Fluorination of anatase TiO2 towards titanium oxyfluoride TiOF2: A novel synthesis approach and proof of the Li-insertion mechanism

Article abstract of DOI:10.1039/c4ta02553a

The reactivity of pure molecular fluorine F₂ allows the creation of new materials with unique electrochemical properties. We demonstrate that titanium oxyfluoride TiOF₂ can be obtained under molecular fluorine from anatase titanium oxide TiO₂, while the fluorination of rutile TiO₂ leads only to pure fluoride form TiF₄. Contrary to most fluorides, TiOF₂ is air-stable and hydrolyses poorly under humid conditions. Such a stability makes it possible for TiOF₂ to be studied as an electrode material in Li-ion secondary battery systems. It shows the capacity as high as 220 mA h g^-1 and good cyclability at high current rates at an average potential of 2.3 V vs. Li⁺/Li. At such a potential, only Li⁺ insertion occurs, as proven by in operando XRD/electrochemistry experiments. the Partner Organisations 2014.

Full text of DOI:10.1039/c4ta02553a

Volume 2 Number 37 7 October 2014 Pages 15211–15648
Journal of
Materials Chemistry A
Materials for energy and sustainability
www.rsc.org/MaterialsA
ISSN 2050-7488
PAPER
N. Louvain et al.
Fluorination of anatase TiO2 towards titanium oxyfluoride TiOF2: a novel
synthesis approach and proof of the Li-insertion mechanism

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
PAPER
Fluorination of anatase TiO2 towards titanium
oxyfluoride TiOF2: a novel synthesis approach and
proof of the Li-insertion mechanism†
Cite this: J. Mater. Chem. A, 2014, 2,
15308
N. Louvain,‡ ab Z. Karkar,ab M. El-Ghozzi,ab P. Bonnet,ab K. Guerinab and P. Willmannc
*
´
The reactivity of pure molecular fluorine F2 allows the creation of new materials with unique
electrochemical properties. We demonstrate that titanium oxyfluoride TiOF2 can be obtained under
molecular fluorine from anatase titanium oxide TiO2, while the fluorination of rutile TiO2 leads only to
pure fluoride form TiF4. Contrary to most fluorides, TiOF2 is air-stable and hydrolyses poorly under
humid conditions. Such a stability makes it possible for TiOF2 to be studied as an electrode material in
Li-ion secondary battery systems. It shows the capacity as high as 220 mA h gꢀ1 and good cyclability at
high current rates at an average potential of 2.3 V vs. Li+/Li. At such a potential, only Li+ insertion occurs,
as proven by in operando XRD/electrochemistry experiments.
Received 21st May 2014
Accepted 15th July 2014
DOI: 10.1039/c4ta02553a
www.rsc.org/MaterialsA
anatase phase. However, capacities of anatase are limited by a
Introduction
phase transition occurring at x ¼ 0.5 that greatly reduces
lithium diffusion, while no such transition occurs in TiO2(B),
and nanostructuration has recently emerged as a way to
improve their capacities.6 Some recent reports claim that uo-
rination of TiO2 is another process that could improve its
electrochemical properties,7–10 as observed for other metal oxide
electrode materials.11–17
The need for new energy materials has become essential for the
development of lithium-ion batteries. Both cathode and anode
materials are the purpose of numerous research studies all over
the world.1 Titanium oxides are attractive anode energy mate-
rials owing to their versatile redox chemistry, relative abun-
dance, and nontoxic nature, and, worth mentioning, they are
industrially produced on a wide scale, up to 5 million metric
tons worldwide in 2010,2 as they found many applications,
including pigments, sunscreen and UV-absorbers, photo-
catalysis and photovoltaics.3,4 In theory, they are able to deliver a
capacity of 1342.5 mA h gꢀ1 upon complete reduction of the
metal, and 335.6 mA h gꢀ1 when only one lithium ion is
considered.
More widely, uorination of titanium dioxide is poorly
documented. Using pure molecular uorine, the uorination of
a stable phase containing Ti4+ should lead to the highest and
most stable oxidation state for a titanium (+IV), which is also the
highest uorine content per atom, i.e. TiF4.18,19 TiF4 is a solid at
room temperature and a gas for a temperature higher than
284 ꢁC.20 It is unstable under air and hydrolyses quickly with
humidity. However, under some special uorination conditions
such as those used for surface uorination, an oxyuoride
TiOF2 has been listed.21 Such TiOF2 is a metastable phase whose
thermodynamic data and synthesis conditions are badly
mastered. Titanium oxyuoride TiOF2 is traditionally obtained
as a product of hydrolysis of titanium tetrauoride. It is also
prepared by the hydrolysis of titanium triuorochloride, and
from the reaction of aqueous or anhydrous hydrogen uoride
with titanium dioxide.22–25 It crystallizes with a cubic ReO3-type
Among different existing TiO2 structural types (anatase,
rutile, brookite, and TiO2(B)), both anatase and TiO2(B) present
interesting properties with capacities of about 0.5 and 0.8
lithium ion in LixTiO2 at 1.7 and 1.6 V vs. Li+/Li, respectively.5,6
Most of the electrochemical articles on TiO2 deal with the
aClermont Universite, Universite Blaise Pascal, Institut de Chimie de
Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France. E-mail: nicolas.
