Which solar-electric technology had the highest EROEI?

Energy Returned On (Energy) Invested, also known as EROI and as EROEI, has become something of a fashionable metric *but see footnote.

It is calculated by dividing the total energy delivered by the system throughout its whole lifetime, divided by the total energy required to build, operate, maintain (and ideally decommission) the system.

Please refer to specific peer-reviewed energy studies where possible.

Within scope for this question:

  • PV - the different and distinct types of photovoltaics:
    • Cadmium Telluride
    • Polycrystalline silicon
    • Monocrystalline silicon
    • Amorphous silicon
    • CIGS
    • organic solar cells (perovskites etc)
  • CPV - concentrating photovoltaics
  • CSP - Concentrated Solar-thermal Power driving a turbine (pictured below)

are all in scope, along with anything I haven't heard of that converts sunlight into electricity, without going through convolutions such as biomass, wind or wave.

enter image description here

Out of scope:

Attempts at doing it purely through prices do not answer the question: price is emphatically not a useful measure of total energy input, because the factors of production (land, capital, enterprise, labour) are not zero-cost, and because the ratio of energy to other inputs varies hugely by manufacturing technique and by product. So that approach is explicitly out of scope, because it's just plain wrong.

Also out of scope:

  • Solar thermal (that delivers heat but no electricity)
  • indirect solar generation such as wind, wave and biomass
  • tidal and geothermal
  • coal, gas, oil, nuclear.

* footnote on EROEI: Whether or not EROEI has much significance is disputed:

  • some tie the very existence of civilisation as we know it to society achieving some minimum arbitrary level;
  • others consider that as long as a generation source has a value greater than one, then the economics matter, and the EROEI does not (except solely for its impact on the economics). IF EROEI is less than one, then it can still be useful: not as a generation source, but as a store or conversion of energy.
  • 3
    You're not going to like this as an "answer", but the best investment is conservation... Insulation, high-efficiency lighting, etc. It generally requires much less energy to produce a simple product like insulation than to produce some exotic energy-conversion device, and the ROI is quite high. Even a fancy LED lighting system with automated controls requires less energy to produce than would be required to produce the generating capacity necessary to power it. Think of it as the difference between paying off your debt and stashing money in a savings account. Paying off the debt reduces the am – Dave Baldwin Nov 23 '13 at 10:04
  • 4
    While Dave is correct it doesn't answer the question "which SOLAR technology has the highest EROEI." Dave is answering a very relevant question, "What is the best energy investment?" – hortstu Jan 11 '14 at 19:02
  • I'm going to suggest biomass, specifically waste biomass, unless you want to explicitly rule it out as being another indirect option. Similarly, second hand PV is likely to come out on top since the cost of acquisition is likely to be very low (and there's a lot of unused/need repair PV systems around). – Móż Jan 31 '14 at 11:51
  • The other caveat with EROEI is that if that's your only measure fission will often do very well indeed, since the problems with that are energy "leakage" rather than energy investment. Although you probably again count fission as indirect solar, since the power isn't coming from our current sun. – Móż Jan 31 '14 at 11:55
  • Any answer to this will be very much a 'as of last Tuesday, the answer was..." as the technology is in rapid flux. In addition, I know of no sources that give an really accurate energy cost of much of anything. You can get approximations for bulk things like concrete or aluminium, but as soon as things have more than one part, you have issues with transportation, assembly, and so on. It's not clear to me that it's possible to get better than price as a measure, particularly since different PV production uses similar processes. – Sherwood Botsford Feb 21 '14 at 1:59

TLDR; cadmium telluride panels have the highest EROI of the mentioned PV panels (around 34.2)

EROI estimates vary widely. This is because of differences in the method of calculation, scope of the study, installation location, assumed lifespan of a panel, etc. This also makes it relatively easy to manipulate EROI calculations.


The most pessimistic estimate I've read is this article where the (apparently pro-nuclear energy) author claims that solar PV has an EROI of 3.9 (in Germany) and 1.6 when you buffer excess energy. He also says solar CSP in a desert has an EROI of 19 (and 9 with buffer). I found the original article on which these numbers are based. To be more precise in that article they've calculated the following numbers:

EROIs of Solar PVs with 1,000 peak hours per year in Germany
                                                EROI     EROI buffered
Solar PV Poly-Si Rooftop                         4.0      2.3
Solar PV Poly-Si Field (extra steel frame)       3.8      2.3
Solar PV Amorphous Rooftop                       2.3      1.6
Solar PV Amorphous Field                         2.1      1.5

EROIs of Solar CSP based on proposed plant in Marocco
  with a solar radiation constant of 2340 kWh/m2**
Solar thermal (CSP) parabolic (phenyl coolant)  21.0      9.6
Solar thermal (CSP) Fresnel (steam coolant)     17.0      8.2


In contrast, in this optimistic paper written by a director of a PV research center they've calculated the following:

EROIs based on the EPBT values for mono-Si, multi-Si and Cadmium Telluride 
  photovoltaics for the Southwest of the US.
Solar PV mono-Si    25
Solar PV multi-Si   25
Solar PV CdTe       50

Truth is the middle?

