Structural, transport, and electrochemical investigation of novel AMSO 4F (A = Na, Li; M = Fe, Co, Ni, Mn) metal fluorosulphates prepared using low temperature synthesis routes

Article abstract of DOI:10.1021/ic100583f

We have recently reported a promising 3.6 V metal fluorosulphate (LiFeSO₄F) electrode, capable of high capacity, rate capability, and cycling stability. In the current work, we extend the fluorosulphate chemistry from lithium to sodium-based systems. In this venture, we have reported the synthesis and crystal structure of NaMSO₄F candidates for the first time. As opposed to the triclinic-based LiMSO₄F phases, the NaMSO₄F phases adopt a monoclinic structure. We further report the degree and possibility of forming Na(Fe_1-xM_x)SO ₄F and (Na_1-xLi_x)MSO₄F (M = Fe, Co, Ni) solid-solution phases for the first time. Relying on the underlying topochemical reaction, we have successfully synthesized the NaMSO₄F, Na(Fe_1-xM_x)SO₄F, and (Na_1-xLi _x)MSO₄F products at a low temperature of 300 °C using both ionothermal and solid-state syntheses. The crystal structure, thermal stability, ionic conductivity, and reactivity of these new phases toward Li and Na have been investigated. Among them, NaFeSO₄F is the only one to present some redox activity (Fe²⁺/Fe³⁺) toward Li at 3.6 V. Additionally, this phase shows a pressed-pellet ionic conductivity of 10 ^-7 S·cm^-1. These findings further illustrate the richness of the fluorosulphate crystal chemistry, which has just been recently unveiled.

Full text of DOI:10.1021/ic100583f

Inorg. Chem. 2010, 49, 7401–7413 7401
DOI: 10.1021/ic100583f
Structural, Transport, and Electrochemical Investigation of Novel
AMSO4F (A = Na, Li; M = Fe, Co, Ni, Mn) Metal Fluorosulphates
Prepared Using Low Temperature Synthesis Routes
†
†
†
†
†
†
€
Prabeer Barpanda, Jean-Noel Chotard, Nadir Recham, Charles Delacourt, Mohamed Ati, Loic Dupont,
Michel Armand,† and Jean-Marie Tarascon*,†,‡
†
ꢀ
ꢀ
ꢀ
Laboratoire de Reactivite et Chimie des Solides, Universite de Picardie Jules Verne, CNRS UMR 6007,
33 rue Saint Leu, 80039, Amiens, France, and ‡Materials Department, University of California Santa Barbara,
California 93106
Received March 29, 2010
We have recently reported a promising 3.6 V metal fluorosulphate (LiFeSO4F) electrode, capable of high capacity, rate
capability, and cycling stability. In the current work, we extend the fluorosulphate chemistry from lithium to sodium-based
systems. In this venture, we have reported the synthesis and crystal structure of NaMSO4F candidates for the first time.
As opposed to the triclinic-based LiMSO4F phases, the NaMSO4F phases adopt a monoclinic structure. We further report the
degree and possibility of forming Na(Fe1-xMx)SO4F and (Na1-xLix)MSO4F(M=Fe, Co, Ni) solid-solution phases for the first
time. Relying on the underlying topochemical reaction, we have successfully synthesized the NaMSO4F, Na(Fe1-xMx)SO4F,
and (Na1-xLix)MSO4F products at a low temperature of 300 °C using both ionothermal and solid-state syntheses. The crystal
structure, thermal stability, ionic conductivity, and reactivity of these new phases toward Li and Na have been investigated.
Among them, NaFeSO4F is the only one to present some redox activity (Fe2þ/Fe3þ) toward Li at 3.6 V. Additionally, this
phase shows a pressed-pellet ionic conductivity of 10-7 S cm-1. These findings further illustrate the richness of the
3
fluorosulphate crystal chemistry, which has just been recently unveiled.
Introduction
(LiMn2O4, LiMn3/2Ni1/2O4),6,7 olivine structure (LiFePO4),8
tavorite structure (LiFePO4F, LiFePO4OH)9 and so
forth. At present, for safety, olivine-LiFePO4 with a redox
Since Sony met success in commercializing its rechargeable
Li-ion batteries, these systems have seen manifold develop-
ments in the last two decades to become an integral part of
portable electronics, telecommunication and computing de-
vices, and are currently on the anvil to enter into the EV/HEV
market. For successful and secure applications, electrodes
should have high operating voltage, energy density, and rate
capability along with low cost and an environmentally benign
nature.1-3 In the quest for better batteries, various electrodes
have been developed with different structures such as layered
structure (LiCoO2, LiCo1/3Ni1/3Mn1/3O2),4,5 spinel structure
voltage of 3.42 V and a theoretical capacity of 171 mA h g-1 is
3
most widely pursued in both academic and industrial arenas.
Benchmarking against LiFePO4, the research world pursues
the identification of novel electrodes combining good cost,
conductivity, capacity, and the same expected safety due to a
scaffold structure. In such an effort, we have recently reported
LiFeSO4F, a 3.6 V electrode delivering a high capacity of
140 mA h g-1 with good rate capability and cycling stability10
3
without any need for carbon coatings or particles down-
sizing. Here, the high open circuit potential is a combined result
of sulfate (SO4)2- substitution of polyanionic phosphates
(PO4)3- and fluorine (F-) induced larger ionicity.11 These
tavorite-structure based metal fluorosulphates have inter-
connected MO4F2 octahedra and SO4 tetrahedra with tunnels
along [100], [010], [101], and [111] directions facilitating ion
insertion/de-insertion.10
*To whom correspondence should be addressed. E-mail: jean-marie.tarascon@
sc.u-picardie.fr.
(1) Tarascon, J.-M.; Armand, M. Nature 2001, 414, 359–367.
(2) Winter, M.; Besenhard, J. O.; Spahr, M. E.; Novak, P. Adv. Mater.
ꢁ
1998, 10, 725–763.
(3) Armand, M.; Tarascon, J.-M. Nature 2008, 451, 652–657.
(4) Mizushima, K.; Jones, P. C.; Wiseman, P. C.; Goodenough, J. B.
Mater. Res. Bull. 1980, 15(6), 783–789.
(5) Yabuuchi, N.; Makimura, Y.; Ohzuku, T. J. Electrochem. Soc. 2007,
154(4), A314–A321.
(6) Tarascon, J.-M.; Guyomard, D. Solid State Ionics 1994, 69, 222–237.
(7) Zhong, Q.; Bonakdarpour, A.; Zhang, M.; Gao, Y.; Dahn, J. R.
J. Electrochem. Soc. 1997, 144(1), 205–213.
(8) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. J. Electro-
chem. Soc. 1997, 144(4), 1188–1194.
(9) Barker, J.; Saidi, M. Y.; Swoyer, J. L. International Patent
WO0184655.
(10) Recham, N.; Chotard, J.-N.; Dupont, L.; Delacourt, C.; Walker, W.;
Armand, M.; Tarascon, J.-M. Nat. Mater. 2010, 9(1), 68–75.
(11) Padhi, A. K.; Manivannan, V.; Goodenough, J. B. J. Electrochem.
Soc. 1998, 145(5), 1518–1520.
r
2010 American Chemical Society
Published on Web 07/14/2010
pubs.acs.org/IC

