Scientists design solar cell that captures nearly all energy of solar spectrum

Scientists design solar cell that captures nearly all energy of solar spectrum
Credit: George Washington University

Scientists have designed and constructed a prototype for a new solar cell that integrates multiple cells stacked into a single device capable of capturing nearly all of the energy in the solar spectrum. The new design converts direct sunlight to electricity with 44.5 percent efficiency, giving it the potential to become the most efficient solar cell in the world.

The approach is different from the solar panels one might commonly see on rooftops or in fields. The new device uses concentrator photovoltaic (CPV) panels that employ lenses to concentrate sunlight onto tiny, micro-scale solar cells. Because of their small size—less than one millimeter square—solar cells utilizing more sophisticated can be developed cost effectively.

The stacked cell acts almost like a sieve for sunlight, with the specialized materials in each layer absorbing the energy of a specific set of wavelengths. By the time the light is funneled through the stack, just under half of the available energy has been converted into electricity. By comparison, the most common solar cell today converts only a quarter of the available energy into electricity.

"Around 99 percent of the power contained in direct sunlight reaching the surface of Earth falls between wavelengths of 250 nm and 2500 nm, but conventional materials for high-efficiency multi-junction solar cells cannot capture this entire spectral range," said Matthew Lumb, lead author of the study and a research scientist at the GW School of Engineering and Applied Science. "Our new device is able to unlock the stored in the long-wavelength photons, which are lost in conventional solar cells, and therefore provides a pathway to realizing the ultimate multi-junction solar cell."

While scientists have worked towards more efficient solar cells for years, this approach has two novel aspects. First, it uses a family of materials based on gallium antimonide (GaSb) substrates, which are usually found in applications for infra-red lasers and photodetectors. The novel GaSb-based solar cells are assembled into a stacked structure along with high efficiency grown on conventional substrates that capture shorter wavelength solar photons. In addition, the stacking procedure uses a technique known as transfer-printing, which enables three dimensional assembly of these tiny devices with a high degree of precision.

This particular solar cell is very expensive, however researchers believe it was important to show the upper limit of what is possible in terms of efficiency. Despite the current costs of the materials involved, the technique used to create the shows much promise. Eventually a similar product may be brought to market, enabled by cost reductions from very high solar concentration levels and technology to recycle the expensive growth substrates.

The study, "GaSb-based Solar Cells for Full Solar Spectrum Energy Harvesting," was published in Advanced Energy Materials on Monday.


Explore further

Solar cell design with over 50% energy-conversion efficiency

More information: Matthew P. Lumb et al. GaSb-Based Solar Cells for Full Solar Spectrum Energy Harvesting, Advanced Energy Materials (2017). DOI: 10.1002/aenm.201700345
Provided by George Washington University
Citation: Scientists design solar cell that captures nearly all energy of solar spectrum (2017, July 12) retrieved 19 November 2018 from https://techxplore.com/news/2017-07-scientists-solar-cell-captures-energy.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
14252 shares

Feedback to editors

User comments

Jul 12, 2017
If you're using lenses to concentrate sunlight onto a pinprick, why bother with stacking the different junctions?

Since a lens is needed anyhow, why not use a prism to separate the wavelenghts into a rainbow, and then just construct the different materials into stripes.

Jul 12, 2017
Actually, quite a good suggestion, E...

Jul 12, 2017
Good one for space use, like a system of these to drive our Mars mission using electric propulsion. Plasma, Shawyer type em-drive, whatever. This system beats burning kerosene and polluting space all the way.

Jul 12, 2017
@Eikka - the record for efficiency was recently held by a team doing basically that.
Prisms spread the light and thus undo some of the concentration, so dichroic mirrors were used instead of prisims, but the idea is the same.


Jul 13, 2017
Eikka:
If you're using lenses to concentrate sunlight onto a pinprick, why bother with stacking the different junctions?

Since a lens is needed anyhow, why not use a prism to separate the wavelenghts into a rainbow, and then just construct the different materials into stripes.


Yes, since the lens shrinks the incident area by the magnification factor of the lens, the room between lenses is available for re-spreading by prism. A continuous bandgap gradient across the prism's projection target could get even more absorption than these discrete layers, covering the full spectrum instead of the frequencies keyed to the entire layers. Plus the layers are not going to be optically perfect, converting some light to heat especially at the refractive interfaces of each stacked layer pair.

