Introduction

The development of microelectronic technology has led to the miniaturization of electronic devices,1,2,3,4 which in turn has given rise to the demand for flexible electronic devices.5,6,7 Currently, popular wearable and flexible electronic products are mostly rigid and rely mainly on the fabrication of miniature integrated circuits on wafers to realize various electronic functions, such as smartwatches, smart glasses, sports monitoring wristbands, and bracelets.8,9,10,11,12 However, these devices have high manufacturing costs, cumbersome production steps, and incompatibility with organic materials; the technology used for these devices does not meet the requirements for the preparation of large-area wearable electronics.13,14 Additionally, rigid microelectronic devices are mostly used as small accessories, which limits the diversity and flexibility of the wearing position. In addition, there is a large difference in Young’s modulus between rigid electronic devices and the human body’s muscles or skin; moreover, when the human body moves, rigid electronic devices will be separated from the surface of the body, resulting in signal artefacts and baseline drift. In contrast, flexible electronics, as opposed to traditional silicon-based electronics, are more adaptable to varied surfaces, may be attached to a range of rigid or flexible complex surfaces, and have greater flexibility to function on curved surfaces and in a stretched condition.15,16,17 Among them, flexible strain sensors are electronic devices that convert mechanical deformation signals into electrical output signals (generally resistance signals), and their unique flexibility and stretchability allow them to be directly affixed to human body joints for real-time monitoring of the biological state of the human body.18,19,20 Thus, flexible strain sensors have great potential for application in the fields of health monitoring, rehabilitation medicine, and human–computer interaction.21,22,23 The materials utilized to make such devices have an enormous impact on the performance of strain sensors. Current materials used for flexible strain sensors include hydrogels,24 elastomers,25 and organic‒inorganic hybrid materials,26 which can meet the requirements for flexibility, but the basic materials used are not degradable, environmentally friendly, or conducive to sustainable development. This is because as the demand for electronic devices grows, electronic waste and environmental problems ensue. Therefore, the development of nontoxic, recyclable, and degradable green devices has become particularly important.

Recently developed hydrogel-based transient electronic devices based on cellulose derivatives, natural waxes, poly(lactic-co-ethanolic acid), poly(lactic acid), poly(vinyl alcohol), and silk can be degraded upon external triggering.27 However, they are complicated to prepare (usually requiring various chemical modifications, in which toxic solvents may be used), followed by poor plasticity (some materials have difficulty adhering to the skin). The traditional type of dough is made by artificially growing wheat and then milling it.28 The shape and function of dough can be easily reassembled and reconfigured through the addition of a certain amount of salt and water, and the internal content of covalent disulfide bonds and a variety of non-covalent hydrogen bonds endow the dough with good mechanical, adhesive, and self-repairing properties. Moreover, the presence of the salt solution imparts ionic conductivity to the dough. Although the lack of resilience of dough makes it challenging to use in resistive strain sensors, reliable capacitance changes can be achieved through the use of an elastic dielectric layer. Thus, conventional dough has the potential for use in flexible electronics. Generally, water-based doughs are subject to functional loss at low temperature. This is because freezing water molecules at subzero temperatures limits ion mobility in the dough while also causing the dough to stiffen and lose its initial mechanical flexibility, which greatly limits its operating temperature range. Accordingly, the development of doughs with anti-freezing properties is expected to solve cold problems and widen the range of application temperatures.

In this work, we prepared flexible, renewable, and reconfigurable capacitive sensors using conventional dough with the aid of the antifreeze properties of ethylene glycol, overcoming the problem of low temperature. The unique rheological properties of the dough endow it with better plasticity and restructuring ability. The addition of NaCl to the dough results in ionic conduction. The capacitive flexible sensors assembled with dough as a dielectric layer have high gauge factor (GF) values and reliable recognition of joint motion. Nevertheless, the design and evaluation principles should be generally applicable and not limited to specific foods or biomaterials.

Experimental

Materials

Flour was purchased from Wuteli Group Xinxiang Flour Co., Ltd. Sodium chloride (NaCl, 99%, analytical research grade [AR]) was obtained from Tianjin BASF Chemical Co., Ltd. Ethylene glycol (EG, 99%, AR) was obtained from Tianjin Fuchen Chemical Reagent Co.

