Open Access
Add to BibSonomy
Abstract
We demonstrate millimeters-long VCSEL linear arrays with SU-8 epoxy-based microlenses that are directly patterned and cross-linked on the output apertures by a simple, photoacid-diffusion-aided photolithography technique. The linear arrays are capable of delivering >7 W of peak pulsed output power. By exploiting the photoacid diffusion effect, it is possible to produce a range of microlens structures with height and radius of curvature ranging from approximately ten to tens of microns. Simulation and experimental results show that the far-field beam divergence can be reduced by a factor of up to 7 in VCSELs integrated with optimal microlens dimensions.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Vertical-cavity surface-emitting laser (VCSEL) arrays have emerged as an important light source for data communication and parallel optical interconnects [1,2] at low optical power, as well as for a range of high optical power applications such as optical pumping, triggering, 3D sensing and LiDAR systems [3–9]. VCSEL arrays are more advantageous than stacked edge-emitting laser arrays in terms of scalability, uniformity, and yield. The aperture diameters of conventional oxide-confined VCSELs are usually smaller than 50 µm in order to achieve uniform current injection and optimal optical mode confinement. This results in 1/e2 divergence (full) angles ranging from ∼10° to <50°, as the beam divergence of a VCSEL is inversely proportional to its beam waist [10]. The inherently large beam divergence of oxide-confined VCSELs tends to degrade the far-field beam profile and severely limits the working distance of a VCSEL-integrated system. Consequently, VCSELs are usually used with external micro-optical components such as microlenses in order to improve beam collimation or focusing.
Microlenses can be fabricated directly on top of light-emitting or detecting devices via a few conventional methods. Thermal reflow is a commonly used simple method, where an array of spheroidal structures is formed by photolithography and thermal reflow of photoresist. Subsequently, photoresist spheroids are either left on the aperture windows of top-emitting devices [11], or used as etch mask to form microlenses on the back surface of bottom-emitting VCSELs after plasma etching [12]. The other commonly used methods are microplastic hot embossing, where microlenses are formed by pressing a heated mold against a polymer substrate, typically polymethyl methacrylate (PMMA) [13]; the microdroplet jetting method, where UV-curable polymer droplets are nozzle-printed on a substrate and subsequently cured by UV to form microlenses [14]; and polydimethylsiloxane (PDMS) molding technique involving thermal and pressure manipulation [15][16], which produces a polymer microlens array film that needs to be transferred and aligned to the device chip. Despite being a well-established technique, thermal reflow of photoresist typically produces thin microlenses (i.e., large radii of curvature) with shape profile limited by the thermal stability of the photoresist. Microlens arrays made from hot embossing and other micromolding processes cannot be readily integrated on device chip or wafer, thus requiring additional optical alignment, transfer and bonding steps. As for the microdroplet jetting method, it is challenging to control the droplet shape at high jetting speed.
Recently, there have been efforts to fabricate microoptics directly on VCSELs and other optoelectronic devices using SU-8 – an epoxy-based negative-tone photoresist. SU-8 has the advantage of high optical transparency over the visible and near infrared range. Using conventional photolithography steps, high aspect-ratio patterns with vertical sidewalls can be defined using SU-8, producing chemically-resistant and thermally-stable structures after a post-exposure bake. The reliability of SU-8 as a core material for microlens integration on VCSELs has been studied and found to have good thermal and mechanical stability [17]. In recent years, a series of work has demonstrated the use of SU-8 micropillar pedestals for coupling drop-dispensed or micro-spotted polymer microlenses to VCSELs [18–20]. However, the shaping of SU-8 into more complex structures deviating from its usual vertical pillar shape requires the use of non-conventional lithography techniques. A relatively inexpensive technique to fabricate SU-8 microlens structures involves using backside 3D diffuser lithography, but the sample or substrate needs to be UV-transparent [21]. Alternatively, costly 3D direct patterning techniques have been used to pattern SU-8 microstructures directly on optoelectronic devices. For example, grayscale electron beam lithography was used in Ref. [22] to fabricate SU-8 microprisms directly onto an array of VCSELs for beam steering purposes. In another recent work [23], aspherical-shaped SU-8 microlenses were patterned directly on VCSELs by femtosecond laser direct writing.
