1. Introduction

Micro-optical elements such as lenticular lens arrays and microlens arrays have been widely used for beam shaping, steering and enhancing light extraction efficiency in displays and imaging such as digital projector, 3D imaging [1–3]. Many techniques such as direct writing, photolithography, and photoresist reflow have been suggested and investigated for the microlens array fabrication. Although the direct writing methods have a higher cost and lower productivity compared to other techniques, they have been considered to be advantageous for fabricating three-dimensional microstructures with arbitrary geometries. Among the direct writing methods, laser direct writing has been applied for rapid prototyping of 3D glass structures because its can be processed faster than other direct writing modalities such as FIB milling [4]. However, laser direct writing requires post-process polishing of the fabricated surface in order to achieve a roughness of only a few nanometers. Various post-processing techniques, such as chemical etching and thermal treatment, have been suggested for the rapid fabrication of high quality three dimensional microstructures [4–7].

Recently, our group developed CO₂ laser-assisted polishing of fused silica which had undergone femtosecond laser machining [8]. The tip of the silica-based optical fiber was machined conically using a femtosecond laser. The femtosecond laser-machined fiber tip was treated with millisecond CO₂ laser pulses in order to reduce the surface roughness. Upon irradiation of the CO₂ laser, the surface of the optical fiber melted and re-solidified resulting in a surface sufficiently smooth to permit total internal reflection.

In this study, we present the fabrication of a fused silica based mold for the microlenticular lens array using a femtosecond laser and a CO₂ laser. We applied the femtosecond laser machining and CO₂ laser polishing process, which was developed in our earlier study, for the fabrication of the glass mold for a plano-convex cylindrical lens. A femtosecond laser was used to machine the base structure of a three dimensional glass mold. The femtosecond laser machined structure has a repeating V-shape grooved surface. The CO₂ laser treatment followed the femtosecond laser process in order to refine the roughness within the V-shaped grooves. To evaluate the performance of the fabricated glass mold, we replicated a PDMS microlenticular lens arrays from the fabricated glass micro lenticular array mold.

2. Femtosecond laser-induced microstructuring on the fused silica glass

Fabrication of a fused silica based mold for the microlenticular lens array was conducted using two processes. In the first process, the basic structure of a microlenticular lens array mold was fabricated on a fused silica plate using a femtosecond laser. In the next process, a secondary treatment of irradiation by a CO₂ laser was used to obtain a smooth surface of a desired curvature. A schematic diagram of the femtosecond laser system and details of the scanning scheme for the V-shaped grooves are shown in Fig. 1. A commercial amplified Yb:KGW femtosecond laser (PHAROS, LightConversion, Vilnius, Lithuania) with a wavelength of 1030 nm was used to machine a base structure consisting of V-shaped grooves. The femtosecond laser has the pulse duration of 290 fs and the repetition rate of 100 kHz with the maximum output power of 6 W. Femtosecond laser pulses were irradiated onto a fused silica plate through an objective lens with the numerical aperture of 0.4 and the working distance of 20 mm. The spot size of the laser beam was measured to be 2 μm on the sample plane with a fluence of 31.9 J/cm² for each pulse. The scanning speed of the beam was set to 8 mm/s. Commercial fused silica plates with a thickness of 1 mm (VIOSIL-SQ, Shin-Etsu Chemical Co., Ltd, Japan) were used as substrates. Because fused silica has excellent optical properties, chemical resistance, and thermal resistance, it has been widely used as micro-optical elements [9, 10] and molds for further fabrication processes [11]. A fused silica plate was placed on a tilt stage for leveling and it was held to the tilt stage by vacuum. The tilt stage was located on an X-Y stage for 2D scanning, and the objective lens could be moved along the Z axis for focusing. The V-shaped groove was designed to have 50-μm width, 50-μm depth, and 60-μm pitch. The machining of each V-groove was conducted in a step-wise manner of layer-by-layer laser ablation (Fig. 1(b)). The first ablated layer had a width of 50 μm and it was engraved on the top of the plate. The second layer was engraved with a smaller width than the first layer on the bottom of the pit created after the first ablation process. The engraving of each successive layer was repeated to a depth of 50 μm.

 

Fig. 1 (a) Schematic diagram of the femtosecond laser system and (b) details of the scanning scheme for the V-shaped grooves.

Download Full Size | PDF

The fabricated fused silica V-shaped grooves were imaged using a scanning electron microscope (SEM) (Fig. 2).The repeating V-shaped grooves were fabricated as basic structures for the microlens array. The high magnification inset image reveals that the fabricated structures have a rough and porous surface.

