Low-temperature photoluminescence of hydrogen Ion and plasma ...

JOURNAL OF APPLIED PHYSICS

VOLUME 96, NUMBER 1

1 JULY 2004

Low-temperature photoluminescence of hydrogen Ion and plasma implanted silicon and porous silicon Zhenghua An,a) Ricky K. Y. Fu, Weili Li,b) Peng Chen, and Paul K. Chuc) Department of Physics and Material Science, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong

K. F. Li, L. Luo, H. L. Tam, and K. W. Cheah Department of Physics, Hong Kong Baptist University, Waterloo Road, Kowloon, Hong Kong

Chenglu Lin State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology (SIMIT), Chinese Academy of Sciences (CAS), 865 Changning Road, 200050 Shanghai, China

共Received 22 January 2004; accepted 14 April 2004兲 Low-temperature photoluminescence in the infrared region of hydrogen implanted single crystalline silicon is investigated. Both beam-line ion implantation and plasma immersion ion implantation 共PIII兲 are used. The beam-line implanted samples show a broad photoluminescence band below the band gap, whereas the PIII implanted samples show at least one more peak at 1.17 eV and a much wider photoluminescence band. The origins are investigated and the peak at 1.17 eV appears to originate from nonphonon emission enhanced by lattice disorder. Our results suggest that PIII may be a better technique than beam-line ion implantation in introducing a certain disorder into the silicon lattice to circumvent the conservation of quasimomentum and consequently enhance the light emission efficiency from the modified Si samples. Our conclusion is further supported by results from plasma implanted porous Si. © 2004 American Institute of Physics. 关DOI: 10.1063/1.1759784兴

I. INTRODUCTION

characteristics and is more stable than traditional porous silicon 共PS兲 produced by anodic etching. However, the luminescence mechanism in this material is not understood and most of the previous experiments8 –10 have focused on the luminescent lines and broad bands below the band gap of silicon. Pavesi et al.10 reported an extremely weak band at about 0.78 ␮m 共visible region兲, but its origin is still unclear. In the work reported here, we investigate the infrared band around the band gap 共both above and below the band gap兲 of hydrogen beam-line and plasma implanted silicon. Different broad luminescent peaks near the band gap of Si are observed for beam-line and plasma implanted samples. The physical mechanisms responsible for the broad luminescent peak and the effects of the implantation-induced disorder are discussed.

Silicon-based luminescence1 is of great interest due to possible integration of optoelectronics into the mature silicon integrated circuit technology. Much work has been conducted to investigate low-dimensional structures such as porous Si,2 Si nanodots embedded in SiO2 共Ref. 3兲 and so on or introducing new luminescence centers into Si materials.4 In contrast, there have been fewer reports on lattice disordering of Si materials to alter the light emission efficiency. Cerofolini et al.5 proposed that disorder-induced weak confinement can widen the band gap and also turn the band gap into a directlike one due to relaxation of momentum conservation caused by disorder. However, no further reports can be published possibly due to the difficulty in realizing the exact microstructure of the proposed disorder in the Si lattice. Hydrogen-implanted silicon has recently attracted much interest6 in the ion-cut technology for silicon-on-insulator fabrication.7 In this method, a large dose of hydrogen is implanted into silicon to create a buried layer consisting of microcavities or bubbles to facilitate layer transfer. This hydrogen-implanted region also exhibits interesting optical

II. EXPERIMENT

Boron-doped p-type Si共100兲 with resistivity of 1–20 ⍀ cm was implanted with hydrogen using conventional beam-line and plasma immersion ion implantation 共PIII兲. During beam-line implantation, a 7° tilt was used to minimize channeling and the wafers were not deliberately heated. H⫹ 2 ions were implanted at 25 kV and the doses were 4 ⫻1015 cm⫺2 and 2⫻1016 cm⫺2 for the two different groups of samples. For PIII, the base pressure was 5.6⫻10⫺6 Torr and the working pressure was about 5⫻10⫺4 Torr. Four different sample voltages: ⫺10 kV, ⫺15 kV, ⫺20 kV, and ⫺25 kV were used to produce four sets of samples. The pulse duration and repetition rate in the PIII experiments were 30

a兲

Present address: Department of Basic Science, University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan. b兲 Present address: State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology 共SIMIT兲, Chinese Academy of Sciences 共CAS兲, 865 Changning Rd., 200050 Shanghai, China. c兲 Author to whom correspondence should be addressed; Electronic mail: [email protected] 0021-8979/2004/96(1)/248/4/$22.00

248

© 2004 American Institute of Physics

J. Appl. Phys., Vol. 96, No. 1, 1 July 2004

FIG. 1. PL spectrum vs annealing temperature from beam-line implanted sample 共25 kV, 2⫻1016 cm⫺2 with H⫹ 2 ions兲 measured at 11 K with an argon laser excitation source (␭⫽514.5 nm, power⫽400 mW).

