Symposium Organizers
Harry A. Atwater California Institute of Technology
Virginia Chu INESC Microsistemas e Nanotecnologias
Sigurd Wagner Princeton University
Kenji Yamamoto Kaneka Corporation
Hsiao-Wen Zan National Chiao Tung University
A1: Transport and Electronic Properties
Session Chairs
Tuesday PM, April 18, 2006
Room 3002 (Moscone West)
9:30 AM - **A1.1
Time-resolved Photoconductivity as a Probe of Carrier Transport in Microcrystalline Silicon.
Steve Reynolds 1
1 Institute of Photovoltaics, Forschungszentrum Juelich, Juelich, NRW, Germany
Show AbstractIn this talk we review the techniques of time-of-flight (TOF) and coplanar transient photocurrent (TPC) spectroscopy and describe how they may be used to study electron and hole drift mobilities and density of localized states distributions in disordered electronic materials, with a focus on microcrystalline silicon films and solar cells prepared over a range of structure compositions. Thin film microcrystalline silicon is a complex material, comprising nanometre-scale crystalline grains, often aggregated into sub-micron columns extending in the direction of film growth, amorphous regions, and cracks or voids. Carrier transport is often described in terms of concepts that have been applied successfully to homogeneous disordered materials such as amorphous silicon; transport paths, mobility edges, band tails, dangling-bond defects. The interaction between localized and transport states, involving multiple-trapping or hopping, may be tailored, at least qualitatively, to fit microcrystalline silicon by ascribing tail states to strained bonds at grain boundaries and defects to unsatisfied bonds at column boundaries, and by introducing band offsets and potential barriers between amorphous and crystalline regions. By applying analytical techniques developed for multiple-trapping transport in homogeneous materials to TOF and TPC data, several interesting features emerge. For example, a large increase in hole time-of-flight mobility, by a factor of over 100, is measured in microcrystalline silicon compared with amorphous silicon. Significant enhancements occur at only 10% crystallinity, below what is normally considered to be the conductivity percolation threshold. However the mobility values, the observed field-dependence of mobility, and hole tail-state distributions deduced from the shape of the photocurrent decay do not appear to be mutually consistent when included in a computer simulation of carrier transport. We discuss what modifications may be required in order to develop a more appropriate analytic transport model.
10:00 AM - A1.2
Carrier Dynamics in Microcrystalline Silicon Studied by Time-Resolved Terahertz Spectroscopy.
Ladislav Fekete 1 , Filip Kadlec 1 , Hynek Nemec 1 , Petr Kuzel 1 , Jiri Stuchlik 1 , Antonin Fejfar 1 , Jan Kocka 1
1 , Institute of Physics, Czech. Acad. Sci., Prague 8 Czech Republic
Show AbstractImprovement of the microcrystalline silicon (μc-Si:H) based devices critically depends on the detailed knowledge of ultrafast charge carrier dynamics. The lifetime and the mobility of the carriers are significantly influenced by the material structure (degree of crystallinity, grain sizes and arrangement, level of doping)[1,2].The time-resolved terahertz (THz) spectroscopy is a contact-free method highly sensitive to mobile charges and, consequently, well suited for the investigation of dynamics of photo-excited semiconductors [3,4,5]. In this contribution we report on optical pump - THz probe experiments in a series of 1 μm thick μc-Si:H layers deposited on sapphire substrates with a variable degree of crystallinity. Our experimental method allows us to obtain broadband transient spectra of the THz conductivity as a function of pump–probe delay. These spectra and their time evolution are found to be significantly influenced by the film crystallinity. We show that the spectra can be decomposed into a contribution of free carriers (Drude term yielding dc conductivity) and a contribution of localized carriers which interact with shallow bandtail levels and with the potential barriers between grains. Characteristic times of all these processes can be deduced from the experimental data. Based on the results a general picture of the transient conductivity mechanisms of these samples can be drawn.[1] J. Kudrna, P. Malý, F. Trojánek, J. Štěpánek, T. Lechner, I. Pelant, J. Meier, and U. Kroll, Mat. Sci. Eng. B69–70, 238 (2000).[2] S. Brehme, P. Kanschat, K. Lips, I. Sieber, and W. Fuhs, Mat. Sci. Eng. B69–70, 232 (2000).[3] C. A. Schmuttenmaer, Chem. Rev. (Washington, D.C.) 104, 1759 (2004).[4] H. Nemec, F. Kadlec, C. Kadlec, P. Kuzel, and P. Jungwirth, J. Chem. Phys. 122, 104504 (2005).[5] P. U. Jepsen, W. Schairer, I. H. Libon, U. Lemmer Ne Hecker, M. Birkholz, K. Lips, and M. Schall, Appl. Phys. Lett. 79, 1291 (2001).
10:15 AM - A1.3
Electronic Properties of a-Si:H/c-Si Heterojunctions.
Lars Korte 1 , Abdelazize Laades 1 , Manfred Schmidt 1
1 Silizium Photovoltaik, Hahn-Meitner-Institut, Berlin Germany
Show AbstractFor application of a-Si:H/c-Si heterojunctions as active part in solar cells, TFTs or field effect devices, it is important to gather information on the interface state density distribution and its modification by contributions of a-Si:H gap states, on band offsets, charge transfer processes etc. We investigated the electronic states at the a-Si:H/c-Si heterojunction by using photoelectron spectroscopy with excitation by near-ultraviolet light (NUV-PES) and surface voltage measurements (SPV). The NUV-excitation (hν=4-8eV) results in an electron escape depth up to 10nm and in an optical excitation probability enhanced by several orders of magnitude compared to excitation with soft X-rays. This allows the direct observation of the distribution of occupied gap states NOCC(E) and of the Fermi level position EF. Using the constant final state yield spectroscopy (CFSYS) mode of NUV-PES which provides the highest sensitivity, NOCC(E) was measured for a-Si:H/c-Si heterostructures with 2-100nm thin intrinsic, n- and p-doped a-Si:H layers of varying doping level. The layers were prepared by PECVD using the precursor gases silane (SiH4), phosphine (PH3) and diborane (B2H6). Applied to an a-Si:H(i) deposition temperature series, CFSYS shows that the Urbach energy is minimal (61meV) and the Fermi level closest to midgap for a deposition temperature of TS=230°C. This correlates with a minimum in interface recombination rate at the same temperature, as derived from the decay of surface photovoltage over time. The corresponding interface state density amounts to about Dit=1×1012cm-2eV-1 at midgap, as estimated by field dependent SPV (FD-SPV). The density of dangling bonds has a value of about 1-3×1018cm-3 and remains nearly independent of the deposition temperature. This is about a factor of ten higher than in a-Si:H bulk, in agreement with findings of other authors on extremely thin films. The Urbach energy increases up to 80meV for layers deposited below and above TS=230°C.FD-SPV measurements clearly show that the interface state density follows the same trend as the Urbach energy. This hints on a direct coupling of the tail states with the fast interface states by charge transfer processes. Thus, by FD-SPV we determine an effective interface state density.For a-Si:H layer thicknesses ≤10 nm, we obtain a contribution of Si valence band states to the photoemission signal. Describing these spectra as the superposition of a-Si:H and c-Si densities of states yields excellent fit results and allows the exact determination of the band offset between a-Si:H(i) and c-Si(p), to ΔEV=0.38eV±0.06eV. Thus, we have a nearly complete description of the major electronic quantities of the a-Si:H/c-Si heterointerface. The relevance of this data for the optimization of a-Si:H/c-Si heterojunction based devices is illustrated by our findings that the maximum of a-Si:H/c-Si solar cell efficiencies is reached when measured effective interface state densities are minimal.
10:30 AM - A1.4
Simulation of Realistic Core-shell Silicon Nanowires.
Rana Biswas 1 , Bicai Pan 2
1 Dept. of Physics & ECpE, MRC, Ames Lab, Iowa State University, Ames, Iowa, United States, 2 Dept of Physics, Univ of Science and Technology of China, Hefei China
Show AbstractTuesday, April 18Poster A5.8 Transferred to Oral A1.4 9:30 AMSimulation of Realistic Core-shell Silicon Nanowires. Rana Biswas
10:45 AM - A1.5
Correlating Light-Induced Degradation in the Performance of Nanocrystalline Silicon Solar Cells with Changes in the Electronic Properties Determined from Junction Capacitance Measurements.
Peter Hugger 1 , Shouvik Datta 1 , J. Cohen 1 , Guozhen Yue 2 , Gautam Ganguly 2 , Baojie Yan 2
1 Department of Physics, University of Oregon, Eugene, Oregon, United States, 2 , United Solar Ovonic Corporation, Troy, Michigan, United States
Show AbstractThe electronic properties of hydrogenated nanocrystalline silicon (nc-Si:H) were studied using drive-level capacitance profiling (DLCP), to establish the deep defect densities near midgap, as well as transient photocapacitance (TPC) and transient photocurrent (TPI) spectroscopies to determine the spectra of defect related optical transitions. These measurements were performed on a series of n-i-p solar cell devices with roughly 1-micron thick nc-Si:H i-layers. Some of these devices employed specular stainless substrates (SS/n+/i nc-Si:H/p+/ITO), while others incorporated a textured Ag/ZnO back reflector (SS/Ag/ZnO/n+/i nc-Si:H/p+/ITO). The intrinsic nc-Si:H layers were deposited using RF or MVHF glow discharge with various hydrogen dilution profiles. Optimized hydrogen dilution profiling improves not only the initial performance but also the stability after prolonged light soaking. As a result, the devices studied exhibited varying degrees of degradation after 1000 hours of 100 mW/cm2 white light exposure, from 10-15% for the highest degradation devices to less than 3% for the lowest degradation devices.The DLCP measurements taken at 11 Hz near room temperature revealed defect densities in the low 1015 cm-3 level for most of the devices. These profiles exhibited only a moderate variation with temperature (increasing by a factor of 2 to 3 between 200K and 325K). Somewhat surprisingly, these profiles exhibited almost no changes after 100 hours of light soaking at 500mW/cm2 using a red filtered (610 nm) ELH light source. In contrast, the TPC optical spectra showed significant changes after light-soaking, indicating a reduction in minority carrier (hole) collection. This was manifested by a marked increase in the TPC signal at optical energies above the silicon bandgap at 1.1eV. That is, nearly equal degrees of electron and hole collection result in a very small photocapacitance signal, and this what is observed for most of the devices in the annealed state near room temperature. After light soaking, however, the TPC signal increased markedly, indicating a reduction in the relative fraction of holes collected. The relative reduction was typically a factor of 3 to 4, depended on the measurement temperature, and also differed in detail between the devices exhibiting more or less stable solar cell performance. The photocapacitance spectra also revealed a deep defect band at lower optical energies (below 0.9eV). However, this did not exhibit any noticeable increase after light soaking. Possible mechanisms for the observed light-induced changes in these nc-Si:H devices will be discussed.
A2: Metastability
Session Chairs
Tuesday PM, April 18, 2006
Room 3002 (Moscone West)
11:30 AM - **A2.1
Metastability in Hydrogenated Nanocrystalline Silicon Solar Cells.
Guozhen Yue 1 , Baojie Yan 1 , Gautam Ganguly 1 , Jeffrey Yang 1 , Subhendu Guha 1
1 , United Solar Ovonic Corporation, Troy, Michigan, United States
Show Abstract12:00 PM - A2.2
Characterization of the Evolution in Metastable Defects Created by Recombination of Carriers Generated by Photo-generation and Injection in p-i-n a-Si:H Solar Cells.
Jingdong Deng 1 , Benjamin Ross 2 , Robert Collins 3 , Christopher Wronski 1
1 Electrical Engineering, Penn State University, University Park, Pennsylvania, United States, 2 Engineering Science and Mechanics, Penn State University, University Park, Pennsylvania, United States, 3 Physics and Astronomy, University of Toledo, Toledo, Ohio, United States
Show AbstractMetastable defect creation by recombination of photo-generated carriers as well as those crated by far forward bias in a-Si:H p-i-n solar cells has been investigated. This has been carried out on cell structures in which the recombination in the p/i interface regions has been minimized so that the dark forward bias currents are dominated by recombination in the bulk even at high forward biases [1]. The evolutions in the defects created with volume absorbed red light and far forward bias were characterized with the Shockley-Reed-Hall recombination obtained from the dark currents under forward bias between 0.2 and 0.7 volts. This methodology allowed the creation of the defects to be obtained in the absence of isothermal annealing which is significant even at 25oC [2]. A detailed study was carried out on the kinetics of the defects creation by carrier injection during which the electron and hole concentrations remain essentially constant. It included a range of temperatures between 25oC and 75oC as well as degradation currents from 1 to 100mA/cm2 which correspond to short circuit currents generated by 0.1 to 10 sun illumination. The kinetics of the changes in the defect densities obtained with both types of recombination are similar to those generally discussed [3]. They exhibit a stretched exponential behavior with a t1/3 relation over limited regimes before the onset of saturation. The approaches to saturation and the saturated values were obtained for different degradation intensities and temperatures. All these kinetics exhibit an (intensity)t2 dependence such as previously reported for fill factors under high intensity illuminations. However, from the detailed characterization clear evidence is obtained that the evolutions of the metastable defects alone exhibit a t1/2 relation over much wider regimes not only for degradation with very high far forward biases but also those corresponding to 1 sun and less. These results on p-i-n cells are compared with those obtained on corresponding i-layer thin films on which similar results are obtained for the light induced metastable defects. The significance of the results on the models that have been proposed for the creation of light induced metastable defects is discussed. [1] J. Deng and C. R. Wronski, J. Appl. Phys. 98, 2005, p. 024509[2] J. Deng, J. M. Pearce, V. Vlahos, R. W. Collins, and C. R. Wronski, Mater. Res. Soc. Symp. Proc., 808, A8.8, 2004.[3] H. Fritzsche, Annu. Rev. Mater. Res. 31, 2001, p.47
12:15 PM - A2.3
Light-Soaking Effects on a-Si:H Solar Cells: Evidence for Self-Limitation?
Jianjun Liang 1 , Eric Schiff 1
1 Physics, Syracuse University, Syracuse, New York, United States
Show AbstractFundamental studies of light-soaking in hydrogenated amorphous silicon (a-Si:H) have often focused on the increase in the dangling bond density during extended illumination of the material. This increase can be an order of magnitude or more before the dangling bond density ultimately saturates for very long illumination times. Several microscopic models for this increase have been proposed; recent NMR studies [1] favor hydrogen-based ideas such as the "hydrogen collision" model. The degradation of the open-circuit voltage (VOC) in solar cells during light-soaking offers additional insights. The principal fact is the relatively small magnitude of the degradation at saturation, which typically amounts to a few percent, or less, of the initial value of Voc. A second intriguing fact reported in our own recent work on United Solar Ovonic Corp. cells is that there is a "latency" period before degradation of Voc commences. In this work we discuss the meaning of these effects. The magnitude of the latency period is consistent with a calculation based on the hydrogen-collision model for defect generation and a fairly simple, "bandtail+defect" recombination model. The relatively small magnitude of the saturated degradation of Voc reflects the approximate coincidence of this latency period and the saturation time. We suggest that this coincidence is evidence for "self-limitation," by which we mean that the decline in Voc is coupled to a decline in the rate of defect generation. The original modeling of light-soaking by Stutzmann, et al. [2], contains such a mechanism. Light-soaking was proposed to be initiated only by bandtail recombination events, so a crossover between photocarrier recombination involving bandtails and recombination involving dangling bonds slows dangling bond generation. While crossover does appear to be important in determining the latency period, it isn't strong enough to account for the observed form of saturation. We speculate that recombination through dangling bonds is the origin of "light-induced annealing" of light-soaking, and we shall present a critical analysis of this model. This research has been supported by the National Renewable Energy Laboratory's Thin Film Photovoltaics Partnership.[1] T. Su, P. C. Taylor, G. Ganguly, and D. E. Carlson, Phys. Rev. Lett., 89, 015502(2002).[2] M. Stutzmann, W. B. Jackson, and C-C. Tsai, Phys. Rev. B 32, 23(1985).
12:30 PM - A2.4
Microscopic Theory of Hydrogen Related Metastability in Disordered Silicon.
Blair Tuttle 1
1 Physics, Penn State Erie; The Behrend College, Erie, Pennsylvania, United States
Show AbstractOur recent density functional electronic structure calculations examined a variety of configurations for hydrogen in silicon. [1] A novel complex has been discovered for hydrogen in amorphous silicon. The complex involves the breaking of a weak bond to form two Si-H bonds with both hydrogens in between the original silicon atoms. This configuration provides a microscopic model for new metastable complexes observed by NMR. Mechanism for hydrogen related metastability are discussed for bulk amorphous and polycrystalline silicon as well as interfacial MOS environments.[1] B. R. Tuttle, Physical Review Letters, vol. 93, 215504 (2004)
12:45 PM - A2.5
The Effects of Oxygen Contamination and Light-Induced Degradation on the Electronic Properties of Hot-Wire CVD Amorphous Silicon-Germanium Alloys.
Shouvik Datta 1 , J. Cohen 1 , Yueqin Xu 2 , James Doyle 2 , A. Mahan 2 , Howard Branz 2
1 Department of Physics, University of Oregon, Eugene, Oregon, United States, 2 , National Renewable Energy Laboratory, Golden, Colorado, United States
Show Abstract We previously reported that the electronic properties of hot-wire CVD a-Si,Ge:H alloys could be improved to a level comparable to the best glow discharge (PECVD) a-Si,Ge:H alloy films when the hot-wire deposition process employed a tantalum filament maintained at ~1800°C. Our studies focused on Ge fractions in the 15 to 50 at.% range yielding Tauc gaps between 1.65eV to 1.3eV. Many of these samples exhibited Urbach energies below 45 meV, even for Ge fractions up to 47at.%. Also, the annealed state defect densities were found to lie well below the 1016cm-3 level for Ge fractions up to 30at.%. These electronic properties were obtained using drive-level capacitance profiling (DLCP), to establish the deep defect densities near midgap, and sub-band-gap spectroscopy (photocurrent and photocapacitance) to determine the spectra of defect related optical transitions. Ion bombardment of the growing film, long thought necessary for deposition of high-quality a-SiGe:H, is absent in the HWCVD growth process. However, a few of the samples were found to exhibit much broader Urbach energies (as high as 60 meV) and significantly higher deep defect densites. These were also found to contain higher levels of oxygen contamination (about 1019 cm-3, instead of about 1018 cm-3). The source of HWCVD oxygen contamination has been identified and eliminated. Controlled oxygen-contamination experiments are now underway, so that the effects of contamination at the 1019 cm-3 level can be produced and characterized in detail. We are also exploring how critically the oxygen contamination level affects the electronic properties of glow discharge a-Si,Ge:H films. Finally, we examined the effects of light-induced degradation on several of these alloy samples using a red-filtered (610 nm) tungsten-halogen light-source, and compared this to the behavior of PECVD grown a-Si,Ge:H alloys of similar optical gaps. The midgap defect densities revealed by DLCP in the 29 at.% Ge samples were found to exhibit an unusual two-step defect creation kinetics, suggesting two distinct types of light induced defects. A similar behavior was also observed in defect creation kinetics of the PECVD alloy samples, although the two regimes of defect creation appeared to be less distinct. Subsequent isochronal annealing of these light-induced defects also indicated a marked difference in behavior of the HWCVD and PECVD alloys samples. A 10 minute dark annealing treatment was found to reduce the defects fairly monotonically for the PECVD samples over the temperature range 360 to 460K but the annealing of defects occurred much more abruptly for the HWCVD samples, between 430 and 460K. This suggests a much narrower range of annealing energy barriers for metastable defects for the HWCVD alloy samples.
A3: Growth Mechanisms
Session Chairs
Tuesday PM, April 18, 2006
Room 3002 (Moscone West)
2:30 PM - **A3.1
In Situ Defect Spectroscopy: Probing Dangling Bonds During a-Si:H Film Growth by Subgap Absorption.
I.M.P. Aarts 1 , A.C.R. Pipino 2 , M.C.M. van de Sanden 1 , W.M.M. Kessels 1
1 Applied Physics, Eindhoven University of Technology, Eindhoven Netherlands, 2 Chemical Science and Technology Laboratory , National Institute of Standards and Technology (NIST), Gaithersburg, Maryland, United States
Show AbstractDetection of ultra-low defect densities in a-Si:H and other thin films is of great importance for state-of-the-art thin-film device applications. From a fundamental point of view, understanding the role of surface defect states during the growth process is also essential. Yet, a paucity of experimental techniques exists for readily detecting surface and/or bulk defect states. Using the technique of evanescent-wave cavity ring-down spectroscopy (EW-CRDS) [1], we demonstrate a unique, absolute absorption measurement of defect states, including dangling bonds, having unprecedented sensitivity that can be applied in situ, during film growth. The minimal detectable absorption of the technique is 3x10-8 optical loss, which is equivalent to 3x108 dangling bonds/cm2. Employing hot-wire chemical vapor deposition of a-Si:H, we deposited films up to 800 nm in thickness onto the total-internal reflection (TIR) surface of a high-finesse (ultra-low-loss) monolithic folded optical resonator. Subgap absorption spectra between 1170 and 1245 nm reveal a broad absorption feature due to dangling bonds defect states present in the bulk and at the interfaces. From the real time experiments, the dangling bond distribution in the film could be established, showing the highest dangling bond density at the a-Si:H / substrate interface, while the a-Si:H/vacuum interface density was approximately ten times smaller. Furthermore, changes in surface dangling bond density could be monitored in real-time under various growth conditions and during flux dependent atomic hydrogen dosing experiments, through formation and decay curves.[1] W.M.M. Kessels, I.M.P. Aarts, J.J.H. Gielis, J.P.M. Hoefnagels, and M.C.M. van de Sanden, Mater. Res. Soc. Symp. Proc. 862, A14.3.1 (2005); I. M. P. Aarts, A. C. R. Pipino, J. P. M. Hoefnagels, W. M. M. Kessels, and M. C. M. van de Sanden, Phys. Rev. Lett. 95, 166104 (2005).
3:00 PM - A3.2
Surface Roughening Transitions in Semiconductor Thin Films
Nikolas Podraza 1 , Jian Li 1 , Christopher Wronski 2 , Robert Collins 1
1 Department of Physics and Astronomy, University of Toledo, Toledo, Ohio, United States, 2 Department of Electrical Engineering, Pennsylvania State University, University Park, Pennsylvania, United States
Show AbstractExtensive research on hydrogenated amorphous silicon (a-Si:H) thin films has shown that optimum device performance is correlated with the smoothest, most stable surfaces versus time in the deposition process – in fact, less than a monolayer of roughening occurs throughout the deposition of optimum films as thick as 0.5 μm. Such research requires smooth crystalline silicon substrates for high sensitivity to surface roughness as well as an in situ probe such as spectroscopic ellipsometry for continuous, real-time analysis. It has been also shown that when the plasma-enhanced chemical vapor deposition (PECVD) conditions are adjusted from the optimum even slightly, e.g., by a reduction in the hydrogen dilution ratio or by an increase in plasma power from its minimum, then a roughening transition is observed in the a-Si:H growth regime. This transition shifts to decreasing bulk layer thickness for films with increasingly deteriorated device properties. As a result, the thickness at which the roughening transition occurs has been entered into deposition phase diagrams, relevant for c-Si substrates, that provide useful insights into the device quality of a-Si:H and how it varies with the deposition parameters. In the present study, the roughening transition has been characterized as a function of deposition variations for PECVD a Si1-xGex:H, an alloy system that poses considerable challenges in optimization. The variations explored include H2-dilution (which is used most widely as the phase diagram variable), alloy content (through G=[GeH4]/{[SiH4]+[GeH4]}), substrate temperature, electrode configuration (anode versus cathode), and He dilution. One common feature of these variations is a maximum in the roughening transition thickness (and hence surface stability) at a H2 dilution level just below the transition to mixed phase (amorphous+microcrystalline) growth. A second feature is a significant enhancement in the roughening transition thickness for cathodic biases of ~—20 V. The close correlation between the amorphous roughening transition thickness and the electronic performance has led to a simple model for the transition in terms of a competition between roughening common to all thin films due to the atomic size and smoothening due to precursor surface diffusion. It is proposed that diffusing precursors are immobilized by defects (or other diffusing precursors), and this process is associated with defects that ultimately reside in the bulk, as well. A simple generic model for the roughening transition is consistent with its widespread appearance in diverse thin film systems. As an example, results are presented for another thin film photovoltaic material CdTe which is prepared in a magnetron sputtering process. REFERENCES[1] R.W. Collins and A. Ferlauto, Current Opinion in Solid State and Material Science 6, 425 (2002).
3:15 PM - A3.3
Reaction Mechanism for Deposition of Silicon Nitride by Hot-Wire CVD with Ultra High Deposition Rate(>7 nm/s).
Vasco Verlaan 1 , Zomer Silvester Houweling 1 , Karine Werf 1 , Hanno D. Goldbach 1 , Ruud E.I. Schropp 1
1 , Utrecht University, Utrecht Netherlands
Show AbstractSilicon Nitride is a material with a wide spectrum of commercial applications. In this research we investigated hydrogenated silicon nitride (SiN:H) deposited by hot-wire chemical vapor deposition (HW-CVD) for application as passivating and antireflection coating on multicrystalline silicon solar cells. Recently, we have shown that multicrystalline silicon solar cells can be obtained with hot-wire deposited SiN:H deposited at 3.0 nm/s, reaching 15.7% efficiency[1]. As high quality SiN:H at even higher deposition rate is needed for low cost mass production, in combination with good control of material composition and structure, knowledge about the deposition mechanism is crucial. Although the mechanism of conventional plasma CVD depositions of SiN:H is relatively well identified as well as the deposition of thin film Si with HW-CVD, the deposition mechanism for SiN:H deposited by HW-CVD has been much less investigated. We studied the deposition mechanism by exploring the effects of wire temperature, process pressure and gas-flow ratio on the composition and optical properties of the deposited SiN:H layers. The composition and density of each of the samples were investigated with Elastic Recoil Detection (ERD) analysis and Fourier Transform InfraRed spectroscopy (FTIR) whereas the optical properties (refractive index and absorption) were determined with reflection/transmission measurements. Additionally, we performed deposition experiments with D2 and deuterated silane (SiD4) to gain further insight in the chemical reactions during deposition.We found that the decomposition of the ammonia (NH3) source gas is more difficult than the decomposition of the silane. Therefore, the properties of the deposited SiN:H layers as well as the potential for fast deposition are largely determined by the ability to decompose the ammonia and to incorporate the nitrogen atoms into the SiN:H coatings. It appeared that both the process pressure and the temperature of the Ta wires greatly affect the efficiency of the ammonia decomposition. The research with D2 and SiD4 silane led to an enhanced understanding of the chemical reactions taking place at the Ta wires and on the substrate. By combining these insights with information of Hot-Wire deposition of thin film Si, we were able to unravel the deposition mechanism for different types of HW-CVD SiN:H. With the knowledge gained we were able to optimize the filament temperature and pressure such that the deposition rate of silicon nitride deposited by HW CVD increased to an ultra high value of 7 nm/s for films (N/Si=1.18, ρ=2.6 g/cm3, n=1.95) with a high transparency, suitable for use as antireflection layer. This is much faster than conventional deposition techniques for SiN:H can offer. By optimizing the flow ratio of the source gasses good control of the composition, and thereby optical properties, of the deposited layers is achieved. [1] Proc. 20th Photovoltiac Solar Energy Conference, 2DV2.4.1. 1434-1437.
3:30 PM - A3.4
Analysis of Chemical Reactions between Radical Growth Precursors Adsorbed on Plasma-Deposited Silicon Thin-Film Surfaces
Tamas Bakos 1 , Mayur Valipa 1 2 , Dimitrios Maroudas 1
1 Chemical Engineering, University of Massachusetts, Amherst, Massachusetts, United States, 2 Chemical Engineering, University of California, Santa Barbara, California, United States
Show AbstractDevice-quality hydrogenated amorphous silicon (a-Si:H) thin films are typically deposited under conditions where the SiH3 radical is the dominant deposition precursor. Upon impingement on the a-Si:H film surface, SiH3 radicals may either react chemically with the surface at the point of contact or diffuse on the surface prior to reaction with surface Si atoms or other surface hydrides. Development of systematic strategies for depositing silicon thin films with desirable properties requires a fundamental understanding of the interactions between reactive radicals originating in the plasma and the growth surface, as well as between the adsorbed radicals themselves. In this presentation, we use first-principles density functional theory (DFT) calculations on the crystalline Si(001)-(2×1):H surface in conjunction with molecular-dynamics (MD) simulations on a-Si:H surfaces to investigate the interactions between SiH3 radicals adsorbed on Si thin film surfaces. Our analysis reveals the existence of two reaction channels: two SiH3 radicals may either form disilane (Si2H6) that desorbs from the surface or undergo disproportionation reactions resulting in formation of an SiH2 adspecies that is incorporated in the film, and a silane (SiH4) molecule that is released in the gas phase. The DFT calculations show that Si2H6 formation from reaction between two adsorbed SiH3 radicals is virtually barrierless if both radicals are in a “diffusive state”, i.e., if they are attached to overcoordinated surface Si atoms. On the other hand, disilane formation exhibits a moderate (0.46 eV) activation energy barrier if one of the two SiH3 radicals is attached to an overcoordinated surface Si atom, while the other one is chemisorbed on the surface. Finally, activation energy barriers in excess of 1 eV are obtained for reactions between two chemisorbed SiH3 radicals. We find that disproportionation displays a similar tendency associated, however, with reaction barriers in excess of 0.4 eV for all possible configurations of the reactants. MD simulations confirm that both disilane formation and disproportionation reactions also occur on a-Si:H growth surfaces, preferentially in configurations where at least one of the SiH3 radicals is in a “diffusive state”. Our results are in agreement with experimental observations and results from plasma process simulators showing that the primary source for disilane in low power plasmas may be the substrate surface.
