A CRYSTALLOCHEMICAL STUDY OF GERSDORFFITE AND
RELATED MINERALS
A thesis presented in 1968 to
The University of New South Wales
for the Degree of Doctor of Philosophy
This thesis has not been previously
presented in whole or part to another University
or Institution for a higher degree
Peter Bayliss (ii)
CONTENTS
Page
Summary 1.
Introduction 3.
Literature Survey: Crystal structure 5.
Physical properties: 9 •
Electrical and magnetic properties 9.
Bonds and atomic radii 12.
Order-disorder and forbidden reflections 13.
Thermodynamic properties 14.
Gersdorffite: Crystallography 15.
Chemistry 16.
Cobaltite: Crystallography 18.
Chemistry 21.
Ullmannite: Crystallography 22.
Chemistry 22.
Arsenopyrite: Crystallography 24.
Chemistry 26.
Glaucodot: Crystallography 27.
Chemistry 28.
Gudmundite: Crystallography 29.
Chemistry 30.
Literature review 31. (iii)
Page
Methods: Synthesis 33.
X-ray diffraction of synthetic powder 37.
Natural powder examination 42.
Single crystal X-ray diffraction 43.
Crystal structure computations 47.
Results: Gersdorffite: Natural material 49.
Crystal structure Slovakia Pa3 54.
Wolfsberg P213 55.
Leichtenberg PI 58.
Synthesis 62.
Cobaltite: Natural material 64.
Crystal stxucture 66.
Synthesis 67.
Ullmannite: Natural material 69.
Crystal structure 70.
Synthesis 72.
Arsenopyrite, glaucodot and gudmundite:
Natural material 73.
Synthesis 74.
Pyrite, cattierite and vaesite 74. (iv)
Page
Discussion: Methods 76.
Natural material 77.
Synthesis 78.
Bernd angles and distances 80.
Crystal structure accuracy 83.
Crystal structure 85.
Crystal structure relations 87.
Acknowledgments 94.
References 95.
Data appendix: Unit cell data 107.
X-ray diffraction powder intensities, optical
anisotropy, zoning and paragenesis 114.
Crystallographic data 118.
Chemical data 133.
Heating experiments 135.
Publication List 137. (v)
LIST OF TABLES
Page
1. Unit Cell Data
la. Gersdorffite: literature and natural 107.
lb. artificial 108.
le. Cobaltite: literature and natural 109.
Id. artificial 110.
le. Ullmannite: literature and natural 111.
If. Ullmannite and gersdorffite:
artificial 112.
lg. Arsenopyrite, glaucodot and gudmundite 113.
2. X-ray Diffraction Powder Intensities, Optical anisotropy,
Zoning and Paragenesis.
2a. Gersdorffite 114.
2b. Cobaltite 115.
2c. Ullmannite 116.
2d. Arsenopyrite, glaucodot and gudmundite 117.
3. Crystallographic Data: Observed reflection amplitudes,
Calculated structure factors, Atomic and Thermal parameters,
lnteratomic distances, Ato'!lic radii, and Tetrahedral and
Octahedral a ogles.
3a. Gersdorffite: Slovakia, Pa3 118. (vi)
Page
3b. Wolfsberg, P213 120.
3c. Leichtenberg, PI 122.
3d. Gersdorffite: synthetic arsenic-rich 124.
3e. synthetic 125.
3f. synthetic sulphur-rich 126.
3g. Cobaltite: Pa3, Giese and Kerr (1965) 127.
3h. Pca21, Giese and Kerr (1965) 128.
3j. Ullmannite: Takeuchi (1957) 129.
3k. synthetic 131.
31. Pyrite, cattierite and vaesite 132.
4. Olemical Data.
4a. hnpurities (ppm) in specpure materials 133.
4b. Spectrographic analyses 134.
5. Heating Experiments.
Sa. Gersdorffite 135.
Sb. Cobaltite 136. (vii)
LIST OF FIGURES
Page
I . Pyrite and marcasite crystal structures 7.
2. FeAsS-CoAsS-NiAsS ternary diagram of Klemm (1965) 16.
3. Cell edge versus gersdorffite composition I 7.
4. Cell edge versus cobaltite composition 67.
5. Cell edge versus ullmannite composition 71.
6. NiSb2-NiAs2-NiS2 ternary diagram 72.
7. Diagrammatic sketch of FeAsS-CoAsS join 74.
8. Positional displacements in crystal structures of gersdorffite 85.
9. Diagrammatic sketch of CoAsS-NiAsS join 92. I.
SUMMARY
Data are presented from natural samples of gersdorffite and related minerals for unit cell size, powder X-ray diffraction reflection intensities, optical anisotropism, zoning, and paragenesis. Three single crystal structure analyses of distinct gersdorffite crystals, and single crystal structure refinements of cobaltite and ullmannite elucidate the non •metal atom ordering and structure distortion. For gersdorffite and related minerals, the absence of both the 001 and 011 reflections in their powder patterns indicate a disordered cubic structure Pa3; the absence of the 001 reflection and the presence of the 011 reflection in their powder patterns indicate a partially ordered cubic structure P213; the presence of both the
001 and OU reflections in their powder patterns indicate a non-cubic distorted structure Pca21 or Pl. For gersdorffite and related minerals, the amount of structure distortion is related to the 001 reflection intensity in their powder patterns and also to their optical anisotropism strength. These are semi quantitatively related to the thermal stability of the distorted structure.
This thermal stability increases with the compositional substitution of cobalt for nickel, which gives a decrease in the unit cell size. An order•disorder change occurs before a distortion release in gersdorffite with a large unit cell size, whereas only a distortion release is observed in cobaltite and gersdorffite with a small unit cell size. 2.
Synthetic gersdorffite results show that ordering of the non •metal atoms decreases with a rise in formation temperature and deviation from the stoichiometric composition. The small distortion (small 001 reflection
intensity in a powder pattern) in synthetic cobaltite is attributed to slow reaction rates and a possible hysteresis cycle. The prepared synthetic cobaltite compositions range from CoAs0 • 86s1. l4 to CoAs0 _42S1. 58 at
550° C. The prepared synthetic ullmannite is stoichiometric NiSbS, al• though arsenic may substitute extensively for antimony and sulphur. Syn thetic materials with stoichiometric compositions and limited metal atom
substitution were produced by the small evacuated tube method in contrast to the LiCl•KCl melt method. 3.
INTRODUCTION
In recent years, extensive research work has been conducted into sulphides. This research work includes phase equilibrium studies such as the geologically significant sulphide-type systems, which were reviewed and evaluated by Kullerud (I 964). The limits of solid solution in these phase equilibrium studies indicate the compositional range of each mineral.
Other lines of research work include crystallographic determinations such as the unit cell data tanilated by Donnay and Donnay (1963), the crystal structures described by Bragg and Claringrull (1965), and the X-ray dif• fraction powder data collated by Berry and Thompson (1962). Other lines of research involve the geological applicability of sulphides, such as the review by Kullerud (1959) on the use of sulphide-type systems as geological thermometers.
The extent of knowledge about gersdorffite and its related minerals is reviewed in the following literature survey under sections on crystal structure, physical properties, and the crystallography and chemistry of each mineral.
The aim of this investigation was to obtain data to fill omissions and check inconsistencies in the published data, which are summarized in the literature review. The methods used to investigate these problems were mainly mineral syntheses from pure elements, polished section 4. studies, and X-ray diffraction powder and single crystal structure analyses.
The data obtained are presented in the results section. These data are then combined with the published data to present a comprehensive theory to account for all the data in the discussion. A major difficulty in the experimental work was to decide when the synthetic minerals had reached equilibrium. After the start of this project in 1964, additional valuable information was added to the literature by Giese and Kerr (1965) on the crystal structure of cobaltite and by Klemm (1965) on the chemical composition of (Fe, Co, Ni)AsS compounds. 5.
LITERATIJRE SURVEY
CRYSTAL STRUCTURE
The crystal structures of AX2 compounds to which gersdorffite and its related minerals belong are grouped into several major types by
Wycoff (1963) of symmetrical (predominantly ionic), layer (moderately ionic), molecular (covalent), and metallic. In a symmetrical structure, A and X differ widely in electro-negativity to produce mainly ionic bonding so the structure type is determined by their ionic radii (r). If r1/r~,,. O. 7 then the resultant 8:4 co-ordination is called the fluorite-type structure; if O. 7 > rA /r)( "" 0.3 then the resultant 6:3 co-ordination is called the rutile or cassiterite-type structure; and if rl/r~ < 0.3 then the resultant
4:2 co-ordination is called the silica-type structure. In a layer structure, a resonance between predominantly covalent and ionic bonding occurs within each layer, whereas the weak bond of Van der Waals holds the layers to• gether. The types of layer structure include cubic close packing for example CdCl2, hexagonal close packing for example CdI2, hexagonal packing for example MoS2, and approximate close packing for example
AlO(OH). The molecular structure varieties are the infinite such as
FeS2 and the discrete such as co2 with covalent bonding within molecules and Van der Waals forces between molecules. A metallic structure, for example Al82, has intermetallic bonding.
Sulphides, arsenides and related AX 2 compounds of elements in the 8 sub-group of the periodic table (sulphide-chemistry) differ from the 6. corresponding oxide compounds because firstly sulphur, selenium, telurium, arsenic, antimony and bismuth atoms are larger and more easily polarized than oxygen atoms; secondly these non-metal atoms can form covalent bonds between each other; and thirdly arsenic, antimony and bismuth have semi-metallic properties. There are several different structure types given for these AX2 compounds by Wyckoff (1963) and
Bragg and Claringbull (1965). No sulphide crystallizes with a typical ionic structure because of its low electro-negativity. Examples of different structure types are melonite (NiTe2) a hexagonal close packed layer structure; molybdenite (MoS2) a hexagonal layer structure; krennerite
(AuTe2) a linear co-ordination structure; pyrite (FeS2) an infinite mole• cular structure, cobaltite (CoAsS) a superstructure derivative of pyrite, and PdS2 a derivative of pyrite by elongation along one of the pyrite axes; marcasite (FeS2) a distorted infinite molecular structure, arsenopyrite
(FeAsS) a superstructure derivative of marcasite~ and IrSe2 an open derivative of marcasite; and lautite (CuAsS) a sphalerite derivative des cribed by Craig and Stephenson (1965).
To this classification, additions have been proposed. Separate marcasite (FeS2) and loellingite (FeAs2) sub-groups and separate arseno• pyrite (FeAsS) and gudmundite (FeSbS) sub-groups are suggested by
&ierger (193 7). Separate cobaltite ( CoAsS) and ullmannite (NiSbS) sub• groups are proposed by Strunz (1966). For the cobaltite group, Wells (1962) 7. describes the following three sub•group possibilities: each s2 group in pyrite has been replaced with As•S (Sb•S) for example ullmannite (NiSbS); one•half of the s2 groups in pyrite have been replaced with As2(Sb2) for example cobaltite (CoAsS); or there is a random arrangement of S+As(Sb) in the s2 positions of pyrite for example gersdorffite (NiAsS).
An extended series of AX2 and AXY transition element sulphides are tanilated by Hulliger and Mooser (1965a, 1965b) under structure types of pyrite and its ternary phase derivatives, and marcasite with its arseno• pyrite derivatives.
Pyrite (figure la from Wycoff, 1963) is regarded by Evans (1964) as constituting a single giant molecule, just as a diamond crystal is a
single carbon molecule. This molecular character (figure lb from
Wycoff, 1963) is more important than the geometric similarity of pyrite to an intermediate crystal structure between rocksalt (NaCl) with replacement of Cl by a non•metal atom pair and fluorite (CaF2) with replacement of F2
by a non-metal atom pair (figure le from Bragg and Claringbull, 1965).
This intermediate crystal structure is derived from parameter x, which is the separation between members of a non•metal atom pair since pyrite
with x = 3/8 lies between rocksalt with x = 1/2 and fluorite with x = 1/4.
In addition pyrite combines an octahedral metal atom co-ordination from
rocksalt and a tetrahedral non-metal atom co-ordination from fluorite 8. with minor distortions to form the most symmetrical structure (figure Id from Bragg and Cl.aringbull, 1965). Pyrite is discussed under the alloy systems of T 2-B2 (transition metal•less metallic B sub-group element), where the range of solid solution becomes restricted and definite structures of varying degrees of complexity are formed. Its characteristics however include metallic appearance and relatively high electrical conductivity.
The marcasite structure (figure le from Wycoff, 1963), which corn• bines octahedral metal atom and tetrahedral non•metal atom co-ordination, is based on a deformed close-packed hexagonal non-metal atom substructure
(figure If from Bragg and Cl.aringrull, 1965), in which each layer of octa• hedral interstices is half filled with metal atoms. The octahedra i11 mar• casite share edges, unlike pyrite where neighbouring octahedra share only corners, so that linear metal atom chains form in marcasite along the shortest axis of /§oj]. Since the shared edges of the (FeS6) octahedra are longer than the unshared edges, Evans (1960) considers this as additional evidence for covalent bonding, because anion polyhedra in ionic crystals rarely share edges and such shared edges are usually shortened owing to the mutual repulsion of the cations. This [ooiJ axis is so compressed in some d4 marcasites that the ratio of the octahedron edges parallel and perpendicular to the short axis is O. 75, but some d6 marcasites have a ratio that exceeds 1. 0. This compression is interpreted by Hulliger and
Mooser (1965b) in terms of Jahn-Teller deformations rather than as direct 9. metal•metal bonding but the expansion is unexplained. Pearson (1965) explains this compression as resulting from mixed bonding. Arsenopyrite is described as a deformed marcasite, when the chains of equidistant metal atom pairs in marcasite are replaced by the chains of metal atom pairs with alternate short (bonded) and long (non•bonded) metal•metal distances.
The mean •square deviation of the octahedral angle from the perfect value of 90°lies betweeri4° to 5° for pyrite-type and anomalous marcasite type structures, but is about 10° for marcasite•type structures.
SOME PHYSICAL PROPERTIES
Under the following four sub-headings in this section, some physical properties such as electrical and magnetic properties, bonds and atomic radii, order-disorder and forbidden reflections, and thermodynamic pro perties are summarized. These properties are considered in this literature survey because they have direct implications on the chemistry and crystal lography of gersdorffite and its related minerals.
Electrical and Magnetic Properties
Some electrical properties, magnetic properties, and the number of d electrons per metal atom are considered for some pyrite-type and marcasite-type compounds with ionic formulae of ~~+(x2)2·, M3+(XY)3- and M4 + (Y 2)4-. These are tabulated and interpreted by Hulliger and 10.
Mooser (1965a, 1965b) in terms of their crystal chemistry on a model in which the metal atom d electrons are assumed to be essentially localized.
Classification based upon electron configuration are high-spin d5, d8 and low-spin d6 and d7 for pyrite-type structures and high-spin d1, d2, d6, d7 and low-spin d4 and d5 for marcasite-type structures. A change from mar casite to pyrite crystal structure is given by Allen and Crenshaw (1914) at
350°C for FeS2 . Similarly both crystal structure-types are exhibited by
CoSe2 and PdBi2 at various temperatures. This crystal structure change also occurs by replacement with heavier metal atoms such as in series
FeTe2 - RuTe2 or with lighter non-metal atoms such as in series FeS2 - 8 7 2+( 2- . FeSe2 . Some d and a few d M X2) compounds have the melomte-type crystal structure. RhTe2 changes from pyrite-type to melonite-type crystal structure with higher temperature, and this crystal structure change also occurs with replacement by heavier non-metal atoms such as in series NiSe2 -
NiTe2 or heavier metal atoms such as in series NiSe2 - PtSe2 . In addition the series, NiSe2 - PdSe2 - PtSe2, has an intermediate compound PdSe2 with a de formed pyrite-type crystal structure, while both the marcasite-type and melonite-type crystal structures are exhibited by CoTe2 . The series RhSe2 -
IrSe2 involves a change from pyrite-type to deformed marcasite-type crystal 9 2+ 2- . structure. The d M (X2) compound group has the krennente-type crystal 4 2+ 2- structure, whereas the d M (X2) compound group has the molybdenite-type crystal structure. 11.
CoAsS and its iso-electronic equivalents with zero magnetic moment
( p = 0) have 20 valence electrons per formula unit. Fourteen electrons are used to form saturated bonds and the six remaining electrons (d6) are localized in three d levels to give diamagnetic and semiconducting pro- perties.
NiAsS with d7 electrons has magnetic moment p ~ 0 and an electron configuration of dE 6dy1• Since this extra dy electron is not localized, the occupation of the two degenerate dy levels by only one electron shows
(1) metallic conductivity, (2) paramagnetism, and (3) superconduction above 1°K. The delocalization of this excess d electron induces metallic conduction in compounds which otherwise would be non -metallic. This is illustrated by the change from semi-metallic to metallic conduction without structural change, which occurs in the series PtAs2-PdAs2 with µ = 0 and dE 6 electron configuration, because the metallic character of the corn - pound is strengthened by replacement with heavier non ·metal atoms (lower electro-negativity) or lighttr metal atoms from the same column in the periodic table. This is also illustrated by the CdI2 type structure in the series PtS2 -PtSe2 -PtTe2 with p = 2.8 and di 6dy2 electron configuration.
FeAsS with d5 electrons has semiconducting and diamagnetic pro perties. In a hypothetical low-spin d5 arsenopyrite, the two singly occupied d levels of the two bonded metal atoms are split up into a filled bonding level and an empty antibonding level. 12.
Bonds and Atomic Radii
The bonds for these compounds are shown by Pauling and Huggins
(1934) to have 6:4 co-ordination, where each non-metal atom has sp3 tetrahedral hybrid covalent bonds with three metal atoms and one non metal atom, and each transition metal atom has d2 sp3 octahedral hybrid covalent bonds with six non-metal atoms. The oxidation states are given by Addison (1961) as plus two for the metal atom and minus two for the non metal atom pair.
The atomic radii determined from the electron pair bond radii of
Pauling and Huggins (1934) ai;e 1.23KX for iron (II), 1.32KX for cobalt (m,
1. 22KX for cobalt (Ill), 1. 39KX for nickel (II), 1. 3 lKX for nickel (III) ,
1. 2 lKX for nickel (IV), 1. 36 KX for antimony, 1. 18 KX for arsenic and
1. 04KX for sulphur. These atomic radii agree with those determined by
Elliott (1960) of l.l73i for iron and l.086R for sulphur in pyrite, of
1.235.R for cobalt and 1.033.R for sulphur in cattierite, and of l.363R for nickel and 1. 062.R for sulphur in vaesite. However these atomic radii disagree with those determined by Buerger (1937b) of l.14KX for iron and
1.1 lKX for sulphur in marcasite. This difference is attributed to additional bonding, which is considered by Pearson (1965) as mixed bonding from con ventional sp3 and d2sp3 to five valence electrons with two in non-metal to non-metal bonds and three in non-metal to metal bonds. 13.
Order-disorder and Forbidden Reflections
The order-disorder phenomena are subdivided by Green, and Hurst
(1964) into 1st order phase transitions such as changes between gas, liquid and solid; and 2nd order phase transitions such as changes in ferromagnetic substances at their Curie point and ordering changes within binary alloys.
An order-disorder transition for a CuAu alloy is described by Evans (1964).
When the CuAu alloy is quenched from above 392°C, the resultant disordered cubic structure contrasts with the ordered tetragonal superstructure with an axial ratio of O. 932 formed when the CuAu alloy is quenched from below
320°C. Intermediate axial ratios of the CuAu alloys indicate partial ordering. An order-disorder transformation for the arsenic and sulphur in cobaltite is given by Giese and Kerr (1965).
The space group Pa3 given for pyrite by Bragg (1913) has no extinction conditions for hkl; whereas extinction conditions occur for hOI with h ~ 2n, and the equivalent conditions of Oki with I.if 2n and hkO with k If 2n together with their derived extinction conditions of hOO with h • 2n,
OkO with k~ 2n and 001 with I~ 2n. From this data, the only systematic absences in the powder pattern occur with N(h2 + k2 + 12 for reflection hkl) = 1, 2 and 10. The other systematic absences may be masked in the powder pattern, since reflections with the same N value can not be dis tinguished. Examples in the hOO series are 300 from 221 (N = 9), and in the hkO series are 210 from 120 (N = 5) and 330 from 411 (N = 18). The space group P2 13 given for ullmannite by Takeuchi (1957) has no extinction 14.
conditions for hlcl, hkO, hOI and Okl; whereas h c: 2n for hOO, and the equival ent conditions are k = 2n for OkO, and I = 2n for 001. From this data, the only systematic absences in the powder pattern are N = 1. The space group
Pca21 given for cobaltite by Giese and Kerr (1965) has no extinction conditions for hkl, hkO and OkO; whereas h = 2n for hOI and hOO, and 1 = 2n for Ok! and
001. From this data, there are no systematic absences in the powder pattern of a pseudo-cubic mineral.
The above data in the powder patterns suggest the following: (I) The absence of reflections when N = I, 2 and 10 result from disordered arsenic and sulphur atoms in a pure end member. (2) Reflections present when
N = 2 and 10 and absent when N = I result from ordered arsenic and sulphur atoms in a pure end member. (3) Reflections present when N = 1, 2 and 10 result from a distortion in a pure end member. An alternative suggestion for these reflections when N = I, 2 and IQ is the substitution of iron and/or cobalt for nickel, and/or the substitution of antimony for arsenic in a pure end member. The chemical similarity within these metal atom and non metal atom groups would indicate a random substitution, which would leave the space group unchanged. The validity of these suggestions needs to be checked.
Thermodynamic Properties
A thermodynamic study of the iron, cobalt, and nickel sulphides is 15. given by Rosenqvist (1954). A simple structure with low order, high entropy and wide homogeneous composition range occurs at high tempera tures, whereas a complex structure with high order, low entropy and small homogeneous composition range occurs at low temperatures. There fore an AX2 compound will have a random distribJtion of non-metal atoms, similar to the pyrite structure, at high temperatures compared to the more ordered distribution of non-metal atoms, similar to the ullmannite structure, at low temperatures. The low entropies (S0 cal. / 0 C), which are given by the Handbook of Chemistry and Physics (1964) as -30 for FeS2, -23 for
CoS2 and -19 for NiS2, indicate negligible composition range. The free energy of formation (G ) for pyrite varies with temperature (T) as given 0 by Toulmin and Barton (1964) in the formula G0 = -71, 280 + 47 .OST cal/g.f.w.
GERSDORFFITE
Crystallography
The pyrite space group Pa3 is given for gersdorffite (NiAsS) by
Peacock and Henry (1948) with nickel atoms at the four-fold position
(0, 0, 0) and an equal distribution of arsenic and sulphur atoms over the eight-fold position (0. 385, 0. 385, 0. 385). The space group P213 is assigned by Ramsdell (1925), Olshausen (1925), Bokii and Tsinokev (1954), Yund
(1962) and Giese and Kerr (1965), which indicates an ullmannite•type structure. From the average measured specific gravity of 5.9 and the 16. cubic unit cell of 5. 7.R for gersdorffite, Palache ~t al. (1944) give four formula units (4NiAsS) per structural cell (Z = 4).
The unit cell size variation with composition is indicated in the
FeAsS-CoAsS-NiAsS ternary diagram (figure 2) of Klemm (1965). Acom pilation from the literature of 16 unit cell sizes for gersdorffite (with varietal names if applicable) and their geographical locations are given in table la. X-ray diffraction powder intensities from two of these gersdorf fites are given in table 2a, but since a cubic mineral is easily indexed and its d-spacings are easily calculated from its unit cell size, only a single column of hkl reflections are tabulated.