´
´
louvain@um2.fr; Fax: +33 4 67 14 33 04; Tel: +33 4 67 14 33 09
structure24 and it is a solid up to 450 C, stable under air.26
ꢁ
b
`
CNRS, UMR 6296, Institut de Chimie de Clermont-Ferrand, F-63177 Aubiere, France
Titanium uorides or oxyuorides have not yet been thor-
oughly investigated as electrode materials because of the
instability of TiF4 under air and the low-related electrochemical
performances of TiOF2. Indeed a study by Chowdari et al.
showed that the TiOF2 structure was destroyed during the rst
discharge at voltage lower than 1 V in TiOF2/Li half-cells.27 They
proposed that the active material aerwards, which led to
capacities as high as 400 mA h gꢀ1 (i.e. less than 2Li+) between
cCentre National d’Etudes Spatiales, Toulouse, France
† Electronic supplementary information (ESI) available: Complementary data
regarding the synthesis conditions; crystallite size calculations; EDX spectra;
complementary TG analyses; Raman spectra; galvanostatic proles of Li/TiO2
half-cells; ex situ XRD/electrochemistry results, detailed in operando
XRD/electrochemistry patterns; prole matching renement of fully discharged
Li/TiOF2. See DOI: 10.1039/c4ta02553a
´
‡ Present address: Institut Charles Gerhardt UMR CNRS 5253 (AIME), Universite
Montpellier 2, CC1502, place E. Bataillon, 34095 Montpellier Cedex 5, France.
15308 | J. Mater. Chem. A, 2014, 2, 15308–15315
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
3.0 and 0.005 V, was in fact amorphous LixTiOy embedded in an simulated by taking into account one mole of TiO2 at 1 atm. The
a priori inactive LiF matrix, where titanium cannot be oxidized system is closed and the Gibbs energy is minimized. The
back to its initial +IV state, and this led to a poor cyclability with simulation is run on a database that contains most thermody-
no plateau.27 These results can be brought closer to those namically stable phases, from which certain metastable phases
obtained for electrode materials made of TiF3, where the may be absent.
formation of an amorphous LixTiFy/LiF composite could display
capacities of 600 mA h gꢀ1 with voltage proles similar to the
previous LixTiOy/LiF composite.28,29 Some authors have also
Titanium oxide uorination
reported that amorphous composites of LixTiOy/TiF3/LiF could Anatase and rutile TiO2 phases have been uorinated under a
show capacities up to 400 mA h gꢀ1 but with the lack of mixture of pure molecular uorine and nitrogen gas at 25 and
reversibility because of a too low potential cutoff.10 It seems 250 ꢁC. The direct F2-gas uorination process was performed in
clear that for further electrochemical investigation, some effort dedicated uorine equipment using special handling proce-
must be made in order to work in an electrochemical window of dures. A cylindrical nickel reactor of approx. 50 cm in length
3–1 V vs. Li+/Li to maintain the initial structure. Even though and offering a volume of approx. 1 litre was used for the reac-
reversible chemical lithiation of TiOF2 seems feasible up to 0.5 tion. The heating was ensured using a horizontal tubular oven
lithium ion per formula unit,30 no report of electrochemical Li- surrounding the reactor, with a domain of homogeneous
(de)insertion is to be found yet.
temperature of approx. 10 cm. All samples were set in a
Going further on getting a pure TiOF2 structure and working passivated nickel boat. Approximately 1 g (12.52 mmol) of tita-
on the electrochemical process that does not alter its pristine nium oxide is placed on a passivated nickel boat, and the
structure (insertion-like mechanism) appears as a challenging powder surface exposed to the reactive atmosphere is manually
objective in order to clearly determine the potentiality of TiOF2. limited to 8–10 cm2, which induced a thickness of the powdered
The most practical uorination technique to get pure TiOF2 is deposit of a few millimetres. The boat was placed at the centre
gas–solid uorination with pure molecular uorine.18,31 The main of the uorination oven, which was rst evacuated in order to
advantage of such a uorination process is the absence of solvent: reach a relative pressure of about ꢀ960 mbar. This step ensures
the reactant can be thermally activated under vacuum in order to the removal of adsorbed water molecules, which may react with
avoid any residual pollutant on the crystal surface that could F2 to form HF that may act as a catalyst, thus resulting in
interact with F2 gas prior to the synthesis and therefore lead to inhomogeneous uorination. Then nitrogen and uorine gases
impurities. Indeed surface pollution of inorganic oxyuorides were slowly and successively introduced up to a relative pres-
synthesized through aqueous media is a known and recurrent sure about 0 mbar (i.e. 1 atm) with approximately 450 mbar (13
problem.32 Here it is demonstrated that TiOF2 can be successfully mmol) of F2 gas. Then, the temperature is increased to 250 C
ꢁ
synthesized by gas–solid uorination of anatase TiO2 under pure with a heating ramp of approximately 2 ꢁC minꢀ1 (in the case of
molecular uorine F2. By optimizing synthesis conditions in room temperature uorination, the temperature is set to 25 ꢁC).
terms of phases and nanostructuration, it is possible to maximize Aer a desired amount of time (4 to 18 h), the system cools
the reaction yield. Then, the obtained materials have been fully down to room temperature, and the reactor oven is ushed with
characterized by XRD, TGA, SEM and Raman. Their electro- nitrogen gas for 1 h.
chemical properties as an electrode in secondary lithium batteries
have been determined at different current densities and, focusing
on an insertion process, this electrochemical mechanism has
Solid-state characterization
been studied through in operando XRD measurements.
