Recyclable Thin‐Film Soft Electronics for Smart Packaging and E‐Skins

Abstract

Despite advances in soft, sticker‐like electronics, few efforts have dealt with the challenge of electronic waste. Here, this is addressed by introducing an eco‐friendly conductive ink for thin‐film circuitry composed of silver flakes and a water‐based polyurethane dispersion. This ink uniquely combines high electrical conductivity (1.6 × 10⁵ S m⁻¹), high resolution digital printability, robust adhesion for microchip integration, mechanical resilience, and recyclability. Recycling is achieved with an ecologically‐friendly processing method to decompose the circuits into constituent elements and recover the conductive ink with a decrease of only 2.4% in conductivity. Moreover, adding liquid metal enables stretchability of up to 200% strain, although this introduces the need for more complex recycling steps. Finally, on‐skin electrophysiological monitoring biostickers along with a recyclable smart package with integrated sensors for monitoring safe storage of perishable foods are demonstrated.

Full text

RESEARCH ARTICLE
www.advancedscience.com
Recyclable Thin-Film Soft Electronics for Smart Packaging
and E-Skins
Manuel Reis Carneiro, Aníbal T. de Almeida, Mahmoud Tavakoli,* and Carmel Majidi*
Despite advances in soft, sticker-like electronics, few efforts have dealt with
the challenge of electronic waste. Here, this is addressed by introducing an
eco-friendly conductive ink for thin-film circuitry composed of silver flakes and
a water-based polyurethane dispersion. This ink uniquely combines high
electrical conductivity (1.6 ×105Sm
−1), high resolution digital printability,
robust adhesion for microchip integration, mechanical resilience, and
recyclability. Recycling is achieved with an ecologically-friendly processing
method to decompose the circuits into constituent elements and recover the
conductive ink with a decrease of only 2.4% in conductivity. Moreover, adding
liquid metal enables stretchability of up to 200% strain, although this
introduces the need for more complex recycling steps. Finally, on-skin
electrophysiological monitoring biostickers along with a recyclable smart
package with integrated sensors for monitoring safe storage of perishable
foods are demonstrated.
1. Introduction
The accumulation of electronic waste (e-waste) is a pressing
global problem that poses environmental threats to natural
ecosystems, economic burdens related to the loss or insuffi-
cient recovery of valuable resources (e.g., precious metals), and
health concerns due to the use and improper discarding of
toxic substances.[1,2 ] This challenge is only expected to deepen
as electronics become more ubiquitous through the use of
wearable, printable, and sticker-like systems that are currently
being developed for use in varied fields from healthcare to
M. Reis Carneiro, C. Majidi
Soft Machines Lab
Department of Mechanical Engineering
Carnegie Mellon University
Pittsburgh, PA, USA
E-mail: cmajidi@andrew.cmu.edu
M. Reis Carneiro, A. T. de Almeida, M. Tavakoli
Institute of Systems and Robotics
Department of Electrical and Computer Engineering
University of Coimbra
Coimbra-, Portugal
E-mail: mahmoud@isr.uc.pt
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/./advs.
©  The Authors. Advanced Science published by Wiley-VCH GmbH.
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in
any medium, provided the original work is properly cited.
DOI: 10.1002/advs.202301673
industry.[3,4,13–18,5–12] Like conventional elec-
tronics, flexible thin-film circuits could add
to the growing e-waste problem since there
are no established recycling methods for
their unique materials and manufacturing
processes.[19,20 ] Moreover, unlike the tradi-
tional rigid electronics that are intended for
long-term use, most thin-film soft electron-
ics are being designed as disposable devices
for use in health monitoring, IoT, and smart
packaging,[21–25 ] having the potential to dra-
matically increase the quantity of e-waste in
the coming decades.
Another emerging environmental chal-
lenge is the use of toxic materials in the
synthesis and decomposition of soft and
wearable electronic systems. Current ef-
forts typically rely on conductive inks and
composites that are printed over polymeric
substrates[26–28 ] and often rely on toxic organic solvents in to
achieve adequate rheology for printing and processing.[29] If im-
properly handled or discarded, such solvents can impact human
health and threaten natural ecosystems.[19] Furthermore, some
of these e-inks have a limited shelf-life and must be stored and
handled within a specific temperature range (usually with re-
frigeration), thereby further adding to their cost and environ-
mental burden on account of increased energy consumption and
pollution.[30–35 ]
Here, we address these challenges by introducing a new class
of printable conductive inks, and fabrication techniques that
enable digital fabrication of sustainable, and eco-friendly soft
electronics. This ink uniquely combines high resolution digital
printing, microchip integration, strain tolerance, and simple,
efficient, and ecological recycling. Moreover, the microchip
integration is performed without the addition of a separate
electrically conductive adhesive (ECA), which is common in
printed electronics,[36–40 ] thus reducing the fabrication cost and
complexity. Finally, all processing steps are performed in ambi-
ent conditions, with no need for thermal sintering of the printed
circuit. This reduces energy consumption during fabrication and
permits printing on a wide range of heat-sensitive substrates,
including sustainable plastics that are of increasing interest for
green electronic technologies such as paper,[41,42] textiles,[43] or
polymers[44,45 ] and bio-synthetic composite materials.[46–49 ]
These inks are composed of various combinations of water-
borne polyurethane dispersion (WPU), Ag flakes, and liquid
metal (LM) alloy and are free of organic solvents. Combinations
of these materials result in printable inks that exhibit high
electrical conductivity, tolerance to mechanical strain, robust
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (1 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 1. Green soft electronics with a circular life cycle. A) Eco-friendly water-based conductive ink composed of silver flakes and polyurethane for
wiring rigid microchips and surface mount devices on a polyurethane film substrate. B) The water-based Ag-WPU ink can be stored at room temperature
for more than  weeks without impacting conductivity or printability. C) Room temperature circuit fabrication process includes digital circuit printing,
microchip pick-and-placing, and electromechanical bonding of components through a soft “soldering” process. D) Applications include ultraflexible cir-
cuits, compliant electronics, and skin-mounted bioelectronics. E) An organic solvent-free decomposition process based on the swelling of polyurethanes
when in contact with Isopropyl Alcohol (IPA) enables separation of the soft circuits in their fundamental components that can be further recycled intoa
new ink or further reprocessed.
adhesive properties, and sinter-free conductivity, making it
compatible for printing on soft elastomeric thin films and other
fragile heat-sensitive substrates. The same ink can also be used
to attach surface mounted devices (SMDs) and other micro-
electronic components for creating complex thin-film circuits,
thereby eliminating the need for additional soldering steps and
materials. Table S1 (Supporting Information) compares this
formulation with other recent efforts on recyclable electronics.
Despite promising advancements with these previous attempts,
no one conductive ink has simultaneously satisfied all of the
necessary requirements for scalable fabrication and large-scale
deployment. These include the ability to integrate surface
mounted microelectronic chips or the ability to support digital
printing, extreme mechanical deformation, and ecologically-
friendly pathways for ink synthesis and decomposition.
One version of this ink is composed of a percolating network
of microscale silver flakes (Figure 1A). This water-based conduc-
tive polymer composite is recyclable, has a conductivity of ≈1.1–
1.6 ×105Sm
−1, and can be stored at room temperature for more
than 4 weeks without impacting its conductivity or printability
(Figure 1B). This eco-friendly ink can be printed digitally to cre-
ate flexible and soft electronic circuits, which can be combined
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (2 of 16)

www.advancedsciencenews.com www.advancedscience.com
with rigid IC chips and SMD components through a “soft sol-
dering” process at room temperature (Figure 1C). The printing
scalability and SMD integration enable numerous applications,
including ultraflexible, surface-conformable, and recyclable cir-
cuits for consumer electronics as a substitute for flexible PCBs
(fPCBs) and molded interconnect devices (MIDs), as well as a
skin-conformable health monitoring sticker with integrated elec-
tronics for skin temperature monitoring and electrophysiology
(Figure 1D).
Recycling of circuits is achieved through a facile process
at room temperature, as depicted in Figure 1E. The process
involves soaking circuits in isopropyl alcohol (IPA) to achieve
decomposition and separation of the circuits into fundamental
components (ink, polymeric substrate, and microchips), without
requiring special equipment or toxic solvents. The recovered ink
aggregates can then be washed with IPA to enable separation
of the PU and silver flakes, which can be reused to synthesize
new inks using the original method. This eco-friendly process
facilitates proper disposal or recycling of the circuit components.
The proposed combination of materials and uncomplicated
techniques results in sophisticated eco-friendly soft electronic cir-
cuits with a fully-circular life-cycle. These materials and methods
encompass the creation of an environmentally friendly conduc-
tive polymer that can be easily utilized in digital printing pro-
cesses, a simple and reliable approach to integrate SMD chips at
room temperature without the need for sintering, and an inex-
pensive method for separating and breaking down the circuits,
allowing for the retrieval and reuse of their components in the
production of new circuits for diverse applications across various
fields.
2. Results and Discussion
2.1. Ink Synthesis and Soft Circuit Fabrication
As depicted in Figure 2A, the ink is prepared by adding Ag flakes
in an organic solvent-free WPU dispersion. Deionized water is
then added to tune the ink rheology so that it can be printed. The
solution is further mixed, resulting in a silvery ink with paste-like
consistency shown in Figure 2B.
