Effects of SiH2Cl2 and H2 addition to SiH4 on photochemical vapor deposition (photo-CVD) of amorphous Si (a-Si) and epitaxial Si films were investigated, and roles of Cl and H in the growth mechanism were discussed. The surface morphology and the defect density of the obtained a-Si films were evaluated by atomic force microscope (AFM), and constant photocurrent method (CPM), respectively. It was found that the H2 dilution of SiH4 is effective to prepare the a-Si films with smooth surfaces and low defect density. Furthermore, nearly flat surface with the root mean square (RMS) of the roughness of 0.35nm was obtained by the SiH2Cl2 addition to H2-diluted SiH4. In the Si epitaxy, the SiH2Cl2 addition is also efficient to improve the film quality. It is suggested that the growing surface is terminated by both Cl and H, and the surface is more stable than that terminated by only H. Therefore, the diffusion on the growing surface of the film precursors such as SiH3 is enhanced, resulting in the improvement of the surface morphology and the film quality.

Hydrogen contents were investigated in a-Si:H films prepared by PECVD under various conditions. The H atom contents were divided into SiH and SiH2 densities by spectral deconvolution of the IR absorption. Dehydrogenation of SiHo simultaneously occurs with Si-Si bonding in the surface reaction. The mechanisms of surface reaction were proposed based on an analysis of the variation in S1H2 density with the deposition condition. There are two kinds of processes, the fast process and the slow process. The fast process incorporates SiH2 into the network. The activation energy of the slow process is 0.3 eV, which is due to the surface migration of adsorbed radicals.

We discuss the preparation, properties and device potential of microcrystalline Si:H (μc-Si:H) material deposited using plasma-CVD techniques. We have been able to deposit high quality μc-Si:H using both glow discharge and ECR plasma CVD techniques. In both cases, the critical parameter was the presence of a high flux of H, and high power. The quality of the material was measured using sub-gap photo-conductivity, in both films and devices. We find that the absorption coefficient in μc-Si:H films follows the c-Si curve very closely, with little excess absorption. We have also fabricated p-i-n devices in these materials. We find that the properties of the devices depend critically upon the p-i interface and on the deposition technique used to make the devices. We use quantum efficiency techniques to study the devices in detail. We find that often, a good device characteristic corresponds to a case where the undoped layer is a mixed phase (amorphous-microcrystalline) and not just microcrystalline. In contrast, using just the microcrystalline phase for i layers in devices leads to severe interface problems, which lead to inflexion points in I (V) curves. Using the mixed phase for i layer, we have made solar cells with 0.46 V open circuit voltage, and a fill factor of 0.58. In contrast, using a microcrystalline phase can lead to devices with high voltages (0.6), but poor fill factors. We also discuss the device potential of microcrystalline devices, and we find that it may be possible to fabricate devices with 8% solar conversion efficiency in this material once all the interface problems are solved. Such devices may have applications as low-gap cells in an a-Si:H/μc-Si:H cell structure.

A scanning tunneling microscope (STM) has been used to study the topology of the surfaces of a series of thin hydrogenated amorphous silicon (a-Si:H) films deposited by rf discharge from pure silane. The substrates were atomically flat, oxide-free, single-crystal Si (100). Films were grown in our laboratory and transferred to the STM with no air exposure between growth and measurement. A series of thin films between 1 and 50 nm in thickness reveals the initial growth stage of a-Si:H on Si (100). Initial nucleation and islanding can be seen on these films. The surface has a distribution of island sizes. The rms roughness and a surface lateral correlation length were measured as functions of film thickness. The rms roughness grows sublinearly with thickness from 0.3–0.5 nm as the film thickness is raised from 1 to 50 nm. The lateral size of the surface features also grows with film thickness. The growth of the roughness and correlation length can be compared with the dynamic scaling model. In addition, the topographs reveal occasional structures of larger size and low density on the film surface. These structures are nanoparticles of silicon deposited from the plasma during film growth. The frequency of these features scales with film thickness, and represents 10−510−4 of the total film volume.

This paper describes the grain formation in very low temperature polycrystalline silicon (poly-Si) growth on SiO2. The silicon films were deposited by electron cyclotron resonance chemical vapor deposition with hydrogen dilution at 250°C and without any thermal annealing. The largest grain sizes observed in the poly-Si film is about 1 μm. The grains have a leaf-like shape as observed in plan-view transmission electron microscopy. The grain morphologies were determined by cross-sectional transmission electron microscopy and atomic force microscopy. Raman scattering spectrum was used to determine the crystalline fraction. X-ray diffraction patterns were used to study the film crystallinity. A simple model of grain formation is proposed.

