Facile Fabrication of Porous Conductive Thermoplastic Polyurethane Nanocomposite Films via Solution Casting

Porous conductive polymers are one of important materials, featuring lightweight, large specific surface area and high porosity. Non-solvent induced phase separation is widely employed to prepare porous polymer sheet materials. Through utilizing water vapor in ambient environment as the non-solvent, a facile approach was developed to produce porous conductive polymer nanocomposites using the conventional solution-casting method. Without using any non-solvent liquids, porous carbon nanofiber/thermoplastic polyurethane (CNF/TPU) nanocomposites were prepared directly by solution casting of their dimethylformamide (DMF) solutions under ambient conditions. The strength of the CNF framework played a key role in preventing the collapse of pores during DMF evaporation. The dependence of porous structures on CNF loading was studied by scanning electron microscopy and porosity measurement. The influence of CNF loading on the mechanical properties, electrical conductivity and piezoresistive behavior was explored.


Introduction
Thermoplastic polyurethane (TPU) is a kind of multi-block copolymers, typically synthesized from a diisocyanate and a long-chain diol with a small molecule diol as the chain extender 1 . Through end-useoriented choice of the di-isocyanates and diols, a huge number of TPUs with various physical properties have been developed 2 . TPU elastomers feature soft segments with glass transition temperature (Tg) lower than room temperature (typically, -30 ~ 20 o C) 3 . The hard segment comprises rigid backbone moieties, rich of potential hydrogen-bonding sites (i.e. urethane groups), forming the hard domain through microphase separation, and serving as the physical cross-linkage in TPU elastomers 4 . Owing to their mechanical robustness, high resilience, small compression set, and excellent resistance to impacts, abrasions, tears, and weather, TPU elastomers have been extensively used over the fields of coatings, footwear, automotives and biomedical industry 5 .
Recently, electrically conductive TPU composites have arisen comprehensive interest in electrical areas like antistatic, electromagnetic shielding and sensing materials 6 . To prepare conductive TPU composites, embedding conductive fillers (e.g. conductive carbon black, nanotubes, nanofibers, graphene, silver nanowires, metal microparticles) within TPU elastomers is one attractive method, in terms of the advantages of manufacture simplicity, cost-effectiveness, and tuning of conductivity 7 . However, it is difficult to develop highly conductive TPU composites by increasing the loading of conductive fillers and simultaneously conserve the outstanding resilience originating from TPU elastomers 8 . This drawback will eventually worsen the stretchability and durability of the material. Porous structures (e.g. foams, sponges, aerogels), which feature lightweight, large specific surface area and high porosity, have been explored as a promising measure to tackle this issue. The application of porous materials have been widely spread over biological scaffolds 9 , catalyst carriers 10 , membrane filters 11 , thermal insulators 12 , super chemical adsorbents 13 , and energy absorbers 14 . Meanwhile, porous conductive polymer composites have demonstrated great potential in novel electronics, including electromagnetic interference shielding 15 , triboelectric generators 16 and piezoresistive sensors 17 . The introduction of appropriate porous structures not only helps to effectively reduce the density and cost, but also enables the materials with improved flexibility, stretchability, and strain at break.
There are many approaches to introduce porous structures to TPU elastomers, such as in situ polymerization/batch foaming 18 29 . A maximum graphene loading of 10 wt.% was achieved, and the porous graphene/TPU composites showed high compression modulus and low thermal conductivity. Porous conductive CNT/TPU composites were also developed by using carbon dioxide as foaming agent 30 . It was found the percolation threshold of porous CNT/TPU nanocomposites was higher than their solid counterparts, attributed to the volume expanding. Freeze drying method is a combination of thermally induced phase separation and template-leaching-like techniques. Porous conductive CNT/TPU and graphene/TPU composites prepared by freeze drying exhibited high porosity (up to 90%) and well-defined piezoresistive behavior 17,31,32 . Phase inversion, involving non-solvent induced phase separation, is considered as a facile and low-energy-consuming approach to prepare membranes with well-defined cell structures 33 . It was found that when exposed to water vapor, TPU dissolved in dimethylformamide (DMF) and N-methyl-2-pyrrolidone (NMP) would be precipitated. If the concentration was high enough, the TPU solution would transform into an organogel. But, the porous structure in the TPU organogel could not be held during the evaporation of solvent because of the evaporation-induced shrinkage, as illustrated in a following section. To avoid the collapse of the porous structure, Chen et al. immersed the organogel into non-solvent water to solidify the framework and the resultant porous graphene/TPU composites exhibited lower modulus, larger elongation at break, and lower hysteresis 7 .
In this study, to avoid the collapse of the porous structure in TPU organogel, we utilized carbon nanofibers (CNFs) to cement the porous structure and developed a facile approach to produce porous conductive TPU nanocomposite films using the conventional solution-casting method. Without using any non-solvent liquids, porous CNF/TPU nanocomposites were prepared directly by solution casting of their DMF solutions in the ambient environment. The porous structures and their dependence on CNF loadings were studied by using scanning electron microscopy (SEM) and porosity measurement. The influences of CNF loading on the mechanical properties, electrical conductivity and piezoresistive behavior of the porous nanocomposites were investigated. To the best of our knowledge, few studies have declared the preparation of microcellular polymer nanocomposites directly by solution casting. This study provides a novel and facile approach to produce lightweight conductive polymer films.

