FEATURE ARTICLES

BULLETIN

Graphene-Based Thermoelectric Materials ZHEN-YU JUANG, CHIEN-CHIH TSENG, CHANG-HSIAO CHEN AND LAIN-JONG LI INSTITUTE OF ATOMIC AND MOLECULAR SCIENCES ACADEMIA SINICA, TAIWAN

ABSTRACT It is well known that the thermoelectric (TE) effect is exhibited in most materials, but very few materials have a large enough TE effect for practical applications. A good thermoelectric material must involve a high Seebeck coefficient and excellent electrical conductivity, as well as low thermal conductivity. The thermoelectric figure of merit ZT is normally used to indicate the energy conversion efficiency. Since 2004 the excellent electrical properties of graphene were discovered, these research efforts for utilizing graphene as a candidate of future thermoelectric materials are ongoing. Some strategies and bottlenecks in the improvement of ZT for graphene are discussed.

ing to recycle waste heat from diverse sources, reasonable conversion efficiency is difficult to be realized using these types of materials [2]. It is well known that the lower flexibility and ductility of bismuth telluride leads to difficulty combining it with the current types of small portable devices such as mobile phones and laptop computers, not to mention some future devices involving irregular surfaces or flexible substrates. Another disadvantage of bismuth telluride-based TE materials is that these materials are not transparent. In other words these materials cannot be applied on the windows glass of buildings or vehicles, which are good power sources for temperatureto-electricity conversion since a large temperature difference normally exists on the two sides of the glass. ENERGY CONVERSION EFFICIENCY

INTRODUCTION Thermoelectric devices are made by materials which can directly convert the temperature difference to an electrical voltage once there is a temperature gradient between both the hot and cold sides of the material. Meanwhile, the temperature difference can be generated by pumping a current through both side of the TE material. These TE phenomena are well known as the temperature-toelectricity conversion (Seebeck effect) and current-totemperature conversion (Peltier effect). In fact, most materials show some TE effect, but very few materials have a large enough TE effect for practical applications. Good TE materials should exhibit an outstanding TE effect which can be used in applications including power generation, refrigeration, and most importantly waste heat recycling [1-3]. Commercial TE devices in such applications are commonly made with bismuth telluridebased materials. However, in some cases when attempt-

14

The energy conversion efficiency of TE materials can be obtained by a dimensionless factor, the figure of merit ZT=S2σT/k, where T, S, σ and k are average temperature, Seebeck coefficient, electrical conductivity and thermal conductivity, respectively. In terms of thermal conductivity, k=ke+kL where ke and kL are the thermal conductivities contributed by charge carriers and phonons. In most cases, the value of ZT ranges between 0.5 and 1.0, which represents the conversion efficiency of between 2%-8% in the common temperature range. One may quickly realize the key for a TE material with high efficiency is that the material should have a high power factor (S2σ) and low thermal conductivity according to above-mentioned equation. To look for high TE efficiency material, it is crucial to find material possessing a high Seebeck coefficient or good electrical conductivity, followed by reducing its thermal conductivity to achieve a high ZT. However, S is the intrinsic property of a mate-

FEATURE ARTICLES

AUGUST 2013 VOL. 23 NO. 4

Fig. 2: CVD-synthesized SLG with different grain size via different CVD conditions. The grain size could be controlled in the CVD process.

Fig. 1: The in-plane thermal conductivity of graphite is up to 5,000 W/mK [10].

rial that is difficult to modulate. Besides, based on the Weideman-Franz Law, the value of σT/ke should be a constant. In other words, modulating kL is actually the only way to reduce the effective thermal conductivity. In recent decades the most popular strategy for reducing kL is to adopt nanomaterials or nanostructures that have the dimensions close to Phonon de Broglie’s wavelength, which allows phonons to scatter during the heat transport process [4-7].

Fig. 3: (a) AFM image of the isolated graphene grains with butterfly-like shape, in which some gaps were partially closed to form grain boundaries. (b) The high magnitude AFM image of square area in (a).

