Am*ican Mineralogist, Volme 70,pages1036-1043,I9E5
The crystal chemistryof iron-nickel thiospinels
Dlvtp
J. VlucnnN
Department of Geological Sciences
Uniuersity of Aston in Birmingham
Birmingham, 84 7ET, England
.lxn Jeuns R. Cn^e,tc
Department of Geological Sciences
Virginia Polytechnic Institute and State Uniuersity
Blacksburg, V irginia 24061
Abstract
The crystal chemistryand mineralogicalpropertiesof the thiospinelsof the seriespolydymite (Ni3s4Fviolarite (FeNirsnlgreigite (Fe.So)have beenstudiedusing syntheticand natural samplesand emphasizingthe intermediatecompositions("violarites,"(Fe, Ni).Sn). In the
seriesNi.Sn-FeNirSo, increasingiron correlateswith increasingthermal stability, a nearly
linear decreasein a and increasein reflectance(R% at 589 nm), although spectralreflectance
curvesshow more complex behavior.Miissbauerspectraof 57Fefor a seriesof compositions
betweenFe625Ni2.rrSoand FeNi2Snconsistessentiallyof a doublet with a small isomer shift
and quadrupolesplitting the former increasingwith increasingiron (0.23-0.53mm/sec).This
is interpreted as low-spin Fe2* in the octahedralsites,although further examination of the
spectrasuggestsa possiblecontribution from iron (-18%) in tetrahedralcoordination' These
data are in agreementwith bonding modelspreviouslyproposedfor the thiospinels.
Although electron microprobe analysesof natural samplesindicate that solid solution
extendscompletelyfrom NirSo to Fe.Sn, bonding models suggestthat compositionsmore
iron-rich than FeNirSn may be metastable.The formation of such possiblymetastablecompositions,the observationin this work of violariteswhich are l-2 wt.7osulfur-poorrelativeto
stoichiometricM.So, and the confusion over the low-temperaurephaserelations involving
violarite are all attributed to its formation by alteration of pre-existingminerals (notably
pentlandite).Reactionmechanismand kinetics are thereforecrucial in the formation of this
mineral in nature.
Introduction
The thiospinel (or sulfospinel) minerals are a group
which exemplifya number of problems of generalinterest
in sulfide mineralogy.This contribution is part of a program of study of the thiospinelsby the authors and aimed
at a fuller understandingof their crystal chemistry,thermochemistry, phase relations and occurrence.Although the
mineral thiospinels,M.Sn sulfideswhere M can be Fe, Co,
Ni, Cu, Cr, In and possibly Pt, are named according to
ideal end-membercompositions,extensivesolid solutions
are common involving two major and one (or sometimes
two) minor cations.Previous work (Vaughanet al., l97l;
Vaughan and Craig, 1978; Craig et al., 1979a, 1979b)
has shown that the thiospinel series carrollite
(CuCorSnlinnaeite (Co.Solpolydymite (Ni.Snfviolarite
(FeNirSo)is characterizedby extensivesolid solution. The
present work is concernedwith the polydymite-violarite
seriesalthough natural compositionsof this seriesalso frequently contain some of the Co3Sa flinnaeite) molecule.
For simplicity in this paper, all iron-nickel thiospinels
10-1036$02.00
0003-o04x/85/09
(Ni,Fe).Sn with or without other minor cations) will be
termed "violarites,"exceptfor the endmembers,polydymite
(Ni3S4)and greigite(Fe.Sa).Theseare all metallic minerals,
the properties of which have been describedin terms of
bonding modelsby Yaughanet al. (1971)and Vaughanand
Tossell(1981).Violaritesare quite common secondaryminerals in pentlandite-bearingores (Misra and Fler.:t,1974)
and constitute at least locally important sourcesof nickel.
Greigite occurscommonly only in recentsediments.
Synthetic violarites: structure' composition and
ProPerties
The violarites have the familiar spinel crystal structure
which is based on a cubic close-packedsulfur sublattice
with half the octahedralsites and one-eighthof the tetrahedral sites occupied by cations (Lundqvist, 1947).The
cubic unit cell contains eight ABrSn formula units and
there are, therefore,one tetrahedral(A) cation site and two
octahedral(B) cation sitesper formula unit which ar€ occupied. The tetrahedral A sites are regular but the octa-
r036
VAUGHAN AND CRAIG: IRON-NICKEL THIOSPINELS
hedral B sites show trigonal distortion along different
[1ll] directionsof the cubic unit cell. Each sulfur atom is
four-coordinatedto three cationsin octahedralB sitesand
one cation in a tetrahedralA site as shown in Figure 1.
