Open Access
Add to BibSonomy
Abstract
We report on UV laser photoresist mask writing as a tool for fabricating integrated optical waveguides and devices. Using 375 nm laser light and a pneumatically controlled direct writing stage, we defined mask features into a 250-nm-thick negative photoresist layer on a silicon nitride film on an oxidized silicon substrate. We investigated the feature size and edge roughness for different laser powers. Using the photoresist mask layer and reactive ion etching, we patterned high-refractive-index-contrast silicon nitride strip waveguides and devices with varying waveguide widths and gaps. We report on the structural and transmission characteristics of a directional coupler, Sagnac interferometer, and ring resonator and demonstrate 50/50 coupling at 1510 nm, 20 dB transmission drop at 1580 nm, and a Q factor of ∼13,000 at 1576 nm, respectively. These results demonstrate that this technique can be applied to a variety of thin film materials and substrates for inexpensive and rapid prototyping of integrated photonic devices.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Direct writing has been demonstrated as a flexible tool for altering material properties and structuring at the nano and micro scale without the need of a physical mask. The use of continuous or pulsed lasers and electron beams has been applied to a variety of material systems to fabricate integrated photonic devices [1,2]. Electron-beam lithography (EBL) has finer resolution in comparison to laser direct write techniques which is advantageous when fabricating features with sub-micrometer dimensions or close proximity [3]. Laser direct write (LDW) is low-cost in comparison to EBL, thus is often more accessible and desirable in the research environment where fast turnaround and rapid and inexpensive prototyping of devices is advantageous. Ultra-violet (UV) pulsed, continuous and interference LDW lithography has been demonstrated to utilize the photochemical response of various materials to permanently induce refractive index changes in order to pattern waveguides and other structures [4–13]. LDW lithography has also been shown to be a flexible platform for structuring photoresist waveguides and microfluidic channels with the use of chemical development for a variety of applications [14–18].
Nevertheless, most LDW research and development has focused on writing waveguide structures directly in the material. The resulting small refractive index change typically limits the selection of materials systems to low-refractive-index-contrast waveguides with large core sizes and bend radii, or alternatively, polymer waveguides. To enable the technique to be applied more widely to a variety of thin film photonic materials and substrates, UV LDW can be instead applied to define a photoresist mask layer on the chip, followed by pattern transfer to the layer below through selective etching. Many established and emerging low- to high-refractive-index-contrast waveguide systems have feature sizes achievable by LDW and their research and development could benefit from UV laser resist mask patterning [19–23].
Here, we demonstrate UV laser resist mask writing as a tool to pattern silicon nitride optical waveguides on silicon. We use a continuous UV LDW system to write features in a negative photoresist layer on a silicon nitride (Si3N4) thin film to act as an etch mask for the definition of Si3N4 waveguides and devices. We investigate feature sizes, edge roughness and gaps versus UV writing parameters. Compared to other work with pulsed laser resist-mask writing [24] we show that smaller resolution is achievable using a UV laser written photoresist mask and that this method offers more flexibility in terms of material choice. The results presented here indicate that UV laser resist mask writing offers a good compromise between cost, resolution, and flexibility next to pulsed LDW and other laser resist-mask writing techniques for prototyping integrated photonic devices.
2. Fabrication
2.1 Experimental setup
The exposure setup consists of a commercial table top patterning system which utilizes a 375 nm laser diode and focusing optics to write a predefined design onto a sample. The write-lens assembly utilizes pneumatic and piezo control with a telescopic lens to focus the beam to the scanning stage, which is mechanically controlled by a roller bearing and air bearing for independent x- and y-axis translation, respectively, during exposure. An internal processor automatically determines the scan speed and write time from run to run, which increases with exposed area. Figure 1 shows a diagram of the exposure setup including the UV laser and focusing optics, the mechanically-controlled vacuum stage for moving the sample and scattering of UV light within the photoresist layer. This scattering plays an important role in feature definition at the relatively higher energies that are required for negative photoresist exposure as compared to positive photoresists. Using negative photoresist also minimizes the area required for exposure, lowering the write time when compared to using positive photoresists in this process.
