The Fabrication of Lithium Niobate Nanostructures by Solvothermal Method for Photocatalysis Applications: A Comparative Study of the Effects of Solvents on Nanoparticle Properties

Abstract

In this work, we report, for the first time, a comparative study on the effects of different solvents on the properties of LiNbO₃ (LN) nanostructures. The solvothermal synthesis method was successfully used with three different solvents: 1—water, 2—methanol, and 3—benzyl. The structural and optical properties of the as-prepared nanoparticles were studied using transmission electron microscopy (TEM), X-ray diffraction (XRD), UV-Vis absorbance, Raman spectroscopy, and photoluminescence (PL). Nanoparticles of a very small size, with an average size between 3 and 10 nm, were obtained for the first time. The photocatalytic activities of the three synthesized LiNbO₃ nanoparticles were studied in relation to the photodegradation of a complex and heavy reactive black 5 dye for a wastewater treatment application. The LiNbO₃ synthesized with deionized water showed a higher photocatalytic activity than those synthesized using other solvents, such as methanol or benzyl.

1. Introduction

A LiNbO₃ (LN) semiconductor has a variety of physical properties, such as birefringence, pyroelectricity, piezoelectricity, electro-optical properties, elasticity, photoelasticity, and bulk photovoltaic properties, that make it the preferred silicon semiconductor in optoelectronic applications [1,2,3,4,5,6,7]. These properties have recently made this semiconductor a promising candidate for new emergent applications related to environmental and sustainable energy, such as solar fuel production, CO₂ reduction, pollutant photodegradation, hydrogen generation, and lithium batteries [8,9,10,11]. Due to their polarity surface properties and their impacts on the charge carrier recombination rate, ferroelectric materials such as LiNbO₃ have been shown to produce 7 times more hydrogen under UV irradiation than TiO₂ and 36 times more under visible irradiation [12]. Ferroelectrics exhibit spontaneous polarization, which plays an important role in enhancing photocatalytic mechanisms. Recently, ferroelectric materials have been reported in different studies as novel photocatalysts in wastewater treatment applications [13,14,15,16]. The internal depolarization fields in these materials conduct photogenerated charge carriers to move in opposite directions, causing the spatial separation of the charges for oxidation and reduction reactions [17]. Fabricating LN nanostructures with a very small size and good crystallinity requires selecting the best synthesis method and the best parameters, such as the solvent, the precursor of niobium, the reaction temperature, and the surfactants [18]. Sol-gel, metal alkoxides, the Pechini approach, non-hydrolytic solution reactions, hydrothermal and solvothermal procedures, and the peroxide route are currently some of the synthesis routes that have been discovered to create LN powders [2,19,20,21,22,23,24]. Among the above techniques, the sol-gel and hydrothermal techniques are the two best types of soft-chemistry synthesis procedures that have most successfully produced LN powders [18,24,25]. According to XRD and IR data, the sol-gel process yields pure LN powders at 600 °C, whereas the hydrothermal method allows for a significantly lower synthesis temperature of 240 °C. By further examining the microstructure of these LN samples, it was shown that, in contrast to the sol-gel approach, the solvothermal process is better for the fabrication of high-quality LN powders due to its relatively lower synthesis temperature, direct synthesis method, and higher-crystallinity products [5,24,25,26]. Although there are reported works on the solvothermal method, no one has reported the effect of solvents on the structural and optical properties of LN powders. In this work, we report, for the first time, a comparative study among three solvents—DI water, benzyl alcohol, and methanol—on the properties of synthesized LN powders. The photocatalytic activities of the powders were studied using the photodegradation of complex and heavy reactive black dye for wastewater treatment applications.

