Electronic structure of erbium suicide ultra-thin films

.,.:r.~.:.:;:~~~~~~:~,~~~,~:;~:~:: i”,~*...,:,:,

A.. ..:.:.:.I:.: x :: : :z...:.:.:. :‘:‘...:.:.s.: 1,...,./,., $:::$...

L

. . . . . . . . .. .

.,.,.,

,,.,.,

~.,_, _,,,,

‘&rface science i-..>A .. ., _..:..:.:::::i ‘““:+.‘~‘~ ‘.‘.‘.“:‘.?.‘I..,._ ...,.,.,,.i,.,.,,,j, ,;;.:,:,:,.;,.,.,:,:: :

Surface Science 293 (1993) 86-92 North-Holland

Electronic

:?,.L.z:i

. .. . :.:.:.:.: . . ..._:,~~ .. .. .. . . . . . .. . .

” .(, .,:::.:‘,~‘,.,I’~ ,...... ......‘.......” :_:,. :.:::::::::::::j:::.:.:.i*:; ,..~.~.:.,,,,_~.~.~.~~,~,,,~ ‘L’,‘. ..............,.:,c .,., __ ,,,,., ,;;,~,,,.,.~.-_.....‘.“. ”‘...‘. ‘.“5:“):.:,:~.::..~.~~~..:,: _,.,..

structure of erbium silicide ultra-thin

films

J.-Y. Veuillen *, T.A. Nguyen Tan and D.B.B. Lollman Luboratoire

d’E&tde des Propri&k

Electroniques

des Solides * * (LEPESJ,

CNRS, BP 166, 38042 Grenobie Cedex 9, France

Received 9 March 1993; accepted for publication 29 April 1993

The electronic structure of erbium silicide ultra-thin films epitaxially grown on Si(ll1) has been studied by means of angle-resolved ultra-violet photoemission spectroscopy (ARUPS) in the coverage range 0.2-4 monolayers (ML). Some peaks probably related to the silicide surface atomic structure are observed at any coverage. Features that appear at normal emission in connection with an R3 superstructure are ascribed to vacancy-induced states in the silicon surface plane. No true interface states could be identified in this study.

1. Introduction Rare-earth (Y, Er, Tb,. . .> silicides can be grown epitaxially on Si(3 31) [l-3]. They form with n-type silicon a low Schottky barrier height (SBH) N 0.3 eV 141. This value is signifi~ntly smaller than that of other transition-metal silitides/silicon interfaces (typically 0.6-0.8 eV). This suggests that the electronic structure of the rare-earth silicide/ silicon interface is peculiar, and that specific interface states may pin the Fermi level in the silicon gap. A calculation of the ideal YSi,/Si(lll) interface has been published, which indicates that such interface states may exist [5]. It has been shown, however, that the electronic structure computed for the disilitide YSi, is significantly different from that computed for the actual composition YSi,., [61. We shall thus not try to relate our experimental results to the theoretical ones of ref. [5]. The aim of the work presented here is to identify true interface states at the erbium silicide (ErSi,)/silicon interface. We follow the usual surface science approach, which consists of depositing silicide layers of increasing thickness 6 on a Si(ll1) sub* To whom correspondence should be addressed. * * Laboratoire associe a l’Universit& Joseph Fourier, Grenoble. 0039-6028/93/$06.00

strate and recording for each film angle-resolved ultra-violet photoemission (ARUPS) spectra in the two inequivalent directions l?MI and KM of the surface Brillouin zone (SBZ) [7]. Since we are looking for states that are specific of the interface region, we have used the following criterion for their identification [S]: the amplitude of a true interface state should increase with coverage up to 8 = 1 ML and decrease in the same way as the substrate contribution for 0 r 1 ML. (The energy of this state should not vary with f?.) Moreover, we are mainly interested in states located close to the Fermi level in the silicon gap. We will thus concentrate on the first few eV below the Fermi level (Er) in the ARUPS spectra (typically in the O-2 eV energy range). The paper is organized as follows: in the first part, we summarize the results published in the literature that are relevant for the present work. In the second part, we describe the experimental procedure. In the third and fourth part, we present and discuss the experimental results. Part of this study has already been published 191. In our previous paper, we have reported results about the SBH formation, the growth mode and the electronic structure for 0 I 1 ML. In this work, we focus on ARUPS results in the whole coverage range.

