A manufacturing technique capable of depositing atomically thin layers of aluminum oxide is reshaping the market for industrial equipment to produce state of the art photovoltaic cells. Manufacturers across China are switching from plasma-enhanced chemical vapor deposition (PECVD) to atomic layer deposition (ALD) as the new method of choice to deposit aluminum oxide passivation layers for PERC solar cells.
“We have so far equipped over 30 GW of manufacturing lines with ALD passivation systems and are still seeing growth,” says Wei-Min Li, CTO of Leadmicro, the leading supplier of ALD passivation equipment. “Almost all major PV cell manufacturers in China now use our system.”
Passivation layers are at the core of PERC technology. They are deposited between materials in these cells to reduce charge recombination at their interfaces. Their inclusion has helped boost energy conversion efficiencies above 22% in cutting edge PERC production. Manufacturers have typically grown aluminum oxide passivation layers using PECVD, a long-established process in the PV industry. ALD entered the PV sector as an alternative approach in 2011, but sparked little initial market interest. The situation changed abruptly last year, when a young company based in Wuxi, China shifted ALD adoption into gear.
New kit on the block
Upstart equipment supplier Leadmicro has brought together early pioneers of ALD technology benefiting from Chinese financing and research experience abroad. The company is hammering the market for manufacturing equipment to deposit aluminum oxide in PERC cells with an in-house variant of ALD tailored for mass production. From a market share of under 2% in 2017, Leadmicro claimed close to 20% of sales in 2018, becoming the second largest supplier of PERC equipment on the market. In just three years, Leadmicro has grown to 300 employees and signed contracts with heavyweight PV manufacturers including Tongwei Solar.
The success of Leadmicro and a handful of ALD competitors has encroached on sales of PECVD equipment. Li says that manufacturers are moving in droves towards ALD because of the technology’s lower cost of ownership. He adds that operation and maintenance expenses to deposit aluminum oxide passivation layers by ALD can be as low as $0.001 a wafer. These costs come on top of other process steps along the factory line, including handling steps to transfer wafers between the ALD system and PECVD equipment when depositing subsequent layers in the cell.
PERC factory lines that incorporate an ALD system still require PECVD equipment, notably to deposit a silicon nitride capping over the aluminum oxide. These layers have conventionally been deposited inside the same PECVD reactor. Using ALD requires interrupting the vacuum deposition process between the two steps and transferring samples by hand. Still, Li claims that the final bill works out favorably for Leadmicro clients.
Leadmicro also reports technological benefits in using ALD to deposit aluminum oxide passivation layers, in particular in terms of the energy conversion efficiency of PERC solar cells. “We have seen from different production sites that ALD has enabled PERC cells to reach energy conversion efficiencies about 0.05% higher than cells with passivation layers deposited by PECVD,” says Li. “ALD is an obvious choice for high-efficiency solar cell manufacturing.”
Improvements to energy conversion efficiency are notoriously challenging to isolate and some companies producing top-range PERC cells, including Hanwha Q Cells and SAS, continue to use PECVD, not ALD. However, if established, even modest benefits in energy conversion efficiency could prove a game-changer for either technology. On the multi-gigawatt scale over which current market leaders compete, each basis-point in energy conversion efficiency accounts for millions of dollars – enough to make or break the business case of PV manufacturers.
Bram Hoex, a specialist in surface passivation and associate professor at the University of New South Wales (UNSW) says that given the potential for further improvements in PERC cells, any efficiency differential between ALD and PECVD is destined to grow more pronounced over time. “The PERC solar cell is relatively new in high volume manufacturing and can reach efficiencies above 24% in production,” he says. “Any imperfections will hurt performance more the closer we get to the fundamental limit of the technology.” At current PV efficiencies, the quality of the rear-side passivation is not a decisive factor when it comes to overall power output. But as efficiency continues to inch higher as manufacturers optimize other manufacturing processes, the importance of the quality of passivation will likely also increase.
Leadmicro claims that ALD also presents tangible benefits for manufacturing lines. The company takes pride in the wafer throughput that its equipment achieves. “When we entered the market in 2016, PECVD could handle only 3,000 wafers an hour,” says Li. “Our first generation of ALD systems processed 5,000 wafers an hour.” At SNEC PV Power Expo in June this year, Leadmicro plans on revealing the latest ALD system in its Kuafu line of products designed to passivate 10,000 wafers an hour.
