Very good! There are more efficient multi-junction solar cells in commercial use, but the existing ones use stacked III-V semiconductors and are very expensive, really only suitable for space applications. This record is for a thin perovskite material solar cell on top of a conventional silicon solar cell. Both materials are inexpensive.
The main obstacle to commercialization is keeping the perovskite material stable over long periods of operation. This family of materials is more sensitive to water/oxygen/light than silicon itself, but they need to last nearly as long as silicon for cells used in solar farms and rooftop panels.
Long term degradation and manufacturer warranty for such is indeed a factor. There's a number of poly and monocrystalline Si 156mm cell based PV modules now which are warranted for something like 83% of their original STC rated output after 25 years. Of course you have to also believe that the manufacturer will still be around in the same corporate form in 25 years and able to honor its warranty.
This certainly may not be the case if you follow PV industry news and are aware of how many PV panel manufacturers have formed as startups and gone bankrupt in the last 8 to 14 years.
edit: the other reply here in the thread asks:
> but can be replaced as a layer every, say, 5 years or so? It's not trivial maintenance, but you have to do some maintenance on the panels anyway
Typical mass market PV panels are cells permanently encapsulated behind glass sandwiched with a back sheet, it is not practical or possible to replace a layer or modify the cells once they're assembled into a panel. The only maintenance done on a series of ground or roof mounted typical PV panels is to wash them.
> Of course you have to also believe that the manufacturer will still be around in the same corporate form in 25 years and able to honor its warranty.
I just bought 26 AEG panels for exactly that reason. They were a bit more expensive but my estimate is that AEG will outlive me so that should be fine. Contrast with many other solar panel producers who seem to go out of business ever five years or so.
There are hobbyists making solar-powered remote-control planes that are just barely able to charge as much as they draw at noon on a sunny summer day.[1]
Even a minor boost to efficiency could push this concept from "doable in ideal scenarios as a tech demo" to "useful in the real world as a way to multiply the range of electric fixed-wing aircraft"
Let's say that truck roof covers about 41.5m² (the area of a standard 53' x 8.5' trailer roof). Let's further assume that 41m² of that is covered with PV panels. Using current technology (~21% efficiency silicon cells) and assuming standard irradiation of 1000W/m² this roof would produce ~200W/m² for a total of ~8.2kW. Add a 40' trailer (~31.5m² roof area, ~6.2kW) for a total of ~14.4kW. Using 31.5% efficient cells would raise this to ~22.7kW. Taking the newly releases Tesla semi as an example (~1 kWh/km, ~900 kWh battery pack) this would add about 14 km of range per hour using current cell technology or about 22 km per hour using these new cells. Assuming an average road speed of 80 km/h this would extend the range of the vehicle by up to ~25% for the new cells, ~18% for current cells during daylight hours with standard irradiation and no shading.
>Let's say that truck roof covers about 41.5m² (the area of a standard 53' x 8.5' trailer roof). Let's further assume that 41m² of that is covered with PV panels. Using current technology (~21% efficiency silicon cells) and assuming standard irradiation of 1000W/m² this roof would produce ~200W/m² for a total of ~8.2kW. Add a 40' trailer (~31.5m² roof area, ~6.2kW) for a total of ~14.4kW.
So you're now pulling two trailers but using the watts/km for one. Also your panels will be fixed to the roof and not tracking the sun. you aren't going to get near 30% for them.
This depends on the load and on the veracity of the data given by Tesla. A load of styrofoam - which is used in construction, just look what goes under the on- and off-ramps of bridges - doesn't weigh all that much after all. Also, this was not an attempt at with 100% accuracy calculating the range effects of PV panels on the roof - notice the liberal use of the '~' sign and the fact that nearly everything is approximated and rounded down. Did you not notice?
What about if the Perovskite degardes, but can be replaced as a layer every, say, 5 years or so? It's not trivial maintenance, but you have to do some maintenance on the panels anyway, so maybe it's practicable. And the price of replacing a detachable layer should be a small fraction the price of the original panels and installation.
Even if the degradation could be repaired (e.g. after five years, restore a 28% panel to 31% efficiency), at grid scale, does it make financial sense to do the maintenance instead of replacing the panel when it falls below X%?
It would need to go to a clean room. And accessing the top lager is at odds with bonding the top glass well enough go protect it.
Methinks a better strategy would be trying to make them keep >17% efficiency even if the perovskite fails and just treating it as a temporary bonus for the early movers. Still sounds hard to manage the voltages as it fails though and would require a different strategy for the busbars even if it were viable.
It may not need to be wasted. There’s currently a price floor on used EV’s with degraded batteries because they can be used for home energy storage after their useful road life.
There could be a solid secondary market 20 years from now for panels that have dropped to say 20% efficiency where the surface area to yield ratio isn’t a factor (rural areas I would guess).
