Embedding liquid metals (LMs) into an elastomer is emerging as a promising strategy for stretchable conductors. Existing manufacturing techniques are struggling between spatial resolution and process complexity and are limited to chemically resistant substrates. Here, we report on a hybrid process combining stencil lithography and centrifugal force-assisted patterning of liquid metal for the development of LM-based stretchable conductors. The selective wetting behavior of oxide-removed eutectic gallium-indium (EGaIn) on metal patterns defined by stencil lithography enables micrometer scale LM patterns on an elastomeric substrate. Stencil lithography allows for defining metal regions without harsh chemical treatments, making it suitable for a wide range of substrates. Microscale LM patterns are achieved by efficiently removing the excess material by the centrifugal forces experienced from spinning the substrate. The proposed approach allows for the creation of LM patterns with a line width as small as 2 μm on a stretchable poly(dimethylsiloxane) (PDMS) substrate. The electrical measurement results show that the fabricated EGaIn devices can endure 40% mechanical strain over several thousands of cycles. Furthermore, a stencil design using microbridges is proposed to address the mechanical stability issue in stencil lithography. An EGaIn conductor with a serpentine structure and an interdigitated capacitor are fabricated and characterized. The results demonstrate that the patterned serpentine conductors retain their functionality with applied mechanical strain up to 80%. The performance of the interdigitated capacitors upon applied strain is in good agreement with the theoretical estimation. Finally, we demonstrate our approach also on poly(octamethylene maleate (anhydride) citrate) (POMaC) substrates to broaden the use of the proposed method to not only flexible and stretchable but also biodegradable substrates, opening a way for in vivo transient microsystem engineering. The work presented here provides a versatile and reliable approach for manufacturing stretchable conductors.
Embedding liquid metals (LMs) into an elastomer is emerging as a promising strategy for stretchable conductors. Existing manufacturing techniques are struggling between spatial resolution and process complexity and are limited to chemically resistant substrates. Here, we report on a hybrid process combining stencil lithography and centrifugal force-assisted patterning of liquid metal for the development of LM-based stretchable conductors. The selective wetting behavior of oxide-removed eutectic gallium-indium (EGaIn) on metal patterns defined by stencil lithography enables micrometer scale LM patterns on an elastomeric substrate. Stencil lithography allows for defining metal regions without harsh chemical treatments, making it suitable for a wide range of substrates. Microscale LM patterns are achieved by efficiently removing the excess material by the centrifugal forces experienced from spinning the substrate. The proposed approach allows for the creation of LM patterns with a line width as small as 2 μm on a stretchable poly(dimethylsiloxane) (PDMS) substrate. The electrical measurement results show that the fabricated EGaIn devices can endure 40% mechanical strain over several thousands of cycles. Furthermore, a stencil design using microbridges is proposed to address the mechanical stability issue in stencil lithography. An EGaIn conductor with a serpentine structure and an interdigitated capacitor are fabricated and characterized. The results demonstrate that the patterned serpentine conductors retain their functionality with applied mechanical strain up to 80%. The performance of the interdigitated capacitors upon applied strain is in good agreement with the theoretical estimation. Finally, we demonstrate our approach also on poly(octamethylene maleate (anhydride) citrate) (POMaC) substrates to broaden the use of the proposed method to not only flexible and stretchable but also biodegradable substrates, opening a way for in vivo transient microsystem engineering. The work presented here provides a versatile and reliable approach for manufacturing stretchable conductors.
Flexible electronics
are circuits and electronic components that
can afford bending while maintaining their electrical functionality.
Significant efforts have been devoted to the development of paper
electronics[1] and flexible displays.[2] As an extension of flexible electronics, stretchable
electronics can be elongated to be conformably attached on the soft
human skin and tissue and follow body motion. For example, epidermal
electronics attached on the human skin allow for electrophysiological
recording,[3] and soft neural implants enable
neuroprosthetic applications.[4] Some wearable
applications require circuits and electronic components with the capability
to monitor large strains. For instance, the bending of human fingers
and wrists experience strain up to 30%.[5,6] Among all the
essential elements for wearable devices, stretchable conductors are
particularly important because they address the problem of the mechanical
mismatch between the soft tissue and the more rigid functional devices
by eliminating stress concentration. To perform this function, stretchable
conductors must remain mechanically intact and electrically conductive
under mechanical deformation.The engineering of structures
and materials has been intensively
studied to enable stretchable conductors. Structuring a rigid metal
film into deterministic fractal motifs in 2D[7,8] or
3D form[9,10] has enabled the extension of the strain
limit of rigid metal films. For example, a stretchability of 800%
has been demonstrated by shaping thin Ag film into meanders.[8] A thin Au film has been shown to be able to endure
over 100% strain with 3D helical design.[10] However, the stretchability of structured metal conductors is mainly
dominated by the rigid metal patterns, which decrease the usable density
of the electronic components because it consumes a large portion of
spaces with meander or helical wiring routing.The immobilization
of an ionic conducting liquid into a polymer
matrix offers an excellent stretchability and high transparency, which
attracts interest for the development of stretchable conductors.[11,12] A stretchable electroluminescent device using an ionic conductor
that shows stable device performance under stretching to 400% strain
over 1000 cycles has been demonstrated.[12] However, its high electrical resistivity (in the order of 107 Ω·μm) limits their applications as electrical
conductors. Embedding nonstretchable conductive nanomaterials into
stretchable insulating polymer matrix provides another design strategy,
which is working with the principle of percolation of the conductive
nanomaterials. The solid state material is not stretched but displaced
so it provides a good stretchability and an acceptable electrical
resistivity.[13,14] Stretchable conductors produced
by spray-coating CNTs on PDMS surfaces via a shadow mask enables the
realization of stretchable transistors that can sustain strain up
to 100% for at least 1000 stretching cycles.[14] Nevertheless, the most-commonly used manufacturing method for this
type of conductor is spray-coating, which results in a relatively
low resolution in the order of tens of micrometers due to the size
of the spray nozzles.[14] The relatively
low resolution limits their use in high density electronic devices.
