1. Introduction

Low optical insertion loss and low cost fabrication are now the key factors in designing optoelectronic devices for high-performance optical fiber links. Optoelectronic devices, such as laser diodes (LD), electroabsorption modulators (EAM), and waveguide devices generally suffer high optical insertion loss owing to the high optical mode mismatch between the waveguide mode and a single Mode fiber (SMF). The concept of spot-size converters (SSC) was thus brought up to overcome such high coupling loss between a SMF and conventional optoelectronic circuit devices [1-8]. In the application of optoelectronic integrated circuits, the SSC-integrated technique has become important because it leads not only to low-loss coupling with a cleaved SMF but also to the larger alignment tolerance with a SMF [8-9]. The cost from the fiber-coupling package issue is then reduced. Tapered waveguides in vertical or lateral directions are usually employed among several SSC-integrated techniques that are classified [9]. Butt-jointing with selective area etching of tapered waveguides and selective area growth can get high transfer efficiency [1-5], but complex processing or material regrowth steps are necessary, giving rise to yield and cost problems. Using waveguide mode coupling, a vertical waveguide directional coupler (VWGDC) with a laterally tapered active waveguide (LTAWG) has been demonstrated and studied [6,7,10], showing that a SSC with low optical-mode transfer loss can be made by one-step material growth. Optical power can be transferred efficiently up or down at the LTAWG region by effective index matching or adiabatically long transfer processing. However, taper structure design, mask design, and waveguide fabrication are still in need of precise control [6,7,10,11], for example, in the submicron photolithography and etching process and material index design. It inevitably leaves a low tolerance in such SSC integration. Therefore, reliable and simple methods for designing and fabricating a SSC are quite important in such research fields.

It has been demonstrated previously that a new processing method, namely the undercutetching-active-region (UEAR) method, can be realized for enhancing high-speed and highefficiency performance of an EAM [12,13]. In this paper, a novel SSC-integrated technique based on a VWGDC structure processed by the UEAR method is proposed and demonstrated. Owing to the selective etching properties between InP and InGaAsP, a tapered structure in the lateral direction can be defined, which is named a laterally tapered undercut-active waveguide (LTUAWG). The whole device processing can be defined by the wet etching method without utilizing any regrowth techniques. A LTUAWG can be processed by a wide tapered ridge waveguide using in situ control, making it a reliable process in which to redeem highly efficient optical mode conversion. Therefore, the complicated processing steps for optimized design and fabrication can be reduced greatly, thus lowering the processing cost. The optical transmission properties of a VWGDC with a LTUAWG are investigated through measuring the electroabsorption (EA) effects of the device, and revealing low-loss mode transfer between large symmetric optical mode and elliptical waveguide mode can be reached by this method. The technique of a LTUAWG shows a high potential for the application of low-cost optoelectronic integration.

 

Fig. 1. Refractive index profile of epitaxial layers. Right side shows schematic cross-section plot of LTUAWG. Waveguide core of width w is formed by undercut etching of active region through wide ridge.

Download Full Size | PDF

 

Fig. 2. Mode effective index for TE polarization against different active region width w of LTUAWG. Right side shows PWG (bottom right) and LTUAWG modes for w=1.8, 2.4, and 3.5 µm.

Download Full Size | PDF

2. Device design and fabrication

InGaAsP/InP-based material is grown on a semi-insulated InP wafer by the metal-organic chemical vapor deposition (MOCVD) system. The refractive index of epitaxial layers is illustrated in Fig. 1, where the inserts are the schematic cross-section plots (right of Fig. 1) of the VWGDC. On the top of the VWGDC, the LTUAWG is defined by laterally undercutetching of the active region of a wide tapered ridge waveguide [12]. The SSC formed by the VWGDC consists of a top LTUAWG and a bottom passive waveguide (PWG). The active region of the LTUAWG contains 10 strain-compensated multiple-quantum-wells (MQW) sandwiched by a p-InP cladding layer (top) and interlaced n-InGaAsP/InP cladding layers (bottom, waveguide separation) forming the active optical waveguide. Since MQWs are used as the active region for the EA effect, the lowest bandgap (highest refractive index) in all of the layers are designed. Underneath the LTUAWG, a PWG is defined by interlaced InGaAsP (0.08µm) and InP (0.12µm) layers. The PWG layers consist of interlaced InGaAsP (0.08µm) and InP (0.12µm) layers. The PWG mode can be designed by adjusting the number of InGaAsP (higher bandgap than the material in quantum well) and InP layers. In the core of the PWG, 6 layers of high-refractive index InGaAsP (λ=1340 nm) are used for confining the optical mode of the PWG, whereas the other layers are set as λ=1300 nm for the cladding layer of the PWG. Three periods of the interlaced InGaAsP/InP layers are used as separation layers (SL) of the coupler between the LTUAWG and the PWG instead of InP. By designing the thickness and composition of InGaAsP, the refractive index of the SLs can be increased to get high coupling efficiency and a larger PWG mode simultaneously [14], so the short coupling length of the VWGDC and a low coupling loss with optical fiber can be expected.

