The use of small-angle x-ray scattering to examine nanostructural features of a-Si:H and the related alloys a-SiGe:H and a-SiC:H will be reviewed. A wide range of H, Ge, and C compositions has been investigated. The films examined came from several film- and device-making groups and represent current state-of-the-art solar cell material or attempts to develop improved and more stable material. A detailed comparison of the three classes of materials reveals dramatic differences in nanostructure. The diffuse component of the small-angle scattering, not recognized or discussed in previous small-angle scattering experiments on these materials by other groups, is shown to contain potentially valuable information on the atomic-scale structure.

Combining real-time ellipsometry and atomic force microscopy (AFM) the growth of hydrogenated amorphous silicon (a-Si:H), deposited on crystalline silicon wafers with a native oxide layer on top and on fused silica from a dc glow discharge, has been studied from initial nucleation to the final morphology. By in-situ ellipsometry we detected the evolution of morphology changes. The surface structure has been determined by AFM in the nucleation phase and in the subsequent growth stage. During nucleation on crystalline silicon only few (about 40 per 1μm) flat islands of a-Si:H (up to 4 nm high and up to 50 nm in diameter) grow with a strongly enhanced rate compared to bulk deposition. Once the crystalline surface has completely been covered by a-Si:H, the fast deposition of these islands stops and further surface structures, comparable with the initial ones, start to grow gradually until a homogeneous final roughness up to 5nm high is formed. Nucleation of a-Si:H on fused silica yields densely distributed nuclei (up to 1.5 nm high and up to 25 nm in diameter), indicating a shorter surface diffusion length on this substrate compared to the growth on silicon wafers. The ongoing film deposition, however, finally results in a morphology comparable to the one of a-Si:H grown on crystalline silicon. Using hydrogen dilution we found that the final roughness is affected by the dilution ratio; furthermore infrared spectroscopy reveals the surface structure to be correlated with the hydrogen content of the a-Si:H films.

We use ab initio pseudopotential local-density-approximation methods to create and study related supercells containing 62 Si atoms and from 5 to 7 H atoms. In particular, we obtain energies and structural properties of H in different charge states that are passivating dangling bonds, in bond centered positions, and in other interstitial sites. The most striking result found is the rather large (of order 1 eV) spread of energies for a given type of defect, depending on its surroundings. We also find that changes in single particle energies or energy eigenvalues from defect to defect are not particular close to changes in the total energy and that the distinction, between bond centered H and a dangling bond plus an Si-H bond, is not as clear cut as one might think. The calculations also provide insights into possible migration mechanisms and energies for H movement.

We have studied the microscopic mechanism of the X-ray-irradiation-enhanced crystallization in amorphous silicon (a-Si) based upon an ab initio molecular-dynamics simulation. We find that the bistable dangling-bonds (sp3- and sp2-like structures) exhibit a large lattice relaxation and are strongly related to the X-ray-irradiation-induced vacancy-interstitial-pair formation. The vacancy-interstitial-pair formation reduces the formation energy of the vacancy to zero and enhances the crystallization with small migration energy of the vacancy. The crystallization rate in X-ray-irradiated a-Si is dominated by the migration energy of the vacancy in this mechanism because the formation energy is zero in X-ray-irradiated a-Si and one order of magnitude larger than the migration energy without X-ray irradiation.

Both photoluminescence (PL) and electroluminescence (EL) energy spectra were studied in the temperature range of 80 to 300 K on a-Si:H p-i-n structures. The EL spectrum depends on several parameters such as the i-layer thickness and the sample structure with or without a buffer layer (b-layer), while the PL spectrum shows no difference when those parameters are varied. Comparing PL and EL in 0.5 μm p-i-n devices with and without a buffer layer, we found that (a) at 80 K, the main-band peak energy is 1.3 eV for PL and 1.2 eV for EL; (b) the PL spectral line shape does not change with the insertion of a buffer layer, but the EL spectra show more enhanced main-band luminescence with the buffer layer; (c) the temperature dependence of the PL intensity shows a slope of 26 K which is similar to that of a-Si:H films, but the EL efficiency shows a weaker temperature dependence that varies with the diode structure.

