HAL Id: jpa-00229691
https://hal.archives-ouvertes.fr/jpa-00229691
Submitted on 1 Jan 1989
HAL
is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire
HAL, estdestinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
INVESTIGATION OF THE Au/Si (111) SURFACE STRUCTURES BY X-RAY DIFFRACTION
R. Feidenhans’L, F. Grey, J. Bohr, M. Nielsen, R. Johnson
To cite this version:
R. Feidenhans’L, F. Grey, J. Bohr, M. Nielsen, R. Johnson. INVESTIGATION OF THE Au/Si (111)
SURFACE STRUCTURES BY X-RAY DIFFRACTION. Journal de Physique Colloques, 1989, 50
(C7), pp.C7-175-C7-179. �10.1051/jphyscol:1989717�. �jpa-00229691�
COLLOQUE DE PHYSIQUE
Colloque C7, suppl6ment a u n010, Tome 50, octobre 1989
INVESTIGATION OF THE Au/Si (111) SURFACE STRUCTURES BY X-RAY DIFFRACTION
R. FEIDENHANS'L, F. GREY', J. BOHR, M. NIELSEN and R.L. JOHNSON*
$is# National Laboratory, DK-4000 Roskilde, Denmark
11 Institute for Experimental Physics, University of Hamburg, D-2000 Hamburg 52, F.R.G.
R e s u m e
Nous avons 6tudi6 des structures "5x1" et f i x f i R 30' de Si(ll1) induites par adsorption d'or dans Ie domaine de la sous-monocouche. Nous avons respectivement mesud 52 et 13 intensites de taches de diffraction non Cquivalentes. Les fonctions de Patterson montrent qu'aucune des structures proposQs jusqu'B present n'est valable. L'analyse de nos rksultats montre que la structure f i x f l correspond B un modkle trimkre, tandis que celle en "5x1" est en fait une structure 5x2 desordom6e qui contient 2 atomes &or par sous-rkseau 5x1.
A b s t r a c t
We have investigated the "5x1" and d 3 ~ d 3 ~ 3 0 0 structures of S i ( l l 1 ) induced by adsorption of Au i n the submonolayer regime. 52 and 1 3 symmetry- inequivalent structure factor intensities for reflections with fractional indices were measured for the "5 X 1" and the d 3 X d 3 structures, respectively. Contour maps of the Patterson (autocorrelation) function show t h a t in none of t h e structures a r e the Au-atoms situated on high-symmetry lattice sites, contrary to recently proposed models. A trimer model is deduced for the d 3 ~ dstructure, 3 whereas the "5 X 1" actually is a disordered 5 X 2 structure. We propose a model with two Au atoms per 5 X 1 subunit for this structure.
Introduction
Au on S i ( l l 1 ) forms a variaty of surface structures prior to the Au/Si interface formation a 5 x 1 structure is formed about 0.4 monolayer (ML) of Au, a d 3 x d 3 ~ 3 0 0 structure is formed a t 1.0 ML and a 6 x 6 structure a t 1.5 ML.
Inbetween these coverages there are disordered structures with diffuse Low Energy Electron Diffraction (LEED) patterns andfor mixed phases2'. To clarify this complicated pattern we have started a structural investigation by x-ray diffraction with the intention to detail the atomic geometry of each structure.
Surface x-ray diffraction i s well suited for t h i s purpose, because t h e interpretation of data is relatively straightforward and the contour map of the Patterson (autocorrelation) function gives direct insight in the atomic structure4'.
We will report on the structure of the Au atoms in the "5X 1" and d 3 X d 3 phases.
Since Au(Z = 79) is scattering x-rays much more than Si(Z = 14), x-ray diffraction is mostly sensitive to the topology of the Au atoms and only marginal sensitive to the Si layers. We will in this paper concentrate about the in-plane structure of the Au-atoms. There has recently been some debate about both the "5 X 1" 3+5) and the d 3 X d 3 6,7) structures and different models were proposed.
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1989717
Experimental
The x-ray diffraction experiments were performed on t h e vertical scattering diffractometer a t the 32-pole wiggler W1 beamline a t Hamburg Synchrotron Radiation Laboratory (HASYLAB). The Si(ll1) surfaces were prepared a t the FLIPPER I1 photoemission beamline. The surfaces were cleaned by repeatedly annealing to 900 O C and slowly cooling down until a sharp 7 x 7 LEED pattern emerged. The Au was evaporated onto the surface a t a rate of 0.3 MLImin by a n effusion cell held a t 1180°C
.
