Crystal arrangement results in more power from ferroelectric solar cells | Hacker News Reader

Crystal arrangement results in more power from ferroelectric solar cells

pv-magazine.com

264 pointsvinnyglennon80 comments

The 1000x efficiency headline is from Figure 4 of the source PDF:

https://advances.sciencemag.org/content/advances/7/23/eabe42...

Units are Amps per miliwatt on the Y axis and electron volts on the X axis. The black BTO line has a minimum of around 1x10^-10 at 2.8 eV, which is the energy of a photon in 450 nm blue light, and the red SBC222 line has a minimum of around 1x10^-7. Yes, the difference between 10^-10 and 10^-7 is 1000x; good job, headline! But no, that's not the key takeaway from the article.

Solar cells have an efficiency curve based on their bandgap and the energy in the incident light. It's theoretically a straight line, and regular silicon cells get pretty close to it, see the figure here:

https://www.pveducation.org/pvcdrom/solar-cell-operation/spe...

Note the units in A/W, not A/mW, also, the research is at a temperature of 77k, the boiling point of liquid nitrogen, while the above plot is close to room temperature. At 450 nm = 2.a silicon cell produces about 0.35 x 10^-3 A/mW, which represents an efficiency of something like 30% in the real world.

This research and that figure 4 plot fundamentally show that conventional barium titanate = BaTiO3 = BTO, produces very little energy from 450 nm light. The researchers fixed this problem with a multilayer cell using strontium titanate, SrTiO3, or STO, then BTO, then calcium titanate, CaTiO3 or CTO, in an STO_2/BTO_2/CTO_2 crystal lattice stack. That's the breakthrough result; they were able to bridge this gap in energy output and take in more wavelengths of visible light.

The actual efficiency of the starting point was about a million times worse than a good silicon cell. There's plenty of room for improvement when you start that low! The end result is still on the order of a thousand times worse than a silicon cell. A production dye-sensitized solar cell today reaches about 10% efficiency, a monocrystaline silicon cell gets about 22%, but the DSSC has potential to be much cheaper to produce, can be flexible, and is not as well understood. This research may help to close that efficiency gap.

A 1000x improvement on a baseline 1000000x worse makes sense. I was wondering for a bit how a PV cell could break conservation of energy. The article has quite a clickbait title.

That explanation was a tour de force. Thank you. You’ve done your good deed for the day!

I absolutely don't understand all the physics but are "barium titanate" and "strontium titanate" rare stuff ?

No. Depends on your definition of rare, of course. You can see prices for various formulations and purities if you Google. Strontium titanate is used for jewelry, as it was one of the main diamond simulants back before cubic zirconia. I did my PhD working with crystals of strontium titanate and contributed to two Wikipedia articles on it.

Looking at the elemental abundances you can see that titanium and strontium are fairly high, above both gallium and arsenic, another solar cell pair. Barium is in the ballpark of gallium and arsenic.

https://en.m.wikipedia.org/wiki/Abundance_of_the_chemical_el...

Any multi junction cells do considerably better than that 22%:

https://en.wikipedia.org/wiki/Multi-junction_solar_cell

I wonder what the difference in energy demands for producing them are.

I do see that a ferroelectric perovskite can do 22%. Maybe this could be the outer protective layer for the perovskite cell. Isn't the main perovskit engineering issue the degradation when exposed to the atmosphere?

Ferroelectric seem to do other interesting thing, like harvest from ultraviolet, and exhibit some "bulk effects" that exceed "theoretical" maximums, but obviously none of that has been found to make a usable commercial cell.

I hope PV research is still being aggressively funded. It seems to me that solar cell costs could be dropped another 50-75% under silicon with the right combination of perovskite and other materials.

The future of our climate hinges on it.

Reminds me of my masters research at university. My work was based on hybrid organic-inorganic solar devices based on boron subphthalocyanine chloride and an optimised optical spacer layer. One day, one variation of the type of cell I was producing was over double the output efficiency of what we usually expected. Was exciting until we realised the solar testing rig had been erroneously set to the equivalent output of two suns...

One day… two suns. Talk about playing God. :-)

That sounds like fascinating research.

I mean, do we really need Jupiter as is? Everything Jupiter has done for us with clearing out the trash from the solar system would have been done by a 2nd smaller sun as well.

Here we are, trying to improve solar cells to avoid global warming, and dissidents like you want to add a sun to the solar system.

It's not the solar system's fault that the inhabitants of one of the bodies in the system are bad stewards.

Also, if you have 30% effeciency from the light of one sun, wouldn't the additional light also improve the solar cell since there is now more light to get 30% out of? I mean, more is always better, right?

I know your comment is in jest, but in point of fact the additional heat from another star would almost certainly reduce the efficiency of the panels.

