Electron spin resonance spectroscopy of the organic conductor: (TMTSF)2PF6

Electron spin resonance spectroscopy of the organic conductor: (TMTSF)2PF6

207—211. Solid State Communications, Vol.35,in pp. Pergamon Press Ltd. 1980. Printed Great Britain. ELECTRON SPIN RESONANCE SPECTROSCOPY OF THE ORGAN...

682KB Sizes 0 Downloads 94 Views

207—211. Solid State Communications, Vol.35,in pp. Pergamon Press Ltd. 1980. Printed Great Britain.

ELECTRON SPIN RESONANCE SPECTROSCOPY OF THE ORGANIC CONDUCTOR:

(TMTSF) 2PF6

H.J. Pedersen* and J.C. Scott Laboratory of Atomic and Solid State Physics Cornell University, Ithaca, New York 14853 and K. Bechgaard Department of General and Organic Chemistry University of Copenhagen The H.C. Orated Institute Universitetsparken 5

DK—2l00 Copenhagen, Denmark (Received 11 Feburary 1980 A. G. Chynoweth)

The ESR g—factor, linewidth and spin susceptibility of (TMTSF)2PF6 are each found to have a distinct temperature dependence. The aniso— tropy of g and of the linewidth for static magnetic fields in the plane perpendicular to the highly conducting axis suggest that spin—phonon scattering is the dominant relaxation mechanism. The metal—insulator transition at 19 K is reflected particularly clearly in the magnetic properties.

The family of organic conductors (TMTSF)2x, where TMTSF is tetramethyltetraselenafulvalene and X is one of the inorganic anions PF6, AsF6, SbF6, BF4 or NO.~,has caused consid— conductivities in excess of 105 have erable recent interestl,Z. Low fllcm—l temperature been measured at ambient pressurel; metal insulator transition temperatures are low (~20 K for all except the BF~complex) in compari— son to most organic conductorsl; conductivity anisotropies are high (> 300 and > l0~ for the two perpendicular directions in

possible relationship between the ESR linewidth and the resistive scattering rate. Resistivity and thertnopower measurements show that the metal—insulator transition 1. It is in assumed (TMTSF)2PF6 that the transition is ± Peierls in nature but the relevant occurs at 19 1 K diffraction studies have not yet been completed. Our ESR measurements reveal several dig— tinct phenomena: the g—factor is temperature dependent, the change with temperature amounting to some 12% of the shift from the free electron value; the linewidth increases by a factor of 30

(TMTSP) 2PF6]; the metallic state is stabilized [in (TMTSF)2PF6 and in (TMTSF)2NO3] by the 2 and of (TMTSF) application only moderate pressure (~ 10 kbar) 2PF6 has recently been dig— 2. The motivation covered to be superconducting below 0.9forK at a pressure studying of this 12 class kbar of organic metals is in— creased further because there is only one type of molecular chain to participate in the con— duction process, and hence it is to be expected that the electronic structure is somewhat sJ.m— pler than in TTF—TCNQ and its two—chain ama— logs. In this communication we present a study of the ESR spectroscopy of one member of the fan— ily discussed in the preceding paragraph: namely (TMTSF) 2PF6. This compound was chosen follows very for the first study because1 the temperature de— closely we wished to explore the pendence p of~ T2, its and resistivity

between the metal—insulator transition and room temperature, but there is no simple relationship the resistive scattering rate; the aniso— tropy in the linewidth scales roughly with the magnitude ofin the as aalso function anisotropy the linewidth g—factor and with of the temperature; the spin susceptibility is temper— ature dependent; the linewidth increases dra— matically as the temperature is lowered below the transition. The samples used in this work were two sin— gle crystals prepared by electrochemical oxi— dation of TMTSF.3 Spectra were recorded using an X—band (9.4 GHZ) homodyne reflection ESR spectrometer and a six—inch Varian magnet. Tern— perature control was achieved with an Oxford In— with

strumentsESR—9 continuous flow cryostat and chromel thermocouple. Temperature stability was DTC 2 temperature controller using an iron:gold/ better than 0.1 K and the accuracy is estimated

to be within 0.5 K. In this apparatus the cavity temperature is constant and hence its Q— factor does not change. The crystal was mounted *Permanent address: Physics Laboratory UI, The Technical University of Denmark, Dk—2800

with the highly conducting a axis parallel to the radio—frequency magnetic field, H 1, and perpen—

