Medium energy ion scattering and STM studies on CuSi(111)

Nuclear Instruments and

J?Jg

Methods in Physics Research B 99 (1995) 495-498

NOMB

Beam Interactions with Materials 8 Atoms

ELSEVIER

Medium energy ion scattering and STM studies on Cu/Si( 111) T. Koshikawa

“3* , T. Yasue a, H. Tanaka b, 1. Sumita b, Y. Kido ’

A Dept. of Applied Electronics

and Fundamental Electronics Research Institute, Osaka Electra-Communication Osaka 572, Japan ’ Matsushita Research Institute of Tokyo, Higashimita, Tama, Kawasaki 214, Japan ’ Dept. of Physics, Ritsumeikan Unirl., Noji-cho, Kusatsu. Shiga 525, Japan

UG.,

Neyagawa,

Abstract The structure of Cu on Si(lll) 7 X 7 deposited at high temperature (300-600” C) was examined by the results of medium energy ion scattering (MEIS) and scanning tunneling microscopy (STM). The structure of the Si bulk that is just under the incommensurate layer, which shows the “5 X 5” electron diffraction pattern, might be the double layer and the first layer of the Si bulk is deduced to be relaxed inward by 0.01 nm after the measurements and the Monte Carlo simulation of the blocking profiles of MEIS. The distance (0.05 nm) behveen the Cu and Si layer in the incommensurate layer was also estimated using the ultra high depth resolution capability (0.16 nm for Si) of MEIS. The Si structure in the incommensurate layer was deduced by the ratio of dark to bright areas of “5 X 5” after counting the number of Si atoms. The ratio was about 0.83 and there is about 1 ML of Si atoms in the incommensurate layer.

1. Introduction The structural and morphological properties of siliconmetal interfaces have been the subject of wide investigation, with particular attention devoted to the noble metalsilicon interface. A special effort for copper-silicon at high formation temperatures has been devoted recently. At high temperatures (130-600” C>, the quasi-5 X 5 layer formed by annealing of about one monolayer (ML) of Cu on Si( 111) has the so-called incommensurate structure that has been discussed by numerous reports, including helium diffraction [l], LEED [2,3], angle-resolved AES [4], a reflection electron microscope (REM) and transmission electron microscope (TEM) [5], STM [6-91, a low energy electron microscope (LEEM) [lo], an X-ray standing wave [ll], an angle-resolved UPS and a band calculation [12], etc. However, the models of the incommensurate structure that were so far reported have many varieties and have not revealed. In the present article, the structure of Si at the bulk, which has no incommensurate layer, was investigated by the blocking profiles of MEIS and the Monte Carlo simulation of the ion trajectories. The atomic layer position of Cu is about 0.05 nm higher than the Si layer in the incommensurate layer, which was estimated by the ultra high depth resolution (- 0.16 nm for Si bulk) measurement of MEIS. The number of Si atoms in the incom-

* Corresponding author. Tel. +81 720 24 1131,

fax +81

720

25 4590. 0168-583X/95/$09.50

0 1995 Elsevier Science

S.SDI 0168-583X(94)00755-1

B.V. All

rights

mensurate layer was investigated by the paired dark and bright area of the “5 X 5” regions with STM images at less than 300°C. There exists about 1 ML Si layer in the incommensurate layer. These results indicate that the missing top layer in the incommensurate layer is on the double layer of Si that is the same as the bulk.

2. Experiments The measurements with MEIS including AES and RHEED, and STM were carried out in the different chambers. The details of the MEIS equipment were described elsewhere [ 131 and here we mention the main point briefly. Primary ions were He+ and the incident energy was 175 keV. Two primary angles were selected in order to check the blocking profiles carefully. The first angle was 35.3”, which is along the [ii01 axis from the surface normal and the ion beam can see the first and second layers if the structure of the Si has the double or missing top layer and there is no displacement. In the second one of 60.5“ along the [%2] axis, the ion beam can see up to the fourth layer. The scattered ions were detected using a toroidal electrostatic analyzer over the scattering angle of 23” and the energy resolution (AE/E) is about 4 X 1O-3 [13]. The base pressure of the chamber was 5 X low9 Pa. The working pressure during measurements was about 2 X 10-s Pa. The typical depth resolution is 0.3-0.4 nm for Cu substrate, as estimated by the semi-empirical formula of Ziegler [14]. The measurement of ultra high depth resolureserved VIII. MATERIALS

ANALYSIS

T. Koshikawa

496

et al. /Nucl.

