Open Access
Add to BibSonomy
Abstract
3D printing techniques have great potential in the direct fabrication of microfluidic and many kinds of molds, such as dental and jewelry models. However, the resolution, surface roughness, and critical dimension uniformity of 3D printing objects are still a challenge for improvement. In this article, we proposed a 405nm light emitting diode (LED) backlight module based on stacks of structured films, and the full width half maximum (FWHM) of the angular distribution of this module is reduced to less than ± 15°. Compared with the commercial lens array optical module, the ten points intensity uniformity of an 8.9” build area is improved from 56% to 80%. Moreover, we found that the surface roughness and the sharpness of the edge of the printing objects are also obviously improved by our novel quasi-collimated LED backlight module. These features give us a promising way for the application of microfluidics and micro-optics components in the future.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Three dimensional (3D) printing technique has grown rapidly in the past decade, and more and more commercial products have been developed recently [1–4]. There are seven types of 3D printing techniques, including material extrusion [5], material jetting, sheet lamination, binder jetting, powder bed fusion [6], directed energy deposition [7], and vat photo-polymerization [8]. For printing objects with micrometer resolution, the vat photo-polymerization is a cost-effective choice [9–13]. The traditional photo-polymerization method, also called stereo-lithography, used Laser as a point scan light source [14]. Recently, for the purpose of high-speed 3D printing, surface light sources such as liquid crystal displays (LCD) or digital light processing (DLP) projectors are developed [15,16]. For the illumination mechanism, there are top-down and bottom-up systems. The top-down system illuminated the build plate from the top and used a blade to recoat the liquid surface for better smoothness. However, the limit of fabrication speed and the usage of large capacity resin inspired another bottom-up system.
For the bottom-up 3D printing system, there are two surface exposure techniques, one is DLP-type and another is LCD-type 3D printing technique. For the DLP-type 3D printing method, the intensity uniformity and collimation are great, however, the drawback is the conflict between the build area (projection field of view) and the resolution [13,12]. Due to the delicate MEMS fabrication process of the digital micro-mirror device, the cost will grow dramatically with the increase of pixel resolution. Moreover, when increasing the build area, the concentrated heat problem of the light source will become a big challenge. On the contrary, traditional LCD-type 3D printing methods have an excellent cost-performance ratio but poor optical performance. Recently, a commercial LCD-type 3D printer (Phrozen, Shuffle XL-2019) used the LEDs with lens array optical module to possess great collimation. However, the commercial LCD-type 3D printers based on lens array modules still have a few fabrication constraints. For example, the surface topography is highly sensitive to the small dim region at the intersections of the lens array, which causes the decline of critical dimension uniformity and accuracy of the printed objects. Moreover, the translucent and non-smooth surface quality are also annoying in some optical and microfluidic applications. [17,18]
In this research, we developed an LCD-type 3D printing system based on a novel UV LED backlight module design. Different from the light source of commercially available LCD-type 3D printer based on lens array and other novel directional backlight modules [19–21], our novel backlight module is based on direct-lit design and composed of stacks of structured films, including a reflector, a diffuser film, two crossed brightness enhancement films (BEF) [22], and a privacy filter (PF). Every functional film has its own purpose, such as improving intensity uniformity, energy recycling, light collimation, and cutoff large-angle leakage light. Finally, we performed our novel quasi-collimated LED backlight module for an LCD-type 3D printing system and demonstrated the merits of high critical dimension uniformity and small surface roughness (Rrms < 0.5 μm) within a 300 µm resolution of the 3D printing objects.
2. Design concepts and simulations
2.1 Design concepts of the quasi-collimated backlight module
Our home build 3D printer system is based on the replacement of an optical engine of the commercial 3D printer system (Phrozen, Shuffle XL-2019). The system is built by an 8.9 inch 2K resolution LCD panel, which corresponds to the 75µm pixel resolution. The resin used in this experiment is Phrozen ABS-like gray resin, which can resolve smaller than 50µm features and is a good choice for resolution tests. The curing energy of resin (Phrozen, ABS-like gray) is 4.87 mJ/cm2, and the curing time for the normal layer depends on the output intensity of different modules. In order to compare with the optical performance of the commercial lens array module directly, we kept LEDs in the same condition as a benchmark and just replace the lens array module with our diffusive module and quasi-collimated backlight module for impartial comparisons. The structures of three different modules for comparison are shown in Fig. 1. Our quasi-collimated backlight module is composed of enhanced specular reflector films (3M, ESR) on the bottom and sidewall of the case, which have a high reflectivity (average of 98% over the visible spectrum). Then, a diffuser film (Entire, EMS-90K3-2.0T) is put on the top of the LEDs for light scattering to improve intensity uniformity. In order to get the desired energy concentrated in the small emitting angle, we put two crossed-orientation BEFs (Yongtek, LEF-77) on the top of the diffuser. The BEF film is composed of a 1D right-angle prism array, the typical cut-off angle (θcutoff) of BEF calculated by Eq. (1) is about 35∼40 degrees (for nprism = 1.6∼1.5), which is derived from the Snell’s law and the prism geometry. [23]
Fig. 1. The schematic diagram of (a) commercial lens array optical module, (b) our diffuser LED-backlight module, and (c) our quasi-collimated LED-backlight module.
