a density functional theoretical study

Surface Science 514 (2002) 389–393 www.elsevier.com/locate/susc Methylthiolate induced vacancy formation on Au(1 1 1): a density functional theoretic...

118KB Sizes 0 Downloads 5 Views

Surface Science 514 (2002) 389–393 www.elsevier.com/locate/susc

Methylthiolate induced vacancy formation on Au(1 1 1): a density functional theoretical study Y. Morikawa

a,b,*

, C.C. Liew a, H. Nozoye

c

a

c

Research Institute for Computational Sciences (RICS), and Research Consortium for Synthetic Nano-Function Materials Project (SNAF), National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan b School of Materials Science, Japan Advanced Institute of Science and Technology (JAIST), 1-1 Asahidai, Tatsunokuchi, Ishikawa 923-1292, Japan Nanotechnology Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8565, Japan Received 5 October 2001; accepted for publication 8 February 2002

Abstract pffiffiffi pffiffiffi We have studied various possible molecular configurations of methylthiolate (MeS) adsorbed Au(1 1 1)cð4 3  2 3Þ surfaces using density functional theory within a generalized gradient approximation (GGA). Assuming unreconstructed substrate, four-chain models which pffiffihave ffi pfour ffiffiffi crystallographically distinct molecules in a unit cell are less stable than a one-chain model, that is, a simple 3  3 structure, being inconsistent with experimental observations. From our GGA calculations, we point out a possibility of vacancy formation in the first Au layer of MeS adsorbed Au(1 1 1) surfaces. Ó 2002 Elsevier Science B.V. All rights reserved. Keywords: Density functional calculations; Chemisorption; Self-assembly; Gold; Surface relaxation and reconstruction

1. Introduction Alkanethiol self-assembled monolayers on Au surfaces have been attracting enormous attention because of their wide range of potential applications such as catalysis, molecular recognition, nanoscale electronic devices, and so on [1,2]. Considering that the ultimate goal of this field is to

*

Corresponding author. Tel.: +81-298-61-2439; fax: +81298-61-3171. E-mail address: [email protected] (Y. Morikawa).

design and create various functional monolayers by controlling molecular arrangements, it is of great importance to clarify the structure and the assembling processes of self-assembled monolayers. Despite of intensive research, the adsorption states and molecular conformations of the monolayers are still a matter of intense debate [3–5]. Recently, we have examined methylthiolate (MeS) and dimethyl disulfide (DMDS) adsorption on the Au(1 1 1) surface using density functional theory (DFT) calculations [6,7]. We concluded that the MeS adsorption is more stable than the DMDS adsorption and that the most stable adsorption sitefor MeS is the bridge site slightly

0039-6028/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 6 0 2 8 ( 0 2 ) 0 1 6 5 7 - 6

390

Y. Morikawa et al. / Surface Science 514 (2002) 389–393

off-centered towards the fcc-hollow site. Our results were further confirmed by the excellent agreement between the theoretically calculated high-resolution electron energy loss spectra (HREELS) with the experimental ones. We confirmed that ethylthiolate (EtS) and butylthiolate (BuS) also prefer the bridge site and the Au–S bonding geometries of the two adsorbates are quite similar to that of MeS. Although previous quantum chemical [8,9] and recent DFT calculations [10,11] had concluded that the hollow site was the most stable adsorption site, very recent DFT studies with wider geometry search supported our results [12–14]. Recent DFT calculations of MeS on Au clusters also reported that S does not prefer threefold Au coordination but prefers lower coordination [15,16]. Having established the local Au–S bonding geometry, we consider possible molecular arrangement in a pffiffiffi pffiffiffi c(4 3  2 3) superstructure which was observed experimentally on CH3 (CH2 )n S (n P 3) adsorbed Au(1 1 1) surfaces by X-ray diffraction, He diffraction, and scanning tunneling microscopy (STM) techniques [17–21]. It is observed stably even at room temperature and comprises four crystallographically distinct molecules. It was suggested that the distinct molecular features come from different twist angles of the C–C–C plane of the hydrocarbon backbones To clarify the pffiffiffi pffiffi[18]. ffi structure of the cð4 3  2 3Þ structure, several classical molecular dynamics calculations have been carried out [22–25]. None of those simulations, however, could predict a four-chain unit cell structure to be the most stable one, presumably because of inaccurate Au–S bonding potential. Vargas et al. [13] and Morikawa et al. [7] examined several molecular arrangements in a pffiffiffi possible pffiffiffi c(4 3  2 3) unit cell but they claimed that the simplest one-chain model is slightly more stable than four-chain models, being inconsistent with the experimental observation. One possible reason for the discrepancy may be due to insufficient accuracy of the present GGA, especially, inability to describe the van der Waals interaction [26,27]. Another possible reason may be a geometrical factor which was not considered previously. In this study, we have intensively studied possible pffiffiffi strucpffiffiffi tures for MeS adsorbed Au(1 1 1)c(4 3  2 3)

