Tailoring the Photomechanical Response of Glassy, Azobenzene

Sep 14, 2012 - Relaxation Dynamics and Strain Persistency of Azobenzene-Functionalized Polymers and Actuators. Amir A. Skandani , Sourav Chatterjee , ...
11 downloads 8 Views 1MB Size
Download PDF
Article pubs.acs.org/Macromolecules

Tailoring the Photomechanical Response of Glassy, AzobenzeneFunctionalized Polyimides by Physical Aging By Kyung Min Lee,†,‡ Hilmar Koerner,†,§ David H. Wang,†,§ Loon-Seng Tan,† Timothy J. White,† and Richard A. Vaia*,† †

Materials & Manufacturing Directorate, Air Force Research Laboratory, Wright-Patterson Air Force Base, Ohio 45433-7750, United States ‡ Azimuth Corp., Dayton, Ohio 45432, United States § UES, Inc., Dayton, Ohio 45432, United States S Supporting Information *

ABSTRACT: Photoresponsive polymers convert a light stimulus input into a mechanical output (work). Photoinduced conformational changes, such as within azobenzene, dictate molecularlevel distortions that summate into a macroscopic strain, which often manifests as a shape change or motion. The transduction of the molecular-level response to a macroscale effect is regulated by mesoscopic features, such as chain packing, free volume, and local molecular orderfactors which depend on chemical composition as well as the process history of the material. Herein, we demonstrate the ability to widely tailor the photomechanical response of a photoresponsive polymer by manipulating the energy state of the glass, rather than formulating new chemical compositions. Physical aging increases the density of the glass, reduces local free volume, and thus reduces the minima in local conformation space, thereby strongly influencing the azobenzene photochemistry (trans−cis−trans isomerization). The subsequent change in the energy landscape of the system reduces the fraction of azobenzene able to undergo reconfiguration as well as increases the probability that those photoinduced conformations will relax back to the initial local environment. The result is a tuning of the magnitude of macroscopic strain and the ability to shift from shape fixing to shape recovery, respectively.

■

INTRODUCTION

The mechanical output of photoresponsive polymeric materials depends on a complex interplay between light absorption (regulated by intensity and wavelength of light as well as the concentration of absorbing species in the material system3), the morphology of the polymer network (amorphous, semicrystalline, liquid crystalline),4−6 thermomechanical properties of the material (glassy or rubbery state, modulus),3,7−9 and the geometry of the sample.7,10,11 The insights into the connections between these variables and the mechanical output have predominantly been derived by varying the composition of the photoresponsive polymeric material. These studies have shown that the efficiency and magnitude of the response depend critically on the collective molecular mobility and the wavelength of illumination. For example, elastomers can yield photostrains greater than 20%.4 Here, collective molecular motion is relatively fast (i.e., T > Tg), and the impact of a molecular photoisomerization event may be amplified through coupling to a phase transition, such as phototriggering of an order−disorder transition.4,12,13 In contrast, photoisomerization in amorphous or liquid crystalline glasses (T < Tg) is constrained to local molecular rearrangements and thus limits

Photomechanical effects in polymeric materials have been the subject of scientific curiosity for more than four decades.1,2 The photomechanical response of these materials can be thought of as an energy transduction process which converts a light stimulus input into a mechanical output (work). As such, they are crucial enablers for a broad range of adaptive technologies in medicine, aerospace, and telecom, where the nature of light provides means for remote triggering and actuation that is not possible with alternative electrical and thermal techniques.1,2 The motif commonly used is to covalently incorporate photochromic moieties that undergo conformational changes (such as the isomerization of azobenzene) within the polymer network. The photoinduced conformational changes dictate molecular-level distortions that summate into a macroscopic strain, which is often visualized as a shape change such as the deflection of a cantilever. To date, the predominant approach to new or improved functional responses has focused on the preparation of new material compositions. The transduction of the molecular-level response to a macroscale effect however will also be regulated by mesoscopic features, such as chain packing, free volume, and local molecular orderall factors that for a given chemical composition also depend on the process history of the material. © 2012 American Chemical Society

Received: July 30, 2012 Revised: August 31, 2012 Published: September 14, 2012 7527

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

Figure 1. (a) Chemical structure of azo-20-CP2 containing three-arm azobenzene net points (n,m,l is degree of polymerization of each arm). Bending of a 5 mm × 1 mm × 20 μm cantilever composed of physically aged azo-20-CP2 samples [(i−iii) RQ: rapidly cooled; (iv−vi) SQ: slowly cooled] before (i, iv), after 2 h of irradiation with polarized 442 nm light (100 mW/cm2) parallel to the long axis of the cantilever (E||x) (ii, v), and 72 h after irradiation (iii, vi). (b) Photomechanical response of cantilevers of RQ (red circles) and SQ (blue circles) during 2 h of irradiation with polarized 442 nm light (100 mW/cm2) and following relaxation in the dark.

