Effect of different preparation conditions on light emission from silicon implanted SiO2 layers G. Ghislottia) and B. Nielsen Department of Applied Science, Brookhaven National Laboratory, Upton, New York 11973
P. Asoka-Kumar and K. G. Lynn Department of Physics, Brookhaven National Laboratory, Upton, New York 11973
A. Gambhir and L. F. Di Mauro Department of Chemistry, Brookhaven National Laboratory, Upton, New York 11973
C. E. Bottani Dipartimento di Ingegneria Nucleare, Politecnico di Milano, 20133 Milano, Italy
~Received 11 December 1995; accepted for publication 15 February 1996! Visible light emission from Si1 implanted SiO2 layers as a function of different annealing conditions ~temperature, time and ambient! is studied. It is shown that a 560 nm band, present in as implanted samples, increases its intensity for increasing annealing temperatures and is still observed after annealing at 1000 °C. The emission time is fast ~0.5–2 ns!. A second band centered at 780 nm is detected after annealing at 1000 °C. The intensity of the 780 nm band further increases when hydrogen annealing was performed. The emission time is long ~1ms–0.3 ms!. Based on the annealing behavior and on the emission times, the origin of the two bands is discussed. © 1996 American Institute of Physics. @S0021-8979~96!09910-7# I. INTRODUCTION
Visible light emission from silicon based nanostructures has attracted interest since the observation of luminescence from porous silicon ~po-Si!.1 Several alternative methods have been proposed to produce silicon nanocrystals ~Si nc): plasma deposition from silane,2 gas evaporation,3 and termination of glass-melt reaction.4 The use of po-Si presented many problems in practical applications ~instability of light emission with spontaneous oxidation,5 fragile mechanical properties!, and the other techniques were found not appropriate for the production of integrated devices. The production of silicon nanocrystals using ion implantation in glass could overcome most of the problems encountered to obtain room temperature visible luminescence from silicon based devices. The choice of a glassy matrix reduces self-absorption problems and improves mechanical and thermal stability. Shimizu-Iwayama et al.6 presented data of room temperature luminescence from 1 MeV Si1 implanted fused silica. They observed a band around 620 nm ~green band! ascribed to defects inside silica and a second one around 730 nm ~red band! which was related to the presence of Si nanocrystals. In particular they proposed that the luminescent center is at the Sinc –SiO2 interface. Komoda et al.7 observed a band centered at 600 nm in low fluence implanted samples and a band at 800 nm in samples implanted at higher fluences and annealed for longer time. They interpreted this behavior as a red shift related to an increase in the size of silicon nanocrystals. Mutti et al.8 in 160 keV Si1 implanted SiO2 layers observed a distinct green light emission on a sample implanted at 131017 cm22 and annealed at high temperature ~T.1000 °C!. A near infrared light emission band was detected in high fluence implant ~at 331017 cm22 dose!. a!
Electronic mail:
[email protected]
8660
J. Appl. Phys. 79 (11), 1 June 1996
The same authors, in preliminary time-resolved measurements, showed9 that the two bands have very different emission times, the green band being characterized by fast ~0.4–6 ns! while the near infrared band has slow (. 250 ns! emission times. A complete investigation of the kinetics of formation of the two emission bands as a function of different preparation conditions and of their emission times has not been conducted so far. This study could provide a better understanding of the physical parameters which influence the luminescence mechanism and, at the same time, it can also be of practical interest to improve the photoluminescence ~PL! efficiency in view of possible optoelectronics applications. Moreover, since the observation of red and blue-light bands presents striking similarities with the emission from po-Si, Si nc produced by implantation could also help to clarify the physical process responsible for the light emission in po-Si. In fact, in spite of the many theoretical and experimental studies this mechanism is still not clear. Quantum confinement within Sinc , 1,10,11 emission from molecular species such as siloxene,12 and recombination near the Si nc –SiO2 interface,13,14 have been proposed as possible mechanisms. To address some of these issues, in this paper continuous and time-gated photoluminescence studies of Si1 implanted SiO2 layers were performed starting from the as-implanted samples. It is shown that a 560 nm band, present in asimplanted samples, increases its intensity for increasing annealing temperatures and is still observed after a 1000 °C annealing. The 780 nm band, detected only after 1000 °C annealing, depends on the annealing conditions and it increases its intensity if an atmosphere containing hydrogen is used. II. EXPERIMENT AND RESULTS
Samples were prepared by implantation of 160 keV Si1 ions into 430 nm thick SiO2 layers thermally grown on
28
0021-8979/96/79(11)/8660/4/$10.00
© 1996 American Institute of Physics
Downloaded¬04¬Oct¬2005¬to¬128.146.34.155.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://jap.aip.org/jap/copyright.jsp
FIG. 1. Photoluminescence spectra for a sample implanted at a dose of 231017 cm22 ~a!, and after 500 °C ~b! and 700 °C 30 min vacuum annealing ~c!.
