We demonstrate a 3x1 fiber-based photonic lantern spatial-multiplexer with mode-selectivity greater than 6 dB and transmission loss of less than 0.3 dB. The total insertion loss of the mode-selective multiplexers when coupled to a graded-index few-mode fiber was < 2 dB. These mode multiplexers showed mode-dependent loss below 0.5 dB. To our knowledge these are the lowest insertion and mode-dependent loss devices, which are also fully compatible with conventional few-mode fiber technology and broadband operation.
© 2014 Optical Society of America
The data capacity carried by a single-mode fiber is rapidly approaching its limits . Multiplexing has been used to increase the capacity of a single-mode fiber (SMF), with various approaches including wavelength division multiplexing (WDM) and polarization. Spatial Division Multiplexing (SDM)  is another approach investigated that has recently being advanced by the research community, albeit being proposed many years ago. This approach can further address this capacity crunch by using a MM (few moded) fiber instead of a SMF, and using its spatial modes [3,4] and even high order orbital angular momentum modes in specialty fiber  as an additional degree of freedom to increase the number of data channels. Another SDM approach is the use of multicore fibers [6,7], here the spatial multiplexing is done by using the uncoupled cores of a multicore fiber as independent data channels.
SMUXes take N signals on many SMF and multiplex them across the N modes of a few-mode fiber (FMF) and vice versa. Mode-selectivity in a SMUX is useful to equalize mode-dependent effects such as differential group delay (DGD) and mode dependent loss (MDL) or gain. It is challenging to produce a SMUX that excites each mode of a FMF without loss since the N spatial modes of the fiber are spatially overlapping and cannot be simply separated. However, a SMUX can have relaxed mode-selectivity requirements for FMF systems due to the nature of waveguide optics mode-dependent effects are very similar for different mode groups. For instance, the 4 vector modes comprising the LP11 group are propagated with nearly the same speed and attenuation. Furthermore, they strongly mix with each other and it is not necessary to individually demultiplex each mode in the group. It is sufficient to demultiplex scrambled but orthogonal combinations of modes within the same group. Similar properties hold for the 4 vector modes within the LP21 and the 2 vector modes within the LP02 group. Therefore, a mode-group selective SMUX would enable DGD and MDL compensation within the FMF communication systems. Nowadays, the best option for mode-selectivity has been phase-mask based SMUXes , however these are very cumbersome. Phase mask SMUXes convert Gaussian beams into different spatial modes and then overlap the shaped beams onto the FMF using passive beam combining . However, these systems suffer from large insertion losses (IL) from the passive beam combining, furthermore their IL increases proportionally to the number of modes that the SMUX is designed for.
Photonic lanterns are one of several breakthroughs that arose from the field of astrophotonics . Interestingly, these devices are now being explored as adiabatic mode converters (i.e. as spatial-multiplexers, SMUX) for their use in SDM systems for coherent multiple-input multiple-output (MIMO) networks [4,10]. Launching into the isolated SMF core modes distributes the information across the FMF modes. This approach will excite an orthogonal combination of modes such that all the channels experience similar modal dependencies. Photonic lanterns can be used as both multiplexer and de-multiplexer with a nearly zero MDL and coupling losses, as previously demonstrated theoretically . Furthermore, photonic lanterns can be nearly lossless, can scale to many modes, and are robust because with the right design they can be spliced directly both to the FMF and the SMFs.
In this paper, we show how to add mode-selectivity to a photonic lantern SMUX by fabricating the photonic lantern with dissimilar optical fibers (Fig. 1)  rather than identical ones as in standard photonic lanterns [11,13]. Along the adiabatic taper, the dissimilarity can control the coupling between the SMF cores and force the propagating light initially launched into certain SMF cores to evolve into specific mode groups (e.g. LP01, LP11, LP21, etc). Similar concepts have been shown in photonic crystal fiber devices [14–17], optical fiber null coupler tapers [18,19] and planar waveguides [20–22].
