Fabrication and Piezoresistive/Piezoelectric Sensing Characteristics of Carbon Nanotube/PVA/Nano-ZnO Flexible Composite

Flexible sensors with a high sensitivity and wide-frequency response are essential for structural health monitoring (SHM) while they are attached. Here, carbon nanotube (CNT) films doped with various PVA fractions (CNT/PVA) and ZnO nanowires (nano-ZnO) on zinc sheets were first fabricated by functionalized self-assembly and hydrothermal synthesis processes. A CNT/PVA/nano-ZnO flexible composite (CNT/PVA/ZnO) sandwiched with a zinc wafer was then prepared by the spin-coating method. The piezoresistive and/or piezoelectric capabilities of the CNT/PVA/ZnO composite were comprehensively investigated under cyclic bending and impact loading after it was firmly adhered to a substrate (polypropylene sheet or mortar plate). The results show that the piezoresistive sensitivity and linear stability of the CNT/PVA films doped with 20%, 50%, and 100% PVA during bending are 5.47%/mm, 11.082%/mm, and 11.95%/mm and 2.3%, 3.42%, and 4.78%, respectively. The piezoelectric sensitivity, linear stability, and response accuracy of the CNT/PVA/ZnO composite under impulse loading are 4.87 mV/lbf, 3.42%, and 1.496 ms, respectively. These merits support the use of CNT/PVA/ZnO as a piezoresistive and/or piezoelectric compound sensor to monitor the static/dynamic loads on concrete structures while it is attached.


Scientific RepoRtS |
(2020) 10:8895 | https://doi.org/10.1038/s41598-020-65771-x www.nature.com/scientificreports www.nature.com/scientificreports/ Fabrication of the self-assembled CNT/PVA film. The processes for the positive and negative functionalization of the CNTs (CNT-COOH and CNT-NH 2 ) and self-assembled CNT films were similar to those in ref. 32 . First, 1 mg/ml CNT-NH 2 and 1 mg/ml CNT-COOH suspensions were ultrasonically treated for 2 h at 40 kHz (KH5500H type, Changzhou, China). After 24 h incubation, the above CNT-NH 2 and CNT-COOH suspensions were doped with PVA solutions with mass ratios of the PVA solution to the CNT-NH 2 or CNT-COOH suspensions of 0.2:1, 0.5:1, and 1:1, and then the suspensions were bath treated at 80 °C for 2 h. The CNT-COOH/PVA suspension was first sprayed on the polished surface of the opposite side of the zinc substrate by a sprayer (WA-101 type, Hangzhou, China). Then, the suspension was deposited by a spin-coater (KW-4B type, Beijing SETCAS Electronic Co. Ltd., China) with low and high speeds of 500 rpm and 1500 rpm, respectively, for 60 s. After drying at 80 °C for 15 min in a vacuum oven (DZF-6210 type, Shanghai, China), a CNT-NH 2 /PVA suspension was sprayed on the first CNT-COOH/PVA layer and spin-coated, and one CNT-COOH/CNT-NH 2 /PVA double layer was finally formed after drying at 80 °C for another 15 min. The above processes were repeated 6 times, and the CNT/PVA film with six double layers was achieved after drying at 80 °C for 30 min. The procedure for the preparation of the CNT-COOH, CNT-NH 2 and self-assembled CNT/PVA film is schematically demonstrated in Fig. 2.
Fabrication of CNT/PVA/ZnO flexible sandwich composite. First, the nano-ZnO array and CNT/ PVA film layer in the CNT/PVA/ZnO flexible sandwich composite were sequentially synthesized in accordance with the abovementioned methods in Sections 2.2 and 2.3, respectively, where a schematic of the fabrication process is shown in Fig. 3a. Second, the electrodes and wires were connected surfaces of the CNT/PVA/ZnO flexible composite. Third, adhesive tape was precisely placed on the edge of the zinc substrate so as to avoid a possible  www.nature.com/scientificreports www.nature.com/scientificreports/ short circuit between the electrodes and the substrate and the fixed wire. A schematic of the configuration is shown in Fig. 3b.
Characterization methods. Scanning electron microscopy (SEM, S3500N type, Hitachi Corp., Japan) was used to observe the morphology and microstructures of the CNT/PVA film, nano-ZnO array, and CNT/PVA/ ZnO flexible composite. A quasi-static d 33 /d 31 measuring instrument (ZJ-6A type, Institute of Acoustics, Chinese Academy of Sciences, China) was employed to measure the d 33 of the nano-ZnO sheet after cutting it into 10 mm × 10 mm × 0.4 mm samples where the size factor is 1.
The piezoresistive and piezoelectric sensing characteristics of the CNT/PVA/ZnO flexible composite were tested by cyclic three-point bending loads and impulse load testing methods, respectively, and the details are demonstrated in supplementary information.
As revealed in Figure S2 from supplementary information, the CNT/PVA/ZnO layer specimen was attached onto a cantilever. Different signal channels of DASP V11 data acquisition system were separately collected the signals from the nano-ZnO and the CNT/PVA layers, with CNT/PVA, nano-ZnO layer respectively acting as a piezoresistive resistor, and a piezoelectric capacitor. The acceleration signals from the nano-ZnO thin layer mainly reflected the dynamic or impact loads, while the voltage signals from CNT/PVA layer could sense the static and quasi-static loads aiding with 2 V DC power supply. It benefits for synchronous and simultaneous detections to the complicate loads in SHM, effectively avoids drawbacks while using traditional either piezoresistive or piezoelectric sensor.

