Abstract
This paper explores the sensing performance displayed by warp-knitted strain sensors under biaxial stretching. These sensors were knitted using silver-plated nylon to be interlooped on a tricot warp-knitting machine. Eight types of warp-knitted sensing fabrics with different loop parameters were prepared and, afterward, electro-mechanical tests were conducted on a biaxial tensile testing machine. These specimens offered similar ground structures but differed in conductive yarn configuration in terms of linear density, number of underlapping wales, open/closed loop type, and guide-bar lapping sequence. Experimental results showed that the loop parameters significantly played a fundamental role in determining sensing performance. It is therefore possible to improve the sensing performance of warp-knitted sensors and engineer them by differing the loop parameters based on certain applications.
Flexible strain sensors have a good reaction to environmental stimuli and therefore are utilized to measure physiological parameters of the human body.1–3 One of the primary issues that determines the sensing effectiveness of strain sensors is the selection of an appropriate fabricating approach to satisfy sensing requirements such as working range, responsivity, and repeatability. The softness, thinness, and stretching performance of fabric materials show advantages for use as flexible wearable devices.4,5 Traditional fabric production approaches cover knitting, weaving, and embroidery. Knitted fabrics are formed by interlocking loops and therefore introduce a better property of large-strain recovery and shape stability, which can realize the design of specifications and parameters of strain sensors in different application scenarios. 6 The sensing mechanism of the knitted strain sensor relies on the electrical resistance inherent in variable conductive loop contact when the sensor is loaded with stress to make deformation. 7
Knitted fabric-based strain sensors are commonly used in health monitoring,8–11 motion monitoring,12–14 identification and interaction,15–17 and some entertainment applications. 18 Frydrysiak and Zieba 19 combined conductive materials with knitted structures to prepare electronic shirts that can monitor breathing patterns. Yao and Soleimani 20 developed smart urinary incontinence pants with conductive yarn combined with circular seamless weft-knitting technology. Five different knitted structures were prepared by Seyedin et al., 21 demonstrating that the strain-sensing behavior can be tailored by simply changing the knitted structures and the positions of conductive versus non-conductive yarns. Among these applications, the electrical properties of knitted strain sensors during stretching are the most explored. They are one of the important indicators to characterize the sensing performance. Raji et al. 22 investigated the effects of elastic yarn types and rib fabric structures on the tensile properties of knitted strain sensors, and the result revealed that these two parameters determined both electrical conductivity and tensile properties. Tohidi et al. 23 exploited the time-varying properties of back-electric waveform peak fitting and tensile fatigue.
Most of the present studies on designing knitted sensors mainly focused on weft-knitted structures, such as jersey stitch, 24 rib stitch, 25 and plating stitch. 26 These structures demonstrate better stretching performance in the weft direction than warp direction because of yarns running horizontally back and forth, and their elastic recovery is likely to be weakened after repeated stretching operations. Compared with weft knitting, yarns in warp knitting are looped vertically down the fabric and simultaneously have horizontal zigzag lapping. This characteristic enables warp-knitted textiles to show good elastic recovery and size stability in biaxial ways. Based on the above performance, it was theoretically feasible to use warp-knitted structures for sensing human motion. In particular, body strain that happens in the warp direction is much greater than that in the weft direction. Thus, this work focused on warp-knitted sensors under biaxial stretching. To basically reveal the sensing mechanism, this study highlighted the effect of loop parameters, such as the linear density of conductive yarns, number of underlapping wales, open/closed loop type, and guide-bar lapping sequence, on sensing performance. It aimed to provide a reference for the development and application of warp-knitted flexible strain sensors.
Methods and experiments
Preparation of warp-knitted strain-sensing fabrics
Detailed information on the yarn raw materials to fabricate warp knitted spacer fabrics (WKSFs) is listed in Table 1. Experimental specimens were prepared on a tricot warp-knitting machine, KS-4, purchased from Karl Mayer (Germany). The gauge of this machine was E28 and three yarn guide bars were used for fabrication. Smooth withdrawal and constant run-in tension of warp yarns were quite fundamental for repeatedly fabricating sensor structures. Therefore, before knitting, the materials were prepared on a warping machine with a linear speed of 150 m/min for silver plated nylon filament (SPNF), 550 m/min for polyamide (PA), and 200 m/min for polyurethane (PU) under constant tension of 10–12 cN. Then in knitting production, a positive run-in system was used to accurately maintain the yarn tension peak under 16 cN for a uniform stitch length and fabric structure. The knitted sensing area was designed around 30 mm * 30 mm with 80 courses and 40 wales. The fabrication details are shown in Figure 1.