Conductive traces and complete digital circuits can be pat-
terned through stencil printing (Figure 2C(i)) or direct ink
writing (Figure 2C(ii)). While the use of a stencil allows for easy
scalability of the printing procedure, direct ink writing allows for
automated stencil-free additive fabrication of circuits with less
ink waste. The ink is printed over a 50 μm thick polyurethane
film and the printed ink is left to dry for 5 min to evaporate all
water in the conductive paste. The full process is done at room
temperature. To ensure the quality of printed or written ink on
the substrate, special care was taken to prevent any interruptions
or failures in the printed traces. Furthermore, postprinting, a
visual inspection was conducted to guarantee a smooth surface,
free from imperfections and discontinuities.
To implement hybrid circuits with integrated microchips, the
rigid electronic components are placed over the printed circuit
using either a pick-and-place system (Figure 2C(iii)) or manually
with tweezers, as would be done for conventional rigid printed
circuit boards (PCBs). The microchips are then integrated into
the circuit through a room-temperature soft bonding process,
as shown in Figure 2C(iv). To achieve both mechanical bonding
and electrical connection between the SMDs and the underly-
ing printed conductive track, Ag-WPU ink droplets are dispensed
on the SMD terminals using a syringe with a stainless steel tip
(200 μm inner diameter) and dried for 5 min at room temper-
ature. Through this “soft soldering” process, a conformal bond
around the rigid component terminals and the underlying con-
ductive track can be achieved.
In Figure 2D(i), an SMD flat cable connector (680 μm pin sep-
aration) is shown bonded to a set of conductive lines printed
through direct ink writing, and it can be observed that the poly-
meric soft solder joints conform well to the rigid terminals of
the SMD component. This circuit can work as a flexible mul-
tisignal conductor which is ultraflexible, and capable of with-
standing 180°bend with minimal (sub-mm) bending radius, as
shown in Figure 2D(ii,iii). Moreover, since the overall circuit is
so thin (≈100 μm) it can be molded to nonplanar surfaces to
work as a molded interconnect device (MID).[50] For instance, in
Figure 2D(iv), the printed circuit is shown adhered to a curved
glass surface only by means of the natural tackiness and stiction
of the elastomer substrate.
Figure 2E(i) depicts a printed circuit with line widths of 1000,
600, and 400 μm, respectively, from top to bottom. In this cir-
cuit 0603-sized (1.5 ×0.8 mm package dimensions and 0.7 mm
pad separation) SMD LEDs and resistors were bonded as shown
in Figure 2E(ii). A detail of the circuit showing the bonded LED
and the printed lines and pads is shown in Figure 2E(iii). The
functioning circuit with illuminated LEDs is shown undeformed
in Figure 2F(i). Even when adhered to nonplanar surfaces and
when crumpled or bent (up to 180°bending with a submm ra-
dius of curvature), the soft printed circuit with integrated SMD
components remains functional as shown in Figure 2F(ii-v). In-
tegration of smaller SMD components in the printed soft circuits
was also shown to be functional, as seen in Figure S1 (Supporting
Information) where 0402-sized resistors and LEDs (1.0 ×0.5 mm
package dimensions and 0.5 mm pad separation) were bonded
to a set of printed lines with 200 μm width which corresponds
to the highest resolution print achieved through DIW. Video S1
(Supporting Information) shows an ultraflexible printed circuit
that is deformed (through bending and crumpling) without los-
ing functionality.
2.2. Characterization
2.2.1. Conductivity and Shelf Life
As shown in Figure 3A, the conductivity of the Ag-WPU (89.2%
Ag wt) printed trace is ≈1.16 ×105Sm
−1soon after the ink dries
(day 0) and slightly improves over time reaching ≈1.54 ×105S
m−1after 30 days of being printed and the printed lines being
left unprotected at room temperature. This behavior can be ex-
plained by moisture slowly evaporating over time after the bulk
H2O (the ink wet medium) is initially evaporated, which leads to
an increase in the ratio of silver filler particles per volume of non-
conductive PU, hence the higher conductivity. This behavior was
previously observed in other conductive particle-filled polymeric
compounds.[51]
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (3 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 2. A) Synthesis of the sinter free water-based conductive ink. B) Synthetized Ag-WPU ink with paste-like consistency. C) Soft circuits fabrication
process. Conductive lines are printed using (i) stencil printing or (ii) direct ink writing using a digital printer. (iii) Microchips and SMD components
are placed on the printed circuit. (iv) Droplets of the Ag-WPU conductive compound are dispensed on the microchip terminals and dried at room
temperature, forming a strong and reliable electromechanical bonding between the rigid component and the underlying conductive tracks. D) (i) Detail
of an SMD flat cable connector bonded to a set of printed conductive lines ( um pin separation). This circuit can work as a flexible multisignal
conductor which is (ii) ultraflexible, (iii) highly deformable, (iv) and can be molded to nonplanar surfaces to work as a MID. E) (i) Example of digitally
printed circuit traces (ii) populated with  size SMD resistors and LEDs. Line width from top to bottom is , , and  μm. (iii) Detail of the
soft bonding points between a  size LED and the underlying conductive tracks. F) (i) The circuit remains functional even when (ii, iii) conformed to
nonplanar surfaces or (iv, v) highly deformed, withstanding crumpling, as well as °and °bends.
In addition, we conducted an investigation into the influence
of silver concentration on the conductivity of ink. To achieve this,
we varied the weight percentage (wt%) of Ag flakes relative to
the solid contents of the WPU. Our findings align with existing
research on percolative networks. We observed that when the sil-
ver concentration in the ink falls below 75%, the printed traces do
not exhibit conductivity. However, as the concentration surpasses
90%, we noted a rapid increase in conductivity, as depicted in
Figure S2 (Supporting Information). It is important to note that
beyond 90% Ag wt%, the ink becomes excessively viscous, resem-
bling a thick paste. This characteristic renders extrusion print-
ing impractical and may also affect the mechanical properties of
the ink due to the decreased polymer content. Nevertheless, we
acknowledge the potential for further refinement of the ink for-
mulation to achieve an optimal balance between higher electrical
conductivity and rheology that suits DIW.
A batch of ink was fabricated, divided into closed vials, and
stored at room temperature without special handling or storing
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (4 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 3. A) Ag-WPU conductivity and aging of exposed printed traces for up to  days. B) Ag-WPU conductivity for ink vials stored up to  days at room
temperature. Error bars represent standard deviation C) Strain versus resistance curve for Ag-WPU traces printed over a thermoplastic polyurethane
(TPU) substrate (three samples from distinct ink batches). D) Strain (–%) vs stress plot for the three printed ink samples. E) Strain at break for the
three ink samples. F) Estimated Young modulus for the  ink samples. G) Cyclic test of the Ag-WPU ink. H) Printed Ag-WPU track with integrated  Ω
SMD resistor at % strain. I) Printed Ag-WPU track with integrated  ΩSMD resistor at % strain, before mechanical failure (electrical failure had
already occurred). J) Strain versus resistance curve for Ag-WPU traces printed over TPU substrate with integrated  ΩSMD resistors (three samples
from distinct ink batches). K) Strain (–%) versus stress plot for the three printed ink samples with integrated resistors. L) Strain at break for the three
ink samples with integrated SMD resistors. M) Mechanical fracture of the samples occurs at the interface between soft printed lines and the rigid SMD
component.
precautions. The vials were exposed to sunlight through windows
and artificial light on a benchtop to test their shelf life. As ob-
served in Figure 3B, the conductivity of the aged ink lies between
1.16 ×105Sm
−1for the freshly synthesized ink, and 1.62 ×105
Sm
−1for the ink stored for 30 days. Conductivity increase is due
to water evaporation during storage, which increases the Ag flake
concentration.
Viscosity also increases over time, but it does not affect print-
ability for the first 30 days. Figure S3 (Supporting Information)
shows the relationship between the viscosities of four Ag-WPU
samples that were stored at room temperature in closed vials for
different periods (0, 15, 30, and 35 days) at different shear rates.
Although in the first 30 days, the viscosity of the inks at a 0.1
s−1increases slightly from 125.65 to 220.91 Pa s, the ink is still
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (5 of 16)

www.advancedsciencenews.com www.advancedscience.com
printable both by stencil printing and DIW. However, between
30 and 35 days of storage the viscosity increases rapidly to 789.79
Pa s due to the curing of the polyurethane caused by low water
content, rendering the ink unusable as it is too viscous for reli-
able printing. Lower temperature storage can address this issue
by reducing water evaporation and keeping the ink’s viscosity sta-
ble for longer periods. In addition, all four samples exhibit shear
thinning behavior which proves beneficial in DIW.
2.2.2. Electromechanical Characteristics of Ag-WPU Ink
As observed in Figure 3C, the normalized resistance (R/R0)for
three Ag-WPU printed traces remains below 2 ΩΩ
−1when the
sample is stretched up to ≈22%, for two of the samples, while the
third sample shows a normalized resistance below 2 ΩΩ
−1up to
≈30% strain. At larger strains, the resistance increases rapidly for
all three samples being that two of the samples fail at 28.3% strain
and one of them at 33.5% strain. Electrical failure was considered
to occur when the normalized resistance increased up to 100 Ω
Ω−1.
To further explain the visible difference between the 3 reported
samples in Figure 3C, six more samples were fabricated from
separate ink batches and tested in the same conditions. As shown
in Figure S5 (Supporting Information), in 5 of the 9 Ag-WPU
samples the normalized resistance (R/R0) remains below 2 ΩΩ
−1
at strains between 22% and 27%, while one of the samples shows
R/R0>2ΩΩ
−1at strains higher than 15%, and a last sample
shows R/R0>2ΩΩ
−1only at strains larger than ≈30%.
