Thermomechanical response of liquid crystal elastomers: role of crosslinker density

In this work thermomechanical properties of main-chain liquid crystal elastomers (MC-LCEs) with different degrees of crosslinking were investigated, and gradual loss of thermomechanical response was observed on repetitive measurements. Specifically, six samples of MC-LCEs were prepared, with crosslinker-to-mesogen relative concentration ranging from 5% to 10% in steps of 1%. The obtained results were then compared to thermomechanical response of side-chain liquid crystal elastomers (SC-LCEs). Additionally, thermomechanical response of polymer dispersed main-chain liquid crystal elastomers (MC-PDLCEs) was investigated. Results indicate that in MC-LCEs the concentration of crosslinker defines thermomechanical response and affects stability of the system. The loss of thermomechanical response is negligible in the case of crosslinker to mesogen ratio being the smallest, namely in 5% sample, and it is unaffected by glasslike to nematic phase transition. SC-LCEs do not show any sign of such behaviour and remain stable after several cycles of thermomechanical measurements.


Introduction
The creation of new polymer composite materials with unique properties and controlled physical characteristics is of great importance for the development of future technologies and medical applications [1,2].Soft materials with shape memory that can change their configuration during a phase transition triggered by external factors such as temperature, an external electric or magnetic field, can be used as active components in soft robotics and microfluidics.Also, such systems can release or absorb heat when a mechanical load is applied, which makes them a promising material for the development of new environmentally friendly cooling systems [3].
These materials include liquid crystal elastomers (LCEs).LCE is a polymer network with mesogenic molecules embedded in it.Materials with such a structure combine the elastic properties of polymer chains and the orientational ordering of liquid crystals [4,5].Commonly studied LCEs exhibit a phase transition from a nematic phase to an isotropic phase associated with macroscopic changes in their physical properties, e.g.length, modulus of elasticity, opacity, etc.In view of the polymer network topology, there are two classes of these materials, main-chain liquid crystal elastomers (MC-LCEs) and side-chain liquid crystal elastomers (SC-LCEs).In SC-LCEs, liquid crystal molecules (mesogens) are attached to polymer chains, which are in turn connected by crosslinker molecules to form the elastomer matrix (figure 1(a)).In MC-LCEs, mesogens are part of polymer chains, which in turn are also linked by crosslinker molecules (figure 1(b)), which leads to large changes in the elastic constants and shape of the elastomer during the phase transition.The most investigated LCE systems are those based on polysiloxane chains [6], which are usually synthesized by hydrosilylation of alkenes in the presence of Platinum-based catalysts [7].There are many types of crosslinkers for both SC-LCEs and MC-LCEs, and it has been shown that symmetry of these crosslinkers plays a significant role in properties of LCE systems [8,9], but in this work there was used only one type of crosslinker for each of the classes of elastomers.During the synthesis of SC-LCEs, a crosslinker with two reactive alkene groups is used.For the synthesis of MC-LCEs, a crosslinkers with five reactive groups is used to synthesize MC-LCEs, as it will be detailed later on.
In the recent years, both SC-LCEs and MC-LCEs were used to prepare soft-soft composites, such as the so called polymer-dispersed liquid crystalline elastomers (PDLCEs) [10,11].It has been shown that, due to giant thermomechanical response and high elastic modulus, the use of MC-LCEs in the form of microparticles, used as fillers in PDLCEs will result in a greater thermomechanical response and elastic modulus, as compared to analogous SC-LCEs.
Despite a significant increase in thermomechanical (TM) response in MC-LCEs with regard to SC-LCEs, a systematic study of MC-LCEs samples showed a gradual decrease in strain upon repeated heating and cooling of the sample.This phenomenon, being little discussed in the literature, may prove to be a limiting factor in future application of this class of materials, so the search for the origin of this behavior is of significant importance.Overcoming the limitation on the number of heating and cooling cycles of the sample can be achieved in several ways.In case of the loss of TM response being related to the composition of the sample [12] or, more specifically, to the concentration of crosslinker molecules, which are point defects distributed throughout the sample, an improvement in TM stability can be achieved by optimizing the concentration of crosslinker.If the gradual loss of the TM response is due to the slow relaxation of the elastomer during the phase transition to the glassy phase of MC-LCEs, then no decrease in response should be observed upon thermal cycling the sample above this temperature.
The mechanical properties of LCEs are conventionally characterized by measuring its thermomechanical (TM) response, typically expressed in term of relative elongation, λ(T), and stress-strain curves at different temperatures.These measurements allow for the identification of the phase transition temperatures, which are usually determined by calorimetric methods [13].For instance, the transition from nematic (N) to isotropic (I) phase, or the transition from a liquid crystalline phase into a glassy-like (or glassy) phase, reflect into thermomechanical response anomalies or even drastic changes.The N-I transition is typically related to the anomaly in TM response, since the disappearance of LC order leads to the change in polymer network topology.Moreover, the transition from ordered to not-ordered phase determines significant changes in several chemicalphysical properties, such as viscosity and elasticity [14][15][16].The glass (G) to nematic transition, on the other hand, typically affects the temperature dependence of elastic coefficient, obtained from stress-strain tests, which for MC-LCEs with a small degree of crosslinking changes from ∼ 10 MPa down to ∼ 100 kPa.
The aim of this work is to investigate the effect of the crosslinking density, which is a fundamental parameter in MC-LCE synthesis, in order to optimize their thermo-mechanical properties and to overcome the observed loss of the mechanical properties, which was revealed in repetitive measurements.This study is important for the preparation of PDLCEs made of MC-LCE microparticles, to greatly increase their thermomechanical response, compared to previously reported SC-LCE based composites [17].

