Bulk-Sensitive Angle-Resolved Photoemission

Spectroscopy on TTF-TCNQ

Kenji KOIZUMI1�, Kyoko ISHIZAKA2, Takayuki KISS

1y,Mario OKAWA

1z, Reizo KATO3, and Shik SHIN1;4

1ISSP, University of Tokyo, Kashiwa, Chiba 277-8581, Japan2Department of Applied Physics, University of Tokyo,

Bunkyo, Tokyo 113-8656, Japan3RIKEN-ASI, Wako, Saitama 351-0198, Japan

4RIKEN SPring-8, Sayo, Hyogo 679-5143, Japan

(Received September 26, 2012; accepted December 26, 2012;

published online January 28, 2013)

KEYWORDS: laser-excited angle-resolved photoemission

spectroscopy, bulk-sensitivity, organic conductor,

TTF-TCNQ, charge density wave

TTF-TCNQ is a quasi-1D molecular conductor exhibitingmetal-insulator transition at 54K accompanied by charge-density-wave (CDW) formation. Its electronic properties inrelation to the Peierls instability and CDW transitions arewell investigated through the several decades.1–7) PastARPES studies indicate the band dispersion consisting ofTCNQ molecules (namely the TCNQ-band) shows thesignature of spin–charge separation, together with thesuppression of the quasiparticle excitation at the Fermilevel.8–10) These observations have been discussed in termsof the 1D character of TTF-TCNQ, nevertheless, there areseveral mysteries remained. One is the band dispersionsobtained by ARPES commonly indicating the band-width(or the band group velocity) larger by a factor of �2 withrespect to band theory.8) To account for this discrepancy, thepossibility of relaxed tilting of the topmost molecules and/orstrong correlation effect have been raised until now.8,11) Thesecond is the anomalous suppression of the spectral weightnear the Fermi level (EF), which gives negligible amount ofspectral intensity at EF even in the highly-conductive stateabove >60K.9,10) The electronic structure of the metallicquasi-1D state has been thus remained to be elucidated.

To investigate the precise electronic structure of TTF-TCNQ, here we utilized the laser-excited ARPES withh� ¼ 6:994 eV.12) Owing to the relatively low photon-energyas compared with past ARPES studies (h� ¼ 20{40 eV), thebulk-sensitive measurement becomes available. The effectof photoirradiation-induced damage peculiar to molecularcompounds is also considered to be weaker compared to themeasurements using higher-energy photons. The high crosssection of s and p electrons for this photon-energy providessome advantage in studying the light-element based organicconductors.

ARPES measurements were performed using a systemconstructed with VG-Scienta R4000 electron analyzer andan ultraviolet (h� ¼ 6:994 eV) laser.12) He II� (h� ¼ 40:8eV) light source was also used to check the correspondence

with past reports. Single crystals of TTF-TCNQ grown bythe diffusion method were cleaved in-situ to obtain freshsurfaces. The pressure was below 5� 10�11 Torr throughoutall the measurements. The energy resolution was �E ¼4:0meV. The Fermi level EF of the sample was referred tothat of a Au film evaporated on the sample substrate, with anaccuracy of �0:1meV.

Figure 1(a) shows the mapping of the ARPES intensity(h� ¼ 6:994 eV) at the Fermi level, obtained by integratingthe ARPES intensity in the energy window of �50meV. Theshape of the Fermi surface thus observed makes a straightline at kF ¼ 0:24� 0:01 �A�1 ¼ ð0:29� 0:01Þb�, indicativeof 1D electronic structure. The image of band dispersionsalong �–Z direction obtained by h� ¼ 6:994 and 40.8 eV areshown in Fig. 1(b,c). The peak positions of the ARPESintensity estimated from momentum distribution curve(MDC) and energy distribution curve (EDC) are plotted bycircle and cross markers, respectively. The red and bluecurves, on the other hand, represent the band dispersionsconsisting of TTF and TCNQ molecules, obtained by thefirst-principles calculation.13) By comparing with the bandcalculation, we can notice that the h� ¼ 6:994 eV data isdominated by the signal from TTF bands, whereas theh� ¼ 40:8 eV data shows both TTF and TCNQ bands thatare well in accord with past reports. This h�-dependence,which should be due to the photoionization cross sectionand/or matrix element effect, also explains the slightdifference of the present ‘‘Fermi-surface’’ shape from thatpreviously reported.14) The Fermi momentum obtained byARPES, kF ¼ 0:24� 0:01 �A�1, is somewhat smaller com-pared with the band calculation, reflecting the overestima-tion of the charge transfer between TTF and TCNQinevitably arising in the calculated data.

