Marko Topič and Franc Smole Faculty of Electrical Engineering, University of Ljubljana, Slovenia
23.09.98
INVITED PAPER MIDEM'98 CONFERENCE - 25.09.98, Rogaška Slatina, Slovenia
Keywors: optoelectronics, semiconductors, color detectors, RGB colors. Red Green Blue colors, thin film technologies, a-Si;H, Hydrogenated amorphous Silicon, PIIIN structures, Positive-Intrinsic-Intrinsic-Intrinsic-Negative structures, two-terminai structures, PIN-PINIP structures, Positive-lntrinsic-Negative-Positive-lntrinsic-Negative-intrinsic-Positive structures, PINIP-PIN structures, Positive-intrinsic-Negative-lntrinsic-Positive-Positive-lntrinsic-Negative structures, three-terminal structures, TFA sensors, Thin-Film-on-Application-specific-integrated-circuits sensors, PECVD, Plasma-Enhanced Chemical Vapor Depositions, numerical modeling, TOO, Transparent Conductive Oxides, bipolar biases, forward biases, reverse biases
Abstract: The operational principle of two-terminal and three-terminal three-color detectors with the a-Si:H-based multi-layer multi-bandgap structures is investigated. Two different approaches (the two-terminal and three-terminal approach, which lead to either unipolar or bipolar bias-controlled three-color detection, are described and evaluated in terms of spectral response, rejection ratio and color suppression with regard to illumination intensity and bias-light. For the two-terminai PIIIN structure, numerical simulation results showed strong negative correlation between color separation and bias-light sensitivity, i.e. the better the color separation the worse insensitivity to bias-light and stronger non-linearity with illumination intensity. For the three-terminal PIN/PINIP and PINIP/PIN structures, the thicknesses of the individual layers were first optimized for the detection of the fundamental chromatic components using the numerical simulator and afterwards fabricated and characterized.
Ključne besede: optoelektronika, polprevodniki, detektorji barv, RGB barve rdeča zelena modra, tehnologije tankoplastne, a-Si:H silicij amorfni hidrogeniziran, PIIIN strukture pozitivno-notranje-notranje-notranje-negativno, strukture dvo-terminalne, PIN-PINIP strukture pozitivno-notranje-negativno-pozitivno-notranje-negativno-notranje-pozitivno, PINIP-PIN strukture pozitivno-notranje-negativno-notranje-pozitivno-pozitivno-notranje-negativno, strukture tro-terminalne, TFA senzorji tankoplastni na vezjih integriranih aplikacijsko specifičnih, PECVD nanosi CVD plazemsko izboljšani, modeliranje numerično, TCO oksidi transparentni prevodni, točke delovne bipolarne, točke delovne propustne, točke delovne zaporne
Povzetek: Prispevek obravnava delovanje dvokontaktnih in trikontaktnih trobarvnih detektorjev, ki temeljijo na večplastnih strukturah iz amorfnega silicija. Dvokontaktni pristop zaznava vse tri barve s spreminjanjem zunanje napetosti reverzne polaritete, trikontaktni pa s spreminjanjem zunanje napetosti obeh polaritet. Barvno zaznavanje smo za oba pristopa ovrednotili s spektralno občutljivostjo, rejekcijskimi faktorji in faktorji barvnega dušenja v odvisnosti od intenzitete osvetlitve in dodane osvetlitve. Za dvokontaktno PIIIN strukturo so numerični simulacijski rezultati pokazali močno negativno korelacijo med kvaliteto ločevanja barv in neobčutljivostjo na dodano svetlobo. Z izboljšano kvaliteto ločevanja barv torej izgubljamo na neobčutljivosti na dodano svetlobo. Hkrati postaja spektralni odziv PlilN strukture v odvisnosti od intenzitete osvetlitve vedno bolj nelinearen. Za trikontaktne PIN/PINIP in PINIP/PIN strukture smo debeline posameznih plasti optimirali za detekcijo osnovnih kromatskih komponent s pomočjo numeričnega simulatorja, jih izdelali in okarakterizirali.
