Surface Science 132 ( 1983) 40-45 North-Holl~d Pubhshing Company
Si(111)2 x 1 STUDIES
F, HOUZAY *, G. GUICHAR **, R. PINCHAUX, G. JEZEQUEL ***, F. SOLAL, A. BARSKY, P. STEINER **** and Y. PETROFF Laboratorre LURE,
209 C, Universit6 de Parts - Sud, F- 9140.5 Orsay CPdex, France
Received 20 October 1982; accepted for publication 30 December 1982
Photoemission experiments on the cleaved Si( 111)surface show that two kinds of cleavages are obtained with different photoemission properties. When the domain is aligned along the cleavage direction the energy dispersion is in very good agreement with the theoretical calculations of Northrup and Cohen which they based on a recent chain model proposed by Pandey. However, the origin of a second structure around j is still an open question.
The cleaved Si( 111)2 X 1 surface has been studied extensively by various experimental techniques: LEED, angle resolved phot~mission, ion scattering, etc. Interpretation of the LEED data led to a buckled surface 2 x 1 geometry, in which alternate rows of surface atoms are raised and lowered in the most probable geometry [l]. However, experimentally and theoretically there have been inconsistencies [2-51. The first angle resolved photoemission experiment on Si( 111)2 X 1 was reported by Rowe, Traum and Smith . Their results were in disagreement with the calculation of Pandey and Phillips , the major discrepancy being the energy location of the dangling bond surface state with respect to the valence band rn~irn~. Subsequently, Guichar et al.  showed that the dangling bond surface state did not have any dispersion along rp. Later on, in new angle resolved photoemission measurements Himpsel et al.  reported the existence of two surface bands at - 0.15 eV and - 0.7 eV with respect to the valence band maximum, which have a very small band dispersion (- 100 meV) along Fj and I?I’. These results and the small charge transfer  between surface atoms as obtained from core shift measurements led these authors and theoretical researchers [9,10] to suggest that correlation
* ** *** ****
And CNET, 196 Rue de Paris, F-92220 Bagneux, France. And Minis&e de 1’Industrie et de la Recherche, 1 Rue Descartes, F-75005 Paris, France. And Laboratoire de Spectroscopic, Universitl? de Rennes I, F-35042 Rennes, France. Permanent address: Fachbereich Physik, Universitat des Saarlandes, D-6600 Saarbrucken, Fed. Rep. of Germany.
0 1983 North-Holland
F. Howay et al. / Si(I1 I)2 X I smdies by ARP
effects might be important. However, the agreement between the measured band dispersion and the theoretical calculations was still not satisfactory. Meanwhile Northrup et al. [lo] showed that the 2 x 1 buckled structure was unstable. Moreover, Uhrberg et al. [ll] questioned the interpretation of Himpsel et al.  and concluded that there was only one state and that the second one was due to a multi-domain effect. In this paper we report angle resolved photoemission experiments on a 2 x 1 single domain cleaved Si( 111). Two kinds of cleavage are obtained: when the domain is aligned along the cleavage direction, the energy dispersion along ry and Tj is in a very good agreement with the calculation of Northrup and Cohen  based on the chain model proposed by Pandey . The width of the surface state band (- 600 meV) is smaller that that reported by Uhrberg et al. [ 1 l] ( - 830 meV) but in very good agreement with the data of Himpsel et al. . We also show that the second structure observed around the 5 point cannot be due to a multi-domain effect and that its origin is still unclear. The angle resolved photoemission spectra have been measured using synchrotron radiation from the AC0 storage ring (540 MeV) at LURE Orsay. The spectra have been obtained for photon energies between 21 and 48 eV with an overall energy resolution (monochromator + analyzer) smaller than 250 meV and an angular resolution of - lo. All energies have been measured with respect to the Fermi level E, of a Au sample. However, the energies can also be referenced to the top of the valence band E, by using E, - E, = 0.48 eV 1141. The samples were n-type and were cleaved in a vacuum of lo-” Torr. Because of the symmetry of the dangling bond surface state (sp,) the measurements were performed with the light beam set at 30” from grazing incidence. After the Si( 111) surface was cleaved, we generally observed one, two or three domains with or without steps. In our case the spot size of the light