A research team from the Department of Mechanical Engineering and Materials Science at Yale University has made a major breakthrough in the field of soft robotics. They succeeded in developing a universal method for converting complex circuit boards into a stretchable form, achieving a high degree of 300% stretchability, and applying it to single-board microcontrollers such as the Arduino, enabling them to be embedded directly in soft robots. This innovation was published in the journal Science Robotics, the top journal in the field of robotics, on September 11, 2024, marking a key step towards smarter and more integrated soft robots.
▍The contradiction between traditional hard circuit boards and soft robots
Soft robots have attracted attention for their flexibility and adaptability, but how to incorporate decision-making computing power into these stretchable structures has been a challenge. Currently, most soft robots still rely on microcontrollers such as Arduino for control, but there is a significant modulus mismatch between these rigid circuit boards and the soft materials. Designers often have to place the electronics in areas where the robot is less stressed, or place it completely outside of the robot itself. This practice obviously limits the functions and application scenarios of soft robots.
To solve this problem, the researchers have proposed a variety of options. These include mechanical computing platforms, soft logic gates, and stretchable electronics. The mechanical computing platform uses high-dimensional dynamic phenomena to calculate and form a distributed information processing network. Soft logic gates, on the other hand, are based on fluid principles (e.g. pneumatics or hydraulics) and offer the possibility of autonomous open-loop behaviors such as motion and arm movements. However, the logic gate density of these mechanical methods is far less than that of traditional electronic computers, and it is difficult to meet the needs of complex control.
Stretchable electronics technology is another direction that is getting a lot of attention. The researchers tried to give the traditional computing platform its stretchability while retaining its computing power. This method connects silicon-based rigid integrated circuits (ICs) through stretchable conductive trajectories and substrates, which not only retains the high computational density of traditional circuits, but also introduces a certain flexibility. However, there are still some limitations to existing stretchable conductor technologies. For example, solid thin film conductors based on geometric patterns, such as serpentine or mesh structures, while well interfacing with rigid microelectronics, often do not achieve the high strain (20% to 1000%) required for soft robotics applications. Liquid metal (LM), LM composites, or conductive elastomers, on the other hand, can achieve higher strain, but there are difficulties in interfacing with rigid IC components, and the resistance varies with strain.
▍Breakthrough stretchable conductor material
To overcome these challenges, the research team has developed a novel stretchable conductor material, gallium indium oxide (OGaIn). The material is a duplex foam containing amorphous gallium oxide particles and eutectic gallium indium alloy (EGaIn). OGaIn is made by thorough mixing of EGaIn in air and can be produced on a large scale.
The researchers performed a detailed characterization of OGaIn. The results show that the content of amorphous gallium oxide in OGaIn is about 1.4wt%, which is close to the previously reported 1.21wt%. In contrast, the previously developed dual-phase gallium indium (BGaIn) material contains crystalline gallium oxide content of about 34wt%. The researchers speculate that OGaIn increases viscosity primarily through air entrainment, which is not reflected in weight percentages. The density of OGaIn (4.65 g/cm³ at 20°C) was significantly lower than that of EGaIn (>6 g/cm³). Scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), and X-ray diffraction analysis show that the hardness of gallium oxide formed at the OGaIn-air interface prevents the deformation of bubbles during stretching, which may explain the viscosity enhancement of OGaIn relative to pure EGaIn.
Further rheological studies revealed the shear thinning behavior of OGaIn, indicating that it is compatible with extrusion printing techniques. OGaIn and BGaIn exhibit similar bulk conductivity of 2.11 × 10⁶ S/m and 2.06 × 10⁶ S/m, respectively.
To evaluate the electromechanical properties of OGaIn, the researchers fabricated a 250 μm wide OGaIn wire (comparable to the thinnest wire width on a commercial Arduino Pro Mini) on a standard dumbbell-shaped acrylic foam tape (VHB tape, 3M) and encapsulated it with rubber cement (Elmers). At 400% strain (limited by the test equipment), the relative resistance change (R/R0) of OGaIn is 7, which is much lower than the 25 predicted by bulk conductor theory. This result is comparable to the previously reported performance of BGaIn over the 100% strain range.
To assess cycling stability, the researchers performed 1000 cycling tests of 150% strain on a single-wire sample at a rate of 15 mm/min. The results show that after the first few cycles, the wire resistance increases by only about 0.5 ohms between the 5th and 1000th cycles. To evaluate the interface stability between the wires and the rigid IC contact pads, the researchers also performed a two-wire cyclic strain test with a 0 ohm resistance bridge and a high strain test
The results show that there is no significant difference between the wires with and without the interface, suggesting that the slight increase in resistance observed is not due to the interface.
