Intrinsic defects generated by iodine during TiO2 crystallization and its relationship with electrical conductivity and photoactivity

Abstract

Defect formation during synthesis is one of the strategies used to improve the photoactivity of polycrystalline semiconductors such as titanium dioxide (TiO₂). Defects can modify the electronic structure of TiO₂ and change the surface of the interaction between the photocatalyst and the reactants. In this study, TiO₂ relationship between processing in the presence of iodine and the consequent formation of intrinsic defects were explored. TiO₂ nanoparticles were synthesized using the polymeric precursor method and exposed to iodine ions at concentrations up to 5 mol%. After calcination at 350°C, detailed chemical analyses revealed that iodine was absent in the samples. However, the TiO₂ properties, such as specific surface area, crystallite sizes, and specific grain boundary area, were affected. Further experiments, such as electron paramagnetic resonance, diffuse reflectance, optical measurements, and electrochemical impedance spectroscopy indicated the presence of defects in the iodine-processed samples. These defects directly influenced the electrical properties of the material, which affected the photoactivity, measured by the degradation of acetaminophen.

INTRODUCTION

Crystalline defects in materials are crucial to understanding most of the materialsˈ properties, and the development of their micro/nanostructures. Defect equilibrium in a ceramic semiconductor oxide such as TiO₂ has as main thermodynamic variables the pressure, temperature, the partial pressure of oxygen, and the concentration of impurities or additives. The defects generated by the presence of additives in these materials can modify fundamental properties such as electrical, thermal, or optical. They can also change the development of micro/nanostructures because of the modification of the chemical composition of the interfaces due to segregation or/and diffusion coefficients.¹

Electronic or point defects in nanoxide semiconductors are generally formed by the solubility of ionic additives.² The general concept is that the ionic additive is soluble by substitution or interstitial in the bulk with the consequent defect formation. However, less attention has been paid to the presence of anionic additives that are susceptible to evaporation during powder crystallization. When evaporation is accompanied by a change in the oxidation state for gas formation, there may be an oxidation/reduction of the matrix oxide species that alters the balance of intrinsic defects.

The introduction of defects in semiconductors such as TiO₂ is relevant in applications such as catalysis, photocatalysis, and electrophotocatalysis. Defects can influence catalytic processes by modifying the electronic structure of TiO₂ and altering the interaction between the surface of the catalyst and the reactants.

Besides these traditional approaches to how defects alter the electronic structure of the material, one relevant aspect of photocatalysis that has been ignored is the electrical conductivity across grain boundaries (GBs). When an additive segregates in the GBs, it can modify the electrical conductivity of the material.³ This is a consequence of the modification of the electrical potential barrier height and the activation energy for conduction. Therefore, the GBs are expected to act as shortcuts for charge transport, while the electrical conductivity of the polycrystalline species tends to increase with the GB density.

This work aims to study the formation of intrinsic defects in TiO₂ nanoparticles caused by the introduction of iodine during synthesis. The defects were detected by color measurements using the CIE L*a*b system, UV–vis diffuse reflectance, electron paramagnetic resonance spectroscopy, and diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy. Then, we used electrochemical impedance spectroscopy (EIS) to measure the overall conductivity of the system, especially at the grain boundary regions. Finally, we studied the photoactivity of the system by degradation of acetaminophen. The goal is to unravel the interdependence between GB electrical conductivity and the photocatalytic properties of nanostructured TiO₂.

EXPERIMENTAL METHODS Synthesis of TiO₂ nanopowders in the presence of iodine

TiO₂ nanopowders were synthesized using the polymeric precursor method. The iodine target molar concentrations during processing were 0.00, 0.05, 0.10, 0.50, 1.00, 2.50, and 5.00 mol%. The titanium-based polymeric precursor was prepared by adding titanium isopropoxide (19.4 wt%) to ethylene glycol (45.4 wt%) at 40°C with constant stirring. After heating the mixture to 70°C, citric acid (35.2 wt%) was added, followed by heating to 120°C for 30 min for polyesterification. The resin was saved, and a thermogravimetric experiment at 900°C for 5 h was performed to determine its yield, which was ∼6%. Iodine was introduced by adding an appropriate amount of an aqueous NH₄I solution (0.1 mol L⁻¹) to the resin. Each composition was mixed for 30 min in an individual container, followed by calcination.