3:45 PM - A3.5
Surface Evolution During Epitaxial and Polycrystalline Silicon Film Growth by Low Temperature Hot-Wire Chemical Vapor Deposition on Silicon Substrates
Christine Richardson 1 , Young-bae Park 1 , Harry Atwater 1
1 Thomas J. Watson Laboratories of Applied Physics, California Institute of Technology, Pasadena, California, United States
Show AbstractThe structure and properties of silicon thin films grown by chemical vapor deposition can vary widely depending upon growth conditions. Previously, many researchers have reported on the evolution roughness of amorphous Si films with substrate temperature and with thickness and of epitaxial films grown by molecular beam epitaxy (MBE). We have in this study deliberately focused on high hydrogen dilution conditions that lead to epitaxial growth with an epitaxial-polycrystalline transition, rather than a crystal-amorphous transition. In this study we address the surface evolution of hot-wire chemical vapor deposited (HWCVD) crystalline Si thin films with temperature, thickness, and hydrogen dilution and discuss the resulting growth regimes.We have investigated the low-temperature growth of crystalline and polycrystalline thin silicon films by hot-wire chemical vapor deposition (HWCVD). Using Raman spectroscopy, spectroscopic ellipsometry, and atomic force microscopy, we find the relationship between surface roughness evolution and the substrate temperature (230 – 350°C) and the hydrogen dilution ratio (H2/SiH4 = 0 – 480). Quantitative height-height correlation functions and scaling exponents determined here are compared with previous reports of silicon growth by other methods, and kinetic growth mechanisms are inferred from the quantitative comparison. Higher hydrogen dilution increases the surface roughness as expected. However surface roughness increases with increasing substrate temperature, in contrast to previous studies of crystalline Si growth. This suggests that H desorption enables more contaminant absorption with increasing temperature, in turn increasing roughness. We develop a general model for temperature-dependent Si surface roughness evolution during H-diluted HWCVD growth and focus on the role of the concurrent atomic hydrogen flux. Specifically, increasing hydrogen dilution changes the growth kinetic regime from i) smooth film growth via random atomic deposition with relaxation to ii) growth flux shadow-dominated growth and finally to iii) a growth kinetic regime dominated by H-abstraction-mediated desorption and re-deposition of growth species. The growth regime determines the structure of the film and we suggest that transitions between these kinetic regimes in turn are the dominant factors governing the epitaxial–polycrystalline transition in low temperature CVD growth, and we discuss the implications of this. We find, significantly, that H-abstraction-mediated desorption and re-deposition does not substantially lower the overall film growth rate, implying efficient redeposition of desorbed species at high dilution. Overall, the model developed here for surface roughness evolution and H-abstraction provides guidance about how best to achieve low-temperature epitaxial Si growth in HWCVD while avoiding a crystal-amorphous transition, thus enabling growth of arbitrarily thick Si crystalline films at low temperatures.
A4: Growth Techniques and Interface Studies
Session Chairs
Tuesday PM, April 18, 2006
Room 3002 (Moscone West)
4:30 PM - **A4.1
Properties of Nanocrystalline 3C-SiC:H and SiC:Ge:H Films Deposited at Low Substrate Temperature.
Shinsuke Miyajima 1 , Akira Yamada 2 , Makoto Konagai 1
1 Physical Electronics, Tokyo Institute of Technology, Tokyo Japan, 2 Quantum Nanoelectronics Research Center, Tokyo Institute of Technology, Tokyo Japan
Show Abstract5:00 PM - A4.2
Dual-Chamber Plasma Co-Deposition of Nanoparticles in Amorphous Silicon Thin Films.
C. Anderson 1 , C. Blackwell 2 , J. Deneen 3 , C. Carter 3 , J. Kakalios 2 , U. Kortshagen 1
1 Dept. of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota, United States, 2 School of Physics and Astronomy, University of Minnesota, Minneapolis, Minnesota, United States, 3 Dept. of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States
Show AbstractSilicon nanocrystals embedded within a hydrogenated amorphous silicon matrix (a/nc-Si:H) have attracted attention for photovoltaic device applications due to their improved charge transport properties. These mixed-phase materials are typically deposited in a single-chamber PECVD system, which relies on a semi-specific set of conditions to deposit both the amorphous and crystalline phases of the film. This represents a limitation of the process, as control over the crystalline formation (e.g. diameter) is not easily achieved. Further, the deposition conditions for nanocrystallite formation are not compatible with those that optimize the electronic quality of the thin-film a-Si:H. We describe a process of remotely synthesizing 3 to 5-nm silicon crystallites which are then “co-deposited” into a growing a-Si:H film in a second capacitive discharge chamber. This approach of separating the particle synthesis from the film deposition has the advantage of being able to independently control either process, allowing the study of the influence of the crystalline phase on the electronic properties of the resulting mixed-phase thin film. We report the optical, infra-red absorption, and electronic properties of these films and discuss the TEM image analysis used to characterize undoped a/nc-Si:H films synthesized in this dual-chamber system. In contrast to materials deposited in a single-chamber PECVD system, the dual-chamber films exhibit a higher dark conductance in the annealed state A, relative to identical materials grown without nanocrystalline inclusions. Even after extended light soaking, the state B dark conductance of the mixed-phase materials is still larger than the state A conductivity of films without nanocrystals. When a/nc-Si:H films are grown with the substrate and RF electrodes at the same temperature (that is, without thermophoretic forces directing the nanocrystals towards or away from the silicon thin film's growing surface) a significant persistent photoconductivity effect is observed. This work was partially supported by NSF grants NER-DMI-0403887, IGERT grant DGE-0114372, and the University of Minnesota.
5:15 PM - A4.3
A Real-time Study of the a-Si/c-Si Interface Formation during Deposition using Ellipsometry, Infrared Spectroscopy, and Second-Harmonic Generation
Peter van den Oever 1 , Joost Gielis 1 , Bram Hoex 1 , Richard van de Sanden 1 , Erwin Kessels 1
1 Department of Applied Physics, Eindhoven University of Technology, Eindhoven Netherlands
Show AbstractThe properties of the a-Si/c-Si interface are of key importance for new developments in the crystalline silicon solar cell research. For example, in the silicon heterojunction (SHJ) solar cell a stack of very thin layers of intrinsic a-Si:H and doped a-Si:H is used to contact the active layer of the cell. Using the SHJ concept, efficiencies exceeding 21 % have been reported by Sanyo. Furthermore, a-Si:H has recently also been used for rear-surface passivation of (homojunction) crystalline solar cells and efficiencies higher than 20 % have been reported. The formation of the interface between c-Si and a-Si:H has been studied in a hotwire chemical vapor deposition process (HWCVD) at a substrate temperature of 150 °C using three different in situ optical diagnostics. Real time spectroscopic ellipsometry (SE) has been used to deduce the optical properties of the a-Si:H as well as the structural information in terms of film thickness and surface roughness evolution. Especially the nucleation phase of the a-Si:H on the c-Si is investigated by focusing on the nucleation-induced surface roughness (typically 15-20 Å), the coalescence phase of the nuclides, and the amorphous-to-amorphous (a-a) roughening transition which takes typically place at approximately 40 Å. The (a-a) roughening transition has been correlated with the hydrogen bonding mode in the film as measured by attenuated total reflection (ATR) infrared spectroscopy. In the first 6 Å, surface hydrides can clearly be distinguished in the infrared spectra with subsequently the formation of a large amount of SiH2 bonds near the interface (up to a thickness of 20 Å) and with finally a transition to mainly SiH bonds during a-Si:H bulk growth. Spectroscopic second-harmonic generation (SHG) in the near-infrared region has been found to be sensitive to dangling bonds at the interface between the c-Si and the a-Si. The evolution of the SHG signal for small film thickness is correlated with the surface roughness evolution and the hydrogen bonding modes while varying the (a-a) roughening transition by diluting the SiH4 gas with H2. Furthermore, measurements of the surface recombination velocity on the c-Si interface are currently in progress using a Sinton lifetime tester. From the combination of the data, very valuable information can be extracted for control of the a-Si:H growth on the c-Si and for further optimization of the performance of heterojunction and homojunction crystalline silicon solar cells.
5:30 PM - A4.4
Interface Study Of Nanocrystalline Silicon/Multicrystalline Silicon Using Microwave Photoconductive Decay.
Mahdi Farrokh Baroughi 1 , Siva Sivoththaman 1
1 Electrical and Computer Engineering, University of Waterloo, Waterloo, Ontario, Canada
Show AbstractNanocrystalline silicon (nc-Si) thin films on top of crystalline silicon material have been recently employed for fabrication of solar cells. Previously, very thin intrinsic amorphous silicon (a-Si) films have been successfully employed for passivation of interfacial defect of heterojunction (HJ) solar cells and wafer passivation purposes. Alternatively, nc-Si can be exploited for passivation of crystalline silicon wafers and interfacial defects states. In this work, we study the interface of nc-Si/mc-Si heterostructure using microwave photoconductive decay (µ-PCD) technique, TEM analysis, and Medici device simulation.Bulk lifetime of p-type mc-Si substrates with resistivity of 1-2 (Ω-cm) was mapped using µ-PCD before film deposition. nc-Si and a-Si films with thicknesses of 50nm were deposited on the mc-Si wafers using PECVD of SiH4 under different deposition conditions. For deposition of the nc-Si films very high hydrogen dilutions in the range of 98 to 99.5 percent were used. For deposition of the a-Si films we used milder hydrogen dilutions in the range of 10 to 80 percent. Crystallinity of nc-Si films, deposited on glass substrates, was measured in the range of %50 to %80 using Raman spectroscopy. Hydrogen contents of a-Si layers were measured in the range of %7 to %12 using infrared spectroscopy. Both conductivity measurement and TEM analysis show that the crystallinity of the films deposited on mc-Si is more than that of the films deposited on glass. In the next step, the second lifetime mapping was performed and the surface recombination velocity (SRV) was extracted for all of the interfaces using a modified asymmetric SRV model. Recombination velocities in the range of 300cm/sec to 2000cm/sec were obtained.An equivalent interface model was developed in Medici based on the measured SRV values. According to the modeling results, interface defect densities in the range of 6x1010 - 4x1011 cm-2 were obtained for nc-Si/mc-Si interface.
5:45 PM - A4.5
Dewetting of Patterned Thin Solid Films.
Erwan Dornel 1 , J-C. Barbe 1 , J. Eymery 2 , F. de Crecy 1
1 LETI/D2NT/LSCDP, CEA, Grenoble France, 2 DRFMC/SP2M, CEA-CNRS-UJF, Grenoble France
Show AbstractIt has been observed that during a thermal annealing between 750°C and 950°C under hydrogenated atmosphere, an 8-13nm Silicon On Insulator (SOI) (001) film patterned along <110> direction can be fully agglomerated in a few minutes. At this temperature (~ 2/3 of the silicon melting temperature), the surface diffusion can be considered as the predominant mass transport mechanism, and the surface energy anisotropy must be taken into account. Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM) have been used to characterize the samples. The facet orientations of dewetted samples have been analyzed to investigate the morphology evolution and to propose a dynamic scheme.In agreement with calculations (see Ref. 2 hereafter) and previous observations,1 the dewetting is initiated from the edge of the pattern and forms a thickened ridge. In this work, it is observed that, the substrate is locally modified to verify the triple line equilibrium, and due to the surface energy anisotropy, the sides of the ridge present {311} and {111} facets whereas the top of the ridge is not a {001} flat surface but is formed with average {1,1,n} orientations (n~10 within our experimental resolution). In this first stage, the retraction is a two dimensional (2D) phenomenon leading to the shedding of lines of matters parallel to the <110> pattern edges. A transition from 2D to 3D retraction is then observed and generates lines of matters along <510>. Whereas the <110> lines stay stable even after a 30 min. annealing, the <510> lines are transformed in single agglomerates by pearling instabilities. The <001> and <510> lines are found to be less stable than the <110> ones. This dewetting scheme and the dewetting from <001> edges highlight that the diffusion coefficient on Si(001) is higher in the <110> directions than in the <001> ones. We will discuss the relative influence of the anisotropy of the surface energy and of the diffusion coefficient leading to faceting and preferential alignments.To supplement the dewetting knowledge, we model the surface diffusion by using a surface chemical potential depending on the local curvature.2 The dewetting of thin films has been computed in 2D for a surface energy anisotropy, described by a so called γ-plot, where cusp points may occur. The sections of the experimental dewetting are correlated with simulated profiles. It is found that the form of the γ-plot influences more the dewetting than the relative surface energy of the dense planes does. Finally, we will show that the facetted experimental surfaces of the Si/SiO2 system are well described by a quite small anisotropy of the γ-plot (several percents).1 Y. Ishikawa and al., Appl. Surf. Sci., 190 (2002), 11.2 E. Dornel and al., submitted to Phys. Rev. B.
A5: Poster Session: Theory of Structure and Transport
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A5.1
Research of Amorphous Silicon Thin-Film Structure and Growth Processes by Nonlinear Dynamics Method
Tatiana Larina 1 , Nikolay Bodyagin 1 , Sergey Vikhrov 1 , Stanislav Mursalov 1
1 BMPE, RGRTA, Ryazan Russian Federation
Show AbstractOur previous investigations show that amorphous materials growth process can be describe as self-organization processes. Material structure is spatially ingomogeous, “frozen’, nonequilibrium system examined from the point of view of the nonlinear dynamics. Structure in the films was investigated by the known in nonlinear dynamics Takens method applying to silicon-based amorphous thin films deposited by GD method and LF PECVD method. The characteristics of films surface were received by AFM. Then the obtained data were processed by of the Grassberger-Procaccia algorithm. The correlation dimension and its dependence on the embedding dimension were defined. On this dependence three various sites were found out. Their character testifies about that surface has extensive chaotic structure, and can be described by limited number of order parameters. The growth process is defined by three various mechanisms. A nature of these mechanisms and order levels is discussed. The average reciprocal information on a films surface was calculated also. The received data clearly specify presence of distant correlations.It is demonstrated, that films structure contains information about its previous growth processes. The analytical connection between parameters of structure and growth process characteristics is established. It is proved that the growth process is defined by chaos with limited number of order parameters. According to the offered methods and algorithms, plasma and surface in-situ diagnostics were processed. Methods of film-growth control by synchronization and chaotic driving are discussed.
9:00 PM - A5.2
Microscopic modeling of Phonon Thermal Conductivity in SiGe Superlattices.
Shang-Fen Ren 1 , Wei Cheng 1 2 , Gang Chen 3
1 Department of Physics, Illinois State University, Normal, Illinois, United States, 2 , Beijing Normal University, Beijing China, 3 Mechanical Engineering Department, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States
Show Abstract9:00 PM - A5.3
Molecular Dynamic Computer Simulation of Thin Film's Heat Dissipation Rate.
Ya-Yun Cheng 1 , Horng-Ming Hsieh 2 , C.C. Lee 1
1 , Institute of Optical Sciences, National Central University, Chung-Li 320 Taiwan, 2 , Institute of Nuclear Energy Research, Lungtan, Taoyuan 325 Taiwan
Show AbstractHeat dissipation rate of the thin film is theoretically related to the thickness of the film. As the film grows on a substrate of constant temperature, the heat dissipation rate would be lower when the film becomes thicker. It is difficult to observe rate change during the film growth through experiment. With MD simulation, this rate can be "observed". A system including Platinum substrate at temperature 300K is setup. The constant temperature is obtained by Phantom method. On the substrate, the aluminum atoms are deposited on the substrate at constant rate simulating the condensation of aluminum vapor on the substrate. The MD simulations were performed by integration Newton's equation of motion for atoms sequentially dropping on the substrate with kinetic energy of 0.5 eV. The EAM potential was applied as the inter-atomic potential that properly describes the interaction force between metal atoms. The average temperature of aluminum thin film is 600K at the end of deposition. For various thickness of the film: 3.656nm, 5.302nm, and 7.567nm, the heat dissipation rates after the deposition are calculated. Through the simple equation T=(T0-Ts)exp(-kt/d2ρcp)+Ts, the dissipation rate is R=k/d2ρcp and the parameter α=k/ρcp is the thermal diffusivity. The simulation observed that the thinner the film the larger the parameter R. The value of R is found to be 15.4×10-3ps-1, 10.7×10-3ps-1, and 6.43×10-3ps-1 for the film of thicknesses 3.656nm, 5.302nm, and 7.567nm.The value of α is found to be 2.06×10-19m2/ps, 3.03×10-19m2/ps, and 3.69×10-19m2/ps respectively. The value of R = 6.04×10-3ps-1 for a film of 140.8nm thickness has been reported on experiment. The simulation provides the data of heat dissipation rate relative to the thickness of thin film in the nanometer range.
9:00 PM - A5.4
A Nature of Structural Irreproducibility and Instability of Amorphous Silicon Thin-Films
Tatiana Larina 1 , Nikolay Bodyagin 1 , Sergey Vikhrov 1 , Stanislav Mursalov 1
1 BMPE, RGRTA, Ryazan Russian Federation
Show AbstractStructural irreproducibility and instability are serious problems commonly encountered in the fabrication of amorphous silicon thin-film at any level, first on micro- and nanoscales. However, until now the above difficulties in microelectronics processing stem from lack of the theoretical understanding of irreproducibility. The prospective approach to a problem is based that the irreproducibility in the fabrication of solid materials by treating the development of three-dimensional structural regularity can be considered as self-organization in a nonlinear system. The dynamics of the system is regarded as a key factor responsible for materials irreproducibility, with the limited degree of precision in process variables considered less important.It has proved theoretically and supported by experiment (on amorphous silicon films, obtained by GD-method and LF PECVD method) that solidification essentially includes a stage of deterministic chaos whatever the material. Preceding from this the conclusion was made that sensitivity to the initial conditions (characteristics of chaotic dynamics) in combination with both the limited precision of process variables and unavoidable fluctuations are responsible for the structural irreproducibility. Some numerical measures of irreproducibility those are suitable for any fabrication technology (Kolmogorov – Sinai entropy, Lyapunov exponents, average mutual information) and connect with invariants of chaotic dynamics, were defined. It is shown that the instability of growth processes implies that reproducibility limit has existed and it is ineffective to raise the precision of the initial conditions and process variables above the level dictated by the reproducibility limit.The analytical connection between the scale of the system (the number of atoms formed an independent micro- or nanogrouping) and a degree of reproducibility is established. It is shown that decreasing of system scale signifies that materials reproducibility is decreasing too. It is shown that a simple tighter control of process variables is an ineffective method to increase reproducibility. Other approaches to its magnification, based on knowledge of dynamic processes of fabrication of a structure are formulated. It is justified that the irreproducibility problem can be treated by techniques developed for controlling chaos: chaotic driving, the amplification of external noise, stabilizing chaotic orbits, and the accelerated transition through a bifurcation point.The problem of metastability of films properties is discussed from positions of nonlinear dynamics.
9:00 PM - A5.5
Thermal Conductivity and Natural Cooling Rate of Excimer-laser annealed Si: A Molecular Dynamics Study.
Byoung-Min Lee 1 3 , Hong Koo Baik 1 , Baek Seok Seong 3 , Shinji Munetoh 2 , Teruaki Motooka 2
1 Metallurgical Engineering, Yonsei University , Seoul Korea (the Republic of), 3 Neutron Physics, Korea Atomic Energy Research Institute, Daejeon Korea (the Republic of), 2 Materials Science and Engineering, Kyushu University, Fukuoka Japan
Show AbstractThermal conductivity of substrate material is a key factor to determin the cooling rate of excimer laser annealed Si. To investigate the relationship between thermal conductivity and cooling rate, we have performed MD simulations based on the combination of Langevin and Newton equations to deal with heat transfer from l-Si to c-Si. The thermal conductivity of c-Si was measured by direct method. In order to deal with finite-size effect, different sizes of MD cell of 4×4×96, 4×4×146, 4×4×196, and 4×4×288 Å3 were used. Extrapolation of the system size to infinity yields values of the thermal conductivity of 58 W/mK at 1000 K and 35.7 W/mK at 1500 K, respectively. The result at 1000 K is in reasonable agreement with the experimental results of isotopically enriched Si, which shows the thermal conductivity of around 50 W/mK. The MD cell with a length of 488.75 Å in the direction of heat flow was used for estimating the natural cooling rate. The initial c/l interface system was obtained by setting the temperature of MD cell at 1000 K and 1500 K, respectivelt, for Z≦35 Å and 3800 K for Z>35 Å. During cooling processes, the temperature of bottom 10 Å of the MD cell was controlled. The cooling rate of 7.4×1011 K/sec for 1000 K and 5.9×1011 K/sec for 1500 K was obtained, respectively. From the ratio of thermal conductivity between c-Si and glass, the cooling rate was calculated to be 3.3×109 K/sec, which is in good agreement with experimental cooling rate of Si thin film on glass substrate.
9:00 PM - A5.6
Phononic Amorphous Silicon: Theory, Material, and Devices.
Samrat Chawda 1 , Jose Mawyin 1 , Gary Halada 1 , Charles Fortmann 2
1 Materials Science and Engineering, Stony Brook University, Stony Brook, New York, United States, 2 Department of Applied Mathematics, Stony Brook University, Stony Brook, New York, United States
Show AbstractThe field of phononic-engineered amorphous silicon is introduced. Specifically the construction of devices and waveguides for information conveyance and manipulation via phonons are considered. Typically the phononic properties of a given material are immutable and the phonons have such a limited diffusion length (nanometers) as to be unsuitable for engineering purposes. Crystalline silicon on the other hand has a reasonably large thermal conductivity and phonon diffusion length at sufficiently low temperatures. Phonon diffusion lengths can measure up to centimeters (e.g., crystal SiO2) at temperature less than 10K but drop to sub microns at room temperature. Amorphous silicon, owing to the inherent scattering structures and owing to localization (of at least some phonon bands), has an anomalously large phonon lifetime [1]. This lifetime maybe indicative of a large phonon diffusion length and/or a fast phonon hop rate from one domain to the next and/or an indication that more than the typical three phonons (umklapp process) are involved in phonon scattering (e.g., see [1]&[2]). Techniques involving small-scaled devices and phonon bands to control umklapp phonon-phonon scattering are described. The potential to exploit inherent amorphous silicon structure as well as the engineering (post film deposition) of di-hydride distributions to induce phonon forbidden bands for significantly reduced multi-phonon scattering is explored. The indirect optical band gap of small domain (and possibly amorphous) silicon [3] provides the physical basis for the transduction of phonon and optical energies. Experimental methods for the post-deposition introduction of phonon scattering structure, the transduction of phononic information to optical information, and experimental approaches including the use of micro-Raman to probe phonon spectra and transport are described. The prospects of a fully integrated phononic, photonic, electronic amorphous silicon technology have been described.
9:00 PM - A5.7
Numerical Simulations of the Stefan Problem on Fully Adaptive Meshes.
Han Chen 2 , Chohong Min 3 , Frederic Gibou 1 2
2 Computer Science, UCSB, Santa Barbara, California, United States, 3 Mathematics, UCSB, Santa Barbara, California, United States, 1 Mechanical Engineering, ucsb, Santa Barbara, California, United States
Show AbstractThe Stefan problem is a model at the core of diffusion dominated phenomena and is used for example in the study of thin films grown by molecular beam epitaxy or crystals grown from a melt. The problem involves the tracking of an interface and the ability to resolve small scale details, for example those associated with the formation of small dendrites. In this talk we will describe a level set methodology to solve this problem in an adaptive mesh refinement setting. In particular, we introduce a method for solving the variable coefficient Poisson equation on irregular domains and on fully adaptive Cartesian grids that yields second order accuracy for the solutions and their gradients. We employ quadtree (in 2D) and octree (in 3D) data structures as an efficient means torepresent the Cartesian grid, allowing for constraint-free grid generation. In particular, the discretization at one cell's node only uses nodes of two (2D) or three (3D) adjacent cells, producing schemes that are straightforward to implement. Numerical results in two and three spatial dimensions will be presented.
9:00 PM - A5.9
A Binary Two-dimensional Colloidal Glass former: Amorphous Structure and Heterogeneous Dynamics Analysis.
Hans Koenig 2 1
2 , University of Konstanz, Konstanz Germany, 1 , University of Mainz, Mainz Germany
Show AbstractA6: Poster Session: Measurements of Film and Interface Properties
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A6.1
Raman and X-ray Reflectivity Study of Annealed Polycrystalline Si/Ge Multilayer Structures.
Shilpa Tripathi 1 , R. Brajpuriya 1 , A. Sharma 1 , T. Shripathi 1 , S.M. Chaudhari 1
1 Beamline, UDCSR, Indore, M.P., India
Show Abstract9:00 PM - A6.2
Environment of Er Doped in a-Si:H and Its Relation with Photoluminescence Spectra
Minoru Kumeda 1 , Yoshitaka Sekizawa 1 , Akiharu Morimoto 1 , Tatsuo Shimizu 2
1 , Kanazawa University, Kanazawa Japan, 2 , NTT Microsystem Integration Labs., Atsugi Japan
Show AbstractEmission of the light at 1540 nm of an Er ion is matched with the wavelength region of the lowest loss of the optical fiber, and many studies have been done on Er ions doped in various host materials. Hydrogenated amorphous silicon (a-Si:H) films are a suitable host material because many Er atoms can be added without segregation and the inversion symmetry inhibiting the optical transition among 4f electronic states is destroyed in the amorphous network. Er atoms tends to get O atoms as nearest neighbors and three- to six-fold coordinations with O are reported by EXAFS measurements. A DV-Xa calculation also suggests the Er atom is slightly lifted from the plane where the four O atoms are located. However, it is not clear whether such an Er environment is consistent with the observed Er photoluminescence (PL) spectrum. We have prepared Er doped a-Si:H films by a magnetron sputtering method and performed PL measurements using He-Ne laser as an excitation source. The films contain O atoms unintentionally introduced during the preparation process. The PL spectrum arising from the Er ion observed at low temperature (19K) was decomposed into several Gaussian lines and was compared with the calculated Stark splitting in the ground state 4I15/2. At this low temperature, only the lowest one of the Stark-split levels of the upper 4I13/2 is thought to be populated and so the PL component lines are attributed to the splitting of the 4I15/2 ground state. As is well known, the crystalline potential around the magnetic ion can be expanded using spherical harmonic functions, and the matrix elements of the crystalline potential perturbation can be calculated using the equivalent angular momentum operators Onm. We consider a crystalline potential around the Er atom which is located in a cage of an octahedron of O atoms but is lifted from the plane of four O atoms. Since the odd-order polynomials in the expansion of the crystalline potential has zero matrix elements, terms such as O20, O40, O44, O60, O64 appear in addition to the terms in the cubic-symmetry case. The perturbation matrix of the crystalline potential with 16 x 16 elements is diagonalized and the energy levels separated by the Stark splitting are obtained. In the cubic-symmetry crystalline potential, only five levels appear in accordance with a group theoretical prediction. Coefficients of the above-mentioned equivalent operator terms are adjusted and the resultant Stark levels can be fit to the observed splittings of the PL lines fairly well. The calculated Stark-split levels have eight components but the experimental data show only four lines probably because the linewidths of the PL lines arising from the upper four levels of the eight Stark-split levels are too broad to be decomposed.
9:00 PM - A6.3
Photocarrier Radiometric Lifetime Measurements of Intrinsic Amorphous-Crystalline Silicon Heterostructure.
Keith Leong 1 , Andreas Mandelis 2 , Nazir Kherani 1 , Stefan Zukotynski 1
1 Electrical & Computer Engineering, University of Toronto, Toronto, Ontario, Canada, 2 Mechanical and Industrial Engineering, University of Toronto, Toronto, Ontario, Canada
Show AbstractIn amorphous-crystalline silicon heterojunctions impurities and defects at the interface play an important role in junction performance. Surface cleaning and native oxide etching are important steps before junction fabrication. The major effect of impurities and defects at a heterojunction interface is increase in the surface/interface recombination velocity, which reduces the effective carrier lifetime.The DC saddle field glow discharge PECVD technique was used to deposit intrinsic hydrogenated amorphous silicon (i-a-Si:H) thin films on crystalline silicon (c-Si) substrates, creating an amorphous-crystalline heterostructure. Photocarrier Radiometry (PCR), a novel all optical technique, was used to simultaneously extract four electronic transport parameters: lifetime, diffusivity, back and front surface recombination velocities of crystalline silicon. PCR measures radiative recombination of modulated laser excited carriers from crystalline silicon. PCR, which is interface defect sensitive, was also used to examine the amorphous silicon passivating layer.A series of crystalline silicon wafers were cleaned in various sequences which included the following wet chemical agents: acetone, trichloroethylene, isopropanol, standard silicon wafer cleaning chemicals (RCA1: NH4OH:H2O2:H20; RCA2: HCl:H2O2:H2O), sulfuric acid/hydrogen peroxide mixture (SPM), SPM with 0.01% hydrofluoric acid. The native oxide was etched and the surface passivated with hydrogen using solutions of hydrofluoric acid (HF), buffered HF, or 40% ammonium fluoride (NH4F). Thin (< 5 nm) intrinsic hydrogenated amorphous silicon, i-a-Si:H, passivating layers were grown on the H-terminated wafers.PCR measurements of the i-a-Si:H - c-Si heterostructure yielded values for the electronic transport parameters. An effective lifetime improvement of a factor of 2.5 was observed for the SPM, RCA1, RCA2 cleaning sequence compared to acetone, isopropanol sequence. This was confirmed by Microwave Photoconductance Decay (microPCD) measurements. PCR bulk lifetime estimates were obtained using quinhydrone in methanol solution to passivate c-Si surface. The amorphous-crystalline silicon interface was explored by monitoring the PCR signal as a function of time at different laser intensities. Annealing of the interface is observed to begin at a laser intensity of 2.4 to 2.9 suns.
9:00 PM - A6.4
Combined Defect Density in Amorphous Si Layer and Amorphous-Crystalline Si Interface Obtained with the Constant Photocurrent Method.
