High-potential reversible Li deintercalation in a substituted tetrahydroxy-p-benzoquinone dilithium salt: An experimental and theoretical study

Article abstract of DOI:10.1002/chem.201103820

Efficient organic Li-ion batteries require air-stable lithiated organic structures that can reversibly deintercalate Li at sufficiently high potentials. To date, most of the cathode materials reported in the literature are typically synthesized in their f

Full text of DOI:10.1002/chem.201103820

FULL PAPER
DOI: 10.1002/chem.201103820
High-Potential Reversible Li Deintercalation in a Substituted Tetrahydroxy-
p-benzoquinone Dilithium Salt: An Experimental and Theoretical Study
[
a, b]
[a, b]
[a]
[a, b]
Anne-Lise Barrꢀs,
Joaquin Geng,
Gaꢁtan Bonnard, Stꢂven Renault,
[a] [b]
[
a, b]
[c]
Sꢂbastien Gottis,
Olivier Mentrꢂ, Christine Frayret, Franck Dolhem, and
[a]
Philippe Poizot*
Abstract: Efficient organic Li-ion bat-
teries require air-stable lithiated organ-
ic structures that can reversibly deinter-
calate Li at sufficiently high potentials.
To date, most of the cathode materials
reported in the literature are typically
synthesized in their fully oxidized
form, which restricts the operating po-
tential of such materials and requires
use of an anode material in its lithiated
state. Reduced forms of quinonic struc-
tures could represent examples of lithi-
ated organic-based cathodes that can
a simple three-step method, lithiated
3,6-dihydroxy-2,5-dimethoxy-p-benzo-
hydrated form of the studied salt was
also characterized by XRD and first-
principles calculations. Various levels
of theory were probed, including DFT
with classical functionals (LDA, GGA,
PBEsol, revPBE) and models for dis-
persion corrections to DFT. One of the
modified dispersion-corrected DFT
schemes, related to a rescaling of both
quinone,
a new redox amphoteric
system derived from the tetralithium
salt of tetrahydroxy-p-benzoquinone.
Electrochemical investigations revealed
that such an air-stable salt can reversi-
+
bly deintercalate one Li ion on charg-
ing with a practical capacity of about
ꢀ
1
100 mAhg at about 3 V, albeit with
a polarization effect. Better capacity
retention was obtained by simply
adding an adsorbing additive. A tetra-
van der Waals radii and s parameter,
6
provides significant improvements to
the description of this kind of crystal
over other treatments. We then applied
this optimized approach to the screen-
ing of hypothetical frameworks for the
delithiated phases and to search for the
anhydrous structure.
+
deintercalate Li at potentials higher
than 3 V thanks to substituent effects.
Having previously recognized the
unique electrochemical properties of
the C O -type ring, we have now de-
Keywords: density functional calcu-
lations · lithiation · lithium-ion bat-
teries · quinones · solid-state struc-
tures
6
6
signed and then elaborated, through
Introduction
burden and affordable price is a major technological chal-
[1]
lenge to promoting a clean energy economy. Among the
various types of EES devices, the rechargeable battery
The global search for electrical energy storage (EES) sys-
tems displaying high performance with low environmental
[2]
system is a critical energy converter in many respects. Day
and night, in stationary or portable mode, on a large or very
small scale, accumulators can directly convert almost rever-
sibly stored chemical energy to direct current. In spite of in-
tensive research and innovation, no universal accumulator
or universal chemistry meets all of the possible requirements
so far; this means managing existing battery technologies to
best meet the required specifications. Basically, the chemis-
try of electroactive species involved in batteries has not sub-
stantially changed since the invention of the first recharge-
[
a] Dr. A.-L. Barrꢁs, Dr. J. Geng, G. Bonnard, Dr. S. Renault,
Dr. S. Gottis, Dr. C. Frayret, Dr. P. Poizot
Laboratoire de Rꢀactivitꢀ et Chimie des Solides - UMR CNRS 7314
Institut de Chimie de Picardie - FR 3085 CNRS
Universitꢀ de Picardie Jules Verne
3
3, rue Saint-Leu, 80039 Amiens cedex (France)
Fax : (+33)322827590
E-mail: philippe.poizot@u-picardie.fr
[
b] Dr. A.-L. Barrꢁs, Dr. J. Geng, Dr. S. Renault, Dr. S. Gottis,
Dr. F. Dolhem
AHCTUNGTRENNUGNa ble battery by G. Plantꢀ in 1859, and is dominated by
Laboratoire des Glucides - UMR CNRS 6219
Institut de Chimie de Picardie - FR 3085 CNRS
Universitꢀ de Picardie Jules Verne
metal-based electroactive components derived from mineral
resources. However, to meet the ever-increasing power
demand while limiting the environmental footprint, organic-
based batteries could be foreseen as an interesting parallel
research path for the future, since organic compounds have
some advantages such as an abundance of raw materials
3
3, rue Saint-Leu, 80039 Amiens cedex (France)
[
c] Dr. O. Mentrꢀ
Unitꢀ de Catalyse et de Chimie du Solide
ꢂquipe de Chimie du Solide - UMR CNRS 8181
ENSC Lille-UST Lille
Citꢀ Scientifique - Bꢃt. C7 - BP 80108
9655 Villeneuve d’Ascq cedex (France)
[2,3]
(
e.g., use of biomass) and easy recycling.
In principle, redox-active organic structures could be im-
5
plemented in a wide variety of existing battery technologies.
However, in practice, development of efficient organic elec-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201103820.
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

trodes is clearly in its early stages, and much remains to be
done, especially if we expect to transpose existing rocking-
chair technology to organic electrode materials. Promising
results have already been obtained, in particular with poly-
mance coupled with an intriguingly high polarization value
(i.e., charge/discharge potential difference).
Benefiting from our observations on quinonic-type struc-
tures and the efficient and unique electrochemical proper-
[3]
[4]
mer-based electrodes (original concept of organic radical
batteries, ORBs) and more recently with small molecules
ties of the Li C O -type unit, we recently designed two
4
6
6
series of organic salts that can reversibly deintercalate lithi-
um. We first focused on lithiated 3,6-dihydroxy-2,5-dime-
[4]
like lithiated quinonic-type structures. To go further, it
seems that one of the remaining problems is to identify
stable, insoluble, and efficient organic structures that can ac-
commodate lithium at very different Fermi levels in order to
enable a large enough output voltage of the cell. Although
several kinds of organic structures can react in reduction
thoxy-p-benzoquinone (Li DHDMQ), an intermediate
2
structure between the tetralithium salt of tetrahydroxy-p-
benzoquinone (Li THQ) and tetramethoxy-p-benzoquinone
4
(TMQ), which has not been reported to date (Figure 1b).
Thanks to its redox amphoteric nature, this salt was indeed
expected to be electrochemically active in reduction
(System 2 in Figure 1c) but also in oxidation according to
a one-electron process (System 1 in Figure 1c). Here we
(
e.g., see the overview in ref. [2]), it appears particularly dif-
ficult to find air-stable lithiated organic materials that can
reversibly deintercalate lithium on cycling at high enough
potentials. For instance, in our previous investigations on di-
report on the chemical synthesis of Li DHDMQ, which is
2
[5]
[6]
lithium rhodizonate (Li C O ) and dilithium chloranilate
based on a simple three-step method, and discuss its electro-
chemical performances versus Li in comparison with those
of parent structures. A theoretical study at the DFT and dis-
persion-corrected DFT (DFT-D) levels of theory was also
performed. The general objective of this computational
study was first to test various methodologies for their abili-
ties to account for powder structural properties. Lattice con-
stants, stacking parameters, intramolecular bond lengths,
and interatomic distances were
2
6
6
(
Li C O Cl ), we observed that these two salts exhibit very
2 6 4 2
poor activity on oxidation whatever the cutoff potential,
[7]
while Xiang et al. reported no electrochemical response
for parent structure Li H C O (Figure 1a). Dahn and co-
2
2
6
4
[8]
workers recently reported a polymer based on a lithiated
naphthazarin unit that reacts at an average potential of 3 V,
but this material exhibits limited electrochemical perfor-
investigated. In parallel to the
study with classical exchange-
correlation (XC) functionals
(LDA, GGA, PBEsol, rev-
PBE), the original Grimme
method for dispersion correc-
[9]
tions (DFT-D2 version) was
tested together with versions of
this scheme modified by chang-
ing the damping function. One
of the modified dispersion-cor-
rected DFT schemes, related to
a concomitant rescaling of van
der Waals radii and the modu-
lated value for the s parameter,
6
provides significant improve-
ments to the geometry descrip-
tion of a known crystal struc-
ture that was experimentally
characterized in this work. To
assist the understanding of the
electrochemical properties, we
also envisaged various deinter-
calated framework models cor-
responding to oxidation of one
carbonyl group. This examina-
tion is related to the topic of
polymorphism prediction of
molecular crystals from the
knowledge of molecular geome-
try, which is a challenging task
for theoretical simulations. In
Figure 1. a) Chemical formulas of lithiated 1,4-benzoquinone derivatives previously studied versus Li.
b) Chemical formulas of TMQ, Li
sertion/deinsertion in Li DHDMQ with oxidation limited to a one-electron process (deeper oxidation is not fa-
vored due to the biradical nature of the thus-produced compound).