X-ray powder diffraction (XRPD). XRPD experiments were
carried out on a PANalytical X-Pert Pro diffractometer equipped
with a diffracted beam monochromator Cu Ka source and an
X'celerator linear detector by using an environment-controlled
cell that allows the data to be collected under a static inert argon
atmosphere. The counting time was 200 s and the angle step
size was 0.0836ꢁ.
Experimental
Materials and methods
Anatase (nanopowder, <25 nm particle size, 99.7%, specic
surface area 45–55 m2 gꢀ1) and rutile (micropowder $99%) TiO2
phases were purchased from Sigma-Aldrich, and anatase TiO2
(micropowder $99%) from Fluka. All powders were used
without further purication. Pure molecular uorine F2 gas (98–
99%, 27 bars) was purchased from Solvay Chemicals, Solvay
Fluor GmbH, Germany. The uorinated products were stored
and handled in an argon-lled glovebox with a humidity level
below 0.05 ppm.
Thermogravimetry analysis. The thermal behaviour of each
precursor and product was investigated by thermogravimetric
analysis under inert argon gas on a Shimadzu TGA-50 ther-
mogravimetric analyser instrument from 25 to 600 ꢁC with a
ꢀ1
ꢁ
heating ramp of 5 C min
.
Scanning electron microscopy (SEM). SEM was carried out
on a SUPRA 55VP ZEISS eld-effect scanning electron micro-
scope equipped with a Gemini® column and high efficiency in-
lens secondary electron detector. The instrument is tted with
an X-MAX Oxford detector for analytical energy-dispersive X-ray
Thermodynamic simulations
Calculations have been run through the Fact-Web interface of spectroscopy (using the INCA 350 soware). EDX analysis was
the FactSage package. The uorination system has been performed at 10 kV acceleration on a wide portion of the sample
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 15308–15315 | 15309

View Article Online
Journal of Materials Chemistry A
Paper
at a 500ꢂ zoom. All samples were coated with graphite prior to using an environment-controlled cell described in Fig. S1.† The
the experiment.
counting time was 500 s and the angle step size 0.0668ꢁ. The
Raman spectroscopy. The samples were characterized in measurements were programed and monitored from both XRD
solid form by Raman spectroscopy at 514.5 nm (2.41 eV) with a and electrochemistry points of view by their respective soware.
Jobin Yvon T64000 Raman spectrometer.
The galvanostatic cycle was set as in a usual measurement,
while XRD patterns were constantly collected. In order to closely
follow the peak position displacement, the (001) peak of the
TiOF2 phase was monitored every 15 min for 2 h. Aer 2 h, a
Galvanostatic charge–discharge measurements
The electrochemical performances of our materials were investi- wide-range XRD pattern was collected between 10 and 50ꢁ 2q for
gated using galvanostatic discharge–charge cycles. The electrodes 40 min. This ‘8 ꢂ 15 min + 40 min’ scheme was repeated during
were composed of the active material (about 72% by weight, w/w), the relaxation and the rst discharge–charge cycle.
acetylene black (18% w/w), to ensure the electronic conductivity
and polyvinylidene diuoride (PVDF, 10% w/w) as a binder. Aer
stirring in propylene carbonate (PC), the mixture was spread
Results and discussion
uniformly onto a metallic current collector disk of 12 mm in
According to thermodynamics, the reaction of molecular uo-
diameter (stainless steel for Swagelok cell type, and Al foil for coin
rine F2 with rutile titanium dioxide TiO2 at 25 ꢁC should lead to
cell type). Aer the PC evaporation, the electrodes were dried in a
an equimolar mixture of TiF4 and O2:
vacuum oven at 80 ꢁC overnight to remove traces of adsorbed
water and solvent before being transferred into an argon-lled
TiO2(s) + 2F2(g) ¼ TiF4(s) + O2(g)
(1)
glovebox. The anode was a lithium metal disk, and the separator
was Celgard 2034. A two-electrode cell was used (Swagelok cell
type, or coin cell type for cyclability measurements) where lithium
was both reference and counter electrodes. The C rate calcula-
tions were based on the theoretical capacity (263.15 mA gꢀ1 for
TiOF2 and 335.6 mA gꢀ1 for TiO2). The electrolyte was 1 M LiPF6
in propylene carbonate–ethylene carbonate–dimethyl carbonate
(PC : EC : 3DMC, 1 : 1 : 3 vol%). The cells were assembled in
an argon-lled glovebox. Relaxation was performed for at
least 5 hours until the open circuit voltage (OCV) stabilization.
Galvanostatic discharge–charge cycles were carried out on a
VMP2-Z instrument from BioLogic and were performed at room
temperature by applying a constant current in the working region
1.2–3.8 V. Current densities ranging from C/100 to C were used.