The habit of gersdorffite given by Palache et al. (1944) is octahedral, cubo•octahedral, pyritohedral, lamellar and massive granular. From polished section studies, evidence by Ramdohr (1950) indicates a possible structure inversion and a non-isometric character. This was confirmed by Klemm (1962), who noted areas of weak anisotropic colours in 8 of his
14 samples, mainly as twin lamellae after (100) and(lll)? (possibly (311) which is similar to the poor cleavage (311) in pyrite described by Frenzel and Boss (1967)), often as zone structure, and occasionally as grains.
Olemistry
Some idea of the variations in the chemical composition of gersdorf fite are given in 52 literature values (figure 2, page 16) and 62 electron 17.
probe analyses by Klemm (1965), 28 electron probe analyses by Klemm
and Weiser (1965), and six chemical analyses and the experimental ternary
diagram Ni-As•S by Yund (1962). No evidence is available from the electron
probe analyses, the chemical analyses and the ternary diagram to indicate a deviation from the stoichiometric ratio Ni :(As + S) = 1 :2. The apparent deviations of Ni:~s + S) from 0.9:2.0 to 1.4:2.0 in the literature survey may
be caused by impure samples or analytical errors. The ratio of As:S from
1.2:0.9 to 0.8:1.2 in the electron probe results and the even wider range from 1.6:0.4 to 0.8:1.2 in the literature survey lie within the range from
1.80:0.20 to 0. 77:1.23 given in the unit cell size diagram (figure 3) of
Yund (1962). These data in addition indicate the majority of the samples have a similar composition to the pure end member. Chemical evidence for zoning was first indicated by Goll (1937), and strongly zoned crystals were detected with electron probe analysis by Lawrence and Markham (1963).
An iron variety of gersdorffite, (Ni, Fe)AsS, is named plessite by
Harcourt (1942). The substitution in gersdorffite of arsenic by antimony
to the extent of Ni(As0 _8sb0 _2)S is designated by Palache et al. (1944) as an
intermediate member in a probable gersdorffite·ullmannite series with a
varietal name of corynite. There is in addition a variety of doubtful validity
called wolfachite with a composition similar to corynite (from an analysis
in 1869), but its crystal structure resembles arsenopyrite. This variety
quoted by Strunz (1957) as an intermediate between gersdorffite and ullmannite is deleted from the later edition of Strunz (1966), whereas Berry and Thompson (1962) express its close similarity to cobaltite from Hakansbo but state reflections 100 and 110 are absent in the powder pattern.
Synthetic gersdorffite as shown in the Ni-As-S system of Yund (1962) was formed from compositions varying from NiAs l. 77s0 . 23 to NiAs0 . 77s1 . 23 at 700°C (figure 3, page 17). The maximum arsenic-rich limit in gersdorffite of NiAsl. 80s0 . 20 is reached at 660°C, which decreases to NiAsl. 72s0 _28 at
450°C to show a nearly vertical solvus in this temperature interval. In the experimental ternary diagram NiAsS-CoAsS-FeAsS (figure 2, page 16) of
Klemm (1965), substitution of nickel in gersdorffite by up to 35 per cent cobalt or by up to 25 per cent iron occurs at 400°C, and at 600°C nickel may be substituted completely by cobalt or by up to 55 per cent iron. Gersdorf fite is considered by Palache !:.t ~- (1944) to be a comparatively rare mineral which occurs in vein deposits with other nickel and sulphide minerals.
COBALTITE
Cr:xstallography
From Laue photographs of cobaltite (CoAsS), the space group P213 was proposed for cobaltite by Mechling (I 921). This apparent isometric character was confirmed by Ramsdell (1925) with powder photographs. It is described as pseudo-cubic orthorhombic by de Jong (1926, 1928) with cubic 19. pseudo space group Pa3. This space group is also indicated by Peacock and Henry (1948) on the basis of intensity measurements from powder photographs. The space group P2 13 was determined also by Bokii and
Tsinokev (1954), who suggest a continuous transition between the two space groups of P2 13 and Pa3.
With intensity data from Weissenberg photographs, Onorato
(1957a, 1957b) proposed the monoclinic space group P21/c, after elimin ation of the orthorhombic space groups Pnc2, Pmna, Pca21, and Pbcm with which he was unable to produce a crystal structure model with a pyrite-type structure. Space group Pca21 was proposed by Winterberger
(1962). No firm conclusion is offered by Oftedal (1963) for the space group.
A nickel-rich variety is described from X-ray diffraction powder data by
Polushkina and Sidorenko (1963) in space group Pnnm, similar to marcasite; but a critical review by Shishkin (1965) shows this mineral to be a high cobalt glaucodot.
Space group Pca21 was confirmed by Giese and Kerr (1965). Their cobaltite crystal structure, after unstated transformation (probably 0,1/4,0) to coincide with a pyrite-type model, has cobalt at (-0.010, 0.006, 0.011), arsenic at (0.380, 0.381, 0.383), and sulphur at (-0.380, 0.380(sic), -0.382)
to give a discrepancy factor R of 0.10 for 41 hkO reflections and O. 12 for
21 hOl reflections. After they heated material from the same locality to 20.
800-850°c for 2 days, the space group changed to Pa3 with cobalt at (0, 0, 0), and arsenic and sulphur equally distributed over (0. 380, 0. 380, 0. 380) to give a 0. 05 7 discrepancy factor R for 21 hkO reflections. The atomic radii calculated from the disordered structure are 1. 16R for cobalt and
I. 16R for the average of arsenic plus sulphur, whereas the atomic radii calculated from the ordered structure are 1.17R for cobalt, 1. 18R for arsenic and 1. 12.R for sulphur.
The unit cell size variation with composition is indicated in the
FeAsS-CoAsS-NiAsS ternary diagram (figure 2, page 16) of Klemm (1965).
A compilation from the literature of 23 unit cell sizes for cobaltite and their geographical locations are given in table le. X-ray diffraction powder intensities from five of these cobaltites are given in table 2b, but since a cubic mineral is easily indexed, and its d•spacings easily calculated from its unit cell size, only a single column of hkl reflections are tabulated.
The habit of cobaltite is stated by Palache ~t ~- (1944) as commonly cubic, pyritohedral, occasionally octahedral, and sometimes as granular massive to compact. Twinning is rare and when present occurs on (011) and (111). Cobaltite was found by Schneiderhohn (1922) to be optically anisotropic in polarized light under a reflecting microscope. This optical anisotropy is suggested by Ramsdell (1925) to result from either defor mation twinning caused by polishing the specimen or the pseudo-isometric character of cobaltite. This optical anisotropy is considered by Ramdohr 21.
(1960) to vary with the extent of deviation from cubic symmetry. Klemm
(1962) did not observe a single optical isotropic cobaltite in his 10 samples but only optical anisotropic zones, lamellae and grains. He also noted that the weak optical anisotropic colours in the newly polished sections became intensified in the older repolished sections. This agrees with the suggestion of Gibbons (1967) that optical anisotropy in pyrite is caused by polishing. When cobaltite was heated above 850°C and quenched by
Florke (1926), it irreversibly changed to an optically isotropic form.
Chemistry
The 18 cobaltite chemical analyses quoted in a literature survey by Klemm (1965) and shown in figure 2 (page 16), indicate that the ratio of Co:(As + S) varies from 0.98:2.0 to 1.2:2.0, but his 40 electron probe results indicate no deviation from the 1:2 stoichiometric ratio. Similarly his literature survey indicates that the ratio of As:S varies from 1.05:0.95 to 0.9:1.1, but his electron probe results indicate no deviation from the stoichiometric 1:1 ratio. These electron probe analyses also show cobalt substitution by iron of up to (Co0 _72 Fe0 _28)AsS and by nickel of up to
(Coo.s6Feo.o6Ni0 _08)AsS. The literature survey values also lie within this range except one at (Co0 _65Fe0 _14Ni0 _21)AsS, however Klemm and
Weiser (1965) indicate a range of up to (Co0 _4Fe0 _2Ni0 _4)AsS with electron probe results. 22.
In the experimental ternary diagram CoAsS•NiAsS•FeAsS (figure 2, page 16) of Klemm (1965), substitution for cobalt in cobaltite by up to 15 per cent nickel or by up to 20 per cent iron occurs at 400°c, and at 600°C cobalt may be substituted completely by nickel or by up to 40 per cent iron. Cobal· tite commonly occurs (Palache ~ !l • 1944) disseminated in metamorphic rocks and more rarely in veins with other cobalt and nickel sulphides and arsenides in high temperature deposits.
ULLMANNITE
Crystallography
The space group of ullmannite (NiSbS) was determined as P213 by
Ramsdell (1925), Peacock and Henry (1948), and Bokii and Tsinokev (1954).
The crystal structure proposed by Peacock and Henry (1948) with nickel at
(0, 0, 0), sulphur at (-0.385, •0.385, -0.385) and antimony at (0.385,
0.385, 0.385) was refined from 40 hkO reflections by Takeuchi (1957) with the sulphur and antimony positions exchanged to give nickel at (-0. 024,
•0.024, •0.024), sulphur at (0.390, 0.390, 0.390) and antimony at (-0.375,
2 •0.375, -0.375) with an overall temperature factor of 0.8.R and a dis• crepancy factor R of O. l15. Therefore the sulphur and antimony atoms in this crystal structure are completely ordered. The interatomic distances calculated from this crystal structure are 2. 57R for Ni•Sb, 2.40R for Sb•S, and 2.34R for Ni-S; which give an atomic radius for nickel of I.26R, for 23. antimony of 1. 32R, and for sulphur of I. osR.
A compilation from the literature of 15 unit cell sizes for ullmannite
(with varietal names if applicable) and their geographical locations are given in table le. X-ray diffraction powder intensities for three of these ullman - nites are given in table 2c, but since a cubic mineral is easily indexed and its d-spacings easily calculated from its unit cell size, only a single column of hkl reflections are tabulated.
The habit for ullmannite is given by Palache et al. (I 944) as mainly cubic, and less frequently as octahedral and pyritohedral. Penetration twins of irregular cubic form and re-entrants have been recorded on the cub& edges. Areas of weak optically anisotropic colours are noted by
Klemm (1962) in seven out of his eight samples, mainly as twin lamellae after (100) and (111)?, frequently as zone structure, and occasionally as grains.
Olemistry
From the three chemical analyses of ullmannite quoted by Palache et al. (1944), no evidence is available to indicate a deviation from the stoichiometric ratio Ni:Sb:S = 1:1 :I. These data show the substitution of nickel by iron to be in negligible amounts, and the substitution of nickel by cobalt which gives the composition Ni0 _5co0 _5SbS for the variety willyamite.
The substitution of antimony by arsenic occurs up to the composition 24.
NiSb As S to give the variety corynite, whereas substitution of 0 . 7 o. 3 antimony by bismuth occurs up to composition NiSb0 _8 Bi0 _1 As0 _1s to give the variety kallilite. The ternary diagram Ni-Sb•S of Lange and Schlegel
(1951) also indicates a stoichiometric ratio of Ni:Sb:S = I :I :I. Ullmannite occurs in veins with other nickel minerals.
ARSENOPYRITE
Crystallography
The arsenopyrite (FeAsS) cell was designated by Huggins (1922) and de Jong (1926, 1928) as orthorhombic Cmmm (first choice). Further investigation by .&terger (1936a, 1936b, 1936c) indicated firstly that no orthorhombic space group satisfied the X•ray diffraction intensity data, and secondly that the apparent orthorhombic symmetry when viewed under the reflecting microscope was found to be caused by twinned composites.
He then derived the arsenopyrite structure from a superstructure based upon the marcasite crystal structure. The first choice is transformed by
001/100/010 into monoclinic 82 1/d space group (second choice) or possibly into triclinic PI for common arsenopyrite. After a crystal structure analysis of gudmundite, an arsenopyrite group member, .&terger (1939) suggested some qualitative corrections to the published arsenopyrite para• meter values. A refinement by Bonnemere and Winterberger (1961) with
109 reflections gave a 0.30 discrepancy factor R. 25.
The unit cell was changed by Morimoto and Oark (1961) on the model suggested by &.lerger (1939), Ramdohr (1954) and Donnay and Donnay (1963) with transformation 1/~D,1/2 /0,1,,0/ 1/2,D,I/2 into monoclinic P2 1/c space group (third choice) for arsenic-rich varieties and into triclinic PI for common arsenopyrite. A crystal structure refinement moves iron to
(0.272, 0.000, 0.289) and (0.275, 0.502, O. 787), sulphur to (0.346,
•0.370, 0.175) and (0.343, -0.131, 0.675), and arsenic to (0.154, 0.371,
0.363) and (0.155, 0.129, 0.863) to give a 0.29 discrepancy factor R with
172 hO 1 reflections and a O. 20 d; screpancy factor R with 135 hkO reflections.
Further structure refinement was impossible, because the relative effects of twinning and partial sulphur•arsenic disorder could not be separated.
The interatomic distances measured from this structure are 2. 35R. for
Fe-As, 2 .25.R for Fe•S and 2. 33.R for As•S, which give an atomic radius of 1. 14R for iron, I. 21.R for arsenic and I. 11.R for sulphur.
A literature compilation of 14 unit cell sizes for arsenopyrite(with vartietal names if applicable)and their geographical locations are given in table lg. One X•ray diffraction powder intensity set together with the pseudo-cubic monoclinic hkl indices and d•spacings of Morimoto and
Oark (1961) are given in table 2d.
The habit of arsenopyrite is given by Palache ~!!· (1944) as prismatic, columnar, granular or compact. Twinning is recorded on 26.
(100) and (001), which produces pseudo-orthorhombic crystals, on (101) as contact or penetration twins, and on (102) as cruciform twins or star shaped trillings. A crystal from Freiberg (Morimoto and Clark, 1961) has
(101) twinning on a large scale and (010) twinning on a fine scale. The optical anisotropism of arsenopyrite is strong.
Chemistry
The chemical composition of arsenopyrite has been considered from
16 chemical analyses by Morimoto and Clark (1961); a literature survey of
52 samples including 18 cobalt-rich samples (figure 2, page 16) and also by
Klemm (1965) 13 electron probe results; and 214 electron probe results by
Clark (1965). All data indicate a stoichiometric ratio of Fe:(As + S) = 1:2 and that any apparent deviation probably results from analytical errors.
The widest As:S ratio varies from 1.2:0.8 to 0. 9:1.1 as given in the literature survey, which agrees with the synthetic study of Clark (1960).
This literature survey also shows the substitution of iron by nickel to be in negligible amounts, and the substitution of iron by cobalt up to composition
(F e0 . 7 Co O• 3)AsS to give the variety danaite. This agrees with Palache ~ .!!_.
(1944), who in addition indicate substitution of arsenic by bismuth in minor quantities and by antimony to a lesser extent.
Variation of the arsenopyrite As:S ratio with temperature is des cribed by Clark (1960). On any isotherm the arsenopyrite composition 27. range is small, for example from FeAsl.08s0 _92 to FeAsl.05s0 _95 at
600°C; however, the composition range is large with temperature variation, for example from FeAs0 _95s1_05 at 300°C to FeAsi. 14s0 •86 at 702°c.
Arsenic-rich arsenopyrites are stable above 400°C; sulphur-rich arseno• pyrites are stable below 500°C, however this limit for sulphur-rich arseno• pyrite rises to 650°c under 2,000 bars confining pressure. The results of both Klemm (1965) and Morimoto and Clark (1961) indicate that sulphur-rich arsenopyrites predominate. Rather than being formed at a low temperature, the predominance of sulphur-rich arsenopyrites may result from the con tamination of the chemically analyzed samples by their associated sulphide
minerals. In the experimental ternary diagram FeAsS•CoAsS 00NiAsS
(figure 2, page 16) of Klemm (1965), iron in arsenopyrite is substituted by cobalt up to 10 per cent or by nickel up to 5 per cent at 400°c, and at 600°C iron is substituted by cobalt up to 20 per cent or by nickel up to 10 per cent.
Arsenopyrite is considered by Palache ~ ~. (1944) to be the most abundant and widespread arsenic mineral. It usually forms at an early stage in diverse types of mineral deposits.
GLAUCODOT
C£tstallography
The unit cell data in table lg of glaucodot, (Co, Fe)AsS, from
Hakansbo was determined by de Jong (1926) and then by Ferguson (1947) 28. on the ortho9'ombic space group Cmmm, which is similar to the ortho rk>mbic space group of arsenopyrite after triplication of the b-axis. The indexed powder data in table 2d of glaucodot and the unit cell data in table lg and the indexed powder data in table 2d of alloclasite were determined by
Berry and Thompson (1962). Klemm (1965) recently determined the unit cell size given in table lg of glaucodot from Hakansbo on the monoclinic space group P2/c, the same space group used by Morimoto and Clark
(1961) for arsenopyrite.
The habit for glaucodot is given by Palache et al. (1944) as prismatic commonly with striations or massive. Glaucodot has occasional twinning on (101) and (102), and less optical anisotropy than arsenopyrite.
Olemistry
The 14 glaucodot chemical analyses in a literature survey (figure 2, page 16) by Klemm (1965) and his eight electron probe results indicate an atomic ratio of (Fe+ Co):(As + S) = 1:2. The As:S ratio varies from
1.1:0.9 to 0.8:1.2 (typographical error assumed in original text) in the literature survey, but no deviation from the stoichiometric 1:1 is indicated by the electron probe results. The metal atom composition range of glaucodot from the electron probe results is Fe0 _83co0 _1 7 to
Fe0 _52eo0 _46 Ni0 _02 ; but since all nine samples come from Hakansbo, the composition gap between glaucodot and cobaltite may result from lack 29. of samples. Only one sample is more iron-rich than Fe0 _64co0 _36 at
Fe0 _83eo0 _1 7, which may be designated as danaite. The metal atom com position range of glaucodot indicated by the literature survey is F e0 • 55 eo0 • 45
(typographical error assumed in original text) to F e0 . 3 Co0 . 7, which indicates no compositional gap between glaucodot and cobaltite but a large compos itional gap between glaucodot and danaite. The compositional limits of glaucodot are defined by Palache ~t.!!_.(1944) as Fe0 _67 Co0 _33 to
Fe0 _1 4co0 _86 • These include the cobalt-rich variety of alloclasite, which lies well within the cobaltite compositional limits. From this data, no firm conclusion is reached on the composition gap between glaucodot and danaite proposed by Ferguson (1947). These data in addition show negligible sub stitution of cobalt and iron by nickel.
Since his electron probe analyses of glaucodot lie within the mis cibility gap in his FeAsS-CoAsS-NiAsS ternary diagram (figure 2, page 16),
Klemm (1965) proposes that glaucodot is metastable. Differential thermal analyses of glaucodot up to 650°C under a nitrogen atmosphere did not indicate a high temperature modification of the crystal structure. Glaucodot generally occurs with cobaltite.
GUDMUNDITE
Crystallography
The space group of gudrnundite (FeSbS) was first designated as 30. orthorhombic Cmmm and then altered to monoclinic 821 /d by Buerger
(1936a, 1936b, 1937a, 1939) and finally changed to monoclinic P2 1/c. The atomic co-ordinates based upon monoclinic P21/c are iron at (-0.300,
0.015, 0. 300), antimony at (0.148, 0.130, 0.132) and sulphur at (0. 355,
0.144, 0.666); but no temperature factors or a discrepancy factor Rare given. The interatomic distances are 2.24R for Fe-S, 2.57.R. for Fe-Sb and 2.61R for S-Sb; which give an atomic radius of 1. 10.R for iron, 1.47.R for antimony and 1.13.R for sulphur. Data for the unit cells of three gudmundites are given in table lg and one set of X-ray diffraction powder intensities with d-spacings are given in table 2d.
The habit of gudmundite is given by Palache ~al. (1944) as pris matic. Penetration and contact twins occur on (101), but they have not yet been demonstrated to produce pseudo-orthorhombic habit similar to arseno pyrite. The optical anisotropism is of medium strength.
Chemistry
The composition of gudmundite given by Palache ~ al. (1944) from only one analysis indicates no deviation from the stoichiometric ratio of
Ft:Sb:S = 1:1:1. A trace of nickel substitutes for iron, but no substitution for antimony or sulphur is given.
An investigation into the Fe-Sb-S ternary system by Lange and
Schlegel (1951) did not produce gudmundite above 200°C, which agrees 31. with the results of Urazov !:!_al. (1960) and Markham (1965). Natural gudmundite was found by Clark (1966) to decompose between 270°C and
295°c into pyrrhotite and antimony. A low formation temperature for gudmundite is indicated by Palache ~ al. (1944) who state that it is "a hydrothermal mineral formed at a relatively late period in sulfide deposits".
LITERATIJRE REVIEW
After a careful appraisal of the preceding literature survey, there seem to be a number of inconsistencies and omissions in the data, which appear worthy of investigation. These problems include (1) a single crystal structure analysis of gersdorffite to determine the atomic co• ordinates and their temperature factors, the octahedral and tetrahedral angles, and the interatomic distances and their derived atomic radii;
(2) a refinement of the single crystal structure analysis by Giese and Kerr
(1965) for cobaltite to determine the temperature factors, and the octahedral and tetrahedral angles; and (3) a refinement of the single crystal structure analysis by Takeuchi (1957) for ullmannite to determine the individual temperature factors, and the octahedral and tetrahedral angles.
Further unit cell size values of natural gersdorffite, cobaltite and ullmannite are needed to indicate their unit cell size range. These natural samples are also needed to ascertain the variation of optical anisotropy, 32. compositional zoning, X-ray diffraction power intensities, and paragenesis.
Heating experiments are needed using these samples to determine the stability of ordered and distorted structures. Heating experiments are also needed with glaucodot to determine its stability.
Preparation of synthetic minerals is required to define (I) the limits of arsenic and sulphur in cobaltite; (2) the limits of arsenic, sulphur and antimony in ullmannite; (3) the substitution limits of cobalt and iron for nickel at various arsenic to suphur ratios in gersdorffite; (4) the sub stitution limits of nickel and iron for cobalt at various aresenic to sulphur ratios in cobaltite; and (5) the substitution limits of cobalt and iron for nickel at various concentrations of antimony, arsenic and sulphur in ullmannite.
All the problems outlined above were investigated. The methods used to investigate these problems, the results obtained, and the discussion of these results are given in the following sections. 33.
METHODS
SYNTHESIS
The chemicals from Johnson, Matthey & Co., Limited used in the syn thesis of gersdorffite and its related minerals were antimony-JM660, arsenic
JM640, cobalt oxide-JM875, iron oxide-JM850, nickel oxide-JM895, and sulphur-JM775. A spectrographic examination supplied by the company indi cated that most chemical elements specifically sought as impurities were either not present or were below the limits of detection by this experimental procedure. The exceptions are given in table 4a. The sulphur content in arsenic was determined by a chemical method.
The metal oxide powder was reduced in a I cm diameter quartz tube by hydrogen flowing through the tube, which was heated to red heat with a meeker burner. Dry panning (tapping) was used to reject large metal par ticles because of their slow reaction rates, and very fine metal particles be cause of transfer difficulties and their faster oxidation rate. The oxidation rate of fine particles is negligible, for example a cobalt powder left exposed to the air by Roseboom (1962) showed an increase in weight of only 0.01 per cent after one month. The metal oxide powder was reduced on the day to be used.
Sufficient sulphur for use during the day was taken from the bottle and crushed small enough to pass into a sample tube. The fines were discarded because of transfer difficulties, but the large sulphur particles were 34. used since they react quickly because of their low melting point. Antimony was treated similarly to sulphur.
