Several possible explanations have been suggested for the low V OC , including bulk non-stoichiometry, [7,[15][16][17] near-surface nonstoichiometry (resulting in a metallic FeS-like surface layer), [18] sulfur vacancies that generate electronic states in the band gap, [13] Fermi level pinning induced by surface states, [19] or small band gap phases (pyrrhotite, troilite, and marcasite) present as domains in bulk pyrite. [8] Orthorhombic marcasite (FeS 2 ) and hexagonal troilite (FeS) are believed to be detrimental phases for photochemical performance, as they have much smaller band gaps (0.04 eV for troilite and 0.34 eV for marcasite), [20] and it has been suggested that even trace amounts of these phases would cause the low V OC reported for pyrite films. [21] The band gap of marcasite was first determined from temperaturedependent (53-370 K) electrical resistivity measurements and a value of 0.34 eV was obtained. [20] However, these measurements were based on the assumption that the carrier mobility is dominated by lattice scattering, [20,22] and the band gap value of marcasite has been rarely verified by other methods, such as optical measurements. Furthermore, several theoretical calculations published in recent years predict that marcasite should have a band gap of 0.8-1.0 eV, which is quite similar to that of pyrite; Gudelli even observed that marcasite has a much larger band gap than pyrite (1.603 eV for marcasite and 1.186 eV for pyrite). [23,24] Very recently, the optical band gap energy of marcasite has been determined by diffuse reflectance spectroscopy to be 0.83 ± 0.02 eV, and the optical absorption of marcasite and pyrite in the near infrared-visible (NIR-Vis) region appears to be quite similar. [25] Therefore, the presence of marcasite is unlikely to undermine the photovoltaic performance of pyrite and it is therefore worth considering whether significant effort should actually be expended on eliminating marcasite traces from pyrite preparations. As the formation of junctions (such as p-n junctions or phase junctions) can efficiently promote charge separation in semiconductor-based photocatalysts, the fabrication of proper junctions in semiconductors is highly desirable in the design and preparation of efficient semiconductor-based photocatalysts. The most conspicuous example is the activity of TiO 2 (P-25, Degussa), which consists of anatase and rutile (4:1 w/w) and exceeds the photocatalytic activity of pure anatase in many reaction systems. [26][27][28][29][30][31][32] Furthermore, Li and co-workers have reported enhanced photocatalytic performance of Ga 2 O 3 with tunable α-β phase junctions. The drastically enhanced activity of mixed α-and β-Ga 2 O 3 over the phase pure oxides was ascribed to efficient charge separation and transfer across the α-β phase junctions. [33][34][35] As to the pyrite-marcasite interface, although most published work has attributed the low performance of pyrite films to the minor presence of marcasite, noThe interest in iron pyrite (cubic FeS 2 ) as a photovoltaic m...
