2064 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 60, NO.

6, JUNE 2013

Thermoelectric Performance of a Single-Layer

Graphene Sheet for Energy Harvesting
Rekha Verma, Student Member, IEEE, Sitangshu Bhattacharya, Member, IEEE,
and Santanu Mahapatra, Senior Member, IEEE

Abstract— We investigate the thermoelectric (TE) figure-of- figure-of-merit ZT, which is defined as [3]
merit of a single-layer graphene (SLG) sheet by a physics-
based analytical technique. We first develop analytical models σ S 2B T
ZT = (1)
of electrical and thermal resistances and the Seebeck coefficient κe + κph
of SLG by considering electron interactions with the in-plane
and flexural phonons. Using those models, we show that both where the transport coefficients σ , S B , κe , and κph are the
the figure-of-merit and the TE efficiency can be substantially electrical conductivity, Seebeck coefficient, and electron- and
increased with the addition of isotope doping as it significantly phonon-dominated thermal conductivities, respectively. Exten-
reduces the phonon-dominated thermal conductivity. In addition, sive experimental studies [3]–[12] report an improvement of
we report that the TE open circuit output voltage and output ZT by varying the chemical composition and structural con-
power depends weakly on the SLG sheet dimensions and sheet
concentration in the strongly diffusive regime. Proposed models finements of the constituent compounds. Recently TE studies
agree well with the available experimental data and demonstrate on Si nanowires bundles demonstrated a large Voc and power
the immense potential of graphene for waste-heat recovery under a temperature difference of 70 K [12]. This is achieved,
application. however, at the cost of large integration of bundles [11], where
Index Terms— Energy harvesting, graphene, phonons, seebeck the open-circuit voltage (Voc ), ZT, and efficiency (η) “per
coefficient, thermal conductivity, thermoelectric figure of merit. wire” could be very low.
Single-layer graphene (SLG) sheet has demonstrated excel-
lent electrical and thermal conductivities [13], [14]. However,
I. I NTRODUCTION
its potentiality as a TE material by analyzing the ZT and Voc

T HERMOELECTRIC materials are the promising energy-

harvesting sources for waste-heat recovery applications.
In last few decades, a vast improvement in thermoelectric (TE)
still requires sufficient investigations. In recent years, there
has been some efforts in determining the ZT for graphene
nanoribbons through various molecular dynamics (MD)
figure-of-merit (ZT) has been resulted due to phonon mean- and nonequilibrium Green’s function (NEGF) simulations
free-path engineering by introducing phonon scattering inter- [15]–[18]. These are particularly based on the numerical
faces or surfaces [1]–[3]. With the increased hot-spots inside evaluations of ZT without considering the interactions between
ICs, nanoscale TE generators (TEGs) are in high demand electrons and the in-plane and out-of-plane (ZA or flexural)
because of their ability to drive other devices efficiently from phonons, which are now reported to control these parameters
the waste heat generated by the chip, where the temperature [19]–[24] when the SLG sheet relaxes with the end contacts
rise can be as high as 400 K. Conventional bulk TE materials, [21]. It is worth noting that in general for most of the
such as Bi2 Te3 , CoSb3 , etc., exhibit a high ZT (∼ 1 near materials, these electrical and thermal transport coefficients
400 K and ∼ 0.8 near 800 K [4]). However, the bulk-to- are dependent on the Fermi energy (E F ) that limits the value
nano-fabrication methodologies involved in realizing the TEG of ZT and hence the TEG performance. However, in case of
arrays are generally very complex [4]–[7]. This is also very SLG sheet since κph  κe [13], one can engineer to maximize
challenging to geometrically cover the entire hot-spot region the ZA phonon interactions to reduce κph significantly. Thus,
of the IC. The TE efficiency is evaluated by the dimensionless the motivation in this work is focused on the analysis of the
role of ZA phonon interactions on the electrical and thermal
Manuscript received January 14, 2013; revised April 5, 2013; accepted conductivities and Seebeck coefficient. This is required to
April 10, 2013. Date of publication May 6, 2013; date of current version
May 16, 2013. The review of this paper was arranged by Editor estimate the TE efficiency, short-circuit current (Isc ), Voc , and
R. Venkatasubramanian. output power, which may find substantial role in recovering
R. Verma and S. Mahapatra are with the Nano-Scale Device Research the waste-heat from the chip hot-spots.
Laboratory, Department of Electronic Systems Engineering, Indian Institute
of Science, Bangalore 560012, India (e-mail: [email protected]; In the following, we formulate a physics-based analytical
[email protected]). solution of the TE transport coefficients of SLG sheet using
S. Bhattacharya was with the Nano-Scale Device Research Laboratory, Landauer’s approach under strongly diffusive transport regime.
Department of Electronic Systems Engineering, Indian Institute of Science,
Bangalore 560012, India. He is with the Department of Electrical Engi- We demonstrate the effect of geometry, temperature, and sheet
neering, Shiv Nadar University, Dadri 203207, India (e-mail: sitangshu. carrier concentration (n 2D ) on the ZT and open circuit TE
[email protected]). voltage. We further show that the ZT and η can be considerably
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. enhanced at room temperature by the addition of isotope
Digital Object Identifier 10.1109/TED.2013.2258159 impurity (C13 over conventional C12 ), which decreases κph to
0018-9383/$31.00 © 2013 IEEE