The thiospinels,like the oxide spinels,have normally
been referredto the cubic spacegroup Fd3m,but Grimes
(1972)has suggestedthat the spacegroup F43m may be
more appropriate to many of these compounds.This is
consistentwith evidencethat the octahedrallycoordinated
metal ions may be "off-cenrer" and displacedalong [111]
directions.Higgins et al. (1975)have found evidencefrom
X-ray'difraction that the thiospinelscarrollite,linnaeite
and siegenitepossessa symmetry lower than Fd3m bfi
consistentwith severalother cubic spacegroups,including
F43m. Grimes(197a),on the other trind, &amined powder
diffraction data and found only a slight improvement in
fitting of the data when the spacegroup F43m wasapplied
to CuCorSo,whereasfor many oxide spinelsthe improvement was appreciable.Other studieshave also found that
cation ordering on octahedral sites in Feo.r.ybr.r.Sn
(Tomas and Guittard, l9j7l and on tetrahedral sites in
FeInrSn (Hill et al., 1978) reduces those thiospinels to
F43rr symmetry.Uncertainty remains,therefore,regarding
the precisestructure of the thiospinelsbut there seemsto
be at least someevidenceto support a structurewith lower
symmetrythan Fd3m.
Although in oxide spinels,divalent and trivalent cations
occur in tetrahedraland octahedralsites gSingnormal and
intsersespinel types, the assignmentof formal oxidation
statesto the cations in thiospinelsis not straightforward.
This is further discussedfor the violarites in relation to
Miissbauerdata presentedin the following section.Structural variations as a function of composition in the series
Ni.So-FeNirSn (-FerSo) will also be discussedmore fully
in a later part of this paper but the most readily monitored
changesare thosewhich occur in the unit-cell parameter.
Precisemeasurements
of the unit-cell dimensionof compositionsbetweenNirSo and FeNirSo were undertakenon
synthetic samplesduring this work. Violarite thiospinels
were syrthesized by conventional silica tube techniques
(Kullerud, 1971; Vaughan and Craig, 1978)and by means
of the multiple reaction techniquespreviously used for
violarite synthesis (Craig, 1971). Reagents used were
99.999+% pure iron, nickel and sulfur as shown by supplier's (esnnco) analyses.Iron was reducedin a stream of
hydrogen at 700'C before being used in the experiments
which were performed in nichrome wound resistancefurnacescontrolled to within + 3"C. Following initial reaction
and homogenizationas (Fe, Ni)r_,S at 7OO"C,sulfur was
added and samples were annealed at 300'C. Quenched
chargeswere examinedby reflectedlight microscopy and
electron microprobe analysisas well as X-ray powder diffraction. Electron microprobe data were obtained using
either an ARL-EMxor an ARL-sEMe
instrument at 15 kV
acceleratingvoltageand 0.15pA samplecurrent. Synthetic
NiS and FeS were usedas standards.In the cell dimension
measurements,
National Bureau of Standardssilicon (o :
5.430884)was usedas an internal standard.
1037
Spinel Struclure
Q anion
@ A - s i f eCotion
o 8 - s i l e Cotion
Fig. 1. The spinelcrystalstructureshowingthe relationships
between
cationsin octahedral
andtetrahedral
coordination.
The variation in cell parameterwith compositionin the
seriesNi.Sn-FeNirSo, is shown in Figure 2. The data show
a linear variation throughout most of the seriesalthough
with a slight flattening out at the FeNirSo end. Greigite,
Fe.So,has a much greatercell parameter(a : 9.876A)than
any ofthe valuesshown hereand the effectofnickel substitution in greigiteon cell parametersis not known at present.
The compositionallimits and stabilitiesof syntheticviolarites can be consideredwith referenceto the Fe-Ni-S
ternary systemand the FeNirSn-NirSn pseudobinaryjoin.
The latter showsthat completeNi.So-FeNirSo solid solution existsbelow 356'C; above 356'C Ni.So decomposes
and only successivelymore iron-rich compositions are
stableto a maximum temperatureof 461'C at the FeNirSn
end (Craig, l97L).It is to be emphasizedthat compositions
more iron-rich than FeNirSn havenot beensynthesized.
Phase relations in the Fe-Ni-S system have been established at high temperaturesby the work of Kullerud
(1963),Naldreu et al. (1967)and Misra and Fleer (1973).
Craig (1971)found that the FeNirSo endmemberbecomes
2
s
P
o
Mol% FeNi254
Fig. 2. Variation in unit cell parameter (a) with composition
betweenFeNirSn and Ni.So. The unit cell parameterof FerSo is
9.8764.
