Diffusion mechanism of lithium ion through basal plane of layered graphene

Article abstract of DOI:10.1021/ja301586m

Coexistence of both edge plane and basal plane in graphite often hinders the understanding of lithium ion diffusion mechanism. In this report, two types of graphene samples were prepared by chemical vapor deposition (CVD): (i) well-defined basal plane gra

Full text of DOI:10.1021/ja301586m

Article
pubs.acs.org/JACS
Diffusion Mechanism of Lithium Ion through Basal Plane of Layered
Graphene
†
,∥
†
†
‡
†
§
Fei Yao, Fethullah Gunes ,̧ Huy Quang Ta, Seung Mi Lee, Seung Jin Chae, Kyeu Yoon Sheem,
Costel Sorin Cojocaru, Si Shen Xie, and Young Hee Lee*
̈
∥
†
,†
†Department of Energy Science, BK21 Physics Division, Graphene Center, Sungkyunkwan Advanced Institute of Nanotechnology,
Sungkyunkwan University, Suwon 440-746, Republic of South Korea
‡
Center for Nanocharacterization, Korea Research Institute of Standards and Science, Daejeon 305-340, Republic of South Korea
§
Samsung SDI Corporate R&D Center, Gongse-dong, Giheung-gu, Yongin-si, Gyeounggi-do 446-577, Republic of South Korea
∥
́
LPICM CNRS, Ecole polytechnique, Palaiseau 91120, France
*
S Supporting Information
ABSTRACT: Coexistence of both edge plane and basal plane in
graphite often hinders the understanding of lithium ion diffusion
mechanism. In this report, two types of graphene samples were
prepared by chemical vapor deposition (CVD): (i) well-defined
basal plane graphene grown on Cu foil and (ii) edge plane-enriched
graphene layers grown on Ni film. Electrochemical performance of
the graphene electrode can be split into two regimes depending on
the number of graphene layers: (i) the corrosion-dominant regime
and (ii) the lithiation-dominant regime. Li ion diffusion
perpendicular to the basal plane of graphene is facilitated by
defects, whereas diffusion parallel to the plane is limited by the steric hindrance that originates from aggregated Li ions adsorbed
on the abundant defect sites. The critical layer thickness (l ) to effectively prohibit substrate reaction using CVD-grown graphene
c
layers was predicted to be ∼6 layers, independent of defect population. Our density functional theory calculations demonstrate
that divacancies and higher order defects have reasonable diffusion barrier heights allowing lithium diffusion through the basal
plane but neither monovacancies nor Stone-Wales defect.
INTRODUCTION
Currently available highly oriented pyrolytic graphite (HOPG),
which is well-known as a highly ordered crystallographic
structure, has a finite size of flakes whose edge planes are still
abundant in addition to basal planes. Therefore, lithium ion
diffusion through the basal plane cannot be exclusively
■
Graphite has been widely used as an anode material in lithium
ion batteries due to its well-defined layered structure for lithium
intercalation, low operating potential, and remarkable interfacial
1
stability. Graphite has two characteristic planes: a basal plane
5
−8
observed.
Thus, a well-defined basal plane of graphite with
and an edge plane, which are parallel and perpendicular to the
c-axis, respectively. It is known in general that the basal plane
and edge plane exhibit different physical and chemical activities
in many aspects, leading to different lithiation capabilities in
a large area is required to have a comprehensive picture of
lithium diffusion mechanism in lithium ion batteries.
Recently, large area monolayer and multilayer graphene have
12,13
2
,3
been synthesized by chemical vapor deposition (CVD).
graphite. The diffusion time constant for Li ion insertion
within the active graphitic flakes is governed by the formula τ =
This paves a new route for exploring numerous new
fundamental sciences and, moreover, developing numerous
technological breakthroughs in electronics and energy
2
L /2D, where L is the diffusion length (or radius of spherical
flake) and D is diffusion coefficient. Although lithium diffusion
through basal plane is rather limited, lithium diffusion may still
4
14−16
storage.
Large area graphene can be transferred onto any
substrate by a simple transfer process and therefore an anode
electrode with layered graphene without leaving an edge plane
occur through several defect sites, such as vacancies and grain
5
,6
boundaries. Lithium diffusion through an edge plane of
graphitic flakes can be easily facilitated but is further
complicated by the presence of different functional groups
such as hydroxyl and carboxyl groups. In other words, lithiation
(
or negligible portion of edge plane) is easily attainable. This
provides an opportunity to study the diffusion of lithium ions
exclusively through the basal plane of graphene.
However, in lithium ion batteries, corrosion of the
5
−9
through these two different planes is highly anisotropic.
conventional current collectors such as Al, Cu, and stainless
One ambiguity in understanding a diffusion pathway of
lithium ions in graphite is the coexistence of both edge planes
and basal planes in the sample. The presence of these two
Received: February 28, 2012
Published: April 30, 2012
1
0,11
different interfaces is unavoidable in conventional graphite.
©
2012 American Chemical Society
8646
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
Figure 1. Schematic of a coin cell fabrication process with Cu-grown SLG or Ni-grown MLG. Bilayer and trilayer graphene coin cells were fabricated
by repeatedly transferring monolayer graphene.
steel (SUS) can adversely affect lifetime and safety through
mechanically polished with FeCl solution for flattening. The prepared
3
Cu foil was then brought into the growth chamber. The temperature
increased internal resistance, passivation of active materials, and
17−23
of the chamber was heated up to 1060 °C with 1000 sccm of Ar gas
consumption of electrolyte/active electrode materials.
and 200 sccm of H gas for 20 min. Methane (5 sccm) was then
2
Anode performance of thin graphene layers can be misguided
by the strong substrate reaction since the most reactive lithium
introduced with 10 sccm H gas for 5 min. After growth, the sample
2
was cooled to room temperature naturally in the same atmosphere. In
the case of MLG synthesis, Ni thin film (300 nm) was deposited on
2
4,25
ions exist in electrolyte.