2. Mode-selectivity principle of operation
A standard photonic lantern is built by placing multiple SMFs into a low refractive index capillary tube that is then adiabatically tapered . The whole composite structure is tapered such that the SMF cores nearly vanish, the SMF cladding becomes the new multi-mode core, and the low-index capillary becomes the multi-mode fiber cladding. Proper geometrical core arrangements  can ensure scalability to large number of modes (> 10) without incurring excess losses and minimizing MDL in SMUXes.
Building a photonic lantern which can couple to individual modes while launching light into the independent SMFs requires: 1) breaking the degeneracy between the modes through the entire photonic lantern transition, and 2) ensuring the propagation constants do not cross or interact with each other during the transition to avoid mode coupling. By making the starting SMF cores dissimilar (i.e. choosing dissimilar fibers to fabricate the photonic lantern) will lead to different propagation constants of the initial guided modes. This breaks up the initial mode degeneracy of the uncoupled core modes, providing a range of non-degenerate propagating modes at the beginning of the transition. The propagation constants order and difference of the initial guided modes at the SMF uncoupled cores will determine the final modal multiplexing properties of the photonic lantern. An adiabatic taper transition from the SMFs to the multimode end will ensure that light in the input core of the n-th greatest propagation constant will excite only the output mode of the n-th greatest propagation constant at the multimode fiber end, and vice versa [15,17]. This should remain the case across all wavelengths for which this photonic lantern transition is adiabatic. Such a mode multiplexer does therefore function over a broad band .
In order to understand how a set of identical modes evolve into an equal number of non-degenerate orthogonal modes, the entire taper transition of the photonic lantern must be modeled. The most effective approach is to model the transition at discrete points along its length, such that at each point the 2D waveguide geometry is considered . Figure 2(A) shows the modal analysis of a photonic lantern with three identical SMFs. The parameters used for this are: index difference of the SMF claddings to the low refractive index jacket (i.e. numerical aperture (NA) of the final multimode waveguide) NA = 0.1; SM waveguides are 3 × 12 µm cores diameter with a core NA = 0.12. The mode effective indexes are plotted against the inner diameter of the capillary along the transition starting at 80 µm inner diameter, which it can be assumed to approximate the multimode fiber core diameter. We assume that during the taper the refractive index profile scales proportionally to the capillary inner diameter and that the fibers are fused together early on the transition. At large inner diameters the modes corresponding to each core are well confined and hence degenerate (i.e., the 3 overlapping blue curves). As the inner diameter decreases, the cores begin to couple and the degeneracy breaks (the blue curves separate, at around 45μm inner diameter, i.e. multimode core size). Adiabatic tapering ensures that light launched into each core evolves into the FMF modes. This lantern has a scrambling transfer matrix since light launched into each core is a superposition of all three lantern modes. Since the propagation constants are degenerate, any perturbation will cause coupling between the three lantern modes, which only enhances the mode-scrambling. At the beginning of the lantern, the lantern modes of the composite waveguide can be approximated by “spots”. Launching light into one of these spots (i.e. independent SMF) will evenly excite each lantern mode. At this point all lantern modes strongly couple since the propagation constants are identical. As the taper continues, the degeneracy between the lantern LP01 and LP11 modes break (splitting of mode index), and there is only strong coupling within the LP11 group.
As stated in section 1, a mode-selective photonic lantern SMUX only needs to multiplex scrambled but orthogonal combinations of modes within the same mode group. For instance, a 3-fiber photonic lantern would only need to distinguish between two mode groups, LP01 and LP11 (6 vector modes). Hence the dissimilar propagation constant distinction has to be applied to only one of the initial fibers. In a simple case like this with only 3 fibers and two mode groups, it is clear that the fiber that will excite the LP01 mode group at the output should have largest propagation constant to start with, hence a larger core. Figure 2(B) shows the modal analysis of this mode-selective lantern. The parameters used for this model are as follows:: index difference of the SMF claddings to the low refractive index jacket (i.e. numerical aperture (NA) of the final multimode waveguide) NA = 0.1; SMF cores, 1 × 14.4 µm diameter and 2 × 12 µm core diameters with a NA = 0.12 for all three cores. At the input, one spatial mode is confined entirely within the larger core (largest mode index), and two spatial degenerate modes are confined within the smaller identical cores. As the inner diameter is decreased, the 2 degenerate modes become the LP11 modes at the FMF end of the lantern, and the larger diameter core mode becomes the LP01 mode. In an adiabatic taper, the difference in effective index inhibits coupling between the different groups. Additionally, the actual mode geometry profiles will favorite a clear evolution of the large diameter core mode into the LP01 mode at the FMF end. The two smaller diameter core modes are strongly mixed, but also evolve only into the FMF LP11 modes. The difference in mode coupling evolution between the standard photonic lantern and the mode-selective one becomes apparent in Fig. 2(A,B).
