Open Access
Add to BibSonomyAbstract
We present inductively-coupled-plasma, reactive-ion-etching (ICP-RIE) techniques with 2 orders of magnitude difference in etch rates, for the AlxGa1-xAs material system. These precise etching processes are used to produce waveguides in a multi-guide vertical integration (MGVI) AlxGa1-xAs chip. The MGVI AlxGa1-xAs chip vertically integrates multiple guiding layers that usually have different material properties. The fabrication of these chips requires precise and anisotropic etching. The first etching recipe used BCl3 and achieved an etch rate of 0.25 nm/s while the second one used Cl2/N2 gases and achieved an etch rate of more than 20 nm/s. Simple AlxGa1-xAs nanowaveguides of 800 nm width were fabricated using these recipes. We measured a propagation loss of 6.7 dB/cm at the wavelength of 850 nm.
© 2017 Optical Society of America
1. Introduction
Photonic integrated circuits (PICs) have been advancing rapidly during the last two decades. Silicon-on-insulator (SOI) has received much attention for its transparency at telecom wavelengths and ability to make compact circuits due to tight confinement of the optical mode. However, silicon has an indirect bandgap, which makes monolithic integration of lasers very difficult and it is not suitable for applications in the 785 – 850 nm spectral range. Heterogeneous integration of III-V semiconductors onto silicon chips provides a solution but significantly increases the difficulty of fabrication [1]. The 785 – 850 nm spectral window has several potential applications where monolithically integrated ciruits are needed. For non-invasive bio sensing, it is the wavelength range where hemoglobin has the lowest absorption [2]. As current data center communications systems use 850 nm vertical cavity lasers, monolithic integration would enable efficient wavelength division multiplexing devices. AlGaAs/GaAs has also been identified as the promising material system for quantum optics [3–5]. The integration of pump lasers, nonlinear optical elements and signal processing waveguides opens the possibility of a true quantum optics technology.
Currently, PICs are based on many different integration schemes including etch-and-regrowth, quantum-well intermixing (QWI) and hybrid material integration [6,7]. These methods have several drawbacks that either limit the integration complexity, or incur high fabrication costs. One alternative integration scheme is multi-guide vertical-integration (MGVI). It is a generalization of the twin-guide scheme but allows more comprehensive photonic integration [8,9]. The MGVI scheme has several advantages over other photonic integration methods. By far, the key advantage of MGVI is that only a single epitaxial growth step is required throughout the fabrication process run. This saves substantial cost in several ways: 1) expensive epitaxial regrowth processes are not necessary, 2) device yield from a single wafer is higher because imperfections and nonuniformity from regrowth are eliminated, and 3) wafer growth and lithography are decoupled and can be performed at separate facilities.
In the MGVI scheme, multiple waveguiding levels are stacked on top of each other with each level hosting a different type of device. The epitaxial layers within each level are optimized for those particular functions. All devices and waveguides are defined lithographically at each waveguiding level using fabrication steps that only involve etching processes. Because of the variety of functionalities, the geometrical dimensions of the devices can vary significantly. Fabrication of the chips relies on highly precise and anisotropic etching. In this paper, we present ICP-RIE fabrication techniques for MGVI AlxGa1-xAs based PICs. We developed two dry etch recipes with almost two orders of magnitude difference in etch rates. The first one used BCl3 and achieved an etch rate of 0.25 nm/s while the second one used Cl2/N2 gases and achieved an etch rate of more than 20 nm/s. We fabricated simple AlxGa1-xAs nanowaveguides of 800 nm width using these recipes and measured a propagation loss of 6.7 dB/cm at the wavelength of 850 nm. While previous attempts at building MGVI chips were mainly made in the InGaAsP material system, our work in this paper is an important step forward in demonstrating the feasibility of making MGVI chips in the AlxGa1-xAs material system.
2. Inductively-coupled plasma (ICP) RIE
Dry etching techniques are essential for the precise fabrication of optoelectronic devices, which require accurate control of etch rate, selectivity, structural profile, and surface morphology while minimizing device damage. The advantages of ICP-RIE are well known in the literature. In an ICP etcher, the ion energy incident on the wafer can be effectively decoupled from plasma generation by independently applying rf power to the wafer chuck. This allows for the possibility of low etch-induced damage at high etch rates. There has been tremendous efforts in analyzing etch characteristics in various gas mixtures, such as CCl2F2/BCl3/Ar, Cl2/BCl3/Ar, Cl2/BCl3/N2, etc. N2 is often used as an alternative of Ar to control the chemical acitivity of the gas composition [10]. In our study, we choose Cl2/N2 as the candidate gas composition for fast etching and BCl3 for slow etching.
