# Do III-V based photovoltaics “glow” (photo-luminesce) when illuminated but not loaded?

In this nice answer I started to learn about how the silicon photovoltaics cells on the International Space Station are managed. In that case, the cells are generally either delivering useful power to the station, or shunted to a dummy load. However, some satellites may leave some fraction of their photovoltaics open circuit if there is low power demand. In that case, the electron-hole pairs produced by sunlight are left to recombine within the junction itself.

Does this mean that photovoltaic cells made from III-V compounds with direct band gaps would produce measurable amounts of photoluminesence under solar illumination, at a wavelength roughly corresponding to the local band gap - there might be heterostructures so the wavelength may not correspond to bulk material band gap.

InP (in the past), InGaAs and GaAs based semiconductor junctions are often used. Many high power telecommunications satellites in geostationary orbits use multi-junction cells that are made from GaAs based semiconductor materials. If I pointed a (near IR capable) telescope at one (for example using this clever technique), and used a filter, could I detect this luminescence?

A lot of exotic processes in devices that include semiconductor junctions can make tiny amounts of light - I'm interested in the main, strong radiative recombination in direct band gap materials with quantum efficiencies of say 1% or greater. Thus the "glow" in the title.

• You could also ask if a forward biased solar cell can be used as an LED? I was not aware that it can, unlike an LED, which can also be used as a (poorly performing) solar cell, however, there seems to be a little bit of a story here, see e.g. "PROGRESS IN SILICON SOLAR CELL CHARACTERIZATION WITH INFRARED IMAGING METHODS" by Kasemann et al. . There have been reports in the past that even regular diodes in glass packages, in conduction mode, will generate some amount of IR radiation which can produce optical crosstalk. – CuriousOne Jul 27 '16 at 1:32
• GaAs transistors light up nicely in operation. Yes, they will make light under solar irradiation. Really, why would one think they wouldn't? – Jon Custer Jul 27 '16 at 2:08
• @CuriousOne wow! Were the diodes in glass indirect band gap (Si, Ge) or some direct band gap material? While silicon can make a little light sometimes, it is usually extremely weak, with very low quantum efficiency. The tiny light from 10V of pulsed reverse bias in that article is not something that happens under normal operation. Here I am interested in unloaded and presumably not externally biased PV cells under full spectrum solar illumination. – uhoh Jul 27 '16 at 2:10
• Yes, the lore was/is that even Si and Ge parts have a very faint "glow", but I don't have any experimental evidence that this is true. Personally I always doubted it. If Jon Custer says that he has seen it on GaAs parts where it's more likely, then I don't doubt it. Makes me almost want to peel a package and see if I can reproduce it. :-) I have certainly used LEDs as photodetectors. They are marginal but workable, but then, that's a no-brainer. – CuriousOne Jul 27 '16 at 3:07
• @uhoh That looks good to me :-) – Tristan Jul 28 '16 at 13:53

III-V photovoltaics will indeed photoluminesce quite nicely (particularly if not loaded). As you note, they are direct band gap materials so if there is any overlap of electrons and holes you can get recombination with the accompanying photons.

If connected to an external circuit, the electron/hole pairs generated will get swept out of the junction fairly quickly, reducing the possibility of recombination. You will still have recombination in the 'bulk' contact areas where the e-h pairs are 'wasted'. In addition, depending on the junction design, you can actually get quite a bit of minority carrier diffusion (not drift) out of the junction in to less hospitable areas.

If not connected to a circuit, you will end up having all the e-h pairs recombine within the material (they aren't going anywhere else), so it is just like doing photoluminescence on bulk material.

Photoluminescence of silicon solar cells, particularly at the wafer level, has been done regularly for the development of high-efficiency material (for, e.g., space applications where that last 0.5% counts). Since silicon is an indirect gap material with nominally long carrier lifetimes, the intensity of the phonon-assisted optical transition is a good indicator of the (non-radiative) defect levels in the material.

As for III-V devices, they all luminesce quite nicely in normal operation. A forward-biased junction results in electrons and holes flowing in from opposite directions to annihilate each other, releasing lots of photons. My lab does radiation effects on devices, and the die have to be bare (unlidded) to hit them with an ion beam. The targeting camera shows the devices lit up brightly. The first time I saw it I could not help a 'Cool - look at that!' coming out of my mouth, at which time the staff member running the experiment commented in an even voice 'GaAs is a direct gap material, Jon'. It is one of those things that you kind of don't directly connect in device physics courses...

• Thanks very much! Science is definitely cool, and anyone who acts otherwise is just bluffing - especially so when things light up! I didn't realize the PV luminescence was a straightforward (though not simple) mechanism in Si PV. I thought it was something exotic and poorly understood - so you've cleared up a number of things for me with this answer. – uhoh Jul 27 '16 at 14:35
• @uhoh - you should look around for a copy of Pankove's 'Optical Processes in Semiconductors', which was Dover reprint 30+ years ago (and still is according to Amazon - only $15 new - my 1975 copy has$7.50 on the cover!). It is very good on, well, all the cool optical measurements you can do on semiconductors, and quite clear and readable. – Jon Custer Jul 27 '16 at 14:45
• – uhoh Jan 20 '19 at 12:55