Chip makers chase the Sun as competition hots up

The solar energy industry is a curious one. Not yet a cost-effective way to produce electricity without huge subsidies, government support of photovoltaics has been enough to drive steep demand over the past two decades.

According to solar industry analyst Michael Rogol, the production capacity of photovoltaic systems shipped during 2006 reached 2.5 GW, double the 2004 figure and 10 times that recorded in 2000. Winfried Hoffman, president of the European Photovoltaic Industry Association (EPIA) and CTO of equipment giant Applied Materials, recently estimated the 2007 figure at a slightly lower 2.3 GW.

Nevertheless, photovoltaic systems are still responsible for less than 0.1% of global energy production. And despite all of the recent interest in technologies using III-V cells, concentrator photovoltaics (CPV) systems are yet to make any real impact, with estimated shipments of just 18 MW in 2006 – less than 1% of the solar total.

However, everybody agrees that this installed capacity is going to rise – and fast. Sarah Kurtz from the US National Renewable Energy Laboratory (NREL) is credited as co-inventor of the multi-junction cells of which so much is hoped. Extrapolating from industry trends, Kurtz says that, by 2020, 5% of global electricity could be produced directly from the Sun. Already valued at $20 billion per year, that level of penetration would represent a further 100- or even 1000-fold expansion of the solar energy industry. Hoffmann expects annual installations to rise from a peak capacity of 2.3 GW in 2007 to 11 GW in 2012, and Goldman Sachs says that total photovoltaic capacity could meet about 0.8% of global electricity demand in 2011.

It is this opportunity for expansion that drives venture capitalists (VCs) towards the emerging technologies that might one day see photovoltaics deliver electricity on a utility scale at an economically viable cost. Eric Emmons, a general partner at the Massachusetts Green Energy Fund, told delegates at the 2008 Key Conference organized by Compound Semiconductor that venture investment into solar start-ups spiked sharply in late 2007. In North America, venture firms sank $400 million into this sector in the third quarter, funding CPV exponents SolFocus and Greenvolts, among others.

That spike may already have set some alarm bells ringing. As Emmons points out, the biofuels sector showed a similar funding spike in 2006. Already, however, there is overcapacity in biofuel production and a serious question mark about the validity of the technology and its negative impact on food production. VCs can get cold feet very quickly.

But it isn t all about VCs in the photovoltaics space. Some of the older start-ups are now established public companies and have reaped huge financial gains for their shareholders. The share price of First Solar, the US-headquartered company that manufactures CdTe cells on glass substrates, rocketed 10-fold over the course of 2007. Remarkably, the firm is now perceived by the stock market to be more valuable than the Ford Motor Company.

Michael Molnar, a vice-president at Goldman Sachs and responsible for equity research in the alternative energy and coal sectors, explained why First Solar was causing so much excitement: "They will potentially generate close to $1 billion in revenue at a gross margin of about 50% in 2008. And the stock market has noticed." Despite the optimism, Molnar says that the solar industry must remember that ultimately it is selling nothing but a commodity product – units of electricity. "Photovoltaics needs to be low cost," said the investment analyst. "[Currently] solar energy costs 20–30 cents per kilowatt-hour and the market [for it] only exists because of subsidies."

For all of the concerns over fossil fuels, coal remains a relatively cheap, reliable energy source that, unlike solar or wind, can be stored easily. The US uses about a billion tonnes of coal each year, and with more than 250 billion tonnes in reserve (according to Molnar s figures), there is no risk of it running out for a long time. So here is the key challenge for solar companies targeting the utility end of the market – achieve a unit price approaching grid parity, meaning 5–10 cents per kilowatt-hour. There are three key parts to meeting this challenge: driving down the cost of the cell materials, installation and system construction; improving cell efficiencies; and, crucially, scaling production to large volumes while maintaining high efficiencies and yields.

First Solar aside, silicon is the dominant material on the photovoltaics scene. But, because of their low conversion efficiency, there is a barrier to driving down the cost of photovoltaics with silicon cells. This is especially pertinent in utility applications, where the unit cost must be even lower than the retail price used as a guide for residential photovoltaic installations. Combined with the constrained silicon supply – huge amounts are needed for panels, and the solar industry gets through more of the material than the semiconductor industry – this has created a window of opportunity for high-efficiency III-V cells and CPV systems to make their mark on the industry with a low-unit-cost, utility-scale solution.

