A Surface Photovoltage Study of Surface Defects on Co-Doped TiO₂ Thin Films Deposited by Spray Pyrolysis

Abstract

Surface photovoltage (SPV) spectroscopy is a powerful tool for studying electronic defects on semiconductor surfaces, at interfaces, and in bulk for a wide range of materials. Undoped and Cobalt-doped TiO₂ (CTO) thin films were deposited on Crystalline Silicon (c-Si) and Flourine doped Tin oxide (SnO₂:F) substrates by chemical spray pyrolysis at a substrate temperature of 400 °C. The concentration of the Co dopant in the films was determined by Rutherford backscattering spectrometry and ranged between 0 and 4.51 at %. The amplitude of the SPV signals increased proportionately with the amount of Co in the films, which was a result of the enhancement of the slow processes of charge separation and recombination. Photogenerated holes were trapped at the surface, slowing down the time response and relaxation of the samples. The surface states were effectively passivated by a thin In₂S₃ over-layer sprayed on top of the TiO₂ and CTO films.

1. Introduction

Surface photovoltage (SPV) spectroscopy is a non-destructive and fast technique for investigating optoelectronic properties at surfaces, interfaces, and in bulk by monitoring the illumination-induced changes in the surface voltage [1]. It provides important information, such as surface band bending, surface and bulk carrier separation and recombination, surface state distribution, and defect states by studying the photo-induced carrier separation and transfer in semiconductor materials. SPV has been used to study nanocrystalline semiconductors for various applications, and has recently received renewed attention for the study of integrated devices and micro-heterogeneous systems [2,3,4].

Doping of metal oxides with transition metals has led to the investigation of new materials with ferromagnetic properties for application in Spintronics devices, photocatalytic applications, as well as for energy conversion and storage [5,6,7,8]. Co dopant is optically active, imparts magnetic properties to TiO₂ suitable for Spintronics device application, and enhances photocatalytic activity through the introduction of additional electronic defects in its lattice [9,10,11,12]. A few Co atoms are sufficient to create specific defect structures that trap charge carriers and induce ferromagnetic properties in TiO₂ [11]. The enhancement of photocatalytic activity of Co-doped TiO₂ (CTO) is associated with the red shift of the band gap due to the sp-d exchange interaction between band electrons and localized d electrons of the Co²⁺ substituting for Ti⁴⁺ cations, which shifts the conduction band downwards and the valence band upwards [13,14]. The red shift has also been associated with an increase in the density of defects, resulting in the formation of impurity bands, which may spread and overlap with the conduction band [5].

Anatase TiO₂ consists of interconnected TiO₆ octahedron, each in contact with eight neighbors via four corners and four edges. Local distortion in the octahedron induces energy levels below the conduction band. Incorporation of Co—which has a larger ionic radius than Ti—introduces additional distortions, increasing the density of defects [15]. Co²⁺ dopant substitutes for Ti⁴⁺, creating additional oxygen vacancies, which leave behind a lone electron pair. The electron pair reduces two Ti⁴⁺ ions to Ti³⁺. The oxygen defects are formed on the surface and in the bulk on TiO₂. The vacancies in the bulk are less stable and may migrate to the surface. The oxygen vacancies, which act as electron donors, can be associated with both deep and shallow defects. The enhanced paramagnetic properties in Co-doped TiO₂ have been linked to increased density of oxygen vacancies [9,10,11,12,13,14,15]. Co²⁺ defects are also introduced in the mid-gap near the oxygen and Ti³⁺-related trap states. Trap-to-trap transition of electrons can take place from a defect to a Co²⁺ state, from where they relax to ground state [10].

The electronic, structural, and electrical properties of thin films depend on the deposition methods. CTO films and nanostructures have been deposited by sol–gel [9,10], ball milling [14], spray pyrolysis [16,17], solvo-thermal [15], pulsed laser deposition [18,19], molecular beam epitaxy [20], reactive co-sputtering [21], and atomic layer deposition [22]. Chemical spray pyrolysis has the advantage that it is vacuum-free, allows for a wide variety of precursor choices and deposition parameters, and different precursor solutions can be mixed in different proportions before deposition to effectively dope the samples.