It has been proposed that
monolayer graphene can be used as a protective layer for
substrate against air oxidation and mild electrochemical
SiO (300 nm)/Si by a thermal evaporator. This was placed in rapid
2
thermal CVD chamber. The temperature was increased to 1000 °C in
2
6,27
reaction.
Therefore, information on the critical layer
5 min in vacuum. Ni surface was reduced by flowing 45 sccm H gas at
2
thickness of graphene (l ) to minimize the substrate effect
1000 °C. The gas mixing ratio of C H /H was optimized to 2:45
sccm and flown for 1 min. After completion of growth, the gas supply
c
2
2
2
and the influence of defects to l are key ingredients to
c
was terminated, and the chamber was cooled to room temperature.
understand electrochemical reaction and protective nature of
graphene layers under severe electrochemical condition.
The main purpose of this work is two-fold: (i) To clarify the
lithium diffusion pathway through the basal plane of graphene
layers and (ii) to investigate the influence of defect population
to lithium ion diffusion and the protective ability of graphene
layers. In this work, we prepared Cu-grown monolayer
graphene (SLG) samples and Ni-grown multilayered graphene
2
8,29
The detail has been described elsewhere.
Transfer Process of a Graphene. PMMA (e-beam resist, 950 k
C4, Microchem) was spin-coated on the graphene/Cu foil (Ni film) at
1
000 rpm for 60 s. To etch away Cu foil (Ni film), the sample was
submerged in a copper etchant (CE-100, Transene) for ∼30 min (4 h
for Ni film). After rinsing by deionized water for a few times, the
PMMA/graphene layer was fished onto the CR 2032 cell case coated
with lithium-reaction resistive polymer. PMMA was removed by
acetone later after graphene was completely dried and attached onto
the cell case. The transferred sample was then annealed up to 650 °C
(
MLG) samples that are dominated with graphene basal planes
and edge planes, respectively. We found that the electro-
chemical performance of few-layer graphenes (FLGs) which are
overlapped up to three layers of SLG is strongly affected by the
substrate reaction. Experiments with Ar plasma treatment
indicated that 6 layers of basal plane-enriched large area
graphene were needed to provide effective substrate protection.
Combing the experimental results and density functional theory
calculations, we proved that basal plane hindered lithium ion
diffusion with a high diffusion barrier height, whereas
divacancies and higher order defects can be shortcuts for
lithium ion diffusion.
−6
for 5 h in high vacuum (1 × 10 Torr) for further removal of
3
0
PMMA.
Argon Plasma Treatment of Graphene. Structural defects of
+
graphene were created by Ar bombardment with different plasma
powers (15 W, 100 W) for 1 min. The transferred graphene was
−6
brought into the vacuum chamber with a base pressure of 1 × 10
Torr and then filled with Ar gas of 100 sccm for a minute, followed by+
the plasma ignition. This was repeated layer by layer to obtain Ar
plasma-treated FLG samples.
Material Characterization. For optical images and confocal
Raman spectroscopy, the CRM 200 (WiTec, Germany) with 100 lens
(
Olympus, N. A. 0.9) and ∼1 mW of 532 nm laser was used. For
micro-Raman spectroscopy (Renishaw, RM-1000 Invia) with an
excitation energy of 2.41 eV (514 nm, Argon laser) and UV−vis-
NIR absorption spectroscopy (Varian, Cary 5000) were used for
EXPERIMENTAL SECTION
■
Material Synthesis. Large area SLG was synthesized on copper
foil by atmospheric pressure (AP) CVD. Cu foil purchased from
Nilaco (Lot No. 113321, 99.96%, 100 μm in thickness) was
preannealed to 1060 °C for 2 h with 100 sccm of Ar gas and 200
sccm of H2 gas to enlarge Cu grain size and then chemico-
characterizing optical properties of the graphene films on SiO /Si and
2
PET substrates, respectively.
Electrochemical Measurement. Electrochemical measurements
were performed with a CR2032 coin cell using VMP3 instrument
8
647
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
Figure 2. Optical micrographs of (a) Cu-grown SLG and (b) Ni-grown MLG on SiO /Si substrate. White dashed lines indicate wrinkles. Some
2
portion of thicker graphene is indicated by arrows. (c) Schematic of (i) SLG with a well-defined basal plane and (ii) edge plane enriched MLG. (d)
Micro-Raman spectra of SLG and MLG. Confocal Raman mapping of D/G intensity ratio of (e) SLG and (f) MLG from squared positions of (a)
and (b). The contrast is normalized to 0.4 to visualize the defect distribution for both images. (g) Wavelength-dependent transmittance (values are
provided at a wavelength of 550 nm) and (h) optical photographs of different number of graphene layers on PET substrate.
Figure 3. (a) Cyclic voltammograms of different number of graphene layers samples at a scan rate of 0.1 mV/s. SUS-related redox reaction peaks
(
S , S ) and lithium intercalation/deintercalation related peaks (Li /Li ) are marked. (b) 1st and (c) 2nd galvanostatic charge/discharge profiles of
O
R
I
n
D
e
2
different number of graphene layers at a current density of 5 μA/cm . (d) The related layer-dependent capacities. The two regimes of corrosion-
dominance and lithiation-dominance are indicated.
