15 ms·

Engineer's solar panels are breaking efficiency records

230 points by belter 2y ago

ziggy_star 2y ago
The super efficient ones are the panels they send to space. This is just like with chips, all about yields and price/performance. I would not hold your breath for the typical consumer panel to improve much beyond 20%-25% any time soon sadly. It does generate a lot of hopeful breathless articles which rubs me the wrong way. It is important to stay realistic in the search for solutions. Solar is already great and cheap and there are lot more wins possible in the actual deployment as most of the cost is now overheads, bureaucracy, labour, 'etc.
- GlibMonkeyDeath 2y ago
  It's not a breathless article - these advances are quite doable from an engineering standpoint (and going from 20-->25% is a HUGE deal - that directly reduces the number of panels needed, and so directly reduces overhead and labor.) The more exotic designs which often are the subject of breathless articles (e.g. perovskite https://en.wikipedia.org/wiki/Perovskite_solar_cell https://en.wikipedia.org/wiki/Perovskite_solar_cell or other multi-junction cells) can get greater efficiencies (like up to 40% for quad junction) but are a lot more expensive from a lifetime cost perspective (they don't last nearly as long as silicon junctions and are much more expensive to produce.)
  - ziggy_star 2y ago
    I will of course happily buy a 25% panel as soon as it is on the market if the price per watt for the total install is not soaked up by a multitude of other factors. When can I expect that to be?
  - IshKebab 2y ago
    I don't know if it's that huge of a deal. The main challenges of solar are not the number of panels now. The incremental cost of installing an extra panel is quite low. The biggest challenges are storage, and - in the UK at least - nimbyism. (Yes people really object to solar panels in fields.) The other thing I would say is that the software for solar inverters is way behind where it could be. You could easily get a 10% improvement just by making them smarter - e.g. using time series prediction, day ahead pricing, etc. Unfortunately they're stuck in the stone ages. I have a QCells inverter (rebranded Solax) - do not buy btw - and they directly told me they are not interested in any of this smart stuff. They also do automatic firmware updates with no opt-out and no notification of changes. And the updates include removing features. Do not buy QCells solar! (Hopefully Google finds this.)
    - jpgvm 2y ago
      IMO with solar it makes the most sense to stick to high volume Chinese stuff. Longi/Jinko/Trina/CanadianSolar/whatever for panels and Deye for inverters. The Deye stuff is particularly good because of aforementioned smart features, especially if you are off-grid or operate on crappy rural power. My favourite feature is called peak-shaving where it uses the battery to supplement said crappy rural power or a generator that is provisioned for average load instead of peak. I haven't found inverters with similar features and configurability anywhere else thus far.
    - justahuman74 2y ago
      Lots of people have limited roof space that isn't in shade for parts of the day. If we're able to cheaply over-provision the panels on the roof then that mean less reliance on the grid (if you use batteries)
- dogma1138 2y ago
  That’s not very correct the new ISS solar panels for example have a 12% efficiency they are optimized for longevity and weight. The relatively high efficiency figures often touted for some spacecraft panels especially in low earth orbit don’t compare apples to apples as they include the additional 70-80% solar radiation that isn’t absorbed, reflected or scattered by the atmosphere. There are some spacecraft that do use multi-junction cells with very high efficiency however those are ones which are sent far into the outer solar system like Dawn and Juno.
szvsw 2y ago
> “Gaining 1 or 2 percent more efficiency is huge. These may sound like very tiny increases, but at scale these small improvements create a lot of value in terms of economics, sustainability, and value to society.” It’s so easy to forget this and the massive scale and its relevance at the massive scale of the systems we need to (and are, to some extent) roll out. It also seems promising when these breakthroughs are happening in R&D groups of industry players trying to dogfood it rather than in labs. At the same time though, it’s starting to feel to me, to some extent, like we have kind of solved solar? It’s everything else around it that needs to advance, particularly grid infra, batteries and electrifying the general class of difficult-to-electrify problems (steel, concrete, freight). I might be totally off-base and blinkered with that assessment. Edit: I guess I should try to clarify my feeling after reading some of the responses below: it feels like solar tech is not really the limiting factor in renewable scaling, and that advances in solar efficiencies won’t drastically/meaningfully simplify the other challenges/limiting factors we currently face (grid infra/batteries, electrification of mfg, duck curve, etc). Children point out that space and cost savings from efficiency gains in solar may still be significant at grid scale though! Still, this is very cool progress to read about!
