Antimony Clustering due to High-dose Implantation

Transcription

1 Mat. Res. Soc. Symp. Vol Materials Research Society Antimony Clustering due to High-dose Implantation Kentaro Shibahara and Dai Onimatsu Research Center for Nanodevices and Systems Hiroshima University Kagamiyama, Higashi-Hiroshima, Hiroshima , Japan ABSTRACT Antimony implantation is a promising technique for fabricating ultra-shallow n + /p junctions for extensions of sub-100-nm n-mosfets. By increasing the Sb + implantation dose to 6x10 14 cm -2, sheet resistance (R s ) of an implanted layer was reduced to 260 Ω/sq. for rapid thermal annealing (RTA) at 800 C. The obtained junction depth of 19 nm is suitable for sub-100-nm MOSFETs. However, the reduction in the sheet resistance showed a tendency to saturate. No pileup at the Si-SiO 2 interface, which was the major origin of dopant loss in lower dose cases was, observed in Sb depth profiles in this case. However, in the case of 900 C RTA, Sb depth profiles indicated that Sb was nearly immobile in the region where Sb concentration exceeded 1x cm -3. These results imply that the major limiting factor of R s reduction under the highdose condition is Sb precipitation, which is different from the lower dose cases. INTRODUCTION Antimony is suitable for the fabrication of ultra-shallow n + /p junctions by ion implantation for sub-100-nm MOSFETs because it is heavier than As or P and its diffusion coefficients are smaller than theirs. Recently, not only junction depth (X j ) but also sheet resistance (R s ) has become an increasingly important factor. This is because the parasitic resistance sometimes limits the MOSFETs performance as their gate length is scaled and channel resistance is lowered. We have demonstrated ultra-shallow (19 nm) and low-resistive (1.4 kω/sq.) n + /p junction formation by 10 kev 1x10 14 cm -2 Sb implantation and applying it to source and drain extensions of 150-nm gate length MOSFETs [1]. Annealing was performed with a furnace at 850 C for 30 min in that case. The R s should be reduced to meet the demand for sub-100-nm generations. Increasing implantation dose is a conventional way to achieve this reduction, but it is not effective because of dopant loss due to pileup at the SiO 2 -Si interface. About half of the implanted Sb was lost due to the pileup [2]. The pileup was suppressed by changing the annealing method from furnace annealing to rapid thermal annealing (RTA) [3,4]. Figure 1 shows Sb depth profiles after RTA. Pileup was observed only for 1000 C RTA. The R s was reduced to 450Ω/sq. for 800 C 10 s RTA by suppressing pileup and increasing implantation dose B8.5.1

2 Sb Concentration [ cm -3 ] SiO2 Si 800 C 900 C Dose 3x10 14 cm -2 Annealing Time 10 s 1000 C Figure 1. Sb depth profiles after RTA. The pileup is observed only for 1000 C RTA. to 3x10 14 cm -2. However, Sb concentration exceeds thermal equilibrium solid solubility (1x cm -3 at 850 C and 2x cm -3 at 1000 C [5]) around a peak region. In the case of furnace annealing for a long time, Sb clusters or precipitates [6]; on the other hand, in the case of RTA carrier depth profiles obtained by the differential Hall measurement method indicated activation higher than the thermal equilibrium solid solubility, or in other words supersaturation of Sb[3,4]. Since annealing time for RTA is much shorter than that for furnace annealing, the limit presupposed by the thermal equilibrium condition cannot be simply applied to RTA. We have investigated the influence of further increasing the Sb implantation dose. The Sb depth profiles and R s were evaluated while changing RTA temperature and time. EXPERIMENTAL Antimony ions were implanted into p-type Si(100) substrates with a 5-nm screen oxide at a dose of 6x10 13 cm -2. Implantation energy was fixed to 10 KeV. RTA was performed in a nitrogen atmosphere for dopant activation. Temperature ramp rate during heating was 100 C/s. Annealing time, which represents holding time at the maximum temperature was selected from 0, 3 and 10 s. The case whose holding time was 0 s is mentioned as spike RTA in this paper. The screen oxide was used to prevent out-diffusion during RTA. Sheet resistance of the Sb + implanted layer was evaluated by two- or four-terminal test structures fabricated with additional processes such as isolation, deep junction formation with As + implantation for contact formation, and Al pad formation. Antimony depth profiles were evaluated by Secondary Ion Mass Spectrometry (SIMS). The primary ion for the SIMS measurement was Cs + and its energy was 1 kv. The screen oxide was not usually removed prior to SIMS measurement. B8.5.2