Preparation of EG-Based Antifreeze Dough Sensors

NaCl was added to a 0.2 M solution. EG was added to the NaCl solution to form a mixed H2O/NaCl solution according to the proportions in Table 1, which was subsequently poured into the flour, kneaded, and homogenized repeatedly to form a smooth dough that was allowed to rise for 30 min. In addition, for comparison, a conventional-style dough without the addition of EG was prepared. Very high bond (VHB) tape as a dielectric layer was sandwiched between two layers of the EG-based dough sample, which was cut into 10 mm × 10 mm × 1 mm squares. The EG-based dough samples were connected to 20 mm × 2 mm × 0.2 mm copper plate electrodes to facilitate connection to the acquisition device. To prevent water evaporation, dough-based flexible capacitive sensors were encapsulated with VHB tape. Therefore, the total size of the device is 10 mm × 10 mm × 8 mm.

Table 1 EG-based antifreeze doughs
Full size table

Characterization

Scanning electron microscopy (SEM; S-3400N, Hitachi Ltd., Japan) was used to analyse the EG-based antifreeze dough. The elemental composition and surface functional groups of flour, gluten, and freeze-dried dough were tested using a Fourier transform infrared spectrometer (FT-IR, Frontier, PerkinElmer, USA) in the wavelength range of 500-4000 cm−1 with a scanning frequency of 32 times. The Horiba LabRAM HR Evolution was employed to obtain the Raman spectra of flour, traditional dough, and EG-based antifreeze dough. An Anton Paar MCR 302 was used to test the rheological properties of the EG-based anti-freezing doughs. Dynamic sweep measurements were carried out at 25°C in oscillation mode with a fixed oscillation strain of 0.5% and a fixed gap of 1 mm. Viscosity tests were carried out at shear rates of 0.1–100 s−1 and at a temperature of 25°C. The viscosity was measured in the range of 10–70°C. Temperature scans of the dough were measured over a temperature range of 10–70°C with an oscillatory strain of 0.5% and a frequency of 1 Hz. Continuous step strain measurements from high-amplitude oscillations (γ = 100%) to low-amplitude oscillations (γ = 1%) were performed on the dough at a frequency of 1 Hz (25°C). The recoverability of the dough was quantitatively assessed after several drying–expansion cycles. The dough was dried at 40°C, an equal amount of water was added, and the dough was kneaded back to its original shape. The recoverability of the dough was quantitatively assessed by G′ recovery measurements after several drying–expansion cycles. The capacitive behaviour of the flexible sensors was recorded by an LCR metre (TH2830) at a sweep rate of 1 kHz and an AC voltage of 1 V. To test the sensing performance, the relative capacitance change (ΔC/C0) during stretching was recorded using an LCR metre (TH2830). In addition, the sensors were attached to different parts of the body (throat, fingers, wrists, elbows, and knees) to monitor human movement. The concentration of each sample was measured three times at room temperature. The sensitivity of strain sensing (GF) was calculated according to Eqs. 1 and 2, respectively:29,30

$$ \frac{\Delta C}{{C_{0} }} = \frac{{C - C_{0} }}{{C_{0} }} $$
(1)
$$ GF = \frac{{\Delta C/C_{0} }}{\varepsilon } $$
(2)

where C0 and C are the capacitance before and after applying the strain, respectively, and GF and ε are the values of strain change before and after stretching the sensor, respectively.

Experiments on the Human Body

All participant experiments were regulated with the approval of the Institutional Review Board at Hunan University of Technology following the provision of written informed consent from the volunteers. The volunteer is the first author himself.