In this work, we have developed a novel technique to fabricate microlenses directly on the output apertures of VCSEL arrays through a simple photolithography method exploiting the photoacid diffusion effect of SU-8. The photoacid diffusion effect can be controlled to result in microlenses with heights on the order of tens of microns and a range of curvatures. Moreover, SU-8 photolithography is compatible with III-V semiconductor processing, allowing wafer-level direct integration of optoelectronic devices with chemically and thermally stable microlens arrays. By shaping the microlens curvature and height, we demonstrate the possibility to significantly reduce far-field beam divergence and to achieve peak power density comparable to state-of-the-art performance.
2. Experimental methods
The wafer used to fabricate VCSEL arrays in this work consists of a standard 850 nm wavelength VCSEL structure, which comprises a p+ top contact layer, a p-type top distributed Bragg reflector (DBR), an AlGaAs oxidation layer, 3 GaAs quantum wells with AlGaAs barriers in the active region, and an n-type bottom DBR mirror. The epitaxial structure is grown on an n+ GaAs (001) substrate. 2 mm-long VCSEL linear arrays and single-aperture VCSELs with different aperture diameters are then fabricated through standard photolithography and mesa wet etching methods. The oxide aperture is formed by steam oxidation. The sample is then planarized with a thick photosensitive benzocyclobutene (photo-BCB) resin, and deposited with bond pads to connect the individual array elements. SU-8 microlenses are patterned directly on the top VCSEL facets via a photoacid diffusion-aided photolithography process. Hereafter, “facet” or “top facet” refers to the emitting area confined within the oxide aperture, and does not include the area covered by the annular top contact. The good match in coefficient of thermal expansion (CTE) between the SU-8 (52×10−6/°C [24]) and photo-BCB (42×10−6/°C [25]), which is smaller than other MLA polymers such as PMMA and PDMS, helps to minimize thermal stress during curing. Moreover, with an excellent thermal stability as indicated by its 5% weight loss temperature at 300°C [24], these SU-8 microlenses are well suited for high-power applications. Finally, the finished sample is wire bonded on a QFP (Quad Flat Package) package and soldered on a printed circuit board (PCB) for beam profiling and pulsed power measurement purposes.
SU-8 photoresist is composed of a mixture of resin, photoinitiator and solvent. The process of photo-polymerization starts after the SU-8 film is soft baked to drive off excess solvent and is exposed to UV light, which generates photoacid through the decomposition of the photoinitiator. During the post-exposure bake (PEB) step, the photoacid serves as a catalyst in a classic acid catalyzed cationic reaction that ends with the resin reticulation, where the epoxy groups of different SU-8 monomers are crosslinked [26]. Although SU-8 is known to produce highly vertical sidewalls with a large aspect ratio (e.g., in excess of 5), it is possible to shape the sidewalls into curved lens structures by coating a second SU-8 layer after soft bake and UV exposure of the first layer but before crosslinking the polymer. A schematic of the microlens fabrication process is shown in Fig. 1(a). To ensure reproducibility, the first SU-8 layer was soft baked for a prolonged duration (at least 2× the recommended soft bake time in the datasheet) to fully remove solvent in the first SU-8 film prior to photoacid generation during UV exposure. Subsequently, the second SU-8 layer enables the diffusion of photoacid from the exposed SU-8 region into the adjacent unexposed region [27]. The photoacid acts as a catalyst in the subsequent polymer cross-linking reaction. The diffusion process is performed either in ambient air for 2 hours, or on a hotplate at 40°C for 30 minutes (the latter was applied in this study). Subsequently, the PEB step is performed where SU-8 is baked on a hotplate for crosslinking and after resist development, microlenses are formed on top of the VCSEL apertures. By adjusting process parameters, the curvature and thickness of the microlenses can be controlled. Figure 1(b) shows scanning electron microscope (SEM) images of VCSEL single-aperture devices and linear arrays integrated with SU-8 microlenses.
