Photosensitive oxide semiconductors for solar hydrogen fuel and water disinfection

Basic reactions

The formation of hydrogen from water, using solar energy, can occur along several alternative pathways, including:

OPEN IN VIEWER

2.

Representation of two water oxidation routes (Φ is the work function (WF), μ(e⁻) is the chemical potential of electrons, and subscripts A and C are related to the anode and cathode, respectively). Reproduced with permission from Ref. 10. Copyright T. Bak, 2010

So far, the most environmentally friendly method of generating solar hydrogen is by water electrolysis using photovoltaic (PV) electricity. This technology, however, is inconvenient as it requires application of two devices: a PV solar panel producing electricity and an electrolyser. Conversely, a photoelectrochemical cell (PEC) for water splitting requires only one device.

The conceptual design of a PEC is represented schematically in Fig. 3. Although their efficiency is currently below the level required for commercialisation, a substantial increase in the energy conversion efficiency (ECE) has been recently achieved. ¹¹ The ECE of the system reported by Kashelev and Turner ¹² was larger (12%); however, the performance of their device rapidly deteriorates because of corrosion. A key aim of the research is the development of a PEC with enhanced ECE and high chemical stability. A critical element of the PEC is the photoelectrode, which is made of a POS.

OPEN IN VIEWER

3.

General concept of photoelectrochemical water splitting using sunlight. Reproduced with permission from Ref. 10. Copyright T. Bak, 2010

Fujishima and Honda ¹³ were the first to show that water may be split using TiO₂ as a photoanode. Since then, TiO₂ has been considered as one of the most promising candidates for water photolysis, although its high band gap (3 eV) has guided the search for alternatives that can absorb light of lower energy but higher intensity in the solar spectrum.

The process of solar water splitting requires multi-electron transfer between the anodic site of a semiconducting oxide surface and adsorbed water molecules. ¹⁴ Under neutral and acidic conditions water oxidation can be described by the following reaction (3a)

Under alkaline condition oxygen evolution takes place according to the following reaction (3b) where e′ and h• denote a quasi-free electron and electron hole (in the oxide lattice), respectively. As seen, reaction (3a), which takes place at anodic sites in neutral and acidic conditions, results in the formation of gaseous oxygen and protons. Then the charge neutrality condition requires that the four generated electrons generated at the anodic sites, are transported to cathodic sites where they react with protons to form H₂ (4a)

In the case of neutral or alkaline solution hydrogen evolution may be represented by the following reaction (4b)

Reactions (3) and (4), which are associated with the transfer of four electrons per two water molecules, represent total water oxidation.

Numerous studies have aimed to develop systems with enhanced ECE. According to the US Department of Energy, an ECE of 10% may be economical. ¹⁵ While higher efficiencies have been reported, these are either too expensive or unstable in aqueous environments.

The key requirement for total water oxidation is the coordinated removal of four electrons from two water molecules. This is not easy since water is a very stable compound. The transfer requires optimisation of both the collective factor, such as the Fermi level and the local factor, which can be considered in terms of atomic site lattice elements. In the case of TiO₂, titanium vacancies are considered to be surface active sites. ¹⁶ The mechanism of total water oxidation in the presence of such active sites is represented schematically in Fig. 4.

OPEN IN VIEWER

4.

Schematic representation of total water oxidation and the associated multi-electron charge transfer at titanium vacancy acting as an active site

Photoelectrochemical water splitting is very attractive for the following reasons:

The reports, which have accumulated so far, allow the following points to be made:

Cost-related aspects

The commercialisation of the solar hydrogen technology is critically determined by cost-related matters. At present, hydrogen is mainly generated by the conversion of natural gas. The direct costs of hydrogen generation using this method are relatively low. Awareness is growing, however, that the indirect costs, associated with the generation of hydrogen from natural gas, are high. These include the costs related to the damaged environment, such as extreme weather conditions and the associated costs in the production of food, increased pollution of air, water and soil and the impact of pollution on health and productivity. So far, these costs, which are substantial, are borne by the community. Governments are beginning to protect the community from pollution-induced health problems, for example, the government of California undertook legal action against car manufacturers (powered by gasoline) in 2009. ¹⁷ This action was reflective of the community pressure applied on governments to reduce pollution by vehicles powered by fossil fuels. There is an increasingly urgent need to develop solar hydrogen fuel, which is environmentally clean in both its generation and combustion.

The production cost of hydrogen by steam reforming of methane is approximately $6/GJ. ¹⁷ On the other hand, the cost of solar hydrogen production using PV electricity is $28/GJ. ¹⁷ This figure is higher than the cost of hydrogen production from natural gas by the factor of 4·7. However, taking into account that the costs associated with air pollution resulting from combustion of fossil fuels is equivalent to $33 GJ, the solar hydrogen fuel is the ultimate winner.

At present, solar hydrogen is generated mainly by water electrolysis using PV electricity. This approach, however, is inconvenient as it requires two devices: electrolyser and PV cell. The development of a device, which allows water photolysis using a single device, PEC is important. So far, however, the ECE of the PECs based on oxide semiconductors remains below the level required for commercialisation.

Charge transfer at anodic site

The charge transfer may be increased by lowering the Fermi level of the anodic site. The lower the Fermi level the larger the driving force to transfer electrons from the water molecule to the surface. Efficient charge transfer also requires that the water molecule is adsorbed on a site, which has the capacity to attract electrons, such as a titanium vacancy. ^8,9 Charge transfer may be considered in terms of the formation of an active complex, which in the case of total water oxidation, is decomposed into oxygen and protons. Charge transfer may be enhanced by the presence of secondary phases, such as IrO₂, which is one of the most commonly described catalyst for the oxygen evolution reaction. Do et al. ³⁸ and Martin et al. ³⁹ postulated that surface loading with small islets of nanoparticulate of metals and metal oxides, acting as a co-catalyst, improves water splitting by suppressing water formation thus allowing the forward reaction, and improving charge separation.

