Thermal poling has been the most reliable methodof inducing large second-order optical nonlinearity(SON) of the order of 1 pm�V in amorphous silicateglasses.1 Mobile alkali ions such as Na� migrate to-wards the cathode under the action of the appliedexternal dc electric field at an elevated temperature,leaving behind a cation-depleted region mostly com-posed of immobile nonbridging oxygen ions. The ex-ternal dc electric voltage is dropped across a verynarrow region in which the resultant electric field isvery strong. Under the action of this strong electricfield, hydrogenated species, H� or H3O
�, are injectedinto the anode glass. With a mobility around fourorders of magnitude lower than that of Na� ions,these H� ions, some neutralized by the negativecharges in the depletion region, accumulate within anarrow region under the anode and form a verystrong electric field that can be frozen in the glassmatrix when the glass temperature drops to roomtemperature with the external field still applied in
the process. This frozen-in electric field can acton the intrinsic third-order susceptibility of the glassthrough ��2� � 3��3�Efrozen and generate a SON insidethe otherwise amorphous silicate glasses.2–4 Forpractical applications, the SON has to be formed inoptical waveguides, either planar waveguides or op-tical fibers.4–8 In such waveguiding structures, themeasured SON is determined by the overlap betweenthe optical modal field and the SON profile, which isclosely related to the distribution of migrating ionsand charges. Unlike the case of thermal poling inuniform samples such as bulk fused silica plates,the existence of the interface between waveguidingcores and claddings could significantly affect theprogress of the migrating ions involved and the sub-sequent SON evolution due to the possible imped-ing effect on the migration of ions and charges fromthis interface. In one experiment it was found thatthe induced SON was located near the interfacebetween the top pure SiO2 cladding layer and theGe-doped SiON core layer of the poled slabwaveguide.9 In another Ge-doped planar wave-guide, it was found that, at the standard polingtemperature and voltage �280 °C, 4 kV), the non-linear layer was mostly confined to the upper clad-ding layer in contact with the anode.10 Consideringthe substantial structural irregularity brought to thepoled region by the presence of circular fiber core, thecharacterization of the exact spatial distribution ofthe SON layer at the fiber core-cladding interfacemay prove to be even more crucial in accurately de-
The authors are with the Optical Fibre Technology Centre, Aus-tralian Photonics Cooperative Research Centre, University of Syd-ney, 206 National Innovation Centre, Australian Technology Park,Eveleigh, NSW 1430, Australia. H. An’s e-mail address is [email protected].
Received 23 January 2006; accepted 17 March 2006; posted 27March 2006 (Doc. ID 67440).
termining the overlap between the induced SON andthe guided core modes.
Chemical etching with hydrofluoric (HF) acid hasbeen used to reveal the SON profile in the poledoptical fibers.11,12 Unfortunately, the etching rate isaltered not only by the presence of electric field in theSON layer but also by the Ge doping in the fiber coreand the stress at the core-cladding interface due tothe difference in refractive indices between the fibercore and cladding. So the real distribution of the SONlayer inside the core region could not be resolved fromthe above HF etching results. We have reported thecharacterization of SON profiles in thermally poledfused silica with second-harmonic (SH) microscopywith submicron spatial resolution.13 This techniquemeasures the true spatial distribution of the inducedSON no matter what the mechanism is. Comparedwith those SON profile retrieving techniques basedon improved Maker fringe methods with differentalgorithms,14,15 this SH microscopy technique is par-ticularly suited to characterizing SON profiles in op-tical fibers. We recently applied this technique tovisualize the SON profiles in thermally poled opticalfibers and clearly showed that the core-cladding in-terface had hindered the progression of the SONlayer into the fiber core.16 In this paper we reportmore comprehensive results about the core-claddinginterface effect on the SON profile with an emphasison overcoming this hindering effect to achieve a bet-ter overlap between the induced SON profile and thefiber core. The polarization dependence of the SON inthe fiber core was also examined.
