2025 Volume 120 Issue 1 Article ID: 250401
Single-crystal X-ray crystal-structure refinements and electron-microprobe analyses of julgoldite from Kouragahana, Shimane Peninsula, and pumpellyite from Sugashima, Mie, Japan yielded the compositions of WCa2.00X(Fe0.69Mg0.20Al0.11)Σ1.00Y(Fe3+1.46Al0.54)Σ2.00Si3.00O14−n(OH)n and WCa2.00X(Al0.56Mg0.35Fe0.09)Σ1.00YAl2.00Si3.00O14−n(OH)n (4 ≥ n ≥ 3), respectively. The latter was classified as pumpellyite-(Al), while the former can only be concluded to be either julgoldite-(Fe3+) or julgoldite-(Fe2+). The hydroxyl groups determined by structure refinement are consistent with those known for pumpellyite. The length of the b-axis most accurately reflects the average size of ionic radii at the Y site based on structural and chemical data for pumpellyite-group minerals. In contrast, the lengths of the a- and c-axes demonstrate a favorable correlation with the mean ionic radius of the Y site; however, they exhibit greater dispersion compared to the b-axis. This is attributed to the lateral expansion or shrinkage of the (010) plane caused by variation in the size of the X site. Therefore, the pumpellyite group minerals with longer a- and c-axes may be rich in divalent cations at the X site. The well-defined, intense Raman peak around 695 cm−1 is characteristic of pumpellyite-group minerals and is likely attributed to the Si-O-Si bending mode. The six to seven Raman peaks resulting from O-H bond stretching reflect the complex hydrogen bond system in the pumpellyite group structure, arising from the presence of multiple hydroxyl groups and variations in the local chemical environment due to compositional differences.
Pumpellyite-group minerals, VIIW2VIXVIY2IVSi3O14−n(OH)n with 4 ≥ n ≥ 3 (Z = 4, Passaglia and Gottardi, 1973), are formed under low-grade metamorphic conditions and at low-temperature hydrothermal activity (e.g., Deer et al., 1986). Due to the compositional and structural complexity of this mineral group, its crystal chemistry is still not well understood. The structure of pumpellyite-group minerals is monoclinic with A2/m space-group symmetry (standard setting C2/m). The W site is subdivided into W1 and W2, predominantly occupied by Ca. In general, both divalent (Me2+) and trivalent (Me3+) cations, such as Mg, Al, Mn2+, Mn3+, Fe2+, Fe3+, V3+, and Cr3+, occupy the X site, whereas the Y site is only occupied by trivalent cations. The pumpellyite structure consists of isolated [SiO4] tetrahedra and disilicate [Si2O6(OH)] groups. The tetrahedral units connect two symmetrically independent edge-sharing chains of X and Y octahedra, extending parallel to the b-axis (Gottardi, 1965). Pumpellyite-group minerals are named after the predominant cation at the Y site, and they are further distinguished by a suffix denoting the predominant cation at the X site (Passaglia and Gottardi, 1973). If 50% of the X site is occupied by divalent cations, 3.5 OH and 10.5 O pfu are required for charge balance. If the X site is only occupied by divalent cations, the formula has four hydroxyl groups, but it has three hydroxyl groups if only trivalent cations occupy the X site (e.g., Yoshiasa and Matsumoto, 1985). The hydrogen bond system varies with the oxidation state of the X-site cations, and at least two hydrogen bond systems have been observed (Nagashima et al., 2010).
Julgoldite is a YFe3+-analogue of the pumpellyite-group mineral, first described by Moore (1971) from Långban, Sweden. Although the pumpellyite-julgoldite series commonly occurs, the occurrence of julgoldite is less frequent. Julgoldite-(Fe3+), julgoldite-(Fe2+), and julgoldite-(Mg) have been listed by IMA-CNMNC (International Mineralogical Association Commission on New Minerals, Nomenclature and Classification). It is known that the type locality of the two former species is Långban, Sweden (Moore, 1971; Passaglia and Gottardi, 1973). According to the IMA-CNMNC list, Japan has been registered as the type locality of julgoldite-(Mg) based on the summarized chemical compositions listed in Passaglia and Gottardi (1973). Furthermore, the database provided by mindat.org recognizes Sugashima, Toba, Mie Prefecture, as its type locality. However, in their list (see their table 3), the specimen from Toba, Shinsen-mura (correctly, Kamiizumi-mura), Central Japan, is categorized as not julgoldite-(Mg) but pumpellyite-(Mg). The analytical results of the pumpellyite-group mineral reported by Seki (1958), which Passaglia and Gottardi (1973) cited as a data source, clearly show that it is pumpellyite due to the enrichment of Al and poor in Fe (23.5-25.4 Al2O3 wt% and 4.6-6.3 wt% Fe2O3 according to Seki, 1958). In addition, the pumpellyite-julgoldite series mineral found in Sugashima in this study shows even lower Fe content (<3 FeO wt%), as discussed later. Since julgoldite-(Mg), in which Mg predominantly occupies the X site, has not been reported from other localities worldwide until now, this mineral species is virtually non-existent.