Here, let us focus on the gradient of the TTF banddispersion. In the h� ¼ 40:8 eV ARPES spectrum, thegradient of the TTF band dispersion is about 2 times largeras compared to the calculation. It shows that the observed

High

Low

Inte

nsity

(a)

hν = 6.994 eV

0.60.40.2Momentum along b* (Å

-1)

0.8

0.6

0.4

0.2

0.0

Bin

ding

Ene

rgy

(eV

) TCNQ

TTF

(c)

hν = 40.8 eV

0.60.40.2Momentum along b* (Å

-1)

0.8

0.6

0.4

0.2

0.0

Bin

ding

Ene

rgy

(eV

)

High

Low

Intensity

TCNQ

TTF

(b)

hν = 6.994 eV

Γ

Y

Z

Fig. 1. (Color) (a) ARPES intensity mapping at EF of TTF-TCNQ,

obtained by using h� ¼ 6:994 eV light. (b, c) ARPES image along �–Z

direction recorded by h� ¼ 6:994 and 40.8 eV, respectively. Red (blue)

curves indicate the band dispersions consisting of TTF (TCNQ) molecules,

obtained by the band calculation. All data are recorded at 60K.

yPresent address: Graduate School of Engineering Science, Osaka

University, Toyonaka, Osaka 560-8531, Japan.zPresent address: Department of Applied Physics, Tokyo University of

Science, Shinjuku, Tokyo 162-8601, Japan.

Journal of the Physical Society of Japan 82 (2013) 025004

025004-1

SHORT NOTES

#2013 The Physical Society of Japan

http://dx.doi.org/10.7566/JPSJ.82.025004

band-width of the TTF band is significantly larger than thatexpected by calculation. On the contrary, the h� ¼ 6:994 eVdata shows that the gradient of the TTF band dispersionis much smaller compared to h� ¼ 40:8 eV, and almostequivalent to the calculation. Since the kz-dependence of thecorresponding band dispersion is very small,13) it can beunderstood by the difference of the bulk sensitivity betweenh� ¼ 6:994 and 40.8 eV ARPES, of which probing depthsare known to be >80 �A and several �A, respectively.15) Thisresult indicates that the long-standing inconsistency of theband-width is due to the surface effect, not the correlationeffect. The tilting of topmost molecules, however, has beenexperimentally ruled out in Ref. 11. The precise mechanismof band-width enhancement at the surface remains to befurther investigated.

Next we show the temperature-dependent EDC at theFermi momentum in Fig. 2(a), obtained by h� ¼ 6:994 eVARPES. With lowering the temperature from 200 to 60K,the spectral weight near the Fermi level increases and formsa broad peak at �0:2 eV. At the Fermi level, the spectralweight is suppressed and shows no quasiparticle peak orclear Fermi-edge cutoff even in the metallic phase (>60K).By closely looking at the Fermi-level, nevertheless, a very

small but finite intensity is found at high temperatures. TheARPES intensity Iðk; !; T Þ at momentum k, energy !,and temperature T is practically given by Iðk; !; T Þ ¼MAðk; !; T ÞfFDð!; T Þ,16) where M is the matrix element,Aðk; !; T Þ is the single-particle spectral function, andfFDð!; T Þ is the Fermi–Dirac distribution function. To seethe temperature dependence of the spectral function itself,we divided the EDCs by the Fermi–Dirac distributionfunction convolved by the gaussian energy-resolutionfunction, as shown in Fig. 2(b). Here we can see that thereis a finite intensity at the Fermi level at 200K, which slightlyincreases on cooling down to 60K, the most conductivetemperature. On further cooling down to 6K, the long-range2kF-CDW formation occurs at 54K, which opens a CDWgap at the Fermi level, observed as the leading edge shift ofabout �30meV. This value of gap is mostly consistent withthat obtained by optical spectroscopy.7)

