Design principles for maximizing photovoltage in metal-oxide-protected water-splitting photoanodes

Journal name:
Nature Materials
Year published:
Published online


Metal oxide protection layers for photoanodes may enable the development of large-scale solar fuel and solar chemical synthesis, but the poor photovoltages often reported so far will severely limit their performance. Here we report a novel observation of photovoltage loss associated with a charge extraction barrier imposed by the protection layer, and, by eliminating it, achieve photovoltages as high as 630mV, the maximum reported so far for water-splitting silicon photoanodes. The loss mechanism is systematically probed in metal–insulator–semiconductor 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 leaky-capacitor model related to the dielectric properties of the protective oxide explains this loss, achieving excellent agreement with the data. From these findings, we formulate design principles for simultaneous optimization of built-in field, interface quality, and hole extraction to maximize the photovoltage of oxide-protected water-splitting anodes.

At a glance


  1. Three junction types showing the extraction barrier imposed by the protection layer illustrated with silicon water-oxidation photoanodes.
    Figure 1: Three junction types showing the extraction barrier imposed by the protection layer illustrated with silicon water-oxidation photoanodes.

    Three types of photoanode junctions have been employed in the literature, the Type 0 semiconductor–liquid (SL), Type I MIS Schottky, and Type II MIS p+n junction. Here, each junction configuration is shown for a silicon cell protected by TiO2. The insulators introduce a series resistance, which when located inside the junction constitutes an extraction barrier. Therefore, the density of holes at the interface and the density of states of the contact play a key role. In Type 0 cells, holes will be accumulated at the semiconductor/insulator interface during current flow, and charge transfer may still be significantly limited owing to a low density of states in the contact. Such a situation may arise from using electrolytes as well as non-metallic catalysts as the hole conduction mediator. In Type 1 cells, moderate hole accumulation is sufficient to efficiently extract minority carriers, but the photovoltage suffers an insulator-thickness-dependent loss. In Type 2 cells, the p+ region ensures a high hole concentration at the interface that is independent of illumination and bias, resulting in no barrier to hole extraction. As a result, these cells exhibit no dependence of photovoltage on the insulator thickness and achieve record photovoltages at all oxide thicknesses and pH values studied.

  2. Photovoltage loss in ALD-TiO2-protected nSi MIS anodes.
    Figure 2: Photovoltage loss in ALD-TiO2-protected nSi MIS anodes.

    a, Water oxidation overpotential at 1mAcm−2 current density—that is, the potential with respect to the thermodynamic potential E0(H2O/O2), at pH values of 14, 7 and 0. Device results are plotted as a function of TiO2 thickness in Ir/TiO2/SiO2/p+Si anodes splitting water in the dark, compared to Ir/TiO2/SiO2/nSi photoanodes under 1 sun illumination. b,c, The trends converge, showing a decreasing photovoltage that starts at a value >500mV (b) and ends with negative apparent photovoltages (c). Panels b and c are both for water oxidation in pH = 0 acid solution, where i is the current density and the voltage is reported with reference to the normal hydrogen electrode (NHE), thus the thermodynamic water oxidation potential is 1.229V.

  3. Photovoltage loss observed with increasing thickness of the SiO2 interlayer with and without TiO2.
    Figure 3: Photovoltage loss observed with increasing thickness of the SiO2 interlayer with and without TiO2.

    The photovoltage decreases with thickness of the SiO2 both with (blue) and without (orange) protective TiO2. Anodes with less than 2nm of SiO2 and no protective TiO2(hollow orange markers) are highly unstable and the results are subject to error. It is possible that these devices have already degraded, leading to an underestimation of the photovoltage loss with respect to thickness for Ir/SiO2/nSi photoanodes (see Supplementary Fig. 4). In all cases, the photovoltage decreases rapidly with SiO2 interlayer thickness and, similar to the TiO2 case, with a linear dependence on thickness, although the slope is much greater.

