Physical and optical properties

At the type locality, schizolite occurs as stacked prismatic columns, up to 2 cm long. These occur as either individual crystals or aggregates of crystals. The mineral is pale pink to brownish with a white streak and vitreous lustre. It is translucent to opaque, non-fluorescent and has a hardness (Mohs) ≈ 5. The mineral is brittle with two perfect cleavages making it splintery or somewhat fibrous. The calculated density is 3.09 g cm^–3.

Pale pink schizolite (NHMD 1899.856) is biaxial (+) with α = 1.626 ± 0.003, β = 1.630 ± 0.002, γ = 1.661 ± 0.002, 2V_meas = 72(4)° and 2V_calc = 40°. Pale brown schizolite (NHMD 1899.8) is biaxial (+) with α = 1.640 ± 0.003, β = 1.643 ± 0.002, γ = 1.658 ± 0.002, 2V_meas = 52(2)°, 2V_calc = 49°. There is no pleochroism observed in either specimen. Optical parameters were measured on a crystal fragment using a spindle stage with Na light, λ = 590 nm, and the measured 2V was determined from extinction curves. It was particularly difficult to measure α in this mineral, as the observation direction is down a crystal fragment edge. Comparative optical properties for current and other members of the series are given in Table 1.

Table 1.

Comparison of schizolite, ‘marshallsussmanite’, serandite and pectolite.

Chemical analysis

The chemical analyses were performed on a JEOL 8230 electron microprobe operating in wavelength-dispersion (WD) mode using Probe for EPMA software (https://www.probesoftware.com). The operating voltage was 20 kV, the beam current was 20 nA and the beam diameter was 10 µm. At these operating conditions, no intensity change due to sample damage was observed. Peak intensities were counted for 20 s and backgrounds were calculated using a mean atomic number correction (Donovan et al., Reference Donovan, Lowers and Rusk2011). The following lines and standards were used for quantification: albite for NaKα; diopside for MgKα and CaKα; sanidine for KKα and AlKα; hematite for FeKα; rutile for TiKα; tephroite for SiKα and MnKα; sanbornite for BaLα; and chromite for CrKα. Raw intensities were converted to concentrations using the default φρZ corrections of the Probe for EPMA software (c.f. Armstrong, Reference Armstrong1988). Results of microprobe analyses are presented in Table 2. The final chemical formulae (Table 1) are calculated for each analyses on the basis of 9 anions per formula unit (apfu); H₂O (wt.% oxide) was calculated on the basis of stoichiometry, with (OH) = 1 apfu.

Table 2.

Chemical analyses: schizolite and pectolite from Tutop Agtakôrfia.

¹Na_1.02Ca_1.09Mn_0.69Fe_0.12Si_3.04H₁O₉; ²Na_0.98Ca_1.12Mn_0.67Fe_0.12Si_3.04H₁O₉; atomic fractions calculated of the basis of 9 oxygen atoms.

* H₂O calculated on the basis of 1 H apfu

Powder X-ray diffraction

X-ray diffraction data (Table 3) for schizolite NHMD 1899.856 were acquired with a Bruker D8 discover equipped with CuKα radiation and a Deskris Eiger2R_500K detector, calibrated at a distance of 17.562 cm using a statistical calibration (Rowe, Reference Rowe2009). Schizolite, as expected, is characterised by an intermediate diffraction pattern between end-members of the pectolite–serandite series. The unit cell was refined in the triclinic system, based on the powder data, thanks to the increased detector resolution and the flexibility the Rietveld software for cell-refinement purposes. The recent improvement in data resolution and area detection technology facilitates the characterisation of lower symmetry structure, for which accurate pattern indexing was historically difficult and often impossible.

Table 3.

Powder X-ray diffraction data for schizolite (Kα₁) (sample# NHMD 1899.856)

* Calculated from structure data modified with Rietveld software.

Refined cell a = 7.85062(14) b = 6.97283(8) c = 6.91131(11) Å, α = 90.695(2), β = 94.5174(18), γ = 102.8861(12)°, V = 367.487(2) ų

The strongest lines are given in bold

The discrepancy between the measured and calculated intensities in Table 3, largely result from the inability of the diffraction experiment to resolve the high-intensity peaks at d = 2.8779 and d = 2.8725 Å, corresponding to hkl indices (120) and (2${\bar 2}$0), respectively. The (120) is the maximum (100%) peak in the calculated powder pattern, whereas (2${\bar 2}$0) is calculated to be 93% of this intensity. As the 100% reference in the experimental pattern consists of both calculated peaks, the observed peak intensities are systematically, proportionally lower than predicted. Accounting for this would result in a better match between observed and calculated data.

Single-crystal X-ray diffraction data and structure refinement

Two crystal-structure analyses on schizolite were carried out. In both cases, single crystals were fixed to glass fibres and a Bruker single-crystal diffractometer equipped with an APEX II area detector and a microfocus sealed X-ray tube operating at 50 kV and 0.99 mA, located at the X-ray Crystallography Laboratory at the University of British Columbia (Vancouver, Canada), was used for the collection of intensity data. In all cases, data were collected with MoKα radiation at 293(2) K and measured φ and ω scans of 0.5° per frame. The APEX III software package was used to determine a data collection strategy that ensured complete coverage of the Ewald Sphere with reasonable data redundancy. The program SAINT (V3.38, Bruker, 2013) was used to index and refine the final unit-cell parameters, as well as reduce, scale and apply an adsorption correction.