Specimen configurations
PA: polyamide; PU: polyurethane; SPNF: silver plated nylon filament.
Schematic of the (a) warping machine, (b) warp-knitting machine and sensory fabrics, and (c) detail of the fabric sensing area with partial enlargement.
A total of eight sensing knitted fabrics with different structures were prepared in this project. These fabrics offered similar ground structures. The specimen configurations are listed in Table 1. The loop diagrams in the table and the three-dimensional (3D) structural model in the article were drawn by the WKCAD 4.0 system, a warp-knitting computer-aided design software developed by Jiangnan University. Samples S1 and S4 were made with SPNF of different linear densities. Samples S2, S3, and S4 had different lapping structures for the SPNFs. Samples S4, S6, and S7 had different lapping structures for the spandex. Samples S4 and S5 had SPNFs with closed and open loop structures. Samples S4 and S8 had different guide-bar lapping sequences. The effects of the loop parameters on the sensing properties were explored by varying the linear density of the conductive yarns, number of underlapping wales, open/closed loop type, and guide-bar lapping sequence.
Electro-mechanical theory of warp-knitted sensors
The sensing theory of the warp-knitted sensors is based on the specific design of SPNF in fabric structures and, when the fabric is loaded with stress, it responds by generating strain and then changing the electrical resistance. The response to strain is related to both the extension of the SPNF and the change of contact points and the contact area between silver-coated filaments and between knitted loops. If the loaded stress does not cause enough structure strain to elongate the SPNF, the sensing performance of the warp-knitted sensor is mainly determined by the rearrangement of contact points between SPNF loops, as Figure 2(a) shows.
Schematic of (a) changes in contact points of loops before and after biaxial stretching and (b) three cases of contact points of silver plated nylon filament (SPNF) loops in fabrics.
According to Holm’s contact theory in Equation (1), for a certain SPNF material with constant hardness and resistance density, the contact resistance mainly relies on the contact pressure and contact points between SPNF knitted loops:
The gauge factor (GF) is used for evaluation of the sensitivity of fabric-based sensors and it is presented by in Equation (2), where ΔR is the resistance change, R is the initial resistance, and ε is the strain that happened during extension:
The above two equations show that when detecting a certain strain, the sensitivity of the designed warp-knitted sensor is determined by the contact resistance change.
Figure 2(b) illustrates SPNF loops in warp-knitted fabrics and contact cases between these loops, which normally occur between two limbs of one SPNF loop (case 1), adjacent loops in two wales (case 2), and in two courses (case 3).
Resistance–strain testing and morphology scanning of SPNF
The extension in the fabricating process would cause certain wear of the silver coating on the SPNF and affect its sensing resistance. The morphology of the SPNF was observed by a DXS-10ACKT (China, Shanghai Tianjing Electronic Optics Technology Co., Ltd) scanning electron microscope (SEM) to evaluate plating damage in the warping and knitting process. The sample was fixed on the SEM stage with conductive glue and then placed on the holder in the instrument. After vacuuming, the morphology of the coating on the yarn surface was observed. Linear resistance of the SPNF was further measured by a hand-held bridge. Each resistance of 10 mm length was recorded around 100 mm to obtain 10 data and then an average value was calculated as the resistance density. Force–strain and resistance–strain tests of the SPNF were also carried out using a YG061FQ (China, Laizhou Electronic Instrument Co., Ltd) tensile machine under a constant extension rate of 500 mm/min with an initial gauge length of 50 mm.
Electro-mechanical testing of warp-knitted strain sensors
Electro-mechanical tests of these eight specimens were performed under biaxial tensile stress using an X-Y fabric biaxial tensile tester (China, Wenzhou Darong Textile Instrument Co., Ltd) to conduct repeated stretching and releasing. Change of fabric resistance was measured and recorded using a KBMS-2400 (America, Tektronix, Inc.) digital multimeter. The experimental setup is shown in Figure 3. Extension levels of 20% in the X direction and 40% in the Y direction are selected to reflect typical human body extension because these sensors are likely used to monitor human motion. The specimen is fixed with two pairs of clamps in both the X and Y directions, and the testing area is 100 mm × 100 mm. Samples are tested with a constant extension rate of 100 mm/min in the X direction and 200 mm/min in the Y direction. The pretension was set as 1 N.