Calcination was carried out in two steps. First, to decompose and eliminate organic traces, each sample was held at 350°C for 4 h. Thus, the samples were ground with a mortar and pestle and heated to 350°C for 15 h under air flow to complete the reaction and stabilize the particle size.

Powder characterization

XRD patterns were obtained using an X'Pert PRO PW 3040/00 Philips diffractometer with Cu Kα radiation, operated at 45 kV and 40 mA. The step size was set at 0.02°, with a step time of 50 s in the 2.5°−90° 2θ range. Crystallite sizes and lattice parameters were calculated by Rietveld analysis using X'Pert Highscore software with MgAl₂O₄ as a standard obtained under the same conditions as the TiO₂ samples. The peak profile used to refine the diffractograms was described by the pseudo-Voigt function, without asymmetry correction, and the FWHM parameters were refined using the Caglioti function.

The specific surface areas of the samples were determined by the adsorption of N₂ gas at 77 K (Micromeritics Gemini III 2375 Surface Area Analyzer) according to the Brunauer-Emmett-Teller (BET) method. Before the experiment, the samples were subjected to thermal pretreatment at 300°C for ∼16 h (100 mmHg—Micromeritics VacPrep 061) for surface degassing.

Powder density measurements were carried out by helium gas pycnometry using a Micromeritics AccuPyc II 1340 pycnometer after 200 purges. Before the experiments, the samples were dried at 105°C for 12 h.

The chemical composition of the powders was determined by X-ray fluorescence spectroscopy (XRF) using a Malvern Panalytical Zetium XRF spectrometer. The iodine concentration was measured using a HI 4111 iodide ion-selective electrode (ISE) connected to a pH Analyzer 300 M. The samples were dissolved in an aqueous sulfuric acid solution (50.0% in volume) and the pH was neutralized with sodium hydroxide. The selective lixiviation method^4,5 was also employed to try to detect iodine on the surface of the samples. The iodine concentration in samples was also visualized by energy-dispersive X-ray spectroscopy (EDS) using an FEI Inspect F50 FEG scanning electron microscope (SEM), operated at 20 kV and 122.7 eV resolution.

TiO₂ defects were identified by electron paramagnetic resonance spectroscopy (EPR). A Bruker EMX electronic paramagnetic resonance spectrometer was used, operated at 150 K, controlled by a Bruker BVT-2000 system with a microwave frequency of 9.498 GHz, and magnetic field range (B0) measured from 3280 to 3480 Gauss.

Diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy was performed using the Thermo-Nicolet Magna 560 spectrometer with a scanning capacity of 400−4000 cm⁻¹ and a resolution of 4 cm⁻¹.

X-ray photoelectron spectroscopy (XPS) measurements were performed using a modular ultrahigh vacuum system (UNI-SPECS UHV Surface Analysis System) equipped with XPS. The Al Kα line (hν = 1254.6 eV) was used as the ionization source and the atomic percentage composition of the surface layer (< 5 nm) was determined by the relative proportions of the peak areas corrected by the Scofield atomic sensitivity factors, with a precision of ± 5%. The spectra were deconvoluted using a Voigtiana-type function, with Gaussian (70%) and Lorentzian (30%) combinations.

For color measurements, the TiO₂ nanopowders were pressed into a 20 mm × 50 mm rectangular matrix using a Carver hydraulic press at 0.3 ton cm^–2 followed by the analysis on a MiniScan XE Plus equipment by the CIE L*a*b* system and standard illuminant D65.

Electrochemical impedance spectroscopy (EIS) was performed on TiO₂ samples processed with 0.00, 1.00, and 5.00 mol% iodine using a SI 1260 / gain-phase analyzer (Solartron) impedance meter connected to a closed furnace with controlled temperature and atmosphere. The EIS data were collected in the frequency range of 30 MHz to 1 Hz with an AC voltage amplitude of 500 mV and without dc bias. Before the experiments, the samples were presintered at 360°C for 1 h and the experiments were carried out from 20°C to 125°C.

