1. Introduction

In optical communications and optical interconnects, high coupling efficiency (e.g., between a laser diode (LD) and a single mode fiber (SMF)) is indispensable. In particular, the large-scale extension of optical access networks will require the development of low-cost and small-size optoelectronic components [1–2]. In such networks, a compact and highly efficient coupling or pigtailing technique for coupling light from a LD into a SMF is important to minimize the operating threshold current of the LD. This is because operating the laser diode at high driving current will increase the heat to be dissipated and hence reduce its long-term stability and reliability. The coupling loss is mainly due to the mismatch of the numerical apertures between the LD and SMF. A butt-joint, in which a laser diode chip is directly in contact with the tip of the SMF at the joint, is a conventionally used technique but its coupling efficiency is rather poor. Owing to the large difference between the numerical apertures of the LD and SMF, the butt-joint technique can only provide ~10% coupling efficiency. Thus effective power coupling between a LD and a SMF is a big concern in optical fiber communication systems. Several coupling methods have been proposed for reducing the coupling loss between the LD and the SMF [3–9]. The methods can be divided into two kinds. In one kind, the form of the fiber end face was transformed into a semi-spherical form or conical form, which made it behave like a lens. The coupling efficiency between the LD and the lensed fiber was about -2.5dB ~ -6.4dB, while the lensed fiber had a working distance shorter than 100 µm [3–6]. In other kinds of instances, specially fabricated fiber was employed, which was made up of several parts with different refractive indexes. The difference of refractive indexes among the parts made the special fiber act as a self-focusing lens. By the fiber subsection method, the coupling efficiency between the LD and the special fiber was about -0.84dB ~ -3dB, and the working distance was shorter than 4500 µm [7–9].

In this paper, we propose a microlens coupling scheme based on a low-cost sol-gel process. We employ the reflow technique for the fabrication of an elliptical microlens array based on the inorganic-organic hybrid SiO₂/ZrO₂ sol-gel material. Sol-gel technology with its excellent intrinsic material and optical properties enjoys a distinct advantage over conventional photoresist-based methods in the fabrication of micro-optical elements because of the use of a single developing step without an etching process [10–13]. The novel fabrication process of the microlens array only requires an ordinary binary photomask to pattern the sol-gel material, and moreover it is a true single-step etch-free fabrication process suitable for low-cost and high-yield mass production. The microlens coupling technique can provide high coupling efficiency between the LD and SMF by matching an elliptical laser mode to a circular fiber mode. We compare the coupling efficiency and misalignment tolerances for the LD-to-SMF system with and without a microlens.

2. Microlens array fabrication

The inorganic-organic hybrid SiO₂/ZrO₂ sol-gel material was prepared by two different sols. The silica sol was prepared using 3-(trimethoxysilyl) propyl methacrylate as a precursor, and it was formed by its hydrolysis in isopropanol and acidified water with a volume ratio of 20:10:1, respectively. The zirconia sol was prepared using zirconium n-propoxide (Zr(OC₃H₇)₄) as a precursor, and it was formed by its hydrolysis in propanol, nitric acid and hydrochloric acid with a volume ratio of 5:3:2:1, respectively. The two sols were further mixed in a molar ratio of 4:1 (SiO₂:ZrO₂). The final hybrid mixture was stirred vigorously at room temperature for 40 hours to hydrolyze throughly. A 4% wt Photoinitiator IRGACURE 184 (CIBA) was added into the SiO₂/ZrO₂ sol-gel material to make it negative-tone UV photosensitive.

As we know, a laser emitted from a LD often has an elliptical mode, but only a circle-mode laser is apt to be transmitted in a SMF. In order to transform an elliptical laser mode into a circular fiber mode, we need to design and fabricate an elliptical microlens array. The reflowing technique for the fabrication of refractive microlens array is summarized as follows. Through a membrane filter with many 0.1-µm-diameter holes, a sol-gel film was spun onto a pre-cleaned quartz substrate at 4000 rpm for 40 seconds. The film was then soft baked on a hotplate at 70 °C for 5 minutes to remove the excess solvent in the film as well as to improve the adhesion of the film to the quartz substrate. Pattern transfer was made by contact printing using UV exposure through a binary chromium mask. The mask designed for a 16×16 elliptical microlens array is a 16×16 elliptical-through-hole array, in which each elliptical hole has a long-axis length of 76 µm, a short-axis length of 36 µm and an edge-to-edge separation of 2 µm. The UV exposure process was carried out for 15 minutes on a contact mask aligner (Quintil Corporation Q2001CT) with a peak emission wavelength of 365 nm and an irradiance of 15 mW/cm². After UV exposure, the sample was developed in acetone for 14 seconds to remove the unexposed parts, whereas the exposed parts of the sol-gel film were cross-linked to form an elliptical cylinder array on the quartz substrate. The patterned sample was heated to 290 °C for 300 minutes in a nitrogen-purged furnace, and the sol-gel material was in the form of fluid at this temperature. After cooling down to room temperature, the sol-gel material was re-solidified yielding a desirable elliptical microlens array due to the effect of surface tension.

