Jun 25, 2009 - thermoelectric figure of merit of the GICs is not as high as that of other thermoelectric materials; this is because of the small Seebeck coefficient.
Materials Transactions, Vol. 50, No. 7 (2009) pp. 1607 to 1611 Special Issue on Thermoelectric Conversion Materials V #2009 The Thermoelectrics Society of Japan
Thermoelectric Properties and Electrical Transport of Graphite Intercalation Compounds Rika Matsumoto1 , Yutaro Hoshina1; * and Noboru Akuzawa2 1 2
Faculty of Engineering, Tokyo Polytechnic University, Atsugi 243-0297, Japan Department of Chemical Science and Engineering, Tokyo National College of Technology, Tokyo 193-0997, Japan
Graphite intercalation compounds (GICs) have high electrical conductivities, large Seebeck coefficients, and low thermal conductivities as compared with their host graphite materials. Due to these properties, GICs are expected to act as effective thermoelectric materials. The thermoelectric figure of merit of the GICs is not as high as that of other thermoelectric materials; this is because of the small Seebeck coefficient and high thermal conductivity of GICs. However, the thermoelectric power factor of GICs is sufficiently high at present. In previous works, it was suggested that an increase in concentration of the intercalated species improves the thermoelectric performance of GICs. To better understand the thermoelectric properties of GICs, the dependence of thermoelectric properties with the electrical carrier density and mobility were investigated through the measurements of galvanomagnetic properties. As a result, the electrical conductivity of GICs slightly increases with the carrier density and the thermal conductivity increases with the carrier mobility. Furthermore, the carrier mobility decreases with an increase in the carrier density. In conclusion, the thermoelectric performance of GICs is suggested to be improved by an increase in the carrier density, that is, by an increase in the intercalate concentration. [doi:10.2320/matertrans.E-M2009813] (Received November 22, 2008; Accepted April 23, 2009; Published June 25, 2009) Keywords: graphite intercalation compound, thermoelectric property, carrier density, carrier mobility
1.
Introduction Intercalation
Graphite consists of hexagonal carbon planes stacked along c-axis by weak van der Waals force. Therefore, many chemical species, can intercalate into the gallery of the graphite interlayer to form graphite intercalation compounds (GICs) as shown in Fig. 1. The GICs have a unique stage structure as shown in Fig. 2. The ‘‘Stage’’ is defined as the number of graphite layers between intercalated layers. As charge transfer occurs between the intercalated species and adjacent graphite layers, the in-plane electrical conductivity of GICs become greater than that of the host graphite. The Seebeck coefficient of pristine graphite is nearly zero, because the number of electrons and the number of holes are almost equal. On the other hand, since GICs have a greater number of electrons (or holes), their Seebeck coefficients also become greater. The thermal conductivities of GICs around room temperature are lower than those of the host graphite.1,2) Due to these properties, GICs can be expected to act as effective thermoelectric materials. Thermoelectric materials have the ability to directly convert thermal energy into electricity. They have recently attracted considerably interest as promising materials for clean power generators. An excellent thermoelectric material should have a high electrical conductivity () to minimize Joule heating, a low thermal conductivity () to prevent thermal shorting, and a large Seebeck coefficient () for the maximum conversion of heat to electrical power. There are many advantages of using GICs as thermoelectric materials, such as the light weight of the materials, high selectivity of their shapes (plate, sheet, fiber, powder, etc.), and their harmless nature. The type of carrier in GICs changes with the intercalated species. Therefore, we can *Undergraduate
Student, Tokyo Polytechnic University
De-Intercalation
Graphite
Intercalate
GIC (Graphite Intercalation Compound)
Fig. 1
1st. stage
2nd. stage
Illustration of intercalation reaction.
3rd. stage
4th stage
5th stage
Graphite
: Graphite layer : Intercalate layer
Fig. 2 Models of stage structure.
fabricate both p and n-type GICs that are necessary to construct thermoelectric power generators. In the GICs, the electrical conductivity is governed mainly by the electrical carriers, while the thermal conductivity is governed mainly by the phonon transport. Therefore, it may be possible to control the electrical conductivity and the thermal conductivity independently. We have paid considerable attention to GICs as a candidate for thermoelectric materials. In previous work,3) we have clarified the thermoelectric properties of Cs-GICs and concluded as follows. The figures of merits (Z ¼ 2 =) of these GICs are not comparable to those of the other thermoelectric materials, although the power factors (P ¼ 2 ) are comparable. The simultaneous increase in the carrier density and decrease in the thermal conductivity were caused by the increase in the intercalate concentration. We consider this performance to be a great advantage for improving the thermoelectric performance of GICs.