Charge transfer at cathodic site

The key function of the cathodic site is to provide electrons to the adsorbed oxygen molecules (see reaction (7)) or other oxidant. The transfer of electrons from TiO₂ to the oxygen molecules requires that the Fermi level of the cathodic site is elevated above a certain critical level. For efficient charge transfer, oxygen molecules should be adsorbed on sites that favour the transfer of electrons from the surface of TiO₂ to the oxygen molecules. Such sites are formed by donor-type defects, such as oxygen vacancies and titanium interstitials. Frequently, the electrons at cathodic sites are scavenged by small islets of secondary phases, such as noble metals. ^24,38–42

Efficient charge transfer associated with the anodic and cathodic reactions (3)–(6), as well as the charge transport required to remove the electronic charge carriers from the reactive sites, are crucial in the formation of a high-performance photocatalytic system. In some instances, metals, such as Pt, are applied as electrical leads and contacts. It is frequently assumed that these electrodes are not reactive. However, it has been established that platinum reacts with air, leading to the formation of a PtO₂ layer, as well as with some oxide semiconductors, leading to the formation of mixed phases. ⁴² These phases may form a pathway for the charge transfer between TiO₂ and the leads.

Electronic structure (KPP-1)

The electronic structure is the key property that has an effect on light absorption (band gap) and the light-induced reactivity of oxide semiconductors. The latter property depends on the position of energy levels of oxide semiconductor and the electrochemical energy levels of the aqueous electrolyte.

The band gap is the property that controls light absorption leading to the generation of electron–electron–hole pairs by band gap ionisation. The band gap of commercially available rutile, which is approximately 3 eV, allows only a small fraction of the solar spectrum to be absorbed. Reduction of the band gap to approximately 1·8–2 eV is needed to maximise absorption of sun light. This may be achieved by doping with foreign ions, such as niobium, iron, chromium, molybdenum, carbon, boron and sulphur.

Charge transfer within photoelectrochemical systems formed by POSs immersed in aqueous electrolytes depends on the difference between the energy of the bottom of the conduction band (E _C) and reduction of the electrochemical couple [E(H⁺/H₂)] as well as the difference between the energy of the top of the valence band (E _V) and the oxidation of the electrochemical couple. Spontaneous charge transfer requires that E _C>E(H⁺/H₂) and E(O₂/H₂O)>E _V. The charge transfer may be modified by shifting the positions of the E _C and the E _V through doping of TiO₂ with foreign ions.

Surface Fermi level (KPP-2)

The Fermi level may be considered as the chemical potential of the electrons within the POSs. While the electrochemical potential in equilibrium is the same across the surface and the bulk, the chemical potential is not. The Fermi level of the outermost surface layer of POSs, which is different than that of the bulk phase, has a profound effect on the reactivity of POSs and the related charge transfer. The Fermi level of the outermost surface layer may be determined using WF measurements. Work function measurements may also be used for monitoring the Fermi level during chemical reactions and to study the reaction kinetics and the associated charge transfer.

The performance of the anode requires that its surface has a tendency to accept electrons from the water molecule. This may be achieved when its Fermi level is relatively low. Conversely the performance of the cathode requires that its Fermi level is high. Then the surface has a tendency to donate electrons.

The position of the Fermi level may be modified in a controlled manner by variation of oxygen activity in the oxide lattice and the concentration of aliovalent ions incorporated into the lattice. For example, amphoteric semiconductors, such as TiO₂, may be modified to form both n- and p-type TiO₂ with controlled Fermi level position. The latter case opens an opportunity to form a PEC equipped with two photoelectrodes (photoanode and photocathode) made of n- and p-type TiO₂, respectively. Such cell would not require a bias voltage to split water (see below).

The surface Fermi level may be radically modified by imposition of a low-dimensional surface layer. Such surface layer may also be formed as a result of segregation when the segregation-induced enrichment surpasses a certain critical limit corresponding to the surface miscibility limit dictated by the local thermodynamic conditions.

Population of reactive surface sites (KPP-3)

While the Fermi level is considered a collective property, which is related to the surface as a continuum, the reactivity of the surface with water strongly depends on local properties of specific sites at which water can be adsorbed leading to the formation of active complexes. The active complex is a precursor of the final product of water splitting leading ultimately to the formation of oxygen gas and protons. The active sites for TiO₂-based anodes are titanium vacancies.

Optimising the concentration of titanium vacancies should improve the photocatalytic performance of the TiO₂ anodes. This may be achieved through ion bombardment (implantation) with Ti and/or O, which will alter the surface of the material without altering the chemical stoichiometry.

Near-surface electric field (KPP-4)

Light-induced ionisation results in the generation of electronic charge carriers (electrons and holes). In order to prevent the recombination, these charge carriers need to be separated. This can be achieved by imposing a local electric field. Such field may be imposed when an oxide semiconductor is immersed in water, then water adsorption results in charge transfer at the oxide/water interface. The electric field leading to charge separation may also be achieved by imposing a chemically induced potential gradient. ⁹

Charge transport (KPP-5)

The light-induced electronic charge carriers, which are formed at a certain distance from the surface, where light is penetrated, must be transported from the site of their generation beneath the surface to the reactivity sites at the surface. The energy losses related to this charge transport could be minimised when the charge transport is enhanced. This may be achieved by doping leading to imposition of higher concentration of electronic charge carriers and, consequently, reduction of ohmic resistance. The mobility term, which is related to the crystalline structure and the chemistry of the periodic lattice, may be modified by imposition of dislocations leading to strains. ⁶

Charge transfer (KPP-6)

As noted above, charge transfer is critical for a number of reactions between the TiO₂ surface and water.

