Article
https://doi.org/10.1038/s41586-019-1538-z
Electrochemically reconfigurable
architected materials
Xiaoxing Xia
1
, Arman Afshar
2
, Heng Yang
1
, carlos M. Portela
1
, Dennis M. Kochmann
1,3
, claudio V. Di leo
2
& Julia r. Greer
1
*
1
Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA.
2
School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, USA.
3
Department of Mechanical and Process Engineering, ETH Zurich, Zurich, Switzerland.
*
e-mail:
jrgreer@caltech.edu
SUPPLEMENTARY INFORMATION
https://doi.org/10.1038/s41586-019-1538-z
In the format provided by the authors and unedited.
12 S
eP
te
MB
er
2019 | VO
l 573 | NA
tU
re
| 205
Supplementary
Information:
Electrochemically
Reconfigurable Architected Materials
Xiaoxing Xia
1
, Arman Afshar
2
, Heng Yang
1
, Carlos M. Portela
1
,
Dennis M. Kochmann
1,
3
,
Claudio V. Di Leo
2
, Julia
R.
Greer
1
*
1
Division of Engineering and Applied Science, California Institute of Tech
nology
, Pasadena,
CA 91125, United States
2
School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United
States
3
Department of
Mechanical and Process Engineering, ETH Zurich, 8092 Zürich, Switzerland
This Supplementary Information document includes:
Section I. Si Microlattice Design and Fabrication
Section II. Electrochemical Testing Method
Section III.
In situ
Observation
of Lith
iation
-
induced Cooperative Buckling
Section IV. Long
-
term Cycling Performance
and Discussion
Section V. Sn Microlattice Fabrication and Lithiation
Section VI. Fabrication Defects and Artificial Defects
Section VII.
Buckling
Domain Map Processing
Section
VIII. Monte Carlo Simulation of the Square
-
lattice Antiferromagnetic Ising Model
Section IX. Coupled Chemo
-
Mechanical Finite Element Analysis
Section X. Reduced
-
order
Chemo
-
Mechanical Model
Section XI. Phononic Dispersion Relation Simulation
Section XII.
Comparison of Reconfiguration Mechanisms for Architected Materials
Additional Supplementary Videos available online:
Supplementary Video 1.
In situ
lithiation of a Si microlattice at a constant current
Supplementary Video 2.
In situ
delithiation of a Si m
icrolattice at a constant current
Supplementary Video 3.
In situ
lithiation of a Si microlattice with a resistor load
Supplementary Video 4.
In situ
cycling of a Si microlattice at high rates
Supplementary Video 5.
In situ
lithiation of a Si microlattice w
ith
programed
artificial defects
Supplementary Video 6.
FEA simulation of a 3D beam that buckles upon lithiation
Supplementary Video 7.
FEA simulation to compare different deformation mechanisms
Supplementary Video 8.
FEA simulation to compare
beams with different slenderness ratios
Supplementary Video 9.
FEA simulation of cooperative buckling of 2D extended unit cells
I.
Si Microlattice
Design and
Fabrication
T
etragonal lattice
s
with 20
μm × 20
μm ×
5
μm (
in
x
,
y
, and z
-
axis respectively
) unit
cells
are
designed in MATLAB
and
imported into a commercial two
-
photon lithography system (
Photonic
Professional GT,
Nanoscribe GmbH). Each sample
is consisted of a 10 ×
10 array of stitch
ed
smaller
lattices written sequentially due to the limited writing
area of the two
-
photo lithography
system. Each s
maller tetragonal lattice has 8 × 8 ×
5 unit cells, and stitch
ed lattices overlap by
one unit cell
. Therefore, each sample has 79 × 79 ×
5 unit cells in total written on a cleaned glass
coverslip substrate (1
8
mm diameter circular No. 2 glass, VWR) with a custom
-
made
photoresist.
