A surface wave plasma source was used for the deposition of amorphous (a-Si) and nanocrystalline (nc-Si) silicon thin films for the manufacture of silicon solar cells.This source was optimized for 900 MHz microwave excitation.The process gases used were silane (SiH4) and hydrogen (H2).The plasma source was shown to be advantageous in depositing films at very high deposition rate, exceeding 2 nm/s for nc-Si, while deposition of a-Si was observed at 10 nm/s and could be increased if a higher flowrate mass flow controller was used.Film thickness was measured via profilometry with verification through SEM imaging, while the crystallinity was determined via peak fitting of Raman spectra.A distinct transition from nc-Si to a-Si was observed between 1% and 2.5% SiH4 concentration, increasing for higher source power and decreasing for lower substrate temperature.An optimal substrate temperature was found for depositing nc-Si: 285°C for 1.0 W/cm2, and 350°C for 0.5 W/cm2.Expansion of the nc-Si process window to higher deposition rates was shown to be possible by higher source power.Film nanostructure of nc-Si was determined by XRD, Raman analysis, TEM, and EPR.Calculation of grain size for 100 nm films yielded 5±1 to 15±2 nm from 200°C to 400°C.EPR analysis of a-Si and nc-Si revealed that defect density increased with crystallinity.Due to adverse deposition conditions, calculated defect densities for the surface wave source ranged from 1.2±0.3•10^16 cm^-3 for a-Si to 7.3±1.2•10^17 cm^-3 for nc-Si.However, a ceteris paribus comparison with films made by radio frequency capacitively-coupled discharge (RF CCP) showed that a-Si made by the latter method had 7.4±0.7•10^17 cm^-3, a factor of 6 worse than a-Si produced via surface wave.The low oscillation height of ions due to high freqency, as well as low sheath potential due to low electron temperature T_e, combined to generate high quality Si thin films relative to the RF CCP industry standard.Numerical modeling of the SiH4 + H2 model was achieved using the volume-averaging formulation of Kim and Lieberman.[49]An extensive literature study for the physical parameters of cross sections and rate coefficients accompanies this work.40 gas species and 62 reactions were tracked, as well as the surface reactions involved in deposition.The result of the plasma simulations predicted experimentally observed trends in deposition rate vs. silane concentration (accurate within 10-25%) and total pressure, although it diverged from what was seen experimentally in varying source power.T_e for 100% H2 was found to be 2.7 eV at 100 mTorr and 2.5 eV at 200 mTorr, coincident with Langmuir probe measurements.Basic plasma trends such as decreasing T_efor higher pressure and constant T_e/increasing n_ewith power were predicted by the model, as well.Most importantly, the influence of hydrogen abstraction via incident H flux was correlated with increased crystallinity, coinciding with what was observed experimentally and what is argued in literature.[68]Finally, solar cells were manufactured on n-type Si wafer with the surface wave and RF sources for comparison.While the RF-made cell did not produce any measurable voltage or current, the surface wave produced a functioning solar cell with very low efficiency.Although this device was far from industry standards due to the CPMI’s and author’s lack of skill and resources to manufacture photovoltaics, it is a basic illustration of the advantages of the low-damage MSWP source over conventional industry methods.
【 预 览 】
附件列表
Files
Size
Format
View
Modeling and experimental process optimization for a SiH4 + H2 surface wave plasma discharge for silicon photovoltaics
PDF
download
878987797
028-85220240
