15-Fold Increase in Solar Thermoelectric Generator Performance
Key topics
A groundbreaking solar thermoelectric generator (STEG) has achieved a 15-fold performance boost, sparking intense discussion around its innovative heat-to-electricity conversion mechanism. Commenters are abuzz, drawing comparisons to Stirling engines and highlighting the Seebeck effect as the key to STEG's direct energy conversion. As the conversation unfolds, a fascinating debate emerges around China's unconventional approach to science and technology, with some arguing it's driving breakthroughs in areas like semiconductors and thorium reactors. Amidst the excitement, others chime in on the bigger picture, noting that solar and battery costs will continue to plummet, making it increasingly tough for fossil fuels to compete.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
3d
Peak period
91
84-96h
Avg / period
16.8
Based on 134 loaded comments
Key moments
- 01Story posted
Aug 26, 2025 at 1:58 PM EDT
4 months ago
Step 01 - 02First comment
Aug 29, 2025 at 11:32 PM EDT
3d after posting
Step 02 - 03Peak activity
91 comments in 84-96h
Hottest window of the conversation
Step 03 - 04Latest activity
Sep 3, 2025 at 2:19 PM EDT
4 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
Given the massive advantage in talent they’ve built up while the Us reverts to Drill Baby Drill we know how this ends.
Eventually the Us with push for atmospheric dimming to “fix” the negative externalities of their approach which had the nice side effect of degrading solar ….
FMl
For enhanced absorbance, alternative technologies that come to mind could be electrodeposition with additives that preferentially bind specific crystal faces, nanoimprint lithography+etching. I was personally involved in developing an electroless deposition technology that created "black gold" that had broadband absorbance across NIR and visible wavelengths. That sucker was black, and easy to make!
No doubt this would be less energy efficient, but perhaps you reap so much savings by not having to worry about water circulation that it's worth it.
But to be fair, you have to consider your « aside », because nuclear has the tremendous advantage of working when it’s cloudy, dark and you need the energy the most in the winter.
I do not think that we can just compare the prices, or maybe we should also add the cost of storage (that is going down too) for solar.
But currently a mix is probably the pragmatic approach.
Nuclear can't compete. https://en.m.wikipedia.org/wiki/Levelized_cost_of_electricit...
Maybe in some far off future nuclear will have a role... But the global energy investment markets paint a very clear picture: solar + wind + battery is the way.
Or when there’s too much and you’re melting your grid.
The main reason for negative electricity prices are inflexible generators, eg. nuclear and coal, because they can't easily (cheaply) ramp down or shut off. Sometimes it is cheaper to let prices go negative than to use emergency mechanisms (that do exist).
Negative prices are not all bad: they are an incentive for storage / flexible demand to step in. Specially, a negative price does not mean the grid is melting.
And like you wrote, that's controlled. Agreed with the rest of your comment, especially the bit that pricing is mostly controlled by the worst parties, not by the best. What we are simply finding out is that a grid designed mostly for baseline loads needs fast response generation (for instance: half of the UK putting their kettle on during half time requires so much extra power that pumped storage becomes a good alternative). And conversely, that if you change the mix considerably that you're going to have to have more control over the cumulative effect of many smaller generators.
But there are already standards for dealing with that even absent remote control of resources: as soon as the local grid voltage that the inverters in modern wind and solar plants see exceeds a very specific maximum for a proscribed period of time they fully autonomously back off their capacity until they are well below those maximums again, and then slowly ramp up to avoid causing grid instability due to oscillation.
What grid balancing is all about is to make this all financially optimal, it has relatively little to do with the safety of the grid, it is simply a way to extract maximum capacity without affecting that safety. A coarser mechanism would simply incur some more waste, but given the amounts of money involved it pays off to tune this.
Ah yes, wind and solar generation crushes the grid (https://x.com/ElectricityMaps/status/1786377006562541825) but that's the fault of all those dastardly nuclear plants germany is littered with, all zero of them (https://en.wikipedia.org/wiki/List_of_commercial_nuclear_rea...)
> Negative prices are not all bad: they are an incentive for storage / flexible demand to step in.
Maybe that'll happen, but currently such events only keep increasing in frequency (https://www.pv-magazine.com/2025/08/26/germany-records-453-h...), and as neighbours also install more solar and wind the ability for germany to maintain their grid stability through exports is going to worsen not improve.