However, liquid metals (LMs) provide an excellent opportunity for
use as stretchable conductors due to the lower electrical resistivity
ranging from 0.29 Ω·μm (eutectic gallium–indium)
to 1 Ω·μm (Hg)[15] compared
to the above-mentioned materials and an intrinsically unlimited strain
limit without the need of complex wire routing.[16]Among the liquid metal candidates for stretchable
conductors, the
alloy of eutectic gallium–indium (EGaIn, 75 wt % of Ga and
25 wt % of In) and gallium–indium–tin (Galinstan, 68.5
wt % of Ga, 21.5 wt % of In, and 10.0 wt % of Sn) are particularly
attractive because of their nontoxicity, low electrical resistivity,
negligible vapor pressure, and low melting temperature. For example,
EGaIn has a melting temperature of 15.5 °C and an electrical
resistivity of 0.29 Ω·μm at room temperature, i.e.,
in liquid state.[15] The electrical resitivity
of EGaIn is one order of magnitude higher than bulk Cu but one order
of magnitude lower than nanomaterial-embedded composites (e.g., AgNW)[13] and eight orders lower than ionic gels,[12] which makes EGaIn an excellent candidate for
stretchable conductors. One of the promising strategies for the use
of liquid gallium alloy is to embed it into an elastomeric polymer.[17,18] Being liquid at room temperature, the liquid gallium alloy can deform
into the shape of the microchannel and remain electrically conducting.
These properties make liquid gallium alloy suitable for wearable applications.The rigid thin oxide skin (∼1–3 nm), formed on the
surface of liquid gallium alloy when exposed to oxygen in the environment,
allows for shape retainment at the micrometer scale and form stable
free-standing structures.[16,19] A broad range of manufacturing
methods have been developed for structuring liquid gallium alloys
at micrometer or submicrometer scale by taking advantage of its moldability.
For example, liquid gallium alloys can be structured at the micrometer
scale by using picosecond laser ablation,[20] printing directly on substrates via nozzles,[19,21] injecting into microfluidic channels,[22,23] screen printing[24] or spray deposition via stencils,[25] or using PDMS stamps to transfer patterns.[26,27] Although using PDMS stamps to transfer liquid metal patterns shows
promising results, the substrate has to be molded to have desired
channels, which prevents this approach to be applied on a wide range
of substrate materials. Recently, a hybrid process utilizing electron-beam
(E-beam) lithography and stamping has been developed to shape EGaIn
at the submicrometer scale.[28] Therein,
the patterned EGaIn is transferred to the desired substrate by immersing
the device in DI water for more than 6 h to laterally etch away the
sacrificial layer. The transfer process makes the entire process time-consuming.
Additional fabrication methods have been developed by taking advantage
of the selective wetting property of oxide-removed liquid gallium
alloy. The native oxide of liquid gallium alloy (Ga2O3) can be removed using aqueous acids (HCl) or bases (NaOH).[29] Without the presence of Ga2O3, liquid gallium alloy can selectively wet different metal
surfaces such as Au and Cu. Therefore, by having patterns of metal
traces on the elastomer surface, where LM adheres, LM patterns can
be produced by removing excess material by using a roller or by a
brief exposure to an aqueous acid/base to cause them to dewet the
region without the metal traces.[30−32] However, in the reported
literature, the metal traces are produced by resist-based photolithography
processes, which puts a limit on what type of substrates can be used.
Moreover, it remains challenging to fabricate liquid gallium alloy
with lateral resolution smaller than 10 μm.[30,31] The limitation presumably comes from the used acid (i.e., HCl) which
etches the patterned metal layer or, alternatively, due to the noncontrollable
forces during the removal of excess liquid gallium alloy. To improve
the resolution, a thick (2 μm) electroplated Cu has been applied
as an adhesive layer for structuring liquid gallium alloys down to
2 μm.[32] However, this approach requires
an additional electroplating process, which normally requires acid
electrolyte, and thus increases the overall fabrication costs, and
finally it is not suitable for chemical-sensitive substrate materials.
Overall, a simple, versatile, and cost-efficient process that enables
patterning liquid gallium alloys at the micrometer scale on a broad
range of substrates is still lacking.An advanced resistless
process, such as stencil lithography (SL),
allows for the local deposition of materials onto a large variety
of substrate materials. It is based on shadow masking a flux of atoms
or molecules to locally deposit them on the substrate via a physical
vapor deposition (PVD) process such as thermal/e-beam evaporation
or sputtering. In contrast to the conventional resist-based process,
the mechanically stable and self-supported stencils can be used multiple
times[33] and also allows for a wafer-scale
processing,[34] which reduce process costs.