Results and discussion
Surface morphology and SEM analysis. Figure 4 presents the surface appearance of the nano-ZnO grown on a zinc substrate in various zinc ion molar concentrations (w zn2+ ). When the w zn2+ is 10 −5 mol/L, the surface appears bright white and there is an uncovered area of substrate in the left corner (circled in Fig. 4a)), which implies that the nano-ZnO layer is extremely thin, and as w zn2+ increases to 2 × 10 −5 mol/L, the surface color gradually changes to yellowish gray. When w zn2+ reaches 10 × 10 −5 mol/L, the surface is no longer smooth, has a grainy appearance (circle in Fig. 4c)), and the color changes to pale yellow. Figure 5 shows the overall and side views of the CNT/PVA/ZnO flexible composite. The images show an obvious interface between both film layers, and the total width of the flexible composite along with the zinc substrate is approximately 200-300 μm, where the thickness of the CNT/PVA film layer, nano-ZnO layer, and substrate is approximately 30-40 μm, 10-20 μm, and 200 μm, respectively. The CNT/PVA layer on the left side of the zinc substrate has a smooth facade, while the nano-ZnO layer on the right side is uneven, and the different  www.nature.com/scientificreports www.nature.com/scientificreports/ morphologies show a good interface contact. The CNT/PVA film surface is smooth with a few light spots, and its appearance is attributed to a uniform dispersion of the CNTs, as shown in Fig. 5b.
The particle size and crystal arrangement of the nano-ZnO determines the apparent color, as revealed in Fig. 5d. A large number of nano-ZnO floccules is deposited; the nano-ZnO crystal arrangement is disorganized, and the mean particle size is over 1 μm, which means that this type of nano-ZnO experiences optical absorption readily compared with that of the nano-ZnO array with a particle size below 100 nm (Fig. 5e). It can be concluded that the larger the particle size is, the more yellow the material becomes as the accumulation of the nano-ZnO increases. Simultaneously, the electrical signal of the corresponding nano-ZnO film is not obvious when subjected to external pressure due to this kind of disordered aggregation and agglomeration of the nano-ZnO (Fig. 5e).
As demonstrated in Fig. 5c, a homogeneous self-assembled CNT system with a high dose and good dispersion can be achieved with a uniform distribution of functionalized CNT-COOH and CNT-NH 2 even though a minor amount of accumulation and agglomeration still exists. As shown in Fig. 5d, the patterns of the nano-ZnO array are regular hexagonal pyramids and hexagonal prisms 33,34 with lengths and diameters less than 1 μm and 0.1 μm, respectively. Notably, a minor amount of flocculent aggregates grew on the nano-ZnO array owing to the aggregation of ZnO colloids, as shown in Fig. 5e 35 . In comparison, Fig. 5f presents a relatively uniform nano-ZnO array whose growth direction is vertical to the zinc substrate. Indeed, the preformed seed layer provides crystal nuclei for the formation of ZnO crystals, inducing the ZnO to grow vertically from the seed layer, thereby forming a crystalline nano-ZnO array, as shown in Fig. 5f. However, ZnO, which is far from the seed layer in the dispersed colloidal system, spontaneously forms a unit cell until it develops into large crystalline grains, causing nano-ZnO deposition on the seed layer by overcoming the buoyancy of the liquid. The neat arrangement and organized array formation allow the nano-ZnO arrays to produce a uniform charge distribution, and the electrical signal increases when subjected to a dynamic load.  Figure 6 shows the relationship between ΔR and the cyclic displacement of the CNT/PVA/ZnO composite.

