Chapter 2 Modal Analysis of Asymmetric 12-fold QPC
3.2. Fabrication of 12-fold QPC Microcavity With Bottom DBR
3.2.2. Process for Direct Wafer Bonding
Surface treatments for all samples are still required before the bonding process. The sample and DBR wafer were cleaned in the acetone solution for 5 minutes, then use D.I.
Water to rinse for 5 minutes. After the cleaning process, two wafers were stick together face to face in the D.I. water without contact to the surrounding air. Then use N2 to drying the stuck sample and checking two wafers were stuck by Van der Waals force. As the Fig. 3.3 shown, the stuck sample was clipped on the wafer bonding fixture. The different thermal expansion coefficient of stainless steel and molybdenum material is the main principle of wafer bonding process. Owing to the different thermal expansion coefficient, high pressure was formed at high temperature.
30
Figure 3.3: The scheme of wafer bonding fixture.
At last, wafer bonding fixture was put into the center of the furnace as shown in Fig. 3.4.
The sample is heated to 600℃ in 90 minutes and holds the temperature for 60 minutes.
Finally, the annealing process was that let the furnace cool down naturally to room temperature. Because of the high temperature and high pressure be occurred in furnace, the interface of two wafers would have a chemical reaction to form the atomic covalent bond.
Furthermore, the function of graphite plate is to prevent the bonding reaction between wafer and stainless steel. After the proper time for annealing, two wafers were bonded together.
Figure 3.4: A illustration of the wafer bonding system.
Molybdenum Screw
Graphite Plate
Stainless Steel Molybdenum Nut
Two wafers for Bonding
Furnace
H2
N2
Gauge
Pump PID controller
etchi
32
CH4/Cl2/H2 mixed gas is used to transfer the patterns into MQWs at 150℃ for 40 seconds in ICP etching mode. Then, the Ar/SiCl4 mixed gas is used to transfer the PC patters into DBR ubstrate at 20℃ for 4 minutes in ICP etching mode. The PC lasers with bottom DBR structure was demonstrated after removing the residual hard mask on the MQWs. In Fig. 3.6, there are overviews of fabrication processes of 2D photonic crystal wafer bonding structure lasers.
InP MQWs PR DBR Si3N4 PMMA InGaAs
Figure 3.6: An overview of fabrication processes of bottom DBR structure.
Channels Define Photoresist Coating
Channels Transfer Remove PR & Clean Wafer Pre-Bonding Surface
Remove Substrate Si3N4 Deposit Wafer Bonding
Patterns Transfer Electron-Beam Define
PMMA Coating
Remove Si3N4
Pattern Transfer Remove PMMA
3.3.
34
encounter a problem during the direct wafer bonding process to combine the double side SiO2
layer. At first, we test two kind of deposit method, one is SAMCO PD-220 plasma enhanced chemical vapor deposition in NDL 10k class clean room and the other is furnace in NDL 10 class clean room. After the direct bonding, only the furnace one combines successfully, the layer deposit by PECVD did not work. In order to analyze this result, we use Atomic Force Microscopy (AFM) to study the surface of both samples as shown in Fig. 3.8. The roughness of SiO2 deposit by PECVD is about 10 times larger than deposit by furnace. The roughness of sample surface will cause the particles to hide inside the surface defects and cause the bubble or other effect to damage the bonding interface.
Figure 3.8: The surface roughness of SiO2 deposited by (a) furnace and (b) PECVE.
The SiO2 surface roughness by furnace one is 0.266 nm and by PECVD is 2.740 nm.
(a) (b)
35
3.4. Fabrication of 12-fold QPC Microcavity With Central-Post
3.4.1. Epitaxial Structure and Dielectric Material Deposition
At first, the epitaxial structure illustrated in Fig. 3.9 is prepared. It consists of four 10 nm 0.85% compressively strained InGaAsP quantum wells (QWs) which are separated by three 20 nm unstrained InGaAsP barrier layers. It has been confirmed that the PL spectrum of the QWs (Fig. 2.1(b)) is centered at 1550 nm with 200 nm span. The QWs is grown on InP substrate by metal organic chemical vapor deposition (MOCVD) and then a 60 nm InP cap layer is deposited on it for protecting the QWs during a series of dry etching processes. In fabrication, at first, a 140 nm thick Si3N4 layer served as a hard mask is deposited by SAMCO PD-220 plasma-enhanced chemical vapor deposition (PECVD) process. The SiH4/NH3/N2
gases mixture are used to deposit dielectric hard mask on 300℃ substrate at 35W plasma power in 100 Pa pressure. The Si3N4 thickness of 140 nm can stand for the whole dry etch process when achieving the depth into our wafer about 800 nm.
Figure 3.9: The scheme of epitaxial structure of InP/InGaAsP MQWs. The thickness of active region is about 220 nm.
57.5 nm InGaAsP barriers
60nm InP Cap
Four 10nm 0.85% compressively strained InGaAsP QWs (E ~1.55μm) λ with three 20nm unstrained InGaAsP Barriers (Eg~1.22μm)
InP Substrate
36
3.4.2. Photonic Crystal Patterns Definition and Dry-Etching Processes
At first, the QPC pattern is designed by using CAD tools. And then we load this file into Nanometer Pattern Generation System (NPGS) system. Then photonic crystal patterns were defined by JEOL JSM-6500F electron-beam lithography (EBL) system. The EBL system is a field-emission scanning electron microscope, which employs a schmasky type fields-emission gun for the electron source and state-of-the-art computer technology for high-resolution image observation. Before EBL, an A5 polymethylmethacrylate (PMMA) resist layer is spin-coated on the wafer after previous dielectric deposition process. The PMMA thickness is 300 nm.
After defining QPC patterns by EBL system, the pattern will be transferred by the following processes. In transferring patterns, Oxford Instruments Plasma Technology Plasma lab 100 inductively coupled plasma / reactive ion etching (ICP/RIE) system is used. At first, the sample is etched by O2 plasma in order to clean the residual PMMA in air holes. And then the Si3N4 hard mask is etched by CHF3/O2 mixed gas in RIE mode dry etching. The Si3N4
etching environment recipes are 150W RF power and 55 mTorr at 20℃ with CHF3 and O2 gas flow rate of 5 sccm and 50 sccm, respectively. The etching rate in CHF3/O2 mixed gas is about 1.5 nm/s in average and the selectivity etching ratio to PMMA is 8. After transferring the pattern into Si3N4 layer, we use O2 plasma to remove the survival PMMA layer. And then the pattern transforming into InP/InGaAsP MQWs layer is achieved by H2/CH4/Cl2 mixed gas in ICP mode dry etching. The MQWs etching environment recipes are 73W RF power, 1000W ICP power, and 4 mTorr at 150℃ with H2、CH4 and Cl2 gas flow pressure of 0.8, 0.4, and 0.3 mTorr, respectively. The etching rate in H2/CH4/Cl2 mixed gas is about 5.5 nm/s in average and the selectivity etching ratio to Si3N4 is 6. After a serious of dry etches process mentioned above, the PC patterns have already been transferred onto the QWs.
37
3.4.3. Construct of Central Post Structure
In order to form the central-post structure, the InP substrate below the MQWs should be removed and leave only a post under the center of our microcavity. This can be achieve by immersion the sample into a mixture solution with HCl : H2O = 3 : 1 at 2.5 for ℃ about 1 minutes, and then the solution goes through the drilled holes to form a small post. This process also removes 60 nm InP cap layer and smoothes the surface and the sidewall of the air holes.
In general photonic crystal patterns, for example, triangular lattice photonic crystal, the CAD design must includes windows to break the InP etching stop plane and achieve membrane structure [29]. However, the etching stop plane can be easily broken due to the lattice structure of 12-fold QPC and the undercut will form. As a result, it is unnecessary to define the windows in our CAD file. Fig. 3.10 shows the SEM picture of our sample after wet etch process. We find that the undercut of outer region of QPC is not formed caused by smaller air-hole in the outer region due to proximity effect during EBL process. Also, in the same sample with larger r/a ratio, the membrane structure has been formed under the same wet-etching time. This also indicates the sensitivity of wet etching time in different r/a ratio pattern.
Figu
39
Figure 3.12: (a)The top view SEM picture of fabricated sample. A 12-fold QPC microcavity with central post. (b) The parallelogram shape of central post. (c) A fabricated 12-fold QPC microcavity membrane with crashed outer region due to the over dosage of EBL is supported by the central post.
Fig. 3.12 (a) shows the top-view SEM picture of fabricated 12-fold QPC microcavity with central post. The parallelogram shape of central post is observed from the SEM picture in Fig. 3.12 (b). This asymmetric shape is caused by the anisotropic etching rate of InP material.In Fig. 3.12 (c), a fabricated 12-fold QPC microcavity membrane with crashed outer region due to the over dosage of EBL is supported by the central post. This structure is very fragile because the absence of outer connections. To solve this problem, we design the other two CAD files and the first one is show in Fig. 3.13. This design will produce contacted bridge to connect membrane and wafer to form a stronger central post structure. Fig. 3.14 shows the top-view SEM picture of fabricated device with contacted bridge between QPC membrane and the wafer. Moreover, we can shift the larger air holes away from the other air hole to make the contacted bridge stronger.
(a) (b)
(c)
Figu
with
42
Figure 3.17: The (a) top-view and (b) side-view SEM picture of 12-fold QPC microcavity with central post. The diameter of top and root of the central post are estimated to be 1.19μm and 0.776μm. And the gap between membrane and substrate is estimate to be 1.5μm.
From the top view SEM picture, the circular shadow can be used to estimate the diameter of central post. From Fig. 3.17 (a), the circular shadow is estimate to be 0.84μm in diameter.
From the direct measurement in side-view SEM picture in Fig. 3.17 (b), the diameter of central post is estimated to be 1.19μm. As a result, there is about 30 % inaccuracy when we estimate post size from the circular shadow.
However, when we reduce the radius of central post, the circular shadow will disappear.
This does not mean the post is no longer under the cavity. The shadow is caused by the different charge distribution in the cavity region and the interface with the central post.
However, when the post size reduces, this charge distribution difference will become difficult to observe, i.e. the shadow is difficult to observe and judge the existence of the central post.
Thus, to confirm the existence of the central post, the side-view or tilted-view SEM pictures are necessary.
(a) (b)
1.19μm
0.776μm 1.5μm
43
As shown in Fig. 3.18 (b), there is no shadow observed in the microcavity region, but the post can be clearly observed from a tilted-view SEM picture. The smaller post size can only be estimated from the tilted-view SEM picture. We can first measure the angle of the post. By the diameter estimation at arbitrary point, we can calculate the diameter at the interface of the central post.
Figure 3.18: The tilted-view SEM picture of 12-fold QPC microcavity with the central post.
(a) The circular shadow can be clearly observed in the microcavity region when the diameter of the central post is large. (b) However, the circular shadow cannot be observed in the microcavity region when the diameter of the central post is small.
3.5 Conclusion
We have fabricated a well design 12-fold QPC microcavity with the central post. After the definition of PC patters and transform by EBL system and ICP/RIE dry etching process, the 12-fold QPC microcavity with central post is formed by HCl wet-etching. The central post size can be controlled by different immersed time. Fig. 3.19 shows an overview of fabrication process to summarize the fabrication process.
44
Figure 3.19: An overview of fabrication processes of 12-fold QPC microcavity with the central post.
Si3N4 Deposit PMMA Coating
Electron-Beam Define Patterns Transfer &
Remove PMMA
Construct of
Central Post Structure
Pattern Transfer &
Remove Si3N4
45
Chapter 4 Measurement Results
4.1. Introduction
In order to measure our epitaxial materials and characterize the fabricated 2D photonic crystal microcavity lasers with central post, a NIR micro-PL system with sub micrometer scale resolution in space and sub nanometer scale resolution in spectrum is used. The simple configuration of the micro-PL system is show in Fig. 4.1.
Figure 4.1: The set-up of our NIR micro-PL system Multi-Port ps
Function Generator
TTL Diode Laser
Optical Spectrum Analyzer
NIR High-Speed Power Detector
CCD
46
In this system, an 845 nm TTL laser is used as the pump source. This TTL laser can be used in pulse operation and continuous-wave (CW) operation by switching a function generator. The pump beam is reflected by a 50/50 beam splitter into a 50x long working distance NIR objective lens with numerical aperture of 0.5. The 48% reflection in angle 45°
of the splitter for 845 nm wavelength is confirmed. And the pump beam is focused to a spot-size with 2 μm in diameter by the objective lens. The light emitted from the microcavity is collected by the same objective lens. We use a collective lens to feed the output signal into a multi-mode fiber, and then the signal is detected by the spectrum analyzer, Ando AQ-6315A, with 0.05 nm resolution. All of the following measurement results were estimated by this micro-PL system.
4.2. Basic Lasing Characteristics
In this section, we will show some basic characteristics of 12-fold QPC microcavity with central post including lasing spectrum, light-in light-out (L-L) curve, mode polarization, and so on. All measurements are done at room temperature by our micro-PL system
4.2.1. Measurement Results from Central Post Structures
The well fabricated devices are optically pumped by 845 nm TTL laser with 0.5 % duty cycle at 0.2MHz repetition rate at room temperature. The influence of the existence of central post will be clearly understood by comparing the lasing characteristics of fabricated microcavities with and without central post with each other. As a result, these two kind of fabricated 12-fold D2 microcavities with the same lattice constant and the slightly different hole-radius are measured in order to make a comparison.
The lasing actions are observed from the 12-fold QPC microcavities with and without
47
central post. At first, the typical near and above threshold lasing spectra of 12-fold QPC microcavity without central post are shown in Fig. 4.2. The lasing wavelength is 1582.6 nm and full-width half-maximum (FWHM) is 0.17 nm. The side-mode suppression-ratio (SMSR) is about 25.5dB. The inset of Fig. 4.2 indicates near-threshold lasing spectrum. The typical lasing spectrua near and above threshold of 12-fold QPC microcavity with central post are shown in Fig. 4.3 (a). The lasing wavelength is 1589.5 nm and its FWHM is 0.2 nm. The SMSR is about 10 dB in this lasing spectrum. Also, in Fig. 4.3 (b), we shows a side-view SEM picture of the 12-fold microcavity with central post. From the SEM picture, one can see that a central post under the microcavity is achieved. The size of the post at the interface between the microcavity and itself is estimated to be 95nm in radius.
From the measured results, the existence of the central post induce the red shift of lasing wavelength about 6.92nm which can be seen by comparing two lasing spectra. The quality factor is reduced by the introducing of the central post. The estimated quality factor of membrane structure is about 9300 and that of cavity with central post is decreased to 7900.
Reduction of the quality factor is because that the vertical confinement might be influenced by the existence of central post which has been investigated theoretically in Chapter 2.4.2.
(See Fig. 2.18)
48
Figure 4.2: The typical lasing spectra near and above threshold of 12-fold QPC microcavity laser. The lasing wavelength is 1582.6 nm and the inset indicates the near-threshold lasing spectrum which the FWHM is estimated to be 0.17 nm.
Figure 4.3: (a) The typical lasing spectra near and above threshold of 12-fold QPC microcavity laser with central post. The lasing wavelength is 1589.5 nm and the inset indicated the near-threshold lasing spectrum which the FWHM is estimated to be 0.2 nm.
1582.6 nm
0.17 nm
1589.5 nm
0.20 nm
(a) (b)
49
Figure 4.4: The L-L curve of12-fold QPC microcavity laser. The thresholds are both estimated to be 0.37 mW.
Furthermore, we compare the thresholds of above two microcavity lasers. From Fig. 4.4, the threshold of membrane and central post structure are both estimated to be 0.37mW. These facts indicate that the existence of the central post is hardly to change the characteristics of WGM as our expectation. In the insert of Fig. 4.4, we can observe the slop efficiency is slight different. There is steeper slop efficiency in membrane structure than that in central post structure. This can be attributed to the additional losses by central post. In addition, the post can be also treated as a heat sinker is another advantage of this structure. It can be observed from Fig. 4.4, when pump power is continuously increased, the L-L curve of membrane structure starts to roll off. However, this phenomenon is not so serious.
50
To further identify the lasing mode, we also measure its polarization as shown in Fig. 4.5.
In our previous work, the polarization ratio of WGM in 12-fold QPC microcavity is about 2, which means the property of no specific resonance direction of WGM. Obviously, Fig. 4.5 shows that the measured polarization ratio is only 1.75 and this agree with the WGM properties both in experiments and simulations in our previous reports. Consequently, the WGM lasing is confirmed.
Figure 4.5: The WGM mode polarization of 12-fold QPC microcavity.
In Fig. 4.6, we compare the thresholds of 12-fold QPC microcavities with different post size.
The threshold of microcavity with post radius of 280 nm and 30 nm are estimated to be 1.25mW and 0.32 mW. It is reasonable that the lasing threshold increases with the increasing of post size due to the increasing optical losses. In Fig. 4.6 (b) (c), we compare the quality factor of two different post radius, 30nm and 280nm. The quality factor decrease with the post size as expected. The quality factor of microcavity with post radius of 30nm and 280nm is estimated to be 7000 and 6000.
51
Figure 4.6: (a) The L-L curves of 12-fold QPC microcavity lasers with different central post size. The near-threshold lasing spectra of 12-fold QPC microcavity lasers with post radius of (b) 30 nm and (c) 280 nm and their FWHM are estimated to be 0.22 nm and 0.26 nm
4.3. Conclusion
In this chapter, the basic characteristics of 12-fold QPC microcavity lasers with central post are obtained and compared with those in membrane structure, including thresholds and quality factors. We also recognize the WGM lasing of 12-fold QPC laser with central post by the measured mode polarization without specific direction. Finally, we compare the quality factor and threshold of different post size to demonstrate the influence of the central post. It is concluded that the existence of small central post slightly affect the characteristics of WGM.
0.22 nm 026 nm
(a)
(c) (b)
52
Chapter 5 Conclusion
In this thesis, the history and basic theory of photonic crystal is introduced in the first chapter. The use of 12-fold QPC to microcavity lasers also being presented. And then, in order to analyze the 12-fold QPC microcavity with bottom DBR and 12-fold QPC microcavity with central post, we calculated the resonance spectra and mode profiles of these structures by 3D FDTD method.
From the calculated result, we find out the quality factor of 12-fold QPC microcavity
From the calculated result, we find out the quality factor of 12-fold QPC microcavity