The photoactivity of the samples was evaluated by the degradation of the model contaminant acetaminophen (ACT) using an aqueous TiO₂ suspension containing ∼10 mg of photocatalyst in 8 mL of a 5 ppm ACT solution. The suspension was irradiated with a monochromatic UV-A LED (LG, λ = 370 nm), which provided an irradiance of 7 mW cm^–2 on the reactor walls. Before the light was turned on, the samples were stirred using a magnetic stirrer at ∼25°C for 30 min to reach an equilibrium of adsorption and desorption of ACT molecules on the TiO₂ surface. During the test, the UV-A LED was the only source of light in the room, eliminating any possible interference of other types of light with the reaction kinetics. Aliquots of the suspension were collected every 15 min and filtered to eliminate any type of contamination by solid particles of the photocatalyst. The aliquots used during the photocatalytic test were 100 μL each, resulting in a total volume of 900 μL after the test, which accounted for approximately 10% of the total volume used in the experiment. During the test, the system was continuously mixed to ensure uniform dispersion of the catalyst, maintaining a constant catalyst concentration in the remaining volume. The concentration of ACT in the aliquots was measured by high-performance liquid chromatography (HPLC) with UV detection at 243 nm (Lc20-A, Shimadzu Co.) after elution through a C18 column with a 0.8 mL min⁻¹ isocratic flow of methanol and deionized water in a proportion of 1:3.

RESULTS AND DISCUSSIONS

Chemical analyses were performed to investigate the chemical elements present in the samples. EDS analysis revealed that there were no iodine peaks in iodine-processed TiO₂ (3.937 keV, Figure S1). The only elements identified were O, Ti, and Au (2.120 keV), the latter due to the coating used for the analysis. To double-check the EDS results, an iodide-selective electrode (ISE) was employed. First, the TiO₂ samples were dissolved in an aqueous sulfuric acid solution (50.0% in volume), then, to reach the pH working range (2.00 to 11.00) of the ISE, the solution was neutralized with sodium hydroxide. No iodide was detected, which confirmed the EDS results. The powders were also leached by the selective lixiviation process, and the supernatant was analyzed by ISE. No iodine was detected, which also confirms the EDS results. Possibly, the ammonium iodide used as an iodine precursor during the synthesis process dissociated into ammonia and iodine gas and evaporated during the reaction.⁶

Although the chemical analyses did not show the presence of iodine in the samples, an investigation was conducted to evaluate the potential effects left by the processing in the presence of iodine. The main characterizations were color measurements, UV–vis diffuse reflectance, EPR, XPS, electrochemical impedance spectroscopy, and photoactivity.

Figure 1 shows the X-ray diffractogram for samples processed in the presence of iodine in concentrations ranging from 0.00 to 5.00 mol%. The predominant phase is anatase, but a small peak of rutile (2θ = ∼27°) can also be observed in some samples. Because the peaks are too small, it was not possible to quantify them in the phase volume. The broad peaks observed are typical of nanostructured materials.

FIGURE 1. X-ray diffractograms of iodine-processed TiO2 samples.

The crystallite sizes, specific surface area (SSA), density, specific grain boundary (SGB) area, and SGB/SSA ratio are listed in Table 1. Crystallite size (D) and pycnometry density (ρ) were used to calculate the SGB and the total specific interface area (TIA). The former was calculated using Equation (1) by assuming that all particles had a tetrakaidecahedral shape. The latter was calculated according to Equation (2) considering that each grain boundary was shared by two particles. [Image Omitted. See PDF][Image Omitted. See PDF]

TABLE 1 Crystallite sizes, specific surface area (SSA), density, specific grain boundary (SGB) area, and SGB/SSA ratio of iodine-processed TiO₂.

