3 ms·

How long would they last? A lot of record breaking cells are lab only due to them degrading so fast that it precludes any practical use. General rule, the pur

How long would they last? A lot of record breaking cells are lab only due to them degrading so fast that it precludes any practical use. General rule, the purer the silicon, the less doping, the longer its life, albeit at low efficiency. Silicon cells above 20% were in labs decades ago, but practical designs with long life only appeared last decade.

_6kzb 5y ago
To my knowledge, having a couple years working photovoltaic research, degradation is only a significant issue in the perovskite solar cells (basically organic molecules that react with or at least see property changes with adsorption of water). Others get maybe a bit of degradation (a couple percent maybe) in the near term, but what they are is what they are. Solid state devices are pretty stable, which is also why CPUs can work for long periods of time (same basic building block, the PN junction, and yet much more complicated). The problem with solar cell efficiency as being the top-line metric is that is that it outright ignores a very complex system. Never mind you got to string them together for panels. Nevermind you just spent $10k making that one cell and your yield is pretty garbage. Never mind that an incrementally more efficient cell doesn't move the needle much when a large fraction of the cost is delivery and installation. Nevermind intermittency is a huge problem for the technology in general. Another important thing to look into for the photovoltaic problem is the Shockley–Queisser limit [1], which shows that we don't even have a lot of room to run in terms of basic efficiency improvements (~50% for Si). That's a fundamental physical limit for single junction cells. In terms of scientific advancements, I would get much more excited to see improvements in energy storage technology. Photovoltaic deployment is probably also going to see more advancement based on improvements in manufacturing, logistics, and building construction. At this point achieving cell efficiency records is more just for the sci-peen. [1] https://en.wikipedia.org/wiki/Shockley%E2%80%93Queisser_limit https://en.wikipedia.org/wiki/Shockley%E2%80%93Queisser_limi...
- ianai 5y ago
  Greatly informative comment! I take that to read this as a way to make a solar panel with an acceptable efficiency rate.
- pfdietz 5y ago
  > I would get much more excited to see improvements in energy storage technology. This whole thread is about silica, so... energy storage with sand! $2/kWh(t) capacity cost, 54% round trip efficiency, baseline scenario $0.05/kWh-cycle cost of storage. https://arpa-e.energy.gov/sites/default/files/2021-03/07%20Day1-Zhiwen%20Ma_NREL.pdf https://arpa-e.energy.gov/sites/default/files/2021-03/07%20D...
- whatshisface 5y ago
  >That's a fundamental physical limit for single junction cells. That sentence contains its own objection.
  - _6kzb 5y ago
    If your implication is that one can simply stack more junctions, that is fine to try from a research standpoint and for mass-constrained applications like space exploration, but the economics get substantially worse for normal applications. It's also the case that multi-junction cells (at the infinite limit) have a max theoretical efficiency that is about double the theoretical max of the best single-junction systems. That also ignores recombination effects at the interfaces between systems, which are also almost pretty gnarly.
    - whatshisface 5y ago
      That's my point, it's an economic limitation masquerading as a physical limitation.
      - _6kzb 5y ago
        Given the definition of economics, I'd say whether it is technically one or the other is kind of meaningless. If you throw enough resources at something, you can achieve all manner of things that don't make any practical sense. Whether such a situation can be more properly understood in terms of "just needing a lot of money," or "having physical limitations making it impractical" is a pants vs. trousers type question. ETA: I didn't follow up my point very strongly earlier, but "double" is important here. Spending many multiples of resources to get cells that at most have double the efficiency of standard single junction cells is not what we usually think of when we think scientific breakthrough. It's certainly not scaling on the level of Moore's law, the expectation that the economy grow something like 3% a year, or various other exponential growth patterns we have come to take as normal. A fun chart to look at for efficiency is provided by NREL [1]. Half the talks on photovoltaics seem to add it somehow. An important point is technologies can get stuck in a particular parameter space for large amounts of time despite tons of money being poured in. Efficiency improvements are not guaranteed, they cost a lot, and they are often tiny when they occur. My macro point, if I have one, is society is best off trying to grab low hanging fruit from multiple disciplines that feed into photovoltaics, and media should try to emphasize that effort rather than pushing cell efficiency improvements, which are not going to do a whole lot by themselves. [1] https://www.nrel.gov/pv/cell-efficiency.html https://www.nrel.gov/pv/cell-efficiency.html
- newyankee 5y ago
  I have seen some Chinese monocrystalline modules being offered with 30 year warranties. This means with < 1% annual degradation (as specified in warranty) no reason well maintained installation can last even longer. I think life of a silicon based cell is no longer an issue, it is the perovskites that are a problem. If perovskites can be made to last longer, we can have very good tandem cells. Another issue is the thin film printed cells, which if they can last even a decade may become a game changer if can be manufactured cheaply enough and have > 10% efficiency.
  - zeusk 5y ago
    but how many would really honor that 30 year warranty? In past, I've had some pretty bad experience getting my alibaba seller to even acknowledge a problem.