Resistance–strain testing of the silver plated nylon filament (SPNF) and electro-mechanical testing of warp-knitted strain sensors. PA: polyamide; PU: polyurethane.
All the experiments involving human volunteers were performed with full compliance with all local laws and institutional ethical guidelines. Full and informed consent was given by each subject for these experiments.
Results and discussion
Electro-mechanical property of SPNF
During the warping and knitting, the silver plating on the surface of the yarn may be abraded due to the friction between the SPNF and the mechanical elements, thus affecting the conductivity of the yarn. In Figure 4(a) it is evident that there are no visible chips or cracks in the silver-plated coating, even after 300× magnification. This shows that the SPNF selected for this project is able to resist friction and tension during high-speed warping and knitting. However, the above processing affected the linear resistance of the SPNF. As seen in Figure 4(b), SPNF with 77 dtex/24F has a slight change of 0.5% after warping (7.78 Ω/cm) and a growth of 0.7% after being knitted into fabrics (7.84 Ω/cm). A similar increase happens in the SPNF with 44 dtex/12 from 9.48 m to 9.83 Ω/cm (3.7%) in warping and then to 10.14 Ω/cm (3.1%) in knitting. The load–strain and resistance change–strain curves show that only when yarn tension is over 25 N does the tensile strain increase to 5% and the resistance density starts to change. When the strain was less than the warping tension (10–12 cN) and knitting tension (<16 cN), quite a small strain was formed to affect SPNF strength and its silver coating and, thus, the SPNF had good stitching performance. The resistance of the SPNF changed primarily because when the SPNF yarn was threaded onto guide needles and bent into loops, it became twisted and buckled, thus changing the surface contact between these coated filaments.
Schematic of (a) scanning electron microscope images of raw, warped, and knitted silver plated nylon filament (SPNF) and the fabric surface morphology, (b) histograms of the resistance density of raw, warped, and knitted SPNFs with 44 dtex/12F and 77 dtex/24F, and (c) load–strain curves and resistance change–strain curves of SPNFs with 44 dtex/12F and 77 dtex/24F.
Effects of SPNF linear density on the sensing properties
The greater the linear density of the yarn, the greater the contact area between the bound yarn and the yarn contact point under the action of biaxially stretched force, and the more obvious the resistance change. Since the SPNF had numerous fiber strands, the merged fibers were in contact with each other and can be regarded as a parallel connection of conductive resistances. Therefore, the yarn with more fiber strands has a smaller resistance, as shown in Figure 5(a).
Schematic of (a) contact point and yarn cross-section of the silver plated nylon filament (SPNF), (b) resistance change over time of S1 and S4 with SPNF linear densities of 44 dtex/12F and 77 dtex/24F, and (c) resistance change rate over time of S1 and S4 with SPNF linear densities of 44 dtex/12F and 77 dtex/24F.
Figure 5(b) describes the resistance performance of samples S1 and S4 with SPNF linear densities of 44 dtex/12F and 77 dtex/24F, respectively, in the tensile tests and verifies the above conjecture. The range of strain variation was 0–20% in the X direction and 0–40% in the Y direction. The resistance value of sample S1 steadily increases from 5.5 to 6.25 Ω, and that of sample S4 steadily increases from 4.4 to 5.25 Ω. Moreover, the resistance value of S4, which has more fiber strands, is consistently smaller than that of S1. Another obvious phenomenon is that the resistance values of the SPNFs of the two conductive samples first increase and then decrease with the stretching time. The change in fabric resistance during biaxial stretching can be divided into two stages. When the fabric is not stretched, the fabric structure is tight and there are many contact points between the loops and coils due to the presence of spandex within the fabric. When stretching begins, the fabric is stressed and deformed, causing the previously overlapping loops to separate. At this point, the contact points between the loops are reduced, the contact resistance of the fabric decreases, and the total resistance increases. In the second stage, when the loops are completely separated from each other and the fabric continues to be stretched, the contact points between the loops do not continue to decrease. Conversely, the originally 3D loops tend to become two-dimensional planar as the tensile strain is further increased. The contact area between the conductive yarns increases with increasing tensile strain, the contact resistance of the fabric increases, and the total resistance decreases.