1038
VAUGHAN AND CRAIG: IRON-NICKEL THIOSPINELS
stable below 461'C with tie lines to nickeloan pyrite (Fe,
Ni)S2,ferroan vaesite(Ni, Fe)S, and the monosulfidesolid
solution (Fe, Ni)r-,S. At lower temperatures,with the
breakdown of the monosulfidesolid solution (mss)below
300'C, phaserelationsinvolving violaritesare still confused
since conflicting evidenceis available from experimental
In parstudiesand the examinationof natural assemblages.
ticular, confusion exists over whether pyrite-millerite or
(Grapentlandite-violariteconstitutethe stableassemblage
terol and Naldrett, l97l; Craig, 1973; Keele and Nickel,
1974;Hudson and Groves,1974;Misra and Fleet, 1974).
Certain mineral properties of the violarites synthesized
in this work havealso beenstudied.In particular, measurementshave beenmadeof the spectralreflectanceof compositions betweenNirSn and FeNirSo. Measurementswere
made with the ReichertSpectralMicrophotometer using a
WTiC standardapprovedby the Commissionon Ore Microscopy. The data are presentedin Figure 3 and show
that at 589 nm there is a systematicdecreasein reflectance
from FeNirSn to NirSn Fig. 3a). Spectral reflectance
curvesbetween420 and 640 nm wavelength(Fig. 3b) show
appreciabledispersionwith maxima towards the red end of
the spectrum.The correspondence
of the shapesof the dispersion curves with compositional variation accountsfor
the subtle, although readily observable,color variations
observedin violarites.
I
,Il
#'
f
x
j reo.rrttir.rrso
i
rx* l f
x
lx
x{xx
ri'{x
'ao
oo
I
xl
j
xX
f
d
x
xxx
X
o
o
T
I
2%l
II
I
-O.5
0
,,r{t
+0.5
+l.O
lsomerShifi (mm s-l)
Fig. 4. M<issbauerspectra for violarites of varying iron content; the zero of the velocity scaleis the center of gravity of an
iron foil spectrum.
I
a
Mii.ssbauerstudiesof syntheticviolarites
Synthetic violarites of compositions FeNirSn,
Feo.rNir.rSnand Feo.rrNir.rrSohavebeen
Feo.rrNi2.rrSo,
studied using sTFe Miissbauer spectroscopy.Room temperature spectra were obtained using a standard spectrometer with sTCo as the source of 7-rays. The specREFLECTIVITY
trometer was calibrated using iron foil and isomer shift
data quoted are all relative to the center of gravity of the
iron spectrumtaken as zero.
The Miissbauer spectrum of FeNirSo (Fig. ) consists
essentiallyof a quadrupole doublet with a small isomer
shift and splitting (Table 1).This is in agreementwith previous reports of M6ssbauerdata for this thiospinel endmember(Vaughanet al., l97l; Townsendet al., 1977).The
spectrum can be fitted to a simple doublet as shown in
Figure 5. In this and all the other attempts at computer
fitting ofthe spectra,the program usedwas that written by
Stone (see Bancroft et al., 1967)which provides a leastsquaresfit of overlapping Lorentzian peaks. Parameters
Fig. 3. Spectralreflectancedata for violarites: (a) variation of obtained from the fitting proceduresare presentedin Table
reflectanceat 589 nm with composition; (b) spectral reflectance 1. This simple doublet has a small isomer shift (0.28
mm/sec)and quadrupolesplitting (0.54mm/sec)which has
curvesfor differentcompositions.
VAUGHAN
AND CRAIG:
IRON-NICKEL
Table l. s?Fe Mtissbauer data for synthetic violarites and relate.d
sulfides (all data with the absorber at room temperature unless
otherwise stated).
(mm sec -rF
Splittint
(mm sec-r)
(a) Synthetic violarirertwo
peak tirs (X2 valres -3J4)
FeNi'S,.