(BioLogic Science Instruments). The cell was assembled in a dry room
Theoretical Calculation. We performed density functional theory
using the CR 2032 cell case with different number of graphene layers
and bare foil (SUS 316) as a working electrode, lithium metal foil as a
counter/reference electrode, and a 1 M of LiPF6 in a 1:1 (v/v)
mixture of ethylene carbonate (EC) and diethyl carbonate (DEC) as
an electrolyte. A glassy carbon microfiber was used as a separator. The
calculations within generalized gradient approximation as implemented
3
2,33
in DMol3 code.
All electron Kohn−Sham wave functions were
expanded in a local atomic orbital basis set with each basis function
defined numerically on an atomic centered spherical mesh. Double
numeric polarized basis sets (DNP) were used for all elements. The
dangling bonds of graphene edge were saturated by hydrogen atoms
and the atomic cluster structure which consists of 120 carbon atoms
and 48 hydrogen atoms were relaxed fully until the force on each atom
cells were charged and discharged galvanostatically between 3.0 and
2
0
.01 V at a constant current of 5 μA/cm . The AC impedance spectra
were obtained by applying a sine wave with an amplitude of 10 mV
31
−4
over a frequency range of 100 kHz to 10 mHz.
is less than 10 eV/Å and the total energy change is less than 5 ×
8
648
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
0− eV. The dampened atom-pairwise dispersion corrections of the
5
1
0.01 to 3 V. The bare SUS electrode showed an anodic peak
−6
34
form C6R
were also considered for calculations. Li adsorption
near 1.03 V (S ) and a cathodic peak around 0.78 V (S ).
O
R
energy was calculated by E (Li) = E (Li + carbon) − E (Li) −
E (carbon), where E (Li) is the self-energy of lithium atom and
Etot(carbon) is the total energy of carbon system. Various local charges
were also calculated using Mulliken, Hirshfeld, and electrostatic
potential (ESP). (See also Supporting Information, SI, S5.)
ad
tot
tot
These redox peaks involve chemical reactions with Li ions and
possibly electrolytes. Both anodic and cathodic peaks were
reduced in the monolayer graphene electrode. These peaks
were reduced consecutively in bilayer and trilayer graphene
electrodes. It is obvious to see that the redox reaction of the
bare SUS electrode was suppressed by the coated graphene
layers. An additional cathodic peak appeared near 0.28 V in
bilayer and trilayer samples. The origin of these peaks could be
tot
tot
RESULTS AND DISCUSSION
■
Figure 1 shows a schematic of the CR 2032 coin cell type
battery fabrication. Once SLG was synthesized on Cu foil by
APCVD (Figure 1b), the graphene layer was transferred onto
the cell case (Figure 1c−e), as described in the Experimental
Section. The half cell was then fabricated with a counter/
reference electrode of Li foil for the test of Li diffusion through
well-defined basal graphene plane (Figure 1f). Bilayer and
trilayer graphene coin cells were also fabricated by repeatedly
transferring the monolayer graphene. MLG was synthesized on
Ni film to represent graphene where the edge plane was
enriched and the half cell was similarly fabricated.
5
ascribed to defect-associated lithium adsorption. At the MLG
(15 layer graphenes) sample, a sharp cathodic peak near 0.01 V
(Li ) is identified as lithium intercalation and a rather broad
In
peak near 0.12 V (Li ) is related to decomposition of graphitic
De
24
intercalation compound (GIC) stages. It is of note that the
bare SUS-related peak was nearly compressed in this case. Both
LiDe and Li peaks appeared in this case, in good contrast with
In
FLG samples in which only a clear LiIn peak was observed,
suggesting that no GIC stages were formed in FLGs. The
distinct CV behavior of FLGs and MLG demonstrates that
lithium ion intercalation becomes more effective in MLG
induced by the considerable amount of edge planes, as shown
in Figure 2b.
In order to clarify the quality and layer number of graphene,
a series of characterization was done, as shown in Figure 2.
Figure 2a,b shows optical micrographs of the transferred SLG
and MLG on SiO /Si substrate. The SLG grown on Cu foil was
Figure 3b shows the first galvanostatic charge/discharge
profile with a voltage sweeping range of 0.01−3 V at a constant
2
rather flat, with the exception of a small portion (∼ 4%) of
bilayer and trilayer graphene domains represented by the dark
spots (arrows) in the image (Figure 2a). Some wrinkles
indicated by the white dashed lines introduced during transfer
process were also visible. Contrary to this, Ni-grown MLG
showed multilayered flakes represented by the white spots
2
current of 5 μA/cm . As the number of graphene layers
increased, a long tail appeared in the charge curve at the low
voltage region. At MLG sample, a plateau appeared in the range
of 1.25−0.6 V. In graphitic material, the solid-electrolyte
interface (SEI) formation via electrolyte decomposition takes
7
,37−39
(
arrows) in Figure 2b, creating numerous edge planes, as can be
place in the range of less than 1.0 V.
The SEI formation
visualized in Figure 2c. Micro-Raman spectra in Figure 2d
potential varies with types of graphite planes. In general, SEI
forms at higher potential in edge plane than in basal plane.