a ~100! oriented p-type Si substrate. Fluences ranged from 331016 cm22 to 331017 cm22. Vacuum annealing was performed in a high-vacuum chamber ~1027 Torr! and the sample was mounted on a resistively heated tantalum foil. Gas annealing was done in a furnace using high purity gases. In the case of hydrogen annealing a 15% H2 , 85% Ar mixture was used. Photoluminescence measurements were performed using 464 nm excitation light from an excimer pumped dye laser. Emitted light was analyzed with a monochromator and detected using an intensified charge coupled device ~ICCD! detector. Time-gated spectra with a temporal resolution of about 30 ns were collected using an external pulse generator to control the acquisition time of the ICCD detector. Nanosecond time-resolved spectroscopy was performed with a mode-locked krypton ion laser oscillating at 413 nm. The experimental setup was described in detail elsewhere.15 Figure 1 represents photoluminescence ~PL! spectra for a sample implanted at a dose of 231017 cm22 and after vacuum annealing at 500 °C and 700 °C for 30 min. On the as-implanted sample a broad band around 550–600 nm is observed. The intensity of this band around 560 nm increases as a function of the annealing temperature up to 700 °C. No band around 750–850 nm is detected. To check the effect of annealing gas, annealing in Ar, N2 and H2 were also performed. Up to 700 °C the PL spectra did not show any difference for different annealing atmospheres. Three samples implanted at 231017 cm22 and annealed in vacuum at 700 °C for 30 min, were further annealed at 1000 °C in Ar, He and N2 . A near-infrared band is observed after 1000 °C annealing. In Fig. 2 the PL spectra for samples annealed for 30 min in Ar and N2 are reported ~the spectrum for He annealing is similar to that for Ar annealing!. The PL intensity for the red band grows for increasing annealing time ~see Fig. 3!. To study the effect of hydrogen, samples already annealed in Ar and N2 ~shown in Figs. 2~a! and 2~b!, were subsequently annealed in H2 atmosphere at 500 °C. Figure 4 shows PL spectra for a sample annealed in argon and after 30 min hydrogen annealing. The intensity in the near-infrared J. Appl. Phys., Vol. 79, No. 11, 1 June 1996
FIG. 2. PL spectra for the sample already annealed in vacuum ~reported in Fig. 1!, after 1000 °C in argon ~a! and nitrogen ~b! annealing were performed.
spectral region increases after hydrogen annealing and saturates after about 3 of hours of annealing ~see Fig. 5!. Figure 6 presents gated PL spectra for a sample implanted at 231017 cm22 and annealed for 30 min in argon at 1000 °C. The green emission is characterized by short emission times, about 2 ns ~see the inset in Fig. 6~a!!. For gates of 100 ns–1m s the emission is centered at about 700 nm. When the time gate is increased the PL emission is centered at a longer wavelength ~780 nm for a gate of 1 m s–10m s, 820 nm when the gate is in the range 10 m s–0.3 ms!. III. DISCUSSION
Results reported in Figs. 1 and 3 show that a green band is observed on the as-implanted sample and also after 1000 °C annealing and its intensity increases as a function of the annealing temperature. The intensity reaches its maximum after 1000 °C annealing. For increasing annealing time at 1000 °C, the green band intensity decreases ~Fig. 3!. The emission times for the green band are in the nanosecond range.
