Defect and Impurity Distributions in Traditionally Cast Multicrystalline and Cast Monocrystalline Silicon for Solar Substrates

by Witting, Ian Thomas

Abstract (Summary)
An examination of two types of cast silicon was performed to correlate observed electrical properties with structural defects and chemical impurities. Results show that many of the significant impurities and defects occur as the result of controllable manufacturing methods. A unique sample of traditionally cast multicrystalline silicon (mc-Si) was obtained from BP Solar in April 2007. This sample was cut longitudinally and has intact the original edges of the ingot typically removed during processing. Impurity concentration profiles are compared directly with minority carrier lifetime maps to determine which impurities are most detrimental to the quality of the silicon. A minority carrier lifetime map was obtained using Microwave Photoconductance Decay (MW-PCD). Light element impurities (C, O) were profiled using Fourier Transform Infrared Spectroscopy (FT-IR) and metallic impurities were profiled by Deep Level Transient Spectroscopy (DLTS). Iron contamination was found to correlate best with reduction in minority carrier lifetime. Silicon samples exposed to the environment of different furnaces used throughout the casting process were examined using Scanning Electron Microscopy (SEM) with Energy Dispersive Spectroscopy (EDS) and Secondary Ion Mass Spectroscopy (SIMS). This examination revealed the sources of many impurities observed in the final cast ingots. SEM and EDS were also used to examine impurity precipitation on samples taken from the tops of cast ingots. Carbide, nitride, and iron were found to be the primary contaminants that segregated to the ingot top. A second unique sample of cast monocrystalline silicon, referred to as âmono-squaredâ and written as âMono2 TMâ, was obtained from BP Solar in December 2007. This sample is cut longitudinally from the original ingot similarly to the mc-Si slice. Examination of defects and impurities in this sample was performed based upon a MW-PCD minority carrier lifetime map. Structural defects and precipitates within the sample were examined using step-wise Secco etching combined with SEM and Nomarski microscopy. FT-IR was used to profile light element impurities and DLTS was used to profile metallic impurities. As with the cast mc-Si sample, minority carrier lifetime was observed to correlate with the DLTS determined iron concentration profile. The iron concentration profile was observed to correlate well with transitions in structural defect type. The structural defects observed suggest a vacancy and interstitial incorporation dependency on crystal growth parameters, as described by Voronkov, for silicon grown using the Czochralski method. Dissolved impurity concentrations within the Mono2 TM ingot were found to be significantly higher than the mc-Si sample, despite having very similar casting parameters. This is perhaps due to a reduction in dislocations and grain boundaries that act as gettering sites. Precipitates were observed to nucleate rows of dislocations that extended in the direction of crystal growth. These rows of dislocations were observed to merge with other rows creating more densely packed rows termed âdislocation cascadesâ. These cascades are observed to branch and merge as crystal growth continues. Branching was observed to occur in the presence of oxide precipitation contaminated with metallic impurities.
Bibliographical Information:

Advisor:Gerd Duscher; Carlton Osburn; George Rozgonyi

School:North Carolina State University

School Location:USA - North Carolina

Source Type:Master's Thesis

Keywords:materials science and engineering


Date of Publication:08/26/2008

© 2009 All Rights Reserved.