The mode-selective photonic lantern is scalable to FMF supporting more than three modes. The modes in the FMF can be grouped into near degenerate groups (e.g., 2 × LP01, 4 × LP11, 4 × LP21, 2 × LP02). The modes comprising these groups strongly couple in the FMF and do not need to be individually demultiplexed in a SDM system; this further simplifies the realization of larger mode counts devices. Figure 2(C) shows a 6-fiber photonic lantern designed to support those groups. There are 6 cores with 4 different sizes whose propagation constants are chosen to match the near degenerate mode-groups of the FMF. The parameters used for this model are: index difference of the SMF claddings to the low refractive index jacket (i.e. numerical aperture (NA) of the final multimode waveguide) NA = 0.1; SMF cores, 1 × 10.7 µm diameter, 2 × 9.6 µm diameter, 2 × 8.5 µm diameter and 1 × 7.35 µm diameter with a NA of 0.12 for all six cores. The largest core maps to the LP01 mode, and the smallest core maps to the LP02 mode. The two 2-nd largest cores map to the LP11 mode, and the two second smallest cores map to the LP21 mode. The propagation constants are separated during the taper, and the modal profiles corresponding to these propagation constants evolve from isolated core modes of the dissimilar SMFs into the modes of the FMF photonic lantern output. In practice this scalability will be limited by the range of fiber types that can be used in reality. However, if suitable fibers could be purchased of indeed manufactured with the desired parameters a photonic lantern SMUX with larger number of modes could indeed be fabricated.
3. Mode-selective photonic lantern fabrication
The only way that these photonic lanterns SMUXes can be made is by making a physical transition in which the single-mode waveguides either stop acting as such, and/or cease behaving as independent uncoupled waveguides. The final aim of this physical transition is to adiabatically form a multimode waveguide in which the single-mode waveguides either vanish or form a composite waveguide formed by strong coupling between them. An all solid optical fiber splitter/combiner photonic lantern fabrication technique  was used for the manufacturing of our devices.
As demonstrated by the modeling in Fig. 2(B), two dissimilar fiber/core types are needed in order to have mode-selectivity between the LP01 and the LP11 mode groups. Three fibers were chosen to produce these devices; a 1550B-HP single-mode fiber from Nufern with a core diameter of 9 μm, and two SM980-5.8-125 single-mode fibers from Thorlabs with core diameters of 5.8 µm. Both fibers had the same cladding diameters and core NAs, 125 µm and 0.13 respectively. The 3-fiber bundle was threaded into a fluorine-doped silica capillary with and ID/OD of 275/900 µm and a lower refractive index than the undoped SMF claddings, giving a relative NA of 0.06 for the final FMF core. The number of vector modes in a step-index fiber can be approximated by M ≈V2/2, where M is the number of modes, V is the fiber normalised frequency (or also called V-parameter) . For the FMF output of the photonic lantern to support the needed 6 vector modes (LP01 and the LP11 mode groups) at 1550 nm wavelength and with a NA of 0.06; the multimode core diameter for the final FMF was calculated to be 31.5µm.
Another important factor to take into account to produce a low-loss device is the adiabaticity. In order to secure the adiabaticity criterion theoretically founded in , the down taper length was selected to be > 20mm so the angle along the transition of the tapered composite multimode core was < 0.69°. The fibers and capillary were then fused and tapered down in our glass processing machine to form an all solid FMF at the other end producing the photonic lantern transition device. The final fabricated SMUX photonic lantern had a 21.5mm transition length, 32µm core size and OD of 105µm. Figure 3 (top panel) shows microscope images of different cross-sections of the device along the tapered length. The fiber cores were illuminated at the SMF input by pointing the fibers to a halogen light bulb while taking the cross-section images. The different core sizes of the two fiber types are clearly noticeable on the images when coupling a white light source.
