Investigation of the High-Pressure Behaviors of Amblygonite by Single-Crystal X-ray Diffraction, Raman Spectroscopy, and DFT Calculations

Abstract

In the present study, we extensively explored the high-pressure behaviors and vibrational properties of amblygonite LiAlPO₄F with elevated pressures up to 34.3 GPa based on single-crystal X-ray diffraction measurements, Raman spectroscopy, and DFT calculations. The compressibility and elastic properties of amblygonite were determined first. Specifically, the obtained isothermal bulk modulus of LiAlPO₄F is 128(4) GPa and the triclinic phase exhibited anisotropic compression with axial compressibility β_c > β_a > β_b with a ratio of 1.11:1.00:1.20. The Raman spectra showed no indication of phase transformation and were used to obtained mode Grüneisen parameters. The average Grüneisen parameter for PO₄ tetrahedral sites was smaller than for the LiO₄F sites. Our results provide new insights into the phase stability and elastic properties of lithium-fluorite granites at extreme conditions.

1. Introduction

The behaviors of lithium-fluorite minerals and their related granites under high-pressure conditions have been studied broadly over a few decades, and knowledge about such minerals has important applications in tracing the pegmatite-forming process [1,2,3,4]. Amblygonite (LiAlPO₄F)–montebrasite (LiAlPO₄OH), cassiterite, and Nb-Ta oxides are the most common accessory minerals in all the dykes of the pegmatite fields, and members of the amblygonite–montebrasite series are common constitutents of Li- and F-rich granitic pegmatites [5,6,7]. Hence, the amblygonite–montebrasite monitor can provide some basic knowledge of the peraluminous granitic and pegmatitic melts. The distribution of fluorine between amblygonite–montebrasite solid solutions has also long been investigated during the crystallization of these magmatic phases and associated phases as well [2,8]. Therefore, the knowledge of high-pressure behaviors of amblygonite helps understand the detailed crystal chemistry of such Li-bearing granites and can reflect the fundamental distinctions in the conditions of magma origin. However, comprehensive studies on the crystal structural evolution and physical properties of amblygonite LiAlPO₄F at extreme conditions have not been fully investigated yet.

Natural faceted gemstones of amblygonite–montebrasite isomorphous series show complete solid solutions with an ideal chemical composition of LiAlPO₄(F,OH). The minerals are open-framework alumino-phosphate series with triclinic structure. As members of the amblygonite–montebrasite series are common constituents of Li and F-rich granitic pegmatites, most previous studies have focused on the determination of the F contents and F/(F+OH) ratio of such solid-solution series [2,9]. Moreover, the F-OH substitution in LiAlPO₄(F,OH) has attracted much attention due to its possible variations in crystal structural and geochemical and physical properties as well [10,11]. For example, F substitution into the lawsonite framework would likely stabilize the crystal structure since it would remove the H-H repulsion [12]. Groat et al. [11] described the crystal structure of the amblygonite–montebrasite series using a C-centered cell and also suggested that the pseudo-monoclinic structure was topologically identical to the titanite group minerals such as lacroixite NaAlPO₄F with a monoclinic phase. Recently, the microstructure of amblygonite–montebrasite was demonstrated by Shirose and Uehare [3] using powder X-ray diffraction and transmission electron microscope techniques. However, the structural behaviors and compressibility of amblygonite at high pressures are more sparse. Further investigation is needed to study the precise and accurate structural information and vibrational properties of amblygonite LiAlPO₄F at high-pressure conditions.

In this work, we undertook high-precision, single-crystal structure determinations of amblygonite up to ~34 GPa in situ using synchrotron-based single-crystal X-ray diffraction and theoretical calculations. High-pressure vibrational properties of amblygonite were also determined by Raman spectroscopy. These results firstly provide a comprehensive understanding of the crystallographic information of such lithium-fluorite granites under high-pressure or high-temperature conditions, thus improving our knowledge of proper elastic and chemical properties of these alumino-phosphates in pegmatites and shedding new light on alkali reservoirs in the Earth’s upper mantle.

