ARTICLE

https://doi.org/10.1038/s41467-018-08233-3 OPEN

Towards maximized volumetric capacity via pore-

coordinated design for large-volume-change
lithium-ion battery anodes
Jiyoung Ma 1, Jaekyung Sung1, Jaehyung Hong2, Sujong Chae 1, Namhyung Kim 1, Seong-Hyeon Choi1,
Gyutae Nam1, Yoonkook Son3, Sung Youb Kim2, Minseong Ko4 & Jaephil Cho1
1234567890():,;

To achieve the urgent requirement for high volumetric energy density in lithium-ion batteries,
alloy-based anodes have been spotlighted as next-generation alternatives. Nonetheless, for
the veritable accomplishment with regards to high-energy demand, alloy-based anodes must
be evaluated considering several crucial factors that determine volumetric capacity. In par-
ticular, the electrode swelling upon cycling must be contemplated if these anodes are to
replace conventional graphite anodes in terms of volumetric capacity. Herein, we propose
macropore-coordinated graphite-silicon composite by incorporating simulation and mathe-
matical calculation of numerical values from experimental data. This unique structure exhibits
minimized electrode swelling comparable to conventional graphite under industrial electrode
fabrication conditions. Consequently, this hybrid anode, even with high specific capacity
(527 mAh g−1) and initial coulombic efficiency (93%) in half-cell, achieves higher volumetric
capacity (493.9 mAh cm−3) and energy density (1825.7 Wh L−1) than conventional graphite
(361.4 mAh cm−3 and 1376.3 Wh L−1) after 100 cycles in the full-cell configuration.

1 Department of Energy Engineering, School of Energy and Chemical Engineering, Ulsan National Institute of Science and Technology (UNIST), 50, UNIST-gil,
Ulsan 44919, Republic of Korea. 2 School of Mechanical, Aerospace and Nuclear Engineering, Ulsan National Institute of Science and Technology (UNIST),
50, UNIST-gil, Ulsan 44919, Republic of Korea. 3 Department of Electric Engineering, Chosun University, 309, Pilmun-daero, Dong-gu, Gwangju 61452,
Republic of Korea. 4 Department of Metallurgical Engineering, Pukyong National University, Busan 48547, Republic of Korea. These authors contributed
equally: Jiyoung Ma, Jaekyung Sung. Correspondence and requests for materials should be addressed to S.Y.K. (email: [email protected])
or to M.K. (email: [email protected]) or to J.C. (email: [email protected])

NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3

T
he ever-increasing demand for high-energy density in lithiation23. Such phenomenon could directly influence the elec-
lithium-ion batteries has stimulated ongoing research on trode swelling and electrical contact loss, and damage the gra-
anode materials. To satisfy this demand, improved anode phite, re-exposing the fresh graphite surface to the electrolyte and
volumetric capacity in high areal mass loading is a prerequisite causing severe volumetric capacity fading during cycling. On the
for practical full-cell systems1–4. However, most studies have other hand, mathematical calculation confirmed that a macropore
focused on increasing the gravimetric capacity, which is the just can have sufficient space to accommodate expansion of the Si-
one of five volumetric capacity determining factors (VDFs) with layers (≥7 nm) contained within this pore (Supplementary Fig. 1
the active material ratio, initial electrode density, electrode and Supplementary Note 2). Accordingly, we thoughtfully con-
swelling ratio, and N/P ratio (defined by areal capacity ratio sidered the effective exploitation of pores to negate the electrode
between anode and cathode), as shown in Fig. 1a and Eq. (1) swelling from Si within the graphite-Si composite (Supplementary
(Supplementary Note 1). More critically, although the electrode Fig. 3).
swelling drastically decreases the volumetric capacity during Notice that inner pore distribution of G can be manipulated
cycling (Fig. 1b), this value has often been calculated with respect during pyrolysis treatment of carbon precursors. In this study, we
to the volume of the pristine electrode before cycling. Therefore, design macropore-coordinated graphite (MG) via chemical vapor
it is strongly desired that the VDFs-defined volumetric capacity deposition process which is mature technology currently used in
during cycles should be considered. the anode industry. And, macropore-coordinated graphite-Si
(MGS) hybrid is developed based on thorough analytical methods
Volumetric capacity ¼ and mathematical calculation (Fig. 1c, d and Supplementary
Gravimetric capacity ´ Active material ratio ´ Initial electrode density Note 3). This unique structure, with no Si-layers located on the
Electrode swelling ratio ´ N=P ratio mesopores because of carbon pre-filling (carbon-blocking),
maintains morphological integrity without cracking and contact
ð1Þ losses during cycling. Notably, volume expansion of Si-layers
selectively situated on internal macropores is well accommodated.
With regard to volumetric capacity, carbonaceous anode is Therefore, the MGS demonstrates a minimized electrode swelling
unprecedented, having low electrode swelling, excellent cyclabil- ratio comparable to that of conventional G under industrial
ity, and outstanding densification properties during electrode electrode fabrication conditions for 500 h (over 100 cycles)
fabrication1,5–7. However, the intrinsic hurdle of low volumetric according to in situ thickness measurement, as shown in Fig. 1e, f
capacity resulting from limited gravimetric capacity encourages and Supplementary Fig. 4. As a result, the strategically synthe-
the battery community to focus on alloy-based anodes (Si, Ge, Sn, sized hybrid anode achieved higher volumetric capacity (493.9
etc.)8,9. Despite their high gravimetric capacity, the most critical mAh cm−3) and energy density (1825.7 Wh L−1) than that of G
challenge concerning alloy-based anodes is poor cyclability due to (361.4 mAh cm−3 and 1376.3 Wh L−1) at 100 cycles in full-cell
their massive volume changes during charge–discharge, which configuration.
induce mechanical fracturing and solid–electrolyte interphase
(SEI) layer instability10,11.
In the past decades, there have been various advanced nano- Results
engineering strategies to alleviate the stress derived from the volume MGS fabrication and characterization. Cross-sectional scanning
changes, through nanoparticle, nanotube, and nanowire design12– electron microscope (SEM) images show that particles of
15. In particular, yolk–shell nano-structures have been successful, spherical-type conventional G, selected as pristine materials for
having empty space to accommodate their volume changes, yielding MGS synthesis, contain many internal mesopores and macro-
further improved cyclability with high gravimetric capacity16,17. pores (Fig. 2a). First, the mesopores were filled via carbon-
However, considering the VDFs, the fabricated electrode yields low blocking to obtain MG through a chemical vapor deposition
volumetric capacity, because nano-properties accompanying the process using ethylene (C2H4), in accordance with the mathe-
low tap density induce excessive use of binder and conductive agent matical calculation in Supplementary Note 3 (Fig. 2b). The
(low active material ratio) and incompatibility with electrode carbon-blocking prevented subsequent Si coating from filling the
calendering (low electrode density)18,19. mesopores. Si-layers were then homogenously distributed in the
Recently, graphite-alloy composites have received considerable macropores and on the G surface via thermal decomposition of
attention, because the nano-engineering advantages are monosilane (SiH4) (Fig. 2c). The unique structural characteristics
strengthened while the nano-property weaknesses are of MGS are schematically illustrated in the cross-section of
compensated20,21. Most previous studies indicate that these Fig. 2d. High-resolution TEM (HR-TEM) images (Fig. 2e) and
composites are the most feasible alternatives for next-generation fast Fourier transform analysis show that carbon-blocking with
anodes, having high gravimetric capacity, superior cyclability, and ~20-nm thickness occurred between the G. As shown in Fig. 2f,
high tap density22–31. And previously reported graphite-alloy the presence of carbon-layers (~30 nm) on the external G surface
composites apparently have satisfactory values for several VDFs implies that most mesopores (~50 nm) could be filled by carbon-
to yield high volumetric capacity. However, to replace the con- blocking, because the carbon-layers formed on both G sides.
ventional graphite (G) anode entirely with regard to volumetric To further verify that the carbon-blocking suitably filled the
capacity, the graphite-alloy composite electrode swelling upon mesopores, the volume of pores with a diameter of 2–50 nm was
battery cycling should be rectified to become comparable to that investigated using the Barrett–Joyner–Halenda (BJH) method
of a conventional G electrode under industrial electrode condi- (Fig. 2g). Both the carbon and Si-layer coating on the G, required
tions (high areal capacity loading ≥3.5 mAh cm−2, high electrode for the MG and Si-layer-coated G (GS) synthesis processes,
density of 1.6 g cm−3, and limited binder, i.e., ≤3 wt%). Thus, respectively, reduced the G pore volume by a similar amount
there remains strong motivation to develop rationally designed (Supplementary Table 2). These results confirm that the
model to negate the electrode swelling of graphite-alloy compo- mesopores could be pre-filled through carbon-blocking before
site for high volumetric capacity. Si-layer deposition on the MG, as designed (Supplementary
In our previous in situ transmission electron microscopy Note 3). However, we observed that the Si-layer coating process
(TEM) analysis, we noted that the Si-layer volume expansion after pre-filling with carbon-blocking (for the MGS synthetic
within the mesopore pushes the Si toward the graphite during process) slightly reduced the pore volume in the mesopore range.

2 NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3 ARTICLE

a b
x<y

High

Al
loy
-b
Volumetric ca

as
Volumetric capacity

Volumetric capacity

ed
Be
n

an
efi
Gr

od
tr

e
ap

eg
ion
hit
High
pacit

e
Initial electrode density
y

Low
Loss
Low Hig region
h
w
Gra Lo ng
li
vim
etric el
cap sw 0% x y
acit Low de
y gh ro
Hi ect Electrode swelling ratio
El

c Si-
d
blocking Carbon-
blocking
Mesopore
(40 nm)

Macropore Pore
(500 nm) control
Si Si
9 µm
coating coating
G GS MG MGS

Porosity: 23% Porosity: 18.5%

Meso.:Macro. Meso.:Macro.
= 1562:1 = 0:0.88

e f
15 15
20 20
Thickness change (µm)

10.5 µm
10 10
15 15
5 5
Pouch cell
10 10

Substrate
5 5
G GS MGS
0 0
0 100 200 300 400 500
A Time (h)

Fig. 1 Critical factors influencing volumetric capacity and in situ thickness changes for MGS structural design. a Three-dimensional (3D) graph, showing a
correlation between volumetric capacity and other factors, including gravimetric capacity, electrode swelling ratio, and electrode density. b Volumetric
capacity curves versus swelling ratio for graphite and alloy-based anode when electrode density is fixed. The benefit (gray) region indicates that the
graphite-Si with specific electrode swelling (below y) has a higher volumetric capacity than that of graphite with value x. c, d Schematic model systemically
computed based on the experimentally analyzed results GS (c) and MGS (d). (The design description is given in Supplementary Fig. 2, Note 3, and Table 1.)
e Diagram of in situ thickness measurement system for pouch-type full-cell (G, GS, and MGS) upon cycling. f Electrode thickness change times of G, GS,
and MGS during 100 cycles. Inset: Two cycles (black arrows indicate thickness changes during charging and discharging) with error bars (±0.5 μm)

