Occurrence

After the investigation of millsite had begun, it was established more specifically that the specimens being studied were from Gråurdfjellet in Oppdal kommune, Norway (62°29'11”N, 9°29'25''E) and were almost certainly part of a find made in 1991 by Kjell Arve Isbrekken which was significant for containing gold and tellurium minerals and a supposed new mineral. It also transpired that a sample of the new mineral had been obtained by one of the authors (FM) from K.A. Isbrekken (specimen KAI-93). The crystal structure of this sample was determined by single-crystal XRD using AgKα radiation (Mo, F., Larsen, F.K., Mathiesen, R., Måseide, K. and Overgaard, J. (Reference Mo, Larsen, Mathiesen, Måseide and Øvergaard2000) Unpublished results). The crystallographic data showed unequivocally that the new mineral is distinct from teineite and likely to be a new mineral. This new mineral is millsite and it is essentially the same as the material donated recently to the NHM via M.S.R.

Millsite occurs as part of an unusual suite of secondary copper minerals found within a granular quartz and chlorite/muscovite matrix. Traces of primary ore minerals were identified within the matrix when the specimen was split for analysis. These have been shown to consist of digenite/chalcocite, hessite and small specks of native gold. Uraninite, monazite, dravite and euxenite have also been reportedly identified within the matrix (http://www.mindat.org retrieved 05/2016) but were not observed during this study. The secondary phases present occur in vughs, cracks and interstices between the granular quartz, and aside from millsite include teineite, malachite, U-rich mcalpineite, schmitterite, a copper sulfate (probably brochantite), goethite and an amorphous, heterogeneous Cu–Te–Si gel-like component with variable chemistry.

On the type specimen, BM 2011,243 (Fig. 1), millsite occurs in one significant cavity that is completely filled by the mineral. Millsite and teineite occur separately on each specimen from Gråurdfjellet in the NHM, other examples from the investigation by Mo et al., (2000) show that the two polymorphs can occur together, intergrown on the same specimen. The geological origin of the millsite-bearing specimens and the entire suite including primary tellurides and gold is unclear. It is reported anecdotally that all the known specimens were extracted from a single boulder, rather than from an in situ geological unit. Recent trips to the area have failed to find any remains of the boulder or any signs of similar mineralization (R. Kristiansen, pers. comm. to M.S.R., 2015). It is possible that the boulder is a glacial erratic from an as yet unidentified Cu–Au–Ag–Te deposit.

Fig. 1.

Holotype millsite BM2011,243. Width of field of view = 6 mm. All the blue phase is millsite.

Millsite and other secondary minerals from the boulder probably formed through the oxidation of primary sulfides and tellurides under near-surface oxidizing conditions in the presence of water. Such an origin is consistent with the secondary tellurium mineralization observed at Otto Mountain and Centennial Eureka mine where similar phases are observed (Kampf et al., Reference Kampf, Housley, Mills, Marty and Thorne2010; Marty et al., Reference Marty, Kampf, Housley, Mills and Weiβ2010; Housley et al., Reference Housley, Kampf, Mills, Marty and Thorne2011).

Physical and optical properties

Millsite occurs on the holotype as cyan blue (HTML 0099CC) to pale royal blue (HTML 0066FF) transparent crystalline grains infilling cavities, fractures and interstices in a granitic matrix. On this specimen the millsite fills a cavity 5 mm × 2 mm × 2 mm in size. Since approval of millsite as a new species, numerous larger specimens have been tentatively identified in private collections and examples can be seen on http://www.mindat.org (retrieved 05/2016). The largest of these aggregates is ~20 mm × 15 mm × 5 mm, with distinct crystals in the mm range. It is of significant note that even though there is a subtle colour difference between teineite (darker royal blue) and millsite on the two specimens at the NHM, the authors do not consider that this is an infallible way to distinguish the two phases. It is clear that this feature has been used by collectors to distinguish the two phases, but tests by one of the authors (MSR) and other researchers have shown that specimens having a cyan colour are still teineite. Not enough of the specimens now known to exist were available to the authors to conclusively determine reliable differences in colour.

There are no discrete terminated crystals and so forms and twinning are difficult to assess macroscopically. Tiny broken fragments show perfect cleavage and appear platy on (100). Millsite is transparent with a vitreous lustre and is non-fluorescent in both shortwave and longwave ultraviolet light. Millsite has a pale green streak. Hardness and density could not be measured due to paucity of sample, but using the determined empirical formula and volume the calculated density is 3.963 g cm^–3.