- bluefirebrand 2y ago
  Completely agree. I'm sure there's still a lot of advances left in solar panels themselves, but we really do need to solve the grid, solve storage, essentially solve the base load We won't be able to transition fully off of fossil fuels until we do
- Kye 2y ago
  I wonder if someone has done a science to compare the carbon impact of batteries and peaker plants. It might be that a lesser of evils fossil fuel is the solution if it's a sometimes thing and not the main thing. It also has the virtue of being supplyable in a renewable way with landfill gas recovery.
  - Tade0 2y ago
    A battery made using the dirtiest grid that produces them at scale (110kg CO2/kWh of battery) breaks even compared to gas after ~200 cycles. LFP batteries commonly used as stationary storage regularly do 2000 cycles. Makes more sense to use that gas to make more batteries.
MBCook 2y ago
Is there a known theoretical maximum efficiency for solar cells that we can’t get past with our current approach? If they’re hitting 25% are we close to that limit?
- Retric 2y ago
  Depends on what you mean by current approach as many different methods are in use. Single junction solar cells are limited to 33.16%. https://en.wikipedia.org/wiki/Shockley–Queisser_limit https://en.wikipedia.org/wiki/Shockley–Queisser_limit Increase the number of junctions and that goes up. Ultimately with infinite layers you can’t beat ~68.7% on earth and 86.8% when much closer to the sun.
  - GeorgeTirebiter 2y ago
    These efficiency limits are true for P-N junction solar cells. There are other solar cell types which can ultimately achieve higher efficiencies, but they are currently lab curiosities with low efficiencies.
    - MBCook 2y ago
      That’s the kind of thing I was trying to get at with my “current approach” comment. Maybe solar cells based on chemical reactions or [insert science things here] can do far better, but right now semiconductors is what everyone sells.
  - crazygringo 2y ago
    What does closeness to the sun have to do with it? If it's a question of the intensity of the sunlight, can't you just focus it with a lens?
    - missedthecue 2y ago
      Concentrator Photovoltaics (CPVs) exist but are expensive not just because of the lenses but because you need a tracking system to follow the movement of the sun and adjust the lens constantly to keep the light focused.
phkahler 2y ago
TLDR they're at 25.5 percent. Didn't read but it seemed like it's probably a practical incremental improvement which is good.
Sparkyte 2y ago
I can tell you I am excited... but when can we realistically afford solar panels with high efficiency? If a person came by my house and said, "Yo, I can do an installation!". Those panels are like running on a 10 year old or greater design and process.
- ehsankia 2y ago
  Right. With something like software, if you get 1% efficiency, you can update most existing devices and that has a huge impact. With this, I feel like it would be far better to focus instead on lowering manufacturing costs and developing technology to make it easier to produce, rather than more efficient. Being able to deploy 50% more panels now is more important than being able to deploy solar panels that return 10% more electricity.
  - Sparkyte 2y ago
    Indeed if solar panel roofs were cheaper than most roofs. People would be able to replace them with something far more sustainable than someone's side project. I really like tesla's idea of solar roofs and how they implemented them. They need to be lightweight and cheap. Easy to replace.
    - thinkcontext 2y ago
      Tesla's solar roofs are anything but cheap, they are priced as super luxury.
      - Sparkyte 2y ago
        We need solar roofs to be cheap though.
- crote 2y ago
  That might be surprisingly soon, actually. The panels themselves are relatively cheap, and you're spending about half of the total cost on everything besides the panels. If switching to higher-efficiency panels allows you to install 5 panels instead of 6, the resulting savings in labor might already pay for a decent chunk of it.
roughly 2y ago
Listening to some of the processes described for eking out that extra 1-2% efficiency, I'm curious if there's a crossover point where the energy required to get that last couple theoretical percentage points exceeds the lifetime return from the efficiency gain to the panel.
- szvsw 2y ago
