Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://danp.sinp.msu.ru/Articles_GSIB/nimb_F_clasteretching.pdf Äàòà èçìåíåíèÿ: Fri Oct 7 20:08:36 2005 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:31:32 2012 Êîäèðîâêà:

ARTICLE IN PRESS

Nuclear Instruments and Methods in Physics Research B xxx (2005) xxx­xxx www.elsevier.com/locate/nimb

Molecular dynamics study of the angular dependence of reactive cluster impacts
Takaaki Aoki
b

a,b,*

, Jiro Matsuo

b

a Collaborative Research Center for Advanced Quantum Beam Process Technology, Japan Quantum Science and Engineering Center, Kyoto University, Sakyo, Kyoto 606-8501, Japan

Available online

Abstract The collisional processes of fluorine clusters impacting at various incident angles on bare silicon surfaces were studied, and the change of the surface profile, sticking probabilities of F atoms, sputtering yield and the distribution of the sputtered particles were examined. When a (F2)30 0 cluster with 10 eV/atom impacts on a Si(1 0 0) surface at normal incident, the cluster penetrates the surface causing a spherical crater structure with the F atoms densely covered. As the incident angle increases, the penetration depth of cluster decreases and the simulation shows an asymmetric crater profile. Especially at an incident angle of 75°, the surface profile did not change but F atoms were deposited over a wide area. The distribution of sputtered particles also depends on incident angle. At normal incidence, the Si atoms are sputtered as well as silicon­fluoride molecules such as SiF and SiF2, while only SiFx molecules are sputtered at 75° of incident angle. From these results the surface etching process caused by reactive cluster impacts are discussed. ñ 2005 Elsevier B.V. All rights reserved.
PACS: 36.40.þc; 83.10.Rs Keywords: Gas cluster ion beam; Molecular dynamics; Sputtering; Fluorine; Silicon

1. Introduction The impact processes of gas clusters on solid surfaces are of great interest for new surface modCorresponding author. Address: Quantum Science and Engineering Center, Gokasho, Uji, Kyoto 611-011, Japan. Tel.: +81 774 38 3977; fax: +81 774 38 3978. E-mail address: aoki@sakura.nucleng.kyoto-u.ac.jp (T. Aoki).
*

ification techniques [1,2]. A Cluster is an aggregated material, which consists of several tens to thousands atoms. When a cluster is accelerated and impacts on a solid target, high-density atomic collisions occur and the kinetic energy of the cluster is deposited in very shallow area of the target. This collision process is different from that of monomer ions, and is the basis of a new surface modification technology. The study of gas cluster ion impact began mainly with argon gas clusters,

0168-583X/$ - see front matter ñ 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2005.07.108


ARTICLE IN PRESS

2

T. Aoki, J. Matsuo / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

which is an inert material and is suitable for understanding the physical effect of the multiple collision effect of clusters. Recently, it has been reported that, gas cluster ion beams can be generated with various gas-phase materials such as SF6, CF4, etc. [3]. These new source materials allow gas cluster ion beam surface modification processes that combine the non-linear effects inherent in cluster impacts with the chemical properties of source gas materials. For example, in the case of sputtering, it has been demonstrated that with (SF6)2000 clusters irradiating Si or W substrates, the sputtering yield is 10 times higher than that of Ar2000 clusters, which in turn is 10 times higher than that of monomer ions with the same total acceleration energy of 20 keV [4]. The high etching rate of reactive cluster ions is now studied for application in industrial precision nano-scale patterning processes [5]. It is important to understand the mechanism of the collisional and etching processes of reactive clusters in order to optimize the surface process for specific industrial requirements. In this study, molecular dynamics (MD) simulations of fluorine clusters impacting on bare silicon surface were performed. The angular dependence of the change of surface profile, the sticking of F atoms and the sputtering of Si atoms are examined. The results of the etching process by reactive cluster ion are discussed.

K before impact using the Langevine dynamics method. After annealing, the fluorine cluster has a spherical and amorphous structure, which avoids artifacts due to the structure of the cluster. The Si(0 0 1) substrate was prepared as the target. The target consists of about 500,000 atoms ° ° ° and its dimension is 360 A · 360 A · 100 A, which is large enough to receive the incident energy when the F atoms collide at glancing angle. A periodic boundary condition is applied to the horizontal plane. The Si atoms in the lower four layers are fixed to keep the bulk status and the bottom region of the substrate (up to 1/4 the thickness) is simulated by the Langevine dynamics method to keep the substrate temperature at 300 K. It should be noted that, the starting target material in this work is bare silicon, so that these simulations represent the very initial stage of reactive cluster ion impact, not the steady state expected for real experiments, where the surface is covered with silicon fluoride compounds.

