Nondestructive photoelastic characterization of monocrystalline silicon photovoltaic wafers

Silicon wafers produced for the photovoltaic industry are becoming thinner. This renders the wafers more susceptible to bow under thermal residual and wafer handling stresses. Furthermore, a saw damage layer exists on the top and bottom surface of each wafer as a result of slicing the silicon ingot into wafers via a wire saw. Thinner wafers are more prone fracture under the combination of residual stress and external loads acting on microcracks in the saw damage layer. This research presents a novel method for characterizing monocrystalline silicon wafers using a photoelastic imaging technique that is capable of measuring the bulk residual stress fields in silicon wafers as well as locally elevated stress fields near microcracks. Infrared photoelastic (PE) imaging is used to reveal the bulk residual stress field of each wafer and locally elevated stress field around each defect. The defect stress fields are first classified by using a support vector machine learning algorithm. The measured local and bulk residual stress fields around each defect are analyzed in order to characterize each crack according to its stress intensity factors. The distribution of cracks on each wafer is subjected to a virtual uniaxial tensile load where the wafer strength is determined by the load at which the first crack is expected to propagate. The distribution of wafer strengths is used to calculate the expected Weibull parameters for the wafer set, under a pure tensile load. Since the method is nondestructive, it can be used to calculate and compare the expected Weibull parameters for the same set of wafers under a variety of externally applied loads. The variation in wafer strength when loaded parallel and perpendicular to the direction of wire motion is in agreement with experimental observations. With the proper modifications to the experimental setup, using this method it is possible to rapidly and accurately characterize large batches of wafers in an automated way.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Mechanical Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2019

Author dc:creator

• Rowe, Logan Perris

Contributors dc:contributor

• Johnson, Harley T.
• Horn, Gavin P.

• photoelasticity, photovoltaic, solar, silicon, machine learning, Weibull, microcrack detection

Statement dc:rights

• Copyright 2018 Logan Rowe

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/102490

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/102490