Content uploaded by Sven C. Vogel
Author content
All content in this area was uploaded by Sven C. Vogel on Apr 27, 2015
Content may be subject to copyright.
ORIGINAL PAPER
Thermal expansion and decomposition of jarosite:
a high-temperature neutron diffraction study
Hongwu Xu ÆYusheng Zhao ÆSven C. Vogel Æ
Donald D. Hickmott ÆLuke L. Daemen Æ
Monika A. Hartl
Received: 22 December 2008 / Accepted: 5 May 2009 / Published online: 24 May 2009
ÓSpringer-Verlag 2009
Abstract The structure of deuterated jarosite,
KFe
3
(SO
4
)
2
(OD)
6
, was investigated using time-of-flight
neutron diffraction up to its dehydroxylation temperature.
Rietveld analysis reveals that with increasing temperature,
its cdimension expands at a rate *10 times greater than
that for a. This anisotropy of thermal expansion is due to
rapid increase in the thickness of the (001) sheet of
[Fe(O,OH)
6
] octahedra and [SO
4
] tetrahedra with increas-
ing temperature. Fitting of the measured cell volumes yields
a coefficient of thermal expansion, a=a
0
?a
1
T, where
a
0
=1.01 910
-4
K
-1
and a
1
=-1.15 910
-7
K
-2
.On
heating, the hydrogen bonds, O1D–O3, through which the
(001) octahedral–tetrahedral sheets are held together,
become weakened, as reflected by an increase in the DO1
distance and a concomitant decrease in the O3–D distance
with increasing temperature. On further heating to 575 K,
jarosite starts to decompose into nanocrystalline yavapaiite
and hematite (as well as water vapor), a direct result of the
breaking of the hydrogen bonds that hold the jarosite
structure together.
Keywords Jarosite Neutron diffraction
Thermal expansion Decomposition Hydrogen bonds
Crystal chemistry
Introduction
Jarosite, KFe
3
(SO
4
)
2
(OH)
6
, and the related sulfates that
comprise the so-called ‘‘alunite supergroup’’ (Jambor
1999), commonly occur in acid drainage environments, as
the weathering products of sulfide ore deposits. They are
found in clays as nodules and in acid soils, where previ-
ously existing pyrite was oxidized into jarosite. They can
also precipitate from aqueous sulfate due to oxidation of
H
2
S in epithermal environments and hot springs associated
with volcanic activity (Papike et al. 2006). In 2004, jarosite
was detected by the Mars Exploration Rover (MER)
Mo
¨ssbauer spectrometer (Klingelho
¨fer et al. 2004), and it
has been interpreted as strong evidence for the occurrence
of large amounts of water (and possibly life) in the history
of Mars. A recent study using laser desorption Fourier
transform mass spectrometry revealed the presence of
organic matters (such as glycine) in several jarosite sam-
ples (Kotler et al. 2008), lending some support to the
hypothesis that life existed on Mars.
In addition to its geological importance, jarosite is of
considerable interest for its industrial applications (Dutrizac
and Jambor 2000). Specifically, in the zinc industry, Zn is
usually extracted from Zn-sulfides (such as sphalerite) by
the so-called ‘‘roast-leach-electrolysis’’ process. However,
these sulfides commonly contain Fe, typically 5–12 wt%,
which needs to be removed. Precipitation of jarosite com-
pounds has been found to be an effective means for the Fe
removal, as they form quickly and are readily filterable and
washable. This process operates at atmospheric pressure,
rather than requiring an autoclave as for many hydrothermal
processes, and is thus economical. Furthermore, the gen-
erated jarosite (in the form of mud) can be combined with
other industrial wastes such as dump ferrous slag (DFS) and
alkaline Al-surface cleaning waste (ASCW) as well as small
H. Xu (&)D. D. Hickmott
Earth and Environmental Sciences Division,
Los Alamos National Laboratory,
Los Alamos, NM 87545, USA
e-mail: hxu@lanl.gov
Y. Zhao S. C. Vogel L. L. Daemen M. A. Hartl
Los Alamos Neutron Science Center,
Los Alamos National Laboratory,
Los Alamos, NM 87545, USA
123
Phys Chem Minerals (2010) 37:73–82
DOI 10.1007/s00269-009-0311-5
portions of Portland cement or lime to produce materials for
construction applications (such as airfield runways and
levee cores) (Mymrin et al. 2005). In addition, jarosite and
its associated alunite-type phases have been proposed as
potential hosts for the long-term immobilization of radio-
active fission products and toxic heavy metals (Ballhorn
et al. 1989; Kolitsch et al. 1999).
The structure of jarosite consists of [SO
4
] tetrahedra
and distorted [Fe(O,OH)
6
] octahedra with K located in a
12-fold coordinated site (space group R
3m) (Fig. 1a)
(Menchetti and Sabelli 1976; Stoffregen et al. 2000;
Basciano and Peterson 2007). Each [Fe(O,OH)
6
] octahedron
corner-shares four hydroxyl groups with neighboring
[Fe(O,OH)
6
] octahedra and two oxygen atoms from two
[SO
4
] tetrahedra (one above the Fe and one below),
forming (001) sheets of [Fe(O,OH)
6
] and [SO
4
] perpen-
dicular to the caxis. There are two types of crystallo-
graphically distinct [SO
4
] tetrahedra: one [SO
4
] tetrahedron
pointing upward along c(c?), and the other [SO
4
] tetra-
hedron pointing downward (c-), which alternate in a
zigzag fashion along the a-axes within the (001) layer
(Fig. 1b). Each K is coordinated by 6 O atoms from [SO
4
]
tetrahedra and 6 OH groups from [Fe(O,OH)
6
] octahedra.
All 6 O atoms and all 6 OH groups are symmetrically
identical, and thus the K site has a highly symmetrical
coordination with 6 identical K–OH bonds and 6 identical
K–O bonds (Papike et al. 2006).