Fig. 1. (a) Schematic of the SU-8 microlens fabrication process. (b) SEM images of VCSEL single-aperture devices and linear arrays integrated with SU-8 microlenses. Scalebars indicate 100 µm. Note: a few microlenses in this sample have some non-uniformities caused by edge bead effect when spin coating thick, viscous SU-8 on a small sample.
By using different viscosity combinations in the first (η1) and second (η2) SU-8 layers, different microlens shape profiles are achieved, as shown in Fig. 2. The low, medium, and high viscosities used in this work correspond to 340, 4400, and 12000 cSt, respectively, which are standard viscosities from off-the-shelf SU-8 products. The microlens height is determined by the viscosity of the first layer, as the SU-8 film thickness is proportional to η1. We omitted the cases where η1 = high, η2 = low or η2 < η1 because they were found to produce undesired microlens shape profiles. For example, η2 = low or η2 < η1 tend to result in flattened and conjoined adjacent microlenses as shown in Fig. 3(a). Moreover, identical η1 and η2 tend to result in a negatively tapered profile, which is more pronounced when η2 = high, as evident in Fig. 3(b).
Fig. 2. SEM images of 4 types of microlens shape profiles achieved by using different viscosity combinations in the first (η1) and second (η2) layers, as indicated by the table. These test structures were patterned on bare Si substrate and reproducible on GaAs substrate. Note: some of the test structures were accidentally damaged during cleaving of sample cross sections.
Fig. 3. (a) Conjoined microlenses resulting from low viscosity in the second SU-8 layer (η2 = low). (b) Negatively tapered microlenses due to identical viscosities in the first and second SU-8 layers, especially when η1 = η2 = high. These test structures were patterned on bare Si substrate and reproducible on GaAs substrate.
A low viscosity indicates a high solvent concentration in the SU-8, and vice versa. According to the Stokes–Einstein equation, the diffusion coefficient of particles in a fluid is proportional to temperature, and inversely proportional to viscosity [28]. The diffusion of photoacid from the exposed region in the first layer into the second layer is aided by the diffusion of solvent in the reverse direction (from the second layer into the first layer). As shown in Fig. 2, type C and D microlenses have a cylindrical base, hereafter known as the pedestal, which is the result of the thicker first SU-8 layer. The nearly vertical pedestal sidewalls suggest that the solvent concentration provided by the second SU-8 layer (with medium or high viscosity) was probably lower near the microlens base, impeding photoacid diffusion. We find that a right balance needs to be achieved between the film thickness (determined by η1) and solvent content in the second layer (determined by η2), in order to result in uniform photoacid diffusion for desired microlens shape formation. The optimal combinations for the present study are shown in Fig. 2. Further room for process development and improvement is possible by thinning standard SU-8 photoresists to obtain different viscosities in order to produce microlens height and curvature that meet application requirements.
Pulsed-power measurements are performed on VCSEL linear arrays with and without microlenses, at far field. Each sample, mounted on PCB, is placed on an optical rail to allow the emitted output power to be measured at far-field distances up to 40 mm away from the aperture of the energy sensor. The low-energy photodiode sensor (Ophir model PD10-pJ-C) used in this experiment has an aperture diameter of 10 mm and is capable of detecting weak light energies down to 10 pJ. The PCB is connected to a high pulsed voltage supply from a commercial pulse generator with a maximum output of 500 V and 10 A. A 50 Ω resistor is connected in series with the sample to achieve impedance matching to the pulse generator. The VCSEL arrays are pulsed with 5 ns-wide pulses at a 5 Hz pulse repetition frequency. In addition, beam profile and divergence measurements are performed using a commercial beam profiling camera (DataRay S-WCD-UHR), which has a sensor size of 6.6 mm × 5.3 mm. The half-angle beam divergence, θ, is given in Eq. (1), where D1 and D2 are the diameters of the beam spots at two different positions away from the sensor, and L is the distance between them. The beam diameter is obtained by taking the 1/e2 width of a Gaussian beam.
