PHYSICAL REVIEW
8
VOLUME 41, NUMBER 7
1
MARCH 1990
Phase transition and non-Ohmic electrical transport in the spin-density-wave state
[(TMTSF) 2PF6]
hexaluorophosphate
of the organic conductor tetramethyltetraselenafulvalinium
W. Kang, S. Tomic, J. R. Cooper, and D. Jerome
Laboratoire de Physique des Solides, Universite Paris-Sud, 91405 Orsay, France
(Received 8 September 1989; revisal manuscript received 4 December 1989)
We report the electric field dependent conductivity in the spin-density-wave (SDW) state of the
organic conductor (TMTSF)2PF6 using two different methods for making contacts. A new
clamped~ntact method reveals that the onset of the SDW transition of (TMTSF)2PF6 may be
first order when the sample is strain free. A non-Ohmic behavior of the conductivity is observed
above a threshold field in the range of 1-5 mV/cm. The temperature dependence of the threshold
between the SDW and the underfield agrees with a pinning mechanism due to commensurability
lying lattice. The use of classical painted contacts leads to a smearing of the originally sharp
transition and modifies the temperature dependence of the threshold field.
hexafluorophosTetramethyltetraselenafulvalinium
phate [(TMTSF)2PF is one of the most studied singlechain organic conductors. At ambient pressure it has a
(SDW) state below 12
semiconducting spin-density-wave
K whose presence has been clearly established by various
X-ray studies show that it is a
magnetic measurement.
pure SDW state in that the 2kF scattering, which is the
(CDW) instability,
precursor of a charge-density-wave
disappears below 50 K. Early experiments have showed
the existence of an extra conductivity in the SDW state
that depends on the applied electric field and the frequency. 4 s The electron-spin resonance signal was also restored by a large microwave field. However, no threshold
field and no evidence for wide- or narrow-band noise or
for a giant dielectric constant could be clearly identified. 5
Subsequent studies were unable to reproduce the current
carrying "spin-resurrected" state and also suggested that
were influenced
measurements
transport
by heating
effects. Moreover, the observed non-Ohmic conductivity
was correlated with the presence of irreversible jumps in
the resistance associated with cracks developing on coolHowever, conductivity measurements at miing down.
crowave frequencies indicate the existence of a collective
response associated with the SDW ground state. s
In a recent Letter" we have reported experiments on
the electric field dependent conductivity in the SDW state
of a related compound (TMTSF)2NO3. There was an evidence for increased conductivity above a finite threshold
field (Er) and the sliding SDW mode was suggested as a
plausible mechanism for the observed nonlinearity. While
the value of ET 40 mV/cm is close to those found in
CDW materials, it was found to be temperature independent below T,/2 in contrast to CDW systems where a slow
increase logioEr A —
BT is generally observed. ' In
Ref. 11, we were unable to follow the T dependence of Er
between T,/2 and T, or to check the sharpness of the
threshold using continuous electric fields, because of heating. '
In this Rapid Communication we report the observation
of non-Ohmic conductivity above a finite threshold field
associated with the SDW phase of (TMTSF)2PF6. We
believe that this is an important result because in contrast
J
'
to (TMTSF)2NO3 the complication of anion ordering is
absent in (TMTSF)2PFs. As far as (TMTSF)2PFs is concerned, the SDW phase has been studied in some detail by
proton NMR experiments. '4'5 Both studies have led to
an
estimate
of the magnetic-distortion
wave
vector
Q (0.5a, 0.24+' 0.03b ) (Ref. 14) and (0.5a, 0.20
~0.05b ) i.e., Q is close (or even equal) to the commensurate value (0.5a, 0.25b ), ignoring the third,
weakly coupled c direction.
were performed on single crystals of
different batches with various
lengths (0.5-2.5 nm) and typical cross sections of about
0.004 mm2. Gold pads were evaporated onto the samples.
