Structures, formation mechanisms, and ion-exchange properties of α-, β-, and γ-Na₂TiO₃

Abstract

α-, β-, and γ-Na₂TiO₃ were prepared from rutile TiO₂ and molten NaOH. Three models of β-Na₂TiO₃ with space groups of R3̄, P1̄, and P3̄ were proposed, and the R3̄ model was refined from the experimental data by using the Rietveld method. The structure of β-Na₂TiO₃ is a superstructure of α-Na₂TiO₃ and supposedly contains Ti₆O₁₉ clusters. The structures of Na₂TiO₃ were mainly determined by the particle sizes of rutile and the reaction temperatures. α-Na₂TiO₃ could be prepared from fine rutile particles (D(50) < 25.8 μm) and molten NaOH at 500 °C or quenching the melt of Na₂TiO₃ at 1000 °C quickly. γ- and β-Na₂TiO₃ were the thermodynamically stable phases of Na₂TiO₃ at around 500 °C and above 800 °C, respectively. α-Na₂TiO₃ was formed far beyond the thermodynamically stable state. The Na⁺ in α-Na₂TiO₃ was easier to exchange with H⁺ in water than that in β or γ phases. They all converted to amorphous phases after the 2nd, 6th, and 4th water washings at 25 °C, respectively. β-Na₂TiO₃ followed similar paths of ion-exchange as α-Na₂TiO₃, which was different from that of γ-Na₂TiO₃.

---

1. Introduction

Various kinds of sodium titanates have been prepared for use as cation exchangers,^1,2 photocatalysts,^3,4 and precursors to prepare TiO₂ nanotubes.^5,6 Most researches have focused on the nanostructures of titanates consisting of negatively charged layered sheets constructed with TiO₆ octahedra and obtained by a hydrothermal method.^7,8 Moreover, sodium titanates can also be prepared from the reaction of Na₂CO₃ and TiO₂.^9,10 There are fewer studies of Na₂TiO₃ (sodium meta-titanate) compared to other sodium titanates. However, Na₂TiO₃ is receiving more attention because of its possible applications for the adsorption of radioactive nuclides,^11,12 CO₂ capture at high temperatures,¹³ electrolysis for producing titanium,¹⁴ and preparation of H₂TiO₃–lithium adsorbents.¹⁵ Furthermore, Na₂TiO₃ hydrolyzes upon contact with water, which could generate NaOH.^16–18 This makes Na₂TiO₃ an ideal intermediate for producing TiO₂ or upgrading titania slag by a process that involves alkaline roasting of titanium minerals and water washing for NaOH recycling.^19–24 As an example, our group used this property of Na₂TiO₃ to develop the molten NaOH process for TiO₂ production.²⁴

According to the phase diagrams of the Na₂O–TiO₂ system shown in Fig. 1,²⁵ Na₂TiO₃ could be obtained from the system of Na₂O–TiO₂ with a TiO₂ content of 40–60 wt% at 400–1000 °C. α-, β-, and γ-Na₂TiO₃ were prepared from the reaction of TiO₂ and Na₂CO₃ or NaOH;^26–28 however, these preparation methods are not very reproducible. α-Na₂TiO₃ exhibits the NaCl-type structure (space group Fm3̄m, Z = 4/3); Na and Ti atoms occupy the same sites in the unit cell with 2/3 and 1/3 probability, respectively.^17,29 γ-Na₂TiO₃ has a chain structure consisting of the rarely observed trigonal-bipyramidal TiO₅ units (space group C2/c, Z = 4).²⁸ The XRD pattern of β-Na₂TiO₃ reported as PDF 00-047-0130 (a = 3.195 Å, c = 7.676 Å) was attributed to the rhombohedrally distorted NaCl-type subcell;³⁰ however, the superlattice reflections were wrongly omitted from this card in the PDF database. The corrected pattern appeared as PDF 00-050-110 (space group R3̄; a = b = 13.927 Å, c = 7.676 Å; Z = 19). β-Na₂TiO₃ was primarily treated as a monoclinic form, indexed as PDF 00-037-0345,²⁶ however, the rhombohedral indexing was more reliable than the monoclinic one.³¹ However, 19 is not a multiple of 3, as is necessary for the rhombohedral lattice. Therefore, some Na/Ti disorder was suggested,³¹ and it was refined in this study.