Results and Discussion

The preparation of EG-based antifreeze dough is simple, merely requiring the addition of H2O/EG binary solvent containing NaCl (Fig. 1a). Similarly, edible dough is a complex polymer–granule system in which a dense and continuous gluten network is reinforced by spherical starch granules of different sizes. During hydration and kneading, some covalently cross-linked disulfide bonds in the gluten and in the bran within the starch granules, as well as intramolecular and intermolecular hydrogen bonding between starch and gluten, contribute to the formation of the elastic network. The morphology of the gluten washed out of the dough showed that it exhibited a typical polymer–particle structure, where the continuous polymer phase was the gluten structure and the particles were the starch granules (Fig. 1b). The formation of disulfide bonds was confirmed by Raman spectroscopy (Fig. 1c). The band near 525 cm−1 corresponds to a trans–gauche–gauche S‒S stretching pattern and is more common in traditional dough than in wheat flour.31 In addition, characteristic bands attributed to -S‒S- were also present in the EG-based antifreeze doughs, suggesting that the addition of EG does not affect the formation of disulfide bonds during the hydration of flour. Non-covalent hydrogen bonding was confirmed by the FT-IR peak at 1650 cm−1 assigned to amine I and at 1550 cm−1 attributed to amine II (Fig. 1d). However, due to the lack of strong covalent cross-links inside the dough, relaxation of the protein polymer chains in the dough also results in plasticity and shapeability. Overall, the synergistic effect of covalent and non-covalent interactions renders the dough uniquely viscoelastic. The rheological profile showed that the dough exhibited shear-thinning behaviour (Fig. 1e). The viscosity and shear stress are related to the shear rate. At a low shear rate (0.1 s−1), the viscosity is approximately 6700 Pa s. When the shear rate is increased (100 s−1), the viscosity decreases to approximately 70 Pa s. In the region of 0.1–100 s–1, we can use the power-law fluid model K = τ γn, where K refers to the consistency index and n is the exponent, to analyse the shear-thinning behaviour quantitatively. The exponent n was calculated from a power-law fit to be 0.38. The Rabinowitsch–Mooney equation was then used to predict the shear rate as the power-law fluid flowed through a cylindrical tube with a radius r. The tip diameter of the tube was 0.6 mm, and the extrusion rate was 6 mm s−1. Therefore, the corresponding viscosity of the dough during extrusion is approximately 45 Pa s. During extrusion, the viscosity of the dough is approximately 45 Pa s, assuming a shear rate much lower than 1 s−1. The dough quickly returned to a high viscosity (>1500 Pa s) and thus retained its shape. Owing to its excellent viscoelasticity, the dough can be plastic and highly stretchable. It can be shaped into a ball or stretched into a twist, then returned to its original shape and stretched again (Fig. 1f). This process can be repeated many times. In addition, the dough can be reconfigured. When the two parts of the dough coloured red and blue are kneaded together, they fit each other without any obvious gaps, and they can be mixed to form a uniformly coloured purple dough (Fig. 1g), which demonstrates the excellent reconfigurability of the antifreeze dough. Such excellent plasticity and reconfigurability make it ideal for proof-of-concept demonstrations of renewable and reconfigurable ionic electronics.

Fig. 1
figure 1

(a) Schematic illustration of the preparation of the EG-based antifreeze dough and gluten; (b) SEM images of dough; (c) Raman spectra of dough, EG-based antifreeze dough and wheat flour; (d) FT-IR spectra of wheat flour, dough, and EG-based antifreeze dough; (e) shear-thinning behaviour of antifreeze dough at 25°C; (f) photo of dough being extruded into a ball and twist; (g) schematic of dough reconfigurability (Color figure online).

Full size image

All the unique viscoelastic properties of dough are closely related to its rheological properties. It is worth noting that some NaCl was added during the softening process, and its addition can introduce free ions to the antifreeze dough on the one hand, playing the role of ionic conductivity, and on the other hand, the ionized ions can strengthen the electrostatic interactions of the biomolecules and enhance the density of the polymer network, which is conducive to the enhancement of the mechanical properties of the dough. From Fig. 2a, the dynamic storage modulus (G′) and loss modulus (G″) decrease in the low-frequency region, and from the overall frequency, it can be seen that G′ is higher than G″, which indicates that the dough exhibits an elastomeric state across the whole frequency range.32 Figure 2b shows the dynamic strain sweep test results for the doughs in the range of 0.1–1000%. At low strain (e.g. 1%), the rheological behaviour of the bread is similar to that of the frequency scan, exhibiting an elastomeric state, while the opposite is true in the high-strain region, indicating that the network of elastomers is broken and presents a plasticized visco-fluid state. Furthermore, from the continuous step strain measurements (Fig. 2c), the 1% strain condition has initial G′ and G″ values of 25,400 Pa and 8430 Pa, respectively. Recovering from 100% strain to 1%, G′ and G″ can be recovered to more than 70% of the original values, and the same values can be recovered in the subsequent three cycles, which fully demonstrates that the EG-based antifreeze doughs have excellent reconfiguration ability.

