- Email: [email protected]
Metallic temperature dependence of resistivity in perchlorate doped polyacetylene
SYflTH|TIIr I-lfl|TRLS ELSEVIER
Synthetic Metals 96 (1998) 81-86
Metallic temperature dependence of resistivity in perchlorate doped polyacetylene Y.W. Park *, E.S. Choi, D.S. Suh Department of Physics and Condensed Matter Research Institute, Seoul National University, Seou1151- 742, South Korea
Received 22 April 1998; received in revised form 4 May 1998; accepted4 May 1998
Abstract We have measured the electrical resistivity (p) and thermoelectric power (TEP) of the perchlorate (C104) doped stretch oriented polyacetylene (PA) film. For the highly conducting samples (~rRT= 41 000 S/cm), the temperature dependence of the fbur-probe resistivity shows positive temperature coefficient of resistivity (TCR) from T= 1.5 to 300 K. For the less conducting samples, the four-probe resistivity data show the crossover of TCR with a broad minimum peak at T = T* = 200 K. For samples of oRx > 20 000 S/cm, p( 1.5 K)/p(300 K) < 1, i.e., the resistivity at 1.5 K is lower than the room temperature resistivity value. The temperature dependence of the TEP shows diffusive linear metallic TEP becoming temperature independent below 40 K. Unlike the others who used C u ( C I O 4 ) 2 for the C104 doping, the initial doping material we used is anhydrous Fe(C104) 3 which is crucial to obtain positive TCR from T= 1.5 to 300 K. © 1998 Elsevier Science S.A. All rights reserved. Keywords: Conductivity;Temperature dependence; Polyacetylene;Perchlorate; Doping
1. Introduction From the first discovery of insulator-metal transition by doping [ 1], metallic polyacetylene (PA) has been widely studied theoretically and experimentally for many years. Metallic properties of heavily doped PA, such as temperature independent magnetic susceptibility, high electrical conductivity of 105 S / c m at room temperature, and linear temperature dependent thermoelectric power (TEP) [ 2 - 4 ] , were observed. From the quasi one-dimensional metallic nature of PA, the room temperature conductivity (~RT) was estimated to be greater than that of copper [5]. But even for the most highly conducting samples, the temperature dependence of resistivity is normally non-metallic, i.e., the temperature coefficient of resistivity (TCR) is negative. There have been some reports about the crossover of TCR from positive to negative near 200 K for AsFs, FeC13, 12 and C 1 0 4 doped PA [6-11 ] and for the other conducting polymers at low temperature [12-14]. It is thought that the non-metallic temperature dependence is due to the substantial disorder which causes localization of the electron wave functions. From the synthesis and the sample preparation approaches, much effort has been made to * Corresponding author. Tel.: +82 2 880 6607; fax: +82 2 873 7037; e-mail: [email protected]
reduce the disorder in the sample to investigate the intrinsic metallic state of conducting polymers. We have doped the stretch oriented PA film with C104and measured the electrical conductivity and TEP. For samples of ~rRV> 20 000 S/cm, we observed p( 1.5 K ) / p ( 3 0 0 K) < 1 and the crossover of TCR below 200 K with a much broader minimum peak than that of the previous reports [ 6 11 ]. In addition, for the first time, we discovered the positive TCR from T = 1.5 to 300 K for the most highly conducting C10 4 doped PA (o-Rv = 41 000 S / c m ) .
2. Experimental The Naarmann type PA film was synthesized by the modified Shirakawa method with the catalyst condition of [ A1 ] / [Ti] = 2.0 [ 15]. Perchlorate doping was done by immersing the stretched film (l/lo = 4) in 0.1 M anhydrous Fe(C104) 3/ acetonitrile solution. Anhydrous Fe (CIO4)3 was prepared by drying Fe(C104)3-H20 powder under dynamic vacuum while heating the powder in the doping apparatus with a heat gun. The doping concentration was controlled by adjusting the immersing time and the concentration of solution. After the film was doped to saturation level, the doping concentration was determined by measuring the weight uptake of the sample. The ionic state of the dopant in the PA film analyzed
0379-6779/98/$ - see front matter © 1998Elsevier Science S.A. All rights reserved. P1lS0379-6779 (98)00078-2
82
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
by energy dispersive spectroscopy (EDS) as well as the ICP emission is identified as in the form of C104- with Fe content less than 0.01%. The extended X-ray absorption fine structure (EXAFS) results, measured by a synchrotron light source, show a small hump at the Fe K-edge energy indicating a small amount of (less than 0.01%) Fe trace in the doped film. The doping concentration is about 7-8% for the samples doped to saturation level. Anhydrous C u ( C 1 0 4 ) 2 w a s also used as an initial doping material. The doping process was similar to the F e ( C I O 4 ) 3 case and analysis shows that the doped PA contains less than 0.02% Cu. FeC13 doping was done for comparison and the doping process is the same as reported earlier [4]. Hereafter, we use CIO4-(Fe) for the material doped with Fe ( C104 ) 3 and C104 (Cu) for the material doped with C u ( C 1 0 4 ) 2. The resistivity was measured by the d.c. four-probe and the Montgomery methods. The typical size of the sample is 8 × 0.5 × 0.01 mm 3 for the four-probe and 5 × 2 × 0.01 mm 3 for the Montgomery methods. Some of the samples were cut into two pieces, and the resistivity was measured by the fourprobe and the Montgomery methods at the same time right after the C 1 0 4 (Fe) doping. The TEP measurement was done by the differential technique [4] with small thermal gradient (typically AT=0.5 K at T = 3 0 0 K and AT=0.1 K at T = 5 K). Since the C I O u (Fe) doped sample was very unstable even under argon atmosphere and at low temperature, the measurements were done immediately after the doping for each sample.
