Hydrogenated Amorphous Silicon Sensors based on Thin Film on ASIC technology

Transcription

1 Hydrogenated Amorphous Silion Sensors based on Thin Film on ASIC tehnology M. Despeisse, D. Moraes, G. Anelli, P. Jarron, J. Kaplon, R. Rusak, S. Saramad and N. Wyrsh Abstrat The performane and limitations of a novel detetor tehnology based on the deposition of a thin-film sensor on top of proessed integrated iruits have been studied. Hydrogenated amorphous silion (a-si:h) films have been deposited on top of CMOS iruits developed for these studies and the resulting Thin-Film on ASIC (TFA) detetors are presented. The leakage urrent of the a-si:h sensor at high reverse biases turns out to be an important parameter limiting the performane of a TFA detetor. Its detailed study and the pixel segmentation of the detetor are presented. High internal eletri fields (in the order of V/m) an be built in the a- Si:H sensor and overome the low mobility of eletrons and holes in a-si:h. Signal indution by generated arrier motion and speed in the a-si:h sensor have been studied with a 660 nm pulsed laser on a TFA detetor based on an ASIC integrating 5 ns peaking time pre-amplifiers. The measurement set-up also permits to study the depletion of the sensor and results are presented. Finally, diret detetion of 5.9 kev X-rays with TFA detetors based on an ASIC integrating low noise pre-amplifiers (27 e - r.m.s) are shown. I. INTRODUCTION IGH resistivity rystalline silion (-Si) is now a well Hestablished detetor tehnology and is widely used in traking appliations in several High Energy Physis experiments. We have investigated an alternative solid state detetor tehnology based on the deposition of a hydrogenated amorphous silion sensor diretly on top of the readout integrated iruit. This tehnology, alled Thin-Film on ASIC (TFA) ([1], [2]), potentially offers radiation hard alternatives to silion devies. The vertial integration of a thin film sensor on the ASIC eliminates the need for bump bonding and enables a level of integration omparable to monolithi pixels while having the advantages of the hybrid pixel approah. Hydrogenated amorphous silion (a-si:h) is used as sensing devie. This material has been studied in the past 30 years and is well known nowadays though it still presents many ontroversial issues. a-si:h is widely industrialized in thin films (< 1 µm) for solar ells and various sensitive imaging devies. Radiation hardness has been studied in the ontext of testing solar ells for spae appliations. It has been shown that proton, neutron and eletron irradiation ([3]-[5]) or photon Manusript reeived November 10, M. Despeisse, D. Moraes, G. Anelli, P. Jarron, J. Kaplon and S. Saramad are with the European Organization for Nulear Researh (CERN), CH-1211 Geneva 23, Switzerland (telephone: , matthieu.despeisse@ern.h). M. Despeisse and R. Rusak are with the shool of physis and astronomy of the University of Minnesota, 116 Churh Street SE, Minneapolis, MN N. Wyrsh is with the Institute of Mirotehnology, IMT, Breguet 2, CH Neuhatel, Switzerland. irradiation (involving the Staebler Wronski effet [4], [6]) all lead to a reation of metastable deep defets that an be annealed out, resulting in a good radiation hardness of the material in omparison to -Si. a-si:h an be deposited at low ost and on large areas at temperatures below 250 C whih are ompatible with proessed integrated iruits. TFA detetor using a-si:h as a deteting devie is then an attrative solution for a radiation hard pixel detetor. However, for partile detetion, thiker a-si:h layers and full depletion of the sensor are needed to provide adequate signal over noise ratio [5]. A first tehnologial hallenge onsists in the deposition of a-si:h layers with thikness higher than 30 µm on ASIC with a minimal defet density to allow full depletion of the sensor and minimal leakage urrents for the high reverse biases needed for the sensor depletion [7]. The voltage V d needed to deplete an a-si:h sensor of thikness d (in µm) an be estimated by onsidering a density of ionized dangling bonds of the material N db * ~ m -3 and V d ~ 0.45 d 2 [1]. Sensors with a thikness higher than 50 µm are then diffiult to grow and would moreover