Melamine sponge skeleton loaded organic conductors for mechanical sensors with high sensitivity and high resolution

In recent years, due to the development of flexible electronics, flexible sensors have been widely concerned and applied in intelligent robots, brain-computer interfaces, and wearable electronic devices. In this paper, we propose a low-cost and high-efficiency sensor component preparation method. The sensor component tetrathiafulvalene-tetracyanoquinodimethane/melamine sponge (TTMS) takes a melamine sponge as a flexible substrate. And the sponge is metallized with the tetrathiafulvalene-tetracyanoquinodimethane (TTF-TCNQ) which is an organic conducting molecule to construct a conductive pathway. The physical load approach is used to ensure the advantages of low cost and efficient manufacturing. TTMS can withstand 8000 compression cycles which exhibits its good mechanical stability. And 1000 cycles of cyclic voltammetry scanning proved it also had good electrical stability. TTMS can distinguish pressure changes of 100 Pa and respond quickly to pressure application and release. The TTMS can be assembled to form an array of sensors that can distinguish the position and intensity of pressure. Therefore, the excellent performance of the sensor is expected to promote the commercial application of the piezoresistive sensor.


Introduction
Nowadays, flexible electronic materials are developing very rapidly, such as flexible display materials [1], flexible electromagnetic shielding materials [2], flexible batteries [3,4], and flexible energy storage materials [5,6]. They are becoming more and more important and common in our lives. We focused on flexible pressure sensors in flexible wearable electronics, whose applications include electronic skins [7], brain-computer interfaces [8,9], and medical monitoring equipment [10][11][12][13][14][15]. According to the working principle of the sensor, pressure sensors could be divided into piezoresistive [16][17][18][19][20][21], capacitive [22,23], and piezoelectric [24][25][26][27][28][29]. Capacitive pressure sensors had high sensitivity and were mostly used in electronic skin and brain-computer interfaces, but their high sensitivity also made them vulnerable to external interference [30]. Piezoelectric sensors were suitable for dynamic measurements and could respond quickly to pressure changes, but not for static tests [31]. The piezoresistive sensor had a simple structure, simple preparation, and little influence by external temperature and humidity, and had a certain scale production and application potential [32]. Flexible piezoresistive sensors were prepared from a variety of materials, including metal nanowires [33,34], carbonbased conductive materials (carbon nanotubes, graphene, etc.) [35][36][37], conductive network constructed by synergistic interaction of multiple metal oxide heterojunctions in the flexible substrate [38], and self-supporting conductive system constructed by conductive fiber structure [39]. For example, Mu et al. [35] proposed an e-skin preparation scheme for anchoring carbon nanotubes (CNTs)/graphene oxide (GO) hybrid 3D conductive networks on porous polydimethylsiloxane (PDMS) layers. The sensor showed excellent sensitivity (gauge factor of 2.26 under a pressure 1 3 loading of 1 kPa) and a highly reproducible response within 5000 cycles of tension, bending, and shear. It not only detected wrist pulses and distinguishes between different surface roughness, but also responded significantly to the slightest tick of a feather (~ 20 mg) and could also be used to monitor human respiration in real time. Cheng et al. [40] fabricated a highly sensitive MXene-based piezoresistive sensor inspired by bioinspired micro-spinous microstructures, which can effectively increase the sensitivity of the pressure sensor and the limit of detectable fine pressure. The obtained piezoresistive sensor showed high sensitivity (151.4 kPa −1 ), relatively short response time (< 130 ms), subtle pressure detection limit of 4.4 Pa, and excellent cycle stability over 10,000 cycles. The sensor showed great performance in real-time remoted monitoring and quantitative detection of pressure distribution. However, structures of heterojunctions and elf-supporting were relatively complex and had higher requirements for experimental conditions [41]. Some materials also had a high cost. The preparation process often used high-temperature annealing, water bath reaction, etc. Metal nanoparticles faced the problem of reduced electrical conductivity caused by oxidation [42]. Therefore, it was very important to develop a simple method to prepare a stable sensor.
The organic conductive material may be an ideal sensing material [43][44][45][46]. In 1954, Akamatu et al. discovered the conductive charge transfer complex salts of perylene bromide complexes [47]. Since then, a variety of organic conductive materials have been developed, including a variety of small molecules, charge transfer complexes, oligomers, and conductive polymers [48]. Their conductive mechanism was mainly due to the interaction between adjacent molecules and the intrinsic electronic structure of the extended molecule. And they were expected to be applied in practice because of their flexibility, easy modification, flexible processing, and high universality. Tetrathiafulvalene-tetracyanoquinodimethane (TTF-TCNQ) was a highly conductive charge transfer complex (CTC) with metal-like conductivity over a wide temperature range (350-59 K) [49][50][51]. It was first reported in 1973 by Bloch et al. [52]. Their chemical formula was shown in Fig S1. Some studies have shown that TTF is the electron donor and TCNQ was the electron acceptor by density functional theory (DFT) [53]. The main skeleton of TTF contains S atoms, which contributed to solid intermolecular S•••S interactions. These S•••S interactions and π-π and •CH-π interactions affected the stacking structure of TTF molecules in the crystal and further affect the electrical conductivity of the material [48]. The high conductivity of TTF-TCNQ CTC was attributed to a "herring bone"-type crystal structure formed by the flat TTF and TCNQ, in which orbitals on adjacent molecules overlapped to form continuous one-dimensional bands [50].
Herein, we proposed an efficient technical method to fabricate pressure sensors using organic conducting molecule TTF-TCNQ. Polyvinyl butyral (PVB) was an adhesive; TTF-TCNQ CTC was physically combined with skeletons of melamine sponge (MS) to form pressure-sensitive sensing element TTMS (Fig. 1). The TTMS had a high resolution of 100 Pa, a rapid response capability of 260 ms, and an ultra-high sensitivity of 90.7 kPa −1 in the range of 20 kPa to 104 kPa. These excellent electrical properties resulted from a dramatic increase in the density of conducting molecules in the compressed state of the sensor. Moreover, a pressure sensor could be obtained by array arrangement of multiple TTMS, which could reflect the strength and position of the force. Its accuracy was closely related to array size, which could be adjusted freely. Ecoflex was used to replace threedimensional-printed polylactic acid (PLA) substrate, and the fully flexible sensor was expected to be applied to intelligent robot behavior sensing and other aspects.