3. Results and discussion Fig. 1 shows the normalized four-probe and Montgomery resistivity results of C 1 0 4 ( F e ) , C 1 0 4 ( C u ) and FeCI4 doped PA. The resistivity data of C I O 4 (Cu) doped polyacetylene measured by Miyamae et al. [ 10,11 ] (data labeled as Miyamae a and b) are also plotted for comparison. The four-probe resistivity o f C 1 0 4 (Fe) doped PA at 300 K varies from 7.39 X 10 -4 1) cm ( 1350 S/cm) to 2.41 × 10 -5 l-I cm (41 000 S/cm). Most of the samples with room temperature conductivity less than 20 000 S/cm show a broad minimum peak in resistivity at the specific temperature T = T*. The peak temperature T* tends to shift to the lower temperature as the room temperature conductivity of the samples increases. Samples with room temperature conductivity higher than 20 000 S / c m show p( 1.5 K ) / p ( 3 0 0 K) < 1, i.e., the resistivity at T = 1.5 K is lower than the room temperature resistivity value. The positive TCR for T > T* was reported for the heavily doped polyacetylene. Miyamae et al. [ 10,11] had measured the four-probe conductivity of C104-(Cu) doped polyacetylene and observed such behavior with T* = 190 or 214 K. The room temperature conductivity, T*, resistivity ratio at T* (p( T*) / p( 300 K) ) for C I O 4 (Fe), C104 (Cu) and FeC14- doped PA are listed in Table 1.
0 (RT)
Sample
2.5
Miyamae (c) (2.2%) 1 67x104 ~)cm
t
C104 (Cu) (3.3%), 1 8 8 x 1 0 4 £_~cm Miyamae (b) (8.2% aged). 7.69x10 ' ~2cm
i\\
2.0
FeCI, 7 5 % . 1 11x10" ~ c m -- Miyamae (a) (7.4%) 5.88x10 ~ L2cm
\\
CI04(Fe), 4-probe, 8.84x10 ~ {)cm
\,~ \
CIO, (Fe), M o n l g o 6.58x10
CIO 4 (Fe), Montgo 8.55x10 ~ f)cm
\
1.5
C[O 4 (Fe). 4-probe. 5.01x10 ~ ~._~cm
v 0 o co v
-Qcm
~
,,
..... i?..
~
Cl04 (Fe), 4_pr0be 2.41x10 ,, ~ c m
..... \ - .
1.0
0.5
0.0
I 0
I
I
I
I
I
100
150
200
250
300
T(K) Fig. 1. Normalizedresistivityof C I O 4 and FeCI4 doped polyacetylnenes vs. temperaturemeasuredby the ~ur-probe and Montgomerymethods. Kaiser et al. [16,17] and Park et al. [18] suggested the heterogeneous model to explain the crossover of TCR at T = T* for the highly conducting polymers and the carbon nanotube network. According to the model, the temperature dependence of conductivity arises from a combination of metallic conduction and Sheng' s fluctuation-induced tunneling model [19]. The metallic conduction term can be expressed as either the three-dimensional metallic or the highly anisotropic ( quasi one-dimensional) metallic conduction formula. For the three-dimensional metallic conduction, the total resistivity, including the metallic and the fluctuation-induced tunneling terms, can be written as follows:
R=A3_dT+B exp[TJ(T+ T~)]
( 1)
where A3_a, B, Z c and T, are constant values. Tc denotes the temperature below which the conduction is dominated by the charge carrier tunneling through the barrier and T~ is the temperature above which the thermally activated conduction over the barrier begins to occur. For quasi one-dimensional metallic conduction, following Kivelson and Heeger's formula [5], the total resistivity can be expressed as
R=Al_d exp(-hwo/kB T)+ B exp[ Tc/( T + T~)]
(2)
where At_d and Wo are constants with h~oo=0.12 eV (1400 K) for undoped polyacetylene. The quasi one-dimensional
83
Y. W. Park et al. /S_vnthetic Metals 96 (199X) 81-M Table I Resistivity data of CIO;. (Fe) doped polyacetylene. Dopant
7.41 x lomJ 1.30x 1.06~ 9.33x 8.84~ 8.31 x 5.80X 5.01 x 4.79x 2.41 x 1.68~
(Cu)
(Cu) and FeClrm doped polyacetylene p( 1.5 K)/p(300
K)
data are listed T*
p(T*)lp(300
K)
(K)
1350
1om4 10m4 lo-.IO-’ lo-’ lo-? loms 1o-s IO-’ lo-“”
1.11X10-4b
FeCI,
the CIO,
cr (300 K) (S/cm)
P (300K) (0 cm)
ClO,- (Fe)
ClO,
For comparison,
I .46
227
0.970
1.07
194 240 220 194 196 165 159
0.894 0.897 0.939 0.894 0.930 0.900 0.916
254
0.977
200
0.905
7670 9390 10720 11300 12000 17200 19960 20900 41000 5950
1.15 0.98 0.66 0.57 I .55
8990
1.38
1.17 I .22
1.07 1.1 I
CIO,
(Cu)
Miyamae
(a) ’
17000
1.19
190
0.917
ClO,
(Cu)
Miyamae
(b) ’
13000
1.38
214
0.935
d 3.3%. h 7.5%. ’ Miyamae (a) (v = 7.4%) is the fresh sample. ’ Miyamde (b) (v = 8.2%) is the aged sample exposed to air for 30 min.
(a) CI04~(Cu) doped
16
15 14 13 12 11 I0 09 0
14
(b) FeCI, 7 5%. 1 11x10 .’ llcm
13
3.3%, 1.68~10.~ clcm
12 % (Miyamae (a)) 5.88~10~~ Rem’ ’ ,O 09 250
CIO, (Fe).8.55x10~5 Qcm (Montgomery)
105
CIO;(Fe). 6.58~10~~ Rem (Montgomery)
100 (f) CIO;(Fe), 5.01~10~ Rem with h~,=0.072 eV D98
095
096
090
094
185
092
0
150
200
250
300
140
200
250
T(K) Fig. 2. Normalized
resistivity of CIO,
(Cu) (a)-(b)
and ClO,- (Fe) (c)-(f)
metallic term represents the elastic scattering of electrons by 2kF phonons of energy hw,,. The large value of &J” causes the increase of resistivity near room temperature. While we used both Eqs. ( 1) and (2) to fit the crossover data of TCR at T= T*, Eq. (2) yielded much better results
doped polyacetylenes
vs. temperature.