This
negative photoresist is composed of 79.1
wt% Acrylo POSS monomer
(MA0736
,
Hybrid Plastics Inc.), 20
.0
wt% dichloromethane solvent (Sigma
-
Aldrich), and
0.9
wt% 7
-
diethylamino
-
3
-
th
enoylcoumarin photoinitiator (Luxottica Exciton)
, and it is placed
on top of the glass substrate
.
Immersion oil is used betwe
en the 63X
objective of the two
-
photon
lithography system and the bottom side of the glass substrate.
After two
-
photon lithography,
the
sample is developed in PGMEA (propylene glycol monomethyl ether acetate, Sigma
-
Aldrich) for
25
min and rinsed in IPA for three times before critical point drying. Each
polymer sample has
elliptically
cross
-
sectioned horizontal beams with a vertically aligned major axis of ~1.8
μm and
a minor axis of ~0.5
μm and cylindrical vertical posts with a diameter of ~1.8
μm with small
sample
-
to
-
sample
variation
s
due to two
-
photon lithography laser degradation.
The bottom layer
of the vertical post is extended to 10
μm
to assist twisting of the vertical posts during lithiation
,
and in the bottom 3
μm of the vertical posts, the diameter gradually increases to ~3.6
μm
to
enhance adhesion
with the substrate
.
The po
lymer samples are cleaned by oxygen plasma
and baked for 2
hr at 250
°
C in an Ar
-
filled
glovebox
before RF magnetron sputtering deposition of ~5
nm of Cr seed layer and ~100
nm of
Ni conductive layer on lattice beams (100
W, 20
sccm Ar flow, 5
mTorr deposition pressure,
AJA International, Inc.). The sputtered Ni film is thicker at the top of each
horizontal
beam and
thinner at the bottom of each
horizontal
beam.
Next, ~300
nm of amorp
hous Si (a
-
Si) is
deposited by plasma enhanced chemical vapor
d
eposition (PECVD, Oxford Instruments) at the
following condition
s
: 200
°
C temperature, 400
mTorr pressure, 250
sccm of 5% s
ilane in Ar
precursor gas flow and 10
W RF power. Finally, ~100
nm of Ni thin film is coated on the back of
the sample substrate by s
puttering with good electrical pathway to the Ni layer on top of the
substrate through good Ni coverage on the edge of the substrate. During two
-
photo
n
lithography,
a 5
μm square grid is written on the substrate underneath and 180
μm around the lattice
(bo
undary marked by red dotted lines in
Supplementary Fig.
1a). A 1.8
mm square shadow mask
is used during PECVD to limit the a
-
Si deposition to only the lattice section within the extent of
the square grid to prevent Si thin film delamination on the substrat
e (
mask
boundary marked by
green dotted lines in
Supplementary Fig.
1a, d).
Supplementary
Fig.
1d shows Si thin film
delamination when a section of the square grid is missing due to
an accidental
interface finding
error during two
-
photon lithography. Finally, non
-
contact support structures are added on the
outside of exterior vertical posts to prevent them from leaning outwards during Si microlattice
lithiation due to the absence of periodic bound
ary conditions (
Supplementary Fig.
1b, c
, e, f
).
The
total Si mass loading on a representative
sample is measured by Cahn C
-
35 microbalance to
be 8.
0±0.4
μg
by
mass
measurements before and after KOH etching of Si on the lattice. Part of
the substrate has t
o be cut
off
by a diamond pen to keep the total sample mass within the range
with 0.1
μg sensitivity so measur
ing Si mass for each sample before
electrochemical
testing is
not practical.
Variation of Si mass loading is
noticed
across samples due to two
-
pho
ton
lithography laser degradation and PECVD chamber condition
s
during Si deposition
. The areal Si
mass loading calculated from the area of the Si deposition shadow mask is ~0.25
mg/cm
2
.
The
theoretical capacity for each Si microlattice sample is ~29
μAh
based on Si’s theoretical specific
capacity
25
of 3600
mAh/g.
Supplementary Figure
1
.
SEM image
s
of Si m
icrolattice fabrication details (a
-
c) before
lithiation and (d
-
f) after lithiation.