The mechanism of BEF is transmitting light of a desired angular distribution and reflecting light of the unwanted state. The reflection is mainly caused by the total internal reflection of two facets of the prism structure. The components below the BEF, such as diffuser, bottom and sidewall reflector can transform this unwanted light into a mixture of both states, and then be reflected again to the BEF for further recycling (desired transmission and unwanted reflection). The optical recycling results in larger energy concentrated in the central emitting angle. However, there are still some unwanted large-angle light rays leaking through the BEF, in order to eliminate the large-angle lobes of the transmission, we put a privacy filter (3M, PF215W9B) on the top of the BEF. The orientation of the micro-louver in the privacy filter is aligned with the prism direction of the outermost BEF, as shown in Fig. 1(c).
2.2 Simulations and analysis of quasi-collimated light module
Before we built up the backlight module, we set proper geometric and optical parameters for the commercial optical films to demonstrate the design concept. The simulation structure of our quasi-collimated LED-backlight module is shown in Fig. 2(a). The simulation region was set as 60 × 60 × 60 mm, and each LED emits 405 nm monochromatic wavelength with ± 26° FWHM distribution angle toward the z-direction. For simplicity, the total flux and ray number of each LED light source were assumed as one million and 100 W respectively. All the simulations in this study were implemented by the commercial ray-tracing software (TracePro, Lambda Research).
Fig. 2. (a)Simulation setup of quasi-collimated backlight module and (b) the influence of terminated flux threshold.
First, we selected a diffuser film with Lambertian distribution to get the best intensity uniformity. Then, two brightness enhancement films (BEFv + BEFh) with 90-degree roof angles and 50 µm pitch are set to be crossed orientation in our simulations. A 98% reflection and 2% absorption surface (3M, ESR) are surrounded at the bottom and side surfaces to prevent the escape of light rays. The refractive index of the BEF and diffuser materials are assumed as 1.5. Moreover, because the quasi-collimated LED-backlight module can effectively converge the emitting light to the normal angle, we put a detector at the location of the backlight output surface and LCD panel (30 and 60 mm away from the LEDs, respectively) to calculate the angular distribution and the intensity uniformity of our LED-backlight modules. In order to compare light distribution between each condition, the unit of gain is the intensity ratio normalized to the maximum intensity of LEDs with a Lambertian diffuser. The transmission property of the Lambertian diffuser is based on the experiments. (see Fig. S1)
For the correct simulation results, we examined the parameter of flux threshold, which is an important parameter for the light recycling process. Once rays fall below the flux threshold, the simulation will be terminated. In Fig. 2(b), we tested different thresholds (1%, 0.5%, 0.1%, 0.05%), and the simulation results indicated that when the flux threshold is lower than 0.1%, the gain of the two crossed BEF will be larger than one BEF. This phenomenon indicated the fully recycling process and the convergence of light in the perpendicular direction (0-degree viewing angle) of the emitting surface by the second BEF is the reason for higher gain. In order to get the balance between correct simulation results and consuming time, we set the flux threshold to be 0.05% in the following simulations.