2. Materials and Methods

2.1. Experimental Setup

High-quality single-crystal samples of natural, cream-colored amblygonite were used for the current study. The amblygonite sample measured in this study had the same chemical composition as that of Dias et al. [1], including 49.58% P₂O₅, 35.19% Al₂O₃, 5.89% H₂O, and 10.25% Li₂O as well as some minor components 0.02% Na₂O and 0.40% F. We screened several chips polished to ~15 μm thickness of the sample for high-pressure measurements. At room P-T conditions, the LiAlPO₄F crystal was characterized with a Bruker SMART CCD diffractometer using a sealed-tube X-ray source (MoKα radiation, λ = 0.71073 Å) operating at 45 kV and 35 mA at the China University of Geosciences (Beijing).

A short symmetric-type DAC fitted with 300 μm Boehler-Almax diamond anvils mounted into a seat with a 56° opening was used for high-pressure measurements. A 160 μm diameter hole was drilled in a pre-indented ~45 μm rhenium gasket with an initial thickness of 250 μm to act as a sample chamber. The polished amblygonite crystal and a gold foil were loaded together into the sample chamber, followed by gas loading with neon as the pressure-transmitting medium using the COMPRES/GSECARS gas-loading system [13]. XRD patterns of gold were collected at each pressure before and after sample data collection [14].

In situ high-pressure single-crystal XRD experiments were conducted at up to 34.4 GPa and room temperature at the 13-BM-C experimental station of the Advanced Photon Source, Argonne National Laboratory. The incident X-ray beam at the 13-BM-C was monochromated to 0.4340 Å with a focal spot size of 15 × 15 μm². XRD patterns were recorded with a MAR165 charge-coupled device detector that was placed about 170 mm away from the sample. The experimental details were also described previously [15,16]. To obtain an adequate number of diffraction peaks of samples and increase the coverage of the reciprocal space, we collected data at four different detector positions. Wide-scan and stepped-φ exposures were collected in a rotation range from –90° to +90° with a typical exposure time of one second per degree. Data were analyzed by the APEX3 Crystallography Software Suite, VESTA software, and the SHELXL package [17,18,19]. P-V data were fitted by the EoSFit7-GUI program [20].

High-pressure Raman spectroscopy for the amblygonite was performed at up to 28.6 GPa and at room temperature using Princeton-style diamond cells with 300 μm ultra-low fluorescent diamond anvils. Rhenium gaskets were pre-indented as sample chambers to a thickness of ~40 μm with a hole of ~160 μm in diameter drilled by laser at the center of the indentation. The diamond cells for Raman spectroscopy were gas loaded with neon at HPSTAR and ruby was used for pressure calibration [21]. Reported pressure uncertainties reflect the difference between pressure measurements performed before and after each data collection, which did not vary by more than ±0.2 GPa. Raman spectra were collected on a Renishaw inVia reflex Raman spectrometer with a 532 nm diode-pumped solid-state laser at Peking University. The spectra were collected in black-scattering geometry using a charge-coupled device detector with a resolution of 1 cm⁻¹. All spectra were fitted with pseudo-Voigt functions to determine the peak positions.

2.2. Computer Simulations

DFT-based first-principle calculations were performed utilizing the projected augmented wave method using the Vienna Ab initio Simulation Package [22,23,24,25,26]. The generalized gradient approximations with the Perdew–Burke–Ernzerhof (PBE) version were adopted to treat the exchange and correlation (XC) function [27]. We employed pseudopotentials to model the ion-electron interaction. The plane wave energy cutoff was set to 800 eV and the energy convergence criterion for the electronic self-consistent calculations was 10⁻⁵ eV. The force difference was converged to 1 × 10⁻³ eV/Å (less than 0.1 GPa). The k-points grids were set as 5 × 5 × 4 in the Brillouin zone for amblygonite in structural relaxation calculations using the Monkhorst–Pack method [28]. The geometry optimization was conducted from about −2 to 27 GPa at 0 K. We changed the pressure by scaling the volume of each crystalline phase. The atomic positions and unit cell parameters were allowed to relax at each given volume to obtain the minimum total energy.

