Stress Dissipation Encoded Silk Fibroin Electrode for the Athlete‐Beneficial Silk Bioelectronics

Abstract The kinetic body motions have guided the core‐shell fabrics of wearable bioelectronics to be elastoplastic. However, the polymeric electrodes follow the trade‐off relationship between toughness and stretchability. To this end, the stress dissipation encoded silk fibroin electrode is proposed as the core electrode of wearable bioelectronics. Significantly, the high degree of intrinsic stress dissipation is realized via an amino acid crosslink. The canonical phenolic amino acid (i.e., tyrosine) of silk fibroin is engineered to bridge the secondary structures. A sufficient crosslink network is constructed when tyrosine is exposed near the amorphous strand. The stress dissipative tyrosine crosslink affords 12.5‐fold increments of toughness (4.72 to 58.9 MJ m−3) and implements the elastoplastic silk fibroin. The harmony of elastoplastic core electrodes with shell fabrics enables the wearable bioelectronics to employ mechanical performance (elastoplasticity of 750 MJ m−3) and stable electrical response. The proposed wearable is capable of assisting the effective workouts via triboelectricity. In principle, active mobility with suggested wearables potentially relieves muscular fatigues and severe injuries during daily fitness.


Investigation of the electromechanical response under the mechanical load.
To concentrate the tensile stress, EPSF was shaped into the dog bone gauge regarding ISO 527-2, 1BA specimen type (Figure 4a). A universal testing machine (UTM; Model 3366, Instron, USA) was used to apply the uniform tensile and bending load to EPSF. For the tension experiment, the gripped region was 50 mm 2 and the tensile rate was 5 mm min -1 . The tension was kept to the maximum strain in Figure 4d. UTM released the specimen when the stain reached 30% and re-stretched the specimen in Figure 4e. The cyclic test was repeated four times. Besides, the bending test was performed on the beam gauge in 60 × 10 mm 2 (width × length). The bending rate (i.e., crosshead speed) was 5 mm min -1 . The cyclic test was conducted, as mentioned above. The relative humidity was kept at 50% during the mechanical deformation.
In situ variation of resistance was measured by the multimeter (Fluke 179, Fluke, USA). [2] Pre-stretched EPSF (ε ≈ 130%) was fixed and vacuum-stored to investigate the crystal properties by Raman spectroscopy.
Multilateral characterizations of silk fibroin electrode. The electronic level of silk fibroin electrode was elicited using kelvin probe force microscopy (KPFM; NX-10, Park systems) and UV-visible spectroscopy (UV-Vis; Evolution 300, Thermo Scientific, USA). KPFM was conducted to measure the work function. The non-contact tip was coated with the platinum with a tip size of 27.5 × 255 × 15 μm 3 (width × length × height). The resonance frequency, spring constant, and sensitivity were 75 kHz, 2.8 N m -1 , and 33.3 V μm -1 , respectively. The driving voltage was 2.0 V. 10 × 10 μm 2 measurement area was divided into 512 pixels and scanned at a 1.00 Hz rate. Prior to the KPFM measurement of silk fibroin electrodes, the work function of highly ordered pyrolytic graphite was estimated as the standard. The optical band gap was calculated from UV-Vis spectra over the range of 200 to 900 nm. The temperature was kept at 25 ℃ by means of the thermoset accessory. The bandwidth, data interval, and scan rate were 2.0 nm, 1.0 nm, and 200 nm min -1 , respectively. Xenon lamp was the light source, and the standard baseline was measured by the empty cell (1 cm length). Fourier transform infrared spectroscopy (FT-IR; FT/IR-4700, JASCO, Japan) was performed to monitor the vibrations of amino acids. Additionally, Raman spectroscopy and XRD were conducted to monitor the inherent characteristics of 1 × 1 cm 2 silk fibroin electrode (e.g., crystal properties and tyrosine environment). The experimental protocol was the identical as mentioned above. 