Photon recycling effect could be observed in case of the presence of appreciable band to band  radiative recombination. If the light emitting effect is pronounced then one can use optical means as you stated to prevent it from escaping out of the material and forcing it to be reabsorbed again.

In order to reach such condition, the radiative recombination must dominate over the nonradiative recombination mechanisms over traps. This would require a very advanced materials with very low defects concentration in the volume and at the interfaces. But as Eugene reported, there are still many defects in the material such that the dark current is dominated by the nonradioactive recombination.Such type of solar cells are characterized by very thin active layers, so it is very probable that the generated electron hole pairs will be disposed from the material in a time that may be much shorter than the radiative recombination lifetime. 

Simply said the recycling can be affected when the material can work as an effective LED.

Yes you want to increase the theoretical limit of the the conversion efficiency by recycling and so minimizing the dark current. From the principle point of view why not? okay, it is conceptually possible even so when the improvement is small.

Nobody observed the PR effect in perovskite-based PV. In my view, the reason is in high level of non-radiative recombination (high density of traps). The junction quality facor is ~2 under 1 sun. For the best GaAsv cells we observed very small improvement of Voc by PR [Braun et al, Energy Environ. Sci. 6(5), 1499–1503 (2013)]. The reason is in the very low QE of radiative recombination even in the best GaAs cells. Perovskite PV are still much far from the realization of the idea

Photon recycling effect could be observed in case of the presence of appreciable band to band  radiative recombination. If the light emitting effect is pronounced then one can use optical means as you stated to prevent it from escaping out of the material and forcing it to be reabsorbed again.

In order to reach such condition, the radiative recombination must dominate over the nonradiative recombination mechanisms over traps. This would require a very advanced materials with very low defects concentration in the volume and at the interfaces. But as Eugene reported, there are still many defects in the material such that the dark current is dominated by the nonradioactive recombination.Such type of solar cells are characterized by very thin active layers, so it is very probable that the generated electron hole pairs will be disposed from the material in a time that may be much shorter than the radiative recombination lifetime. 

Simply said the recycling can be affected when the material can work as an effective LED.

Yes you want to increase the theoretical limit of the the conversion efficiency by recycling and so minimizing the dark current. From the principle point of view why not? okay, it is conceptually possible even so when the improvement is small.

Thanks for your reply very much! Now I have some new ideas, we will do some experiments and get your back soon.

For direct bandgap materials with very high absorption coefficient, light trapping or photon recycling is almost useless for practical devices. Only if you can get perfect materials, the so-called photon recycling might help a bit for the efficiency limit. How can you improve the performance of a polycrystalline solar cells with such a concept?

Light trapping is still useful even if the absorption of materials is high. Because light-trapping solar cells achieve the same absorption with a thin active layer. Hence, the recombination could be reduced. Auger recombination is very large in polycrystalline solar cells and thus photon recycling is useless. However, for GaAs solar cells, the device efficiency can be improved. You can see experimental results in http://onlinelibrary.wiley.com/doi/10.1002/aenm.201400919/full

Never forget that the light trapping structures may considerably generate higher defects density. In direct bandgap solar cells, all the state-of-the-art devices (including GaAs) are based on flat structure (no light trapping, only AR). As for polycrystalline solar cells, such as Perovskites, the SRH recombination dominates, thus photon-recycling is useless from a practical way. For very high-quality GaAs with extremely low defects density, it is true that photon-recycling can help to improve the efficiency, but only a bit.

nice talk, I do learn a lot!

http://science.sciencemag.org/content/351/6280/1430

Recent paper in Science studied the "Photon recycling in lead iodide perovskite solar cells". But I am not sure it is really useful to improve solar cell efficiency. I hope to listen to your views~

Dear Wei,

The phenomenon raised in Science deserves dissension but unfortunately i could not see the full text. I am interested in such effect that the authors observed photo emission  at a distance of 50 um from the the photo luminescence spot.. Since the charge carriers can not reach this distance, then it most probable results from recycling of a previously emitted photon.

For more insight i need to see the full paper. 

Thank you for exciting this issue.

