Microcrystalline silicon (μc-Si:H) intrinsic layer for application in micromorph tandem photovoltaic solar cells has to be optimized in order to achieve cost-effective mass production of solar cells in large area radio frequency capacitively-coupled PECVD reactors. The optimization has to be performed with regard to the deposition rate as well as to the crystallinity uniformity over the substrate area. This last prerequisite is difficult to achieve since the optimal solar grade μc-Si:H is deposited at the limit between a-Si:H and μc-Si:H material, where the film crystallinity is very sensitive to the plasma process.In this work, a controlled RF power nonuniformity was generated in a large area industrial reactor. The resulting film uniformity was studied as a function of the deposition regimes. Results show that the higher the input silane concentration, the more the uniformity of the crystallinity is sensitive to the RF power nonuniformity for films deposited at the limit between a-Si:H and μc-Si:H. The effect of the input silane concentration on the microstructure uniformity could be explained on the basis of an analytical plasma chemistry model.It has been recently shown that the plasma composition of SiH₄-H₂ discharges plays a major role in determining the silicon thin film crystallinity, and that the lower the silane concentration in the plasma, the higher the crystallinity [1]. The silane concentration is determined not only by the input silane concentration, but also by the fraction of silane dissociated in the plasma. Therefore, the nonuniform silane dissociation induced by the RF power distribution profile makes the silane concentration in the plasma nonuniform over the substrate area. This effect is much more pronounced as the input silane concentration is increased, since at high silane concentrations the plasma composition is mostly determined by the silane dissociation fraction, whereas at low silane concentrations it is mostly dominated by the input silane concentration [1].This result is important for reactor design. In reactors generating nonuniform plasma the input silane concentration has to be limited to low values – where the deposition is inefficient [2] – in order to deposit films with uniform microstructure. If we want to benefit from high deposition efficiency encountered only for higher silane concentrations [2], the RF power distribution has to be as uniform as possible over the whole substrate area. This means that the standing-wave effect in large area parallel plate radio frequency reactors has to be accurately compensated and that the plasma and the thick dielectric substrate have to be taken into account for the calculation of the correct shape of the electrode to balance the standing-wave nonuniformity [3].[1] B. Strahm et al, Plasma Sources Sci. Techol. 16 (2007) 80-89.[2] B. Strahm et al, Solar Energy Mater. Solar Cells 91 (2007) 495-502.[3] L. Sansonnens et al, Appl. Phys. Lett. 82 (2003) 182-184.

Epitaxial silicon thick films, reaching 25 µm in thickness in 10 min, have been deposited with no trace of epitaxial breakdown at low temperature by mesoplasma chemical vapor deposition. The deposition rate increases linearly with the silane gas flow rates while it remains as high as 40 nm/s even if the substrate temperature decreases as low as 360C. It should be further noted that Hall mobilities of such epitaxial films was maintained around 280 cm2/V-s independently of the deposition rates and substrate temperatures. These unique qualities could be attributed to the specific characteristics of mesoplasma where one can expect simultaneous attainment of low gas and electron temperatures and high flux transport of growth precursors on the plasma flow. In particular, clusters formed through rapid vapor condensation within the thermal boundary layer between the plasma and substrate was identified to be in a ~2 nm spherical form with loosely-bound structure by in-situ small angle x-ray scattering. Taking account of the growth morphology of the films, the difference in the cluster characteristics between the epitaxial and polycrystalline film conditions suggests the thermally-activated nano-sized silicon clusters possibly facilitates fast rate surface migration of the constituent silicon atoms on the film, despite low temperature and adhesive growth mode.

Our ‘cone kinetics’ model, which explains development of cone-shaped inclusions during nanocrystalline silicon film growth and during low-temperature silicon epitaxy breakdown, is extended to protocrystalline (‘edge’) amorphous silicon, Si heterojunctions and other Si film morpholologies. We generalize the physics underlying cone formation and present a diagram that delineates the many deposition regimes giving rise to different film morphologies; these regimes are determined by by the nucleation rate of the second phase and the relative growth rate of the phases present. The model predicts cone growth during thin-film deposition by plasma-enhanced and other chemical vapor deposition techniques when there is 1) isolated nucleation of a second phase which grows faster than the first phase and 2) isotropic growth. Experimental measurements are consistent with the morphologies predicted by the model for both nanocrystalline silicon cone formation and low-T epitaxy failure. Protocrystalline amorphous silicon and other embedded crystallite forms are understood as nucleation of a slightly slower-growing phase. The cone kinetics phase diagram explains simply the observations by Vallat-Sauvain et al. [J. Appl. Phys. 87, 3137 (2000)] and other groups of continuously varying nanocrystalline film morphology with increasing H-dilution of silane precursors. In this paper, we describe measurements of several film morphologies, present the general deposition phase diagram for several substrates, and discuss factors that control the morphology during growth. We thank Baojie Yan and Jeffrey Yang of United Solar Ovonics LLC for providing several samples and acknowledge support from the U.S. DOE under Contract DE-AC36-99GO10337.

