Search

Metal-insulator-semiconductor (MIS) structures are promising candidates for integrated solar driven water splitting devices. Often, the metal serves a dual purpose, catalyzing water oxidation while also setting the built-in field for extracting photogenerated carriers from the semiconductor. Recently, incorporation of additional protection layers into the MIS junction has made it possible to use semiconductor materials, such as silicon, that would normally be unstable under the conditions required for water oxidation.1 Although the theoretical maximum photovoltage for silicon photovoltaics is 700-800 mV,2 there is a tradeoff between photocurrent and photovoltage in adventitiously-doped TiO2.3 For example, maximum photovoltages of ~400 mV have been reported for nSi / TiO2 / metal photoanodes protected by highly conductive TiO2 layers,4 with some devices achieving only 200-250 mV.5 Our goal is to develop a more ideal protection layer using transition metal oxide-titania alloys synthesized by ALD. An ideal protection layer should be corrosion resistant, highly conductive, and enable a high photovoltage. For n-type silicon Schottky photoanodes, this means that the protection layer must also possess a high work function. We previously demonstrated that TiO2-RuO2 alloys composed of 13-46% Ru were highly conductive and consistently achieved photovoltages >525 mV without any post-processing anneals.6 The built-in field was set by the TiO2-RuO2 alloy, which possessed a sufficiently high work function of 5.02 eV. Although TiO2-RuO2 alloys were ideal Schottky contacts to nSi, they lacked the long-term corrosion resistance required of a protection layer. RuO2 is known to be unstable at the conditions required for water oxidation. We also report on ALD alloys of TiO2 and IrO2, which exhibit high conductivity and photovoltage > 600 mV on nSi, but with improved stability. Additionally, TiO2-IrO2 alloys are capable of catalyzing oxygen evolution. Thus, TiO2-IrO2alloys have the potential to be an “all-in-one” catalyst, Schottky contact, and corrosion resistant protection layer. References: 1. Chen, Y.C. et al. Atomic Layer-Deposited Tunnel Oxide Stabilizes Silicon Photoanodes for Water Oxidation. Nature Mater. 10, 539-44 (2011). 2. Green, M. A. Limits on the Open-circuit Voltage and Efficiency of Silicon Solar Cells Imposed by Intrinsic Auger Processes. IEEE Trans. Electron Devices ED-31,671–678 (1984). 3. Scheuermann, A.G. et al. Photovoltage Design for Metal Oxide Protected Solar-Water-Splitting Photoanodes. Nature Mater. 15, 99-105 (2016). 4. Hu, S. et al. Amorphous TiO2 coatings stabilize Si, GaAs, and GaP photoanodes for efficient water oxidation. Science (80-. ). 344,1005–1009 (2014). 5. McDowell, M. T. et al. The Influence of Structure and Processing on the Behavior of TiO 2 Protective Layers for Stabilization of n-Si/TiO 2 /Ni Photoanodes for Water Oxidation. ACS Appl. Mater. Interfaces 7,15189–15199 (2015). 6. Hendricks, O. L. et al. Isolating the Photovoltaic Junction : Atomic Layer Deposited TiO2 − RuO2 Alloy Schottky Contacts for Silicon Photoanodes. (2016). doi:10.1021/acsami.6b08558

Metal oxide protection layers for photoanodes may enable the development of large-scale solar fuel and solar chemical synthesis. ALD-TiO2 remains the most widely used material because of its excellent stability under water oxidation conditions and potential for high electrical conductivity both as an ultrathin film and with thicknesses exceeding 100 nm [1-3]. However, the most conductive ALD-TiO2films exhibit poor photovoltages of ~ 400 mV and less [3]. Even assuming near-ideal fill factors, these voltages fall far short of the values needed for viable water splitting devices. Photovoltage optimization is especially difficult in MOS photoanodes due to the range of factors which influence the conductivity and stability of the protection layer as well as the presence of electrically active defect states at the semiconductor/oxide interface. Therefore, understanding how to optimize photovoltage and stability is of utmost concern for the advancement of the field. Here we report a novel observation of photovoltage loss associated with charge transfer in these metal-oxide protected devices, and by eliminating it, achieve photovoltages as high as 630 mV, the maximum reported to date for single-junction water-splitting silicon cells. The loss mechanism is systematically probed in MOS Schottky junction cells compared to buried junction p+n cells, revealing the need to maintain a characteristic hole density at the semiconductor/insulator interface. A capacitor model that predicts this loss is developed, and is related to the dielectric properties of the protective oxide, achieving excellent agreement with the data. From these findings, we extract design principles for simultaneous optimization of charge transfer resistance and interface quality to maximize the photovoltage of metal-oxide protected MOS water splitting devices. [1] Y.W. Chen, et al. Nature Mat. 2011, 10, 539-544. [2] A. G. Scheuermann, et al. Energy Environ. Sci. 2013, 6, 2487–2496. [3] S. Hu, et al. Science 2014, 344, 1005−1009. Figure 1 | Charge transfer in three cell types for water splitting applied to silicon: The Type 0 Semiconductor-Liquid (SL), Type 1 Metal-Insulator-Semicondcutor (MIS), and Type 2 p+n junction. The density of states on either side of the oxides and the excition splitting position with respect to these layers play a crucial role in mediating efficient charge transfer. These effects are so strong that Type 0 protected silicon cells exhibit essentially no photovoltage, Type 1 nSi cells show a linear photovoltage loss with oxide thickness, and Type 2 cells--where the hole concentration on the Si/SiO2 interface is always high--exhibit record photovoltages at all oxide thicknesses and pH values studied. Figure 1

Photoelectrochemical cells are becoming increasingly attractive for the on-site production of clean chemicals and fuels, both as end-use products and as an integrated energy storage solution for mitigating the intermittency of solar radiation. One persistent challenge in this field is the difficulty of simultaneously achieving high photovoltaic efficiency and chemical stability of semiconductor photoelectrodes during water oxidation under wide-ranging pH conditions. Wide band gap semiconducting metal oxides are frequently stable under water oxidation conditions but have inherent photocurrent limitations. We have recently demonstrated that atomic layer deposition (ALD) of a thin and chemically-protective TiO2 layer can avoid oxidation of the surface of a silicon photoanode during long-duration oxygen evolution. Our initial work showed that it was possible to decouple efficient light absorption by the silicon anode from the oxygen evolution reaction (OER) occurring on the surface of a water oxidation catalyst layer by the presence of an interposed ALD-TiO2 layer. This conformal and pinhole-free oxide layer was found to be dense enough to block the solution from oxidizing the underlying semiconductor (greatly increasing its stability), but still thin enough to allow efficient hole transport between an n-type silicon substrate and the iridium OER catalyst [1]. In more recent work, we have showed that ALD-TiO2 could be used with a variety of other catalysts to produce efficient devices with more than 600mV of open circuit voltage, high saturation currents, and similar stability. In addition, by studying thicker ALD-TiO2 films, we found that protection could be increased with a minor increase in anode resistance. The additional resistance is associated with bulk conduction of electronic carriers through states in the TiO2 band-gap between the catalyst and anode [2]. Together, these findings indicate that atomic layer deposited coatings in metal-insulator-semiconductor (MIS) photoanodes serve as a general protection method for a variety of semiconductor-catalyst systems. In addition to summarizing prior reports and the most recent results on such MIS photoelectrodes, this presentation will touch on future directions for efficient artificial photosynthesis using ALD surface protection layers. [1] YW Chen, J.D. Prange, et al. Nat. Mater. 2011, 10, 539-44. [2] A.G. Scheuermann, et al. Energy Environ. Sci. 2013, 6, 2487 – 96.