Basically, RCA cleaning is necessary for a wafer to eliminate organic, inorganic attach-ment, and other contaminants. After RCA cleaning, we used the Chamber-B of the Oxford PECVD system to deposit SiOx …lms for the core and the second cladding lay-ers. The PECVD system can deposit thick …lms with lower stress, and the deposition rate is faster than a horizontal furnace system. Moreover, the PECVD system produces oxide …lms with high conformality and low viscosity at a low deposition temperature.
The designed thickness of the second cladding layer is 1 m, and the thickness of the core layer is 2 m. The deposition parameters for 1- m SiOx are listed in Table 4.1.
Table 4.1: Deposition parameters for 1- m SiOx :
Time (sec) 965 RF Power (50kHz) 40
TEOS (sccm) 50 O2 (sccm) 300 Pressure (mTorr) 500 Temperature ( C) 350
Then we used the HDPCVD system to deposit the amorphous silicon …lm (the …rst cladding layer) with a thickness of 0.120 m. The chemical reaction is
SiH4 ! Si + 2H2. (4.1)
By try and error, the better parameters were found to deposit thin …lms with high refractive index, low stress and good uniformity. The deposition parameters for 0.120
m a-Si are listed in Table 4.2. Here, we repeated three times in the steps 1 to 3.
Table 4.2: Deposition parameters for 0.120- m amorphous silicon.
Steps 1 2 3
Time (sec) 180 1 600 ICP Power (W) 0 45 45 Pressure (mTorr) 200 200 200
Ar (sccm) 50 50 50
H2(sccm) 800 800 800 SiH4 (sccm) 50 50 50 Temperature ( C) 200 200 200
The thicknesses of the …ve points and the average refractive indices of cladding layers
are shown in Figure 4-2 and Table 4.3. The average thickness of the core layer was deposited as 2.07 m. The thicknesses and the indices of each layer were measured by using “N&K analyzer 1500”, it is a popular thin …lm measurement system based on the patented “N&K method” [27]. The method can use to calculate the thickness and the refractive index of thin …lms by utilizing Forouhi and Bloomer dispersion equations based on quantum mechanics.
Then we used the thermal coater system to deposit the aluminum layer with 3000 Å thickness as a hard mask. Since we expected to transfer the image from the photoresist (P.R.) to the underneath …lm without any critical dimension loss, therefore, a perfect etching pro…le is favored.
1
2 3
4
5
Figure 4-2: The locations of measuring …ve points on a 6-inch Si wafer.
Table 4.3: The thicknesses and the refractive indices of the …ve points of cladding layers at = 1.55 m.
Material SiOx a-Si
Thickness / Refractive Index d ( m) n d ( m) n Location 1 1.13 1.455 0.123 3.85 Location 2 1.10 1.457 0.118 3.82 Location 3 1.11 1.454 0.120 3.85 Location 4 1.07 1.455 0.117 3.81 Location 5 1.13 1.457 0.119 3.84
Average 1.11 1.456 0.119 3.83
4.3 Lithography
Figure 4-3 shows the layout pattern of the dual ARROW power splitter which includes three parts: the input region, the coupling region, and the decoupling region, which was designed by R-soft and L-edit software. After completing the deposition processing steps, we used the “Clean Track MK-8” system for the lithography process. If the wafer was bended too much, the machine arm of the exposure system could not grab it.
Therefore, before coating P.R., we used the stress measuring instrument to measure the radius of curvature of the wafer. The lithography procedures are described as follows:
a. Spin Coating and Soft Baking
On the wafer surface, we added the HMDS, (CH3)3SiNHSi(CH3)3, for enhancing the adhesion between the wafer and P.R.. The temperature is 90 C and the baking time is 60 seconds. After cooling the wafer, the negative photoresist was applied. Initially, the speed of spin coating should be slow for the P.R. spread uniformly on the wafer surface.