1. INTRODUCTION
Hydrogenated amorphous silicon (a-Si:H) thin film optoelectronic devices are not only easily applicable in intelligent image thin-film-on-application-specific-inte-grated-circuits (TFA) sensors /1/, but they can also be simply integrated upon amorphous, poly- or mono-crystalline readout electronics /2,3/. The optoelectronic properties of a-Si:H based films can be changed by deposition parameters of plasma-enhanced chemical vapour deposition (PECVD) process and by modifying the optical gap by the addition of carbon or germanium atoms in the plasma during the deposition. Such an approach enables detection from ultraviolet to the infrared /4,5/. High photosensitivity in the visible light spectrum, homogeneous deposition over large areas by PECVD, and low-cost fabrication make a-Si:H and its alloys a promising candidate for color detectors /6/. In multi-layer a-Si:H based structures for detection of two, three or more colors, spectral response is bias-controlled. Since all the signals (e.g. red, green and blue
(RGB)) are bias-controlled at the same spatial detector position without the need of optical filters, the color-moire effect can be prevented. The possibility of producing large area photosensing arrays makes a-Si:H-based devices even more attractive.
Recent advances in the two-terminal a-Si:H based three-color detectors exhibit the potential of these devices for the color sensor arrays. Unresolved speed limitation of the two-terminal transparent conducting oxide (TCO)/NIPI[N/metal detectors /7/ speaks in favour of the alternative two-terminal TCO/PIIIN/metal detectors /8,9/. In order to achieve bias-controlled spectral separation of fundamental chromatic components RGB, the PIIIN structures need to be band-gap profiled. To examine such an approach, the ASPIN numerical simulator is used.
Beside two-terminai devices, a family of three-terminal three-color detectors based on stacked a-Si:H based structures is theoretically and experimentally investigated. The detectors have the structure
TCO/PliN/TCO/Pl2Nl3P/metal or TCO/PliNl2P/TCO/Pl3N/metal. Using the ASPIN numerical simulator, device physics of different design concepts is analyzed and presented. Optimization criteria deduced from simulation and experimental results and their comparison are investigated and discussed.
2.	NUMERICAL MODELING
The ASPIN computer model /10/ is used for steady-state analysis of different PIIIN structures with the aim to explain the red-green-blue (RGB) three-color detection mechanism and to gain detailed insight into the operating principle of the PIIIN structures. Förtha three-terminal structures, the ASPIN simulator was used to optimize the thickness of the individual layers.
The light generation model is based on the particle nature of light and takes into account only the reflection at the front glass surface. The flux of photons is taken to decay exponentially. The absorption coefficient in each layer is wavelength dependent and corresponds to the imaginary part of the complex refractive index. Although the model does not account either for interference or for the numerous reflections in the multi-layer structure, the model gives a good agreement with the experimental results for structures deposited on rough TCO (with haze) /11,12/. For fiat TCOs (without haze), an accurate numerical modeling should include the wave nature of light that accounts for light interference effects, especially in the long wavelength range of the visible light spectrum /13,14/.
3.	UNIPOLAR BIAS-CONTROLLED COLOR DETECTION PRINCIPLE IN PIIIN DETECTORS
For high short-circuit current in PIN a-Si:H solar cells, the collection efficiency is to be as high as possible throughout the whole visible spectrum. Under reverse bias, the collection efficiency (CE) usually only slightly improves. In a-Si:H color detectors, the CE must be bias controlled. Since PIIIN color detectors operate only under reverse bias, this detection principle is described as unipolar bias-controlled detection principle. By increasing the externally applied reverse bias the CE of PIIIN device can only improve. Thus, to enable different spectral response as a function of reverse bias, the CE at short-circuit conditions must be worse than that one of the PIN solar cell. The electric field plays a key role in governing the CE. To worsen CE of the PIIIN structure at short-circuit conditions, the built-in electric field must be weaker throughout the structure and strongly nonuniform. There are several possibilities:
to insert appropriate compensational doped layers next to the P and N layer, ~ to reduce the doping concentration in the P and N layer,
- to use worse quality I layers with higher defect density.
The thickness of constituent layers with their optical absorption coefficients (a) determine the spatial distribution of the photogeneration of excess carriers. Fig. 1 shows typical generation rate profiles in a P(a-SiC:H)li(a-SiC:H)l2(a-Si:H)l3(a-SiGe:H)N(a-Si:H)
three-color detector. Since the P and H layer have a high optical gap (Eopt=2.2 eV), the photogeneration region of blue monochromatic illumination (?t=450 nm) dominates in the front part of the PIIIN structure. The abrupt changes of profiles denote interfaces between layers and they are due to higher a in the subsequent layers with the lower optical gap. For red monochromatic illumination (X=630 nm), the photogeneration region spreads in the last third of the PIIIN structure.