From the simulation results shown in Fig. 3, we can separate the contribution of each structure film by a sequence of simulations. First, when the light rays from LEDs pass through a diffuser, the angular distribution is divergent and follows Lambert's cosine law (Fig. 3(a)). Second, when we put a BEF with horizontal orientation (Fig. 3(b)), the prism structures will mainly converge light in the y-direction, increase the gain (∼1.4) and reduce FWHM (∼± 35°). We also find some leakage light at an angle larger than 60°, which is caused by reflection from the first facet of the prism but transmission through the second facet of the prism. Third, when we put two BEFs with horizontal and vertical orientations (Fig. 3(c)), the emitting light will converge majorly in the x-direction and minorly in the y-direction. Moreover, the gain increased to 1.5, and the FWHM of angular distribution reduced to about ±20°. However, there are still some leakage light rays close to ±50°. Finally, to eliminate this leakage light, we put two crossed 1D privacy filters on the top of BEFs. The simulation results are shown in Fig. 3(d), we found that the large-angle leakage light rays are eliminated and the FWHM of angular distribution is reduced further to about ±13.5°.
Fig. 3. Simulation results of angular distribution in our quasi-collimated LED-backlight module (a) LED+ diffuser, (b) LED + diffuser + one BEF, (c) LED + diffuser + two crossed BEFs, (d) LEDs + diffuser + two crossed BEFs + two crossed 1D privacy filter.
In addition, we discussed the relation of LED quantities to the output energy and the angular distribution. The simulation results are shown in Fig. 4. We found that the angular distribution of 1 LED module and 4 LEDs module are almost the same, the difference is the output power, which is proportional to the number of LEDs.
Because the intensity uniformity before and after the piracy filter is almost the same, in order to save the consuming time of simulations, we just take off the piracy filter and analysis the influence of LEDs uniformity on the intensity uniformity of the backlight module. Figure 5(a) shows the simulation setup, the intensity of one of four LEDs is changed from 100% to 0%. The intensity uniformity of the backlight module is calculated by the ratio of minimal luminance (Lmin) to the maximum luminance (Lmax) within the region between four LEDs. The simulation result is shown in Fig. 5(b), we found that the intensity uniformity of the quasi-collimated backlight module is great (LED uniformity = 100%, backlight module uniformity = 88.7%). Moreover, the resistance to the variance of LEDs is also great in our quasi-collimated backlight module even if one of four LEDs is broken (LED uniformity = 0%, backlight module uniformity = 77.4%). This property is important because the intensity variance of LEDs is common. Although the LEDs uniformity could be improved by better quality and current control, the asynchronous decline of each LEDs may still cause the non-uniformity problem.
Fig. 5. (a) schematic of simulation setup and (b) the relation of backlight uniformity to LED uniformity.
3. Experimental Results
3.1 Analyzing optical properties of the quasi-collimated backlight module
In this article, we measured the intensity uniformity and angular distribution of emitting light to compare different optical modules. All the measurements are based on the same LEDs and driving conditions. For the intensity uniformity (or output power) measurement, we followed the concept of ANSI 9 method and used a power meter (Ophair Optronics, Nova II) with a power sensor (PD300RM-UV) to measure 10 points spread all over the entire LCD panel for uniformity, as shown in Fig. 6(d). The PD300RM-UV power sensor has a flat response in the spectral range, which allows the measurement of broadband light sources, like LEDs. The angular distribution of emitting light is measured by a goniometer (Nippon Denshoku, GC5000). We put a pinhole on the height of the LCD panel to measure the emitting light angular distribution of our diffusive and quasi-collimated LED backlight module. The measurement range is ± 75° against the normal direction of the samples.
Fig. 6. Photos of emitting light uniformity of (a) commercial lens array optical module, (b) our diffusive LED-backlight module, and (c) our quasi-collimated LED-backlight module. The related 10 points intensity uniformity measurement setup and results are shown in (d) and (e).
3.2 Intensity Uniformity
The design concept of our quasi-collimated LED-backlight module is trying to combine the advantages of collimation property in the traditional lens array module and the intensity uniformity property in the diffusive-backlight module. We used stacks of structure films to adjust the distribution of light of the entire 8.9” LCD panel and to narrow down the angular distribution of emitting light at the same time. For the intensity uniformity, we used a low driving current to measure 10 points spread all over the entire LCD panel, the photos and output power measurement results are shown in Fig. 6. We can obviously see the profile of square units caused by the lens array, and the dark (or bright) intersections will reduce the uniformity to only 51.8%. When we used a diffuser to replace the lens array, the uniformity increased a little to 55.8%. In comparison, the intensity uniformity of our quasi-collimated optical module is 80%. The improvement is due to light scattering by the diffuser and the light recycling process by the BEF, which could blur the image of LEDs and make the intensity of each LEDs mix together. Our experiments demonstrated that even if the decline of LEDs is not perfectly the same, the diffused and recycling process in our optical module can improve the intensity uniformity efficiently. Here, the 80% uniformity is not the upper limit, because the initial uniformity of LED array in the commercial lens array module is poor. If we control the quality of LEDs, the uniformity can be larger than 90%.