The surface was then transferred under UHV- conditions to a small portable UHV x-ray cell, which was mounted on the diffractometer. The pressure in the cell was in the 10-l0 mbar region and no decay of the samples could be observed over the measurement period (- 1 day). The samples were aligned on their optical surface by total reflection, such t h a t t h e angle of incidence of the x-rays to the surface could be kept fixed during the data collection. In order to maximize the diffracted intensity and minimize the background t h e angle of incidence was s e t to t h e critical angle for total reflection4). For the " 5 x 1 " structure a s e t of 52 in-equivalent, in-plane, fractional-order structure factor intensities was obtained, and 13 for the d 3 X d 3 structure. It is on those intensities we base the further analysis.The "5 X l" S t r u c t u r e
The "5 X 1" structure was first found in 1969~) and has been shown by microprobe RHEED to nucleate a t stepsg). The 5 X 1 LEED pattern is accompanied by diffuse streaks running through the half-order positions10), although the LEED spots belonging to the 5 X 1 pattern are sharp. Lipson and Singer proposed that the structure is not 5 X 1, but rather 5 x 2 with one-dimensional disorder between the 5 X 1 subunitslO), a s shown in figure l. We start by discussing the 5 X 1 pattern, but note t h a t we also observed the streaks by x-ray diffraction.
5x2 cell ~ ( 1 0 x 2 ) cell
Figure 1 . The disordered 5x2 structure proposed by Lipson and singerzo).
Note that the structure is a random mixture between rows of 5x2 and c ( l 0 X2) cells, respectively.
Information about the atomic geometry inside the unit cell can be obtained from a contour plot of the Patterson function
P(x,y) =
1
l ~ ~ ~ l ~ c o s ( 2 r 1 ( h x+
ky))hk (1)
where l'Fhk12 are the structure factor intensities. This gives a map of the inter- atomic vectors between the atoms in the unit cell. On basis of such maps wrong models may be excluded. The contour map for the "5X 1" structure is shown in figure 2a. The full 5 X 1 unit cell is shown and contains two irreducible units for the Patterson function which are related by inversion symmetry. Firstly the map contains four strong non-origin peaks which must correspond to Au-Au interatomic vectors. Hence, a t least four Au-atoms must be present in the 5 X 1 unit cell.
Figure2. (a) A contour map of the Patterson function for the 5x1 structure.
The origin rises 10 contour levels. The triangle is the irreducible unit.
(b) A model for the Au-structure with a 5x1 unit cell. Note the Au coverage is 0.8 ML.
(C) A model proposed for the 5 x 2 structure. The coverage is 0.4 ML.
The two 5 x 1 subunit~projected into one 5x1 cell gives the model in (b).
Secondly, non of the interatomic vectors are close to vectors between atoms residing on lattice sites. This means that the Au-atoms are situated a t low- symmetry sites, in correspondance with a recent x-ray standing wave experiment by Berman e t Thirdly, if the 5 x 1 unit cell contains four Au-atoms the coverage would be 0.8 ML, contradicting previous coverage d e t e r m i n a t i ~ n s ~ ~ ~ ) . However, if the four Au-atoms are distributed in a 5 X 2 unit cell, the coverage is 0.4 ML. By only measuring reflections corresponding to a 5 X 1 structure, the two 5 X 1 subunits in the 5 x 2 cell are projected into one 5 X 1 unit cell. So if the structure i s 5 x 2 with two Au atoms in each 5 X 1 subunit (giving the required coverage) i t will show up in our Patterson map as a four atom model.