The charge transfer may be enhanced by deposition of small islets of co-catalysts and can be further tuned through engineering of the size, shape and chemical composition of such surface functional islets. The charge transfer may also be modified by imposition of an optimised concentration of the reactivity surface sites.

Electrical conductivity

The electrical conductivity measurements of metal oxides at elevated temperatures are commonly used in the determination of the effect of oxygen activity on concentration of electronic charge carriers. Such interpretation of the dependence requires to assume that the mobility term is independent of oxygen activity. ²⁰ Electrical conductivity measurements may also be used to monitor the electrical properties of oxides during processing at elevated temperatures. ²⁰ The dependence between the electrical conductivity and oxygen activity may be used to determine defect disorder models. ²⁰

For amphoteric semiconductors such as TiO₂, the band gap can be estimated from the temperature dependence of the minimum in electrical conductivity ⁵⁷ (15) where N represents the density of states of n- and p-type levels, and μ the mobility of the electrons and holes, respectively, β is the temperature coefficient of the band gap that is represented by the following expression (16)

Thermoelectric power

Thermoelectric power is the electrical property related to the electrical potential, which is formed along the temperature gradient. The TP for n-type materials is dependent on the concentration of electrons (17) where N is the density of states and A is the kinetic term. Therefore, simultaneous measurements of both electrical conductivity and TP may be used to determine both the concentration and the mobility terms.

Work function

Information about the semiconducting properties of the surface layer may be obtained by WF measurements. ^9,26 Changes in the WF are reflective of changes in the surface layer and its defect disorder. Work function measurements at elevated temperatures may be used for the determination of defect disorder of the outermost surface layer and the reaction kinetics at and in this layer. The WF changes of TiO₂ during oxidation and reduction at 1073 K are illustrated in Fig. 16. ¹⁰

OPEN IN VIEWER

16.

The work function (WF) changes during oxidation and reduction of TiO₂ at elevated temperatures ¹⁰

The picture is entirely different at room temperature (Fig. 17) ²⁶ when the chemical reactions are markedly slower. In the latter case, the fast WF changes during oxidation are reflective of oxygen chemisorption and the subsequent slow changes are determined by oxygen incorporation into the TiO₂ lattice. The latter process becomes extremely slow.

OPEN IN VIEWER

17.

Work function changes of TiO₂ during oxidation at room temperature showing the ranges related to oxygen chemisorptions and oxygen incorporation ²⁶

Surface photovoltage

The effect of light, of controlled energy, on the surface charge of a POS may be monitored by measurements of the surface photovoltage (SPV). ^58,59 Changes in the SPV are reflective of surface photoreactivity with the adsorbed species such as water and its decomposition products, including oxygen and hydrogen.

The SPV is extremely sensitive to the behaviour of surface and bulk states close to the surface of a POS, which change as a function of photon energy, temperature and gas pressure. ⁵⁹ The key components of a instrument to measure the SPV are illustrated in Fig. 18.

OPEN IN VIEWER

18.

Schematic of the set-up used to measure surface photovoltage of a photosensitive oxide semiconductor

OPEN IN VIEWER

19.

Surface photovoltage (SPV) curves of an n-type α-Fe₂O₃ (haematite) layer deposited by sol–gel method before and after heat treatment. The enhanced SPV signals after annealing of the haematite layer correspond to an increased activity as a photoanode in water oxidation. Black dots represent the SPV of the film as prepared, red triangles after annealing at 550°C in air and blue stars after treatment in an oxygen plasma at 800°C. The increase in SPV at photon energies above 1 eV in the upper part of the graph can be explained by states in the band gap or surface states of the POSs film. The increase above 2 eV is explained by a band to band transition of electrons in the oxide. From the lower part of the graph, conclusions can be met on the recombination velocity of surface states and bulk properties ⁵⁸

Specific attributes of nuclear techniques

This section provides a brief overview of specific nuclear techniques suitable for characterisation and modification of POS materials. The nuclear techniques offer a wide range of unique approaches to characterise the bulk and surface properties of POSs and to establish the effect of modification of these on photoelectrochemical response. The techniques include: ^77–97

The nuclear techniques described above have the potential to provide unique insight into the properties of POSs and to monitor the effects of modification in order to achieve desired performance.

Intensive research aimed at the development of solar cells for the production of hydrogen fuel by photoelectrochemical water splitting is continuing, but the commercialisation of such solar cells, PECs, requires that the efficiency of the conversion of solar energy into the chemical energy for water splitting needs to be further improved to make this economically viable.

The focus of this section is on nuclear techniques, and in particular, their specific application in the characterisation of surface versus bulk chemical composition and microstructure as well as their modification capabilities in order to optimise the KPPs.

A wide range of techniques can be used to modify the chemical composition of POSs leading to the formation of solid solutions. The obvious limitation of this approach is determined by the properties of specific structures that are thermodynamically stable and the related solubility limits. Application of nuclear techniques offers an approach to overcome these limits through ion implantation. It has been established that implantation and post-implantation temperature treatment allows modification of both microstructure and chemical composition in a controlled manner, leading to modification of the electronic structure and the related semiconducting properties.

In addition, the use of ion beams offers the possibility depositing isotopic thin films that may exhibit outstanding properties. The deposition of isotopic thin films offers a unique approach in formation of the outermost surface layer with enhanced properties that are needed to achieve maximised performance. This approach may be used for tuning the local electronic structure and the charge transport to the level that is needed to maximise light absorption and light-induced reactivity with water.