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 VERMA et al.: THERMOELECTRIC PERFORMANCE OF A SINGLE-LAYER GRAPHENE SHEET 2065

Fig. 1. (a) Schematic of an SLG sheet of dimension L × W . The contacts

1 and 2 are assumed to be ideal and at temperature TH and TL , respectively.
(b) Averaging of the carrier back-scattering length over the entire SLG sheet.

a large extent. We find that an SLG under strongly diffusive

regime can result a Voc of around 14 mV in the temperature
difference of around 100 K at a carrier concentration of Fig. 2. R as a function of T with varying carrier concentration. The symbols
1016 m−2 . Our theoretical formulation of Voc and ZT can are the experimental data from [25], while the lines are our analytical curves.
be compared with the experimental data from Si nanowire
bundles arrays [12]. The methodology presented here exhibits
a viable analytical technique to assess the applicability of the Here, g ≈ 3 eV is the screened deformation potential constant,
SLG-based TE device. We also find that the TE transport coef- β ≈ 2–3, ς = 1 eV is the bending rigidity [24], a = 1.4Å
ficients, electrical resistance (R), and S B exhibit an excellent is the distance between nearest carbon atoms, ρ = 7.6 ×
match with the available experimental data over a wide range 10−7 kg m−2 is the mass density, υ L = 2.1 × 104 m s−1 and
of temperatures [25], [26]. υT = 1.4×104 ms−1 are the longitudinal and transverse sound
velocities, respectively [24], k B is the Boltzmann’s constant,
II. M ODEL D EVELOPMENT AND R ESULTS h̄ = (h/2π ), and h is the Planck’s constant. In deriving (5)
and (6), it is assumed that the Bloch-Gruneisen temperature
We consider an SLG sheet as shown in Fig. 1(a), whose
(TBG = 57, 38 and 0.1 K) is far less than the room temperature
length (L) and width (W ) are along x and y directions with the
for longitudinal, transverse, and flexural phonons, respectively,
left and right contacts at temperatures TL and TH , respectively.
with the domination of the absorption or emission of two
Because of the distribution of carrier velocity along both the
phonons. The parameter ωc is the infrared phonon cutoff
longitudinal and transverse directions, one should consider the
frequency and can be determined from the quadratic phonon
net average electron velocity along the x-direction, can be
dispersion relation ωc = αqc2 where α(= 4.6 × 10−7 m2 s−1 )
written as  +π/2 is the ZA phonon diffusion constant and q is the ZA phonon
  −π/2 cos θ dθ 2 wave vector with qc = 1 nm−1 [19].
v x+ = v F = vF (2)
π π The use of the aforementioned definitions leads to the trans-
where v F is the Fermi velocity with a value of 106 m s−1 . mission coefficient at the Fermi level under strong diffusive
The isotropic energy-dependent electron mean free path back- regime as  (E F ) ≈ (λ2D /L), which leads to the diffusive
scattering length in such 2-D system can then be written as electrical resistance as
[27] h L
π R= (7)
λ2D (E F ) = v F τ (E F ) (3) 2e2 λ2D (E F ) M (E F )
2
in which λ2D (E F ) has been averaged over the entire length where e is the electron charge and the number of channels in
+x to −x direction of the sheet [see Fig. 1(b)]. The resultant 2-D SLG sheet is
energy-dependent scattering rate (1/τ (E F )) is assumed to be 4W E F
M (E F ) = (8)
quasi-elastic and is due to the in-plane and infrared flexural hv F
phonons [24], which can be written from Matthiessen’s rule
at the Fermi level. Given a Dirac-type electron energy dis-
as
1 1 1 persion relation for 2-D SLG sheet, the Fermi energy can be
= + (4) calculated from the relation
τ (E F ) τ (E F )|in−plane τ (E F )|flexural
√
where E F = h̄v F n 2D π (9)
  
