High-Pressure Behavior and Thermal Stability of Water-Bearing TiO₂-II Formed by Phase Transition of Natural Rutile

Abstract

In situ high-pressure Raman spectroscopy (up to 38 GPa) and thermal stability analysis (up to 900 °C) based on Raman and infrared spectroscopy were performed on water-bearing TiO₂-II. The samples were synthesized from natural rutile through pressure-induced phase transition experiments. The results demonstrate that at ambient temperature, water-bearing TiO₂-II transforms into akaogiite at 12–15 GPa, which is higher than the phase-transition pressure reported in previous studies for anhydrous TiO₂-II samples transforming to akaogiite (approximately 10 GPa). Hydroxyl groups at different positions within the TiO₂-II structure exhibit distinct thermal stabilities. Although the water-bearing TiO₂-II sample underwent partial dehydration between 300 and 400 °C under atmospheric pressure, the majority of the hydroxyl content persisted at higher temperatures (even after the phase transition from TiO₂-II to rutile). At ambient pressure, water-bearing TiO₂-II transforms to rutile at 500–600 °C, which is significantly lower than the previously reported transition temperature (730 °C) for anhydrous TiO₂-II. The partial dehydration of water-bearing TiO₂-II during heating may account for the phase transition occurring at a reduced temperature.

1. Introduction

TiO₂ is a technologically important material with diverse industrial applications. Naturally occurring TiO₂ exists in three primary polymorphs: rutile (tetragonal, P4₂/mnm), anatase (tetragonal, I4₁/amd), and brookite (orthorhombic, Pbca) [1,2]. Under atmospheric pressure, anatase and brookite transform to rutile at approximately 600–700 °C [3]. Consequently, among these three polymorphs of TiO₂, rutile is the most abundant in nature due to its greater thermal stability [2]. As an accessory mineral in ultrahigh-pressure metamorphic rocks, rutile participates in Earth’s deep-water cycle through structural hydrogen incorporation [4,5,6]. Under elevated pressures, TiO₂ exhibits several high-pressure polymorphs: TiO₂-II (orthorhombic, Pbcn), akaogiite (monoclinic, P2₁/c), TiO₂-OI (orthorhombic, Pbca), and TiO₂-OII (orthorhombic, Pnma) [2,7,8,9]. Experimental studies under high-temperature and high-pressure conditions have revealed that TiO₂-II can be stable above 2 GPa in subducting slabs. Its occurrence—either as discrete grains coexisting with rutile or as nano-crystalline lamellae within rutile hosts—serves as an indicator of high-pressure conditions reaching the diamond stability field in subduction zones [3,10,11,12].

The compressibility and pressure-induced phase transitions of rutile, anatase, and brookite have been extensively studied by using in situ high-pressure X-ray diffraction and Raman spectroscopy [2,13,14,15,16]. The research findings indicated that the phase transition processes under high pressure for these TiO₂ polymorphs are inconsistent. Swamy et al. [14] demonstrated that anatase gradually transforms into TiO₂-II and akaogiite (baddeleyite-type TiO₂) at 7–9 GPa, completing its transition to pure akaogiite at approximately 17 GPa. After this phase transition, akaogiite retains its crystal structure at pressures of up to 40 GPa. Liu et al. [2] revealed that Nb-bearing natural brookite undergoes a phase transition to akaogiite at 12 GPa while simultaneously transforming into TiO₂-II and minor rutile. Akaogiite maintains its structural integrity in a single-phase state between 21 and 33 GPa but transforms further into TiO₂-OI when the pressure exceeds 33 GPa. High-pressure spectroscopic studies indicated that rutile begins transitioning to akaogiite above 10 GPa, completing the transformation at approximately 16–18 GPa [15]. Gerward and Olsen [13] confirmed that after the rutile-to-akaogiite transition, the latter maintains stability below 60 GPa.