The observed variations among the samples are hypothesized
to be a result of minor inconsistencies inherent to the sample
preparation process. This encompasses a number of factors
including inconsistencies in the preparation of the ink batches.
The ink preparation process can be affected by variations in
mixing or ambient room temperature and humidity, which
can lead to disparities in the distribution of silver flakes within
the polyurethane matrix. In turn, these disparities could be a
contributing factor to alterations in the percolation threshold.
The percolation threshold—a pivotal factor in the composite’s
conductivity and strain sensitivity—is thus subjected to variation,
culminating in the discernible differences in these properties
among all samples. Moreover, the fabrication process of the
dogbones introduces another layer of variability. The procedure
entails printing on a thermoplastic polyurethane (TPU) sub-
strate, a process that may introduce minor variations due to
factors such as printing repeatability, the specific conditions of
the printing process, or slight inconsistencies in the TPU sub-
strate itself. Collectively, these factors might lead to additional
variability in the final samples.
In terms of the mechanical properties of the ink, the strain–
stress curve between 0% and 35% strain for all three tested sam-
ples is shown Figure 3D. For all samples, the yield strength oc-
curs at around 30% strain, with a magnitude of ≈738–816 kPa.
As shown in Figure 3E, the mechanical failure of the samples oc-
curred between 453.8% and 484.5% strain (much above the elec-
trical failure strain), and the young modulus of the Ag-WPU ink
(plotted in Figure 3F) was estimated to be in the 10.5–12.8 MPa
range.
To test the cyclic performance of the Ag-WPU composite, it
was first stretched by 5% and then relaxed to its original length,
and the strain was increased by 5% up to 30% for each succes-
sive cycle as shown in Figure 3G. While the maximum stress for
each cycle follows the same values as for the strain–stress curve
in Figure 3D, a hysteresis loop between loading and unloading
cycles can be seen, which has been previously observed in other
PU-based compounds.[52] As well, after 20% strain, some degree
of plastic deformation can be observed, leading to a permanent
stretch of ≈3.5% in the sample for the following cycles.
The SEM images in Figures S6 and S7 (Supporting Informa-
tion), corresponding to Ag-WPU show the structure and mor-
phology of the ink, where only silver flakes are observable. Ag
flakes have a disk-like shape with widths below 5 μm. In terms
of orientation, the silver flakes appear randomly oriented in the
sample, with no predominant alignment or preferred direction
observed. The silver flakes are uniformly and evenly distributed
across the entire sample, without visible clusters or void areas.
2.2.3. Electromechanical Characterization of Soft Solder Joints
A solid-state resistor was embedded in the printed traces using
the proposed soft soldering method as depicted in Figure 3H
and stretched (Figure 3I) until mechanical failure. As shown in
Figure 3J, the normalized resistance for three printed samples
with integrated SMD resistors remains below 2 ΩΩ
−1for strains
of up to 20%, while full electrical failure (normalized resistance
above 100 ΩΩ
−1) occurs above 31% strain for all three samples.
The plot in Figure 3K provides evidence that the integration of
the solid-state chip does not significantly affect the morphology of
the strain–stress curve when compared with samples that do not
contain an integrated resistor. Yield occurs at around 30% strain,
with a magnitude of ≈775–854 MPa for the 3 tested samples. Fur-
thermore, mechanical failure of all three samples occurred above
386.5%, once again, much above their electrical failure, as plotted
in Figure 3L. Figure 3M depicts the failure mode of the samples
with integrated solid-state technology (SST) resistors. In all three
samples, mechanical fracture occurred at the interface between
the printed lines and the resistor. Nevertheless, no delamination
of the Ag-WPU ink from the TPU substrate was ever observed,
despite the occurrence of buckling in the printed ink layer due to
plastic deformation at high strains. Plots of the full strain–stress
curves until fracture are provided in Figure S8 (Supporting Infor-
mation). Video S2 (Supporting Information) shows a printed cir-
cuit with integrated SMDs being stretched and crumpled without
delamination of the rigid components or the conductive traces.
2.3. Circuit Separation and Recycling
One of the advantages of the Ag-WPU conductive polymer is that
it can easily be degraded through a simple decomposition process
and the surface mounted circuit components can be individually
recovered. To do this, the soft circuits are soaked in an IPA-filled
beaker, as shown in Figure 4A(i,ii). The beaker is then placed on a
magnetic stirrer and its content is stirred for 30 min at 500 RPM,
as seen in Figure 4A(iii). At this stage, one can observe that the
solution turns gray due to the disintegration of the conductive
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (6 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 4. A) Circuit degradation and separation process. The circuit is soaked in IPA (i, ii) and stirred in a magnetic stirrer for  min (iii). At this
point, the clean TPU substrate (iv) and rigid components (v) can be removed from the solution. The suspended Ag flakes and PU residues are left to
precipitate (vi) and are decanted, while any remaining IPA traces is then evaporated. B) Separation eciency. While the rigid electronic components and
TPU substrate can be fully recovered with trace amounts of ink residues, the ink (Ag flakes and PU residues) can be separated with ≈% eciency due
to losses during decanting. C) SEM image depicting no changes in morphology of the Ag flakes before and after the separation process (scale bar 
μm).
ink from the substrate. As shown in Figure 4A(iv,v). The TPU
substrate as well as the rigid components can be removed from
the beaker with just trace amounts of ink and the Ag flakes and
conductive ink residues are left to precipitate for 1 h, as shown
in Figure 4A(vi). The IPA can then be decanted and evaporated,
leaving the ink residue and Ag flakes. Soaking of the circuits in
water was also tested but no degradation was observed in this
case.
The recycling process described above is based on the way
IPA interacts differently with TPU and WPU due to their under-
lying chemical compositions and structure. Despite both being
polyurethanes, TPU has a highly crosslinked structure compared
to WPU and, when in contact with IPA, it can cause the TPU to
swell as the alcohol molecules get lodged between the polymer
chains, but it does not break down the actual chains. Once the al-
cohol evaporates, the TPU returns to its original state. WPU, on
the other hand, is designed to be dispersible in water thanks to
a less crosslinked structure and the presence of polar groups—
hydrophilic segments—which are not as prevalent in TPU, mak-
ing WPU more vulnerable to other solvents, specifically polar sol-
vents, such as IPA.[53] As such, when IPA comes into contact with
WPU, it interacts with the polar soft segments breaking the poly-
mer chains apart, causing the material to dissolve or degrade.
As shown in Figure 4B, the process is efficient in enabling
the recovery of all of the TPU substrate with trace amount of
ink residue, as well as all the rigid SMD components that can
be cleaned by spraying them with IPA above the recovery beaker.
The precipitation, decantation, and IPA evaporation processes al-
low for ≈91.18% of the initial ink (by weight) to be recovered,
while ≈7.78% of it is lost during processing (the remaining 1.04%
is bonded to the TPU substrate). This separation efficiency esti-
mate was obtained by comparing the weight of recovered compo-
nents relative to the amount of components in the initial circuit.
From the SEM image in Figure 4C, it can be observed that the
morphology of the Ag flakes in the ink is not affected by the sep-
aration process, indicating that the flakes can be further recycled
and reused. This circuit degradation and separation method is
intended for separating the soft circuits into their basic compo-
nents (shown in Figure S9, Supporting Information) so that some
components, such as the PU substrate, and rigid components can
be directly reused in other circuits. This enables the main objec-
tive of achieving a simple recycling process that is eco-friendly
through the use of IPA, which has negligible toxicity when at low
concentrations.
To enable reuse of aggregates of Ag flakes formed during the
recycling process, the aggregates were soaked in clean IPA, left
to precipitate, and decanted four times to allow most of the PU
residues to be separated from the silver. After the fourth wash,
the IPA is decanted, and the silver is dried at room temperature.
At this point, large pieces of PU residue (still with trace amounts
of Ag), as seen in Figure S10 (Supporting Information), can be
removed with a pair of tweezers. The recovered Ag-rich powder
(Figure S11, Supporting Information) can then be used to make
a new ink by mixing them in pristine WPU dispersion following
the initial method and quantities. The full process for washing
the Ag flakes and recycling them in a new ink is detailed in the
Experimental Section. The downside of this method for washing
the Ag Flakes multiple times in IPA is that, at the end, around
19.83% of the initial ink is lost in the process. As described in
the Experimental Section, from an initial chunk of ink weighting
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (7 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 5. A) Inclusion of liquid metal (EGaIn) in the Ag-WPU ink leads to a biphasic stretchable conductive compound. B) Strain versus resistance curve
for Ag-EGaIn-WPU traces printed over TPU substrate (three samples from distinct ink batches). C) Strain versus stress plot for the three printed ink
samples. D) Strain at break for the three ink samples. E) Estimated Young modulus for the  ink samples. F) Cyclic test of the Ag-WPU ink. G) Strains
versus resistance curves for both Ag-WPU and Ag-EGaIn-WPU.
12 g (containing 10.701 g of pristine Ag flakes and 1.298 g of
polymer), we were able to recover 8.89 g of Silver powder with PU
residues (which was used in the recycled ink), as well as 0.73 g of
large PU chunks with silver residues that were discarded. More
efficient separation processes, such as electrowinning[54] could
also be employed for recovery of the rest of the Ag, at the cost of
higher toxicity and increased complexity.