4. Characterization and results
In order to measure the transmission losses of the fabricated devices the cut-back technique was used. A fibre-coupled laser at λ = 1550 nm with a SMF-28 fibre connected at the output was used as the input source. One fibre from the photonic lantern along with the end of the SMF-28 fibre were both cleaved and spliced together, and then the output power at the FMF photonic lantern end was recorded with a large area photo-detector. The input fiber was cleaved a few centimetres after the initial splice and the power was measured again. The transmission losses were calculated by using these two values. This technique ensured that the losses introduced to the system by light coupling and splicing were not taken into account on the final value, giving a true transmission device loss. Furthermore, those devices were connectorized and glued to a bare fiber adaptor, ensuring that light coupled to the cladding along the transition was stripped by absorption and not measured by the photo-detector at the fiber output. This procedure was repeated for each SMF input of the photonic lantern SMUX, and average transmission loss of < 0.3 dB in the SMF to FMF direction was typical for the fabricated devices.
Light at λ = 1550 nm was coupled into each SMF of the photonic lantern SMUX in turn. The near-field images at the FMF output of the photonic lantern SMUX were recorded with an infrared camera. We compared two 3-mode photonic lanterns SMUXs: a conventional lantern with identical SMFs, and the mode-selective lantern fabricated with dissimilar fibers (1 x Nufern 1550B-HP and 2 x Thorlabs SM980-5.8-125), both of them were fabricated with the same low-index capillary and to the same core size; giving an NA of 0.06 at the FMF end of the photonic lanterns. One can appreciate core size differences between the conventional and the mode-selective photonic lanterns in the microscope images of Fig. 4 (A and B, left panels). In those images it is still possible to observe this effect since the cores still confine light at lower visible wavelengths, far away from the device operating wavelength of 1550 nm. Figure 4(A) shows the output of the conventional lantern. Since the device is rotationally symmetric, the output filed images rotate by 120 degrees. This effect should be expected whilst changing the input coupling core one at the time in symmetric triangular core geometries, if equal coupling conditions from all the three cores along the transition are preserved. This symmetric output field result also shows that the quality of the tapered transitions is very good. We assume that any local perturbation (that does not affect the three cores equally) resulting from fabrication imperfections will lead to a non-rotationally symmetric output. Figure 4(B) shows the output of the mode-selective photonic lantern. These output patterns are not longer rotationally symmetric and instead resemble the LP01 and LP11 modes. The mode profiles with non-zero intensity between the LP11 lobes and unequal lobe intensities and geometry, can indicate the extent to which the modes are not purely excited by the photonic lantern. However, the output patterns still resemble the FMF modes and the depth of the valley in the LP11 modes indicates at least 5.5dB mode selectivity. The geometry and unequal lobe intensity could be further explained by the geometry of the FMF output core lantern itself. When fusing three fibers together is very likely for the final core to retain certain triangular symmetry as it is apparent from the images of the FMF outputs shown in Fig. 4. Figure 4(C) shows the simulation of the supported modes of a triangular shaped core similar to that obtained during the photonic lantern fabrication. The simulation was carried out by the finite element method showing that the unequal shape of the mode profiles could be simply a result from the core’s triangular geometry.
