The Effects of Network Architecture on the Photomechanical Performance of Azo-Acrylate Liquid Crystal Elastomers

Abstract

Azo-containing liquid crystal elastomers are photomechanical materials that can be actuated via illumination. The photomechanical response is a result of the photoisomerization of the azo moiety, which produces bulk stresses in the material. These stresses arise via two distinct and competing mechanisms: order parameter change induced stress and direct contractile stress. We describe thermomechanical and photomechanical experiments aimed at assessing the relative contributions of these. We show that the details of the attachment of azo dyes to the network can greatly influence the photoresponse. We discuss our results and summarize our findings.

1. Introduction

Liquid crystal elastomers (LCEs) are a unique class of advanced materials that combine the properties of liquid crystals and rubber-like materials. They consist of a crosslinked polymer network polymerized with monomers possessing liquid crystalline phases. LCEs, first proposed by P.G. de Gennes [1] and realized by H. Finkelmann [2], are remarkable materials [3] due to their exceptional responsiveness to external stimuli, such as heating or illumination. This responsiveness originates from the sensitivity of liquid crystalline order to system parameters near a phase transition. LCEs hold significant potential for applications, as they can sustain and exert shear stresses as elastic solids. This makes them highly versatile for various applications requiring shape changes and stress response, such as soft robotics [4,5,6,7], artificial muscles [8,9], adaptive optics [10], and smart textiles [11,12,13], as well as biomedical applications [14,15,16]. The combination of mechanical flexibility and tunable properties makes LCEs well suited to next-generation responsive and adaptive technologies.

A great variety of LCEs has been synthesized, ranging from Finkelmann’s early polyhydrosiloxane-based ones [2,17,18,19,20,21,22] to more recent acrylate-based [6,23,24,25,26,27,28,29,30,31,32,33], polyester-based materials [34,35], and others [36,37]. Actuation mechanisms used to change the liquid crystal order parameter include heating [6,8,15,23,24,26,27], photoactuation [3,4,7,17,18,19,20,21,22,25,28,30,31,38,39,40,41,42,43,44,45,46,47,48,49,50], the application of electric [51,52,53,54,55] and magnetic [56] fields, and chemical stimuli [57].

In LCEs, photoactuation can be achieved by incorporating azobenzene moieties into the polymer network. Upon illumination, the azobenzene undergoes photoisomerization, inducing internal stresses that can result in shape changes and enable the material to perform mechanical work. The photomechanical response can be attributed to two competing mechanisms: changes in the liquid crystal orientational order parameter and direct contractile stress exerted on the network.

Acrylate-based LCEs containing azobenzene moieties have attracted considerable interest in recent years [7,25,28,45,46,47,48,49,50]. In this work, we focus on amine–acrylate LCEs doped with azo dyes with different degrees of attachment to the network. We measured and compared the thermal and photoresponse of three types of samples: those in which the azobenzene moiety is covalently bonded into the network at both ends (2-azo), at one end (1-azo), and not bonded at all (free-azo). In this paper, we describe our sample materials, report experimental results, discuss our findings in light of the two competing mechanisms, and conclude by briefly summarizing our results.

2. Sample Composition and Structure

2.1. Materials

The chemical structure of the constituents of our samples is shown in Figure 1. The liquid crystal monomer 1,4-Bis[4-(6-acryloyloxyhexyloxy)benzoyloxy]-2-methylbenzene (RM82) was purchased from Jiangsu Hecheng Advanced Materials Co., Ltd. (Nanjing, China). The azo dyes 4,4’-Bis(6-acryloyloxyhexyloxy)azobenzene (2-azo) and acrylic acid (4-((4-hexyloxyphenyl) diazenyl)phenoxy)hexyl ester (1-azo) were purchased from SYNTHON Chemicals GmbH & Co., KG (Bitterfeld-Wolfen, Germany). The azo dye 1,2-Bis(4-(hexyloxy)phenyl)diazene (free-azo) was purchased, as a result of custom synthesis, from Henan Daken Chemical Co., Ltd. (Zhengzhou, China). The chain extender n-Butylamine and the photoinitiator phenylbis(2,4,6-trimethylbenzoyl)phosphine oxide (Irgacure 819) were purchased from Sigma-Aldrich, Merck KGaA (Darmstadt, Germany). All materials were used as received; no further chemical or structural characterization was carried out, and monomer purity was not taken into account in determining molar ratios. Polyimide SE-2170 (Nissan Chemical Corp., Tokyo, Japan) and its thinner 21 (Nissan Chemical Corp., Tokyo, Japan) for liquid crystal alignment were purchased from Brewer Science, Inc. (Rolla, MO, USA).