After printing the recycled ink, we observed changes in the
morphology of the stencil printed traces. As shown in Figure S12
(Supporting Information), some PU residue from the initial ink
are “trapped” inside the newly synthesized polymer, leading to a
nonuniform rough surface, compared to the smooth surface of
the initial printed traces. Nevertheless, the traces printed with re-
cycled ink presented a conductivity of 1.13 ×105Sm
−1,which
represents a decay of only ≈2.4% compared to the conductivity
of the pristine ink (1.16 ×105Sm
−1, as shown in the previous
section).
In terms of electromechanical properties, as shown in Figure
S13 (Supporting Information), the recycled ink shows reduced
stretchability when compared to the ink made of pristine silver.
While the initial Ag-WPU tracks can withstand more than 25%
strain without impacting the conductivity, in the recycled coun-
terpart the resistance is shown to increase rapidly above 20%
strains, leading to loss of functionality. While the issue of reduced
stretchability of the recycled ink could be solved through the in-
clusion of LM, this would come at the cost of less straightfor-
ward and eco-friendly recycling. In applications that do not re-
quire stretchability but only flexibility (for instance smart labels
as shown later), the benefits of scalable low-cost fabrication, and
straightforward eco-friendly recycling process surpass the disad-
vantage of reduced stretchability, given that high electrical con-
ductivity is maintained.
Figure S14 (Supporting Information) compares the strain–
stress curves of the pristine TPU film that is used as a substrate,
with the TPU film that is recovered after undergoing the IPA-
based circuit-degradation method. As can be observed, the pro-
posed method for separation of the circuit’s components does not
impact the mechanical properties of the TPU film, which can be
directly reused as a substrate for new prints after being dried.
This happens since the IPA only leads to swelling of the TPU
film, which is reversed after all IPA is evaporated.
2.4. Liquid Metal Inclusion
To overcome the electrical failure of the Ag-WPU ink at low
strains (<30%), the inclusion of eutectic gallium-indium
(EGaIn) in the conductive compound was tested, aiming at cre-
ating a biphasic structure which, ideally, could deform without
impacting its conductivity (Figure 5A). Details of the synthesis
of the Ag-EGaIn-WPU biphasic mixture are presented in the
Experimental Section. As observed in Figure 5B, the normal-
ized resistance (R/R0) for three Ag-EGaIn-WPU printed traces
remains below 2 ΩΩ
−1when the sample is stretched up to
≈20%, for two of the samples, while the third sample shows a
normalized resistance below 2 ΩΩ
−1up to ≈24% strain. At larger
strains, the resistance increases for all three samples up until
electrical failure at 203%, 222%, and 323.5% strain, respectively.
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (8 of 16)

www.advancedsciencenews.com www.advancedscience.com
This failure was defined as occurring when the normalized
resistance increased up to 100 ΩΩ
−1. The presence of EGaIn
in the ink allows the samples to stretch ≈10 times more before
electrical failure when compared to the ink without liquid metal.
Similarly, the electrical conductivity, reported in Figure S15
(Supporting Information), decreases slowly at a constant rate up
to 200% strain from ≈105Sm
−1down to 104Sm
−1for 2 of the
samples, while for a 3rd sample the conductivity decrease occurs
only at ≈400% strain. After these limits, the conductivity starts
to decrease rapidly until complete electrical failure is observed.
Interestingly, as depicted in Figure S2 (Supporting Informa-
tion), the integration of EGaIn in the Ag-WPU ink does not lead
to a noticeable change in the bulk initial conductivity of the ink
which remains stable between 0% and 75% EGaIn wt%. Instead,
the role of EGaIn in the ink is to create conductive paths that
keep the percolative network created by Ag flakes stable at larger
strains that when no EGaIn is used.
In terms of the mechanical properties of the Ag-EGaIn-WPU,
the full stress–strain curves for the three tested samples are
shown in Figure 5C. As shown in Figure 5D, the mechanical fail-
ure of the samples occurred between 454.3% and 483.7% strain,
similar to what occurs in the samples without liquid metal. The
elastic modulus of the Ag-EGaIn-WPU ink (Figure 5E) was esti-
mated to be in the 13.79–15.9 MPa range, which is slightly higher
than when EGaIn is not present. No smearing of EGaIn from the
polymeric compound was observed during the tests.
In Figure 5F, the cyclic performance of the Ag-EGaIn-WPU
composite is shown. The sample was first stretched by 5% strain
and then relaxed to its original length, and the strain was in-
creased by 5% up to 30% for each successive cycle. While the
maximum stress for each cycle follows similar values as for the
strain–stress curve, a hysteresis loop between loading and un-
loading cycles can be observed. After 20% strain, plastic defor-
mation appears to occur, leading to a permanent stretch of ≈3.5%
in the sample for the following cycles, similar to the ink without
EGaIn.
Figure 5G compares the strain–resistance curve for both poly-
mers (with and without EGaIn). From the plots the resistance of
the printed Ag-EGaIn-WPU traces increases much more slowly
with stretch compared to Ag-WPU traces. Moreover, the bipha-
sic Ag-EGaIn-WPU mixture can support much greater strains
(i.e., >200%) before losing conductivity. However, these advan-
tages in material performance are at the expense of greater
complexity with recycling EGaIn-filled composites. When EGaIn
is present in the conductive ink, the proposed IPA-based sep-
aration process no longer works since EGaIn alloys with the
Ag flakes. In this case, we must modify our recycling pro-
cess to incorporate the complex EGaIn separation processes
described.[55]
In order to compare the cyclic stability of the inks, specimens
of Ag-WPU and Ag-EGaIn-WPU were subjected to repetitive
loading up to a strain of 10%, as depicted in Figure S16 (Support-
ing Information). The Ag-WPU sample exhibited an escalating
trend in relative resistance with each successive loading cycle,
resulting in a 23-fold increase in the relative resistance within
the 50 cycles. Conversely, the sample incorporating liquid metal
demonstrated enhanced stability to cyclic loading during the ex-
amination, displaying merely a 1.5-fold rise in its relative resis-
tance after 50 loading cycles.
Despite the mechanical hysteresis observed in the samples,
particularly in Figures 3G and 5F for the Ag-WPU and Ag-EGaIn-
WPU materials, it is important to note that the impact of hystere-
sis on the electrical performance of thin film electronics can vary
depending on the specific application. In the case of the Ag-WPU
material, which exhibits lower stretchability and cyclic stability,
the observed hysteresis may not have a significant impact on its
electrical performance in certain applications. This ink formula-
tion, although less stretchable, offers a level of flexibility that can
be advantageous in certain scenarios. For instance, in the context
of smart labels integrated into thin plastic wraps (as shown later),
where moderate flexibility is required, the Ag-WPU material can
be suitable. The hysteresis in this case is unlikely to hinder the
functionality of the electronics.
On the other hand, if the application demands stretchability
and higher stability under cyclic loading, the ink formulation
containing EGaIn becomes more appropriate. The presence of
EGaIn in the Ag-EGaIn-WPU ink enhances the stretchability and
mechanical stability of the resulting thin film. This ink formula-
tion demonstrates lower hysteresis, enabling better performance
under repeated stretching and cycling.
Finally, we observe that the microstructure of the Ag-EGaIn-
WPU ink contains agglomerates of liquid metal that are fully en-
capsulated in PU, leading to a relatively rough surface for the
printed traces compared to Ag-WPU (Figure S17A, Supporting
Information). To prevent nozzle clogging by the EGaIn agglom-
erates, the smallest usable nozzle for direct ink writing with Ag-
EGaIn-WPU composite (Figure S17B, Supporting Information)
is 250 μm in diameter. Further research is necessary to fully un-
derstand the processes behind such EGaIn-induced agglomera-
tion.
SEM and EDS imaging of Ag-EgaIn-WPU ink (Figures S18
and S19, Supporting Information) shows a clear separation of Ag
and Ga, while In is present all over the sample. This can be ex-
plained by the high affinity between Ag and In. Ag bonds to In,
which is pulled out of EgaIn toward Ag particles, helping to an-
chor the EgaIn particles.
Moreover, in Figures S20 and S21 (Supporting Information),
the formation of intermetallic Ag-In microparticles can be ob-
served. These correspond to AgIn2, which is formed at tempera-
tures below 100 °C and, while having the same size of Ag flakes
(<5μm), Ag-In particles have a distinctive spherical shape.
2.5. Applications
The Ag-WPU materials architecture and fabrication techniques
enable the rapid implementation of soft circuits with integrated
microchips and SMD components that can be easily recycled.
Possible application domains for this system include recyclable
product packaging and on-skin electronics for health monitoring,
which rely on highly flexible, skin-conformal circuits that must
be comfortable to wear. As a first use-case, we present a fully recy-
clable smart package that continuously monitors the temperature
of perishable products during handling and provides informa-
tion to consumers regarding any mishandling or previous stor-
age in nonideal conditions. For the second use-case, we present
a family of physiological sensing stickers based on Ag-WPU ink
and integrated microchips for the acquisition of multiple digital
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (9 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 6. Fully recyclable smart packages. A) Printed recyclable smart label integrated in a plastic water bottle. B) A printed smart label integrated in
a package of perishable fresh food monitors its storage temperature over time and informs consumer through two LEDs whether the item is safe to
consume or whether there is the risk of contamination due to poor refrigeration. Historic and current temperature data can be transmitted via BLE to a
smartphone. C) Both the food package and the integrated smart label can be fully recycled using conventional means and the proposed separation and
recycling method, respectively. Materials that can be recycled through conventional methods are identified by their United Sates recycling codes (PET –
; PVC – ; other plastics – ; paper – ).
biomarkers, including continuous axillary temperature monitor-
ing, single-lead electrocardiography (ECG) monitoring, and sur-
face electromyography (sEMG) recording for hand gesture and
facial expression recognition. In these demonstrations, the Ag-
WPU ink formulation (without liquid metal) was used to enable
fully recyclable systems.
