CoSi2 formation by high current ion implantation with a metal vapor vacuum arc ion source

CoSi2 formation by high current ion implantation with a metal vapor vacuum arc ion source

Nuclear Instruments and Methods in Physics Research B 211 (2003) 358–362 www.elsevier.com/locate/nimb CoSi2 formation by high current ion implantatio...

276KB Sizes 0 Downloads 54 Views

Nuclear Instruments and Methods in Physics Research B 211 (2003) 358–362 www.elsevier.com/locate/nimb

CoSi2 formation by high current ion implantation with a metal vapor vacuum arc ion source Y. He, J.Y. Feng *, W.Z. Li Department of Materials science and Engineering, Key Laboratory of Advanced Materials, Tsinghua University, Beijing 100084, PR China Received 25 October 2002; received in revised form 23 April 2003

Abstract Crystalline CoSi2 (1 0 0) buried layers with low sheet resistivity were synthesized by direct ion implantation into Si(1 0 0) substrates at a low temperature of 400 C using a metal vapor vacuum arc ion source. X-ray diffraction patterns showed that these layers had a strong (1 0 0) preferred orientation. Rutherford backscattering spectrometry and fourpoint probe measurement showed that a high implantation dose is beneficial to forming good electrical properties.  2003 Elsevier B.V. All rights reserved. Keywords: CoSi2 ; Sheet resistivity; Ion implantation

1. Introduction Due to its line-width independence, good thermal stability, low resistivity, easy formation on narrow Si lines and small lattice mismatch with Si substrates [1–3], CoSi2 has become a promising candidate for future generation contact materials used in ultra-large-scale integrated (ULSI) circuits. Over the past decades, various techniques have been employed to synthesize metal silicides, among which, solid-state reaction [4] and ion-beam mixing [5] of metal overlayers on Si wafers have been studied extensively. Typically, solid-state reaction requires a relatively high temperature, which is not suitable for the Si-device technology. *

Corresponding author. Tel.: +86-10-62772617; fax: +86-1062771160. E-mail address: [email protected] (J.Y. Feng).

A new kind of ion source, the metal vapor vacuum arc (MEVVA) ion source was developed by Brown et al. in the 1980s [6]. The MEVVA ion source has a high-current capability which can produce a mean ion current density up to 100 lA/ cm2 , and the current density of Co-FIB (focused ion beam) has been in the range of A/cm2 [7]. This enables the high-dose implantation to be completed in much shorter time compared with implantation by conventional ion implanters. Furthermore, the high current density can produce a notable heating effect [8], which is beneficial for the synthesis of CoSi2 by direct implantation of Co [9]. CoSi2 is of CaF2 type structure, which is similar to that of Si. At room temperature, the lattice constant of CoSi2 is 1.2% smaller than that of Si [10–12]. The Co atoms can possibly be ‘‘added’’ into the Si lattice to convert Si to CoSi2 structure

0168-583X/$ - see front matter  2003 Elsevier B.V. All rights reserved. doi:10.1016/S0168-583X(03)01388-0

Y. He et al. / Nucl. Instr. and Meth. in Phys. Res. B 211 (2003) 358–362

with almost no volume change. This conversion can be completed by Co ion implantation into Si wafers at low temperature, as such silicatization is a dynamic but not a thermal excitation process like in solid-state reaction of the thick cobalt film on Si requiring annealing at 650 C [13]. Zhu et al. have investigated the formation of CoSi2 by MEVVA ion implantation [14]. Experimental results showed that a fraction of the implanted Co formed CoSi2 in the buried layer by X-ray diffraction (XRD) and Rutherford backscattering spectrometry (RBS) measurement. In those experiments, the Si substrates were heated merely by the ion beam during the implantation process. In present experiments, the substrates were heated directly by passing a current through them to keep a constant temperature. In this paper, we report the direct formation of a crystalline CoSi2 layer by MEVVA ion source implantation at 400 C without the post-annealing step. By heating the samples, we can prevent an amorphization of the substrate in the CoSi2 formation progress [9].

