Measurements and microscopic model of quantum capacitance in

graphene
Huilong Xu, Zhiyong Zhang, and Lian-Mao Peng

Citation: Appl. Phys. Lett. 98, 133122 (2011); doi: 10.1063/1.3574011

View online: http://dx.doi.org/10.1063/1.3574011
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v98/i13
Published by the American Institute of Physics.

Related Articles
Ab initio study of the interactions between boron and nitrogen dopants in graphene
J. Appl. Phys. 112, 034304 (2012)
Band gaps and structural properties of graphene halides and their derivates: A hybrid functional study with
localized orbital basis sets
J. Chem. Phys. 137, 034709 (2012)
Abnormal pseudospin-degenerate states in a graphene quantum dot with double vacancy defects
J. Appl. Phys. 112, 014308 (2012)
Electronic band gaps and transport in aperiodic graphene superlattices of Thue-Morse sequence
Appl. Phys. Lett. 100, 252402 (2012)
Charge transfer equilibria in ambient-exposed epitaxial graphene on (000) 6H-SiC
J. Appl. Phys. 111, 113706 (2012)

Additional information on Appl. Phys. Lett.

Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors

Downloaded 08 Aug 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
 APPLIED PHYSICS LETTERS 98, 133122 共2011兲

Measurements and microscopic model of quantum capacitance

in graphene
Huilong Xu, Zhiyong Zhang,a兲 and Lian-Mao Penga兲
Department of Electronics and Key Laboratory for the Physics and Chemistry of Nanodevices,
Peking University, Beijing 100871, People’s Republic of China
共Received 25 January 2011; accepted 12 March 2011; published online 1 April 2011兲
Metal-oxide-semiconductor 共MOS兲 structures based on graphene were fabricated with ultrathin
Y2O3 films as the top gate oxide. While the quantum capacitance of graphene was measured using
the MOS structure and shown to agree well with theory for ideal graphene at large channel potential,
it deviates significantly from theory near the Dirac point. A general microscopic capacitance model
is developed and used to describe the quantum capacitance anomaly near the Dirac point. Excellent
agreement with experiment results was achieved using this model and key parameters including
potential fluctuation and local carrier density fluctuation were retrieved. © 2011 American Institute
of Physics. 关doi:10.1063/1.3574011兴