The mineral is optically biaxial (+) with optical constants measured in white light: α = 1.756(5), β = 1.794(5) and γ = 1.925_calc. Due to the reaction of millsite with index liquids with n > 1.8, γ was calculated based on α, β and 2V. The 2V angle was determined by conoscopic observation on a spindle stage as 60(1)° and obvious r < v dispersion was observed. The optical orientation is Y = b, X ^ a = 41° in obtuse β; one optical axis ≈ ⊥ to (001). The crystal studied is weakly pleochroic: X ≈ Z > Y, greenish blue.

The Gladstone-Dale compatibility index, 1–(K_p/K_c) as defined by Mandarino (Reference Mandarino1981), provides a measure of the consistency among the average index of refraction, calculated density and chemical composition. For millsite of end-member CuTeO₃·2H₂O composition the compatibility index is 0.0095, indicating ‘superior’ agreement among these data. Based on the empirical composition the compatibility index is 0.056, indicating ‘good’ agreement.

Chemical composition

A small fragment of holotype specimen BM 2011,243 was removed, mounted in epoxy and polished using aluminium oxide. The polished block was carbon-coated for analysis using a Cameca SX100 electron microprobe at the NHM. The Cameca microprobe was operated in wavelength-dispersive mode at 20 kV, a probe current of 20 nA and a beam diameter of 10 µm. Only Cu, Te and Se were detected, although S, Fe, Co, Ni, Zn, Ag and U were all sought but found to be below detection limits. Table 1 summarizes the microprobe data. The anhydrous empirical formula (O = 3) is Cu_0.99(Te_0.98Se_0.02)O₃, which is consistent with the hydrous end-member formula determined in the crystallographic study, CuTeO₃·2H₂O. The ideal formula requires 28.91 wt.% CuO, 58.00 wt.% TeO₂ and 13.09 wt.% H₂O and is identical to the composition of teineite.

Table 1.

Electron microprobe analyses of millsite from holotype specimen BM 2011,243 and cotype specimen KAI-93.

* The H₂O content is that calculated on the basis of the structure determination.

S.D. – Standard deviation.

A small piece of the cotype specimen KAI-93 was embedded in a drop of adhesive on a microscope glass slide and polished with 1 µm diamond powder. Quantitative element analyses were carried out using a JEOL JXA-8500 F electron microprobe (field-emission electron gun) equipped with a high-resolution wavelength X-ray dispersive spectrometer. A teineite standard was used for CuO and TeO₂, and for SeO₂ a Se metal standard. Raw data were reduced to weight percentages using the ZAF algorithm. The analyses were done using 15 kV accelerating voltage and an electron beam current of 20 nA with a defocused probe size of 10 μm. Selenium was measured as a trace element, and the beam was defocused to avoid beam damage to the specimen and limit the charging effects because of long counting times on Se (300 s). A further reason for employing such mild conditions is that several previous electron microprobe analyses implied a structure with only one water molecule, at variance with the X-ray crystal structure, and indicating the loss of ~1 H₂O per formula unit (pfu) due to the charging and local heating of the sample under the impact of the electron beam. For the elements Cu and Te counting times of 10 s were adequate. The spectra showed weak emission lines for SeLα_1,2 and SeLβ₁. As the latter is well resolved from any Te lines we believe that the sample does contain a small amount of Se, presumably as a substitute for Te. The results from 5 analyses are given in Table 1.

There was not enough material on the holotype specimen to perform CHN analysis or directly quantify water content by any other means other than structure refinement. However, both crystal structure determinations demonstrate clearly that there are two H₂O pfu. These two non-equivalent H₂O groups form part of the octahedral coordination of Cu and are not ‘zeolitic’, but a fixed and structurally well-defined feature. Consequently, the H₂O content of millsite correlates with the Cu content, and as such can be inferred from the stoichiometry of the anhydrous empirical formula.

Raman spectroscopy

An unpolarized Raman spectrum was obtained for a crystal of millsite (30 µm × 75 µm × 115 µm) removed from the holotype specimen 2011,243. Excitation induced by the 473 nm line of a 50 mW Cobalt Blues TM laser focused down to a 10 µm spot. The spectrum is shown in Fig. 2. Peaks in the 100–1000 cm^–1 range are assigned by comparison with the Raman spectrum of teineite (Frost and Keeffe, Reference Frost and Keeffe2009). The 749 cm^–1 band and a shoulder are assigned to symmetric and antisymmetric modes of the TeO₃ group; a band at 339 cm^–1 is assigned to the TeO₃ bending mode; the 484 cm^–1 band is a possible H₂O libration mode. Two low-intensity but distinct bands occur at 1532 and 1662 cm^–1 and are attributed to the bending modes of two non-equivalent H₂O molecules (see Crystal structure). Two bands at 3288 cm^–1 and 3367 cm^–1 are due to O–H stretching modes of H₂O molecules.

Fig. 2.

Raman spectrum of millsite, showing O‒H vibrational modes in the insets.