▍Conductor-substrate compatibility guide
The wettability and adhesion of liquid and duplex metals to substrates are important parameters in stretchable electronics applications. In this study, OGaIn needed to adhere well to both the underlying substrate and the IC component. Based on the results of the cycle stability test, the researchers deduced that OGaIn adheres well enough to VHB tape. To embed complex stretchable circuits into soft robotics, the researchers also characterized the adhesion of OGaIn to other commonly used soft robotics materials and explored how existing materials could be modified to improve adhesion.
The researchers compared the benchmark VHB tape substrate with four different ratios of silicone rubber elastomer (Dragon Skin 10, DS10, SmoothOn) substrate. The results showed a positive linear correlation between substrate adhesion and OGaIn-substrate adhesion (R² = 0.96). All substrates exhibit some degree of OGaIn adhesion, even though the adhesion is < 0.1 N. This may be due to the fact that the oxide particles in the OGaIn promote wetting.
In practice, the researchers found that substrates with lower adhesion values, such as pure DS10 and Slacker 1, increased the likelihood of failure modes such as wire defects and IC displacement. Therefore, they recommend choosing a substrate with an adhesion value of at least 0.18 N to ensure adequate OGaIn-substrate adhesion, stable IC placement, and overall conductor-substrate compatibility for stretchable circuits.
▍Stretchable conversion method for complex circuits
After identifying a suitable stretchable conductor and summarizing its compatibility with soft, stretchable substrates, the research team developed a method for converting complex board designs as-is into stretchable circuits. Throughout the method development process, they emphasize accessibility with the aim of eliminating the need for extensive equipment, materials expertise, or circuit design expertise. They applied the method to make the popular Arduino Pro Mini, a reprogrammable single-board microcontroller, as well as stretchable versions of the Arduino Lilypad, SparkFun sound detector, and SparkFun RGB and gesture sensors.
The circuit fabrication process uses laser cutting and stencil printing techniques. The substrate (e.g. VHB tape) is sandwiched between two layers of 0.1mm thick sticker that acts as a mask. A carbon dioxide (CO2) laser is used to cut the circuit board outline and vias, and a ultraviolet (UV) laser is used to etch the wire profile on the bottom mask. OGaIn is then applied to the bottom wire, the mask is removed, and the circuit is encapsulated with a thin layer of rubber cement. Next, the top wire is made using the same laser and smearing procedure, after which the element is placed and the rubber cement seal is made. Finally, a silicone adhesive (SilPoxy, SmoothOn) is added to a small area around the microprocessor to reduce the stress caused by the stiffness gradient.
Compared to the more restrictive methods that used transfer technology before, this scalable process based on screen printing and laser cutting enables the creation of encapsulated multilayer circuits using substrates that meet compatibility requirements (adhesion ≥~0.18 N) and can accommodate high-density IC components while withstanding high strains. In addition, the increased adhesion improves the robustness of the IC-OGaIn interface, enabling researchers to directly convert commercial circuits into stretchable forms.
When performing a strain at break test on the soft Arduino Pro Mini (defined as a disconnection of serial communication with a computer; strain rate 15 mm/min), the researchers found that serial disconnection always occurs due to loss or short circuit of electrical contact of the wire, rather than mechanical failure of the substrate. The average breaking strain is 328% and the highest is 344%. This is far beyond the strain range required for most soft robotics applications.
To further evaluate the performance of the stretchable Arduino Pro Mini, the researchers performed a cyclic strain test. After 120 cycles at 100% strain, the circuit remains functional. This excellent cyclic stability opens up the possibility of integrating these stretchable microcontrollers into real-world soft robotics applications.
To demonstrate the versatility of the method, the research team also produced stretchable versions of several other popular open-source circuits, including the Arduino Lilypad, the SparkFun sound detector, and the SparkFun RGB and gesture sensor. It is important to note that these circuits use the same package and IC package as the original version, including leadless (e.g., double-row flat unleaded) and leaded (e.g., thin quad flat package). This suggests that the method can be directly applied to a variety of existing commercial circuit designs without major modifications.
Finally, the researchers embedded the stretchable Arduino Pro Mini into the high-strain position of the soft robot and used it for embedded computing. These demos mark a shift from a one-off presentation with limited functionality to a robust, reliable, and complex multilayer stretchable circuit.
This research opens up new possibilities in the field of soft robotics. By integrating high-performance computing power directly into soft, deformable structures, researchers are paving the way for the creation of smarter, more autonomous soft robots. This approach is not only applicable to robotics, but may also find a wide range of applications in wearables, flexible electronics, and other fields that require the combination of computing power and flexible structures. As this technology continues to evolve and expand, we can expect to see more exciting applications of soft robotics emerging to push this field forward.