1608
R. Matsumoto, Y. Hoshina and N. Akuzawa
In this paper, we report the thermoelectric properties of various types of GICs not only Cs-GICs. Moreover, we selected two types of graphite material, Grafoil and PGS, as host graphite materials, because the properties of the host graphite strongly affect those of the resulting GICs. Furthermore, we discuss the dependence of thermoelectric properties of GICs with the electrical carrier density and mobility, to find a strategy for improving thermoelectric performance of GICs.
-1 -1
κ / WK m
200
100
Experimental
2.2 Determination of properties The in-plane electrical conductivity, Seebeck coefficient, and thermal conductivity of the above mentioned GICs and their host graphite were measured at room temperature. The measurements of GICs were performed after the samples were exposed to air. Samples cut into 25-mm-long strips were used for these measurements. The electrical conductivities were measured by the four-terminal method in air. The thermal conductivity and Seebeck coefficient were measured under vacuum using basically the same system.3) The thermoelectric performance of a material was evaluated using the figure of merit Z ( 2 =) and power factor P ( 2 ), where is the electrical conductivity, is the Seebeck coefficient, and is the thermal conductivity. The Hall coefficient (RH ) and magnetoresistance (=) were also measured in air at room temperature by a conventional five-terminal method with magnetic fields (B) up to 0.5 T.6)
50 -1
2.1 Preparation of GICs Grafoil (GrafTech Co.), and PGS graphite sheets (Panasonic Co.) were used as the host graphite material. Grafoil is made from pure graphite flakes that have been exfoliated and sheeted by applying extreme heat and pressure.4) PGS is produced by the thermal decomposition of a polyimide film with excellent orientation.5) Grafoil sheets with a thickness of 0.30 mm were cut into 3:0 30 mm2 rectangles and heat-treated under vacuum at 900 C before use. PGS sheets with a thickness of 0.10 mm were cut into 2:0 30 mm2 rectangles. Alkali metals such as lithium (Li), potassium (K), and cesium (Cs) for n-type and metal chlorides such as FeCl3 , CuCl2 , and MoCl2 for p-type were used as the intercalate species. The GICs were prepared by allowing graphite to react with the vapor of the intercalate species at specific temperatures under vacuum. For some alkali metal-GICs prepared from Grafoil, molecules such as ethylene (C2 H4 ) were then absorbed into the vacant nano-space in the interlayer to improve their air stability.
|α | / µ VK
2.
300
30 20 10
3
4
10
5
10
10 -1
σ / Scm
Fig. 3 Dependence of thermal conductivity () and absolute Seebeck coefficient (jj) with electrical conductivity (). : Alkali metal (AM)GICs, : AM-ethylene-GICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AM-GICs, : AM-molecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
¼ ne
ð1Þ
1 RH ¼ ne
ð2Þ
In the case that magnetoresistance is detected, the transport system is considered as two-carrier conduction by both electrons and holes. Therefore, eqs. (3)–(5) are used with some suitable approximations. ¼ eðne e þ nh h Þ ð3Þ 1 ne e 2 nh h 2 e ðne e þ nh h Þ2 ne nh e h ðe þ h Þ2 2 B = ¼ ðne e þ nh h Þ2
RH ¼
3. 2.3 Estimation of carrier density and mobility The electrical carrier densities (ne and nh ) and mobilities (e and h ) were estimated from the electrical conductivity, Hall coefficient, and magnetoresistance using the following relations.6–10) In the case that magnetoresistance is not detected, the transport system is considered as one-carrier conduction by only electrons or holes. Therefore, the carrier density (n) and mobility () are estimated by eqs. (1) and (2).
40
ð4Þ ð5Þ
Results and Discussion
3.1 Thermoelectric properties of GICs We have investigated the thermoelectric properties of the GICs prepared from Grafoil and PGS. Figure 3 shows the dependence of the thermal conductivity () and absolute Seebeck coefficient (jj) with the electrical conductivity () for the GICs. The thermal conductivity depends on their host materials. In the case of GICs prepared from PGS, the thermal conductivities increase with the electrical conduc-
Thermoelectric Properties and Electrical Transport of Graphite Intercalation Compounds
Figure 4 shows the dependence of the figure of merit and power factor with the electrical conductivity in the GICs discussed in Fig. 3. The figures of merit increase with the electrical conductivity and there are two lines for each graphite host. Though the power factor also increases with the electrical conductivity, all the values lie on one line. This shows that the power factor of the GICs depend only on their electrical conductivity. Thus, the thermoelectric performance of GICs is strongly affected by their electrical conductivity. The electrical and thermal conductivity of GICs are 1 or 2 digits higher than those of other thermoelectric materials, and their absolute Seebeck coefficient is 1 digit smaller. As a result, the figures of merit of GICs are fairly small compared with those of thermoelectric materials in practical use. However, the power factors of GICs are sufficiently high compared to other materials. The power factors of almost all the GICs prepared from PGS are greater than 103 Wm1 K1 , which is the value for thermoelectric materials in practical use. We consider that the potential of GICs for use as a practical thermoelectric material is sufficiently high.
