15
Charge ordering, symmetry and electronic structure issues and Wigner crystal structure of the quarter-filled band Mott insulators and high pressure metals d-(EDT-TTF-CONMe 2 ) 2 X, X ¼ Br and AsF 6 Leokadiya Zorina, ab Sergey Simonov, ab C ecile M ezi ere, a Enric Canadell, c Steve Suh, d Stuart E. Brown, d Pascale Foury-Leylekian, e Pierre Fertey, f Jean-Paul Pouget e and Patrick Batail * a Received 15th April 2009, Accepted 6th July 2009 First published as an Advance Article on the web 4th August 2009 DOI: 10.1039/b906287d We report on the synthesis and application of an internal chemical pressure to effectively control, and reduce, the Mott gap in the system d-(EDT-TTF-CONMe 2 ) 2 X, X ¼ Br, AsF 6 ; the detailed accounts of its Pmna, averaged room temperature structure and reversible phase transition at ca. 190 K towards a low temperature P2 1 /a structure; the synthesis of ( 13 C-EDT-TTF-CONMe 2 ) 2 Br, where one carbon atom of the inner double bond is 100% 13 C-enriched and single crystal 13 C solid state NMR spectroscopy and relaxation revealing that charge ordering occurs at room temperature and ambient pressure; the discovery of weak superstructure Bragg reflections in d-(EDT-TTF-CONMe 2 ) 2 Br and subsequent analysis of the superstructure symmetry and refinement of an exhaustive synchrotron radiation data set; suggesting an alternation at room temperature of neutral and oxidized molecules along both the stacking a and transverse b directions in orthorhombic, non-centrosymmetric space group P2nn, a CO pattern compatible with ferroelectricity. The charge disproportionation and long range order crystallization of the electron gas onto every other molecular site within a three- dimensional Wigner lattice is coupled to a concerted activation-deactivation of large collections of transverse C sp2 –H/O hydrogen bonds and an anti-phase, static modulation of the bromide anions displacements along b. Despite the occurrence of charge ordering, the stacks remain essentially uniform, in agreement with the rich low temperature Mott physics of the system. Introduction The ability to control what makes electrons (or holes) travel long distances, as fast as one thousandth of the speed of light (the Fermi velocity, 10 5 ms 1 ), over molecular sites along a metallic wire or, instead, become prevented to do so and stay put, localized onto the molecular backbone, is critical for the development of molecular circuitry and electronic nanodevices, and all comes down to playing with, and mastering, the interplay of electron correlations and Coulomb interactions. This interplay is recognized to be at the heart of the mechanism of metal-to- insulator transitions (MIT), one of the most fascinating key phenomena in condensed matter materials chemistry and physics whose approach was initiated by pioneering considerations by Wigner, Mott and Hubbard more than 50 years ago, 1–3 one which pervades over, and keep inspiring, much current inter- disciplinary research. 4 For example, systems with an odd number of electrons per site (sites here may be either discrete atomic entities or covalent molecular units instead) can undergo a Mott- Hubbard electron localization for large U/t ratios, where U is the on-site Coulomb repulsion and t is the transfer integral. 2,3 Elec- tron localization can also occur for a non-integer number of electrons per site in the presence of long range Coulomb repul- sions, of which the nearest neighbour’s interaction V is the most prominent, 5 leading to a so-called Wigner crystallization. While the study of such transitions in materials with open d and f electron shells has a long history, 6 MIT of similar nature were recently observed in low dimensional molecular conductors. 7 These materials, with their soft anisotropic lattices directed by a subtle balance of a series of weak intermolecular interactions such as hydrogen bonds, 8 are currently receiving much attention because the MIT can be tuned not primarily by doping but rather by physical 9,10 or chemical pressure, providing an avenue for bandwidth control of the Mott-Hubbard gap. 2,3,11 The importance of long range Coulomb interactions in organic conductors was established very early by the discovery of the so- called 4k F instability in the one dimensional (1D) charge transfer salt TTF-TCNQ, 12 the signature, as seen from the reciprocal space, of a 1D Wigner lattice of localized charges in direct space a Laboratoire de Chimie et Ing enierie Mol eculaire d’Angers, CNRS-Universit e d’Angers, 49045 Angers, France. E-mail: patrick. [email protected] b Institute of Solid State Physics, Russian Academy of Sciences, 142432 Chernogolovka, MD, Russia c Institut de Ci encia de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, E-08193 Bellaterra, Spain d Department of Physics and Astronomy, University of California, Los Angeles, USA e Laboratoire de Physique des Solides, CNRS-Universit e Paris-Sud, 91405 Orsay, France f Synchrotron SOLEIL, L’Orme des Merisiers - Saint Aubin, B.P. 48, 91192 Gif-sur-Yvette, France † Electronic supplementary information (ESI) available: Schemes S1–S3, Table S1. CCDC reference numbers 727931–727934. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/b906287d 6980 | J. Mater. Chem., 2009, 19, 6980–6994 This journal is ª The Royal Society of Chemistry 2009 PAPER www.rsc.org/materials | Journal of Materials Chemistry Published on 04 August 2009. Downloaded by University of Wales Aberystwyth on 29/03/2017 20:19:20. View Article Online / Journal Homepage / Table of Contents for this issue

Charge ordering, symmetry and electronic structure issues

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

PAPER www.rsc.org/materials | Journal of Materials Chemistry

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online / Journal Homepage / Table of Contents for this issue

Charge ordering, symmetry and electronic structure issues and Wignercrystal structure of the quarter-filled band Mott insulators and highpressure metals d-(EDT-TTF-CONMe2)2X, X ¼ Br and AsF6†

Leokadiya Zorina,ab Sergey Simonov,ab C�ecile M�ezi�ere,a Enric Canadell,c Steve Suh,d Stuart E. Brown,d

Pascale Foury-Leylekian,e Pierre Fertey,f Jean-Paul Pougete and Patrick Batail*a

Received 15th April 2009, Accepted 6th July 2009

First published as an Advance Article on the web 4th August 2009

DOI: 10.1039/b906287d

We report on the synthesis and application of an internal chemical pressure to effectively control,

and reduce, the Mott gap in the system d-(EDT-TTF-CONMe2)2X, X ¼ Br, AsF6; the detailed

accounts of its Pmna, averaged room temperature structure and reversible phase transition at ca. 190 K

towards a low temperature P21/a structure; the synthesis of (13C-EDT-TTF-CONMe2)2Br, where one

carbon atom of the inner double bond is 100% 13C-enriched and single crystal 13C solid state NMR

spectroscopy and relaxation revealing that charge ordering occurs at room temperature and ambient

pressure; the discovery of weak superstructure Bragg reflections in d-(EDT-TTF-CONMe2)2Br and

subsequent analysis of the superstructure symmetry and refinement of an exhaustive synchrotron

radiation data set; suggesting an alternation at room temperature of neutral and oxidized molecules

along both the stacking a and transverse b directions in orthorhombic, non-centrosymmetric space

group P2nn, a CO pattern compatible with ferroelectricity. The charge disproportionation and long

range order crystallization of the electron gas onto every other molecular site within a three-

dimensional Wigner lattice is coupled to a concerted activation-deactivation of large collections of

transverse Csp2–H/O hydrogen bonds and an anti-phase, static modulation of the bromide anions

displacements along b. Despite the occurrence of charge ordering, the stacks remain essentially

uniform, in agreement with the rich low temperature Mott physics of the system.

Introduction

The ability to control what makes electrons (or holes) travel long

distances, as fast as one thousandth of the speed of light

(the Fermi velocity, 105 m s�1), over molecular sites along

a metallic wire or, instead, become prevented to do so and stay

put, localized onto the molecular backbone, is critical for the

development of molecular circuitry and electronic nanodevices,

and all comes down to playing with, and mastering, the interplay

of electron correlations and Coulomb interactions. This interplay

is recognized to be at the heart of the mechanism of metal-to-

insulator transitions (MIT), one of the most fascinating key

aLaboratoire de Chimie et Ing�enierie Mol�eculaire d’Angers,CNRS-Universit�e d’Angers, 49045 Angers, France. E-mail: [email protected] of Solid State Physics, Russian Academy of Sciences, 142432Chernogolovka, MD, RussiacInstitut de Ci�encia de Materials de Barcelona (ICMAB-CSIC), Campusde la UAB, E-08193 Bellaterra, SpaindDepartment of Physics and Astronomy, University of California, LosAngeles, USAeLaboratoire de Physique des Solides, CNRS-Universit�e Paris-Sud, 91405Orsay, FrancefSynchrotron SOLEIL, L’Orme des Merisiers - Saint Aubin, B.P. 48,91192 Gif-sur-Yvette, France

† Electronic supplementary information (ESI) available: Schemes S1–S3,Table S1. CCDC reference numbers 727931–727934. For ESI andcrystallographic data in CIF or other electronic format see DOI:10.1039/b906287d