1400 to 1500 cm -1 , 1640 to 1720 cm -1 , 800 to 900 cm -1 spectra individually are related to amide Ⅲ, amide Ⅰ, and tyrosine doublet, respectively. XRD pattern in the range of 3 to 40º informed the (110) direction of the β-sheet crystal. Furthermore, the fluorescence spectrometer (FP-8300, Jasco, Japan) and rheometer (MCR 302, Anton Paar, Austria) were conducted for the in-depth characterizations. The silk fibroin electrodes (2 cm diameter) were subjected to excitation at 315 nm. The emission was recorded at 350 to 500 nm in 2 nm increments. The scan rate was 500 nm min -1 , and the slits were set to yield sufficient fluorescence resolution (ca. 2.5 nm). The frequency-dependent storage modulus of 1 × 1 cm 2 silk fibroin electrode was measured under the frequency sweep mode (set as 1.0% strain amplitude) and the constant temperature of 25℃. The gap was 500 μm, corresponding to the thickness. Finite element simulation was conducted with a commercial software ABAQUS (ABAQUS Inc. Student edition). The unit silk fibroin was considered as the elastic beam. The detailed parameters utilized in the simulation are summarized in Table S1. Engineering strain vs. stress curves was recorded during the tension using the UTM. The detailed experimental protocol (e.g., tensile rate) was accorded with the above.
Fabrication of EPSF-incorporated triboelectric fabric band. EPSFs were cut into ten pieces of 1 × 6 cm 2 each. Every piece was fully packaged using commercial PTFE (thickness of 0.08 mm, Chukoh Chemical Industries, Korea) and nylon (thickness of 0.05 mm, KAWAGUCHI, Korea) tapes. PTFE-packaged EPSF (EPSF_P) and Nylon-packaged EPSF (EPSF_N)corresponding to warp and weft fabric band-were woven into a plain weave pattern. The final sample of the 5 × 5 fabric band has a natural woven structure.
Characterization of triboelectricity. Vertical excitation input was applied to a 5 × 5 fabric band device using a vibration tester (ET-126B-4, Labworks, USA), a function generator (AFG3021C, Tektronix, USA), and an amplifier (pa-151, Labworks, USA). The vertical vibration amplitude and frequency were set at 3 mm and 6 Hz, respectively. The voltage output and current output were measured using a mixed domain oscilloscope (MDO 3014, Tektronix Co, USA) and a low-noise current preamplifier (MODEL SR570, Stanford Research Systems, USA), respectively. The voltage probe 1 and 2 were connected to the warps (EPSF_P) and wefts (EPSF_N). The voltage output was obtained from the electrical potential difference between probes 1 and 2. The current output was measured exploiting one current probe, directly connected with the oscilloscope. Principal component analysis (PCA) was conducted as the statistical approach to evaluate the obtained waveform.
In vitro cell experiments. Here, the fibroblast (e.g., human dermal fibroblast (HDF)) and myoblast (C2C12) were the model cell lines that are essential in the aerobic resistance exercise. Prior to the various assays, the model cells were fully sub-cultured and stabilized by incubation at 37 °C in a humidified incubator with 5% CO2 in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum (Gibco Life Technologies, USA). To verify the no cytotoxicity of EPSF_P and EPSF_N, prewashed fabric bands of 1 × 1 cm 2 were thoroughly sterilized by the overnight UV exposure and transferred 24-well culture plate. The cells (HDF and C2C12) with a density of 5.0 × 10 4 cell mL -1 were seeded in each well with a pre-sterilized fabric band. After 24 h incubation, a cell counting kit assay (CCK assay: D-Plus CCK cell viability assay kit, Dongin LS, Korea) was performed following the manufacture's protocol. Specifically, the wells were treated with 10% (v/v) CCK assay solution and incubated for 2 h. The relative viability was monitored by tracking the optical density at 450 nm. The lab-made multi-well culture plate, assembly of indium tin oxide (10 × 10 cm 2 ), and removable chambers (three well chambers, ibidi) were exploited for in vitro electrostimulation experiments. Before the cell culture, the lab-made culture plates were sterilized deliberately with 70% (v/v) ethanol and UV exposure. The power source of electrostimulation was triboelectricity of 5 × 5 fabric band device or equivalent triboelectricity (i.e., VOC ≈ 20 V with 6 Hz frequency). Both sides of the culture plate were individually connected with the warp (EPSF_P) and weft (EPSF_N) of the 5 × 5 fabric band device inserted in the vertical vibration tester. Moreover, HDF and C2C12 (in 4.0 × 10 4 cell mL -1 ) were incubated overnight for sufficient adherence. After that, 2, 4, and 6 h of electrostimulations (n=3) were performed according to triboelectricity of 5 × 5 fabric band.
