High-Extraction-Rate Ta2O5-Core/SiO2-Clad Photonic Waveguides on Silicon Fabricated by Photolithography-Assisted Chemo-Mechanical Etching (PLACE)

We demonstrate high-extraction-rate Ta2O5-core/SiO2-clad photonic waveguides on silicon fabricated by the photolithography-assisted chemo-mechanical etching technique. Low-confinement waveguides of larger than 70% coupling efficiency with optical fibers and medium propagation loss around 1 dB/cm are investigated in the experiment. Monolithic microring resonators based on Ta2O5 waveguides have shown the quality factors to be above 105 near 1550 nm. The demonstrated Ta2O5 waveguides and their fabrication method hold great promise in various cost-effective applications, such as optical interconnecting and switching.


Introduction
Tantalum pentoxide (Ta 2 O 5 ) is a well-known dielectric material used in high-quality mirror coatings for optical experiments requiring high laser power [1].The CMOS compatibility of such material also enables a large-scale photonic integration of optical waveguides and resonators, which are further profited by the high refractive index (~2.05at 1550 nm), large bandgap (~4 eV), low optical loss, and broad transmission range (300 nm-8 µm) of Ta 2 O 5 [2,3].In addition, compared with the prevalent silica and silicon nitride (Si 3 N 4 ) photonic integration platform, the nonlinear refractive index of Ta 2 O 5 has been found to be an order of magnitude and three times higher than these two materials, respectively [4][5][6].Such an advantage facilitates various nonlinear applications, such as supercontinuum generation and optical switching, to be realized in reduced interaction lengths and thus a more compact footprint with high scalability [7][8][9][10][11][12][13][14][15][16].
Waveguide fabrication in Ta 2 O 5 has progressed rapidly in the last decade.Due to the low film stress of Ta 2 O 5 compared with Si 3 N 4 , fabricating either low-mode-confinement waveguides with ~100 nm thickness or high-mode-confinement waveguides with 500-800 nm thickness is quite flexible [3,[11][12][13][14][15][16].Both waveguide configurations have been employed to realize propagation losses as low as 3 dB/m, which are determined by the residual interface roughness resulting from the dry etching process [3].In this work, by using the photolithography-assisted chemo-mechanical etching (PLACE) technique [17,18], Ta 2 O 5core/S i O 2 -clad waveguides on a silicon substrate are fabricated with a very low surface roughness of ~1 nm.This kind of Ta 2 O 5 waveguide supports high coupling efficiency with the ultrahigh-NA optical fiber and the retrieved coupling loss between the waveguide, and the fiber is 1.3 dB at 1550 nm by the cut-back method, while the waveguide propagation loss is deduced to be 1 dB/cm.The high waveguide propagation loss is discussed to be induced by the inherent absorption loss in the deposited Ta 2 O 5 thin film and the oxide cladding layer.Microring resonators based on Ta 2 O 5 waveguides are also fabricated by the PLACE technique, showing quality factors above 10 5 at 1550 nm.Further reduction in material losses through optimized film deposition process can enable large-scale applications of the Ta 2 O 5 waveguides fabricated by the high-throughput PLACE technique.