Barzin Bahardoust 1 , Nazir Kherani 1 , Stefan Zukotynski 1
1 Electrical & Computer Engineering, University of Toronto, Toronto, Ontario, Canada
Show AbstractAmorphous-crystalline silicon heterojunctions have recently drawn much attention owing to their low-temperature fabrication and high efficiency photovoltaics. The defect density in the amorphous layer(s) and the defect density at the amorphous-crystalline interface play a significant role in the performance of the device. p-type hydrogenated amorphous silicon was deposited on n-type crystalline silicon substrates using the dc saddle-field plasma deposition technique. The doped amorphous layers were deposited at a pressure of 160 mTorr on a heated substrate holder at a temperature of 180 deg.C. The diborane in silane concentration was varied from 1 at.% to 2.5 at.%; the total gas flow rate was 15 sccm. The anode current was 15mA. The film thickness of the amorphous layer was ~10 nm.Using the constant photocurrent method (CPM), the electronic density of states in the amorphous-crystalline silicon samples were studied. We obtained a combined density of defect states in the doped amorphous layer and the amorphous-crystalline silicon interface based on CPM derived absorption and quantum efficiency measurements. CPM inverse intensity at higher energies (>1.4 eV) is observed to increase with decreasing dopant concentration in the p-type amorphous layer. This is attributed to a decrease in the density of defect states in the emitter layer and the amorphous crystalline-silicon interface. A combined defect density of 2.8E+18/cu.cm was measured for the best device. These results correlated well with IV and CV characteristics of the devices. Our results show that the constant photocurrent method can be used to study the defect density of amorphous-crystalline silicon heterojunction post device fabrication.
9:00 PM - A6.5
Microstructure Characterization of Amorphous Silicon-Nitride Films by Effusion Measurements.
Wolfhard Beyer 1 , H.F.W. Dekkers 2
1 Institute of Photovoltaics, Forschungszentrum Juelich, Juelich Germany, 2 , IMEC vzw, Leuven Belgium
Show AbstractAmorphous silicon nitride (a-Si:N:H) films are of interest for passivation of multicrystalline silicon solar cells. Such a-Si:N:H films are deposited on top of the solar cells and, by applying annealing (“firing”) procedures, passivation of defects in the bulk of the crystalline material is achieved. This passivation involves hydrogen diffusion from the a-Si:N:H into the multicrystalline silicon. In order to understand these passivation effects better, we studied the microstructure of the amorphous Si:N:H films by effusion measurements, monitoring e.g. the effusion of D2, HD and H2 from sandwich structures of hydrogenated and deuterated material and the effusion of He and Ne for samples implanted with such atoms. Both measurements can give information on the diffusion of atomic and molecular hydrogen. In the first case, the relative amount of effusing HD is analyzed [1] while in the latter case the effusion data of neon are used to estimate the long-range motion of H2, as Ne and H2 are known to be of similar size [2]. The results show that an increase in substrate temperature from TS = 250°C to 500°C results in a densification of the material as visible in a shift of the He and Ne effusion peaks to higher temperatures. While for TS = 250°C material hydrogen diffuses at temperatures between 500 to 900°C predominantly as H2, for TS = 450 to 500°C material diffusion largely by atomic H must be concluded for temperatures up to 700°C. Annealing is also found to have a strong influence on the type of diffusing species. The results are discussed in focus of optimized H passivation effects.1.W. Beyer, Physica B 170 (1991) 1052.W. Beyer. Phys. Stat. Solidi (c) 1 (2004) 1144
9:00 PM - A6.6
Annealing Effects on the Phase and Electronic Structure Evolutions of SiO Film by Electron Energy Loss Spectroscopy.
Juan Wang 1 , Xiaofeng Wang 1 , Alkiviathes Meldrum 2 , Quan Li 1
1 Physics, The Chinese University of Hong Kong, Hong Kong Hong Kong, 2 Physics, University of Alberta, Edmonton, Alberta, Canada
Show AbstractComposite films consisting of Si nanoclusters embedded in silicon oxide (also known as silicon rich oxide (SRO)) have attracted much attention due to their potential technological importance in the Si-based optoelectronic industry. The annealing process significantly affects the luminescence of the as-deposited SRO films, causing major changes in the integrated luminescence intensity and the peak wavelength (which, in fact, can be tuned across the visible spectrum). As the optical properties of the films are determined by the microstructure/electronic structure of the material, it is important to understand the films’ structure change as a function of the annealing temperature. In the present work, we have carried a systematic study on the microstructure and electronic structure evolution of the SRO film when the annealing temperature increases, using combined transmission electron microscopy and electron energy loss spectroscopy related techniques. We found that the as-deposited SRO film is basically a single phase SiO1.0, as suggested by its electronic structure characteristics disclosed by the valence electron energy loss spectrum. Such single phase undergoes a continuous but incomplete phase decomposition to Si and SiO2 for films annealed from 300-1100oC. The resulted Si phase first appears as small amorphous clusters, which continues to grow to larger sizes at higher annealing temperatures, but only crystallize at a critical temperature of ~800-900oC. Such cluster/matrix configurations of the annealed SiO films are also consistent with the appearance of the interfacial plasmon and its oscillation strength change with the annealing temperature. An interesting correlation between the films’ phase/electronic structure evolution and the trend of their photoluminescence property change is identified. The possible origins of the luminescence are also discussed.
9:00 PM - A6.7
Modeling Dynamical Crack Propagation in Silicon Films Using the ReaxFF Reactive Force Field.
Markus Buehler 1 , Adri Duin 2 , William Goddard 2
1 Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States, 2 Chemistry, California Institute of Technology, Pasadena, California, United States
Show Abstract9:00 PM - A6.9
Direct Observation of the Surface Dynamics of Dangling Bonds During H interaction with a-Si:H.
I.M.P. Aarts 1 , A.C.R. Pipino 2 , W.M.M. Kessels 1 , M.C.M. van de Sanden 1
1 Applied Physics, Eindhoven University of Technology, Eindhoven Netherlands, 2 Chemical Science and Technology Laboratory , National Institute of Standards and Technology (NIST), Gaithersburg, Maryland, United States
Show AbstractUnderstanding the interaction of atomic hydrogen H with the hydrogenated amorphous silicon (a-Si:H) surface is of great importance to obtain knowledge of the growth process of this important semi-conductor material. However, the surface abstraction, passivation and diffusing properties of H are not yet fully understood. Here, using the technique of evanescent-wave cavity ring-down spectroscopy [1], we have realized a unique and absolute absorption technique capable of probing surface dangling bonds that can be applied in situ and during H dosing on hot-wire chemical vapor deposited a-Si:H thin film. The technique probes the dangling bond defect concentration at the surface in absolute terms with an accuracy of 5×109 dangling bonds/cm2 at a time resolution of 0.03 s, which enables dynamic studies of the interaction of H with the a-Si:H surface. In general, upon dosing the a-Si:H surface with H at a substrate temperature of 150°C, we see an sharp rapid increase in dangling bond density due to abstraction of H from the surface, followed by a steady-state situation in which abstraction and passivation of dangling bonds are in equilibrium. When dosing is terminated a complete recovery to the original dangling bond defect density through a rapid passivation and a slow recombination process is observed. Furthermore, flux dependent measurements using a quantified H source show first order kinetics for the abstraction reaction (abstraction probability = ~0.01) while also a flux dependent decay is found. Moreover, we observe a steady-state dangling bond concentration which depends sub-linear on the H flux, which cannot be explained by simple abstraction and passivation processes only. Therefore we present a model that incorporates abstraction, passivation and recombination terms in a set of coupled differential equations that describe the formation and decay of dangling bonds as well as the flux dependent steady-state dangling bond concentration at the surface in agreement with the measurements. On the basis of these results the role of H during silicon-based film growth is addressed.[1] W.M.M. Kessels, I.M.P. Aarts, J.J.H. Gielis, J.P.M. Hoefnagels, and M.C.M. van de Sanden, Mater. Res. Soc. Symp. Proc. 862, A14.3.1 (2005); I. M. P. Aarts, A. C. R. Pipino, J. P. M. Hoefnagels, W. M. M. Kessels, and M. C. M. van de Sanden, Phys. Rev. Lett. 95, 166104 (2005).
A7: Poster Session: Amorphous Silicon Growth
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A7.1
Electrical Characterization of SiCN Films Deposited by HW-CVD Method Using Hezamethyldisilazane.
Takashi Nakayamada 1 , Akira Izumi 1
1 , Kyushu Institute of Technology, Fukuoka Japan
Show Abstract9:00 PM - A7.2
Effect of Hydrogen Dilution on Structure and Electronic Properties of Ge:H and GeY-Si1-Y Films Deposited by Low Frequency Plasma
Andrey Kosarev 1 , Liborio Sanchez 1 , Alfonso Torres 1 , Thomas Felter 2 , Alexander Ilinski 3 , Yurii Kudriavtsev 4 , Rene Asomoza 4
1 Electronics, Inst.Nat.for Astrophysics, Optics and Electronics, Puebla, Puebla, Mexico, 2 Physics, Lawrence Livermore National Laboratory, Livermore, California, United States, 3 Physics, Benemerita Universidad Autonoma de Puebla, Puebla Mexico, 4 Ingeneria Electrica, SEES, CINVESTAV-IPN, Puebla, Mexico DF, Mexico
Show AbstractWe report on systematical study of growth rate, surface morphology, hydrogen and oxygen incorporation, optical and electrical properties in Ge:H and Ge Y–Si1-Y:H, Y> 0.85, films deposited in capacitive reactor by low frequency PE CVD. Silane and germane were used as feed gases diluted by hydrogen. Hydrogen dilution defined as R= QH2/[QSiH4+QGeH4], where QH2, QSiH4, and QGeH4 are gas flows of hydrogen, silane and germane, respectively, was varied in the range of R=20 to 80. Other deposition parameters were: substrate temperature Ts=300 C, power W=300 W, pressure P= 0.6 Torr and total flow QSiH4 + QGeH4= 50 sccm. GeY Si1-Y films were deposited with QSiH4=25 sccm and QGeH4 = 25 sccm. Composition of the films was characterized by SIMS profiling. We observed no significant change of deposition rate Vd in Ge-Si films in all the range of R, while for Ge films Vd significantly reduced after R=50. AFM characterization of the surface morphology demonstrated that at R=50 average height (R) reached maximum in both types Ge and Ge-Si films, while average diameter (R) had minimum in Ge-Si films and maximum in Ge films. To characterize optical properties we used optical gap E04, characteristic energy E03 determined as photon energy at which absorption α(E04)=104 cm-1 and α(E03)=103 cm-1, respectively, and ΔE=E04-E03 that reflects density of band tail states. Both Ge and Ge-Si films have no significant change of E04 in the studied range of R, but demonstrate clearly ΔE minimum at R=50-60 suggesting significant reducing weak bonds in these films. Activation energy of conductivity Ea has no significant change with R, while room temperature conductivity σRT shows maximum at R=40 in both types of the films that reflects behavior of γ factor and probably corresponds less rigid lattice of these films. Both FTIR and SIMS data show general trends of reducing hydrogen and oxygen content with R, though Ge-H bondings have complex structures revealed in spectra. Thus we have found the correlations between structure and electronic properties and showed that they are different in Ge:H and GeYSi1-Y:H , Y> 0.85.The co-authors from INAOE acknowledge the support of this research by CONACyT project No.42367 (CIAM-2002). Work of T.F. was performed under auspices of the US DOE by the University of California Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.
9:00 PM - A7.3
Photoactive Thin Films of GeC Deposited Using a Unique Hollow Cathode Sputtering Technique
Rodney Soukup 1 , Jason Schrader 1 , James Huguenin-Love 1 , Natale Ianno 1 , Vikram Dalal 2
1 Electrical Engineering, University of Nebraska, Lincoln, Nebraska, United States, 2 Electrical and Computer Engineering, Iowa State University, Ames, Iowa, United States
Show AbstractExperimental results on thin films of the new material GexC1-x, deposited by a unique dual plasma hollow cathode sputtering technique are presented here. The (Ge, C) system is extremely promising since the addition of C to Ge has reduced the lattice dimensions enough to allow a lattice match to silicon, while increasing the bandgap close to that of c-Si. The sputtering is accomplished by igniting a dc plasma of the Ar and H2 gases which are fed through Ge and C nozzles, cylindrical tubes 30mm in length with an 8mm O.D. and a 3mm I.D.The basic material, optical, and structural properties were analyzed. Film characterization was performed using Fourier transform infra-red spectroscopy (FTIR), X-ray diffraction (XRD), Raman spectroscopy, x-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), and Auger electron spectroscopy. Additional measurements such as Tauc bandgap, conductivity as a function of temperature and light intensity, and film uniformity have been made. The film properties from a variety of deposition conditions are discussed. The measurements made indicate that he films can be grown so that the C enters the material at lattice sites. In addition, the GexC1-x films are absorb photons much more efficiently than either c-Si or c-Ge.Initial results on doped films will be presented.
9:00 PM - A7.4
The Influence of Deposition Conditions on the Electronic Properties of a-Si:H Prepared in Expanding Thermal Plasmas.
Monica Brinza 1 , Guy Adriaenssens 1
1 Halfgeleiderfysica, University of Leuven, Leuven Belgium
Show AbstractExpanding thermal plasma (ETP) proved itself advantageous for the deposition of a-Si:H films for photovoltaic applications due to the high deposition rates (up to 10 nm/s) that can be achieved and the excellent hole mobility values measured in ETP samples [1]. For a deposition rate of 6-7 nm/s, a substrate temperature of 400-450°C was required for obtaining the best hole mobility of 1.1 cm2V-1s-1. The deposition regime characterized by high substrate temperature and high growth rates produces films with a field independent drift mobility. This property was explained by a band tail having a gaussian rather than exponential profile [2]. Since the high substrate temperature needed for good quality films prevents the production of pin solar cells, it is desirable to devise new methods for lowering the substrate temperature while preserving the high deposition rates and the good electronic quality of the samples. Specifically, the rf biasing of the substrate is currently being investigated, since it is thought that an additional ion bombardment of the growth surface, normally missing in the ETP method, can have beneficial effects. The present study examines the influence of the rf bias on the carrier drift mobility and density of states using the Time-of-Flight (TOF) transient photoconductivity. Although an increase in the photosensitivity of the rf biased samples has been reported [3], our first measurements show a decrease in the hole mobility of the samples subjected to extra ion bombardment. Also, the electron transport properties have worsened. Other questions to be addressed are the manner in which the rf biasing of the substrate during deposition changes the field dependence of the drift mobility and how it influences the deep defect density. A photo-induced barrier-lowering mechanism that was identified before in Cr/ETP a-Si:H sandwich cells [4], proved to be insensitive to the rf bias. [1] W.M.M. Kessels et al., J. Appl. Phys. 89, 2404 (2001).[2] M. Brinza et al., Phys. Rev. B 71, 115209 (2005).[3] A.H.M. Smets et al., Mat. Res. Soc. Symp. Proc. 808, 383 (2004).[4] M. Brinza and G.J. Adriaenssens, J. Mater. Sci.: Mater. Electron. 14, 749 (2003).
9:00 PM - A7.5
Grain Nucleation and Grain Growth During Crystallization of HWCVD a-Si:H Films.
S. Ahrenkiel 1 , A. Mahan 1 , B. Roy 2 , D. Ginley 1
1 , NREL, Golden, Colorado, United States, 2 , Colorado School of Mines, Golden, Colorado, United States
Show AbstractThe crystallization of hydrogenated amorphous silicon (a-Si:H) is being investigated as a viable approach for the production of inexpensive, device-quality, polycrystalline thin films. The crystallization process is comprised of two stages: an initial incubation period during which the majority of the bonded hydrogen is released, and a period of grain nucleation and grain growth. The initial hydrogen content of a-Si:H films increases as the growth temperature is decreased, and this increased hydrogen content has been shown to extend the total crystallization time. After crystallization, material with initially high hydrogen levels also shows increased broadening of x-ray diffraction peaks, indicating an influence of hydrogen on the grain size, defect density, or both [1].For this study, a-Si:H films of high and low hydrogen content were deposited directly on molybdenum, carbon-coated TEM grids by HWCVD and annealed at 600°C for variable times to achieve various degrees of crystallinity. 100-nm-thick films were characterized by TEM without additional thinning. The grain growth in such thin films is nearly two-dimensional, allowing clear identification of crystalline and amorphous regions. Thus, the crystalline volume fraction can be tracked by simple image-processing methods. The evolution of crystallization by grain nucleation and growth for these films is described quite accurately in these films by classical phase-change kinetics. Particle analysis of the randomly distributed grains at early stages of crystallization also provides the areal grain number density and grain size. The final grain size can then be estimated by extrapolation to the fully crystallized state, assuming a constant nucleation rate. We will try to correlate the TEM and x-ray diffraction measurements and identify the structural influences of the initial film hydrogen content on crystallization.[1] A.H. Mahan, B. Roy, R.C. Reedy Jr., D.W. Readey, and D.S. Ginley, J. Applied Physics (2005), in review.
9:00 PM - A7.6
Surface Smoothening Mechanism of Plasma-Deposited Amorphous Silicon Thin Films.
Mayur Valipa 1 2 , Tamas Bakos 1 , Eray Aydil 3 , Dimitrios Maroudas 1
1 Chemical Engineering, University of Massachusetts, Amherst, Massachusetts, United States, 2 Chemical Engineering, University of California, Santa Barbara, California, United States, 3 Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States
Show Abstract9:00 PM - A7.7
The Influence of Thermophoresis Effects During Deposition of Hydrogenated Amorphous Silicon Thin Films.
C. Blackwell 1 , C. Anderson 2 , J. Deneen 3 , C. Carter 3 , U. Kortshagen 2 , J. Kakalios 1
1 School of Physics and Astronomy , University of Minnesota, Minneapolis, Minnesota, United States, 2 Dept. of Mechanical Engineering, University of Minnesota, Minneapolis, Minnesota, United States, 3 Dept. of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States
Show AbstractReports of improved charge transport properties in mixed-phase films of hydrogenated amorphous silicon containing silicon crystalline inclusions have motivated studies of the specific role played by the nanocrystallites in these materials. Silicon cluster formation is known to occur within silane plasmas when a capacitively coupled deposition reactor is operated at high gas chamber pressures. These clusters are sensitive to thermophoretic forces that, depending on the sign of the thermal gradient, direct them towards or away from the silicon film's growing surface. We have developed a deposition system that produces small crystalline silicon particles (3-5 nm diameter) in a flow-through reactor, and injects these particles into a separate capacitively-coupled plasma (CCP) chamber. The structural, optical and electronic properties of these mixed-phase materials are investigated as a function of the controllable thermal gradient applied across the CCP during deposition. These results for the dual-chamber system are compared to mixed-phase films deposited in a single-chamber PECVD system, where TEM imaging confirms that the presence of nanocrystals in the silicon thin film is indeed sensitive to the orientation of the thermal gradient during deposition. The dark conductance is lowest, and the infra-red absorption at 2090 cm^-1 is highest for films grown with a thermal gradient that directs clusters toward the growing film. The properties of nominally homogeneous a-Si:H films, that are without nanocrystalline inclusions, are also found to be sensitive to the thermophoretic force across the silane plasma, with the trend with thermal gradient being qualitatively similar to that observed in materials containing nanoparticles. These results suggest that sub-nanometer clusters may be present in a-Si:H films synthesized under standard deposition conditions. This work was partially supported by NSF grants NER-DMI-0403887, IGERT grant DGE-0114372, and the University of Minnesota.
9:00 PM - A7.8
Gas phase Chemistry and Film Stoichiometry in Hot-Wire CVD of a-SiGe:H
James Doyle 2 1 , Yueqin Xu 1 , Robert Reedy 1 , Howard Branz 1 , A. Mahan 1
2 Physics and Astronomy, Macalester College, St. Paul, Minnesota, United States, 1 , National Renewable Energy Laboratory, Golden, Colorado, United States
Show AbstractHot-wire chemical vapor deposition (HWCVD) amorphous silicon-germanium films (a-SiGe:H) made with a Ta wire at about 1800°C has been used to grow low bandgap films with narrow bandtails and low defect densities [S. Datta, et al., J. Non-Cryst. Solids., in press]. We used mass spectrometry to study silane and germane gas decomposition and depletion in order to understand their relationship to film stoichiometry and growth rate. Under conditions used to deposit our best quality a-SiGe:H, germane dissociates at a rate two to three times that of silane. No higher silanes, germanes, or silylgermane molecules were detected, which implies that stable gas species are not formed in the feed gas dissociation. The Si fraction in the film observed with secondary ion mass spectrometry (SIMS) is somewhat smaller than would result from incorporation of all silane and germane dissociation products in the growing film. This is likely due to preferential alloying of Si compared to Ge in the Ta filament. The dissociation rates of both pure silane and pure germane fall about a factor of 2 over the range 1 to 12 mTorr, but moderate dilution in hydrogen or argon has only a small effect on the feed gas dissociation. Comparisons between HWCVD deposition chemistry for a-Si:H, a-Ge:H and a-SiGe:H will be presented and interactions between silane and germane species in the gas phase and on the filament surface will be discussed. Based on these results, a model will be presented which can be used to predict gas depletion, film growth rate, and film stoichiometry from the filament temperature, gas flow rates, and pressure. This work was supported by the U.S. Dept. of Energy under Contract # DE-AC39-98-GO10337 and the National Science Foundation under NSF-RUI program grant DMR-0513775.
A8: Poster Session: Nanocrystalline and Microcrystalline Silicon Films
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A8.1
Incubation Layer-Free Nanocrystalline-Si Thin Film Fabricated by ICP-CVD at 150oC for Flexible Electronics
Sang-Myeon Han 1 , Joong-Hyun Park 1 , Sang-Geun Park 1 , Min-Koo Han 1 , Young-Kwan Cha 2 , YoungSoo Park 2
1 , Seoul National University, Seoul Korea (the Republic of), 2 , Samsung Advanced Institute of Technology, Yong In, Gyeong Gi, Korea (the Republic of)
Show Abstract Nanocrystalline silicon (nc-Si) thin film transistor (TFT) may be a promising device due to rather simple process and uniformity compared with poly-Si TFT and higher mobility and stability compared with a-Si TFT. To fabricate high performance bottom gate nc-Si TFT, fully crystalline nc-Si film as an active layer is essential. However, during the deposition of the nc-Si film, the incubation layer is formed at the bottom of nc-Si film, which is a kind of a-Si layer. It has been reported that this incubation layer leads inferior TFT characteristics such as low field effect mobility and threshold voltage shift. The efforts to reduce the incubation layer of the nc-Si film were made in order to fabricate good bottom gate nc-Si TFT.Conventional plasma enhanced chemical vapor deposition (PECVD) has been widely employed to deposit nc-Si film. On the other hand, inductively coupled plasma chemical vapor deposition (ICP-CVD) may provide a certain advantages such as high deposition rate and improved crystallinity over PECVD. The purpose of our work is to report the deposition of nc-Si film without any troublesome incubation layer deposited by ICP-CVD. It may be suitable for low temperature bottom gate nc-Si TFTs which also can be applied to flexible displays. We deposited nc-Si film by ICP-CVD at 150oC. ICP power was 400W. The process gas was SiH4 diluted with He, H2 and He/H2 mixture. The flow rate of dilution gas such as He, H2 and He/H2 mixture was varied from 20sccm to 60sccm and that of SiH4 was 3sccm. The crystalline volume fractions evaluated from the Raman spectrum of the nc-Si film were above 70%. The crystalline growth structure and thickness of incubation layer of each nc-Si film was observed by cross-sectional high resolution transmission electron microscopy (HR-TEM). All nc-Si films have crystalline structure of columnar type from the bottom to the top of the nc-Si film. When the nc-Si film was deposited in the dilution ratio of 20:3 and the dilution gas was He, the thickness of incubation layer was thin, which was 20nm. When the He dilution ratio increased to 40:3 and 60:3, the incubation layer was eliminated. In the case of H2 dilution (H2:SiH4=40:3) and He/H2 mixture dilution (He:H2:SiH4=10:10:3 or 20:20:3), the incubation layer of each nc-Si film was also not observed. The deposition rate of these incubation-free nc-Si film is considerably high, from 3.4Å/s to 4 Å/s. In every high resolution TEM image, the Si lattice pattern was observed clearly. The grain size of 30nm was measured by cross-section TEM image. The absence of incubation layer in nc-Si film may be attributed to ICP-CVD which generates remote plasma of high density, the role of hydrogen, and the dilution effect on the growth of crystalline.Our experimental results show that incubation-free nc-Si film deposited by ICP-CVD may be suitable for active layer of bottom gate nc-Si TFTs.
9:00 PM - A8.2
Role of Hydrogen in the Crystal Formation in Microcrystalline Silicon Films.
Gyu-Hyun Lee 1 , Jong-Hwan Yoon 1
1 Physics, Kangwon National University, Chuncheon, Kangwon-do, Korea (the Republic of)
Show AbstractMicrocrystalline silicon (μc-Si:H) films can be easily grown at substrate temperatures of 200-300 oC by plasma-enhanced chemical vapor deposition (PECVD) using highly hydrogen-diluted silane gas. Many studies have shown that hydrogen is the most important factor in the nucleation and growth of μc-Si:H films. However, the detailed role of hydrogen in the μc-Si:H formation remains still unclear. In this work we performed post-growth hydrogenation on μc-Si:H films, and show that chemical annealing[1] by atomic hydrogen is most likely to be responsible for crystal formation in μc-Si:H films.Post-growth hydrogenation was performed by exposing μc-Si:H films of about 185 nm thick to hydrogen plasma in situ after deposition. μc-Si:H films with low crystalline volume fraction (~10 %) were deposited on Corning 7059 glasses at a substrate temperature of 220 oC by plasma-enhanced chemical vapor deposition using highly argon-diluted silane gas, and subsequently exposed to hydrogen plasma. Raman spectra revealed that the intensity near 520 cm-1 caused by crystalline phase increases with hydrogenation time, accompanied by a decrease in the intensity near 480 cm-1 caused by amorphous phase. Cross-sectional transmission electron micrographs (TEM) of μc-Si:H films exposed to hydrogen plasma show the crystalline grains with approximately inverse cone shape. Whilst little, or no grain is observed in the normal samples which were not exposed to plasma. These data provide strong support that crystal formation in the μc-Si:H films can be caused by chemical annealing rather than others. The detailed mechanisms will be discussed.[1] K. Nakamura, K. Yoshida, S. Takeoka and I. Shimizu, Jpn. J. Appl. Phys. 34, 442 (1995).
9:00 PM - A8.3
Influence of Annealing on Crystallinity and Conductivity of p-type Nanocrystalline Si films.
Vikram Dalal 1 , Durga Panda 1
1 Elec. and Computer Engr., Iowa State University, Ames, Iowa, United States
Show Abstract9:00 PM - A8.4
The Influence of the Hot Wire Temperature on the Crystallization of µc-Si:H Films Prepared by Hot Wire Assisted ECR CVD.
Li Ying 1 , Li Zhizhong 2 , Chen Guanghua 2 , Minoru Kumeda 1
1 , Kanazawa University, Kanazawa Japan, 2 , Beijing University of Technology, Beijing China
Show Abstract9:00 PM - A8.5
“Effects of HWCVD-deposited Seed Layers on Hydrogenated Microcrystalline Silicon Films on Glass Substrates”
Michael Adachi 1 , Wing Fai Lydia Tse 1 , Karim Karim 1 , Karen Kavanagh 2
1 School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada, 2 Department of Physics, Simon Fraser University, Burnaby, British Columbia, Canada
Show AbstractMicrocrystalline silicon (μc-Si:H) deposited by the hot-wire chemical vapor deposition (HWCVD) technique has gained considerable attention in recent years due to the potential of fabricating more efficient and in particular, stable solar cells at higher deposition rates. In this study, μc-Si:H seed layers has been grown by HWCVD using a graphite catalyzer filament with the aim of improving the nucleation and thus homogeneity of the μc-Si:H film on glass substrates. Silane with high hydrogen dilution is well known to yield microcrystalline films but at the cost of a reduced deposition rate. In contrast, films deposited with lower hydrogen dilution tend to deposit faster but are in the amorphous phase. For example, in this research,, we found that HWCVD films deposited directly on glass at a 3.3% silane concentration diluted in hydrogen displayed amorphous characteristics. When such films deposited at a high silane concentration were incorporated with an ultra thin, HWCVD deposited, μc-Si:H seed layer of 20 nm thickness, μc-Si:H growth was achieved with improved crystallinity and film homogeneity. The results presented are encouraging for the development of large area, high deposition rate, stable thin film solar cells.
9:00 PM - A8.6
Organic Dye Molecules Incorporated into Hot Wire-CVD Grown µc-Silicon
Ulrich Weiler 1 , Yvonne Gassenbauer 1 , Thomas Mayer 1 , Wolfram Jaegermann 1
1 Surface and Solar Energy Research, Darmstadt University of Technology, Darmstadt Germany
Show AbstractGrowth of µc-Si / organic dye compounds has been explored as a way of incorporating typical functionalities of organic molecules to µc-Si. Compounds with different dye concentrations have been produced via sequential deposition of silicon by remote hot wire-CVD and zink phthalocyanines by PVD. The films were characterized by Raman spectroscopy, transmission spectroscopy and time resolved microwave conductivity. Compared to pure dye films these measurements show only minor changes in the phthalocyanine features suggesting that the organic material survives the radical bombardment of the CVD-Si process. Furthermore a systematic gradual shift of the dye HOMO state to higher binding energy in dependence of the number of substitute F atoms has been observed with photoelectron spectroscopy. The HOMO maxima are found 0.45, 1.0, 1.0 and 2.05 eV below the Si valence band edge for ZnPc, F4ZnPc, F8ZnPc, and F16ZnPc, respectively indicating a direction of rational electronic state line up engineering for optimizing charge transfer processes between dye and µc-Si.
9:00 PM - A8.7
Spatial Distribution of the SiH3 Radicals in VHF Plasmas Under μc-Si Thin Film Growth Conditions.
Takehiko Nagai 1 , Arno Smets 1 , Michio Kondo 1
1 Research Center for Photovoltaics, National Institute of Advanced Industrial Science and Technology (AIST), Ibaraki Japan
Show Abstract9:00 PM - A8.8
Photoluminescence in Silicon Rich Silicon Oxide Thin Films.
