Intermediate band photovoltaics hold the promise of being highly efficient and cost effective photovoltaic cells. Intermediate states in the band gap, however, are known to facilitate nonradiative recombination. Much effort has been dedicated to producing metallic intermediate bands in hopes of producing lifetime recovery -an increase in carrier lifetime as doping levels increase. We show that lifetime recovery induced by the insulator-to-metal transition will not occur, because the metallic extended states will be localised by phonons during the recombination process. Only trivial forms of lifetime recovery, e.g., from an overall shift in intermediate levels, are possible. Future work in intermediate band photovoltaics must focus on optimizing subgap optical absorption and minimizing recombination, but not via lifetime recovery.The development of novel highly-efficient photovoltaic (PV) devices has the potential to significantly address the global energy and carbon problems. The vast majority of commercial solar cells are made from singlejunction semiconductors, an architecture which Shockley and Queisser showed has an absolute efficiency limit of 41% (with concentrated sunlight) [1]. Among the proposals to break this limit is the intermediate band (IB) photovoltaic, which has an efficiency limit of 63%, considerably higher than the single-junction limit [2,3].A standard single-junction semiconductor PV must optimise its band gap to maximise the product of current and voltage; these compete because it can absorb only photons with energy greater than the band gap E g , and the supplied voltage can be no larger than E g /e, where −e is the electron charge. An IBPV device, illustrated in Fig. 1, has an extra set of levels inside the semiconductor band gap; two subgap photons can be absorbed by the IB layer, producing a single electron-hole pair. Electrical contact is made only to the standard n-and p-type layers, so the IB layer produces extra current while allowing the full band gap to set the limit on the voltage, giving the considerably elevated efficiency bound of 63% [2]. The IBPV effect has been demonstrated in a number of systems [4][5][6], though it has not yet produced high efficiency cells.One method for making a material with an IB is to dope a semiconductor with large concentrations of dopants that form donor (or acceptor) levels deep inside the band gap. There is an obvious problem with this recipe: levels deep in the band gap are well known to cause nonradiative recombination [7,8]. It was proposed that a sufficiently high concentration of dopants could cause an insulator to metal transition (IMT) in the IB, which would suppress the nonradiative recombination rate and cause lifetime recovery in which adding additional dopants decreases the nonradiative recombination rate [9]. A great deal of work has gone into looking for such an IMT in doped semiconductors [10][11][12][13][14][15], including a report of lifetime recovery [16]. A b s o r b e d i n h o s t A b s o r b e d i n I B IB n p E g CB VB V μ e...