3. Experimental results of microlens array and analysis

The refractive index (n) of the sol-gel thin film and the quartz substrate were measured as 1.442 and 1.458, respectively, at a wavelength of 1.55 µm by a prism coupler (Metricon Corporation). Figure 1 shows the surface morphology of the fabricated 16×16 elliptical microlens array. The image taken by a scanning electron microscope (Jeol SEM - JSM5300), as shown in Fig. 1(a), shows the dimensions of the elliptical apertures in a 2-D imaging field. To find out more details on the height of the profile, a surface profiler (Veeco Metrology - Detak-III) was used to scan across the microlenses, and the surface profiles along the directions of the long axis x and short axis y of the elliptical microlens are shown in Figs. 1(b) and (c), respectively. It can be seen that the elliptical microlens array has good structural and dimensional uniformity as well as excellent surface smoothness. Due to the large difference between the abscissa scale and the ordinate scale, the microlenses shown in Figs. 1(b) and (c) all appear to be elongated. In fact, the microlenses have relatively gentle and flat profile with a sag (h) of only 0.89 µm and an elliptical aperture (D) with a short-axis length and a long-axis length of 36.0 µm and 76.0 µm, respectively. To quantitatively analyze the uniformity of the microlens elements, we randomly chose a sample of 100 microlens elements and measured their sags using the surface profiler. From these measured results, we can obtain an arithmetic average of the sags (h̅) of 0.90 µm, sample standard deviation of 0.07 µm, and non-uniformity of the microlens elements of 7.78%. The root-mean-square roughness of the surface of the microlenses was measured as 1.4 nm using an atomic force microscope (AFM), characterizing a very smooth surface.

 

Fig. 1. Surface characterization of the 16×16 elliptical microlens array in sol-gel material (a) SEM image, (b) profile along the long-axis of elliptical aperture (unit: µm), (c) profile along the short-axis of elliptical aperture (unit: µm)

Download Full Size | PDF

The parameters of the fabricated elliptical microlens array are shown in Fig. 2. For a planar-convex lens in air, it has the same focal length and F# number in both the image space and object space. Because the elliptical microlens has different apertures in the x direction (D_x) and y direction (D_y), it has different average radii of curvature (R̅_x and R̅_y), different average focal lengths (f̄_x and f̄_y), and different average F# numbers (F̅#_x and F̅#_y) in the x and y directions, respectively. Using simple geometry, the average radius of curvature (R̅_i), average focal lengths (fİ_i), and average F# numbers (F̅#_i) of the microlenses are respectively given by

 

Fig. 2. Characteristic parameters of an elliptical microlens array

Download Full Size | PDF

where i=x, y. Using Eqs. (1)–(3) and the parameter values measured in the experiment (i.e., D_x=76.00 µm, D_y=36.00 µm, h̅=0.90 µm, n=1.442), the following microlens parameters are calculated as: R̅_x=802.67 µm, f̄_x=1816.00 µm, F̅#_x=23.89, R̅_y=180.45 µm, f̄_y=408.26 µm, F̅#_y=11.34.

As there is a relationship between the original thickness (H) of the sol-gel thin film and the parameters of the fabricated microlens array (e.g., h, D and R), the profile of the fabricated microlens array can be precisely determined by designing the value of H in advance. During reflow, the sol-gel block was transformed from an elliptical cylinder into a segmented ellipse, and its net volume would remain unchanged (V=V′) if the existence of volatilization was negligibly small. As shown in Fig. 2, an elliptical microlens is a segment (or crown) of an ellipse which can be described by x ²/a ²+y ²/b ²+z ²/c ²=1, where a, b and c are lengths of the three axes of the ellipse. Using simple geometry, the volume of a sol-gel block before reflow (elliptical cylinder) is given by V=πabH[1- (1- h̅/c)²], and the volume of a sol-gel block after reflow (segmented ellipse) is given by V′=(πabh̅ ²/c)(1-h̅/3c), and due to V=V′, the original thickness of the sol-gel thin film (H) is therefore given by

where h̅≪c because the fabricated elliptical microlens array is very gentle and flat. Using equation (4) and the average sag h̅ calculated from the 100 elements of the microlens array, we obtain H=0.45µm. By a thin-film measuring system, the thickness of the sol-gel thin film for fabricating elliptical microlens array is measured as 0.47 µm. The calculated value tallies with the measured value very much. The error between the calculated value and the measured value originated from the volatilization of the sol-gel material in the course of the melting.