Wei Pan 1 , Malcolm Corroll 1 , Roberto Dunn 1
1 , Sandia National Labs, Albuquerque, New Mexico, United States
Show AbstractMuch progress has been made in highly luminescent silicon nanocrystal (SNCs). Effects of nanocrystal size and density are known to strongly affect the SNC optical properties. Yet their mechanisms are not completely understood. At Sandia, SNCs are formed by first depositing silicon rich silicon oxide (SRSO) films on a [100] Si wafer, using high-density plasma chemical vapor deposition. The as-deposited SRSO films are then annealed at ~ 1000C in the nitrogen ambient for 10 minutes, which phase segregates the extra Si into SNCs in the oxide. Compared to these high-thermal-budget (HTB) samples, the SRSO films without N2 annealing are called the zero-thermal-budget (ZTB) samples. In this talk, we will report photoluminescence (PL) results in these two kinds of samples. In the first experiment, we studied the temperature dependence of the PL intensity. In the HTB samples, the PL intensity shows a non-monotonic temperature dependence, with the peak at ~ 60K. This non-monotonic temperature dependence of PL has been observed in SNC samples, and is believed to be due to the formation of the exchange-interaction-induced singlet and triplet levels [1]. In the ZTB samples, on the other hand, the PL intensity shows a monotonic temperature dependence and decreases with increasing temperature. In the second experiment, we studied the PL intensity at the fixed temperature of ~ 77K, but with/without a post-annealing in the forming gas. It is observed that after the forming gas annealing, the PL intensity in the HTB samples increases, while it decreases in the ZTB sample. It is known that post-annealing in the forming gas helps to reduce the density of defect states [2]. Consequently, it is expected that the PL intensity should increase if PL is due to the formation of SNCs. On the other hand, if PL is due to the radiative defect states, post-annealing will reduce the PL intensity. Based on these two observations, we thus conclude that the PL in our HTB samples probably is due to the formation of SNCs, while the PL in the ZTB samples most likely is due to the radiative defect states. Having sought out the origin of PL in our different thermal budget samples, it remains puzzling to observe that in both kinds of samples the PL peak occurs at the same wavelength. In fact, we have systematically looked over 20 samples of different thermal budgets, and the PL peak always occurs at ~ 1000 nm and varies little from sample to sample. At the present time, its origin remains unclear to us. Sandia National Labs is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin company, for the United States Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.References[1] M. L. Brongersma et al, Appl. Phys. Lett. 76, 351 (2000).[2] K. S. Min et al, Appl. Phys. Lett. 69, 2033 (1996).
9:00 PM - A8.9
Nucleation and Growth of Quasicrystalline Silicon Thin Films on Glass by Ceramics Hot Wire Chemical Vapor Deposition
Abdul Middya 1 , Jian-Jun Liang 2 , Kartik Ghosh 3
1 Physics, Syracuse University, Syracuse, New York, United States, 2 Physics, Syracuse University, Syracuse, New York, United States, 3 Physics,Astronomy and Materials Science , Missouri State University, Missouri, Missouri, United States
Show AbstractA9: Poster Session: Defects and Metastability
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A9.1
Relationship between Phase Shift, Square-Wave Response and Density of States in Modulated Photocurrent Spectroscopy
Steve Reynolds 1 , Charlie Main 2
1 Institute of Photovoltaics, Forschungszentrum Juelich, Juelich, NRW, Germany, 2 Division of Electronic Engineering and Physics, University of Dundee, Dundee, Angus, United Kingdom
Show AbstractThe density of states distribution (DOS) in a semiconductor may be probed by means of modulated photocurrent spectroscopy (MPC), in which the complex photocurrent response to a modulated light source is measured as a function of modulation frequency. Depending on the experimental conditions, behavior may be governed by carrier trapping and/or emission involving states above the Fermi level [1] (the so-called high-frequency (HF) regime), or by recombination via states close to the Fermi level (the LF regime) [2]. It has been demonstrated that both methods, when applied to the same sample, yield a consistent interpretation of the DOS and, additionally, may provide information on capture coefficients [3]. Here we investigate the MPC response in both HF and LF regimes under sinusoidal and square-wave excitations, in more detail than reported hitherto. We show that for coplanar thin-film silicon samples, the response to low modulation-index square-waves and sinusoids may differ significantly from that expected for a ‘single-pole’ network, there being for example no simple quantitative link between square-wave ‘decay time’ and sinusoidal ‘phase shift’. This suggests that significant contributions from states located more than kT from the probe energy may occur, which could lead to incorrect estimations of the DOS. Observations and computer simulations over a range of temperatures and excitations [4] enable us to check the validity of HF and LF MPC as DOS probes, and the self-consistency of the analyses.[1] R Brueggemann, C Main, J Berkin and S Reynolds, Phil. Mag. B 62, 29 (1990).[2] RR Koropecki, JA Schmidt and R Arce, J. Appl. Phys. 91, 8965 (2002).[3] JP Kleider, C Longeaud and ME Gueunier, J. Non-Cryst. Solids 338-340, 390 (2004).[4] A Dussan, JA Schmidt, RD Arce, RH Buitrago and RR Koropecki, Thin Solid Films 449, 180 (2004).
9:00 PM - A9.2
Thermally-Stimulated Currents in Thin-Film Semiconductors: Analysis and Modelling
Charles Main 1 , Nacera Souffi 2 , Steve Reynolds 3 , Rudi Brueggemann 2
1 Department of Electronic Engineering and Physics, University of Dundee, Dundee United Kingdom, 2 Physics Institute, Carl von Ossietzky University Oldenburg, Oldenburg Germany, 3 Institute for Photovoltaics, Forschungszentrum Juelich, Juelich Germany
Show AbstractAbstractWe follow up a recent publication [1] in which the authors illustrated the surprising robustness of the thermally stimulated current technique (TSC) as a method to determine the density of states distribution (DOS) in thin film semiconductors under a wide range of conditions. In the present paper, we use numerical simulation to solve the non-linear time-dependent rate equations for free and trapped charge in systems with continuous and also highly structured DOS profiles. We examine for these very different systems the limits of the method’s apparent immunity to varying conditions of strong and weak retrapping, and investigate more closely the corrections required for variations in carrier lifetime with temperature.[1] C. Main, N. Souffi, S. Reynolds, R. Brueggemann, Paper presented at ICANS 21, Lisbon, Sept 4 – 9, 2005, to be published J. non-Cryst Solids.
9:00 PM - A9.4
A Comparison of 1H NMR Characteristics for Stable and Metastable Paired Hydrogen Sites in a-Si:H.
David Bobela 1 , Tining Su 2 , Craig Taylor 2 , Gautam Ganguly 3
1 Physics, University of Utah, Salt Lake City, Utah, United States, 2 Physics, Colorado School Of Mines, Golden, Colorado, United States, 3 , United Solar Ovonics Corp., Troy, Michigan, United States
Show AbstractIn a defective sample of a-Si:H we have recently reported the presence of a stable paired hydrogen configuration, (SiH2)n (n ≥ 1) where the hydrogen-hydrogen separation is on average 1.8 Å as determined from detailed fittings of the lineshape [1]. For n = 1 the configuration corresponds to dihydride bonding, which is known to exist from infrared absorption measurements. Recent calculations for a-Si:H place the hydrogen-hydrogen separation for SiH2 at approximately 2.4 Å [2]. In addition, a metastable paired hydrogen site has also been observed in light-soaked samples of a-Si:H and tied to the defects created in the Staebler-Wronski effect [3]. A crude analysis of the lineshape of this metastable site suggests that the hydrogen-hydrogen separation is approximately 2.3 Å, but this lineshape is not known accurately enough to distinguish it from the stable hydrogen doublet observed in a defective sample. On the other hand, the temperature dependence of the spin-lattice relaxation times appears to be different for the stable and metastable sites [4]. Therefore, the evidence is ambiguous as to whether the metastable sites are a subset of the stable sites or are microscopically different. In an attempt to resolve these ambiguities we report measurements on a large volume sample of device grade a-Si:H, for which half has been light soaked.
A10: Poster Session: Optical Properties
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A10.1
Dielectric Functions of a-Si1-xGex:H versus Ge Content and Temperature: Advances in Optical Function Parameterization
Nikolas Podraza 1 , Deepak Sainju 1 , Christopher Wronski 2 , Robert Collins 1
1 Department of Physics and Astronomy, University of Toledo, Toledo, Ohio, United States, 2 Department of Electrical Engineering, Pennsylvania State University, University Park, Pennsylvania, United States
Show AbstractParameterizations of the optical functions of semiconductors have two important applications. First, they can be used in performance simulations of optoelectronic devices that involve absorption of photons -- solar cells being one important example. In such simulations, device performance can be predicted even with material compositions and operating temperatures for which specific dielectric function ε = ε1 + iε2 sets are unavailable, as long as the wavelength-independent free parameters that describe ε can be approximated as known polynomial functions of composition and/or temperature in a comprehensive database. Second, such parameterizations can be applied in fitting transmittance, reflectance, or ellipsometric spectra collected on multilayer stacks that include one or more of the layers in order to extract structural and compositional parameters, or even temperature. The most popular parameterization for amorphous semiconductors is based on a model for ε2 that includes a parabolic-band, constant-momentum-matrix-element (Tauc) expression at near gap energies, coupled to a Lorentz oscillator at high energies.1 Then ε1 is determined by a Kramers-Kronig transformation. More recent research has shown that a similar approach that adopts instead a constant-dipole-matrix element (Cody) formalism as well as an additional parameter that controls the transition energy between the near gap absorption onset and the Lorentz oscillator behavior provides a superior fit with improved internal consistency.2,3 We have applied this more recent approach to analyze data for the dielectric functions of a-Si1-xGex:H alloys as a function of measurement temperature at selected Ge contents, obtained in situ by spectroscopic ellipsometry upon cooling freshly-deposited samples from 200oC to room temperature. The resulting parameterizations provide insights into enhanced ordering due to protocrystallinity in the alloys, as well electron-phonon interactions and their changes with alloying. The parameterization also provides predictions for the change in short-circuit current with operating temperature in multijunction solar cells. REFERENCES[1] G.E. Jellison, Jr., and F.A. Modine, Appl. Phys. Lett. 69, 371 (1996); 69, 2137 (1996).[2] A.S. Ferlauto, G.M. Ferreira, J.M. Pearce, C.R. Wronski, R.W. Collins, X. Deng, and G. Ganguly, J. Appl. Phys. 92, 2424 (2002). [3] J. Price, P.Y. Hung, T. Rhoad, B. Foran , and A.C. Diebold, Appl. Phys. Lett. 85, 1701 (2004
9:00 PM - A10.2
The Optical Response of Silicon Films Prepared Through Molecular Beam Deposition.
Li-Lin Tay 2 , David Lockwood 2 , Jean-Marc Baribeau 2 , Mario Noel 3 , Joanne Zwinkels 3 , Farida Orapunt 1 , Stephen O'Leary 1
2 Institute for Microstructural Sciences, National Research Council of Canada, Ottawa, Ontario, Canada, 3 Institute for National Measurement Standards, National Research Council of Canada, Ottawa, Ontario, Canada, 1 Faculty of Engineering, University of Regina, Regina, Saskatchewan, Canada
Show AbstractUsing an ultra-high vacuum molecular beam epitaxy deposition system, we have deposited eleven silicon films on quartz substrates for a variety of different growth temperatures, ranging from 98 to 572 oC. From measurements of the specular reflectance spectrum at near normal incidence and the regular transmittance spectrum at normal incidence, we have determined the spectral dependence of the refractive index, the extinction coefficient, the optical absorption coefficient, and the real and imaginary components of the dielectric function. These optical dispersion relationships are contrasted with those corresponding to other forms of amorphous silicon. From this analysis, we have determined how the growth temperature influences the optical response of these deposited silicon films. We find that the optical response become more crystalline-silicon like as the growth temperature increases, and that there is a particularly dramatic transition in the nature of the optical response as the growth temperature increases beyond 450 oC. We suggest that this dramatic change in the nature of the optical response corresponds to a transition in the structural nature of the deposited silicon films, from disordered amorphous silicon to a more ordered form.
A11: Poster Session: Mechanical Properties
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A11.1
Stress Transfer and Protection Layer for Advanced Semiconductor Technology.
Frank Wirbeleit 1 , Michael Belyansky 2 , Deepal Wehella-gamage 2 , Marc Passaro 2 , Anupama Mallikarjunan 2 , Huilong Zhu 2 , Sey-Ping Sun 2 , John Pellerin 2 , Kai Frohberg 1 , Martin Gerhardt 1 , Manfred Horstmann 1 , Rolf Stephan 1 , Michael Raab 1
1 Technology Department, AMD Saxony LLC & Co. KG, Dresden, Sachsen, Germany, 2 IBM Systems & Technology Group, AMD cooperation at IBM Semicconnductor Research and Development Center (SRDC), Hoopewell Junction, New York, United States
Show AbstractA12: Poster Session: Other Materials
Session Chairs
Virginia Chu
Arokia Nathan
Wenchang Yeh
Wednesday AM, April 19, 2006
Salons 8-15 (Marriott)
9:00 PM - A12.1
Role of Surface on the Persistent Photoconductivity in Porous Silicon and Boron Doped a-Si:H*.
S Agarwal 1 , N Mandal 1 , Abhishek Kumar 1
1 Department of Physics, I.I.T. Kanpur, India, U.P., India
Show Abstract9:00 PM - A12.2
Pulsed Laser Deposition of Boron Doped Si70Ge30
Sherif Sedky 1 2 , Ibrahim El Deftar 2 , Omar Mortagy 2
1 Physics, The American University in Cairo, Cairo Egypt, 2 , The Science and Technology Research Center, Cairo Egypt
Show AbstractSilicon germanium (Si1-xGex) is considered as an attractive material for a wide variety of applications. The incorporation of Ge into heavily doped p-type poly Si causes the gate work function to be noticeably reduced, allowing both PMOS and NMOS surface channels to be realized. Its low thermal conductivity as compared to silicon makes it suitable for the realization of high performance uncooled thermal detectors. The capability of Si1-xGex to absorb solar radiation falling in the near infrared improves the efficiency of solar cells. In addition, the low transition temperature from amorphous to polycrystalline is attractive for low-thermal budget applications such as the fabrication of high-performance thin-film transistors on glass substrates. Recently, LPCVD and PECVD Si1-xGex films have been used as structural and sacrificial layers for MEMS that can be monolithically integrated with the driving electronics. For these studies, the deposition temperature was reduced down to 370οC to achieve compatibility with standard CMOS backend and at the same time minimize the mean stress and stress gradient to realize flat suspended structures at such low temperature. The main objective of this work is to investigate the possibility of combining pulsed laser deposition (PLD) and pulsed laser annealing to realize high quality Si1-xGex thin films suitable for post-processing MEMS on top of standard pre-fabricated driving electronics. The main advantage of this approach is that the substrate is kept at room temperature throughout the deposition and thus the MEMS integration process will have no thermal impact on the underlying layers. In addition, it is demonstrated in this work that PLD Si1-xGex can be directly deposited on SiO2 and there is no need for a silicon nucleation layer. Furthermore, this technique is much more economical than PECVD and LPCVD as it does not imply using expensive gas precursors such as germane and silane. This paper reports on depositing SiGe by ablating Boron doped Si70Ge30 targets using KrF excimer laser operated at 50 Hz and at a pulse fluence varying from 0.8 J/cm2 to 3.3 J/cm2. The effect of the type of background gas, being either Ar, N2 or vacuum, on the quality of the deposited films is investigated using field emission scanning electron microscopy (FESEM), atom force microscopy (AFM), four point measurements and bow measurements. It is demonstrated that the optimal deposition conditions that yield a smooth film having low cluster density is a pulse fluence of 0.8 J/cm2 at 50 Hz and a deposition pressure of 5 x 10-5 mbar. In this case the as-deposited films are amorphous having a sheet resistance of 30 MΩ/sq. These films are then crystallized, locally, using pulsed laser annealing with a pulse having a uniform intensity over an area of 0.48 x 0.48 cm2. After a single laser pulse having a fluence of 182 mJ/cm2, the sheet resistance drops by more than four orders of magnitude, and the RMS surface roughness is 70 nm.
9:00 PM - A12.3
Blue Light Emission from PECVD Deposited Nanostructured SiC.
Liudmyla Ivashchenko 1 , Andriy Vasin 2 , Volodymyr Ivashchenko 1 , Mykola Ushakov 1 , Andriy Rusavsky 2
1 Laboratory 61, Institute for Problems of Material Science, NAS, Ukraine , Kyiv Ukraine, 2 , Institute of Semiconductor Physics, NAS, Ukraine, Kyiv Ukraine
Show Abstract9:00 PM - A12.4
Novel Semiconducting Phase of Amorphous Carbon Nickel Composite Films.
Somnath Bhattacharyya 1 , S Henley 1 , N Blanchard 1 , S Silva 1
1 , ATI, University of Surrey, Guildford United Kingdom
Show Abstract
Symposium Organizers
Harry A. Atwater California Institute of Technology
Virginia Chu INESC Microsistemas e Nanotecnologias
Sigurd Wagner Princeton University
Kenji Yamamoto Kaneka Corporation
Hsiao-Wen Zan National Chiao Tung University
A13: Nanocrystalline Silicon Growth
Session Chairs
Wednesday AM, April 19, 2006
Room 3002 (Moscone West)
9:30 AM - **A13.1
Electronic Properties of Nanocrystalline Si and (Si,Ge)
Vikram Dalal 1
1 Elec. and Computer Engr., Iowa State University, Ames, Iowa, United States
Show AbstractNanocrystalline Si:H and (Si,Ge):H alloys are becoming increasingly important for photovoltaic energy conversion. In this paper, we review the growth chemistry and the fundamental electronic and optical properties of these materials and their relationship to growth. It is known that the material properties such as mobility and diffusion lengths of holes depend both on crystallinity and are also dependent upon the direction in which they are measured. For photovoltaic devices, we need a measurement of electronic properties in vertical or growth direction, not in horizontal direction. The optical properties were measured in device-type structures using combination quantum efficiency under far reverse bias in novel structures which prevent back-reflection. The mobility of both electrons and holes was measured in device type structures (n+nn+ or p+pp+) fabricated on stainless steel substrates. Mobilities were measured using space charge limited current techniques and by combining measurement of diffusion lengths with independent measurement of carrier lifetimes. Electron mobility was found to increase with grain size. Hole lifetimes were measured using reverse recovery techniques in actual solar cell devices. Diffusion length was measured using quantum efficiency techniques. Diffusion lengths and carrier lifetimes were correlated with the measurement of defect density measured using variable frequency capacitance techniques. The critical recombination centers were shown to lie in the range of 0.35 -0.5 eV below the conduction band edge. The capture cross-section of these traps was also measured and found to be in the range of 2x10-16 cm2. All the parameters were found to depend upon the degree of crystallinity, and it is shown that the presence of amorphous tissue at the grain boundaries significantly improves carrier transport.
10:00 AM - A13.2
Numerical Simulation of Microcrystalline Silicon Growth on Structured Substrate.
Martin Python 1 , Evelyne Vallat-Sauvain 1 , Julien Bailat 1 , Christophe Ballif 1 , Arvind Shah 1
1 , Institute of Microtechnology IMT, Neuchatel Switzerland
Show AbstractThe growth of thin-film silicon close to the amorphous/microcrystalline transition is qualitatively described by a 3D - discrete dynamical growth model on a cubic lattice. The result of this simulation is a representation of the microstructure of the layer as function of time, i.e. computer-generated animations of growing microcrystalline silicon layers. It permits to follow the evolution of the nucleation and of the growth of the crystalline phase, the surface roughness and the average crystalline volume fraction, the void volume fraction, as well as the surface roughness. In these computer simulations, the substrate surface morphology, the angle of incidence of the adatoms, their surface diffusion length and the rules governing the binding of atoms to the growing films, as well as the layer thickness can be varied.In a previous work [1], the incoming angle of incidence of adatoms was considered vertical, and relaxation was implemented by searching the lowest site within the first and second nearest neighbors around the adatom impinging position. Here, we present new results considering the effects of random, isotropic incidence (with cosine distribution) and lateral sticking probability. The introduction of isotropic incidence leads to more pronounced shadowing effects due to the substrate roughness. Furthermore, in this new version, local relaxation is achieved by the maximization of the number of nearest neighbors around the incoming adatoms arrival site.In this study, we describe and discuss the influence of the substrate surface morphology on the computer-generated film microstructure. The substrate topography can be selected either flat or with pyramidal structures in 1D or 2D, or with periodic trenches in 1D or 2D. Shadowing effects, due to the combination of random isotropic incidence of the adatoms and of the substrate morphology, are found to play an important role for the nucleation and the growth of microcrystalline phase.These results are then compared with layers grown by VHF-PECVD on substrates with periodic structures. The microstructure of the layers is studied by transmission electron microscopy (TEM). This comparison helps us to understand the mechanism of growth of microcrystalline Si and gives particular insights into the generation of micro-cracks in the layer. These cracks are generally found to appear above the lowest points of the textured substrates, both in the simulation and in the TEM cross-sections of the films.[1] J. Bailat et al., Phase diagram and microstructure of microcrystalline and amorphous silicon: a numerical growth simulation, Proceedings of the Materials Research Society, vol 808, 2004, pp. 221-226.
10:15 AM - A13.3
Fast Deposition of Highly Crystallized Microcrystalline Si Films Utilizing a High-Density Microwave Plasma Source for Si Thin Film Solar Cells.
Haijun Jia 1 2 , Hajime Shirai 2 , Michio Kondo 1
1 Research Center for Photovoltaics, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Japan, 2 , Graduate School of Science and Engineering, Saitama University, Saitama Japan
Show Abstract A high-density and low-temperature (∼1012 cm-3 and 1.5∼2.5 eV in Ar) microwave plasma utilizing a spoke antenna for the fast deposition of microcrystalline silicon (μc-Si) films [1] has been studied. In this work, we investigate the influences of source gas supply method as well as substrate bias on the film deposition rate and structural properties. We also demonstrate the high rate deposition of highly crystallized μc-Si films with low defect density. The optical emission spectroscopy (OES) measurements reveal that the SiH emission intensity and Hα/SiH intensity ratio are possible monitors for the film deposition rate and Raman crystallinity, respectively. The moderate negative dc bias at the substrate provides appropriate ion bombardment that is effective to improve film crystallinity, especially the preferred (220) orientation, and to reduce defect density with maintaining a high deposition rate. After optimizing the source gas supply configuration and plasma conditions, a fast deposition rate of 65 Å/s has been achieved for the μc-Si films with high Raman crystallinity Ic/Ia>3 and low defect density of 1-2×1016 cm-3. These results imply that a microwave plasma using a spoke antenna is beneficial to supply the sufficient amount of deposition precursors such as SiH3 as well as atomic hydrogen without ion damage for the synthesis of highly crystallized μc-Si films with low defect density. A novel gas supply configuration is also beneficial for the film homogeneity. The high rate grown μc-Si films with optimized properties are integrated into the p-i-n structure solar cells and the preliminary results will be presented.[1] H. Jia, M. Nakajima, A. Nakao and H. Shirai: Jpn. J. Appl. Phys. 43 (2004) 7960.
10:30 AM - A13.4
Development of Hydrogen Dilution Graded Nanocrystalline Silicon Thin Films and Comprehensive Analysis with Spectroscopic Ellipsometry.
Chandan Das 1 , W. Du 1 , J. Stoke 1 , N. J. Podraza 1 , R. W. Collins 1 , X. Deng 1
1 Department of Physics & Astronomy, The University of Toledo, Toledo, Ohio, United States
Show AbstractThin film hydrogenated nanocrystalline silicon (nc-Si:H) has received considerable attention either as the intrinsic layer of a single junction solar cell or as the bottom i-layer component of a multijunction a-Si:H-based solar cell. During the nc-Si:H deposition process, if the hydrogen dilution of the Si source gas is kept constant, the crystalline volume fraction in the layer increases with thickness due to a well-recognized evolutionary pattern of mixed-phase film growth. To suppress the natural structural evolution of nc-Si:H, hydrogen dilution grading (to decreasing dilution with increasing thickness) has been recommended so that a constant (but adjustable) crystalline volume fraction can be obtained throughout the growth process. In this paper, we present research results on the development of nc-Si:H films with graded hydrogen dilution as well as on the associated characterization of the structural phases using ex-situ spectroscopic ellipsometry (SE). Two different substrates are used: glass and stainless-steel/Ag/ZnO, the former due to its smoothness and the latter as a preliminary step towards applying real-time SE in the actual device configuration. To mimic in-situ measurements in the ex-situ studies, nc Si:H films have been prepared with different accumulated thicknesses, while maintaining the same grading profile of hydrogen dilution, thus enabling us to study different growth stages in the ex-situ mode. The intrinsic nc-Si:H films have been grown by VHF-PECVD from a (Si2H6 + H2) mixture on top of a thin n-type a Si:H layer deposited on both types of substrates. For the series of samples on glass, when the hydrogen dilution ratio (R=[H2]/[Si2H6]) was varied from 200 to 118 in 37 minutes, the corresponding crystalline volume fractions (fc) evaluated by SE are: 69%, 82%, and 70%, at 15 min (bulk thickness, db=175 nm), 22 min (db=310 nm), and 37 min (db=533 nm), respectively. The value of fc at 37 min is suppressed towards the initial value by the dilution grading; however, an increase is observed in the intermediate stage of 22 mins. Such information enables one to tune the grading of R to maintain a required constant fc. These results have been compared with those measured by Raman spectroscopy and are in good agreement, strengthening validity of the analytical procedure. In spite of the progress achieved, ex-situ characterization of films >1 μm has its limitations due to the number of individual depositions required to establish a stable profile in fc. As a result, we are currently installing in-situ analysis capabilities using spectroscopic ellipsometry (SE). Our first results in the real-time mode will be presented at this meeting.
10:45 AM - A13.5
Low Temperature Fabrication of Microcrystalline Silicon Germanium Films by RF Reactive Magnetron Sputtering Method.
Isao Nakamura 1 , Toru Ajiki 1 , Masao Isomura 1
1 Department of Electrical and Electronic Engineering, Tokai university, Kanagawa Japan
Show AbstractA14: Laser Crystallization
Session Chairs
Wednesday PM, April 19, 2006
Room 3002 (Moscone West)
11:30 AM - **A14.1
Remarkable Enlargement of Grain Size in Excimer-laser Crystallization of Si Film and Fabrication of Single Crystalline Si Array.
Wenchang Yeh 1
1 Electronics Engineering, National Taiwan University of Science & Technology , Taipei Taiwan
Show Abstract12:00 PM - A14.2
High Efficiency Crystallization of Silicon Thin Films Using Continuous Wave Infrared Laser.
Naoki Sano 1 2 , Masato Maki 2 , Toshiyuki Sameshima 2
1 , Hightec Systems Corporation, Yokohama, kanagawa, Japan, 2 , Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan
Show Abstract We propose a high efficiency crystallization method for silicon thin films using the diamond-like- carbon (DLC) photo-absorption layer. DLC films had low refractive indices from 1.3 to 1.9 and high extinction coefficients from 0.8 to 0.9 for wavelengths from 250 to 1100 nm. These properties result in optical absorbance higher than 0.7 for 200-nm-thick DLC films for wavelengths shorter than 1100 nm. DLC films are also highly heatproof [1]. A high power infrared laser can be applied for crystallization of silicon thin films by the DLC photo absorption layer [2]. 50-nm-thick amorphous silicon films were formed on glass substrates by the plasma enhanced chemical vapor deposition (PECVD) with a SiH4 gas. DLC films with a thickness of 200 nm were subsequently formed on the silicon films by PECVD with a C2H2 gas. Continuous wave (CW) Nd:YAG laser with wavelength of 1064 nm with 23W was normally irradiated onto samples of DLC/Si/glass with a beam diameter of 400 μm at room temperature. The laser beam was scanned at a constant speed of 4~30 cm/s keeping the size and the shape stable by a galvanic mirror and an fθ lens unit. After removing the DLC films by oxygen plasma, Raman scattering spectral measurements were carried out using a 514.5 nm excitation laser for structural analysis of the silicon films. A high scattering intensity and a sharp phonon band of crystalline silicon were observed in the laser irradiated regions. There was no residual amorphous band. This means that the silicon strip regions with a width of 400 μm were completely crystallized by the present method. The DLC top layer efficiently absorbed Nd:YAG laser light and was heated to a high temperature above 2000oC. The silicon films were heated and crystallized by heat diffusion from the hot DLC layer. On the other hand, no crystallization was observed when CW Nd:YAG laser was directly irradiated to 50-nm-Si/quartz, because silicon has no optical absorption coefficient at 1064 nm. We will also discuss crystallization of silicon films using an infrared laser diode with an energy conversion efficiency above 40%. REFERENCES[1] T. SAMESHIMA and N. ANDOH, Mat. Res. Soc. Symp. Proc. 849, KK9.5, 2004.[2] T. SAMESHIMA and N. ANDOH, Jpn. J.Appl. Phys. 44 (2005)7305-7308.
12:15 PM - A14.3
The Crystallization Mechanism of poly-Si Thin Film Using High-power Nd:YAG Laser with Gaussian Beam Profile.
Hsiao Wen Zan 1 , Chang-Yu Huang 1 , Kazuya Saito 3 , Kouichi Tamagawa 4 , Jack Chen 2 , Tung Jung Wu 2
1 National Chiao Tung University, Department of Photonics and Display Institute, HsinChu Taiwan, 3 ULVAC Inc., Chiba Institute for Super Materials, Chiba Japan, 4 ULVAC Inc., Chigasaki , Kanagawa Japan, 2 , ULVAC Taiwan Inc., HsinChu Taiwan
Show Abstract12:30 PM - A14.4
Explosive Crystallization of Amorphous Solid Films in the Presence of Melting
Costas Grigoropoulos 2 , Matthew Rogers 2 , Seung Hwan Ko 2 , Alexander Golovin 1 , Bernard Matkowsky 1
2 Mechanical Engineering, University of California, Berkeley, California, United States, 1 Engineering Sciences and Applied Mathematics, Northwestern University, Evanston, Illinois, United States
Show AbstractExplosive crystallization (EC) of thin amorphous solid films of germanium is investigated experimentally and theoretically. EC regime characterized by a propagating melting layer between the amorphous and the crystalline phases is considered. Laser-induced, linear EC fronts, uniformly propagating over large distances are achieved in films with various thicknesses deposited on quartz substrate. Depending on the front speed, the film thickness and the substrate temperature, different types of morphology of the resulting crystal phase are observed: columnar, scalloped and mixed. A theory of EC in the presence of melting is developed. The EC front propagation speed is calculated as a function of the substrate temperature and the film thickness; it is found to be in a good agreement with experiments. Linear stability analysis of a uniformly propagating planar EC front is performed. It is shown that for the parameter values where the columnar crystalline structure was observed the front is unstable with respect to a fingering instability similar to the Mullins-Sekerka instability of a solidification front in an undercooled melt. Nonlinear evolution of this instability is simulated numerically and is shown to exhibit a structure similar to the columnar one.