Two different mounting techniques were used. Electrical
contacts were made either by silver paint or by mechanical clamping of fine gold wires. 's For samples with paint2 K/h) were employed
ed contacts, slow cooling rates
but all the samples measured showed some irreversible
resistance jumps associated with microcracks. However,
the non-Ohmic behavior disappeared above T, and the
threshold field observed for a given batch of samples was
the same, irrespective of the resistance changes associated
with jumps. Apart from the sharp jumps all samples measured showed the well-known metallic behavior down to
the SDW transition at T, 11.5~0.25 K as defined by
the maximum of dlogioR/dT. We found a different resistivity ratio (RR) between 300 and 13 K ranging from 3.3
to 42. However, samples with clamped contacts did not
show even the slightest visible crack in the whole temperature region. Transition temperatures were in the range
12+ 1 K, and the RR was around 90.
A comparison between curves a and b in Fig. 1 reveals a
marked difference between painted and clamped contacts
as far as the temperature dependence of the resistance is
concerned. For the clamped contact technique an extremely sharp metal-insulator (MI) transition is observed
12-5 K a nearly
domain
and in the temperature
temperature-independent
gap can be derived from the
data in Fig. 1, curve a (2h 23 K). In addition, a resistance increase is clearly observed below 4 K or so. On the
other hand, using the painted contact technique, a temperature dependence close to the classical BCS behavior was
The measurements
(TMTSF)2PFs from
(=
4862
@1990The
American Physical Society
PHASE TRANSITION AND NONWHMIC ELECTRICAL.
~
10
0
0pp
p
pp
~
~
I
I
I
I
I
(n)
I.SK
0
+ 7.4K
~ 8.7K
C)
6
& 12.7K
iiIQ2 "
10
W
QD
D
CO
UJ
10
100
I
8
'
'
'5'
'
'
'iO'
I
I
I
I
100
200
300
400
k p
~~~ gQ
I
I
d
r
r
500
10
0.6
Iooo/T (K ')
FIG. I. Overall temperature
dcpcndcncc of (TMTSF)iPF~
with two different methods for making contacts. Inset shows the
temperature dependence of the threshold 6eld. Curve a, for the
sample with clamped contacts and curve b for that with painted
contacts.
I
4-
Q.
C3
02-
+
+ + +
Q.
-0
++/+++A
~
o
s
K
24K
IIQQK
o
4 ~
IS
XO~ee&%C
~
4
a+
aa ma
i
5
aagS
4
4
P
found with a similar value for 2h (28 K).
The experimental apparatus used to search for electric
field dependent transport was similar to the one used in
Ref. 11 with pulse durations of about 40 ps and a repetition rate of 50 ms. We also performed regular checks for
sample heating to rule out spurious effects and made additional measurements in the metallic region, at about 25 K,
and between 80 and 100 K.
Figure 2(a) shows the field dependent conductivity for
clamped contact samples at various temperatures versus
the applied pulsed electric field. The pulse method avoids
most sample heating problems but only has a resolution of
1%. Thus data in Fig. 2(a), e.g. , those at 4.6 K, reveal
an Ohmic character of the conductivity, n ec at low
fields in a broad field domain and a threshold field at
ET 5.5 mV/cm, which we define as the intercept between ere and the extrapolation of the field dependence
above Er. Furthermore, under favorable circumstances,
e.g. , a thin sample at low temperatures or one directly immersed in superfluid helium we were able to observe a fiat,
field independent conductivity with the much more sensitive dc technique which implies that the threshold field
must be finite (of the order of 5 mV/cm or so). However,
at higher temperatures sample heating becomes important
even below ET as higher currents must be passed through
the sample. Then there is a curvature below Er but we
were still able to see a wellMefined effect near to Er by
plotting the apparent sample temperature (as determined
from the conductivity itself) versus power. In the latter
case the continuous curvature of o(V) vs V was similar to
that reported in Ref. 5. Therefore, in our opinion, heating
is one possible reason for the absence of a clear threshold
reported in Ref. 5. However, we have also found that
due to microsamples with very low RR values
cracks do not show a well-de6ned threshold.
