See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303872327

Dielectric relaxation and hopping conduction in reduced graphite oxide

Article in Journal of Applied Physics · June 2016

DOI: 10.1063/1.4953357

CITATIONS READS

37 5,936

4 authors, including:

Ji Yu Min Gu
Nanjing University Independent Researcher
13 PUBLICATIONS 164 CITATIONS 109 PUBLICATIONS 1,299 CITATIONS

SEE PROFILE SEE PROFILE

Tong B. Tang
Asian Power Development (Group) Ltd
179 PUBLICATIONS 2,243 CITATIONS

SEE PROFILE

All content following this page was uploaded by Tong B. Tang on 25 April 2018.

The user has requested enhancement of the downloaded file.

 JOURNAL OF APPLIED PHYSICS 119, 224102 (2016)

Dielectric relaxation and hopping conduction in reduced graphite oxide

Guidan Wei,1 Ji Yu,1 Min Gu,1,a) and Tong B. Tang2,b)
1
National Laboratory of Solid State Microstructures and Department of Physics, Collaborative Innovation
Center of Advanced Microstructures, Nanjing University, Nanjing 210093, People’s Republic of China
2
Department of Physics, H.K. Baptist University, Kowloon, Hong Kong, People’s Republic of China
(Received 21 March 2016; accepted 24 May 2016; published online 8 June 2016)
Graphite oxide reduced by sodium borohydride was characterised and its electrical conduction
investigated with impedance spectroscopy. Thermal dependence of electrical modulus (instead of
permittivity, its inverse) was calculated from complex impedance spectra, an approach that prevents
any peak in dielectric loss (imaginary component) from being swarmed by large dc conductivity.
Two loss peaks appeared at each tested frequency, in a sample of either degree of reduction. The set
of weaker peak should arise from the relaxation of some polar bonds, as proposed earlier by us. The
stronger loss peaks may correspond to the hopping of conduction electrons; variable range hopping is
also consistent with the observed thermal dependence of conductivity. However, nearer ambient
temperature there is a change in mechanism, to band transport, with an activation energy of fairly
similar values as derived from both loss peaks and conductivity. Published by AIP Publishing.
[http://dx.doi.org/10.1063/1.4953357]

I. INTRODUCTION II. EXPERIMENTAL AND DATA ANALYSIS

1
Graphite oxide (GO) was first obtained by Schafhaeutl A. Experimental details
and Brodie,2 who intercalated polar groups such as carbonyl,
Our GO samples were synthesized from graphite via a
hydroxyl, and epoxide into naturally occurring graphite
modified Hummers’ method, and reduced by using NaBH4
through treatment with an oxidising agent in the presence of
as a reducing agent, for details see supplementary material.36
a strong acid. Its designation as an oxide is historical, though
X-ray diffraction (XRD) was performed in a Dmax-rB
in fact it is a hydrocarbon. More than a century later, despite
(Rigaku) Diffractometer using Cu Ka1 radiation at 0.02"
some uncertainties in details of its highly disordered struc-
steps over 2h from 5" to 55" ; Raman spectroscopy, in a
ture, the model formulated by Lerf and Klinowski3–5 finds
LabRam HR800 Microscope (Horiba-Jobin Yvon), that uti-
common acceptance. In their non-stoichiometric model, GO
lised excitation at 633 nm wavelength. X-ray photoelectron
consists of aromatic “islands” of variable sizes representing
spectroscopy (XPS) measurements proceeded in a K-Alpha
remnants of its parent material, separated from one another
(Thermo Scientific) instrument generating Al Ka radiation.
by buckled, aliphatic 6-membered rings with various C¼O,
Dielectric spectra were recorded by an Agilent E4980A
C-OH, and C-O-C attached. A more recent 13C nuclear mag-
Precision LCR Meter, in three-terminal configuration, at a
netic resonance study has added refinements to this model,
working voltage of 1.0 V. The impedance analyser scanned
proving the existence also of O-C-OH.6
at 12 chosen frequencies from 111 Hz to 2 MHz, as the sam-
Renewed interest in GO derives partly from its potential
ple under study originated from above liquid nitrogen tem-
use in the production of graphene.7–9 Moreover, chemically
perature to 310 K at 1 K/min, sitting in a 30 Pa vacuum, with
reduced GO, being soluble in many organic solvents, has
its temperature monitored by a Cu-CuNi thermocouple accu-
potential in molecular electronics, such as transparent and
racy to 0.1 K. Silver and gold-plated copper served, respec-
flexible conductive films for displays and electrodes,10,11 con-
tively, as upper and lower electrodes. Samples were dried
ductive inks for printed electronics,12,13 molecular sensors,14
GO, or rGO, pressed under the same uni-axial pressure of
and energy storage devices.15,16 GO is electrically insulating,
5 MPa into the pellets of 8 mm in diameter and approxi-
due to the localisation of its p electrons as a result of sp3
mately 0.6 mm thick.
hybridisation between C and O atoms, but after reduction and
fractional loss of oxygen-containing groups, its conductivity
approaches that of a semiconductor.17 However, the detach- B. Impedance data analysis
ment of those functional groups gives rise to multitude struc- The dielectric response of a solid under some ac electric
tural defects, as just described. Its electronic applications call field of angular frequency x can be described by its permittiv-
for detailed understanding of its conduction mechanism and in ity. The imaginary part of this complex function represents
particular, the role of defects, hence the present work. dielectric loss, which, when due to charge re-distribution, has
a)
strong frequency as well as temperature dependences; further-
Author to whom correspondence should be addressed. Electronic mail: more, when the charge relaxation (or retardation) shows
mgu@nju.edu.cn. Tel.: þ86 25 83593508. Fax: þ86 25 83595535.
b)
Now at Asia Power Development (Group), 2502 Win Plaza, San Po Kong, Debye behaviour with a single time constant s, the loss has a
Hong Kong, People’s Republic of China. peak value at x ¼ 1/s.