Site occupancy at the X site is essential for the nomenclature of pumpellyite group minerals, and its determination is not straightforward, especially for Fe-bearing species. Iron in the pumpellyite-julgoldite series commonly exists in both Fe2+ and Fe3+, making it difficult to determine the oxidation state and site distribution of cations between the X and Y sites using conventional chemical and structural analyses. Allmann and Donnay (1973) estimated the Fe2+ and Fe3+ occupancies at the X site using the expected X-O distances calculated from the effective ionic radii. Although the 57Fe Mössbauer spectroscopy was applied as a powerful tool to investigate the behavior of Fe in the pumpellyite-julgoldite series (Artioli and Geiger, 1994; Akasaka et al., 1997; Artioli et al., 2003; Nagashima et al., 2006; Kasatkin et al., 2021; Akasaka et al., 2023), the different assignments of the Mössbauer doublets lead to some possible Fe distribution among the polyhedra. Even two recent studies, by Akasaka et al. (2023) and Kasatkin et al. (2021), have proposed different results for Fe distribution at the X and Y sites. Akasaka et al. (2023) assigned Fe2+ to the X site and Fe3+ to the X and Y sites in a study of the Fe-rich pumpellyites from the Northern Chichibu Belt, western Shikoku, Japan. Kasatkin et al. (2021) investigated the julgoldite-(Fe3+) from the Kradag volcanic massif. According to their determined formula, the site occupancy at the X and Y sites can be written as Fe3+0.77-0.62Fe2+0.13-0.28Mg0.05Mn0.01 and Fe3+0.92-1.00Fe2+0.08-0.00Al0.01, respectively, implying both Fe2+ and Fe3+ ions distribute the X and Y sites. In addition, they suggested the presence of a small amount of water molecule in their julgoldite-(Fe3+), based on the result of the infrared spectrum.
The Me2+:Me3+ ratio is generally known to vary at the X site between 0.4:0.6 and 0.6:0.4 (e.g., Passaglia and Gottardi, 1973). The suggested formula of julgoldite by Allmann and Donnay (1973) was Ca2(Fe2+1−xFe3+x)Fe3+2Si3O10+x(OH)4−x, where x ≈ 0.5. However, Artioli et al. (2003) reported the presence of Bombay julgoldite, which contains 95% trivalent cations at the X site, representing a significant deviation from the typical Me2+:Me3+ ratio. This finding clarifies the longstanding idea, proposed by Passaglia and Gottardi (1973) and Allmann and Donnay (1973), that approximately half of the X site is occupied by divalent cations and the other half by trivalent cations, is no longer valid. Despite the contributions of earlier studies on pumpellyite-group minerals, a method for estimating the cation distribution at the X site has yet to be established. Thus, it is essential to examine pumpellyite-group minerals with various compositions precisely and accurately from a crystal-chemical perspective.
To gain insights into the comprehensive crystal chemistry of the pumpellyite-group minerals, we investigated the relationship between cation distribution and structural variation, as well as the hydrogen-bond system of these minerals from two localities: (1) Kouragahana, Shimane Peninsula, Shimane Prefecture, Japan, and (2) Sugashima, Toba, Mie Prefecture, Japan, using microprobe analysis, X-ray single-crystal structural analysis, and Raman spectroscopy.
Julgoldite was found at Kouragahana, Shimane Peninsula, Japan. At this site, a variety of secondary minerals are typically found in Miocene dolerite that has experienced hydrothermal alteration (e.g., Kano et al., 1986). Secondary minerals, such as prehnite, pumpellyite, babingtonite, and thomsonite, have been reported from these altered rocks, replacing primary minerals or filling cavities (Kano et al., 1986; Akasaka et al., 1997, 2003; Nagashima et al., 2006, 2016, 2018a, 2021). The studied julgoldite is one of these secondary minerals, and the crystals are prismatic or bamboo-leaf-like in shape (<120 µm in length). They are closely associated with radial aggregates of ferriprehnite (IMA2020-057; Nagashima et al., 2021) or are embedded as inclusions in calcite (Fig. 1a). Matsubara et al. (1992) reported julgoldite-(Fe2+) from the same locality. Based on its assemblage and chemical composition, the julgoldite-(Fe2+) described by Matsubara et al. (1992) closely resembles the sample examined in this study.
Pumpellyite from Toba, Mie Prefecture, Japan, occurs as a main constituent of a white vein (<1.5 cm in width) cutting through the greenstone (Fig. 1b). Aggregates of fibrous pumpellyite crystals were associated with grossular and prehnite, as well as a small amount of epidote, pectolite, and albite. In the sampling area, the Shurei Igneous Rocks are distributed and correlated with the Mikabu greenstones of the Kanto Mountains based on the similarity of their petrological characteristics (Nakamura, 1971; Uchino et al., 2017). The Shurei Igneous rock comprises metamorphosed peridotite, hornblendite, and gabbro, along with mafic meta-volcanic (Nakamura, 1971; Uchino et al., 2017). According to Tomiyoshi and Takasu (2010), the K-Ar ages from phengite in meta-pelite associated with the Shurei Igneous rock were 99-93 Ma (the earliest Late Cretaceous).
The chemical compositions of the Kouragahana julgoldite and the Sugashima pumpellyite were determined using a JEOL JXA-8230 electron microprobe analyzer (EPMA) at Yamaguchi University, Japan, and a JEOL JXA-8800 at Shimane University, respectively. Operating conditions were an accelerating voltage of 15 kV, a beam current of 20 nA, and a beam diameter of 1 µm. Wavelength-dispersion spectra were collected using LiF, PET, and TAP monochromator crystals to identify interfering elements and determine the optimal wavelengths for background measurements. The abundances of Si, Ti, Al, Cr, V, Fe, Mn, Mg, Ca, Sr, Ba, Na, K, Ni, Cu, Zn, P, F, and Cl were measured. Several elements not listed in Table 1 are below the detection limit. The ZAF method was used for data correction.