7402 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
On another note, cost and resource availability do impact
the battery community, which has led to the growing use of
Mn-, Ti-, and Fe-based electrodes instead of Co- or Ni-based
ones. Along the same line, increasing fears regarding Li
availability and prices are driving new chemistries. Within
this context, Na-based compounds are being considered for
the next generation energy-storage devices, resulting in the
renewed interest for reversible Na-based electrodes such as
Na2FePO4F and NaVPO4F.12-15 Although at a nascent
stage, Na-based intercalation compounds are promising with
their capacity for fast Na-Li ion exchange and successful
use of hard carbons as anodes reaching a capacity as high as
taken as precursors. Here, monohydrate precursors are specifi-
cally used as the product formation occurs via topochemical
reaction.10,18 These monohydrate precursors can be easily pre-
pared from commercially available MSO4 7H2O powders as
3
described in a previous report.18 Stoichiometric amounts of these
precursors [with slight excess (≈ 5 mol%) of NaF] were mixed for
5 minina Spex-5000 shock-type millertoform 1 g offinal product.
The precursor mixture was pressed into pellets by uniaxial press
(10 MPa). These pellets were loaded inside Teflon-lined Parr reac-
tors sealed under Ar atmosphere and were annealed at 300 °C for
40-50 h with an autogenous pressure of 3 bar. The recovered
pellets were ground, washed with ethyl acetate/dichloromethane,
and oven-dried for further analysis. NaMnSO4F only can be
synthesized by annealing a stoichiometric mixture of NaF and
anhydrous MnSO4 at 500 °C for 2 h in air or 550 °C for 12 h in Ar
atmosphere. The 3d-metal substituted Na(Fe1-xMx)SO4F (M =
Co, Ni, Mn) phases were prepared by following the same method
and using an equimolar mixture [with slight excess (≈5 mol%)]
300 mA h g .
-1 16,17 Encouraged by this, we pursued to extend
3
the Li-based metal fluorosulphate systems (LiMSO4F) to
their Na-counterparts (NaMSO4F).
This opens up a variety of questions. Do the NaMSO4F
phases adopt the same crystal structure like LiMSO4F? Is it
feasible to exchange Na with Li and study the underlying ion-
exchange mechanism? What is the feasibility of substituting the
M site for various transition metal elements? Can we form a
solid solution between NaMSO4F-LiMSO4F phases? In an
effort to answer these queries, we attempted to synthesize
NaMSO4F (M = Fe, Co, Ni, Mn) end members as well as Na-
(Fe1-xMx)SO4F (M=Co, Ni, Mn) and (Na1-xLix)MSO4F
(M=Fe, Co, Ni, Mn) mixed phases by adopting both a low-
temperature (300 °C) ionothermal synthesis route and a solid-
state (ceramic) synthesis. We have described both syntheses of
various metal fluorosulphate materials. The structure of var-
ious isostructural NaMSO4F phases has been reported for the
first time. The effect of crystal structure on the ionic conduc-
tivity, ion-exchange reaction, and overall electrochemical per-
formance has been described to gauge the potential of these
novel sodium-based metal fluorosulphate phases for energy-
storage applications.
of NaF and (Fe1-xMx)SO4 H2O precursors. The preparation
3
of (Fe1-xMx)SO4 H2O monohydrate precursors is described
3
elsewhere.18 Further, mixed cations-based (Na1-xLix)MSO4F
phases were prepared following a similar method using stoichio-
metric mixtures of NaF, LiF, and MSO4 H2O precursors.
3
Ionothermal Synthesis. When specified within the text, using
mixtures of the same precursors as that of the ceramic route, the
ionothermal synthesis was conducted using EMI-TFSI [1-Ethyl-
3-methylimidazolium bis(TriFluoromethanesulfonyl Imide)]
ionic liquid media. Stoichiometric amounts of precursor mix-
ture were taken in Teflon-lined autoclaves, 5 cm3 of EMI-TFSI
was added and stirred for 20 min to ensure proper mixing of
precursors inside the ionic liquid. The autoclaves were left un-
disturbed for 15 min, placed inside an oven and heat-treated at
300 °C for 9 h (heating rate 5 °C.min-1). Upon cooling to room
temperature, the product was recuperated from ionic liquid by
centrifugation, washed with dichloromethane (CH2Cl2) and
oven-dried at 60 °C. The ionic liquid can be recuperated for
further usage as described in an earlier report.19
Chemical Oxidation and Ion Exchange Reaction. Ion-
exchange of Naþ for Liþ in the Na-based fluorosulphates
was conducted, for times ranging from hours to days, by
refluxing NO2BF4-based acetonitrile/cyclopentadione solutions
containing fluorosulphate powders or dissolved LiCl/LiBr as a
source of Li.
Experimental Section
Synthesis of AMSO4F Fluorosulphates (A = Na, Li; M = Co,
Ni, Mn). Thanks to the use of ionothermal synthesis, practically
unknown to inorganic chemists, we have succeeded in preparing
LiFeSO4F,10 which turns out to decompose at 360 °C as well as
in water, explaining why this metastable phase was never made
before. Ionic liquids are presently costly. Although they can be
recovered, large-volume production suggests to look for synth-
esis alternatives. Ceramic approaches under autogenous pres-
sure were tried to stabilize fluorosulphates. This was feasible
with a few but not all of them, LiFeSO4F being the less
cooperative. Consequently, the materials reported herein were
prepared by either a ceramic or an ionothermal approach under
the conditions described below.
XRD and Rietveld Refinement. Powder X-ray diffraction
(XRD) patterns of as-synthesized materials were obtained either
by a Bruker D8-Advantage powder diffractometer using Cu-KR
radiation (λ1=1.5405 A, λ2 = 1.5443 A) equipped with a LynxEye
detector or by a Bruker D8-Advantage Diffractometer with
Co-KR radiation (λ1=1.7892 A, λ2=1.7932 A) equipped with a
Vantec-1 detector both operating at 40 kV and 40 mA. The
diffraction patterns were collected in the 2θ range of 10-60° at a
scan speed of 1°/min. The powder patterns were indexed using
DICVOL, and cell parameters were refined by the full pattern
matching method using the FullProf program.20,21 Following, the
temperature-controlled XRD was performed in a Bruker D8
diffractometer (Cu-KR radiation, operating at 40 kV and 30 mA)
with an attached HTK 1200 °C Anton Paar chamber. Powder dif-
fraction patterns were recorded in the 2θ range of 14-32° for
temperatures varying from room temperature (RT) to 450 °C, with
a scan duration of 20 min at every intermediate temperature.
Thermal Gravimetric Analysis. Around 10 mg of powder
sample, kept inside an alumina crucible, was heated from RT
Solid-State Synthesis. For solid-state (ceramic) synthesis of
sodium-based metal fluorosulphates, corresponding metal sulfate
monohydrates (MSO4 H2O, M=Fe, Co, Ni, Mn) and NaF were
3
(12) Ellis, B. L.; Makahnouk, W. R. M.; Makimura, Y.; Toghill, K.;
Nazar, L. F. Nat. Mater. 2007, 6, 749–753.
(13) Ellis, B. L.; Makahnouk, W. R. M.; Rowan-Weetaluktuk, W. N.;
Ryan, D. H.; Nazar, L. F. Chem. Mater. 2010, 22(3), 1059–1070.
(14) Recham, N.; Chotard, J.-N.; Dupont, L.; Djellab, K.; Armand, M.;
Tarascon, J.-M. J. Electrochem. Soc. 2009, 156, A993–A999.
(15) Barker, J.; Saidi, M. Y.; Swoyer, J. L. Electrochem. Solid-State Lett.
2003, 6(1), A1–A4.
(19) Recham, N.; Dupont, L.; Courty, M.; Djellab, K.; Larcher, D.;
Armand, M.; Tarascon, J.-M. Chem. Mater. 2009, 21(6), 1096–1107.
(20) Rodrıguez-Carvajal, J. Recent Developments of the Program FULL-
´
(16) Stevens, D. A.; Dahn, J. R. J. Electrochem. Soc. 2000, 147(4), 1271–
1273.
(17) Komaba, S.; Murata, W.; Ozeki, T. Abstracts of Papers, 216th
Meeting of the Electrochemical Society, Vienna, Austria, Oct 4-9, 2009.
(18) Barpanda, P.; Recham, N.; Chotard, J.-N.; Djellab, K.; Walker, W.;
Armand, M.; Tarascon, J.-M. J. Mater. Chem. 2010, 20(9), 1659–1668.
PROF. CPD Newslett. 2001, 26, 12; available at http/::www.iucr.org/iucr-top/
news/index.html. The program/ documentation can be obtained from http://www.
ill.fr/dif/Soft/fp.
(21) Rodrıguez-Carvajal, J. Physica B 1993, 192, 55–69.
´

