Strategy of the channel-free method for 3D stretchable conductors
The phase transition and plastic deformation of LM are key factors for the fabrication of LM 3D flexible electronics using our new approach without the need for channel construction. We first optimized the content of In to obtain a good balance between melting point and plasticity in the Ga-10In alloy (Fig. 1a). The differential scanning calorimetry (DSC) analysis of the Ga-10In alloy shows that the Ga-10In alloy starts to melt at 16.3 °C and melting is complete at 22.7 °C (Fig. 1b). Supercooling describes the process of lowering the temperature of a liquid below its freezing point without it becoming a solid until reaching crystallization temperature. Interestingly, by supercooling the alloy it can remain in a liquid state at temperatures as low as -38 °C. After solidification at -43.6 °C, the Ga-10In alloy completes the transition from a liquid to a solid state and can remain in the solid state even as the temperature is increased again up to ~ 16 °C. A tensile test (Fig. 1c) carried out at 10 °C showed that the Ga-10In alloy possesses a moderate yield strength and excellent elongation which are both crucial for plastic deformation. Ideally, the Ga-10In alloy can be plastically processed after solidification within the solid-state temperature range. In our experiment we used a temperature range of 0 ~ 15 °C.
In our approach, we first induce a liquid-solid phase transition of the Ga-10In alloy via the freeze casting method using silicon tubes with tunable inner diameter as a mold. This allowed us to obtain Ga-10In alloy based solid wires with diameters as small as ~ 60 µm (Fig. S1c, Supporting Information (SI)). These Ga-10In solid wires have a low melting point, moderate strength, excellent plasticity, and can be mechanically processed into any 3D structure as needed for the preparation of 3D-structured stretchable conductors (Fig. 1d). Another key step in the channel-free construction of flexible electronics is the encapsulation process using soft elastomer. We used uncured elastomer to encapsulate the pre-shaped 3D conductive structure after placing it onto a cured elastomer substrate (Fig. 1e). After the two parts of the elastomer have cross-linked to form a single entity, the 3D electronic device is heated to 60 °C to melt the solid Ga-10In wire and regain stretchability (Fig. 1f). Meanwhile, attributing to the supercooling of LM alloy44, the Ga-10In alloy can remain liquid state at or far below room temperature after melting at high temperature, which ensures that the stretchability of the fabricated devices is maintained across a wide range of applications and usage conditions. The proposed method does not require any microchannel construction, thus eliminating the complicated processes of microchannel mold preparation and troublesome LM filling, while paving the way for the design and manufacture of highly complex 3D circuits.
Quantify and characterize the mechanical and thermal properties of Ga-In alloys
The specific phase transition behavior and mechanical properties of the Ga-In LM alloy, which can be regulated by adjusting the In element content, are crucial to successfully apply our method to the pre-construction of 3D conductive structures. As shown in the Ga-In alloy phase diagram (Fig. S2, SI)45, in hypoeutectic (In element < 24.5 wt.%)/hypereutectic (In element > 24.5 wt.%) alloys the melting point decreases/increases with increasing In element content. For metals, the content of alloying elements has a significant effect on the overall strength and plasticity of the alloy. Therefore, we synthesized Ga-5In, Ga-10In, and Ga-15In LM alloys (all hypoeutectic) to investigate the influence of In element content on the alloy’s microstructure, melting point, and mechanical properties.
Back-scattered electron (BSE) imaging was used to characterize the microstructure of the solid-state Ga-10In alloy (Fig. 2a). Similar to the low-temperature phase separation of LM nanoparticles46,47, a second phase (A6 phase) is introduced to the matrix with the addition of additional In. The volume fraction of the A6 phase increases with increasing In element content (Fig. 2b, BSE images of pure Ga, Ga-5In, and Ga-15In are shown in Fig. S3, SI). Energy dispersive X-ray spectroscopy (EDS) element mapping (Fig. S4a, SI) of Ga-10In alloy reveals that the matrix and A6 phase can be clearly distinguished by the element distribution. EDS spot analysis (Fig. S4b, SI) further illustrates that the A6 phase of Spot 1 is a solid solution with Ga atoms dissolved in the In matrix, while there are no In atoms solute in the Ga matrix (Spot 2, A11 phase). The phase composition of Ga-10In and Ga-15In alloys was further investigated by X-ray diffraction (XRD) at a temperature of 0 °C detecting only the α-Ga (PDF-#05-0601) and In phase (PDF-#05-0642) (Fig. 2c). The crystal structures of the two phases are orthorhombic and tetragonal, respectively.