As a plateau is clearly observed on each galvanostatic curve, the
(dis)charge potential is measured at half of the total (dis)charge
From room temperature, the reaction is supposedly quanti-
tative, with solid TiF4 formed (Fig. 1). When the temperature
increases, TiF4 starts to sublimate from approximately 200 C,
ꢁ
and the presence of gaseous TiF4 becomes non-negligible above
250 ꢁC, which is lower than expected (Tsub ¼ 284 ꢁC).20 The
major difficulty is that only TiF4 is the only stable phase
reported and no other, hypothetical or not, metastable uoride
or oxyuoride phases can be included in the calculation.
Moreover the difference of reactivity between anatase and rutile
under molecular uorine has never been analysed. In order to
investigate the reactivity of titanium dioxide, anatase and rutile
TiO2 powders were uorinated with a mixture of molecular
uorine and nitrogen gases at 250 ꢁC for 18 hours in dedicated
uorination equipment.
Fig. 2 displays the results of the in situ pressure data
collected during the static uorination process. First, let us
consider the case of micrometre-sized powders of rutile and
anatase TiO2. When rutile is heated up to 250 ꢁC in one atm of a
F2/N2 mixture, the pressure reachesꢁabout 1350 mbar. Then,
when the temperature is kept at 250 C, the pressure keeps on
capacity and is noted E1/2
.
Ex situ XRD/electrochemistry measurements. The samples
studied here were taken from partially or totally discharged
Swagelok cells. The cells were cycled at C/10 and prepared
according to the procedure described in the article. The
discharge was programed to stop at a desired potential (2.5, 2.0
or 1.2 V vs. Li+/Li), and the cells were opened in an argon-lled
glovebox. The active material was dried overnight and measured
in an environment-controlled cell.
In operando XRD/electrochemistry measurements. The
active material preparation for the in operando XRD/electro-
chemistry measurements is described in the aforementioned
procedure, except that the mixture was uniformly spread onto a
glass side with a 14 mm diameter. Aer a vacuum-drying
process, the electrode was removed from the glass slide with a
scalpel, and placed in an environment-controlled cell. The
current collector was composed of an iron–nickel alloy disk with
a 12 mm diameter aperture. Celgard 2034 and Whatman glass
bre membranes separate the electrode from the lithium
disk. The TiOF2/Li cell was cycled at C/10 for the discharge and
at C/20 for the charge (to ensure complete delithiation). In
Fig. 1 Thermodynamic calculations for various temperatures of a
operando XRD/electrochemistry experiments were carried out by TiO2 : F2 ratio of 1 : 2.
15310 | J. Mater. Chem. A, 2014, 2, 15308–15315
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 2 In situ total pressure data collected during static fluorination of Fig. 3 XRD patterns of rutile and anatase TiO2 precursors and rutile
rutile and anatase TiO2 with 1 atm of a mixture of N2 and F2 gases at and anatase TiO2 fluorinated at 250 ꢁC for 18 h under static conditions.
250 ꢁC; the inset shows a zoom on the isotherm part of the curves.
Characteristic peaks coming from anatase, rutile and cubic TiOF2 are
noted.
slowly increasing. This indicates the formation of a gaseous
phase of TiF4 because of its high tensile strength at such a
temperature, in good accordance with thermodynamics (Fig. 1).
In contrast, when anatase is uorinated under similar condi-
tions (Fig. 2), the pressure prole obtained is clearly dissimilar,
suggesting a different reactivity with pure uorine between both
polymorphs, contrary to what was expected. At 250 ꢁC, the
pressure drastically decreases from 1350 to 1200 mbar, indi-
cating the formation of a new solid phase with a net
consumption of F2 gas. With all parameters being kept iden-
tical, the only reason for such a difference is polymorphism.33
The rst interface encountered by F2 gas is the crystal surface,
thus crystal structure and morphology may affect the reactivity
during a gas–solid process.34 Indeed, such a difference between
rutile and anatase is oen discussed in adsorption of active
species in photocatalysis.
The XRD patterns of micrometre-sized anatase and rutile
TiO2 uorinated at 250 ꢁC for 18 h are shown in Fig. 3. For
uorinated rutile, the XRD signature is characteristic of the
rutile phase. This is in good agreement with our conclusions
from both the in situ pressure data and thermodynamic
assumptions: rutile titanium dioxide partially reacts with
molecular uorine to form gaseous titanium tetrauoride TiF4,
leaving only unreacted rutile TiO2 at the end of the experiment
(Fig. 3b). Crystallization of a TiF4-derived compound, namely
NiTiF6, in the cold end of the uorination oven, where it is very
likely that gaseous TiF4 may have condensed and subsequently
reacted with nickel uoride NiF2 from the passivation layer of
the reactor, is indirect proof of the formation of gaseous TiF4
during the uorination of rutile TiO2 (Fig. S2†). Although
equimolar quantities of reactants are used, the reaction is not
completed aer 18 h. Etching of rutile with F2 might be kinet-
ically limited.
hkl planes, respectively (ICSD #160661). Nevertheless peaks
from the second minor phase can be identied at 2q ¼ 25.3ꢁ,
corresponding to the (101) hkl plane of anatase (ICSD #9855).
This reveals that residues of unreacted anatase are still present
in the uorinated powder.
