Assembly and Experimental Characterization of Fiber Collimators for Low Loss Coupling

Transcription

1 Assembly and Experimental Characterization of Fiber Collimators for Low Loss Coupling Ruby Raheem Dept. of Physics, Heriot Watt University, Edinburgh, Scotland EH14 4AS, UK ABSTRACT The repeatability of assembled fiber collimator performance was investigated using Corning SMF28 fiber and 4.4mm SLW GRIN lens (0.23 pitch) from NSG with a 1550nm ASE source in a laboratory environment by measuring the beam width and optical coupling loss of collimators assembled in the laboratory. Coupling loss as low as 0.11 db was obtained consistently, which was a considerable improvement over the standard performance spec of 0.15 to 0.2dB on the GRIN lens. Source fluctuation was measured to be 0.003dB. Intensity profile showed reasonable gaussian beam behavior in the two perpendicular directions. The average coupling loss was 0.112dB with a standard deviation of 0.012dB. The average mode field diameter was 351 microns with standard deviation of 4.6 microns. Keywords: Fiber collimators, GRIN lens, low loss coupling 1. INTRODUCTION A variety of new micro-optic lenses have emerged in the market for free space communication devices and it is necessary to couple light from or to these devices efficiently. The experimental study of the coupling efficiency of these lenses can be used to compare their performance and at the same time yield information on their optical transmissivity and beam quality. The focus of this paper is on the experimental technique used to build and optimize the collimator performance. The ability of the micro-lens to efficiently collect and collimate the highly divergent beam from a reference single mode fiber with minimal distortion and loss is used as a measure of its performance capability. Imperfections in the optical material give rise to scattering and absorption of the signal, degrading the beam profile and reducing the beam intensity. In order to make performance tests it is necessary to demonstrate capability to build collimators with consistent and repeatable beam profile and consistently low coupling loss, using micro-lenses with known optical characteristics, in a uniform test environment. The test environment includes the optical fiber, the light source, the assembling station, the housing devices, the collimator aligning station, fiber splices and the power meter for optical loss measurement. Each component in the test environment is selected such that the statistical variation in their physical and optical parameters are significantly less than that of the test micro-lenses, thus not contributing to the lens to lens variation in coupling loss. The micro-lens used as control and the micro-lenses whose optical performance will be compared, have to be otherwise compatible in their physical and optical characteristics, such that these components could be used interchangeably. Single mode beam propagation through the fiber can be approximated by a gaussian beam profile. The refractive index within the selected graded-index (GRIN) lens has cylindrical symmetry with index maximum at the center (axis) and decreasing as the square of the axial separation with a parabolic index profile. A 0.23 pitch GRIN less has air gap between the fiber (source) and lens and the beam waist. Coupling loss is sensitive to the fiber lens separation. The anti-reflective (AR) coating at the end face of the pigtail and GRIN lens minimized loss due to Fresnel reflection. The coupling loss varies as a function of the transmitting and receiving collimator separation and the location of the beam waist. For this study, the beam waist was close to the exit face of the collimator to narrow the gap between the transmitting and receiving collimators, as would be the case with fiber optic connectors. When the separation between two collimators increase, not all the transmitted light reach the receiver and the loss increases with separation due to the gradual divergence of the transmitted beam.