 

Fig. 2 SEM image of the Fused silica glass microstructured using a femtosecond laser.

Download Full Size | PDF

3. CO₂ laser treatment of rough V-shaped groove

A microlenticular lens should have a smooth surface and a curvature that depends on the desired focal length. Since the femtosecond laser machined V-grooves have a rough surface, a CO₂ laser treatment was conducted to obtain a smooth lens surface with a desired curvature. A schematic diagram of the CO₂ laser system and the details of the scanning scheme for the process are shown in Fig. 3.A CO₂ laser (C-55L, Coherent, Santa Clara, CA, USA) with the maximum power of 55 W at a wavelength of 10.6 μm and a bare beam diameter of 1.8 mm was used. The bare CO₂ laser beam with a Gaussian profile was irradiated through a 2D scanner (hurrySCAN®II 10, SCANLAB, USA) with the focal length of 160 mm and the scan area of 110 x 110 mm². After passing the 2D scanner, the CO₂ laser beam was focused on the surface of a fused silica plate. The CO₂ laser was irradiated in zigzags with a scanning pitch of 20 μm by the 2D scanner (Fig. 3(b)).The focused CO₂ laser beam had a diameter of approximately 200 μm on the glass surface. Scan speed, laser irradiation power and the CO₂ laser treatment repetition time were found through the preliminary optimization process in order to make the microlenticular lens mold with a smooth surface and the radii of curvature of 20 μm and 30 μm. When the scanning speed of the beam was set to 50 mm s⁻¹, the repetition rate of the CO₂ laser was 5 kHz, the duty cycle was 40%, and the laser power was set to 10 W, 2 and 6 cycles of the CO₂ laser treatment produced the microlenticular lens array with 20μm and 30μm radii, respectively. The calculated fluence from the beam spot size and irradiated power was 6.4 J/cm² on the sample plane. A CO₂ laser with a mid-infrared wavelength has been applied for melting and ablating various glass materials because of its strong linear absorption for glass. In the CO₂ laser glass melting and ablation, the damage threshold for fused silica is generally associated with a critical temperature such as the softening point or melting point [9]. In order to minimize possible effects of the ambient gas and temperature, the CO₂ laser treatment process was performed in a constant temperature and humidity room. Upon irradiating the glass surface with a CO₂ laser, the temperature in the laser-irradiated zone rapidly increases up to the softening point and the onset of melting takes place on the surface. We believe that a significant overlap between the successive scan lines together with the larger beam spot size (200 μm) compared to the pitch (20 μm) minimizes the stress non-uniformity in the fused silica sample after the CO₂ laser treatment. However, a more thorough, quantitative evaluation of residual stress will be needed in the future for complete optical characterizations.

 

Fig. 3 (a) Schematic diagram of the CO₂ laser system and (b) details of the scanning scheme for the post-process.

Download Full Size | PDF

The melted surface of CO₂ laser-irradiated fused silica mold was observed by scanning electron microscopy (SEM) and typical surface images of fused silica after the CO₂ laser irradiation are shown in Fig. 4.For all treatment conditions, the V-shaped grooves were transformed into plano-convex cylindrical microlenticular lens arrays, and the surface of the microlenticular lens became smooth. We observed that the pores visible on the surface of the microstructure before the CO₂ laser irradiation had disappeared and the surface roughness was significantly decreased. In order to obtain desired curvatures for the microlenticular lens, the CO₂ laser treatments scanning were performed with different repetitions (two and six cycles). In the optimization process of CO₂ laser irradiation conditions, we observed that the engraved pattern by femtosecond laser irradiation remained on the fused silica surface after one irradiation by the CO₂ laser, and the radii of the curvature of microlenticular lens arrays increased as the number of CO₂ laser irradiation cycles increased. As the number of the treatment cycles increased, more material at the crest of the grooves melted and flowed down to fill the troughs of the grooves. As a result, the curvature of the lenticular lens decreased with the number of treatment cycles (Fig. 4(b), 4(d)). The radii of the curvature of microlenticular lens arrays treated with 2 and 6 cycles of CO₂ laser irradiation were approximately 20.7 μm (large curvature) and 29.0 μm (small curvature), respectively. In the SEM image evaluation, repeated exposures to CO₂ laser beams did not result in further improvement on the degree of roughness as the samples with two and six exposures showed virtually no difference in roughness (Fig. 4(a), 4(c)). Next, we measured surface roughness by atomic force microscopy (AFM) for a quantitative analysis. We first took topographic data from 1 x 1 μm² areas from samples treated with 2 and 6 cycles of CO₂ laser irradiation. Then the arithmetic average roughness, R_a, was computed from the obtained topographic height data. R_a values for samples with 2 and 6 treatment cycles were 15.3 ± 4.2 nm and 9.5 ± 3.0 nm, respectively. Despite we did not investigate further evolution of surface roughness as a function of CO₂ laser irradiation, it is already clear to see substantial improvements in surface smoothness before and after the CO₂ laser treatment.

