All-solid photonic bandgap fiber-ol-29-20-2369

October15,2004/Vol.29,No.20/OPTICS LETTERS2369 All-solid photonic bandgap fiberF.Luan,A.K.George,T.D.Hedley,G.J.Pearce,D.M.Bird,J.C.Knight,and P.St.J.RussellDepartment of Physics,University of Bath,Bath BA27AY,UKReceived June22,2004We describe the design and fabrication of a photonic bandgap fiber formed with two different glasses.As ina hollow-core fiber,light is guided in a low-index core region because of the antiresonances of the high-indexstrands in the fiber cladding.The structure described represents an ideal bandgap fiber that exhibits nointerface modes and guides over the full width of multiple bandgaps.©2004Optical Society of AmericaOCIS codes:060.2270,060.2280,060.2310.Hollow-core photonic crystal fibers(HC-PCFs)in which light is guided by a photonic bandgap cladding material have attracted great interest in recent years.1–3Most reported HC-PCFs consist of a central air core surrounded by a periodic array of air holes in a silica background.When the air holes in a HC-PCF are large,the structure resembles a honeycomb array of thin silica strands supported by a mesh of even finer webs.This structure is a compromise between performance and feasibility.The fiber performance would be improved if the connecting webs of silica were thinner or were completely absent,but such a structure is not self-supporting.In the simpler case of an array of unconnected high-index rods in a low-index background,the bandgaps of the cladding are easily seen to arise from the antiresonance of each individual high-index cladding rod.4Such fibers have been demonstrated5–7by putting high-index f luid into a short length of standard solid-core silica–air photonic crystal fiber.8,9However,those fibers are hard to use and have an intrinsically limited length, as well as being unstable to temperature and other environmental effects.In this Letter we report the realization and characterization of a similar fiber based on two thermally matched silicate glasses.Our fibers are practical and stable and can be readily drawn down for access to the fundamental bandgap. They provide an ideal system to study the basic features of photonic bandgap guidance,which are complicated in hollow-core fibers by interface-mode effects.2The structure that we have studied is a two-dimensional triangular array of high-index rods (refractive index n1෇1.79)embedded in a low-index background(index n2෇1.54).Computations havebeen performed with a variation of the plane-wave method10in which we compute the photonic states occurring in a range of values of the propagation constant b(the wave-vector component along the fiber axis)for a fixed free-space propagation constant k0.Figure1is a bandgap diagram based on the computed density of photonic states10for three arrays of circular high-index rods͑n1෇1.79͒of the same diameter but spaced by different amounts in a low-index matrix͑n2෇1.54͒.The scaling of the y axes is normalized by rod diameter d,while the x axis shows ͑b22n22k02͒d2,so that the glass line representing the cutoff for the continuous core material n2appears as a vertical line at zero.11The most obvious feature of the these plots is that the frequency of the funda-mental bandgap remains approximately unchanged even though the pitch of the array has changed by a factor of3,direct evidence that the mechanism behind bandgap formation is antiresonances of the individual rods.Although the pitch does not(to first order)change the bandgap frequency,it does affect the shape and depth of the bandgap as a result of the changing coupling strength.In moving from Fig.1(a) to Fig.1(c)the bandgap gets narrower and deeper, which will affect the confinement of waveguiding fibers designed using this type of cladding.To realize the above structure we used two ther-mally compatible silicate glasses,LLF1and SF6 from Schott Glass,Ltd.LLF1has an index near 1.54at visible and near-IR wavelengths,while the index of SF6isϳ1.79.Similar glasses were used in recently reported work12to make index-guiding photonic crystal fibers.In this demonstration we used a multiple-stacking method in which each SF6 rod was surrounded by6LLF1rods.The central area,which would have contained7SF6rods and12 LLF1rods,was replaced with one big LLF1rod to form the core.The stack was drawn to fiber in two drawing stages,between which a thick SF6jacket was added.Figure2(a)shows a backscattering image of the endface of a fiber taken with a scanningelectron Fig.1.Bandgap plots of the same size rods placed at dif-ferent distances,with the white regions having zero den-sity of photonic states.The background index was set to n2෇1.54,and the index of the rods was n1෇1.79.0146-9592/04/202369-03$15.00/0©2004Optical Society of America2370OPTICS LETTERS/Vol.29,No.20/October15,2004Fig.2.