2. Experimental procedure In this study, n-type Si(1 0 0) wafers with a resistivity of 2–4 X cm were used as substrates. The Si substrates were cut into 1 · 2.5 cm2 wafers. Before being loaded into the vacuum chamber, the wafers were ultrasonically cleaned in acetone for about 10 min, dipped in diluted HF solution for about 5 min and then rinsed in de-ionized water. The Co implantations were performed with a MEVVA ion source with an extra voltage of 40 kV. The implantation current density was maintained at 80 lA/cm2 . The beam spot size was 100 mm in diameter. The Co ion beam was composed of 47% Coþ , 49% Co2þ , 4% Co3þ [15], and the corresponding ion energy was therefore 40, 80 and 120 keV. A thermocouple was attached to the backs of the substrates to monitor the substrate temperature. The implantations were conducted in a background vacuum better than 3 · 104 Pa. Before implantation, the substrates were heated to 600 C for approximately 2 min to degas and to get rid of surface oxide. During implantation, the average temperature of the substrates enhanced about

359

100 C. Then we decreased the voltage to keep the temperature stability. No post-annealing step was performed. These as-implanted samples have been characterized by XRD and RBS respectively. A D41-5/ZM type four-point probe was used to measure the sheet resistivity.

3. Results and discussion The h–2h XRD patterns of these as-implanted samples are shown in Fig. 1. In all of the diffraction patterns of samples with different implantation doses, the CoSi2 (1 0 0) diffraction peak is observed. It is well known that the CoSi2 phase has not been formed by solid-state reaction at a temperature of 400 C without annealing. The temperature that CoSi2 phase forms is beyond 650 C [13], which is much higher than the substrate temperature. Thus, the CoSi2 phase must have been formed during the implantation process without the post-annealing step. This may be attributed to the pulse mode of the MEVVA ion source. As the peak ion current density, it can cause very high instantaneous and local temperature, which is beneficial to the nucleation of CoSi2 [8]. In these XRD patterns, only CoSi2 (1 0 0) and Si(1 0 0) diffraction peaks can be seen. Other diffraction peaks of CoSi2 , such as CoSi2 (1 1 1) at 2h ¼ 28:8 and (2 2 0) at 2h ¼ 47:9 were not observed in XRD patterns. This is due to the strong (1 0 0) preferred orientation of the CoSi2 formed by direct Co implantation into Si(1 0 0). It can be seen that the intensity of the CoSi2 (1 0 0) diffraction peak increase obviously with the rise of implantation dose. In Fig. 1(c) and (d), the diffraction intensity of CoSi2 (1 0 0) is almost the same as the Si(1 0 0), indicating the crystallinity of formed CoSi2 layers in these two wafers were better than that in those two wafers implanted with low dose. Fig. 2 shows the random RBS spectrum for Si wafer implanted with high current density confirming the formation of the silicide, as the spectrum shows a clear step just behind the Si leading edge. A 1-mm-diameter collimated beam was used in the RBS analysis. The analyzed thickness of the buried CoSi2 layer was about 52 nm. In the RBS pattern, the low-energy edge of Co peak drops

360

Y. He et al. / Nucl. Instr. and Meth. in Phys. Res. B 211 (2003) 358–362

Fig. 2. Random 2 MeV Heþ RBS spectrum of a CoSi2 layer, implanted with the dose of 8 · 1017 /cm2 at the temperature of 400 C.

slowly, demonstrating that there are also some CoSi2 clusters below the CoSi2 layer [7]. According to the equation of consistence [16], we can get the approximate ratio of the amount.  2 NSi HSi  DESi ZCo ¼  ; ð1Þ NCo HCo  DECo ZSi

Fig. 1. h–2h XRD patterns of the samples implanted at 400 C with a dose of (a) 2 · 1017 /cm2 , (b) 4 · 1017 /cm2 , (c) 6 · 1017 /cm2 , (d) 8 · 1017 /cm2 .