The most commonly applied nuclear technique used for the modification of materials properties is ion implantation. While this technique is sometimes referred to as doping, the effect of this kind of doping is entirely different from that achieved by the doping using chemical processing procedures. The main difference is that it is a non-equilibrium process, it is limited to the surface layers and it is usually associated with structural damage. Post-implantation temperature treatments may be used to recover some of the damage and to diffuse or to coalesce the implanted ions, corresponding to the local solubility limit. Proton-induced X-ray emission (PIXE) may be used for very precise chemical analysis, including the determination of impurities, with the maximum sensitivity at parts per million level around Z = 26 (Fe). Rutherford backscattering offers chemical analysis for surface and near-surface layer, and depth profile of elements, and is most suited to heavy elements imbedded in light matrices. Nuclear reaction analysis (NRA) may be used for determination of chemical composition and depth profile of light elements, such as hydrogen. Positron annihilation lifetime spectroscopy (PALS) allows the determination of structural defects, such as voids (agglomerations of vacancies) and single vacancies. Isotopic oxygen exchange has been commonly used in the determination of the kinetics of oxygen mobility the oxide lattice.

Oxygen exchange kinetics

This technique may be applied for better understanding of the oxygen reduction reaction (ORR). A novel approach using oxygen-isotope exchange, called isothermal isotope exchange (IIE), has been developed to extract ORR rate coefficients and mechanisms. ⁷⁷ An example of ¹⁸ O exchange is shown in Fig. 23 demonstrating an ability to distinguish surface oxygen dissociation and incorporation into the lattice. Moreover, this can be used to determine catalytic effects of surface decoration ⁷⁸ and heterogeneous mixtures of oxides. Integration of these heterogeneous catalysis techniques with electroanalytical techniques, such as electrochemical impedance spectroscopy (EIS) and cyclic voltammetry will provide direct correlation between kinetic mechanisms and cell polarisation.

OPEN IN VIEWER

23.

Isothermal oxygen-isotope exchange of LaFeO₃, ¹⁶O₂ is from bulk lattice, ¹⁸O₂ is gas phase, and ¹⁶O¹⁸O is surface combined scrambled isotope product ⁷⁹

The oxygen-isotope exchange technique is a powerful combination of isotope diffusion and isotope depth profile characterisation, and it allows for diffusion kinetics to be determined for specific markers in various materials. It makes use of isotope-specific nuclear reactions capable to distinguish subtle variations in concentrations.

For the purpose of understanding, the TiO₂ surface properties, partial replacement of ¹⁶O with ¹⁸O, or the use of D₂ ¹⁶O, or D₂ ¹⁸O instead of H₂ ¹⁶O may offer advantages in monitoring surface adsorption of water.

For example, the bridging oxygen in Fig. 24, removed by ion bombardment or by thermal treatment, may be replaced by ¹⁸O instead of ¹⁶O, from a gas phase or from H₂ ¹⁸O. Surface and sub-surface ¹⁸O can then be detected by elastic recoil detection analysis (ERDA) with low energy Ne⁺ as projectile, or by the use of specific nuclear reactions such as ¹⁸O(p, α)¹⁵N using 845 keV proton irradiation.

OPEN IN VIEWER

24.

Schematic representation of defect-free TiO₂ (110) surface

Implantation

Implantation is the process of bombarding the solid with specific ions, which possess the kinetic energy that is required to penetrate the surface and come to rest at a particular depth inside the target (crystalline or amorphous). In this process, the original chemical composition is changed and some atoms of the crystal lattice will be recoiled from their equilibrium position. Therefore, implantation cannot be considered as doping in terms of the formation of solid solutions that can be obtained using either solid-state reactions or soft chemistry routes. Implantation results in the formation of systems that are remote from thermodynamic equilibrium. The properties of implanted solids depend on the selected ions and their energy used for implantation. As a result, thermodynamics cannot be applied to describe the properties of solids formed by implantation. However, the post-implantation annealing may be used to gradually transfer the system from its non-equilibrium state to a system that can be equilibrated at the local atomic level. In this process, the implanted ions may diffuse inside the target or may diffuse and coalesce into nanoparticles, under the target surface or on the surface. ⁸⁰ Therefore, the post-implantation annealing process may be used to form systems that are partially equilibrated, and which present characteristic properties related to both implantation and chemical doping.

Ions of most elements, and of some molecular species, can be implanted at both low (few keV) and medium energy (few MeV). Ion implantation is a very versatile and a powerful tool for surface and near-surface modifications of materials mostly because this is a non-equilibrium technique, and thus it can be used to implant any ion in any matrix. The choice of energy is a function of the mass of implanted ion, the required implantation depth and the target properties. For a particular ion, the lower the energy, the shallower the implantation depth, but below a specific threshold energy, the ion is no longer able to penetrate the energy surface barrier of the target, and comes to rest on the target surface. For some ion-energy-target combinations, this results in sputtering of target atoms, associated with surface damage. If the ion energy is further reduced, it may result in a deposition process. At the higher end of implantation energy, sputtering of the target surface may be the dominant process. Implantation offers precise control of composition and when it is carried out through masking or with a focused ion beam, it also offers precise control of location on the surface of the sample.

A variation of this technique, called ‘micro-implantation’, is achieved when the ion beam is focused down to a micrometre-size spot. If the ion current is sufficiently high, for some target materials this leads to micro-implantation or micro-sputtering and can be used to create implanted or damaged islands, or to ‘write’ complex shapes on the surfaces. ⁸¹

There are numerous examples of the use of ion implantation for materials, such as semiconductors, corrosion resistant materials, polymers, glasses, etc. In the case of oxide semiconductors, implantation offers the capability to change the surface composition and defect disorder. In particular for TiO₂-based POSs, shallow implantation is one of the few ways to control the surface and sub-surface doping. ⁸² This leads to controlled changes in chemical composition at the surface and sub-surface regions, with the direct effect of electronic structure modifications. In addition, surface and sub-surface doping can lead to nanometre cluster formation on the surfaces and the enhancement of plasmonic effects.