Since TiO₂-II is relatively rare in nature and cannot be synthesized artificially under ambient pressure conditions, research on its high-pressure behavior presents greater challenges compared with studies of rutile, anatase, and brookite. Wang et al. [16] synthesized Al-bearing and Al-free TiO₂-II samples using a multi-anvil press and conducted in situ high-pressure spectroscopic studies on them at up to 20 GPa. Their research revealed that Al-free TiO₂-II undergoes a phase transition to akaogiite at pressures as low as 10 GPa. In contrast, Al-bearing TiO₂-II exhibits no structural changes at significantly higher pressures. For example, the Ti_0.957(AlH)_0.043O₂ phase of TiO₂-II requires a pressure of approximately 20 GPa to transform into akaogiite. Furthermore, Wang et al. [16] also demonstrated that both Al-bearing and Al-free TiO₂-II revert to rutile at temperatures above 680 °C under ambient pressure. However, the high-pressure behavior and thermal stability of TiO₂-II samples (especially natural samples derived from subducting slabs) with distinct compositions require further experimental investigation.

In this study, in situ high-pressure Raman spectroscopy (up to 38 GPa) and thermal stability analysis (up to 900 °C) were conducted on water-bearing TiO₂-II. The samples were synthesized from natural Al-free rutile in subducted continental crust through pressure-induced phase transition experiments using diamond anvil cells (DACs). Consequently, these specimens serve as reliable proxies for natural TiO₂-II by precisely replicating its inherent chemical signatures and physical properties. The experimental results reveal the influence of water (hydrogen) incorporation on the high-pressure/high-temperature stability of TiO₂-II, contributing to a deeper understanding of the existing form of TiO₂ within Earth’s interior.

2. Sample Preparation and Experimental Methods

TiO₂-II samples were synthesized from natural rutile through pressure-induced phase transition experiments. The rutile sample was extracted from ultrahigh-pressure eclogite (No. B199R176P6i) within subducted continental crust [17]. Its chemical composition was analyzed using a JEOL JCM-7000 benchtop scanning electron microscope at Center for High Pressure Science and Technology Advanced Research, Beijing, China. The results (Figure S1) demonstrate that the sample is nearly pure TiO₂, with only trace Fe impurities detected. To analyze the water content in rutile, unpolarized infrared spectra (Figure 1) were collected using a Bruker INVENIO-R FTIR spectrometer equipped with a 20× objective on a Bruker HYPERION 1000 microscope, a CaF₂ beam-splitter, and an MCT detector at Center for High Pressure Science and Technology Advanced Research, Beijing, China. Water content was determined through the Lambert–Beer Law, where water concentration (c, mol·L⁻¹) was calculated as c = (A/t)/ε, with A representing the measured absorbance, t denoting the crystal plate thickness (cm), and ε being the integrated molar absorption coefficient (L·mol⁻¹·cm⁻¹) [18]. The total absorbance of a crystal is the sum of absorbance over its three polarization directions. Therefore, for non-polarized infrared spectroscopy measurements, the total absorbance of the OH bands is estimated by multiplying the measured absorbance by a factor of three [19,20].

The synthesis experiment for TiO₂-II utilized a symmetric-type diamond anvil cell (DAC) equipped with a pair of type Ia diamond anvils (culet diameter: ~0.3 mm). A rhenium gasket was pre-indented to a thickness of ~60 μm, and a sample chamber ~0.15 mm in diameter was drilled. Due to the limitations of the sample chamber dimensions, the grains of rutile were crushed into fragments (ranging from 30 to 60 μm in size). For each pressure-induced phase transition experiment, a fragment was loaded into the sample chamber with ruby spheres, and the chamber was backfilled with neon as the pressure-transmitting medium. Based on pressure calibration via the shift of the ruby R1 luminescent line [21], the rutile fragment was first compressed to 18–19 GPa and then gradually decompressed to ambient pressure. During decompression to ambient pressure, the rutile fragment transformed into TiO₂-II. To obtain a sufficient quantity of TiO₂-II samples (as well as backup samples), this experiment was repeated multiple times. The recovered TiO₂-II samples (Figure 2) were subsequently used for further experiments in this study. The infrared spectrum (acquired using the same method as for the rutile sample) of the TiO₂-II sample exhibited strong absorption peaks in the 2800–3600 cm⁻¹ region (Figure 3), indicating that water in rutile can be retained in a TiO₂-II sample after undergoing a pressure-induced phase transition to TiO₂-II.