FIG. 3. PL emission from a sample implanted at 331017 cm22 fluence and annealed in nitrogen at 1000 °C for different times: 30 min ~a! and 5 h ~b!. Ghislotti et al.
8661
Downloaded¬04¬Oct¬2005¬to¬128.146.34.155.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://jap.aip.org/jap/copyright.jsp
FIG. 4. Effect of hydrogen annealing. A sample annealed for 30 min in argon at 1000 °C ~a! was subsequently annealed in hydrogen at 500 °C ~b!. While the band centered at 560 nm slightly changes, the band around 750 nm shows a strong increase in intensity.
A green-blue band in silicon nanocrystals inside a SiO2 matrix was observed3,4 and related to defects ~twofold coordinated Si atoms bonded to two oxygen atoms! inside the oxide.16 These defects are characterized by radiative times of the order of m s,17 and their number decreases as a function of the annealing temperature. The increase in intensity as a function of the annealing temperature, the observation of the same band after 1200 °C annealing,8 and the fast decay time seem to exclude the possibility that this band originates from the point defects mentioned above. The present results could be consistently explained assuming that after implantation in the Si-rich region corresponding to the end of range extended defects like clusters or chain of silicon are formed. These structures can act as radiative sites. The PL intensity increase as a function of the annealing temperature could be explained in terms of annealing of defects that can give rise to non-radiative recombination, while no significant Si rearrangement takes place below 1000 °C due to the low silicon diffusivity.18 The intensity decrease as a function of the annealing time at 1000 °C ~Fig. 3! is related to the
FIG. 5. Intensity of the PL signal detected at around 750 nm as a function of the annealing time in hydrogen for samples prepared by annealing at 1000 °C in argon ~full squares! and nitrogen ~open circles! a sample implanted at a dose of 231017 cm22. 8662
J. Appl. Phys., Vol. 79, No. 11, 1 June 1996
FIG. 6. Time-gated spectra of 231017 cm22 fluence implant after annealing at 1000 °C for 30 min in Ar. Time-gates are 0–100 ns ~a!, 100–1 m s ~b!, 1–10 m s ~c! and 10–200 m s ~d! after the laser pulse. The inset in ~a! refers to a nanosecond resolved time-gated spectrum ~time gate of 0–2 ns!.
growth process of Sinc ~as will be discussed later on! and therefore to the disappearance of these precursors. This picture is consistent with the shell model for SiOx 19 for which Si clusters are embedded inside SiO2 . Since the radiative times for the emission at 560 nm are fast, the luminescence intensity is not influenced by nonradiative processes which are usually characterized by capture rates of the order of ms.20 This may explain why the intensity is not affected by hydrogen annealing. Kanemitsu et al.21 reported that linear or ladder structures of Si atoms can give a luminescence signal in the spectral region around 500 nm and they have radiative times of the order of ns, while cubic silicon clusters have radiative times of the order of ms.22 Moreover Si clusters are greatly influenced by the presence of non-radiative centers at their surface.22,23 While based on annealing temperature behavior both small silicon clusters and Si-chain or ladder structures can be responsible for the 560 nm band, the fast lifetime and the insensitivity to annealing atmosphere ~argon, nitrogen, hydrogen! seem to indicate that the latter are the most probable candidates. The observation of a band in Si implanted SiO2 layers has also been reported also by Shimizu-Iwayama et al. in 1 MeV Si1 implanted SiO2 . 6 They observed an intensity decrease after thermal annealing at 600 °C which leads them to relate this band to defects inside a Si-rich SiO2 . This discrepancy can be explained if we consider that at a fixed dose, the local Si concentration decreases as a function of the implantation energy, and therefore, if MeV ions are used, a dose of 231017 cm22 can be not high enough to produce Si structures after implantation. In fact, in our case, samples implanted at a fluence of 331016 cm22 do not show green Ghislotti et al.