2.5.1. Fully Recyclable Smart Packaging
Foodborne illness caused by ingestion of contaminated food can
pose a serious public health threat. One of the most prominent
causes for contamination is related to poor refrigeration of fresh
and perishable products since bacteria can multiply rapidly when
products are left above 5 °C for extended periods.[56] Aprinted
smart label is introduced to monitor the storage temperature of
perishable products. The recyclable label can be integrated into
various packaging forms, such as a recyclable water bottle (Figure
6A) or a fresh fish package (Figure 6B), to help merchants and
consumers identify poor handling rapidly. As shown in Figure 6B
for a package of fresh salmon, a green LED is lit up whenever the
perishable good has been properly stored at temperatures below
5°C. If the temperature rises above 5 °C for longer than 1 h, a red
LED lights up and stays on even if the package is later brought
back to a safe storage temperature, as shown in the temperature
plots from Figure S22 (Supporting Information). The BLE con-
nection to the smart tag allows the user to review the historical
temperature data as well as the current storage temperature of
the package. The functioning of the smart label system is shown
in Videos S3 and S4 (Supporting Information).
As shown in Figure 6C, the full package and smart label can
be recycled. The plastic package is composed of PET, PVC, and
paper that can be recycled in conventional ways (as identified by
the United States recycling codes in the figure), while the smart
label can be recycled through our proposed method for recycling
the printed electronic circuits.
2.5.2. Body Temperature Monitoring Wearable
A microcontroller-based system with WiFi capabilities was de-
signed to measure body temperature through a thermistor. The
circuit shown in Figure 7A was digitally printed through direct
ink writing over a soft TPU substrate using the Ag-WPU conduc-
tive polymer and it was then transferred to a double-side medical-
grade adhesive film. All rigid SMD components were bonded
to the printed lines through the proposed soft soldering process
(Figure 7B) and the outline of the patch was laser-cut.
After calibrating the system, the patch was transferred to the
user’s chest, with the measuring thermistor placed directly on
the user’s axilla, as shown in Figure 7C. The sensor was worn
for 1 h, while the user worked on a computer sitting at a desk,
and it showed no signs of delamination or functionality issues.
The temperature was measured by the patch at a frequency of
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (10 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 7. A) Design of multilayer printed patch which includes a microcontroller, LEDs, voltage regulator and analog signal conditioning circuitry for
thermal sensing through a skin-contact thermistor. B) Detail of the printed circuit where the interface between rigid SMD components and the soft printed
circuit can be observed. C) Thermal sensing patch adhered to the skin, with the thermistor in the armpit region. D) Axillary temperature measured in
the armpit for  h using the patch (in red, data acquired every  s, filtered with a moving average filter with a  s window) and a ground truth obtained
using a commercial digital thermometer (in blue, data acquired every  s).
0.2 Hz. As ground truth, the temperature was also measured us-
ing a commercial digital thermometer. Figure 7D shows the ac-
quired signals, indicating that the patch temperature follows the
ground truth measurement trend. The average temperature for
both the patch and ground truth is 36.2 °C.
2.5.3. Multielectrode Patch for Electrophysiology
To record biopotentials from the body, a soft patch with three con-
ductive electrodes was printed on a TPU substrate, and a SMD
connector and medical adhesive were added to the patch to en-
sure good contact with the skin, as detailed in the Experimental
Section. The patch’s structure is displayed in Figure 8A.
The multielectrode patch was placed over the flexor carpi radi-
alis muscle in a volunteer’s forearm to acquire electromyography
(EMG) signals, as shown in Figure 8B. The Ag-WPU was also
used to detect surface EMG activity produced by a finger perform-
ing flexion or extension, as shown in Figure 8C. When gripping a
handheld dynamometer (Figure 8D), the wrist flexors all contract
simultaneously and generate EMG signals, as shown in the plots
from Figure 8E, with the amplitude growing proportionally with
the applied gripping force.
The same bioelectronic patch can also be placed on the fore-
head near the temple, as shown in Figure 8F. In this case, the
task consisted of performing twice the same sequence of facial
expressions: smiling, jaw clenching, eye blinking, and eyebrow
flashing, before returning to a resting state. In Figure 8G, the top
plot shows evidence of the repeatability of the signal morphology
for each facial expression across the two repetitions. Likewise, the
bottom plot corresponding to the amplitude RMS envelope of the
above signal shows that both the amplitude and signal shape is
the same for the repetitions and distinct enough among different
actions.
In the above facial gesture case, the various actions performed
in each task are quite distinct from one another and similar be-
tween repetitions, as evidenced by the Dynamic Time Warping
plots in Figures S23–S26 (Supporting Information). Figure S27
(Supporting Information) shows the Euclidean distance between
pairs of facial EMG signals for the 2 repetitions of the task, which
correlates to the similarity between pairs of signals. In this figure,
the lowest distance between two gestures of distinct repetitions
is highlighted, and accurately corresponds to the same gesture
in the two repetitions for the four tested cases. In this sense, a
simple classifier based on dynamic time warping could be reli-
ably employed for basic task classification and their eventual ap-
plication in human–machine interfaces.[57–59 ] Moreover, the pre-
sented use cases provide a measurable way to assess everyday
motor tasks, serving as digital biomarkers of movement. Clini-
cians can adopt these biomarkers to monitor the progression of
neuromuscular disorders and other clinical conditions with mo-
tor symptoms.[60–63 ]
In Figure 8H, the same electrode patch as before was adhered
to the chest over the sternal portion of the left pectoralis major
muscle and used to acquire electrocardiography (ECG) signal,
which is shown in Figure 8I. In the depicted signal, the standard
features of a normal ECG wave can be observed: the QRS com-
plex, as well as P and T waves, which are labeled in the figure. The
RR interval (corresponding to the time between two consecutive
R peaks) was calculated as being 630 ms, from which a heart rate
(HR) of 95 bpm can be estimated. This HR value is within the
normal limits stipulated by physicians for a healthy heart.[64]
As shown in Figure S28 (Supporting Information), the
electrode-skin impedance of the printed Ag-WPU electrodes was
measured and compared to that of conventional Ag/AgCl elec-
trodes, which are commonly used in healthcare practice. One of
the conclusions that can be drawn is that the printed electrodes
show a lower intrinsic electrical resistance than the Ag/AgCl
material, evidenced by their lower impedance at high frequen-
cies (≈100 kHz). Nevertheless, Ag/AgCl electrodes present
lower interface resistance with the body, supported by the lower
impedance at low frequencies (20 kΩat 1 Hz, compared to 400
kΩof the Ag-WPU electrodes at the same frequency). Moreover,
the overall lower impedance of Ag/AgCl electrodes in the tested
frequency range (1 Hz to 100 kHz) indicates that they allow for a
low interface capacitance. This can be explained by the presence
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (11 of 16)

www.advancedsciencenews.com www.advancedscience.com
Figure 8. A) Printed multielectrode patch with integrated flat-cable connector for skin-surface electrophysiology. B) Electrophysiology patch adhered to
the right forearm over the flexor carpi radialis muscle. C) EMG signals and corresponding RMS amplitudes produced by the flexion/extension of dierent
fingers. D) Hand dynamometer used to measure hand grip strength. E) EMG signal and corresponding RMS amplitude envelope acquired during a task
consisting of squeezing the dynamometer with increasing gripping forces, with rests in between. F) Multielectrode patch adhered on the face near
the temple. G) Bioelectronic signals produced by various facial expressions and subtle movements, including smiling, jaw clenching, eye blinking, and
eyebrow flashing. Bottom plot shows the corresponding RMS amplitude envelope for each expression. H) Electrophysiology patch adhered to the user’s
chest, over the sternal portion of the left pectoralis major muscle, connected to a biopotential recording system and battery. I) ECG signal acquired with
the printed patch, where ECG features (P and T waves, QRS complex) can be observed. RR interval and heart rate (HR) can also be calculated.
of a wet hydrogel, which creates a soft ion-rich medium that
conforms to the skin and improves charge flow between the
body and the recording electronics. This contrasts with what
happens with dry printed Ag-WPU electrodes. Nonetheless, de-
spite their slightly higher impedance (which may be translated
to noisier measurements), the dry thin-film printed electrodes
show some advantages related to the straightforward fabrication
of user and application-specific biopotential recording patches
with custom number and positioning of electrodes as well as
intricate electrode shapes. Moreover, the fact that these printed
electrodes are dry (i.e., no wet gel or electrolyte-rich conductive
paste is interfacing them with the skin) decreases the chances of
electrode cross-talk or electrode-skin interface degradation over
time due to electrolyte drying, as previously discussed.[[65 ]]
3. Conclusion
In this study, we introduce a novel eco-friendly class of soft
conductive inks composed of combinations of water-based
polyurethane, silver, and liquid metal that exhibit high electrical
conductivity (1.1–1.6 ×105Sm
−1), is suitable for digital printing
methods like DIW, and can be printed on thin-film substrates
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (12 of 16)

www.advancedsciencenews.com www.advancedscience.com
to create highly flexible, high resolution (down to 200 μm line
width), sticker-like electronics. Contrary to previous works re-
porting printable conductive mixtures, the Ag-WPU ink remains
stable for up to 4 weeks at room temperature without affecting
conductivity or printability. By taking advantage of the robust
adhesive properties of the proposed conductive polymer, a sim-
ple “soft soldering” process is proposed allowing for direct in-
tegration of IC chips and miniaturized SMD components into
the printed circuits without the need of other electrically con-
ductive adhesives or complex bonding processes. Moreover, the
entire process can be performed at room temperature without
the need for sintering, resulting in low-cost and fast fabrication
of functional chip-integrated circuits. Electrical failure occurs at
≈30% strains, but the strong mechanical bonding between the
printed Ag-WPU traces and SMD components allows for circuits
to stretch up to 380% before mechanical failure.
Another key feature is that the Ag-WPU ink is fully recyclable
and compatible with environmentally sustainable practices: The
ink is dispersed in water and free of organic solvents, and the
chip-integrated soft circuits can be disassembled by soaking and
stirring in isopropyl alcohol, leading to the separation of the rigid
components, TPU substrate film, and ink residues. At this stage,
they can be properly discarded, reprocessed (in the case of the
ink), or directly reused (in the case of the SMD components and
substrate). To complete the circular life cycle, the separated ink
aggregates can be washed multiple times in IPA to separate Ag
flakes and PU residues. The recovered silver is then used to syn-
thesize a new ink that exhibits 97.6% of the initial ink’s conduc-
tivity. This straightforward recycling method can be implemented
without the need for complex equipment or harmful chemicals,
contrasting with the few degradation and recycling processes pro-
posed for soft electronics that are complex, labor intensive and
usually lead to environmentally harmful chemical byproducts.
In order to make the ink more mechanically robust, EGaIn was
added to form a biphasic composite. This led to a conductive ink
that can be stretched to more than 200% strain without electrical
failure. This added stretchability with EGaIn, however, is at the
expense of a more complex decomposition and recycling method,
which is needed to allow full separation of the metals.
We showcase the potential of this material system by creat-
ing on-skin bioelectronic patches that acquire multiple digital
biomarkers, including body temperature, and electrophysiolog-
ical signals related to heart activity and muscle activity during
motor tasks. Finally, we present a smart package in which Ag-
WPA circuitry is used to create an electronic printed label that
records the storage temperature of perishable items and warns
consumers about unsafe storage conditions that can lead to prod-
uct contamination. The full smart package is recyclable by em-
ploying both conventional methods (for the food packaging) and
the proposed separation and recycling method for the soft elec-
tronic temperature monitor. Together, these implementations
demonstrate the potential for Ag-WPA inks to support electronic
sticker functionality while reducing material waste.
Summing up, we demonstrate for the first time a conjunc-
tion of simple materials and methods that enable complex eco-
friendly soft electronic circuits with a fully circular life-cycle.
These materials and methods include the development of an
eco-friendly conductive polymer compatible with digital printing
methods, a facile method for room-temperature, sinter-free, re-
liable integration of SMD chips using the same conductive poly-
mer, and a method for straightforward and low cost separation
and degradation of the fabricated circuits that enables the recov-
ery and reuse of the old circuit’s constituents for use in new cir-
cuits across various fields.
4. Experimental Section
Conductive Ink Preparation:First, . g of an organic-solvent-free
WPU dispersion (U % solid content, Alberdingk) were transferred
to a  mL capacity glass vial. Next, . g of silver flakes (Silflake ,
Technic inc.) were mixed into the WPU solution (.: Ag:WPU wt%)
using a planetary mixer (Thinky AR-) for  min at  rpm. After mix-
ing in the Ag flakes, a thick paste was obtained. It was diluted by adding
. g of deionized water and the compound was again mixed in the plane-
tary mixer for  min at  rpm. The presented quantities lead to .%
concentration of Ag flakes in the final ink after all water was evaporated.
For the ink samples including liquid metal (Ag-EGaIn-WPU), first the
Ag-WPU ink as described above was synthesized. After this, . g of EGaIn
(.% Ga, .% In) were heated to  °C and added to the Ag-WPU ink.
After the inclusion of EGaIn, the polymer was mixed for  min at  rpm
using a vertical overhead stirrer instead of the planetary mixer, since the
vertical mixer reduces the size of EGaIn-PU agglomerates due to the higher
shear,thus preventing nozzle clogging. The process is shown in Figure S
(Supporting Information).
Stencil Printing:Using a COdesktop laser cutting system (Universal
Laser Systems VLS.), the shapes to be printed were patterned on the
stencil material (Blazer Orange Laser Mask, Ikonics Imaging). The stencil
was then adhered to the desired substrate and ink was spread using a
single edge razor blade. The stencil was immediately lifted and removed,
and the print was let to dry at room temperature for  min before further
processing steps. The used stencil material leads to printed ink layers with
 μm height.
Direct Ink Writing:The Ag-WPU solution was loaded into a syringe
barrel for direct ink write (DIW) printing. The ink was then dispensed over
a TPU film (Bemis  TPU hotmelt film) following defined circuit paths
using a Voltera ink dispensing system (Voltera V-One PCB printer). Next,
the film was left to dry at room temperature for  min before integrating
the microchips and SMD components. The direct ink writing process is
show in Video S (Supporting Information).
When printing with Ag-EGaIn-WPU, the nozzle diameter limit is  μm
since at lower gauges nozzles will clog due to LM-PU agglomerates.
Conductivity Measurement and Aging Tests:Each sample in the con-
ductivity test consists of five stencil-printed traces (dimensions  cm ×
mm× μm) printed over an FR rigid substrate (FR Substrate ″×
″, Voltera), as shown in Figure S (Supporting Information). At the each
end of each trace, a droplet of EGaIn was dispensed to reduce the contact
resistance between the printed polymer and the multimeter probes. The
resistance of each sample was measured by a digital desktop multimeter
(, Keithley) with a four-point probe. For each track,  measurements
were taken.
For the aging test of the printed lines, the print was stored uncovered
at room temperature. For the storage test, various ink batches were fabri-
cated and stored in closed glass vials at room temperature before printing
the samples over time. Each day, a new set of five traces was printed us-
ing the aged ink vial after mixing the ink for  min at  rpm, and its
resistance was measured after letting the print dry.
Tensile Testing:TPU samples were patterned into a standardized dog-
bone shape (Die C, ASTM D) by cutting with a COlaser cutter
(VLS.; Universal Laser System). For characterization of the Ag-WPU
ink, an ink trace was printed on each dogbone using a stencil, as previously
described. Dimensions are shown in Figure S (Supporting Information).
To analyze the electromechanical failure modes of the “soft solder” joints,
two conductive traces were printed on each dogbone, and  ΩSMD re-
sistors ( and  sizes) were bonded to the traces, as depicted in
Figure S (Supporting Information).
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (13 of 16)

www.advancedsciencenews.com www.advancedscience.com
Tensile testing was done on a materials testing machine (; Instron)
with a  N load cell to which the samples were clamped. For the qua-
sistatic strain–stress characterization and cyclic testing, the strain loading
rate was set to ≈.% s−until the sample broke.
To measure sample resistance, a digital desktop multimeter (,
Keithley) with a four-point probe was connected to a laptop via USB and
measurements were acquired using KickStart Instrument Control Software
(Version ... for windows ).
To estimate conductivity of the ink samples upon stretching, it was as-
sumed that the conductive materials were incompressible, maintaining
a constant volume during deformation. The volume (V) of the conduc-
tive sample is given by Equation () where lcorresponds to the sample’s
length, tto its thickness, and wto its width
V=l×t×w⇔t×w=V∕l()
The conductivity, assuming constant volume, is then given by Equa-
tion ()
𝜎=l
R×w×t⇔𝜎=l
R×(V
l)⇔𝜎=l
R×V()
It is important to note that the assumption of incompressibility of the
conductive materials results in only an approximate estimate of the sam-
ple’s conductivity upon stretching.
Separation and Recycling of Silver from Recovered Ink Chunks:To sepa-
rate the Ag flakes from the PU residues, the recovered ink aggregates were
soaked in clean IPA.  g of clean IPA was used for each  g of recovered
ink. After manually stirring the contents, these were left to precipitate for
 h, and the IPA was decanted. The process was repeated four times to
allow most of the PU residues to be separated from the silver. After the
fourth wash, the IPA was decanted, and the solid contents were dried at
room temperature and scraped for the wall and bottom of the container
using a spatula. The larger pieces of cured PU (still with traces of Ag) were
removed with tweezers and discarded. The remaining powder, consisting
of recovered Ag flakes and PU residues can then be used to make a new
ink by mixing . g of pristine WPU with . g of recovered silver powder
and . g of DI water).
To test the conductivity of the recycled ink,  g of cured ink pieces
(. g PU, . g Ag) were subjected to the recycling process. The
weight of solid contents after the fourth IPA wash was . g (.% of
the initial weight), from which . g of large PU chunks were discarded
and . g of Ag flakes-rich powder were recovered.
After synthesizing a new ink with the recovered Ag-rich powder, five
traces were stencil-printed and their conductivity measured, as described
above in Section .. Printed traces of recycled ink are shown in Figure S
(Supporting Information).
Smart Packaging:A circuit containing an IC digital humid-
ity/temperature sensor (HIH, Honeywell) and a Bluetooth low
energy (BLE) enabled microcontroller (Seeed Xiao nRF, Seeed Stu-
dio) was designed and printed using the proposed Ag-WPU conductive
ink, as shown in Figure S (Supporting Information). This circuit also
contains a . V rechargeable LiPo coin cell battery, SMD resistors and
capacitors, and two LEDs to provide information to the user.