Tensile tests of pure Ga, Ga-5In, Ga-10In, and Ga-15In alloys were conducted at a temperature of 10±3 °C. According to metallurgy theory, dislocation slip is a major mechanism controlling plastic deformation48,49. The second phase often tends to hinder dislocation slip thereby increasing alloy strength, a mechanism known as “second phase strengthening”50,51. However, stress concentration occurs at the interface between the matrix and second phase and will result in the formation of microcracks that continue to expand into macrocracks and finally cause the interface to be pulled apart. The fracture morphology of the Ga-10In alloy indicates that the microcracks are mainly distributed along the interface between the A6 phase and matrix (Fig. 2d). Hence, the A6 phase is more likely to cause fracture despite increasing alloy strength. The tensile stress-strain curve, yield strength, and elongation vs. alloy In element content are shown in Fig. 2e and f. More detailed results of the stress-strain curve and corresponding yield strength, tensile strength, and elongation are provided in Table S1 (SI) which are all consistent with theoretical expectations. Tensile samples of pure Ga, Ga-5In, and Ga-10In after fracture (Fig. S5, SI) exhibit obvious necking and indicate that the fracture mode is plastic fracture. In comparison, the fracture surface of the Ga-15In alloy is flat and presents brittle fracture characteristics. Bending experiments with the Ga-10In and Ga-15In solid wire further verified the superior plasticity of the Ga-10In alloy compared to the Ga-15In alloy (Movie S1, SI). Therefore, Ga-In alloys with an In element content of less than 10 % possess good plastic processing characteristics. Besides, the measured yield strength of below 37.8 MPa requires a relatively low external force to deform the LM alloy, which is convenient for the plastic processing of the Ga-10In solid wire into different conductive structures.
Clearly, the thermal behavior of Ga-In alloys can be significantly influenced by the In element content (Fig. S2, SI). Results from a DSC analysis of pure Ga, Ga-5In, Ga-15In, and Ga-24.5In alloys all showed a peak freezing temperature of about − 40 °C during the cooling cycle (Fig. 2g). However, all these exothermic peaks seem to consist of multiple overlapping peaks suggesting that multiple phase transition processes may be involved52. During the heating cycle, pure Ga has an endothermal peak with a melting onset temperature of ~ 30 °C. The endothermic peaks of the Ga-5In, Ga-10In (Fig. 1b), and Ga-15In alloys exhibit bimodal characteristics. From the Ga-In phase diagram (Fig. S2, SI) we can deduce that the left and right exothermic peaks represent the melting processes of the eutectic phase (A6 phase + partial matrix) and remanent matrix, respectively. We consider the temperature at which the matrix melts completely as the melting point and the melting points are summarized in Fig. 2h. Clearly, the melting point decreases with increasing In element content. We also investigated the dynamics of Ga-10In based wire solidification and melting (Fig. S6, SI) using an optical microscope equipped with a custom-built temperature control system (Fig. S7 and Movie S2, SI). Due to significant supercooling of the LM, the liquid Ga-10In wire does not crystallize when cooled to ~ 0 °C. We therefore introduced a solid LM wire at the right end of the liquid wire to inhibit supercooling and act as a crystallization nucleus which resulted in the liquid wire crystallizing quickly and the solid-liquid interface moving 3.1 mm to the left within a time period of 3 s. On the other hand, the melting process is relatively slow and the non-uniform. The significant supercooling of LM ensures that the Ga-10In alloy remains liquid state across a relatively large range of temperatures keeping stretchable electronics functional even at relatively low temperatures.