3. Results

3.1. Sample Characterization

The natural sample used in this work was obtained from Divino das Laranjeiras, the district of Linópolis, Minas Gerais in Brazil; the composition in this study is close to that published by Dias et al. [1] which is also from the same location. A small chip of amblygonite single crystal, extracted from a larger specimen, was used for the sample characterization with single-crystal diffraction, which agrees well with previous studies [29]. All observed reflections could be indexed to a triclinic primitive unit cell. The results confirmed that the amblygonite sample possess space group P-1 structure, and its lattice parameters were a = 5.0338(3) Å, b = 5.191(1) Å, c = 7.0160(6) Å, α = 106.63(1)°, β = 109.26(1)°, γ = 97.933(1)°, and V = 160.3(1) ų. The crystallographic information file for amblygonite at room P-T is provided in the supporting information. The disorder in the Li site is caused by the substitution of F for OH; therefore, it would also give the OH/F disorder in the nearest sites within the structure [10]. The amblygonite structure is characterized by chains of corner-linked AlO₄F₂ octahedra that run parallel to the c-axis. These octahedral chains are linked laterally by edge-sharing PO₄ tetrahedra and larger LiO₄F-polyhedra [16] (Figure 1).

3.2. Room-T compressibility

Lattice parameters and the unit–cell volume of amblygonite at high-pressure conditions were analyzed using the APEX3 v2018.7-2 software (Bruker) Figure 2 shows the single-crystal diffraction pattern at 1.0 GPa, and all diffraction spots collected can be readily indexed with the triclinic LiAlPO₄F structure. There was no indication of phase transition up to 34.4 GPa at room temperature (Table S1, see in Supplementary Materials). However, the high-pressure crystal structure could not be refined due to the limited opening angle of the diamond cell and the low symmetry of amblygonite. The obtained unit–cell reference volume, V_T0 = 160.27(1.7) ų, is consistent with the previously reported value [29]. To obtain the EoS parameters, we then fit the P-V data to a third-order Birch–Murnaghan equation of state (BM3-EoS) using error-weighted least squares with EoSFit7c [30]. The resulting BM3-EoS parameters were as follows: V_T0 = 160.3(1) ų, K_T0 = 128(4) GPa, and K_T0′ = 3.4(4) (Figure 2). The P-V data yielded values of K_T0 = 122(1) GPa when assuming a pressure derivative of K_T0′ = 4.

To verify the XRD experimental results, we then carried out the theoretical calculation using Density Functional Theory. Energy–volume results were also fitted with the BM3-EoS [31,32], which yields V_T0 = 163.2(1) ų, K_T0 = 102(1) GPa, and K_T0′ = 4.2(2). The bulk modulus from the DFT calculation was slightly smaller than the experimental results (Figure 3). The calculated equilibrium volumes were generally underestimated by 3% in comparison to the experimental results, which is typical for PBE method computations. Our fitted values are comparable with those reported typical values of phosphates [33].

The trend of lattice parameters relative to a (a/a₀), b (b/b₀), c (c/c₀), α, β, and γ angles with pressure are plotted in Figure 3. To determine the axial compressibility of each axis in amblygonite, we used a linearized BM3 fitting where each axial dimension is cubed and treated as a volume in the BM formulation [30]. The zero-pressure axial compressibility of linear dimension l, defined as β_l0 = −(l⁻¹)(δl/δP)_P=0, is related to the linear modulus (linear incompressibility) by K_l0 = (β_l0)⁻¹. For amblygonite, our fitted linear moduli to a, b, and c were 387(7), 430(5), and 359(9) GPa, respectively, corresponding to axial compressibility values of β_a = 2.58(1) × 10⁻³, β_b = 2.33(2) × 10⁻³, and β_c = 2.79(3) × 10⁻³ GPa⁻¹. There is a considerable anisotropy in axial compressibility with β_c > β_a > β_b and the ratio of zero-pressure axial compressibility in amblygonite is 1.11:1.00:1.20. The DFT calculated results are more scattering than the experimental data (experimental results: solid symbols, DFT: open circle), but both results are in agreement with each. The interaxial angles α, β, and γ showed a steady trend in compression up to ~35 GPa, supported by the experimental and calculation evidence (Figure 4).

3.3. Polyhedral Compression and Distortion

Bond lengths, angles, and polyhedral volumes are useful parameters to characterize the distortion of each polyhedron. The compressibilities calculated based on theoretical (DFT) results of three polyhedra (AlO₄F₂, PO₄, and LiO₄F) are significant. There were two distinct octahedral sites (Al1 and Al2), two distorted hexahedral sites (Li1 and Li2), and two P positions (P1 and P2) (Table S2). Bond length (Å), polyhedral volume (ų), distortion index (D), and quadratic (λ) and bond angle variance (σ²) for various polyhedra at different pressures are shown in Table S2.