2 4
DHDMQ, and Li THQ. c) Schematic of the expected reversible Li-ion in-
2
&
2
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
this work, we restricted our scope to phases originating from
the lithiated compound, without extending the search to the
whole set of crystals, considering all possible structural ar-
rangements. The most stable phase among various hypothet-
ical structural frameworks was selected with a view to pro-
viding a computational estimation of the deintercalation po-
tential.
of the organic structure. From this first series of characteri-
zation it was possible to ascribe Li DHDMQ·3H O as
2
2
chemical formula for the brown compound.
The solid appears to be crystalline according to X-ray
powder diffraction (XRPD) measurements (Figure 3). Small
purple needles suitable for single-crystal XRD were grown
Results and Discussion
Synthesis and characterization of Li DHDMQ and related
2
compounds: The chosen synthetic route to prepare
Li DHDMQ starts with the preparation of TMQ by the pro-
2
[10]
tocol described by Verter et al., which consists of the re-
action of chloranil (2,3,5,6-tetrachlorocyclohexa-2,5-diene-
1
,4-dione) with a solution of methanolate. Inspired by the
[11]
work of Eistert et al., which described the synthesis of the
corresponding potassium salt K DHDMQ, we prepared the
2
lithium salt by a simple saponification procedure in which
TMQ was treated with LiOH·H O in water. The final brown
2
Figure 3. XRPD pattern of Li
Li DHDMQ·4H O (bottom).
2 2
DHDMQ·3H O (top) compared to that of
product was isolated in 93% yield. Chemical/physical prop-
erties of the as-prepared compound were characterized by
thermal/elemental analyses and IR/NMR spectroscopy (Fig-
ures S1 and S2 in the Supporting Information), which led to
the conclusion that a hydrated form was obtained. Clear
proof was readily obtained by thermal analysis coupled with
mass spectrometry (Figure 2). Indeed, the thermogravimet-
ric (TG) curve shows a first weight loss of about 20% asso-
ciated with several endothermic phenomena and the release
of water molecules (as detected by MS). Then a second
weight loss beyond 2508C involves thermal decomposition
2
2
by slow evaporation of a concentrated aqueous solution of
Li DHDMQ·3H O. Surprisingly, the crystal structure clearly
2
2
showed a tetrahydrate phase instead of the expected trihy-
drate. Li DHDMQ·4H O crystallizes in monoclinic space
2
2
group P2 /n with two independent water molecules per qui-
1
none half-molecule in the asymmetric unit. One of the two
independent water molecules has a disordered hydrogen
atom occupying two sites denoted H2b’ and H2b’’ (Figure 4).
Molecules are stacked along the a axis and linked together
through a lithium coordination network in the bc plane
forming a layered structure. Lithium is pentacoordinate,
with three oxygen atoms from the p-benzoquinone deriva-
ꢀ
tive (i.e., C=O, CO , and MeO) and two water molecules.
One of these water molecules is included in the molecular
layer, and the other lies between the layers. The large dis-
tance (4.66 ꢅ) between two stacked molecules could be ex-
plained by the cumbersome methyl groups pointing out of
the layer and especially by the presence of this intercalated
water molecule. In the crystal structure of TMQ, this inter-
[12]
planar distance is only 3.51 ꢅ.
Furthermore, the bulky methyl terminal groups prevent
face-to-face stacking of the p-benzoquinone and induce slip-
ping of the overlying sheet. Finally, contrary to anhydrous
TMQ, which has a weak intermolecular hydrogen-bond net-
work based on CꢀH···O interactions from methyl groups,
this structure exhibits a hydrogen-bond network developed
through the hydrogen atoms of the water molecules. They
are engaged in five different hydrogen bonds, involving two
oxygen atoms from the p-benzoquinone and the oxygen
atoms of the water molecules (Figure 4b). More precisely,
the two oxygen atoms from unsubstituted CO groups inter-
act with two hydrogen atoms from the first water molecule
(OꢀH···O 1.82(2) ꢅ, 176(2)8; OꢀH···O 1.88(2) ꢅ, 168(2)8)
Figure 2. Coupled thermal analysis and mass spectrometry of
ꢀ
1
Li
counting time for MS is 20 ms per m/z value with a resting time of 1 s
m/z 18 is ascribed to H O, 28 to CO, and 44 to CO ).
2 2
DHDMQ·3H O at a heating rate of 108Cmin under argon flow. The
(
2
2
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
Figure 5. Coupled thermal analysis and mass spectrometry of
ꢀ
1
Li
counting time for MS is 20 ms per m/z value with a resting time of 1 s
m/z 18 is ascribed to H O, 28 to CO, and 44 to CO ).
2 2
DHDMQ·4H O at a heating rate of 108Cmin under argon flow. The
(
2
2
water loss now involves two distinct steps but without in-
volving a trihydrated phase as intermediate, although the
quite similar appearance of the two TG/DSC traces beyond
Figure 4. a) View along the
b axis of the layered structure of
Li DHDMQ·4H O. Hydrogen atoms of methyl groups are omitted for
2
2
2
508C reveals a common decomposition reaction path (Fig-
clarity. b) View of a layer along the a axis showing lithium coordination
polyhedra and highlighting the hydrogen-bond network (dashed lines).
Nomenclature for water molecules is also detailed, whereby H2b’ and H2b’’
are relative to the disordered hydrogen atom with an occupancy of 0.5.
ures 2 and 5). We attempted to produce the anhydrous
Li DHDMQ phase for testing as electrode material versus
2
Li and to reveal more details of the thermal dehydration
process of the two hydrates by performing temperature-con-
trolled powder XRD experiments on the two hydrates from
and with one of the three hydrogen atoms belonging to the
room temperature to 5508C under N at a heating rate of
0.28Cs , in accordance with the thermal analysis experi-
ments (Figure 6).
Unexpectedly, the tetrahydrate undergoes direct amorphi-
zation beyond 608C, while dehydration of the trihydrate
2
ꢀ
1
second water molecule (OꢀH···O 2.23(7) ꢅ, 144(6)8). The
last two hydrogen bonds are observed between the two re-
maining hydrogen atoms from the latter water molecule and
the oxygen atom from the two other water molecules (Oꢀ
H···O 2.03(4) ꢅ, 150(4)8; OꢀH···O 2.15(7) ꢅ, 179(7)8).