−1
40−42
clearly show G-band near 1590 cm , which is related to the
optical E2g phonon at the Brillouin zone center indicating sp2
hybridization of carbon network, and G′-band around 2694
Therefore, we ascribed this plateau in MLG to edge plane-
related SEI formation. In the second cycle, the voltage profile
shows a gradual change in a wide range of voltages during
charge/discharge, revealing a V-shape curve, i.e., no plateau
region, as shown in Figure 3c. This is in good contrast with a U-
shaped curve in graphite electrode, where the edge plane
intercalation is dominant in the plateau region of low voltage
−1
cm , which is also known as 2D-band, an overtone of D-band,
3
5
in both samples. Large G′/G intensity ratio (∼2) with a small
−1
D-band near 1350 cm , which corresponds to transverse
3
optical phonon near the K point and indicates sp hybridization
of carbon network, was observed in SLG, indicating high
quality monolayer graphene. However, the intensity ratio of G′/
G which is less than one reveals multilayered properties of Ni-
grown graphene. Defect distribution was shown in the images
of confocal Raman mapping of D/G intensity ratio in Figure
39,43
within 0.1 V.
Capacities of graphene-coated electrodes in
Figure 3c were consistently smaller than those of the bare
electrode, and furthermore much smaller by about 30 times
16
than the recently reported graphene battery result. The huge
capacity difference comes from the use of different substrates
(See Figure S1 of the SI). This implies that in spite of graphene
layers coated on the electrode with well-defined basal plane, the
reaction with electrode did inevitably occur.
2e,f. Defects indicated by bright spots were scattered uniformly
over the surface, while grain boundary lines were faintly visible
in SLG. Small flakes were visible in MLG (Figure 2f). Although
D-band intensity was barely visible in Figure 2d, we clearly
observed from D/G band mapping that some defects were
distributed in both samples. Transmittance of each graphene
layer transferred onto PET substrate is provided in Figure 2g.
The transmittance of SLG was 96.5%, slightly smaller than
HOPG value of 97.7%, which may be attributed to some
portion of multilayered domain formation as described in
The related layer-dependent capacities are summarized in
Figure 3d. As the number of graphene layers increased, the first
charge capacity increased rapidly up to trilayer graphene
electrode and saturated at the MLG electrode. As described in
the schematic of Figure 2c, the basal plane is exposed during
lithiation up to three layers, whereas both the edge plane and
basal plane are present in 15 layers. Two different types of SEI
are formed: (i) basal-plane associated SEI (b-SEI) which is
formed up to 3 graphene layers and (ii) edge-plane associated
SEI (e-SEI) which is formed in MLG sample. It has been
known that b-SEI formed at lower potential is associated with
solvent reduction, while e-SEI formed at higher potential is
3
6
Figure 2a. Correspondingly, bilayer and trilayer graphene
samples revealed a systematic reduction in the transmittance.
The Ni-grown MLG showed 63.6% of transmittance,
corresponding to 15 layers in average by assuming 2.3%
36
absorption per each layer. Optical photographs were provided
to visualize different transmittances with different numbers of
graphene layers in Figure 2h.
40−42,44
associated with salt ions.
Since our basal plane contains
Figure 3a shows cyclic voltammograms (CV) of different
numbers of graphene layers at a scan rate of 0.1 mV/s from
abundant defect sites, as observed from Figure 2d,e, some
decomposed solvent molecules may further diffuse into the
8
649
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
Figure 4. (a) Raman spectra, (b) cyclic voltammograms at a scan rate of 0.1 mV/s, and (c) 2nd galvanostatic charge/discharge profiles at a current
2
density of 5 μA/cm for monolayer graphene treated by Ar plasma with different plasma powers (15 and 100 W). (d) Capacity of 2nd charge as a
functional of number of graphene layers under different Ar plasma powers. Absolute slopes according to different plasma powers and critical layer
thickness (l ) are indicated. (e) Schematics of proposed Li diffusion mechanism through defects on the basal plane with different defect population.
c
Broad down arrows indicate Li ion diffusion through defect sites of basal plane. Red glows represent steric hindrance for Li ion diffusion formed by
the accumulated Li ions or functional groups. The inset in the right indicates the relative magnitude of diffusion coefficient. (f) Relationship of D/G
ratio with the extracted slope from (d).
subjacent layers along with Li ions or in a form of lithium
salvation and form additional b-SEI layer. This is why b-SEI
increases as the number of graphene layers increases at FLG
samples. At the MLG electrode, both b-SEI and e-SEI are
formed. Although e-SEI increases in this case, b-SEI is reduced
compared to FLG electrodes due to the decrease of effective
basal plane area of 15 layers (See Figure 2c) and therefore the
capacity from SEI formation is saturated in the first charge.
On the other hand, the first discharge capacity decreased
gradually up to three layer graphene electrode and increased at
curves in FLGs (Figure 3a). This implies several facts: (i) SUS
substrate reaction is systematically suppressed with an
increasing number of graphene layers (For impedance
measurement of pristine graphene layers, see Figure S2 of the
SI). (ii) Because lithium ions can diffuse through the basal
plane of graphene, monolayer graphene is not sufficient to
prohibit substrate reaction. Since the pure basal plane
presumably does not allow Li diffusion, the diffusion may be
provoked through some defect sites that exist on the graphene
plane, as observed from the D/G intensity ratio of confocal
Raman mapping in Figure 2e. This will be described later in
detail. In FLG samples, if we presume capacity only to be
15 layer electrode. Similar trend was also observed in the
second charge/discharge profile. The discharge capacities of the
second cycle were not much different from those of the first
cycle. A large capacity of the bare SUS electrode was reduced
by coating graphene layers up to three layers. This gradual
reduction was also expected from the reduced areas of CV
−2
contributed from intercalation (0.028 μAhcm /interlayer in
the case of LiC ), then the intercalation capacity reaches 0.056
6
2
μAh/cm at trilayer graphene sample (see Figure S3 of the SI).