Quantum capacitance measurement is not only a direct Cg = CQCox / 共CQ + Cox兲. This Cg was measured under various
method to explore the density of states of graphene1–8 but gate voltages in vacuum at room temperature using Agilent
also an efficient way to explore various anomaly occurring B1500A semiconductor analyzer with a high frequency of 50
near the Dirac point which is difficult to probe only by trans- KHz. The gate leakage current was measured right before the
port measurements.9–14 To accurately measure quantum ca- CV measurement using Keithley 4200 semiconductor ana-
pacitance of graphene, metal-oxide-semiconductor 共MOS兲 lyzer, and found to be smaller than 40 pA 共corresponding to
structures with ultralarge oxide capacitance comparable with 1.26 pA/ ␮m2兲 under the gate voltage of 0.4 V as shown in
quantum capacitance should be fabricated on graphene. But the inset of Fig. 1共c兲. The small leakage current implies little
it remains a challenge to realize experimentally.4–7 Several contribution of the gate conductance to the impedance of the
groups have already observed and retrieved quantum capaci- device during the CV measurement and thus ensures the re-
tance in graphene,4,7,8 and some measurement results5,6 have liability of the capacitance measurement. The gate voltage
shown good agreement with theory at large bias far from the dependent total gate capacitance is shown in Fig. 1共c兲. Using
Dirac point. However, deviation from theory near the Dirac
point has been found in all measurements.4–8 This deviation
was generally attributed to electron-hole puddles induced by
charged impurities5,6,9–11 but a microscopic capacitance
model designed to quantitatively describe the quantum ca-
pacitance near the Dirac point is absent.5,6 In this letter, uti-
lizing ultrathin Y2O3 gate dielectric, the capacitance of
graphene MOS device is measured, and the quantum capaci-
tance of graphene is extracted with high accuracy. To de-
scribe the behaviors near the Dirac point, we developed a
general microscopic capacitance model based on the
electron-hole puddle model using a single parameter, i.e.,
potential fluctuation. The experimental results are found to
agree quantitatively at all bias with the single parameter
model.
A MOS capacitor consisting of a graphene-Y2O3 – Ti/ Au
sandwich was used to measure the quantum capacitance of
graphene as shown in Fig. 1共a兲.2,15 The optical picture of the
as fabricated device is shown in Fig. 1共b兲. The device fabri-
cation process can be found in Ref. 16, with additional form-
ing gas 共Ar: H2 = 5 : 1兲 annealing at 300 ° C for 1 h in furnace
so as to improve the quality of Y2O3. High resistive silicon FIG. 1. 共Color online兲 共a兲 Diagram showing the geometry of a graphene
substrates 共18 k⍀ cm兲 were used here to reduce parasitic MOS structure, where a graphene-Y2O3 – Ti/ Au capacitor lies on a SiO2 / Si
substrate. 共b兲 Optical image showing an as fabricated device and experimen-
capacitance. tal set up for measurements. The graphene channel is in the dashed frame
The total gate capacitance Cg may be considered as re- indicated by the black arrow. The channel area covered by top gate electrode
sulted from the top gate oxide capacitance Cox and the quan- is W / L = 4.5 ␮m / 7 ␮m. The separation between source/drain electrodes to
tum capacitance CQ of graphene connected in series, i.e., the top gate electrode is 1.2 ␮m. 共c兲 Total gate capacitance of the as fabri-
cated graphene FET. Inset: gate leakage current density as a function of gate
voltage. 共d兲 Quantum capacitance of graphene extracted from 共c兲. The
a兲
Authors to whom correspondence should be addressed. Electronic ad- dashed line is the theoretical value of quantum capacitance of graphene at
dresses: [email protected] and [email protected]. 300K with a Fermi velocity 1.15⫻ 106 m / s.

0003-6951/2011/98共13兲/133122/3/$30.00 98, 133122-1 © 2011 American Institute of Physics

Downloaded 08 Aug 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
133122-2 Xu, Zhang, and Peng Appl. Phys. Lett. 98, 133122 共2011兲

plane.9–11 Similar to the nonzero minimum conductivity at

the Dirac point, the deviation of the measured quantum ca-
pacitance from ideal value near the Dirac point is also attrib-
uted to LDOS共Ef兲 fluctuation induced by fluctuating poten-
tial. To calculate the average quantum capacitance of a large
graphene with fluctuating local potential, we consider a mi-
croscopic capacitance model in which the macroscopic
graphene plane is divided into N small cells with the same
area. When the cells are small enough we can assume that
each cell posses a uniform potential V and a uniform micro-
scopic quantum capacitance Cq. The graphene plane can be
viewed as composed of many cells connected in parallel.
Therefore, the quantum capacitance per area of the ith cell
with a potential V共i兲 can be written as

Cq共i兲 =
2q2kT
␲共បvF兲
冋冉
2 ln 2 1 + cosh
qV共i兲
kT
冊册 , 共3兲

FIG. 2. 共Color online兲 Microscopic capacitance model and fits of the ex- with i being the cell index. If the total cell number N tends to
periment data. 共a兲 Illustration of the key assumption of the model: local infinite, the average quantum capacitance is then given by
voltage fluctuating in the graphene plane around a gate induced channel

