Sie sind auf Seite 1von 7

Electrical properties of plasma enhanced chemical vapor deposition a-Si:H and aSi1xCx:H for microbolometer applications

Hang-Beum Shin, David Saint John, Myung-Yoon Lee, Nikolas J. Podraza, and Thomas N. Jackson Citation: Journal of Applied Physics 114, 183705 (2013); doi: 10.1063/1.4829013 View online: http://dx.doi.org/10.1063/1.4829013 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/114/18?ver=pdfcov Published by the AIP Publishing

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

JOURNAL OF APPLIED PHYSICS 114, 183705 (2013)

Electrical properties of plasma enhanced chemical vapor deposition a-Si:H and a-Si12xCx:H for microbolometer applications
Hang-Beum Shin,1,2 David Saint John,3 Myung-Yoon Lee,2 Nikolas J. Podraza,4,5 and Thomas N. Jackson2,3
LG Chem, Corporate R&D, Daejeon 133-791, South Korea Department of Electrical Engineering, Penn State University, University Park, Pennsylvania 16802, USA 3 Department of Material Science and Engineering, Penn State University, University Park, Pennsylvania 16802, USA 4 Department of Physics and Astronomy, University of Toledo, Toledo, Ohio 43606, USA 5 Wright Center for Photovoltaics Innovation and Commercialization, University of Toledo, Toledo, Ohio 43606, USA
2 1

(Received 17 August 2013; accepted 21 October 2013; published online 11 November 2013) Electrical properties for resistive microbolometer sensor materials including resistivity, temperature coefcient of resistance (TCR), and normalized Hooge parameter were explored in n-type a-Si:H and a-Si1xCx:H prepared by plasma enhanced chemical vapor deposition. The complex dielectric function spectra (e e1 ie2) and structure were measured by spectroscopic ellipsometry. Two-dimensional drift-diffusion simulations were used to understand the band-tail slope dependency C 2013 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4829013] of TCR and 1/f noise. V

I. INTRODUCTION

II. EXPERIMENTAL DETAILS

Hydrogenated amorphous silicon (a-Si:H) prepared by plasma enhanced chemical vapor deposition (PECVD) and vanadium oxide prepared by ion-beam deposition are commonly used as the thin lm imaging layer in infrared sensing uncooled microbolometers.13 Compared with vanadium oxide, more is known about the transport and band structure of PECVD a-Si:H, perhaps owing to its variety of device applications including solar cells, photodetectors, and thin lm transistors.4,5 However, the temperature dependence of conductivity in a-Si:H lms has not still been fully explained in terms of band structure and relevant defect states. Variations in the deposition process have been performed to achieve higher temperature coefcient of resistance (TCR) for sensitivity and lower 1/f noise at a given resistivity level in an effort to achieve better microbolometer performance. We seek to investigate the physics of TCR and 1/f noise. Amorphous lms typically have defect states or trap distributions in the bandgap, i.e., band-tail and gap-states, which originates from imperfections such as vacancies, micro-voids, or dangling bonds.6,7 The trap distributions interact with temperature by Fermi-Dirac distributions8 and affect the resulting TCR and 1/f noise. For microbolometers applications, the following electrical properties determine the device performance: pixel resistance (should be well matched to standard read-out circuitry), TCR (in the case of resistive microbolometer), and systematic noise at the frame frequency. The resistivity, TCR, and 1/f noise characteristics of the imaging layer in turn directly affect the electrical properties of the microbolometer pixel.9,10 Using n-type hydrogenated amorphous silicon and carbon alloys (a-Si1xCx:H), electrical conduction was investigated for relatively ordered and disordered amorphous networks. Also, a Sentaurus11 simulation has been used to correlate measured TCR and 1/f noise with interpretations about the band structure and tails.
0021-8979/2013/114(18)/183705/6/$30.00