Amine-acrylate-based LCEs were first proposed by T.J. White et al. [24]; a number of recipes are available for fabrication today [6,23,24,25,26,27,28]. Below, we briefly describe our fabrication process.

2.2. Network Synthesis and Sample Production

Plain glass substrates, $10\mathrm{cm}\times 10\mathrm{cm}\times 1\mathrm{mm}$, were cleaned in an ultrasonic bath with detergent at 60 °C for $15\min$, followed by the rinsing of substrates with deionized water and isopropanol. Following this, the substrates were dried in an oven for $15\min$ at 82 °C. The polyimide mixture SE-2170, in the proportion of 1:3 with its thinner, was spin-coated onto the clean substrates with the following sequence: 1 s at 500 rpm, 30 s at 1500 rpm, 1 s at 50 rpm. After spin-coating, the polyimide was soft-baked at 80 °C on a hotplate for $5\min$ and then hard-baked in an oven at 200 °C for $45\min$. The glass substrates were then cooled to room temperature, and they were rubbed unidirectionally with velvet cloth on a rubbing block 10–15 times. They were subsequently cut into $4.5\mathrm{cm}\times 4.5$ $\mathrm{cm}$ squares and assembled, anti-parallel, into cells using Mylar film (DuPont Teijin Films, Chester, VA, USA) as spacers and Norland Optical Adhesive 68T (Norland Products Inc., Jamesburg, NJ, USA) as glue.

The liquid crystal monomer, azo dye, and photoinitiator were added to the amine and melted at 110 °C while stirring for 30 $\mathrm{s}$. The mass of the photoinitator added was $2.5\mathrm{wt}.\%$ of the combined mass of the other constituents. Relative concentrations of the components of samples are shown in

Table 1. Compositions of sample types.

Sample Type	RM82	Azo Dye	Acryl./Amine	Irgacure 819
	(Mole Fraction)	(Mole Fraction)	(Mole Ratio)	(wt.%)
no azo	1	0	1.1:1	2.5
free-azo 2%	0.98	0.02	1.1:1	2.5
1-azo 2%	0.98	0.02	1.1:1	2.5
2-azo 2%	0.98	0.02	1.1:1	2.5
free-azo 5%	0.95	0.05	1.1:1	2.5
1-azo 5%	0.95	0.05	1.1:1	2.5
2-azo 5%	0.95	0.05	1.1:1	2.5

.

Empty cells with planar surface alignment on the hotplate at 85 °C were filled with the mixture via the capillary effect. Filled cells were moved to an oven at 77 °C and left overnight for oligomerization.

The LCE cells were cured the following day using a mercury lamp (OmniCure s1500, Excelitas Technologies Corp., Pittsburgh, PA, USA). A filter with a cut-on wavelength of $400\mathrm{nm}$ (20CGA-400, Newport Corp., Irvine, CA, USA ) was placed on each cell to reduce the photoisomerization of azobenzene during curing. Polymerization was carried out with light intensity of 75 $\mathrm{mW}/{\mathrm{cm}}^2$ at 70 °C for 20 $\min$. The light intensity was measured at 405 $\mathrm{nm}$ with an Optical Power Meter (1830-R, Newport Corp., Irvine, CA, USA).

After curing, the cells were heated to 100 °C and opened using a thin blade. The LCE samples were removed from the substrate with tweezers.

