May 1, 2002 / Vol. 27, No. 9 / OPTICS LETTERS 713
Nanoparticle-modified metal–oxide–silicon
structure enhancing silicon band-edge
electroluminescence to near-lasing action
Ching-Fuh Lin, Peng-Fei Chung, and Miin-Jang Chen
Graduate Institute of Electro-Optical Engineering and Department of Electrical Engineering, National Taiwan University, Taipei 106, Taiwan
Wei-Fang Su
Institute of Materials Science and Engineering, National Taiwan University, Taipei 106, Taiwan
Received September 10, 2001
With the insertion of SiO2 nanoparticles in the oxide layer, near-lasing actions such as threshold behavior
and resonance modes are observed at the Si bandgap energy of metal – oxide – silicon (MOS) structure. The threshold current is⬃12 mA. The SiO2 nanoparticles cause simultaneous localization of electrons and holes
to enhance phonon-assisted radiative recombination. Electroluminescence at Si bandgap energy is increased to orders of magnitude larger than in similar MOS structures without SiO2 nanoparticles. The efficient
light emission at the Si bandgap energy indicates that a direct bandgap nature is not necessarily the basic requirement for radiative recombination. © 2002 Optical Society of America
OCIS codes: 230.3670, 140.5960, 140.2020, 160.3380, 230.0250.
Because of the mature ultralarge-scale-integration technology based on Si, Si has been the most impor-tant material for the integrated-circuit industry and is also attractive for applications in the emerging area of optoelectronics. Unfortunately, the indirect-bandgap nature of Si makes it ineff icient as a light source, so op-toelectronic integration on the Si chip is a challenging issue. Many efforts have thus been devoted to over-coming the indirect-bandgap obstacle. Those efforts include use of porous-silicon-based devices,1,2 doping
impurities into Si,3 – 5
use of Si1-implanted SiO 2,6
alloying Si with Ge or Sn,7,8
use of nanocrystalline Si (Refs. 9 –11) or quantum-conf inement structures,12
deposition of polymer upon Si substrates,13
and growth of GaN upon Si substrates.14
Either the Si crystal is signif icantly modified or materials other than Si are used for light emission, so the emissions are not related to the Si bandgap energy. Recently, efficient light emission at the Si bandgap energy was achieved by introduction of dislocation loops or texturing sili-con.15,16
Here we report that, by use of SiO2
nanopar-ticles in the oxide layer in metal–oxide– semiconductor (MOS) structures, the differential quantum efficiency of electroluminescence (EL) at the Si bandgap en-ergy could be enhanced to nearly 1024, orders
of magnitude larger than that of similar MOS struc-tures without SiO2 nanoparticles.17,18 Moreover,
near-lasing actions at the Si bandgap energy, such as threshold behavior and resonance modes, are observed at room temperature.
The main hindrance to light emission from indirect-bandgap semiconductors is the momentum mismatch between electrons and holes, which leads to the low radiative recombination rate of electrons and holes. Several mechanisms for overcoming the momentum mismatch between electrons and holes in indirect-bandgap materials have been proposed.17
One of them is the spatial localization of carriers,
which could have two effects: One is to cause the spread of carrier momentum as a result of the uncer-tainty principle and thus to relax the momentum mis-match between electrons and holes. The other is to cause the easy formation of excitons such that phonons can more easily be involved in the electron–hole radia-tive recombination. Either mechanism will change the optical matrix element to enhance the radia-tive recombination in indirect-bandgap materials. Lifetime measurement of electroluminescent MOS tunneling diodes has shown that the ratio for Shock-ley–Read –Hall, radiative, and Auger recombinations can be 1:0.196:0.096.19
Therefore the difference between the radiative recombination rate and the nonradiative recombination rate for indirect-bandgap materials can be less than 1 order of magnitude.
According to the theoretical model20
that explains the spectrum of electroluminescence from a MOS tunnel-ing diode, phonons and excitons are both involved in radiative recombination. As an electron and a hole form an exciton, the probability of radiative recombina-tion is increased. The reason for this is that the pro-cess is now more like a two-particle collision (phonon and electron –hole pair) than a three-particle (electron, hole, and phonon) collision. However, even if an elec-tron and a hole do not form an exciton but are spatially confined together, they still have an increased proba-bility of radiative recombination.
