GreenMountain Engineering

Standards for Photovoltaics

I recently returned from a week spent as one of the US representatives on the IEC Technical Committee 82 (TC82), which develops standards for solar photovoltaics. My focus is on Working Group 7, where we are working on standards including solar trackers and power and energy rating of concentrating photovoltaic modules. However, meeting with the rest of the TC82 community also gave me an opportunity to discuss issues in the design and testing of conventional wafer-based and thin-film modules.

While there are a number of groups working on standards applicable to solar (including the UL, ASTM, and NEC in the United States, CENELEC in Europe, and IEC and ISO internationally), the IEC plays an especially significant role because photovoltaics is a global market: major producers of polysilicon, cells, modules, and tools are spread across Europe, Asia, and the Americas, and customers are worldwide as well. Similarly, TC82 has members from 29 countries (including the major markets; China, Spain, Germany, the US, Japan, and France) working together to develop standards.

Standards are an important part of any growing industry.  For example, the SEMI International Standards Program is widely credited with speeding the growth of the semiconductor industry since the 1970s. Looking back, it’s hard to believe that at one point wafer sizes and shapes were not standard, and “custom-made solutions for each individual customer were the norm”[1].

For an industry whose value proposition depends on long product lifetimes in outdoor environments, standards that govern design qualification, accelerated testing, and safety of products are especially important. Additionally, standards for power rating, energy rating, and measurement are critical for allowing side-by-side comparison of different products.  This is especially apparent when trying to compare between crystalline PV, thin-film PV, and CPV.

Some of the PV standards we find most relevant in our work[2] are listed below. Where possible, I’ve also linked to free previews of the table of contents of each standard:

• IEC60904: Photovoltaic Devices
• This is a large, ten-part standard (IEC60904-1 is the numbering scheme for part 1, and so on) covering a number of device characterization areas such as measurement of I-V curves, spectral response, and solar simulators.

• IEC61215 (ed2.0, 2005): Crystalline silicon terrestrial photovoltaic (PV) modules: Design Qualification and Type Approval
• Note that a third edition is in development within the IEC.
• UL1703: Flat-Plate Photovoltaic Modules and Panels
• Note that this is also applied to thin film modules and in some cases in the past concentrating modules, though see also UL8703 below.
• IEC61646 (ed2.0, 2008): Thin-film terrestrial photovoltaic (PV) modules: Design Qualification and Type Approval
• IEC62108 (ed1.0, 2007): Concentrator photovoltaic (CPV) modules and assemblies: Design Qualification and Type Approval
• Note that for low concentration (<10x) modules, it is less clear whether they will be tested under IEC 62108 or an adapted form of IEC 61215. And some concentrating systems such as heliostats differ from the main focus (no pun intended…) of IEC 62108.
• On the IEC committee we are actively soliciting feedback on the first edition, as we work on a second edition.
• UL8703: Concentrator Photovoltaic Modules and Assemblies
• UL1741 and the just-published European standard EN50530 cover inverters
• The PV Resources web site contains a more exhaustive list of standards, though it is somewhat out of date and does not mention some of the newest standards. And the standards above cover a significant portion of what companies we work for care about.
• You may also find this UL diagram of UL/IEC PV standards by system component informative.

For anyone who was new to the industry, I hope this list of information is useful. 

That said, qualification standards only outline the bare minimum testing. It’s important to design tests to simulate other failure modes and environmental conditions not included in the standards. In addition, testing identifies certain failure modes that are more systematic (damp heat for the previous generation of thin-film modules, for example), helping guide areas for design.  Including testing early in the development path is important: we have seen some companies develop a first prototype, only to require major design revisions once they begin thinking about reliability, DFM, and qualification. 