Melting would imply that currents exceed rated capacity of the lines that is entirely impossible due to how the grid is set up. What does happen is that loads that are otherwise not economical to run get turned on and that sources that are remote controllable (which is all wind installations > 2 MW and all solar farms > 10 KW except for residential) are switched off. This is a fascinating subject and worth some study, the thing you want to read up on is called grid balancing.
Typically the day-ahead and the 15 minute ahead markets take care of this with pricing alone and there have been no meaningful excursions due to overproduction of renewables, that's just FUD and it does not contribute to the discussion.
What you could argue if you had read up on this is that there are market operators that do both sides of the market, which sets you up for an Enron like situation because they can make money by front-running. After all, they have a little bit of time between the moment where they know what they're going to do and the moment when they actually do it. Market makers that are also traders is always a dangerous combination and this has already led to some trouble, especially early on in the energy balancing market process. Now it is much better.
https://en.m.wikipedia.org/wiki/List_of_countries_by_renewab...
I'm an empiricist.
FUD about "what about where renewables aren't available " is just rhetorical handwaving. The answer, which already exists at nation-level scale is storage and infrastructure.
That table also doesn't say what you apparently think it does: it lists Luxembourg as 89% renewable, which is true, but does not include that Luxembourg only covers about 28% of the electricity it uses, and imports the rest.
Thus Luxembourg's production being 89% renewable is worthless information as to the viability and reliability of wind and solar for baseload: Luxembourg relies on its neighbours for reliable electricity supply.
Nuclear has its own failure modes. In Switzerland, one of the nuclear plants will be offline for winter (!) due to "unplanned repairs". This will cost the owners of the plant millions.
And also when better tech comes along, you can partially transition a farm to newer panels and resell the old ones after market.
Plus you don't have to build Onkalo Repository like systems to store waste for 100,000 years after you've produced your electricity.
It's wildly more feasible.
Of course the tech and science is cool, possibly useful in space or other niche environments, but whenever I see fusion proposed as some general energy solution, I just roll my eyes and move on.
There's already a convenient fusion reactor fairly close by, and it's unlikely to stop operating any time soon.
https://knowledge.energyinst.org/new-energy-world/article?id...
It's a 12-1 over OGC in what the IEA labels "advanced economies" https://www.iea.org/reports/world-energy-investment-2025
France and China have built nuclear plants in 6 years, and they provide stable power for over 40 years, unlike wind turbines and panels which last maybe 20 for panels (if you're lucky), and a few years for turbine failures, and neither provide stable power.
Renewables have their place but people really need to stop with this panacea nonsense.
Why do think the two countries you mention as being capable of quickly building nuclear are in fact much more quickly deploying renewables?
> Why do think the two countries you mention as being capable of quickly building nuclear are in fact much more quickly deploying renewables?
Short-term political expediency is not an argument for technical superiority or fitness for purpose.
In Fukushima, TEPCO was required pay $1.5 billion. But the real cost is / was around $150 billion. So, the bulk of the disaster was not covered / covered by the taxpayer. So: the public.
According to the French nuclear industry itself a major accident on one reactor may cost more than 430 billion euro (2013). Source (French): https://www.irsn.fr/savoir-comprendre/crise/cout-economique-...
Biz as usual... https://sites.google.com/view/electricitedefrance/accueil#h....
Basically.
The price of solar and battery storage has collapsed. It's really dramatic
This is a log scale https://ourworldindata.org/grapher/solar-pv-prices?time=earl...
Maybe it will reach that point, maybe not but anyways, you can't plan a grid on non-existing tech. Otherwise I'd pick some better non-existing one
Systems like these are just getting started.
https://stateofgreen.com/en/solutions/storing-heat-for-a-col...
https://reneweconomy.com.au/a-near-100-per-cent-renewable-gr...
I don’t feel like doing napkin math on Saturday morning, but you’d need an obscene amount of batteries, the US uses 500+ GWh per day.
Ideally battery storage density will keep advancing to the point where we can use grid scale backup batteries for long durations but we are not there yet.
Case in point: France. A household consumes an average of 14 kWh of electricity per day. The capacity of electric cars will exceed 500 GWh before 2035 and 2000 GWh between 2040 and 2050.
Trucks, utility vehicles, and stationary batteries (domestic and industrial) will add to this. Batteries from retired vehicles will increasingly be converted into static batteries before being recycled (see "Redwood Materials" in the US).