0.29 (10.01)
0.54 (:0.02)
F.o.ziNi2.2.:5+
0.2t (:0.01)
0.59 (!0.02)
FeO.jONi2.50Sg
0.26 (:0.01)
0.52 (:0.0I)
Fe9.25Ni2.7554
0.23 (r0.01)
0.6r (!0.02)
(b) Svnthetic violarite-four Fak fits O(2 vatue 268)
FeNi2s4
[tzro area] (a)
0.25 (:0.01J)
0.J6 (:0.0t)
[re* area] 6)
0.38 (r0.015)
0.48 (!0.02)
0.31 (!0.002)
0.61 (!0.06)
0,36 (t0.002)
0,37 (!O.O02)
0.70
0.10
(d Related sulfides*)
FeS2 (pyrire)
Ge,Ni)95t(penttandite)
(tetrahedral site Fe)
Fe3S4 kreigite)
at 4.2K
(a)
(HIoc: 322 Kga6s){
0.45
(H1*
0.40
= 455 6t.u."1+
0.40
(H1*
0.0
= 436'6*"u,"1*
* all isomer shifts quoted relative to rhe centre oI
travity of an iron toil sp*trum
as zero
** se vaughan and Crait, 1978, tor data surces
+ internal
matnetic field at the iron nucleus
beeninterpretedas due to low-spin Fe2* in the octahedral
sites in the spinel structure (Vaughan et al., 1971).Townsend et al. (1977)reported thermoelectricpower and magnetic susceptibility results for violarite in addition to
Mdssbauerspectraat temperaturesdown to 5 K. All the
resultsunambiguouslyshow that violarite exhibitsmetallic,
Pauli paramagneticbehavior.The assumptionthat the iron
in FeNirSa occurs as low-spin Fe2n in octahedralsites is
supported by the correlation in isomer shift and quadrupole splitting values with low-spin Fe2+ in pyrite (see
Table l).
If this simple interpretation of the Mdssbauerspectrum
of FeNirSn is accepted,two-peak fits can be applied to the
spectrafrom the compositionswhich contain successively
smalleramounts of iron (Fig. 4). The results(Table l) show
that with successivedecreasein iron content there is a
systematicdecreasein isomer shift and increasein quadrupole splitting (0.28mm/sec-O.23
mm/secand 0.53mm/sec0.61mm/secrespectively).This could be interpretedas the
result of changesin the occupancyof next-nearest-neighbor
(cation)sitesproducingslightly more asymmetricelectronic
environmentsaround the nucleus,and slight changesin the
s electrondensityat the nucleus.
Closer examination of the spectra of the violarites shown
in Figure 4 shows a distinct, though small, difference in
1039
THIOSPINELS
intensity of the two major peaks.(The low velocity peak
has - 52 to - 54o/oof the total intensity).Although there
are a numhr of possibleexplanationsfor this asymmetry,
the most obvious would be that a second quadrupole
doublet is present and the experimentally observed enveloperesultsfrom a superimpositionof the two doublets.
Attempts at computer fitting were therefore undertaken
employingtwo quadrupoledoublets.This resultsin significantly better fits (e.g.,X2 for FeNirSn: 2 peak fit:354; 4
peak fit:268) and the four peak fit for FeNirSois illustrated in Figure 6 with data presentedin Table 1. The
seconddoublet comprisesl8o/' (+4yo\ of the overall intensity. It is possiblethat this doublet representsiron in tetrahedral sites in violarite. Since the material is a metallic,
Pauli paramagneticphase,suchiron should be comparable
to that in the tetrahedralsitesin pentlandite(seeTable l)
and the valuesdo show reasonableagreementwith those
reportedfor pentlandite.
Additional doublets may also arise through different
next-nearest-neighbor
coordination, as observedby Reidel
and Karl (1981)for tetrahedralsite Fe2+ ions in the system
FeCrrSn-Fe.So.However, although such differencesproduce large changesin quadrupole splitting, the changes
producedin isomershifts are very small comparedto those
observedhere.Also, attempts to obtain a consistenttrend
in parametersand in the intensity of the seconddoublet
with varying Fe :Ni ratio proved unsuccessful.Another
possibility is that some octahedralsite iron has been oxidized to Fe3+.This also seemsunlikely as everyprecaution
was taken in sample preparation to prevent oxidation.
However the parameterof the seconddoublet (b) in Table
1) in the FeNirSo spectrum are very similar to those as-
l--:
--J,
V
I\
.t
,l
lr
ll
I
l*l
I
Ir
bomrSlih(rmrr)
-"_Y
_l_:_ _v_:__'__,_
.rh r
^a-.
lq
A.
.l
^A
. ^,r
Fig. 5. Computer fit of a simple doublet to the Miissbauer
,p."t.u- of FeNirsn.
VAUGHAN AND CRAIG: IRON.NICKEL THIOSPINELS
10,10
.E
a
.t
E
o
Fig. 6. Computerfit of two doubletsto the Miissbauerspectrumof FeNirSn.
signedby Reideland Karl (1981)to high-spinFe3* ions in
octahedralsitesin Fer**Crr-,Sn.