surfaces including substrate reconstructions and pointed out a possibility of Au vacancy formation.

2. Method All calculations were carried out using a program package STATE (Simulation Tool for Atom TEchnology), which has been successfully applied for semiconductor as well as transition and noble metal surfaces [6,7,28,29]. We adopted a generalized gradient approximation (GGA) in the density functional theory (DFT) [30,31] and the Perdew, Burke and Ernzerhof formula [32] as the exchange-correlation energy functional. We constructed pseudopotentials of H 1s, C 2p, and Au 5d states by Vanderbilt’s scheme [33], while other components of pseudopotentials by norm-conserving scheme [34,35]. Two projectors are used for ultrasoft pseudopotentials. The cut-off energy for the wave function is 25 Ry and that for the augmentation charge is 225 Ry. A repeated slab model, in which slabs consisting of four atomic  layers are separated by a vacuum region of 14.4 A is used to simulate the Au(1 1 1) surface. Adsorbates are introduced on one side of each slab. For the structural optimization, adsorbates and the three top layerspof atoms are allowed to ffiffiffi substrate pffiffiffi relax. For cð4 3  2 3) unit cell, surface Brillouin zone is sampled by 2  3 or 3  4 uniform mesh of k-points and the Fermi level is broadened by Methfessel and Paxton scheme [36].

3. Results and discussion In our previous study [7], we have examined several molecular for MeS adsorbed pffiffiffi arrangements pffiffiffi Au(1 1 1)cð4 3  2 3) surfaces. Fig. 1 shows the three most stable structural models and the adsorption energies of those models are summarized in Table 1. The adsorption energy is defined by pffiffiffi pffiffiffi Ead ¼ ðE½CH3 S=Auð1 1 1Þcð4 3  2 3Þ pffiffiffi pffiffiffi  E½Auð1 1 1Þcð4 3  2 3Þ  2  E½CH3 SSCH3 Þ=4;

ð1Þ

Y. Morikawa et al. / Surface Science 514 (2002) 389–393

391

pffiffiffi pffiffiffi Fig. 1. Top views of MeS adsorbed Au(1 1 1)cð4 p3ffiffiffi 2 p3ffiffiÞffi surfaces: (a) one-chain and (b,c) four-chain models. The calculated most stable structure is one-chain model of (a). The c(4 3  2 3) unit cell is indicated by solid lines. Numbers 1–4 indicate positions of Au vacancies. In the case of one vacancy in a c(4  2) unit cell, atoms numbered 1 were removed. In the case of two vacancies, atoms numbered 1 and 2 were removed, and so on.

Table 1 inffiffiffi kcal/mol) of MeS in various The adsorption energy (Ead p pffiffiffi possible configurations of c(4 3  2 3) superstructure Configuration

4L 6k

4L 12k

6L 6k

(a) 1–1 (b) 4–2 (c) 4–3

6.75 6.46 6.39

6.46 6.18 5.85

5.31 4.75 4.90

The effect of layer thickness (4L: 4 layers and 6L: 6 layers) and number of k-points (6k: 6k-points and 12k: 12k-points) are checked.

pffiffiffi pffiffiffi where E½CH 1 1Þcð4 3  2 3Þ, E½Au pffiffiffi 3 S=Auð1 pffiffiffi ð1 1 1Þcð4 3  2 3Þ, and E½CH3 SSCHp 3 ffiffiffiare total pffiffiffi energies of MeS adsorbed Au(1 1 1)c(4 3  2 3), pffiffiffi pffiffiffi clean Au(1 1 1)cð4 3  2 3Þ and gas phase DMDS respectively. Although we have checked the convergence with respect to number of kpoints and slab thickness, the most stable adsorption configuration is the one-chain model shown in Fig. 1(a), and four-chain models shown in Fig. 1(b) and (c) are slightly less stable than the one-chain model, being inconsistent with the experimental results. Vargas et al. [13] also reported that the four-chain model of Fig. 1(c) is 0.5 kcal/ mol less stable than the one-chain model of Fig.