Extreme efficiencies in STC W per square cm (or meter) are primarily of interest to things where room to mount PV cells is extremely constrained. Such as on satellites. Look at triple junction GaAs based cells used in satellite applications for example.
There are a large number of research-lab-only PV cells made in the last 10-12 years which greatly exceed 23% but are economically unfeasible or impossible to purchase for ordinary use. Some of this tech does trickle down eventually, however.
Of more practical real world interest is $ per STC watt for a panel you can buy in a 20-panel pallet load from an ordinary PV wholesaler. Like a figure of $0.28 USD/W for nominally 380W rated 72-cell monocrystalline Si panels for rooftop or ground mount applications. Meaning that a pallet of 20 panels would be somewhere around $2100 to $2200 USD to purchase plus freight.
In approximately the last 12 years we've seen things go from if you buy a pallet of "cheap" mass market 72-cell panels, you'd get 320W rated per panel (STC rating of about 4.44W per cell), to now being able to buy something that is 380W rated as mentioned above, approximately 5.27W per cell. All under STC measurement conditions which are only a rough approximation of real world sunlight of course. The same panels typically measure 1.99 x 0.99 meters so you can do the math on the improvement in STC W per square meter if mounting space is a limiting factor.
The cost of the panels is already such a small fraction of residential install cost. My panels were definitely under a buck a watt (maybe 15% of the total cost). The installation cost (i.e. paying for a team of a dozen guys to dangle in harnesses on my steep roof for three days) absolutely dominated the invoice (50-60%). The second most expensive was the batteries (about 20%). The remainder went into the hundreds of other parts (inverter, cut-off, conduit, circuit breakers, cables, MPPTs, brackets, safety stickers…) needed for a functioning & legal system. And a couple hundred bucks of fees to the city for permitting and inspection, and a couple hundred bucks for the off-shore engineers to draw the system design docs.
I won't make a profit for decades, unless the price of grid power shoots up. It would make more sense in a place with higher power costs. But I can keep the lights on if there's a power outage, without the maintenance costs (and noise) of a hydrocarbon-fueled generator.
If you're installing solar for monetary gain, don't put it on your roof, buy/lease cheap land and build a solar farm. My electricity provider even lets you buy in on syndicated solar farm deals, if you don't want to manage the process yourself; you get the generated kWh credited back on your power bill!
There was a story last week about commercial solar farms having panel flat on the ground and some people posted about putting home solar setups flat (or slightly elevated with a brick and tied down).
Saving money with the setup beats getting the last percent out of the install. Although probably only some rural home-owners are going to be able/allowed to just put the panels flat on the ground.
Haha, love it. Sure, you're losing a few percentage points of efficiency, but not THAT much, as long as you're within a couple dozen degrees latitude of the equator (COS(X) ≈ X for small values of X). And if you're far from the equator, why are you messing with solar anyway??
For a residential install (10s of square meters, not thousands). I would still feel more comfortable laying down some concrete first to minimize the chance of spoilage from vegetation. Plus something cheap for elevation from any flooding (bricks, cinder blocks, etc).
Yep, as soon as you start talking about flat panels on the ground, you start introducing problems with moisture, and/or snow/ice buildup in the winter.
Might be ok in some areas, but not at northern latitudes, and definitely not their winters.
I'd gotten a couple quotes for grid-tied solar installations recently.. and they were pretty outrageous @ over $5/watt installed (with no battery backup & no option to stay on with the grid out).
So now I'm in the process of putting together a 1.8kW solar system for the roof of a cargo trailer. Not the biggest solar setup, but I'm also not a huge consumer of power - this should offset a significant portion of my consumption in summer, and it's also expandable to ~2.7kWh (though I should've bought a 6th panel for that).
I've paired the panels up with a 3kW hybrid inverter and a 5.1kWh LFE rackmount battery. Should do pretty well together, and like you said... I'll be able to keep the lights (+fridge/furnace) on if there's a power outage - a huge plus.
Total cost for the main components and most of the necessary wire/hardware? ~$4000, before any tax credit.
Not just that, the other trend is that panels used for solar farms have massively increased in size & capacity. The panels we are now buying for our solar farms are 660W, and over 2 meters long. While not as feasible for roof mount (although we're speccing 575W panels for commercial rooftops), these large panels reduce our associated balance of system costs on a $/W basis.
Keep an eye on the panel construction though, the gold standard is glass-glass but there are plenty of other materials used for the sandwich (and the sealing!) and not all of them will stand the test of time.
This is also relevant to large-scale installations. For a single family home, perhaps you don't care if you use 30% of the roof area or 32.5%; but if you're building a solar farm for a whole city, then those extra 2.5% are a good number of Km^2 which you can save; or have as extra safety-margin production capacity.