Simulations showed that strong non-uniform electric field profiles in the PIIIN structures always lead to an increased electric field in the front part of the structure, especially at the PH interface. Therefore, the PIIIN devices under low reverse bias detect only shorter wavelengths and under higher reverse bias the electric field strengthens throughout the structure, so the whole
1.	I2	■3
(a-SiC:H)	(a-Si:H)	(a-SiGe:H)
10^' 1020 f 10" I w
s g
o 10"
X=450 nm
X=530 nm
>.=630!n
rV.
\
I
__J

i_
0 100 200 300 400 Position (nm)
Fig. 1 Generation rate profiles throughout a
Ph I2I3N structure for different monochromatic iiiuminations (1 mW/cm^).
S
JJ >
w
Fig. 2
Position (nm)
Calculated electric field profile of the optimized PV/1/2/3TC/V structure for a three-color detector under different monochromatic illuminations and two biases.
range of the visible light spectrum is detected. In case of three-color detector, such a detection principle can generate three signals with the information: blue (B), blue-hgreen (B-hG), blue-fgreen-hred (B-^G + R). The signals are to be transformed into the RGB components. However, the extraction is justified only if the three-color detector behaves as a linear system, i.e. linearly as a function of illumination intensity and insensitively to the bias-light.
4. THREE-COLOR PINN DETECTORS
In the study of Pli I2N structures /15/, the built-in electric field was profiled using the compensational doping approach in the Pvli l27tN structure or using the high-de-fect-density h* layer approach in the Pli*l2N structure. Both analyzed structures suffered the non-linearity and bias-light sensitivity.
In the PI1I2I3N detectors, an a-SiGe:H layer with Eopt=1.6 eV is added as the third I layer (I3) to improve the detection of the long wavelength illumination. The germanium content in a-SiGe:H increases the slope of the tail states and defect density. From /16/, a 5-times higher density of dangling bond states than in the l2(a-Si:H) layer was selected. The calculated generation rate profiles of the optimized PI1I2I3N structure with 10-100-150-115-25 nm thickness are shown in Fig. 1.
Under short-circuit condition, the electric field should only assist in the collection of excess carriers from short wavelength photons (up to 450 nm). Thus, the built-in electric field should be high only in the front part of the structure, otherwise it should even change its direction. For this purpose, we inserted between the P and h layer a compensational N layer (v layer) and between the I3 and N layer a compensational P layer (n layer). The built-in electric field profile of the optimized structure is shown in Fig. 2 (full line). Under short-circuit condition with monochromatic illumination, the electric field changes due to recharging of defects (Fig. 2). Increasing the reverse bias, the electric field recovers first for blue illumination, afterwards for green and finally, also for red illumination. For a good color separation, the doping concentrations in the compensational v and n layers are to be high or the same effect can be achieved by selecting the P and N layer with lower doping concentration.
Calculated spectral response of the optimized PI1I2I3N structure as a function of reverse bias is presented in Fig. 3. Calculated current-voltage (J-U) characteristics of the optimized PI112I3N device for different monochromatic illuminations (1 mW/cm^) is plotted in Fig. 4. Current-voltage behavior under blue and green illumination is qualitatively similar to the behavior of the PUN structures. In contrast to the PUN J-U characteristics under red illumination, a postponed red response with regard to the reverse bias occurred. From the color-separation point of view, this postponed or sometimes even an S-shape behavior is beneficial, since it provides better rejection ratios under lower reverse bias, and different groups /8,9/ experimentally observed it. The origin comes from increased defect states in the I3 layer, hindering the extraction of excess carries therein due to increased recombination. The extent of increase of the defect states in the I3 layer increase determines the
400 500 600 700 Wavelength (nm)
Fig. 3 Calculated spectral response of the optimized P/1/2/3A/ structure for a three-color detector as a function of reverse bias.
6
ž
ß
0.0
-0.1
-0.2
I
c i
0 -0.4
-0.5
>.=430 nm
/
// //
/
>.=630 nm
X=530 nm"
-2
-1
Voltage (V)
Fig. 4 Calculated J-U characteristics of the optimized P/1/2/3/V device for different monochromatic illuminations (1 mW/cm ).
triggering reverse bias, beyond which the red response steeply starts to increase.