1(a) and they ascribed the discrepancy to the van der Waals interaction which is not included in the present GGA functional. Here, we have examined another possibility. All previous theoretical studies of alkanethiolates on Au(1 1 1) assumed 1  1 unreconstructed substrate but we suspected this assumption. We investigated Au vacancy formation on MeS adsorbed Au(1 1 1) surfaces and the vacancy formation energy is summarized in Table 2. The vacancy formation energy EV is defined by pffiffiffi pffiffiffi EV ¼ ðE½CH3 S=Auð1 1 1Þcð4 3  2 3Þ  nV þ n  E½Aufcc  pffiffiffi pffiffiffi  E½CH3 S=Auð1 1 1Þcð4 3  2 3ÞÞ=n; ð2Þ pffiffiffi pffiffiffi where E½CH3 S=Auð1 1 1Þcð4 3  2 3Þ  nV and E½Aufcc  are p the ffiffiffi total pffiffiffienergies of a MeS adsorbed Au(1 1 1)cð4 3  2 3Þ surface with n Au vacancies and bulk pffiffiffi fccpffiffiffiAu, respectively. E½CH3 S= Auð1 1 1Þcð4 3  2 3p Þffiffiffiis theptotal energy of MeS ffiffiffi adsorbed Au(1 1 1)c(4 3  2 3) in one chain configuration (Fig. 1(a)). Positions of Au vacancies are shown in Fig. 1 with pffiffinumbers ffi pffiffiffi 1–4. In the case of one vacancy in a cð4 3  2 3Þ unit cell, atoms

392

Y. Morikawa et al. / Surface Science 514 (2002) 389–393

Table 2 The Au vacancy formation energy (EV in kcal/mol) of MeS pffiffiffi p ffiffiffi adsorbed Au(1 1 1)cð4 3  2 3Þ surfaces calculated by a generalized gradient approximation (GGA) and a local density approximation (LDA) Configuration

# of vacancies

GGA

LDA

4L 12k

7L 12k

4L 12k

7L 12k

(a) 1–1

1 2 3 4

0.88 2.18 1.91 1.91

1.63 2.59 2.39 2.51

þ4.37 þ3.08 þ3.33 þ2.89

þ3.21 þ2.49

(b) 4–2

1 2 3 4

0.43 þ0.41 1.66 1.57

(c) 4–3

1 2 3 4

þ5.29 þ0.36 þ0.58 þ0.06

EV is defined in the text. Molecular configurations of (a)–(c) are shown in Fig. 1 and positions of Au vacancies are indicated by numbers 1–4. For example, for the case of one vacancy in a c(4  2) unit cell, atoms numbered 1 were removed. For the case of two vacancies, atoms numbered 1 and 2 were removed, and so on. The effect of number of layers (4L: 4 layers and 7L: 7 layers) is checked. Negative value of the vacancy formation energy means that formation of Au vacancy is exothermic.

numbered 1 were removed. In the case of two vacancies, atoms numbered 1 and 2 were removed, and so on. Negative value of the vacancy formation energy means that formation of vacancies is exothermic. GGA calculations predict that vacancy formation on the clean Au(1 1 1) surface is 10.1 kcal/mol endothermic while those on the MeS adsorbed Au(1 1 1) surfaces can be exothermic depending on the molecular configurations and positions of vacancies. The most negative vacancy formation energy is achieved in the case of one chain model with two vacancies (Fig. 1(a)). The stabilization of surface vacancy can be traced back to strengthening of S binding to Au atoms surrounding a vacancy. As shown in Fig. 1(a), all first-layer Au atoms which are at nearest neighbour sites of Au vacancies are bound to S atoms. The S-Au bond length near by vacancies is 0.4–0.8  shorter and height of S atoms near by a vacancy A  lower than those of MeS on unreis 0.3 A constructed Au(1 1 1) surfaces, indicating stronger