Very interestingly, this seems similar to the strategy used by plants to absorb energy from sunlight in the red and blue parts of the spectrum with pervoskite absorbing the blue and silicon the green. However there is no indication this is intentional bio mimicry. Do we not have materials which can absorb the middle green part of the spectrum?
It's not really very interesting. Those parts of the spectrum are where the energy is concentrated. There's quite a bit in infrared, but it's spread out over a wide range of wavelengths, so difficult to use (except as heat).
Assuming this[0] is accurate, it looks like solar irradiance reaching the surface is pretty flat from mid-blue up to red, including green. I would guess either there are other evolutionary reasons for reflecting green, or no reason at all and chlorophyll happened to be the first/easiest substance to evolve for photosynthesis.
The most successful aquatic phototrophs, e.g. brown algae and diatoms, absorb a much larger fraction of the solar light, by having complementary pigments that transfer the energy to chlorophyll.
On the other hand, the terrestrial space is dominated by the green plants because here the problem is not absorbing more of the abundant light, but avoiding overheating from absorbing too much light (because the efficiency of photosynthesis is low, so most of the absorbed energy becomes heat, and not only there is much more light in air, but also the cooling is more difficult than in water).
These perovskites are notoriously difficult to deposit at scale in a uniform enough layer to be effective, with small variances causing dramatic drops in efficiency.
On the cool side, it's theoretically possible to make them translucent (without the silicon substrate ofc) which could make for cool power generating windows in the future.
On the not so cool side you really don't want the materials anywhere near you or your water table in the event of a panel being damaged. Lead halide perovskites, methylammonium lead iodide, are insanely toxic and a race to the bottom on price if they become widespread could be an environmental disaster waiting to happen.
Not to take from the achievements described here, but there isn't any mention of it. There is some hope in taking the lead out (tin based perovskites) but that tends to result in a drop in efficiency.
Give me the TSMC or AMD of solar panel manufacturing. If they can do single digit improvements in panels every year or so, they could enable new use cases each generation. I bet those (3nm that computers now have) fabs tools have useful technology for manufacturing solar panels for example.
Also you can’t lay massive solar farms withou native manufacturing. What you’ll just buy them from China for hundreds of billions of dollars coughEuropecough but also the US??
Removes the cost of the frame, reduces efficiency. Presumably whether it's a net gain depends upon the site, eg terrain shape, latitude, need for passive cooling, wind-blown dirt, local fauna, etc.
None; they are all hitting up against the same limits with the current technology someone would need to figure out something completely new to trigger another solar boom.
Quite seriously. I think out of nearly all of the energy options out there (green or not), it's certainly the least "offensive", plus, it's cheap, efficient, and definitely green (other wastage aside), to boot.
Now if we can only figure out an equally amazing way to get potable water and sewage taken care of without storage or digging then you'd have a truly magic shed.
I believe around 22% is standard for PERC which is dominant. TOPCON and HJT are a smidge over 25% but are a bit more expensive for now and use more silicon and silver.
Single junction caps out around 28% at the module level so surpassing this at scale requires commercialising a new technology like these perovskite tandem cells.
The other benefit is it reduces consumption of the main limiting critical mineral of silver down to insignificant levels by increasing the voltage on the metal layer.
So, obviously, there's a difference between efficiency in lab settings and efficiency in the field for mass-produced cells. But - this is an impressive achievement for the Silicon + Perovskite technology.
But I have a question to the more knowledgeable here: The chart in the story shows other technologies which achieve significantly higher efficiency figures:
> As of 2014 multi-junction cells were expensive to produce, using techniques similar to semiconductor device fabrication, usually metalorganic vapour phase epitaxy but on "chip" sizes on the order of centimeters.
I'm curious how this process works. Is it that the researchers are experimenting with different ingredients and seeing what works best? Or do they have a clear idea what sort of structure they want to build and the research complexity is in how to get the various ingredients to assemble into the required structure? Basically wondering how the researchers go about planning a research program and how clear the goals and timelines are.
I'm not working in solar cells but do research in optics and have a general understanding of this (also from discussions with people who understand this better).
The main limiter for solar cells from silicon is the overlap of the bandgap with the spectrum of the sunlight and also the loss in efficiency when cells get hot. So what people do is they use multiple materials (often in different layers) that cover different parts of the spectrum to absorp most of the sunlight. Researchers tend to have a good idea which these are (although there is research in creating new organic ones), but the challenge is to combine this in fabrication with the silicon and in a cheap easy to fabricate way and with materials that don't degrade over time.
So short answer researchers have a general idea what needs to be done, but as usual the devil is in the details.
Again not an expert in the exact area, so anyone who is please correct if something isn't right.
Also not an expert, but from my reading it's all engineering. The physics is sufficiently well understood and variations can be simulated.
It's about preventing reflections at various surface boundaries (or more generally, controlling reflections everywhere) and structuring things so that different wavelengths go to the right layers, and preventing recombination, and a zillion other details. Material quality and process control to get the right surface properties are very important.
