Srinivas Devayajanam

Understanding light-induced degradation of c-Si solar cells

We discuss results of our investigations toward understanding bulk and surface components of light-induced degradation (LID) in low-Fe c-Si solar cells. The bulk effects, arising from boron-oxygen defects, are determined by comparing degradation of cell parameters and their thermal recovery, with that of the minority-carrier lifetime (τ) in sister wafers. We found that the recovery of t in wafers takes a much longer annealing time compared to that of the cell. We also show that cells having SiN:H coating experience a surface degradation (ascribed to surface recombination). The surface LID is seen as an increase in the q/2kT component of the dark saturation current (J02). The surface LID does not recover fully upon annealing and is attributed to degradation of the SiN:H-Si interface. This behavior is also exhibited by mc-Si cells that have very low oxygen content and do not show any bulk degradation.

Characterizing damage on Si wafer surfaces cut by slurry and diamond wire sawing

We have measured and compared surface roughness and the degree of damage for wafers cut by three different sawing techniques - slurry, Ni-based diamond wire, and resin-based diamond wire sawing. The local damage was determined by angle polishing followed by defect etching, TEM, SEM/EBSD imaging and Raman imaging. It showed that each of the cutting processes produces a thin layer of amorphous Si at the surface and dislocation loops that can go about 1 μm deep below the surface. A new approach was used to quantify the average damage over a large area. We determined the effective surface recombination (SRV) as a function of depth. Because the effective SRV is a function of the carrier loss close to the surface, it is well-suited to define damage distribution at and below the surface. Wafers with surface damage were step etched in (HF:HNO3:CH3COOH::1:1:5), and the effective lifetime was measured with a Sinton system after each etching step, with iodine-ethanol passivation. The SRV plots as a function of depth, representing depth distribution of the damage, were compared for large groups of wafers cut by each technique. Our results show that for optimized cutting, all three cutting methods produce damage depth of about 5μm (each surface). However, the degree of damage is higher for slurry cut wafers.

A reflectance spectroscopy-based tool for high-speed characterization of silicon wafers and solar cells in commercial production

Some new applications of reflectance spectroscopy using the GT FabScan are described, which make this system highly desirable for process monitoring in commercial Si solar cell fabrication. These applications include grain orientation, grain size distribution, dislocation density distribution, and antireflection coating thickness on a finished solar cell. These measurements are performed very fast, typically in less than 10 ms over the entire wafer.

A new method for rapid measurement of orientations and sizes of grains in multicrystalline silicon wafers

We describe a new technique for rapid measurement of orientations and sizes of various grains in a multicrystalline silicon (mc-Si) wafer. The wafer is texture etched to expose (111) faces nearest to each surface. Because grains of different orientations result in uniquely different texture shapes, they also have well-defined reflectance values. Hence, the process of determining the grain orientations is brought down to making reflectance maps. Reflectance maps are produced by PVSCAN or reflectometer (GT FabScan), and then transformed into orientation maps. Because the grain boundaries are very well delineated in the reflectance maps, they are also excellent for making measurements of size and distribution of grains. We will compare the results of this technique with other standard techniques.

Some challenges in making accurate and reproducible measurements of minority carrier lifetime in high-quality Si wafers

Measurement of the minority carrier lifetime (τ) of high-quality wafers (having bulk minority carrier lifetime, τb > few milliseconds) requires surface passivation with very low surface recombination velocity, typically <; 1cm/s. Furthermore, for mapping large (e.g., 156 x156 mm) wafers, the passivation must also be stable and uniform over the entire wafer surfaces. These are very demanding requirements and it is a common experience that they are very difficult to achieve. Yet, they are necessary for performing defect analyses of the current N-type wafers. To understand the problems associated with these measurements, we have studied effect of wafer preparation (cleaning procedures, handling) and the passivation characteristics (stability, sensitivity to light, thickness of the passivation medium required for stable passivation) for many commonly used passivation media-iodine-ethanol (IE), quinhydrone-methanol (QHM), aluminum oxide (Al2O3), amorphous-silicon (a-Si), and silicon dioxide (SiO2). Here, we will discuss main factors that influence the accuracy and repeatability of lifetime measurements.

Dissolution of Oxygen Precipitate Nuclei in n-Type CZ-Si Wafers to Improve Their Material Quality: Experimental Results

We present experimental results which show that oxygen-related precipitate nuclei (OPN) present in p-doped, n-type, Czochralski wafers can be dissolved using a flash-annealing process, yielding very high quality wafers for high-efficiency solar cells. Flash annealing consists of heating a wafer in an optical furnace to temperature between 1150 and 1250 °C for a short time. This process produces a large increase in the minority carrier lifetime (MCLT) and homogenizes each wafer. We have tested wafers from different axial locations of two ingots. All wafers reach nearly the same high value of MCLT. The OPN dissolution is confirmed by oxygen analysis using Fourier transform infrared spectra and injection-level dependence of MCLT.

Analyses of diamond wire sawn wafers: Effect of various cutting parameters

We have evaluated surface characteristics of diamond wire cut (DWC) wafers sawn under a variety of cutting parameters. These characteristics include surface roughness, spatial frequencies of surface profiles, phase changes, damage depth, and lateral non-uniformities in the surface damage. Various cutting parameters investigated are: wire size, diamond grit size, reciprocating frequency, feed rate, and wire usage. Spatial frequency components of surface topography/roughness are influenced by individual cutting parameters as manifested by distinct peaks in the Fourier transforms of the Dektak profiles. The depth of damage is strongly controlled by diamond grit size and wire usage and to a smaller degree by the wire size.

Bulk defect generation during B-diffusion and oxidation of CZ wafers: Mechanism for degrading solar cell performance

We describe results of our experimental study to investigate the effect of B diffusion and drive-in/oxidation on minority carrier lifetime of the wafer. We have observed that B diffusion generates stacking faults that can be attributed to injection of Si interstitials into the wafer by formation of a boron rich layer at the wafer surface. These Si interstitials are also believed to enhance interactions between the native point defects and impurities (such as O, Fe) in the wafers during subsequent processing leading to the development of swirl patterns. Spatial variation of the lifetime degradation follows the point defect interactions and impurity segregation/precipitation. Lifetime can be partially recovered by Phosphorous (P) gettering. The overall effect on the cell performance due to Si interstitial generation, impurity/point defect interactions, and P-gettering is briefly discussed.

Defect Generation and Propagation in Mc-Si Ingots: Influence on the Performance of Solar Cells

This paper describes results of our study aimed at understanding mechanism (s) of dislocation generation and propagation in multi-crystalline silicon (mc-Si) ingots, and evaluating their influence on the solar cell performance. This work was done in two parts: (i) Measurement of dislocation distributions along various bricks, selected from strategic locations within several ingots; and (ii) Theoretical modeling of the cell performance corresponding to the measured dislocation distributions. Solar cells were fabricated on wafers of known dislocation distribution, and the results were compared with the theory. These results show that cell performance can be accurately predicted from the dislocation distribution, and the changes in the dislocation distribution are the primary cause for variations in the cell-to-cell performance. The dislocation generation and propagation mechanisms, suggested by our results, are described in this paper.

Online monitoring for Si solar cell manufacturing

We have developed an online monitoring system that can image various parameters of wafers and cells as they are transported over a conveyer belt. For silicon applications, these parameters include: sawing irregularities, texture quality and uniformity, AR coating thickness, metallization statistics on finger and busbar widths, and final “visual inspection.” For a multicrystalline wafer, it also measures grain size and grain orientations. The system involves a light source and a line camera, to record the reflectance image of the wafer/cell moving at speeds up to 5 inches per second (ips). A high-speed computer then transforms the reflectance images into the appropriate parameter images. We will describe the essential principles of the system, image processing methods, and discuss examples of the applications in a manufacturing facility. The current system is a single wafer line, carrying wafers/cells at a speed of 2 inches per second. The imaging system can be adapted to existing conveyor belt assembly.