The energy distribution of localized states above the Fermi level in undoped a-Si:H and a-SiGe:H has been determined by phase shift analysis of modulated photocurrents. (1) A peak in the DOS with 0.56 – 0.65 eV activation energy has been found, reflecting the D− -state of isolated dangling bonds. A second peak with 0.35 eV activation energy has been detected which is attributed to the T3+-state of T3+-T3− -pairs or to the antibonding state of weak Si-Si bonds, respectively. Strong illumination raises the lower peak and quenches the upper one supporting a shallow state - deep state conversion model. (2) From temperature-dependent measurements a shift of the dominant electron transport path into the tail with decreasing temperature - associated with tail state hopping - has been obtained.

Amorphous-crystalline heterostructures composed of sub-micron thick undoped amorphous hydrogenated silicon deposited on lightly doped n-type crystalline silicon substrates have been studied by capacitance techniques. Capacitance vs. temperature scans along with model calculations on these samples are used to investigate the amorphous-crystalline silicon interface region. We deduce the existence of an anomalously defective a-Si:H region extending roughly 1000 Angstroms from the growth interface. We also deduce considerable fluctuations in the interface potential which indicate lateral variations of ±20% in the defect distritribution over a 1000 Angstrom scale.

High-quality p-type a-SiC films were prepared using a new doping gas, B(CH3)3, instead of B2H6. a-SiC films deposited with B(CH3 )1 show high photoluminescence intensity and low defect density. Doping efficiency was also improved by usiyg B(CH3 )3. A module efficiency of more than 9% was achieved for a 100 cm2 a-Si solar cell whose p-layer was doped with B(CH3 )3. A new plasma CVD method, called controlled plasma magnetron (CPM) method was developed to realize high electron-density plasma with low damage to substrate. High conductive p-type μc(microcrystalline)- SiC films were prepared by CPM method with B(CH3 )3. CPM method was found to be suitable not only for preparation of μc films but for high-speed deposition.

Over the past twelve years large strides have been made in improving the conversion efficiency of amorphous silicon based solar cells from 2.4% to 13%. The history and status of the material and device developments that have led to this five-fold improvement in the conversion efficiency are reviewed. Prospects for future improvements are discussed.

Although single junction a-Si:H p-i-n solar cells with a conversion efficiency of about 12% have been demonstrated [1], the photo response of devices is limited by several factors. At short wavelengths, the most obvious limitation is imposed by the optical bandgap of the topmost doped layer. However further limitations to the short wavelength response have also been noted. Homojunction p-i-n devices of a-Si:H suffer from poor short wavelength response owing not only to high optical absorption in the p-layer, but also recombination at the p/i interface. The latter effect is enhanced due to the “back diffusion” of electrons towards the p/i interface and recombination that further reduces the short wavelength response. The short wavelength response of devices is improved by the use of a heterojunction front contact such as p-type a-Si1-x Cx :H. The resulting conduction band discontinuously reduces in part the back diffusion and recombination and reduces optical losses. Despite these improvements, however, the short wavelength response of p-i-n devices is frequently lower than expected on the basis of optical absorption within the topmost layers. Evidence for these losses comes from measurement of devices in reverse bias, where improvements in the short wavelength response are indicative of recombination losses. For example, Figure 1 compares the normalized photoresponse of two a-Si1-x Cx p-i-n devices. The photoresponse is normalized by taking the ratio of the quantum efficiency at -3V and zero bias at each wavelength. Cells with a photoresponse ratio greater than unity have additional, nonoptical losses. In Figure 1 the “normal” devices is a standard a-Si1-x Cx :H p-i-n solar cell while the graded interface device has a silicon-carbon alloy grading extending several hundred Angstroms away from the p/i interface. It is clear from the photoresponse data that recombination lowers the short wavelength response of the ungraded device.

We present data of materials and devices produced from GSI's continuous mass production system for large area solar cells. We also discuss our recent development of a multichamber system which is capable of producing uniform a-Si:H layers over areas of 2 m2.

Procedures for analyzing hydrogenated amorphous silicon p-i-n solar cells are presented. Two closed form expressions, each containing a single unknown i layer transport parameter, are adequate to describe the light-intensity and temperature dependence of the current-voltage curve and photocapacitance for the majority of situations. Excellent agreement between theory and experiment is obtained.