Main-chain liquid crystal elastomers and composites
MC-LCEs were prepared using two different approaches, with either stress induced alignment [7,18], or with magnetically induced alignment [10], with 2, 4, 6, 8, 10-tentamethylcyclopentasiloxane (denoted by HD5) as a crosslinker, 1, 1, 3, 3-tetramethyldisiloxane (denoted by TDMS) as a chain extender and 1, 4 bis [4-(butyl 3-ethylenoxy) benzoyloxy] 2-ethyl benzene (denoted as N1) as a bifunctional mesogen.The mesogen, which was produced in our laboratory and otherwise cannot be obtained on the market [8], chain extender and crosslinker are shown on figure 2(a).For the synthesis 1 mol.% of N1, 1 mol.% of TDMS and various amounts (from 0.05 mol.% to 0.10 mol.%) of HD5 were used.The samples were labelled as MC-LCEx, where x is a crosslinker to mesogen ratio in per cents.The LCEs were synthesized using 1.3 ml of toluene as a solvent, and 30 μl of 71 mmol l −1 P −1 t −1 -catalyst solution (with dichloromethane as a solvent).During preparation using mechanical stress, the mixture was filtered into the Teflon chamber, which was left into the centrifuge at 5500 rpm at 343 K for about 18 h.After that, partially crosslinked LCEs were mechanically strained with different loads (from 10 mg up to 200 mg) depending on the crosslinker to mesogen ratio and kept in the universal oven (Memmert company, Germany) at 343 K until the crosslinking reaction is complete.When prepared by the procedure employing magnetic alignment, the mixture is left in the magnet with B = 9 T, at T = 343 K for two days in order to partially orient mesogenic molecules while the polymer matrix forms.It is important to point out that samples of pure MC-LCEs can be used for thermomechanical measurements only if they are prepared with standard procedure involving mechanical stress.Magnetic synthesis is mostly used in this paper in order to prepare composites.In order to prepare composites, partially oriented MC-LCE (with 10% crosslinker to mesogen molar ratio) material is cut into pieces smaller than 1 mm 3 , mixed with PDMS base (Sylgard 184 silicon elastomer) in 1 to 1 ratio and the freeze-fractured in a cryogenic mill CryoMill (by Retsch) with the following settings: pre-cooling at 5 Hz for 2 min, followed by three milling cycles at 30 Hz for 12 min, carried out in 3 min intervals that were separated by a 30 s intercooling cycle at 5 Hz.The sample was further diluted in PDMS base and then mixed with PDMS hardener resulting in volume ratio base/hardener being 40 to 1.After that still liquid sample was put in cylinder form and left in the magnet at T = 343 K overnight in order for the PDMS matrix to polymerize.Resulting composite had 37 vol.% of the particles.

Side-chain liquid crystal elastomers
SC-LCEs consisting of 0.85 mol.% of side-mesogenic units 4-methoxyphenyl 4-(3-butenyloxy) benzoate nematogens (denoted by MBB) attached to a poly(methylhydrosiloxane) backbone (denoted by PMHS) and interconnected by 0.15 mol.% of 1,4-bis(10-undecenyloxy) benzene cross-linker (denoted by 11UB).The samples were prepared so that the alignment of mesogens was induced by external stress.The mesogen, polymer chains and crosslinker are shown on figure 2(b).All the details except for the abovementioned chemical composition and the weights used to apply mechanical stress are the same as for MC-LCE.Details about the synthesis and preparation of these SC-LCEs are also reported in [13].