neessitate reverse voltages of about 1.1 kv for full depletion. Sensors with thiknesses up to 36 µm have then been deposited in our studies, whih an be onsidered thik in omparison to standard a-si:h opto-eletroni devies but thin in omparison to standard rystalline silion detetors with thiknesses higher than 150 µm. Different groups have performed haraterization of a-si:h sensors deposited on a glass substrate and demonstrated that the signal reated by a partile in the a-si:h sensor is small [5], [8], [9] (Minimum Ionizing Partiles are expeted to reate few hundred of eletrons in 20 µm thik sensors). One of the resulting hallenges in the development of this tehnology is to design low noise pre-amplifiers able to readout the small signal indued by a partile going through the thin a-si:h sensor. The potential of a pixel sensor based on this novel tehnology has been investigated together with the design of adapted integrated iruits and together with the haraterization of the a-si:h eletrial properties, i.e. signal indution speed, detetion harateristis and radiation hardness. II. TFA SENSORS DEVELOPED The a-si:h sensor is built on top of the ASIC by onseutive depositions of n-doped, intrinsi and p-doped (n-ip) a-si:h films. Depositions have been performed at the Institute of Miro-tehnology of Neuhâtel (IMT) by Very High Frequeny Plasma Enhaned Chemial Vapor Deposition at 70 MHz and 200 C using an hydrogen dilution of Silane, and a deposition rate of 15.6 Å/s was obtained [7]. The n-i-p diodes were deposited on different ASICs developed

2 for our studies. The n-layer and p-layer have high defet density resulting in the reombination of harge arriers generated in these layers. The doped layer thus only permit the appliation of reverse voltages on the i-layer and are therefore deposited very thin ( 30 nm). The diode thikness is then equivalent to the i-layer thikness. The bottom ontats are defined by metal strutures integrated in the ASIC and designed in the last metal layer of the tehnology. The different metal strutures integrated in the ASIC are interonneted through the n-layer. In order to avoid a patterning of this layer, it is speially designed with low ondutivity providing pixel isolation. A Transparent Condutive Oxide (TCO) layer is deposited on the p-layer and defines the global top eletrode; this TCO layer is usually made of ITO (Indium Tin Oxide). A shemati ross setion of a TFA detetor is presented in Fig 1. external aess. Openings are normally performed inside the metal pad so that the 5 µm thik passivation layers indue unevenness of the iruit surfae and of the TFA detetor (Fig 1). The asihtest hip integrates metal strutures for whih the openings were performed outside the metal strutures in order to study the impat of the passivation layer step on the detetor performane. Ring of output metal pads Bottom eletrodes integrated: test strutures a-si:h sensor Fig. 2. Piture of a TFA detetor based on a 20 µm thik n-i-p a-si:h diode deposited on top of an asihtest integrated iruit. Fig. 1. Shemati ross setion of a TFA detetor based on the deposition of n-i-p a-si:h layers and TCO eletrode on top of an ASIC with 3 levels of metal. a-si:h sensors were deposited on 3 different ASICs (AFP, MACROPAD and asihtest) eah designed in a CMOS 0.25 µm tehnology. The 3 integrated iruits ontain different integrated metal strutures and pre-amplifiers optimized for different speed and noise performane. The AFP hip ontains µm 94 µm metal pads eah onneted to an integrated Ative Feedbak Pre-amplifier whih has a 5 ns peaking time for an Equivalent Noise Charge (ENC) of about 400 e- rms [10]. The MACROPAD iruit onsists of 48 otagonal pads with about 140 µm width and 380 µm pith, eah pad being onneted to a transimpedane amplifier followed by a shaper stage. The iruit has a peaking time of about 160 ns, and an ENC of 27 e - rms [11]. The asihtest hip has been designed as a test hip to haraterize the TFA tehnology. Different metal strutures are integrated in the enter of the hip and define the detetor ative area (Fig. 2). 