We have theoretically studied the role of the doped contacts on the open circuit voltage (Voc) of amorphous silicon, pin solar cells by numerical solution of coupled Poisson and current continuity equations for the bulk and the doped regions. The built-in potential (Vbi) in the cell was found to be split asymmetrically between the p+/intrinsic and i/n+ regions. The first one forms a p+/n junction (dominant junction) since the undoped intrinsic is slightly n-type; the latter forms a n/n+ or low/high junction.

We found that Voc is determined by the built-in potential at the dominant junction (and hence by the conductivity activation energy of the p+ layer) and is fairly insensitive to the activation energy in the n+ region. If one dopes the intrinsic with phosphorus, Voc is even less dependent on the n+/i junction potential but becomes very sensitive to the activation energy in the p+ layer. However, when the i-layer becomes p-type by boron doping, the dominant junction is created at the n+/i interface and Voc then becomes insensitive to the quality of the p+ layer and is determined by the n+ contact potential. The results are interpreted by considering the modification of the electric field within the bulk by the junction potentials and the recombination mechanisms at the two interfaces.

Lateral photovoltages were measured in a 20 cm long strip of hydrogenated amorphous silicon deposited on a platinum foil. The position dependence was nonlinear. It was antisymmetric with respect to the center of the strip when the platinum substrate was electrically floating. If the platinum was grounded, the photovoltaic signal was zero at one end of the sample and increased monotonically to the other. The magnitude of the photovoltage was linearly dependent on the light intensity. The polarity of the photosignals when the sample was exposed to a flash of light at one position of its length was opposite to that of the semi-steady photovoltage when it was exposed to a steady light source. The phenomena can be explained on the basis of a model of charge carrier separation in the Schottky barrier junction at the point of illumination.

Both the optical and transport properties of a-Si: H have been suitably parametrized as a function of Urbach's tail logarithmic slope Eo, surface density of states Ns and optical absorption normalization parameter αo, in order to reduce the number of parameters used for p-i-n solar cell S, simulation.

The transient photovoltaic effect has been simulated by a Monte Carlo multitrapping computer model in order to investigate the possibility of its use for p-i-n solar cell diagnostics.

Amorphous silicon (a-Si:H) solar cells based on p.i. n junctions have been produced by a Two Consecutive Decomposition and Deposition Chamber (TCDDC) system [1], either using Wide Band Gap-μc (WBG-μc) films or normal doped ones, as contact layers. The role of such doped layers on device's performance (open circuit voltage, Voc, short circuit current density, Jsc, fill factor, FF, and conversion efficiency, η) have been determined by spectral response and I-V curve measurements. The obtained data show that Voc is enhanced over more than 25% by using WBG-μc films with thicknesses bellow 300 Å as doped contacts, when compared with normal p.i.n junctions [2]. Besides this, if a smooth grade interlayer, based on a-Si:C doped alloys, is added between junctions [3], Voc higher than 1.0 V with η ≥7.5% are obtained in single structures.

We survey experimental evidence, obtained by optoelectronic measurements, and theoretical models for the density of gap states in a-Si,Ge:H(F) alloys. This survey shows that increasingly reproducible densities are obtained in the various segments of the energy gap. The results are becoming understood quantitatively. It is not yet clear if the reproducibility signals that the best material has been reached. The electron transport properties of low-gap (1.2 eV < Eopt < 1.4 eV) alloys are adequate for solar cells. Improvement of the hole transport properties, particularly (μr) which is set by the valence band tail width and the density of deep defects, is achieved through a reduction of the Urbach energy.

Paramagnetic defects in the mobility gap of doped a-Si:Ge:H alloys with a Ge-content of 20% and 30% are studied by electron spin resonance (ESR and LESR). The singly occupied states of Si- and Ge-dangling bonds are found to be centered around midgap with no detectable difference in energy position. The ESR-spectra exhibit at least two paramagnetic resonances in both n- and p-type samples which are attributable to singly occupied tail states. At a given distance of the Fermi level from the mobility edge of the conduction band the density of states is much higher in the alloys than in a-Si:H. In phosphorus doped material hyperfine structure is observed which arises from two different phosphorus states: fourfold coordinated phosphorus (P°4).and twofold coordinated phosphorus (P°2).