Preparation of TTF-TCNQ
TTF and TCNQ are dispersed in an alcohol solution; then, the liquid is transferred to a mortar and ground to a dry powder. The grinding process can be repeated to fully grow organic molecular conductor crystals. For a more detailed preparation process, see reference [54].

Preparation of TTMS
The sponges were cut into 0.7 cm × 0.7 cm × 0.5 cm squares by a mold. 0.5 wt.% PVB/ethanol solution was prepared by dissolving 0.5 g PVB in 99.5 g ethanol. Then, 1.8 g TTF-TCNQ was added to 30 g 0.5 wt.% PVB/ethanol solution. Ultrasonic dispersing for 30 s, followed by rapid stirring for 5 min. Finally, the sponge was immersed in TTF-TCNQ/ PVB/ethanol solution and kept in a vacuum at 1000 Pa for 3 min. Finally, TTMS can be obtained after natural drying [55].

Preparation of sensor
The sensor was 3D printed by using polylactic acid (PLA). The sensor slot was designed to be 0.7 cm × 0.7 cm × 0.2 cm. The first and third layers of the sensor were copper tape perpendicular to each other, which was used to construct a conductive network that uniquely identified the location of the signal source. TTMS were then placed into each sensing unit. TTMS could be connected to copper tape more closely through conductive silver paste. Ecoflex was applied to fabricate a fully flexible sensor with a conductive channel width of 0.5 cm. The Ecoflex substrate differs from the PLA substrate in that the Ecoflex was divided into upper and lower parts.

