Columbite supergroup of minerals: nomenclature and classification

Abstract

The columbite supergroup is established. It includes five mineral groups (ixiolite, wolframite, samarskite, columbite and wodginite) and one ungrouped species (lithiotantite). The criteria for a mineral to belong to the columbite supergroup are: the general stoichiometry MO₂; the crystal structure based on the hexagonal close packing (hcp) of anions (or close to it); the six-fold coordination number of M-type cations (augmented to eight-fold in the case of slight distortion of hcp); and the presence of zig-zag chains of edge-sharing M-centred polyhedra. The ixiolite-type structure is considered as an aristotype with the space group Pbcn, the smallest unit cell volume, and the basic vectors a₀, b₀ and c₀. Based on the multiplying of the ixiolite-type unit cell the following derivatives are distinguished: ixiolite type [ixiolite-group minerals; a = a_0, b = b₀ and c = c₀; space group Pbcn; the members are ixiolite-(Mn²⁺), ixiolite-(Fe²⁺), scrutinyite, seifertite and srilankite]; wolframite type [wolframite-group minerals, ordered analogues of the ixiolite type with a = a_0, b = b₀ and c = c₀; P2/c; the members are ferberite, hübnerite, huanzalaite, sanmartinite, heftetjernite, nioboheftetjernite, rossovskyite and riesite]; samarskite type [samarskite-group minerals; a = 2a₀, b = b₀ and c = c₀; P2/c; the members are samarskite-(Y), ekebergite and shakhdaraite-(Y)]; columbite type [columbite-group minerals; a = 3a_0, b = b₀ and c = c₀; Pbcn; the members are columbite-(Fe), columbite-(Mn), columbite-(Mg), tantalite-(Fe), tantalite-(Mn), tantalite-(Mg), fersmite, euxenite-(Y), tanteuxenite-(Y) and uranopolycrase]; and wodginite type [wodginite-group minerals; a = 2a_0, b = 2b₀ and c = c₀; C2/c; the members are wodginite, ferrowodginite, titanowodginite, ferrotitanowodginite, tantalowodginite, lithiowodginite and achalaite]. Samarskite-(Yb), ishikawaite and calciosamarskite are insufficiently studied, tentatively considered as possible members of the samarskite supergroup. Qitianlingite, yttrocolumbite-(Y), yttrotantalite-(Y) and yttrocrasite-(Y) are questionable and need further studies. Polycrase-(Y) is discredited as identical to euxenite-(Y). Ixiolite has been renamed as ixiolite-(Mn²⁺), with the end-member formula (Ta_2/3Mn²⁺_1/3)O₂. Ta- and Nb-dominant analogues of ixiolite with different schemes of charge balancing have the end-member formulae (M1⁵⁺_0.5M2³⁺_0.5)O₂, M1⁵⁺_2/3M2²⁺_1/3)O₂, M1⁵⁺_0.75M2⁺_0.25)O₂ or M1⁵⁺_0.8□_0.2)O₂ and the root name ‘ixiolite’ (for M1 = Ta) or ‘nioboixiolite’ (for M1 = Nb).

Introduction

Among Ti⁴⁺-, Sn⁴⁺-, ^VIGe⁴⁺-, ^VISi-, ^VIMn⁴⁺-, ^VIPb⁴⁺-, ^VITe⁴⁺-, Nb-, Ta-, Sb⁵⁺-, Mo⁶⁺- and W⁶⁺-oxide minerals with the stoichiometry MO₂, there are numerous mineral species related structurally to columbite. Although they display substantial common features, these minerals differ from each other in many aspects, including different kinds of cation ordering, symmetry, unit-cell dimensions, and coordination numbers of cations. Attempts to elaborate a general crystal-chemical classification of columbite-type minerals and other related mineral species with the stoichiometry MO₂ have been undertaken repeatedly (Graham and Thornber, 1974a; Sugitani et al., 1985; Hanson et al., 1999). This paper summarises available data on minerals with the stoichiometry MO₂ that are topologically related to columbite and constitute the columbite supergroup. The nomenclature and classification of the columbite supergroup has been approved by the Commission on New Minerals, Nomenclature and Classification of the International Mineralogical Association (IMA–CNMNC).

The root-name columbite is the oldest one among all of the names of mineral species that are discussed in this nomenclature report. Minerals belonging to the columbite group are important from petrological, geochemical and practical points of view.

The name is after the chemical composition: the mineral columbite was described originally as an iron and columbium oxide. Columbium is an old and today obsolete name for the chemical element that was later re-named niobium. The mineral, however, retained its name. The root columbium is also maintained in ‘coltan’, an acronym which refers to the niobium/tantalum oxides.

General definitions

The following criteria are applied to define the minerals of the columbite supergroup:

1. (1) The general stoichiometry MO₂ is required.

2. (2) The crystal structure is based on the hexagonal close packing (hcp) of anions (or close to it).

3. (3) Only octahedral voids of hcp are occupied. As a result, the coordination number of M-type cations is 6 (sometimes augmented to 8 in the case of a slight distortion of hcp).

4. (4) The presence of zig-zag chains of edge-shared octahedra (the idealised symmetry described by the rod group þ2/c11; Fig. 1).

Fig. 1. The zig-zag and straight chains of edge-shared MO₆ octahedra and their rod groups.