  Let’s assume a system outputs 2 MWh annually at 24%. Let’s bump up to 25.5% efficiency, and we get 2.125 MWh. Let’s assume a 30 year lifespan. This gives us an extra 3.75 MWh total over the lifetime. Let’s round down to 3 MWh to very conservatively account for performance degradation. That would be the raw energy budget. It seems unlikely to me that these processes would require more than 3 MWh additional energy for a system of this size compared to the baseline 24% system, but that’s just on vibes! At the same time, it’s probably better to view it in terms of carbon, in which case the situation changes a little bit. You would need to know how carbon intensive the source power production is for the manufacturing process, and additionally how the grid decarbonizes: if the grid decarbonizes substantially due to (eg) massive wind scale up and deep geothermal breakthroughs, then the efficiency gains from PV aren’t as valuable in the future from a carbon perspective as they are in the year 2025.
szundi 2y ago
Isn’t installation costs like frames, transportation and labor makes it irrelevant now?
- szvsw 2y ago
  Lots of discussion of this in other comments in this thread, but at least theoretically, for a system of the same size, a change from 20 to 25% efficiency would reduce: the number of frames to manufacture, the weight of the transported goods, the number of panels needed to be installed, and the space needed all of which do plausibly reduce the total cost and time to deploy, no? That change is itself a 25% performance improvement (over the 20% baseline), meaning you need significantly less space and potentially weight/labor to install. Obviously we aren’t making that jump all at once.
- jillesvangurp 2y ago
  No, those are just the next cost to decimate. And from a technical point of view rather low hanging fruit. Here in Germany the government recently approved incentives for balcony solar. These are cheap panels that you hang (zip ties) from your balcony railing and then plug into a wall socket. No need for approval. These things are exempt from rules intended to protect the look of buildings. The idea is that that puts a little bit of power on the wires in your home and that that is enough for things like your fridge to run off. It partially runs of your solar instead of grid power and lowers your bills a bit. Even a few hundred wh of capacity can already make a difference. You can level up the experience with a portable battery. Of course don't expect miracles from setups like this but if the cost is low enough, why not? Why do I mention this: no professional installers involved at all. You order these things on Amazon or wherever, plug them, in and done. You can get a battery to go with them as well. A friend of mine got himself some panels and he's definitely not an electrician. His balcony is tiny and partially in the shade. I doubt it's very effective. But it didn't cost him an arm and a leg. There's no good technical reason for bigger solar setups to be much more difficult. You can get setups for your garden shed, country house, boat, etc. and it's all relatively easy and straightforward DIY stuff. It's only when these things get connected to the electrical grid and installed on roofs that a lot of rules start applying and a lot of bureaucracy kicks in. Often for good reason like fire safety. But it's gotten way out of hand in a few countries and the cost bares no relationship to the complexity of the problem.
dv_dt 2y ago
Perovskite cells have not been demonstrated to last nearly as long as silicon cells. If you trade cheaper, conversion efficient cells, but they last 1/5 the age, the system cost is much higher because you'll have to re-pay for replacement / reinstallation costs much more frequently.
- boringg 2y ago
  While I can speak to the veracity of your statement. If true it would feel that this is industry finding a way to create a long term subscription fee to buying their product? It would be the worst outcome possible - the beauty of solar is that it lasts for 40-50 years while only having to replace inverter / maintain the system. Thankfully commercial grade and investors wouldn't bank on that.
  - dv_dt 2y ago
    It's not actually clear to me if this article is talking about Perksovite cells at all, or a different set manufacturing techniques improving on current common techniques for the current state of practice. I think should have hung my comment off of some of the other threads on this that were in a side discussion of Perk cells.
  - bradknowles 2y ago
    Things that are exposed to the weather frequently need to be replaced more often than every 40-50 years. Like roofs. And cables. And anything that is an external attachment to the structure. You might not have to replace the solar panels themselves except on a 40-50 year basis, but if you’ve had to replace everything else that’s been exposed on a more frequent basis, I would have to ask Mr. Theseus how much of that solar system is really the same, and how many of those costs would have to be re-incurred over that longer period of time due to the shorter life span of the other products.
    - secondcoming 2y ago
      How often are you replacing your roof!?
rvba 2y ago
The PhD here talks about 1% gains, meanwhile some random website, windows process or chrome itself takes a lot of power - and it seems nobody cares. Yet it adds up too. Not only via ecology, also often poor customer experience.
walrushunter 2y ago
Hell yeah. Big win for the engineers here. Nobody builds better solar panels than us.