Figure 5(c) further shows the relationship between biaxial tensile strain and the resistance change rate. It can be seen that when the fabric is stretched to the specified strain, the resistance change rates of S1 and S4 are 0.13 and 0.19, respectively. This means that the resistance change rate of the conductive fabric increases with the fineness of the SPNF during biaxial stretching The fabric made of finer yarn had fewer contact points. Furthermore, when the yarn was thinner, the equilibrium point was reached earlier. The S1 curve reaches the inflection point earlier than the S4 curve. When the stretching time was less than 7.8 s, the resistance change rates of S1 and S4 were similar. When the stretching time was greater than 7.8 s, the resistance change rate of S1 with a fineness of 44 dtex was smaller than that of S4 with a fineness of 77 dtex, which stabilized earlier.
Effect of the lapping structure on the sensing properties
The change of the loop lapping structure had a great impact on the change in the resistance of the conductive fabric. Knitted fabrics were composed of loops. In the presence of elastic yarn, there was a certain overlap between the upper and lower layers of the loops and the same layers. Figure 6(a) depicts the change in morphology of laps of different lengths before and after biaxial stretching. As shown in Figure 6(a), for conductive loops, the longer the length of the loop lap when unstretched, the greater the overlap and contact area between the laps, and the more contact points are reduced after biaxially stretching, the more pronounced the change in resistance. For spandex loops, the longer the lap length, the tighter the knitted fabric shrank, and the larger the overlap area between the loops. As before, the more contact points that were separated after biaxial stretching, the more pronounced the change in resistance.
Schematic of (a) different lapping structure changes under biaxial stretching, (b) resistance change rate over time with different lapping structures of the silver plated nylon filament, and (c) resistance change rate over time with different lapping structures of spandex yarn.
Figures 6(b) and (c) illustrate the effect of the lapping structure on the electrical properties of warp-knitted conductive fabrics during biaxial stretching. Figure 6(b) shows that a longer lap length of SPNF in the fabric after biaxial stretching led to a higher resistance change rate. When the fabrics are stretched in the X direction by 20% and in the Y direction by 40%, the resistance change rates of S2, S3, and S4 are 0.07, 0.13, and 0.19, respectively. The longer the lapping length, the more stitches there would be, and the larger the contact area. When the fabric was stretched, the contact points between the stitches would be further apart, resulting in a decrease in the number of contact points and an increase in resistance. The above analysis confirms the previous conjecture. Figure 6(c) reveals that a longer spandex lap led to a higher resistance change value under the biaxial stretching condition. When the fabric is stretched 20% in the X direction and 40% in the Y direction, the resistance change rates of S4, S6, and S7 are 0.19, 0.24, and 0.30, respectively. This was due to the fact that spandex had good elasticity, which caused the fabric to shrink in the transverse direction when the fabric was stretched in one direction. When the fabric was stretched in both directions, the loops of the fabric were pulled apart in both directions, and a denser fabric created more contact points, resulting in a larger resistance change rate.
Effect of loop type on the sensing properties
The loop structure of warp-knitted fabric can be divided into open and closed loop forms and the difference between them is whether the lower part of the loop overlaps or crosses, as shown in Figure 7(a). From the viewpoint of the structural unit of warp-knitting fabric, the warp-knitting loop is formed by the yarn in 3D space. When the fabric is stretched under force, the yarn is first stretched from the initial 3D curvature along the direction of force gradually to a two-dimensional plane, which is a combination of the process of straightening of the curved yarn without elongation, and the yarn elongation. Yarn segments are more likely to transfer between open loops than between closed loops when biaxially stretched. Open loops have more contact points and a larger contact area with other loops, thus leading to a greater resistance change rate. Therefore, loops’ opening and closing has a great influence on the performance of the fabric; open loops are preferred where possible.
Schematic diagram of (a) loop opening and closing when biaxial stretched and (b) resistance change rate over time with different loop types of silver plated nylon filament (SPNF).
The impact of loops’ opening and closing on the performance of biaxially stretched conductive fabric was explored. Results showed that the resistance change rate of an open loop was greater in the open loop than in the closed loop. Specifically, when the fabric strain is 20% in the X direction and 40% in Y direction, the resistance change rate of S4 is 0.19 and that of S5 is 0.56, as shown in Figure 7(b).
