Color-sensitive photodetector based on porous silicon superlattices

Share Embed


Descripción

Thin Solid Films 297 (1997) 241–244

Color-sensitive photodetector based on porous silicon superlattices M. Kru¨ger a, M. Marso a, M.G. Berger a, M. Tho¨nissen a, S. Billat a, R. Loo a, W. Reetz a, H. Lu¨th a, S. Hilbrich b, R. Arens-Fischer b, P. Grosse b a

Institut fu¨r Schicht- und Ionentechnik, Forschungszentrum Ju¨lich GmbH, D-52425 Ju¨lich, Germany b I. Physikalisches Institut, RWTH Aachen, D-52056 Aachen, Germany

Abstract

Color-sensitivity of Si photodiodes was achieved by integrating porous silicon (PS) Bragg reflectors and Fabry–Perot filters. The PS was formed in the pq-type part of the pqn junction which required illumination of the samples during anodization. The optimal illumination power density turned out to be a compromise: high power densities are necessary to enable high anodization current densities, but this results in a degraded filter performance. The PS layers had no significant influence on the electrical characteristics of the photodiodes, but as expected they strongly modified the spectral response. The results are in good agreement with the reflectance spectra of the filters. q 1997 Elsevier Science S.A. Keywords:

Photodetectors; Porous silicon; Superlattices; Colour

1. Introduction

The refractive index of porous silicon (PS) can easily be varied over a wide range by changing the anodization current density. By this means we were able to fabricate interference filters [1,2] and optical waveguides [3] from pure Si wafers. The main advantage of this method compared to conventional ones is the very cheap and fast fabrication, because no expensive deposition processes (e.g. for Si3N4, SiO2 or TiO2) are necessary. Moreover, the refractive index of PS can be varied continuously, which results in a large freedom in filter design (rugate filters [4]). First applications of PS interference filters concerned the narrowing of the PS luminescence spectra [1,2,5,6]. Besides these application in optoelectronics, the usage of PS Bragg reflectors turned out to be very suitable in low-cost optical sensors [7]. Recently, we could demonstrate that PS Bragg reflectors can also be integrated in Si photodiodes as colorselective layers [8]. These preliminary results are extended in this work with special emphasis on the problems of PS integration in pn junctions. 2. Device geometry

A schematic view of the photodiode structure is shown in Fig. 1. The pn junction consists of an epitaxial, 1.8 mm thick p-type layer grown by low-pressure chemical vapour depo-

Fig. 1. Device geometry of the Si photodiode with a PS multilayer stack in the pq-type top layer. The photocurrent flows through the crystalline Si because of the very high resistivity of the PS.

sition (LPCVD) on top of a n-type substrate. The PS layers which determine the photodiodes’ spectral response are formed in the upper, p-type layer. We used highly-doped (1=1019 cmy3) p-type Si because of the wide range of porosities which is available using this kind of material. This corresponds to a large range of refractive indexes [4] which is desired for the fabrication of thin, high-quality PS filters [9]. On the contrary, the n-type substrate was low doped (5=1014 cmy3) in order to obtain a photodiode with a high spectral response. Processing of the samples just required standard photolithographical techniques well known from other Si devices. The diameters of the photodiodes ranged from 50 to 2000 mm,

0040-6090/97/$17.00 q 1997 Elsevier Science S.A. All rights reserved PII S0040-6090(96)09414-X

Journal: TSF (Thin Solid Films)

Article: 9414

242

M. Kru¨ger et al. / Thin Solid Films 297 (1997) 241–244

but even smaller dimensions of just some microns seem to be achievable. As ohmic contacts, Cr/AuSb/Au (5/150/20 nm) was evaporated on the backside of the samples and Cr/ Au (50/100 nm) on the front side. Then the samples were annealed (370 8C; 150 s) and the mesa structure was defined by reactive ion etching (RIE). Next the whole front side except the filter area was covered with photoresist which was baked out at 130 8C for 10 min to withstand the etch solution during anodization. Special care had to be taken for tightly covering the steep mesa edges. Otherwise the n-type substrate is etched favourably, as the anodization current then not needs to pass the pn junction. The last step was etching the PS multilayer stacks using an electrolyte composition of H2O:HF:C2H5OH 1:1:2. Two different types of PS superlattices were investigated: Bragg reflectors and Fabry–Perot (FP) filters. The Bragg reflectors consisted of l/4 stacks which can be denoted as (HL)n, whereby n is the number of the layers with high (H) and low (L) refractive index. The FP filters consisted of a l/2 cavity embedded in between two Bragg reflectors and can analogously be denoted as (HL)m(LH)n. The filter characteristics turned out to become better with increasing ratio of the refractive indexes nH and nL [9] which corresponds to a large difference of the anodization current densities jH and jL. We used jHs20 mA cmy2 for the H layers and jLs400 mA cmy2 for the L layers which resulted in refractive indexes of nHs2.85 and nLs1.49 at ls600 nm [4]. The thickness of the multilayer stacks was limited by the thickness of the epitaxial layer and varied from 0.75 to 1.44 mm. This means that a 0.36 to 1.05 mm thick crystalline layer remained in between the PS and the pn junction, so that the photocurrent was not forced through the highly resistant [10] PS layer during the measurements. I–V measurements performed in the dark showed no crucial influence of the PS, just the ideality factor seemed to increase slightly. However, this effect was so small that it could not be separated clearly from fluctuations in the quality of the ohmic contacts. 3. Influence of illumination on the filter characteristics