Au–S bonding near by a Au vacancy. The strengthening of Au–S bonding near the vacancies make positions of MeS molecules at vacancies closer to the surface, giving slight corrugation in molecular height. Therefore, one-chain molecular configuration with two vacancies is qualitatively consistent with experimentally observed STM images, in which two molecules in a cð4  2Þ unit cell are imaged brighter than the others [19,20]. Furthermore, although the density of first-layer Au atoms is reduced by vacancy formation, positions of Au atoms are quite close to those of an ideal Au surface, being again qualitatively consistent with X-ray standing wave (XSW) results [37]. For example, the maximum displacement and the mean displacement of first-layer Au atoms from  and 0.17A , their ideal surface positions are 0.21A respectively for the unreconstructed MeS/Au(1 1 1)  and 0.15–0.28 surface while they are 0.24–0.30 A , respectively for reconstructed MeS/Au(1 1 1) A surfaces. Local Au–S bonding geometry, i.e., S–C bond length and the tilting angle of S–C bond from the surface normal are also close to those of unreconstructed MeS=Au(1 1 1) surfaces. These facts make experimental detection of Au vacancies difficult. However, we think that surface X-ray diffraction may be possible to clarify the existence of Au vacancies proposed here. Experimentally, dramatic structural change was observed during the formation of self-assembled monolayers [38], which might be a result of vacancy formation. Actually, substrate reconstruction was reported in the case of octanethiol adsorbed Cu(1 1 1)[39]. One serious drawback of calculations using the present GGA is that GGA tends to underestimate the vacancy formation energy [40] and LDA results contradict to those of GGA as indicated in Table 2. Further study including more accurate functionals for inhomogeneous systems [41] as well as the van der Waals interaction are necessary but at present, it is not a easy task.

4. Summary We have extensively examined possible molecular configurations of MeS adsorption on the

Y. Morikawa et al. / Surface Science 514 (2002) 389–393

pffiffiffi pffiffiffi Au(1 1 1)cð4 3  2 3Þ surface using first-principles theoretical calculations. If we assume unreconstructed Au(1 1 1) surfaces as substrates, the most stable is the one-chain model, i.e., pffiffiffi structure pffiffiffi simple 3  3 structure, being inconsistent with the experimental observations. We have pointed out a possibility of Au vacancy formation in the first-layer of Au substrate based on our GGA calculations. The most negative (exothermic) vacancy formation energy is achieved in the case of one-chain model with two vacancies. This structure can be qualitatively consistent with experimental results. At present, LDA calculations give contradictory results to GGA and further investigations using more accurate functionals are necessary.

Acknowledgements We would like to thank Mr. T. Hayashi for fruitful discussions and collaboration in the early stage of the present study. The numerical calculations were performed at the computer centers of JRCAT, Tsukuba Advanced Computing Center (TACC), and the Institute of Physical and Chemical Research (RIKEN). The present work is partly supported by New Energy and Industrial Technology Development Organization (NEDO) and also by a Grant-in-Aid for Scientific Research from Ministry of Education, Science and Culture of Japan.

References [1] J.D. Swalen, D.L. Allara, J.D. Andrade, E.A. Chandross, S. Garoff, J. Israelachvili, T.J. McCarthy, R. Murray, R.F. Pease, J.F. Rabolt, K.J. Wynne, H. Yu, Langmuir 3 (1987) 932. [2] J. Chen, M.A. Reed, A.M. Rawlett, J.M. Tour, Science 286 (1999) 1550. [3] L.H. Dubois, R.G. Nuzzo, Ann. Rev. Phys. Chem. 43 (1992) 437. [4] A. Ulman, Chem. Rev. 96 (1996) 1533. [5] G.E. Poirier, Chem. Rev. 97 (1997) 1117. [6] T. Hayashi, Y. Morikawa, H. Nozoye, J. Chem. Phys. 114 (2001) 7615. [7] Y. Morikawa, T. Hayashi, C.C. Liew, H. Nozoye, Surf. Sci. 507–510 (2002) 46.