The differences between samples were as follows: the 2-azo samples contained azo molecules with two covalently bondable double bonds, the 1-azo samples contained azo molecules with one covalently bondable double bond, and the free-azo samples contained azo molecules with no covalently bondable double bonds. All samples were produced using identical procedures; only the type (2-azo, 1-azo, and free-azo) and mole percentage ($2\%$ or $5\%$) of the azo compounds were changed.

A simple schematic of the resulting polymer network, shown in Figure 2, illustrates the key structural differences.

3. Experimental Results

3.1. Elastic Moduli

Young’s modulus of each samples was measured using a custom-built apparatus [29]. For these measurements, the LCE samples were cut into $4\mathrm{mm}$ × $20\mathrm{mm}$ strips with the director oriented parallel to the long edge. Samples marked with black dots were clamped at the top and bottom, and a tensile force was applied. A photograph of the sample was taken for each load, and the strain was determined using ImageJ software (Version 1.54f) [29]. A schematic of the device and experimental setup, together with typical data, is shown in Figure 3.

The LCE samples without any applied stress are not completely flat. Elongating a sample initially, therefore, requires very little stress, which then increases nonlinearly until the linear regime is reached. Measurements are shown in the linear regime. Data and Young’s moduli for all samples are shown in Figure 4.

The elastic moduli measurements of Figure 4 are included to indicate that the elastic properties of our different samples are essentially identical, and differences in thermal and optomechanical properties do not originate from differences in elastic properties. It is also included to enable calculations of strains resulting from both initial stresses, as well as from thermal and photostresses measured in our experiments.

3.2. Thermal Stress Measurements

Our samples for thermal and photostress measurements were $30\mathrm{mm}$ long, $10\mathrm{mm}$ wide, and 27–32 $\mathsf{\mu }\mathrm{m}$ thick, with nematic alignment along their long edge. They were held with the long edges vertical, and their two $10\mathrm{mm}$ edges were clamped between two rigid horizontal plexiglass struts, leaving a 10 $\mathrm{mm}\times 10$ $\mathrm{mm}$ sample area between holders. The bottom edge was held in a fixed position by a rigid support, while the top was connected to an Entran force sensor, whose height could be adjusted, with a nearly inextensible thin filament (Fireline jewelry thread). The sample and its holders were underwater in a thermostatted container. An image of the setup, used for both thermal and photostress measurements, is shown in Figure 5.

Thermal stress was determined via the following procedure. The initial stress on the sample was set by adjusting the height of the Entran sensor. The water temperature was increased in steps. At each temperature, the Entran voltage was recorded continuously for $1min$, and it was then averaged over time to provide the force and the stress exerted by the sample. The thermal stress was obtained by subtracting the initial stress from the measured stress. The heating rate between measurements was 0.25 °C/min.

Thermal stress data from azo-containing samples are presented below, together with photostress.

3.3. Transmission Spectra

Photoactuation in these samples is due to the photoisomerization of the azo moiety. When azo isomers, predominantly in the extended trans-configuration, absorb light at $365\mathrm{nm}$, they undergo photoisomerization to the more compact cis-form, which has an absorption peak near $450\mathrm{nm}$. Information about the dynamics of the population distribution can be inferred from transmission spectra.

We illuminated our samples with light from a $500\mathrm{mW}$ LED at $365\mathrm{nm}$. The intensity of the illumination was 250 $\mathrm{mW}/{\mathrm{cm}}^2$ with duration $10\mathrm{s}$. We then measured the transmittance spectra, whose minima provide information about the relative density of the trans- and cis-isomers, as a function of time. Transmission measurements were obtained using an HR4000CG-UV-NIR (Ocean Optics Inc., Orlando, FL, USA) spectrometer. Spectra were normalized with respect to the sample with no azo dye. The transmittance relaxation data are shown in Figure 6.

Material and optical parameters of the samples are given in

Table 2. Sample material and optical parameters.

