A highly stretchable strain sensor based on electrospun carbon nanofibers for human motion monitoring

Abstract

Highly stretchable and sensitive strain sensors are in great demand for human motion monitoring. This work reports a strain sensor based on electrospun carbon nanofibers (CNFs) embedded in a polyurethane (PU) matrix. The piezoresistive properties and the strain sensing mechanism of the CNFs/PU sensor were investigated. The results showed that the CNFs/PU sensor had high stretchability of strain up to 300%, a high sensitivity of gauge factor as large as 72, and superior stability and reproducibility during the 8000 stretch/release cycles. Furthermore, bending of finger, wrist, or elbow was recorded by the resistance change of the sensor, demonstrating that the strain sensor based on CNFs/PU could have promising applications in flexible and wearable devices for human motion monitoring.

---

1. Introduction

Strain sensors detect the change of electrical characteristics such as resistance or capacitance in response to mechanical deformation.¹ Recently, strain sensors for human motion monitoring, structural health monitoring, and human–machine interface, etc. have attracted considerable attention.^2–6 To detect human motions, e.g. the vigorous motions like bending of fingers, arms, or legs, strain sensors need to have high stretchability and sensitivity.^7,8 However, conventional strain sensors, which are made of thin metal foils or semiconductors, typically detect only small strain (<5%) and have limited sensitivity (gauge factor ∼2).^9–11 Therefore, strain sensors with good stretchability and high sensitivity for human motion monitoring are in great demand.

The general strategy to achieve strain sensors with high strain range and sensitivity is to fabricate piezoresistive elastic composite by mixing conductive filler with stretchable rubber/elastic polymer.^12,13 The change of resistance due to the change of the inter-filler distance during the stretch of the composite is used to detect the strain. Various nanomaterials such as metallic nanoparticle/nanowire,^14–16 Si nanowire,¹⁷ carbon nanotube (CNT),^7,18,19 and graphene^8,20,21 are used as the conductive fillers. Strain sensors using metal nanoparticle as a filler may achieve high sensitivity because of the high resistance change during stretch; however the disconnect between the nanoparticles under high strain leads to formation of irreversible inter-particle gaps/cracks, and the devices can only operate with limited strain range.^14,22 The strain sensors based on one dimensional metal nanowire (such as Ag nanowire) and CNT can measure high strain but have relatively low sensitivity.^7,16 Additionally, the preparation methods for these nanomaterial fillers and the piezoresistive elastic composites may be time-consuming and expensive. For example, Si nanowire, CNT, and graphene, are prepared by chemical vapor deposition.^7,23 To fabricate the strain sensors, the fillers are dispersed in a solvent to form “inks”, and patterning/deposition of the inks may involve complicated processing steps.^15,16,18

In this work, we report a highly stretchable and sensitive strain sensor composed of the free-standing electrospun carbon nanofibers (CNFs) embedded in a polyurethane (PU) elastomer. The CNFs were prepared by electrospinning of polyacrylonitrile (PAN) followed by stabilization and carbonization. The strain sensor was assembled by sandwiching the free-standing CNF mat between two PU substrates. The piezoresistive properties and the sensing mechanism of the CNFs/PU strain sensor were investigated. The results showed that the sensor had high stretchability of strain up to 300%, high sensitivity of gauge factor up to 72, and good durability and stability during the 8000 cycles of stretch/release test. Monitoring of bending of finger, wrist and elbow was also demonstrated using the strain sensor.

2. Experimental

2.1 Materials

Polyacrylonitrile (PAN) (M_W = 150 000) powder, tetrahydrofuran (THF, ≥99.9%) and N,N-dimethylmethanamide (DMF, 99.8%) were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA). Thermoplastic polyurethane (PU85A) was purchased form Shenzhen Huayang plastic raw materials factory (Shenzhen, China). All chemicals were used as received.

2.2 Preparation of electrospun carbon nanofibers (CNFs)

CNFs were prepared by the electrospinning technique.^24,25 First, a 12 wt% PAN solution was prepared by dissolving PAN in a mixture solvent of DMF/THF (mass ratio 9 : 1). Then, the PAN solution was filled in a syringe with a metal spinneret. During the electrospinning process, a DC positive voltage of 15 kV was applied to the spinneret. The feed rate of the PAN spin dope was set as 1.0 mL h⁻¹. A grounded aluminum foil was used as the collector, and the distance between the spinneret and the collector was about 20 cm. After electrospinning, a nonwoven PAN nanofibrous mat was obtained. The mat was dried in a vacuum oven at 100 °C for 6 hours prior to the oxidative stabilization and carbonization treatments.