1 g2 h̄ 2 υ F2 β 2 1 1 EF where n 2D is the sheet concentration. Using (7), Fig. 2 exhibits
≈ + + 2 kB T
τin−plane 2υ L2 4a 2 υL2 υT 2ρ h̄ 3 υ F2 the variation of R as function of the temperature for different
(5) sheet concentrations. We have kept the dimension of the
and sheet to be 1 × 1.5 μm2 to correlate our theoretical model
  with the experimental data from [25]. It appears that our
1 g2 h̄ 2 υ F2 β 2 (k B T )2 kB T analytical model fits well with increasing n 2D , which increases
≈ + ln . (6)
τflexural 2 4a 2 2
64π h̄ς E F h̄ωc the E F . This is an alternative approach to the experiment done

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 2066 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 60, NO. 6, JUNE 2013

with the back gate biasing where the gate voltage changes
the Fermi level position. At this point, it should be noted
that the electrical conductivity σ = (L/RW ) is generally
treated by considering a perfect 2-D layer with zero thickness.
It should be carefully noted that SLG is a 2-D material with
respect to electronic transport. The layer has a wall thickness
δ (= 0.34 nm). With respect to electronic transport, (8) is
2-D in nature. However, considering a finite wall-thickness,
the electrical conductivity of graphene is σ = (L/RW δ).
This is also true for realizing the SLG layer thermal resis-
tance, which is given as (L/κph W δ). Additionally, we would
like to state that the authors [28] have recently proposed a
theoretical model for the determination of R by considering
the electron and in-plane and electron and flexural phonons
scatterings. The formulation by the authors in [28] is carried
out under the presence of a longitudinal electric current from Fig. 3. Magnitude of Seebeck coefficient as a function of temperature for
varying n-type 2-D carrier concentration. The symbols are the experimental
the right to left contact [see Fig. 1(a)], which resulted in self- data from [26] with hole concentration of 2 and 2.8 × 1016 m−2 and at 4
heating. However, the analysis as done in this paper reflects and 6 μm widths, respectively. However, because of symmetric band structure
the dependency of R on n 2D via the Landauer transmission about the Dirac point, S B takes the same value for the corresponding electron
concentration. Solid lines: our analytical model.
approach when there is zero applied electrical current. From
Fig. 2, it appears that at low temperature, R follows a linear
relation with T because of the large interaction with the in- T and can span a value from 25 to 110 μV K−1 between an
plane phonons. This behavior is, however, dominated by a n 2D sweep of 0.1–1 × 1017 m−2 at room temperature.
parabolic behavior at higher T due to the emission of infrared To estimate the dimensionless ZT, one should also need
phonons. Following the Mott-Jones relation [29], the Seebeck the temperature-dependent thermal conductivities. It is now
coefficient under high carrier concentration can be written as well known that the thermal conductivity of SLG is entirely
due to the phonons [13]. The contribution from the electrons
π 2 k 2B T 1 ∂G are being poor, which can be evaluated from Landauer for-
SB = (10)
3e G ∂E E=E F malism as κe = (16/27h 2 )(λ2D (E F )/L)(E 3F /v F T ) for an
average energy of (2/3)E F and gives a value of about κe =
in which G = (1/R) is the electrical conductance. Substituting 0.12 W m−1 K−1 at 300 K for n 2D = 1017 m−2 , L = 5 μm
the value of R, S B can be written as and W = 4 μm. Thus it becomes straightforward to derive
π 2 k 2B T  (E F ) M  (E F ) the phonon contribution to the thermal conductivity, which
SB = + (11) lies in the range of 600–7000 W m−1 K−1 with a phonon
3e  (E F ) M (E F )
mean free path of about 775 nm (for 2000 W m−1 K−1 ) at
where the primes denotes the differentiation with respect to room temperature [30]–[32]. Recently, Verma et al. [33] have
energy, ( (E F )/ (E F )) = (2λ2D (E F )/πv F )[(D/E 2F )−C], shown an analytical model of thermal conductivity of SLG
and (M  (E F )/M (E F )) = (1/E F ) in which the coeffi- sheet, which is based on the quadratic dispersion relation of the
cients C and D can be evaluated from (5) and (6) as flexural phonon wave vector. The authors used an anharmonic
(1/τ (E F )) = C E F + (D/E F ). Thus, it seems that in the second-order three-phonon Umklapp, the edge roughness, and
strong diffusive regime, S B becomes independent of the sheet isotopic scatterings to demonstrate the variation of the thermal
dimension. However, in the near ballistic regime,  (E) = conductivity with temperature. It should be noted that the
(λ2D (E)/(λ2D (E) + L)) and under such conditions, S B starts isotopic scatterings in the present case mainly arise because
depending on the sheet length. Fig. 3 exhibits the varia- of the addition of C13 isotopes on C12 , which makes the
tion of the magnitude of S B with respect to T at different sheet impure. Using these rates, it is exhibited by the authors
sheet concentration. The sheet dimensions were chosen to be that κph follows a T 1.5 and T −2 law at lower and higher
5 × 4 μm2 to compare with the experimental data from temperatures in the absence of C13 isotopes, respectively.
[26]. These data has been taken at 300 K under the hole In their presence, however, the κph sharply deviates from the
concentration of 2 and 2.8 × 1016 m−2 and at 4 and 6 μm T −2 law at higher T . In these analyses, we incorporate this
widths, respectively. Although the data seems to be different analytical model of κph which can be written as [33]
for different widths, Grosse et al. [26] agreed that for wider −1 −1 −1
κph = κph + κph (12)
devices, S B stands independent of W . It is evident that the low high

