Skip to main content

Full text of "Infrared Investigation of the Charge Ordering Pattern in the Organic Spin Ladder Candidate (DTTTF)2Cu(mnt)2"

See other formats


Infrared Investigation of the Charge Ordering Pattern in the Organic Spin 

Ladder Candidate (DTTTF)2Cu(mnt)2 

> J. L. Musfeldt 1 , S. Brown 1 '*, S. Mazumdar 2 , R. T. Clay 3 , M. Mas-Torrent 4 , C. Rovira 4 , J. C. Dias 5 

R. T. Henriques 5 , M. Almedia 5 

1 Department of Chemistry, University of Tennessee, Knoxville, TN 37996, USA 
Q\ 2 Department of Physics, University of Arizona, Tuscan, AZ 85721, USA 

3 Dept. of Physics and Astronomy, HPC 2 Center for Computational Sciences, Mississippi State University, Mississippi State, MS 39762, USA 
I i 4 Institut de Ciencia de Materials de Barcelona, CSIC Campus de la UAB, E-08193 Bellaterra, Spain 

5 Departamento de Quimica, Instituto Tecnoldgico e Nuclear /Centero de Fisica da Materia Condensada da Universidade de Lisboa, 



P-2686-953 Sacavem, Portugal 



Abstract 



I , We measured the variable temperature infrared response of the spin ladder candidate (DTTTF)2Cu(mnt)2 in order to distinguish 
^ between two competing ladder models, rectangular versus zigzag, proposed for this family of materials. The distortion along the 
S stack direction below 235 K is consistent with a doubling along b through the metal-insulator transition. While this would agree 
with either of the ladder models, the concomitant transverse distortion rules out the rectangular ladder model and supports the 
i i zigzag scenario. Intramolecular distortions within the DTTTF building block molecule also give rise to on-site charge asymmetry. 



> 

OO Quantum spin ladders have attracted considerable in- 
terest as intermediaries between one-dimensional chains 
and two-dimensional square lattices |H2l3l4j . Additional 
• interest has arisen from theoretical studies which find that 
; hole-doped spin ladders can support superconductivity 

j, |2l5j . Whereas most systems are structural ladders, organic 

ladder-like compounds such as dithiophentelrathiafulva- 
!* lene copper maleonitrile dithiolate, (DTTTF)2Cu(mnt)2 
. — i are formed by the coupling of molecular building blocks 
(Fig.JTJa)). This system is particularly attractive because 
^ it belongs to a family of quasi-isostructural compounds, 
with tunable properties depending on the counterfoil 
(M = Pt, Cu, Au, Ni) [3E]. That the cation stacks of 
(DTTTF) 2 Cu(mnt ) % have a ^-filled band with delocalized 
charge and spin at high temperatures, and are therefore 
described by an electronic extended Hubbard Hamiltonian, 
as opposed to the localized Heisenberg spin Hamiltonian, 
brings additional complexity to the field. 

Two different theoretical models, the rectangular 
|4|6|7|8j and the zigzag ladder [9], have been proposed for 
this family of coupled stack materials (Figs. [TJd) and (e)) 
Although determination of the wavevector dependence of 
the magnon bands [TO] is the strongest test of spin lad- 
der character and can in principle distinguish between 
the models, inelastic neutron scattering experiments on 
(DTTTF)2Cu(mnt)2 are limited by sample quantity and 



isotopic decoration requirements |llj . The cations have 
homogeneous charge within the rectangular model and are 
charge ordered within the zigzag model [9]. An alternate 
approach to distinguishing between the models involves 
detailed comparison of vibrational property measurements 
with theoretical predictions. 

(DTTTF)2Cu(mnt)2 is a donor-acceptor salt that dis- 
plays the characteristic herringbone chain structure of 
many organic molecular solids (Fig. [Tja)). The ^-filled 
band donor stack is weakly metallic at room temperature 
with a dc ~ 12 tt- 1 cm- 1 [12 . (DTTTF) 2 Cu(mnt) 2 has 
a sharp second-order metal-insulator transition at 235 K, 
which doubles the unit cell along the stacking direction, b 
|12l4j . Within the rectangular ladder model, each dimer 
unit cell of (DTTTF) 2 + ' radical cations along the stack- 
ing 6-axis is thought to act as a single site with spin S = 
i below this transition temperature |13l3l4j . The rungs 
of the proposed ladder lie along c and are formed by 
close S"'S contacts between DTTTF molecules in adja- 
cent stacks. From susceptibility measurements, magnetic 
exchange strengths are estimated to be Jy ~ 121 K and 
J ± ~ 218 K g]. The ratio of Jj_/Jy is 1.8, indicative of 
intermediate coupling. The spin gap is 130 K [3]. Interest- 
ingly, application of pressure recovers the metallic state 
[12 . The structural, magnetic, and electrical properties of 
doped DTTTF-based materials were also studied, [7] but 