Fig. 1 Phase diagrams for the Na₂O–TiO₂ system: (a) TiO₂ content <60%, and (b) TiO₂ content >60%.²⁵

There are some misunderstandings related to the structures of Na₂TiO₃. For instance, Na₆Ti₂O₇ in Fig. 1 should be γ-Na₂TiO₃.²⁶ Other kinds of sodium titanates prepared by a hydrothermal method were wrongly identified as Na₂TiO₃ in some studies.^32,33 A sodium titanate coating formed on the titanium surface by NaOH treatment was thought to be Na₂TiO₃.³⁴ However, the X-ray diffraction (XRD) patterns could not be indexed to the three phases discussed above; a likely structure is that of Na₂Ti₅O₁₁.³⁵ Additionally, the space group of α-Na₂TiO₃ was mistaken for Immm.¹³

The properties among the three kinds of Na₂TiO₃ structures were found to be different. In the molten NaOH process for TiO₂ production as mentioned above, the crystal structures of Na₂TiO₃ affected the Ti yield coefficient, and α-Na₂TiO₃ was the ideal phase that could be prepared from fine natural rutile.²¹ However, the reaction mechanisms of molten NaOH and TiO₂ in forming α-, β-, and γ-Na₂TiO₃ remained unclear, especially for the transformations among the three phases.^26,27 Moreover, many further applications of Na₂TiO₃ must consider the ion-exchange properties, e.g., the molten NaOH process for TiO₂ production need Na₂TiO₃ with good ion-exchange ability for the recycle of NaOH medium. Thus, a comparison of the ion-exchange properties in water should be studied.

In this study, α-, β-, and γ-Na₂TiO₃ were successfully prepared from the reaction of molten NaOH and rutile TiO₂. The structure of β-Na₂TiO₃ was refined. The effects of particle sizes of rutile and reaction temperatures were studied to reveal the formation and transformation mechanisms of α-, β-, and γ-Na₂TiO₃. Their ion-exchange properties in water were investigated for comparison.

2. Experimental

2.1 Materials

Analytical-grade NaOH, rutile TiO₂, and deionized water were used throughout the experiment. Natural rutile concentrate (claret colour, Hubei province, China) was used as a substitute for rutile TiO₂ at different particle sizes. The chemical composition (weight fraction) of the natural rutile was 90.62% TiO₂, 4.01% SiO₂, 1.49% ∑Fe, 0.54% MgO, and 1.16% Al₂O₃. The main phase of natural rutile was rutile TiO₂.

2.2 Experimental details

NaOH pellets were ground into powders in a mortar made of WC–Co (tungsten carbide 84.5 wt% and cobalt 15.5 wt%) with the protection of N₂ in the glove-box. NaOH powder and rutile TiO₂ were mixed in the mortar with a molar ratio of 2.2 : 1. The mixture was put into the nickel crucibles that were then heated in a muffle furnace. α- and γ-Na₂TiO₃ were prepared by heating the mixture of NaOH and TiO₂ at 500 °C for 90 min. β-Na₂TiO₃ was prepared at 800 °C for 90 min and cooled slowly to room temperature. Excessive NaOH was used to guarantee the complete reaction. In preparing α- and β-Na₂TiO₃, analytical-grade TiO₂ powder with an average particle size (D(50)) of 0.5 μm was used. γ-Na₂TiO₃ was prepared from the coarse TiO₂ particles with secondary particle sizes of about 100 μm, obtained from prilling the above TiO₂ powder.

After the reaction, the products were ground and washed with ethanol to remove a part of excess NaOH in order to avoid the moisture of the samples. γ-Na₂TiO₃ sample was contaminated by a small amount of α-Na₂TiO₃ which was converted to an amorphous form by additional washes with ethanol.

In the ion-exchange tests, α-, β-, and γ-Na₂TiO₃ were washed with water in a 250 mL flask heated by an oil bath with a liquid/solid mass ratio of 3 : 1 at 25 °C and 60 °C. Each washing lasted for 30 min, and the pulp was filtered after each washing. Then, the filter cake was sampled and dried for further analysis. The removal of Na⁺ in the water washing was used to characterize the ion-exchange abilities of α-, β-, and γ-Na₂TiO₃ in water.

2.3 Characterisation

XRD patterns were obtained using an X-ray diffractometer (X'Pert PRO MPD, PANalytical, Netherlands), recorded within 5–90° 2θ using CuKα radiation. The phase semi-quantification of Na₂TiO₃ was obtained using the GSAS-EXPGUI software packages with the Rietveld method.^36,37 In phase semi-quantification of Na₂TiO₃, the zero-point correction, unit cell parameters, scale factor, background, and peak width were refined, while the atomic coordinates, occupancies, and thermal parameters were fixed. The weight fraction (w_i) for each phase was obtained from the refinement relation:where i is a particular phase among the N phases, S_i is the refined scale factor, Z is the number of formula units per cell, M is the formula mass, and V is the volume of the unit cell.³⁸

The particle size distributions of the TiO₂ precursors were measured using a laser particle analyser (Mastersizer 2000, Malvern Instruments, UK).

The reactions of NaOH and rutile TiO₂ were investigated in a microfurnace by Raman spectrometer (LABRAM HR800, Horiba Jobin Yvon) in situ. It was equipped with an intensive charge-coupled device (ICCD), and the pulsed exciting light (514.5 nm) from a Q-switch pulsed THG-Nd:YAG laser was focused using an Olympus BH-2 microscope. To investigate the ion-exchange process of Na₂TiO₃, the resulting slurry after the water washing was placed on a small glass plate to collect the Raman spectra.