The non-Ohmic conductivity behavior of painted contact samples is displayed in Fig. 2(b). At 24 and 100 K
the conductivity remains constant over two decades of
o0
4 Q.SK
vQ
0
+f
o 42K
4 8.5K
LU
(-I)
I
T(K}
~ f.7 K
=
I
I
E(mV/cm)
b
'0'
Og p
Og
D
0
~ po ao
Pn
CC
44 o
D
0
1
M
I
I
I
0 6.0K
o P.I. -
h
CI
U)
4863
r
10
~
CJ
..
Inst
&~ It
aa
4 aa aaa
0
0 II
I
I
&0
50
4
44
4
v
XQ
. .
~
XOa
I
100
E(mV/cm)
FIG. 2.
e(E
Non-Ohmic
the logarithm
conductivity
Ie(E) —e(E
0)]/
of electric field (E) at various tcmpcratures of (TMTSF)iPFt; (a) for sample with clamped contacts and (b) for that with painted contacts. Inset of (b) shows
0) vs
the comparison between the experiments
current and pulse current at 4.2 K.
with
continuous
electric field. Below 12 K a non-Ohmic field dependence
is observed above a certain threshold field ranging from 20
to 7.5 mV/cm at 12 and 2 K, respectively. As indicated in
thc inset of Fig. 2(b), dc measurements at 4.2 K confirm
the existence of the threshold field and the value obtained
corresponds fairly well to the pulsed data also shown in
the inset. We have also noted that the magnitude of the
excess conductivity is smaller for samples with smaller
RR as already observed for (TMTSF)sNO3 and that samples from another batch with a broader MI transition have
140 mV/cm at 1.7 K). The
much larger Ez values
observation that Er can be 15-30 times larger for crystals
with broad MI transitions is important since it implies
that ET is very sensitive to the impurity content and this
could explain why some previous investigators have failed
to detect any threshold behavior. Note that if there are
microcracks the RR is not a good indicator of the impurity content.
The onset of non-Ohmic conductivity at the temperature of three-dimensional
SDW order strongly suggests
that the nonlinearity is associated with the establishment
of a SD%'. Indeed, as was previously argued, it is difBcult
to explain thc observed effects using models based on a
Moreover, another highlight of
single-particle picture.
the present data is the behavior of the low electric field
conductivity versus temperature in samples studied using
the clamped contact technique. The sharpness of the
(=
"
W. KANG, S. TOMIC,
4864
J. R. COOPER,
resistive transition as shown in Fig. 1, curve a is somewhat
suggestive of the establishment of three-dimensional magnetic ordering via a first-order transition with the magnetic order parameter jumping from zero to a finite value at
the transition. Indeed, a similar behavior of the order parameter (e.g. , non-BCS temperature dependence) has already been observed in the study of the temperature
dependence of the magnetic broadening of the proton
NMR spectrum which jumps from zero to half the zerotemperature value at 12 K. '
On the other hand, experiments sensitive to the lattice
dynamics have not shown any measurable discontinuity in
either the lattice parameter'
or the sound velocity. '
Consequently, the first-order character of the MI transition seems to involve only the magnetization
(e.g. , the
spin-phonon coupling is small). At this stage we can think
of one possibility for the first-order character of the transition if the wave vector of the magnetic modulation is
commensurate, i.e., Q (0.50a, 0.25b ) an hypothesis
which is compatible with the present knowledge of NMR
properties. A similar first-order transition between a normal state and a commensurate CDW state has already
been encountered in the study of the phase diagram
(TTFof tetrathiafulvalene-tetracyanoquinodimethane
TCNQ). n In a narrow domain around of 19 kbar the
CDW of TTF-TCNQ becomes commensurate (x 3) with
the underlying lattice and the MI transition becomes first
order instead of second order. A temperature hysteresis
5%) is thus observed at the transition of T'1FTCNQ. However, in case of the SDW transition we have
failed so far to detect any hysteresis between cooling and
warming runs. The existence of a first-order SDW transition is also reminiscent of that reported in unstrained
chromium at TN 311.5 K.