The entire process is free from harsh chemical treatments or high
temperature and thus is compatible also with polymer substrates. Stencils
for micro- and nanopatterning on chemically fragile substrates such
as stretchable[35,36] and biodegradable polymers[37,38] have already been studied. SL offers a reliable method for high-resolution
patterning on a wide range of substrates. However, there still exists
a challenge in SL for complex shapes such as a long serpentine structure
due to the low mechanical robustness of the thin suspended planar
membrane. By introducing corrugations in the SiN membrane, the moment
of inertia of the membrane can be greatly increased which reduces
the stress-induced stencil bending and thereby greatly improves the
pattern definition.[39] Producing the corrugated
membrane requires another fabrication process, which increases the
overall stencil fabrication time and costs. Furthermore, a long serpentine
or interdigitated structure has not been demonstrated yet by using
the reported corrugated membrane. Incorporating with thicker (3 μm)
electroplated Cu as the stencil membrane has enabled an interdigitated
structure.[24] However, the pattern resolution
is limited to few tens of micrometer due to the use of the thicker
membrane. In short, a process capable of producing a complex pattern
structure in the micrometer scale by SL without additional stencil
fabrication steps has not yet been reported.For facile and
reliable patterning, the selective deposition of
materials utilizing the centrifugal force experienced from spinning
the substrate is a well-established approach. It is normally used
to create uniform films on substrate by spinning off excess materials
from the surface. Selective deposition of materials can be achieved
by controlling interfacial interactions between the substrate and
the deposited material. Patterns can be achieved by efficiently removing
the excess materials by the well-controlled centrifugal force experienced
from spinning the substrates.[40,41] This approach provides
a way for high-resolution and size-scalable manufacturing of patterned
liquid structures.In this paper, we describe a hybrid process
combining stencil lithography
and centrifugal force-assisted patterning of LM to provide a simple
but reliable method to shape LM on an elastomer surface at the micrometer
scale. It allows for the creation of microscale EGaIn features on
a stretchable poly(dimethylsiloxane) (PDMS) substrate. We also propose
a new stencil design using microbridges to address the mechanical
stability issue in SL and to further enlarge its applicability of
complex design patterns, such as long meanders and interdigitated
electrodes. The EGaIn resistors with straight line and serpentine
shapes are both fabricated on PDMS to demonstrate the electrical and
electromechanical performances. Furthermore, we demonstrate the possibility
to extend this approach to a stretchable and biodegradable poly(octamethylene
maleate (anhydride) citrate) (POMaC) substrate. This represents an
important enabling step toward its use in applications such as recyclable
electronics.[42,43]
Results and Discussion
Microfabrication
Based on a Hybrid Process
The fabrication
process based on SL and centrifugal force-assisted patterning of LM
to shape EGaIn at the micrometer scale is illustrated in Figure a. The substrate
is composed of a soft PDMS layer having a thickness of 1 mm with a
lateral size of 30 × 15 mm2 on a 525 μm thick
silicon wafer, which is coated with a self-assembled trimethylsilyl
chloride silane (TMCS) (Sigma-Aldrich) layer to enable the peeling
of PDMS in the last step. The stencil mask used here consists of 500
nm thick low-stress silicon nitride (LS-SiN) membranes supported by
a silicon frame. The stencils are fabricated by patterning the apertures
on LS-SiN membranes, followed by the etching of the backside silicon
carrier to release the nitride membranes. First, the fabricated stencil
is brought into contact with the substrate and then 4 nm thick Cr
and 60 nm thick Au are locally sputter deposited on the substrate
via the aperture of the SiN stencil mask. The deposition is performed
using DC sputtering with powers of 350 and 250 W for Cr and Au, respectively.
Here, Cr is used as an adhesion layer for Au on PDMS, and Au is used
to have a selective wetting contrast with EGaIn. After the sputter
deposition, the stencil mask is removed from the PDMS substrate by
immersing the sample in isopropyl alcohol (IPA). Due to the sticky
property of PDMS, we use IPA to assist the removal of the stencil
without damaging it. IPA swells PDMS, which increases the strain at
the interface of the stencil and the PDMS, resulting in the PDMS film
to peel off.[36] In a second experiment,
we performed this method on a flexible POMaC substrate to validate
the possibility of extending this approach on a biodegradable substrates.
On POMaC substrate, the stencil can be released directly after sputter
deposition without the use of IPA. After the creation of the Cr/Au
pattern and the removal of the stencil mask, 20 μL of 4% NaOH
treated EGaIn is drop-casted onto the substrate to cover the entire
Cr/Au patterns, followed by spinning the substrate. Here, 4% NaOH
is applied to remove the native oxide of EGaIn (Ga2O3) so that it can selectively wet on the patterned Au surfaces.
