Transient Absorption Spectroscopy of Anatase and Rutile: the Impact of Morphology and Phase on Photocatalytic Activity

Xiuli Wang,ab‡ Andreas Kafizas,b‡* Xiaoe Li,b Savio J. A. Moniz,c Philip J. T. Reardon,c Junwang Tangc Ivan P. Parkind and James R. Durrantb*

aState Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, 457 Zhongshan Road, Dalian, China

bDepartment of Chemistry, Imperial College London, South Kensington Campus, London, SW7 2AZ, UK

cDepartment of Chemical Engineering, University College London, Torrington Place, London WC1E 7JE, UK

dDepartment of Chemistry, University College London, 20 Gordon Street, London, WC1H 0AJ, UK

‡These authors contributed equally.

ABSTRACT

We employed transient absorption spectroscopy (TAS) to investigate the kinetic dependences of photocatalysis in anatase and rutile TiO2 films of varying morphology. In mesoporous films, anatase was ~ 30 times more efficient than rutile in the photocatalytic degradation of an intelligent ink model system. Independent of phase, up to 100 lower levels of photocatalysis was found in dense films. Charge carrier lifetimes were probed by TAS on the μs – s timescale. For both rutile and anatase, recombination was independent of morphology. Rutile exhibited up to 10 times slower recombination kinetics than anatase. Efficient, irreversible hole scavenging by alcohols was present in mesoporous anatase alone, resulting in the generation of long lived electrons (τ ≈ 0.7 s) which, upon the addition of the dye reduction target resazurin, enabled efficient electron transfer (τ ≈ 3 ms). Hole scavenging by alcohols on mesoporous rutile was substantially less efficient and more reversible than anatase, resulting in only a marginal increase in electron lifetime. The lower activity of rutile was not due to differences in recombination, but rather from the deficiency of rutile holes to drive efficient and irreversible alcohol oxidation.

KEYWORDS: Transient absorption spectroscopy, photo-generated carrier lifetime, photocatalysis, morphology, intelligent ink

INTRODUCTION

TiO2 is the most widely employed material for photocatalytic water splitting and degradation of organic pollutants. Its importance can be attributed to its unique physicochemical properties.1–3 For photocatalytic applications, anatase and rutile are the two most studied TiO2 polymorphs. Despite possessing the same chemical composition, the different arrangements of TiO6 octahedra in anatase (4 edge sharing partly distorted octahedra) and rutile (2 edge sharing non-distorted octahedra) result in different physicochemical properties. Aside from phase mixtures and composites, the anatase form is generally regarded as the more active phase in photocatalysis,1 especially for environmental applications2, despite rutile often showing superior photocatalytic performance in specific cases, such as water oxidation.4,5 This is further supported by its commercial application, where anatase is used exclusively in self-cleaning windows (Pilkington Activ, Saint-Gobain Bioclean, PPG Sunclean) and tiles (TOTO Hydrotect).6 Understanding why anatase often shows a higher activity compared with rutile in most photocatalytic reactions remains a key challenge that will support strategies to enhance TiO2 photocatalysis.

Nowadays, most studies of photocatalytic reactions are focused on particulate systems (as opposed to their dense thin film counterparts) due to the apparent reality that a high surface area is needed to achieve high quantum yields.7 Based on similar arguments, mesoporous TiO2 films have been actively developed.8 However, it is the dense form of TiO2 that is found in commercial self-cleaning products, where efficiency has been compromised for material durability. Although substantial efforts have been made to control the morphology of TiO2 to improve photocatalysis, the role of morphology in photocatalysis is not clear from the aspect of kinetics. To understand this, in this study we will examine the difference in photocatalytic activities for mesoporous and dense thin films of both pure anatase and rutile phase. We will then compare these differences in activity with the lifetime and population of photo-generated charges using transient absorption spectroscopy (TAS).

TAS is a form of laser flash spectroscopy that tracks transient changes in absorption after an excitation pulse. 9–11 It has been shown that the dynamics specific to photo-generated electron or hole carriers could be studied in TiO2 by tracking transient absorbance changes at particular wavelengths.12 This has allowed many research groups to use TAS to understand the dynamics of charge transfer in many photocatalytic processes, such as the reduction of O2,13 the oxidation of NO14 and the oxidation of water.15 To date, such studies have largely been limited to anatase, where no direct comparison of the charge carrier dynamics in anatase and rutile TiO2 has been made.