Non-merohedral twinning was observed in sample NHMD 1899.856. Using CELL_NOW (V. 2008/4; Sheldrick, Reference Sheldrick2008), more than two domains were found; these were verified visually using the RLATT routine available in the APEX III software package. Only the two most prominent domains (rotated 2.9° about the b axis) were used during the integration routine. The structure was refined on a merged data set of the two domains. TWINABS-2012/1 (Bruker, 2012) was used to calculate a multi-scan absorption correction. The twinned structure was solved against an HKLF4 file, containing reflection intensities averaged over the two twin domains, and was refined again with a HKLF5 file, containing domain-separated data. No twinning was observed in the data for sample NHMD 1899.8.

For the crystal taken from NHMD 1899.856, the structure was solved with Direct Methods using SHELXS-97 (Sheldrick, Reference Sheldrick1997) and refined by the Least Squares Method using version 2014 of SHELXL (Sheldrick, Reference Sheldrick2015). For the crystal taken from NHMD 1899.8, the structure was refined against the structure presented in Ohashi and Finger (Reference Ohashi and Finger1978). The transformation matrix from Ohashi and Finger (Reference Ohashi and Finger1978) to our cell is [−½−½0 / 00−1 / −½−½0]. The M1 and M2 sites were refined with split-occupancies (with the constraint Mn + Ca = 1). All positions were refined anisotropically, with the exception of hydrogen (U _iso = 0.05Ų), which, in all cases was readily observed in the difference-Fourier prior to being included in the refined model. The O–H distance was allowed to refine freely to O3–H1 = 1.22(1), O4–H1 = 1.28(1) and O3–H1····O4 = 2.4726(4) Å. Scattering factors of neutral atoms were taken from the International Tables for X-ray Crystallography (Ibers and Hamilton, Reference Ibers and Hamilton1974). For all crystals, assigning phases to a set of normalised structure-factors gave mean values of |E ² – 1| in excess of 0.95, in accord with a centrosymmetric space-group. Weighting parameters were manually updated during final Least Square cycles, and this process was repeated until convergence. For sample NHMD 1899.856, details of collection and refinement are given in Table 4, and final atomic coordinates are given in Table 5; structural formulae, calculated by site-refinement are given in Table 1. Compatible structure data for sample NHMD 1899.8 have been deposited as supplementary material (see below). The final structure is entirely consistent with previously reported intermediate compositions of the serandite–pectolite solid-solution series (Fig. 2). It shows the tri-periodic [Si₃O₁₀] chains, dreierinfach chains that zigzag parallel to c between double-wide ribbons of octahedra are made up of two crystallographically-unique octahedrally-coordinated sites, M1 and M2 (Fig. 2). The final structures for NHMD 1899.856 and NHMD 1899.8 are very similar, but not identical i.e. small differences in Mn²⁺ and Ca content but consistent in the ordering of the M1 and M2 sites. Summary data for four minerals are given in Table 1. Consistent with previous authors, (Nagashima et al., Reference Nagashima, Imaoka, Fukuda and Pettke2018; Ohashi and Finger, Reference Ohashi and Finger1978; Prewitt, Reference Prewitt1967; Rozhdestvenskaya and Vasilieva, Reference Rozhdestvenskaya and Vasilieva2014) Ca and Mn preferentially order at M1 and M2, respectively, resulting in the ideal structural formula for schizolite being written as Na(^M1Ca^M2Mn)Si₃O₈(OH).

Fig. 2.

The schizolite crystal structure. It shows the tri-periodic [Si₃O₁₀] chains that zigzag parallel to c between double-wide ribbons of edge-sharing octahedra. The ribbon is formed by dimers of M1 (yellow) and M2 (orange) octahedra.

Table 4.

Data and experimental details for schizolite, Tutop Agtakôrfia NHMD 1899.856 (cotype), Greenland.

$R_1 = \sum \vert {\vert {F_{obs}} \vert - \vert {F_{calc}} \vert } \vert / \sum \vert {F_{obs}} \vert $; $wR_2 = \left[ {\sum \left \lsqb {w{ \left \lpar {F_{obs}^2 -F_{calc}^2} \right \rpar }^2} \right \rsqb / \sum w{\left \lpar {F_{obs}^2} \right \rpar }^2} \right]^{1/2} $ G$oF = \left[ {\sum \left \lsqb {w{\left \lpar {F_{obs}^2 -F_{calc}^2} \right \rpar }^2} \right \rsqb /\lpar {n - p} \rpar } \right]^{1/2}$w = 1/[δ²(F ₀²) + (0.0269 x P)² + 0.20 x P] where P = (Max (F ₀², 0) + F _c²) / 3

Table 5.

Final refined atomic coordinates, site-occupancy factors (s.o.f.) and thermal parameters for schizolite Tutop Agtakôrfia, Greenland (NHMD 1899.856).

M1* s.o.f. Ca 0.792(1) and Mn 0.208(1); M2* s.o.f. Ca 0.792(1) and Mn 0.208(1)

U _iso parameter fixed during refinement; BVS – bond-valence sum.