The oxidative stabilization of the PAN nanofibrous mat was carried out in a muffle furnace. The mat was first heated up to 250 °C at a rate of 1 °C min⁻¹, and then the temperature was maintained at 250 °C for 3 hours. During the process, a constant air flow was used to facilitate sufficient oxidative stabilization of PAN. Thereafter, the stabilized PAN nanofibers were transferred into a tube furnace for carbonization. The temperature was increased stepwise to 600, 800, 1000, and 1200 °C at a rate of 5 °C min⁻¹, and the furnace was held at each temperature for 1 hour. Finally, CNFs were obtained after cooling down the furnace to room temperature. Inert argon atmosphere was used during the carbonization and cooling steps.

2.3 Assembly of strain sensor

First, PU granules were dissolved in DMF to prepare 15 wt% PU solution, then, a thin layer of PU film was coated on a clean glass substrate by film casting of the PU solution. After the PU film was dried, a piece of CNFs mat was placed on the PU film. The piece was about 4 mm wide and 20 mm long, cut readily with a scissor. The two ends of the CNFs mat were brushed with silver paste, and the copper wires were glued for the electric contacts. Thereafter, a layer of PU was cast on the surface of the CNFs mat to encapsulate the device. After the PU film was cured and peeled off from the glass substrate, a free-standing and flexible strain sensor was obtained.

2.4 Characterization

The morphologies of the PAN nanofibers, CNFs, and the strain sensor were characterized by Field Emission Scanning Electron Microscopy (FE-SEM, Zeiss Supra 40 VP). The tensile tests of the CNFs fibrous mat, the neat PU film, and the strain sensor were carried out by a QTEST™/10 mechanical testing machine with tensile speed of 10 mm min⁻¹. A custom-built apparatus was used to stretch the strain sensor repeatedly during the stretch/release cycles. The resistance of the strain sensor was measured by Keithley 2612 Source Meter.

3. Results and discussion

Electrospinning is a versatile technique that can prepare polymeric, ceramic, carbonaceous, and composite fibrous materials with the diameters ranging from tens of nanometers up to several microns.^26–28 Compared to the commercial micro-sized carbon fibers, the carbon nanofibers have the advantages of high electrical conductivity, high aspect ratio, and high specific surface area, and have been considered as outstanding materials for applications in energy conversion and storage devices, catalysts, and electrochemical sensors.^24,29–33 In this work, CNFs were prepared by electrospinning of the PAN solution followed with oxidative stabilization and carbonization treatments. Fig. S1† shows the SEM images of the as-spun PAN nanofibers. The continuous PAN nanofibers were randomly overlaid with diameters of 508 ± 34 nm (inset of Fig. S1b†). After stabilization and carbonization treatments, the obtained CNFs had smaller diameters of 400 ± 36 nm (Fig. 1a).

Fig. 1 (a) SEM image of electrospun CNFs. Inset on the upper right: photo of a free-standing and flexible CNFs mat; inset on the lower left: histogram of the diameters of the CNFs. The histogram was obtained by measuring the diameters of 40 individual nanofibers in the SEM image. (b) Pieces of the CNFs mat readily cut with different shapes and sizes.

The CNFs mat was free-standing and flexible, as shown in the optical image in the inset of Fig. 1a. The free-standing CNFs mat could be readily cut into pieces with different shapes and sizes (Fig. 1b), which enabled us to integrate the CNFs mat directly into devices. This approach is different from the general strategy to fabricate strain sensors with high strain range and sensitivity, in which the conductive filler (e.g. carbon nanotube) is mixed with stretchable rubber/elastic polymer.^12,13 The assembly process of the CNFs/PU strain sensor is schematically depicted in Fig. 2a. The as-spun CNFs mat was first cut into a long stripe with desired dimensions. The piece was then placed on a PU film on a glass substrate. Finally, another layer of PU was cast to seal the device. After the PU film was cured, the device was peeled off from the glass substrate. Optical image of a strain sensor is shown in Fig. 2a. The thickness of the CNFs layer was about 60 μm (Fig. 2b), and each PU layer was about 120 μm. The cross-sectional SEM images of the device (Fig. 2c and d) show that the CNFs were embedded in the PU matrix. Continuous nanofibers, broken nanofiber ends, and holes were observed in the cross-section SEM of the CNFs layer, indicating that the CNFs were randomly overlaid in the PU matrix. The nanofibers parallel with the cut direction were continuous, and the nanofibers not aligned with the cut direction were either broken or pulled out when the device was fractured in liquid nitrogen for SEM characterization.