The cells that make it out of the lab are those where the cells are robust and processes can be adapted and scaled to mamufacturing quantities, competitively with current products.
> Various teams from HZB had achieved a record value in late 2021 with an efficiency of 29.8% that was realized by periodic nanotextures. More recently, in summer 2022, the Ecole Polytechnique Fédérale de Lausanne, Switzerland, first reported a certified tandem cell above the 30% barrier at 31.3%, which is a remarkable efficiency jump over the 2021 value.
> More recently, in summer 2022, the Ecole Polytechnique Fédérale de Lausanne, Switzerland, first reported a certified tandem cell above the 30% barrier at 31.3%, which is a remarkable efficiency jump over the 2021 value
Probably build a small solar-powered light, or see how much usage I could get out of a solar-powered microcontroller, or even just see how small of a thing I could make that still collects power.
The technological advances we (society) make are endlessly fascinating to me. I just want to have more of them in my hands.
Solar is pretty ubiquitous already in a lot of places… Around here (Australia), more than 1/3rd of homes have solar and plenty of commercial premises.
The efficiency doesn’t matter much, it’s just the cost. Most people don’t pack as much solar on their roof as they can possibly fit on anyway, so higher efficiency would just mean less of the roof is taken up.
The problem right now isn't so much panel efficiency as the lump cost of doing a project like this and the affordability / justifiability of that.. Especially when solar companies take advantage of people receiving tax credits & gouge on labor pricing.
in other forms of progress the goal post is pretty clear, I believe it is clear here but need it explained to me, let me know if you can help. what is the metric, how close are we to it.
I mean, I don't think there is one. There's no magic number that's the goal (or well, I suppose in the broadest possible view there's the theoretical maximum efficiency, but that's a long, long way off). In general, better efficiency is better than worse efficiency.
> in other forms of progress the goal post is pretty clear
I'm not sure that's true. Incremental progress is fairly common. A 7nm feature size microchip is generally better than a 10nm one (from the same manufacturer, at least; annoyingly, people are increasingly fudging the numbers there), which in turn was better than a 13nm one and so on, as an example.
There are a bunch of physical reasons why you can't have a ~100% efficient solar cell – obviously thermodynamics chief among them, but also effects like the fact that the bandgap is finite (yet the sun's black-body spectrum has energy radiated below it) and the fact that solar cells themselves will heat up and re-radiate photons they absorb.
For a long time, solar cell efficiency was around ~10-20%. There's actually a theoretical upper limit, the Shockley–Queisser limit, of a bit less than 34% for a single p-n junction photocell. [1] This is a tandem cell – for which higher efficiencies are possible but costs go up. The thermodynamic limit is reached if you have an infinite number of layers.
Shitty efficiencies[1] haven't stopped us from using combustion engines in cars for a century. But while combustion engines use expensive fuel, these solar cells use sunlight, which is free. So which technology should we celebrate?
[1]: 20% is already quite a good efficiency for a car in mixed traffic. That doesn't even account for all the energy used to provide the fuel.
So a panel that looks like Vanta Black, but also at non-optical wavelengths? Better efficiency is being pursued but the engineering behind real products will always be the result of balancing design constraints:
Has the potential for an upside previously unavailable at low cost. By stacking layers of different solar cells you can surpass the theoretical limit of a single layer which is 32% (or around 28% in practical terms including spaces that aren't solar cell).
Single layer silicon is in the 22-25% range now with the 22% modules being the optimal cost in most areas.
The modules also only make up a small fraction of the cost of a PV install (around 25-30%), so increasing efficiency from 22% to 32% reduces BOS costs by almost as much as the entire module currently costs
Perovskites are also dirt cheap, so are a good candidate for increasing efficiencies. The difficulty is in making them last. Accelerated aging tests are promising but they haven't hit market yet.
BOS: balance of system, i. e., everything else besides the panels (mountings, wiring and connectors, labor and land costs being the main parts of BOS affected by panel efficiency.)
There's a vast number of esoteric PV cell types you can't actually buy at any reasonable price, or buy at all at any price, unless you're in an extremely specialized market niche.
Note the PDF link above to best research cell efficiencies, which means somebody made a very small number of PV cells in a lab environment. Not something that's produced for sale to end-users in any real quantity.
Because this one is very practical and has a really good chance to make it to market. One can build very esoteric structures in a lab environment - which is of course also an achievement - but they are not easily mass produced (too expensive, too unstable, too difficult to manufacture at scale, too unpredictable, etc.).
The main obstacle to commercialization is keeping the perovskite material stable over long periods of operation. This family of materials is more sensitive to water/oxygen/light than silicon itself, but they need to last nearly as long as silicon for cells used in solar farms and rooftop panels.