| Oxide (wt%) |
Kouragahana | Sugashima | ||
| Ave. | Std. | Ave. | Std. | |
| n = 8 | n = 15 | |||
| SiO2 | 33.13 | 0.27 | 37.92 | 0.22 |
| TiO2 | 0.11 | 0.07 | 0.03 | 0.04 |
| Al2O3 | 6.10 | 0.61 | 26.97 | 0.30 |
| Cr2O31) | 0.03 | 0.04 | 0.01 | 0.02 |
| V2O31) | 0.08 | 0.04 | 0.03 | 0.04 |
| Fe2O31) | 32.17 | 0.94 | 2.67 | 0.30 |
| MnO1) | 0.02 | 0.03 | 0.07 | 0.03 |
| MgO | 0.84 | 0.11 | 2.68 | 0.30 |
| NiO | 0.01 | 0.01 | 0.02 | 0.03 |
| CaO | 20.64 | 0.18 | 23.29 | 0.14 |
| Na2O | 0.01 | 0.01 | 0.11 | 0.05 |
| K2O | 0.00 | 0.01 | 0.05 | 0.02 |
| F | 0.05 | 0.04 | 0.00 | |
| -O=F | 0.02 | |||
| Total | 93.17 | 93.85 | ||
| Σcations = 8 | ||||
| Si | 3.01 | 0.01 | 3.00 | 0.01 |
| Ti | 0.01 | 0.00 | 0.00 | 0.00 |
| Al | 0.65 | 0.06 | 2.52 | 0.03 |
| Cr | 0.00 | 0.00 | 0.00 | 0.00 |
| V | 0.01 | 0.00 | 0.00 | 0.00 |
| Fe | 2.20 | 0.07 | 0.16 | 0.02 |
| Mn | 0.00 | 0.00 | 0.00 | 0.00 |
| Mg | 0.11 | 0.02 | 0.32 | 0.04 |
| Ni | 0.00 | 0.00 | 0.00 | 0.00 |
| Ca | 2.01 | 0.01 | 1.98 | 0.01 |
| Na | 0.00 | 0.00 | 0.02 | 0.01 |
| K | 0.00 | 0.00 | 0.00 | 0.00 |
| Total | 8.00 | 8.00 | ||
| F | 0.01 | 0.01 | ||
1) Cr as Cr2O3, V as V2O3, Fe as Fe2O3, and Mn as MnO.
X-ray diffraction data for single crystals of Kouragahana julgoldite (0.02 × 0.04 × 0.10 mm) and Sugashima pumpellyite (0.01 × 0.02 × 0.08 mm) were collected with MoKα radiation (λ = 0.71073 Å) using an XtaLAB Synergy-R/DW (Rigaku Co. Ltd.) installed at Yamaguchi University, Japan. Each crystal was mounted on a glass fiber, and the intensity data were measured at room temperature. Preliminary lattice parameters and an orientation matrix were obtained from six sets of frames and refined during the intensity data integration process. The diffraction data were processed using CrysAlisPro (Agilent, 2014). An empirical absorption correction was applied using CrysAlisPro (Agilent, 2014). Structural refinement was performed using the program SHELXL-2019/3 (Sheldrick, 2015). Scattering factors for neutral atoms were employed. At the primary stage, populations of Ca at W1, W2, and Si at Si1, Si2, and Si3 were refined. However, these sites were fully occupied within one standard deviation. Thus, the site occupancies at these sites were fixed at 1.0. The site occupancies of the X and Y sites were refined with Fe versus Mg and Fe versus Al, respectively. X-ray data cannot discriminate between Al and Mg due to similar scattering factors. Thus, the Al content may also include Mg and vice versa. The site occupancy at the Y site in Sugashima pumpellyite was found to be fully occupied by Al within one standard deviation. Thus, it was fixed as 1.0Al in the final refinement. The positions of the hydrogen atoms of the hydroxyl groups were derived from difference Fourier syntheses and were refined with fixed Uiso = 0.05 Å2. In addition, a bond-distance constraint of O-H = 0.98(2) Å (Franks, 1973) was applied.
Raman spectroscopy was conducted using a Renishaw inVia Reflex spectrometer with a green diode laser (λ = 532 nm) installed at the ISSP, University of Tokyo, Japan. The laser power at the sample surface was approximately 10 mW. The Raman spectra were collected from unoriented crystals in the 100-4000 cm−1 range. Figure 2 shows the spectrum in the ranges 118-1215 (lattice vibration region) and 2650-3800 (OH region) cm−1 with the spectra of pumpellyite-(Al), R070131, and julgoldite-(Fe2+), R070725 (unoriented with 532 nm laser, Lafuente et al., 2015) as references.
The average chemical compositions of Kouragahana julgoldite and Sugashima pumpellyite are listed in Table 1. The sum of cations was normalized to 8. The Kouragahana julgoldite has a high Fe content, with the total Fe2O3 content (including FeO) being approximately 32.2 wt%. In contrast, the Sugashima pumpellyite is rich in Al but poor in Fe. The Fe2O3 content does not reach 3 wt%. The chemical formulas (Z = 4) are represented as Ca2.01(Fe2.20Mg0.11Al0.65Ti0.01V0.01)Σ2.98Si3.01O14−n(OH)n for the Kouragahana julgoldite, and (Ca1.98Na0.02)Σ2.00(Al2.52Mg0.32Fe0.16)Σ3.00O14−n(OH)n for the Sugashima pumpellyite, with n as a variable depending on the oxidation state of iron. A method for estimating valence by fixing the stoichiometry is commonly used (e.g., Munoz et al., 2010; Kováčik, 2011; Austrheim et al., 2022); however, this approach cannot be applied to pumpellyite-group minerals because the amount of OH is correlated with the oxidation state of iron.