4.1. Examination of bias-light and illumination intensity
For correct extraction of an RGB signal, the PIIIN three-color detectors should exhibit no monochromatic bias-light dependence. We examined the bias-light dependence under the short (450 nm) and long wavelength (650 nm) bias-light. The optimized three-color PIIIN structure, exhibits weak bias-light dependence under short-circuit (detection of B) and under high reverse bias (detection of B-t-G-fR). Unfortunately, strong bias-light dependence occurs in the middle
range of reverse bias (Fig. 5). Significant variation of the spectral response is due to the redistribution of the electric field caused by the recharging of defects in the front or the rear part of the device for short wavelength or long wavelength bias-light, respectively, We managed to mitigate the bias-light dependence by reducing the thickness of the device. At the same time, reduced thickness almost proportionally shrinks the reverse bias range of detection, but it affects the spectral response only in the long wavelength region. The results showed that the bias-light dependence correlates with the reverse bias dependence of the long wavelength spectral response (under no bias-light). The higher the variation of red response as a function of reverse bias is, the more bias-iight dependent is the device. To reduce the bias-liglit dependence, we have to sacrifice the suppression of long wavelength response under short-circuit conditions, resulting in a worse color separation. Thus, a trade-off between good color separation and low bias-
C
a
lA O
cx;
S
Fig. 5
500 600 Wavelength (nm)
Calculated spectral responses of the optimized 400 nm thick PIIIN color detector for (0 V, -0.5 V, -2 V) bias without (full line), with 450 nm (dash-dot-dot line) and 650 nm (dashed line) monochromatic bias-light
(1 mW/cm").
400 500 600 700 Wavelength (nm)
Fig. 6 Calculated spectral responses as a function of illumination intensity (0.001, 0.1, lOmW/cm^).
light sensitivity is in the unipolar bias-controlled PIIIN structures unavoidable.
All structures were also examined for different illumination intensities, ranging from /jW/cm^ up to 100 mW/cm2. Again, larger differences in spectral response arose only in the middle reverse bias range (Fig. 6). Simulations showed that thinner devices exhibit better linearity. Again, a trade-off between good color separation and illumination linearity together with bias-light sensitivity is therefore necessary.
5. BIPOLAR BIAS-CONTROLLED
DETECTION PRINCIPLE IN THREE
TERMINAL DETECTORS
A family of bipolar bias-controlled three-terminal three-color detectors based on stacked a-Si:H based structures has recently been proposed /17,11,12/. The detectors have the structure TCO/PhN/TCO/Pl2Nl3P/metal or TCO/PhNlaP/TCO/PlsN/metal. Numerical analysis of both stacked structures, and the optimization of their layer thicknesses for the detection of the fundamental chromatic components - blue, green and red - was performed using the ASPIN numerical simulator /17/.
5.1. Bipolar bias-controlled detection principle
The TCO/PhN/TCO/PbNIsP/metal structure (Fig. 7a) consists of a top PHN diode and two anti-serial diodes in the sequence PI2NI3P accompanied by three contacts (TCO1, TCO2 and metal). The PI1N diode independently detects the blue color under reverse bias (Ui <0 V), while the PI2NI3P structure acts under differ-
a)
a	sT			f 1' x>	jS		
Ü	CJ					K	
I				lÜF Äf	t		
		z	HJ			i/	
Fig. 7
b)
Schematic view of a three-terminal detectors
a)	TCO/PI^N/TCO/PlzNIsP/metaland
b)	TCO/PliNl2P/TCO/Pl3N/metalstructure.
ent polarity of the externally applied voltage U2 as the photodetector for the green or red color. For a negative bias (U2<0 V), the PI2N diode operates as a photodetector and the NI3P diode as an innpedance. While most of the high energy photons (blue light) are already absorbed in the top PI1N diode, the photons of green light generate electron/hole pairs mainly in the PI2N diode. The red light has the longest penetration depth, and it should be collected in the NI3P diode under a positive bias (U2>0 V), under which the PI2N diode operates as an impedance and the NI3P as a photodetector, Since both polarities of bias are used, this color detection principle is called bipolar bias-controlled principle.
lntheTCO/PliNl2P/TCO/Pl3N/metal structure (Fig. 7b), the PI3N diode detects the red color independently (U2<0 V), while the PI1NI2P structure acts under application of different bias voltages as a photodetector for the blue and green color (analogous to the operation of PI2NI3P discussed above).
