Structural and compositional variations of basic Cu(II) chlorides in the herbertsmithite and gillardite structure field

Abstract Natural samples of the substituted basic Cu(II) chloride series, Cu4–x M x 2+(OH)6Cl2(M = Zn, Ni, or Mg) were investigated by single-crystal X-ray diffraction in order to elucidate compositional boundaries associated with paratacamite and its congeners. The compositional ranges examined are Cu3.65Zn0.35(OH)6Cl2 – Cu3.36Zn0.64(OH)6Cl2 and Cu3.61Ni0.39(OH)6Cl2 – Cu3.13Ni0.87(OH)6Cl2, along with a single Mg-bearing phase. The majority of samples studied have trigonal symmetry (R3̄m) analogous to that of herbertsmithite (Zn) and gillardite (Ni), with a ≈ 6.8, c ≈ 14.0 Å. Crystallographic variations for these samples caused by composition are compared with both published and new data for the R3̄m sub-cell of paratacamite, paratacamite-(Mg) and paratacamite-(Ni). The observed trends suggest that the composition of end-members associated with the paratacamite congeners depend upon the nature of the substituting cation.

This crystallographic investigation of naturally occurring samples from the series was carried out to elucidate the compositional boundary between the R 3 and R 3m structures in terms of Zn and Ni substitution.

Samples and analysis
Specimens of the basic Cu(II) chlorides were obtained from the Mineralogical Museum, Hamburg, Germany, and from several private collections for compositional and crystallographic analysis. The authors analysed samples of paratacamite from the British Museum, London, UK (specimen BM86958), paratacamite-(Mg) from the Natural History Museum of Los Angeles County, USA (specimen 64041) and paratacamite-(Ni) from the Western Australian Museum, Western Australia, Australia (specimen WAM M365.2003), in this study, but full data of the analyses appear in the separate publications Welch et al., (2014), Kampf et al. (2013a) and Sciberras et al. (2013), respectively. Additional analyses of these samples are included in this paper. The remainder of samples and their localities are reported in Table 1.

Crystallographic measurements
Crystals of Ni-bearing specimens from the 132 N deposit G8502, G8568 and G7751, were measured at 293(2) K using a Bruker Smart 1000 CCD diffractometer with graphite-monochromated MoKα radiation. The remaining samples from the Carr Boyd Rocks mine, the Murrin Murrin mine and the San Francisco mine, CB03, CB07, MM02, MD166-2 and MD166-3 were analysed at 294(2) K on a Nonius Kappa CCD diffractometer with MoKα radiation. Final unit-cell dimensions were determined by a least-squares refinement of the full data sets and all structure refinements were made using SHELXL (Sheldrick, 2008) based on atom coordinates reported for analogous phases (Braithwaite et al., 2004;Clissold et al., 2007). Special attention was given to the identification of weak reflections at half integer positions of h and k, which correspond to the paratacamite superstructure. Pseudo-precession diffraction patterns reconstructed from the full data collections for each sample indicated the R 3m substructure (Table 2), 2a* superlattice reflections being absent.
Similarly, Zn-bearing samples exhibited unit-cell parameters related to herbertsmithite (a ≈ 6.8, c ≈ 14.1 Å). The range detected expressed the varying contribution of Zn content, increasing from 14.046(9) to 14.062(4) Å, as Zn content increases. The reported unit cell for herbertsmithite is a = 6.834, c = 14.075 Å for material of end-member composition Cu 3 Zn (OH) 6 Cl 2 (Braithwaite et al., 2004) and is in line with the composition vs. unit-cell relationship determined here. These results are also in accord with the variation in cell parameters reported for synthetic trigonal Zn-bearing members of the basic Cu(II) chlorides by Jambor et al. (1996).
Due to the absence of any super-lattice reflections and the similarity of these unit cells with those reported for herbertsmithite and gillardite, structural refinements were made in space group R 3m for all data sets. All structures were refined based on the atom coordinates established by Braithwaite et al. (2004) and Clissold et al. (2007) for herbertsmithite and gillardite, respectively, and converged to acceptable residuals and anisotropic thermal parameters. Structure refinement details can be found in Table 2. Selected crystallographic data are given in Table 3.
The paratacamite R 3m sub-cell structure is an average representation of the full R 3 super-cell structure (Fleet 1975;Welch et al., 2014). Crystallographic data for the substructures of samples identified as paratacamite (BM86958) (Welch et al., 2014), paratacamite-(Mg) (64041) (Kampf et al., 2013a) and paratacamite-(Ni) (WAM M365.2003) (Sciberras et al., 2013), were refined in space group R 3m after data reduction of the full set of structure factors to include only the sublattice reflections. Selected crystallographic data for the sub-cell structure of these paratacamite samples is given in Table 3.