Special precautions were taken when using arsenic to reduce its atmospheric contact time because of its fast oxidation. For example, arsenic with a maximum size of 1 mm was found by Roseboom (1962) to gain weight in air at 0.1 per cent per hour. The arsenic, which was sup plied in a large evacuated sealed container, was opened and quickly re sealed with a minimum of crushing in 2 gm. lots under 0. 2mm of mercury vacuum. The arsenic was taken from the sealed container for use in the mixtures within an hour. It was crushed small enough to pass into a sample tube, and the fines were discarded because of transfer difficulties and their fast oxidation rate. The large arsenic particles were used since they react quickly because of their low melting point.
The quantity of each chemical element in the sample was calculated from its atomicity multiplied by its atomic weight in mg. This method was used to avoid errors, since it involved little calculation and the resultant sample of approximately 0.2g. was also sufficient to cover an X-ray dif• fraction powder slide. Each chemical element was weighed to the nearest
0.0001g. on a mettler balance. For the volatile chemical elements of sulphur, arsenic and antimony, an extra 0.0001 g. was added to compen sate for the vapour from each chemical element that would fill the space 35. above the sample. The powders were stirred together and although the mixture was not homogenous, it was not crushed to avoid transfer losses.
The powder sample was transferred via a glass filter funnel into a glass sample tube, both of which had been oven dried at 100°c in order to eliminate transfer losses, as a powder may adhere to a moist glass surface.
A sample tube for the simple tube method was made from pyrex glass tubing of 3 I/2mm internal diameter by 3/4mm wall thickness. The glass tubing was heated by an oxygen-coal gas flame and pulled into a Imm neck at one end of the tubing and then sealed at the other end. Silica glass tubes of similar shape formed under heat from an oxygen -hydrogen flame were used for samples heated above 550°C. Each sample tube was sealed under
0.2mm of mercury vacuum across the neck with the 50 cubic mm volume more than half filled with a sample. Since the sample tube was constructed to leave a minimum space above the sample, a glass rod as used by Kullerud and Yund (1962) was not needed to reduce this space. Natural material was also sealed under vacuum in similar tubes to determine temperatures of crystal structure change and decomposition. Siderite-contaminated natural material was first heated to soo 0 c, and then the sample tube was opened to release the carbon dioxide pressure, and next resealed under vacuum before heating to the required temperature.
Synthetic compounds were also prepared by the LiCI-KCI melt 36. method of Klemm (1965). The eutectic point of this melt occurs at 358°c with 42 mol per cent Li Cl and 58 mol per cent KC!. Special care was needed to keep the extremely deliquescent LiCl dry. One gram of the Li Cl-KC! eutectic mixture was placed inside a sample tube made from pyrex glass tubing of 12mm internal diameter by 1 I/2mm wall thickness. A 5 per cent non-metal atom excess was added to each sample for the determination of metal atom substitution. Each tube was sealed under vacuum with the 4cc volume more than three-quarters filled with a sample.
Each tube was stacked into a metal block holder and then placed in a vertical tube furnace, whose 6 cm hot zone lies within.± 5°C at 550°C. The furnace temperature was controlled either by a hand variable transformer to within+- 20°c or a temperature controller to within+- 3°c. Each tube was heated for a period of up to nine months. The approach to equilibrium was faster with (1) higher temperatures, (2) a non-metal atom excess,
(3) a LiCI-KCl melt, (4) nickel compounds compared to iron compounds, and with (5) simple mixtures compared to complex mixtures. After each tube was opened the product was crushed, rut the loss due to this crushing had negligible effect since the product was homogenous.
Synthetic compounds were also prepared by the vacuum distillation method of Klemm ( 1965). This method is similar to the transportation, via the gas phase, of solids under a temperature gradient; when volatile products 37. are formed by heterogeneous reversible reactions as described by Schafer
(1961), and later more rigorously theoretically treated by Mendel (1962).
An example of crystal growth is the needle-shaped arsenopyrite crystals formed on a fresh arsenopyrite surface by Morimoto and Clark (1961).
In this case the arsenopyrite crystals grew in a 3S0°C environment, where the decomposition of a little arsenopyrite and arsenic, originally inter grown at 600°C, filled a long sealed tube with vapour. In the vacuum dis tillation method used, each sample was sealed under vacuum in a sample tube made from a 30 cm length of pyrex tubing with a 12mm internal diameter by 1 1/2 mm wall thickness. Each tube was placed in a horizontal furnace with one end in the centre of the furnace at sso0 c and the other end outside the furnace at room temperature to create a temperature gradient.
The sample placed at the hot end migrated along the tube and its deposition in a cooler environment led to crystal growth.
A hydrothermal method used involved an exceedingly thick walled pyrex tube, which was filled with a sample and 20 per cent water. The maximum temperature used was 200°c. The tube was covered with gauze to prevent furnace damage in case of explosion.
X-RAY DIFFRACTION OF SYNTHETIC POWDERS
The synthetic powder was first analysed with a diffractometer, which consists of a Philips X-ray diffraction generator PW1010 (full wave rectification), a goniometer PWIOS0, and an electronic circuit panel with 38.
recorder PWI051. The generator operated at !KW (50KV and 20ma) with a
copper tube, even though samples containing iron and cobalt fluoresce and
give a high background count. Later this disadvantage was eliminated with
a !KW cobalt tube. The goniometer scanned from 14 to 64°29 (Bragg
angle) at 1/8° per minute over a specimen 2 cm long by 1 cm wide and 0. 15 mm
deep with the reflected radiation set to pass slits of 1°, 0.1 mm, 1°, and a
nickel filter. The electronic panel settings were rate meter of 16, multi-
plier of 1, and time constant of 16 seconds; and the chart speed of the
recorder was 1 inch per degree 29. These values were changed to a scanning
range from 16 to 74°29 at 1/2° per minute, time constant of 4 seconds, and an iron filter with a cobalt tube. These values give the most suitable
resolution, intensity, accuracy and operation time, after consultation with
Klug and Alexander (1959).
The resultant X-ray diffraction powder pattern was used to check for
each N value (h2 + k2 + i2) in the synthesized mineral and also to identify
other extraneous phases. The X-ray diffraction powder patterns used to
identify these ore minerals are listed in the extensive compilations by
Hanawalt ~t .!!_. (1938), by Harcourt (1942), in the Peacock Atlas by Berry
and Thompson (1962), and in the X-ray Diffraction Powder Data File issued
and reviewed annually by the American Society for Testing and Materials.
The unit cell size (a) of the synthetic cubic mineral was determined 39. with a Philips 11704 unit (half wave rectification), which operated at 350W
(37KV and 14ma) with cobalt radiation and an iron filter. From the powder specimen plus 10 per cent gum trajacanth and water, a 0. 2mm diameter cylindrical specimen rod was prepared. It was mounted in a 114.6mm diameter Debye-Scherrer camera with a 0. 5mm diameter inlet collimator and exposed for 15 hours to record on an Ilford Industrial G X-ray film at room temperature of 20 ±. 3°C. After this film was developed, the back angle reflections (360-49) were measured. The unit cell size (a= /N~/2sin8) was calculated for each reflection using the wavelength (A) of Bragg (1947) and the table of Parrish~ al. (1953). These values were plotted against the function l/2(cos29/sin9 + cos28/9) of Nelson and Riley (1945) and the value was extrapolated to 8 = 90°, which gave the unit cell size corrected for film shrinkage and minimized absorption errors. The probable error for the unit cell size is estimated from the extrapolation curve. These unit cell sizes were compared to those in the extensive compilation by Donnay and Donnay
(1963).
The unit cell size accuracy limit given by Ekstein and Segel (1949), due to the X-ray beam spectral width, is 0. 0005 per cent, whereas a significantly lower agreement value of 0.01 per cent was calculated from the results of the International Crystallography Union precision lattice parameter project collated by Parrish (1960). Materials used in the
International Crystallography project were obtained from Dr. Parrish and 40. a unit cell size of 5.4302 :!: 0.0003R was measured for silicon compared to
5. 4305f:!: 0.00017R in the project, and a unit cell size of 3. 1650 ± 0. 0002.R was measured for tungsten compared to 3. 16522 :!: 0. 00009.K in the project.
Consideration was given to the errors given by Cohen (1936) and Straumanis
(1960) of camera radius, sample eccentricity, sample deformation by setting of the bonding material, and film hole distortion when the holes are punched.
Neither the refraction correction of Straumanis (1955), nor the absorption correction of Straumanis (1959) was used. The temperature expansion was not taken into account, although a slight error was introduced as shown by average temperature expansion coefficients of 10-5 /°C calculated for some
AX2 compounds by Straumanis et.!!_. (1964) and Chrystall (1965). A summary for the attainment of high precision lattice dimensions is given by Nuffield
(1966).
The powders of the synthetic cubic minerals were used in crystal structure analysis. Little preferred orientation occurred when the synthetic powder was packed into the sample holder because of its fine particle size and sintered nature. Preferred orientation is induced during sample packing by perfect cleavage planes. It is reduced by crushing the powder down to less than 2 microns in particle size (Klug and Alexander,
1959), and by packing through the rear of the sample holder (McCeery,
1949) or the side of the sample holder (Niskanen, 1964). 41.
The X-ray diffraction powder patterns were recorded up to 160°29, which included N up to 48. The collimator slits 1°, 0.1mm, and 1° were used up to 12°20 and then they were changed to the wider slits 4°,- 0.2mm and 4°; since all the radiation from these wider slits falls upon the 2 cm long specimen above the 72°29 angle. The two slit sizes were correlated by measurement of several reflections through both slit sizes. A small time constant of 4 seconds was used to give more reproducable peaks, and a fast chart speed of 4 inches per degree 29 was used to give greater accuracy.
Each peak was cut out from the X-ray diffraction powder pattern and weighed to give its peak intensity. The variation in paper weight was allowed for by the division of the peak intensity by the weight of paper per square inch, which was cut from beneath the peak. No correction was made for the small dead time characteristics of the proportional counter, which is quoted by Philips Electrical as 2 micro-seconds. The intensity was subjected to asymmetric Lorenz and polarization corrections taken from Henry~ al. (1961) and to a multiplicity correction taken from Klug and Alexander (1959). Its square root gave the observed structure factor (Fo). Only 16 reflections up to N = 48 are contributed by one set of structure factor planes, since in the space groups Pa3 and P2 13 (hkOa (khO).
In addition unobserved reflections could be individually indexed with zero intensity. No correction was made for absorption or extinction. 42.
NA1URAL POWDER EXAMINATION
Some natural materials donated by museums were gently crushed.
Pure material was selected under a binocular microscope for X-ray diffractometer powder analysis, similar to the analysis of synthetic powder on page 37, to check for the presence of all reflections and also to identify other extraneous minerals. The reflection intensities of N up to 11 for cubic minerals were taken from the peak heights on an X-ray diffractometer powder pattern with the most intense reflection as 100 and all the other reflection intensities proportional to this most intense reflection. For non cubic minerals such as monoclinic glaucodot, the unit cell size was deter mined by a trial and error procedure. Significant preferred orientation errors may occur because of perfect cleavage and large crystal sizes.
The reflection shape indicates the crystallinity of the powder.
Distinct Ktt-1 and Kd'2 reflections in the 60°28 region of the X-ray diffraction powder pattern indicate well crystallized samples, which allow the deter mination of the unit cell size to 0.01 per cent probable error by the Debye
Scherrer method, similar to the method of synthetic powder analysis on page 38. Visible Kd-1 and Kd-2 reflections in the 60°29 region of the X-ray diffraction powder pattern indicate average crystallized samples, which allow determination of the unit cell size to O. 02 per cent probable error by the De bye-Scherrer method. Non separable Kl 1 and Kcl- 2 reflections in the
60°28 region of the X-ray diffraction powder pattern indicate poorly 43. crystallized samples, which allow determination of the unit cell size to only 0.2 per cent probable error with an X-ray diffractometer by measure ment of the reflections at two thirds of their height to the nearest 0.01° for each of three oscillations, as per Chayes and MacKenzie (1957).
Distinct splitting of reflection peaks indicates two components of similar compositions. Their percentages were estimated from relative peak heights and their unit cell sizes were calculated from peak positions. Only a single list of reflection peak heights is tabulated, since the overlapping reflection peaks are difficult to separate especially at low 29 angles.
Polished sections of natural materials were prepared from several large chips mounted in plastic. The sections were finally polished in a vibrating machine for 12 hours with 1/2 micron alumina polishing powder.
The identity of these samples was confirmed by the reflected light proper ties of shape, colour, relative reflectance, polishing hardness, reflection pleochroism, polarization colours, extinction, texture relations and cleavage. Special attention was given to optical anisotropism and compos itional zoning.
SINGLE CRYSTAL X-RAY DIFFRACTION
Since single crystals of the relatively complex AXY compounds are extremely difficult to grow by the methods reviewed by Van Hook (1961), which are hydrothermal, from a melt, from a solution or from a vapour; natural materials donated by museums were used. From the crushed 44. natural material, a fragment less than 50 microns with some crystal faces of a cube was selected, because the small size reduces X-ray absorption and extinction errors and also the small size and cubic habit decreases the pro bability of multiple or twinned crystals. This crystal was attached with silicone grease to a glass fibre, which was then mounted in an unknown position with plasticine on to a goniometer head.
Next the crystal had to be aligned with one crystallographic axis horizontal to the camera base and at right angles to the X-ray beam.
Since the crystal was too small to align by optical methods, a 20° oscillation photograph of 2 hours exposure was taken with unfiltered molybdenum radiation using a PW1010 generator and a Nonius Weissenberg camera with a 0. 5mm inlet collimator. This photograph was used to locate a crystallo graphic axis, but if an axis could not be found then another 20° oscillation photograph was taken after the crystal had been rotated 20°. From this axis the two rotation angles as shown by Henry et al(l961) were measured to determine the adjustments to the arcs of the goniometer head. This cor rection was repeated till no further adjustment was required. From this final oscillation photograph a unit cell edge was calculated. With these crystal settings, a zero layer Weissenberg photograph over 180° was taken with unfiltered radiation for 20 hours. This photograph was used to measure the other two unit cell edges. These three unit cell edges are sufficient data to confirm the identity of the crystal. If a multiple crystal was detected 45. during this experiment, it was discarded and another crystal was selected.
Intensity data, after 160 hours exposure of the single crystal to nickel filtered copper radiation, were obtained from the zero layer
Weissenberg photographs. A series of Weissenberg photographs with de creasing intensity was obtained by packing four films into the casket during the exposure, since a film substantially absorbs the soft copper radiation to give a 3:1 interfilm density ratio from one film to the film below. The harder molybdenum radiation was not used, because it is difficult to insert the aluminum sheet foil needed between the films to reduce the radiation intensity which passes through to the film below. Molybdenum radiation is generally preferred to copper radiation because it reduces the absorption coefficient by up to an order of magnitude as shown in the wollastonite
(CaSi03) example by Buerger (1960). The linear absorption coefficient for gersdorffite of 50mm -l with copper radiation is less than double the 30mm -l with molybdenum radiation. The series of intensity photographs was required, because spot blackness is proportional to the spot intensity at low intensities only. Therefore the weak reflection spots were measured on the strong top photograph, whereas the stronger reflection spots were measured on a weaker lower photograph. The spot intensity was found by visual comparison with a series of spots obtained under known conditions (calibrated comparison strip). 46..
The equi -inclination angle ( µ) of the Weissenberg camera for each
upper reflection layer (n) was calculated from the formula JI = n/2a, and the
position (d) of the screen with the radius (r) needed to eliminate reflections
from the other layers was calculated with the formula d = rtan p. Occasion
ally an oscillation photograph was taken to check that the layer line fell
centrally in the screen split. Data were usually collected up to the fourth
layer with a Weissenberg camera.
The reciprocal lattice of each crystal was studied from two pre
cession photographs. The Buerger precession camera was fitted with the
single crystal mounted on to a goniometer head. For setting the crystal, a 10° precession photograph of 4 hours exposure was taken with unfiltered
molybdenum radiation using a 1010 generator. The photograph recorded only
the zero layer by the elimination of the upper layers with a suitable screen
setting. From this photograph the adjustments were calculated for each of
the three arcs; and if large adjustments were made, a further setting photo
graph was taken. With these corrected settings, a 25° zero layer precession
photograph (hOI layer) was exposed for 100 hours to unfiltered molybdenum
radiation; and then a second photograph was taken after a 90° crystal
rotation (hkO layer).
The reciprocal lattice of each crystal indicates its Laue symmetry,
its systematic absences indicate a non-primitive lattice or screw and glide
operations, its intensity distribution is either centric or acentric, and each 47. abnormal intensity average indicates a particular symmetry element. From this data the possible space groups were deduced.
Intensity data for a poorly coherent single crystal were recorded on a precession camera instead of a Weissenberg camera with a series of timed exposures, which gave a 10:I interfilm density ratio. These intensi ties were measured with a densitometer. The asymmetric Lorentz and polarization corrections were made for the zero-layer precession intensity data with the transparent overlay chart of Waser (1951), and for the Weissen berg intensity data the corrections were made with program NSW94 on a
Deuce machine. No correction was made for absorption or extinction.
Nuffield (1966) states that it is common practice to make no correction for absorption, but to use very small crystals of 0.1mm in size, thereby re• ducing its effect presumably to negligible proportions. A similar statement is also made by Buerger (1960), who in addition states that the correction for absorption is often so large that it is better to apply an approximate correction based for example on a sphere or cyclinder than none at all.
CRYSTAL STRUCTURE COMPUTATIONS
A Patterson vector map was calculated on a Deuce machine with unit correlation factors between layers to determine possible trial structures.
The scattering curves of Thomas-Fermi for nickel, cobalt, iron, arsenic, antimony and sulphur were taken from the International Tables for X-ray
Crystallography (I 962) together with the real component of the anomalous 48. dispersion correction for each radiation used.
Methods used to refine a trial structure included fourier, difference fourier and least-squares. The fourier was calculated on a Deuce machine with the crystallographic program of Rollett (1961), whereas the least
squares was calculated on an IBM7040 or IBM360/50 computer with the
Oak Ridge Fortran program of Busing !:!_al. (1962) as modified by Cox
(1966). In addition, powder data or twinned single crystal data may be re• fined on I Fol 2 with the modified least-squares program of Kennicott (1963) to account for overlapped data. The function I w( I Fo I - I Fe I >2 was minimised after consultation with de Vies (1965) using the following weighing schemes: (1) unit weights, (2) weights inversely proportional to
I Fo/10 I 2 and if Fo < 4Fominimum then Fo is taken as 4Fominimum'
(3) weights inversely proportional to Fo and if Fo< 4Fominimum then Fo is taken as 4Fominimum, and (4) weights inversely proportional to ( f Fol-f Fc I )2 •
In the initial refinement, an average non-metal atom scattering curve, assumed thermal parameters, unit scale factors, and atom positions (estimated from a trial structure) were used to determine the scale and occupancy factors.
Further least-squares cycles were undertaken with the above atomic dis tribution to refine the overall scale factor, the positional parameters and
the individual isotropic temperature factors. The interatomic distances and angles with their standard deviations were computed on an IBM7040 machine
with the program of Busing ~t.!!_. (1964). 49.
RESULTS
GERSDORFFITE
Natural Material
The geographical locations and collection numbers of 14 additional natural gersdorffite samples together with their unit cell sizes are recorded in table la. Their unit cell sizes are similar to the literature values from
Lobenstein, Mitterburg, Musen and Sudbury; which are also recorded in table la for easy comparison. The unit cell sizes of all these samples, except for the ones from Cobalt and Sudbury, lie close to the 5.693.R unit cell size for synthetic NiAsS of Yund (1962) and Klemm (1965).
This NiAsS composition agrees with (1) the spectrographic analyses given in table 4b for Ferro, Wolfsberg and Leichtenberg; (2) the literature chemical analyses for Leichtenberg by Doelter (1926), Southern Rhodesia by Lightfoot (1931), and Lobenstein and Musen by Doelter (1926); and
(3) the British Museum values for Musen. An arsenic-rich composition is indicated by all nine chemical analyses from Dobschau by Sipocz (1886),
Doelter (1926) and Goll (1937) which agrees with its larger unit cell size as well as the substitution for nickel by iron and some cobalt. The high
iron and cobalt content in Dobschau, Cobalt and Sudbury material is indicated
by the high background counts with copper X-ray radiation due to fluoresence.
An iron-rich composition is indicated by the chemical analyses of samples so.
from Sudbury by Thomson (1921) and Doelter (1926). A Ni0 _4co0 _6AsS
composition for USNMR830 (Sudbury) and a Ni0 •4co0 _3Fe0 _3AsS com
position for BM1922, 145 (Cobalt) are indicated by their unit cell sizes in
table la upon figure 2 (page 16) of Klemm (1965) and their spectrographic
data in table 4a .
The X-ray diffraction powder reflection intensities from the two
literature and the 14 additional natural samples together with the optical
anisotropism data for these natural samples are recorded in table 2a. From
the X-ray diffraction powder patterns, six samples have both 001 and 011
reflections, five samples have 011 reflection rut not 001 reflection, and
three samples have neither 001 nor 011 reflections. The 001 reflection is
quantitatively related to the optical anisotropism, because the two samples
with medium 001 reflections exhibit medium optical anisotropism, and the four samples with weak 001 reflections exhibit weak optical anisotropism;
whereas of the eight samples with unobserved 001 reflections, three exhibit
weak optical anisotropism and the other five are completely optically iso
tropic. Both optically anisotropic and isotropic areas were noted in Loben•
stein and Musen samples. In the samples from Sudrury, USNMR830 has
medium optical anisotropism whereas BMI917,285 is optically isotropic.
Optical anisotropism is semi-quantitatively related to the unit cell
size, because the two samples with medium optical anisotropism have
medium unit cell sizes, the seven samples with weak optical anisotropism have large unit cell sizes, and the five optically isotropic samples have very large unit cell sizes.
Zoning is observed optically (with oil immersion only) in most samples as shown in table 2a, although zoning was only detected by the
X-ray diffraction powder technique in four of these fourteen samples. No correlation was found between the optical and X-ray diffraction powder zoning results. The large 0.09R unit cell size difference, strong variation in optical properties, and interstitial texture indicate that material from
Cobalt (D25174) has two distinct minerals rather than compositional zoning.
The O. 03R difference between unit cell sizes of material from Dobschau indicates a substantial composition range, which agrees with the chemical data. The o.01R differences in unit cell sizes of material for Hager and
Sudrury (BM1917, 285) indicate minor compositional variations. Differ ences up to o.02R could be masked in the three poorly crystallized samples, up to o.01R in the four average crystallized samples and up to 0.005.R in the seven well crystallized samples. If the CuKd- 1 and CuK62 reflections for a gersdorffite reflection are considered to represent two CuKd-1 re flections from a zoned gersdorffite, then their unit cell difference is
0.014.iR. Zoning frequently occurs in patches, often as a mottled pattern or in broad bands, and occasionally in concentric zones with both sharp and gradational boundaries. Zoning is expressed optically as a colour change from lilac-yellow (occasionally isotropic) to yellow (anisotropic). 52.
Twinning is extensive in Leichtenberg material and also occurs in
Lobenstein material. A few specimens possess a brecciated texture. Their paragenesis is given in table 2a. The gersdorffites with small unit cell sizes commonly occur with calcite and sulphides such as pyrite and pyrrhotite, wherease those gersdorffites with larger unit cell sizes commonly occur with siderite and arsenides such as skutterudite and safflorite. Other common associates of gersdorffite are quartz, chalcopy rite and sphalerite. Of these minerals only siderite, which decomposes in a carbon dioxide atmosphere at soo 0 c, indicates any limitation to formation temperature.
Seven natural gersdorffite samples were heated for I week periods at 50° C temperature intervals. These experiments which show X-ray diffraction powder reflection intensity changes are listed in table Sa. A quantitative relationship in four optically anisotropic samples shows that reflections 001 and 011 are retained longer in samples with smaller unit cell sizes. The four samples with large unit cell sizes indicate the dis appearance of the 011 reflection at about 550°c. An alternative inter pretation is that neither the 001 nor the 0ll reflection always disappears before the other, rut that the strongest reflection is retained the longest.