10% (v/v) CCK assay solution was treated and incubated for 2 h simultaneously with electrostimulation to analyze the cellular dehydrogenase activity. The optical density was measured as described above. The enzyme-linked immunosorbent assay (ELISA) was performed to quantify the amount of secreted protein and growth factor considering the origin of model cells. The cell culture supernatant of HDF (1 mL) directly after the electrostimulation was subject to the type Ⅰ collagen ELISA (Human Pro-Collagen Ⅰ alpha, R&D systems, USA and Mouse Pro-Collagen Ⅰ alpha, abcam, USA). Furthermore, the lysis procedure was followed to quantify the amount of basic fibroblast growth factor (bFGF) from HDF (Human FGF ELISA Kit, Thermo Fisher, Scientific, USA) and the insulin-like growth factor 1 (IGF-1) from C2C12 (Mouse IGF-1 ELISA Kit, Thermo Fisher, USA). In detail, the electrostimulated cells were gathered from 10% (v/v) trypsin-EDTA solution and exposed to the pH 7.4 icy lysis buffer for 30 min. The component of NP-40 lysis buffer was 150 mM NaCl, 1% (v/v) Triton X 100, and 50 mM Tris-Cl. The cleaved plasma membranes were pellet down by 13,000 rpm centrifugation for 10 min. Processed supernatant was subject to bFGF and IGF-1 ELISA. All ELISAs were proceeded according to the manufacturer's protocol. The supernatants from culture media or lysis buffer were diluted 1/20 (type Ⅰ collagen from HDF; R 2 = 0.9855), 1/10 (type Ⅰ collagen from C2C12; R 2 = 0.9855), 1/2 (bFGF from HDF; R 2 = 0.9973), and 1/4 (IGF-1 from C2C12; R 2 = 0.9729) for the reliable data within the ELISA sensitivity range (pg mL -1 to ng mL -1 ).  Raman spectrum of β-chitin was in accordance with previous studies. [3] (b) Near 1.0 value of D/G ratio (ID band/IG band = 1.04) suggested that the physical reduction resulted in the highly reactive carbon structure. [4,5] XRD monitorings of (c) β-chitin, (d) rCDC. (c) XRD patterns of β-chitin were observed. [3] (d) (002), (100) peaks suggested the formation of sp 2 -hybridized carbon hexagonal structure within rCDC. [6] XPS spectra of (e) β-chitin, (f) rCDC. O --, 25.5 mol%) nitrogen atom, respectively. [9,10] (e) The schemes suggested that the hydrothermal-carbonization and physical reduction of β-chitin formed the electrical filler (rCDC). The chemical structure of rCDC was conjectured from the three spectroscopies above.