3. Results and discussion The (F2)300 cluster was given a total energy of 6 keV and irradiated at various incident angles on the Si(1 0 0) target. Each F atom in the cluster shares the total incident energy, so each F atom carries 10 eV. Fig. 1 shows snapshots of the impact processes of a fluorine cluster. In Fig. 1 black and gray circles indicate F and Si atoms, respectively. When the cluster impact is normal to the surface, the cluster penetrates into the target. During the collisional process, large numbers of collisions occur between the cluster and the substrate atoms. Because of this multiple collision effect, the incident atoms are directed not only vertically but also horizontally, which results in a hemispherical crater formation on the target surface. After the impact, some fluorine atoms remain stuck on the surface while others desorb into the vacuum. As the incident angle increases, the penetration depth of the cluster becomes shallower, and the incident F atoms tend to evaporate in different directions from that of the incident cluster. At an incident angle of 75°, the incident cluster slides on the target surface with its upper part remaining in the cluster

2. Simulation model Our group has studied the collisional processes of fluorine atoms, molecules and clusters on silicon targets using the molecular dynamics method [6,7]. In this work, the angular dependence of the collisional process of fluorine clusters is reported. In order to describe the interactions between Si­Si, Si­F and F­F, the Stillinger and Weber (SW) type potential model was applied. The SW type potentials were developed by Stillinger et al. [8­10] and modified by Weaklime et al. [11]. This potential model was modeled from the results of ab initio calculations and well represents the energy and structure of silicon­fluoride molecules and surface sticking processes An (F2)300 cluster was used as a projectile; it was annealed in the simulation at 10


ARTICLE IN PRESS

T. Aoki, J. Matsuo / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

3

Fig. 1. Snapshots of (F2)30 0 clusters with 10 eV/atom impacting on bare Si(1 0 0) surface at different incident angles.

state. After several picoseconds, the cluster breaks up and many F atoms leave the target. Fig. 2 shows the surface profiles of the targets at 16 ps after F clusters impact with various incident angles. As the incident angle increases, the depths of the craters decrease and the shapes become asymmetric. Finally, at 75° of incident angle, the crater-like shape does not occur. The previous

100 90 80 70 60 50 40 30 20 10 0 -10 -20 -30 -200 -150 -100 -50 0 50 100 150 200 0 deg. 60 deg. 45 deg. 30 deg. 15 deg. 75 deg.

(F 2 )

300

1 0 e V /a to m

X Position [å]
Fig. 2. Surface profiles after impact by (F2)30 different angles, 16 ps after impact.
0

clusters at

work on Ar clusters impacting on Si targets [12] has shown that when a cluster impacts in the normal direction, there is threshold energy of several eV per atom where the cluster will penetrate the surface of the target and create crater-like damage. For this study, such a threshold energy would as apply to the vertical component of the incident energy, which decreases with a cos2 h law. This value is calculated to be about 0.67 eV/atom at 75° and 2.5 eV/atom at 60°. The rapid decrease of the vertical component of incident energy causes the difference in the profiles in Fig. 2. Fig. 3 shows the angular dependence of the fluorine sticking probability for a (F2)300 cluster with 10 eV/atom. It is natural that the sticking probability depends on the number of adsorption sites on the surface. For glancing angle ion irradiation, the sticking probability should increase with incident angle because the projection area of the cluster obeys 1= cos h. However, Fig. 3 shows the sticking probability decreases as incident angle increases. Two reasons are considered. The first is that, when a (F2)30 0 cluster impacts at a very large incident angle, some part of F atoms in the cluster do not contact the silicon target, as shown in the bottom line of Fig. 1. In other words, too many F atoms are supplied at the impact. Another reason is that, when the incident angle is small (very

Surface [å]


ARTICLE IN PRESS

4
1.0 0.9 0.8 0.7

T. Aoki, J. Matsuo / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

T o ta l 6 k e V (F 2 )3 0 0

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 10 20 30 40 50 60 70 80 90

Incident Angle [deg.]
Fig. 3. Incident angle dependence of the sticking probability of F atoms, 16 ps after impact.

near to the normal), the cluster can penetrate the surface and make a crater, which increases the number of adsorption sites. From Fig. 3, at normal incidence, more than 300 F atoms remain ° ° within an area of about 100 A · 100 A on the target surface, which implies that precursors can be formed at very high density with only one impact of the cluster and suggests an enhancement of the chemical reaction between fluorine and silicon atoms. As can seen in Fig. 1, not only fluorine but also silicon atoms evaporate at all incident angles, and
50 45 40
50