2
This value is negligible to the capacity (0.73 μAh/cm )
8
650
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
observed in our experiment. This tells us that even if
intercalation of lithium ions was invoked, the observation was
still obscured by the dominant SUS redox reaction. By noting a
linear decrease of the capacity and hence extrapolating to a
substrate redox reaction was also suppressed, which is identified
by the capacity decrease with increasing number of graphene
layers similar to that of pristine graphene samples.
Smaller capacity was increased in FLG electrodes compared
to that of SLG after plasma treatment and generated different
slopes, as shown in Figure 4d. The absolute slope increased
from 0.26 to 0.56 with increasing the plasma power.
Extrapolation of these slopes, which determines the critical
minimum capacity, l to effectively prohibit the SUS redox
c
reaction is predicted to be ∼6 layers (See Figure S2 of the SI
for the impedance measurement of 6 layer graphene sample).
As the number of graphene layers increases, the capacity from
the SUS redox reaction decreases, while the capacity reduction
will be compensated by the intercalation capacity between
graphene layers. After 6 layers, the capacity starts increasing by
the pure intercalation. We can define substrate corrosion-
dominant region up to 6 layers and lithiation-dominant region
after 6 layers, as visualized in Figure 3d.
layer thickness to prohibit substrate reaction, gave rise to l of
c
∼6 layers independent of the plasma power, i.e., defect
population. This is rather surprising, because creation of
more defects in the basal plane is expected to increase basal-
plane diffusion of Li ions which will eventually increase
substrate redox reaction (See Figure 4b,c) and thereby larger
critical layer thickness should be required after strong plasma
treatment. In order to explain this contradictory phenomenon,
a schematic of Li diffusion through defects in the basal plane is
provided in Figure 4e. In the case of SLG, Li ion diffusion is
allowed through defect sites and no lateral diffusion limitation
is expected, since the graphene layer is fully surrounded by Li
ions in electrolyte. Therefore, a higher defect population will
enhance Li ion adsorption and also substrate reaction. In the
case of FLGs, where large area basal plane is dominant,
graphene layers are overlapped with each other so that Li ions
will diffuse through defects perpendicular to the plane of the
top layer first and diffuse along the plane of subjacent graphene
layer until they meet another defect site. Since these Li ions
may accumulate near the defect sites generated by Ar plasma, Li
diffusion along the plane direction will be limited by the steric
hindrance from aggregated Li atoms, which is different from the
SLG case. Therefore, further Li diffusion through graphene
basal planes in FLGs is constrained severely by the lateral
diffusion at higher defect density. Thus, when FLGs are used as
a protective layer, the defects-related lithium adsorption on
subjacent graphene layers and actual lithium ion reaction with
substrate are suppressed, which is again consistent with the
reduction of the peaks in CV diagrams (See Figure S4a,b of the
SI). As a consequence of these phenomena, the critical layer
thickness gives rise to the same value, independent of the defect
population. It will be worth mentioning the possibility of
forming oxygen-related functional groups on defect sites. Li
ions can also be adsorbed on such sites and thus our argument
of lateral diffusion suppression by the steric hindrance is still
valid.
It is intriguing to see the relationship between D/G intensity
ratio from Raman spectra and the slope extracted from charge/
discharge profiles, as shown in Figure 4f. The slope which
indicates Li diffusion through graphene layers is correlated to
the population of defects in the graphene plane. The larger
slope implies the slower diffusion rate and vice versa. Li ion
diffusion is limited by the Li aggregates adsorbed on the
increased defect sites described in the schematic (e). Thus,
information of Li diffusion obtained from electrochemical test
could be used as a metric for evaluating the defect population of
graphene, an important material parameter of graphene.
In order to understand what type of defects allows Li ion
diffusion through the basal plane of graphene, we conducted
density functional theory calculations for various defects: ideal
hexagonal site (H site), Stone−Wales defect (SW), mono-
vacancy (V1), and divacancy (V2). Li atom adsorbs on the H
site with a bond length of 2.35 Å above the graphene plane and
with an adsorption energy of −1.69 eV, as shown in the upper
panel of Figure 5a and Table 1. Li ESP charge at H site is
The theoretically estimated capacity at 15 layers (or
2
effectively 9 layers), is 0.2 μAh/cm (see Figure S3 of the
SI). However, this value is still far smaller than the observed
2
value of 1.30 μAh/cm . This extra capacity could be ascribed to
the lithium adsorption on defects of the graphene surface,
which can be supported by the widely distributed defects
observed from confocal Raman mapping in our experiments
(
Figure 2f). In the second cycle, the discharge capacity was
consistently smaller than the charge capacity, nearly independ-
ent of the thickness of graphene layers. This difference of 0.35
μAh/cm2 in the charge/discharge capacity is irreversible
capacity and can be ascribed to strongly adsorbed lithium
ions on defects such as vacancies or grain boundaries formed
on the graphene layers. This will be discussed in the theory
section later.