To realize high rate deposition of high quality microcrystalline silicon (µc-Si:H) thin films for cost-effective solar cell applications, a microwave induced high density plasma source has been developed. By optimizing plasma conditions, highly crystallized µc-Si:H films can be deposited from SiH4+He mixture on quartz fiber at an ultrafast rate higher than 700 nm/s. The µc-Si:H growth process at such high deposition rates, however, still remains ambiguous. In this work, we investigate the evolution of film microstructure during the high-rate deposition process to get insight into the film deposition process. The film depositions using the microwave source were performed at a working pressure of 2 Torr. We employed SiH4 flow rate of 5~15 sccm with a fixed He flow rate of 400 sccm for the depositions. To observe the microstructure evolution, the deposition time was varied from 5 s to 40s, leading to the thickness of 0.5 ~ 10 µm. Our results showed that both film deposition rate and Raman crystallinity increased with increasing the film thickness. In addition, ESR spin density increased with the thickness while H concentration in the films decreased. To confirm the evolution of crystallinity along the growth direction, reactive ion etching was performed for the ~6 µm-thick film together with Raman measurement after each etching turn. From this experiment, we found that crystallinity of a Si film formed initially near substrate was enhanced significantly during the thick film formation. Other results also indicated that there is an annealing effect during the deposition process. A possible mechanism, the annealing assisted chemical vapor deposition, is proposed to describe the observed results. Moreover, optical emission spectroscopy showed that the discharging gas had a relatively high temperature, which may be the main parameter responsible for the annealing effect.

Mixed phase amorphous-crystal (a-c) silicon systems have a great technological relevance, ranging from microelectronics, to optoelectronics and photovoltaics. Such systems undergo a large number of process steps during which the amorphous phase can spontaneously recrystallize. In general, the transformation kinetics is controlled by the mobility of the a-c boundaries. For example, in the case of a planar a-c boundary, the recrystallization gives rise to solid phase epitaxy (SPE), a technique routinely used in microelectronics in order to incorporate high concentration of dopants into crystalline silicon. On the other hand, curved a-c boundaries are rather observed during the solid phase crystallization (SPC) of amorphous silicon. In this case, c-Si grains nucleate and grow within the amorphous phase, eventually transforming into nano- or polycrystalline silicon, with applications in photovoltaics and optoelectronics. Molecular dynamics simulations represent a powerful tool to understand the microstructure evolution in such complex, two-phase systems. Since the kinetics mainly result from a trade-off between orientational, strain effects and the defects at the interface, this approach is promising in order to validate alternative, sometimes conflicting models of microstructure evolution and to improve the process engineering of these materials.In this work we report on a joint effort to properly model the a-c boundary mobility by means of atomistic simulations based on model potentials. Firstly, by focusing on the case of a planar amorphous-crystalline interface we present a comprehensive study of the available model potentials for solid phase epitaxy. The analysis takes into account several important physical properties such as the velocity of the interface, the temperature dependency, the structural properties of the amorphous phase. Examples are discussed in the case of pure silicon [1] and boron doped a-c systems[2].Secondly, we describe the case of curved a-c boundaries as they occur during the recrystallization of nano-sized crystalline grains embedded into the amorphous phase. We prove that the mobility of the grain a-c boundaries depends on the actual grain size and, as a result, the grain radius evolution turns out to be nonuniform. A simple power-law growth model, with an exponent depending on the temperature, is able to describe the atomistic data in the whole range of temperatures considered[3].[1] C. Krzeminski, Q. Brulin, V. Cuny, E. Lecat, E. Lampin and F. Cleri, J. Appl. Phys. 101, 123506 (2007)[2] A. Mattoni and L. Colombo, Phys. Rev. B 69, 045204 (2004)[3] A. Mattoni and L. Colombo, Physical Review Letters, (2007) in press