0021-8979/2016/119(22)/224102/5/$30.00 119, 224102-1 Published by AIP Publishing.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 205.175.215.52 On: Wed, 08 Jun
2016 16:22:43
 224102-2 Wei et al. J. Appl. Phys. 119, 224102 (2016)

FIG. 2. Raman spectra of GO and rGO.

FIG. 1. XPS spectra of GO and rGO.
intensities among resolved components, we see that, as chemi-
An alternative property to e*(x), proposed by McCrum cal reduction progressed, sp2 carbons increased in concentra-
et al. in their polymeric study,18 is the dielectric or electrical tion whilst oxygen-containing functional groups became
modulus M*(x), simply defined as the reciprocal of complex scarcer.
permittivity. M*(x) relates to the time-domain function U(t) As indicated in Fig. 2, GO displays two vibrational
of electric field within the dielectric material via Fourier bands: D at 1348 cm'1, attributed to K-point phonons of
transform A1g symmetry, and G at 1593 cm'1, ascribed to zone-center
" # phonons of E2g symmetry.25–27 The former vibration modes
ð '1 " #
$ 'd/ðtÞ are active only along the edges of carbon sheets, where
M ðx Þ ¼ M 1 1 ' expð'ixtÞdt ; (1) translational symmetry breaking relaxes selection rules. A
0 dt
gain in the D to G intensity ratio, ID/IG, corresponds there-
which may also be measured directly.19 Representing experi- fore to greater discontinuity in graphene layers. Table I
mental data in the form of M*(x) instead of e*(x) is particu- presents quantitative deductions from both XPS and Raman
larly useful when the material exhibits significant dc spectroscopies.
conductivity within the temperature range of interest,19–21 as
we shall see in our case. B. Impedance spectroscopy
The thermal dependence of dielectric data is presented
III. RESULTS in Fig. 3. The imaginary part of permittivity, or dielectric
loss, at a low frequency is illustrated in Fig. 3(a): near room
A. Characterisation
temperature in rGO3 it exceeds by five orders of magnitude
X-ray diffractometry, Fig. S1, confirms that our GO than in pristine GO. After reduction of the oxide, e00 becomes
samples have been chemically reduced. dominated by dc conduction, especially at high temperature
Figure 1 shows representative XPS C1s spectra, where or low frequency.
the energy scale has been calibrated internally by setting the Figs. 3(b) and 3(c) depict both real and imaginary parts,
binding energy of C¼C to 284.3 eV.22 Deconvolution23,24 for respective rGO samples at certain frequencies. (Data for
helps to identify five types of carbon. From the relative other frequencies have been omitted, for clarity.) No loss

TABLE I. XPS intensity (with binding energy within brackets) and Raman bands intensity ratio (last column) in GO and rGO.