Lyngby, Denmark 207

208

ELECTRON SPIN RESONANCE SPECTROSCOPY OF (TMTSF)

2PF6

dicular to the static magnetic field, H0. The sample could be rotated around that axis, there— by changing 4 Datathe were direction collected of using H0 in conventional the azimuthal derivative techniques, with immediate conversion plane. to digital form for storage and subsequent transfer to a computer for analysis. Care was taken to ensure that the modulating field was much less than the linewidth. It was found in— possible to saturate the line even with the maximum power available from the klyston (equivalent to H measured by comparison with a DPPH 1 = 1 gauss). The marker, g—factor was At all temperatures and for all orien— tations the ESR line was found to be, to within the noise, Lorentzian. Field sweeps were per— formed to lOx the linewidth at high tempera— ture and to more than 20x the linewidth near 20 K. The linewidth, LW, is defined via A

=

A(LW) 2 /[(H—H.~) 2 + (LW) 2

Vol. 35 No. 3

group, it is certain that the active spins reside entirely on the TMTSF chains. We attri— bute spin—orbit the change admixture in g of to unoccupied the variation molecular in the orbitals as the crystal undergoes thermal ex— pansion6 i.e. ~g(T) ~ AJA(T) where A is the spin—orbit coupling and I~ is some typical mole— cular electronic excitation energy and depends on the molecular environment. Similar mecha— nisms must be presumed to affect the g—values of all chains of organic radical ions. There— 7 clearly identifies the chains involvedet. in the fore, although the work of Tomkiewicz al. various phase transitions in the two—chain complexes, the quantitative decomposition of susceptibilities into separate chain contribu— tions must be viewed with caution. The temperature dependence of the ESR linewidth is displayed in Figure 2. Between the metal—insulator transition (at 19.0 I

I

I

I

±0.5

K

I

where A is the absorption of peak amplitude A 0 and HR is the center field of the absorption line. The data were computer fit to the deny— ative of this expression. The spin suscepti— bility (in arbitrary units) was then calculated as the product, A0 x (LW). is shown in Figure 1.to In The g—factor5 for H0 perpendicular the wide crystal face 2.048 ~ I

200

‘~‘2

-

6

-

(TMTSF)

.

2PF6 2.047-.

=

-



I—

2.046

-

-

2.045—

-

S

:

2.044-

-

-





S

2.O43

-

2.042—

.S.c

S

2.041



o

-

0 2.040

I

0

I

I

I

100 200 TEMPERATURE (K)

I 100

I

200

I

300

TEMPERATURE(K)

300 2. Temperature dependence of the ESR linewidth, H0 parallel to 001.

1. Temperature dependence of the g—factor of (TMTSF)2PF6, with H~,in the 001 direction. The

scatter in the data arises from the large line— width at high temperature. Below about 30K the uncertainty in g is comparable to the size of

and clearly identifiable in the data) and room temperature the linewidth increases by a factor 2 in does (TMTSF) of 30. This increase clearly not follow

the data points, and the rapid change below 20K is a real effect,

the dependence of p[p, T 2PF6] which is found for a three dimensional 8: metal where the transverse relaxation rate T —l ~ (ISg/g)2r’~ 2 and r is the resistive scattering time. Also, T 2]~at room temperature is an order of magni— tude smaller than the 3—U theory predicts.

spite of the experimental uncertainty (which is due primarily to the relatively large line— width) it is clear that there is a distinct in— crease in g as the temperature is lowered. In view of the high electron affinity of the PP6

Vol. 35, No, 3

ELECTRON SPIN RESONANCE SPECTROSCOPY OF (TNTSF)