-4
10
-5
Z/K
-1
10
-6
10
-2
-1
10
-1
P / Wm K
1609
-3
10
3.2
Dependence of thermoelectric properties with electrical carrier of GICs All the thermoelectric properties: electrical conductivity, Seebeck coefficient, thermal conductivity, are dependent on the properties of their electrical carriers. To obtain additional details of the above observations and make a strategy for improving the thermoelectric performance of GICs, we attempted to determine the galvanomagnetic properties and estimate the density and mobility of the electrical carriers, electrons or holes. Table 1 summarizes the observed values of electrical conductivity (), Seebeck coefficient (), thermal conductivity (), Hall coefficient (RH ), and magnetoresistance (=) of the some of the GICs mentioned above. The estimated values of electron density (ne ), hole density (nh ), electron mobility (e ), and hole mobility (h ) are also shown in Table 1. The metal chloride-GICs, such as CuCl2 GIC and MoCl2 -GIC, are p-type materials with hole conduction and the alkali metal-GICs are n-type materials with electron conduction. If a GIC sample has an ideal composition and structure, it exhibits one-carrier conduction
-4
10
3
4
10
5
10
10 -1
σ / Scm
Fig. 4 Dependence of figure of merit (Z) and power factor (P) with electrical conductivity (). : Alkali metal (AM)-GICs, : AM-ethyleneGICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AM-GICs, : AM-molecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
tivity; this is similar to the behavior of simple metals. In contrast, in the case of GICs prepared from Grafoil, the thermal conductivity appears to decrease with the electrical conductivity. The absolute Seebeck coefficient decreases with the electrical conductivity; this behavior is also similar to that of metals.
Table 1 The thermoelectric properties, galvanomagnetic properties and the estimated carrier densities and mobilities for GICs and host graphite. Electrical Seebeck Thermal Hall coefficient Magnetoresistance conductivity coefficient conductivity at 0.4 T at 0.4 T
Density
Mobility
electron
hole
electron
hole
ne
nh
e
h
RH
=
/Scm1
/mVK1
/WK1 m1
/cm3 C1
/—
/cm3
/cm2 V1 s1
1:2 103
0:24
200
1:0 102
9:8 103
1:6 1018 1:6 1018
2:4 103
CuCl2 -GIC
4:3 103
34
84
5:1 102
3:5 103
3:7 1016 6:6 1019 1:1 104 4:0 102
CuCl2 -GIC
5:3 103
25
48
1:8 102
—
MoCl2 -GIC
2:7 103
30
57
5:7 102
2:2 103
4:9 1016 5:0 1019 7:6 103 3:3 102
4:5 103
9:4
480
1:6 101
8:3 102
2:2 1018 1:8 1018
190
3
Sample Grafoil
PGS
Symbol in Figures
5:1 10
4
K-GIC
5:5 10
4
28
170
4:3 10
Li-GIC CuCl2 -GIC
3:7 104 3:4 104
18 29
130 300
5:3 103 4:3 102
Cs-GIC
36
8:1 10
3
— — 2:2 103 7:2 103
3:4 1020
—
—
9:7 101
7:1 103
20
—
4:1 102
—
21
—
2:4 102
—
7:7 10 1:4 10
1:0 1021 2:4 1015 2:2 102 5:0 104 3:0 1016 1:3 1020 3:1 104 1:3 103
1610
R. Matsumoto, Y. Hoshina and N. Akuzawa (a)
(b) 5
5
10
10
2
-1
σ / Scm
-1
3
σ / Scm
10
2
µ / cm V s
-1 -1
10
4
10
3
10
4
10
3
10
20
10
10
21
10
2
-3
-3
n / cm
Fig. 6 Dependence of electrical conductivity () with carrier density (n) and carrier mobility (). : Alkali metal (AM)-GICs, : AM-ethyleneGICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AM-GICs, : AM-molecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
Fig. 5 Dependence of carrier mobility () with carrier density (n). : Alkali metal (AM)-GICs, : AM-ethylene-GICs, : FeCl3 - and CuCl2 GICs prepared from PGS, : AM-GICs, : AM-molecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
(b) 300
-1
300
-1
κ / Wm K
by only electrons or holes. Some GIC samples exhibited two-carrier conduction by both electrons and holes like the host graphite. This is due to the partial decomposition of the samples by exposure to air, or to the non-intercalated part of the graphite. However, the GIC samples with two-carrier conduction can be approximated as one-carrier conduction, because the density of minor-carrier is quite smaller than that of major-carrier. Figure 5 shows the correlation between the carrier density (ne or nh ) and the carrier mobility (e or h ) of the GICs presented in Table 1, together with the differences of their host graphite. This correlation is very clear. Although the absolute values depend on their host graphite, the tendencies are nearly identical. As the intercalate species intercalate into the