2 2. EQUIPMENT AND COMPONENTS The first step in bench marking and comparing the performance of devices require repeatability and reproducibility of test results and a fundamental understanding of the theory and performance of the devices. The theory of GRIN lens, thin and thick lens optics, single mode fiber and the relationship between the fiber-grin lens separation and beam quality, which explain some of the principles behind this experimental work are found in many optics books [1,2] and papers [3-6]. Industry standards like Corning SMF28 fiber and 4.4mm SLW GRIN lens with 0.23 pitch from NSG with a 1550nm ASE source from Agilent were selected for the study. A single spool of SMF28 fiber was used throughout the study so as to minimize any possible batch to batch variability of the single mode fiber. The source fluctuation and coupling loss were measured using ILX Lightwave meter. The average source fluctuation was about 0.003dB. The beam profile was measured using Beamscan from Photon Inc. where dual slits on a rotating drum scanned the beam to reproduce the intensity profile of the spot to within a few micron accuracy. The mode field diameter of the collimated beam was in the range of 300 to 400microns. Angle polished (8 o ) pigtails purchased from AC Photonics were used to build the collimators. The free end of the fiber was cleaved and spliced to the SMF28 reference fiber using Furukawa fusion splicer. Ferrule for housing the pigtail were 9 ± 0.1mm long and the inner diameter was ± 0.005mm. A custom built base plate with 10 micron precision was used for mounting the ferrule/grin lens assembly and positioning the pigtail. A microscope with ring light and camera was used to align the angle facets of the fiber and GRIN lens within the ferrule, by aligning the back reflected images of the ring light from the two facets. The Beamscan positioned at a predetermined fixed distance from the collimator (the distance was much greater than the dimensions of the collimator) was used to monitor the beam width before permanently fixing the pigtail in the ferrule. A custom-built collimator assembly unit was used to hold two collimators in position to couple light between them. One of the two holders was mounted on a 5-axis system with large z-axis displacement capability between them. 2.1 Minimizing Error and Coupling Loss Micro -lenses come in a variety of shapes and sizes with different NAs and wavefront errors. Choosing the right kind of micro -lens for the right collimator is essential to minimize coupling loss. Misalignments are chiefly due to lateral, angular and displacement errors. The critical source of error in building collimators comes from fiber to lens angular misalignment [3,6]. In this study, the ferrule used for housing the collimator was also used for minimizing misalignment between the fiber and the lens. With ferrule inner diameter specification of /-0.005mm and GRIN lens diameter specification of 1.8+/-0.005mm, the maximum possible lateral displacement error would be 0.018mm and the maximum possible angular error would be about 0.3 o when about 1.5mm of the GRIN lens protruded from the ferrule. The actual alignment error is typically a fraction of these extreme cases. 0.3 o 0.018mm Figure 1: The inner diameter and error of the ferrule has to be such that any tilt error between the pigtail and lens is minimised

3 During assembling, the major source of signal loss was found to be surface contamination. Hence cleaning the GRIN lens surfaces and the pigtail end surface prior to mounting was critical. Acetone and lint-free paper were used to clean the AR coated components even though acetone is not recommended for the purpose. The pigtail fiber was spliced to 9+/- 0.1mm Fibre in ferrule /-0.005mm GRIN 1.8+/ mm Glass sleeve Figure 2(a): Arrangement of the pigtail and GRIN lens in ferrule 1.5mm +/- 0.1mm Figure 2(b): The GRIN lens is fixed first with about a third of it protruding outside the ferrule. The pigtail is positioned in place using the beam profile for a beam waist close to the exit face of the lens. the reference fiber during assembling and testing. Splices with loss in the range of 0.01dB were considered acceptable and splices with higher losses were rejected so that the splice loss contribution could be maintained at the same low level. Cut back method was used to measure splice contribution to the total loss and subtracted to isolate the lens contribution to coupling loss. Each grin lens was cleaned and mounted in the ferrule. Using a droplet of UV curable epoxy, the lens was fixed in position such that about 1.5mm of the lens was outside the ferrule. 3. ASSEMBLING AND TESTING The GRIN lens was inserted in to the ferrule and held in position vertically with about 1.5mm jutting out. UV curable glass bonding epoxy was used sparingly to fix the GRIN lens in place in the ferrule. A droplet anywhere around the rim would disperse evenly around the narrow space between the lens and the ferrule due to surface tension, centering the lens in the ferrule. The UV exposure and curing was done with the ferrule and lens standing on end. Placing the ferrule on the side could result in uneven thickness distribution of the epoxy and create misalignment. In the vertical position, the lens would be centered inside the ferrule, minimizing both lateral and angular misalignment (Figure 2). Angle (8 o ) polished fiber and GRIN lens were aligned using the back reflected images of the ring light attached to the viewing microscope (Figure:3). The GRIN lens and fiber separation within the ferrule were adjusted to provide minimum beam width very close to the collimator, so as to enable low loss coupling of two collimators in close proximity. Collimators were built to obtain a specific beam width at the detector positioned at a fixed distance from the collimator. Prior tests were used to obtain the beam width needed at this particular detector position, such that the beam waist was formed very close to the exit face of the GRIN lens. Collimators built to this specification demonstrated high repeatability in coupling efficiency and beam profile. The output beam width of each collimator at different distances from the beamscan was measured to obtain a profile of the beam width as a function of distance.