Since defects on graphene basal plane seem to play an
important role in lithium diffusion, a systematic study is
required for comprehensive analysis. Figure 4a shows Raman
spectra of Ar plasma-treated monolayer graphene. At 15 W
plasma power, D/G intensity ratio was increased to 0.56 from
0.19 in no plasma-treated pristine graphene, implying structural
defect formation in the graphene plane. At 100 W, one
−1
additional peak near 1620 cm (D′) appeared in addition to
further increase of D-band intensity (D/G intensity ratio is
4
5
1
.66), indicating plausible formation of structural defects. No
+
−
peak splitting of G-band into G and G peaks indicates that
46
our process does not involve strain-induced effect. Figure 4b
shows CV diagrams for SLG electrode with different plasma
powers at a scan rate of 0.1 mV/s. (For CV diagrams of plasma-
treated graphene samples with different numbers of graphene
layers, see Figure S4 of the SI.) It is obvious to see that the
redox reaction peak intensities of S and S related peaks were
O
R
enhanced and the related peak positions were also shifted after
plasma treatment. Those peaks are a combination of defect-
associated adsorption and SUS substrate reaction, as mentioned
in Figure 3a. Additional redox reaction due to the generated
basal plane defects by plasma treatment is provoked. Since the
protective layer is monolayer graphene, extra lithium ions
adsorbed on defects could easily reach the SUS substrate, thus
increasing the substrate redox reaction. The second galvano-
static charge/discharge capacity of SLG increased accordingly
compared to the pristine graphene sample, as shown in Figure
4c. (For charge/discharge profiles of plasma-treated bilayer and
trilayer graphene electrodes, see Figure S4 of the SI.) The
enhanced capacity was attributed to the increased adsorption of
Li ions on defects and increased substrate reaction, as
mentioned in Figure 4b. The second charge capacity kept
increasing with increasing plasma power, independent of the
number of graphene layers, as summarized in Figure 4d. The
8
651
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
Figure 5. Side and top views of atomic configurations (top panel), isosurface images of electrostatic potential (second panel), bond lengths and local
charge distributions at the barrier states (third panel), and diffusion barrier profiles of Li (bottom panel) through (a) graphene hexagonal site (H
3
site), (b) Stone-Wales (SW) defect (c) monovacancy (V1), and (d) divacancy (V2). Isovalue for rendering isosurfaces is 0.25 e/Å . The insets in the
third panel show the isosurface image of electrostatic potential for each corresponding structure without Li ion. Bond lengths (yellow color) and
electrostatic potential charges (white color) are in units of Å and electrons, respectively.
Table 1. Defects Related Li Adsorption Energy, And Li Atomic Charges Calculated by Mulliken, Hirshfeld, And Electrostatic
a
Potential (ESP) at the Minimum Energy Configurations (M) and the Barrier States (B)
Li atomic charge (e)
Mulliken
Hirshfeld
ESP
binding site
adsorption energy (eV)
diffusion barrier height (eV)
M
B
M
B
M
B
b
H site
SW
VI
−1.69
−1.94
−3.12
−2.36
10.2 (10.7)
0.55
0.53
0.46
0.53
−1.64
−1.00
−1.05
−0.04
0.48
0.44
0.38
0.40
0.20
0.20
0.14
0.18
0.62
0.63
0.71
0.62
0.28
0.67
0.53
0.56
6.35
b
8.86(10.3)
V2
2.36
a
b
Positive charge indicates charge depletion from lithium atom. See ref 47.
8
652
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
partially depleted to 0.62 e. The ESP charge of Li atom at
barrier state is 0.28 e, much less compared to that at the binding
site (see Figure S5 of the SI for details). This charge difference
between adsorption and barrier state is an important variable in
determining the Coulomb interaction energy. As the Li
approaches to the barrier site, the available space for Li is
narrow with a short separation distance of 1.52 Å, invoking
severe charge overlapping between Li and adjacent carbon
atoms, as can be seen in the electrostatic potential contour in
the second panel of Figure 5a. This increases repulsive forces,
giving rise to large diffusion barrier height of 10.2 eV, similar to
ASSOCIATED CONTENT
Supporting Information
■
*
S
Figures showing the corrosion effect of different types of SUS
substrate, AC impedance spectra of pristine graphene samples,
relationship between intercalation capacity and different
numbers of graphene layers, cyclic voltammogram and 2nd
charge/discharge behavior of graphene samples with different
number of layers under different plasma powers, charges (Li
site) calculated by various models with varying the normal
distance from barrier site to the stable position of Li adsorption
at the H site. This material is available free of charge via the
Internet at http://pubs.acs.org.
47
48
the previous report (Table 1), validating our approaches.
A
similar situation takes place in the SW defect, which is
49
abundant in the graphene grain boundary. The Li adsorption
energy near the heptagon is −1.94 eV, slightly stronger than
that of H site. Although the charge overlapping is still severe, a
longer separation distance of 1.60 Å and also much less charge
difference between adsorption and barrier state (0.04 e) forms a
relatively smaller activation barrier height of 6.35 eV than that
of H site, as shown in Figure 5b. In the case of V1, Li adsorbs at
the vacancy site with an adsorption energy of −3.12 eV,
keeping closer distance (2.03 Å), as shown in the top panel of
Figure 5c. The excess charge difference of Li atom between
adsorption and barrier site is 0.18 e and the closest separation
distance at the barrier site is 1.36 Å. Charges are distributed not
only on the Li and carbon sites, but also between them,
implying both covalent bonding and ionic bonding characters
due to charge depletion from Li atom. This produces a large
diffusion barrier height of 8.86 eV. However, V2 provides a
rather large open space with an adsorption energy of −2.36 eV
near the middle of the two dimers (top panel of Figure 5d)
such that a large separation distance of 2.90 Å is maintained.
This gives minimizes electrostatic charge overlapping and a
large bond length of 1.83 Å at the barrier state, i.e., steric
hindrance is minimized, as shown in the second and third
panels in Figure 5d. The charge difference between the
adsorption and the barrier states is 0.04 e. All of these factors
induced a smallest diffusion barrier height (2.36 eV) among the
defects we studied. This barrier height can be overcome under
the typical charging conditions of the battery.