A model reproducing the Patterson map is shown in figure 2b. The structure is least-square refined by allowing the atoms to move along the (2,l) symmetry direction. The chi-square is defined as
where N is the number of measured reflections, p the number of free parameters in the fit, ohk the uncertainties on the measured intensities and w a n arbitrary scale factor. With three atomic displacements and one isotropic Debye-Waller factor for all atoms, the best fit gives x2 = 18. A dramatic improvement can be obtained by including a n anisotropic Debye-Waller factor. This i s already suggested by the Patterson map where all the peaks are elongated in the (0,l) direction. The best fit now has $=9.8 with a Debye-Waller factor By along the (2,l) direction of B! = 0.0k0.4
A2
and B, = 7.0f 1.0A2
perpendicular to it. The reason for the large B, is probably not thermal vibrations but could be due to the disorder in that direction. The refinement of the structure must be done by including reconstruction/relaxation in the underlying Si-substrate, but the main features contained in the x-ray diffraction pattern are described by the model i n figure 2b. On basis of that model we propose the model in figure 2c for the 5 X 2 structure. The Au-atoms are concentrated in rows fairly widely apart, which properly is the reason for the disorder.The d 3 X d 3 ~ 3 0 O S t r u c t u r e
Two different models have recently been proposed for the d 3 ~ d 3 structure, a trimer model6) and a honeycomb model7). These are shown in figure 3b and 3c, respectively. The trimer model has a coverage of 1.0 ML, whereas the honeycomb model has 213 ML. The Patterson map for our dataset is shown in figure 3a. There is only one non-origin peak in the irreducible unit corresponding to a n Au-Au interatomic vector. This vector can clearly not be reproduced by the honeycomb model, which hence must be excluded. However, the vector is contained in the trimer model. A least-square refinement of t h a t model gives
$=
10. A better fit ofx2 = 5.5 can be obtained by including a strongly relaxed Si-layer, however, more accurate data are needed to pinpoint the Si-atoms.
a ) Patterson map b) Trimer model c ) Honeycomb model
Figure3. ( a ) A c o n t o u r m a p of t h e P a t t e r s o n f u n c t i o n f o r t h e
~ i ( 1 1 1 ) d 3 X ~ ~ R ~ O O - A U structure. The origin rises 1 0 contour levels.
(b) The trimer model proposed by Oura et ~ 1 . ~ ) . The vector between the Au-atoms corresponds to the peak in (a).
(C) The honeycomb modelproposed by Huangand williams7).
The G e ( l l 1 ) surface also has a d 3 ~ d 3 structure induced by A U ~ ) . If this structure is similar to that for S i ( l l l ) , substrate relaxation would be easier to identify because Ge(Z = 32) is scattering more than Si.
Conclusions a n d Acknowledgements
We have determined the structure of the Au atoms in the "5 X 1" and d 3 X d 3 structures on the Si(111) surface by surface x-ray diffraction. In none of the structures do the Au-atoms reside on high S rnmetry sides of the substrates. The
"5 X 1" structure is, a s previously proposedro), a disordered 5 X 2 structure. The trimer model proposed by Oura e t aL6)for the d 3 ~ d 3 structure is confirmed. To obtain the reconstruction/relaxation of the Si substrate, more accurate data are needed.
T h i s work was supported by t h e Bundesministerium fiir Forschung u n d Technologie, the Max-Planck Society and the Danish National Research Council.
We would like to thank the staff of HASYLAB for their kind assistance.
References
+
Also a t Max-Planck-Institute for Solid State Research, Heisenbergstrasse 1, Stuttgart,1) G. Le Lay. Surf. Sci. 132,169 (1983).
2) K. Higashiyama, S. Kono and T. Sagawa. Jap. Jour. App. Phys. 25, L117 (1986).
3) Y. Yabunchi, F. Shoji, K. Oura and T. Hanawa. Surf. Sci. 131, L412 (1983).
4) For a review see: R. Feidenhans'l. Surf. Sci. Rep. (1989) to be published.
5) L.E. Berman, B.W. Batterman and J.M. Blakely. Phys. Rev. B38, 5397 (1988).
6) K. Oura, M. Katayama, F. Shoji and T. Hanawa. Phys. Rev. Lett. 55, 1486 (1985).
7) J.H. Huang and R.S. Williams. Phys. Rev. B38,4022 (1988).
8) H.E. Bishop and J.C. Riviere. Brit. J. Appl-Phys. (J. Phys. D) 2,1635 (1969).
9) M. Ichikawa, T. Doi and K. Hayakawa. Surf. Sci. 159,133 (1985).
10) H. Lipson and K.E. Singer. J. Phys. C7,12 (1974).