thick n-type a-Si:H or a-Si1xCx:H lms were 500 A prepared in a load-locked system with a base pressure of 106 Torr using capacitive radio frequency (13.56 MHz) PECVD with 6 in. diameter shower head. Films were deposited onto crystalline silicon (c-Si) wafers coated with of thermally grown silicon dioxide (SiO2). The lms 1000 A were deposited using a mixture of silane (SiH4, >99.99% purity), phosphine (PH3), hydrogen (H2), and methane (CH4) precursor gases. The gas ow ratios used were the hydrogento-silane ratio, R [H2]/[SiH4] 8; the doping gas ratio, D [PH3]/[SiH4], which ranged from 0  D  0.039; and the alloying gas ratio, Z [CH4]/{[CH4] [SiH4]}, which ranged from 0  Z  0.33. Other deposition parameters included a substrate temperature Ts 200  C, RF plasma power density of 0.33 W/cm2, and a process pressure of 0.5 Torr. Ellipsometric spectra (in D, w) were collected for the lms over a range from 0.75 to 5.15 eV at multiple angles of incidence from 50 to 80 using a multichannel ellipsometer based on dual rotating compensator principles.12 The complex dielectric function spectra (e e1 ie2) and structural parameters (bulk lm thickness and surface roughness thickness) were extracted using a least squares regression analysis with an unweighted error function.13 The structural model thermally consisted of a semi-innite c-Si substrate/1000 A grown SiO2/bulk a-Si:H or a-Si1xCx:H lm/surface roughness/air ambient. e for the bulk a-Si:H or a-Si1xCx:H lm is represented by a Tauc-Lorentz oscillator14 and a constant additive term to e1, e1. A Bruggeman effective medium approximation of 0.5 bulk material and 0.5 void fractions was used to represent e for the surface roughness.15 e for c-Si and SiO2 are from Ref. 16. Deposition was followed by a subsequent device patterning process for electrical measurements including sheet resistance, TCR, and normalized Hooge parameter, a
C 2013 AIP Publishing LLC V

114, 183705-1

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

183705-2

Shin et al.

J. Appl. Phys. 114, 183705 (2013)

representative value for 1/f noise. Two photolithographic layers of patterning was applied, dening a-Si:H and a-Si1xCx:H patterns with CF4 O2 dry etching and subsequently sputtered titanium lift-off for electrode formation. Transmission line method (TLM) from patterns with ve different spacings was used for extracting accurate sheet resistance and contact resistance, using a 1 to 1 V sweep of HP 4156 parametric analyzer. TCR measurement was facilitated with a computer controlled Duo-chuck and HP 4145 parametric analyzer. A 2 to 2 V voltage sweep determined electrical resistance for each temperature from 22 to 55  C. TCR was extracted from the slope of the natural logarithm of resistance plotted versus temperature.17 Low frequency 1/f noise has been characterized by Hooges empirical equation. The material inherent value, the Hooge parameter, aH, which is independent of measuring voltage or current, is dened as   SI f aH 1 aH 1 1 ; (1) 2 N f n Vf I where SI(f) is the power spectral density or the uctuation of the current, N is the number of carriers, and the f is the probing frequency.18,19 The equation indicates that the uctuating portions in the square of electrical current are inversely proportional to the number of carriers at a given frequency, i.e., 1 Hz. When the mobility is known, N can be derived from the measured resistivity and the conductivity relationships, r qln. For amorphous materials, however, it is difcult to determine the mobility due to the presence of trap distributions. Although the mobility was not measured, a normalized Hooge parameter, described as aH/n can be derived using the relation, N nV. In this case the word normalized means with respect to the volume as well as measuring current. The normalized Hooge parameter is a stand-alone parameter without known mobility or carrier concentration and is applicable to evaluate material characteristics as long as the dimensions are known. As a result, it does not require measuring the ambiguous carrier concentration or the mobility. The normalized Hooge parameters instead of Hooge parameters are used in Refs. 9 and 10, for example. The normalized Hooge parameter is still useful for evaluating 1/f noise for each material, which means the uctuating portions in owing current at a given volume is determined by the pixel size of microbolometer at 1 Hz, as expressed in Eq. (1). This noise parameter was extracted from the power spectral density over the frequency range of 0.1 to 2 kHz using a low noise current amplier (Stanford Research 570) with an internal battery source voltage up to 5 V. Once the amplied voltage signal was band-pass ltered for antialiasing within 0.1 and 1 kHz, the output voltage was collected into 16-bit data acquisition board (NI-6036E) instead of conventional way of using dynamic signal analyzer (DSA). Automated data averaging of more than twenty measurements increases the accuracy of the noise measurement. Since the empirical Hooge equation assumes a volume dependence, a lithographic mask was designed to include three devices having identical resistance values but different volumes of 5, 20, and 80 lm3 as shown in Fig. 1. Those three