In Figure S2g, the crystallographic characterization of electrical filler (rCDC) was featured according to the D/G ratio ( Figure S2b) and XRD pattern ( Figure S2d). Following Equation S1, the crystallite size (La) was determined as 42.3 Å.
here, C(λ) (ca. 44 Å) was the prefactor of the 532 nm laser for the Raman scattering. Then the d-spacing (d) along with the lattice directions were measured using Bragg's law (Equation
here, n was the order of reflection; λ is the wavelength of the incident X-rays; θ was the angle of incidence. In our experimental environment, nλ was 1.54. (~1165 cm -1 ), amide Ⅲ (~1230, ~1260 cm -1 ), Asn (~1400 cm -1 ), amide Ⅱ (~1505, ~1525 cm -1 ), and amide Ⅰ (~1620, ~1650 cm -1 ).  Table S1. [12] The extended finite element simulation (here, the tyrosine crosslink was off the table) further evidenced the essential role of interfacial tyrosine crosslink in ΓD. Figure S5 and Supporting Video S1 and S2 inform that the in situ crack propagations resulted from the tensile stress. According to the symmetric coarse-grained model (i.e., all around propagating crack is crystal domain), the uniformly distributed stress maximally converged to 31.1 MPa at the crack tip. However, the symmetric crystal structure hardly represented the realistic model of the silk fibroin. [13] Therefore, the asymmetric beam was evaluated, in which the crack was flanked by both crystal and amorphous domains. In the asymmetric model, the stress was nonuniformly biased to the amorphous region. Additionally, the amorphous side crack tip underwent significant tensile stress of 66.7 MPa, ~2.14 times enhanced stress of the symmetric model. Without the stress dissipative crosslink of two disparate domains, the delicate amorphous strand (i.e., 9.9 pN Å -1 ) would preferentially be fractured; that is, the stumbling block of elastoplasticity.    [17] The various fillers have been introduced to produce mechanically and electrically high-performance conductors, such as Hofmeister series ions, [18] inorganic nanoparticles, [19] carbon derivatives, [20] and polysaccharides. [21] For instance, 2D carbon elements helped to produce highly conductive biopolymers. [22,23] However, the introduced fillers may present the gravitational sedimentation and interrupt the polymer strands according to Stoke's law. [24,25] Since the density of electrical filler (rCDC; ~2.26 g cm -3 ) was higher than the value of silk fibroin (~1.40 g cm -3 ), the spontaneous sedimentation should be investigated.
When ΦrCDC < ΦC, the introduced fillers solely formed the percolation network. Otherwise, the sedimentation would be spontaneous upper ΦC. According to the power law in Equation S3 and S4, ΦC was rationally figured out as a function of sediment thickness (tsediment). [17] In Equation   S4, α was the index to judge ΦC.
Technically, when the appropriate ΦC was taken into account, α would be in the range of 1.6 to 2.5. As proof of concept, the linear slope (α) in Figure S8 was 2.27, assuming ΦC = 1.0 wt%. Therefore, ΦrCDC = 1.0 wt% was the upper limt of percolation threshold.   When weaving the fabric with near-unrealistic construction, the cloth's fell would creep beyond the reed's forward position, and the bumping would occur. [27] The coverage factor (k) was the weavability index to predetermine whether the realistic or unrealistic fabric construction. Peirce's geometric parameters of the plain weave were summarized in Figure   S11. In the suggested model, Equation S5 should be satisfied to be the weavable structure. [ where k1 = 28d1/p1, k2 = 28d2/p2, and β = t2/t1. The variation of thickness (β) was computed as 3.23, indicating the unit intersection should be deformed substantially for the sustainable weaving. In detail, the unit cell underwent 300% stretching maximally in the warp (or weft) axis to be weavable. The corresponding deformation stress was localized ~150 MJ m -3 . The estimated stress was the function of workouts dynamicity and could reach up to 450 MJ m - Figure S12. Enhanced intrinsic orders of EPSF according to the tensile load. Raman studies were conducted to elicit the order of crystal from amide Ⅲ bands when (a) no tension, (b) 130% tensile strain was applied. The Raman marker (ca. 1235 cm -1 ) represented the ordered crystal (i.e., Φorder). Otherwise, the Raman components (ca. 1250 cm -1 ) corresponded to the disordered random coil conformation (i.e., Φdistort). [11,28] As the tensile strain reached 130%, the order of intrinsic structures simultaneously increased ~193%. [23]   Here, the suitability of EPSF for triboelectrification was inquired. The single electrode mode was adopted as general triboelectrification-based electricity generation method. Figure S14 where each V(t), I(t), and T represents measured peak voltage output, peak current output, and measurement period.