Experimental Details
Figure 1a shows a schematic of the Ta 2 O 5 -core/S i O 2 -clad waveguide.A Ta 2 O 5 thin film is first deposited on the thermally oxidized silicon wafer, and then the Ta 2 O 5 layer is patterned into the waveguide core, with the ensuing oxide over-cladding deposited by the plasma-enhanced chemical vapor deposition process (PECVD).Similar waveguide configurations have been validated in the low-loss Si 3 N 4 platform [19].The simulated fundamental transverse electric (TE) mode profile of the Ta 2 O 5 -core/S i O 2 -clad waveguide is shown in Figure 1b.The simulation is completed with the finite-element method (FEM), using the refractive index distribution in the waveguide region (n clad = 1.444 for SiO 2 and n core = 2.058 for Ta 2 O 5 at λ = 1550 nm).The width and height of the Ta 2 O 5 core are selected to be w = 1.3 µm and h = 96 nm in the simulation, in accordance with the profile of the fabricated waveguides tested later.In addition, the Ta 2 O 5 -core sidewall angle is set to θ = 10 • in accordance with the fabricated sample.It can be seen from Figure 1b that most of the guided optical power is distributed in the SiO 2 -clad, giving the waveguide an effective index of n eff = 1.469.The modal propagation properties of the Ta 2 O 5 waveguide are further simulated at different widths and wavelengths while keeping the core height and sidewall angle constant (h = 96 nm and θ = 10 • ).The results are shown in Figure 1c,d.The singlemode propagation condition (ignoring polarization dependence) is clearly illustrated in Figure 1c, where only the fundamental modes (TE 0 and TM 0 ) are allowed to propagate when the waveguide width is less than 2.2 µm.The waveguide dispersion curves for the TE 0 and TM 0 modes at the waveguide width of 1.3 µm (the star points in Figure 1c) are further shown in Figure 1d, and the corresponding group index dispersion curves are also plotted, giving the group index of n g = 1.568 and n g = 1.448 at 1550 nm for the TE 0 and TM 0 modes, respectively.
The Ta 2 O 5 thin film is deposited by electron-beam evaporation (EBE) method on the thermally oxidized silicon wafer [20].The silicon wafer has a diameter of 4 inches and a thickness of 450 µm, with 10-µm-thick oxide layers at both the top and bottom faces.The EBE process is conducted at a temperature of 280 • C and an oxygen-filled vacuum pressure of 2.5 × 10 −2 Pa.The evaporation rate is 3 A/s with an ion beam flux of 30 mA.The surface profile of the deposited Ta 2 O 5 film is examined by the profilometer, and the result is shown in Figure 2a.An average film thickness of 141 nm with ±1 nm fluctuation is achieved.The Ta 2 O 5 waveguide is fabricated with the process flow shown in Figure 2b.A thin layer of chromium is first deposited on the Ta 2 O 5 film by magnetron-sputtering at room temperature.The deposition rate is 2 A/s in the argon-filled environment of 2.1 × 10 −1 Pa.The thickness of the deposited chromium film is 250 nm.Using femtosecond laser ablation, the chromium thin film is prepared into the desired mask pattern.The employed laser parameters are a central wavelength of 1030 nm, a single pulse width (full width at half maximum, FWHM) of 190 fs, and an average laser power of 0.08 mW, with a repetition rate of 250 kHz.The laser beam is tightly focused onto the sample by a microscope objective (NA = 0.7, 100×, M Plan Apo NIR, Mitutoyo Corporation, Kawasaki, Kanagawa, Japan) at ambient conditions (i.e., room temperature, dry air), and the resolution of the laser ablation is about 200 nm [17].The sample is mounted on an air-bearing motorized stage (Aerotech, Inc., Pittsburgh, PA, USA), with a translation resolution of 100 nm and moving with a speed of 1 mm/s