The particle morphologies and elemental compositions of the products were characterized using a scanning electron microscope (SEM; JSM-7001F, Electron Company, Japan) coupled to energy dispersive X-ray spectroscopy (EDS; INCA X-MAX, Oxford Instruments, US).

Thermogravimetry (TG) and differential scanning calorimetry (DSC) experiments (STA 449C, NETZSCH, Germany) were performed from 25 °C to 900 °C in a Pt crucible.

The chemical compositions of the samples and solutions were examined by inductively coupled plasma atomic emission spectroscopy (ICP-OES; Optima 5300DV, PerkinElmer, USA).

3. Results and discussion

3.1 Crystal structures and characterisation of Na₂TiO₃

As mentioned in the introduction, the structures of α- and γ-Na₂TiO₃ have been well studied. α- and γ-Na₂TiO₃ were successfully prepared in this study. The XRD patterns after lattice refinement and ball-stick models (created by the VESTA³⁹) of α- and γ-Na₂TiO₃ are shown in Fig. 2. All XRD peaks were well indexed.

Fig. 2 XRD patterns after the refinement and ball-stick models (inset) of α-Na₂TiO₃ (a), β-Na₂TiO₃ (b, space group of R3̄), and γ-Na₂TiO₃ (c) (red, O; yellow, Na; blue, Ti).

As suggested by V. B. Nalbandyan,³¹ there may be Na/Ti mixing in some positions of β-Na₂TiO₃ with the space group of R3̄. The original atomic coordinates were obtained from the procedures shown in Fig. S1† by Density Functional Theory (DFT) calculations and revising the atomic site of Na (0, 0, 0) by 2/3 Na and 1/3 Ti. Lattice parameter refinement was done firstly, and the obtained simple R3̄ model (its atomic sites were shown in Table S1†) was used to do structural refinement and create the P1̄ and P3̄ models in the following text. Then structural refinement on the XRD data in Fig. 2b was done. The atomic coordinates, occupancies, and thermal parameters were all optimized. The results were shown in Tables 1 and 2. The structural residual factors from Rietveld refinement were small. In Fig. 2b, all XRD peaks were well indexed with this R3̄ model, and its Miller indices were added.

Table 1 Crystallographic data of β-Na₂TiO₃ after refinement

Formula	Na38Ti19O57
Space group	R3̄
Crystal system	Trigonal
Lattice parameters	a = b = 13.9252(5) Å, c = 7.7024(3) Å
Gaussian (G) and Lorentzian (L) profile coefficients	GU = 0, GV = 45.48, GW = 18.76, LX = LY = 1.00
Reliability factors	Rp = 7.03%, Rwp = 9.48%, χ^2 = 2.426

---

Table 2 Atomic positions, occupancies, and thermal parameters of β-Na₂TiO₃ with space group of R3̄ after refinement

Atom	Site	x/a	y/b	z/c	Occ.	Uiso
O1	18f	0.6796	0.4540	0.5257	1.0000	0.0525
O2	18f	0.0799	0.8491	0.5019	1.0000	0.0173(7)
O3	18f	0.2234	0.9642	0.7946	1.0000	0.0120(4)
O4	3b	0.6667	0.3333	0.8333	1.0000	0.0252
Na1	18f	0.4253	0.3669	0.0110	0.9444	0.0251(1)
Ti1	18f	0.4253	0.3669	0.0110	0.0556	0.0251(7)
Na2	18f	0.1442	0.1238	0.3326	0.1111	0.0276(7)
Ti2	18f	0.1442	0.1238	0.3326	0.8889	0.0276(8)
Na3	3a	0.0000	0.0000	0.0000	0.3333	0.0306
Ti3	3a	0.0000	0.0000	0.0000	0.6667	0.0305(5)
Na4	18f	0.1089	0.8467	0.0109	1.0000	0.0106(6)

---

As shown in Table 2, there are another two Na/Ti mixing positions in the unit cell except for (0, 0, 0). The ball-stick model of β-Na₂TiO₃ (space group R3̄) in Fig. 2b indicates that it is not a framework or layered structure, and it contains Ti₆O₁₉ clusters some of which are connected into Ti₁₃O₃₈ dimers via edge sharing with the Ti2 octahedron. However, the Ti2 site is partially substituted by Na2 and this precludes formation of infinite oxotitanate chains along the c direction. When the clusters are stacked according to the face centred lattice, the NaCl-type crystal structure will be obtained. Thus, β-Na₂TiO₃ can be considered as a superstructure of α-Na₂TiO₃ which is in accord with the prediction of V. B. Nalbandyan.³¹

Table 3 shows the average bond lengths of Na/Ti and O atoms in β-Na₂TiO₃. The ionic radii sums of six-coordination Na(+1) and O(−2), Ti(+4) and O(−2) are 2.42 and 2.005 Å, respectively.⁴⁰ The average bonds lengths in Table 3 were all close to the predicted values. According to the Bond Valence model,^41,42 the central Na distributes its valence for six neighboured O atoms, and the calculated Na–O bond length is 2.47 Å, which is very close to the bond length of Na4–O (2.40 Å). Similarly, the calculated Ti–O bond length for the central six-coordination O atom from Bond Valence model is 2.22 Å. For the O4 atom, it is surrounded by six Na2/Ti2 mixing positions with the Ti occupancy of 0.889. The bond length of Na2/Ti2–O4 is 2.28 Å, a little larger to the calculated value. The structure is highly disordered due to substitution of cations that differ significantly in size. Then, positions of nominally identical anions contacting with them should also be disordered, and some tabulated distances should not be taken for real interatomic distances.