Photothermal deflection spectroscopy, infrared, and Raman measurements are presented for a series of a-SiC:H samples containing different amounts of microstructure. The Urbach edge is seen to widen systematically as the film microstructure increases, but the width of the Raman transverse optical mode remains unchanged. We argue that the increase in Urbach edge width with increasing C content is due entirely to the increase in microstructure. In addition, from the invariance of the Raman measurements with increasing microstructure and the assumed sensitivity of the Raman technique to surface states on the interior surfaces of these microstructural features, we are able to place lower limits on the size of these features. Such sizes are corroborated by preliminary small angle xray scattering measurements on identically prepared samples.

A novel technique has been proposed for determining the density-of-state (DOS) distribution in the mobility gap of highly resistive amorphous semiconductors, using amorphous/crystalline heterojunction structures. This technique has been tested and applied on undoped a-Si:H and a-SiGe:H films, covering the optical gap (E0 ) range of 1.30 to 1.76 eV. For undoped a-Si:H with E0 =1.76 eV, the peak of the midgap DOS distribution has been locatedst 0.85 eV below the conduction band edge, EC , with a value of 5.6×1015 cm-3 eV-1 . For undoped a-SiGe:H (E0 =1.55 eV) the same has been obtained 0.71 eV below EC with a magnitude of 7.9×1016 cm-3 eV-1. Those midgap states have been found to be correlated with singly-occupied dangling bonds.

The effects of substrate temperature and hydrogen dilution on mobilitylifetime product and hydrogen bonding are reported for amorphous silicongermanium films with bandgaps from 1.1 to 1.74 eV deposited by Hg-sensitized photo-CVD. The ratio of dihydride to monohydride bonding and preferential H attachment to silicon both decreased with hydrogen dilution. The best films were deposited with hydrogen dilution at substrate temperatures between 230 and 250 ° C. Model equations used to calculate radical fluxes to the substrate are described.

We present a detailed study of the electron transport parallel to the layers in a a-Si,Ge:H,F/a- Si:H,F amorphous semiconductor multilayers. The optical gap of the well layer is shifted from the bulk alloy due to quantum confinement. This shift W can be modeled using a finite barrier height, and electron effective masses for the barrier and for the well.

Multilayer-induced changes in σd and its activation energy Ea,d are completely determined by W. The photoconductivity σph, its temperature dependence (Eaph) and its dependence γ on carrier generation rate reflect a stretching of the conduction band tail near the band edge as the quantum shift W increases.

We have carried out combined measurements of dark ESR, light-induced ESR and low photon energy absorption by Constant Photocurrent Method (CPM) for various thickness a-Si:H, a-Si1-x-.Nx Cx.:H and a-Si1-x-.Nx:H films. As a result, both the surface density of dangling bonds in the disordered surface layer and the density of dangling bonds in the bulk region were obtained discriminating the neuttal and charged dangling bonds. In a-Si:H, it is found that a large fraction of dangling bonds is charged ones. The incorporation of C decreases and the incorporation of N increases the fraction of charged ones. It is concluded from the present result that charged defects in a-Si1-xNx,:H are pairs of N+4and Si-3. It is possible that charged defects in a-Si:H originate from a small amount of N contamination.

Different properties of hydrogenated amorphous silicon nitride films, prepared by various deposition techniques, are reported in this paper. Special attention has been devoted to the influence of film stoichiometry and hydrogen content on these properties. It has been established that the hydrogen can influence the physical, chemical, mechanical, and magnetic properties of the N-rich film. However, for a given film stoichiometry there is no clear evidence for the hydrogen content influence on optical and electrical properties; the film stoichiometry provides the main influence on these properties. The best quality gate dielectric silicon nitride is N-rich film deposited at temperatures exceeding 350°C.

Ambipolar diffusion lengths in a-SiGe:H and a-SiC:H have been analyzed by Steady State Photocarrier Grating Technique. Diffusion lengths and photoconductivity of a-Si:H are considerably affected by alloying. Photoconductivity in a-SiGe:H can be improved by special deposition methods, diffusion lengths, however turn out to remain nearly unchanged. The comparison of diffusion lengths and -ho~oconductivity yields hole mobilities in a-Si1-xGex:H of 10-2cm2 /Vs for O≤×≤O.2. For a-Si:H changes in photoconductivity by generation of defects (light soaking) result in noticeable changes in diffusion lengths, whereas different photoconductivities in a-SiGe:H caused by different deposition methods end up in the same diffusion lengths. Consequently the improvement of a-SiGe:H photoconductivity by changes in deposition condition is by far a much more complex process than only decrease of density of midgap states.