12:45 PM - A14.5
Substrate Influence on the Appearance of Segregation in Laser-crystallized Polycrystalline SiGe Thin Films.
Moshe Weizman 1 , Norbert Nickel 1 , Ina Sieber 1 , Baojie Yan 2
1 , Hahn-Meitner-Institut, Berlin Germany, 2 , United Solar Ovonic Corporation, Troy, Michigan, United States
Show AbstractA15: Controlled Crystallization
Session Chairs
Wednesday PM, April 19, 2006
Room 3002 (Moscone West)
2:30 PM - **A15.1
Low-temperature Silicon Epitaxy and its Breakdown to Amorphous Silicon.
Charles Teplin 1 , Eugene Iwaniczko 1 , Qi Wang 1 , Kim Jones 1 , Robert Reedy 1 , Bobby To 1 , Dean Levi 1 , Helio Moutinho 1 , Howard Branz 1
1 , National Renewable Energy Laboratory, Golden, Colorado, United States
Show AbstractLow-temperature and economical silicon epitaxy could be useful in microelectronics and in achieving large-area, crystal silicon films. The first low-T epitaxy was achieved using molecular beam epitaxy (MBE) at temperatures as low as 100°C, but it was found that epitaxy was limited to a characteristic thickness, hepi, after which epitaxy broke down [D.J. Eaglesham et al., Phys. Rev. Lett. 65, 1227 (1990)]. Recent efforts have focused on using more economical growth techniques for growing epitaxial films at temperatures compatible with glass (<600°C). This research has led to important advances, including growth of thick epitaxial films at higher base pressures than is possible using MBE [A. Straub et al., J. Crystal Growth 268, 41 (2004)]. However, low-temperature epitaxy has yet to be achieved at high rates using any technique that has been demonstrated to be scalable to large areas. Further, the physics determining hepi and the mechanism of the subsequent breakdown is not yet well-understood.Our work provides new insights into the mechanisms that affect epitaxial growth and breakdown. We have grown 400 nm thick epitaxial films at 380°C using the scalable technique of hot-wire chemical vapor deposition, with a Ta filament. The dependence of hepi on substrate temperature and growth rate agree generally with the results using other techniques: hepi increases with increasing temperatures and lower growth rates. Surface and cross-sectional electron microscopy, as well as real-time spectroscopic ellipsometry, reveal that in all of our films, breakdown proceeds via nucleation of amorphous silicon cones with spherical caps; these cones expand isotropically and take over the film. Defect and impurity studies suggest that interface contamination cannot explain the failure of epitaxy. Atomic force microscopy shows that even at thicknesses greater than hepi, the surface roughness in epitaxial regions remains small. Measurements of the film morphology and of defect nucleation statistics will be discussed in relation to the existing theoretical pictures for epitaxial growth and breakdown.We acknowledge the U.S. Department of Energy for financial support under Contract DE-AC36-99GO10337.
3:00 PM - A15.2
Growth of Orientation-Controlled Long and Narrow Si Grains.
Tomoya Kato 1 , Yukio Taniguchi 1 , Kazufumi Azuma 1 , Masakiyo Matsumura 1
1 Research Dept.2, Advanced LCD Technologies Development Center Co.,Ltd., Yokohama Japan
Show AbstractINTRODUCTIONHigh-performance thin film transistors (TFTs) for next-generation ‘system displays’ are typically fabri-cated using lateral growth methods for Si grains. The most promising method appears to be phase-modulated excimer-laser annealing (PMELA). The grains in the channel area of PMELA-grown poly-Si films are long and narrow with a roughly rectangular-shape. Since carriers do not have to cross grain boundaries, high mo-bility can be achieved. The channel width needs to be broad in comparison with the grain widths so that many grains exist within a channel area resulting in suppres-sion of TFT performance fluctuations caused by crystal orientation fluctuations of the grains.In this paper, we report lateral growth characteristics for various Si layer thicknesses, and demonstrate arrays of long and narrow Si grains grown by a single-shot PMELA. In particular, we demonstrate for the first time that orientation of grains can be controlled when the Si films are very thin, which is crucial for TFTs.Experimental setupWe used a XeCl excimer laser and an optical system that included a phase modulator, illumination optics and projection lens. The sample had a stacked structure consisting of a 320 nm-thick capping SiOx layer and an a-Si layer on a thermally oxidized Si wafer. A periodic V-shaped one-dimensional 100% modulated light in-tensity distribution was directed onto the sample surface. The microstructure of the crystallized Si film was ana-lyzed using electron backscatter diffraction patterns (EBSP).Morphology of Si filmsThe grain length was 7~8 mm for a 30~100 nm-thick Si layer. For a 100 nm-thick Si layer, the average grain width increased with increasing distance from the start-ing point of growth and reached 0.70 mm. On the other hand, for a Si layer of under 50 nm thickness, the aver-age grain width saturated at a >1.5 mm position from the starting point and reached 0.32, 0.22 and 0.18 mm for 50, 40 and 30 nm-thick Si layers, respectively.Orientation of Si filmsFor Si layer of thickness under 50 nm, the orientation along the growth direction changed from {100} to {110} for a thinner Si layer. In particular, the {110} orientation predominated for the 30 nm-thick Si layer. The orientation of the normal and width directions swung from {001} to {112} and from {110} to {111}, respectively, rotating around the <110> axis which was the growth direction, for a 30 nm-thick Si layer.CONCLUSIONSAn array of highly oriented striped grains suitable for use in next-generation high-performance TFTs can be grown by single-shot PMELA.The crystal orientation was {110} along the growth direction and within {001} ~ {112} along the normal direction for a 30 nm-thick Si layer.These results indicate that (1) the TFT channel width can be reduced without increased fluctuations by mak-ing the Si layer thinner, and (2) the width is drastically reduced when the Si layer is less than 30 nm in thick-ness due to the aligned orientation.
3:15 PM - A15.3
Periodic alignment of Silicon Dot Fabricated by Linearly Polarized Nd:YAG Pulse Laser.
Kensuke Nishioka 1 , Susumu Horita 1
1 School of Materials Science, Japan Advanced Institute of Science and Technology, Nomi, Ishikawa, Japan
Show AbstractThis paper describes a novel technique for fabrication of Si dot periodically aligned by only irradiation of linearly polarized Nd:YAG pulse laser. Interference between a scattered surface wave and the incident beam leads to the spatial periodicity of beam energy density distribution on the surface of the irradiated samples. The periodic spacing Λ of the energy density distribution on the surface depends on the wavelength λ and the incident angle θ of the laser beam, and is formulated from Rayleigh’s diffraction conditions as λ/(1±sinθ) for p-polarized beam. This periodic energy density distribution induces the periodic temperature distribution on the irradiated surface. So, we applied the spontaneously induced periodic temperature distribution to fabrication of Si dots.A 10-nm-thick amorphous Si (a-Si) film was deposited on a thermal oxidized Si substrate with the 30-nm-thick SiO2 film at 350oC by electron beam evaporation and was irradiated by linearly polarized Nd:YAG pulse laser (λ: 532 nm, Repetition frequency: 10 Hz, Pulse width: 6-7 ns) at 250oC. The laser energy density and number of laser pulse were 60 mJ/cm2 and 1000, respectively.The a-Si film was melted by laser beam, and then, the molten thin Si film was split and condensed due to its surface tensile according to the periodic temperature distribution. Thus, the polycrystalline Si fine lines with the width of 200 nm were formed periodically on the SiO2 film, and the period was about 532 nm. The direction of the Si fine line was perpendicular to the electric field vector of the incident beam. After the first irradiation, the sample was rotated by 90o, and the laser beam was irradiated under the same condition as the first one. By the second irradiation, the periodic temperature distribution was generated on the Si fine lines. Then, the lines were split off and condensed according to the periodic temperature distribution. As a result, the Si dots periodically aligned were fabricated. We evaluated the Si dots by scanning electron microscopy and atomic force microscopy. It was found that the shape of the Si dots was hemisphere, and the diameter, height and period were 240 nm, 90 nm and 532 nm, respectively. Moreover, they were uniform. We successfully fabricated the submicron Si dots periodically aligned, using the spatially periodic temperature distribution induced by laser irradiation on the surface of samples. We can control the period of Si dots array by changing the wavelength and incident angle of the laser beam.The authors acknowledge the supports of The Foundation of Ando Laboratory and The Konica Minolta Imaging Science Foundation.
3:30 PM - A15.4
The Investigation of High Performance TFT by Thin Beam Directional X’stallization Method.
Chihwei Chao 1 , Chia-tien Peng 1 , CW Cheng 1 , Bernd Burfeindt 2
1 LTPS, AU Optronics Corporation, Hsinchu Taiwan, 2 , TCZ GmbH, San Diego, California, United States
Show AbstractA method, TDX (thin beam directional X’stallization), was proposed to form directional laterally grown poly-si film. Without any mask, hundreds micrometer-long poly-Si grain could be formed by TDX method. The influence of film thickness and laser energy parameter (energy and scan pitch) on the properties of poly-Si film has been investigated in this paper. The grain size and roughness of TDX poly-Si films that were observed by SEM and AFM changed with incident laser energy and laser scan pitch. According to AFM results, the roughness of TDX poly-Si films more flat than that of ELA poly-Si films is due to the lateral growth. The growth mechanism of TDX method was discussed in this paper. The high performance of n and p type TFT with optimum TDX laser condition has been made on the glass substrate. The carrier mobility of n and p type TFT is 360 and 160 cm2/V-S.
3:45 PM - A15.5
Solid-Phase Crystallization of Hydrogenated Amorphous Silicon for Thin Film Si Photovoltaics.
Paul Stradins 1 , David Young 1 , Yanfa Yan 1 , Dan Williamson 2 , Charles Teplin 1 , Yueqin Xu 1 , Eugene Iwaniczko 1 , Robert Reedy 1 , A. Mahan 1 , Howard Branz 1 , Qi Wang 1
1 , National Renewable Energy Laboratory, Golden, Colorado, United States, 2 , Colorado School of Mines, Golden, Colorado, United States
Show AbstractTwo distinct modes of hydrogenated amorphous silicon crystallization are studied: random crystallization in which crystallites nucleate and grow everywhere in the bulk, and solid-phase epitaxy (SPE) in which a near-planar growth front nucleates from an underlying crystalline template. By avoiding the nucleation step, SPE allows separation of the crystal growth rate from the homogeneous nucleation rate. We investigate both hot-wire (HW) and plasma-enhanced (PE) chemical vapor deposition hydrogenated amorphous silicon (a-Si:H) with a goal of developing improved thin-film Si solar cells on inexpensive substrates. We find, surprisingly, that the initial SPE growth rate on (100) c Si is far slower in thicker a-Si:H films -- whenever the initial thickness of the amorphous film is above 0.5 microns. Possible causes of this effect include difficulties in evolving H or expelling voids during thick-film crystallization, or an effect of high built-in stress at the interface of the thicker films. Film interface preparation plays a critical role in SPE both on c-Si and foreign lattice-matched substrates. The random crystallization time in HW a Si:H changes very little with deposition rate, void density, and hydrogen content. On the other hand, it is increased by 4x in PE a-Si:H as compared with HW [1], despite nearly identical residual H content and void density. Therefore, there is another, yet to be identified structural property of a-Si:H that strongly influences both the nucleation and growth rates. In-situ optical monitoring allows us to differentiate SPE and random crystallization, detect their rates [2] and probe these issues. We complement this technique with TEM, X-ray diffraction, small-angle x-ray scattering (SAXS), Raman spectroscopy, secondary ion mass spectrometry (SIMS), and in-situ monitoring of SPE by real-time surface reflection during the crystallization of a thickness-wedged a-Si:H film. Supported by DOE contract #DE-AC36-99G010337. 1.D. Young, P. Stradins, H. M. Branz, M. Page and Q. Wang, Mat. Res. Soc. Proc. 862 (2005) 233. 2.P. Stradins, D. Young, H. M. Branz, M. Page and Q. Wang, Mat. Res. Soc. Proc. 862 (2005) 227.
A16/L9: Joint Session: AMOLED Backplane Electronics
Session Chairs
Wednesday PM, April 19, 2006
Room 3002 (Moscone West)
4:30 PM - **A16.1/L9.1
Backplane Requirements for Active Matrix Organic Light Emitting Diode Displays.
Arokia Nathan 1
1 Electrical and Computer Engineering, University of Waterloo, Waterloo, Ontario, Canada
Show AbstractOrganic light emitting diode (OLED) displays are a serious competitor to liquid crystal displays in view of their superior picture quality, higher contrast, faster on/off response, thinner profile, and high power efficiency. While the passive matrix OLED display structure is simple, its pulse-driven nature leads to high power consumption and low lifetime, restricting its use to small, low-resolution displays. For large size and/or high-resolution applications, an active matrix OLED (AMOLED) addressing scheme is vital, requiring use of a thin-film transistor (TFT) pixel circuit to regulate the OLED current. The active matrix backplane can be made with amorphous silicon (a-Si), polysilicon, or organic technology, all of which have significant drain-current degradation due to threshold voltage (VT) shift or mismatch problems, which cause temporal or spatial variations in the OLED brightness. In addition, the efficiency of the OLED itself degrades over time. Despite material weaknesses, considerable progress is being made in designing AMOLED displays with stable and uniform brightness using circuit solutions to compensate for low lifetime and differential aging. Indeed the design of AMOLED pixel circuits, particularly in low-mobility TFT technologies such as a-Si, is challenging due to the stringent requirements of timing, current matching, and low voltage operation. However, circuit solutions, while necessary, are not sufficient, which raises the question of whether process improvements can be made to enhance TFT performance. This paper will review pertinent material requirements of AMOLED backplanes along with design considerations that address pixel architecture, contact resistance, and more importantly, the VT-stability and gate overdrive. In particular, can conventional PECVD be deployed for high mobility and high VT-stability nano-crystalline silicon TFTs, including thin film CMOS for eventual system-on-panel integration?
5:00 PM - A16.2/L9.2
Flexible AMOLED Backplane Based on Excimer Laser Annealed Poly-Si TFT on Metal Foil.
JengPing Lu 1 , Yu Wang 1 , Chinwen Shih 1 , Yunan Pei 1 , Jackson Ho 1 , Robert Street 1 , Keith Tognoni 2 , Bob Anderson 2 , Dave Huffman 2 , Anna Chwang 3 , Richard Hewitt 3 , Ken Urbanik 3 , Michael Hack 3 , Julie Brown 3 , Teresa Ramos 4 , Lorenza Moro 4 , Nicole Rutherford 4
1 , Palo Alto Research Center, Palo Alto , California, United States, 2 , L3 Communications, Alpharetta, Georgia, United States, 3 , Universal Display Corporation, Ewing, New Jersey, United States, 4 , Vitex Systems, San Jose, California, United States
Show AbstractTFT backplanes on flexible substrates have become the current research trend because of the recent surge of interest in flexible displays. Based on our previously developed, conventional Exicmer Laser Annealed poly-Si TFT on glass process, which has shown excellent performance and has been successfully used to implement many demanding applications, such as active pixel flat panel imagers, PARC has developed a poly-Si TFT on thin metal foil technology. By carefully modifying the process steps and conditions, we have successfully demonstrated the TFTs on flexible substrates with similar performance as TFTs on glass such as high mobility (78 cm2/Vs for p-channel), low off-state leakage current (0.1pA/μm), and sharp turn on (S=0.38 V/decade). Targeting the application of driving low power consumption phosphorescent OLED displays, PARC, UDC, Vitex, and L3 have jointly demonstrated a prototype, thin film encapsulated AMOLED display on stainless steel foil with a 100 dpi resolution. In this talk, we will report the current status of the development of poly-Si TFT on thin metal foil. TFT performance, uniformity, stability, as well as pixel circuit will be discussed.
5:15 PM - A16.3/L9.3
a-Si:H 2-TFT AMOLED Pixel Circuits on Stainless Steel Foils.
Alex Kattamis 1 , I-Chun Cheng 1 , Sigurd Wagner 1 , Yongtaek Hong 2
1 Department of Electrical Engineering and The Princeton Institute for the Science and Technology of Materials (PRISM), Princeton University, Princeton, New Jersey, United States, 2 Display Science and Technology Center, Eastman Kodak Company, Rochester, New York, United States
Show Abstract5:30 PM - **A16.4/L9.4
Flexible Substrate All Organic Active Matrix Displays.
Tom Jackson 1
1 Electrical Engineering, Penn State University, University Park, Pennsylvania, United States
Show AbstractThere is great interest in active matrix organic light-emitting diode (AMOLEDs) displays because of their potential for thin, light weight, low-power, and possibly also flexible displays. AMOLED displays have been demonstrated on stainless steel and polymer substrates [1, 2], using poly-silicon or amorphous silicon TFTs as the active elements. Organic thin film transistors (OTFTs) are of interest for pixel control devices because of the possibility of reduced processing temperature and cost.We have fabricated 48 x 48 pixel pentacene OTFT driven AMOLED displays on flexible polyethylene terephthalate (PET) substrates. The displays use a simple two transistor per pixel drive scheme with a pixel pitch of 500 μm and an aperture ratio of 0.52. The drive and select transistors have a W/L ratio of 10 (200 μm/20 μm) and 1 (20 μm/20 μm), respectively, and the storage capacitor is 1.2pF. For display fabrication, 2.5 × 2.5 inch, 125 μm thick PET substrates are laminated to glass plates using a removable pressure-sensitive silicone gel to maintain a flat PET surface. The pentacene OTFTs are patterned using photosensitive polyvinyl alcohol and isolated from the OLEDs by a parylene layer. After display fabrication the PET display substrate is easily removed from the glass carrier. Although the pixel and line defect density is significant in our simple displays, flexible OTFT-AMOLED display function is demonstrated. This indicates that OTFT backplanes are viable candidates for active-matrix OLED flexible displays. 1. Afentakis, T., et al. “Poly-silicon TFT AM-OLED on thin flexible metal substrates,” Poly-Silicon Thin Film Transistor Technology and Applications in Displays and Other Novel Technology Areas. SPIE 5004-30 (2003).2. Nichols, J.A., et al. “a-Si:H TFT phosphorescent OLED active Matrix Pixels fabricated on polymeric substrates,” 2004 Device Research Conference Technical Digest, pp. 59-60 (2004).
Symposium Organizers
Harry A. Atwater California Institute of Technology
Virginia Chu INESC Microsistemas e Nanotecnologias
Sigurd Wagner Princeton University
Kenji Yamamoto Kaneka Corporation
Hsiao-Wen Zan National Chiao Tung University
A17: Sensors
Session Chairs
Thursday AM, April 20, 2006
Room 3002 (Moscone West)
9:30 AM - **A17.1
Amorphous Silicon as an Active Material in Optical Resonators.
Dennis Hohlfeld 1 , Hans Zappe 1
1 Department of Microsystems Engineering, University of Freiburg, Freiburg Germany
Show AbstractThe application of amorphous silicon as a thermo-optically active material in MEMS-based, tunable optical filters is presented. The thin film interference filter employs a solid-state silicon cavity and silicon-based distributed Bragg reflectors (DBR). Tuning is achieved by varying the optical resonator thickness through thermal modulation. The filter is configured as a membrane and operates up to a temperature of 450°C. Such a filter is essential for monitoring and reconfiguration of optical data communication networks and is also applicable for chemical and gas sensing.The transmission wavelength of an etalon may be tuned by varying the resonator thickness. This tuning is usually accomplished by mechanically varying the distance between the mirrors using electrostatic, thermal or magnetic actuation. Varying the temperature of the membrane changes the refractive index of the cavity and accordingly the optical resonator thickness of the filter. In this work thin-film metal heaters are deposited on top of freestanding filter membranes. This configuration increases thermal isolation and improves heating efficiency by a factor of 6 when compared to similar filters deposited on fused silica substrates. In the infrared spectral region, amorphous silicon shows a large thermo-optic coefficient combined with low absorptance and therefore is especially well suited as a material for the thermally tunable cavity layer. By using amorphous silicon with silicon dioxide or silicon nitride for multilayer film stacks, DBRs of nearly 100 % reflectance at a center wavelength of 1550 nm were fabricated. Plasma enhanced chemical vapor deposition (PECVD) was used for deposition of the optical layers onto silicon substrates and platinum resistors were structured on top of the filter layers. Control of mechanical stress is of crucial importance for fabrication of robust freestanding membranes. To reduce stress within the membrane to a minimum, we used silicon nitride and amorphous silicon for deposition of DBR layers. Since the fabrication process of silicon nitride allows the deposition of layers under defined tensile stress, this procedure was used to efficiently compensate the compressive stress within the adjacent amorphous silicon layers. For filter characterization, light was coupled perpendicularly through the membrane. A minimum filter linewidth of 0.282 nm with insertion loss of 5.5 dB were achieved for the optimized filter structure. The thermo-optic properties of amorphous silicon were determined by using a new model for the tuning behavior of optical resonators.
10:00 AM - A17.2
Band Gap Engineering and Electrical Field Tailoring for Voltage Controlled Spectral Sensitivity.
Manuela Vieira 1 , Alessandro Fantoni 1 , Miguel Fernandes 1 , Paula Louro 1 , Reinhard Schwarz 1 , Guilherme Lavareda 2 3 , Carlos N Carvalho 2 3
1 DEETC, ISEL, Lisbon Portugal, 2 C1, IST, Lisbon Portugal, 3 DCM, FCT-UNL, Lisbon Portugal
Show AbstractMultilayered laser scanned photodiode (LSP) color sensitive imagers are analyzed. The sensors provide a voltage controlled spectral sensitivity in the visible range, allowing a complete color analysis to be performed with a single two terminal detector element and an optically addressed readout. The imager consists of a thin wide band gap p-i-n a-SiC:H spectrally sensitive element deposited on top of a large area a-SiC:H(-p)/a-Si:H(-i’)/a-SiC:H(-n) image sensor, presenting voltage controlled color selectivity properties. As only two transparent contacts are present, the electronic front-end is composed of a single current to voltage converter and signal amplifier. The image to acquire is optically mapped onto the front photodiode and a low-power light spot scans the device by the opposite side. The photocurrent generated by the moving spot is recorded as the electronic image signal, and its magnitude depends on the local illumination conditions, like the wavelength and intensity without the need of any additional signal processing. To improve the red-green-blue separation the approach in this paper applies a supplemental front a-SiC:H (-i) layer in the bottom diode resulting in a pinpii’n layer sequence. This approach leads to regionally different collection parameters resulting in multispectral photodiodes, coding for red (R), blue (B), and two green (G) components. Spectral and light-to-dark sensitivity data obtained from the detectors will be reported. Results show that band gap engineering and electrical field tailoring allows a voltage controlled shift of the collection regions. For each RGB wavelength, color rejection is achieved at a read-out voltage that cancels the forward or reverse self-bias induced by the impinging photons across the a-Si:H back absorber. To achieve flat band condition across the device, positive bias is needed under blue irradiation and moderate reverse bias under green. The threshold voltage between green and red sensitivity depends on the thickness of the supplemental a-SiC:H (-i) layer and corresponds to the complete confinement of the absorbed green photons across the pinpi sequence. As the a-SiC:H thickness increases the bottom diode reverse biasing caused by the light absorption in the front one will decrease as back the absorption in the back one increase, leading to a shift of the threshold voltage to lower reverse bias.The various design parameters and the optical readout process tradeoffs are discussed and supported by a 2D numerical simulation. A self-bias model will be proposed to explain the color selectivity properties.
10:15 AM - A17.3
Image Sensors Based on Thin-film on CMOS Technology: Additional Leakage Current due to Vertical Integration of the a-Si:H Diodes.
Clement Miazza 1 , Nicolas Wyrsch 1 , Gregory Choong 1 , Sylvain Dunand 1 , Christophe Ballif 1 , Arvind Shah 1 , Nicolas Blanc 2 , Felix Lustenberger 2 , Rolf Kaufmann 2 , Danielle Moraes 3 , Mathieu Despeisse 3 , Pierre Jarron 3
1 Insititute of Microtechnology, University of Neuchatel, Neuchatel Switzerland, 2 Photonic Division, CSEM SA, Zurich Switzerland, 3 Experimental Physic Divison, CERN SA, Geneva Switzerland
Show AbstractImage sensors based on thin-film on CMOS technology have been developed, showing very high geometrical fill factors (FF = 92 %) and increased sensitivity (S > 60 V/(µJ/cm2)). In this approach, amorphous silicon (a-Si:H) detectors are vertically integrated directly on top of a CMOS readout chip so as to form monolithic image sensors. In order to reduce as far as possible the dark current density (Jdark) of the TFC sensors, we have focused on analyzing and understanding the behavior of Jdark in this type of detectors.In the case of large area a-Si:H n-i-p photodiodes (several mm2) with an n-i-p structure, the lower limit for Jdark is given by the thermally-generated charge in the intrinsic layer. The density of thermally-generated current depends on various factors: the deep defect density, the mobility gap of the semiconductor material and the width of the intrinsic layer.However, in the case of vertical integration, additional leakage currents which lead to an increase in Jdark are observed. Edge effects at the periphery and in the corner of the pixel, linked to the presence of a passivation layer, are found to be responsible for this increase. In fact, thanks to a simple model which the authors use to analyse the measured values of Jdark as plotted in function of the detector size, a corner effect is shown to be present. Furthermore, the presence of this corner effect, as well as the presence of a peripheral effect at the pixel side is clearly demonstrated by electron beam induced current (EBIC) analysis of a TFC sensor.In order to minimise these effects, responsible for the observed increase of Jdark, different approaches have been evaluated. This lead to the choice of a new and adapted solution which combines the use of a metal-i-p a-Si:H diode configuration and that of an unpassivated CMOS chip. With such a TFC sensor the best performances are attained, and in particular a value as low as 12 pA/cm2 has been measured after thermal annealing, on 40x40 μm2 pixels polarised at -1 V. The results of the overall characterization of these TFC sensors in terms of noise, dynamic range, linearity, light-induced degradation and sensitivity are also given.
10:30 AM - A17.4
Role of the Oxide Layer on the Performances of a-Si:H MIS Structures Applied to PDS Fabrication
Hugo Aguas 1 , Luis Pereira 1 , Daniel Costa 1 , Leandro Raniero 1 , Elvira Fortunato 1 , Rodrigo Martins 1
1 DCM, FCT-UNL, Caparica Portugal
Show Abstract10:45 AM - A17.5
Un-cooled Micro-bolometer with Sandwiched Thermo-sensing Layer Based on Ge Films Deposited by Plasma
Andrey Kosarev 1 , Mario Moreno 1 , Alfonso Torres 1 , Roberto Ambrosio 1
1 Electronics, Inst.Nat.for Astrophysics, Optics and Electronics, Puebla, Puebla, Mexico
Show AbstractFor improving the performance of a micro-bolometer, it is necessary to have high temperature coefficient of resistance, TCR (α) and low output resistance (Rout≤106 Ohm), which is compatible with the commonly used driver circuits. In planar micro-bolometer structures this is solved by doping the thermo-sensing layer [1], solution, which results in reducing the activation energy and, consequently, TCR and responsivity. To avoid this effect, an elegant solution is the use of structures with the thermo-sensing layer sandwiched between the contact electrodes, which reduces significantly Rout without loss of TCR. However, up to date, there have been only few publications on this approach e.g.[2]. In this work we report on fabrication and characterization of an un-cooled micro-bolometer, which uses as thermo-sensing an a-Ge:H film sandwiched between metal electrodes. The micro-bolometer structures were fabricated on a Si wafer. Firstly a SiN bridge was formed by surface micromachining technique for obtaining sufficient thermo-isolation. Then, the bottom electrode (Ti) was deposited and patterned. Thermo-sensing Ge:H layer was deposited on the bottom electrode and coated by top electrode (Ti). A SiN layer optimized for improvement IR absorption around λ=10μm and protection was deposited on the top electrode. SiN and Ge:H layers were deposited by low frequency(LF) PE CVD at a substrate temperature Ts=300 C. Ti electrodes were deposited by electron beam evaporation. Parameters of the depositions process were optimized in order to provide better performance of both materials. The active area of the cell is S=70x66 μm2. Bridge height is 2.5 μm. Resistance of the cell is Rcell=1.4x105 Ohm. The activation energy of the thermo-sensing film measured in a test structure is Ea=0.34eV providing TCR= 0.043 K-1. Current responsivity (ΡI) was determined from current-voltage characteristics measured in dark and under IR illumination from a “Globar” source, which provided intensity I=3.1x10-2 W/cm2 on the sample surface. At DC bias voltage U=3 V the value of ΡI= 1,43 A/W was obtained. Noise measurements, thermal and response characterization are ongoing and will be discussed in comparison with reported data in literature.The authors acknowledge the support of this research by CONACyT project No.42367 (CIAM-2002).References:1. A.J.Sillaios, T.R.Schimert, R.W.Gooch, W.L.McCardel, B.A.Ritchey, J.H.Tregilgas. Mat.Res.Soc.Symp.Proc. 609 (2000) A14.4.1.2. A.H.Z.Ahmed, R.NTait., J.Vac.Sci.Technol.A 22(3), May/June 2004. p
A18: Flexible Electronics
Session Chairs
Thursday PM, April 20, 2006
Room 3002 (Moscone West)
11:30 AM - A18.1
Substrate Effects on TFT Performance
I-Chun Cheng 1 , Alex Kattamis 1 , Sigurd Wagner 1 , Yongtaek Hong 2
1 Electrical Engineering, Princeton University, Princeton, New Jersey, United States, 2 Research and Developement, Eastman Kodak Company, Rochester, New York, United States
Show AbstractPlastic or metal foil substrates are attracting great attention for next-generation large-area electronics, because they can be rugged, lightweight, flexible or even deformable. For transferring the fabrication process from conventional rigid glass substrate to these novel flexible foil substrates, preservation of the device performance becomes an important issue. Here we use amorphous silicon thin-film transistors (a-Si:H TFTs) as an example to study the dependence of TFT performance on substrate types. It has been demonstrated that TFTs made on plastic substrates can have electrical performance comparable to their counterparts on glass substrates [1,2]. Both, I-V characteristics and stability depend mainly on the process temperature, which in the case of plastic substrates is limited by the glass transition temperature. In contrast to plastic, metal foil allows high process temperatures. We chose ~ 75-μm thick stainless steel foil as the substrate for its dimensional stability and resistance to process chemicals. Prior to device fabrication, the rough steel foil substrates are smoothed with spin-on-glass and passivated with a PECVD deposited SiOx or SiNx layer to provide electrical isolation for the device. Sufficiently thick planarization/passivation is required to suppress substrate leakage current and reduce capacitive coupling during TFT circuit operation. The TFTs are fabricated in a staggered, bottom-gate structure at a maximum process temperature of ~ 280°C. Initially, dry etching was non-uniform when parts of the steel substrate were directly exposed to the plasma. This caused large variations in the threshold voltage across the sample. The non-uniformity is minimized by completely sealing the front surface of the steel substrate with the passivation layer during any plasma process. The TFT performance on steel is comparable to that of TFTs made on glass. Effective electron mobilities range from 0.4 to 0.7 cm2V-1s-1 for channel lengths from 5 μm to 40 μm. The low mobility at short channel length indicates that it is limited by contact resistance. TFT stability is investigated under constant-voltage and constant-current bias stressing. Similarly to TFTs on glass, the degree of threshold voltage shift and mobility reduction increases with bias stressing time, and a more pronounced shift is observed when the device is under constant-current than constant-voltage bias stress. At the symposium we will provide a summary of our work on the fabrication and properties of a-Si:H TFTs on steel.[1] H. Gleskova, S. Wagner, V. Gasparik, and P. Kovac, “150°C amorphous silicon thin-film transistor technology for polyimide substrates,” J. Electrochem. Soc., 148, pp. G370-G374, 2001[2] K. Long, I-C. Cheng, A. Kattamis, H. Gleskova, S. Wagner and J. C. Sturm, “Stability of amorphous-silicon thin-film transistors deposited on clear plastic substrates at 250°C to 280°C,” submitted to Electron Device Lett.