Crystal-structure refinement: site occupancy and hydrogen positionsCrystallographic data and refinement parameters are summarized in Table 2. The Rint and Rσ are slightly high due to poor crystal quality. However, enhancing the data quality from needle- or thin-platy-pumpellyite and julgoldite crystals proved to be difficult. Structure refinements converged at R1 values of 5.43% for the Kouragahana julgoldite and 7.20% for the Sugashima pumpellyite. The atomic positions, refined occupancies, and anisotropic displacement parameters are listed in Tables 3 and 4, respectively. Selected interatomic distances and angles, as well as the volume and calculated distortion parameters of octahedral sites, are listed in Table 5. The crystal structure of the Kouragahana julgoldite is shown in Figure 3.
| Sample | Kouragahana julgoldite | Sugashima pumpellyite | |
| Crystal size (mm) | 0.02 × 0.04 × 0.10 | 0.01 × 0.02 × 0.08 | |
| Space group | A2/m | ||
| Unit-cell dimensions |
a (Å) | 8.8669(10) | 8.8754(10) |
| b (Å) | 6.0400(7) | 5.8912(4) | |
| c (Å) | 19.328(2) | 19.0796(14) | |
| β (°) | 97.608(12) | 97.398(8) | |
| V (Å3) | 1026.0(2) | 980.39(15) | |
| Dcalc (g/cm3) | 3.46 | 3.20 | |
| Radiation | MoKα (λ = 0.71073 Å) | ||
| Monochromator | VariMax optics | ||
| Diffractometer | Rigaku XtaLAB Synergy-R/DW HyPix-6000HE | ||
| Scan type | ω scan | ||
| Absorption correction | CrysAlisPro | ||
| θmin-max (°) | 2.1-29.6 | 2.3-28.3 | |
| μ (mm−1) | 4.67 | 1.90 | |
| Collected reflections | 4142 | 4077 | |
| Unique reflections | 1540 | 1330 | |
| Rint (%) | 6.13 | 7.53 | |
| Rσ (%) | 7.82 | 9.72 | |
| Miller index limits | −12 ≤ h ≤ 12, −4 ≤ k ≤ 8, −26 ≤ l ≤ 22 | −11 ≤ h ≤ 9, −6 ≤ k ≤ 7, −22 ≤ l ≤ 25 | |
| Refinement on F2 using | SHELXL-2019/3 (Sheldrick, 2015) | ||
| R1 (%) | 5.43 | 7.20 | |
| wR2 (%) | 15.80 | 18.99 | |
| No. of parameters | 131 | 128 | |
| Weighting scheme1) | w = 1/[σ2(Fo2) + (0.0869P)2] | w = 1/[s2(Fo2) + (0.1151P)2] | |
| Δρmax (e Å−3) | 1.48 at 0.76 Å from W1 | 1.67 at 0.85 Å from O9 | |
| Δρmin (e Å−3) | −0.98 at 1.18 Å from O4 | −1.02 at 0.95 Å from W2 | |
1) The function of the weighting scheme is w = 1/(σ2(Fo2) + (a·P)2 + b·P), where P = (Max(Fo2) + 2Fc2)/3, and the parameters a and b are chosen to minimize the differences in the variances for reflections in different ranges of intensity and diffraction angle.
| Kouragahana julgoldite | Sugashima pumpellyite | ||||||||||
| Site | W2) | Refined occupancy | x | y | z | Ueq | Refined occupancy | x | y | z | Ueq |
| W1 | 4i | Ca1.0 | 0.2526(2) | ½ | 0.33952(10) | 0.0190(4) | Ca1.0 | 0.2499(2) | ½ | 0.33935(9) | 0.0160(5) |
| W2 | 4i | Ca1.0 | 0.1892(2) | ½ | 0.15522(10) | 0.0213(4) | Ca1.0 | 0.1891(3) | ½ | 0.15427(10) | 0.0191(5) |
| X | 4f | Fe0.687(11)Mg0.313 | ½ | ¼ | ¼ | 0.0173(5) | Mg0.910(11)Fe0.090 | ½ | ¼ | ¼ | 0.0106(9) |
| Y | 8j | Fe0.728(10)Al0.272 | 0.25443(11) | 0.24670(16) | 0.49573(5) | 0.0156(3) | Al1.0 | 0.2542(2) | 0.2462(3) | 0.49589(9) | 0.0130(5) |
| Si1 | 4i | 1 | 0.0515(3) | 0 | 0.09310(13) | 0.0144(5) | 1 | 0.0503(3) | 0 | 0.08963(12) | 0.0130(6) |
| Si2 | 4i | 1 | 0.1643(3) | 0 | 0.24839(13) | 0.0162(5) | 1 | 0.1667(3) | 0 | 0.24759(12) | 0.0134(6) |
| Si3 | 4i | 1 | 0.4645(3) | 0 | 0.40180(13) | 0.0153(5) | 1 | 0.4654(3) | 0 | 0.40314(12) | 0.0118(6) |
| O1 | 8j | 1 | 0.1389(5) | 0.2219(6) | 0.0746(2) | 0.0170(9) | 1 | 0.1378(6) | 0.2243(7) | 0.0708(2) | 0.0134(10) |
| O2 | 8j | 1 | 0.2639(5) | 0.2277(6) | 0.2466(2) | 0.0180(9) | 1 | 0.2676(6) | 0.2308(7) | 0.2463(2) | 0.0156(10) |
| O3 | 8j | 1 | 0.3658(5) | 0.2188(6) | 0.4137(2) | 0.0182(9) | 1 | 0.3668(6) | 0.2227(7) | 0.4176(2) | 0.0141(10) |
| O4 | 4i | 1 | 0.1275(7) | ½ | 0.4430(3) | 0.0155(12) | 1 | 0.1296(8) | ½ | 0.4446(3) | 0.0136(14) |
| O5 | 4i | 1 | 0.1260(7) | 0 | 0.4580(3) | 0.0181(13) | 1 | 0.1321(9) | 0 | 0.4581(3) | 0.0184(16) |
| O6 | 4i | 1 | 0.3737(7) | ½ | 0.0454(4) | 0.0199(14) | 1 | 0.3709(8) | ½ | 0.0451(3) | 0.0124(14) |
| O7 | 4i | 1 | 0.3761(7) | 0 | 0.0327(4) | 0.0206(14) | 1 | 0.3674(8) | 0 | 0.0326(3) | 0.0164(15) |
| O8 | 4i | 1 | 0.0344(7) | 0 | 0.1779(3) | 0.0187(13) | 1 | 0.0354(8) | 0 | 0.1759(3) | 0.0145(14) |
| O9 | 4i | 1 | 0.4746(7) | ½ | 0.1754(3) | 0.0192(13) | 1 | 0.4785(9) | ½ | 0.1761(3) | 0.0177(15) |
| O10 | 4i | 1 | 0.0650(7) | 0 | 0.3136(4) | 0.0218(14) | 1 | 0.0679(9) | 0 | 0.3136(3) | 0.0238(17) |
| O11 | 4i | 1 | 0.5016(7) | ½ | 0.3145(3) | 0.0194(14) | 1 | 0.5021(8) | ½ | 0.3139(3) | 0.0170(15) |
| H5 | 4i | 1 | 0.0149(13) | 0 | 0.447(6) | 0.05(Uiso) | |||||
| H7 | 4i | 1 | 0.4873(12) | 0 | 0.044(6) | 0.05(Uiso) | 1 | 0.4797(10) | 0 | 0.037(6) | 0.05(Uiso) |
| H10 | 4i | 1 | 0.115(8) | 0 | 0.3619(13) | 0.05(Uiso) | 1 | 0.120(8) | 0 | 0.3622(14) | 0.05(Uiso) |
| H11 | 4i | 1 | 0.506(13) | ½ | 0.3654(6) | 0.05(Uiso) | 1 | 0.508(16) | ½ | 0.3655(7) | 0.05(Uiso) |
1) Thermal parameter for H atoms was fixed as 0.05 Uiso.