21 strutures with areas of µm 2 or µm 2 and with different shapes are integrated to study the leakage urrent of the a-si:h sensor and the detetor pixel segmentation. Eah struture is onneted to urrent mirror stages whih provide a urrent gain of 10 2 to Final proess steps during the fabriation of the ASIC onsist in depositing passivation layers on top of the hip and in opening windows in these layers on top of eah metal pad to permit 24 other metal strutures integrated in the enter of the iruit are eah onneted either to a fast pre-amplifier with 5 ns peaking time for an ENC of 160 e - rms or to a low noise preamplifier with a 25 ns peaking time for an ENC of 70 e - rms. The strutures are otagons with an area of µm 2 and 2070 µm 2 and strips with an area of 1864 µm 2 for a pith varying from 23 µm to 3 µm. The strip strutures have been integrated in order to study the potential of the TFA tehnology for high resolution strip detetors. III. EXPERIMENTAL RESULTS A. Eletrial haraterization and pixel segmentation Low dark reverse bias urrents of the a-si:h sensor are needed in order to be able to fully deplete the sensor and to have an adequate signal over noise ratio. The leakage urrent of thin ( 1 µm) n-i-p a-si:h sensors has been shown to originate from thermal generation in the intrinsi layer [12]. However, thiker layers are needed for partile detetion. High reverse biases are required for full depletion of the sensor resulting in elevated eletri field (> 10 4 V/m) in the depleted region, and in inreased leakage urrents [13]. Results obtained on thik a-si:h sensors (up to 32.6 µm) deposited on a glass substrate are presented in [2]. TFA detetors developed on a MACROPAD and on an AFP hip showed leakage urrent up to 3 orders of magnitude higher than for similar sensors deposited on a glass substrate ([11], [14]), showing a strong inrease of the a-si:h sensor leakage urrent linked to its deposition on an ASIC. The thermal generation of harges is enhaned by the eletri field by Poole-Frenkel mehanisms [13], so that loal elevated fields in a pixel might reate higher leakage urrents. The pixel ative area of a detetor made of a 32.6 µm thik diode and of an AFP hip has been studied using a Sanning Eletron Mirosope (SEM). A 20 kev eletron beam sans a pixel and the signal indued on the bottom eletrode is amplified by the integrated pre-amplifier and fed bak to the

3 SEM operating system. Eletron Beam Indued Current (EBIC) images are built with ontrast proportional to the amplitude of the indued signal, whih depends on the depleted thikness, on the internal eletri field whih modifies the drift veloities of arriers generated and on the weighting field. The SEM piture and EBIC image obtained on a pixel are presented in Fig. 3, for a 260 V reverse bias applied to the sensor. Fig. 3. EBIC study of a pixel of a TFA detetor made of a 32 µm thik a- Si:H sensor on top of an AFP hip. Left: SEM piture showing geometrial effets aused by the passivation layer steps. Right: EBIC image. White zones orrespond to higher indued urrent. The SEM piture shows a peripheral edge and some fissures at the orners of the pixel, aused by the unevenness of the ASIC surfae as represented in Fig. 1. The EBIC image learly shows the effet of this partiular geometry on the detetor eletri field: higher EBIC urrents are observed at the edges and orners, indiating higher eletri fields at these regions whih might ause the higher leakage urrents. The additional urrents suggested by the EBIC measurements have been studied on different detetors based on the asihtest hip by measuring leakage urrents on the different metal strutures integrated. Tendenies measured on the different samples are illustrated by results obtained on a 20 µm thik sample (Fig. 4). otagon with an area of m 2 presents a urrent density a fator 6 lower than an otagon with an area 9 times smaller. This is a first indiation of urrents arising from the pixel edges and orners, as the ratio perimeter/area is higher for smaller strutures, and as the pixel orners should not depend on the pixel size. A strip struture has a higher leakage urrent than an otagon beause of a higher perimeter/area ratio (Fig. 4). These differenes demonstrate a strong ontribution to the TFA detetor leakage urrent arising from the periphery and orners of a pixel. Otagonal strutures with openings in