Can you explain why?

It'd be twice (>1<2) the energy

And I thought we were supposed to reduce global warming ;-)

There's something hilarious in adding a sun so you can improve the efficiency of your solar panels, so that you can switch to renewable energy and offset warming caused by your current single sun (and polution).

> set to the equivalent output of two suns...

I didn't realize we have achieved a Kardashev Type II civilization already :)

Achieving the effect locally for a few milliseconds has been possible for like 70 years now.

Damn hard to get a license for that nowadays, though, no matter how important your science is.

Is...this a nuclear explosion joke?

That would be a... double-whoosh?

http://wordpress.mrreid.org/2015/04/18/the-nuclear-double-fl...

Haha, very good.

In fact, we had some equipment similar to that on this site https://solarlight.com/product_category/products/solar-simul...

Basically a super bright light source with a similar spectral output to our sun. A roughly 1x1cm solar cell in a nitrogen filled chamber is wired up to check the output under the incident light. Very cool production process. I was extremely lucky to have had the opportunity to help with the research.

For those like me, who didn't get this at first.

It's about improving output from ferroelectric solar cells, a specific type of solar cells, which are easier to build than regular ones made of silicon. But they usually aren't as effective as the silicon ones.

yup, my thoughts exactly. otherwise a 1000x increase in 30%-ish efficient pv silicon-based solar cells would be far above unity

Depends on how you count.

If your process is 30% efficient, it wastes 70% of input. If you increase efficiency 1000x, it wastes 70% / 1000 = 0.007% of input, so it's 99.993% efficient.

It could also mean that the ferroelectric cells were harvesting 0.000001% of sunlight, while now they harvest 0.001%.

The fact that no absolute number is given in the article makes me want to boycott it.

That would be a 1000x decrease in inefficiency. Very, very impressive but not even remotely as impossible as a 1000x increase assuming that you start with more than 0.1%.

I'd expect the headline to resolve to some perspective where the numbers do make sense, in some very special, limited way, assuming that they didn't just roll dice to decide how many orders of magnitude to falsely claim.

Unfortunately, there's a huge crowd out there that combines, in a surprising way, radical science scepticism with radical science optimism. The "science said man cannot fly, then came the Wright Brothers" crowd who refuse all painful achievable improvements based on their hope for some convenient miracle. These headlines are dangerous.

I'm not certain about the "easier" part, but with efficiency higher than a semiconductor cell can ever be, this device should draw attention, no matter how hard is it to produce currently.

Ah, good to know.

Yes, easier might be not right here, but it sounded like overall they are less complex.

Reading the article and the comments... No, these are not more efficient than silicon cells. They took an inefficient, simpler to produce PV material and found a way to dramatically increase its efficiency on certain wavelengths of light. It's still worse than silicon panels overall.

So, without a p-n junction, then we can break the Shockley–Queisser limit. Which states that in traditional photovoltaic cells you'll never get past 1/3rd efficiency. And until now I didn't even know that ferroelectric absorbers were a thing. Pretty amazing! https://www.nature.com/articles/nphoton.2016.143

I thought solar conversion efficiency was a large fraction already. There's no room for 1000x improvement. Maybe 20%. What is this title talking about?

A similar question was in the comments of the article and reply from the Author:

---

Ralph A Hutton August 6, 2021 at 1:40 am

1000 times? That means a normal size panel ,area about one square meter, that produces 200 watts , would be capable of an output of 200 kW. I look forward to that. I believe, and this needs confirming, the maximum available power from one square meter is about one kW. What have I misunderstood? Reply

---

    Ryan Kennedy
    August 6, 2021 at 1:46 am

    Our apologies for not making this clear. The photovoltaic effect was increased by 1000 times compared to previous output achieved from cells made of ferroelectric crystals, not from prevailing solar cells made of silicon or other conventional materials.**

Can we change title to reflect this?

I don’t know if the title was already changed since you wrote this comment, but the current title of “Crystal arrangement results in 1,000x more power from ferroelectric solar cells” is accurate.

Can someone explain this to me like I'm 5 for me?

the previous best output for this type of solar cell was nowhere near the state of the art. they found a more efficient structure that narrows the gap. the advantage of this type of solar cell is that it’s potentially much easier to produce, but before now it’s efficiency has been very low compared to the state of the art.

This Red widget is 1000x better than older Red widget designs. Even though Green widgets are still much better.

But red widgets might be cheaper to mass produce. Maybe. I don't know I'm knee deep in analogies now.

But how long do red widgets last compared to the green widgets? Since red widgets have been so woefully under performing compared to the green widgets, has anyone seriously looked at the red widget for long term viability or is this just flash in the pan type performance? Just to beat up the analogy further.