some sputtered atoms has molecule state. Fig. 4 shows the distribution of sputtered particles. In this study, the sputtered molecule was defined as a cluster in which atoms are linked with each other ° within 3.8 A of interatomic distance, but are not linked with substrate atoms at 16 ps after impact. Therefore, it is noted that Fig. 4 eliminates the long-term thermal desorption process and only represents the sputtering process under high-density energy and momentum deposition in very short time range. At 0° and 45° incident angles, non-fluoridated silicon was observed as sputtered material as well as silicon­fluoride molecules such as SiF and SiF2. Under these collisional conditions, many Si atoms are moved horizontally along the surface, due to the multiple-collision process. During the multiple collisions, some target atoms leave from the edge of crater. This lateral sputtering effect is an important collisional process in large cluster impacts and leads to surface smoothing [13]. On the other hand, when the incident angle is as large as 75°, most of sputtered particles are silicon­fluoride molecules. In this case, an enhanced chemical reaction due to the high-density irradiation effect is the dominant mechanism of sputtering. As for the total desorption yield, the maximum for both Si and F occurs at 45°. In order to achieve higher etching, both the formation and desorption of the precursors should be enhanced. The results graphed in Fig. 4 indicate that at 45° incident angle, crater formation contributes to the former enhancement mechanism and asymmet50

F Stick Probability

(F 2)

300

0 deg

45 40

(F 2)

300

45 deg

45 40

Yield [atoms/impact]

Si (total 50) F (total 51)

(F2)

300

75 deg

Yield [atoms/impact]

30 25 20 15 10 5 0 Si Si2 Six SiF SiF SiF SiF4 SixFy

30 25 20 15 10 5 0 Si Si2 Six SiF SiF2 SiF3 SiF4 Six Fy

Yield [atoms/impact]

35

Si (total 117) F (total 104)

35

Si (total 55 ) F (total 76 )

35 30 25 20 15 10 5 0 Si Si2 Six SiF SiF2 SiF3 SiF4 SixFy

2

3

Species

Species
30 0

Species

Fig. 4. Distributions of sputtered particle species caused by the impact of (F2) after impact.

clusters at 0°, 45° and 75° of incident angle, 16 ps


ARTICLE IN PRESS

T. Aoki, J. Matsuo / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx

5

ric momentum transfer due to glancing angle irradiation contributes to the latter one.

Acknowledgement This study is supported by Ministry of Economy, Trade and Industry (METI) and New Energy and Industrial Development Organization (NEDO) in Japan.

4. Conclusions Molecular dynamics simulations of fluorine cluster impacts were performed in order to examine the surface etching process by reactive gas cluster impacts. It was shown that, by changing the incident angle, the shape of the crater could be controlled. Additionally, when the vertical component of the incident energy is less than several eV, a cluster cannot penetrate the target surface and chemical reactions only occur very near to the surface. The predicted sticking probability and sputter rate for (F2)30 0 clusters with 6 keV are large enough to form a high density of silicon­fluoride precursors on the target surface. The study of the sputtered products indicates that the sputtering process can be enhanced due to both the physical and the chemical processes of the cluster impact. The former mechanism is the horizontal momentum transfer effect termed as lateral sputtering and the latter is the high-density particle irradiation effect. These collisional processes occur in the region very near to the surface, shallower than ° 100 A, so it is expected that the reactive gas cluster ion beam technique will achieve high efficiency with low damage when used as a nano-scale etching processes.

References
[1] I. Yamada, J. Matsuo, N. Toyoda, T. Aoki, E. Jones, Z. Insepov, Mater. Sci. Eng. A 253 (1998) 249. [2] I. Yamada, J. Matsuo, Z. Insepov, T. Aoki, T. Seki, N. Toyoda, Nucl. Instr. and Meth. B 164­165 (2000) 949. [3] A. Kirkpatrick, Nucl. Instr. and Meth. B 206 (2003) 830. [4] N. Toyoda, H. Kitani, N. Hagiwara, J. Matsuo, I. Yamada, Mater. Chem. Phys. 54 (1998) 106. [5] E. Bourelle, A. Suzuki, A. Sato, T. Seki, J. Matsuo, Jpn. J. Appl. Phys. 43 (2004) L1253. [6] T. Aoki, J. Matsuo, I. Yamada, Nucl. Instr. and Meth. B 164­165 (2000) 546. [7] T. Aoki, J. Matsuo, I. Yamada, Nucl. Instr. and Meth. B 180 (2001) 164. [8] F.H. Stillinger, T.A. Weber, J. Chem. Phys. 88 (1988) 5123. [9] F.H. Stillinger, T.A. Weber, Phys. Rev. Lett. 62 (1989) 2144. [10] T.A. Weber, F.H. Stillinger, J. Chem. Phys. 92 (1990) 6239. [11] P.C. Weakliem, C.J. Wu, E.A. Carter, Phys. Rev. Lett. 69 (1992) 200. [12] T. Aoki, J. Matsuo, G. Takaoka, Nucl. Instr. and Meth. B 202 (2003) 278. [13] T. Aoki, J. Matsuo, Z. Insepov, I. Yamada, Nucl. Instr. and Meth. B 121 (1997) 49.