The selective UEAR method is employed to form the LTUAWG [12]. At the end of the LTUAWG, an AWG is connected and used for the EAM. Base on direction coupler theory [15], mode-effective index matching is a key issue to bring up for high coupling efficiency between two waveguides of a coupler. Since the highest refractive index region is in MQWs, the larger optical mode in the PWG (lower refractive index region) is allowed to attain the mode index matching condition by gradually reducing the size of the undercut-wet-etched active region in the LTUAWG. By use of the BPM, Fig. 2 shows the calculated-mode effective indices of the TE-polarized PWG mode and the LTUAWG mode against different active region widths. On the right side of Fig. 2, the calculated LTUAWG modes with UEAR waveguides for w=1.8, 2.4, and 3.5 µm are plotted. As shown, in the LTUAWG mode, an optical waveguide with a narrower active region has a lower mode effective index and also a smaller optical mode, indicating that the mode index matching between the LTUAWG and the PWG can be realized [6,10]. In this design, the region of the mode effective index matching in the LTUAWG is at waveguide widths from w=2 µm to 2.5 µm, where the corresponding coupling length is 200 µm. Once out of this condition, most of optical power will be kept in individual waveguides because of high mode index mismatching. A LTUAWG width of w=4 µm is set at the connection point to the AWG. In order to get high coupling efficiency, the design of the LTUAWG width range and length is set to ensure that the mode-index region can be located in the LTUAWG within our processing tolerance. In processing, the LTUAWG can be defined by wet-etching-based processing from a wide ridge waveguide. Tapered active regions and thus optical mode conversion efficiency can be engineered through undercut wet etching.

Figure 3 plots the schematic structure of the whole device, including the PWG, the VWGDC with the LTUAWG, and the AWG. Both the PWG and the AWG end with optical waveguide facets for two port optical coupling devices, where the 250 µm-long AWG serves as an EAM for testing optical transmission and also for modulation against voltage. The LTUAWG in the VWGDC is set at a length of 350 µm based on the BPM simulation of the VWGDC. In fabricating the LTUAWG, an HBr- and then HCl-based etching solution is first used to define the top p-cladding (1.5 µm thick) of the tapered ridge waveguide, ranging from 6 µm (in the side of the PWG) to 8 µm (connecting with the AWG). An H₂O₂-base solution is then utilized to selectively undercut-etch MQWs (InGaAsP material) from InP material. Through in situ monitoring of the test ridges fabricated on the side of the waveguide, the waveguide core (active region) width can be decided. The same processing steps with LTUAWG steps are also utilized for fabricating the AWG simultaneously, except that the top p-cladding is uniform with a width of 8 µm. After forming the AWG and the LTUAWG, the PWG is then defined by a HBr-based etching solution through aligning the top LTUAWG. Based on the in situ monitoring method, it is found that the reliable- and reproducible-processing steps can be realized by checking the sacrificial test ridges. In an EAM device (AWG), Ti/Pt/Au and Ni/AuGe/Ni/Au are deposited by an e-beam evaporation system as p-and n-type metalizations. The final cleaved facets on the AWG and the PWG are shown in the SEM pictures of Fig. 3, where the corresponding waveguide widths of the AWG and the PWG are 3.8 µm and 8 µm, respectively. PMGI (Polymethylglutarimide) is used for passivating the etching surface, planarization, and metal bridging in the EAM.

 

Fig. 3. Schematic structure of VWGDC based on LTUAWG; SEM pictures of cleaved facets on AWG and PWG.

Download Full Size | PDF

3. Measurement and results

In order to characterize the performance of a VWGDC based on a LTUAWG, an integrated EAM (AWG) serves as the device under test. The optical power of 0 dBm at a wavelength of 1570 nm is the optical source testing the optical transmission of the EAM with bias. Owing to the dual coupling properties of the EAM, a cleaved SMF and a fiber focuser (OZ optics; 3.5 µm FWHM mode size) are employed for coupling optical power from the PWG facet (through the VWGDC) and the AWG facet (EAM side), respectively. In the transmission measurement, two directions of optical power excitation are applied for comparison: (a) the forward direction is defined as from the PWG to the AWG; (b) the reverse direction is as from the AWG to the PWG. During the forward direction excitation, the non-complete mode transfer in the VWGDC will leave the optical power in the PWG region forming the radiation mode under the AWG in the EAM region, deteriorating the extinction ratio of the EAM. Such behavior will be different if the backward direction is generated. In addition, the optical loss and modulation extinction ratio will then be different in both directions. However, no significant difference is observed in this measurement. As shown in Fig. 4(a), the measured optical transmission with bias for both directions exhibits the same insertion loss of -13 dB (at bias=0 V). The analogous transmission properties with bias are obtained, suggesting that the lights for both directions propagate through the same paths and undergo the same loss budget. It can also be verified by collecting the photocurrent through the EAM. 72 µA and 69 µA for forward and backward directions at high bias are observed, revealing that the high mode transfer efficiency is attained in this VWGDC.