For high-pressure Raman spectroscopic measurements, a symmetric-type diamond-anvil cell containing a pair of type Ia diamond anvils with 0.3 mm culet diameters was employed to compress the TiO₂-II sample. A chip featuring flat surfaces (approximately 40 μm × 30 μm × 30 μm in size), along with two ruby spheres for pressure calibration, was selected and loaded into a rhenium gasket hole measuring 150 μm in diameter and 50 μm in depth. Neon was loaded as a pressure-transmitting medium at 2.56 GPa. Raman spectra were acquired using a WITec confocal Raman imaging microscope (WITec alpha300R) equipped with a 20× microscope objective at Center for High Pressure Science and Technology Advanced Research, Beijing, China. All spectra were measured within the 100 to 1000 cm⁻¹ wavenumber range, utilizing a 1800 gr/mm grating, five accumulations, 10 seconds of exposure time, and 40 mW laser power. Pressure was determined via the ruby R1 fluorescence line shift [21]. The measurements spanned the 2.4 to 38.6 GPa range during both compression and decompression cycles.

For thermal stability analysis, the fragments of TiO₂-II were placed on sapphire sheets and subjected to successive half-hour heating treatments in a muffle furnace at 100, 200, 300, 400, 500, 600, 700, 800, and 900 °C. Following each heating cycle, the crystals were removed and subjected to Raman and infrared spectroscopic analyses after cooling to room temperature. Subsequent heating procedures were continued after the analyses were completed. Raman spectra were collected by a WITec confocal Raman imaging microscope (WITec alpha300R) with a 100× microscope objective. All spectra were recorded in the 100 to 1000 cm⁻¹ wavenumber range using a 1800 gr/mm grating, five accumulations, 10 seconds of exposure time, and 30 mW laser power. Infrared spectra were recorded using a Bruker INVENIO-R FTIR spectrometer with a Bruker HYPERION 1000 microscope (20× objective) over the 2400–4000 cm⁻¹ range at Center for High Pressure Science and Technology Advanced Research, Beijing, China. Both background and sample spectra were acquired with 128 scans.

3. Results and Discussion

3.1. High-Pressure Behavior of TiO₂-II

The Raman spectrum of the sample obtained from the pressure-induced phase transition experiment under ambient conditions exhibited scattering peaks at 170, 277, 311, 339, 352, 422, 529, and 568 cm⁻¹ (Figure 4). These spectral features match those reported in previous studies for TiO₂-II samples synthesized via compression experiments [15], confirming the successful transformation of natural rutile into TiO₂-II.

As shown in Figure 5 and Figure 6 and Table S1, within the pressure range from 2.56 to 11.98 GPa, the Raman scattering peaks of the TiO₂-II sample exhibit a linear shift toward higher frequencies with increasing pressure. This indicates that although the crystal structure of TiO₂-II undergoes compression under high-pressure conditions, it maintains its structural integrity within this pressure range. The Raman spectral characteristics of TiO₂-II undergo significant changes when the pressure reaches 15.31 GPa. The linear shift of the original Raman scattering peaks is interrupted, accompanied by the emergence of a series of new scattering peaks represented by strong peaks at 224, 244, 270, and 407 cm⁻¹ (Figure 5 and Figure 6). These modifications signify the initiation of a phase transition from TiO₂-II to akaogiite. Upon reaching 16.91 GPa, the Raman spectrum of TiO₂-II evolves further, displaying characteristic scattering peaks at 164, 226, 246, 272, 321, 360, 412, 444, 490, 520, 638, 709, and 826 cm⁻¹ (Figure 5 and Figure 6). This confirms the complete transition to akaogiite [2,15].

