Integrated Optical Mode Field Adapters for Multimode ...

3 downloads 0 Views 869KB Size Report
Integrated Optical Mode Field Adapters for Multimode 40Gbit/s. Optical Ethernet Systems. U. H. P. Fischer (member IEEE)a, Th. Windela , S. Hemrumgrotea.
Integrated Optical Mode Field Adapters for Multimode 40Gbit/s Optical Ethernet Systems U. H. P. Fischer (member IEEE)a, Th. Windela , S. Hemrumgrotea Harz University of Applied Sciences, Friedrichstr. 57, 38855 Wernigerode, Germany E-mail: [email protected]


ABSTRACT In this paper we present the fabrication of optical mode field adaptors at the end of single mode and multimode optical fibers, which act as a micro lens, for fiber optical communications devices, capable up to 40Gbit/s data. The mode field adaptors were used to focus the optical output field (1550nm wavelength) of the fiber to receiver and transmitter OEICs. Based on the measurement of a singlemode fiber in accordance with ITU Recommendation G.652 the optical mode fields are measured in a new set-up, which is demonstrated and discussed in comparison to conventional methods. The work was performed in cooperation with the Heinrich-Hertz-Institute in Berlin. Keywords: spot-size measurement, optical far field, optical communications systems, fiber-chip-coupling

1. INTRODUCTION As the demand for high-speed digital communication such as data, video, and the broadband Internet increases, the 1 required throughput of the modules in communications systems must also increase. Ethernet is well established up to 10Gbit/s in local area networks. Now, the next extension up to 40 Gbit/s is in the definition phase and several development steps must be performed. As in well known SDH-long-haul optical transmission systems, fast transmitter and receiver modules are basic elements of these systems. In the transmission of optical signals, the coupling of the optical output of the laser diodes into the optical fiber, and of the fiber to the photodiode at the end of the transmission 2 line, with high efficiency and low cost is always a challenge for the optical communications industry . In this paper we present the fabrication of an optical mode field adaptor which acts as a micro lens for 40Gbit/s Ethernet fiber optical communications devices using a newly developed type of 50µm core graded index multimode fiber (GIMMF), capable of transmitting up to 40Gbit/s data over 1km. The mode field adaptor was used to focus the optical output field (1550nm wavelength) of the GI-MMF onto a photodiode with a 10µm active region. The work was performed in cooperation with the Heinrich-Hertz-Institute in Berlin, who performed the system tests in their well known 40Gbit/s ring test bed. To characterise the output field distribution, we developed a new method for spot size measurement for multimode optical components which is applicable in mass production. In the classical far field-method where measurements are made along a spherical surface, equidistant from the element under test, the mounting of rotary stages and the long measurement time are great disadvantages. The new method implements a planar measurement surface which overcomes these problems and provides the opportunity to quickly characterize optical mode fields of arrayed fibers, passive waveguides or laser bars. This method will be described in comparison to common far field and near field techniques. The mode field adaptors are fabricated by drawing/melting in a fusing process. A reliable method was identified that could manufacture reproducible spot-size adaptors with radii from 5µm to 90 µm. A study of more than 50 devices with identical drawing parameters was performed. The average fiber end radius of 10.1µm was accomplished with a very small standard deviation of σ=0.16µm. The measured maximum values for the fiber lens radii were 9.8µm and 10.27µm. Additionally BPM (OPTIWAVE) simulations were performed to predict the spot sizes at different fiber end diameters.