Maybe the lens/stack design is simpler to fab than lens/prism using existing semiconductor process. But it sounds like even more than 44.5% efficiency is within next steps, maybe >50%

Jul 13, 2017
The only difficulty I can see with the prism idea is that the different bandgaps all generate a different voltage, which therefore can't be used in parallel, so a continuous bandgap wouldn't work either - it would leak charge from the higher voltage to the lower.

The different bandgaps must be wired in series, which is easily done if they are stacked, which however limits the current because not all wavelengths contribute the same amount, yet each junction must pass electrons to the other.

In other words, the efficiency of the cell is limited one way or the other. Either by the weakest voltage, or by the weakest current.

Jul 13, 2017
The different bandgaps must be wired in series, which is easily done if they are stacked, which however limits the current because not all wavelengths contribute the same amount, yet each junction must pass electrons to the other.

In other words, the efficiency of the cell is limited one way or the other. Either by the weakest voltage, or by the weakest current.


Exactly. In stacked tandem cells the sub-cells are usually photo-current matched by tuning the band-gaps so that each one harvests the same number of photons, and the sub-cell with the weakest current does indeed limit the whole stack.

Normally the cells are grown as crystalline layers on the same substrate and so they ALSO have to have matched lattice constants, but the current record efficiency of 46% is held by mechanically bonding the cells (much like this paper), allowing two different lattice matches to be used.

Jul 13, 2017
Eikka, RealScience:
In other words, the efficiency of the cell is limited one way or the other. Either by the weakest voltage, or by the weakest current.


Could the "bandgap gradient" have a series of transformers alongside that normalizes the voltage/current as it travels through the gradient? That's probably not a photoresist process, maybe a CNT layer bonded to the Si lattice. Or perhaps there's a transformer trick to straining the Si lattice alongside the bandgap gradient...

Jul 13, 2017
@ Osiris1, "burning kerosene and polluting space all the way"?
Good one , polluting space, LOL..
Jurassic period was 3000 CO2.

Jul 13, 2017
@EMC2 - "Limited" in the sense that it puts another set of constraints on on the cells, making everything a bit harder and less efficient.

As for transformers, one DC equivalent is to put more things in series. For instance, if your sub-cells have 1V, 1.5V and 2V, then one could put 3 of the 2V sub-cells in series, 4 of the 1.5V, and six of the 1V, and groups would then B 6V and could be put in parallel.
But if you do this, then the same band-gap sub-cells that are in series need to be current-matched, which means that they need the same focal intensity. That's easy on a linear focus like a trough plus a long prism, but the concentration is too low to afford these expensive cells.
However if perovskites can be made durable enough, then one could do this with a band of perovskites and a band of silicon...

But would it be cheaper than stacking those same materials?

Jul 13, 2017
Yes and like every other solar break through it will be available in 20 years or more.

Jul 13, 2017
Porgie:
Yes and like every other solar break through it will be available in 20 years or more.


This is a science site. It is not a Santa Claus site. From scientific breakthrough to commercial product takes a lot of work, by very different parties, usually taking years. This is true for all the products you buy.

Except with solar, the last 20 years have seen a complete transformation of the products, due to lots of breakthroughs reaching the market far faster.

At least they're doing something to get there. At least the others in this thread are enjoying the breakthrough for its own sake, which is what drives science.

All you are doing is whining.

Jul 13, 2017
RealScience:
However if perovskites can be made durable enough, then one could do this with a band of perovskites and a band of silicon...

But would it be cheaper than stacking those same materials?


Stacking is relatively expensive, compared to a single wafer. Plus the stack will likely have thermal dissipation issues (until it's 100% efficient :).

It looks like a lot of progress is being made in perovskite material properties, including durability and cost. How would a perovskite avoid the current/voltage limits you described?

Jul 13, 2017
Although the world record cells use mechanical stacking, most tandem cells are made by growing the different layers one after another and are almost as efficient (44% for Azur commercial cells vs 46% for the world record, so it is still roughly twice as efficient as silicon.

While the tandem cells currently cost ~1000 times as much per area as silicon solar cells, if one concentrates ~1000x onto the twice-as-efficient cell one gets 2000 times the power per area, so the CELL cost per watt is cheaper than silicon. Currently the 1000x optics are too expensive so silicon still wins on overall cost, but progress is being made on the optics, too.