Table 3 The average bond lengths of Na/Ti and O atoms in β-Na₂TiO₃ after refinement (Å)

Atom 1	Atom 2	Average bond length	Prediction from ionic radii sums^a
a Prediction = 2.42 × occupancy(Na) + 2.005 × occupancy(Ti).
Na1/Ti1	O	2.39	2.40
Na2/Ti2	O	2.02	2.05
Na3/Ti3	O	2.18	2.14
Na4	O	2.40	2.42

---

To obtain the β-Na₂TiO₃ models with definite atomic sites (the need for DFT calculations to study its properties in future studies), another two models with space groups P1̄ and P3̄ were obtained by geometric transformation from the rhombohedral and hexagonal cell of the R3̄ model with the 2/3 Na and 1/3 Ti at site (0, 0, 0) in Table S1.† For this simple R3̄ model of β-Na₂TiO₃, the bond valence calculations in Tables S2 and S3† indicated that this structure was reasonable. For the P1̄ model, the lattice parameters are Z = 19, a = b = 8.440 Å, c = 19.689 Å, α = 85.23°, β = 94.77°, and γ = 68.83°. For the P3̄ model, Z = 57, a = b = 13.925 Å, c = 23.107 Å. Their atomic sites obtained from DFT calculations are shown in Tables S4 and S5.† Thus, these two models were not directly proved by the Rietveld analysis which should be done in the future study.

The morphologies of α-, β-, and γ-Na₂TiO₃ were slightly different. α-Na₂TiO₃ had a more porous surface and a larger specific surface area (see Fig. S2 and S3†).

Fig. 3 shows the Raman spectra of α-, β-, and γ-Na₂TiO₃, and they are in agreement with a previous report.²⁷ α- and β-Na₂TiO₃ have similarities in their Raman spectra. However, the poor disordered structure because of the random distribution of Na and Ti atoms endows α-Na₂TiO₃ with very broad Raman peaks.²⁷ β- and γ-Na₂TiO₃ have clearly different and numerous sharply resolved peaks. The peaks at 500–1000 cm⁻¹ are mostly connected to Ti–O–Ti stretching vibrations, and the peak at 800–815 cm⁻¹ can be ascribed to the difference in Ti–O bond lengths.⁴³ The band at approximately 1080 cm⁻¹ is related to the CO₃²⁻ symmetric stretching vibration from Na₂CO₃ that remained in the three samples.⁴⁴ The peaks at approximately 454 and 657 cm⁻¹ are assigned to the Ti–O–Ti stretching in the edge-shared TiO₆ units in β-Na₂TiO₃, while the peaks at approximately 176 and 296 cm⁻¹ are attributed to the Na–O–Ti bending modes.^45,46 For γ-Na₂TiO₃, the peaks at approximately 385 and 632 cm⁻¹ indicate the edge-shared TiO₅ units.⁴⁷

Fig. 3 Raman spectra of the prepared α-, β-, and γ-Na₂TiO₃.

3.2 Formation and transformation of α-, β-, and γ-Na₂TiO₃

3.2.1. Effect of the particle sizes of rutile TiO₂. To investigate the effect of the particle sizes of the rutile TiO₂ precursor, natural rutile which were milled to D(50) = 2.9, 9.1, 25.9, 58.5, 67.5, 76.2, 85.3, and 103.0 μm substituted for rutile TiO₂. The XRD patterns of the products and the quantification analysis of the α-Na₂TiO₃ phase in each product are shown in Fig. 4. When the D(50) of natural rutile was smaller than 25.9 μm, the α-Na₂TiO₃ phase formed, and when D(50) > 58.5 μm, γ-Na₂TiO₃ was the main phase of the product with about 15 wt% of α-Na₂TiO₃.

Fig. 4 XRD patterns (a) and weight fraction of α-Na₂TiO₃ (b) in the products obtained from natural rutile with different average particle sizes.

