An IR and multiple quantum NMR (MQNMR) study of hydrogen microstructure in three boron doped a-Si:H is discussed. The total Si-bonded H content of all films was 6.5 ± 1.0 at.% as determined by the 640 cm-1 IR wagging mode, but their boron content, which was determined by secondary ion mass spectrometry, ranged from 0.02 to 0.3 at. %. The number of correlated hydrogen, as measured at a preparation time of 600 μSwas found to be more weakly dependent on the boron content than previously observed in phosphorous-doped glow-discharge films. Upon annealing at 220 °C the MQNMR spectrum show a moderate increase in the number of correlated hydrogen in all three samples.

A series of hydrogenated amorphous silicon (a-Si:H) films was deposited by rf glow-discharge deposition using various processing conditions. We have studied microstructure in the films by means of infrared absorption spectroscopy. Small-angle X-ray scattering measurements were used to determine the microvoid fractions of a few selected samples. Our results show that both the void fraction and the amount of microstructure can be varied either by changing the substrate temperature or by H2 dilution. Bond-angle variation in the films was probed by Raman scattering measurements. The Raman data indicate that the substrate temperature is the main variable that determines the bond-angle variation. We conclude that the presence of microvoids in a-Si:H does not influence the structural disorder of the amorphous matrix surrounding the voids. Our results are in agreement with experimental work on microvoids in a-Si1-xCx:H, and model calculations on voids in a-Si.

A SAXS and IR study of microvoid distribution and dynamics in a-Si:H deposited by rf sputtering at 200 – 600 W on nominally unheated substrates is described and discussed. The 640 cm-1 band of the 200 W film yielded a total Si-bonded H content CH=21 at. %; the 840 – 890 cm-1 band yielded a dihydride content CH2 3.4 at. %. The SAXS measurements yielded a microvoid volume fraction vf=8.2 vol.%, and tilting SAXS data indicated elongated voids consistent with a columnar microstructure. In the other films, 9<CH<12 at. % and CH2 was negligible, vf was -2 vol. %. Annealing from 250°C to 310°C for 6 hrs resulted basically in no changes of CH and vf. However CH decreased and vf increased significantly after annealing at 350°C and above. The results showed a strong correlation between the IR determined CH and CH2 and the SAXS determined vf.

We have performed ab initio calculations and determined the bond-energies and vibrational frequencies of Si-H groups that are: i) attached to Si-atoms as their immediate, and also more distant neighbors; and ii) attached to three O-atoms as their immediate neighbors, but are connected to an all Si-atom matrix. These arrangements simulate bonding geometries on Si surfaces, and the calculated frequency for i) is in good agreement with that of an Si-H group on an Si surface. To compare these results with a-Si:H alloys it is necessary to take into account an additional factor: the effective dielectric constant of the host. We show how to do this, demonstrating the way results of the ab initio calculations should then be compared with experimental data.

The dipolar interaction of hydrogen (∼ 10 at. %) in boron-doped amorphous silicon has been studied using the Jeener-Broekaert pulse sequence. The sample was prepared on an Al foil substrate at a temperature of 230°C using a standard glow discharge system (operating at 1 W rf power). The diborane/silane ration was 10-4. The same sample was removed from the Al substrate using dilute hydrochloric acid. The Jeener-Broekaert pulse sequence consists of three pulses: π/2|x′ – τ1 – π/4|y′ – τ2 – π/4|y′ – echo. We measured T1D, the dipolar spin lattice relaxation time, for τ1 = 82 μs, τ1 = 100 μs and τ1 = 50 μs at 299°K. The value of τ1D was found to be independent of τ1. At 335°K we found τ1D to be much longer than at room temperature. The values of τ1D at 299 K, 314 K and 335 K are, respectively, 0.7 ms, 1.2 ms and 1.8 ms. From the data we estimate an activation energy for microscopic motion to be ∼ 0.2 eV.