Low perovskite cost allows Eikka's suggestion to work. Low cost allows using the ~70x linear focus of a trough, which could be split into two (or more) long bands of different wavelengths. Each band could have a long row of cells in series with the lower-voltage cells narrower so that the voltages per width were equal.

Jul 13, 2017
RealScience:
While the tandem cells currently cost ~1000 times as much per area as silicon solar cells, if one concentrates ~1000x onto the twice-as-efficient cell one gets 2000 times the power per area, so the CELL cost per watt is cheaper than silicon. Currently the 1000x optics are too expensive so silicon still wins on overall cost, but progress is being made on the optics, too.


Well, even if 5% of maybe 50% efficiency is lost (so 47.5% efficient) aggregating bandgaps across a layer instead of stacked, that seems like a small price to pay for shrinking the cells price to 0.1%. If innovation moves to packaging and installation that system could cost under $3K to install, and completely power a house even through Northeast Winters (given grid storage). Every private home could install one.

Jul 13, 2017
"Well, even if 5% of maybe 50% efficiency is lost (so 47.5% efficient) aggregating bandgaps across a layer instead of stacked, that seems like a small price to pay for shrinking the cells price to 0.1%."


The balance would be closer. First the efficiency is lower under lower concentration, so the 44% 3-sub-cell stack would come down to a ~~33% linear 2-sub-cell 'rainbow' cell, or a loss of 25%. Depending on concentration, the 44% cells are ~~15 cents/Watt, and the cell placement, substrate, cover glass, frame, shipping, installation, etc are 50 cents to $1 per Watt, so these would go up by 44/33, or 17 to 33 cents/Watt, so even if the 33% cell is free, it would have to use cheaper optics ... but troughs are cheaper, so it would come out roughly even today (if perovskite cells were stable).

But if other costs come down faster than cell costs, then it would start to make economic sense.

Jul 13, 2017
RealScience:
The balance would be closer. First the efficiency is lower under lower concentration


I don't think the cell has to lose the concentrator lens, just add a prism between the lens and the PV surface.

I also think the prism and lens could be first more effective, and then perhaps combined into a single component, by using a metamaterial (or topo insulator) instead of traditional optics. Indeed that wouldn't even necessarily be a separate component, but rather just an etched surface of the PV.

Jul 14, 2017
The 2D focus of the lens is higher concentration , but it makes the matching of either current or voltage hard (as Eikka pointed out), unless the sub-cells are stacked.
Selective absorbers on the PV cell surface is a whole other area - possible, but how do you match currents or voltage (even more important with a sea of tiny elements.

There are very many ways for concentrating optics and PV absorbers to be combined. The catch is that if a way has even one expensive thing, then is can't compete on cost with acres of silicon flat panels.
But many combinations have never been properly analyzed so there may be good ways waiting to be discovered.

Jul 14, 2017
The practical problems involved in getting 1000x concentration of sunlight, or even 70x make the proposition a pure fantasy, as besides lenses (which are not 100% efficient either), you also need to provide cooling, and the whole system cost turns ridiculous.

1000x concentrated sunlight is over 1 MW/m^2 heat.

-"Except with solar, the last 20 years have seen a complete transformation of the products, due to lots of breakthroughs reaching the market far faster."-


The actual "improvements" in practically viable solar panels has come from the manufacture of bog-standard low-efficiency polysilicon panels using older equipment, in massive volumes, with cheap fossil fuels, labor, and lax environmental regulations in China, which now commands about 80% of the market.

All the esoteric types of solar panels, such as CdTe or concentrated photovoltaics are more or less academic curiosities, as nobody's buying them.

Jul 14, 2017
Besides, the color of sunlight as seen on earth isn't constant, so the current/voltage matching of multi-junction cells drifts throughout the day and depends on the atmospheric conditions.


Jul 14, 2017
"The practical problems involved in getting 1000x concentration of sunlight, or even 70x make the proposition a pure fantasy, as besides lenses (which are not 100% efficient either), you also need to provide cooling, and the whole system cost turns ridiculous."

70x is pretty simple and low cost - it has been done for over 100 years. Even 1000x is not hard technically. Cooling is indeed an issue for a large focus, but each each focal point is small enough then a simple aluminum heat spreader works.