冕
N
potential. 共b兲 Calculated quantum capacitance of graphene with different 1 +⬁
fluctuation levels ␦V which varies from 0 to 120 mV with step of 30 mV
from bottom to top. Inset: carrier density fluctuation ␦n as a function of ␦V2.
CQ共Vch兲 = 兺 Cq共i兲 =
N i=1 −⬁
Cq共V兲P共Vch兲dV, 共4兲
The solid line is the numerical result and the dots are the results of Eq. 共8兲.
Best fits of the experiment data with 共c兲 a single fitting parameter ␦V where P共Vch兲 is the potential distribution function of
= 110 mV for quantum capacitance and 共d兲 another device with small fluc- graphene with an average potential of Vch. To calculate the
tuation ␦V = 72 mV. average quantum capacitance using Eq. 共4兲, the potential dis-
tribution function P共Vch兲 must be known at first. Since the
a relative dielectric constant 10 and thickness 3.9 nm 共mea- potential distribution of the SiO2 substrate was measured and
sured by atomic force microscope兲, the Y2O3 oxide capaci- found to obey a Gaussian distribution,10 it is reasonable to
tance is calculated to be 2.28 ␮F / cm2, which is the largest assume here that the local potential fluctuations of the
solid state oxide capacitance measured on graphene. The graphene surrounded by oxide also obeys a Gaussian distri-
quantum capacitance of the graphene is further retrieved by bution. Therefore, we write
CQ共Vch兲 = CoxCg/共Cox − Cg兲
where Vch = Vg − 兰Vg
0 Cg / CoxdVg is the channel potential
共1兲
P共Vch, ␦V兲 =
1
冑2␲␦V 冋
exp −
共V − Vch兲2
2␦V2
, 册 共5兲
of graphene. The Vch dependent CQ is shown as the solid line
in Fig. 1共d兲. The symmetrical electron and hole branches where, Vch and ␦V are, respectively, the average potential
with respect to the Dirac point clearly reveal the band sym- and potential fluctuation of graphene. Substituting Eq. 共5兲
metry of graphene. For an ideal graphene with a uniform into Eq. 共4兲, we obtain the final expression for the quantum
channel potential,2 the capacitance is given by capacitance

CQ =
2q2kT
␲共បvF兲2
冋冉
ln 2 1 + cosh
qVch
kT
冊册 , 共2兲 CQ共Vch, ␦V兲 =
冑2q2kT
␲ ␦V共បvF兲
3/2 2 冕 冋冉
+⬁

−⬁
ln 2 1 + cosh
qVch
kT
冊册
where ប, q, k, and T are the Planck constant, electron charge,
Boltzmann constant, and temperature, respectively, and vF is
the Fermi velocity of the Dirac electrons. The dashed line in
⫻exp − 冋 共V − Vch兲2
2␦V2
dV. 册 共6兲

Fig. 1共d兲 is the best fits using Eq. 共2兲 and a Fermi velocity For a graphene with an average potential Vch, the quantum
1.15⫻ 106 m / s for graphene at 300 K. This fitted value of capacitance is determined entirely by the variance of the
Fermi velocity agrees excellently well with that resulted channel potential ␦V. Therefore, a microscopic quantum ca-
from previous experiments6,12,17 and theories.1,18,19 pacitance model, explicitly including local potential fluctua-
It must be noted that the retrieved quantum capacitance tions, is obtained. Compared to the simple model of Ref. 5,
only agrees well with theory at large channel potential and in which the quantum capacitance is simply fitted by a car-
significant deviation occurs near the Dirac point. This devia- rier concentration n dependent CQ ⬀ 冑n, the microscopic
tion is due to the fact that Eq. 共2兲 is derived for a perfectly quantum capacitance model developed here is based on the
uniform and clean graphene sample with a uniform potential. more basic physical parameters such as potential fluctuation
In a real graphene on a substrate, the local potential fluctu- which may be directly related to the electron-hole puddles in
ates spatially as shown in Fig. 2共a兲 due to charged impurities graphene. It is obvious that Eq. 共6兲 is not easy to solve ana-
above and below the layer. Electron and hole puddles near lytically, but which can be analyzed numerically.
the Dirac point were predicted in theory and observed in The numerical simulation of the quantum capacitance
experiment as a result of potential fluctuations in graphene under different potential fluctuation levels is shown in Fig.
and the local density of states at a given Fermi energy 2共b兲. It can be seen that as potential fluctuation ␦V increases,
关LDOS共Ef兲兴 fluctuates in the two dimensional graphene the minimum quantum capacitance increases and the region

