The size of the challenge for OPV

In a recent paper titled Green chemistry for organic solar cells, published (open access) in Energy and Environmental Science, the authors Burke and Lipomi ask an interesting question: how much organic semiconductor do we need in order to make a sizeable dent in global energy consumption with organic photovoltaics (OPV)?

[I]f 10 TW of the 30 TW of power demanded in the year 2050 is to be generated by photovoltaics, and if organics account for 500 GW to 5 TW, then 10–100 kilotonnes of organic semiconductors will be required… given an average solar flux of 200 W m-2, a module efficiency of 5 %, a typical thickness of the active layer of 200 nm, and a density 1000 kg m-3. This extremely rough estimate assumes 100 % yield of working modules, no waste in the coating processes, and infinite lifetime of devices.

To put 10,000–100,000 tonnes in perspective, polyethylene is the most common plastic with an annual production of 80 million tonnes (according to Wikipedia, but it sounds reasonable). Taking the higher end of their rough estimate, we’re talking 0.1 million tonnes over nearly 40 years—basically nothing compared to commodity polymers. But organic semiconductors have more complicated structures than commodity polymers and require more complicated chemistry. Burke and Lipomi also compare their estimate with pharmaceuticals, with their figure “2–3 orders of magnitude greater than those of top-selling small-molecule drugs of similar structural complexity”.

Making this amount of high quality polymer—with a specific molecular weight and acceptable PDI, high purity and low batch-to-batch variation—is a challenge. Material currently on the market is terrible, meeting none of these requirements.

Both conjugated polymers and structurally complex drugs require synthetic sequences of 5–10 steps to produce. The multi-tonne synthesis of conjugated polymers will be a challenge in process chemistry with few precedents, and will in consequence the materials that could be seriously considered for installations that cover many square kilometers.

The challenge is made even greater by the need for it to be cheap because if OPV isn’t cheap, it’s commercially inviable. If OPV is to stand a chance researchers must remember that they’re working on a technology that has to cost <$10 m-2 to compete with fossil fuels.[^price] Burke and Lipomi point out is roughly the same price as carpet.

No matter how efficient your polymer, if its synthesis can’t be scaled up or is too expensive (e.g. PCDTBT), it will fail. It worries me a little that this is lost in the quest for high device efficiency (because that’s what makes for a HIGH IMPACT[^IMPACT] paper). Taking a “not my problem” approach is unacceptable. We must stop passing the responsibility off to process chemists and engineers and instead remember the scale of the problem that OPV is trying to solve, otherwise it’ll never make the slightest dent in our global energy consumption.

*[PCDTBT]: poly[N-9′-heptadecanyl-2,7-carbazole-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′-benzothiadiazole)]

[^IMPACT]: Just to be clear, I’m being flippant. [^price]: Burke and Lipomi get this price from Lewis and Nocera.

Making semiconducting polymers in flow

Organic electronics has a problem with batch-to-batch variability in the quality of materials, particularly the active semiconducting layer. A fellow PhD student in my office described to me the trouble he often experiences. He made one batch of solar cells last week and measured an average efficiency of 5 %. This week, another batch had only 0.5 % efficiency, depsite having the same device structure and processing conditions. Further investigation revealed that the lab had recently switched to a new batch of polymer, even though they’re supposedly identical.

(This raises the question of how many positive literature reports describing organic electronic devices are “correct” and not anomalous? But here I want to focus on this problem in the context of large scale device production.)

Batch-to-batch variation is bad enough with the sub-gram quantities used for small area device fabrication. Printing optimised large area devices—measured in square metres/minute rather than square centimetres/day—would be next to impossible. You need to know that this week’s batch of polymer is the same as last week’s because different molecular weight size distributions, chemical defects and impurity levels have a big effect on the required processing conditions and final device performance.

It’s quite difficult to produce organic semiconductors on the large scale required for device printing because the polymerisation reactions scale up poorly from round bottom flasks to large batch reactors. You have to re-optimise at each scale, which costs time and money.