Instr. and Meth. in Phys. Res. B 99 (1995) 495-498

tion (- 0.16 nm for Si) was performed by the grazing emission angle (3-5” from the surface). STM images were taken with a field ion-scanning tunneling microscope (FI-STM) [15] with a constant current (50 PA) mode. The base pressure was 5.5 X lop9 Pa and the pressure during measurements was better than 1.3 X lo-* Pa. The samples used in the measurements with MEIS and STM were the mirror-polished B-doped p-type (l-100 fl cm) and P-doped n-type Si(ll1) (l-10 n cm), respectively. The samples were annealed at about 1200” C by direct resistive heating under a pressure below 6.7 X lo-’ Pa and then cooled down slowly to room temperature. Cu of 99.9999% was evaporated from Cu beads on hot tungsten wires. The deposition rate was about 1 ML/min (1 for MEIS and about ML = 7.83 X 1014 atoms/cm’) f ML/min for STM. The evaporation rate was monitored by the quartz oscillator.

Double layer A’: =-O.Olnm

A;

=O.Onm

I

1

(a) 65

70

75

80

65

Exit Angle (deg) Missing top layer A: =-O.Olnm

A”: =O.Onm

(b)

3. Results and discussion

-5

65

The structure at the top of the Si bulk was estimated by the blocking profiles of MEIS experiments and the simulation with the Monte Carlo method. The two models are assumed as shown in Fig. 1. Fig. la shows the double layer and Fig. lb the missing top layer. These two are the main models proposed so far. Each layer can be displaced in the simulation in order to fit the experiments. Fig. 2 shows the blocking profiles of the experiment (circles) and the simulation (solid lines: (a) for the double layer and (b) for the missing top layer) for the primary angle of 35.3”. Assuming the displacement (0.01 nm) of the first layer, the

Double layer

(al

h

h

h

Mtssing top layer (b)

Fig. l.(a) The model of the double layer of the Si bulk. (b) The missing top layer model. iach layer can be displaced in the

Exit Angle (deg)

Fig. 2.(a) Blocking profiles of the experiment (circles) and the simulation (solid line) results from Cu/Si(lll) “5 X 5”. The Primary angle is 35.3” from the surface normal. The simulation profile is based on the double layer model (see Fig. la) and inward displacement of the first Si layer by 0.01 nm. (b) The simulation profile (solid line) is based on the missing top layer model (see Fig. lb). The first layer of Si is also relaxed by 0.01 nm inward.

simulation profiles follow the experiments for the main dips of [?i3], [ii41 and [iiS], however, the medium size dip at around 78” of the simulation based on the missing top layer model is not clearly seen in the experiments. In order to know more details, the blocking profiles from each Si layer are simulated as shown in Fig. 3. The blocking dip at around 77” of the missing top layer is caused by the scattering from the fourth and fifth layers, however the blocking dip from the fifth layer of the double layer is canceled by the peak from the third and fourth layers. For better reliability, the case of the primary angle of 60.5” was also investigated with experiments and simulations. The dip at around 77” for the missing top layer that was not observed in the experiment was emphasized due to the higher encounter probability of the primary ions to the fourth layers. The double layer model is plausible from the above results. The detailed results for the position between Cu and Si layers in the incommensurate layer were estimated under the ultra high depth resolution using the grazing emission angles (3-5” from the surface). The typical depth resolution for an emission angle of 4.4” and the primary angle of 70” (random condition) from the surface normal is 0.16

T. Koshikawrz et al. /Nucl.

65

70 Exit

75 Angle

80

65

85

(deg)

70

Exit

75 Angle

Ins&. and Meth. in Phys. Res. B 99 (1995) 495-498

80

85

(deg)

Fig. 3. The blocking profiles from each layer are simulated for the double (left) and the missing top (right) layer. The remarkable blocking dip at around 77” from the fourth and fifth layers exists in the missing top layer model, however, the dip from the fifth layer is cancelled with the peak from the third and fourth layers in the double layer model. nm and 0.11 nm for Si and Cu substrates. The energy peaks from each double layer of Si were separated as in Fig. 4 and the first Cu and Si peak positions were compared with no inelastic energy loss peak positions that are shown as arrows in the figure. The peak position of the first Si layer in the “5 X 5” layer shifted to the low energy side due to the inelastic energy loss, which predicts that the Cu layer is on the Si layer and the distance between them is about 0.05 nm. The value was estimated using the semi-emperical formula of the stopping power by Ziegler [ 141.