We have developed an improved analysis of constant photocurrent method (CPM) data. It is based on a numerical simulation of CPM spectra taking into account the full set of optical transitions between localized and extended states, capture and emission processes as well as the position of the Fermi level. Comparing measured and simulated CPM spectra provides information about the density of localized states in a-Si:H, i.e. the valence band tail, the integrated defect density, the energy distribution and the charge state of defect states. Based on these results we examine the predictions of the defect-pool model. The defect distribution in undoped and doped a-Si:H can be described by the defect-pool model taking into account the doping level dependence of principal parameters including the valence band tail, the equilibration temperature, and the width of the defect-pool.

The structural memory model of slow defect relaxation in a-Si:H is extended to the limit of long defect filling times. The model was proposed in order to explain unusual, defect filling-time dependent capacitance transients that were observed for short defect filling times. For long defect filling pulses however, the experiments show normal charge emission transients that saturate into filling-time independent transients. We present two possibilities for explaining the approach to saturation within the structural memory model. Results of Monte Carlo simulations of the models are discussed.

We present the results of studies on the defect properties and the effect of light soaking for various hot wire deposited (HW) films. We employ junction capacitance measurements together with the transient photocapacitance spectroscopy to measure the deep defect densities in as-grown state (state A) and in light soaked state (state B). Good agreement is found between the defect densities measured from both measurements. The HW film with a hydrogen content of 10 – 12 at.% shows physical characteristics and defect densities similar to conventional PECVD films. The HW films with hydrogen content, CH, in the range 2 – 9 at.% show a smaller defect density in state B than the defect density of the film with higher CH. However, the film with a hydrogen level of less than 1 at.% exhibits markedly inferior physical properties.

Metastable light-induced increases in the dark conductivities of a-SiSx:H alloys are explained as photo-activation of hydrogen-passivated sulfur donor sites. For a sulfur concentration (sulfur-to-silicon ratio) of 5.6 × 103 the excess dark conductivity as a function of illumination temperature is thermally activated with an activation energy of 0.72 eV. When the sulfur concentration is 3.3 × 102, the temperature dependence is very weak. This dramatic difference in the temperature dependence of the creation of increased dark conductivity is explained by a lowering of the annealing temperature for the metastable changes as the sulfur concentration increases. We discuss the influence of this new metastability on the possibility of obtaining more stable films.

We have deposited Si-nitride thin films by remote plasma-enhanced chemical-vapor deposition using different combinations of hydrogen and deuterium source gases. In one set of experiments, NH3 and SiH4 were injected downstream from a He plasma and the ratio of NH3 to SiH4 was adjusted so that deposited films contained IR-detectable bonded-H in SiN-H arrangements, but not in Si-H arrangements. Similar results were obtained using the same ND3 to SiD4 flow ratio; these films contained only SiN-D groups. However, films prepared from ND3 and SiH4 displayed both SiN-D and SiN-H groups in essentially equal concentrations establishing that H and D atoms bonded to N are derived from both source gases SiH (D) 4 and NH (D) 3, and further that inter-mixing of H and/or D atoms occurs at the growth surface. This reaction pathway is supported by additional studies in which films were grown from SD4 and ND3 with either i) He or ii) He/H2 mixtures being plasma excited. The films grown from the deuterated source gases without H2, displayed only SiN-D bands, whereas the films grown using the He/H2 mixture displayed both SiN-H and SiN-D bands. The total concentration of N-H and N-D bonds in the films grown from the He/H2 excitation was the same as the concentration of N-D, supporting the surface reaction model. In-situ mass spectrometry provides additional insights in the film deposition reactions.

