Pulsed KrF laser annealing of Mo/Si0.76Ge0.24

Pulsed KrF laser annealing of Mo/Si

Ge

Jian-Shing Luo

_{, Wen-Tai Lin}

_{, C.Y. Chang}

_{, P.S. Shih}

_{, T.C. Chang}

b_{Department of Electronics Engineering, National Chiao Tung University, Hsinchu, Taiwan, ROC} c_{Department of Physics, National Sun Yat-Sen University, Kaohsiung, Taiwan, ROC}

Abstract

Interfacial reactions of Mo/Si0:76Ge0:24were studied by pulsed KrF laser annealing as a function of the energy density

and pulse number. Vacuum annealing was also performed on some samples for comparison. For the samples annealed at a temperature of 500±700°C a continuous hexagonal Mo(Si1ÿxGex)2 (h-Mo(Si1ÿxGex)2) ®lm was formed, while Ge

segregation from the h-Mo(Si1ÿxGex)2®lm to the underlying Si0:76Ge0:24occurred with the extent becoming more severe

at higher annealing temperatures. Concurrently, amorphous structures appeared in the Si0:76Ge0:24substrate. At 700°C

h-Mo(Si1ÿxGex)2 transformed to tetragonal Mo(Si1ÿxGex)2 (t-Mo(Si1ÿxGex)2). Multiple-pulse laser annealing could

produce a continuous h-Mo(Si1ÿxGex)2®lm without forming amorphous structures in the Si0:76Ge0:24substrate,

how-ever, it could not suppress Ge segregation. In the present study, no t-Mo(Si1ÿxGex)2was formed upon pulsed KrF laser

annealing even at higher energy densities. Ó 2000 Elsevier Science B.V. All rights reserved. Keywords: Mo(Si1ÿxGex)2; Pulsed KrF laser annealing; Ge segregation; Amorphous structures

1. Introduction

Recently, strained Si1ÿxGex epitaxial layers on

Si have been extensively studied for potential applications in optoelectronic and high-speed electronic devices [1]. For device applications the

formation of metal=Si1ÿxGex ohmic or rectifying

contacts is required. Thus, the interfacial reac-tions of some metals such as Ni [2±4], Pt [5,6], Pd [6±9], Ti [10±15], Co [16±20], W [21,22], Cr [23],

Cu [24,25] and Zr [26] on Si1ÿxGex by

conven-tional furnace annealing, rapid thermal anneal-ing, and pulsed laser annealing have been studied, respectively. For conventional furnace annealing the formation of a ternary phase is generally accompanied with Ge segregation. Ad-ditionally, an agglomeration structure also ap-pears at higher annealing temperatures. These phenomena could be attributed to the higher heat of formation for metal-Si than for metal-Ge [27]. Rapid thermal annealing [12,13] and pulsed laser annealing [3,4,9] can shorten the annealing time, resulting in a reduction of Ge segregation. Fur-thermore, pulsed laser annealing can produce a smooth and continuous germanosilicide ®lm without inducing strain relaxation in the unre-acted Si1ÿxGex ®lm [9].

*_{Corresponding author. Tel.: 275-7575; fax:} +886-6-2745-985.

E-mail address: [email protected] (W.-T. Lin).

0168-583X/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 0 0 ) 0 0 0 2 9 - X

Refractory metals and refractory metal sili-cides have a higher temperature stability that makes them attractive candidates for contact metallizations in ultra-large-scale integration circuits. The interfacial reactions of W on

Si1ÿxGex have been studied by conventional

an-nealing and rapid thermal anan-nealing, respectively [21,22], where Ge segregation may increase the barrier height. As we know, however, the

Mo=Si1ÿxGex system has not been studied yet. In

this work, the interfacial reactions of Mo on

Si0:76Ge0:24 were studied by pulsed KrF laser

annealing as a function of energy density and pulse number. Simultaneously, vacuum

anneal-ing of Mo/Si0:76Ge0:24 was also conducted for

comparison.

2. Experimental

Epitaxial Si0:76Ge0:24 ®lms about 1500 A thick

were grown on n-type Si(1 0 0) at 550°C by ultra-high-vacuum chemical-vapor deposition (CVD). Prior to Mo deposition, the substrates were cleaned by the RCA method and then

immedi-ately loaded into the chamber. Mo about 200 A

thick was deposited onto the Si0:76Ge0:24 ®lms at

200°C by electron gun evaporation at a rate of 1 

A/s. The base pressure was around 1±2  10ÿ6

Torr. Vacuum annealing was carried out at a temperature of 450±700°C in a vacuum of

1±2  10ÿ6 _{Torr. Rapid thermal annealing was}

performed at a temperature of 600±1100°C in a

N2 atmosphere. Pulsed KrF laser annealing was

performed at an energy density of 0.5±3.2 J/cm2

in a vacuum around 2  10ÿ2 _{Torr. The pulse}

length was 14 ns. The laser beam was focused

onto an area of 4  4 mm2_{. For each laser}

an-nealing the samples were illuminated by a single pulse unless otherwise speci®ed. Phase transfor-mation, the distribution of chemical species, and the microstructures of the annealed samples were examined by using transmission electron micros-copy (TEM) in conjunction with energy disper-sive spectrometry (EDS), which were equipped with a ®eld emission gun with an electron probe 12 A in size.