As for cost, 1000x CPV has come way down. NREL's "A Bottom-up Cost Analysis of a High-Concentration PV Module" shows current costs <$0.70/W (and a 2-axis tracker only adds ~0.25/W). So CPV already costs less than silicon PV did 5 years ago, and that's at ~1% of the volume.

CPV's challenge is that silicon PV is a moving target and now costs <$0.40/W to make.

So ridiculously expensive? Not any more (but still not as cheap as silicon).


Jul 14, 2017
"The actual "improvements" in practically viable solar panels has come from the manufacture of bog-standard low-efficiency polysilicon panels using older equipment, in massive volumes, with cheap fossil fuels, labor, and lax environmental regulations in China, which now commands about 80% of the market."

Partly correct. But crystalline silicon, not poly, and efficiency has also gone up (16% is now common even in cheap panels). But most of the price drop is a an accumulation of individually minor improvements driven by volume, combined with cheap labor and lax environmental regs (and also cheap capital and below-zero margins to squeeze out other countries).

"All the esoteric types of solar panels, such as CdTe or concentrated photovoltaics are more or less academic curiosities, as nobody's buying them."

CdTe sell multiple gigaWatts per year. Electricity itself was once an academic curiosity, so curiosities can become mainstream.

Jul 14, 2017
"Besides, the color of sunlight as seen on earth isn't constant, so the current/voltage matching of multi-junction cells drifts throughout the day and depends on the atmospheric conditions."

This is a minor factor with today's CPV cells - they have excess lower-junction photo-current anyway so that the important higher-voltage junctions are fully utilized. Yes, they get even more excess 'red' light when the sun is low, but that just wastes lower-energy photons, keeping its impact low.

But the more junctions a cell has, the more sensitive to spectrum it tends to be, so changing spectrum may indeed become a significant factor in limiting efficiency as more junctions are added.

Jul 14, 2017
I just want to note that this is an excellent discussion. A real dialectic from disagreements without being disagreeable. I am learning a lot, and my imagination is stimulated by how Eikka's prism idea is playing out among the existing facts of PV manufacture. Thanks for it.

Jul 14, 2017
Yes, it is nice to have a technical discussion - what this forum was meant for!

If you are interested in more, search on:
"spectrum splitting" cpv

It is a moderately active sub-field - definitely of academic interest.

The question is which (if any) area of CPV will become of MORE than academic interest. Competing against silicon panel manufacturers who combine ingenious cost cutting with lax environmental rules and low labor and capital costs in an industry with a suicidal desire to gain market share by selling below the already low cost is not easy...

Jul 14, 2017
"Partly correct. But crystalline silicon, not poly"

I believe you'll find that polycrystalline silicon is cheaper than monocrystalline, because it requires less refining, and most of the panels on the market are polycrystalline.

-"70x is pretty simple and low cost"


Yes, if you only intend to heat something with a magnifying glass.

-"but each each focal point is small enough then a simple aluminum heat spreader works."


Not quite. The efficiency of the junction drops rapidly with temperature, which should not rise much above room temperature so you need really efficient cooling. At 100x concentration, you're dealing with power densities of around 12 Watts per square centimeter, and you want to keep it below 50 C, so you need a heatsink the size and general shape of a common CPU cooler inside a desktop computer, especially if you want it to run fanless.

"silicon PV ... costs <$0.40/W to make."


You'd pay that just for the aluminium heatsink.

Jul 14, 2017
-"CdTe sell multiple gigaWatts per year."


That's a misleading metric. CdTe has only 5% share of the global market.

CdTe was initially promising because it was heading down to sub $0.50/W but then polysilicon panels became cheaper and the whole thin film PV market took a nosedive. See Solyndra for a famous example. Its main limitation is the rarity of Tellurium, which isn't really all that abundant and it's mainly produced in dubious places.

Jul 14, 2017
"-"CdTe sell multiple gigaWatts per year."

That's a misleading metric. CdTe has only 5% share of the global market.

CdTe was initially promising because it was heading down to sub $0.50/W but then polysilicon panels became cheaper and the whole thin film PV market took a nosedive. See Solyndra for a famous example. Its main limitation is the rarity of Tellurium, which isn't really all that abundant and it's mainly produced in dubious places."

I'm not a fan of CdTe - in the earth's crust tellurium is five times rarer than gold, and tellurium could drive destructive deep-sea mining. But I would't describe 2.5 GW per year as "nobody's buying them"

At a solar show back in 2007 I asked Solyndra: "is it easier or harder to deposit the material inside a glass cylinder than on flat glass?" When they answered "harder", it was clear that they had no chance. Even if silicon prices hadn't plummeted, they would have been beaten by other thin film.