11:45 AM - A18.2
Hot-wire CVD a-Si:H TFT on Plastic Substrates
Farhad Taghibakhsh 1 , Karim Karim 1
1 Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada
Show AbstractRugged substrates, especially mechanically flexible plastic, are an excellent candidate for future display technology; they are inexpensive, lightweight and durable. On-going research focuses on reducing the TFT fabrication temperature for deposition on plastic substrates where current state-of-the-art a-Si:H TFT technology uses PECVD deposition at 100°C or less [1]. However, hot-wire CVD (HWCVD) a-Si:H TFTs have been shown to be superior to traditional PECVD fabricated TFTs with excellent stability, while having high switching current ratios and low leakage currents. High stability devices are particularly important for emerging applications of a-Si:H technology that require high TFT duty cycles such as organic light emitting diode (OLED) displays. In addition, HWCVD is a high deposition rate technique (approximately one order of magnitude higher than PECVD), which is easily scalable to large area substrates due to its simple and inexpensive setup [2]. In this research, we present the first HWCVD a-Si:H TFTs fabricated on thin polyamide films. Here, the primary challenge lies in minimizing thermal damage to the plastic substrate due to the high thermal radiation from the hot filament (> 1500°C). A top-gate TFT configuration was employed with a 100 nm intrinsic a-Si:H film deposited by HWCVD of a graphite filament. PECVD was used for the n+ a-Si:H source and drain contact layers as well as for the a-SiNx:H gate insulator. Molybdenum was used for the gate while aluminium was used for the source and drain metallization. Electrical measurements show a low leakage current of less than 0.3 pA with a switching current ratio of > 107, for a device aspect ratio of 10. Threshold voltage and field effect mobility were measured to be 5.8 V and 0.35 cm2/V.s respectively. Here, we will present details of the fabrication process, device characterization and results of the gate bias stress analysis. References:1. H.J. Lee, A. Sazonov, and A. Nathan, “Low Temperature Gate Dielectrics for Silicon-on-Plastic,” Abstract, 12th Canadian Semiconductor Technology Conference, Ottawa, August 15-19, 2005.2. B. Stannowski, R.E.I. Schropp, R.B. Wehrspohn, M.J. Powell, "Amorphous silicon Thin Film Transistors deposited by VHG-PECVD and Hot-Wire CVD", J. Non-Cryst. Solids 299-302 (2002) 1340-1344.
12:00 PM - A18.3
Self-aligned Thin Film Transistor Fabrication with Ultra Low Temperature Polycrystalline Silicon Process on a benzocyclobutene Planarized Stainless Steel Foil Substrate.
Jaehyun Moon 1 , Choong-Heui Chung 1 , Yong-Hae Kim 1 , Dong-Jin Park 1 , Sun Jin Yun 1 , Jung Wook Lim 1 , Jin Ho Lee 1
1 Basic Research Lab., Electronics and Telecommunications Research Institute, Daejeon Korea (the Republic of)
Show Abstract12:15 PM - A18.4
Mechanical Design Of A-Si TFT’s Fabrication On High-Temperature Clear Plastic Substrate
Ke Long 1 , I-Chun Cheng 1 , Alex Kattamis 1 , Helena Gleskova 1 , Sigurd Wagner 1 , James Sturm 1
1 ELectrical Engineering, Princeton University, Princeton, New Jersey, United States
Show AbstractHigh quality amorphous-silicon (a-Si) thin-film transistors (TFT’s) fabrication on clear plastic substrates is essential for flexible displays. A high process temperature (~300°C) is required for optimum TFT properties and minimum electron trapping in the gate dielectric for long-term stability. However, clear plastics with a glass transition temperature (Tg) in excess of 300°C have a coefficient of thermal expansion (CTE) larger than that of the silicon nitride and a-Si in the TFT’s. This can lead to cracking of the device films that limits the process temperature to well below that of the plastic glass transition temperature. In this work, the mechanical interaction of the TFT stack and the plastic substrate is modeled to develop design guidelines to avoid cracking. This methodology was then used to successfully fabricated a-Si TFT’s on novel clear plastic substrates with a maximum process temperature of up to 280°C. The TFT’s made at high temperatures have higher mobility and lower leakage current, and high stability compared to TFT’s made on conventional low-Tg clear plastic substrates. The thermal expansion mismatch leads to a strain in the device films and a stress which must be supported by an interfacial force. Experiments show that the critical factor leading to failure is not a strain level, but rather a critical interfacial force, above which the device film cracks. Experimentally, this is shown by the fact that the maximum thickness before cracking of a PECVD nitride buffer on the plastic decreases as the temperature increases, so that the critical product of CTE mismatch times thickness is approximately constant, while the strain in the nitride is independent of thickness. The interfacial force required by the device stack can be reduced by the built-in strain of the PECVD nitride buffer(in-situ tension), the thickness of the device stack, and the CTE of the substrate. The PECVD deposition power can be used to control the built-in strain in the SiNx film, to compensate for the thermal mismatch strain. From these basic considerations, for a given process temperature, to avoid cracking, quantitative guidelines are developed for the maximum allowable buffer nitride thickness (for fixed substrate thickness and CTE) and for the maximum substrate thickness and CTE for a given device stack requirement. These mechanical considerations were then applied in practice to successful TFT fabrication on novel clear plastic substrates supplied by Dupont Company with a Tg in excess of 300°C.
12:30 PM - A18.5
a-Si TFT Fabricated on PEN Plastic Substrate under External Stress.
TeChi Weng 1 , Jian-Shu Wu 1 , Jung-Fang Chang 1
1 , ITRI, Hsinchu Taiwan
Show Abstract12:45 PM - A18.6
Nano-crystalline Silicon Thin Film Transistors on PET Substrates Using a Hydrogenation-assisted Metal-induced Crystallization Technique.
Ashkan Behnam 1 , Saber Haji 1 , Farshid Karbasian 1 , Shams Mohajerzadeh 1 , Aida Ebrahimi 1 , Yaser Abdi 1 , Michael Robertson 2 , Craig Bennet 2
1 Electrical and Computer Engineering, University of Tehran, Tehran, Tehran, Iran (the Islamic Republic of), 2 Physics Department, Acadia University, Wolfville, Nova Scotia, Canada
Show AbstractA19: TFT Stability
Session Chairs
Thursday PM, April 20, 2006
Room 3002 (Moscone West)
2:30 PM - **A19.1
Mechanisms for Defect Creation and Removal in Hydrogenated and Deuterated Amorphous Silicon Studied using Thin Film Transistors.
Andew Flewitt 1 , Ralf Wehrspohn 2 , Shufan Lin 1 , Martin Powell 3 , William Milne 1
1 Engineering Department, Cambridge University, Cambridge United Kingdom, 2 Department of Physics, University of Paderborn, Paderborn Germany, 3 , 252, Valley Drive, Kendal United Kingdom
Show AbstractIt has been widely observed that thin film transistors (TFTs) incorporating an hydrogenated amorphous silicon (a-Si:H) channel exhibit a progressive shift in their threshold voltage with time upon application of a gate bias. This is attributed to the creation of metastable defects in the a-Si:H which can be removed by annealing the device at elevated temperatures with no bias applied to the gate, causing the threshold voltage to return to its original value. This threshold voltage shift is negligibly small over the lifetime of TFTs in active matrix liquid crystal displays, where the duty cycle is only ~10%. However, the duty cycle is much greater in the next generation of active matrix organic light emitting diode displays, and this threshold voltage shift is unacceptably large. There is, therefore, a renewed interest in understanding the microscopic mechanisms by which defect creation and removal occur.In this work, the defect creation and removal process has been investigated using both fully hydrogenated and fully deuterated amorphous silicon (a-Si:D) TFTs. In both cases, material was deposited by rf plasma enhanced chemical vapour deposition over a range of gas pressures to cover the α-γ transition. The variation in threshold voltage as a function of gate bias stressing time, and annealing time with no gate bias, was measured. Using the thermalisation energy concept, it is possible to quantitatively determine the distribution of energies required for defect creation and removal as well as the associated attempt-to-escape frequencies.In the case of defect creation, activation energies in the range of 0.93 to 0.98 eV are measured with an associated attempt-to-escape frequency of 109 Hz for both hydrogenated and deuterated devices. It is shown that a-Si:H and a-Si:D deposited at the same growth rate (rather than under the same nominal growth conditions) is structurally similar. When the activation energy is plotted as a function of amorphous silicon growth rate, both hydrogenated and deuterated devices lie on the same curve, within experimental errors. This insensitivity to the mass of the hydrogen atom suggests that defect creation is limited by the breaking of an Si-Si bond.The attempt-to-escape frequency for defect removal, however, is measured to be ~1013 Hz. A molecular dynamics simulation of hydrogen and deuterium in amorphous silicon reveals that this frequency correlates with the stretching mode of the Si-H or Si-D bond, which is (4.4±0.3)×1013 Hz and (5.7±0.3)×1013 Hz respectively. Furthermore, a difference in the activation energies for defect removal is then observed between a-Si:H and a-Si:D which mirrors the difference in the Si-H and Si-D bond energies. Therefore, it would appear that the rate limiting step for defect removal is the breaking of an Si-H bond.These results are discussed in context with molecular dynamics simulations of the stable configurations of two hydrogen atoms in close proximity in a-Si:H.
3:00 PM - A19.2
Defect States in Laser Crystallized Silicon Based TFTs Studied with Isothermal Charge Deep-level Transient Spectroscopy.
Vojtech Nadazdy 1 , Stefan Lanyi 1 , Rudolf Durny 2 , Vikas Rana 3 , Ryoichi Ishihara 3 , J. Wim Metselaar 3 , C.I.M. Beenaker 3
1 , Institute of Physics Slovak Academy of Sciences, Bratislava Slovakia, 2 Department of Physics, Slovak University of Technology, Bratislava Slovakia, 3 DIMES, Delft University of Technology, Delft Netherlands
Show Abstract3:15 PM - A19.3
Effect Of Light Illumination On Threshold Voltage And Field Effect Mobility Of Amorphous Silicon Thin Film Transistors.
Lihong (Heidi) Jiao 1 , Jingdong Deng 2 , C. R. Wronski 2 , T. N. Jackson 2
1 School of Engineering, Grand Valley State University, Grand Rapids, Michigan, United States, 2 Electrical Engineering, Pennsylvania State University, University Park, Pennsylvania, United States
Show AbstractSince the discovery of Staebler-Wronski effect in amorphous silicon materials, many studies have been carried out on light induced defect states in both a-Si:H thin film materials and corresponding cell structures (Schottky barrier and p-i-n solar cells)1,2. In this study the light induced defect state creation under 1 sun illumination is investigated through changes in threshold voltage, VT, and field effect mobility, μFET, in a-Si:H TFT structures. Threshold voltage changes can be due to either charge trapping or defect states creation near the interface of gate insulator and a-Si:H active layer. By incorporating high quality thermal silicon dioxide in the a-Si:H TFT, the charge trapping effects have been minimized and the changes in threshold voltage can be attributed to induced defect states in the a-Si:H. Since the field effect mobility is closely related to the states near the conduction band edge, the threshold voltage and field effect mobility can be used as a new probe to study the light induced defect creation in the a-Si:H. In this study, the information obtained about the nature of gap defect states from the analysis of changes in VT and μFET under light illumination is presented. The results obtained show that under 1 sun illumination there are very fast increases (within 30 seconds) in VT with subsequent slow increases, similar to those observed in the light induced photoconductivity changes in a-Si:H thin films3. The increase in VT under the initial light illumination are associated with the Fermi Level in a-Si:H active layers moving downward and causing band bending near the interface region of the gate insulator and a-Si:H active layer. Results are also presented on changes in VT under gate-bias stressing as well as on the degradation in the field effect mobility. Comparisons are then made between the results obtained under 1 sun light illumination and those under gate-bias stressing. The information obtained in this study about the nature of the gap defect states derived from characterization of the sub-threshold regimes is discussed.1.C. R. Wronski, J. M. Pearce, R. J. Koval, X. Niu, A. S. Ferlauto, J. Koh, and R. W. Collins, Mater. Res. Soc. Proc. 715, 459 (2002)2.H. Liu, Y. Lee, T. Jamali-beh, Z. Lu, R. Collins and C. R. Wronski, 25th IEEE Photovoltaic Specialists Conf., 1125 (1996)3.L. Jiao, H. Liu, S. Semoushikiana, Y. Lee and C.R. Wronski, 9th International Photovoltaic Science and Engineering Conf.641 (1996)
3:30 PM - A19.4
Post Deposition Ultraviolet Treatment of Silicon Nitride Dielectric: Modeling and Experiment
Vladimir Zubkov 1 , Mihaela Balsenau 1 , Li-Qun Xia 1 , Hichem M'Saad 1
1 , Applied Materials, Inc., Santa Clara, California, United States
Show Abstract The use of high tensile stress silicon nitride (SiN) has been shown capable to improve transistor performance in NMOS devices. The post-deposition ultraviolet (UV) treatment enhances the tensile stress of PECVD silicon nitride films. In this work we discuss the relation between UV-induced composition changes and tensile stress increase. UV treatment brings about excited states in SiN. Increase in tensile stress is achieved by breaking N-H and Si-H bonds in excited states with release of H atoms. Generation of radicals =N and ≡Si facilitates formation of new Si-N bonds. In order to make UV cure more effective it is desirable to understand bond dissociation in excited states. Ab initio methods were applied to determine absorption wavelengths in SiN in the range 200-300nm and effect of UV excitation on bond strengths. Time-dependent DFT applied to SiNxHy clusters of various sizes to yielded absorption bands of significant intensity in the range 200 – 240nm. Effect of excitations on bond strengths has been measured by increase in energies upon initial bond stretch in ground and excited states for chain and ring clusters. It turned out that an initial stretch of the N-H bond is noticeably alleviated in excited state contrary to the case of the Si-H bond. Thus, modeling indicates that H abstraction from N-H is easier than from Si-H upon UV treatment contrary to their relative bond strengths in the ground state. As for Si-N bonds, a more realistic ring cluster model indicates that these bond are only marginally weakened in excited states. Experimentally, the Si-H and N-H content in the nitride film was monitored using transmission FT-IR for different UV cure conditions. Within the measurement error, both Si-H and N-H content are decreasing at the same rate with increasing cure time. This apparent contradiction with ab initio results suggests the existence of additional reactions taking place in the nitride film during UV cure. The hydrogen released from the N-H bonds can potentially assist abstraction of hydrogen from Si-H bonds and formation of new Si-N bonds. Ab initio calculations indicate that H abstraction from Si-H bond by H atom proceeds much more easily than that from N-H bond.
3:45 PM - A19.5
The Effects of Crystal Filter on electrical Characteristics of Low Temperature Poly-Si Thin Films Transistor.
Min Sun Kim 1 , Seung-Ki Joo 1
1 School of Material Science and Engineering, Seoul National University, Seoul Korea (the Republic of)
Show AbstractWhen the amorphous silicon(a-Si) film is crystallized into poly-Si, the mobility can be increased up to more than 200 cm2/Vs. This enables to fabricate system on panels, OLED, and flexible substrate can be used for mobile display products when low temperature crystallization of a-Si is realized. Recently, metal-induced lateral crystallization(MILC) process has been introduced, by which the a-Si thin films can be crystallized at below 500°C. However, it has been reported that the leakage current of poly-Si TFTs using MILC is higher and mobility is lower than those of poly-Si TFTs prepared by laser annealing process. Therefore, we fabricated novel MILC-TFT to get superior electrical characteristics of poly-TFT using crystal filtering system. Control of the orientation of Si crystal grain enables us to control the silicide location, thus artificially remove contaminated the silicide from the device area. The a-Si thin film was then deposited by LPCVD. And then, the samples were defined by photolithography and dry etch process. The neck widths of crystal filter were 1, 2, 4, 6, and 8 μm and the neck length of crystal filter was fixed at 5 μm. After etching a-Si layer, 100Å Ni dots were formed by a lift-off process. The a-Si was crystallized to poly-Si by MILC. We compared the difference of MILC growth lengths between filtered and unfiltered area. The crystalline orientations of poly-Si grains before and after crystal-filtering were observed by electron back scattering diffraction. Also p-type poly-Si TFTs were fabricated as follows. The a-Si thin film was patterned by lithography process. The SiO2 gate oxide and MoW gate were formed by PECVD and sputtering, respectively. In order to define the source/drain junction, the samples were doped with B2H6. And the samples were annealed in hydrogen ambient at a temperature of 550 °C for 30 hrs.We observed differences of microstructures of poly-Si between non-crystal filtered area and crystal filtered area. It is easily noticed that poly-Si at crystal-filtered area shows narrower crystal grain widths and longer grain lengths than those of non-crystal filtered area MILC poly-Si. Networks of needle-like grains having a uniform surface-crystallographic orientation could be selected through CF in CF-MILC. MILC growth length was depended on the width of crystal filter. It is well known that the growth rate of MILC is practically independent of geometries of a-Si layers. It is also reported that the growth rate is about 4 μm/h at 550°C. However, we proved that the wider the width of the crystal filter is, the higher the growth rate is. Also, we found that MILC growth rates are different among the non-crystal filtered area, the crystal filter area, and the crystal-filtered area.We found that electrical characteristics of CF-MILC TFT dependence on the width of crystal filter. The narrower the width of the crystal filter is the electrical characteristic of CF-MILC TFT was improved.
A20: Novel Devices and Films
Session Chairs
Thursday PM, April 20, 2006
Room 3002 (Moscone West)
4:30 PM - **A20.1
Application of Thin-Film Amorphous Silicon to Chemical Imaging.
Tatsuo Yoshinobu 1 , Werner Moritz 2 , Friedhelm Finger 3 , Michael Schoening 4 3
1 , Tohoku University, Sendai Japan, 2 , Humboldt University Berlin, Berlin Germany, 3 , Research Centre Juelich, Juelich Germany, 4 , University of Applied Sciences Aachen, Campus Juelich, Juelich Germany
Show AbstractThe light-addressable potentiometric sensor (LAPS) is a semiconductor-based chemical sensor with an electrolyte-insulator-semiconductor (EIS) structure. A depletion layer is formed in the semiconductor layer in response to the potential on the sensing surface of the insulator, which is a function of the ion concentration in the solution. When the ion concentration is spatially distributed on the sensing surface, the distribution is mapped into the width and therefore the capacitance of the depletion layer. The local value of the capacitance can be read out in the form of a photocurrent generated by a focused laser beam.This light-addressability of the LAPS allows its application in two directions, i.e., the multisensor application and the chemical imaging application. In the multisensor application, various membranes with sensitivities to different ions are integrated on the sensing surface and individually illuminated. In this way, a single LAPS chip can detect and measure more than one kind of ion species. In the chemical imaging application, a scanning laser beam is employed to measure the local values of the ion concentration at each point on the sensing surface, which are gathered to construct a chemical image, a visualization of the ion concentration. In both applications, the spatial resolution is of critical importance. In the multisensor application, the spatial resolution limits the density of measuring sites on the sensing surface without crosstalk between adjacent sites. In the chemical imaging application, the spatial resolution limits the smallest size of structures that can be observed.The spatial resolution of the LAPS is determined by the spot size of the laser beam and the lateral diffusion of photocarriers in the semiconductor layer, which is affected by the diffusion length or the lifetime of minority carriers. When the EIS structure is illuminated, the incident light is gradually absorbed according to the absorption coefficient of the semiconductor material. The generated photocarriers still need to travel across the width of the semiconductor layer to arrive at the semiconductor-insulator interface before contributing to the photocurrent. During this process, lateral diffusion of photocarriers takes place, which determines the spatial resolution. Therefore, reduction of the width of the semiconductor layer and the diffusion length of minority carriers is expected to result in higher spatial resolution.In this study, thin-film amorphous silicon deposited on a glass substrate is employed as the semiconductor layer of the EIS structure to realize a high spatial resolution of the LAPS device. Fabrication of the device and the sensor performance are reported as well as the theory and experimental results on the spatial resolution.
5:00 PM - A20.2
Performance of Thin-film a-Si:H Microresonators in Dissipative Media
Teresa Adrega 1 , Virginia Chu 1 , Joao Conde 1 2
1 , INESC-MN, Lisbon Portugal, 2 Dept. of Chemical and Biological Engineering, Instituto Superior Tecnico, Lisbon Portugal
Show AbstractThe application of thin-film silicon to microelectromechanical systems (MEMS) has been developed to benefit from the advantages of thin-film technology such as low temperature processing (< 150 °C) and large area deposition, which allow the use of substrates such as glass, plastic and stainless steel sheets. In addition, thin-film MEMS are CMOS compatible enabling the monolithic integration of MEMS with its control electronics.Previously, we have reported the performance of electrostatically actuated thin-film amorphous silicon (a-Si:H) microresonators in vacuum and air. For resonator lengths (L) between 10-100 μm and with a 1 μm air gap, the resonance frequency (fres) is between 1-20 MHz and the quality factors (Q) are approximately 1000 in vacuum and 100 in air.There has been growing interest in using MEMS as chemical and biological sensors. These transducers are recognized as promising platforms for real-time, in situ, measurements because they can operate in vacuum, gas and liquid environments. Although electrostatically actuated microresonators are well studied in vacuum and air, there have been relatively few studies of their behavior in a liquid medium such as water. A liquid medium presents technological challenges such as high dissipation, electrode polarization or electrolysis. Resonance frequency detection in aqueous media is fundamental for the development of sensitive MEMS biological sensors applications.This work presents thin-film a-Si:H electrostatic microresonators operating in dissipative media such as air and water. Flexural and torsional microresonators are fabricated at temperatures below 110 °C on glass substrates using amorphous silicon-based thin-film technology and surface micromachining. For fres measurements the microstructures are electrostatically actuated by applying a voltage with both DC and AC components between the bridge and the gate counter electrode. The resulting deflection is detected optically.The flexural fres of microbridges was successfully measured in air and in de-ionized water using electrostatic actuation. Neither electrolysis nor electrode polarization was observed. As the medium becomes more dissipative, the fres peak shifts to lower values and there is a broadening of the peak, contributing to lower Q. For a 30 μm long microbridge, the fres shifts 0.02 MHz from vacuum to air, and shifts a further 4 MHz in water. The Q is around 1000 in vacuum and decreases, with the inverse of the pressure, to 100 at atmospheric pressure. In water the Q is approximately 10. For all the media under study, the fres was proportional to 1/L2. A strong dependence of Q with the length of the structure was also observed. Quantitative studies of resonance frequency and quality factor measured in liquid media with different viscosities and electrical conductivities will be presented. The effect of the microresonator geometry and of the type of excited vibration mode on the dissipation energy will also be discussed.
5:15 PM - A20.3
Dynamic Measurements of MEMS-Based Field Effect Transistors Using Scanning Capacitance Microscopy.
Meredith Anderson 1 , Ralph Young 1 , Craig Nakakura 1
1 Microelectronics Development Laboratory, Sandia National Laboratories, Albuquerque, New Mexico, United States
Show AbstractDue to the increasing complexity of integrated circuits, quantitative 2D techniques that analyze the dopant characteristics of state-of-the-art devices have become a necessity. Scanning Capacitance Microscopy (SCM) is a powerful tool for visualizing free carrier profiles and is commonly used to define junction positions and measure channel lengths of static devices. We employ the SCM and a novel simulation approach to quantify 2D carrier profiles of a MEMS device during operation. The ability to predict device behavior in MEMS-based products is particularly valuable because of the complex nature of integrating electronic and mechanical components. The n-channel SUMMiTTM [1, 2] field-effect transistor (SFET) is a model system for quantitative SCM due to its large channel length and device dimensions. Independent bias voltages were applied to the source, gate, drain, and well regions, while SCM was used to measure changes in carrier distribution of the cross section [3]. Device operation was confirmed by simultaneously measuring the drain current. The SCM image contrast directly beneath the gate changed from p-type to n-type as a function of applied gate bias voltage. When the device was in the ‘on’ state, the n-type inversion layer within the channel distinctly connected the source and drain. Simulations of this device were performed by modifying the MEDICI simulation package to include the SCM tip, surface oxide, and experimental biasing conditions. The model SCM tip was moved along a one dimensional carrier profile of the device region of interest, as defined by TSUPREM-4 and SIMS measurements, while two small signal ac MEDICI simulations were completed to extract the change in capacitance at each tip position. The results were compared to the SCM measurements and show excellent agreement. The simulation accurately maps the change in capacitance beneath the biased gate. This comprehensive method of imaging and simulating active devices under a variety of biasing conditions can provide direct feedback to engineers during the many stages of new product development.[1] B.L. Draper, M. Okandan, S.S. Mani, R.S. Bennett, J. Microelectromechanical Sys. 13, 500 (2004).[2] Sandia Ultra-planar Multi-level MEMS Technology (SUMMiT V™)[3] C.Y. Nakakura, P. Tangyunyong, D.L. Hetherington, M.R. Shaneyfelt, Rev. Sci. Instrum. 74, 127 (2003).Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
5:30 PM - A20.4
Black Thin Film Silicon - Light Trapping and Photoconductivity
Svetoslav Koynov 1 , Martin Brandt 1 , Martin Stutzmann 1
1 E 25, Walter Schottky Institut - Technische Universitat Munchen , Garching b. Munchen Germany
Show AbstractRecently we proposed a new process for the modification of Si surfaces, which results in an almost complete suppression of the reflectivity in a broad spectral range leading to black Si surfaces. The process results in a nano-scale texturing of the topmost 150-200 nm of the material, independent of surface orientation and doping. Thus, the process can be applied to thin films as well as to bulk silicon. In this presentation, we focus on the efficient light trapping and on the significant increase of the photoconductivity, which accompany the suppression of reflectivity in black a-Si:H and µc-Si:H thin films. The enhancement of the weak optical absorption in such films is investigated by Photo-thermal Deflection Spectroscopy (PDS) and is compared to the theoretical limits predicted for Lambertian light tripping. Measurements of the dark- and photo-conductivities were carried out in both coplanar and sandwich contact geometries. A significant increase of the photosensitivity and anisotropy of the electron transport are observed. Measurements of the lateral conductivity, induced by illumination with a light spot of variable offset in respect to the gap between the contacts, are used to estimate the distribution of the internally scattered light. The applicability of black thin film Si in photovoltaic devices is demonstrated by test Schottky barrier structures on black a-Si:H films prepared by the new method.
5:45 PM - A20.5
Combination of Metal Nano-Imprint and Excimer Laser Annealing for Location Control of Si Thin-Film Grain.
Gou Nakagawa 1 , Tanemasa Asano 1
1 , Center for Microelectronic Systems, Kyushu Institute of Technology, Iizuka Fukuoka Japan
Show AbstractPoly-Si thin film transistors(TFTs) are being extensively studied to realize active-matrix displays with signal processing circuits and so on. Grain boundaries in poly-Si films significantly degrade the TFT performance. One promising approach for fabricating high performance poly-Si TFTs is to eliminate grain boundaries in the channel region. In the previous work, we have proposed metal nano-imprint technology to grow large grains at controlled sites using solid phase crystallization(SPC). Besides, we reported that single-grain TFTs fabricated by aligning the position of the channel and the location controlled grain show superior performance to conventional-SPC poly-Si TFTs. In this work, we report a novel method of locating Si grains by combining metal(Ni) imprint and excimer laser annealing(ELA) for the lowering of crystallization temperature. In the experiment, an array of tips was prepared at the surface of (001) oriented Si by anisotropic etching with an alkali solution. Tip surface was coated with vacuum evaporated Ni. Amorphous-Si(a-Si) film(130 nm) was deposited using an UHV evaporator on SiO2 substrate. Imprint was carried out in such a way that Ni-coated tip-array was faced down to a-Si film surface and was pressed by applying pressure of 0.15 MPa. As a result, a small amount of Ni was transferred in desired position at a-Si surface. After detaching tip-array, a-Si was annealed at temperatures below 723K in an N2 ambient to form Si nuclei which act as the seed during ELA. After removal of Ni, a 20 ns-single pulse of XeCl laser(λ:308 nm) was irradiated at room temperature to crystallize Si film. From SEM observations of Si film after the laser irradiation at 516 mJ/cm2 and Secco's etching, over 2 μm-sized Si grains were found to be formed at the sites imprinted with the tips. This result indicates that the location control of crystal nucleation in ELA is certainly realized by using Ni-imprint. However, most of location controlled grains consist of several sub-grains radially grown from the grain center. From electron back-scattering pattern analysis, boundaries in the location controlled grains were classified as random boundaries(45%) and coincidence site lattice(CSL) boundaries(55%). These CSL boundaries which include low-angle tilt boundaries and <011> symmetrical tilt boundaries are considered to be electrically inactive boundaries. From these facts, the growth mechanism is considered as follows. The laser irradiated regions starts to melt from the film surface. Then several solid phase nanocrystallites survive at imprinted regions and the surrounding a-Si regions completely melts by the difference of absorption coefficient between crystal-Si and a-Si. The unmelted crystallites as multi-seed preferentially grow after the laser irradiation ceases. And the lateral overgrowth with twin growth and mechanical twinning radially occurs from imprinted sites. As a result, the location controlled grains consist of several sub-grains.