for fitting the overall temperature dependence of resistivity as well as the crossover of TCR. This indicates that the quasi one-dimensional charge transport is dominant in heavily doped polyacetylene. Fig. 2 shows the fitting results using Eq. (2). The value of r%wOwas fixed as 0.12 eV for each
84
Y. W. Park et al. /Synthetic Metals 96 (1998) 81-86
sample except one, for which the best estimated value of hwo was found to be 0.072 eV. The fitting is rather good but a small discrepancy appears below 50 K for C I O 4 (Fe) doped samples. It is remarkable that the overall temperature dependence of the most highly conducting samples cannot be explained by either the heterogeneous model or the metallic conduction alone. Fig. 3 shows the resistivity data and the dotted curve represents the quasi one-dimensional metallic conduction formula (the first part of Eq. (2)) with the residual resistivity of/9o = 1.54 X 10 -5 1) cm. The inset shows the linear fitting of the data below 40 K. For this sample, other contributions such as the fluctuation induced tunneling or the disordered metallic conduction could not explain the resistivity data at both low and high temperatures. However, the resistivity data obtained by different methods (four-probe and Montgomery methods) for the sample doped in the same batch show different behaviors. While the four-probe resistivity is metallic in the whole temperature range (Fig. 3), the Montgomery resistivity shows a broad minimum peak as shown in Fig. 2(e) although the room temperature conductivity is lower than that of the four-probe result. The anisotropy (trll / trl ) of resistivity measured by the Montgomery method for the sample shown in Fig. 2 is temperature independent from T = 300 to 1.5 K with the magnitude of approximately 50. For the highly anisotropic samples, the four-probe resistivity often shows anomalous temperature dependences. In fact we have measured quite a few four-probe negative resistances at low temperature in these samples. But the data change upon thermal recycling. This is a rather well-known artifact arising from the back flow resistance in the highly anisotropic samples. We have ruled out such data in our data presentation. The four-probe data shown in Fig. 3 are reproducible upon thermal recycling. After four thermal recyclings from T = 1.5 up to 200 K, the genuine feature of the temperature dependence remains the same, although the absolute values become bigger by 12%. Therefore, we conclude that the observed differences of the four-probe and the Montgomery resistivities do not originate from an artifact of the four-probe resistivity measurements for the highly anisotropic samples. It is rather due to the room temperature conductivity differences. The resistivity of the sample doped in the same batch can vary from part to part of the sample. That is, the doping is homogeneous on the microscopic scale but it can vary on the macroscopic scale of the sample. The room temperature resistivities of the C 1 0 4 (Fe) doped samples and their corresponding temperature dependences ( see Figs. 2 ( c ) - ( f ) and 3 ) show systematic correlations regardless of whether it is the four-probe results (Figs. 2(c), (f) and 3) or Montgomery results (Fig. 2(d), (e)). As the room temperature resistivity decreases, the broad minimum peak temperature T* tends to shift down to lower temperature and the p(l.5 K ) / 0 ( 3 0 0 K) becomes smaller, as shown in Fig. 4. For the most highly conducting sample (Fig. 3), there is no broad minimum of resistivity. Instead, the resistivity con-
2.6x10
22x10
s
20xlO ~
]
1 8x10 ~
0
5
10 15 20 25 30 35 40
14×10 ~ 510
~ 1 O0
i 150
i 250
200
I 300
T(K)
Fig. 3. Temperature dependence of the resistivity of the highly conducting C104- ( Fe ) doped polyacetylene ( trRT= 41 000 S /cm ). The inset show s the data below T= 40 K.
.
13
I 240
12 ~,
220
O0 0
1 1
2OO o
10
re 180 -x: ,n
09 160 0.8140 07120
06-
05 20xl
i
0 ~
40x10
i
~
60xlO p (30OK)
100
i
"5
8,0x105
10xlO
"
(_Qcm)
Fig. 4. Plots of p(l.5 K)/p(300 K) and T* vs. p(300 K) of C104 (Fe) doped polyacetylenes.
tinues to decrease from T = 300 to 1.5 K. The initial rapid decrease of resistivity upon cooling tends to be slowed down, becoming more or less constant in the region at 40 < T < 100 K. For T < 4 0 K, the resistivity changes linearly as a function of temperature, as shown in the inset of Fig. 3. The data shown in Fig. 2(f) also show a resistivity slope change at T < 30 K with p( 1.5 K ) / p ( 3 0 0 K) = 0.975 which is less than unity. That is, the resistivity at T = 1.5 K is lower than the room temperature resistivity value, although there exists a broad minimum peak at T* = 150 K. More recent data for another C104 (Fe) doped PA whose room temperature resistivity is slightly lower (pRT=4.9× 10 -5 1) cm) than that of the sample plotted in Fig. 2 (f) show very similar temperature dependence with the data in Fig. 2(f). These observations confirm that the four-probe resistivity data shown in Figs. 2(f) and 3 are the intrinsic temperature dependences of the C 1 0 4 (Fe) doped polyacetylene: ( 1 ) If the p(300 K) < 5 . 0 0 × 10 -3 1"1cm, the p(l.5 K ) / p ( 3 0 0 K) becomes less than unity. (2) The broad minimum tempera-
85
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
ture T* becomes significantly lower for the sample of p(300 K ) = 5 . 0 0 × 10 -5 lq cm and it disappears for the p(300 K) = 2.6 × 10-5 f~ cm sample. (3) Another resistivity slope change exists at T = 4 0 K showing the linear temperature dependence of resistivity at lower temperature. Fig. 5 shows the temperature dependence of the TEP for C I O 4 - ( C u ) , C I O 4 (Fe) and FeC14 doped polyacetylene. The overall temperature dependence of the TEP is quasi linear with anomalous behavior below 40 K. The linear temperature dependent term can be attributed to the diffusive metallic TEP. From Motrs formula [20]: '11"2
kB
2st"~ 2e ~ ~
25-
~
~ ' 1 O1 ~ ~
,
05
20
~
~ .05]
7sU'~.... . "~
°
FeCl442% ClO4 (Fe) 7.0%
~
Cl04(Cu )33%
•
CIO 4 (Fe) 8 0%
~e
..%~_
,,., 8
=
15 o
20
4o
T(K) >
60
8o
lO
~,~
~.-~
10
O3
where ~(EF) is the density of the states at the Fermi level. The value of 7?(EF) is about 0.14 states/(eV C) for C i O 4 ( F e ) doped polyacetylene, consistent with the typical value obtained by magnetic susceptibility measurements of heavily doped polyacetylene (about 0.2). When extrapolated to T = 0 K, it is found that there is a negative contribution of TEP with magnitude - 2 IxV/K for the FeCI 4 doped sample and - 0.4 to - 0.8 txV/K for the C I O 4 (Fe or Cu) doped one. The inset of Fig. 5 shows the contribution of TEP (AS) at low temperature which deviates from the metallic diffusion TEP. It is obtained by
AS=S-AT
(4)
AS increases as temperature decreases except for one sample which shows a positive hump near 50 K. The increase of TEP at low temperature can be due to the electron-phonon interaction or the impurity scattering. It reminds us that the TEP of normal metals containing impurities with localized spin increases at low temperature [21]. The effect of high magnetic field on TEP of the metal-halide doped PA has indicated such spin-spin couplings between conduction electrons in the polymer chain and the localized spins in the dopant [22]. Further investigations on the high magnetic field effect of the resistivity and the TEP for the different doping concentrations of CIO 4 (Fe) are continuing in detail at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee, FL.