The In element content determines the mechanical properties and thermal behavior of the Ga-In alloy as it affects the volume fraction of the eutectic second phase. Compared with the EGaIn alloy with 24.5 wt.% In element content (Ga-24.5In), the lower In element content in our Ga-10In alloy resulted in excellent plasticity and moderate strength. Although the melting point of Ga-10In is slightly higher than EGaIn, the significant supercooling of the Ga-In alloy ensures that the alloy remains in the liquid state for temperatures down to about − 40 °C44,53. Therefore, the Ga-10In alloy with its excellent plasticity, low melting point, and moderate strength is the best candidate from among the Ga-In alloys series, to be used in the channel-free construction of 3D flexible electronics.
Before exploring this technique on 3D structured circuits, we first tested our approach to prove the reliability of the channel-free method by fabricating a 2D strain sensor, encapsulating the Ga-10In alloy based LM conductive wires within elastomer and systematically testing the sensor's performance (Fig. 3). Its electromechanical response (Fig. 3a) was similar to previously publish results54 and the resistance-strain relationship close to theoretical predictions using: R = R0(1 + ε)2. However, with increasing strain, some increasing hysteresis appeared in the process of loading and unloading. A previous report55 suggests that this hysteresis is associated with the channel diameter and may be due to the high flux in narrower channels which causes the relative resistance to fluctuate during the stretching-recovery process. This hysteresis we found in our sensor has also been observed in previous studies of LM strain sensors55. Besides, the average gauge factor increases linearly with strain and follows the relationship ε + 2 (Fig. 3b). We found no creep behavior when straining the sensor by up to 80 % in the longitudinal and transverse direction and holding it there for 5 s (Fig. 3c), the absence of creep is crucial for sensing stability. To test for long-term stability, we subjected the sensor to 7500 cycles of 50 % strain and did not find any change in its electromechanical response (Fig. 3d). To demonstrate the strain sensor function, we attached five LM strain sensors on the knuckles of a glove (Fig. 3e), and monitor the motion of the finger in real-time. Signals from five different motions can be obtained and each motion is accurately distinguished (Fig. 3f). The good electromechanical response, long-term stability test, and finger motion detection capability of the 2D strain sensor, all together reflect the high reliability and stability of the device, which demonstrate the practicability and feasibility of further fabricating LM based 3D flexible electronics by the channel-free strategy.
Interconnect Arches For An Integrated Led Array
The cross-arrangement of interconnects has been extensively used in active-matrix displays where 2D structures of transverse and longitudinal interconnects are typically deposited on separate layers to avoid short circuits34. 3D interconnects can achieve a cross-arrangement structure on a single plane, avoiding short circuits without requiring separate layers. This can be used to increase the density of interconnects and decrease design complexity, thus allowing the integration of higher density devices in electronic products. Stretchable interconnects can provide an extra degree of freedom for flexible electronic systems while maintaining the same high performance as rigid devices. The Ga-10In solid wire possesses excellent plastic deformability and is therefore easily deformed into a 3D interconnect arch. Figure 4a displays the solid and liquid states of an interconnect arch made of Ga-10In wire on a bending PDMS substrate. Due to the mechanical support provided by the rigid ultrathin Ga2O3 film on the surface, the 3D interconnect arch can still maintain its initial status after melting the Ga-10In alloy inside. Simulations of the stress distribution using finite element modeling (FEM) showed that the stress is concentrated near both corners of the arch and is mainly induced by the gravitational weight of the arched area which depends on the radius of curvature (Fig. S8, SI). The bigger the radius of curvature of the arch, the larger its weight and resulting strain on the corner areas, which may lead to the complete collapse of the arch if its weight becomes too large. Based on our experiments, we found that the 3D interconnect arch can persist after melting when the radius of the curvature is less than 3 mm.