4. Coupling using microlens array

The conventional butt joint method and the proposed microlens coupling scheme for coupling a LD into a SMF are investigated by measuring their coupling efficiency and misalignment tolerances. As the beam divergence angle of LD and the numerical aperture of SMF can affect the coupling efficiency and misalignment tolerances, the same LD and SMF were employed in the experiments to obtain a fair comparison of the two coupling methods. The measured far-field patterns of LD at 1.55 µm were 39.3° and 20.2° for perpendicular and parallel to the junction plane, respectively. The SMF has a core diameter of 8.6 µm, a refractive-index difference of 0.42%, and a numerical aperture calculated as 0.096.

The butt joint method was carried out by directly coupling light from a LD into a SMF. The microlens coupling scheme, in which a microlens element of an independent elliptical microlens array was introduced between a LD and an optical fiber to reduce the coupling loss, is illustrated in Fig. 3. In aligning of the LD-Microlens-SMF system, we employed not a microlens but a microlens element of a microlens array because a microlens had too small size (76 µm × 36 µm) to be seen by the naked, which made the alignment very difficult. The LD, the microlens array and the SMF were put on three high-precision muti-axis positioning stages with five degrees of freedom for micro-movement, translation precision of 0.1 µm in X, Y and Z directions (Z direction is the propagation direction of the laser beam), rotation precision of 5″ in θ_X and θ_Y. The LD, driven by LDC500-EC LD driver, produced a maximum light power of 1.8 mW at a wavelength of 1.55 µm. The light power distribution of a laser beam in the space was measured by connecting a one-meter long SMF to a power meter. An infrared sensor card, which can emit visible light when irradiated by infrared light, was employed to assist the alignment process because the 1.55 µm light used is invisible. First, the LD and the SMF were aligned, and LD chip was directly in contact with the end of the SMF, this is corresponded to the butt joint method. Laser output powers were then measured when the LD drive current were changed from 9.0 mA to 18.0 mA. Second, the LD, the microlens array and the SMF were put on the three translation stages and aligned by adjusting the five knobs of the stages, and this is to the microlens coupling scheme. With the microlens coupling scheme, the laser beam was incident on the back of microlens array (the side of quartz substrate). Laser output powers were also measured when the LD drive current was changed from 9.0 mA to 18.0 mA. The measured results of the laser output power from the SMF pigtail as a function of the LD drive current for the two methods are shown in Fig. 4, whereby the microlens coupling scheme has improved the coupling efficiency by nearly 8 fold. The measured coupling efficiency has improved to -2.41 dB (75.8%) by the microlens coupling scheme, compared with the coupling efficiency of -20.1 dB (9.7%) by the butt joint method. Furthermore, in the alignment and assembly process of the LD, the micolens and SMF, misalignment tolerances are found to affect the coupling efficiency significantly. Figures 5 (a)–(c) show the normalized measured coupling efficiency as a function of the lateral misalignment, axial offset and angular misalignment for the two methods. Compared with the butt joint method, the microlens coupling scheme relaxes greatly on the misalignment tolerances of the lateral misalignment and axial offset, but tightens on the misalignment tolerances of the angular misalignment.

 

Fig. 3. Schematic diagram of the microlens coupling scheme

Download Full Size | PDF

 

Fig. 4. Laser output power from the SMF pigtail as a function of the laser diode drive current of the two coupling methods

Download Full Size | PDF

5. Conclusion

An elliptical refractive microlens array was fabricated using a single-step fabrication process involving the use of low-cost inorganic-organic hybrid SiO₂/ZrO₂ sol-gel glass together with the simple reflow technique. Application of the reflowing technique to sol-gel material is a new and successful method, and brings significant improvement in the fabrication technology of a refractive microlens array because the fabricated refractive microlens array has excellent surface smoothness, and structural and dimensional conformity with the designed parameters. The microlens coupling scheme, an available, practical and cost-effective solution, has been shown to outperform the conventional butt-joint method by providing higher coupling efficiency of light from a LD into a SMF, relaxing the lateral and axial misalignment tolerances, and reducing undesirable back-reflected light into the LD. This method can also be used to improve poor coupling efficiency between a LD and its pigtail fiber.

 

Fig. 5. Normalized measured coupling efficiency as a function of (a) lateral misalignment, (b) axial offset, (c) angular misalignment

Download Full Size | PDF