This study begins with a comparison of the oxidation of methanol, a widely used hole scavenger in photocatalysis. Although there have been TAS studies of the hole scavenging effect of alcohols with anatase,16 to our knowledge, there has been no such study with rutile. This study is then extended to intelligent ink, previously shown to be an effective model system for quantifying photocatalytic activity.17,18 The ink consists of three components dissolved in water: (i) a polymer thickener, (ii) glycerol and (iii) a resazurin redox dye. Intelligent ink is made rapid acting through the use of a hole scavenger, glycerol. Such alcohols are well known for their hole scavenging potency and are regularly used to increase the efficiency of photocatalysis, including water splitting.2 In removing holes, electron-hole recombination is retarded and photo-generated electrons have more time to take part in reduction processes of the dye, as summarized in Scheme 1.

Scheme 1. Two-step reduction of resazurin (royal blue) to resorufin (pink) and then dihydro-resorufin (colorless).

The focus of this study is to understand how the phase and morphology of TiO2 impacts on the kinetic competition between electron-hole recombination and their reaction with organics in solution – enabling better materials design for future optimization of the photocatalytic function.

EXPERIMENTAL SECTION

Sample preparation

Mesoporous TiO2 film

A colloidal anatase paste was prepared from the aqueous hydrolysis of titanium isopropoxide, as described in our previous publications.19 A colloidal rutile paste was prepared from a commercial powder (Shanghai ST-Nano Science and Technology Co.).20 Mesoporous films were formed by doctor-blading each colloidal paste onto glass microscope slides, cleaned with acetone prior to use. The pastes were allowed to air-dry for 30 min before calcination at 450 oC for 30 min.

Dense TiO2 film

Dense anatase thin-films were grown on quartz substrates by chemical vapor deposition from the reaction of TiCl4 and ethyl acetate vapors carried by N2 gas at 500 oC. Dense rutile films were made by simply annealing the dense anatase films at 1050 oC in air for 5 hours to ensure the complete conversion to its thermodynamic phase.21

Physical characterization

Surface microstructures were investigated by scanning electron microscopy (SEM) on a JOEL-6301F field emission instrument. X-ray diffraction (XRD) patterns were measured with a modified Bruker-Axs D8 diffractometer with parallel beam optics equipped with PSD LinxEye silicon strip detector. Patterns were fit to a Rietveld refined model using GSAS-EXPGUI software.22 UV-visible transmittance and reflectance spectra were recorded from 250 – 1100 nm using a Helios double beam instrument equipped with an integrating sphere. Specific surface area measurements were measured using the Brunauer–Emmett–Teller (BET) method23 in N2 using a TriStar 3000 Micromeritics device. Pore diameter distributions were determined using the Barret-Joyner-Halenda (BJH) method.24 Mesoporous film thickness was measured using an Alpha 200 Profilometer.

Photocatalytic activity

The photocatalytic test involved spray-coating intelligent ink25 evenly over the surface of the film.18 The intelligent ink (royal blue in color) consisted of hydroxyl ethyl cellulose polymer (0.45 g), glycerol (3.0 g) and the redox dye resazurin (40 mg) dissolved to 30 mL in distilled water. Samples were spray-coated with the ink using an aerosol spray gun (SIP Emerald Spray Gun, Halfords, Plc.) at an air-pressure feed of 3.5 bar until a blue coating was formed. UV-visible absorption spectra were measured from 450 – 700 nm using a PerkinElmer Lambda 25 device. A soft UVA light source was used as the excitation source for the photocatalytic reaction (365 nm –Vilber Lourmat 2 x 8W, Figure S1), which was shone from above (i.e. lamp – ink – sample – substrate). The lamp flux was measured using a UVX-Radiometer equipped with a UVX-36 sensor (λ = 365 ± 15 nm).

Previous studies have shown the photocatalytic reduction of intelligent ink on anatase TiO2 is first order, where the rate constant for the reduction of resazurin to resorufin is almost 4 times larger than the reduction of resorufin to dihydro-resorufin.26 As such, the rate of the first reduction (Scheme 1) can be determined by assessing the decrease in absorbance at 608 nm (λmax, resazurin) at early times (i.e. times in which the rate of decrease is linear). The thickness of the ink coating could be derived using the Beer-Lambert law given the dye was 5.8 mM in concentration and possesses an extinction coefficient of 43,500 M-1cm-1 at 608 nm. The rate of decrease at 608 nm could then be converted into a molecular rate. Thus with knowledge of the light flux from the 365 nm excitation source (450 μW.cm-2 ≈ 1.1 x 1015 photons.cm-2.s-1) coupled with the material’s absorbance at 365 nm the quantum yield (QY) was derived:

QY %= molecules degraded (cm-2s-1)photons absorbed (cm-2s-1)×100

(1)

Transient absorption spectroscopy

Charge carrier dynamics of TiO2 films were measured using transient absorption spectroscopy (TAS) from the μs – s timescale at room temperature. The TAS apparatus has been described in detail elsewhere.27 In brief, a 75 W Xe lamp is used as a probe beam with a monochromator placed before the sample. The change in transmitted light is measured by a Si PIN photodiode after a UV laser excitation pulse is applied on the sample using the third harmonic of a Nd: YAG laser (355 nm, 6 ns pulse width). Reasonably low laser intensities were used (~ 30 - 500 µJ cm-2 pulse-1) with a laser repetition rate of 1 Hz. Each TAS trace is the result of averaging between 50 – 500 scans.

RESULTS

Materials characterization

X-ray diffraction showed that the mesoporous and dense films of anatase (I41/ amd) and rutile (P42/ mnm) were phase-pure (Figure S2). Scherrer line broadening studies28 showed that the average width of crystallites were ~ 40 nm in the mesoporous anatase film, ~ 40 nm in the dense anatase film, ~ 110 nm in the mesoporous rutile film and ~ 80 nm in the dense rutile film. Dense films showed strong preferred growth effects, akin to epitaxially grown films by CVD.29 No preferred growth was observed in mesoporous films, typical of particulate growth.

The specific surface areas of mesoporous anatase and rutile (Figure S4) as determined by the Brunauer–Emmett–Teller (BET) method were measured to be 100 m2.g-1 and 72 m2.g-1 respectively. The specific surface areas of our dense films were lower than the detection limits of the device (< 1 m2.g-1). Barret-Joyner-Halenda (BJH) analysis showed that the average pore diameters in mesoporous anatase (~ 25– 30 nm) and rutile (~ 24 – 31 nm) were similar. Anatase was more porous, possessing an average pore volume of 0.82 cm3.g-1 compared with 0.20 cm3.g-1 for rutile. Film topography was investigated by SEM (Figure 1). Mesoporous films were rough and porous, with anatase possessing primarily spherical particles ~ 20 nm in width and rutile possessing more fluffy particles that ranged from 100 – 500 nm in width. Dense films were more compact and flat, with anatase possessing tetragonal domains 100 – 200 nm wide and rutile possessing larger and more rounded domains 200 – 600 nm wide.

Figure 1. SEM images of (a) mesoporous anatase, (b) mesoporous rutile, (c) dense anatase and (d) dense rutile films.

Photocatalytic activity

The photocatalytic activity of mesoporous anatase and rutile films was examined using the intelligent ink test as a model system.25 Figure 2a depicts the changes observed by UV-visible spectroscopy at selected times during the reaction on mesoporous rutile. At 0 s, a spectrum characteristic of the sole presence of resazurin is observed.30 At the early stages of the reaction, predominantly resazurin (λmax = 608 nm) is reduced to resorufin (λmax = 575 nm). By 450 s most of the resazurin was consumed and a maximum concentration of resorufin was reached. By 1000 s the resazurin dye was completely reduced and a spectrum characteristic of the sole presence of resorufin is observed.26 Further irradiation caused the resorufin spectrum to decrease in intensity as it was reduced to its colorless counterpart dihydro-resorufin.

By plotting the change in absorbance at 608 nm against irradiation time (Figure 2b) one can track the kinetics of dye reduction from resazurin to resorufin.26 Almost no change was observed in the blank glass sample, proving dye reduction was due solely to semiconductor mediated photocatalysis. The photocatalytic conversion of resazurin to resorufin was two orders of magnitude faster in mesoporous anatase than dense anatase, two orders of magnitude faster than mesoporous rutile and more than three orders of magnitude faster than dense rutile.

Figure 2. (a) The changing UV-visible absorption spectra during the photocatalysis of intelligent ink on mesoporous rutile under excitation at 365 nm and (b) the change in normalized absorbance at 608 nm (λmax resazurin) with time (s, log timescale) during the photocatalysis of intelligent ink for mesoporous anatase [open circles], mesoporous rutile [black circles], dense anatase [open squares], dense rutile [black squares] and a blank piece of glass [grey circles].