(a) describes the boundaries of the square shadow mask
(marked by
green dotted lines) used during PECVD is in between the edges of the microlattice
and the edges of the square grid on the substrate (marked by red dotted lines). (b) shows Si thin
film delamination when a section of the square grid is missing due to interf
ace finding error
during two
-
photon lithography, which demonstrates the square grid is important for preventing
Si delamination on the substrate. (b, c, e, f) show non
-
contacting support structures on the
outside of exterior vertical posts that effectively
prevent them from leaning outwards during
lithiation despite the absence of periodic boundary conditions at the edges.
T
he rationale for choosing the specif
ic tetragonal lattice geometry is briefly discussed below.
The cross
-
sectional dimensions of indivi
dual beams were mainly dictated by the resolution of the
two
-
photon lithography
process; we chose the thickness of Si layer to be below the critical length
scale for fracture and delamination through so
-
called size effects in the mechanical properties of
S
i at small scales during lithiation and delithiation. The elliptical shape of the beam cross
-
section
with vertically aligned major axis constrains the lowest energy buckling modes to be in
-
plane
and also minimizes feature size because the writing voxel
in
two
-
photon lithography
is an
ellipsoid; beams with circular cross
-
sections require hatching, which expands their dimensions.
The ratio of length over
radius of gyration
of the horizontal beams defines the beams’
slenderness ratio and their propensity for b
uckling instabilities, which is analyzed in details in
Fig. 3. We chose
the
tetragonal lattice geometry (square lattice in the lateral plane) for its
simplicity in design and fabrication. We also fabricated other, higher
-
symmetry lattices with
equivalent b
eam dimensions and similarly adjoined and supported by vertical posts, such as
hexagonal and triangular lattices, as shown in Supplementary Fig. 2a
-
f. Upon lithiation, we
found the hexagonal lattice to buckle into an ordered geometry (Supplementary Fig. 2b
), closely
re
sembling
that
reported in ref. 23
, and the triangular lattice buckled into a “frustrated” geometry
(Supplementary Fig. 2e), similar to what is reported in ref. 15. We learned that these higher
-
symmetry lattices were more susceptible to fabrica
tion defects, for example stitching
inaccuracies during fabrication, as shown by the periodic distortions in zoomed
-
out SEM images
in Supplementary Fig. 2c, f. This is most probably because the large samples are stitched from
smaller lattices during two
-
ph
oton lithography in x and y directions, the effective defects due to
stitching are more pronounced for
lattices with higher symmetry
and
non
-
orthogonal coordinates.
This observation also illustrates the importance of defects in reconfigurable architected m
aterials.
The horizontal beams in tetragonal lattices with wider, 3.8
μm
-
diameter vertical posts, also
buckled cooperatively as a result of lithiation, but the domain boundaries had frequent overlaps
with periodic stitching sites (Supplementary Fig. 2i),
which indicates that the larger torsional
stiffness of the vertical posts exaggerates the influence of stitching inaccuracies. Through
empirical, iterative exploration, we found that vertical posts with diameters of 2.6
μm had the
best combination of struc
tural stability and minimal stitching influence on domain formation.
Narrower vertical posts would snap in the bottom layer upon lithiation driven by the greater
degree of rotation. The total number of vertical layers and the lateral size of Si microlattic
es were
chosen to optimize the trade
-
off between higher active material loading and reasonable
fabrication time.
Supplementary Figure 2
.
(a
-
c)
SEM images of hexagonal microlattices
(a)
before and
(b, c)
after lithiation
.
(d
-
f)
SEM images of
triangular
microlattices
(d)
before
and
(e, f)
after lithiation.
(g
-
i) )
SEM images of tetragonal microlattices with a larger vertical post diameter
(g)
before and
(h, i)
after lithiation
.
Dotted horizontal lines in (i) help to mark the stitching sites that have a
str
ong influence over the domain boundary location when a larger vertical post diameter is used.
II.