This phenomenon was also noted when artificial graphite (aG) Note 3), the majority of the Si located in the mesopores was
without internal pores was coated on the Si-layers (aGS); thus, the successfully moved to the macropores by manipulating the pore
reduced pore volume stems from the surface, not the mesopore distribution using carbon-blocking. Therefore, Si-layer pushing of
inside G (Supplementary Figs. 6 and 7). the G within the mesopores (Si-blocking) is completely prevented
We performed mercury porosimetry to demonstrate the Si- in MGS. Further, MGS exhibits high tap density, low porosity,
layer coating on the macropores. Figure 2h indicates that the MG and low specific surface area because of the carbon-blocking
pore volume decreased dramatically following Si-layer coating, (Fig. 2i and Supplementary Table 2). These outstanding powder
implying that most Si-layers filled the macropores. As the overall characteristics (low porosity and specific surface area) induce side
Si content of MGS is similar to that of GS (Supplementary reaction reduction during electrochemical reactions, yielding a

NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3

a b c
G Mesopore MG MGS
Si-layers

Macropore

d e
Carbon-blocking Si-layers in the macropore
• Blocking of Si coating • Electrode swelling decrease
• Surface area reduction • Stress relief Graphite
• Strong adhesion Carbon
-blocking
• Stable SEI formation
• Energy density improvement Graphite
(002)
0.34 nm

f
Graphite

Carbon-layers

Si-layers

g 0.03 h i 26
G GS 0.20 G GS Specific surface area
MG MGS MG MGS Porosity
6
24

0.15
Pore volume (cm3 g–1)

Pore volume (cm3 g–1)

Surface area (m2 g–1)

0.02 5 22

Porosity (%)
0.10 20
4

0.01
18
0.05 3

16
2
0.00 0.00
10 20 30 40 50 0 500 1000 G GS MG MGS
Pore diameter (nm) Pore diameter (nm)

Fig. 2 MGS fabrication and characterization. Cross-sectional SEM images of G (a), MG (b), and MGS (c), after ion beam milling, with insets showing top
views of SEM images. d Cross-sectional schematic illustration of MGS providing detailed characteristics of each component. e, f High-resolution TEM
images at inner region (e) and surface (f) of MGS with fast Fourier transform inset images. The white line indicates the (002) plane of graphite spanning
an inter-layer distance of 0.34 nm. The white dotted lines indicate the boundaries between the graphite, carbon-blocking, and Si. Mesopore size
distribution determined via BJH method (g) and macropore size distribution obtained through mercury intrusion porosimetry (h) of G (black), GS (green),
MG (blue), and MGS (red). i Specific surface area (black) and porosity (red) of G, GS, MG, and MGS. Scale bars, 1 μm (a–c), 5 μm (inset in a–c), 5 nm (e),
10 nm (f)

high initial Coulombic efficiency (CE) of 93% and stable shown in Fig. 3a–c, the initial G thickness (63 μm) exceeded that
cyclability32,33. of GS (43 μm) and MGS (43 μm) under industrial electrode con-
ditions. After 100 cycles, the GS-containing electrode revealed a
Electrode swelling testing and analysis via simulation. The 35% (15 μm) swelling ratio, approximately twice the 17% (11 μm)
electrode (G, GS, and MGS anodes) thickness variation in full-cell value for the G electrode (Fig. 3d, e). However, the MGS electrode
configuration was examined via SEM before and after 100 cycles exhibited only 19% (8 μm) swelling after 100 cycles (Fig. 3f).
in the fully lithiated state (detailed electrode information and To investigate the influence of lithiated Si on the electrode
electrochemical conditions are given in the Methods section). As swelling, we calculated the Si distribution of each sample in

4 NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3 ARTICLE

a b c

63 µm 43 µm 43 µm

Cu Cu
Cu

G bare electrode GS bare electrode MGS bare electrode

d e f

74 µm 58 µm 51 µm

Cu Cu Cu

G after 100 cycles GS after 100 cycles MGS after 100 cycles

g Si-blocking h High
GS
stress

Te
Lithiation ns
ile

Macro
pore

i MGS Carbon- j
blocking
R

r
Lithiation
Macro
pore Low
stress

k 30 l 120
GS
20 40 GS
Expansion

MGS 100
ratio (%)
Volume increase (%)

MGS
Stress,  (GPa)

10 20
80
0
60 0
Meso Macro
–10
40
–20 GS
20 MGS
–30

–40 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Normalized distance (r /R ) Lithiation state (x of Lix Si)

Fig. 3 Series of cross-sectional SEM images of electrode thickness differences before and after 100 cycles with FEM of GS and MGS upon lithiation.
a–f Cross-sectional views of G, GS, and MGS electrodes before cycling (a–c) and after 100 cycles (d–f), respectively. Finite element model for GS (g) and
MGS (i), where photographs show each particle. Diffusion-induced hoop stress in lithiation state (h, j). The red dotted line indicates the diagonal path.
k Comparison of hoop stress along with diagonal path. l Representative volume increase versus lithiation state. Inset: Expansion ratio depending on
different pore types. Scale bars, 50 μm (a–f) and 5 μm (g, i)

accordance with the numerical values from the experimentally blocking in the mesopores strongly contributed to the electrode
analyzed results, e.g., those for the pore volume by BJH, porosity, swelling, as it pushed directly toward the G surface. However, no
bulk density, true density and pure Si capacity (Supplementary Si-blocking occurred in the MGS mesopores; the MGS contained
Note 3). For the GS, Si (total weight fraction in GS is 6.3 wt% Si (the same amount of Si in GS) in the macropores (4.8 wt%, 76
indicating 100 vol% as total Si volume) was distributed in the vol%) and on the surface (1.5 wt%, 24 vol%). Note that a large
mesopores (2.6 wt%, 40.5 vol%), macropores (2.9 wt%, 46.6 vol%), amount of Si in the macropores yields low electrode swelling,
and on the surface (0.8 wt%, 12.9 vol%). We believe that the Si- because the Si volume expansion is highly mitigated by the

NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3

macropore vacant space during lithiation. Nevertheless, the Figs. 9, 13 and 16). After 100 cycles, some G cracks and Si-
reason why MGS electrode swelling ratio slightly exceeds than blocking fracturing were observed, as shown in the SEM images
that of G is that the one Si-layer on the external surface is likely to of the GS electrode (Fig. 4e). Furthermore, the TEM images
influence the electrode swelling (Supplementary Fig. 4). For more confirm severe Si-blocking fracturing, and related elemental
details, we conducted numerical simulations using the finite mapping images show fluorine and oxygen elements, which are
element method (FEM) at single-particle level to further elucidate evidence of SEI-layer39–41 (Fig. 4f). Since additional SEI layer can
the electrode swelling (detailed description of numerical simula- occur at the newly exposed, the G cracks and Si-blocking
tions is in the Methods section). fracturing may induce a continuous SEI-layer during cycling,
Schematic illustrations of pristine GS and MGS particle and leading to the low CE and severe capacity fading. On the other
cross-sectional SEM images are shown in Fig. 3g, i. Computation hand, the Si-layers in the MGS macropores preserved its
of the diffusion-induced stress based on two-phase lithiation of morphological integrity without physical damage to the G
Si34 with concentration-dependent properties35,36 indicates that (Fig. 4i). Therefore, a relatively stable SEI-layer was preserved
tensile stress (red region) was strongly applied to the GS at the on the Si-layers, which maintained attachment to the G (Fig. 4j),
core and the Si-blocking boundary in the fully lithiated (x = 3.75, ensuring high CE and cycle stability (see in situ TEM analysis of
LixSi) state (Fig. 3h, k and Supplementary Movie 1). In contrast, Si volume changes in mesopore and macropore during lithiation
there was little tensile stress in the lithiated MGS particle (Fig. 3j, in ref. 23).
k and Supplementary Movie 2). The corresponding volume For detailed understanding of the Si-blocking behavior in the
increase was also calculated through numerical simulation. As mesopores, we simulated the lithiation/delithiation process of the
shown in Fig. 3l, MGS exhibited a low volume increase ratio mesopore Si-blocking through FEM. Noting that the critical
arising from stress relaxation due to the macropore vacant space. fracture energies for Si and G crack formation are reported to be 9
However, the GS volume increase ratio far exceeded that of MGS and 17–69 J m−2, respectively42,43, we examined the possible
because of the high Si-blocking volume expansion in the growth of a crack applied to the Si and G adjacent to the Si-
mesopores. Therefore, stress relaxation due to the accommoda- blocking. The calculated Si and G fracture energies with a pre-
tion of the Si volume expansion by the macropore empty space existing 5 nm-sized crack in GS during lithiation were 10.44 and
induces a low electrode swelling ratio (additional TEM images are 71.04 J m−2, respectively (6.88 and 3.80 J m−2, respectively, in
given in Supplementary Fig. 17). MGS); thus, results exceeding the critical values indicate crack
growth during the lithiation/delithiation process. Furthermore,
compressive stress (blue region) was strongly applied to the G
Morphological changes and SEI-layer formation. To allow around the Si-blocking in the lithiated state, suggesting that the Si
comprehensive understanding of both the Si-blocking in the was strongly attached to the bottom and upper G, as this region
mesopores (GS) and the Si-layers in the macropores (MGS) functions as a cohesion zone (Fig. 4k). For delithiation, the stress
during cycling, the morphological changes and SEI-layer forma- distribution was opposite to the lithiation case. Consequently,
tion are schematically shown in Fig. 4a, b. Briefly, when the Si- tensile stress, which functions as a crack expander in delithiation,
blocking expands towards the G upon lithiation, it applies was applied to both Si-blocking sides (Fig. 4l, m). These results
extremely high tensile stress on the G, inducing crack formation. coincide with the Fig. 4a illustration.
The stresses allow the Si-blocking to attach strongly to the bottom
and upper side of the G during lithiation. The adhesive Si
maintains this contact upon contraction during delithiation, Electrochemical characterization considering electrode swel-
causing Si fracturing. After many cycles, the G crack formation ling. To confirm MGS anode practical viability, electrochemical
and Si-blocking fracturing induces electrical contact loss and tests were performed in half- and full-cell configuration under
newly exposes a fresh surface to the electrolyte, causing con- industrial electrode conditions (see the additional data in Sup-
tinuous SEI-layer formation. Previous studies have reported Si plementary Figs. 11, 12 and Supplementary Note 4). The rever-
nanoparticle critical sizes of approximately 150 and 300 nm, and sible capacities of the synthesized MGS, GS, and conventional G
Si fracturing can be avoided under these sizes37,38. Although <50 were measured using the constant current and constant voltage
nm is far smaller than the Si critical size, isolated Si-blocking may technique, from 0.005 to 1.5 V (first cycle in half-cell). The MGS,
cause morphological deformation and fracture upon cycling GS, and G exhibited first-cycle gravimetric reversible capacities of
(detailed interpretation about the reason of morphological 527, 525, and 360 mAh g−1, respectively, with initial CEs of
deformation and fracture of Si below 50 nm are given in 93.0%, 92.2%, and 92.0%, respectively (Fig. 5a). Based on the half-
Fig. 4k–m). On the other hand, the Si-layers in the macropore, cell electrochemical results, we designed a pouch-type full-cell
having sufficient space to accommodate the volume expansion, containing MGS anode and LiCoO2 (LCO) cathode (Supple-
generate far less tensile stress on the G upon lithiation. Therefore, mentary Fig. 10). The anode and cathode areal charge capacities
no G crack formation and Si-layer fracturing occur during were fixed at 3.8 and 3.5 mAh cm−2, respectively, yielding a
cycling, yielding a relatively stable SEI-layer on the Si and good theoretical N/P ratio of 1.1. As shown by the normalized capacity
Si-to-G electrical contact. profiles (Fig. 5b), MGS revealed an initial CE of 91.3% with a 3.16
Ex situ TEM and SEM images were obtained before and after mAh cm−2 discharge capacity, exceeding that of G (3.11 mAh cm
100 cycles (delithiated state) in the full-cell configuration, to −2 with 90.0%) for the first cycle within a voltage range of
confirm both the morphological changes and SEI-layer. As shown 2.5–4.35 V (detailed results of full-cell comprising MGS and
in Fig. 4c, d, Si-blocking with various thicknesses (≤40 nm) were LiNi0.6Co0.2Mn0.2O2 are given in Supplementary Fig. 14). The
found inside the GS particle. Si-layers with ~20-nm thickness low specific surface area, which reduced the side reaction during
were on the surface; thus, Si-blocking (~40 nm) could be formed electrochemical reaction, caused the increased initial CE of MGS.
in the mesopores because the Si-layers were coated on both G The volumetric capacity was defined as the discharge capacity
sides. Further, Si-layers with ~30-nm size were observed inside divided by the maximum thickness during cycling, which was
the MGS macropores (Fig. 4g, h, Supplementary Figs. 8 and 18). determined through in situ thickness measurement for 100 cycles.
As the Si had the same weight percent in the MGS and GS, the As shown in Fig. 5c, MGS exhibited a higher initial volumetric
MGS Si-layers were thicker than those of the GS because of the capacity (632.0 mAh cm−3) than GS (588.6 mAh cm−3). Even
intensively coated Si-layers in the macropores (Supplementary after 100 cycles, the MGS volumetric capacity (493.9 mAh cm−3)