␮s and 200 Hz, respectively. The PIII time was 60 min to minimize sample heating and the final dose was estimated to be about 1016 cm⫺2 based on the current wave form and similar experiments conducted on the same instrument. Compared to conventional beam-line ion implantation, PIII offers advantages such as instrument simplicity, high efficiency, high sample throughput, and easy variation of implantation energy. Hence, with increasing silicon wafer size, PIII is more appealing for some niche applications11 and details about this technique can be found elsewhere.12 The implanted samples were subsequently annealed at 200 °C, 400 °C, or 600 °C for 30 min in nitrogen ambient. For the photoluminescence 共PL兲 studies, the light source was an Ar laser with a wavelength of 514.5 nm and with a power of 400 mW or a He-Cd laser with a wavelength of 325 nm. The PL signal was analyzed using a single monochromator and detected by a liquid nitrogen cooled InSb photo detector connected to a lock-in amplifier. The samples were placed on a cold finger with a continuous He flux cryostat. III. RESULTS AND DISCUSSION

Figure 1 exhibits a typical PL spectrum from the beamline implanted samples as a function of annealing temperature, measured at 11 K with a 514.5 nm argon laser. For the as-implanted sample, a weak PL band below the silicon band gap peaking at ⬃1.4 ␮m is observed. No significant changes can be detected after annealing at 200 °C for 30 min, but annealing at 400 °C results in a blueshift and a new, stronger luminescence peak at a higher energy appears. However, after annealing at 600 °C, the entire luminescence band vanishes. This dependence with annealing temperature is quite similar to those of peaks at ⬃0.78 ␮m, ⬃1.25 ␮m, and ⬃1.5 ␮m reported previously.10 It can be explained as follows. Large dose H implantation into Si creates many kinds of defects, platelets, and also Si-H and/or Si-OH complexes, all of which possibly contribute to the final PL band. During annealing, H interacts with these features and thus modulates the PL band. With increasing annealing temperature 共T兲, the hydrogen becomes more mobile and decorates some point defects. Passivation of the dangling bonds prevents nonradiative recombination at these centers resulting in an increase

An et al.

249

FIG. 2. PL spectrum acquired at 11 K from the 400 °C annealed samples. 共I兲 beam-line, 25 kV, 2⫻1016 cm⫺2 ; 共II兲 beam-line, 25 kV, 4⫻1015 cm⫺2 ; 共a兲 PIII, ⫺10 kV; 共b兲 PIII, ⫺15 kV; 共c兲 PIII, ⫺20 kV, and 共d兲 PIII, ⫺25 kV. The dotted line depicts the single crystal Si band edge emission, which acts as a reference line. The inset gives a Guassian fitting for sample 共d兲. Five peaks are used but the actual peak may have more than five components.

of PL intensity. At T⬎500 °C, H out diffuses and the dangling bonds become nonradiative recombination centers again. Therefore, in Fig. 1, the PL band vanishes in the 600 °C sample. We believe that the PL band observed in our experiments includes two bands at ⬃1.25 ␮m and ⬃1.5 ␮m assigned either to the recombination of carriers at hydrogen stabilized platelet defects13 or to recombination of carriers localized by the strain field present in the damaged regions.14 Our results show that the 400 °C sample exhibits the strongest PL intensity. Figure 2 depicts the results also measured at 11 K for all the samples annealed at 400 °C. For comparison, the PL result from an unimplanted silicon 共virgin Si兲 sample is also given as a dotted line. The spectrum shows two narrow PL bands/lines located at ⬃1134 nm 共strong兲 and 1200 nm 共weak兲. They are due to the boronbound-exciton 关B共BE兲兴 recombination. The dominant emission at 1134 nm comes from momentum-conserving transverse optical (BTO) phonon replicas of the B共BE兲 at 1.093 eV 共1134 nm兲. The one at 1200 nm may originate from several phonon replicas such as BTO⫹O⌫ (O⌫ is the zone-center optical phonon兲 at 1.021 eV and the two-hole transition leaving the acceptor hole in an excited state after recombination, 14 BTO⫹B* h at 1.050 eV. After hydrogen implantation and annealing at 400 °C for 30 min, all samples exhibit a broad PL band near the band gap. The original peaks at 1134 nm and 1200 nm cannot be distinguished clearly because the surface layer has been damaged by ion implantation and the PL signal mainly comes from this surface layer. In Fig. 2, it is obvious that for the beam-line implanted samples, only emission below the band gap can be observed, while emission both below and above 共very near兲 the band gap can be observed from the PIII and annealed samples. In Fig. 2, for beam-line implantation, the lower dose (4⫻1015 cm⫺2 ) sample shows stronger intensity than the higher dose one. It is thus believed that a lower dose creates a smaller number of defects such as dangling bonds resulting in a smaller number of nonradiative recombination centers. For the PIII samples, in addition to the peaks observed for

250

An et al.


