There’s only so much you can cover in fifteen minutes.

Conference presentations, by design, have to leave out many specifics. However, I find this makes it very difficult to tell if there was a lack of rigor in some aspect of an experiment, or if the standard method is widely known and thus not described. When I’m surrounded by people with decades of experience, I tend to assume that if they’re not calling the presenter out, then I definitely shouldn’t be. My ignorance is far more likely than insightfulness.

This state of confused skepticism was my headspace while listening to the presentation by Shu-Hung Yu of Motech Industries on their new bifacial photovoltaic technology at the 40th IEEE Photovoltaics Specialists Conference (PVSC) in Denver. Sipping my (ever-present) coffee with a furrowed brow, I weighed the option of asking a (hostile) question. Luckily, I didn’t have to. Dr. Steven Hegedus of the University of Delaware instead took to the microphone. I can’t remember his words exactly (I’m no journalist), but it his questions were along the lines of:

–          What is the point of bifacial cells anyways?

–          What was the particular experimental arrangement during the system performance measurements and will it ever play a significant role in the real world?

These are crazy-important questions and ones that the speaker could not answer.

Dr. Hegedus was not alone in his skepticism. A lunch with PVIN colleagues during the week brought up similar questions. I myself was not completely won over by the talks discussing bifacial concepts. It did not help that a couple of the bifacial talks were in the same session where Sharp, Sunpower, and Panasonic all announced new record efficiencies based on all-rear contact solar cells.

It was not entirely the fault of the presenter that the above questions were not easily answered. They are not ones that most photovoltaic talks have to deal with because certain assumptions have already been made by the larger community and encompassed in standards. So, while both of Dr. Hegedus’ questions can be answered to some degree, there are still clear and challenging gaps. Below, I’ll give a brief description of what bifacial cells are and their situational applicability. Following that, the current gaps in bifacial reporting standards will be outlined. With this information, the “offending” talk can then be properly critiqued and we’ll see how community knowledge gaps can lead to frustration with small talks. I’ll suggest some steps we should take in order to ensure credibility in analysis of bifacial technologies. This is a pretty editorial post, but I’ve attempted to back up most of my claims.

What are Bifacial Solar Cells?

Most cells or modules contain opaque metallic reflectors on their rear to increase their efficiencies. The idea is to reflect any light that isn’t absorbed on the first pass through the cell back through the absorbing layer. This means that they only absorb sunlight that is incident on their front surface. These are called monofacial cells and modules.

If one removes this reflective layer, the solar cells can generate power from illumination incident on both the front and rear faces. While the front-side efficiency will be reduced, the light that reflects of the surrounding environment and onto the rear of the module may more than make up for it. Cells and modules that allow light to be absorbed from both sides are referred to as bifacial.

When Should Bifacial Solar Cells be Used?

Hopefully the description itself shows why their potential use is somewhat controversial. While monofacial solar cells are optimized for solar illumination on the front and have a performance only* dependent on the atmospheric conditions, the power generated by bifacial cells depends much more on the surrounding environment. Depending on what materials are behind or around bifacial panels, how closely the panels are spaced, and their height off the surface, the power generation will differ significantly.

There isn’t really any point to use bifacial modules in arrangements like the one shown in figure 1a. They’re just too packed together and there would be no significant rear illumination. However, the arrangement in figure 1b is much more amenable to the use of bifacial cells. The modules are at an angle over a flat, diffuse surface (in this case, grass and dirt). Light illuminating the grass and dirt can be partially absorbed by the modules if they have transparent rears. If the surface that the modules in figure 1b were installed over was a white diffuse surface (say, the roof of a “big box” store), then the results would be even better

Figure 1 of Andrews Blog

Figure 1: a) a rooftop solar array (“Solar panels on a roof” by Pujanak – Own work. Licensed under Public domain via Wikimedia Commons), b) a ground mounted array (“Solar panels in Ogiinuur” by Chinneeb – Own work. Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons)

Most manufacturers of bifacial modules provide some guidance with regards to what arrangements generate the most power. Prism Solar Technologies provide a detailed design guide that gives guidelines for height of installation, angle, and distance between cells based on the albedo (amount of diffuse reflection) of the mounting surface. Various correction factors are included that allow a design to be made. This is useful as it allows one to estimate the power generated from a given installation and compare it with monofacial cells.