Characterization
X-ray powder diffraction data were collected using an XRD (D/max 2500, Rigaku, Japan) with Cu Kα radiation (λ = 1.54178 Å). Micromorphological images were recorded using a field emission scanning electron microscope (FE-SEM, LEO-1530, Zeiss, Germany) with an EDX attachment module. An X-ray photoelectron spectrometer (ESFalab 250Xi, Thermo Fisher, America) equipped with an Al Kα radiation source (1487.6 eV) and a hemispherical analyzer with a pass energy of 30.00 eV was employed to obtain surface element information. The thermogravimetric analysis (TGA) and differential thermal gravity (DTG) analysis were performed using a thermogravimetric analyzer (STA 449 F3, Jupiter, Germany). Fourier transform infrared spectroscopy (FT-IR, VERTEX 70 V).

Mechanical and electrical test
The real-time resistance and current of sensors upon various deformations were obtained by a combined instrument consisting of a computer-controlled electrochemical workstation (CHI 660E, CH Instrument, China) and a universal mechanical tester (Zwicki-Z1.0, ZwickRoell GMbH&Co. KG, Germany) with a double silver electrode system. The relative current change was defined as [56,57]: The conductivity calculation formula was: where σ (S/m) is the conductivity, ρ (Ω/m) is the resistivity, L (m) is the width of the PVA/AgNPs hydrogel sensor, A (m 2 ) is the cross-sectional area of the sensor, and U (V) and I (A) are the applied voltage and the corresponding current which was available through the electrochemical workstation, respectively.
And the formula for calculating GF was [58]: And the formula for calculating sensitivity was [59]: where k (kPa −1 ) is the sensitivity, and P (kPa) is the pressure.