The spinning starts with 100 rpm for 5 s, followed by 500 rpm for
7 s to remove the excess material. At the end, the substrate slows
down to 0 rpm in about 2 s. During the spinning, EGaIn wets the metal
traces and the excess material is removed from the substrate by the
centrifugal force. The EGaIn confinement on the metal traces is attributed
to the wetting contrast between PDMS and metal, and the pinning effect
on the metal mesa edge.[44] The competition
between the adhesion forces between EGaIn and Au, the cohesion forces
within EGaIn, and the centrifugal force determines the geometry of
the obtained EGaIn structures. Compared to the existing fabrication
methods using a roller or manual shaking to remove excess materials,[30,32,45] centrifugal force-assisted patterning
of EGaIn with microscale wetting contrast patterns provides a way
to efficiently remove the excess material in a controllable manner
and is size-scalable for large area manufacturing. Furthermore, the
removed excess EGaIn remains in the spin-coater as a droplet, which
allows to be simply recollected for further use (Figure S1, Supporting Information, SI). After the spinning, the device
is rinsed in DI water to remove unwanted NaOH residues. Finally, the
patterned LM device is sealed with a 1 mm thick layer of PDMS obtained
by gently pouring a self-curing liquid PDMS solution over the device
to protect LM patterns for further stretching test. At the end, the
encapsulated device is peeled off from the silicon carrier. The developed
hybrid process enables the creation of EGaIn patterns on the substrate
with minimal exposure to harsh chemicals. Only a few seconds of exposure
to 4% NaOH and DI water is needed in the process, which allows the
process to be extended to a wide range of substrates. To elucidate
the resolution limit, we have created on a PDMS substrate squares
with a minimum side of 5 μm and lines with a minimum width of
2 μm (Figure b). Arrays of lines and narrow lines with large contact pads with
a width of 2 μm are also successfully fabricated by the same
method (Figure S2, SI). Furthermore, arbitrary patterns can be produced, such
as the EPFL logo (Figure b(iii)). It is noted that in the stencil design, the letter
“P” is connected with bridges because the suspended
membrane in the stencil limits the production of a close-loop structure,
as known as the “Doughnut problem”. To solve this challenge,
we report in a later section a solution utilizing the blurring effect
to achieve complex and continuous LM patterns. The fabricated line
patterns on a flexible and biodegradable POMaC substrate (Figure S3, SI) demonstrates
the possibility to apply the proposed approach on a wide range of
substrates.
Figure 1
Hybrid process for the fabrication of EGaIn stretchable conductors.
(a) Schematic pictures showing the liquid metal patterning process
on a Si (0.5 mm)/PDMS (1 mm) substrate. (i, ii) The fabricated stencil
is placed on the substrate, followed by sputter deposition of Cr (4
nm) and Au (60 nm). (iii) EGaIn treated with 4% NaOH is drop-casted
onto the substrate. (iv, v) EGaIn wets the metal traces and the excess
material is removed by spinning the substrate. (vi) Sealing with PDMS
(1 mm) and peeling off from the Si support. (b) Various EGaIn patterns
obtained by the proposed process. (i) Patterned EGaIn square dot arrays
with width from 5 to 20 μm. (ii) Patterned EGaIn line patterns
with width from 2 to 20 μm. (iii) Patterned EGaIn EPFL logo.
All patterns are imaged with an optical microscope at step (v) of
the process.
Hybrid process for the fabrication of EGaIn stretchable conductors.
(a) Schematic pictures showing the liquid metal patterning process
on a Si (0.5 mm)/PDMS (1 mm) substrate. (i, ii) The fabricated stencil
is placed on the substrate, followed by sputter deposition of Cr (4
nm) and Au (60 nm). (iii) EGaIn treated with 4% NaOH is drop-casted
onto the substrate. (iv, v) EGaIn wets the metal traces and the excess
material is removed by spinning the substrate. (vi) Sealing with PDMS
(1 mm) and peeling off from the Si support. (b) Various EGaIn patterns
obtained by the proposed process. (i) Patterned EGaIn square dot arrays
with width from 5 to 20 μm. (ii) Patterned EGaIn line patterns
with width from 2 to 20 μm. (iii) Patterned EGaIn EPFL logo.
All patterns are imaged with an optical microscope at step (v) of
the process.To assess the geometry of the
fabricated EGaIn patterns on a PDMS
substrate, confocal microscopy (S neox, Sensofar) is used to measure
the profile of the patterned EGaIn square dots having different side
widths (Figure ).
As expected, the height of each square pattern increases with its
width (Figure a).
The same result has been shown by using spin-coating to selectively
deposit polymer thin film on a heterogeneous substrate.[41] Furthermore, a linear relationship can be obtained
between the patterned EGaIn width and height (Figure b). The error bar shows the spread of data
from three random samples, with each sample produced from a single
20 μL NaOH treated EGaIn droplet. The process is performed under
the same spinning parameters for all samples. Next, to study the effect
of the spinning speed, which leads to different centrifugal forces
on the geometry of the obtained EGaIn structure, geometrical analysis
is carried out on samples fabricated with different spinning speed.
Five spinning speeds ranging from 250 to 4000 rpm are used to create
EGaIn square array patterns, which consists of square patterns with
side widths of 5, 10, and 20 μm, respectively. The results (Figure S4, SI) show
a maximum standard deviation in heights of about 20% when applying
different spinning speeds. This demonstrates that using centrifugal
force experienced from spinning the substrate to structure LM provides
not only promising results but also a large process window for reproducible
manufacturing. Furthermore, it shows the possibility to extend the
presented method to the full wafer scale by adapting a well-established
full wafer stencil.[34] After patterning
metals on a wafer, several EGaIn droplets can be dispensed on the
metal structures. Followed by spinning the substrate to remove the
excess materials. Although the centrifugal force differs based on
the pattern location with respect to the rotation axis, the results
shown in Figure S4 in the SI demonstrate a weak dependence of the patterned EGaIn structure
dimensions on the spinning speed (i.e., on the centrifugal force),
which paves a way for wafer scale patterning.