The spin coating speed was set as 500 rpm for 10 seconds. Then it ramps to the …nal speed of 5000 rpm for 45 seconds. The thickness of negative P.R. is 3800 Å.
b. Hot-plate Unit
The step was used to remove the solvent of P.R. and to enhance the adhesion between the Al and P.R. layers. The temperature of the step is 90 C and the baking time is 90 seconds.
c. Exposure, Development and Hard Baking
The dual ARROW power splitters were patterned by the “Leica E-beam” (Leica weprint 200) with exposing dosage 4.5 C=cm2. We used the “Clean Track MK-8” for exposure and development. After developing, we used hard bake to harden the resin within the P.R. and stop the reaction between the developer and P.R. layers. The tem-perature of the step is 120 C and the baking time is 90 seconds. Finally, we used the in-line SEM to inspect the fabrication results.
Decoupling Region 7850 µm
Coupling Region 2150 µm (Lc/2)
Input Region 4000 µm
Figure 4-3: Layout diagram of the dual ARROW power splitter.
4.4 Etching Process and AEI (after etching inspec-tion)
After completing after-develope-inspection (ADI), we used the metal etcher, “ILD-4100”, for etching aluminum of 3000 Å and SiOx of 20000 Å. The etching recipes are shown in Tables 4.4 and 4.5. The etching time for aluminum of 3000 Å is about 38 seconds and for SiOx of 20000 Å is about 600 seconds.
Table 4.4: The recipe for etching aluminum 3000 Å by using metal etcher, “ILD-4100”.
Steps 1 2
Time (sec) 3 35
RF Power (W) 1000 1000 Bias Power(W) 100 100 Pressure (mTorr) 5 5
Cl2 (sccm) 85 100
N2(sccm) 15 0
Load Value Auto Auto Tune Value 4050 4050
In the recipes, CHF3 was used to produce polymers and CF4 was used to produce the ions of ‡uorine. The ions of ‡uorine were used to etch the designed segment. After
Table 4.5: The recipe for etching SiOx 20000 Å by using metal etcher, “ILD-4100”.
Step 1
Time (sec) 600 RF Power (W) 1000 Bias Power (W) 75 Pressure (mTorr) 4
CF4(sccm) 20 CHF3 (sccm) 20 Load Value Auto Tune Value 3250
etching process, we used the wet-bench system to remove the residual aluminum …lm on the wafer with the chemical solution of H2SO4 : H2O2 =3 : 1 at 130 C for 10 minutes.
Then we used the P-10 surface pro…ler to meausre the depth and the in-line SEM to measure the width. After etching inspection (AEI) images were photographed by the in-line SEM.
Figures 4-4 (a) to (c) show the AEI topview SEM images of the coupling region and the decoupling region of the dual ARROW power splitters, and the average etching depth of the trench is 1.98 m.
Coupling Region
Decoupling Region
(a) (b)
(c)
Figure 4-4: The AEI topview SEM images of (a) the dual ARROW power splitter, (b) the coupling region, and (c) the decoupling region.
Chapter 5
Characterization and Discussion
5.1 The Setup of the Optical Measurement System
After accomplishing all of the fabrication process, we used the lensed …bers coupled with the waveguide input and output in the optical measurement system to perform the experiments for characterizing the performance for our designed devices at the operation wavelength of 1.55 m, and the setup is illustrated in Figures 5-1 and 5-2.
The optical measurment instruments contain:
(1) The 1.55- m diode laser: It can be used to generate the light of of 1.55 m.
(2) Lensed …bers: They are used to provide a convenient way to improve coupling between an optical …ber and a waveguide device.
(3) Optical microscope (OM): It assists the alignment of the …bers with the input and output ports of the waveguide.
(4) Infrared (IR) charge-coupled device (CCD) camera: It can view the light spot at of 1.55 m, and we put a 20X lens in front of IR camera to focus the light spot.
(5) TV monitor: It can display the light spot images from the IR camera to facilitate observation and photography.
(6) Photodetector: It is used to detect and measure the optical power.
(7) Power meter: It shows the power intensity.
First, we aligned the lensed …ber with the input port of the dual ARROW waveguide.
Next, we put the IR camera on the output port of the dual ARROW waveguide, and move the position of the input lensed …ber near the input port of the waveguide. At this time, we tuned the x, y, and z positions of the stage and the IR camera until two clear
Optical Breadboard
Figure 5-1: The optical measurement setup for the alignment of the input lensed …ber with the IR camera.
Figure 5-2: The optical measurement setup with photodetector for the power measure-ment.
light spot images on TV monitor were obtained, as shown in Figure 5-3. The stringent requirements for end-face ‡atness and alignment accuracy at the input and output of the waveguides are necessary. After we aligned the lensed …ber, we removed the IR camera and put the output lensed …ber near to the output port of the devices, as shown in Figure 5-2, and tuned the x, y, and z positions of the input and output ports stages until the power meter read a maximum value. Finally, the output power was detected by the photodetector, and the power intensity were recorded and displayed on the power meter.
The Light Spot
30 µm
Figure 5-3: The IR camera image of the light spot from the output port of the dual ARROW power splitter with a separation width of 30 m.
5.2 Cut-back Method for Propagation Loss of the Dual ARROW Power Splitters
For verifying the performances of the power splitters with a coupling length Lc of 4300 m, we designed …ve di¤erent lengths power splitters, 1950, 2050, 2150 (Lc/2), 2250, and 2350 m, and we launched a power only into the left core or only into the right core to measure the output power. In the experiment, guided waves were excited and we used
the cut-back method to measure the propagation losses of the waveguides. The method is often used for measuring the total attenuation of a waveguide device and has the advantage that a relatively accurate measurement is possible with a simple con…guration, but it is a destructive method and information about mode-order dependence of losses can not be obtained. The propagation losses of the transmittances with di¤erent lengths of the device can be obtained as
=
10 log PP2
1
L2 L1 (dB/cm), (5.1)
where P1, P2 and L1, L2 are the transmittances and the lengths before and after cutting, respectively. Here, we de…ne the P1 and P2 are the sum of Pcore1 and Pcore2 with di¤erent lengths, where Pcore1 is the output power of core 1 by launching a power only into the left core or only into the right core and Pcore2 is the output power of core 2 for launching a power only into the left core or only into the right core. The power in decibel relative to 1 mW is de…ned as
P (dBm) = 10 log Pcore1+ Pcore2
1 mW . (5.2)
We measured …ve samples to analyze the characteristics of the power splitters. The relations between the values of the normalized output power and the propagation lengths of the samples 1, 2, 3, 4, and 5 by launching a power only into the left core and only into the right core are shown in Figures 5-4 to 5-8. Then, the values of propagation losses from the slope of the linear …tting line were obtained. Tables 5.1 to 5.5 show the propagation losses of the samples 1, 2, 3, 4, and 5 by linear …tting in Figures 5-4 to 5-8.
The results are quite consistant.
Table 5.1: The propagation loss measurement results of …ve samples for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 1950 m.
(a) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 1.92 2.89 4.99 2.26 2.42
(b) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 1.96 1.76 2.01 1.18 1.24
Table 5.2: The propagation loss measurement results of …ve samples for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2050 m.
(a) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 2.39 2.94 3.26 2.88 3.48
(b) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 2.94 1.12 1.01 1.43 2.08
Table 5.3: The propagation loss measurement results of …ve samples for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2150 m (Lc/2).
(a) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 3.45 1.95 2.33 1.51 1.88
(b) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 2.98 1.27 1.55 2.08 1.84
Table 5.4: The propagation loss measurement results of …ve samples for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2250 m.
(a) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 1.75 1.96 2.08 2.77 2.73
(b) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 1.78 2.35 1.71 1.56 1.72
Table 5.5: The propagation loss measurement results of …ve samples for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2350 m.
(a) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 1.94 2.19 3.37 2.78 2.77
(b) Sample Number 1 2 3 4 5
Propagation Loss (dB/cm) 2.56 2.33 2.34 2.05 2.58
(a) Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -1.92 -2.89 -4.99 -2.26 -2.42
0.4 0.6 0.8 1.0 1.2 1.4 Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -1.96 -1.76 -2.01 -1.18 -1.24
Figure 5-4: The propagation loss for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 1950 m.
(a) Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -2.39 -2.94 -3.26 -2.88 -3.48
0.4 0.6 0.8 1.0 1.2 1.4 Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -2.94 -1.12 -1.01 -1.43 -2.08
Figure 5-5: The propagation loss for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2050 m.
(a) Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -3.45 -1.95 -2.33 -1.51 -1.88
0.4 0.6 0.8 1.0 1.2 1.4 Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -2.98 -1.27 -1.55 -2.08 -1.84
Figure 5-6: The propagation loss for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2150 m (Lc/2).
(a) Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -1.75 -1.96 -2.08 -2.77 -2.73
0.4 0.6 0.8 1.0 1.2 1.4 Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -1.78 -2.35 -1.71 -1.56 -1.72
Figure 5-7: The propagation loss for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2250 m.
(a) Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -1.94 -2.19 -3.37 -2.78 -2.77
0.4 0.6 0.8 1.0 1.2 1.4 Linear Fit of Sample1 Linear Fit of Sample2 Linear Fit of Sample3 Linear Fit of Sample4 Linear Fit of Sample5
Pcore1+Pcore2 (dBm)
Propagation Length (cm)
Sample 1 2 3 4 5
Slope -2.56 -2.33 -2.34 -2.05 -2.58
Figure 5-8: The propagation loss for launching a power (a) only into the left core and (b) only into the right core, respectively. The length of the coupling region is 2350 m.
5.3 The Measurement Results of the Dual ARROW Power Splitters
For dual ARROW power splitters, the imbalance is an important parameter. The output power imbalance in dB is de…ned as
imbalance = 10 log Pcore1
Pcore2 , (5.3)
where Pcore1 and Pcore2 are the output powers of core 1 and core 2 of the dual ARROW power splitters, respectively. If the value of imbalance is lower, the performance of the designed power splitter is better. Here, we separately show the imbalances by launching a power only into the left core or only into the right core. Figures 5-9 and 5-10 show the imbalance measurement results of …ve samples for our designed dual ARROW power splitters. Tables 5.6 and 5.7 show the measurement results of …ve samples for output powers of core 1 and core 2, Pcore1 and Pcore2, and the imbalance. In Tables 5.6 and 5.7, the length of the coupling region is 2150 m (Lc/2). The Pcore1 and Pcore2 average imbalance and standard deviation are 0.45 and 0.12 dB by launching a power into the left core of …ve samples, and 0.57 and 0.05 dB for launching a power into the right core.
Table 5.6: The measurement results of …ve samples by launching a power into the left core for output powers of core 1 and core 2, Pcore1 and Pcore2, and the imbalance. The length of the coupling region is 2150 m (Lc/2).
Sample Number 1 2 3 4 5
Pcore1 ( W) 6.12 8.03 6.79 8.14 6.67
Pcore2 ( W) 5.37 7.03 6.23 7.35 6.26
Imbalance (dB) 0.57 0.58 0.37 0.44 0.28
Table 5.7: The measurement results of …ve samples by launching a power into the right core for output powers of core 1 and core 2, Pcore1 and Pcore2, and the imbalance. The length of the coupling region is 2150 m (Lc/2).
Sample Number 1 2 3 4 5
Pcore1 ( W) 9.17 7.98 8.84 7.89 6.68
Pcore2 ( W) 8.02 6.91 7.93 6.88 5.84
Imbalance (dB) 0.58 0.63 0.47 0.59 0.58
1 2 3 4 5
Standard Deviation 0.1152
Imbalance (dB)
Sample Number
Sample Number
Figure 5-9: The imbalance measurement result of …ve samples for launching a power into the left core. The length of the coupling region is 2150 m (Lc/2).
1 2 3 4 5
Standard Deviation 0.0522
Imbalance (dB)
Sample Number
Sample Number
Figure 5-10: The imbalance measurement result of …ve samples for launching a power into the right core. The length of the coupling region is 2150 m (Lc/2).
Figures 5-11 and 5-12 show the relations between the normalized output power and the length of coupling region of our designed dual ARROW power splitters by launching a power into the left core and the right core, respectively. Figures 5-11(a) and 5-12(a) are the simulation results for the length of the coupling region from 150 m to 4150 m. Figures 5-11(b) and 5-12(b) show the simulation and measurement results of our designed dual ARROW power splitters with the length of the coupling region as 1950, 2050, 2150, 2250, and 2350 m, respectively.
0 1000 2000 3000 4000
0
The Length of Coupling Region (µm) Simulation Results of Core1 Simulation Results of Core2
1900 2000 2100 2200 2300 2400
0
The Length of Coupling Region (µm) Simulation Results of Core1 Simulation Results of Core2 Measurement Results of Core1 Measurement Results of Core2
(a) (b)
Figure 5-11: (a) The simulation results with the length of the coupling region from 150 to 4150 m. (b) The simulation and measurement results of our designed dual ARROW power splitters for launching a power into the left core.
0 1000 2000 3000 4000
0
The Length of Coupling Region (µm) Simulation Results of Core1 Simulation Results of Core2
1900 2000 2100 2200 2300 2400
0
The Length of Coupling Region (µm) Simulation Results of Core1 Simulation Results of Core2 Measurement Results of Core1 Measurement Results of Core2
(a) (b)
Figure 5-12: (a) The simulation results with the length of the coupling region from 150 to 4150 m. (b) The simulation and measurement results of our designed dual ARROW power splitters for launching a power into the right core.
5.4 Discussion
In Chapter 3, the parameters of our designed devices shown in Figure are as follows:
the refractive indices of ng=nh=nl=nh=nl are 1.45/3.7/1.45/3.7/1.45, the thicknesses of dg=dh=dl=dh=dl are 2.0/0.12/1.0/0.12/1.0 m, the widths of wg1=wg2=wh1=wl1=wh2=wl2= wh3=wl3 are 7.0/7.0/ 1.24/3.5/0.5/4.24/0.5/19.2 m, and the etching depth dx is 2.0 m. In Chapter 4, the fabrication results of the dual ARROW power splitters are as follows: the average refractive indices of SiOx and amorphous silicon are 1.46 and 3.83, the average thickness of SiOx and amorphous silicon are 1.11 and 0.12 m, the shape of trench is trapezoid, and the average etching depth dx is 1.98 m. In this Chapter, the imbalance measurement results of the average imbalance are 0.45 and 0.57 dB for launching a power only into the left core and only into the right core, respectively.
From Figures 5-11 and 5-12, we can observe the measurement results and the sim-ulation results are di¤erent. These di¤erences between the simsim-ulation results and the measurement results of the dual ARROW power splitters arise from many possible fabri-cation errors. Because of the imperfection of the waveguides, the shape of the waveguide is trapezoid, as shown in Figure 4-4, wg1, wg2, wh1, wl1, wh2, wl2, wh3, and wl3are changed.
Thus the length of the coupling region of our designed power splitter is changed. More-over, when the etching depth dx is changed, the e¤ective indices of quasi even and odd mode are changed, so the length of the coupling region is changed accordingly. If the dual ARROW waveguides are not certainly symmetric, the radiation losses due to mode conversion from the guided modes to radiation modes of the two waveguides are di¤erent, which results in the imbalance of the power splitter. Moreover, if the densities of defects in two waveguides are di¤erent, these also a¤ect the imbalance of the power splitter.
The measurement results of propagation losses of …ve samples with a length of the coupling region of 2150 m (Lc/2) by launching a power only into the left core are 3.45, 1.95, 2.33, 1.51, and 1.88 dB/cm and by launching a power only into the right core are 2.98, 1.27, 1.55, 2.08, and 1.84 dB/cm, respectively. The propagation losses of the dual ARROW power splitters include [29]: (1) Absorption loss: Impurities in the dual ARROW waveguides materials result in interband absorption, and impurity absorption.
(2) Scattering loss [30]-[32]: It is caused by imperfection of waveguide structure during fabrication processes, such as inclined sidewall of the guiding layer and some defects in the guiding layer. Moreover, the surface between layers is rough, which also a¤ects