1 © 2020 IOP Publishing Ltd Printed in the UK

1. Introduction

Stretchable electronics and flexible circuits are one of key tech-nologies for developing future robots especially involved with soft materials and structures. It has recently been paid more attention due to the increased possibility of physical interac-tion of robots with humans. Different methods have been proposed in order to make soft conductive structures, which

include wavy circuits [1, 2], carbon nanotube films [3, 4], silver nanowire embedded polymers [5, 6], and liquid-metal embedded soft structures [7–9]. Feng et al, have manufactured stretchable metal oxide silicon circuits using a wavy structure [10]. Takeo et al, have fabricated wearable and stretchable strain sensors for human-motion detection by using carbon nanotube films [11]. Amjadi et al, have developed conductive and stretchable polymer by embedding silver nanowires [12].

Among the aforementioned technologies, we focused on a particular class of conductive material, a room-temperature

Journal of Micromechanics and Microengineering

Direct printing of sub-30 µm liquid metal patterns on three-dimensional surfaces for stretchable electronics

Gyowook Shin1, Byungjun Jeon1 and Yong-Lae Park

Department of Mechanical Engineering; Institute of Advanced Machines and Design (IAMD); Institute of Engineering Research, Seoul National University, Seoul 08826, Republic of Korea

E-mail: ylpark@snu.ac.kr

Received 2 September 2019, revised 29 December 2019Accepted for publication 20 January 2020Published 13 February 2020

AbstractIn this study, a liquid metal is directly printed on various types of surfaces using an automated dispensing system. A particular class of liquid metals called eutectic gallium–indium (Ga: 75.5% In: 24.5% by weight ratio) was chosen and printed on flat, inclined (20°, 30°, 40°, and 50°), and curved (κ = 0.02, 0.03, 0.04, and 0.05 mm−1) surfaces. The inner diameter of the dispenser nozzle, the distance between the nozzle tip and the surface of the substrate, turned out to be the crucial parameters that determine the performance of printing, based on the experimental evaluation of the relationship between the trace width and the parameters. We were able to control the trace width under 200 µm as small as 22 µm by adjusting the parameters we tested. To the best of our knowledge, an EGaIn trace 22 µm in width is the smallest one achieved by direct printing of a liquid metal on three-dimensional (3D) surfaces. Also, we were able to print not only straight lines but also curved patterns, such as spiral shapes. This will lead to the miniaturization of stretchable electronics with any pattern shapes consisting of straight lines and curves. As an example of applications of the proposed method, a micro-scale pressure sensor with a spiral trace pattern was fabricated, and its performance was evaluated with loading and unloading tests. Another application of the proposed method includes direct printing of stretchable electronics on surfaces with arbitrary shapes and curvatures. It was demonstrated with a seven-segment display circuit and soft sensors printed on a mannequin hand. We believe the proposed method and its applications will open a new space in development of soft electronics and robots.

Keywords: stretchable electronics, soft robotics, liquid metal, direct printing, eutectic gallium–indium

S Supplementary material for this article is available online

G Shin et al

Direct printing of sub-30 µm liquid metal patterns on three-dimensional surfaces for stretchable electronics

Printed in the UK

034001

JMMIEZ

© 2020 IOP Publishing Ltd

30

J. Micromech. Microeng.

JMM

10.1088/1361-6439/ab6dbc

Paper

3

Journal of Micromechanics and Microengineering

IOP

1 Gyowook Shin and Byungjun Jeon contributed equally.

2020

1361-6439

1361-6439/ 20 /034001+10$33.00

https://doi.org/10.1088/1361-6439/ab6dbcJ. Micromech. Microeng. 30 (2020) 034001 (10pp)

G Shin et al

2

liquid metal called eutectic gallium–indium (EGaIn), one of main conductive materials used in research and development of soft electronics. It is applicable to stretchable electronics since it can be easily embedded into a polymer structure not affecting the stiffness and the properties of the host material [13, 14]. EGaIn is a mixture of gallium 75.5% and indium 24.5% by weight [15]. It is a conductive fluid at room temper-ature and known to be non-toxic unlike mercury. Up to date, various types of sensors using EGaIn have been developed for measuring changes in mechanical strain and pressure. Due to its conformability, sensors made of a liquid metal embedded in a soft structure can easily detect changes of resistance, capaci-tance, or inductance depending on their geometries [16–18]. It has even been embedded in or integrated with actuators for acquiring proprioceptive feedback information of the actua-tors directly [19, 20].