Samples	Crystallite size (nm)	SSA (m2 g–1)	Density (g cm–3)	SGB (m2 g–1)	SGB/SSA
0.00%	13.5 ± 0.1	60.9 ± 0.1	3.444 ± 0.005	46.1 ± 0.3	0.76
0.05%	13.7 ± 0.1	58.5 ± 0.1	3.465 ± 0.006	45.6 ± 0.4	0.78
0.10%	14.1 ± 0.1	57.3 ± 0.1	3.469 ± 0.007	44.1 ± 0.4	0.77
0.50%	14.7 ± 0.1	57.5 ± 0.1	3.487 ± 0.008	40.5 ± 0.4	0.70
1.00%	11.7 ± 0.1	62.6 ± 0.1	3.446 ± 0.007	56.9 ± 0.4	0.91
2.50%	12.9 ± 0.1	57.9 ± 0.1	3.505 ± 0.013	49.3 ± 0.4	0.85
5.00%	11.3 ± 0.1	49.9 ± 0.1	3.432 ± 0.005	66.3 ± 0.5	1.33

The crystallite sizes were slightly affected by the presence of iodine during the synthesis. They increased for samples from 0.00 to 0.50 mol% iodine, and then dropped to values below the pristine TiO₂. The SSAs did not present any specific trend and can be better analyzed together with the SGB and the SGB/SSA ratio. The SGB showed a tendency toward higher values. The SGB/SSA ratio increased especially for TiO₂ samples processed with 1.00 and 5.00 mol% iodine. These results indicate that the presence of iodine during synthesis can affect the stability of the interfaces in this system, stabilizing the GB against the surface.⁷

The color measurements shown in Figure 2 indicate that the color of the samples changed with the amount of iodine during synthesis. The CIE L*a*b* system shows a tendency toward the orange color (positive a and b) with a darker tonality (lower L). The change in sample color confirmed that the TiO₂ samples contained many defects.⁸ Among the several color-changing mechanisms that may influence the materials optical properties, oxides are usually affected by three main mechanisms: color change because of the charge transfer between ions and defects, color change deriving from the band theory in semiconductors, with the introduction of a donor or acceptor level in the band gap by defects, and the presence of color centers generated by electrons transfer from the semiconductor lattice to defects on the surface.⁹ To better understand the reasons why the color of the samples changed, the band gap of the samples using UV–vis diffuse reflectance measurements was investigated.

FIGURE 2. CIE L*a*b* color system showing the color change in iodine-processed TiO2 samples.

The UV–vis diffuse reflectance spectra are shown in Figure 3. As expected for the TiO₂ anatase phase, the samples showed a low UV reflectance between 5.0 and ∼3.2 eV, and high visible reflectance at energies lower than ∼3.1 eV. The reflectance decreased in the visible spectrum at wavelengths close to violet and blue, mainly for TiO₂ processed with 1.00, 2.50, and 5.00 mol% iodine. These results suggest that the TiO₂ samples absorbed light at these wavelengths, which was observed in the material through the color variations from white to orange/yellow, and complementary colors of violet/blue, confirmed by the colorimetry results.

FIGURE 3. Diffuse reflectance spectra.

The band gap of the samples was estimated by applying the Kubelka–Munk function using the Tauc model in the diffuse reflectance spectra. The results are shown in Table 2. Although the band-gap values tended to decrease, they were all virtually equal, suggesting that processing did not affect the energy gap of the samples. Therefore, the color change seen in Figures 2 and 3 was not due to the narrowing of the material band gap or the creation of intermediate energy levels in the CB.

TABLE 2 TiO₂ band gap obtained by the Tauc method.

Sample	Band gap (eV)
TiO2 0.00% I	3.11 ± 0.03
TiO2 0.05% I	3.10 ± 0.03
TiO2 0.10% I	3.09 ± 0.03
TiO2 0.50% I	3.08 ± 0.03
TiO2 1.00% I	3.09 ± 0.03
TiO2 2.50% I	3.08 ± 0.03
TiO2 5.00% I	3.07 ± 0.03