3.3 Angular distribution
The light control progress of our quasi-collimated module is analyzed by angular distribution measurements of emitting light to illustrate the function of each structure film, as shown in Fig. 7. In order to compare the relative light throughput of each condition, the voltage and current for LEDs are the same, and the calculations of gain are relative to the maximum intensity of LEDs with only a diffuser. First, a diffuser film is placed at 10 mm on top of the LEDs, the angular distribution of emitting light is broad, which is used to improve intensity uniformity. However, the broad angular distribution is bad for the resolution of 3D printing. To improve the resolution, we put two crossed aligned BEFs on the top of the diffuser. The distance from the diffuser to BEF is 20 mm. The orientation of the prism structure of the first BEF is horizontal, which constrained the emitting light not only in the vertical (y-axis) but also in the horizontal (x-axis) direction, as shown in the red and green lines of Fig. 7(b). Then a second BEF with the vertical orientation of prism structure is stacked on top of the first BEF. We found that the angular distribution of emitting light after passing through two cross-oriented BEFs becomes narrower and stronger at the zero-degree emitting angle, as shown in Fig. 7(c). The gain is about 1.68 fold compared to the LEDs with only a diffuser. However, there are two distinct side lobes after passing through BEFs, and the FWHM of central emitting light is still too large (±27°), therefore, we finally used a light control film to eliminate the side lobes and improve the collimation. The light control film, which contains closely spaced micro-louver structures with a 30-degree cutoff angle, is usually used as a privacy protection function. After passing through two crossed alignment 1D privacy filters, the FWHM of the angular distribution of emitting is narrowed down to ±15°, as shown in Fig. 7(d). Here, the BEF film and recycling process not only improves the intensity uniformity but also increases the throughput energy. The emitting energy of the optical module with and without BEFs are shown in the green and black curves in Fig. 7(d). The improvement of 1.67 fold is contributed to the recycling and selective angular filter effect to reshape of emitting light by BEFs. Comparing the angular distribution with previous simulation results (Fig. 3), the measurements (Fig. 7) present good consistency.
Fig. 7. Angular distribution analysis in our quasi-collimated LED-backlight (a)LEDs + diffuser, (b) LEDs + diffuser + one BEF, (c) LEDs + diffuser + two crossed BEFs, (d) LEDs + diffuser + two crossed BEFs + two crossed 1D privacy filter.
3.4 Appearance of 3D printing results
The optical power of our quasi-collimated LED module during 3D printing experiments is 2.1 mW/cm2, and the efficiency is about 4.2%. (more detail in Table S1, S2 in the Supplement 1). To evaluate the influence on the 3D printing objects, we used a 3D printer composed of different light source modules to print the line pair test pattern and measured the lateral resolution, size deviation (uniformity), and surface roughness. The line pair pattern is designed by the integer multiple of the unit pixel, the line width varies from 225 µm to 600 µm. The line width of the 3D printed objects is measured by microscope (WHITED, WM-100) with high-resolution CCD (WHITED, UC-1800) to analyze the lateral accuracy and size deviation. The surface roughness is measured by the surface profiler (KLA-Tencor, P-11) with a 500 µm scan length.
The photos and microscope images of 3D printed objects by commercial LED lens array module and our quasi-collimated LED backlight module are shown in Fig. 8. In Fig. 8(a), we printed two flat slabs, the left one is fabricated by a commercial lens array module, and the right one is fabricated by our quasi-collimated LED backlight module. There is a distinct difference in the surface smoothness of the two objects, which is related to the uniformity of intensity distribution of the light source as shown in Fig. 6(a) (c). If we take a closer look at the printed test pattern (Fig. 8(b)), we found the optical properties of the two objects are obviously different. One is a diffusive reflection surface and another is a specular reflection surface. Moreover, the line pair patterns of microscopic images of the two printed objects indicate a jagged or a smooth edge of line pattern respectively, as shown in Fig. 8(c). In addition, we also used a microscope to measure the relation of the designed line width to the actual printed line width, the results are shown in Fig. 9. In Fig. 9, we summarized ten 3D printed test pattern objects spread all the building area, as shown in Fig. 9(c). The accuracy is defined based on the regression line of design line width to actual printed line width and quantized by the coefficient of determination (R2). We can see that both optical modules have a good linear relation between design and printed pattern dimension, and our module (R2 = 0.9993) is slightly better than the lens array module (R2 = 0.9857). However, the size deviation implied by the error bar indicated the improved critical dimension uniformity of our quasi-collimated LED backlight module. The result of critical dimension uniformity of printed objects also corresponds to the intensity uniformity in the building area.