3. Results and discussion
In Fig. 4, the radius of curvature (ROC) as a function of microlens diameter for the four types of microlens structures are compared. The ROC, R, is calculated based on a spherical cap defined by Eq. (2), where r is the microlens radius and hL is lens height; the pedestal height for microlens types C and D, hP, can be obtained by subtracting the lens height from the total height (hS) of the microlens structure, as given in Eq. (3). The SEM image at the bottom of Fig. 4 shows the hP and hL of three type D microlenses with varying diameter.
Fig. 4. Radius of curvature (left y-axis) and corresponding focal length (right y-axis) vs microlens diameter for the 4 types of microlens shape profiles. The SEM image below the graph shows the pedestal height (hP) and lens height (hL) positions of three type D microlenses with decreasing diameter from left to right.
Since the ROC is approximately half of diameter, the microlenses are nearly hemispheres. The corresponding focal length on the right y-axis is calculated using the lens maker’s formula for a plano-convex lens. The radii of curvature (ROCs) produced in this work range from ∼15 to >60 µm, corresponding to focal length range from ∼26 to >106 µm. The ROCs for all four types of microlens profiles are similar at diameters smaller than 50 µm. At larger diameters, types A and B produce more curved lens surfaces (i.e., smaller ROCs) than types C and D. Because the diffusion of photoacid is isotropic [27] and the same diffusion process was applied in all four cases, the resulting microlens curvature profiles are not significantly different between the four types. The trend of increasing ROC with increasing diameter is the result of greater lens height formed by more photoacid diffusion from the UV-exposed SU-8 covering the larger top facet surface area. For practical integration with VCSELs, the microlens profile needs to match the corresponding VCSEL mesa diameter, which is typically ≤50 µm, and the top VCSEL facet needs to be placed at a distance equal to the focal length from the lens. Since all four lens types have large focal lengths >20 µm, the distance between the top VCSEL facet and microlens needs to be compensated by a pedestal with a height of at least 20 µm. Such pedestals are inherently found in microlens types C and D, whereas types A and B will require an additional photolithography step to create pedestals on which the microlenses will sit.
Simulations based on the beam propagation method (BPM) [29] were performed using the RSoft BeamPROP software to assess the improvement of beam divergence with microlens integration. Figure 5 compares the far-field beam propagation profile of a beam with 15° half-angle divergence before and after microlens integration. The simulation parameters consist of an SU-8 refractive index of 1.562 at 850 nm [24], a lens diameter of 50 µm, a lens ROC of 25 µm, and two pedestal heights of 20 µm and 50 µm. According to Fig. 4, the focal length of such microlens dimensions is ∼45 µm. The resulting beam divergence is reduced by 2× to 7.65° and by 7× to 2.23° at pedestal heights of 20 µm and 50 µm, respectively. Therefore, when closely matched to the focal length of the microlens, the optimal beam divergence reduction achievable by the SU-8 microlens design in this study is predicted to be 7×.
Fig. 5. [Best viewed in color] Simulated far-field beam propagation profile of (a) a beam with half-angle divergence of θ = 15° reduced to (b) θ = 7.65° through an SU-8 microlens with 20 µm pedestal height and (c) θ = 2.23° through the same SU-8 microlens with 50 µm pedestal height. The microlens has a diameter of 50 µm and ROC of 25 µm.