Among these three types of polyhedra, PO₄ tetrahedra showed little or no compression, while LiO₄F polyhedra showed the greatest compression (Figure 5). The average AlO₄F₂ volume was 9.125 ų at 1.9 GPa and decreased to 7.725 ų at 28.3 GPa, and the average volume in LiO₄F polyhedra was about 7.072 ų at 1.9 GPa and changed to 5.377 ų at the highest pressure. Between 1.9 and 28.3 GPa, the PO₄ tetrahedra showed the least compression, decreasing from 1.919 to 1.793 ų in the amblygonite sample. Slopes of the polyhedral volumes as a function of pressure were obtained from a linear fit. It can be concluded that the distorted trigonal prismatic geometry of the LiO₄F unit is more compressible than the PO₄ rigid polyhedral unit. On the other hand, the partial replacement of Li by Na might decrease Li-O bond strength, thus making the LiO₄F polyhedra softer. The OH-F substitution is also giving rise to the less rigid LiO₄F and AlO₄F₂ polyhedra. Therefore, the unit–cell volume decrease on compression is dominated by the compression of the AlO₄F₂ octahedra and LiO₄F polyhedra.

We first calculated the distortion index, quadratic elongation, and bond angle variance to quantify the distortion of each polyhedron. The distortion index (D), defined as D =$\frac{1}{n}\sum _{\mathrm{i}=1}^n\frac{l_{\mathrm{i}}-l_{\mathrm{a}\mathrm{v}}}{l_{\mathrm{a}\mathrm{v}}}$, where l_i is the distance from the central cation to the ith surrounding oxygen and l_av is the average distance. Quadratic elongation (λ) and bond angle variance (σ²) were employed to describe the deviation from the regular shape of polyhedra [34,35]. The bond angle variance and quadratic elongation were defined as σ² = $\sum _{\mathrm{i}=1}^n\frac{{{(\theta }_{\mathrm{i}}-{\theta }_0)}^2}{n-1}$ and λ = $\sum _{\mathrm{i}=1}^n\frac{{{(l}_{\mathrm{i}}-l_0)}^2}{n}$, respectively, where θ_i is the ith angle, θ₀ is the ideal bond angle for the perfect regular polyhedron, l_i is the ith center-to-vertex distance, and l₀ is the ith center-to-vertex distance for the regular polyhedron.

As a consequence, the degree of distortion within the LiO₄F polyhedra showed a significant increase with pressure, while the other AlO₄F₂ octahedra and PO₄ tetrahedra showed little or no increase as indicated by the D values (Figure 6). At 1.9 GPa, both the AlO₄F₂ and PO₄ polyhedra in LiAlPO₄F were close to being regular with average D = 0.006 and changes to 0.014 and 0.005 at 28.3 GPa, respectively. The evolutions of the values of σ² and λ for AlO₄F₂ and PO₄ polyhedra showed similar trends as the pressure changed (Figure 7). The σ² value of PO₄ tetrahedra slowly reduced from 12.25 at 1.9 GPa to 11.25 at 28.3 GPa (for λ value, from 1.0032 to 1.0029), and then increased to 13.96 (1.0035 for the λ value) at the highest pressure. The resulting values of σ² and λ for AlO₄F₂ octahedra were more responsive to pressure change than the PO₄ tetrahedra. The λ value of Al₂O₄F₂ octahedra rapidly increased from 1.001 to 1.0096 (for σ², from 3.50 to 34.19) between 1.9 and 28.3 GPa, while the values of σ² and λ for Al1O₄F₂ octahedra displayed a moderate trend as pressure elevated, which increased from 1.0022 to 1.0034 and from 6.57 to 9.09 for λ and σ², respectively. (Figure 6). This difference is likely due to the different structural configurations in the two distinct Al sites.