When manufacturing stretchable electronics with a liquid metal, the patterning technique of liquid metal traces on a soft material is critical [21]. Several methods have been pro-posed, and one of the most common processes is injection of the liquid metal into microchannels embedded in a soft material, such as silicone rubber [8, 22]. After manufacturing microchannels inside the soft material, needles are commonly used for injection. Although this method has an advantage of being low-cost, it is time-consuming and does not guarantee the product quality, since it involves multiple manual steps in the process. Moreover, due to the limitation in resolution of the molds that are usually made by three-dimensional (3D) printing, a channel width less than 200 µm is extremely dif-ficult to manufacture. Another method is lithography [23–27], in which a patterned mask covers the substrate, and the liquid metal is transferred on top [28]. Although this method has an advantage of further miniaturization compared with the injection method, the mask preparation process is costly and time-consuming. In addition, masked printing without lithog-raphy has been proposed by fabricating precise masks using a mechanical cutter and laser [29, 30]. Although masks can be easily removed using a water-soluble material [31] or mask-less printing using a selectively wettable stamps is possible, additional processes are still necessary.

Therefore, a simple but more reliable method has been developed by directly printing liquid metal traces on a poly mer substrate. In this method, complex patterns can be easily achieved on a polymer substrate without molds or masks [33, 34]. Direct printing of liquid metal is possible due to the presence of an oxide skin as soon as the liquid metal is exposed to the air. This oxide skin provides a stable structure of the trace during printing, since it makes good bonding to the polymer sub-strate maintaining the trace intact. Even though the oxide skin of EGaIn provides mechanical stability of the printed traces, EGaIn itself has relatively low wettability on silicone substrates. This results in high contact angles, making it challenging to directly print EGaIn [35]. Due to this issue, most of the results are with flat surfaces only. Boley et al, have demonstrated a capability of printing a trace of 83 µm in width [33]. Wu et al, have proposed a method of direct printing of liquid metal trapped inside viscoe-lastic ink achieving trace width less than 100 µm [36]. Kim et al, have developed a glove sensor using direct writing of a liquid

metal [35]. Dickey et al, and Park et al, found ways to form 3D structures of EGaIn on flat substrates [37, 38]. However, when a soft sensor in a form of a flat layer is applied to a curved sur-face, it is possible that the sensor signals become unreliable or the sensor itself may even fail. Since the sensor may experience undesired strain or sharp bending, resulting in loss of connec-tion or degradation of sensitivity while transferring the sensor to a 3D surface, it is highly useful if the liquid metal can be directly printed on the curved surface in 3D. Therefore, we pro-pose a method of printing EGaIn not only on a flat surface but also on an uneven surface with a controlled trace width using a sensor feedback system. Moreover, we were able to print liquid metal patterns consisting not only of straight lines but also of lines with continuous curvatures, such as spirals, which is chal-lenging without optimizing the values of multiple parameters. Using this method, we were able to reliably print EGaIn traces on the 3D surfaces, as thin as 22 µm in width. As a simple appli-cation, a pressure sensor with a spiral pattern was fabricated and characterized. We also demonstrate the capability of printing in 3D surfaces, which has a potential in developing stretchable electronics and wearable devices.