Fig. 2
figure 2

(a) Frequency dependence of G′ and G" of EG-based antifreeze dough; (b) strain sweep measurements from 0.1% to 1000% for EG-based antifreeze dough; (c) continuous step strain oscillation measurements of EG-based antifreeze dough from 0.1% to 100% and then back to 0.1% strain; (d) variation of G′ and G″ versus temperature in the original dough and EG-based antifreeze dough (fixed oscillatory strain of 0.5%, fixed frequency of 1 Hz); (e) frequency dependence of G′ and G" for initial dough and dough recovered after one drying–hydration cycle.

Full size image

Additionally, the recyclability of dough is related to its rheological properties. From the variable temperature rheological properties in the temperature range of 0–70°C (Fig. 2d), it can be seen that an increase in temperature results in a decrease in both G′ and G″, which is mainly attributed to starch gelatinization and gluten denaturation by heat-induced conformational changes. Notably, G′ and G″ did not decrease significantly and still maintained G′ greater than G″ throughout the temperature range, suggesting that the effect of temperature on the rheological properties of EG-based antifreeze doughs was not significant. These results indicate that the dough does not undergo irreversible changes in its mechanical properties in the range of 0–70°C. When the temperature is less than zero, the G′ of the dough in the pure H2O system increases significantly to approximately −0°C, indicating a hard solid; in contrast, this phenomenon is delayed at approximately −20°C for the dough in the binary H2O/EG system, which suggests that the addition of EG causes a decrease in the freezing point temperature of H2O. The recyclability of the EG-based antifreeze doughs was further quantified by dehydration–hydration cycling tests. After the dough was fully dehydrated at 40°C and then rehydrated by adding the same amount of water, the G′ of the dough was essentially restored to 80% of the original value (Fig. 2e), indicating that the dehydration–hydration process of the dough was reversible. Moreover, these cycles can be looped 10 times (see supplementary Fig. S1). Thus, EG-based antifreeze dough is a promising candidate for the fabrication of renewable, reconfigurable, antifreeze devices.

To validate the dough for applications such as renewable and reconfigurable skin-like ionotronics, it was connected to copper plates and assembled into a capacitive sensor by sandwiching the dough in VHB tape (3M 4905). To improve signal stability and prevent dough dehydration, the dough-based ionotronics were sealed between two layers of VHB tape (Fig. 3a). During stretching, the dough was compliant and remained in close contact with the VHB tape without slipping (see supplementary Fig. S2). It is possible to predict the capacitance changes in the parallel-plate configuration theoretically according to the relationship between the capacitance and area changes without considering the changes in permittivity and volume. As a strain sensor, the linearity of sensing is an extremely important index, as shown in Fig. 3b for the sensing characteristics of dough base sensors at 100% strain. It is stretched by a factor of λ at room temperature, and the capacitance C linearly scales as C = λC0. Both the conventional and EG-based doughs exhibited almost linear sensing and similar GF values (161% for EG-based dough, 160% for dough), indicating that the addition of EG did not adversely affect the room temperature sensitivity of the doughs. In contrast, at subzero temperatures, the conventional dough does not possess the sensing ability, while the addition of EG can cause the dough to possess low-temperature sensing ability; however, this ability is also affected by the content of EG (Fig. 3c and see supplementary Fig. S3), which is mainly manifested by the fact that both too much EG and too little EG cause the sensing performance to gradually deviate from linearity. This is mainly because EG is not a good solvent for NaCl, so too much EG leads to the precipitation of NaCl at low temperatures, thus reducing the electrochemical performance. On the other hand, the inability to form competitive hydrogen bonds with H2O when there is too little EG can lead to poor frost resistance in the dough and thus to a decrease in capacitance. Accordingly, future dough-based flexible strain sensing should use a H2O:EG ratio of 2:1. Furthermore, the dough-based flexible sensor exhibited a linear relationship not only during stretching but also during recovery and almost coincided with the stretching process, indicating that there was no hysteresis in the sensing process (Fig. 3d). The sensing stability is assessed by assessing the reliability of the sensor and guaranteeing its long-term stability. Supplementary Fig. S4 shows that highly smooth sensing signals can be output repeatedly for up to 100 cycles without visible baseline shifts. Subsequently, the response time and relaxation time of the dough-based sensor were evaluated using the approximate time difference method, and the results were only 232 ms and 192 ms, respectively (see supplementary Fig. S5). The synchronization of the output signal with the real-time strain is also highly important. Supplementary Fig. S6a and S6b depict the synchronization curve with the sensing signal at various stretching speeds. When the stretch rate is low (100 mm min−1), the sensing signal of the dough-based sensor is nearly coordinated with the real-time stretch curve; as the stretch rate increases (500 mm min−1), a slight hysteresis of the sensing signal occurs at approximately 0.3 s, which is due to the viscoelasticity of the stretched chain segment of the hydrogel during rebound. Thus, when the rebound time is wider than the relaxation time of the chain segment, no hysteresis occurs, whereas the opposite effect occurs dramatically. Generally, the sensing behaviour of ionic sensors is more susceptible to humidity and temperature. As shown in Fig. 3e, the dough-based sensor can maintain good capacitive stability (less than 2% fluctuation) at humidity levels ranging from 30% to 100%. Supplementary Fig. S7 also shows that this sensor ensures good sensing linearity at these humidity levels. In terms of temperature, it exhibits small capacitance fluctuations in the range of −20 to –70°C, similar to humidity (Fig. 3f). In addition, we evaluated the sensing performance of the dough-based sensor at −20°C. A stable step signal is generated with a quick response when the knuckle is bent, but the sensor takes a long time to recover from higher strains to the initial value (see supplementary Fig. S8). This is because the motility of the molecular chain segments is reduced at low temperatures and therefore takes longer to relax.33 These findings suggest that the platform can basically adapt to the needs of different seasons and is a candidate for renewability, reconfigurability, and anti-freezing to replace nondegradable, nonrecyclable components in current flexible and soft ionotronics and electronics (see supplementary Table S1).