Fig. 8. Photographs of 3D printed objects of (a) flat slabs, (b) line pair test patterns, and (c) the microscopic images of the line pattern. (Note: the left one is printed by a commercial lens array module and the right one is by our quasi-collimated optical module.)
Fig. 9. Critical dimension uniformity and accuracy testing of (a) commercial lens array module, and (b) our quasi-collimated LED backlight module summarized by (c) ten 3D printed test objects inside a 16 x9.6 cm area
4. Discussions and conclusions
The surface roughness comparison of our proposed optical module and the commercial benchmark is shown in Fig. 10, which indicated the great reduction of root mean square surface roughness (Rrms) of the printing object from 5.25 µm to 0.35 µm. Moreover, we found either the width of the jagged shape on the edge of a line pattern in Fig. 8 (c) or the pitch of surface roughness in Fig. 10 (a) is very close to 75µm, which is exactly the size of a unit pixel. Therefore, we attributed this phenomenon to the structure inside the LCD panel. For the 405 nm LEDs, the emitting light may be blocked by the red and green color filters inside the LCD panel, therefore, even when the liquid crystal is turned ON, the output light is not uniform but pixelated. On the other hand, our quasi-collimated backlight is a better tradeoff between critical dimension uniformity, surface roughness, and resolution.
Fig. 10. Surface profile image of 3D printed object on a smooth area by (a) commercial lens array module and (b) our quasi-collimated optical module.
In summary, we developed an LCD-type 3D printing system based on the design of a novel quasi-collimated UV LED backlight. The advantage of our proposed module is not sensitive to the intensity variance of LEDs and is uniform (without pixeled) intensity distribution compared to the traditional lens-array optical module. We also find out the surface roughness, critical dimension uniformity, and accuracy of printing objects can be improved. We have demonstrated the optical performance (e.g., intensity uniformity, angular distribution) by simulations and measurements. Compared to the lens array module of a commercial product, the diffused and recycling process in our optical module can improve the intensity uniformity (52% →80%) efficiently. Moreover, the quasi-collimated property can reduce the surface roughness (Rrms = 5.25 µm →0.35 µm) of 3D printing objects. To our best knowledge, this paper is the first UV backlight design based on the light recycling process used in an LCD-type 3D printing system. Our simulations also implied that this kind of optical module has great potential to extend to larger building area 3D printers easily in the future.
Funding
Ministry of Education (Higher education sprout project); Ministry of Science and Technology, Taiwan (109-2222-E-011-005-MY2).
Acknowledgments
We deeply appreciate the instrument support from the Materials Research Laboratory, Industrial Technology Research Institute (ITRI), Taiwan.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but maybe obtained from the authors upon reasonable request
Supplemental document
See Supplement 1 for supporting content.
References
1. M. Zastrow, “The new 3D printing,” Nature 578(7793), 20–23 (2020). [CrossRef]
2. A. Nazir, A. Azhar, U. Nazir, Y.-F. Liu, W. S. Qureshi, J.-E. Chen, and E. Alanazi, “The rise of 3D Printing entangled with smart computer aided design during COVID-19 era,” Journal of Manufacturing Systems 60, 774–786 (2021). [CrossRef]
3. W. Xu, S. Jambhulkar, D. Ravichandran, Y. Zhu, M. Kakarla, Q. Nian, B. Azeredo, X. Chen, K. Jin, B. Vernon, D. G. Lott, J. L. Cornella, O. Shefi, G. Miquelard-Garnier, Y. Yang, and K. Song, “3D Printing-Enabled Nanoparticle Alignment: A Review of Mechanisms and Applications,” Small 17(45), 2100817 (2021). [CrossRef]
