1. INTRODUCTION

Birefringence, an optical property of numerous materials, is of fundamental importance in the design of various laser systems to control the polarization of a light beam. Although first discovered in crystalline solids, birefringence is not a property exclusive of crystals. Various materials exhibit form birefringence, direction-dependent refractive index due to their organized microstructure representing an interesting alternative to traditional materials that possess bulk birefringence properties [1–6]. Low-loss transmissive nanostructured wave plates fabricated by glancing-angle deposition (GLAD) made from oxides such as ${{\rm SiO}_2}$ and MgO have been recently demonstrated [7–9]. A key advantage of these materials is their scalability to large-size optical elements with a relatively low fabrication cost and high surface-quality uniformity, resulting in excellent optical performance.

Polarization control optics used in current-generation high-power laser systems are typically based on the birefringent properties of bulk materials, which remain homogenous as a function of position. However, spatial control of optical birefringence has become a highly desirable and increasingly important capability to enable a new generation of optics for controlling light propagation and its properties [10,11]. The latter requires the development of wave plates showcasing spatial variation of their birefringence in the 10 mm spatial scale [12]. Current-generation polarization control optics that can provide spatially tailored polarization control are based on liquid crystal devices, which are arguably not suitable for high-power laser applications in the ultraviolet spectral region [13,14]. Alternative technologies such as metasurfaces are an attractive candidate for complex manipulation of light; however, large-scale fabrication remains a challenge [15–17]. More traditional approaches have been tested for post-deposition modification of optical properties. Among these are annealing of chiral sculptured films leading to homogeneous blueshifting of their spectral characteristics [18]. Another example is all-silica nanostructured wave plates patterned by photolithography to provide spatial variation of the retardance [8]. This middle ground is very promising; however, a methodology enabling the realization of complex patterns in GLAD wave plates has not yet been achieved.

Development of techniques to spatially tailor the birefringence in nanostructured wave plates based on large-bandgap dielectric materials can address the low-performance characteristics of liquid crystal materials, enabling the fabrication of optics for complex control of polarization in high-power lasers including those in the ultraviolet spectral region. Such advanced optics can be utilized for improved beam shaping [19], distributed phase rotators [20], light valves, apodizers [21], beam spot blockers [22,23], and other applications. When utilized in inertial confinement high-power laser systems, these UV-compatible optics have the potential to enable new ways to deliver laser power onto a target more effectively, minimizing adverse laser–plasma interactions.

In this work, we demonstrate direct-write patterning of form birefringence in all-silica nanostructured coatings, utilizing the energy of a ${{\rm CO}_2}$ laser to achieve spatial modification and control of the induced birefringence change. Laser irradiation assisted in localized melting of nanostructured coatings, subsequently modifying the microstructure and altering the local retardance. We show that this process is only successful when performed under high vacuum to avoid trapping air within the coating. The change in birefringence was evidenced through in-line monitoring of the process with a polarization-sensitive camera as well as off-line traditional Mueller matrix polarimetry. The change in microstructure was observed and analyzed through electron microscopy.

 

Fig. 1. Schematic description of the coating used for this work overlaid on scanning electron micrographs illustrating the essential role of processing in a vacuum environment to avoid trapping air within the coating. Cross-section images of processed coatings were taken at the center of irradiation.

Download Full Size | PDF

2. MATERIALS AND METHODS

A. Coating Fabrication

Nanostructured coatings are deposited at a substrate-source angle ($\alpha $) that deviates from zero achieved by tilting the substrate. Initially, $\alpha \ge {85}^\circ$ was considered glancing angle while $\alpha \lt {85}^\circ$ was considered oblique incidence. However, these terms have been used interchangeably more recently. Here, we will refer to our coatings as GLAD which is the term most commonly used in our subfield of optics. ${{\rm SiO}_2}$ has been the material of choice in near-ultraviolet high-power laser systems, and we thus focus this work on ${{\rm SiO}_2}$ GLAD coatings. The all-silica coatings were fabricated through glancing-angle serial bideposition following the methodology reported by this group elsewhere [8]. In short, clean fused-silica substrates were mounted onto a custom Angstrom Engineering GLAD stage inside an e-beam deposition chamber. The chamber was evacuated overnight and heated to 25°C with the use of quartz lamps. The base pressure was better than ${1} \times {{10}^{- 6}}\;{\rm Torr}$ and deposition was carried out without the addition of reactive gases. The deposition rate was controlled to 9 Å/s using feedback from a quartz crystal microbalance. Figure 1 shows a schematic representation of the cross section, annotated with the deposition angles as well as the measured effective refractive indices [24]. The number and thickness of birefringent layers determines the overall retardance of the wave plate. The coating used in the present work consisted of 24 alternating layers deposited at 0° (normal incidence) and 73°, terminated with an antireflection layer deposited at 82°. This quarter-wave plate stack was designed to be used for the ultraviolet (351 nm) in a vacuum environment and has documented low-loss, high laser-induced damage threshold, and a wide design bandwidth [8].