  4. Simulated band diagrams reveal differences between Type I Schottky junction structures and Type 2 p+n buried junction structures.
    Figure 4: Simulated band diagrams reveal differences between Type I Schottky junction structures and Type 2 p+n buried junction structures.

    a,b, Band diagrams are simulated under 1 sun of AM 1.5G illumination and device operation at 1mAcm−2 for the Type 1 MIS Schottky (a) and the Type 2 buried junction case (b). EC and EV denote the conduction band and valence band. EFn and EFp denote the quasi-electron and quasi-hole Fermi levels under illumination. The high-work-function metal creates a thin inversion layer of holes in the Type 1 nSi structure, reaching 3 × 1017cm−3 at the interface, but it is insufficient to maintain a high hole density at the interface, regardless of operating conditions, compared with a 5 × 1019cm−3 implanted p+ region. The optical generation is shown here in arbitrary units. Full details and the integrated photogeneration can be found in Supplementary Section 7.

  5. Charge extraction photovoltage loss eliminated and record photovoltages achieved with Type 2 p+n buried junction cells.
    Figure 5: Charge extraction photovoltage loss eliminated and record photovoltages achieved with Type 2 p+n buried junction cells.

    ac, Electrochemical performance of Ir/ALD-TiO2/SiO2/Si anode of varying TiO2 thickness (for example, 20c = 20 TDMAT/H2O ALD cycles = 1.2nm) in ferro/ferrocyanide (FFC) for a p+Si reference substrate (a), nSi Type 1 MIS junction under 1 sun illumination (b) and p+nSi Type 2 buried junction under 1 sun (c). A large, asymmetric loss (asymmetric stretching of the CV) is observed for Type 1 nSi MIS structures, compared with the p+Si reference, which is eliminated in the Type 2 p+nSi structure. TDMAT, tetrakis(dimethylamido)titanium. d, This translates to a near-constant photovoltage for Type 2 cells (red squares) compared with steep losses when increasing TiO2 (green squares) or SiO2 (blue squares) in the Type 1 cell. e, For water oxidation, the same effect is observed, where the Type 2 p+nSi leads to a constant overpotential shift with respect to the p+Si (dark) reference, corresponding to record photovoltages of 550mV to >600mV at all pH values. In the Type 1 nSi cell, the inferred photovoltage goes from positive (green double-headed arrow) 500mV to negative (red double-headed arrow) 200mV. f, Water-oxidation cyclic voltammetry for a 4nm TiO2 layer showing an ~100mV photovoltage shift in the nSi MIS compared with an ~600mV shift for the equivalent p+nSi MIS.