The wafer used in the experiment consists of five AlxGa1-xAs layers with different aluminum concentrations. They are a 5 nm GaAs, a 120 nm Al0.2Ga0.8As, a 250 nm Al0.34Ga0.66As, a 4 μm Al0.5Ga0.5As and a 3.5 μm Al0.63Ga0.37As from top to bottom and are grown on an n-type (100) GaAs substrate. The first three layers on the top are suitable for implementing high refractive index contrast waveguides for on-chip signal processing. The two bottom layers are good for making large-dimension waveguides that facilitate end-fire coupling. Schematic of the wafer structure is shown in Fig. 1(a). Modal fields were simulated using Lumerical® MODE solutions. The electric field intensity of TE modes of high confinement waveguide and large waveguide is shown in Fig. 1(b) and 1(c), respectively.
Fig. 1 a) Wafer schematic showing layer thicknesses and refractive indices. b) TE mode profile for a 800 nm wide waveguide in the high confinement upper layer. c) TE mode profile for a 4 μm wide waveguide in the lower layer.
Test samples were prepared using the following process steps. Firstly, a large sample was soaked in acetone for 10 mins and then in IPA for 10 mins. Secondly, the sample was soaked in 30% NH4OH:DI water = 1:10 for 30 seconds and then rinsed with running DI water for 30 seconds. Lastly, either ma-N2400 series photoresist, or HSQ was spin coated on to the sample. Simple waveguides with various widths were exposed using e-beam lithography (EBL). The resulting samples were masked with 500 nm ma-N2405, 1000 nm ma-N2410 or 600 nm HSQ. The samples prepared using HSQ were hard-baked at 250 °C for 60 mins to improve hardness. We cleaved the large sample into smaller pieces with area about 25 mm2. Etching was carried out in Oxford Instruments Plasmalab system 100 with ICP 380 which can handle 4” diameter wafers and has a 380 mm diameter ICP source. Substrate DC bias is independently controlled by an RF generator at 13.56 MHz. The inductively coupled plasma source is a 380 mm diameter coil powered by a 1.7-2.1 MHz generator that can deliver up to 5 kW. Oxford Instruments Plasmalab system 100 with ICP 380 etch sources produce a high density of reactive species at low pressure. All samples were etched for 4 mins unless otherwise noted. The two baseline recipes for fast and slow etch are listed in Table 1.
Table 1. Baseline recipes for fast and slow etch recipes.
2.1 Slow etch recipe with BCl3
In order to fabricate the highly confining waveguides on the upper layer very precise control of the etch depth is required. To achieve this we developed the following BCl3 etch process. It is known that BCl3 scavenges aluminum oxides and eliminates lag times [11]. This is crucial in achieving precise shallow etch because a fast etch recipe with unpredictable lag times inevitably diminishes the accuracy of the etch depth. It has also been reported that BCl3 achieved equi-rate etching for AlxGa1-xAs of different aluminum concentrations [12]. These two factors are important since the slow etch recipe was used in etching the top layers which were prone to native oxides and empassed several AlxGa1-xAs layers. We kept the chamber pressure constant at 5 mTorr in the experiment but varied other parameters, including rf power, ICP power, gas flow rate and temperature. Samples with 500/1000 nm ma-N2405/2410 were used. Etch depths were obtained from at least four locations on the small sample, which had an area of ~25 mm2. Etch rates were calculated as the averaged depths divided by etch times. The same etch depth measurement and etch rate calculation methods were followed throughout this paper.
The etch rate showed little change when the rf power was varied from 50 to 200 W. This can be understood as the dominant etch mechanism was chemical etching. Increasing the rf power basically increased ion energy and physical sputtering, which remained insignificant under the conditions explored. When the ICP power was varied from 300 to 800 W, the etch rate increased significantly due to increased ion density and flux. We recorded the DC bias voltage of the sample chuck. For samples etched using recipes from Fig. 2, sidewalls were found to be overcut when the DC bias was higher than 400 V. A close examination revealed that the overcut emerged because of photoresist reflow which then covered the sidewalls as etching proceeded. As a result, the widths of the waveguides were larger at the bottom of the structure. On the other hand, samples showed undercut sidewalls when DC bias voltage was about 200 V. This could be attributed to no resist reflow and insufficient ion energy for ion assisted chemical etching. Overall, an rf power of 100 W seemed to acquire a balance. Samples etched using recipes from Fig. 2(a) are shown in Fig. 3.