6980 | J. Mater. Chem., 2009, 19, 6980–6994

phenomena in condensed matter materials chemistry and physics

whose approach was initiated by pioneering considerations by

Wigner, Mott and Hubbard more than 50 years ago,1–3 one

which pervades over, and keep inspiring, much current inter-

disciplinary research.4 For example, systems with an odd number

of electrons per site (sites here may be either discrete atomic

entities or covalent molecular units instead) can undergo a Mott-

Hubbard electron localization for large U/t ratios, where U is the

on-site Coulomb repulsion and t is the transfer integral.2,3 Elec-

tron localization can also occur for a non-integer number of

electrons per site in the presence of long range Coulomb repul-

sions, of which the nearest neighbour’s interaction V is the most

prominent,5 leading to a so-called Wigner crystallization. While

the study of such transitions in materials with open d and f

electron shells has a long history,6 MIT of similar nature were

recently observed in low dimensional molecular conductors.7

These materials, with their soft anisotropic lattices directed by

a subtle balance of a series of weak intermolecular interactions

such as hydrogen bonds,8 are currently receiving much attention

because the MIT can be tuned not primarily by doping but rather

by physical9,10 or chemical pressure, providing an avenue for

bandwidth control of the Mott-Hubbard gap.2,3,11

The importance of long range Coulomb interactions in organic

conductors was established very early by the discovery of the so-

called 4kF instability in the one dimensional (1D) charge transfer

salt TTF-TCNQ,12 the signature, as seen from the reciprocal

space, of a 1D Wigner lattice of localized charges in direct space

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

accompanied by a concomitant lattice deformation.13 The 4kF

instability was found to be enhanced in radical cation or anion

salts, with a 2:1 or 1:2 stoichiometry, respectively, where the net

charge of conducting mixed valence molecular stacks is balanced

by neighboring strings of closed shell anions or cations which

here do not provide for mobile charge to screen the Coulomb

repulsions.14 In these salts, whose (electron or hole like) band is

quarter filled, the insulating ground state is achieved for

reasonable U/t and V/t ratios.15–17 With one carrier every two

sites, the 4kF insulating ground state is achieved by a doubling of

the chain repeat periodicity, a consequence of the localization of

the carriers either on one intermolecular bond out of two between

successive molecules (hence, here all molecules are in a mixed

valence state), or rather right onto one molecular site only out of

two successive molecular sites. The first case corresponds to the

formation of a mixed valence Mott dimer associated with a stack

dimerization, as observed for example at the 335 K MIT in

MEM(TCNQ)2.18 The second case, where charge rich molecules

alternate with charge poor molecules along the stack, corre-

sponds to a charge ordering (CO) or charge dis-

proportionation.6d,e This charge disproportionation is expected

to yield a doubling of the number of NMR lines at the CO

transition, as indeed observed in (DI-DCNQI)2Ag19 as well as for

the (TMTTF)2X series with X ¼ PF6, SbF6, etc.20 For

(DI-DCNQI)2Ag it has been found that the 210 K transition

corresponds to the divergence of a 1D 4kF structural instability,21

yet the 50 K Wigner structure consists of different stacks where

the charges are localized either on the sites, or on the bonds

(dimers) or in an intermediate position.19c A single type of stack is

identified in the CO state of (TMTTF)2X, but as this stack was

not uniform to start with at room temperature but already

slightly yet significantly dimerized at higher temperature, the low

temperature structure is postulated to consist of a mixture of 4kF

bond and site orders, in agreement with the observation of

ferroelectricity due to the removal of all inversion centers at

CO.22 Although the Extended Hubbard Hamiltonian produces

the CO ground state,3,16 the role of coupling to the lattice is

necessary to describe the complete temperature/pressure phase

diagram of materials such as (TMTTF)2X.23,24 The counterion

sublattice is likely also important,22,25–27 but there are no sup-

porting structural data. In contrast, the present report will

conclude there is a strong involvement of the lattice in producing

the CO state at ambient pressure.

In this context, the report in 2003 of the quasi-one-dimen-

sional, 3⁄4 -filled band Mott insulator d-(EDT-TTF-CON-

Me2)2AsF6 (3⁄4 -filled with electrons or ¼-filled with holes),28

where there is no centre of symmetry between the molecular units

along the stack whose strict uniformity is imposed by glide plane

symmetry, has triggered an in-depth exploration of its rich phase

diagram.29 The Mott gap in d-(EDT-TTF-CONMe2)2AsF6, an

insulator at ambient pressure and room temperature, decreases

upon hydrostatic pressure up to 20 kbar where it is suppressed

and the system becomes metallic. Here, we report on d-(EDT-

TTF-CONMe2)2Br and demonstrate that substitution of AsF6�

for Br�, whose volume is 76 A3 smaller, builds up an internal

chemical pressure which results in a shrinking of the stacking axis

and, conversely, an increase of the transfer integral along the

stack, as discussed. This crystal engineering approach effectively

allows for a control and a sizeable decrease of the Mott gap, as

This journal is ª The Royal Society of Chemistry 2009

the localization in d-(EDT-TTF-CONMe2)2Br is suppressed at

a smaller hydrostatic pressure with a net gain of as much as

7 kbar, as reported independently.29

Then, for the reasons described previously, it is likely that in

the insulating state the charge density will no longer remain

uniform along the stacks, that is, charge disproportion is

expected to occur. Therefore, in order to search for CO, we set

out to synthesize (13C-EDT-TTF-CONMe2)2Br, where one

carbon atom of the inner double bond is 100% 13C-enriched, in

order to perform high resolution 13C solid state NMR experi-

ments, reported here, and IR reflectivity and Raman experi-

ments, reported independently,30 which both reveal that CO and

Wigner crystallization occur already at room temperature and

ambient pressure in these salts. This prompted the present

complementary investigation of the crystal structure of d-(EDT-

TTF-CONMe2)2Br using a combination of X-ray diffuse scat-

tering experiments and synchrotron radiation leading to the

discovery of superstructure Bragg reflections which in turn

imposes two independent molecules to co-exist in a double unit

cell. An assessment of the symmetry base and the successful

refinement of the ambient pressure, room temperature synchro-

tron data are further discussed herein, allowing for a complete

description of the three-dimensional pattern of CO, here in the

non-centrosymmetric space group P2nn, which proved to

confirm earlier models.

Experimental

Synthesis of 13C-enriched 4,5-ethylenedithio-40-(N,N-dimethylcarbamoyl)

tetrathiafulvalene

13C-enriched 4,5-ethylenedithio-40-(N,N-dimethylcarbamoyl)-

tetrathiafulvalene is prepared with one important modification

of the procedure reported earlier.31 Since the synthesis of the13C-enriched 4,5-ethylenedithio-2-thioxo-1,3-dithiole necessary

to obtain the key molecule 13C-EDT-TTF-(CO2Me)2 is not

practical because it requires large amounts of CS2, we chose to

prepare its oxo equivalent instead and then adjust the coupling

conditions to obtain a correct yield. The enriched precursor

4,5-ethylenedithio-2-oxo-1,3-dithiole is prepared as described via

the Larsen-Lenoir route (ESI,† Scheme S1).32

13C-enriched potassium O-(isopropyl)dithiocarbonate

3.64 g (0.065 mol) of potassium hydroxide are dissolved in 35 mL

of 2-propanol distilled on CaO with heating. The hot solution is

filtered into a 100 mL round bottom flask and cooled to 0 �C with

an ice bath. 13C-enriched CS2 from Sigma-Aldrich Company

(5 g, 0.065 mol) is added dropwise by a syringe with stirring.

After 30 minutes the potassium xanthate is collected by filtration

and washed with four 10 mL portions of diethyl ether and dried

under pressure. Recrystallization from ethanol affords 6.88 g of

the desired compound (Mp 265 �C).

Coupling procedure

The coupling of the 13C-enriched 4,5-ethylenedithio-2-oxo-1,3-

dithiole moiety with the 4,5-dimethyloxycarbonyl-2-oxo-1,3-

dithiole has been optimized and the best yields are obtained using

five equivalents of the diester moiety (ESI,† Schemes S2 and S3).

J. Mater. Chem., 2009, 19, 6980–6994 | 6981

Fig. 1 Atom labelling and molecular structure for the averaged, room

temperature orthorhombic structure of d-(EDT-TTF-CONMe2)2Br

(DIAMOND diagram with 50% thermal ellipsoids). Atoms S5, C7, C8

and C11 are disordered about the mirror plane in Pmna.

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

13C-EDT-TTF-CONMe2

13C-EDT-TTF-COCl (800 mg, 2.23 mmol) dissolved in dry THF

(32 mL) was added dropwise to a 2M THF solution of dime-

thylamine (5.6 mL, 11.2 mmol) diluted in dry THF (32 mL).

After stirring for one hour the solution is filtered and evaporated.

The orange solid is recrystallized in CH3CN to afford the

compound as orange platelets (620 mg, 76%), mp 166 �C; anal-

ysis found: C, 36.28; H, 2.77; N, 3.91. 12C1013C1H11ONS6

requires: C, 36.31; H, 3.02; N, 3.82%; NMR (500 MHz, DMSO-

d6) dH 7.28 (s, 1H, C]CH), 3.38 (s, 4H, CH2–CH2), 3.00 (s, 6H,

CH3); dC 160.36 (C]O), 130.23 (C–CO), 124.58 (]CH), 112.82–

116.91 (C]C), 103.97 (13C]C), 35.90 (CH3) and 29.58 (CH2–

CH2); m/z (EI) 366 (M+, 100%), 338 (M+ � CH2 � CH2, 50%).

Electrocrystallization experiments. (EDT-TTF-CONMe2)2Br

Black, elongated crystalline platelets were grown at 25 �C upon

constant current (1.5 mA) oxidation at a Pt wire electrode (length

2 cm, diameter 1 mm) of a dichloromethane/acetonitrile 1/1

solution of EDT-TTF-CONMe2 (5 mg) containing PPh4Br

(35 mg) as electrolyte. The crystals were harvested on the anode

after one week.

(13C-EDT-TTF-CONMe2)2AsF6

Black, elongated crystalline platelets were grown at 30 �C upon

constant current (0.5 mA) oxidation at a Pt wire electrode (length

2 cm, diameter 1 mm) of a 1,1,2-trichloroethane solution (24 mL)

of 13C-EDT-TTF-CONMe2 (10 mg) containing TBAAsF6

(100 mg) as electrolyte. The crystals were harvested on the anode

after two weeks.

(13C-EDT-TTF-CONMe2)2Br

Black, elongated crystalline platelets were grown at 20 �C upon

constant current (3 mA) oxidation at a Pt wire electrode (length 2

cm, diameter 1 mm) of a dichloromethane/acetonitrile 1/1 solu-

tion (12 mL) of 13C-EDT-TTF-CONMe2 (10 mg) containing

PPh4Br (70 mg) as electrolyte. The crystals were harvested on the

anode after one week.

Standard X-ray diffraction experiments

Experimental X-ray structural data were collected on a d-(EDT-

TTF-CONMe2)2Br crystal mounted on a glass fiber using

a Bruker Nonius KappaCCD diffractometer with mono-

chromatized MoKa radiation (l ¼ 0.71073A, graphite mono-

chromator) equipped with an Oxford Cryosystems cryostream

cooler. Preliminary measurements of unit cell parameters were

performed in the temperature range of 300–100 K with 10 degree

steps. The as-grown crystal has an orthorhombic unit cell (space

group Pmna) characterized by sharp discrete Bragg reflections at

room temperature. Significant changes in the diffraction patterns

occur upon lowering the temperature. The former sharp reflec-

tions broaden upon reaching 200 K to the point where two

different systems of reflections are identified at 180 K and below

(down to 100 K). The latter define two equivalent monoclinic

lattices, as revealed by the DIRAX program,33 both in space

group P21/a. This qualifies an orthorhombic-to-monoclinic

phase transition accompanied by twinning.