X 




0.15r 



-I mm 






ts 


tt 








1 J s 


> — 4 


l < 


1 1 




t — • 




Fig. 1. (Color online) (a) Crystal structure of (DTTTF) 2 Cu(mnt) 2 
g], (b) photograph of the (DTTTF)2Cu(mnt)2 single crystal and 
the aperture used during our infrared measurements, (c) proposed 
low-temperature arrangement of charge supporting the rectangular 
spin ladder picture [4], (d) schematic of the bond distortion pattern 
within the rectangular ladder model [6], and (e) schematic charge 
ordering and bond distortion pattern within the zigzag ladder model. 
The grey circles in (d) denote site charge of 0.5 while the filled 
(unfilled) circles in (e) denote site charges greater than (less than) 
0.5 [9]. 

superconductivity was not observed. 

In order to investigate the charge ordering and bond dis- 
tortion patterns (if any) of (DTTTF) 2 Cu(mnt) 2 below the 
metal-insulator transition, we measured the temperature- 
dependent infrared vibrational properties of this material. 
We compare the results with those of (DTTTF)2Au(mnt)2 
[8 and with theoretical predictions of intermolecular charge 
ordering in the rectangular and zigzag ladders [4|6|9j . We 
find that the zigzag ladder model provides a more appropri- 
ate description of the large transverse structural distortion 
that accompanies the 235 K metal-insulator transition. A 
mode analysis provides the microscopic basis for this distor- 
tion. We briefly discuss the consequences of intramolecular 
distortions for charge ordering within the DTTTF building 
block molecule in both the charge transfer salt and neutral 
molecular solid. 

Single crystals of (DTTTF)2Cu(mnt)2 were grown as de- 
scribed previously |3l4j . Typical dimensions were 2 x 0.3 
x 0.1 mm. Our work was done on the largest (be) crystal 
face. As shown in Fig.[TJb), the crystal surfaces were some- 
what striated. Variable temperature infrared reflectance 
measurements were carried out on these small crystals us- 
ing a Bruker 55 Fourier transform infrared spectrometer 
and attached microscope accessory. Both wire grid and 
polaroid film polarizers were used, as appropriate. Mea- 
surements were taken at several different temperatures us- 
ing a cryostat and temperature controller setup, concen- 
trating on the temperature range near the phase transi- 
tions. The cooling rate was slow (~ a few K/min) to avoid 




650 700 750 800 850 900 
Frequency (cm" ) 



1200 1300 1400 
Frequency (cm" ) 




600 650 700 750 800 
Frequency (cm ) 



1000 1050 1100 1150 1200 
Frequency (cm" ) 



0.4 

0.3 
0.2 
0.1 
0.0 



\ (e) ^ 


E||c 










300K 


— t 



0.10 



1200 1250 1300 1350 1400 
Frequency (cm ) 



0.05 



(f) 


E||c 







3050 3100 3150 

Frequency (cm" ) 



Fig. 2. (Color online) Close-up views of several vibrational features 
of (DTTTF)2Cu(mnt)2 in reflectance mode, (a) in-phase breathing 
mode of the outer ring of DTTTF cation involving C-S stretch and 
C=C-S bend, (b) symmetric C-C stretching mode of the DTTTF 
cation, (c) C-S-C bending and C-S stretching modes of the DTTTF 
cation, (d) asymmetric C-C + C-S stretch + C-H wag of DTTTF, 
S-C=C-H wagging motion of the DTTTF cation, and C-C stretch of 
the Cu(mnt)2 counterion, (e) out-of-phase C-C stretch in DTTTF, 
and (f) C-H stretching mode of the DTTTF cation. Measurement 
polarizations and temperatures are indicated in each panel. The 
arrows in (e) denote features used in the charge analysis. 

breakage. Careful aperaturing minimized the impact of sur- 
face quality and small crystal size (Fig.jljb)). A Kramers- 
Kronig analysis was tried in an attempt to relate the mea- 
sured reflectance to the optical constants of the material. 
In many cases, the procedure worked well, but in a few 
cases, less reasonable results were obtained. We therefore 
elect to present and discuss the absolute reflectance spec- 
tra of (DTTTF)2Cu(mnt)2, which captures the important 
physics of this system over the full temperature range of 
investigation without the variability of a Kramers-Kronig 
analysis. 