In this work, CTO thin films were prepared by spray pyrolysis from precursors with different compositions of Ti and Co salts. The composition was analyzed by Rutherford backscattering spectrometry (RBS), while optoelectronic properties were investigated using UV-Vis and SPV techniques. SPV is a photoelectric method that is very sensitive to changes in electronic states participating in charge separation, independent of whether it is at the surface, interface, or in the bulk [23].

2. Materials and Methods

To prepare the precursor solutions, 19.2 mL of Titanium isopropoxide (TTIP) was mixed with 28.8 mL of 2,4-Pentanedione solution. Absolute ethanol solvent was added and stirred to make 480 mL of solution. This was called solution A. Solution B was prepared by dissolving 43.2 mg of anhydrous CoCl₂ in 50 mL of acetonitrile and stirring for 3 hrs. Different volumes of solution B were added to solution A in volumetric percentages varying from 0% to 50%. TiO₂ films were obtained from solution A and Co-doped TiO₂ (CTO) thin films from the mixture of solutions A and B by spraying for 5 min at a substrate temperature of 400 °C. RBS experiments were performed using a 2.0 MeV He⁺ ion beam at zero incidence and exit angle of 12°. The backscattered ions were detected at a scattering angle of 168°. The charge was 5 coulombs with measurement time of about 12 min. The optoelectronic defects were investigated by modulated surface photovoltage in the fixed parallel capacitor configuration [24,25], with the incident light modulated at 8 Hz. The SPV signals were detected using a high impedance buffer and a lock-in amplifier.

3. Results

3.1. Compositional Analysis

Figure 1a shows the RBS spectra for the TiO₂ and CTO thin films deposited on Crystalline Silicon (c-Si) substrates. The RBS signals from Co, Ti, Si substrate, and O were identified as indicated in the figure. The channel position of these signals depends on the atomic number of the elements detected. The heavier elements appear at higher channel numbers, corresponding to higher energy of the backscattered He⁺ ions after collision with the nucleus of that specific atom. The height/yield of the RBS signal is proportional to the amount of material present in the sample, and it depends on the scattering cross-section of the detected element. The width of the peak is proportional to the thickness of the probed sample and depends on the energy loss per unit depth of the detected element [26].

A decrease in the width of the RBS signals from Ti and Co and a shift of the signal from the surface of the c-Si substrate to higher channels was observed with an increase in the amount of CoCl₂ in the precursor solution. A decrease in the width of the RBS signal is concomitant with a decrease in the film thickness. The height of the Co signal increased as a function of increasing amount of solution B in the precursor, which confirms the increase in the amount of Co in the CTO film. Figure 1b shows the Ti and Co signals from the RBS spectra fitted using Spewa software (Version 1.0, Institute of solid state Physics, Jena, Germany) based on the Nuno’s Data Furnace (NDF) code [27].

The thicknesses and atomic concentrations of the TiO₂ and CTO thin films were extracted from fitting the spectra and from the integrals of the Ti and Co signals. The thickness for undoped TiO₂ was 175 nm, and for the CTO films it ranged between 175 nm for 5 vol % and 60 nm for 50 vol % of CoCl₂ in the precursor solution, as given in

Table 1. Thin film thickness and Co dopant concentrations as a function of the CoCl₂ vol % in the precursor solution.

CoCl2 vol %	Thickness (nm)	Co (at %)
0	175	0
5	175	0.43
10	140	0.60
20	140	1.25
30	105	1.50
40	90	2.40
50	60	4.51

. The film deposition rate was dependent on the composition of the precursor solution, and decreased with an increase in the amount of CoCl₂ in the solution. The Co atomic concentrations in the thin films obtained from fitting the RBS spectra as shown in Figure 1b amounted to between 0.43 at % and 4.51 at % for CTO films deposited from solutions containing 5 vol % and 50 vol % of CoCl₃, respectively, as presented in Table 1.