The Miissbauer spectmm of greigite (Fe3Sn),the pure
iron analogueof violarite, contrastsstrongly with the violarites (Vaughan and Ridout, 1971).At 4.2 K, it has been
resolvedinto three setsof six-line magnetichyperfinespectra attributed to high-spin Fe3* ions in tetrahedralA sites
and octahedral B sites. Greigite is, therefore,an inverse
spinel and the sulfur analogueof magnetite as also confirmed by the work of Coey et al. (1970) although they
observedonly a singlespecieson the octahedralsitesas a
result of electronhopping betweenaFez+' and "Fe3+".
Fe.Sn. Misra and Fleet (1974)have alreadyreported natural "violarite" compositionsindicating a solid solution extending to nickel contentsas low as 17 atomic percent;the
data in Figure 7 show essentially complete violaritegreigite solid solution. For Figure 7 perfect M3Sn stoichiometry is assumedbut the nature and extent of any variation in M:S ratio in this seriescan also be considered
using a triangular Fe-Ni-S plot as shown in Figure 8- It is
remarkablethat of the 74 analysesshown here, only two
plot in the field ofexcesssulfur.A few ofthe analyseslie on
the line joining stoichiometricFeNirSn and NirSn but most
are sulfur-deficientby l-2 wt.oh sulfur. This strongly suggests a real trend to sulfur-poor compositionsin natural
violarites, an observationfirst noted by Desboroughand
Czamanske(1973)and later supportedby Misra and Fleet
(1974\ and by Craig and Vaughan (1976).Like Misra and
Fleet(1974)we find no clear evidenceto support the alternative stoichiometry of (Fe, Ni)"St, suggestedby Desborough and Czamanske(1973).A note of caution should be
added, however, becausemost natural violarites are extremely porous and many of the microprobe analysesreported have totals less than 98%. Any preferentialloss of
SK, X-rays as a result of the porosity would bias the results shown in Figure 8. Nevertheless,the consistencyof
low sulfur contentsobservedby so many workers suggests
that it is a real phenomenon.
Polydymite is an uncommon mineral known primarily
from scatteredoccurrencesin hydrothermal vein deposits.
Violarite, on the other hand, is a widespreadand wellknown secondary mineral formed in the weathering of
nickeliferouspyrrhotite oresassociatedwith mafic rocks.In
such occurrences,it is generallypresentas rims on hypogene pentlandite where it apparently forms through the
selectiveremoval ofcations from the pre-existingstructure.
Natural violarites: compositions, occurrence and
associations
During this work the compositionsof numerousnatural
violarites(as well as certain of the coexistingsulfides)have
beendeterminedby electronmicroprobe analysisusing the
instrumentsand standardsalready described.In addition,
numerousanalyseshave beenassembledfrom data already
in the literature. These data are presentedin Table 2.1
Compositions of natural iron-nickel thiospinels are also
plotted on Figure 7 in terms of relative amounts of
Ni.So-Fe.So-Co.Snendmemberssinceminor cobalt is frequently presentin violarites.A very important observation
immediatelyseenfrom this diagram is that, unlike the synthetic iron-nickel thiospinels which do not form compositions more iron-rich than FeNirSn, the natural samples
show a complete compositional range from Ni.Sn to
1 To obtain a copy of Table 2, order Document AM-85-275
from the Mineralogical Society of America, Business Offrce, 2000
Florida Ave., N.W., Washington, D.C. 20009. Please remit $5.00 in
advance for the microfiche.
Fig. 7. Compositions of natural iron-nickel thiospinelsshown
in terms of FerSo-NirS*{o.So endmembersto illustrate the variation in cation contents(for sourcesof data seeTable 2).
VAUGHAN AND CRAIG: IRON-NICKEL THIOSPINELS
Wf A Nickel
Fig. 8. Compositions
of naturaliron-nickelthiospinels
shown
in termsof Fe,Ni andS content(forsources
of data.seeTable2).
Nickel (1973)o
Misra and Fleet(1974),and,Imaiet al. (1978)
have noted that the iron/nickel ratios (and even the cobalt
contents)of thesc secondaryviolarites are largely a function of the initial pentlandite composition. Consequently,
many violarites exhibit iron/nickel ratios far in excessof
the ll2 stoichiometry seenin synthetic violarites and expectedfrom crystal chemicalconsiderations.Nickel (1973)
has further suggestedthat very iron-rich compositions
result from replacementof pyrrhotite (possibly nickeloan
pyrrhotite). The thermodynamicstability of theseiron-rich
compositionsis neither known nor probably consequential
in terms of their formation and persistence.Misra and
Fleet (1974)have noted that the similarities of the crystal
structuresreadily permit conversionof pentlanditeto violarite regardlessof stabilities.This sameeaseof conversion
may also account for the problem of what constitutes
stablelow temperatureassemblages.