Fig. 1. Diagram of the UV laser write system showing lateral UV light scattering during exposure of negative photoresist. The quoted system minimum feature size is 0.9 µm with 120 σ nm edge roughness, an estimated write speed of 5 mm2/minute, and output powers of 0–70 mW at 375 nm. The maximum substrate diameter is 10 cm.
2.2 Negative photoresist mask patterning
We used the UV write setup to pattern a negative photoresist mask layer on a 7.5-cm-diameter silicon substrate with a 6-µm thermal SiO2 layer and a 100-nm-thick low-pressure chemical vapor deposited (LPCVD) silicon nitride thin film. We selected NR7-250P negative photoresist, as the mask layer. Primer and the resist layer were spun on top of the substrate and material stack at 3000 rpm for 30 seconds. This leads to a 250-nm-thick resist layer which was then baked at 120°C for 3 minutes before UV exposure. After UV exposure, we baked the sample at 150°C for 5 minutes to harden the polymerized bonds and ensure high chemical contrast during development. We developed the sample at room temperature in a solution of 3:1 RD6 developer to deionized water for 10 seconds with manual agitation.
Standard GDSII layout software was used to create various designs, which are translated into the photoresist layer by the movement of the scanning stage during exposure. In this manner, maskless lithography is performed without the use of a virtual mask, and the UV laser writes the pattern directly into the photoresist layer. We defined straight waveguide features in the negative photoresist layer with designed widths varying from 0.6 to 1.4 µm and lengths of 2 cm using UV exposure powers ranging from 40 to 70 mW, as shown in Fig. 2. The designed waveguide features were rotated by 45° in order to simultaneously account for the resolution difference between each axis of movement by the independent air and roller bearings of the write stage during exposure. In order to fabricate waveguide designs in arbitrary x and y directions without any difference in resolution translating to the structure during exposure, we developed a high power write process whereby the resolution is limited by lateral scattering of UV light in the resist layer. Higher exposure powers also ensure the feature edge roughness is dominated by optical scattering of the UV diode laser, rather than the mechanical motion of the stage. This methodology gives rise to a designed-to-fabricated feature increase, which has been characterized via scanning electron microscopy (SEM), as displayed in Fig. 2. Figure 2(a) demonstrates these results for waveguide features rotated by 45° relative to the write stage, showing the evolution of feature resolution at various exposure powers. Figures 2(b) and 2(c) show representative cross-sections of the resist mask profile after exposure and development of y axis straight and rotated features respectively.
Fig. 2. (a) Various widths of exposed and developed negative photoresist on Si3N4 films on thermally-oxidized silicon substrates. Features at narrow designed widths and lower energies demonstrate the x-y limitations of the stage’s mechanical motion. (b) Resist feature cross section for 0.45 µm designed width exposed using only y-axis of stage motion at 70 mW. (c) Resist feature cross section for 0.65 µm designed width exposed using both x-y stage motion at 70 mW.
The width and sidewall roughness as measured from the SEM images are plotted in Fig. 3(a) and Fig. 3(b), respectively. As shown in Fig. 3(a), the minimum feature size increases with exposure power, and the relative feature expansion with respect to designed width is constant among different powers, varying from 2.1 µm at 50 mW to 2.2 µm at 70 mW. To quantify the LER in units of σ nm, image processing was used following the methodology provided by [25] to express the area of saturated pixels surrounding each straight line as line-edge roughness (LER) in units comparable to, and values below the quoted system specifications mentioned in Fig. 1. For the formation of features which are dominated by UV scattering as opposed to mechanical stage motion, it was found that higher powers of 60 and 70 mW are required during exposure. As seen in Fig. 3(b), 40 and 50 mW demonstrate significantly higher edge roughness at smaller designed widths, which we attribute to stage motion, whereas 60 and 70 mW show relatively width-independent LER, expected by a scattering dominated process.
Fig. 3. (a) Measured widths vs. designed width for photoresist features exposed at various laser diode powers. Inset: SEM of the 60 mW exposed set of features. (b) Calculated edge roughness vs. designed width exposed at various laser diode powers. Inset: SEM of resist feature for 0.8 µm designed width and 60 mW exposure power.