The aim of this work is to achieve detailed information about surface and bulk density of states in device quality a-Si:H films by means of deconvolution of PDS spectra. Unambiguous deconvolution of the spectra has been performed by calculating the first derivative vs. energy of the absorption coefficient of the samples. The results are compared with those obtained by means of the integration rule. Calculations of the density of states carried out on samples having different thickness suggest that the defect distribution differs from surface to bulk. The surface defect peaks are found to lie closer to the corresponding mobility edges, with respect to the bulk defect peaks: these results agree well with those obtained by means of other techniques, like total yield spectroscopy and can be explained in terms of a defect pool model.

The present deconvolution procedure when applied to samples having different thickness provides new results about the film's homogeneity in terms of deep defect and valence tail states.

We have calclulated the ESR hyperfine parameters of threefold-coordinated Si atoms and twofold-coordinated P and N atoms in Si-based amorphous semiconductors using the density functional theory with a local-spin-density approximation. These calculated results have been compared with the observed ESR results.

The microstructure of amorphous silicon-based alloys prepared by reactive magnetron sputter (RMS) deposition has been examined on a scale from 1 to 25 nm using small-angle x-ray scattering (SAXS) and compared to that from films prepared by standard glow-discharge (GD) technology. Device-quality RMS a-Si:H material is found to have a larger microvoid fraction than device-quality GD a-Si:H and to have a significantly different size distribution. Addition of by the RMS technique using CH produces enhanced SAXS similar to the introduction of C via CH4 in GD material. A significant difference in the SAXS from RMS a-Si:H and a-SiC:H films compared to GD films is the observation of some oriented microstructure, most likely columnar in nature. Flotation density measurements of the same films examined by SAXS support the assumption that the SAXS originates primarily from microvoids.

The effects of H concentration on H transport has been measured by deuteration of H depleted samples which were re-hydrogenated at various concentrations and on samples subjected to various annealing protocols. A remarkable observation is that the trap density appears to increase with H concentration. Extended thermal annealing appears to increase the trap depth but does not markedly alter the number of traps. The results suggest that H introduced from the plasma is less strongly bound than H residing in the film during annealing. Nucleation and growth of H clusters is introduced as a possible model accounting for these transport results.

Relaxation data provide evidence that thermally generated defects with spin observed in thick, undoped hydrogenated amorphous silicon (a-Si:H) films can be stabilized with alternate structural configurations. Some quenched-in metastable spins, equilibrating further at a lower temperature, relax and become more resistant to annealing. Some of these defects take longer to anneal than any of the defects equilibrated only at the higher temperature. An obvious crossing of annealing data curves occurs on one of two differently prepared films. Crossover has important physical implications in deciding between defect models.

Such an observation necessitates the use of defect models with more complexity than variations based on two-level energy-configuration coordinate diagrams. Long annealing times may not be controlled simply by high energy barriers; the annealing of a spin may require many dispersive steps with smaller energy barriers. Models that stablize a spin by either hierarchically constrained defect configurations or the trapping of a mobile hydrogen in a distribution of sites can be used to explain this result. The fact that crossover is more obvious on the sample that has an oriented microstructure implicates film structure.

We present an atomistic and quantum mechanical model of light-induced defects (the Staebler-Wronski effect). The model is based in part on our observations of molecular dynamics simulations with an ab initio code and requires a change in the charge of a well localized state in the gap, such as a dangling bond, to nucleate the new defects. Besides the new defects, a substantial rearrangement of the supercell is observed.

A physical interpretation is given for the widespread observation of stretched-exponential time dependences for both light-induced generation and thermal anneal behavior of a-Si:H. This interpretation is based on a distribution of simultaneous rate processes, each of which represents a subset of normal metastable defects having a simple exponential behavior; it is thus an adaptation of one used for relaxation in glasses. The distribution can be evaluated from an observed stretched exponential, and for the case of thermal anneal with activated time constants, the stretch parameter should increase with temperature quite generally. The shape and width of the distributions of time constants and activation energies are inferred from these relations and properties of observed stretched exponentials.