Fig. 2 Etch rate and DC bias voltage as a function of (a) rf power at 600 W ICP and (b) ICP power at 200 W rf, respectively, at 5 mTorr, 50 sccm BCl3 and 10 °C. Samples were etched for 2 mins. ma-N2410 was the mask.
Fig. 3 Scanning electron micrographs of samples etched using rf power (a) 50 W, (b) 100 W and (c) 200W at 5 mTorr, 600 W ICP, 50 sccm BCl3 and 10 °C.
The change of etch rate as a function of gas flow rate was examined. As shown in Fig. 4(a), the gas flow rate didn’t noticeably influence the etch rate at an ICP power level of 300 W. A similar trend was reported by [13] and therefore running excessive gas was discouraged. At an ICP power level of 600 W, the etch rate was about 50% higher at a BCl3 flow rate of 50 sccm owing to an increased ion concentration coupled with a higher ion flux at the elevated ICP power. Overall, the etch rate was reduced by about half when the ICP power was changed from 600 W to 300 W, indicating a diffusion-limited etch: higher ICP power induced a higher ion flux which carried away etch byproducts faster and resulted in higher etch rates. The photoresists ma-N2405 and ma-N2410 were used for the 300 W and 600 W ICP power, respectively. Due to strong physical sputtering, overcut sidewalls resembling Fig. 3(c) were observed in both cases. Therefore, we decreased the rf power to 100 W and set the 300 W ICP power as a ‘new’ baseline. As shown in Fig. 4(b), etch rate was almost invariant at 10, 20 and 30 sccm BCl3 at 10 °C and 20 °C. Figure 6 shows samples that were etched using 10 and 20 sccm BCl3 at 10 °C.
Fig. 4 Etch rate as a function of BCl3 flow rate. (a) Two ICP power levels were examined at 5 mTorr, 200 W rf and 10 °C. 512 V and 500 V bias voltages were recorded for both ICP power levels. (b) Two temperatures were examined at 5 mTorr, 100 W rf and 300 W ICP. 350 V bias voltage was recorded for both temperatures.
To further confirm temperature dependence of etch rate, we varied temperature up to 40 °C. A decline in etch rate can be seen in Fig. 5(a). This trend could be explained as follows. It’s known that ions traversing the ion sheath region would assume the substrate temperature long before they reach the substrate. A higher temperatre resulted in stronger scattering thus fewer ions reached the substrate. While the elevated temperate didn’t increase the chemical reactivity, it led to weaker desorption of byproducts thus lower etch rate. Lowering the ICP power would decrease etch rate and give reasonably low etch rates. As shown in Fig. 5(b), etch rate as low as 2.5 Å/s could be obtained and resulting DC bias was reasonably low. Selectivy of ma-N 2400 series photoresist was larger than 2:1. The rapid drop of etch rate when ICP power was reduced to 200 W could be attributed to a combination of reduced ion flux, which reduced desorption, and insufficient reactive ions, which limited reaction with the substrate. Further exploration was not carried out. Samples etched under these temperatures and ICP power had almost the same profile as shown in Fig. 6.
Fig. 5 (a) Etch rate as a function of temperature at 5 mTorr, 100 W rf, 300 W ICP and 20 sccm BCl3. 350 V bias voltage was recorded. (b) Etch rate as a function of ICP power at 5 mTorr, 100 W rf, 10 sccm BCl3 and 20 °C. 300 V bias voltage was recorded between 50 W and 200 W. 350 V bias voltage was recorded at 300 W ICP.
Fig. 6 SEMs of samples etched using (a) 10sccm and (b) 20sccm BCl3 at 5 mTorr, 100 W rf, 300 W ICP and 10 °C.
2.2 Fast etch recipe with Cl2/N2
The large coupling waveguide in the lower layer requires a deeper etch. To facilitate the fabrication of vertically integrated waveguides, we developed the following fast etch process based on a combination of Cl2/N2. Compared to Ar, N2 is preferred for balancing chemical acitivity and physical sputtering [14] [15]. N2 can also passivate sidewalls and gives more flexibility to control sidewall profile. In addition, in a two-compound composition, it’s easier to see the interplay of chemical etching and physical sputtering. Samples with 600 nm HSQ were used.
