Design and Evaluation of 3D-Printed 4 Structures Coated By CWPU/Graphene as Strain Sensor: Truss, Honeycomb, Chiral Truss, Re-entrant

A strain sensor characterized by elasticity has recently been studied in various ways to be applied to monitoring humans or robots. Here, 4 types of 3D-printed auxetic lattice structures using thermoplastic polyurethane (TPU) as raw material were characterized: truss and honeycomb with positive Poisson's ratio and chiral truss and re-entrant with negative Poisson's ratio. Each structure was fabricated as a exible and stable strain sensor by coating graphene through a dip-coating process. The fabricated auxetic structures have excellent strength, exibility, and electrical conductivity desirable for a strain sensor and detect a constant change in resistance at a given strain. The 3D-printed auxetic lattice 4 type structures coated with CWPU/Graphene suggest potential applications of multifunctional strain sensors under deformation.


Introduction
Strain sensors are used to detect deformation or structural changes occurring in the surrounding environment as well as internal activities in human bodies. A multitude of studies have been conducted to develop e cient strain sensors, and materials that respond to structural changes and exhibit such change have been mainly explored [1][2][3] . Strain sensors require great exibility and play an important role in a variety of applications such as electronic skin 4,5 and soft robots 6,7 .
3D printing is a method of manufacturing an object with a desired shape through the process of continuously depositing designed materials under computer control. It has the advantage of being able to easily create the shape of a complex model that overcomes the limitations of the existing cutting process method 8-10 . Ever-evolving new technologies are being reported, starting with the stereo lithography-based production of Charles Hull's solid imaging process technology, which has made the greatest contribution to the advancement of 3D printing 11 . Numerous studies are underway on 3D printers, including various prototypes [12][13][14] , output processes and product completeness [15][16][17][18] , as well as strain sensors 19 , and innovative developments are being made to apply 3D printers more e ciently and inexpensively in various elds of both industry and academia, such as engineering 20,21 , construction 22,23 , aviation 24,25 , fashion 26-28 , and biology eld [29][30][31] .
Among the versatile 3D printing materials mentioned above, thermoplastic polyurethane (TPU) is a very attractive material. TPU is a low-cost material, highly suitable for 3D printing 32 , has the advantages of both the rubber and rigid thermoplastic resin, so it can be easily plastically deformed under the in uence of heat, does not stick like rubber, and has high elasticity because it is manufactured with excellent adhesion. These outstanding properties have the potential to develop strain sensors that respond to stress and strain as well as low operating voltages and to develop multifunctional sensory materials.
Graphene, a two-dimensional hexagonal material that will be responsible for the electrical signal of the strain sensor, is useful for detecting tensile strain with high sensitivity in strain sensors that require exibility and is being intensively applied to monitoring human movement 33,34 . This is due to its exceptional conductivity and mechanical strength. However, in order for graphene to optimally detect electrical signals from sensors, it must compensate for its lack of exibility. Waterborne polyurethane (WPU) is one of the best materials for dispersing graphene. It is friendly to the human body and the environment in terms of odor and toxicity of organic solvents and has excellent viscosity control and lmforming properties 35 . Moreover, several studies on a WPU-based coating solution using graphene as a ller have been conducted in various elds. This is because, as a composite material, the exibility and elasticity of WPU and the durability of graphene are improved, and at the same time, graphene forms a network as a ller, enhancing electrical, mechanical, and thermal properties 36-38 .
Caster-oil-based waterborne polyurethane (CWPU) can synthesize WPU from renewable raw materials such as vegetable oil-based materials, which are non-edible, accordingly no need to compete with edible oils. A major challenge in the development of CWPU is increasing the use of biomass to replace fossil fuels due to economic and environmental concerns. However, the performance of CWPU is not inferior to that of solvent-borne polyurethane in terms of mechanical properties, exibility, and mobility, so its applications are diverse 39 .
In this study, TPUs of four different structures were printed through a fused deposition modeling (FDM) 3D printing method and used as strain sensors, and FDM is one of the safest, simple, and most widely used 3D printing approaches. First, a chiral truss (CT) and re-entrant (RE) structure were used as auxetic lattice materials with a negative Poisson's ratio, a new type of metamaterial that expands transversely in response to axial stretch. For comparison, a truss (TR) and honeycomb (HN) structure were analyzed as two types of regular structures with a positive Poisson's ratio. Also, in this study, a dip-coating method using graphene was used, which is one of the easy and e cient methods of e ciently imparting electrical conductivity to materials 40,41 . Therefore, all structures were designed to coat the surface by applying the dip-coating method with CWPU/Graphene solution. The four structures coated with up to 5 layers of graphene were analyzed as sensors that respond to strain centered on Poisson's ratio by elongation.

