Dynamics of photogenerated holes in nanocrystalline α-Fe₂O₃ electrodes for water oxidation probed by transient absorption spectroscopy

Abstract

Transient absorption spectroscopy on the μs–s time scale is used to monitor the yield and decay dynamics of photogenerated holes in nanocrystalline hematite photoanodes. In the absence of a positive applied bias, these holes are observed to undergo rapid electron–hole recombination. The application of a positive bias results in the generation of long-lived (3 ± 1 s lifetime) photoholes.

---

The rapid depletion of fossil fuel reserves and the concerns regarding the environmental impact of increasing emission of CO₂ to the atmosphere have prompted the search for alternative, clean and renewable energy sources. In this context, much research in recent years has focused on the development of photocatalyst materials that can work towards efficient conversion and storage of solar energy in the form of molecular fuels such as H₂ (obtained from the photoelectrolysis of water) or CH₃OH (resulting from the reduction of CO₂).^1,2

Although TiO₂ is the most widely studied material for photoelectrochemical (PEC) water splitting, its wide band gap (3.0–3.2 eV) restricts absorption to the UV region. In order to utilise a greater fraction of the solar spectrum, other metal oxides, such as WO₃ and Fe₂O₃, have been investigated.^3,4Hematite (α-Fe₂O₃) is an n-type semiconductor with a band gap of ca. 2.1 eV, corresponding to light absorption in the visible; its valence band edge energy is suitable for water oxidation. It is also non-toxic, abundant and stable in most electrolytes. Nanocrystalline, mesoporous α-Fe₂O₃ films are considered one of the most promising candidates for PEC applications.^5–7 However, a positive electrical bias is required for water photolysis to occur, generally considered necessary to increase the reduction potential of electrons for proton reduction. Hematite is thought to have an extremely short hole diffusion length (2–4 nm⁵)—an indicator of rapid electron–hole recombination—and it has been suggested that slow hole transfer kinetics at the semiconductor–liquid junction (SCLJ) may also limit the efficiency of this material.⁸ A thorough understanding of charge carrier dynamics and their sensitivity to applied bias are, therefore, key to devising strategies for enhancing the performance of such photoanode materials.

The dynamics of charge separation, transport and recombination in dye-sensitised solar cells employing n-type, mesoporous titania films have been extensively studied and are now reasonably well understood.^9,10 However, the analogous processes in PEC water photolysis cells are not yet well characterised. Transient absorption spectroscopy (TAS) allows the study of charge carrier dynamics within the semiconductor film and has recently been used in our group to probe the lifetime of holes in nanocrystalline TiO₂ films, and provide evidence that non-geminate e⁻–h⁺ recombination limits the efficiency of water oxidation in this material.¹¹ Recently, TAS has also been used to study the mechanism of water oxidation in TiO₂ in a complete PEC cell for the first time, indicating that water oxidation occurs on a millisecond timescale under positive electrical bias.¹²

To the best of our knowledge, the only previous TAS studies of iron oxide focused on fast (fs–ns) timescales—often employing relatively high excitation conditions—identifying charge carrier decay times of hundreds of picoseconds.^13–15 In this paper we focus on the transient absorption in the μs–s timescale, likely to correspond to the timescale of water oxidation, employing relatively low intensity excitation conditions, more comparable to solar irradiation conditions. We focus upon the dynamics of photogenerated holes as a function of hole scavengers in the electrolyte and under applied bias. The undoped α-Fe₂O₃ films used in our study were prepared by atmospheric pressure chemical vapor deposition (APCVD) and have a dendritic nanoporous structure.¹⁶ Transient absorption measurements on the μs–s timescale were obtained using band-gap excitation at 337 nm or 355 nm (∼0.2 mJ cm⁻², 0.33–2.0 Hz, illuminated through the substrate (SE)), and a monochromatic probe beam, as described in detail elsewhere.^11,12

The inset in Fig. 1 shows the transient absorption (TA) spectrum of nanoporous hematite films measured in an argon atmosphere, which is characterised by a long-lived (μs–ms) absorption peak at 580 nm and tail that extends to the near IR. This photoinduced absorption at ∼580 nm is assigned primarily to holes, consistent with data collected under electrical bias (see below). Charge carrier trapping is expected to occur in the subnanosecond timescale, therefore the spectral features observed are likely to correspond to trapped holes, possibly surface bound OH radicals. Typical transient decays in the absence of applied bias (see Fig. 1) show dispersive power-law decay dynamics, typical of bimolecular recombination in the presence of charge trapping, as we have reported previously for nanocrystalline TiO₂ films.^11,12

Fig. 1 Transient absorption decay traces of nanocrystalline hematite films probed at 580 nm upon laser excitation at 337 nm, in different media. All decays shown were obtained with SE illumination but EE illumination results are identical. Inset: transient absorption spectra in argon, recorded 5 μs and 80 μs after laser excitation.