makes formation of crystallized Li DHDMQ possible; the
2
To obtain the tetrahydrate in a larger amount and further
characterize it, we considered its synthesis from the trihy-
drate. Interestingly, the latter can be readily converted to
the tetrahydrate analogue by further hydration in a water-
saturated atmosphere. The brown powder turned purple on
further hydration, in agreement with the color of the crystals
of Li DHDMQ·4H O previously obtained. Finally, the ob-
latter is obtained in pure form for temperatures ranging
from 150 to 2308C. This fast amorphization step is support-
ed by the DSC measurement showing a first endothermic
peak near 708C (Figure 2), which is not present at all in
Figure 5. On heating, a common new phase arises between
300 and 3508C for the two hydrates, and turned out to be
Li CO , in accordance with the expected thermal decompo-
2
2
2
3
tained tetrahydrate phase was fully confirmed by means of
a profile-matching refinement of the XRPD pattern with
cell parameters obtained from single-crystal XRD structure
resolution (Figure S3 in the Supporting Information). Good
agreement between observed and calculated patterns
sition of such compounds. As discussed below, it is better to
use anhydrous compounds as electrode materials to avoid
side reactions with water in the electrochemical cell. We ex-
ploited dehydration of the trihydrate to produce anhydrous
Li DHDMQ. However, we used the freeze-drying technique
2
proved the purity of our crude material (R =7.60, R
=
wp
(lyophilization) to remove water, since this method, besides
being commonly used in organic and biochemistry for frag-
p
7.77). However, the corresponding thermal analysis data
[13]
showed some differences in the dehydration process in com-
ile compounds like proteins, can produce very small parti-
[14]
parison with Li DHDMQ·3H O (Figure 5). While the pres-
cles because it leaves microscopic pores. Residual water
was then removed by a short thermal annealing step under
2
2
ence of four water molecules is confirmed by the TG curve,
&
4
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
Figure 6. Temperature-controlled XRPD experiment performed under N
2 2 2 2 2
flow on a) Li DHDMQ·4H O and b) Li DHDMQ·3H O. Circles denote peaks
of Li CO phase, and stars peaks emanating from the sample holder (reported only at T=300 and 5508C for the sake of conciseness).
2
3
vacuum at 1808C. Finally, Scheme 1 recaps the synthesis
corresponding anhydrous compound 4 obtained by freeze-
routes to get Li DHDMQ and the related compounds ob-
served in our investigation.
drying were first measured versus Li by a similar procedure
2
[4–6]
to our previous studies.
In particular, we used Swagelok-
type electrochemical cells, simple hand-milling of the active
materials with carbon black without binder, and a molar
Electrochemical reactivity of Li DHDMQ·yH O (y=4, 3, 0)
2
2
versus Li: Electrochemical behavior of hydrates 2, 3 and the
LiPF solution in ethylene carbonate/dimethyl carbonate as
6
electrolyte. Under such experimental conditions, a prelimina-
ry investigation showed that degradation of the compounds
may occur at both too high and too low potentials, and
hence the potential window was restricted to the range 3.5–
+
0
1
.5 V versus Li /Li . Figure 7 shows the corresponding first
charge/discharge profiles recorded at a typical cycling rate
+
of one Li exchanged in 10 h. As expected, reversible elec-
trochemical activity was systematically observed on oxida-
tion, unlike dilithium rhodizonate or dilithium chloranilate.
During the charging process all curves exhibited a well-de-
fined plateau near 3.4 V related to a capacity of x ꢁ0.8
Li
(
i.e., ꢁ80% of the theoretical value of System 1 in Fig-
ure 1c). In the subsequent discharge process the plateau is
mainly recovered for 3 and 4 and completed with a sloping-
type profile up to x =2 (starting composition), whereas the
Li
tetrahydrate exhibits a more complex behavior. Like for the
[8]
polymer based on a lithiated naphthazarin unit, this elec-
trochemical deintercalation/intercalation process unfortu-
nately suffers from a large polarization value of about
9
00 mV, which greatly reduces the energy efficiency of such
an electrochemical reaction. Note that the polarization
effect is slightly decreased after the first cycle for com-
pounds 3 and 4. Beyond x =2, the redox amphoteric
Li
nature of such compounds makes reduction of the quinonic
skeleton possible, too (Figure 1c, System 2). Within this po-
tential–composition domain, the two hydrated phases in-
volve a reversible plateau followed by a sloping-type profile.
Scheme 1. Synthesis of Li
2
DHDMQ·yH
2
O (y=4, 3, 0).
The lithium uptake is, however, restricted to x =3, which
Li
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
Figure 8. Potential–composition curve focusing on the high-potential
region for a Li half-cell using Li DHDMQ prepared by freeze-drying and
2
+
cycled in GITT mode at a rate of one Li exchanged in 10 h for Dx=0.1
followed by relaxation period of 20 h (T=21ꢃ0.18C). The dashed line
correspond to the GITT charge/discharge near-equilibrium potential
curve.
+
rent equivalent to a cycling rate of one Li exchanged in
1
0 h was applied during 1 h (i.e., Dx=0.1), followed by a re-
laxation period of 20 h at zero current (Figure 8). As expect-
ed, the shape of the GITT curve measured under current is
similar to that shown in Figure 7, exhibiting pronounced
electrode polarization but restricted to the composition
range 1.1ꢂDxꢂ 2. Conversely, the charge/discharge near-
equilibrium potential curve (Figure 8, dashed line) now
shows a polarization value reduced by 75% within this par-
ticular composition range, which clearly confirms both the
Figure 7. Typical potential–composition curve of Li
2 2
DHDMQ·yH O/Li
cell (y=4, 3 and 0) galvanostatically cycled between 3.5 and 1.5 V versus
+
0
+
Li /Li at a rate of 1Li exchanged in 10 h with EC–DMC/LiPF
electrolyte.
6
1m as
+
existence of some kinetic limitations for the specific Li
·
ꢀ
seems to indicate a limited reduction process. Interestingly,
compound 4 shows quite a different electrochemical feature,
as it involves a series of sloppy plateaus dropping step by
step, which leads to a larger capacity value. Within this
lower potential window (i.e., 1.5–2.5 V), the polarization
effect is much more reasonable for the three compounds. In
ACHTUNGTNERNGNU( DHDMQ )/Li DHDMQ biphasic system (System 1) and
2
that the corresponding formal potential should be quite
+
0
close to 3 V versus Li /Li . Several parameters were then
adjusted with the aim of decreasing the polarization effect,
such as changing the electrolyte composition, Li exchange
rate, milling procedure, and nature of the carbon. However,
no clear improvement was observed, which seems to con-
firm that the activation energy is intrinsic to System 1.
Some attempts to improve both the specific capacity and
[5–7]
accordance with other studies,
it appears that the anhy-
drous phase exhibits the best electrochemical performance
on cycling. Indeed, the presence of water generally induces
some side reactions with the electrolyte and a worse electro-
chemical performance as a consequence. Nevertheless, a no-
ticeable decrease in capacity on cycling is still noticed with
the stability on cycling of Li DHDMQ were also investigat-
2
ed. First, we took advantage of the redox amphoteric nature
of this compound and the fact that System 2 involves, at the
beginning of the process, a flat plateau located at an average
value of 2.45 V (Figure 8) to probe the electrochemical per-
formances within a new potential window of 3.6–2.15 V. Sec-
ondly, with the aim of simply circumventing the solubility
issue at this stage without trying to design efficient compo-
site electrodes, we assessed addition of a load of an adsorb-
ing (inert) material additive during preparation of the com-
posite electrode. In particular, we benefited from recent
work that demonstrated the ability of g-Al O nanoparticles
Li DHDMQ, due to a certain solubility in the liquid carbon-
2
ate-based electrolyte, like that previously observed with
[6]
Li C O Cl , another parent structure. Importantly, it was
2
6
4
2
also checked that the electrochemical feature of the poten-
tial–composition curve is maintained whatever the adopted
cycling direction, that is, by first discharging the cell till
1
.5 V before subsequent cycling within the potential range
+
0
of 3.5–1.5 V versus Li /Li .
Further electrochemical investigations focused on the an-
hydrous compound. First, to gain insight into the hysteresis
2
3
to sustain the capacity performance in a lithium/sulfur cell
by adsorbing lithium polysulfides and preventing their disso-
effect at high potential, another Li DHDMQ/Li cell was
2
[15]
cycled by using the galvanostatic intermittent titration tech-
nique (GITT) to obtain a charge/discharge curve closer to
thermodynamic equilibrium. Experimentally, a constant cur-
lution in the electrolyte. We thus prepared another com-
posite electrode by milling Li DHDMQ/carbon black (SP)/
2
g-Al O in 56.6/33.3/10 weight ratio. Its electrochemical re-
2
3
&
6
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
sponse was then studied under stronger conditions by apply-
+
ing a galvanostatic cycling test at a rate of one Li ex-
changed in 2 h. While the charge/discharge profile is inter-
estingly not affected by the presence of this inactive and in-
sulating additive, it appears that capacity retention is clearly
improved, with 74% capacity recovered after 20 cycles and
ꢀ1
a residual capacity of 80 mAhg after 50 cycles (Figure 9).