same S B would reflect for an n-type concentration as the SLG in which

band structure is symmetric about the Dirac-point. The error 1 3
Fk B LW 2 kB T 2
bars fitting of S B in [26] are about +/−20 μVK−1 or almost κphlow = 4.58 (13)
50% uncertainty. Thus, within these error bars the S B between 2δ α π h̄
3 2
the two devices are quite consistent with each other, as they 27 kB Mα θ4 1
κphhigh = (14)
should be. It appears from our model that S B increases with 16πδ h̄ |γ Z A |2 h̄ω B T2

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 VERMA et al.: THERMOELECTRIC PERFORMANCE OF A SINGLE-LAYER GRAPHENE SHEET 2067

Fig. 4. ZT as a function of T with varying carrier concentration for

isotopically impure sheet.
Fig. 5. ZT as a function of E F for both isotopically pure and impure SLG
sheet.
for isotopically pure and
kBα θ4 appears that ZT at 300 K can reach a value up to 0.62 at a
κ ph high = ln ζ (15)
πδS0 m T2 sheet concentration of 1016 m−2 . We further see that, with
for isotopically impure SLG sheet. In the aforementioned the decrease in sheet length at very high carrier concentration
equation, the effect of edge-roughness, three-phonon Umklapp and isotope scattering strength, ZT decreases. This behavior is
and the isotope scattering rates are also same at higher concentration for isotopically pure case.
√ At low electron concentration, the dimensional dependence of
1 π ∂ωq ZT reduces at large sheet lengths.
= √ (16)
τE 2F LW ∂q Using the definition of S B for an n-type SLG sheet, the
 2
1 32 kB T negative open-circuit TE voltage under the presence of a
= |γ Z A | 4
 2 ω B (17) temperature difference TH − TL can be written as
τU 27 M ∂ωq /∂q
1 1  −1 TH
= S0 m qωq2 ∂ωq /∂ (18) Voc = − S B (T ) d T.
τI 4 (19)
respectively. The parameters F, M, |γZA |, θ , ω B , S0 , m , ζ , TL
are the geometric factor, mass of the carbon atom, Gruneisen This can be evaluated as
parameter for the ZA mode (−1.24), Debye temperature  
(1000 K), ZA phonon branch frequency (28 GHz), cross- π 2 k 2B TH2 − TL2
Voc =− +φ (20)
section area per one atom (= δ × r0 ) (where r0 = 0.14 nm 3q 2E F
is the C-C distance), strength of the impurity scattering and
(27S0 m α/16ω B )(k B M/h̄ 2 |γZA |2 )2 , respectively, and is fur- where the second term φ can be numerically obtained from
ther discussed in detail in our previous work [33]. It should  xH
1 xlnx − z
be noted that (14) has been realized through a simplified unity φ= 2 xdx (21)
model of the Dirac integrand [33] rather than extensive analy- p E F xL xlnx + z
sis of the phonon cutoff frequency [32]. Using this, Figs. 4 in which z = (bp/a)E 2F , p = (k B /h̄ωc ), C = bT , D =
and 5 exhibit the variation of ZT with T and E F , respectively, aT 2 ln ( pT ) and x = pT . Fig. 6 exhibits the variation of
for different sheet concentration and isotope density. We see the magnitude of Voc as function of TH for different sheet
that a substantial increment in ZT occurs at 300 K even for concentration. It can be seen from the figure that with the
n 2D = 1016 m−2 at L = 10 μm and W = 5 μm for an isotope increase of Fermi energy, Voc falls down at a very high rate
scattering strength of 2 × 10−2 . At this high impurity strength, due to the inverse dependency of S B on E F . Further, it appears
κph ≈ 44 W m−1 K−1 (the effect of isotope addition on κph that within a difference of 100 K, Voc is nearly −14 mV for
of SLG has been further experimentally exhibited in [34] in n 2D = 1016 m−2 , while the length effects on Voc is negligible
which our theoretical model [33] agrees well). However, it is in the entire diffusive range.
