Benitoite (ideally BaTiSi₃O₉, a∼6.64 Å, c∼9.76 Å, space group P6‾c2) is an extremely rare mineral species, and its occurrence has been confirmed in just nine locations worldwide. Only the Benitoite Gem mine (California, USA) and four other mines nearby have produced gem-quality material (Wise and Gill, 1977). Here, benitoite was found in hydrothermally altered serpentine near the headwaters of the San Benito River (California, USA), from which it takes its name (Louderback and Blasdale, 1907). There is considerable controversy regarding the first discovery of this phase, and it seems that it was initially misinterpreted as sapphire (Louderback and Blasdale, 1909; Austin, 1988; Laurs et al., 1997) due to its appearance. The colour of benitoite crystals is quite variable and ranges from blue to violet blue (totally and pink benitoite samples are extremely rare), while the intensity varies from very light to medium dark. Strong dichroism, from colourless to blue, is evident to the eye and through a dichroscope (Gaft et al., 2004). Blue benitoite crystals mined in the Benito and Fresno counties are currently used as gemstones. The study of Louderback and Blasdale (1909) contains plenty of details regarding the geological setting of the discovery site, and the authors also provided wet chemical analyses and described its physical and optical properties. Later, the chemical composition was re-investigated by Laird and Albee (1972) via electron microprobe analysis, whereas its structure was initially solved by Zachariassen (1930) and was later refined and essentially confirmed by Fisher (1969). In benitoite, groups of three SiO₄ tetrahedra are connected to form rings with 6‾ symmetry. The Ba and Ti atoms lie on three-fold symmetry axes parallel to the c crystallographic direction in two-fold positions (2e and 2c respectively) on either side of a mirror plane. At ambient pressure, barium displays an unusual [6+6]-fold coordination, but if only the six shortest distances are considered (Ba–O ∼2.76 Å), Ba is in a distorted octahedral coordination.

Titanium is octahedrally coordinated to six oxygen atoms (Ti–O ∼1.95 Å) belonging to the unshared corners of the triangular [Si₃O₉]⁶⁻ ring. Therefore, the crystal structure of benitoite (Fig. 1) can be viewed as a stacking of titanium silicate complexes (with [Ti(Si₃O₉)₆]³²⁻ stoichiometry) in which the Ba site fills the gaps (Henry, 1998; Gaft et al., 2004; Hejny et al., 2012). The high-pressure behaviour of benitoite has been previously studied by Hejny et al. (2012), who detected a second-order phase transition at approximately 4.2 GPa. Diffraction experiments indicate a second-order phase transition with a′=a√3‾ and c′=c relative to the P6‾c2 sub cell. The high-pressure phase of benitoite (hereafter referred to as benitoite-II) was resolved by Hejny et al. (2012) in the P31c space group.

The symmetry breaking, as reported by the authors, leads to a splitting of the single Ba site into three sites occupying the Wyckoff positions 2a (0, 0, z) and 2b (1/3, 2/3, z; 2/3, 1/3, z). Moreover, the Ti site changes its special position, from 2c (point symmetry 32) in the ambient-pressure benitoite to 6c (point symmetry 1). The geometrical relationship and the structures of the two polymorphs are displayed in Fig. 1. Hejny et al. (2012) also reported that the structural changes induced by the phase transition have an effect on the bulk modulus (i.e. KV0=β-1, where β is the compressibility parameter), decreasing its bulk modulus from K0=117.9(7) GPa for benitoite to K0=88.9(1.6) GPa for benitoite-II. However, because of the relatively low hydrostatic limit of the used pressure medium (water-free 4:1 methanol–ethanol mixture), the authors decided to conclude the experiment at 6.14 GPa in order to avoid shear stresses which could decrease the quality of the diffraction peaks (Angel et al., 2007; Klotz et al., 2009). Nevertheless, benitoite is rather interesting due to the presence of the (Si₃O₉)⁶⁻ silicate-ring units, which is a relatively uncommon feature in silicates. Only CaSiO₃ pseudowollastonite present a structure with three-fold silicate ring units with the same topology as benitoite (Yang and Prewitt, 1999). However, it is known that wollastonite or pseudowollastonite transform with pressure and temperature into new high-pressure polymorphs, namely breyite and parabreyite, featuring three-fold silicate rings but with different topologies (Brenker et al., 2021; Chrappan Soldavini et al., 2025). Considering the interest of the CaSiO₃ system for the Earth's inner mineralogy, a high-pressure study on benitoite is a rather relevant comparative system to be investigated. Moreover, recent high-pressure research on carbonates, for instance, Fe-bearing MgCa(CO₃)₂ and Fe-bearing MgCO₃, has revealed that, at pressures above 80 GPa, both the compounds transform into new structures based on rigid three-fold tetrahedral carbonate rings (Merlini et al., 2017; Chariton et al., 2020). Due to limited possibilities to systematically investigate the crystal chemistry of candidate high-pressure carbonate minerals, studies conducted on silicates with similar structural features do allow comparative studies relevant to identifying possible different polymorphs in the same way studies on germanates drove the pioneering crystal chemistry of high-pressure silicates a few decades ago (e.g. Ringwood and Major, 1967).