The unique distribution of [Fe(O,OH)
6
] octahedra
within the (001) layer coupled with the magnetic properties
of Fe
3?
makes jarosite a model compound for studying the
spin frustration in two-dimensional kagome
´lattices (com-
posed of magnetic ions located at corners of triangles that
are linked via corner-sharing) (Wills et al. 2000). Low-
temperature neutron diffraction experiments reveal that
jarosite exhibits long-range magnetic ordering when cooled
below 65 K, as evidenced by the appearance of several
magnetic reflections at hkl/2, l=odd (Inami et al. 2000).
The c-dimension of the magnetic unit cell is twice that of
the conventional unit cell, and the magnetic structure
belongs to the so-called ‘‘q =0, 120°type’’ with triangles
of the spins having only positive chirality (Inami et al.
2000). The magnetic ordering is interpreted as a result of
the coupling between the jarosite (001) layers exhibiting a
net magnetization, which is mainly due to Dzyaloshinsky-
Moriya (DM) anisotropic interactions (Grohol et al. 2003;
Yildirim and Harris 2006).
Fig. 1 a Crystal structure of
jarosite, KFe
3
(SO
4
)
2
(OH)
6
;ba
sheet of [Fe(O,OH)
6
] octahedra
and [SO
4
] tetrahedra projected
along the c-axis; cball-and-
stick representation of the
jarosite structure. Tetrahedra
represent [SO
4
] units, octahedra
represent [Fe(O,OH)
6
] units,
pink balls represent K, green
balls represent Fe, brown balls
represent S, blue balls represent
O(light blue O1 and O2;
dark blue O3), and red balls
represent H. Blue lines in aand
boutline the unit cell, and the
dash line in cmarks the
hydrogen bond between O1 and
H. In aand c, the c-axis of the
jarosite structure is vertical
74 Phys Chem Minerals (2010) 37:73–82
123
Despite the detailed structural studies of jarosite at room
and low temperatures, no information is available about its
high-temperature structural behavior. The recent discovery
of jarosite on Mars has spurred interest in its stability at
various temperatures, pressures, and aqueous conditions
(such as solution pH). A number of thermochemical studies
of jarosite and its analogues have been performed to
determine their decomposition temperatures, enthalpies of
formation, and enthalpies of dehydroxylation (Drouet and
Navrotsky 2003; Drouet et al. 2004; Forray et al. 2005;
Frost et al. 2005; Navrotsky et al. 2005). However, the
changes in the jarosite structure upon heating are still
poorly constrained. In particular, its coefficients and
mechanisms of thermal expansion remain unknown. Since
the high-temperature structural behavior of jarosite is likely
to be related to changes in its hydroxyl behavior and since
neutron scattering is sensitive to the position of hydrogen,
high-temperature neutron diffraction studies are particu-
larly useful to unravel the mechanisms of its thermal
expansion and decomposition.
In this study, we carried out in situ neutron diffraction of
jarosite using a pulsed neutron source at temperatures up to
650 K (the sample started to decompose into yavapaiite
KFe(SO
4
)
2
, hematite Fe
2
O
3
and water vapor D
2
O between
550 and 575 K). To avoid the large incoherent scattering of
neutrons by hydrogen, we synthesized deuterated jarosite,
KFe
3
(SO
4
)
2
(OD)
6
, using hydrothermal methods. Rietveld
analysis of the time-of-flight neutron data allowed deter-
mination of structural parameters as a function of temper-
ature. In particular, the atomic positions and atomic
displacement parameters of jarosite at high temperatures
have been obtained for the first time, and the structural
effects on jarosite thermal expansion and stability are
discussed.
Experimental methods
Sample synthesis
The jarosite sample used in this study was prepared via
hydrothermal methods. First, 8.1 g of Fe(NO
3
)
3
9D
2
O
(20 mmol) (Acros Organics, [99%) and 3.5 g of K
2
SO
4
(Acros Organics, [99%) were dissolved separately in
25 mL D
2
O. Second, the two solutions were mixed and
stirred thoroughly in a 100 mL Teflon cup, which was then
placed in a standard Parr autoclave. Third, the autoclave
was sealed and heated at 433 K for 3 days. After cooling
down to room temperature, the autoclave was opened, and
the contents filtered and washed with cold D
2
O. Lastly, the
resulting solid product was dried in air for one hour, placed
in a vacuum oven at 383 K overnight and then stored in a
desiccator. The product, a brown, well-crystallized powder,
was confirmed to be single-phase jarosite by powder X-ray
diffraction (Rigaku Ultima III, 40 keV, 50 mA, CuKa
radiation). The K, Fe and S contents of the sample were
measured by inductively coupled plasma atomic emission
(ICP-AE) spectroscopy. The determined weight concen-
trations are 7.66% K, 35.0% Fe and 12.6% S. The water
content, 11.7%, was determined by heating weighted
powders of the material to 723 K for *4 h and attributing
the weight loss to D
2
O. These values are very close to the
stoichiometric compositions of 7.71% K, 33.1% Fe, 12.7%
S and 11.9% D
2
O in KFe
3
(SO
4
)
2
(OD)
6
, respectively. Thus
in the following structural analysis, we treated the sample
as having the ideal formula.
Neutron diffraction
Time-of-flight neutron diffraction experiments were per-
formed at the High-Pressure Preferred Orientation (HiPPO)
beamline of the Manuel Lujan, Jr. Neutron Scattering
Center, Los Alamos National Laboratory. Sample powders
were placed in a vanadium can 0.95 cm in diameter, and
the can was mounted in an ILL-type high-temperature
furnace with vanadium heating elements and heatshields
for contamination-free diffraction data collection (Vogel
et al. 2004). Data were collected under vacuum at room
temperature and at temperatures from 350 to 650 K with an
interval of 25 K. For each temperature point, three detector
banks with nominal diffraction angles of 40°,90°and 140°
were simultaneously used. The heating rate was 5 K/min,
and the dwell time at each targeted temperature (including
an equilibration time of 5 min) was *4h.