2. OEIC-fiber coupling It is necessary in all optical transmission systems to couple the maximum amount of light from a single mode laser diode into a standard monomode or multimode fiber (SMF/MMF). But, it also is important to provide good coupling from the fiber to the photodiode and good coupling from the fiber to/from InP passive waveguides, or to/from GaAs devices. For low-cost coupling, typically a butt-coupled fiber is used and fixed with adhesive in front of the OEIC facet. For Micro-Optics, VCSELs, and Photonic Interconnects II: Fabrication, Packaging, and Integration, edited by H. Thienpont, M. R. Taghizadeh, P. Van Daele, J. Mohr, Proc. of SPIE Vol. 6185 618511, (2006) · 0277-786X/06/$15 · doi: 10.1117/12.660149 Proc. of SPIE Vol. 6185 618511-1

lasers and passive waveguides this technique allows only 10-15% coupling efficiency into SMF and 30-60% into MMF. The use of adhesive also introduces problems with long-term stability of the fixation of the fiber-chip coupling. 34 Saruwatari , has shown the analytical calculation for the fiber to the OEIC coupling problem in detail. SI GI

40° horizontal




Θ=10-30° 60° Θ⊥ vertical Fig. 1: Optical far field of glass fibers and laser diodes (GI graded index, SI step index core)



n 1 (r)


For higher coupling efficiencies one has to adapt the mode fields of the OEIC and the SMF for good overlap. The numerical aperture (NA) of the SMF of 0.15 or MMF of 0.5 is low in comparison to an InP DFB-Laser of 0.3-0.7. A lens or a system of lenses can perform this adaptation of the mode fields. A commonly used system consists of two lenses with a parallel beam part in between where an optical isolator can be inserted to prevent back reflections into the emitting device. Commercial solutions for high-speed networks for more than 10Gb/s transmission rates typically use these two lens systems with an optical isolator with 60dB isolation and are available from different suppliers. The relatively high cost of the components, in combination with a high amount of work time handling many discrete parts, presents a great disadvantage and increases packaging cost. 3. INTEGRATED OPTICAL MODE FIELD TRANSFORMER Reducing the cost of the packaging can be effected by using a micro lens at the end of the SMF - a mode field 5 transformer, or fiber taper. We can fabricate these fiber tapers by melting the fiber in a RF-plasma and drawing it in the longitudinal direction (see Fig. 2). Other approaches used to form such lenses are photolithography, etching, 67 89 micromachining , or dipping into molten glass and fusing , . For the realization of the fiber tapers we used a specially adapted fiber splicing machine from Fujikura. The splicing machine was modified mechanically and in software to get a long drawing length along the z-axis and to control all parameters via automated software (LabView) programs. The ends of two pre-fabricated butt-ended glass fibers are fusion spliced in the first step. In the second step, we add an additional heat pulse by current flow between two electrodes activating a RF-plasma and pull one of the fiber ends at a specific velocity. In this step, the splice is melted and torn off. Based on several specified parameters, a lens like fiber end is formed. Several parameters must be controlled very precisely to form the desired shape and radius of curvature of the fiber lens (fig 3): • Heat injection power • Injection time • Pulling speed • Pulling length All types of glass fibers can be processed (9 µm/50 µm/62,5µm), as well step index (SI) and graded index multimode fibers (GI-MMF). For MMF fibers, the lens function also will be effective, but with larger spot-size compared to the lens radius. The end of the glass fiber is formed conical and works like a mode field transformer with radii from 10 µm to 90 µm. With these different radii one can make an optimum adaptation to the field modes of the OEIC. Coupling efficiencies from 30% to 50% can be obtained with these tapers. At the end of standard single mode fibers we have manufactured reproducible mode field transformers with diameters from 5µm to 90 µm as shown in fig. 5. After melting, the following mechanical and optical parameters of the lens are quantified: • • •

Length of pulled region Radius of fiber tip end Symmetry

Proc. of SPIE Vol. 6185 618511-2

Optical mode field/spot size (depicted in fig. 12.)