2PF6

Nor is the behaviour like that of9 TTF—TCNQ, where T TSP— TCNQ, and many of their analogues 2l decreases with increasing temperature. Quali— peaks slightly above the Peierls transition then tatively the behaviour is more like that of th~ single—chain conducting complexes based on TTP 0 or TMTTF6, although as one might expect for the selenium containing molecule the linewidth is two orders of magnitude larger. It is tempting to view the function T 2l(T) as falling into two regimes. At high temperature (T ~ 100 K) the dependence is quite linear in T, whereas at low

temperature the dependence is much faster. that is well cubic Over the behaviour limited range 20
S

,

I

I

decrease in bandwidth, interchain correlations 12 and the depression of the density of states The present with dataPeierls—Fröhlich merely confirm that this is associated fluctuations. common behaviour, and do nothing to clarify its origin. The rapid decrease in x(T) below the tram— sition is worthy of comment. The accompanying sudden increase in both the linewidth and the g— factor might suggest that the transition has some magnetic component. This is particularly puzzling in view of the preliminary static mag— netic susceptibility 1 These measurements measurements which areshow cur—no rently being in order to clarify the anomaly at 19 repeated K. situation, in particular to determine the magnitude of the magnetic energy gap in the low temperature region where the rapidly increasing linewidth and weakening signal preclude9 measurement by integration of the ESR signal. Finally we turn our attention to the anisotropy in the plane perpendicular to the a axis. Measurements of g and LW were made at several different temperatures and orientations as indicated in Figures 4 and 5. There is clearly a correlation between the anisotropies of the two quantities; Sg(Q) and LW(9), both of which reach maxima near 140°and minima near 50°. Therefore one is led to believe that the relaxation is due to spin—orbit coupling and electron—phonon scattering. If we extend the the relationship of Elliott8 to the aniso— tropic case, then —l 2 —1 T 2 (~,T) [dg(9,T)] r(T,9) F~l~1 J. 2

I

I



6

I

I

I

I

I

I

(TMTSF)2PF6

I- I’ 1ik.~

(TMTSF)2 PF6 7

209

II



I

~ I’ I

2.045k\.

I

I

I

“•-~

./

/

L~.,”.



I

. 1P/./



[‘<‘~..‘~

• 5

-

S



[



\“.

•\

-

\

2.040



‘•. •

\\ .

Z

.1

\.

s\”,.

—o./

~ ~.

~\

.

2.035



I

/

\ST~5O~/’

\‘a 2

.

~/

0

I

I

I

I 100

I I

I

I

200

..

I

-

/

I I

I

I

(‘~

-

~

I





\

/_

~

/

1=100K

1=200K

-

~‘

‘/

°T~K.~/

“.. \

.

i

~.//~/

\.

I I

I

I

I

300

/

-

TEMPERATURE (K) 3.

Temperature dependence of the spin—suscepti—

1=300K

bility. 2.025

I

0 line as described above, reveals the increase with temperature commonly observed in highly—conducting organics. This behaviour has been variously attributed to band—structure effects, dilation of the sample with ensuing

I

I

I

I

I

I

I

I

I

50 00 Angle of Rotation (degrees)

I

I

I

I

50

4. Anisotropy of the g—factor in the plane per— pendicular to the a—axis for several different temperatures. The direction marked 2 is (001).

210

ELECTRON SPIN RESONANCE SPECTROSCOPY OF (TMTSF)

2PF6

[~.2

I

I I I (TMTSF)

I 6I

I

I

2PF

200

I

I

~

~

I

I

I

I

that this is not the back—scattering, k’—k

-

2kg, since this process has no spin—flip 13]. If channel the same ofscattering mechanism app’ies for all because time reversal invariance field orientations one expects r.,. to be independent of 9. We show that this is not the case in Table 1 where the ratio [T 2/(6g)2]~50/[T2’~’1./(~5g)2]9..,140o is evaluated. At low temperature this ratio is unity,

-

•__=— 10

~ 50

-

but increases by aonefactor 2.5 byfor300phonon— K. This is the behaviour might ofexpect

-

=

scattering across the Fermi surface of a metal at low temperature, crossing over to anisotropic diffusion at high temperature. To summarize: this ESR study of the single— organic—chain conductor, (TMTSF)2PF6, has shown that the g—tensor, linewidth and susceptibility

I—

a w

z

~ 100

~

-

Vol. 35, No. 3

Te— is the electron—phonon scattering where rate whic~contributes to the relaxation. [Note

V

each has a distinct temperature depen— dence. The temperature and orientation depen— dence of the linewidth suggest that it would be fruitful to develop a theory describing a low -

-

temperature regime of band motion with diffusion

-

becoming dominant as the temperature increases. —o

The metal—insulator transition at 19 K appears very sharply in the magnetic behaviour.