6982 | J. Mater. Chem., 2009, 19, 6980–6994

In order to clarify the nature of the reversible phase transition,

data were collected at room temperature and 150 K for the

orthorhombic and monoclinic phases, respectively, by

a combined 4- and u-scan method. Diffraction intensities from

both twin domains obtained at 150 K were integrated using the

twinning option in the EVALCCD package of programs34 and

combined into SHELX HKLF5 file with additional indexes 1 or

2 marking the reflections from one or another twin cell, respec-

tively. Analytical absorption correction of experimental intensi-

ties taking into account the crystal shape and size (prism

0.05 � 0.23 � 0.04 mm) was applied for both phases using the

WINGX package of programs.35

The structures were solved by a direct method followed by

Fourier syntheses and refined by a full-matrix least-squares

method in an anisotropic approximation for all non-hydrogen

atoms using the SHELX-97 programs.36 The twin relationship

was taken into account in the HKLF5 refinement of the low-

temperature data with SHELXL-97.37 The twin fraction was

refined to a value of 0.4723(4).

Orientational disorder affects one amide methyl group and the

outer ethylene group at room temperature (Fig. 1). A slight

disorder of the ethylene group (10% of the alternative position)

remains in the monoclinic phase at 150 K. H atoms in both

phases were introduced in calculated positions with isotropic

displacement parameters fixed at 120% (or 150% for –CH3

groups) of the corresponding parameters of the attached

C atoms. Torsion angles of idealized hydrogen positions in the

–CH3 groups were refined from electron density (by HFIX 137

instruction). Details of unit cell data, data collection and

refinement are summarized in Table 1.†

A similar temperature-dependent study performed for one

single crystal of the isostructural salt, d-(EDT-TTF-CON-

Me2)2AsF6, concludes that the very same twinning occurs in the

low-temperature monoclinic phase. The complete experimental

data set was collected at 100 K. Twin data integration combining

intensities of both twin domains into the HKLF5 file as well as

structure solution and refinement were conducted with proce-

dures identical to those used for d-(EDT-TTF-CONMe2)2Br at

150 K.

X-Ray diffuse scattering experiments

The diffuse scattering study was performed using copper radia-

tion (l ¼ 0.1542 nm) issued from a rotating anode equipped with

a confocal multilayer monochromator. The investigation was

first performed with the so-called fixed film-fixed crystal method

in order to detect any weak X-ray diffuse scattering. Then,

This journal is ª The Royal Society of Chemistry 2009

Table 1 Crystal data, data collection and refinement details for d-(EDT-TTF-CONMe2)2X (X ¼ Br, AsF6)

X ¼ Br X ¼ AsF6

l/A 0.71073 0.71073 0.92072 0.71073Temperature/K 295(2) 150(2) 293(2) 100(2)Chemical formula C22H22BrN2O2S12 C22H22BrN2O2S12 C22H22BrN2O2S12 C22H22AsF6N2O2S12

Molecular weight 811 811 811 920Crystal system Orthorhombic Monoclinic Orthorhombic Monoclinica/A 7.1097(4) 6.9762(9) 7.1126(5) 7.0157(8)b/A 6.5049(9) 6.480(1) 13.0250(7) 6.422(1)c/A 32.691(3) 32.671(9) 32.7590(13) 35.465(5)a/� 90 90 90 90b/� 90 90 90 90g/� 90 92.45(2) 90 92.26(1)V/A3 1511.9(3) 1475.6(5) 3034.8(3) 1596.6(3)Space group, Z Pmna, 2 P21/a, 2 P2nn, 4 P21/a, 2F(000) 822 822 1644 926rcalc./g$cm�3 1.782 1.825 1.775 1.914m/cm�1 22.16 22.71 22.61 19.15Tmin, Tmax 0.732, 0.922 0.735, 0.920 0.467, 0.918 0.748, 0.9152qmax/� 60.0 60.4 95.4 64.1Reflections collected 18647 37386 42335 46120Independent reflections 2333 — 12895 —Rint 0.067 — 0.077 —No. of parameters 134 189 356 208GooF on F2 1.002 1.002 1.006 1.012R1 [I > 2s(I)] 0.0331 0.0571 0.0521 0.0784wR2[I > 2s(I)] 0.0599 0.1274 0.1418 0.1832

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

accurate measurements of the superstructure reflection intensity

and of the possible forbidden reflections were performed using

a four-circle diffractometer. The set-up was equipped with

a closed-cycle helium cryostat operating from 300 K down to

15 K. For the fixed film-fixed crystal technique, it was possible to

investigate the temperature range 300–1.8 K using a home-made

cryocooler equipped with a Joule-Thomson He closed-cycle third

stage. Three single crystals (�1 mm � 0.1 mm2) were used in the

experiments.

Synchrotron experiments at SOLEIL

X-Ray synchrotron data were collected at the Cristal beamline

using a Newport 4-circle diffractometer equipped with a 2D

CCD Ruby detector from Oxford Diffraction. A 13.466 keV

energy was chosen (0.92072 A) just below the Br absorption edge

in order to reinforce the anomalous signal of the Br atom

(f0 ¼ �7.74289 e�). 4- and u-scans were collected using the

oscillation method (1�/frame) at four different detector positions

(2qD¼�26� and 2qD¼�55�) with 1 s/� and 3 s/� exposure times,

respectively. During these data collections, the incident beam

was attenuated to limit the detector saturation. However, an

extra data set was collected without attenuation of the incident

beam to improve the signal of weak reflections. All the frames

were then processed using the CrysAlis software suite (integra-

tion of the intensities, incident beam monitor corrections, frame

scaling, empirical absorption corrections and reciprocal space

layer reconstructions).38

The structure was solved and refined in different space groups

with SHELX-9736 using special dispersion and absorption coef-

ficients for all chemical elements for a synchrotron wavelength of

0.92072 A (the DISP instruction was applied). A direct method

followed by Fourier syntheses and a full-matrix least-squares

method in an anisotropic approximation for all non-hydrogen

This journal is ª The Royal Society of Chemistry 2009

atoms were used for the structure solution and refinement,

respectively. Results in space group P2nn were considered as the

best description of the data, as discussed below. Large thermal

parameters are observed for one C atom only within the ethylene

outer group of each EDT-TTF-CONMe2 molecule, pointing to

two essentially equiprobable positions with anisotropic

displacement parameters restrained to be equal. Orientational

disorder affects some other atoms (S, Me groups), yet with lesser

thermal anisotropy and two atom sites could not be properly

distinguished. Hydrogen atoms were introduced as described for

the standard X-ray diffraction data. Details of unit cell data, data

collection and refinement are summarized in Table 1.†

Single crystal 13C NMR

The 13C spin-labelled samples with Br and AsF6 counterions were

studied over a range of temperatures (4–300K) and with a field

strength of 10 T oriented in the bc plane. The range of temper-

atures is accessed using a standard variable temperature cryostat

inserted into a cold-bore superconducting magnet. The rf tuning

coil was wound specifically for the small single crystals used in

this study so as to optimize signal/noise, then mounted onto

a rotatable platform. Evidence for two inequivalent environ-

ments, consistent with charge disproportionation, was obtained

from spectroscopy and spin–lattice relaxation measurements.

Electronic structure

The tight-binding band structure calculations used an extended

H€uckel type Hamiltonian.39 The off-diagonal matrix elements of

the Hamiltonian were calculated according to the modified

Wolfsberg-Helmholz formula.40 A basis set composed of double-

z Slater type orbitals for all atoms was used. The exponents and

ionization potentials (eV) employed were 1.817 and �20.0 for S

J. Mater. Chem., 2009, 19, 6980–6994 | 6983

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

3s, 1.817 and�13.3 for S 3p, 1.626 and�21.4 for C 2s, 1.625 and

�11.4 for C 2p, 1.95 and �26.0 for N 2s, 1.95 and �13.4 for N

2p, 2.275 and �32.3 for O 2s, 2.275 and�14.8 for O 2p, 1.30 and

�13.6 for H 1s.

Results and discussion

The synthesis of 13C-EDT-TTF-CONMe2, described in Schemes

S1–S3 (ESI†), was adapted from earlier procedures31 for EDT-

TTF-CONMe2. Electrocrystallization of the p-donor molecule

on a platinum wire electrode provides elongated, ruler-like

crystals appropriate for X-ray, transport, ESR, IR and Raman

and solid state NMR experiments.

The high and low temperature, averaged Pmna and P21/a

structures, both have uniform stacks imposed by glide plane

symmetry

Standard X-ray diffraction single crystal determinations revealed

the room and low temperature structures of d-(EDT-TTF-

CONMe2)2Br where discrete anions are interspersed between

successive radical cation slabs along c, as exemplified in the two

projections of the 150 K monoclinic structure shown in Fig. 2.

The unit cell contains one bromine anion on an inversion centre

and one independent donor molecule located in a general posi-

tion in the low-temperature monoclinic phase or on the mirror

plane in the room temperature orthorhombic form, some of the

Fig. 2 Projections of the averaged, P21/a structure of the low-temper-

ature monoclinic form of d-(EDT-TTF-CONMe2)2Br at 150 K along the

a-direction (a) and b-direction (b). Note that the locations in the lattice of

atoms C1, which will eventually be 13C-enriched (see text), have been

single out by coloring the atom in black.

6984 | J. Mater. Chem., 2009, 19, 6980–6994

atoms (methyl and ethylene groups) being disordered about the

m-plane, as shown in Fig. 1. These are denoted as the averaged

Pmna and P21/a structures, on account that, in each one of them,

the single independent molecule bears the formal charge 0.5+.

It is important to note that a defining, characteristic feature of

the average structures of this 2:1 mixed-valence system is the

presence of exactly uniform stacks of half-charged p-conjugated

molecules generated by the repetition of any donor molecule by

the crystallographic glide plane a. Adjacent molecules in the

stack adopt a criss-cross mode of overlap and are 3.488(1) and

3.555(1) A apart in the monoclinic and orthorhombic phases,

respectively, i.e., a/2. Within the stack, CEt–H/O hydrogen

bonds (Fig. 2b, Table 2) are eventually strong enough to provide

additional, slightly shortened van der Waals S/C as well as

C/C intrastack contacts, while all S/S intrastack distances

exceed 3.7 A. Coplanar molecules from neighboring stacks are

connected by cell translation along b, creating a d-type layer, and

interact by one Csp2–H/O contact and three S/S contacts, two

of which are very short (ESI,† Table S1).