GRAPHENE-BASED IN-PLANE TE EFFECT Since the excellent electrical properties of graphene have been discovered [8, 9], this 2D nanomaterial is expected to become a potential TE material. Although graphene has extremely high carrier mobility, which results in excellent electrical conductivity [11], the high thermal conductivity could be up to 5,000 W/mK [10] (based on the known thermal conductivity of graphite in Figure 1) with a poor Seebeck coefficient of 30–60 µV/K [12, 13] which leads to low energy conversion efficiency. In 2007, Dragoman et al. reported the first experimental report for graphene-based TE material [12]. They used a graphene-based interference device with the structure of dual-gate field-effect transistors. By modulating the gate voltage, the Seebeck coefficient could rise to 30 mV/K, which is 10 times higher than most commercial materials. Even though the acceptable thermal conductivity should be lower than 10 W/mK for commercial devices, their results truly opened the possibility of using graphene for future TE applications.

Grain boundaries of synthetic graphene such as those obtained by chemical vapor deposition (CVD) [14, 15] could scatter phonons and result in a slight reduction of kL [16], which indicates that the kL reduction might be well controlled by the CVD process. For example, the optical microscope images in Figure 2 show the CVDsynthesized single-layer graphene (SLG) via different synthetic conditions. The grain size and the morphology of the grain boundaries could be controlled. Note that the gaps in-between graphene grains were generated on purpose during the CVD process followed by an oxidation procedure of copper substrates [17]. The oxidation procedure makes the gaps visible for observation. Figure 3 shows the AFM image of the isolated graphene grains, in which some gaps were partially closed to become grain boundaries. Obviously, the result implies the kL could be modulated by the density of the grain boundaries which are dominated by the CVD process. Meanwhile, the chemical adsorption or particle decoration on the graphene grain boundaries has been achieved[18-20].

15

FEATURE ARTICLES

BULLETIN

The in-plane thermal conductivity might be dramatically reduced by ad-atoms or ad-molecules since these atoms, particles or molecules may hinder the phonon transport. BAND GAP MODULATING Another possible way to improve ZT is through the band-structure modulation of graphene, including band gap opening and modification of carrier concentrations. These methodologies focus on increasing the power factor, which directly connects to the excellent electrical conductivity of graphene. In the efficiency equations of ZT, both thermal and electrical conductivities are the functions of carrier mobility, and compete with each other. In an insulator, naturally low carrier mobility results in a high Seebeck coefficient and low electrical conductivity. Conversely, the conductor has a low Seebeck coefficient and high electrical conductivity. Therefore, semiconductors usually have the appropriate carrier mobility and reasonably efficient power factor S2σ [21]. It is expected that the power factor can improve since the chemical adsorption on the graphene basal plane modulates the electronic structures of graphene, making it to less-metallic [18]. Recently, researchers have shown that a slight O2-plasma treatment on a few layers of graphene (FLG) could improve the Seebeck coefficient up to 700 µV/K at 575K, as shown in Figure 4. Meanwhile, the electrical conductivity was kept at a high value of ~104 S/m. Thus, the max power factor achieved was ~4.5×10-3 W/K2m [22]. The process of O2-plasma treatment in-

Fig. 5: Edge roughness and vacancies of GNR could reduce the thermal conductivity and improve the Seebeck coefficient; however the ZT is still suppressed by the reduced electrical conductivity[27].

duced defects on the FLG leads to the band gap opening, modification of carrier concentration and enhancement of phonon scattering. GRAPHENE NANORIBBONS Graphene nanoribbons (GNRs) have qusi-one-dimensional geometry with hexagonal carbon lattices, which are narrow stripes of graphene. Typically the width of GNR is just a few nanometers. The structures, electronic and magnetic properties of GNRs have been experimentally and theoretically studied in detail [23-26]. Because of the various edge structures, GNRs present different identities from normal semiconductors and semi-metals, leads to the possibility of GNRs as future electric devices.