during the femtosecond laser ablation.The Ta2O5 thin film is deposited by electron-beam evaporation (EBE) method on the thermally oxidized silicon wafer [20].The silicon wafer has a diameter of 4 inches and a thickness of 450 μm, with 10-μm-thick oxide layers at both the top and bottom faces.The EBE process is conducted at a temperature of 280 °C and an oxygen-filled vacuum pressure of 2.5 × 10 −2 Pa.The evaporation rate is 3 A/s with an ion beam flux of 30 mA.The surface profile of the deposited Ta2O5 film is examined by the profilometer, and the result is shown in Figure 2a.An average film thickness of 141 nm with ±1 nm fluctuation is achieved.The Ta2O5 waveguide is fabricated with the process flow shown in Figure 2b.A thin layer of chromium is first deposited on the Ta2O5 film by magnetron-sputtering at room temperature.The deposition rate is 2 A/s in the argon-filled environment of 2.1 × 10 −1 Pa.The thickness of the deposited chromium film is 250 nm.Using femtosecond laser ablation, the chromium thin film is prepared into the desired mask pattern.The employed laser parameters are a central wavelength of 1030 nm, a single pulse width (full width at half maximum, FWHM) of 190 fs, and an average laser power of 0.08 mW, with a repetition rate of 250 kHz.The laser beam is tightly focused onto the sample by a microscope objective (NA = 0.7, 100×, M Plan Apo NIR, Mitutoyo Corporation, Kawasaki, Kanagawa, Japan) at ambient conditions (i.e., room temperature, dry air), and the resolution of the laser ablation is about 200 nm [17].The sample is mounted on an air-bearing motorized stage (Aerotech, Inc., Pittsburgh, PA, USA), with a translation resolution of 100 nm and moving with a speed of 1 mm/s during the femtosecond laser ablation.The laser-ablated chromium mask pattern is transferred onto the Ta 2 O 5 layer underneath by chemo-mechanical polishing (CMP).The CMP process is conducted using a wafer polishing machine (NUIPOL802, Kejing, Inc., Hefei, China) with a piece of velvet polishing cloth.An amorphous colloidal silica suspension with a particle diameter of 60 nm is used as the polishing slurry.The polishing time is about 7 min.Then, the residual chromium mask is removed by immersion in a standard Cr-etching solution for 10 min.An additional CMP is conducted to further smooth the edge of the fabricated structure, which can attain a high surface quality comparable with the surface tension limited roughness.The height of the Ta 2 O 5 structure after the second CMP is about 100 nm.Afterwards, an oxide layer of 3.  The laser-ablated chromium mask pattern is transferred onto the Ta2O5 layer underneath by chemo-mechanical polishing (CMP).The CMP process is conducted using a wafer polishing machine (NUIPOL802, Kejing, Inc., Hefei, China) with a piece of velvet polishing cloth.An amorphous colloidal silica suspension with a particle diameter of 60 nm is used as the polishing slurry.The polishing time is about 7 minutes.Then, the residual chromium mask is removed by immersion in a standard Cr-etching solution for 10 minutes.An additional CMP is conducted to further smooth the edge of the fabricated structure, which can attain a high surface quality comparable with the surface tension limited roughness.The height of the Ta2O5 structure after the second CMP is about 100 nm.Afterwards, an oxide layer of 3.5 μm thickness is deposited on top of the fabricated structure by PECVD.The PECVD process is carried out by the Oxford PlasmaPro-100 PECVD (Oxford Instruments, Abingdon, Oxon, UK) equipment at a temperature of 300 °C, a deposition rate of 58 nm/min, and an RF power of 20 W. The environment pressure is 1000 mtorr, with an SiH4 flux of 710 sccm and a N2O flux of 170 sccm.