The trends of tie lines
in higher temperatureexperimentsand the associationsobserved in what are interpreted as hypogenemillerite occurrencessuggest that pyrite and millerite may coexist
stably at low temperature.On the other hand, the common
occurrenceof secondaryviolarite forming on pentlanditein
weatheredores has been interpretedas indicating a stable
violarite-pentlanditetie line at low temperature.It appears
likely that pyrite + millerite representsthermodynamicstability but that pendlandite+ violarite represents a
common situation in nature becauseof the readv conversion ofpentlanditeto violarite.
Discussion
The crystal chemistry of thiospinel minerals has been
discussedin terms of qualitative molecular orbital and
band theory bonding modelsby Vaughanet al. (1971)and
l04t
these models have been supported by molecular orbital
calculationsin the casesof Co.Sn, CuCorSnand Fe.Sn
(Vaughanand Tossell,1981).Sincethe highestenergylevels
containingelectrons(and the lowest empty levels)are those
associatedwith the transition metal 3d orbitals, only those
need be consideredin detail. Schematicenergy level diagrams for the 3d orbitals in Ni.Sn, FeNirS* and FerSnare
shownin Figure9. In Ni.So and FeNirSo,the lower energy
3d levels (tzr, e) are consideredas filled with electrons,
whereas the higher energy levels (comprised of e, and
tz-type orbitals) which are antibonding in characteroverlap to form delocalizedbands of collective electrons(of,
and ofi from cations on A and B sites).In Ni.So, the o*
band contains 6 electronsper formula unit, whereaswith
substitution of iron in the series NirSn-FeNirSn, the
number drops to 4 at FeNirSn. No fundamental change
occurs in the electronic structures of thiospinels in this
polydymite-violarite series. Furthermore, the systematic
changesin propertiesin the seriescan be directly linked to
the electron occupancyof this o* band (Vaughan et al.,
l97lF--correlations further substantiatedby the more detailed data on propertiespresentedin this paper.Thus, the
systematicincreasein cell parameterfrom FeNirSo-Ni.Sn
(Fig. 2) correlateswith the increasingelectronoccupancyof
the antibonding orbitals of the o* band which are more
proximal to the sulfur ligands tending to push theseaway.
The systematicdecreasein reflectanceat 589 nm along the
sameseries(Fig. 3a)can also be relatedto a decreasein the
available empty levels in the o* band; levels into which
electronsmay be excitedresultingin high absorption coefficients and high reflectances(Burns and Vaughan, 1920;
Vaughan et al., l97l). The stabilities of these thiospinels
have also been discussedin terms of the bonding models
and the lower thermal stability of polydymite (353+ 3.C)
compared to violarite (461+3'C) related to the greater
electronoccupancyof the o* band in polydymite(Vaughan
et al., l97l).
The Mcissbauerdata presentedfor the violarites in this
paper have beenusedto suggestthat someiron may occur
in tetrahedral sites in these minerals. Such substitution
would not affect the electronoccupancyof the o* band so
that the correlationsnoted abovewould remain valid. Further detailed spectroscopicstudiesare required to confirm
the presenceof iron in tetrahedral coordination in violarites and determineif its distribution is dependenton the
conditions(annealingtemperature,time) of formation.
The electronicstructureofgreigite illustrated in Figure 9
contrastsstrongly with the polydymite-violariteseries.The
model shown hereincorporatesthe resultsof the molecular
orbital calculationsof Vaughanand Tossell(1981).Greigite
contains localized 3d electronswith unpaired spins which
are coupled antiferromagneticallyat lower temperatures.
This essentiallymore "ionic' substancehas a much larger
cell parameterthan the violarites and the large number of
electronsin antibonding orbitals contribute to its instability. Vaughan et al. (1971)suggestedthat the lack of experimental solid solution betweenFeNirSn and FerSn is
due to this fundamentaldifferencein electronicstructure.
VAUGHAN AND CRAIG: IRON-NICKEL THIOSPINELS
lu2
Ni3s4
FeNi2S4
,,:,r+htsd!- *er$H*sA
tzs"rr".r..
,+"_,.1,.1,*#r
o-.js-+to
e++F ,, ,,
Ni3*in B-.ir.