2. Method

2.1. Instruments and image acquisition

To print EGaIn directly on a substrate, a motorized x − y stage (Shotmaster 300ΩX, Musashi) and a pneumatic dis-pensing system (SuperΣ CMIII V2 , Musashi) were used with a laser distance sensor (LK-G32, Keyence) with a resolution of 0.4 µm, as shown in figure 1(a). The laser sensor measures the stand-off distance (SOD) which indicates the distance between the top surface of the substrate and the nozzle tip in z-axis. The range of distance sensing in z-axis was ±5 mm from the neutral level. The sensor data were used as feedback for controlling the SOD. The sensor collects data at every interpo-lation points which can be programmed using computer-aided design software (Mu-CAD, Musashi). The distance data for each pre-programmed points were collected before printing EGaIn. In this step, the sensor follows the exact same path as the pattern to be printed by moving the stage only in x and y axes. After all the data are collected, the printing procedure takes place, the robot arm provides appropriate motion in z-axis maintaining a constant SOD using the pre-recorded dis-tance data at each point. In this way, the SOD was controlled to remain the same throughout the whole printing procedure even on 3D surfaces. The maximum and the minimum pres-sure levels that the dispenser was able to generate was 200 kPa and 5 kPa, respectively, with a 0.1 kPa interval. The motorized stage moves in x and y axes while the single axis robotic arm provides motions in z-axis. The system receives a pre-planned path trajectory from CAD software and traces the path on the stage with a pre-determined pressure output.

To accurately measure the width of the printed trace, high resolution (1024 × 768 px) ‘png’ format images of the liquid metal traces were acquired using a microscope (SZX16-3111, Olympus) shown in figure 1(b). The EGaIn trace was magni-fied 40 times of its original size, and image capture software

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

3

(Leopard, Olympus) was used as a reference to compare the width measured from the software with that measured from Matlab using an edge detection process. A digital filter was used to change the original image to a grey-scale image, and the edges of the traces were found using the Sobel algorithm [39]. The actual width was measured by counting the number of pixels between the two edges of the trace. A trace with the same parameters was printed six times, and eight microscopic images of each of the six traces were taken. After gathering all the data, the mean and the standard deviation values of the trace width were plotted for evaluating the effects of the parameters.

2.2. Printing setup

A syringe was filled with EGaIn covered by 5% hydro-chloric acid (HCl) solution on top so that EGaIn would not make a direct contact with the air for preventing oxidation. Polytetrafluoroethylene (PTFE) nozzles with three different inner diameters (100 µm, 200 µm, and 300 µm) covered with stainless steel guides were used for printing EGaIn traces. Different methods of printing were used depending on the inner diameter (ID) of the nozzle. While printing, unwanted EGaIn droplets are easily formed due to its high surface energy (624 mN m−1) unless right printing parameters are chosen. Also, the EGaIn’s moldability (i.e. the ability to maintain a stable microstructure within its oxide skin at room temper-ature) is a function of the ratio of the surface area to volume (s/v ratio). The moldability increases as the s/v ratio increases, and the s/v ratio increases as the size of the structure decreases [15]. When using a nozzle with a large inlet diameter, the s/v ratio decreases due to the excessive amount of EGaIn out of the nozzle, resulting in formation of unwanted droplets in the middle of the trace rather than maintaining its original trace structure. Therefore, the dispensing pressure was applied only at the start of printing and shut off while printing for nozzles with relatively large IDs, such as 200 µm and 300 µm. In this case, the relatively low viscosity of and the gravitational force to EGaIn helped pulling EGaIn out of the nozzle even without applying a further pressure, making a consistent trace without a droplet. However, a constant dispensing pressure was needed for the smaller nozzle (100 µm) during the entire printing pro-cess, since the ID of the nozzle was small enough to prevent

formation of EGaIn droplets. The starting pressure was 5 kPa for the nozzles with 200 µm and 300 µm IDs, and the same 5 kPa was applied for the entire period of printing for the 100 µm nozzle. Printing was completed by lifting the nozzle off the substrate for the 200 µm and 300 µm nozzles and by simply cutting the dispensing pressure for the 100 µm nozzle.

We printed traces 30 mm long with different parameters on silicone substrates. The printing parameters that we consid-ered important were printing speed (i.e. velocity of the stage), ID, and SOD whereas the trace width was set as the output. To figure out the relationships between the parameters and the output, we conducted a printing test with various combina-tions of the parameters. SOD was first set as a parameter to be evaluated for its influence on the trace width for the nozzles with different IDs. A different SOD was used depending on the sizes of the nozzle because the size of the droplet formed at the tip of the nozzle was different. Then, the SOD was fixed to investigate the effects of the stage velocity and the ID of the nozzles. The stage velocity was controlled from 1 mm s−1 to 6 mm s−1 with a 1 mm s−1 interval. Once the printing was complete, the silicone substrates with EGaIn traces were placed under the microscope to acquire digital images.

