Micro-mechanics of nanostructured carbon/shape memory polymer hybrid thin film.

This paper investigates the mechanics of hybrid shape memory polymer polystrene (PS) based nanocomposites with skeletal structures of CNFs/MWCNTs formed inside. Experimental results showed an increase of glass transition temperature (Tg) with CNF/MWCNT concentrations instead of a decrease of Tg in nanocomposites filled by spherical particles, and an increase in mechanical properties on both macro- and μm-scales. Compared with CNFs, MWCNTs showed a better mechanical enhancement for PS nanocomposites due to their uniform distribution in the nanocomposites. In nanoindentation tests using the Berkovich tips, indentation size effects and pile-up effects appeared obviously for the nanocomposites, but not for pure PS. Experimental results revealed the enhancement mechanisms of CNFs/MWCNTs related to the secondary structures formed by nanofillers, including two aspects, i.e., filler-polymer interfacial connections and geometrical factors of nanofillers. The filler-polymer interfacial connections were strongly dependent on temperature, thus leading to the opposite changing trend of loss tangent with nanofiller concentrations, respectively, at low and high temperature. The geometrical factors of nanofillers were related to testing scales, further leading to the appearance of pile-up effects for nanocomposites in the nanoindentation tests, in which the size of indents was close to the size of the nanofiller skeleton.


Introduction
Shape memory polymers (SMPs) can recover their original shapes from deformed states upon various external stimuli including heat, electric field, magnetic field and light, and their shape memory mechanism is based on reversible energy conversion in polymer chain movements, as shown in Fig. 1. [1][2][3][4][5][6] The programmable shape memory ability of the SMPs has potential applications in sensors/actuators 7 , biomedical devices [8][9] and aerospace industries 10 . Currently, great effort has been made to modify the existing SMPs for the improved mechanical, functional and shape memory properties by adding various nanofillers, aiming to overcome the low stiffness and strength of the pure SMPs which remarkably limit the applications when comparing to other shape memory materials such as shape memory alloys or ceramics. [11][12][13][14] Both carbon nanofibers (CNFs) and carbon nanotubes (CNTs) received great attentions recently as they can effectively improve both the mechanical and functional (such as electrical and optical) properties of the SMP matrix. For example, Leng, et al. 15 reported novel infrared activating ability of shape memory nanocomposites with various carbon nanofillers. Ni, et al. 16 found the dramatic improvement in the macroscopic mechanical properties and shape recovery ability by adding CNTs inside shape memory polyurethane. Gunes, et al. 17 synthesized shape memory polyurethane based nanocomposites by adding carbon black and CNFs for the enhanced electrical properties and electro-active recovery ratio. Jung, et al. 18 prepared carboxyl groups modified CNT reinforced shape memory polyurethane nanocomposites by cross-linking polymerization, and achieved high-performance shape memory.
There are a lot of reports for the characterization of mechanical and shape recovery properties of the SMP nanocomposites, most of which focus on the macroscopic mechanical tests, such as tensile tests, dynamic thermomechanical analysis (DMA), etc. [19][20] . Limited effort has been made to understand the nm-/μm-scale mechanical Fig. 1 A schematic diagram of shape recovery of shape memory nanocomposites with rod/tube shape fillers. enhancement of those SMP nanocomposites. However, this micro-mechanism is essential to understand the enhancement mechanism, to grasp the stretch induced softening effect, and to utilize shape memory polymers/nanocomposites in microdevice applications. [19][20][21] Nanoindentation is a well-known method for the nm-/μm-scale mechanical characterization of materials, which could be used to study the deformation of reinforcement phase. 22 Previous studies have been made on the nanoindentation of pure SMPs. For example, Wornyo, et al. 23 investigated the deformation behavior of diethylene glycol dimethacrylate and polyethylene glycol dimethacrylate shape memory copolymer networks with various organic components based on the nanoindentation results. Fulcher, et al. 24 provided a detailed approach on the thermomechanical characterization of a thermosetting epoxy based SMPs by nanoindentation tests at different temperatures. Nelson, et al. 25 reported a temperature-dependent nano-scale recovery of a thermosetting epoxy based SMPs using the tip of atomic force microscopy (AFM). However, only a few studies are available for the nanoindentation research on the SMP nanocomposites. [26][27] As a further exploration of our previous studies on the thermal-mechanical properties and shape memory performance of nanocomposites reinforced by spherical particles, [27][28] this study is focused on the micro-mechanics and strengthening mechanism of polystyrene (PS) based CNF/MWCNT nanocomposites. Scanning electron microscope (SEM) was used to identify the micro-structures of CNF/MWCNT nanocomposites. Dynamic mechanical analysis (DMA) was used to determine the influence of nanofillers on thermalmechanical properties of nanocomposites, and the nanoindentation tests were used to characterize the micromechanical properties. Atomic force microscope (AFM) was employed to quantitatively study the indents left by the Berkovich tips. Finally, theoretical analysis was done to understand the mechanical enhancement mechanism of CNFs/MWCNTs. This work is focused on the enhancement mechanism of CNFs/MWCNTs on shape memory polymer, which would be helpful for understanding the Mullins' effect [19][20] , establishing constitutive equations 29 and designing tunable wrinkle devices 30 .