Extensometry
TM experiments were carried out on a homemade extensometer, the main components of which are a stepmotor-driven linear translator, a strain-gauge in a feedback loop and a heating chamber.The Oxford ITC502 was used for controlling the temperature with a precision of 0.1 K, the load M was measured with a precision of 0.1 mg, and the length of the sample L with a precision of 10 μm.Measurements of temperature dependence of thermomechanical response were carried out in the range 300 K-430 K for MC-LCEs and in the range 300 K-410 K for SC-LCEs.A very small mechanical stress (approximately 10 Pa) was applied along the direction of the orientation of the mesogens in the elastomer in order to be as close to a 'free hanging sample case' as possible.The data on thermomechanical response was collected upon cooling with rate 0.2 K min −1 .
During the measurements of thermomechanical response, the constant load was close to zero.Thermomechanical response, expressed in terms of percentage elongation, λ(T), can be written as where L I is the length of the sample in the isotropic phase.The nematic-isotropic phase transition in these two samples takes place at different temperatures, (T N-I ≈ 393 K for MC-LCE5 and T N-I ≈ 405 K for MC-LCE10), and similar shifts are commonly observed in highly crosslinked LCEs.The anomaly at T G-N ≈ 330 K is clearly visible in MC-LCE5, although the glasslike phase was observed in both samples, whereas for MC-LCE10 the transition from nematic to glasslike phase is seemingly continuous.

Results and discussion
Figure 4 demonstrates the behavior of TM response of MC-LCE with 8% of crosslinker, when it is thermally cycled above the temperature of the transition from glass-like state to nematic phase of MC-LCE (T Glass-Nem ≈ 325 K).In this case TM response drops from ≈ 80% down to ≈ 62% in just four cycles.We thus conjure that loss of TM response in MC-LCEs is unaffected by low temperature phase transition into a glass-like state, and therefore this experiment allows to exclude from consideration one of hypotheses of the origin of thermomechanical degradation.
Let us take a look on the TM response of SC-LCE when they are thermally cycled.As it is shown on figure 5., SC-LCE is very stable even though its TM response upon the first cooling is almost three times smaller than the TM response in MC-LCE (≈ 50% against ≈ 149%).Since SC-LCEs have been reported to be thermally and mechanically stable, this experiment should remove any doubts about persistence of their TM response.
Let us now speculate that putting MC-LCEs in the polymer network might decrease the loss of TM response after thermal cycling.In the first publication introducing the concept of PDLCEs [10] there was an estimation for the domain sizes inside LCE particles after freeze-fracturing.It was shown, by optical microscopy, that typical size of microscopic LCE particles produced by freeze-fracturing according to the procedure used in this study is up to 30 μm [19].If the size of domains inside LCEs is larger than this value, and if the dynamics of domains is responsible for the extremely slow relaxation of mesogenic molecules after thermal cycling, using MC-PDLCE could prevent the loss of TM response.As one can see on the figure 6, it is not the case.As it was previously shown [20], typical size of nematic domains in liquid crystal elastomer varies from about 5 μm to 10 μm, but it increases rapidly on approaching the phase transition temperature.
Figure 7 demonstrates that the general tendency in the behavior of TM response is the shifting of the T N-I to the higher temperature with increasing of concentration, accompanied by the smearing of the phase transition.Such behavior is expected to be present in MC-LCEs, since the energy required for disruption of liquid crystalline order and tendency of polymer network to form coiled structures becomes larger with increase of degree of crosslinking [21,22].However, MC-LCE8 and MC-LCE9 do not follow this general trend.This can be explained in terms of an extremely fast evaporation rate of TDMS chain extender, leading to deviations from expected composition of liquid crystal elastomers.