With the possibility to isolate TiOF2 from the anatase TiO2
phase being established, the crystallite size inuence on the
reactivity can be briey described. For such purposes, a nano-
powder of anatase TiO2 was uorinated, as it is known that
nanostructuration of uorides is highly desirable for electro-
chemical applications.35 When the nanosized powder of anatase
ꢁ
is heated up to 250 C under uorine, the pressure reaches a
maximum of 1180 mbar before lessening towards 1100 mbar
(Fig. 2), that is 100 mbar lower than its microsized parent,
indicating both the formation of a new solid phase with a net
consumption of F2 gas and an increased yield. It is noteworthy
that the reactivity of nanosized anatase with uorine is
enhanced, as can be seen from Fig. 2. While microsized
powders of TiO2 do not display any unforeseen reactivity before
reaching the target temperature of 250 ꢁC, the pressure for
nanosized anatase slows down its rise ꢁaround 175 ꢁC and
ꢁ
climbs to its maximum at 220 C, i.e. 30 C lower than for the
micrometric powders (Fig. S3†). As for microsized anatase TiO2,
uorination of nano-anatase leads to TiOF2 (Fig. 3f). The
resulting powder does not display any peak from its parent
anatase phase, indicating that the reaction came close to
completion. Interestingly, the crystal size did not suffer from
the uorination conditions, and the initial size of anatase
crystals (25 nm) is kept nearly constant aer synthesis, and the
size of nanocrystals of TiOF2 is approximately 30 nm (Table
S1†). Fig. 4 shows SEM images of nano-anatase before and aer
uorination. The pristine powder is composed of sponge-like
aggregates, with various sizes larger than 1 mm, made up of self-
assembled nanoparticles, and the powder morphology does not
seem to have evolved during uorination.
When anatase is uorinated under the same conditions, the
resulting powder is unambiguously composed of TiOF2
(Fig. 3d), as shown by the intense peaks at 2q ¼ 23.4ꢁ, 33.3ꢁ,
47.8ꢁ and 53.9ꢁ, corresponding to (001), (011), (002) and (012)
Thermogravimetric analyses of pristine and uorinated
titanium dioxide powders are displayed in Fig. 5. While neither
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 15308–15315 | 15311

View Article Online
Journal of Materials Chemistry A
Paper
TiOF2 (ꢀ7.6%); the presence of TiOF2 in uorinated rutile was
not evidenced from XRD and may stem from the anatase
impurity in the pristine rutile TiO2 powder (Fig. 3a). EDX
measurements clearly identied uorine in uorinated rutile
powder (Fig. S4†). Fluorinated micrometric anatase is mainly
composed of TiOF2 (Fig. 3d), and it decomposes in a single step
at 538 ꢁC (ꢀ35.6%), leading to anatase TiO2 (Fig. S5†). The
supposedly higher yield of uorinated nano-anatase, compared
to micro-anatase, is positively demonstrated by the single-step
decomposition at 525 ꢁC with an increased weight loss of
45.5%, a result of which is in good accordance with in situ
pressure data and XRD measurements.
It is commonly accepted that TiOF2 decomposes into equi-
molar amounts of TiO2 and TiF4, which would give rise to a 60%
weight loss.26 Nonetheless such a value has never been reported,
and most of the recent sol–gel derived TiOF2 crystals present
weight losses of approximately 30 to 45%.9,26 The fact that our
TiOF2 nanocrystals do not show a 60% weight loss, even though
their purity is apparently high, may arise from either incom-
plete uorination or a wrong, therefore more complicated,
decomposition equation. If the rst assumption is considered
true, then the Raman spectrum of TiOF2 nanocrystals should
reveal the presence of an anatase impurity, which cannot be
pinpointed (Fig. S6†). Consequently, the second assumption
must be the right one, and the decomposition pathway of TiOF2
crystals may be more complex than expected.
The XRD measurement aer thermal treatment shows only
the presence of anatase TiO2 (Fig. S5a†). Placing a uorine
scavenger such as Li2CO3 along with TiOF2 during a thermal
treatment up to 600 ꢁC must lead to a 24% weight loss and to a
LiF–TiO2 mixture in the following proportions:
Fig.
4 SEM images of anatase (a) and fluorinated anatase (b)
nanopowders.
anatase nor rutile TiO2 shows any signicant loss up to 600 ꢁC,
uorinated rutile decomposes in two steps at 127 and 503 C
2Li2CO3(s) + 5TiOF2(s) ¼ 4LiF(s) + 5TiO2(s) + gaseous products(2)
ꢁ
and displays weight losses of 11.2% and 7.6%, respectively.
According to XRD, uorinated rutile is composed of crystalline
rutile TiO2. The rst step might be correlated with the presence
of the etching product adsorbed on the grain surface,7 while the
second step at 503 ꢁC can be attributed to the decomposition of
Both TG and XRD analyses (Fig. S5b and c†) are in perfect
accordance with eqn (2), which does not involve any gaseous
titanium uoride intermediate products. Further investigation
by coupled techniques with TGA such as mass spectrometry
should be made in order to revisit the decomposition products
of TiOF2.
We can thus condently conclude that, based on the results
from in situ pressure data, XRD and TG analyses, 30 nm nano-
crystals of pure TiOF2 were obtained from a nanopowder of
anatase TiO2.
