Growth and characterization of the interface

surface science ELSEVIER Surface Science 352-354 (1996) 628-633 Growth and characterization of the Re/Si(111) interface A. Siokou a, S. Kennou a,1, S. Ladas a,., T.A. Nguyen Tan b, j._y. Veuillen b a Surface Science Laborctory, Department of Chemical Engineering, University of Patras, and Institute o f Chemical Engineering and High Temperature Chemical Processes~FORTH, GR-26500 Rion, Patra, Greece b CNRS-LEPES, BP-166, 38042 Grenoble Cedex, France Received 5 September 1995; accepted for publication 31 October 1995 Abstract The structure and electronic properties of the interfaces formed by Re evaporation on clean Si(111)-7 X 7 and subsequent annealing were investigated by low energy electron diffraction (LEED), Photoelectron spectroscopy (XPS/XAES, UPS) and work function (WF) measurements. Rhenium was deposited from submonolayer up to 50 ML coverages and gradually annealed up to 1100 K. Deposition causes rapid disappearance of the 7 X 7 reconstruction and the substrate UPS features, a gradual increase in the WF from 4.5 to 5.1 eV and negligible chemical shift in the XPS/XAES peaks. The interface remained abrupt with no intermixing, as metallic Re grew in a simultaneous mulfilayers fashion above 0.8 ML coverage. Annealing of 50 ML of Re causes a Si enrichment of the overlayer above 600 K and the formation at around 950 K of a continuous layer of silicide which remains stable up to 1100 K, whereas annealing of Re layers close to 1 ML creates ReSi 2 islands on the substrate surface. Keywords: Growth; Rhenium; Silicides; Silicon; Work function measurements; X-ray photoelectron spectroscopy 1. Introduction The semiconducting character of ReSi 2 and its thermal stability make it a very interesting material for applications in optoelectronic devices [1], in particular if it can be grown epitaxially on silicon substrates. Epitaxial ReSi 2 films, 1500 ,~ thick, of good crystalline quality have been prepared on Si(100) by Mahan et al. [2] using reactive deposition and by Kim et al. [3] using ion implanafion. Chu et al. [4] have observed that upon furnace annealing of * Corresponding author. Fax: +30 6199 3255; e-mail: ladas@ rea.iceht.forth.Gr. l Permanent address: University of loannina, Department of Physics, P.O. Box 1186 GR 45110 Ioannina, Greece. ~ 400 ~ Re deposited on Si(111) and Si(100) epitaxy of ReSi 2 occurred only in a fraction of the silicon surface. Electrical and optical measurements have concluded that ReSi 2 is a narrow-gap semiconductor [5,6]. The formation and the crystallographic, electronic and optical properties of 2 0 - 3 0 0 A ReSi 2 films grown on S i ( l l l ) using in situ (LEED, XPS and UPS) and ex situ (GIXD) techniques have been studied by Nguyen Tan et al. [7]. Glancing-incidence X-ray diffraction (GIXD) showed that the ReSi 2 films obtained upon annealing above 600°C are epitaxial although LEED did not exhibit any ordered patterns. However, the early stages of the R e / S i interface formation have not been studied so far. This paper reports the results o f XPS, XAES, UPS, LEED and W F measurements used for the 0039-6028/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0 0 3 9 - 6 0 2 8 ( 9 5 ) 0 1 2 1 7 - 6 A, Siokou et a l . / Surface Science 352-354 (1996) 628-633 629 investigation of the Re/Si(111) interface formation when Re is vapour deposited in UHV from submonolayer up to 50 ML coverages at 400 K and gradually annealed up to 1100 K. min duration. the electronic adlayer were measurements 2. Experimental procedure 3. Results and discuss.ion The measurements were performed in two different apparatuses previously described [8,9]. The XPS, XAES and WF measurements were carried out in a conventional turbo-pumped UHV chamber whereas the LIPS and