B. Laser-Assisted Processing

The processing experimental system consisted of five main components, including (1) a ${{\rm CO}_2}$ laser with exposure time and power control, (2) a collimated UV LED source equipped with a polarizer, (3) a motorized stage with XYZ control, (4) a polarization-sensitive camera for in-line birefringence monitoring of the exposed area of the sample, and (5) a vacuum chamber to allow for processing under vacuum at a pressure ${\lt}{2.5} \times {{10}^{- 5}}\;{\rm Torr}$. Figure 2(a) presents a schematic representation of the experimental system. The ${{\rm CO}_2}$ laser used has a wavelength of 10.6 µm (SYNRAD, Firestar ti–series) that produces a circular beam with linear polarization. The laser beam was directed through a pinhole with a 2.7 mm diameter aperture before reaching the samples. A fiber-coupled LED operating at a wavelength of 365 nm complemented by a wire grid linear polarizer was used for back-illumination of the sample to support imaging that could detect retardance modification in real time using a polarization-sensitive (PS) camera (Thorlabs, CS505MUP1). The PS camera is equipped with a polarized sensor (5.0 MP, monochrome complementary metal-oxide semiconductor) with a wire grid polarizer array comprised of a repeating pattern of polarizers (0°, 45°, ${-}{45}^\circ$, and 90° transmission axes) placed on-chip to form an image array of four pixel blocks constituting each calculation unit.

 

Fig. 2. Experimental setup used in this work: (a) schematic representation of the main parts comprising the processing table; (b) example images taken with the in-line polarization-sensitive camera allowing for birefringence tracking during irradiation.

Download Full Size | PDF

The image at the sample plane generated under illumination with the polarized LED light propagating through the sample was captured by the PS camera equipped with a suitable lens and after passing through a 400 nm short-pass filter (UG11) to suppress thermal emission generated on the sample during laser irradiation. This experimental arrangement enabled in-line tracking of the changes in the propagated light polarization state arising from birefringence modification of the coating during processing. An example of the in-line birefringence tracking is shown in Fig. 2(b) from a coating irradiated for up to 40 s. These representative images, captured by the PS camera, show only the 90° transmission axis image component of the PS camera under linearly polarized at 0° LED illumination and the fast axis of the sample placed at 45°. As a result, the complete loss in retarding power causes the light from the source to pass through the optic unchanged (thus remaining polarized at 0°) and the 90° transmission image component capturing this modification as a feature that shows complete loss of any detected light. The in-line imaging capability facilitated rapid testing at various exposure conditions toward the development of an efficient processing protocol.

Experiments were initially performed in ambient air, but the results indicated it is necessary that the process be carried out in a vacuum environment to avoid trapping air into voids within the coating. This is demonstrated by the cross-section scanning electron micrographs taken from two coatings processed either in air or in vacuum shown in Fig. 1. The micrographs were taken from the center of the irradiated areas and demonstrate the significant difference in morphology. The coating processed in air contains numerous voids, which contribute to excessive light scattering. The coating processed in vacuum resulted in a full-density film virtually indistinguishable from the substrate. Prior to processing, the sample was held in the vacuum chamber at ${\lt}{2.5} \times {{10}^{- 5}}\;{\rm Torr}$ for approximately 2 days to ensure removal of the air within the GLAD structure.

C. Characterization

The change in birefringence was further characterized after exposure using the polarization-sensitive camera discussed in the previous section. The camera is able to probe the first three Stokes parameters. In addition, Mueller matrix polarimetry was carried out with an Exicor 450XT from Hinds Instruments capable of measuring the full Mueller matrix at a wavelength of 355 nm with a repeatability of ${3}\sigma \lt {0.03}\;{\rm nm}$ or 1% and a noise floor of ${3}\sigma \lt {0.03}$. Maps were taken with a 1 mm diameter beam size scanned across the area of interest along a 0.5 mm square grid to increase area coverage. Average retardance values for locations of interest were taken from a ${1.5} \times {1.5}\;{\rm mm}$ area, and error was computed as the standard deviation. Finally, the microstructure of the film as a function of the position was evaluated through scanning electron microscopy for which the samples were coated with a thin metallic layer to avoid charging effects.