~

0~__1_~I~”T~ 0 _____ 50 ~______ I I

I

I

100 I

I ______ I I

I

T=22K~ ______ 150 I I j

Acknowledgement—We Silsbee cal reading for helpful of the manuscript. are discussions grateful HJP and to Prof. for aR.H. thanks criti— the

Angle of Rotation (degrees) 5. Anisotropy of the linewidth in the plane perpendicular to the a—axis at various temper—

Danish Research Council for support during his visit to Cornell University. JCS acknowledges partial support from the Cornell Materials Science Center under the NSF—MEL program,

atures as indicated,

grant #DMR—76—81083AO2.

TABLE

I

2 K Temp.

Gauss Linewidth(LW)

LW/(ôg)

g-shift(ög)

Ratio [ui/(6g)2~ 0e

50°

140°

50=

140

50°

300

213

238

.024

.040

3.7xl0

200

163

186

.033

.044

100

80

99

.034

50

27

36

22

6

8

140° l.5xl05

2.5

±0.7

1.5

.96

1.6

±0.3

.043

.69

.53

1.3

±0.1

.036

.044

.21

.19

1.1

±0.1

.038

.046

.042

.038

1.1

±0.1

5

Anisotropy of the linewidth and g-shift at several temperatures.

The values given are

those corresponding to the maxima (140°) and minima (50°)of figures 4 and 5. These quantities are used to compute the values of LW/(5g)2 and the ratio of LW/(Og)2 between the two orientations.

-

Vol. 35, No. 3

ELECTRON SPIN RESONANCE SPECTROSCOPY OF (TMTSF)

2PF6

REFERENCES 1. K. Bechgaard, C.S. Jacobsen, K. Mortensen, H.J. Pedersen and N. Thorup, Commun. (in press). 2. D. Jerome, A. Mazaud, N. Ribault and K. Bechgaard, J. Phys. Lett. (Paris), (in press). 3. K. Bechgaard (unpublished results). 4. Note that the crystal structure is triclinic (see Reference 1) and so the plane of H0 is not the bc plane, but it is perpendicular to the highly conducting direction. 5. This direction (denoted as “.2” in Refer— ence 1) is the (001) direction. 6. A similar g—dependence is found in the sul— phur analogues; P. Delhaes, C. Coulon, J. Amiell, S. FLandrois, E. Toreilles, J.N. Fabre and L. Giral, Mol, Cryst. and Liq. Cryst. 50, 43 (1979). 7. Y. Tomkiewic~7A.R. Taranko and R. Schu— maker, Phys. Rev. Bl6, 1380 (1977). and references therein.

8. R.J. Elliott, Phys. Rev. 96, 266 (1954). See also Reference 13. 9. Y. Tomkiewicz, Phys. Rev. Bl9, 4038 (1979), and references therein. 10. Y. Tomkiewicz and A.R. Taranko, Phys. Rev. !~., 733 (1978). 11. A.N. Bloch, p. 317 in Proceedings of the Conference on Organic Conductors and Semiconductors, Siofok, Hungary 1978. (Springer—Verlag, Berlin 1976). 12. For extensive recent reviews see A.J. Heeger, pp. 69—145 and T.D. Schultz and R.A. Craven, pp. 147—225 in Highly Con— ducting One—Dimensional Solids, J.T. Devreese, R.P. Evrand and V.E. van Doren, eds. (Plenum, New York, 1979). 13. Y. Yafet, p. 1 in Solid State Physics, Vol. 14, F. Seitz and D. Turnbull (Acad. Press, New York, 1963).

211