2) W, Wykoff notation of point position.
| Kouragahana julgoldite | Sugashima pumpellyite | |||||||||||
| Site | U11 | U22 | U33 | U23 | U13 | U12 | U11 | U22 | U33 | U23 | U13 | U12 |
| W1 | 0.0241(9) | 0.0163(8) | 0.0165(10) | 0 | 0.0036(8) | 0 | 0.0164(12) | 0.0147(9) | 0.0164(9) | 0 | −0.0004(8) | 0 |
| W2 | 0.0339(11) | 0.0130(8) | 0.0161(10) | 0 | 0.0014(8) | 0 | 0.0284(14) | 0.0093(9) | 0.0176(9) | 0 | −0.0040(8) | 0 |
| X | 0.0216(9) | 0.0114(7) | 0.0191(9) | 0.0011(6) | 0.0038(6) | 0.0001(6) | 0.0118(17) | 0.0042(13) | 0.0156(14) | −0.0006(9) | 0.0014(10) | 0.0004(10) |
| Y | 0.0199(6) | 0.0110(5) | 0.0162(6) | 0.0002(4) | 0.0032(4) | 0.0007(4) | 0.0142(12) | 0.0076(9) | 0.0162(9) | 0.0004(6) | −0.0014(7) | −0.0004(8) |
| Si1 | 0.0196(12) | 0.0087(10) | 0.0154(13) | 0 | 0.0042(10) | 0 | 0.0143(16) | 0.0079(11) | 0.0158(12) | 0 | −0.0022(10) | 0 |
| Si2 | 0.0215(12) | 0.0127(11) | 0.0141(13) | 0 | 0.0015(10) | 0 | 0.0140(16) | 0.0068(11) | 0.0182(12) | 0 | −0.0029(10) | 0 |
| Si3 | 0.0187(12) | 0.0116(10) | 0.0151(13) | 0 | 0.0015(9) | 0 | 0.0105(15) | 0.0088(11) | 0.0156(12) | 0 | −0.0006(10) | 0 |
| O1 | 0.020(2) | 0.014(2) | 0.018(2) | −0.0028(17) | 0.0063(18) | −0.0032(16) | 0.017(3) | 0.0068(19) | 0.0151(19) | 0.0007(14) | −0.0012(17) | −0.0022(17) |
| O2 | 0.023(2) | 0.0122(19) | 0.020(2) | −0.0036(18) | 0.0057(18) | 0.0000(16) | 0.020(3) | 0.008(2) | 0.018(2) | −0.0019(15) | −0.0012(18) | −0.0033(17) |
| O3 | 0.021(2) | 0.012(2) | 0.021(2) | 0.0024(16) | 0.0039(18) | 0.0046(15) | 0.017(3) | 0.009(2) | 0.0161(19) | 0.0033(15) | 0.0017(18) | 0.0024(17) |
| O4 | 0.019(3) | 0.016(3) | 0.011(3) | 0 | 0.001(2) | 0 | 0.015(4) | 0.006(3) | 0.018(3) | 0 | −0.003(3) | 0 |
| O5 | 0.018(3) | 0.018(3) | 0.018(4) | 0 | −0.002(3) | 0 | 0.025(5) | 0.013(3) | 0.015(3) | 0 | −0.007(3) | 0 |
| O6 | 0.022(3) | 0.017(3) | 0.020(4) | 0 | 0.001(3) | 0 | 0.011(4) | 0.008(3) | 0.016(3) | 0 | −0.004(2) | 0 |
| O7 | 0.021(3) | 0.016(3) | 0.022(4) | 0 | −0.004(3) | 0 | 0.008(4) | 0.013(3) | 0.027(3) | 0 | −0.003(3) | 0 |
| O8 | 0.017(3) | 0.022(3) | 0.017(3) | 0 | 0.001(3) | 0 | 0.013(4) | 0.015(3) | 0.014(3) | 0 | −0.005(3) | 0 |
| O9 | 0.024(3) | 0.020(3) | 0.015(3) | 0 | 0.005(3) | 0 | 0.012(4) | 0.027(4) | 0.015(3) | 0 | 0.002(3) | 0 |
| O10 | 0.020(3) | 0.030(3) | 0.016(3) | 0 | 0.005(3) | 0 | 0.021(5) | 0.030(4) | 0.018(3) | 0 | −0.005(3) | 0 |
| O11 | 0.024(3) | 0.019(3) | 0.016(3) | 0 | 0.007(3) | 0 | 0.011(4) | 0.023(3) | 0.017(3) | 0 | 0.000(3) | 0 |
| Kouragahana julgoldite |
Sugashima pumpellyite |
Kouragahana julgoldite |
Sugashima pumpellyite |
||||
| W1 -O2 | ×2 | 2.448(4) | 2.401(4) | Si1 -O1 | ×2 | 1.612(4) | 1.594(4) |
| -O3 | ×2 | 2.359(4) | 2.358(4) | -O4 | 1.647(7) | 1.632(7) | |
| -O4 | 2.413(6) | 2.387(7) | -O8 | 1.666(7) | 1.667(6) | ||
| -O8 | 2.522(6) | 2.489(8) | Mean | 1.634 | 1.622 | ||
| -O11 | 2.323(6) | 2.331(8) | |||||
| Mean | 2.410 | 2.389 | Si2 -O2 | ×2 | 1.637(4) | 1.626(4) | |
| -O8 | 1.663(7) | 1.673(6) | |||||
| W2 -O1 | ×2 | 2.294(4) | 2.279(4) | -O10 | 1.629(7) | 1.619(8) | |
| -O2 | ×2 | 2.440(4) | 2.400(4) | Mean | 1.642 | 1.636 | |
| -O6 | 2.847(7) | 2.785(7) | |||||
| -O9 | 2.509(7) | 2.526(8) | Si3 -O3 | ×2 | 1.618(4) | 1.616(5) | |
