Perovskites Didn’t Start the Fire
A recent editorial in Matter (https://www.cell.com/matter/fulltext/S2590-2385(21)00560-9?rss=yes#%20), written by its Editor-in-Chief, Steve Cranford, breaks down some of the fuss about perovskite materials and why they dominate the field of materials science these days. It's an excellent introduction to both the materials themselves, the research areas, and it even offers up some commentary on how perovskites have changed the culture of materials science.
Where it falls a bit short in my mind is in the history. Ask anyone in the perovskite field their biggest gripes and they'll probably tell you about how, as Steve points out, new papers are published where the sole innovation is, "we got a higher PCE than our control by doing X". And it's true, you can write perovskite papers like clickbait titles and never run out of content. "Caffeine wakes up perovskites". "A pinch of salt to perk your PCE". I had a meeting with my then supervisor in 2019 telling him I could add sodium or potassium chloride to a tin oxide layer, improve the PCE, characterize the device and have a paper out in about a month. He declined and informed me that was not the type of science he wanted to do and fair enough. But sure enough the paper I just described was published by a different group a few months later in Advanced Energy Materials. This was hardly a surprise to me. The group in question was led by a researcher who had published a similar paper in Science two years prior using a titanium dioxide layer, during his postdoc. That paper had given me the idea, and evidently he hadn't forgotten it either. Now, both of those studies were very thorough, the type that Steve is specifically lifting up. They didn't just add salt, run some JV curves, collect some XRD and call it a day. They went to great effort to examine why the ions made a difference and they are what I would largely label "good science".
The thing that strikes me as misleading in this editorial though is the idea that perovskites as a field are responsible for this PCE-focused style of reporting. While I think it's undeniable that they cemented the approach, perovskite researchers inherited the metric and the style from other solar materials research. Organic photovoltaics (OPVs) had been doing this long before the perovskite solar cell (PSC) appeared in ~2012 (As well as dye-sensitized cells, kesterites, CdTe, and many more).
Not mentioning that legacy is, in my opinion, a huge oversight. Partially because perovskites inherited the incremental improvement and reporting from other solar fields. More importantly, because the differences between these materials is the reason why they have changed the face of materials research. Imagine it's 2012. Konarka has just filed for bankruptcy and the funding outlook for OPVs, the once promising, game-changing solar material, is starting to look bleak. Their PCE has been stuck at just over 10% for years and it will be another 2 years until non-fullerene acceptors kick the race off all over. When they do, OPVs will be living in a post-perovskite world and it will be very different.
Organic photovoltaics are curious. The focus in that field is often on designing new molecules to serve as the photoactive material. That process is damn difficult and incredibly time consuming. If you want to design a new molecule from scratch you might very well spend months doing so, only to end up with a small amount of the photoactive compound. Enough for a small study of devices. Add in a few variations with different atoms or alkyl chains and you have a full study that is probably a year's work easily. The whole process is often rate limited by the synthesis, as are the experiments you can reasonably hope to run. Oh, and the reagents probably aren't cheap.
Then PSCs come along. They use salts that you can buy in huge quantities. You have 50 mg of your precious OPV material that took 6 months to make? How about 500 g of a common chemical that you can order on Tuesday and get on Wednesday for 50 bucks? Perovskites completely obliterated the publishing model created by OPVs. Suddenly the researchers making devices didn't have to contend with material scarcity, and they did the logical thing to do…. Do more. A lot more.
So what's the problem? Well, OPVs (and other solar materials) had already set the publishing model. Make a few changes, study the result in great detail, hopefully report a better efficiency. Single atom substitutions, changing the processing solvent, adding an alkyl chain, trying a new copolymer, swapping out a transport layer in the device. All of these changes were rate-limited by synthesis and materials availability and perovskites threw everything out the window, except the part where the PCE is the only thing that mattered. When I switched fields from OPVs to perovskites in 2017 I did a survey of some of the bigger names in the field. Unsurprisingly, many researchers whose work I had been following in OPVs had become major pillars in the perovskite community. At a simple level much of the infrastructure for making and testing both devices is the same. You can pretty much buy the perovskite compounds and get started if you're already making OPVs. For years it felt like I had deja-vu. New perovskite studies would be published where the researchers basically took an OPV manuscript, and ran a find and replace to swap "OPV" for "PSC". Entertainingly, I had several colleagues who hadn't worked with OPVs at all who thought these were ground breaking, novel, and insightful. Some were quite surprised to find out that the clever use of a certain transport layer was actually something that had been perfected for the past decade in a different type of solar cell.
But plenty of researchers did remember their history of working with OPVs. And how did the field respond? By publishing more perovskite work than anyone could imagine. And in doing so they moved the goal posts for other solar materials. When OPVs had their non-fullerene acceptor (NFA) renaissance in 2014/2015 the field started grinding the same way. The smallest changes possible were made to new NFAs. A chlorine added to one position. Months later a fluorine added to the same position. Published separately with record PCEs. Always chasing the now distant champion perovskite cells. They still haven’t caught up.
The thing is that these cookie cutter studies do add value, but realistically they should exist as table entries in a database, not as full articles and certainly not as the pinnacle of research in the field. It's no surprise that machine learning, automation, and high-throughput screening are championed by some members of the PSC community. They make sense as research tools for the fast-paced, multi-parameter type of work that perovskites enable.
I agree with Steve Cranford that some of this work is more engineering than science. At the same time, I think that increasingly the engineering is challenging our community to re-evaluate how we define "science" in light of the new tools at our disposal. And, to re-iterate the main point, perovskites didn't start the fire, they just stepped on the gas.