The only sophistication is the three-terminal approach that requires some additional technological steps (deposition of TCO2, interconnections), which have already been successfully utilized for parallel-connected tandem solar cells /18/. With regard to the electronic detection system, this three-terminal approach for the detection of three colors is even more simple than the two-terminal approach.
5.2. Simulation results
Simulations and optimization of the TCO/PliN/TCO/Pl2Nl3P/metal and TCO/PliNl2P/TCO/Pl3N/metal structure was performed using the ASPIN numerical device simulator. The optimized thicknesses of the individual layers are listed in Table 1.
For the top li-layer and all P-layers, a wide bandgap (a-SiC:H) material is used. The N layer in the PI1NI2P structure is thicker in order to provide spectral separation between blue and green color, and hence to have good rejection ratios (>2.0).
The calculated spectral responsivity of the TCO/PliNl2P/TCO/Pi3N/metal structure is plotted in Fig. 8. The structure exhibits narrow spectral responses (full width half magnitude - FWHMs below 150 nm) and
high rejection ratios: for the PI1NI2P structure both at 430 nm (R-iv/R-nv= 3.0) and at 530 nm (R-nv/R-iv= 5.3). The color suppression is R43onm/R53onm= 5.6 at Ui = -1 V and R53onm/R43onm= 2.8 at Ui = V . For the PI3N diode, which independently detects red color at reverse applied voltage, the color suppression is also good (R63onm/R53onm= 2.7 at U2= -1 V).
0.25
0.00
300 400 500 600 700 800 Wavelength X (nm)
a)
Fig. 8
0.00
300 400 500 600 700 800 Wavelength A. (nm) b)
Calculated (lines) and measured (lines drawn as guides for the eyes with symbols) spectral responsivity of
a)	TCOIPhNITCOIPIzNIsPlmetaland
b)	TCO/PhNl2P/TCO/Pl3N/metal structure.
Table 1 Optimised geometrical parameters of three-terminal three-color detectors
		TCO	P	h	N	TCO	P	I2	N	I3	P
i ds	(nm)	1000	5	40	5	1000	11	50	140	369	11
dE	(nm)	740	10	30	10	1000	10 1	50	130	270	20 1
											
		TCO	P	li	N	I2	P	TCO	P	13	! P
: ds	(nm)	1000	5	35	45	60	11	1000	30	365	20
dE	(nm)	1000	10	35	60	60	11	1000	20	365	20
8- simulation; E - experiment
E
o <
•ž"
tA
C Hi T3
O j: Du
TT-. . .......
;t=430nm, U,=-\ V X=530nm, (7,=+! V X=630nm, V
Photon flux (cm'^s"')
a)
e
<
C u -a
c
t
o ft.
10-^ 10-3 lO-* 10-5 lO-"^
10-'
i—.....'"1—........1 . 0 l=430ntn, U,'	.....' '""I ' ' ■ ""i =-! V :
-a-- l=530nm, U,'	=+l V
r l=630nm, Uf	=-l V ^ 1
	
	
	
	
	//
	
	
	
	
	
r	
	
y	
........1 .........	.......J—........
10'
10'
10'
10''
Photon flux (cm'^s"')
b)
Fig. 9 Calculated (lines) and measured (symbols) photocurrent vs. photon flux under blue (430 nm) and green (530 nm) and red (630 nm) illumination of
a)	TCO/PhN/TCO/Pl2Nl3P/metal and
b)	TCO/PhNl2P/TCO/Pl3N/metal structure.
Simulations showed that both structures had a linear photocurrent/generation-rate relationship for all three colors at peak wavelengths 450, 530, 635 nm, applying a bias of ±1 V or more (Fig. 9). This linearity allows all three colors to be easily detected with adequate rejection ratios.
5.3. Experimental results
A conventional two-chamber PECVD deposition system (30 X 30 cm2) was used for the deposition of PIN and PINIP structures. A smooth glass/indium tin oxide substrate was used for the front TCO, and sputtered ZnO for the second TCO. Patterning steps were made by laser scribing /18/.