The as-prepared nanocrystals of TiOF2 were tested as an
electrode of Li-ion batteries in Swagelok-type half-cells. The
galvanostatic charge–discharge voltage proles at three
different current rates (C/10, C/20 and C/100, where C is set at
263.15 mA h gꢀ1 for a 1 ꢀ eꢀ process) of the rst and eighth
cycles for TiOF2 versus Li between 3.8 and 1.2 V are shown in
Fig. 6. Such a potential window was chosen according to the
studies by Chowdari et al.27 showing that under 1.0 V no more
potential plateau occurs in reduction. For each current density,
the discharge curves show a relatively at plateau around 2.25 V
vs. Li+/Li. The TiOF2 nanocrystal electrode delivers a reversible
specic capacity of 145 mA h gꢀ1 at C/10 (Fig. 6a) over the rst
eight cycles. The differences in shape and potential between the
Fig. 5 TG profiles of pristine TiO2 and TiO2 samples fluorinated at
250 ꢁC for 18 h; rutile profile, not shown here, is identical to the
anatase one.
15312 | J. Mater. Chem. A, 2014, 2, 15308–15315
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
Fig. 7 In operando evolution of the XRD pattern recorded at C/20
with a close view on the 001 reflection (left) and its corresponding
voltage profile (right). Miller indices for the cubic (black) and rhom-
bohedral (blue) cells are shown. Peaks between 45ꢁ and 47.5ꢁ originate
from the measurement cell.
Fig. 6 Galvanostatic voltage profiles for TiOF2/Li cells cycled at C/10
(a), C/20 (b), and C/100 (c) between 3.8 and 1.2 V, and a comparison of
their first cycle (d).
rst cycle and the following ones are thus relatively small,
indicating that the reaction mechanism is most probably the
same. At slower rates (Fig. 6b–d), the reversible specic capacity
is increased up to 155 mA h gꢀ1 (C/20) and impressive 220 mA h
gꢀ1 (C/100). The fact that the specic capacity increases when
reducing the current density would indicate a kinetically limited
ionic or electrical diffusion mechanism, which could be
explained either by the insulating nature of TiOF2, a common
characteristic of uorinated materials, or a none convenient
volume/area ratio of the TiOF2 grains. It is noteworthy that the
potential of TiOF2 is 0.6 V higher than the one of anatase TiO2
(Fig. S7 and Table S2†), as expected from the high electroneg-
ativity of the uoride ion.
We endeavoured to study the Li-insertion mechanism into
TiOF2 in order to discriminate between a typical insertion,
yielding a LixTiOF2 phase, and a conversion reaction, yielding
a composite mixture of LiF and LiyTiO2 (the formation of Ti0 in
our working potential window has been ruled out). In the rst
step, an ex situ XRD study has been carried out on electrodes
that were subjected to a C/10 rate and stopped at 2.5, 2.0,
and 1.2 V during their rst discharge (Fig. S8†). The impact of
Li-insertion is relatively small on the XRD prole and no
lithium uoride can be detected, thus a priori indicating that
no conversion reaction may occur in the chosen potential
window.
In order to go further into detail in the reaction mechanism
of TiOF2 with Li, and to probe cell parameter modulation, the
rst charge–discharge cycle was investigated by in operando
XRD and the results are shown in Fig. 7. During the open-
circuit, only the well-dened peaks of crystalline cubic TiOF2
are visible in the diffraction pattern (Fig. 7, patterns A and B) at
2q ¼ 23.4ꢁ, 33.3ꢁ, and 47.8ꢁ, corresponding to (001), (011), and
(002) hkl planes, respectively (ICSD #160661), and no evolu-
tion, either in intensity or in position, can be detected
(Fig. S9†). At the beginning of the reaction (Fig. 7, pattern C),
the intensities of these peaks slightly decrease without
observing any new peak in the diffraction pattern until half
discharge, where Bragg peaks broaden and shi to higher
angles (Fig. 7, pattern D). This is clearly noticeable on the close
view on the 001 reection (Fig. 7 and S10†). Please note that at
this point of discharge a new peak unambiguously appears at
2q ¼ 40.1ꢁ, and that such XRD behaviour is characteristic of Li-
intercalation materials.36 At the end of discharge (Fig. 7,
pattern E), the set of reections of the initial cubic TiOF2 phase
are broadened and shied to higher angles, and their intensity
is slightly lower. The angle shi may point to the cubic-to-
rhombohedral phase transition described for chemically
lithiated TiOF2,30 and the hypothesis has been veried by a
prole matching renement of the fully discharged pattern
using a rhombohedral model (R-3c). All the diffraction features
are well modelled by this phase with cell parameters a ¼ 5.206
˚
˚
A and c ¼ 13.334 A (Fig. S11 and Table S3†), and this allows us
to conclude that the phase at the end of the discharge is a
lithiated TiOF2 of general formula LixTiOF2; thus, the present
peaks at 2q ¼ 23.8ꢁ, 33.4ꢁ, 34.4ꢁ, 40.1ꢁ, and 48.7ꢁ would
correspond to (012), (104), (110), (113), and (024) hkl planes,
respectively. While at the early oxidation process the diffrac-
tion pattern can still be indexed by the rhombohedral cell
(Fig. 7, pattern F), the cubic non-lithiated TiOF2 phase is
fully recovered at the end of charge (Fig. 7, pattern G), there-
fore corroborating our hypothesis for Li-insertion and con-
rming the reversibility of the electrochemical performances
obtained. With a concrete picture of the reactions of lithium
with titanium oxyuoride, it is now possible to deduce the
lithium content, i.e. x in LixTiOF2 formula, from our electro-
chemical measurements. At a C/10 rate, a specic capacity of
145 mA h gꢀ1 was obtained (Fig. 6a) and this leads to a lithi-
ated LixTiOF2 with x ¼ 0.55 (with a theoretical capacity of
263.15 mA h gꢀ1 for 1 lithium); the lithium content can be as
high as x ¼ 0.87 at C/100, for a specic capacity equal to 220
mA h gꢀ1 (Fig. 6c).
