Panda is the name of the high efficiency module line from Yingli Green Energy (NYSE: YGE). Yingli is one the largest solar PV manufacturers in the world. It just surpassed Suntech (NYSE: STP) in Q1, 2012 to become the number one company in category of module shipments. Developing high efficiency modules is a response to an industrial trend in the past few years. It is also a way for the top PV companies to differentiate themselves from smaller competitors. While all the top module companies have their own high efficiency module line, Yingli’s Panda is well received and has shipped in volume since its debut in 2010. In contrast, Suntech’s Pluto has been plagued by manufacturing problems from its start back in 2007, although recently the situation appears to have improved. Last month, we also reported the mass production problem with Canadian Solar’s (NASDAQ: CSIQ) ELPS.

The technology behind Panda involves selective emitter cell built on n-type monocrystalline wafer. It is well-known that cells made with monocrystalline wafers carry higher efficiency than the main-stream multicrystalline wafers. While typical monocrystalline wafers are p-type when doped with boron, Panda’s n-type wafers are doped with phosphorus. Since n-type wafers are more tolerant to chemical crystallographic defects, they have exceptional potential for high efficiency commercial solar cells. Also, there is no light-induced degradation for n-type cells. It is estimated that n-type wafers can achieve efficiencies 5-8% higher than p-type. A second benefit is that it is easier to make cells to be bifacial with n-type wafers. Double-sided cells have better performance under weak sunshine and can better absorb scattered sunlight. A third benefit with n-type wafers is that they can be cut thinner than p-type – hence saving some polysilicon material. Of course, the savings from thinner wafers are less significant these days due to the plummeting polysilicon prices. Also, n-type wafers tend to have lower efficiency degradation rate than p-type wafers.

Panda is the result of collaborations between Yingli, ECN of The Netherlands, and Tempress Systems B.V, which is also based in the Netherlands. Tempress is a wholly-owned subsidiary of Amtech Systems, headquartered in Arizona. ECN is a research institute which provides key technologies while Tempress is a leading equipment maker, which excels at phosphorus doping (a key process in making n-type wafer). Tempress also provides complete cell solution for Yingli, which includes saw-damage removal & texturing, self-aligned selective emitter & back-surface-field formation, glass removal, passivation & anti-reflective coating, contact printing & firing etc.

Yingli stated Panda achieved an average cell efficiency of 18.5% in 2010 and 19% in 2011. Its latest module efficiency, based on the power output data published at its website, ranges from 14.9% (245W) to 16.2% (265W). However, Yingli often claims it has achieved a commercialization efficiency of 16.5%. This is seemingly related to the non-uniform power output within the same nameplate modules. Indeed, there are tests showing that Panda modules produce an output of 270W, equivalent to 16.5% efficiency. Since there are likely fluctuations in power output even in the same batch of modules, Yingli wants to be conservative in labeling its modules in order to avoid complaints.

Compared to latest cell technologies, Panda’s cell efficiency is not particularly impressive. Last month, an equipment maker, Centrotherm, announced that it achieved an average of 19.5% cell efficiency on over 50MW production lines using its “centaurus Technology”. Though lacking specifics, 19.5% cells use typical p-type mono wafer. However, modules made with Panda cells have very good CTM (cell to module degradation), supporting further benefits of n-type wafers.

While Yingli wanted Panda to be a household name, details on the progress of Panda’s production lines and shipments are murky. From Yingli’s annual report, it said it started Panda project in June 2009 and reached a first-phase capacity of 300MW in 2010. By the end of 2011, it stated it has accomplished a second-phase expansion of additional 300MW, reaching total of 600MW capacity. A further investigation revealed Yingli finished the construction of first 300MW in mid-2010 and the full production started in Q2, 2011. Despite the 600MW capacity at the end of 2011, it is unlikely that full utilization has been reached.

Yingli did not disclose details about Panda’s shipments in its 2010 and 2011 annual reports. In a recent interview, a Yingli official said that the company shipped 200 MW Panda modules in 2011, a figure which beat easily any other manufacturer’s high-efficiency shipments. In the same interview, for the first time, manufacturing costs were revealed. On average Panda module costs $0.15 per watt more than typical multicrystalline module. This is not a surprise as the technology used is unique and relatively expensive. The extra costs seemed to be in line with prior estimates: typical p-type mono wafers cost $0.06 to 0.07 more than p-type multi wafers; n-type wafer should be $0.04 to 0.05 more than p-type due to the differences in the doping process and relatively lower production scale; the rest of the costs are in the cell manufacturing contributing remaining few cents. Yingli stated its goal is to reduce the cost gap from 15 to 10 cents (compared to multi modules).

There are critics of the technology Yingli chose for Panda. Renesola’s (NYSE: SOL) outspoken CEO, Xianshou Li, once said n-type wafer is one hyped technology because the efficiency improvement does not justify the extra cost. Li is an expert on wafers so one should respect his view. Employees at Trina Solar (NYSE: TSL) also commented that Panda costs trailed far behind its Honey in contributing to company’s bottom line. Indeed, at $0.15 in extra costs, Panda does not appear to bring financial benefits to Yingli. In Q1, 2012 conference call, Yingli’s CFO said Panda’s modules have 10% premium over regular multicrystalline modules. While pricing of multi modules expected to dip below $0.80 per watt, conditions to extract positive gross margins on Panda would only deteriorate.

While those results question value of the line to the company’s bottom line, Yingli is already looking for a second generation Panda-cell development. The company teamed up with Heraus in $70M worth R&D and supply project, to develop new silver metalization pastes to reduce costs. Collaboration with DuPont to improve cell efficiency was announced earlier this year. The relationship with ECN and Tempress carries on with development of MWT (metal wrapped through) n-type cell. In the lab setting, this type of a cell reached 19.7% efficiency and module built with them delivered 17.6% conversion worth 288W of output. The new cell costs were estimated to be similar to currently mass produced cells,  with added cost for drilling holes etc for the MWT structure.

In pursuit of the best case scenario there seems to be a market shift to cheaper technology like multicrystalline to reduce overall costs, but using delivery system of high-conversion cells to achieve the work of more expensive material like mono. While some companies like Nexolon see the gap between mono wafer pricing and observe gross margin benefits in producing n-type wafers, a dawn of high conversion multicrystalline modules will certainly dent a demand on high-cost, premium required mono products. Unless mono products, including Panda module, can garner significant benefits and exceed significantly conversion of multi modules, the popularity of Panda may decrease, cooling off Yingli’s enthusiasm for this product’s growth and further development.