LEED measurements in a two-chamber, multitechnique system. Rhenium was evaporated by electron bombardment heating of a Re tip, on the S i ( l l l ) - 7 × 7 surface of a n-doped single crystal with a resistivity 1 12-cm. During evaporation the temperature of the sample was maintained at 400 K. The pressure during evaporation was < 5 × 10 - 1 ° mbar. After prolonged exposure to the background, small amounts of oxygen contamination were observed on the Re layers. The presence of Re also facilitated the oxygen uptake by the silicon substrate. To reduce oxygen contamination the measurements were taken as quickly as possible after evaporation. The evaporation rate, J, was estimated by XAES from the initial decay of the SiLVV signal. The I/I o = exp exponential attenuation law (--d/ASRiLVV), with h siLvv ~, 4 ,~ [10], was used to obtain an average adlayer thickness d. Assuming a sticking coefficient S = 1, the adlayer thickness d is related to the evaporation rate J, by d = Jtade. The parameter aR3e is equal to the equivalent volume of the Re atom, given by CtR3e= M p - 2NA 2, where p = 21 g cm -3 the Re metal density, M = 186.2 its atomic weight and NA= 6.023 × 1023 the Avogadro's number. Using the above equations and the first few experimental points of the Si LVV intensity versus time curves, one can estimate J, a typical value of which was 3.1 × 1012 at cm -2 s -2. The total Re coverage was expressed in monolayers (ML). The thickness of one dense monolayer (dr,l) was defined to be d M = a r e = 2.45 ,~, corresponding to an adatom surface density of 1.66× 1025 atoms 3.1. Rhenium deposition cm -2. After the desired amount of Re had been deposited on the Si surface, the crystal was annealed at increasing temperatures up to 1100 K, by steps of 10 After each step, the composition and and structural properties of the Re checked by XPS, XAES, LIPS, WF and LEED. Fig. 1 shows the variation with coverage of the SiLVV and Re4f7/2 peak heights, as well as of the WF upon Re deposition on S i ( l l l ) at 400 K. No change of the energy position and the shape of the Auger and photoelectron peaks occurred during deposition. Since the inelastic mean free path (IMFP) of the Si LVV electrons in Re is ~ 4 ,~ most of the signal originates from the first 3 - 4 atomic layers. The fast attenuation of SiLVV during Re evaporation down to noise level when the Re coverage is ~9 ~ 4 ML, indicates that no formation of 3D-crystallites and no substantial intermixing between Re and Si atoms occur at this temperature. In order to determine the growth mode of the Re films on the Si substrate we used the procedure first proposed by Biberian [11] and plotted the XPS (or XAES) signal of a substrate peak ( I s) as a function of the XPS (or XAES) signal of an adsorbate peak it50f: t 5,0 -~ o " 0 ".. 1 2 3 4 4,5 5 Re COVERAGE, e / ML Fig. 1. The variationof the SiLVV and Re4f7/2 peak heights and of the work function during Re evaporation on Si(lll) as a function of the Re coverage, (9, expressedin monolayers. A. Siokou et aL / Surface Science 352-354 (1996) 628-633 630 16 14 12 . ~ • 10 i i i I i i ~k(SiLW)=4/~o ~,(Re4f)=15.5A iRe,==470 4 2 01~ 50 100 150 200 250-300 IRe4fzn / a.u. Fig. 2. The variation of the SiLVV peak height as a function of the Re4fT/2 peak height during Re evaporation. With solid lines are shown the theoretical curves for the layer by layer growth mode (FM) the simultaneous multilayers (SM) and the monolayer plus simultaneous multilayers (ML +SM). (IA) at quite different energies. Using such plots instead of I versus t plots one can avoid errors arising from possible changes of the sticking coefficient or the evaporation rate. Fig. 2 shows the variation of the Si LVV (KE = 92 eV) signal as a function of the Re4f7/2 (KE = 1214 eV) signal