Characterizing the designed-to-exposed feature-width increase allows waveguides to be designed accounting for a nominal decrease in all feature sizes to achieve targeted dimensions. It is expected smaller features can be made than those shown by using designed widths below 0.6 µm, and solely exposing features using the y axis rather than x to utilize the finer control provided by the air bearing for y motion stage control, as shown in Fig. 2(b). After analyzing the edge roughness, it is clear that operating above a certain power threshold is desirable for the formation of smooth sidewalls. To discover the optimal exposure power an investigation into gap resolution after waveguide fabrication was performed.
2.3 Waveguide design and fabrication
We selected UV laser powers of 60 and 70 mW to write mask features and pattern silicon nitride waveguides with varying widths and gaps. The exposures were carried out at 60 and 70 mW to optimize for sidewall roughness. The varying gaps were studied to further investigate the patterning technique and its capability for fabricating waveguides with close proximity to one another. After defining the resist mask, we etched the Si3N4 film using reactive ion etching (RIE) with a CF4-O2 plasma. The etch parameters included gas flow rates of 22.5 sccm CF4 and 2.5 sccm O2 with 120 W of forward power at a pressure of 25 mTorr. The etching selectivity of Si3N4 to photoresist was determined to be 1.2:1.0 with a selectivity of 2:1 for Si3N4 to SiO2. An etch time of 2 minutes was used with a 60 nm/min Si3N4 etch rate to ensure an etch depth of 100 nm. After stripping the resist in acetone and cleaning in isopropyl alcohol, a 1.8-µm-thick plasma-enhanced chemical vapour deposited (PECVD) SiO2 film was deposited on top of the Si3N4 waveguides with 25 sccm O2 and 2.5 sccm SiH4/Ar (70/30 % dilute) at 2.5 mTorr with 500 W forward power. After top cladding deposition the sample was cleaved to form end facets. The fabrication steps are shown in Fig. 4, as well as the electric field profile of the fundamental transverse electric (TE) mode at 1550 nm wavelength, calculated using a finite element method (FEM) modesolver.
Fig. 4. Si3N4 waveguide fabrication steps and calculated electric field profile of the fundamental TE mode at 1550 nm for a 1.6 µm × 0.1 µm Si3N4 strip waveguide.
SEM images of fabricated waveguides prior to top-cladding deposition are shown in Fig. 5. Figure 5(a) shows a top-down view of a variety of Si3N4 waveguides with varying designed widths and gaps at 70 mW exposure power. Figures 5(b) and 5(c) show SEM waveguide cross-section images for single, and adjacent waveguides respectively. It can be seen that the minimum gap starts to close at designed gap widths of 4.0, 5.2 and 5.8 µm for designed waveguide widths of 0.7, 0.8 and 1.0 µm, respectively. The gap width after fabrication versus the designed gap is plotted in Fig. 6 for write powers of 60 and 70 mW. The realized gap width is influenced by the combined effects of stage motion resolution, exposure-power-dependent waveguide width expansion, and designed gap width collectively. By selecting the correct write parameters, minimum gaps in the range of 1–2 µm are possible. It is observed by comparing identical designed waveguide widths, that on average ∼170 nm smaller gaps are achievable with 70 mW exposure powers compared to 60 mW, with larger designed gaps required to compensate for the feature increase at higher powers.
Fig. 5. (a) Various fabricated Si3N4 waveguides with increasing width and gap demonstrating gap resolution at 70 mW exposure power. (b) Final fabricated Si3N4 waveguide cross section for a 0.7 µm designed width waveguide fabricated using 70 mW UV laser power. (c) Parallel fabricated waveguides in a coupler region for 0.7 µm designed widths at 70 mW with a designed gap of 5 µm.
Fig. 6. Measured gaps between fabricated waveguides of varying designed widths at 60 and 70 mW exposure powers. The measured waveguide widths are 2.1–2.8, 2.9–3.3 and 3.3–3.9 µm at 60 mW and 3.0–3.6, 3.4–4.1 and 4.1–4.7 µm at 70 mW for 0.7, 0.8 and 1.0 µm designed widths, respectively. Inset SEM image of waveguides with 1.0 and 7.0–6.2 µm designed width and gap respectively, with 70 mW exposure power.