the passivation outside the metal pad are also integrated in the asihtest hip. A lear redution of leakage urrent density is observed for these strutures (Fig. 4), saling with the distane of the opening loation to the metal pad edges. Opening the passivation layers outside the metal pad then leads to a redution of the additional urrents, as these urrents ontribute to the total leakage urrent of the pixel only after ondution through the low ondutivity n-layer. The mehanisms responsible of the elevated dark urrents at high reverse biases have been studied by measuring the evolution of the urrent with temperature (Fig. 5). Fig 5. Evolution of leakage urrent with temperature measured on a 20 µm thik a-si:h sensor on top of an asihtest iruit. The leakage urrent indued by thermal generation enhaned by eletri field by Poole-Frenkel mehanisms an be written as [13]: I = I 0 exp( EA / kt ) (1); EA = EG 2 γ E / (2). Fig. 4. Leakage urrent density measured on a TFA detetor made of a 20 µm thik sensor on top of an asihtest hip. The sensor is not fully-depleted for the range of applied biases. L-otagon is an otagonal struture and L-strip a strip struture, both strutures with an opening inside the metal pad and with an area of m 2. S-in, S-out6 and S-out40 are otagons with an area of m 2 and with passivation openings respetively inside the metal pad and outside, at 6 µm and at 40 µm from the metal pad edges. The leakage urrent density of the different strutures varies exponentially with the square root of the applied voltage. An E G is the material bandgap, E A is the ativation energy, i.e. the energy needed to ativate ionisable defets and γ is the Poole- Frenkel onstant. Experimental data are well fitted by equations (1) and (2) (Fig. 5), showing that the leakage urrent an be attributed to Poole-Frenkel redution of the ativation energy, enhaning the thermal generation of harges. The ativation energies have been extrated for different strutures and show similar results, thus demonstrating that additional urrents arising from the pixel edges and orners are also mainly determined by this effet. The ativation energy dereases for inreasing applied voltage and is equal to 0.49 ev and 0.42 ev for 80 V and 160 V reverse biases.

4 B. Signal indution and speed in TFA detetors The motion of harge arriers generated by radiation inident on an a-si:h sensor indues a urrent on the sensor eletrodes. The indued urrent speed is a ruial parameter of the detetor performane and has been studied by measuring a TFA detetor response to a 660 nm pulsed laser light. The laser pulse impinges on the top eletrode and sine 660 nm photons have a mean free path of 1 µm in a-si:h, eletronhole pairs are generated lose to the p-i interfae of the sensor. Generated eletrons drift down through the whole depleted thikness and generated holes drift on a short distane up to the p + layer where they are absorbed or reombine. The signal indued on the detetor bottom eletrode integrated in the ASIC is then mainly determined by the eletron motion. The moving harges indue a signal proportional to the total harge, to the veloity of the harges and to the weighting field of the detetor [15]. For under-depleted a-si:h sensors, the non-depleted i-layer has a low ondutivity (σ < S m -1 ), so that a urrent is indued only while the reated harge paket is moving [15]. Tests with a 660 nm pulsed laser have been performed on a 32.6 µm thik sensor deposited on an AFP iruit. Pre-amplifier output signals measured for 3 ns FWHM pulses for different detetor reverse biases are presented in Fig. 6. Fig 6. Response of a 20 µm thik a-si:h sensor on top of an AFP iruit to a 3 ns FWHM 660 nm laser pulse for different detetor biases. Fast deay of the output pulses orrespond to the eletron drift, long tail of signal orrespond to hole drift. The signal rise orresponds to the pre-amplifier peaking time (~ 5ns) and to the time of generation of harges, whih orresponds to the laser pulse duration here of 3 ns. The peaks orrespond to the signal indued by the eletron paket moving in the high field region. The paket drifts under deaying eletri field so that the indued signal then dereases. By inreasing the detetor reverse