2.3. Printing procedure

We first prepared a flat mold using a 3D printer (Objet30, Stratasys) to fabricate a flat substrate of silicone. After the mold was prepared, two parts of highly stretchable silicone elastomer (Ecoflex 0030, Smooth-On) and a white pigment (Silc Pig, Smooth-On) was mixed with a ratio of 1:1:0.1 by weight using a centrifugal mixer (ARE-310, Thinky) and degassed in a vacuum chamber. The pigment was used to pre-vent any measurement errors from laser sensing. It was then poured in the mold and cured in an oven at 60 °C. Once the silicone was fully cured, it was removed from the mold and treated in an ultrasonic cleaner filled with isopropyl alcohol for 30 s to remove any contamination on the surface that could interfere adhesion of EGaIn to the surface. After the silicone substrate was cleaned, it was dried using the oven of 60 °C again for a few minutes to remove any residue of the isopropyl alcohol. Finally, the flat substrate was firmly placed on the x − y stage to prepare for the actual printing process. The same method was used to fabricate flat silicone substrates that

Figure 1. Setup for EGaIn direct printing. (a) Overall system and syringe setup with EGaIn and HCl solution. (b) Image acquisition by microscope and acquired image of EGaIn trace.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

4

were used for experiments on different types of substrates, such as curved surfaces, and inclined surfaces, as shown in figure 2. Molds with curvatures from 0.02 mm−1 to 0.05 mm−1 with a 0.01 mm−1 interval, and with inclined surfaces from 20° to 50° with a 10° interval (printing direction: upward), were prepared using the 3D printer, as shown in figure 2. Then, flat silicone layers were carefully placed on top of the molds. During this process, the silicone layers were neither stretched nor compressed.

3. Results

3.1. Flat surface

Figure 3 shows the mean and the standard deviation values of three different widths of EGaIn traces. The ID of the nozzle and the SOD were found to be critical, and the width became larger as the values of these two parameters increased. While the mean of the largest printable widths was approximately

190 µm using the 300 µm nozzle when the SOD was 0.3 mm, the mean of the smallest printable widths was 33 µm using the 100 µm nozzle when the SOD was 0.1 mm. The smallest width actually achieved in our experiments was 25 µm although it is not shown in figure 3(a), since the plot shows only the mean and the standard deviation values. These results indicate that it is possible to control the trace width by adjusting certain parameters of direct printing. The mean and the standard deviation values are available in table 1 in the supplementary document (SD1 (stacks.iop.org/JMM/30/034001/mmedia)).

According to the experiments, the SOD was most related to printing failures that included formation of random EGaIn droplets and disconnections on the trace during printing. When the SOD was too large, EGaIn droplets were easily formed due to its high surface energy, or the trace was easily disconnected since EGaIn did not make a consistent contact with the silicone surface [40]. On the other hand, when the SOD was too small, insufficient amount of EGaIn was dis-pensed, causing disconnection. By adjusting the parameters,

Figure 2. Different types of molds used for printing experiments: (a) flat mold, (b) curved molds with different curvatures, and (c) inclined molds with different angles.

Figure 3. EGaIn trace widths with respect to different SODs on flat substrate for different nozzle sizes: (a) 100 µm, (b) 200 µm, and (c) 300 µm. The stage velocity was set to 3 mm s−1.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

5

we were able to fabricate an EGaIn trace as thin as 25 µm on a flat surface. Figure 4(a) compares the 25 µm EGaIn trace with a human hair. To check the continuity of the trace, a light-emitting diode (LED) was connected to the printed traces, as shown in figure 4(b). Furthermore, we were able to print not only straight but also curved ones. Curved patterns were much more difficult to print than straight ones since the direction of the velocity was always changing, causing trace discon-nection more frequently unless right parameter values were chosen. Figure 4(c) shows a spiral pattern of EGaIn printed without any fluctuation in width or disconnection. To examine the microstructure of an EGaIn trace made by direct printing, a microscopic image of the cross-section of an actual EGaIn trace was taken, as shown in figure 4(d). As expected, the cross-section was symmetric, and the contact angle was 137°. These results show a potential of fabricating micro-scale cir-cuits with any shapes from straight lines to curves, embedded in an extremely small space.