The topic of PV module design for reliability could be a whole separate discussion, but two documents to get you started down this path are:

• NREL's "non-comprehensive summary of the most common types of failures for each PV technology", with 53 references to additional papers
• Another paper out of NREL, published in 2009: "History of Accelerated and Qualification Testing of Terrestrial Photovoltaic Modules: A Literature Review"

Standards themselves don’t necessarily inspire passion and dedication in everyone, but their purpose overlaps with the desire to design and build high-quality, reliable, cost-effective technologies that can solve some of our pressing energy supply and environmental issues. And doing that makes us at GreenMountain very excited. Come back to this blog next week for a post about the scalability of solar, examined from a variety of perspectives (land usage, capital requirements, labor, and growth rate). 

[1] The SEMI International Standards Program – History, Successes and Lessons Learned to Address Compound Semiconductor Manufacturing Challenges, http://www.csmantech.org/Digests/2006/2006%20Digests/4A.pdf

[2] I'm putting the "about us" blurb down here in a footnote as many readers may already be familiar with us: We offer design engineering for hire, including engineering of products, automated tools, and software for many companies in the cleantech sector. This includes extensive experience in the solar industry (we’ve done design engineering for dozens of solar companies). The product-related standards I mention in this post are less relevant when we develop manufacturing tools, but do come into play when we design solar modules, encapsulation, interconnects, CPV receivers, or a range of other solar components.

Thin-Film Solar: Interpreting Efficiencies

Reporting the efficiency of a solar cell or module depends on a number of assumptions, and unrealistic assumptions are sometimes made in order to report the highest possible efficiency.

There are some cases where it is difficult to make fair and consistent assumptions. For example, when comparing different technologies (thin film, crystalline silicon, and CPV) which have different values for temperature coefficient, spectral dependence, land coverage, and other properties, it can be difficult to come up with a completely standard, comparable, and realistic method of rating power and efficiency.

However, within a particular type of technology (thin-film solar, for example), there is really no excuse for making non-standard assumptions and then omitting this critical information when you report your results. For this post I’m going to focus on particular misleading reporting practices I have seen used multiple times in the thin-film industry (that is, solar technologies such as CIS, CIGS, CdTe, and thin-film silicon, which includes amorphous silicon, and “micromorph” tandem devices).

I do understand that this poses a challenge to each individual company: If your competitors are reporting their efficiency in a misleading way, it is difficult to report your own results in the “correct” way if it gives the appearance that you have a lower-efficiency product. This is a form of arms race or “Prisoner’s Dilemma”, but this is also not my problem: my point here is to make sure people who read results know the right questions to ask to make sure companies are giving them realistic numbers with full disclosure on the most critical measurement assumptions.

Assumption: Aperture Area

One detail often glossed over is the notion of aperture area compared to full area. An aperture area efficiency for a module divides the output power by only the area covered with the active absorber, not including the border around the module where the edge seal is or the frame. In addition, in some cases (especially during R&D), companies mask or scribe away regions of the absorber where there is significant thickness variation due to edge effects in the deposition chamber, and don't include those edge regions in the measurement. A 25mm border on a Gen 5 glass sheet (1.1 x 1.4m) corresponds to about an 8% relative difference in area, or a roughly 1% difference in reported absolute efficiency for a 12% efficient module.

Note that in some cases aperture area is a reasonable way to report efficiency, especially during R&D when you want to report the overall potential of a technology and ignore some process issues, or if the deposited materials are a large fraction of your cost and your substrate is a low-cost plastic (in which case extra area of this plastic may not add as much cost as active module area). However, it’s important to clearly state that an aperture efficiency is what’s being reported.

Assumption: Light-Induced Degradation

Another reporting issue most relevant to thin-film silicon is the reporting of initial efficiency compared to stabilized efficiency. Stabilized efficiency is the relevant number for real-world applications, and refers to performance after the degradation that amorphous silicon undergoes when exposed to light for the first time (the Staebler-Wronski effect, also referred to as LID for Light Induced Degradation). However, some companies report the higher initial/unstabilized efficiencies, and only mention this is what they’re doing if you ask them directly “is that a stabilized efficiency?” This is important because a typical Staebler-Wronski degradation can be 10% relative, or even more. At the recent EUPVSEC, a number of companies mentioned that their LID at the cell level was 10%, but then were vague about their LID at the module level, or said something along the lines of “there are some challenges with LID on large area [Gen 8.5 glass = 2.2m x 2.6m!] modules,” effectively telling the audience that LID was significantly worse than 10% in their modules.