3 RESULTS AND DISCUSSION Figure 2 shows Sb depth profiles for the implantation dose of 6x10 14 cm -2. In this case, the screen oxide was removed before SIMS analysis for accurate junction depth measurement. The junction depth defined at 5x10 17 cm -2 is 19 nm and 25 nm for 800 C and 900 C RTA, respectively. These depths are suitable value for the sub-100-nm MOSFETs extensions. Figure 3 shows the relationship between implantation dose and R s. By increasing the implantation dose to 6x10 14 cm -2, R s is reduced to 260 Ω/sq. for 800 C 10 s RTA. The sheet resistance values obtained with the twoor four- terminal test structures are smaller than those obtained with the four-point-probe method in our previous work [3,4]. We believe the values in this work are more accurate than the previous ones because the test structures of various dimensions showed identical results. Although R s reduces as the implantation dose is increased, reduction between 3x10 14 cm -2 and 6x10 14 cm -2 is very small, especially in the case of 900 C RTA. The depth profile for 800 C in Fig. 2 is almost identical to that for the as implanted case. However, that for 900 C shows the following features that are not clear in the 3x10 14 cm -2 case in Fig. 1. The immobile atoms form a narrowed peak that accompanies shoulder formation due to relatively large diffusion in a distribution tail region. The immobile Sb peak is attributed to Sb precipitation, as revealed by Solmi et al. [7]. They pointed out that Sb precipitation occurs very quickly when Sb concentration exceeds a threshold concentration of 3-4x cm -3, and as a result activated Sb is less than the threshold. The shoulder height that corresponds to the peak concentration of mobile Sb is about 2x cm -3 in Fig. 2, which coincides with their results. The tendency to saturate in the R s reduction in Fig. 3 supports the precipitation formation. As shown in Fig. 1, when the implantation dose is 3x cm -2, the peak Sb concentration is about Sb Concentration [ cm -2 ] As Implanted RTA 800 C 10 s After Oxide Stripping Dose 6x10 14 cm C 10 s Figure 2. Antimony depth profiles after and before RTA. The screen oxide was stripped before SIMS analysis. Sheet Resitance Rs [ k /sq. ] Four-point Probe [3,4] This Work 800 C 10 s RTA 900 C 10 s 1x x x10 14 Dose [ cm -2 ] Figure 3. Relationship between implantation dose and sheet resistance R s. Though R s is reduced as the dose increases, the R s reduction shows the tendency to saturate. B8.5.3

4 Sb Concentration [ cm -3 ] Spike 3 s 10 s RTA 900 C Dose 6x10 14 cm -2 SiO 2 Si Figure 4. Variation of Sb depth profiles due to annealing time. Distribution peak gradually moves to the surface side and becomes higher as annealing time is extended. Sb Concentration [ cm -3 ] SiO 2 Si 900 C 10 s 1000 C Spike 1000 C 10 s Figure 5. Comparison of the 1000 C RTA cases with 900 C. The profile for 1000 C spike RTA is almost identical to that for 900 C 10 s. 3x cm -3. Since the value is slightly lower than the precipitation threshold, the effective reduction in R s due to increasing the implantation dose from 1x cm -2 to 3x cm -2, can be understood as a result of the absence of Sb precipitation. Figure 4 shows time dependence of Sb depth profiles for 900 C RTA. Diffusion at the tail region between the spike RTA and 3 s is much smaller than that between 3 s and 10 s. Antimony is known as a vacancy diffuser, and no transient enhanced diffusion was reported. We speculate that precipitation formed between 3 s and 10 s and accompanies point Sb Concentration [ cm -3 ] As Annealed After Oxide Stripping RTA 1000 C 10 s Dose 6x10 14 cm Figure 6. Comparison of Sb depth profiles for as annealed and after the 5-nm-tchick screen oxide stripping. Pileup at the SiO 2 /Si interface was not removed by the stripping. B8.5.4