Results
We used Fourier transform infrared spectroscopy (FT-IR) to analyze the difference of chemical bond information between MS and TTMS (Fig. 2a). The enhancement of the peak at 900-1300 cm −1 was due to the joint action of C-S, C = S, and benzene ring, and the newly formed peak at 2200 cm −1 was due to the introduction of C≡N in TCNQ. The increase of peak at 2850 cm −1 and 2920 cm −1 was due to the increase of C-H, while the newly formed peak at 3070 cm −1 is caused by C = C vibration. Raman spectra (Fig. 2b) showed that the characteristic principal vibration modes at 980 cm −1 , 1202 cm −1 (C = CH bending), and 1603 cm −1 (C = C ring stretching) confirmed the presence of the TTF and TCNQ phases [60]. The composition differences between MS and TTMS were analyzed by thermogravimetric analysis (TGA) and derivative thermogravimetry (DTG) in the N 2 atmosphere (Figs. 2c and S1), which showed that TTMS prepared using 6% TTF-TCNQ concentration contained approximately 45% TTF-TCNQ by weight. TTF-TCNQ began to decompose at about 185 ℃ and gradually completes decomposition at 400 ℃. The curve decline after 400 ℃ was due to the decomposition of MS at high temperature until the decomposition is completed at 700 ℃. At 900 ℃, the residual matter after TTMS decomposition was about 3.8%, which may be formed under high temperature. The blue curve (6% compressed) in the figure reflected the material after 10000 compression tests. It could be seen that the physical composite mode using PVB as the adhesive was very stable, and the content of TTF-TCNQ was the same as that without compression. XRD curve could obviously show that TTF-TCNQ has been attached to MS in TTMS (Fig. 2d). We marked multiple crystal planes of TTF-TCNQ, which could be well corresponding to literature [61]. Then, the C, N, O and S elements in TTMS were analyzed by X-ray photoelectron spectroscopy (XPS) (Figs. 2e, f and S1b-d). There were two binding energy peaks of N 1 s [62], 399.1 eV and 397.5 eV, corresponded to N≡C and N≡C − in TCNQ, respectively. The binding energy peak of S 2p was complicated, including 2p satellite peak, 2p 1/2 and 2p 3/2 [63][64][65][66][67]. The S element came from TTF, and its electron energy was affected by TCNQ. The peak at 168.8 eV was not classified into the above three electron states but was commonly found in metal sulfates. In both metal sulfates and TTF-TCNQ, S element acts as an electron donor, thus explained the origin of the peak at 168.8 eV. In the binding energy peak of C element C 1 s, 288.7 eV corresponded to C-S in TTF, 286.1 eV corresponded to C≡N in TCNQ and C-O-C in PVB, and 284.8 eV corresponded to carbon in benzene ring and C-C. The peak of C1s at 283.3 eV was commonly seen in the carbon and metal-binding material, so it may be formed by Raman spectra. c TGA curves. d XRD curves. e, f XPS spectra. g, h, i SEM images and mapping of S element the influence of the C element in TTF or TCNQ. This unique peak position in C and S elements also supported the charge transfer mentioned above. All of the O element came from PVB, the binding energy peak of the O element correspond to C-O-C, C-O, and C = O. We observed the microstructure of TTMS with a scanning electron microscope (Fig. 1g), and it could be seen that TTF-TCNQ CTC was loaded on the MS skeleton, and there would be more accumulation of TTF-TCNQ at the skeleton joints (red arrow). Then, we scanned the distribution of S elements on the surface through mapping (Fig. 1h, i), and it could be seen that the distribution of S elements was completely consistent with TTF-TCNQ.
Next, we studied the mechanical and electrical properties of the sensing element. We measured the modulus of TTMS prepared with three different TTF-TCNQ usage amounts (Fig. 3a). It could be seen that the increase of TTF-TCNQ usage would increase the modulus of TTMS, but also increased its conductivity. Thus, we determined the usage of 6%TTF-TCNQ. The stress-strain curve could also reflect the influence of usage on mechanical properties of TTMS (Fig. 3b), and the increase of stress of this material would also change significantly with the increase of strain. Subsequently, we tested the response of TTMS to current under different pressure states (Fig. 3c). And optical photographs of 0-70 kPa compression are shown in Fig. S2. The higher the pressure, the higher the current represents, the lower the resistance. Therefore, a piezoresistive sensor could be constructed based on this sensing mechanism. We further  (Fig. 3d), which could maintain the stability of electrical signals under 1000 cycles of cyclic voltammetry (CV). In addition, TTMS could keep a stable rate of current change even after 10000 cycles of compression (Fig. 3e). We also gave the detail diagram of the current change rate in the first, middle and last three periods in the compression process (Fig. S1e). The periods were 5th-9th, 3998th-4002th and 7903th-7908th, respectively. After 8000 compression cycles, the current change rate decreased by about 5%.
Then, we measured several main evaluation indexes of the sensor, including response time, resolution, gauge factor and sensitivity. TTMS could respond quickly to stress application and release (Fig. 3f), with response times of 260 ms and 160 ms respectively. There was no obvious limit on the left end of the stress holding platform (red arrow). Therefore, we took the time from the stress applied to 90% of the maximum stress as the response time of TTMS to the stress application. This problem did not exist in the stress release stage. The sensor had a very high GF value (Fig. 3g) and sensitivity (Fig. 3h). In 65-70 kPa, GF was 131704.6. The sensitivity of 5-15 kPa was 8.5 kPa −1 , 15-20 kPa was 37.0 kPa −1 , and 20-105 kPa was 90.2 kPa −1 . This excellent performance came from the state change of TTF-TCNQ. In the beginning, TTF-TCNQ existed in the form of aggregates, which were bonded to the sponge skeleton by PVB, and the contact of aggregates forms a conductive path similar to the sponge skeleton. When TTMS was compressed, the density of TTF-TCNQ increased. At this time, there was not only a conductive path on the horizontal plane, but also a more complex conductive path on the vertical direction due to the pressure promoted more TTF-TCNQ contact. Therefore, the electrical conductivity of TTMS was greatly increased under compression, further affecting GF value and sensitivity. In Fig. 4 The sensor responded to different pressure and different shapes. a 30% compression. b 40% compression. c 50% compression. d Box. e Rectangle. f Circle this paper, the rate of change of current rather than the rate of change of resistance was chosen to calculate GF value and sensitivity. The conductivity of TTMS ranged from 2.3 × 10 −4 S/m initially to 1.79 S/m at 70% compression (Fig. S1f). This value increased by 5 orders of magnitude, and the significant increase in conductivity was due to the increase of conductive paths under compression. We tested current conditions from 0.1 kPa to 0.5 kPa pressures, and TTMS was able to resolve a minimum stress difference of 0.1 kPa (Fig. 3i). The higher the resolution of the sensor, the more accurate the signal transmitted [68].
Finally, several sensor elements TTMS were used to construct a 5 × 5 array sensor (Fig. S3a). The sensor substrate was 3D printed, and the conductive channels were pre-designed on the substrate and filled with conductive copper tape to construct the conductive channels. To achieve complete flexibility of the sensor, we used the outer surface of the sensor prepared by Ecoflex and also used conductive copper tape to construct the conductive path ( Fig. S3b-d).
We tested the response of the sensor to different normal force (Fig. 4a-c). The results showed that the sensor could clearly distinguish different normal force, and the approximate range of normal force could be obtained from calorific value. Then, we also tested the response of the sensor under different shapes of pressure (Fig. 4d-f). The shapes of pressure were box, rectangle and circle. The electrical signal data was presented in the form of a heat map. It could be seen that the sensor could reflect the position of pressure applied under the pressure of these three shapes. In addition, we could determine the corresponding relationship between a certain calorific value and pressure, which was through the relationship between the pressure measured in the previous experiment and the current change rate. The calculation of calorific value is the same as the current change rate. But at the same time, we could also see that the sensors did not accurately represent the original shape of the pressure. Such defects resulted from the low precision caused by the small size of the sensor. Due to the small number of pixels in the sensor, part of the information would be lost in the feedback. This defect could be remedied by adjusting the size of the sensor. In this paper, the fast and efficient preparation method and simple sensor assembly method could greatly reduce the difficulty of expanding the sensor scale.