Sample sp2 % (eV) C-OH % (eV) C-O-C % (eV) O¼C-OH % (eV) O¼C-OH % (eV) ID/IG

GO 25 (284.3) 18 (285.9) 29 (286.6) 15 (288.1) 12 (289.7) 1.1

rGO1 50 (284.3) 17 (285.1) 24 (286.5) 8 (288.4) 1.25
rGO3 55 (284.3) 17 (285.1) 16 (286.4) 11 (288.4) 1.3

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 205.175.215.52 On: Wed, 08 Jun
2016 16:22:43
 224102-3 Wei et al. J. Appl. Phys. 119, 224102 (2016)

FIG. 4. Imaginary component of dielectric modulus M* (scale on the left)

and its real component (scale to the right) for (a) rGO1 and (b) rGO3. Insets:
enlarged plots of M00 , showing second peaks at lower temperatures.

FIG. 3. Permittivity e as function of temperature T: (a) its imaginary compo- out that every plot exhibits the highest linearity, as quantified
nent, for all three samples; (b) real and imaginary components, for rGO1; by standard error in the slope generated by least squares fit,
and (c) those for rGO3. if T'1/3 was chosen as the independent variable; but this lin-
ear range is broader when the measure-ment frequency was
peaks are discernible in either, apart from a shoulder as the lower, and also broader in rGO3 than in rGO1. To illustrate
results of a masking effect origin from dominating dc con- the more reduced case, Fig. 7 shows all three alternative
duction. Accordingly, we re-plot the same set of data in plots at one particular frequency, whilst Fig. S2 in the sup-
terms of electrical modulus instead, in Fig. 4. The loss modu- plementary material36 depicts the ln r ' T'1/3 plots at nine
lus does exhibit a sharp peak when plotted against log x. x (curves for 316, 626, and 1 MHz omitted for clarity).
This peak frequency readily gives a conductivity relaxation On the high-temperature side, ln r vs. T'1 are more lin-
time, while the corresponding characteristic time of the same ear (Fig. 8, where graphs for some x are not shown to avoid
data represented in terms of e*(x) would be difficult to deter-
mine.28 In each material at all frequencies, one strong and
one weak loss peak (see inset of Fig. 4) are now apparent;
indeed, each set of peak frequencies satisfies the Arrhenius
relation

f ¼ f0 expð'Ea =kTÞ; (2)

where the activation energy Ea assumes the respective value,

from linear fits (Fig. 5), of 0.16 and 0.19 eV for rGO1, and
0.10 and 0.11 eV for rGO3.
We have seen that dielectric loss rises drastically in GO
after chemical reduction, particularly at high temperature
and low frequency. Based on e00 data together with informa-
tion deduced from resolved components in XPS C1s peaks
(Table I), Fig. 6 illustrates a correlation between ac conduc-
tivity r ¼ e0xe00 (e0 denoting vacuum permittivity) and the
ratio of sp2 to sp3 carbon atoms.
Finally, we examine the thermal dependence of ac con- FIG. 5. Arrhenius plots of the relaxation peaks at (a) high and (b) low
ductivity by plotting ln r against T'1, T'1/2, or T'1/3. It turns temperatures.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 205.175.215.52 On: Wed, 08 Jun
2016 16:22:43
 224102-4 Wei et al. J. Appl. Phys. 119, 224102 (2016)

FIG. 8. Log conductivity (2p x ¼ 111 Hz) vs. T, (a) rGO1 and (b) rGO3.
The red points are the experimental data and the black lines are fit curves.