voltage, eletri field and drift distane (given by the width of the depleted region) are inreased resulting in global onstant signal deay time of about 20 ns. Tails are observed in the output signals, varying with the eletri field. Signal indued by holes as well as eletron reemitted from deep traps auses this long tail. Hole transport is dispersive and hole mobility is field dependent [16] so that variations of the detetor bias strongly impat the hole motion. A simple model of the urrent indued by the eletron motion has been developed onsidering an eletri field linearly deaying from its maximum at the p-i interfae down to the end of the depleted region where it is nil. Simplified expressions for the eletri field, the depleted thikness d and the eletron motion x(t) are presented in [1]. Eletron transport is non-dispersive in a-si:h at room temperature and its drift mobility µ d is onsidered as onstant, so that eletrons reated at the p-i interfae will have a motion along the whole depleted thikness d defined by: x () t d( 1 exp( t / τ )) = (3); * ( qµ N ) τ = ε ε (4) 0 asi / d db The eletron indued signal time onstant τ depends on the the inverse of the produt drift mobility-density of ionized dangling bonds. The urrent reated by eletrons generated at the p-i interfae and drifting in the deaying eletri field an be expressed as: q N d A t for t < t d 1 tot τ t1 I() t = q N d A exp( ( t t1 )/ τ ) for t > t 1 dtot τ (5) The fator A is equal to 2τ /(2τ +t 1 ) where t 1 is the duration of the laser pulse. d tot is the sensor thikness (1/d tot is a simplified expression of the sensor weighting field) and q N is the total harge moving.) Simulations of the pre-amplifier output response to input urrents defined by equation (5) have been performed with HSPICE. Current signals rising in ~ 3 ns (orresponding to the laser pulse width) and then deaying as defined by equation (5) permit a very good between the simulated and measured AFP output response (Fig 7). Indued urrent (µa) 3 ns rise time that orresponds to the laser pulse width 70 V 260 V τ 1 ~ 5.7 ns Time (ns) τ 1 ~ 5.7 ns 160 V τ 1 ~ 5.6 ns Fig 7. Left: Calulated urrents reated by eletrons generated at the p-i interfae during 3 ns and then drifting in the dereasing eletri field. Right: simulation of the AFP output response to urrents presented on left. A good fitting of the fast deay omponent is observed with measured signals. The parameter τ an be extrated from these simulations and we obtained τ = 5.6 ns. The produt µ d N db * an also be extrated and is equal to m -1 V -1 s -1. For an expeted value of N db * of m -3 the eletron mobility an be extrated and µ d ~ 2 m 2 V -1 s -1, whih is in agreement with

5 usual values of 1 to 5 2 m 2 V -1 s -1 in a-si:h. The high eletri field inside the a-si:h sensor (> 10 4 V/m) leads to reasonable drift veloity of the eletrons even though they have a low mobility, and signal indued by the eletrons lasts for ~ 16.8 ns. The signal tails are indued by the movement of holes on a mean distane of 1 µm. A slow signal strongly depending on the eletri field is observed. For partile detetion, only a small fration of the signal from holes will then be readout, whih will limit the detetion effiieny. Full depletion of the sensor is needed for an optimized signal from eletrons but depletion of the sensor annot be studied via standard C-V measurements as it is usually done with rystalline silion sensors. Due to the high resistivity of the non-depleted i-layer, no variations of apaitane are observed hanging V and therefore d. We propose a tehnique based on the photo-generation of eletron-hole pairs lose to the p-i interfae with the 660 nm pulsed laser. Most of the readout harge depends on the integration of the urrent indued by eletron motion. For a time t > 3 τ, this harge Qe an be obtained by integrating equation (5) and depends on the eletron harge moving and on the ratio of the depleted thikness over the sensor thikness: density of ionized dangling bonds an also be extrated from this measurement and N db * ~ m -3. The extrated value is in agreement with the expeted value of N db * ~ m -3. C. Radiation detetion with TFA detetors Diret detetion of harged partiles and of soft X-rays