FIG. 1. Normalized power spectral density as functions of frequency for three a-Si:H resistors with equivalent resistance but different volumes.

power spectral densities were used to extract the normalized Hooge parameter. Thus, normalized Hooge parameter, as a representative value of 1/f noise, is a normalized standard deviation value at 1 Hz with respect to current as well as volume.

III. RESULTS A. n-type a-Si:H

Fig. 2(a) shows the changes in the resistivity, TCR, and normalized Hooge parameter for a series of a-Si:H lms

FIG. 2. Resistivity, TCR, and normalized Hooge parameter (a) as a function of doping gas ratio, D [PH3]/[SiH4], for a-Si:H and (b) methane gas ratio, Z [CH4]/{[CH4] [SiH4]}, (D 0.019), for a-Si1xCx:H.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

183705-3

Shin et al.

J. Appl. Phys. 114, 183705 (2013)

prepared at varying doping gas ratio, D. As D was initially increased until D 0.02, more phosphorus dopant atoms incorporated into the lms, and the resistivity decreased. The resistivity value saturated for D > 0.02, so we believe that the maximum doping efciency was obtained at D 0.02. e obtained from spectroscopic ellipsometry measurements for these lms is shown in Fig. 3(a) and indicates that these materials are amorphous.20 For D < 0.02, there was not a signicant change in e with D. This behavior implies that the structure of the amorphous networks were not signicantly altered by low amounts of phosphorous which is expected for substitutional doping.21,22 The lm prepared at D 0.039 showed a lower amplitude of e compared to the other lms, indicative of a structural change in the network. The lower amplitude may imply a decrease in material density or in the probability of particular electronic transitions arising from a more defective structure. Overall, the normalized Hooge parameter and TCR appears to be proportional to the resistivity value of these n-type a-Si:H lms.
B. n-type a-Si12xCx:H

C. Comparison

Fig. 4 compares the TCR and normalized Hooge parameter for the a-Si:H and a-Si1xCx:H lms exhibiting resistivities of interest for microbolometer applications (<10 000 X cm). Based on the range of resistivity, it can be inferred that these two data sets have similar distances between the conduction band and the Fermi level. These two series of samples suggest that the current transport depends on the degree of disorder in the network, provided primarily by carbon incorporation in this case. Shallower band-tail slope in the a-Si1xCx:H results in lower TCR at a given resistivity but also a lower normalized Hooge parameter for a given TCR. As a function of methane ow during deposition, the bandgap and absorption edge shift to higher energies, indicating that a shallower band-tail slope is present in lms presumably incorporating more carbon. Absorption features in e have been previously linked to band tail behavior.25,26,28
IV. TWO-DIMENSIONAL DRIFT DIFFUSION SIMULATION

A series of a-Si1xCx:H lms were prepared at xed D 0.019, the value for maximum doping efciency for a-Si:H, and variable alloying gas ow ratio, Z.23,24 Carbon content in the lms was controlled by increasing methane gas ratio, Z [CH4]/{[CH4] [SiH4]}, from 0 to 0.33, and was evaluated by using e.2527 The carbon content in the a-Si1xCx:H lms was compared with the estimates developed by Basa et al., who correlated variations in e with carbon content extracted from Rutherford backscattering.28 Based on the behavior of e in Ref. 28, the a-Si1xCx:H lms in this series prepared at Z 0 to 0.33 contained less than <7% carbon relative to silicon.