Figure 10. Comparison of the typical first discharge/charge profiles of
TMQ, Li DHDMQ, and Li THQ cycled galvanostatically versus lithium
2
4
+
at a rate of one Li exchanged in 10 h under equivalent experimental
conditions.
is, the relationship between substituents and the reduction
[18]
potentials of quinones also exists in the solid state.
DFT calculations and methodology
Figure 9. Capacity retention curves of Li
cally cycled between 3.6 and 2.15 V versus Li /Li at a rate of one Li
exchanged in 2 h, with and without a 10 wt% load of g-Al additive.
Inset: corresponding potential–composition curve for composite elec-
trode containing alumina as additive.
2
DHDMQ/Li cell galvanostati-
Geometry optimizations: The first step of this theoretical ap-
proach was validation of the applied methodology for struc-
tural description. In organic crystals, such studies encompass
not only identification of the most suitable density function-
als, but also inclusion of van der Waals interactions. We first
used solvated Li DHDMQ·4H O as benchmark system to
+
0
+
2
O
3
2
2
This capacity is higher than for an alumina-free electrode
even when the total electrode mass is considered, that is, the
capacity gain over cycling offsets additional mass due to ad-
ditive. Such a result demonstrates that better cycling perfor-
mance is expected by designing elaborate composite elec-
trode architectures, in particular by appropriate use of poly-
mer binders and other additives to prevent residual dissolu-
tion phenomenon (e.g., promoting specific interactions to-
define the most suitable methodology to account for geome-
try features in our typical phases. We first addressed the rel-
ative abilities of various exchange-correlation functionals
[19]
(LDA,
GGA,
Perdew–Burke–Ernzerhof (PBE) variant of
[20]
[21]
[22]
PBEsol,
and revPBE ) to account for both
intra- and intermolecular geometries. The experimental
single-crystal XRD structure of Li DHDMQ·4H O was
2
2
energy-minimized in full, including unit cell parameters and
atomic positions. However, standard DFT methods are
known to fail in describing the long range van der Waals in-
teractions. We therefore used a DFT code incorporating dis-
persion corrections in density functionals, that is, the pseu-
dopotential VASP package (see Computational Details sec-
[16]
wards organic-based electrode materials).
This is an
alternative pathway to the use of redox-active polymers or
grafting processes to make more insoluble active organic
moieties.
Finally, having also prepared TMQ (1) as an intermediate
[23]
for the synthesis of Li DHDMQ (Scheme 1) and knowing
tion).
2
[4]
the electrochemical reactivity of Li THQ, we compared in
Three methods adding semi-empirical correction for dis-
persion were considered (PBE-D, PBE-D*, corr-PBE-
D*_0.52) and described hereafter. Using the original van
der Waals radii may lead to overestimation of the dispersion
forces and hence underestimation of the lattice constants.
Results may thus be improved by modification of the semi-
empirical vdW correction. In addition to DFT-D2 calcula-
tions (labeled PBE-D), the original Grimme model was also
modified here by rescaling van der Waals radii in the damp-
ing function (scaling factors: 1.30 for hydrogen and 1.05 for
all other atoms), whereas the value of the global scaling
4
a single layout the typical discharge/charge profile of this
series of similar p-benzoquinone derivatives under equiva-
lent experimental conditions (Figure 10). As previously
[4,5,17]
mentioned,
we first confirmed with this series that
better stability of the organic electrode material is obtained
when the structure is negatively charged (lower solubility).
For instance, Li THQ reaches its full theoretical capacity
4
(
two-electron reduction) after the first discharge due to its
higher stability towards the electrolyte. Conversely, the aver-
age redox potential is decreased simultaneously due to elec-
tron enrichment of the redox-active quinonic structure, that
factor s was set to 0.75. This parameters modification was
6
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
proposed by Civalleri et al. (PBE-D*) and normally enables
roughly the same (ꢄ4%). The most satisfying results for V
and b are observed for PBE-D* and corr-PBE-D*_0.52,
with deviations from experimental V of ꢀ2.8 and ꢀ1.1%, re-
spectively. For b, we find deviations on the order of only 1–
28 with these two DFT-D methods. Rev-PBE gives the worst
result (DV/Vꢄ16%) and therefore completely fails to cor-
rectly account for the geometry originating from distinct
kinds of interactions in this crystal (electrostatic, dispersion,
polarization, etc.).
[24,25]
softening of the dispersion interaction.
In the method
labeled corr-PBE-D*_0.52, an optimized s6 value (lower
than that of 0.75 proposed by Grimme and used for PBE-D/
PBE-D* calculations) was also employed while leaving RvdW
fixed at the PBE-D* values (see Computational Details sec-
[26]
tion).
Li DHDMQ·4H O and related dehydrated or partially hy-
2
2
drated phases: The quality of structural reproduction for the
Li DHDMQ·4H O crystal can be evaluated by comparing
In the Li DHDMQ·4H O crystal, stacking of molecules
2
2
occurs along the a direction. On the other hand, due to the
orientation of molecules within the crystal, dispersion effects
also involve the c lattice parameter. As a consequence, we
expect bad descriptions of these two quantities from non-
dispersion-corrected XC functionals. Large deviations are
indeed found for one or both directions for rev-PBE, LDA
and PBE methods, which prevent them from being used for
a proper description of the system. Figure 11b shows that
the relaxation systematically gives rise to an underestima-
tion of the a lattice parameter, except for the rev-PBE func-
tional, which leads to an overestimation. Especially large
underestimations of a are found for the PBEsol (ꢀ5.1%)
and PBE-D (ꢀ5.4%) methods, and an even larger one for
LDA (ꢀ9.0%). Corr-PBE-D*_0.52 and PBE-D* decrease
this discrepancy to about ꢀ4%. On the whole, due to the
lack of involvement in the dispersion interactions, the b lat-
tice parameter is better accounted for by the various tested
methodologies, and the most suitable treatment for this pa-
rameter is PBE. Only a slight overestimation is introduced
within DFT-D (corr-PBE-D*_0.52 and PBE-D*) compared
to the PBE. The variation of the calculated interlayer spac-
ing (i.e., interplane distance), d, as a function of the level of
theory is depicted in Figure 12, which shows that corr-PBE-
D*_0.52 still provides the best result among all tested meth-
ods, with an underestimation of less than ꢀ2%. The calcu-
lated d values are not satisfactory for pure DFT methods
except for PBEsol, which properly represents the interlayer
distance (slight underestimation by ꢀ1.5%). On the other
2
2
the calculated structures with the available X-ray crystal
structure. This allows for a quantitative assessment of the
performance for each exchange-correlation functional or
DFT-D method for the same set of computational parame-
ters (k-points grid/cutoff energy).
The changes in unit-cell volume V, monoclinic angle b,
and lattice dimensions with respect to the experimental X-
ray data for the full-geometry optimizations are presented
in Figure 11. Unit-cell volumes are underestimated by all
methods, except PBE and rev-PBE, which lead to overesti-
mation. The absolute discrepancies of calculated V for PBE-
D and PBEsol methods with respect to the experiment are
2 2
Figure 11. Li DHDMQ·4H O: Relative errors [%] of the DFT or DFT-D
calculated values to the experimental ones for a) the unit-cell volume V
2 2
Figure 12. Li DHDMQ·4H O: Relative errors [%] of the DFT or DFT-D
and monoclinic angle b and b) the optimized lattice parameters, a, b, and
c.
calculated values to the experimental ones for the optimized inter-plane
distance d.
&
8
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
hand, PBE-D and PBE-D* do not offer a better description
of the stacking parameter than the PBE model, that is, ade-
rev-PBE, PBE, and LDA functionals all fail to describe such
interactions, which belong to the electrostatic type, while
the best experimental description is observed for the
PBEsol method (RMSD as low as 0.03 ꢅ). Addition of the
dispersion correction to the PBE functional lowers the
RMSD values by more than half, to a more satisfying value
of 0.08 ꢅ (i.e., a discrepancy with the experiment of ꢄ4%).