have been arried out on a TFA detetor based on a 15 µm thik sensor deposited on a MACROPAD iruit. Signals from the detetor were readout on an osillosope in self-trigger mode. Signal pedestals and peak amplitudes were reorded via a Labview program. A maximum reverse bias of 145 V ould be applied to the sensor, and the noise level of the readout eletronis was measured equal to 40 e - rms. The spetrum obtained with 5.9 kev X-rays from a Fe 55 soure is shown in Fig 9. Qe = qnd / (6); d 2ε 0 ε /( qn V d tot * asi db ) = (7) Tests have been arried out on a 20 µm thik sensor deposited on an asihtest hip. Strutures onneted to preamplifiers with 70 e - rms noise and 25 ns peaking time have been haraterized. The readout eletroni peaking time is higher than 3 τ so that the measured output pulse maximum orresponds to the harge Qe readout. Evolution of Qe for varying detetor biases is presented in Fig 8. Fig 8. Charge reated by a 3 ns FWHM 660 nm laser pulse as a funtion of the reverse voltage applied on the a-si:h sensor. Results obtained on a 20 µm thik sensor deposited on an asihtest hip for 3 different laser amplitudes are shown. Qe inreases with the square root of the voltage, whih is onsistent with equations (6) and (7). The slope depends on qn and then inreases for inreasing amplitude of laser. Saturation starts to our for a reverse voltage of ~ 193 V, whih then orresponds to the depletion voltage, as the reated harge saturates for d = d tot. This method is suessful in determining experimentally the a-si:h sensor depletion. The Fig 9. Spetrum of 5.9 kev X-rays from Fe 55 soure obtained on a 15 µm thik a-si:h sensor deposited on a MACROPAD iruit, for a reverse bias of 145 V and a threshold of 200 e -. A broad peak is observed with a peak harge at about 650 e - rms. The signal orresponds to harges indued by the transport of both eletrons and holes. Fast onstrution of the signal indued by the eletron drift has been demonstrated, but low and field dependent signal indued by holes has been observed. The total harge reated by a partile or a photon in the a-si:h sensor will depend on the loation of reation of the eletron-hole pairs. The signal orresponding to pairs reated lose to the p-i interfae will be built prinipally by the eletron drift and is expeted to be omplete. As pairs are reated further into the depleted region, lower signal is reated from eletron drift and higher ontribution from holes is expeted. However, due to the low drift mobility of holes, only part of the signal from holes is readout. A mean energy of eletron-hole pairs in a-si:h of 4.8 to 6 ev [17] is onsidered. X-rays interating lose to the p-i interfae indue a maximum signal and a total harge of 1000 to 1230 e - is expeted, while a maximum signal of about 900 to 1000 e - is readout. X-rays reating pairs further into the sensor will indue a lower signal beause of a lower olletion of signal from holes, thus the peak has lower amplitude at about 650 e -. No saturation of the peak harge was observed, as a peak harge of 510 e - and of 582 e - were readout for applied biases of 110 V and 130 V even though full depletion of the film is expeted for biases higher than 110 V. This an be explained by a non full

6 depletion of the sensor and by the inrease of the signal reated by holes as the internal eletri field is inreased. IV. CONCLUSION TFA detetors based on n-i-p a-si:h sensors with thikness from 15 µm to 32.6 µm and on 3 different integrated iruits have been developed. Higher leakage urrents have first been measured on TFA detetors in omparison to similar a-si:h sensors deposited on a glass substrate. These high leakage urrents prevent the full depletion of sensors with thiknesses up to 32.6 µm. However, fully depleted sensors with suh thikness are needed for optimized detetion effiieny, so that the redution of the leakage urrent is a ruial parameter for the development of this tehnology. Leakage urrents an be attributed to field enhaned thermal generation of harges by Poole-Frenkel mehanisms. The ativation energy dereases for inreasing bias voltage and equals ~ 0.42 ev for a 160 V reverse voltage applied to a 20 µm thik diode. High urrents have been found originating from the pixel