230 points by belter 2y ago

The super efficient ones are the panels they send to space. This is just like with chips, all about yields and price/performance. I would not hold your breath for the typical consumer panel to improve much beyond 20%-25% any time soon sadly. It does generate a lot of hopeful breathless articles which rubs me the wrong way. It is important to stay realistic in the search for solutions. Solar is already great and cheap and there are lot more wins possible in the actual deployment as most of the cost is now overheads, bureaucracy, labour, 'etc.

GlibMonkeyDeath 2y ago
It's not a breathless article - these advances are quite doable from an engineering standpoint (and going from 20-->25% is a HUGE deal - that directly reduces the number of panels needed, and so directly reduces overhead and labor.) The more exotic designs which often are the subject of breathless articles (e.g. perovskite https://en.wikipedia.org/wiki/Perovskite_solar_cell https://en.wikipedia.org/wiki/Perovskite_solar_cell or other multi-junction cells) can get greater efficiencies (like up to 40% for quad junction) but are a lot more expensive from a lifetime cost perspective (they don't last nearly as long as silicon junctions and are much more expensive to produce.)
- ziggy_star 2y ago
  I will of course happily buy a 25% panel as soon as it is on the market if the price per watt for the total install is not soaked up by a multitude of other factors. When can I expect that to be?
- IshKebab 2y ago
  I don't know if it's that huge of a deal. The main challenges of solar are not the number of panels now. The incremental cost of installing an extra panel is quite low. The biggest challenges are storage, and - in the UK at least - nimbyism. (Yes people really object to solar panels in fields.) The other thing I would say is that the software for solar inverters is way behind where it could be. You could easily get a 10% improvement just by making them smarter - e.g. using time series prediction, day ahead pricing, etc. Unfortunately they're stuck in the stone ages. I have a QCells inverter (rebranded Solax) - do not buy btw - and they directly told me they are not interested in any of this smart stuff. They also do automatic firmware updates with no opt-out and no notification of changes. And the updates include removing features. Do not buy QCells solar! (Hopefully Google finds this.)
  - jpgvm 2y ago
    IMO with solar it makes the most sense to stick to high volume Chinese stuff. Longi/Jinko/Trina/CanadianSolar/whatever for panels and Deye for inverters. The Deye stuff is particularly good because of aforementioned smart features, especially if you are off-grid or operate on crappy rural power. My favourite feature is called peak-shaving where it uses the battery to supplement said crappy rural power or a generator that is provisioned for average load instead of peak. I haven't found inverters with similar features and configurability anywhere else thus far.
  - justahuman74 2y ago
    Lots of people have limited roof space that isn't in shade for parts of the day. If we're able to cheaply over-provision the panels on the roof then that mean less reliance on the grid (if you use batteries)
dogma1138 2y ago
That’s not very correct the new ISS solar panels for example have a 12% efficiency they are optimized for longevity and weight. The relatively high efficiency figures often touted for some spacecraft panels especially in low earth orbit don’t compare apples to apples as they include the additional 70-80% solar radiation that isn’t absorbed, reflected or scattered by the atmosphere. There are some spacecraft that do use multi-junction cells with very high efficiency however those are ones which are sent far into the outer solar system like Dawn and Juno.

szvsw 2y ago

> “Gaining 1 or 2 percent more efficiency is huge. These may sound like very tiny increases, but at scale these small improvements create a lot of value in terms of economics, sustainability, and value to society.” It’s so easy to forget this and the massive scale and its relevance at the massive scale of the systems we need to (and are, to some extent) roll out. It also seems promising when these breakthroughs are happening in R&D groups of industry players trying to dogfood it rather than in labs. At the same time though, it’s starting to feel to me, to some extent, like we have kind of solved solar? It’s everything else around it that needs to advance, particularly grid infra, batteries and electrifying the general class of difficult-to-electrify problems (steel, concrete, freight). I might be totally off-base and blinkered with that assessment. Edit: I guess I should try to clarify my feeling after reading some of the responses below: it feels like solar tech is not really the limiting factor in renewable scaling, and that advances in solar efficiencies won’t drastically/meaningfully simplify the other challenges/limiting factors we currently face (grid infra/batteries, electrification of mfg, duck curve, etc). Children point out that space and cost savings from efficiency gains in solar may still be significant at grid scale though! Still, this is very cool progress to read about!