seen from Fig. 5 that with the decrease in L, ZT falls, which is The TE efficiency has been exhibited in Fig. 7 as function
due to the inverse dependency of the transmission coefficient of carrier concentration for both isotopically pure and impure
on Seebeck coefficient. Although, increase in length increases SLG sheet. It is evident from the equation
the thermal conductivity, however, for strongly diffusive ranges √
(1 μm and above), the variation of the thermal conductivity TL 1 + ZTav /2 − 1
η (%) = 1 − √ (22)
becomes almost length independent. From the same figure, it TH 1 + ZTav /2 + TL /TH

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 2068 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 60, NO. 6, JUNE 2013

Fig. 6. TE open circuit voltage, Voc as a function of TH with varying carrier

concentration.
Fig. 8. Short-circuit current as function of temperature at contact 1 for
different lengths and widths.

Fig. 7. TE efficiency as a function of n 2D for both isotopically pure and

impure SLG sheet.
Fig. 9. Output power as function of temperature at contact 1 for different
lengths and widths.

that higher the ZT, higher the η. It appears from Fig. 7 that at a
sheet concentration of 1016 m−2 , the efficiency reaches around Figs. 9 and 10. The rate of variation of power with TH is sim-
3.2% when the other SLG end is at 500 K. These ZT and η ilar to that of Fig. 8 with larger power output for larger widths
values can be compared with the MD and NEGF solutions and smaller lengths in the diffusive regime. We see that almost
for nanoribbons of graphene [15]–[18], which is considered 0.7 μW output power can be achieved from 1×5 μm2 at 100 K
without incorporating these necessary electron and phonon temperature difference. As seen from Fig. 10, however, there is
scatterings. The short-circuit current Isc as function of TH no significant variation of the power over carrier concentration
and sheet concentration is shown in Fig. 8. It should be noted at higher lengths. The variation trend is certainly not similar
that the main assumption we considered in this analysis is the to what is observed for other materials where the power factor
negligible resistance because of the two contacts at the ends. σ S 2B exhibits a maxima. This is a clear advantage of using SLG
However, this can be achieved experimentally through careful sheet over the TE conventional nanomaterials. At this point,
selection of contact materials. It can be seen that with the it is worth mentioning that the ballistic values of R, S B , and
increase in the length, Isc decreases. This is due to the increase κph are (h/2e2 M(E F )), (π 2 k 2B T /3eE F ), and (4π 2 k 2B T /3h)
in the SLG resistance with length. The open circuit voltage in respectively, which gives ZT = (M(E F )/18)(k B T /E F )2 and
this case remains almost dimensional independent as shown in |Voc | = (π 2 k 2B /6eE F )(TH2 − TL2 ). it should be noted that
Fig. 6. We see that with the increase in the width, Isc increases obtaining a ZT of the order of 1 or even greater for T = 450 K
up to a value near 200 μA for a temperature difference of is an extremely challenging task [3], [4]. In this case, however,
100 K even at an extreme concentration of 1017 m−2 . One can SLG exhibits a dramatically superior ZT, Voc , and η values
also notice that Isc also exhibits an increased variation with even in the presence of electron and in-plane and ZA phonon
n 2D . The power delivered for similar condition is shown in scatterings because of the very high σ , S B , and a controllable