Then, a styrene-based precursor (styrene content ≥ 85 wt.%) was added in the solution and the mixture was agitated ultrasonically for 3 hours at a rotation speed of 1000 rpm. Finally, benzoyl peroxide based curing agent (Luperox ATC50, SIGMA, UK) was added and the mixture was stirred ultrasonically at 1000 rpm for another 1 hour. Film samples of SMP and nanocomposites with a thickness of 0.2 mm were casted into the PTFE mould and baked at 75 °C for 36 hours. Samples with exact shape were processed according to ASTM D638 for tensile testing.
SEM (Tescan Lyra FIB/SEM-FEG) was used to observe the micro-structures of the nanocomposites. DMA tests were carried out in tension mode with a TA Instruments DMA 2980 at a frequency of 1 Hz with a default 0.1% peak to peak amplitude, a heating rate of 2 °C/min, and a temperature range from 25 °C to 100 °C. Nanoindentation was carried out using a Tribo-Indenter system (Triboscope, Hysitron Inc., Minneapolis, USA) with a standard diamond Berkovich tip. The measurement was taken at room temperature (~20 °C). An acoustic enclosure was adopted to prevent the acoustic interferences from the environment. The indentation procedures were programmed into three steps, as shown in Fig. 2. The first step is to increase the load to a maximum value with a loading rate of 200 μN/s, followed by a 5-second holding time at the maximum load. This load holding step was applied to minimize the effects of material creep on the estimated values of modulus and hardness, and to investigate the viscous effect on the nm-/μm-scale. 31 The third step is to retrieve the indenter tip from the sample with an unloading rate of 200 μN/s. Finally, two 2× 2 arrays of indents were left on each sample. f nanocomposites which is mainly due to the micro-defects introduced by the aggregates of CNFs. The T g variation of nanocomposites strengthened by CNFs/MWCNTs is totally different from those nanocomposites filled by spherical particles, whose T g decreases slightly with the nanofiller concentration as reported in our previous work. 27 This phenomenon indicates that there is a strong filler-filler interaction for CNFs/MWCNTs inside the hybrid system.
The DMA results characterize the macroscopic thermomechanical properties of shape memory nanocomposites, and will be discussed at both low temperature (T l ) and high temperature (T h ), i.e., corresponding to the two operating temperatures in a standard shape memory programming prucedure. 4,27 To study thermal-mechanics and shape recovery of shape memory polymers/nanocomposites, previous researchers proposed a phenomenological approach to divide the SMP matrix into two phases, i.e., in glassy state and rubbery state, and further assumed that polymer matrix should be fully frozen into glassy state at T l and also fully activated into rubbery state at T h . 29,36 The thermomechanical properties of nanocomposites at T l and T h determine their shape memory performances in shape fixing step and in shape recovery step. In this study, the T l and T h were set as 25 o C and 90 o C (~T g +20 o C).
As shown in Fig. S3, storage modulus E' of these two types of nanocomposites at T l increases with nanofiller concentrations, and reach a maximum value of 711 MPa for the 3 wt.% MWCNT/PS nanocomposites. Due to the softening effect of polymer matrix, values of E' decrease rapidly in the vicinity of T g . At T h , values of E' still increase with nanofiller concentrations, and then reach a maximum value of 3 MPa for the 3 wt.% MWCNT/PS nanocomposites.