Regardless of the polarity of surface charge (representatively, polar directions of peak outputs), Al and EPSF generated comparable electrical performances, suggesting that EPSF could behave as the electrode similar to Al. In Figure S14a, Al has produced a maximum peak VOC of 43.2 V (RMS 7.83 V) and peak ICC of 2.8 µA (RMS 0.41 µA). In Figure S14b, EPSF has generated a maximum peak VOC of 51.2 V (RMS 9.14 V) and a peak ICC   Previous studies on triboelectric fabric or textile hardly represented the woven structure, and the simplified working mechanism was proposed as summarized in Figure   S16a. [29] In other words, a slightly nonrealistic mechanism, considering only single contact with cotton, was featured; that is, a small number of free-electrons flowed due to the no additional surface charges by internal contact within the woven structure ( Figure S16b). The electrical output of the model device in Figure S16c was measured. In Figure S16d, the single contact mechanism sample generated a maximum peak VOC of 3.6 V. Notably, the maximum peak VOC of a single contact mechanism was 6.5 times lower than VOC of double contact mechanism. In this work, among fabric-related triboelectric series, each PTFE (i.e., Gore-tex) and Nylon was adopted for packaging EPSFs as the most negative and positive material. Owing to the large differences in electron affinities, the packaged EPSFs can be more efficiently charged via (iii) internal contact with each other ( Figure S17a). Moreover, cotton was used as a counterpart material for (i, ii) external contact due to its neutral position compared to PTFE and nylon. The possibility of connections is (i) PTFE-cotton, (ⅱ) nylon-cotton, and (ⅲ) PTFE-nylon.
If other slightly positive fabric-related triboelectric material such as wool is considered as counterpart material, there is no significant effect on electrical output performance ( Figure   S17b). It is because although the amount of charge generation at external contact (ii) reduces, the charge generation amount at external contact (i) increases, and thus total charge generation amount remains similar to that of Figure S17a. The possibility of connections is (ⅰ) PTFE-wool, (ⅱ) nylon-wool.
The tendency above will be identical for polyethylene as counterpart material which is a slightly negative material; in other words, decreased surface charges at external contact (i) are compensated by external contact (ii) ( Figure S17c). The possibility of connections is (i) PTFEpolyethylene, (ⅱ) nylon-polyethylene.
Meanwhile, suppose other fabric-related triboelectric materials such as polyethylene and wool are exploited to package EPSFs instead of PTFE and nylon respectively when the counterpart material is cotton ( Figure S17d). In that case, the total amount of surface charges will decrease during both external (i, ii) and internal contact (iii) because the relative difference of electron affinity becomes small between each internal and external contact material.
Therefore, the small number of free electrons will flow between each EPSF electrode by electrostatic induction. Here, the possibility of connections is (i) polyethylene-cotton, (ⅱ) woolcotton, (ⅲ) polyethylene-wool.

Figure S18. Peak VOC plots of 2 × 2 unit as a function of contact-separation frequency.
During the workouts, various body motions may cause irregular contact-separation frequency. In this regard, the electrical output of 2 × 2 unit was evaluated depending on vertical input frequency ( Figure S18). As the frequency increases from 3 Hz to 6 Hz, a similar peak VOC output of ca. 23 V was measured (left column in Figure S18). But peak ICC output increased from 1.24 to 2.16 µA (right column in Figure S18), following the previous studies on triboelectrification-based electricity generation. [30,31] Figure S19b showed the measured VOC outputs (i.e., 11.2 V) when 5 × 5 fabric band device underwent vertical contact with the sportswear (e.g., during cable side lateral raise). In Figure S19c, the sliding contact motion (e.g., during running) has generated the maximum peak VOC of 8.6 V and produced the double peaks originated from the two times of contact-separations during one swaying arm back and forth.  The type of sports performance considerably affected the source of ATP supplement. [32] In terms of the sprint events lasting seconds, the dominant ATP was replenished from the anaerobic metabolism (e.g., breakdown of phosphocreatine and glycogen). Otherwise, the aerobic metabolism (e.g., phosphorylation of carbohydrate and fat) supplied the vast ATP for the long-term muscle contraction events lasting hour scale. In detail, the aerobic metabolism solely relied on carbohydrate oxidation to perform the higher intensity, vice versa. [32] Since the daily workout routine usually carried on at least one hour, the aerobic metabolism was taken into consideration.