bluefirebrand 2y ago
Completely agree. I'm sure there's still a lot of advances left in solar panels themselves, but we really do need to solve the grid, solve storage, essentially solve the base load We won't be able to transition fully off of fossil fuels until we do
Kye 2y ago
I wonder if someone has done a science to compare the carbon impact of batteries and peaker plants. It might be that a lesser of evils fossil fuel is the solution if it's a sometimes thing and not the main thing. It also has the virtue of being supplyable in a renewable way with landfill gas recovery.
- Tade0 2y ago
  A battery made using the dirtiest grid that produces them at scale (110kg CO2/kWh of battery) breaks even compared to gas after ~200 cycles. LFP batteries commonly used as stationary storage regularly do 2000 cycles. Makes more sense to use that gas to make more batteries.

MBCook 2y ago

Is there a known theoretical maximum efficiency for solar cells that we can’t get past with our current approach? If they’re hitting 25% are we close to that limit?

Retric 2y ago
Depends on what you mean by current approach as many different methods are in use. Single junction solar cells are limited to 33.16%. https://en.wikipedia.org/wiki/Shockley–Queisser_limit https://en.wikipedia.org/wiki/Shockley–Queisser_limit Increase the number of junctions and that goes up. Ultimately with infinite layers you can’t beat ~68.7% on earth and 86.8% when much closer to the sun.
- GeorgeTirebiter 2y ago
  These efficiency limits are true for P-N junction solar cells. There are other solar cell types which can ultimately achieve higher efficiencies, but they are currently lab curiosities with low efficiencies.
  - MBCook 2y ago
    That’s the kind of thing I was trying to get at with my “current approach” comment. Maybe solar cells based on chemical reactions or [insert science things here] can do far better, but right now semiconductors is what everyone sells.
- crazygringo 2y ago
  What does closeness to the sun have to do with it? If it's a question of the intensity of the sunlight, can't you just focus it with a lens?
  - missedthecue 2y ago
    Concentrator Photovoltaics (CPVs) exist but are expensive not just because of the lenses but because you need a tracking system to follow the movement of the sun and adjust the lens constantly to keep the light focused.

phkahler 2y ago

TLDR they're at 25.5 percent. Didn't read but it seemed like it's probably a practical incremental improvement which is good.

Sparkyte 2y ago

I can tell you I am excited... but when can we realistically afford solar panels with high efficiency? If a person came by my house and said, "Yo, I can do an installation!". Those panels are like running on a 10 year old or greater design and process.

ehsankia 2y ago
Right. With something like software, if you get 1% efficiency, you can update most existing devices and that has a huge impact. With this, I feel like it would be far better to focus instead on lowering manufacturing costs and developing technology to make it easier to produce, rather than more efficient. Being able to deploy 50% more panels now is more important than being able to deploy solar panels that return 10% more electricity.
- Sparkyte 2y ago
  Indeed if solar panel roofs were cheaper than most roofs. People would be able to replace them with something far more sustainable than someone's side project. I really like tesla's idea of solar roofs and how they implemented them. They need to be lightweight and cheap. Easy to replace.
  - thinkcontext 2y ago
    Tesla's solar roofs are anything but cheap, they are priced as super luxury.
    - Sparkyte 2y ago
      We need solar roofs to be cheap though.
crote 2y ago
That might be surprisingly soon, actually. The panels themselves are relatively cheap, and you're spending about half of the total cost on everything besides the panels. If switching to higher-efficiency panels allows you to install 5 panels instead of 6, the resulting savings in labor might already pay for a decent chunk of it.

roughly 2y ago

Listening to some of the processes described for eking out that extra 1-2% efficiency, I'm curious if there's a crossover point where the energy required to get that last couple theoretical percentage points exceeds the lifetime return from the efficiency gain to the panel.

