Optoelectronic characterization of a-SIC:H stacked devices

Optoelectronic characterization of a-SIC:H stacked devices

Journal of Non-Crystalline Solids 338–340 (2004) 345–348 www.elsevier.com/locate/jnoncrysol Optoelectronic characterization of a-SIC:H stacked device...

477KB Sizes 0 Downloads 68 Views

Journal of Non-Crystalline Solids 338–340 (2004) 345–348 www.elsevier.com/locate/jnoncrysol

Optoelectronic characterization of a-SIC:H stacked devices P. Louro *, A. Fantoni, M. Fernandes, A. Macßarico, R. Schwarz, M. Vieira ISEL, Electronics Telecommunications and Computer Department, R. Conselheiro Emidio Navarro, P 1949-014 Lisbon, Portugal Available online 15 April 2004

Abstract The aim of this work is the optoelectronic characterization of double p–i–n stacked devices based on a-Si alloy materials, in order to evaluate their suitability in large area optical sensors. Photogeneration, collection efficiency and carrier transport are investigated from dark and illuminated current–voltage characteristics and spectral response measurements, with and without additional background illumination and different electrical bias conditions. Results show that the collection efficiency depends on the device configuration and on the optical and electrical bias. The carrier collection is mainly dependent on the front and back intrinsic layers thickness and on the composition of the p-type doped layers. When wide band gap p-layers are used, the asymmetric distribution of the electrical field controls the transport mechanism. Under red optical bias the electrical field is enhanced at the front cell and decreased at the back one leading to an increased red light-to dark sensitivity. A numerical simulation supports the discussion of the experimental results. Considerations about induced electric field and inversion layers at the interfaces and generationrecombination process are used to explain the device output. Ó 2004 Elsevier B.V. All rights reserved.

1. Introduction

2. Experimental details

Hydrogenated amorphous silicon (a-Si:H) is a material that exhibits excellent photosensitive properties, especially in the visible range of the spectrum. Therefore, it has been largely employed in imaging devices, as well as in colour devices, as its spectral response can be modulated through the applied bias voltage. Various structures and sequences have been suggested [1,2]. In our group efforts have been devoted towards the development of a new kind of colour sensor, the Color Laser Scanned Photodiode Sensor (CLSP) [3–5]. In these sensors simultaneous image and colour detection are achieved by combining the wavelength filtering property of silicon with the sensor responsivity dependence on the applied electrical and optical bias. The optimization of this trade-off demands a full understanding of the transport mechanism. The aim of this work is to describe the results of experiments as well as to discuss the usefulness of the a-SiC:H doped layers in the improvement of the sensor performance.

2.1. Device structure

*

Corresponding author. Tel.: +35-12 1831 7181; fax: +35-12 1831 7114. E-mail addresses: [email protected], [email protected] (P. Louro). 0022-3093/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2004.02.070

Amorphous silicon and silicon carbide single and double stacked p–i–n structures were produced by plasma enhanced chemical vapor deposition. The devices were deposited on glass substrates coated with a transparent conductive material, either ITO or a thin semitransparent metallic coating. The top metallic contact was produced by thermal evaporation. Two single p–i–n structures were studied and characterized [6]. Both consist of a transparent conducting oxide (TCO)/p–i–n/metal structure. One of them is a homostructure (#1), as all films are made of hydrogenated amorphous silicon (a-Si:H), and the other one is an heterostructure (#2) as the n-doped and intrinsic films are similar to those used in the homostructure but the p-doped film is made of amorphous silicon carbide. A schematic view of these structures is shown in Fig. 1(a) and (b)). The stacked structure, #3 (Fig. 1(c)), consists of two p–i–n cells produced in the assembly metal/n–i–p–n–i–p0 /metal, where both intrinsic films and doped n and p films are made of hydrogenated amorphous silicon, and the p0 doped film is made of amorphous silicon carbide. The other stacked structure (#4)

P. Louro et al. / Journal of Non-Crystalline Solids 338–340 (2004) 345–348

2

Current density (A/cm )

346

10

-4

10

-5

10

-6

10

-7

#3 #4

-1.2 -0.8 -0.4 0.0

0.4

0.8

1.2

1.6

2.0

Voltage (V) Fig. 2. Density current dependence on the applied voltage for stacked devices (#3, #4).