An understanding of the width and gap resolutions using the demonstrated methodology enabled the design of various proof-of-concept passive waveguide structures for test. We selected standard passive integrated optical components, including a directional coupler, ring resonator and Sagnac loop mirror [26]. Waveguide mode and bend loss simulations were carried out to determine a minimum bending radius of 300 µm, which was chosen for a point coupled ring resonator, as well as supermode analysis [27] to discover a coupling length of 40 µm for a single mode 50/50 splitter at 1550 nm both based on a waveguide width of 1.6 and a gap of 1.1 µm. The coupling region of the Sagnac loop mirror was based on and identical to the included 50/50 splitter, with a loop radius identical to the ring resonator. These feature sizes were achieved using 70 mW exposure powers and designed widths and gaps of 0.7 µm and 2.8 µm respectively. We note that even smaller feature sizes in the straight sections and designed gaps than those shown in Fig. 3(a) and Fig. 6 were achieved due to the independent use of the y axis during exposure of the waveguides lengthwise, and resolution-limited x axis motion of the roller bearing for gap formation. By keeping the coupler gaps parallel with respect to the y axis rather than rotated, it is expected that smaller gaps than those shown here can be achieved. Variations in designed widths to compensate for varying axis resolution for bent waveguides can also be considered.
3. Optical measurements
We measured the transmission spectra of the waveguides and devices using an edge coupling setup. The setup included a tunable 1510–1640 nm laser source, polarization maintaining fiber paddles for selecting TE polarization, lensed input and output fibers with 2.5-µm spot size for coupling to and from the chip and a photodetector for measuring the transmitted signal. Figure 7 shows the overall footprint of the fabricated devices as well as each coupling region prior to cladding revealing waveguide coupling gaps of 1.1, 2.1, and 1.5 µm for the Sagnac loop, directional coupler, and ring resonator respectively. Figure 8(a) shows the transmission spectra for each arm of the directional coupler which was measured with a fixed input arm and output to a through and cross port to measure the coupled power from 1510–1580 nm. As seen from the transmission spectra in Fig. 8(a), a 50/50 splitting ratio is achieved at 1510 nm, shown by the similar signal powers. Figure 8(b) shows the transmission spectra for the Sagnac loop interferometer, demonstrating a 20 dB extinction ratio (ER) at 1580 nm, where the directional coupler is operating as a 50/50 splitter, causing the reflection of light. Figure 8(c) shows the transmission spectra for the ring resonator demonstrating resonances from 1510–1580 nm, with a wavelength-dependent ER ranging from 0.5 to 0.9 dB and a free spectral range (FSR) of ∼1 nm. Note the measured insertion loss includes fiber loss, fiber-chip coupling loss and waveguide loss not subtracted from the measurements. A Lorentzian fit of a resonant dip is included in Fig. 8(d) at 1576 nm which was used to characterize and extract an internal quality factor of ∼13,000, assuming the ring is under coupled due to the large gap and small ER.
Fig. 7. SEM images of total footprint (bottom) and coupling regions (top) for Si3N4 integrated optical devices fabricated using UV laser resist mask patterning. (i) 300-µm-radius Sagnac loop interferometer with 40 µm coupling length and with (w) and gap (g) of 1.9 and 1.1 µm respectively. (ii) Directional coupler with 40 µm coupling length and fabricated w and g of 1.1 and 2.1 µm respectively. (iii) 300-µm-radius point coupled ring resonator with w and g of 1.6 and 1.5 µm respectively.
Fig. 8. Transmission spectra for Si3N4 integrated optical devices fabricated using UV laser resist mask patterning. (a) Directional coupler showing 50/50 coupling at 1510 nm for 40 µm coupling length. (b) 300-µm-radius Sagnac loop interferometer with repeated coupler length, demonstrating a 20 dB extinction ratio at ∼1580 nm (c) 300-µm-radius point coupled ring resonator. (d) Lorentzian resonance fit demonstrating a Q of 12,980 at 1576.2 nm.
Based on these proof-of-concept results, we propose that various waveguiding structures and photonic devices can be readily prototyped using this process. Investigation of UV patterning of additional resist mask materials, optimization of the resist processing (e.g. thickness and bake times) and improved waveguide etch recipes can lead to lower waveguide losses and higher Q factor resonators. This technique is also promising for post-processing of novel device layers with relaxed feature size and alignment tolerances onto existing photonic integration platforms for new functionalities (e.g. in silicon photonic microsystems).