Materials
In order to 3D print the auxetic lattice structures, NinjaFlex (Ninjatek, Fenner Inc., USA and Shenzhen Esun Industrial Co., Ltd, China), a TPU-based lament, with a diameter of 1.75 mm and a hardness of 85A was used. A CFDM 3D printer (Blackbelt 3D B.B., Netherland), which uses a continuous fused deposition modeling, with a nozzle diameter of 0.6 mm was used.
The anionic castor oil-based waterborne polyurethane was prepared through the following steps: (1) raw material mixing step of adding isophorone diisocyanate (IPDI) dropwise to a mixture of castor oil and dimethylolbutanoic acid (DMBA) while continuously stirring, (2) synthesizing urethane step in which urethane synthesized while lowering the viscosity of the pre-polymer with acetone and, (3) neutralization step of the carboxylic group of DMBA with triethylamine, (4) forming water-borne polyurethane (WPU) particles step by dropping distilled water into prepolymer solution, (5) acetone removal step in the WPU aqueous solution. The content of natural substances in the prepared WPU is about 50%, the equivalent ratio of OH (castor oil) / NCO (IPDI) / OH (DMBA) is 1:2.2: 1.19, and the content of the hard segment of the WPU is 59.4%. The particle size is 180 ± 10 nm, and the zeta potential is -30 ± 2 mV.
Modelling of 3D printed pattern of auxetic cellular 4structure Table 1 shows 3D modeling and printed products for each pattern of 3D-printed auxetic lattice 4 type structures. To prepare the TR, HN, CT, and RE auxetic lattice materials, we designed a single repeating unit of 10 mm (length) × 10 mm (width) with a 2 mm line width (17 mm × 17 mm for the CT). The nal model with a size of 120 mm × 1200 mm was created by arranging 12 repeating units horizontally and 120 repeating units vertically. 3D modeling was performed with a thickness of 1 mm using the 123D Design program of Autodesk Co., Ltd, (Korea). During printing work, the nozzle temperature, printing angle, printing speed, and in ll density were set at 240 ℃, 15°, 15 mm/s, and 100%, respectively. It was transformed into a g-code le to be printed. The obtained 4 samples were named TR, HN, CT, and RE, and all the structures were cut to 50 mm × 50 mm after removing the edges to maintain the auxetic structure.
Preparation for coating of CWPU/Graphene water-based solution on 4-structure.
A coating solution for the 3D-printed auxetic lattice 4 type structures was prepared as follows. 8.0 wt% graphene was added to a 30.0 wt% CWPU solution dispersed in water. The prepared CWPU/Graphene water-based solution was stirred at least 24 hours, followed by 1 hour of sonication, before being used as the coating solution. We continued to stir the solution before use. The 4 structures printed by the 3D printer were dip-coated with the coating solution prepared above and dried in an oven at 40 ℃ for 1 hour before use. One hour of drying was considered the end of one cycle of coating, and the coating was repeated 5 times.

Characterization
The mechanical properties of 3D-printed auxetic lattice 4 type structures were characterized using a constant rate of extension type tensile testing machine (AGS-500D, Shimadzu, Japan). The distance between the clamps gripping the sample in the machine direction (MD) was set to be the length of the single repeating unit of each structure: 10 mm for the TR, HN, and RE, and 17 mm for the CT. The maximum elongation based on the gauge length was tested up to 60%, and the initial modulus, tensile strength, and toughness were analyzed.
The electrical properties of the coated auxetic lattice 4 type structures were characterized using a source meter (2450 Source Meter, Keithley Instruments). To investigate the e ciency of increasing electrical performance by coating with graphene solution and to explore the electrical properties applied in MD and counter direction (CD), the current-voltage curves of the structure for each coating cycle were measured in MD and CD and analyzed. The electrical conductivity was calculated using the following equation (1): where σ is the electrical conductivity, ρ is the resistivity, R is the resistance, A is the cross-sectional area of the sample, and L is the length of the sample. The surface area of each structure was calculated using Image J (software), and the apparent conductivity was obtained from the calculated surface area. The practical conductivity of the coating was calculated only by the actual increment of the coating.
The relative resistance change was calculated by grab the sample at the constant rate of extension machine as above mentioned, connecting electrodes to both ends, and applying 0.01V through the source meter. The strain was elongated by 1, 3, 5, 10, 15, 20, 25, and 30% against the grab distance, respectively.
The relative resistance change (R r ) was calculated using the following Equation (2): where R is the resistance during elongation and R 0 is the initial resistance. The image of each strain was taken with a camera (HDR-CX550, Sony, Japan) to measure the change in relative resistance. Then, the longitudinal and transverse extension points were identi ed to calculate the strains in each direction, and the Poisson's ratio was calculated using the following Equation (3): where ν is the Poisson's ratio, ε t is the transverse strain, and ε l is the longitudinal strain.