to embrittlement.25 Therefore, controlling thermal process history of a photoresponsive polymer glass provides an equally powerful approach to modifying the local structure, morphology, and dynamics of a polymer glass, and as such it should impact the macroscopic photomechanical behavior of the glass in a manner distinct from composition and polymer network connectivity. We report here the ability to widely tailor the photomechanical response, from shape fixing to shape recovering, of a polymer glass of a set composition by manipulating the energy state of the glass via physical aging. Physical aging (also referred to as thermal annealing, aging, or structural relaxation) increases the density of the glass, reduces local free volume, and thus reduces the minima in local conformation space. These factors strongly influence the azobenzene photochemistry (trans−cis−trans isomerization), reducing the fraction of azobenzene able to undergo reconfiguration as well as increasing the probability that those photoinduced conformations will relax back to the initial local environment. The result is a tuning of the magnitude of macroscopic strain and the ability to shift from shape fixing to shape recovery, respectively.

options to amplify the event through collective molecular processes. This results in strains of 0.1−2% but with a higher force output due to increased stiffness.14 In concert with the glassy−rubbery state of the material, the wavelength of the exposing irradiation determines the specific steps of the photochemical process. For example, UV light will isomerize azobenzene from the rod-like trans conformation to the bent cis conformationa reduction in the length of the molecular axis from 9 to 5.5 Å.15 Recovery from the metastable cis conformation to the thermodynamically stable trans conformation occurs by heating or by irradiation at the absorption maximum of the cis isomer (∼530 nm). As such, two light sources are used to cycle between the setting of the initial shape and subsequent recovery. Note that the set shape is metastable and ultimately governed by the kinetics of the thermal-driven cis−trans isomerization in the dark. Alternatively, blue-green wavelengths (440−514 nm), which are absorbed nearly equivalently by both the trans and cis isomers,16,17 simultaneously drives both trans−cis and cis−trans processes (trans− cis−trans). When the blue-green light is linearly polarized, trans-azobenzene is reoriented normal to the electric field vector of the polarization due to the rotational freedom of the azo bond and the polarization dependent dichroic absorbance of both the trans and cis isomeric forms.18 The local reorientation of trans isomers results in a highly stable, optically fixed shape in polymeric glasses without a substantial concentration of the thermally sensitive cis isomer5,6,9,14,19−23 and the ability to reverse the deformation at room temperature.24 Collectively, these prior efforts provide substantial evidence that the local structure, morphology, and dynamics of the polymer regulate the efficiency, kinetics, and magnitude of the photomechanical response. Surprisingly though, prior research on the development of new photomechanical polymers has exclusively focused on exploring new compositions of matter to optimize the coupling between the local environment and the chromophore. However, the local structure and dynamics of glasses in particular are also highly sensitive to prior process history. For example, physical aging of amorphous polycarbonate slightly below its glass transition temperature (Tg) for an extended period of time can transform ductile deformation

■

EXPERIMENTAL SECTION

Materials. The synthesis of polyimide networks containing trifunctional azobenzene moieties at cross-link sites has been reported elsewhere.5 In brief, the diamine (BAPB, 0.4093 g, 1.400 mmol) and DMAc (8 mL) were added to a 50 mL three-necked flask equipped with a magnetic stirrer, nitrogen inlet, and outlet and stirred under dry nitrogen at room temperature for 30 min. The dianhydride (6FDA, 0.8885, 2.000 mmol) was then introduced to the resulting solution. The light yellow solution was agitated at room temperature for 24 h to afford a poly(amic acid) (PAA) solution. Then, the tris(azobenzeneamine) cross-linker (0.3568 g, 0.400 mmol) was added to this solution. Herein, azo-20-CP2 (Figure 1) refers to a polyimide with 20 mol % tris(azobenzeneamine). Cantilevers of 20 μm thickness were prepared by casting the above solution onto 2 in. × 2 in. glass plates. 1−2 μm films for UV−vis measurements were prepared by spin-coating the 20 wt % PAA solution onto glass slides at 2000−3000 rpm. Both films were heattreated at 50 °C (vacuum, overnight), 100 °C/2 h, 150 °C/2 h, 175 °C/1 h, 200 °C/2 h, 250 °C/1 h, and 300 °C/1 h to form cross-linked polyimides. After drying, the films are removed from the glass substrates and cut into 5 mm × 1 mm × 20 μm cantilevers. Cantilevers 7528