The photodiode structure shown in Fig. 1 implies one main technological problem: PS formation requires the supply of holes from the backside contact of the sample, which means that the pn junction is reverse-biased during anodization. Fig. 2 shows that without illumination the high current densities necessary for PS filter fabrication cannot be applied using a moderate bias. Therefore we illuminated the samples with a 150 W halogen lamp during anodization. The illumination has not only the effect of increasing the reverse current of the diode, but also the undesired effect that electron–hole pairs are generated in the region where the PS is formed. The consequences can be seen in Fig. 3. With increasing illumination power density the FP filter wavelength shifts to longer wavelengths. Moreover, for illumination power densities G2.6 W cmy2 the width of the high

Fig. 2. Current density vs. voltage of a reference photodiode without PS, illuminated with different power densities. The measurements were performed without electrolyte, so the voltage was applied between the backside and the frontside ohmic contact. We could not measure the illumination power density that reaches the sample when the electrolyte is on top, but as an estimate we measured it at the sample position without electrolyte. The horizontal lines indicate the anodization current densities used for PS filter fabrication.

Fig. 3. Reflectance spectra of Fabry–Perot filters ((HL)4(LH)4) fabricated from pq-substrates using different illumination power densities. The anodization current densities (jHs20 mA cmy2; jLs400 mA cmy2) and etching times (tHs1.99 s; tLs0.42 s) were the same for all of the filters.

transmission range suddenly increases and a double peak structure occurs. In order to understand this effect, we fabricated f1 mm thick PS single layers on pure pq substrates under the same illumination conditions (Fig. 4). The etch rate was measured by SEM and turned out to increase with increasing illumination power density. While the difference was smaller than 8% for js400 mA cmy2, a drastic increase of about 30% was observed for js20 mA cmy2. As the FP filters consist of both types of layers, the increase of their thicknesses is between the values of the single layers. Besides the change of the etch rate, also the porosity of the layers is drastically changed. The porosity values given in Fig. 4 were determined by fitting the reflectance spectra of the single layers. The decrease of the porosity causes an

Journal: TSF (Thin Solid Films)

Article: 9414

243

M. Kru¨ger et al. / Thin Solid Films 297 (1997) 241–244

Fig. 4. Influence of the illumination power density on the layer thickness and porosity of PS formed on pq-substrates. The anodization current densities of the single layers were 20 mA cmy2 (———) and 400 mA cmy2 (∆) and the etching times 36.76 s and 4.43 s, respectively. The same current densities were used for the fabrication of the Fabry–Perot filters (- - -), whose reflectance spectra are shown in Fig. 3.

Fig. 5. External quantum efficiency of a photodiode with integrated Bragg reflector (———) compared to a reference diode without PS (∆). The transmission through the PS is roughly estimated by 1 minus the reflectance (- - -) of the Bragg reflector.

increase of the refractive index. This means that both effects, the increase of the etch rate and the decrease of the porosity, lead to an increasing optical thickness of both the H and the L layers of the FP filters. This effect explains the red-shift of the filters in Fig. 3. As the change of the optical thickness is different for the H layers and the L layers, the interference effects are disturbed in the case of strong illumination and cause the increase of the FWHM. Therefore in future the current density and etching time should be adjusted to compensate the effect of illumination. However, for the fabrication of the filters shown below, this adjustment was not yet performed so that the filter characteristics are not optimal.

is much lower than the one of the reference diode without PS. The transmission spectrum of the filter becomes dominant in the range between. In Fig. 5 the transmission of the filter is roughly estimated by one reflectance neglecting the absorption in the PS. The photodiode is nearly insensitive to light with wavelengths around 600 nm, corresponding to the maximum of the reflectance of the Bragg reflector. As an undesired effect, oscillations of the EQE occur between 700 and 1000 nm because of the sidelobes in the reflectance spectrum. The maximum EQE nearly reaches the value of the reference diodes which shows that absorption in the PS layers is not the dominant effect, except of the blue spectral range. This was not obvious before because in the case of multiple reflections in an absorbing multilayer stack the absorption can reach very high values. Fig. 5 demonstrates to what great extent the spectral response can be modified by PS interference filters. According to the requirements, the spectral characteristics of the PS filters can widely be varied by: c variation of the thickness of the layers [4] ´filter wavelength; c variation of the number of layers [9] ´variation of the height of the reflectance maximum; c variation of nH/nL [9] ´variation of the height and width of the reflectance maximum; c usage of a continuous refractive index profile (rugate filter [4]) ´reduction of the sidelobes; c oxidation of the PS [4] ´reduced absorption in the PS. Besides these reflectance filters, we also integrated transmission filters in the photodiodes. Fig. 6 shows the effect of a FP filter on the EQE. The high transmission at the filter wavelength (ls600 nm) results in a peak of high sensitivity with a FWHM of 52 nm and a peak-to-valley ratio of 3.8. This is in good agreement with the reflectance spectrum of the filter. We suppose that far better values can be achieved if the influence of the illumination during the anodization is adjusted, as indicated above. For example, on pure substrates