_6kzb 5y ago

To my knowledge, having a couple years working photovoltaic research, degradation is only a significant issue in the perovskite solar cells (basically organic molecules that react with or at least see property changes with adsorption of water). Others get maybe a bit of degradation (a couple percent maybe) in the near term, but what they are is what they are. Solid state devices are pretty stable, which is also why CPUs can work for long periods of time (same basic building block, the PN junction, and yet much more complicated). The problem with solar cell efficiency as being the top-line metric is that is that it outright ignores a very complex system. Never mind you got to string them together for panels. Nevermind you just spent $10k making that one cell and your yield is pretty garbage. Never mind that an incrementally more efficient cell doesn't move the needle much when a large fraction of the cost is delivery and installation. Nevermind intermittency is a huge problem for the technology in general. Another important thing to look into for the photovoltaic problem is the Shockley–Queisser limit [1], which shows that we don't even have a lot of room to run in terms of basic efficiency improvements (~50% for Si). That's a fundamental physical limit for single junction cells. In terms of scientific advancements, I would get much more excited to see improvements in energy storage technology. Photovoltaic deployment is probably also going to see more advancement based on improvements in manufacturing, logistics, and building construction. At this point achieving cell efficiency records is more just for the sci-peen. [1] https://en.wikipedia.org/wiki/Shockley%E2%80%93Queisser_limit https://en.wikipedia.org/wiki/Shockley%E2%80%93Queisser_limi...

ianai 5y ago
Greatly informative comment! I take that to read this as a way to make a solar panel with an acceptable efficiency rate.
pfdietz 5y ago
> I would get much more excited to see improvements in energy storage technology. This whole thread is about silica, so... energy storage with sand! $2/kWh(t) capacity cost, 54% round trip efficiency, baseline scenario $0.05/kWh-cycle cost of storage. https://arpa-e.energy.gov/sites/default/files/2021-03/07%20Day1-Zhiwen%20Ma_NREL.pdf https://arpa-e.energy.gov/sites/default/files/2021-03/07%20D...
whatshisface 5y ago
>That's a fundamental physical limit for single junction cells. That sentence contains its own objection.
- _6kzb 5y ago
  If your implication is that one can simply stack more junctions, that is fine to try from a research standpoint and for mass-constrained applications like space exploration, but the economics get substantially worse for normal applications. It's also the case that multi-junction cells (at the infinite limit) have a max theoretical efficiency that is about double the theoretical max of the best single-junction systems. That also ignores recombination effects at the interfaces between systems, which are also almost pretty gnarly.
  - whatshisface 5y ago
    That's my point, it's an economic limitation masquerading as a physical limitation.
    - _6kzb 5y ago
      Given the definition of economics, I'd say whether it is technically one or the other is kind of meaningless. If you throw enough resources at something, you can achieve all manner of things that don't make any practical sense. Whether such a situation can be more properly understood in terms of "just needing a lot of money," or "having physical limitations making it impractical" is a pants vs. trousers type question. ETA: I didn't follow up my point very strongly earlier, but "double" is important here. Spending many multiples of resources to get cells that at most have double the efficiency of standard single junction cells is not what we usually think of when we think scientific breakthrough. It's certainly not scaling on the level of Moore's law, the expectation that the economy grow something like 3% a year, or various other exponential growth patterns we have come to take as normal. A fun chart to look at for efficiency is provided by NREL [1]. Half the talks on photovoltaics seem to add it somehow. An important point is technologies can get stuck in a particular parameter space for large amounts of time despite tons of money being poured in. Efficiency improvements are not guaranteed, they cost a lot, and they are often tiny when they occur. My macro point, if I have one, is society is best off trying to grab low hanging fruit from multiple disciplines that feed into photovoltaics, and media should try to emphasize that effort rather than pushing cell efficiency improvements, which are not going to do a whole lot by themselves. [1] https://www.nrel.gov/pv/cell-efficiency.html https://www.nrel.gov/pv/cell-efficiency.html
newyankee 5y ago
I have seen some Chinese monocrystalline modules being offered with 30 year warranties. This means with < 1% annual degradation (as specified in warranty) no reason well maintained installation can last even longer. I think life of a silicon based cell is no longer an issue, it is the perovskites that are a problem. If perovskites can be made to last longer, we can have very good tandem cells. Another issue is the thin film printed cells, which if they can last even a decade may become a game changer if can be manufactured cheaply enough and have > 10% efficiency.
- zeusk 5y ago
  but how many would really honor that 30 year warranty? In past, I've had some pretty bad experience getting my alibaba seller to even acknowledge a problem.