Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://danp.sinp.msu.ru/Articles_GSIB/nimb_B_smallclustersions.pdf Äàòà èçìåíåíèÿ: Fri Oct 7 22:02:22 2005 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:35:27 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: arp 220

ARTICLE IN PRESS

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

Small B-cluster ions induced damage in silicon
Jiarui Liu *, Xuemai Wang, Shao Lin, Hui Chen, Wei-Kan Chu
Department of Physics, and TcSAM, University of Houston, TX 77204-5932, United States Available online

Abstract Small molecule-cluster ions, such as BF2, BSi, BGe, B10H14 and Bn can be used for shallow junction formation. We studied the cluster induced damage in silicon with different small clusters in keV energy range. The radiation damage was measured by glancing angle RBS/channeling. The measurement shows that the cluster induced damage per atom is quite different from the monomer ion induced damage at the same velocity. The small-cluster ions show strong enhanced radiation damage per atom. In this paper we will show non-linear effects on small boron-cluster induced damage in silicon. The cluster size dependence of this non-linear effect will also be presented. ñ 2005 Elsevier B.V. All rights reserved.
PACS: 36.40.þc; 36.40.þWa; 61.46.+W Keywords: Cluster; Damage; Silicon

1. Introduction As the scale of semiconductor devices decreases, the fabrication of high quality shallow p-type junction becomes more difficult. In order to get a sub0.1 lm p-MOS device, boron ions are expected to be implanted at an energy less than 1 keV. At this low energy, the conventional ion implantation techniques used to dope sub-micron silicon devices have throughput problems. The problems are
Corresponding author. Tel.: +1 713 743 8255; fax: +1 713 743 8201. E-mail address: jrliu@uh.edu (J. Liu).
*

attributed to the basic physics of ion extraction: the space charge limitation or I­V3/2 law. The ion beam current from the ion source drops significantly due to the space charge limitation at very low ion energy. The attempts to meet these major technology challenges have led to the development of new processes which include: (1) cluster ion implantation; (2) other doping methods such as plasma immersion ion implantation; (3) low thermal budget process with spike rapid thermal annealing or laser annealing. Each of these methods has its advantages and disadvantages. The boron-cluster ion implantation technique using decaborane (B10H14) has been proposed for

0168-583X/$ - see front matter ñ 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2005.07.106


ARTICLE IN PRESS

2

J. Liu et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx
10
5

shallow junction formation by Goto et al. [1] and successfully demonstrated for 40 nm p-MOSFETs fabrication [2]. The radiation damage due to the ion implantation by small boron-clusters Bn (n = 1­4) and transient enhanced diffusion (TED) were studied in keV energy range by Jin et al. [3]. This type of ion implantation by small cluster ions exhibit some other positive effects, such as reduction of the channeling effect and reduction of TED [2,3]. On the other hand, some basic processes in small cluster applications, such as the range and straggling in cluster implantation and radiation damage in cluster ion implantation are not well studied. In this paper we will discuss small boron-cluster induced damage in silicon.

a

Cone angle

Ion current (nA)

10 10 10 10

4

90

o o

3

120

2

1

60 50 40 30 20 10 0 0 1 2 3 4 5 6 7
n

b

8

9

10

11

Number of n in C

2. Source for small clusters The small boron-clusters were extracted from a source of negative ions by cesium sputtering (SNICS), which is a negative sputtering ion source for monomer ions. To get higher cluster ion output, optimization of the operation parameters is necessary. The beam current of some cluster ions can be increased by one to two orders of magnitude by optimization. The details were published elsewhere [4,5]. Here we just list the principal modifications of the SNICS ion source and the latest results of the small boron-cluster ion beam. In order to obtain a high current for various species of cluster ions, we investigated the sputter target geometry, the materials of the sputter target, sputtering voltage and sputtering cesium ion current. 2.1. Sputtering target geometry Due to the difficulties of getting the boron target in different geometries from powder material, the geometries of the sputter target were tested by high density graphite targets with a cone tip in the front. It was surprising that the cluster beam greatly depends on the geometry with the maximum cluster beam current difference of 1 order of magnitude, while the monomer ion beam depends on the geometry only by a factor less than 2. Fig. 1(a) shows the Cn-cluster ion beam current

Fig. 1. C-cluster ion beam current for different front target surface cone angle of 90° and 120° (a) and the beam ratio for these two geometries (b).

dependence on the cluster size for the flat target surface and the target with a 60° cone tip on the target surface. Fig. 1(b) shows the cluster ion beam current ratio for these two geometries. We can see that the cluster ion beam current is enhanced by an order of magnitude for larger clusters (n $ 10) due to the target surface shape modification. It is interesting to mention that the optimum geometry for cluster ions is different from the optimum geometry for monomer ions. We noticed a general trend that with the increase of cluster size, the cluster ion beam current decreases with some exceptions. 2.2. Material selection Three different sputtering targets were used for B-cluster extraction: Boron-10 target from National Electrostatic Corp. (NEC), Home-made Boron-10 target from enriched Boron-10 powder and Home-made Boron-11 target. The typical cluster ion beam current dependence on the cluster size is shown in Fig. 2. It is interesting to mention that the boron monomer beam current is almost the same for the three different targets. The B2cluster ion current is slightly higher than B1 for three different targets.