Electrochemical
Testing
Method
Modified CR2032 coin cell
s
are
used to test Si microlattice
s
f
or long
-
term cycling with accurate
electrochemical data
and minimized side reaction
s
. As shown in
Supplementary Fig.
3
a
, a
0.79
mm thick polyethylene washer is adhered
to the sample substrate via re
-
solidified paraffin
wax (Sigma
-
Aldrich) to create a small leak
-
free cavity
around the Si microlattice, which
sig
nificantly reduce
s
the amount of electrolyte used and the contact area between electrolyte and
Ni thin film on the substrate.
Approximately 30
μl of electrolyte is used in each coin cell, and the
electrolyte consists of 90
vol% of 1
M LiPF
6
in EC/DEC
=
5
0/50 (v/v) (battery grade,
Sigma
-
Aldrich) and 10
vol% FEC additive (BASF). A Li
foil
counter electrode with a 25
μ
m
-
thick
separator (Samsung) is placed on top of the polyethylene washer cavity filled with electrolyte.
The modified coin cells are sealed by
a crimper inside an Ar
-
filled glovebox before taking out
for electrochemical testing.
Elevated temperature experiments are conducted inside an
environmental chamber using coin cells. For each sample, we wait for 1hr before lithiation after
putting the cell
inside the environmental chamber at the set temperature for the cell to reach
thermal equilibrium.
Supplementary Figure 3
. (a) Illustration of modified coin cells. (b, c) Images of the
in situ
optical microscopy setup and the custom electrochemical cell with a quartz viewing window.
A custom
-
made electrochemical cell with a quartz window for
in situ
optical observation is
shown in
Supplementary Fig.
3
b,
c.
A
Li foil is punched into a ring
shape
to unblock the
top
-
down
view of
the
Si microlattice during
in situ
observation. Approximately 4
00
μl
of
1
M LiPF
6
in EC/DEC
=
50/50 (v/v) (battery grade,
Sigma
-
Aldrich)
electrolyte is used for each
in situ
cell
.
The large electrolyte amount gives rise to
significant side reaction
s
from electrolyte
decomposition and impurities like water and oxygen, which leads to
larger
and inaccurate
lithiation capacity. During electrochemical lithiation
/delithiation
,
Keyence VW
-
9000 digital
microscope records the
dynamics of cooperative buckling
/unbuckling
in
the
Si microlattices.
All lithiation, delithiation and cycling tests are conducted galvanostatically with a constant
current using a battery cycler (BCS 805, Bio
-
Logic Science Instruments)
or a potentiostat (
SP
200, Bio
-
Logic Science Instruments)
unless otherwise specified. The applied current is
quantified
by
the
C
-
rate in the
main
text, where
a C
-
rate of x·C is defined as the current under
which the electrochemical reaction can be completed in 1/x hours base
d on the theoretical
capacity of the active material. The theoretical capacity of the Si microlattice samples is
approximated to be 30
μAh when calculating the C
-
rate. Therefore, a constant current of 5
μA,
i.e. a current density of 0.15
mA/cm
2
normalized
by the Si coated area, corresponds to a C
-
rate of
~
C/6. For the Si microlattice
-
Li half cells, the lithiation (di
scharge) cutoff
voltage
is 0.01
V vs.
Li/Li
+
and the delithiation (charge) cutoff
voltage
is 1.5
V vs. Li/Li
+
for full delithiation and
0.6
V
vs. Li/Li
+
for partial delithiation. The first cycle Co
ulo
mbic efficiency is ~70
% with
the
0.6
V delithiation cutoff voltage, which indicates about 30
% of inserted Li remains in the Si
microlattices. Cyclic voltammetry (CV)
in Fig. 2e
is conducted at a s
can
ning
rate of 0.1
mV/s
between 0.01
V and 1.5
V vs. Li/Li
+
in modified coin cells. The shape and the current peaks of
the CV plot are consistent with previously published results of various Si anodes
27,28
.