0 1993 - Elsevier Science Publishers B.V. All rights reserved

J.-Y. Veuillen et al. / Electronic structure of erbium silicide ultra-thin jilms

2. Background The electronic and crystallographic structure of rare-earth silicides (TRSi,, with x = 1.7 and TRY, Er, Tb, Gd, . . . > epitaxially grown on Si(ll1) have been extensively studied in the last few years [7,10-141. These silicides have a common crystallographic structure (defective AlB, type> [l]. It consists of alternate (hexagonal) Er and Si planes that are parallel to the sample surface (or interface). One out of six Si atoms is missing in the silicon planes relative to the ideal (stoichiometric) AlB, (TRSi,) structure, leading to the composition TR : 1, Si : 1.7 [1,3]. Vacancy ordering is thought to take place in the bulk material. Various superstructures have been observed in thick films by electron diffraction [l-3,1.5]. From LEED and X-ray photoelectron diffraction experiments, the atomic structure of ErSi, [16,17] and YSi, [18] surfaces seems to be identical for epitaxial films. They are terminated by a buckled Si plane, the geometric structure of which is quite similar to that of the bilayer of an ideal Si(ll1) surface [18]. The 6 X fiR30” (R3) diagram generally observed by LEED on thick films has been attributed to an ordering of vacancies in this Si surface plane [18]. (These vacanies should be located on the outermost plane with respect to the bilayer of the ideal Si(ll1) surface.) The electronic structure of this erbium silicide surface has been investigated by means of ARUPS [7]. Sharp and intense structures have been observed in the O-2 eV binding energy (BE) range. They have been ascribed to surface states. Some results on the electronic and crystallographic structure of erbium silicide ultra-thin films have already been published [9,16,19]. For 1 ML of Er annealed at 400-6OO”C, a (1 x 1) structure is observed [16,19]. It corresponds to a twodimensional phase (hereafter named (1 x 1) structure or 2D silicide [16]). This 2D silicide has a geometric structure similar to that of the bulk silicide surface. It is terminated by a buckled Si layer, as is the ideal SK1111 surface. The (1 x 1) LEED pattern suggests that Si vacancies do not exist in this plane [16]. The electronic structure of this phase has been studied. ARUPS spectra also show very sharp structures [19]. Similar results

87

have been reported [9] at various coverages in the submonolayer range. In our previous paper [91, we have shown that the (1 X 1) structure transforms to an R3 superstructure at about 0.7 ML (the difference between this value and that given in ref. [19] is not critical). From XPS experiments, it was concluded that the SBH at the ErSi, interface is close to the value found for a buried interface for coverages larger than 1 ML of silitide.

3. Experiment The experimental procedure has been described in ref. [9]. Er ultra-thin films were deposited under UHV conditions (pressure in the lo- lo mbar range) on a clean Si(111)7 X 7 substrate held at room temperature. This deposit was annealed at 600°C for 5 to 10 min to promote silicide formation. The cleanliness of the reacted layer was checked by means of ARUPS (He11 line) prior to the ARUPS study. The thickness of the film (0) is given in units of silicide coverage: 1 ML corresponds to 1 Er(0001) plane in the silitide, this is about 4 A. 19was varied in the range 0.2-4 ML.

4. Results The ARUPS spectra (He1 radiation) for the critical points of the (1 X 11 and R3 SBZ are shown in figs. 1 and 2 for various coverages: 0 I 0.7 ML ((1 X 1) structure or 2D silicide), 8 = 1 ML, 8 = 2-3 ML and for thick films (R3 structure for 8 > 0.7 ML). The corresponding dispersion curves energy versus K,, are shown in figs. 4 and 5. All the structures that show up in the He I spectra have been reported in figs. 4 and 5, which does not mean a priori that they have a two-dimensional character for thick films. One observes that the shape of the UPS spectra and of the dispersion curves are quite different for the (1 x 1) (0 5 0.7 ML) and for the R3 (0 2 1 ML) structure. However, except at the r point (fig. 31, the spectra and the dispersion curves do not evolve much with increasing coverage for 0 2 1

88

J.-Y. Veuillen et al. / Electronic structure of erbium silicide ultra-thin jZms

ML, and are similar to those recorded on thick layers. This is clear for the O-2 eV BE range. (Some differences are observed at about 3 eV BE for the K’ and M points (IIMI, but we believe that they are due to changes in the intensity of the structures rather than in their BE.) This behavior may be related either to the occurrence of a three-dimensional (3D) growth or to the fact that the spectra are dominated by structures that originate mainly from the surface of the film. The latter hypothesis is consistent with our previous findings that the most intense structures that show up in the O-2 eV BE range in the ARUPS spectra of thick films are due to surface states [71. In the former case, sufficiently thick 3D islands should be formed (even for 1 ML nominal coverage) so that their electronic structure closely resembles that of thick films. This would require 3D islands several ML thick, thus leaving a large fraction of the substrate uncovered, at least for 0 I 2-3 ML. However, LEED and Auger observations do not give any evidence for a strong