Increasing the rf power increases the ion energy and therefore the sputter rate. In order to protect the passivation layer, the rf power needs to be reasonally low. Another reason is Cl2 dissociates easily and the chemical reacitivy of the gas composition can be too high to achieve anisotropic etching. We found a quadratic relationship between etch rate and chlorine concentration. A very similar relationship was reported in [14]. A thin passivation layer can be seen in Fig. 7, the average thickness of which was about 20 nm. Increasing the N2 concentration increased the passivation effect. Overcut sidewalls (roughtly 87°) were observed when Cl2 concentrations were 20% and 40%, indicating sidewall passivation proceeded too fast. In addition, the passivation layer on two closely positioned waveguides appeared thinner, which was due to sputter effect of gas molecules bounced off two facing sidewalls. The effect of varying ICP power was examined at two different chlorine concentrations. Etch rate variation was more notable at 20% Cl2 than at 80% Cl2. At 80% Cl2, the etch rate decreased monotonically, which could be attributed to sputter desorption of reactive Cl(I) from the surface prior to reaction. At 20% Cl2, the increase in etch rate after 600 W could probably be ascribed to increased sputter desorption effect. Figure 8 also shows etch rates are relatively invariant at 600 W ICP power and indicates precise etch depths can be achieved under these conditions. Moreover, we recorded much lower DC bias voltages, which is good for lowering etch-induced damage to the devices. Etch selectivity of photoresist generally remained above 20:1.
Fig. 7 (a) Etch rate as a function of Cl2 concentration at 5 mTorr, 50 W rf, 600 W ICP and 20°C. The total flow rate of (Cl2 + N2) was kept constant at 25 sccm. (b) SEMs of samples etched using 80% Cl2 concentration for 3 mins. The inset shows the 20 nm thick passivation layer. (c) SEMs of samples etched using 60% Cl2 concentration for 4 mins.
Fig. 8 Etch rate and DC bias voltage as a function of ICP power at (a) 20% and (b) 80% Cl2 of total 25 sccm (Cl2 + N2) flow, 5 mTorr, 50 W rf and 20°C.
2.3 Summary of etch recipes
Highly anisotropic etching has been achieved using both slow and fast etch recipes. Slow etch recipe gives etch rates between 0.25 nm/s and 3.8 nm/s while fast etch recipe gives etch rates between 2.0 nm/s and 21 nm/s. These recipes give two orders of magnitude difference in etch rates and facilitate the fabrication of waveguides with different dimensions, located on different vertically separated guiding layers. The slow etch recipe can be used for extremely shallow etching between tens of nanometers and 500 nm. The fast etch recipe can etch a couple of hundred nanometers to a few microns. This broad range is ideal for fabricating MGVI chips in the AlxGa1-xAs material system. One additional recipe with an intermediate etch rate was added. Final results are shown in Table 2. The fast etch recipe has also been used to fabricate suspended, 2D high-contrast AlGaAs waveguides for nonlinear applications in a different wafer structure [16]. It has also been used to fabricate spot-size converters in the same wafer but with the first three epitaxial layers firstly removed [17]. In both cases, similar etch rates were reported, demonstrating the stability and reproducibility of these recipes.