As shown in Fig. S4, loss modulus E'' of MWCNT/PS nanocomposites at T l increases with MWCNT concentrations and reaches a maximum value of 90 MPa. For CNF/PS nanocomposites, values of E'' at T l firstly decrease, and then increase with CNF concentrations, which is mainly due to the aggregates of CNFs inside the hybrid system. However, at T h E'' values of these two types of nanocomposites increase with nanofiller concentrations, and reach a maximum value of 1.7 MPa for the 3 wt.% MWCNT/PS nanocomposites.
Results of DMA tests reveal that MWCNTs have a better macroscale enhancement effect than CNTs, which could be attributed to the uniform distribution of MWCNTs, as shown in Fig.3. Compared with the mechanical properties of the pure PS, the increases of storage modulus and loss modulus with filler concentrations indicate that the abilities of MWCNT/PS composites to resist both elastic deformation and creep deformation are strengthened by the addition of MWCNTs at both T l and T h .
As shown in Fig. S5, values of loss tangent tanδ for these two types of nanocomposites decrease with nanofiller concentrations at T l , but increase at T h . The hybrid system are composed of three components, i.e., polymer matrix, nanofillers and polymer-nanofiller interface. Because of the glass transition, the mechanical properties of polymer matrix are strongly dependent on the temperature. However, the nanofillers with much higher moduli (~500 GPa -1 TPa) than the polymer matrix are usually treated as rigid bodies in studying filled rubbery, and could be assumed independent of temperature. 37 Thus, these two opposite trend of tanδ values varied with nanofiller concentrations at T l and T h indicate that the polymer-nanofiller interfacial strength also depends on temperature.
The addition of nanofillers with high elastic modulus, on one hand improves the mechanical properties of pure PS, but on the other hand brings in the temperature dependent polymernanofiller interface. As revealed from Fig. S5, at T l , the polymer-nanofiller interfacial connection is strong, and the addition of nanofillers reduces the damping effect of composites. However at T h , the polymer-nanofiller interfacial connection is weak, and the addition of nanofillers further increases the damping effect of composites.
Therefore, the enhancement effect of nanofillers is dependent on the synergistic influences of mechanical properties of nanofillers, nanofillers distributions, filler-filler interactions and the polymer-nanofiller interfacial strength. The variation of polymer-nanofiller interfacial strength with temperature and its influence on mechanical properties of polymer nanocomposites will be further discussed in Section 3.4.    5 shows typical load-displacement curves for nanoindentation tests of the pure PS and PS based nanocomposites with 1 wt.% CNFs and MWCNTs, respectively. Considering the pyramid geometry size of the Berkovich tip on the μm scale, the smooth curves indicate that there is no significant porosity in testing samples. 38 The loaddisplacement curves for CNF/PS and MWCNT/PS samples show larger slopes of dP/dh in the unloading step and lower values of h max compared with pure PS samples, reflecting the significant enhancement effect of nanofillers on micromechanical properties of nanocomposites. Furthermore, the increase in penetration depth Δh during the load holding step is due to viscous effect. 39 Compared with that for the pure PS, adding MWCNTs significantly reduces the chain mobility of polymer matrix, while CNFs slightly reduces the chain mobility. Similar to the results from the macroscale DMA tests, Fig. 5 reveals that MWCNTs have a better enhancing effect on the mechanical properties than CNFs on the μm scale, which could be attributed to several factors offered by MWCNTs, such as larger surface area, uniform dispersion, etc. As a result, stable and uniform load transferring points are created inside the MWCNT/PS composites leading to a dramatic enhancement in the modulus/hardness.

Results of nanoindentation tests
The calculated values of hardness and elastic modulus as a function of the indentation load are plotted in Fig. 6. Indentation size effect (ISE) is clearly observed as both hardness and modulus values decrease with increasing indentation loads (or indentation depth), before reaching a plateau value. Fig. 6 also reflects that the ISE becomes more obvious with the increase of nanofiller concentrations. As reported in many articles, ISE could be attributed to the intrinsic structures such as indentation elastic recovery, secondary structure, etc. [40][41][42] Here, considering no obvious ISE appearing in pure PS samples, the ISE could be mainly attributed to the secondary structure in the hybrid system. Considering the working condition of nanoindentation and the secondary structures of this hybrid system, Berkovich tips would contact more nanofillers with the increase of filler concentrations, thus leading to an increase of measured modulus and hardness. 22 In Fig. 6 a&b, the measured curve of composites filled by 3 wt.% CNFs is very close to those filled by 2 wt.% CNFs, which indicates that enhancement effects of CNFs is weakened with nanofiller concentrations. This downward trend could be mainly attributed to the random aggregation of CNFs as revealed from Fig.3. Fig. 7a shows the calculated values of the dissipation energy in nanoindentation tests, which generally decrease with the increase of nanofiller concentrations. The dissipation energy, as a quantitative reflection of viscous effect, is due to the internal friction or plastic deformation energy inside polymer/composites. Dissipation energy of the MWCNT/PS nanocomposites decreases continuously until the MWCNT concentration reaching 3 wt.%. Furthermore, there is an abnormal increasing phenomenon for the 3 wt.% CNF/PS nanocomposites, as shown in Fig. 4, which is attributed the obvious agglomeration of CNFs in nanocomposites.