2. Optical Fibers and Second Harmonic Microscopy
The optical fiber used in this experiment was a Dfiber fabricated by the modified chemical vapor dep-osition method with a Ge-doped core. There is acapillary in the fiber cladding for inserting a 50 �mdiameter aluminum (Al) wire as the electrode. Thediameters of the D fiber and the capillary are �260and 75 �m, respectively. Due to the asymmetry in theD-shaped fiber geometry, the core takes a slightlyoval shape in the drawn fiber. The distance betweenthe flat surface and the nearest edge of the hole isaround 32 �m. The fiber core, 5.8 �m in diameter, is�3.6 �m (edge to edge) away from the fiber flat.Therefore to ensure good overlap between the SONlayer and the fiber core, the D fiber was negativelypoled with its flat in contact with the top surface of agrounded Al heater, and the Al wire electrode applieda �1.5 to �4.5 kV voltage. For SH microscopy, thepoled fibers, Al wires removed, were either embeddedin epoxy inside a glass capillary or sandwiched be-tween supporting thin glass plates and silicon wafers.Short sections of around 1.5 mm were then cut per-pendicular to the fiber axis and polished on both sidesto an optical finish. It should be stressed that our aimin this paper is to characterize SON spatial profiles;therefore no attempt at measuring the absolute valueof the induced SON by comparing its SH signal withthat from a reference sample of known SON wasmade.
The SH microscopy was conducted with an in-verted Leica model DMIRBE microscope equippedwith a Leica TCS2MP confocal system and Coher-ent Verdi-Mira tunable pulsed titanium sapphirelaser of 830nm. The microscope was also equippedwith dual photomultiplier transmitted light detec-tors, with a model 505 DCLP dichroic mirror dividingthe detectable spectrum �380–680 nm� at 505 nm be-tween the two channels. The detecting spectrumrange of the longer wavelength channel (channel 1) isfrom 505 to 650 nm. This channel will therefore re-ceive two-photon fluorescence over a wide range. Insome cases we also used it to detect a transmittednonconfocal image of the sample using a 543 nmhelium–neon (He–Ne) laser. The shorter wavelengthchannel (channel 2) was fitted with a 415�10 nm nar-
Fig. 1. SH micrographs of a D fiber poled at 2.0 kV and 280 °C for30 min. (a) SH signal from channel 2; (b) overlay image fromchannels 1 and 2. The SH signal near the top left corner of theimages comes from the cross section of another poled fiber in thesame glass capillary.
row bandpass filter to receive only the SH signal andreject any two-photon excited fluorescence. Both20� and 40� plan-fluorite objectives were used. Thespatial resolution was estimated to be around0.4–0.6 �m. The fundamental laser beam was mainlylinearly polarized along a direction 15° counterclock-wise relative to the horizontal direction of the micro-graphs.
3. Results and Discussions
A. Impeding Effect at the Core-Cladding Interfaceto the Second-Order Optical Nonlinearity at LowerPoling Voltages
A typical SH microscopy result from a D fiber poledat 2 kV at 280 °C for 30 min is shown in Fig. 1. The543 nm He–Ne laser beam was used as illuminationsource. It can be seen that the induced SON, gener-ating a bright SH signal in Fig. 1(a), is distributed ina thin layer beneath the anode surface. The SON wasnot just induced near the fiber flat directly in contactwith the anode, but also in other areas of the claddingaround the cathode hole. It seems that the wholeouter surface of the D fiber had acted as the anode.The more interesting feature is that, close to the core-cladding interface, the SON layer lagged behind theother parts of the layer under the fiber flat, forming aslight curve around the interface. This finding seemsto suggest that the core-cladding interface has actedas an extra barrier for the migrating cations andprevented them from entering the fiber core. A linescan was conducted across the fiber core and the SONlayer and the obtained digital information was ana-lyzed with the result shown in Fig. 2. The SON layerblocked at the interface is 4.16 �m away from thecore center, indicating that the SON layer has not yetreached the fiber core.
