Recently, TiO₂ received renewed attention in the field of photoelectrochemical water splitting as corrosion protection layer for unstable, small-bandgap photoelectrodes.^[1–4] Two approaches are being pursued for protecting photoanodes with TiO₂: (1) using ultra-thin films to enable hole-transport via a tunnelling mechanism or (2) preparing TiO₂ with hole-conducting properties. In particular the hole-conducting properties of TiO₂ (also referred to as ‘leaky’ TiO₂) proved key towards a stable PEC device.^[4]

For the thin-film approach, it is of particular importance to study the nucleation behaviour on specific substrates. Although the ALD process of TiO₂ is one of the most widely studied deposition processes, only a few papers report on the initial growth.^[5,6] Here, we studied the nucleation mechanism of ALD-grown TiO₂ on silicon with a 2 nm native surface oxide (which should not affect the charge transport^[3]).

The film deposition and analysis are carried out in our in-line ALD/XPS reactor that allows for XPS measurements after each precursor pulse, without breaking the vacuum. This enables us to analyze the elemental composition and chemical state of each species during nucleation and subsequent film growth. While our TiO₂ process exhibited no nucleation delay, we found that especially during the nucleation phase longer precursor exposure times are necessary to reach saturated growth than the exposure times typically found from growth-per-cycle saturation curves. Furthermore, prolonged water exposure times were found to reduce the amount of Cl impurities built in at the interface.

The resulting films did not provide significant protection for the Si anodes, probably due to the presence of pinholes. In addition, the thicker films (~10 nm) were found not to be hole-conducting and are therefore not suited to protect photoanodes. The origin for the hole-conducting properties of TiO₂ is not yet fully understood. A likely cause are defect states present in the bandgap formed by the presence of Ti³⁺ species^[7,8], presumably in the form of interstitial ions. We found that we can control the amount of incorporated Ti³⁺ with the deposition temperature and the purge time after the Ti-precursor step. The implications of these findings for the fabrication of TiO₂ protection layers on photoelectrodes will be discussed.  

[1]       A. Paracchino et al., Nat. Mater. 2011, 10, 456–61

[2]       B. Seger et al., RSC Adv. 2013, 3, 25902

[3]       Y. W. Chen et al., Nat. Mater. 2011, 10, 539–544

[4]       S. Hu et al., Science 2014, 344, 1005–9

[5]       S. Y. Lee et al., Jpn. J. Appl. Phys. 2012, 51, 031102

[6]       R. Methaapanon et al., J. Phys. Chem. C 2010, 114, 10498–10504

[7]       G. D. Wilk et al., J. Appl. Phys. 2001, 89, 5243–5275

[8]        M. T. Mcdowell et al., ACS Appl. Mater. Interfaces 2015, 7, 15189–15199