We started the experimental investigation with the TCO/PHN/TCO/PlaNlsP/metal structure. It had already been demonstrated /11/ that the PI2NI3P structure exhibits very good (>3.0) rejection ratios and color suppressions between green and red color (Fig. 8a, lines with symbols). However, the thickness of the top PI1N
diode (10-30-10 nm) was too small to prevent the shunt defects, thus hindering the detection of the blue color, We managed to reduce the top PI1N diode thickness (while preserving its functionality) to 90 nm. But, the spectral response is too broad (FWHM=210 nm) and the maximum is located at 455 nm. Despite the ii layer thickness reduction, the spectral responsivity of PI1N diode decreases very slowiy in the wavelength region above 500 nm, indicating that an a-SiC:H material with a higher optical gap should be preferred.
The problem of the top PI1N diode functionality was solved with the TCO/PliNl2P/TCO/Pl3N/metal stacked structure. Since the whole Ph NI2P (10-35-60-60-15 nm) structure was around 180 nm thick, local defects impeding the photodetection function were eliminated. The fabricated PI1NI2P structure shows good detection of blue and green color (Fig. 8b; lines with symbols). It exhibits narrow spectral responses (FWHM(ui =-i v) = 165nm and FWHM{ui = -nv) = 180 nm) and high rejection ratios, at >.=450 nm with R-iv/R+iv= 7.8 and at >.=530 nm with R+iv/R-iv= 1-3. The measured color suppression is high, R45onm/R53onm= 4.1 at Ui= -1 V and R53onm/R45onm= 2.3 at Ui= -H V, and the dynamic range is also very high.
The fabricated bottom PI3N (20-365-20 nm) diode detects independently the red color under reverse bias. Its measured spectral responsivity at U2= -1 V is plotted in Fig. 8b (lines with triangles). It has the narrowest FWHM (140 nm) in the structure and a high color suppression (R63onm/R53onm= 3.1 at U2= -1 V).
The linear photocurrent/generation-rate dependence of both three-terminal structures derived from the simulations was confirmed by measurements under different monochromatic illumination intensities ranging from 10^^ to 10^^ photons per cm^s (Fig. 9; lines with symbols).
Comparison between simulated and measured results in Fig. 8 reveals the following: a) losses in the glass and front TCO were higher than assumed; b) the li layer absorbs too large a part of the long-wavelength light; c) the second TCO acts as a good reflector, resulting in a large difference between simulated and measured results for the PI3N diode.
Thickness optimization of the second TCO will be necessary. Good reflectivity of the second TCO and thus low responsivity of the PI3N diode could be partly mitigated with an improved reflection at the back contact.
6. CONCLUSIONS
We used numerical modeling as a tool for analysis of thin film color detectors based on a-Si:H.
The device physics of two-terminal a-Si:H based three-color detectors with the multi-layer multi-bandgap PIIIN structure was investigated. They operate under different reverse (unipolar) biases, The PIIIN structures suffer from high bias-light sensitivity and non-linearity of the illumination intensity, although this sensitivity can be mitigated with the higher built-in electric field at the expense of worse color separation. The optimized PIIIN
structure operates under lower range of reverse bias and the I3 layer with an increased defect density results in a postponed red response. The results lead to the conclusion that for a more linear and bias-light independent color detection in the PIIIN devices, we have to sacrifice the quality of three color separation or vice versa.
The three-terminal structures operate under forward and reverse (bipolar) biases. They exhibit excellent linearity, a high dynamic range and they directly generate a RGB-signal.
Due to stringent thickness conditions (<60 nm) for the top PI1N diode in the TCO/PliN/TCO/Pl2Nl3P/metal structure, and additionally due to high reflection of the second TCO, the steady-state experimental results indicate that the TCO/PliNl2P/TCO/Pl3N/metal structure gained an advantage over the TCO/PliN/TCO/Pl2Nl3P/metal structure.
ACKNOWLEDGMENTS
The authors want to thank J. Furlan (University of Ljubljana) for helpful discussions, W. Kusian (Siemens AG) for the encouragement and the fabrication of three-terminal samples. The work was supported by the Ministry of Science and Technology Republic of Slovenia, which is gratefully acknowledged.
REFERENCES
/1/ M. Böhm eta!., "Image sensors inTFAtechnology", Mat. Res. Soc. Symp. Proc. 507 (1998) in print.
/2/ R. A, Street, R, L, Weisfield, R. B. Apte, S. E. Ready, A. Moore, M. Nguyen, W. B. Jackson, P. Nylen, "Amorphous silicon sensor arrays for X-ray anddocument imaging", Thin Solid Films 296 (1997) 172.