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

Figure 2. (a) Heat capacity cp of physically aged (RQ, red line; SQ, blue line) azo-20-CP2. The overshoot seen in SQ is a result of the release of enthalpy that was gained when the sample physically aged during slow cooling (see Supporting Information for details). (b) Wide-angle X-ray diffraction of cantilever in normal (i) and perpendicular (ii) direction through film, before and after physical aging and physical aging thermally erased (three patterns overlap). The local order remains the same, irrespective of direction. Data in (a) and (b) have been offset for clarity. were heated to Tg + 50 °C (270 °C) for 30 min under vacuum to thermally erase all prior process history. Physical aging of the samples was accomplished via rapid quenching (RQ) by immediately immersing the cantilever in liquid nitrogen (−196 °C) and slow quenching (SQ) by cooling the cantilever in air at a rate of 1 °C/min. Characterization. The active response was visualized using deflection of a thin cantilever of the azo-20-CP2 clamped at one end. Note that although bending angles (radius of curvature) can exceed 90°, for example in the case of twisted nematic LCN glasses,26 the photogenerated strain response on the illuminated surface is small (0.1−2%). It is important to note that for the cantilever to bend light must also be attenuated (e.g., absorbed) nonuniformly across the sample thickness to yield a strain gradient. Thus, for the given chromophore content, the thickness of the cantilever was chosen such that illumination was attenuated by 1/e ∼3 μm into the cantilever (attenuation coefficient of azo-20-CP2 at 442 nm is 3460 cm−1). This created an exponential intensity decay through the thickness of the cantilever that is mirrored by an equivalent decay in the photoisomerization events that ultimately manifests in a strain gradient across the cantilever thickness. Temperature increases due to laser exposure at room temperature are only 3−5 °C maximum as determined by calibrated thermal imaging. Physical aging studies were carried out on a TA Instruments Q1000 DSC following standard procedures reported in the literature (Supporting Information).27,28 Wide-angle X-ray experiments were carried out on a Statton box camera in transmission mode using Cu Kα generated by a Rigaku ultraX 18 system.

mW/cm2 of 442 nm illumination (helium cadmium (HeCd) laser), linearly polarized parallel to the long axis of the cantilever. The relation between deflection and photoinduced strain gradient within the cantilever is well described by absorptivity and beam mechanics.14 Upon illumination, the RQ azo-20-CP2 cantilever immediately bends to a large angle (15°) that is followed by slower continual bending over 2 h. Upon removal of the illumination (dark state), the bent cantilever quickly recovers a few degrees but retains the majority of the 25° deflection for more than 3 days. Thus, RQ azo-20-CP2 exhibits shape fixing (Figure 1). To recover its original shape, the cantilever requires a subsequent triggerheating above Tg or by irradiation with circularly polarized/unpolarized light at 442 nm. In contrast, the SQ azo20-CP2 cantilever also immediately bends upon illumination, but to a reduced angle (10°) which is constant despite continuing irradiation. Upon removal of the illumination, the SQ-azo-20-CP2 cantilever returns to the original vertical position in ∼15 min. Thus, the SQ azo-20-CP2 material exhibits shape recovery (elastic) behavior. Recall, glasses are supercooled liquids, formed when the rate of volume reduction upon cooling exceeds the rate of molecular relaxation; therefore, trapping the system at a free energy state greater than the ideal, infinitely relaxed structure (Figure S1).29 As a polymer is cooled through its glass transition temperature, Tg, molecular crowding within the polymer results in a dramatic change in viscosity, accompanied by a reduced molecular mobility. For a given polymer composition and network structure, the rate of temperature change determines the extent to which the glass deviates from the ideal equilibrium upon cooling. The extent of deviation determines the polymer glass morphology and its local distribution of volume, density, and enthalpy. Subsequent relaxation processes at T < Tg, commonly referred to as structural relaxation, physical aging, or thermal annealing, drive this initial distribution of local volume, density, and enthalpy toward that of the lowest free energy amorphous state.30 Heuristically, the breadth of the distribution of relaxation processes, and their temperature dependence, is reflected in the breadth of the glass transition region upon cooling, such as seen in energy loss processes in dynamic mechanical analysis or impedance spectroscopy. Glasses

■

RESULTS AND DISCUSSION To demonstrate the concept that physical aging will alter photomechanical response of a polymer glass, Figure 1 summarizes the radically different responses for cantilevers (20 μm thick) composed from amorphous, azobenzenefunctionalized polyimide (20 mol % azobenzene, referred to herein as azo-20-CP2), cooled from above Tg to room temperature at different rates. The cantilevers were heated to Tg + 50 °C (270 °C) for 30 min under vacuum to thermally erase all prior process history. “Rapid quenching” (RQ) refers to the immediate immersion of the cantilever in liquid nitrogen (−196 °C) while “slow quenching” (SQ) refers to slowly cooling the cantilever in air at a rate of 1 °C/min. Empirically, RQ creates a glass that is substantially farther from ideal equilibrium than SQ. Both samples were irradiated with 100 7529

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

Figure 3. (a, b) Polar plots of the normalized absorption value at 355 nm for the physically aged azo-20-CP2 samples (a, RQ azo-20-CP2; b, SQ azo20-CP2). (●) Before irradiation, (▲) after irradiation with linearly polarized 442 nm light polarized along the y direction (90−270° axis), (▼) after irradiation with linearly polarized 442 nm light polarized along the x direction (0−180° axis), and (■) 4 days after irradiation with linearly polarized 442 nm light polarized along the x direction (0−180° axis). (c) Schematic of polarization direction relative to beam geometry and lab coordinates.