a-Si:H device has been modeled and reported in several ways.4,29,30 The dependence of TCR on band structure can be simulated crudely by calculating Fermi-Dirac distribution function as a function of temperature. Commercial software (Sentaurus11) can simulate material parameters as well as device structures, such as band structure, electrical conduction model in the density of state, and tail/gap state and various electrical properties. A simplied model was constructed with conduction/valence band-tail and gap states to examine the temperature dependence of electrical conduction, as illustrated in Fig. 5(a). Most of the carriers in the a-Si:H are trapped below the mobility gaps and contribute little to the electrical conduction.31,32 Only the small portion of the carriers at the end of Fermi-Dirac distribution are in extended states and provide most of the electrical conduction, as shown in Fig. 5(b). The number of carriers above the mobility gap or free carriers increases with temperature. As a result, resistivity decreases. The resistivity at elevated temperature can also be calculated by integrating a density of states and Fermi-Dirac distribution function that varies with temperature. TCR was extracted in the same manner as for the experimental data by computing the slope of the natural

FIG. 3. Complex dielectric function spectra (e e1 ie2) for the (a) doping gas series and (b) methane gas series.

FIG. 4. Resistivity, TCR, and the normalized Hooge parameter for a-Si:H prepared as a function of D and a-Si1xCx:H prepared as a function of Z.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

183705-4

Shin et al.

J. Appl. Phys. 114, 183705 (2013)

FIG. 5. Simulated resistivity changes as a function of temperature by FermiDirac distribution with carriers in the trap distribution (up to the extended state) contributing to the electrical conduction.

logarithm of resistance plotted versus temperature from 27 to 97  C. In the Fig. 5(a), a simplied model was constructed to investigate the electron carrier conduction for comparison with experimental studies of n-type a-Si:H, with hole mobility substantially smaller than the electron mobility. A bandtail trap density distribution was modeled as " # E ET 0 ; Ntail E NT 0 exp ETsig

(2)

where NT0 is a band-tail trap concentration, which ranged from 1019 to 1022 cm3, ET0 is a conduction band or extended states position, 1.72 eV, and ETsig is a characteristic value of band-tail slope, which ranged from 0.02 to 0.08 eV. In addition, electrical conduction below the mobility gap, i.e., hopping conduction, was neglected for a simplied calculation. For each band-structure, the resistivity and TCR were calculated depending on the phosphorous doping concentration from 1 1016 to 1 1019 cm3. Fig. 6 shows the resistivity and TCR behavior predicted according to the doping concentration and band structure

FIG. 6. Simulated resistivity and TCR change as a function of the doping concentration depending on the band-tail distribution (ETsig). A variation in the bandtail slope (a), (b), (c) and the trap concentration (d), (e), (f) showed that the variation in band-tail slope has signicantly affected the resistivity-TCR relation.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

183705-5

Shin et al.

J. Appl. Phys. 114, 183705 (2013)

conduction band of kT is increasing as the band-tail gets shallower. By this reason, the noise can be expected to be higher with shallower band-tail or with higher TCR.
V. DISCUSSIONS

FIG. 7. At xed resistivity and different band-tail slope (0.03, 0.04, 0.05 eV). (a) Fermi level shifts as a function of temperature (Fermi-level started to be pinned at a higher trap distribution); (b) trap distribution and trapped charge distribution at 300 K (The empty traps (trap distribution lled traps) below the bandgap increases as the band-tail has shallower slope).