Table 2. Fast, slow and intermediate etch recipes
3. Propagation loss measurement of nanowaveguides
In order to verify the applicability of the recipes developed above, we fabricated high confinement waveguides with varying lengths. The high confinement waveguide has a width of 800 nm and a height of 3.175 μm, the waveguide schematic and TE mode profile of which are shown in Fig. 1(a). Detailed fabrication procedures are as follows. Samples were spin-coated with HSQ and baked at 250 °C for 3 mins. The samples were then exposed using Vistec EBPG 5000 + EBL System and developed in 25% TMAH at 80 °C for 45 seconds. The samples were hard-baked at 250 °C for 60 mins before finally being etched using the fast etch recipe (5 mTorr, 50 W rf, 600W ICP, 20 sccm Cl2/5 sccm N2 and 20 °C) shown in Table 2. A cross-sectional view of the fabricated nanowaveguides can be seen in Fig. 9(a). The sidewall and surface roughnesses are shown in Fig. 9(b). We used a Ti:sapphire laser operating at a wavelength of 850 nm as the light source. A polarizing beam splitter was used to select the polarization of the input light in order to excite the TE modes of the waveguides. We used two Newport M-40x objective lenses for coupling light in and out of the waveguides. The optical power was kept constant at 1 mW before the input lens. The output power from waveguide was measured using Newport 883-SL optical detectors and plotted versus the waveguide lengths. The coupling efficiency between nanowaveguides and objective lenses was about 10% due to mode mismatch. An average propagation loss of 6.7 dB/cm was measured, which represented a worst case scenario result. At such a narrow waveguide width, the mode field intensity at the waveguide/air interface was quite large. We want to point out that, smooth HSQ patterns were developed during the lithography. Sidewall roughness was due to the erosion of the resist during the etch. Propagation loss largely resulted from narrow waveguide dimensions and sidewall roughness. Compared to some propagation loss characteristics found in the literature [18], results in this paper are quite competitive. Though some researchers used resist-reflow methods to improve sidewall roughness [19], whether it’s applicable to MGVI AlxGa1-xAs chips remains to be seen. While the resist reflow methods are generally good for fabricating straight waveguides, it may not be applicable in MGVI AlxGa1-xAs chips where dense features exist and multiple etching steps are needed.
Fig. 9 (a) Scanning electron micrograph of the high confinement waveguides with a width of 800 nm and a height of 3.175 μm. HSQ still remained on the sample. (b) Zoomed-in view of the bottom of the waveguides showing the sidewall and surface roughnesses. (c) Propagation loss of nanowaveguides.
4. Summary
We successfully developed two etch recipes that meet the stringent requirements of fabricating MGVI AlxGa1-xAs chips. The slow etch recipe used BCl3 and achieved an etch rate of 0.25 nm/s while the fast etch recipe used Cl2/N2 gases and achieved an etch rate of more than 20 nm/s. We fabricated simple high aspect ratio AlxGa1-xAs nanowaveguides of 800 nm width using the fast etch recipe and measured a propagation loss of 6.7 dB/cm at the wavelength of 850 nm. We proved that highly precise and anisotropic etching can be achieved. As fabrication of state-of-the-art optoelectronic devices essentially relies on ICP-RIE technieques, our work is an important step forward in demonstrating the feasibility of making MGVI chips in the AlxGa1-xAs material system.
Funding
This work was supported by Applied Science in Photonics and Innovative Research in Engineering (ASPIRE), the Natural Sciences and Engineering Council (NSERC), the Collaborative Research and Training Experience (CREATE) program of Canada. We would also like to thank CMC Microsystems for the support for fabrication costs through the MNT program.
References and links
1. G. Roelkens, A. Abassi, P. Cardile, U. Dave, A. de Groote, Y. de Koninck, S. Dhoore, X. Fu, A. Gassenq, N. Hattasan, Q. Huang, S. Kumari, S. Keyvaninia, B. Kuyken, L. Li, P. Mechet, M. Muneeb, D. Sanchez, H. Shao, T. Spuesens, A. Z. Subramanian, S. Uvin, M. Tassaert, K. van Gasse, J. Verbist, R. Wang, Z. Wang, J. Zhang, J. van Campenhout, X. Yin, J. Bauwelinck, G. Morthier, R. Baets, and D. van Thourhout, “III-V-on-Silicon photonic devices for optical communication and sensing,” Photonics 2(3), 969–1004 (2015). [CrossRef]