The 300 K optical properties of (DTTTF) 2 Cu(mnt) 2 are 
highly anisotropic and similar to those of the Au analog 
[8 , with a combination of electron-molecular vibrational 
and infrared active modes of B u symmetry plus a strong 
charge transfer band in the chain direction. Along 6, the 
largest modes appear at ^770 and 1300 cm -1 in the re- 
flectance spectrum. We attribute these features to terminal 
C-S and C-C stretching vibrations of the DTTTF cation 
based on our dynamics calculations [8]. These modes are 



totally symmetric and activated along the chain direction 
due to electron-molecular vibrational coupling |14j . In the 
c direction, there are many infrared-active vibrations of B u 
symmetry |14J . Below, we focus on trends in the C-S-C 
bending and C-S stretching features of the DTTTF cation 
between 600 and 800 cm -1 , the asymmetric C-C + C-S 
stretch + C-H wag and the S-C=C-H wagging motion of 
the DTTTF cation between 1000 and 1200 cm" 1 , the out- 
of-phase C-C stretch of the DTTTF cation at -1335 cm" 1 , 
and the C-H stretch of the DTTTF cation near 3100 cm" 1 . 

Figure [2] displays close-up views of the reflectance of 
(DTTTF)2Cu(mnt)2 as a function of temperature. As 
shown in Fig. |2ja) and (b), the spectral response above 
and below 235 K is consistent with a doubling along b 
through the metal-insulator transition |12)4j . There is also 
a striking transverse distortion as evidenced by changes in 
the c-polarized reflectance spectra. Analysis of the vibra- 
tional modes that probe the interchain interactions shows 
that symmetry breaking involves the C-S bending and 
stretching modes of the DTTTF cation, combined C-C + 
C-S stretching and C-H wagging, and S-C=C-H wagging 
perpendicular to the chain (Fig. [2^c) and (d)). This result 
has important implications for the intermolecular charge 
ordering pattern as discussed below. 

The variable temperature spectral response of (DTTTF)2 
Cu(mnt)2 is different than that of the Au analog. Two 
separate transitions are observed (220 and 70 K) in the lat- 
ter. The broad 220 K metal-insulator transition is driven 
by massive symmetry breaking along the rung direction, 
whereas the 70 K magnetic transition is associated with 
mode modifications in the rail direction [8|. In contrast, 
the rung- and rail-directed lattice distortions are coupled 
in (DTTTF) 2 Cu(mnt) 2 . Both occur near 235 K, consistent 
with the observation of a sharper transition. This may be a 
consequence of stronger interchain interactions (see below). 

We now discuss the implications of the combined fr-axis 
doubling and large transverse distortion at the metal- 
insulator transition for the intermolecular charge ordering 
pattern in (DTTTF) 2 Cu(mnt) 2 . Within the rectangular 
ladder model of Fig.jljd), bond alternation along the rails 
is a necessity but charge ordering is not anticipated [6]. 
The symmetry between the chains implies that this model 
cannot account for the strong transverse distortion in 
(DTTTF)2Cu(mnt)2: specifically, the symmetry breaking 
involving C-S, C-C, and complex wagging motion perpen- 
dicular to chain direction (Fig.[2|c) and (d)). The proposed 
Heisenberg spin ladder model with (DTTTF)2 + ' dimers 
acting as the S = | sites (Fig. [ljc)) |3l4j is therefore not 
applicable to this class of materials. 