6 NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3 ARTICLE

a
Graphite cracking Continuous SEI-layer

Lithiation

lithiation

cycles
Con-

Many
Si Expansion

De-
traction
Graphite High stress Morphological change Poor electrical contact
Si-blocking in mesopore
b
Thin SEI-layer

Low
stress Good electrical contact
Si-layers in macropore
c d e f
Si-blocking Graphite cracking
Si-blocking
Fracture of Si
F

Si Si O
Bare GS After 100 cycles
g h i j
Cavity
Carbon-
blocking
F

Si Si O

Bare MGS After 100 cycles

k l m 30
Expanding (lithiation) Contracting (delithiation) GS
Compressive stress
20
er
Stress,  (GPa)
Si

p
de

Up
R 10
r
0
Si

m
tto
de

Bo
Tensile stress
–10

High stress Low stress 0.0 0.2 0.4 0.6 0.8 1.0
Normalized distance (r /R)

Fig. 4 Structural changes and SEI formation of GS and MGS before and after 100 cycles with FEM of GS upon delithiation. a, b Schematic of SEI formation of
Si in mesopore (a) and macropore (b). c–j Cross-sectional SEM and scanning TEM (STEM) images after ion beam milling with elemental mapping through
energy-dispersive spectroscopy of GS and MGS electrode before cycling (c, d, g, h) and after 100 cycles (e, f, i, j), respectively. k, l Estimation of structural
change based on finite element results during lithiation and delithiation. m FEM of diffusion-induced hoop stress along with diagonal path in GS delithiation
state. Scale bars, 1 μm (c, e, g, i) and 50 nm (d, f, h, j)

exceeded that of GS (367.5 mAh cm−3), because of the low Although the GS seemed to exhibit higher volumetric capacity
swelling ratio with improved cyclic stability (Figs. 1f and 5e, (588.6 mAh cm−3) than G (420.2 mAh cm−3) before cycling,
Supplementary Fig. 15). Moreover, MGS exhibited a remarkable the GS value (367.5 mAh cm−3) became similar to that of G
CE increase exceeding 99.5% after only 4 cycles, whereas the CE (361.4 mAh cm−3) after 100 cycles due to high electrode volume
of GS was only reached after 10 cycles. These results imply that expansion with poor cycle retention. These results imply that
the macropores relieve the stress due to Si volume expansion, volumetric capacity should be considered based on electrode
allowing the formation of a stable SEI-layer during cycling. swelling and capacity retention after cycling. As shown in Fig. 5d,

NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3

a 1.5 b 4.4
G
GS 4.0
MGS
1.0
Voltage (V)

Voltage (V)
3.6

3.2
0.5 G
2.8 GS
MGS
0.0 2.4
0 100 200 300 400 500 600 0 20 40 60 80 100
Capacity (mAh g–1) Nomalized capacity (%)

c d
2400
Volumetric capacity (mAH cm–3)

1200 100 G

Coulombic efficiency (CE)

Energy density (Wh L–1)

2200 GS
1000 MGS
99.5 98
2000
800
0 2 4 6 8 10 96 1800
600

400 1600
94
G 1400
200
GS MGS
0 92 1200
0 20 40 60 80 100 1 20 40 60 80 100
Cycle number Cycle number

e
10 10
capacity (mAH)
Discharge

9 9

8 GS 8
MGS
7 7

0 25 50 75 100 125 150 175 200

Cycle number

Fig. 5 Electrochemical characterization of various anodes. Voltage profiles of G, GS, and MGS half-cell (a) and full-cell (b) plotted for first cycles at 0.1 C.
c Full-cell volumetric capacity as reversible areal capacity divided by maximum electrode thickness at each cycle for 100th cycle at 1 C discharge rate and
0.5 C charge rate in 2.7–4.35 V potential range. The mass loading of the GS and MGS was 6.9 mg cm−2 (G: 10.1 mg cm−2). The CE of each sample is
plotted on the secondary y-axis. Inset: Magnified graphs of CE with 99.5 line for 10 cycles to indicate the number of cycles for CE to reach 99.5%.
d Volumetric energy densities for 1st, 20th, 40th, 60th, 80th, and 100th cycles. e Full-cell discharge capacity of GS and MGS paired with conventional
lithium cobalt oxide. The mass loading of the GS and MGS was 3 mg cm−2