The enthalpy (endothermic) of hydrogen evolution from p-type (boron doped) amorphous silicon with 17 at. % H was 4.8, 10.3, 15.8 and 17.3 kJ/g, the evolution temperature was 585, 606, 625 and 644 °C and the entropy of evolution was 5.6, 11.7, 17.5 and 18.9 J/g.K at heating rates of 5, 10, 20 and 30 °C/min respectively. That the enthalpy and entropy increased with heating rate means that the evolution involves not only Si-H bond breaking, but also Si-Si bond breaking and other defect formation. The Si-Si bond breaking and defect formation were enhanced at high heating rates, which caused high rates of hydrogen evolution. For n-type (phosphorous-doped) and intrinsic amorphous silicon with 25 and 23 at. % H respectively, the enthalpy and entropy of hydrogen evolution were higher than the p-type case, due to severe defect formation resulting from the higher hydrogen content. The activation energy of hydrogen evolution was 1.38, 2.5 and 4 kJ/g for the p-type, intrinsic and n-type materials respectively. Crystallization which occurred at temperatures higher than hydrogen evolution, was delayed for the amorphous silicon film in a higher disordered state after hydrogen evolution, suggesting that hydrogen evolution influenced the crystallization process.

Hydrogen bulk mobility plays an important role in determining a wide range of materials and electronic properties of hydrogenated amorphous silicon (a-Si:H). The existence of two types of hydrogen traps plays an important role in controlling hydrogen mobility in, and evolution of hydrogen from a-Si:H, however, theoretical and experimental literature values for the trap energetics vary considerably. We have developed a mean-field reaction-diffusion model which explicitly includes two trap states and realistic surface processes to model hydrogen evolution from a-Si:H. Modern numerical techniques were required to solve this challenging problem over the wide range of temperatures and concentrations encountered in typical hydrogen evolution experiments. The model is based on a number of experimentally established parameters. Comparison of our rigorous model with temperature programmed hydrogen evolution experiments provides a powerful method for characterizing the energetics, trap concentrations and diffusivity of hydrogen in a-Si:H.

We do not observe any immobile deuterium in secondary ion mass spectrometry D profiles taken after long anneals of hydrogenated amorphous silicon sandwich structures with a thin deuterated interior layer. This suggests that a single deep H level (∼1.4 eV deep) controls H diffusion. On its face, our result contradicts nuclear magnetic resonance and H effusion measurements that show about 30% of H in a-Si:H is “isolated” and deeply bound (∼ 2.1 eV deep). We reconcile our experimental results with the existence of isolated deep H by assuming there is a low-barrier (<< 1.4 eV) exchange process between free H and deep D. In tracer experiments, exchange has the effect of increasing the apparent emission rate of the deep D to nearly that of the shallowest trapped H. We solve for the D profiles and confirm that a deep-trapped D component is consistent with our D tracer profiles if and only if exchange processes are important. We also find that the mean distance D travels before retrapping (100–200Å) is determined by an exchange process of free D with trapped H.

PECVD silicon rich (x<4/3) and nitrogen rich (x>4/3) silicon nitride films have been deposited using a silane, ammonia, and nitrogen mixture at 400°C in a dual frequency parallel plate reactor. From FTIR measurements, most of the bonded hydrogen was observed to be present as N-H in the nitrogen rich films and as Si-H in the silicon rich films. In this study, the movement of hydrogen in the various nitride films as a result of Rapid Thermal Annealing (RTA) in N2 for 30 seconds has been investigated. While the intensity of the N-H stretching band was always found to decrease from that measured for the as-deposited film, the intensity of the Si-H stretching band was observed to increase for anneal temperatures up to 650°C. During the anneal process some hydrogen effusion takes place and some of hydrogen changes its bonding to silicon atoms from previous nitrogen atoms in the film. Consistent with the FTIR results, Hydrogen Forward Scattering (HFS), shows that only a small amount of hydrogen effuses from the silicon nitride film as a result of the RTA process below 650°C.

We report results of a search for a unifying rate law for the annealing of metastable defects in hydrogenated amorphous silicon (a-Si:H). We tested the hypothesis that defect-annealing by both heating or illumination is driven by the density of free electrons. This hypothesis is formulated via the rate equation - dN/dt = A nα N f (T), where N is the defect density, t the time, A a constant, n the free electron density, and f (T) a function of temperature derived from a distribution of annealing energies. The model fits two sets of data, with light-intensity and electrical conductivity as the independent variables, reasonably well, with a ranging from 0.39 to 0.76, but not the third set, where we varied the temperature.