Downloaded 08 Aug 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
133122-3 Xu, Zhang, and Peng Appl. Phys. Lett. 98, 133122 共2011兲

showing significant deviation from Eq. 共2兲 broadens. There- of 72 mV near the Dirac point is very close to deviation of
fore, the deviation of measured quantum capacitance from potential below threshold in silicon MOS,20 and both 1.2
the ideal value may be attributed to result from potential ⫻ 1012 and 4.3⫻ 1011 / cm2 are typical values of carrier den-
fluctuations near the Dirac point. At large channel potential, sity fluctuation observed in transport measurement.13
the gate induced carrier density is much larger than the car- In conclusion, graphene based top-gated MOS devices
rier fluctuations and smears out the fluctuation effect, and were fabricated with ultrathin Y2O3 gate insulator films. The
then CQ remains unchanged under varying potential fluctua- gate capacitances were measured under varying gate voltage
tions. The CQ deviation from ideal value becomes significant and the channel potential dependent quantum capacitances
only near the Dirac point with 兩Vch兩 ⬍ 2␦V. were retrieved. The measured quantum capacitance agrees
In addition to potential fluctuation, charged impurity in- perfectly well with theory for ideal graphene at large channel
duced carrier density in graphene is another concerned pa- potential, but begins to deviate from the theory due to
rameter in electron-hole puddles model of graphene. The car- electron-hole puddles near the Dirac point. A general micro-
rier density n and its variance ␦n at the Dirac point can be scopic capacitance model based on electron-hole puddles is
calculated based on the microscopic capacitance model for developed to describe the quantum capacitance anomaly near
any given potential fluctuation. The relation between ␦n and the Dirac point. The experimental quantum capacitance re-
␦V2 is shown in Fig. 2共b兲, which indicates a linear depen- sults were well fitted by the microscopic capacitance model,
dence of ␦n on ␦V2 with a slope about 9.6⫻ 1017 m−2 V−2 in and the local channel potential fluctuation and carrier density
the density fluctuation range from 1 ⫻ 1011 cm−2 to 2 fluctuation were obtained.
⫻ 1012 cm−2. Noting that for qVch Ⰷ kT, Eq. 共2兲 can be sim-
plified to CQ = 共2q3 / ␲共បvF兲2兲兩Vch兩, or CQ = q2DOS共E兲. Thus, This work was supported by the Ministry of Science
we have and Technology of China 共Grant Nos. 2011CB933001 and
2011CB933002兲, the Fundamental Research Funds for the
q2
n= ⫾ V2 , 共7兲 Central Universities, and National Science Foundation of
␲共បvF兲2 ch China 共Grant No. 61071013兲.
where “+” stands for hole and “⫺” for electron accumula- 1
A. H. Castro Neto, F. Guinea, N. M. R. Peres, K. S. Novoselov, and A. K.
tion, respectively. As a result, the density fluctuation can
Geim, Rev. Mod. Phys. 81, 109 共2009兲.
be analytically related to the potential fluctuation at the 2

