4.3 Passive devices
4.3.1 Curved waveguides loss
Fig. 4.3 Measurement setup for the characterization of the devices.
The measurement setup of the devices is shown in Fig. 4.3. The measurement is done connecting a tunable laser with a polarization controller at the input waveguide. The transmitted signal is detected in an InGaAs material photo diode and a lock-in amplifier. The tunable laser with a polarization controller is coupled to the input waveguide by using a tapered fiber, which can be adjusted by a five axis stage. The near field of the output waveguide is collected by a lensed fiber and detected by the photo diode. The specimen is placed on a thermoelectric (TE) cooler so that all measurements can be performed at a definite temperature. The passive devices are measured at 25 0C (room temperature) and the active devices are measured at 150C. All measurements are performed by using TE-polarized light. From the ratio of the peak-to-valley value of the Fabry-Perot (FP) oscillations in the output spectra of waveguides without the AR coating, the total losses of the device are determined. [4.4, 4.5]
The measurement method for the propagation losses of the waveguide is so called Fabry-Perot resonance method. [4.6] The chip is regarded as a Fabry-Perot resonator for waveguide losses, where the facets of the chip use as the mirrors of the resonator. The optical wave is reflected back and forth within the chip waveguide depending on the intrinsic losses and the reflectivity factor of the facets calculated by about 0.28. The optical length is changed by varying the temperature of the whole chip or the wavelength exploiting the group velocity dispersion. On this experiment, changing the optical wavelength from 1550 nm to 1551 nm, the optical length is changed. The measured wavelength starts from 1550 nm to1551 nm and the step of 0.01 nm. The more accurate method for waveguide loss measurement is a transmission spectrum of a Fabry-Perot resonator. The transmission intensity can be given by [4.4, 4.7]
Here, I0 is the input light intensity, λ is the optical wavelength, αinsertion is the insertion losses of the waveguide, n is the waveguide effective index, R and T are the intensity reflection and transmission coefficients of the cleaved end facet, respectively, and L is the waveguide length. As the wavelength exploiting the group velocity dispersion is changed, the optical length is changed, resulting Fabry-Perot fringes in the output intensity. The transmitted maximum intensity, Imax, is obtained when the resonance condition is met and the transmitted minimum intensity, Imin, occurs when the anti-resonance condition is met.
1
Assuming zero linewidth, the waveguide propagation loss is given by [4.13]
Imin
Fig. 4.4 The Fabry-Perot resonator is for the straight waveguide with the width of 2.2 μm and the length of 2000 μm
The straight ridge waveguide with the width of 2.2 μm and the length of 2000 μm is measured by the Fabry-Perot resonance method. The measurement result is shown in Fig. 4.4.
The measurement is approximated by a straight line from which the reflectivity factor is directly taken by 0.28. The waveguide losses αpropagation are given by:
1)
Using Fig.4.4 data, the waveguide losses αpropagation are calculated to 30 (dB/cm). The intrinsic losses of the chip using Fabry-Perot method are easily determined than using the cutback method. The cutback method is necessity to cut the waveguide several times, but the Fabry-Perot method is not necessity. When the reflection factor is known, the cutback method becomes less reliable and a more accurate Fabry-Perot method for waveguide loss measurement is used.
Fig. 4.5 The design patterns of the cured waveguide
To determine bend losses, we separately tested S-bends with radius R= 60, 80, 110, 170, 260 μm. The tested patterns of the cured waveguide are shown in Fig. 4.5. The measurement result of bend waveguides with and without deep etching on the outer side is shown in Fig. 4.6. The value for bend waveguides is depicted in Table 4.1. Minimum bending radius of 260 μm can be realized with negligible bending losses. The minimum bending radius for outer deep etched bend waveguides is as low as 110 μm without significant loss.
The experimental results are corresponding with the theoretical values. The designed waveguides with outer deep etching would be used on the ring resonators with MMI turning mirrors for consideration.
Fig. 4.6 Calculated theoretical bending losses of the waveguide with outer deep etching and experimental bending losses with and without outer deep etching.
TABLE 4.1 The measured value for bend waveguides and straight waveguides.