Description of the structures
The R 3m structure is characterized by layers of (4+2) Jahn-Teller distorted octahedra of composition [CuCl 2 (OH) 4 ] (centred at the M(2) site), which are linked together in the interlayer M(1) site by an M 2+ O 6 octahedron. This interlayer metal position is bonded to six symmetry equivalent O atoms and exhibits a slight angular distortion. While the M(2) site is completely composed of Cu 2+ , the M(1) site bears the extent of Cu substitution by other divalent cations with similar ionic radius. This is the same scheme of metal distribution adopted for the related R 3m phases herbertsmithite (Braithwaite et al., 2004), gillardite , leverettite (Kampf et al., 2013b) and tondiite . The
It is important to note that the R 3m structure shared by herbertsmithite, gillardite, leverettite and tondiite, is topologically, but not crystallographically, identical to that of paratacamite (R 3) and its congeners. The former minerals, sensu stricto, are defined as having an interlayer site that is dominated by Zn, Ni, Co or Mg respectively (Braithwaite et al., 2004;Clissold et al., 2007;Kampf et al., 2013b;Malcherek et al., 2014). Guidelines for nomenclature of topologically identical phases defer to the 'dominant-constituent rule' (Hatert and Burke, 2008). Therefore, those samples exhibiting the R 3m structure but with Cu dominance in the interlayer, represent a separate species that deserves a unique name. This issue will be addressed in a future manuscript.
An examination of selected crystallographic data (Table 3) for samples containing Zn 2+ as the primary substituting cation shows that a and c axes decrease towards the monoclinic-trigonal transformation boundary, in line with the observations of powdered material in Jambor et al. (1996). There is a small contraction of M-O bond lengths for both metal sites with decreasing Zn content. All For Zn-bearing samples, there is no significant change in the O···Cl distance with changes in composition. The Ni-bearing samples show only a minor decrease in the O···Cl distance with increasing Ni-content. Data from the paratacamite R 3m structure are generally consistent with trends observed for herbertsmithite and gillardite (R 3m) samples.
The lattice strain induced by composition was calculated by determining the corresponding strain tensor of the aristotype unit cell as well as the transformed paratacamite sub-cell for samples listed in Table 3. The strain tensors were then used to calculate the scalar strain. According to the crystallographic data in Table 3, the paratacamite substructure offers a good comparison with samples exhibiting the aristotype structure (sensu stricto). Therefore, the corresponding unit-cell strain observed for this substructure should also be comparable with the compositional trends observed for the aristotype structure. The tensor components for the hexagonal setting can be determined from the following equations: The above equations are from Carpenter et al. (1998) and are discussed in the context of this mineral series by Malcherek and Schlüter (2009 This material is not ideal as a reference for the lattice parameters expected for pure Cu 3 Ni(OH) 6 Cl 2 , but was retained here because it exhibits the smallest lattice volume and highest substitution of the available gillardites in the literature of this study.
Calculations were made using the unit-cell parameters in Table 3 for all Zn-and Ni-bearing samples. The trace amount of Co present in some of the gillardite samples is not expected to contribute significantly to the unit cell volume. The scalar strain and calculated tensor components can be found in Table 4 in the final column.
The distortion of the aristotype unit cell increases towards the trigonal→monoclinic transformation as the critical interlayer Cu content is approached. The  Fig. 1. The sub-cell of paratacamite (BM86958) shows the greatest strain of all Zn-bearing samples. The upper compositional limit proposed for the stability of clinoatacamite, at x ≈ 0.33, appears to be a critical composition in terms of the aristotype unit-cell strain. Extrapolation of the trend for Zn-bearing samples indicates that the Zn composition of holotype paratacamite examined by Fleet (1975), with a scalar strain of 0.0028 associated with the sub-cell, is between c. Cu 3.70 Zn 0.30 (OH) 6 Cl 2 and Cu 3.67 Zn 0.33 (OH) 6 Cl 2 . The distortion of the M(1) octahedron in the R 3m aristotype structure was calculated for Zn-and Nibearing material in this study using the formulation for quadratic elongation (QE) and bond-angle variance (BAV) of Robinson et al. (1971), as implemented in the program VESTA (Momma and Izumi, 2008). The data are displayed in Fig. 2. Both the QE and BAV values for herbertsmithite and gillardite samples show significant changes that can be related to composition. The single representative QE and BAV value determined from the paratacamite (BM86958) R 3m structure, with a composition of Cu 3.71 Zn 0.29 (OH) 6 Cl 2 (Welch et al., 2014), has the highest distortion of all Zn-bearing samples. With increasing Zn content, both QE and BAV values decrease to a minimum for compositions above x ≈ 0.6 and are unaffected by increased Zn content. Similarly, gillardite samples show a significant and reproducible decrease for both QE and BAV values with excess Ni content. However, the decrease in these values appears to be sharper and occurs at a composition x > 0.7. The R 3m structure of paratacamite-(Ni) gives comparable QE and BAV values with samples having lower Ni contents.
The holotype paratacamite of Fleet (1975) has QE and BAV values associated with the interlayer octahedron of the average sub-cell structure of 1.053 and 207.64 deg 2 , respectively. Extrapolation of the trends in Fig. 2 indicate a compositional range in agreement with that suggested from the scalar strain results described above.