Several series of amorphous silicon nitride thin films have been grown by plasma-enhanced chemical vapour deposition, where the ratio of ammonia and silane feed gases was held constant for each series while the deposition temperature was varied from 160 °C to 550 °C, and all other deposition conditions were held constant. Photothermal Deflection Spectroscopy measurements were used to determine the Urbach slope E0 and the defect density ND. It is found that ND is determined by E0 for most of these samples, suggesting that defect equilibration occurs in a-SiNx:H for x up to at least 0.6. The growth temperature at which the disorder is minimised increases to higher values with increasing x, which is explained in terms of a hydrogen-mediated bond equilibration reaction. Fourier Transform Infra Red spectroscopy measurements were performed to determine the changes in hydrogen bonding with growth temperature. The results suggest that a second bond equilibration reaction also occurs at the growing surface, but that equilibrium cannot be reached at higher temperatures because of hydrogen evolution from Si-H bonds.

The stability of amorphous silicon-based solar cells is related to the i-layer carrier mobilities as these determine the carrier densities under illumination. Materials with high mobilities quickly diffuse the carriers to high recombination regions such as the contacts and interface regions. The quick out diffusion of carriers lowers the i-layer carrier densities and therefore the driving force for defect generation. Increased compressive stress appears to be correlated to increased electron mobility and is related to the hydrogen content. Solar cell configurations also play a role in the i-layer carrier density. For example, factors that increase the V∝ tend to decrease the stability because the associated decrease in interface recombination results in increased i-layer carrier densities. Factors that reduce the blue response such as increased interface recombination tend to increase the stability as these results in reduced i-layer carrier densities.

Hydrogenated amorphous silicon (a-Si:H) solar cells are prepared by plasma enhanced chemical vapor deposition (PECVD). Before quenching the solar cells, the short circuit current (Jsc), open circuit voltage (Voc), fill factor (F. F.) and conversion efficiency (η) are 17.79 mA/cm2, 0.79 V, 53.29, and 7.49 %, respectively. After thermal quenchine the solar cells from 200°C, Jsc, Voc, F. F., and η are 18.64 mA/cm2, 0.8 V, 53.79, and 8.02 %, respectively. We investigated the thermal equilibrium processes of each P, I, and N layers. Also, we obtained the dark current-voltage characteristics of a-Si:H solar cells before and after quenching. We analyze the results in terms of the change of the internal electric field in a-Si:H solar cells, caused by the shift of the Fermi level of P layer toward valence band.

The influences of sub-gap illumination and thermal annealing on the light-induced neutral dangling bonds (DB's) created in the fast process (FDB's) and those created in the slow process (SDB's) are prominently different in Si-rich a-Si1-xNx:H. The results of photobleaching show that the sub-gap illumination annihilates FDB's, more effectively at 77 K, whereas it appears not to affect SDB's, suggesting that FDB's are produced by a carrier capture at preexisting charged DB's. The annealing behaviors for both FDB's and SDB's are in agreement with the monomolecular kinetics. The attempt-to-anneal frequency, ν0, is ∼ 103 Hz for FDB's and ∼ 107 Hz for SDB's. Annealing-time dependence of the density of light-induced spins at various annealing temperatures can be fit with a distribution of annealing activation energies. In addition, clearly separated peaks of the distribution of annealing activation energies related to FDB's and SDB's are observed. A comparison indicates that SDB's have a nature similar to the light-induced metastable spins in a-Si:H, which are likely to be produced by a creation of new defects.

We report the results of a study of metastable defect creation by pulsed light soaking in undoped hydrogenated amorphous silicon (a-Si:H). An illumination time dependence of the defect density, a saturated defect density, and light-induced annealing under pulsed laser light have been studied. Measurements show approximately a t1/2 time-dependence of the defect creation, which is independent of light intensity. It is observed that the saturation value of the defect density is about one order of magnitude higher than by cw illumination in device quality films. It has been suggested that these results would be due to the difference in the light-induced defect annealing rate between cw and pulsed lights, in which it is found that the light-induced annealing rate by pulsed light is lower than by cw light.