| -O10 | 2.410(7) | 2.418(9) | -O6 | 1.647(7) | 1.638(7) | ||
| Mean | 2.462 | 2.441 | -O9 | 1.653(7) | 1.650(7) | ||
| Mean | 1.634 | 1.630 | |||||
| X -O2 | ×2 | 2.090(4) | 2.039(5) | Y -O1 | 1.956(4) | 1.870(5) | |
| -O9 | ×2 | 2.079(4) | 2.031(4) | -O3 | 1.983(4) | 1.900(5) | |
| -O11 | ×2 | 1.957(4) | 1.910(4) | -O4 | 2.084(4) | 2.030(4) | |
| Mean | 2.042 | 1.993 | -O5 | 1.957(4) | 1.891(5) | ||
| V(VI) (Å3) | 11.25 | 10.47 | -O6 | 1.998(4) | 1.947(4) | ||
| DI (oct) | 0.028 | 0.028 | -O7 | 1.952(4) | 1.881(4) | ||
| <λ oct> | 1.007 | 1.007 | Mean | 1.988 | 1.920 | ||
| σθ (oct)2 | 19.71 | 18.61 | V(VI) (Å3) | 10.34 | 9.29 | ||
| DI (oct) | 0.018 | 0.024 | |||||
| O5···O12) | 2.889(7) | <λ oct> | 1.010 | 1.011 | |||
| O5···O52) | 2.931(13) | σθ (oct)2 | 30.67 | 35.09 | |||
| O5···O102) | 2.772(10) | 2.742(8) | |||||
| O7···O32) | 2.927(7) | 2.909(8) | Si1-O8-Si2 | 131.5(4) | 132.3(5) | ||
| O7···O72) | 2.676(15) | 2.781(15) | |||||
| O7···O112) | 3.008(10) | 3.005(8) | |||||
| O11···O32) | 2.936(7) | 2.934(8) | |||||
1) DI is the bond-length distortion parameter (Baur, 1974). <λ> and σθ2 are the bond-length and angle variance, respectively (Robinson et al., 1971).
2) D···A oxygen donor-acceptor distances.
Table 3 lists the determined hydrogen positions. The oxygen atoms at O5, O7, O10, and O11 positions behave as donors of the OH groups, which is supported by previous studies (Allmann and Donnay, 1971, 1973; Yoshiasa and Matsumoto, 1985; Brigatti et al., 2006; Nagashima et al., 2006; Nagashima and Akasaka, 2007; Hamada et al., 2010; Nagashima et al., 2010, 2018a, 2018b). In the Kouragahana julgoldite, the hydrogen positions at H5, H7, H10, and H11 were determined, and those, except for H5, were determined in the Sugashima pumpellyite. However, based on the bond-valence sum (Brown and Altermatt, 1985), the value of the O atom at O5 was 1.04 valence units, indicating that the O atom at O5 in the Sugashima pumpellyite acts as a donor for the OH group. Although the disordered distributions of H atoms for O5 and O7 have been reported by Nagashima et al. (2010, 2018a, 2018b), they are not confirmed in this study, likely due to inadequate data quality.
Raman spectra of julgoldite and pumpellyiteRaman spectra were collected from randomly oriented crystals of julgoldite and pumpellyite. The Raman spectra collected in this study are in good agreement with those in the RRUFF database (Fig. 2). Based on these spectra, the well-defined intense peak around 695 cm−1 is characteristic of the pumpellyite-group minerals, which can be attributed to the Si1-O8br-Si2 bending mode. In julgoldite, this peak shifts to a lower wavenumber than in pumpellyite. Six to seven peaks were observed in the O-H stretching region. While Kasatkin et al. (2021) suggested the presence of a limited quantity of water molecules within julgoldite-(Fe3+) based on their infrared spectral analysis, the Raman spectra collected repeatedly from several crystals in our study did not show the H-O-H bending mode near 1650 cm−1, indicating the absence of water molecules in our samples. It is worth noting that the cations at the X site do not participate in Raman-active modes, despite being infrared-active based on group theory.