Here, we report the results of a high-pressure X-ray experiment on single crystal of benitoite using a diamond anvil cell and He as the pressure-transmitting medium up to 22.36(5) GPa (numbers in parentheses are the standard uncertainties (s.u.) of the last significant digits according to IUCr guidelines). The main aim of this study is to extend the pressure range of the experiments reported by Hejny et al. (2012), providing new insights into the high-pressure behaviour and polymorphism of benitoite. We observed three distinct high-pressure phase transitions, two of which have not been reported previously.

A sample of benitoite from San Benito River (California, USA), provided by the Earth Sciences Museum Ardito Desio (Milan), was used in this study. A small single crystal fragment of about 20×10×10 µm was loaded into a membrane-driven diamond anvil cell (DAC), with 600 µm culet Boehler–Almax design anvils, along with a ruby sphere for pressure determination (pressure uncertainty ±0.05 GPa, Mao et al., 1986). A stainless-steel foil (with an initial thickness of∼250 µm) was pre-indented to 90 µm and then drilled by spark erosion, leading to a P chamber of ∼300 µm in diameter; He was used as the pressure medium. An in situ high-pressure single-crystal synchrotron X-ray diffraction experiment was performed at the ID15b beamline, ESRF, Grenoble (France). A convergent monochromatic beam (E∼30 keV, λ=0.41069 Å, spot size ∼5×6 µm) was used for the diffraction experiment. Diffraction images were acquired with a Mar555 flat-panel detector, using an in-house script for collecting step scan diffraction images. Sample-to-detector distance (∼260 mm) was calibrated using an Si standard and an enstatite (MgSiO₃) crystal. Further details on the beamline setup are reported in Merlini and Hanfland (2013). Load was applied to the diamond anvil cell with an automated gas-membrane-driven system, and pressure in the sample chamber was determined with the online ruby–alignment system after an average equilibration time of 15 min. The adopted data collection strategy consisted of a simple ω scan (-32° ≤ω≤+32°), with 0.5° step width and 0.25 s exposure time per step. The diffraction images were then converted into the “Esperanto” format in order to be analysed with the software CrysAlis (Rothkirch et al., 2013; Rigaku Oxford Diffraction, 2018). Indexing of the diffraction peaks and integration of their intensities (corrected for Lorentz-polarization effects) were performed using the CrysAlis package (Rigaku Oxford Diffraction, 2018). Corrections for absorption (caused by the DAC components) were applied using the semi-empirical ABSPACK routine implemented in CrysAlis (Rigaku Oxford Diffraction, 2018). All the structure refinements were performed using the package JANA2006 (Petříček et al., 2014). To assess the reversibility of the detected phase transitions and investigate their hysteresis, a few measures were taken in decompression. Selected reconstructions of reciprocal lattice planes in the four polymorphs are shown in Fig. S1 in the Supplement. The unit-cell parameters at high pressure are listed in Table 1, and the evolution with pressure of the volume per formula unit is shown in Fig. 2.

For benitoite and benitoite-II, the structures reported by Hejny et al. (2012) were used as starting models, refining the displacement parameters of all atoms as isotropic (Table S1 in the Supplement); no restraint on bond distances or angles was used. Two phase transitions were observed between 12.44(5) and 13.05(5) GPa and between 16.36(5) and 17.08(5) GPa, respectively. The newly discovered polymorphs will be referred to as benitoite-III and benitoite-IV. The structures of benitoite-III and benitoite-IV were solved using the Superflip software (Palatinus and Chapuis, 2007), implemented in JANA2006, to locate most of the crystallographic sites. A careful analysis of the difference–Fourier electron density maps allowed for the location of the remaining missing atoms; crystallographic information files (CIFs) are submitted as part of the Supplement. The structural changes associated with the individual phase transitions will be discussed in the next sections. The principal statistical parameters of the structure refinements are listed in Table S1, and inter-atomic angles and relevant distances, calculated by the Diamond and Vesta software (Momma and Izumi, 2008), are reported in Tables S2, S3, and S4.

The high-pressure evolution of benitoite structure reveals the existence of four polymorphs in the pressure interval of 0–20 GPa, characterized by a progressive distortion of the three-fold tetrahedral silicate ring and Ba and Ti polyhedra. Benitoite, BaTiSi₃O₉-II, and BaTiSi₃O₉-III present similar topology of the tetrahedral units, while BaTiSi₃O₉-IV presents a different configuration of two out of three silicate rings, with a tetrahedral base that is almost parallel. This resulted in a significant overall density increase. A similar behaviour is observed in the CaSiO₃ system, where breyite and parabreyite, which feature distorted three-membered rings, are significantly denser compared to pseudo-wollastonite. Both benitoite and CaSiO₃ provide valuable examples to explore the mechanism governing three-fold ring deformation under pressure and consequent structural variation. Other relevant examples of three-fold tetrahedral ring-based structures are constituted by high-pressure carbonates. Although only a few high-pressure polymorphs have been experimentally resolved for these carbonates with sp³-hybridized carbon (e.g. MgCO₃-II, dolomite-IV, and predicted CaCO₃ phases), the complexity observed in benitoite suggests that similar structural variability could exist in carbonate systems. These may include multiple carbonate polymorphs and extensive solid solutions stabilized under lower mantle conditions, highlighting the potential richness under lower mantle conditions.