model. The calculation suggests that TCR is mainly affected by the band-tail rather than the gap-state distribution. Specically, the shallower band-tail (larger band tail slope) or higher trap concentration has higher TCR at a given resistivity. Figs. 6(a)6(c) show that the shallower band-tail has higher TCRs among the variation in the tail slope. When ETsig 0.05 eV, the simulation results coincide with experimental data points, which are near the range of the reported values 0.0450.094 eV.3335 In Figs. 6(d)6(f), the trap concentration affects the TCR-resistivity relation but not to the extent that the band-tail slope does. For example, increasing the trap concentration with ETsig 0.04 eV does not permit the simulated TCR to match experimental data. Consequently, trap concentration has less effect on the resistivity-TCR relation. Thus, the band-tail contribution dominates the TCR-resistivity relation in these simulations. In Fig. 7, electrostatics determines the Fermi-level position for a given trap distributions. The Fermi-Dirac distribution is a function of temperature and Fermi-level position. Variation of the Fermi-level position is plotted at the three band-tail slopes, ETsig 0.03, 0.04, and 0.05 eV, in the Fig. 7(a). The larger band tail slope, ETsig 0.05 eV, has less movement of the Fermi-level as the temperature variesas if the Fermi-level was pinned. Considering the Fermi-Dirac distribution shape changes as a function of temperature, the larger number of free carriers appears at a xed Fermi-level position. However, at the band-tail slope of ETsig 0.03 eV, the Fermi-level moves downward so that thermal expansion of the Fermi-Dirac distribution was compensated with the longer distance between the conduction band and Fermilevel. As a result, TCR at the band-tail of ETsig 0.03 eV becomes smaller than that of the band-tail of ETsig 0.05 eV. Fig. 7(b) shows the trap distributions and trapped charges as a function of energy at the same conditions as Fig. 7(a). The trap distributions increase as the energy level approaches to the conduction band edges while the lled states are at a similar level. Empty traps are the trap distribution subtracted by the trapped charge density. In the shallower slope of band-tail, ETsig 0.05 eV, empty trap density is much larger than those of the steeper slopes. Among the three slopes of band-tail, the empty trap density beneath the

A negative value of TCR is mainly attributed to the temperature dependency of Fermi-Dirac distribution. Since the Fermi-Dirac distribution shape changes as temperature increases, the number of carriers above the mobility gap also increases at elevated temperatures. Specically in a-Si:H, most of the carriers are trapped in the gap states or tail states, i.e., below the mobility gap, and only a small portion of carriers contribute to the electrical conduction in the extended states so that the lm can have a higher magnitude of TCR than that observed for c-Si.36 For a given band structure, TCR increases as the Fermilevel position is initially located farther from the band edge. Experimental results from a-Si:H in Fig. 3(a) indicates that TCR increases as the Fermi-level position moves farther from the conduction band or the resistivity increases in Fig. 2(a). In that case, the TCR value increases as the Fermi-level position is displaced from the band-edge or at high resistivity. The simulation results shown in Figs. 6(c) and 6(f) implies that TCR always increases as the resistivity increases for a given band-tail slope. Thus, in a sense, TCR is not a direct measure of either band-tail slope or trap concentration due to its dependence on the Fermi-level position. Simulations showed that a shallower band-tail slope has higher TCR, as in Fig. 6(c). Within the working range of a uncooled microbolomter (2050  C), a steeper band-tail slope of c-Si, ideally without a band tail, results in magnitude of TCR  0.2%/K,36 which is much smaller than TCR near 2%4%/K from a-Si:H at similar carrier concentrations. Interestingly, deep traps or gap-states do not affect the TCR-resistivity relation at all in the simulation, suggesting that a temperature sweep may be a good evaluation of the band-tail. Since deep traps or the gap-states were being completely covered with the Fermi-Dirac distribution for a range of temperature, the probing range by temperature was limited only to the band-tail region as shown in the Fig. 7(b). When an amorphous lm has both conduction and valence band-tails, the typical light absorption in the sub-gap measurement suffers from the multiple optical transitional paths. For example, valence band to conduction band-tail or valence band-tail to conduction band can occur.37 The temperature sweep is quite limited to a few kT. As a result, the temperature sweep can selectively characterize the band-tail. Researchers have interpreted 1/f noise as generation/recombination process or trapping/de-trapping at a path of owing current.38 In that perspective, the thermal voltage, kT, portion below the conduction band can be subject to the 1/f noise measurement so that the free carriers in the conduction band are interacting with the trapped charge region within the thermal voltage. As the band-tail slope gets shallower in Fig. 7(b), the 1/f noise is expected to be higher. Thus, the higher TCR leads to higher 1/f noise in the simulation. Experimentally, the incorporation of carbon species can suppress 1/f noise at the expense of TCR, as shown in