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 VERMA et al.: THERMOELECTRIC PERFORMANCE OF A SINGLE-LAYER GRAPHENE SHEET 2069

[3] R. Venkatasubramanian, E. Siivola, T. Colpitts, and B. O’Quinn, “Thin-

film thermoelectric devices with high room-temperature figures of
merit,” Nature, vol. 413, pp. 597–602, Oct. 2001.
[4] G. J. Snyder and E. S. Toberer, “Complex thermoelectric materials,”
Nature Mater., vol. 7, no. 2, pp. 105–114, 2008.
[5] T. M. Tritt, “Recent trends in thermoelectric materials research III,”
in Semiconductors and Semimetals, vol. 71, R. K. Willardson and
E. R. Weber, Eds. San Francisco, CA, USA: Academic Press, 2001.
[6] H. Y. Chen, X. B. Zhao, Y. F. Lu, E. Mueller, and A. Mrotzek,
“Microstructures and thermoelectric properties of Fe0.92 Mn0.08 Six
alloys prepared by rapid solidification and hot pressing,” J. Appl. Phys.,
vol. 94, no. 10, pp. 6621–6626, 2003.
[7] T. C. Harman, P. J. Taylor, M. P. Walsh, and B. E. LaForge, “Quantum
dot superlattice thermoelectric materials and devices,” Science, vol. 297,
pp. 2229–2232, Sep. 2002.
[8] L. D. Hicks, T. C. Harman, X. Sun, and M. S. Dresselhaus, “Experimen-
tal study of the effect of quantum-well structures on the thermoelectric
figure of merit,” Phys. Rev. B, vol. 53, no. 16, pp. R10493–R10496,
1996.
[9] A. I. Hochbaum, R. Chen, R. D. Delgado, W. Liang, E. C. Gar-
nett, M. Najarian, A. Majumdar, and P. Yang, “Enhanced thermo-
Fig. 10. Output power as function of sheet concentration for different lengths electric performance of rough silicon nanowires,” Nature, vol. 451,
when contact 1 is at 400 K. pp. 163–167, Jan. 2008.
[10] B. Xu, C. Li, K. Thielemans, M. Myronov, and K. Fobelets, “Ther-
moelectric performance of Si0.8 Ge0.2 nanowire arrays,” IEEE Trans.
κph through addition of isotopes. Finally, we understand that Electron Devices, vol. 59, no. 12, pp. 3193–3198, Dec. 2012.
the SLG thermal conductivity can still be lowered down [11] Y. Li, K. Buddharaju, N. Singh, G. Q. Lo, and S. J. Lee, “Chip-level
thermoelectric power generators based on high-density silicon nanowire
by exploiting the edge scattering of phonons by casting the array prepared with top-down CMOS technology,” IEEE Electron Device
SLG into its nanoribbon [13]. This might see a promising Lett., vol. 32, no. 5, pp. 674–676, May 2011.
result, however, the electronic mobility can also deteriorate [12] Y. Li, K. Buddharaju, B. C. Tinh, N. Singh, and S. J. Lee,
“Improved vertical silicon nanowire based thermoelectric power gener-
due to increased scattering from the edges. Additionally to ator with polyimide filling,” IEEE Electron Device Lett., vol. 33, no. 5,
keep the mathematics simple, we have assumed the SLG pp. 715–717, May 2012.
surface to be ultra-flat, which reduces the electron-hole puddle [13] A. A. Balandin, “Thermal properties of graphene and nanostructured
densities [35]. The results of our theoretical model can be put carbon materials,” Nature Mater., vol. 10, pp. 569–581, Jul. 2011.
[14] S. V. Morozov, K. S. Novoselov, M. I. Katsnelson, F. Schedin,
forward to estimate the TE performance of graphene-based D. C. Elias, J. A. Jaszczak, and A. K. Geim, “Giant intrinsic carrier
TE arrays for recovering waste-heat from a dissipating active mobilities in graphene and its bilayer,” Phys. Rev. Lett., vol. 100, no. 1,