Figure 2
Geometrical analysis
of the patterned EGaIn on a PDMS substrate.
(a) Profile of the patterned EGaIn square dots with different widths.
The inset shows the optical image of the sample. Scale bar: 50 μm.
(b) Relationship between patterned EGaIn width and height and its
linear fitting curve. The width and height values are taken from the
stencil aperture size and the measured profile, respectively. Each
data point contains measurement results from three random samples.
The error bar represents the standard deviation.
Geometrical analysis
of the patterned EGaIn on a PDMS substrate.
(a) Profile of the patterned EGaIn square dots with different widths.
The inset shows the optical image of the sample. Scale bar: 50 μm.
(b) Relationship between patterned EGaIn width and height and its
linear fitting curve. The width and height values are taken from the
stencil aperture size and the measured profile, respectively. Each
data point contains measurement results from three random samples.
The error bar represents the standard deviation.
Electrical Characterization of the Patterned EGaIn
Next
we evaluate the electrical performance of the patterned EGaIn
on a PDMS substrate (Figure ). The devices used for resistance measurement contain lines
with lengths of 50, 100, and 150 μm and widths of 2, 5, and
10 μm, and are connected by two square electrical pads having
a width of 100 μm at their terminal ends. After patterning EGaIn
with the method described above, the electrical measurements are performed
before encapsulating the devices. The parasitic resistance, which
consists of the resistance of the probes and the contact resistance
between the probes and droplets, are quantified and subtracted from
the measured total resistance. The parasitic resistance is measured
by inserting two probes into the same droplet whereby the measured
values varied between 5 to 10 Ω. The expected resistances of
patterned EGaIn are calculated from a parallel circuit formed by the
underneath adhesive Au layer and EGaIn structure. Here, because the
thickness of Cr (4 nm) is ten times thinner than Au (60 nm) and Cr
is six times more resistive than Au (0.12 Ω·μm for
Cr[46] and 0.02 Ω·μm for
Au[47]), the contribution of Cr to the electrical
conductance is neglected. The resistivity of the adhesive Au layer
is characterized by measuring the resistance of the sample before
patterning EGaIn. The result shows a resistivity of 0.31 Ω·μm
for the sputter deposited Au by using a stencil mask (Figure S5, SI). The
measured resistiviy of a sputtered Au thin film is more than ten times
of the value of bulk Au, which is attributed to random cracks and
surface wrinkles in the Au film formed during the sputtering process.[48,49] To calculate the resistivity of patterned EGaIn, the resistance
is plotted as a function of the inverse of the EGaIn cross-section
area for different lengths (Figure a). The cross-section area of each device is calculated
by integrating the profile measured by confocal microscopy (S neox,
Sensofar) and subtracted from the area of the underneath adhesive
layer. We observe variations in the cross-section area among different
samples having the same device geometry. The variations can be explained
by the resulting height differences of the patterned structures (Figure ). The measured resistivity
of EGaIn is 0.39 ± 0.02 Ω·μm (Figure b), which is close to the reported
resistivity of bulk EGaIn (0.29 Ω·μm).[15] It is important to note that the calculation
is based on a parallel circuit model without considering the occurring
cracks on the Au film. If the cracks are filled with EGaIn, then the
effective resistivity of the Au film that contributes to the conduction
is reduced. Therefore, the presence of such cracks would result in
an underestimation of the calculated resistivity of EGaIn.
Figure 3
Electrical
characterization of the patterned EGaIn on a PDMS substrate.
(a) Resistance measurement results as a function of the inverse of
the cross-section area with different lengths and widths. The inset
shows the geometry of the device and its measured cross-section profile
together with a sketch of the two-point probe configuration used for
the resistance measurement. Scale bar: 100 μm. (b) Extracted
resistivity of each measured data, which are 0.40, 0.37, and 0.39
Ω·μm for lengths of 50, 100, and 150 μm, respectively.
(c) Resistance versus applied current. The maximum current which does
not produce irreversible damage is about 7 mA. (d) Long-term stability
measurements on a sample with 10 μm in width and 100 μm
in length. The strain is applied with a period of 24 s per cycle.
Four cycle tests are performed in the same device under different
interval periods. The inset shows detailed resistance changes under
the maximum applied strain of 40% with an applied current of 1 mA.
Electrical
characterization of the patterned EGaIn on a PDMS substrate.
(a) Resistance measurement results as a function of the inverse of
the cross-section area with different lengths and widths. The inset
shows the geometry of the device and its measured cross-section profile
together with a sketch of the two-point probe configuration used for
the resistance measurement. Scale bar: 100 μm. (b) Extracted
resistivity of each measured data, which are 0.40, 0.37, and 0.39
Ω·μm for lengths of 50, 100, and 150 μm, respectively.