2. D. D. Cunningham, In Vivo Glucose Sensing, (Wiley, 2010).
3. K. Dolgaleva, P. Sarrafi, P. Kultavewuti, K. M. Awan, N. Feher, J. S. Aitchison, L. Qian, M. Volatier, R. Arès, and V. Aimez, “Tuneable four-wave mixing in AlGaAs nanowires,” Opt. Express 23(17), 22477–22493 (2015). [CrossRef] [PubMed]
4. P. Kultavewuti, E. Y. Zhu, L. Qian, V. Pusino, M. Sorel, and J. Stewart Aitchison, “Correlated photon pair generation in AlGaAs nanowaveguides via spontaneous four-wave mixing,” Opt. Express 24(4), 3365–3376 (2016). [CrossRef] [PubMed]
5. J. Wang, A. Santamato, P. Jiang, D. Bonneau, E. Engin, J. W. Silverstone, M. Lermer, J. Beetz, M. Kamp, S. Höfling, M. G. Tanner, C. M. Natarajan, R. H. Hadfield, S. N. Dorenbos, V. Zwiller, J. L. O’Brien, and M. G. Thompson, “Gallium arsenide (GaAs) quantum photonic waveguide circuits,” Opt. Commun. 327, 49–55 (2014). [CrossRef]
6. J. H. Marsh, “Quantum well intermixing,” Semicond. Sci. Technol. 8(6), 1136–1155 (1993). [CrossRef]
7. B.-C. Qiu, X.-F. Liu, M.-L. Ke, H.-K. Lee, A. C. Bryce, J. S. Aitchison, J. H. Marsh, and C. B. Button, “Monolithic fabrication of 2 x 2 crosspoint switches in InGaAs/InAlGaAs multiple quantum wells using quantum well intermixing,” IEEE Photonics Technol. Lett. 13(12), 1292–1294 (2001). [CrossRef]
8. P. V. Studenkov, M. R. Gokhale, and S. R. Forrest, “Efficient coupling in integrated twin-waveguide lasers using waveguide tapers,” IEEE Photonics Technol. Lett. 11(9), 1096–1098 (1999). [CrossRef]
9. Y. Logvin, F. Wu, H. Deng, V. Tolstikhin, and C. Brooks, “On-chip wavelength-division (de)multiplexers for multi-guide vertical integration in InP,” IEEE Photonics Conference 2012, Burlingame, CA, 2012, pp. 907–908. [CrossRef]
10. K. A. Atlasov, P. Gallo, A. Rudra, B. Dwir, and E. Kapon, “Effect of sidewall passivation in BCl3/N2 inductively coupled plasma etching of two-dimensional GaAs photonic crystals,” J. Vac. Sci. Technol. B 27(5), L21–L24 (2009). [CrossRef]
11. H. Tamura and H. Kurihara, “GaAs and GaAlAs Reactive Ion Etching in BCl3-Cl2 Mixture,” Jpn. J. Appl. Phys. 23(9), L731–L733 (1984). [CrossRef]
12. C. Constantine, R. J. Shul, C. T. Sullivan, M. B. Snipes, G. B. McClellan, M. Hafich, C. T. Fuller, J. R. Mileham, and S. J. Pearton, “Etching of GaAs/AlGaAs rib waveguide structures using BCl3/Cl2/N2/Ar electron cyclotron resonance,” J. Vac. Sci. Technol. B 13(5), 2025–2030 (1995). [CrossRef]
13. W. Lim, I. Baek, P. Jung, J. Lee, G. Cho, J. Lee, K. Cho, and S. Pearton, “Investigation of GaAs dry etching in a planar inductively coupled BCl3 plasma,” J. Electrochem. Soc. 151(3), G163–G166 (2004). [CrossRef]
14. R. Shul, G. McClellan, R. Briggs, D. Rieger, S. Pearton, C. Abernathy, J. Lee, C. Constantine, and C. Barratt, “High-density plasma etching of compound semiconductors,” J. Vac. Sci. Technol. A 15(3), 633–637 (1997). [CrossRef]
15. T. Brückl and H. Zull, “Actinometry of inductively coupled Cl2/N2 plasmas for dry etching of GaAs,” J. Appl. Phys. 98(2), 023307 (2005). [CrossRef]
16. P. Kultavewuti, Z. Liao, and J. S. Aitchison, “2D high-contrast AlGaAs waveguides for nonlinear applications,” IEEE Photonics Conference 2016, Waikoloa, HI, 2016, pp. 589–590. [CrossRef]
17. Z. Liao, M. Alam, and J. S. Aitchison, “Experimental demonstration of vertically integrated AlGaAs/GaAs waveguides,” IEEE Photonics Conference 2016, Waikoloa, HI, 2016, pp. 625–626. [CrossRef]
18. C. Lacava, V. Pusino, P. Minzioni, M. Sorel, and I. Cristiani, “Nonlinear properties of AlGaAs waveguides in continuous wave operation regime,” Opt. Express 22(5), 5291–5298 (2014). [CrossRef] [PubMed]
19. G. A. Porkolab, P. Apiratikul, B. Wang, S. H. Guo, and C. J. K. Richardson, “Low propagation loss AlGaAs waveguides fabricated with plasma-assisted photoresist reflow,” Opt. Express 22(7), 7733–7743 (2014). [CrossRef] [PubMed]