Effect of threading sequence on the sensing properties
The order of laying yarns in warp-knitted fabrics had a significant effect on their position in the fabric. When the yarn was laid on the first bar, its loops and laps were both located in the outermost layers. Conversely, when the yarn was laid on the last bar, its loops and laps were both situated in the innermost layers. Conductive fabrics were able to generate electrical signals by modifying the contact points between yarns. Thus, the number of bars worn by conductive yarns affected their electrical properties. To explore this effect of biaxially stretched warp-knitted conductive fabric, the relationship between the resistance change rate and time of S4 and S8 is studied and documented in Figure 8(a).
Schematic diagrams of (a) silver plated nylon filaments (SPNF) contact points in loops with different threading sequences and (b) the resistance change rate over time with different threading sequences of SPNF. PA: polyamide; PU: polyurethane.
Figure 8(b) shows that the resistance change rate of S4 is higher than that of S8 when subjected to biaxial stretching. At the strain of 20% in the X direction and 40% in the Y direction, the resistance change rate of the S4 reaches 0.19, while that of S8 reaches 0.04. This difference in resistance change rate was attributed to the difference in threading sequence of the conductive yarns. The conductive yarn of S4 was placed in the outermost layer, resulting in almost no contact between the circle arc and its lap. Conversely, the conductive yarn of S8 was wrapped by the outer non-conductive yarn, resulting in a tight shrinkage state under normal conditions and contact between the loop arc of the conductive yarn and its lap. This caused a reduction in the amount of loop transfer during biaxial stretching, leading to a lower resistance change rate. Therefore, the threading sequence of the conductive yarns played an important role in the performance of the fabric.
Linear fitting of two curves is performed as shown in Figure 8(b). The range of the correlation coefficient R2 is 0–1; the closer to 1, the better the fitting. The correlation coefficient R2 of the resistance variation rate–time curve of S4 tended to 1, which indicated that the fitting degree of the two-way stretching resistance variation rate curve of the conductive yarn knitted in the first comb was very good. The R2 of S8 was 0.7183, which was not as good as that of S4. Thus, the conductive yarn passing through the first comb improved the sensitivity and linearity of the fabric.
Structural optimization and motion detection
Optimal design of warp-knitted strain-sensing fabrics
The above study demonstrates that the structural parameters of loops have an effect on the electro-mechanical properties and tensile response of warp-knitted strain sensors and that warp-knitted strain-sensing fabrics with tensile resilience and wearing comfort are promising candidates for wearable device applications. The tissue structure and process parameters of sensing fabrics were optimized based on the previous study to improve responsiveness in limb motion monitoring, and nine samples of warp-knitted sensing fabrics were knitted. The higher the number of contact point separations between the loops at the same time, the greater the rate of change in resistance of the warp-knitted transducer. Based on this characteristic, the atlas stitch (a pattern repeat of eight courses), which combines open and closed laps where the loops slide more easily when pulled, is considered in the design. In addition, the way the SPNFs are threaded is also taken into consideration in this design. The nine samples can be divided into three groups according to the different yarn-threading methods of the SPNFs. Samples 1–3 are in group 1, samples 4–6 are in group 2, and samples 7–9 are in group 3. The samples in the same group have different tissue structures, and the SPNFs in the samples of different groups are threaded in different ways in order to investigate the effect of dual factors on the electro-mechanical performance of the fabric sensors. The warp-knitting machines used for knitting the specimens were the same as those used in the Preparation of warp-knitted strain-sensing fabrics section. The structural parameters of the specimens are shown in Table 2.
Optimized specimen configurations
PA: polyamide; PU: polyurethane; SPNF: silver plated nylon filaments.
Electro-mechanical test results of warp-knitted strain sensors
The electrical and mechanical properties of the nine specimens were tested according to the experimental method in the Electro-mechanical testing of warp-knitted strain sensors section, and the results are shown in Figure 9. As can be seen in Figure 9(a), the conductive fabric whose tissue of SPNFs is stain lap with open loops has the largest change rate of resistance in the case in which the threading method of the SPNFs is full in. It is evident from the loop diagrams of samples 1–3 in Figure 9(i) that in this case, the tissues that were stain lap had the longest underlap, the largest contact area with each other, and the largest change in contact area after being stretched biaxially. This also verifies the analysis in the Effect of the lapping structure on the sensing properties section. In the second group of samples, samples 4 and 6 have an empty course or wale between neighboring conductive loops in both the vertical and horizontal directions, whereas the conductive loops of sample 5 are one after the other in the vertical direction. After biaxial stretching, the contact area of the conductive yarns in the vertical direction of samples 4 and 6 decreases more, which leads to a slightly larger resistance change rate with time for samples 4 and 6 than for sample 5. In addition, benefiting from the loop structure with a combination of openings and closures, sample 4 has the largest variation of conductive yarn contact points and contact areas, revealing the best electro-mechanical properties. As for the third group of samples, the SPNFs on the conductive bar were too sparse, resulting in a much-reduced contact area of the conductive yarns when unstretched. This fundamentally limits the amount of change in the contact area and contact points between SPNFs after the samples are subjected to biaxial stretching.