processes. For azo-20-CP2, the activation energy, ΔH, and the breadth of the relaxation process, β (β < 1: broad relaxation distribution; β = 1: narrow relaxation distribution), are 1500 kJ/mol and 0.46, respectively, as determined from excess heat capacity measurements by approaches developed by Tool, Narayanaswamy, and Moynihan (TNM) (Figures S2−S5).27,32 The associated fragility index, m, is ∼150 compared to 11033 for the un-cross-linked parent CP2 polyimide. This represents a highly fragile glass (extreme temperature sensitivity of viscosity and relaxation processes near Tg), most likely reflecting the significantly rigid nature of the backbone and cross-linker. In concert, the relaxation distribution β decreases substantially from the un-cross-linked parent CP2 (0.84) indicating a much broader relaxation spectrum, possibly due to an inhomogeneous network topology arising from the high amount of crosslinks. Together, thermal analysis, X-ray diffraction, and the density measurements confirm that RQ and SQ azo-20-CP2 samples have large differences in the local free volume of the amorphous polymer network morphology. Thus, physical aging does not drive crystallinity or induce anisotropy due to film geometry, but rendered SQ and RQ as equivalent glasses with different local packing densities that enable tuning of the photomechanical response across the entire gamut of shape fixing (RQ) to shape recovering (SQ) through physically aging. However, the question remains, what is the specific correlation between the local glass structure and the isomerization of the azobenzene? Prior work has demonstrated that the rate and efficiency of azobenzene isomerization depend on the dynamics of the molecular environment near the chromophore.34−44 For example, the isomerization of azobenzene in liquid media has been extensively studied34 and is known to depend on a number of factors but principally on the functionalization of the azobenzene chromophore (e.g.,

exhibiting a substantial departure from classic Arrhenius temperature dependence of viscosity are referred to as fragile glasses.31 These characteristics depend on the molecular architecture of the polymer, including chemical composition, molecular rigidity, cross-link density, and molecular weight distributions. Practically then, different glasses, deviating from ideality and thus local molecular structure and dynamics, can be created using different rates of cooling from the polymer melt (T > Tg) or by annealing the polymer for various times at a temperature near, but less than Tg. Thus, for a single chemical composition, a manifold of local molecular environments can be created through a variation of thermal process histories. Differences in the specific heat capacity of the RQ and SQ azo-20-CP2 confirm the different energetic state of the glasses (Figure 2a). SQ azo-20-CP2 exhibits an enthalpic relaxation peak superposed on the step-like glass transition. This excess enthalpy is absent in the RQ azo-20-CP2. The area under the peak reflects the energy the glass gained while it physically aged toward equilibrium during slow cooling. This reflects an increased local packing between polymer chains. Furthermore, the introduction of excess free volume upon rapid quenching is apparent by the lower density of RQ azo-20-CP2 (1.403 ± 0.001 g/cm3) relative to SQ azo-20-CP2 (1.411 ± 0.001 g/ cm3). Wide-angle X-ray diffraction confirmed that although local packing increased, the local molecular order within SQ did not increase relative to RQ (Figure 2b). Note that X-ray also shows that the differences in RQ and SQ azo-20-CP2 do not change after illumination with 442 nm light, indicating that the general energetic state of the glass does not change with photomechanical processes. A priori prediction of the physical state of the glass arising from different thermal aging profiles depends on the activation energy and time constants of the associated relaxation 7530

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

Figure 4. Schematic depiction of the energy landscape framework for glasses.46,47 (a) Notional energetic differences between configurational space of a rapidly quenched glass with greater free volume (left) than a slowly quenched glass (right). Each energy basin contains potential energy wells and barriers reflective of the impact of local reconfiguration of molecular units (open, closed circles). (b) Portion of the energy landscape where 442 nm illumination provides an athermal path to a higher energy well (ΔU) which is recovered via thermal processes (Ua/kT ∼ 1 shape recovery; Ua/kT ≫ 1 shape fixing).

azobenzene, amino-azobenzene, or pseudostilbene)35 and polarity of the medium.36 The efficacy of azobenzene isomerization is considerably reduced by embedding the moiety in a polymeric material, as evident in the work of Morawetz which contrasts the kinetics and concentration of the photostationary state of photoisomerized azobenzene in dilute solution, plasticized, and bulk forms.37 Others, most notably Torkelson and co-workers,38−42 have used azobenzene photochemistry as a probe of local molecular environment and dynamics to examine free volume in polymeric materials including physically aged glasses. Theoretically, the impact of free volume in photochemistry has been discussed recently by Giacometti.43 Weiss presented a discussion and model of the relationship between free volume and photoisomerization, where he describes the volume element surrounding the chromophore as an “effective reaction cavity” with extensive treatment of the impact of so-called “hard” and “soft” walls, free volume, and the nature of azobenzene connectivity (if any).44 Figure 3 summarizes polarized UV−vis spectroscopy23,45−48 of the azo-20-CP2 samples before and after 442 nm illumination. The absorption dipole of the π−π* transition at 355 nm coincides with the long axis of the trans isomer, and the 0−180° axis of the polar plot corresponds to the long direction (length) of the cantilever. The relative absorption of this transition therefore reflects the relative concentration of the trans isomer at different orientations relative to the cantilever axis. Before irradiation (●), the orientational distribution of azobenzene is uniform and independent of process history (RQ or SQ) as shown by the independence of the absorption value on the polarization of the probe light. However, after illumination with 442 nm light (▼) that is linearly polarized parallel to the cantilever length, both the RQ and SQ azo-20CP2 samples exhibit elliptically shaped curves indicative of dichroism. Rotating the polarization of the 442 nm illumination orthogonal to the cantilever axis (90°−270° axis, ▲) results in similar ellipticity that is rotated 90°, confirming that linearly polarized 442 nm light causes a neat enhancement of trans isomer oriented 90° from the polarization direction, irrespective of previous azobenzene distribution or process history. The magnitude of the photoinduced reorientation can be quantified as a dichroic ratio (R = A∥/A⊥) and dye order parameter (S = (R − 1)/(R + 2)), where A∥ and A⊥ refer to the absorption magnitude parallel and perpendicular to the long axis of the cantilever.45−48 The dichroic ratio and dye order parameter are greater for RQ than SQ (RRQ,i = 1.10 > RSQ,i = 1.02; SRQ,i = 0.03