Conclusions
The difference in trend evolution of QE and BAV values between the Zn-or Ni-bearing aristotype Compositional error bars are smaller than the size of the symbol. structure may be attributed to the difference in crystal-chemical behaviour of these cations. These results show that the distortion exhibited by the M(1)O 6 octahedron varies with changes in composition in the aristotype structure. It may be inferred that the analogous interlayer position in the paratacamite superstructure at M(1), which is invariant with temperature (Welch et al., 2014), varies with composition. Therefore, it is likely that the Zn-and Ni-bearing samples of paratacamite would have a different set of end-members. This could also be true of other paratacamite congeners. However, the end-members associated with Zn or Ni substitution in paratacamite could not be identified from this study. Both paratacamite-(Ni) and paratacamite-(Mg) examined here have >50% interlayer occupancy of the substituting cation. This may indicate that the R 3 super-cell may also exist across much of the substitution series. One must consider also the multitude of structural refinements for the R 3m aristotype structure with end-member or near endmember stoichiometry from the literature Braithwaite et al., 2004;Chu et al., 2010Chu et al., , 2011Han et al., 2011;Chu, 2011;Wulferding et al., 2010;Schores et al., 2005). The aristotype structure appears to be thermodynamically stable near the end-member composition Cu 3 M (OH) 6 Cl 2 . As the presence of Cu 2+ becomes significant in the interlayer the R 3 structure may become metastable. Based on the quantifiable distortion of the interlayer position in the aristotype structure, the substituting cation defines the range of stability (or metastability) for the phase. This implies that under the right conditions paratacamite congeners would crystallize before their corresponding aristotype phase, herbertsmithite or gillardite for Zn and Ni, respectively, and by extension tondiite and leverettite for Mg and Co, respectively, described by the Ostwald step rule (Ostwald, 1897). The particular conditions which promote the nucleation and growth of the aristotype structure may serve to inhibit the nucleation and growth of R 3 domains.