Likewise, the ethylene and methyl groups are hydrogen

bonded to bromide atoms (Fig. 2), the CEt–H/Br hydrogen

bonds being significantly shorter than CMe–H/Br contacts

(Table 2). Such dependence of hydrogen bond strength on the

nature of the carbon atom is a common tendency for C–H/X

contacts.8 One of the two independent methyl C-atoms in EDT-

TTF-CONMe2 (C11 in Fig. 1) does not form any hydrogen

bonds at 150 K despite a C/O separation as short as 3.254(2) A

from the carbonyl oxygen atom. Hence, C11 is more loosely

packed in the crystal and exhibits the largest thermal parameter

at low temperature in comparison with all other atoms.

Structural phase transition accompanied by twinning

The lattice distortion identified in d-(EDT-TTF-CONMe2)2Br

below 190 K proved to be associated with an orthorhombic-to-

monoclinic phase transition. The occurrence of twinning

complicates the analysis of this transition. Two sets of Bragg

reflections are distinguished in the diffraction patterns of the

twinned monoclinic crystals at 180 K and below. This structural

phase transition is found to be entirely reversible upon warming

the crystal back to room temperature. Extensive unit cell

parameter determinations performed over a wide temperature

range demonstrate that there is no evidence for any cell volume

discontinuity or temperature hysteresis. Therefore, this is

a second order phase transition.

The temperature dependence of the monoclinic g angle, shown

in Fig. 3, qualifies g as an order parameter characteristic of the

transition. Note that the lattice distortion is also revealed by

a change of slope of the linear temperature dependence of the

ESR linewidth and the narrowing of the latter at 190 K, both for

the AsF6� and Br� salts,41 as well as by a deviation at this

temperature of the spin susceptibility from its high temperature

Bonner-Fisher dependence.41 The separation in terms of the g

angle between the two twin parts becomes more appreciable

upon further cooling, as illustrated in Fig. 3.

The nature of the twinning is an important question. One of

two ways is possible: either the crystals are twinned already at

room temperature or twin micro-domains form during the phase

transition. Note that, as exemplified in Fig. 4a, several crystals are

This journal is ª The Royal Society of Chemistry 2009

Table 2 C–H/O donor/donor and C–H/Br donor/anion hydrogen contacts in d-(EDT-TTF-CONMe2)2Br at room temperature and 150 K

Contact type

Geometry of D–H/A contactsa

150 K 295 K

293 K (double lattice)

H-bond donor I0 H-bond donor II�+

CEt–H/O (intrastack) 2.32 [3.26A, 163�] 2.39 [3.29A, 154�] 2.45 [3.38A, 162�] 2.39 [3.33A, 163�]2.48 [3.37A, 153�] 2.54 [3.46A, 157�] 2.56 [3.30A, 134�] 2.67 [3.54A, 150�]

Csp2–H/O (interstack) 2.89 [3.80A, 165�] 2.93 [3.84A, 165�] 2.96 [3.86A, 165�] 2.91 [3.82A, 167�]

CEt–H/Br 2.73 [3.70A, 175�] 2.76 [3.72A, 173�] 2.84 [3.72A, 152�] 2.78 [3.75A, 175�]2.95 [3.64A, 129�] 2.98 [3.68A, 130�] 3.19 [3.86A, 127�] 3.05 [3.71A, 127�]

CMe–H/Br 3.21 [4.10A, 155�] 3.38 [4.21A, 145�] 3.32 [4.15A, 145�] 3.36 [4.19A, 146�]3.28 [4.19A, 159�] 3.40 [4.09A, 130�] 3.37 [4.22A, 148�] 3.42 [4.27A, 148�]

a Values for each donor–H/acceptor contact are given in following order: H/A distances, A [D/A distances, D–H–A angles].

Fig. 3 Change in the g lattice angle across the structural phase transi-

tion. Blue rhombuses and red circles below the transition are data points

for the two different twin domains in the monoclinic phase.

Fig. 4 Different shapes of the crystals of d-(EDT-TTF-CONMe2)2Br.

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

composed of two identical fragments separated by a long thin

crystal plate in between, as though the inner thin plate is actually

the twin boundary. However, X-ray experiments on those crystals

as well as on crystal cut-offs along the assumed twin boundary, or

alternatively on crystals of strictly rectangular shape (Fig. 4b),

always yielded the orthorhombic symmetry at room temperature

and the same twin diffraction pattern at low temperature (the full,

final X-ray data set was obtained from a rectangular crystal).

The observation of positional disorder in the orthorhombic

form of the crystal also suggests possible twinning at room

temperature. As mentioned earlier, in the orthorhombic phase

This journal is ª The Royal Society of Chemistry 2009

the asymmetric donor molecule lies flat on a mirror plane, hence

the methyl and ethylene groups are disordered over two equi-

probable sites about this plane, as exemplified in Fig. 1 and 5a. In

other words, the donor disorder in the structure of the high-

temperature phase is imposed by the existence of the mirror

plane; yet the choice of another, non-centrosymmetric space

group (with possible twinning) is ruled out by taking into

account the nature of the observed phase transition. In agree-

ment with a second order transition, the symmetry group of the

orthorhombic phase should necessarily be a super-group of that

of the monoclinic phase.42 Analysis of systematic absences makes

for the unambiguous assignment of P21/a as the space group for

the low temperature monoclinic form. The twin structure

refinement based on this space group delivers a high quality

structure without any indication of the disorder caused by

symmetry elements 21 or a. Thus, the centrosymmetric space

group Pmna, a supergroup of P21/a, was chosen for the high-

temperature orthorhombic average structure.

Note finally that the twinning operation is deciphered as the

result of a two-fold rotation about a in the direct lattice (Fig. 5b),

and that this two-fold axis is an integral part of the complete set

of symmetry elements of P2/m2/n21/a. Hence, no evidence for the

presence of a twinning similar to that found below the phase

transition is expected at room temperature. Across the phase

transition, although the 2x symmetry is lost at the macroscopic,

full length-scale of the crystal, it is kept locally as a twin opera-

tion. Temperature and pressure-induced structural transitions

are susceptible to this kind of twinning and even more so for

molecular compounds whose structure is directed by non-cova-

lent intermolecular interactions such as hydrogen bonds and van

der Waals interactions and where alternative orientations of

molecules in the single crystal and across the twin boundary can

be energetically similar.43 As all the crystals tested (more than

ten, from different syntheses) have repeatedly shown the same

twinning across the phase transition, d-(EDT-TTF-CON-

Me2)2Br appears to be another typical example of concomitant

twinning and structural transition.

As shown in Fig. 5b, the orientational disorder affecting the

methyl groups in the room temperature orthorhombic phase has

vanished in the monoclinic structure at 150 K, as any of the two

twin domains now selectively contains one or the other methyl

group orientations. Note that a similar situation applies for the

ethylene-end groups albeit each of the two twin domains still

J. Mater. Chem., 2009, 19, 6980–6994 | 6985

Fig. 5 (a) d-Layer of the orthorhombic Pmna phase viewed along c.

(b) The twinned, monoclinic structure of the low-temperature P21/

a phase finds its origin in a stacking fault at the domain ac-interface in

keeping with the local two-fold axial symmetry about a. Two twin

domains are shown in different colours. Note here that the methyl and

ethylene groups are ordered in the low temperature, monoclinic structure

with a neat partition inside individual twin domains. That is, the outer,

ordered ethylene C–C bonds are parallel inside each twin domain while

the spatial orientations of these bonds differ from one domain to the

other. The same applies to all ordered C–Me bonds.

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

remains affected by some small amounts (like 10%) of disorder.

Most likely, some amounts of disorder remain because the

transition appears not to be completed at 150 K, yet the data in

Fig. 3 suggest that a fully ordered structure is expected at lower

temperatures. The two twin domains are almost equiprobable at

150 K as the refined partition is 53% and 47%.

A similar twinning also affects the low temperature monoclinic

phase of d-(EDT-TTF-CONMe2)2AsF6. Here, analysis of single

crystal X-ray diffraction data collected at 100 K reveals that the

Table 3 Chemical pressure effect in the series d-(EDT-TTF-CONMe2)2X (X

X ¼ Br�

Temperature/K 150 295a/A 6.9762(9) 7.1097(4)b/A 6.480(1) 6.5049(9)c/A 32.671(9) 32.691(3)a/� 90 90b/� 90 90g/� 92.45(2) 90V/A3 1475.6(5) 1511.9(3)

6986 | J. Mater. Chem., 2009, 19, 6980–6994

molecules are fully ordered and that, within the two twin

domains, the same difference in spatial orientations of the methyl

and ethylene groups prevails. The twin domain ratio amounts

to 62:38.

Chemical pressure

Several monoclinic phases with a d-type, criss-cross pattern of

overlap within the stack and similar lattice parameters have been

identified amongst BEDT-TTF salts and analyzed using

CSD.44,45 Note that they all have other symmetry groups with

their monoclinic axis always parallel to the d-slab while it is here

normal to the layers. Hence, the sole isostructural salt known to

date is d-(EDT-TTF-CONMe2)2AsF6.28

The structures of the Br� and AsF6� salts and their trans-

formation at low temperatures are quite similar yet the donor

stacks in d-(EDT-TTF-CONMe2)2Br are both more compact

than those in d-(EDT-TTF-CONMe2)2AsF6 and exhibit more

and shorter S/C and C/C intra-stack contacts as a result of the

chemical pressure which sets in upon installing the bromide

anion, whose effective volume is 76 A3 smaller than that of AsF6�

(Table 3). Note that, conversely, the cell dimension in the

direction b transverse to the stacks increases in the Br� salt so

that the effective area of the donor layer in the ab plane barely

changes by 1%. Accordingly, side-by-side S/S and H/O

contacts are slightly shorter in the AsF6� salt. The largest

compression of the lattice upon changing the anion from AsF6�

to Br� occurs along the direction across the layers. Remarkably

here, AsF6� is found to be strongly disordered upon cooling the

crystal from room temperature down to 100 K while Br� is

ordered in the whole temperature range, hence providing for

a better fit to the donor packing.

Uniform stacks and chemical pressure effect both nicely

exemplified in the compared electronic structures of Br� and AsF6�

salts

Because of the symmetry of the crystal, there are only two

different types of donor/donor interactions in the donor layers

of these salts: one along the stack (//) and one along the

perpendicular direction (t). Although the S/S contacts along

the stacks are all longer than 3.7 A whereas there are some quite

short contacts along the interstack direction (around 3.3 A or

shorter; see ESI† Table S1), the HOMO/HOMO interaction

along the stacks is considerably stronger. For instance, the

calculated transfer integrals for d-(EDT-TTF-CONMe2)2Br at

150 K are t// ¼ 87 meV and tt ¼ �32 meV. With the previous

¼ Br�, AsF6�)

X ¼ AsF6�

100 150 2957.0157(8) 7.0618(9) 7.2419(4)6.422(1) 6.4459(8) 6.4622(3)

35.465(5) 35.562(4) 35.557(2)90 90 9090 90 9092.26(1) 91.73(1) 90

1596.6(3) 1618.0(3) 1664.0(2)

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

values of the S/S contacts in mind, the latter value can seem

surprisingly small. However, a closer look shows that the above-

mentioned short contacts are associated with one (or two)

S atoms of the six-member rings, which have small contributions

to the HOMO (i.e. less than one-third that of the inner core

S atoms). In addition, the associated orbital interactions are of

the intrinsically weak p-type. Thus, the relatively weak lateral

interaction is not that surprising. In contrast, the interactions

along the stacks, although associated with longer S/S contacts,

are of the stronger s-type. This leads to the pseudo-one dimen-

sional (pseudo-1D) band structure of Fig. 6a. The calculated

Fermi surface assuming a metallic filling of the bands

(see Fig. 6b) is a pair of warped but perfectly nested open lines.

The pressure effect discussed above is clearly seen in the

calculated band structures. We report in Table 4 the main

characteristics of the calculated band structures for the mono-

clinic structure of d-(EDT-TTF-CONMe2)2Br at 150 K and

d-(EDT-TTF-CONMe2)2AsF6 at 100 K as well as the ortho-

rhombic room temperature structure of d-(EDT-TTF-CON-

Me2)2Br. It is clear from the comparison of the low-temperature

values for the Br and AsF6 salts that there is a pressure effect

affecting both the intra and interstack interactions, which is

consistent with the fact that the Br salt becomes metallic for

a pressure which is 7 kbar lower than for the AsF6 salt.29 Thermal

contraction also noticeably affects the electronic structure since

there is an increase of approximately 15% in the intrastack

interaction for the Br salt when going from room temperature to

150 K. However thermal contraction is also associated with

a decrease of the interstack interactions (compare the values for

Fig. 6 Calculated band structure (a) and Fermi surface assuming

a metallic filling of the bands (b) for the donor layers of the low-

temperature, P21/a monoclinic form of d-(EDT-TTF-CONMe2)2Br at

150 K. G ¼ (0, 0), X ¼ (a*/2, 0), Y ¼ (0, b*/2).

Table 4 Main parameters of the calculated band structure for d-(EDT-TTF-CONMe2)2Br at room temperature (RT) and 150 K, and d-(EDT-TTF-CONMe2)2AsF6 at 100 K

Wc, meV Wa, meV Wa0, meV

d-(EDT-TTF-CONMe2)2Br at RT 371 84 194d-(EDT-TTF-CONMe2)2Br at 150

K423 66 149

d-(EDT-TTF-CONMe2)2AsF6 at100 K

309 42 103

This journal is ª The Royal Society of Chemistry 2009

Wa and Wa0 in Table 4) despite the fact that the S/S contacts

decrease somewhat (see ESI† Table S1), a reminder of the

importance of the orbital overlap effects which do not always

follow purely metric considerations. Thus, pressure and

thermal contraction seem to have slightly different effects on the

electronic structure although both reinforce the intrastack

interactions.

Let us note the similarity of the electronic band structure of

these salts with that of (TMTTF)2Br,46 even if one must bear in

mind that the stacks in the two salts are slightly different

because of the peculiar criss-cross intrastack overlap of the d-

type salts.47 This leads to the fact that the band dispersion

along the interstack direction goes down slightly in

(TMTTF)2Br but up slightly in the present d-type salts. Indeed,

when comparing the room temperature structures of the two Br

salts, the total dispersion along the stack direction (Wc)

amounts to 0.371 eV in d-(EDT-TTF-CONMe2)2Br, i. e., only

4.8% larger than in (TMTTF)2Br (0.354 eV). The interstack

dispersions are somewhat stronger in d-[EDT-TTF-CON-

Me2]2Br (0.084, 0.194 eV for the upper and lower bands along

G / Z, to be compared with 0.030, 0.082 eV in (TMTTF)2Br).

Thus, the two bromine salts are electronically quite similar but

(TMTTF)2Br has an enhanced 1D character. An important

difference comes from the fact that the present d-type salts do

not exhibit a dimerization gap (D in Fig. 6a) because of the

existence of the glide planes, whereas there is a small but non-

zero band dimerization (0.027 eV) in (TMTTF)2Br. Thus the

present d-type salts can be described as genuine quarter-empty

pseudo-1D systems.

Single crystal 13C NMR spectroscopy and relaxation in (13C-

enriched EDT-TTF-CONMe2)2X (X ¼ AsF6� and Br�)

NMR measurements were carried out at UCLA on samples 13C

spin labeled on the C1 site (Fig. 1–2 and ESI† Scheme S2) for

both AsF6 and Br salts. The objective for the experiments was to

determine whether the materials are in a charge-ordered phase

for T < 300K. Owing to the relatively larger crystals available for

the former, a more detailed set of experiments were made on the

AsF6 compound. However, over the range of temperatures

measured, both materials exhibited spectra and spin–lattice

relaxation consistent with unusually large CO amplitude.

Specifically, the charge ratio for the two sites of the AsF6 salt is

estimated at rB:rA � 9:1, and essentially independent of

temperature over the range 10K < T < 200K. The Br salts are

also in a CO phase over the range 16K < T < 150K, most likely

with a similarly large charge imbalance.

The results reported here were obtained from crystals with

approximate dimensions 1.5 � 0.2 � 0.05 mm3 and 1.5 � 0.1 �0.05 mm3 for the AsF6 and Br salts, respectively. They were

oriented so that the stacking (a) axis is orthogonal to the applied

field. Then, the 13C sites are magnetically equivalent in the

orthorhombic structure when the magnetic field is directed along

b or c; for other orientations, two independent sites should be

observed when all molecular environments are equivalent. Below

we present the results of spectroscopy and relaxation measure-

ments taken on the AsF6 crystal and conclude with a brief

discussion of spectroscopy on the Br crystal.

J. Mater. Chem., 2009, 19, 6980–6994 | 6987

Fig. 8 Shift vs. angle for the four spectral lines. The rotation is about the

a-axis. The results in Fig. 7 were recorded at 4 ¼ �36�.

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

A spectrum recorded at T ¼ 200 K and 4 ¼ �36� relative to

the c-axis is shown in Fig. 7, in which four absorption lines are

resolved and labelled A1, A2, B1, B2. The spectrum provides the

evidence that the AsF6 salt is in the CO phase: the doubling of

the number of resolved peaks indicates that for each of the

two molecular orientations, there are two distinct magnetic

environments.

A rotation pattern, also obtained at T ¼ 200 K, is shown in

Fig. 8, and in which the four distinct absorption features are

labelled as above. The shifts along a were not measured, so the

full shift tensor is not known. However, the variation Kii(4) in

the bc plane is rotated away from the crystal axes, and

approximately determined by the molecular orientations.

For the two distinct molecular environments, we obtain

Kxx(A,B) ¼ 150, �260 ppm and Kyy(A,B) ¼ 50,�400 ppm at

T ¼ 200K, with principal axes rotated relative to the crystal

axes by 4A1,2 ¼ �14(5)�, 4B1,2 ¼ �68(10)�. The results are

consistent with rotation of the principal axes in the bc plane by

approximately 90� for sites A,B. Any additional shifts produced

by the relatively small distortion below the orthorhombic/

monoclinic transition at TOM ¼ 190 K are not resolved in these

experiments.

The variation with temperature of the spectra in environ-

ments A,B as well as the spin–lattice relaxation rate T1�1 vs. T

for sites A,B are discussed separately in a physics paper

reporting on the (P,T) phase diagram.29 The spin–lattice

relaxation rates are shown to differ by about 2 orders of

magnitude (AT1�1 � O(100 BT1

�1) over the entire range of

measured temperatures. The observation of very different rates

is qualitatively similar to that observed in other charge transfer

salts with CO symmetry breaking, such as (TMTTF)2X,20b

Ag(DI-DCNQI)2,19 or a-(BEDT-TTF)2I3.48 Following the

discussions of, for example, the TMTTF materials, we consider

the charge imbalance ratio as

T�1

1 ðBÞT�1

1 ðAÞ¼�

QB

QA

�2

and consequently the charge ratios for sites A,B is roughly 9:1.

Finally, we note that there is no distinction between the two

compounds in the physical properties as measured by NMR at

ambient pressure. For example, consider the spectra for the Br

salt recorded for a sequence of temperatures and shown in

Fig. 9. The distinction between the A, B sites is again clear,

Fig. 7 13C NMR spectrum recorded at T ¼ 200 K and applied field

B ¼ 10.007 T. For arbitrary orientations, there are four non-equivalent

sites (labelled A1, A2, B1, B2; see text).

6988 | J. Mater. Chem., 2009, 19, 6980–6994

even at T ¼ 150 K. Upon cooling, one of the two absorption

lines broadens significantly. In this case, the orientation of the

crystal relative to the magnetic field is sufficiently close to the

crystalline axes that no splitting of either the A or B site is

observed.

In summarizing the results from the NMR, it is useful to

contrast with other compounds exhibiting the CO state, such as

the TMTTF family.20 In this case, the charge imbalance is

significantly larger. Presumably, this is a consequence of the

combination of a large V/t, and strong coupling to the counterion

sublattice. Such a coupling is necessary to explain the details of

the CO order parameter revealed by the X-ray diffraction

measurements described below.

Fig. 9 Temperature evolution of 13C spectra for the Br salt. The CO is

evident already at T ¼ 150 K, and significant line broadening occurs as

the sample is cooled.

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

Back to X-ray experiments: charge ordering consistent with

a doubling of b and change of symmetry

13C NMR, IR and Raman30 all provide evidence for charge

ordering already above the orthorhombic-to-monoclinic transi-

tion for both the AsF6 and Br salts. Therefore, a re-investigation

of their crystal structures is in order to try and identify the

Wigner crystal structure. As the structures described above

contain one independent molecule only, allowing for charge

ordering necessitates two non-equivalent molecules, which

implies either a change of symmetry or a doubling of one lattice

parameter. Therefore, an exploration of the reciprocal space of

(EDT-TTF-CONMe2)2Br was conducted using highly sensitive

experimental set ups at Orsay and readily revealed additional

weak Bragg reflections satisfying both of these criteria. A

doubling of the unit cell occurs along b, the direction transverse

to the stacking axis. Fig. 10 shows one example of a scan across

one superstructure reflection using an X-ray diffractometer with

a point detector. Furthermore, a number of additional Bragg

reflections are found to violate the extinction conditions for

several symmetry elements of Pmna, namely n t y, a t z and

21 k z, yet their intensity does not exceed 1& of the intensity of

the main Bragg reflections. Accordingly, an extensive explora-

tion of the reciprocal space using a standard CCD detector in

Angers, with very long X-ray data collection times, only deliv-

ered a very few of these additional reflections which ultimately

proved to be useless for structure refinements in the double

lattice.

In contrast, single crystal synchrotron experiments for the

detection of the superstructure proved very successful. Crisp

superstructure patterns are obtained upon using the SOLEIL

synchrotron source. Fig. 11a represents the (0kl) reciprocal plane

drawn out of the experimental diffraction data. Rows of super-

structure Bragg reflections are located in between the lines of

main strong reflections. Of particular note in Fig. 11b is the (h7l)

plane entirely composed of superstructure weak reflections.

Notice the sharpness of the reflections, with evidence of reflec-

tions at high diffraction angles indicating long range ordering

with low positional shift. These features are seen to be primarily

associated with the ordering of the heavy bromide atoms since

the experiment was done at the bromine absorption edge.

Fig. 10 Scan across (2, �1.5, 3) at room temperature. Note that the net

intensity of this reflection is 3 to 4 orders of magnitude weaker than that

of the neighbouring Bragg reflections qualifying the averaged Pmna

structure.

This journal is ª The Royal Society of Chemistry 2009

In contrast with the former clear case for a doubling of the

lattice parameter along b, the pattern of systematic absences for

the new lattice is rather a complex one, as exemplified in Table 5

where the comprehensive set of synchrotron data intensities is

systematically averaged for all possible symmetry elements.

Space group determination proved to be challenging since there

are quite a few possibilities to choose from on the basis of the

inventory of symmetry elements satisfying the observed system-

atic absence conditions.

First, recall that EDT-TTF-CONMe2 has no symmetry

elements. In the double lattice of the orthorhombic unit cell two

independent molecules, bearing different charges, can exist either

Fig. 11 (a) (0kl) and (b) (h7l) reciprocal planes drawn out of the room

temperature synchrotron data.

J. Mater. Chem., 2009, 19, 6980–6994 | 6989

Table 5 Number of systematic absence exceptionsa in the orthorhombic double unit cell of d-(EDT-TTF-CONMe2)2Br at room temperature

a ¼ 7.1126(5) A b ¼ 13.0250(7) A (double axis) c ¼ 32.759(1) A

b t x 21 k x n t y 21 k y a t z b t z n t z 21 k z

Nb 6678 31 2549 79 1017 1097 828 217N, I > 3s(I) 231 4 60 0 17 24 17 6<I/s(I)> 0.7 1.2 0.6 0.3 0.5 0.5 0.4 0.6Absence

conditions0kl:k ¼ 2n + 1

h00:h ¼ 2n + 1

h0l:h + l ¼ 2n + 1

0k0:k ¼ 2n + 1

hk0:h ¼ 2n + 1

hk0:k ¼ 2n + 1

hk0:h + k ¼ 2n + 1

00l:l ¼ 2n + 1

a Only possible symmetry elements (i.e. with low <I/s(I)> ratio) are listed in the table. b N is the number of reflections satisfying the absence conditions.

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

in a non-centrosymmetric space group where they occupy general

positions or in an m-containing centrosymmetric group if the two

molecules lie on a mirror plane. In order to allow for the possi-

bility of charge ordering within the stacks (and also, if any, of an

intrastack dimerization), one excludes glide plane a t z. Then,

the symmetry elements 21 k x and b t z are not allowed as they

would move the molecules over into improper positions in the

structure. b t x cannot exist since rows of weak superstructural

peaks with odd k indices are found in the 0kl reciprocal plane.

One can also exclude 2 k y since it connects molecules inside the

stack and therefore is not compatible with a charge ordered

stack. Thus, according to the structure pattern and those

systematic absences remaining after deleting the former dis-

carded symmetry elements, one is left with the following

symmetry elements: 2 k x, m t x, 21 k y, n t y, 21 k z, n t z.

These elements are consistent with three possible orthorhombic

space groups that would all ultimately allow for non-uniform

molecular stacks with two independent molecules, that is, Pmnn,

P2nn, P22121.

Fig. 12 Three dimensional pattern of alternating neutral and charged

molecules based on refinements of the room temperature synchrotron

data taking into account weak superstructural reflections in space group

P2nn: (a) along the stacking axis direction a; and (b) for one single layer

viewed along b, the direction transverse to the stacking axis. The geom-

etry of hydrogen contacts is given in Table 2. Anti-phase displacements of

several bromide anions belonging to this layer are symbolized by red

arrows.

Refinement of room temperature synchrotron data favors non-

centrosymmetric group P2nn and reveals the three-dimensional

pattern of charge ordering

Refinements in all of the former three groups, Pmnn, P2nn and

P22121, each consistent with a double lattice with two indepen-

dent molecules of different charge, yield approximately the same

picture for the three-dimensional charge disproportionation

pattern, the essence of the Wigner crystal structure. Yet, refine-

ment in the acentric P2nn group is seen as the best outcome of the

analysis and refinement of the room temperature synchrotron

data, as discussed below.

The primary outcome of the synchrotron data refinements in

the double cell of P2nn symmetry is that the two independent

molecules have different charges (Fig. 12) as demonstrated by the

striking difference between the bond lengths of the inner C]C

double bonds within the TTF cores. Indeed, the calculated

HOMO energy of donor I is 0.100 eV lower than that of donor II,

suggesting a smaller charge for I. Hence, as exemplified in

Fig. 12, molecule I0 with a short inner C]C of 1.354(2) A is

formally neutral. Conversely, it is likely that most of the charge is

localized on radical cation IIc+ whose inner C]C bond is elon-

gated up to 1.392(2) A in agreement with the respective bonding

character of the adjacent inner carbon p orbitals contribution to

the HOMO, and entirely consistent with the former analysis

of the temperature dependence of the single crystal 13C NMR

6990 | J. Mater. Chem., 2009, 19, 6980–6994

spin–lattice relaxation rate T1�1 for the two discrete molecular

sites which suggested a record high charge ratio of 9:1. Therefore,

radical cations IIc+ and neutral molecules I0 alternate along the

stacks, as shown in Fig. 12a.

The Wigner structure keeps uniform stacks

Although P2nn symmetry does allow for any amount of

dimerization, there is only a minute, if any, dimerization along

the stacking direction since the separations between successive

molecular planes, amounting to 3.550(9) and 3.563(8) A, are

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

identical within experimental accuracy. Nevertheless, the

difference in the CEt–H/O contact distances is significant.

Each pair of molecules within the stack is connected by two

hydrogen contacts, and both of them are shorter in one pair

(interaction A) in comparison with another (interaction B,

Fig. 12a and Table 2).

Therefore, a remarkable feature of the Wigner crystal struc-

ture is that, despite the occurrence of charge ordering, the

stacks appear to remain essentially uniform. For instance, the

calculated transfer integrals along the stack (t// and t0//) differ by

only 2.2 meV. A detailed analysis of the electronic structures of

the Pmna and P2nn room temperature structures shows that it

is the difference in HOMO donor energies which mostly

controls the differences. Thus, the lowering of the HOMO

energy of donor I with respect to that of donor II leads to an

increase in the relative participation of donor I in the lower

levels whereas the opposite is true in the upper levels. This leads

to the development of deeper minima in the central part of the

density of states (DOS) in the region of the HOMO bands

(see Fig. 13a and c). However, the electronic structure and in

particular the nature of the interactions along the stacks is not

sizeably altered. For instance, the calculated Fermi surfaces

assuming a metallic filling of the bands for the Pmna and P2nn

structures are shown in Fig. 13b and d, respectively. The Fermi

surface of Fig. 13d is just a folded version of that of Fig. 13b,

as required by the doubling of the cell along b.

Fig. 13 Calculated density of states, (a) and (c), and Fermi surface

assuming a metallic filling of the bands, (b) and (d), for the room

temperature Pmna and P2nn structures of d-(EDT-TTF-CONMe2)2Br,

respectively. The DOS is given in states per eV and per repeat unit of the

donor layer. The dashed line in the DOS diagrams refers to the Fermi

level assuming a metallic filling of the bands.

This journal is ª The Royal Society of Chemistry 2009

Activation-deactivation upon charge ordering of large collections

of transverse Csp2–H/O hydrogen bonds

Another significant outcome of the double cell, synchrotron data

refinement in P2nn is that interstack interactions along b are

controlled by the localization of the full positive charge onto only

one of the two independent molecules, via a modulation, or

activation, of in-plane, transverse Csp2–H/O hydrogen bonds,

as shown in Fig. 12b and 14. The prior assumption, before

carrying out the synchrotron diffraction experiment, was that

doubling of the lattice in the charge ordered Wigner crystal could

be a manifestation of the alternation of neutral and charged

molecules along the double b parameter. Since molecular units

are coplanar in the b-direction transverse to the stacks (Fig. 2b

and 12b), an alternation of long and short Csp2–H/O contacts

was indeed anticipated. Stronger Csp2–H/O hydrogen bonding

interactions, and shorter contacts, are expected from IIc+

towards I0, that is, from the redox activated, enhanced Csp2–H

hydrogen bond donor to the sizeable hydrogen bond acceptor

carbonyl oxygen atom of neutral EDT-TTF-CONMe2 than the

same involving the reversed pairs of weaker donor I0 and weaker

acceptor IIc+.8,49 Remarkably, the actual charge ordering

revealed by the refinement taking into account the observed

superstructure is entirely consistent with the former assumption.

As shown in Fig. 12b, adjacent stacks are packed in such a way

that indeed I0 and IIc+ alternate along b, the direction transverse

to the stacks. Accordingly, the Csp2–H/O hydrogen bond in

interaction C is stronger than in D (Fig. 12b and Table 2). Note

that, in contrast to the remarkable activation-deactivation of

side-by-side hydrogen bonds upon charge alternation, S/S side-

by-side contacts remain almost equivalent for both C and D

interactions (ESI† Table S1).

Anti-phase, bromide anion sub-lattice static modulation wave

A last significant outcome of the successful refinement of the

ambient pressure, room temperature charge ordered structure is

Fig. 14 Projection along a of the room temperature, three-dimensional

charge ordered structure of d-(EDT-TTF-CONMe2)2Br based on the

refinement of synchrotron data in space group P2nn. Comparison with

Fig. 2a exemplifies the doubling of the unit cell along b. The darker and

lighter blue solid lines, as with Fig. 12b, represent the stronger and

weaker Csp2–H/O hydrogen bonds, respectively, activated or deacti-

vated upon charge localization. Anti-phase displacements of bromide

anions are symbolized by red arrows.

J. Mater. Chem., 2009, 19, 6980–6994 | 6991

Scheme 1 Schematics of the full set of four possible arrangements of

neutral (open ovals) and charged (solid black ovals) EDT-TTF-

CONMe2 molecules, and Br anions (pink circles), in the bc plane of

a double lattice, perpendicular to the stacking a-direction. Closed and

dashed symbols refer to positions at a ¼ 0 or 1/2, respectively. The

charge ordered state is associated with an activation of hydrogen bond

interactions between Br� and [EDT-TTF-CONMe2]+c, shown by black

dotted lines, resulting in anti-phase shifts of the Br anions towards the

radical cations as indicated by red arrows. (a) and (b) correspond

to structures where stacks of [EDT-TTF-CONMe2]+c and stacks of

[EDT-TTF-CONMe2]0 would be segregated. In (a), charge-rich and

charge-poor stacks alternate along c, and there is no Br shift due to

symmetrical hydrogen interactions. In (b), with no alternation of charge

along c, the modulation waves are in-phase for all Br� strings. (c) and

(d) correspond to mixed stacks, that is where charges order along both

the stacking a and transverse b directions. There is no bromide atoms

displacement in (c), while in (d) displacement waves for adjacent Br�

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

that the Br/Br separations along b, which are equivalent in the

averaged Pmna and P21/a structures, are found to differ in P2nn,

amounting to 6.4985(6) and 6.5265(6) A. Strings of Br� anions

are threading along in the bc-planes across pockets bordered by

ethylene groups of differently charged donors. Electrostatic

interactions between I0 and IIc+ and Br ions are not equivalent,

pointing to a modulation of the anion surroundings upon charge

ordering, which leads the bromide anions to shift away from

neutral organic molecules towards the positively charged donors

(Fig. 12b and Table 2 for CEt–H/Br contacts). In agreement

with the pattern of charge alternation in the bc-plane, adjacent

anions along the Br ion strings experience counter displacement

primarily along b in such a way that pairs of charged molecules

become bracketed in between pairs of closer bromide anions. It

should be emphasized here that the concerted anti-phase move-

ment of heavy Br atoms within any such string accounts for most

of the intensity of the 2b superstructure reflections measured with

the synchrotron radiation (Fig. 11). As a result, the anion string

becomes slightly dimerized. Moreover, as exemplified in Fig. 14,

the static longitudinal modulation wave of the bromine anions

string running along [0,b,1/2] is in anti-phase with the similar

wave along [1/2,b,0], in tune with the pattern of charge ordering

within neighboring donor stacks, hence the characteristic 3D

pattern of dimerization of the anion sub-lattice (Fig. 12).

It is of interest to note that the demonstration here of

a synchronized displacement of the cation and anion sites in the

crystal upon charge ordering provides convincing support for

earlier similar models of displacements postulated for the

(TMTTF)2X series which could not be tested by such full

structure determination.27

strings along c are necessarily in anti-phase, as observed in the present

P2nn structure.

Final considerations on symmetry and possible patterns of charge

ordering

In order to reach a comprehensive view of the symmetry of the

CO state, we now consider all possible patterns of charge

ordering in the radical cation sub-lattice and assess their

compatibility with relevant atomic displacements in the anion

sub-lattice. Let us remember that, right at the onset of the

choice of space group, it was assumed that the structure in the

double lattice contains charge ordered stacks. Yet, the charge

ordering pattern could be principally different if one allows for

segregated stacks, that is, either charge-rich or charge-poor

(neutral) stacks to be repeated along b. Scheme 1 gives an

overview of the full set of four possible CO patterns with either

segregated stacks (a and b) or mixed stacks (c and d) where

neutral and charged molecules are shown by white and black

ovals, respectively. Quite remarkably, each of the diverse, four

CO topologies (a-d) appears to be associated with divergent

types of transformations of the anion sub-lattices. In fact, in

each case the movement of the anions is seen as being directed

by the activated hydrogen bond interactions between Br� and

charge-rich radical cations represented by dotted lines in

Scheme 1. Accordingly, for segregated stacks, the bromide

anion sub-lattice would eventually be uniform (Scheme 1a), or

one single, anti-phase longitudinal modulation wave for the

bromide strings could prevail and it should be in-phase in all

chains (Scheme 1b). For mixed stacks, there is either no

bromide atoms displacement (Scheme 1c) or anti-phase

6992 | J. Mater. Chem., 2009, 19, 6980–6994

modulation waves in adjacent anion chains (Scheme 1d). In this

X-ray experiment carried out at the Br absorption edge, the Br

atom sub-lattice is better defined than the organic sub-lattice

and thereby captures most of the structural features of the

concerted radical cation-anion ordering. Indeed, refinement of

the structure in an asymmetric P1 double lattice allowing for

unconstrained displacements of the heavy Br ions demonstrates

that only pattern d, which associates mixed stacks and anti-

phase Br modulation, ultimately holds.

Note that pattern d is compatible with two space groups

only, P2nn and P22121, both of which are non-centrosym-

metrical. However, refinement in P22121 indicates that the

differences in bond lengths and intermolecular contact distances

between I0 and IIc+ are quite subtle and, besides, this space

group merely accounts for the full pattern of systematic

absences, given in Table 5, in comparison with P2nn which

describes it all. Note also that centrosymmetric Pmnn, the last

of three previously selected space groups, corresponds to c

where the Br modulation wave cannot be observed. Thus, we

conclude that the acentric space group P2nn, compatible with

a ferroelectric charge ordered state, is the only one to provide

for a full description of the orthorhombic room temperature

charge ordered structure.

Finally, we note in closing that the present report meets with

the challenge outlined in the introductory comments by

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

delivering a full structure determination of a Wigner crystal

structure which hereby demonstrates a rather simple 3D pattern

where the charges are found to order on every other molecules

both along the stacking direction a as well as from one stack to

another along b in an orthorhombic unit cell and non-centro-

symmetric space group. This appears to be in sharp contrast with

the complex pattern proposed recently for (DI-DCNQI)2Ag19c

where the charge is localized both on molecular sites and onto

dimeric molecular units, with a complex helico€ıdal pattern in

between, and also for the (TMTTF)2X series whose CO pattern

has been postulated to consist of a mixture of 4kF bond (mixed

valence dimers) and (single molecular) site orders.20,22,25,27

Conclusions

The whole of the foregoing results, which include the synthesis

and successful application of an internal chemical pressure to

effectively control, and reduce, the Mott gap in d-(EDT-TTF-

CONMe2)2Br; the detailed accounts of its Pmna, averaged

room temperature structure as well as reversible phase transi-

tion at ca. 190 K, accompanied by twinning; the full description

of its averaged low temperature P21/a structure; the synthesis of

(13C-EDT-TTF-CONMe2)2[Br and AsF6], where one carbon

atom of the inner double bond is 100% 13C-enriched, allowing

high resolution 13C solid state NMR experiments to be con-

ducted, as well as IR reflectivity and Raman experiments

reported independently, which both reveal that charge ordering

occurs already at room temperature and ambient pressure; the

discovery of weak superstructure Bragg reflections in d-(EDT-

TTF-CONMe2)2Br and subsequent analysis and refinement of

an exhaustive synchrotron radiation data set; suggests an

alternation at room temperature of neutral and oxidized

molecules along both the stacking a and transverse b directions

in orthorhombic, non-centrosymmetric space group P2nn, and

demonstrates a rather simple 3D CO pattern compatible with

a ferroelectric charge ordered state. The charge disproportion-

ation, and subsequent long range order crystallization of the

electron gas within the three-dimensional Wigner lattice is

found to be coupled to a concerted activation-deactivation of

large collections of transverse Csp2–H/O hydrogen bonds

coupled to an anti-phase, static modulation of the bromide

anions displacements along b, thereby demonstrating a cooper-

ative participation of the anion and cation sub-lattices to the

stabilization of the CO state. In addition to (DI-DCNQI)2Ag

discussed earlier, where superstructure Bragg reflections

condensate from diffuse scattering at the CO transition, only

a few CO structures of molecular systems have been investi-

gated using synchrotron radiation. The analysis of powder

synchrotron data for (EDO-TTF)2PF6 suggested a doubling of

the unit cell upon CO.50 Single crystal synchrotron data were

used for the investigation of the P1 CO structure51 of a-(BEDT-

TTF)2I3 where no superstructure reflections are identified, in

contrast to q-(BEDT-TTF)2RbZn(SCN)4 where a doubling of

the unit cell was found52 upon CO transition at high pressure.