Dirac point as ␦n = 共冑3q2 / ␲共បvF兲2兲␦V2, that is ␦n = 9.6

T. Fang, A. Konar, H. Xing, and D. Jena, Appl. Phys. Lett. 91, 092109
共2007兲.
⫻ 1017 ␦V2 m−2 V−2, the slope of which is exactly what we
3
F. Schwierz, Nat. Nanotechnol. 5, 487 共2010兲.
4
obtained through numerical method as presented above. In Z. Chen and J. Appenzeller, Tech. Dig. - Int. Electron Devices Meet.
2008, 509.
addition, we find that in the density fluctuation region from 5
J. Xia, F. Chen, J. H. Li, and N. J. Tao, Nat. Nanotechnol. 4, 505 共2009兲.
1 ⫻ 1011 to 2 ⫻ 1012 cm−2, the numerical result can be fitted 6
L. A. Ponomarenko, R. Yang, R. V. Gorbachev, P. Blake, A. S. Mayorov,
by K. S. Novoselov, M. I. Katsnelson, and A. K. Geim, Phys. Rev. Lett. 105,
冑3q2 7
136801 共2010兲.
F. Giannazzo, S. Sonde, V. Raineri, and E. Rimini, Nano Lett. 9, 23
␦n = ␦V2 + nT , 共8兲 共2009兲.
␲共បvF兲2 8
S. Dröscher, P. Roulleau, F. Molitor, P. Studerus, C. Stampfer, K. Ensslin,
where nT = 5.8⫻ 1010 cm−2 is the room temperature density and T. Ihn, Appl. Phys. Lett. 96, 152104 共2010兲.
9
of electrons or holes for an ideal graphene. The result is E. H. Hwang, S. Adam, and S. Das Sarma, Phys. Rev. Lett. 98, 186806
共2007兲.
shown in the inset of Fig. 2共b兲. 10
J. Martin, N. Akerman, G. Ulbricht, T. Lohmann, J. H. Smet, K. Von
Through the microscopic capacitance model, the mea- Klitzing, and A. Yacoby, Nat. Phys. 4, 144 共2008兲.
sured quantum capacitance in Fig. 1共d兲 of graphene can be 11
Y. Zhang, V. W. Brar, C. Girit, A. Zettl, and M. F. Crommie, Nat. Phys. 5,
perfectly fitted with a single fitting parameter ␦V = 110 mV 722 共2009兲.
12
as shown in Fig. 2共c兲. Since potential fluctuation on a bare K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson,
SiO2 substrate is about 50 mV,10 ␦V = 110 mV is a reason- I. V. Grigorieva, S. V. Dubonos, and A. A. Firsov, Nature 共London兲 438,
197 共2005兲.
able potential fluctuation amplitude for a graphene embedded 13
S. Adam, E. H. Hwang, V. M. Galitski, and S. Das Sarma, Proc. Natl.
between SiO2 and Y2O3. The voltage dependent gate capaci- Acad. Sci. U.S.A. 104, 18392 共2007兲.
14
tances can also be fitted using this model as shown in Fig. Y.-W. Tan, Y. Zhang, K. Bolotin, Y. Zhao, S. Adam, E. H. Hwang, S. Das
2共c兲. The carrier density fluctuation at the Dirac point is cal- Sarma, H. L. Stormer, and P. Kim, Phys. Rev. Lett. 99, 246803 共2007兲.
15
culated to be 1.2⫻ 1012 cm−2, indicating a large amount of S. Luryi, Appl. Phys. Lett. 52, 501 共1988兲.
16
Z. X. Wang, H. L. Xu, Z. Y. Zhang, S. Wang, L. Ding, Q. S. Zeng, L. J.
charged impurities in the surroundings, which in turn Yang, T. Pei, X. L. Liang, M. Gao, and L. M. Peng, Nano Lett. 10, 2024
induces a very large minimum quantum capacitance of 共2010兲.
about 1.68 ␮F / cm2 at the Dirac point. For another 17
Y. B. Zhang, Y. W. Tan, H. L. Stormer, and P. Kim, Nature 共London兲 438,
metal-Y2O3-graphene MOS device, the retrieved minimum 201 共2005兲.
18
quantum capacitance is only about 1.21 ␮F / cm2, and the P. E. Trevisanutto, C. Giorgetti, L. Reining, M. Ladisa, and V. Olevano,
Phys. Rev. Lett. 101, 226405 共2008兲.
measured CQ − Vch and Cg − Vg curves 关Fig. 2共d兲兴 can be fit- 19
L. Yang, J. Deslippe, C. H. Park, M. L. Cohen, and S. G. Louie, Phys. Rev.
ted very well using the microscopic capacitance model with Lett. 103, 186802 共2009兲.
a smaller potential fluctuation ␦V = 72 mV and carrier den- 20
G. Slavcheva, J. H. Davies, A. R. Brown, and A. Asenov, J. Appl. Phys.
sity fluctuation ␦n of about 4.3⫻ 1011 / cm2. The retrieved ␦V 91, 4326 共2002兲.

Downloaded 08 Aug 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions