Supplementary Materials for
3D-printed epifluidic electronic skin for machine learning–powered
multimodal health surveillance
Yu Song
et al.
Corresponding author: Wei Gao, weigao@caltech.edu
Sci. Adv.
9
, eadi6492 (2023)
DOI: 10.1126/sciadv.adi6492
The PDF file includes:
Note S1
Figs. S1 to S45
Table S1
Legends for movies S1 to S5
References
Other Supplementary Material for this manuscript includes the following:
Movies S1 to S5
Note S
1
.
In situ
synthesis of MXene
-
Prussian blue (MX
-
PB)
Prussian blue (PB) nanoparticles is synthesized
in situ
on exfoliated MXene nanosheets
when mixed in a solution containing ferric chloride and ferricyanide (
Met
hods
). The
detailed synthetic pathway of MX
-
PB is depicted in
fig. S14a
. When exposed to the Fe
3+
ions, the electrostatic
-
repulsive forces between the MXene nanosheets are destroyed as the
positively charged metal ions spontaneously bind to the negatively charged surface of
MXene (
5
4
). This is evident by the instantaneous aggregation of MXene which fo
rms
sediments at bottom of the dispersion. At the same time, it can also be observed that the
dispersion turned blue instantaneously which suggests spontaneous formation of PB
nanoparticles. This process is similar to the one
-
step hybridization of metallic
nanoparticles
with MXene using aqueous metal salt precursors (
5
5
–
5
7
). The formation mechanism can
be explained by an
in situ
reduction
-
oxidation (redox) process, where the Ti on the surface
of MXene serves as reductant for reducing ferric ions (
5
8
,
5
9
):
F
e
3+
→
Fe
2+
(1)
Fe(CN)
6
3
-
→
Fe(CN)
6
4
-
(2)
A series of material characterization techniques including XRD, Raman, EDS and TEM
were used to determine the formation of MX
-
PB (
figs. S14
and
S
15
). From the XRD and
Raman spectra of MX
-
PB,
characteristic peaks originating from both MXene and PB were
observed (
fig. S14B
and
S14C
). EDS mapping of MX
-
PB deposited on porous CNT
-
styrene
-
butadiene
-
styrene (CNT
-
SBS) revealed that all the elemental compositions of MX
-
PB (i.e., C, Ti, Fe and N)
were uniformly distributed (
fig. S14D
and
S14E)
. The
microstructure of the MX
-
PB was further investigated using TEM (
fig. S15)
. As can be
easily observed, many PB nanoparticles were decorated uniformly on the surfaces of the
MXene nanosheets. High resoluti
on TEM images in
Fig. 3F
unveiled the presence of PB
nanoparticles with measured D
-
spacings of 0.36 nm and 0.23 nm corresponding to (220)
and (420) crystal planes of PB, respectively. The average size of the PB nanoparticles were
estimated to be ~8 nm. It
is noteworthy to mention that no leaching of PB nanoparticles
from the MX
-
PB was observed after repeated centrifugation indicating strong electrostatic
attraction between the MXene and PB nanoparticles.
Fig. S1.
3D printing with customizable inks.
(
A
)
Optical image of the semi
-
solid extrusion
(SSE)
-
based 3D printing system. Scale bar, 10 cm.
(
B
)
Optical image of the tapered tips and
stainless
-
steel straight tips for inks with different properties.
Scale bar, 1 cm.
(
C
)
Optical
image of the customizable
inks for 3D printing.
(
D
and
E
)
Typical plots of viscosity (
D
) as
a function of shear rate and storage modulus (G′) and loss modulus (G′′) (
E
) as a function
of shear strain for 3D
-
printable inks. The green and purple boxes represent shear
conditions
following and during extrusion, respectively.