For a proof-of-concept demonstration, the interconnect arches were used to construct an LED array (Fig. 4b) with six parallel branches. Three vertical and three horizontal branches cross at right angles using interconnect arches to avoid short circuiting. A specific welding method needs to be developed to resolve the problem of unstable contact between the Ga-10In wire and the LED electrodes (Fig. S9, SI). The stability of the LED array was investigated by bending the array up to radii of curvature of about 20 mm without any visible change in the current-voltage relationship (Fig. 4c). We also found that the current remained constant after bending the array 20,000 times using a 22 mm radius of curvature (Fig. S10, SI), which attests to the long-term stability of the array and Ga-10In interconnect arches. Each branch of our LED array can be independently controlled using a control system composed of a single-chip microcomputer and electric relays (Fig. 4d and e). The timing and sequence for turning the six branches on and off can be controlled by a program (Fig. 4f and Movie S3, SI). These results demonstrate that our interconnect arch is functional and can be used to achieve independent control of the different branches in our example LED array while minimizing the number of interconnections. This method can greatly simplify the circuit designs and reduce manufacturing costs of flexible electronic products.
Wearable Electronic Fingerstall For Remote Monitoring Of Finger Movement
In recent years, considerable attention has been paid to flexible electronics that are deformable and comfortable due to their great potential in wearable electronics56,57. However, keeping wearable devices integrated, comfortable and stable has always been a technical difficulty for LM-based wearable electronics. In current work, we further demonstrated a wireless monitoring device that can sense finger movement as shown in Fig. 5a. This device consists of a 3D structured sensor and a module with the function of data processing and wireless transmission. Both 3D structured sensor and multilayer flexible circuit board with voltage stabilizing function are prepared based on the plastic deformation and phase transition mechanisms of the developed Ga-10In, which are very difficult to be prepared using other methods like channel injection.
A technology utilizing the automatic winding machine and Ga-10In solid wire (Fig. S11, SI) was employed to prepare the 3D structured wearable sensor, and the three-step fabrication process is shown in Fig. S12, SI. The wearable sensor (Fig. 5b (i)) consists of a hollow-cylindrical 3D structure with a wall thickness of only 0.94 mm and thus conformably integrates with the finger joints and skin (Fig. S13, SI). This sensor was able to survive tensile tests in the axial (Fig. 5b (ii)) and mechanical squeezing using a tweezer (Fig. 5b (iii)), demonstrating its excellent stretchability and circuit stability which are essential for wearer comfort and sensing stability. The as-designed sensor can quantify the degree of finger bending based on specific changes in its resistance (Fig. 5c), and the normalized resistance change increased with increasing bending angle. Furthermore, ΔR/R0 remained stable even after 1000 bending cycles using a 32 mm radius of curvature (Fig. 5d), which demonstrates the high level of stability and reliability of this 3D structured wearable sensor. More importantly, the 3D hollow cylinder structured wearable sensor can be directly used as a thermal therapy device to treat the injured joints through thermal expansion of the vascular system and surrounding collagen tissue when applied a direct current. A thermal image shows the effectiveness of this approach in joining the middle joints of the fingers (Fig. S14a, SI), and the heating temperature can be actively adjusted by wearer bend the finger 20 times by a certain angle (Fig. S14b, SI).
The signal measurement of typical wearable sensors mostly uses a rigid circuit board, which would constrain the movement of the human body in certain cases. The miniaturized and integrated flexible circuit board is more comfortable to wear because they are fully attachable to human skin. Here, we prepared a flexible circuit board with 3D wiring in multiple layers, which could conduct voltage regulating function (Fig. S15, SI). The fabrication process is shown in the Fig. S16, SI. The data transmission was accomplished by a Bluetooth module. We employ the Ga-10In solid wires to connect the electronic components including capacitors, resistors, and a voltage regulator (Fig. 5e and Fig. S17, SI), by the welding method explained in Fig. S9, SI. Particularly, the design of “L-shaped” Ga-10In solid wire can connect the circuits in the bottom and top layers (Fig. 5f, g), which provides great convenience for the preparation of the flexible circuit of multilayer as compared to previous work that requires multiple procedures of elastomer cutting and LM filling23. We then directly integrated the wearable sensor with the flexible circuit board, and wear them on the finger joints and hand back, respectively (Fig. 5a). The whole device can sensitively distinguish the movement of the finger and wirelessly transmit data to the computer, achieving remote monitoring of finger movement (Fig. 5h and Movie S6, SI). In contrast to previous works of wearing rigid circuit board on the body58,59, here the device we developed can closely contact with the skin surface and thus increase the wearing stability and comfortability, which is advantageous merit for the future advancement of wearable electronics.