3. Results and discussion 3.1. Furnace annealing

For the samples annealed at a temperature of

500±600°C hexagonal Mo(Si1ÿxGex)2, hereafter

referred to h-Mo(Si1ÿxGex)2, was formed. Above

700°C all h-Mo(Si1ÿxGex)2 was transformed to

tetragonal Mo(Si1ÿxGex)2, hereafter referred to

t-Mo(Si1ÿxGex)2. From XTEM examination the

Mo(Si1ÿxGex)2 layers were continuous without

agglomeration. From EDS analysis the Ge con-centration in the Mo(Si1ÿxGex)2 layer was

signi®-cantly de®cient with the extent becoming more severe at higher annealing temperatures. One ex-ample is shown in Fig. 1, in which Ge segregates

from the h-Mo(Si1ÿxGex)2 layer to the underlying

Si0:76Ge0:24, forming a Ge-rich Si1ÿxGex layer. In

the h-Mo(Si1ÿxGex)2layer the upper part is nearly

Ge-free, while the lower part contains little amounts of Ge. It is interesting to note that in Fig. 1 two isolated areas with bright contrast

underlying the Mo(Si1ÿxGex)2 layer are present,

which even penetrate to the Si substrate. From

Fig. 1. (a) XTEM micrograph and (b) the depth pro®les of the chemical species for the Mo/Si0:76Ge0:24 sample annealed at 550°C.

microdi€raction analysis the two areas are amor-phous. Furthermore, the EDS analysis in Fig. 2 shows that the two areas are Ge-free and de®cient in Si. It is well known that in the thermal reac-tions of Mo/Si Si is the dominant di€usion spe-cies. It seems that upon vacuum annealing Ge di€uses out of the amorphous areas to react with Si0:76Ge0:24, forming Ge-rich Si1ÿxGex, meanwhile

Si di€uses out of them to react with the Mo

overlayer, forming MoSi2 and Ge-de®cient

Mo(Si1ÿxGex)2.

In our previous studies such as Ni [4], Pd [9]

and Co [16] on Si0:76Ge0:24, the agglomeration of

Ge-de®cient germanosilicide started to appear at a temperature of 400±500°C. However, in the

pre-sent study the Mo(Si1ÿxGex)2 layer remained

con-tinuous even after 700°C annealing. It is conjectured that the interfacial energy of Mo(Si1ÿxGex)2/Si1ÿxGex is signi®cantly lower than

those of Co, Ni and Pd germanosilicides on Si1ÿxGex.

Upon rapid thermal annealing h-Mo(Si1ÿxGex)2

and t-Mo(Si1ÿxGex)2 were formed at 750°C and

950°C, respectively. At 750°C, the same phenom-ena, i.e., Ge segregation and the formation of amorphous structures, as observed in vacuum annealing also occurred.

3.2. Pulsed laser annealing

For the samples annealed at an energy density

of 0.9±1.2 J/cm2 _h-Mo(Si

1ÿxGex)2 was formed

concurrently with signi®cant amount of remaining

Mo. At an energy density of 1.4±2.8 J/cm2_{Mo was}

completely transformed to h-Mo(Si1ÿxGex)2 and

Mo5(Si1ÿxGex)3 as shown in Fig. 3. At 2.6 J/cm2

constitutional supercooling apparently occurred

with the Si1ÿxGex islands surrounded by the

ger-manosilicide. At an energy density of 3.0±3.2 J/cm2

Mo5(Si1ÿxGex)3 was completely transformed to

h-Mo(Si1ÿxGex)2. The sluggish transformation from

Mo5(Si1ÿxGex)3 to h-Mo(Si1ÿxGex)2 may be

at-tributed to the rapid melt/solidi®cation process for pulsed laser annealing, in which the reaction time is too short to allow the intermixing between the species to be complete. In addition, it is worth

to note that even after annealing at 3.2 J/cm2

t-Mo(Si1ÿxGex)2 was still absent.