In California, when the sun is at its peak (midday), solar power produces up to three-quarters of the electricity. Batteries are charged in the afternoon, when solar electricity is cheap, and released in the evening, when Californians return home. At their peak consumption, around 8 p.m., batteries can supply up to 30% of the state's electricity.
https://www.energy-storage.news/edwards-sanborn-california-s...
You want a 320 GWh installation?
You do realize HVDC grids can do 3,000km energy travel, right? That's basically anywhere to anywhere, continental US. There's already installs like the PDCI https://en.wikipedia.org/wiki/Pacific_DC_Intertie that take 3GW from north oregon to LA.
There's even transcontinental energy links in the works like this: https://en.wikipedia.org/wiki/Australia-Asia_Power_Link
That's closer to 1% of what California needs by itself then even 1% of the USA's need. We aren't even taking into account the large and continual growth in electricity demand yet either.
Unless I'm mistaken, the US consumption is 500GWh/day with peaks at 700GW/day, so 3GWh isn't going to do much
Which has to factor in the design and cost calculation.
Note in case SMR become part of our grid: what if something similar happens to your hundreds of produced and deployed SMR?
That would be go a very long way to convince me.
Energy got increasingly expensive in Germany the further the Energiewende agenda advanced, to the point that we’re now rapidly deindustrialising.
Turns out base load kinda matters.
That's not exclusively due to the price of energy, though it is a factor, there are other factors (such as the price of wages) that are much larger factors.
The biggest simply being that China is outcompeting Germany on its own strengths through a combination of a lack of environmental regulations, cheap labor and state subsidies at a level that the EU would not tolerate.
You can't really set aside the reality of the electric grid, you have to do with it.
I'd love to have more modern nuclear, but I don't see it happening anymore, no expertise in building them anymore, cost and time overruns all over...
https://en.wikipedia.org/wiki/Radioisotope_thermoelectric_ge...
Neat tech, but very inefficient, to make it efficient fluids start needing to be moved around which hurts reliability. The next step up are pebble bed reactors, I don't think any have been built but the idea is to have self contained fuel "pebbles" enriched enough to get hot but with enough built in moderation so they can never melt down. Then a traditional heat engine is bolted on.
https://en.wikipedia.org/wiki/Pebble-bed_reactor
And I'm not convinced that particular discovery would yield that kind of performance increase for such an application. There are just too many things different in the environment alone.
If you use ambient temperature for cooling, you are severely limited in your total power output. Like, we're talking about less than a megawatt output (depending on how big the ambient heat dispensers are) compared to the ~1GW of a regular old nuclear plant.
You might say: that's fine, let's just build many small ones. But you still need to track your radioactive material, make sure it's not stolen etc., which is a lot of overhead per installation.
Also, I guess you could have the hot end very hot too..? This improving efficiency. Especially if, by virtue of cooling being safer, you could run it at a higher temperature (less safety margin needed).
Explanation starts at ~minute 4. https://www.youtube.com/watch?v=tmbZVmXyOXM
All the designs I know of have a pumped (active) cooling loop for the reactor, then a secondary loop where the coolant (typically water) evaporates and drives a turbine, and that secondary loop is then coupled to either river water cooling or evaporation towers. (There might even be another intermediary cooling loop, not entirely sure).
(You don't want potentially radioactive water to interact with your turbine directly, makes it a nightmare to maintain).
> Also, I guess you could have the hot end very hot too..?
There are some limiting factors:
* melting point of your fuel (you really want that to remain solid, so you can control where it is)
* The reactor core is usually contained in a pressure vessel, which is made out of steel. Steel becomes weaker with higher temperatures. Switching to other materials is super expensive (harder to machine, fewer people are good at machining it etc.)
* You really want to be able to reliably move fuel rods and control rods in and out of the reactor; thermal expansion must be taken into account. At the same time, you want as little leakage as possible out of the reactor core.
A "Boiling Water Reactor" (BWR) has the reactor and the turbine on the same cooling loop. The radioactivity in the water going through the turbine is not a "nightmare", it is a manageable trade-off.
Some major currently-operating BWRs are Leibstadt (Switzerland, 1.2 GWe), Oskarshamn (Sweden, 1.4 GW) and several dozen in the USA. Germany also had some, they were shut down a few years ago (e.g. Grundremmingen).
https://www.gevernova.com/nuclear/carbon-free-power/large-re...
There's many ways in which this can happen in existing reactors. You may have a catastrophic leak and lose the coolant - and you can't just send some welders in, what with radiation, superheated steam etc. The pumps that push the coolant around might fail. Etc. etc.