M-bQl?4L0.’ c

,m.k

-

4 3 2 1 fJ BINDING ENE.RGY(eV) Fig. 2. ARUPS M’ (a), K (b) direction of the structure; 02 1

4

3

2

1

0

BINDING ENERGY(eV) Fig. 1. ARUPS spectra (He1 radiation) recorded close to the K’ (a) and M (b) critical points along the TMT direction of the surface Brillouin zone (SBZ) for various coverages: 0 = 0.7 ML, (1 X 1) structure; 0 2 1 ML, R3 superstructure. The spectra for thick films are labelled “bulk”.

spectra (He1 radiation) recorded close to the and M (cl critical points along the TKM SBZ for various coverages: 0 = 0.7 ML, (1 X 1) ML, R3 superstructure. The spectra for thick films are labelled “bulk”.

islanding above 1 ML [9]. Moreover, the ARUPS spectra recorded at the I point (fig. 3) change with 0 above 1 ML up to at least 4 ML, showing that the electronic structure of the film in this coverage range is still different from that of thick films. We thus conclude that the (relative) invariance of the spectra in the energy range of interest above 1 ML is due to the large weight of surface structures in the spectra. (It additionally shows, as quoted in ref. 191, that the surface structure does not change much above 1 ML.) From the point of view of the present paper this is quite unfortunate, since we can hardly find a structure satisfying the criterion for being a true interface state in the Si gap, close to the Fermi level. As one can see from figs. 1, 2, 4 and 5, the structures

J.-Y. Veuillen et al. / Electronic structure of erbium silicide ultra-thin film

observed at the M’, K’, M and K points in the O-2 eV BE range for 8 I 0.7 ML either vanish when the R3 superstructure appears, or survive at any larger coverage. (The sharp peak located at 1.6 eV BE at the M point for the 2D silicide is strongly attenuated for 0 2 1 ML. However, a structure remains visible at about the same energy for thick films: see figs. 2c and 3c of ref. [7].) At the I point, the evolution of the spectra with the silicide thickness seems to be more or less continuous. The sharp peak located around 0.8 eV appears at about 0.85-1.0 ML, in connection with the R3 LEED pattern. For larger coverages, it shifts slightly towards larger BE and finally splits in two peaks for 8 = 4 ML. It corresponds to the broad peak centred at 1 eV for thick films. Structures in this energy range at the l? point seem to be characteristic of the R3 structure, since they are present only when this superstructure is observed in LEED and in ARUPS for any value of 13. Since they do not show up at submonolayer coverage (fig. 3 and ref.

K’

0

89

M

K’

Q!5 I.0 O11 K// (A-

Fig. 4. Dispersion curves (binding energy versus parallel wavevector K,,) along TMF in the SBZ for: (a) 0 = 0.7 ML; (b) 0 = 1.0 ML and (c) thick film. The original ARUPS spectra were recorded using He1 radiation. ((0) and (+) denote strong and weak structures, respectively).

BINDING ENERGY(eV) Fig. 3. ARUPS spectra (He I radiation) at normal emission (P point of the SBZ) for various coverages: 13< 0.7 ML, (1 x 1) structure; 13> 0.7 ML, R3 superstructure. For 0 = 0.7 ML, faint spots of the R3 superstructure are detected in LEED 191, although the ARUPS spectra show essentially a (1 X 1) symmetry (see fig. 4a).

[19]) they cannot be considered as true interface states using the criterion of section 1. In summary, we have not been able to identify unambiguously true interface states in this study of the ErSi,/Si(lll) interface. One can see from figs. 2 and 3 that the sharp peaks located just below E, at the M point and at 1.4-1.5 eV at the K point are observed for both the 2D silicide (see also ref. [19]) and for thick films, as well as for any intermediate coverage [21]. We know from refs. [193 and [7] that these structures have a two-dimensional behavior for the 2D and bulk silicide films. Now, it has been shown that the surface atomic structure of these two phases are somewhat similar (an Er plane sandwiched between two Si planes, where the Si plane on the vacuum side has a structure similar to that of a Si(ll1) bilayer of bulk Si, see section 2). The observation of peaks at about the same binding energy at the M and K points for

90

J.-Y. Veuillen et al. / Electronic

structure of erbium silicide ultra-thin films

1ML

point for thick films. The creation of vacancies in the Si surface plane generates many additional dangling bonds (DB) relative to the (1 X 1) structure: each DB of the initial (1 X 1) structure has 3 DB located on nearest neighbor Si atoms in the R3 structure. These DB are expected to interact, thus modifying the surface electronic structure. Irrespective of their origin, substantial differences are indeed observed in the dispersion curves between the (1 X 1) and the R3 superstructure (see figs. 4 and 51, but, as mentioned above, some features remain at about the same energy at the K and M critical points.