Fig. 1. Schematic structure of the devices: (a) single p–i–n homostructure (#1), (b) single p0 –i–n heterostructure (#2), (c) stacked p–i–n– p0 –i–n structure (#3), (d) stacked p0 –i–np0 –i–n structure (#4).

is similar to the previous, but contains carbon in both p layers. The assembly configuration is ITO/n–i–p0 –n– i–p0 /metal and, in this configuration, the light is always shinning through the semitransparent metal contact. In Fig. 1 the thickness of the layers are also displayed. 2.2. Characterization The characterization of the single layer was performed in order to infer the absorption coefficient, optical gap, defect density, conductivity and hydrogen and carbon content of the single layer. These results are reported elsewhere [7]. The optical characterization has been done through the absorption spectra obtained from transmission and reflection measurements and complemented with the constant photocurrent method (CPM). The conductivity measurements were performed using a coplanar geometry. Structural properties were analyzed through infrared spectroscopy. The characterization of the devices was achieved through the analysis of the photocurrent dependence on the applied voltage and spectral response, with and without red optical bias. The responsivity was obtained by normalizing the photocurrent to the incident flux under different light wavelengths (in the range of 400– 800 nm), in dark and under a steady optical bias (kL ¼ 650 nm, UL ¼ 2 mW/cm2 ). To suppress the dc components this measurement was performed using the lock-in technique. Fig. 2 shows, for the stacked devices (#3, #4), the experimental current–voltage (I–V ) characteristics under red illumination (kL ¼ 550 nm, UL ¼ 2 mW/cm2 ). The spectral response was analyzed in order to determine its dependence on the applied electrical bias, with and without optical bias. In Fig. 3 the spectral response of the tandem structures is displayed.

In Fig. 4 and for all the devices it is displayed the light to dark sensitivity (the difference between the normalized spectral responsivities, RN , with and without optical bias). For a better comparison, each spectral response was normalized to its maximum value without applied optical bias (at open circuit voltage). 2.3. Numerical simulation For both the single and the double structures, in Fig. 5, the potential profile in dark and under steady state bias illumination (2 mW cm2 , 450 and 650 nm) is shown. The results were obtained using the ASCA simulator [8]. As input parameters we took into account the device configurations (Fig. 1) and the experimental characterization of the individual layers. Details about the program and experimental input parameters are described elsewhere [9].

3. Discussion Experimental results show that the I–V characteristics under red illumination (Fig. 2) and the spectral response (Fig. 3) are strongly dependent on the configuration of the device, i-layers thickness and doped layers composition (a-SiH or a-SiC:H). The double structures show well shaped I–V characteristics (Fig. 2) due to the electrical field tailoring at the a-Si:H/a-SiC:H internal interface. The current density in the double p0 –i–n structure (#4) is high due to an optimized band offset engineering (i-layer thickness and player optical gap) that overrides the effect of the reverse electrical field (Fig. 5) at the internal interface. The reversed electrical field leads to charge accumulation at the internal n–p interface in the tandem p0 –i–n/p–i–n structure (#3). Therefore, this structure presents lower collection efficiency and a smaller open circuit voltage. As can be seen from Figs. 3 and 4, the spectral response is dependent on both optical and electrical bias. As the applied voltage increases the spectral sensitivity

P. Louro et al. / Journal of Non-Crystalline Solids 338–340 (2004) 345–348

347

Fig. 3. Spectral response dependence on the applied voltage in dark (top row) and under steady state light, hL ¼ 650 nm, UL ¼ 2 mW/cm2 (down row).

decreases. The decay rate depends on the device configuration. In the configurations #2 and #4, as the applied voltage increases, the maximum spectral response shifts to lower wavelengths due to the decrease of the electrical field in one (#2) or both the p (a-SiC:H)–i(aSi:H) interfaces (Fig. 5). This shift is particularly evident in dark. Under illumination the shift is almost imperceptible and the decay is much slower as the applied voltage results in an overriding of the internal field asymmetry. The effect of the p0 doped layer is mainly visible under illumination. In the single p0 –i–n structure, #2, the potential is asymmetric, it drops mainly at the p0 –i interface and flattens inside the absorber (Fig. 5). The doped a-SiC:H layer has a blocking effect [3] and prevents the carriers from being injected into the i-layer

RN(ΦL=0) - RN(ΦL)

1.0

#1 #2 #3 #4

0.8 0.6 0.4 0.2 0.0 400

500

600

700

800

Wavelength (nm) Fig. 4. Difference between the spectral responsivities with and without red bias for different device configurations.

1.4

1.4 1.2 1.0

Potential (V)

Potential (V)

1.0 0.8 0.6 λ L=530 nm

0.4

Dark

0.2

1.2

Dark

1.0

0.8 0.6 0.4

λ L=530 nm

0.2

0.8

0.4

0.0

0.0

-0.2

-0.2

0.2

0.4

0.6

0.0

0.2

0.4

0.6

0.8

1.0

Dark

0.2

0.0

Position (µm)

λ L=530 nm

0.6

-0.2

0.0

λL=650 nm

1.4 λ L=650 nm

Potential (V)

λ L=650 nm

1.2

0.0

0.2

0.4

0.6

Position (µm)

Fig. 5. Numerical simulations of the potential profiles under different illumination conditions.