B. Overcoming This Impeding Effect
In glasses, ion migration is a thermally activated pro-cess, and the ion mobility � can be expressed as17
��T� �qx0
2
KT �0 exp��E
KT�, (1)
where q is the electric charge, x0 the ion hoppingdistance, �0 the hopping rate, and E the migrationactivation energy. In optical fibers, the existence ofthe core-cladding interface presents an extra poten-tial barrier to the migration ions.16 There are twoways of overcoming this barrier: increasing either thepoling voltage or the poling temperature. The appli-cation of a higher voltage could decrease the activa-tion energy in the poling direction,16 while highertemperature could increase the activity of migrationions. Taking original E as 1.0 eV,18 and keeping otherparameters unchanged, when temperature is in-creased from 280 °C to 320 °C it can be shown thatthe ion mobility is increased by a factor of �4.
In the experiment, both higher poling voltages andtemperatures were tried in order for the ions to reach
Fig. 2. Profile-retrieving result from a line scan across both thefiber core and the core-cladding interface. The scanning line isshown in the inset.
Fig. 3. SH micrographs of a D fiber poled at 2.5 kV and 280 °C for30 min. (a) SH signal from channel 2; (b) overlay image fromchannels 1 and 2.
the fiber core. It was found that with poling voltages�2.5 kV, the SON layer could overcome the interface.A typical result for samples poled at 2.5 kV and280 °C for 30 min is shown in Fig. 3. The distancebetween the anode surface and the SON spot in thecore was measured to be �5.3 �m. Since the corecenter is �6.5 �m away from the anode surface, it isclear that in this case the SON layer, although havingpartly overcome the interface and penetrated into thecore region, had not yet reached the core center. Inthe cladding region, the SON layer is almost parallelto the anode surface and is �4.2 �m under the anode.It can be seen that the SON spot in the core is furtheraway from the anode surface than the SON layer inthe cladding. This seems to suggest a higher ion mo-bility in the Ge-doped fiber core. To drive the SONspot further into the fiber core, samples were poled ata higher voltage �4.5 kV� for longer time �60 min�.
The new SON profile is shown in Fig. 4. Now the SONspot is about 6.8 �m away from the anode and hasreached the fiber core.
To test the effect of high poling temperature, severalfibers were thermally poled at 320 °C with an applieddc voltage of 2.0 KV for 30 min. The correspondingSON profile is shown in Fig. 5. The ion mobility ismuch higher in this case, and the SON layer, �14 �maway from the anode surface, had obviously passedthe whole core region. The slowing-down effect of thecore-cladding interface is still visible from the factthat the SON layer near the interface is still slightlybehind the rest of the SON layer.
C. Polarization Dependence of the Second-Order OpticalNonlinearity in the Fiber Core
From our experience, the electro-optic coefficientsmeasured in thermally poled optical fibers with a
Fig. 4. SH micrographs of a D fiber poled at 4.5 kV and 280 °C for60 min. (a) SH signal from channel 2; (b) overlay image fromchannels 1 and 2.
Fig. 5. SH micrographs of a D fiber poled at 2.0 kV and 320 °C for30 min. (a) SH signal from channel 2; (b) overlay image fromchannels 1 and 2.
Mach–Zehnder interferometer have little dependenceon the polarization of the detecting laser beam. Herewe checked the SH signal from a fiber sample poled at4.5 kV and 280 °C for 60 min placed under the mi-croscope with the poling direction either nearly par-allel or perpendicular to the polarization direction ofthe incident fundamental laser beam. The result isshown in Fig. 6. In accordance with our previousexperience, it can be seen that while the SON layeroutside the core-cladding interface became muchweaker when the poling direction became perpendic-ular to the polarization of the fundamental laserbeam, the intensity of the SH signal from the fibercore remained almost unchanged. We believe suchpolarization independence is mainly due to complexdistribution around the core-cladding interface of theinvolved charges that are responsible for the creation
of the electric field in the fiber core. Indeed the SONregions in the two perpendicular configurations arein a similar ellipse shape, suggesting a radial distri-bution for the electric field in the fiber core.