Function Without outer deep etching With outer deep etching Straight
waveguide loss 0.00279 (dB/μm) 0.0059 (dB/μm)
Bend losses dB/μm per 900 curve dB/μm per 900 curve
R= 60 μm 0.037 (dB/900) 0.032 (dB/900)
R= 80 μm 0.026 (dB/900) 0.015 (dB/900)
R= 110 μm 0.016 (dB/900) 0.01 (dB/900)
R= 170 μm 0.008 (dB/900) 0.006 (dB/900)
R= 260 μm 0.011 (dB/900) 0.001 (dB/900)
4.3.2 1x1 MMI waveguides of 4.4 μm-wide and 5 μm-wide pattern
The measured result of the straight 2.2-μm-wide waveguide (in/output waveguide) is shown in Fig. 4.7(a). The straight 2.2-μm-wide waveguide loss is of 4.6 dB/mm. From Fig.
4.30(b), 1x1 4.4-μm-wide and 88-μm-long MMI waveguide has a total loss of 0.33 dB. From Fig. 4.30(c) 1x1 5-μm-wide and 112-μm-long MMI waveguide has a total loss of 1.33 dB.
The higher total loss for the 5-μm-wide MMI waveguide is related to mode-mismatch at boundary between the access waveguide and the MMI waveguide, and the longer length of MMI waveguide.
Fig. 4.7 (a) Fabry-Perot oscillations for straight waveguide of width= 2.2 μm. (b) Fabry-Perot oscillations for MMI waveguide of width = 4.4 μm, and Lmmi= (3/2)L = 88 μm. (c) Fabry-Perot oscillations for MMI waveguide width = 5 μm, and Lmmi= (3/2)L = 112 μm.
4.3.3 90-degree MMI waveguide crossing and turning mirror pattern
The measured result of the 90-degree MMI crossing is shown in Fig. 4.8(b). The 1x1 90-degree MMI waveguide crossing is with MMI waveguide width = 5 μm and length = 112 μm. The total loss of 90-degree MMI waveguide crossing is 3.22 dB. Compared to the data in Fig. 4.7(c), the 90-degree crossing causes a loss of 1.89 dB. This value is considerably larger than the expected, and is attributed to the imperfection in fabricating the deep-etched 90-degree junction. For the 90-degree MMI turning mirror device, the result is shown in Fig.4.8(d). The total loss is 4.97 dB. Therefore, the optical loss from the deep-etched reflector is 1.75 dB. The rather high mirror loss is related to the difficulty for obtaining the precise position of reflector plane, and the smooth and sharp vertical surface during the fabrication process. Modified fabrication techniques are needed to develop, and new designs such as using photonic crystal effect might improve the reflector loss.
Fig. 4.8 (a) SEM picture for the 90-degree MMI crossing pattern (MMI waveguide width = 5 μm). (b) Fabry-Perot oscillations for the 90-degree MMI crossing pattern. (c) SEM picture for the 90-degree MMI turning mirror pattern. (d) Fabry-Perot oscillations for the 90-degree MMI turning mirror pattern.
4.3.4 60-degree MMI waveguide crossing and turning mirror pattern
The measurement of the 60-degree MMI waveguide crossing is shown in Fig. 4.9(b).The 1x1 60-degree MMI waveguide crossing is with MMI waveguides of width = 4.4 μm and length = 88 μm. The total loss of 60-degree MMI waveguide crossing is 0.48 dB. The 60-degree crossing loss is 0.15 dB. For the 60-degree MMI turning mirror device, the result is shown in Fig. 4.9(d). The total loss of single 60-degree MMI turning mirror is 2.54 dB. The 60-degree reflector loss is 2.06 dB.
Fig. 4.9 (a) SEM picture for the 60-degree MMI crossing pattern (MMI waveguide width = 4.4 μm ). (b) Fabry-Perot oscillations for the 60-degree MMI crossing pattern. (c) SEM picture for the 60-degree MMI turning mirror pattern. (d) Fabry-Perot oscillations for the 60-degree MMI turning mirror pattern.
4.3.5 2x2 90-degree MMI waveguide turning mirror pattern
The cross/bar output power ratio measured of 0.15/0.85 at the operating wavelength of 1.55 μm with the 2x2 90-degree MMI turning mirror is shown in Fig. 4.10. The 2x2 90-degree MMI waveguide turning mirror is with a MMI waveguide width of 10 μm, a length of 216 μm and a gap of 2.8 μm.
Fig. 4.10 The cross/bar output power ratio with the 2x2 90-degree MMI turning mirror at the operating various wavelength are measured of almost 0.15/0.85.