D. Modeling

Computational modeling was performed to inform and support observations including a 1-D analytical model to understand the effects of the thermal properties on the temperature profile and temperature variation within the multilayer. Furthermore a 2-D numerical model was employed to study the effects of temperature-dependent absorption depth, thermal conductivity, and specific heat. The most important results of the theoretical work have been included in the main text; the detailed modeling work has been included as a Supplement 1.

3. RESULTS

A. Laser-Assisted Modification of Form Birefringence

The ability to monitor in real time the laser-induced modification enabled rapid evaluation of the required laser fluence and/or exposure time. The representative results presented in Fig. 3 were obtained with a single irradiation performed by ramping the laser power up to 27 W. The microstructure was observed at different points within the irradiated spot and cross-section micrographs were recorded. These results present evidence of the gradual loss of microstructure as a function of position relative to the laser beam as can be seen from the lack of features at the center of the spot where the coating is indistinguishable from the substrate. It is noteworthy that the loss in microstructure due to melting does not occur layer by layer. The latter is indicated by the identifiable presence of the 25 layers even when the microstructure has almost vanished close to the center of the irradiated spot. A 30% decrease in thickness was estimated between the panels (a) and (c) in Fig. 3.

 

Fig. 3. Modified microstructure resulting from ramping up the laser power to 27 W as shown from cross-section scanning electron micrographs. The images from left to right are representative of the microstructure when moving toward the center of the spot.

Download Full Size | PDF

Form birefringence arises from anisotropy in the microstructure, which is composed of vertical columns surrounded by elongated voids oriented in a direction perpendicular to the fast axis. Figure 4(a) corresponds to the retardance map representative of the characteristic pattern arising from localized melting. The pattern includes (1) close to zero retardance in the middle, (2) decreased retardance in a direction parallel to the fast axis, and (3) increased retardance in a direction perpendicular to the fast axis. The area of increased retardance indicates that the void fraction increases preferentially in that direction (perpendicular to the fast axis), which has the concomitant effect of increasing the form birefringence. Figure 4(b) provides top-view scanning electron micrographs taken from a representative coating, demonstrating the anisotropic melting in the area outside of the center spot. Overall, the retardance change outside the central irradiated area resulting from material redistribution is direction-dependent and defined by the microstructure.

 

Fig. 4. Effect of anisotropic microstructure on laser-assisted melting. (a) Modified retardance resulting from ramping up the laser power to 27 W. The size of the irradiated area has been marked with a 2.7 mm diameter circle, corresponding to the pinhole used in the system. (b) Top-view scanning electron micrographs taken from a single-layer coating before and after irradiation showing increased void fraction in a direction perpendicular to the fast axis.

Download Full Size | PDF

B. Effect of Laser-Irradiation Conditions

Two specific tests conducted as part of this work are discussed in order to understand the effect of laser irradiation on the modified retardance of the coating. Several areas were irradiated on a single GLAD wave plate, carefully spaced to avoid merging. The experiment was performed in one session and the sample was promptly loaded into the birefringence mapper for analysis; therefore, increasing confidence in the changes observed and the acquired data. First, the laser power was varied to a set point in the range of 20.5 to 33 W (358 to ${576}\;{{\rm W/cm}^2}$ estimated average value at the sample by considering the size of the pinhole used, 2.7 mm in diameter) while maintaining a constant 5 s irradiation time at the desired set point. For this experiment, the power was increased at a rate of ${\sim}{1}\;{\rm W/s}$ to reach the prescribed set point. In a second test the laser power, without a ramp, was set constant at 27 W, while the irradiation time was increased from 10 to 40 s. Similar results were obtained, indicating that the laser-assisted process can be controlled by either power or time.

1. Increasing Laser Power

Figure 5(a) plots the average retardance at the center of irradiation as a function of laser power and selected retardance maps have been included at the top. Overall, we observed an initial retardance increase at the lower power settings tested before a steady decrease was recorded at higher power. Specifically, an initial retardance increase to 123.5 nm resulted from irradiating at 20.5 W, while a decrease to $({\le} {18.3}\;{\rm nm}$ was obtained for laser power ${\ge} {30}\;{\rm W}$. The initial 19.9% retardance increase can be attributed to densification of the GLAD columnar microstructure, which leads to increased void fraction and improved form birefringence. However, irradiating at laser power ${\gt}{27}\;{\rm W}$ results in rapid change in retardance with an 82.2% decrease due to localized melting, resulting in the loss of microstructure and form birefringence.

