Light-Driven Reversible Shaping of Individual Azopolymeric Micro-Pillars

Azopolymers are known to exhibit a strong light responsivity known as athermal photofluidization. Although the underlying physics is still under debate, athermal photofluidization has been demonstrated to trigger mass-migration according to the polarization of a proper illumination light. Here, a polymer blend is proposed wherein a commercial azo-polyelectrolyte is mixed with a passive polymer. The blend is patterned as an array of micro-pillars that are individually exposed to visible laser illumination. Thanks to the interplay between the two blend components, a reversible and controlled deformation of the micro-pillars by periodically tuning the laser polarization in time is demonstrated. A reduced mobility of the azo-compound allows to repeatibly elongate and rotate micro-pillars along specific directions, with no significant material flow outisde the initial volume and no significant degradation of the structure morphology over several cycles. The proposed work suggests new degrees of freedom in controlling the mechanical features of micro-patterned light-responsive materials that can be usefully exploited in many application fields.

is rotated by 90 degrees with respect to a previous irradiation. A typical result is shown in Fig. 2c. Despite some detrimental effects on the morphology quality, a substantially circular cross section is restored by triggering a laser-induced stretching along the minor axis of the elongated pillars. A similar result has been recently found on the light-tuning of the pore shape in azopolymeric breath figures 27 . The reversible elongation of the pillars can be iterated several times, with a slow progressive degradation of the structures. For a quantitative analysis, in Fig. 2d a mean roundness parameter is plotted against time, as a polarization-variable laser illumination is performed. The mean roundness is evaluated time by time as the average of the ratio Am(t)/AM(t) over 20 micro-pillars included in the optical images that are collected, binarized and analyzed as described above. In Fig. 2e-i, exemplary binarized images are shown corresponding to relevant times during the light-induced modification of the pillars. More specifically, Fig. 2e is related to the initial pillar state (such as in Fig. 2a), Fig. 2f,h are related to an elongated state (such as in Fig. 2b) and Fig. 2g,i are related to a restored circular state (such as in Fig. 2c). In Fig. 2d we observe that elongation-restoration cycles can be repeated upon irradiation with the very same laser power, by simply rotating the incident polarization by 90 degrees, from a vertical to a horizontal orientation and vice-versa. In time cycles lasting about 400 s, almost circular pillars (mean roundness about 0.9, as estimated from Fig. 2e) are elongated along the vertical direction (mean roundness down to 0.6, as estimated from Fig. 2f,h) and then restored back to a circular shape (Fig. 2g,i). This reversible pillar modification is advantageously promoted by an improved stiffness of the polymeric mixture due to the presence of a stabilizing PMMA component, as suggested elsewhere in other co-polymer matrices containing azobenzenes 5 . In this way, the mass migration triggered by photofluidization is prevented to flow outside the pillar volume even without external constraints 29 .
Worth to note here that illuminated PAZO-only micro-pillars are prone to quickly lose their regular shape because of the strong mass diffusion away from the volume initially occupied in their as-fabricated state. For comparison purposes, a SEM picture of all-PAZO pillars before and after an overall irradiation time of 400 s is shown in Fig. 3a,b. The softening of PAZO pillars due to the athermal photofluidization hampers the full recovery of a regular pillar shape because of the progressive leakage of polymer out of the initial confined volume. The elongation-restoration mechanism can be hardly repeatable over many cycles, as demonstrated by the time-dependent mean roundness plot shown in Fig. 3c. In fact, when several elongation-restoration steps are performed in sequence, the PAZO pillars get completely melted and the pillar cross section cannot be optically recognized anymore.
The stretching/shrinking effect described above can be extended for longer time periods. In such a case, a new deformation of the micro-pillars can occur, eventually along a different orientation as determined by the illumination polarization. Figure 4a,b illustrates the time-evolution of the mean roundness and the mean orientation angle θ (t) calculated over 30 micro-pillars exposed to a laser beam sequentially switching between two orthogonal ; (e-i) binarized optical images of micro-pillars at relevant times during pillar light-modification. An accelerated (30x) live sequence of the two-cycle elongation-restoration process depicted here is provided in movies S1 (raw frames) and S2 (binarized frames) in the Supporting Information.
Scientific RepoRts | 6:31702 | DOI: 10.1038/srep31702 linear polarizations. During the first 100 s, as-fabricated circular micro-pillars ( Fig. 4c) are illuminated with an H-polarized beam such that a horizontal elongation is produced, corresponding to a mean roundness of about 0.7 (estimated from Fig. 4d). When the polarization is rotated by 90 degrees (V-polarization), the initial circular shape is restored (mean roundness 0.9) in 130 s irradiation (estimated from Fig. 4e). However, if the V-polarized illumination is kept for additional time, the mean roundness parameter starts to decrease again, the elongation occurring along a different direction. In fact, from 100 s to 400 s, the mean orientation angle θ experiences a rather quick change from 0 degrees (horizontally oriented micro-pillars) to nearly − 70 degrees (almost vertically oriented micro-pillars). After 700 s, the micro-pillars are well elongated (roundness 0.5) and flipped as shown in Fig. 4f. Afterwards, upon a further switch to H-polarization, a restoring force is triggered, such that the roundness parameter starts to increase again. Worth to note here that the maximum orientation change we could induce on already elongated pillars with a given orientation has been generally limited to angles smaller than 90 degrees only. In the specific case illustrated in Fig. 4, after the first horizontal elongation, pillars are hampered to fully orient in a vertical direction, despite a vertically oriented polarization has been used for illumination. As an explanation, we speculate that some degree of polarization-induced irreversible plasticity within the PAZO-PMMA matrix might occur during the very first exposure 14 . In alternative, a phase separation between PMMA and PAZO due to the light-induced diffusion might constrain the mass-migration mobility 31 .
By rotating the polarization of the incident laser by angular steps smaller than 90 degrees, it is possible to induce micro-pillars to smoothly orient accordingly, while still keeping a given elongation. For example, in Fig. 5a, elongated pillars are observed to maintain a roughly constant (mean) roundness value during a 400 s illumination wherein the polarization is rotated by 45 degrees each 100 s (green arrows in the figure). In this case, light-induced deformations are provided such that an overall rotation of the elongated pillars directly occurs without an intermediate restore of the initial circular shape. As shown in Fig. 5b, the mean orientation angle suggests that pillars are substantially experiencing a smooth rotation of about 80 degrees according to the instantaneous polarization direction (with some delay due to the viscosity of the polymer). In order to better illustrate this effect, starting from the estimated mean values for A m (t), A M (t) and θ (t) in each time frame, the trajectories defined by the ray vectors associated to the pillar minor axis and major axis are traced on an xy-plane that ideally contains the pillar top surface (black circles for the minor axis and orange circles for the major axis in Fig. 5c). As a result, a 'mean' elliptical pillar can be drawn having a minor axis and a major axis oriented according to the instantaneous value of θ (t). Such a graphical representation of the 'mean' elliptical pillar in time provides an intuitive description of the substantially rigid rotation experienced during the polarization changes. A movie of the full rotation occurring in the 400 s time interval is shown as video S5 in Supporting Information.
The fine control on the elongation and orientation of PAZO-PMMA structures demonstrated above can also allow a tuning of the surface pattern on an individual-pillar basis. In fact, if illumination is provided as a localized light spot (e.g. by means of a a focusing system), micro-pillars can be sequentially modified according to the desired polarization state. In Fig. 5d, a SEM image of several pillars sequentially elongated in different directions is illustrated.