Results and Discussion
Scanning electron microscopy (SEM) is first used to characterize the fabricated Ta2O5 waveguide.The waveguide cross-section images are shown in Figure 3a,b.The Ta2O5-core profile before the deposition of oxide cladding is shown in Figure 3a, featuring a slant sidewall with small angles (~10°) typical of the PLACE technique.The full profile of the Ta2O5-core/SiO2-clad waveguide is shown in Figure 3b, where the small Ta2O5 core is labeled in the green dashed box for clarity.The optical micrographs of the Ta2O5 waveguide are further shown in Figure 3c,d, and the SEM image of the Ta2O5 waveguide is shown in Figure 3e.The waveguide surface profile is further measured by atomic force microscopy (AFM), with the result shown in Figure 3f.The residual surface roughness is deduced to

Results and Discussion
Scanning electron microscopy (SEM) is first used to characterize the fabricated Ta 2 O 5 waveguide.The waveguide cross-section images are shown in Figure 3a,b.The Ta 2 O 5 -core profile before the deposition of oxide cladding is shown in Figure 3a, featuring a slant sidewall with small angles (~10 • ) typical of the PLACE technique.The full profile of the Ta 2 O 5 -core/S i O 2 -clad waveguide is shown in Figure 3b, where the small Ta 2 O 5 core is labeled in the green dashed box for clarity.The optical micrographs of the Ta 2 O 5 waveguide are further shown in Figure 3c,d, and the SEM image of the Ta 2 O 5 waveguide is shown in Figure 3e.The waveguide surface profile is further measured by atomic force microscopy (AFM), with the result shown in Figure 3f.The residual surface roughness is deduced to be R a = 1.2 nm from the AFM measurement.Such a level of surface roughness will induce very little scattering loss (<0.1 dB/cm) for waveguide propagation.
The output mode profile of the Ta 2 O 5 waveguide is measured by the microscope imaging system equipped with an infrared charge-coupled device (InGaAs camera, HAMA-MATSU, Inc., Hamamatsu City, Shizuoka, Japan), as shown in Figure 4a.The input light is coupled into the waveguide using an ultrahigh-NA fiber (UHNA7) with NA = 0.41 and a mode field diameter (MFD) of 3.2 µm at 1550 nm.An in-line fiber-based polarization controller is used to adjust the input polarization for the TE-mode excitation in the Ta 2 O 5 waveguide.The output light from the waveguide is collected by the microscope objective (NA = 0.2, 10×) and imaged into the infrared camera.The measured mode profile at 1550 nm is depicted in Figure 4b, where the 2D and 3D color views are given in the right side and left side, respectively.For comparison, the output mode profile from the UHNA7 fiber is also measured by the same imaging system and is shown in Figure 4c.Similar mode profiles from the Ta 2 O 5 waveguide and UHNA7 fiber can be noticed, indicating the high coupling efficiency between the two components.
Nanomaterials 2024, 14, 1466 5 of 10 be Ra = 1.2 nm from the AFM measurement.Such a level of surface roughness will induce very little scattering loss (<0.1 dB/cm) for waveguide propagation.The output mode profile of the Ta2O5 waveguide is measured by the microscope imaging system equipped with an infrared charge-coupled device (InGaAs camera, HAMA-MATSU, Inc., Hamamatsu City, Shizuoka, Japan), as shown in Figure 4a.The input light is coupled into the waveguide using an ultrahigh-NA fiber (UHNA7) with NA = 0.41 and a mode field diameter (MFD) of 3.2 μm at 1550 nm.An in-line fiber-based polarization controller is used to adjust the input polarization for the TE-mode excitation in the Ta2O5 waveguide.The output light from the waveguide is collected by the microscope objective (NA = 0.2, 10×) and imaged into the infrared camera.The measured mode profile at 1550 nm is depicted in Figure 4b, where the 2D and 3D color views are given in the right side and left side, respectively.For comparison, the output mode profile from the UHNA7 fiber is also measured by the same imaging system and is shown in Figure 4c.Similar mode profiles from the Ta2O5 waveguide and UHNA7 fiber can be noticed, indicating the high coupling efficiency between the two components.