Raman spectra were collected continuously during the reactions of NaOH and rutile of different average particle sizes, as shown in Fig. 5. The rutile crystals belong to the space group of P4₂/mnm with Raman active modes occurring around 143 cm⁻¹ (B_1g), 445 cm⁻¹ (E_g), 612 cm⁻¹ (A_1g), and 243 cm⁻¹ (second order Raman scattering) in Fig. 5.⁴⁸ As shown in Fig. 5a and b, the E_g mode shifted to lower frequencies, the band at 243 cm⁻¹ shifted to higher frequencies, while the B_1g and A_1g modes remained unchanged with the increasing temperature (<380 °C). The Raman measurements were performed at ambient conditions; thus, NaOH could readily absorb CO₂ from the air and form Na₂CO₃. The intense band at 1081–1074 cm⁻¹ is attributed to Na₂CO₃,⁴⁴ and the bands shifted to lower frequencies with increasing temperatures. In both Fig. 5a and b, the Raman active modes of rutile slowly disappeared with the appearance of the bands of α- and γ-Na₂TiO₃ at 806 and 798 cm⁻¹, respectively. Blue shifts occurred at above 380 °C compared to the Raman spectra at room temperature (Fig. 3). During the reaction of NaOH and rutile with different average particle sizes, no intermediate phases were observed during the formation of the α- and γ-Na₂TiO₃ phases.

Fig. 5 In situ Raman spectra of the mixture of NaOH and rutile from 30 °C to 500 °C. Natural rutile (a) D(50) = 2.9 μm and (b) D(50) = 76.2 μm.

TG/DSC measurements were used to study the details of the reactions with rutile of different particle sizes. The TG graphs in Fig. 6 show a continuous weight loss at about 300–600 °C. This weight loss can be related to the release of water from the reaction. The end of this weight loss indicated the end of the reaction. Fig. 6 indicates that fine rutile requires less time to complete the reaction, which means that fine rutile particles could improve the reaction rate. There are only two obvious endothermic peaks at about 62 °C and 282 °C in the DSC curve, corresponding to the evaporation of free water and the melt of the NaOH + Na₂CO₃ eutectic mixture,⁴⁹ respectively. No other exothermic or exothermal peaks were observed below 600 °C. Thus, it is reasonable to conclude that there is no crystal transformation between α- and γ-Na₂TiO₃ in the reactions below 600 °C.

Fig. 6 TG/DSC curves of the mixture of NaOH and natural rutile from 25 °C to 900 °C. Natural rutile (a) D(50) = 2.9 μm and (b) D(50) = 76.2 μm.

The XRD patterns of the product obtained at different reaction time are shown in Fig. 7. The main peak of rutile at approximately 27.5° is close to the main peak of γ-Na₂TiO₃ at 27.6–27.8°. This shows that the rutile phase decreased with the increased reaction time, and γ-Na₂TiO₃ formed gradually. A very small amount of α-Na₂TiO₃ formed at the beginning of the reaction, and its amount remained nearly constant in the reaction. Thus, the crystal transformation did not occur between α- and γ-Na₂TiO₃ in the reaction, which was in agreement with Fig. 6.

Fig. 7 XRD patterns of the products obtained from natural rutile D(50) = 76.2 μm for different reaction time.

The product was formed by inter-diffusion upon thermal activation.⁵⁰ Fig. 8a is backscattered SEM of the intermediate product after 10 min being embedded in the resin and polished, while Fig. 8b is the general SEM of the same intermediate product. The surface of the rutile particle reacted with NaOH at the beginning of the reaction, while the interior part of the particle remained unchanged (see EDS analysis in Fig. 8c and d). A compact product layer consisting of the sodium titanate formed and wrapped the surface of the rutile particle.

Fig. 8 Backscattered SEM (a) and general SEM (b) of the intermediate product at 10 min prepared from natural rutile D(50) = 85.3 μm; (c) and (d) EDS analysis of spot ① and ② in (a) and (b), respectively.

In conclusion, at 500 °C, the fine rutile particles of D(50) < 25.8 μm reacted with NaOH very fast, thus favoured the formation of α-Na₂TiO₃ in which Na and Ti atoms were distributed randomly. Conversely, the product layer formed on the surface of rutile of larger particle sizes made the diffusion of NaOH slow and difficult, thus the coarse rutile particles obtained γ-Na₂TiO₃ in which the atoms were orderly arrayed.

3.2.2. Effect of temperatures. As shown in Fig. 9a and b, α- and γ-Na₂TiO₃ changed into β-Na₂TiO₃ after heating at 800 °C, which means that β-Na₂TiO₃ is the thermodynamically stable phase at above 800 °C. There was an obvious endothermic peak at 815 °C in Fig. 6. Since the phase transition to β-Na₂TiO₃ could occur at 750 °C according to our previous study,²¹ 815 °C may be a eutectic point of Na₂CO₃ + Na₂TiO₃. After slowly cooling the melt of β-Na₂TiO₃ at 1000 °C, the product remained as β-Na₂TiO₃, while quenching the melt yielded α-Na₂TiO₃ (see Fig. 9c and d).

Fig. 9 XRD patterns of the product obtained by heating Na₂TiO₃ at different temperatures. (a) α-Na₂TiO₃, 800 °C; (b) γ-Na₂TiO₃, 800 °C; (c) β-Na₂TiO₃, 1000 °C, slow cooling; (d) β-Na₂TiO₃, 1000 °C, quenching. (e) Fine rutile powder mixed with a 50 wt% NaOH solution at ambient conditions, 500 °C.