Even when you "switch off" the chain reaction, the fuel rods keep emitting heat from the decay of transient radioactive elements, enough to need active cooling for days or weeks.
So a lot of new reactor designs revolve around eliminating such failure modes. NuScale for example, IIRC, don't use pumps to circulate the coolant, and that's one thing less that can break.
What I'm daydreaming about simply cannot stop working, in terms of cooling. You have something hot in the middle, you let all the heat get our naturally, and you harvest some of it along the way.
What makes them potentially unsafe is nuclear technology having an incredible energy density, which can be misused and the radioactive material being active even without prior activation. The latter makes many radioactive isotopes a very effective poison.
And misuse or bad practices are a general problem. One can build awful buildings, toys or government structures, too.
And if you have a look at what the worst reasonable non-political consequences from a nuclear powerplant meltdown can be, they're surprisingly harmless.
We have to Soviets to thank for their absolutely incompetent response to the Chernobyl meltdown, that we have a good idea of the long term effects. The powerplant never stopped operating, people kept working there every day for decades. Hundreds of people were never evacuated and hundreds more returned within weeks.
Just to put this into perspective: Chernobyl was effectively a dirty super-bomb dispersing 50t of highly active radioactive materials and yet the death count among anyone who didn't approach into rock throwing distance remains 0.
The question becomes is the power production worth the operation and maintenance costs.
Outside of RTGs.
Turns out there’s companies that do hybrid systems! Water is used to cool the PV, increasing the efficiency of the panels in the process, and then the heated water is used wherever you need it.
Unfortunately it seems there’s only a couple of providers, it’s rare to find installers that do it, and it’s ssuuuppppeeerrr expensive relative to the normal options. Such a shame. I wish there were more options here. It seems like a great approach.
With photovoltaic panels being dirt cheap, we decided to rather heat our pool with a heat pump that is powered by our own electricity.
If you fully price it out you'll find it's more cost effective to just spam more PV and use a heat pump, unless you've reached space limits.
You're looking what the cost would be now and I don't think they were suggesting that, but rather as a direction of development for panels.
Luckily this is exactly how things work and why we have continues progress in the area, including with the batteries. Because 10 years ago you wouldn't even bother with super expensive Lithium batteries for home energy storage and go with NiCd, right?
Some parts get very hot, and any electricity produced without engine or fuel add to range / efficiency.
So my very hot take is that a conventional forced air finned radiator treated with this laser process would show an improvement, it is unlikely to be economically viable versus just using a bigger radiator (at desktop/server CPU/GPU scales). At laptop scales it might be more viable given space constraints.
I feel like someone should have caught that before publication.
mumbles something else about 2^15
Why is that? I can imagine doing two folds on a sheet of paper and ending up with three layers of paper. Imo one fold adds one layer.
Words are arbitrary, but there really isn't any dispute what -fold means as a suffix. See the dictionary entry for it https://www.merriam-webster.com/dictionary/-fold
Also as far as i know, this isn't just an english being weird thing, most germanic languages use "fold" the same way.
What you described is the Log2 Fold Change (log2(A/B)), meaning that if A has a log2FC of 15 over B, its signal is 2^15 times higher, hence ≈32,000x.
You can use this to improve the efficiency of a regular solar panel and as a way to still produce electricity when there is less direct light but enough temperature difference.
I’m even willing to accept a lot of inefficiency because new AC’s in the US can cost 10-20k now and you can need them as frequently as every 7-10 years.
The whole industry is really getting absurd.
It is impossible to make a solid state peltier cooler that is even 25% as efficient as a vapor-compression cycle heat pump at removing heat, therefore it will never gain traction in applications where a heat pump can be used.
Over the lifetime of an air conditioner, the cost of the energy used to run the equipment will be larger than the cost of the equipment itself, so efficiency is important if you want to reduce your total cost of ownership. SEER ratings eventually have diminishing returns per dollar spent, but finding the most efficient unit per dollar is time well spent.
My point is this is rapidly becoming not true.
Thanks for the info on the efficiency though. I’m still hoping we can figure something out. If not , we should switch to propane for the coolant. Way cheaper than whatever they keep making manufacture switch to.
Propane is cheap but it has a couple drawback. Flammability, and the refrigerant must be pure so no mercaptan can be added which means if there’s a propane leak, nobody will smell it. This is more of a problem for commercial and industrial units than residential.
Laser patterning is a lot easier than improving thermoelectric device efficiency.
16 more comments available on Hacker News