Fig. 2 (a) Schematic illustration of the fabrication process of the stretchable CNFs/PU strain sensor. (b–d) Cross-sectional SEM images of the strain sensor.

Thermoplastic polyurethane (PU) is a highly elastic polymeric material that can tolerate strain higher than 500%.^34,35 Fig. 3a shows the optical images of a strain sensor with initial length of 10 mm being stretched to 100%, 200% and 300% strains. The stress–strain curves of the PU and the CNFs/PU (Fig. 3b) had typical elastomer behaviour. The stress of the neat PU film was about 15 MPa at the break strain of about 510%. The stress of the CNFs/PU material was about 20 MPa at the break strain of 480%; both values of the stress and strain were similar to those of the neat PU film. The difference was likely related to the stress–strain property of the electrospun carbon nanofibers. As shown in Fig. S2,† the CNFs mat could only stand strain less than 4% before break. Even though the strain of the CNFs/PU sensor was as high as 480% before break, the strain sensor was measured up to 300% strain. The CNFs/PU sensor could not fully recover to the initial length when the strain was higher than 300% (Fig. S3†).

Fig. 3 (a) Photographs of a strain sensor being stretched to different strains. (b) Typical stress–strain curves of the neat PU film and the CNFs/PU strain sensor. (c) Resistance versus applied strain of the strain sensor during the first stretch cycle. (d) Relative resistance change (ΔR/R₀) versus applied strain of the strain sensor during the first stretch cycle. (e) Gauge factor (GF) versus applied strain of the strain sensor at the first stretch cycle. The regions i, ii, and iii are corresponding to “break stage”, “slide stage”, and “disconnect stage”, respectively.

Fig. 3c shows the resistance (R) versus the applied strain (ε = ΔL/L₀) of the CNFs/PU strain sensor at first stretch cycle. Here, L₀ is the initial length of the strain sensor, ΔL is the length difference of the sensor at the stretched condition and the initial length. The sensor had a low initial resistance (R₀) of about 85 Ω, indicating the good conductivity of the CNFs mat. When stretched, the resistance of the sensor increased. The resistance of the device at the strain of 300% was about 10 kΩ. Fig. 3d plots the relative resistance change (ΔR/R₀) versus the applied strain, where ΔR is the resistance change (ΔR = R − R₀). At the strain of 300%, the resistance of the device increased about 2 orders of magnitude.

The gauge factor (GF), defined as (ΔR/R₀)/ε, measures the sensitivity of a strain sensor. Fig. 3e shows the variation of GF with the strain for the CNFs/PU device at the first stretch cycle. For the CNFs/PU sensor being stretched to 300% strain, the change of GF could be divided to three different regions. In the strain range between 0 and 20%, GF increased sharply and reached to 12 at 20% strain; in the range of 20–100% strain, GF was nearly constant; in the strain range of 100–300%, GF increased significantly from 14 to 40.

The change of the gauge factor can be explained by the morphological evolution of CNFs during the stretch of the device. Fig. 4a shows the schematic of the morphology of CNFs at different strains; the corresponding SEM images of the cross-sections of the device parallel and perpendicular to the stretch direction are shown in Fig. 4b and c, respectively. The change of morphology can be broken down into three stages, correlated to the three regions of GF in Fig. 3e. At the initial stage (A), the random-overlaid CNFs embedded in the PU matrix (SEM images A1–4) are continuous. For the strain range of 0–20% (stage i, “break stage”), the CNFs start to break during stretch. As seen in the SEM images (B1–4) of the device after stretched to 20% strain, the CNFs are broken into short fibers, which leads to rapid increase of the resistance and the GF. For the strain range of 20–100% (stage ii, “slide stage”), more fibers may be broken into short fibers, and the shortened fibers may slide and align with each other. The SEM images of the device after stretched to 100% strain are shown in C1–4. Parallel to the stretch direction (C1, C2), the shortened nanofibers are observed with orientation along the stretch direction, and the wavy structure of the CNFs mat indicates the sliding of the nanofibers; perpendicular to the stretch direction, few holes are observed in C4, suggesting that there are few long fibers after the device is stretched to 100% strain. In this stage, even though the shortened carbon fibers may slide in the matrix, the connection between the nanofibers remains relatively constant, hence the GF is nearly constant in stage ii (Fig. 3e). When the strain is higher than 100% (stage iii, “disconnect stage”), the CNFs begin to form wavy islands of shortened fibers parallel to the stretch direction (SEM images D1–4), and the connection between these islands determines the resistance of the device. With continuous stretch, the wavy islands of the CNF bundles become less connected, leading to significant increase of resistance and the resultant GF in Fig. 3e.