For the TiO₂ 5.00% I sample to obtain an orange color, its band gap should be approximately 2.60 eV, not 3.07 eV. Possibly, the interface defects that were acting as adsorption sites were also generating color centers. Color centers occur in the material when a hole-generating species, A, coexists with an electron-receiving species, B. Upon absorbing energy, the hole-generating species will eject an electron to the electron-receiving species, for example, an oxygen vacancy, forming an A⁺ center and a B⁻ center, also known as F-center (F, F⁺, F⁺⁺, and Ti³⁺). For this electronic transition to occur, light absorption is necessary, often in the visible spectrum, which makes a colorless material obtain color.⁹ Serpone¹⁰ suggests that visible light absorption by doped TiO₂ does not occur by narrowing the band gap or creating intermediate energy levels in the semiconductor. This is due to the generation of defects that act as color centers that absorb the visible spectrum. Moreover, manipulation of the band gap through doping is a known phenomenon, but it requires a large amount of dopant, which can drastically modify the chemical composition of the photocatalyst and form new crystalline phases.

Following the experiments to identify the type of defects created by iodine processing in TiO₂, EPR measurements were performed. Oxygen vacancies can trap unpaired electrons from the material, being detectable by EPR, while Ti³⁺ sites are paramagnetic and are usually associated with the removal of oxygen atoms from the lattice. The EPR g-tensor values for Ti³⁺ are in the 1.99-1.96 range. Figure 4 shows the EPR spectra of the TiO₂ 0.00% I, TiO₂ 1.00% I, and TiO₂ 5.00% I samples. The g-tensor values were identified at 2.0077 for all samples and at 1.9860 and 2.0238 for TiO₂ 5.00% I. These values can be associated with oxygen vacancies or Ti³⁺ defects. Studies have shown that Ti³⁺ in regular positions of the anatase lattice, generated by the compensation of an extra charge carried by a cation (e.g., Nb⁵⁺) or an anion (e.g., F⁻) as dopants, exhibits axial EPR signals in g = g_xx = g_yy = 1.992 and g_∥ = g_zz = 1.962 for spectra recorded at 10 and 77 K. The same signal appears in reduced anatase annealed under vacuum.¹¹ Although the signal in Figure 4 has the same axial shape, the g-factor values are slightly different from those reported in the literature. Similar signals have been observed in doped anatase,¹¹ anatase under irradiation,¹² or after thermal treatments.¹³

FIGURE 4. EPR spectra for TiO2 0.00% I, TiO2 1.00% I, and TiO2 5.00% I samples.

In most cases, defects are identified after UV or visible light irradiation. In this study, EPR measurements were performed in a closed sample holder without any illumination, further indicating that defects were generated during the synthesis as a result of iodine evaporation and that the process method influenced the formation of defects in the material. This feature agrees with previous findings of electron holes induced by valence change (O⁻) identified by EPR in anatase.¹² Electron holes trapped onto an O²⁻ ion have g values between g = 2.003 and g = 2.027. The same values were obtained for anatase under illumination.

A similar result was found by Li et al.,¹⁴ where g-factor values of g_zz = 2.023, g_yy = 2.006, and g_xx = 1.987 of an orthorhombic EPR signal were detected for N-doped TiO₂. These signals were attributed to O₂⁻ species generated from the interaction of atmospheric O₂ molecules with the localized electrons in the oxygen vacancies on the surface. Although these signals do not directly represent oxygen vacancies, they are indicative of the presence of paramagnetic species and probably the existence of Ti³⁺ in the lattice. The presence of O₂⁻ species on the surface of TiO₂ is consistent with other reports.¹⁵ The formation of O₂⁻ groups on the surface of reducible oxides is possible when oxygen adsorption occurs after a preliminary reductive treatment (e.g., chemical reduction). If the electrons occupying donor levels in the band gap of a reduced oxide are excited to the conduction band, they can be transferred toward adsorbed oxygen even at low temperatures (e.g., room temperature). Electron transfer causes a depletion of charges in the conduction band, leading to a decrease in the material conductivity.¹⁶

To investigate the chemical states of the samples, measurements of I 3d, Ti 2p, O 1s, and C 1s were performed on TiO₂ 5.00% I sample using XPS. The analysis revealed the presence of Ti, O, and C located in the binding energies of 458.25 eV, 529.50 eV, and 285.00 eV, respectively. The spectra can be seen in Figure 5. The Ti 2p_3/2 and Ti 2p_1/2 spectra are associated with TiO₂ bonds.¹⁷ Ti³⁺ peaks, usually detected around 457.2 eV (Ti 2p_3/2) and 462.9 eV (Ti 2p_1/2), were not observed in any sample.^18–20 The main component in the oxygen spectrum was related to O-Ti bonds. Surface hydroxyl groups and oxidated carbon groups were also identified. The main component in the carbon spectrum was related to hydrocarbon groups (CH) on the surface and other oxidized carbon groups at a lower content. No iodine 3d peaks were detected, which corroborates the EDS results.