Article
Inorganic Chemistry, Vol. 49, No. 16, 2010 7403
to 600 °C (heating rate =10 °C/min) in air (50 mL.min-1
)
ions into the structure, the overall reaction being
regulated atmosphere, and the TGA data were collected with
a Simultaneous Thermal Analyzer STA 449C Jupiter unit
(Netzsch Inc.). The isothermal drift and sensitivity values are
0.6 μg h-1 and 0.1 μg, respectively.
MSO4 H2OðsÞ þ NaFðsÞwNaMSO4FðsÞ þ H2OðgÞ
3
As the sulfate compounds are prone to decomposition at
high temperature, we conducted the synthesis at a mod-
erate temperature of 300 °C for 40 h to ensure complete
reaction (as Naþ diffuses slowly because of its larger size).
These materials were single-phased as determined by XRD
and TEM studies. The XRD patterns of the NaMSO4F end
members (M=Fe, Co, Mn) are presented in Figure 1 panels
a, b, and c, respectively (the case M = Ni is shown in the
Supporting Information). NaMSO4F samples containing
Fe, Co, and Ni were successfully indexed in a monoclinic
cell with a space group P21/c. Further, using the Rietveld
refinement method, the atomic coordinates were determined
for NaFeSO4F and NaCoSO4F phases as given in Table 1
and 2, respectively. Selected interatomic distances of these
compounds are reported in Table 3. For each NaMSO4F
phase (M=Fe, Co, Ni), 38 parameters were refined in the
following order: 1 for the scale, 1 for the sample shift, 4 for the
cell parameters, 3 for the profile of peaks, 24 for the atomic
positions, and finally 5 for the isotropic Debye-Waller
factors. Although we are aware of our accuracy limitation,
it is worth noting that the Fe atoms are displaced from the
geometrical centers of the octahedra resulting in alternat-
inglong(2.00A) and short (1.88 A) Fe-F bonds (e.g., a 0.06 A
displacement from the average octahedral center in the þc
direction, which occurs as well for the Co atomic positions in
NaCoSO4F) (Figure 2a). A similar displacement for the Fe
or Co cations, but in the opposite direction, occurs within the
second chain within the unit cell, which is related to the
first one by a symmetric center. Similar to our earlier
encounter with Mn-based phases (LiMnSO4F), we found
that NaMnSO4F have completely different XRD patterns
than other end-members (Figure 1c), which can be fully
indexed to a monoclinic structure possessing I2/c space
groups with lattice parameters a = 12.1635(5) A, b =
6.6364(2) A, c = 10.3488(4) A, β = 105.068(2)°, and V =
806.65(5) A3 (with Z = 8). Interestingly, this NaMnSO4F
compound has lattice parameters very close to those of
Triplite (Mn,Fe,Mg,Ca)2(PO4)(F,OH) with lattice para-
meters a = 12.085(1) A, b = 6.536(1) A, c = 9.91(1) A,
β=105.633°, V= 753.81 A3 (Z= 8) or of Magniotriplite
(Mg0.89Fe0.88Mn0.23)(PO4)F with lattice parameters a =
12.035(5) A, b=6.432(4) A, c=9.799(2) A, β=108.12(2)°,
V=720.91 A3 (Z = 8) so that one could think our com-
pounds as (Mn,Li)2(SO4)F or (Mn,Na)2(SO4)F analogues.24
This will imply inter site mixing that could explain both (i)
the space group change from P21/c to I2/c in going from
NaFeSO4F to NaMnSO4F and (ii) the inability of the
LiMnSO4F to reversibly insert Li ions as previously encoun-
tered because of poorly mobile 3d cations blocking the Li-ion
diffusion path. However, these structural comments, based
solely on comparisons, are highly speculative and should be
treated as such for the time being. We are currently in the
process of solving the exact structure of Mn-based fluorosul-
phates (i.e., NaMnSO4F and LiMnSO4F) and of under-
standing their structure-property relationships.25
Electron Microscopy. The morphology of all samples was
examined with a FEI Quanta 200 F field-emission scanning electron
microscope (FESEM) equipped with an energy-dispersive X-ray
(EDX) spectrometer. The SEM unit was operated at 20 kV
under low vacuum to avoid any charging effect. Elemental
analysis and mapping was performed at several different spots
to assess homogeneous distribution of constituting elements.
A transmission electron microscopy (TEM) study was conducted
on the powder sample deposited on a TEM holey (with carbon-
copper grid) using an FEI-Tecnai F20 S-Twin electron micro-
scope operating at 200 kV and fitted with an EDAX EDS
system.
Electrical Conductivity Measurement. For conductivity mea-
surement, dense pellets were made by pressing the metal fluoro-
sulphate powder samples first by uniaxial press (500 kg cm-2
)
followed by isostatic press (2500 bar). As fluorosulphates are
prone to decomposition at high temperatures, the densification
process was devoid of any sintering step. The ionically-blocking
electrodes were gold sputtered on both faces of the pellets.
Alternating current (AC) and direct current (DC) conductivity
was measured by employing (AC) impedance spectroscopy (200
kHz to 10 mHz) and 1 V DC polarization (which is below the
decomposition potential). These measurements were conducted
at different temperatures (from RT to 140 °C) using Bio-Logic
VMP3 as detailed elsewhere.22 The activation energies (Ea) for
AC and DC conductivities were calculated as per the Arrhenius
relation: log(δT) = log(δ0T0) - Ea/kB(1/T-1/T0) (where δ =
conductivity (S cm-1) and kB = Boltzmann constant).
3
Electrochemical Testing. The metal fluorosulphate samples
were mechanically milled with Super P carbon black in 80:20
weight ratio for 10 min (under Ar atmosphere) using a Spex-800
shock-milling machine. The electrochemical performance of
these active materials was tested in standard Swagelok-type
cells, using pasted sodium as anode and two sheets of Whatman
GF/D borosilicate glass fiber disks as separator soaked with
either 1 M NaClO4 or Na-TFSI solution in propylene carbonate
(PC). The cells were assembled inside an argon-filled glovebox
with a typical cathode loading of 9-10 mg cm-2. Galvanostatic
3
charge-discharge tests were performed using a MacPile (Claix,
France) controller at 20 °C. Typically, the cells were cycled
between 2.5 to 4.2 V versus Naþ/Na0 at a rate of C/20 (1 Naþ
exchanged per 20 h). Cells containing Co- and Ni-based cath-
odes were charged up to 5 V using an aluminum current collector
to avoid any stainless steel corrosion at high potential.
Results and Discussion
Synthesis of NaMSO4F End Members. Our ionother-
mal synthesis approach of lithium metal fluorosulphate
compounds (LiMSO4F; M = Fe, Co, Ni, Mn),10,16 to-
gether with its recently developed low temperature solid-
state (ceramic) one,23 was implemented to synthesize the
corresponding sodium metal fluorosulphates, with a spe-
cial focus on the solid-state route for its feasibility and
economy. Overall, the synthesis of NaMSO4F com-
pounds involves two parallel steps; first the removal of
the structural H2O from the monohydrate precursor
followed by the introduction of Naþ and F- (from NaF)
(22) Delacourt, C.; Laffont, L.; Bouchet, R.; Wurm, C.; Leriche, J.-B.;
Morcrette, M.; Tarascon, J.-M.; Masquelier, C. J. Electrochem. Soc. 2005,
152(5), A913–921.
(23) Ati, M.; Barpanda, P.; Recham, N.; Moulay, M.; Jumas, J. C.;
Armand, M.; Tarascon, J.-M. J. Electrochem. Soc., manuscript in preparation.
(24) http://webmineral.com/data/.
(25) Barpanda, P.; Chotard, J.-N.; Dupont, L.; Delacourt, C.; Armand,
M.; Tarascon, J.-M. Inorg. Chem., manuscript in preparation.