c

5. Discussion i * 7’. t

0

++,/t++

li

*

t ,

I

I

++*a+

/

;++

,

bulk

Q5 K/ (kf

‘15

Fig. 5. Dispersion curves (binding energy versus parallel wavevector K,,) along lXM in the SBZ for: (a) 0 = 0.7 ML, (b) B = 1.0 ML and fc) thick film. The original ARUPS spectra were recorded using He1 radiation. C(O) and (i-1 denote strong and weak structures, respectively.1

the 2D and bulk silicide suggests that they are typical of this peculiar geometry. Of course, the atomic structure of the bulk silicide surface is not really identical to that of the 2D silicide since Si vacancies presumably give rise to the R3 superstructure [lg]. As can be seen from fig. 5 the dispersion of the state located at 1.4-1.5 eV at the K point is actuahy different for the (1 X 1) and R3 structures - as expected - and its BE is larger (by 0.1-0.15 eV> in the (1 x 1) structure than in the R3 structure. Anyway, a priori, it looks surprising that some surface states remain at about the same BE at the critical points K and M whatever vacancies are present or not in the Si surface plane. For the 2D silicide, there are indications that the band which disperses from E, close to F to 1.5 eV at the K point derives from the dangling bonds of the surface Si layer [19]. If our assumption is correct, this should also be true for the state located at about 1.4 eV at the K

As shown in section 4, no true interface state could be identified in the silicon gap (within 2 eV of the Fermi level) in this study of the silicide/ silicon interface. One can find several reasons for this. Such states may not exist: in the case where interface states are derived from bulk metal states tunnehing in the silicon gap (as in the MIGS model [22]), our criterion does not allow one to identify them. Alternatively, they may not be detectable due to an overlap with some surface related states. As discussed previously, some of the most intense features of the ARUPS spectra (at about 1.5 eV at the K point and close to E, at the M point) may be characteristic of the geometry of the 2D or of the bulk silicide surface. One may think of building the ErSi,/Si(lll) interface by simply connecting a semi-infinite Si crystal (with an ideal Si(ll1) termination) to the (Si terminated) silicide surface, in the way suggested in ref. 157. This will result in a direct bonding between the Si atoms in the outermost planes of both the silicide and silicon surfaces, thus removing the dangling bonds (normal to the interface) on these atoms. It may happen, however, that some states characteristic of the silicide surface but not related to dangling bonds will be preserved after interface formation. This might be the case for states derived from Er-Si bonding between the surface Si plane and the subsurface Er plane. Since these two planes become the interface planes after connection to the semi-in-

J.-Y. Veuillen et al. / Electronic structure of erbium silicide ultra-thin films

finite Si(ll1) crystal (i.e. the former Si terminal plane of the silicide film may also be regarded as the last plane on the semiconductor side after interface formation due to their similar geometric structure), some “surface states” may become “interface states”. This is merely a conjecture, however, since it requires for instance that the electronic structure at the surface is not much affected by interface formation, except for the DB states. We now turn to the ARUPS spectra collected at normal emission (I point). Significant changes are observed with increasing coverage, up to at least 4 ML. Some structures, however, remain at fixed binding energies for 0 2 1 ML. They are located at 0.4, 1.4, 1.8 and 2.2 eV. The last two ones have been attributed to surface states [7,13]. As for the other surface states observed at the M’, K’, M and K points, their position does not depend much on 8. The second one results from the backfolding of the sharp peak recorded at 1.4 eV at the K point (the I and K points should be equivalent in the R3 SBZ, see e.g. ref. [7]). This is clearly seen from the dispersion curve along lKM in fig. 5b. Below about 2 eV, one observes broad structures that seem to shift as a function of 8. Their origin is not clear [9]. They are located far outside the silicon gap at the I point on the Si(ll1) surface [20]. As mentioned in section 4, the existence of peaks located at about 0.8-1.0 eV for 8 2 1 ML is strongly correlated with the observation of the R3 superstructure with LEED (they appear at the same coverage). If one admits that the R3 originates from an ordering of Si vacancies, then one can ascribe the peaks found at about 1 eV BE to vacancy-induced states. A recent calculation proposes that vacancy-induced states are located at about l-l.5 eV BE along the IA direction (the surface normal) for the bulk silicide [13], and the present results seem to corroborate these findings. One should consider this agreement with some care, however, for several reasons: (i> the ARUPS technique is highly surface sensitive, and it is not certain that the vacancy-induced states are at the same energy in the surface plane and in the bulk planes, due to buckling and/or asymmetric bonding (note, however, that the structures at about 1 eV BE for

91

thick films are not very sensitive to oxygen adsorption [7,131; this suggests that they are not pure surface features). (ii) There are some discrepancies between different calculations of the bulk silicide electronic structure. One paper does not give any indication of clearly identified vacancy-induced states [6], whereas another one indicates that they are located between 2 and 4 eV BE [23].

