In this work, we demonstrate that manganese oxide deposited via physical vapor deposition behaves as an effective barrier to the oxidation of silicon photo-anodes for water oxidation and water splitting applications while still generating in excess of 500 mV in photo-voltage under 100 mW/cm² illumination, with currents of up to 7 mA/cm² without an additional catalyst. The optimal thickness of MnOx which strikes a balance between photo-current and protection of the underlying silicon appears to be between 7 and 10 nm.  

X-ray photoelectron spectroscopy data has confirmed that a drop in photocurrent in cells depending on unprotected silicon as the photo-anode is due to anodic oxidation. An initial current of 1 mA/cm² at an applied potential of 1.2 V vs. an Ag/AgCl electrode in the light will grow 1.3 nm of sub stoichiometric silicon oxide within 400 seconds and the current density drops by a factor of 200 as a result. Even a 2 nm manganese oxide layer keeps the grown SiO_x thickness below 0.5 nm at the same conditions mentioned above. The surface chemistry is such that when Manganese oxide is deposited on native silicon oxide, it appears to form a thin layer of MnSiO­₃ in accord with previous studies, but this layer quickly dissociates during water oxidation. However, the remaining SiO_x/Mn-oxide is sufficiently stable to prevent oxygen transport from the electrolyte to the underlying silicon. XPS studies show that even a 2 nm manganese oxide layer keeps the grown SiO_x thickness below 0.5 nm at the same conditions mentioned above.

Our results also show that determination of the particular stoichiometry of manganese oxide which forms is difficult, and it appears that layers grown in a variety of conditions, including atomic oxidation in UHV, all tend towards the same mixed phase oxide upon atmospheric exposure. Attempting to improve the quality of the Si/Mn-oxide interface without invoking a phase change which we show dramatically affects conductivity is an attractive goal.

Finally we examine the role of the native silicon oxide, showing that that on hydrofluoric acid treated silicon substrates which strips the native oxide, we see an improved initial condition, but this comes at the expense of long term barrier reliability.

Our talk will also outline potential routes to understanding the degradation mechanisms which lead to anode failure in the presence of Mn-oxide including how the Mn-oxide layer degrades to allow oxygen to contact the silicon.