7404 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
Figure 1. Powder XRD patterns for (a) NaFeSO4F, (b) NaCoSO4F, and (c) NaMnSO4F prepared by solid state synthesis using an equimolar mixture of
corresponding metal sulfate monohydrate precursors and NaF. A typical synthesis was carried out at 300 °C for 40 h. For each case, the experimental
diffraction pattern (red dots), calculated patterns (black lines), Bragg positions (green ticks), and the difference curve (blue line) are shown. The inset figures
depict the respective crystal structure. While NaFeSO4F and NaCoSO4F have monoclinic structure with P21/c symmetry, NaMnSO4F has monoclinic I2/c
symmetry. For NaMnSO4F, the unindexed peak around 32° is due to a slight amount of Na2SO4 or Na2Mn(SO4)2 impurity.
Table 1. Lattice Parameters, Cell Volume, and Atomic Coordinates for NaFeSO4F with Monoclinic (P21/c) Symmetry
NaFeSO4F
a (A)
b (A)
8.7060(2)
β = 113.516(2)°
Rp = 18.3
c (A)
volume (A3)
383.48(2)
6.6799(2)
R = 90°
χ2 = 2.79
7.1914(2)
γ = 90°
Rwp = 8.91
RBragg = 5.28
atom
Wyckoff position
x
y
z
Biso
occupancy
Na
Fe
S
4e
4e
4e
4e
4e
4e
4e
4e
0.226(3)
0.7479(19)
0.259(3)
0.249(3)
0.561(2)
0.360(3)
0.887(2)
0.151(3)
0.9164(6)
0.7572(10)
0.9273(7)
0.1688(8)
0.178(3)
0.488(3)
0.341(3)
0.041(3)
0.253(3)
-0.001(2)
0.762(3)
0.261(4)
0.087(3)
0.152(3)
0.890(4)
0.859(3)
1.35(19)
2.50(8)
1
1
1
1
1
1
1
1
1.88(16)
0.74(18)
0.94(10)
0.94(10)
0.94(10)
0.94(10)
F
O1
O2
O3
O4
Except for the Mn-homologue, all other phases have
similar structures as shown in Figure 1 (insets), which
belong to the titanite class of compounds (e.g., CaTiO-
SiO4, NaTaOGeO4, NaVOPO4) originally reported by
Zachariasen.26-28 Similar to the LiMSO4F (triclinic),
NaMSO4F (monoclinic) have chains of MO4F2 octa-
hedra units (along c-axis) connected to each other by F
vertices in a trans position. These MO4F2 chains are con-
nected to each other by O vertices of SO4 tetrahedra units,
thus forming a 3-dimensional framework. However, the
structure of NaMSO4F differs from that of LiMSO4F in
terms of orientation and distortion of MO4F2 octahedra,
(26) Zachariasen, W. H. Z. Kristallogr. 1930, 73, 7–16.
(27) Malcherek, T. Acta Crystallogr. 2007, B63, 545–550.
(28) Lii, K.-H.; Li, C.-H.; Chen, T.-M.; Wang, S.-L. Z. Kristallogr. 1991,
197, 67–73.

Article
Inorganic Chemistry, Vol. 49, No. 16, 2010 7405
Table 2. Lattice Parameters, Cell Volume, and Atomic Coordinates for NaCoSO4F with Monoclinic (P21/c) Symmetry
NaCoSO4F
a (A)
b (A)
8.6174(4)
β = 114.343(3)°
Rp = 5.5
c (A)
volume (A3)
373.89(3)
6.6664(3)
R = 90°
χ2 = 2.20
7.1436(3)
γ = 90°
Rwp = 7.4
RBragg = 1.68
atom
Wyckoff position
x
y
z
Biso
occupancy
Na
Co
S
4e
4e
4e
4e
4e
4e
4e
4e
0.260(3)
0.7468(16)
0.257(3)
0.244(4)
0.585(3)
0.335(3)
0.916(4)
0.126(3)
0.9132(6)
0.7500(13)
0.9319(8)
0.1681(14)
0.177(4)
0.489(3)
0.340(4)
0.046(3)
0.239(4)
3.7(3)
1
1
1
1
1
1
1
1
-0.006(3)
0.761(3)
0.262(8)
0.099(5)
0.132(4)
0.890(5)
0.846(3)
3.00(16)
2.3(3)
2.6(3)
1.00(19)
1.00(19)
1.00(19)
1.00(19)
F
O1
O2
O3
O4
Figure 2. (a) Chains of interconnected MO4F2 (M = Fe) octahedra and SO4 tetrahedra depicting the displacement of Fe atoms from their geometric centers,
as shown by the white square marks. (b) The comparative crystal structures of monoclinic sodium-based and triclinic lithium based metal fluorosulphates.
as shown in Figure 2b. The shape of cavity (marked
by 1) is more distorted in NaMSO4F structure. There-
fore, the major difference occurs in the sodium site. While
the (smaller) Li can occupy 2 sites with equal prob-
ability,10,29 there is a fixed site for the (larger) Na atom. We
intend to conduct neutron diffraction studies in the near
future to obtain further structural insights on these newly-
found materials. For all these phases (with four formula
unit per unit cell), we could calculate the theoretical
Table 3. Selected Interatomic Distances (A) in NaFeSO4F and NaCoSO4F
Focusing on the MO4F2 Octahedra and SO4 Tetrahedra
distance (A)
NaFeSO4F
NaCoSO4F
M - O1
M - O2
M - O3
M - O4
M - F
1.99(2)
2.37(2)
2.40(2)
2.01(3)
1.88(3)
2.00(3)
2.12(2)
2.23(3)
2.19(3)
2.08(3)
1.86(6)
1.98(6)
M - F
density to be 3.36, 3.49, and 3.51 g cm-3 for NaFeSO4F,
3
S - O1
S - O2
S - O3
S - O4
1.56(2)
1.43(3)
1.36(3)
1.55(3)
1.45(3)
1.41(4)
1.45(3)
1.59(3)
NaCoSO4F, and NaNiSO4F, respectively. Figure 3(a-d)
illustrates the morphology of solid-state synthesized
NaMSO4F phases with micrometric particles. For NaFe-
SO4F, the TEM study reveals particle sizes ranging
from 0.5-1 μm (Figure 3e), and the corresponding
SAED pattern shows an array of pseudo-hexagonal
symmetry dots, which is indexed using [6-25]* zone axis
(Figure 3f). Following the current work, we have recently
succeeded in the low-temperature preparation of the
monoclinic-structured NaCuSO4F, NaZnSO4F, and NaMg-
SO4F phases to be reported in a future communication
(29) Sebastian, L.; Gopalakrishnan, J.; Piffard, Y. J. Mater. Chem. 2002,
12, 374–377.

7406 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
Figure 3. SEM images of solid-state synthesized (a) NaCoSO4F, (b) NaFeSO4F, (c) NaNiSO4F, and (d) NaMnSO4F. For the NaFeSO4F homologue,
a TEM image (e) and corresponding SAED pattern (f) are shown.
addressing their synthesis, structure, and conductivity
data.30
Synthesis of Na(Fe1-xMx)SO4F Solid-Solution Phases.
As the synthesis of metal fluorosulphate compounds in-
volves a topotactic reaction mechanism having a striking
structural similarity between the mother precursor mono-
hydrate phase (monoclinic with C2/c symmetry) and the
product phase (monoclinic with P21/c symmetry), we have
used (Fe1-xMx)SO4 H2O precursors (where, M = Co, Ni,
3
Mn) to fabricate 3d-transition metal-substituted NaFeSO4F
compounds. These solid-solution precursors containing
Fe-Co, Fe-Ni, and Fe-Mn mixing at the molecular level
were synthesized by dissolution and precipitation methods as
described earlier.18 The powder XRD patterns and corres-
ponding lattice parameters of these (Fe1-xMx)SO4 H2O
3
precursors are given in the Supporting Information. These
precursors were reacted with NaF to obtain the correspond-
ing Na(Fe1-xMx)SO4F phases using the same experimental
protocol as for NaMSO4F end members.
The XRD patterns of Co, Ni, and Mn-substituted Na-
(Fe1-xMx)SO4F members were collected and are reported
in Figure 4 for the Na(Fe0.5M0.5)SO4F phases (M = Co, Ni,
and Mn) only (see the Supporting Information for the
complete series of the XRD). The related lattice parameters
and cell volume data are summarized in Tables 4, 5, and 6,
respectively, with the variation of the volume as a function of
x shown in Figure 5. The substituting 3d-metals occupy the
Fe-site in the FeO4F2 octahedra, keeping the 3-dimensional
framework structure intact. On the basis of simple ionic
radius considerations [MnII (97 pm)>FeII (92 pm)>CoII
(88.5 pm)>NiII (83 pm)]31 one would expect the unit cell
volume to decrease with Co and Ni substitutions and to
Figure 4. Comparative XRD patterns of 3d transition metal substituted
sodium iron fluorosulphates for the Na(Fe0.5M0.5)SO4F series for M = Co,
M=Ni, and M=Mn. These compounds were prepared via solid state
synthesis using equimolar mixtures of corresponding (Fe1-xMx)SO4
3
H2O monohydrate precursors and NaF. The major impurity peak around
23° is due to the presence of MSO4 4H2O intermediate phases.
3
increase with the Mn substitution as we experimentally
observed (Figure 5). Note also the existence of a complete
solid solution for the Co- and Ni-based Na(Fe1-xMx)SO4F
series having isostructural end members. In contrast, the
solid solution does not extend beyond x = 0.6 for the Na-
(Fe1-xMnx)SO4F series. This does not come as a surprise as
NaMnSO4F is not isostructural with NaFeSO4F. We could
not extend the range of this solid solution regardless of
whether we used solid-state or ionothermal synthesis routes.
Besides, it is worth noting that we could not prepare these
solid-solution phases either by ionothermal synthesis or by
solid-state thermal treatment of a mixture of NaFeSO4F and
(30) Barpanda, P.; Chotard, J.-N.; Dupont, L.; Armand, M.; Tarascon,
J.-M. Chem. Mater., manuscript in preparation.
(31) http://www.webelements.com.