4 Source Microscope to view the collimator Microscope used for angle face alignment of the fiber and lens Beamscan Figure 3: The collimator assembly setup. The beam width of the signal from the source is measured using the Beamscan which is positioned at a fixed distance from the ferrule. The horizontal microscope ring light illumination is used to align the two angled faces inside the ferrule. The vertical microscope camera unit is used for viewing the assembly. 3.1 Coupling Loss Measurement To minimize alignment error during coupling loss measurement and to speed the coupling process, a parallel faced mirror was positioned between the two collimators to parallel align the z-axis of the two mounted collimators using the tip/tilt knobs. Once the axis of the two collimators were rendered parallel, the two collimators were centered on the z- axis using the x-y axis controls until the 1550nm signal launched through one collimator was detected when coupled to the second collimator (Figure:4). Fine tuning to maximize the signal output was done using all five axis. This was found to be a fast and efficient technique to avoid spurious off-axis signals. Various initial tests were conducted before building the 10 collimators. The collimator used as reference collimator was exchanged with the test collimator to see if there was any difference between the coupling loss as a result of the change of the reference collimator. The error between the loss data from the two sets were negligible, indicating that the test setup was sufficiently robust and the assembled collimators had reproducible loss performance as a function of separation. Source Detector rc z-axis Separation Figure 4: The collimators were first aligned for parallel z-axis and then using the x-y axis, their z-axis were linked coupling the signal between the two collimators.

5 4. RESULTS AND CONCLUSION Loss measurement was based on the optimized loss when the beam from one collimator was coupled to another similar reference collimator and corrected for associated splice loss. Loss (in db) and beam width (in micro ns) as a function of the collimator separation was plotted for ten assembled collimators to show the repeatability of the collimator performance. Beam Width Beam Profile Width of Profile 10 Collimators Beam Width in microns Beam width in microns NSG-1 NSG-3 NSG-5 NSG-7 NSG-2 NSG-4 NSG-6 NSG NSG-9 NSG Separation in mm Collimator separation in mm Figure 5: Beam width profile as a function of collimator separation for ten assembled collimators Intensity profile showed reasonable gaussian beam behavior in the two perpendicular directions. Data (Figure:5) indicated the collimators had a consistent mode field diameter between 340 to 360um at the exit face of the collimator. Average mode field diameter was 351microns with a standard deviation of 4.6 microns. The loss was minimum when two collimators were separated by a millimeter or less (Figure:6). Minimum measured Loss was 0.09dB and the maximum Loss was 0.13dB with average Loss at 0.112dB and Standard deviation at dB. Parameters linked to alignment errors include the uniformity and tightness of the spec in the inner diameter of the ferrule, the surface qualities of the optical elements and the accuracy in the alignment between the collimators. By cleaning the components and minimizing misalignment during assembly, through the choice of ferrules, vertical positioning of GRIN lens in ferrule during curing - all contributed significantly to improving the fiber lens alignment and the signal through the collimator. Predefining the beam width such that the beam waist was formed close to the collimator exit face and assembling the collimators using the beam width at a fixed distance from the collimator (alternate approach to measuring the sub-millimeter fiber lens separation) also contributed to the improved coupling efficiency. This approach resulted in collimators with repeatable coupling loss that was a considerable improvement over the standard performance spec of 0.15 to 0.2dB on the SLW GRIN lens, at the time.

6 Loss Profile of 10 Collimators Comparison of 10 NSG Collimators built to 460um Beam Width Loss in db Loss in db Separation in mm Collimator separation in mm Figure 6: Loss profile as a function of collimator separation for ten assembled collimators Acknowledgement: D.M.Trotter(retired) CorningInc. (This experimental work was completed in 2001during my employment at CorningInc. and released for my use.) REFERENCES 1. Hecht, Eugene, Optics, 3rd Ed, Addison Wesley, Buck, J.A., Fundamentals of Optical Fibers, John Wiley and Sons, New York, Robert W. Gilsdorf, Joseph C. Palais, Single -mode fiber coupling efficiency with graded-index rod lenses, APPLIED OPTICS, Vol. 33, No June Garland Best And Ömür M. Sezerman, Shedding Light On Hybrid Optics: A Tutorial in Coupling, Optics and Photonics News, February Joseph C. Palais, Fiber Coupling using graded-index rod lenses, Applied Optics, Vol. 19, No.12, June Martin van Buren, Nabeel A. Riza, Foundations for Low-Loss Fiber Gradient-Index Lens Pair Coupling with the Self-Imaging Mechanism, Applied Optics, Volume 42, Issue 3, , January 2003.