Upon multiple-pulse annealing pure

h-Mo(Si1ÿxGex)2®lms can be grown at lower energy

densities. In the present study, pure and

continu-ous h-Mo(Si1ÿxGex)2 ®lms have been grown at 0.5

J/cm2_{for 50 pulses, 0.6 J/cm}2_{for 30 pulses, 0.8 J/}

cm2 _{for 20 pulses, and 1.0 J/cm}2 _{for 10 pulses,}

respectively. One example is shown in Fig. 4, in

which no amorphous structures appear in the Si1ÿxGex®lm. Evidently, the microstructures of the

laser-annealed samples are superior to those of the furnace-annealed samples. This result may be ex-plained in terms of the rapid melt/solidi®cation process for pulsed laser annealing. During pulsed laser annealing although the pulse length, 14 ns, is very short, it allows the species such as Mo, Si, and Ge to intermix between each other in the melt and then form the Mo germanosilicide in a rapid quench. The interfacial reactions can be proceeded by each individual pulse. Inversely, for furnace annealing Si and Ge are the di€usion species and the annealing time is long enough to allow them to

di€use out of the Si0:76Ge0:24 ®lm, forming the

amorphous structures as shown in Fig. 1. How-ever, upon multiple-pulse annealing signi®cant amounts of Ge segregation out of the germano-silicide to the underlying Si0:76Ge0:24 still occurred

as seen in Fig. 4(b). It has been found that for Pd

[9] and Co [20] on Si0:76Ge0:24 multiple-pulse

annealing at an appropriate energy density is an e€ective method to produce a continuous ger-manosilicide ®lm without inducing Ge segregation to the unreacted Si0:76Ge0:24 ®lm and strain

relax-ation. This discrepancy may be attributed to the larger di€erence between the heats of formation

for MoSi2, )44 kJ/mol, and MoGe2, )14 kJ/mol,

as compared with those for Pd2Si, )65 kJ/mol,

Pd2Ge, )58 kJ/mol, CoSi2, )33 kJ/mol, and

CoGe2, )12 kJ/mol [27].

In the present study, h-Mo(Si1ÿxGex)2remained

inert without transforming to t-Mo(Si1ÿxGex)2

Fig. 4. (a) XTEM micrograph and (b) the depth pro®les of the chemical species for the sample annealed at 1.0 J/cm2 _{for 10} pulses, (c) DP of the h-Mo(Si1ÿxGex)2layer.

Fig. 3. (a) Plan-view micrograph of the sample annealed at 1.4 J/cm2_{. (b) Electron di€raction patterns (DP) of (a)} showing that h-Mo(Si1ÿxGex)2 was formed concurrently with Mo5(Si1ÿxGex)3.

even after annealing at higher energy densities

such as 3.2 J/cm2 _{for one pulse and 2.6 J/cm}2 _for

20 pulses. One example is shown in Fig. 5. It is conceived that during pulsed laser annealing the rapid melt/solidi®cation process renders the reac-tion kinetics to play an important role in deter-mining the phase formation. The present results indicate that the reaction kinetics favors the formation of the low temperature phase, h-Mo(Si1ÿxGex)2.

4. Summary and conclusions

1. Upon conventional annealing continuous

Ge-de®cient Mo(Si1ÿxGex)2 ®lm was formed, while

Ge-free amorphous structures appeared in the

Si0:76Ge0:24 substrate. Ge segregated from the

amorphous structures and the Mo(Si1ÿxGex)2

®lm to the remaining Si0:76Ge0:24 substrate,

forming Ge-rich Si1ÿxGex. At higher

tempera-tures, h-Mo(Si1ÿxGex)2 was transformed to

t-Mo(Si1ÿxGex)2.

2. Upon multiple-pulse laser annealing at an

ener-gy density of 0.5±1.0 J/cm2 _{a continuous and}

smooth h-Mo(Si1ÿxGex)2 ®lm was formed, no

amorphous structures appeared in the

Si0:76Ge0:24®lm. However, Ge segregation from

the Mo(Si1ÿxGex)2 ®lm to the underlying

Si0:76Ge0:24®lm still occurred.

3. In the present study, even after annealing at

higher energy densities such as 3.2 J/cm2 _for

one pulse and 2.6 J/cm2 _{for 20 pulses,}

respec-tively, h-Mo(Si1ÿxGex)2 remained inert

with-out transforming to t-Mo(Si1ÿxGex)2. It is

conjectured that the rapid melt/solidi®cation process for pulsed laser annealing renders the reaction kinetics to favor the growth of h-Mo(Si1ÿxGex)2.

References

[1] S.S. Iyer, G.L. Patton, J.M. Stork, B.S. Meyerson, D.L. Harame, IEEE. Trans. Electron. Dev. 36 (1989) 2043. [2] R.D. Thompson, K.N. Tu, J. Angillelo, S. Delage, S.S.