Fig. 4: (a) HRTEM image of FLG after O2-plasma treatment. The inset of the SAED pattern confirms the amorphous status after O2-plasma treatment. The yellow circles highlight small

Fig. 6: The specially designed halogen-substituted bianthryl monomers was coated onto

crystals of graphene in such films while the red circles point out the disordered arrangement

gold and silver surfaces in a vacuum. After a two-step thermolysis process at 250℃ and

of carbon atoms. (b) and (c) are the Seebeck coefficient (thermopower) and the power

440℃, the monomers were linked up to form polyphenylene chains followed by the

factor for the FLG films after different O2-plasma treatments, respectively[22].

interconnection of polymer chains to construct planar, aromatic GNRs [29].

16

FEATURE ARTICLES

AUGUST 2013 VOL. 23 NO. 4

Low-dimensional materials are expected to show good TE effects [4-7, 28]. The first GNR TE effect was published in 2009 [27]. In the report the authors discussed the effect of edge roughness and vacancies on TE phenomena. As shown in Figure 5, although edge roughness and vacancies could reduce the thermal conductivity and thus improve the Seebeck coefficient, the ZT is still suppressed by the reduced electrical conductivity. This report clearly demonstrated that the ZT should be improved through the independent adjustment of S, σ and k. According to Weideman-Franz’s Law, the Sσ/ke should be constant and the reduction of kL in GNRs becomes the major challenge for improving ZT.

substrate. Apparently, more efforts should be made to enhance the TE performance of graphene materials as well as graphene-based TE devices.

Reference [1] Karni J., Solar energy: The Thermoelectric Alternative. Nature Materials. 011;10(7):481-2. [2] Vining CB., An Inconvenient Truth About Thermoelectrics. Nature Materials. 2009;8(2):83-5. [3] Majumdar A., Thermoelectric devices: Helping Chips to Keep their Cool. Nature Nanotechnology. 2009;4(4):214-5. [4] M. S. Dresselhaus GDXSZZSBCTKJYYGC., THE PROMISE OF LOW-DIMENSIONAL THERMOELECTRIC MATERIALS. Microscale Thermophysical Engineering. 1999;3(2):89-100. [5] Venkatasubramanian R, Siivola E, Colpitts T, O'Quinn B., Thin-film Thermoelectric Devices with High Room-temperature Figures of Merit. Nature. 2001;413(6856):597-602.

Since vacancies can reduce either the thermal or the electrical conductivities of graphene, controlling edge roughness to independently reduce kL and keeping the power factor at a reasonably high level seem desirablefor achieving high ZT. In other words, the use of kinked GNRs would be a good approach [30-35]. However, the synthesis of GNRs or kined GNRs is technically difficult [36, 37]. Fortunately, the first synthesis of kinked GNRs was realized in 2010 [29]. In this report, a simple, surface-based bottom-up chemical process was introduced as shown in Figure 6. Researchers first used the specially designed halogen-substituted bianthryl monomers, and performed the designed two-step thermolysis on gold and silver surfaces in a vacuum to form various GNRs. In most cases the edges of the GNRs were smooth and armchair-shaped, and the ribbons themselves were either straight or kinked, depending on the monomers they designed [29, 38]. Unfortunately, these as-grown GNRs are extremely difficult to be transferred from gold or silver substrates to desired substrates for testing due to the strong coupling in-between GNRs and catalytic metal substrates. Also, the size of the synthetic GNRs is still too small for TE measurements. In other words, the outstanding TE properties of GNRs were predicted in theory, but are hard to prove at this moment.