interlayer of graphite, the density of electrons or holes on graphite planes bounded to the intercalated species increase. The carrier mobility clearly decreases with the carrier density. It suggests that the intercalation of species into the interlayer of graphite causes a decrease in the carrier mobility, in addition to an increase in the carrier density. This is considered to be because that the intercalation process produces many defects in the graphite crystallites and the intercalated species have a scattering effect on the carrier transport. This is also the reason why the thermal conductivity of GICs is smaller than that of the host graphite. The correlations between the thermoelectric properties (, , and ) and carrier density (n) and mobility () are summarized in Figs. 6–8. Here, the carrier density and mobility refer to the density and mobility of their major carrier. The variation of electrical conductivity () with the carrier density is shown in Fig. 6(a), with the mobility in Fig. 6(b). As there is a direct correlation between carrier mobility and density as shown in Fig. 5, these two plots are symmetrical. It seems that the electrical conductivity very slightly increases with the carrier density. It is known that the electrical conductivity of GICs is determined from the valance of high-density and low-mobility. In case of alkali metal-GICs prepared from high-quality graphite like highly oriented pyrolytic graphite (HOPG), the electrical conductivity is the greatest at stage 5, having fairly low concen-
(a)
-1
21
200
-1
10
κ / Wm K
20
3
-1 -1
µ / cm V s
n / cm
10
10 2
100
20
200
100
21
10
10 -3
0
2
3
10
10 2
-1 -1
µ / cm V s
n / cm
Fig. 7 Dependence of thermal conductivity () with carrier density (n) and carrier mobility (). : Alkali metal (AM)-GICs, : AM-ethylene-GICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AM-GICs, : AMmolecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
tration of intercalated species,11) because HOPG have highmobility (1:1 104 cm2 V1 s1 10)). On the other hand, in case of GICs prepared from Grafoil, we have found that the electrical conductivity is the greatest at stage 1, having the highest concentration of intercalated species. Figure 7(b) shows the dependence of thermal conductivity () with carrier mobility (). In the GICs prepared from Grafoil, the change of thermal conductivities with the carrier density is quite small. On the other hand, the thermal conductivities increase with the carrier mobility in the GICs prepared from PGS. This is because the high thermal conductivity of PGS arises from its large carrier mobility. In the same way, Fig. 7(a) shows the dependence of thermal conductivity () with carrier density (n). The thermal conductivity decreases with the carrier density. This result indicates that the intercalated species have a scattering effect on the carrier and phonon transport. The thermal conductivity in a solid can be written as ¼ L þ E
ð6Þ
where L and E are the lattice and electronic contribution to total thermal conductivity, respectively. The electronic contribution (E ) can be estimated the by the WiedemannFranz law. Figure 8(a) shows the dependence of the lattice contribution (L ) with carrier density (n). Figure 8(a) is almost similar with Fig. 7(a); this indicates that the thermal conductivity of GICs is governed mainly by the lattice
Thermoelectric Properties and Electrical Transport of Graphite Intercalation Compounds (a)
(b)
50 300 -1 -1
κ E / Wm K
-1
κ L / Wm K
-1
40
200
100
30 20 10 0
0
20
21
10
20
10
21
10
-3
10 -3
n / cm
n / cm
Fig. 8 Dependence of electronic contribution (L ) and lattice contribution (E ) to total thermal conductivity with carrier density (n) and carrier mobility (). : Alkali metal (AM)-GICs, : AM-ethylene-GICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AM-GICs, : AMmolecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil. (a)
(b)
40
-1
30
|α | / µ VK
|α | / µ VK
-1
40
20
20
21
10
10 -3
n / cm
30
20
2
3
10
10 2
-1 -1
µ / cm V s
Fig. 9 Dependence of absolute Seebeck coefficient (jj) with carrier density (n) and carrier mobility (). : Alkali metal (AM)-GICs, : AMethylene-GICs, : FeCl3 - and CuCl2 -GICs prepared from PGS, : AMGICs, : AM-molecules-GICs, : FeCl3 - and CuCl2 -GICs prepared from Grafoil.