The application of these criteria obviously excludes compounds with rutile-related structures (e.g. tapiolite-group minerals) which are characterised by straight chains of edge-sharing octahedra with the idealised symmetry described by the rod group þ112/m (Fig. 1). Alumotantite (Ercit et al., 1992d) only matches criteria I-III and is not considered a member of the columbite supergroup. A short outline of minerals with the general MO₂ stoichiometry which do not belong to the columbite supergroup (and as such were not part of the IMA-approved report) has been deposited with the Principal Editor of Mineralogical Magazine and is available as Supplementary material (see below).

Using the approach applied for the perovskite supergroup (Mitchell et al., 2017), the ixiolite-type structure is considered as an aristotype with the space group Pbcn, the smallest unit cell volume and the basic vectors a₀, b₀ and c₀. The following derivatives can be distinguished on the basis of the multiplying of the initial ixiolite-type unit cell (Fig. 2a): ixiolite type with a = a_0, b = b₀ and c = c₀; space group Pbcn; wolframite type (an ordered analogue of the ixiolite type) with a = a_0, b = b₀ and c = c₀; P2/c; samarskite type with a = 2a_0, b = b₀ and c = c₀; P2/c; columbite type with a = 3a_0, b = b₀ and c = c₀; Pbcn; and wodginite type with a = 2a_0, b = 2b₀ and c = c₀; C2/c.

Fig. 2. General comparison of the unit cells (a), and symmetry reduction from the initial aristotype with the ixiolite-type unit cell and the space group Pbcn induced by the different kinds of ordering of cations (b).

Different schemes of ordering of M cations control both the symmetry lowering and multiplying of the basic ixiolite-type unit cell. The Bärnighausen tree (Müller, 2004) shown in Fig. 2b illustrates the symmetry relations between different structures.

Minerals belonging to the columbite supergroup

Ixiolite group

Minerals belonging to the ixiolite group with the general formula M1O₂ (orthorhombic, Pbcn, a = a_0, b = b₀, c = c₀ and Z = 4) are characterised by a disordered distribution of the cations: in the crystal structure of ixiolite-group minerals (Fig. 3), all cations occupy a single M1 site. In these minerals, edge-sharing M1O₆ octahedra form chains along the c direction. In the a direction, the chains are connected with each other via common vertices of the octahedra.

Fig. 3. The crystal structure of ixiolite-group minerals. The unit cell is outlined.

Ixiolite was first described by Nordenskiöld (1857) as a tantalum oxide, with subordinate Fe and Mn and minor Sn. The sample originated from Skogsböle, Kimito Island, Finland. The chemical analysis of the sample from Skogsböle is incomplete and corresponds to the approximate formula Ta_0.6(Fe,Mn)_0.3Sn_0.1O₂. The Fe:Mn ratio was not determined. On the basis of goniometric measurements, the mineral was assumed to be orthorhombic with a:b:c = 1:0.5508:1.2460. D _meas = 7.0–7.1; H(Mohs) = 6–6½.

In another ixiolite sample from Skogsböle, the Mn:Fe ratio is 1.04:1 in atomic units (Rose, 1858). Mn-rich ixiolite (with 9.35 wt.% MnO) has been also discovered in pegmatites of the Kalbinskiy range, Russia (Chukhrov and Bonshtedt-Kupletskaya, 1967). The crystal structure of Mn-rich ixiolite with the charge-balanced empirical formula (Ta_0.43Nb_0.24)Mn²⁺_0.23Mn³⁺_0.07(Ti_0.02Sn_0.01)O₂ from the Tanco pegmatite, Bernic Lake, Manitoba, Canada was solved by Grice et al. (1976).

The chemical formula of ixiolite is currently given as (Ta,Mn,Nb)O₂ which corresponds to an ixiolite-group mineral with Mn as the main charge-balancing component, but samples with Fe > Mn are also known. In most analyses of ixiolite from Skogsböle, Fe prevails over Mn, with Fe:Mn up to 13.8:1 (Rose, 1858). Nickel et al. (1963a) investigated the crystal structure of an ixiolite sample from Skogsböle with the charge-balanced empirical formula (Ta_0.43Nb_0.12)(Fe²⁺_0.13Mn²⁺_0.12)Fe³⁺_0.05(Sn_0.13Ti_0.01Zr_0.01)O₂. The sample is deposited in the Royal Ontario Museum with the catalogue number M-6591. A synthetic compound with the formula NbFe³⁺O₄ and ixiolite-type structure has been described by Harrison and Cheetham (1989).

Scrutinyite, α-PbO₂ was discovered in two natural occurrences situated in Bingham, New Mexico, USA and Mapimi, Mexico (Taggart et al., 1988). The crystal structure of synthetic α-PbO₂ was solved by Zaslavskij and Tolkachev (1952).

Seifertite, SiO₂, is an orthorhombic high-pressure silica polymorph with the ixiolite-type structure. The mineral is a constituent of high-pressure assemblages typical of shock-affected Martian meteorites belonging to the shergottite group (Dera et al., 2002; El Goresy et al., 2008; Zhang et al., 2016).

Srilankite, TiO₂, was described as a new mineral from Rakwana, Sabaragamuva province, Sri Lanka (Willgallis et al., 1983). The chemical composition was given originally as (Ti,Zr)O₂, with Zr:Ti = 1:2. The ixiolite-type structure of srilankite has been confirmed by a single-crystal X-ray diffraction (XRD) study of a natural sample (Willgallis and Hartl, 1983) and its synthetic analogue (Troitzsch et al., 2005). Similarly to transition metals in other ixiolite-group minerals, Ti and Zr in srilankite occupy the same crystallographic M1 site. Zirconium, having an ionic radius larger than titanium, plays an essential role in stabilising the ixiolite-type structure of srilankite at ambient pressure. Zirconium-free srilankite, pure TiO₂, was described as a quenched ‘TiO₂-II’ polymorph from the Ries impact structure (El Goresy et al., 2001), the Xiuyan crater in China (Zhang et al., 2009) and in the high-pressure mineral assemblages of subduction zones (Chen et al., 2013).