2. Materials and Methods

2.1. Preparation Methods

We used the solvothermal synthesis method to prepare LiNbO₃ using three solvents: deionized water (DI), benzyl alcohol, and methanol. For each solvent, we separately added 400 μL of niobium (V) ethoxide (Chem Cruz Biotechnology, Dallas, TX, USA) to 40 mL of deionized water, benzyl alcohol (99–100% GC—Sigma Aldrich, Schnelldorf, Germany), or methanol (99.5%—Thermo Fisher Scientific, Loughborough, England). After 30 min of stirring without heating, the color became slightly milky. After that, we added 148.4 mg of lithium 2,4-pentanedionate (98%—Chem Cruz Biotechnology, Dallas, TX, USA) and stirred the mixtures for 6 h; the pH for the mixtures was measured to be 10.6 (Hanna instrument—ph211 microprocessor pH meter, Woonsocket, RI, USA), 9.4, and 9, respectively, for the three solvents (DI, benzyl alcohol, and methanol). All the chemicals were used as received without further purification. The resulting product was placed directly in a 45 mL, Teflon-lined, stainless-steel autoclave (Parr-instrument, Moline, IL, USA) and steamed at 220 °C for 96 h. After waiting for the autoclave to reach room temperature, white precipitates were separated from the reaction mixture via the centrifugation (benchtop high-speed centrifuge—Biolab, Markham, ON, Canada) of the suspension at a speed of 8000 rpm for 30 min. The obtained powders were washed three times by placing them in an ethanol suspension of 10 mL and repeating the process of centrifugation and decanting the solution. We purified the products three times in 10 mL of deionized water. The purified products were dried at 70 °C for 24 h prior to further analysis to remove any moisture present from the purification process.

2.2. Characterization Techniques

Transmission electronic microscopy (TEM) measurements were performed by using TEM equipment (JEOL, JEM-2100F, Tokyo, Japan) with a voltage of 200 kV. A solution of 10 µL of nanoparticles was deposited on a carbon-covered copper grid after sonication for 20 min. To measure the crystal structure of the powders, we used an X-ray diffractometer (Philips type PW 1710, Amsterdam, The Netherlands) with CuKα radiation. The XRD patterns were measured within the range of [10°–90°] with a scanning frequency of 2°/min. A confocal Raman microscope (LabRAM HR800, HORIBA FRANCE SAS, Palaiseau, France) was used to measure the Raman spectra at room temperature with a blue laser (He-Cd), with a wavelength of 442 nm and an output power of 20 mW. The Raman spectra were recorded in a backscattering configuration at ambient temperature. A grating of 1800 l/mm, an acquisition time of 90 s, and a 2 mW excitation power were used to record the Raman spectra. The photoluminescence measurements were carried out using a He-Cd laser with a wavelength of 325 nm and an output laser power of 20 mW as an excitation source. The PL spectra were recorded using the following parameters: a grating of 1800/mm, an objective with a magnification of 40× to record, an acquisition time of 30 s, and an excitation power of 5 mW.

2.3. Photocatalytic Performance Tests

For the photodegradation measurement, 4 mg of photocatalysts (DI water, benzyl, and methanol) was immersed in 20 mL of reactive black 5 (Sigma Aldrich, Schnelldorf, Germany, Molar Mass = 991.82 g/mole, CAS number 306452) dye solution. The photocatalytic study of the three samples (DI water, benzyl, and methanol) was achieved by measuring the photo absorption in the wavelength range [450–700 nm]. The initial concentration of reactive black 5 dye was 10 mg/L. The absorption spectrum of the dye was measured solely before adding the catalyst. After adding the catalyst, the solution was placed in the dark for 15 h to exclude the adsorption effect from the photodegradation process, see Figure S3. To measure UV–Vis absorbance spectra of reactive black 5 dye, a Genesys 10 s UV-VIS spectrophotometer with a scanning rate of 5.0 nm/s was utilized. This system was also used to measure the UV-Vis absorbance spectra of photocatalysts in an ethanol suspension. To excite the photo-electron generation in the photocatalyst, a high-intensity discharge 400 W iron-doped metal halide UV bulb with a wavelength range [315–400 nm] with the model (UV 400 HL230 Fe Clear-A, UV Light Technology Limited, Birmingham, UK) was used. The light was irradiated to a Pyrex reactor with an intensity of 100 mW/cm². The photodegradation efficiency of the dye was determined by recording the variation of the maximum absorbance intensity located at a wavelength of 605 nm.