With the Li-insertion mechanism being validated, the
cycling performances of our TiOF2 nanocrystals were
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 15308–15315 | 15313

View Article Online
Journal of Materials Chemistry A
Paper
Conclusions
In conclusion, it has been demonstrated that nanocrystals of
TiOF2 can successfully be obtained by gas–solid uorination of
TiO2 anatase nanocrystals, and not from rutile, with molecular
uorine. This material gave us the opportunity to investigate the
reaction of lithium with titanium oxyuoride in TiOF2/Li half-
cells. By taking advantage of the different stable oxidation states
of Ti ions, in particular Ti4+ and Ti3+, it was possible to revers-
ibly intercalate up to 0.87 Li+ into TiOF2, thus obtaining Li0.87
-
TiOF2. Using in operando XRD/electrochemistry, it has been
established that TiOF2 can act as a typical Li-insertion material
with an average potential of 2.3 V vs. Li+/Li.
Further work will be devoted to the unravelling of the
decomposition pathway of TiOF2 by TGA coupled with mass
spectrometry, as well as improvements of the electrochemical
performances in order to reach the theoretical capacity of
263.15 mA h gꢀ1, by focusing on further nanotexturation and
electrode formulation. In the light of such knowledge regarding
the Li-insertion mechanism of TiOF2, its electrochemical
performances below 1 V may now be re-investigated.
Acknowledgements
´
The authors would like to gratefully thank Profs Andre Hamwi
and Daniel Avignant for fruitful discussions on oxyuorides,
Nicolas Batisse and Lawrence Frezet for the automatized uo-
¨
rination equipment, Joel Cellier for the development of the in
Fig. 8 Composition–voltage profiles for one TiOF2/Li cell cycled at C/
20, C/10, C/20, C and C/20 consecutively between 3.8 and 1.2 V (a);
cycling performance curves of TiOF2 nanocrystals at current densities
of C and C/100 (b).
operando XRD/electrochemistry cell, and Claire Fonquernie for
her global help on all our measurements.
Notes and references
investigated (Fig. 8). The galvanostatic charge–discharge
composition–voltage proles of key cycles of a TiOF2/Li cell
consecutively cycled at C/20, C/10, C/20, C, and nally C/20
between 3.8 and 1.2 V are displayed in Fig. 8a. The most
striking feature of this material is that the specic capacity
obtained at C/20 is systematically resumed aer being cycled
at higher C/10 and C rates. In addition to its excellent cyclic
capacity retention at sequential current rates, TiOF2/Li cells
demonstrate a stable specic capacity at the high current
density of C for more than 30 charge–discharge cycles between
3.0 and 1.2 V (Fig. 8b). The discharge capacity slowly increases
during the rst thirty cycles and reaches 60 mA h gꢀ1. This
relatively high capacity retention at C may stem from the
presence of O2ꢀ ions in TiOF2, which makes it a semi-
conductor with a band-gap of approximately 2.5 eV,37 whereas
metal uorides are commonly referred to as insulators with
band-gaps larger than 4 eV (ref. 38–40) that cannot be cycled at
such a high density current without nanocarbon graing.41–43
An impressive capacity of 200 mA h gꢀ1, close to the theoretical
one, can even be obtained when the half-cell is cycled at C/100
(Fig. 8b). The system retains a high capacity of 180 mA h gꢀ1
aer tens of cycles, a value that can surely be improved by
optimizing the battery engineering.
1 P. G. Bruce, B. Scrosati and J. M. Tarascon, Angew. Chem., Int.
Ed., 2008, 47, 2930–2946.
2 TDMA, Cec – The European Chemical Industry Council,
2010.
3 G. Bedinger, in US Geological Survey – Mineral Commodity
Summaries, 2013, pp. 172–173.
4 X. Chen and S. Mao, Chem. Rev., 2007, 107, 2891–2959.
5 J.-Y. Shin, D. Samuelis and J. Maier, Adv. Funct. Mater., 2011,
21, 3464–3472.
6 D. Deng, M. G. Kim, J. Y. Lee and J. Cho, Energy Environ. Sci.,
2009, 2, 818–837.
7 M. Saito, Y. Nakano, M. Takagi, T. Maekawa, A. Tasaka,
M. Inaba, H. Takebayashi and Y. Shodai, Key Eng. Mater.,
2014, 582, 127–130.
8 Y. Zeng, W. Zhang, C. Xu, N. Xiao, Y. Huang, D. Y. Yu,
H. H. Hng and Q. Yan, Chem.–Eur. J., 2012, 18, 4026–4030.
9 L. Chen, L. Shen, P. Nie, X. Zhang and H. Li, Electrochim.
Acta, 2012, 62, 408–415.