have little or no protonic and ionic conduction. The high

conductivity in our reduced GO with high fraction of sp2,
FIG. 6. Changes in conductivity with C bonds hybridisation. though yet incompletely graphitised, should arise from p
electrons within islands of graphene layers, which however
crowding). Interpreted as Arrhenius plots, the least squares suffer discontinuity due to clusters of sp3 carbons as well as
fits imply an activation energy of 0.2 eV for rGO1 and holes in the graphene sheets.
0.14 eV for rGO3. Two relaxation peaks have been observed in each
rGO. The weaker peak at low temperature in less reduced
IV. DISCUSSION rGO1 has the same activation of 0.16 eV as reported ear-
lier,32 and attributed there to dipoles formed from the
In GO, oxygen-containing functional groups are ran- breaking of a hydrogen bond (-O-H)))O<) between a
domly distributed on the two surfaces of individual graphene hydroxyl attached to a carbon grid and an epoxy group on
sheets,3 and they desorb during reduction, so that an rGO has the adjacent carbon grid. The other one in rGO3 has almost
less sp3 C but more sp2 C atoms, as Table I verifies. Its broad the same Ea as that in more reduced oxide, namely,
XRD (Fig. S1 in the supplementary material36) peak of d ( 0.08 eV,32 and probably arise from hydrogen bonds
3.4 Å suggests that the conversion to 2D coordination is not (-O-H)))p) between a hydroxyl group and the p-
random but cooperative, leading to nucleation followed by conjugation of the opposite carbon grid. The stronger
growth, and resulting in patches of sp2 atoms with large relaxation peaks at high temperature (corresponding to
enough areas to diffract. This explains the formation of low-frequency range) should then derive from the ther-
regions with residue functional groups29–31 surrounding mally assisted hopping of p electrons, the measured Ea of
islands of smooth hexagonal structure, as imaged by atomic 0.19 and 0.11 eV being the respective barrier height, in
force and scanning electron microscopies.31 Additionally, rGO1 and rGO3. This relaxation by hopping electrons was
the desorption of functional groups removes some C atoms not observed in the earlier study32 where dielectric spectra
of the sheets, leaving holes and producing more edge C were measured with the use of blocking electrodes.
atoms, which explains the rise in Raman D band in rGO, as The clusters of sp3 carbons exist in randomly distributed
shown in Table I. sizes and therefore the mechanism of this conduction by par-
Our samples were adequately anhydrous and no metallic tially localised electrons should be variable range hopping
impurities had been detected by XPS; therefore, they could (VRH).17,33,34 This hopping mechanism implies that

FIG. 7. Log conductivity (at 2p

x ¼ 111 Hz) vs. (a) T'1, (b) T'1/2, or
(c) T'1/3 for rGO3. The pink points are
the experimental data and the olive
lines are fit curves.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 205.175.215.52 On: Wed, 08 Jun
2016 16:22:43
 224102-5 Wei et al. J. Appl. Phys. 119, 224102 (2016)