Note in closing that a remarkable feature of the Wigner crystal

structure is that the stacks appear to remain essentially uniform

in the charge ordered state in agreement with the rich low

temperature Mott physics of the system.29

This journal is ª The Royal Society of Chemistry 2009

Acknowledgements

This work was supported by the French National Research

Agency, ANR Project CHIRASYM 2005-08 (NT05-2 42710),

the Interdisciplinary ANR project 3⁄4 -Filled 2009-2011 (ANR-

08-BLAN-0140-01); the CNRS; the INTAS Grant 04-03-4001;

the Spanish Ministerio de Educaci�on y Ciencia (Project FIS2006-

12117-C04-01) and the Generalitat de Catalunya (2005 SGR

683); and NSF grants DMR-0520552 and DMR-0804625. We

thank Jean-Charles Ricquier for expert assistance in preparing

the graphical abstract illustration. We thank the CNRS for an

associated researcher position (LZ) and the R�egion des Pays de

Loire for a post-doctoral fellowship (SS).

References

1 N. F. Mott, Metal-insulator transitions, Taylor and Francis, London/Philadelphia, 1990.

2 T. Giamarchi, Chem. Rev., 2004, 104, 5037.3 H. Seo, C. Hotta and H. Fukuyama, Chem. Rev., 2004, 104, 5005.4 P. P. Edwards, H. B. Gray, M. T. J. Lodge and R. J. P. Williams,

Angew. Chem. Int. Ed., 2008, 47, 6758.5 See Fig. 9 of reference 2.6 (a) P. Fulde, Electron Correlations in Molecules and Solids, 3rd

edition, Springer-Verlag; Berlin, 2002; (b) N. Grewe, F. Steglich, inHandbook on the Physics and Chemistry of Rare Earths, Volume 14,K. A. Gscheidner Jr. and L. Eyring, Eds., Elsevier; Amsterdam -New York, 1991, p. 343; (c) M. Imada, A. Fujimori and Y. Takura,Rev. Mod. Phys., 1998, 63, 1; (d) J. P. Attfield, Solid State Sci.,2006, 8, 861–867; (e) J. P. Wright, J. P. Attfield and P. G. Radaelli,Phys. Rev. Lett., 2001, 87, 266401.

7 (a) P. Batail, Special Issue on Molecular Conductors, Chem. Rev.,2004, 11, p104; (b) Special Topic: Organic conductors, J. Phys. Soc.Jpn, 2006, 5, 75.

8 M. Fourmigu�e and P. Batail, Chem. Rev., 2004, 104, 5379.9 (a) D. J�erome, Chem. Rev., 2004, 104, 5565; (b) L. Degiorgi and

D. J�erome, J. Phys. Soc. Jpn, 2006, 75, 051004.10 S. Kagoshima and R. Kondo, Chem. Rev., 2004, 104, 5593.11 R. Rousseau, M. Gener and E. Canadell, Adv. Funct. Mater., 2004,

14, 201.12 J. P. Pouget, S. K. Khanna, F. Denoyer, R. Com�es, A. F. Garito and

A. J. Heeger, Phys. Rev. Lett., 1976, 37, 437.13 In reciprocal space, 4kF corresponds to the first Fourier component of

a 1D Wigner lattice of localized charges in direct space (in reciprocalunits, 4kF amounts to the inverse average charge separation), see:(a) J. Kondo and K. Yamaji, J. Phys. Soc. Jpn, 1977, 43, 424;(b) J. Hubbard, Phys. Rev. B, 1978, 17, 494.

14 A. J. Epstein, J. S. Miller, J. P. Pouget and R. Com�es, Phys. Rev. Lett.,1981, 47, 741.

15 F. Mila and X. Zotos, Europhys. Lett., 1993, 24, 133.16 H. Seo, J. Merino, H. Yoshioka and M. Ogata, J. Phys. Soc. Jpn,

2006, 75, 051009.17 M. Dressel and N. Drichko, Chem. Rev., 2004, 104, 5689.18 S. Huizinga, J. Kommandeur, A. Sawatzky, B. T. Thole, K. Kopinga,

W. J. M. de Jonge and J. Joos, Phys. Rev. B, 1979, 19, 4723.19 (a) K. Hiraki and K. Kanoda, Phys. Rev. Lett., 1998, 80, 4737;

(b) K. Miyagawa, K. Kanoda and A. Kawamoto, Chem. Rev.,2004, 104, 5635; (c) T. Kakiuchi, Y. Wakabayashi, H. Sawa,T. Itou and K. Kanoda, Phys. Rev. Lett., 2007, 98, 066402.

20 (a) D. S. Chow, E. Zamborszky, B. Alavi, D. J. Tantillo, A. Baur,C. A. Merlic and S. E. Brown, Phys. Rev. Lett., 2000, 85, 1698;(b) F. Zamborzky, W. Yu, W. Rass, S. E. Brown, B. Alavi,C. A. Merlic and A. Baur, Phys. Rev. B, 2002, 66, 081103.

21 Y. Nogami, R. Moret, J. P. Pouget, Y. Yamamoto, K. Oshima,K. Hiraki and K. Kanoda, Synth. Met., 1999, 102, 1778.

22 P. Monceau, F. Ya Nad and S. Brazovskii, Phys. Rev. Lett., 2001, 86,4080.

23 R. T. Clay, S. Mazumdar and D. K. Campbell, Phys. Rev. B, 2003, 67,115121.

24 M. Kuwabara, H. Seo and M. Ogata, J. Phys. Soc. Jpn., 2003, 72, 225.25 J. Riera and D. Poilblanc, Phys. Rev. B, 2001, 63, 241102(R).

J. Mater. Chem., 2009, 19, 6980–6994 | 6993

Publ

ishe

d on

04

Aug

ust 2

009.

Dow

nloa

ded

by U

nive

rsity

of

Wal

es A

bery

stw

yth

on 2

9/03

/201

7 20

:19:

20.

View Article Online

26 W. Yu, F. Zhang, F. Zamborszky, B. Alavi, A. Baur, C. A. Merlic andS. E. Brown, Phys. Rev. B, 2004, 70, 121101.

27 M. de Souza, P. Foury-Leylekian, A. Moradpour, J. P. Pouget andM. Lang, Phys. Rev. Lett., 2008, 101, 216403.

28 K. Heuz�e, M. Fourmigu�e, P. Batail, C. Coulon, R. Cl�erac,E. Canadell, P. Auban-Senzier, S. Ravy and D. J�erome, Adv.Mater., 2003, 15, 1251.

29 P. Auban-Senzier, C. Pasquier, D. J�erome, S. Suh, S. E. Brown,C. M�ezi�ere and P. Batail, Phys. Rev. Lett., 2009, 102, 257001.

30 T. Yamamoto, C. M�ezi�ere, P. Batail, unpublished work.31 K. Heuz�e, M. Fourmigu�e and P. Batail, J. Mater. Chem., 1999, 9, 2373.32 J. Larsen and C. Lenoir, Synthesis, 1989, 2, 134.33 A. J. M. Duisenberg, J. Appl. Cryst., 1992, 25, 92.34 A. J. M. Duisenberg, L. Kroon-Batenburg and A. M. M. Schreurs,

J. Appl. Cryst., 2003, 36, 220.35 L. J. Farrugia, J. Appl. Cryst., 1999, 32, 837.36 G. M. Sheldrick, SHELXS-97 and SHELXL-97, University of

G€ottingen, Germany, 1997.37 R. Herbst-Irmer and G. M. Sheldrick, Acta Cryst., 1998, B54, 443.38 Oxford Diffraction (2007). Oxford Diffraction Ltd., Xcalibur CCD

system, CrysAlisPro Software system, Version 1.171.32.39 M.-H. Whangbo and R. Hoffmann, J. Am. Chem. Soc., 1978, 100, 6093.40 J. Ammeter, H.-B. B€urgi, J. Thibeault and R. Hoffmann, J. Am.

Chem. Soc., 1978, 100, 3686.41 B. N�afr�adi, R. G�aal, A. Olariu, L. Forr�o, C. M�ezi�ere, P. Batail,

A. J�anossy, unpublished work.

6994 | J. Mater. Chem., 2009, 19, 6980–6994

42 B. K. Vainstein, V. M. Fridkin, V. L. Indenbom, ModernCrystallography, 2nd ed., Springer-Verlag; Berlin – Heidelberg –New York, 1995, V. 2.

43 S. Parsons, Acta Cryst., 2003, D59, 1995.44 Cambridge Structural Database. Version 5.26. University of

Cambridge, UK.45 I. J. Bruno, J. C. Cole, P. R. Edgington, M. Kessler, C. F. Macrae,

P. McCabe, J. Pearson and R. Taylor, Acta Cryst., 2002, B58, 389.46 (a) L. Balicas, K. Behnia, W. Kang, E. Canadell, P. Auban-Senzier,

D. J�erome, M. Ribault and J. M. Fabre, J. Phys. I, 1994, 4, 1539; (b)for an earlier band structure, although not directly comparable tothe present results because of the different computational details, see:L. Ducasse, A. Abderraba and B. Gallois, J. Phys. C, 1986, 19, 3805.

47 T. Mori, Bull. Chem. Soc. Jpn., 1999, 72, 2011.48 (a) Y. Takano, K. Hiraki, H. M. Yamamoto, T. Nakamura and

T. Takahashi, Synth. Met., 2001, 120(1–3), 1081; (b) S. Moroto,K. Hiraki, Y. Takano, Y. Kubo, T. Takahashi, H. M. Yamamotoand T. Nakamura, J. Phys. IV, 2004, 114, 339.

49 S. A. Baudron, P. Batail, C. Rovira, E. Canadell and R. Cl�erac, Chem.Commun., 2003, 1820.

50 S. Aoyagi, K. Kato, A. Ota, H. Yamochi, G. Saito, H. Suematsu,M. Sakata and M. Takata, Angew. Chem. Int. Ed., 2004, 43, 3670.

51 T. Kakiuchi, Y. Wakabayashi, H. Sawa, T. Takahashi andT. Nakamura, J. Phys. Soc. Jpn., 2007, 76, 113702.

52 M. Watanabe, Y. Noda, Y. Nogami and H. Mori, J. Phys. Soc. Jpn.,2004, 73, 921.

This journal is ª The Royal Society of Chemistry 2009