during Re deposition. In the same figure the solid lines represent the theoretical curves for the layer-by-layer (Frank van der Merwe, FM), the simultaneous multilayers (SM), in which each successive Re layer begins to grow before the completion of the preceding layer, and the monolayer plus simultaneous multilayers (ML + SM) growth modes. For the FM mode straight line plots are obtained. Breaks occur at the completion of each layer and the beginning of the next [12]. The curve for the SM mode exhibits an exponential decay and has been obtained following the work of Barthes and Rolland [13]. For the ML + SM mode the curve coincides with that for the FM mode up to one monolayer and follows an exponential decay (SM) beyond the first monolayer. The theoretical curves in Fig. 2 have been obtained using the following IMFP values in Re: 4 A for Si LVV electrons and 15.4 ,~ for Re4f7/2 electrons [10] whereas the intensity of the 4f7/2 electrons from a thick Re film (IR~,~o) was measured under the same experimental conditions. In Fig. 2 one can see that the first experimental points follow closely the straight line that represents the formation of the first monolayer, up to /Re = 50 which corresponds to O -- 0.8 ML. After this value the points deviate from the lines describing the layer by layer growth, being more close to the SM plots within experimental error. At high IRe the /SiLVVis close to noise level and the experimental error is larger. In this region the three theoretical curves begin to merge together again. The change of the WF during Re evaporation as a function of the Re coverage shown in Fig. 1 supports the above conclusion. The value • = 4.5 eV corresponds to the work function of a clean Si(111)-7 X 7 surface [14] with the same doping level as the one used in this work. The work function increases monotonically with Re coverage and when (9 = 1.3 ML it saturates at • - - 5.1 eV which is the value for metallic Re [15]. The WF does not saturate at O = 1 ML indicating that Re grows in a SM mode. At O - - 1 the Si surface is not yet completely covered by Re, something which happens only at (9 = 1.3 ML. The work function is very sensitive to the surface composition. In the case of reactive interfaces like Nb/Si(111) and Ta/Si(111) [14] the variation of WF with coverage is not monotonic but there exist intermediate plateaus indicating metalsilicon intermixing and the formation of new phases with stable composition. On the contrary, in the case of abrupt interfaces like Mo/Si(111) and W/Si(111) [14] the WF changes monotonically as in the case of Re. An estimate of the initial dipole moment of the Re adatoms is obtained from the initial slope of the WF versus O curve using the Helmholtz equation [16]. A value of P0=0.11_+0.05 D is obtained which indicates very small charge transfer between Re and Si in agreement with the XPS and XAES measurements where there is no energy shift of the Si and Re peaks during deposition. The abruptness of the Re/Si(111) interface near room temperature is also confirmed by the UPS measurements. Fig. 3 shows with solid lines some representative energy distribution curves (EDC) of ultrathin Re deposits. The Fermi level position, E F, was determined by using the polycrystalline Ta sample holder. The lower curve corresponds to the clean A. Siokou et al./ Surface Science 352-354 (1996) 628-633 i 8 i i' i ! t UH;IS2 ! i i 3.2. Annealing of the Re deposits 50ML Re/-~ ...." "....... pP 6 ...... ...