  1. Fujishima, A. & Honda, K. Electrochemical photolysis of water at a semiconductor electrode. Nature 238, 3738 (1972).
  2. Heller, A. Conversion of sunlight into electrical power and photoassisted electrolysis of water in photoelectrochemical cells. Acc. Chem. Res. 14, 154162 (1981).
  3. Bard, A. J. & Fox, M. A. Artificial photosynthesis: Solar splitting of water to hydrogen and oxygen. Acc. Chem. Res. 28, 141145 (1995).
  4. Turner, J. A. A realizable renewable energy future. Science 285, 687689 (1999).
  5. Grätzel, M. Photoelectrochemical cells. Nature 414, 338344 (2001).
  6. Lewis, N. S. Light work with water. Nature 414, 589590 (2001).
  7. Lewis, N. S. & Nocera, D. G. Powering the planet: Chemical challenges in solar energy utilization. Proc. Natl Acad. Sci. USA 103, 1572915735 (2006).
  8. Bassett, R. K. To the Digital Age: Research Labs, Startup Companies, and the Rise of MOS Technology (John Hopkins Univ. Press, 2007).
  9. Chen, Y. W. et al. Atomic layer-deposited tunnel oxide stabilizes silicon photoanodes for water oxidation. Nature Mater. 10, 539544 (2011).
  10. Scheuermann, A. G., Prange, J. D., Gunji, M., Chidsey, C. E. D. & McIntyre, P. C. Effects of catalyst material and atomic layer deposited TiO2 oxide thickness on the water oxidation performance of metal–insulator–silicon anodes. Energy Environ. Sci. 6, 24872496 (2013).
  11. Hu, S. et al. Amorphous TiO2 coatings stabilize Si, GaAs, and GaP photoanodes for efficient water oxidation. Science 344, 10051009 (2014).
  12. Lichterman, M. F. et al. Stabilization of n-cadmium telluride photoanodes for water oxidation to O2(g) in aqueous alkaline electrolytes using amorphous TiO2 films formed by atomic-layer deposition. Energy Environ. Sci. 7, 33343337 (2014).
  13. McDowell, M. T. et al. Improved stability of polycrystalline bismuth vanadate photoanodes by use of dual-layer thin TiO2/Ni coatings. J. Phys. Chem. C 118, 1961819624 (2014).
  14. 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, 671678 (1984).
  15. Green, M. A., Emery, K., Hishikawa, Y., Warta, W. & Dunlop, E. D. Solar cell efficiency tables (version 45). Prog. Photovolt. 23, 19 (2015).
  16. Hu, S. et al. An analysis of the optimal band gaps of light absorbers in integrated tandem photoelectrochemical water-splitting systems. Energy Environ. Sci. 6, 29842993 (2013).
  17. Kenney, M. J. et al. High-performance silicon photoanodes passivated with ultrathin nickel films for water oxidation. Science 342, 836840 (2013).
  18. Mei, B. et al. Protection of p+nSi photoanodes by sputter-deposited Ir/IrOx thin films. J. Phys. Chem. Lett. 5, 19481952 (2014).
  19. Mei, B. et al. Iron-treated NiO as a highly transparent p-type protection layer for efficient Si-based photoanodes. J. Phys. Chem. Lett. 5, 34563461 (2014).
  20. Green, M. A. & Godfrey, R. B. MIS solar cell—General theory and new experimental results for silicon. Appl. Phys. Lett. 29, 610612 (1976).
  21. Green, M. A. Effects of pinholes, oxide traps, and surface states on MIS solar cells. Appl. Phys. Lett. 33, 178180 (1978).
  22. Godfrey, R. B. & Green, M. A. 655mV open-circuit voltage, 17.6% efficient silicon MIS solar cells. Appl. Phys. Lett. 34, 790793 (1979).
  23. Singh, R., Green, M. A. & Rajkanan, K. Review of conductor–insulator–semiconductor (CIS) solar cells. Solar Cells 3, 95148 (1981).
  24. Green, M. A. & Blakers, A. W. Advantages of metal–insulator–semiconductor structures for silicon solar cells. Solar Cells 8, 316 (1983).
  25. Grauvogl, M. & Hezel, R. The truncated-pyramid MIS inversion-layer solar cell: A comprehensive analysis. Prog. Photovolt. Res. Appl. 6, 1524 (1998).
  26. McDowell, M. T. et al. The influence of structure and processing on the behavior of TiO2 protective layers for stabilization of n-Si/TiO2/Ni photoanodes for water oxidation. ACS Appl. Mater. Interfaces 7, 1518915199 (2015).
  27. Mei, B. et al. Crystalline TiO2: A generic and effective electron-conducting protection layer for photoanodes and -cathodes. J. Phys. Chem. C. 119, 1501915027 (2015).
  28. Levy, D., Ponpon, J. P., Grob, A., Grob, J. J. & Stuck, R. Rapid thermal annealing and titanium silicide formation. Appl. Phys. A 38, 2329 (1985).
  29. Mann, R. W., Clevenger, L. A., Agnello, P. D. & White, F. R. Silicides and local interconnections for high-performance VLSI applications. IBM J. Res. Dev. 39, 403417 (1995).
  30. Sekine, K., Saito, Y., Hirayama, M. & Ohmi, T. Silicon nitride film growth for advanced gate dielectric at low temperature employing high-density and low-energy ion bombardment. J. Vac. Sci. Technol. A 17, 31293133 (1999).
  31. Sugawara, T. et al. Characterization of ultra thin oxynitride formed by radical nitridation with slot plane antenna plasma. Jpn. J. Appl. Phys. 44, 12321236 (2005).
  32. Scheuermann, A. G., Lu, D. Q., Ito, T., Chidsey, C. E. D. & McIntyre, P. C. The effect of SPA-SiO2 tunnel oxide thickness for metal–insulator–silicon photoelectrochemical cells. ECS Trans. 64, 265276 (2014).
  33. Alamo, J. A. & Swanson, R. M. Modelling of minority-carrier transport in heavily doped silicon emitters. Solid-State Electron. 30, 11271136 (1987).
  34. Tyagi, M. S. & Van Overstraeten, R. Minority carrier recombination in heavily-doped silicon. Solid-State Electron. 26, 577597 (1983).
  35. Chen, L. et al. p-type transparent conducting oxide/n-type semiconductor heterojunctions for efficient and stable solar water oxidation. J. Am. Chem. Soc. 137, 95959603 (2015).
  36. Sze, S. M. Semiconductor Devices: Physics and Technology (John Wiley, 2008).
  37. Swaminathan, S. & McIntyre, P. C. Titania/alumina bilayer gate dielectrics for Ge MOS devices: Frequency- and temperature-dependent electrical characteristics. Electrochem. Solid-State Lett. 13, G79G82 (2010).
  38. Bard, A. J., Bocarsly, A. B., Fan, F. R. F., Walton, E. G. & Wrighton, M. S. The concept of Fermi level pinning at semiconductor/liquid junctions. Consequences for energy conversion efficiency and selection of useful solution redox couples in solar devices. J. Am. Chem. Soc. 102, 36713677 (1980).
  39. Scheuermann, A. G., Lawrence, J. P., Gunji, M., Chidsey, C. E. D. & McIntyre, P. C. ALD-TiO2 preparation and characterization for metal–insulator–silicon photoelectrochemical applications. ECS Trans. 58, 7586 (2013).
  40. Fuke, N. et al. Influence of TiO2/electrode interface on electron transport properties in back contact dye-sensitized solar cells. Solar Energy Mater. Solar Cells 93, 720724 (2009).
  41. Pattantyus-Abraham, A. G. et al. Depleted-heterojunction colloidal quantum dot solar cells. ACS Nano 4, 33743380 (2010).
  42. Devore, J. R. Refractive index of rutile and sphalerite. J. Opt. Soc. Am. 41, 416417 (1951).
  43. Haynes, W. M., Bruno, T. J. & Lide, D. R. (eds) CRC Handbook of Chemistry and Physics 12124 95th edn (CRC Press, 2014).