The Nb-dominant analogue of ixiolite (with Nb > Ta) has been known for a long time (von Knorring and Sahama, 1969; Wise et al., 1998; Zubkova et al., 2020). This mineral was described as the new mineral species ‘ashanite’ with the formula (Nb, Ta,U, Fe, Mn)₄O₈ (Z = 1) (Zhan et al., 1980). However, in 1998, ‘ashanite’ was discredited by the IMA–CNMNC. This decision was made based on unsatisfactory compositional data for this mineral, suggestive of a mixture of ixiolite, samarskite and uranmicrolite (Shen, 1998).

Although there is only one cationic M1 site in the ixiolite-type structure, a charge-balanced end-member formulae of ixiolite and its Nb-dominant analogue cannot be written with a single cationic component. Thus, the dominant-charge-compensating cations (either a lower-valency cation or vacancy) should be taken into account, as discussed by Hatert and Burke (2008).

Wolframite group

The wolframite-type structure (M1M2O₄, monoclinic, P2/c, a = a₀, b = b₀, c = c₀, β ≈ 91° and Z = 2) is a derivative of the ixiolite-type structure characterised by the ordering of the cations with lowering of the symmetry. It can be represented as a sequence of two kinds of structurally identical, but chemically different, octahedral layers of parallel zig-zag chains alternating along the a axis of the ixiolite quasi-framework (Fig. 4). The larger-radius cations occupy the octahedral M1 site, whereas the smaller-radius cations reside at the M2 octahedron. Consequently, members of the wolframite group are double oxides with the general formula M1²⁺M2⁶⁺O₄ (M1 = Mg, Mn, Fe and Zn; M2 = W) or M1³⁺M2⁵⁺O₄ (M1 = Sc and Fe; M2 = Nb and Ta). The Ti⁴⁺Ti⁴⁺O₄ oxide, riesite, represents a slightly distorted variant of the wolframite structure. The layered ordering of different-sized cations in wolframites results in monoclinic distortion of the ixiolite framework, whereas the unit-cell dimensions of the parent ixiolite remain unchanged (Fig. 5).

Fig. 4. General view of the wolframite-type structure.

Fig. 5. The scheme of splitting of atomic sites (the upper row) and their coordinates in the ixiolite- and wolframite-type structures in accordance with the relations between the mineral groups (see Fig. 2b). One cationic M1 site and one oxygen O1 site in the ixiolite-type structure split into two symmetrically non-equivalent M1 and M2 as well as O1 and O2 sites in the wolframite-type structure due to the cation ordering and reducing of the symmetry from the space group Pbcn to P2/c.

The wolframite group inherits its name from ‘wolframite’, which is now considered to be an obsolete mineral species. The first scientific description of this mineral with the name ‘Wolfram’ (‘wolf-cream’, from German Wolfram or Wolfrahm) was made by Henckel (1725).

Historically, wolframites represent intermediate members of the solid solution between pure Fe²⁺WO₄ and pure Mn²⁺WO₄. In particular, the term wolframite indicated minerals with compositions ranging between (Fe_0.8Mn_0.2)WO₄ and (Fe_0.2Mn_0.8)WO₄. The species having Fe > 0.8 and Mn > 0.8 atoms per formula unit (apfu) were called ferberite and hübnerite, respectively. Subsequently, compositional fields of ferberite and hübnerite have been expanded according to the 50% rule and the term ‘wolframite’ has been abandoned. For historical reasons, however, it seems convenient to keep wolframite as the name for the group of ordered structures with an ixiolite-type unit cell, but with the space group P2/c.

Ferberite was first described by Liebe (1863). The type locality is the Niña mine, Sierra Almagrera, Andalusia, Spain. The crystal structure of ferberite has been refined by Cid-Dresdner and Escobar (1968).

Hübnerite was first described by Credner (1865). The type locality is the Ellsworth mine, Nevada, USA. The crystal structure of ferberite has been refined by Dachs et al. (1967).

Huanzalaite is the Mg-dominant analogue of ferberite and hübnerite. It was first described by Miyawaki et al. (2010). The type locality is the Huanzala mine, Ancash Department, Peru. The crystal structure of its synthetic analogue has been refined by Macavei and Schulz (1993).

Sanmartinite, ideally ZnWO₄, was first described by Angelelli and Gordon (1948). The type locality is the Department of San Martín, San Luis province, Argentina. The crystal structure of sanmartinite has been refined by Redfern et al. (1995).

Heftetjernite, ScTaO₄, was first described by Kolitsch et al. (2010), who also refined its crystal structure. The type locality is the Heftetjern pegmatite, Tørdal, Telemark, Norway.

Nioboheftetjernite, ScNbO₄, was first described by Lykova et al. (2021), who also refined its crystal structure. The type locality is the Befanamo pegmatite, Madagascar.

Rossovskyite was first described by Konovalenko et al. (2015), who also refined its crystal structure. The type locality is Bulgut, Altai Mountains, Mongolia. The chemical formula of the mineral is given as (Fe³⁺,Ta)(Nb,Ti)O₄. According to the dominant-valency rule and the site-total-charge approach (Bosi et al., 2019), the end-member formula is Fe³⁺NbO₄.