edges and orners. These additional urrents are aused by the unevenness of the ASIC surfae due to the passivation layer steps. We have shown that opening the passivation layers outside metal pads permits a redution of the leakage urrent. Signal indution and speed have been studied and a onstrution of signal from eletron drift in a maximum time of 16.8 ns has been shown. The laser measurement tehnique permits to measure the full-depletion voltage whih has been found to be ~ 193 V for a 20 µm thik sample. Full-depletion is required to obtain a maximum signal from eletrons generated in the sensor depleted region. Hole mobility is field dependent and therefore the signal indued by hole motion in few tens of ns (whih orrespond to the pre-amplifier shaping time) strongly depends on the reverse voltage applied to the sensor and will not be omplete. Detetion of 5.9 kev X-rays has been performed on a TFA detetor with a 15 µm thik sensor. In order to optimize the detetor effiieny for partile detetion, thiker a-si:h sensors fully depleted are needed. They would permit a higher interation volume for harged partiles and at the same time higher internal fields thus enhaning the detetion effiieny. Radiation hardness of the developed detetors is a ruial parameter and is under study. Instability of the sensor leakage urrent has been observed during the development and haraterization of this novel tehnology. Moreover, long term stability has not been haraterized and proven yet and is a ruial parameter under study. Finally, problems of yield of a-si:h sensors deposited on ASIC have been observed and need further investigations. Snigireva for the EBIC measurements performed at the ESRF in Grenoble. REFERENCES [1] P. Jarron, G. Anelli, S. Commihau, M. Despeisse, G. Dissertori, C. Miazza, D. Moraes, A. Shah, G. Viertel, N. Wyrsh, Nul. Instr. And Meth. In Phys. Res. A518 (2004) pp [2] N. Wyrsh, S. Dunand, C. Miazza, A.Shah, G. Anelli, M. Despeisse, A.Garrigos, P. Jarron, J. Kaplon, D. Moraes, S.C. Commihau, G. Dissertori and G.M. Viertel, Physia Status Solidi (), 2004, Vol. 1(5), pp [3] N. Kishimoto, H. Amekura. N. Kishimoto, H. Amekura, K. Kono, C.G. Lee, Journal of nulear materials, (1998) [4] J.R. Srour, G. J. Vendura, D. H. Lo, C.M.C. Toporow, M. Dooley, R.P. Nakano and E.E. King, IEEE Trans. On Nul. S., vol. 45, no. 6, (1998), pp [5] V. Perez Mendez et al., Nul. Instr and Meth. A273, 127 (1988). [6] A. Kolodziej. Opto-eletronis review 12(1), (2004). [7] N. Wyrsh, C. Miazza, S. Dunand, A. Shah, N. Blan, R. Kaufmann, L. Cavalier,G. Anelli, M. Despeisse, P. Jarron, D. Moraes, A. G. Sirvent, G. Dissertori, G. Viertel. Proeedings of the MRS Spring Meeting, San Franiso, April 2003, MRS Vol. 762, pp [8] J. Dubeau, T. Pohet, L. A. Hamel, B. Equer, A. Karar. Nul. Instr. And Meth. In Phys. Res. B, 1991, Vol. 54, pp [9] C. Hordequin, A. Brambilla, P. Bergonzo, F. Foulon. Nul. Instr. And Meth. In Phys. Res. A, 2001, Vol. 456, pp [10] G. Anelli, K. Borer, L. Casagrande, M. Despeisse, P. Jarron, N. Pelloux and S. Saramad. Nul. Instr. And Meth. In Phys. Res. A 512 (2003), p [11] D. Moraes, G. Anelli, M. Despeisse, P. Jarron, J. Kaplon, A.Garrigos Sirvent, N. Wyrsh, C. Miazza, S. Dunand, A. Shah, G. Dissertori, G. Viertel. J. Non-Cryst. Solids (2004), pp [12] F. Meillaud, A. Shah, C. Droz, E. Vallat-Sauvain, C. Miazza, to be published in Solar Energy and Materials and Solar Cells. [13] J. B. Chévrier and B. Equer, J. Appl. Phys. 76 (11), 1994, pp [14] M. Despeisse, G. Anelli, S. Commihau, G. Dissertori, A. Garrigos, P. Jarron, C. Miazza, D. Moraes, A. Shah, N. Wyrsh, G. Viertel. Nul. Instr. And Meth. In Phys. Res. A518 (2004), pp [15] W. Riegler, Nul. Inst. And Meth. In Phys. Res. A, 2004, Vol. 535, pp [16] Q. Gu, E.A. Shiff, J-B. Chévrier and B. Equer. Phys. Rev. B, 1995, Vol. 52, n 8, pp [17] S. Najar, B. Equer and N. Lakhoua. J. Appl. Phys., 1991, Vol. 69, No. 7, pp ACKNOWLEDGMENT The authors are glad to thank olleagues and ollaborators for their important ontribution and help in the work presented: W. Riegler for very useful disussions, C. Miazza and S. Dunand for the depositions of the a-si:h sensors, D. Campos- Garia and M. Fransen for their help and J. Morse and I.