Sample	Thickness	Max	Number	Decay Length	Absorption	Cis-Lifetime
	(μm)	Absorbance	Density of Dye	(μm)	Cross-Section	(s)
			(m−3)		(m2)
free-azo 2%	27	0.43	$1.86\times {10}^{25}$	63	$8.50\times {10}^{-22}$	$1.2\times {10}^4$
1-azo 2%	27	0.57	$1.94\times {10}^{25}$	47	$1.10\times {10}^{-21}$	$9.3\times {10}^3$
2-azo 2%	28	0.82	$1.97\times {10}^{25}$	34	$1.49\times {10}^{-21}$	$4.4\times {10}^3$
free-azo 5%	31	1.36	$4.70\times {10}^{25}$	23	$9.20\times {10}^{-22}$	$5.9\times {10}^3$
1-azo 5%	32	1.61	$4.64\times {10}^{25}$	20	$1.08\times {10}^{-21}$	$6.5\times {10}^3$
2-azo 5%	27	1.87	$4.84\times {10}^{25}$	14	$1.48\times {10}^{-21}$	$3.6\times {10}^3$

. The absorption cross-section is $\sigma =1/\left(\rho l_d\right)$, where $\rho$ is the number density, and $l_d$ is the decay length.

Although transmittance is clearly higher for the $2\%$ than for the $5\%$ azo containing samples, there is very little difference in the extent of isomerization in the free-azo, 1-azo, and 2-azo samples. We note here that the isomerization process is only weakly dependent on temperature, given the large excitation energy needed for azo trans–cis isomerization [30].

3.4. Photostress and Thermal Stress Measurements

Photostress measurements were carried out using the same setup as thermal stress measurements, shown in Figure 5. The sample was illuminated with UV light at $365\mathrm{nm}$ from a $500\mathrm{mW}$ LED (Prizmatix Ltd., Holon, Israel) at a distance of $6\mathrm{cm}$ from the sample. The light source was unpolarized, so polarization-dependent effects [31,43] need not be considered. The illuminated area of the sample was $10\mathrm{mm}\times 10\mathrm{mm}$. The illumination sequence consisted of the following: UV OFF for $10\mathrm{s}$, then, repeatedly, UV ON for $200\mathrm{ms}$, and UV OFF for $1\mathrm{s}$. Since the cis-thermal relaxation time was very slow, as indicated in Table 2, after each measurement the sample was illuminated with light at $460\mathrm{nm}$ for $1\mathrm{s}$ to drive the cis–trans isomerization. The sample temperature as monitored via our thermometer was constant to within ±0.5 °C.

To illustrate that the photomechanical effect is a result of the presence of azo dyes, we show the thermal and photoresponse of samples without azo dye in Figure 7. There is no observable photostress in our thermostatted samples without azo dyes.

Photostress measurement results, together with those from thermal stress measurements, are shown in Figure 8 and Figure 9. The results of measurements clearly indicate that the thermal stresses are essentially identical for free-azo, 1-azo, and 2-azo samples with the same initial stresses; however, photostresses of 2-azo samples differ dramatically from those of free-azo and 1-azo samples with the same initial stresses for both 2% and 5% azo concentrations. Our interpretations of these results are elucidated in Section 4.

4. Discussion

Heating changes the orientational order of nematic liquid crystals, which is necessarily coupled with mechanical strain [1,38]. This coupling persists in solid LC elastomers [3] and even in LC glasses [39]. Photoisomerization changes the shape of the azo-containing molecules. This gives rise to two competing mechanisms: 1. a change in orientational order due to the azo-containing molecules being less liquid crystal-like (the shape is less elongated, and the polarizability is less anisotropic) and 2. direct contractile stress due to the contraction of the two ends of the aligned azo-mesogens because of photoisomerization.

Perhaps these two mechanisms can be more clearly understood by considering the schematic of the network structure in Figure 2. The three rows of the right-hand column show the three network architectures studied. The elongated dark blue ellipsoids indicate the RM82 liquid crystal molecules. The extent of the nematic liquid crystal order, measured using the liquid crystal order parameter, is the extent to which these molecules are parallel. In a pure liquid crystal, these molecules tend to be parallel both due to attractive London dispersion forces and due to shape-related steric interactions. These interactions are present in our samples, but they are modified by the topology of the network. Increasing the temperature reduces the degree of liquid crystalline order; the RM82 molecules become less parallel—the distance between neighbors along the chain is shortened, while the distance between neighbors along neighboring chains is increased. This is the essence of mechanism 1; this process was expected to be essentially the same in all three types of samples.