The mode field overlaps between the Ta2O5 waveguide and the optical fibers of variable NAs are calculated using the overlap analysis where the fiber modes are approximated by vectorial Gaussian beams in the calculation [21].The results are shown in Figure 4d.The largest overlap factor is 75% for the fiber, with an NA = 0.42, which is quite close to the UHNA7 fiber used in the experiment.The coupling losses between the waveguide and the fiber, considering the mode overlap and the index mismatch (Fresnel reflection), are also calculated and shown in Figure 4d.The minimum coupling loss is about 1.3 dB for the UHNA7 fiber, denoted by the dashed elliptical circle in Figure 4d.The mode field overlaps between the Ta 2 O 5 waveguide and the optical fibers of variable NAs are calculated using the overlap analysis where the fiber modes are approximated by vectorial Gaussian beams in the calculation [21].The results are shown in Figure 4d.The largest overlap factor is 75% for the fiber, with an NA = 0.42, which is quite close to the UHNA7 fiber used in the experiment.The coupling losses between the waveguide and the fiber, considering the mode overlap and the index mismatch (Fresnel reflection), are also calculated and shown in Figure 4d.The minimum coupling loss is about 1.3 dB for the UHNA7 fiber, denoted by the dashed elliptical circle in Figure 4d.
The waveguide propagation losses are characterized by the cut-back method, using the same experiment setup shown in Figure 4a.A series of Ta 2 O 5 waveguides with identical profiles are fabricated.To reduce the required footprint, waveguides longer than 2 cm are designed to have curved sections.The bending radii for the waveguide lengths of 2.2 cm, 2.4 cm, 2.6 cm, 2.8 cm, 3.0 cm, 3.2 cm, and 3.4 cm are 1 mm, 2 mm, 3 mm, 4 mm, 5 mm, 6 mm, and 7 mm, respectively.The measured fiber-to-fiber insertion losses of the Ta 2 O 5 waveguides are shown in Figure 4e, and the inset shows the digital picture of the sample under test.A linear fit to the measured data, excluding the second point, gives a coupling loss of 1.3 dB per facet and a propagation loss of 1 dB/cm.The large deviation of the second data point should come from the small bend radius used (1 mm), while the consistence of the unit propagation losses of curved waveguides with straight waveguides implies the bending loss is very little when the bending radius is greater than 2 mm.
The measured relatively high propagation loss of 1 dB/cm for the Ta 2 O 5 -core/SiO 2clad waveguides should mainly come from the inferior quality of the film deposition process since the scattering loss incurred by the interface roughness is greatly suppressed through the PLACE fabrication technique.Ta 2 O 5 film deposition by electron-beam evaporation can reduce the intrinsic stress, though the residual metal contaminants could induce absorption loss.In addition, the OH absorption induced by the oxide cladding deposition through PECVD has long been a problem for photonic devices working around the telecom C-band.The employed low-mode-confinement Ta 2 O 5 waveguide is very susceptible to the absorption loss in the oxide cladding.High temperature annealing can remedy such a problem, though the annealed temperature is unfeasible with the employed Ta 2 O 5 waveguide configuration.Further optimizations on the film deposition processes are required to reduce the attainable propagation loss fabricated by the PLACE technique.The waveguide propagation losses are characterized by the cut-back method, using the same experiment setup shown in Figure 4a.A series of Ta2O5 waveguides with identical profiles are fabricated.To reduce the required footprint, waveguides longer than 2 cm are designed to have curved sections.The bending radii for the waveguide lengths of 2.2 cm, 2.4 cm, 2.6 cm, 2.8 cm, 3.0 cm, 3.2 cm, and 3.4 cm are 1 mm, 2 mm, 3 mm, 4 mm, 5 mm, 6 mm, and 7 mm, respectively.The measured fiber-to-fiber insertion losses of the Ta2O5 waveguides are shown in Figure 4e, and the inset shows the digital picture of the sample under test.A linear fit to the measured data, excluding the second point, gives a coupling loss of 1.3 dB per facet and a propagation loss of 1 dB/cm.The large deviation of the second data point should come from the small bend radius used (1 mm), while the consistence of the unit propagation losses of curved waveguides with straight waveguides implies the bending loss is very little when the bending radius is greater than 2 mm.