Modifications of the DFT-D2 parameters also tend in this
direction, for instance, with an RMSD value of 0.04 ꢅ for
the corr-PBE-D*_0.52 functional (i.e., a discrepancy with
the experiment of ꢄ2%).
quate selection of the s value is crucial. From the above-
6
mentioned results, we conclude that corr-PBE-D*_0.52 gives
the best compromises in the description of crystal structure
parameters.
To complete the discussion of the geometry optimization,
we also consider the effect of the XC functional or DFT-D
method on bond lengths and interatomic distances. Indeed,
despite weak influence on the electronic structure, intramo-
lecular geometry directly affects the structure and energy of
the systems in the computations, which are meaningful espe-
cially for estimation of the intercalation potential. For
Li DHDMQ·4H O, root mean square deviations (RMSDs)
In conclusion, on the whole, geometric features in
Li DHDMQ·4H O crystal are most reliably described by
2
2
corr-PBE-D*_0.52. Therefore, calculations on the partially
2
2
for intramolecular bond lengths and intermolecular distan-
ces with respect to the experimental data are gathered in
Figure 13. The RMSD values for intramolecular bond
delithiated phase LiDHDMQ·4H O and those of
2
Li DHDMQ or LiDHDMQ·3H O crystals have been per-
2
2
formed with this methodology, because similar kinds of in-
teractions occur in these compounds, too.
The nature of the crystal phase for Li DHDMQ was in-
2
vestigated by removing all water molecules within the re-
laxed structure of Li DHDMQ·4H O and allowing subse-
2
2
quent full-geometry optimization with no constraints on the
crystalline symmetry (i.e., P1). The resulting relaxed struc-
ture remains in the same space group (P2 /n). Nevertheless,
1
despite preservation of the symmetry, crystallographic data
change significantly. The relaxed lattice dimensions and
monoclinic angle for Li DHDMQ are a=6.068, b=6.408,
2
c=14.118 ꢅ, and b=109.858. Relaxation thus led to reduc-
tion of the b lattice parameter, while a and c increase by
1
2.7 and 10.4% in comparison to Li DHDMQ·4H O. These
2 2
variations correspond to a volume diminution of ꢀ18.97%
compared with the solvated phase and can be related to the
modification of the stacking direction within the lattice,
which now involves a and b lattice parameters.
For the partially dehydrated phase Li DHDMQ·3H O,
2
2
2 2
Figure 13. Li DHDMQ·4H O: RMSD values with respect to experimen-
which was also relaxed without symmetry constraints, the
optimized lattice dimensions and angles of the resulting tri-
clinic cell are a=4.827, b=9.662, c=13.156 ꢅ and a=90.16,
b=109.85, g=87.138. Since our experimental attempt at
tal data for the intramolecular bond lengths and intermolecular distances
without taking into account covalent bonds involving an H atom or H-
bonding.
structure
determination
of
Li DHDMQ
2
and
lengths which do not involve any H atoms or H-bonding are
fairly uniform across all calculations (except for rev-PBE),
and this indicates somewhat consistent reproduction of the
covalent molecular structure in full-geometry optimizations
regardless of the correction for dispersion forces. As expect-
ed, introduction of dispersion does not modify the CꢀC, Cꢀ
Li DHDMQ·3H O crystals was not successful, we only com-
2
2
pared the experimental XRPD pattern to that generated
from our calculated phase (Figure S4 in the Supporting In-
formation). However, due to the mismatch of these two pat-
terns, we consider that our relaxed phase generated by this
methodology is not the relevant one.
O, and C=O bond lengths with respect to conventional PBE
calculation. The overall RMSD value of about 0.015 ꢅ for
these various DFT and DFT-D methods is of similar order
of magnitude to typical values found in Li C O , which does
LiDHDMQ·4H O: We have applied the most suitable
2
methodology identified for the Li DHDMQ·4H O crystal
2
2
above (corr-PBE-D*_0.52) to relaxation of selected hypo-
thetical crystal structures of the delithiated phases. Three
models were screened corresponding to different sites for
removal of lithium atoms. The lattice constants, unit-cell
volume, and stacking parameters of these relaxed frame-
works are listed in Table S1 of the Supporting information.
Among all structural hypotheses, the lowest stable phase
was 0.2 eV per formula unit (p.f.u.) from the most stable
2
4
4
[27]
not have any H atoms in the structure. The corresponding
RMSD value for the rev-PBE functional is almost seven
times higher than the best reproduction among all methods,
highlighting the bad description of this feature by this func-
tional.
Figure 13 considers solely intermolecular bond lengths
+
ꢀ
that correspond to Li ···O interactions. In this case, the
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
the most stable LiDHDMQ·4H O framework, they are lo-
2
cated in a distorted tetrahedron (dmin =1.922, dmax =1.988 ꢅ;
Figure S8 in the Supporting Information). On lithium ex-
traction from Li DHDMQ·4H O, a volume decrease of
2
2
ꢀ
7.1% is observed in our most stable crystal. The c lattice
parameter shows a contraction of ꢀ15.7%, while the b lat-
tice dimension is highly augmented (+12.6%). In compari-
son, the a lattice parameter is much less affected (ꢀ2.0%).
Energetics and average potential: Calculated equilibrium po-
tential corresponding to oxidation of Li DHDMQ·4H O is
2
2
compared with experimental data obtained from electro-
chemical measurements. The equilibrium average potential
of the deintercalation (oxidation) phenomenon for
Li DHDMQ·4H O experimentally determined through
2
2
open-circuit voltage is about 3.0 V (Figure 8). The equilibri-
um potential V(x) relative to lithium metal for lithium inser-
tion into a host material M can be estimated according to
Equation (1)
GðMÞ þ xGðLiÞ ꢀ GðLi MÞ
x
VðxÞ ¼
ð1Þ
xz
where G is the Gibbs free energy p.f.u. and z the elementary
charge per lithium ion (z=1).
By neglecting small changes in entropy (TDS) and volume
[28]
(
PDV), which are expected to be relatively small,
the
Figure 14. Li
moved from the compound in order to perform optimization of
LiDHDMQ·4H O are enclosed within dotted circles; b) Optimized crys-
2 2
DHDMQ·4H O: a) Relaxed crystal structure; Li atoms re-
Gibbs free energy (DG) can be approximated by the inter-
nal energy (DE) and the average potential can be expressed
as Equation (2)
2
tal structure showing Li polyhedra.
EðMÞ þ xEðLiÞ ꢀ EðLi MÞ
x
VðxÞ ¼
ð2Þ
xz
one. This most stable delithiated structure, LiDHDMQ·4
H O, is obtained by removing Li atoms at the following po-
2
where E is the total energy of each system p.f.u and E(Li)
sitions: (0.72, 0.14, 0.64) and (0.77, 0.64, 0.85). This phase is
denoted 1-4 in reference to Li atom positions (Figure 14a).
Other hypothetical frameworks (i.e., 1-2 and 1-3, Figures S5
and S6 in the Supporting Information) correspond to remov-
al at atomic positions (0.77, 0.64, 0.85) and (0.27, 0.86, 0.35);
the total energy for lithium metal in the bcc structure).
After geometry optimizations of both lithiated
(
Li DHDMQ·4H O) and delithiated (LiDHDMQ·4H O)
2
2
2
phases, the average lithium extraction potential can thus be
extracted from Equation (2), in which x=1, since one lithi-
um ion p.f.u. is considered to be removed from
Li DHDMQ·4H O. The resulting equation is as follows
(
0.72, 0.14, 0.64) and (0.27, 0.86, 0.35), respectively. Al-
though model 1-4 is slightly more stable (by 0.01 eV) than
-2, the crystal structures of these two forms emerging from
the relaxation are quite similar (DV/Vꢄꢀ7.1%; Dd/dꢄꢀ36/
37%). Model 1-3 is characterized by less structural change
2
2
1
[
Eq. (3)].
ꢀ
E ¼ EðLiDHDMQ ꢅ 4 H OÞ þ EðLiÞ ꢀ EðLi DHDMQ ꢅ 4 H OÞ
2
2
2
(
DV/Vꢄꢀ3.5%; Dd/dꢄ +2.4%) and moderately higher
energy (by 0.2 eV). Due to the higher stability of the 1-4
form, this phase is the only one considered in the following,
including computed estimation of the deintercalation poten-
tial. The crystal structure of the Li DHDMQ·4H O and op-
ð3Þ
The calculated Li DHDMQ·4H O deintercalation poten-
2
2
tial (3.75 V) exhibits quite a noticeable discrepancy of
+0.75 V compared to experiment. We note that the selected
method of calculation may noticeably impact the total
energy of both lithiated and delithiated phases. Thus, other
methodologies are currently being tested (including hybrid
functionals). In addition, the procedure used in this work
does not completely guarantee reaching a global minimum
for the delithiated phase in the potential-energy surface of
2
2
timized 1-4 LiDHDMQ·4H O phases are depicted in Fig-
2
ures 4b and 14b, respectively. The symmetry of the relaxed
1
-4 LiDHDMQ·4H O phase is lowered (space group P1).
2
In the non-deintercalated Li DHDMQ·4H O phase that
2
2
issued from the corr-PBE-D*_0.52 calculation, Li ions lie in
a distorted bipyramidal environment (dmin =1.935, dmax
=
2.338 ꢅ; Figure S7 in the Supporting Information), while in
&
10
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
our system, due to limited probing of the relative arrange-
ment of Li ions and molecules. We should therefore
extend our screening of hypothetical structures in a forth-
coming work.
Experimental Section
+
1
13
General methods: H and C NMR spectra were recorded on a Bruker
DPX-300 at 300 MHz and 75 MHz, respectively; chemical shifts (d) are
1
3
given in ppm relative to TMS. For compound 2, the C NMR spectrum
was recorded in D O by using a capillary filled with [D ]DMSO as inter-
2
6
nal standard. Infrared spectra were recorded with a Thermo Scientific
Nicolet iS10 FTIR equipped with an attenuated total reflectance (ATR)
probe. ESI-HRMS experiments were performed in positive-ion mode on
a Q-TOF Ultima Global instrument (Waters-Micromass, Manchester,
UK) equipped with a pneumatically assisted electrospray ion source (Z-
spray) and an additional sprayer for the reference compound (Lock-
Spray). Elemental analyses were performed on a Thermo Finnigan EA
1112 Series Flash elemental analyzer controlled by Eager 300 software.
The DSC measurements were carried out with a Netzsch DSC204 calo-
rimeter under argon flow by using sealed aluminum crucibles with cap
layers perforated just before analysis to allow gas release. Thermogravi-
metric (TG) measurements were carried out under argon with a Netzsch
STA 449 C coupled with mass spectrometry for gas analysis by using alu-
mina crucibles. TG/DSC analyses were obtained at a heating rate of
Conclusion
[
4]
Building on our previous investigations on Li THQ, we
have designed and synthesized, through a simple three-step
method, Li DHDMQ, another derivative of the Li C O -
type unit. As expected, an electrochemical investigation
4
2
4
6
6
versus Li revealed that Li DHDMQ can reversibly deinter-
2
calate lithium according to a one-electron process at an
average potential of 3 V (System 1), while the p-benzoqui-
none redox center of the structure reacts below 2.5 V
System 2). A noticeable decrease in capacity on cycling was
also noticed due to certain solubility in the electrolyte,
(
ꢀ
1
1
08C min .
X-ray diffraction
which can partly be suppressed by simply adding g-Al O
2
3
nanoparticles to the composite electrode; this suggests that
much better performance can be expected by designing effi-
cient composite electrodes. As previously observed by Dahn
et al. for a polymer based on a lithiated naphthazarin unit,
the reversible deintercalation process occurring in
Powder XRD: Powder XRD was performed on a Bruker D8 Advanced
diffractometer with Cu radiation (l =1.54056, l =1.54439 ꢅ) at 40 kV
1 2
and 40 mA equipped with a Linkseye detector. Temperature-controlled
experiments were performed on a Bruker D8 diffractometer with a Co
1 2
radiation (l =1.788970, l =1.792850 ꢅ) equipped with an Anton Parr
[8]
Chamber HTK from room temperature to 5508C. Each pattern was re-
corded under nitrogen flow with a step of 0.02998 and an acquisition time
Li DHDMQ (System 1) also suffers from quite large polari-
2
ꢀ1
of 2.7 s per step (heating rate of 0.28Cs ). All diffraction profiles and
structure refinements were generated by using the FullProf Suite soft-
ware. This set of programs is developed and constantly updated by Ro-
zation. Efforts to reduce this effect failed, and this suggests
a kind of intrinsic limitation phenomenon, as is typically ob-
served for conversion reactions involving metal oxides as
[30]
drꢆguez-Carvajal and co-workers.
[29]
active materials. Further investigations are in progress, in-
cluding electrochemical impedance spectroscopy measure-
ments to determine the electrode kinetics and to isolate con-
tributions from different physical and kinetic processes.
In addition to electrochemical characterization of hydrat-
Single Crystal XRD: XRD data were collected on a Bruker X8 Apex2
CCD4K at room temperature (MoKa radiation, l=0.71073 ꢅ). Reflec-
tions were collected up to q=26.148 to a completeness of 99% by follow-
ing a strategy based on f and w scans. Intensities were corrected for ab-
sorption effects by a semi-empirical method based on redundancy by
[31]
using the SADABS program. The structure was solved by direct meth-
[
32]
ed and solvent-free Li DHDMQ, the crystal structure of the
ods using SHELXS, and least-squares refined with the JANA2006 Soft-
2
[
33]
solvated form of this phase (Li DHDMQ·4H O) has been
ware package.
Crystal data for Li
group P2 /n (no. 14), a=5.3318(5), b=9.5162(9), c=12.7812(13) ꢅ, b=
6.490(10)8, V=644.34(11) ꢅ , Z=2, 1calcd =1.464 gcm , crystal dimen-
sions: 0.3ꢇ0.2ꢇ0.1 mm, 26652 reflections collected, 1284 unique, Rint
.66%, R(obsd) =3.16%, wR(all) =3.21%, R indices based on 978 reflections
with I>3s(I) (refinement on F), 120 parameters, max./min. residual elec-
2
2
ꢀ
1
determined. Crystallographic data, at room temperature, of
this compound were used to test various functionals and dis-
persion energy corrections within DFT calculations. The
original PBE-Grimme method for dispersion corrections
2 2
DHDMQ·4H O: M=284 gmol , monoclinic, space
1
3
ꢀ3
9
=
4
(
DFT-D2) was tested together with versions of this scheme
ꢀ
3
modified by changing the damping function. The shortcom-
ings of classical functionals (e.g., GGA, LDA, and revPBE)
were outlined, whereas a modified DFT-D2 scheme involv-
ing changes in particular parameters was identified as the
most suitable methodology for description of this particular
crystal type. This variant of DFT-D2 was then selected for
tron densities: 0.17/ꢀ0.14 eꢅ . Hydrogen positions from methyl groups
and from water molecules were located on Fourier difference maps and
introduced in the refinement. The hydrogen atoms were refined isotropi-
cally and thermal agitation factors were restrained to the same value for
hydrogen atoms belonging to the same water molecule (i.e., H1a and H1b
for one part and H2a, H2bꢈ and H2b“ for the other. CCDC 856689 contains
the supplementary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif.
the relaxation of 1) solvent-free Li DHDMQ and partially
2
dehydrated phase Li DHDMQ·3H O, and 2) selected hypo-
2
2
Electrochemistry: Electrochemical studies were performed in Swagelok-
thetical crystal structures after lithium extraction (both un-
known experimentally). Due to the used methodology, not
all possible frameworks for these two kinds of phases were
tested in this work. To complement this first approach, more
sophisticated crystal-structure prediction tools will thus be
applied in forthcoming work in order to extend our studies
to a more complete range of crystal structures.
type cells with an Li metal disk as negative electrode and a fiberglass
separator soaked with a molar LiPF solution in ethylene carbonate/di-
6
methyl carbonate (1/1 v/v) as electrolyte. The positive electrode was pre-
pared without binder by mixing organic compounds with 33% carbon SP
(
in total mass). Cells were typically cycled in galvanostatic mode by using
a Macpile or VMP system (Biologic S.A., Claix, France) at a typical rate
of one lithium ion exchanged in 10 h unless otherwise specified.
Synthesis: Methanol was purchased from VWR Prolabo (HPLC grade).