___SMF optieches cable Kahel

H DC h sp an flu n Electrodes for ci ektroc n fiber heating m t

/ Drawing direction

/ Lichthogen

:1 L 290,990 tm 60,449 mm

Drawing fiber Zugrase


____ Justage of Y, xyX adjusting her Zuptaser

drawing fiber

Fig. 2: schematic view of the drawing process

Fig. 3: parameters for taper lens analysis

The advantages of these fiber tapers are the small dimensions and the integration onto the end of the SMF which results in a very small opto-electronic package. The major disadvantage is the high sensitivity to small mechanical dislocations: 0.3 µm lateral dislocation results in ca. 2-3dB optical loss. Typical optical lenses realized by this method, which are integrated within the fiber-end face, are shown in fig.4 and exhibit a coupling efficiency of more than 50 % (-3 dB).


50 samples

97,971 mm

36,465 mm

125 tm

Fig. 4: taper with radii of 3µm and 80µm

97O-9SO 9SO-99O 9$J-1OOO pm





lOjOpm 1O2On


Fig. 5: stabile drawing process of 50 samples with 10µm radii with standard deviation σ = 0.16µm

As shown in fig. 5, fabrication with low standard deviation σ = 0.16µm has been achieved. This To adapt the outline of the fiber taper the optical mode field of the laser diode must be characterized very precisely. In the following section, we describe several common methods for mode field acquisition and propose new technique.

Proc. of SPIE Vol. 6185 618511-3

4. MODE FIELD MEASUREMENT 4.1. Far-field-methode 10,11 12

, typically used to characterise the optical mode field of photonic devices. The most There are several methods 13, 14 . Here the optical far field is measured by a scanning photodiode, common method is the so-called far-field method which is rotated around the output of an optical waveguide. The measurement is performed at a relatively long distance of several centimetres from the end face of the waveguide and can be realized without any additional optics and only 15 simple rotating stages. This method also is described in the EIA standard RS-455-47 and in the TIA FOTP-191 test procedure which is used for qualifying optical fibers. Here, the resolution of this method is limited by the scanning resolution. Additionally, in the far field a smearing of the fine contour of the near field can be observed. The set-up for this method, shown in fig. 6, is relatively simple, but requires a large space around the DUT because the photodiode 16 detector needs several centimetres scanning space. Based on the Petermann-fomular, the spot-size (2w0) easily can be determined by the optical far-field angle θ: Detektor


2 w0 = (λ / π )

2 ∫ I (θ ) sin(θ ) cos(θ )dθ −90 90

∫ I (θ ) sin


(1) 3


(θ ) cos(θ )dθ



λ-wavelength, I- Intensity Fig. 6: Optical far-field scanning technique

The field width is obtained at the 1/e2 level of the maximum optical power of the measured light. As a rough approximation, the optical field width, w0, can be written as:

⎡ 1 ⎤ w0 = (λ / π ) ⎢ ⎥ ⎣ n tan(θ ) ⎦


where n is the index of the medium between the detector and the waveguide, wo > wo Because the medium between the detector and the measured waveguide typically will be air, the refractive index n = 1.0 for 1,55µm wavelength and a far-field angle of 10°. The field width wo = 2.8 µm, which is a typical value for an integrated fiber lens of 20 µm radius. 4.2. Near-field-methode The second commonly used method uses a microscope objective to enlarge the surface spot of the waveguide face into a 17 18 camera system. This method is referred to in the literature as the near-field technique , . The resolution of this method is strongly limited by the diffraction limit of the exitant light of the laser diode which is fed through the waveguide under test. At 1550 nm the resolution limit typically is in the region of 1.6 µm in both lateral dimensions. For typical optical fields of laser diodes the resolution of this method is not sufficient to obtain a detailed characterization of the optical field. Additional problems are caused by the lens aberrations, which broaden the image on the camera screen. Also, the DUT must be aligned with micrometer precision in front of the set-up which takes a very long time and limits this method to laboratory use only. This measurement technique is standardised in the DIN32562 recommendation for optical waveguide testing. The spot-size can be directly determined by enlarging the OEIC-field by a microscope lens, and is depicted in fig.7. The field width, w0, is directly measured at 1/e2 of the maximum intensity of the incoming light at the

Proc. of SPIE Vol. 6185 618511-4

camera array, including the enlargement factor of the microscope objective. Both of the above methods suffer from long measurement and positioning times required by the operator, which restricts these methods for high throughput, automated industrial applications.