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

183705-6

Shin et al.
7 8

J. Appl. Phys. 114, 183705 (2013) P. A. Fedders, J. Non-Cryst. Solids 266269, 218 (2000). A. A. Sherchenkov and A. B. Apalkov, Semiconductors 42(13), 1503 (2008). 9 L. M echin, J. Routoure, S. Mercone, F. Yang, S. Flament, and R. A. Chakalov, J. Appl. Phys. 103, 083709 (2008). 10 C. Barone, S. Pagano, L. M echin, J.-M. Routoure, P. Orgiani, and L. Maritato, Rev. Sci. Instrum. 79, 053908 (2008). 11 See http://www.synopsys.com/Tools/TCAD/DeviceSimulation/Pages/ default.aspx for TCAD Sentaurus 2010, Synopsys. 12 C. Chen, I. An, G. M. Ferreira, N. J. Podraza, J. A. Zapien, and R. W. Collins, Thin Solid Films 455456, 14 (2004). 13 Y. Cong, I. An, K. Vedam, and R. W. Collins, Appl. Opt. 30(19), 2692 (1991). 14 G. E. Jellison and F. A. Modine, Appl. Phys. Lett. 69(3), 371 (1996); 69, 2137(E) (1996). 15 H. Fujiwara, J. Koh, P. I. Rovira, and R. W. Collins, Phys. Rev. B 61(16), 10832 (2000). 16 C. M. Herzinger, B. Johs, W. A. McGahan, J. A. Woollam, and W. Paulson, J. Appl. Phys. 83(6), 3323 (1998). 17 S. K. Ajmera, A. J. Syllaios, G. S. Tyber, M. F. Taylor, and R. E. Hollingsworth, Proc. SPIE 7660, 766012 (2010). 18 F. N. Hooge, T. G. M. Kleinpenning, and L. K. J. Vandamme, Rep. Prog. Phys. 44(5), 479 (1981). 19 A. V. D. Ziel, Proc. IEEE 76(3), 233 (1988). 20 D. B. S. John, H. B. Shin, M. Y. Lee, S. K. Ajmera, A. J. Syllaios, E. C. Dickey, T. N. Jackson, and N. J. Podraza, J. Appl. Phys. 110, 033714 (2011). 21 W. E. Spear and P. G. L. Comber, Solid State Commun. 17(9), 1193 (1975). 22 M. Taniguchi, M. Hirose, and Y. Osaka, J. Cryst. Growth 45, 126 (1978). 23 Z. Hu, X. Liao, H. Diao, G. Cong, X. Zeng, and Y. Hu, J. Cryst. Growth 264(13), 7 (2004). 24 A. Tabata, M. Kuroda, M. Mori, T. Mizutani, and Y. Suzuoki, J. NonCryst. Solids 338340, 521 (2004). 25 M. Losurdo, M. Giangregorio, P. Capezzuto, G. Bruno, and F. Giorgis, J. Appl. Phys. 97, 103504 (2005). 26 E. Pascual, J. L. Andtujar, J. L. Fernamdez, and E. Bertran, Diamond Relat. Mater. 4, 702 (1995). 27 H. Fujiwara, J. Koh, and R. W. Collins, Thin Solid Films 313314, 474 (1998). 28 D. K. Basa, G. Abbate, G. Ambrosone, U. Coscia, and A. Marino, J. Appl. Phys. 107, 023502 (2010). 29 C. Longeaud, G. Fournet, and R. Vanderhaghen, Phys. Rev. B 38(11), 7493 (1988). 30 K. Rerbal, J.-N. Chazalviel, F. Ozanam, and I. Solomon, J. Non-Cryst. Solids 299302, 585 (2002). 31 K. Murayama, Phys. Status Solidi C 8(1), 198 (2011). 32 J. J. Thevaril and S. K. OLeary, Solid State Commun. 151, 730 (2011). 33 N. Hern andez-Como and A. Morales-Acevedo, Sol. Energy Mater. Sol. Cells 94(1), 62 (2010). 34 X. Wen, X. Zeng, W. Liao, Q. Lei, and S. Yin, Sol. Energy 96, 168 (2013). 35 L. Shen, F. Meng, and Z. Liu, Sol. Energy 97, 168 (2013). 36 P. Norton and J. Brandt, Solid-State Electron. 21(7), 969 (1978). 37 L. Jiao, I. Chen, R. W. Collins, C. R. Wronski, and N. Hata, Appl. Phys. Lett. 72(9), 1057 (1998). 38 F. N. Hooge, Physica B 336, 236 (2003). 39 D. Franta, D. Necas, and L. Zajickova, Opt. Express 15(24), 16230 (2007). 40 J. A. Schmidt, M. Hundhausen, and L. Ley, J. Non-Cryst. Solids 266269, 694 (2000).