FIGURE 5. XPS spectra of Ti 2p, O 1s, and C 1s of TiO2 5.00% I. I 3d and Ti3+ were not detected.

DRIFT analyses were performed to identify the species adsorbed on the surface of TiO₂ samples and their relationship with defects. The results are shown in Figure S2 and the chemical groups are presented in Table 3.

TABLE 3 Chemical groups of TiO₂ and iodine-processed TiO₂ DRIFT spectra, and their respective vibration bands.

Chemical group	ν (cm−1)	References
OH	3700–2700	21, 22
O = C = O	2350	23–25
CO	2285	25
CO	2215	25
OH	1621	21
b-CO32− bidentate	1556	22, 25–26
Carbonates	1356–1377	23, 25
Ti—O	650–1100	21

DFT and experimental studies have shown that CO₂ adsorption, activation, and dissociation processes are significantly influenced by the distribution of defects on the TiO₂ surface, as well as the phases present in the material. These studies indicate that electrons trapped in photocatalyst defects can induce the formation of CO₂⁻ and its subsequent dissociation into CO in the dark.^27,28 In situ DRIFT studies also confirmed that CO₂ can spontaneously dissociate into CO in reduced Rh/TiO_2–x²⁹ and Cu(I)/TiO_2–x (P25) systems due to surface defects,³⁰ even without UV–vis irradiation. In these studies, exposure of photocatalyst surface defects to CO₂ led to the formation of CO₂⁻ species bonded to the Ti⁴⁺ surface site, which was detected by DRIFT in the vibration bands of 1673  and 1248 cm⁻¹. This fact suggests that adsorbed CO₂ can be activated by the spontaneous migration of excess electrons trapped in defects, such as Ti³⁺ or oxygen vacancies.³¹ The CO₂ species in angular vibration mode present DRIFT bands at 1673  and 1248 cm⁻¹ and are adsorbed at Ti³⁺ sites on the surface.³⁰ These groups were not identified in the TiO₂ samples in the present study, suggesting that the reduction of CO₂ to CO₂⁻ does not occur spontaneously in this system.

Figure 6 shows the electrochemical impedance spectra obtained at 20°C for TiO₂ 0.00% I, TiO₂ 1.00% I, and TiO₂ 5.00% I samples. Each curve can be divided into two parts: one at high frequency (from ∼10 MHz to 100 Hz), characterizing the electrical properties of the samples and a spike at low frequency (< 100 Hz) related to electrode reactions. The high-frequency portion of the diagrams was deconvoluted and extrapolated to find their interception with the real axis Z', which represents the total resistivity of the samples. It was observed that both capacitance and resistivity increased with the amount of iodine during the synthesis. Generally, TiO₂ is recognized as an n-type semiconductor at room temperature due to the presence of oxygen vacancies and interstitial titanium atoms that supply electrons to the conduction band.³² More recently, p-type conductivity has been obtained at room temperature for undoped TiO₂.^33,34 This transition from n-type to p-type semiconductor occurs through the formation of titanium vacancies (electron acceptor defect), which can be generated from rutile oxidation at high temperatures (∼1000°C),³³ and also from the solvothermal synthesis of anatase followed by calcination at 470°C.³⁴

FIGURE 6. Electrochemical impedance spectra showing the increase in electrical resistance for iodine-processed samples. The numbers on the curve are the logarithms of the measuring frequency, in Hertz.