Downloaded¬04¬Oct¬2005¬to¬128.146.34.155.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://jap.aip.org/jap/copyright.jsp
emission after 1000 °C annealing, 131017 cm22 being the critical fluence to observe this band. This is in agreement with the previous results.8 The behavior observed for the green light emission presents some similarities with results recently obtained by Augustine et al.24 on amorphous silicon oxynitrides produced by plasma-enhanced chemical vapor deposition. They reported a fast ~less than 10 ns! green emission which does not depend on annealing atmosphere and which shows anologies with organosilane chemical compounds.25 When the annealing temperature is raised to ~or above! 1000 °C SiOx becomes unstable26,27 and Si precipitates are known to be formed depending on the local level of supersaturation, thermal history, and annealing time.18,28,29 Transmission electron microscopy observations made on samples implanted at the same doses showed the presence of Si nanocrystals inside the SiO2 matrix having average dimensions of 5 nm.8 Similar results were obtained by Komoda et al.7 Therefore the observation of a red band is related to the formation of bigger Si precipitates ~although not necessarily related to a recombination within the Sinc). When the annealing time was increased the intensity of the red band increases with respect to the green one ~Fig 3!. This is consistent with a Ostwald ripening process in which smaller Si structures disappear and bigger ones are formed. If a diffusion-controlled growth is assumed, increasing the annealing time from 30 min to 5 hours at 1000 °C would change the mean radius from 4 nm to 7–9 nm.18 Accordingly, if the near infrared emission comes from recombination inside Sinc , we would expect a wavelength shift. Such a shift is not observed ~Fig. 3! which seems to exclude quantum confinement effects. The near-infrared emission is characterized by times from 1 m s to 0.3 ms. These values are similar to those obtained for porous silicon.30,31 In the case of the near-infrared band the temporal scale is comparable to that of nonradiative recombinations, hence centers like Si dangling bonds can represent important non-radiative channels. Lannoo et al.23 showed that even a few Si dangling bonds can completely quench the PL emission from a Sinc . The intensity increase, observed after hydrogen annealing for both Ar and N 2 annealed samples ~Fig. 4 and Fig. 5!, can be explained based on this point. As a matter of fact hydrogen is known to passivate the Si–SiO2 interface.32,33 A similar mechanism can be effective in the case of the Sinc –SiO2 interface too. IV. CONCLUSIONS
SiO2 layers implanted with Si1 ions at doses of 331016 cm –331017 cm22 are studied under different annealing conditions. Two emission bands are detected. The first centered at 560 nm is present in as-implanted samples, increases its intensity as a function of annealing temperature, and does not change appreciably for different annealing atmosphere. A second band around 750 nm is detected after 1000 °C annealing was performed. The fast emission time and the annealing behavior suggest that the green emission is related 22
J. Appl. Phys., Vol. 79, No. 11, 1 June 1996
to extended defects like silicon chains, while the nearinfrared light is related to the presence of silicon precipitates. ACKNOWLEDGMENTS