LiOH·H
none (99%) by Acros.
2
O was supplied by Fluka (ꢁ99%), and tetrachloro-p-benzoqui-
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
11
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
Tetramethoxy-p-benzoquinone (1): A sodium methanolate solution was
prepared in a closed vessel by adding sodium (0.930 g, 40 mmol) by por-
tion to methanol (20 mL) cooled at 08C. After complete reaction of
sodium with methanol, a heterogeneous solution of chloranil (2.46 g,
In the VASP program, the DFT-D2 approach of Grimme is implemented,
corresponding to semi-empirical dispersion corrections. They are includ-
6
ed by a summation of damped interatomic C
6
/R terms (additional atom-
atom potentials that vanish for covalent bonding distances). The Grimme
scheme using such pairwise interaction terms is defined according to
Equation (4)
1
0 mmol) in methanol (5 mL) was added to the methanolate solution.
The mixture, which immediately turned reddish, was then heated at 858C
for 6 h in the closed vessel. The product crystallized on cooling to room
temperature. The orange needles were recovered by filtration, washed
Nꢀ1
N
ij
6
6
ij;g
X X X C
ꢀ
ꢁ
E
disp ¼ ꢀs
6
f
dmp
R
ij;g
ð4Þ
several times with water and finally dried under vacuum overnight to
R
i¼1 j¼iþ1
g
1
give the desired compound (1.860 g, 82%). H NMR (300 MHz, CDCl
3
):
),
1
3
3 3
d=3.98 ppm (s, 12H, CH ); C NMR (75 MHz, CDCl ): d=61.41 (CH
3
where the energy is the summation over all atom pairs and g lattice vec-
tors, N is the number of atoms, s is a functional-dependent global scaling
1
1
(
42.85 (COMe), 180.5 ppm (C=O); IR (ATR): ~n =2969–2842 (CꢀH),
6
666, 1660 (C=O), 1461, 1438 (C=C), 1272 (C-O-C), 1072 (C=C); HRMS
ij
6
factor, C are the dispersion coefficients of atom pair ij, and Rij,g is the in-
+
+
10 12 6
ESI): m/z calcd for C H O +Na : 251.0532 [M+Na ]; found:
ternuclear separation of the atom pair. The interaction between each pair
of atoms is approximated by a damped multipole expansion, often (but
2
51.0528.
,6-Dihydroxy-2,5-dimethoxy-p-benzoquinone dilithium salt trihydrate
O (0.092 g,
.2 mmol) were placed in a three-neck round bottomed flask under argon
ij
C
6
3
not always) truncated after the first term R
6
. To avoid near-singularities
ij;g
(
2): Ground compound 1 (0.250 g, 1.1 mmol) and LiOH·H
2
for small Rij,g values and double-counting effects of correlation at inter-
mediate distances, a damping function fdmp must be used [Eq. (5)]
2
atmosphere. Distilled water (2 mL), previously degassed under a flux of
argon, was added to the mixture. After heating at 908C for 2 min, the ob-
tained purple solution was placed in the freezer for 2 h. Addition of ace-
tone to the frozen solution led to the formation of a brown precipitate
which was immediately filtered off. The compound was finally washed
twice with acetone affording pure 2 (0.267 mg, 1.02 mmol, 93% yield).
ꢀ
ꢁ
1
f
dmp
R
ij;g
¼
ð5Þ
ꢀ
dðRij =Rr ꢀ1Þ
1 þ e
where R
r
is the sum of atomic vdW radii RvdW
.
In the DFT-D2 approach (applied here to the PBE functional), the total
energy of the system is thus defined as a sum of the self-consistent
Kohn–Sham energy terms as obtained from the chosen XC functional
1
13
H NMR (300 MHz,
75 MHz, O+[D
71.73 ppm (C=O); IR (ATR): ~n =3000–3500 (OH), 3070–2750 (CꢀH),
623 (C=O), 1500, 1429 (C=O), 1370 (CH
D
2
O): d=3.56 ppm (s, 6H, CH
3
);
C NMR
(
D
2
6
]DMSO): d=58.84 (CH ), 134.27 (COMe),
3
(
E
KSꢀDFT) and the semi-empirical correction Edisp [Eq. (6)].
1
1
9
4
), 1053, 1033 (CꢀO), 952,
3
ꢀ
1
E
¼ EKSꢀDFT þ Edisp
ð6Þ
32 cm (C=C); elemental analysis calcd (%) for Li
2
C
8
H
12
O
9
: C 36.11, H
DFTꢀD
.55; found: C 35.7, H 4.59.
All calculations were performed by using the default values of
C parameters for each species reported in the original article of Grimme
describing this method. A value of d=20 (dampening parameter) was
selected in order to specify the steepness of the dampening function
Eq. (5)]. A cutoff radius of 30 ꢅ for pair interactions was used to trun-
cate the summation over lattice vectors.
3
,6-Dihydroxy-2,5-dimethoxy-p-benzoquinone dilithium salt tetrahydrate
ij
6
(
3): Compound 2 (150 mg) was placed in a watch glass installed in a Petri
[
9]
dish containing hot water. The set was then covered and a purple powder
was recovered after 12 h. This powder was identified as the pure tetrahy-
drate phase by comparison of the powder XRD pattern with the pattern
simulated from the X-ray crystallographic data of 3. Single crystals of 3
[
6
To determine an optimal value for the s global scaling factor, a screening
were grown from a concentrated solution of 2 (20 mg) in water (1 mL)
1
of this parameter was tested on the known experimental system
Li DHDMQ·4H O within the range 0.52–0.75, the extreme values of
which correspond respectively to the corr-PBE-D*_0.52 (s =0.52) and
PBE-D* (s =0.75) methods (Figure S9 in the Supporting Information).
According to our results, b rises when s decreases, while the volume of
by slow evaporation in
a
partially covered watch glass. H NMR
2
2
(
(
1
300 MHz, D
2
O): d=3.56 ppm (s, 6H, CH
3
); IR (ATR): ~n =3000–3500
),
053, 1033 (CꢀO), 952, 932 cm (C=C); elemental analysis calcd (%) for
10: C 33.82, H 4.97; found: C 33.58, H 4.93.
,6-Dihydroxy-2,5-dimethoxy-p-benzoquinone dilithium salt (4): Com-
6
OH), 3070–2750 (CꢀH), 1623 (C=O), 1500, 1429 (C=O),1370 (CH
3
ꢀ
1
6
6
2 8 14
Li C H O
the unit cell undergoes a reverse trend (Figure S9a in the Supporting In-
formation). Due to the moderate absolute value reached for both unit-
3
pound 2 (2.64 g) was dissolved in distilled water (45 mL) and the solution
frozen in liquid nitrogen. Lyophilization was carried out for five days
before refilling the flask with argon. Residual water was completely re-
moved by subsequent annealing at 1808C under vacuum for 2 h. The
6
cell volume and monoclinic angle with s =0.52, the corr-PBE-D*_0.52
method was selected to provide an improvement of PBE-D*. In addition,
the interplane distance, d, is closer to the experimental one for s =0.52
6
(Figure S9b in the Supporting Information) and the stacking parameter
1
dark green, fluffy compound was stored in a glove box. H NMR
a is also better described by this value (minor relative error of ꢀ3.2%
(
(
1
300 MHz, D
2
O): d=3.56 ppm (s, 6H, CH
3
); IR (ATR): ~n =3000–3500
),
obtained for this s ).
6
OH), 3070–2750 (CꢀH), 1623 (C=O), 1500, 1429 (C=O), 1370 (CH
3
ꢀ1
053, 1033 (CꢀO), 952, 932 cm (C=C).
Computational details: All calculations were performed using the pseu-
dopotential VASP package. Full geometry optimization was achieved
through a conjugate gradient algorithm minimizing the Hellmann–Feyn-
[
34]
Acknowledgements
This work was supported by GCEP program (from Stanford University),
the Rꢀgion Picardie and the FEDER program. The authors deeply thank
Matthieu Courty for thermal measurements, Jꢀrꢉme Musqua for comple-
mentary experiments and Michꢁle Nelson for helpful assistance. C.F. and
G.B. thank the DSI-CCuB from the University of Bourgogne, the com-
puter center of the University Bordeaux 1 within the M3PEC intensive
calculations project, and the CINES from Montpellier for allowing us to
access their computer facilities. We gratefully acknowledge generous allo-
cations of computing time from the CINES.
man forces. For the Li
taking the experimental structure of the crystal at room temperature.