The varying trend of the dissipation energy with nanofiller concentrations in nanoindentation tests is coherent with that of loss modulus E'' in the DMA tests, both of which represent the ability of materials to resist the viscous deformation. The DMA tests were conducted under a strain control, but the nanoindentation tests were conducted under a stress control.     The restriction or enhancement effects of nanofillers on the motion of polymer molecules depends on the temperature, which causes the changes of filler-polymer interfacial strength. As shown in Fig.S5, the damping effect of nanocomposites decreases with nanofiller concentrations at T l , but increases at T h . That phenomenon indicates that the energetic hole of chain localization on the filler surface is strongly dependent on temperature.
The restriction or enhancement effect of nanofillers on the motion of polymer molecules also depends on the length scale, which is achieved by geometrical factors of nanofillers. As shown in Fig.3, inside the nanocomposites, CNFs/MWCNTs form a skeleton due to the strong filler-filler interactions. Therefore, there are two types of networks in the hybrid system, a polymer network and a filler network. The restriction effect of nanofillers works, only if the testing scale is much larger than the size of the filler network. However, for nanoindentation tests, the size of indents (4-6 μm) is very close to the size of nanofiller networks (about 1-5 μm), as shown in Fig.3 and Fig.8, respectively. Thus, the addition of nanofillers only hinders the long rang motion of polymer chains, but not affecting their local motions, which further lead to the appearance of the pile-up edge beside the indents when the measured modulus and hardness are both increased.

Conclusions
In this paper, the CNF/MWCNT PS based nanocomposites were fabricated and their micro-mechanics and enhancement mechanisms were experimentally investigated.
SEM images reveal the skeleton formed by CNFs/MWCNTs in the SMP nanocomposites. Experimental results reflect a better enhancement effect of MWCNTs than CNFs. As reflected in DMA tests, the glass transition temperature of nanocomposites enhanced by CNFs/MWCNTs is significantly increased by nanofiller concentrations from 60 o C to 75 o C, and this trend is totally different for those of the nanocomposites filled by spherical particles. The addition of nanofillers can lead to an increase in the mechanical properties of nanocomposites, except for the 3 wt.% CNF/PS composites. The storage modulus and loss modulus of nanocomposites both increase with nanofiller concentrations at T l (25 o C) and T h (90 o C). However, the loss tangent decreases with nanofiller concentrations at T l , but increases at T h . As reflected in nanoindentation tests, the addition of nanofillers leads to an increase in hardness and modulus of the materials on the μm scale. In indentation tests by the Berkovich tips, the indentation size effect and the pile-up effect both obviously appear in nanocomposites, but not in those of pure PS.
Theoretical analysis shows that mechanical properties and enhancement mechanisms of nanocomposites are both strongly dependent on their secondary structures, which could be further divided into the filler-polymer interfacial connection and geometrical factors of nanofillers. The filler-polymer interfacial connection, which is due to the localization of polymer chains on the surface of nanofillers, relies on the temperature. The high energetic hole at T l and low energetic hole at T h of polymer chain localization lead to the opposite trends of loss tangent with nanofiller concentrations at T l and at T h . The geometrical factors of CNFs/MWCNTs are due to the large ratio of length and diameter, and the enhancement effects of nanofillers on the motion of polymer molecules only work when the testing scale is larger than the size of filler network. Because in the nanoindentation tests, the size of indents (4-6μm) is very close to the size of the nanofiller networks (about 1-5μm), the pile-up effects are obvious in the nanocomposites compared with those in the pure PS, further leading to the failed prediction from the O&P criterion for the polymer based nanocomposites.