Fig. 10 provides a summarized schematic illustration for the formation and transformation of α-, β-, and γ-Na₂TiO₃. γ- and β-Na₂TiO₃ are the thermodynamically stable phases of Na₂TiO₃ at approximately 500 °C and above 800 °C, respectively. α-Na₂TiO₃ can be prepared from the reaction of fine rutile particles and NaOH at 500 °C or quenching the melt of Na₂TiO₃ at 1000 °C quickly. It means that the fast kinetics of these processes prevents the Na₂O–TiO₂ system from reaching the thermodynamically stable state, and leads to the random distribution of Na and Ti atoms, thus forming the α-Na₂TiO₃ phase. It is reasonable to conclude that α-Na₂TiO₃ is formed far beyond the thermodynamically stable state. To verify the above deduction, TiO₂ powder of D(50) = 1.5 μm was mixed evenly with a 50 wt% NaOH solution with a NaOH/TiO₂ molar ratio of 2.2 : 1 in a nickel crucible and placed in a muffle furnace at 250 °C; the furnace was then heated to 500 °C for 90 min. Compared to the preparation of Fig. 2a, the addition of water accelerated the diffusion of Na₂O, and allowed the O and Na atoms to reach the ordered array (thermodynamically stable state) more easily, which could break through the impediment of the reaction kinetics. In this case, γ-Na₂TiO₃ was obtained, as shown in Fig. 9e.

Fig. 10 Schematic illustration for the formation and transformation of α-, β-, and γ-Na₂TiO₃.

3.3 Ion-exchange properties of α-, β-, and γ-Na₂TiO₃

Na₂TiO₃ remained unstable during the water washing because the Na⁺ in Na₂TiO₃ easily underwent ion-exchange with the H⁺ in water.^17,22 This can be treated as the hydrolysis of Na₂TiO₃, which is described below:

Na₂TiO₃ + xH₂O → Na_(2−x)H_xTiO₃ + xNaOH (2)

As shown in Fig. 11a, the removal of Na⁺ increased with the increasing of water washing times at 25 °C. Na⁺ in β-Na₂TiO₃ exchanged slower than that in the α- and γ-Na₂TiO₃. The solubility of Ti(IV) was less than 0.001 mol L⁻¹,⁵¹ which means that almost all Ti remained in the solids. Further study shows that the removal of Na⁺ after the 5th water washing at 65 °C were 84.5%, 79.5%, and 82.2% for α-, β-, and γ-Na₂TiO₃, respectively. The ion-exchange abilities can be arranged as follows: α-Na₂TiO₃ > γ-Na₂TiO₃ > β-Na₂TiO₃.

Fig. 11 (a) The removal of Na⁺ of Na₂TiO₃ for different water washing times at 25 °C. XRD patterns of (b) α-, (c) β-, and (d) γ-Na₂TiO₃ after different water washing times and calcined at 250 °C.

XRD and Raman measurements were used to examine the samples after each water washing. Fig. 11 shows that α-, β-, and γ-Na₂TiO₃ converted amorphous phases after the 2nd, 6th, and 4th water washings at 25 °C, respectively. As shown in Fig. 11a, the exchange of Na⁺ was almost the same (about 70%) for the above washed sample (α-2, β-6, and γ-4) which was very interesting.

For α- and β-Na₂TiO₃, the XRD peaks of the washed samples in Fig. 11 were broadened and shifted, which may indicate that the lattice parameters changed because of a partial H substitution for Na. α-1 and β-5 exhibited similar peaks. α-1 had a typical FCC pattern, while β-5 was the same but with a rhombohedral distortion and a superstructure. However, the peaks of γ-Na₂TiO₃ became weaker and disappeared during the washing process, and no intermediate phase was found.

As shown in Fig. 12, the original Raman peaks of α-, β-, and γ-Na₂TiO₃ became weaker during the water washing. For α-Na₂TiO₃, the peak at 819 cm⁻¹ moved to 884 and 888 cm⁻¹ after the first and second water washings, and its strength rapidly decreased in the following washing. The Raman spectra of β-Na₂TiO₃ in the 5th and 6th water washings were very similar to that of α-Na₂TiO₃ in the first two water washings. The exchange of Na⁺ led to a symmetry reduction and a loosening of the structures, causing the blue shifts of these peaks and the distortions of the TiO₆ units in α- and β-Na₂TiO₃. The peaks at 880–900 cm⁻¹ corresponded to the very-short Ti–O bond between the terminal O atom and the central Ti atom of the distorted TiO₆ octahedra.⁵² Our previous study showed that the layered structures formed during the ion-exchange process of α-Na₂TiO₃.¹⁷ β-Na₂TiO₃ was treated as the superstructure of α-Na₂TiO₃, and both had TiO₆ units. Deduced from the XRD and Raman analysis, β-Na₂TiO₃ may break down at the beginning of the ion exchange and follow similar paths of ion-exchange with α-Na₂TiO₃. For γ-Na₂TiO₃, there were no new XRD or Raman peaks during the water washing. The rare TiO₅ units caused it to follow completely different paths during the ion-exchange process, which will be studied in depth in our future study.