Article
Inorganic Chemistry, Vol. 49, No. 16, 2010 7407
Table 4. Lattice Parameters and Cell Volume of Na(Fe1-xCox)SO4F Solid Solution Phases Having Monoclinic (P21/c) Symmetry
materials
NaFeSO4F
Na(Fe0.8Co0.2)SO4F
Na(Fe0.6Co0.4)SO4F
Na(Fe0.5Co0.5)SO4F
Na(Fe0.4Co0.6)SO4F
Na(Fe0.2Co0.8)SO4F
NaCoSO4F
a (A)
b (A)
c (A)
β (deg)
V (A3)
6.6799(2)
6.6759(4)
6.6721(9)
6.6760(6)
6.6698(5)
6.6670(7)
6.6664(3)
8.7060(2)
8.6804(3)
8.6632(2)
8.6634(5)
8.6534(4)
8.6392(3)
8.6174(4)
7.1914(2)
7.1742(3)
7.1612(5)
7.1564(3)
7.1535(3)
7.1458(3)
7.1436(3)
113.516(2)
113.72(3)
113.91(6)
114.03(4)
114.05(4)
114.24(2)
114.343(3)
383.48(2)
380.62(4)
378.39(4)
378.02(3)
377.02(2)
375.27(6)
373.89(3)
Table 5. Lattice Parameters and Cell Volume of Na(Fe1-xNix)SO4F Solid Solution Phases Having Monoclinic (P21/c) Symmetry
materials
NaFeSO4F
Na(Fe0.8Ni0.2)SO4F
Na(Fe0.6Ni0.4)SO4F
Na(Fe0.5Ni0.5)SO4F
Na(Fe0.4Ni0.6)SO4F
Na(Fe0.2Ni0.8)SO4F
NaNiSO4F
a (A)
b (A)
c (A)
β (deg)
V (A3)
6.6799(2)
6.6632(5)
6.6479(3)
6.6397(2)
6.6278(8)
6.6193(9)
6.6099(3)
8.7060(2)
8.6616(6)
8.6192(3)
8.6024(1)
8.5818(1)
8.5503(2)
8.5252(3)
7.1914(2)
7.1671(2)
7.1534(1)
7.1391(4)
7.1321(3)
7.1078(6)
7.1032(1)
113.516(2)
113.83(4)
114.04(1)
114.11(5)
114.18(2)
114.48(1)
114.67(1)
383.48(2)
378.37(3)
374.31(1)
372.17(5)
370.06(3)
366.11(6)
363.72(1)
Table 6. Lattice Parameters and Cell Volume of Na(Fe1-xMnx)SO4F Solid Solution Phases Having Monoclinic (P21/c) Symmetry
materials
NaFeSO4F
a (A)
b (A)
c (A)
β (deg)
V (A3)
6.6799(2)
6.6970(9)
6.7132(7)
6.7298(8)
6.7310(8)
6.7437(6)
8.7060(2)
8.7156(2)
8.7283(6)
8.7360(7)
8.7337(5)
8.7464(6)
7.1914(2)
7.2088(7)
7.2177(5)
7.2475(7)
7.2598(9)
7.2787(4)
113.516(2)
113.58(7)
113.71(3)
113.65(1)
113.76(5)
113.69(1)
383.48(2)
385.62(9)
387.21(6)
390.30(7)
390.58(5)
393.15(5)
Na(Fe0.9Mn0.1)SO4F
Na(Fe0.8Mn0.2)SO4F
Na(Fe0.6Mn0.4)SO4F
Na(Fe0.5Mn0.5)SO4F
Na(Fe0.4Mn0.6)SO4F
Na(Fe0.2Mn0.8)SO4F
NaMnSO4F
12.1635(5)
6.6364(2)
10.3488(4)
105.06(2)
806.65(5)
and more polarizing Li tends to form a triclinic structure
(LiMSO4F) where the Li-ions are disordered over two
sites occupied with equal probability. So a legitimate
question regards the degree of compatibility between
structurally different Li- and Na-based end phases, hence,
the need to study the various (Na1-xLix)MSO4F series
with M = Fe, Co, and Mn.
As Mn-based fluorosulphates show identical mono-
clinic structures regardless of whether the guest alkali ion
is Na or Li, we first attempted to prepare (Na1-xLix)Mn-
SO4F compounds via solid-state synthesis at 300 °C. The
collected XRDs for the various members of the series are
graphically presented in Figure 6a. There is a noticeable
and gradual shift of the main Bragg peaks toward a lower
angle, with increasing the Na content, implying the
existence of a complete solid solution. However, a keen
look reveals the possible presence of a narrow biphasic
domain around (Na0.5Li0.5)MnSO4F. The shift toward
the lower angle (e.g., higher d-spacing and larger unit cell
volume, Figure 6b) is consistent with the fact of replacing
small Li-ions (76 pm) by larger Na-ions (94 pm).30 Thus,
the increase in Na-content systematically leads to larger
unit cell parameters as reported in Table 7.
Figure 5. Graph showing the variation in the unit cell volume of 3d
transition metal substituted Na(Fe1-xMx)SO4F (M = Co, Ni, Mn) com-
pounds. Because of its larger size, Mn substitution increases the cell
volume, whereas Co and Ni substitution results in decrease in unit cell
volume. The black dashed lineindicates the volumeof pristine NaFeSO4F
as a reference.
Next, we attempted to synthesize the following (Na1-x
-
NaMSO4F end members, further signifying the importance
of the underlying topochemical reaction (Supporting In-
formation).
Synthesis of (Na1-xLix)MSO4F Phases. The nature of
alkali in these polyanionic metal fluorosulphate com-
pounds (AMSO4F, A = Na, Li) plays a pivotal role in
shaping the orientation, symmetry, and final crystal
structure because of its size and polarization. While Na
favors a monoclinic structure (NaMSO4F), the smaller
Lix)MSO4F series (M = Fe and Co) bearing in mind that the
Na- and Li-end members are not isostructural. The collected
XRD profiles of Fe- and Co-based compounds were refined,
and the corresponding lattice parameters reported in Table 8
and 9, respectively (only those for x = 0.2 and 0.3 are shown
in Figure 7; refer to the Supporting Information for all com-
positions). Preparing mixed-cation Ni-based phases was
relatively difficult. Whatever the nature of M, we were able
to produce cationic solid-solution(Na1-xLix)MSO4Fphases