6. Conclusion We have performed a study of erbium silicide ultra-thin films epitaxially grown on Si(ll1). We have found indications that some peaks in the ARUPS spectra close to the K and M points are specific of the peculiar geometry of the silicide surface. Structures observed at 1 eV BE at normal emission are probably related to vacancy-induced states (at least in the surface Si plane). We have not identified any true interface states.

References

Knapp and ST. Picraux, Appt. Phys. Lett. 48 (1986) 466. 121F.H. Kaatz, J. Van der Spiegel and W.R. Graham, J. Appl. Phys. 69 (1991) 514. 131 F. Arnaud d’Avitaya, A. Perio, J.C. Oberlin, Y. Campidelli and J.A. Chroboczek, Appl. Phys. Lett. 54 (1989) 2198. J.A. 141 F. Arnaud d’Avitaya, P.A. Badoz, Y. Campidelli, Chroboczek, J.-Y. Duboz and A. Perio, Thin Solid Films 184 (1990) 283. 151 H. Fujitani and S. Asano, Appl. Surf. Sci. 56-58 (1992) 408. J. Phys. 161L. Magaud, J.-P. Julien and F. Cyrot-Lackmann, Condensed Matter 4 (1992) 5399. [7] J.-Y. Veuillen, L. Magaud, D.B.B. Lollman and T.A. Nguyen Tan, Surf. Sci. 269/270 (1992) 964. [8] J.E. Houston, C.H.F. Peden, P.J. Feibelman and D.R. Hamann, Phys. Rev. Lett. 56 (1986) 375. 191 J.-Y. Veuillen, D.B.B. Lollman, T.A. Nguyen Tan and L. Magaud, Appl. Surf. Sci. 65/66 (1993) 712. [lo] J.-Y. Veuillen, S. Kennou and T.A. Nguyen Tan, Solid State Commun. 79 (1991) 795. [ill W.A. Henle, M.G. Ramsey, F.P. Netzer, R. Cimino and W. Braun, Solid State Commun. 71 (1989) 657. 1121 0. Sakho, F. Sirotti, M. De Santis, M. Sacchi and G. Rossi, Appl. Surf. Sci. 56-58 (1992) 568. [ll J.A.

92

J.-Y Veuillen et al. / Electronic structure of erbium silicide ultra-thin films

[13] L. Stauffer, C. Pirri, P. Wetzel, A. Mharchi, P. Paki, D.

[14]

[15] [16]

[17] [18]

Belmont, G. Gewinner and C. Minot, Phys. Rev. B 46 (1992) 13201. L. Magaud, J.-Y. Veuillen, D.B.B. Lollman, T.A. Nguyen Tan, D.A. Papaconstantopoulos and M.J. Mehl, Phys. Rev. B 46 (1992) 1299. T.L. Lee, L.J. Chen and F.R. Chen, J. Appl. Phys. 71 (1992) 3307. P. Paki, U. Kafader, P. Wetzel, C. Pirri, J.C. Peruchetti, D. Bolmont and G. Gewinner, Phys. Rev. B 45 (1992) 8490. D.B.B. Lollman, Thesis, Universite J. Fourier, Grenoble, 1992, unpublished. R. Baptist, S. Ferrer, G. Grenet and H.C. Poon, Phys. Rev. Lett. 64 (1990) 311.

1191 P. Wetzel, C. Pirri, P. Paki, J.C. Peruchetti, D. Bolmont and G. Gewinner, Solid State Commun. 82 (1992) 235. [20] See, e.g., RIG. Uhrberg and G.V. Hansson, Crit. Rev. Solid State Mater. Sci. 17 (1991) 133. [21] This seems to be true also for the broad structures located at 2.0-2.5 eV at the K point and at about 2.7-3.0 eV at the M point. [22] See, e.g., F. Flores and J. Ortega, Appl. Surf. Sci. 56-58 (1992) 301; J. Tersoff, Phys. Rev. Lett. 52 (1984) 465. [23] G. Allan, I. Lefebvre and N.G. Christensen, private communication.