It conveys
the reversible Si
-
Li alloying and dea
lloying reactions indicated by the reduction peaks around
0.03
V and 0.21
V and the oxidation peaks around 0.33
V and 0.49
V respectively. The initial
lithiation of pristine Si occurred at a lower voltage around 0.11V, and weak reduction peaks
around 0.40
V appeared in the second and third cycles possibly caused by irreversible Li
insertion; these features are consistent with reports for various binder
-
free amorphous Si
electrodes
28,50,51
.
III.
In situ
Observation
of Lithiation
-
induced Cooperative Buckling
Supp
lementary
V
ideo
1 and
Supplementary
Video
2
present
in situ
lithiation and delithiation of a
Si microlattice at a constant current of 5
μA (~C/6). The lithiation (discharge) cutoff voltage is
0.01
V vs. Li/Li
+
and the delithiation (charge) cutoff voltage is 1.5
V vs. Li/Li
+
. The video is
played at a speed of 2700X. The lithiation capacity
in the
in situ
cell
reached
122% of the
theoretical capacity of Si
, whereas the first lithiation capacity in modified coin c
ells
is
consistently ~80% of the theoretical
capacity under
the same galvanostatic conditions.
The f
irst
cycle Co
ulo
mbic efficiency wa
s 44% compared with that of ~90% in coin cells under the same
cycling conditions. The
s
e discrepancies
demonstrate the significantly large
r
side reactions in the
in situ
cell due to the large amount of electrolyte used.
Therefore, we refer
to different stages of
lithiation and delithiation in the
in situ
experiments by the corresponding voltages in Fig. 2a
,
b
instead of the
attained
capacities, and accurate electrochemical analysis and long
-
term cycling
are
conducted in modified coin cells.
Supplementary Fig. 4
are
SEM images of a representative
Si microlattice after the first
in situ
delithiation with a 1.
5
V
delithiation cutoff voltage showing
the fractured nodes.
Supplementary
Figure
4
.
SEM images of a representative Si microlattice aft
er the first
delithiation with
a
1.
5
V
delithiation cutoff voltage showing the fractured nodes.
Supplementary Video
3
shows
lithiation
-
induced buckling at a playing speed of 150X when a
2000
Ω
resistor load was applied between the Si microlattice and the Li counter electrode.
The
Si
-
Li alloying reaction is a spontaneous discharge process, which means that the alloy has a
lower free energy than that of the two electrodes combined. This implies that the observed
lithiation
-
induced cooperative buckling does not require additional energy supply to be activated
or to proceed. Supplementary Video 3 presents thermodynamically dri
ven lithiation and buckling
of a Si microlattice drawing current from the alloying reaction for joule
-
heating of the
2000
Ω
resistor.
The Si microlattice sample had artificial defects that favor the single
-
domain buckling
configuration.
A
ll beams
buckled
c
oherently as expected
and a single domain was formed.
Supplementary Video
4 shows
stable
and
reversible structural transformations
of
the
3
rd
charg
e
,
the 4
th
discharge, the 4
th
charge,
and
the 5
th
discharge at
high
lithiation/delithiation
rate
s
of the
same sample as in
Supplementary
Video
3 at a playing speed of 150X. The 3
rd
and the 4
th
charge
were conducted at a constant voltage of 0.6
V with a current cutoff of 10
μA and took ~9
min to
complete. The 4
th
discharge was conducted with a 221
Ω resistor l
oad and a cutoff voltage of
0.005
V, which took ~14min to complete. The 5
th
discharge was conducted at a constant voltage
of 0.01
V with a cutoff
current
of 20
μA, which took ~15
min to complete. The cutoff current for
constant
voltag
e discharge was relatively high because a significant amount of side reactions
would continue to
sustain the current
when the current dropped below 20
μA
, which was
confirmed in other samples
.
In these constant voltage and resistor load
discharge/charge
ex
periments,
the initial currents were very high (above 4C) and gradually slowed down as
lithiation/delithiation proceeded so
the majority of the
buckling/unbuckling deformation
happened in the first half of the lithiation/delithiation processes.