4. Summary The metallic conduction of C l O 4 (Fe) doped polyacetylene was studied by electrical resistivity and TEP measurements. The crossover of TCR was fitted using the heterogeneous model with the quasi one-dimensional metallic conduction formula. For the most highly conducting sampie, TCR is positive in the whole temperature range ( T = 1.5 to 300 K). For this sample, the linear temperature dependence of resistivity appears below 40 K and the rapid increase of resistivity near room temperature can be attributed to the quasi one-dimensional metallic conduction. The temperature
-5
0
5~0
100 ~
150 ~ T (K)
200 ~
250 ~
300 p
Fig. 5. TEP of ClO4 and F e C L doped polyacetylenes vs. temperature. The inset shows AS below T= 100 K.
dependence of TEP indicates the diffusive metallic TEP with extra scattering mechanisms at low temperature.
Acknowledgements The ICP emission and EDS analysis for doped polyacetylne were done at the Inter-University Center for Natural Science Research Facilities, Seoul National University. This work was supported by the Korea Science and Engineering Foundation (KOSEF) and the Ministry of Education (MOE), Korea.
References [ 1] C.K. Chiang, C.R. Fincher Jr., Y.W. Park, A.J. Heeger, H. Shirakawa, EJ. Louis, S.C. Gau, A.G. MacDiarmid, Phys. Rev. Lett. 39 (1977) 1098. [2] A.J. Epstein, H. Rommelmen, M.A. Druy, A.J. Heeger, A.G. MacDiarmid, Solid State Commun. 38 ( 1981 ) 683. [3] Th. Schimmel, W. Rie6, J. Gmeiner, G. Denninger, M. Schwoerer, H. Naarmann, N. Theophilou, Solid State Commun. 65 (1988) 1311. [4] Y.W. Park, Synth. Met. 45 ( 1991 ) 173. [5] S. Kivelson, A.J. Heeger, Synth. Met. 22 (1988) 371. [ 6] Y.W. Park, C. Park, Y.S. Lee, C.O. Yoon, H. Shirakawa, Y. Suezaki, K. Akagi, Solid State Commun. 65 (1988) 147. [7] J.F. Kwak, T.C. Clarke, R.L. Greene, G.B. Street, Solid State Commun. 31 (1979) 335. [ 8 ] Y.W. Park, J.C. Woo, K.H. Yoo, W.K. Han, C.H. Choi, T. Kobayashi, H. Shirakawa, Solid State Commun. 46 (1983) 731. [9] W. Pukacki, R. Zuzok, S. Roth, W. G6pel, in H. Kuzmany, M. Mehring, S. Roth (eds.), Electronic Properties of P o l y m e r s - Orientation
86
[ 10] [ 11 ] [ 12 ] [ 13 ] [ 14] [ 15]
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
and Dimensionality of Conjugated Systems ( Kirchberg IV), Springer, Berlin, 1992, p. 106. T. Miyamae, T. Mori, K. Seki, J. Tanaka. Synth. Met. 69-71 (1995) 59. T. Miyamae, T. Mori, K. Seki, J. Tanaka, Bull. Chem. Soc. Jpn. 68 (1995) 803. C.O. Yoon, M. Reghu, A.J. Heeger, Phys. Rev. B 49 (1994) 10 85 I. M. Ahlskog, M. Reghu, A.J. Heeger, Phys. Rev. B 53 (1996) 15 529. W.P. Lee, Y.W. Park, Y.S. Choi, Synth. Met. 84-86 (1997) 841. K. Akagi, M. Suezaki, H. Shirakawa, H. Kyotani, M. Shimomura, Y. Tanabe, Synth. Met. 27-29 (3) (1989) DI.
[ 16] A.B. Kaiser, Synth. Met. 45 ( 1991 ) 183. [ 17] A.B. Kaiser, G. Diisberg, S. Roth, Phys. Rev. B 57 (1998) 1418. [ 18l Y.W. Park, A.J. Heeger, M.A. Druy, A.G. MacDiarmid, J. Chem. Phys. 73 (1980) 946. [ 19] P. Sheng, Phys. Rev. B 21 (1980) 2180. [20] N.F. Mott, E.A. Davis, Electronic Process in Non-Crystalline Materials, Clarendon Press, Oxford, 1979, p. 52. 121 ] D.K.C. MacDonald, Thermoelectricity: An Introduction to the Principles, Wiley, New York, 1962, p. 24. [22] E.S. Choi, Y.H. Seol, Y.S. Song, Y.W. Park, S.T. Hannahs, Synth. Met. 84-86 (1997) 685.