Structure refinement
The neutron data were analyzed using the Rietveld method
with the General Structure Analysis System (GSAS) pro-
gram of Larson and Von Dreele (2000). The starting
structural parameters for KFe
3
(SO
4
)
2
(OD)
6
at 298 K were
taken from the neutron diffraction study of Menchetti and
Sabelli (1976). We then used the refined structural param-
eters at 298 K as the starting parameters for the next highest
temperature and continued this procedure systematically
with increasing temperature. For the runs at 575 and 600 K,
since a portion of the sample decomposed into yavapaiite,
hematite and water vapor, we included yavapaiite and
hematite as secondary phases in the Rietveld analyses. The
starting structural parameters for yavapaiite and hematite
were taken from the X-ray diffraction studies of Anthony
et al. (1972) and Maslen et al. (1994), respectively. For the
two highest temperature runs (625 and 650 K), only
yavapaiite and hematite were present, and thus jarosite was
excluded from the analyses. For each temperature point,
two datasets from the detectors at 2h=90°and 140°were
Phys Chem Minerals (2010) 37:73–82 75
123
simultaneously analyzed (the 40°dataset was not used
because of its relatively low resolution).
The refinements proceeded as follows: after the scale
factor and four background terms (Shifted Chebyshev
function) for each histogram had converged, lattice
parameters and phase fractions (for the runs at 575, 600,
625 and 650 K) were added and optimized. Fourteen or
eighteen additional background terms were then added for
each histogram, and the peak profiles were fitted to a TOF
profile function (Von Dreele et al. 1982). On convergence
of the preceding parameters, atomic coordinates and iso-
tropic atomic displacement parameters for K, Fe, S, O, and
D were refined, yielding R
wp
values ranging from 1.33 to
1.43%, R
p
from 0.87 to 0.98%, and v
2
from 2.8 to 3.9. The
refined unit-cell parameters, atomic coordinates, atomic
displacement parameters, and selected bond parameters are
listed in Tables 1,2,3, and 4, respectively. A representa-
tive pair of fitted patterns is plotted in Fig. 2.
Results and discussion
Stability of jarosite
Our high-temperature neutron diffraction patterns indicate
that the deuterated jarosite sample was stable up to 550 K.
However, it started to decompose into yavapaiite, hematite
and D
2
O vapor when the temperature reached 575 K:
KFe3SO4
ðÞ
2ODðÞ
6!KFe SO4
ðÞ
2þFe2O3þ3D2O:ð1Þ
As shown in Fig. 3, at 575 K, new diffraction peaks
indicative of yavapaiite and hematite appeared, and the
molar ratio for jarosite:yavapaiite:hematite obtained from
Rietveld analysis was 74.8:12.6:12.6. When the temperature
was increased to 600 K, these new peaks grew, and more
obviously, the original jarosite peaks (such as 003) became
significantly weaker. The refined molar ratio for
jarosite:yavapaiite:hematite at 600 K was 28.6:35.7:35.7.
With increasing temperature to 625 K, jarosite decomposed
completely, as revealed by the disappearance of its
diffraction peaks, and only yavapaiite and hematite were
present. Thus the onset temperature of the jarosite
dehydroxylation (T
d
) lies between 550 and 575 K. This is
generally consistent with previous thermal analyses of
potassium jarosite, which show that the mass loss due to
the dehydroxylation occurs in the temperature range 403–
603 K (Frost et al. 2005). On the other hand, as in other
hydroxyl-containing compounds such as portlandite (Xu
et al. 2007), the measured dehydroxylation temperature
can vary with sample purity, sample crystallinity and
experimental conditions such as heating rate and water
vapor pressure. These factors may account for some of the
discrepancies in the reported T
d
values for jarosite (Frost
et al. 2005; Drouet and Navrotsky 2003).
Table 1 Unit-cell parameters of deuterated jarosite and agreement
indices of the refinements
T(K) a(A
˚)c(A
˚)V(A
˚
3
)R
wp
(%) R
p
(%)
298 7.29013(6) 17.1921(2) 791.28(1) 1.33 0.87
350 7.29109(6) 17.2293(2) 793.20(1) 1.33 0.87
375 7.29275(6) 17.2514(2) 794.58(1) 1.33 0.87
400 7.29422(6) 17.2713(2) 795.82(1) 1.33 0.88
425 7.29541(7) 17.2916(2) 797.02(1) 1.33 0.89
450 7.29603(7) 17.3129(3) 798.13(1) 1.35 0.92
475 7.29644(7) 17.3316(3) 799.08(1) 1.35 0.93
500 7.29702(7) 17.3478(3) 799.96(1) 1.42 0.97
525 7.29759(7) 17.3632(3) 800.79(2) 1.43 0.98
550 7.29760(8) 17.3761(3) 801.39(2) 1.42 0.96
575 7.29775(10) 17.3854(4) 801.85(2) 1.42 0.89
Table 2 Atomic coordinates of deuterated jarosite
T(K) z(S) z(O1) x(O2) z(O2) x(O3) z(O3) x(D) z(D)
298 0.3077(2) 0.3913(1) 0.22320(7) -0.05488(5) 0.12731(7) 0.13499(6) 0.19585(8) 0.10988(5)
350 0.3070(2) 0.3907(1) 0.22323(7) -0.05518(5) 0.12741(7) 0.13500(6) 0.19555(8) 0.10978(5)
375 0.3066(2) 0.3904(1) 0.22338(7) -0.05532(5) 0.12746(7) 0.13504(6) 0.19533(8) 0.10971(5)
400 0.3061(2) 0.3901(1) 0.22339(7) -0.05547(5) 0.12751(8) 0.13497(6) 0.19523(8) 0.10966(5)
425 0.3057(2) 0.3898(1) 0.22346(7) -0.05562(5) 0.12754(8) 0.13497(7) 0.19506(8) 0.10957(6)
450 0.3053(2) 0.3895(1) 0.22347(7) -0.05573(6) 0.12755(8) 0.13502(7) 0.19490(8) 0.10959(6)
475 0.3051(2) 0.3893(1) 0.22347(7) -0.05581(6) 0.12761(8) 0.13502(7) 0.19480(9) 0.10961(6)
500 0.3045(2) 0.3889(1) 0.22349(7) -0.05581(6) 0.12765(9) 0.13491(7) 0.19467(9) 0.10967(6)
525 0.3041(3) 0.3887(1) 0.22348(8) -0.05578(6) 0.12761(9) 0.13492(8) 0.19449(9) 0.10979(6)
550 0.3038(3) 0.3885(1) 0.22355(9) -0.05562(7) 0.1276(1) 0.13487(9) 0.1942(1) 0.11000(7)
575 0.3035(4) 0.3883(2) 0.2236(1) -0.05566(9) 0.1276(1) 0.1349(1) 0.1941(2) 0.10994(9)
x(K) =y(K) =z(K) =0; x(S) =y(S) =0; x(Fe) =-y(Fe) =-z(Fe) =1/6; x(O1) =y(O1) =0; x(O2) =-y(O2); x(O3) =-y(O3);
x(D) =-y(D)
76 Phys Chem Minerals (2010) 37:73–82
123
Note that the overall intensities of the patterns at 600
and 625 K are much weaker than those at lower tempera-
tures (Fig. 3). More specifically, diffraction peaks for the
newly formed phases, yavapaiite and hematite, are broad
and not well resolved. This behavior suggests that these
phases are probably nanocrystalline in nature, presumably
due to the relatively low temperatures of their formation.