(c) Resistance versus applied current. The maximum current which does
not produce irreversible damage is about 7 mA. (d) Long-term stability
measurements on a sample with 10 μm in width and 100 μm
in length. The strain is applied with a period of 24 s per cycle.
Four cycle tests are performed in the same device under different
interval periods. The inset shows detailed resistance changes under
the maximum applied strain of 40% with an applied current of 1 mA.In order to verify the suitability of the fabricated
EGaIn structures
as stretchable conductors, the measurement of the maximum operation
current is performed. After patterning EGaIn on a PDMS substrate with
the presented method, the terminal droplets are administered on the
large electrical pads at the ends of the tested EGaIn wire. The terminal
droplets allow the copper wire to be added on to perform electrical
measurements. The device is encapsulated by gently pouring a self-curing
liquid PDMS solution on the device and curing it at room temperature
for 48 h. The experiment on the maximum operation current is conducted
by measuring the electrical resistance with increased input currents.
We observe that, at currents below 7 mA, no permanent modification
of the device resistance is introduced. The resistance increases as
a function of applied currents due to Joule heating but recover to
its initial value once the current is reduced (Figure c). The measured resistance change indicates
a temperature increase of around 65 °C based on the equation
ΔT = ΔR/(R0α), where R is the resistance,
and α the temperature coefficient of resistivity of EGaIn (∼10–3/°C).[50,51] The result shows that
LM conductors fabricated by the presented method could meet the requirements
for wearable devices operating at least in a temperature range between
20 and 85 °C. At larger currents (7.5 mA), the resistance starts
to gradually increase as a function of time and a permanent modification
of the device resistance is observed. The phenomena is tentatively
attributed to the void formation inside the embedded EGaIn channel
when the device is heated up to a certain temperature.[50,51] The maximum operation current for the fabricated EGaIn sample is
around 7 mA, which corresponds to a current density of around 2 ×
109 A/m2. The measured current density is about
4 orders of magnitude higher than the values for stretchable conductors
based on the percolation of conductive silver nanowires.[52] Next, the electrical characterization of a EGaIn
resistor under applied uniaxial deformation is performed. It shows
an excellent electrical stability over 1000 cycles with 40% maximum
applied strain in a sample with 10 μm in width and 100 μm
in length (Figure d). The electrical resistance increases from about 17 Ω to
45 Ω in the first 100 cycles and then remains constant up to
at least 1300 cycles (8.6 h). The mechanism of the observed phenomenon
is not clear. In previous literature,[53] the initial increase of the electrical resistance is tentatively
attributed to the deformation and reformation of the oxide skin. Because
PDMS is permeable to oxygen,[54] oxygen might
permeate inside the PDMS channel and results in the reoxidation process,
which forms a different amount of Ga2O3 within
the wire.[53] Another possible explanation
for the observed phenomenon is the increased crack density of the
underneath Au during stretching.[48] In the
first few cycles of stretching, EGaIn fills in the cracks produced
on the Au film and the entire metal structures are stabilized afterward.
After 1300 cycles of stretching in 8.6 h, the device is released to
its original shape. To further study the stability of the fabricated
device, three additional cycle tests are performed with different
interval breaks. The second cycle test is performed after 15 h from
the first cycle test. The results of the second cycle indicates that
there is no significant change in electrical resistance. The third
and the fourth cycle tests are performed after 10 min and 1 h from
the second and the third cycles, respectively. No significant electrical
resistance variations are observed. It can be concluded that after
a certain number of stretching cycles, the electrical resistance of
the fabricated device is no more affected by the subsequent stretching
cycle. The calculated gauge factor of the fabricated device is 1.3
within the applied strain range from 0% to 40% (the inset in Figure d) due to the geometrical
change of the embedded liquid conductor, which is in agreement with
previously reported values.[21] As an example
for an application of the proposed process, we show the stretchable
connection to a commercial LED chip. The connection to the LED chip
can withstand bending and twisting and remains functional (Figure S6, SI). These
measurements demonstrate that the bilayer conductive pattern (Au/EGaIn)
produced with the method proposed in this work has excellent and stable
electrical properties and can be used to fabricate stretchable conductors.
Microbridge Stencil for More Complex Patterns
Although
stencil lithography has been demonstrated to enable the deposition
of materials on a variety of substrates, the low mechanical robustness
of the suspended thin SiN membrane limits its use for more complex
patterns. To enable a wider variety of structures, in this work we
propose a microbridge stencil to enable geometries such as long serpentine
and interdigitated structures. The proposed microbridge stencil consists
of several bridges with a width of 1 μm uniformly distributed
across the entire structure to stabilize the suspended SiN membrane
(Figure a(i)). The
applied microbridges result in a significant reduction of the membrane
bending due to stresses and enable the realization of fragile serpentine
structure (Figure S7, SI). By lifting the microbridge stencil at a certain gap distance
above the substrate surface, the line-of-sight evaporation results
in material blurring and fills the shadowed bridge region (Figure a(ii,iii) and Figure S8, SI). In
the last, EGaIn is formed on the stencil-patterned metal trace by
spinning the excess EGaIn off the substrate. (Figure a(iv)). The blurring effect causes pattern
enlargement of the evaporated adhesive Au layer and consequently also
of the final LM structure. For example, we observe LM pattern enlargements
of 2.5 μm per side when using a 70 μm gap stencil-substrate
distance. The result indicates that the best resolution of using microbridge
stencil to pattern LM is around 6 μm, which results from the
1 μm aperture opening plus the 5 μm pattern enlargement.