Results and discussion
Preparation of porous CNF/TPU nanocomposites Figure 1. Digital photographs of a TPU DMF solution (0.08 g/mL) (a), TPU DMF organogel at room temperature (b) and at 50 o C (c). Cross-section SEM images of TPU samples prepared from TPU organogel by solvent exchanging with water (d) and DMF evaporation (e).
As discussed above, TPU DMF solution is not stable when exposed to the humid air at room temperature, which is known as water-vapor induced phase separation 7 . Figure 1a illustrates a TPU DMF solution (0.08 g/mL). Once removing the cap, the gelation occurred under the ambient conditions, starting at the liquid surface and extending into the depths beneath. In this case, the gelation fully completed in 12 h and the resultant TPU DMF organogel is shown in Figure 1b. The TPU DMF organogel was found thermally responsive such that the sample enclosed in a vial could regain the capability to flow at a temperature higher than 50 o C (as shown in Figure 1c) and recovered after cooled down to room temperature. To obtain TPU sheets, we removed DMF from the TPU DMF organogel using two methods. One was solvent exchanging with water (immersing the organogel in water followed by drying to remove water); the other was allowing the evaporation of DMF in a fume cardboard at room temperature. The cross-section SEM images of the resultant two TPU sheets are shown in Figure 1d and 1e. It can be seen that, without the solidification using water, the porous structure in the TPU DMF organogel was not able to retain after the evaporation of DMF (Figure 1e). It was because that the TPU-phase framework in the organogel was not strong enough to withstand the shrinkage caused DMF evaporation. By contrast, owing to the restriction from the container, the uneven bottom surface of the DMF-evaporation sample still preserved the traces of the porous structure in the TPU DMF organogel. It was also noted that the top surfaces of both samples were fully covered by numerous uniform, connected TPU micropartiles (size ~5 µm), suggesting the growth of TPU DMF organogel by coalescence processes. 34 The morphologies derived from TPU aggregation at the liquid surface and in the bulk liquid phase were different, which might be related to the diffusion of water vapor and the convective flow induced by DMF evaporation.  (e, f), and TPU40 (g, h).
Rigid conductive fillers were proposed to cement the TPU-phase framework in the organogel so that porous structures could be achieved using the conventional solution-casting method. Figure 2a illustrates the pathway to utilize CNFs to reinforce the organogel and prepare porous CNF/TPU nanocomposites directly by solution casting. The obtained CNF/TPU DMF organogel was quite strong as shown in Figure   2b. After removing DMF through evaporation, the resultant CNF/TPU nanocomposite sheets were porous and flexible, as shown in Figure 2c. It can be seen from Figure 2d that the top surface of CNF/TPU nanocomposites was smoother than that of the neat TPU sample as shown in Figure 1e. This indicates the presence of CNFs significantly suppressed the formation of TPU microparticles at the liquid surface because they spatially blocked the coalescence of small TPU nucleus particles. 34 35,36 . It was noted that TPU showed only one well-defined peak at 1730 cm −1 , corresponding to free C=O 37 , indicating most of carbonyl groups in TPU were not involved in hydrogen bonding. For TPU cast from its THF solution, 56% of carbonyl groups were involved in hydrogen bonding 37 . The difference in the hydrogenbonding state of carbonyl groups might be related to the solvent used for solution casting. CNFs show no obvious IR absorption peaks and the percent transmittance monotonically increases with the wavenumber decreasing, due to their highly graphitic structure, which is in great agreement with previous report 38   Mechanical properties of porous CNF/TPU nanocomposites were investigated by tensile testing. Figure   5a and 5b show the tensile stress-strain curves. The averaged results are summarized in Figure 5c and 5d, with the error bars referring to standard deviations. TPU is not an ideal linear viscoelastic material, where the modulus changes with the strain 42 . Ultimate tensile strength, elongation at break, and Young's modulus (at initial linear stage) were 7.41 MPa, 798%, and 2.92 MPa for neat TPU, respectively. In comparison to neat TPU, tensile strength and Young's modulus of TPU10 (with a porosity of 24.3%) were raised by 34% and 22%, respectively, due to the reinforcement effect of CNFs. For highly porous samples, namely TPU20 (80.1% porosity), TPU30 (87.3% porosity) and PU40 (84.5% porosity), their tensile strength and Young's modulus were remarkably lower than that of neat TPU and TPU10 owing to high porosity. It was noticed that there was no clear trend in the Young's modulus. This was caused by the fact that a higher CNF loading is expected to lead to a higher Young's modulus in its non-porous TPU nanocomposite; yet the Young's modulus of a porous nanocomposite is also inversely affected by the porosity which is again dependent on the CNF loading. The values of elongation at break were 810%, 149%, 87% and 12% for TPU10, TPU20, TPU30 and TPU40, respectively. The dramatic decline in elongation at break from TPU10 to TPU20 can be explained by the fact that the latter had a much higher porosity which reduced the uniformity of the material, resulting in more defects and other stress concentrations. Although these values were lower than that of neat non-porous TPU, the mechanical performance of TPU20, TPU30 and TPU40 was commendable considering such high porosities. Specific stiffness and specific strength of samples are shown in Figure 5e. It can be seen that TPU30 presented the best stiffness-to-weight ratio while TPU10 showed the best strength-to-weight ratio.