In addition, the resistive anomaly which is located
around 4 K in Fig. 1, curve a may be related to the transition towards a "more insulating" state detected by NMR
at the same temperature. 3 z4 The difference in behavior
between clamped and painted contacts is still unclear.
However, we tend to believe that the method which
guarantees the absence of microcracks during cooling
down is more likely to provide an intrinsic behavior.
Finally, we come to the discussion of the threshold field
ET which displays different behavior with temperature depending on the contact technique (inset of Fig. 1). For
painted contacts the increase of Er close to T, is qualitatively reminiscent of the behavior found in the CDW system. However, the threshold field does not diverge at T„
(=1.
'
address: Institute of Physics of the University, P.O.
Box 304, Zagreb, 41001 Yugoslavia.
'K. Mortensen, Y. Tomkiewicz, T. D. Schultz, and E. M.
Engler, Phys. Rev. Lett. 46, 1234 (1981).
W. M. Walsh, F. Wudl, E. Aharon-Shalom, L. W. Rupp, J. M.
Vandenberg, K. Andres, and J. B. Torrance, Phys. Rev. Lett.
49, 885 (1982).
J. P. Pouget, R. Moret, R. Comes, K. Bechgaard, J. M. Fabre,
and L. Giral, MoL Cryst. Liq. Cryst. 79, 129 (1982).
4W. M. Walsh, Jr., F. Wudl, G. A. Thomas, D. Nalewajek, J. J.
Permanent
AND D. JEROME
as for most CDW materials. ' Furthermore, the negative
temperature coefficient of ET below T, which has been observed for CDW is clearly absent here. Data in the inset
of Fig. 1, curve b, are consistent with those obtained in
(TMTSF)2NO3 using the painted contact technique"
with ET constant below T,/2. The inset of Fig. 1, curve b,
shows ET(T, )/ET(1. 7 K)
5 in fair agreement with the
theory of Maki and Virosztek
predicting an increase of
Er of 3. 13 and 1.77 for the weak pinning limit in D 3
and 2 dimensions, respectively. However, in samples with
clamped contacts ET display a minimum below T, and a
steep divergence at T, . The different behavior of ET vs T
may again be understood in terms of commensurability
(or incommensurability)
of the SDW. There exist two
limiting situations: In a clean sample the SDW is pinned
potential leading to a temperature
by a commensurability
dependence of ET as shown in Fig. 1 of Ref. 26; on the
other hand, for a sample containing more defects (possibly
introduced by microcracks) the impurity potential may be
the dominant pinning mechanism. It is worth noting that
if this interpretation
is correct, it implies that the
fourfold-commensurate
SDW considered here still has a
small threshold field. This can be understood within the
theoretical model of Maki and Virosztek. 26 In contrast,
for the CDW mentioned above, there was a large increase
in ET in the region of commensurability.
In conclusion, using a new method for electrical contacts, the resistive transition at the onset of the SDW of
(TMTSF)2PF6 appears to be first order. A careful examination of the thermodynamic
properties at the SDW
transition of a strain-free (TMTSF)2PF6 crystal should be
enlightening.
A non-Ohmic behavior of the conductivity
is observed above a threshold field in the range of 1-5
mV/cm. The temperature dependence of the threshold
field agrees with a pinning mechanism due to commensurability between the SDW and the underlying lattice.
The use of classical painted contacts leads to a smearing
of the originally steep transition and modifies the temperature dependence of the threshold field.
-2.
We thank P. Batail, K. Bechgaard, and their colleagues
for the sample preparation. This work has been partially
supported
for
by a European Strategic Programme
Research in Information Technology Contract No. 3121.
The Laboratoire de Physique des Solides is "Laboratoire
Associee au Centre National de la Recherche Scientifique
No. 0002.