ARTICLE IN PRESS

J. Liu et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx
10
5
10

3

10

4

10 11

B from NEC B Home-made B Home-made

10

10

2

10

1

10

0

10

-1

Icluster/Imonomer ratio

3

13 12 11 10 9 8 7 6 5 4 3 2 1 0 0 1 2 3 4 5 6

IB2 / IB1 IB3 / IB1

Ion Current (nA)

0

2

4

6

8

10

7

8

B Cluster Size

Sputtering Voltage (kV)

Fig. 2. B-cluster ion beam current for three different sputtering target materials at sputtering voltage of 5 kV.

Fig. 3. The voltage dependence of I B2 =I B1 and I B3 =I sputtering voltage range from 0.5 kV up to 7 kV.

B

1

in a

The difference in B5 beam current is a factor of 2 for three different boron targets. The difference in B8 and B9 is about two orders of magnitudes, but with considerable errors. The highest B8 and B9-cluster ion beam was obtained from Boron-10 target from NEC. This unusual result was confirmed by nuclear reaction analysis of boron. 2.3. Sputtering voltage

The SNICS source parameter optimization can increase the cluster ion current by one to two orders of magnitude for small clusters. Small clusters from some boron-containing compounds, such as SiB6, SiB4 and GeB4, were also studied for ultrashallow junction study. Ultra-shallow junction was successfully studied with these molecular-cluster ions [3,6].

3. Non-linearity in cluster induced damage The cluster ion yield of different materials shows a general behavior dependence on the sputtering voltage: the cluster beam current decreases at lower sputtering voltage. Remarkable differences were observed in sputtering voltage dependence of the cluster ion current ratio of I Bn =I B1 for different boron cluster sizes. Fig. 3 shows the voltage dependence of I B2 =I B1 and I B3 =I B1 in a sputtering voltage range from 0.5 kV up to 7 kV. At 0.5 kV sputtering voltage, the cluster ion yield of B2 and B3 is higher than B-monomer by a factor of 12.5 and 2.1 respectively. We believe that the sputtering source, both high power positive and negative, can deliver mA of B2 and B3-cluster ion beams at this low sputtering voltage. Due to the gain in cluster ion transport at higher voltage proportional to the cluster size, application of boroncluster beams for shallow junction formation in device fabrication is a practical approach. Small boron-cluster ions, such as BF2, BSi, BGe, B10H14 and Bn can be used for shallow junction formation. We successfully studied the junction formation by the BSi, BGe and Bn (n = 1­4) in our lab. On the other hand, cluster induced damage in silicon with different small clusters in keV energy range is an unknown factor in small cluster­solid interaction. We studied this non-linear effect in small cluster induced damage in silicon with some small clusters such as Cn, Cun and Aun. The radiation damage was measured by glancing angle RBS/ channeling. The measurement shows that the cluster induced damage per atom is quite different from the monomer ion induced damage at the same velocity. The small cluster ions show strong enhanced radiation damage per atom. Such an enhanced non-linear effect in radiation damage is an important factor in low energy ion implantation


ARTICLE IN PRESS

4

J. Liu et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx
4500 4000 3500 3000 2500 2000 1500 1000 500 0 210

by cluster ions. In the non-linear effect in cluster induced damage by Cn,Cun and Aun small clusters at 6 keV per atom, the highest non-linearity of a factor of 15 was for Cu clusters and the lowest non-linearity of a factor of 1.5 was for Au clusters. Here we will study this non-linearity induced by small boron clusters in silicon. The Bn-cluster beam current must be reduced for the sample bombardment to keep low beam heating. The cluster ion beam was scanned over a 8 · 10 mm2 area of a silicon wafer. All measurable contaminations, such as C, O and O2 were filtered out by an analyzing magnet. P-type [1 0 0] Si wafers were irradiated by clusters of different sizes with the same energy per atom (1 keV and 6 keV). Each set of samples was bombarded by Bn with n = 1­4 to the same atomic fluence (in number of B-atoms/cm2). The low fluence rate was kept approximately the same for all clusters to avoid an annealing effect by beam heating. The beam power density during irradiation was in the order of a few mW/cm2, where the maximum temp erature increase was about 10­20 °C, so the annealing effect due to this beam heating was negligible. This upper-bound estimate was based on the radiation energy loss only. The heat reduction due to contact to the sample holder was not counted. The defect production was measured by RBS/ channeling with 2 MeV a-particles along [1 0 0] Si normal incidence direction. We used a three-detector system to perform this RBS/channeling measurement. No. 1 detector at 165° backward angle was used for the sample alignment for channeling. No. 2 detector at a glancing angle of 98° was used for RBS/channeling spectrum measurement with high depth resolution and quantification of silicon defects. No. 3 detector at 155° with a 10 lm Mylar foil in the front was used for a-particle signal detection from 11B(p,a)8Be(!a + a) reaction to get the B fluence measurement. Fig. 4 shows the RBS/channeling spectra of 6 keV B1 and 18 keV B3 bombarded Si to the atomic fluence of 1 · 1015 B-atoms/cm2 along with the spectrum of a virgin Si. The spectra were taken at a glancing angle to get high depth resolution. The difference between the spectra for B1 and B3 for the same B-atomic dosage is a clear indication