To acquire temperature data over BLE, a debug app (nRF Connect,
Nordic Semiconductors) was installed in a smartphone (S G, Sam-
sung). Temperature and humidity data were transferred to the mobile
phone every time the app was paired with the printed smart tag.
For the fully recyclable smart package demo, a polyethylene terephtha-
late (PET) deli container was used. The transparent plastic cover was made
of polyvinyl chloride (PVC) plastic wrap.
On-Skin Temperature Monitor:A circuit containing a WiFi-enabled mi-
crocontroller (ESP), NTC thermistor ( kΩat  °C, ), and volt-
age divider for signal conditioning, as well as  LEDs was designed and
printed in a TPU substrate (Bemis  TPU hotmelt) through direct ink
writing. After printing, the flexible circuit was transferred to a double side
medical-grade adhesive (A Medical Transfer Adhesive,  ) and the
SMD chips were bonded in place. The system was powered through an
external . V LiPo battery connected by copper wires.
The temperature sensing patch was calibrated using a hot plate
(Cimarec, Thermo Scientific) with an attached digital thermometer (TP-,
ThermoPro), as shown in Figure S (Supporting Information). The patch
was placed on the hotplate, with the thermistor near the digital thermome-
ter.
The ADC output read from the microcontroller was sent to a nearby
computer via an ESP-Now wireless protocol. Simultaneously, the tempera-
ture of the hot plate was also recorded. The samples were tested between a
range of  and  °Cin°C increments. After waiting for  min between
each data point recording to ensure that the temperature had reached
equilibrium.  trials ( with the temperature increasing and  with tem-
perature decreasing) were performed and the calibration curve was then
calculated. Due to the narrow temperature range, the sensor’s response
could be approximated linearly as shown in Figure S (Supporting Infor-
mation).
The patch was adhered to the user’s chest after carefully cleaning the
skin with rubbing alcohol. Attention was given so that the thermistor
would be placed directly on the user’s axilla. Temperature was measured
every  s for a total of  h using the patch. Also, axillary temperature was
manually measured every  min using a digital thermometer (KD-,
 Innovations, . °C accuracy) in order to obtain a ground truth record-
ing. The data acquired from the patch was filtered using a moving average
filter ( samples) and the values were sent in real time to a nearby com-
puter via ESP-Now wireless protocol.
Biopotential Recording Electrodes:The three conductive electrodes
were printed simultaneously by stencil printing over a TPU substrate film
(Bemis  TPU hotmelt) and a flat cable connector (Molex -,
 mm pitch) was integrated in the patch using the described soft solder-
ing method. The electrode patch was then transferred to a double side
medical-grade adhesive (A Medical Transfer Adhesive,  ) that was
previously patterned to shape using a COdesktop laser cutting system
(Universal Laser Systems VLS.) and finally, the outline of the patch was
cut using the COlaser.
The patch was connected to a miniaturized biopotential recording sys-
tem previously presented[] through a flat cable and the electrophysiolog-
ical recordings were sent by the analog front end (AFE) via UDP protocol
and acquired by the OpenBCI GUI (v.. for Mac) and later converted to
CSV files. The full acquisition setup is shown in Figure S (Supporting
Information).
For recording of the biopotential signals, the user’s skin was first care-
fully cleaned with rubbing alcohol and let to dry, before laminating the
electrode patch and recording analog front-end into the recording site.
In this setup, acquired signals were sampled at  Hz and amplified
×. They were then filtered in MATLAB using a notch filter ( Hz) and
High/Low pass second-order Butterworth filters as needed for each mon-
itored signal. For ECG, the relevant frequencies were assumed to be in
the – Hz band. For surface EMG (both in the arm and face), the rel-
evant frequencies were in the – Hz band. For each acquired EMG
signal, its corresponding root-mean square (RMS) envelope was as well
calculated using a window with a length of  samples for the finger pose
experiment and  samples for the gripping force and facial expressions
experiments.
Electrode-Skin Impedance Measurement:The electrode-skin
impedance for the fabricated electrodes was measured using a Palm-
Sens impedance analyzer.  impedance points (≈. per decade) were
measured between and Hz. Before placing the electrodes on the
right inner forearm of the volunteer, the skin was cleaned by wiping with
rubbing alcohol and dried for  min. The electrode patch was then placed
and left to rest for  min, and the impedance measurement was taken.
After the measurement was complete, the electrodes were removed, and
any residue of adhesive was wiped o with rubbing alcohol.
Impedance of medical-standard Ag/AgCl electrodes (RedDot,  ) was
measured in the same frequency range by adhering three electrodes to the
user’s forearm after following the same skin cleaning procedure described
above. Gripping force was measured through a handheld dynamometer
(Digital Hand Dynamometer, Handeful).
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (14 of 16)

www.advancedsciencenews.com www.advancedscience.com
The experiments in human subjects were approved by the Carnegie Mel-
lon University Institutional Review Board (STUDY_) in ac-
cordance with the US HHS regulations for the protection of human sub-
jects in research at CFR . Informed consent was obtained from the
volunteer, who is also the first author of the manuscript, and all experi-
ments were performed in accordance with the applicable regulations.
SEM Imaging:To analyze morphologic changes in Ag flakes, samples
of pristine and recovered Ag flakes were prepared and imaged using a FEI
Quanta  FEG scanning electron microscope with an acceleration volt-
age of . kV and magnification of x. Images were taken on carbon
adhesive disks and the scale bar corresponds to  μm.
The Ag-EGaIn-WPU samples were immersed for  s in liquid nitro-
gen and fractured through a mechanical impact, leading to a clean cross-
section fracture. EDS surface scanning was used to build the color map of
the distribution of the elements (Ga, Ag, and In) on the sample’s surface.
Rheological Characterization:Rheological measurements were per-
formed on a DHR- stress-controlled rheometer (TA Instruments) at 
°C for Ag-WPU samples that had been stored at room temperature for –
 days. A  mm parallel plate configuration was used with a  μmgap
height.
Statistical Analysis:Unless otherwise stated, one measurement per
sample was taken. Data treatment was performed in Matlab_Rb and
Microsoft Excel V. for Mac.
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
Support for this research was provided by the Fundação para a Ciên-
cia e a Tecnologia (Portuguese Foundation for Science and Technol-
ogy) through the Carnegie Mellon Portugal Program under Grant No.
SFRH/BD//. Support also came from the European Com-
mission through the European Research Council project LiquidD | GA
 | ERC--COG and from the CMU-Portugal project WoW
(), which had the support of the European Regional Development
Fund (ERDF) and the Portuguese State through Portugal  and COM-
PETE .
Conflict of Interest
The authors declare no conflict of interest.
Data Availability Statement
The data that support the findings of this study are available from the cor-
responding author upon reasonable request.
Keywords
direct ink writing, e-waste, flexible electronics, green electronics, mi-
crochip integration, printed electronics, recyclable electronics, smart pack-
aging, soft circuits
Received: March , 
Revised: June , 
Published online: July , 
[] A. Kumar, M. Holuszko, D. C. R. Espinosa, Resour., Conserv. Recycl.
2017,122, .
[] K. Grant, F. C. Goldizen, P. D. Sly, M.-N. Brune, M. Neira, M. van den
Berg, R. E. Norman, Lancet Global Health 2013,1, .
[] M. Tavakoli, M. H. Malakooti, H. Paisana, Y. Ohm, D. G. Marques, P.
A. Lopes, A. P. Piedade, A. T. de Almeida, C. Majidi, Adv. Mater. 2018,
30, .
[] I. D. Joshipura, H. R. Ayers, C. Majidi, M. D. Dickey, J. Mater. Chem.
C2015,3, .
[] R. K. Kramer, C. Majidi, R. J. Wood, Adv. Funct. Mater. 2013,23, .
[] J. Alberto, C. Leal, C. Fernandes, P. A. Lopes, H. Paisana, A. T. De
Almeida, M. Tavakoli, Sci. Rep. 2020,10, .
[] G.Costa,P.A.Lopes,A.L.Sanati,A.F.Silva,M.C.Freitas,A.T.De
Almeida, M. Tavakoli, Adv. Funct. Mater. 2022,32, .
[] M. R. Carneiro, M. Tavakoli, IEEE Sens. J. 2021,21, .
[] S. Kang, J. Lee, S. Lee, S. Kim, J.-K. Kim, H. Algadi, S. Al-Sayari, D.-E.
Kim, T. Lee, Adv. Electron. Mater. 2016,2, .
[] S. Liu, D. S. Shah, R. Kramer-Bottiglio, Nat. Mater. 2021,20, .
[] D.-H. Kim, R. Ma, Y.-S. Kim, R.-H. Kim, S. Wang, J. Wu, S. M. Won,
H. Tao, A. Islam, K. J. Yu, T.-I. Kim, R. Chowdhury, M. Ying, L. Xu,
M. Li, H.-J. Chung, H. Keum, M. Mccormick, P. Liu, Y.-W. Zhang, F.
G. Omenetto, Y. Huang, T. Coleman, J. A. Rogers, Science 2011,333,
.
[] A. Libanori, G. Chen, X. Zhao, Y. Zhou, J. Chen, Nat. Electron. 2022,
5, .
[] T. V. Neumann, M. D. Dickey, Adv. Mater. Technol. 2020,5, .
[] M. Baharfar, K. Kalantar-Zadeh, ACS Sens. 2022,7, .
[] J. Yang, W. Cheng, K. Kalantar-Zadeh, Proc. IEEE 2019,107, .