The interpretation that the red shift of the a-Si:H photoluminescence spectrum is due to carrier hopping is incompatible with Monte-Carlo simulations. We suggest that thermalization is instead the result of slow, gradual, electronic self-trapping. We show how this model reconciles the narrowness of the photoluminescence lifetime distribution with the observed red-shift of the energy spectrum.

PL measurements on device-quality samples of a-Si:H (dark ESR bulk signal, Ns < 3 × 1015spins/cm3) and on a sample with Ng = 1016 spins/cm3 have been performed using excitation energies, E×, from approximately 1.3 to 2.4 eV. Sample temperatures investigated cover the range 4 to 300 K, and PL energies both below and above the excitation energies were measured. PL spectra were recorded from approximately 0.7 to 1.8 eV with a dynamic range exceeding 1012 at constant incident power density. The results permit the separation of the one-photon (band-tail to band-tail) and two-photon (two consecutive excitations through silicon dangling bonds) excitation processes. In contrast to the well-known, strong temperature dependence of the PL efficiencies near the PL peak, the PL at the highest energies (> 1.7 eV) depends only weakly on the temperature at constant Ex.

We report extensive optical and electronic transport data on silicon-implanted a-Si:H, annealed in steps in the dark or with additional illumination. All measured properties relax gradually with increasing annealing temperature. The dark conductivity of the as-implanted film is dominated by hopping conduction via midgap defects. This channel is pinched off during the initial stages of annealing. The midgap defect density and the Urbach energy follow an annealing path that agrees qualitatively with the trajectory postulated by the equilibrium theory of the dangling-bond density. Therefore, the silicon network and the defect density equilibrate continuously during network relaxation.

A systematic study of both proton and deuteron NMR in hydrogenated amorphous silicon films has revealed significant resonance shifts among various resolved components. The shifts include both paramagnetic and diamagnetic displacements of resolved spectral features from trapped molecular hydrogens. The shifts depend on film quality and deposition conditions. Some of the shifts vary as 1/T and reflect Curie susceptibilities characteristic of local regions of differing dangling bond densities. Spectra from relatively immobile hydrogen molecules trapped in nanovoids are shifted diamagnetically and broadened as temperature decreases. Hydrogens tightly bound to Si do not show similar changes and thus are more remote from dangling bonds and other magnetic defects. Similar spectrally-resol ved shifts have been observed in a-Ge and a-SiGe films and are correlated with film photovoltaic quality as measured by mobility-lifetime products.

Nuclear magnetic resonance (NMR) measurements of 1H dipolar echoes in a-Si:H indicate that the dipolar spin-lattice relaxation time, T1D, can be used as a measure of local hydrogen motion. In general, the microscopic hydrogen motion is orders of magnitude faster, and the microscopic activation energies are much smaller, than those measured macroscopically using secondary ion mass spectroscopy (SIMS). Although no detailed model exists to explain these apparently divergent results, these discrepancies can be understood phenomenologically by assuming a very broad distribution of rates. In this case the measured activation energies can be shown to scale roughly as the logarithm of the time over which the measurement is made.

High quality a-Si:H thin films with varied optical gaps in the range from 1.55 to 2.1 eV were fabricated by various methods, i.e., the standard RF glow discharge of silane, “Chemical Annealing” and ECR-H-plasma from SiCl2H2 under in situ monitoring with an ellipsome try. Despite marked differences in the local structure, all these films showed low defect density as low as (3–5) × 1015 cm3. In addition, the stability for light soaking was improved markedly for the films made by promoting intensively structural relaxation with atomic hydrogen.

Many researchers have found the photoresponse of germanium-based films to be directly related to the homogeneity of the film's microstructure, with more homogeneous films having better photoresponse. Transmission electron microscopy (TEM) gives the resolution needed for studying microstructure, but is typically limited to imaging films with a thickness of less than 200 nm. Surface measurement by atomic force microscopy (AFM) gives a three-dimensional representation of the film surface and can be used for measuring any thickness film. A more complete idea of the microstructure can be gained by combining AFM with small-angle x-ray scattering (SAXS) measurements. SAXS measurements yield the size of internal voids or less dense regions and indicate whether the inhomogeneities have a preferred orientation, such as might be associated with columnar growth. Films of amorphous germanium and germanium carbon alloy were grown by the PECVD method from SiH4, GeF4, H2 and CH4 source gases. The films were then characterized using AFM, SAXS, and FTIR. The effect that various growth conditions had on the microstructure and photoresponse will be reported.