Download references

Author information


  1. Department of Materials Science and Engineering, Stanford University, Stanford, California 94305, USA

    • Andrew G. Scheuermann,
    • John P. Lawrence,
    • Kyle W. Kemp,
    • T. Ito &
    • Paul C. McIntyre
  2. Tokyo Electron Limited, Technology Development Center, 650, Hosaka-cho Mitsuzawa, Nirasaki, Yamanashi 407-0192, Japan

    • T. Ito
  3. Tyndall National Institute, University College Cork, Cork, Ireland

    • Adrian Walsh &
    • Paul K. Hurley
  4. Department of Chemistry, Stanford University, Stanford, California 94305, USA

    • Christopher E. D. Chidsey


A.G.S. and J.P.L. prepared all samples and performed all measurements for the initial TiO2 photovoltage series first observing the photovoltage loss. A.G.S. prepared all samples and performed all measurements for the SiO2 photovoltage series, capacitance voltage measurements, and buried junction p+nSi experiments. A.W. prepared all p+n buried junction substrates and performed all physical characterization of these junctions. K.W.K. performed Sentaurus modelling for light absorption in Type I and Type II cells, as well as simulating band diagrams. T.I. performed the SPA plasma oxidation for the experiments varying SiO2 thickness. A.G.S. and K.W.K. maintained the ALD chamber for TiO2 depositions and A.G.S. qualified and performed the runs. A.G.S., C.E.D.C. P.K.H. and P.C.M. designed the experiments and developed the solid-state capacitor model to explain the loss. All authors helped in the preparation of the manuscript.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Supplementary information

PDF files

  1. Supplementary Information (2,137 KB)

    Supplementary Information

Additional data