7408 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
Table 7. Lattice Parameters and Cell Volume of (Na1-xLix)MnSO4F Mixed Cation Solid Solution Phases Possessing Monoclinic (I2/c) Symmetry
materials
NaMnSO4F
a (A)
b (A)
c (A)
β (deg)
V (A3)
12.1635(5)
12.2007(5)
12.1549(5)
12.1110(6)
12.0789(8)
12.0910(3)
12.0256(0)
11.9997(7)
11.9605(9)
11.9277(9)
11.8833(11)
6.6364(2)
6.6201(3)
6.6327(7)
6.5970(2)
6.5600(3)
6.5564(3)
6.5174(5)
6.4833(3)
6.4651(5)
6.4416(5)
6.4130(6)
10.3488(4)
10.3453(1)
10.3373(3)
10.2981(0)
10.2749(3)
10.2278(3)
10.2028(8)
10.1436(1)
10.1519(7)
10.1140(4)
10.0294(10)
105.06(2)
105.35(8)
105.11(5)
105.29(4)
105.57(6)
105.49(8)
105.67(7)
105.72(6)
105.94(8)
106.06(7)
106.15(3)
806.65(5)
805.74(8)
804.54(0)
793.66(5)
784.27(2)
781.34(4)
769.94(5)
759.61(2)
754.80(5)
746.74(1)
734.51(1)
(Na0.9Li0.1)MnSO4F
(Na0.8Li0.2)MnSO4F
(Na0.7Li0.3)MnSO4F
(Na0.6Li0.4)MnSO4F
(Na0.5Li0.5)MnSO4F
(Na0.4Li0.6)MnSO4F
(Na0.3Li0.7)MnSO4F
(Na0.2Li0.8)MnSO4F
(Na0.1Li0.9)MnSO4F
LiMnSO4F
substitute Li for Na, even a minute amount (say 5%) of Na-
substitution led to the simultaneous (and independent)
formation of Li- and Na- phases. From our experience,
we observe that the triclinic LiMSO4F-tavorite phases are
on the verge of structural instability, and a slight distortion
created by non-isostructural substitution results in structur-
al decomposition or phase separation. A synergistic study of
NMR and density functional theory (DFT) calculation may
shed light on the structural instability of Li-rich (Na1-xLix)-
MSO4F phases.
Thermal Analysis. Phosphate-based electrodes gener-
ally exhibit high thermal stability owing to the strong P-O
and M-O bonds, which is crucial for safe battery operation.
In our earlier report on LiMSO4F as well as Li(Fe1-xMx)-
SO4F, we found that these compounds, as deduced by
thermogravimetric and temperature-controlled XRD study,
are stable up to 350 °C, amply sufficient for batteries but
reflecting the lesser stability of the S-O bond.10,18 Similar
measurements were used to determine the thermal stability
of NaMSO4F end members.
The TGA analysis (in air) of NaMSO4F end members is
shown in Figure 8a. NaFeSO4F is stable up to 350 °C, after
which there is a gradual weight loss, which can be due to the
partial conversion of SO4 species into SO2 gas. The thermal
degradation is more severe in case of NaFeSO4F (∼16%
weight loss) in comparison to LiFeSO4F (∼5% weight loss),
which indicates that the weaker bonding is associated with
Na-based compounds. Other Na-homologues show similar
nature, dissociating around 300 °C. NaMnSO4F, which
adopts a different XRD pattern than other homologues,
have excellent thermal stability with little decomposition up
to 600 °C in air (Figure 8a). As a rule of thumb, the reticular
energy is usually higher for compounds having similar sized
ions (cations and anions). Likewise for NaMSO4F phases,
thermal stability should be higher for the compounds having
the closest sized ions (i.e., M2þ cations and SO4F3- anions)
knowing that Mn2þ, Fe2þ, and Co2þ have ionic radii of 97,
92, and 88.5 pm in octahedral coordination, respectively.31
Besides, the observed stability for the Mn-based fluoro-
sulphate can be further explained by the less reductive effect
of Mn2þ as compared to the divalent Co and Fe ions. This
reducing strength argument could also explain the observed
stability of the Mg-based fluorosulphate at temperatures as
high as 600 °C.
Figure 6. (a) Comparative XRD patterns of solid solution between
NaMnSO4F and LiMnSO4F, both having monoclinic structure with
space group P21/c. The numbers alongside the patterns refer to the x
value in (Na1-xLix)MnSO4F. (b) Graph showing the variation of cell
parameters and cell volume of (Na1-xLix)MnSO4F compounds as a
function of Na content.
up to ∼30% Li still bearing the monoclinic NaMSO4F
structure in all cases with minor changes in lattice para-
meters. However, with higher Li content (40-100%), we
systematically failed to form solid-solution phases despite
several attempts both by solid-state and by ionothermal
routes. Instead, a mixture of stoichiometric amount of
NaMSO4F and LiMSO4F end phases was obtained. Inter-
estingly, when we started from the Li-rich side and tried to
For further confirmation on thermal stability, a tem-
perature-controlled XRD study was performed on a
representative end member, that is, NaFeSO4F (as shown
in Figure 8b). The pristine phase was retained up to 325 °C
and decomposition started occurring ∼350 °C, as depicted
by the appearance of a few extra peaks. Further heating

Article
Inorganic Chemistry, Vol. 49, No. 16, 2010 7409
Table 8. Lattice Parameters and Cell Volume of (Na1-xLix)FeSO4F Mixed Cation Solid Solution Phases Possessing Monoclinic (P21/c) Symmetry
materials
NaFeSO4F
a (A)
b (A)
c (A)
R (deg)
β (deg)
γ (deg)
V (A3)
6.6799(2)
6.6729(5)
6.6717(6)
6.6615(1)
8.7060(2)
8.7046(6)
8.7005(9)
8.6903(4)
7.1914(2)
7.1885(4)
7.1893(6)
7.1666(9)
90.0
90.0
90.0
90.0
113.51(2)
113.48(8)
113.51(1)
113.64(7)
90.0
90.0
90.0
90.0
383.48(2)
382.96(9)
382.66(4)
380.03(3)
(Na0.9Li0.1)FeSO4F
(Na0.8Li0.2)FeSO4F
(Na0.7Li0.3)FeSO4F
(Na0.6Li0.4)FeSO4F
(Na0.5Li0.5)FeSO4F
(Na0.4Li0.6)FeSO4F
(Na0.2Li0.8)FeSO4F
(Na0.1Li0.9)FeSO4F
LiFeSO4F
5.1747(3)
5.4943(3)
7.2224(3)
90.0
107.2(3)
97.7(3)
182.55(1)
Table 9. Lattice Parameters and Cell Volume of (Na1-xLix)CoSO4F Mixed Cation Solid Solution Phases Possessing Monoclinic (P21/c) Symmetry
materials
NaCoSO4F
a (A)
b (A)
c (A)
R (deg)
β (deg)
γ (deg)
V (A3)
6.6664(3)
6.6824(5)
6.6958(6)
6.7019(7)
8.6174(4)
8.6423(6)
8.6366(5)
8.6337(7)
7.1436(3)
7.1469(3)
7.1458(3)
7.1482(3)
90.0
90.0
90.0
90.0
114.343(3)
114.42(6)
114.49(5)
114.48(6)
90.0
90.0
90.0
90.0
373.89(3)
375.81(6)
376.05(6)
376.40(6)
(Na0.9Li0.1)CoSO4F
(Na0.8Li0.2)CoSO4F
(Na0.7Li0.3)CoSO4F
(Na0.6Li0.4)CoSO4F
(Na0.5Li0.5)CoSO4F
(Na0.4Li0.6)CoSO4F
(Na0.2Li0.8)CoSO4F
(Na0.1Li0.9)CoSO4F
LiCoSO4F
5.1721(7)
5.4219(7)
7.1842(8)
90.0
107.7(6)
97.9(5)
177.80(4)
system, colloquially termed as NASICON,33-36 was in-
troduced by Hong and Goodenough. As our materials
have a Na-based polyanionic structure like NASICON
although the 3D-packing is different, and knowing
furthermore that conductivity is a key parameter for
successful electrode application, we investigated the
transport properties of NaFeSO4F NaCoSO4F, and their
solid-solution systems.
The conductivity results are summarized in Figure 9. Typi-
cal impedance spectra (Figure 9, inset) consist of a slightly
depressed semicircle (at high frequency) and a defined War-
burg diffusion tail (at lower frequency), suggesting the ionic
nature of AC conductivity. This is further confirmed by the
lower values of DC conductivity and the inability to measure
DC conductivity at lower temperatures (300-340 K). The
AC (ionic) conductivity value of NaFeSO4F was measured to
be 7.14ꢀ10-7 S cm-1 at ambient temperature (Figure 9a),
3
which is four orders of magnitude higher with respect to
LiFeSO4F (7.0 ꢀ 10-11 S cm-1).10 The NaFeSO4F has a
3
much lower AC activation energy (0.29 eV) in comparison to
that of LiFeSO4F (0.99 eV). It is evident that these results on
cold-pressed pellets suggest a bulk conductivity of several
orders higher magnitude. NMR diffusion coefficient mea-
surements are currently underway. The monoclinic structured
NaFeSO4F has a rigid oxo-fluoride framework of corner-
shared FeO4F2 octahedra and SO4 tetrahedra. Such a high
ionic transport suggests that the Naþ ions are well connected
by channels of low energy barriers allowing fast ion transport.
Furthermore, unlike highly polarizing Liþ that “sticks” to the
walls of the galleries, Naþ, with a lower surface charge
density, is likely to stay in the center as in β-alumina. This
NaMSO4F structure permits a wide range of M-site substitu-
tion forming a versatile family of superionic conducting solids
soon to be reported.30 Ongoing first principle calculations
Figure 7. Comparative XRD patterns of solid solution between mono-
clinic sodium metal fluorosulphates and triclinic lithium metal fluorosul-
phates for the x = 0.2 and 0.3 members (Na1-xLix)MSO4F series with
M = Fe and Co. For M = Ni, it was difficult to produce (Na1-xLix)Ni-
SO4F phases, so they are not shown here.
aggravate the decomposition forming secondary phases
like Fe2O3, Na2SO4, Fe2(SO4)3, and FeS. From this study,
a general conclusion can be drawn: all the NaMSO4F
homologues as well as their derivatives are stable at least
up to ≈325 °C.
Electrical Conductivity of NaFeSO4F. Since the discov-
ery of high ion mobility in sodium β-Al2O3 ((Na2O)1þx
3
11Al2O3),32 there has been active search for superionic
conductors for batteries and fuel cells. One such material
(35) Goodenough, J. B.; Hong, H. Y-P.; Kafalas, J. A. Mater. Res. Bull.
1976, 11, 203–220.
(36) Padma Kumar, P.; Yashonath, S. J. Chem. Sci. 2006, 118(1),
135–154.
(32) Yao, Y. F.; Kummer, J. T. J. Inorg. Nucl. Chem. 1967, 29, 2453–2475.
(33) Tran Qui, D.; Capponi, J. J.; Gondrand, M.; Saıb, M.; Joubert, J. C.;
Shannon, R. D. Solid State Ionics 1981, 3-4, 219–222.
(34) Hong, H. Y-P. Mater. Res. Bull. 1976, 11, 173–182.