From what the article stated, it seems the longest term testing they did was firing a laser at the red widget for 6 months straight and it maintained constant current the entire time

Say ferroelectric crystals used to be 0.01% efficient, and this paper brings it to 10%. That would be an increase of 1000x, but still below the current best which is ~20%-30%. (The 0.01%/10% percentages are illustrative only, I don't know what is claimed in the paper.)

in the email above, "that produces 200 watts" is the wrong baseline. That's the current value for a different technology.

just look for the qualifier

in this case "ferroelectric"

I think the confusion people are having with "Crystal arrangement results in 1,000x more power from ferroelectric solar cells" is that it doesn't have a qualifier like "more power than current ferroelectric solar cells." It only tells you what type the new ones are.

You have to infer that it's comparing new ferroelectric cells to old ferroelectric cells, and not comparing new ferroelectric cells to our typical silicon cells.

But an article doing the latter could be written with exactly the same title.

The article says “the current flow was up to 1,000 times stronger”. I don’t know what that means.

These are ferroelectric, rather than the more efficient silicon cells used for most solar energy.

I did a quick Google for the Power Conversion Efficiency of Barium Titanate based solar cells and they are currently mostly in single-digit percentages [1]. A thousandfold increase would be over 1000%. Unless I'm missing something important, they probably mean 1000x over a super basic BaTiO3 system, not 1000x over the current state of the art.

[1] For example, this 5.13% PCE from "hierarchical barium titanate micro flowers": https://iopscience.iop.org/article/10.1088/2053-1591/aace86

From the comments section of the article:

> Our apologies for not making this clear. The photovoltaic effect was increased by 1000 times compared to previous output achieved from cells made of ferroelectric crystals, not from prevailing solar cells made of silicon or other conventional materials.

In other words they made something work that really didn't before.

Even if they made the best theoretically possible solar cell, the starting point of 1/1000th of that would not have been considered useful for anything.

To get the 1000x figure, the research is comparing photocurrents from bulk BTO and their special crystal-aligned BTO at a specific wavelength of incident light (306nm).

They also compare photocurrents under a solar simulator and achieve 100x increase vs bulk BTO.

At no point do they make comparison to state of the art BTO cells, which almost certainly aren't just bulk BTO but are likely doped, nanostructured, or have some other novelty.

That is true. The research is good. But the headline, the article itself, and especially this comment by the author

misleadingly make it seem like they made 1000x progress from the current best.

The other day, I watched this documentary on a huge off-shore wind project off the coast of the UK. It uses some of the largest windmills ever created.

This project has a mind-boggling amount of complexities and involved a special large ship, one of the only kinds of its type, to move things around—very impressive project.

It made me realize just how revolutionary solar is. Also, why it can compete and be so cheap compared to other forms of energy production.

You just put this almost magical thing on your roof, and it silently creates green renewable energy for decades. Power plants can be complex and expensive buildings. Solar is just stupidly simply.

I was told that an offshore oil exploration drilling operation costs as much as the most recent Mars rover mission ever will. When it comes to keeping the world moving, there's room for ungodly amounts of complexity.

Solar cell installation is pretty easy, but there is some non-negligible complexity in producing them.

What is:

Power output per m2?

Durability (life span)?

Production cost?

At least theoretical ...

The article tells you the answer to your question. Search for "1,000".

Edit: Apologies, I misread the parent's question. I thought they were saying "1,000 what?". Sorry about the unhelpful answer.

No it doesn't. The 1000x factor refers to the photo-response of the material, not power output, not cost to produce, not durability.

All these are properties of a (prototype-)product, while this research is basic research and didn't create any prototypes. The researchers measured the photo-response by shining laser light onto a sample, they didn't build a panel and measured the power output.

Not 1000x net output since we are in what now, 10-20% efficiency with current available cell technology? That would not make sense.

From one of the comments on the article:

  Ryan Kennedy
  August 6, 2021 at 1:46 am
  Our apologies for not making this clear. The photovoltaic 
  effect was increased by 1000 times compared to previous 
  output achieved from cells made of ferroelectric crystals, 
  not from prevailing solar cells made of silicon or other 
  conventional materials.

Ok, now we just need to know how ferroelectric crystals compare to conventional solar cells to find out what this "1000x" really means...

i would guess that they are about 1000x worse than conventional silicon

Or much worse yet, otherwise the authors could have used clear language.

Yes, it would be extremely helpful if we knew how they compared to what people are able to purchase today.

Non of that can be answered just yet. This is basic research in its earliest stages.

It can't even be answered theoretically, because the researchers are in the process of developing the theory that could answer these questions.

So far they only made an observation and need to go on from there.