FIG. 4. PL spectrum from unimplanted and plasma implanted porous Si 共PS兲 measured at room temperature with a He-Cd laser excitation source (␭⫽325 nm).

FIG. 3. XTEM image of plasma implanted sample at ⫺10 kV. The inset gives the electron diffraction pattern of the surface layer.

the beam-line samples, at least one more PL peak at higher energy 共1.17 eV兲 that is approximately equal to the band gap 共1.166 eV at 0 K兲 of Si can be observed. From the viewpoint of band-gap evolution, our results provide direct evidence of band gap widening5 after PIII. The intensity of this peak increases with increasing bias voltage, whereas the intensity of the peaks below the band gap does not exhibit an obvious monotonic dependence on the bias. In addition, the PL results acquired at different temperatures 共not shown here兲 indicate that all the features in the PL band decrease with increasing measurement temperature. In order to elucidate the mechanisms of the PL results, we deconvolute the PL band into several 共⭓5兲 peaks as shown in the inset of Fig. 2 suggesting that no less than five components contribute to the final PL spectrum. At present, we are not able to exactly assign the origin of each component. Below the band gap, it is believed that the peak at ⬃1360 nm 共⬃0.91 eV兲 comes from the hydrogen stabilized 兵111其 platelet.15 It is believed that the other two peaks below the band gap are due to recombination of carriers at hydrogen stabilized platelet defects or carriers localized by the strain field, as mentioned above. The peak at ⬃1.1 eV originates from phonon replicas of B共BE兲 in our samples. For the peak at 1.17 eV, we analyze its origin according to the difference 共see Ref. 9 and also our previous work兲 between beam-line implantation and PIII. Since voltage pulses are applied in PIII, there are many low energy ions during the rise time and fall time of each sample pulse and they cause substantial damage and even amorphization.16 –19 Figure 3 depicts a cross-sectional transmission electron microscopy 共XTEM兲 image of a plasma implanted sample at ⫺10 kV. Obviously, the surface layer is greatly damaged during implantation as indicated by the electron diffraction pattern. It has two possible effects. First, the band gap is enlarged since the band gap of a-Si 共1.6 eV兲 is higher than that of crystalline Si 共1.1 eV兲. Second, momentum conservation is no longer necessary due to the disordered lattice, resulting in enhanced efficiency of nonphonon emission. Thus,

we attribute the peak at 1.17 eV to nonphonon emission in the heavily damaged surface layer. This agrees well with the fact that a higher bias voltage results in a thicker surface damaged layer. In fact, according to high resolution transmission electron microscopy results,20 the PIII damaged layer is different from an amorphous microstructure that is completely disordered. Although PIII destroys the original long-distance order in the crystal lattice, the final structure retains a certain order or short-distance order and thus can be considered as a medium state between single crystal 共longdistance order兲 and amorphous structure 共disorder兲. This unique microstructure, which agrees well with the so-called ‘‘quantum sieves,’’ 5 produces Anderson localization and hence weak confinement. Consequently the nonphonon emission at about 1.17 eV is greatly enhanced. Elicited by the results described above, we made some porous Si 共PS兲 samples using traditional anodic etching method21 and then performed plasma implantation at ⫺10 kV at a base pressure of 4.6⫻10⫺6 Torr and a working pressure of 5.8⫻10⫺4 Torr. As expected, the PL results in Fig. 4 acquired at room temperature with a He-Cd laser source 共325 nm兲 confirm that the PL band at 550– 850 nm is greatly increased by plasma implantation, whereas the weak band at about 420 nm remains almost unchanged. It is reasonable since the PL band at ⬃420 nm comes from O-related complexes and the strong PL peak at 550– 850 nm originates from nanosize Si in PS.22 This proves that the emission efficiency in these nanosize Si structures is improved by implantation and thus corroborates the conclusion drawn from hydrogen-implanted silicon. IV. CONCLUSION