Virgin B1 B3

Counts

220

230

240

250

260

270

280

290

300

Channel

Fig. 4. RBS/channeling spectra of 6 keV B1 and 18 keV B3 bombarded Si to the atomic fluence of 1 · 1015 B-atoms/cm2 along with the spectrum of a virgin Si.

on the non-linear effect in the B-cluster induced damage in Si. The quantitative results of the Si displacements per B-atom obtained by subtracting the virgin Si surface peak from the bombarded sampleós nearsurface peak and then normalized by the amount of B-atoms are given in Fig. 5 for B-clusters with fluence of 1 · 1015 B-atoms/cm2. The number of Si displacements per B-atom systematically increased for B1, B2, B3 and B4 respectively. The statistical error is estimated to be ±5%, much less than the non-linearity observed in cluster induced damage. The cluster size dependence of the Bn induced defects was investigated with atomic fluences from 1 · 1015 B-atoms/cm2, which is the lowest dosage that could be measured by RBS/ channeling. Fig. 5 shows these normalized numbers of Si displacements per B-atom versus B-cluster size for two energies of 1 · 1015 B-atoms/cm2 at 6 keV/atom and 1 · 1015 B-atoms/cm2 at 1 keV/ atom. The measured non-linearity for the higher energy of 6 keV/atom is higher than that for the lower energy of 1 keV/atom at the same fluence of 1 · 1015 B-atoms/cm2. Our measurements show that the non-linearity in B-cluster induced damage is significant, and it is very close to the non-linearity due to Cn-clusters.


ARTICLE IN PRESS

J. Liu et al. / Nucl. Instr. and Meth. in Phys. Res. B xxx (2005) xxx­xxx
7
1keV B-cluster 6keV B-cluster 6keV C-cluster

5

6

Normalized damage enhancement

5

cluster bombardment in keV energy range is very close to the C-cluster induced damage. The method of measuring the non-linear effect by using the cluster ions from SNICS ion source and subsequent RBS/channeling measurement are sensitive, reliable and simple.

4

3

Acknowledgements This project was supported by The State of Texas through the Texas Center for Superconductivity and Advanced Materials, partially by the 003652-797 ARP and the National Science Foundation through the Materials Research Science and Engineering Center.
0 1 2 3
Cluster size

2

1

0 4 5

Fig. 5. The non-linearity of B-cluster induced damage in Si for 6 keV B-cluster at 3 · 1014 B-atom/cm2 and 1 keV at 1 · 1015 B-atom/cm2 along with 6 keV C-cluster at 1 · 1015 C-atom/

References
[1] K. Goto, J. Matsuo, Y. Tada, T. Moriyama, T. Sigii, I. Yamada, IEDM Tech. Digest. (1997) 471. [2] I. Yamada, J. Matsuo, N. Touada, T. Aoki, AIP Conference Proceedings 475. p. 379. [3] J.-Y. Jin, J. Liu, P.A.W. van de Heide, W.-K. Chu, Appl. Phys. Lett. 6 (2000) 574. [4] X. Wang, J. Liu, L. Shao, H. Chen, W.-K. Chu, Nucl. Instr. and Meth. B 196 (2002) 198. [5] X. Wang, X. Lu, L. Shao, J. Liu, W.-K. Chu, 16th International Conference on Application of Accelerators in Research and Industry, CAARI 2000, 1­4 November 2000, Denton, TX, USA, AIP Conference Proceedings, Vol. 576, 2001, 1007. [6] X. Lu, L. Shao, J.-Y. Jin, Q. Li, I. Rusakova, Q.Y. Chen, J. Liu, W.-K. Chu, P. Ling, MRS Symp., Vol. 610, 2000, B4.5.1. San Francisco, 4/24/2000.

4. Conclusion In conclusion, we have observed a non-linear effect in B-clusters of 1­4 atoms induced damage in single crystals Si for the cluster energy of 1 and 6 keV per atom. The fluence range was from 3 · 1014 B-atoms/cm2 to 1 · 1015 B-atoms/cm2. The damage was quantified with glancing angle RBS/channeling with improved sensitivity. Our results showed significant non-linearity in defect production in Si by B-clusters. We found that the non-linearity in defect production due to B-