Piezoresistivity of the CNT/PVA/ZnO flexible composite. Cyclic three-point bending test.
As shown in Fig. 6, at the mid-span on the tensile side, the dose of PVA can significantly affect R and ΔR, and the ΔR/R 0 peak of the CNT/PVA/ZnO film is approximately 29%, 54%, and 62% when m ( documented that his self-assembled CNT films had a piezoresistive sensitivity ranging from 5%/mm~20%/mm 33 ; however, the fluctuation of the piezoresistive sensitivity of his sensor was higher than ours, which was a drawback for its potential SHM applications. The R and ΔR of the CNT/PVA component of the CNT/PVA/ZnO film both increase as the tension increases because the connecting pathways among the conductive but well-distributed CNTs are accordingly decreased or the barrier widths between the CNTs are increased, which dramatically reduces the current densities of the electrons inside the CNTs; therefore, ΔR has the highest value when the displacement (tension) reaches its highest value. Conversely, the connecting pathways among the CNTs are increased, or the barrier widths between CNTs are decreased on the compression side, and ΔR accordingly has the lowest value when the displacement (compression) reaches its highest value 36,37 .
where ε = ΔL/L 0 , ΔL is the change in length of the film, L 0 is the initial length of the film, and ΔR/R 0 is the resistance change rate vs. initial R (R 0 ). As revealed in Table 1, the S e of each sample in the same group is close and has a low deviation, which indicates that each group of CNT/PVA/ZnO flexible composites possesses excellent repeatability. The mean S e of the CNT/ PVA/ZnO flexible composite is 5.47%/mm, 11.08%/mm, and 11.95%/mm when m(CNT):m(PVA) is 0.2, 0.5, and 1, respectively, under tension, and the value of the mean S e is 5.607%/mm, 9.74%/mm, and 9.78%/mm under compression, respectively.     Table 2    The repeatability (r e ) of the CNT/PVA/ZnO composites under cyclic loading can be calculated by the following formula 38,39 : where ∆max1 and (ΔR/R 0 ) F•S represent the maximum output nonrepetitive error and full-scale output of ΔR/ R 0 , respectively. As shown in Table 4, the r e values of the CNT/PVA/ZnO composite with m(CNT):m(PVA) values of 0.2, 0.5, and 1 are 3.0%, 2.36% and 1.27% on the tension side, respectively, whereas they are 1.86%, 2.27%, and 2.59% on the compression side, respectively. All r e s are below 5%, which indicates good sensing stability for the flexible composite [40][41][42][43] . The data show that the linearity index between individuals is not very volatile and has a high similarity.
As revealed in Table 5, excellent sensing parameters and stable linear responses render our CNT/PVA/ZnO composite feasible as a stress/strain sensor. An excessively high sensitivity results in failure during large deformation of the structure, and a sensitivity that is too low greatly increases the monitoring difficulty of the system 44 . The CNT/PVA/ZnO composite in our study and ref. 43 both show high sensitivity, but the linearity and repeatability in this study are superior to that in the cited reference.
piezoelectric/piezoresistive properties of cnt/pVA/Zno sheets under impulse loading. Figure 9 shows the vibration history after 12 impulses of the hammer within 3 s on the CNT/PVA/ZnO sheet. First, we observe the degree of coincidence at the responding time node, as shown in Fig. 10a, after which we convert the relationship between the load and the electrical signal only if the time node is matched (Fig. 10b). Figure 9 shows data points for the impulse voltage peaks that correspond to their impulse load peaks along with their response time, and the correlation of these sample points for the voltage and load peaks along with the response time is calculated. The equation for the linear fit is = .
+ . y x 1 00005 14 96156 2 2 , the correlation coefficient R 2 is 1.0 and the response time is 1.496 ms; also, y 2 , y 4 , and x 2 , x 4 represent the time of the peak impulse and time of peak the voltage, respectively, and y 3 , y 5 , and x 3 , x 5 represent the peak voltage and peak impulse, respectively.     Fig. 10b, the voltage fitting line is = .
− . y x 1 09953 70 9372 3 3 with an R 2 of 0.99132, a sensitivity of 4.87 mV/ lbf, and a maximum linearity error of 5.7%. It can be seen from R 2 that the peak linearity is improved; that is, the piezoelectric response capability can meet the demand. The sensitivity is compared with piezoelectric sensor products produced in America by DYTRAN 45 , and the model and relative parameters are shown in Table 6. Our sandwich sensor is suitable as a piezoelectric sensor to detect impulse loads (4.93 × 10 5 Pa). with an R 2 and response time of 1.0 and 1.276 ms, respectively. In Fig. 11b), the voltage fitting line is = .
− . y x 0 64845 22 30897 5 5 , with an R 2 of 0.97178, and sensitivity of 0.64845. It can be seen from the R 2 values that the peak linearity is appealing; that is, the piezoresistive response capability can meet the dynamic stress/strain detection for SHM applications.
Moreover, as revealed in Table 7, the sensitivities are basically consistent with the sensor requirements. The sensing range of our sensor reaches 0.5 MPa, which is much higher than that shown for other piezoresistive sensors, with most of the full-scale range below 50 kPa. The overall results on piezoelectric/piezoresistive behaviors indicate that our CNT/PVA/ZnO can be a good candidate for attachment to various types of infrastructure for in situ monitoring.

conclusions
Oriented nano-ZnO arrays were successfully grown on a zinc substrate by a hydrothermal process. PVA-doped functionalized CNTs were well self-assembled on the opposite side, and a CNT/PVA/ZnO flexible composite was finally integrated by a spin-coat process to investigate its piezoresistive and piezoelectric performances. (2) CNT/PVA/ZnO can effectively respond to an impulse load with a high sensitivity, large sensing amplitude, and fast response time of 4.87 mV/lbf, 4.93 × 10 5 Pa, and 1.496 ms, respectively.
These consequences render CNT/PVA/ZnO flexible composites excellent candidates for piezoresistive and/or piezoelectric sensors in SHM applications.  Table 7. The piezoelectric sensing parameters of the CNT/PVA/ZnO composite under impulse loading for our study and compared to those from other references.