0.01
0.1
1
10
100
10
100
1k
10k
Shear strain
(%)
Storage modulus (G')
Loss modulus (G'')
Modulus
(
Pa
)
0.01
0.1
1
10
100
1
10
100
1k
10k
Shear rate
(s
-1
)
Viscosity
(
Pa·s
)
Straight tip
Tapered tip
Shear
-
thinning
Yield
-
stress
SSE-
based
3D
printing system
MX
ene
MX-PB
SBS
CNT
-
SBS
CNT
-
SBS-
PANI
Graphite
-
SBS
CB
-SBS
Ag
CNT
-
PDMS
PVA-
H
2
SO
4
C
E
A
B
D
Fig. S2.
Fabrication process of the 3D
-
printed e
3
-
skin.
Schematic illustration of the
sequential printing and assembly of the e
3
-
skin consisting of 3 components: 3D
-
printed
microfluidics, 3D
-
prin
ted biosensors, and 3D
-
printed energy storage device.
3D
-
printed
CNT
-
SBS
-
PANI
3D
-
printed
Ag
electrode
3D
-
printed microfluidics
3D
-
printed biosensors
3D
-
printed
SBS substrate
3D
-
printed
SBS inlet layer
3D
-
printed
SBS
channel layer
3D
-
printed carbagel
3D
-
printed
e
3
-skin
3D
-
printed
CNT
-
SBS electrode
3D-
printed MXene electrode
3D
-
printed
sensing layer
3D
-
printed micro
-
supercapacitor
3D
-
printed
SBS substrate
3D
-
printed
MX
ene electrode
3D
-
printed
gel electrolyte
Freeze
-
drying
treatment
Immerse
in DI for 1h
Assemble
All the customizable inks require drying process after printing.
* R
emove the 3D
-
printed device from the printing bed for further treatment.
*
*
*
F
ig. S3.
Schematic illustration of the preparation process of concentrated MXene inks.
MXene nanosheets were synthesized from MAX phase precursor using the minimally
intensive layer delamination (‘M
ILD’) method (
60
).
HCl
-
LiF
etch
‘MI
LD’ method
S
tir
C
entrifuge
Shake & centrifuge
Collect sedimentation
Collect
s
upernatant
C
entrifuge
Collect sedimentation
Vacuu
m-
dry
Concentrated
MXene ink
MXene
‘clay’
Delaminated MXene
&
unetched
MAX
Multilayered MXene
&
unetched
MAX
MAX precursor
(
Ti
3
C
2
T
X
)
MAX precursor
(
Ti
3
C
2
T
X
)
Fig. S4.
Characterization of the 3D
-
printed filament with the MXene inks.
(
A
)
Schematic of the SSE
-
based 3D printing of MXene filament.
V
, printing speed;
P
, pressure;
D
, nozzle diameter;
H
, printing height;
d
, diameter of printed filament;
C
, extrusion speed;
α
, die swelling ratio.
(
B
)
Theoretical model for the diameter of the printed filament as a
function of the printing speed based on the equation for volume conservation
8
:
d
=
α
D
/
√
V
/
C
(1)
For printing of straight filament, the printing speed,
V
, must be equal or greater than the
critical extrusion speed,
CS
. As
V
increases, the diameter of the printed filament,
d
,
decreases and the filament becomes thinner. When
V
exceeds the limiting speed,
LS
, the
filament cannot be conformally deposited on the substrate resulting in discontinuity.
(
C
and
D
)
Viscosity as a function of shear rate (
C
) and storage modulus and loss
modulus as a
function of shear strain (
D
) of the MXene ink (60 mg mL
-
1
).
(
E
)
Microscopic image of the
printed MXene filaments with gaps from 10 μm to 50 μm. Scale bar, 100 μm. (
F
to
H
)
Linewidth distribution (F and G) and uniformity (
H
) of the 100 printed
MXene filaments
with
an ultrahigh spatial uniformity of 0.93%.
