Solar and silicon have gone together ever since Bell Labs devised the world’s first functioning PV cell back in 1954. From then to the first commercial solar cells produced for space applications, and onwards to this day, wafer-based crystalline silicon technology has been synonymous with the term “photovoltaics.”
Until fairly recently, the solar market’s relatively small size meant that the polysilicon market predominantly served the semiconductor manufacturing industry, with the bulk of the world’s polysilicon output being used to create similar, but purer wafers compared to those used for PV cells. All the way into the mid-2000s, the solar industry subsisted quite literally on the scraps of semiconductor manufacturing, procuring its feedstock needs from off-spec or unused silicon rejected by that sector.
How things have changed. With solar exploding over the last decade and c-Si technology staving off the challenges of CSP and thin film, demand for polysilicon has shifted dramatically -- and likely irrevocably -- toward solar.
The chart below illustrates solar’s rapid path toward dominance of the polysilicon industry: as estimated by GTM Research, from constituting only 27 percent of global polysilicon demand in 2001, solar is expected to devour almost 90 percent of the world’s production in 2014. For all practical purposes, the future of the polysilicon industry depends on the solar market more than ever before.Source: GTM Research, Sage Concepts, Prometheus Institute
To a great extent, polysilicon’s reliance on solar (or more correctly, PV) has been reciprocated in recent years. Historically, polysilicon has easily been the single largest cost center in a PV system. The Great Polysilicon Shortage of 2003-2008, followed by The Great Polysilicon Price Collapse of 2009-2013, were defining developments in the PV manufacturing industry and were responsible for determining the fate of possibly hundreds of solar firms in one way or another, with Solyndra’s insolvency and First Solar’s transition to a fully integrated power plant EPC firm being two prominent examples.
At the same time, silicon’s importance to solar economics at large has undeniably decreased over the years. From a cost perspective, silicon may still be the largest individual contributor in the module, or even system, bill of materials -- but what was once a vast gulf (between silicon and other cost centers) has now shrunk to a small gap. As the chart below shows, silicon has dropped from 71 percent of the all-in module cost for Chinese manufacturers in Q4 2008, to just 18 percent in Q4 2013E.Source: GTM Research Global Competitive Intelligence Tracker
The main reason for this, of course, is the precipitous fall in polysilicon prices during this time -- but what often goes ignored is that manufacturers have also reduced the unit consumption of silicon tremendously over the years, from over 15 grams per watt in 2000 to around 5.2 grams per watt today (mostly due to improvements in cell efficiency, manufacturing yields and wafer thickness). Partly because of this, the future threat of polysilicon price increases to overall solar costs has diminished. According to GTM Research’s current base-case forecast (as discussed in detail in our recent report on 2014 global PV pricing dynamics), polysilicon pricing increases by almost 30 percent over the course of 2014, which would cut into profit margins for wafer, cell and module suppliers, but would still only increase overall module costs by 5 percent and system costs by a mere 1 percent.
In the long term, a stable, sub-$25/kg polysilicon pricing environment is an important component of most $1/W system cost roadmaps, and some may be concerned about the possibility of another drastic increase in pricing as occurred in 2003-2008. But while intermittent periods of short supply accompanied by meaningful price increases can be expected (GTM’s high-case silicon price for 2014-2017 is $28/kg), I believe the days of silicon as the gating factor for solar demand may be over for good. The supplier landscape is larger and more diverse; there is a slew of hopeful new entrants from China and the Middle East; and next-generation Siemens and FBR reactors can already achieve fully loaded production costs in the mid-teen levels.
In the end, what this means is that the goal of lowering module, system, and ultimately, solar electricity costs has become less monolithic and more nuanced. At the module level, no single technology innovation has the potential to be disruptive on its own -- the problem needs to be attacked on many fronts simultaneously. A penny here, a penny and a half there -- all the pieces will matter from here on out. And while silicon rightfully remains a subject of much discussion and debate in the solar industry, it is perhaps time that innovations in other areas -- system design, balance-of-systems components, customer acquisition and financing -- share the stage.