The effect of UV illumination is fundamentally different; it essentially causes the azo-containing molecules, represented by the elongated yellow ellipsoids, to change their shape and become shorter and wider due to photoisomerization. Although all azo-containing molecules are expected to undergo photoisomerization from the elongated trans- to the more compact cis-isomer shape, the effect on the network is dramatically different. In free-azo samples, shown in the top row, the shape change due to isomerization would have a slight effect on orientational order in the vicinity of the azo moiety, but only slight, since the concentration of the azo moiety is small. In one-azo samples, in the middle row, isomerization again will slightly reduce the orientational order, and due to the covalent bond at one end, it could give rise to a slight contractile force along the chain. However, in 2-azo samples, in addition to slightly reducing the orientational order, photoisomerization and the effective contraction of the molecule along its long axis will significantly shorten the chain, giving rise to a direct contractile force. This direct contractile force is the essence of mechanism 2.

The first of these, the change in the orientational order parameter, has often been referred to as the major mechanism for photoactuation in LCEs [17,31,38,40,42,43,44], and it has been seen as playing essentially the same role in both thermal and photoactuation [41,42,44]. The second, direct contractile stress, was already recognized by Finkelmann and Sanchez [20], who referred to the mechanism as a ‘cooperative effect’. Earlier studies have probed the effects of different azo dye attachments [18,19], but the different structures of the dyes used made comparisons difficult. Later work on siloxane elastomers [20] with two different kinds of 2-azo and 1-azo dyes (also different from ours) indicated that higher photostress was exhibited by 2-azo samples. Although the justification is not clearly given in their paper, it is suggested that mechanism 2 provides $60\%$, and mechanism 1 $40\%$ of the stress.

Our three samples, with the 2-azo, 1-azo, and free-azo architectures, exhibit essentially identical thermal stress, as shown in the bottom rows of Figure 8 and Figure 9. This implies that the differences in network architecture do not affect the thermally induced change in order parameter or the stress resulting from the order parameter change. The thermal stress is due to the change of order parameter caused by a change in temperature, and the stress is primarily the consequence of the coupling of the order parameter and the stress.

The three samples, with the 2-azo, 1-azo, and free-azo architectures, exhibit the same photoisomerization, as indicated by the spectroscopic results shown in Figure 6. If the photostress was primarily due to the change in order parameter via mechanism 1, then the photostress would be essentially the same for all samples since the photoisomerization is essentially the same. If the photostress was primarily due to a direct contractile stress due to mechanism 2, then the result for the 2-azo sample would be dramatically different from the response of the 1-azo and free-azo samples. Since the difference in the observed photostress is dramatically larger for the 2-azo samples than those for the 1-azo and free-azo samples, we conclude that the primary mechanism is the contractile stress due to photoisomerization in mechanism 2.

In light of the above, it is interesting to ask the following: what, then, is the role of liquid crystallinity in the photoresponse of 2-azo samples? Likely the role is to provide the alignment of the azo moiety when the sample is prepared. Acrylate LCEs, with or without azo compounds, are essentially aligned nematic samples, which would also align the predominantly trans-azo isomers. Our samples produce uniaxial stress because the stress-producing azo moiety is aligned, and the contraction of the unidirectionally aligned azo-nematogens, with their covalent bonds at two ends, efficiently transfer stress to the bulk network. It appears, then, that the main role of liquid crystallinity in the 2-azo samples is to provide the alignment of the stress-producing azo moieties for the production of unidirectional photostress.

5. Summary

Our experimental results indicate that photostress created in 2-azo acrylate LCEs is due to the direct unidirectional contraction of the network caused by the shape change of the azo moiety during photoisomerization. Stress associated with an order parameter change also exists, as observed in the 1-azo and free-azo samples, but it is about one order of magnitude smaller than photostress in the 2-azo materials. In the 2-azo samples, the traditional actuating mechanism via an order parameter change is, therefore, not significant; the primary role of liquid crystallinity is simply to align the photoresponsive azo-containing molecules during network formation.