Riesite was reported as a new TiO₂ polymorph from impact-affected rocks (suevites) at the Ries impact crater, Germany (Tschauner et al., 2020). Similarly to formerly described Zr-free srilankite, riesite was formed by shock-induced transformation of rutile at pressures of 20–25 GPa. In the crystal structure of riesite, the M1 and M2 sites are insignificantly displaced from the general positions of the wolframite-type framework, becoming statistically half-occupied. By analogy with other wolframite-group minerals, the ideal formula of riesite can be written as TiTiO₄.

Samarskite group

The samarskite group includes three valid species, namely, samarskite-(Y), ekebergite and shakhdaraite-(Y). These minerals are monoclinic (space group P2/c, a = 2a₀, b = b₀, c = c₀, β ≈ 93° and Z = 2), cation-ordered double niobates and tantalates with the general formula AM1M2₂O₈ (A = Y and Th; M1 = Fe²⁺, Fe³⁺ and Sc³⁺; M2 = Nb and Ta) and unit-cell parameters a = 9.8–9.9, b = 5.6–5.7, c ≈ 5.2 Å, and β = 92–94° (Z = 2). Unlike other columbite-supergroup minerals, members of the samarskite group contain a relatively large cation at the A site with 6 + 2-fold coordination (Fig. 6) due to the slight distortion of the hcp (Lima-de-Faria, 2012). Such insertion of a large cation transforms parallel zig-zag chains into a rigid layer of edge-sharing AO₈ polyhedra with the preservation of the cation distribution between the ‘octahedral’ voids of hcp (Fig. 7). There are also three insufficiently studied metamict minerals, namely, samarskite-(Yb), ishikawaite and calciosamarskite, that are tentatively assigned to the samarskite group based on their stoichiometry and the powder XRD patterns of annealed samples.

Fig. 6. General view of the samarskite-type structures.

Fig. 7. Transformation of parallel zig-zag chains of edge-sharing octahedra into a solid layer of edge-shared eight-vertex polyhedra with the increasing of the ionic radii of the cation in the samarskite-type structures.

The name samarskite was introduced into the mineralogical literature by Rose (1847) who described a sample from Ilmen Mountains, Chelyabinsk region, Russia. Subsequently, the mineral name was changed to samarskite-(Y) according to general nomenclature rules for the REE-bearing minerals (Levinson, 1966). According to Hanson et al. (1999), the name samarskite-(Y) is attributed to the samarskite-group mineral in which the A site is dominated by REE cations, among which Y³⁺ prevails.

Samarskite-(Y) is the first member of the samarskite group whose crystal structure was published. A recent finding of non-metamict samarskite-(Y) allowed the refinement of its crystal structure, and the re-definition of the mineral as YFe³⁺Nb₂O₈ (Britvin et al., 2019). These authors confirmed that this new chemical formula, with Fe³⁺ as a species-forming constituent, corresponds to the formula of holotype samarskite-(Y).

Ekebergite, ideally ThFe²⁺Nb₂O₈, was approved as a new mineral species in 2018 (Kjellman et al., 2018). This mineral originates from the pumice quarry ‘In den Dellen’ (Bimsgrube Zieglowski), Mendig, Laacher See (Laach Lake) complex, Eifel, Rhineland-Palatinate, Germany. Ekebergite is isostructural with samarskite and forms a solid-solution series with samarskite. The full description of the mineral has not as yet been published.

Shakhdaraite-(Y), YScNb₂O₈, was described as a new mineral from Tajikistan (Pautov et al., 2022). It is the Sc-dominant analogue of samarskite-(Y).

Samarskite-(Yb), YbFe³⁺Nb₂O₈, was described as a new mineral by Simmons et al. (2006). It occurs as a metamict mineral at the Little Patsy pegmatite, South Platte district, Jefferson Co., Colorado, USA. The mineral recrystallised after heating at 1100°C for 12 h.

Ishikawaite was first described as an unnamed mineral from Ishikawa, Iwaki province, Japan, by Shimata and Kimura (1922a) and then named ishikawaite after the type locality (Shimata and Kimura, 1922b). Its chemical formula is currently given as (U,Fe,Y)NbO₄. According to Hanson et al. (1999), the name ishikawaite should be attributed to the samarskite-group mineral in which the A site is dominated by U⁴⁺. Under this assumption, ishikawaite should be considered as the analogue of ekebergite with U⁴⁺ > Th and the end-member formula U⁴⁺Fe²⁺Nb₂O₈.

Calciosamarskite was first described by Ellsworth (1928a, 1928b) as the Ca-dominant analogue of samarskite. Its chemical formula is currently given as (Ca,Fe,Y)(Nb,Ta,Ti)O₄. The mineral was supposed to be discredited (see Hanson et al., 1999), but actually it is still considered a valid, grandfathered species. According to Hanson et al. (1999), the name calciosamarskite should be attributed to the samarskite-group mineral in which the A site is dominated by Ca. However, the end-member formula CaFe³⁺Nb₂O₈, which would be expected for a Ca-dominant samarskite-group mineral, is not charge-balanced even with trivalent iron. The formula CaFe³⁺Nb₂O₇(OH) is neutral, but the presence of OH groups in calciosamarskite is questionable. Probably, this problem could be solved based on data for the synthetic analogue.