Figure 6 compares the far-field beam profile of a VCSEL with ∼46 µm-wide top facet integrated with type D SU-8 microlens, at distances up to 30 mm away from the beam profiling camera, with and without microlens. The half-angle beam divergence of the 46 µm-wide aperture without microlens was measured to be 15.2°. The microlens diameter is approximately 55 µm, with a pedestal height of ∼30 µm. These dimensions are limited by our fabricated VCSEL samples and the standard SU-8 viscosities used. With the microlens, the beam divergence in Fig. 6 is reduced to 3.5°, which represents a 4.3× reduction. This beam divergence reduction is consistent with the BPM simulation prediction. Further improvement is expected by increasing the SU-8 pedestal height to ∼50 µm, by using a thicker SU-8 first layer. On a side note, it is apparent from Fig. 6(a) that the beam profile of the VCSEL without microlens is more Gaussian-shaped, while the microlens-integrated VCSEL in Fig. 6(b) shows donut-mode emission. While all the fabricated VCSELs in this study emit multiple modes due to their large aperture sizes, it is difficult to ascertain the origin of the donut-shaped higher order modes in Fig. 6(b) without a detailed analysis of the VCSEL structure. An interesting possibility is related to the mode selection properties reported on microlens-integrated VCSELs [30], which offers the potential of emission mode shape engineering. The presence of the lens may have altered the VCSEL cavity laterally in such a way that lasing is slightly less efficient where the lens is thickest.
Fig. 6. [Best viewed in color] VCSEL beam profile (a) without microlens at 10 mm distance and (b) when integrated with type D SU-8 microlens at different far-field distances up to 30 mm. Note: the diffraction pattern is caused by contaminants and scratches on the glass lid covering the packaged sample and the neutral density (ND) filter covering the camera’s sensor.
In Fig. 7(b), the far-field peak pulsed output power of two similar 2 mm-long, 46 µm-facet-diameter VCSEL linear arrays with and without microlenses (type D, ∼30 µm pedestal height) are compared. The experimental setup for measuring pulsed optical power is described in the previous section and photographed in Fig. 7(a). The VCSEL arrays were pulsed at 500 V (10 A) with a pulse repetition frequency of 5 Hz, and pulse widths of 5, 10, 20 and 40 ns. The peak optical output power is obtained from dividing the measured optical energy by the pulse width. At a far-field distance of ∼2 mm from the VCSEL facet, the photodiode sensor with 10 mm-diameter aperture captured a peak pulsed power of 8.3 W from the VCSEL array integrated with type D microlenses, compared to 7.4 W from similar array without microlenses due to beam divergence loss caused by the beam spot size exceeding the sensor aperture. The measured peak power of the VCSEL arrays decreases more rapidly with increasing far-field distance >12 mm, as more of the diverging beam spot falls outside of the sensor aperture. At a further distance of 40 mm, the output power measured from microlens-integrated VCSEL arrays is 3× higher than those without microlenses.
Fig. 7. (a) Photograph of a pulsed optical power measurement in progress. (b) Far-field peak pulsed output power of two similar VCSEL linear arrays with and without integrated microlenses (type D), as captured by a photodiode sensor with 10 mm-diameter aperture, at various far-field distances from the VCSEL facet.
4. Conclusions
Using a simple photolithography technique based on the photoacid diffusion effect in SU-8, we have shown that it is possible to fabricate near-hemisphere microlenses with a range of diameters directly on top of the VCSEL facets. The microlenses are formed by a two-layer SU-8 coating process, where the first layer defines the lens dimensions including pedestal height after UV exposure, and the second layer produces a curved lens surface by photoacid diffusion from the exposed first layer. The microlens dimensions and curvature can be shaped using SU-8 layers with different viscosities. We have demonstrated experimentally that such microlenses can be patterned directly on 2 mm-long VCSEL linear arrays delivering >7 W peak pulsed output power, leading to over 4× reduction in beam divergence. Further beam divergence reduction up to 7× is possible by perfectly matching the SU-8 pedestal height to the focal length of the microlens.
Acknowledgments
G. Grenci acknowledges financial support for this work from the Mechanobiology Institute (MBI), National University of Singapore.
Disclosures
The authors declare no conflicts of interest.
References
1. T. Kagawa, Y. Ohiso, K. Tateno, O. Tadanaga, H. Uenohara, and C. Amano, “850-nm VCSEL arrays for optical interconnection and transmission applications,” in LEOS 2000 IEEE Lasers and Electro-Optics Society 2000 Annual Meeting (LEOS 2000), 800–801 (2000).