Substrate
H
P
d
0
25
50
75 100
50
100
150
200
0
Linewidth
(
μm
)
Number
Average linewidth: 160
μM
0.7
1.3
1.6
1.9
1
Limiting speed (LS)
Critical speed (CS)
0
Speed
V
(
mm s
-1
)
d
/
D
∞
0.01 0.1
1
10 100
10
100
1000
10000
Shear strain
(%)
Storage modulus
Loss modulus
Modulus
(
Pa
)
0.01 0.1 1 10 100 1000
100m
10
1k
100k
Viscosity
(
Pa·s
)
Shear rate
(s
-1
)
-3 -2 -1
0123
16
32
48
0
Counts
Width variation (%)
<w>: 0.93%
MX
ene
(60 mg mL
-1
)
MX
ene
(
60 mg mL
-1
)
154
157
160
163
166
Printed MXene filament
Linewidth
(
μm
)
Curve
filament
Straight filament
Discontinuous filament
Equal
-
diameter
Die
-
swelling filament
Thinned filament
Printed filament
V
D
MXene
Nozzle
αD
C
C
D
F
G
E
H
A
B
Fig. S5. Characterization of the MXene nanosheets.
(
A
and
B
) Atomic force microscopy
(AFM) image of MXene nanosheet (A) and its corresponding height profile (B). Scale bar,
500 nm. (
C
)
The distribution of lateral size of 60 sampled MXene nanosheets.
R2 C1
AFM & lateral size of
MXene nanoflakes
Scale bar 1 um
Fig R2 S5
A
B
Mxene
nanosheets
12345
0
8
16
24
Counts
Lateral size (μm)
Aver
age
lateral
siz
e
:
2.
6
3
μm
Avg thickness = 3.54
±
0.315 nm
Avg lateral size = 2.625
±
0.783 μm
Fig
.S5.
Characterization
of the
MXene
nanosheets
.(
A
and
B
)
Atomic
force
microscopy
(AFM)
image
of
MXene nanosheet
(A)
and
its
corresponding
height
profile
(B)
.
Scale
bar,
500
nm
.(
C
)
The distribution of
lateral
size
of 60
sampled MXene
nanosheets
.
C
0
500
1000
1500
0
2
4
6
Height (nm)
Distance (nm)
Thickness : 3.54 nm
Fig. S6. Characterization of 3D
-
printed temperature sensors with different linewidth.
(
A
) Optical image showing the linewidth of the temperature sensor in our e3
-
skin. Scale
bar, 500 μm.
(
B
and
C
) Dynamic response of the MXene
-
based temperature sensors with
different linewidth under varying temperature (B) and the calibration plot within
physiological temperature range (C).
25 30 35 40 45 50
-30
-20
-10
0
ΔR/R
(%)
350 μm
300 μm
400 μm
Temperature
(°C)
0
100
200
300
2.5
3.5
4.5
5.5
R
(
kΩ
)
Time
(s)
350 μm
300 μm
400 μm
A
B
L
inewidth
350
μm
R2 C2
Linewidth & temperature sensitivity
A scale bar 500 um
Fig R3 S6
T
emperature
25
°
C
30
°
C
35
°
C
40
°
C
45
°
C
50
°
C
C
Fig
.S6.
Characterization
of 3D-printed
temperature sensors
with
different
linewidth
.(
A
)
Optical
image showing
the
linewidth
of
the
temperature
sensor
in
our
e
3
-
skin
.
Scale
bar, 500
μm
.(
B
and
C
)
Dynamic
response
of
the MXene
-
based
temperature
sensors
with
different
linewidth
under
varying
temperature
(
B
)
and
the
calibration
plot
within
physiological
temperature range
(
C
).
Fig.
S7
.
Dynamic response of the MXene temperature sensor.
(
A
and
B
)
Response time
of a MXene temperature sensor upon contact with and removal from the human skin (
A
)
and an ice cube (
B
).
0
100
200
300
3.3
3.7
4.1
4.5
Resistance
(
kΩ
)
Time (s)
0
100
200
300
4
5
6
Resistance
(
kΩ
)
Time (s)
Ice cube (0 º
C)
Contact
R
emoval
Skin (37 º
C)
Contact
Removal
A
B