integrated circuits. pp. 016602-1–016602-4, 2008.
[15] X. Ni, G. Liang, J.-S. Wang, and B. Li, “Disorder enhances thermo-
electric figure of merit in armchair graphane nanoribbons,” Appl. Phys.
III. C ONCLUSION Lett., vol. 95, no. 19, pp. 192114-1–192114-3, 2009.
[16] H. Zheng, H. J. Liu, X. J. Tan, H. Y. Lv, L. Pan, J. Shi, and X. F. Tang,
In this paper, we theoretically demonstrated the TE effi- “Enhanced thermoelectric performance of graphene nanoribbons,” Appl.
ciency, ZT, and open circuit output voltage through the in- Phys. Lett., vol. 100, no. 9, pp. 093104-1–093104-5, 2012.
plane and ZA phonon scatterings with electrons in an SLG [17] H. Karamitaheri, N. Neophytou, M. Pourfath, R. Faez, and H. Kosina,
“Engineering enhanced thermoelectric properties in zigzag graphene
sheet. We presented an analytical result for evaluating the elec- nanoribbons,” J. Appl. Phys, vol. 111, no. 5, pp. 054501-1–054501-9,
trical resistance and Seebeck coefficient in SLG sheet for the 2012.
analytic estimation of ZT and η. Thus, it is shown that ZT can [18] G. S. Kliros and P. C. Divari, “Thermoelectric performance of electrons
reach a value 0.17–1.02 for isotope-doped SLG sheet under a in ballistic graphene ribbons,” in Proc. IEEE Int. Semicond. Conf., vol. 1.
Oct. 2011, pp. 45–48.
sheet concentration range of 1016–1017 m−2 and temperature [19] E. V. Castro, H. Ochoa, M. I. Katsnelson, R. V. Gorbachev, D. C. Elias,
range of 300–450 K. We found that our analytical models are K. S. Novoselov, A. K. Geim, and F. Guinea, “Limits on charge carrier
in good agreement with the available experimental data. mobility in suspended graphene due to flexural phonons,” Phys. Rev.
Lett., vol. 105, no. 26, pp. 266601-1–266601-4, 2010.
[20] M. T. Pettes, I. Jo, Z. Yao, and L. Shi, “Influence of polymeric residue
ACKNOWLEDGMENT on the thermal conductivity of suspended bi-layer graphene,” Nano Lett.,
vol. 11, no. 3, pp. 1195–1200, 2011.
The authors would like to thank Prof. E. Pop from the [21] E. Mariani and F. von Oppen, “Flexural phonons in free-standing
Department of Electrical and Computer Engineering, Univer- graphene,” Phys. Rev. Lett., vol. 100, no. 7, pp. 113401-1–113401-4,
sity of Illinois at Urbana-Champaign, Urbana, IL, USA, for 2008.
[22] L. Lindsay, D. A. Broido, and N. Mingo, “Flexural phonons and
his contribution to valuable discussions. thermal transport in graphene,” Phys. Rev. B, vol. 82, no. 11,
pp. 115427-1–115427-6, 2010.
R EFERENCES [23] L. Lindsay, D. A. Broido, and N. Mingo, “Flexural phonons and thermal
transport in multilayer graphene and graphite,” Phys. Rev. B, vol. 83,
[1] L. D. Hicks and M. S. Dresselhaus, “Effect of quantum-well structures no. 23, pp. 235428-1–235428-5, 2011.
on the thermoelectric figure of merit,” Phys. Rev. B, vol. 47, no. 19, [24] H. Ochoa, E. V. Castro, M. I. Katsnelson, and F. Guinea, “Scattering by
pp. 12727–12731, 1993. flexural phonons in suspended graphene under back-gate induced strain,”
[2] G. D. Mahan and S. F. Sofo, “The best thermoelectric,” Proc. Nat. Acad. Phys. E, Low-Dimensional Syst. Nanostruct., vol. 44, no. 6, pp. 963–966,
Sci. United State Amer., vol. 93, no. 15, pp. 7436–7439, 1996. 2012.