7410 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
mity of 10-7 S cm-1, with very similar values of acti-
vation energy. This affirms that these calcite-based NaM-
SO4F monoclinic phases have a very high ionic conductivity
at room temperature. We are currently in the process of
further investigating the transport properties of NaMSO4F
phases, specially having a non paramagnetic metal center
(M = Mg, Zn), using NMR spectroscopy.
Electrochemical Studies of Metal Fluorosulfates. To
gauge the electrochemical activity and possibility of
reversible Naþ extraction from these novel fluoro-
sulphate phases, we first conducted chemical oxidation
of pristine NaFeSO4F with NO2BF4. Irrespective of the
relative molar amount of NO2BF4, reacting time (1 to 4
days) or temperature (20-80 °C) at which the reaction
was conducted, no chemical oxidation was observed, with
the NaFeSO4F structure remaining intact (Supporting
Information). This is somewhat surprising as very effec-
tive chemical oxidation has been earlier reported for Na-
based compounds like Na2FePO4F, having a layered
orthorhombic structure favoring facile Na extraction.13
Besides, we also failed to perform any ion exchange by
chemical reaction with excess LiCl or LiBr in ionic liquid/
acetonitrile/cyclopentadione medium (Supporting Infor-
mation). For both NaFeSO4F and NaCoSO4F, there was
absolutely no success in Na-ion exchange by Li.
Next, the NaFeSO4F end member was tested in an
electrochemical cell to check its activity. An 80:20 powder
mixture of NaFeSO4F and SP carbon black was cycled at
C/20 rate in a sodium cell. Though NaFeSO4F has a
theoretical capacity of 137 mAh g-1, we could hardly
3
achieve ∼6% of capacity (Figure 10a). Like LiFeSO4F,
NaFeSO4F too has a FeII/FeIII signature plateau centered
around 3.6 V. The other NaMSO4F (M=Co, Ni, Mn)
homologues were cycled up to 5 V in Swageloks, still
giving no redox reaction. A point to note is that we used
solid-state synthesized NaFeSO4F for electrochemical
study, which gives similar results to the ionothermal syn-
thesized NaFeSO4F, as there is no particle coarsening/
grain-growth at moderate temperature of 300 °C.
Figure 8. (a) Thermogravimetric analysis (TGA) of NaMSO4F homo-
logues (M = Fe, Co, Mn) conducted from RT to 600 °C in air. Barring
NaMnSO4F, which is thermally stable, all other homologues decompose
above 350 °C. The TGA curve of LiFeSO4F is also provided for reference.
(b) Temperature-controlled XRD patterns of NaFeSO4F conducted from
RT to 450 °C. The number along the XRD pattern refers to the heating
temperature (°C). Several extra peaks start appearing after 350 °C (as
marked) indicating thermal decomposition of parent material.
As LiFeSO4F delivers excellent electrochemical cycling,10
we prepared chemically oxidized Li0.75FeSO4F, Li0.25Fe-
SO4F, and FeSO4F from parent LiFeSO4F. These materials
were cycled in a sodium cell architecture with 1 M Na-TFSI:
PC electrolyte to notice the effective Na insertion capa-
bility. The resulting electrochemical curves are shown in
Figure 10b (and Supporting Information). Similar to earlier
data, we observe very inefficient Na-insertion into these
materials upon cycling. Overall, at this nascent stage,
sodium metal fluorosulphates are electrochemically limited,
the main reason being most likely associated to their struc-
tures as discussed next.
under the DFT þ U framework should help in clarifying
transport properties within this novel family of fluorosul-
phates.
Following, DC conductivity was tested by applying a
steady external voltage of 1 V to the pellet and recording
the steady-state current (iSS), which is solely due to the
electrons. Owing to low electronic conductivity, we could
not measure any current below 80 °C, because of too
small currents. Extrapolating the log(δ) ∼ (1000/T) curve,
the DC conductivity value was measured to be 1.31 ꢀ
10-12 S cm-1 at ambient temperature, which is slightly
lower than for LiFeSO4F (5.2ꢀ 10-11 S cm-1). The DC
Discussion
3
activation energy for NaFeSO4F (0.70 eV) was compar-
able to that of LiFeSO4F (0.75 eV).
We have exploited the rich chemistry offered by the fluoro-
sulphate to successfully prepare novel Na-based 3d metal
fluorosulphates NaMSO4F (M = Fe, Ni, Mn, and Co). Such
phases structurally differ from their Li counterparts previously
reported as they crystallize in a monoclinic (P21/c) rather than
in a triclinic (P1) structure except for Mn which crystallizes in a
monoclinic cell whose space group has not yet been deter-
mined. Such a Na-driven triclinic-monoclinic transition, which
is associated with a unit cell volume of ≈9 A3 from LiFeSO4F
We observed similar values of AC and DC conductivity as
well as activation energy values for NaCoSO4F (Figure 9c),
further proving the superionic conductivity of these NaM-
SO4F phases. Eventually, we examined the transport proper-
ties of Na(Fe1-xCox)SO4F solid-solution systems, using
three intermediate compositions. As shown in Figure 9b,
their AC (ionic) conductivity values are in the close proxi-