10 D. Dambournet, K. Chapman, P. Chupas, R. Gerald,
N. Penin, C. Labrugere, A. Demourgues, A. Tressaud and
K. Amine, J. Am. Chem. Soc., 2011, 133, 13240–13243.
11 Z. G. Wang, Z. X. Wang, H. J. Guo, W. J. Peng, X. H. Li and
J. X. Wang, Mater. Lett., 2014, 123, 93–96.
15314 | J. Mater. Chem. A, 2014, 2, 15308–15315
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry A
12 K. M. Wiaderek, O. J. Borkiewicz, E. Castillo-Martinez, 28 H. Li, P. Balaya and J. Maier, J. Electrochem. Soc., 2004, 151,
R. Robert, N. Pereira, G. G. Amatucci, C. P. Grey, A1878–A1885.
P. J. Chupas and K. W. Chapman, J. Am. Chem. Soc., 2013, 29 H. Li, G. Richter and J. Maier, Adv. Mater., 2003, 15, 736–
135, 4070–4078. 739.
13 Y. Ma, B. Ding, G. Ji and J. Lee, ACS Nano, 2013, 7, 10870– 30 D. W. Murphy, M. Greenblatt, R. J. Cava and S. M. Zahurak,
10878. Solid State Ionics, 1981, 5, 327–330.
14 M. Ati, M. Sougrati, G. Rousse, N. Recham, M. L. Doublet, 31 N. Louvain, J. Peyroux, M. Dubois, W. Simond and F. Leroux,
J. Jumas and J. M. Tarascon, Chem. Mater., 2012, 24, 1472–
Inorg. Chem., 2014, 53, 852–860.
1485.
32 A. P. Wilkinson, R. E. Josefsberg, L. C. Gallington,
C. R. Morelock and C. M. Monaco, J. Solid State Chem.,
2014, 213, 38–42.
15 Q. Luo, T. Muraliganth and A. Manthiram, Solid State Ionics,
2009, 180, 703–707.
16 K. S. Lee, S. T. Myung, K. Amine, H. YashiroB and Y. K. Sun, 33 D. S. Torkhov, P. E. Meskin, Y. V. Kolen'ko, V. A. Ketsko,
J. Mater. Chem., 2009, 19, 1995–2005.
17 I. D. Gocheva, I. Tanaka, T. Doi, S. Okada and J. Yamaki,
Electrochem. Commun., 2009, 11, 1583–1585.
A. A. Burukhin, B. R. Churagulov and N. N. Oleinikov,
Dokl. Chem., 2004, 394, 36–38.
34 S. C. Li and U. Diebold, J. Am. Chem. Soc., 2010, 132, 64–66.
18 N. Louvain, A. Fakhry, P. Bonnet, M. El-Ghozzi, K. Guerin, 35 G. G. Amatucci and N. Pereira, J. Fluorine Chem., 2007, 128,
M. T. Sougrati, J. C. Jumas and P. Willmann,
CrystEngComm, 2013, 15, 3664–3671.
19 P. Hagenmuller, Inorganic Solid Fluorides, Academic Press,
Inc., 1985.
20 G. Brauer and G. Brauer, Handbook of Preparative Inorganic
Chemistry, Academic Press, Inc., 1963.
21 D. Li, H. Haneda, S. Hishita, N. Ohashi and N. Labhsetwar,
J. Fluorine Chem., 2005, 126, 69–77.
243–262.
36 J. N. Reimers and J. R. Dahn, J. Electrochem. Soc., 1992, 139,
2091–2097.
37 C. Xue, T. Narushima and T. Yonezawa, J. Inorg. Organomet.
Polym., 2012, 23, 239–242.
38 Z. Yang, Y. Pei, X. Wang, L. Liu and X. Su, Comput. Theor.
Chem., 2012, 980, 44–48.
39 R. Li, S. Wu, Y. Yang and Z. Zhu, J. Phys. Chem. C, 2010, 114,
16813–16817.
22 J. Moss and A. Wright, J. Fluorine Chem., 1975, 5, 163–167.
23 K. S. Vorres and F. B. Dutton, J. Am. Chem. Soc., 1955, 77, 2019. 40 R. E. Doe, K. A. Persson, Y. S. Meng and G. Ceder, Chem.
24 K. Vorres and J. Donohue, Acta Crystallogr., 1955, 8, 25–26. Mater., 2008, 20, 5274–5283.
25 H. M. Haendler, S. F. Bartram, R. S. Becker, W. J. Bernard 41 R. Ma, Y. Dong, L. Xi, S. Yang, Z. Lu and C. Chung, ACS Appl.
and S. W. Bukata, J. Am. Chem. Soc., 1954, 76, 2177–2178. Mater. Interfaces, 2013, 5, 892–897.
26 S. Xie, X. Han, Q. Kuang, J. Fu, L. Zhang, Z. Xie and L. Zheng, 42 C. Li, L. Gu, J. Tong and J. Maier, ACS Nano, 2011, 5, 2930–
Chem. Commun., 2011, 47, 6722–6724. 2938.
27 M. V. Reddy, S. Madhavi, G. V. Subba Rao and 43 S. W. Kim, D. H. Seo, H. Gwon, J. Kim and K. Kang, Adv.
B. V. R. Chowdari, J. Power Sources, 2006, 162, 1312–1321.
Mater., 2010, 22, 5260–5264.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. A, 2014, 2, 15308–15315 | 15315