The time dependence of reflection intensity changes is shown by material from Leichtenberg (USNMR862), in which the 0()1 reflection exists after 1 week at o00°C but is destroyed after 4 weeks at sso0 c. No change was 53. observed in the material from Slovakia with no 001 and 011 reflections. Care was taken to avoid oxidation, since a reflection from an unidentified oxidation product occurs near the 011 reflection in the X-ray diffraction powder pattern.
Crystal Structure
The powder X-ray diffraction data and the associated optical aniso tropy results indicate three different crystal structures: pyrite-type Pa3 with
001 and 011 reflections absent, ullmannite-type P213 with 001 reflection absent and 011 reflection present, and cobaltite-type Pca21 with 001 and 011 reflections present. To substantitate these results and to determine their crystal struc tures, single crystal analyses were undertaken for the three varieties of gersdorffite from Ferro, Slovakia (UNSW); Wolfsberg, Harz, Germany
(HMM); and Leichtenberg, Fichtelgerbirge, Germany (USNM R862). Their impurities, determined from the spectrographic analyses, are stated in table 4b. The iron content of up to 6 per cent occurs in the extraneous minerals detected by powder X-ray diffraction of pyrite, pyrrhotite, chal copyrite and siderite; whereas the 2 per cent silicon in the Wolfsberg material occurs in quartz. The substitution of nickel by 2 per cent cobalt decreases the unit cell size of gersdorffite but the substitution of arsenic by l per cent antimony has the opposite effect of increasing its unit cell size. Their net effect on the unit cell size therefore is assumed to be zero. Their unit cell sizes from table la indicate from the curve in figure 3 (page 17) of Yund (1962) 54. a composition of NiAsl.23s0 _77 for Slovakia, NiAs0 . 82sl. IS for Wolfsberg, and NiAs0 . 8s I. 2 for Leichtengerg. The similarity of these chemical corn - positions indicates that any crystal structure differences are unrelated to composition.
Slovakia, Pa3. The weighted reciprocal lattice of Slovakia, from zero layer precession photographs of hOI and hkO layers with MoK J. radiation, exhibits the Laue symmetry 3. Systematic absences occur 2m for hOI with h ~ 2n and the equivalent systematic absences are Ok! with
I :\: 2n and hkO with k :\ 2n together with their derived systematic absences of hOO with h :\: 2n, OkO with k \ 2n and 001 with I :\ 2n, which indicate the space group Pa3. Forty two hkl reflections above background were recorded about axis [o.Q} on a Weissenberg camera with CuK (I 948). After least-squares refinement, the resultant parameters with their stan dard deviations are listed in table 3a2 . In addition the angles and interatomic distances with their standard deviations and derived atomic radii are recorded in 55. table 3a2, and the 67 calculated structure amplitudes (F c> are listed along• side the observed reflection amplitudes (F0 ) in table 3a 1 • This structure gave an overall discrepancy factor R, with both observed and unobserved reflections, of O. 111 . This crystal structure is similar to the pyrite structure. The 4. 0 nickel atoms lie in special position 4(a) with x = 0. 0000. The 4. 92 arsenic and 3.08 sulphur atoms are equally distributed over the variate-atom equi point of the eightfold position 8(c) with x = 0.3785 and henceforth will be called arsenic-sulphur atoms. Each nickel atom is octahedrally co ordinated to six arsenic-sulphur atoms. Each arsenic-sulphur atom is tetrahedrally co-ordinated to three nickel atoms and one arsenic-sulphur atom. The bond distances to the arsenic-sulphur atom represent a dis tance intermediate between that to an arsenic atom and that to a sulphur atom, because the X-ray diffraction results give the average effect from superposition of all cells. In addition the arsenic-sulphur atom, due to its dual nature, occupies a position between the real centres of arsenic and sulphur, and its shape is a result of these 0.615 arsenic atoms in a given position and the 0.385 sulphur atoms in a nearby position. Wolfsberg, P2 13. The weighted reciprocal lattice of Wolfsberg from zero layer precession photographs,hOl and hkO layers,with MoK'- radiation exhibits the Laue symmetry ..!.. 3. Systematic absences occur for hOO with m h \ 2n, and the equivalent absences of OkO with k ~ 2n and 001 with 1 \ 2n; 56. which indicate the space group P2 13. Thirty eight hkO reflections above background were recorded on a Buerger precession camera with a 30° precession angle and MoKr radiation. The intensities of these reflections were converted to observed reflection amplitudes (F 0 ), which are recorded in table 3b1. A Patterson vector map indicates the similarity of this gersdorffite structure to the pyrite structure rather than the ullmannite structure, so a trial structure was based upon the space group P2 13 and the co-ordinates of Peacock and Henry (1948). In the refinement with an average arsenic plus sulphur scattering curve and assumed thermal parameters for each atom, the occupancy factom for all the atomic positions were varied. The nickel occupancy factor indicated the overall scale factor, while the other occupancy factors indicated that 4.0 sulplmr atoms are located in position 4(a) with x = 3/8 and the remaining O. 7 sulphur and 3. 3 arsenic atoms are equally distributed over the variate-atom equipoint of the fourfold position 4(a) with x = 5/8. After further least-squares refinement, the resultant parameters with their standard deviations are listed in table 3bi. In addition the angles and interatomic distances with their standard deviations and derived atomic radii are recorded in table 3b2, and the 49 calculated structure amplitudes (F c> are listed alongside the observed data (F 0 ) in table 3b1 . This structure gave an overall discrepancy factor R, with both observed and unobserved reflections, of 0.144. 57. This structure is distorted from a pyrite-type structure similarly to the ullmannite structure. The 4. 0 nickel atoms are moved from the special position in the pyrite structure along a three-fold axis to position 4(a) with x = -0.0065. The arsenic and sulphur atoms are not equally dis tributed over the eightfold position of the pyrite structure, but there are 4 .0 sulphur atoms in four-fold position 4(a) with x = 0. 3825, while the remaining 0. 7 sulphur and 3. 3 arsenic atoms are equally distributed over position 4(a) with x = 0.6164 and henceforth will be called arsenic-rich atoms. In this case the eightfold position of the pyrite structure is destroyed by ordering and also by position shift, although each atom is in a position with point symmetry 3. Each nickel atom is octahedrally co-ordinated to three sulphur atoms and three arsenic-rich atoms. Each sulphur (or arsenic-rich) atom is tetrahedrally co-ordinated 1:D three nickel atoms and one arsenic-rich (or sulphur) atom. The bond distances to the arsenic-rich atom lie nearer the distances to the arsenic atom than to the sulphur atom, because the X-ray diffraction results give the average effect from superposition of all cells. Each atom, due to thermal motion, may have an anisotropic shape by an elongation along th..: three-fold axis and would then require two independent thermal parameters to define its ellipsoid. In addition the arsenic-rich atom, due to its dual 58-. nature created by superposition of all cells, occupies a position between the real atomic centres of sulphur and arsenic, and its shape is a result of these 0.18 sulphur atoms in a given position with two independent thermal parameters and the 0.82 arsenic atoms in a nearby position also with two independent thermal parameters . Leichtenberg, PI. The weighted reciprocal lattice as shown by two zero layer precession photographs of hOI and hkO layers with MoK d radiation exhibits no systematic absences. The precession photographs show that a 0 = b0 = c 0 and d-= f, =Y•90°, and thus the unit cell is geometric ally isometric. One hundred and ninety-four independent hkl reflections above background on five layers were recorded about axis [iocfJ on a Weissenberg camera with CuKd- radiation. The intensities of these re flections were converted to observed reflection amplitudes (Fo), which are recorded in table 3c 1. These intensity data do not show intensity relation ships of the type hkl = klh = lhk and also show no four-fold axes. There fore the structure is not cubic or tetragonal. All the Weissenberg layers show intensity relationships of the type hkl = hkl = hkl = hkl which indicates orthorhombic symmetry, although a few reflections did not show this relation - ship. A three-dimensional Patterson function, calculated with data from one-eighth of the weighted reciprocal lattice, indicates that the crystal 59. structure is similar to the pyrite structure. The space groups of lower symmetry with no systematic absences are the ortha.hombic space groups of P222, Pmm2 and Pmmm; the monoclinic space groups of P2/m, Pm and P2; and the triclinic space group of Pl, however only in Pl could a pyrite type structure be promulgated, even with the technique of moving from the origin as used by Giese and Kerr (1965) for cobaltite with space group Pca21. A trial structure with space group Pl was based upon the co-ordinates of Peacock and Henry (1948), because there are no reasonable alternatives. Full matrix least-squares procedures were then carried out using data from one-half of the weighted reciprocal lattice and allowing each atom to vary its parameters independently according to triclinic symmetry. Scale factors for each Weissenberg layer were carried as variants in each least-squares cycle, since the correlation data obtained from precession photographs were not considered to be sufficiently accurate. In the last cycle of least-squares these interlayer scale factors were kept constant and the atomic temperature parameters were allowed to vary anisotropically. The resultant parameters with their standard deviations are listed in table 3c2 . In addition the range of angles and interatomic distances taken from the eight different octahedra and the four different tetrahedra are recorded in table 3c2. and the 648 calculated structure amplitudes (Fe) are listed alongside the 219 observed data (Fo) in table 3cl" This structure gave 60. an overall discrepancy factor R, with both observed and unobserved reflections, of 0. 19, whereas the discrepancy factor R for each individual layer was 0.24 for zero layer, 0.17 for 1st layer, 0.18 for 2nd layer, 0.18 for 3rd layer, and 0. 19 for 4th layer. There are satisfactory features of this gersdorffite refinement, such as temperature factors for the nickel atom of 1. 4R2 and for the sulphur• arsenic atom of 1. oR.2 which are similar to values obtained for like structures. Standard deviations were calculated with data from one•half of the weighted reciprocal lattice and assuming the space group Pl. These standard deviations are low, and therefore the small shifts undergone by the individual metal and non-metal atoms from the ideal positions on the three-fold axes are significant. The average positions of all similar atoms lie on the three-fold axes as indicated by the values in table 3c2 . The movement of the nickel atoms, from the ideal metal atom positions found in the pyrite structure, is x =·0.009 (fractional shift), while the sulphur arsenic atoms occupy the two sets of four-fold positions with average fractional co-ordinates of x = 0. 381 and 0. 618. These values are in fact similar to corresponding atomic co-ordinates in partially ordered gersdorffite from Wolf sberg. An average arsenic plus sulphur scattering curve was used during least-squares refinement, and the occupancy factors for non-metal atoms did not significantly deviate from unity. This result would seem to indicate 61. that no ordering of atoms occurs in this gersdorffite,. although the displace• ment of all atoms from the ideal pyrite positions and the triclinic symmetry of this gersdorffite would suggest that ordering of non-metal atoms does occur. Another disturbing feature of the refined structure is the persist ent! y high value for the discrepancy factor R of O. I 9. A possible explanation is twinning by reticular merohedry, which is not revealed by reflection splitting in the X-ray diffraction data from the geometrically isometric unit cell. An optical examination of polished material from the same specimen used for the crystal structure analysis revealed a complex twin pattern with a strong suggestion of some lamella twinning diagonal to the cubic cleavage faces. The only twin type found to give a similar unit cell is twinning through an inversion centre. This will produce averaged disordered non-metal atom positions from individual twin lamella with ordered non-metal atoms, but also will centre the metal atom positions around the ideal position in the pyrite structure. Therefore twinning is not an acceptable explanation. Another possible explanation is that this gersdorffite is similar to ordered cobaltite. To obtain a reciprncal lattice with no systematic absences, composite twins about the [ OlcfJ and foof] two-fold axes are required. With this type of twinning, the atoms from different unit cells do not fall in approximately equivalent positions. Therefore this is not an acceptable explanation. 62. Synthesis Initially eight mixtures with compositions ranging from NiAsl. 7s0 _3 to NiAs0 _7 s1. 3 when heated at 550tfor I month produced gersdorffite. Their unit cell sizes in table lb lie within the limits of experimental error on the curve of Yund (1962) as shown in figure 3 (page 17). The unit cell sizes of sulphur-rich gersdorffites lie along the join between NiAsS and NiS2(5.667R from Elliott, 1960). Six compositions ranging from Ni0 •9co0 _1Asl.4s0 _6 to Ni0 •9co0 _1As0 _9sl.l when heated at 550°c for I month produced gersdorff ite with a trace of cobaltite. These unit cell sizes of gersdorffite in table lb are also plotted in figure 3 (page 17), and the average unit cell size decrease of 0.004.R indicates from figure 2 (page 16) of Klemm (1965) that cobalt substitutes for nickel up to only Nio. 97co0 . 03 . The presence of cobalt has limited the substitution of arsenic in these gersdorffites to As1 _4s0 _6 due to the formation of the most nickel-rich skutterudite (Ni0 •6 Co0 _4As3_x) given by Roseboom (1962). With the LiCl-KCl melt method, a composition of Ni0 • 5 Co0 • 5AsS was formed with traces of both cobaltite and gersdorffite. Similarly five compositions ranging from Nio. 9Fe0 _1Asl. 4s0 _6 to Ni0 _9Fe0 . 1As0 _9s1. l when heated at 550°c for I month produced gersdorffite. These unit cell sizes of gersdorffites in table lb are also plotted in figure 3 (page I 7), and the average unit cell size decrease of 63. 0.004R indicates from figure 2 (page 16) of Klemm (1965) that iron sub stitutes for nickel up to only Ni0 _99Fe0 _01 . The presence of iron has limited the substitution of arsenic in these gersdorffites to Asl. 4s0 .6 due to the formation of the most nickel-rich skutterudite (Ni0 • 7Fe0 _3As3_x) given by Roseboom (I 962). All gersdorffites produced at 5S0°C, including those produced by the LiCl-KCl melt method, show a 011 reflection but no trace of a 001 reflection in their X-ray diffraction powder patterns. In addition six mixtures ranging from NiAsl. 6s0 .4 to NiAs0 _8s1. 2 were heated at 250°c for 9 months. The two sulphur-rich gersdorffites produced show a Oll reflection even though their crystallinity is poor. A similar set of six mixtures ranging from NiAsl. 6s0 _4 to NiAs0 _8s1. 2 were heated at 400°C for 9 months. The sulphur-rich gersdorffites produced show a Oll reflection and have reached equilibrium in contrast to the arsenic-rich gersdorffites, which are poorly crystallized and have not reached equilibrium. Synthetic NiAsl.SSS0.4S' NiAsS, and NiAs0 _8s1. 2 were prepared at sso 0 c and 700°c to investigate the effect of temperature and composition on the ordering of arsenic and sulphur. Up to 16 hkl reflections were recorded from each of the three powders on an X-ray diffractometer with Cul<# radiation. The intensities of the reflections were converted to observed reflection amplitudes (F 0 ), which are recorded in table 3d for 64. NiAsl. 55 s0 _45 , table 3e for NiAsS, and table 3f for NiAs0 _8s1. 2 . Trial crystal structures based upon the co-ordinates of Peacock and Henry (1948) were refined with least-squares techniques and their resultant parameters, space groups, and discrepancy factors R are also recorded in these tables; while the calculated structure amplitudes (F c> are listed alongside the observed reflection amplitudes (F 0 ) in these tables (3d, 3e and 3f). The structure of NiAsS formed at both sso0 c and 700°C shows the arsenic and sulphur atoms to be completely ordered. However both the arsenic-rich and sulphur-rich gersdorffites are only partially ordered at sso 0 c, whereas at 700°C the arsenic-rich gersdorffite is disordered and the sulphur-rich gersdorffite is still partially ordered. COBALTITE Natural Material The geographical locations and collection numbers of six additional cobaltite samples together with their unit cell sizes are recorded in table le. These unit cell sizes are similar to literature values, which are also recorded in table le for easy comparison. These literature values, except for Hakansbo, are slightly larger than the 5. 572.R unit cell size of the synthetic material of Klemm (1965). The spectrographic analyses of some natural samples given in table 4b indicate little deviation from a CoAsS composition. This agrees with a high cobalt and iron content in all samples, which is indicated by the high background counts obtained with copper X-ray radiation due to fluoresence. The X-ray diffraction powder reflection intensities from the five literature and the six additional natural samples, together with the optical anisotropism data for the six additional natural samples, are recorded in table 2b. All six samples show both 001 and 011 reflections in their X-ray diffraction powder patterns. Intensity of the 001 reflection is quantitatively related to the optical anisotropism, because the four samples with strong 001 reflections exhibit strong optical anisotropism and the two samples with medium 001 reflections exhibit medium optical anisotropism. A semi quantitative relationship is found in the six additional natural samples as well as in the five literature values between the 001 reflection intensity and the unit cell size, because the six samples with strong 001 reflections have small unit cell sizes and the five samples with medium 001 reflections have medium sized unit cells. No zoning was observed, as shown in table 2b, either optically in oil after etching with ION HNO3 or from X-ray diffraction powder data. Tuinning is extensive in all samples. One sample (UNSW61) has a well developed brecciated texture. Cobaltite, as shown in table 2b, commonly occurs with quartz, mica and chlorite; and the cobaltite sample with the largest unit cell size (UNSW) also occurs with safflorite. 66. Four samples were heated for 1 week periods at 25°C temperature intervals. These experiments which show X-ray diffraction powder reflection intensity changes are listed in table Sb. A quantitative relation ship in these four optically anisotropic samples shows that reflections 001 and 011 are retained longer in samples with smaller unit cell sizes. The stronger 001 reflection is retained longer than the weaker 011 reflection. The time dependence of reflection intensity changes is shown by sample UNSW 233, where the 001 reflection intensity has gradually decreased after 14 weeks at 700°C to the stage reached after 1 week at 825°c. Crystal Structure No chemical analysis is given for the cobaltite used in the single crystal structure analysis by Giese and Kerr (1965). Their unit cell size of 5.582 + 0.002.R indicates a CoAsS composition. Their cobaltite with space group Pa3 was refined further by least-squares techniques and the resultant parameters are listed in table 3g. In addition the angles and interatomic distances with their standard deviations and and derived atomic radii are recorded in table 3g. This structure was refined to a discrepancy factor R, with both twenty observed reflections and one unobserved reflection, of 0.055 f,offl 0.057. The cobaltite with space group Pca21 of Giese and Kerr (1965) was also refined further by least-squares techniques and the resultant para meters are listed in table 3h. In addition the angles and interatomic 67. distances with their standard deviations are recorded in table 3h. This structure was refined to an overall discrepancy factor R of 0 .112 and the layer hk0 with 41 reflections was refined from 0. 10 to 0.087 and the layer h0I with 21 reflections from 0.12 to 0.144. Negative temperature factors have no physical reality. An attempt to obtain positive temperature factors using the real component of the anomalous dispersion correction for the atomic scattering curves was unsuccessful and also increased the discrepancy factor R. An attempt with slight variations in the chemical composition was unsuccessful and also increased the discrepancy 2 factor R. An attempt with a perfect crystal and use of IF 0 1 instead of a 2 mosaic crystal and use of IFO I produced temperature factors of +o. 9 R , but substantially increased the discrepancy factor R from 0. 