Cam era or 2-D sc an Wa veguide DUT

Microsc ope lens NA > 0.6 corrected @ 1,55µm Fig. 7: Optical near-field scanning technique

4.3. New Median-field-methode To overcome the disadvantages of the conventional methods, we have developed a new technique for scanning the optical field linearly without additional optics or µm-precise adjustment of the DUT. The measurement of the intensity takes place by means of a photodiode in a 2-axis system, where r is the in-line distance from the test object to the detector (see fig. 8). The distance d can be fixed in the set-up. It does not need to be changed during the measurement.

measure plane

spherical surtace


Fig 8: Median-field measurement set-up.

In relation to the classical far field method a conversion of the measured intensities must be made when the photodiode is moved off the Z-axis since the photodiode no longer aligns directly with the test object. The photodiode would have to be rotated in order to keep the effective surface area constant. Additionally, the distance of the photodiode to the test objects changes. Both factors thus affect the effective surface area of the detector device and must be compensated by a distance coefficient. The coefficient is determined uniquely for each measuring point and retained for later use when measurements are performed again. The surface area of the detector device (usually a photodiode) is assumed here to be rectangular. The corners of the rectangle can be represented by vectors originating at light-emitting face of the test object, see fig. 9. The surface shown in the illustration corresponds to the surface of the photodiode By fixing the length of the vectors to the length of the vector impinging the center of the rectangle, one obtains the effective illuminated surface depicted schematically in fig. 9.

Proc. of SPIE Vol. 6185 618511-5

Movement of the photodiode from the Z-axis reduces this surface area. Also it should be considered that the surface must be distorted, forming an irregular square at many locations in the measurement plane. The surface area is determined according to the formula for the general square:


1 * d1 * d 2 * sin ∠(d1 , d 2 ) 2


where d1, d2 are the diagonals between the corner points of the effective measuring surface. Replacing these, the surface can be calculated by: AEN =

⎯ ⎯→ ⎯ ⎯→ ⎯ ⎯→ ⎛ ⎯⎯→ 1 * AEN C EN * B EN D EN * sin ∠⎜⎜ AEN C EN , B EN D EN 2 ⎝

⎞ ⎟⎟ ⎠


The difference between a flat plane area and the spherical surface is depicted in fig. 10. For the determination of the surface area on the spherical surface the vectors of the corner points of the measurement-receiving device are represented by the radius of the sphere. The radius, therefore, corresponds exactly to the distance of the measurement receiving device from the test object along the Z-axis. It stretches to an irregular square, whose surface area can be determined as: Measurement plane

Measurement plane

Spherical plane



Fig. 9: Reception plane and DUT

AK =

Fig. 10: Difference of plane areas of flat measurement surface and spherical surface

⎯⎯→ ⎯⎯→ ⎛ ⎯⎯→ ⎯⎯→ ⎞ 1 * AK C K * BK DK * sin ∠⎜⎜ AK C K , BK D K ⎟⎟ 2 ⎝ ⎠


The relationship of the two surface areas is used to determine the distance. This factor depends on the coordinates of the measurement-receiving device within the measurement plane

κ (coordinates ) =



Using this factor the intensities measured at the respective coordinates must be calibrated in order to determine the real intensity values. To control the 6-axis-motion system (Physic Instruments (PI) F-604) which moves the photodiode device in the measurement plane, a Labview program was developed. The Labview program also determines the distance coefficient. To calculate the optical spot-size from the assembled median-field-data, the same calculation is used as in the far-fieldmethod, which is described in detail in chapter 4.a). A virtual instrument (VI) was developed which controls the 6-axismotion system and converts the measurement data to far-field-data and estimates the field width w0 in both lateral directions.