Columbite group

The columbite group includes double oxides with the general formula M1²⁺M2⁵⁺₂O₆ (orthorhombic, Pbcn, a = 3a₀, b = b₀, c = c₀ and Z = 4; M1 = Mg, Ca, Mn and Fe; M2 = Nb and Ta). In the crystal structure of these minerals (Fig. 8), M1O₆ octahedra share edges to form infinite zig-zag chains along the c axis. Similar chains are formed by the M2O₆ octahedra. Thus, alternating [100] ‘layers’ are formed: a single ‘layer’ consisting of chains of M1O₆ octahedra and double ‘layers’ comprising chains of M2O₆ octahedra. The chains of the neighbouring layers are linked via common vertices.

Fig. 8. The general view of the columbite-type structure. The unit cell is outlined.

Columbite-(Fe), Fe²⁺Nb₂O₆, is the current name of the mineral originally described as ‘columbite’ and later named ferrocolumbite. Columbite was first described by Jameson (1805). The type locality is likely to be either Haddam or Middletown, both in Connecticut, USA (cf. Dana, 1892). The mineral was renamed to columbite-(Fe) after Burke (2008). The crystal structure of natural columbite-(Fe) from S. José de Safira, Minas Gerais, Brazil has been refined by Tarantino and Zema (2005).

Columbite-(Mn), Mn²⁺Nb₂O₆, was first described by Dana (1892) under the name manganocolumbite. This mineral was considered initially to be a Mn-dominant variety of columbite. The mineral was renamed to columbite-(Mn) after Burke (2008). The crystal structure of natural columbite-(Mn) from Kragero, Norway has been refined by Tarantino and Zema (2005).

Columbite-(Mg), MgNb₂O₆, the Mg-dominant member of the columbite solid-solution series, was first found in the Muzeinaya vein, Gorno-Badakhshan, Tajikistan (Mathias et al., 1963). The mineral was originally named magnocolumbite and then renamed to columbite-(Mg) after Burke (2008). The crystal structure of synthetic MgNb₂O₆ has been refined by Pagola et al. (1997).

Tantalite-(Fe), Fe²⁺Ta₂O₆, is the current name of the mineral originally described as ‘tantalite’ and then named ‘ferrotantalite’. Tantalite was first described by Thomson (1836). The type locality is Upper Bear Gulch, Tinton pegmatite district, Lawrence Co., South Dakota, USA. The mineral was renamed to tantalite-(Fe) after Burke (2008). An overwhelming majority of analysed tantalite-(Fe) samples contain significant amounts of Mn and/or Nb. Samples with compositions close to the Fe²⁺Ta₂O₆ end-member have the tapiolite structure (Ercit et al., 1995).

Tantalite-(Mn), Mn²⁺Ta₂O₆, was first described as ‘manganotantalite’, a Mn-dominant variety of tantalite by Nordenskiöld (1877). The type locality is the Utö Mines, Stockholm Co., Sweden. The mineral was renamed to tantalite-(Mn) after Burke (2008). The crystal structure of natural tantalite-(Mn) from the Tanco pegmatite, Manitoba, Canada has been refined by Grice et al. (1976).

Tantalite-(Mg), MgTa₂O₆, was described as a new mineral ‘magnesiotantalite’ from Lipovka, Central Urals, Russia by Pekov et al. (2003). The mineral was renamed to tantalite-(Mg) after Burke (2008).

Similarly to the samarskite-type structures (Fig. 7), the insertion of cations with large ionic radii into the columbite-type structure causes the parallel zig-zag chains to transform into a rigid layer. Such layers of edge-shared AO₈-polyhedra (A = Ca and Y) have been found in the euxenite derivative of the columbite-type structure, where they alternate with double ‘layers’ containing zig-zag chains of M2O₆ octahedra (Fig. 9). Despite the distortion of the initial hcp, the distribution of the cations over the ‘octahedral’ void in the euxenite derivative are exactly equal to those in the columbite-type structure (Lima-de-Faria, 2012).

Fig. 9. General view of the euxenite derivative of the columbite-type structure, containing layers of edge-shared eight-vertex polyhedra.

Fersmite, CaNb₂O₆, was discovered in the pegmatites of the Vishnevye Mountains, Central Urals (Bohnstedt-Kupletskaya and Burova, 1946). The crystal structure of fersmite was solved by Aleksandrov (1960). The presumed synthetic analogue of fersmite is orthorhombic in space group Pcan, with a = 5.75, b = 14.03 and c = 5.20 Å and Z = 4 (Cummings and Simonsen, 1970). Unlike other tantalite-group minerals, fersmite contains a rather large Ca cation having 8-fold coordination. Fersmite is dimorphous with the aeschynite-group mineral vigezzite.

Based on the stoichiometry, powder XRD patterns of annealed samples, and crystal structures of presumed synthetic analogues, four minerals whose natural samples are usually metamict [namely, euxenite-(Y), polycrase-(Y), tanteuxenite-(Y) and uranopolycrase] can be assigned tentatively to the columbite group (Palache et al., 1944; Weitzel and Schröcke, 1980; Aurisicchio et al., 1993).