Fig. 4 (a) Schematic illustration of the morphological changes of CNFs at the first stretch cycle. (b) Cross-sectional SEM images of the strain sensor cut parallel to the stretch direction after being stretched at different strains. (c) Cross-sectional SEM images of the strain sensor cut perpendicular to the stretch direction after being stretched at different strains.

After being stretched to 300% for the first time, the CNFs/PU strain sensor returned to the initial length when the stress was released. However, because the CNFs were broken into short fibers during stretch, the resistance of the device increased to about 2 kΩ. In the next few stretch/release cycles, the irreversible change of the resistance with the strain continued, until the morphology of the CNFs/PU composite became stable. Fig. 5 shows the behaviour of the strain sensor for the first 100 stretch/release cycles. At the second cycle, the resistance of the device increased to about 360 kΩ at 300% strain, indicating the continuous shortening of the carbon nanofibers. After the 10 stretch/release cycle, the resistance values at 0% and 300% strains were stabilized at ∼2.5 kΩ and ∼500 kΩ, respectively. As shown in Fig. 5b and c, the ΔR/R₀ and GF followed the similar trend. The GF reached as large as 72 at 300% at the stretch/release cycle 5. Compared with the one at the first cycle, the GF versus strain curve at stretch/release cycle 100 did not have a distinguished stage i (“break stage”) region, but had similar “slide stage” (0–100%) and “disconnect stage” (100–300%) behaviour. Note that the GF at 300% strain for cycle 100 was slightly lower than those at cycle 5 and 10. This was likely related to the subtle evolution of the morphology after repeated stretch, which led to the increased initial resistance and decreased resistance at high strain.

Fig. 5 Resistance (a), ΔR/R₀ (b), and gauge factor (c) of the strain sensor at the stretch/release cycle of 1, 2, 5, 10 and 100.

The results of the strain sensor behaviour in Fig. 5 can be correlated to the morphology of the CNFs/PU composite during the stretch/release cycles. As we have discussed, during the first stretch/release cycle, the long carbon nanofibers are shortened and the shortened nanofibers form bundles (Fig. 6A1 and A2). The shortening of the nanofibers leads to the irreversible resistance change after the first stretch/release cycle. The irreversible shortening of CNFs continuous in the next few cycles. After stretched for 5–10 cycles, most of the CNFs are broken to short fibers, and the stable bundles of the shortened CNFs are formed. As observed in the SEM images (Fig. 6A1–C1), uniform nanofiber bundles are formed after 10 cycles (C1), and the nanofiber bundles have wavy structures with a period of about 20 μm. The SEM images A2–C2 show that the CNFs become shorter and are well embedded in the PU matrix with repeated stretch/release of the device. After most of the CNFs are shortened and form bundles during the stretch/release cycle, the resistance change of the device is determined by the sliding and connection of the CNF bundles, resulting in stable and reversible strain sensor behavior.

Fig. 6 Cross-sectional SEM images of the strain sensor after being stretched to 300% at cycle 1 (A1, A2), 5 (B1, B2), 10 (C1, C2), and 100 (D1, D2). A1–D1 are the cross-sectional images cut parallel to the stretch direction; A2–D2 are the cross-sectional images cut perpendicular to the stretch direction.

The stability, reversibility, and durability of the CNFs/PU strain sensor were tested using a custom-built stretch/release apparatus (inset of Fig. 7). The stretch and release cycle was set up by a computer-controlled motor. The sensor was first manually stretched to 100% for one cycle to shorten the CNFs before used for cycle test. The CNFs/PU sensor was stretched and released between 100% and 0% strains for 8000 times continuously with a stretch/release period of 5 seconds. The ΔR/R₀ versus cycle number is shown in Fig. 7. The ΔR/R₀ decreased during the first 100 cycles, and then became stable. This result is consistent with our morphological observation (Fig. 6) that the irreversible shortening of the carbon nanofibers is complete after about 100 cycles. For practical application, we can consider the first 100 cycles as a device “aging” step. The device shows superior stability and reversibility during the 8000 cycle test.