055 to 0. 235. No conclusion can be given from this investigation on the reason for negative temperature factors. Synthesis Seventeen mixtures with compositions ranging from CoAsS to CoAs0 _4sl.6 were heated at 550°c for I month to produce cobaltite. Their unit cell sizes in table Id are plotted against their compositions in figure 4 to show continuous solid substitution from CoAs0 . 86s l. ! 4 to CoAs0 . 42s I. 58 • Although these points lie close to a straight line, a better fit is obtained with a concave curve. The slope of this curve is shown in figure 4 to be less than 68. the slope of the line to join CoAsS and CoS2(5. 523R from Elliott, 1960). This curve indicates a CoAs0 _75 sl.25 composition for the 5.572.K synthetic value of Klemm (1965). Four mixtures with compositions ranging from CoAs0 _9sl.l to CoAs0 _6sl.4 were heated at 250°C for 9 months but did not produce cobaltite. Seven compositions ranging from Coo. 85Ni0 _1 5As0 _85 sl. 15 to Co0 _85Nio. 15As0 _55sl.45 when heated at 550°c for 1 month produced cobaltite with a trace of gersdorffite. These unit cell sizes of cobaltite stated in table Id are also plotted in figure 4, and the average cell size decrease of 0.001.R indicates from figure 2 (page 16) of Klemm (1965) that nickel substitutes for cobalt up to only co Ni . Substitution of nickel for cobalt appears not 0 . 99 0.01 to effect the sulphur to arsenic composition range. With the LiC!-KCl melt method, extensive nickel substitution in cobaltite was obtained, similar to figure 2 (page 16) although complete equilibrium had not been reached after 5 weeks at 550 0 C. Similarly seven compositions ranging from Co0 _9Fe0 _1As1 _1S0 .!) to eo0 _9Fe0 _1As0 _5sl. 5 when heated at 550°C for 1 month produced cobaltite. Their unit cell sizes are recorded in table Id and plotted in figure 4 (page 67), and indicate a sharper decrease in the unit cell size of cobaltite with increased sulphur substitution for arsenic. A unit cell size increase of O. 03.R for arsenic-rich cobaltite changes to a unit cell size decrease of 0.03.R for sulphur-rich cobaltite. The amount of iron substitution for cobalt is undeter mined but is well below the maximum Co0 . 9Fe0 . 1. The size for no unit cell 69. size change with iron substitution in figure 4 (page 67) is similar to the size for CoAsS in figure 2 (page 16) for no unit cell size change with iron substit ution. Substitution of iron for cobalt has limited the sulphur to arsenic corn - position range in cobaltite from Co0 _9Fe0 _1As0 _8s1. 2 to Co0 _9Feo.IAs0 _5s1. 5 . With the LiCl-KCl melt method, an extensive range of iron substitution for cobalt similar to figure 2 (page 16) was produced. Instead of the 5.572.i. unit cell size measured by Klemm (1965), the values measured were 5.577R., which is similar to the unit cell size obtained from the simple tube method. Neither reflection 001 nor 011 was recorded in the X-ray diffraction powder patterns from those cobaltites produced at 550°c by the simple tube method. With the LiCI-KCl melt method, the 001 reflection was recorded by all the cobaltites produced, whereas the Oll reflection was only recorded by those cobaltites with strong X-ray diffraction powder patterns. ULLMANNITE Natural Material The geographical locations and collection numbers of four additional natural ullrnannite samples with varietal names of corynite, kallilite and willyarnite, if applicable, together with their unit cell sizes are recorded in table le. These unit cell sizes lie within the range indicated by the literature values, which are also recorded in table le. Individual samples from the same deposit, such as the willyarnite from Broken Hill, the 70. kallilite from Obersdorf and the ullmannite from Salchendorf (Sakhendorf is equivalent), are similar. The unit cell sizes of natural corynite concentrate around the unit cell sizes of the end members NiSbS and NiAsS, and so indicate a miscibility gap between them. The wide range of unit cell sizes, from a small unit cell size for corynite to a large unit cell size for willyamite, suggests a wide composition range. The spectrographic analyses in table 4b confirm this deviation from a NiSbS composition by arsenic substitution for antimony in BM69114 to NiSb0 • 7As0 . 3s and by cobalt substitution for nickel in AM D17751. The X-ray diffraction powder reflection intensities from the three literature and four additional natural samples, together with optical anisotropism data for the natural samples, are recorded in table 2c. All these samples show reflection 011 in their X-ray diffraction powder patterns; but although reflection 001 is absent, two samples are weakly optically anisotropic. No change was noticed in the X-ray diffraction powder reflection intensities of ullmannite upon heating, before its decom position at 710 ± s0 c. Distinct zoning, as shown in table 2c, was observed only in BM69114, where one phase is distinctly harder than the other. Ullmannite as shown in table 2c occurs with pendlandite, chalcopyrite, tetrahedrite, pyrrhotite, galena and quartz. Crystal Structure In the crystal structure analysis of ullmannite by Takeuchi (1957), 71. no chemical analysis is given for his sample. Its unit cell size, from the NiSbS to NiAsS curve (figure 5), indicates a NiSb0 . 8As0 . 2s composition. Evidence to support this composition is as follows: refinement to a lower discrepancy factor R, atomic multipliers closer to unity, and more accept able thermal parameters. The resultant co-ordinate changes however are less than half one standard deviation. This crystal structure with Sbo_ 8As0 _2 substituted for Sb was refined. After least-squares refinement, the resultant parameters with their standard deviations are listed alongside his original values in table 3j2 • In addition the angles and interatomic distances with their standard deviations and derived atomic radii are recorded in table 3j2 • The calculated structure amplitudes (Fe) are listed alongside his observed reflection amplitudes (FoT) and his calculated structure amplitudes (FcT) in table 3j 1• This crystal structure was refined to a discrepancy factor R of 0.068, with both thirty six observed and four unobserved reflections of layer hkO, from his O. 115. The 4. 0 nickel atoms are moved from the special position in pyrite structure along a three-fold axis to position 4(a) with x =-0.0183. The antimony, arsenic and sulphur atoms are not equally distributed over the eightfold position of the pyrite structure; but there are 4. 0 sulphur atoms in position 4(a) with x = 0.3838, whereas the antimony and arsenic atoms are equally distributed over the variate-atom equipoint of the fourfold position 4(a) with x = 0. 624 7, and henceforth will be called antimony-rich 72. atoms. In this case the eightfold position in the pyrite structure is destroyed by ordering and also by position shift, although each atom is in a position with point symmetry 3. Synthesis Seven mixtures with compositions ranging from NiSb1. 1s0 . 9 to NiSb0 _9sl.l were heated at 550°C for I month to produce ullmannite. Their unit cell sizes in table If indicate little deviation from stoichiometric NiSbS. Substantial arsenic substitution for both antimony and sulphur in these cubic minerals is shown by the 36 unit cell sizes, which are tabulated in table If and plotted in figure 6 (a ternary diagram NiSb2 -NiAs2 -NiS2). Across the NiSbS-NiAsS join, a miscibility gap is shown from NiSb0 _6As0 _4s to NiSb0 _3 As0 _7 s in figure 5 (page 71). From NiSbS to NiSbAs, continuous solid solution is shown in figure 5 (page 7l)from NiSbS up to NiSbS0 _4As0 _6 . Although little cobalt or iron substitution for nickel is indicated by the unit cell sizes of a series of mixtures ranging from NiSbS up to Ni0 • 7co0 . 3Sb5 or Ni0 _7 Fe0 _3sbS by the simple tube method, the LiCI-KCI melt method produced a solid solution from NiSbS to Ni0 _6co0 _4SbS, as indicated by the unit cell sizes of six mixtures tabulated in table If and plotted in figure 5 (page 71). Synthetic NiSbS was prepared at sscfc and 700°C to investigate the effect of temperature on the ordering of antimony and sulphur atoms. Up to 17 hk 1 reflections were recorded from the NiSbS powder on an X-ray 73. diffractometer with CuKd-- radiation. These reflection intensities were con- verted to observed reflection amplitudes (F0 ), which are recorded in table 3k. Trial crystal structures based upon the co-ordinates of Takeuchi (1957) were refined with least-squares techniques, and their resultant parameters, space groups, and discrepancy factors R are also recorded in table 3k; while the calculated structure amplitudes (Fc> are listed alongside the observed reflection amplitudes (F 0 ) in table 3k. The crystal structure of NiSbS at both 550°c and 700°C has the antimony and sulphur atoms completely ordered. All ullmannites produced at 2so0 c, 400°c, sso0 c, 700°C by the simple tube method show a strong OU reflection hlt no trace of a 001 reflection in the X-ray diffraction powder pattern. ARSENOPYRITE, GLAUCODOT AND GUDMUNDITE Natural Material The geographical locations of fourteen arsenopyrite, four glaucodot and three gudmundite samples, with varietal names of danaite and allocasite if applicable, from the literature, together with their unit cell sizes, are recorded in table lg alongside the value for one additional natural glaucodot. The composition of this glaucodot is shown by the spectrographic analysis in table 4b to be approximately Fe0 _6co0 _4AsS. The X-ray diffraction powder reflection intensities from four literature samples and one additional natural sample are recorded in table 2d together with the indices (hkl) of Morimoto and Oark (I 961) based upon the monoclinic cell. 74. Synthesis A series of mixtures ranging from Fe0 •9co0 _1Asl. 1s0 _9 to Fe0 _9eo0 _1AsS when heated to 550°c for 1 month produced arsenopyrite with a trace of cobaltite. Similarly a series of mixtures ranging from Fe0 _9Ni0 _1Asl. 1s0 _9 to Fe0 _9Ni0 • 1AsS produced arsenopyrite with a trace of gersdorffite, which indicates little metal atom substitution as compared to Klemm (1965). All the arsenopyrite produced is poorly crystalline. It gives no indication of deviation from a pseudo-orthorhombic symmetry and it gives low accuracy for measurement of the reflection 131, which Morimoto and Oark (1961) used to measure the sulphur to arsenic ratio. No reaction was observed after glaucodot had been heated for 1 month at 600°C, although there were several changes in the X-ray diffraction powder reflection intensity. After glaucodot had been heated for 1 week at 650°C, a small amount of cobaltite was detected by the powder X-ray dif fraction technique. The relationships between cobaltite, glaucodot and arsenopyrite from present limited knowledge are shown in figure 7. PYRITE, CATTIERITE AND VAESITE Optical anisotropy in pyrite has been discussed by Stanton (1955 and 1957) and Gibbons (1967). A probable explanation involves an extremely thin skin of nonisometric crystalline structure caused by polishing and controlled by the orientation of the underlying lattice. The isometric 75. character of pyrite is taken from the early work of Bragg (1913), which may not be correct in fine detail. To investigate the possibility of the optical anisotropy in pyrite being caused by distortions instead of by polishing; the octahedral and tetra hedral angles for the disulphide series of pyrite, cattierite and vaesite, which are tarulated in table 31, were determined from the unit cell sizes and atomic co-ordinates of Elliott (1960). Since these angular distortions of pyrite are similar to those of gersdorffite, the optical anisotropy in pyrite may result from a non-cubic structure, in a similar manner to the optical anisotropy caused by the non -cubic structure of gersdorffite from Leichtenberg. In further investigation of this possibility, 18 pyrite powder samples taken from hydrothermal to sedimentary environments were examined by X-ray diffraction. No trace of either the 001 or 011 reflection was found, however the possibility of a non-cubic structure remains, since optical anisotropy was found to be a more sensitive detector of distortion than X-ray diffraction in the work on gersdorffite. 76. DISCUSSION METHODS The major advantage of using the simple tube method is the formation of compounds with exact known compositions. Non-metal atoms substitute easily for each other, but there is only limited metal atom substitution. This limited metal atom substitution probably represents a non-equilibrium assemblage, especially in the FeAsS-CoAsS join. The long reaction times, such as 750 days at 700°c for sulphur-rich gersdorffite by Yund (1962), may be substantially reduced by elimination of large particles; since a gersdorf fite coating on a large particle delays its reaction with arsenic and sulphur. Well crystallized sulphur-rich gersdorffites were formed from mixtures heated at 550°C for 1 month or at 250°for 9 months. The sulphur-rich gersdorffites approach equilibrium faster than the arsenic-rich gersdorffites. In the LiCl-KCl melt method, a wide metal atom composition range may be produced although the exact non-metal atom composition is unknown. The greater mobility of atoms in this LiCl-KCl melt method, compared to the simple tube method, accelerates the formation of equilibrium assem - blages to allow distorted and ordered structures (cobaltite) to form. After the sample has been taken from the furnace and quenched, the attainment of one homogeneous equilibrium phase is often indicated by the absence of a condensed vapour phase in the top of the sample tube. In the 77. initial experiments, the samples were taken from the tubes and crushed. These were resealed under vacuum in new tubes and reheated in the furnace to be annealed for the accurate determination of their unit cell sizes. This regrinding process was found undesirable, as the condensed phase formed at the top of some sample tubes was difficult to retain. Since these synthetic minerals are hard, grinding does not distort the crystal structure so annealing is not required. Grinding did not increase the approach rate of the sample to equilibrium, because the metal powder used had a fine particle size. Samples near the stoichiometric composition have better crysrsllinity, than samples with extensive substitution, because they reach equilibrium faster. Neither the vacuum distillation method nor the hydrothermal method was used successfully. NATURAL MATERIAL All the old unit cell sizes from the literature a"te tabulated in KX units, which may be converted into R by the multiplier 1.00202. The unit cell sizes of Bokii and Tsinokev (1954) for gersdorffite, cobaltite and ullmannite lie well outside the general range; which suggests measurement errors. In the comprehensive paper by Klemm (1965), the extensive typo graphical errors decrease the value of his tabulations. Ute small unit cell size of 5.551KX for cobaltite from Hakansbo indicates a high sulphur content or possibly an error in measurement. The small unit cell size of 5.593KX given for a gersdorffite by Peacock and Berry (1940) seems to 78. indicate a cobaltite composition from its small unit cell size using figure 2 (page 16) of Klemm (1965). The large unit cell size of 5. 7238..R for gersdorf fite from Styria by Yund (1962) suggests the presence of antimony. Zoning may be caused by either metal atom replacement or non-metal atom replacement. For example, cobalt and iron substitute for nickel in material from Cobalt (D25174) and arsenic partially replaces sulphur in material from Dobschau. These results indicate that further valuable data on zoning could be obtained by electron probe analysis. The paragenesis of all natural gersdorffites indicates the association of arsenides and iron minerals with gersdorffites of large unit cell size, and the association of sulphides with gersdorffites of small unit cell size. Therefore gersdorffite with a large unit cell size is formed in an arsenic-rich environ ment, and gersdorffite with a small unit cell size is formed in a sulphur-rich environment. The iron in the arsenic-rich gersdorffite environment forms separate iron minerals, while the iron in the sulphur-rich gersdorffite environ ment is absorbed into the sulphur-rich gersdorffite. The maximum tempera ture for gersdorffite formation is probably 500°c. This is suggested by the common association of gersdorffite with siderite which decomposes in carbon dioxide at 500°c. These results and those discussed under crystal structure relations provide only a limited indication of geothermometry. SYNTHESIS For synthetic gersdorffite, an explanation is needed for the apparently 79. abrupt slope change at the NiAsS stoichiometric composition as shown by the unit cell curve in figure 3 on page 17 (See N. B.). This slope change appears to be independent of structure variations such as order-disorder changes. The greater antimony substitution in arsenic-rich gersdorffite as compared to the antimony substitution in sulphur-rich gersdorffite shown in figure 6 (page 7Z) is attributed to the size similarity of antimony and arsenic atoms compared to the size dissimilarity between antimony and sulphur atoms. The absence of very arsenic-rich gersdorffite in nature is explained by the formation of skutterudite with the ubiquitous cobalt and iron in arsenic-rich environments. For cobaltite, the unit cell curve in figure 4 (page 67) for sulphur rich cobaltites lies above the join CoAsS-CoS2 and its concave shape needs explanation (See N. B.). Nickel substitution only slightly increases the unit cell size. Iron substitution has increased the slope of the cell edge versus cobaltite composition curve, to approach the slope of the cell edge join between CoAsS and CoS2, which needs explanation (See N.B.). Iron has also decreased the range of arsenic for sulplmr substitution. The occurrence of only apparently stoichiometric cobaltite in nature is attributed to the low stability of sulphur-rich cobaltite, as shown by its fast decomposition under atmospheric conditions, which implies that sulphur rich cobaltite is metastable. Ullmannite is shown in figure 6 (page 7l.) to accept substantial 80. arsenic substitution, which is attributed to its intermediate size between the sulphur and antimony atomic sizes. Although the arsenic atom is larger than the sulphur atom, its substitution for sulphur decreases the unit cell size of ullmannite (figure 5, page 71). N. B. The deviations from straight line unit cell curves with substitution are attributed to (1) the alteration of packing characteristics, and/or (2) possible bonding changes due to the electron loss caused by the substitution of antimony and/or arsenic for sulphur. BOND ANGLES AND DISTANCES All the pyrite-type compounds examined show significant distortions from the theoretical octahedral angle of 90° and the theoretical tetrahedral angle of 109.5° as shown in tables 3a2, 3b2, 3c2, 3g, 3h, 3j2, and 31. The maximum distortions occur in ullmannite and orthorhombic cobaltite, and range from 83° to 99° for octahedral angles and from 99° to U 9° for tetra hedral angles. The distortion of the octahedral angle, as shown in table 31, increases from 3 1/2° for vaesite to 4° for cattierite and then to 5° for pyrite. This shows the substitution of transition metal atoms by transition metal atoms with less electrons, that is iron substitutes for cobalt and then cobalt substitutes for nickel, which increases the distortion of the octahedral angle. The crystal structure change from a pyrite-type structure to a marcasite-type structure as shown by Hulliger and Mooser (1965b) occurs 81. at the average distortion of 4° to 5° in the octahedral angle. Cobaltite and gersdorffite have both cubic and non-cubic structures and their average distortion in the octahedral angles are 4° to 5°. Since ullmannite has an average distortion in the octahedral angle of 5°,. both cubic and non-cubic structures appear possible. Both interatomic distances, and the atomic radii calculated from these interatomic distances with simultaneous equations, are shown in tables 3a2, 3b2' 3c2, 3g, 3h, and 3j2 • Normal covalent bonding is indicated by these interatomic distances. All atomic radii in these tables are similar except for gersdorffite Pa3 in table 3a2, where the nickel atom is smaller and the arsenic-sulphur atom is larger. For sulphur, the 1.04KX radius of Pauling and Huggins (1934) is similar to the radii 1.033.R, 1.062.R and 1.t86R of Elliott (1950); but both sets of radii are less than the large sulphur radii of I. 12R from gersdorf fite (table 3b2) and 1. 14.R from ullmannite (table 3j2). The large radius for sulphur is also indicated by the sulphur-arsenic radii of 1.20.R from gersdorffite (table 3a2) and I. 16.R from cobaltite (table 3g). For arsenic however the 1. 