Proc. of SPIE Vol. 6185 618511-6

4.3.1. Measurement set-up For verification of the method a SMF manufactured according to ITU-G.652 was measured. This reference measurement was realized with the following measurement set-up, shown in fig. 11:

Fig. 11: measurement set-up a) schematic

b) photograph of set-up

The light source MU951001A, integrated in the Anritsu optical test set MT9810A, was used. This module consists of two Fabry Perot laser diodes with the wavelengths 1310nm and 1550nm which alternatively can be selected. For the detector a photodiode from RCA (InGaAs) with a 50µm active surface was used. The fiber and the receiver were installed on a motion controller system from PI which was controlled by an additional controller unit. However, the controller served only as interface for the PC where the measurement data recording and calculating was performed. For fast measurements, the scanning photodiode was replaced by an infrared camera from Spiricon (1550m). 4.3.2. Results For calibrating the Median-field-method we used a reference-SMF from NIST in which the optical field width, w0, was well-known (2w0=10.4nm @ 1550nm). For these measurements, the infrared camera from Spiricon was used. The 3dimensional field distribution is shown in fig. 12a. A true Gaussian behaviour is observed. As an example of a field distribution for the fabricated fiber tapers, fig. 12b shows the broader median-field width of the lens function which is expected due to the inverse behaviour between far-field angle and field width w0. Taper Field (Setting no.16)

Standard Field

1,00000 0,90000 0,80000 0,70000

1,00000 0,90000 0,80000 0,70000















Fig. 12: measured mode field widths with Median-method: a) Reference fiber of NIST (G652-SMF): w0=5,2µm b) Tapered fiber with 11µm fiber end radius: w0=2,7µm

Later, we measured the field of a butt-ended MMF (62.5/125µm) which is depicted in fig. 13. For MMF a set-up must be realised which guaranties a flat mode distribution by a 2km MMF field test ring and a consecutive mode stripper to select

Proc. of SPIE Vol. 6185 618511-7

only the core modes. For excitation, a laser diode (1540nm) with high NA was used which was coupled directly to a buttended MMF fiber. The special GI-MMF fiber of the HHI was shaped to a tapered end with a 12.65µm radius and a 500µm pulling region length.

Fig. 13: GI-MMF taper of 40Gbit/s fiber In figure 14a) the result of farfield measurement of a butt ended fiber is depicted. From this distribution you can obtain the profile exponent g from the graded index fiber. In good agreement with the producer given value the profile exponent is calculated to be g = 1.8 ± 0.1. Figure 14b) depicts the farfield of a tapered fiber. The field distribution is widely broadened as expected. The Numerical Aperture increases from NA = 0.28 for butt ended to NA = 0.55 for the fiber taper.




5 10

values[a.u] 1

fit -6

4 10


g=1.8 ± 0.1




3 10



2 10



1 10


acceptance angle = 15.9° NA=0.28

acceptance angle = 30.1° NA=0.55




















angle [°]

Fig 14a): Farfield measurement butt fiber with profile exponent g

b) farfield of tapered fiber

5. SUMMARY In this paper we presented the fabrication of an optical mode field adaptor which acts as a micro lens, for 40Gbit/s Ethernet fiber optical communications devices with the newly developed 50µm core graded index multimode fiber (GIMMF), capable of transmitted up to 40Gbit/s data over 1km. The mode field adaptors were used to focus the optical output field (1550nm wavelength) of the GI-MMF onto a photodiode with a 10µm active region. The work was performed in cooperation with the Heinrich-Hertz-Institute in Berlin, who performed the system tests in their well known 40Gbit/s ring test bed. In the classical far field-method where measurements are made along a spherical surface, equidistant from the element under test, the mounting of rotary stages and the long measurement time are great disadvantages. The new method implements a planar measurement surface which overcomes these problems and provides the opportunity to quickly characterize optical mode fields of arrayed fibers, passive waveguides or laser bars. In combination with a Labview program the planar measurement-method shows great potential for automated high-speed high, precise mode field measurement in accordance with ITU Recommendation G.652.