However, it is noticeable that they specifically recommend there be at least 1 m of space between rows of cells. Since a Prism modules are 984mm wide according to their specifications, their recommendation is to only cover half the surface at most. This is where one starts to wonder exactly when bifacial technology will be useful. Even in the case of high albedo, rooftop mounted bifacial modules will only be 1.5 times as efficient as monofacial cells, but require twice the area. If module cost was the biggest factor in solar installations, then this might make sense… but it’s not. Both CanSia and the International Renewable Energy Agency report that the modules themselves are generally less than 1/2 of the overall cost for crystalline silicon based systems (although the actual amount appears to vary wildly). If you consider the fact that extra preparation of the installation surface must be performed in order to ensure a large albedo, bifacial systems become even less attractive. They don’t appear to be a good candidate for use in traditional situations.

So when should you consider using bifacial cells? Well, if one already has a flat rooftop surface with a high albedo (such as many big box stores), you may have more rooftop space than you plan on using for solar modules. It would make sense then to just spread them out and generate more power. Bifacial modules also have uses in architecturally-specific situations, like being used as horizontal barriers or as the roof of a car park. The main point here is that they become more attractive when used in less conventional situations.

Bifacial Standards Aren’t so Standard

So let’s you want to start comparing these bifacial cells and modules. The efficiency of new designs are usually first discussed outside of their actual installed context. Instead, standard test conditions (STC) are used. For monofacial cells, these conditions are well established and involve direct illumination of the module/cell with a specific spectrum of light. For earth-based non-concentrator photovoltaics, the AM1.5-g spectrum is used. It roughly corresponds to the average latitude of the U.S.A with the modules tilted at 37° in the direction of the equator. While the metric itself may have flaws, the cells and modules can be directly compared to one-another.

Bifacial cells, on the other hand, do not have well-established STCs. Companies sometimes report on the expected increase in short circuit current (and minor increase open-circuit voltage), or some estimated increase in efficiency when illuminated from the rear by some percentage of the AM1.5-g spectrum in their specifications. However, the argument for bifacial cells generally ends up being supported by system-level data that does not necessarily match any standard. As an alternative, some papers report the efficiency for both the front and rear side independently under STC. For instance, there was a talk given by Fabian Fertig on the newly developed BOSCO cell by Fraunhofer that used this method of characterization. Finally, and most importantly, the ASTM and IEC standards for solar simulator measurements and spectral response measurements contain no mention of bifacial arrangements. I had an informal email discussion with Dr. Keith Emery of NREL about this lack of standards, and he confirmed that there are no US or IEC standards for bifacial technologies and that an agreement between the different independent measurement labs on procedures has not been reached. Basically, right now it’s difficult to compare bifacial cells and modules to each other. Some serious work needs to be done to address these gaps.

A Critique of the MoTech Talk

Let’s return to the conference, where this topic got jammed into my skull. The talk by Yu et al is a perfect demonstration of why the issues outlined above make it difficult for bifacial cells to gain traction. The talk itself was not terrible. The results were not unusual or strange. They reported a 20% improvement in generated power for their bifacial system. It’s similar to other reported results.

The problem is that the method of reporting invited speculation. Absolutely no details were given of the geometric and environmental setup. Details like angle of inclination, cardinal direction, and roofing surface (it looked white-ish) were not provided. When your improvement is heavily based on the installed environment, you need to report in explicit detail what that environment was if the data is to stand up to critique. In addition, they did not use the same technology when comparing the bifacial and monofacial cells. A significantly less efficient module was used for the monofacial condition. While their “20%” claim appears to account for this, it’s still poor practice if your goal is to show that bifacial modules are a good idea. The angle of inclination and cardinal direction was the same for both modules rather than optimized independently.

Here’s the thing we have to ask ourselves… what’s the point of an experiment like this? Assuming that they did a better job reporting their results, could they have achieved this goal? If not, what’s preventing them from doing so?