3.2. Inclined and curved surfaces

As seen in figures 5 and 6, the mean of the widths decreased but the standard deviation increased when the EGaIn traces were drawn on inclined or curved surfaces, compared with those of the traces on flat surfaces even though the same parameters were used. In addition, it was not possible to reliably print traces on 40° and 50° surfaces with the 100 µm and the 200 µm nozzles, since the corner of the nozzle became too close to the substrate when the angle was too steep. The trace width became smaller although the standard deviation increased as the curvature became larger. The results show that EGaIn traces can be printed without disconnection even on uneven surfaces although the trace width changes slightly. The smallest traces measured in our experiments on an inclined (20°) and a curved (0.04 mm−1) surfaces were 21 µm and 11 µm in width, respec-tively, using the 100 µm nozzle with 0.1 mm SOD and 3 mm s−1 velocity. The mean and the standard deviation values are avail-able in tables 2 and 3 in the supplementary document (SD1).

Figure 4. EGaIn traces printed on flat surface: (a) straight line of 25 µm drawn by 100 µm nozzle compared with a human hair at the bottom, (b) LED connected to EGaIn trace of 25 µm in width, (c) spiral pattern drawn by 300 µm nozzle, and (d) magnified cross-sectional view of EGaIn trace (150 µm in width) encapsulated in cured silicone.

Figure 5. EGaIn trace widths for nozzle IDs and stage velocities on inclined surfaces with angles of (a) 20°, (b) 30°, (c) 40°, and (d) 50°. The SOD was set to 0.15 mm, 0.13 mm, and 0.1 mm for 300 µm, 200 µm, and 100 µm nozzles, respectively.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

6

To verify the printing parameters, different types of pat-terns were printed on flat, curved, and inclined surfaces, as shown in figure 7. A spiral pattern was drawn on the curved surface with the curvature of 0.02 mm−1, and a strain gauge

pattern was printed on a 20° inclined surface. The traces were printed without any disconnection. The printing videos are available in the supplementary videos (SV1, SV2, SV3, and SV4).

Figure 6. EGaIn trace widths for nozzle IDs and stage velocities on curved surface. The SOD was set to 0.15 mm, 0.13 mm, and 0.1 mm for 300 µm, 200 µm, and 100 µm nozzles, respectively. (a) κ = 0.02 mm−1, (b) κ = 0.03 mm−1, (c) κ = 0.04 mm−1 and (d) κ = 0.05 mm−1.

Figure 7. EGaIn traces on inclined and curved surfaces: (a) strain gauge pattern printed on 20° inclined surface, (b) spiral pattern printed on curved surface with curvature of 0.02 mm−1, and magnified top views of (c) strain gauge pattern and (d) spiral pattern.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

7

4. Applications

To verify the practicality of the proposed method, we dem-onstrate three different applications of stretchable electronics that could be beneficial if fabricated by direct printing.

4.1. Micro-scale soft sensor

We were able to fabricate a soft pressure sensor with a spiral pattern as an application (see figure 8). The mechanism of this sensor has been introduced with a much larger form factor [8, 22]. The sensor size was comparable to the tip of a ball-point pen, which is smaller than any of direct-printed EGaIn sensors previously reported. Although other previous sensors so far have mostly focused on a serpentine pattern composed of only straight lines, this is, to the best of our knowledge, the first curved EGaIn pattern made by direct printing and charac-terized as a sensor in this form factor. A 100 µm ID nozzle was

used, and the stage velocity and the SOD were set to 6 mm s−1 and 0.14 mm, respectively. The fabrication method is available in the supplementary document (SD2).