 

Fig. 4 SEM image of the microlenticular lens with the fused silica glass.

Download Full Size | PDF

In order to evaluate the light focusing ability of the microlenticular lens array, the focal plane images of the fabricated microlenticular lens mold were captured with a CCD-based imaging system in a dark room. Figure 5(a) presents a schematic diagram of the focal ability measurement. A He:Ne laser with 632 nm wavelength (HNL050R, Thorlabs, USA) was used as a light source. The He:Ne laser was collimated and expanded by an objective lens (LC plan FI 60X, NA 0.7, Olympus) and a plano-convex lens (LA1401-A, f = 60 mm, Ø = 50.8 mm, Thorlabs). The collimated light was irradiated onto the microlenticular lens arrays. The image for the focal plane of the microlenticular lenses was captured by an objective lens (UPlanSApo 4x, NA 0.16, Olympus), a plano-convex lens (AC254-100-A1, Ø = 25.4 mm, Thorlabs) and a CCD camera (ARTCAM-150PIII, ARTRAY Co.Ltd, Japan). Figure 5(b) and 5(c) presents two images of the focal plane of the microlenticular lenses treated with 2 and 6 cycles of CO₂ laser treatments, respectively. It is clearly seen that the emitted light from the light source was focused on the focal planes of the microlenticular lens arrays. The widths of the focused line obtained with lenses treated with 2 and 6 treatment cycles were about 1.3 μm and 1.9 μm, respectively. The theoretical focused line widths of the fabricated micro lenticular lenses were calculated using the thin lens equation and the circular aperture formula. If collimated light with a wavelength of 632 nm is irradiated onto fabricated microlenticular lens arrays, where the refractive index of fused silica is 1.5, the theoretical focused line widths of microlenticular lenses with 20 μm and 30 μm radii of curvature is approximately 0.6 μm and 0.9 μm, respectively. Although the measured focused line widths were larger than the theoretical values, likely due to aberrations, the fabricated microlenticular lens was proven to be capable of focusing light.

 

Fig. 5 (a) Schematic diagram of the focus ability measurement and focal plane images of (b) 2cycles CO2 laser treated glass mold and (c) 6cycles CO₂ laser treated glass mold.

Download Full Size | PDF

4. Replication of a polymer mold (PDMS soft mold) and polymer lenticular lens array

The fabrication of a micro-optical element using a femtosecond laser was relatively straightforward and have been considered advantageous for fabricating 3D microstructures with arbitrary geometries compared to other fabrication schemes such as photolithography and photoresist reflow. However, the direct writing method using a femtosecond laser has a limitation in regard to mass production. For a higher productivity, a mold-based replication method is adopted. Generally, hard materials such as metals, silicon, and quartz have been used as molds in the replication method. As a hard mold is used repeatedly in a replication process, the features and patterns on the surface of the hard mold tend to collapse or get distorted by high pressure-induced stress, friction and fatigue accumulation. Especially, a hard mold made of fused silica is vulnerable to impact because it is brittle. In order to overcome these problems, soft polymer molds have been used after replicating from the hard mold (master mold). Compared with the hard mold, the soft polymer mode is robust to the impact and is cost-effective, and a review of surface feature replication techniques is found elsewhere [12]. In order to evaluate the feasibility of mass production, we tried to fabricate soft polymer molds from a previously fabricated fused silica microlenticular lens array as a master mold. Then, polymer microlenticular lens arrays could be fabricated from the soft polymer molds. The fabrication of polymer microlenticular lens arrays using polymer molds was conducted in two processes. These two processes were very similar except for the type of molds used. The entire manufacturing process is shown in Fig. 6.The first process is the replication of a polymer mold from the hard fused silica mold (Fig. 6(a)-6(e)), and the second process is the fabrication of a polymer microlenticular lens array by replication using the soft polymer mold (Fig. 6(e)-6(i)).

 

Fig. 6 Schematic diagram of the replication process. Replication of a polymer mold from the hard fused silica mold (a-e), Replication of a polymer microlenticular lens array from the polymer mold (f-i).