The mode-dependent loss (MDL) of the lanterns spliced to a 30 metres length of FMF was measured using a swept wavelength interferometer with spatial diversity operating in reflection mode . The reflection mode measurement allows characterization of the MDL of a single lantern coupled to a FMF. The reflection is provided by Fresnel reflection from the cleaved facet of the FMF. The NA and core size of the FMF end of the photonic lantern SMUX, are 0.06 and 32µm respectively. In order to efficiently couple to the smaller core of the graded-index FMF used in this experiment, the mode field profiles of the photonic lantern output required a demagnification of 3.7 × using a telescope lens system. With the imaging optics, the average insertion loss from the independent SMF inputs of the lantern into the graded-index FMF was 2 dB. Figure 5(A) shows the output of 30 m graded-index FMF when excited by the different input SMFs of the mode-selective photonic lantern SMUX. Even though, the excited modes are not perfectly pure, one can appreciate the effectiveness of the mode-selectivity of the devices. Figure 5(B) shows the reflection transfer matrix of the mode-selective photonic lantern SMUX in the time-domain. In this piece of fiber there is 20-ps dispersion group delay (DGD) between the LP01 and the LP11 modes which “temporally” demultiplexes the mode, facilitating the characterization of the devices. The signal was launched into each independent fiber at the time and the output was also recorded at the different input fibers acting as an output in reflection mode. In the 01-01 cell, the LP01 peak is largest, and in the 11-11 cells, the LP11 peak is largest. When launching LP01, the mode selectivity is 3 dB over the LP11 modes. When launching the LP11 modes, the mode-selectivity is 6.5 dB over the LP01 mode. The MDL was extracted from the transfer matrix following the method explained in . The MDL measured for the mode-selective photonic lantern SMUX was below 0.5 dB.
We have demonstrated a new fiber-based mode multiplexer technology with to our knowledge the lowest reported system insertion loss and mode-dependent loss, of 2 dB and 0.5 dB respectively. These photonic lantern based SMUX devices also have mode selectivity in excess of 5.5 dB. The performance of these devices can indeed be optimized further. For instance, the mode-selectivity could be improved increasing the non-degeneracy of the initial SMF inputs even further, by for example getting bigger core size/NA differences. A unique feature of this SMUX technology is the compatibility with existing FMF based systems. The photonic lantern FMF output could be designed to almost perfectly match the mode field diameters and core/cladding size of the FMF used for transmission. This could be achieved simply by choosing the right capillary to fabricate the photonic lantern SMUX (i.e. NA and size). This will produce a spliceable device further reducing the insertion loss by a significant amount from the currently measured 2 dB. From this value, a large part was due to the optical relay fiber-to-fiber coupling and mode mismatches. The already lower insertion losses have shown drastic improvements in FMF transmission systems already, with record distances in excess of 1500 kms (to be reported).
The photonic lantern SMUX technology offers also the possibility to scale up in number of modes without the penalty increase on insertion and mode-dependent loss that other current methods would have, such as for example spot-based mode couplers . We have shown by simulations that a 6-fiber mode-selective SMUX would be feasible. In theory mode-selective photonic lanterns SMUX with even larger would be possible, however other factors such as adiabaticity of the lantern transitions should be taken into account. The adiabaticity criterion will become harder as the number of modes increases in the system. Furthermore, the parameters needed from the dissimilar cores to fulfill the requirements for mode selectivity may become also difficult in practice when going to a very large number of modes.
Sergio G. Leon-Saval would like to thank Alexander Argyros for useful and stimulating discussions about the principle of operation of photonic lanterns with mode selectivity. All devices reported on this work were manufactured in the Astrophotonics facilities at the University of Sydney.