4. Conclusion
We have introduced UV laser resist-mask patterning as a low-cost prototyping technique for the development of integrated optical waveguides and devices. Using a continuous UV laser source to pattern a negative photoresist etch mask, we defined waveguides in a silicon nitride thin film on silicon. We performed dose tests and characterized the negative resist features to measure the increase in designed feature dimensions and quantify the edge roughness. We discovered that higher energy exposures with scattering-limited sidewall roughness is ideal for negative photoresist development for high contrast, straight, smooth sidewalls. The gap resolution was also investigated before fabricating various devices to ensure minimum spacing between adjacent waveguides is achievable. A directional coupler, Sagnac interferometer and ring resonator were fabricated and measured demonstrating 50/50 coupling at 1510 nm, 20 dB drop at 1580 nm, and a Q factor of ∼13,000 at 1576 nm respectively. The ability to prototype passive optical components was demonstrated with promise towards applying the patterning technique to a wide variety of applications and material systems.
Funding
Canada Foundation for Innovation (CFI) (35548, 21134); Natural Sciences and Engineering Research Council of Canada (NSERC) (06423, 494306).
Acknowledgements
We thank Doris Stevanovic and Shahram Tavakoli of the Centre for Emerging Device Technologies (CEDT) at McMaster University for their assistance with fabrication.
References
1. R. R. Gattass and E. Mazur, “Femtosecond laser micromachining in transparent materials,” Nat. Photonics 2(4), 219–225 (2008). [CrossRef]
2. Y. Huang, G. T. Paloczi, A. Yariv, C. Zhang, and L. R. Dalton, “Fabrication and replication of polymer integrated optical devices using electron-beam lithography and soft lithography,” J. Phys. Chem. B 108(25), 8606–8613 (2004). [CrossRef]
3. T. Barwicz, M. A. Popovic, P. T. Rakich, M. R. Watts, H. A. Haus, E. P. Ippen, and H. I. Smith, “Microring-resonator-based add-drop filters in SiN: fabrication and analysis,” Opt. Express 12(7), 1437–1442 (2004). [CrossRef]
4. R. R. Krchnavek, G. R. Lalk, and D. H. Hartman, “Laser direct writing of channel waveguides using spin-on polymers,” J. Appl. Phys. 66(11), 5156–5160 (1989). [CrossRef]
5. L. Eldada, C. Xu, K. M. T. Stengel, L. W. Shacklette, and J. T. Yardley, “Laser-fabricated low-loss single-mode raised-rib waveguiding devices in polymers,” J. Lightwave Technol. 14(7), 1704–1713 (1996). [CrossRef]
6. M. Rosenberger, B. Schmauss, and R. Hellmann, “UV-writing of a superstructure waveguide Bragg grating in a planar polymer substrate,” Sensors 17(9), 1964 (2017). [CrossRef]
7. J. Missinne, A. Vasiliev, A. Elmogi, N. Teigell Beneitez, E. Bosman, B. Van Hoe, and G. Van Steenberge, “Bragg grating sensors in laser-written single mode polymer waveguides,” Procedia Eng. 120, 878–881 (2015). [CrossRef]
8. M. Svalgaard, “Direct writing of planar waveguide power splitters and directional couplers using a focused ultraviolet laser beam,” Electron. Lett. 33(20), 1694–1695 (1997). [CrossRef]
9. G. D. Marshall, P. Dekker, M. Ams, J. A. Piper, and M. J. Withford, “Directly written monolithic waveguide laser incorporating a distributed feedback waveguide-Bragg grating,” Opt. Lett. 33(9), 956–958 (2008). [CrossRef]
10. S. Pelli, G. C. Righini, A. Scaglione, M. Guglielmi, and A. Martucci, “Direct laser writing of ridge optical waveguides in silica-titania glass sol-gel films,” Opt. Mater. 5(1-2), 119–126 (1996). [CrossRef]
11. M. Svalgaard, C. V. Poulsen, A. Bjarklev, and O. Poulsen, “Direct UV writing of buried single mode channel waveguides in Ge-doped silica films,” Electron. Lett. 30(17), 1401–1403 (1994). [CrossRef]
12. E. Zgraggen, I. Murat Soganci, F. Horst, A. La Porta, R. Dangel, B. J. Offrein, S. A. Snow, J. K. Young, B. W. Swatowski, C. M. Amb, O. Scholder, R. Broennimann, U. Sennhauser, and G.-L. Bona, “Laser direct writing of single-mode polysiloxane optical waveguides and devices,” J. Lightwave Technol. 32(17), 3036–3042 (2014). [CrossRef]
13. G. D. Marshall, M. Ams, and M. J. Withford, “Direct laser written waveguide-Bragg gratings in bulk fused silica,” Opt. Lett. 31(18), 2690–2691 (2006). [CrossRef]
14. M. Dubov, S. R. Natarajan, J. A. R. Williams, and I. Bennion, “Mask-less lithography for fabrication of optical waveguides,” Proc. SPIE 6881, 688110 (2008). [CrossRef]
15. V. Vaidyanathan and S. Wang, “A maskless fabrication approach of integrated optical waveguides for engineering technology students,” presented at the American Society for Engineering Education Annual Conference & Exposition, Chicago, Illinois, 18–21 June 2006.