This type of cheeky criticism is probably fair when you have a startup claiming crazy numbers with no prototype, but this is reporting on a paper from a research institution. No need for snark.

criticism, snark ?!

What are you talking about?! All I am trying to get some numbers, if anyone has them, but it seems amount of unhelpful comments begins to be overwhelming on HN.

Seems like some people are unable to handle actual questions on anything that appears challenging and want to dismiss as being hyper critical and snarky for not just drinking the kool-aid.

There was no snarky tone-of-voice in your reply when I read it. Maybe cultural differences? Please, kindly inform me of the output of the power output per m2? Also, is there any information available regarding the durability (lifespan) of the product? Who's got time to be that formal on something as an internet forum? I read topic, now here's some questions I have is perfectly fine. Some people possibly just need to get thicker skins and realize that not everyone out there is trying to be an asshole and just stop reading things in that manner.

Honestly, I do not know how much technical my question can be, there is no tone, there is 14 words in total. I was googling and I could not find anything useful so I hoped I could get some answer from some technical person (material science/physic person). In my mind finding those number is the first thing I would do on the test bench, they give you prospect of technology some kind a range you would attract investors. You would have test sample 1X1cm you shine that amount of light and you get output. And then you know this is feasible and can help in future or not... simple as that. Instead I am getting some kind responses like I am talking with emotional teenagers tripping on some cry baby drugs... for heaven sake ... (yes that last is emotional response, I mean seriously what is going on with HN lately it used to be place with tech guys talking tech ...)

I suppose the tone could come exactly from the brevity. Imagine stepping up to a bakery and hearing “what?” instead of “how can I help you today dear gentleperson?”.

Which one would you find more snarky?

Even if the bakery employee yelled "What?" at me, I would not hear snark in that. I would probably hear rude, but not snark. I would also be a little less surprised if it was a very busy place and the employee was visibly harried by the commotion. Or if I was just in any store in NYC.

There's a difference between terse and to the point vs being rude. It's not the terse response's fault the recipient cannot tell the difference.

Also, are we using snark as a synonym for rude nowadays? I'm going to need to update my definitions as I was always understanding snark to be sarcasm. I just want to be using the same words as everyone else on the internet so I don't offend. (that was snark).

Edit: actually, if I went to a store and someone asked "how can I help you today dear gentleperson?", I would immediately assume they themselves were being snarky. That's like saying "bless your heart". It's just dripping with snark.

I always assigned “rude sarcasm” to snark but apparently it is just sarcasm.

FWIW, I didn't detect any snark in your words either. Perhaps brevity was mistaken for snark?

For the rest of us that read no snark, please enlighten us into what words/phrasing led to your reading of snark in the parent's post. Curious minds would like to know answers to the actual questions posed by the parent, but also to what sensitivities are being upset by asking those questions.

They don't cite power output, but cite current output: 17 mA / cm² under full sun, pretty impressive if you ask me.

The key promise of this approach is "more efficient that silicon can even theoretically be". Overcoming the Shockley limit is the key finding.

Thank you, is there any way we could find out what is probable life span, is it more durable (hailing/dust/corrosion) or less than silicon crystal? From time to time it would be nice if we had AMA on HN like on Reddit for the research team.

Also mA does not says much if we do not have voltage, and I am not sure what full sun means as in Africa full sun can mean 4.5 and 6.5kWh per m2 while in Europe we can get between 1 and 2.5kWh per m2 or radiation.

From the underlying paper[1], I found this excerpt:

“The photocurrent or the short-circuit current density (JSC) extracted from BTO is around 0.415 μA/cm2. … The JSC value from SBC555 is around 11.03 μA/cm2 and is about 25 times higher than measured in BTO. The open-circuit voltage (VOC) in the case of BTO was found to be around −0.007 V, as opposed to −0.058 V in SBC555.”

Here, BTO is the existing material and SBC555 is the new material from the paper. Multiplying JSC by VOC to get power density , we see that BTO yields 2.9 nW/cm^2, while SBC555 yields 639.74 nW/cm^2.

For reference, according to [2], we have that “typical external parameters of a crystalline silicon solar cell as shown are; Jsc ≈ 35 mA/cm2, Voc up to 0.65 V.” Notice the unit difference of mA/cm^2 here vs uA/cm^2 above. Multiplying these parameters yields 22,750,000 nW/cm^2. So we see that these cells are still approximately 100,000x worse at producing power than the current crystalline silicon cells.

[1] https://advances.sciencemag.org/content/7/23/eabe4206

[2] https://ocw.tudelft.nl/wp-content/uploads/solar_energy_secti...

Thank you sir!

Better title: "Extremely awful solar cell technology improved to be just really bad"

Is 1000x from current efficiency good enough?

Reply on news.ycombinator.com