Here, the percentage of maximum oxygen uptake (% VO2,max) was the quantified index of workout intensity to conduct the mathematical and theoretical biology (Figure 6b-d). Since the respiratory exchange ratio collected at the mouth presented the metabolic activity at mitochondrion, the regarding work intensity as a function of % VO2,max was one of the general approaches in the exercise physiology (e.g., spirometry device). [32,33] The cross-validation procedure was conducted to verify whether % VO2,max represented the workout intensity in the mathematical biology approach.
First of all, the clinical database of the college-aged students was gathered referring to the prior studies (Table S2). Then, the intuitive index, power output (P in watt (W)), was explored in the green line plot (•) of Figure 6c and S21. Here, P indicated the muscular contraction activity performed by specific oxygen uptake (% VO2,max). The empirical dependence of the two variables was as the following relationship; % VO2,max = 13.895P+151. [34] The comparative analysis between calculated and clinical P values verified the reliability of the approach above; that is, P(75% VO2,max) was 203 W in the mathematical biology ( Figure S21d), and 200 W from the clinical measurement. [32] Therefore, the theoretical investigation of % VO2,max accurately informed the workout intensity.
In the blue line plot (■) of Figure S21, the ATP utilization rate to perform the specific intensity of muscle contraction (-∆H in mmol ATP kg -1 s -1 ) was estimated following Equation S7 and S8. [35] For the workouts below the maximal intensity (0~75% 2, ); were substantially dependable since the mathematics was accorded with the clinics; that is, -∆H (75% VO2,max) was 0.442 mmol ATP -1 kg -1 in Figure S21d and 0.4 mmol ATP -1 kg -1 in the clinical results. [32] Therefore, the vast ATP (i.e., higher -∆H) was required as the muscle activity enhanced (i.e., higher % VO2,max).
The first compensation of the required ATP resynthesis (-∆H) was the consumption of intramuscular ATP storage. [32] The general intramuscular ATP storage was not sufficient, i.e., 5 mmol kg -1 . Therefore, ATP stock would be exhausted within 15 s to perform the submaximal muscle contraction event since -∆H(75% VO2,max) was 0.442 mmol ATP kg -1 s -1 . Figure 6c was obtained regarding ATP storage would be positively expanded as the cellular activity improved ( Figure S22). The symbols * (P value < 0.05), ** (P value < 0.01), and *** (P value < 0.001) indicated the statistical significance. The error bars and statistical analysis were originated from n=3 results.
In vitro experiments were designed since theoretical biology has assumed that the cellular metabolism could represent the whole-body event. Notice that the electroceutical devices (e.g., functional electrical stimulation to induce muscle contraction [36] ) have clinically influenced human activity. In Equation S9 and S10, the required ADP concentration to perform the specific muscle contraction (ΦADP-Required) was deduced from Michaels-Menten kinetics (middle term in Equation S9) and spirometry principal (right term in Equation S9). [37] 2, where VO2,op was the activity of phosphorylation; KS1 was the 50% activity constant of phosphorylation (i.e., 0.0631 mmol 2 kg -2 ); P, W, and VO2,max was the power output, the body weight, and maximum oxygen uptake, respectively ( Figure S21, Table S2). Ks4 was the oxygen-workload constant. ΦADP-Required was the function of Ks4 since the respiratory exchange ratio collected at the mouth represented the mitochondrion activity. [33]  48 mL kg -1 min -1 . [38]