-" Re ! ~t 1Mk n. . ..... -t. J ~ "" 2 0 ' ' 631 ' ' ' -8 -7 -6 -5 -4 - 3. I 11 -2 - 0 0 1 BINDING ENERGY Es / eV Fig. 3. The energy distribution curves obtained with HeI (21.2 eV) on Re deposits near room temperature (solid lines) and after annealing at 950 K (dotted lines). Si(111) surface and the upper one to a thick ( ~ 50 ML) Re film• We note that deposits of a few tenths of a ML remove rapidly the characteristic electronic features of the 7 X 7 surface. At coverage of 1 M L the EDC is close to that of the metallic Re. The Fermi step - characteristic property of the conductors - has already been formed indicating the metallic character of the Re deposit. At the initial stages of Re deposition, LEED observation shows a rapid removal of the 7 X 7 reconstruction. At @---0.1 ML only spots of the 1 × 1 structure remain, and become very weak near 19 = 0.5 ML, which corresponds to a R e / S i ratio of about 1 for the top Si layer• Above this coverage only a strong diffuse background remains indicating the disordered nature of the adlayer. Similar observations have been obtained for Mo on S i ( l l l ) [17] where the spots of the 7 x 7 reconstruction disappear at 19 = 0.3 and after O ~ 0.7 the 1 X 1 structure disappears too. For W on S i ( l l l ) the same phenomena occur at 19 = 0.1 and @ = 1 respectively [18]. The dotted lines in Fig. 3 show the energy distribution curves (EDC) obtained with He I on the corresponding Re deposits after annealing at 950 K. The top dotted curve of the figure corresponds to a thick ReSi 2 film. The main characteristic of those curves is the drastic loss of electron density at the Fermi level, which confirms the semiconducting nature of the new phase formed on the surface upon thermal treatment, even for submonolayer quantities of deposited Re. Fig. 4 shows the variation of the S i L V V , Si2p and Re 4f7/2 peak heights as a function of temperature upon annealing 1.5 M L Re deposited on S i ( l l l ) at 400 K. The same figure also shows with dotted line the variation of the IRe4fT/2/lsi2p ratio. The energy position and the shape of all the XPS and XAES peaks does not change upon annealing. The intensity changes (increase o f Si and decrease of Re) start above 700 K for the 1.5 M L deposit. For comparison, similar changes start at ~ 550 K in the case of thick Re deposits [7]. These changes result from Si diffusion in the Re deposit and silicide formation. In the case of 50 M L deposits, the semiconducting silicide ReSi 2 is formed at 950 K and 120 . . . . . MgKa XPS • 100 . / 3,0 1.5ML Re t ] 2,5 * ~ ..-i 80 z 2.0 , - ............=.............. 60 1,5 ~, UJ I--z_ N ,~. 40 20 " o 4;0 1,0 ~-- 0,5 Isa_w (x4) II---- :. • 6;0 8;o lO;O ,o,o 1200 Annealing Temperature T, K Fig. 4. The variation of the SiLVV, Si2p and Re4fT/2 peak heights and of the /ReafT/2/Isi2p intensity ratio (dotted lines) as a function of temperature, upon annealing 1.5 ML Re deposited on Si(ll 1). 632 A. Siokou et al. / Surface Science 352-354 (1996) 628-633 remains stable up to 1100 K in agreement with previous results [7,19]. Between these temperatures (950-1100 K) the intensities of Si2p and R e 4 f peaks are stable. In the case of the ultrathin (1,5 ML) Re film, silicon diffusion in Re starts at 700 K but up to 1100 K the Re and Si signals do not stabilize. The ratio IRe4f7/2/Isi2~v from Fig. 4 is used for the estimation of the relative atomic concentration of Re and Si, in the annealed layer. In the case of a thick silicide film this ratio has the value 3.5 between 950 and 1100 K [19]. In Fig. 4 at 1100 K this ratio has the value 0.78. If the whole amount of Re (1.5 ML or 3.67 ,~) had been transformed to ReSi 2 then the silicide film would have a thickness 2.55 times as much, that is 9.4 ,~ [20]. The expected value of the IRe4f7/2/Isi2ov ratio for a 9.4 A silicide film on a silicon substrate is 1.95. The fact that the experimental value of this ratio, 0.78, is smaller may have two possible explanations: either islanding of the ReSi 2 film leaving bare silicon from the substrate, or formation of a homogenous ReSi 2 