A possible effect for the increase in electrical resistance for iodine-processed samples is based on the presence of Ti³⁺ species segregated at the grain boundaries of the material. The EPR results (Figure 4) indicated the presence of oxygen vacancies and Ti³⁺ in the TiO₂ samples. When oxygen vacancies are generated, the formation of Ti³⁺ species occurs to maintain charge neutrality. However, the Ti³⁺ species were not identified by XPS on the surface of the material, suggesting that it is soluble in the bulk or at the grain boundaries. The increase in the SGB/SSA ratio (Table 1) for samples processed with iodine indicates the presence of species segregated at grain boundaries, which change the chemical composition and stabilize this interface.^3,35,36 The SGB/SSA ratio ranges from 0.76 to 1.33 for samples prepared with 0.00 and 5.00% iodine, respectively. As iodine was not identified during chemical analyses, the species segregated in GBs may be Ti³⁺ ions that received an electron during I₂ evaporation, with a simultaneous formation of an oxygen vacancy, as shown in Equation (3): [Image Omitted. See PDF]

Studies have shown that the segregation of defects and electrons occurs in the GBs of TiO₂ and contributes to the increase in the potential barrier and the depletion layer at the grain boundary, influencing the conductivity of the material.^37–39 EPR spectra of anatase samples with and without grain boundaries suggest that Ti³⁺ is associated with the sublayer as it is not affected by surface adsorption, indicating that the titanium defect is located at the solid-solid interface.⁴⁰ However, oxygen vacancies are generated on the surfaces and adsorb molecular oxygen from the gas phase (or soluble in water), generating paramagnetic species that can be analyzed by EPR⁴¹ and are related to the photocatalytic activity of TiO₂ anatase.⁴²

Photodegradation of acetaminophen (ACT) was performed for TiO₂ 0.00% I, TiO₂ 1.00% I, and TiO₂ 5.00% I samples under UV radiation to understand the contribution of defects and interfaces in anatase photoactivity. The possible chemical reactions of ACT degradation can be represented by the equations presented in Table 4.

TABLE 4 Possible chemical reactions of ACT degradation by oxidation⁴³ ($$e_{{\mathrm{CB}}}^ - \ $$is the electron in the conduction band and $$h_{{\mathrm{VB}}}^ + $$ is the hole in the valence band).

$${\mathrm{Ti}}{{\mathrm{O}}_2} + h{\nu _{{\mathrm{UV}}}} \to e_{{\mathrm{CB}}}^ - + h_{{\mathrm{VB}}}^ + $$	(4)
$$e_{{\mathrm{CB}}}^ - + h_{{\mathrm{VB}}}^ + \to h{\nu _{{\mathrm{IR}}}}$$ (recombination)	(5)
$${{\mathrm{H}}_2}{\mathrm{O}} + \ h_{{\mathrm{VB}}}^ + \to {\mathrm{OH}} \cdot \ + {{\mathrm{H}}^ + }$$	(6)
$${\mathrm{O}}{{\mathrm{H}}^ - } + \ h_{{\mathrm{VB}}}^ + \to {\mathrm{OH}} \cdot \ $$	(7)
$$e_{{\mathrm{CB}}}^ - + {{\mathrm{O}}_2} \to {\mathrm{O}}_2^ - $$	(8)
$${\mathrm{ACT}} + \ h_{{\mathrm{VB}}}^ + \to {\mathrm{oxidation}}$$	(9)
$${\mathrm{ACT}} + {\mathrm{OH}} \cdot \to {\mathrm{oxidation}}$$	(10)

Some steps are fundamental for the success of the ACT oxidation reaction, among them, the formation of the electron/hole pair and its recombination, the reaction of charges with molecules adsorbed on the TiO₂ surface, either water or hydroxyls, or even molecular oxygen. Note that the transfer of electrical charges occurs through sites on the surface, which are usually crystalline defects, such as oxygen vacancies.

The results displayed as the concentration of ACT over time are shown in Figure 7. Although no iodine was detected in the samples, the presence of defects hinders the photodegradation reaction, as evidenced by the higher apparent kinetics of decomposition seen for the TiO₂ 0.00% I sample.

FIGURE 7. Kinetics of acetaminophen photodegradation under UV light of 370 nm wavelength.