In summary, in the vicinity of Si band gap 共infrared region兲, hydrogen beam-line ion-implanted Si only exhibits broad PL band below Si band gap, while hydrogen plasma ion-implanted Si shows a much broader PL band at both above and below the band gap. The additional PL component above the band gap comes from the nonphonon emission enhanced by PIII-induced disorder. PIII may thus be a more effective technique to introduce the disorder into Si lattice needed for Anderson localization and weak confinement. Our work suggests an interesting possibility to improve the emis-

An et al.


sion efficiency in silicon/porous silicon by introducing certain disorder through plasma implantation. This may be a complement to previous approaches for developing Si-based light emission, which have mainly been on the reduction of the size of Si such as PS and Si quantum dots, etc. We are conducting more work to further fathom the optical phenomena in plasma implanted Si. ACKNOWLEDGMENTS

The work was supported by Hong Kong Research Grants Council 共RGC兲 Competitive Earmarked Research Grants 共CERG兲 Grant Nos. CityU 1052/02E and HKBU 2154/03E. 1

K. D. Hirschman, L. Tsybeskov, S. P. Duttagupta, and P. M. Fauchet, Nature 共London兲 384, 338 共1996兲. 2 A. Loni, A. J. Simons, P. D. Calcott, and L. T. Canham, J. Appl. Phys. 77, 3557 共1995兲. 3 M. Zhu, G. Chen, and P. Chen, Appl. Phys. A: Solids Surf. 65, 195 共1997兲. 4 N. Q. Vinh, H. Przybylinska, Z. F. Krasil’nik, and T. Gregorkiewicz, Phys. Rev. Lett. 90, 066401 共2003兲. 5 G. F. Cerofolini, L. Meda, D. Bisero, F. Corni, G. Ottaviani, and R. Tonini, Mater. Sci. Eng., B 36, 108 共1996兲. 6 S. J. Pearton, J. W. Corbett, and M. Stavolla, in Hydrogen in Crystalline Semiconductors 共Berlin, Springer, 1991兲. 7 M. Bruel, Nucl. Instrum. Methods Phys. Res. B 108, 313 共1996兲.

8

251

A. V. Mudryi, F. P. Korshunov, A. I. Patuk, I. A. Sharkin, T. P. Larionova, A. Ulyashin, R. Job, W. R. Fahrner, V. V. Emtsev, V. Y. Davydov, and G. Oganesyan, Physica B 308, 181 共2001兲. 9 A. Gulyashin, R. Job, W. R. Fahrner, A. V. Mudryi, A. I. Patuk, and I. A. Shakin, Mater. Sci. Semicond. Process. 4, 297 共2001兲. 10 L. Pavesi, G. Giebel, R. Tonini, F. Corni, C. Nobili, and G. Ottaviani, Appl. Phys. Lett. 65, 454 共1994兲. 11 P. K. Chu, X. Lu, S. S. K. Iyer, and N. W. Cheung, Solid State Technol. 40, S9 共1997兲. 12 P. K. Chu, S. Qin, C. Chan, N. W. Cheung, and L. A. Larson, Mater. Sci. Eng., R. R17„6-7…, 207 共1996兲. 13 N. M. Johnson, in Semiconductors and Semimetals, edited by R. K. Willardson and A. C. Beer 共Academic, New York, 1991兲, Vol. 34, p. 113. 14 H. Weman, J. L. Lindstrom, G. S. Oehrlein, and B. G. Svensson, J. Appl. Phys. 67, 1013 共1990兲. 15 K. H. Hwang, J. W. Park, E. Yoon, K. W. Whang, and J. Y. Lee, J. Appl. Phys. 81, 74 共1997兲. 16 L. W. Wang, R. K. Y. Fu, X. C. Zeng, P. K. Chu, W. Y. Cheung, and S. P. Wong, J. Appl. Phys. 90, 1735 共2001兲. 17 P. K. Chu, Surf. Coat. Technol. 156, 244 共2002兲. 18 P. K. Chu, Plasma Phys. Controlled Fusion 45, 555 共2003兲. 19 P. K. Chu, J. Vac. Sci. Technol. B 22, 289 共2004兲. 20 W. L. Liu, in Doctoral thesis, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Science, 2002. 21 W. L. Liu, M. Zhang, C. L. Lin, Z. M. Zeng, L. W. Wang, and P. K. Chu, Appl. Phys. Lett. 78, 37 共2001兲. 22 V. A. Joshkin, M. N. Naidenkov, V. N. Pavlenko, A. V. Kvit, and S. R. Oktyabrsky, Phys. Rev. B 52, 12102 共1995兲.