Redetermination of olivenite from an untwinned single-crystal.

The crystal structure of olivenite, ideally Cu(2)(AsO(4))(OH) [dicopper(II) arsenate(V) hydroxide], was redetermined from an untwinned and phosphate-containing natural sample, composition Cu(2)(As(0.92)P(0.08)O(4)), from Majuba Hill (Nevada, USA). Olivenite is structurally analogous with the important rock-forming mineral andalusite, Al(2)OSiO(4). Its structure consists of chains of edge-sharing, distorted [CuO(4)(OH)(2)] octa-hedra extending parallel to [001]. These chains are cross-linked by isolated AsO(4) tetra-hedra through corner-sharing, forming channels in which dimers of edge-sharing [CuO(4)(OH)] trigonal bipyramids are located. The structure is stabilized by medium to weak O-H⋯O hydrogen bonds. In contrast to the previous refinements from powder and single crystal X-ray data, all non-H atoms were refined with anisotropic displacement parameters and the H atom was located.

The crystal structure of olivenite, ideally Cu 2 (AsO 4 )(OH) [dicopper(II) arsenate(V) hydroxide], was redetermined from an untwinned and phosphate-containing natural sample, composition Cu 2 (As 0.92 P 0.08 O 4 ), from Majuba Hill (Nevada, USA). Olivenite is structurally analogous with the important rock-forming mineral andalusite, Al 2 OSiO 4 . Its structure consists of chains of edge-sharing, distorted [CuO 4 (OH) 2 ] octahedra extending parallel to [001]. These chains are crosslinked by isolated AsO 4 tetrahedra through corner-sharing, forming channels in which dimers of edge-sharing [CuO 4 (OH)] trigonal bipyramids are located. The structure is stabilized by medium to weak O-HÁ Á ÁO hydrogen bonds. In contrast to the previous refinements from powder and single crystal X-ray data, all non-H atoms were refined with anisotropic displacement parameters and the H atom was located.
Toman (1977) proposed that olivenite has actually monoclinic symmetry, and that most of the crystals are twinned. Structure refinements based on single-crystal X-ray diffraction data, uncorrected and corrected for twinning, yielded reliability factors R(F) of 0.090 and 0.065, respectively. However, Toman (1977) did not report any atomic displacement parameters or the position of the H atom. To avoid the complication of interpreting X-ray diffraction intensity data due to twinning, Burns & Hawthorne (1995) performed structure refinements of olivenite using the Rietveld method from powder X-ray diffraction data. By assuming a single isotropic displacement parameter for all O atoms and no H atom position, they attempted refinements both in space group Pnnm and P2 1 /n and obtained nearly identical R Bragg factors (~ 0.074) and goodness-of-fit values (~ 2.30). In our efforts to understand the relationships between the hydrogen environments and Raman spectra of hydrous minerals, we concluded that the structural information of olivenite needs to be improved. During the course of sample identification for the RRUFF project, we discovered an untwinned and phosphate-containing single-crystal of olivenite from Majuba Hill, Pershing County, Nevada, USA, and conducted a detailed structure refinement.  (Fig. 1). Although the average <As1-O>, <Cu1-O>, and <Cu2-O> distances of our study agree well with those given by Toman (1977) and Burns & Hawthorne (1995), the corresponding individual bond distances and angles from the three structure refinements (including ours) vary significantly. For example, the shortest and longest As-O bond distances within the AsO 4 tetrahedra are 1.618 Å (As-O5) and 1.731 Å (As-O4), respectively, from Toman (1977) (1995). Furthermore, the AsO 4 tetrahedron reported by Burns & Hawthorne (1995) is remarkably distorted, as measured by the tetrahedral angle variance (TAV) and quadratic elongation (TQE) (Robinson et al., 1971), which are 105.3 and 1.0281, respectively. In comparison, the TAV and TQE values are 6.1 and 1.0027, respectively, from Toman (1977), and 2.8 and 1.0008 from this study.

supplementary materials sup-2
The H atom is bonded to O3, at a separation of 0.67 (4) Å. This distance is in fairly agreement with that (0.77 Å) reported for adamite (Hill, 1976). Our Raman spectra of olivenite (http://rruff.info/olivenite/R040181) show two major bands in the hydroxyl stretching (ν OH ) region: one at 3440 cm -1 and the other at 3464 cm -1 . Similar wavenumbers (ν OH = 3437 and 3464 cm -1 ) were also obtained by Frost et al. (2002). Given the correlation between ν OH and O-H···O distances in minerals (Libowitzky, 1999), one would expect two O-H···O distances between 2.8 and 2.9 Å in olivenite. Our structural data indeed show that the O3(=OH) atom is at a distance of 2.79 Å from O4 and 2.98 Å from O5. Nevertheless, the angles O3-H···O4 (125°) and O3-H···O5 (112°) appear to be too small for hydrogen bonding. Note that, based on the structure refinement of libethenite, the phosphate analogue of olivenite, Cordsen (1978) proposed a bifurcated hydrogen bonding model for this mineral, in which there are two O-H···O bonds at the same distance of 2.84 Å and two bonding angles of 110°.

Refinement
The setting in P2 1 /n11 with the a-axis as the monoclinic axis was chosen to keep consistency with previous studies on this mineral (Toman, 1977;Burns & Hawthorne, 1995). The minor vacancies for Cu and OH determined from electron microprobe analysis were ignored during the final refinement and all corresponding sites were assumed to be fully occupied.
The relative ratio of As versus P at the As1 site was allowed to vary freely, with the sum of site occupation factors constrained to unity. The results agree well with those obtained from electron microprobe analysis. The H atom was located in a difference Fourier map, and its position was refined freely. The highest residual peak in the difference Fourier maps was located 0.80 Å from atom O1, and the deepest hole was located 0.63 Å from Cu1. Fig. 1. The crystal structure of olivenite given in the polyhedral representation. The H atoms (blue spheres) are drawn with an arbitrary radius.

(I)
Crystal data Refinement. Refinement of F 2 against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F 2 , conventional R-factors R are based on F, with F set to zero for negative F 2 . The threshold expression of F 2 > σ(F 2 ) is used only for calculating Rfactors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F 2 are statistically about twice as large as those based on F, and R-factors based on ALL data will be even larger.