Polysilicon falls to 28.3% of module cost, crucible rises to 3.7%

This week saw the polysilicon price fall below the 200 yuan per kilogram mark for the first time since January – it is now $28 per kilogram per the Silicon Industry Branch, so $70 per kW assuming a usage rate of 2.5 grams per Watt. Another $1.2 billion, 100,000-ton factory was put into operation by Daqo New Energy days ago, and 2023 will see production capacity double to at least 2.4 million tons. It’s only downhill from here for the polysilicon price.

In contrast, the price of high-purity quartz sand is expected to soon reach $54,000 per ton, which will push wafer prices up. Price reporting for the crucibles themselves is sparse, but a rough calculation featuring 33 kilograms each of outer, inner, and middle-layer grade high-purity quartz sand at respective prices of $54,000, $29,000, and $13,000 per ton, brings us up to a $3,200 cost for a 100 kilogram crucible – with another $300 production cost, that’s $3,500, another 25% price rise for crucibles just like the one we reported in March. If each crucible is responsible for producing 378 kW, the crucible price is $9.26 per kW of finished product.

Note that this graph starts the Y axis at zero!

Last week’s module price from Infolink was $247 per KW, so polysilicon has fallen to 28.3% of module cost. If the polysilicon cost falls to $22 per kilogram by the end of the year (if not, it will be lower), that will remove $15 per kW (6%) from the module cost.

Crucible cost is at 3.7% of module cost per our calculations. Given that consistent nature of the shortage and the small number of relevant companies, we can’t see why quartz sand prices and thereby crucible prices can’t or won’t be hiked up far, far higher than that. Nothing stopped polysilicon makers from charging seven times their production cost – they kept hiking until people were only barely willing to buy. If polysilicon was able to seize 40% of module price, could quartz sand grab 20%?

These calculations are muddied by variations in the speed of ingot-pulling, the fact that up to 15% of the material can be recycled (now that the prices have increased), and that crucible depth has increased to the extent that crucible weight has in some cases increased from 90 kilograms to 130 kilograms. Not to mention differences in crucible purity and lifespan. In another example of the diversity of this supply chain element, a small-size 28-inch crucible can avoid using imported high-purity sand at all if it is used for no more than 200 hours.

The quartz sand shortage is most acute with the 4N8 (99.998%) purity used in the inner coating layer which contacts the polysilicon during the Czochralski process. This high-purity sand is almost exclusively the preserve of the two Western manufacturers who make 30,000 tons per year. Unimin has yet to decide on expansion plans while TQC has reportedly decided against expansion.

The leading Chinese quartz sand producer Pacific Quartz is able to make 5,000 tons a year of inner-layer, and Triumph Technology, also Chinese, is working on 5,000 tons of synthetic quartz production capacity.

Pacific Quartz’s total high-purity Quartz sand expansion this year will be 50,000 tons with another 60,000 tons next year, but it is unclear how much of this if any will be suitable for the inner layer. Pacific Quartz’s expansions dwarf this year’s expected overall quartz sand shortage of 15,000 tons out of 100,000 tons demand, but they barely keep ahead of the expansion of demand from the rest of the solar industry including the wafer segment. A manufacturing demand of 400 GW (less than the current pace of the industry) would require up to 132,000 tons of high-purity quartz sand, including 53,000 tons of 4N5 purity inner-face grade sand.

Mitsubishi’s synthetic quartz can be used, but it costs something like $43,000 per ton or even more according to reports, with a production cost (if the same methods were adopted in China) estimated at $25,000 per ton. Its excessive purity makes it less durable, thus needing to be replaced even more frequently – the equivalent cost might be something over $50,000 per ton with that in mind.

However, a recent interview by Triumph Technology claims that it can cut product cost of its synthetic quartz sand – for 5N purity, lower than typical synthetic sand for semiconductors – to just $5,000, which would return prices to January 2022 levels – a return to normalcy which would make crucibles a forgettable element of the solar module cost once again. But we can assume that it will take years to reach that point. Triumph Technology will also be making 6N and 7N purity sand for semiconductors.

There is also a shortage, less acute, with the middle layer sand supply which is sourced by Pacific Quartz from various mines around the world, especially Rajasthan and Gujarat. Generally speaking the price of middle-layer sand is two-thirds that of inner-layer and will follow a similar price trend in the near-term, while the price of outer-layer is 20% lower than middle-layer and will not increase.

The other two layers in a Czochralski process crucible, the vacuum transparent middle layer and the opaque outer layer, nominally use 4N5 purity. However the middle layer supply comes from Pacific Quartz while outer-layer supply is increasingly the preserve of smaller Chinese start-ups whose product is lower-quality, so the requirement is apparently less strict for the outer layer.

Ingot-pulling of wafers intended for N-type cells requires not just that the crucible be replaced about 50% more frequently, than for P-type, but also that the crucible be higher-quality. So to match the adoption of N-type cell technology, which has just begun in the past couple of years, there is an upgrade in the quality of crucibles used in the N-type supply chain.

The other change in downstream manufacturing which has affected crucibles is the adoption of larger cell sizes, namely 182mm and 210mm, which explains the shift from 28-inch to 32-inch, 36-inch and even 40-inch crucibles.

For geometrical reasons, these larger crucibles have a higher proportion of inner-face quartz sand usage, reaching a 4:3:3 ratio between inner, middle and outer layers. But other reports say there has been a divide between 2:2:6 ratio and 3:4:3 ratio crucibles, with the former having 350-400 hour lifespan and a 15% slower ingot-pulling rate, while the latter had a 300-hour lifespan or less. TCL Zhonghuan and LONGi have been able to persist with usage of 3:4:3 crucibles due to their scale and secure supply chain.

That size increase has been spearheaded by crucible maker Oujing Technology. Amusingly, the company’s website is currently down and an article published just today on the Chinese net exposes several problems with the company’s filings with the China Securities Regulatory Commission, made in the context of its bid to be listed on the Shenzhen Stock Exchange. This reflects the reality that crucible making companies have always been tiny in terms of revenue because they contributed only a fraction of a percentage point to overall module costs, until the recent shortage. Last year the company’s revenue reached $209 million, up 69% from 2021.