I'm not sure how reliable any of the mentioned numbers are. The truth probably lies somewhere in the middle. This seems to be confirmed by the results of this 2015 meta study where they reviewed 232 studies published between 2000 and 2013. 23 studies that met their criteria were analyzed further (11 on mono-Si, 11 on poly-Si, 7 on amorphous Si, 11 on CdTe and 8 on CIGS). The conclusion was that the mean harmonized EROI varied from 8.7 for mono-crystalline silicon panels to 34 for cadmium telluride panels (see also the picture below). Sadly they didn't investigate CSP or CPV.

Mean harmonized EROI with error bars representing one standard deviation

Source: K. P. Bhandarib et al. (2015) "Energy payback time (EPBT) and energy return on energy invested (EROI) of solar photovoltaic systems: A systematic review and meta-analysis", in Renewable and Sustainable Energy Reviews, Vol 47, Pages 133–141. Link to full paper

  • Solar EROI also depends on location. For Germany, it's 3.9 according to this page. – Cees Timmerman Feb 16 '15 at 11:15
  • @CeesTimmerman Yes, EROI does depend on location, but also on your method of calculation (which processes/parts of construction and maintenance do you include or not). The article you link to seems to refer to the same paper I mentioned in my answer. I didn't mention Germany so thank you for pointing this out, I will edit my answer and include it. On another note; one thing that is still unclear to me is how they derived the 3.9 (in figure 3) since that number cannot be found anywhere else in the paper. – THelper Feb 16 '15 at 11:41
  • 1
    It's 2102 MJ (roof) vs 2172 MJ (field), so close enough to round like that. – Cees Timmerman Feb 16 '15 at 15:41
  • 3
    @NeilTyson My main argument is that a definite answer is not possible because there is no single agreed-upon method of calculating EROI. – THelper May 26 '16 at 11:28
  • 4
    @NeilTyson primary sources are a bad idea in this case, meta-analyses are much better for broad comparisons. Otherwise you end up saying "this exact item has these values" but have no way of knowing how unusual "this exact item" is, and how it compares to anything else. The measurements are likely to be non-comparable as well, since they'll probably be done in ways that are easy for the exact item. Much better to use standard measures and a comparative study with consistent criteria.That's if you actually want an answer, of course, rather than "I'm just asking questions" that never stop. – Móż May 26 '16 at 22:55

http://www.ce.cmu.edu/~gdrg/readings/2005/12/07/Williams_EnergyChemicalUse.pdf following the above link you will find there on page 2(185):"According the 1997 National Technology Roadmap for Semiconductors, average electricity consumption was 1.4 kWh per square centimeter of silicon wafer processed". I do not think figures are much lower now. Besides, they are do not include assembly, transportation and buffer. If there was EROI=1 it would be a magic.

  • 1
    There have in fact been significant improvements in solar cell technology since the late 1990s: NREL's solar PV efficiency chart shows progress for all major (and emerging) solar cell technologies. – LShaver Jan 20 at 21:00
  • This answer has very limited validity: it is from 1997, it is about chips like DRAM, processors and not photovoltaic cells. – Jan Doggen Jan 21 at 12:53


Solar EROI is far below 1. The above-one EROI estimates are due to various shenanigans, such as excluding infrastructure costs..

You can also see that solar prices just track oil prices.

The comments below also ignore infrastructure costs and use industry-reported data despite obvious conflicts of interest.

  • 5
    Two things caught my eye in this study. First, in the conclusion they state EROI is 0.82, and maybe +/-15%. But, their range and error calculations throughout the paper are absent and/or inconsistent. Second, in at least two places, they cite as their source their personal experience. I found a rather thorough critique of the paper here. Finally, compared to Ferroni et al's meta-study of 28 papers, this study of 231 papers found an EROI value of 11-12. – LShaver Jun 28 '16 at 9:56
  • There's no point in you replying if you're just going to repeat the same methodological errors. – D J Sims Jun 28 '16 at 16:18
  • 2
    @DJSims The study you quote included finance costs as part of the EROEI cost, which is unusual and probably excluded in the question above. Just dropping that element from the paper puts the return above 1. Alternatively, you could re-do the same analysis including finance-as-energy for other electrical generation systems, and I expect you'll find they all drop below 1 as well. Which suggests to me that the authors should revisit their decision to include purely financial aspects in an energy calculation. – Móż Jun 30 '16 at 1:20
  • 2
    @DJSims since you seem unable to explain the errors you mean in enough detail that other people can understand them, would it be reasonable to conclude that our error is in responding to you at all? – Móż Jul 1 '16 at 1:12
  • 1
    @LShaver A group of scientists wrote a response paper in which they identified "several critical methodological and calculation flaws in that paper": sciencedirect.com/science/article/pii/S0301421516307066 – THelper Jan 21 '17 at 22:33

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.