Euxenite-(Y) is orthorhombic, with the end-member formula YNbTiO₆ and unit-cell parameters a ≈ 14.6, b ≈ 5.55 and c ≈ 5.2 Å. For example, the empirical formula of euxenite-(Y) from Lyndoch Township, Ontario, Canada (Ellsworth, 1927) calculated on 2(Nb + Ta + Ti + Fe³⁺+Al) apfu is [(Ca_0.31Fe²⁺_0.04Mn_0.02Pb_0.01)_Σ0.38(Y_0.58Ce_0.10)_Σ0.68(Th_0.07U_0.01)_Σ0.08][(Fe³⁺_0.06Al_0.01)_Σ0.07Ti_0.74(Nb_1.13Ta_0.06)_Σ1.19]O_6.34. Numerous chemical data of euxenite-(Y) are given in the reference book Minerals (Chukhrov and Bonshtedt-Kupletskaya, 1967). All of them correspond to the end-member formula YNbTiO₆. The unit-cell parameters of a metamict euxenite-(Y) sample with the empirical formula (REE _0.92Ca_0.08U_0.11Th_0.06Mn_0.01)_Σ1.18(Nb_0.84Ta_0.09Ti_0.84Fe_0.12)_Σ1.89O₆ from a rare-metal pegmatite, which was annealed at 900°C, are a ≈ 14.68, b ≈ 5.56 and c ≈ 5.18 Å (Sokolova, 1959). The unit-cell parameters of synthetic YNbTiO₆ (Weitzel and Schröcke, 1980) are a = 14.64, b = 5.55 and c = 5.20 Å.

‘Polycrase-(Y)’, which was considered an analogue of euxenite-(Y) with Ti > Nb in atomic units (Johnsen et al., 1999), is rarer. The empirical formula of metamict polycrase-(Y) from Birkenes, Norway is (Y_0.47Ln_0.20Ca_0.19U_0.18Th_0.06)_Σ1.10(Ti_1.19Nb_0.71Ta_0.07)_Σ1.97O₆ (Tomašić et al., 2004).

Non-metamict polycrase-(Y) with the unit-cell parameters a =14.82, b = 5.66 and c = 5.22 Å was described by Guastoni et al. (2019). It occurs in the Fiume pegmatite dyke, Vigezzo Valley, Central Alps, Italy. Its simplified empirical formula (analysis 9/1 in the cited paper) is (Ca,Mn,Fe²⁺)_0.085REE _0.78(U,Th)_0.19Ti_1.14Si_0.01(Nb,Ta)_0.78W_0.01.

Another sample described by Guastoni et al. (2019) originates from the Bosco dyke situated in the same region. It is an intermediate member of the euxenite-(Y)–polycrase-(Y) solid-solution series and has the simplified formula (Ca,Mn,Fe²⁺)_0.165REE _0.84(U,Th)_0.10Ti_0.96Si_0.01(Nb,Ta)_0.96W_0.01. This sample is also non-metamict and has the unit-cell parameters a = 14.736, b = 5.605 and c = 5.184 Å. All available analyses of polycrase-(Y) correspond to the end-member formula Y(NbTi)O₆.

Thus, euxenite-(Y) and polycrase-(Y) (including those of annealed samples) are minerals with identical unit-cell parameters and the common end-member formula Y(NbTi)O₆. Consequently, these minerals should be considered as the same mineral species. The name euxenite-(Y), as the older of the two, has priority.

Tanteuxenite-(Y), YTaTiO₆, is a rare mineral first described from Western Australia (Simpson, 1928) and reported from a few other localities. The mineral is usually metamict.

Uranopolycrase, ideally UTi₂O₆, was described as a new mineral from Elba Island, Italy. Because the mineral is metamict, its crystal structure has been refined on a sample annealed at 900°C for 10 h (Aurisicchio et al., 1993).

Wodginite group

The wodginite group includes monoclinic minerals (space group C2/c; a = 2a₀, b = 2b₀, c = c₀, β ≈ 91° and Z = 4) with the general formula M1M2M3₂O₈. The dominant cations at the M sites are: M1 = Mn²⁺, Fe²⁺ and Li; M2 = Ti, Sn⁴⁺ and Ta; M3 = Ta. The structure of these minerals (Ercit et al., 1992a) is based on alternating (100) ‘layers’ consisting of chains of edge-sharing MO₆ octahedra running along the c axis (Fig. 10). The ‘layers’ of the first type contain chains of M3O₆ octahedra, whereas the ‘layers’ of the second type contain chains of alternating M1O₆ and M2O₆ octahedra (Fig. 11). The chains of the neighbouring layers are linked via common vertices. The structures of wodginite-group minerals are characterised by a different degree of ordering of cations among the M sites; the heating of samples at 1000°C for 16 hours induces a full order of cations in wodginite-group minerals (Ercit et al., 1992a, 1992b, 1992c).

Fig. 10. General view of the wodginite-type structure.

Fig. 11. Two types of layers containing zig-zag chains in the wodginite-type structure.

Wodginite, ideally MnSnTa₂O₈, was described as a new mineral from two localities, Wodgina, Western Australia and Bernic Lake, Manitoba, Canada (Nickel et al., 1963b). On the basis of powder XRD data, its crystal structure was recognised as a superstructure of ixiolite. The crystal-structure refinements have been carried out by Ercit et al. (1992a), who have shown that different samples have different degrees of Ta disorder. Partially ordered samples are structurally intermediate between wodginite and ixiolite. The crystal structure of wodginite from Wodgina was investigated by Graham and Thornber (1974b). Later, the crystal structure of wodginite from Bernic Lake was solved by Ferguson et al. (1976).

Ferrowodginite, FeSnTa₂O₈, was characterised as a new mineral species by Ercit et al. (1992c). In the type specimen, ferrowodginite occurs as 0.01 to 0.2 mm inclusions in cassiterite from a granitic pegmatite near Sukula, southwestern Finland.

Titanowodginite, MnTiTa₂O₈, holotype material occurs as euhedral crystals up to 1 cm across at the Tanco pegmatite, Bernic Lake, Manitoba, Canada. Its crystal structure was solved by Ercit et al. (1992c).