Discussion
It is clear that the observed inability of MC-LCEs to fully restore their original length is not of a chemical origin, since their transition temperatures do not shift and the thermomechanical response for each sample retains its temperature profile regardless of the number of heating/cooling cycles.One of the possible explanations suggests that, during the cooling of the system, the nematic order parameter does not equilibrate with temperature on approaching the transition into glasslike phase.This may be due to slow stress relaxation, related to hairpin dynamics [23], as it has been shown for acrylate-based MC-LCEs [24].In order to verify this for our system, a sample of MC-LCE8 was selected for additional measurements at higher temperatures.Figure 5 shows the TM response measured repetitively in the temperature range 370 K-430 K.The loss of TM response is still observed, and it proves that the low temperature phase transition does not influence the loss of the actuation.Figure 3 also shows that the N-I phase transition in MC-LCE8 changes from 404 K down to 393 K and reaches the temperature of N-I phase transition in MC-LCE5 specimen.It has been shown that crystallites can form in acrylate-based MC-LCEs in a wide range of temperatures, substantially enhancing mechanical properties of MC-LCEs, but retarding stress relaxation at the same time [25].We conclude that, in our system, the formation of crystallites may begin well above 370 K, and is responsible for the slow relaxation also in nematic phase.
Initial and subsequent TM responses of a series of samples used in the experiments after three cycles of heating/cooling in the temperature range 300 K-430 K are presented in figure 7.During the measurements, the following steps have been taken: heating up the LCE from 300 K to 430 K with the heating rate of 2 K min −1 under the load of 40 mg; cooling down the composite back to room temperature with cooling rate of 0.2 K min −1 under the 40 mg load; removing the load and waiting for 30 min; then repeating the procedure several times.
The responses of the first three sample are almost the same, but the loss of the TM response is small only for the MC-LCE5.MC-LCE6 loses almost 15% of response only after three cycles of measurements.LCE9 has a slower rate of loss of TM response, similar to what is observed in LCE6.

Conclusions
Thermomechanical response of MC-LCEs was studied in a wide range of concentrations of crosslinker.A gradual loss is observed during repetitive cycling.The results of these experiments are compared with thermomechanical measurements of SC-LCEs, which do not exhibit any loss of TM response.From the concentration dependence we find that the concentration of crosslinker greatly influences the thermomechanical response, as well as the stability towards the thermal cycling.The loss of thermomechanical response decreases with lowering the concentration and is found to be negligible in the 5% sample.One possible explanation suggests the formation of crystallites in nematic phase well above TG-N, which in turn slows down stress relaxation in the system.It turns out that by cryo-milling the elastomers and using them as a filler for PDLCE composite, it was not possible to avoid the formation of crystallites.Therefore, the use of MC-LCE with smallest possible crosslinker to mesogen ratio is proposed for future applications.

Figure 1 .
Figure 1.Two kinds of liquid crystal elastomers MC-LCEs (a) and SC-LCEs (b).Large ovals represent mesogens, black lines are polymer chains, stars are crosslinkers for main-chain elastomers and small ovals are crosslinkers for side-chain elastomers.

Figure 2 .
Figure 2. Chemical composition of the LCEs.Three formulas on (a) are the reagents used in preparation of MC-LCEs and on (b) are the chemicals used in synthesis of SC-LCEs.

3. 1 .
ResultsOn figure 3 the TM responses of two sample with extreme values of crosslinker to mesogen ratio (10% on figure3(a) and 5% on figure 3(b)) are shown.As can be seen the TM response of MC-LCE10 decreases with repetitive measurements from 114% to 79%, whereas for MC-LCE5 the change is almost negligible from 149% to 143%.Another prominent feature of MC-LCE5 is the response observed during the first cooling, which is 35% bigger than in MC-LCE10.

Figure 3 .
Figure 3. Thermomechanical response of MC-LCEs with 10% of crosslinker (a) and 5% of crosslinker (b).Cooling curves of five cycles of TM measurements are shown.

Figure 4 .
Figure 4. Evolution of thermomechanical response of MC-LCEs with 8% of crosslinker above T G-N in temperature range 370 K-430 K over four cycles of heating and cooling.Only cooling part of the dependence is shown.

Figure 5 .
Figure 5. Thermomechanical response of SC-LCEs over five cycles of heating and cooling.

Figure 6 .
Figure 6.Thermomechanical response of MC-PDLCE over five cycles of heating and cooling.The crosslinker to mesogen ration is 10%, and the concentration of the particles in polymer matrix is 37%.

Figure 7 .
Figure 7. Thermomechanical response of MC-LCEs with different crosslinker to mesogen ratio (a) and the change of the thermomechanical response of several MC-LCEs depending on crosslinker to mesogen ration in the system after three cycles of measurements (b).