Similar behavior has been observed in simple hydroxides
such as brucite [Mg(OH)
2
], where nanocrystalline MgO
forms upon brucite dehydroxylation at 600 K (Sharma
et al. 2004).
Thermal expansion
Although jarosite has trigonal symmetry (space group
R
3m), its structure is conventionally treated in terms of a
hexagonal cell (defined by two unit-cell parameters aand c).
On heating, both aand cincrease, and thus cell volume
Valso increases (Fig. 4). However, as shown in Fig. 4a
and b (plotted on the same scale), the structural expan-
sion of jarosite occurs at a much higher rate along the
c-axis than along the a-axis and is thus highly aniso-
tropic, which is consistent with the layered nature of its
structure. To obtain the mean coefficients of thermal
expansion (CTEs), we fitted the cell-parameter data to
linear relations:
a¼7:2818 þ2:9756 105Tð2Þ
c¼16:9810 þ7:2339 104Tð3Þ
V¼779:727 þ3:9835 102Tð4Þ
The derived mean CTEs of KFe
3
(SO
4
)(OD)
6
in the
temperature range 298–575 K are: a
a
=4.0814 9
10
-6
K
-1
;a
c
=4.2066 910
-5
K
-1
; and a
V
=5.0322 9
10
-5
K
-1
. Thus the c-axis expands *10 times more
rapidly than the a-axis with increasing temperature.
The cell volume data can also be fitted to a more general
equation for thermal expansion:
VTðÞ¼V0exp ZaTðÞdT
ð5Þ
where V
0
is the volume at a chosen reference temperature,
T
0
, and a(T) is the thermal expansion coefficient, having
the form:
aTðÞ¼a0þa1T:ð6Þ
Using T
0
=298 K, the fit yielded the following
parameters: V
0
=790.99 A
˚
3
,a
0
=1.01 910
-4
K
-1
, and
a
1
=-1.15 910
-7
K
-2
. This fit is excellent, as indi-
cated by an R
2
value of 0.995 and by the fact that the
refined V
0
is approximately the same as the measured V
0
(791.28 A
˚
3
).
Table 3 Isotropic atomic displacement parameters of deuterated
jarosite
T(K) U
iso
(K) U
iso
(S) U
iso
(Fe) U
iso
(O)
a
U
iso
(D)
298 2.9(1) 1.10(7) 0.83(2) 1.21(1) 2.72(3)
350 3.0(1) 1.19(7) 0.89(2) 1.29(1) 2.89(3)
375 3.2(1) 1.27(7) 0.94(2) 1.38(1) 3.08(3)
400 3.4(1) 1.40(7) 1.00(2) 1.46(1) 3.22(3)
425 3.5(1) 1.46(7) 1.05(2) 1.53(2) 3.38(3)
450 3.7(1) 1.54(7) 1.09(2) 1.59(2) 3.50(4)
475 3.8(1) 1.54(8) 1.12(2) 1.64(2) 3.59(4)
500 3.9(1) 1.63(8) 1.18(2) 1.67(2) 3.57(4)
525 3.8(1) 1.70(8) 1.21(2) 1.74(2) 3.60(4)
550 3.8(1) 1.77(9) 1.26(2) 1.78(2) 3.62(4)
575 3.9(2) 1.76(11) 1.34(3) 1.86(3) 3.60(6)
The unit of U
iso
:A
˚
2
/100
a
The U
iso
’s for the three O atoms are set to be equal
Table 4 Selected bond parameters of deuterated jarosite
T (K) K–O2(A
˚) K–O3(A
˚) S–O1(A
˚) S–O2(A
˚) Fe–O2(A
˚)
a
Fe–O3(A
˚)
b
D–O3(A
˚)DO1(A
˚) Fe–O3–Fe(°)
298 2.9721(8) 2.823(1) 1.437(3) 1.479(1) 2.0501(9) 1.9815(4) 0.967(1) 1.952(1) 133.77(6)
350 2.9751(8) 2.828(1) 1.442(3) 1.477(1) 2.0493(9) 1.9825(4) 0.964(1) 1.961(1) 133.69(6)
375 2.9786(8) 2.832(1) 1.447(3) 1.474(1) 2.0501(9) 1.9831(4) 0.962(1) 1.966(1) 133.67(6)
400 2.9805(8) 2.834(1) 1.450(3) 1.472(1) 2.0499(9) 1.9841(4) 0.961(1) 1.970(1) 133.58(6)
425 2.9830(9) 2.836(1) 1.454(4) 1.471(1) 2.0500(9) 1.9847(5) 0.960(1) 1.974(1) 133.55(6)
450 2.9843(9) 2.839(1) 1.459(4) 1.469(2) 2.050(1) 1.9849(5) 0.958(1) 1.979(1) 133.55(6)
475 2.9852(9) 2.842(1) 1.459(4) 1.468(2) 2.051(1) 1.9853(5) 0.957(1) 1.983(2) 133.51(7)
500 2.9860(9) 2.843(1) 1.464(4) 1.465(2) 2.053(1) 1.9863(5) 0.954(1) 1.988(2) 133.40(7)
525 2.986(1) 2.844(1) 1.469(4) 1.463(2) 2.055(1) 1.9873(5) 0.951(1) 1.993(2) 133.40(7)
550 2.986(1) 2.845(1) 1.471(5) 1.460(2) 2.059(1) 1.9866(6) 0.946(1) 2.001(2) 133.38(8)
575 2.987(1) 2.846(2) 1.474(6) 1.458(2) 2.060(2) 1.9866(8) 0.946(2) 2.003(2) 133.37(10)
a
Average of two Fe–O2 edges
b
Average of four Fe–O3 edges
Phys Chem Minerals (2010) 37:73–82 77
123
To the best of our knowledge, the obtained CTEs rep-
resent the first measurement of thermal expansion for
jarosite and its related alunite group. The a
V
value of
5.0322 910
-5
K
-1
falls within the a
V
range for common
compounds. However, it is significantly smaller than the a
V
values of many other hydroxyl-bearing minerals with a
layered structure. For example, brucite, Mg(OH)
2
, has an
a
V
of 10.9 910
-5
K
-1
(Redfern and Wood 1992), about
two times that of jarosite. On the other hand, like brucite,
jarosite exhibits large anisotropy in axial thermal expan-
sion with a much higher CTE along the caxis (normal to
the layer) than perpendicular to c. The mechanisms that
underlie this anisotropic thermal expansion are detailed
below.