we proved that MGS has an outstanding energy density of 1825.7 These outstanding features eventually demonstrated a volumetric
Wh L−1 being higher than those of G and GS (1376.3 and 1374.5 capacity of 493.9 mAh cm−3 and energy density of 1825.7 Wh L−1
Wh L−1, respectively). Detailed electrode and energy density exceeding those of G (361.4 mAh cm−3 and 1376.3 Wh L−1) after
information is given in Supplementary Table 3. 100 cycles. Further, our material design satisfies the most rigorous
requirements concerning high-energy density, being sufficiently
large to replace a conventional G anode even after cycling. Thus,
Discussion the exquisitely designed MGS paves the path for next-generation
We introduced strategically designed MGS through pore- anodes with both high-energy density and commercial feasibility;
distribution manipulation, based on a calculation model from this could also be a significant breakthrough for the development
experimentally analyzed results with insight into the vacant space of an electric vehicle with more extended driving mileage of
in composite. Interestingly, the developed MGS exhibited a several times than before on a single charge.
minimized electrode swelling ratio (19%) comparable to that of
conventional G (17%) even after 100 cycles in full-cell config- Methods
uration under industrial electrode fabrication conditions (≥ 3.1 Synthesis of MGS. For the fabrication of MGS, 5 kg of spherical graphite was
mAh cm−2, 1.6 g cm−3). In addition to the high initial CE of introduced into a rotary tube furnace. To fill mesopore with carbon-blocking,
93.0% and 527 mAh g−1, a rapid cycling efficiency increase thermal decomposition of ethylene gas (99.9%) was executed at 900 °C for 3 h
(5 L min−1). In succession, high-purity monosilane gas (99.9999%) flowed in the
exceeding 99.5% over 4 cycles was exhibited, because of the crack- same furnace at 475 °C for 1 h (5 L min−1). In case of all samples for electro-
and contact-loss-free morphological integrity during cycling. chemical test, additional carbon coating was performed for 2 h (5 L min−1).