beam was focused down to 0.5 X 0.5 mm2, to avoid contributions from different parts of the sample. We have found that multi-domains can yield photoemission curves with totally different properties, so in this paper we discuss only results obtained on a single domain. Our results show that quite different photoemission curves can be obtained depending of the cleavage. When the crystals are cleaved along the  direction, two kinds of result are obtained: (a) The domain is aligned along the cleavage direction as shown in fig. la. In that case the LEED pattern is very bright, the intensity of the half-order spots being of the same order of magnitude as the integral ones. The 2 x 1 single domain covers the whole sample (5 X 5 mm2). The photoemission experiments were performed at Aw = 48 eV to emphasize the surface contribution. In that case the surface state is very strong in each part of the Brillouin zone (in contrast to experiments performed at 10 eV). Data were taken for positive and negative angles to determine precisely the normal emission
F. Houzay et al. / Si(111)2 X I studies by ARP
Fig. 1. Two kinds of cleavages obtained for the Si( 11I)2 X 1: (a) domain along the cleavage direction [zl I]; (b) domain at 60’ from the cleavage direction.
(6 = 0). The results are presented in fig. 2. The lines correspond to the calculations of Northrup and Cohen [ 121 (based on Pandey’s chain model [ 131). The solid circles correspond to the data of Uhrberg et al. [ 1 I]; the squares to those of Himpsel et al. . To avoid the problem of the different values used for E, - E, [ 141 we have arbitrarily fixed the experimental point at 5 as in ref. [ 121. We observe the following trends: (1) There is a remarkable agreement between our data and the data of Himpsel et al. . There is also a good agreement with the data of Uhrberg et al. [ 1 l] except for the two points at the centre of the Brillouin zone (between r and 5). This discrepancy is important because it changes appreciably the width of the surface state (600 instead of 840 meV). In a simple tight binding model  it means that the second nearest neighbour interaction has to be very strong (V, = 0.6 ev> which is not compatible with Pandey’s model. It can also be seen that the agreement with the theoretical calculation is quite good except that in our experiment there is no minimum between r and 5. However, it is necessary to recall that the structure in this part of the BZ is a surface resonance which means that the shape of the dispersion could be changed easily. (2) The other important feature is the observation of a second state around 5 as shown in fig. 2 (indicated by x). We will now discuss the different explanations possible. (i) Uhrberg et al. [ 1 l] have suggested that this weak peak was a multi-domain effect. This is very difficult to believe in our case considering the fact that the intensity of this second structure is 30% to 40% of the intensity of the
el al. / Si(lll)Z
X I studies by ARP
INITIAL ENERGY ( ev I Fig. 2. Surface state dispersion from 7 to 5 for the chain model obtained by Northrup and Cohen. Notice that the experimental points of Uhrberg et al. have been shifted upwards by 0.3 eV so that the theory and experiment agree at 5. To avoid the problem of the value of E, - Ev we have also used the same approach. Fig. 3. Angle resolved photoemission line obtained at Ao = 48 eV.
angles 8 afong the fi symmetry
surface dangling bond state (see fig. 3) and that in our LEED pattern there is no trace of a second domain. (ii) Electron correlation effects as in the case of the buckled model can be invoked but it seems that the effect in the chain model is very small . (iii) The coexistence of the chain model and a molecular model as it has been suggested by Chadi . In this model the atomic displacements and rebonding leading to the new structure are very similar to those for the B bonded chain model, except that the shear distortion of the first double layer of atoms occurs at 120” with respect to the long axis of the unit cell instead of parallel to it as in the chain model, However, the relative stability of the ?r bonded chain and molecule structure have not yet been tested and the calculation of the energy position of the dangling bond is not yet available. (iv) Finally the last possibility is a strong indirect transition due to surface
I? Houzay et al. / Si(I Ii)2 x I studies by ARP
z: 3 i
Fig. 4. Surface state dispersion photon energies.