Structural variation
Figure 5shows variation of isotropic displacement factors
(U
iso
) for K, Fe, S, O and D with temperature. As expected,
for a given element, its U
iso
increases with increasing
temperature. At a given temperature, U
iso
(Fe) \U
iso
(S) &
U
iso
(O) \U
iso
(D) &U
iso
(K). These trends are consistent
with the decreased bond strengths from Fe to S/O to D/K
(with their neighboring atoms), as U(= kT/f, where kis the
Boltzman constant, Tabsolute temperature, and fthe bond
force constant) is inversely proportional to the bond force
constant. Generally, the lighter the element, the weaker the
bond strength and thus the larger the U
iso
. However,
exceptions do occur, depending on the bonding configu-
ration of a given atom in the structure. In jarosite, K is
situated between the (001) [Fe(O,OH)
6
]/[SO
4
] sheets
(Fig. 1a) and thus has relatively weaker electrostatic
interactions with its adjacent O and OD. As a result, K
exhibits U
iso
values that are similarly high to those of D,
although it is much heavier.
As describe above, the jarosite structure is based on (001)
sheets of [Fe(O,OH)
6
] octahedra and [SO
4
] tetrahedra
(Fig. 1a). [Fe(O,OH)
6
] octahedra are linked via corner-
sharing, forming six- and three-membered rings perpen-
dicular to the c-axis (Fig. 1b). Each three-membered
[Fe(O,OH)
6
] ring is connected to one [SO
4
] tetrahedron
through one of the two sets of apical vertices, and
(A) 2θ= 90º
(B) 2θ= 140º
Intensity (a.u.)
1.0 4.0
3.0
2.0
d (Å)
Fig. 2 A pair of fitted neutron diffraction patterns of deuterated
jarosite collected at a2h=90°and b2h=140°at 298 K. Data are
shown as plus signs, and the solid curve is the best fit to the data. Tick
marks below the pattern show the positions of allowed reflections, and
the lower curve represents the difference between the observed and
calculated profiles
625 K: yavapaiite + hematite
600 K: jarosite + yavapaiite + hematite
575 K: jarosite + yavapaiite + hematite
Hematite 104 Yavapaiite 111/201
Jarosite 003
Intensity (arbitrary unit)
550 K: jarosite
08 282313 18 3833 4843 53 58 63.... .......
d (Å )
-
Fig. 3 Neutron diffraction patterns (2h=90°) of the deuterated
jarosite sample collected at 550, 575, 600, and 625 K. At 575 K,
jarosite started to decompose into yavapaiite and hematite, as
evidenced by the appearance of their diffraction peaks (e.g., hematite
104 and yavapaiite 111/
201). The decomposition was completed at
625 K, as indicated by the disappearance of jarosite peaks such as 003
78 Phys Chem Minerals (2010) 37:73–82
123
neighboring [SO
4
] tetrahedra point in opposite directions
(c?or c-). Therefore, among the four O atoms in a [SO
4
]
tetrahedron, three of them (O2) are each shared by one S and
one Fe, but the fourth O (O1) is bonded only to one
S (Fig. 1c). Since O1 is underbonded relative to O2, the
S–O1 distance is expected to be shorter than S–O2; structure
refinement of jarosite at room temperature shows that its
S–O1 and S–O2 distances are 1.437 and 1.479 A
˚, respec-
tively. These values fall within the observed S–O distances
for 112 refined sulfate structures, which vary from 1.394 to
1.578 A
˚(Hawthorne et al. 2000). However, the S–O2 value
of 1.479 A
˚is very close to the grand mean S–O distance,
1.473 A
˚, calculated from the 112 sulfate structures, whereas
S–O1 is significantly lower. This observation can be
explained using a formal charge model that was initially
developed to explain local structures in alkaline-earth
boroaluminates (Bunker et al. 1991). In this model, only
network-forming cations are considered. The charge dona-
ted by different cations is taken to be the cation charge
divided by the cation coordination number, as in Pauling’s
second rule (Pauling 1960) and Brown and Shannon’s
treatment of bond strengths (Brown and Shannon 1973). In
the case of jarosite, each S
6?
cation donates a charge of ?6/4
(or ?1.5), and each Fe
3?
donates a charge of ?3/6 (or ?0.5).
Because O2 is bonded to one S
6?
and one Fe
3?