The refined site occupancies at the X and Y sites in the Kouragahana julgoldite were Fe0.687(11)Mg0.313 and Fe0.728(10)Al0.272, respectively (Table 3). The total Fe content in the final refinement was approximately 2.14 atoms per formula unit (apfu), which is consistent with the Fe content of 2.20(7) apfu obtained by EPMA within a standard deviation (Table 1). As a result of EPMA, the Al and Mg contents are 0.65 and 0.11 apfu, respectively. In terms of the small ionic radius of Al (0.535 Å; Shannon, 1976), Al ions predominantly occupy the smaller octahedral Y site. Thus, the refined Al occupancy at the Y site (= 0.272) is interpreted as all Al ions. All Fe atoms in the Y site can be interpreted as Fe3+. Although different assignments of Fe2+ and Fe3+ have been proposed by several Mössbauer studies (Artioli and Geiger, 1994; Akasaka et al., 1997; Nagashima et al., 2006, 2018a; Kasatkin et al., 2021; Akasaka et al., 2023), the correlation between the b-axis and the mean ionic radius of the Y site, which will be described in the next chapter, also supports the idea that all Fe at the Y site is in the ferric state. The refined Mg occupancy at the X site includes Al due to their similar X-ray scattering factors. Therefore, compared to the results from EPMA, the excess Mg obtained by structural refinement is considered to be Al. Based on the results of the chemical analysis, the variation in Mg content was relatively constant compared to that of Al content; therefore, the Mg content was fixed at the value obtained by EPMA. Consequently, the site occupancy of the X site was revised to Fe0.69Al0.20Mg0.11 (Table 6). The dominant cation at the X site is Fe, but it is impossible to determine whether ferric or ferrous iron predominates. Currently, we can only conclude that Kouragahana julgoldite is either julgoldite-(Fe3+) or julgoldite-(Fe2+).
| Site | Determined occupancies | No. e− |
| Kouragahana julgoldite | ||
| X | Fe0.69Al0.20Mg0.11 | 21.86 |
| Y | Fe0.73Al0.27 | 22.49 |
| Sugashima pumpellyite | ||
| X | Al0.56Mg0.35Fe0.09 | 13.82 |
| Y | Al1.0 | 13 |
The refined site occupancies at the X and Y sites in the Sugashima pumpellyite were Mg0.910(11)Fe0.090 and Al1.0, respectively (Table 3). The Fe content in the final refinement was ∼ 0.09 apfu, which is less than the Fe content of 0.16(2) apfu derived from EPMA (Table 1). The crystal used for the structural analysis is possibly poor in Fe. The measured Al and Mg contents determined by EPMA were 2.52 and 0.32 apfu, respectively. Since the Y site is fully occupied by Al, the remaining Al, 0.52 apfu (= 2.52-1.0 × 2), is assigned to the X site. The Fe ions occupy the X site with the Al and Mg. Since the compositional ranges of Al, Mg, and Fe are similar, the Al:Mg ratio derived from the average composition, 0.52:0.32, was applied to the site occupancy of the X site. Therefore, the X site occupancy in the Sugashima pumpellyite can be written as Al0.56Mg0.35Fe0.09 (Table 6). Thus, the dominant cations of the X and Y sites are both Al. Similar to Kouragahana julgoldite, while the oxidation state of Fe cannot be determined directly, it may be trivalent concerning closely associated phases, such as epidote and prehnite. Although a chemical composition thought to correspond to pumpellyite-(Mg) was reported from the same locality by Seki (1958), the Sugashima pumpellyite we studied is richer in Al and is classified as pumpellyite-(Al).
Influence of cation behavior on the crystal structure of pumpellyite-julgoldite series mineralsPrevious studies have reported that the unit-cell volume of pumpellyite-group minerals is strongly dependent on the mean ionic radius of the Y site, and the <X-O> distance hardly reflects the cation distribution at the X site (e.g., Artioli et al., 2003; Nagashima et al., 2006). The unit-cell volume of Sugashima pumpellyite is one of the smallest among the previously reported pumpellyite-group minerals, in which Al ions fully occupy the Y site (Fig. 4). In particular, the variation of the b-axis has been considered the best indicator to reflect the mean ionic radius of the Y site despite both the X and Y octahedra forming independent edge-shared octahedral chains running along the b-axis. However, julgoldite-(Fe3+), reported by Artioli et al. (2003), is an exception, with a short b-axis deviating from the regression line. In general, although the mean ionic radius of the X site in pumpellyite-group minerals is larger than that of the Y site, this specimen is the only one in which the mean ionic radius of the X site (0.616 Å) is smaller than that of Y (0.645 Å). Nagashima et al. (2018a) interpreted that this short b-axis is due to the small mean X ionic radius, and the cation assignment at the X site also affects the variation of the b-axis. The variation of octahedra, which has a smaller mean ionic radius, may be reflected in the b-axis. The a- and c-axes of julgoldite tend to be spread more than Al-rich ones. It is explained by the lateral extension or shrinkage in the (010) plane. If the b-axis is governed by the size of the Y site [julgoldite-(Fe3+) studied by Artioli et al. (2003) is an exception.], the (010) plane acts as a buffer to compensate for the larger X site (Nagashima et al., 2018a). Samples with longer values for the a- and c-axes and the unit-cell volume than estimated by the regression lines can be rich in the divalent cations at the X site. In other words, they are likely to be julgoldite-(Fe2+). The a- and c-axes of julgoldite-(Fe3+) studied by Artioli et al. (2003) are shorter than the expected values using the regression line. Thus, the a- and c-axes of julgoldite provide clues for the cation assignment to the X site. However, the amounts of ferric and ferrous iron in the X site can only be estimated if the Mg and Al contents assigned to the X site are very low or if each is determined accurately. The degree of hydroxylation should also be considered because the ratio of Me2+:Me3+ directly affects the number of OH groups. In pumpellyite structures, the disordered distributions of H atoms for O5 and O7 have been reported by Nagashima et al. (2010, 2018a, 2018b), implying that multiple hydrogen bonding systems are present in the crystal structure. The lack of disordered hydroxyl groups in this study cannot be denied as a reflection of the ratio of Me2+:Me3+ at the X site, in addition to the quality of the data collected. However, it is evident that the charge of the cations distributed at the X sites affects the number of OH groups, so further investigation into these relationships is required.
