**1. Introduction**

420 Solar Cells – Thin-Film Technologies

68-75, ISBN 0-7803-1460-3, Waikoloa, Hawaii, USA, December 5-9, 1994 Minemoto, T.; Negami, T. Nishiwaki, S. Takakura, H. & Hamawaka, Y. (2000). Preparation

Minemoto, T.; Hashimoto, Y. Satoh, T. Negami, T. Takakura, H. & Hamakawa, Y. (2001).

372, (August 2000), pp. 173-176, ISSN 0003-6951

ISSN 0021-8979

BASED SOLAR CELLS: PROCESSIMG OF NOVEL ABSORBER STRUCTURES, *Proceedings of the First World Conference on Photovoltaic Energy Conversion 1994*, pp.

of Zn1-xMgxO films by radio frequency magnetron sputtering. *Thin Solid Films*, Vol.

Cu(In,Ga)Se2 solar cells with controlled conduction band offset of window/ Cu(In,Ga)Se2 layers. *Journal of Applied Physics*, Vol. 89, (June 2001), pp. 8327-8330,

> The advances and promises of thin-film photovoltaics (PV) are much discussed these days, typically using the viewpoint that a picked technology and process approach would provide "the" solution to many problems experienced implementing PV commercialization. In 2009, a thin-film PV company, First Solar, garnered world-leadership as a PV company, being the first company to produce or ship more than 1 GW of PV modules in a single year. This makes it timely to discuss the advantages and limitations of thin-film PV technology, as compared to the currently prevailing crystalline Si PV industry. Traditionally, the following technologies are considered constituting "thin-film PV:"


In the amorphous silicon (a-Si:H) based category, several approaches are pursued, ranging from amorphous silicon single junction modules to spectrum splitting multijunction cell structures using either a-SiGe:H cell absorbers or a-Si:H/nc-Si:H multijunctions. Pros and cons will be given for these different approaches that lead to this multitude of device structures. It is argued that as long as the advantages of the aforementioned materials are not understood, it would be difficult to "design" materials for more efficient solar cell operation.

This review will recap what is currently known about these materials and solar cell devices, keeping in mind that there will always be some unexpected "surprises," while there were many other approaches that did not result in anticipated cell/module performance improvements. This knowledge leads the author to ask the following question: "Was improper implementation or inadequate process choice responsible for the lack of solar cell/module performance improvement, or was the expectation for improved device performance or decreased device cost simply not warranted?"

The chapter of this book is written such as to not prejudge an outcome, i.e., an a priori assumption that a given measure would result in a commensurate expected performance improvement. The impact (i) of an improvement is broken down into probability (p) of achieving a projected improvement times the effectiveness (e) of such improvement, where

What is Happening with Regards to Thin-Film Photovoltaics? 423

was, however, found that a critical CdCl2-anneal step is crucial to achieve best solar cell or module performance (McCandless, 2001). Anneal temperatures on the order of 400 oC are typically used after the CdCl2 exposure. For industrial production rates, it is important to limit the time for such anneal step in order to achieve an appropriate throughput. Looking at current commercial throughput rates, one has to conclude that this is possible. It was also attempted to substitute this CdCl2 anneal step (where CdCl2 is often applied as an aqueous solution] with a gaseous anneal step using HCl dry gas (McCandless, 2001). While this approach resulted in similar results as the aqueous CdCl2 anneal step, a superiority using

CdTe cells can be made stable and lasting, but not all production schemes result in stable cells. It was reported that excessive reliance on the CdCl2-anneal step to obtain the highest cell or module efficiencies often led to less stable devices (Enzenroth et al., 2005), with processes leading to the highest pre-anneal efficiency often resulted in the most stable manufacturing recipes. It is now known that Cu, applied to many back-contacting schemes, is correlated with the stability of CdTe cells. While it has been established that "too much" Cu results in unstable cells, some rather stable cell deposition schemes were developed that use Cu-doped back contact recipes. The degradation process shows a mixture of diffusive and electromigration behavior (Townsend et al., 2001). Alternatives to using Cu for the back contact were developed (e.g., P-doping, N-doping) (Dobson et al. 2000). These 'Cu-free' recipes also showed instabilities and did so far not improve cell performance over that achieved with stable Cucontaining back-contact recipes. Perhaps, it is a flaw to ask: "Is Cu in the back contact good or bad for cell stability?" The appropriate question may well be: "When is Cu good, when is it

While all commercial CdTe solar modules are currently fabricated in a superstrate configuration (using a glass superstrate), the question has been posed whether such process could be inverted and/or be applied to flexible substrates. Flexible substrates (like polyimide foil) limit the temperature that can be applied during the position process. Also, the issue of low-cost hermetic packaging of such transparent foils has to be addressed in greater detail in a cost-effective manner. Because glass-encapsulated PV works, the cost of glass (on the order of \$10/m2 for a single sheet) can often be used as a cost-guideline for terrestrial flexible packaging schemes for power modules. It is clear at this juncture that CdTe PV and CIGS PV have greater moisture sensitivity than many Si PV schemes, requiring a more hermetic seal than Si PV might require. A point of research continues to be the "edge delete" for modules. Typically, SnO2 coated superstrates are coated with all layers of the entire glass surface. A fast removal of such films, including the SnO2-layer, along the module edges is required. For CdTe modules, often rather crude methods (like beadblasting or using grinding wheels) are employed for this "edge delete" step were employed. The drawback of employing these methods is that glass surfaces are damaged using such processes, resulting in greater water penetration rates from the module edges. Also, such processes tend to weaken the glass. However, less damaging edge delete techniques like

this "dry" process could not be established .

bad, and when is it irrelevant for cell performance and stability?"

laser ablation methods are rapidly becoming feasible and more cost effective.

In order to make a monolithically interconnected module, cell "strips" have to be created that carry CdTe currents through the SnO2. Typically, 1 cm-wide cell strips are used for CdTe modules. These strips require 3 scribes sometimes labeled P1 (SnO2), P2 (semiconductor layer), and P3 (back-contact) scribe line. The area including and between scribes P1 and P3 is electrically "dead" and does not contribute to module power, hence reducing the total area module efficiency. Therefore, scribe lines should be narrow and close

It is of interest to note that while impact is costed and/or priced by many companies, the right hand side of the above equation also has associated cost elements associated with effectiveness e plus an estimated probability p. Probabilities (p between 0 and 1 or 0% and 100%) are often assumed to be either 0% (for an unsuccessful project) or 100% (for a successful project), with the benefit of hindsight. This is true only with the benefit of hindsight, forward looking probabilities should be estimated and accounted for as accurately as possible. In financial terms, a probability between 0 and 1 should be accounted for by applying appropriate financial discounts to probabilities falling outside the extreme values, 0 or 1. Instead, often p=1 is being "assumed," but strictly speaking, this is inadmissible in forward-looking situations. Whenever p is increased at the expense of e, the total benefit for i may not be achieved as planned. Typically, p has to be empirically assessed, which is important for appropriate financial "discounting" leaving much room for discussion as to what value (between 0% and 100%) to assign to p. The foregoing statement is valid for all PV technologies (not just thin-film PV), but in the following, mainly elucidated picking thin-film PV examples. This chapter does not want to chime in on a debate about what appropriate probabilities or discount factors should be used, but rather serve as a reminder to the fact that projected probabilities occur with less than 100% probability.