, it has a net
charge of zero, resulting in a typical S–O distance associated
with O2. On the other hand, as O1 is bonded only to one S
6?
,
it receives a charge of ?1.5 and has a net charge of –0.5. To
compensate for this charge deficiency, S–O1 bond needs to
be strengthened, thereby causing the contraction of S–O1.
With increasing temperature, S–O1 increases from
1.437 A
˚at 298 K to 1.474 A
˚at 575 K (Table 4). This
behavior is consistent with the larger thermal expansion of
the c-dimension, as S–O1 is parallel to the c-axis (Fig. 1c).
On the other hand, S–O2 decreases from 1.479 A
˚at 298 K
to 1.458 A
˚at 575 K. This S–O2 shortening can be
explained in terms of [Fe(O,OH)
6
] octahedral tilting. On
heating, the Fe-O3-Fe bond angle becomes smaller (from
133.77°at 298 K to 133.37°at 575 K, Table 4). Since
each [SO
4
] tetrahedron shares three O2 atoms with a
three-membered [Fe(O,OH)
6
] ring (one O2 from each
[Fe(O,OH)
6
] octahedron) (Fig. 1b), the narrowing of the
Fe–O3–Fe angel effectively decreases the O2–O2 distance
7.36
7.40
a (Å)
7.20
7.24
7.28
7.32
A
7.16
17 34
17.38
17.42
c (
Å
)
17.22
17.26
17.30
17.
B
17.18
800
802
804
V (Å3)
792
794
796
798
C
T (K)
280 320 360 400 440 480 520 560 600
790
280 320 360 400 440 480 520 560 600
280 320 360 400 440 480 520 560 600
Fig. 4 Variation of unit-cell parameters aa,bc, and ccell volume V
of deuterated jarosite with temperature
4.0
4.5
Å2)
3.0
3.5
K
D
Uiso (1/100
2.0
2.5
D
O
S
Fe
U
1.0
1.5
T(K)
250 300 350 400 450 500 550 600
0.5
T
Fig. 5 Variation of isotropic atomic displacement parameters (U
iso
)
of K, S, Fe, O and D in deuterated jarosite with temperature
Phys Chem Minerals (2010) 37:73–82 79
123
of the [SO
4
] tetrahedron, which causes shortening of the
S–O2 bond. Moreover, as shown in Fig. 1a, [Fe(O,OH)
6
]
octahedral layers are puckered, rather than being flat planar
(which would correspond to a Fe–O3–Fe angle of 180°).
Thus the decrease in the Fe–O3–Fe angle results in an
increase in the degree of the (001) layer puckering via
octahedral tilting. As a result, the overall structure expands
along the c-axis but contracts parallel to the (001) plane.
On the other hand, individual [Fe(O,OH)
6
] octahedra
expand with increasing temperature, as reflected by the
larger Fe–O2 and Fe–O3 distances. This causes expansion
of the structure along both a- and c-axis. It appears that the
net increase in aresulting from the thermal expansion of
[Fe(O,OH)
6
] octahedra is largely canceled by the decrease
due to the octahedral layer puckering, and thus aonly
shows a slight expansion. By contrast, both octahedral
expansion and tilting contribute to the structural expansion
along the c-axis, which, together with the S–O1 lengthen-
ing, leads to a much larger expansion along c.
Although jarosite exhibits a larger thermal expansion
along the c-axis (normal to the layer) than along a,as
observed in other hydrous minerals with a layered struc-
ture, its volume expansion coefficient (a
V
) is significantly
smaller. In other words, the overall jarosite structure is less
flexible in terms of expansion at elevated temperatures.
This behavior can be interpreted on the basis of its unique
structural characteristics. For many layered hydrous min-
erals (such as brucite [Mg(OH)
2
]), their structures can be
treated as consisting of a structural layer (e.g., the [MgO
6
]
layer in brucite) and the interlayer in which weak bonds
(e.g., van der Waals forces) are operating. Thus the thermal
expansion of these layered structures is controlled by both
the structural layer and interlayer, but the latter typically
plays a more significant role. In contrast, the jarosite
structure is comprised only of layers of [Fe(O,OH)
6
]
octahedra and [SO
4
] tetrahedra (Fig. 1a), lacking a distinct
interlayer as in other hydrous structures. The absence of the
interlayer is due to the constraint that neighboring [SO
4
]
tetrahedra linked to different [Fe(O,OH)
6
] octahedral lay-
ers must have the same height along the c-axis (there is
only one crystallographically distinct S in the unit cell of
jarosite) (Fig. 1a). Hence, the thermal expansion of jarosite
is determined solely by the flexibility of its [Fe(O,OH)
6
]/
[SO
4
] sheets, resulting in a smaller a
V
compared with those
of other layered hydrous compounds.
Despite the absence of a distinct weak-bonding inter-
layer in jarosite, the sheets of [Fe(O,OH)
6
] and [SO
4
]
polyhedra are held together by the interstitial K
?
cation via
the K–O2 and K–O3 bonds and the hydrogen bonding
between O1 and D, O1D–O3 (Fig. 1c). As in other lay-
ered hydrous structures, these bonds are weaker than the
bonds within the octahedral/tetrahedral sheets. As a result,
the K–O2, K–O3 and DO1 distances exhibit relatively
larger increases with increasing temperature (Table 4). In
particular, the DO1 attraction, which operates between a
given D (or H) and its closest O1 from the [SO
4
] tetrahe-
dron of the neighboring [Fe(O,OH)
6
]/[SO
4
] sheet (Fig. 1c),
becomes weakened, as manifested by the increase in
DO1 distance (Fig. 6a). In contrast, the O3–D bond
length shows decreases on heating (Fig. 6b), suggesting
that the O3–D bond becomes somewhat stronger. In other
words, in the O1D–O3 bonding configuration, with
increasing temperature, the O3 atoms pull the D atoms
closer, thereby effectively weakening the DO1 attraction.
Hence, the interatomic interactions of D with its neigh-
boring O atoms are interdependent and are largely driven
by the thermal motion of D at elevated temperatures.