In many cases, photocatalytic materials obtained by ion implantation are called second generation photocatalytic materials. In the case of thin films of polycrystalline, TiO₂ made by ion cluster beam (ICB) on glass, ^83,84 and implanted by V⁺, Cr⁺, Mn⁺, and Fe⁺, at 150 keV and a dose of 10¹⁶ and 6×10¹⁶ cm⁻², the UV-VIS transmittance show a shift towards higher wavelength with the increase of implanted dose. XAFS indicated the replacement of Ti⁴⁺ by V³⁺, V⁴⁺, and ab initio DFT calculation showed a mixture of Ti(d), V(d) and O(d) orbitals, which suggest an explanation for the red shift.

In a similar study, ⁸⁵ Cr and V ions implanted in TiO₂ thin film show two to three times higher photocatalytic activity than the analogous non-implanted material, but the film thickness and the substrate temperature during the film deposition play a critical role in UV-VIS absorption. In the case of Cr implantation, it was shown that it also led to a decrease in the band gap and to absorption of visible light by TiO₂. It was concluded that implanted Cr does not form recombination centres unlike chemically doped Cr. Also, V, Mn, Ni, Ar, Mg, Ti, and Fe were implanted in TiO₂, and the red shift achieved increased in the following order: V>Cr>Mn>Fe> Ni, after annealing at 723–823 K in O₂. Molecular orbital calculations showed that V³⁺, V⁴⁺, and V⁵⁺ substitute for octahedral Ti⁴⁺, and therefore Ti(d) and implanted ion-metal (d) orbital mixing led to narrowing of the band gap. Preliminary investigations show that ion implantation is advantageous as compared to other semiconductors and dye sensitised materials in which the e⁻ mediator and sacrificial agents are necessary. This is not the case for implanted TiO₂.

In the same study, ⁸⁵ polycrystalline TiO₂ and WO₃, loaded with metals (Pt, Au, Pd, Rh, Ni, Cu, and Ag) and/or doped with metal ions (Fe, Mo, Ru, Os, Re, V, and Rh) lead to an increased photocatalytic activity. Also, out of a range of elements (La, Ce, Er, Pr, Gd, Nd, and Sm), Gd was the best in increasing the photocatalytic activity. Doping with Be at near the surface is beneficial, but when Be was implanted deeper the benefit is lost. In the case of WO₃, doping with Ni achieved better photocatalytic activity than doping with Fe, Co, Cu, Zn, and Mn. It was shown that Cu, Mn and Fe can trap both e⁻ and h⁺. Anion doping of polycrystalline TiO₂ with N, C, F, and S can also narrow the BG, and is less likely to form recombination centres (unlike metal ions), and so be more effective. C and P were less effective as the introduced states were too deep, and photo-generated charge carriers were difficult to be transferred to the surface. Sulphur has shown to provide S 3p states mixed with valence band of TiO₂ by shifting upwards, with no change in the width of conduction band, and consequently narrowing the band gap. Similar behaviour was observed for N-doping.

Ion implantation is one of the most promising technologies for improving the properties of TiO₂ because it allows for a high impurity filling factor, beyond the equilibrium limit of impurity, solubility, and also provide control on the nanoparticles formed at different depth levels. The anatase phase appears to be better than rutile because of native higher band bending, deeper under the surface, and with a steeper gradient, enhancing the trapping of surface holes. ⁸⁶ In rutile, the e⁻/h⁺ recombination is pronounced in the bulk regions, so only a smaller fraction of holes are trapped and transferred to the surface. The low ion irradiation dose of less than 10¹⁴ cm⁻² results in well dispersed implanted ions. For the dose ranging between 10¹⁵ and 10¹⁷ cm⁻², the implanted ions exceed the solubility limit, and form nano-particles (NP) by nucleation and growth. Low implantation energies between 10 and 300 keV is of higher interest because of the proximity to the surface of the implanted region. Implantation with Fe, V, Cr, Mn, Co, Ni, and Cu was effective in the red shift of TiO₂, while Ti, Ar, and Na was not. From XAFS, Fe³⁺ appears to replace Ti⁴⁺ and the same for V and Cr. Also, implanted Nb and Sn are effective for increasing the matrix conductivity, in particular Sn, which is substitutional up to ∼1 at.%, when increases conductivity very high σ = 5–30 (Ω cm)⁻¹, compared to typical σ = 1 (Ω cm)⁻¹, obtained by conventional doping. Implantation with Cr at room temperature increases σ from 10⁻¹³ to 10⁻² (Ω cm)⁻¹. ⁹⁶

Synthesis of magnetic nano particles MNP for optic and magnetic properties by ion implantation is very attractive because the dimension and structure of the MNP can be well controlled. Au, Ag and Cu show strong surface plasmon resonance (SPR) in visible or near IR range for TiO₂ implanted with 2 MeV Au at RT. Also, TiO₂ implanted with 40 keV Ag at 10¹⁵–10¹⁶ cm⁻² shows poly-colour photochromism and antibacterial properties. ⁹⁶

Thin films of TiO₂ grown by reactive magnetron sputtering to 75 nm thick, were doped by ion implantation with Au at 150 keV, and a dose between 5×10¹⁵ and 10¹⁷ cm⁻². At 5×10¹⁶ cm⁻², Au nanoclusters of 3–5 nm diameter by X-ray diffraction (XRD) start to form in the film at RT, and during annealing at 350°C in reduced atmosphere, the nanoclusters start to coalesce, following an Oswald ripening mechanism. When annealed at 600°C, the precipitate dimensions reach 40 nm (all by in situ XRD), and annealing above 700°C promotes the formation of two Au-rich regions, at the interface and at the surface, shown by He-RBS. ⁹⁷

Rutile (110) single crystals were implanted with Er at 40, 80, 150, 200, and 350 keV, at doses of 10¹⁴, 1·5×10¹⁴, 2×10¹⁴, 1·5×10¹⁴, and 6×10¹⁴ cm⁻², respectively, [100]. Some samples were annealed in Ar for 30 min at 800, 900 and 1000°C. Non-Rutherford resonant scattering/channelling with 3·05 MeV He along <110>, showed no detectable crystal damage, and Er diffuse a little above 800°C, suggesting an interstitial position [100].