Jul 14, 2017
RealScience:
Competing against silicon panel manufacturers who combine ingenious cost cutting with lax environmental rules and low labor and capital costs in an industry with a suicidal desire to gain market share by selling below the already low cost is not easy...


Certainly true. Even in solar thermal (ie. water heating) the evacuated tubes that are much more efficient (including more incident angle) were killed by the far less efficient cheapest PV. Those rockbottom PV prices are mostly enabled by minimal labor and environment regulations, bankrolled by dumping for to kill competition (especially foreign).

Jul 15, 2017
-"But I would't describe 2.5 GW per year as "nobody's buying them""


That is practically nobody, considering how large the market is and how big it's going to get.

You have to remember that a gigawatt of panels isn't a gigawatt of electricity. For a rough estimate you can divide the amount by ten, to about 250 MW actual production over a year, which gives you a metric: CdTe sells about one medium size powerplant per year, in the whole world, and the sales are declining.

-" Even in solar thermal (ie. water heating) the evacuated tubes that are much more efficient (including more incident angle) were killed by the far less efficient cheapest PV."


That's more to do with subsidy politics. If you're going to put a solar collector on your roof, you get more money out of the feed-in-tariffs or net metering for solar PV, than saving on your gas bill by heating your hot water. The most economic case is electricity with PV, reap the subsidies, and then heat with gas.

Jul 15, 2017
For example, in the US natural gas has 3.2 cents/kWh average price, vs. 11 c/kWh for electricity. In Europe the gas prices are double, but the electricity prices are 2-3x

http://shrinkthat...-compare

So if you get a subsidy for solar PV, it saves you more money than solar heating, which in turn would save you more energy, be easier to store, etc. and therefore reduce more CO2 output. Just another example of why renewable subsidies distort the market and produce worse outcomes.


Jul 15, 2017
I'm not a fan of CdTe, but I'd take a billion dollars a year in revenues from panel sales and another billion a year from installing them. And while First Solar's sales growth hasn't matched the industry, they also haven't gone insane borrowing money to add so much capacity that they make the whole industry unprofitable, and have historically had the highest margins in the industry. Their shipments dropping this year is a transition from gen-4 to Gen-6, having cancelled Gen-5 as not enough of an improvement.

Many a company has died in a product-generation change, especially when the new generation is delayed (the Osborne effect), but some have also come back from low market share. And some influence the industry even with low market share - Apple, for example, had low market share in PCs (but the industry's highest margins), and significantly changed how we use computers.

So while the odds are against CdTe (and I'm not a fan of it), I am not dismissive of it.

Jul 15, 2017
"Just another example of why renewable subsidies distort the market and produce worse outcomes."

As for renewable subsidies distorting the market, I mostly agree. The mainstream panel architecture got its huge lead by working less poorly in Germany than solar thermal and CPV. If solar had grown naturally in high-sun areas, silicon flat panels might have dominated anyway but at least it would have been a fair fight. Solar hot water would have had the advantage of higher energy gain, CSP would have had the advantage of storage, the CPV would have had the advantage of higher efficiency and the time to work out low-cost optics, and silicon's initial scaling advantage of wafers from the chip industry wouldn't have been important.

However to single out renewable subsidies is unfair - for example, we subsidize the fossil fuel industry by not charging for pollution and with trillions in mid-east wars as well as the obvious subsidizing exploration.

Jul 15, 2017
Eikka:
-" Even in solar thermal (ie. water heating) the evacuated tubes that are much more efficient (including more incident angle) were killed by the far less efficient cheapest PV."


That's more to do with subsidy politics. If you're going to put a solar collector on your roof, you get more money out of the feed-in-tariffs or net metering for solar PV, than saving on your gas bill by heating your hot water. The most economic case is electricity with PV, reap the subsidies, and then heat with gas.

No. Feed-in tariffs and net metering can only get your bill to zero. There is practically nobody reaping PV subsidies to pay for gas heat. You can't claim 2.5GW CdTe PV generation is "practically nothing" while citing PV subsidies + gas heat as a "practically something" economic model.

Jul 17, 2017
"Jurassic period was 3000 CO2."

And no Humans.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more