Mechanism of jarosite decomposition
It is conceivable that the stability of jarosite is dictated by
the stability of the hydrogen bond, O1D–O3, as it, along
with K
?
, holds the structural sheets of [Fe(O,OH)
6
] octa-
hedra and [SO
4
] tetrahedra together. Once this hydrogen
(Å)
198
1.99
2.00
2.01
D...O1
194
1.95
1.96
1.97
.
A
.
0.965
0.970
B
O3-D (
Å
)
0.950
0.955
0.960
T (K)
250 300 350 400 450 500 550 600
0.940
0.945
250 300 350 400 450 500 550 600
Fig. 6 Variation of interatomic distances aDO1 and bO3–D in
deuterated jarosite as a function of temperature
80 Phys Chem Minerals (2010) 37:73–82
123
bond is broken due to high-temperature dehydroxylation,
the jarosite structure disintegrates into yavapaiite, hematite
and water vapor. More specifically, [Fe(O,OH)
6
] octahedra
become [FeO
6
] octahedra after dehydroxylation, and
one-third of the [FeO
6
] octahedra combine with [SO
4
]
tetrahedra, via corner-sharing, forming [Fe(SO
4
)
2
] sheets
parallel to the (001) plane. These sheets are linked together
by interstitial 10-coordinated K
?
, resulting in a layered
compound, yavapaiite. In the meantime, the remaining
two-thirds of the [FeO
6
] octahedra are connected via edge-
sharing to form gibbsite-type octahedral layers, and the
latter are stacked, via face-sharing, along the c-axis,
forming hematite. Given the structural relations among
jarosite, yavapaiite and hematite, there may be certain
topotactic relations between the decomposed jarosite and
newly formed yavapaiite and hematite. Specifically, the
layered nature of all three phases may result in the fol-
lowing relation: (001)
jarosite
//(001)
yavapaiite
//(001)
hematite
or
c
jarosite
//c
yavapaiite
//c
hematite
. This type of topotactic reaction
(i.e., structurally controlled) mechanism has been found
responsible for the thermal decomposition of many min-
erals including hydroxides, oxyhydroxides and carbonates
(Sharma et al. 2004; Floquet and Niepce 1978). The
occurrence of topotactic relations among jarosite, yavap-
aiite and hematite, however, would require verification by
other techniques such as high-resolution transmission
electron microscopy.
Conclusions
We have studied the stability and structural behavior of
deuterated jarosite in the temperature range 298–650 K
using neutron diffraction in conjunction with Rietveld
analysis. Our results show that jarosite is stable up to
550 K, above which it starts to decompose into nanocrys-
talline yavapaiite and hematite. With increasing tempera-
ture, both the aand cdimension of jarosite expand, but the
latter expands at a rate *10 times larger, as is consistent
with the layered nature of its structure. On the other hand,
because of the lack of a distinct weak-bonding interlayer
between adjacent (001) sheets of [Fe(O,OH)
6
] octahedra
and [SO
4
] tetrahedra, the volume expansion coefficient of
jarosite is significantly smaller than those of many other
hydroxyl-bearing minerals with a layered structure. At a
given temperature, the amplitudes of thermal vibration of
D and K are much larger than those for Fe, O and S,
implying their weaker bonding with surrounding atoms.
Correspondingly, on heating, the DO1 distance of the
hydrogen bond O1D–O3 increases, which suggests
weakened hydrogen bonding between neighboring (001)
tetrahedral/octahedral sheets. By contrast, the O3–D bond
becomes stronger with increasing temperature, a trend also
observed in simple hydroxides such as portlandite (Xu
et al. 2007).
Acknowledgments We thank P.J. Heaney and an anonymous
reviewer for helpful comments, and M.S. Rearick for carrying out
compositional analysis of the jarosite sample. This work has benefited
from the use of the Lujan Neutron Scattering Center at LANSCE,
which is funded by the Department of Energy’s Office of Basic
Energy Sciences. Los Alamos National Laboratory is operated by Los
Alamos National Security, LLC, under DOE Contract DE-AC52-
06NA25396.
References
Anthony JW, McLean WJ, Laughon RB (1972) The crystal structure
of yavapaiite: a discussion. Am Mineral 57:1546–1549
Ballhorn R, Brunner H, Schwab RG (1989) Artificial compounds of
the crandallite type: a new material for separation and immo-
bilization of fission products. Sci Basis Nucl Waste Manage
12:249–252
Basciano LC, Peterson RC (2007) Jarosite-hydronium jarosite solid-
solution series with full iron site occupancy: mineralogy and
crystal chemistry. Am Mineral 92:1464–1473. doi:10.2138/am.
2007.2432
Brown ID, Shannon RD (1973) Empirical bond-strength–bond-length
curves for oxides. Acta Crystallogr A 29:266–282. doi:10.1107/
S0567739473000689
Bunker BC, Kirkpatrick RJ, Brow RK (1991) Local structure of
alkaline-earth boroaluminate crystals and glasses: I. Crystal
chemical concepts—structural predictions and comparisons to
known crystal structures. J Am Ceram Soc 74:1425–1429. doi:
10.1111/j.1151-2916.1991.tb04123.x
Drouet C, Navrotsky A (2003) Synthesis, characterization and
thermochemistry of K–Na–H
3
O jarosites. Geochim Cosmochi-
mica 67:2063–2076. doi:10.1016/S0016-7037(02)01299-1
Drouet C, Pass KL, Baron D, Draucker S, Navrotsky A (2004)
Thermochemistry of jarosite-alunite and natrojarosite-natroalu-
nite solid solutions. Geochim Cosmochimica 68:2197–2205. doi:
10.1016/j.gca.2003.12.001
Dutrizac JE, Jambor JL (2000) Jarosite and their application in
hydrometallurgy. In: Alpers CN, Jambor JL (eds) Sulfate
minerals: crystallography, geochemistry, and environmental
significance, reviews in mineralogy and geochemistry, vol 40.