(a)Backscattered electron images of the all-solid photonic bandgap fiber.The black structure is the LLF1 rods and the white areas are the SF6rods.(b)Measured near-field patterns of guided modes.A3-nm bandpass fil-ter was used at the input.The wavelengths for the top and bottom images were690and575nm,respectively. The transmission spectrum of the fiber sample is shown in Fig.4(a),below.microscope.The bright white regions are SF6,and the dark areas are LLF1.Observed near-field images of guided modes in this fiber are shown in Fig.2(b). During drawing,the SF6rods turn into star shapes, because at the drawing temperature SF6is softer than LLF1.Figure3presents numerical simulations of an array of star-shaped high-index rods derived from electron microscope images of real fibers,demon-strating that the shape of the high-index rods has a relatively small effect on the position and shape of the photonic bandgaps,especially the lower-order ones. The modeled structure consists of an array of touching low-index circles superimposed on a high-index back-ground(see Fig.3,top left).A transmission spectrum recorded with a fiber sample of20-cm length is shown in Fig.4.The sample had a 1.83-m m pitch,and the transmis-sion spectrum was recorded with a supercontinuum white-light source.13The transmitted spectrum shows several low-attenuation windows,which stand out from the background by more than30dB.To model the structure precisely we performed com-putations based on the real structure over a broad frequency range and included the material dispersion of the two component materials.14The density of states plotted in Fig.4was computed along a line of b෇n2k0.By comparing the computed density of states and the observed transmission bands,we can see that the low-loss regions observed experimentally correspond to the second-,third-,and fourth-order bandgaps[Fig.4(a)].From our numerical computa-tions we predict that the fundamental bandgap occurs in the wavelength range near 1.8m m,beyond the limit of our experimental capability.In preparing Fig.4we used commercially available data on n1and n2and their dispersions,14and structural information was directly measured from the electron microscope images.There were no adjustable parameters in the fit.The relative width of each band fits well with the simulation,which is not the case with HC-PCFs.2,15In hollow-core fibers the transmission bands are reduced because of surface-confined guided modes.These modes match the propagation constant of the hollow-core modes at certain wavelengths, causing increased coupling out of the fundamental mode that manifests as attenuation.In the fiber being described here the interface between the core and the cladding is better matched than in previ-ously reported hollow-core fibers,16and we see no evidence of surface-confined guided modes.We have drawn smaller fibers in which we have identified the fundamental bandgap:for example,a fiber with a pitch of1.17m m exhibited the fundamental bandgap in the wavelength band from1050to1330nm,with higher-order bandgaps near626and766nm.We measured the attenuation in our fiber by use of a cutback technique and found it rather high at ϳ18dB͞m at the bandgap center.The intrinsic at-tenuation of the core material is less than1dB͞m over the visible and near-IR wavelength range.We at-tribute this to the presence of bubbles at the interfaces between the two materials,which arise during our multiple-stacking process and are visible in some fibers.It is possible to reduce or eliminate this effect by improving fabrication procedures.12We used low-coherence interferometry to measure the group index over two guiding bands as shown in Fig.4(b).Our system uses a tunable3-nm bandwidth source that we obtained by passing a supercontinuum generated from a1064-nm pump source13through a monochromator.As in hollow-core fibers,the group index curve in each band drops to a minimum before rising again.In hollow-core fibers the minimum (which corresponds to the zero in group-velocity dis-persion)is shifted toward the short-wavelength edge of the bandgap as a result of the effects of anomalous waveguide dispersion.17However,in these solid-core fibers the strongly normal dispersion of the core material becomes increasingly dominant at short wavelengths,shifting the minimum point of the group index curve to the long-wavelength side of the trans-mission band.The contribution from the waveguide dispersion is comparatively weak near thecenterparison of bandgap plots for the idealized array of circular high-index rods and for the array of star-shaped high-index rods found in the actual f ibers. High-index material is represented in black.October15,2004/Vol.29,No.20/OPTICS LETTERS2371Fig.4.