The zigzag ladder model provides a more promising 
framework for understanding intermolecular charge order- 
ing in (DTTTF)2Cu(mnt)2- The bond-charge density wave 
ground state shown in Fig.[lje) is a consequence of coopera- 
tive charge and bond ordering |9J . Predictions of the model 
include a large spin gap, charge ordering, and a complex 
structural modification involving both leg and diagonally 
transverse distortions. The strong c-polarized distortions 




Fig. 3. (Color online) Schematic mode displacement pattern for the 
out-of-phase C-C stretch of the DTTTF cation. 

observed in the spectra (Fig. [2]) are consistent with these 
predictions. Indeed, the zigzag ladder model can poten- 
tially also explain the occurrence of two distinct transitions 
in (DTTTF)2Au(mnt)2 but single transition behavior in 
(DTTTF) 2 Cu(mnt) 2 . Recall that the co-operative distor- 
tion of Fig. 1(e) requires a minimum interchain hopping 
tint > 0.59t, where t is the intrachain one-electron hopping 
integral 0. We believe that t int in (DTTTF) 2 Au(mnt) 2 
may be smaller than this threshold value at high tempera- 
tures, and the nearly independent stacks undergo the usual 
high temperature 4kp charge ordering transition observed 
also in the quasi-one- dimensional TMTTF-based ^-filled 
band systems |15j . The charge disproportionation is then 
exactly as in Fig. 1(e), but all bonds are uniform. This 
would give rise to the observed asymmetry between the 
stacks [5]. As the temperature is lowered to below 70 K, 
either because of lattice contraction or molecular rotation 
[TBI . ti n t is larger than the threshold value, the system 
behaves as the zigzag ladder in the spin singlet subspace, 
and the co-operative bond distortions of Fig. 1(e) occur. 
Within this scenario tint in the Cu-analog is already larger 
than 0.59i, which leads to the single observed transition 
involving both charge and spin. Complete theoretical proof 
of this scenario requires the demonstration of a single tran- 
sition in the | -filled band zigzag ladder with large tint and 
work is currently in progress. 

In order to evaluate the molecular charges in the low tem- 
perature phase, we followed the approach outlined by Ya- 
mamoto et. al. |17j and selected an unperturbed, infrared- 
active vibrational mode that, from our dynamics calcula- 
tions |8 , is sensitive to charge. Several authors have ob- 
served that vibrational frequencies depend linearly on av- 
erage molecular charge in the absence of strong coupling 
[14 . The advantage of using an unperturbed (rather than 
strongly coupled) mode for the analysis is that complemen- 
tary Raman data is not required |17j . Our analysis focused 
on an infrared-allowed out-of-phase C-C stretch of DTTTF 
(Fig. [3j. This mode is similar in character to ^27 in the 
BEDT-TTF building block molecule [17]. Accounting for 
the typical 3% overestimate of calculated frequencies and 
the flat vs. boat-shaped nature of the neutral vs. cationic 
building block molecules (see Fig.[4]and discussion, below), 
we compared the calculated results and the associated lin- 
ear fit (p — -0.022a-> + 29.82) with the experimental optical 
conductivity. From features at ~1319.5 and 1334.5 cm -1 in 
cti(w), we estimate the difference in the site charges in the 
charge-ordered low temperature phase to be ~ 0.35 - 0.4. 
These numbers should be compared with those obtained 
by Raman spectroscopy. The absence of strong low tem- 



•3 



[(DTTTF) 2 ][Cu(mnt) 2 ] Salt Neutral DTTTF 

S8 




Fig. 4. (Color online) (a) Bond lengths and relative orientation of 
the two independent cations in (DTTTF)2Cu(mnt)2, and (b) those 
of neutral DTTTF in the molecular crystal at 300 K The 
cations display an out of plane distortion in the salt (with a 5.78° 
angle between the thiophene rings and the TTF core) in addition 
to strong molecular asymmetry. In the neutral compound, DTTTF 
is nearly flat, although still asymmetric; both molecules in the unit 
cell are identical. Interestingly, the 300 K vibrational spectrum of 
the neutral DTTTF molecular crystal (not shown) also displays 
mode splitting that we take as an indication of intramolecular charge 
inhomogeneity. Asymmetry in the local molecular structure supports 
this assignment. Distances are in A. 

perature splitting of the C-H stretching mode on the cation 
(Fig. |2jf)) indicates that charge rich and charge poor re- 
gions are located at the heart of the DTTTF building block 
molecule. 