To spatially confine electrons and holes together, we used SiO2 nanoparticles in the insulation layer of
the MOS structure. The processing steps were as fol-lows: First the native oxide upon the Si wafer was removed. Then the solution with SiO2 nanoparticles
was spun onto the wafer and dried in an oven. Fig-ure 1 shows a field-emission type of scanning-electron microscope photo of SiO2nanoparticles spun upon a Si
wafer. Nanoparticles with a feature size of ⬃12 nm are shown. After the nanoparticles were deposited, a
714 OPTICS LETTERS / Vol. 27, No. 9 / May 1, 2002
Fig. 1. Scanning-electron microscope photos of SiO2 nanoparticles spun upon a Si surface.
thin layer of Al metal was deposited upon the nanopar-ticles by evaporation. Then Ag paint was applied on top of the Al to hold a Au wire for electrical contact. A thick layer of Al was also evaporated onto the back side of the Si to form another electrical contact.
A schematic of the device is shown in Fig. 2. Be-cause the spun SiO2 nanoparticles are not tightly
packed, there are many hollow spaces between those particles. During evaporation, the Al atoms can get into those hollow spaces, so Al metal comes very close to the Si surface, which makes the structure similar to a metal– insulator – silicon structure. The measured I– V characteristics show that the device behaves similarly to a MOS tunneling diode. The measured EL spectrum is similar to previously reported spec-tra from MOS upon Si without nanoparticles,20,21
indicating that similar physics is involved in the radiative recombination. However, the measured efficiencies at Si bandgap energy are much larger than those previously reported for MOS structures without nanoparticles.17,18,21
More than 100 devices were measured. The differential quantum efficiency of measured EL from the periphery of the Ag paint pad on the devices varied from 1 3 1026 to 1 3 1024
at an injection current of 50 mA. This eff iciency does not take into account light blocked by the thick Ag paint. If the blocked light is also included, the best efficiency should be far beyond 1024.
The temporal response of EL was measured for sev-eral samples as a square-current pulse was injected with a scanning-gate integrator –boxcar averager sys-tem.19
The measurements show that the rise time varies from 4.5 to 20 ms and the fall time varies from 1.5 to 15 ms. Also, the longer rise time corresponds to a longer fall time and stronger EL emission. The non-radiative recombination states usually dominate the transition of electrons from the conduction band to the valence band and thus strongly inf luence the tempo-ral response. More nonradiative recombination states speed up the recombination process and thus the tem-poral response. In the meantime, they also reduce the radiative emission. The varying rise and fall times in-dicate that those samples experience different amounts
of nonradiative recombination states and so have dif-ferent EL efficiencies. For 20 ms of rise time the non-radiative recombination rate is no more than 5 3 104兾s,
implying that some portion of the Si wafer has good quality and could possibly serve for eff icient radiative recombination.
With the enhanced radiative recombination rate, threshold behavior and resonance modes were ob-served. The light – current curve [curve (a) of Fig. 3] shows a sudden increase in output power after the threshold current 共⬃12 mA兲. The corresponding spectrum at 50 mA is shown in Fig. 4(a). Some of the resonance modes are clearly shown in Fig. 4(a). Both measurements were made for cw operation at room temperature. For comparison, the light–current curve of another device without threshold is shown by curve (b) in Fig. 3. Its output power has only a gradual increase with the injection current. Its corresponding spectrum is shown in Fig. 4(b). No resonance modes were observed for the devices without threshold behavior.
The output power shown in Fig. 3 was measured with the detector directly facing the device. There-fore the threshold shown in curve (a) of Fig. 3 is the behavior of the total output power, not only of the power of the sharp curves shown in Fig. 4(a). Because our monochromator (CVI, CM110) does not have suf-ficient resolution, only some resonance modes can be seen clearly. As a result, the spectrum of Fig. 4(a) appears to be a superposition of sharp spikes with a
Fig. 2. Schematic of the SiO2nanoparticle-modif ied MOS structure.
Fig. 3. Room-temperature light – current curves: (a) the device with threshold, (b) the device without threshold.
May 1, 2002 / Vol. 27, No. 9 / OPTICS LETTERS 715
Fig. 4. Measured spectra: (a) with resonance modes, (b) without resonance modes. The spectra correspond to the Si bandgap energy.
broad background. The background in fact consists of many resonance modes that are not resolved by the monochromator.
The SiO2 nanoparticles cause the thickness of the
insulation layer between the metal and Si to be nonuni-form, as shown schematically in Fig. 2. The tunneling probability is greater, and band bending of Si is more severe, for the region with a thin insulation layer than in the thick region, so electrons and holes are spatially confined together near the Si SiO2interface.
There-fore radiative recombination is enhanced, as explained above. Also, the presence of SiO2nanoparticles could
lead to a local increase of the current density and so enhance radiative recombination. However, the local current density cannot be increased for orders of mag-nitude and so should not be the main reason for the enhanced EL.