2. A. Larsson, “Advances in VCSELs for Communication and Sensing,” IEEE J. Sel. Top. Quantum Electron. 17(6), 1552–1567 (2011). [CrossRef]
3. D. Francis, H.-L. Chen, W. Yuen, G. Li, and C. Chang-Hasnain, “Monolithic 2D-VCSEL array with >2 W CW output power,” in Conference Digest. ISLC 1998 NARA. 1998 IEEE 16th Int’l Semicon. Laser Conf., 99–100 (1998).
4. M. Grabherr, M. Miller, R. Jager, R. Michalzik, U. Martin, H. J. Unold, and K. J. Ebeling, “High-power VCSELs: single devices and densely packed 2-D-arrays,” IEEE J. Sel. Top. Quantum Electron. 5(3), 495–502 (1999). [CrossRef]
5. M. Miller, M. Grabherr, R. Jager, and K. J. Ebeling, “High-power VCSEL arrays for emission in the watt regime at room temperature,” IEEE Photonics Technol. Lett. 13(3), 173–175 (2001). [CrossRef]
6. L. Goldberg, C. McIntosh, and B. Cole, “VCSEL end-pumped passively Q-switched Nd:YAG laser with adjustable pulse energy,” Opt. Express 19(5), 4261–4267 (2011). [CrossRef]
7. A. R. Weidberg, “VCSEL reliability in ATLAS and development of robust arrays,” J. Instrum. 7(01), C01098 (2012). [CrossRef]
8. K. J. Ebeling, R. Michalzik, and H. Moench, “Vertical-cavity surface-emitting laser technology applications with focus on sensors and three-dimensional imaging,” Jpn. J. Appl. Phys. 57(8S2), 08PA02 (2018). [CrossRef]
9. T. R. Fanning, J. Maynard, C. J. Helms, L. Yang, D. Podva, J. Earls, J. Lopez, M. Chung, J. Foresi, M. E. Warren, and G. Bacchin, “Performance, manufacturability, and qualification advances of high-power VCSEL arrays at TriLumina Corporation,” Proc. SPIE 11300, 1130002 (2020). [CrossRef]
10. S. Okur, M. Scheller, J.-F. Seurin, A. Miglo, G. Xu, D. Guo, R. V. Leeuwen, B. Guo, H. Othman, L. Watkins, and C. Ghosh, “High-power VCSEL arrays with customized beam divergence for 3D-sensing applications,” Proc. SPIE 10938, 109380F (2019). [CrossRef]
11. A. Kroner, I. Kardosh, F. Rinaldi, and R. Michalzik, “Towards VCSEL-based integrated optical traps for biomedical applications,” Electron. Lett. 42(2), 93–94 (2006). [CrossRef]
12. E. M. Strzelecka, G. D. Robinson, L. A. Coldren, and E. L. Hu, “Fabrication of refractive microlenses in semiconductors by mask shape transfer in reactive ion etching,” Microelectron. Eng. 35(1-4), 385–388 (1997). [CrossRef]
13. D. Xie, X. Chang, X. Shu, Y. Wang, H. Ding, and Y. Liu, “Rapid fabrication of thermoplastic polymer refractive microlens array using contactless hot embossing technology,” Opt. Express 23(4), 5154–5166 (2015). [CrossRef]
14. X. Zhu, L. Zhu, H. Chen, M. Yang, and W. Zhang, “Fabrication of multi-scale micro-lens arrays on hydrophobic surfaces using a drop-on-demand droplet generator,” Opt. Laser Technol. 66, 156–165 (2015). [CrossRef]
15. D. Zhang, Q. Xu, C. Fang, K. Wang, X. Wang, S. Zhuang, and B. Dai, “Fabrication of a Microlens Array with Controlled Curvature by Thermally Curving Photosensitive Gel Film beneath Microholes,” ACS Appl. Mater. Interfaces 9(19), 16604–16609 (2017). [CrossRef]