film with extra silicon on the surface. The latter is less likely since it is not observed in the case of the thick deposits. The islanding process could start at even lower temperature than that at which the silicide formation is completed. The LEED observations support the first explanation i.e. the formation of ReSi 2 islands. At 950 K the spots of S i ( l l l ) 1 X 1 structure appear, and at T > 950 K appear the spots of the 7 X 7 reconstruction. The value 0.78 for the ratio corresponds to thick ReSi 2 islands which cover about 40% of the substrate area. That means that there exist wide areas of bare silicon, which are enough to show the 7 X 7 reconstruction. The work function of the adlayer upon annealing decreases with increasing temperature and reaches a value of 4.5 + 0.1 eV at 950 K that is close to the WF of the clean Si(111) surface. For comparison the WF value for a bulk ReSi 2 film is ~ 4.7 eV. lRe4fT/2//lSi2p 4. Conclusions The results concerning the formation of the R e / S i ( l l l ) interface lead to the following conclusions. (1) The R e / S i ( l l l ) is an abrupt interface at 300-400 K. There ~is no intermixing between the R e - S i atoms. Re grows on Si in the simoultaneous multilayer mode, following the formation of 80% of the first monolayer. At coverages > 1.3 ML the Re film exhibit a clear metallic character. During deposition the charge transfer between R e - S i atoms is negligible. (2) Annealing of ultrathin Re deposits leads to their enrichment in Si which starts at T ~ 700 K. At 950 K, islands of semiconducting ReSi 2 are formed, leaving bare areas of the Si substrate, where the 7 X 7 reconstruction is again visible. Acknowledgement Financial support from a Greek-French Scientific and Technological Cooperation (PLATON) project is gratefully acknowledged. References [1] J. Derrien, J. Chevrier, V. Le Thanh and J.E. Mahan, Appl. Surf. Sci. 56-58 (1992) 382. [2] J.E. Mahan, K.M. Geib, G.Y. Robinson, R.G. Long, Yan Xinghwa, G. Bai, M.A. Nicolet and M. Nathan, Appl. Phys. Lett. 56 (1990) 2439. [3] K.H. Kim, D.H. Kim, S,T. Nam, J. Jule, I.H. Kim, S.C. Kim, J.Y. Lee, M.A. Nicolet and Gabg Bai, J. Appl. Phys. 74 (1993) 1046. [4] J.J. Chn, J.L. Chen, K.N. Tu, J. Appl. Phys. 62 (1987) 461. [5] T. Siegrist, F. Hullinger and Travaglini, J. Less-Common Met. 92 (1983) 119. [6] C. Krontiras, L. Gr6nberg, I. Suni, F.M. d'Heurle, J. Tersoff, J. Engstr6m, B. Karlsson and C.S. Peterson, Thin Solid Films 161 (1988) 167. [7] T.A. Nguyen Tan, J.Y. Veuillen, P. Muret, S. Kennou, A. Siokou, S. Ladas, F.L. Razafindramisa and M. Brunel, J. Appl. Phys. 77 (1995) 2514. [8] A. Siokou, S. Kennou and S. Ladas, Surf. Sci. 307-309 (1994) 810. [9] D.B. Lollman, T.A. Nguyen Tan, N.Y. Veuillen, P. Muret, K. Lefki, M. Brunel and J.C. Dupuy, Appl. Surf. Sci. 65/66 (1993) 704. [10] S. Tanuma, G.J. Powel and D.R. Penn, Surf. Interf. Anal. 20 (1992) 77. [11] J.P. Biberian and G.A. Somorjai, Appl. Surf. Sci. 2 (1979) 352. [12] J. Werckmann, A. Mosser and J.P. Deville, Analysis 9 (1981) 213. A. Siokou et aL / Surface Science 352-354 (1996) 628-633 [13] M.G. Barthds and A. Rolland, Thin Solid Films 76 (1981) 45. [14] M. Azizan, T.A. Nguyen Tan and J.Y. Veuillen, Vacuum 41 (1990) 1132. [15] H.B. Michalson, J. Appl. Phys. 48 (1977) 4729. [16] N.D. Lang and W. Kohn, Phys. Rev. B 3 (1971) 1215. [17] H. Balaska, R.C. Cinti, T.T.A. Nguyen and J. Derrien, Surf. Sci. 168 (1986) 225. 633 [18] M. Azizan, T.A. Nguyen Tan, R.C. Cinti, R. Baptist and G. Chanmet, Surf. Sci. 178 (1986) 17. [19] S. Kermou, T.A. Nguyen Tan and J.Y. Veuillen, in: Proc. 4th Int. Conf. Formation of Semiconductor Interfaces (World Scientific, Singapore, 1993) p. 621. [20] N.A. Nicolet, VLSI Electronics (1983), Ed. N.G. Einspurch (Academic Press, New York, 1983) ch. 6.