(262b) Reviving Pyrite FeS2 as a Photovoltaic Material

Pyrite FeS₂ has long been considered an ideal semiconductor for low-cost, sustainable solar cells because it is composed of earth-abundant, non-toxic, inexpensive elements, has a suitable band gap (0.95 eV), and absorbs light so strongly that a 100-nm-thick film can absorb >90 % of incoming sunlight. Moreover, minority charge carrier diffusion lengths (100-1000 nm) exceed the film thickness required for complete light absorption, a significant advantage for efficient collection of photogenerated carriers. For these reasons, pyrite FeS₂ was pursued vigorously in the 1980’s as a potential absorber in thin film solar cells. All attempts failed, however, producing solar cells with disappointing power conversion efficiencies, less than 3 % and a factor of ten below the theoretical maximum. With the rise of other thin film photovoltaic materials, such as CdTe and CuInGaSe₂, with higher power conversion efficiencies, enthusiasm for pyrite FeS₂ vanished. Interest in FeS₂ reemerged around 2009, motivated in part by the sustainability, cost, and toxicity concerns with commercialized thin film solar cells based on CdTe and CuInGaSe₂. This time, however, a few groups are pursuing the fundamental origins of the disappointing performance of FeS₂ rather than attempting to produce efficient cells via the trial-and-error approach that previously failed. Three issues have been identified: the possible presence of secondary phases and their deleterious effects, anomalous surface conduction, and a lack of understanding and control of doping. All of these issues are now yielding to rigorous research. First, concerns about secondary phases have largely abated, with well-characterized phase-pure single crystals and thin films synthesized. Second, significant progress has been made in understanding the anomalous conduction at the FeS₂ surface. While the physical reasons for this are not fully understood yet, pyrite FeS₂ single crystals have been shown to exhibit surprisingly diverse surface conduction, with surface conductivity varying by more than eight orders-of-magnitude, and, more remarkably, both p-and n-type behavior have been observed while the bulk is always n-type. This sheer diversity of surface behavior hints that eventual mitigation of these surface effects may be achievable. Third, and most importantly, defects and doping in pyrite FeS₂ have also started to yield to understanding. For three decades, electronic transport data from thin FeS₂ films were interpreted as p-type, while single crystals have been unambiguously established as n-type. Recently, we resolved this discrepancy and showed that rigorously phase-pure pyrite FeS₂ films are in fact n-type but in thin films with mobility below 1 cm²V^-1s^-1 hopping conduction artificially inverts Hall and Seebeck coefficients, incorrectly indicating p-type conduction. In single crystals and high quality thin films with mobilities over 1 cm²V^-1s^-1, these artifacts caused by hopping conduction are absent and electronic transport measurements reveal that FeS₂ pyrite thin films are in fact n-type. These conclusions were reached by making temperature-dependent resistivity, Hall effect, and thermopower measurements on over one hundred single crystals and thin films. Importantly, during this investigation we discovered a universal dependence of electron mobilities in both thin films and crystals on Hall coefficient, suggesting that a common dopant is responsible for this unintentional n-type doping. Most recently, we have amassed the strongest evidence to date that this common dopant is the sulfur vacancy. Significantly, we achieved control over sulfur vacancy concentrations and electron densities by varying the sulfur vapor pressure during crystal growth via chemical vapor transport.

Work supported by the Xcel Energy Renewables Development Fund and the University of Minnesota NSF MRSEC under DMR-1420013.