Fig. 12 Raman spectra of the slurry obtained from washing (a) α-, (b) β-, and (c) γ-Na₂TiO₃ for different times at 25 °C. Numbers indicate the water washing times.

4. Conclusions

α-, β-, and γ-Na₂TiO₃ were prepared from rutile TiO₂ and molten NaOH. Three models of β-Na₂TiO₃ with the space group of R3̄, P1̄, and P3̄ were proposed, and the R3̄ model was refined from the experimental data by Rietveld method. The structure of β-Na₂TiO₃ is a superstructure of α-Na₂TiO₃ and supposedly contains Ti₆O₁₉ clusters.

The crystal structures of Na₂TiO₃ were mainly determined by the particle sizes of rutile and the temperatures in the reaction of rutile and NaOH. Fine particle sizes of rutile favoured the formation of α-Na₂TiO₃. There was no crystal transformation between α- and γ-Na₂TiO₃ in the reactions below 600 °C. α-Na₂TiO₃ could be prepared from the reaction of fine rutile particles (D(50) < 25.8 μm) and molten NaOH at 500 °C or quenching the melt of Na₂TiO₃ at 1000 °C quickly. γ- and β-Na₂TiO₃ were the thermodynamically stable phases of Na₂TiO₃ at around 500 °C and above 800 °C, respectively. Thus, α-Na₂TiO₃ was formed far beyond the thermodynamically stable state.

The Na⁺ in α-Na₂TiO₃ was easier to exchange with H⁺ in water than that in β or γ phases. They all converted to amorphous phases after sufficient water washing. β-Na₂TiO₃ followed the similar paths of ion-exchange as α-Na₂TiO₃ in water washing process, which was different from that of γ-Na₂TiO₃.

Acknowledgements

This work was supported by the National Basic Research Program of China (Grant No. 2013CB632604), the National Science Foundation for Distinguished Young Scholars of China (Grant No. 51125018), the Key Research Program of the Chinese Academy of Sciences (Grant No. KGZD-EW-201-2), and the National Natural Science Foundation of China (Grant No. 51374191 and 51402303).