Crystal structure

The structure of a small prismatic crystal (70 µm × 40 µm × 30 µm) from the holotype specimen BM 2011,243 was determined. This crystal was mounted on an XcaliburE four-circle diffractometer equipped with an Eos CCD detector (Rigaku Oxford Diffraction) at the Natural History Museum, London. Graphite-monochromated MoKα X-rays were used at 45 kV and 40 mA. Reflection intensities were corrected for Lorentz-polarization and absorption effects, and converted to structure factors using CrysalisPro (®Rigaku Oxford Diffraction). A whole sphere of data was collected to 60°2θ with 99.9% completeness. Structure solution by direct methods and refinement by least-squares were carried out using SHELX (Sheldrick, Reference Sheldrick2008) within the WinGX environment (Farrugia, Reference Farrugia1999). Neutral scattering factors for all atoms were taken from the International Tables for Crystallography Volume C (Wilson, Reference Wilson1992).

Systematic absences are consistent with space group P2₁/c. Structure solution using direct methods (SHELX, Sheldrick, Reference Sheldrick2008) located the Cu and Te atoms (one of each). All five O atoms of the asymmetric unit were found after 10 cycles of least-squares refinement. All four H atoms of the asymmetric unit were found in subsequent stages of weighted least-squares isotropic refinement, and their coordinates refined with their U _iso values set to 1.2 times those of the oxygen atoms to which they are bonded. A soft restraint on O–H bond lengths was used. All other atoms were then refined anisotropically. Information relating to the data collection and structure refinement is summarized in Table 2. Atom coordinates and displacement parameters are given in Table 3. Key interatomic distances, angles and polyhedral parameters are given in Table 4. Bond-valence calculations are summarized in Table 5. A crystallographic information file is provided as Supplementary material.

Table 2.

Summary of information relating to data collection and structure refinement of millsite from holotype specimen BM 2011,243.

Table 3.

Atom coordinates, anisotropic and equivalent-isotropic displacement parameters (Ų) for millsite from holotype specimen BM 2011,243.

*Value coupled to 1.2 × refined U _eq of O(4). **Value coupled to 1.2 × refined U _eq of O(5). See text for explanation.

Table 4.

Interatomic distances (Å), bond angles (°) and polyhedral parameters for millsite.

wa = oxygen of a H₂O molecule.

Table 5.

Bond valences and bond-valence sums (BVS, valence units) for millsite.

Notes: Atoms O(3) and O(4) each receive a hydrogen bond. Atom O(2) receives two hydrogen bonds. Atoms O(4) and O(5) belong to H₂O molecules. Values calculated using the parameters for the Te^IV–O bond of Mills and Christy (Reference Mills and Christy2013). Cu^II–O bond valences were calculated using the parameters of Brown and Altermatt (Reference Brown and Altermatt1985).

* This is a secondary bond in the sense of Mills and Christy (Reference Mills and Christy2013).

Structural features of millsite are shown in Figs 3–7. All atoms lie on general positions (4e). Cu forms four short, nearly coplanar, Cu–O bonds at 1.95–1.99 Å, one of which Cu–O(4) (1.99 Å) is to an H₂O molecule. In addition to these short bonds there are two much longer bonds Cu–O(5)(H₂O) at 2.387(2) Å and Cu–O(1') at 2.651(3) Å. These two long bonds lie trans with respect to the strongly-bonded planar CuO₃(H₂O) grouping. Thus, Cu²⁺ in millsite has a highly Jahn-Teller-distorted (prolate) octahedral coordination. A pair of CuO₄(H₂O)₂ octahedra share a long edge (3.38 Å) to form a centrosymmetric Cu₂O₆(H₂O)₄ dimer (Figs 3 and 4). It is conceivable that the long Cu–O(5)(H₂O) bond may be more easily broken than the stronger Cu–O(4)(H₂O) bond, leading to the loss of ~1 H₂O pfu during electron microprobe analysis mentioned above.

Fig. 3.

(a) The fundamental structural unit of millsite in which each Cu₂O₆(H₂O)₄ dimer is decorated with four TeO₃ groups which connect to four adjacent dimers to form the heteropolyhedral sheet; (b) and (c) different views of the intersheet connectivity of millsite. H(1) forms an intrasheet hydrogen bond shown here as a dashed red line. Hydrogen atoms H(2,3,4) form intersheet hydrogen bonds.

Fig. 4.

The extended millsite topology showing channels and intersheet connectivity, with a single heteropolyhedral sheet highlighted in (b). Hydrogen bonds shown as dashed red lines.

Fig. 5.