16. Q. Xu, B. Dai, Y. Huang, H. Wang, Z. Yang, K. Wang, S. Zhuang, and D. Zhang, “Fabrication of polymer microlens array with controllable focal length by modifying surface wettability,” Opt. Express 26(4), 4172–4182 (2018). [CrossRef]
17. B. Reig, V. Bardinal, J.-B. Doucet, E. Daran, T. Camps, M. Aufray, A. Lamure, and C. Tendero, “Study of SU-8 reliability in wet thermal ambient for application to polymer micro-optics on VCSELs,” Jpn. J. Appl. Phys. 53(8S2), 08MC03 (2014). [CrossRef]
18. C. Levallois, V. Bardinal, C. Vergnenègre, T. Leïchlé, T. Camps, E. Daran, and J.-B. Doucet, “VCSEL collimation using self-aligned integrated polymer microlenses,” Proc. SPIE 6992, 69920W (2008). [CrossRef]
19. V. Bardinal, B. Reig, T. Camps, C. Levallois, E. Daran, C. Vergnenegre, T. Leichle, G. Almuneau, and J. B. Doucet, “Spotted Custom Lenses to Tailor the Divergence of Vertical-Cavity Surface-Emitting Lasers,” IEEE Photonics Technol. Lett. 22(21), 1592–1594 (2010). [CrossRef]
20. V. Bardinal, T. Camps, B. Reig, S. Abada, E. Daran, and J.-B. Doucet, “Advances in Polymer-Based Optical MEMS Fabrication for VCSEL Beam Shaping,” IEEE J. Sel. Top. Quantum Electron. 21(4), 41–48 (2015). [CrossRef]
21. J.-H. Lee, W.-S. Choi, K.-H. Lee, and J.-B. Yoon, “A simple and effective fabrication method for various 3D microstructures: backside 3D diffuser lithography,” J. Micromech. Microeng. 18(12), 125015 (2008). [CrossRef]
22. C. Reardon, A. D. Falco, K. Welna, and T. Krauss, “Integrated polymer microprisms for free space optical beam deflecting,” Opt. Express 17(5), 3424–3428 (2009). [CrossRef]
23. Q.-S. Li, L.-J. Wang, Z.-N. Tian, X.-F. Lin, T. Jiang, J. Zhang, X. Zhang, J.-H. Zhao, A.-W. Li, and L. Qin, “Direct integration of aspherical microlens on vertical-cavity surface emitting laser emitting surface for beam shaping,” Opt. Commun. 300, 269–273 (2013). [CrossRef]
24. Kayaku Microchem, “SU-8 3000 Permanent Epoxy Negative Photoresist,” (Kayaku Microchem, 2019). https://kayakuam.com/wp-content/uploads/2019/09/SU-8-3000-Data-Sheet.pdf
25. The Dow Chemical Company, “CYCLOTENE 4000 Series Advanced Electronics Resins (Photo BCB)” (Dow, 2012). http://msdssearch.dow.com/PublishedLiteratureDOWCOM/dh_08b4/0901b803808b4eda.pdf
26. E. Mitri, G. Birarda, L. Vaccari, S. Kenig, M. Tormen, and G. Grenci, “SU-8 bonding protocol for the fabrication of microfluidic devices dedicated to FTIR microspectroscopy of live cells,” Lab Chip 14(1), 210–218 (2014). [CrossRef]
27. Q. Chen, G. Li, and J. Zhao, “Curved SU-8 structure fabrication based on the acid-diffusion effect,” in 2011 IEEE 24th International Conference on Micro Electro Mechanical Systems (2011), pp. 225–228.
28. I. Avramov, “Relationship between diffusion, self-diffusion and viscosity,” J. Non-Cryst. Solids 355(10-12), 745–747 (2009). [CrossRef]
29. R. Pregla, W. von Reden, H. J. W. M. Hoekstra, and H. V. Baghdasaryan, “Beam propagation methods,” in Photonic Devices for Telecommunications, G. Guekos, ed. (Springer, 1999), pp. 35–65.
30. I.-S. Chung, P. Debernardi, Y. T. Lee, and J. Mørk, “Transverse-mode-selectable microlens vertical-cavity surface-emitting laser,” Opt. Express 18(5), 4138–4147 (2010). [CrossRef]