3.4. Vibrational Properties

To date, there is no detailed high-pressure Raman spectroscopic study on amblygonite. Figure 8 shows in situ Raman spectra of LiAlPO₄F recorded from ~100 to 1300 cm⁻¹ at various pressures up to 28.6 GPa. All observed Raman peaks shifted to a higher frequency with increasing pressure. In the 1000–1150 cm⁻¹ region, there were prominent peaks located at 1019, 1053, 1066, and 1116 cm⁻¹ that can be assigned to various P–O stretching modes [36]. The spectra were more complex below 650 cm⁻¹, and the mixed region composed of P–O bending and AlO₄F₂ stretching vibrations mainly contributed to the absorption in this domain [37]; therefore, some vibration modes merged together with increasing pressure. The intense band at 486 cm⁻¹ is attributed to the symmetric Al–O–Al stretching mode in the network. Some Raman modes located between 400–650 cm⁻¹ may be assigned to the O–P–O deformation modes, which are in accordance with the results from Cooper et al. [38]. Lower frequencies from 140 to 330 cm⁻¹ can be attributed to the O–Li–O (or F) bending and Li–O bond vibrations. which are in agreement with the spectroscopic study of lithium pyroxenes [39,40]. This shows that modes v1, v2, and v3 spread out non-monotonically with increasing pressure and it might be caused by the asymmetric stretching and bending of O–Li–O and O–Li–F bonds (Figure 8). It should be noted that there were no obvious peaks broadening observed for all spectra, indicating that the structural disorder over the lattice sites was insignificant up to the highest pressure.

Grüneisen parameters (γ_i) describe the effect of changing volume of solid matter on vibrational motions of atoms; thus, the γ_i, particularly its volume dependence, is profound in the field of thermoelastic properties of solids. Mode Grüneisen parameters (γ_i) were calculated using γ_i = (K₀/ν_i) (d_νi/d_P)_T, where ν_i is the wavenumber of the ith mode and K₀ is the bulk modulus at room temperature. Here, we used the isothermal bulk modulus K₀ of 128(4) GPa for amblygonite to calculate the mode Grüneisen parameters. The resulting mean pressure coefficient of the whole structure for LiAlPO₄F was 2.22(3) cm⁻¹/GPa, and correspondingly, the calculated mode Grüneisen parameters determined in this study ranged from 0.22(1) to 1.44(2), with average value from all observed bands of 0.62(2). As for the PO₄ tetrahedral site vibrations, the calculated γ values generally fell in the range of 0.43(2)–0.44(2), and were smaller than the 0.97(3) found in LiO₄F polyhedral sites.

The amblygonite–montebrasite series from petalite-bearing localities can include various contents of lacroixite, while the lower-temperature pegmatite with spodumene LiAlSi₂O₆ either did not possess or involved very low lacroixite contents [3]. Some researchers also indicated that the fluorine contents had a positive correlation with the sodium contents in the amblygonite–montebrasite series [11]. Moreover, the enriched fluxing components such as Li, F, and P in the amblygonite–montebrasite series can provide relevant information on the crystallization of the fluxing component during pegmatite formation. Although alkalis, such as lithium and sodium, are significantly enriched in continental crust compared with oceanic crust, Li and Na are important incompatible elements in the upper mantle [41,42]. Silicates are almost the exclusive hosts for Li, and Li⁺ and Na⁺ are expected to replace each other in the structure [43]. The detailed investigation of the structural evolution of such lithium-fluorite minerals under compression will give some insight into possible systematic trends in the amblygonite–montebrasite series and provide direct evidence on the alkali-rich minerals at extreme conditions in the pegmatite-forming process.

4. Conclusions

The high-pressure behaviors of amblygonite were investigated by synchrotron-based single-crystal XRD, Raman spectroscopy, and DFT calculations at pressures up to ~34 GPa. The isothermal pressure–volume relationship of LiAlPO₄F is described by the BM3-EoS, yielding K_T0 = 128(4) GPa, and K_T0′ = 3.4(4). In combination with Raman measurements performed to 28.6 GPa, we found no evidence for phase transformation over this pressure range. The OH-F substitution gave rise to the less rigid of both LiO₄F and AlO₄F₂ polyhedra and the distorted LiO₄F geometry was more compressible than the PO₄ tetrahedral unit. The average γ value of PO₄ vibration modes was 0.44(3), which is smaller than the γ value for the LiO₄F polyhedra. The findings contribute to broadening our knowledge of the crystal chemistry of amblygonite at high-pressure conditions, thus giving a better understanding of lithium-fluorite minerals and some related granites [44].