The Resident Season 2 Episode 2 | Isle of Dogs (2018) | Bolo Yeung

Relation between electroluminescence and photoluminescence in porous silicon

Relation between electroluminescence and photoluminescence in porous silicon

Materials Science and Engineering B72 (2000) 138 – 141 www.elsevier.com/locate/mseb Relation between electroluminescence and photoluminescence in por...

179KB Sizes 0 Downloads 2 Views

Materials Science and Engineering B72 (2000) 138 – 141 www.elsevier.com/locate/mseb

Relation between electroluminescence and photoluminescence in porous silicon E. Savir a,*, J. Jedrzejewski a, A. Many a, Y. Goldstein a, S.Z. Weisz b, M. Gomez b, L.F. Fonseca b, O. Resto b b

a Racah Institute of Physics, The Hebrew Uni6ersity, Jerusalem 91904, Israel Department of Physics, Uni6ersity of Puerto Rico, Rio Piedras PR 00931, USA

Abstract We present combined measurements of electroluminescence (EL) and photoluminescence (PL) in p-type porous silicon. The EL spectra were measured using an electrolyte contact for electron injection into the porous face of the sample. Upon applying the current, the EL intensity first rises with time, reaches a maximum, and then decays to zero. (The whole process takes about half an hour.) At the same time, the peak of the EL spectrum shifts from : 850 nm in the beginning to : 600 nm at the end of the process. The PL, which was measured simultaneously, peaked at : 750 nm in the beginning and was much wider than all of the EL spectra. Towards the end of the EL process, the red part of the PL spectrum practically disappears. This shifts the PL peak towards the blue, to about the same wavelength as the EL peak ( : 600 nm) and the spectrum becomes much narrower, comparable to the EL spectrum. The voltage across the sample during the EL process shows a moderate increase up to the point where the EL disappears, and then the voltage rises steeply. This behavior is associated with the build-up of a thin oxide layer on the porous surface. The combined results of EL and PL, and especially the disappearance of the red part in the photoluminescence spectrum at the end of the EL process, suggest that in addition to quantum confinement, localized surface states play an important role in the luminescence process, at least in the red part of the spectrum. Such states may be associated with adsorbed species and disappear upon oxidation. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Electroluminescence; Photoluminescence; Porous silicon; Localized surface states

1. Introduction Porous silicon [1 – 4], obtained by electrochemical etching procedures applied to crystalline silicon surfaces, exhibits high luminescence efficiencies in the visible range. It is generally accepted that the visible luminescence originates from the band-gap enlargement due to quantum confinement in the porous silicon nanocrystallites [3,4]. At the same time, the reasons for the high-efficiency luminescence are still somewhat under debate [3–5]. It was suggested that it is the amorphous or microcrystalline nature of the porous silicon that is responsible for the phenomenon, or that the formation of silicon compounds such as siloxene (Si6O3H6) or SiH, SiO and SiF bonds are involved in the luminescence [3,4].

* Corresponding author.

In this paper we present combined measurements of electroluminescence (EL) and photoluminescence (PL) in p-type porous silicon. It is shown that our combined results of EL and PL point to the important conclusion that in addition to quantum confinement, localized surface states play an important role in the luminescence process at the red part of the spectrum. Such states may be associated with adsorbed species and disappear upon oxidation.

2. Experimental The starting material was high-grade p-type (100) silicon of resistivity in the range 0.5–1.5 Vcm. A p+ layer was formed by diffusing metallic A1 into the back faces of the silicon wafers to obtain an ohmic contact. The sample was attached to a cylindrical Teflon cell via a Kalrez O-ring. The sample constituted the bottom of

0921-5107/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 1 0 7 ( 9 9 ) 0 0 4 8 8 - 2

E. Sa6ir et al. / Materials Science and Engineering B72 (2000) 138–141

the cell, with its front surface facing upwards. Before anodization, the samples were etched in 20% HF. In order to prepare the porous surface [4], a solution of HF, ethanol and water (1:2:1) was poured into the cell. A platinum electrode was immersed in the solution and a spring contact was attached to the bottom p + contact. The anodization of the Si surface was carried out with a current density of 15 – 50 mA cm − 2. The EL and PL were measured in the same cell using a CCD-based computer board spectrometer. In the former case the cell was filled with an electrolyte consisting of an aqueous solution of NaCl. The EL was obtained under constant current conditions, the current densities ranging from l – 3 mA cm − 2. The PL was excited by a 45-mw argon laser beam of 488 nm wavelength.

3. Results In order to determine the EL characteristics we applied to the cell a constant current (the electrolyte negative) and continuously measured the EL spectra. In Fig. 1 we show some of these spectra (assorted symbols) obtained at different times of the EL current flow as marked on the curves. We see that upon applying the current the EL intensity first rises with time, reaches a maximum, and then decays to zero. The whole process usually takes about half an hour. At the same time, the peak of the EL spectrum shifts from :850 nm in the beginning to :600 nm towards the end of the process.