3.2. Modulated Charge Separation

The electronic defects in TiO₂ and CTO thin films deposited on SnO₂:F substrates were investigated by studying photogenerated charge separation under chopped light. In modulated SPV experiments, four signals are normally recorded using a lock-in amplifier. The in-phase (X-signal) and the phase shifted by 90° (Y-signal) with respect to the reference modulated signal gives information about the direction of separation of the photogenerated charge carriers and the time response of the sample to the modulation of the incident light. The spectral dependent SPV amplitude is the square root of the sum of the squares of the X and Y signals, while the phase angle is the inverse tangent of the ratio between the Y and X signals [27,28].

Figure 2a shows SPV amplitudes for the TiO₂ and CTO thin films doped with 0.43, 0.60, 1.25, and 2.40 at % of Co. The onset energy of the signal was 0.96 eV for all samples (which is very low), compared to the absorption edge of TiO₂ which has a band gap of 3.2 eV. The onset energy of the SPV signal is characteristic of the mobility edge and is expected to be close to the band gap energy for an ideal semiconductor. The SPV signal was therefore dominated by charge separation below the TiO₂ band gap, which indicates the possibility of there being a high density of sub-band gap defect states.

The amplitude signal shows two peaks centered at 1.4 and 1.9 eV. Such peaks cannot be related to absorption peaks or peaks observed in spectroscopy, because the generation of an SPV signal could be a net result of several charge separation and recombination processes [28]. The two peaks therefore relate to energetic sub-band gap defects participating in charge separation and relaxation processes. The magnitude of the SPV amplitude at 1.9 eV increased from 0.036 mV for undoped TiO₂ to 0.13, 0.32, 0.40, and 0.70 mV for dopant concentrations of 0.43, 0.60, 1.25, and 2.40 at %, respectively, while the film thickness decreased. The SPV amplitude is proportional to the number of photogenerated charge carriers separated in space, and the perpendicular distance between them. The magnitude of the SPV signal was not proportional to the film thickness, so it follows that the increase in the SPV amplitude was not due to an increase in the charge separation distance, but probably due to an increase in the density of the photogenerated charge carriers.

If the charge separation is caused by only one mechanism and the relaxation is due to recombination, the X and Y signals will have opposite signs [24,25,28], as shown in Figure 2b. The X-signal is positive, while the Y-signal is negative, indicating preferential separation of photogenerated electrons towards the substrate and holes towards the top surface. The X signal increased from 0.03 mV to 0.20 mV when the Co concentration in the film was increased from zero to 1.50 at %, but then dropped to 0.10 mV for higher Co amounts. There was no direct relationship between the Co concentration and magnitude of the X signal.

The absolute magnitude of the Y signal increased from 0.01 mV for undoped TiO₂ to 0.11, 0.23, 0.32, and 0.67 mV when the amount of Co was increased to 0.6, 1.25, 1.50, and 2.4 at %, as shown in Figure 2b. This shows a direct correlation between the magnitude of the Y signal and the Co concentration in the CTO films. It can be concluded that the systematic increase in the SPV amplitude observed in Figure 2a with the increase in Co was mainly due to the concomitant increase in the Y signal. The Y-signal characterizes the phase shift between the SPV signal and the chopped light, and it corresponds to the slow increase or decrease of the SPV signal in relation to the half period of modulation [28]. The increase in the Y signal implies that the charge separation and/or relaxation were much slower than the modulation period of the chopped light. It follows that an increase in the Co dopant in the films slows down the charge separation and recombination processes.

Incorporation of Co ions introduces distortions in the octahedron of the TiO₂ structure due to the difference in the ionic radii of Ti (0.605 Å) and Co (0.745 Å), enhancing the density of defect states and the concentration of the photogenerated charge carriers in the different trap levels [15]. Photoluminescence studies have shown that Co doping of TiO₂ increases the lifetime of photogenerated carriers by introducing mid-band gap trap states associated with d states of cobalt, where an increase in the relaxation time of conduction electrons is observed before it undergoes recombination [10]. This explains the increase in the Y-signal with increasing Co dopant discussed above.