Ni3tin B site
F"2*inB-rir"
,,
lt
-H4
-eg
_ttzs
-eg
11er-t29
Ni2*in B site
-t2
-e
#tz
+e
dP
ff-eg
ftrzg
aB
frlt2o
Fe3'i. B-iite
F.3*ina-.it"
Fe2*inB-rite
Fig. 9. Schematic energy level diagrams for the 3d orbitals in
Ni.So, FeNirSo and FerSo. a and fl refer to spin directions of the
electrons; E. is the Fermi level.
The new data show that although FeNirSn-FerSo solid
solution is not observedin synthetic (high temperature)
experiments,it is observedin natural phases(Fig. 7).This is
not necessarilyin conflict with the proposed electronic
structure theory for thiospinels,sincethe compositionsbetween FeNirSo and FerSnare very likely to be metastable.
They occur in small quantities and frequently result from
the alteration of pre-existingpentlandite-the "predominant mode of formation of violarite" (Misra and Fleet,
1974).There is increasingevidenceto show the importance
of reaction mechanismsand kinetic factors in the formation of sulfides at low temperatures(e.g. Putnis, 1976;
Putnis and McConnell, 1980).The metastabletransformation of pentlandite to violarite would readily explain the
occurrenceof iron-enriched violarites and also the frequently observedsulfur-poor compositions(Fig. 8) which
lie on the pentlandite site of the stoichiometric violarite
composition. It is not surprising, therefore,that the low
temperaturephaserelations involving violarite are controthat the
versial.Furthermore,Nickel (1973)has suggested
compositions of secondary violarites are not dependent
upon stability but rather upon the initial phasereplacedor
altered.
The properties of compositions between FeNirSo and
FerSn are particularly interesting and attempts to synthesizethesein aqueousmedia are underway.Such materials must exhibit a transition from delocalizedto localized
d-electron behavior at some point in the series.It is interestingto note that such a transition has been observed
in CuCrr-*V,Sn thiospinels(Robbinset al., 1970).
Acknowledgments
Financial support from NATO grant No. 966 and NSF grants
DMR75-038?9 and DMR78-09202is gratefully acknowledgedas
are funds from the University of Aston for purchase of a
Mcissbauerspectrometer.We thank Dr. D. O'Connor (Birmingham University) for providing some of the Mcissbauerfacilities
and Dr. A. D. Law and Mr. M. Linskill for the help with computer fitting of M<issbauerspectra.Natural violarite sampleswere
kindly provided by the U.S. National Museum, and the British
Museum(Natural History).
References
Bancroft, G. M., Ong, W. K., Maddock, A. G. and Prince, R. H.
Journal of
(1967)Mcissbauerspectraof Fe(III) ac€tylac€tonates'
ChemicalSocietyA, 1969.
Burns, R. G. and Vaughan,D. J- (1970)Interpretation of the reflectivity behavior of ore minerals.American Mineralogist, 55,
1576_1586.
Coey, J. M. D., Spender,M. R. and Morrish, A. H. (1970)The
magneticstructure of the spinel, Fe.Sn. Solid State Communications,S, 1605-1608.
Craig, J. R. (1971)Violarite stability relations.American Mineralogist,56,1303-1311.
Craig, J. R. (1973)Pyrite-pentlanditeand other low temperatule
relations in the Fe-Ni-S system.American Journal of Science'
2734,,496-510.
Craig, J. R. and Vaughan,D. J. (1976)Thiospinels:crystal chemistry of the polydymite-violariteseries.(abstr.)GeologicalSociety
of AmericaAnnual Meeting,Denver,824.
Craig J. R., Vaughan, D. J. and Higgins, J. B. (1979a)Phase
relations in the Cu-Co-S systemand mineral associationsof
the carrollite (CuCorS*)-linnaeite(CorSo)series.EconomicGeology,74,657-6'll.
Craig J. R., Vaughan, D. J. and Higgins, J. B. (1979b) Phase
equilibria in the copper-cobalt-sulfursystem. Materials ResearchBulletin. 14,149-154.
Desborough,G. A. and Czamanske,G. K. (1973)Sulfidesin eclogite nodules from a kimberlite pipe, South Africa, with comments on violarite stoichiometry. American Mineralogist, 58'
t95-202.
Graterol, M. and Naldrett, A. J. (1971)Mineralogy of the Marbridge No. 3 and No. 4 nickel-iron sulfide deposits.Economic
Geology,66, 886-900.
'Off centre' ions in compoundswith spinel
Grimes, N. W. (1972)
structure.PhilosophicalMagazine,26, l2l7 -1226.
