The Opto-Electronic Physics That Broke the Efficiency Record in Solar Cells

The solar cell field is changing. We are finally approaching the Shockley-Queisser (SQ) limit for single junction solar cell ~33.5% efficiency under the standard solar spectrum. Previously, the record had been stuck at 25.1%, during 1990-2007. Why then the 8% discrepancy between the theoretical limit 33.5% versus the previously achieved efficiency?

It is usual to blame material quality. But in the case of GaAs double heterostructures, the material is almost ideal with an internal fluorescence yield of >99%. This deepens the puzzle as to why the full theoretical SQ efficiency was not achieved?