Download Full Size | PDF

Polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning Corporation, Midland, MI, USA) was used in the replication process of a soft polymer mold and microlenses. In order to decrease the adhesion between the molds and PDMS, a surface treatment was applied to the surface of the molds (Fig. 6(b), 6(f)) using dichlorodimethylsilane (440272-100ML, Sigma-Aldrich Co. LLC.). In the replication process of a soft mold from the hard fused silica, the hard mold was put in a vacuum desiccator for the surface treatment. A very small amount of the surface treatment agent was dropped onto a glass petri dish in the vacuum desiccator at once. The glass petri dish and hard mold were left for twenty minutes in the vacuum desiccator. After the surface treatment, a silicone elastomer base and a silicone elastomer curing agent for PDMS were mixed in a volume ratio of 10:1. Once the base and the curing agent were mixed, the desiccator was evacuated to remove air in the mixture. PDMS with air removed was dropped on the hard mold, and then the vacuum desiccator was used once more to remove air in PDMS (Fig. 6(c)). Then, PDMS and the hard mold were placed in an oven to cure PDMS for thirty minutes at 90°C. After taking the parts out of the oven, PDMS was separated from the hard mold (Fig. 6(d)-6(e)). After the replication of a soft mold, the replicated soft mold was used for replication of polymer microlenticular lens arrays. The surface of soft mold was treated by the surface treatment agent (Fig. 6(f)). The mix and bubbled removed PDMS was dropped on the soft mold, and then the PDMS and the soft mold degassed by a vacuum desiccator (Fig. 6(g)). The PDMS on the soft mold was cured in the oven and the cured PDMS was separated from soft mold (Fig. 6(h)-6(i)).

In the replication process, the soft PDMS mold was replicated from the hard glass mold, and then the PDMS microlenticular lens array was replicated with the soft PDMS mold. Because the hard glass mold has a plano-convex shape (Fig. 4), the soft PDMS mold has a plano-concave shape (Fig. 7(a)-(d)).Then, the corresponding PDMS microlenticular lens array has a plano-convex shape (Fig. 7(e)-(h)). The radii of curvature of the PDMS soft molds obtained from the fused silica molds after 2 and 6 cycles of CO₂ laser treatments were approximately 20.5 μm and 30.1 μm, respectively. The radii of curvature of the PDMS microlenticular lens array obtained from the corresponding soft PDMS molds were approximately 19.2 μm and 30.4 μm, respectively. The hard glass mold, soft mold, and PDMS microlenticular lens had almost the same curvature.

 

Fig. 7 SEM image of the PDMS soft molds (a-d) and the micro lenticular lens array with PDMS (e-f). (a) Replicated PDMS soft mold from 2-cycle-treated glass mold. (b) Cross-sectional image of replicated PDMS soft mold from 2-cycle-treated glass mold. (c) Replicated PDMS soft mold from 6-cycle-treated glass mold. (d) Cross-sectional image of replicated PDMS soft mold from 6-cycle-treated glass mold. (e) Replicated PDMS micro lenticular lens array from 2-cycle PDMS soft mold. (f) Cross-sectional image of replicated PDMS micro lenticular lens array from 2-cycle PDMS soft mold. (g) Replicated PDMS micro lenticular lens array from 2-cycle PDMS soft mold. (h) Cross-sectional image of replicated PDMS micro lenticular lens array from 2-cycle PDMS soft mold.

Download Full Size | PDF

5. Conclusion

In this paper, we present the fabrication of a fused silica based mold for the microlenticular lens array using a femtosecond laser and a CO₂ laser. In order to make the 3D microlens array mold, a commercial amplified Yb:KGW femtosecond laser was used as a primary machining tool and the subsequent treatment process was conducted with a CO₂ laser to polish the rough surface and to obtain a desired curvature for the microlenticular lens array. The surface roughness of the microstructure after femtosecond laser machining was significantly improved after the 2 cycles of CO₂ laser irradiation, and the curvature of the microlenticular lens could be controlled by the number of CO₂ laser irradiation cycles. To evaluate the performance of the fabricated glass mold, we employed a replication method using a soft polymer mold. The fabricated glass microlenticular array mold was used as a hard glass mold to make a soft PDMS mold, and the polymer microlenticular lens array was replicated from this soft PDMS mold. The glass mold, soft mold, and PDMS microlenticular lens had almost the same curvature. We expect that the suggested method can be widely used in various microlens fabrications for the rapid prototyping of high quality 3D microstructures at a low cost and in a high volume. Particularly, the fabricated soft mold and hard glass mold obtained by the suggested method can be useful for a roll to roll process. Additional studies for a large area machining better suited for industrial applications are currently underway.