References

1. J. Huang, Y. Cao, Z. Deng and H. Tong, J. Solid State Chem., 2011, 184, 712–719.
2. D. Nepak, RSC Adv., 2015, 5, 47740–47748.
3. I. S. Grover, S. Singh and B. Pal, RSC Adv., 2014, 4, 51342–51348.
4. X. Fu, D. Y. C. Leung and S. Chen, CrystEngComm, 2014, 16, 616–626.
5. H. Zhang, L. Cao, W. Liu and G. Su, Appl. Surf. Sci., 2012, 259, 610–615.
6. H. Lu, RSC Adv., 2015, 5, 89777–89782.
7. D. V. Bavykin, J. M. Friedrich and F. C. Walsh, Adv. Mater., 2006, 18, 2807–2824.
8. Y. Zhang, Z. Jiang, J. Huang, L. Y. Lim, W. Li, J. Deng, D. Gong, Y. Tang, Y. Lai and Z. Chen, RSC Adv., 2015, 5, 79479–79510.
9. A. Rudola, K. Saravanan, S. Devaraj, H. Gong and P. Balaya, Chem. Commun., 2013, 49, 7451–7453.
10. C. Y. Xu, J. Wu, P. Zhang, S. P. Hu, J. X. Cui, Z. Q. Wang, Y. D. Huang and L. Zhen, CrystEngComm, 2013, 15, 3448–3454.
11. I. El-Naggar, E. Mowafy, I. Ali and H. Aly, Adsorption, 2002, 8, 225–234.
12. I. Ali, J. Radioanal. Nucl. Chem., 2004, 260, 149–157.
13. P. Sanchez-Camacho, I. C. Romero-Ibarra, Y. Duan and H. Pfeiffer, J. Phys. Chem. C, 2014, 118, 19822–19832.
14. K. Zhao, Y. W. Wang, S. H. Tao and N. X. Feng, Adv. Mater. Res., 2013, 734, 2430–2433.
15. D. Tang, D. Zhou, J. Zhou, P. Zhang, L. Zhang and Y. Xia, Hydrometallurgy, 2015, 157, 90–96.
16. F. Meng, T. Xue, Y. Liu, W. Wang and T. Qi, Hydrometallurgy, 2016, 161, 112–116.
17. Y. Liu, W. Zhao, W. Wang, X. Yang, J. Chu, T. Xue, T. Qi, J. Wu and C. Wang, J. Phys. Chem. Solids, 2012, 73, 402–406.
18. F. C. Meng, T. Y. Xue, Y. H. Liu, G. Z. Zhang and Q. I. Tao, Trans. Nonferrous Met. Soc. China, 2016, 26, 1696–1705.
19. A. J. Manhique, W. W. Focke and C. Madivate, Hydrometallurgy, 2011, 109, 230–236.
20. S. Middlemas, Z. Z. Fang and P. Fan, Hydrometallurgy, 2013, 131–132, 107–113.
21. F. Meng, Y. Liu, J. Chu, W. Wang and T. Qi, Can. J. Chem. Eng., 2014, 92, 1346–1352.
22. Y. Li, Y. Yang, M. Guo and M. Zhang, RSC Adv., 2015, 5, 13478–13487.
23. F. Meng, D. Wang, T. Xue, W. Wang, Y. Liu and T. Qi, Asia-Pac. J. Chem. Eng., 2016, 11, 14–23.
24. Y. Liu, F. Meng, F. Fang, W. Wang, J. Chu and T. Qi, Dyes Pigm., 2016, 125, 384–391.
25. R. S. Roth, T. Negas, L. P. Cook and G. Smith, Phase Diagrams for Ceramists, The American Ceramic Society, 1981.
26. W. Antony Hill, A. R. Moon and G. Higginbotham, J. Am. Ceram. Soc., 1985, 68, C-266–C-267.
27. C. E. Bamberger and G. M. Begun, J. Am. Ceram. Soc., 1987, 70, C-48–C-51.
28. S. Mao, X. Ren and Z. Zhou, Chin. J. Struct. Chem., 2008, 27, 553–557.
29. Y. Li, H. Y. Yu, Z. T. Zhang, M. Zhang and M. Guo, ISIJ Int., 2015, 55, 134–141.
30. V. B. Nalbandyan and I. L. Shukaev, Zh. Neorg. Khim., 1990, 35, 1085–1086.
31. V. Nalbandyan, Russ. J. Inorg. Chem., 2000, 45, 1652–1658.
32. C. Xu, Y. Zhan, K. Hong and G. Wang, Solid State Commun., 2003, 126, 545–549.
33. D. S. Seo, J. K. Lee, E. G. Lee and H. Kim, Mater. Lett., 2001, 51, 115–119.
34. S. Y. Kim, K. K. Yu, I. S. Park, G. C. Jin, T. S. Bae and H. L. Min, Appl. Surf. Sci., 2014, 321, 412–419.
35. H. M. Kim, F. Miyaji, T. Kokubo and T. Nakamura, J. Biomed. Mater. Res., 1996, 32, 409–417.
36. A. C. Larsen and R. B. von Dreele, General Structure Analysis System (GSAS), Los Alamos National Laboratory Report LAUR, 2004, pp. 86–748.
37. B. H. Toby, J. Appl. Crystallogr., 2001, 34, 210–213.
38. S. K. Rout and S. Panigrahi, Indian J. Pure Appl. Phys., 2006, 44, 606–611.
39. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
40. R. D. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32, 751–767.
41. I. D. Brown, The Chemical Bond in Inorganic Chemistry. The Bond Valence Model, 2002.
42. I. D. Brown and D. Altermatt, Acta Crystallogr., Sect. A: Found. Crystallogr., 1985, 41, 244–247.
43. Y. V. Kolen'ko, K. A. Kovnir, A. I. Gavrilov, A. V. Garshev, J. Frantti, O. I. Lebedev, B. R. Churagulov, G. Van Tendeloo and M. Yoshimura, J. Phys. Chem. B, 2006, 110, 4030–4038.
44. B. Buchholcz, H. Haspel, Á. Kukovecz and Z. Kónya, CrystEngComm, 2014, 16, 7486–7492.
45. B. C. Viana, O. P. Ferreira, A. G. Souza Filho, A. A. Hidalgo, J. Mendes Filho and O. L. Alves, Vib. Spectrosc., 2011, 55, 183–187.
46. X. Meng, D. Wang, J. Liu and S. Zhang, Mater. Res. Bull., 2004, 39, 2163–2170.
47. S. K. Barbar and M. Roy, J. Mol. Struct., 2012, 1024, 132–135.
48. K. Tsuzuku and M. Couzi, J. Mater. Sci., 2012, 47, 4481–4487.
49. G. W. Morey and J. S. Burlew, J. Phys. Chem., 1964, 68, 1706–1712.
50. M. Jansen, Angew. Chem., Int. Ed., 2002, 41, 3746–3766.
51. X. B. Li, S. W. Yu, N. K. Liu, Y. K. Chen, T. G. Qi, Q. S. Zhou, G. H. Liu and Z. H. Peng, Hydrometallurgy, 2014, 147, 73–78.
52. D. Yang, H. Liu, L. Liu, S. Sarina, Z. Zheng and H. Zhu, Nanoscale, 2013, 5, 11011–11018.

---

Footnote

† Electronic supplementary information (ESI) available: Procedures of obtaining the structures of β-Na₂TiO₃; atomic sites and bond valence calculations of β-Na₂TiO₃; morphologies of the prepared α-, β-, and γ-Na₂TiO₃. See DOI: 10.1039/c6ra16984h

---