Light-induced changes in electron spin resonance (ESR) signal intensity and photoluminescence (PL) intensity in device quality a-Si:H are compared. Different dependences on the light soaking temperature for the ESR spin density and the PL peak intensity at ∼ 1.3 eV suggest that a non-radiative process, that does not directly involve neutral dangling bonds, influences the photoluminescence at the peak. After light soaking no clear increase of the peak in the low energy luminescence region (∼ 0.8 eV) was observed even at higher temperatures (> 200 K), where the low energy luminescence band is relatively more important.

We provide experimental evidence for the fact that hydrogen diffusion in hydroge-nated amorphous silicon is controlled by the concentration of electronic carriers. It is experimentally demonstrated that the hydrogen diffusion coefficient (a) is enhanced if the carrier population is increased by illumination and (b) is strongly suppressed if carriers are extracted from the diffusion region by the application of an electric field.

Transient light-induced electron spin resonance (LESR) at 120 K, has been used to investigate deep defects in a-Si:H through changes in the lineshape. When the lineshape is deconvoluted into narrow and broad components, the narrow component is found to decrease, relative to the broad component, with increasing light-soaking time. Similar changes are not observed, however, in as-deposited or annealed films regardless of deposition and annealing conditions. An important role for charged dangling bonds is proposed to explain these changes and we suggest that intrinsic (stable) and light-induced (metastable) defects play different roles in transient-LESR and may occupy different energy distributions in the gap.

The light induced degradation of a-Si:H p-i-n solar cells under electrical bias has been systematically studied. By comparing the results with the light intensity dependence of cell degradation under open circuit condition, we show that the only recombination mechanism, which can be consistent with the experimental data in both cases, is based on the bimolecular recombination between a free hole and a trapped electron at the “weak” bond site. Other possibilities for defect creation are also pointed out.

We describe a new technique for the deposition of a-Si:H, in which the growing film surface is periodically irradiated with high-intensity xenon light pulses. The films show high stability of the photoconductivity under illumination, but also exhibit a high dark conductivity, so that the photosensitivity is reduced. Infrared and Raman scattering measurements indicate little difference in the Si-H and Si-Si bonding structures compared to reference films; however, there is a large increase in the sub-bandgap absorption in addition to a shift of the Fermi level, implying some subtle modification of the film growth process caused by the irradiation. Results on the preparation and stability testing of solar cells fabricated using this technique are also described.

The spin dynamics of deuteron magnetic resonance (DMR) components have been studied in deuterated plasma-deposited thin films of a-Si, Ge and SiGe. Transient recoveries of perturbed deuteron magnetization components evolve via magnetization transport in the limit of large inhomogeneity. The evolution is well-described by an error function expression whose parameters correlate with the film photovoltaic quality as measured by the photoresponse ημτ and the photosensitivity, ΔI/Id. These relaxation results are correlated with our continuing DMR measurements of reversible metastable photo-induced rearrangements in high-quality a-Si:D,H films. These studies indicate that about 0.5% of the hydrogens interact with a light-induced defect.

The thermodynamic equilibrium framework first presented at the Spring ′91 MRS meeting is refined and applied. The effect of temperature on band gap is added, resulting in a larger estimate of the loss in entropy associated with dangling bond formation. The magnitude of the entropy loss is consistent with a structural rearrangement. At the temperatures of interest, dangling bond defect formation is exothermic with negative entropy and free energy changes. The thermodynamic framework was used to address some practical issues including the estimation of the saturated dangling bond density resulting from low temperature light soaking and the electronic energy levels of the various dangling bond charge states.

We observe reversible quenched-in metastability in the dark conductivity of extremely low-H-content amorphous silicon (a-Si). The sample is sputtered, P-implanted a-Si containing 0.1 at.% of H. Except for the high value of its equilibration temperature (T*), 355° ± 20°C, this metastability is similar to that observed in doped hydrogenated a-Si (a-Si:H). The ∼10 at.% H present in a-Si:H is not essential for a-Si metastability. After hydrogenation with about 10 at.% of H, T* falls to 175° ± 10°C. We propose that higher H content reduces T* by increasing Si network flexibility.