The $M_2^{{\rm 2 +}} {\rm O}_6\left( {{\rm H}_2{\rm O}} \right)_4$ dimer units in (a) millsite, (b) monoclinic CuSeO₃·2H₂O, (c) ahlfeldite, (d) teineite and (e) chalcomenite. Only millsite has a sheet topology, whereas the others are frameworks. Teineite and chalcomenite do not have dimer units. Instead they have single CuO₃(H₂O)₂ groups coordinated to three Te atoms. Water oxygens of teineite and chalcomenite are shown as black spheres. All distances are in Å.

Fig. 6.

The bonding environment and bond valences of the TeO₃ group of millsite. A fourth secondary Te−O(1)' bond contributing 0.18 vu lies on the lone-pair side of the Te atom.

Fig. 7.

The fundamental structural units of millsite and the related structures of monoclinic CuSeO₃·2H₂O and synthetic MgSeO₃·2H₂O. Water oxygens of monoclinic CuSeO₃·2H₂O are shown as light blue spheres. CuO₄(H₂O)₂ octahedra = blue, MgO₄(H₂O)₂ octahedra = yellow, TeO₃ = brown and SeO₃ = green.

The Cu₂O₆(H₂O)₄ dimer is decorated with four TeO₃ groups and it is terminated by a pair of H₂O molecules at each end (Fig. 4), which prevent polymerization in those directions. Dimers are connected to each other via TeO₃ groups and hydrogen bonding. The dimer−TeO₃ connectivity defines a hetero-polyhedral sheet (100); the only connections between these sheets perpendicular to (100) are via hydrogen bonds. This topology underlies the (100) cleavage of millsite (Fig. 4b). The unit-cell volume of millsite is ~2.5% smaller than that of teineite, implying a denser packing of the sheet structure in the former.

The coordination of Te⁴⁺ is strongly asymmetric. Te lies at one apex of a TeO₃ trigonal pyramid and forms three Te–O bonds at 1.86−1.88 Å, with three very similar bond angles at 94−99°. TeO₃ groups are not polymerized.

As part of the mineral characterization supporting the first proposal submitted to the IMA in 2001 (IMA2001-000), a structure determination was carried out on a large crystal (415 µm × 180 µm × 130 µm) taken from the cotype specimen KAI-93. A 4-circle Huber diffractometer equipped with a point detector was used with graphite-monochromated AgKα radiation (λ = 0.56089 Å). A total of 4953 intensities which included reflections to a resolution of 0.885 Å^–1 were collected, corrected for Lorentz-polarization, absorption and anomalous dispersion, and averaged to yield a total of 2253 reflections, R _int = 0.0159. Refinement of all atoms converged at R ₁(F) = 0.0236 and R _w(F ²) = 0.0528 for all 2253 F, Goof = 1.154. A full structure report has not been published. To assess the agreement between the two structure determinations we have compared the nine (Cu, Te)–O distances and thirteen O–(Cu, Te)–O angles. The maximum and average differences in distances are 0.007 Å and 0.003 Å, respectively; for the angles the maximum and average differences are 0.22^o and 0.09^o, respectively. The largest differences are comparable to 2σ. On the basis of the crystallographic results, as well as on observations of other physical properties, it is clear that the crystals from specimens BM 2011,343 and KAI-93 are both millsite.

Powder X-ray diffraction

Due to the very small quantities of millsite available from specimen BM 2011,243, it was not possible to collect powder X-ray diffraction data for this sample. However, powder diffraction was carried out successfully on a small sample from the cotype specimen KAI-93 using synchrotron radiation with λ=0.79948 Å at the Swiss-Norwegian Beamlines, ESRF. Table 6 lists hkl, d _hkl and I/I _o values for millsite from sample KAI-93. Unit-cell parameters derived from the Rietveld refinement are shown in Table 7. Rietveld refinement of the powder data confirmed the identity of the powder as millsite. For small amounts of sample, a synchrotron beam provides a spatial (2θ) resolution of the diffraction peaks (very narrow, highly symmetric peak-shapes) that is superior to what can be achieved with radiation from a standard X-ray source.

Table 6.

Powder X-ray diffraction data for millsite sample KAI-93. Reflections with I/I _max <5% are omitted.

Table 7.

Unit-cell parameters of millsite and related phases.

(1) Effenberger (Reference Effenberger1977); (2) Pasero and Perchiazzi (Reference Pasero and Perchiazzi1989); (3) Robinson et al. (Reference Robinson, Sen Gupta, Swihart and Houk1992); (4) This study, holotype specimen BM 2011,243; (5) KAI-93, Mo et al. (2000); (6) This study, powder-diffraction of cotype specimen KAI-93; (7) Luo et al. (Reference Luo, Chen and Ma1984); (8) Larranaga et al. (Reference Larranaga, Mesa, Pizarro, Pena, Chapman, Arriortua and Rojo2005); (9) Johnston and Harrison (2001). *Transformed from the reported P2₁/n cell by ($\bar 1$00/0$\bar 1$0/101).