Results And Discussion
Morphology and weight and thickness increase rate Table 2 shows the images of the 3D-printed auxetic lattice 4 type structures taken in the dip-coating process where CWPU/Graphene water-based solution was used, and the coating was conducted up to 5 times. In all the structures, the color change was clearly observed as the coating cycle progressed, and, accordingly, the weight and thickness increased linearly. The dip-coating process using the graphene coating solution is e cient and has a great in uence on increasing electrical conductivity for electrical signal sensing 42 . In particular, the number of dip-coatings has a clear correlation with physical durability and electrical conductivity and varies from a few to as many as 20 times depending on the study 43,44 . In our previous study, we have demonstrated that 5 dip-coating using the WPU and graphene composite solution is appropriate and e cient and applied in the present study 36 . Figure 1 shows the change rate in weight and thickness of the 4 type structures when the CWPU/Graphene water-based solution was coated 5 times. The CWPU/Graphene coating layer increased in weight and thickness linearly in all structures. This indicates that the coating proceeded smoothly and that a coating layer is physically and gradually formed on the outer layer of the TPU after the CWPU/Graphene solution is dried in the water dispersion state. Particularly, RE has the highest initial weight since it has a denser repeating unit than other structures due to its compact structure. Therefore, the rate of increase in weight is measured to be the lowest.
Mechanical property Figure 2 and Table 3 compare the mechanical properties before and after the coating to analyze the mechanical strength and stretchability of 3D-printed auxetic lattice 4 type structures. Figure 2 (a-d) shows the tensile stress-strain curves before and after the coating when elongated to 60% for the 4 type structures. Overall, the mechanical strength appeared to increase as the coating cycle progressed. In general, although there is a difference depending on the content, coating single graphene or composite solution imparts an appropriate amount of mechanical strength to the applied matrix 45 . Also, as shown in Table 3, the coating increases the initial modulus at 1% strain and toughness at 60% strain, regardless of the structure. This shows that coating of auxetic TPU with CWPU/Graphene water-based solution 5 times increases all mechanical properties without degrading strength, resulting in stable coating e ciency.

Electrical property
In order to analyze the electrical properties of the 3D-printed auxetic lattice 4 type structures that serve as the strain sensors after coating, the current-voltage (IV) characteristics in the CD and MD were measured for each coating cycle as shown in Figure 3 and Figure 4. The sensing range of current and voltage for sensing under strain should conform to Ohm's law for reliable results of the sensor 46 . As can be seen from Figure 3(a-d), the current increases linearly with the applied voltage as the coating progress in all 4 type structures. In general, it was con rmed that at least fourth cycles of the coating were required for the electrical current to ow to all structures. This is presumably because a su cient network between graphene having electrical conductivity may not be formed until after two or three cycles of coatings. Also, this result is consistent with the time when the overall color turns gradually black, as shown in Table  2. In particular, all CT structures have relatively disadvantageous paths through which current can ow compared to other structures due to the inherent complexity of the structures. In Figure 4(a-d), the current owing through the MD of the 4 type structures was compared with the current owing through the CD. As a result, it was found that current ows smoothly in all structures regardless of direction. However, in the case of the RE structure, the slopes of the I-V curves in the MD and CD are different, unlike other structures, indicating that the pathway through which the current ows is related to the structural characteristics of each structure. Unlike TR, CT, and HN, which have little or no structural difference in their pathways in the MD and CD, RE has a huge structural difference, implying that such structural characteristics should be considered when using RE as a strain sensor. Table 3 summarizes the apparent and practical conductivity according to the actual coating thickness by calculating the electrical conductivity after obtaining the surface area of the 3D-printed auxetic lattice 4 type structures using Image J. The apparent conductivity of the uncoated structures was measured within a range of ~10 −11 -10 −12 S/cm digits of the nonconductor level. The electrical conductivity increased with each cycle of coating, and after 5 cycles of the coating, each structure exhibited an electrical conductivity ranging from ~9.08 × 10 −06 to ~1.96 × 10 −03 S/cm. In addition, the electrical conductivity with coating thickness actually increases signi cantly from 3.23×10 −04 to 1.96×10 −03 S/cm. The electrical conductivity of each sample coated 5 times is su cient to detect currents as a strain sensor, even including voids created due to the structural properties of the 4 type structures.