The phase angle of the SPV signal gives information about the direction of charge separation and the time response of the sample in relation to the period of modulation. For example, a positive X signal and a phase angle between 0° and −90° implies a single charge separation process with photogenerated electrons separated preferentially towards the internal interface [29]. In Figure 2c, the phase angle was observed to shift from −30° for undoped TiO₂ to −48°, −60°, and −80°, with an increase in Co concentration to 0.6 at %, 1.25 at %, and 2.40 at %, respectively. A shift of the phase angle towards −90° means that the charge separation and/or relaxation processes became slower with increasing Co concentration [29,30,31,32].

The phase angle was independent of the photon energy, and so it can be concluded that in all samples, there was only one transport mechanism involved with charge separation and recombination processes within this spectral range. The shift in the phase angle corresponds to the increase in the Y-signal to more negative values, as discussed above (Figure 2b). The response times of the CTO samples to modulated light were increased by Co-doping. This could be due to enhanced trapping of the photogenerated holes in surface defect states.

Figure 3 shows the SPV signals of the bare In₂S₃ and CTO/In₂S₃ multilayer as a function of the Co concentration in the thin films. The onset of the SPV signal was 2.0 eV for bare In₂S₃ and 1.92 eV for TiO₂/In₂S₃ and all CTO/In₂S₃ samples, independent of the Co concentration. The optical band gap of Ion Layer Gas Reaction deposited In₂S₃ thin films similar to those prepared in this work has been reported to be 2.0 eV [23]. The formation of SPV signal below the In₂S₃ band gap was therefore due to charge separation involving tail states below the In₂S₃ band gap and probably due to defect states at the CTO/In₂S₃ interface. The magnitude of the SPV amplitude (Figure 3a) of the TiO₂/In₂S₃ was almost twice that of the CTO/In₂S₃ films. The SPV amplitude below 2.3 eV is higher for the TiO₂/In₂S₃ and CTO/In₂S₃ multilayer’s compared to that of the bare In₂S₃ films, while there was barely any signal below 1.7 eV for all samples. There was dependence of the SPV signal on the Co dopant concentration. Compared to the SPV signals from the bare TiO₂ and CTO films in Figure 2, we can conclude that the shallow defects centered at 1.4 eV were effectively passivated by the In₂S₃ over-layer.

The X signal was positive and the Y signal negative, which confirmed that the photogenerated electrons were preferentially separated toward the substrate and holes towards the surface. The phase angle for In₂S₃ was −47°, but shifted to about −60° for all the CTO/In₂S₃ samples (Figure 3b). This means that charge separation and recombination processes were slower for CTO/In₂S₃ samples compared to bare In₂S₃ films. The interface states at CTO/In₂S₃ heterojunctions played a role in the generation of the SPV signal in these samples by slowing down the processes.

3.3. Band Gap Diagram for CTO/In₂S₃ Double Layer

Charge separation across the CTO/In₂S₃ interface can be understood using the band diagram given in Figure 4. The band gap of TiO₂ was taken as 3.2 eV and that of In₂S₃ as 2.0 eV. Figure 4a indicates the presence of energetic surface defects centered at 1.9 and 1.4 eV below the conduction band, as deduced from Figure 2a. When the sample is illuminated, photon energies in this range are absorbed and the electrons trapped in this energetic defect state get excited into the conduction band, resulting in the SPV signal peaking at 1.4 and 1.9 eV. Photogenerated electrons excited into the conduction band are drawn into the bulk and holes towards the surface by the electric field created by the charge separation. The holes can be trapped at the surface states, resulting in delayed recombination. This is detected as an increase in the SPV Y-signal or an increase in the relaxation time of the conduction electrons in Photoluminescence measurements [6]. Increasing the Co dopant concentration enhanced the density of the surface states, where a larger number of holes got trapped, resulting in the higher amplitude of the Y-signal with increasing dopant amount, as observed in Figure 2b.