Using the mean ionic radius of the Y site, 0.615 Å, calculated from the site occupancy at the Y site in Kouragahana julgoldite, Fe3+0.73Al0.27, the a- and c-axes can be estimated as 8.894 and 19.328 Å, respectively, by using the regression lines shown in Figure 4. The estimated c-axis is consistent with the observed one within the standard deviation, while the estimated a-axis is longer than the observed one [a = 8.8669(10) and c = 19.328(2) Å, Table 2]. The site occupancy of the X site in this sample, determined by considering the chemical composition, is Fe0.69Al0.20Mg0.11. The shortage of an a-axis suggests the enrichment of trivalent cations at the X site; however, the effect of the Al ion is also significant.
The size of the X site shows little correlation with the mean ionic radius of the X site, which can be interpreted as either the structural topology requirements governing the X site or as a correlation remaining undetected due to insufficient structural data. Nagashima et al. (2018a) indicated that the weak correlation between the mean ionic radius of the X site and the volume of the X site exists if julgoldite-(Fe3+) is recognized as an outlier (Fig. 5). For samples containing both ferric and ferrous iron, only those for which the ratios have been directly determined, are plotted. The solid regression line in Figure 5 is obtained from such results, except for the outlier julgoldite-(Fe3+) by Artioli et al. (2003). The dashed line assumes a regular octahedron with the X-O distance obtained from the mean ionic radius of the central cation plus oxygen atom (1.35 Å; Shannon, 1976). Given the potential to control the X site by the structural topology requirements, the application of the dashed line, assuming an ideal octahedron, is questionable. On the other hand, the regression line (solid line) shows a weak correlation; therefore, its accuracy is poor. Both have problems, but they may be helpful as guidelines. Horizontal solid lines span the range of ionic radii at the X site if all Fe is either assumed ferric or ferrous for Kouragahana (filled green star; 0.631-0.724 Å in this study), Kreimbach/Kaulbach (filled green square, 0.638-0.729 Å in Nagashima et al., 2018a), and Långban julgoldite samples (filled green circle, 0.645-0.780 Å in Allmann and Donnay, 1973). Fe2+ is predominant in the Kreimbach/Kaulbach julgoldite, and Fe2+ rather than Fe3+ also dominates in the Långban julgoldite (Fig. 5). Therefore, these two specimens can be identified as julgoldite-(Fe2+), which is supported by their long a- and c-axes beyond regression lines (Fig. 4). However, the Kouragahana julgoldite is almost in the middle, with Fe2+:Fe3+ = 1:1. The estimated cation occupancies at the X site are Fe3+0.40Fe2+0.29Al0.20Mg0.11 using the solid regression line and Fe2+0.39Fe3+0.30Al0.20Mg0.11 assuming the ideal octahedron (dashed line). Combined with the short a-axis observed in the Kouragahana sample, there may be a slight predominance of Fe3+; however, this cannot be determined. A comprehensive examination of the hydroxyl groups is also essential for understanding the pumpellyite group minerals due to their significant relationship with the oxidation states of cations at the X site. In this study, the determined H positions are assumed to be fully occupied by hydrogen. However, regarding the hydrogen bonding system, the O atoms at O5 and O7 act as donor oxygen and acceptors, so the H5 and H7 positions would not be fully occupied by the H atom. In addition, the O11-H11 hydroxyl, directly bonded to the X cation, may exhibit partial occupation. The six to seven Raman peaks arising from the O-H bond stretching reflected the complicated H-bond system in the pumpellyite group structure (Fig. 2). These multiple O-H stretching peaks are due not only to the presence of multiple hydroxyl groups, O5-H5, O7-H7, O10-H10, and O11-H11 but also to differences in the local chemical environment caused by compositional variation. For example, when four species of cations, such as Fe3+, Fe2+, Al, and Mg, occupy the X site, like in our julgoldite, XFe3+-O11-H11, XFe2+-O11-H11, XAl-O11-H11, and XMg-O11-H11 are expected to generate the different O-H stretching mode because O···O distance varies with each cation. The shortage of local O···O distance resulting from the substitution of the larger cation increases the strength of the H···O bond, which in turn reduces the O-H bond strength. In this case, a lower-energy peak appears. A similar phenomenon occurs at least at the Y site, which typically has the Al-Fe3+ substitution. Therefore, the assignment of the O-H stretching peaks in pumpellyite-group minerals remains unresolved.
While a close relationship between the distribution of cations and the hydrogen bond system is evident, a complete understanding of their connection requires further investigation. Specifically, enhanced regression line analysis using high-quality diffraction data from pumpellyite-julgoldite series minerals, along with the established oxidation state of iron and a better understanding of OH group behavior, will help address this issue. Despite the common occurrence of pumpellyite-julgoldite series minerals, accurately identifying their proper names remains challenging due to complex chemical compositions and crystal structures.
This study uses research equipment shared in the MEXT Project for Promoting Public Utilization of Advanced Research Infrastructure (Program for supporting construction of core facilities) grant no. JPMXS0440400024 and also supported by Core Clusters for Research Initiative of Yamaguchi University. We thank the editor, Dr. M. Hamada, as well as two anonymous reviewers for their constructive comments. We gratefully acknowledge the financial support of Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science, No. JP23K0355, and Sasakawa Scientific Research Grant from Japan Science Society to MN, No. 17-105.
Supplementary CIF files are available online from https://doi.org/10.2465/jmps.250401.