5 defect formation, which enhances Sb diffusion. We should note that the distribution peaks in Fig. 2 and Fig. 4 move to the surface side as the annealing temperature is raised or annealing time is expanded. This tendency is clearer by comparing the profiles for the spike and 10 s RTA at 1000 C in Fig. 5. It is also significant that the depth profile for 900 C 10 s is almost identical to that for 1000 C spike RTA. Although the profile for 1000 C 10 s in Fig. 5 shows pileup at the SiO 2 /Si interface, it is different from the pileup for low dose cases [1,2]. This is because the pileup is not removed by the screen oxide stripping as shown in Fig. 6. In our previous work [4], we evaluated the Sb depth profile for the high-dose (1x10 15 cm - 2 ) long-time (30 min at 900 C) annealing case. In that case, the pileup peak was not completely removed by the oxide stripping and Sb precipitation was observed at the SiO 2 /Si interface by TEM. The shapes of the precipitates indicated that precipitation nucleated at the interface. Variations in the depth profiles in Fig. 5 indicates that immobile precipitated Sb is transferred to the SiO 2 /Si interface. Solmi et al. reported that Sb precipitation grows at the expense of smaller ones [7]. In contrast to their case, since here the interface is so close, mobile Sb atoms attributed to decomposition of the bulk precipitation fall into the interface sink and forms the pileup. We speculate that the behavior of Sb is as follows based on the results described above. 1st Stage ( 800 C 10 s and 900 C 3 s ) A part of the Sb atoms begin to form precipitation in a high-concentration region, but this does not greatly affect depth profiles. 2nd Stage ( 900 C 10 s and 1000 C spike ) As Sb precipitation progresses, generated point defects increase Sb diffusivity. As a result, precipitated Sb is revealed as the immobile peak. 3rd Stage ( 1000 C 10 s ) Large precipitation begins to grow at the SiO 2 /Si interface. The precipitation formed in bulk is consumed as a source of the precipitation at the interface. CONCLUSION Depth profiles and R s of high dose Sb + implanted layers were evaluated. By increasing Sb + implantation dose to 6x10 14 cm -2, R s of the implanted layer was reduced to 260 Ω/sq. for 800 C RTA. However, reduction in R s showed a tendency to saturate. This saturation tendency is attributed to Sb precipitation. As annealing time was expanded or annealing temperature was raised, the precipitaeded Sb redistributed to the SiO 2 /Si interface. From the viewpoint of practical MOSFET fabrication, if the Sb + dose is larger than 6x10 14 cm -2, maximum heat treatment temperature after Sb implantation should be limited to below 800 C in order to avoid deterioration in R s and distribution spreading. B8.5.5

6 ACKNOWLEDGMENTS Part of this work was supported by the Core Research for Evolutional Science and Technology (CREST) of Japan Science and Technology Corporation (JST) and a Grant-in-aid for Scientific Research (C) from the Ministry of Education, Science, Sports and Culture. REFERENCES 1. K. Shibahara, M. Mifuji, K. Kawabata, T. Kugimiya, H. Furumoto, M. Tsuno, S. Yokoyama, M. Nagata, S. Miyazaki and M. Hirose, Tech. Digest Int. Electron Devices Meeting, 579 (1996). 2. K. Shibahara, H. Furumoto, K. Egusa, M. Koh and S. Yokoyama, Mater. Res. Soc. Symp. Proc (1998). 3. K. Shibahara, K. Egusa and K. Kamesaki, Extend. Abst Int. Conf. Solid State Devices and Materials 314 (1999). 4. K. Shibahara, K. Egusa and K. Kamesaki, Jpn. J. Appl. Phys (2000). 5. D. Nobili, R. Angelucci, A. Armigliato, E. Landi and S. Solumi, J. Electrochem. Soc (1989). 6. M. Hashimoto, T. Deguchi, S. Yokoyama and M. Hirose, Jpn. J. Appl. Phys. 33 L1799 (1994). 7. S. Solmi, F. Baruffaldi and M. Derdour, J. Appl. Phys (1992). B8.5.6