The total size of the sensor was 1.8 mm × 3.6 mm. The trace width and the spacing were approximately 70 µm and 120 µm, respectively. The specified sensor dimensions were chosen to show the capability of fabricating a fully functional micro-scale soft pressure sensor with an EGaIn trace, which was robust enough to be functional under repeated applications of external force. The spacing (120 µm) between the traces was the smallest we were able to achieve using the 100 µm nozzle. Spacing smaller than 120 µm was not possible due to the size of the outer diameter of the nozzle, which was 160 µm, since the nozzle would touch and destroy the adjacent trace while printing. The specific width was chosen to ensure that the trace would maintain its original shape after multiple times of deformation. The sensor was tested by applying normal forces up to 0.33 N using an indenter (φ: 1.9 mm) attached

Figure 8. EGaIn pressure sensor with spiral pattern printed by 100 µm nozzle. (a) Microscopic view of the sensor. (b) Comparison with the tip of a ballpoint pen.

Figure 9. Test setup and results of micro-pressure sensor: (a) setup for loading and unloading test, (b) test result (red: force, blue: resistance change), and (c) resistance change for applied forces.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

8

to a tensile tester (ESM 303, Mark-10), and the sensor output was measured by a commercial load cell (Nano17, ATI Industrial Automation). A loading and unloading test was con-ducted multiple times, as shown in figure 9(a). The resistance changed from 1.1 Ω to 20 Ω when a force of 0.1 N to 0.33 N was applied, and the sensor signal was reliable. The sensor showed a high sensitivity when the force was over 0.13 N (∆R/R · 1

F ≈ 90 N−1). Also, the figure 9(c) shows the hysteresis loop of the sensor during the test which is common in soft sensors [41].

4.2. EGaIn printed circuit board (E-PCB)

The second example shows a potential of miniaturization of soft electronics. Figure 10(a) is the schematic of a seven-seg-ment display circuit. We were able to fabricate a circuit made of EGaIn traces that connected a development board in the form factor of an Arduino Nano board, a 220 Ω resister, and a seven-segment display, as shown in figure 10(b). Since there were crossing lines in the circuit, it was not possible to com-plete the circuit only with the printing skill for a flat surface.

This means that the amount of space needed for the entire circuit would significantly increase as the circuit becomes complicated. However, we found a way of minimizing the space to increase the circuit density by printing the traces on top of each other. This method requires a small bump struc-ture over an EGaIn trace and then printing another trace over. Figure 10(c) shows a trace printed on a bump avoiding inter-ference to the traces underneath. Even though the width was slightly different at the boundaries of the bump, the trace was fully connected, and the circuit was functional.

For fabrication, an uncured silicone droplet was first dis-pensed on the traces for crossing. After the silicone droplet was cured, forming a bump structure, the other trace in a dif-ferent direction was printed directly on the bump. This was possible due to the capability of printing EGaIn traces on uneven surfaces by means of laser sensing. The advantage of this method is that traces do not need to detour existing traces to reach the target areas. Traces can cross each other using this method, minimizing the required space, increasing the circuit density, and consequently reducing the form factor of the cir-cuit significantly.

Figure 10. Seven-segment display circuit wired by EGaIn traces. (a) Schematic circuit diagram. (b) Actual circuit. (c) Traces crossing each other without touching.

Figure 11. EGaIn traces directly printed on 3D surface. (a) Soft sensor array printed on hand. (b) Sensor signal with respect to applied force for different fabrication methods.

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

9

4.3. 3D soft sensor array

The last example is a soft sensor array entirely made of an EGaIn-printed circuit on a 3D surface. We were able to print seven soft sensors on the unstructured 3D surface of the back of a mannequin hand, as shown in figure 11(a). The center LED validates that the entire circuit and the sensors are fully connected and functional. This shows a potential in manufac-turing wearable devices that perfectly fit human bodies with a high sensitivity and reliability. The printing videos are avail-able in the supplementary video (SV4). To verify this concept, we conducted an experiment that compares the performance of the sensor directly fabricated on the curved surface (i.e. back of the mannequin hand) with that of the sensor fabri-cated on a flat surface and later attached to the curved surface. A commercial load cell (Nano17, ATI Industrial Automation) and an indenter (φ: 6 mm) integrated with a tensile tester (ESM 303, MARK-10) were used for the test. The same loading and unloading test was conducted ten times for each sensor with maximum applied pressure of 141 kPa, and the result is shown in figure 11(b). The black curves are the data obtained from the sensor fabricated on the flat surface and later attached to the back of the mannequin hand. The red curves indicate the data from the sensor that was printed directly on the silicone sur-face of the mannequin hand, and this sensor showed not only a higher sensitivity but also less hysteresis than that printed on a flat surface. This was mainly because the entire area of the flat sensor did not fully conformed with the 3D surface. Even though the sensor directly printed on a 3D surface still showed some hysteresis, it was not because of the quality of the EGaIn trace but due to the viscoelastic material property of the host material, silicone rubber, and this effect is common in most soft sensors made of highly deformable polymer. From this result, we can verify that our proposed method for fabricating soft sensors and soft electronics directly on 3D surfaces pro-vides a higher performance and reliability.