By using a Nd: YAG-pumped optical parametric oscillator (OPO) as excitation light source (8 ns pulse width), the dynamic range of photothermal deflection spectroseopy (PDS) is increased by a factor up to 1000. This enables the study of nonlinear defect absorption in intrinsic, phosphorus and boron doped a-Si:H. To probe nonlinear absorption of defects, the intensity of the fundamental emission of the Nd:YAG (1064 nm) has been varied over three orders of magnitude. For intensities greater than 1024 photons/ (cm2s), a significant increase of a by a factor 1.4 in p-, 1.2 in i-, and 1.15 in n-a-Si:H is detected. From electron spin resonance (ESR) and transient photoconductivity experiments on intrinsic a-Si:H, a bleaching of the defect density by electron excitation into the conduction band is measured. A hole-like signature in the ESR spectra indicates that electrons from the valence band are excited into the defect band as well. The data are interpreted based on a model for IR-induced nonlinear optical effects.

We report the first yield spectroscopy study on well characterized c-Si/a-Si:H heterojunctions grown in situ under UHV conditions. We find that this spectroscopy, when operated in the constant final state mode, allows a direct and precise determination of the valence-band discontinuity at the interface. A value of δEv = 0.44 ± 0.02 eV was found for the discontinuity.

The localized density of states (LDOS) at interfaces between intrinsic silicon and silicon nitride (Si3N4 films are studied using transient voltage spectroscopy (TVS). In the TVS technique, the transient of the voltage across a MIS-diode after a trap filling voltage pulse is measured using a high-impedance voltage probe. This allows us to make a precise measurement of the LDOS at undoped Si/insulator interfaces. The LDOS in a-Si:H/Si3N4systems has a broad peak around the energy of 0.9 eV below the conduction-band edge. A modification of the LDOS at a-Si:H/Si3N4 interfaces by bias-annealing is clearly observed using this technique. The results are consistent with the defect pool model. The LDOS in laser annealed poly-Si/Si3N4 systems has a peak centered 0.6eV below the conduction-band edge, which seems to be the Si dangling bond states in the poly-Si films.

We report results of a transient modulated photocurrent technique which allows us to observe the time evolution of the D0 sub-band under the application of optical bias light and after turning off this bias light Our measurements show that the D0 band shifts monotonically to shallower thermal energies after the bias light is applied, with roughly 10 seconds to saturation at 300K and to deeper thermal energies after removing the bias light, with a decay time of over 1000 seconds. We have also found there exists an intimate relation between the motion of the D0 band and that of the quasi Fermi level as deduced from the transient photoconductivity and therefore, in particular, to the long time photoconductivity decay. This relation is exactly reproduced by the assumption of a D0 band whose energy position evolves in time, together with a recombination process dominated by changes in the charge state of a deeper defect band under light bias.

We discuss the subtle effects involved in observing slow dangling bond relaxation by studying capacitance transients in p-type hydrogenated amorphous silicon (a-Si:H). The data suggest that neutral dangling bonds are reversibly converted into metastable positive charged dangling bonds by hole trapping. These metastable positive dangling bonds reconvert to neutral dangling bonds upon annealing at elevated temperature. The annealing kinetics for this process are the same as those observed for annealing of quenched in conductivity changes in p-type a-Si:H.

Measurement of the dependence of emission capacitance transients on filling pulse duration has been extended to devices with Ohmic back contacts. Capacitance transients on devices possessing identical bulk 20-ppm P doped a-Si:H but either Ohmic contacts or blocking contacts were compared. The devices with blocking contacts completely reproduced in quantitative detail the previously observed anomalous dependence of capacitance transients on filling pulse duration. The diodes with Ohmic contacts showed no evidence of the anomalous filling pulse effect even for light-degraded, resistive samples. Current injection measurements show that blocking contacts delay the charge injection into the device by about 10–100 msec.

In order to clarify the microscopic origin of the photodegradation in a-Si:II, two kinds of approaches have been done. One approach is to carry out the light soaking both at room temperature (RT) and 77 K to cause the photodegradation and to investigate the recovery by thermal annealing. The other approach is to investigate the light-induced ESR in hydrogenated amorphous silicon-nitrogen alloys (a-Si1-xNx:H) and to see the x-dependence down to x = 0. From the first approach, it is found that there is a very broad continuous distribution of annealing activation energies from a few eV through 1.4 eV. It appears less likely that such a wide distribution of annealing activation energy is related with the bond-breaking model. On the other hand, it is found from the second approach that there exist two distinct origins for the light-induced spins, the fast process and the slow process, which are likely to be the carrier capture and the bond breaking, respectively. From the continuation of the observed behavior to a-Si:H, that is, x = 0 in a-Si1-xNx:H, the RT photodegradation in a-Si:H appears to be the bond breaking.

An undoped and a compensated a-Si:H sample have been degraded by 17–34 ns laser pulses and by steady light (CW) at 78K and 300K. The light-induced defect concentration N is monitored by the increase in subgap absorption a. For the same change in a pulses degrade the photoconductivity σp more than CW light and more strongly for exposures at 78K than at 300K. The lack of correlation between σp and N suggests that light soaking causes additional damage besides an increase in N. This additional effect is more pronounced for pulse and low temperature exposures.