393

[8] H. Sellers, A. Ulman, Y. Shnidman, J.E. Eilers, J. Am. Chem. Soc. 115 (1993) 9389. [9] K.M. Beardmore, J.D. Kress, N. Grønbech-Jensen, A.R. Bishop, Chem. Phys. Lett. 286 (1998) 40. [10] H. Gr€ onbeck, A. Curioni, W. Andreoni, J. Am. Chem. Soc. 122 (2000) 3839. [11] Y. Yourdshahyan, H.K. Zhang, A.M. Rappe, Phys. Rev. B 63 (2001) 081405R. [12] Y. Akinaga, T. Nakajima, K. Hirao, J. Chem. Phys. 114 (2001) 8555. [13] M.C. Vargas, P. Giannozzi, A. Selloni, G. Scoles, J. Phys. Chem. B 105 (2001) 9509. [14] J. Gottschalck, B. Hammer, J. Chem. Phys 116 (2002) 784. [15] H. H€akkinen, R.N. Barnett, U. Landman, Phys. Rev. Lett. 82 (1999) 3264. [16] D. Kr€ uger, H. Fuchs, R. Rousseau, D. Marx, M. Parrinello, J. Chem. Phys. 115 (2001) 4776. [17] P. Fenter, P. Eisenberger, K.S. Liang, Phys. Rev. Lett. 70 (1993) 2447. [18] N. Camillone III, C.E.D. Chidsey, G. Liu, G. Scoles, J. Chem. Phys. 98 (1993) 3503. [19] G.E. Poirier, M.J. Tarlov, Langmuir 10 (1994) 2853. [20] E. Delamarche, B. Michel, Ch. Gerber, D. Anselmetti, H.J. G€ untherodt, H. Wolf, H. Ringsdorf, Langmuir 10 (1994) 2869. [21] T. Hayashi, C. Kodama, H. Nozoye, Appl. Surf. Sci. 169– 170 (2001) 100. [22] W. Mar, M.L. Klein, Langmuir 10 (1994) 188. [23] A.J. Pertsin, M. Grunze, Langmuir 10 (1994) 3668. [24] R. Bhatia, B.J. Garrison, Langmuir 13 (1997) 765. [25] T.W. Li, I. Chao, Y.T. Tao, J. Phys. Chem. B 102 (1998) 2935. [26] S. Serra, S. Iarlori, E. Tosatti, S. Scandolo, G. Santoro, Chem. Phys. Lett. 331 (2000) 339. [27] S. Tsuzuki, H.P. L€ uthi, J. Chem. Phys. 114 (2001) 3949. [28] Y. Morikawa, K. Iwata, K. Terakura, Appl. Surf. Sci. 169– 170 (2001) 11. [29] Y. Morikawa, Phys. Rev. B 63 (2001) 033405. [30] P. Hohenberg, W. Kohn, Phys. Rev. 136 (1964) B864. [31] W. Kohn, L.J. Sham, Phys. Rev. 140 (1965) A1133. [32] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. [33] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892. [34] G.B. Bachelet, D.R. Hamann, M. Schl€ uter, Phys. Rev. B 26 (1982) 4199. [35] N. Troullier, J.L. Martins, Phys. Rev. B 43 (1991) 1993. [36] M. Methfessel, A.T. Paxton, Phys. Rev. B 40 (1989) 3616. [37] P. Fenter, F. Schreiber, L. Berman, G. Scoles, P. Eisenberger, M.J. Bedzyk, Surf. Sci. 412–s413 (1998) 213. [38] G.E. Poirier, Langmuir 13 (1997) 2019. [39] S.M. Driver, D.P. Woodruff, Langmuir 16 (2000) 6693. [40] K. Carling, G. Wahnstr€ om, T.R. Mattsson, A.E. Mattsson, N. Sandberg, G. Grimvall, Phys. Rev. Lett. 85 (2000) 3862. [41] J.P. Perdew, S. Kurth, A. Zupan, P. Blaha, Phys. Rev. Lett. 82 (1999) 2544.