Mineralogical Society of America, Washington DC, pp 405–452
Floquet N, Niepce JC (1978) Three-fold transformation twin in the
topotactic decomposition of cadmium carbonate crystals. J Mater
Sci 13:766–776. doi:10.1007/BF00570511
Forray FL, Drouett C, Navrotsky A (2005) Thermochemistry of
yavapaiite KFe(SO
4
)
2
: formation and decomposition. Geochim
Cosmochimica 69:2133–2140. doi:10.1016/j.gca.2004.10.018
Frost RL, Weier ML, Martens W (2005) Thermal decomposition of
jarosites of potassium, sodium and lead. J Therm Anal Calorim
82:115–118. doi:10.1007/s10973-005-0850-z
Grohol D, Nocera DG, Papoutsakis D (2003) Magnetism of pure iron
jarosites. Phys Rev B 67:064401-1-13
Hawthorne FC, Krivovichev SV, Burns PC (2000) The crystal
chemistry of sulfate minerals. In: Alpers CN, Jambor JL (eds)
Sulfate minerals: crystallography, geochemistry, and environ-
mental significance, reviews in mineralogy and geochemistry,
vol 40. Mineralogical Society of America, Washington DC, pp
1–112
Inami T, Nishiyama M, Meegawa S, Oka Y (2000) Magnetic structure
of the kagome
´lattice antiferromagnet potassium jarosite
Phys Chem Minerals (2010) 37:73–82 81
123
KFe
3
(OH)
6
(SO
4
)
2
. Phys Rev B 61:12181–12186. doi:10.1103/
PhysRevB.61.12181
Jambor JL (1999) Nomenclature of the alunite supergroup. Can
Mineral 37:1323–1341
Klingelho
¨fer G, Morris RV et al (2004) Jarosite and hematite at
Meridiani Planum from Opportunity’s Mo
¨ssbauer spectrometer.
Science 306:1740–1745. doi:10.1126/science.1104653
Kolitsch U, Tiekink ERT, Slade PG, Taylor MR, Pring A (1999)
Hinsdalite and plumbogummite, their atomic arrangements and
disordered lead sites. Eur J Mineral 11:513–520
Kotler JM, Hinman NW, Yan B, Stoner DL, Scott JR (2008) Glycine
identification in natural jarosites using laser desorption Fourier
transform mass spectrometry: implications for the search for life
on mars. Astrobiology 8:253–266. doi:10.1089/ast.2006.0102
Larson AC, Von Dreele RB (2000) GSAS—general structure analysis
system. Los Alamos National Laboratory Report No. LAUR 86–
748. Los Alamos National Laboratory, Los Alamos
Maslen EN, Strel’tsov VA, Strel’tsova NR, Ishizawa N (1994) Syn-
chrotron X-ray study of the electron density in alpha Fe
2
O
3
. Acta
Crystallogr B 50:435–441. doi:10.1107/S0108768194002284
Menchetti S, Sabelli C (1976) Crystal chemistry of the alunite series:
crystal structure refinement of alunite and synthetic jarosite.
N Jahrb Min Monatsh H 9:406–417
Mymrin VA, Ponte HA, Impinnisi PR (2005) Potential application of
acid jarosite wastes as the main component of construction
materials. Construct Build Mater 19:141–146. doi:10.1016/j.
conbuildmat.2004.05.009
Navrotsky A, Forray FL, Drouet C (2005) Jarosite stability on Mars.
Icarus 176:250–253. doi:10.1016/j.icarus.2005.02.003
Papike JJ, Karner JM, Shearer CK (2006) Comparative planetary
mineralogy: Implications of martian and terrestrial jarosite. A
crystal chemical perspective. Geochim Cosmochim Acta 70:
1309–1321. doi:10.1016/j.gca.2005.11.004
Pauling L (1960) The nature of chemical bond, 3rd edn. Cornell
University Press, Ithaca
Redfern SAT, Wood BJ (1992) Thermal expansion of brucite,
Mg(OH)
2
. Am Mineral 77:1129–1132
Sharma R, Mckelvy MJ, Bearat H, Chizmeshya AVG, Carpenter RW
(2004) In situ nanoscale observations of the Mg(OH)
2
dehydr-
oxylation and rehydroxylation mechanisms. Philos Mag
84:2711–2729. doi:10.1080/14786430410001671458
Stoffregen RE, Alpers CN, Jambor JL (2000) Alunite-jarosite
crystallography, thermodynamics, and geochronology. In: Alpers
CN, Jambor JL (eds) Sulfate minerals: crystallography, geo-
chemistry, and environmental significance, reviews in mineral-
ogy and geochemistry, vol 40. Mineralogical Society of
America, Washington DC, pp 453–479
Vogel SC, Hartig C, Lutterotti L, Von Dreele RB, Wenk HR,
Williams DJ (2004) Texture measurements using the new
neutron diffractometer HIPPO and their analysis using the
Rietveld method. Powder Diffr 19:65–68. doi:10.1154/1.
1649961
Von Dreele RB, Jorgensen JD, Windsor CG (1982) Rietveld
refinement with spallation neutron powder diffraction data.
J Appl Cryst 15:581–589. doi:10.1107/S0021889882012722
Wills AS, Harrison A, Ritter C, Smith RI (2000) Magnetic properties
of pure and diamagnetically doped jarosites: model kagome0
antiferromagnets with variable coverage of the magnetic lattice.
Phys Rev B 61:6156–6169. doi:10.1103/PhysRevB.61.6156
Xu H, Zhao Y, Vogel SC, Daemen LL, Hickmott DD (2007)
Anisotropic thermal expansion and hydrogen bonding behavior
of portlandite: a high-temperature neutron diffraction study.
J Solid State Chem 180:1519–1525. doi:10.1016/j.jssc.2007.03.004
Yildirim T, Harris AB (2006) Magnetic structure and spin waves in
the Kagome jarosite compound KFe
3
(SO
4
)
2
(OH)
6
. Phys Rev B
73:214446. doi:10.1103/PhysRevB.73.214446
82 Phys Chem Minerals (2010) 37:73–82
123