(a)Comparison of the transmission spectrum mea-sured through a fiber sample of20-cm length with the nor-malized density of states11(DOS)computed along the line b෇n2k0.The spikes near the1064-nm wavelength are the residual pump that was used to form the supercon-tinuum for the measurement.(b)Measured group index of the fundamental guided mode and the group-velocity dis-persion(GVD)across two of the bandgaps.The dashed curve is the core material group index.of the bandgap.17One interesting feature of this kind of fiber is that it is possible to match the group indices at widely different wavelengths because of the presence of multiple bandgaps.Relatively weak bire-fringence was found due to the slight asymmetry in the actual fiber structure.We observed propagation of higher-order core modes in our fibers,as predicted by a supercell computation,as shown in Fig.2(b). The number of guided modes in photonic bandgap fibers is determined by the depth of the bandgap and the core size1:for our cladding design we would need a considerably smaller core to guide just a single mode.However,fibers with smaller cores are difficult to excite efficiently and have higher attenua-tion.Experimentally,we can easily excite purely the fundamental mode in the fibers described here,and the fundamental mode does not couple to higher-order guided modes even when the fiber is bent or twisted. It is worth noting that exclusively higher-order modes are observed near the long-wavelength edges of the transmission bands.In conclusion,all-solid photonic bandgap fibers have been successfully made and characterized.We have observed transmission through the first five bandgaps in these fibers,and we attribute the bandgap forma-tion to antiresonances of the high-index strands in the cladding.The structures represent model systems for photonic bandgap fibers,with isolated high-index rods in a low-index background,and do not exhibit effects from surface modes because of the detail of the core–cladding interface.F.Luan’s e-mail address is pypf l@. References1.R.F.Cregan,B.J.Mangan,J.C.Knight,T.A.Birks,P.St.J.Russell,P.J.Roberts,and D.C.Allan,Science 285,1537(1999).2.B.J.Mangan,L.Farr,ngford,P.J.Roberts,D.P.Williams, F.Couny,wman,M.Mason,S.Coupland,R.Flea,H.Sabert,T.A.Birks,J.C.Knight,and P.St.J.Russell,in Optical Fiber Com-munication Conference(OFC),Vol.96of OSA Trends in Optics and Photonics Series(Optical Society of America,W ashington,D.C.,2004),postdeadline paper PDP24.3.J.C.Knight,Nature424,847(2003).4.N.M.Litchinitser,S. C.Dunn, ner, B.J.Eggleton,T.P.White,R. C.McPhedran,andC.M.de Sterke,Opt.Express11,1243(2003),http://.5.J.Jasapara,R.Bise,and R.Windeler,in Optical FiberCommunication Conference(OFC),Vol.70of OSA Trends in Optics and Photonics Series(Optical Society of America,W ashington,D.C.,2002),p.519.6.R.T.Bise,R.S.Windeler,K.S.Kranz,C.Kerbage,B.J.Eggleton,and D.J.Trevor,in Optical Fiber Com-munication Conference(OFC),Vol.70of OSA Trends in Optics and Photonics Series(Optical Society of Amer-ica,W ashington,D.C.,2002),p.466.rsen, D.S.Hermann,J.Broeng,andA.Bjarklev,presented at the29th European Confer-ence on Optical Communications(ECOC’03),Rimini, Italy,September21–25,2003.8.J.C.Knight,T.A.Birks,P.St.J.Russell,and D.M.Atkin,Opt.Lett.21,1547(1996).9.J.C.Knight,T.A.Birks,P.St.J.Russell,and D.M.Atkin,Opt.Lett.21,484(1997),erratum of Ref.8. 10.J.M.Pottage,D.M.Bird,T.D.Hedley,T.A.Birks,J.C.Knight,P.St.J.Russell,and P.J.Roberts,Opt.Express11,2854(2003),.11.T.A.Birks,D.M.Bird,T.D.Hedley,J.M.Pottage,and P.St.J.Russell,Opt.Express12,69(2004),http:// .12.X.Feng,T.M.Monro,P.Petropoulos,V.Finazzi,and D.Hewak,Opt.Express11,2225(2003),http:// .13.W.J.W adsworth,N.Joly,J.C.Knight,T.A.Birks,F.Biancalana,and P.St.J.Russell,Opt.Express12,299(2004),.14.Schott Optical Glass Catalog-Data Sheets,http:///optics_devices/english/download.15.C.M.Smith,N.Venkataraman,M.T.Gallagher,D.Muller,J.A.West,N.F.Borrelli,D.C.Allan,andK.W.Koch,Nature424,657(2003).16.M.J.F.Digonnet,H.K.Kim,J.Shin,S.Fan,andG.S.Kino,Opt.Express12,1864(2004),http://.17.D.G.Ouzounov,F.R.Ahmad,D.Muller,N.Venkatara-man,M.T.Gallagher,M.G.Thomas,J.Silcox,K.W.Koch,and A.L.Gaeta,Science301,1702(2003).。