Ferrotitanowodginite, FeTiTa₂O₈, has been described from the San Elías pegmatite, Sierra de la Estanzuela, San Luis Province, Argentina (Galliski et al., 1999).

Tantalowodginite, (Mn_0.5□_0.5)TaTa₂O₈, was found in the Emmons granite pegmatite dyke in Oxford County, Maine, USA (Hanson et al., 2018).

Lithiowodginite, LiTa₃O₈ or LiTaTa₂O₈, was discovered at the Ognevka and Yubileinoye tantalum deposits, Kalba Mountains, eastern Kazakhstan (Voloshin et al., 1990).

Achalaite, Fe²⁺TiNb₂O₈, is the first niobium-dominant member of the wodginite group and was described from the La Calandria granite pegmatite, Cañada del Puerto, Córdoba province, Argentina (Galliski et al., 2016).

Ungrouped columbite-supergroup mineral

Lithiotantite, LiTa₃O₈, with space group P2₁/c, a = 7.44 b = 5.04 c = 15.25 Å, β = 107.2° and Z = 4, is chemically and topologically identical to lithiowodginite (Fig. 12) (Voloshin et al., 1990; Ercit et al., 1992a, 1992c).

Fig. 12. The crystal structure of lithiotantite (left) and the proposed structure of qitianlingite (right).

Insufficiently studied minerals

The minerals listed below are not currently included in the columbite supergroup, pending reliable data on their chemical composition and crystal structure.

Qitianlingite is a mineral related to the members of the columbite and tantalite solid-solution series. It was described as a new mineral species with the ideal formula Fe²⁺₂Nb₂W⁶⁺O₁₀ (Yang et al., 1985). Qitianlingite was named after the type locality (Qitianling granite, Hunan Province, China). The crystal structure of qitianlingite has been refined by Peng et al. (1988), who described it as a superstructure of ixiolite with ordered cation distribution and a unit cell with the a axis approximately 5 times larger than the a axis of ixiolite (Fig. 12). However, calculated powder diffraction data confirming the superstructure of qitianlingite are not given in these papers. Indexing of all assumed superstructure reflections in the measured powder data is not in accordance with the pattern calculated from the proposed structure; all observed reflections can be indexed using an ixiolite-type cell. The holotype material of this mineral needs additional investigation.

Yttrocolumbite-(Y), (Y,U,Fe²⁺)(Nb,Ta)(O,OH)₄, is a questionable mineral described by Lepierre (1937). This mineral has been considered to be the Nb-dominant (with Nb > Ta) analogue of yttrotantalite-(Y). Natural yttrocolumbite-(Y) is metamict. The idealised formula of yttrocolumbite-(Y) coincides with those of fergusonite-(Y) and fergusonite-β-(Y).

Yttrotantalite-(Y) was described as a new mineral from Sweden (Ekeberg, 1802). Its chemical formula is currently given as (Y,U,Fe²⁺)(Ta,Nb)(O,OH)₄. Actually, its ideal chemical formula should be reduced to YTaO₄. Natural yttrotantalite-(Y) is metamict. It is considered to be a polymorph of formanite-(Y). Crystal structure refinements of yttrotantalite-(Y) have been carried out on presumed synthetic analogues; Keller (1962) described it with a samarskite-like unit cell, whereas Wolten (1967) described it with a wolframite-like unit cell.

Yttrocrasite-(Y) is an ill-defined mineral described as an yttrium–thorium–uranium titanate from Burnet County, Texas, USA (Hidden and Warren, 1906). Its chemical formula is currently given as (Y,Th,Ca,U)(Ti,Fe)₂(O,OH)₆.

A mineral with the empirical formula [REE _0.52(U,Th)_0.25(Fe²⁺,Mn,Ca)_0.20]_Σ0.97[(Nb,Ta)_1.26Fe³⁺_0.43(Ti,Zr,Sn,Hf)_0.28W_0.03]O₆ and with Y as the predominant REE was described by Nakajima and Kurosawa (2006) as ‘euxenite’. If this sample is isostructural with euxenite, its end-member formula should be Y(Nb_1.5Fe³⁺_0.5)O₆. Unfortunately, no X-ray diffraction data have been provided for this mineral.

Ginzburg et al. (1969) described a so-called ‘wolframoixiolite’ from an unknown locality. The empirical formula of this sample is (Nb_0.54W_0.46Fe_0.40Mn_0.30Ta_0.10Zr_0.06U_0.05Ca_0.03Mg_0.01Ti_0.01)_Σ1.96O₄⋅0.84H₂O. The powder diffraction data were indexed with a monoclinic cell P2/c, a = 4.750, b = 5.72, c = 5.06 Å and β = 90°. A monoclinic cell was required because not all lines could be indexed with the ixiolite cell. Wang et al. (1988) described a homogeneous material with the composition (Nb_0.70Fe_0.50W_0.38Mn_0.23Ta_0.12Ti_0.03Sn_0.01)_Σ1.97O_4.00, monoclinic, space group Pc, with a = 4.674, b = 3.673, c = 5.050 Å and β = 90°. Borneman-Starynkevitch et al. (1974), during a reinvestigation of the type material by electron microprobe analysis, found a Nb–Ta–Mn mineral without W as the main phase. The authors discuss whether wolframoixiolite is really a homogenous mineral or a mixture of ferberite with columbite. Eventually Nickel and Mandarino (1987) listed wolframoixiolite as a discredited mineral. Taking into account the relationships Nb > W, Fe + Mn > W, and Fe > Mn and under the assumption of a disordered cation distribution, the end-member formula of ‘wolframoixiolite’ could be (Nb_2/3Fe²⁺_1/3)O₂. However, this mineral also needs additional investigation.