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.
 2070 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 60, NO. 6, JUNE 2013

[25] K. I. Bolotin, K. J. Sikes, J. Hone, H. L. Stormer, and P. Kim, Rekha Verma (S’12) is currently pursuing the Ph.D.
“Temperature-dependent transport in suspended graphene,” Phys. Rev. degree in electrothermal transport in 1-D and 2-D
Lett., vol. 101, no. 9, pp. 096802-1–096802-4, 2008. layered materials for applications in interconnects
[26] K. L. Grosse, M.-H. Bae, F. Lian, E. Pop, and W. P. King, “Nanoscale and thermoelectric devices at the Department of
Joule heating, Peltier cooling and current crowding at graphene-metal Electronic Systems Engineering, Indian Institute of
contacts,” Nature Nanotechnol., vol. 6, pp. 287–290, Apr. 2011. Science, Bangalore, India.
[27] C. Jeong, R. Kim, M. Luisier, S. Datta, and M. Lundstrom, “On Lan-
dauer versus Boltzmann and full band versus effective mass evaluation
of thermoelectric transport coefficients,” J. Appl. Phys., vol. 107, no. 2,
pp. 023707-1–023707-7, 2010.
[28] R. Verma, S. Bhattacharya, and S. Mahapatra, “Physics-based solution
for electrical resistance of graphene under self-heating effect,” IEEE
Trans. Electron Devices, vol. 60, no. 1, pp. 502–505, Jan. 2013.
[29] N. F. Mott and H. Jones, The Theory of the Properties of Metals and
Sitangshu Bhattacharya (M’13) received the M.Sc.
Alloys. London, U.K.: Oxford Univ. Press, 1936.
degree from the University of Calcutta, Kolkata,
[30] D. L. Nika, E. P. Pokatilov, and A. A. Balandin, “Theoretical description India, in 2003, and the Ph.D. degree from Jadavpur
of thermal transport in graphene: The issues of phonon cut-off frequen- University, Jadavpur, India, in 2009.
cies and polarization branches,” Phys. Status Sol. B, vol. 248, no. 11, He is currently a DST Inspire Faculty Member and
pp. 2609–2614, 2011. Assistant Professor at Shiv Nadar University, Dadri,
[31] D. L. Nika and A. A. Balandin, “Two-dimensional phonon trans- India.
port in graphene,” J. Phys., Condens. Matter, vol. 24, no. 23,
pp. 233203-1–233203-18, 2012.
[32] D. L. Nika, S. Ghosh, E. P. Pokatilov, and A. A. Balandin, “Lattice ther-
mal conductivity of graphene flakes: Comparison with bulk graphite,”
Appl. Phys. Lett., vol. 94, no. 20, pp. 203103-1–203103-3, 2009.
[33] R. Verma, S. Bhattacharya, and S. Mahapatra, “A physics-based flex-
ural phonon dependent thermal conductivity model for single layer
graphene,” Semicond. Sci. Technol., vol. 28, no. 1, pp. 01509-1–01509-6, Santanu Mahapatra (M’08–SM’10) received the
2013. Ph.D. degree in electrical engineering from the
[34] S. Chen, Q. Wu, C. Mishra, J. Kang, H. Zhang, K. Cho, W. Cai, École Polytechnique Fédérale de Lausanne, Lau-
A. A. Balandin, and R. S. Ruoff, “Thermal conductivity of isotopically sanne, Switzerland, in 2005.
modified graphene,” Nature Mater., vol. 11, pp. 203–207, Jan. 2012. He is currently an Associate Professor with the
Indian Institute of Science, Bangalore, India.
[35] J. Xue, J. S. Yamagishi, D. Bulmash, P. Jacquod, A. Deshpande,
K. Watanabe, T. Taniguchi, P. J. Herrero, and B. J. LeRoy, “Scan-
ning tunnelling microscopy and spectroscopy of ultra-flat graphene
on hexagonal boron nitride,” Nature Mater., vol. 10, pp. 282–285,
Feb. 2011.

Authorized licensed use limited to: INDIAN INSTITUTE OF INFORMATION TECHNOLOGY. Downloaded on September 21,2023 at 10:40:10 UTC from IEEE Xplore. Restrictions apply.