[] W. Wu, Nanoscale 2017,9, .
[] B. Tian, W. Yao, P. Zeng, X. Li, H. Wang, Li Liu, Yu Feng, C. Luo, W.
Wu, J. Mater. Chem. C 2019,7, .
[] B. Tian, Y. Fang, J. liang, K. Zheng, P. Guo, X. Zhang, Y. Wu, Q. Liu, Z.
Huang, C. Cao, W. Wu, Small 2022,18, .
[] J. Wiklund, A. Karakoç, T. Palko, H. Yi˘
gitler, K. Ruttik, R. Jantti, J.
Paltakari, J. Manuf. Mater. Process. 2021,5, .
[] B. Ghosh, M. K. Ghosh, P. Parhi, P. S. Mukherjee, B. K. Mishra, J.
Cleaner Prod. 2015,94,.
[] A. Moin, A. Zhou, A. Rahimi, A. Menon, S. Benatti, G. Alexandrov, S.
Tamakloe, J. Ting, N. Yamamoto, Y. Khan, F. Burghardt, L. Benini, A.
C. Arias, J. M. Rabaey, Nat. Electron. 2021,4, .
[] P. Tan, H. Wang, F. Xiao, X. Lu, W. Shang, X. Deng, H. Song, Z. Xu, J.
Cao, T. Gan, B. Wang, X. Zhou, Nat. Commun. 2022,13, .
[] M. Held, A. Pichler, J. Chabeda, N. Lam, P. Hindenberg, C. Romero-
Nieto, G. Hernandez-Sosa, Adv. Sustainable Syst. 2022,6, .
[] S. P. Lee, G. Ha, D. Wright, Y. Ma, E. Sen-Gupta, N. R. Haubrich, P.
C.Branche,W.Li,G.L.Huppert,M.Johnson,H.B.Mutlu,K.Li,N.
Sheth, J. A. Wright Jr., Y. Huang, M. Mansour, J. A. Rogers, R. Ghaari,
NPJ Digit. Med. 2018,1,.
[] H. Lee, C. Song, Y. S. Hong, M. Kim, H. R. Cho, T. Kang, K. Shin, S.
H. Choi, T. Hyeon, D.-H. Kim, Sci. Adv. 2017,3, .
[] D. F. Fernandes, C. Majidi, M. Tavakoli, J. Mater. Chem. C 2019,7,
.
[] Y. Ohm, C. Pan, M. J. Ford, X. Huang, J. Liao, C. Majidi, Nat. Electron.
2021,4, .
[] Z. Huang, Y. Hao, Y. Li, H. Hu, C. Wang, A. Nomoto, T. Pan, Y. Gu, Y.
Chen, T. Zhang, W. Li, Y. Lei, N. Kim, C. Wang, L. Zhang, J. W. Ward, A.
Maralani, X. Li, M. F. Durstock, A. Pisano, Y. Lin, S. Xu, Nat. Electron.
2018,1, .
[] M. G. Saborio, S. Cai, J. Tang, M. B. Ghasemian, M. Mayyas, J. Han,
M. J. Christoe, S. Peng, P. Koshy, D. Esrafilzadeh, R. Jalili, C. H. Wang,
K. Kalantar-Zadeh, Small 2020,16, .
[] M. Vaseem, G. Mckerricher, A. Shamim, ACS Appl. Mater. Interfaces
2016,8, .
[] P. A. Lopes, D. F. Fernandes, A. F. Silva, D. G. Marques, A T. De
Almeida, C. Majidi, M. Tavakoli, ACS Appl. Mater. Interfaces 2021,13,
.
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (15 of 16)

www.advancedsciencenews.com www.advancedscience.com
[] W. Zu, Y. Ohm, M. R. Carneiro, M. Vinciguerra, M. Tavakoli, C. Majidi,
Adv. Mater. Technol. 2022,7, .
[] J. Li, C. M. Johnson, C. Buttay, W. Sabbah, S. Azzopardi, J. Mater.
Process. Technol. 2015,215, .
[] C.-W. Chang, T.-Y. Cheng, Y.-C. Liao, J. Taiwan Inst. Chem. Eng. 2018,
92,.
[] N. Ibrahim, J. O. Akindoyo, M. Mariatti, J. Sci.: Adv. Mater. Devices
2022,7, .
[] Z. Li, R. Zhang, K.-S. Moon, Y. Liu, K. Hansen, T. Le, C. P. Wong, Adv.
Funct. Mater. 2013,23, .
[] P. Lall, J. Narangaparambil, S. Miller, in Proc. ASME 2021 Int. Tech.
Conf. Exhib. Packag. Integr. Electron. Photonic Microsystems, InterPACK
2021, ASME, New York 2021, https://doi.org/./IPACK-
.
[] P. Lall, K. Goyal, K. Schulze, S. Miller, in Intersoc. Conf. Therm. Ther-
momechanical Phenom. Electron. Syst. ITHERM, IEEE,2021, pp. –
.
[] H. Wu, C. Yang, J. Liu, X. Cui, B. Xie, Z. Zhang, in Proc. Electron.
Packag. Technol. Conf. EPTC, IEEE,2014, pp. –, https://doi.
org/./ICEPT...
[] J. Lee, J. Sjoberg, D. T. Rooney, D. A. Geiger, D. Shangguan, in Proc.
IEEE/CPMT Int. Electron. Manuf. Technol. Symp, IEEE, 2008, https:
//doi.org/./IEMT...
[] A. Sudheshwar, N. Malinverno, R. Hischier, B. Nowack, C. Som, Re-
sour., Conserv. Recycl. 2023,189, .
[] E. Jansson, J. Lyytikäinen, P. Tanninen, K. Eiroma, V. Leminen, K.
Immonen, L. Hakola, Mater 2022,15, .
[] M. Dulal, S. Afroj, J. Ahn, Y. Cho, C. Carr, Il-D Kim, N. Karim, ACS
Nano 2022,16, .
[] V. Arroyos, M. L. K. Viitaniemi, N. Keehn, V. Oruganti, W. Saunders,
K. Strauss, V. Iyer, B. H. Nguyen, in Conf. Hum. Factors Comput. Syst.
– Proc, ACM., 2022, https://doi.org/./..
[] H. Borland, Y. Bhatti, A. Lindgreen, Corp. Soc. Responsib. Environ.
Manag. 2019,26, .
[] Ye Zhou, Z. Xu, H. Bai, C. E. Knapp, Adv. Mater. Technol. 2023,8,
.
[] F. Hoeng, A. Denneulin, J. Bras, Nanoscale 2016,8, .
[] Z. Xu, M. Wu, W. Gao, H. Bai, Adv. Mater. 2020,32, .
[] R. Sliz, M. Karzarjeddi, H. Liimatainen, T. Fabritius, in Proc. 2021 IEEE
11th Int. Conf. ’’Nanomaterials Appl. Prop. N. 2021, IEEE, Piscataway,
NJ 2021, https://doi.org/./NAP...
[] A. Islam, H. N. Hansen, P. T. Tang, J. Sun, Int. J. Adv. Manuf. Technol.
2009,42, .
[] M. Reis Carneiro, C. Majidi, M. Tavakoli, Adv. Eng. Mater. 2021,24,
.
[] X.Wang,X.Liu,D.W.Schubert,Nano-Micro Lett. 2021,13, .
[] S.Y.Chen,R.Q.Zhuang,F.S.Chuang,S.P.Rwei,J. Appl. Polym. Sci.
2021,138, .
[] Z. Wang, C. Peng, K. Yliniemi, M. Lundström, ACS Sustainable Chem.
Eng. 2020,8, .
[] M. Tavakoli, P. Alhais Lopes, A. Hajalilou, A. F.Silva, M. Reis Carneiro,
J. Carvalheiro, J. Marques Pereira, A. T. De Almeida, Adv. Mater. 2022,
34, .
[] L. M. Deac, J. Clin. Rev. Case Rep. 2020,5, .
[] M. Tavakoli, C. Benussi, J. L. Lourenco, Exp. Syst. Appl. 2017,79, .
[] M. Hamedi, S.-H. Salleh, M. Astaraki, A. Noor, Biomed. Eng. Online
2013,12, .
[] A. Moin, A. Zhou, A. Rahimi, S. Benatti, A. Menon, S. Tamakloe, J.
Ting, N. Yamamoto, Y. Khan, F. Burghardt, L. Benini, A. C. Arias, J. M.
Rabaey, in Proceedings – IEEE International Symposium on Circuits and
Systems, IEEE, Piscataway, NJ 2018.
[] S. M. Rissanen, M. Koivu, P. Hartikainen, E. Pekkonen, Clin. Neuro-
physiol. 2021,132, .
[] E. D. Pasmanasari, J. A. Pawitan, Biomed. Pharmacol. J. 2021,14,.
[] A. M. Kotov-Smolenskiy, A. E. Khizhnikova, A. S. Klochkov, N. A.
Suponeva, M. A. Piradov, Hum. Physiol. 2021,47, .
[] B.-Y. Youn, Y. Ko, S. Moon, J. Lee, S.-G. Ko, J.-Y. Kim, Diagnostics 2021,
11, .
[] F. G. Yanowitz, Introduction to ECG Interpretation, Intermountain
Healthcare, Murray, UT 2018.
[] M. Reis Carneiro, C. Majidi, M. Tavakoli, Adv. Funct. Mater. 2022,32,
.
Adv. Sci. 2023,10,  ©  The Authors. Advanced Science published by Wiley-VCH GmbH
2301673 (16 of 16)