szvsw 2y ago
Let’s assume a system outputs 2 MWh annually at 24%. Let’s bump up to 25.5% efficiency, and we get 2.125 MWh. Let’s assume a 30 year lifespan. This gives us an extra 3.75 MWh total over the lifetime. Let’s round down to 3 MWh to very conservatively account for performance degradation. That would be the raw energy budget. It seems unlikely to me that these processes would require more than 3 MWh additional energy for a system of this size compared to the baseline 24% system, but that’s just on vibes! At the same time, it’s probably better to view it in terms of carbon, in which case the situation changes a little bit. You would need to know how carbon intensive the source power production is for the manufacturing process, and additionally how the grid decarbonizes: if the grid decarbonizes substantially due to (eg) massive wind scale up and deep geothermal breakthroughs, then the efficiency gains from PV aren’t as valuable in the future from a carbon perspective as they are in the year 2025.

szundi 2y ago

Isn’t installation costs like frames, transportation and labor makes it irrelevant now?

szvsw 2y ago
Lots of discussion of this in other comments in this thread, but at least theoretically, for a system of the same size, a change from 20 to 25% efficiency would reduce: the number of frames to manufacture, the weight of the transported goods, the number of panels needed to be installed, and the space needed all of which do plausibly reduce the total cost and time to deploy, no? That change is itself a 25% performance improvement (over the 20% baseline), meaning you need significantly less space and potentially weight/labor to install. Obviously we aren’t making that jump all at once.
jillesvangurp 2y ago
No, those are just the next cost to decimate. And from a technical point of view rather low hanging fruit. Here in Germany the government recently approved incentives for balcony solar. These are cheap panels that you hang (zip ties) from your balcony railing and then plug into a wall socket. No need for approval. These things are exempt from rules intended to protect the look of buildings. The idea is that that puts a little bit of power on the wires in your home and that that is enough for things like your fridge to run off. It partially runs of your solar instead of grid power and lowers your bills a bit. Even a few hundred wh of capacity can already make a difference. You can level up the experience with a portable battery. Of course don't expect miracles from setups like this but if the cost is low enough, why not? Why do I mention this: no professional installers involved at all. You order these things on Amazon or wherever, plug them, in and done. You can get a battery to go with them as well. A friend of mine got himself some panels and he's definitely not an electrician. His balcony is tiny and partially in the shade. I doubt it's very effective. But it didn't cost him an arm and a leg. There's no good technical reason for bigger solar setups to be much more difficult. You can get setups for your garden shed, country house, boat, etc. and it's all relatively easy and straightforward DIY stuff. It's only when these things get connected to the electrical grid and installed on roofs that a lot of rules start applying and a lot of bureaucracy kicks in. Often for good reason like fire safety. But it's gotten way out of hand in a few countries and the cost bares no relationship to the complexity of the problem.

dv_dt 2y ago

Perovskite cells have not been demonstrated to last nearly as long as silicon cells. If you trade cheaper, conversion efficient cells, but they last 1/5 the age, the system cost is much higher because you'll have to re-pay for replacement / reinstallation costs much more frequently.

boringg 2y ago
While I can speak to the veracity of your statement. If true it would feel that this is industry finding a way to create a long term subscription fee to buying their product? It would be the worst outcome possible - the beauty of solar is that it lasts for 40-50 years while only having to replace inverter / maintain the system. Thankfully commercial grade and investors wouldn't bank on that.
- dv_dt 2y ago
  It's not actually clear to me if this article is talking about Perksovite cells at all, or a different set manufacturing techniques improving on current common techniques for the current state of practice. I think should have hung my comment off of some of the other threads on this that were in a side discussion of Perk cells.
- bradknowles 2y ago
  Things that are exposed to the weather frequently need to be replaced more often than every 40-50 years. Like roofs. And cables. And anything that is an external attachment to the structure. You might not have to replace the solar panels themselves except on a 40-50 year basis, but if you’ve had to replace everything else that’s been exposed on a more frequent basis, I would have to ask Mr. Theseus how much of that solar system is really the same, and how many of those costs would have to be re-incurred over that longer period of time due to the shorter life span of the other products.
  - secondcoming 2y ago
    How often are you replacing your roof!?

rvba 2y ago

The PhD here talks about 1% gains, meanwhile some random website, windows process or chrome itself takes a lot of power - and it seems nobody cares. Yet it adds up too. Not only via ecology, also often poor customer experience.

walrushunter 2y ago

Hell yeah. Big win for the engineers here. Nobody builds better solar panels than us.