Iyer, J. Electrochem. Soc. 135 (1988) 3161.

[3] J.S. Luo, W.T. Lin, C.Y. Chang, W.C. Tsai, J. Appl. Phys. 82 (1997) 3621.

[4] J.S. Luo, W.T. Lin, C.Y. Chang, W.C. Tsai, Mater. Chem. Phys. 54 (1998) 160.

[5] Q.Z. Hong, J.W. Mayer, J. Appl. Phys. 66 (1989) 611. [6] H.K. Liou, X. Wu, U. Gennser, S.S. Iyer, K.N. Tu, E.S.

Yang, Appl. Phys. Lett. 60 (1992) 577.

[7] A. Buxbaum, M. Eizenberg, A. Raizman, F. Scha‚er, Appl. Phys. Lett. 59 (1991) 665.

[8] A. Buxbuam, E. Zolotoyabko, M. Eizenberg, F. Scha‚er, Thin Solid Films 222 (1992) 157.

[9] D.R. Chen, J.S. Luo, W.T. Lin, C.Y. Chang, P.S. Shin, Appl. Phys. Lett. 73 (1998) 1355.

[10] O. Thomas, S. Delage, F.M. dÕHerule, G. Scilla, Appl. Phys. Lett. 54 (1989) 228.

[11] W.J. Oi, B.Z. Li, W.N. Huang, Z.G. Gu, H.Q. Lu, X.J. Zhang, M. Zhang, G.S. Dong, D.C. Miller, R.G. Aitken, J. Appl. Phys. 77 (1995) 1086.

[12] D.B. Aldrich, Y.L. Chen, D.E. Sayers, R.J. Nemanich, S.P. Ashburn, M.C. Ozturk, J. Appl. Phys. 77 (1995) 5107. [13] D.B. Aldrich, H.L. Heck, Y.L. Chen, D.E. Sayers, R.J.

Nemanich, J. Appl. Phys. 78 (1995) 4958.

[14] A. Eyal, R. Brener, R. Beserman, M. Eizenberg, Z. Atzmon, D.J. Smith, J.W. Mayer, Appl. Phys. Lett. 69 (1996) 64.

[15] W. Freiman, A. Eyal, Y.L. Khait, R. Beserman, K. Dettmer, Appl. Phys. Lett. 69 (1996) 3821.

[16] J.S. Luo, W.T. Lin, C.Y. Chang, W.C. Tsai, S.J. Wang, Mater. Chem. Phys. 48 (1997) 140.

Fig. 5. (a) Plan-view micrograph of the sample annealed at 2.6 J/cm2_{for 20 pulses. (b) DP of (a) showing that h-Mo(Si}

1ÿxGex)2 was formed concurrently with poly-Si1ÿxGex.

[17] M.C. Ridgway, R.G. Elliman, N. Hauser, J.M. Baribeau, T.E. Jackman, Mater. Res. Soc. Symp. Proc. 260 (1992) 857.

[18] F. Lin, G. Sarcona, M.K. Hatalis, A.F. Cserhati, E. Austin, D.W. Greve, Thin Solid Films 250 (1994) 20. [19] Z. Wang, Y.L. Chen, H. Ying, R.J. Nemanich, D.E. Sayer,

Mater. Res. Soc. Symp. Proc. 320 (1994) 397.

[20] J.S. Luo, Y.L. Hang, W.T. Lin, C.Y. Chang, P.S. Shih, J. Mater. Res. 14 (1999) 3433.

[21] V. Aubry, F. Meyer, R. Laval, C. Clerc, P. Warren, D. Dutartre, Mater. Res. Soc. Symp. Proc. 320 (1994) 299. [22] V. Aubry, F. Meyer, R. Laval, C. Clerc, P. Warren, D.

Dutartre, Appl. Surf. Sci. 73 (1993) 285.

[23] Q.Z. Hong, J.W. Mayer, Mater. Res. Soc. Symp. Proc. 181 (1990) 145.

[24] E.J. Jaquez, A.E. Bair, T.L. Alford, Appl. Phys. Lett. 70 (1997) 874.

[25] J.S. Luo, W.T. Lin, C.Y. Chang, P.S. Shih, Thin Solid Films 346 (1999) 207.

[26] Z. Wang, D.B. Aldrich, R.J. Nemanich, D.E. Sayers, J. Appl. Phys. 82 (1997) 2342.

[27] F.R. Deboer, R. Boom, W.C. Mattens, A.R. Miedema, A.K. Niessen (Eds.), Cohesion in Metals: Transition Metal Alloys, North-Holland, Amsterdam, 1988.