Because SiO2 nanoparticles induce no damage
to the Si crystal, nonradiative centers remain low for the portion of a Si wafer that has good quality. The near-lasing actions occur probably because the optimal enhancement of radiative recombination by the SiO2nanoparticles and the minimal nonradiative
recombination states coincidentally occur at the same spot on the Si wafer, resulting in signif icant gain. Also, the high ref lection of the top Ag-paint pad and the bottom Al layer provides the optical feedback for resonance. In addition, the small absorption of Si at the bandgap energy causes the gain possibly to surpass the absorption loss of the substrate, leading to the near-lasing actions.
In conclusion, EL at the Si bandgap energy from a MOS structure is greatly enhanced by use of SiO2
nanoparticles for the oxide layer. The reason is that the SiO2nanoparticles cause electrons and holes to be
localized, making the radiative recombination of an electron –hole pair more like a two-particle collision than a three-particle collision. The measured differ-ential quantum eff iciency of EL could be near 1024,
even with signif icant light blocked by the Ag paint. In addition, for the first time to our knowledge, near-lasing actions such as threshold behavior and reso-nance modes that correspond to Si bandgap energy have been observed.
This research is supported in part by National Science Council, Taipei, Taiwan, under contract NSC90-2622-L-002-002. C.-F. Lin’s e-mail address is cf [email protected].
References
1. A. G. Cullis and L. T. Canham, Nature 353, 335 (1991). 2. M. V. Wolkin, J. Jorne, P. M. Fauchet, G. Allan, and
C. Delerue, Phys. Rev. Lett. 82, 197 (1999).
3. G. Franzo, F. Priolo, S. Coffa, A. Polman, and A. Carn-era, Appl. Phys. Lett. 64, 2235 (1994).
4. M. Matsuoka and S. Tohno, Appl. Phys. Lett. 71, 96 (1997).
5. S. G. Pavlov, R. Kh. Zhukavin, E. E. Orlova, N. V. Shastin, A. V. Kirsanov, H.-W. Hubers, K. Auen, and H. Riemann, Phys. Rev. Lett. 84, 5220 (2000).
6. W. Skorupa, R. A. Yankov, I. E. Tyschenko, H. Frob, T. Bohme, and K. Leo, Appl. Phys. Lett. 68, 2410 (1996).
7. S. Oguz, W. Paul, T. F. Deutsch, B.-Y. Tsaur, and D. V. Murphy, Appl. Phys. Lett. 43, 848 (1983). 8. C. W. Liu, J. C. Sturm, Y. R. J. Lacroix, M. L. W.
Thewalt, and D. D. Ferovic, Mater. Res. Soc. Symp. Proc. 342, 37 (1994).
9. D. B. Geohegan, A. A. Puretzky, G. Duscher, and S. J. Pennycook, Appl. Phys. Lett. 73, 438 (1998).
10. L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franz, and F. Priolo, Nature 408, 440 (2000).
11. M. H. Nayfeh, N. Barry, J. Therrien, O. Akcakir, E. Gratton, and G. Belomoin, Appl. Phys. Lett. 78, 1131 (2001).
12. F. Buda, J. Kohanoff, and M. Parrinello, Phys. Rev. Lett. 69, 1272 (1992).
13. D. R. Baigent, R. N. Marks, N. C. Greenham, R. H. Friend, S. C. Moratti, and A. B. Holmes, Appl. Phys. Lett. 65, 2636 (1994).
14. R. Birkhahn and A. J. Steckl, Appl. Phys. Lett. 73, 1700 (1998).
15. W. L. Ng, M. A. Lourenco, R. M. Gwilliam, S. Ledain, G. Shao, and K. P. Homewood, Nature 410, 192 (2001). 16. M. A. Green, J.-H. Zhao, A.-H. Wang, P. J. Reece, and
M. Gal, Nature 412, 805 (2001).
17. C.-F. Lin, C. W. Liu, M.-J. Chen, M. H. Lee, and I. C. Lin, J. Appl. Phys. 87, 8793 (2000).
18. C. W. Liu, M. H. Lee, M.-J. Chen, I. C. Lin, and C.-F. Lin, Appl. Phys. Lett. 76, 1516 (2000).
19. M.-J. Chen, C.-F. Lin, M. H. Lee, S. T. Chang, and C. W. Liu, Appl. Phys. Lett. 79, 2264 (2001).
20. M.-J. Chen, E.-Z. Liang, S.-W. Chang, and C.-F. Lin, J. Appl. Phys. 90, 789 (2001).
21. M.-J. Chen, C.-F. Lin, W. T. Liu, S. T. Chang, and C. W. Liu, J. Appl. Phys. 89, 323 (2001).