Resistance change rate over time for (a) silver plated nylon filaments (SPNFs) full in of three different tissue structures, (b) SPNFs one in one out of three different tissue structures, (c) SPNFs one in two out of three different tissue structures, (d) SPNFs structured as atlas stitch (a pattern repeat of eight courses) with three different yarn-threading methods, (e) SPNFs structured as cord lap with three different yarn-threading methods, (f) SPNFs structured as stain lap with three different yarn-threading methods, (g) samples 1–9, (h) resistance change with time for sample 4, and (i) loop diagrams of SPNFs for samples 1–9.
As shown in Figures 9(d)–(f), analyzing from another perspective, for samples with the same tissue structure and different threading methods, taking the cord lap as an example, sample 5, compared to sample 2, only reduces the contact area of the underlap before stretching, while sample 8, although it also reduces the contact area of the conductive yarns, increases the separation of the conductive loops after the biaxial stretching, which has a significant impact. Likewise, for the stain lap, sample 9 has the worst electro-mechanical properties. Sample 6, although it increased the number of separations between the conductive loops before and after stretching, did not compensate for the loss caused by the substantial reduction of the contact area between the underlaps. Considering all of the above factors, sample 4, which has the advantages of open loops, loop separation, and a proper contact area, showed the best sensing performance. The results are shown in Figure 9(g).
Moreover, after the optimal sensing fabric tissue structure and yarn-threading method were found, repeated biaxial stretching up to 2000 s was performed on sample 4 in order to verify the long-term stability of the sensing fabric in detecting motions. The test results are shown in Figure 9(h), and the resistance change shows that the sensing fabric is able to detect each stretching process in a long-term, stable, and clear way. This indicates that the textile sensor designed in this topic has application stability.
Application of warp-knitted strain sensors in human motion detection
The smart wearable limb motion detection product designed in this project for monitoring upper limb motion can be divided into a carrier zone consisting of 55.5 dtex/13F nylon and 44.4 dtex spandex and a sensing zone consisting of 77.7 dtex/24F SPNF, 55.5 dtex/13F nylon, and 44.4 dtex spandex. Sample 4 with the best strain response capability among the above test results was selected as the effective tissue for the final product. As shown in Figures 10(a) and (c), the participant wore the sensing device with the sensing zone close to the elbow joint and flexed the elbow at slow, medium, and fast speeds for 60 s. The amplitude of arm movement was 90 and 145 degrees, respectively. As seen in Figures 10(b) and (d), the sensor is highly sensitive to the cyclic motion of the elbow joint at the three speeds and can identify each motion clearly and accurately. Furthermore, the resistance change of the sensor remains smooth after several cycles of motion.
Resistance changes for human motion monitoring.
Conclusions
After extensive analysis of the data, it can be concluded that (1) the SPNF has excellent knitting properties and its fineness has an impact on its initial resistance; (2) the retraction of the fabric and its density is affected by the underlap of the SPNF; (3) the resistance change rate of the warp-knitted conductive fabric is highly dependent on the loop type, as well as the threading sequence of the yarn. All these results indicate the importance of considering the aforementioned factors to maximize the performance of the warp-knitted conductive fabric. The results of real-time human motion detection show that warp-knitted elastic strain-sensing fabrics have excellent sensitivity and stability and have enormous potential in the field of smart wearables.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (61902357, 51973027), The Ministry of Education's Industry and Academy Cooperative Education Project (220706429061416), the Fundamental Research Funds for the Central Universities (2232020D-14, 2232020A-08), and the International Cooperation Fund of Science and Technology Commission of Shanghai Municipality (21130750100). This work was also supported by the Chang Jiang Scholars Program and the Innovation Program of Shanghai Municipal Education Commission (2019-01-07-00-03-E00023) to Prof. Xiaohong Qin.