In other cases, thin-film manufacturers would report the power output and area of modules, but not the efficiency, or even the area, initial power, initial efficiency, and stabilized power, but not the stabilized efficiency. This is a bit silly, as anyone in the room can calculate the stabilized efficiency from these numbers.

So, thin-film manufacturers: If you’re going to show the performance of a new cell or module in a conference presentation, I suggest you include a summary table like the below, rather than making your audience guess or ask you about your assumptions. It only takes one slide, and saves everyone time; and if you don’t do this, savvy people in the audience realize you’re trying to pull the wool over their eyes when you are vague about your measurement conditions.

			Area	Power [W]	Efficiency [%]	LID [% relative]
Cell	Initial	Aperture Area	1 cm2		11.1%
Cell	Stabilized	Aperture Area	1 cm2		10.0%	-10%
Module	Initial	Aperture Area	1.5m2	143	9.3%
Module	Stabilized	Aperture Area	1.5m2	123	8.0%	-14%
Module	Initial	Total Area	1.5m2	136	9.1%
Module	Stabilized	Total Area	1.5m2	116	7.8%	-14%

(the numbers above are arbitrary, just to provide an example)

Or if you really don’t want to use a table, a least attach the most relevant test details to your reported values, for example:

• 10.4% (1cm² Cell, Aperture, Stabilized)
• 9.3% (Gen 5: 1.1x1.4m, Module, Aperture, Initial)
• 8.7% (Gen 8.5: 2.2x2.6m, Module, Aperture, Initial)
• 7.8% (Gen 8.5: 2.2x2.6m, Module, Aperture, Stabilized)

Assumption: Deposition Rate

Another question to ask when new efficiency results are reported is: “What was the deposition rate used to achieve the reported efficiency, and is that a realistic throughput according to your cost models?”

Typically, film quality and electrical efficiency go down as the deposition rate goes up. So a company may be able to get a higher efficiency “champion” module by using a very low deposition rate (one that is too low a throughput for volume manufacturing) when making that particular module. This result does have some value: it shows that a certain module efficiency is technically possible on large areas, and perhaps in the future with improved tools and processes, that efficiency might be achievable in manufacturing.

However, in production, companies can run cost models and pick some deposition rate that is a balance of throughput and module efficiency that leads to the best overall cost per watt and cost per area for the module. It’s not a trivial calculation because as efficiency goes down, module area at a given power goes up, so balance of system costs such as installation and land also go up slightly. But it’s also not economically sensible to use a process with dramatically lower throughput just to get an extra 0.1% of efficiency.

Assumption: Deposition Process

Another issue to consider is whether the fabrication processes used are industry standard ones, or less-used processes. If the latter, is there a path to scale-up of the process? It is certainly acceptable to explore new deposition processes in pursuit of better performance, lower cost, or even proof-of-concept devices. However, if a company is presenting results in an established industry where there is a common deposition tool and method (for example, a-si deposited by 13.56MHz PECVD), the company should identify if their devices were made with a different process (VHF plasma, HWCVD, or so on).

Conclusions

There are a number of common ways in which efficiencies reported by thin-film solar companies can be difficult to interpret and compare, based on variation in how the numbers are calculated. Three of the most critical are whether an aperture-area measurement is made, whether light-induced degradation of thin-film silicon is included, and whether the deposition rate was unrealistically low. It is the responsibility of companies to clearly disclose their measurement assumptions, and also of people receiving results to demand this disclosure. I have included a very simple table for reporting results which would avoid the need for perhaps 40% of the questions I have heard asked of thin-film manufacturers at conference presentations. Now let’s all get back to engineering and solving problems!