By Jeff Nestel-Patt, Scott Kroeger

The PV Cost Model

GT Solar‘s turnkey group has done extensive cost modeling on a wide range of large, fully-integrated PV manufacturing facilities across different equipment sets to come up with a viable cost model for the PV manufacturing value chain. The PV manufacturing cost per watt from silicon to module is approximately US$1.32/watt. As shown in Table 1, wafering is the most costly component of this total at US$0.43.

The wafering operation can be broken down into its functional components for further cost analysis to see which parts of the overall operation are the most costly. Table 2 shows that ingot growth, wafering, and slurry operations are the three most costly functions with wafering representing the single highest cost.

Based on the data in Table 2, PV manufacturers should develop strategies that optimize their wafering operation to achieve a higher return on their investment. One way of achieving this goal is to increase mass ingot yield. Mass ingot yield is a measure of the total amount of the multi-crystalline ingot that gets processed into good wafers. While there are many factors that impact mass ingot yield, for the purposes of this article we will not discuss factors that occur during the ingot growth phase, rather we will address the wafering operation and explore a concept that increases the number of wafers that can be produced from an ingot, thus increasing mass ingot yield.

Rethinking Brick Cropping Methodologies

One way of getting more wafers out of an ingot is to change the way bricks are cropped prior to being sawed into wafers. Wafering data that we have collected from a wide range of companies suggests that conventional cropping methodologies send good waferable material to scrap that is either lost or recycled. Manufacturers typically make two cropping cuts to remove top and bottom brick sections. These cuts are usually set to standard positions that actually cut into good waferable silicon. This causes two problems; it results in inefficient recycling because good material is being scrapped, and because the crop is being made in otherwise waferable material, additional good material is being lost as kerf.

The decision about where the top and bottom crops should be made is typically based on brick level lifetime testing. The problem with brick level lifetime testing is that the analysis is performed by scanning the side of the brick, which is actually perpendicular to the eventual surface of the wafer (Figure 1). If the goal of lifetime testing is to eliminate low lifetime material and maximize the amount of good material that can be turned into cells, then wafer level lifetime testing may be a more accurate indicator of determining where to make cropping cuts on the brick.

 Optimizing Mass Ingot Yield

Increasing mass ingot yield is like finding additional profit previously hidden in your wafering operations because it drops directly to the bottom line--the more wafers produced out of a given ingot, the higher the profit. The goals of optimizing your mass ingot yield should be to produce more wafers per brick by reducing the amount of good material to kerf loss and minimizing the amount of good material that gets recycled. This will increase your overall factory output, lower cost of ownership (reduced cost/Wp), and will have no impact to downstream cell yield processing. By implementing a wafer level testing methodology into your wafering operations, mass ingot yields of greater than 70% with typical production feedstock mix are readily achievable.

Wafer lifetime testing results from a conventional brick cropping methodology are plotted on the graph in Figure 2. Conventional cropping cuts (indicated by the two red vertical lines) are intended to keep ‘bad’ wafers from being processed into cells so wafers that fall outside the two vertical lines typically get recycled. The red horizontal line indicates the threshold below which wafers should be recycled due to low wafer lifetime testing results. What Figure 2 clearly points out is that in an effort to keep ‘bad’ wafers from entering the cell line, manufacturers are losing otherwise waferable material to kerf and recycling and therefore not optimizing their mass ingot yield.

Another way of visualizing this is found in Figure 3. Figure 3 presents two conventional brick cropping methodologies typically found in wafering operations. The brick on the left labeled Typical 1 shows a large crop area on the bottom and two crop area on the top. This approach reduces scrap material but sends otherwise good waferable material to recycling. The brick labeled Typical 2 results in slightly more waferable material, but scraps a higher amount of good material. In both cases though, good material is either scrapped or recycled, which reduces mass ingot yield.

An Optimized Approach to Brick Cropping

To increase the amount of waferable material, wafer manufacturers should take a different approach to brick cropping. By examining wafer lifetime testing data from multiple production runs operators can determine the optimal way to preserve as much of good material as possible while minimizing scrap and recycled material. Figure 4 shows how a manufacturer can increase mass ingot yield by 2.5% or an equivalent of 18 wafers for each brick. This is not a trivial increase as we will see when we translate this into additional profit.

The optimized brick on the right of Figure 4 produces the same amount of scrap material from the top of the brick as the Typical 1 example, but it greatly reduces the amount of recycled material by using wafer lifetime testing data to determine the optimal place the cut. On the bottom of the brick, careful analysis of wafer lifetime testing data allows the manufacturer to locate the optimal place to make the crop that will keep good waferable material from becoming recycled material. Even with the additional kerf loss due to the additional crops in our optimized example, the amount of good material sent to wafering is still greater than either of the other two examples.

Impact on Revenue and Profit

Table 3 quantifies the difference in mass ingot yield improvement between the base case scenario and the optimized brick cropping methodology described previously. The base case is based on a 420 Kg ingot, annual production capacity of 25,500,000 wafers at 200 micron thickness, and polysilicon pricing of US$45/Kg. Material utilization improves from 6.7 grams/watt in the base case to 6.5 grams/watt for the optimized case. Wafers per kilogram of material improve from 37.3 wafers/Kg to 38.3 wafers/Kg. Based on the base case scenario, the improved material utilization translates into approximately 1,000,000 additional wafers produced annually, or a 3.4% increase in revenue.

Jeff Nestel-Patt is Director of Marketing Communications, and Scott Kroeger is Director of Marketing at GT Solar (http://www.gtsolar.com/).