16. J. C. Dingeldein, Direct Write Fabrication of Waveguides and Interconnects for Optical Printed Wiring Boards (Ph.D. Thesis, Michigan Technological University, 2012).
17. K. Kruse and C. Middlebrook, “Laser-direct writing of single mode and multi-mode polymer step index waveguide structures for optical backplanes and interconnection assemblies,” Photonics Nanostruct. Fundam. Appl. 13, 66–73 (2015). [CrossRef]
18. H. Hartwig, E. Sergeeva, N. Thomas, and D. Hohlfeld, “Precision laser direct writing of polymer optical waveguides and optofluidic systems,” in DGaO Proceedings (The German Branch of the European Optical Society, 2016), A16.
19. M. Belt, M. L. Davenport, J. E. Bowers, and D. J. Blumenthal, “Ultra-low-loss Ta2O5-core/SiO2-clad planar waveguides on Si substrates,” Optica 4(5), 532–536 (2017). [CrossRef]
20. K. Wörhoff, J. D. B. Bradley, F. Ay, D. Geskus, T. P. Blauwendraat, and M. Pollnau, “Reliable low-cost fabrication of low-loss Al2O3:Er3+ waveguides with 5.4-dB optical gain,” IEEE J. Quantum Electron. 45(5), 454–461 (2009). [CrossRef]
21. S. A. Vázquez-Córdova, M. Dijkstra, E. H. Bernhardi, F. Ay, K. Wörhoff, J. L. Herek, S. M. García-Blanco, and M. Pollnau, “Erbium-doped spiral amplifiers with 20 dB of net gain on silicon,” Opt. Express 22(21), 25993–26004 (2014). [CrossRef]
22. J. F. Bauters, M. J. R. Heck, D. D. John, M.-C. Tien, W. Li, J. S. Barton, D. J. Blumenthal, J. E. Bowers, A. Leinse, and R. G. Heideman, “Ultra-low-loss single-mode Si3N4 waveguides with 0.7 dB/m propagation loss,” in 37th European Conference and Exposition on Optical Communications, OSA Technical Digest, OSA Technical Digest (CD) (Optical Society of America, 2011), paper Th.12.LeSaleve.3.
23. G. C. Righini and A. Chiappini, “Glass optical waveguides: a review of fabrication techniques,” Opt. Eng. 53(7), 071819 (2014). [CrossRef]
24. A. Kiani, K. Venkatakrishnan, B. Tan, and V. Venkataramanan, “Maskless lithography using silicon oxide etch- stop layer induced by megahertz repetition femtosecond laser pulses,” Opt. Express 19(11), 10834–10842 (2011). [CrossRef]
25. J. R. Bickford, “Automatic stripe analysis tool” (Army Research Laboratory, Report No. ARL-TR-6469, 2013). https://www.arl.army.mil/arlreports/2013/technical-report.cfm?id = 6765
26. Y. Zhang, S. Yang, H. Guan, A. E.-J. Lim, G.-Q. Lo, P. Magill, T. Baehr-Jones, and M. Hochberg, “Sagnac loop mirror and micro-ring based laser cavity for silicon-on-insulator,” Opt. Express 22(15), 17872–17879 (2014). [CrossRef]
27. L. Chrostowski and M. Hochberg, Silicon Photonics Design: From Devices to Systems (Cambridge University Press, 2015).