where N is the ratio of the amount, H is the height of signal peaks, DE is the width of energy, Z stands for atomic number. As the RBS pattern shows, HSi ¼ 230:60 and HCo ¼ 308:23, DESi ¼ 8:55 and DECo ¼ 11:72. ZCo and ZSi equal to 27 and 14 respectively. To solve Eq. (1), the ratio of the amount of Si to that of Co was close to 2:1, in agreement with the CoSi2 composition basically. As mentioned above, due to the similarities of both structure and the atomic density in Si and CoSi2 , the conversion of Si and CoSi2 would cause no significant volume change. Such conversion can therefore be realized at relatively low temperature as it requires no thermal activation like that in traditional ion beam synthesis and solid-state reaction [17]. Table 1 lists the sheet resistivity of the samples. Considering the error from the sheet resistivity as well as the thickness measurement, the relative error of the resistivity was assessed to be within 18.9%. The dependence of resistivity versus implantation dose is displayed in Fig. 3, from which one sees that the sheet resistivity decreases with

Y. He et al. / Nucl. Instr. and Meth. in Phys. Res. B 211 (2003) 358–362 Table 1 Resistivity (q) of CoSi2 layers as a function of implantation dose Sample Dose (·1017 /cm2 ) Sheet resistivity (X=) q (lX cm)

1

2

3

4

2 9.6

4 6.8

6 5.6

8 4.9

49.9

35.3

29.1

25.4

e¼E

Fig. 3. The electrical sheet resistivity of CoSi2 versus implantation dose in Co-ion implanted Si(1 0 0) at the temperature of 400 C.

increasing implantation dose, and ranged between 9.6 and 4.9 X= for implantation doses of 2 · 1017 –8 · 1017 Co/cm2 . It is found that the implantation dose plays an important role in the electrical properties of CoSi2 layer. In the experimental process, an increase of dose means also an increase of implantation time and so a longer and more perfect annealing. That is the reason that a high implantation dose is better to synthesize a higher quality CoSi2 layer. On the other hand, the saturation dose is lower than the applied doses, and the sputtering must be considered. According to MatsumanÕs formula [18]: "  1=2 # aðM2 =M1 Þ Eth Y ðEÞ ¼ 0:042 Sn ðEÞ 1  ; U0 E ð2Þ

361

pffiffi 3:44 e logðe þ 2:718Þ pffiffi pffiffi ; Sn ðEÞ ¼ 1 þ 6:355 e þ e½1:708 þ 6:882 e aM2 ; Z1 Z2 e2 ðM1 þ M2 Þ

ð3Þ

ð4Þ

a ¼ 0:4137 þ 0:6092ðM2 =M1 Þ0:1708 ; (  n Z 1 x  Rp þ Sð/Þ 0 pffiffiffi N ð/Þ ¼ erf 2Y 2 DRp 0 ) x  Rp  erf pffiffiffi dx; 2 DRp

ð5Þ

Sð/Þ ¼ ðND =n0 ÞY :

ð7Þ

ð6Þ

Here Y ðEÞ is the ratio of sputtering; M1 ¼ 28, M2 ¼ 59; Z1 ¼ 14, Z2 ¼ 27; E ¼ 40 keV; Eth ¼ n  U0  E (U0 ¼ 2:32 eV-atomic binding energy). From Eq. (2), we can get: ‘‘Y ðEÞ ¼ 0:00152’’. ND ¼ 8  1017 /cm2 (implantation dose), n0 ¼ 5:2  1022 /cm2 (density of silicon); Rp ¼ 52 nm, DRp ¼ 9:1 nm. Doing this, Eq. (6) is N ð/Þ ¼ 2:8  1016 /cm2 which is just 3.5% of implantation dose. Thus, it can be seen that the ratio of reserved ions is too low. The calculation of temperature rise upon implantation will be discussed in the following. Considering the implantation of an ion beam of 10 cm diameter a Si wafer which was heated by passing a current, the equation of temperature rise can be expressed as W0 ¼ EI=A;

ð8Þ

P ¼ e0 Ai r0 ðTT4  T04 Þ;

ð9Þ

TT ¼ ððW0 =2Ae0 r0 Þ þ T04 Þ

1=4

;

ð10Þ

where W0 is the thermal power; I the density of the ion beam; E the energy of the ion beam; A the implanted area of Si wafer; P the energy of radiation; e0 the emissivity of the Si wafer; r0 the Stephan–Boltzman constant. The temperature rise of the Si wafer TT can be calculated by solving Eq. (4), where E ¼ 40 keV; I ¼ 6:4  103 A; A ¼ 1 cm2 ; T0 ¼ 400 C, e0  0:5; r0 ¼ 5:67  1012

362

Y. He et al. / Nucl. Instr. and Meth. in Phys. Res. B 211 (2003) 358–362

W cm2 K4 , so TT  516 C, TT  T0 ¼ 116 C, which accords with the experimental data. In the process of experiment, the Si wafer was hold by Mo slices which was also implanted by Coþ ions. So the thermal conductivity of thermal conductivity of the sample holder can be ignored. Therefore, the temperature rise was about 100 C which accords with the experimental data.

4. Conclusions In summary, we have reported the results of an experimental study of CoSi2 thin layers. It is concluded that CoSi2 surface layers with low sheet resistivity are synthesized by direct Co implantation into Si(1 0 0) substrates at the temperature of 400 C using a MEVVA ion source. The implantation dose has a significant influence on the crystallinity and electrical properties of CoSi2 layers. But the ratio of reserved implantation dose is only about 3.5% because of sputtering.

Acknowledgements This work was supported by National Natural Science Foundation of China. Professor L.E. Rehn is gratefully acknowledged for fruitful discussion for this study.

References [1] J.P. Gambino, E.G. Colgan, Mater. Chem. Phys. 52 (1998) 99. [2] G.B. Kim, J.S. Kwak, H.K. Baik, J. Appl. Phys. 82 (1997) 2323. [3] C.M. Osburn, J.Y. Tsai, J. Sun, J. Electron. Mater. 25 (1996) 1725. [4] M.-A. Nicolet, S.S. Lau, in: N.G. Einspruch (Ed.), VLST Electronics Microstructure Science, Vol. 6, Academic Press, New York, 1983. [5] S. Matteson, M.A. Nicolet, Annu. Rev. Mater. Sci. 13 (1983) 339. [6] L.G. Brown, J.E. Gavin, R.A. MacGill, Appl. Phys. Lett. 47 (1985) 358. [7] S. Hausmann, L. Bischoff, J. Teichert, M. Voelskow, W. Moller, Jpn. J. Appl. Phys. 38 (1999) 7148. [8] D.H. Zhu, K. Tao, F. Pan, B.X. Liu, Appl. Phys. Lett. 62 (1993) 2356. [9] A.E. White, K.T. Short, R.C. Dynes, J.P. Garno, J.M. Gibson, Appl. Phys. Lett. 50 (1987) 95. [10] A. Vantomme, M.-A. Nicolet, N. David Theodore, J. Appl. Phys. 75 (1994) 3882. [11] P. Liu, B.-Z. Li, Z. Sun, Z.-G. Gu, Wei-Ning Huang, Z.-Y. Zhou, R.-S. Ni, C.-L. Lin, S.-C. Zou, F. Hong, G.A. Rozgonyi, J. Appl. Phys. 74 (1993) 1700. [12] C.W.T. Bulle-Lieuwma, Appl. Surf. Sci. 68 (1993) 1. [13] H.N. Zhu, B.X. Liu, Surf. Coat. Tech. 131 (2000) 307. [14] D.H. Zhu, Y.G. Chen, B.X. Liu, Nucl. Instr. and Meth. B 101 (1995) 394. [15] J.Z. Zhang, X.Y. Ye, J. Chang, S. Bernard, Mater. Lett. 32 (1997) 121. [16] L.C. Feldman, J.W. Mayer, Fundamentals of Surface and Thin Film Analysis, Elsevier Science Publishing Co., 1986. [17] K.Y. Gao, B.X. Liu, Nucl. Instr. and Meth. B 148 (1999) 615. [18] E.C.H. Silk et al., Philos. Mag. 4 (1959) 970.