Synthetic Metals 96 (1998) 81-86
Metallic temperature dependence of resistivity in perchlorate doped polyacetylene Y.W. Park *, E.S. Choi, D.S. Suh Department of Physics and Condensed Matter Research Institute, Seoul National University, Seou1151- 742, South Korea
Received 22 April 1998; received in revised form 4 May 1998; accepted4 May 1998
Abstract We have measured the electrical resistivity (p) and thermoelectric power (TEP) of the perchlorate (C104) doped stretch oriented polyacetylene (PA) film. For the highly conducting samples (~rRT= 41 000 S/cm), the temperature dependence of the fbur-probe resistivity shows positive temperature coefficient of resistivity (TCR) from T= 1.5 to 300 K. For the less conducting samples, the four-probe resistivity data show the crossover of TCR with a broad minimum peak at T = T* = 200 K. For samples of oRx > 20 000 S/cm, p( 1.5 K)/p(300 K) < 1, i.e., the resistivity at 1.5 K is lower than the room temperature resistivity value. The temperature dependence of the TEP shows diffusive linear metallic TEP becoming temperature independent below 40 K. Unlike the others who used C u ( C I O 4 ) 2 for the C104 doping, the initial doping material we used is anhydrous Fe(C104) 3 which is crucial to obtain positive TCR from T= 1.5 to 300 K. © 1998 Elsevier Science S.A. All rights reserved. Keywords: Conductivity;Temperature dependence; Polyacetylene;Perchlorate; Doping
1. Introduction From the first discovery of insulator-metal transition by doping [ 1], metallic polyacetylene (PA) has been widely studied theoretically and experimentally for many years. Metallic properties of heavily doped PA, such as temperature independent magnetic susceptibility, high electrical conductivity of 105 S / c m at room temperature, and linear temperature dependent thermoelectric power (TEP) [ 2 - 4 ] , were observed. From the quasi one-dimensional metallic nature of PA, the room temperature conductivity (~RT) was estimated to be greater than that of copper [5]. But even for the most highly conducting samples, the temperature dependence of resistivity is normally non-metallic, i.e., the temperature coefficient of resistivity (TCR) is negative. There have been some reports about the crossover of TCR from positive to negative near 200 K for AsFs, FeC13, 12 and C 1 0 4 doped PA [6-11 ] and for the other conducting polymers at low temperature [12-14]. It is thought that the non-metallic temperature dependence is due to the substantial disorder which causes localization of the electron wave functions. From the synthesis and the sample preparation approaches, much effort has been made to * Corresponding author. Tel.: +82 2 880 6607; fax: +82 2 873 7037; e-mail: [email protected]
reduce the disorder in the sample to investigate the intrinsic metallic state of conducting polymers. We have doped the stretch oriented PA film with C104and measured the electrical conductivity and TEP. For samples of ~rRV> 20 000 S/cm, we observed p( 1.5 K ) / p ( 3 0 0 K) < 1 and the crossover of TCR below 200 K with a much broader minimum peak than that of the previous reports [ 6 11 ]. In addition, for the first time, we discovered the positive TCR from T = 1.5 to 300 K for the most highly conducting C10 4 doped PA (o-Rv = 41 000 S / c m ) .
2. Experimental The Naarmann type PA film was synthesized by the modified Shirakawa method with the catalyst condition of [ A1 ] / [Ti] = 2.0 [ 15]. Perchlorate doping was done by immersing the stretched film (l/lo = 4) in 0.1 M anhydrous Fe(C104) 3/ acetonitrile solution. Anhydrous Fe (CIO4)3 was prepared by drying Fe(C104)3-H20 powder under dynamic vacuum while heating the powder in the doping apparatus with a heat gun. The doping concentration was controlled by adjusting the immersing time and the concentration of solution. After the film was doped to saturation level, the doping concentration was determined by measuring the weight uptake of the sample. The ionic state of the dopant in the PA film analyzed
0379-6779/98/$ - see front matter © 1998Elsevier Science S.A. All rights reserved. P1lS0379-6779 (98)00078-2
82
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
by energy dispersive spectroscopy (EDS) as well as the ICP emission is identified as in the form of C104- with Fe content less than 0.01%. The extended X-ray absorption fine structure (EXAFS) results, measured by a synchrotron light source, show a small hump at the Fe K-edge energy indicating a small amount of (less than 0.01%) Fe trace in the doped film. The doping concentration is about 7-8% for the samples doped to saturation level. Anhydrous C u ( C 1 0 4 ) 2 w a s also used as an initial doping material. The doping process was similar to the F e ( C I O 4 ) 3 case and analysis shows that the doped PA contains less than 0.02% Cu. FeC13 doping was done for comparison and the doping process is the same as reported earlier [4]. Hereafter, we use CIO4-(Fe) for the material doped with Fe ( C104 ) 3 and C104 (Cu) for the material doped with C u ( C 1 0 4 ) 2. The resistivity was measured by the d.c. four-probe and the Montgomery methods. The typical size of the sample is 8 × 0.5 × 0.01 mm 3 for the four-probe and 5 × 2 × 0.01 mm 3 for the Montgomery methods. Some of the samples were cut into two pieces, and the resistivity was measured by the fourprobe and the Montgomery methods at the same time right after the C 1 0 4 (Fe) doping. The TEP measurement was done by the differential technique [4] with small thermal gradient (typically AT=0.5 K at T = 3 0 0 K and AT=0.1 K at T = 5 K). Since the C I O u (Fe) doped sample was very unstable even under argon atmosphere and at low temperature, the measurements were done immediately after the doping for each sample.