Dear Abdelhalim:

I have sent it to you by internal email of researchgate.

Thank you Wei!

I went through the paper of photon recycling in the pervoskite solar cell. The authors elaborated on the the photon recycling in perovskite solar cells and demonstrated its occurrence experimentally and theoretically. It is so that there is an appreciable recycling because of the high absorption coefficient ate the the band edge of the perovskite and the confinement of the the charge carriers and the photons in the thin perovskite layer. So, one can say the structure is a good solar cell and at the same time good photon emitter. This is in agreement with my conclusion in the first post where i wrote:Simply said the recycling can be affected when the material can work as an effective LED. 

And i add,recycling raises directly the output power by the amount which the photons converted to electrical energy.

It was a very fruitful discussion which was really very useful to me.

Best wishes

In fact, I found at the band edge, the absorption coefficient of CH3NH3PBI3 is sufficiently small compared to GaAs (from 750-950 nm), and thus the non-radiative current loss by SRH recombination will be much larger than the radiative current. I mentioned this point in my APL paper (http://arxiv.org/abs/1506.09003). We know that the black body spectrum of 300 K has the peak around 10 um. Hence, only infrared absorption will make a significant contribution to dark current. I do this calculation and comparison. Note that the authors used log scale for the absorption coefficient in the Science paper.

This is the reason why I found something is strange in the work. Moreover, the structure in the Science paper is "lateral" structure not traditional solar cell structure.

I attached the imaginary part of refractive index (log scale) for two materials. Another figure is from other paper. GaAs has a much larger long wavelength absorption than CH3NH3PBI3.

Dear Wei,

I agree with you that the experimental structure of the Science paper is different from the conventional one. Also, they did not estimated how much the photon recycling could raise the conversion efficiency compared to negligible recycling. Even the opencircuit voltage of their structure at the standard illumination reached only about .5 V which is largely lower than that pf the reported for the conventional perovsite structure. In addition, when they demonstrated the variation of the photocurent outside the illuminated spot, they used logarithmic scale. The picture will change if the photcurrent distribution is represented on a linear scale.

However, they demonstrated clearly the photo recycling in the pervosite solar cells  and proved its occurrence. 

I would expect the occurrence of photo recycling in the conventional structure since the active pervoskite material is still thin and extended between to optical reflecting interfaces. Therefore the confinement of charge carriers and photons is prevailing for large distances around the generation point of the photons. That is any generated photon will propagate in the pervoskite layer till it recycled even if its absorption coefficient is relatively smaller since the available distances is sufficient. We have to take into consideration that the emitted photons have a relatively wide spectrum not only that corresponding to the band edge because of the thermal energy of the recombining electrons and holes. The emission pea is shifted from the that of the band edge. 

Yes, one has to estimate the contribution of recycling for practical perovskite solar cells. I agree with you in this point.

Best wishes

Thanks for your discussion. I may write a paper to draw more attentions to this issue, The urbach tail absorption will be included. 

"We have to take into consideration that the emitted photons have a relatively wide spectrum not only that corresponding to the band edge because of the thermal energy of the recombining electrons and holes. The emission peak is shifted from the that of the band edge. " This is indeed a good insight, Abdelhalim!

Last day, I calculated the dark current at Voc [J0*exp(qVoc/kT)] by detailed balance theory. I found the dark current is quite small compared to nonradiative SRH current loss, when Voc=1.2 V that goes beyond the state-of-art result of 1.1 V for CH3NH3PBI3. So I still suspect why photon recycling can be supported or observed in the Science paper. In my calculation, I believed it is rigorous, the urbach tail absorption has been included. For GaAs, using Eli Yablonovitch paper parameters, the dark current at Voc is comparable to nonradiative Auger current loss. I will consider to write a simple eLetters to Science.

Recently a new paper about photon recycling efficiency in perovskite single crystals has appeared in Nature Com.

https://www.nature.com/articles/ncomms14417?WT.feed_name=subjects_techniques-and-instrumentation

It seems that photon recycling of single crystal is different than polycrystal of perovskite.

What do you think about this paper?