A21: Poster Session: Metal-Induced, Laser-Induced and Other Crystallization Techniques
Session Chairs
Janez Krc
Menno van den Donker
Sigurd Wagner
Friday AM, April 21, 2006
Salons 8-15 (Marriott)
9:00 PM - A21.1
Polysilicon Films Formed On Alumina By Aluminium Induced Crystallization Of Amorphous Silicon.
Etienne Pihan 1 , Abdelilah Slaoui 1 , Claude Maurice 2
1 InESS, CNRS, Strasbourg France, 2 Ecole des Mines , SMS Centre , Saint Etienne France
Show AbstractIn this work we investigated the structural properties of polysilicon films fabricated by the aluminium induced crystallisation of amorphous silicon on alumina ceramic substrates. We analyzed the films in terms of grain size distribution and grains orientation versus crystallization temperature. For these studies, we used extensively the orientation imaging micrograph (OIM) technique, a very powerful tool that allows elucidating the inner-grain structure, the grain boundaries, the grain orientation and the overall crystallographic quality of the poly-Si films. From our analysis, we may conclude that the polysilicon films formed by AIC on foreign substrates have the following orientation features: (i) for all investigated temperatures, most of the silicon grains have a deviation angle from (100) orientation between 5 and 25° ; (ii) increasing the annealing temperature tends to decrease the (100) preferred orientation; (iii) the angular boundary distribution revealed that the main defects are those that have been observed inside isolated dendrites: low angle boundaries (<2°), Σ3, Σ9 and Σ27 CSL boundaries.
9:00 PM - A21.10
Fabrication of Poly-silicon Thin Films on Glass and Flexible Substrates using Laser Initiated Metal Induced Crystallization of Amorphous Silicon.
Husam Abu-Safe 1 , Hameed Naseem 1 , William Brown 1
1 Electrical Engineering Department, Univerisyt of Arkansas, Fayetteville, Arkansas, United States
Show AbstractPoly-silicon thin films on glass and Kapton® substrates are fabricated using laser initiated metal induced crystallization of hydrogenated amorphous silicon films. The process starts by depositing 200 nm amorphous silicon films on the substrates. Following this, 200 nm of sputtered aluminum films were deposited on top of the silicon layers. The samples are irradiated with an argon ion cw laser beam for annealing. Laser power densities ranging from 4 to 9 W/cm2 were used in the annealing process. Each area on the sample is irradiated for a different exposure time. Optical microscopy was used to examine any cracks in the films and loss of adhesion to the substrates. X-ray diffraction patterns from the initial results indicated the crystallization in the films. Scanning electron microscopy shows dendritic growth. The composition analysis of the crystallized films was conducted using energy dispersive x-ray spectroscopy.
9:00 PM - A21.12
Preparation of Large, Location-controlled Si Grains by Excimer Laser Crystallization of α-Si Film Sputtered at 100°C.
Ming He 1 , E.J.J. Neihof 1 , Y. Van Andel 1 , R. Ishihara 1 , J.W. Metselaar 1 , C. I. M. Beenakker 1
1 , Dimes_Tudelft, Delft Netherlands
Show Abstract9:00 PM - A21.14
Low Thermal Budget Techniques For Controlling Stress In Si1-XGeX Deposited At 210°C.
Sherif Sedky 1 2 3 , Omar Mortagy 2 , Ann Witvrouw 3
1 Physics, The American University in Cairo, Cairo Egypt, 2 The Science and Technology Research Center, The American University in Cairo, Cairo Egypt, 3 PMT, IMEC, Leuven Belgium
Show AbstractThe demand for implementing MEMS in various systems is increasing tremendously. To increase integration density, and to improve performance and system reliability, MEMS should be merged monolithically with their driving and control electronic circuitry. For high-density integration, it is preferred to post-process MEMS on top of prefabricated electronics as this allows using standard CMOS wafers from a foundry, and at the same time significantly improve the fill factor. Post-processing restricts the MEMS thermal budget, as it should not introduce any damage or degradation to the performance of the prefabricated driving electronics. This work reports, for the first time, on the possibility of realizing surface micromachined silicon germanium structures at 210°C, which have extremely low strain gradient (10-1 μm-1). This extremely low strain gradient is obtained by tuning the physical properties of SI1-XGEX, locally, without affecting the underlying layers, by excimer laser annealing. Tuning the laser annealing condition to optimize the physical properties of PECVD SI1-XGEX is challenging, especially for films deposited at low temperatures (~ 250°C or lower) due to the high hydrogen content and the poor adhesion of these films. Furthermore, optimizing some properties might be at the cost of others. To clarify this issue, it is interesting to note that reducing the electrical resistivity implies using high laser pulse fluence. This however will increase mean stress, strain gradient and surface roughness. This paper presents a detailed study of the effect of excimer laser annealing on the physical properties of SI1-XGEX (0 ≤ x ≤ 0.69) deposited on silicon dioxide sacrificial layers at temperatures varying between 210°C and 250°C. The deposition conditions of silicon germanium are adjusted to have a good adhesion to silicon dioxide, to yield a growth rate of 22 nm/min at 210°C and to obtain an initial stress gradient that can be tuned by excimer laser annealing. This is achieved by varying the germanium concentration across the film thickness from 0% to 28%. Furthermore, the effect of varying the laser pulse fluence, rate and number on mean stress, stress gradient, electrical conductivity and surface roughness is presented. The range of Ge contents under consideration is selected to broaden the use of SI1-XGEX to a wide variety of applications, which include, but are not limited to, uncooled thermal imagers, inertial sensors, RF filters, micro-mirrors, etc. In conclusion this work shows the possibility of processing high quality MEMS layers at temperatures not exceeding 210°C. Besides their use for CMOS integration, these low temperature SiGe-based MEMS could be integrated onto more exotic substrates such as polymer films.
9:00 PM - A21.15
Thermal and Stress Modelling for the Flash Lamp Crystallization of Amorphous Silicon Films.
Mark Smith 1 , R McMahon 1 , K Seffen 1 , D Panknin 2 , W Skorupa 2
1 , Department of Engineering, University of Cambridge, Cambridge United Kingdom, 2 , Forschungszentrum Rossendorf, Dresden Germany
Show Abstract9:00 PM - A21.16
Growth of Biaxially Textured CeO2 Template Layers on Glass by Magnetron Sputtering.
Maikel van Hest 1 , Andrew Leenheer 1 , David Ginely 1 , John Perkins 1 , Charles Teplin 1
1 , National Renewable Energy Laboratory, Golden, Colorado, United States
Show AbstractRecently, we have proposed a new approach to growing crystal Si on glass for large-area electronics applications, such as displays or photovoltaics [C.W. Teplin, D.S. Ginley, and H.M. Branz, submitted to J. Non-Crystalline Solids]. In this approach, a foreign template layer is initially deposited with biaxial texture directly on glass. Subsequently, a heteroepitaxial crystal silicon layer is grown on the template; the high degree of crystalline order in the silicon will result in improved electronic properties because any grain boundaries will be low-angle and electrically benign. In our first experiments, we have explored magnetron sputtering of the template material CeO2 from CeO2 targets. CeO2 was chosen as a template candidate because CeO2 is lattice matched to silicon and because previous work has shown that it is possible to obtain biaxially textured CeO2 films on disordered substrates. We have used two techniques to successfully grow biaxially textured CeO2. Using inclined substrate deposition, where the deposition beam is at a oblique angle to the substrate, we have grown biaxially textured (220)-oriented CeO2. Using ion-beam assisted deposition, where a beam of ions is directed toward the substrate at an angle, we have grown biaxially textured (200)-oriented CeO2. We will present results using both of these techniques and discuss the potential of using CeO2 as a template layer for subsequent epitaxial growth of crystal silicon.
9:00 PM - A21.17
Fabrication of Crystallized Si Film Deposited on a Polycrystalline YSZ Film/Glass Substrate at Low Temperature.
Susumu Horita 1 , Keisuke Kanazawa 1 , Kensuke Nishioka 1 , Mikio Koyano 1
1 School of Materials Science, Japan Advanced Institute of Science and Technology, Nomi,, Ishikawa, Japan
Show Abstractμc-Si thin film on glass substrate has attracted a considerable attention for driving circuit of AMLCD and pixel TFT of OLED due to improved mobility and stability compared with a-Si TFT. Also, μc-Si thin film is more useful material for photovoltaic devise, e.g., solar cell. Although there are many reports of μc-Si thin film produced on glass substrate at low temperature, it has an incubation layer of amorphous phase with some thickness. This incubation layer grows just from the surface of the substrate and affects the device performance of TFT, in particular, bottom gate TFT. So, for low temperature crystal growth of Si film from the substrate, we tried to use a seed layer deposited on the glass substrate to induce the crystal growth from it. The requirements of the seed layer material are cubic crystal structure of Si and small lattice mismatch with Si as well as high chemical and thermal stability. So, we used a polycrystalline YSZ(yttria-stabilized zirconia) film as a seed layer because YSZ has 5% lattice mismatch with Si and can be epitaxially grown on it. The 80-nm-thick YSZ film was deposited on a glass substrate at 80°C by reactive sputtering with Ar +O2 gas and Zr metal target on which 8 pieces of 1×1 cm2 Y sheet were placed concentrically. The XRD measurement showed the deposited YSZ film was preferentially (111)-oriented and the RHEED pattern showed clear spots to indicate its good crystalline quality. After dipping the glass substrate covered with the YSZ film into the diluted HF solution, we deposited a 50-nm-thick Si film on it by vacuum evaporation method at 500°C. The color of Si film deposited on the glass substrate without YSZ film is brown and the some region of Si film on the YSZ film is yellowly transparent. We estimated the degree of crystallization of the deposited Si film by using the Raman spectroscopy with λ = 514.5 nm and the beam diameter of ~1μm. From the Si film deposited on the glass substrate without the YSZ film, the Raman spectrum showed a broad peak around 480 cm-1, which indicates the Si film is amorphous phase. On the other hand, from the transparent regions, the Raman spectrum showed a strong peak at ~515 cm-1, which indicates that the Si film is crystallized. However, we observed some brown regions of amorphous phase in the Si film deposited on the YSZ film. That means that the whole deposited Si film on the YSZ film is crystallized not entirely but partially. This is due to the SiOx layer formed partially by the interface reaction between the oxygen of YSZ film and the deposited Si. This interface layer has no crystalline information of YSZ film so that it may prevent crystalline information of the YSZ film from transmitting to the deposited Si film. We still investigate the reason for the partial crystallization induced by the seed layer more in detail. But, from this result, it can be said that a YSZ film is useful as a seed layer to crystallize an Si film at low temperature without plasma.
9:00 PM - A21.18
Correlation between Annealing Temperature and Crystallinity of Si Films Prepared by Thermal Plasma Jet Crystallization Technique
Hirotaka Kaku 1 , Seiichirou Higashi 1 , Tatsuya Okada 1 , Hideki Murakami 1 , Seiichi Miyazaki 1
1 Grad. School of Advanced Science of Matter, Hirosima univ., Higashi-Hiroshima, Hiroshima, Japan
Show AbstractThe low temperature crystallization of amorphous Si (a-Si) films is a key for the fabrication of thin film transistors (TFTs) on glass substrates. Recently, we have proposed the application of thermal plasma jet (TPJ) [1] to the crystallization of a-Si films and have demonstrated good performance TFTs with the maximum field effect mobility of 70cm2/Vs and minimum threshold voltage of 3.3V are successfully fabricated [2]. In this work, we investigated the correlation between the annealing temperature and the Si films' crystallinity. The experimental set up are as follows. The W cathode and the water-cooled Cu anode separated 1 mm each other is connected to a power supply. Arc discharge was performed by supplying DC biases of 13.4–13.7V and 150A between the electrodes with an Ar gas flow of 9.4L/min. The TPJ was formed by blowing out the arc plasma through a 4mmφ nozzle. The a-Si films were deposited on quartz substrate by PECVD method and the substrate was linearly moved by a motion stage in front of the TPJ with scanning speed ranging from 150 to 700mm/s. The distance between the plasma source and the substrate was set at 3.0mm. The substrate surface temperature during the annealing was measured by an optical probe method where the details of the measurement have been reported elsewhere [3]. The crystallinity of the films was evaluated by Raman scattering spectroscopy. The crystalline volume ratio was calculated by deconvoluting the spectra with crystalline Si, microcrystalline Si, amorphous Si and SiO2 (substrate) peaks.The a-Si films were crystallized at scan speed slower than 550mm/s. Under this condition, the maximum surface temperature (Tmax) reached to 1343K and the crystalline volume ratio was 59.1%. By reducing the scan speed to 400 and 300mm/s, the Tmax increased to 1562 and 1691K, respectively, and the crystalline volume ratio increased to 68.4 and 84.6%, respectively. At scan speed of 250mm/s, the film was heated to 1740K and we hardly observed amorphous phase in the Raman scattering spectra (crystalline volume ratio was 100%). Similarly, the peak width (FWHM) of the crystalline Si TO phonon band decreased from 11.6 cm-1 to 6.0 cm-1 by increasing the annealing temperature from 1343 to 1740K, while the peak position of it hardly changed around 514 to 515cm-1. These results indicate that the crystallinity of the Si films improved by the annealing at higher temperature. Since the Tmax exceeds the melting point of Si by annealing with 250mm/s scan speed, liquid phase crystallization may have occurred and this improves the crystallinity of the films significantly. A part of this work is supported by the Industrial Technology Research Grant Program in 2005 from New Energy and Industrial Technology Development Organization (NEDO) of Japan.[1] H. Kaku et al., Appl. Surf. Sci. 244, (2005), pp.8-11.[2] S. Higashi et al., Jpn. J. Appl. Phys. 44, (2005), pp.L108-L110. [3] T. Okada et al., Tech. Dig. Pap. AM-LCD'05, pp. 171-174.
9:00 PM - A21.2
Electron Field Emission from Aluminium Induced Crystallised PECVD thin silicon films.
Mohammed Zubair Shaikh 1 , Saydulla Persheyev 1
1 Electronic Engineering and Physics Division, University of Dundee, Dundee United Kingdom
Show AbstractThis paper reports for the first time on Aluminium Induced Crystallised PECVD thin silicon films, to produce an electron source for use as a prospective cold cathode field emission device. This results in a new phase of conducting crystallites in an insulating medium, tested through the performance of field emission measurement. We analysed changes in the surface morphology and field emission properties of the aluminium crystallised thin silicon films with varying annealing temperatures. Formation of micron-submicron tip structures, the density of these structures and the relative thresholds for field emission under varying conditions were measured.Thin amorphous silicon films were deposited using PECVD (Plasma Enhanced Chemical Vapour Deposition) on glass with various backplane metals such as Cr and Mo. These thin films were then sputtered with a thin layer of Al (~50 nm) and annealed in vacuum at temperatures varying from 350 to 500 degree Celsius. The thin Al layer was then etched away to expose the crystallised thin silicon films. The surface morphology of the thin silicon films was characterised using Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM) to study the affect of various annealing temperatures on the density of uniformly spaced “island-like” micro-crystallites on the surface of both Cr and Mo backplane samples. An increase in surface roughness was also seen after aluminium induced crystallization of thin silicon films.
9:00 PM - A21.20
Effects of Mechanical Stress on Metal-Induced Lateral Crystallization Growth Rate.
Nam-Kyu Song 1 , Seung-Ki Joo 1
1 School of Materials Science and Engineering, Seoul National University, Seoul Korea (the Republic of)
Show AbstractRecently, polycrystalline silicon (poly-Si) has drawn increasing attention for active matrix liquid crystal display (AMLCD) devices. The most common method used to form poly-Si is the Solid Phase Crystallization (SPC) which is involves annealing in a conventional furnace to crystallize amorphous silicon (a-Si) films at around 600°C. However, the crystallization temperature is too high for glass substrates, and thus intensive studies have been done to lower the crystallization temperature of a-Si films. It is well known that the crystallization temperature of a-Si can be lowered by adding some metals. This is called Metal-Induced Crystallization (MIC). However, the MIC process has a serious drawback in that it incorporates undesirable metal impurities into Si, hence making it is impossible to fabricate the TFTs. Recently, it was reported that a thin layer of Ni or Pd deposited and patterned on a-Si thin films induces lateral crystallization at the temperature of as low as 550°C. This phenomenon is called Metal-Induced Lateral Crystallization (MILC). MILC process enables crystallization of a-Si thin films without metal contamination. But, there is little known about the exact mechanism of MILC. Therefore, in this study, effects of external mechanical stress on the growth behaviors of MILC were investigated. Results showed that the MILC growth rate was 4 times faster by the tensile stress than without stress and 2 times slower by the compressive stress. In the microstructure analysis, the needle-like grains was observed to grow along the <111> direction when the tensile stress was applied, but the wider grains grew in random directions when the compressive stress was applied. Those results confirm the applicability of the MILC growth model.
9:00 PM - A21.3
Crystallization of Amorphous Si Thin Films Using Nanoscale Nickel Oxide Thin Layers Deposited through Atomic Layer Deposition
Jinha Hwang 1 , Yil-Hwan You 1 , Byung-Soo So 1 , Young-Hwan Kim 1 , Hyoung-June Kim 1 , Won-Tae Cho 2 , Ki-Seok An 2 , Yunsoo Kim 2 , Young-Cheol Kim 3 , Sung-Ryong Ryu 4 , Dong-Hoon Shin 4
1 Dept. of Mat. Sci. & Eng., Hongik Univ., Seoul Korea (the Republic of), 2 Thin Film Mat. Lab., Korea Research Institute of Chemical Technology, Daejeon Korea (the Republic of), 3 Dept. of Mat. Sci. & Eng., Korea University of Technology and Education, Cheonan Korea (the Republic of), 4 , Viatron Technologies, Seoul Korea (the Republic of)
Show AbstractLow temperature polycrystalline Si (LTPS) technology has opened up a novel approach for the next generation displays, due to higher mobility of charge carriers, compared to that of the conventional amorphous Si-based transistors. Polycrystalline Si transistors allow the wide range of applications including from the current liquid crystal displays to the organic light-emitting diodes (OLED). The combination with OLED requires a highly stringent control in transistor parameters, e.g., threshold voltage, mobilities of charge carriers, S-slopes, and leakage current. Such requirement is strongly dependent upon polycrystalline Si channels, gate dielectrics, and their interfaces. Among those three components, the crystallization of amorphous Si has generally been performed using solid phase crystallization, metal-induced crystallization, excimer laser annealing, etc. Metal-induced crystallization has suffered from high leakage current due to difficulty in thickness control of metal agents, typically Ni, along with higher level of Ni contaminations. Such limitations have hindered the application of the metal-induced crystallization to mass production in OLED combined with LTPS technology. The current study has combined the metal-induced crystallization with atomic layer deposition (ALD) of nickel oxide, in order to eliminate the barriers in thickness control and to reduce the contamination level of Ni in polycrystalline Si channel regimes. The nanoscale thickness in nickel oxide thin film was controlled using ALD with high precision and reproducibility. The microscopic features were investigated using a variety of microscopy techniques, such as optical microscopy, FESEM, and HRTEM. The crystallinity was estimated using Raman spectroscopy and UV-Vis spectrophotometry which incorporate both amorphous and polycrystalline portions. The implications of nanoscale NiO thin films in Si crystallization will be discussed in conjunction with active matrix thin film transistors on glass substrates.
9:00 PM - A21.4
Epitaxial Silicon Thin Films by Low Temperature Aluminum Induced Crystallization of Amorphous Silicon.
Khalil Sharif 1 , Husam Abu-Safe 1 , Hameed Naseem 1 , William Brown 1
1 Electrical Engineering Department, University of Arkansas, Fayetteville, Arkansas, United States
Show Abstract9:00 PM - A21.5
Low Temperature Poly-Si Sputtering Deposition Through Metal-induced Crystallization and its Application.
Hsiu-Wu Guo 1 , Chen-Luen Shih 2 , Joe Ketterl 3 , Scott Dunham 1
1 Electrical Engineering, University of Washington, Seattle, Washington, United States, 2 Materials Science and Engineering, University of Washington, Seattle, Washington, United States, 3 , MicroConnex, Snoqualmie, Washington, United States
Show AbstractGrowing polycrystalline silicon thin films via metal-induced crystallization (MIC) has been demonstrated by several research groups due to its capability of growing polysilicon at lower temperature compared to solid phase crystallization (SPC). The crystallization temperature can be reduced to different levels with different metals. Nickel is one of the most popular metals that often is adopted for this purpose. It is believed that Ni reacts with Si to form silicide (NiSi 2) so that it can prompt epitaxial growth due to the close lattice match to Si (0.4 %: 5.408 Å of NiSi2 compares to 5.430 Å of silicon).In this work, a 20-50 nm Ni layer is deposited by dc magnetron sputtering onto a 500 nm silicon oxide layer grown on silicon wafers, followed by Si deposition at 500 °C. Commercially available silicon wafers with different doping types and concentrations were used as Si sputtering targets. It is ideal not to break vacuum condition between Ni and Si sputtering in order to avoid contamination at the interface. X-ray diffraction results and cross-sectional transmission electron microscopy (XTEM) confirmed the formation of polysilicon in a columnar structure with grain size in 100-300 nm ranges. XTEM and XPS confirm that NiSi2 is formed at the Si-Ni interface which acts as seeds for Si epitaxial growth. No nickel is observed in the bulk and surface of polysilicon films, which indicates nickel did not migrate into polysilicon films. This mechanism agrees with the observation from Guliants et al. [Thin Solid Films 385 (2001) 74-80]. XTEM shows that different doping types and concentrations do not have a significant impact on the grain structure. Secondary ion mass spectrometry (SIMS) shows no significant loss in doping concentration during sputtering.After successfully growing poly-Si films, PN junctions are fabricated through this method. Both N-type and P-type silicon films are sputtered sequentially onto Ni layers, followed by patterning and lithography to form Ni contacts on both silicon layers. Post annealing is applied to reduce contact resistance. At last, characterization of PN junction is performed.
9:00 PM - A21.6
Characterization of Nickel Induced Crystallized Silicon by Spectroscopic Ellipsometry
Luis Pereira 1 2 , Hugo Aguas 2 , Rui Martins 3 , Manfred Beckers 3 , Elvira Fortunato 2 , Rodrigo Martins 2 1
1 , CEMOP-UNINOVA, Caparica Portugal, 2 Materials Science Department, FCT/UNL, Caparica Portugal, 3 , Institute of Ion Beam Physics and Materials Research, Forschungszentrum Rossendorf, Dresden Germany
Show AbstractThe Spectroscopic Ellipsometry (SE) has proved to be a very effective tool in studying polycrystalline (poly-Si) films, in a quick and nondestructive way. Several works compared the results obtained by SE with the ones obtained by other techniques such as Atomic Force Microscopy (AFM), X-Ray Diffraction (XRD), Raman Spectroscopy (RS) and Rutherford Backscattering Spectroscopy (RBS), confirming that SE gives reliable results in determining the thickness, surface roughness, crystalline fraction and composition. In this work SE was used to characterize MIC silicon films in order to analyze the influence of different annealing conditions on the structural properties of the films produced. The variation of the metal thickness has shown to be determinant on the time needed to full crystallize silicon films with different thicknesses. Films with 80 nm crystallize after 2h at 500°C after depositing 0.5 nm of Ni on it, as confirmed by XRD. Thicker films will require more time to crystallize. However the same 80 nm film will need almost 5 hours to get totally crystallized when using 0.1 nm of Ni. Using a new approach on the modeling procedure of the SE data, in this work is also shown that this technique may also be used to determine the Ni remaining inside the crystallized films. The method consists in using Ni as reference on the Bruggeman Effective Medium Approximation (BEMA) layers that will simulate the optical response of the crystallized silicon. Silicon samples and metal layers with different thickness were analyzed and this new method has shown to be sensible to changes on the initial metal/silicon ratio. The nickel concentration inside the silicon layers was independently measured by RBS and AES. The SE results show a good conformity with the RBS and AES data, confirming to be sensitive to changes on the Ni concentration, even below 1%.
9:00 PM - A21.7
Metal-induced Nickel Silicide Nanowire Growth Mechanism in the Solid State Reaction.
Joondong Kim 1 , Jong-Uk Bae 1 , Hyun-Mi Kim 2 , Ki-Bum Kim 2 , Wayne Anderson 1
1 Electrical Engineering, University at Buffalo, Buffalo, New York, United States, 2 School of Materials Science and Engineering, Seoul National University, Seoul Korea (the Republic of)
Show Abstract9:00 PM - A21.8
Fabrication of Location-Controlled Silicon Crystal Grains by Combining Excimer Laser Irradiation with Nanometer-sized A-Si.
Chun-Chien Tsai 1 , Ting-Kuo Chang 2 , Hsiu-Hsin Chen 1 , Bo-Ting Chen 1 , Huang-Chung Cheng 1
1 Department of Electronics Engineering and Institute of Electronics, National Chiao-Tung University, Hsinchu Taiwan, 2 , Toppoly Optoelectronics Corporation, Miao-Li Taiwan
Show Abstract9:00 PM - A21.9
Analytical Studies of the Capping Layer Effect on Aluminum Induced Crystallization of Amorphous Silicon.
Husam Abu-Safe 1 , A.S. Islam 1 , Hameed Naseem 1 , William Brown 1
1 Electrical Engineering Department, University of Arkansas, Fayetteville, Arkansas, United States
Show AbstractThe capping layer effect on metal induced crystallization of amorphous silicon was studied. Three sets of samples were prepared for this study. All samples had the basic layer structure of amorphous silicon layer deposited on a glass substrate. This deposition was followed by the deposition of a thin aluminum layer. In the second and third sets, however, a third layer of amorphous silicon was deposited on top of the aluminum layer. The thickness of this layer was 10% and 25% of that of the aluminum in the second and third sets, respectively. The samples were annealed at 400°C for 15, 30 and 45 minutes. The crystallization fraction in the resultant films was analyzed using X-Ray diffraction patterns. The surface morphology was examined using scanning electron microscope and atomic force microscopy. The composition analysis of the crystallized films was conducted using energy dispersive x-ray spectroscopy. Transmission electron microscopy was used to examine the grain size in the fabricated films.
A22: Poster Session: Thin Film Transistors
Session Chairs
Janez Krc
Menno van den Donker
Sigurd Wagner
Friday AM, April 21, 2006
Salons 8-15 (Marriott)
9:00 PM - A22.1
Improvement of Threshold Voltage Degradation Characteristics in a-Si:H TFT by Pre-electrical Bias-Aging for AMOLED Display
Jae-Hoon Lee 1 , Sang-Geun Park 1 , Kwang-Sub Shin 1 , Min-Koo Han 1
1 Electrical Engineering and Computer science, Seoul National University, Seoul Korea (the Republic of)
Show AbstractHydrogenated amorphous silicon thin-film transistors (a-Si:H TFTs) have gained much attention as pixel elements for AMOLED due to uniform a-Si deposition and well matured technology. However, the stability problem, such as threshold voltage (Vt) shift, may be critical for practical application. In a-Si:H TFT, the mechanism of Vt degradation due to gate and drain bias are classified into two categories, (1) defect state creation in the a-Si layer and (2) charge trapping into the gate nitride. In the AMOLED display application, low gate-voltages ( < 15V) are required so that the Vt shift of a-Si:H TFT is mainly due to creation of dangling bond defects. The purpose of our work is to reduce Vt shift of a-Si:H TFT by pre-electrical bias-aging for AMOLED display. Although pre-electrical bias-aging increases Vt of a-Si:H TFT, the quantity of Vt shift is reduced. The rate of Vt shift depends on the rate of creation of Si dangling bonds proportional to the density of weak bond sites, the occupancy of the tail state, and to dispersive diffusion of hydrogen. The density of weak bond sites and hydrogen atoms are determined during a-Si deposition. As a stress time increases, the density of weak bond sites and dispersive diffusion of hydrogen decreases so that the rate of Vt shift reduces. The TFT sample used in this work (W=200μ and L=4μ) was inverted staggered structure widely used for a commercial one. We performed a pre-electrical bias-aging at room temperature (27oC) during 50,000 sec, such as VGS=15V and VDS=0V. The widely used Vt shift equation, during the pre-electrical bias-aging, is ΔVt = 0.001256×(VGS-Vt)×t0.43. After pre-electrical bias-aging, the stress gate-voltage was increased up to keep a same electron concentration in the channel, such as VGS=16.9V and VDS=0V. The Vt shift equation after an initial bias-aging is ΔVt = 0.0006199× (VGS-Vt)×t0.43. Please note that A in equation ΔVt = A×(VGS-Vt)×tβ is proportional to weak bond density and hydrogen diffusion rate. When the same stress duration and stress voltage (VGS-Vt) is applied after pre-electrical bias-aging, the created dangling-bond density due to an electrical bias would be reduced. When the same stress duration of 50,000 sec is applied to TFT sample, the created dangling bond density (ΔNDB) after pre-electrical bias-aging is decreased from 1.38×1011/cm2 to 0.685×1011/cm2. Our experimental results indicate that although an electron breaking weak-weak Si bonding is same level, a created dangling bond is decreased so that weak bond density and hydrogen diffusion may be decreased after the pre-electrical bias aging. In summary, the proposed pre-electrical bias-aging would be effective to reduce Vt shift of a-Si:H TFT due to the reduction of weak-bond density and hydrogen diffusion rate.
9:00 PM - A22.10
Defect Passivation of Ultra Low Temperature Poly-Si Thin Film Transistors.