Conclusion
Due to the high conductivity, stability, and flexibility of organic molecular conductors, we constructed a conductive system by using TTF-TCNQ. And it was applied to mechanical sensing with a flexible substrate. The electrical and mechanical stability of the sensor TTMS benefited from the double guarantee of TTF-TCNQ and MS stability. The high conductivity of 1.79 S/m at the maximum compression was derived from the diversification of conductive paths, which also made the sensor have a high sensitivity of 90.2 kPa −1 . The response time of TTMS was 260 ms and 160 ms when the force was applied and released, which ensured that the sensor could respond quickly to stress changes. At the same time, TTMS's high resolution of 100 Pa pressure greatly improved the detection range of the sensor. In general, the performance of the flexible pressure sensor prepared by TTF-TCNQ was excellent, and it had application potential in artificial intelligence, soft robot, and other aspects. The simple and efficient preparation method also increased the possibility of the sensor being widely used.
Author contribution Wu Yufeng was responsible for all the experimental and paper writing parts. Wu Jianbo and Lin Yan provided the characterization test equipment. He Xian helped to complete the characterization of SEM and XRD. Liu Junchen and Pan Xiaolong helped to revise the writing of the paper. Lei Ming and Bi Ke provided experimental ideas, experimental direction, and financial support.

Data availability
The data sharing is not applicable to this article.

Conflict of interest
The authors declare no competing interests.