The in situ high-pressure Raman spectroscopy analysis conducted by Wang et al. [16] on synthetic TiO₂-II revealed that an anhydrous TiO₂ sample began to transform into akaogiite at approximately 10 GPa. In contrast, samples containing Al and H in their structure required higher pressures for this phase transition to occur. Specifically, the sample with the chemical formula Ti_0.96(AlH)_0.04O₂ initiated the transformation at ~19 GPa, while the sample with the formula Ti_0.93(AlH)_0.07O₂ exhibited no transition to akaogiite within the studied pressure range (0–20 GPa).

In this study, the pressure required for the Al-free TiO₂-II sample to begin transforming into akaogiite (12–15 GPa) is significantly lower than that needed for the two previously reported Al-bearing samples to undergo the same phase transition. This also demonstrates that Al substantially elevates the pressure threshold for this transformation. A study by Escudero and Langenhorst [22] on Al incorporation into high-pressure polymorphs of TiO₂ (at 10–20 GPa and 1300 °C) demonstrated that Al exhibits lower affinity for akaogiite compared to TiO₂-II. This finding may explain how Al extends the stability field of TiO₂-II to higher pressures. Furthermore, in the study by Wang et al. [16], the Al-bearing TiO₂-II samples were hydrous. Therefore, Al can enter the TiO₂-II structure via substitution of Ti⁴⁺ by Al³⁺ coupled with H⁺. This coupled substitution mechanism can suppress the formation of oxygen vacancies associated with the direct substitution of Ti⁴⁺ by Al³⁺. Such oxygen vacancies may destabilize TiO₂-II at higher pressures [22].

The transformation pressure (12–15 GPa) observed for the Al-free TiO₂-II sample in this study is notably higher than that (10 GPa) reported by Wang et al. [16]. The size of the TiO₂-II crystal (~40 μm) used for high-pressure Raman spectroscopy analysis in this study is comparable with that of the TiO₂-II crystals (30–100 μm) used by Wang et al. [16]. The key difference between the TiO₂-II sample in this work and the pure TiO₂ sample in that of Wang et al. [16] lies in the presence of water (≤0.12 wt% H₂O) in our sample. In water-free samples, the presence of a small amount of Ti³⁺ could result in the simultaneous formation of oxygen vacancies in the crystal structure. In water-bearing samples, however, H⁺ can partially compensate for the charge imbalance resulting from the reduction of Ti⁴⁺ to Ti³⁺ [23]. Therefore, it is possible that the oxygen vacancies in the water-bearing sample investigated in this study are fewer than those in the previously reported water-free sample. And this may be one of the possible reasons why the water-bearing sample undergoes a phase transition to akaogiite at higher pressures. Whether other factors (such as crystal orientation) can exert similar influence on the pressure conditions for this phase transition requires further experimental investigation (e.g., in situ high-pressure Raman spectroscopy and XRD analysis) to elucidate.

At 31.60 GPa, the Raman spectrum of akaogiite exhibits significant changes, characterized by the appearance of new scattering peaks at 366, 479, and 634 cm⁻¹, indicating the onset of its phase transition to TiO₂-OI (Figure 5 and Figure 6). When the pressure reaches 33.89 GPa, the sample’s Raman spectrum shows scattering peaks at 201, 255, 262, 291, 332, 370, 409, 455, 485, 515, 604, 639, 669, 694, 732, and 794 cm⁻¹ (Table S1). These spectral features confirm the complete transformation from akaogiite to TiO₂-OI [2]. In the pressure range of 33.89 to 38.55 GPa (the maximum pressure studied), while these peaks continue to shift with increasing pressure, no further significant changes occur in the Raman spectrum characteristics (Figure 5 and Figure 6), suggesting no additional phase transitions within this range.