The loss budget analysis is also used for further investigation of the LTUAWG-VWGDC. As shown in Fig. 4(a), through the cutback method, the optical coupling loss (including mode-mismatching loss and reflection loss) for the PWG and AWG facets are -3 dB and -4.7 dB, respectively. The extracted propagation loss along the 250 µm-long AWG (EAM) waveguide is -3.7 dB. In addition, the overall device transmission at 0 V is -13 dB, giving a -1.6 dB transfer loss in the VWGDC. As for the justification of the measurement, the 3D BPM mode evolution calculation is used. As shown in Fig. 4(b), the calculated transfer loss is about 70% (-1.55 dB), which is quite consistent with the measured result. In order to characterize the optical modes in the PWG and the AWG, the measurements of alignment tolerance and far-field angle are also utilized to examine the VWGDC. Figure 5 shows the misalignment measurements (dot). The 1-dB misalignment tolerance of ±2.9 µm (horizontal) and ±2.2 µm (vertical) is obtained from the PWG side, which is better than the results of ±1.9 µm (horizontal) and ±1.6 µm (vertical) from the AWG side. By the mode mismatch calculation (solid line of Fig. 5) between the BPM-calculated optical modes and the SMF mode, the simulation curves are in agreement with the measured data (dot) for both the PWG and the AWG. Also, as shown in the inserted plots in Fig. 5, the measured far-field angles for the PWG and the AWG are 6.0° (horizontal) ×9.3° (vertical) and 11° (horizontal) ×20° (vertical), respectively. Based on BPM-calculated modes, the simulated far-field angles of the PWG and the AWG are 6.6° (horizontal) ×11.2° (vertical) and 12.6° (horizontal) ×19.5° (vertical), respectively. Higher misalignment tolerance with lower and more symmetric far-field angles is observed from the PWG, suggesting that a larger and more symmetric optical mode can be obtained from a VWGDC with a LTUAWG. It reveals the capability of high-efficiency mode transfer in such a LTUAWG-VWGDC.

 

Fig. 4. (a) Optical output power against reverse bias of forward (PWG to AWG) and backward (AWG to PWG) transmission through VWGDC. Insert shows semi-log scale and results from cutback method. (b) BPM calculated transfer efficiency of VWGDC.

Download Full Size | PDF

 

Fig. 5. Alignment tolerance and far-field angle measurement (inserts) for PWG (left) and AWG (right). Solid dot curve is for horizontal direction and open dot curve is for vertical direction. All measurement is consistent with calculation curves. Corresponding alignment tolerance of 1 dB drop for PWG and AWG are ±2.9µm (horizontal), ±2.2µm (vertical) and ±1.9µm (horizontal), ±1.6µm (vertical), respectively. Measured far-field angles for PWG and AWG are 6.0° (horizontal) ×9.3° (vertical) and 11° (horizontal) ×20° (vertical), respectively.

Download Full Size | PDF

4. Conclusion

We have first demonstrated a novel processing technique, namely laterally tapered undercut etching (LTUE), for fabricating a SSC. Using the selective UEAR method on a tapered ridge, a LTUAWG can be defined. A large PWG aligned below the LTUAWG is then processed to form a VWGDC structure. An EAM is monolithically integrated with the LTUAWG for testing optical transmission and measuring mode transfer efficiency. Through optical modulation in the EAM, -1.6 dB of mode transfer loss is observed in 1570 nm optical excitation, verified by a 3D BPM calculation. In order to test coupling efficiency in two reverse directions of optical transmission through an EAM-integrated LTUAWG-VWGDC, almost identical transmission relations with bias are found, showing the low mode transfer loss in this coupler. Through optical misalignment testing, 1 dB tolerance of ±2.9 µm (horizontal) and ±2.2 µm (vertical) is observed in the PWG. It is better than ±1.9 µm (horizontal) and ±1.6 µm (vertical) in the AWG. It is also confirmed by the far-field angle measurement from the PWG and the AWG, which have corresponding angles of 6.0° (horizontal) ×9.3° (vertical) and 11° (horizontal) ×20° (vertical), respectively. By in situ controlling of the undercut wet etching process, a tapered structure is realized through a wide tapered ridge, avoiding regrowth steps and complicated processing.