3. Results and Discussion

3.1. TEM

The morphology of the LN structures for different solvents, DI water, methanol, and benzyl, was determined using TEM microscopy; see Figure 1. The TEM images show more dispersed particles in DI water than in methanol and benzyl. The particles are more agglomerated in benzyl than methanol. The higher resolution of the images permitted us to extract the average size of the nanoparticles in the three samples. The average size was extracted from the statistics of the size distribution (see Figure S1). The average size of particles μ= 2.95, 5.97, and 4.65 nm, and the standard deviation are σ= 0.42, 0.48, and 2.13 nm for DI, benzyl, and methanol, respectively. This confirms that the size is smaller in DI water than in methanol and benzyl. It is clearly seen that methanol has a greater deviation in size distribution. The shape is spherical for benzyl and DI water. However, it is a square shape for methanol.

3.2. XRD

Figure 2 shows XRD patterns of LN powders synthesized by methanol, benzyl, and DI water. A commercial powder was used as a reference for comparison and verification. The XRD peaks observed in the commercial powder are attributed to the crystal planes (012), (104), (110), (006), (113), (202), (024), (116), (122), (018), (214), (300), and (208) that belong to the LiNbO₃ phase (space group R3c, JCPDS No. 020-0631) [18,27]. The crystal planes of LiNbO₃ were observed in methanol. However, this is not the case for benzyl and DI water. The peaks are sharper in the case of methanol than in benzyl and DI water. This is due to the large size of nanostructures observed in methanol, as shown in the TEM image (see Figure 2). The broadness of DI and benzyl peaks is due to the very small size effect, as shown in TEM images. In addition, XRD patterns show that the crystallinity order decreases from methanol to benzyl. This confirms that the formation of LN powders is not complete in benzyl. The crystal planes of the LiNbO₃ phase in benzyl are (202) and (214). There also are other peaks observed, but they are very weak and broad. In DI water, the crystal planes are (012), (104),(110), (113), (116), (214), and (310). For benzyl, the crystal planes are (202) and (300). We can also observe the niobium oxide (Nb₂O₅) phase marked by an asterisk [18,20,28]. This phase is observed more in methanol and less in benzyl and DI water. There are some patterns attributed to an unknown phase in methanol; it could be that they are attributed to the Li₃NbO₄ [27,29] phase; this has been confirmed by Raman spectroscopy and reported in different studies [29]. The latter confirms the presence of multiphase in LN fabricated in methanol solvent.

3.3. Raman Spectroscopy

Raman spectroscopy was performed on LN nanostructures prepared with the three solvents and the commercial for comparison. When comparing the Raman spectra of the synthesized LN to the commercial sample, we were able to confirm the presence of the LiNbO₃ phase in the three samples (see Figure 3). All the peaks of LN observed in the commercial are observed in the methanol sample. The Raman peak intensities attributed to the LiNbO₃ phase in DI are much higher than benzyl. This observation correlates well with the XRD patterns. It is clearly seen the peaks related to the LiNbO₃ phase are blue- and red-shifted due to distortions, the incorporation of ions, stoichiometry effects, etc. The peaks assigned to LiNbO₃ are located at 151 cm⁻¹, 236 cm⁻¹, 320 cm⁻¹, 370 cm⁻¹, 430 cm⁻¹, 580 cm⁻¹, 625 cm⁻¹, and 875 cm⁻¹ [18,27]. The peak located at 151 cm⁻¹ corresponds to Nb-O bonds (E-TO), while 236 cm⁻¹ and 320 cm⁻¹ are assigned to NbO₆ octahedra deformation (E-TO). The Raman bands centered at 370 cm⁻¹ and 430 cm⁻¹ are associated with the bending modes of the Nb-O-Nb bond (E-TO). The Raman peak located at 625 cm⁻¹ is attributed to the symmetric stretching of Nb-O-Nb bonds (A₁-TO). The Raman band at 875 cm⁻¹ corresponds to the antisymmetric stretching of Nb-O-Nb (E-LO). The peak located around 904 cm⁻¹ belonging to LiNbO₃ (phase) is attributed to octahedral distortions of surface species such as $\mathrm{O}=\mathrm{N}\mathrm{b}\cdots \mathrm{O}\mathrm{H}$. All the phonon modes related to LiNbO₃ phase are present in methanol; however, some peaks within the spectral range [150, 400 cm⁻¹] are absent in DI water and benzyl. Some peaks attributed to Li₃NbO₄ are observed in methanol [27]. The LiNb₃O₈ phase is observed in benzyl and DI water [27,30]. This confirms the formation of polycrystalline structures in the three solvents.