/3/ P. Mei, J. B. Boyce, D. K. Fork, G. Anderson, M. Hack, R. Lujan, "Hybrid amorphous and polycrystalline silicon devices for large-area electronics", Mat. Res. Soc. Symp. Proc. 507 (1998) in print.
/4/ G. de Cesare, F. Irrera, F. Palma, M. TuccI, E. Jannlttl, G. Naletto, P. Nicolosi, "Amorphous silicon/silicon carbide pho-todlodes with excellent sensitivity and selectivity In the vacuum ultraviolet spectrum", Appl. Phys. Lett. 67 (1995) 335.
/5/ Y. K. Fang, S. B. Hwang, K. H. Chen, 0. R. Liu, L. C. Kuo, "A Metal-Amorphous Silicon-Germanium Alloy Schottky Barrier for Infrared Optoelectronic 10 on Glass Substrate Application", IEEE Trans, on Electron Devices 39 (1992) 1350.
/6/ H. Stiebig, D. Knipp, J. Folsch, F. Finger, H. Wagner, "Optimized three-color detector based on a-SiGe:H heterojunc-tlons", Mat. Res. Soc. Proc. 420 (1996) 153.
/7/ H. Stiebig, 0. Ulrichs, T. Kulessa, J. Fölsch, F. Finger, H. Wagner, J. Non-Orystalline Solids 198-200 (1996) 1180.
/8/ T. Neldlinger, M. B. Schubert, G. Schmid, H. Brummack, "Fast color detection with two-terminal P-l-l-N devices". Mat. Res. Soc. Proc. 420 (1996) 147.
/9/ D. KnIpp, H. Stiebig, J. Fölsch, R. Carlus, H. Wagner, "Improved concept for NIPIIN and PIIIN color sensitive two-terminal devices with high linearity". Mat. Res. Soc. Proc. 467
(1997)	931.
/10/ F. Smole, M. Topic, J. Furlan, "Analysis of TCO/p(a-SiC:H) heterojunctlon and Its Influence on p-i-n a-Si:H soiar cell performance", J. Non-Orystalline Solids 194 (1996) 312.
/11/ M. Topic, F. Smole, J. Furlan, W. Kusian, "Stacked
a-SIC:H/a-Si:H heterostructures for bias-controlled three-colour detectors", J. Non-Crystalline Solids 198-200 (1996) 1180.
/12/ M. Topic, F. Smole, J. Furlan, W. Kusian, "Stacked
a-Si:H-based three-colour detectors", J. Non-Crystalline Solids 227-230 (1998) 1326.
/13/ G. Tao, M. Zeman, J. W. Metselaar, "Accurate generation rate profiles In a-Si:H solar cells with textured TOO substrates". Solar Energy Materials and Solar Cells 34 (1994) 359.
/14/ P. Popovlč, J. Furlan, W. Kusian, "The wave nature of light In computer analysis of solar cells". Solar Energy Materials and Solar Cells 34 (1994) 393.
/15/ M. Topic, F. Smole, J. Furlan, "Investigation of a-SI:H PUN color detector operation". Mat. Res. Soc. Symp. Proc. 507
(1998)	in print.
/16/ J. Folsch, F. Finger, T. Kulesa, F. Siebke, W. Beyer, H. Wagner, "Improved ambipolar diffusion length In a-Sh -xGexiH alloys for multi-junction solar cells". Mat. Res. Soc. Symp. Proc, 377 (1995) 517.
/17/ M. Topic, F. Smole, J. Furlan, " New bias-controlled three-color detectors using stacked a-SiO:H/a-Si:H heterostructures", Mat. Res. Soc. Symp. Proc. 377 (1995) 779.
/18/ W. Kusian, R. D. Plättner, J. Furlan, G. Conte, "A new double stacked a-Si:H solar cell having the structure pln/TCO/nlp", Proc. 23"^ IEEE PVSC (1993) 955.
Marko Topič and Franc Smole Faculty of Electrical Engineering University of Ljubljana Tržaška 25, S I-1000 Ljubljana, Slovenia E-mail: Marko. Topic@fe. uni-lj.si
Prispelo (Arrived): 21.9.1998 Sprejeto (Accepted): 16.11.1998