The core methodology of theoretical investigations are first-principles calculations based on density functional theory (DFT). These calculations yield accurate defect formation energies and the position of defect levels within the band gap of the material. The information about formation energies and defect levels is then used to narrow-down the list of candidate defects that could contribute to Shockley-Read-Hall recombination.
Regarding the formation energy the requirement is that this formation energy is sufficiently low so that the considered defects occur in appreciable concentrations. If the formation energy is high, such defects are not expected to form, and thus they cannot be at the origin of the experimentally observed Shockley-Read-Hall recombination. Regarding the defect levels, the conventional wisdom tells that these levels be position at approximately the mid-gap region so that both electrons and holes are captured efficiently. This makes the defect a detrimental center at which Shockley-Read-Hall recombination is very efficient. However, for defects in pure GaN (band-gap 3.5 eV) there are no obvious candidates that fulfil both of these requirements. However, there were strong experimental proof that non-radiative recombination can be very efficient both in InGaN alloys with smaller band gaps as well as GaN itself. This made this project very challenging and prompted us to search for alternative mechanisms that could explain Shockley-Read-Hall recombination.
In this project we identified two different mechanisms that makes specific defects detrimental non-radiative recombination centers in nitrides. Importantly, it was discovered that the behaviour of point defects in InGaN alloys cannot be understood solely from their behaviour in pure GaN. There is a class of defects that in GaN have their defect levels in the lower part of the band gap, enabling efficient hole capture, but inefficient electron capture. Thus, these defects do not contribute to Shockley-Read-Hall recombination in gallium nitride. However, when the band gap of the nitride material decreases due to alloying with indium (as required for light emission in the visible range of the spectrum), defect levels moves towards mid-gap and turn into efficient non-radiative centers. This behaviour was found for certain gallium vacancy complexes and some more unexpected defects such as calcium. We have also found out that some defects posses electronically excited states, enabling efficient carrier capture even if the position of defect levels would indicate otherwise. This discovery is the most important fundamental result of the entire project. In InGaN specific gallium vacancy complexes turn into efficient recombination centers because of this mechanism. Even more unexpectedly, we have discovered that this is precisely the mechanism that involves excited states that turns Fe impurities in GaN into extremely detrimental non-radiative centers. The same mechanism is proposed to be operational for many other transition metal impurities in wide band-gap materials.
As of April 2017 the results have been published in six peer-review articles (two more articles reporting results of the Marie Sklodowska-Curie project are in preparation). The results of the project, especially the work on gallium vacancy complexes and iron in GaN, received numerous press releases worldwide (most importantly, in a monthly magazines Compound Semiconductors and Photonics Spectra, as well as news web-sites such as Phys.Org EurekAlert, Newswise, etc). The results have been also presented in international scientific conferences, such as the March meeting of the American Physical Society, the International Conference on Defects in Semiconductors, the Gordon Research Conference on Defects in Semiconductors, the International Workshop on Nitride Semiconductors, the Psi-k 2015 conference.