Let’s assume that this was done in order to really test out their modules and assumptions (and not just for advertising purposes). I can see two main questions they’d like to answer:

  1. Are bifacial modules a better choice than monofacial modules in this environment?
  2. How do our bifacial modules compare to other bifacial modules?

I assert that they could possibly have made a good argument for item one had they used their own technology for the monofacial modules. In addition, they would’ve had to do a rudimentary cost analysis to convince me of the fact that the additional space required by bifacial cells would be a worthwhile decision. They would also probably need to qualify how similar the installed environment is compared to the rest of the world. That kind of work would’ve convinced me that rooftop bifacial modules had a future.

There’s nothing they could really do to address the second point. Unless you’re using an established standard, it gets really hard to convince people that you’re not creating a test biased in favour of your own technology. Developers of this technology are effectively hamstrung by the fact that these standards do not exist. The incredulity bleeds into the audience at every talk.

In the end, it’s questionable as to why every manufacturer of bifacial modules should have to undergo these tests in the actual installed environment. One of the benefits of the current monofacial test standards is that you don’t need to go to a rooftop to check how well your cell or module works. This needs to be extended to bifacial cells so that they can do the same thing.

Suggestions to Fill the Standards Gap

A 2011 paper by C. Duran et al. outlined the importance of the chuck material on bifacial cell characterization. Although the difference in generated current between highly reflective and non-reflective materials is less than I would’ve suspected (under 2%), it should still be a design consideration in solar simulator fixtures. In addition, a dual-illumination system (both front and rear) was developed and demonstrated by that group. The illumination intensity on each surface can be varied, more closely mimicking operating conditions. Although a good step forward, I still think that a key difference between the expected illumination on the rear of the cell and the front of the cell (in a conventional arrangement) is that the rear illumination is almost exclusively diffuse, and this is not taken into account in their setup.

A more recent paper from J.P. Singh et al. outlines a method to characterize the performance of bifacial cells based on the independent measurements of the front and rear illumination conditions at 1 sun. While short-circuit currents were summed, corrections to the open-circuit voltage and fill-factor based on things like increased series resistance were added to increase the accuracy of the predictions. How well their analysis matches to actual photovoltaic cells was not performed, but (assuming such tests yield a positive result) it has potential in terms of analyzing bifacial cells without requiring dual light sources or complex optical arrangements. Certainly, the analysis technique itself is applicable to both diffuse and direct illumination conditions.

From the papers above, it is clear that there is some work going on in terms of establishing these measurements standards, yet it doesn’t appear that there is a whole-hearted industry push.

What Do We Need to Do?

There are two main things that I think we really need to look into. The first is that we need to get better spectral data with regards to actual illumination conditions that bifacial cells are exposed to. White (or concrete) rooftops are particularly low-hanging fruit, although snow and grass are two others with high potential. Alternatively, an integration of the solar spectrum with the diffuse reflection spectra of materials in a database would be a good first start. These become the spectra you want to emulate for rear illumination. However, I’m personally in favour of direct measurement and trying to capture a wide variety of unconventional situations. If the photovoltaic community ever wants to develop wearable solar or indoor solar products, we have to get better at capturing this kind of data so we can use it in design.

From this rear illumination data we can then start to establish appropriate standards for measuring bifacial cells and modules. It is possible that the current monofacial standards can simply be modified, with additional sections pertaining to the testing of bifacial cells. I don’t think it makes sense to establish brand new documents when the ASTM ones on solar simulation are already so extensive.

With that, I finish my “rant”. Bifacial cells have potential when applied in the right situations, but work is still required before they will be fully accepted as an element worth using in an optimized photovoltaic system. Hopefully I haven’t made any mistakes in my assessment, but there’s always the possibility that I’ve missed something. Please feel free to contact me!


Andrew Flood's picture

Andrew Flood
PhD Student, Year 1
APD Group, University of Toronto
Associated with: Project 10

Acknowledgements: I’d like to thank Dr. Keith Emery of NREL for confirming that my claim of a lack of standards was not entirely spurious. He also pointed me in the direction of the PTB publication regarding the angular dependence issue of the current solar spectrum standard.

*If one wants to be pedantic, the AM1.5-g spectrum that is currently standard for terrestrial photovoltaics attempts to take into account the effect of reflections from “dirt”