However, while there is potential for economic electricity production with III-V cells, its deployment will be restricted to geographies with clear blue skies. High-quality tracking systems that can follow the path of the Sun precisely and reliably are key. In the US, the Solar America Initiative (SAI) is funding the development of all kinds of photovoltaic technology. Its key goal is to deliver solar energy at a cost equivalent to conventional grid prices by 2015. Prior to forming the SAI, the target year had been 2020, and Kurtz says that the project has added "a real sense of urgency" to the photovoltaic sector. Within III-V cells and CPV, its incubator has funded six projects, the most recent being headed up by EnFocus, MicroLink Devices and SolFocus.

Along with its involvement in the SAI, SolFocus is one of the companies leading the effort to commercialize CPV. It is concentrating on the parts of the world where energy costs are high – places like Italy, Japan and California. This seems to be paying off, with recent deals to equip the ISFOC demonstration project in Spain and to power radio transmitters used by the San Francisco radio station KGO. Mark MacDonald from the company reports that the mirror system used by SolFocus can tolerate tracking errors particularly well – crucial when the intense beams of light created by CPV could damage other components if they are directed anywhere but directly at the center of the III-V cell.

Over in Europe, Spanish company Isofotón has been in pilot CPV production for just over a year and expects to have an annual module assembly capacity of 10 MW in place by the end of 2008. Founded in 2001, Isofotón is one of the best-known solar energy companies in the region and, thanks to a succession of projects funded by the European Commission, it has wide-ranging expertise in optical technologies. "We are committed to getting to 1000-sun concentrations with III-V cells," said Vicente Díaz, manager of the company s CPV business unit, adding that the impending commercial realization of CPV would represent the culmination of a decades-old dream since the approach was first postulated in the 1970s. "Terrestrial applications are now, finally, creating demand," he said. Raising concentration to the 1000-sun level would help the drive to reduce the cost-per-unit figure, although it would present even greater challenges for the cell manufacturers and system integrators.

One of the difficulties is with the uniformity of wafers containing the III-V cells. As a result, Isofotón has developed its own SAMCEL test equipment. Based on a flashlamp set-up, it allows engineers to test the uniformity of 4000 III-V cells per hour, although Díaz says that this crucial part of the production process still needs to be improved.

Díaz also shed some light on the likely volumes of germanium wafers that would be required to support the broader deployment of CPV. He reckons that as much as 100 MW may be realized annually just 2–3 years from now. This would demand 150,000 4 inch germanium wafers per year.

At the moment, however, Isofotón s CPV projects remain relatively small, with a 16.2 kW (peak) system recently completed and installations in Barcelona, ISFOC and Malaga University in progress.

Something that has characterized the III-V solar cell industry thus far is the relative paucity of cell producers, with only Spectrolab and Emcore manufacturing in anything like production-scale volumes. This now looks certain to change, despite the high level of expertise required. Speaking in Key West, Kurtz listed new US entrants, including Spire Semiconductor and MicroLink Devices, and a whole host of other potential suppliers from as far afield as Japan, Italy and Taiwan (table 1). More recently, MOCVD epitaxy foundry Kopin has thrown its hat into the ring too, marking a return to one of the first applications that the company targeted, back in the late 1980s. Kopin CEO John Fan said: "We have substantial intellectual property and manufacturing know-how around this technology."

Kurtz also highlighted some major challenges specific to III-V cell makers. For space applications, with small volumes and large-area cells, manual testing and mounting is the norm. But for terrestrial CPV, with far smaller cells and higher expected volumes, automated production will be essential. Cells must also be attached to heat sinks and optical components, demanding high-performance bonding materials. For this, the photovoltaic cell makers may be able to learn from the high-brightness LED industry, which has had to go through a similar learning curve to solve such thorny issues.

The LED industry s use of flip-chip and substrate removal processes could also be useful for a new type of triple-junction cell pioneered by Emcore and NREL. The technique, which uses an inverted semiconductor material stack to achieve a cell efficiency of up to 38.9%, also involves removing the GaAs substrate, leaving behind a very thin cell that must be mounted and bonded carefully.

Further challenges are certain to emerge, but, as the list of companies involved in CPV grows (see box "Companies working on CPV systems"), the collective expertise should grow, and the emerging community should ensure that it benefits from the experiences of LED and RFIC manufacturers.

What seems to be required is a healthy awareness of the challenges that remain. Molnar said: "Many start-ups have been able to produce interesting photovoltaics at a pilot scale but have found it challenging to scale their technology to commercial size without problems with efficiency and yields."

However, he is confident that, with so many companies and so much venture finance now at stake, the probability of a handful of them emerging successfully from the current wave of new entrants is very high. With a collaborative effort from cell makers, CPV companies and the existing III-V industries, at least one of those successes ought to be based on compound semiconductors.

• Copies of the 2008 Key Conference proceedings are now available to purchase from this website. Visit compoundsemiconductor.net/key for details.