[6] Boukai AI, Bunimovich Y, Tahir-Kheli J, Yu J-K, Goddard Iii WA, Heath JR., Silicon Nanowires as Efficient Thermoelectric Materials. Nature. 2008;451(7175):168-71. [7] Hochbaum AI, Chen R, Delgado RD, Liang W, Garnett EC, Najarian M, et al., Enhanced Thermoelectric Performance of Rough Silicon Nanowires. Nature. 2008;451(7175):163-7. [8] Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, et al., Electric Field Effect in Atomically Thin Carbon Films. Science. 2004;306(5696):666-9. [9] Novoselov KS, Geim AK, Morozov SV, Jiang D, Katsnelson MI, Grigorieva IV, et al., Twodimensional Gas of Massless Dirac Fermions in Graphene. Nature. 2005;438(7065):197-200. [10] Balandin AA., Thermal Properties of Graphene and Nanostructured Carbon Materials. Nature Materials. 2011;10(8):569-81. [11] Geim AK, Novoselov KS., The Rise of Graphenes. Nature Materials. 2007;6(3):183-91. [12] Dragoman D, Dragoman M., Giant Thermoelectric Effect in Graphenes. Applied Physics Letters. 2007;91(20):203116-3. [13] Li X, Yin J, Zhou J, Wang Q, Guo W. Exceptional High Seebeck Coefficient and Gas-flowinduced Voltage in Multilayer Graphenes. Applied Physics Letters. 2012;100(18):183108-4. [14] Su C-Y, Lu A-Y, Wu C-Y, Li Y-T, Liu K-K, Zhang W, et al., Direct Formation of Wafer Scale Graphene Thin Layers on Insulating Substrates by Chemical Vapor Deposition. Nano Letters. 2011;11(9):3612-6. [15] Ching-Yuan S, Dongliang F, Ang-Yu L, Keng-Ku L, Yanping X, Zhen-Yu J, et al., Transfer Printing of Graphene Strip from the Graphene Grown on Copper Wires. Nanotechnology. 2011;22(18):185309. [16] Lu Y, Guo J., Thermal Transport in Grain Boundary of Graphene by Non-equilibrium Green's Function Approach. Applied Physics Letters. 2012;101(4):043112-5. [17] Lu A-Y, Wei S-Y, Wu C-Y, Hernandez Y, Chen T-Y, Liu T-H, et al., Decoupling of CVD Graphene by Controlled Oxidation of Recrystallized Cu. RSC Advances. 2012;2(7):3008-13. [18] Zhang W, Lin C-T, Liu K-K, Tite T, Su C-Y, Chang C-H, et al. Opening an Electrical Band Gap of Bilayer Graphene with Molecular Doping. ACS Nano. 2011;5(9):7517-24. [19] Shi Y, Kim KK, Reina A, Hofmann M, Li L-J, Kong J., Work Function Engineering of Graphene Electrode via Chemical Doping. ACS Nano. 2010;4(5):2689-94. [20] Chen T-Y, Loan PTK, Hsu C-L, Lee Y-H, Tse-Wei Wang J, Wei K-H, et al., Label-free Detection of DNA Hybridization Using Transistors Based on CVD Grown Graphene.

SUMMARY Since the Fukushima Daiichi nuclear disaster in 2011, recycling waste heat from diverse sources using TE materials with reasonable conversion efficiency has become crucial in future power management. The TE research using graphene-based materials is rising and has made admirable progress in recent years. The irreplaceable advantage of graphene-based TE material is that the device could be manufactured on atransparent/flexible

Biosensors and Bioelectronics. 2013;41(0):103-9. [21] Vining CB. ,Semiconductors Are Cool. Nature. 2001;413(6856):577-8. [22] Xiao N, Dong X, Song L, Liu D, Tay Y, Wu S, et al., Enhanced Thermopower of Graphene Films with Oxygen Plasma Treatment. ACS Nano. 2011;5(4):2749-55. [23] Barone V, Hod O, Scuseria GE., Electronic Structure and Stability of Semiconducting Graphene Nanoribbons. Nano Letters. 2006;6(12):2748-54. [24] Brey L, Fertig HA., Electronic States of Graphene Nanoribbons Studied with the Dirac Equation. Physical Review B. 2006;73(23). [25] Owens FJ., Electronic and Magnetic Properties of Graphene Nanoribbons. Molecular Physics. 2006;104(19):3107-9. [26] Son YW, Cohen ML, Louie SG., Half-metallic Graphene Nanoribbons. Nature. 2006;444(7117):347-9.