3. Results and discussion Fig. 1 shows the normalized four-probe and Montgomery resistivity results of C 1 0 4 ( F e ) , C 1 0 4 ( C u ) and FeCI4 doped PA. The resistivity data of C I O 4 (Cu) doped polyacetylene measured by Miyamae et al. [ 10,11 ] (data labeled as Miyamae a and b) are also plotted for comparison. The four-probe resistivity o f C 1 0 4 (Fe) doped PA at 300 K varies from 7.39 X 10 -4 1) cm ( 1350 S/cm) to 2.41 × 10 -5 l-I cm (41 000 S/cm). Most of the samples with room temperature conductivity less than 20 000 S/cm show a broad minimum peak in resistivity at the specific temperature T = T*. The peak temperature T* tends to shift to the lower temperature as the room temperature conductivity of the samples increases. Samples with room temperature conductivity higher than 20 000 S / c m show p( 1.5 K ) / p ( 3 0 0 K) < 1, i.e., the resistivity at T = 1.5 K is lower than the room temperature resistivity value. The positive TCR for T > T* was reported for the heavily doped polyacetylene. Miyamae et al. [ 10,11] had measured the four-probe conductivity of C104-(Cu) doped polyacetylene and observed such behavior with T* = 190 or 214 K. The room temperature conductivity, T*, resistivity ratio at T* (p( T*) / p( 300 K) ) for C I O 4 (Fe), C104 (Cu) and FeC14- doped PA are listed in Table 1.
0 (RT)
Sample
2.5
Miyamae (c) (2.2%) 1 67x104 ~)cm
t
C104 (Cu) (3.3%), 1 8 8 x 1 0 4 £_~cm Miyamae (b) (8.2% aged). 7.69x10 ' ~2cm
i\\
2.0
FeCI, 7 5 % . 1 11x10" ~ c m -- Miyamae (a) (7.4%) 5.88x10 ~ L2cm
\\
CI04(Fe), 4-probe, 8.84x10 ~ {)cm
\,~ \
CIO, (Fe), M o n l g o 6.58x10
CIO 4 (Fe), Montgo 8.55x10 ~ f)cm
\
1.5
C[O 4 (Fe). 4-probe. 5.01x10 ~ ~._~cm
v 0 o co v
-Qcm
~
,,
..... i?..
~
Cl04 (Fe), 4_pr0be 2.41x10 ,, ~ c m
..... \ - .
1.0
0.5
0.0
I 0
I
I
I
I
I
100
150
200
250
300
T(K) Fig. 1. Normalizedresistivityof C I O 4 and FeCI4 doped polyacetylnenes vs. temperaturemeasuredby the ~ur-probe and Montgomerymethods. Kaiser et al. [16,17] and Park et al. [18] suggested the heterogeneous model to explain the crossover of TCR at T = T* for the highly conducting polymers and the carbon nanotube network. According to the model, the temperature dependence of conductivity arises from a combination of metallic conduction and Sheng' s fluctuation-induced tunneling model [19]. The metallic conduction term can be expressed as either the three-dimensional metallic or the highly anisotropic ( quasi one-dimensional) metallic conduction formula. For the three-dimensional metallic conduction, the total resistivity, including the metallic and the fluctuation-induced tunneling terms, can be written as follows:
R=A3_dT+B exp[TJ(T+ T~)]
( 1)
where A3_a, B, Z c and T, are constant values. Tc denotes the temperature below which the conduction is dominated by the charge carrier tunneling through the barrier and T~ is the temperature above which the thermally activated conduction over the barrier begins to occur. For quasi one-dimensional metallic conduction, following Kivelson and Heeger's formula [5], the total resistivity can be expressed as
R=Al_d exp(-hwo/kB T)+ B exp[ Tc/( T + T~)]
(2)
where At_d and Wo are constants with h~oo=0.12 eV (1400 K) for undoped polyacetylene. The quasi one-dimensional
83
Y. W. Park et al. /S_vnthetic Metals 96 (199X) 81-M Table I Resistivity data of CIO;. (Fe) doped polyacetylene. Dopant
7.41 x lomJ 1.30x 1.06~ 9.33x 8.84~ 8.31 x 5.80X 5.01 x 4.79x 2.41 x 1.68~
(Cu)
(Cu) and FeClrm doped polyacetylene p( 1.5 K)/p(300
K)
data are listed T*
p(T*)lp(300
K)
(K)
1350
1om4 10m4 lo-.IO-’ lo-’ lo-? loms 1o-s IO-’ lo-“”
1.11X10-4b
FeCI,
the CIO,
cr (300 K) (S/cm)
P (300K) (0 cm)
ClO,- (Fe)
ClO,
For comparison,
I .46
227
0.970
1.07
194 240 220 194 196 165 159
0.894 0.897 0.939 0.894 0.930 0.900 0.916
254
0.977
200
0.905
7670 9390 10720 11300 12000 17200 19960 20900 41000 5950
1.15 0.98 0.66 0.57 I .55
8990
1.38
1.17 I .22
1.07 1.1 I
CIO,
(Cu)
Miyamae
(a) ’
17000
1.19
190
0.917
ClO,
(Cu)
Miyamae
(b) ’
13000
1.38
214
0.935
d 3.3%. h 7.5%. ’ Miyamae (a) (v = 7.4%) is the fresh sample. ’ Miyamde (b) (v = 8.2%) is the aged sample exposed to air for 30 min.
(a) CI04~(Cu) doped
16
15 14 13 12 11 I0 09 0
14
(b) FeCI, 7 5%. 1 11x10 .’ llcm
13
3.3%, 1.68~10.~ clcm
12 % (Miyamae (a)) 5.88~10~~ Rem’ ’ ,O 09 250
CIO, (Fe).8.55x10~5 Qcm (Montgomery)
105
CIO;(Fe). 6.58~10~~ Rem (Montgomery)
100 (f) CIO;(Fe), 5.01~10~ Rem with h~,=0.072 eV D98
095
096
090
094
185
092
0
150
200
250
300
140
200
250
T(K) Fig. 2. Normalized
resistivity of CIO,
(Cu) (a)-(b)
and ClO,- (Fe) (c)-(f)
metallic term represents the elastic scattering of electrons by 2kF phonons of energy hw,,. The large value of &J” causes the increase of resistivity near room temperature. While we used both Eqs. ( 1) and (2) to fit the crossover data of TCR at T= T*, Eq. (2) yielded much better results
doped polyacetylenes
vs. temperature.
for fitting the overall temperature dependence of resistivity as well as the crossover of TCR. This indicates that the quasi one-dimensional charge transport is dominant in heavily doped polyacetylene. Fig. 2 shows the fitting results using Eq. (2). The value of r%wOwas fixed as 0.12 eV for each
84
Y. W. Park et al. /Synthetic Metals 96 (1998) 81-86
sample except one, for which the best estimated value of hwo was found to be 0.072 eV. The fitting is rather good but a small discrepancy appears below 50 K for C I O 4 (Fe) doped samples. It is remarkable that the overall temperature dependence of the most highly conducting samples cannot be explained by either the heterogeneous model or the metallic conduction alone. Fig. 3 shows the resistivity data and the dotted curve represents the quasi one-dimensional metallic conduction formula (the first part of Eq. (2)) with the residual resistivity of/9o = 1.54 X 10 -5 1) cm. The inset shows the linear fitting of the data below 40 K. For this sample, other contributions such as the fluctuation induced tunneling or the disordered metallic conduction could not explain the resistivity data at both low and high temperatures. However, the resistivity data obtained by different methods (four-probe and Montgomery methods) for the sample doped in the same batch show different behaviors. While the four-probe resistivity is metallic in the whole temperature range (Fig. 3), the Montgomery resistivity shows a broad minimum peak as shown in Fig. 2(e) although the room temperature conductivity is lower than that of the four-probe result. The anisotropy (trll / trl ) of resistivity measured by the Montgomery method for the sample shown in Fig. 2 is temperature independent from T = 300 to 1.5 K with the magnitude of approximately 50. For the highly anisotropic samples, the four-probe resistivity often shows anomalous temperature dependences. In fact we have measured quite a few four-probe negative resistances at low temperature in these samples. But the data change upon thermal recycling. This is a rather well-known artifact arising from the back flow resistance in the highly anisotropic samples. We have ruled out such data in our data presentation. The four-probe data shown in Fig. 3 are reproducible upon thermal recycling. After four thermal recyclings from T = 1.5 up to 200 K, the genuine feature of the temperature dependence remains the same, although the absolute values become bigger by 12%. Therefore, we conclude that the observed differences of the four-probe and the Montgomery resistivities do not originate from an artifact of the four-probe resistivity measurements for the highly anisotropic samples. It is rather due to the room temperature conductivity differences. The resistivity of the sample doped in the same batch can vary from part to part of the sample. That is, the doping is homogeneous on the microscopic scale but it can vary on the macroscopic scale of the sample. The room temperature resistivities of the C 1 0 4 (Fe) doped samples and their corresponding temperature dependences ( see Figs. 2 ( c ) - ( f ) and 3 ) show systematic correlations regardless of whether it is the four-probe results (Figs. 2(c), (f) and 3) or Montgomery results (Fig. 2(d), (e)). As the room temperature resistivity decreases, the broad minimum peak temperature T* tends to shift down to lower temperature and the p(l.5 K ) / 0 ( 3 0 0 K) becomes smaller, as shown in Fig. 4. For the most highly conducting sample (Fig. 3), there is no broad minimum of resistivity. Instead, the resistivity con-
2.6x10
22x10
s
20xlO ~
]
1 8x10 ~
0
5
10 15 20 25 30 35 40
14×10 ~ 510
~ 1 O0
i 150
i 250
200
I 300
T(K)
Fig. 3. Temperature dependence of the resistivity of the highly conducting C104- ( Fe ) doped polyacetylene ( trRT= 41 000 S /cm ). The inset show s the data below T= 40 K.
.
13
I 240
12 ~,
220
O0 0
1 1
2OO o
10
re 180 -x: ,n
09 160 0.8140 07120
06-
05 20xl
i
0 ~
40x10
i
~
60xlO p (30OK)
100
i
"5
8,0x105
10xlO
"
(_Qcm)
Fig. 4. Plots of p(l.5 K)/p(300 K) and T* vs. p(300 K) of C104 (Fe) doped polyacetylenes.
tinues to decrease from T = 300 to 1.5 K. The initial rapid decrease of resistivity upon cooling tends to be slowed down, becoming more or less constant in the region at 40 < T < 100 K. For T < 4 0 K, the resistivity changes linearly as a function of temperature, as shown in the inset of Fig. 3. The data shown in Fig. 2(f) also show a resistivity slope change at T < 30 K with p( 1.5 K ) / p ( 3 0 0 K) = 0.975 which is less than unity. That is, the resistivity at T = 1.5 K is lower than the room temperature resistivity value, although there exists a broad minimum peak at T* = 150 K. More recent data for another C104 (Fe) doped PA whose room temperature resistivity is slightly lower (pRT=4.9× 10 -5 1) cm) than that of the sample plotted in Fig. 2 (f) show very similar temperature dependence with the data in Fig. 2(f). These observations confirm that the four-probe resistivity data shown in Figs. 2(f) and 3 are the intrinsic temperature dependences of the C 1 0 4 (Fe) doped polyacetylene: ( 1 ) If the p(300 K) < 5 . 0 0 × 10 -3 1"1cm, the p(l.5 K ) / p ( 3 0 0 K) becomes less than unity. (2) The broad minimum tempera-
85
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
ture T* becomes significantly lower for the sample of p(300 K ) = 5 . 0 0 × 10 -5 lq cm and it disappears for the p(300 K) = 2.6 × 10-5 f~ cm sample. (3) Another resistivity slope change exists at T = 4 0 K showing the linear temperature dependence of resistivity at lower temperature. Fig. 5 shows the temperature dependence of the TEP for C I O 4 - ( C u ) , C I O 4 (Fe) and FeC14 doped polyacetylene. The overall temperature dependence of the TEP is quasi linear with anomalous behavior below 40 K. The linear temperature dependent term can be attributed to the diffusive metallic TEP. From Motrs formula [20]: '11"2
kB
2st"~ 2e ~ ~
25-
~
~ ' 1 O1 ~ ~
,
05
20
~
~ .05]
7sU'~.... . "~
°
FeCl442% ClO4 (Fe) 7.0%
~
Cl04(Cu )33%
•
CIO 4 (Fe) 8 0%
~e
..%~_
,,., 8
=
15 o
20
4o
T(K) >
60
8o
lO
~,~
~.-~
10
O3
where ~(EF) is the density of the states at the Fermi level. The value of 7?(EF) is about 0.14 states/(eV C) for C i O 4 ( F e ) doped polyacetylene, consistent with the typical value obtained by magnetic susceptibility measurements of heavily doped polyacetylene (about 0.2). When extrapolated to T = 0 K, it is found that there is a negative contribution of TEP with magnitude - 2 IxV/K for the FeCI 4 doped sample and - 0.4 to - 0.8 txV/K for the C I O 4 (Fe or Cu) doped one. The inset of Fig. 5 shows the contribution of TEP (AS) at low temperature which deviates from the metallic diffusion TEP. It is obtained by
AS=S-AT
(4)
AS increases as temperature decreases except for one sample which shows a positive hump near 50 K. The increase of TEP at low temperature can be due to the electron-phonon interaction or the impurity scattering. It reminds us that the TEP of normal metals containing impurities with localized spin increases at low temperature [21]. The effect of high magnetic field on TEP of the metal-halide doped PA has indicated such spin-spin couplings between conduction electrons in the polymer chain and the localized spins in the dopant [22]. Further investigations on the high magnetic field effect of the resistivity and the TEP for the different doping concentrations of CIO 4 (Fe) are continuing in detail at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee, FL.
4. Summary The metallic conduction of C l O 4 (Fe) doped polyacetylene was studied by electrical resistivity and TEP measurements. The crossover of TCR was fitted using the heterogeneous model with the quasi one-dimensional metallic conduction formula. For the most highly conducting sampie, TCR is positive in the whole temperature range ( T = 1.5 to 300 K). For this sample, the linear temperature dependence of resistivity appears below 40 K and the rapid increase of resistivity near room temperature can be attributed to the quasi one-dimensional metallic conduction. The temperature
-5
0
5~0
100 ~
150 ~ T (K)
200 ~
250 ~
300 p
Fig. 5. TEP of ClO4 and F e C L doped polyacetylenes vs. temperature. The inset shows AS below T= 100 K.
dependence of TEP indicates the diffusive metallic TEP with extra scattering mechanisms at low temperature.
Acknowledgements The ICP emission and EDS analysis for doped polyacetylne were done at the Inter-University Center for Natural Science Research Facilities, Seoul National University. This work was supported by the Korea Science and Engineering Foundation (KOSEF) and the Ministry of Education (MOE), Korea.
References [ 1] C.K. Chiang, C.R. Fincher Jr., Y.W. Park, A.J. Heeger, H. Shirakawa, EJ. Louis, S.C. Gau, A.G. MacDiarmid, Phys. Rev. Lett. 39 (1977) 1098. [2] A.J. Epstein, H. Rommelmen, M.A. Druy, A.J. Heeger, A.G. MacDiarmid, Solid State Commun. 38 ( 1981 ) 683. [3] Th. Schimmel, W. Rie6, J. Gmeiner, G. Denninger, M. Schwoerer, H. Naarmann, N. Theophilou, Solid State Commun. 65 (1988) 1311. [4] Y.W. Park, Synth. Met. 45 ( 1991 ) 173. [5] S. Kivelson, A.J. Heeger, Synth. Met. 22 (1988) 371. [ 6] Y.W. Park, C. Park, Y.S. Lee, C.O. Yoon, H. Shirakawa, Y. Suezaki, K. Akagi, Solid State Commun. 65 (1988) 147. [7] J.F. Kwak, T.C. Clarke, R.L. Greene, G.B. Street, Solid State Commun. 31 (1979) 335. [ 8 ] Y.W. Park, J.C. Woo, K.H. Yoo, W.K. Han, C.H. Choi, T. Kobayashi, H. Shirakawa, Solid State Commun. 46 (1983) 731. [9] W. Pukacki, R. Zuzok, S. Roth, W. G6pel, in H. Kuzmany, M. Mehring, S. Roth (eds.), Electronic Properties of P o l y m e r s - Orientation
86
[ 10] [ 11 ] [ 12 ] [ 13 ] [ 14] [ 15]
Y. W. Park et al. / Synthetic Metals 96 (1998) 81-86
and Dimensionality of Conjugated Systems ( Kirchberg IV), Springer, Berlin, 1992, p. 106. T. Miyamae, T. Mori, K. Seki, J. Tanaka. Synth. Met. 69-71 (1995) 59. T. Miyamae, T. Mori, K. Seki, J. Tanaka, Bull. Chem. Soc. Jpn. 68 (1995) 803. C.O. Yoon, M. Reghu, A.J. Heeger, Phys. Rev. B 49 (1994) 10 85 I. M. Ahlskog, M. Reghu, A.J. Heeger, Phys. Rev. B 53 (1996) 15 529. W.P. Lee, Y.W. Park, Y.S. Choi, Synth. Met. 84-86 (1997) 841. K. Akagi, M. Suezaki, H. Shirakawa, H. Kyotani, M. Shimomura, Y. Tanabe, Synth. Met. 27-29 (3) (1989) DI.
[ 16] A.B. Kaiser, Synth. Met. 45 ( 1991 ) 183. [ 17] A.B. Kaiser, G. Diisberg, S. Roth, Phys. Rev. B 57 (1998) 1418. [ 18l Y.W. Park, A.J. Heeger, M.A. Druy, A.G. MacDiarmid, J. Chem. Phys. 73 (1980) 946. [ 19] P. Sheng, Phys. Rev. B 21 (1980) 2180. [20] N.F. Mott, E.A. Davis, Electronic Process in Non-Crystalline Materials, Clarendon Press, Oxford, 1979, p. 52. 121 ] D.K.C. MacDonald, Thermoelectricity: An Introduction to the Principles, Wiley, New York, 1962, p. 24. [22] E.S. Choi, Y.H. Seol, Y.S. Song, Y.W. Park, S.T. Hannahs, Synth. Met. 84-86 (1997) 685.