> SSQ,i = 0.007), indicating the excess free volume in RQ enhances the ability for the azobenzene to undergo photoisomerization and thereby generate greater photoinduced dichroism, strain gradient, and cantilever deflection. Note that in both samples the dichroism is accompanied by a similar reduction in the overall trans isomer absorbance (area inscribed by the loop), reflecting the generation of 3−5% cis isomer along with trans isomer reorientation. Upon removal of the illumination (dark), the ellipticity only partially recovers for RQ (Figure 3a, ■) but is fully recovered for SQ (Figure 3b, ■) (RRQ,d = 1.078 > RSQ,d = 1.003; SRQ,d = 0.025 > SSQ,d ∼ 0.001). Relative to the initial state, the relaxed RQ-azo-CP2-20 after illumination (72 h storage in the dark) exhibits a larger absorbance perpendicular to the cantilever axis (90°), reflecting a net increase of azobenzene chromophores in this direction. The population gradient of the aligned transazobenzene chromophores is maintained through the sample thickness, and thus the cantilever remains deflected at an angle toward the illumination source. The increase in ascribed area upon removal of illumination is similar for both SQ and RQ and consistent with the thermal recovery of the comparable concentration of cis isomers in both samples. Also upon removal of illumination, the initial relaxation rate of the photoinduced dichroism is similar for both SQ and RQ (Figure S6). This implies that a fraction of reoriented azobenzenes in RQ occupy similar local environments as the majority of reoriented azobenzenes in SQ. Most importantly, this initial recovery rate of the dichroism is the same as the initial rate of recovery of the photoinduced strain (∼102 s, Figure 1b), further confirming the direct connectivity between molecular reorientation processes and macroscopic strain behavior. With these observations, we can understand the mechanistic differences in transduction of molecular isomerization to macroscopic photomechanical response in these glasses. Following the energy landscape framework for interpreting complex phenomenology in glasses,49,50 one can visualize the impact of physical aging as an alteration of the local transition states and basins in potential energy space as defined by the different states of the glass (i.e., collective configurational coordinates). The initial state of the glass corresponds to an energy basin surrounded by potential energy barriers separating other local energy minima which correspond to alternative collective configurations of the polymer chains as schematically illustrated in Figure 4. Physical aging and the resulting molecular relaxations result in a different collective configura7531

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

recovery. Macroscopically, however, the glass consists of a distribution of local environments, and thus there will be a distribution of local potential energy barriers. The ratio between the amount of initial photostrain that is fixed versus that which is recovered will then depend on the details of the volume-averaged distribution of energy barriers as well as sample temperature. In other words, the amount of recovered strain will depend on the local probability distribution of the lower energy barriers, whereas the initial rate of photostrain recovery will be approximately independent of physical aging. This framework is consistent with the similar initial recovery rates of strain and dichroism in both SQ and RQ (∼102 s, Figure 1 and Figure S6). Furthermore, transients over these barriers will require cooperativity with the local relaxation modes of the network, and thus we would also expect comparable rates between the recovery of dichroism and local molecular dynamics. Prior impedance spectroscopy studies on un-cross-linked CP2 identified two secondary relaxation processes occurring in the glass and corresponding to rotation of pendent trifluoromethyl groups within the anhydride (β′) and rotation of phenyl ether linkages within the diamine (β). At room temperature (25 °C) the lower energy β process occurs on the order of tens of seconds with a breadth of about 50 °C, whereas the higher energy β′ process is effectively frozen (τ ∼ 107 s). The similarities between the rate of rotation of the phenyl ether linkages and reorientation of azobenzene strongly imply that these moieties form the local environment of the photoactive azobenzene.