" #
B 2012CB934000, and by the N.S.F. of China, through
r ¼ A exp ' 1=n ; (3) 10674060.
T
1
where n depends on details concerning the density of states 2
C. Schafhaeutl, J. Prakt. Chem. 21, 129 (1840).
for the conduction electrons, being either equal to 2 (as B. C. Brodie, Ann. Chim. Phys. 45, 351 (1855).
3
A. Lerf, H. He, M. Forster, and J. Klinowski, J. Phys. Chem. B 102, 4477
derived by Efros and Shklovskii), or is determined by the (1998).
dimensionality of the sample (Mott), such that n ¼ 3 in our 4
H. He, J. Klinowski, M. Forster, and A. Lerf, Chem. Phys. Lett. 287, 53
case. Both possibilities have been proposed for rGO,17,33,34 5
(1998).
H. He, T. Riedl, A. Lerf, and J. Klinowski, J. Phys. Chem. 100, 19954
whereas the present work favours the latter, but firmer con- (1996).
clusion awaits further experiments, particularly in the regime 6
W. Gao, L. B. Alemany, L. Ci, and P. M. Ajayan, Nat. Chem. 1, 403
below liquid nitrogen temperature. The common slope of the 7
(2009).
twelve conductivity plots in their low-temperature regions S. Pei and H.-M. Cheng, Carbon 50, 3210 (2012).
8
S. Pei, J. Zhao, J. Du, W. Ren, and H.-M. Cheng, Carbon 48, 4466 (2010).
(Fig. S2) is an expression dependent on the density of hop- 9
S. Park and R. S. Ruoff, Nat. Nanotechnol. 4, 217 (2009).
ping electrons and a parameter characterising electron 10
G. Eda, G. Fanchini, and M. Chhowalla, Nat. Nanotechnol. 3, 270 (2008).
11
localization.35 12
J.-H. Ahn and B. H. Hong, Nat. Nanotechnol. 9, 737 (2014).
E. B. Secor, B. Y. Ahn, T. Z. Gao, J. A. Lewis, and M. C. Hersam, Adv.
Thermal dependence of conductivity suggests that at a Mater. 27, 6683 (2015).
low temperature, variable range hopping occurs in reduced 13
B. W. An, K. Kim, M. Kim, S.-Y. Kim, S.-H. Hur, and J.-U. Park, Small
GO but with rising temperature, thermal excitation of free 11, 2263 (2015).
14
electrons may become the dominant rate-limiting step for J. T. Robinson, F. K. Perkins, E. S. Snow, Z. Wei, and P. E. Sheehan,
Nano Lett. 8, 3137 (2008).
conduction, as argued above. Arrhenius plots put its activa- 15
D. Yu, K. Gon, H. Wang, L. Wei, W. Jiang, Q. Zhang, L. Dai, and Y.
tion energy at 0.2 eV and 0.14 eV, respectively, lower in the Chen, Nat. Nanotechnol. 9, 555 (2014).
16
more reduced material. These values are moderately consist- J. Sun, H.-W. Lee, M. Pasta, H. Yuan, G. Zheng, Y. Sun, Y. Li, and Y.
Cui, Nat. Nanotechnol. 10, 980 (2015).
ent with those measured from impedance spectroscopy. 17
G. Eda, C. Mattevi, H. Yamaguchi, H. Kim, and M. Chhowalla, J. Phys.
Chem. C 113, 15768 (2009).
V. CONCLUSION 18
N. G. McCrum, B. E. Read, and G. Williams, Anelastic and Dielectric
Effects in Polymeric Solids (Dover, New York, 1991).
X-ray diffractometry confirms that our GO samples 19
R. Richert and H. Wagner, Solid State Ionics 105, 167 (1998).
20
have been chemically reduced and locally reverted to graph- A. Molak, M. Paluch, S. Pawlus, J. Klimontko, Z. Ujma, and I. Gruszka,
ite in structure, with substantial disordering. X-ray photo- J. Phys. D: Appl. Phys. 38, 1450 (2005).
21
I. M. Hodge, K. L. Ngai, and C. T. Moynihan, J. Non-Cryst. Solids 351,
electron spectroscopy reveals corresponding decreases in 104 (2005).
oxygen-containing functional groups and the fraction of sp3 22
D. W. Davis and D. A. Shirley, J. Electron Spectrosc. Relat. Phenom. 3,
carbons. Raman data show the existence of more edge car- 137 (1974).
23
A. Ganguly, S. Sharma, P. Papakonstantinou, and J. Hamilton, J. Phys.
bon atoms, indicative of more holes and islands on graphene Chem. C 115, 17009 (2011).
sheets. Above all, reduced GO exhibits hugely increased 24
C. K. Chua, A. Ambrosi, and M. Pumera, Analyst 138, 7012 (2013).
25
electrical conductivity. We analyze electrical conductivity F. Tuinstra and J. L. Koenig, J. Chem. Phys. 53, 1126 (1970).
26
mechanism and point to a change from variable range hop- K. N. Kudin, B. Ozbas, H. C. Schniepp, R. K. Prud’homme, I. A. Aksay,
and R. Car, Nano Lett. 8, 36 (2008).
ping over the energy barrier at island edges to band conduc- 27
A. C. Ferrari and J. Robertson, Phys. Rev. B 61, 14095 (2000).
tion by p electrons at higher temperature. 28
N. L. Ngai, Relaxation and Diffusion in Complex Systems (Springer,
Electron localization and hopping conduction may play 2011).
29
M. M. Lucchese, F. Stavale, E. H. Martins Ferreira, C. Vilani, M. V. O.
significant roles in electronic devices. However, a clear
Moutinho, R. B. Capaz, C. A. Achete, and A. Jorio, Carbon 48, 1592
understanding of electronic transport in rGO sheets is lack- (2010).
30
ing. Various studies report different conduction mechanisms X. Gao, J. Jang, and S. Nagase, J. Phys. Chem. C 114, 832 (2010).
31
such as Mott variable-range hopping (VRH)17,34 and Efros- K. Erickson, R. Erni, Z. Lee, N. Alem, W. Gannett, and A. Zettl, Adv.
Mater. 22, 4467 (2010).
Shklovskii VRH;33 our present work supports Mott VRH. 32
G. Wei, J. Yu, M. Gu, X. Ai, X. Xu, and T. B. Tang, Carbon 95, 374
Understanding of the electron transport properties of rGO is (2015).
33
of great significance to realize its promising electronic 34
D. Joung and S. I. Khondaker, Phys. Rev. B 86, 235423 (2012).
application. C. G!omez-Navarro, R. T. Weitz, A. M. Bittner, M. Scolari, A. Mews, M.
Burghard, and K. Kern, Nano Lett. 7, 3499 (2007).
35
N. F. Mott, Electronic Processes in Non-Crystalline Materials (Oxford
ACKNOWLEDGMENTS University Press, Oxford, UK, 1971).
36
See supplementary material at http://dx.doi.org/10.1063/1.4953357 for
This work was supported by the National Basic Research sample preparation, X-ray diffractometry, and Log conductivity vs. T'1/3
Programme of China through Grant Nos. 2011CB933400 and for rGO3 at nine frequencies.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 205.175.215.52 On: Wed, 08 Jun
2016 16:22:43
View publication stats