Irradiation of surfaces with ions implies energy transfer from ions to the target atoms, which can lead to displacement of target atoms, associated with the formation of defects, and this process is called radiation damage. The (110) structure of rutile is shown schematically in Fig. 24, where the two characteristic positions for O (in red) are visible: the in-plane and the out-of-plane locations. Ion irradiation can more easily remove the out-of-plane (bridging) surface oxygen, forming defects, and introducing an inter-band donor state, below the Fermi level, with a Ti-3d character, leading to the formation of n-TiO₂. Heat treatment in reduced oxygen partial pressure may have a similar effect.

In well defined experiments, energy loss mechanisms governing the ion-target interaction are responsible in producing point defects or defect clusters, and this defect engineering may be a useful tool in enhancing specific properties of materials, such as surface reactivity (including photocatalysis), magnetic flux pinning and others.

In the case of non-stoichiometric oxides, implantation may be used to drive the deviation from stoichiometry in the desired direction. For example, implanting the host metal cation (self-implantation) results in an increased concentration of oxygen vacancies and a decrease in the concentration of metal vacancies. However, the implantation process results in destruction of the crystal lattice. Restoration of the crystalline structure requires post-implantation annealing.

A polished polycrystalline TiO₂ surface, annealed in air at 1273 K, Fig. 25, shows a clear delimitation of grain boundaries and the annealing re-construction after the polishing damage of the surface of individual grains with various orientations, indicating the re-constructed growth steps. This surface was implanted with Ti (self-implantation) and Pt ions at 50 keV and a dose of 1×10¹⁶ at cm⁻², Figs. 26 and 27, respectively. As compared with Fig. 25, Fig. 26 shows fewer details of grain boundaries and growth steps. Figure 27 shows barely visible grain boundaries and no growth steps. These differences can be explained in terms of damage to the crystal lattice during ion implantation. In Fig. 28, the calculated displacement damage in displacements per atom (DPA) using a Monte-Carlo based simulation programme (stopping range of ions in matter, SRIM) is shown, ⁸⁹ and it indicates that the displacement damage created by Ti is spread over a larger distance from the surface with a peak of 1·8 dpa, at 40 nm, while the damage created by Pt is higher and concentrated closer to the surface, with a peak of 4·5 dpa at 15 nm under the TiO₂ surface. This suggests that the differences in the appearance of the SEM micrographs shown in Figs. 25 and 26 may be explained by the higher number of vacancies and their proximity to the surface, and represent the different degree of surface damage imparted to the TiO₂ crystals surface by Ti and Pt implantation at the same energy and dose.

OPEN IN VIEWER

25.

The SEM micrograph for pure TiO₂ surface after polishing and annealing at 1273 K in air (reproduced with permission from Ref. 10. Copyright T. Bak, 2010)

OPEN IN VIEWER

26.

The SEM micrograph for pure TiO₂ surface after polishing, annealing at 1273 K in air and subsequent implantation of titanium (50 keV at a dose of 1×10¹⁶ at cm⁻²,); (reproduced with permission from Ref. 10. Copyright T. Bak, 2010)

OPEN IN VIEWER

27.

The SEM micrograph for pure TiO₂ surface after polishing, annealing at 1273 K in air and subsequent implantation of platinum (50 keV at a dose of 1×10¹⁶ at cm⁻²); (reproduced with permission from Ref. 10. Copyright T. Bak, 2010)

OPEN IN VIEWER

28.

Calculated displacement damage of TiO₂ by implantation of Ti and Pt with 50 keV at 10¹⁶ at cm⁻² (reproduced with permission from Ref. 10. Copyright T. Bak, 2010)

Rutherford backscattering

In RBS, the probe ions of a few million electron volts, typically below the coulomb barrier for target atoms, are directed towards the sample surface at an angle between 0° and 90° relative to the sample normal. When a probe ion encounters a heavier sample atom at, or below the sample surface, a precise amount of kinetic energy will be exchanged, and the probe ion will be backscattered with a precise amount of kinetic energy, measured by a detector. This measured backscattered energy, together with the initial energy of the probe ion and the experiment geometry, are used to calculate the mass of the target atom and the location of the collision relative to the sample surface. For these reasons, RBS is a standard-less technique, and in particular H-RBS and He-RBS, is a quasi-non-destructive technique for characterising the surface, and near-surface regions. It provides the depth distribution of elements and the distribution of impurities. This can reveal additional information on the condition of interfaces and the thickness of thin films and of implanted layers. Under ideal conditions, a depth resolution of up to 0·05 ML may be achieved, for example, for Au evaporated on cleaved Mica in UHV.

The thickness of thin films determined by RBS differs from that provided by the optic-based techniques, as it measures the number of atoms per unit area of the beam and is therefore independent of the sample density. The flexibility in the choice of ions, energies and the experiment geometry offers the possibility to maximise the signal from a specific element, but the method is most suitable for analysing thin films and surfaces composed of elements heavier than the substrate or matrix, respectively. In addition, under specific circumstances, this technique offers information on multi-layers and on interfaces, such as inter-diffusion.

For TiO₂ implanted with elements heavier than Ti, RBS can offer non-destructive depth distribution of the implanted ion, and therefore, the measured samples can be further used for other characterisation work, such as photocatalytic properties, thus reducing the necessity to infer identical properties for different samples characterised by the same techniques.

Elastic recoil detection analysis is another technique for the characterisation of the surface and near-surface regions, and it differs from RBS only in its experimental geometry. In ERDA, the probe ions, which are heavier than the matrix elements, are directed at the sample surface under a large angle (60–85°) from the sample normal. Thus, the heavier projectile having an adequate kinetic energy recoils the sample atoms and their energy is measured. The sample damage in this case can be higher than in RBS, but the beam flux is very small and a number of repeated examinations of the same area still provide indistinguishable results. The advantage is a complete depth resolution analysis for light, as well as for heavy elements. In particular, when the energy detection is carried out with a time-of-flight (ToF) detector, the depth resolution can be greatly enhanced to 1–10 nm. ⁹⁰

Elastic recoil detection analysis and RBS can be carried out in situ or ex situ, in conjunction with ion implantation or heat treatment procedures. In Fig. 29, the result of in situ RBS on (110) TiO₂ single crystal annealed in oxygen at 900°C and in Ar at 600°C is shown. This shows an unchanged oxygen stoichiometry for the as-grown and oxygen-treated sample. However, for the Ar-annealing case, the oxygen depletion is readily measurable, which allows the oxygen stoichiometry and the depth distribution of oxygen depletion in the sample under the surface to be determined.

OPEN IN VIEWER

29.

Rutherford backscattering (RBS) with 2·5 MeV protons on ‘as-grown’ (110) TiO₂ single crystal, and on the same crystal annealed in O₂ at 900°C and Ar at 600°C (Copyright M. Ionescu, 2013)

Ion channelling

Ion channelling is a variation of the RBS technique, described above. The main difference consists in the direction of primary beam of probe ions, which in this case is almost exclusively He, directed along specific crystallographic directions. For single crystals, under specific orientation conditions, the direction of probe ions may be aligned with specific crystallographic channels in the sample. In that case, He probe ions can penetrate along the channel and encounter a much smaller number of the target atoms with which to collide, and thus He probes the channel integrity. If the channel is distorted as a result of the presence of interstitial atoms, or defects, the He probe ions will collide with a larger number of target atoms, and will be back scattered more often than otherwise. This may lead to the possibility of extracting additional information on the sample surface orientation, the interstitial atoms and vacancies. ⁹¹

For example, for a TiO₂ single crystal implanted with an ion heavier than Ti, if channelling is carried out after ion implantation, it may reveal the type of implanted ion, its depth distribution as well as the damage that the implanted ion has created and the depth of the damage. In Fig. 30 is shown the main crystallographic directions in P4₂/mnm system of rutile, looking down along the [001] direction. If the single crystal surface is represented by the (110) plane, as in the previous example, then for a channelling experiment, the He probe ions will be directed along [100] direction, and the detector normal will be along [010] direction. In this configuration, the crystallographic channels parallel to these directions will be probed for distortions and interstitials.

OPEN IN VIEWER

30.

Schematic representation of basic crystallographic directions in TiO₂

In Fig. 31, the results of channelling on (100) Si implanted with Ru at various equivalent concentrations are shown, measured with 1·8 MeV He⁺. For comparison, the same figure also shows the He-RBS of a randomly oriented Si and of a (100) oriented Si crystal before Ru implantation. These results indicate that not only the Ru concentration and its depth distribution but also the associated Si damage peak, and its depth distribution, noting the direct correlation between the Ru concentration and the level of damage.

OPEN IN VIEWER

31.

Channelling Rutherford backscattering (RBS) on (100) Si implanted with Ru, using 1·8 MeV He⁺ (Copyright M. Ionescu, 2013)

Nakamura et al. ⁹² reported the effect of indium implantation at 100 keV and 5×10¹⁴ and 1×10¹⁵ cm⁻², at 7° from the [100] and [001] directions of TiO₂ single crystal (rutile). Channelling RBS in [100] direction showed that there is only one Ti peak, but in [001] two Ti peaks were seen. This was interpreted as irradiation damage (higher energy) and the presence of In (lower energy). Annealing at 650 K gives a small recovery while annealing at 973 K gives a large recovery and a migration of In atoms towards the surface. The ψ scan around [001] and [100] indicate 88% In substitution for Ti up to 5×10¹⁴ cm⁻², and 11% of In sitting in interstitial positions along c-axis at the sites (0, 0, 1/4). With increasing dose to 5×10¹⁵ cm⁻², In precipitate in clusters, with 3% substituting Ti, 15% interstitial at (0, 0, 1/4) and 82% random position, but with annealing at 973 K, the In distribution was 20% substitution and 80% random. The increase of In interstitial occupancy after implantation and its disappearance after annealing suggest that In is interacting with defects resulting from implantation, which could be in the form of: indium–titanium vacancy pair; indium-Schottky defect complex; indium–oxygen divacancy complex; and indium–titanium mixed dumbbell.

Elastic recoil detection analysis

As mentioned in the RBS section, ERDA differ from RBS in its experimental geometry, with the direction of the incident probe ions and the detector normal forming a large angle (60°–85°) from the sample normal. The advantage of this configuration is that the heavy and energetic projectiles used can recoil all the sample atoms, and thus the complete sample composition, from the lightest (H) to the heaviest element present, can be determined, as well as their depth distribution, with a depth resolution between 1 and 10 nm.

In Fig. 32, the ERDA result using 82 MeV I⁺¹⁰ probe ions, for a LiNbO₃ thin film deposited by CVD on (100) Si, and a time-of-flight (ToF) detector is shown. On the right the schematics of ERDA experimental geometry and on the left the detection result of the (ToF, energy) plane are shown.

OPEN IN VIEWER

32.

Typical (t-E) elastic recoil detection analysis (ERDA) result obtained with 82 MeV I⁺¹⁰ probe ions for a LiNbO₃ thin film deposited on (100) Si, and a time-of-flight (ToF) detector; on right is experimental geometry (Copyright M. Ionescu, 2013)

This result indicates that the depth distribution of all elements present in the LiNbO₃ film, including the Li isotopes, the thickness of the film (∼100 ML), the presence of a SiO₂ film (∼6 ML thick) at the interface, and the diffusion of Nb in Si substrate.