4. Spectral characteristics of the photodiodes

While the I–V measurements in the dark showed no difference between the photodiodes with and without PS layer, the influence of the PS becomes evident under illumination. For that purpose we measured the spectral response of the photodiodes under short circuit conditions, i.e. without applied voltage. From these measurements we calculated the external quantum efficiency (EQE), which is the number of electron–hole pairs generated per incident photon. Reference photodiodes without PS were fabricated from the same wafer using identical processing steps. In Fig. 5 the EQE of such a reference diode is compared to the one of a photodiode with integrated Bragg reflector. In the long-wavelength range (lG1000 nm) the absorption length of the photons in the Si substrate is comparable to the width of the depletion zone, so that this ratio dominates the influence of the PS filter. Therefore the EQE of both diodes becomes equal. In the short-wavelength range (lF500 nm) the situation is reversed: most of the photons are absorbed near the surface of the sample. As the density of recombination centers is very high in the PS, the photons absorbed in the PS do not contribute to the photocurrent and the sensitivity

Journal: TSF (Thin Solid Films)

Article: 9414

244

M. Kru¨ger et al. / Thin Solid Films 297 (1997) 241–244

Acknowledgements

We would like to thank H.-P. Bochem for the SEM investigations and Th. Eickhoff and W. Theiss for helpful discussions. References

Fig. 6. External quantum efficiency of a photodiode with integrated Fabry– Perot filter (———) compared to a reference diode without PS (∆). The filter wavelength is 600 nm.

and without illumination we could recently fabricateFPfilters with a FWHM of just 11 nm and a maximum reflectivity )99% [9]. 5. Conclusion

Porous silicon (PS) multilayer stacks were integrated in conventional pqn photodiodes in order to modify their spectral response. As the pn junctions were reverse biased during anodization, the samples had to be illuminated. This illumination caused an increase of the etch rate and a decrease of the porosity, which influenced the reflectance spectra of PS interference filters. Despite these problems, Bragg reflectors and Fabry–Perot filters could be integrated in the photodiodes. The spectral response of these photodiodes corresponds very well to the spectral characteristics of the filters. Further progress is expected through an adjustment of the anodization current densities.

[1] S. Frohnhoff and M.G. Berger, Porous silicon superlattices, Adv. Mater., 6 (12) (1994) 963–965. [2] M.G. Berger, M. Tho¨nissen, R. Arens-Fischer, H. Mu¨nder, H. Lu¨th, M. Arntzen and W. Theiß, Investigation and design of optical properties of porosity superlattices, Thin Solid Films, 255 (1995) 313– 316. [3] A. Loni, R.J. Bozeat, M. Kru¨ger, M.G. Berger, R. Arens-Fischer, M. Tho¨nissen, H.F. Arrand and T.M. Benson, Application of Porous Silicon technology to optical waveguiding, Proceedings of the IEE colloquium Microengineering Application in Optoelectronics, 27th February, 1996, London, Digest No. 96/39.

[4] M.G. Berger, R. Arens-Fischer, M. Tho¨nissen, S. Hilbrich, W. Theiß, M. Kru¨ger, S. Billat, P. Grosse and H. Lu¨th, Dielectric filters made of PS: Advanced performance by oxidation and new layer structures, Thin Solid Films, to be published. [5] C. Mazzoleni and L. Pavesi, Application to optical components of dielectric porous silicon multilayers, Appl. Phys. Lett., 67 (20) (1995) 2983–2985. [6] M. Araki, H. Koyama and N. Koshida, Jpn. J. Appl. Phys., 35 (2B) (1996) 1041. [7] S. Hilbrich, W. Theiß, R. Arens-Fischer, M.G. Berger, M. Kru¨ger and M. Tho¨nissen, The application of Porous Silicon interference filters in optical sensors, Thin Solid Films, to be published. [8] M. Kru¨ger, M.G. Berger, M. Marso, M. Tho¨nissen, S. Hilbrich, W. Theiss, R. Loo, Th. Eickhoff, W. Reetz, P. Grosse and H. Lu¨th, Integration of porous silicon interference filters in silicon photodiodes, Proceedings ESSDERC96, G. Baccarani and M. Rudan (eds.), ISBN

2-86332-196-X, pp. 891–894. [9] M. Kru¨ger et al., Optimization and first applications of porous silicon Fabry–Perot filters fabricated on pq-silicon, to be published. [10] V. Lehmann, F. Hofmann, F. Mo¨ller and U. Gru¨ning, Resistivity of porous silicon: a surface effect, Thin Solid Films, 255 (1995) 20–22.

Journal: TSF (Thin Solid Films)

Article: 9414

Lihat lebih banyak...

Comentarios

Copyright © 2017 DATOSPDF Inc.