tional state that in essence moves the glass to a different energy basin within this energy landscape, with a different distribution of local features (potential energy wells and barriers).51 In general, this landscape is rougher, with more features, for a molecularly less crowded local environment, such as for RQ azo-20-CP2, than for a denser, more compact environment, such as for SQ azo-20-CP2.49 Also, the local structure of the glass, outside of the reorientation of a fraction of the azobenzene moieties, does not change during the photodriven processes as experimentally demonstrated above via X-ray and DSC. The photochemical processes can then be visualized as an athermal translation of a fraction of azobenzene moieties within this energy landscape in a manner analogous to photochemical addition processes within a reaction space (Figure 4b). For example, light absorption and associated excited electronic states provide paths to suprafacial [2 + 2] cycloaddition which are thermally disallowed.52 The trans−cis−trans isomerization can be thought of as a similar excited state process enabling athermal transmigration to a neighboring collective configurational state with a nonuniform orientational distribution of chromophores and a higher potential energy (i.e., strain). First consider the initial photoisomerization and associated macroscopic strain generation. The local molecular density will determine the probability, and thereby fraction of the population of azobenzene that can reorient, and thus occupy a neighboring well in the collective configuration state space. A denser local molecular environment (SQ azo-20-CP2) with corresponding deeper, isolated energy wells will reduce the concentration of successful transits to surrounding local minima. The reduced concentration of reoriented azobenzene reduces the local strain generation that summate into the photomechanical response, which is visualized here as a slightly reduced dichroism and steady state bending upon irradiation with linearly polarized 442 nm for SQ azo-20-CP2. This reduced dichroism in the denser glass is in line with prior reports of Torkelson38−42 and Weiss,44,53 who concluded that inadequate free volume reduces or altogether negates the ability of azobenzene to isomerize upon irradiation. Similar conclusions were drawn for the reduction of photomechanical response as crystalline content increased in semicrystalline azobenzne containing polymers.6 Thus, the magnitude of the initial photostrain is related to how the local packing environment of the glass affects the probability of successful trans−cis−trans transit to a neighboring state in the collective configuration space. Perhaps more interesting is the impact of the local potential energy features on the mechanistics of the recovery of photostrain. Upon removal of illumination, thermal processes will drive recovery of the reoriented azobenzene over the local potential energy barriers (Figure 4b). The recovery rate will thus reflect the potential energy barrier between the trans−cis− trans photoinduced state and the initial state of the glass (Ua). For a sufficiently large potential energy barrier, ambient thermal energy will be insufficient and the molecular distortions will be trapped. Thus, the photoinduced strain will be stored and necessitates thermal annealing or additional illumination (e.g., circular polarized 442 nm light) for complete recovery. For the latter, circular polarized illumination will drive an athermal randomization of the azobenzene, effectively eliminating the dichosim and the corresponding strain gradient. Alternatively, for a smaller local barrier that is comparable to ambient thermal energy, the photodriven molecular distortions can relax, and the photoinduced strain will relax, resulting in ambient shape

■

CONCLUSIONS The dependence of the photomechanical response of the azo20-CP2 system on thermal processing confirms that the free volume of the polymer network, generated by controlling the thermal history of the glass, has a profound influence on the orientation and recovery of the azobenzene chromophore and thus the macroscopic photomechanical behavior. Photoinduced dichroism in glasses exhibiting optically fixable shape memory (RQ azo-20-CP2) is stable over many daysconfirming that the reoriented azobenzene units are trapped within the glass, thereby providing the means by which the strain is fixed. In contrast, materials exhibiting shape recovery (SQ azo-20-CP2) are unable to maintain the reoriented azobenzene units and correspondingly the photoinduced strain. This indicates that the local molecular environment within these glasses is not able to inhibit (or trap) the recovery of the orientationally redistributed azobenzene. Furthermore, the recovery of RQ and SQ azo-20-CP2 is qualitatively consistent within an energy landscape framework for interpreting complex phenomenology in glasses. On average, the more feature-rich landscape corresponding to the less dense RQ azo-20-CP2 results in a larger fraction of azobenzene that can reorient, which in turn have a greater probability to be trapped within neighboring potential energy wells. In contrast, isomerization in the denser SQ azo-20-CP2 occurs across a less structured landscape. This results in a smaller fraction of azobenzene that reorient and of these an increased likelihood to occupy shallower local minimum with barriers on the order of kT. Also, as expected for an amorphous solid where there is a broad distribution of local environments, shallow wells exist in both RQ and SQ resulting in similar initial recovery rates, which correspond well to local secondary relaxation modes of the host matrix. In general, the dependence of the photomechanical output of the azo-20-CP2 system on thermal processing has strong 7532