for a domain
at 60’ from the cleavage
[zl 11, for different
defects. One has to be reminded that for most of the metals (Cu, Ag, Au) indirect transitions are observed. A phonon contribution could also be important even if the Debye temperature is large for the volume (655 K). Finally, the origin of the second structure around 5 is still unclear. (b) When the domain is at 60” of the cleavage direction, the sample shows a LEED pattern (fig. lb) with half-order spots weaker that the integral ones. The surface has often different parts: single domain, two or three domains with or without steps. The photoemission spectra obtained on a single domain give the results shown in fig. 4. A flat state between r and 5 and some dispersion around the comer of he Brillouin zone (which is not a symmetry point for the 1 x I) are observed. The experiment was repeated on two cleavages and done at three photon energies Aw = 21, 30 and 48 eV to avoid any ~sa~~rnent. Again a second structure appears around this point. Although no steps are observed by LEED we think that this different dispersion is related to the existence of steps on the surface. In fact a single domain with steps shifts the dispersion. Measurements to study the direction and the number of steps are underway. In conclusion, we have shown that two kinds of cleavages are obtained on the Si( 111)2 x 1 surface. When the domain is along the cleavage direction [zl 1] the agreement with a calculation based on the chain model is very good. However, the origin of the second structure around 5 is still unclear. Experiments at 10 K to study the 1 x 1 and the 2 x 1 reconstructions of Si and Ge are in progress.
[I] D.J. Chadi, J. Vacuum Sci. TechnoI. 15 (1980) 1244.  J.E. Rowe, M.M. Traum and N.V. Smith, Phys. Rev. Letters
33 (1974) 1333.
et al. / Si(Iii)2
X 1 studies by A RP
j3] KC. Pandey and J.C. Pbithps, Phys. Rev. Letters 32 (1974) 1433.  M.W. Parke, A. McKinley and R.H. Williams, J. Phys. Cl 1 (1978) L993; A. McKinley, R.H. Williams and A.W. Parke, J. Phys. Cl2 (1979) 2447.  G.V. Hansson, R.Z. Bachrach, R.S. Bauer, D.J. Chadi and W. Gopel, Surface
Sci. 99 (1980)
 2: Guichar, F. Houzay, R. Pinchaux and Y. Petroff, Solid State Commun. 38 (1981) 809: F. Houzay, G. Guichar, R. Pmchaux and Y. Petroff, J. Vacuum Sci. Technol. 18 (1981) 860.  F.J. Himpsel, P. Heimann and D.E. Eastman, Phys. Rev. B24 (1981) 2003. [S] F.J. Himpsel, P. Heimann, T.-C. Chiang and D.E. Eastman, Phys. Rev. Letters 45 (1980) 1112; S. Brennan, J. St&r, R. Jaeger and J.E. Rowe, Phys. Rev. Letters 45 (1980) 1414.  R. Del Sole and D.J. Chadi, Phys. Rev. B24 (1981) 7431. [IO] J.E. Northrup, J. Bun and M.L. Cohen, Phys. Rev. Letters 47 (1981) 1910. [1 I] R.I.G. Uhrberg, G.V. Hansson, J.M. Nicholls and S.A. Flodstrom, Phys. Rev. Letters 48 (1982) 1032. [ 121 J.E. Northrup and M.L. Cohen, Phys. Rev. Letters 49 (1982) 1349.  KC. Pandey, Phys. Rev. Letters 47 (1981) 1913. [ 141 We use a value of 0.48 eV for E, - E, determined by G. Guichar (Thesis, Paris (1978), unpublished). We believe that this value is more accurate that the value of 0.33 eV reported by Gobeli and Allen. [IS] D.J. Chadi, to be published. [ 16) J.E. Northrup and M.L. Cohen, unpublished.  D.J. Chadi, Phys. Rev. B26 (1982) 4762.