The optical images of the sample taken during the process indicate
that the blurring underneath the microbridges in the stencil allows
for a continuous metal pattern on the substrate surface (Figure b(i)) and enable
a continuous LM pattern across the entire serpentine structure (Figure b(ii,iii)). Compared
to the reported strategy of using a corrugated SiN[39] or an electroplated Cu[24] membrane
to increase its mechanical robustness, the proposed microbridge stencil
enables a wider variety of designs in the micrometer scale without
introducing additional processes.
Figure 4
The use of a microbridge stencil for patterning
EGaIn and the fabrication
results (a) Schematic drawings showing the concept of microbridge
stencil (i) and fabrication processes (ii–iv). (ii) A microbridge
stencil is attached above the Si/PDMS substrate with a certain gap
distance (70 μm) and fixed with Kapton tape. (iii) Cr (4 nm)
and Au (150 nm) are evaporation deposited on the substrate. (iv) The
EGaIn is structured by using the process mentioned in Figure . (b) Optical images of the
sample during the fabrication steps: (i) Deposited continuous Cr/Au
pattern after removing the stencil mask. (ii) Patterned EGaIn serpentine
structure and its zoom-in view (iii). (c) Electromechanical characterization
results of the patterned LM structures with a microbridge stencil.
The inset shows the optical image of the tested sample. Scale bar:
100 μm. (d) Relative resistance change as a function of the
applied strain. Error bar shows the standard deviation of the measured
two samples. (e) An optical image of the fabricated microbridge stencil
with interdigitated apertures. (f) Measured and computed relative
capacitance change as a function of the applied mechanical strain.
The inset shows the optical image of the tested sample. Scale bar:
100 μm. Error bar shows the standard deviation of the measured
two samples.
The use of a microbridge stencil for patterning
EGaIn and the fabrication
results (a) Schematic drawings showing the concept of microbridge
stencil (i) and fabrication processes (ii–iv). (ii) A microbridge
stencil is attached above the Si/PDMS substrate with a certain gap
distance (70 μm) and fixed with Kapton tape. (iii) Cr (4 nm)
and Au (150 nm) are evaporation deposited on the substrate. (iv) The
EGaIn is structured by using the process mentioned in Figure . (b) Optical images of the
sample during the fabrication steps: (i) Deposited continuous Cr/Au
pattern after removing the stencil mask. (ii) Patterned EGaIn serpentine
structure and its zoom-in view (iii). (c) Electromechanical characterization
results of the patterned LM structures with a microbridge stencil.
The inset shows the optical image of the tested sample. Scale bar:
100 μm. (d) Relative resistance change as a function of the
applied strain. Error bar shows the standard deviation of the measured
two samples. (e) An optical image of the fabricated microbridge stencil
with interdigitated apertures. (f) Measured and computed relative
capacitance change as a function of the applied mechanical strain.
The inset shows the optical image of the tested sample. Scale bar:
100 μm. Error bar shows the standard deviation of the measured
two samples.The electromechanical characterization
is performed on EGaIn resistors
patterned with a microbridge stencil on a PDMS substrate. The serpentine
EGaIn resistor is characterized by stretching the device up to 80%
strain with a step of 10% strain. After each measurement, the device
is released to its original shape. The results indicate that the resistance
recovers to its initial value even after stretching to 80% strain
(Figure c). A relative
resistance change along with the applied strain shows a maximum relative
resistance change of 106% with an applied strain of 80% (Figure d). In addition to
meander shaped resistors, the proposed microbridge stencil also enables
interdigitated structures in the suspended membrane (Figure e). The confocal image shows
a greatly reduced membrane bending in the stencil (Figure S9, SI) and thus allows
the realization of an interdigitated capacitor. The measured initial
capacitance is around 0.3 pF, and the relative capacitance change
shows a linear relationship with the applied strain having a slope
of ∼0.9, which agrees well with the theoretical model derived
by Fassler and Majidi[55] (Figure f). The equation used for the
theoretical estimation is given in Equation S1 in the SI.
Conclusions
The
results reported in this work demonstrate an original process
that combines stencil lithography and centrifugal force-assisted patterning
of liquid metal for the realization of EGaIn-based stretchable conductors.
Using the selective wetting property of oxide-removed EGaIn, micrometer
patterns with resolution down to 2 μm can be obtained with micrometer
scale wetting contrast patterns. Stencil lithography allows patterning
metal traces, that EGaIn wets without harsh chemical treatments and
can thus be also applied to soft, biodegradable polymer materials.
The stencil can also be used several times to reduce fabrication costs.
The centrifugal force experienced from spinning the substrate allows
for the removal of excess materials in a highly controllable manner,
and thus enables micrometer EGaIn patterns. The patterned EGaIn on
a flexible and biodegradable POMaC substrate demonstrates its suitability
for a wider range of substrates. The presented method enables shaping
EGaIn into a variety of patterns. Furthermore, the fabricated EGaIn
patterns show low electrical resistivity (0.39 Ω·μm)
and excellent electromechanical response under large uniaxial deformation.