8 NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3 ARTICLE

Material characterization. The samples were visualized using scanning electron Received: 6 July 2018 Accepted: 21 December 2018
microscopy (Verios 460, FEI) with energy-dispersive spectroscopy (XFlash 6130,
Bruker) and high-resolution transmission electron microscopy (JEM-2100F, JEOL).
Sample preparation for the cross-sectional view was carried out using ion milling
system (IM-40000, Hitachi). Dual-beam focused ion beam (Helios 450HP, FER)
was used for cross-section view of TEM images. To observe the volume expansion References
of the electrode at lithiated state after cycling, the cells are disassembled, and the 1. Obrovac, M. N. & Chevrier, V. L. Alloy negative electrodes for Li-ion batteries.
electrodes are rinsed with dimethyl carbonate in a dry room. Specific surface area Chem. Rev. 114, 11444–11502 (2014).
and mesopore diameter distribution were estimated with the 2. Sun, Y. et al. Promises and challenges of nanomaterials for lithium-based
Brunauer–Emmett–Teller (BET) method using the nitrogen adsorption–desorption rechargeable batteries. Nat. Energy 1, 16071 (2016).
analyzer (TriStar II, Micromeritics). Prior to measurement, the samples were 3. Choi, J. W. & Aurbach, D. Promise and reality of post-lithium-ion batteries
degassed at 120 °C for 2 h. And porosity and macropore diameter distribution were with high energy densities. Nat. Rev. Mater. 1, 16013 (2016).
determined by mercury-porosimetry (Autopore V9500, Micromeritics). 4. Chae, S. et al. Confronting issues of the practical implementation of Si anode
in high-energy lithium-ion batteries. Joule 1, 47–60 (2017).
5. Dahn, J. R. et al. Mechanisms for lithium insertion in carbonaceous materials.
Electrochemical characterization. Working electrode was prepared by mixing Science 270, 590–593 (1995).
slurry composed of the active material (G, GS, MGS), the conductive agent (Super 6. Yoshio, M. et al. Improvement of natural graphite as a lithium-ion battery
P, TIMCAL), sodium carboxymethylcellulose (CMC) and styrene butadiene rubber anode material, from raw flake to carbon-coated sphere. J. Mater. Chem. 14,
(SBR) at mass ratio of 96:1:1.5:1.5 and then the homogeneously blended slurry was 1754–1758 (2004).
cast onto the Cu current collector up to 6.9 mg cm−2. In the case of graphite, the 7. Zheng, H. et al. Electrochemical intercalation of lithium into a natural graphite
loading level was 10.1 mg cm−2 for same areal capacity. All electrodes were dried at anode in quaternary ammonium-based ionic liquid electrolytes. Carbon 44,
80 °C for 0.5 h and then calendared for 1.6 g cm−3 of electrode density with the 203–210 (2006).
electrode thickness of 43 μm of the MGS and GS electrode and 63 μm of the G 8. Lu, J. et al. The role of nanotechnology in the development of battery materials
electrode excepting the Cu current collector. In sequence, the electrode was finally for electric vehicles. Nat. Nanotechnol. 11, 1031 (2016).
vacuum-dried at 110 °C for 8 h. The cathode electrode was made by casting slurry 9. McDowell, M. T. et al. 25th anniversary article: Understanding the lithiation
on a Al current collector with conventional lithium cobalt oxide, carbon black, and of silicon and other alloying anodes for lithium-ion batteries. Adv. Mater. 25,
polyvinylidene fluoride binder in a mass ratio of 96:2:2. The mass loading level of
4966–4985 (2013).
the cathode was 20 mg cm−2 and then pressed until the density of the electrode
10. Kumar, R. et al. In situ and operando investigations of failure mechanisms of
became 3.6 g cm−3 with the electrode thickness of 55 μm excepting for the Al
the solid electrolyte interphase on silicon electrodes. ACS Energy Lett. 1,
current collector. N-methyl-2-pyrrolidone (NMP) was used as the solvent. CR2032
689–697 (2016).
(half-cell) and pouch (full-cell) type cells were assembled in the dry room using
these working electrodes. The electrolyte was 1.3 M LiPF6 in mixture of ethylene 11. Erickson, E. M. et al. Review—development of advanced rechargeable
carbonate/ethyl methyl carbonate/diethyl carbonate (3/5/2, by volume) with 10% of batteries: a continuous challenge in the choice of suitable electrolyte solutions.
fluoroethylene carbonate, 0.2% of lithium tetrafluoroborate, 0.5% of vinylene car- J. Electrochem. Soc. 162, A2424–A2438 (2015).
bonate, 3% of succinonitrile, and 1% of propane sultone (Panax Starlyte) and 12. Yao, Y. et al. Interconnected silicon hollow nanospheres for lithium-ion
microporous polyethylene was used as a separator with a thickness of 20 μm. As a battery anodes with long cycle life. Nano Lett. 11, 2949–2954 (2011).
counter electrode, pure Li metal foil (1 mm) was used for half-cell, and LCO was 13. Park, M.-H. et al. Silicon nanotube battery anodes. Nano Lett. 9, 3844–3847
utilized for full-cell. Electrochemical tests of the half-cell were carried out in a (2009).
voltage range of 0.005–1.5 V at 0.35 mA cm−2 (0.1 C) for the first cycle, and 14. Chan, C. K. et al. High-performance lithium battery anodes using silicon
between 0.005 and 1.0 V at 1.75 mA cm−2 (0.5 C) for the rest of the cycles. Elec- nanowires. Nat. Nanotechnol. 3, 31 (2007).
trochemical performances of the full-cell, designed with an N/P ratio of 1.1, were 15. Wu, H. et al. Stable cycling of double-walled silicon nanotube battery
evaluated in the voltage range between 2.5 and 4.35 V at 0.1 C for the first cycle, anodes through solid–electrolyte interphase control. Nat. Nanotechnol. 7, 310
and between 2.7 and 4.35 V at a discharge rate of 1 C and charge rate of 0.5 C for (2012).
the rest of the cycles. The electrolytes and separator in the full-cell were the same as 16. Liu, N. et al. A yolk-shell design for stabilized and scalable Li-ion battery alloy
those in the half-cell above. All the cell tests were done using a battery cycler anodes. Nano Lett. 12, 3315–3321 (2012).
(TOSCAT-3100, TOYO SYSTEM) at 25 °C. 17. Liu, N. et al. A pomegranate-inspired nanoscale design for large-volume-
change lithium battery anodes. Nat. Nanotechnol. 9, 187–192 (2014).
18. Su, X. et al. Silicon-based nanomaterials for lithium‐ion batteries: a review.
Dilatometry. Thickness change of pouch (full-cell) type cell during 100 cycles was Adv. Energy Mater. 4, 1300882 (2014).
measured using an electrochemical dilatometer (Mitutoyo). All electrochemical 19. Jin, Y. et al. Challenges and recent progress in the development of Si anodes
conditions were the same as those in the full-cell above. for lithium-ion battery. Adv. Energy Mater. 7, 1700715 (2017).
20. Chevrier, V. L. et al. Evaluating Si-based materials for Li-ion batteries in
commercially relevant negative electrodes. J. Electrochem. Soc. 161,
Numerical simulations. For finite element simulations, we used the commercial A783–A791 (2014).
software ABAQUS/Standard 6.14 (Dassault System). Thermal-mechanical coupled 21. Dimov, N. et al. Practical silicon-based composite anodes for lithium-ion
model that is equivalent to the lithium-diffusion model was utilized to describe the batteries: fundamental and technological features. J. Power Sources 171,
volume expansion and contraction of GS and MGS particles during lithiation and 886–893 (2007).
delithiation processes. Two-phase lithiation model of Si34 and linear elastic 22. Wang, C. et al. Lithium insertion in carbon–silicon composite materials
materials with concentration-dependent mechanical properties which had been produced by mechanical milling. J. Electrochem. Soc. 145, 2751–2758 (1998).
measured by experiments35,36 were adopted in all simulations. To reduce com- 23. Ko, M. et al. Scalable synthesis of silicon-nanolayer-embedded graphite for
putational cost, we employed circular-shaped models of which dimensions were high-energy lithium-ion batteries. Nat. Energy 1, 16113 (2016).
smaller than the GS and MGS particles in the experiment. Radiuses of the particles, 24. Xu, Q. et al. Watermelon-inspired Si/C microspheres with hierarchical buffer
macropores, and mesopores were 1100, 500, and 20 nm, respectively. Thicknesses structures for densely compacted lithium-ion battery anodes. Adv. Energy
of the Si-layer on macropores in the GS and MGS particles were 20.0 and 36.8 nm, Mater. 7, 1601481 (2017).
respectively, so that the total weight percent (wt%) of Si for both GS and MGS 25. Holzapfel, M. et al. Chemical vapor deposited silicon/graphite compound
particles was the same. In static fracture simulations, we calculated fracture ener- material as negative electrode for lithium-ion batteries. Electrochem. Solid
gies (i.e., J-integrals) on five independent pathways encircling a pre-existing nan- State Lett. 8, A516–A520 (2005).
ometer-sized crack (5 nm) on the graphite side at the interface of the mesopores 26. Zhang, Y. et al. Composite anode material of silicon/graphite/carbon
and compared with critical fracture energies (i.e., fracture toughness) of the cor- nanotubes for Li-ion batteries. Electrochim. Acta 51, 4994–5000 (2006).
responding materials. We note that we intentionally set the models stress-free at 27. Lee, J.-H., Kim, W.-J., Kim, J.-Y., Lim, S.-H. & Lee, S.-M. Spherical silicon/
the beginning of lithiation and delithiation processes during the fracture simula- graphite/carbon composites as anode material for lithium-ion batteries. J.
tions, and thus the residual stress field resulting from the lithiation process was not Power Sources 176, 353–358 (2008).
considered when we started the delithiation process. We did not conduct crack 28. Martin, C. et al. Graphite-grafted silicon nanocomposite as a negative
propagation simulations. electrode for lithium-ion batteries. Adv. Mater. 21, 4735–4741 (2009).
29. Yoon, Y. S., Jee, S. H., Lee, S. H. & Nam, S. C. Nano Si-coated graphite
composite anode synthesized by semi-mass production ball milling for lithium
Data availability secondary batteries. Surf. Coat. Technol. 206, 553–558 (2011).
The authors declare that the data supporting the findings of this study are available 30. Gan, L. et al. A facile synthesis of graphite/silicon/graphene spherical
within the article and its Supplementary Information Files. All other relevant data composite anode for lithium-ion batteries. Electrochim. Acta 104, 117–123
supporting the findings of this study are available on request. (2013).

NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-018-08233-3

31. Xiao, C. et al. Walnut-structure Si–G/C materials with high Coulombic Author contributions
efficiency for long-life lithium ion batteries. RSC Adv. 8, 27580–27586 (2018). J.M., J.S., and M.K. conceived and designed the experiments. J.M. and J.S. prepared the
32. Yoshio, M. et al. Spherical carbon-coated natural graphite as a lithium-ion samples and carried out the main experiments including construction of models for G,
battery-anode material. Angew. Chem. Int. Ed. 115, 4335–4338 (2003). GS, and MGS. J.H. performed structure and fracture analyses using FEM. S.C. conducted
33. Li, H. & Zhou, H. Enhancing the performances of Li-ion batteries by carbon- dilatometry. N.K. participated in electrochemical measurements. S.-H.C. and G.N.
coating: present and future. Chem. Commun. 48, 1201–1217 (2012). assisted with sample preparation. Y.S. provided experimental insights into the electro-
34. Pharr, M. et al. Kinetics of initial lithiation of crystalline silicon electrodes of chemical test. J.M., J.S., J.H., S.Y.K., M.K., and J.C. co-wrote the paper. J.M., J.S., J.H., S.Y.
lithium-ion batteries. Nano Lett. 12, 5039–5047 (2012). K., M.K., and J.C. discussed the results and revised or commented the manuscript.
35. Choi, Y. S. et al. A simple technique for measuring the fracture energy of
lithiated thin-film silicon electrodes at various lithium concentrations. J.
Power Sources 294, 159–166 (2015).
Additional information
Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467-
36. Qi, Y. et al. Threefold increase in the Young’s modulus of graphite negative
018-08233-3.
electrode during lithium intercalation. J. Electrochem. Soc. 157, A558–A566
(2010).
Competing interests: The authors declare no competing interests.
37. Liu, X. H. et al. Size-dependent fracture of silicon nanoparticles during
lithiation. ACS Nano 6, 1522–1531 (2012).
Reprints and permission information is available online at http://npg.nature.com/
38. McDowell, M. T. et al. Studying the kinetics of crystalline silicon nanoparticle
reprintsandpermissions/
lithiation with in situ transmission electron microscopy. Adv. Mater. 24,
6034–6041 (2012).
39. Nie, M. et al. Silicon solid electrolyte interphase (SEI) of lithium ion battery Journal peer review information: Nature Communications thanks the anonymous
characterized by microscopy and spectroscopy. J. Phys. Chem. C 117, reviewers for their contribution to the peer review of this work. Peer reviewer reports are
13403–13412 (2013). available.
40. Nguyen, C. C. & Lucht, B. L. Comparative study of fluoroethylene carbonate
and vinylene carbonate for silicon anodes in lithium ion batteries. J. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
Electrochem. Soc. 161, A1933–A1938 (2014). published maps and institutional affiliations.
41. Schroder, K. et al. The effect of fluoroethylene carbonate as an additive on the
solid electrolyte interphase on silicon lithium-ion electrodes. Chem. Mater. 27,
5531–5542 (2015). Open Access This article is licensed under a Creative Commons
42. Shi, F. et al. Failure mechanisms of single-crystal silicon electrodes in lithium- Attribution 4.0 International License, which permits use, sharing,
ion batteries. Nat. Commun. 7, 11886 (2016). adaptation, distribution and reproduction in any medium or format, as long as you give
43. Etter, T. et al. Strength and fracture toughness of interpenetrating graphite/ appropriate credit to the original author(s) and the source, provide a link to the Creative
aluminium composites produced by the indirect squeeze casting process. Commons license, and indicate if changes were made. The images or other third party
Mater. Sci. Eng. A 386, 61–67 (2004). material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
Acknowledgements the copyright holder. To view a copy of this license, visit http://creativecommons.org/
This work was supported by the Korea Institute of Energy Technology Evaluation and licenses/by/4.0/.
Planning (KETEP) and the Ministry of Trade, Industry & Energy (MOTIE) of the
Republic of Korea (No. 20172410100140). Also, financial support from the 2019
Research Fund of UNIST is greatly acknowledged. © The Author(s) 2019

10 NATURE COMMUNICATIONS | (2019)10:475 | https://doi.org/10.1038/s41467-018-08233-3 | www.nature.com/naturecommunications