Grimes, N. W. (1974)An X-ray diffraction investigationof sulfide
spinels.Journal of PhysicsD: Applied Physics,7, 1-6.
Higgins, J. B., Speer,J. A. and Craig J. R. (1975) A note on
thiospinel spacegroup assignment.PhilosophicalMagazite, 32,
685.
Hill, R. J., Craig, J. R. and Gibbs, G. V. (1979)Systematicsof the
spinel structure type. Physics and Chemistry of Minerals, 4,
317-339.
Hudson, D. R. and Groves,D. l. (1974)The composition of viol'
arite coexistingwith vaesite,pyrite and millerite. EconomicGeology,69,1335-1340.
Imai, N., Mariko, T. and Shiga, Y. (1973) Siegenitefrom the
Nippo ore deposit of the Kamaiski Mine, Iwate Prefecture,
Japan.Mining Geology(Japan),24,357-364.
Keele, R. A. and Nickel, E. H. (1974\The geology of a primary
and supergenealteration at
millerite-bearingsulfideassernblage
the Otter Shoot, Kambalda, WesternAustralia. Economic Geology,69,ll02-lll7.
VAUGHAN AND CRAIG: IRON-NICKEL THIOSPINEL9
Kullerud, G. (1963)The Fe-Ni-S system.Camegie Institute of
WashingtonYearBook,67,175-189.
Kullerud, G. (1971) Experimental techniquesin dry sulfide research. In G. C. Ulmer, Ed., ResearchTechniquesfor High
Pressureand High Temperature,p. 288-315. Springer-Verlag,
Berlin.
Lundqvist, D. (1947)X-ray studieson the ternary systemFe-Ni-S.
Arkiv Kemi MineralogiGeologi,24a,No.22.
Misra, K. C. and Fleet,M. E. (1973)The chemicalcompositionsof
synthetic and natural pentlandite assemblages.
Economic Ge_
ology,68,518-539.
Misra, K. C. and Fleer, M. E. (1974)Chemical composition and
stability of violarite.EconomicGeology,69,391-40?..
Naldrett, A. J., Craig, J, R. and Kullerud, G. (19671The central
portion of the Fe-Ni-S systemand its bearing 04 pendlandite
exsolution in iron-nickel sulfide ores. Economic Geology, 62,
826-847.
Nickel, E. H. (1973) Violarite-a key mineral in the supergene
alteration of nickel sulphideores.In Australian I.M.M. Conference,May 1973,p. 111-116,AustralianInstituteof Mining and
Metallurgy, Parkville,Australia.
Putnis, Andrew (1976) Alternative transformation behaviour in
sulfides: direct observation by transmission electron microscopy.Science,193,4174lB.
Putnis, Andrew and Mc{onnell, J. D. C. (1980)principlesof Mineral Behaviour.Blackwells,Oxford.
Reidel, E. and Karl, R. (1981)Mcissbauerstudiesof thiospinels.
IV. The systemFeCrrSn-Fe.So.Journal of Solid State Chemistry,38, 48-54.
Robbins, M., Menth, A., Miksovsky, M. A. and Sherwood,R. C.
(1970)Magnetic and crystallographicevidencefor localization
to delocalizationof V d-electronlevelsin CuCrr_*V,Sn spinels.
Journal of Physicsand Chemistryof Solids,31,423430.
Tomas, A. and Guittard, M. (1977\Une structure deriveede celle
du spinelle: la structure de Feo.ruYb,,uSo.Materials Research
Bulfetin. 12. 1043-10/;6.
Townsend,M. G., Gosselin,J. R., Horwood, J. L., Tremblay,R. J.
and Ripley, L. c. (1977) Violarite, a metallic narural spinel.
PhysicaStatusSolidi !\ Q,K25-K29.
Vaughan,D. J. and Craig, J. R. (1978)Mineral Chemistryof Metal
Sulfides.CambridgeUniversity Press,Cambridge.
Vaughan, D. J and Ridout, M. S. (1971) Mcissbauerstudies of
somesulfideminerals.Journal of Inorganic and Nuclear Chemistry,33,741-747.
Vaughan, D. J. and Tossell,J. A. (1981)Electronic structure of
thiospinel minerals; results from MO calculations.American
Mineralogist,66, 1250-1253.
Vaughan,D. J., Burns, R. G. and Burns, V. M. (1971)Geochemistry and bonding of thiospinel minerals.Geochimicaet CosmochimicaActa.35.361381.
Manuscript receiued,September 14, l9B4;
acceptedfor publication, April 25, 1985.