 

Fig. 5. Modified retardance upon varying the irradiation conditions: (a) the effect of increasing laser power while keeping irradiation time constant at 5 s; and (b) the effect of increasing the irradiation time while keeping the laser power constant at 27 W. Average retardance values were taken from a ${1.5 \times {1.5}\;{\rm mm}}$ area, error was computed as the standard deviation.

Download Full Size | PDF

2. Increasing Irradiation Time

Similar results were obtained when increasing irradiation time at a constant laser power of 27 W. The results obtained from this test are presented in Fig. 5(b) where the retardance at the center as well as parallel and perpendicular to the fast axis have been plotted; retardance maps have been included at the top. The center and parallel retardance follow a similar trend toward zero, while the perpendicular retardance increases. An average maximum retardance value of 173.1 nm was obtained (perpendicular to the fast axis) for an irradiation time of 40 s. The retardance pattern observed is dictated by the microstructure as explained in Section 3.A.

C. Modeling

The 1-D analytical model established indicated that the temperature at the area of the surface irradiated by the laser beam can exceed the melting temperature with only a small gradient of temperature within the underlying GLAD multilayer (axial direction). This explains the observed uniform melting of the GLAD at the location of beam irradiation. The 1-D model also established the importance of thermal conductivity, and its increase with temperature, as important film material properties.

Our 2-D numerical model assumed a Gaussian beam and a GLAD multilayer on a 50 mm diameter substrate with a thickness of 3 mm. The thermal conductivity and specific heat used for the GLAD coating were adjusted by considering the porosity of its layers. The findings obtained from the 2-D numerical model generally agreed well with the experimental results presented in previous sections. The modeling results further indicated that although power and irradiation time can be used as tuning parameters, controlling the laser power is more efficient in producing the desired pattern. The numerical simulations for various conditions ultimately showed a linear dependence of the maximum temperature as function of laser power, despite the nonlinear dependence of thermal properties on temperature. The spatial details of thermal absorption near the surface were not found to be important. Furthermore, our numerical models revealed two distinct heat transfer modes affecting the GLAD structure during irradiation. For short times (${\sim}{2.2}\;{\rm s}$), heat transfers primarily in the axial direction, through the thickness of the sample. For longer times, the heat flux occurs also in the radial direction. A detailed description of the modeling performed for the irradiated GLAD coatings has been included in the Supplement 1. Altogether, experimental and theoretical data demonstrated that the deposited laser power is the main driver of temperature evolution and therefore a crucial processing parameter for controlling the change in form birefringence.

D. Direct-Write Pattern

To demonstrate the potential of this method for direct writing on a GLAD coating, a simple pattern was generated by moving the stage at a constant speed of 0.2 mm/s while irradiating the sample with the laser power set to 30 W. Under these conditions the retardance pattern shown in Fig. 6(a) was produced in five connecting segments shown on the map with arrows pointing in the scanning direction. Figure 6(b) shows the coating as imaged through crossed polarizers. As seen in the photograph, the pattern edges are not straight; this is due to instabilities in the laser energy. Further optimization of the irradiation conditions is needed to improve the pattern produced (during these experiments the laser output was fluctuating by about $\pm {5}\%$ with a period of the order of 15 to 20 s). Nevertheless, the pattern reported here serves as proof-of-concept of the method introduced in this work to spatially control the polarization properties of optical coatings with form birefringence.

 

Fig. 6. Direct-write pattern on an all-silica GLAD coating: (a) retardance map and (b) as seen through crossed polarizers.

Download Full Size | PDF

4. CONCLUSIONS

We have introduced a new methodology affording spatial control over the polarization output of birefringent wave plates by modifying the microstructure of coatings fabricated by glancing-angle deposition. The technique requires that it be performed in a vacuum environment to avoid the formation of large pores, and a ${{\rm CO}_2}$ laser that can be used in a direct-write fashion to locally melt the coating along a predesigned pattern. Our experimental and theoretical work demonstrated laser power to be the dominant processing parameter. The elicited change in microstructure modifies the form birefringence with the concomitant effect of controlling the output polarization of the processed optic. The work reported here is paramount because it opens new avenues for the experimental realization of optics for complex polarization control that require spatial control of birefringence. Although ${{\rm SiO}_2}$ was used in the GLAD coatings utilized in this work, we believe there is no inherent limitation to applying the same methods with other materials.