T.H. Wang

Growth of silicon thin layers on cast MGSi from metal solutions for solar cells

Abstract In pursuit of device-quality layer formation on cast, metallurgical-grade silicon (MG Si) substrates for solar cells, the growth kinetics of silicon liquid phase epitaxy (LPE) from metal solutions was studied. We found an ideal solvent system, Cu Al, for growth of Si layers with thicknesses of tens of microns on cast MG Si substrates by LPE at temperatures near 900°C. This solvent system utilizes Al to ensure good wetting between the solution and substrate by removing silicon native oxides, and employs Cu to control Al doping into the layers. Isotropic growth is achieved because of a high concentration of solute silicon in the solution and the resulting microscopically rough interface. As a result, macroscopically smooth Si layers have been grown on cast MG Si that are suitable for device fabrication. With the microscopically rough interface, the growth rate has been studied with a diffusional model involving a boundary layer that takes the melt convection into account. The model was found to be in good agreement with experimental results, indicating only a small boundary layer (∼ 0.1 cm) and a silicon diffusivity of ∼ 2 × 10−4 cm2 s−1 in the liquid. The thin layer (∼ 30 μm) grown on the MG-Si substrate has a minority-carrier diffusion length greater than the layer thickness.

Si thin layer growth from metal solutions on single-crystal and cast metallurgical-grade multicrystalline Si substrates

A simple, vertical-dipping, liquid-phase epitaxy (LPE) method for growth of Si layers from Cu/Si solution at temperatures less than 950/spl deg/C has been shown to be a promising technique for thin crystalline Si photovoltaic (PV) applications. Solar cells with more than 15% AM1 efficiency were fabricated on 5-/spl mu/m-thick layers grown from Cu/Si solution on (111) Czochralski (CZ) substrates. To extend the application of this technique to low-cost substrates, we grew thin (5-40 /spl mu/m) Si layers on cast multicrystalline metallurgical-grade (MG) substrates from Cu/Si solution as well as from Al/Si, Al/Cu/Si, Bi/Si, Ga/Cu/Si, and Sn/Si solutions. The conditions of growth, morphology, solvent incorporation characteristics and problems that arise with the use of multicrystalline Si substrates are discussed. A diagnostic solar cell with efficiency equal to 0.42 and open-circuit voltage equal to 0.89 of the values for a single crystal control cell was obtained, without any light-trapping scheme, on a 15-/spl mu/m-thick layer grown on a MG Si substrate.<<ETX>>

Effect of nitrogen doping on microdefects and minority charge carrier lifetime of high-purity, dislocation-free and multicrystalline silicon

We studied the effects of Si growth in atmospheres containing N2 on minority charge carrier lifetime τ using a high-purity, induction-heated, float-zone (FZ) crystal growth method. Ingots were grown with purge gases that ranged from pure argon (99.9995%) to pure N2 (99.999%). τ was measured as a function of position along the ingots using the ASTM F28–75 photoconductive decay (PCD) method. We found that Ga-doped, multi-crystalline silicon ingot growth in a partial or total nitrogen ambient has a negligible effect on minority charge carrier lifetime and no significant grain boundary passivation effect. Values of 40 μms < τ < 100 μs were typical regardless of ambient. For dislocation-free (DF) growth, the degradation of τ is minimal and τ values above 1000 μs are obtained if the amount of N2 in the purge gas is below the level at which nitride compounds form in the melt and disrupt DF growth.

Grain boundary and dislocation effects on the PV performance of high-purity silicon

To quantify the effects of grain size and dislocation defects on the minority charge carrier lifetime /spl tau/ and photovoltaic (PV) efficiency of silicon, the authors grew high-purity, float-zoned (FZ) ingots with a range of grain sizes from single crystalline (dislocated and dislocation-free) down to 4/spl times/10/sup -4/ cm/sup 2/. In situ ingot cooling rates of 18/spl deg/ and 89/spl deg/C min/sup -1/ were used. Bulk ingot /spl tau/ ranged from less than 30 /spl mu/s for the multicrystalline ingots to 2,500 /spl mu/s for the dislocation-free crystals. Wafers from different positions in the ingots were used for /spl tau/ measurements and the fabrication of mesa-isolated, 0.04-cm/sup 2/ diagnostic PV device structures. They found that /spl tau/ decreased to 4 /spl mu/s and normalized solar cell efficiency decreased to 0.6 for the smallest average grain areas (4/spl times/10/sup -4/ cm/sup 2/).<<ETX>>

Impurity segregation in LPE growth of silicon from CuAl solutions

Al segregation at the solid-liquid interface and Cu segregation at the free silicon surface were studied in liquid-phase epitaxy (LPE) of silicon thin layers from mixtures of Cu-Al-Si. Using the multi-component regular solution model and experimental results, we found that Si-Al and Si-Cu interactions in the liquid solution are repulsive, and Al-Cu interaction is attractive. As a result, Al doping in silicon epitaxial layers is controlled by both Cu and Al compositions in the growth solution to allow epitaxy at about 900°C, with a substantial amount of Al present in the liquid for substrate surface-oxide removal. On the other hand, Cu concentration in the grown layers is determined by both the solid-liquid interface segregation during growth and segregation at the silicon surface after growth. The surface segregation phenomenon can be used to getter Cu from the bulk of silicon layers so that its concentration is much lower than its solubility at the layer growth temperature and the reported 10 17 cm -3 degradation onset for solar-cell performance.