AUTHOR INFORMATION
■
Corresponding Author
leeyoung@skku.edu
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
0
This work was supported by the Star Faculty program (2010-
029653) and WCU program (R31- 2008-10029) of the NRF
of Korea funded by MEST. The density functional theory work
was supported by KRCF through KRISS. Authors gratefully
acknowledge Drs. Feng Li and Li Liu for valuable discussions.
REFERENCES
1) Winter, M.; Besenhard, O. J.; Spahr, E. M.; Novak
998, 10, 725.
■
(
́
, P. Adv. Mater.
1
(
1
(
1
2) Winter, M.; Novak
45, 428.
3) Bar-Tow, D.; Peled, E.; Bursteinb, L. J. Electrochem. Soc. 1999,
46, 824.
(4) Ren, Y.; Armstrong, R. A.; Jiao, F.; Bruce, G. P. J. Am. Chem. Soc.
́
, P.; Monnierb, A. J. Electrochem. Soc. 1998,
2010, 132, 996.
(5) Tran, T.; Kinoshita, K. J. Electroanal. Chem. 1995, 386, 221.
6) Persson, K.; Sethuraman, A. V.; Hardwick, J. L.; Hinuma, Y.;
Meng, S. Y.; Ven, A.; Srinivasan, V.; Kostecki, R.; Ceder, G. Phys.
Chem. Lett. 2010, 1, 1176.
(
(
7) Placke, T.; Siozios, V.; Schmitz, R.; Lux, S. F.; Bieker, P.; Colle,
C.; Meyer, H.-W.; Passerini, S.; Winter, M. J. Power Sources 2012, 200,
83.
(8) Jungblut, B.; Hoinkis, E. Phys. Rev. B 1989, 40, 10810.
(9) Yamada, Y.; Miyazaki, K.; Abe, T. Langmuir 2010, 26 (18),
CONCLUSIONS
■
We have studied lithium diffusion pathways with two types of
graphene samples prepared by CVD; (i) well-defined basal
plane graphene grown on Cu foil and (ii) edge plane-enriched
graphene layers grown on Ni film. We have discovered that
electrochemical reaction of the electrode (substrate/graphene)
is not only related to the number of graphene layers but also
relies on the defect sites on the basal plane of graphene. The
experimental and calculated results related to the specific types
of defects, such as divacancies and higher order defects, that can
assist lithium ion diffusion through the basal plane could help
us in designing high capacity and highly conductive corrosion-
free electrodes for lithium ion batteries. It would be reasonable
to expect that the substrate-protective nature of few-layer
graphenes could be the basis for further investigation of the
preparation of original substrate, which remains unaltered with
respect to its properties and has a longer lifetime under the
severe electrochemical corrosion conditions found in batteries.
Furthermore, by correlating the lithium diffusion in graphene
layers to the D/G intensity ratio from Raman spectra, we
developed a way of predicting the defect population of
graphene with an electrochemical method.
1
(
4990.
10) Tarascon, J. M.; Guyomard, D. G. Extended Abstracts, Spring
Meeting of the Electrochemical Society, May 16−21, 1993, 93−1,
Honolulu, HI, no. 69.
(
11) Takamura, T.; Endob, K.; Fuc, L.; Wu, Y.; Lee, K. J.;
Matsumoto, T. Electrochim. Acta 2007, 53, 1055.
12) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, S. K.;
(
Ahn, J.-H.; Kim, P.; Choi, J.-Y.; Hong, B. H. Nature 2009, 457, 706.
(13) Li, X. S.; Cai, W. W.; An, J. H.; Kim, S.; Nah, J.; Yang, D. X.;
Piner, R. D.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.;
Colombo, L.; Ruoff, R. S. Science 2009, 324, 1312.
(14) Biswas, C.; Lee, Y. H. Adv. Func. Mater. 2011, 21, 3806.
(15) Yu, W. J.; Chae, S. H.; Lee, S. Y.; Duong, D. L.; Lee, Y. H. Adv.
Mater. 2011, 23 (16), 1889.
16) Reddy, A. L. M.; Srivastava, A.; Gowda, R. S.; Gullapalli, H.;
Dubey, M.; Ajayan, M. P. ACS Nano 2010, 4, 6337.
̈
17) Vetter, J.; Novak, P.; Wagner, M. R.; Veit, C.; Moller, K.- C.;
(
(
́
Besenhard, J. O.; Winter, M.; Wohlfahrt-Mehrens, M.; Vogler, C.;
Hammouche, A. J. Power Sources 2005, 147, 269.
(18) Arora, P.; White, E. R. J. Electrochem. Soc. 1998, 145, 3647.
(19) Wijesinghe, T.; Blackwood, D. Corros. Sci. 2007, 49, 1755.
8
653
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654

Journal of the American Chemical Society
Article
(
(
20) Kadry, S. Eur. J. Sci. Res. 2008, 22, 508.
21) Hyams, C. T.; Go, J.; Devinea, M. T. J. Electrochem. Soc. 2007,
(49) Banhart, F.; Kotakoski, J.; Krasheninnikov, A. V. ACS Nano
2011, 5, 26.
1
(
54, C390.
22) Song, S.-W.; Richardson, J. T.; Zhuang, V. G.; Devine, M. T.;
Evans, W. J. Electrochim. Acta 2004, 49, 1483.
23) Zhang, X.; Winget, B.; Doeff, M.; Evans, W. J.; Devine, M. T. J.
Electrochem. Soc. 2005, 152, B448.
24) Pollak, E.; Geng, B.; Jeon, K.-J.; Lucas, T. I.; Richardson, J. T.;
Wang, F.; Kostecki, R. Nano Lett. 2010, 10, 3386.