High throughput is an unconventional sales pitch for an ALD equipment supplier. The deposition process has traditionally been regarded as inherently slow, and has in the past been valued more for its tight control than for high throughput. ALD emerged in the semiconductor industry 40 years ago to deposit flawless thin films for electroluminescent displays. The process achieves unprecedented deposition conformity using self-limiting chemical half-reactions. In the case of aluminum oxide deposition, wafers are exposed to trimethyl aluminum (TMA) which reacts with hydroxyl groups on the sample surface resulting in a methyl group covered surface. TMA does not react with methyl groups, so film growth automatically stops. After removing the TMA from the reactor, water is injected to form a hydroxyl-terminated surface on which TMA can react again. The gas sequence is repeated until the desired film thickness is obtained.
The low pinhole density and high conformity of ALD are critical to applications in the integrated circuit industry such as coating deep trench structures. But the slow deposition rate has presented barriers for adopting ALD in PV manufacturing. Whereas PECVD lines in the PV industry can reach deposition rates above five atoms a second, ALD counterparts lag far behind. “In the lab, ALD is quite a slow process with every ALD cycle taking at least several seconds to complete,” says Hoex. Another long-standing barrier is that material deposited by ALD tends to wrap around the edges of samples. “This is less of a concern with PECVD,” says Hoex. “In the case of ALD, in essence every exposed surface will be coated.”
In recent years, Leadmicro has been instrumental in overcoming both these challenges. Rather than focus on masking the back side of samples, the company has learned to embrace the conformity of ALD deposition and put it to PV’s advantage. Building on research conducted by Hoex’s group in 2013, the company has adapted the properties of its aluminum oxide to enable it to passivate both the p-type and n-type surface of PERC cells simultaneously. “We used advanced computer simulations to identify the optimum charge concentration for passivating the p-type silicon without hampering the n-type emitter on the front of the PERC cell,” says Hoex. His team subsequently adapted ALD recipes to tailor the charge level and allow ALD to safely deposit aluminum oxide on both sides of the cell.
This know-how has since freed Leadmicro from the need to mask samples inside its ALD system. In its batch process, wafers stand next to each other, densely packed, as they are exposed to oxygen and aluminum precursors. This vastly increases the number of wafers that can be processed in a single run. “ALD used to load one wafer at a time, today we feed 800 into each batch,” says Li. The new configuration has required strenuous optimization on issues ranging from gas flow and automation to mechanical constraints. But according to Hoex, batch processing has produced a simple workaround to low ALD deposition rates and transformed the market for ALD.
Hoex also says that the spread in PERC cell results is remarkably narrow when using passivation layers deposited by ALD. This year, his group also reported at Silicon PV, a scientific conference, that the density and conformal nature of ALD-deposited aluminum oxide layers help reduce light and elevated temperature induced degradation (LeTID) in PERC silicon solar cells. Last year, UNSW announced a collaboration with Leadmicro to investigate broader applications for ALD. “We are looking at passivating contacts, transparent conducting oxides, buffer layers…” says Hoex. “We hope that aluminum oxide is just the beginning of the story and are confident that ALD will introduce new materials to drive down costs and improve efficiencies across the PV industry.”
The future of ALD may present challenges as well as opportunities. As field observations build up on the ultimate cost of ownership of PERC manufacturing lines incorporating ALD systems, the PV market is already trending towards higher efficiency architectures such as TOPCon. These upgrades stand to benefit from integrated production processes. Although ALD is already being used in industry to deposit passivation layers for TOPCon solar cells, best-in-class efficiencies are achieved with PECVD. To continue growing, ALD will have to prove its adaptability in line with evolutions in PV technology.
In addition to issues of technology, PERC equipment suppliers also face legal hurdles. This year, Hanwha Q Cells filed lawsuits against PV manufacturers for adopting passivation layers that the company claims are protected by its intellectual property.
The South Korean-headquartered firm claims that its patents cover any PERC passivation stack made of aluminum oxide capped with a hydrogen containing dielectric. The original patent refers repeatedly to the use of ALD in manufacturing these structures. Manufacturers need to stay clear of intellectual property issues if they want to sell their products abroad. As a result, the outcome of these cases may influence the ongoing market adoption of ALD and the future of PECVD equipment suppliers alike.
“In China, good news travels fast,” says Hoex, referring to the remarkable momentum with which ALD has spread so far across the country’s PV industry. “The flip side is that bad news travels even faster.”