5. Conclusion

Direct printing of room-temperature liquid metals on sili-cone substrates is becoming common these days, and more applications are proposed. It has been shown that liquid metal patterns can be printed on both flat and uneven substrates [35, 42]. However, a thorough study on parameters that affect the printing quality has not been reported yet. Therefore, we focused on investigation of the parameters that determine the performance of printing on substrates with uneven surfaces as well as with flat surfaces.

To find right parameters, different SODs, stage velocities, and nozzle IDs were tested using a robotic stage and an auto-mated liquid dispenser. EGaIn traces were printed on silicone substrates with varied parameter values. The EGaIn traces were observed using a microscope, and their magnified images were processed to accurately measure the widths by detecting the edges. The processed data were evaluated by examining the mean and the standard deviation values, which can be used for controlling the trace widths for different applications and

fabrication environments. This data can be used to print EGaIn traces on flat or curved surfaces with desired widths with dif-ferent printing conditions when a certain parameter needs to be constrained. The capability of printing EGaIn traces on uneven surfaces has a great potential especially in the field of wearable devices where dealing with 3D surfaces is mostly the case. Furthermore, patterns of not only straight lines but also curves can be printed with stability on various types of curved surfaces. Based on the parameters found, we were able to fabricate a micro-scale soft sensor with a spiral pattern as an application, which can be highly useful for detecting mechanical pressure (or force) when the pressure area is extremely limited. Another application area of the proposed method includes direct printing of stretchable electronics on surfaces with arbitrary shapes, as demonstrated with examples of the seven-segment display circuit and the soft sensor array on a human hand. We believe the proposed method and its applications will open a new space in soft robotics with an emphasis on wearable technologies.

Acknowledgment

This work was supported by Samsung Science and Technology Foundation (Project No. SRFC-IT1701-12).

ORCID iDs

Yong-Lae Park https://orcid.org/0000-0002-2491-2114

References

[1] Gu G, Xu H, Peng S, Li L, Chen S, Lu T and Guo X 2019 Soft Robot. 6 368–76

[2] Itoigawa K, Ueno H, Shiozaki M, Toriyama T and Sugiyama S 2005 J. Micromech. Microeng. 15 S233

[3] Wang X, Li Z, Xu W, Kulkarni S A, Batabyal S K, Zhang S, Cao A and Wong L H 2015 Nano Energy 11 728–35

[4] Cao Q, Kim H S, Pimparkar N, Kulkarni J P, Wang C, Shim M, Roy K, Alam M A and Rogers J A 2008 Nature 454 495

[5] Xiong W et al 2016 Adv. Mater. 28 7167–72 [6] Liu C H and Yu X 2011 Nanoscale Res. Lett. 6 75 [7] Hayes G J, So J H, Qusba A, Dickey M D and Lazzi G 2012

IEEE Trans. Antennas Propag. 60 2151–6 [8] Park Y L, Chen B R and Wood R J 2012 IEEE Sens.