KZ” Exit anr~k4.4~

I

A

491

The parts of the reconstructed “5 X 5” region after Cu evaporation on the Si(ll1) kept at 400” C have random distinct protrusions and holes at the sample bias under 2 V. These protrusions and holes have no clear relation with the “5 x 5” periodicity [7]. However, when the same part was displayed at the bias of 2.5 V. the “5 X 5” part shows clear regular domain protrusions whose periodicity is 5.5 + 0.2 times that of the silicon bulk, which is similar to the results of Mortensen [9]. However, the images at the bias of 0.22 V were observed by him. We could observe the regular domains at the low bias such as 0.22 V after several times scanning on the surfaces. This bias dependence predicts that the tunneling images are strongly influenced by surface electronic conditions and the top of the tip condition. The spectra of an inverse photo emission of the “5 X 5” surface were published by Nicholls et al. [ 161. The results indicate that a strong surface state at around 2.2 V above the Fermi level is present for a wide interval of emission angles. This surface state might cause the clear 5.5 times periodicity of the STM images at V, = 2.5 V. Takayanagi et al. [5], tried to propose a model of Si surface after the estimation of the ratio of the “5 X 5” area on the upper and lower terraces at the single steps. They made their observations with a reflection electron microscope (REM). However, Bauer’s group obtained the different LEEM images in which there were some “5 X 5” regions on the center of terraces [lo]. We also observed that the “5 X 5” region at the double step expanded to only the upper or lower terrace. These observations predict that the growth process of the “5 X 5” region is more complicated and might have a temperature dependence. We tried to get the image at a lower temperature (less than 300” Cl in order to estimate the number of the Si atoms in the incommensurate layer. The paired “5 X 5” regions were seen on the same terrace [17]. The ratio of dark and bright areas is estimated to be 0.83 after averaging the data. Here we assume that the structure at the top of the Si bulk has the double layer. which was predicted by the results of MEIS. We can estimate the coverage 0; of Si in the incommensurate layer is about 1 ML.

4. Conclusion

I

I

I

I

ENERGY

1

I

170

166

174 keV)

Fig. 4. The energy spectrum from Cu/Si(Xll) that is measured in the ultra high resolution condition. Primary angle is 70” (random condition) from the surface normal and the emission angle from the surface is 4.4”. The depth resolution is 0.16 nm for Si bulk. The each double layer of Si is separated. The spacing between Cu and Si layer in the incommensurate layer is estimated by the position of the first peak of Si and the Cu peak (see text).

The structure of Cu on Sit1111 at high temperature was proposed by the results of scanning tunneling microscopy @TM) and medium energy ion scattering (MEIS). The structure of the Si bulk layer was proposed by the blocking profiles of the MEIS. The double layer structure is plausible and the top layer of the bulk Si displaces to the bulk by 0.01 nm. The position of Cu and Si layer in the incommensurate layer was discussed under the ultra high depth resolution measurements. The Si layer is 0.05 nm under the Cu layer. The STM image at the high sample bias voltage shows the very clear domains with 5.5 times periodicity that coincide to the “5 X 5” incommensurate

VIII. MATERIALS

ANALYSIS

498

T. Koshikawa et al. / Nucl. Ins@. and Meth. in Phys. Res. B 99 (1995) 495-498

structure. The coverage (1 ML) of Si in the incommensurate layer was estimated by the area ratio of the paired dark and bright “5 X 5” regions.

Acknowledgements We acknowledge Prof. E. Bauer of TLJ Clausthal for showing us the unpublished LEEM movies and for his invaluable discussions.

References [l] R.B. Doak and D.B. Nguyen, Phys. Rev. B 40 (1989) 1495. [2] E. Duagy, P. Mathiez, F. Salvan and J.M. Layet, Surf. Sci. 154 (1985) 267. [3] H. Kemman, F. Muller and H. Neddermeyer, Surf. Sci. 192 (1987) 11. [4] S.A. Chambers, S.B. Anderson and J.H. Weaver, Phys. Rev. B 32 (1985) 581. [5] K. Takayanagi, Y. Tanishiro, T. lshizuka and K. Akiyama, Appl. Surf. Sci. 41/42 (1989) 337.

[d R.J. Wilson, S. Ching and F. Salvan, Phys. Rev. B 38 (1988) 12696. [71 St. Tosch and H. Neddermeyer, Surf. Sci. 211/212 (1989) 133. 181J.E. Demuth, U.K. Korler, R.J. Hamers and P. Kaplan, Phys. Rev. Lett. 62 (1989) 641. [91 K. Mortensen, Phys. Rev. L&t. 66 (1991) 461. DO1 M. Mundschau, E. Bauer, W. Telieps and W. Swiech, J. Appl. Phys. 65 (1989) 4747. [ill J. Zegenhagen, E. Fontes, F. Grey and J.R. Patel, Phys. Rev. B 46 (1992) 1860. 1121 D.D. Chamblis and T.N. Rhodin, Phys. Rev. B 42 (1990) 1674. R. Kikuchi, K. Takagi, T. Uchiyama, Y. [131 T. Koshikawa, Mihama, Y. Agawa, S. Matsuura, E. Inuzuka and T. Suzuki, Nucl. Instr. and Meth. B 33 (1988) 623. [141 J.F. Ziegler, He Stopping Powers and Ranges in All Elements (Pergamon, New York, 1977). 1151 T. Sakurai, T. Hashizume, I. Kamiya, Y. Hasegawa, T. Ide, M. Miyao, A. Sasaki and S. Hyodo, J. Vat. Sci. Technol. A 7 (1989) 1684. [I61 J. Nicholls, F. Salvan and B. Reihl, Phys. Rev. B 34 (1986) 2945. [171 T. Koshikawa, T. Yasue, H. Tanaka, I. Sumita and Y. Kido, Surf. Sci., in press.