The measured relatively high propagation loss of 1 dB/cm for the Ta2O5-core/SiO2clad waveguides should mainly come from the inferior quality of the film deposition process since the scattering loss incurred by the interface roughness is greatly suppressed through the PLACE fabrication technique.Ta2O5 film deposition by electron-beam evaporation can reduce the intrinsic stress, though the residual metal contaminants could induce absorption loss.In addition, the OH absorption induced by the oxide cladding Monolithic microring resonators coupled with bus waveguides are also fabricated, employing the Ta 2 O 5 -core/S i O 2 -clad waveguide and the PLACE technique.Such microresonators have great use in high-power and high thermal-load applications [22].To minimize the bending loss, the diameter of the microring is set to 10 mm.The optical microscope image of the microring is shown in the middle of Figure 5, and the enlarged SEM image for the coupling region between the microring waveguide and the bus waveguide is shown as well.The coupling gap is 3.5 µm.The transmission spectrum of the microring is measured using a tunable C-band external cavity diode laser (Toptica).The input and output lights are coupled through the UHNA-7 fibers into the bus waveguides.The output lights are sent to a photodetector connected with an oscilloscope to record the transmission spectrum during wavelength scanning.To characterize the overall mode structure, a broadband spectrum is first measured by coarse scanning the input wavelength through mechanically changing the external cavity length of the tunable laser.Then, a triangular wave signal, generated by a signal generator, is applied to the high-resolution piezo-electric movement accessory of the laser for fine wavelength tuning to measure the resonance profile of each mode.
is measured using a tunable C-band external cavity diode laser (Toptica).The input and output lights are coupled through the UHNA-7 fibers into the bus waveguides.The output lights are sent to a photodetector connected with an oscilloscope to record the transmission spectrum during wavelength scanning.To characterize the overall mode structure, a broadband spectrum is first measured by coarse scanning the input wavelength through mechanically changing the external cavity length of the tunable laser.Then, a triangular wave signal, generated by a signal generator, is applied to the high-resolution piezo-electric movement accessory of the laser for fine wavelength tuning to measure the resonance profile of each mode.A dense spectrum of optical resonance is measured from 1545 nm to 1555 nm, and the results are shown in Figure 6a.An enlarged spectrum, around 1550 nm, is further shown in Figure 6b, where the free spectral range (FSR) of the microring resonator is found to be 0.05 nm from the spacing between adjacent resonances.The expected FSR can be calculated, using the equation ∆λ = λ πD × n (D = 10 mm is the diameter of the microring resonator, and n = 1.568 is the group index of the waveguide obtained from the simulation), to be 0.048 nm, which is close to the retrieved FSR in the experiment.The transmission profile of a single resonance curve around 1550.12 nm is fitted by the Lorentz function shown in Figure 6c, from which the quality factor is obtained to be Q = 1.8 × 10 5 .A dense spectrum of optical resonance is measured from 1545 nm to 1555 nm, and the results are shown in Figure 6a.An enlarged spectrum, around 1550 nm, is further shown in Figure 6b, where the free spectral range (FSR) of the microring resonator is found to be 0.05 nm from the spacing between adjacent resonances.The expected FSR can be calculated, using the equation ∆λ = λ 2 / πD × n g (D = 10 mm is the diameter of the microring resonator, and n g = 1.568 is the group index of the waveguide obtained from the simulation), to be 0.048 nm, which is close to the retrieved FSR in the experiment.The transmission profile of a single resonance curve around 1550.12 nm is fitted by the Lorentz function shown in Figure 6c, from which the quality factor is obtained to be Q = 1.8 × 10 5 .The microring resonator with different waveguide widths is also fabricated and characterized.From the measured Q-factors, the waveguide propagation losses are extracted by the equation α = 2πn g /(λ × Q) and shown in Figure 6d.The Q-factors range from 0.8 × 10 5 to 1.8 × 10 5 , and the corresponding propagation losses are from 3.2 dB/cm to 1.5 dB/cm.The higher propagation losses retrieved from the microring resonator Q-measurement could come from the unsymmetrical coupling region due to the anisotropic chemical polishing process, which could be improved by further optimization of the coupling region.
terized.From the measured Q-factors, the waveguide propagation losses are extracted by the equation α = 2πn λ × Q ⁄ and shown in Figure 6d.The Q-factors range from 0.8 × 10 5 to 1.8 × 10 5 , and the corresponding propagation losses are from 3.2 dB/cm to 1.5 dB/cm.The higher propagation losses retrieved from the microring resonator Q-measurement could come from the unsymmetrical coupling region due to the anisotropic chemical polishing process, which could be improved by further optimization of the coupling region.