4. T. D. Ngo, A. Kashani, G. Imbalzano, K. T. Q. Nguyen, and D. Hui, “Additive manufacturing (3D printing): A review of materials, methods, applications and challenges,” Composites Part B: Engineering 143, 172–196 (2018). [CrossRef]
5. M. Spoerk, C. Holzer, and J. Gonzalez-Gutierrez, “Material extrusion-based additive manufacturing of polypropylene: A review on how to improve dimensional inaccuracy and warpage,” J Appl Polym Sci 137(12), 48545 (2020). [CrossRef]
6. A. Kumar, L. Collini, A. Daurel, and J.-Y. Jeng, “Design and additive manufacturing of closed cells from supportless lattice structure,” Addit. Manuf. 33, 101168 (2020). [CrossRef]
7. K. M. Abate, A. Nazir, and J.-Y. Jeng, “Design, optimization, and selective laser melting of vin tiles cellular structure-based hip implant,” Int J Adv Manuf Technol 112(7-8), 2037–2050 (2021). [CrossRef]
8. H. Quan, T. Zhang, H. Xu, S. Luo, J. Nie, and X. Zhu, “Photo-curing 3D printing technique and its challenges,” Bioactive Materials 5(1), 110–115 (2020). [CrossRef]
9. D. Ahn, L. M. Stevens, K. Zhou, and Z. A. Page, “Rapid High-Resolution Visible Light 3D Printing,” ACS Cent. Sci. 6(9), 1555–1563 (2020). [CrossRef]
10. M. Mohamed, H. Kumar, Z. Wang, N. Martin, B. Mills, and K. Kim, “Rapid and Inexpensive Fabrication of Multi-Depth Microfluidic Device using High-Resolution LCD Stereolithographic 3D Printing,” JMMP 3(1), 26 (2019). [CrossRef]
11. K. J. Krieger, N. Bertollo, M. Dangol, J. T. Sheridan, M. M. Lowery, and E. D. O’Cearbhaill, “Simple and customizable method for fabrication of high-aspect ratio microneedle molds using low-cost 3D printing,” Microsyst Nanoeng 5(1), 42 (2019). [CrossRef]
12. K. Kowsari, B. Zhang, S. Panjwani, Z. Chen, H. Hingorani, S. Akbari, N. X. Fang, and Q. Ge, “Photopolymer formulation to minimize feature size, surface roughness, and stair-stepping in digital light processing-based three-dimensional printing,” Addit. Manuf. 24, 627–638 (2018). [CrossRef]
13. E. Behroodi, H. Latifi, and F. Najafi, “A compact LED-based projection microstereolithography for producing 3D microstructures,” Sci Rep 9(1), 19692 (2019). [CrossRef]
14. B. Busetti, B. Steyrer, B. Lutzer, R. Reiter, and J. Stampfl, “A hybrid exposure concept for lithography-based additive manufacturing,” Addit. Manuf. 21, 413–421 (2018). [CrossRef]
15. X. Wu, C. Xu, and Z. Zhang, “Flexible film separation analysis of LCD based mask stereolithography,” J. Mater. Process. Technol. 288, 116916 (2021). [CrossRef]
16. D. S. Kim, J. Suriboot, M. A. Grunlan, and B. L. Tai, “Feasibility study of silicone stereolithography with an optically created dead zone,” Addit. Manuf. 29, 100793 (2019). [CrossRef]
17. G. D. Berglund and T. S. Tkaczyk, “Fabrication of optical components using a consumer-grade lithographic printer,” Opt. Express 27(21), 30405 (2019). [CrossRef]
18. B. G. Assefa, M. Pekkarinen, H. Partanen, J. Biskop, J. Turunen, and J. Saarinen, “Imaging-quality 3D-printed centimeter-scale lens,” Opt. Express 27(9), 12630 (2019). [CrossRef]
19. Y. Zhang, D. Yi, W. Qiao, and L. Chen, “Directional backlight module based on pixelated nano-gratings,” Opt. Commun. 459, 125034 (2020). [CrossRef]
20. Y.-J. Wang, S.-H. Ouyang, W.-C. Chao, J.-G. Lu, and H.-P. D. Shieh, “High directional backlight using an integrated light guide plate,” Opt. Express 23(2), 1567 (2015). [CrossRef]
21. Y.-J. Wang, J.-G. Lu, W.-C. Chao, and H.-P. D. Shieh, “Switchable viewing angle display with a compact directional backlight and striped diffuser,” Opt. Express 23(16), 21443 (2015). [CrossRef]
22. M.-W. Wang and C.-C. Tseng, “Analysis and fabrication of a prism film with roll-to-roll fabrication process,” Opt. Express 17(6), 4718 (2009). [CrossRef]
23. G. Boyd, “Optical Enhancement Films,” in Handbook of Visual Display Technology (Springer, 2012), pp. 1625–1646.