Subsequent calculations on solvent-free Li DHDMQ or partially dehy-
drated Li DHDMQ·3H O along with selected hypothetical crystal struc-
tures on lithium extraction (LiDHDMQ·4H O) were then performed
2 2
DHDMQ·4H O phase, calculations were done by
2
2
2
2
starting from the energy-minimized crystal structure. All minimizations
were considered complete when energies converged to better than 1ꢇ
ꢀ
ꢀ
5
1
1
0
0
eV per atom and maximum residual forces were lower than 1ꢇ
3
ꢀ1
eVꢅ . Projector augmented wave (PAW) potentials were used to
[
35]
describe the electron–ion interaction,
while the wave functions were
expanded in plane waves with energy below 520 eV. Brillouin zone sam-
[
36]
pling was performed by using the Monkhorst–Pack scheme, with a k-
point grid of 4ꢇ4ꢇ2.
&
12
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Li Deintercalation in a Tetrahydroxy-p-Benzoquinone Dilithium Salt
FULL PAPER
[
[
18] X.-Q. Zhu, C.-H. Wang, J. Org. Chem. 2010, 75, 5037–5047 and ref-
erences cited therein.
19] J. P. Perdew, A. Zunger, Phys. Rev. B 1981, 23, 5048–5079.
[
1] a) H. Ibrahim, A. Ilinca, J. Perron, Renewable Sustainable Energy
Rev. 2008, 12, 1221–1250; b) N. Armaroli, V. Balzani, Angew. Chem.
2
007, 119, 52–67; Angew. Chem. Int. Ed. 2007, 46, 52–66.
[20] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77,
3865–3868.
[21] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuse-
ria, L. A. Constantin, X. Zhou, K. Burke, Phys. Rev. Lett. 2008, 100,
136406–1–136406–4.
[22] Y. Zhang, W. Yang, Phys. Rev. Lett. 1998, 80, 890.
[23] G. Kresse, J. Furthmuller, Phys. Rev. B 1996, 54, 11169–11186.
[24] B. Civalleri, C. M. Zicovich-Wilson, L. Valenzano, P. Ugliengo, Crys-
tEngComm 2008, 10, 1693–1694.
[25] P. Ugliengo, C. M. Zicovich-Wilson, S. Tosoni, B. Civalleri, J. Mater.
Chem. 2009, 19, 2564–2572.
[26] M. D. King, W. D. Buchanan, T. M. Korter, Phys. Chem. Chem.
Phys. 2011, 13, 4250–4259.
[27] C. Frayret, E. I. Izgorodina, D. R. MacFarlane, A. Villesuzanne, A.-
L. Barrꢁs, O, Politano, D. Rebeix, P. Poizot, unpublished results.
[28] M. K. Aydinol, A. F. Kohan, G. Ceder, J. Power Sources 1997, 68,
664–668.
[29] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J.-M. Tarascon, Nature
2000, 407, 496–499; J. Cabana, L. Monconduit, D. Larcher, M. R.
Palacꢆn, Adv. Mater. 2010, 22, E170-E192.
[
[
2] P. Poizot, F. Dolhem, Energy Environ. Sci. 2011, 4, 2003–2019, and
references therein.
3] T. Suga, H. Nishide, in Stable Radicals - Fundamentals and Applied
Aspects of Odd-Electron Compounds, Chap. 14, Eds: R. G. Hicks,
WILLEY-VCH, Weinheim, 2010, 507–519 and references cited
therein.
[
[
[
[
[
[
4] H. Chen, M. Armand, M. Courty, M. Jiang, C. P. Grey, F. Dolhem,
J.-M. Tarascon, P. Poizot, J. Am. Chem. Soc. 2009, 131, 8984–8988.
5] H. Chen, M. Armand, G. Demailly, F. Dolhem, P. Poizot, J.-M. Tar-
ascon, ChemSusChem 2008, 1, 348–355.
6] H. Chen, P. Poizot, F. Dolhem, N. I. Basir, O. Mentrꢀ, J.-M. Taras-
con, Electrochem. Solid-State Lett. 2009, 12, A102–A106.
7] J. Xiang, C. Chang, M. Li, S. Wu, L. Yuan, J. Sun, Cryst. Growth
Des. 2008, 8, 280–282.
8] A. Kassam, D. J. Burnell, J. R. Dahn, Electrochem. Solid-State Lett.
2
011, 14, A22–A23.
9] S. Grimme, J. Comput. Chem. 2006, 27, 1787–1799.
[
[
[
10] H. S. Verter, J. Rogers, J. Org. Chem. 1966, 31, 987–988.
11] B. Eistert, G. Bock, Chem. Ber. 1959, 92, 1239–1246.
12] a) E. M. D. Keegstra, B.-H. Huisman, E. M. Paardekooper, F. J.
Hoogesteger, J. W. Zwikker, L. W. Jenneskens, H. Kooijman, A.
Schouten, N. Veldman, A. L. Spek, J. Chem. Soc. Perkin Trans. 2
[30] J. Rodriguez-Carvajal, Physica B+C 1993, 192, 55–69; Fullprof
available from http://www.ill.eu/sites/fullprof/, last accessed Nov.
2011.
1
996, 229–240; b) H. Bock, S. Nick, W. Seitz, C. Nꢃther, J. W. Bats,
[31] SADABS: Area-Detector Absorption Correction; Siemens Industri-
al Automation, Madison, WI, 1996.
[32] G. M. Sheldrick, SHELXTL, Suite vers. 5.1, Bruker AXS, Madison,
WI, 1998.
[33] V. Petricek, M. Dusek, L. Palatinus, JANA2006, Institute of Physics,
Academy of Sciences, Praha, Czech Republic, 2006.
[34] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558–561.
[35] P. E. Blçchl, Phys. Rev. B 1994, 50, 17953–17979.
[36] H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188–5192.
Z. Naturforsch., B: J. Chem. Sci. 1996, 51, 153–171.
13] H. R. Costantino, M. J. Pikal, Lyophilization of biopharmaceuticals,
American association of pharmaceutical scientists Press, Arlington,
VA, 2004.
14] L. Qian, H. Zhang, J. Chem. Technol. Biotechnol. 2011, 86, 172–184.
15] Y. J. Choi, B. S. Jung, D. J. Lee, J. H. Jeong, K. W. Kim, H. J. Ahn,
K. K. Cho, H. B. Gu, Phys. Scr. 2007, T129, 62–65.
[
[
[
[
16] a) D. Guy, B. Lestriez, D. Guyomard, Adv. Mater. 2004, 16, 553–
5
57; b) D. Guy, B. Lestriez, F. Boucher, D. Guyomard, J. Electro-
chem. Soc. 2006, 153, A679–A688.
17] S. Renault, J. Geng, F. Dolhem, P. Poizot, Chem. Commun. 2011, 47,
Received: December 5, 2011
Revised: March 23, 2012
[
2
414–2416.
Published online: && &&, 0000
Chem. Eur. J. 2012, 00, 0 – 0
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
13
&
ÞÞ
These are not the final page numbers!

P. Poizot et al.
Organic Electrode Materials
A.-L. Barrꢀs, J. Geng, G. Bonnard,
S. Renault, S. Gottis, O. Mentrꢁ,
C. Frayret, F. Dolhem,
P. Poizot* ...................... &&&& — &&&&
High-Potential Reversible Li Deinter-
calation in a Substituted Tetrahydroxy-
p-benzoquinone Dilithium Salt: An
Experimental and Theoretical Study
+
0
A new redox amphoteric system,
derived from the tetralithium salt of
tetrahydroxy-p-benzoquinone, reversi-
3 V versus Li /Li (see figure). A sol-
vated form of the studied salt
(Li DHDMQ·4H O) was also charac-
terized by X-ray diffraction and first-
principles calculations.
2
2
+
bly deintercalates one Li on charging
and has a practical capacity of about
ꢀ
1
1
00 mAhg at an average potential of
&
14
&
www.chemeurj.org
ꢄ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!