0.8

1.0

348

P. Louro et al. / Journal of Non-Crystalline Solids 338–340 (2004) 345–348

where the low field region is not high enough to sweep the photocarriers to the contacts before they recombine. Thus, there is a much lower collection under steady state optical bias than in dark (Fig. 3). This blocking layer is responsible for the device high light-to-dark sensitivity (Fig. 4, #2). In the double structures (#3, #4) the effect of the p0 -layer is also visible. In dark a reverted electric field peak at the internal n–p interface is observed, but it can be outlined that while in device #3 it leads to a reverse sign in the electrical field, in structure #4 no sign reversing is observed and the field remains negative all over the device. Under red light exposure, the potential distribution is similar in both configurations. When compared with its distribution in dark, the electrical field is enhanced in the front sub-cell that becomes full depleted and decreases at the back one. In the high wavelength range the spectral response decreases while it increases in the low range (Fig. 3). When only the front doped layer is based on a-SiC:H (#3) the light-to-dark sensitivity decreases (Fig. 4). Here, the red light is absorbed mainly inside the back, not sensitive, homojunction (#1). When the front and back cells (#4) are both based on heterojunctions and if the back diode is thick enough to absorb almost all the incoming steady state light (500 nm, #4) the increased recombination of the photogenerated carriers decreases the spectral response mainly in the red spectral region (Fig. 3). The blue light is absorbed only at the front thin heterojunction and so the carriers are collected due to the increased electrical field in this region. It can be outlined that the light-to-dark sensitivity (Fig. 4) is mainly controlled by the back heterojunction and so, it has the same trend as the single p0 –i–n junction (#2) except in the low wavelength range were it decreases due to the thin front layer increased collection (Fig. 5). The back homojunction in #3 configuration is also the responsible by the lower light-to-dark sensitivity presented as in the homojunction the spectral response is insensitive to the optical light bias (#1).

4. Conclusions An optoelectronic characterization of a-SiC:H p–i–n/ p–i–n stacked devices was presented. The obtained results show that an accurate engineering of the interfaces,

in terms of optical gap and band-bending control can improve collection efficiency in dark and decrease it under illumination giving to the device light-to-dark sensitivity. It was demonstrated that the spectral response and the light to dark sensitivity are strongly dependent on the configuration of the device, i-layers thickness and doped layers composition. Results show that the light to dark sensitivity depends on the optical bias wavelength and is mainly controlled by the back sub-junction characteristics. The light to dark sensitivity is high when the back sub-cell is a heterojunction and low when it is based on a homojunction. The importance of a well balanced photogeneration between the two sub-cells has also been outlined, aiming to indicate the need to determine a light-to-dark sensitive tandem optimal configuration, which will take advantage of the local fields created by band discontinuities at the internal n/p interface.

Acknowledgements The authors are grateful to the IPE-Stuttgart for the sample depositions, and to Y. Vygranenko for his collaboration. This work has been financially supported by IPL and POCTI/ESE/38689/2001 project.

References [1] T. Neidlinger, R. Bruggemann, H. Brummak, M.B. Schubert, J. Non-Cryst. Solids 227–230 (1998) 1335. [2] D. Caputo, G. Cesare, J. Non-Cryst. Solids 198–200 (1996) 1334. [3] M. Vieira, M. Fernandes, J. Martins, P. Louro, A. Macßarico, R. Schwarz, M. Schubert, IEEE Sens. J. 1 (2) (2001) 158. [4] M. Vieira, M. Fernandes, J. Martins, P. Louro, A. Macßarico, R. Schwarz, M. Schubert, Mater. Res. Soc. Symp. Proc. 609 (2000) A14. [5] M. Fernandes, P. Louro, J. Martins, A. Macßarico, R. Schwarz, M. Vieira, Sens. Actuat. A 92 (2001) 60. [6] P. Louro, M. Vieira, Yu. Vygranenko, M. Fernandes, R. Schwarz, M. Schubert, Appl. Surf. Sci. 184 (2001) 144. [7] P. Louro, A. Fantoni, I. Rodrigues, A. Macßarico, M. Vieira, Sens. Actuat. A (2004) in press. [8] A. Fantoni, M. Vieira, R. Martins, Math. Comput. Simul. 49 (1999) 381. [9] A. Fantoni, P. Louro, I. Rodrigues, A. Macßarico, M. Vieira, Sens. Actuat. A (2004) in press.