In conclusion, we investigated the electronic structure ofTTF-TCNQ by using laser-excited ARPES. The observedTTF band shows that the band-width agrees well with theband theory, and small but finite spectral intensity exists inthe metallic state. These results show the effectiveness of thebulk-sensitive laser ARPES measurement for studying thevariety of organic conductors.

�koizumi@issp.u-tokyo.ac.jp

1) M. J. Cohen, L. B. Coleman, A. F. Garito, and A. J. Heeger: Phys.

Rev. B 10 (1974) 1298.

2) S. Etemad: Phys. Rev. B 13 (1976) 2254.

3) T. Ishiguro, S. Kagoshima, and H. Anzai: J. Phys. Soc. Jpn. 41 (1976)

351.

4) J. R. Cooper: Phys. Rev. B 19 (1979) 2404.

5) S. Kagoshima, T. Ishiguro, and H. Anzai: J. Phys. Soc. Jpn. 41 (1976)

2061.

6) K. S. Khanna, J. P. Pouget, R. Comes, A. F. Garito, and A. J. Heeger:

Phys. Rev. B 16 (1977) 1468.

7) D. B. Tanner, C. S. Jacobsen, A. F. Garito, and A. J. Heeger: Phys.

Rev. Lett. 32 (1974) 1301.

8) M. Sing, U. Schwingenschlogl, R. Claessen, P. Blaha, J. M. P.

Carmelo, L. M. Martelo, P. D. Sacramento, M. Dressel, and C. S.

Jacobsen: Phys. Rev. B 68 (2003) 125111.

9) F. Zwick, D. Jerome, G. Margaritondo, M. Onellion, J. Voit, and M.

Grioni: Phys. Rev. Lett. 81 (1998) 2974.

10) R. Claessen, M. Sing, U. Schwingenschlogl, P. Blaha, M. Dressel, and

C. S. Jacobsen: Phys. Rev. Lett. 88 (2002) 096402.

11) M. Sing, J. Meyer, M. Hoinkis, S. Glawion, P. Blaha, G. Gavrila, C. S.

Jacobsen, and R. Claessen: Phys. Rev. B 76 (2007) 245119.

12) T. Kiss, T. Shimojima, K. Ishizaka, A. Chainani, T. Togashi, T. Kanai,

X.-Y. Wang, C.-T. Chen, S. Watanabe, and S. Shin: Rev. Sci. Instrum.

79 (2008) 023106.

13) S. Ishibashi and M. Kohyama: Phys. Rev. B 62 (2000) 7839.

14) T. Ito, A. Chainani, T. Haruna, K. Kanai, T. Yokoya, S. Shin, and R.

Kato: Phys. Rev. Lett. 95 (2005) 246402.

15) M. P. Seah and W. A. Dench: Surf. Interface Anal. 1 (1979) 2.

16) A. Damascelli, Z. Hussain, and Z. X. Shen: Rev. Mod. Phys. 75 (2003)

473.

Inte

nsity

(a.

u.)

0.8 0.4 0.0Binding Energy (eV)

(a)

6 K 60 K 200 K

hν = 6.994 eV

EDC at kFIn

tens

ity /

FD

(a.

u.)

0.3 0.0Binding Energy (eV)

(b)

6 K 60 K 200 K

Fig. 2. (Color) (a) Temperature-dependent EDC at the Fermi momentum

of TTF-TCNQ, obtained by h� ¼ 6:994 eV. (b) EDC in (a) divided by the

Fermi–Dirac function convolved by the energy-resolution gaussian function.

K. KOIZUMI et al.J. Phys. Soc. Jpn. 82 (2013) 025004 SHORT NOTES

025004-2 #2013 The Physical Society of Japan