The Raman peak located at 236 cm⁻¹ was adjusted with a Voight function to extract the FWHM and Raman peak variation for the three solvents (see Figure S2). DI water was shifted more toward the higher phonon energy; this could be due to its very small size effect. However, for benzyl and methanol, they were shifted closely to the commercial sample. Concerning the FWHM features, the three solvents have slightly equal values and are broader than the commercial. This explains how the broadness of the Raman line was affected by the crystallinity’s order and the presence of dislocations. If we look profoundly into the FWHM variation, it clearly shows that methanol and DI exhibited broader peaks than benzyl. In conclusion, methanol promoted formation in the majority of the LiNbO₃ phase with the very small contribution of Li₃NbO₄; however, DI water and benzyl promoted formation in the majority of LiNbO₃, with a very small contribution of LiNb₃O₈. Raman also shows the appearance of a broad Raman band located around 500 and 800 cm⁻¹ in methanol and benzyl; this could be due to the presence of an amorphous phase.

3.4. UV-Vis Absorbance

The UV-Vis absorption spectra of the LN fabricated with different solvents, DI water, benzyl, and methanol, in a suspension are shown in Figure 4. The obtained absorption spectra of the three solvents exhibit a maximum absorbance that is below 300 nm, which is due to the electronic transition from the orbital Sp of Nb covalence to the orbital d conduction bands [12,25]. The absorption coefficient $\alpha$ was calculated using the Beer–Lambert law relation.

$$\mathrm{I}={\mathrm{I}}_0\times {\mathrm{e}}^{-\mathsf{\alpha }\mathrm{t}}$$

(1)

Meanwhile, the absorption coefficient (α) is determined by the formula, A is the absorbance, and t is the thickness.

$$\alpha ={\displaystyle \frac{2.303\times A}{t}}$$

(2)

The energy band gaps were calculated using Tauc’s relation and is given by

$${\left(\alpha h\nu \right)}^{1/n}=A(h\nu -E_g)$$

(3)

where n = 1/2 for direct band gap semiconductors, h is the constant Planck, $\nu$ is the photon frequency, and E_g is the optical band gap. The band gap for DI, benzyl, and methanol is presented in

Table 1. Band gap values of LN fabricated with DI water, benzyl, and methanol solvents.

Samples	Band Gap (eV)
DI water	3.6
Methanol	3.9
Benzyl	3.17

. Benzyl shows a smaller band gap, while methanol has a larger one, and that of DI is between both. The strong effect of the solvent on the optical properties of LN is clearly seen. This variation could be due to the size, shape, facet, and defect effects of crystal structures. The peaks observed in the absorbance spectra located at 240, 246, and 258 nm, for methanol, DI, and benzyl correspond to the excitonic peaks that prove the crystallinity of nanoparticles.

3.5. Photoluminescence

Figure 5 shows the photoluminescence spectra of the three samples: DI water, methanol, and benzyl. It is clearly seen that the PL intensity is much higher in benzyl than in DI water and methanol. The intensity of methanol is higher than that of DI water. These features confirm that DI water LN has a slower recombination rate of electron–hole pairs than methanol and benzyl. This observation matches well with the photodegradation efficiency, where DI water exhibited a higher kinetic rate than methanol and benzyl. Different PL peaks located at 420, 430, 477, 487, 520, 580, and 631 nm are observed. Our observations are consistent with those of some reported studies [31,32,33]. These peaks, located within the wavelength range of [350, 700 nm], promoted the excitation of electrons from valence bands to intermediate levels (trapping, defects, etc.) and, in turn, these can recombine to the hole present in the valence band or climb to the conduction band. Then, they can de-excite to the intermediate levels, and then again go down to the valence band.

3.6. Photocatalytic Study

Figure 6a–c show the absorbance spectra versus wavelength for the different irradiation time varying from 0 to 280 min for the three samples. A decrease in absorbance is expected when compared to the absorbance of the dye; this is due to the adsorption effect that takes place in the dark. The decrease in the absorbance is much faster in the case of DI water than methanol and benzyl. The photodegradation efficiency η of the reactive black 5 dye was calculated using Equation (1):

$$\eta \mathrm{\%}={\displaystyle \frac{C_0-C_t}{C_0}}\times 100$$

(4)

where $C_0$ and $C_t$ represent the initial and instantaneous concentration of reactive black 5 dye at irradiation time t.