contribution. Figure 8(b) shows the dependence of the electronic contribution (E ) with carrier density (n). In the GICs prepared from Grafoil, the change of the electronic contribution with the carrier density is quite small. On the other hand, the electronic contribution increase with the carrier density in the GICs prepared from PGS having highcarrier density. However, as the dependence of the electronic contribution on the carrier density is smaller than that of the lattice contribution, it can be concluded that the thermal conductivity of GICs decreases with the carrier density. Figure 9(a) and (b) show the change of absolute Seebeck coefficient (jj) with carrier density (n) and mobility (). It seems that the absolute Seebeck coefficient values decrease with carrier density and also increase with carrier mobility. In case of K-GICs12) and SbCl2 -GIC13) prepared from HOPG, it was reported that the absolute Seebeck coefficient is also greatest around stage 5 similar to the electrical conductivity. However, the change in the values of absolute Seebeck coefficient is very small compared with the electrical and thermal conductivity. Therefore, we consider that controlling of the Seebeck coefficient is difficult and not effective to improve the thermoelectric performance of GICs. From these observations, we gained some important insight on how to increase the thermoelectric performance of GICs. We consider an increase in the carrier density to be
1611
the most important and effective way. An increase in the carrier density is accompanies by a decrease in carrier mobility in GICs. The increase in carrier density causes an increase in the electrical conductivity and the decrease in carrier mobility causes a decrease in the thermal conductivity. A method of increasing the carrier density of GICs is to increase the density of the intercalated species in the interlayer of graphite. Furthermore, it is also important to select a suitable host graphite material. It is found that PGS is more suitable for improving the thermoelectric performance of GICs compared with Grafoil. This is attributed to its large carrier mobility which originates from the large graphite crystallite size and high orientation of graphite layers. 4.
Conclusions
The thermoelectric property of some types of GICs prepared from Grafoil and PGS and their correlation with the electrical carrier density and mobility are investigated. The power factors of GICs are sufficiently high as compared with those of practical thermoelectric materials in spite of the small Seebeck coefficient of GICs. The electrical conductivity of GICs slightly increases with the carrier density and the thermal conductivity increases with the carrier mobility. Furthermore, the carrier mobility decreases with an increase in the carrier density. In conclusion, the thermoelectric performance of GICs is suggested to be improved by an increase in the carrier density, that is, by an increase in the intercalate concentration. Taking into account the advantages of using GICs as thermoelectric materials, it is confirmed that GICs have significant potential to be used as thermoelectric materials. Acknowledgments The present work was partly supported by funding from the Thermal & Electric Energy Technology Foundation. REFERENCES 1) J. Heremans, J.-P. Issi, I. Zabala-martinez, M. Shayegan and M. S. Dresselhaus: Phys. Lett. 84A (1981) 387–389. 2) J.-P. Issi, H. Heremans and M. S. Dresselhaus: Phys. Rev. B 27 (1983) 1333–1347. 3) R. Matsumoto, N. Akuzawa and Y. Takahashi: Mater. Trans. 47 (2006) 1458–1463. 4) C. Uher: Phys. Rev. B 25 (1982) 4167–4172. 5) M. Murakami, N. Nishiki, K. Nakamura, J. Ehara, H. Okada, T. Kouzaki, K. Watanabe, T. Hoshi and S. Yoshimura: Carbon 30 (1992) 255–262. 6) R. Matsumoto, Y. Takahashi and N. Akuzawa: TANSO 2001 [No. 198] (2001) 129–133. 7) R. Matsumoto: TANSO 2003 [No. 209] (2003) 174–178. 8) A. Marchand and R. Mathur: Carbon 27 (1989) 349–357. 9) N. Akuzawa, S. Takei, M. Yoshioka, Y. Hishiyama and Y. Takahashi: Carbon 29 (1991) 899–903. 10) N. Akuzawa, S. Kondow, Y. Kaburagi, Y. Hishiyama and Y. Takahashi: Carbon 31 (1993) 963–968. 11) E. Mcrae, D. Billaud, J. F. Mareche and A. Herold: Physica 99B (1980) 489–493. 12) L. C. F. Blackman, J. F. Mathews and A. R. Ubbelohde: Proc. R. Soc. 258A (1960) 339. 13) M. Elizinga, D. T. Morelli and C. Uher: Phys. Rev. B 26 (1982) 3312– 3319.