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

(7) van Oosten, C. L.; Corbett, D.; Davies, D.; Warner, M.; Bastiaansen, C. W. M.; Broer, D. J. Macromolecules 2008, 41 (22), 8592−8596. (8) Kondo, M.; Miyasato, R.; Naka, Y.; Mamiya, J.-i.; Kinoshita, M.; Yu, Y.; Barrett, C. J.; Ikeda, T. Liq. Cryst. 2009, 36 (10−11), 1289− 1293. (9) Lee, K. M.; Koerner, H.; Vaia, R. A.; Bunning, T. J.; White, T. J. Macromolecules 2010, 43 (19), 8185−8190. (10) Koerner, H.; White, T. J.; Tabiryan, N. V.; Bunning, T. J.; Vaia, R. A. Mater. Today 2008, 11 (7−8), 34−42. (11) Corbett, D.; Warner, M. Liq. Cryst. 2009, 36 (10−11), 1263− 1280. (12) Cviklinski, J.; Tajbakhsh, A. R.; Terentjev, E. M. Eur. Phys. J. E 2002, 9 (5), 427−434. (13) Hogan, P. M.; Tajbakhsh, A. R.; Terentjev, E. M. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys. 2002, 65 (4−1), 041720/1−041720/ 10. (14) Lee, K. M.; Smith, M. L.; Koerner, H.; Tabiryan, N.; Vaia, R. A.; Bunning, T. J.; White, T. J. Adv. Funct. Mater. 2011, 15, 2913. (15) Hugel, T.; Holland, N. B.; Cattani, A.; Moroder, L.; Seitz, M.; Gaub, H. E. Science 2002, 296 (May), 1103−1106. (16) Viswanathan, N. K.; Kim, D. U.; Bian, S.; Williams, J.; Liu, W.; Li, L.; Samuelson, L.; Kumar, J.; Tripathy, S. K. J. Mater. Chem. 1999, 9, 1941−1955. (17) Natansohn, A.; Rochon, P. Chem. Rev. 2002, 102 (11), 4139− 4175. (18) Lee, K. M.; Koerner, H.; Vaia, R. A.; Bunning, T. J.; White, T. J. Soft Matter 2011, 7, 4318−4324. (19) White, T. J.; Tabiryan, N.; Tondiglia, V. P.; Serak, S.; Hrozhyk, U.; Vaia, R. A.; Bunning, T. J. Soft Matter 2008, 4, 1796−1798. (20) Hrozhyk, U.; Serak, S.; Tabiryan, N.; White, T. J.; Bunning, T. J. Opt. Express 2009, 17 (2), 716−722. (21) White, T. J.; Serak, S. V.; Tabiryan, N. V.; Vaia, R. A.; Bunning, T. J. J. Mater. Chem. 2009, 19 (8), 1080−1085. (22) Serak, S.; Tabiryan, N.; White, T. J.; Vaia, R. A.; Bunning, T. J. Soft Matter 2010, 6, 779−783. (23) Lee, K. M.; Tabiryan, N. V.; Bunning, T. J.; White, T. J. J. Mater. Chem. 2012, 22, 691−698. (24) Tabiryan, N.; Serak, S.; Dai, X.-M.; Bunning, T. Opt. Express 2005, 13, 7442−7448. (25) Engels, T. A. P.; van Breemena, L. C. A.; Govaert, L. E.; Meijer, H. E. H. Polymer 2011, 52 (8), 1811−1818. (26) Van Oosten, C. L.; Harris, K. D.; Bastiaansen, C. W. M.; Broer, D. J. Eur. Phys. J. E 2007, 23 (3), 329−336. (27) Hodge, I. M. J. Non-Cryst. Solids 1994, 169, 211−215. (28) Andreozzi, L.; M, F.; Giordano, M.; Palazzuoli, D. Macromolecules 2002, 35, 9049−9056. (29) Hutchinson, J. M. Prog. Polym. Sci. 1995, 20 (4), 703−760. (30) Hodge, I. M.; Berens, A. R. Macromolecules 1982, 15, 762. (31) Martinez, L.-M.; Angell, C. A. Nature 2001, 410, 663−666. (32) Svoboda, R.; Honcová, P.; Málek, J. J. Non-Cryst Solids 2012, 358, 804−808. (33) Jacobs, D. J.; Arlen, M. J.; Wang, D. H.; Ounaies, Z.; Berry, R.; Tan, L.-S.; Garrett, P. H.; Vaia, R. A. Polymer 2010, 51 (14), 3139− 3146. (34) Tiberio, G.; Muccioli, L.; Berardi, R.; Zannoni, C. ChemPhysChem 2010, 11 (5), 1018−1028. (35) Ikeda, T.; Zhao, Y. Smart Light-Responsive Materials: AzobenzeneContaining Polymers and Liquid Crystals; Wiley: Hoboken, NJ, 2009; p 514. (36) Serra, F.; Terentjev, E. M. Macromolecules 2008, 41 (3), 981− 986. (37) Paik, C. S.; Morawetz, H. Macromolecules 1972, 5 (2), 171−177. (38) Royal, J. S.; Torkelson, J. M. Macromolecules 1992, 25 (18), 4792−6. (39) Royal, J. S.; Victor, J. G.; Torkelson, J. M. Macromolecules 1992, 25 (2), 729−34. (40) Victor, J. G.; Torkelson, J. M. Macromolecules 1988, 21 (12), 3490−7.