Previous studies have shown that pure akaogiite retains its crystal structure at pressures of up to 40 GPa [14], whereas Nb-bearing akaogiite begins transforming into TiO₂-OI at approximately 33 GPa [2]. The pressure observed in this study for the phase transition from akaogiite to TiO₂-OI (31.60 GPa) is close to that required for the transition in Nb-bearing akaogiite (33 GPa). However, considering that the samples reported in the previous studies [2,14] were finely ground powders, the influence of water on this phase transition requires further experiments on water-bearing powdered samples to delineate.

As shown in Figure 5, the Raman spectral characteristics indicate that TiO₂-OI maintains its structural integrity at 30.11 GPa during decompression, whereas it exhibits instability at the same pressure during pressurization. Furthermore, akaogiite persists at 13.10 GPa during decompression, though unstable at this pressure during pressurization (Figure 5). This indicates that the pressure conditions required for the two sequential phase transitions of TiO₂-II occurring from ambient pressure to 38.55 GPa during pressurization do not fully correspond to those necessary for their respective reversed phase transitions during decompression. Due to reconstructive phase transitions involving the breaking and reconstruction of primary chemical bonds, such transitions require overcoming a substantial thermodynamic barrier [24]. Consequently, initiating the forward transition during compression necessitates overpressure, while initiating the reverse transition during decompression requires underpressure. This pressure asymmetry likely constitutes the primary driver of the observed irreversibility in the phase transition path.

3.2. Thermal Stability of TiO₂-II

The Raman spectral features of the TiO₂-II samples recovered after heating to 100, 200, 300, 400, and 500 °C remain consistent with those of the untreated sample (Figure 7). The primary scattering peaks of TiO₂-II (observed at 170, 277, 311, 339, 352, 422, 529, and 568 cm⁻¹) persisted after heating to 500 °C, with no significant changes in their relative intensities. Furthermore, no new peaks appeared in the Raman spectra of the recovered samples (Figure 7). These results indicate that the TiO₂-II sample retained structural stability at temperatures of up to 500 °C under ambient pressure. Compared with the untreated TiO₂-II sample, the recovered TiO₂-II sample after heating to 600 °C shows significant changes in the Raman spectral features. The Raman spectrum of the recovered sample exhibits three new strong scattering peaks at 255, 436, and 609 cm⁻¹, which are characteristic of the rutile phase [15,16]. These observations indicate that the phase transformation from TiO₂-II to rutile initiates within the temperature range of 500–600 °C. In the Raman spectrum of the sample recovered after heating to 700 °C, the scattering peaks (170, 311, 339, and 352 cm⁻¹) previously observed in the Raman spectrum of TiO₂-II disappear (Figure 7). This change suggests the completion of the phase transformation from TiO₂-II to rutile.

The infrared spectral characteristics of the TiO₂-II sample in this study (Figure 3) differ from those reported in previous studies for Al-bearing TiO₂-II samples [16]. Although the hydroxyl vibration peak at approximately 2940 cm⁻¹, documented in prior research [16], was also observed in the infrared spectrum of the TiO₂-II sample in this study (Figure 3), the spectrum of our sample additionally exhibited a series of broad and strong absorption peaks within the 3000–3600 cm⁻¹ region (Figure 3). In mineral crystal structures, O-H stretching frequencies are governed by O-H···O hydrogen bond lengths. Higher wavenumbers of O-H bands typically correspond to weaker hydrogen bonding, while lower wavenumbers indicate stronger hydrogen bonding [25]. During the synthesis of TiO₂-II in this study, the original tetragonal rutile sample undergoes multiple phase transitions under pressure cycling (compression and decompression) to ultimately form the lower-symmetry orthorhombic TiO₂-II phase [15]. This process involves the breaking and reconstruction of primary chemical bonds and the rearrangement of hydrogen atoms. Consequently, compared with the hydrogen-bonding configurations within the original rutile, those in the resultant TiO₂-II phase are likely more complex. Therefore, the presence of the peak at 2940 cm⁻¹ alongside broad absorptions across 3000–3600 cm⁻¹ suggests that hydroxyl groups occupy multiple nonequivalent sites with distinct hydrogen-bonding environments in the crystal structure.