"
Hauser, P. A. Lee, and
T. Poehler, Phys. Rev. Lett. 45, 829
(1980).
5P. M. Chaikin, G. Gruner, E. M. Engler, and R. L. Greene,
Phys. Rev. Lett. 45, 1874 (1980).
A. Zettl, G. Gruner, and E. M. Engler, Phys. Rev. B 25, 1443
(1982).
7A. Janossy, M. Hardiman, and G. Gruner, Solid State Commun. 46, 21 (1983).
H. H. S. Javadi, S. Sridhar, G. Gruner, L. Chiang, and F.
Wudl, Phys. Rev. Lett. 55, 1216 (1985).
PHASE TRANSITION AND NON-OHMIC ELECTRICAL.
L. I. Buravov, V. N. Laukhin, and A. G. Khomenko, Zh. Eksp.
Teor. Fiz. $$, 2185 (1985) [Sov. Phys. JETP 61, 1292
(1985)].
'oG. Griiner, Synth. Met. 29, F453 (1989).
"S.Tomic, J. R. Cooper, D. Jerome, and K. Bechgaard, Phys.
Rev. Lett. 62, 462 (1989).
' P. Monceau, in Lo~-Dimensional Conductors and Superconductors, edited by D. Jerome and L. G. Caron, NATO Advanced Study Institute, Series B, Vol. 155 (Plenum, New
York, 1987), p. 369.
' Recently for (TMTSF)2NO3 a sharp threshold was observed
in the differential resistance at 1.8 K using continuous electric
fields [S. Tomic et al. , in Proceedings of the First Institute for
Solid State Physics International Symposium on the Physics
attd Chemistry of Organic Superconductors, 1989, edited by
G. Saito (Springer-Verlag, Berlin, in press)].
'4T. Takahashi, in Lo~-Dimensional Conductors and Superconductors, Ref. 11, p. 195.
' J. M. Delrieu, M. Roger, Z. Toffano, E. Wope Mbougue, and
R. Saint James, Synth. Met. 19, 283 (1987).
' V. B. Ginodman, A. V. Gudenko. P. A. Kononovich, V. N.
Laukhin, and I. F. Shchegolev, Zh. Eksp. Teor. Fiz. 94, 333
(1988) [Sov. Phys. JETP 67, 1055 (1988)].
' T. Takahashi, H. Kawamura, T. Ohyama, Y. Maniwa, K. Mu-
..
4865
rata, and G. Saito, J. Phys. Soc. Jpn. 5$, 703 (1989).
' G. Creuzet, C. Gaonach, and B. Hamzic, Synth. Met. 19, 245
(1987).
' P. M. Chaikin, T. Tiedje, and A. N. Bloch, Solid State Commun. 41, 739 (1982).
R. H. Friend, M. Miljak, and D. Jerome,
1048 (1978); R. H. Friend, Ph. D. thesis,
bridge, 1979, p. 198.
'A. Arott, S. A. %'erner, and H. Kendrick,
1022 (1965).
P. R. Gamier and M. B. Salamon, Phys.
Phys. Rev. Lett. 46,
University of CamPhys. Rev. Lett. 14,
Rev. Lett. 27, 1523
(1971).
T. Takahashi, Y. Maniwa, H. Kawamura, K. Murata, and G.
Saito, Synth. Met. 19, 225 (1987).
24T. Takahashi, T. Oyama, T. Harada, K. Kanoda, K. Murata,
and G. Saito, in Proceedings of the First Institute for Solid
State Physics International
Symposium on the Physics and
Chemistry of Organic Superconductors,
1989, edited by G.
Saito (Springer-Verlag, Berlin, in press).
zsK. Maki and A. Virosztek, Phys. Rev. B 39, 9640 (1989).
K. Maki and A. Virosztek, in Proceedings of Dubrovnik 1989
[Fizika (to be published)].
R. C. Lacoe, J. R. Cooper, D. Jerome, and F. Creuzet, Phys.
Rev. Lett. 5$, 262 (1987).