implications to the development of photoresponsive materials, as it illustrates the importance of thermal history, local molecular packing, and secondary relaxation processes on the mechanical response. Direct comparisons of mechanical output of photomechanical materials are valid only when the differences in process history, whether thermal, solvent, or cross-linking, are removed such as via a standardized thermal erasure. The demonstrated importance of small changes in molecular packing on the magnitude of generated photostrain and the recovery process reveal new molecular design considerations. For example, what is the ideal composition of matrix and network to tune local free volume, control trap density, and couple recovery rate to chain relaxation modes. Toward these ends, numerous issues are critical to understand, including a quantitative coupling between the distribution of local environments in the glass and the magnitude and recovery rates of the photoinduced strain as well as the correspondence between temperature, strain recovery, and secondary relaxation processes in the glass. Another key problem to understand is the impact of molecular and mesoscopic (domain) orientational order, such as found in nematic glasses and uniaxially deformed amorphous polymers, on local packing and these processes. The resulting deeper connections between glass physics, photomechanical response, and mechanics will greatly improve the development of physics-based design tools and optimization of material and device architectures, thereby facilitating the use of these novel materials in medical, optics, and aerospace technologies.

■

ASSOCIATED CONTENT

S Supporting Information *

Experimental and analysis details of the physical aging behavior of the materials discussed. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was completed at Air Force Research Laboratory (AFRL) at Wright Patterson Air Force Base with funding from Materials and Manufacturing Directorate as well as the Air Force Office of Scientific Research. We thank Ian Hodge of Rochester Institute of Technology for providing his MATLAB code for physical aging analysis.

■

REFERENCES

(1) Lovrien, R. Proc. Natl. Acad. Sci. U. S. A. 1967, 57 (2), 236−242. (2) Agolini, F.; Gay, F. P. Macromolecules 1970, 3 (3), 349−351. (3) Kondo, M.; Sugimoto, M.; Yamada, M.; Naka, Y.; Mamiya, J.-i.; Kinoshita, M.; Shishido, A.; Yu, Y.; Ikeda, T. J. Mater. Chem. 2010, 20 (1), 117−122. (4) Finkelmann, H.; Nishikawa, E.; Pereira, G. G.; Warner, M. Phys. Rev. Lett. 2001, 87 (1), 015501/1−015501/4. (5) Wang, D. H.; Lee, K. M.; Yu, Z.; Koerner, H.; Vaia, R. A.; White, T. J.; Tan, L.-S. Macromolecules 2011, 44, 3840−3846. (6) Lee, K. M.; Wang, D. H.; Koerner, H.; Vaia, R. A.; Tan, L.-S.; White, T. J. Angew. Chem., Int. Ed. 2012, 124, 4193−4197. 7533

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534

Macromolecules

Article

(41) Victor, J. G.; Torkelson, J. M. Macromolecules 1987, 20 (11), 2951−4. (42) Victor, J. G.; Torkelson, J. M. Macromolecules 1987, 20 (9), 2241−50. (43) Dall’Agnol, F. F.; Oliveira, O. N., Jr.; Giacometti, J. A. Macromolecules 2006, 39 (14), 4914−4919. (44) Weiss, R. G.; Ramamurthy, V.; Hammond, G. S. Acc. Chem. Res. 1993, 26 (10), 530−6. (45) Wu, Y.; Demachi, Y.; Tsutsumi, O.; Kanazawa, A.; Shiono, T.; Ikeda, T. Macromolecules 1998, 31 (2), 349−354. (46) Wu, Y.; Demachi, Y.; Tsutsumi, O.; Kanazawa, A.; Shiono, T.; Ikeda, T. Macromolecules 1998, 31 (14), 4457−4463. (47) Wu, Y.; Kanazawa, A.; Shiono, T.; Ikeda, T.; Zhang, Q. Polymer 1999, 40 (17), 4787−4793. (48) Wu, Y.; Mamiya, J.-i.; Kanazawa, A.; Shiono, T.; Ikeda, T.; Zhang, Q. Macromolecules 1999, 32 (26), 8829−8835. (49) Debenedetti, P. G.; Stillinger, F. H. Nature 2001, 410 (6825), 259−267. (50) Angell, C. A.; Ngai, K. L.; McKenna, G. B.; McMillan, P. F.; Martin, S. W. J. Appl. Phys. 2000, 88, 3113−3166. (51) Heuer, A. J. Phys.: Condens. Matter 2008, 20, 373101. (52) Hein, S. M. J. Chem. Educ. 2006, 83 (6), 940−945. (53) Wang, C.; Weiss, R. G. Macromolecules 2003, 36 (11), 3833− 3840. (54) Narayanaswamy, O. S. J. Am. Ceram. Soc. 1971, 54, 491. (55) Moynihan, C. T.; Easteal, A. J.; DeBolt, M. A.; Tucker, J. J. Am. Ceram. Soc. 1976, 59, 12. (56) Kohlrausch, F. Ann. Phys. Chem. 1866, 128, 1. (57) Williams, G.; Watts, D. C. Trans. Faraday Soc. 1970, 66, 80. (58) Cowie, J. M. G.; Ferguson, R. Polymer 1993, 34, 2135. (59) Bohmer, R. J. Non-Cryst. Solids 1994, 172, 628.

7534

dx.doi.org/10.1021/ma3016085 | Macromolecules 2012, 45, 7527−7534