18KX radius of Pauling and Huggins (I 934) is similar to the 1. I 9.R radius from gersdorffite (table 3b2). Their radius for antimony of l.36KX is similar to the 1.32.R radius from ullmannite (table 3j2). For nickel (II), the I. 39KX radius of Pauling and Huggins (1934) is similar to the I. 36.R radius of Elliott (1960); but both radii are smaller than the nickel 82 .. radii of I. 17j and 1.22.R from gersdorffite (tables 3a2 and 3b2) and I.23R from ullmannite (table 3h). For cobalt (II), the I.32KX radius of Pauling and Huggins (1934) is similar to the 1.2sR radius of Elliott (1960); but both radii are smaller than the cobalt radius of 1. 16R from cobaltite (table 3g). The calculated atomic radii vary according to the method used, as different radii are obtained from X-ray diffraction, neutron diffraction, spectroscopy and nuclear magnetic resonance. In fact different radii may be determined with X-ray diffraction from the simultaneous equation method and from fourier density maps. In addition, thermal vibrations effect the radius of the atom. Therefore an anisotropic atom has a variable shape so that its atomic radius only represents an average value. For triclinic gersdorffite and orthorhombic cobaltite a range of interatomic distances are listed in tables 3c2 and 3h, so a range of atomic radii could be calculated. The physical significance of atomic and ionic radii are discussed by Slater (1964), who also tabulates atomic radii to the nearest o.osR based upon the best empirical agreement with observed bond distances. The larger sulphur (-tO. 1R) and the smaller nickel ( -0. 15.R) and cobalt ( -0. lR) atomic radii of cobaltite, gersdorffite and ullmannite appear significantly different to the sulphur, nickel and cobalt atomic radii of pyrite, cattierite and vaesite even though their co-ordination is identical. A possible explanation for this difference is the crystal structure distortion and bonding in the crystal structure. CRYSTALSTRUC1UREACCURACY The accuracy of a crystal structure analysis indicated by the dis crepancy factor R has no physical meaning, although this factor is used in crystal structure refinement. A correct structure is considered by Buerger (1960) usually to have a discrepancy factor R less than 0.25. When the discrepancy factor is reduced to 0.20, Lipson and Cochran (1966) state that the broad features of the crystal or molecular architecture are already apparent. Their minimum value for a discrepancy factor R obtained in practice after absorption corrections is 0. 06, which is attributed to random errors of measurement, systematic errors associated with secondary extinction and double Bragg reflection, and redistribution of electrons associated with bonding. Complete three dimensional data are normally collected in crystal structure analysis to obtain numerous observations (intensities of hkl reflections) relative to variables (co-ordinates of atoms and their tem perature factors and scale factor), for the accurate determination of these variables in least-squares refinement. These data are collected about a second axis to scale the different layers, when anisotropic temperature factors are to be refined by least-squares techniques, since a singular solution is not attained if scale factors and anisotropic temperature 84. factors are refined together. Extinction is stated by Buerger (1960) to reduce the observed intensity of the high intensity reflections near the origin. In the structures described in this thesis, extinction has not significantly reduced the observed intensity of these high intensity reflections near the origin. The absorption correction is shown by Buerger (I 960) to resemble a residual temperature error, so the absorption error in the structures described is partially compensated for by the temperature factor, which therefore makes the temperature factors less reliable. The refinement of the single crystal data, which was uncorrected for absorption, gave relatively large discrepancy factors R, and one structure is only just below 0. 2. These single crystal structure analyses accomplish their main purpose, which is to show the broad crystal structure features. Although the large discrepancy factor R would indicate a poor reliability for the variables, the errors in the atomic co-ordinates at the one standard deviation level are small. Therefore the derived atomic radii and the bond angles have a greater accuracy than expected. The crystal structures refined from powder data have a small dis crepancy factor R. The poor accuracy of these crystal structures however was indicated by the large standard deviations for each variable, since only 16 reflections were recorded to locate 8 variables (3 position, 3 temperature, 1 occupancy and 1 scale). The sole purpose however of 85. these crystal structure determinations from powder data was to determine the occupancy of the non-metal atom positions, which was accomplished with satisfactory accuracy. CRYSTAL STRUCTURE In pyrite-type structures where each metal atom(M) is surrounded by an octahedron of the same type or of "average" atoms(A), the metal atom lies on a three-fold axis and at a centre of symmetry as shown in figure 8a. Examples of such structures are pyrite (FeS2) by Elliott (1960) with M = Fe and A = S, disordered cobaltite (CoAsS) by Giese and Kerr (1965) with M = Co and A = (As+ S)/2, and disordered gersdorffite (NiAsS) from Slovakia with M = Ni and A =(As+ S)/2. These crystal structures are isostructural and have space group Pa3. When non-metal atom ordering occurs, so that the non-metal atoms surrounding the metal atoms are no longer equivalent, the symmetry of the crystal structure is lowered. In the case of gersdorffite from Wolfs berg shown in figure 8b, the partial ordering of arsenic and sulphur atoms causes the nickel atoms to move (-0.0065, -0.0065, -0.0065 fractional co-ordinates) from their centres of symmetry towards the sulphur atoms, since the Ni-S bond distances are shorter than the Ni-As bond distances. Since the atoms in the plane above the nickel atom in figure 8b are all 86. sulphur atoms and the atoms in the plane below the nickel atom are also equivalent (equal to 0.18S + o.82As due to disordering), the nickel atom remains on a three-fold axis. The unit cell of this gersdorffite remains cubic but the space group degenerates to P2 13. Another example is ullmannite. In the case of ordered cobaltite shown in figure 8c by Giese and Kerr (1965), the non-metal atoms are ordered into a set of four arsenic atoms and a set of four sulphur atoms. In addition the cobalt has moved significantly (-0.010, 0.006, 0.011 fractional co-ordinates) from the cobalt atom position in disordered cobaltite, whereas the arsenic and sulphur atoms have not moved from the non-metal atom positions in dis ordered cobaltite. Although the unit cell is geometrically isometric, this cobalt movement is reflected in the lower symmetry of space group Pca21. Further non-metal atom ordering may occur in gersdorffite as shown in figure 8d. If in 4 Ni(As, S)2, there is a whole number of each atomic species in the formula, such as Ni4As3s5, triclinic symmetry would arise when four sulphur atoms occupy a set of pseudo-four-fold positions and the remaining one sulphur atom and three arsenic atoms occupy the other pseudo-four-fold position with the atoms ordered, as shown in figure 8d. If there are fractional numbers of atoms in the formula, such as Ni4As3_2s4 . 8 for gersdorffite, then disordering occurs in one of the 87. atomic sites to give an average atom. The extent of this disordering depends on the deviation of each atomic species from a whole number of atoms in the formula 4Ni (As, S)2 • The unit cell remains geometrically isometric, although the space group degenerates through ordering into PI. In a gersdorffite with non-metal atoms completely ordered, fractional numbers of atoms are a voided by using a larger unit cell. The size of this larger unit cell may be determined from the superlattice reflections. The space group of ordered gersdorffite is also PI. Gersdorffite from Leichtenberg has a pyrite-type structure in which the sulphur and arsenic atoms are disordered about the nickel atoms. The average atomic displacements of each set of four atoms indicate movement along the three-fold axes, with a resultant environment which closely resembles that found in partially ordered gersdorffite from Wolfsberg. In addition there are small random displacements of atoms from their ideal positions on the three-fold axes as shown in figure Se. Since this sulphur rich gersdorffite deviates significantly from stoichiometry, most nickel atoms occupy sulphur-rich octahedra and slight anisotropy in temperature factors is observed. These random atom displacements result in the degeneration of the crystal class to triclinic PI. CRYSTAL STRUCTURE RELATIONS Ordering of non-metal atoms and distortion of their crystal 88~ structures are elucidated from the results of gersdorffite, cobaltite and ullmannite in tables 3a2, 3b2, 3c2, 3d, 3e, 3f, 3g, 3h, 3j2 and 3k. The relationships are considered below between the following: (I) single crystal structure analyses of natural material, (2) presence and intensity of OCH and 011 reflections in X-ray diffraction powder patterns of natural materia]. (3) optical anisotropism of natural material, (4) unit cell sizes of natural material, (5) compositions of natural material, (6) the thermal stability of natural material, (7) powder crystal structure analyses of synthetic material, (8) the presence and intensity of 001 and 011 reflections in X-ray diffraction powder patterns of synthetic material, (9) unit cell sizes of synthetic material, (10) compositions of synthetic material, and (11) formation temperatures of synthetic material. The three crystal structure types determined by single crystal methods are related to the 001 and 011 reflections found in the X-ray diffraction powder patterns from natural material. The absence of both reflections 001 and 011 in the X-ray diffraction powder pattern indicates a cubic structure Pa3 with the non-metal atoms disordered (cobaltite heated to sso0 c and gersdorffite). The absence of 001 reflection and the presence of 011 reflection in the X-ray diffraction powder pattern indicates a cubic structure P2 13 with the non-metal atoms ordered (ullmannite and gersdorffite). The presence of both 001 and 011 reflections in the X-ray diffraction powder pattern indicates a non-cubic 89. structure with the non -metal atoms either ordered or disordered (cobaltite orthorhombic, gersdorffite-triclinic). The 001 reflection intensity in the X-ray diffraction powder pattern is related quantitatively to the optical anistropism strength in the polished section. Material with strong 001 reflection intensity in its powder pattern has strong optical anisotropism (cobaltite Pca21), material with medium 001 reflection intensity in its powder pattern has medium optical aniso tropism (cobaltite Pca21 and gersdorffite PI), and material with no 001 reflection in its powder pattern has weak optical anisotropism or is optically isotropic (cobaltite Pa3, gersdorffite Pa3 and P2 13 and ullmannite P2 13). Both the 001 reflection intensity (which represents the 001 series with 1 \ 2n and is the only observable member in the powder pattern) and the optical anisotropism strength are interpreted as being proportional to the crystal structure distortion from cubic symmetry. The weak optical anisotropism found in gersdorffite and ullmannite without detection of the 001 reflection in their powder patterns is interpreted as minor crystal structure distortion below the level detectable by powder X-ray diffraction, provided that optical anisotropism is not caused during the polishing process (Gibbons, 1967). In other words optical anisotropism appears to be a more sensitive method than powder X-ray diffraction for detecting crystal structure distortion. For example, the weak optical anisotropism in gersdorffite from Wolf sberg, which is also noted as numerous twin 90. lamellae (100) by Ramdohr (1950), is only shown by X-ray diffraction as a negligible effect on the single crystal structure, because the standard deviations of the crystal structure parameters are small. Optical anisotropism is related semi -quantitatively to the unit cell size. Materials with strong optical anisotropism have small unit cell sizes (comltite), materials with medium optical anisotropism have medium unit cell sizes (comltite and gersdorffite), materials with weak optical anisotropism have large unit cell sizes (gersdorffite and ullman nite) and optically isotropic materials have very large unit cell sizes (gersdorffite and ullmannite). The unit cell size is quantitatively related to the composition as shown in figure 2 (page 16) of Klemm (1965), where the unit cell size decreases with the substitution of cobalt for nickel. Therefore the substitution of cobalt for nickel is semi-quantitatively related to the crystal structure distortion. The unit cell size is quantitatively related to the disappearance of reflections 001 and 011 in the X-ray diffraction powder pattern upon heating, (thermal stability). The loss of these 001 and 011 reflections in the X-ray diffraction powder pattern is interpreted as crystal structure distortion release. After heating optically anisotropic gersdorffite powder samples to 600°C; samples with large unit cell sizes lost both reflections, 91. samples with intermediate unit cell sizes lost one reflection, and samples with medium unit cell sizes were unaffected. After heating cobaltite samples to 840° C; samples with medium unit cell sizes lost both reflections, samples with intermediate unit cell sizes were partially affected, and samples with small unit cell sizes were unaffected. These data indicate that the temperature of the crystal structure distortion release is quantitatively related to both the unit cell size and the chemical composition of the crystal structure. Therefore the substitution of cobalt for nickel increases the thermal stability of the distorted crystal structure. These thermal stability experiments also indicate order-disorder changes. Gersdorffite without the 001 reflection but with the 011 reflection in its X-ray diffraction powder pattern loses its 011 reflection when heated, which indicates an order-disorder change. In the non-cubic gersdorffite with large unit cell size, the crystal structure distortion release occurs before the order-disorder change. In the non-cubic gersdorffite with small unit cell size and in cobaltite, the crystal structure distortion release appears to be after the order-disorder change, and occurs at a higher temperature than the distortion release for non-cubic gersdorffite with large unit cell size. To check the validity of this hypo thesis, single crystal analysis is needed on material heated to just below the crystal structure distortion release temperature. Ullmannite decom - poses at 710°C before this order-disorder change may occur. A diagram- 92 .. matic sketch of these crystal structure results is given in figure 9. Most synthetic gersdorffites and ullmannites produced at 250°C, 400°c, 550°C and 700°C show the 011 reflection but no 001 reflection in their X-ray diffraction powder patterns. The intensity of the 011 reflection in the X-ray diffraction powder pattern represents the degree of non -metal atom ordering. This was shown by crystal structure powder analyses of gersdorffite, where ordering decreased with higher formation temperatures and with deviation from the stoichiometric composition. The strong 011 reflection and the absence of the 001 reflection in the X-ray diffraction powder pattern for ullmannite is shown by powder crystal structure analyses to represent an ordered undistorted crystal structure at all temperatures. The absence of 001 and 011 reflections in the X-ray diffraction powder pattern of cobaltite produced at 550° C by the simple tube method indicates a disordered undistorted crystal structure; whereas from a LiCl-KCl melt, both 001 and 011 reflections are present in the X - ray diffraction powder pattern of cobaltite which indicates an ordered distorted crystal structure. An explanation is needed for the differences found between the crystal structures of natural and synthetic materials. The heating experiments with natural materials indicate a slow solid state rate of crystal structure change as the temperature decreases; and since the transformations were not reversed, the actual transformation temperatures are lower than the temperatures used in the heating experiments. The production of synthetic material is slow as shown by its poor crystallinity after long periods at low temperatures, so there may have been insufficient time and lack of atom mobility to allow the formation of ordered and distorted crystal structures. These transformations of order-disorder and distortion release are second order, so these changes may involve a hysteresis cycle. Therefore the crystal structure differences between natural and synthetic materials are explained by slow solid state reaction rates and a possible hysteresis cycle. 94. ACKNOWLEDGMENTS The author wishes to express his sincere thanks to the following individuals: Prof. J. J. Frankel of the School of Applied Geology who made available the departmental facilities. Dr. N. L. Markham of the School of Applied Geology who gave many valuable suggestions and criticisms in his capacity as supervisor. Prof. N. C. Stephenson of the School of Chemistry who made available, and gave advice on the use of, single crystal X-ray diffraction equipment, the computer facilities of Deuce and IBM 360 in the School of Electrical Engineering, the IBMI620 in the Faculty of Applied Science, and the IBM7040 in the Australian Atomic Energy Commission at Lucas Heights. Mr. P. G. Embrey of the British Museum (B.M.), Prof. C. Frondel of the Harvard Mineralogical Museum (H.M.M.), Mr. P. E. Desautels of the United States National Museum (U.S. N. M.) and Mr. 0. Charmers of the Australian Museum (A . M.) who provided excellent natural materials. Mr. L. Rannit of the School of Chemistry who produced the spectrographic analyses. 95. REFERENCES Addison, W. E., 1961. Structural Principles in Inorganic Compounds, Longmans, Edinburgh, 183 p. Allen, E. T., and Crenshaw, J. L., 1914. Effect of temperature and acidity on the formation of marcasite and wurzite, a contribution to the genesis of unstable forms, Amer. J. Sci., 33. Bashenow, A. J., 1958. Kobalt- und Nickelmineralien auf der Lagerstatte Karagem, Ref. Zbl. Mineral. II, 1962, 95. Bernard, J. H., 1954. Gersdorffit z. Rudnan, Ref. Zbl. Mineral. I, 1958, 100. Berry, L. G., and Thompson, R. M., 1962. X-ray Powder Data for Ore Minerals, Geol. Soc. Amer. Mem. 85, 281 p. Bokii, G. B., and Tsinokev, L. I., 1954. X-ray examination of cobaltite, gersdorffite and ullmannite, Trudy Inst. Krist., Akad. Nauk. SSSR,9,239. (Abstr. C.A. 48, 13553H). Bonnemere, C., and Wintenberger, M., 1961. Crystal structure of arsenopyrite, Compt. Rend., 252, 1344. Bragg, W. L., 1913. The analysis of crystals by the X-ray spectrometer, Proc. Roy. Soc., A89, 468-489. Bragg, W. L., 194 7. The conversion factor for kX units to angstrom units, J. Sci. lnstr., 24, 27. 96. Bragg, W. L., and Claringbull, G. F., 1965. Crystal Structures of Minerals, Bell, London, 409 p. Buerger, M. J., 1936a. The symmetry and crystal structure of the minerals of the arsenopyrite group, Z. Krist., 95, 83-113. Buerger, M. J., 1936b. The crystal structure of the arsenopyrite group, Amer. Mineral., 21, 203. Buerger, M. J., 1936c. A systematic method of investigating super-structure applied to the arsenopyrite crystal structural type, Z . Krist., 94, 425-438. Buerger, M. J., 1937a. A common orientation and a classification for crystals based upon a marcasite-like packing, Amer. Mineral., 22, 48-56. Buerger, M. J., 1937b. Interatomic distances in marcasite and notes on the bonding in crystals of lollingite, arsenopyrite, and rnarcasite types, Z. Krist., 97, 504-513. Buerger, M. J. , 1939. The crystal structure of gudrnundite (F eSbS) and its bearing on the existence field of the arsenopyrite structural type, Z. Krist., 101, 290-316. Buerger, M. J. , 1960. Crystal Structure Analysis, Wiley, New York, 668 p. Busing, W.R., Martin, K. 0., and Levy, H. A., 1962. OR FLS, a fortran crystallographic least-squares program, Oak Ridge National Labor atory, TM305, 75 p. Busing, W.R., Martin, K. 0., and Levy, H. A., 1964. ORFFE, afortran crystallographic function and error program, Oak Ridge National Laboratorv. TM306. 