Article
Inorganic Chemistry, Vol. 49, No. 16, 2010 7411
Figure 9. (a) Conductivity of NaFeSO4F pellet at different temperatures coated with ionically blocking Au electrodes. Inset image shows representative
AC impedance spectra (Nyquist plot). They have a slightly depressed high-frequency semicircle and a low-frequency Warburg diffusion line. (b) The AC
conductivity values ofthree compositions inNa(Fe1-xCox)SO4F (x = 0.4, 0.5, 0.6) solid-solution systemwith their ACactivationenergy reported. The inset
image at top right corner shows the equivalent circuit consisting of a resistance (R1) and a constant phase element (CPE1) in parallel connection, along with
Warburg resistance (W1). The ionic conductivity value was calculated from the resistance of high-frequency semicircle. (c) Conductivity of ionically
blocking NaCoSO4F pellet at different temperature having similar transport properties as NaFeSO4F.
(182.56 A3) to NaFeSO4F (191.73 A3), is most likely nested in
the larger ionic radii of Naþ as compared to Liþ.
at first attribute such failure to our solid-state synthesis process,
but it is not so since no noticeable differences were found with
respect to the extent of the obtained solid solution when the
samples were made by ionothermal synthesis. Such a limited
solubility existence for the end-member phases prepared at
300 °C is reflective of what does occur at room temperature as
well, since we could not succeed in electrochemically removing
at room temperature more than 0.2 Na from NaFeSO4F and
replacing it by the equivalent amount of Li. More so, the
limited tolerance of the triclinic (P1) structure to sodium is
reflected by our inability to reversibly insert more than 0.2 Na
ions in our fully delithiated FeSO4F structure.
Although all of these structures preserve the same sequence
of octahedral (FeO4F2) and tetrahedral (SO4) arrangements,
they differ by subtle reorientations of these polyhedra with
respect to each other as well as by the position of the alkali
atom which is disordered over two different sites occupied at
50% for Li as compared to a single site for Na. From studies
of the alkali [(Na1-xLix)MSO4F] and 3d-metal [Na(Fe1-x
-
Mx)SO4F] substituted series, we show the inability to sub-
stitute any Li for Na in the end member LiMSO4F (M = Fe,
Co, Ni) phase showing its resilience against sodium substitu-
tion while solely 0.3 Na can be replaced by 0.3 Li in the end
solution member NaMSO4F phases. As most of these phases
are not stable for temperatures greater than ≈325 °C, we could
In contrast, our ability to prepare the full (Na1-xLix)Mn-
SO4F solid solution is not a surprise as both end members
(LiMnSO4F and NaMnSO4F) have a monoclinic structure;

7412 Inorganic Chemistry, Vol. 49, No. 16, 2010
Barpanda et al.
could be more mobile than Liþ could be nested, like for beta-
alumina, in a better fit of Naþ cations to the size of the
galleries. We must on the other hand make certain that the
other singly charged ion, F-, is not the charge carrier.
Fluoride is appreciably mobile in fluorite-type compounds,
especially with soft Pb2þ or Sn2þ, but the more polarizing
transition metals are unlikely candidates, independently of
the favorable tunnel structure. NMR experiments are under-
way with the Zn or Mg equivalents, as the paramagnetic
d-cation screen the signal and will be reported within the
scope of a systematic investigation of ionic conductivity in
theses new host structures.
Prior to carrying on the above track, it should first be
recalled that ion size consideration does not universally rule
ionic conductivity, with the best example against it being the
highest conductivity in Na-β alumina than either Li or K
counterparts. Besides, high ionic conductivity and limited
insertion properties are not incompatible within a poorly
electronic conducting electrode material, as redox reactions
require both electronic and ionic conducting properties.
Regarding insertion reactions, they can enlist either solid
solutions or two-phase reactions depending upon whether
the fully lithiated or delithiated phases have the same struc-
ture. Interestingly, the monoclinic NaFeSO4F phase has a
volume of 191.73 A3 as compared to 163.64 A3 for the
triclinic (P1) delithiated phase. Therefore, the removal of
sodium implies a huge volume contraction (≈16%) asso-
ciated to a phase change, which is energetically unfeasible,
and this is the reason why such a Na deintercalation process
does not proceed. To this end, existing computer modeling
methods, such as quantum chemical calculations using DFT
procedures, are intensively needed to provide a rationale for
the stability of these phases and for their redox reactions and
transport features. Undoubtedly, the richness and complex-
ity of the structural behavior of these fluorosulphates will
challenge DFT promoters.
Figure 10. (a) Electrochemical voltage profile of NaFeSO4F cycled in a
sodium cell with metallic Na anode and 1 M Na-TFSI:PC electrolyte at a
rate of C/20. The NaFeSO4F was prepared by the solid-state route, and
the electrode was made by ball-milling NaFeSO4F and SP carbon black
on80:20 ratio for 10 min. (b)Electrochemical voltageprofileofchemically
delithiated LiFeSO4F powders cycled in a sodium cell with metallic Na
anode and 1 M Na-TFSI: PC electrolyte at a rate of C/20.
Conclusions
The synthesis of various sodium metal fluorosulphate
compounds is reported for the first time. Using both low
temperature solid-state synthesis and ionothermal synthesis,
different end members (NaMSO4F, M=Fe, Co, Ni, Mn)
were prepared at a moderate temperature of 300 °C, and their
monoclinic structures were solved. Owing to the isostructural
nature, various 3d transition metals (Co, Ni and Mn) could
neither is the existence of a complete solid solution among the
3d-metal [Na(Fe1-xMx)SO4F] substituted series, as the end
members NaFeSO4F and NaMSO4F (M = Co, and Ni)
adopt the same monoclinic structure. We presently view
our 60% limited Mn solubility in Na(Fe1-xMnx)SO4F series
as a strong indication of a different space group for the two
end members monoclinic unit cells. Synchrotron experiments
are presently being planned to clarify this point.
be substituted to the Fe site to form series of Na(Fe1-x
-
Mx)SO4F (M = Co, Ni, Mn) phases. All these sodium-based
phases were found to be monoclinic, unlike their lithium
counterparts. These phases were found to be structurally
sensitive; thus, we were unable to formulate a complete series
of solid-solution (Na1-xLix)MSO4F phases. With open tun-
nel structures, the NaFeSO4F compounds were found to
Turning to the transport properties, we found NaFeSO4F
to show a drastically higher ionic conductivity (≈7.14 ꢀ 10-7
S cm-1) than its Li counterpart (≈7 ꢀ 10-11 S cm-1 for
3
3
LiFeSO4F). This comes as a surprise from structural con-
siderations as (i) the sodium ions are less disordered than the
lithium ones in the structure and (ii) the 3D-channels dis-
played by the NaFeSO4F phase for sodium migration are less
open than those displayed by the LiFeSO4F, namely, along
the [111] direction; both of these observations lead to the
expectation of a lower ionic conductivity for the Na phase.
This is an expectation initially supported by our inability to
exchange sodium for Li into the Na-based phase or even to
remove Na from these compounds regardless of the oxidant
we have tried. Thus, such a counter intuitive greater ionic
conductivity measured for the Na-phase as compared to the
Li-one is worth of further consideration. The fact that Naþ
have excellent ionic conductivity (∼7.14 ꢀ 10-7 S cm-1),
3
which is impressive for solid-state electrochemical appli-
cations. However, we could extract only 0.7 Na from
this structure electrochemically involving a 3.6 V FeII/
FeIII redox plateau. We hope that this paper, which high-
lights the richness of the fluorosulphate crystal chemistry
together with their attractive ion transport properties, will
call for intense theoretical calculations using DFT proce-
dures or others so as to boost our research for better ionic
conductors.

Article
Acknowledgment. The authors would like to thank
Inorganic Chemistry, Vol. 49, No. 16, 2010 7413
equipment at the University of California Santa Barbara
ꢁ
M. Courty, M. Reynaud, K. Djellab, W. Walker,
C. Masquelier, E. Baudrin, and J.-B. Leriche for their tech-
nical assistance and the ALISTORE-ERI nanopositive
team members for fruitful discussions. J-M.T. is grateful
to F. Wudl’s group for hosting him and sharing their
laboratory synthesis facilities as well as Materials
Research Laboratory (MRL) for use of their XRD
(UCSB). The authors further acknowledge Ms. Michele
Nelson for proof-reading and improving the linguistic
aspects of the article.
Supporting Information Available: Further details are given in
Figures S1-S8 and Table S1. This material is available free of
charge via the Internet at http://pubs.acs.org.