17

FEATURE ARTICLES

BULLETIN

[27] Ouyang Y, Guo J., A Theoretical Study on Thermoelectric Properties of Graphene Nanoribbons. Applied Physics Letters. 2009;94(26):263107-3. [28] Nakada K, Fujita M, Dresselhaus G, Dresselhaus MS., Edge State in Graphene Ribbons: Nanometer Size Effect and Edge Shape Dependence. Physical Review B. 1996;54(24):17954-61.

[34] Xie Z-X, Tang L-M, Pan C-N, Li K-M, Chen K-Q, Duan W.,Enhancement of Thermoelectric Properties in Graphene Nanoribbons Modulated with Stub Structures. Applied Physics Letters. 2012;100(7):073105-4. [35] Yang K, Chen Y, D'Agosta R, Xie Y, Zhong J, Rubio A., Enhanced Thermoelectric Properties in Hybrid Graphene/Boron Nitride Nanoribbons. Physical Review B.

[29] Cai J, Ruffieux P, Jaafar R, Bieri M, Braun T, Blankenburg S, et al., Atomically Precise Bottom-up Fabrication of Graphene Nanoribbons. Nature. 2010;466(7305):470-3. [30] Chen Y, et al., Thermoelectric Properties of Graphene Nanoribbons, Junctions and Superlattices. Journal of Physics: Condensed Matter. 2010;22(37):372202. [31] Sevin, Ccedil, li H, Cuniberti G., Enhanced Thermoelectric Figure of Merit in Edgedisordered Zigzag Graphene Nanoribbons. Physical Review B. 2010;81(11):113401. [32] Huang W, Wang J-S, Liang G., Theoretical Study on Thermoelectric Properties of Kinked Graphene Nanoribbons. Physical Review B. 2011;84(4):045410.

2012;86(4):045425. [36] Jiao LY, Zhang L, Wang XR, Diankov G, Dai HJ., Narrow Graphene Nanoribbons from Carbon Nanotubes. Nature. 2009;458(7240):877-80. [37] Jiao L, Wang X, Diankov G, Wang H, Dai H. Facile, Synthesis of High-quality Graphene Nanoribbons. Nat Nano. 2010;advance online publication. [38] Muller S, Mullen K., Expanding Benzene to Giant Graphenes: Towards Molecular Devices. Philosophical Transactions of the Royal Society a-Mathematical Physical and Engineering Sciences. 2007;365(1855):1453-72.

[33] Liang L, Cruz-Silva E, Girão EC, Meunier V., Enhanced Thermoelectric Figure of Merit in Assembled Graphene Nanoribbons. Physical Review B. 2012;86(11):115438.

18

Dr. Lain-Jong Li obtained his PhD in condensed matter physics from Oxford University, United Kingdom in 2006. He was an assistant professor in the school of materials science and engineering at Nanyang Tech Univ Singapore. Since 2010, he has become an associate research fellow at Academia Sinica Taiwan.

Dr. Chang-Hsiao Chen obtained his PhD in the Institute of Nano Engineering and Micro Systems at National Tsing Hua University, Taiwan in 2011. After one year of military service, he became a postdoctoral fellow at the Institute of Atomic and Molecular Sciences at Academia Sinica, Taiwan.

Chien-Chih Tseng obtained his MSc in OptoMeachatronics at National Chung Cheng University in 2012. Since 2012, he has become a research assistant at Academia Sinica Taiwan.

Dr. Zhen-Yu Juang obtained his PhD in the Department of Engineering and System Science from National Tsing Hua University, Taiwan in 2004. He was a postdoctoral researcher of Max-Planck Institute of Polymer Research in Germany until early 2012. Since 2012, he has become a postdoctoral researcher at Academia Sinica Taiwan.