Introduction of Co²⁺ ions into a TiO₂ lattice is known to induce structural distortion and hence an increase of defect states. Because Co²⁺ is incorporated with a charge of +2, oxygen vacancies are created to fulfill the charge neutrality condition [5]. Surface trap states have been attributed to the Ti³⁺ trap states and oxygen vacancies. Photoluminescence experiments have shown radiative recombination of trapped electrons at 0.7–1.6 eV below the conduction band edge of TiO₂, which are associated with uncoordinated Ti³⁺ defects. Distribution of hole traps energetically positioned at 2.5–1.8 eV below the band gap are associated with hole traps due to oxygen vacancies [33,34,35]. Oxygen vacancies can form both shallow donor states as well as relatively deep trap levels located on the surface of TiO₂. The defect states observed at 1.4 eV can be assigned to acceptor-like electron traps resulting from the Ti³⁺ defects, while the donor-like states localized at 1.9 eV below the conduction band can be associated with oxygen vacancies. With Co doping, the new oxygen vacancies create lone pair electrons in the vicinity of Ti⁴⁺ and reduce it to Ti³⁺, increasing the density of both oxygen vacancy and Ti³⁺ defect states [33].

By covering the TiO₂ and CTO films with a thin In₂S₃ over-layer, the electronic defects on the surfaces were passivated and the SPV signals below the band gap disappeared. The SPV signal observed below the In₂S₃ band gap in Figure 3 was due to interface states existing at the CTO/In₂S₃ heterojunctions, which were not fully passivated. Some deep trap states remained at the interface and contributed to charge separation. As shown in Figure 4b, the SPV signal measured from the CTO/In₂S₃ double layer was due to several charge separation processes—namely, (i) photon absorption and excitation of electrons from the CTO/In₂S₃ interface states into the CTO conduction band; (ii) charge separation involving In₂S₃ tail states; (iii) band-to-band absorption in In₂S₃; and (iv) excitation from In₂S₃ valence band into CTO conduction band. It is also possible to have electrons excited from the defects states below the In₂S₃ band gap to states below TiO₂ band gap. The SPV measurement does not distinguish between the different processes because it only captures the net charge separation mechanisms.

It can be concluded that the enhancement of the photocatalytic activity of CTO films and nanostructures observed in visible light is mainly due to an increase in the surface defects that are involved in charge separation. Substitution of Ti⁴⁺ by Co²⁺ causes distortions in the lattice structure and the creation of additional oxygen vacancies, which increases the density of the deep and shallow defect states in CTO films [15]. Co doping does not introduce new electronic defects, but increases the density and distribution of the already existing Ti³⁺ and oxygen vacancy defects. This behavior has led to an increase in the possible applications of TiO₂ doped with transition metals in optoelectronics, especially under visible light.

4. Conclusions

Increasing the amount of Co dopant in the TiO₂ thin film resulted in a decrease in the deposition rates. The optoelectronic properties of the CTO films showed dependence on the concentration of the Co dopant. The increase in the amplitude of the SPV signal was due to enhanced charge trapping by the surface defect states created by the substitution of Ti⁴⁺ by Co²⁺ in the TiO₂ lattice. The formation of the SPV signal in the CTO/In₂S₃ double layer resulted mainly from charge separation in the In₂S₃ due to photon absorption and the excitation of electrons from the valence band into the conduction band of the In₂S₃ over-layer. The SPV onset energy of the CTO/In₂S₃ layers (1.92eV) was close to but slightly lower than the band gap of In₂S₃ (2.0 eV), and this can be due to the excitation of electrons from tail states below the band gap of In₂S₃. The holes can be trapped at the surface states, resulting in delayed recombination. This is detected as an increase in the SPV Y-signal or an increase in the relaxation time of the conduction electrons in PL measurements. This explains the enhanced photocatalytic activity in CTO films and nanostructures compared to undoped TiO₂ reported in literature.