References and links
1. R.-J. Essiambre, G. Kramer, P. J. Winzer, G. J. Foschini, and B. Goebel, “Capacity limits of optical fiber networks,” J. Lightwave Technol. 28(4), 662–701 (2010). [CrossRef]
2. D. J. Richardson, J. M. Fini, and L. E. Nelson, “Space Division Multiplexing in Optical Fibers,” Nat. Photonics 7(5), 354–362 (2013). [CrossRef]
4. R. Ryf, S. Randel, A. H. Gnauck, C. A. Bolle, A. Sierra, S. Mumtaz, M. Esmaeelpour, E. C. Burrows, R.-J. Essiambre, P. J. Winzer, D. W. Peckham, A. McCurdy, and R. Lingle, “Mode-division multiplexing over 96 km of few-mode fiber using coherent 6 × 6 MIMO processing,” J. Lightwave Technol. 30(4), 521–531 (2012). [CrossRef]
5. N. Bozinovic, Y. Yue, Y. Ren, M. Tur, P. Kristensen, H. Huang, A. E. Willner, and S. Ramachandran, “Terabit-Scale Orbital Angular Momentum Mode Division Multiplexing in Fibers,” Science 340(6140), 1545–1548 (2013). [CrossRef] [PubMed]
6. S. Matsuo, Y. Sasaki, T. Akamatsu, I. Ishida, K. Takenaga, K. Okuyama, K. Saitoh, and M. Kosihba, “12-core fiber with one ring structure for extremely large capacity transmission,” Opt. Express 20(27), 28398–28408 (2012). [CrossRef] [PubMed]
7. B. Zhu, J. M. Fini, M. F. Yan, X. Liu, S. Chandrasekhar, T. F. Taunay, M. Fishteyn, E. M. Monberg, and F. V. Dimarcello, “High-Capacity Space-Division-Multiplexed DWDM Transmissions Using Multicore Fiber,” J. Lightwave Technol. 30(4), 486–492 (2012). [CrossRef]
8. R. Ryf, M. A. Mestre, S. Randel, X. Palou, A. H. Gnauck, R. Delbue, P. Pupalaikis, A. Sureka, Y. Sun, X. Jiang, and R. Lingle, “Combined SDM and WDM transmission over 700-km Few-Mode Fiber,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2013, OSA Technical Digest (online) (Optical Society of America, 2013), paper OW1I.2. [CrossRef]
9. J. Bland-Hawthorn and P. Kern, “Molding the flow of light: photonics in astronomy,” Phys. Today 65(5), 31–37 (2012). [CrossRef]
11. N. K. Fontaine, R. Ryf, J. Bland-Hawthorn, and S. G. Leon-Saval, “Geometric requirements for photonic lanterns in space division multiplexing,” Opt. Express 20(24), 27123–27132 (2012). [CrossRef] [PubMed]
12. N. K. Fontaine, S. G. Leon-Saval, R. Ryf, J. R. Salazar-Gil, B. Ercan, and J. Bland-Hawthorn, “Mode-Selective Dissimilar Fiber Photonic-Lantern Spatial Multiplexers for Few-Mode Fiber,” in 39th European Conference and Exposition on Optical Communications, OSA Technical Digest (CD) (Optical Society of America, 2013), paper PD1.C.3. [CrossRef]
13. S. G. Leon-Saval, A. Argyros, and J. Bland-Hawthorn, “Photonic lanterns: a study of light propagation in multimode to single-mode converters,” Opt. Express 18(8), 8430–8439 (2010). [CrossRef] [PubMed]
14. S. Yerolatsitis and T. A. Birks, “Three-Mode Multiplexer in Photonic Crystal Fibre,” in 39th European Conference and Exposition on Optical Communications, OSA Technical Digest (CD) (Optical Society of America, 2013), paper Mo.4.A.4.
19. D. O. Culverhouse, T. A. Birks, S. G. Farwell, and P. S. J. Russell, “All-fiber 3 × 3 acousto-optic switch,” in Conference on Lasers and Electro-Optics, J. Bowers, D. Miller, D. Scifres, and A. Weiner, eds., Vol. 9 of OSA Technical Digest (Optical Society of America, 1996), paper CWK4.
22. N. Riesen and J. D. Love, “Tapered Velocity Mode-Selective Couplers,” J. Lightwave Technol. 31(13), 2163–2169 (2013). [CrossRef]
23. D. Noordegraaf, P. M. Skovgaard, M. D. Nielsen, and J. Bland-Hawthorn, “Efficient multi-mode to single-mode coupling in a photonic lantern,” Opt. Express 17(3), 1988–1994 (2009). [CrossRef] [PubMed]
24. W. Snyder and J. D. Love, Optical Waveguide Theory (Chapman & Hall, 1983).
25. N. K. Fontaine and R. Ryf, “Characterization of Mode-Dependent Loss of Laser Inscribed Photonic Lanterns for Space Division Multiplexing Systems,” in 2013 18th OptoElectronics and Communications Conference held jointly with 2013 International Conference on Photonics in Switching, (Optical Society of America, 2013), paper MR2_2.