Fig. 7 Relative resistance change (ΔR/R₀) versus cycle number for a strain sensor being stretched to 100% strain for 8000 cycles. Inset: the apparatus for the stretch/release cycle test.

Table S1† compares the performance of the CNFs/PU strain sensor in this work and several strain sensors (based on different materials and/or fabricating processes) in the published works in recent years. The CNFs/PU strain sensor shows excellent stretchability, sensitivity and durability. Preparation of the CNFs by electrospinning makes it easy to tailor the size and the shape of the conductive component of the strain sensor, and the sandwiched device structure which embeds the CNFs in a polymeric elastomer can be readily processed and integrated into arrays of strain sensors.

Finally, the capability of the CNFs/PU strain sensor for detecting human motions was demonstrated. Bending of finger, wrist and elbow was monitored using the CNFs/PU strain sensor. A device “aging” step was carried out prior to the test. The response of the device to the bending of finger, wrist and elbow is shown in Fig. 8 and Movie S1–S3 (ESI†). The relative resistance changes (ΔR/R₀) during bending of the finger, the wrist, and the elbow were about 6, 3 and 20, respectively. Obviously, the motion of the elbow gave higher strain (deformation), which was measured by the large resistance change; on the other hand, the bending of the wrist was limited, resulting in the small resistance change.

Fig. 8 The response (ΔR/R₀) of the CNFs/PU strain sensors following the bending motion of finger (a), wrist (b) and elbow (c). Inset: digital images of the flat and bent states of finger, wrist and elbow.

4. Conclusion

This work presents a simple method to assemble a highly stretchable and highly sensitive strain sensor. Carbon nanofibers prepared by electrospinning of PAN followed by stabilization and carbonization, were sandwiched and embedded between two layers of elastomer PU. The CNFs/PU sensor showed large resistance change at strain range up to 300%, high sensitivity with gauge factor up to 72, and superior stability and durability during 8000 cycles of stretch/release. These parameters were among the best ones of the piezoresistive strain sensors reported in the recent literature. Additionally, the CNFs/PU strain sensor showed fast, stable and reproducible response following the bending motion of finger, wrist, and elbow. The stretchable CNFs/PU strain sensor with these excellent properties could have broad applications in wearable devices for human motion monitoring.

Acknowledgements

This research is supported by the National Aeronautics and Space Administration (NASA) of the United States (Cooperative Agreement Numbers: NNX13AD31A), the National Science Foundation/EPSCoR Cooperative Agreement #IIA-1355423, the South Dakota Research and Innovation Center, BioSNTR, and the State of South Dakota.