Proc. of SPIE Vol. 6185 618511-8

6. ACKNOWLEDGEMENT The Department of Research and Development of the Federal Republic of Germany and the Ministry of Sachsen-Anhalt supported this work. We thank Ronald Freund and Christoph Caspar at the Heinrich Hertz-Institute in Berlin for the GIMMF fiber. 1 T. Naito:” One Terabit /s Transmission over 10.000km using C-Band and L-Band”, Networks and Optical Communications I, IOS Press, pp 2-9 (2000) 2 U. H. P. Fischer (Member IEEE), S. Zech, K. Peters :“Transmitter modules with reusable fiber-chip coupling method for optical communications systems”, TechOnline ( (2002) 3 M. Saruwatari, K. Nawata:“Semiconductor laser to single-mode coupler“, Appl. Optics, Vol 18, No 11, (1979) 4 M. Saruwatari, T. Sugie:“Efficient laser diode to to single-mode fiber coupling using a combination of two lenses in confocal condition“, IEEE Jour. Quant. Opt., Vol 18,pp 1847-1856, (1981) 5 U. H. P. Fischer, O. Krips, E. Müller, A. Jacob:“Laser micro welding for fiber-chip coupling modules with tapered SMF-fiber ends for optical communication systems”,Optical Engineering, Vol 41, No 12, pp 3221-3229 (2002) 6 L. Landany et al. :“Wedge coupling of lasers into multoimode fibers“, Appl, Opt., Vol 22, No 7, pp 960-961 (1983) 7 M. Presby et al.:“Efficient Coupling of Polarisation-Maintaining Fiber to Laser Diodes“, IEEE Phot. Techn. Lett. , Vol 4, No 8, pp 897-899 (1992) 8 Mathyssek, K. et. al :”Fabrication and Investigation of Drawn Fiber Tapers with Spherical Microlenses”, Journ. of Opt. Comm. , Vol 6, pp. 142-b 146, (1985) 9 U. H. P. Fischer, K. Peters, R. Ziegler, D. Pech, Th. Eckhardt, G. G. Mekonnen, G. Jacumeit: „Packaging of OEIC`s with Tapered Fibers for Optical Communications Systems with up to 45GHz Modulation Bandwidth“, Broadband Access and Technology, Faulkner and Harmer (editors), IOS Press pp 296-300, (1999) 10 Electronic Industries Association Standard RS-455-47, Output Far-Field Radiation Pattern Measurement, American National Standard, (1983) 11 Firester, A.H., Heller, M.E., Sheng, P. : Knife-edge scanning measurements of subwavelength focused light beams, Applied Optics, Vol. 16, No.7, pp 5605 – 5619 (1977) 12 W. Anderson, D. Philen:”Spot Size Measurements for Single-Mode Fibers - A Comparison of Four Techniques”, J. Lightw. Techn. Vol LT-1, NO 1, pp 20-26 (1983) 13 TIA: “Fiberoptic Test Procedures FOTP-191 (1998) 14 Young, M. Mode-field diameter of single-mode optical fiber by far-field scanning,, Applied Optics , Vol. 37, No.24 , pp 5605 – 5619 (1998) 15 Electronic Industries Association Standard RS-455-47, Output Far-Field Radiation Pattern Measurement, American National Standard, 1983 16 Petermann:“Contraints for fundamental-mode spot-size for broadband dispersion-compensated dingle-mode fibers“, Electr. Lett. 19, pp712-714 (1983) 17 . U. H. P.Fischer:“Optoelectronic Packaging“, Berlin/Offenbach :VDE-Verlag GmbH, pp 125, (2002) 18 “DIN32562“

Proc. of SPIE Vol. 6185 618511-9