The behaviour of photogenerated holes was further investigated by the immersion of the film in 0.1 M NaOH electrolyte (pH ∼12.8) and by the addition of various hole scavengers, including methanol and iodide (Fig. 1), thiocyanate and isopropanol (not shown). In all cases, the yield and decay dynamics of photogenerated holes, as monitored by μs–ms transient absorption signal at 580 nm, were essentially insensitive to the chemical environment of the film. This is in contrast to typical behaviour observed for mesoporous TiO₂, where hole scavengers such as methanol result in the rapid (often nanosecond) quenching of photogenerated holes.¹¹ It appears that, although holes at the top of the Fe₂O₃ valence band (ca. +2.2 V vs. RHE) are thermodynamically able to oxidise iodide (E⁰ = +1.35 V),¹⁷thiocyanate (+1.64 V)¹⁷ and methanol (+0.02 V),¹⁸ the holes observed in our μs–ms transient absorption studies, collected without any applied bias, do not have a long enough lifetime to oxidise these species.‡

In order to investigate the hole kinetics of hematite in a fully-functional PEC cell, a three-electrode cell with Ag/AgCl/sat. KCl reference electrode, Pt gauze counter electrode and de-aerated 0.1 M NaOH electrolyte with/without ca. 0.2 mM MeOH was used. Typical current/voltage data obtained in the dark and under white light irradiation (150 W ozone-free xenon lamp) are shown in Fig. 2. As expected, significant photocurrent is only observed with the application of positive potential to the hematite photoanode, assigned to water photo-oxdiation. The presence of methanol results in a shift in the onset and significant enhancement of photocurrent, assigned to methanol photo-oxidation.

Fig. 2 Effect of bias in the photoelectrochemical response of α-Fe₂O₃ electrodes. Current–voltage curves in 0.1 M NaOH in the absence (black curve) and presence (red curve) of methanol, under white light illumination (SE). Dark current is negligible in the potential range shown.

Fig. 3 shows corresponding transient absorption data collected using the complete PEC cell under light irradiation, measured under applied biases versusAg/AgCl of −0.1 V (corresponding to approximately open circuit) and +0.4 V (corresponding to significant photocurrent generation). At −0.1 V, the transient exhibits microsecond decay dynamics, very similar to those observed for isolated films (as shown in Fig. 1), both with and without methanol. However, an applied +0.4 V bias results in the appearance of a much longer lived decay phase, (Fig. 3). This slow decay phase can be fitted with a stretched exponential function with a lifetime of 3 ± 1 s, decreasing to 400 ± 100 ms in presence of methanol.

Fig. 3 Effect of electrical bias in the transient absorption of photogenerated holes in nanocrystalline α-Fe₂O₃ electrodes measured in a complete PEC cell. In 0.1 M NaOH under negative applied bias of −0.1 V vs.Ag/AgCl (blue curve) and under positive applied bias of +0.4 V vs.Ag/AgCl (black curve). Upon the addition of methanol in the positive bias condition (red curve), the faster decay indicates the oxidation of this substrate by photoholes. All decays are obtained upon laser excitation at 355 nm and probing at 580 nm.

Our observation of the appearance of a long lived transient signal under positive applied bias is strongly indicative of the formation of long lived holes which have avoided rapid recombination. The faster decay of this long lived signal in the presence of methanol indicates that the transient lifetime is determined by surface oxidation kinetics, confirming the assignment of this signal to surface active holes. In this context, the faster kinetics in the presence of methanol are consistent with the relatively facile oxidation of methanol compared to water. We note that the amplitude of this long lived hole signal is only ∼10% of the initial hole photoinduced signal (Fig. 1), indicating that even under positive applied bias, the majority of photogenerated holes still undergo rapid electron–hole recombination on the microsecond timescale.

These results could be interpreted in terms of rapid bulk recombination, or slow hole transfer kinetics at the SCLJ resulting in recombination of surface-trapped holes. It is often considered that the low intrinsic faradaic rate constant for water oxidation on Fe₂O₃ limits the performance of this material.⁸ On the other hand, fast bulk recombination (indicated by the short hole diffusion length^5,8) is thought to prevent holes generated in the semiconductor bulk from reaching the surface. Our observation that a positive bias is necessary to generate long lived holes capable of driving surface oxidation reactions is consistent with the perception that rapid recombination may be the key loss process in hematite photoelectrodes. According to Gerischer's electric double-layer theory of the semiconductor–electrolyte junction,²⁰ a positive bias will increase the width of the depletion layer and facilitate removal of electrons. However, it has been suggested that this band bending model does not apply to the nanoporous structures investigated here.²¹ In the latter case, the effect of positive bias can be interpreted as a decrease of the background electron density throughout the film (Scheme 1). Under either interpretation, the overall effect will be a decrease in electron–hole recombination, increasing the yield of longer-lived holes, as observed in this work. This provides an additional explanation for the requirement of applied positive bias for water photo-oxidation by iron oxide, specifically that decreased electron density is necessary to reduce recombination and allow long-lived photoholes to diffuse to the surface and oxidise water.¹⁶Scheme 1 illustrates a system where recombination is assumed to be mediated by free conduction band electrons, but the results observed are also consistent with trapped electron mediated recombination.