85 o. 97. Cha yes, F., and Mackenzie, W. S., 1957. Experimental error in deter mining certain peak locations and distances between peaks in X-ray (powder) diffractometer patterns, Amer. Mineral., 42, 534-547. Chrystall, R. S. B., 1965. Thermal expansion of iron pyrites, Trans. Faraday Soc., 61, 1811-1815. Clark, A. H. , 1965. The composition and conditions of formation of arsenopyrite and lollingite in the Ylojarvi copper-tungsten deposit, Southwest Finland, Bull. Comm. Geol. Finlande, 217, 56 p. Clark, A. H., 1966. Heating experiments on gudmundite, Mineral. Mag., 35, 1123-1125. Clark, L. A., 1960. The Fe•As•S system: phase relations and applications, Econ. Geol., 55, 1345-1381, 1631-1652. Cohen, M. U., 1936. The calculation of precise lattice constants from X-ray powder photographs, Z. Krist., 94, 288-298. Cox, G. W., 1966. COXFLS - a crystallographic least-squares program for use at Lucas Heights, Aust. Atomic Energy Comm., MP/TN 25, 6 p. Craig, D. C., and Stephenson, N. C., 1965. The crystal structure of lautite, CuAsS, Acta Cryst., 19, 543-547. Doelter, C., 1926. Handbuch der Mineralchemie, Vol. 4, Part 1, Steinkopf, Dresden . Donnay, J. D. H., and Donnay, G., 1963. Crystal Data Determinative Tables, Amer. Cryst. Assoc. Monograph 5, 1302p. 98. Ekstein, H., and Siegel, S., 1949. Limits of accuracy in the determin ation of lattice parameters and stresses by the Debye-Scherrer method, Acta Cryst., 2, 99-104. Elliott, N., 1960. Interatomic distances in FeS2, CoS2, and NiS2, J. Chem. Phys., 33, 903-905. Evans, R. C., 1964. An Introduction to Crystal Structure, Univ. Press., Cambridge, 410 p. Ferguson, R. B., 1947. The unit cell of glaucodot, Univ. Toronto Studies, Geol. Ser. No. 51, 41-47. Florke, W., 1926. Zur Polymorphie des CoAsS, Cbl. Mineral., 337-338. Frenzel, G., and moss, F. D., 1966. Cleavage in pyrite, A1ut. Mineral., 52, 994-1002. Gibbons, G. S., 1967. Optical anisotropy in pyrite, Amer. Mineral., 52, 359-370. Giese, Jr., R. F., and Kerr, P. F., 1965. The crystal structures of ordered and disordered cobaltite, Amer. Mineral., 50, 1002•1014. Goll, J. J., 1937. Etude metallographique des minerais de nickel provenant des environs de Dobsina (Slovaquie), Ceska Akad. Bull., Internat., 38, 11-16. Green, H. S., and Hurst, C. A., 1964. Order-Disorder Phenomena, Interscience, New York, 363 p. 99. Hanawalt, J. D., Rinn, H. W., and Frevel, L. K., 1938. Olemical analysis by X-ray diffraction: Classification and use of X-ray diffraction patterns, Anal. Chem., 10, 457-512. Handbook of Chemistry and Physics, 1964. (ed. C. D. Hodgman, R. C. Weast, and S. M. Selby), Chem. Rubber, Cleveland, 3513 p. Harcourt, G. A., 1942. Tables for the identification of ore minerals by X-ray powder patterns, Amer. Mineral., 27, 63-113. Henry, N. F. M., Lipson, H., and Wooster, W. A., 1961. The Inter pretation of X-ray Diffraction Photographs, Macmillan, London, 282 p. Huggins, M. L., 1922. The crystal structure of marcasite, arsenopyrite and loellingite, Phys. Rev., 19, 369. Hulliger, F., and Mooser, E., 1965a. Semiconductivity in pyrite, marcasite and arsenopyrite phases, J. Phys. Chem. Solids, 26, 429-433. Hulliger, F., and Mooser, E., 1965b. The bond description of semi conductors: polycompounds, Prog. Chem. Solid State, 2, 330-377. International Tables for X-ray Crystallography, 1962 . Vol. 3, Kynock Press, Birmingham, 362 p. Jong, W. F., de, 1926. Bepaling van de absolute aslengten van markasiet en daarmee isomorfe mineralen, Physica, Nederl. Tijds. Naturrk., 6, 325-332. 100. Jong, W. F. de, 1928. Over de kristalstructuren van arsenopyriet, borniet, en tetraedriet, Gedrukt bij de Technische Boekhandel en D;rukkerij, Delft. (Abstr. Z. Krist., Strukturbericht, 1913-1928, p. 283) Kennicott, P. R., 1963. A modification of the Busing-Levy least-squares program to account for overlapped data, General Electric Report No. 63-Rl-(3321 G), 13 p. Klemm, D. D. , 1962. Anisotropieeffekte bei kubischen Erzminerlien, N. Jb. Mineral., Abh., 97, 337-356. Klemm, D. D., 1965. Synthesen und Analysen in den Dreieckdiagrammen FeAsS-CoAsS-NiAsS und FeS2 -eos2 -NiS2, N. Jb. Mineral., Abh., 103, 205-255. Klemm, D. D., and Weiser, T., 1965. Hochtemperierte Glanzkobalt Gersdorffit-Mischkristalle von Outokumpu/Finnland, N. Jb. Mineral. Mh., 236-241. Klug, H.P., and Alexander, L. E., 1959. X-ray Diffraction Procedures, Wiley, New York, 716 p. Kullerud, G., 1959. Sulfide systems as geological thermometers, in Abelson, P. H., ed. Researches in Geochemistry, Wiley, New York, 301-335. Kullerud, G., 1964. Review and evaluation of recent research on geo logically significant sulfi de-type systems, Fortschr. Mineral, 41, 221-270. 101. Kullerud, G. , and Yund, R. A . , 1962. The Ni -S system and related minerals, J. Petrology, 3, 126-175. Lange, W., and Schlegel, H., 1951. Die Zustandsbilder der Systeme Eisen-Antimon -Schwefel und Kobalt-Antimon-Schwefel, Z. Metallz., 42, 257-268. Lawrence, L. J., and Markham, N. L., 1963. The petrology and mineralogy of the pegmatite complex at Bismuth, Torrington, N. S. W., Geol. Soc. Aust., 10, 343-364. Lightfoot, B., 1931. On occurence of gersdorffite in Southern Rhodesia, Rhodesia Sci. Assoc. Proc., 30, 45-47. Lipson, H., and Cochran, W., 1966. The Determination of Crystal Structures, Bell, London, 414 p. Mc Ceery, C. L. , 1949. Improved mount for powdered specimens used on the Geiger-counter X-ray spectrometer, J. Amer. Ceramic Soc., 32, 141-146. Markham, N. L., 1965. Private communication. Mechling, M., 1921. Uber die Kristallstruktur des Kobaltglanzes, Abh. Math Phys. Kl. Sachs. Akad. Wiss. Leipzig, 38, 37-51. Mendel, G., 1962. Vapor transport of solids by vapor phase reactions, J. Phys. Chem. Solids, 23, 587-598. Morimoto, N., and Clark, L. A., 196 I. Arsenopyrite crystal-chemical relations, Amer. Mineral., 46, 1448-1469. 102. Nelson, J.B., and Riley, D. P., 1945. An experimental investigation of extrapolation methods in the derivation of accurate unit-cell dimensions of crystals, Proc. Phys. Soc., 57, 160-177. Niskanen, E., 1964. Reduction of orientation effects in the quantitative X- ray diffraction analysis of kaolin minerals, Amer. Mineral., 49, 705-714. Nuffield, E. W. , 1966. X -ray Diffraction Methods, Wiley, New York, 409 p. Olshausen, S. K., v., 1925. Strukturuntersuchungen nach der Debye Scherrer-Methode, Z. Krist., 61, 463-514. Oftedal, I., 1963. Note on the crystal structure of cobaltite, Norsk, Geol. Tidsskr., 43, 401-404. Onorato, E., 1957a. Rontgenographische Untersuchungen uber den Kobalt- glanz, N. Jb. Mineral., Abh., 91, 41-51. Onorato, E., 1957b. X-ray examination of cobaltite, Acta Cryst., 10, 764-765. Palache, C., Berman, H., and Frondel, C., 1944. The System of Mineralogy, Vol. 1, Elements, Sulfides, Sulfosalts, Oxides, Wiley, New York, 834 p. Parrish, W., 1960. Results of the I.U. Cr. precision lattice-parameter pro ject, Acta Cryst., 13, 838-850. Parrish, W., Ekstein, M. G., and Irwin, B. W., 1953. Data for X-ray Analysis, II. Tables for Computing the Lattice Constant of Cubic Crystals, Phillips' Tech. Library, 81 p. Pauling, L., and Huggins, M. L., 1934. Covalent radii of atoms and inter• atomic distances in crystals containing electron pair bonds, Z. Krist., 87, 205-238. 103. Peacock, M. A., and Berry, L. G., 1940. Rontgenographic observations on ore minerals, Univ. Toronto Studies, Geol. Ser. No. 44, 47-69. Peacock, M. A., and Henry, W. G. , 1948. The crystal structures of cobaltite (CoAsS), gersdorffite (NiAsS), and ullmannite (NiSbS), Univ. Toronto Studies, Geol. Ser. No. 52, 71•80. Pearson, W. B., 1965. Compounds with the marcasite structure, Z. Krist., 121, 449-462. Polushkina, A. P., and Sidorenko, G. A., 1963. A structural variety of cobaltite, Doklady Akad. Nauk. SSSR. 153, 1420-1423. (Abstr. Amer. Mineral., 1964, 49, 818.) Ramdohr, P., 1954. Klockmann 's Lehrbuch der Mineralogie, F. Enke-Verlag, Stuttgart. Ramdohr, P., 1960. Die Erzmineralien und ihre Verwachsungen, Akademie - Verlag, Berlin, 1089 p. Ramsdell, L. S., 1925. The crystal structure of some metallic sulfides, Amer. Mineral., 10, 281-304. Rollett, J. S., 1961. General programs for crystal structure analysis on the English Electric Deuce computer, in Pepinsky, R., Robertson, J. M., and Speakman, J. C. , ed. Computing Methods and the Phase Problem in X-ray Crystal Analysis, Pergamon Press, London, 87-101 p. Roseboom, E. H., 1962. Skutterudites (Co, Ni, Fe) As3_x: composition and cell dimensions, Amer. Mineral., 47, 310•327. 104. Rosenqvist, T., 1954. A thermodynamic study of the iron, cobalt, and nickel sulphides, J. Iron Steel Inst., 176, 37-57. Schafer, H., 1961. Olemische Transportreaktionen und ihre Anwendungen VII. Mitteilung uber chemische Transportreaktionen, Angew. Olem., 73, 11-14. Schneiderhohn, H., 1922. Anleitung zur mikroskopischen Bestimmung und Untersuchung von Erzen und Aufbereitungsprodukten, besonders im auffallenden Licht, Selbstverl. Ges. dt. Metallh. Bergl. Berlin. (Abstr. Econ. Geol., 1923, 18, 604). Shishkin, N. N., 1958. D"Lhulukite-a new cobalt mineral, C. R. Acad. Sci. USSR, 121, 724-726 (Abstr. M.A. 14-140). Shishkin, N. N. 1965. Not an orthorhombic variety of cobaltite but a high cobalt glaucodot, Zap. Vses. Mineral. Obshch., 94, 477-481. (Abstr. M.A. 17-689). Silbermann, J. R., Baljter, B. L. , and Erenburg, B. G. , 1958. Der Kobaltglanz aus der Lagerstatte Wladimirowskoe im Gornyj Altaj, Ref. Zbl. Miner. I, 1962, 95. Sipocz, L., 1886. Uber die chemische Zusammensetzung einiger seltener Minerale aus Ungarn, Z. Krist., 11, 209-219. Slater, J. C., 1964. Atomic radii in crystals, J. Olem. Phys., 41, 3199-3204. Stanton, R. L., 1955. The anisotropy of pyrite, Aust. J. Sci., 18, 98-99. Stanton, R. L., 1957. Studies of polished surfaces of pyrite and some implications, Can. Mineral., 6, 87-118. 105. Straumanis, M. E., 1955. The refraction correction for lattice constants calculated from powder and rotation crystal photographs, Acta Cryst., 8, 654-656. Straumanis, M. E., 1959. Absorption correction in precision determination of lattice constants, J. Appl. Phys., 30, 1965-1969. Straumanis, M. E., 1960. Some sources of error in precision determin ation of lattice parameters, Acta Cryst., 13, 818-821. Straumanis, M. E., Amstutz, G. C., and Chan, S., 1964. Lattice para meters and expansion coefficients of FeS2 (natural and synthetic) and of CoS2, Amer. Mineral., 49, 206-211. Strunz, H., 1957. Mineralogische Tabellen, Akademie•Verlag., Leipzig, 418 p. Strunz, H., 1966. Mineralogische Tabellen, Akademie-Verlag., Leipzig, 560 p. Takeuchi, Y., 1957. The absolute structure of ullmannite, NiSbS, Mineral. J., Japan, 2, 90-102. Thomson, E., 1921. A mineralographic study of the pyrite group, Univ. Toronto Studies, Geol. Ser. No. 12, 32-39. Toulmin, III, P., and Barton, Jr., P. B., 1964. A thermodynamic study of pyrite and pyrrhotite, Geochim. et Cosmochim. Acta, 28, 641-671. Urazov, G. G., Bol'shakov, K. A., Fedorov, P. I., and Vasilevskaya, I. I., 1960. The antimony-iron-sulphur ternary system. A contribution to the theory of precipitation smelting of antimony, Russian J. Inorganic Chem., 5, 214-218. 106. Van Hook, A., 1961. Crystallization Theory and Practice, Reinhold, New York, 325 p. Vies, A. de, 1965. On weights for a least-squares refinement, Acta Cryst., 18, 1077-1078. Waser, J ., 1951. Lorentz and polarization correction for the Buerger pre• cession method, Rev. Sci. lnstr., 22, 567-568. Wells, A. F., 1962. Structural Inorganic Chemistry, Clarendon Press, Oxford, 1055 p. Winterberger, M., 1962. Etude electrique et magnetique de composes sulfures et arsenies d'elements de transition. III. Propri~tes electriques et magnetiques et liaisons dans l 'arsenopyrite, la cobaltite et la loellingite, Bull. Soc. Franc. Mineral. Cryst., 85, 107-119. Wyckoff, R. W. G., 1963. Crystal Structures, Vol. I, Interscience, New York, 467 p. Yund, R. A., 1962. The system Ni-As-S: phase relations and mineralogical significance, Amer. J. Sci., 260, 761-782. Zachariasen, W., 1927. Strukturbericht I, 284. 107. Table la. Unit cell data for gersdorffite from 16 literature and 14 additional natural samples. Numbers in parentheses denote probable error in last figure. 5.59 No locality given Bokii and Tsinokev 1954 5. 593 Sudhlry, Ontario, C.anada Peacock and Berry 1940 5 o 616(1) II II II USNM R830 5.645 II II ff Peacock and Berry 1940 5 o 655 II tf tf " " 5 o 65 tf II II Peacock and Henry 1948 5.6728(3) + 5.68 (I) (1.0%) tf II BM 1917, 285 5.63(1) Cobalt, Ontario, Canada BM 1922, 145 5 • 63(1) + 5 o 72(1) (30%) II II AM 025174 5.68 Germany Ramsdell 1925 5.6827 Timagami, Ontario, Canada Yund 1962 5.6846(3) (amoibite) Leichtenberg, Fichtelgerbirge, Germany USNM R862 5.6870(8) Mitterburg, Salzburg, Austria BM 1929, 12 5.5910 II tf II Yund 1962 5.6885(3) Wolfsberg, Harz, Germany HMM 5.693(1) + 5.68(1) (30%) Hager a.d. Dill, Germany HMM 5.693 Musen, Westphalia, Germany Klemm 1965 5.694(1) II II II BM 1434 5. 693 Synthetic Yund 1962 5.693 " Klemm 1965 5.6941(5) Lobenstein, Russ, Germany BM 57562 5 .694 II tf tf Klemm 1965 5 o 70 II II tf Zachariasen 1927 5.69(1) + 5. 72(1) (60%) (dobscanite) Dobschau, Hungary USNM 0650 5.696 Rudoan, Slowakei Bernard 1954 5.6957(5) Farvic Mine, Gwanda, Sth. Rhodesia BM 1933,371 5.696(1) Cochabamba, Boliva BM 1959, 462 5. 7050(3) Ferro, Slovakia UNSW 5. 719 No locality given Olshausen 1925 5.7238 Schladming, Styria, Germany Yund 1962 108. Table lb. Unit cell data and other phases formed from 19 artificially pre- pared gersdorffites. Numbers in parentheses denote probable error in last figure. Composition Unit Cell Other Phases Fe Co Ni As s R I.00 1. 70 0.30 5. 730 (5) NiAs2 1.55 0.45 5. 725 (5) 1.40 0.60 5.717 (3) 1.25 0.75 5. 704 (3) 0.92 1.08 5.695 (3) 0.88 1.12 5.690 (3) 0.84 1. 16 5.691 (3) 0.70 1.30 5.687 (3) 0.08 o. 92 1.4 0.6 5. 710 (3) NiAs3 1.2 0.8 5. 702 (3) 1.0 1.0 5.691 (3) 0.9 I.I 5.688 (3) 0.1 0.9 1.2 0.8 5. 702 (3) CoAsS 1.0 1.0 5.692 (3) CoAsS 0. I 0.9 1.4 0.6 5. 709 (5) NiAs3 1.3 0.7 5. 703 (5) 1.2 0.8 5.692 (5) 1.0 1.0 5.692 (5) 0.9 I.I 5.690 (5) 109. Table le. Unit cell data for cobaltite from 23 literature and 6 additional natural samples. Numbers in parentheses denote probable error in last figure. 5.551 Hakansbo, Sweden Mee n in Peacock and Berry 1940 5.57 " " Berry and Thompson 1962 5.569 Karagem, Altai, U.S. S. R. Ba.shenow 1958 5.572 " " " " It 5.570 Skutterud, Norway Oftedal 1963 5.572 Synthetic Klemm 1965 5.574 Tunaberg, Sweden Mean in Peacock and Berry 1940 5.579 " " Klemm 1965 5 .581 (calc) " " Giese and Kerr 1965 5.59 " " Onorato 1957 5.60 " " Mechling 1921 5.575 No locality given Shishkin 1958 5.577 Riddarhytan Klemm 1965 5.5766(3) N.S. W., Australia UNSW 362 5.577 Cobalt, Ontario, Canada Meen in Peacock and Berry 1940 5.582 " " " Giese and Kerr 1965 5.591 (calc) " " " " " " 5.60 " " " Peacock and Henry 1948 5.577(2) Concurry, Qld., Australia UJ'.ISW 361 5.5780(5) Bimbowrie, S.A., Australia UNSW 61 5. 5790(5) Torrington, N. S. W., Australia UNSW 233 5.579(2) Mt. Cobalt, Qld., Australia AM D24919 5.58 No locality given Ramsdell 1925 5.58 Modum, Norway Zachariasen 1927 5.58 Gornyj, Altai, U. S. S. R. Silbermann et al. 1958 5.617 " " " " " 5.581(1) South Broken Hill, N. S. W. , Australia UNSW 5.61 No locality given de Jong 1926 5.67 " " " Bokii and Tsinokev 1954 110. Table Id. Unit cell data and other phases formed from 31 artificially pre- pared cobaltites • Numbers in parentheses denote probable error in last figure. Composition Unit Cell Other Phases Fe Co Ni As s R 1.0 0.98 1.02 5.5766(2) CoAs2 0.94 1.06 5.5767(2) CoAs2 0.90 1.10 5.5767(2) CoAs2 0.86 1. 14 5.5766(2) 0.82 1.18 5.5748(4) 0.78 1.22 5.5732(4) 0.74 1.26 5.5716(4) 0.66 1.34 5.5694(6) 0.61 1.39 5. 5681(6} 0.56 1.44 5.5660(6) 0.54 1.46 5 .5658(6) 0.52 1.48 5.5652(6) 0.46 1.54 5 .5640(8) 0.44 1.56 5.5638(8) 0.43 1.57 5. 5623(8) 0.42 1.58 5.5630(8) CoS2 0.40 1.60 5.5630(8) CoS2 0.90 0.10 I.I 0.9 5.5776(3) CoAs2 0.9 I.I 5. 5777(3) 0.8 1.2 5.5746(5) 0.66 1.34 5.5698(5)' o. 85 o. 15 0.85 1.15 5.5765(5) NiAsS 0.70 1.30 5.5754(8) NiAsS 0.55 1.45 5.567 (1) NiAsS 0.1 0.9 I.I 0.9 5.5773(3) CoAs2 0.9 I.I 5.5773(3) CoAs2 0.8 1.2 5.5774(5) 0.7 1.3 5.5716(8) 0.66 1.34 5.5670(8) 0.6 1.4 5.5660(8) 0.55 1.45 5.554 (I) Fe S2 111. Table le. Unit cell data for ullmannite from 15 literature and 4 additional natural samples. Numbers in parentheses denote probable error in last figure. 5.69 No locality given Bokii and Tsinokev 1954 5. 724 (corynite) No locality given Peacock and Berry 1940 5. 735 " " " " Meen 1940 in Berry & Thompson 1962 5. 755 " Olsa, Carinthia, Austria Berry and Thompson 1962 5. 8 76 (1) Harzgerode, Hartz Mts., Germany BM 69114 5.881 Sakhendorf, Siegerland, Germany Takeuchi 1957 5. 882 Salchendorf, Westphalia, Germany Berry and Thompson 1962 5.899 " " Peacock and Berry 1940 5. 89 Siegen, Germany Zachariasen 1927 5. 894(1) Petersbach, Siegen, Germany AM 09984 5. 905 Montenarba, Sardinia, Italy Peacock and Berry 1940 5.91 Prussia, Germany Harcourt 1942 5.91 No locality given Ramsdell 1925 5.915 (kallilite) Obersdorf, Siegen, Germany Peacock and Berry 1940 5 • 92 7 " " " " Berry and Thompson 1962 5.918 (willyamite) Broken Hill, N.S. W., Australia Peacock and Berry 1940 5. 9203(5) " " AM 017751 5. 9203(5) " " AM 027860 5.930 " Berry and Thompson 1962 112. Table If. Unit cell data from 49 artificially prepared ullmannites and gersdorffites. Numbers in parentheses denote probable error in last figure. Comeosition Uni~ Cell Comeosition Unit Cells Ni Sb As s R Co Ni Sb As s .R R 1.0 I.I 0.9 5. 9355(3) 1.0 0.6 0.2 1.2 5. 868(2) 1.0 1.0 5. 9355(3) 0.3 I.I 5. 854(2) 0.99 1.01 5. 9346(3) 0.4 1.0 5. 846(2) 0.98 1.02 5. 9344(3) 0.6 0.8 5. 855(3) 5. 767(5) 0.96 1.04 5. 9344(3) 0.8 0.6 5. 872(3) 5.811(5) 0.94 1.06 5. 9344(3) 1.0 0.4 5. 866(3) 5.809(5) 0.9 I.I 5. 9344(3) 1.0 0.5 0.5 1.0 5.842(3) 5. 720(3) 1.0 I.I 0.2 0.7 5. 9390(5) I.I 0.3 0.6 5. 9266(5) 1.0 0.4 0.4 1.2 5. 806(3) 5. 743(3) 0.6 1.0 5. 843(3) 5. 730(3) 1.0 1.0 0.1 0.9 5.9316(5) 0.8 0.8 5. 843(3) 5. 753(3) 0.2 0.8 5. 9282(5) 1.0 0.6 5.874(3) 5.822(3) 0.3 0.7 5. 9259(5) 1.2 0.4 5.8130(8) 0.4 0.6 5. 9226(8) 0.6 0.4 5. 9165(8) 1.0 0.3 0.7 1.0 5. 834(5) 5. 725(1) 0.8 0.2 5.88 (1) 0.9 0.8 5.820(5) 5. 7436(8) 1.0 o. 7 5. 840(5) 5. 757(1) 1.0 0.8 0.2 1.0 5.8994(8) 0.4 0.8 5.894(1) 1.0 0.2 0.6 1.2 5. 7020(5) 0.6 0.6 5. 907(2) 0.8 1.0 5. 7305(5) 0.8 0.4 5.899(3) I.0 0.8 5. 7360(5) 1.0 0.2 5.88 (1) 1.2 0.6 5. 7493(5) 1.4 0.4 5. 7650(5) 1.0 o. 7 0.2 I.I 5.877(1) , 0.3 1.0 5.861(1) 1.0 0. I 0.9 1.0 5. 7105(3) 0.1 0.9 1.0 1.0 5. 9260(8) 0.2 0.8 5. 911(1) 0.3 0.7 5. 899(3) 0.4 0.6 5.885(3) 0.5 0.5 5.89 (1) 0.6 0.4 5.89 (1) 113. Table lg. Unit cell data for arsenopyrite , glaucodot, and gudmundite from 21 literature and one additional natural samples. Numbers in parentheses denote probable error in last figure. Arsenopyrite a b C J3 3.80 4.52 5.15 Huggins 1922 6.44 9.52 5.64 (danaite) Sulitjelma Norway de Jong 1926 9.51 5.65 6.42 90 Spindelmuhle, Czecho• slovakia Buerger 1936a 9.55 5.71 6.42 90 Franklin, U. S. A. " 5.828 5.720 5. 792 113.20 Synthetic Morimoto and Clark 1961 5.753 5.690 5. 787 112.2 Llallagua, Bolivia " 5.738 5.680 5.782 112.1 Calceranica, Italy " 5.744 5.675 5.785112.17 Freiberg, Germany " 5.737 5.667 5. 770 111. 7 Nogare, Italy " 5.704 5.663 5. 785 111.4 Boliden, Sweden " 5.771 5.659 5. 736 111.20 (danaite) Sulitjelma, Norway Klemm 1965 5.749 5. 724 5.812 113.20 Synthetic " 5.881 5.665 5 . 698 111. 80 Freiberg, Germany " 5.839 5.684 5. 752 112.20 St. Christoph " Glaucodot 9.64 5.75 6.68 Hakansbo, Sweden de Jong 1926 6.63 28.33 5.63 " Ferguson 1947 6.032 5.648 5 • 545 109. 55 " Klemm 1965 6.80 27.75 5.62 (alloclasite) Orawitza, Hungary Berry and Thompson 1962 5. 716(3) 5.646(3) 5. 781(3) 110.34(8) Hakansbo, Sweden BM Gudmundite 10.04 5. 93 6.68 90 Gudmundstorp, Sweden Buerger 1936a 10.00 5.93 6. 73 90 " Buerger 1939 6.02 5. 93 6.02 112 .13 II Buerger 1939 114. Table 2a. X-ray diffraction powder intensities for gersdorffite literature and natural samples. Optical anisotropy, zoning and paragenesis for natural samples. hkl 100 110 111 200 210 211 220 300 310 311 222 Sample No. 221 I. USNM R830 10 5 10 100 75 80 15 1 3 75 5 2. BM 1917,285 0 0 100 35 30 10 1 1 15 1 3. BM 1922, 145 6 3 6 90 100 90 30 1 1 40 6 4. AM D25174 0 0 0 55 100 70 30 1 15 35 10 5. USNM R862 11 2 5 100 70 35 10 1 1 35 11 6. BM 1929, 12 1 9 2 100 50 25 9 1 8 30 3 7. HMM Wolfsberg O 6 3 35 100 75 10 3 2 25 3 8. HMM Hager 3 10 6 80 100 60 20 3 3 60 4 9. BM 1434 0 8 4 100 65 60 20 1 7 30 2 10. BM 57562 2 4 2 100 55 20 25 1 3 50 1 11. USNM D650 1 30 100 60 10 0 40 3 12. BM 1933, 371 1 35 45 100 15 1 70 13. BM 1959, 462 7 4 75 75 45 20 1 1 100 3 14. UNSW 0 90 100 55 30 0 50 3 15. a=5.66,Sudbury 6 9 8 3 10 2 16. a=5 .694, artificial 4 6 60 100 90 35 80 10 Yund(l961) Anisotropy Zoning Paragenesis 1. medium pyrrhotite, calcite 2. nil medium niccolite, chalcopyrite, pyrrhotite, pyrite 3. weak medium calcite \magnetite, chlorite 4. weak strong skutterudite, sphalerite, siderite, calcite 5. medium medium pyrite 6. weak strong skutterudite, dolomite 7. weak chalcopyrite, pyrrhotite, calcite, quartz 8. weak medium pyrrhotite, galena, quartz 9. v. weak medium jamesonite, siderite 10. weak medium sphalerite, siderite ll. nil weak skutterudite, safflorite, siderite, quartz 12. nil weak safflorite, sphalerite 13. nil nil siderite, quartz 14. nil strong siderite, dolomite 115. Table 2b. X-ray diffraction powder intensities for cobaltite literature and natural samples. Optical anisotropy, zoning and paragenesis for natural samples. hkl 100 110 111 200 210 211 220 300 310 311 222 Sample No. 221 1. UNSW 10 5 20 80 100 80 20 2 2 20 5 2. UNSW 61 10 2 5 80 80 80 20 5 2 100 20 3. UNSW 362 50 5 5 100 50 20 8 10 1 50 1 4. UNSW 233 50 4 2 100 20 20 5 2 1 50 2 5. UNSW 361 50 5 5 100 90 40 15 3 1 40 2 6. AM 024919 20 5 2 100 50 50 10 5 2 20 2 7. a=5.582, 2 1 2 8 10 9 6 1 1/2 10 1 Cobalt 8. a=5.581 (calc) 2 1 2 8 10 9 3 10 1/2 Tunaberg 9. a=5.591 (calc) 3 7 10 9 2 10 1/2 Cobalt 10. a=-5. 