Performance as strain sensor and Poisson's ratio
For each 3D-printed auxetic lattice 4 type structure, Poisson's ratio was calculated by analyzing the images depicted under strain as a deformation property, and the rate of change of relative resistance during elongation was measured, as a strain sensor. Table 4 Poisson's ratio, and they expanded transversally as they are axially stretched 47 . The auxetic structure is one of the metamaterials that obtain properties from the structure 48 , and all 4 type structures printed as designed exhibited a Poisson's ratios suitable for their characteristics. Figure 5 are graphs showing the Poisson's ratio extracted from each image by plotting it for each structure. The 4 type structures maintained either positive or negative Poisson's ratio until the 60% of strain, as shown in Figure 5(a-d). At large levels of uniaxial stretching, the auxetic structures experience nonlinear nite deformations aligned closely to the loading direction by huge loadings 49 . The TPU-based 4 type structures we designed were shown to exhibit stable structures at strains up to 60%.
Finally, the performance as strain sensors was compared with the TR and HN with positive Poisson's ratio and CT and RE with negative Poison's ratio, as shown in Figure 6. For the measurement, the sensor was elongated for 20 seconds for a predetermined strain range and held for 10 seconds before it returned to its original position. After mounting the electrodes in the CD on each structure, the strain signal was connected to the source meter, and the change in the resistance was outputted to the computer through the signal.
Each sensor was reliably sensitive to a displacement of 1% for all structures, and the detection of this initial resistance change is the result of an e cient CWPU/Graphene water-based coating solution process. However, the limit of the graphene coating of TPU at 60% strain was different for each structure. TR and HN with positive Poisson's ratio detected changes in the resistance up to 20% and 30% strains, which are related to the durability of the coated CWPU and the magnitude of reversible deformation with each structural change. In fact, as shown in Table 4 and 5, TR and HN show these results due to damage to the coating at the intersection of the structures at the elongation of 20% and 30%. CT and RE with a negative Poisson's ratio detected changes in the resistance up to 15% and 25% strains, and also microscopic damages at the intersection of these structures were con rmed as shown in Table 6 and, 7. In addition, CT seems to have a limit of 15% due to the greater reversible deformation during strain, further reinforcing this result. However, all structures transmitted signals very stably for a period of 10 seconds to respond. In applications, the sensors can reliably identify a constant strain by the change in resistance 50 . Although we only showed the design of 4 models for the strain sensor, it was shown that the 3D printed TPU materials stably coated with the CWPU/Graphene water-based solution could act as a sensor. Our results provide new opportunities for TPU-based auxetic lattice structures to be used in various electronic products using the dip-coating method.

Conclusion
In conclusion, we 3D-printed TPU through FDM printing to generate two of the 4 types with positive and negative Poisson's ratio respectively and coated them with graphene to evaluate them as strain sensors.
After coating the surface of the printed TPU with the CWPU/Graphene water-based solution more than 5 times and evaporating moisture, the graphene network was combined to form a stretchable material and a conductive coating lm. In addition, the mechanical properties after coating 5 times improved the strength of all the structures. TR and HN with positive Poisson's ratio have resistance sensing ranges of 20% and 30%, respectively, while CT and RE with negative Poisson's ratio have 15% and 25%, respectively, and act as sensors. In particular, all of them can be used as exible wearable sensors that identify various strain motions through linear and stable resistance changes over the entire range of each strain.