The maximum of the absorbance peak, located at 604 nm, was used to estimate the photodegradation efficiency for all the spectra. Figure 6d shows the degradation efficiency of the three samples. The degradation efficiency reached 95, 24, and 16% for DI water, methanol, and benzyl, respectively. A big difference was observed in the efficiency of DI water compared to that of methanol and benzyl.

The decrease in the concentration of the dye as a function of the time can be expressed in the first-order reaction kinetics of the Langmuir–Hinshelwood model with the linear relationship of ln(C_t/Co) versus the irradiation time t using the following equation [34,35,36]:

$$\ln \left({\displaystyle \frac{C_t}{C_o}}\right)=-kt+K$$

(5)

where $C_o$ is the dye concentration at t = 0 without irradiation, $C$ is the rest dye concentration after the irradiation time (t), k is the first-order reaction rate constant, and K is the degradation equilibrium constant. The semi-logarithmic relationship of the dye concentration versus the irradiation time is depicted in Figure 6e. The kinetics of the reaction rate constant and the decomposition equilibrium constant can be extracted from Equation (2). The constants k and K were extracted and are displayed in

Table 2. First-order reaction rate constants (k) and the degradation equilibrium constant (K) for DI water, benzyl, and methanol.

Sample	k1 (min−1)	k2 (min−1)	K1	K2	R2
DI water	3 × 10−3	2.7 × 10−2	0.0716	4.91	0.973
Benzyl	5.81 × 10−4	0	0.0346	0	0.960
Methanol	5.88 × 10−4	0	0.110	0	0.980

. It was found that k values are 3 × 10⁻³, 5.81 × 10⁻⁴, and 5.88 × 10⁻⁴ min⁻¹ for DI water, benzyl, and methanol, while K values are 0.0716, 0.0346, and 0.110, respectively. The kinetic of degradation is much faster in DI water than in methanol and benzyl. For the case of DI, the process of the kinetic of degradation was established in two steps. The first was between 0 and 180 min, with k₁ = 3 × 10⁻³ min⁻¹, and the second was between 200 and 280 min, with k₂ = 2.7 × 10⁻² min⁻¹. The latter confirms how the photodegradation speed increased at 180 min. This analysis confirmed the speed of the kinetic process of photodegradation in DI compared to the other solvents. This origin could be due to different effects: the large surface area of LN fabricated with DI; the presence of the LiNb₃O₈ phase, which showed its capability to increase the photocatalytic activities in different studies [30]; and the large size of benzyl and the presence of an amorphous phase, which affected its photocatalytic activity. For methanol, the low kinetic of photodegradation could be due to its large size and band gap. A PL study confirmed that DI has lower emissions than methanol and benzyl. In addition, the PL of methanol is lower than benzyl. The results correlate well with the photodegradation efficiency. This can be explained by the slow recombination rate of electron–hole observed in DI water. Table 3 shows a comparative study between the first-order reaction rate constants (k) of LN fabricated in our work and other materials, such as LiNbO₃, LiNb₃O₈, BaTiO₃, and TiO₂, in terms of their size. It is clearly seen that the very small size of our LN nanostructures exhibited a higher kinetic reaction rate than the other with the presence of reactive black 5, which is more complex than other tested dyes.

Table 4. Recapitulative table of some parameters extracted from characterization studies and photodegradation for the sake of discussion.

Samples	Average Size/nm	Main Crystalline Phase	Residual Phases	Optical Band Gap/eV	Pseudo First-Order Rate Kinetic /min−1
DI water	2.95	LiNbO3	LiNb3O8	3.6	2.7 × 10−2
Benzyl	5.97	LiNbO3	LiNb3O8	3.17	5.81 × 10−4
Methanol	4.36	LiNbO3	Li3NbO4	3.9	5.88 × 10−4