James Bannock, who is a PhD student in the same group as me at Imperial, has been working on the use of droplet flow reactors to make poly(3-hexylthiophene) and get around the problem of batch scale up. He’s recently published his work in Advanced Functional Materials (DOI: 10.1002/adfm.201203014) and it’s open access so you can read it yourself for free.

I encourage you to take advantage of the paper being open access and have a look. There’s lots of photographs and figures. Briefly, activated monomer solution is injected into a narrow plastic tube containing a flowing stream of immiscible carrier fluid dispersed with the catalyst. Microlitre-sized droplets form, like tiny individual reaction flasks, and travel through the tubing. Because the droplets are small the chemical environment is highly homogeneous and each droplet is essentially identical. The tubing is heated in an oil bath and the polymerisation reaction (a Grignard metathesis) happens as the droplets travel through the tubing.

You can produce more material by using longer tubing and a higher flow rate. Crucially, this can be done independently of droplet size. So unlike batch reactors, nothing happens to the chemical environment and you get exactly the same polymer, with a low polydispersity index and high regioregularity. The plan is to apply this method to other organic semiconducting polymers. It provides a way to produce device-grade polymer on large scale required for large area fabrication, with minimal batch-to-batch variation, and will hopefully drive the industrial production of organic electronic devices.

Reference: J. H. Bannock, S. H. Krishnadasan, A. M. Nightingale, C. P. Yau, K. Khaw, D. Burkitt, J. J. M. Halls, M. Heeney & J. C. de Mello, Continuous Synthesis of Device-Grade Semiconducting Polymers in Droplet-Based Microreactors, Advanced Functional Materials, 2012. DOI:

claimtoken-50ca5f32167d2

Put down that bottle of chlorobenzene

Recently I’ve been searching the literature for some good references on the roll-to-roll printing of organic electronics for the introduction of my MRes report, which is due at the start of September.

The performance of organic semiconductors is much lower than that of the conventional inorganic semiconductors. One reason why researchers are interested in organic semiconductors is that they can be produced using printing presses (literally the same technology to print things like magazines, newspapers and T-shirts) so cheaply compared to inorganic semiconductors that it no longer matters that their performance isn’t as good. The biggest application is probably solar cells and it’s really important that costs are kept as low as possible for them to be economocially viable.

Despite being part of the Centre for Plastic Electronics at Imperial, I don’t know of anyone who completely prints solar cells, let alone using roll-to-roll processes. There’s a lot of reasons why (e.g. printing presses take up a lot of space, use a lot of material) but it does my head in that people are use techniques like spin coating and vacuum deposition, anneal at high temperatures in inert atmospheres for long times and use environmentally-unfriendly cholorinated solvents. Papers describing devices made using these techniques frequently laud the scalable and low cost nature of plastic electronics. But are these techniques scalable to large area, high throughput, continuous printing processes? No, not without spending a lot of money, but then it’s not economically viable.

Today I came across a good review[^ref] in Materials Today (open access!) on the roll-to-roll fabrication of polymer solar cells. The first paragraph sums this issue up nicely (emphasis added):

In order to reach its full potential, the imminent realization of the 10 %-10 yr target[^target] within the laboratory must transcend into a realistic industrial process. While this may seem trivial to many and even obvious to some, there are challenges that have perhaps been taken too lightly in laboratory reports. Often tiny spin coated devices prepared on rigid glass through toxic solvent processing and metal evaporation is said to be roll-to-roll and industry compatible. The view held here is that claiming to be roll-to-roll and industrially compatible without such instruments is similar to claiming that one can learn how to swim on a floor.

I love that last sentence so much I’m tempted to quote it at the start of my report.

To summarise: I think researchers need to stop simply writing in the introductions of papers about scalabilty and low fabrication costs and actually start considering it in the lab or, even better, dropping these unscalable techniques and compounds from the lab altogether.

[^ref]: R. Søndergaard, M. Hösel, D. Angmo, T. T. Larsen-Olsen, F. C. Krebs, Materials Today, 15 (1-2), 36-49. DOI: 10.1016/S1369-7021(12)70019-6

[^target]: 10% device efficiency from devices that last >10 years.