Fig. 4(b). The volume dependency of the noise in the Hooge equation implies that a larger number of paths for carrier conduction reduce the noise. Likewise, introduction of carbon into the lm increases disorder even over that already present in a-Si:H and is expected to increase the number of the electrical conduction paths.
VI. CONCLUSIONS

We have correlated the effects of band-tails from simulations with the measured electrical and optical properties of n-type a-Si:H and a-Si1xCx:H. In simulations, the Fermi-level position moves more slowly with temperature at a shallower band-tail. Consequently, higher TCR can be achieved at the shallower band-tail for a given resistivity. As seen from the simulation result of TCR-resistivity relation, it is desirable to have the shallow slope as long as the resistivity limit allows. In addition, the 1/f noise as well as TCR increases at the shallower band-tail, owing to the large empty trap-states concentration below the conduction band. Thus, the shallower band-tail is advantageous for TCR but not 1/f noise. Experimentally, carbon incorporation into a-Si:H lms increases disorder in the network27,39 by alloying and resulted in a more shallow band-tail slope than a-Si:H. The relative amount of carbon incorporation is expected to directly affect both the mobility40 and carrier concentration. In agreement with simulation results, carbon incorporation increases TCR and 1/f noise, but the incremental change was moderate relative to variations of a-Si:H itself. Adding carbon can at least result in lower 1/f noise for similar TCR values.
ACKNOWLEDGMENTS

Research was sponsored by the U.S. Army Research Ofce and U.S. Army Research Laboratory and was accomplished under Cooperative Agreement No. W911NF-0-2-0026.
1

T. Schimert, J. Brady, T. Fagan, M. Taylor, W. McCardel, R. Gooch, S. Ajmera, C. Hanson, and A. J. Syllaios, Proc. SPIE 6940, 694023 (2008). 2 J. L. Tissot, C. Trouilleau, B. Fieque, A. Crastes, and O. Legras, OptoElectron. Rev. 14(1), 25 (2006). 3 R. T. R. Kumar, B. Karunagaran, D. Mangalaraj, S. K. Narayandass, P. Manoravi, M. Joseph, and V. Gopal, Smart Mater. Struct. 12(2), 188 (2003). 4 S. Sherman, S. Wagner, and R. A. Gottscho, Appl. Phys. Lett. 69(21), 3242 (1996). 5 C. R. Wronski and R. E. Daniel, Phys. Rev. B: Condens. Matter 23(2), 794 (1981). 6 H. C. Kang, J. Non-Cryst. Solids 261, 169 (2000).

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 143.248.157.20 On: Mon, 24 Feb 2014 02:07:29

Das könnte Ihnen auch gefallen