shows a summary of the parameters (the average size, main crystalline phase, residual phase, band gap, and first-order rate kinetic k) extracted from the structural, optical, and the photodegradation studies. DI showed a smaller average size, benzyl showed a larger one, and methanol was between both. This correlates well with their band gap values, except for methanol, where it was contradictory. Raman and XRD spectra showed that methanol contains residual phases such as Li₃NbO₄, which, in a previous study, showed a band gap of 4–4.08 eV [37]. This could explain why methanol exhibited a larger band gap than others. Other factors, such as the crystallinity, shape, etc., could also contribute. On the other hand, the values of the kinetic reaction rate constant k correlate well with the variation in size; this confirms the importance of the effect of the surface area on the photocatalytic mechanism. Table 4 also shows the appearance of LiNbO₃ as the major phase in all; this was confirmed by XRD and Raman spectroscopy. XRD and Raman peaks of the LN phase totally appeared in methanol; however, this was not the case for DI and benzyl, where they lack some XRD patterns and Raman modes. The difference between XRD and Raman is that Raman exhibited the appearance of a contribution of the LN phase in the case of benzyl and DI more than XRD. XRD and Raman also showed the appearance of residual phases such as LiNb₃O₄ in methanol and LiNb₃O₈ in DI and Benzyl. The difference between XRD and Raman spectra is that Raman exhibited the appearance of a contribution of LiNb₃O₈ in DI and benzyl and Li₃NbO₄ phases in methanol more than XRD. It is known that Raman spectroscopy is very similar to anisotropy and polymorphic crystals. Finally, we concluded that the smaller size of nanoparticles in DI, its higher crystallinity, and its constituent of the LiNb₃O₈ phase compared to benzyl and methanol made it an excellent solvent for the fabrication of efficient photocatalysts. Another factor that increased its photocatalytic reaction mechanism is its band gap of 3.6 eV located in the middle of the wavelength range of the light source. Although the different factors enhanced the catalytic activity of the DI sample, the very small size obtained from the fabrication using DI water remains the dominant factor. This work paves the way to further improve the synthesis of very small nanoparticles with higher catalytic activity based on the solvent-change strategy.

Table 3. Comparison between pseudo first-order rate kinetic k in this work and those of other reported works.

References	Samples	Average Size	Dye	Pseudo First-Order Rate Kinetic /min−1
This work	LN nanostructures	2.95 nm	Reactive black 5	0.027
C. Yu et al. [38]	BaTiO3 nanowires	200 nm	Methyl orange	0.0011
D. Mazkad et al. [39]	LiNbO3	100 μm	Rhodamine	0.0020
H. Zhai et al. [30]	LiNb3O8	100 nm	Methylene blue	0.016
F. Azeez et al. [40]	Pure TiO2	10 nm	Methylene blue	0.018

Table 3. Comparison between pseudo first-order rate kinetic k in this work and those of other reported works.

References	Samples	Average Size	Dye	Pseudo First-Order Rate Kinetic /min−1
This work	LN nanostructures	2.95 nm	Reactive black 5	0.027
C. Yu et al. [38]	BaTiO3 nanowires	200 nm	Methyl orange	0.0011
D. Mazkad et al. [39]	LiNbO3	100 μm	Rhodamine	0.0020
H. Zhai et al. [30]	LiNb3O8	100 nm	Methylene blue	0.016
F. Azeez et al. [40]	Pure TiO2	10 nm	Methylene blue	0.018

4. Conclusions

Very small nanostructures of LN were successfully synthesized with an average size of around 4–10 nm with the solvothermal method using three solvents—DI water, methanol, and benzyl. Structural studies confirmed the presence of a LiNbO₃ phase in all the powders with the small presence of Li₃NbO₄ and LiNb₃O₈ phases. Methanol promoted the majority formation of LiNbO₃ with a very small contribution of Li₃NbO₄. However, DI water and benzyl promoted the majority formation of LiNbO₃ with a very small contribution of LiNb₃O₈. Benzyl showed a lower-order crystallinity than other solvents and confirmed the incomplete formation of LN nanocrystals. The optical band gap was extracted from the Tauc plot equation with values of 3.9, 3.6, and 3.17 eV for methanol, DI, and benzyl, respectively. The photodegradation of reactive black 5 dye experiments showed that DI nanopowders exhibited faster kinetics of degradation than methanol and benzyl due to their large surface area and the presence of the LiNb₃O₈ phase. The latter was confirmed by the slow recombination rate of electron–hole pairs proved by PL. This work paves the way to further improve the synthesis of very small nanoparticles of LN with higher catalytic activity based on the solvent-change strategy.