Conclusion
In conclusion, we demonstrated that the mixture of an azopolymer with a passive component such as PMMA can improve the mechanical characteristics of individual micro-objects suitable for deformation using light. As an exemplary micro-structure, we considered micrometer-sized pillars that can be reversibly elongated and rotated with a rather good degree of control, upon optimal choice of the radiation wavelength and polarization. While micro-pillars have been chosen in order to allow a quick and statistically meaningful optical characterization, an analogous mechanical light-responsivity is expected to occur also in nano-sized objects. The findings here presented are promising in useful applications such as tunable smart architectures including, e.g. photonic crystals and optical metamaterials 32,33 , and dynamically wettable surfaces [34][35][36] .

Experimental Methods
Sample Fabrication. A mixture of commercially available poly [1-[4-(3-carboxy-4-hydroxyphenyl-azo)benzene sulfonamido]-1,2-ethanediyl, sodium salt] (PAZO) and (poly(methyl methacrylate) (PMMA), (both from Sigma-Aldrich) is prepared as follows. A solution of PMMA with molecular weight (Mw) of ~15000 g/mol. is dissolved in N,N-Dimethylformamide at a concentration of 4 wt%. PAZO dissolved in methanol (concentration 25 mg/mL) is added to the PMMA solution, at 10:1 ratio and then mechanically stirred and sonicated for few minutes until a homogeneous blend is formed. An amount of 20 μ l of PAZO-PMMA is casted onto a glass substrate, previously washed in acetone, rinsed with isopropanol and dried with a nitrogen flow.
An elastomeric stamp, fabricated using conventional soft lithography molding techiniques, is used to pattern a centimetric area of micro-pillar array. The PDMS stamp is prepared by mixing Sylgard 184 Dow Corning elastomer with the curing agent at a 10:1 ratio. The mixture is then casted onto a silicon master whose surface has been patterned with the complementary structures, subsequently degassed in a dessicator and left to cure at 60 °C for 2 hours. The prepared PDMS stamp is soaked in ethanol for few minutes and then gently pressed on the PAZO-PMMA film on glass.
The sample is placed in a vacuum dessicator for 2 hours in order to enhance patterning and let the polymer blend to fill in the PDMS template. As a next step, the sample is dried in oven, until the solvent is completely evaporated (3 hours at 60 °C) before the final removal of the PDMS stamp. An array of well-defined azo-polymeric micro-pillars over a large area is then obtained.
Optical Conditioning and Monitoring. A slightly divergent pulsed laser beam (10 ps pulse duration, wavelength λ = 490 nm) with an initial circular polarization is polarization-filtered through a polarizer. As the polarizer can be rotated about the optical axis, the resulting beam can be linearly polarized along any arbitrarily controlled polarization direction. A low Numerical Aperture (NA = 0.2) lens focuses the laser beam onto the back of the fabricated sample. A laser power of 1.3 mW covers a homogeneous circular illumination area with radius about 25 μ m. The micro-pillar surface is monitored by means of a wide-field white-light microscope in reflection mode, employing a 100x, NA = 0.95 objective (Olympus). A computer-controlled 24 bit CCD (Apogee Ascent) records the white light images after proper laser-filtering (Semrock RazoEdge MaxLine 532). The CCD is operated in such a way that optical images are collected at a frame rate of 1 Hz. An on line image processing code takes the raw data from the CCD, performs a proper thresholding, then pixel binarization and pattern recognition operations of unconnected areas. As a result, a set of generally elliptical individual objects is recognized from each image. Some geometrical parameters, including the major axis A M , the minor axis A m and the orientation θ of the ellipses, where θ is the angle between the major axis and a horizontal axis, are extracted for all individual objects in each image and made available for further analysis. In addition, mean values are calculated in order to add significance and robustness to the overall description of the collective shape modification involving a plurality of pillars.