J. 12 2711–8 [9] Kim T and Park Y L 2018 IEEE Robot. Autom. Lett. 3 881–7[10] Feng X, Yang B D, Liu Y, Wang Y, Dagdeviren C, Liu Z,

Carlson A, Li J, Huang Y and Rogers J A 2011 ACS Nano 5 3326–32

[11] Yamada T, Hayamizu Y, Yamamoto Y, Yomogida Y, Izadi-Najafabadi A, Futaba D N and Hata K 2011 Nat. Nanotechnol. 6 296

[12] Amjadi M, Pichitpajongkit A, Lee S, Ryu S and Park I 2014 ACS Nano 8 5154–63

[13] Wang Q, Yu Y, Yang J and Liu J 2015 Adv. Mater. 27 7109–16[14] Palleau E, Reece S, Desai S C, Smith M E and Dickey M D

2013 Adv. Mater. 25 1589–92[15] Dickey M D, Chiechi R C, Larsen R J, Weiss E A, Weitz D A

and Whitesides G M 2008 Adv. Funct. Mater. 18 1097–104

J. Micromech. Microeng. 30 (2020) 034001

G Shin et al

10

[16] Chossat J B, Park Y L, Wood R J and Duchaine V 2013 IEEE Sens. J. 13 3405–14

[17] Fassler A and Majidi C 2013 Smart Mater. Struct. 22 055023[18] Varga M, Ladd C, Ma S, Holbery J and Troster G 2017 Lab

Chip 17 3272–8[19] Hao Y, Wang T, Xie Z, Sun W, Liu Z, Fang X, Yang M and

Wen L 2018 J. Micromech. Microeng. 28 024004[20] Wirekoh J, Valle L, Pol N and Park Y L 2019 Soft Robot.

6 768–77[21] Joshipura I D, Ayers H R, Majidi C and Dickey M D 2015 J.

Mater. Chem. C 3 3834–41[22] Arabagi V, Felfoul O, Gosline A H, Wood R J and Dupont P E

2015 IEEE Sens. J. 16 1294–303[23] Guo C, Yu Y and Liu J 2014 J. Mater. Chem. B 2 5739–45[24] Kramer R K, Majidi C and Wood R J 2013 Adv. Funct. Mater.

23 5292–6[25] Kim M G, Alrowais H, Pavlidis S and Brand O 2017 Adv.

Funct. Mater. 27 1604466[26] Liu J, Yang S, Liu Z, Guo H, Liu Z, Xu Z, Liu C and Wang L

2019 J. Micromech. Microeng. 29 095001[27] Jung T and Yang S 2015 Sensors 15 11823–35[28] Shi X, Cheng C H, Zheng Y and Wai P 2016 J. Micromech.

Microeng. 26 105020[29] Jeong S H, Hjort K and Wu Z 2015 Sci. Rep. 5 8419

[30] Wang B, Xin W, Hjort K, Guo C and Wu Z 2019 ACS Appl. Polym. Mater. 1 145–51

[31] Wang B, Wu K, Hjort K, Guo C and Wu Z 2018 Soft Robot. 6 414–21

[32] Zhang S, Wang B, Jiang J, Wu K, Guo C F and Wu Z 2019 ACS Appl. Mater. Interfaces 11 7148–56

[33] Boley J W, White E L, Chiu G T C and Kramer R K 2014 Adv. Funct. Mater. 24 3501–7

[34] Zheng Y, Zhang Q and Liu J 2013 AIP Adv. 3 112117[35] Kim S, Oh J, Jeong D, Park W and Bae J 2018 Soft Robot.

5 601–12[36] Wu K, Zhang P, Li F, Guo C and Wu Z 2018 Polymers 10 330[37] Park Y G, An H S, Kim J Y and Park J U 2019 Sci. Adv.

5 eaaw2844[38] Ladd C, So J H, Muth J and Dickey M D 2013 Adv. Mater.

25 5081–5[39] Sobel I 1990 An isotropic 3 × 3 gradient operator Machine

Vision for Three-Dimensional Scenes (New York: Academic) pp 376–9

[40] Plech A, Klemradt U, Metzger H and Peisl J 1998 J. Phys.: Condens. Matter 10 971

[41] Case J C, White E L and Kramer R K 2015 Soft Robot. 2 80–7[42] Yoon Y, Kim S, Kim D, Kauh S K and Lee J 2018 Adv. Mater.

Technol. 4 1800379

J. Micromech. Microeng. 30 (2020) 034001