Nuclear reaction analysis

When the energy of the probe ions is higher than the coulomb screening barrier of the target atoms, a specific nuclear reaction can be triggered between the probe ion and the target atomic nucleus. There is a large number of nuclear reactions, and specific nuclear reactions can be used for surface and near-surface characterisation, when, for example, the direct measurement of a particular element is difficult to achieve through other techniques. Light elements (up to Al) usually fall in this category, and nuclear reactions can yield prompt reaction products. In particular, nuclear reactions with narrow resonances and γ-ray products are a first choice for depth profiling of light elements, with a depth resolution of around 10 nm.

NRA can be used for both crystalline as well as for amorphous samples, and in specific cases, it can be combined with the isotopic exchange kinetics, representing a powerful combination of isotope diffusion and isotope depth profile characterisation. An example is presented in Fig. 33 of the use of the¹⁵N(H, αγ) ¹²C nuclear reaction for the depth profiling of hydrogen in DLC thick films. This nuclear reaction has a narrow reaction resonance of 1·8 keV, and as the energy of probe ¹⁵N ions of 6·4 MeV collide with the surface H, prompt γ photons of 4·43 MeV are created and detected. As the energy of the probe ions is increased above 6·4 MeV, they penetrate the sample surface, and start losing their kinetic energy. When this energy decreases down to 6·4 MeV, and ¹⁵N collide with another H, a new reaction is triggered and the new γ-ray photon is detected. For this nuclear reaction, the depth of analysis range from a few nanometres, up to a few hundreds of nanometres, with a depth resolution of 5–20 nm.

OPEN IN VIEWER

33.

Nuclear reaction analysis (NRA) of H in DLC thick films (Copyright M. Ionescu, 2013)

Particle-induced X-ray emission

When the probe ions collide with target atoms on or under the surface, one possible consequence is the ionisation of the target atoms, and the subsequent emission of characteristic X-rays. Proton-induced X-ray emission is a surface and near-surface analysis technique, and is suitable for analysis of elements heavier than Al. The depth of analysis is dictated by the type of the sample matrix, the energy of probe ion and the type of probe ion. The analysis is usually carried out with protons between 1 and 3 MeV, but for specific samples other probe ions such as He or C may be used.

The characteristic X-rays are detected, and the quantitative analysis is achieved by a calibration procedure with known standard samples. Optimum detection sensitivity is near Fe (1 ppm), and other element sensitivities range from 1 to 100 ppm, which is about 100 times better than EDX, and is mainly because of a much lower background. For proton PIXE, the typical depth of analysis ranges from 20 to 50 μm, but the depth information is absent. ^93,94

Proton-induced X-ray emission can be carried out with a broad ion beam or with a focused ion beam. In the latter case, the ion beam is focused down to micrometre, or even sub-micrometre, size and rastered over area to be analysed, which confers the additional capacity to extract spatial information on the surface distribution of a particular element, with a lateral resolution of less than 0·5 μm. In this way, surface X-ray mapping of element distribution can be measured.

Proton-induced X-ray emission is a powerful tool in determining the concentration of a wide range of secondary elements, which are added intentionally (doping) or unintentionally (impurities). Since the presence of aliovalent ions already at the level 5–10 ppm results in substantial changes in properties, the characterisation of POSs should include the determination of the impurities.

Positron annihilation lifetime spectroscopy

The positrons emitted from a radioactive source such as ²² Na can penetrate the surface of a conductor or an insulator and quickly thermalise. When the positron energy falls down to 511 keV, they annihilate with the free electrons existent in the material, or can undergo a spin coupling with electrons and form two quasi-particles, para-positronium and ortho-positronium. In all cases, the annihilation reaction is producing the characteristic back-to-back γ-ray photons of 511 keV, but the ortho-positronium lifetime is approximately 140 times longer. If defects (vacancies, pores) are present in the material, then the lifetime of ortho-positronium will further increase, and a correlation can be established between the additional increase in the lifetime and the size of the defects.

A refinement of this technique consists of the use of a micro beam of pulsed positrons for enhanced lateral resolution which afford the possibility to measure both the lifetime and the Doppler broadening of the annihilation reaction, and produce 2D defect maps. ⁹⁵

An example is taken from Ref. 95, where fluorine was implanted in TiO₂ single crystal at 200 keV, and at doses between 10¹⁶ and 10¹⁷ cm⁻², and after annealing in air at 1200°C, SIMS shows fluorine migration towards the surface. Channelling RBS with 2 MeV He shows damage, and XPS suggest F replacing O. When PALS was performed with positron energies between 0·2 and 25 keV, it measured the broadening and S parameter of the annihilation response before and after implantation and recovery annealing, which indicated the creation of vacancies. ^95,96

Summary

The nuclear techniques reviewed here can offer major benefits in advancing the technology of POSs, from sample preparation to sample characterisation. They are adequate for POSs requirements of surface and near-surface modification and characterisation, and can be beneficial in tailoring the POSs properties bringing these materials closer to industrial applications.

The use of nuclear techniques requires the building and maintenance of major facilities, and therefore their cost can be high. In some cases, the need of high energies requires the use of accelerators or the use of radioactive sources. In addition, the use of nuclear techniques and facilities require specific expertise, and are subject to strict safety regulations, and these are the main reasons why these techniques are mainly concentrated at national facilities.

The nuclear techniques are frequently associated with the use of isotopes, which are required in order to distinguish the tracer particle from the background formed by the same elements.

While the nuclear techniques are frequently considered as hazardous if related to radioactivity, the present work represents an example of the application of nuclear techniques for protection of the environment through the development of a new generation of solar materials that can be used for solar cells for the production of environmentally friendly fuel and for water purification.