25) Reddy, A. L. M.; Srivastava, A.; Gowda, R. S.; Gullapalli, H.;
Dubey, M.; Ajayan, M. P. ACS Nano 2010, 4, 6337.
26) Chen, S.; Brown, L.; Levendorf, M.; Cai, W. W.; Ju, S.-Y.;
(
(
(
(
Edgeworth, J.; Li, X.; Magnuson, C.; Velamakanni, A.; Piner, D. R.;
Kang, J.; Park, J.; Ruoff, S. R. ACS Nano 2011, 5, 1321.
(
27) Prasai, D.; Tuberquia, J. C.; Harl, R. R.; Jennings, G. K.; Bolotin,
K. I. ACS Nano 2012, 6 (2), 1102.
28) Han, G. H.; Gunes, F.; Bae, J. J.; Kim, E. S.; Chae, S. J.; Shin, H.-
J.; Choi, J.-Y.; Pribat, D.; Lee, Y. H. Nano Lett. 2011, 11, 4144.
29) Chae, S. J.; Gunes, F.; Kim, K. K.; Kim, E. S.; Han, G. H.; Kim,
(
̈
̧
(
̈
̧
S. M.; Shin, H.-J.; Yoon, S.-M.; Choi, J.-Y.; Park, M. H; Yang, C. W.;
Pribat, D.; Lee, Y. H. Adv. Mater. 2009, 19 (16), 2553.
(
̈
30) Gunes ,̧ F.; Han, G. H.; Shin, H.-J.; Lee, S. Y.; Jin, M.; Duong, D.
L.; Chae, S. J.; Kim, E. S.; Yao, F.; Benayad, A.; Choi, J.-Y.; Lee, Y. H.
Nano 2011, 6, 409.
(
31) Naoi, K.; Ogihara, N.; Igarashi, Y.; Kamakura, A.; Kusachi, Y.;
Utsugi, K. J. Electrochem. Soc. 2005, 152 (6), A1047. Gong, J. B.; Wu,
H. Q. Electrochim. Acta 2000, 45, 1753. Funabiki, A.; Inaba, A.; Ogumi,
Z. J. Power Sources 1997, 68, 227. Zhang, H.; Li, F.; Liu, C.; Cheng, H.
J. Phys. Chem. C 2008, 112, 7767.
(
32) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
865.
33) Delley, B. J. Chem. Phys. 1990, 92, 508; J. Chem. Phys. 2000,
No. 113, 7756.
3
(
(
34) Tkatchenko, A.; Scheffler, M. Phys. Rev. Lett. 2009, 102, 073005.
(
35) Casiraghi, C.; Pisana, S.; Novoselov, K. S.; Geim, A. K.; Ferrari,
A. C. Appl. Phys. Lett. 2007, 91, 233108.
36) Nair, R. R.; Blake, P.; Grigorenko, A. N.; Novoselov, K. S.;
Booth, T. J.; Stauber, T.; Peres, N. M. R.; Geim, A. K. Science 2008,
20, 1308.
37) Edstro
006, 153, 380.
38) Li, X.; Geng, D.; Zhang, Y.; Meng, X.; Li, R.; Su, X. Electrochem.
Commun. 2011, 13, 822.
39) Bruce, P. G.; Scrosati, B.; Tarascon, J. - M. Angew. Chem., Int. Ed.
008, 47, 2930.
40) Hirasawa, K.; Sato, T.; Asahina, H.; Yamaguchi, S; Mori, S. J.
Electrochem. Soc. 1997, 144 (4), L81.
41) Chu, C. A.; Josefowicz, J. Y.; Farrington, G. C. J. Electrochem.
Soc. 1997, 144 (12), 4161.
42) Tang, M.; Miyazaki, K.; Abe, T.; Newman, J. J. Electrochem. Soc.
012, 159 (5), A634.
43) Yoo, E. J.; Kim, J.; Hosono, E.; Zhou, H. S.; Kudo, T.; Hunma, I.
Nano Lett. 2008, 8, 2277.
44) Yan, J.; Zhang, J.; Su, Y. C.; Zhang, X. G.; Xia, B. J. Electrochim.
Acta 2010, 55, 1785.
45) Kudin, K. N.; Ozbas, B.; Schniepp, H. C.; Prud’homme, R. K.;
Aksay, I. A.; Car, R. Nano Lett. 2008, 8, 36.
46) Mohiuddin, T. M. G.; Lombardo, A.; Nair, R. R.; Bonetti, A.;
(
3
(
̈
ma, K.; Herstedt, M.; Abraham, D. P. J. Power Sources
2
(
(
2
(
(
(
2
(
(
(
(
Savini, G.; Jalil, R.; Bonini, N.; Basko, D. M.; Galiotis, C.; Marzari, N.
K.; Novoselov, S.; Geim, A. K.; Ferrari, A. C. Phys. Rev. B 2009, 79,
2
(
2
(
05433.
47) Zheng, J.; Ren, Z.; Guo, P.; Fang, L.; Fan, J. Appl. Surf. Sci. 2011,
58, 1651.
48) For Li diffusion barrier calculations, the position of Li atom was
fixed and carbon atoms were allowed to relax fully simultaneously,
except the saturating hydrogen atoms and the carbon atoms bound to
hydrogen atoms. The diffusion barrier height obtained through this
process may result in higher values due to the plausible improper
pathways taken in this study.
8
654
dx.doi.org/10.1021/ja301586m | J. Am. Chem. Soc. 2012, 134, 8646−8654