Fig. 3
figure 3

(a) Digital photo of the dough-based sensors sealed in the VHB tapes and schematic design of dough-based sensors; (b) sensing characteristics of dough-based sensors at room temperature; (c) sensing characteristics of dough-based sensors before and after freezing treatment; (d) reversible capacitance–extensile strain curves of freeze-resistant dough-based flexible strain sensors. Capacitive stability of dough-based flexible strain sensors exposed to different (e) humidity (100% relative humidity) and (f) temperature (−20 to –70°C).

Full size image

Stable signal output and better environmental adaptability allow dough-based flexible sensors to be used for monitoring human movement. Figure 4a shows the relationship between finger bending and the capacitive signal. For every angle of finger bending, the dough sensor causes the capacitance to rise by one position due to the change in area, and it is clearly observed that the relationship with the capacitance signal remains highly linear within bending angles of 30°, 60°, and 90°. Furthermore, the capacitance signal correspondingly shows a better stepped signal with negligible noise during the step-by-step bend-holding process (Fig. 4b). To further evaluate the practicality of the dough-based sensors, a variety of tests on human motion were conducted to investigate the sensing ability of the sensors. Dough-based sensors are attached directly to different joints of the body to detect motion signals. Dough-based sensors attached to knuckles, elbows, and knees, as shown in Fig. 4c–e, accurately detect capacitive signals during the stretching and bending process of these joints. Dough-based sensors attached to the knuckles are able to continuously output a sharp capacitive signal five times, and the same results are observed for the elbows and knees. All these applications involving large-scale human motion as well as small strains show the great potential of dough-based sensors for applications in artificial intelligence and the Internet of Things technologies.

Fig. 4
figure 4

(a) Variable values of capacitance corresponding to fixed bending angles (0°, 30°, 60°, 90°); (b) real-time capacitance changes when the finger is bent at different angles; sensing performance of antifreeze dough-based flexible strain sensors for various human motions. A student’s capacitance changes of the antifreeze dough-based flexible strain sensor induced by (c) finger bending, (d) elbow bending, and (e) knee bending.

Full size image

Conclusion

In conclusion, to overcome the increasing electronic waste in the era of electronic information, in this work, dough-based flexible strain sensors, which are frost-resistant, mouldable and reconfigurable, were prepared from wheat flour using the H2O/EG binary system. The unique rheological properties of the dough were investigated to illustrate its advantages in terms of flexible electronics. Owing to its rheological properties, the flexible strain sensor can be shaped and reconfigured. The strain sensor is capable of detecting human joint motions through stable and reproducible capacitive signals, demonstrating its potential application value in human motion detection and human–computer interaction. In addition, good sensing performance was maintained at −20°C by adding EG and salt solution. This work illustrates the potential of sustainable, eco-friendly, and mechanically adaptable materials for Internet of Things technologies, which also provides insights for other green biomaterials, such as agar and carrageenan, to replace nondegradable and nonrenewable conductive components in bioelectronic devices.