We gratefully acknowledge Dr. L. Meda, Dr. G. F. Cerofolini and Dr. P. Mutti for many helpful discussions. The research at the Chemistry Department of Brookhaven National Laboratory was carried out under Contract No. DEAC02-76CH00016 with the U.S. Department of Energy and supported by its Division of Chemical Science, office of Basic Energy Sciences. The work was supported by U.S. Department of Energy under Contract No. DE-AC0276CH00016. L. T. Canham, Appl. Phys. Lett. 57, 1046 ~1990!. T. Kawaguchi and S. Miyamiza, Jpn. J. Appl. Phys. 32, L215 ~1993!. 3 H. Morisaki, H. Hashimoto, F. W. Ping, H. Nozawa, and H. Ono, J. Appl. Phys. 74, 2977 ~1993!. 4 S. H. Risbaud, L. C. Liu, and J. F. Shackelford, Appl. Phys. Lett. 63, 1648 ~1993!. 5 M. A. Tichler, R. T. Collins, J. H. Satish, and J. C. Tsang, Appl. Phys. Lett. 60, 239 ~1992!. 6 T. Shimizu-Iwayama, K. Fujita, S. Nakao, K. Saitoh, T. Fujita, and N. Itoh, J. Appl. Phys. 75, 7779 ~1994!. 7 T. Komoda, J. Kelly, F. Cristiano, A. Nejim, P. L. F. Hemment, K. P. Homewood, R. Gwilliam, J. E. Mynard, and B. J. Sealy, Nucl Instrum. Methods B 96, 387 ~1995!. 8 P. Mutti, G. Ghislotti, S. Bertoni, L. Bonoldi, G. F. Cerofolini, L. Meda, E. Grilli, and M. Guzzi, Appl. Phys. Lett. 66, 851 ~1995!. 9 P. Mutti, G. Ghislotti, L. Meda, E. Grilli, M. Guzzi, L. Zanghieri, R. Cubeddu, A. Pifferi, P. Taroni, and A. Torricelli, Mater. Res. Symp. Proc. ~to be publishd!. 10 V. Lehmann and U. Gosele, Appl. Phys. Lett. 58, 856 ~1991!. 11 B. Delley and E. F. Steigmeier, Phys. Rev. B 47, 1397 ~1993!. 12 M. S. Brandt, H. D. Fuchs, M. Stutzmann, J. Weber, and M. Cardona, Solid State Commun. 81, 307 ~1991!. 13 F. Koch, Mater. Res. Soc. Symp. Proc. 298, 319 ~1993!. 14 S. M. Prokes and O. J. Glembocki, Phys. Rev. B 49, 2238 ~1994!. 15 R. Cubeddu, F. Docchio, W. Q. Liu, R. Ramponi, and P. Taroni, Rev. Sci. Instrum. 59, 2254 ~1988!. 16 L. Skuja, J. Non-Cryst. Solids 149, 77 ~1992!. 17 L. Skujia, Solid State Commun. 84, 613 ~1992!. 18 L. A. Nesbit, Appl. Phys. Lett 46, 38 ~1985!. 19 P. Bruesch, Th. Stockmeier, F. Stucki, and P. A. Buffat, J. Appl. Phys. 73, 7677 ~1993!. 20 D. Goguenheim and M. Lannoo, Phys. Rev. B 44, 1724 ~1991!. 21 Y. Kanemitsu, K. Suzuki, S. Kyushin, and H. Matsumoto, Phys. Rev. B 51, 13103 ~1995!. 22 Y. Kanemitsu, K. Suzuki, M. Kondo, S. Kyushin, and H. Matsumoto, Phys. Rev. B 51, 10666 ~1995!. 23 M. Lannoo, C. Delerue, and G. Allan, J. Lumin. 57, 249 ~1993!. 24 B. H. Augustine, E. A. Irene, Y. J. He, K. J. Price, L. E. McNeil, K. N. Christensen, and D. M. Maher, J. Appl. Phys. 78, 4020 ~1995!. 25 S. Kyushin, H. Matsumoto, Y. Kanemitsu, and M. Goto, J. Phys. Soc. Jpn. 63, 46 ~1988!. 26 D. Dong, E. A. Irene, and D. R. Young, J. Electron. Soc. 125, 819 ~1978!. 27 F. Rochet, G. Dufar, H. Roulet, B. Pelloie, J. Perrie`re, E. Fogarassy, A. Slaoui, and M. Froment, Phys. Rev. B 37, 6468 ~1988!. 28 C. Jaussand, J. Margail, J. Stoemenos, and M. Bruel, Mater. Res. Soc. Symp. Proc. 107, 17 ~1988!. 29 S. Mantl, Mater. Sci. Rep. 8, 1 ~1992!. 30 J. C. Vial, A. Bsiesy, F. Gaspard, R. Herino, M. Ligeon, F. Muller, R. Romestain, and R. M. Macfarlane, Phys. Rev. B 45, 14171 ~1992!. 31 Y. H. Xie, W. L. Wilson, F. M. Ross, J. A. Mucha, E. A. Fitzgerald, J. M. Macaulay, and T. D. Harris, J. Appl. Phys. 71, 2403 ~1992!. 32 K. L. Brower, Phys. Rev. B 42, 3444 ~1990!. 33 M. L. Reed, Semicond. Sci. Technol. 4, 980 ~1989!. 1 2
Ghislotti et al.
8663
Downloaded¬04¬Oct¬2005¬to¬128.146.34.155.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://jap.aip.org/jap/copyright.jsp