Notes and references

1. M. L. Hammock, A. Chortos, B. C. K. Tee, J. B. H. Tok and Z. Bao, Adv. Mater., 2013, 25, 5997–6038.
2. T. Yamada, Y. Hayamizu, Y. Yamamoto, Y. Yomogida, A. Izadi-Najafabadi, D. N. Futaba and K. Hata, Nat. Nanotechnol., 2011, 6, 296–301.
3. Y. Wang, L. Wang, T. Yang, X. Li, X. Zang, M. Zhu, K. Wang, D. Wu and H. Zhu, Adv. Funct. Mater., 2014, 24, 4666–4670.
4. K. Inpil, J. S. Mark, H. K. Jay, S. Vesselin and S. Donglu, Smart Mater. Struct., 2006, 15, 737–748.
5. I. M. Alarifi, A. Alharbi, W. Khan and R. Asmatulu, J. Appl. Polym. Sci., 2016, 133, 43235.
6. X. Wang, L. Dong, H. Zhang, R. Yu, C. Pan and Z. L. Wang, Adv. Sci., 2015, 2, 1500169.
7. S. Ryu, P. Lee, J. B. Chou, R. Xu, R. Zhao, A. J. Hart and S.-G. Kim, ACS Nano, 2015, 9, 5929–5936.
8. J. J. Park, W. J. Hyun, S. C. Mun, Y. T. Park and O. O. Park, ACS Appl. Mater. Interfaces, 2015, 7, 6317–6324.
9. M. Hempel, D. Nezich, J. Kong and M. Hofmann, Nano Lett., 2012, 12, 5714–5718.
10. A. A. Barlian, P. Woo-Tae, J. R. Mallon, A. J. Rastegar and B. L. Pruitt, Proc. IEEE, 2009, 97, 513–552.
11. H. Zhang, W. Ding and D. K. Aidun, J. Nanosci. Nanotechnol., 2015, 15, 1660–1668.
12. S. Mark, S. Yi, H. Adam and S. Vesselin, Sens. Rev., 2014, 34, 209–219.
13. X. Li, R. Zhang, W. Yu, K. Wang, J. Wei, D. Wu, A. Cao, Z. Li, Y. Cheng, Q. Zheng, R. S. Ruoff and H. Zhu, Sci. Rep., 2012, 2, 870.
14. J. Lee, S. Kim, J. Lee, D. Yang, B. C. Park, S. Ryu and I. Park, Nanoscale, 2014, 6, 11932–11939.
15. S. Gong, D. T. H. Lai, Y. Wang, L. W. Yap, K. J. Si, Q. Shi, N. N. Jason, T. Sridhar, H. Uddin and W. Cheng, ACS Appl. Mater. Interfaces, 2015, 7, 19700–19708.
16. M. Amjadi, A. Pichitpajongkit, S. Lee, S. Ryu and I. Park, ACS Nano, 2014, 8, 5154–5163.
17. B.-C. Zhang, H. Wang, Y. Zhao, F. Li, X.-M. Ou, B.-Q. Sun and X.-H. Zhang, Nanoscale, 2016, 8, 2123–2128.
18. E. Roh, B.-U. Hwang, D. Kim, B.-Y. Kim and N.-E. Lee, ACS Nano, 2015, 9, 6252–6261.
19. G. T. Pham, Y.-B. Park, Z. Liang, C. Zhang and B. Wang, Composites, Part B, 2008, 39, 209–216.
20. Y. Cheng, R. Wang, J. Sun and L. Gao, Adv. Mater., 2015, 27, 7365–7371.
21. C. S. Boland, U. Khan, C. Backes, A. O'Neill, J. McCauley, S. Duane, R. Shanker, Y. Liu, I. Jurewicz, A. B. Dalton and J. N. Coleman, ACS Nano, 2014, 8, 8819–8830.
22. J. Herrmann, K.-H. Müller, T. Reda, G. R. Baxter, B. Raguse, G. J. J. B. de Groot, R. Chai, M. Roberts and L. Wieczorek, Appl. Phys. Lett., 2007, 91, 183105.
23. X. Wang, Y. Qiu, W. Cao and P. Hu, Chem. Mater., 2015, 27, 6969–6975.
24. L. Zhang, A. Aboagye, A. Kelkar, C. Lai and H. Fong, J. Mater. Sci., 2013, 49, 463–480.
25. G. Duan, H. Zhang, S. Jiang, M. Xie, X. Peng, S. Chen, M. Hanif and H. Hou, Mater. Lett., 2014, 122, 178–181.
26. A. Greiner and J. H. Wendorff, Angew. Chem., Int. Ed., 2007, 46, 5670–5703.
27. D. Li and Y. Xia, Adv. Mater., 2004, 16, 1151–1170.
28. S. Agarwal and S. Jiang, in Encyclopedia of Polymeric Nanomaterials, ed. S. Kobayashi and K. Müllen, Springer, Berlin Heidelberg, 2014, pp. 1–15,  DOI:10.1007/978-3-642-36199-9_370-1.
29. X. Peng, W. Ye, Y. Ding, S. Jiang, M. Hanif, X. Liao and H. Hou, RSC Adv., 2014, 4, 42732–42736.
30. Z. Zhou, C. Lai, L. Zhang, Y. Qian, H. Hou, D. H. Reneker and H. Fong, Polymer, 2009, 50, 2999–3006.
31. Y. Liu, J. Huang, H. Hou and T. You, Electrochem. Commun., 2008, 10, 1431–1434.
32. Y. Zeng, X. Li, S. Jiang, S. He, H. Fang and H. Hou, Mater. Lett., 2015, 161, 587–590.
33. W. Xu, Y. Feng, Y. Ding, S. Jiang, H. Fang and H. Hou, Mater. Lett., 2015, 161, 431–434.
34. S. Jiang, G. Duan, H. Hou, A. Greiner and S. Agarwal, ACS Appl. Mater. Interfaces, 2012, 4, 4366–4372.
35. S. Jiang, A. Greiner and S. Agarwal, Compos. Sci. Technol., 2013, 87, 164–169.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16236c

---