Nanomaterials 2024, 14 , 1466 3 of 10 Figure 1 .
Figure 1.Design and property of Ta2O5 waveguide: (a) a cross-section schematic of the Ta2O5 waveguide, (b) the simulated fundamental TE-mode profile of the Ta2O5 waveguide at 1550 nm, (c) the effective refractive index vs waveguide width for different modes, (d) the effective index and group index vs wavelength for the fundamental modes.

Figure 1 .
Figure 1.Design and property of Ta 2 O 5 waveguide: (a) a cross-section schematic of the Ta 2 O 5 waveguide, (b) the simulated fundamental TE-mode profile of the Ta 2 O 5 waveguide at 1550 nm, (c) the effective refractive index vs waveguide width for different modes, (d) the effective index and group index vs wavelength for the fundamental modes.
5 µm thickness is deposited on top of the fabricated structure by PECVD.The PECVD process is carried out by the Oxford PlasmaPro-100 PECVD (Oxford Instruments, Abingdon, Oxon, UK) equipment at a temperature of 300 • C, a deposition rate of 58 nm/min, and an RF power of 20 W. The environment pressure is 1000 mtorr, with an SiH 4 flux of 710 sccm and a N 2 O flux of 170 sccm.

Figure 2 .
Figure 2. Fabrication processes of the Ta2O5 waveguide: (a) the thickness profile of the deposited Ta2O5 film, (b) the process flow for the Ta2O5 waveguide fabrication.

Figure 2 .
Figure 2. Fabrication processes of the Ta 2 O 5 waveguide: (a) the thickness profile of the deposited Ta 2 O 5 film, (b) the process flow for the Ta 2 O 5 waveguide fabrication.

Figure 3 .
Figure 3. Microscopy characterization of the Ta2O5 waveguide: (a) SEM image of the Ta2O5 waveguide before oxide cladding deposition, (b) SEM image of the Ta2O5 waveguide after oxide cladding deposition, (c,d) top-view optical microscope images of the Ta2O5 waveguide, (e) top-view SEM image of the Ta2O5 waveguide, (f) top-view AFM image of the Ta2O5 waveguide.The region in the green dashed boxes is the Ta2O5 core layer.

Figure 3 .
Figure 3. Microscopy characterization of the Ta 2 O 5 waveguide: (a) SEM image of the Ta 2 O 5 waveguide before oxide cladding deposition, (b) SEM image of the Ta 2 O 5 waveguide after oxide cladding deposition, (c,d) top-view optical microscope images of the Ta 2 O 5 waveguide, (e) top-view SEM image of the Ta 2 O 5 waveguide, (f) top-view AFM image of the Ta 2 O 5 waveguide.The region in the green dashed boxes is the Ta 2 O 5 core layer.

Figure 4 .
Figure 4. Guided mode and propagation loss: (a) the experimental setup for waveguide mode and loss measurement, (b) the output mode profile of the Ta2O5 waveguide, (c) the output mode profile of the UHNA7 fiber, (d) the calculated mode overlap factor and coupling loss between the Ta2O5 waveguide and the UHNA7 fiber, (e) the measured insertion losses of the Ta2O5 waveguide of variable lengths and bending radii.

Figure 4 .
Figure 4. Guided mode and propagation loss: (a) the experimental setup for waveguide mode and loss measurement, (b) the output mode profile of the Ta 2 O 5 waveguide, (c) the output mode profile of the UHNA7 fiber, (d) the calculated mode overlap factor and coupling loss between the Ta 2 O 5 waveguide and the UHNA7 fiber, (e) the measured insertion losses of the Ta 2 O 5 waveguide of variable lengths and bending radii.

Figure 5 .
Figure 5. Microresonator characterization setup.The microscope image of the fabricated microring resonator is shown in the middle.Red lines denote a fiber connection and black lines represent an electrical connection.

Figure 5 .
Figure 5. Microresonator characterization setup.The microscope image of the fabricated microring resonator is shown in the middle.Red lines denote a fiber connection and black lines represent an electrical connection.

Figure 6 .
Figure 6.Microresonator Q-factor characterization.(a) The full transmission spectrum of the microresonator from 1545 nm to 1555 nm.(b) The enlarged part of the transmission spectrum showing the FSR of 0.05 nm.(c) The Lorentz fitting of the resonance profile around 1550.12 nm.(d) The Qfactors and corresponding propagation losses retrieved from microring resonators of different waveguide widths.