The term ln(C₀/C) was used to calculate a pseudo-first order reaction rate constant. Moreover, the conversion achieved at 120 min was divided by the initial specific surface area of each composition to provide a normalization of the photocatalytic degradation capacity to account for differences in surface area. These parameters are summarized in Table 5.

TABLE 5 Summary of photocatalysis results.

Sample	Apparent pseudo-first order rate (min−1)	Specific conversion at 120 min (μmol ACT m−2)
TiO2 0.00% I	4.64 × 10−3	0.188
TiO2 1.00% I	3.54 × 10−3	0.139
TiO2 5.00% I	3.41 × 10−3	0.181

The lower photoactivity of the iodine-processed samples is not a result of the change in SSA, since the SSA of the TiO₂ 1.00% I sample is greater than that of the TiO₂ 0.00% I sample (Table 1). However, increasing the GBA (Table 1) can contribute to charge recombination by modifying the mean free path of electrons and holes or by promoting a region of charge recombination at the grain boundaries.⁴⁴ Another consideration is that the total number of defects is almost the same for both samples, as demonstrated by EPR measurements (Figure 4). Therefore, the increase in grain boundary resistivity (Figure 6) can be associated with the poor efficiency of the photodegradation reaction.

By comparing TiO₂ 5.00% I with the pristine TiO₂ and TiO₂ 1.00% I samples, a strong reduction in SSA simultaneously with the increase in GBA and the intensification of electrical resistance is observed (Table 1, Figures 6 and 7). In this case, an even lower photocatalytic activity was expected. However, the ACT photodegradation for the TiO₂ 5.00% I sample is very similar to that of the TiO₂ 1.00% I sample. The observed efficiency can be justified by the sharp increase in the number of defects detected by the EPR analysis (Figure 4), which could be associated with surface defects. Hence, although there is a decrease in degradation efficiency due to the increase in the grain boundary electrical resistance and the decrease in SSA, the increase in the number of defects seems to partially compensate for those effects and increase photoreactivity.

SUMMARY

Intrinsic defects were formed in TiO₂ nanoparticles synthesized in the presence of iodine. The iodine ions evaporate during calcination and promote the reduction of Ti⁴⁺ to Ti³⁺, which segregates at the grain boundaries (GBs), stabilizing this interface regarding the surface. The Ti³⁺ species segregated in the GBs decrease the overall electrical conductivity of the system, reducing the photodegradation of acetaminophen in an aqueous solution under UV irradiation (Figure 8).

FIGURE 8. Summary of the article designated for the graphical table of contents.

CONCLUSIONS

TiO₂ nanoparticles were processed in the presence of iodine as an additive up to 5 mol%. The results indicated that the photocatalytic activity is limited by the electrical conductivity in polycrystalline TiO₂. Experimental data indicate that Ti³⁺ species, probably generated by I₂ formation, segregate at the GBs and decrease the overall electrical conductivity of the system, reducing the photodegradation of acetaminophen in an aqueous solution under UV irradiation. The iodine ions evaporation promoted the reduction of Ti⁴⁺ to Ti³⁺, which segregate at the GBs, stabilizing this interface regarding the surface. These results can be used as a guideline to design catalysts with better properties by controlling doping and dopant segregation that favor the formation of active surface species for photocatalytic reactions that are not hindered by low grain boundary electrical conductivity.

ACKNOWLEDGMENTS

We are grateful to the support from the Research Centre for Gas Innovation (RCGI), hosted by the University of São Paulo (USP) and sponsored by the São Paulo Research Foundation (FAPESP, 2014/50279-4 and 2020/15230-5) and Shell Brazil. Furthermore, we are grateful to the support provided by the ANP (Brazilian National Oil, Natural Gas, and Biofuels Agency) through the R&D levy regulation. We also acknowledge the support from FAPESP (grants 2019/10109-6 and 2017/11937-4) and Prof. Dr. M. Teresa Lamy and Dr. Evandro L. Duarte from the Biomembranes Laboratory at the Institute of Physics of the University of São Paulo for their assistance with EPR experimental procedures. DG and FCF are fellows of CNPq (Brazilian National Council for Scientific and Technological Development).

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.