Electrical properties of porous CNF/TPU nanocomposites
The CNF network provides a continuous conductive network in the insulating TPU matrix. The experimentally determined electrical conductivity (σ) increases with the CNF loading, from 2.66×10 -3 S/cm with 10 wt.% CNFs to 2.98×10 -2 S/cm with 40 wt% CNFs (Figure 6a). It was found that the resistance of CNF/TPU nanocomposites changed under stretching. At 50% stain, the resistance was raised by 17.2 and 1.7 times for TPU10 and TPU20, respectively. The increment at 30% stain was 5.0, 0.8 and 1.7 times for TPU10, TPU20 and TPU30, respectively. It suggests that highly porous samples, TPU20 and TPU30, exhibited less sensitive dependence on strain, in comparison with TPU10. This means the high porosity could significantly reduce the interference of stretching operations on the conductivity and benefit the stability of the electrical conductive performance of materials in the application of flexible electronics. The relaxation behavior of each sample at different strains is shown in Figure 6c. The sample was stretched to a desired strain (e.g., 5% and 10%) at a speed of 50 mm/min and then waited for 1 min before further stretching. It is apparent that all the samples had a remarkable creep and the relaxation amplitude increased with increasing strain. Taking TPU30 for an example, the relative resistance relaxed by 0.11 at 5% strain and 0.35 at 30% strain. This overshoot behavior was commonly observed in conductive elastomer composites 43,44 . This was caused by the relaxation of conductive networks in conductive elastomer nanocomposites, which was expected to be controlled by the relaxation behavior of elastomer matrix. Therefore, to determine the relaxation time, the experimental relaxation curves of the CNF/TPU nanocomposites as a function of time (t) at different strains were fitted with the stretched exponential Kohlrausch's equation (1) Here, R∞, R1, τ, and β refer to the fitting constants. τ is the relaxation time and β is the stretching parameter (0< β≤1). The experimental data were well fitted with the theoretical values and the obtained relaxation times are given in Figure 7.

Conclusions
Porous conductive TPU nanocomposites were successfully prepared by a conventional solution-casting method. As a rigid conductive filler with high aspect ratio, CNFs were utilized to cement the porous structure in the water vapor-induced TPU organogel. The porosity (void fraction) of samples was calculated from the true density measured on AccuPyc-II-1340 pycnometer (Micromeritics Instrument Corp.) and the bulk density.

Data Availability Statement
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request and shared in Zenodo repository.