In particular, it can endure an applied strain of 40% over thousands
of cycles without significant change in its electrical performance.
Furthermore, the proposed microbridge stencil broadens the design
flexibility of stencil lithography. With a microbridge stencil, an
EGaIn resistor with a serpentine structure and an interdigitated capacitor
are demonstrated. The fabricated meander resistor can be stretched
up to 80% strain and recovered to its original shape without significant
change in its electrical resistance. The performance of the fabricated
interdigitated capacitor under the applied strain agrees well with
theoretical calculation. Comparing to the reported selective wetting
process for patterning liquid gallium alloy, the presented work shows
an improved pattern resolution without harsh chemical treatments and
thus enables extending the method to a wide range of substrate materials
such as biodegradable polymer materials (Table S1, SI). The work presented here
provides a simple and versatile method to fabricate liquid metal-based
stretchable conductors.
Experimental Section
Stencil
Fabrication
First, a 500 nm low-stress silicon
nitride films is grown on both sides of a bare silicon wafer by low
pressure chemical vapor deposition (LPCVD). Second, a photolithograpy
step is performed: 1 μm thick photoresist (AZ ECI 3007, MicroChemicals
GmbH) is spin-coated on the wafer and patterned by direct laser writer,
followed by the development of the resist (AZ 726 MIF, MicroChemicals
GmbH). Reactive ion etching (RIE) is then performed to etch LS-SiN
with a mixture of He, H2, and C4F8 gases. The photoresist is then removed with oxygen plasma. Third,
by the same method, a photolithography step is performed on the backside
of the wafer but with a 8 μm thick photoresist (AZ 10XT, MicroChemicals
GmbH), followed by etching the backside LS-SiN film with RIE. Last,
the Bosch process is used to etch ∼400 μm of the silicon
bulk, and KOH is used to etch the remaining silicon. At the end, the
wafer is cleaved into 10 × 10 mm2 chips.
PDMS Substrate
Preparation
First, a 4 in. silicon wafer
is cleaved into 30 × 15 mm2 chips. The chips are exposed
to oxygen plasma with 1000 W for 1 min. Then, a self-assembled layer
of trimethylsilyl chloride silane (TMCS) (Sigma-Aldrich) is coated
on the chips in a desiccator. The poly(dimethylsiloxane) (PDMS) solution
is prepared by mixing prepolymer and cross-linker with a mass ratio
of 10:1 (Sylgard 184, Dow Corning) and poured on the chips. Later,
the chips are placed in a desiccator for degassing. At the end, they
are cured in an oven at 80 °C for 2 h.
Spinning Procedure
The chip is first attached on a
vacuum sample holder in a spin-coater. Then 20 μL of 4% NaOH
and EGaIn (Eutectic Gallium–Indium, ≥ 99.99%, Sigma-Aldrich)
are drop-casted on the chip using a pipet, followed by spinning the
chip. Spinning program is setup in three sequential steps using an
in-built controller. Each step is defined by the target spinning speed
and duration time. The time for acceleration is also included in the
duration time. The parameters used for each step are as follows: (1)
100 rpm for 5 s with an acceleration rate of 82 rpm/s. (2) 500 rpm
for 7 s with an acceleration rate of 82 rpm/s. (3) 0 rpm for 5 s with
a deceleration rate of 410 rpm/s. At the end, the chip is rinsed with
DI water.
Uniaxial Electromechanical Characterization
The sample
chip with size of 30 mm by 15 mm is peeled from the silicon carrier
and mounted on a commercial tensile testing stage (TST350, Linkam
Scientific) with an in-built controller to actuate one of the two
clamps moving along the horizontal plane. The electrical resistance
is measured using a two-point probe method (2400 source-meter, Keithley)
controlled by a LabVIEW program. The resistances are measured with
an applied current of 1 mA. The capacitance is measured using a LCR
meter (4285A, HP) with an applied voltage of 500 mV and a frequency
of 500 kHz.
Authors: J A Rogers; Z Bao; K Baldwin; A Dodabalapur; B Crone; V R Raju; V Kuck; H Katz; K Amundson; J Ewing; P Drzaic Journal: Proc Natl Acad Sci U S A Date: 2001-04-24 Impact factor: 11.205
Authors: Serap Aksu; Min Huang; Alp Artar; Ahmet A Yanik; Selvapraba Selvarasah; Mehmet R Dokmeci; Hatice Altug Journal: Adv Mater Date: 2011-08-24 Impact factor: 30.849
Authors: Seung-Kyun Kang; Rory K J Murphy; Suk-Won Hwang; Seung Min Lee; Daniel V Harburg; Neil A Krueger; Jiho Shin; Paul Gamble; Huanyu Cheng; Sooyoun Yu; Zhuangjian Liu; Jordan G McCall; Manu Stephen; Hanze Ying; Jeonghyun Kim; Gayoung Park; R Chad Webb; Chi Hwan Lee; Sangjin Chung; Dae Seung Wie; Amit D Gujar; Bharat Vemulapalli; Albert H Kim; Kyung-Mi Lee; Jianjun Cheng; Younggang Huang; Sang Hoon Lee; Paul V Braun; Wilson Z Ray; John A Rogers Journal: Nature Date: 2016-01-18 Impact factor: 49.962