Choong-Heui Chung 1 , Yong-Hae Kim 1 , Jaehyun Moon 1 , Myung-Hee Lee 1 , Jung Wook Lim 1 , Sun Jin Yun 1 , Dong-Jin Park 1 , Jin-Ho Lee 1
1 Basic research Lab, ETRI, Daejon Korea (the Republic of)
Show Abstract Fabrication of poly-Si thin film transistor (TFT) on plastic substrates has challenged nearly all conventional Si based unit processes. This is mainly due to the inherent plastic thermal instability which restricts processing temperature below 250 oC. However, low temperature processes introduce a large number of defects in devices. Therefore, to improve the device performance, the defect passivation emerges as one of the most key processes in ultra low temperature poly-Si (ULTPS) TFTs fabrication. Hydrogenation has been proved to effectively passivate defects and, usually, is achieved by plasma with H2 or mixed gas containing H2. While the importance of H passivation cannot be stressed enough, literature shows scanty research on this issue for ULTPS TFTs. In this study, we report on ULTPS TFTs fabrication with SiNx:H interlayer dielectrics (ILD) below 150 oC and by annealing and deriving H from the SiNx :H interlayer dielectric without plasma application. There were few reports on plasma-free hydrogenation even though some exits poly-Si TFTs fabricated at elevated temperature. Because plasma-free hydrogenation can be carried out by a batch process, from the viewpoint of cost effectiveness, it would be preferred to plasma hydrogenation at which only one individual product can be treated. The Si dangling bonds were effectively passivated around 170 oC, and the Si strained bonds were properly passivated at higher temperature about 35 oC than the dangling bonds passivation temperature. The activation energy of H diffusion into the TFTs was estimated to be 0.87 eV. By annealing at 250 oC without plasma application, threshold voltage and electron field effect mobility were improved from 11.5 V to 3.5 V and from 86 cm2/Vs to 212 cm2/Vs, respectively.
9:00 PM - A22.11
A Novel Self-Aligned Field Induced Drain Polycrystalline Silicon Thin Film Transistor with a Vacuum Cavity by using a Selective Side Etch Process
Liao Ta-Chuan 1 , Wu Chun-Yu 2 , Tsai Chun-Chien 1 , Chen Hsiu-Hsin 1 , Chien Feng-Tso 3 , Kung Chung-Yuan 2 , Cheng Huang-Chung 1
1 Department of Electronic Engineering, National Chiao Tung University, Hsinchu Taiwan, 2 Department of Electrical Engineering, National Chung Hsing University, Taichung Taiwan, 3 Department of Electronic Engineering, Feng Chia University, Taichung Taiwan
Show Abstract9:00 PM - A22.12
The Electrical Properties of Low Temperature Polycrystalline Silicon Thin Film Transistors Prepared on Flexible Steel Foil.
Yih-Rong Luo 1
1 Electronics Research & Service Organization, Industrial Technology & Research Institute, Chutung, Hisnchu Taiwan
Show Abstract9:00 PM - A22.13
Nanocrystalline Silicon Films Deposited by RF PECVD for Bottom-gate Thin-film Transistors...
Mohammad Reza Esmaeili Rad 1 , Czang-Ho Lee 1 , Andrei Sazonov 1 , Arokia Nathan 1
1 Electrical Engineering, Unversity of Waterloo, Waterloo, Ontario, Canada
Show AbstractThin-film transistors (TFTs) in active-matrix organic light emitting diode (AMOLED) displays are required to supply high and stable driving current to OLEDs. Top-gate TFTs with nanocrystalline silicon (nc-Si) active layer have shown promise to render high mobility and stable driving current. However, to be compatible with current production facilities, bottom-gate TFTs are demanded. Currently, bottom-gate nc-Si TFTs show insufficient field effect mobility and exhibit driving current instability due to presence of amorphous incubation layer at the interface with gate dielectric. Our research is motivated by the need to eliminate the incubation layer. In order to do so, we studied nc-Si deposition process to find the RF PECVD deposition regimes which lead to minimum incubation layer.We have deposited a set of undoped nc-Si films by 13.56 MHz PECVD at 250C by varying RF power, reactor pressure, silane and hydrogen flow rates. Raman spectroscopy, constant-photocurrent method (CPM) and optical absorption have been used to measure film crystallinity, defect density and optical bandgap, respectively. Carrier transport in the films has been studied using dark conductivity, photoconductivity and conductivity activation energy measurements.Our results reveal that silane and hydrogen flow rates are the most contributing factors to film characteristics. The results also indicate that the reactor pressure does not have a significant effect on the film crystallinity. However, CPM data confirm that to obtain lower defect density, medium or high deposition pressures are preferred. We obtained films with dark conductivity and Raman crystallinity in the order of 10-6-10-7S/cm and 60-80 %, respectively. Furthermore, we have deposited nc-Si films as thin as 20nm with 60% crystallinity, which is crucial for bottom-gate TFTs. Finally, four different sets of bottom-gate TFTs have been fabricated by changing gate dielectric compositions and changing [SiH4]/([SiH4] + [H2]) gas flow ratio. The device performance, relationship to the film structure and deposition process, and future improvements will be discussed in details.
9:00 PM - A22.14
A Study of PECVD Silicon Oxynitride Films for nc-Si TFT Gate Insulator Applications...
Mohammad Reza Esmaeili Rad 1 , Czang-Ho Lee 1 , Andrei Sazonov 1 , Arokia Nathan 1
1 Electrical Engineering, University of Waterloo, Waterloo, Ontario, Canada
Show AbstractNanocrystalline silicon (nc-Si) thin-film transistors (TFTs) directly deposited by plasma enhanced chemical vapour deposition (PECVD) have been shown to promise to render high-mobility devices for organic displays and system-on-panel applications [1]. However, electrical quality of PECVD gate insulator is still insufficient to meet performance and stability requirements for nc-Si TFTs. Silicon oxide is suffering from high leakage current whereas silicon nitride interface with nc-Si channel layer yields high charge trapping. Silicon oxynitride is expected to have low leakage current and to form low defect density interface with TFT channel layer. In this research, we studied the suitability of silicon oxynitride as an alternative gate insulator.We deposited silicon oxynitride films by 13.56 MHz PECVD at 260C using He-diluted mixtures of SiH4, N2O, and NH3 or N2. Two series of NH3-diluted and N2-diluted samples were fabricated by varying the ratios of [N2O]/([N2O]+[NH3]) and [N2O]/([N2O]+[N2]), while keeping constant total gas flow rate, RF power and reactor pressure. To study electrical properties, metal-insulator-semiconductor (MIS) structures were fabricated using 200nm thick oxynitride films as the insulator layer. From current-voltage (I-V) measurements, leakage current and breakdown field were measured. Flat-band voltage was determined by high frequency capacitance-voltage (C-V) measurements. Etching rate in diluted HF acid and deposition rate were also obtained.The samples grown using N2 dilution show lower leakage current and flat-band voltage than those deposited from NH3-diluted mixtures. Our results reveal that for N2-diluted samples, the variations in N2O and N2 flow rates do not affect the leakage current. However, they largely affect the flat-band voltage. On the other hand, the oxynitride films deposited by NH3 dilution exhibit lower leakage current, etching and deposition rates, and higher flat-band voltage at higher NH3 flow rate. These results will be analyzed along with the electrical stability measurements, and the possibility of “device quality” silicon oxynitride TFT gate dielectric fabrication by RF PECVD will be discussed. [1] Czang-Ho Lee, Andrei Sazonov and Arokia Nathan, “High-mobility nanocrystalline silicon thin-film transistors by plasma enhanced chemical vapor deposition,” Appl. Phys. Lett. 86, 222106 (2005)
9:00 PM - A22.2
Bias Stress Stability of Asymmetric Source-Drain a-Si:H Thin Film Transistors
Kwang-Sub shin 1 , Jae-Hoon Lee 1 , Sang-Geun Park 1 , Min-Koo Han 1
1 School of Electrical Engineering, Seoul National University, Seoul Korea (the Republic of)
Show AbstractRecently, a-Si:H TFTs are being employed as a current source in AMOLEDs as well as a switching element in AMLCDs. The instability of a-Si:H TFTs caused by electrical stress over time may be a critical problem for using a-Si:H TFT as a current driver for OLEDs. Especially, the electrical stability under drain bias stress is of great importance for the current drivers, since they have a common constant voltage source (VDD) at the drain terminal.The purpose of our work is to report the electrical stability of asymmetric Source-Drain a-Si:H TFTs under drain bias stress. The asymmetric Source-Drain TFTs have been employed as switching devices in AMLCDs due to the advantages, such as suppressing the feed-through voltage and reducing transistor size. We have fabricated two types of asymmetric Source-Drain TFTs and one type of conventional symmetric TFT. Compared with the symmetric TFT (‘I’ shape, WD = WS), the asymmetric structures (‘L’ and ‘J’ shape) have the different width of source and drain electrode (WD > WS), but the W/L ratio was maintained to be the same (W/L = 200/4), regardless of Source-Drain shape.Before bias stressing, the transfer and output current characteristics of asymmetric TFTs showed no significant difference compared with symmetric TFT, since all the structures have the same W/L ratio. We performed 30,000sec of DC bias stressing for each structures. In the absence of a drain bias (VG=15V, VD=0V), the (VT) shifts did not vary significantly with the Source-Drain structure. However, in the presence of drain bias (VG=15V, VD=20V), the VT shifts of asymmetric TFTs were less than that of symmetric TFTs. The VT shifts of ‘L’ and ‘J’ shaped TFT were 0.29V, 0.24V respectively, while The VT shift of ‘I’ shaped TFT was 0.42V.The small degradation in the asymmetric TFTs can be explained by the defect creation model that the VT shift would be proportional to the number of induced carriers in the a-Si:H channel. In addition, it was reported that when the drain bias is applied, the induced carriers in the channel decreases, hence the defect creation decreases, finally the VT shift decreases. Since the actual drain width of the asymmetric TFTs is longer than that of symmetric TFTs, at the same W/L ratio, it can be seen that, even though the same drain bias is applied, the amount of depleted charges in the asymmetric TFTs is larger than that of symmetric TFTs. Consequently, the number of carriers in the asymmetric TFTs is smaller than that of symmetric TFTs so that asymmetric TFTs exhibit less degradation, when the drain bias is applied.In summary, under drain bias stress, the threshold voltage shift of asymmetric Source-Drain TFTs was smaller than symmetric TFTs, due to the reduced carrier concentration caused by the increased depletion. Our experimental results show that, in the current drivers for AMOLEDs, where the VDD exist at the drain terminal, the asymmetric Source-Drain a-Si:H TFT is more reliable than the symmetric TFT.
9:00 PM - A22.3
Characterization of Amorphous Silicon Thin Film Transistors Fabricated Entirely by RF Magnetron Sputtering Below 200 °C.
Seung-Ik Jun 1 , Philip Rack 1 , Michael Simpson 2 , Timothy McKnight 2 , Anatoli Melechko 2
1 Materials Science and Engineering, The University of Tennessee, Knoxville, Tennessee, United States, 2 Molecular Scale Engineering and Nanoscale Technologies Research Group, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States
Show AbstractWe have characterized the current densities and breakdown voltages of silicon nitride films (SiNx) as a function of direct current (DC) substrate bias during sputter deposition. The breakdown voltage of low temperature sputter-deposited SiNx with 20 W (125 V) substrate bias is 7.65 MV/cm which is equivalent to that of high-quality plasma enhanced chemical vapor deposition (PECVD) SiNx deposited films. Subsequently, amorphous silicon thin film transistors (a-Si TFTs) with an inverted staggered and back channel etched structure (BCE-TFTs) were fabricated and all thin film layers were entirely sputter deposited at low temperature below 200 °C. The thin film transistors were characterized as a function of the TFT geometry, post-annealing conditions, back channel etch depth, a-Si thickness. In this presentation, we will outline the process flow of the BCE-TFT process including relevant materials integration issues. The material characteristics of the a-Si films as a function of sputtering conditions (temperature and substrate bias) will be presented. Furthermore, we will discuss the characteristics of the SiNx as a function of sputtering conditions and will compare the TFT characteristics of optimized SiNx versus SiO2.
9:00 PM - A22.4
The Effect of Electrical Stress on the Leakage Current of Poly-Si TFTs Fabricated by Metal Induced Lateral Crystallization.
Shinhee Han 1 , Seungki Joo 1
1 School of Material Science and Engineering, Seoul National University, Seoul Korea (the Republic of)
Show Abstract9:00 PM - A22.5
Ambipolar Thin-Film Transistors and Invertors Fabricated by PECVD Nanocrystalline Silicon
Czang-Ho Lee 1 , Andrei Sazonov 1 , Mohammad R. Esmaeili Rad 1 , G. Reza Chaji 1 , Arokia Nathan 1
1 ECE, University of Waterloo, Waterloo, Ontario, Canada
Show AbstractNanocrystalline silicon (nc-Si:H) thin-film transistors (TFTs) deposited by plasma-enhanced chemical vapor deposition (PECVD) are very promising for the total integration of peripheral circuit drivers and switching devices in active-matrix TFT backplanes for system-on-panel. Recently, we reported high-mobility nc-Si:H TFTs showing a field-effect electron mobility (μeFE) of ~150 cm2/(Vs) at 260 oC [1]. However, to realize building blocks pertinent to complementary digital circuits, high hole channel mobility TFTs are required. In this paper, we present for the first time ambipolar nc-Si:H TFT based inverter fabricated using Cr silicide drain/source contacts. The resulting TFTs show both p- and n-channel operation. The typical p-channel TFTs show field-effect hole mobility (μhFE) of ~30 cm2/(Vs), threshold voltage (VT) of ~3.9 V, subthreshold slope (S) ~0.2 V/dec, and ON/OFF current ratio ~106. On the other hand, the typical n-channel TFTs show μeFE ~170 cm2/(Vs), VT ~2 V, S ~0.3 V/dec, and ON/OFF current ratio ~107. Such high carrier mobilities are attributed to the reduced contact resistance as well as to the good interface integrity between the channel and the gate dielectric, along with the low valence and conduction band-tail state densities. The complementary metal oxide semiconductor (CMOS) TFT invertors operated by simple connection of p- and n-channel ambipolar TFTs show good transfer characteristics with a gain of 9.97 at an input voltage of 8 V. These results demonstrate the feasibility of low-temperature and high-speed nc-Si:H TFTs for system-on-panel integration in flat panel displays and imagers, although the issues related to device stability are the subject of further improvement.[1] Czang-Ho Lee, Andrei Sazonov, and Arokia Nathan, Appl. Phys. Lett. 86, 222106 (2005).
9:00 PM - A22.6
An Asymmetric Dual Gate Poly-Si TFTs for Improving Hot Carrier Stress Stability and Kink Effect Suppression
Joong Hyun Park 1 , Woo Jin Nam 1 , Jae Hoon Lee 1 , Min Koo Han 1
1 School of Electrical Engineering, Seoul National University, Seoul Korea (the Republic of)
Show AbstractLow temperature poly-Si (LTPS) thin-film transistors (TFTs) fabricated by excimer laser annealing (ELA) have attracted considerable attention due to high mobility and on-currents. However, poly-Si TFTs still suffer from the kink currents which originate from the inherent floating body effect and rather poor reliability.The purpose of this paper is to propose a kink suppressed poly-Si TFT by employing an asymmetric dual-gate. The proposed device employs a long gate and a short gate in the dual-gate structure, which does not only improve hot carrier stress but also suppress kink effect. We have fabricated the various asymmetric dual-gate. The single-gate TFTs (L=30 μm) suffer from the kink-effect severely as the VDS increase up to 25V. The conventional dual-gate TFT (L=15+15 μm) rather suppresses the kink effect, but it still exhibits the kink current at a high drain field. On the other hand, the asymmetric dual-gate devices reduce the kink current considerably. The kink-free current saturation is observed dominantly as the asymmetry of dual-gate lengths is increased from 19+11 μm to 23+7 μm. The kink suppression of the dual-gate is induced by controlling the operation point between two sub-TFTs. As the drain voltage increases, the total transistor operates in a saturation regime, in which the sub-TFT near the source operates in a linear regime and the sub-TFT near the drain operates in a saturation regime. The total current saturation is limited and confined by the long gate TFT near source so that the IDS is no more than the saturation current level of the long gate TFT. As a result the asymmetric dual-gate suppresses kink better than the single-gate as well as the conventional dual-gate.In order to investigate a reliability of asymmetric dual-gate poly-Si TFT as well as single-gate and symmetric dual-gate poly-Si TFTs, we applied hot-carrier stress. After the stress, the single-gate TFT is degraded due to the kink and the on-current is decreased and the off-current is increased. The symmetric dual-gate TFT and asymmetric dual-gate TFT are less degraded because the dual-gate structures suppress the kink. However, although the asymmetric dual gate suppress kink more than dual-gate, reliability would be degraded compared with symmetric dual gate. Since sub-TFT near drain severe kink due to high drain field, the mobility degradation and VTH shift is larger. The degraded poly-Si film property of the short gate TFT increases the overall the channel resistance, resulting in the mobility degradation and VTH shift. However, the asymmetric still exhibit a good current saturation even after the stress because the long-gate TFT operates in the linear regime and is not degraded. Our results show that the asymmetric dual-gate suppress kink and reduce mobility degradation from 64% to 49% under the hot carrier stress compared with single-gate.
9:00 PM - A22.7
p-channel MOSFET Devices in n-type Nanocrystalline Si:H Films.
Vikram Dalal 1 , Durga Panda 1
1 Elec. and Computer Engr., Iowa State University, Ames, Iowa, United States
Show Abstract9:00 PM - A22.8
Direct Deposition of Microcrystalline Silicon TFT at Low Temperature by ICP-CVD.
TeChi Weng 1 , Chih-jeng Huang 1 , I-Hsuan Peng 1
1 , ITRI, Hsinchu Taiwan
Show Abstract9:00 PM - A22.9
Characteristics of Low-Temperature Polysilicon Thin-Film Transistor with Gate Insulator Grown by Atomic Layer Deposition.
Woo-Jung Lee 1 , Min-Ho Cheon 2 , Sa-Kyun Rha 2 , Youn-Seoung Lee 1 , Won-Jun Lee 3
1 Division of Information Communication and Computer Engineering, Hanbat National University, Daejeon Korea (the Republic of), 2 Department of Materials Engineering, Hanbat National University, Daejeon Korea (the Republic of), 3 Department of Advanced Materials Engineering, Sejong University, Seoul Korea (the Republic of)
Show AbstractThe formation of high-quality SiO2/polysilicon interfaces at a low temperature is one of the key technologies to manufacture high-performance low-temperature polysilicon thin-film-transistor (TFT) arrays for display device applications. Although plasma enhanced chemical vapor deposition (PECVD) is the most popular technique in TFT manufacturing processes for growing SiO2 films at below 400 °C, plasma damages during PECVD degrade the quality of SiO2/Si interfaces significantly. There are several approaches to reduce the plasma damages, for example electron cyclotron resonance (ECR) plasma CVD. In this work, we adopted atomic layer deposition (ALD) technique to produce high-quality SiO2 film without plasma damages. ALD SiO2 thin films were deposited at 350–400 °C by alternating exposures of SiH2Cl2 and an O3/O2 mixture, and the characteristics of the deposited films and the SiO2/polysilicon interfaces were comparatively studied with the conventional PECVD SiO2 films. The ALD films exhibited physical and electrical characteristics better than those of PECVD films, especially in terms of leakage current, roughness and wet etching rate. The charge densities at the SiO2/Si interface were examined by measuring the capacitance-voltage (C-V) plots for metal-oxide-silicon (MOS) capacitor structures. Smaller densities of interface and oxide charges were obtained for ALD films. The top-gated PMOS and NMOS TFT arrays were also fabricated with gate insulator deposited by using either ALD or PECVD technique, and the transfer characteristics of TFT arrays were compared for both types of gate insulator films.
A23: Poster Session: Solar Cells
Session Chairs
Janez Krc
Menno van den Donker
Sigurd Wagner
Friday AM, April 21, 2006
Salons 8-15 (Marriott)
9:00 PM - A23.1
Analysis of Amorphous-Microcrystalline Si:H Tandem Solar Cell
Andrzej Kolodziej 1 , Pawel Krewniak 1
1 Department of Electronics, AGH University of Science and Technology, Krakow Poland
Show Abstract9:00 PM - A23.2
Boron Doped Polycrystalline Silicon Produced By Step-by-Step XeCl Excimer Laser Crystallization
Rosari Saleh 1 , Nobert Nickel 2
1 of Physics, Universitas Indonesia, Depok Indonesia, 2 , Hahn-Meitner Institute, Berlin Germany
Show AbstractExcimer laser crystallization is an efficient technology to produced high-quality polycrystalline silicon (poly-Si) through the crystallization of amorphous silicon (a-Si). Commonly, a-Si with low H content is used for laser crystallization because the presence of large amounts of hydrogen causes difficulties for the crystallization process. To obtain large crystalline grains, a laser fluence exceeding the melt-through threshold must be applied. Under these conditions a-Si film ablates due to explosive out-diffusion of hydrogen. This can be prevented using laser dehydrogenation and crystallization method, also known as step-by-step crystallization.In this work, boron doped hydrogenated amorphous silicon films were prepared by rf-glow discharge decomposition of silane on a quartz substrate. Doping was achieved by mixing silane with diborane. The nominal gas doping was varied up to 2000 ppm. These samples were crystallized using a step-by-step laser dehydrogenation and crystallization procedure at room temperature and at a pressure of about 3 x 10-5 mbar using the 308-nm laser line of a XeCl excimer laser. To avoid explosive out-diffusion of hydrogen, the samples were crystallized employing a step-by-step crystallization scheme. Laser crystallization starts at a fluence of 100 mJ/ cm2 and ending at the desired final laser fluence using steps of 40 or 20 mJ/ cm2 depending on the hydrogen content of the starting materials and the film thicknesses. The influence of laser crystallization on structural change, hydrogen bonding and hydrogen diffusion is investigated employing Raman spectroscopy and hydrogen effusion measurements.Our results show that in the first step of the laser crystallization procedure a structural transformation takes place. Initially, the sequential crystallization process leads to the formation of a two-layer system. The Raman spectra of completely crystallized heavily doped films are distorted. This is due to a resonant interaction between optical phonons and direct intraband transitions that is well known as a Fano resonance. The structural change is accompanied by a significant change in Si-H bonding and a decrease of the total hydrogen concentration in specimens. The hydrogen bonding is influenced by the presence of dopant. Samples crystallized with a laser fluence higher than 300 mJ/cm2 can no longer be characterized by Raman spectroscopy. This drawback can be overcome by hydrogen effusion measurements. From this measurement, the hydrogen density-of-states distribution can be derived. In completely crystallized samples two prominent peaks around 1.1 and 1.5 eV below the hydrogen transport states are observed. While a third peak that is observed in amorphous starting material located at about 0.8 eV below the hydrogen transport statesdisappeared. The results of Raman spectroscopy will be correlated with the results obtained from hydrogen diffusion measurements
9:00 PM - A23.3
Double-heterojunction Hot-wire CVD Silicon Solar Cells with High Open-circuit Voltage.
Eugene Iwaniczko 1 , Matt Page 1 , Yueqin Xu 1 , Qi Wang 1 , Lorenzo Roybal 1 , Dean Levi 1 , Russel Bauer 1 , Howard Branz 1 , T Wang 1
1 , NREL, Golden, Colorado, United States
Show Abstract9:00 PM - A23.4
Thin-Film Polycrystalline-Silicon Solar Cells on Ceramic Substrates Made by Aluminium-Induced Crystallization and Thermal CVD.
Dries Van Gestel 1 , Ivan Gordon 1 , Lode Carnel 1 , Kris Van Nieuwenhuysen 1 , Guy Beaucarne 1 , Jef Poortmans 1
1 Silicon Solar Cells, IMEC, Leuven Belgium
Show AbstractAt present about 50% of the cost of crystalline-silicon solar modules arises from the expensive silicon material. Thin-film Si solar cells on low-cost foreign substrates can therefore lead to a large cost reduction if sufficiently high efficiencies can be obtained. A promising approach is to make thin-film polycrystalline-silicon (pc-Si) solar cells by epitaxial thickening of a large-grained but thin seed layer made by aluminium-induced crystallization (AIC) of amorphous silicon. We used a commercial single-wafer atmospheric-pressure CVD system at 1130°C to grow epitaxial absorber layers onto AIC layers on ceramic alumina substrates. This enables us to easily reproduce the crystal structure of the seed layers into the absorber layers at high growth rates above 1 μm/min. To reduce the surface roughness of the alumina substrates we applied flowable oxides of various thicknesses onto the substrate surfaces prior to the seed layer preparation. The typical thickness of the epitaxial layers was between 2 and 6 μm. We compared cells with diffused homojunction emitters to cells with low-temperature heterojunction emitters, and developed simple laboratory processes to obtain both mesa cells (with base contacts at the periphery of the cell) and cells with interdigitated top contacts. We found that the high surface roughness of the ceramic substrates leads to small grains and poor cell results. The use of thick flowable oxide layers led to absorber layers with much larger lateral grain sizes up to 20 μm, and to markedly higher open-circuit voltages up to 520 mV. Heterojunction emitters in combination with interdigitated top contacts led to the best cell results. Without light trapping, our best pc-Si solar cells so far showed energy conversion efficiencies up to 6%. These solar cells are the best pc-Si cells obtained on ceramic substrates where no (re)melting of Si was involved. AIC seed layers are clearly well suited to obtain thin-film polycrystalline-silicon solar cells of high quality. An interdigitated cell structure combined with a heterojunction emitter seems to be the best cell concept for thin-film pc-Si solar cells. By further improving the absorber layer quality and by applying an effective light trapping, we expect to obtain much higher efficiencies in the near future.
9:00 PM - A23.6
Local Current Flow in Mixed-Phase Silicon Solar Cells and Correlation to Light-Induced Open-Circuit Voltage Enhancement.
Baojie Yan 1 , C.-S. Jiang 2 , H. R. Moutinho 2 , M. M. Al-Jassim 2 , Jeffrey Yang 1 , Subhendu Guha 1
1 , United Solar Ovonic Corp., Troy, Michigan, United States, 2 , National Renewable Energy Laboratory, Golden, Colorado, United States
Show AbstractIt is well known that the best a-Si:H solar cells are made close to the amorphous-nanocrystalline transition but still in the amorphous regime; while the best nc-Si:H solar cells are close to the nanocrystalline-amorphous transition but still in the nanocrystalline regime. Solar cells made in the transition regime are called mixed-phase solar cells. A better understanding of the mixed-phase solar cell may help in improving both the a-Si:H and nc-Si:H solar cells. One interesting phenomenon in the mixed-phase solar cells is the light-induced Voc enhancement [1]. A parallel-connected diode model was proposed to explain the light-induced Voc increase [2].In this paper, we use conductive atomic force microscopy (C-AFM) to measure the local current flow in the mixed-phase n-i-p solar cell structure without the top ITO contact. The forward biased C-AFM images reveal that for the fully amorphous region the current is very low in the entire surface. However, high current spikes appear in the mixed-phase region, where the current spikes correspond to the formation of clusters of nanocrystallites. In addition, the density of the current spikes increases from the mixed-phase to the substantially nanocrystalline region. Adding a 50-nm thick a-Si:H buffer layer between the p and i layers significantly reduces the magnitude of the high current spike, although the top morphology appears mainly unaffected. The C-AFM measurements provide additional supporting explanation for the light-induced Voc increase in the mixed-phase solar cells [2].[1] K. Lord, B. Yan, J. Yang, and S. Guha, Appl. Phys. Lett. 79, 3800 (2001).[2] B. Yan, J. Yang, and S. Guha, Proc. of 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, May 11-18, 2003, p. 1627.
A24: Poster Session: Other Devices
Session Chairs
Janez Krc
Menno van den Donker
Sigurd Wagner
Friday AM, April 21, 2006
Salons 8-15 (Marriott)
9:00 PM - A24.1
Light Filtering Properties in a-SiC:H Multilayer Structures: A SPICE model.
Joao Martins 1 , Miguel Fernandes 1 , Alessandro Fantoni 1 , Yuriy Vygranenko 1 2 , Manuela Vieira 1
1 Electronics, Telecommunications and Computer Engineering, Univ Lisbon, Lisboa Portugal, 2 , University of Waterloo, Waterloo, Ontario, Canada
Show AbstractEarly work on characterization and modeling of the laser scanned photodiode electrical behavior gave some insight for the development of the current design. Then, the large-area pin structures used for pattern detection of a B/W image were modeled by a consistent non-planar array of photodiodes. In this paper the results from different light wavelengths are the main concern. Further, a model for tandem structures like pin-pin stacked photodiodes is established. Both models are merged in order to achieve image pattern and color detection. The model follows SPICE descriptions with a regular program for analysis and simulation of integrated circuits.The circuit includes elements trying to capture several aspects of the sensor structure and the observed data: diodes including their physical parameters for p-i-n structures; current sources discriminating image light wavelength cases of study; capacitors showing charge and phase dependencies; extra parallel and series resistors; and current dependent current sources for some transmission and feedback aspects. Adjusting the values of some components parameters or the type of the simulation analysis, it is possible to explain some of the observed singularities of the sensor, aiming to explain color detection. Depending on the wavelength of the scanner beam, this modulated light source can be modeled through a sine-wave current source in parallel with the target diode. The current measured at the electrical contacts has a steady state and a sinusoidal component, the former due to the diode currents and the steady state photocurrent, and the later due to the scanner induced photogeneration.The simulation procedure has two main phases: the circuit construction and the simulation process. In the first phase, the input parameters and the array dimensions are defined in order to include the image area and the surrounding dark region. In the second phase, a current distribution composed of a mxn matrix is obtained. Each run step corresponds to a different assigned node where the scanner source is applied. The stored values are the amplitudes of the steady state signals and the amplitudes of the sinusoidal small signals taken in each simulation process step.Electrical simulations were performed for different transducer configurations and illumination conditions and compared with the experimental data. A physical model supported by the electrical simulation gives insight into the methodology used for image representation and color discrimination.
9:00 PM - A24.2
Silicon Etching Study in a RT-CVD Reactor with the HCl/H2 Gas Mixture.
Nicolas Loubet 1 , Alexandre Talbot 1 , Didier Dutartre 1
1 FEOL R&D, STMicroelectronics, Crolles France
Show Abstract