Summary of the approved report

Establishment of the supergroup

The columbite supergroup is established. It is divided into the ixiolite group, the wolframite group, the samarskite group, the columbite group and the wodginite group (Table 1).

Table 1. Minerals belonging to the columbite supergroup.

Note: Names of insufficiently studied minerals are italicised.

Redefined species

Currently, the IMA-accepted formulae of some mineral species belonging to the columbite supergroup do not correspond to their end-members. An introduction of end-member formulae for these minerals implies their redefinition. All these changes are summarised in Table 2.

Table 2. Changes in the formulae of columbite-supergroup minerals.

Discredited species

The currently IMA-accepted formula for polycrase-(Y) is Y(Ti,Nb)₂(O,OH)₆. Its end-member formula is Y(NbTi)O₆, which is identical to the revised formula of euxenite-(Y) (cf. Table 2). As euxenite (Scheerer, 1840) is older than polycrase (Scheerer, 1844), polycrase-(Y) should be discredited.

New species within the ixiolite group

As noted above, Nb-dominant analogues of ixiolite with different schemes of charge balancing are known from numerous localities. In order to distinguish minerals with different kinds of dominant charge-compensating cations (DCCC), the end-member formula will depend on the dominant cation within the dominant valence state of the charge-compensating cation. Accordingly, formulae will have the form:

$${\rm for \;DCCC = 3 + :}\;\Big( {{\rm T}{\rm a}_{{\rm 0}{\rm .5}}M^{{\rm 3 + }}_{{\rm 0}{\rm .5}} } \Big){\rm O}_{\rm 2}\,{\rm and }\;\Big( {{\rm N}{\rm b}_{{\rm 0}{\rm .5}}M^{{\rm 3 + }}_{{\rm 0}{\rm .5}} } \Big){\rm O}_{\rm 2}{\rm;}$$

$${\rm for \;DCCC = 2 + :}\;\left( {{\rm T}{\rm a}_{{\rm 2/3}}M^{{\rm 2 + }}_{{\rm 1/3}} } \right){\rm O}_{\rm 2}\;{\rm and }\;\left( {{\rm N}{\rm b}_{{\rm 2/3}}M^{{\rm 2 + }}_{{\rm 1/3}} } \right){\rm O}_{\rm 2}{\rm;}$$

$${\rm for \;DCCC = 1 + :}\;\Big( {{\rm T}{\rm a}_{{\rm 0}{\rm .75}}M^{\rm + }_{{\rm 0}{\rm .25}} } \Big){\rm O}_{\rm 2}\,{\rm and }\;\Big( {{\rm N}{\rm b}_{{\rm 0}{\rm .75}}M^{\rm + }_{{\rm 0}{\rm .25}} } \Big){\rm O}_{\rm 2}{\rm;}$$

$$\!\!\!\!\!\!\!\!\!\!\!{\rm for \;DCCC = 0:\Big(T}{\rm a}_{{\rm 0}{\rm .8}}\squ _{{\rm 0}{\rm .2}}{\rm \Big)}{\rm O}_{\rm 2}{\rm \;and\; \Big(N}{\rm b}_{{\rm 0}{\rm .8}}\squ _{{\rm 0}{\rm .2}}{\rm \Big)}{\rm O}_{\rm 2}{\rm .}$$

The DCCC will be appended to the root name ‘ixiolite’ (for Ta-dominant end-members) or ‘nioboixiolite’ (for Nb-dominant end-members). Accordingly:

1. (1) The current ‘ixiolite’ will become ixiolite-(Mn²⁺) with the formula (Ta_2/3Mn²⁺_1/3)O₂.

2. (2) Because Fe²⁺-dominant ‘ixiolite’ is also known to occur at the same locality (Rose, 1858; Nickel et al., 1963a), ixiolite-(Fe²⁺) is now considered a distinct mineral species, with the formula (Ta_2/3Fe²⁺_1/3)O₂. The type locality for ixiolite-(Fe²⁺) is Skogsböle, Kimito, Finland. A similar procedure was adopted recently for the two grandfathered minerals ‘tetrahedrite’ and ‘tennantite’: both were redefined into two distinct species, after the IMA-approved report on the tetrahedrite group (Biagioni et al., 2020).

The names with no suffixes: ‘ixiolite’ and ‘nioboixiolite’, will not refer to any specific mineral species and will have the status of series names.

The status of the ixiolite-related mineral qitianlingite remains unclear until more reliable data on the crystal structure of the holotype sample is solved.

Change of status

The crystal structures of three metamict minerals tentatively assigned to the samarskite group [namely, samarskite-(Yb), approved with the current formula YbNbO₄, ishikawaite, grandfathered with the current formula (U,Fe,Y)NbO₄, and calciosamarskite, grandfathered with the current formula (Ca,Fe,Y) (Nb,Ta,Ti)O₄] are unknown. Provided that these minerals are isostructural with samarskite-(Y), their end-member formulae could be written as YbFe³⁺Nb₂O₈, U⁴⁺Fe²⁺Nb₂O₈, and CaFe³⁺Nb₂O₇(OH), respectively. However, before making effective the changes in their end-member formulae, all these minerals need further study and so currently should be considered as questionable species; for instance, according to the type description of samarskite-(Yb) (Simmons et al., 2006), the mineral is iron-depleted, with only 0.11 Fe apfu, and all iron tentatively given as Fe²⁺.

The status of yttrotantalite-(Y) is changed from Rn (renamed) to Q (questionable).