Growth of thin crystalline silicon layers for photovoltaic device use

Abstract Because of the current interest in thin-layer (5–50 μm) crystalline silicon for photovoltaic (PV) applications, we have grown Si layers on (111) single crystal Si substrates from the metal solvents Ga, Sn, and Cu with a particular emphasis on Cu. A substrate dipping technique was used, and the growth temperature was typically near 950°C. The conditions of growth, morphology, qualitative solvent incorporation characteristics, and electrical properties of the solution-grown layers are presented and discussed. Thin single crystal layers with suitable impurity characteristics for photovoltaic use were obtained, and solar cells with 15% conversion efficiency were fabricated on 5 μm thick layers grown on single crystal Si substrates. This is 97% of the efficiency obtained on substrates without a grown layer, and indicates that the Cu solution growth process does not significantly degrade cell performance.

Liquid phase epitaxy for thin-layer silicon PV devices

We have studied liquid phase epitaxial growth of thin‐layer silicon (≤60‐μm thick) on substrates of cast metallurgical‐grade silicon, silicon‐dip‐coated graphite, and single‐aluminum. A solar cell made using a 60‐μm‐thick silicon layer grown on a heavily‐doped single‐crystal silicon substrate has 83% of the efficiency of a control cell made using 300‐μm‐thick float‐zone silicon. In this paper, we discuss a number of issues related to thin‐layer silicon growth including choice of solvents and substrates, thin‐layer silicon material characteristics, and solar cell performances.

Hydrogen passivation and junction formation on APIVT-deposited thin-layer silicon by hot-wire CVD

Abstract The hot-wire chemical vapor deposition (HWCVD) technique was employed to deposit μc-Si emitters and a-SiNx:H passivation/antireflection films, and to hydrogenate silicon thin layers grown by atmospheric-pressure iodine vapor transport (APIVT). Photovoltaic devices with HWCVD μc-Si emitters on APIVT epitaxial silicon exhibit greater than 8% efficiency, similar to those made with diffused junctions. On polycrystalline APIVT-Si layers, a HWCVD-deposited μc-Si emitter reduces open-circuit voltage loss caused by grain boundaries. Hot-wire hydrogenation improves Hall mobility by approximately 50%. HWCVD a-SiNx:H films improve minority-carrier lifetime significantly after thermal annealing at temperatures up to 500 °C.

A silicon ingot lifetime tester for industrial use

A specially designed lifetime measurement instrument has been developed to characterize silicon ingots before they are subjected to expensive slicing and solar-cell processing, thereby saving needless processing costs of inferior materials in a solar-cell production line. The instrument uses the direct-current photoconductance decay (DC-PCD) method for linear detection of the transient photoconductance signal and localized probing / illumination for necessary sensitivity on low resistivity and large samples. The instrument also has a compact and high-power laser diode as the light source, data averaging capability, a pneumatic ingot transport and probe positioning mechanism, and a user-friendly graphical interface for data acquisition / lifetime calculation / data storage / hardcopy for factory-floor use with quick turnaround. A 3-dimensional finite-element analysis indicates that the as-cut surface finish is adequate for measuring the bulk lifetime on the order of 50 ms or less. Measurement repeatability and clear distinction among different grades of feedstock materials have been demonstrated.

Properties of iron-doped multicrystalline silicon grown by the float-zone technique

Multicrystalline Fe-doped Si ingots were float-zoned from high-purity feed rods. Fe was introduced by pill-doping, which gives uniform impurity content for small segregation coefficients (k/spl sim/10/sup 15/ for Fe in Si). Fe concentrations were calculated from the initial weight of the Fe pill, the molten zone geometry and the growth parameters. Values in the range of 10/sup 12/-10/sup 16/ atoms/cm/sup 3/ were targeted. No additional electrically active dopants were introduced. Minority charge carrier lifetime (via YAG-laser-excited, 430 MHz ultra-high-frequency-coupled, photoconductive decay) was measured on the ingots, and wafers were cut to examine grain structure and electron-beam-induced current response of grain boundaries. Observed lifetimes decreased monotonically with increasing Fe content for similar grain sizes (from /spl sim/10 /spl mu/s to 2 /spl mu/s for <10/sup -3/ cm/sup 2/ grains, from /spl sim/30 us to 2 us for /spl sim/5/spl times/10/sup -3/ cm/sup 2/ grains, and from /spl sim/300 /spl mu/s to 2 us for >10/sup -2/ cm/sup 2/ grains) as the Fe content increased to 1/spl times/10/sup 16/.