60, Cobalt 1/2 6 10 8 2 9 1/2 11. a=5.57, 1 1/2 1/2 5 10 7 3 2 9 1 Hakansbo Anisotropy Zoning Paragenesis 1. medium nil safflorite, quartz, mica, chlorite 2. medium nil quartz 3. strong nil 4. strong nil quartz, mica, chlorite 5. strong nil quartz 6. strong nil quartz 116. Table 2c. X-ray diffraction powder intensities for ullmannite literature and natural samples. Optical anisotropy, zoning and paragenesis for natural samples. hkl 100 110 111 200 210 211 220 300 310 311 222 Sample No. 221 1. BM 69114 20 5 50 100 50 10 0 5 20 2 2. AM D9984 20 10 80 100 20 8 0 10 50 8 3. AM D27860 20 8 20 100 50 20 1 8 50 8 4. AM D17751 50 8 80 100 50 10 1 10 50 10 5. a=5. 882, West- phalia 3 1 2 10 6 2 1 7 1 6. a= 5. 775, Olsa 3 1 6 10 6 3 1 8 2 7. a=5.91, Prussia 5 1 5 10 8 3 3 3 8 1 Anisotropy Zoning Paragenesis 1. weak medium pentlandite, chalcopyrite, pyrrhotite, galena, quartz 2. nil nil chalcopyrite, tetrahedrite 3. nil nil 4. weak nil 117. Table 2d. X-ray diffraction powder intensities for arsenopyrite, glaucodot and gudmundite literature and natural samples. hkl d -I ~ I d I g ! d I 010 5.66 5 5.58 1/2 5.63 1 100 5.34 3 110 3.89 3 3.95 3 4.09 5 111 3.658 40 3.64 1 3.62 5 3.57 5 3.87 1/2 101 3.29 2 3.01 1/2 3.02 1/2 3.00 2 102 2.866 5 2.85 2 2.93 1/2 2.95 2 020 2.838 30 2.83 1 2. 82 20 2.81 4 111 2.798 10 2.83 50 002 2.678 100 2. 72 10 2. 71 50 2.75 10 2.68 3 200 2.660 100 2.68 50 2.64 1/2 2.63 1 112 2.558 10 2.54 10 2.55 10 121 2.440 90 2.45 8 2.43 50 2.47 8 012 2.422 95 2.43 7 2.45 100 210 2.408 95 2.42 20 2.39 6 2.31 1 2.32 1/2 212 2.204 25 2.18 1 2.18 10 2.21 1 102 2.096 20 2.13 1/2 2.13 1 2.13 2 2.07 1/2 122 2 .016} 2.02 1/2 2.01 5 2.02 1 221 2.010 5 2.00 2 112 I.966} 5 211 1.960 1.98 1 022 1.948} 1.964 2 1.96 5 1.961 3 1.917 7 220 1. 941 25 1.95 3 1: Arsenopyrite, Freiberg, Morimoto and Clark (1961) 2: Glaucodot, Hakansbo, Berry and Thompson (1962) 3: Glaucodot, Hakansbo 4: Alloclasite, Orawitza, Berry and Thompson (1962) 5: Gudmundite, Gudmundstop, Berry and Thompson (1962) 118. Table 3a1. Observed hkl reflection amplitudes (Fo) and calculated structure factors (Fe) of gersdorffite from Slovakia with space group Pa3. hkl Fo Fe hkl Fo Fe hkl Fo Fe 020 99 84 151 91 74 136 29 30 040 60 65 161 36 38 146 <8 1 060 10 18 122 <8 4 222 71 55 012 98 110 132 57 54 232 <8 4 022 61 64 142 <8 0 242 50 39 032 90 92 152 58 47 252 <8 2 042 35 38 162 <8 2 262 31 25 052 63 55 123 69 65 233 56 50 062 27 26 133 12 11 243 78 67 014 •8 7 143 "8 5 253 28 31 024 42 38 153 <8 0 263 <8 7 034 <8 7 163 32 36 234 <8 0 044 116 116 124 76 70 244 42 33 054 <8 4 134 <8 4 254 <8 0 064 23 30 144 <8 5 235 56 45 016 48 53 154 <8 4 245 46 42 026 28 26 164 25 41 236 <8 2 036 52 51 125 40 39 333 79 62 046 22 30 135 <8 0 343 <8 4 111 14 15 145 <8 3 353 46 49 121 80 76 155 <8 6 344 <8 5 131 79 101 126 <8 7 fit 49 !1 141 <8 5 119. Table 3a2 • Atomic parameters (x), thermal parameters (B), interatomic distances, tetrahedral and octahedral angles with standard deviations, and atomic radii for gersdorffite from Slovakia with space group Pa3. Numbers in parentheses denote number of angles or distances. Atom Point --X Atomic Position Radii (R) Ni 4a 0.0000 1.2+ 0.1 1.17 s + As 8c: 0.3785+ 0.0005 0.9+ 0.1 1.20 Interatomic distances (R) Ni-S+As(3) 2.372 + 0.002 S+As•s+As(l) 2.400 + 0.002 0 Tetrahedral angles { ) Ni-s+As-s+As(3) 100.8 + 0.1 Ni-s+As-Ni(3) 116. 5 + 0.1 Octahedral angles (0 ) S+As-Ni-S+As(3) 85.1 + 0.1 S+As-Ni-s+As(3) 94. 9 + 0 .1 120. Table 3b1 · Observed hkO reflection amplitudes (Fo) and calculated structure factors (Fe) of gersdorffite from Wolfsberg with space group P2 13. hk Fo Fe hk Fo Fe hk Fo Fe 02 86 107 27 41 43 51 16 19 04 35 34 31 16 18 52 17 18 06 19 18 32 20 17 53 c8 2 11 34 24 33 24 23 54 <8 2 12 64 45 34 <8 4 55 <8 1 13 16 19 35 17 20 56 17 15 14 16 16 36 17 16 60 19 18 15 26 17 37 <8 1 61 40 40 16 <8 4 40 35 34 62 38 33 17 <8 2 41 13 13 63 42 42 20 86 107 42 36 45 64 45 41 21 76 93 43 11 12 65 12 19 22 82 79 44 122 106 71 17 18 23 97 84 45 <8 7 72 <8 7 24 39 45 46 37 41 73 <8 1 25 37 37 47 13 8 74 <8 4 26 29 33 121. Table 3b2 • Atomic parameters (x)., thermal parameters (B), interatomic distances, tetrahedral and octahedral angles with standard deviations, and atomic radii for gersdorffite from Wolfsberg with space group P2 13. As atom is As0 . 82s0 • 18 . Numbers in parentheses denote number of angles or distances . Atom Point X Atomic Position Radii (R} Ni 4a -0.0065 + 0.0005 1.4+ 0.1 1.22 s 4a 0.3825 + 0.0010 1.0 + 0.2 I. 12 As 4a 0.6164 + 0.0005 I.6+0.1 1.19 Interatom ic distances cR> Ni -S( 3) 2.338 + 0.006 Ni-As(3) 2.409 + 0.003 S-As(l) 2. 306 + 0.010 0 Tetrahedral angles ( ) Ni-As•S(3) 102.4 + 0.1 Ni•As-Ni(3) ll5.5+0.l Ni-S-As(3) 101.4 + 0.2 Ni•S-Ni(3) 116.2 + 0. l Octahedral angles (0 ) As•Ni-S(3) 85.3 + 0.2 S-Ni-As(3) 85.6 + 0.1 As-Ni-As(3) 92. 7 + 0. l S-Ni-S (3) 96.3 + 0.2 1zz. 'i':--.hl~ ,c,. oo 3 erved h'd reflection a1"1pl1 tud~e (Fo) and calculated otructu.re ~;:-(,c) tor hkl, nil, h~I and hAl rrflection~ o! gersdorffite from Lo 1c~tenberg with apace group Pl. Unobserve~ hkl are not entered under Fo. "'.,, .. 1 Fr, ~~ '!"'~ ,...,~ "?e ----·------·- - -- : . '· 'j ) , ~:,. 1 •. \ ~' _'t • 1• 1 r,,. r : , .:_ .L;, I. ,.j ,,.. l , ' ;··_\j, 'J 1 6 3·~ .• (, 5 '-' ,, I _~-,,'. J') '/, \5 36 35 i:5 47 41 40 41 40 Cu~ 53 4'', ~: 3 ~;:-' ~,:., 'J2 9;·· r :· L .' :..6 8 8 7 ~2~ 5 ~ 4 5 3 -i,,:\) 15 8 8 c.~s "' o }41 6 7 11 6 11 oos 9 11 115 qo 69 ~7 ~0 61 2?6 26 20 ?? 2? ?3 C·J7 l 116 33 30 29 30 31 2.:,0 8 3 10 342 6 6 3 6 l OJ.J :o l 1?0 59 91 91 231 55 58 57 58 59 343 6 5 . 6 6 6 011 14 5 3 121 53 67 67 67 66 232 15 8 11 7 8 344 5 10 9 11 10 ou> 73 9? 93 122 14 7 11 9 9 233 38 40 41 39 41 345 3 6 4 4 3 013 11 10 4 123 61 59 59 60 60 234 6 9 1 1 4 350 5 13 014 24 9 10 124 75 58 57 59 56 235 55 41 40 40 43 351 4_ 4 3 2 015 13 6 9 1?5 36 30 31 ?9 29 236 8 6 6 .7 352 43 43 43 42 42 o::.i; 52 ~3 44 :~~ 10 15 9 11 13 2t.0 33 31 31 353 40 tl 40 40 39 c,-., .~". r,:: - .• , ,..:, -: i.. ') ?• 1 ~~ ~~ 5~ ~q 5R ,54 G 10 4 7 6 •·•·· (' :1 C. : '• ;_1 ~, .~ • •' . r. ~. :.;. ~-.· J li;, )·; '1 r. , , '• 2 3 -:: 3 :, :)) 17 l) lJ l~ ~1 '.' I ,,I) :1) ;,) jl, }J 3u2 3 1 5 5 6 c.:,, 31; j'.) ,:;'; l}~ 15 6 12 9 8 2~5 34 31 32 31 31 400 71 48 (,25 9 2 2 135 6 3 4 4 4 250 2 2 401 17 14 13 026 2J 22 23 1)5 22 2~ 25 25 23 251 31 29 29 30 30 402 38 35 33 03') 10 5 l~O 12 9 8 252 5 7 5 5 6 403 19 12 12 0 3:.. 11 3 l 141 11 7 7 9 8 253 23 23 25 24 24 404 99 97 97 c ~-' -~ '. eo n 1~2 6 6 1 G 2 254 4 2 8 4 4 405 6 11 8 O)j :.: 10 4 143 9 9 7 9 255 2 25 24 26 23 406 28 31 31 C:; 20 10 8 1~4 9 7 8 8 260 23 23 22 410 4 11 15 C -; 9 7 9 145 5 2 5 4 4 261 9 14 10 15 9 41112 9 111110 c:i 45 45 44 150 30 8 11 262 24 20 21 23 23 412 64 57 58 59 57 o.,J ~9 ,9 151 73 69 67 67 68 263 14 11 13 10 14 413 15 8 10 9 11 C .i 15 9 13 152 41 41 40 41 41 270 2 2 1 41,,4 7 10 8 10 7 C,. _,, 3;:, 33 34 153 4 6 4 2 272 7 5 5 5 7 415 6 8 8 10 11 o.} 6 7 7 154 7 9 11 7 300 10 3 416 17 33 34 33 33 c.;; 85 98 98 155 9 8 8 8 301 10 6 420 24 34 34 0.:5 8 8 0 160 37 43 43 302 6 6 8 421 6 5 7 2 3 o.;G ?4 30 29 161 28 31 32 32 30 303 10 5 422 35 35 36 36 35 oi:;o 6 6 162 6 4 3 4 304 7 7 6 423 5 5 2 0 O:il 6 4 2 163 2q 31 31 31 32 305 6 8 7 424 45 32 30 31 31 01;2 48 43 43 170 15 6 13 306 4 6 6 425 5. 3 4 6 5 c,.3 5 8 2CO 63 90 307 5 4 3 430 6 7 6 o~.~ 7 7 ~01 79 93 92 310 6 4 2 43110 12 1110 13 C'.i5 6 5 202 68 63 63 ·311 56 89 89 90 90 432 53 58 57 58 57 0(-:J 12 12 203 87 79 80 312 80 61 60 60 60 433 5 11 6 8 0.;1 6 6 204 35 32 3:? 313 15 17 18 18 19 434 10 10 11 9 12 0:.2 28 2) 23 205 49 43 42 314 13 10 10 9 9 435 4 10 7 7 8 c.:.3 7 7 206 30 23 23 315 6 5 5 4 · 5 440 96 98 97 cs.; 21 30 29 207 3 45 45 316 36 32 32 32 33 441 7 1, 10 12 10 0~5 7 7 208 16 17 17 320 75 82 82 442 35 31 30 32 32 l0J ll l 210 12 6 l:? 321 44 42 41 40 40 443 7 11 13 12 13 ~.:,: l& o 4 211 60 65 66 67 66 322 10 9 4 7 8 444 35 33 33 33 33 l-~2 19 7 9 212 20 10 12 ll 16 323 44 38 40 38 40 450 6 6 1 l-- :-, 6 5 l 213 47 42 43,243 324 74 61 60 61 59 451 3 , 6 4 1c.; 6 10 12 214 6 9 2 l 5 3?5 26 22 '5 23 21 452 33 32 32 31 31 105 5 l 215 ~3 40 43 41 43 326 9 ll 13 14 11 453 9 8 5 5 4 10.; 7 7 116 7 4 6 , 330 7 7 6 460 22 29 1.07 , 3 2 220 52 63 63 331 ?l 17 18 19 18 123. Table 3c2 • Atomic parameters (X, Y, Z), thermal parameters (B), inter atomic distances, tetrahedral and octahedral angles for gersdorffite from Leichtengerg with space group PI. Numbers in parentheses denote number of angles or distances. Atom Point X y z ~R2> Position Ni la 0.995+ 0.002 0. 988+ 0.001 0.986+ 0.001 1.5 + 0.1 Ni la 0.494+ 0.002 0.007+ 0.001 0.499+ 0.001 1.4 + 0.1 Ni la 0.013+ 0.002 0. 503+ 0.001 0.487+ 0.001 1.4+0.I Ni la 0.512+ 0.002 0.487+ 0.001 0.004+ 0.001 1.4 + 0.1 As +S la 0.387+ 0.001 0.378+ 0.001 o. 38o+ 0.001 1.1+0.l As +S la 0.618+ 0.001 0. 113+ 0 • 001 0.874+ 0.001 1.0+0.1 As +S la 0.121+ 0.001 0.872+ 0.001 0.619+ 0.001 I.0+0.1 As +S la 0.622+ 0.001 0.617+ 0.001 0.617+ 0.001 0.9+0.l As+ S la 0.114+- 0.001 o. 379+ 0.001 o. 878+ 0.001 0.9+0.l As +S la 0.387+ 0.001 o. 875+ 0.001 0. 114+ 0.001 1.0+0.I As +S la 0.887+ 0.001 0.112+ 0.001 0.381+ 0.001 1.0 + 0.1 As t 5 ta o. 8 80! o. 001 Q. 616t 0, OOL O. ll+± 0°001 l.2 t 0.1 Interatomic distances cR> Ni·S+As(3) 2.30-2.45 s+As-s+As(l) 2.33 Tetrahedral angles ( 0 ) Ni-s+As-s+As(3) 99-103 Ni·S+As•Ni(3) 115-117 Octahedral angles ( 0 ) s+As-Ni•S+As(6) 84-87 s+As-Ni•S+As(6) 91-98 124. Table 3d. Observed hkl reflection amplitudes (Fo) and calculated structure factors (Fe) of synthetic arsenic-rich gersdorffite formed at 550°c and 700°C with atomic parameters (x), thermal parameters (B), and their standard deviations. 550°c 700°C hkl Fo Fe Fo Fe 200 92 92 92 88 400 75 76 67 66 ll0 5 4 <5 0 220 72 70 64 62 440 144 143 97 97 lll ll 10 13 6 2ll 85 86 76 80 311 112 113 94 97 221 8 9 9 7 331 4 17 9 11 222 59 62 52 51 422 47 49 35 33 332 59 56 51 45 533 55 67 444 21 43 Seace Point DiscreEancy Teme. Group Atom Position X B(A o2) factor R 550°c P2 13 Ni 4a 0.986-t-0.005 -0.6-t0.4 0.074 4a 80.27As0. 73 0.384-t-0.003 o. 5+1.6- 80.18As0.82 4a 0.623-t0.002-- 0. 5+1.6 700°C Pa3 Ni 4a 0.00 2.0-t-0.4 0.051 80.23As0. 77 4a 0.38tt0.001 1.9-t0.3 125. Table 3e. Observed hkl reflection amplitudes (Fo) and calculated structure factors (Fe) of synthetic gersdorffite formed at 550°C and 700°c with atomic parameters (x), thermal parameters (B), and their standard deviations. 550°c 700°C hkl Fo Fe Fo Fe 200 84 85 81 81 400 68 52 43 43 ll0 14 22 38 23 220 63 62 73 60 440 99 98 75 75 111 14 25 23 22 2ll 68 69 77 65 311 95 95 87 87 221 9 9 15 14 331 13 15 20 19 222 51 59 53 59 422 30 35 33 33 622 17 22 17 17 332 42 42 35 35 533 41 41 38 30 444 30 36 21 19 S~ace Atom Point 2 Discrepanci Temp. Group Position -X B 700°C P2 13 Ni 4a 0.992-t0.005 2.3-t-0.3 0.083 s 4a 0. 369+0. 003 l.7±0. 3 As 4a 0.616-t0.001 2.8-t-0.2- 126. Table 3f. Observed hkl reflection amplitudes (Fo) and calculated structure factors (Fe) of synthetic sulphur-rich gersdorffite formed at 550°c and 700°c with atomic parameters (x), thermal parameters (B), and their standard deviations. 550°C 700°C hkl Fo Fe Fo Fe 200 93 93 80 78 400 50 35 38 38 110 8 8 34 11 220 72 70 94 69 440 111 108 99 98 111 25 25 33 31 211 59 60 60 60 311 95 96 109 92 221 11 11 16 16 331 32 32 30 30 222 65 64 51 62 422 42 48 40 48 622 25 40 36 38 332 43 38 36 36 533 54 55 38 43 444 15 15 48 25 Seace Point DiscreEancr Group Atom Temp. Position -X B 550°C P2 13 Ni 4a 0.995-+-0.003 0.0-+0.1 0.069 50.82As0.18 4a 0.380-+0.004 1.4-t0.6 50.48As0.52 4a 0.615-+-0.002 0.3-t0.3- 700°C P2 13 Ni 4a 0.998-t0.004 0.9-t0.2 0.142 SO. 82Aso. 18 4a 0.357-t0.008 I.i+o.9- 5o.48As0.52 4a 0.612-t0.002- -0.3-t0.4 127. Table 3g. Atomic parameters (x), thermal parameters (B), interatomic distances, tetrahedral and octahedral angles with standard deviations, atomic radii, scale factor and discrepancy factor R for cobaltite with space group Pa3 followed by values of Giese and Kerr (1965). Numbers in parentheses denote number of angles or distances. Point Atom Position -X -X Co 4a 0.0000 0.000 s+As Be 0.3802 + 0.0002 0.380 Atom BCR2> BCR2> Atomic Atomic Radii(~) Radii (R) Co -o.o6+o.o5 0.0 1. 16 1. 16 s+As -0.lo+o.04 0.0 1. 16 1.16 Interatomic distances (R) Co-s+As(3) 2. 323+o. 001 2.31 s+As-s+As(l) 2.317+o.002 2.31 0 Tetrahedral angles ( ) Co-s+As-Co(3) 116. 3+ 0.1 Co-s+As-s+As(3) 101.2:±_0. l 0 Octahedral an&les ( ) s+As•Co-S+As(3) 85.2+o.l s+As•Co-s+As(3) 94.8+o.l Scale factor 0.9429 DiscreQancy factor R 0.055 0.057 128. Table 3h. Atomic parameters (X, Y, Z), thermal parameters (B), inter- atomic distances, tetrahedral and octahedral angles with standard deviations, scale factors and discrepancy factor R for cobaltite with space group Pca21 derived from data of Giese and Kerr (1965). 0 Tetrahedral angles ( ) As-S-Co 102+1 S-As-Co 99+1 As-S-Co 103+1 S-As-Co 99+1 As-S-Co 104+1 S-As-Co 101+1 Co-S-Co 113+1 Co-As-Co 115+1 Co•S-Co 115+1 Co-As-Co 117+1 Co-S-Co 118+1 Co-As-Co 119+1 0 Octahedral angles ( ) S-Co-S 86+1 S-Co•As 83+1 S-Co-S 88+1 S-Co-As 88+1 S-Co-S 95+1 S-Co-As 92+1 As-Co-As 82+1- S-Co•As 94+1 As-Co-As 85+1 S-Co-As 95+1 As-Co-As 94+1 S-Co-As 98+1 Interatomic distances (R) As-S 2.31+o.02 s-co 2.25+o.0l 2.26+o.0l 2.37+o.03 As-Co 2.28+o.0l 2.39+o.0l 2.4o+o.0l Point Atom Position- X y z B(R2> Co 4a 0.006+o.001 0.24o+o.001 0.014+o.006 0.o+o.1 As 4a 0.381+o.001 0.63o+o.001 0.38o+o.007 -o.1+o.1 s 4a 0.62o+o.001 0.869+o.001 0.62o+o.006 0.l+o.2 Scale factors 0.903-hk0, 0.925-h0l Discrepancy factor R 0.112 129. Table 3j 1• Observed hkO reflection amplitudes (FoT) and calculated structure factors (FcT) by Takeuchi (1957), and calculated structure factors (Fe) of ullmannite with space group P213. hk FoT FcT Fe hk FoT FcT Fe 11 14.6 13.5 13.0 36 15.7 14.7 13.8 12 23.8 25.0 23.5 40 26.8 24.1 25.7 13 13.6 13.3 14.9 41 0 0.3 0.6 14 5.5 5.9 4.3 42 9.3 10.7 8.9 15 9.8 9.4 9.4 43 0 I.I 0.3 16 7.3 4.6 5.3 44 33.8 36.0 34.2 17 8.5 7.4 6.3 45 0 2.1 2.4 20 20.1 19.0 19.7 46 6.2 6.2 5.8 21 32.7 33.9 34.7 51 15.6 17.1 15.0 22 18.4 17.2 IS.I 52 9.0 8.3 9.3 23 27.7 28.8 28.0 53 3.7 3.0 4.4 24 ll.7 10.9 8.9 54 3.3 4.3 3.6 25 20.3 21.5 20.1 55 0 2.0 3.3 26 5.7 6.7 5.4 60 4.2 5.0 4.2 27 16.6 17.6 15.9 61 17.0 17.5 18.0 31 6.6 6.6 6.5 62 5.0 6.3 5.4 32 13.0 11.3 14.4 63 17.5 17.8 17.6 33 12.7 16.3 14.3 64 5.5 6.7 5.8 34 3.7 6.5 4.9 71 9.3 11.6 9.8 35 8.2 11.0 9.0 72 10.6 13.6 10.2 130. Table 3j2 . Atomic parameters (x), thermal parameters (B), interatomic distances, tetrahedral and octahedral angles with standard deviations, atomic radii, scale factor and discrepancy factor R for ullmannite followed by values of Takeuchi (1957). Numbers in parentheses denote number of angles or distances. Sb represents Sb0 . 8s0 . 2 • Point Atom Position -X xT Ni 4a -0.0183-t0.0010 -0.024 s 4a 0.3838-t0.0015 0.390 Sb 4a 0.6247-t0.0004 0.625 Atom B(i2) Atomic Atomic Radii (X) Radii (X) T Ni 0.9-t0.2 0.8 1.22 1.26 s 1.3-t0.2 0.8 1.14 1.08 Sb 1.2-tO.- l 0.8 1.32 1.32 Interatomic distances (i) Ni-Sb(3) 2.542-t0.006 2.57-t0.03 Ni-S(3) 2.361-t0.009 2.34-t0.03 Sb-S(l) 2.454-t0.005 2.40-t0.03 0 Tetrahedral angles ( ) Ni-S-Sb(3) 101.0-t0.4 Ni-Sb•S(3) 101.1-tO.l Ni-S-Ni(3) 116.4-t0.2 Ni-Sb-Ni(3) 116.4-tO.l 0 Octahedral angles ( ) S-Ni-Sb(3) 86.3-t0.3- Sb-Ni-Sb(3) 90.2-t0.3 S-Ni-Sb(3) 83.5-t0.2- S-Ni-S(3) 99.5-t0.3 Scale factor 0.2387 Discreeancr factor R 0.068 0.115(T) 131. Table 3k. Observed hkl reflection amplitudes (Fo) and calculated structure factors (Fe) of synthetic ullmannite formed at 550°C and 700°C with atomic parameters (x), thermal parameters (B), and their standard deviations. 550°c 100°c hkl Fo Fe Fo Fe 200 83 83 56 71 400 113 117 110 62 62 64 65 410 ,20 1 220 67 68 73 56 620 <20 13 140 <20 3 440 144 165 87 90 260 .c20 13 111 64 58 51 51 211 115 114 99 99 311 152 135 108 108 221 11 12 <20 3 331 38 36 <20 23 222 113 136 100 100 322 <20 6 422 45 45 59 25 622 91 92 332 74 71 81 51 533 89 73 444 114 73 Space Point Discrepancy Temp. Group Atom Position -X B Table 31. Tetrahedral and octahedral angles for pyrite, cattierite and vaesite derived from data of Elliott (1960). Numbers in parentheses denote numbers of angles. Pyrite . 0 Tetrahedral angles ( ) Fe•S-Fe(6) 115.7 S-S-Fe(6) 102.1 Octahedral angles (0 ) S-Fe•S(6) 94.9 S-Fe-S(6) 85.6 Cattierite Tetrahedral angles (0 ) Co-S-Co(6) 114.9 S•S•Co(6) 103.3 Octahedral angles (0 ) s-co-S(6) 94.0 s-co-S(6) 86.0 Vaesite 0 Tetrahedral angles ( ) Ni-S-Ni(6) 113.8 S-S-Ni(6) 104. 7 0 Octahedral angles ( ) S-Ni-S(6) 93.5 S-Ni-S(6) 86.5 133. Table 4a. Impurities (ppm) in specpure materials. Specpure Materials As ~8 Fe2_Q3 NiO s Sb Imeurities Ag C 1 < 1 Al 1 0.1 B 0.3 Ba 0.01 Ca • 1 0.1 cl Cd C 1 Cu cl 2 ' 1 C 1 0.03 ◄ l Fe 2 2 0.03 Mg < 1 1 < 1 1 0.03 C 1 Mn 2 Na 1 0.1 Ni 2 Si 1 5 3 5 3 C 1 s C 5 Ti 0.1 134. Table 4b. Spectrographic analyses of natural material, a partial analysis from British Museum and end member compositions. Ni Co Fe Sb As s Si NiSbS 28 57 15 BM69114 25-30 0.5 1 30 25-30 0.3 017751 4 10 0.05 50 1 NiAsS 36 45 19 BM1434 nil 34.39 18.23 HMM Wolfsberg 25-30 2 1 1 40 2 USNM R862 25-30 2 4 1 40 0.3 UNSW 25-30 2 6 1 40 0.2 USNM R830 25 10 10 nil 40 1 BM1922, 145 20 15 1 nil 50 0.3 CoAsS 36 45 19 UNSW 362 2 30 4 nil 40 0.4 UNSW 61 1 30 4 nil 40 2 Glaucodot 1 15 20 nil 40 20 FeAsS 34 46 20 135. Table Sa. Unit cell data and X-ray diffraction powder intensities for gers- dorffite natural samples heated to various temperatures. Sam,ele No. Un\5 Cell 001 011 Temp. Time A co weeks BM1959,462 5.696 7 6 550 1 0 600 1 BM1434 5.694 0 8 8 550 1 0 600 1 BM57562 5.694 2 4 0 3 550 1 0 0 600 1 BM1929, 12 5.687 1 9 0 4 550 1 0 0 600 1 USNM R862 5.865 8 2 3 1 550 1 2 0 600 1 0 0 550 4 BM1922, 145 5.633 6 3 6 3 550 1 2 2 600 1 136. Table Sb. Unit cell data and X-ray diffraction powder intensities for cobaltite natural samples heated to various temperatures. Sampl~ No. Unit Cell 001 011 Temp. Time g co Weeks UNSW 5.581 10 5 2 0 825 1 0 0 850 1 UNSW 233 5.579 50 4 40 4 700 2 20 4 700 4 10 4 700 14 10 4 825 1 4 0 850 1 UNSW 362 5.577 50 5 10 5 825 1 3 0 850 1 UNSW 361 5.577 50 5 25 1 825 1 8 1 850 1 137. PUBLICATION LIST 1. Golding, H. G., and Bayliss, P. , 1960. Dehydration and rehydration of ferromolybdite from Lowther, New South Wales, Amer. Mineral., 45, 1111-1113. 2 . Loughnan, F. C. , and Bayliss, P. , 1961. The mineralogy of the bauxite deposits near Weipa, Queensland, Amer. Mineral., 46, 209-217. 3. Bayliss, P., and Warne, S. St. J., 1962. The effects of the controllable variables on differential thermal analysis, Amer. Mineral., 47, 775-778. 4. Warne, S. St. J., and Bayliss, P., 1962. The differential thermal analysis of cerussite, Amer. Mineral., 47, 1011-1023. 5. Bayliss, P. , and Loughnan, F. C. , 1963. Mineralogical evidence for the penecontemporaneous laterization of the basalts from New England, N.S. W., Amer. Mineral., 48, 410•414. 6. Loughnan, F. C., and Bayliss, P., 1963. A chromium-bearing dyke clay from Cowan, N. S. W., Aust. J. Sci., 26, 185. 7. Bayliss, P., 1964. Effect of particle size on differential thermal analysis, Nature, 201, 1019. 8. Loughnan, F. C., Ko Ko, M., and Bayliss, P., 1964. The red-beds of the Triassic Narrabeen Group, J. Geol. Soc. Aust., 11, 65-78. 9. Bayliss, P., and Loughnan, F. C., 1964. Mineralogical transformations accompanying the chemical weathering of clay-slates from New South Wales, aay Minerals Bull., 5, 353-362. 138. 10. Bayliss, P., 1964. Effect of chemical weathering on parent material, Nature, 203, 327. 11. Bayliss, P., 1964. Some properties of alunogen from New South Wales, Amer. Mineral., 49, 1763•1766. 12. Bayliss, P. , Loughnan, F. C. , and Standard, J. C. , 1965. Dickite in the Hawkesbury Sandstone of the Sydney basin, Australia, Amer. Mineral., 50, 418•426. 13. Bayliss, P., 1965. Differential thermal analysis: effect of particle size, Nature, 207, 284. 14. Bayliss, P., and Males, P.A., 1965. The mineralogical similarity of precious and common opal from Australia, Mineral. Mag., 35, 429-431. 15. Frankel, J. J., and Bayliss, P., 1966. Ferruginized surface deposits from Natal and Zululand, South Africa, J. Sed. Petrology, 36, 193-201. 16. Bayliss, P., Lawrence, L. J., and Watson, D., 1966. Rare copper arsenates from Dome Rock, South Australia, Aust. J. Sci., 29, 145-146. 17. Bayliss, P., and Stephenson, N. C., 1967. The crystal structure of gersdorffite, Mineral. Mag., 36, 38-42. 18. Bayliss, P. , and Standard, J. C. , 196 7. Further mineralogical data on native lead balls, Mineral. Mag., 36, 141-143. 19. Golding, H. G., and Bayliss, P., 1967. Altered chrome ores from the Coolac Serpentine Belt, N .S. W., Australia, Amer. Mineral., in press. 20. Bayliss, P., 1967. The crystal structure of disordered gersdorffite, Amer. Mineral. , in press.