Although not part of the aforementioned zigzag model for 
intermolecular charge ordering, on-site charge asymmetry 
within the DTTTF molecule is expected from the chemical 
and structural point of view. Examination of the structure 
(Fig.ga)) shows that the DTTTF cations are locally dis- 
torted, with irregular bond lengths (and angles) |3l4j . This 
molecular asymmetry combined with the boat-type distor- 
tion of the cation gives rise to intramolecular charge inho- 
mogeneities, where the local concentration of charge is dif- 
ferent from the average charge on an individual molecule. 
An inhomogeneous on-site charge distribution is evident in 
the vibrational spectrum as additional mode splitting (for 
instance, fine structure on the C-S bending and stretching 
modes of the DTTTF cation as well as the combined C-C + 
C-S stretch + C-H wag and the complex S-C=C-H wagging 
motion of the DTTTF cation), overlaying the aforemen- 
tioned intermolecular charge ordering effects. Based on the 
small size of the additional mode splitting, the local charge 
distribution deviates only slightly from the average distri- 
bution at room temperature. The fine structure is more 
well-defined at low temperature, but the energy scale does 
not change very much. We therefore estimate | Ap\ = \p ave - 
Pioc I 'CO.l with respect to the average 4 K charge distribu- 



tion. Mode splitting of a similar size (~10 cm -1 ) has been 
observed in (TTM-TTP)l3 and attributed to an asymmet- 
ric deformation of the organic building block molecule and 
consequent non-uniformity of the intramolecular charge 
distribution |18j . Pressure may suppress the intermolecu- 
lar charge ordering and on-site charge inhomogencitics in 
(DTTTF) 2 Cu(mnt) 2 , leading to the metallic state [12] . 

We measured the variable temperature infrared response 
of the spin ladder candidate (DTTTF)2Cu(mnt)2 in order 
to assess the intermolecular charge ordering pattern within 
the context of two recent recent theoretical predictions. 
The low temperature distortion along the chain direction 
through the 235 K metal-insulator transition is consistent 
with expectations based on structural studies for a doubling 
along b. At the same time, there is a large transverse distor- 
tion. The presence of this transverse lattice distortion rules 
out rectangular ladder models and suggests that the zigzag 
ladder model with its cooperative charge and bond ordering 
may be more appropriate for (DTTTF)2Cu(mnt)2- Com- 
parison of these results with our previous work on the Au 
analog material shows that the metal-insulator transitions 
are quite different. Coupled chain- and rung-directed dis- 
tortions are observed in (DTTTF)2Cu(mnt)2, whereas they 
are separate in the Au compound. We suspect that this 
is a consequence of stronger interchain interaction in the 
Cu-based system. A mode analysis provides microscopic in- 
formation on the vibrational processes driving the metal- 
insulator transition in (DTTTF)2Cu(mnt)2- 

This research is supported by NSF-DMR-0600089 
(UT), DOE DE-FG02-06ER46315 (UA and MSU), DGI- 
CTQ2006-06333 (Spain), POCI/QVI/57528/2004 (Por- 
tugal), and MAGMANET. We thank J.T. Haraldsen, I. 
Olejniczak, and H. Mikre for useful discussions. 

* Presently at University College London. 



References 

[I] E. Dagotto and T. M. Rice, Science, 271, 618 (1996). 

[2] E. Dagotto, Reports on Progress in Physics 62, 1525 (1999). 
[3] C. Rovira, Structure and Bonding, Vol. 100 Springer- Verlag, 

Berlin Heidelberg, (2001). 
[4] X. Ribas et al., Adv. Funct. Mater. 15, 1023 (2005). 
[5] S. Maekawa, Science, 273, 1515 (1996). 
[6] Y. Yan et al., Phys. Rev. B. 73, 224418 (2006). 
[7] X. Ribas et al., Inorg. Chem. 44, 2358 (2005). 
[8] R. Wesolowski et al., Phys. Rev. B. 68, 134405 (2003); Phys. 

Rev. B. 70, 059901(E) (2004). 
[9] R.T. Clay and S. Mazumdar, Phys. Rev. Lett. 94, 207206 (2005). 
[10] T. Barnes, Phys. Rev. B. 67, 024413 (2003). 

[II] Sample requirements include a large volume (at least 1 cm 3 ) 
and full deuteration to avoid the large scattering cross section 
of hydrogen. 

[12] J.C. Dias et al., J. de Phys. IV. 114, 497 (2004). 

[13] C. Rovira et al., Angew. Chem. Int. En. 36, 2324 (1997). 

[14] M. Dressel and N. Drichko, Chem. Rev. 104, 56 89 (2004). 

[15] R.T. Clay et al. |http://arxiv.org/abs/0704.1656| 

[16] T. Mori et al. Bull. Chem. Soc. Jpn. 72, 179 (1999). 

[17] T. Yamamoto et al. J. Phys. Chem. B 109, 15226 (2005). 



A 



[18] R. Swietlik et al. Synth. Met. 150, 83 (2005).