139

The solid curve represents the total luminescence power as a function of wavelength, obtained by integrating the EL emission intensity. It is seen that this curve forms an envelope of the EL spectra at the different stages of the current flow, as expected. The width of the different EL spectra was found to be about 120 nm and is essentially the same for all samples studied. This is illustrated also by the data in Fig. 2. Here we plot the EL peak amplitude as a function of the area of the spectrum. The various symbols (nine altogether) correspond to samples prepared under different anodization conditions, while the recurrence of the same symbol represents measurements at different times of the EL current flow. We note that all data lie on a straight line, showing the proportionality between the emitted EL power (area) at any time and the peak amplitude. This indicates that both the shape and the width of the various spectra are constant. Accordingly we used the peak amplitude to represent the emitted EL power. The main conclusions from Fig. 1 are summarized in Fig. 3. Here we plot the emitted EL power and the wavelength of the spectral peak as functions of time. Also plotted is the time variation of the voltage across the sample. Note that during the time that the EL persists, the voltage rises moderately, but shortly after the disappearance of the EL, it rises steeply. This behavior is associated with the buildup during the EL current flow of an insulating layer, most probably an oxide, on the porous surface. In fact we have measured the buildup process of such an oxide layer on crys-

Fig. 1. Several EL spectra (assorted symbols) obtained at different times of the EL current flow (current density J =3.3 mA cm − 2) as marked on the curves. The solid curve represents the total luminescence power.

140

E. Sa6ir et al. / Materials Science and Engineering B72 (2000) 138–141

Fig. 4. Growth of the oxide-layer thickness on crystalline silicon with time of applied current.

Fig. 2. EL peak amplitude as a function of the area of the spectrum. The various symbols (nine altogether) correspond to samples prepared under different anodization conditions, while the recurrence of the same symbol represents measurements at different times of the EL current flow.

talline silicon during current flow under similar conditions to those used for EL in porous silicon. Use was made of the space charge capacitance technique of the silicon–electrolyte interface [5]. Fig. 4 shows the growth of the oxide thickness with the time of the applied current. We see that the current flow causes an appreciable enhancement in the oxide thickness. It is only to

be expected that the EL current induces an oxide growth in porous silicon as well. Fig. 5 displays the PL spectra (solid curves) at the onset of the EL (curve 1) and after the EL disappears (curve 2). We see that due to the EL current flow, the red part of the PL practically disappears. As a result, the PL spectrum is appreciably narrower and blue shifted. For comparison purposes we replotted from Fig. 1 (dotted curves) the total EL emitted power (curve 3) and one of the last measured EL spectra (curve 4). The two curves were normalized to the respective PL curves. It is seen that each pair of curves coincide quite closely.

Fig. 3. Emitted EL power and wavelength of the spectral peak as functions of time. Also plotted is the time variation of the voltage across the sample.

E. Sa6ir et al. / Materials Science and Engineering B72 (2000) 138–141

141

Fig. 5. PL spectra (solid curves) at the onset of the EL (curve 1) and after the EL disappears (curve 2). For comparison purposes, we replotted from Fig. 1 (dotted curves) the total EL emitted power (curve 3) and one of the last measured EL spectra (curve 4).

4. Discussion The blue shift of the EL peak with the EL time and the accompanying moderate increase of the voltage across the sample are somewhat similar to results obtained by Bsiesy and Vial [6] and are probably due to the same mechanism. Because of quantum confinement, the energy gap associated with the nanocrystallites is higher, the smaller the size of the crystallite. At the beginning, the contribution to the EL is mainly from the larger nanocrystallites. This is because such particles are associated with smaller energy gaps so that electron injection from the electrolyte requires less energy (lower voltage). With time, these nanocrystallites become covered with an oxide due to the EL current flow. At this point the current is diverted to smaller nanocrystallites and electron injection into such crystallites requires a higher voltage. As pointed out above, the measurements were performed under constant current conditions and we see indeed from Fig. 3 that the voltage across the cell rises with the time of the EL current flow. The EL disappears when all of the nanocrystallites, large and small, become covered with an oxide. One might conceivably assume that the EL originates from band-to-band transitions and thus account for the blue shift as the EL current switches from larger to smaller nanocrystallites. Such a model, however, is inconsistent with some PL data. In the first place we reported previously [7] that the excitation of the PL is practically nonexistent in the range of wavelengths encompassed by the PL spectrum (above 600 nm). This suggests that the PL involves some intermedi.

ate localized states, and that the luminescence results from transitions between such states and either energy band. As the gap of the nanocrystallites involved increases, so does the energy separation between the states and the relevant band. Further information about the nature of the localized states is provided by the observation that the red part of the PL spectrum disappears at the end of the EL process (Fig. 3). This suggests that at least the localized states responsible for the red part of the spectrum reside at the surface of the nanocrystallites. Such states may be associated with adsorbed species and disappear upon oxidation. In conclusion, it appears that in addition to quantum confinement, localized states play an important role in the luminescence process in porous silicon. Our combined results of EL and PL suggest that the states responsible for the red part of the spectrum reside at the surface.

References [1] L.T. Canham, Appl. Phys. Lett. 57 (1990) 1046. [2] I. Amato, Science 252 (1991) 22. [3] Z.C. Feng, R. Tsu, Porous Silicon, World Scientific, Singapore, 1994. [4] A.G. Cullis, L.T. Canham, P.D.J. Calcott, J. Appl. Phys. 82 (1997) 909 and references therein. [5] E. Savir, A. Many, Y. Goldstein, S.Z. Weisz, J. Avalos, Surf. Rev. Lett. 2 (1995) 765. [6] A. Bsiesy, J.C. Vial, J. Lumin. 70 (1996) 310. [7] T.V. Torchinskaya, N.E. Korsunskaya, M.K. Sheinkman, L. Yu, Khomenkova, A.L. Kapitanchuk, Y. Goldstein, E. Savir, A. Many. Proc. Int. Semiconductor Conf., CAS’98, Sinaia, October 1998 (Romanian Academy of Sciences, Bucarest, 1999), p. 451.