The infrared spectral features of the water-bearing TiO₂-II recovered after heating to 100, 200, and 300 °C remained consistent with those of the untreated sample (Figure 8). After heating to 400 °C, the vibrational peak at 2900–3000 cm⁻¹ disappeared, while the peaks in the range of 3000–3600 cm⁻¹ persisted. This demonstrates that hydroxyl groups at different positions within the TiO₂-II crystal structure exhibit varying thermal stability. After the phase transformation of TiO₂-II to rutile (after heating to 600 °C), the peaks in the 3000–3600 cm⁻¹ range remained visible (Figure 8). However, their peak heights decreased after heating to 700 °C, indicating dehydration of the sample. After heating to 900 °C, the infrared spectrum of the sample was characterized by a relatively sharp vibrational peak at 3290 cm⁻¹ (Figure 8). The position of this peak aligns with the hydroxyl vibrational band observed in the infrared spectra of the natural rutile samples used in this study (Figure 2) and those reported in previous studies [23,26].

High-temperature Raman spectroscopy analysis by Wang et al. [16] on synthesized TiO₂-II samples revealed that, under ambient pressure, anhydrous pure TiO₂ samples undergo a phase transition to rutile at 730 °C. By comparison, the temperature required for the phase transition to rutile of the water-bearing TiO₂-II sample in this study (500–600 °C) is significantly lower (Figure 7). This implies that water (hydrogen) incorporated into TiO₂-II can alter the temperature conditions for the TiO₂-II-to-rutile phase transition. According to Wang et al. [16], hydrous Al-bearing TiO₂-II samples transformed into rutile at approximately 630–680 °C. Prior to this phase transition, the samples underwent dehydration within a temperature range of approximately 530–630 °C. This dehydration process was marked by a significant decrease in the intensity of the OH band, which was located at 2900–3000 cm⁻¹ at room temperature. Similarly, the Al-free TiO₂-II sample investigated in this study underwent partial dehydration at 300–400 °C before the phase transition (500–600 °C). This dehydration is characterized by the disappearance of the OH band at about 2900–3000 cm⁻¹ (Figure 8). This indicates that under high-temperature conditions, dehydration may exert a more pronounced effect on the stability of TiO₂-II than Al content does.

The high-pressure polymorphs of SiO₂ are key components in subducted oceanic crust and serve as important carriers of water (hydrogen), transporting it into Earth’s deep interior [27,28]. α-PbO₂-type SiO₂ (seifertite), which shares the same structure as TiO₂-II [29,30], is likely the dominant SiO₂ polymorph existing at the base of the lower mantle (110–130 GPa). The experimental results from this study suggest that the potential implications of water (hydrogen) incorporation in seifertite on its thermal stability should be fully considered in future investigations regarding the form of SiO₂ at the base of the lower mantle.

4. Conclusions

(1)

At room temperature, water-bearing TiO₂-II (with water content of up to 0.12 wt% H₂O) transitions to akaogiite at pressures of approximately 12–15 GPa. Although this transition pressure is lower than the values previously reported for hydrous Al-bearing samples (>19 GPa), it is higher than those reported for anhydrous pure TiO₂ samples (10 GPa). This suggests that water (hydrogen) can affect the pressure stability of TiO₂-II.

(2)

Hydroxyl groups at distinct sites within the crystal structure of the TiO₂-II sample display differing thermal stabilities. Although the sample underwent partial dehydration between 300 and 400 °C under atmospheric pressure, marked by the disappearance of the ~2900–3000 cm⁻¹ OH band, the majority of the hydroxyl content persisted at higher temperatures (even after the phase transition from TiO₂-II to rutile).

(3)

The transition temperature from water-bearing TiO₂-II to rutile observed in this study is at least 130 °C lower than that reported for anhydrous pure TiO₂-II and markedly lower than that for hydrous Al-bearing TiO₂-II. The dehydration or partial dehydration of the water-bearing TiO₂-II during heating may be the underlying reason for the phase transition from TiO₂-II to rutile occurring at a lower temperature.