Scheme 1 Schematic representation of the effect of bias potential on the Fermi level position and relative rates of electron–hole recombination and surface charge transfer in a hematite photoanode for water photoelectrolysis. The applied positive bias decreases the background electron density relative to the open circuit condition, decreasing recombination and resulting in longer-lived photoholes.

Electron–hole recombination has been shown to be the key loss process limiting water photolysis by TiO₂ photoelectrodes.¹²Titania, however, appears to differ from the hematite films studied herein in that scavenging of photogenerated charge carriers—e.g. by methanol and Ag⁺—can be competitive with recombination, even in the absence of applied bias.¹¹ Recombination may be more dominant in hematite due to (i) greater electron (donor) density, (ii) greater distance for holes to travel to reach the surface, (iii) lower hole mobility. However, such detailed investigation is beyond the scope of this communication.

The study reported herein indicates that the photogeneration of long lived holes is correlated with water oxidation by hematite photoelectrodes. Further studies are currently under way to quantify this dependence and elucidate the effects of morphology and surface co-catalysts on electrode performance.

Funding from EPSRC is gratefully acknowledged. The authors thank Dr Piers Barnes, Dr Wenhua Leng and Dr Stephen Dennison for helpful discussions, and Dr Christopher Barnett for equipment development and maintenance.

Notes and references

1. J. Barber, Chem. Soc. Rev., 2009, 38, 185–196.
2. N. S. Lewis and D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 15729–15735.
3. C. Santato, M. Ulmann and J. Augustynski, J. Phys. Chem. B, 2001, 105, 936–940.
4. F. E. Osterloh, Chem. Mater., 2008, 20, 35–54.
5. J. H. Kennedy and K. W. Frese, J. Electrochem. Soc., 1978, 125, 709–714.
6. R. Shinar and J. H. Kennedy, Sol. Energy Mater., 1982, 6, 323–335.
7. K. Sivula, R. Zboril, F. Le Formal, R. Robert, A. Weidenkaff, J. Tucek, J. Frydrych and M. Grätzel, J. Am. Chem. Soc., 2010, 132, 7436–7444.
8. M. P. Dare-Edwards, J. B. Goodenough, A. Hamnett and P. R. Trevellick, J. Chem. Soc., Faraday Trans. 1, 1983, 79, 2027–2041.
9. S. A. Haque, E. Palomares, B. M. Cho, A. N. M. Green, N. Hirata and J. R. Durrant, J. Am. Chem. Soc., 2005, 127, 3456–3462.
10. B. O'Regan and J. R. Durrant, Acc. Chem. Res., 2009, 42, 1799–1808.
11. J. Tang, J. R. Durrant and D. R. Klug, J. Am. Chem. Soc., 2008, 130, 13885–13891.
12. A. J. Cowan, J. Tang, W. Leng, J. R. Durrant and D. R. Klug, J. Phys. Chem. C, 2010, 114, 4208–4214.
13. N. J. Cherepy, D. B. Liston, J. A. Lovejoy, H. M. Deng and J. Z. Zhang, J. Phys. Chem. B, 1998, 102, 770–776.
14. V. A. Nadtochenko, N. N. Denisov, V. Y. Gak, F. E. Gostev, A. A. Titov, O. M. Sarkisov and V. V. Nikandrov, Russ. Chem. Bull., 2002, 51, 457–461.
15. G. Xiong, A. G. Joly, G. P. Holtom, C. M. Wang, D. E. McCready, K. M. Beck and W. P. Hess, J. Phys. Chem. B, 2006, 110, 16937–16940.
16. A. Kay, I. Cesar and M. Gratzel, J. Am. Chem. Soc., 2006, 128, 15714–15721.
17. P. Wardman, J. Phys. Chem. Ref. Data, 1989, 18, 1637–1755.
18. T. Iwasita, Electrochim. Acta, 2002, 47, 3663–3674.
19. A. J. Bard, J. Am. Chem. Soc., 2010, 132, 7559–7567.
20. H. Gerischer, J. Electroanal. Chem., 1975, 58, 263–274.
21. I. Cesar, K. Sivula, A. Kay, R. Zboril and M. Graetzel, J. Phys. Chem. C, 2009, 113, 772–782.

---

Footnotes

† Electronic supplementary information (ESI) available: Experimental details and sample characterisation. See DOI: 10.1039/c0cc03627g

‡ When evaluating the lack of methanol and water oxidation on the μs–ms timescales by Fe₂O₃ holes it is also important to consider the thermodynamics of the one-electron intermediates in these oxidation reactions which would lead to a considerably lower thermodynamic driving force.¹⁹

---