4. Conclusion
We have measured the spatial progression of theSON in thermally poled optical fibers with second-harmonic microscopy. The results show that, at lowerpoling voltages and a standard poling temperature of280 °C, the progression of the SON layer has beenimpeded at the core-cladding interface, which actedas an extra potential barrier to the migrating ions.Poling at both higher voltages and temperatures canhelp the nonlinear layer to partially overcome thisbarrier and extend into the fiber core. The opticalnonlinearity in the fiber core was polarization inde-pendent on the input fundamental laser beam due tocomplex distribution of charges around the core-cladding interface.
The authors acknowledge the facilities as well asscientific and technical assistance from staff in theNanostructural Analysis Network Organisation Ma-jor National Research Facility at the Electron Micro-scope Unit, the University of Sydney.
References1. R. A. Myers, N. Mukherjee, and S. R. J. Brueck, “Large second-
order nonlinearity in poled fused silica,” Opt. Lett. 16, 1732–1734 (1991).
2. P. G. Kazansky and P. St. J. Russell, “Thermally poled glass:frozen-in electric field or oriented dipoles?” Opt. Commun. 110,611–614 (1994).
3. T. G. Alley, S. R. J. Brueck, and R. A. Myers, “Space chargedynamics in thermally poled fused silica,” J. Non-Cryst. Solids242, 165–176 (1998).
4. D. Wong, W. Xu, S. Fleming, M. Janos, and K. M. Lo,“Frozen-in electrical field in thermally poled fibers,” Opt. FiberTechnol. 5, 235–241 (1999).
5. A. C. Liu, M. J. F. Digonnet, and G. S. Kino, “Electro-opticphase modulation in a silica channel waveguide,” Opt. Lett. 19,466–468 (1994).
6. P. G. Kazansky, L. Dong, and P. St. J. Russell, “High second-order nonlinearities in poled silicate fibers,” Opt. Lett. 19,701–703 (1994).
7. X.-C. Long, R. A. Myers, and S. R. J. Brueck, “A poled elec-trooptic fiber,” IEEE Photon. Technol. Lett. 8, 227–229(1996).
8. J. Arentoft, M. Kristensen, K. Pedersen, S. I. Bozhevolnyi, andP. Shi, “Poling of silica with silver-containing electrodes,” Elec-tron. Lett. 36, 1635–1636 (2000).
9. J. Arentoft, K. Pedersen, S. I. Bozhevolnyi, M. Kristensen, P.Yu, and C. B. Nielsen, “Second-harmonic imaging of poledsilica waveguides,” Appl. Phys. Lett. 76, 25–27 (2000).
10. D. Faccio, A. Busacca, D. W. J. Harwood, G. Bonfrate, V.Pruneri, and P. G. Kazansky, “Effect of core-cladding interfaceon thermal poling of germano-silicate optical waveguides,”Opt. Commun. 196, 187–190 (2001).
11. P. Blazkiewicz, W. Xu, D. Wong, and S. Fleming, “Mechanismfor thermal poling in twin-hole silicate fibers,” J. Opt. Soc. Am.B 19, 870–874 (2002).
12. N. Myrén and W. Margulis, “Time evolution of frozen-in fieldduring poling of fiber with alloy electrodes,” Opt. Express 13,3438–3444 (2005).
Fig. 6. SH micrographs of a D fiber poled at 4.5 kV and 280 °C for60 min with the poling direction either (a) parallel or (b) perpen-dicular to the polarization of the fundamental laser beam.
13. H. An, S. Fleming, and G. Cox, “Visualization of second-ordernonlinear layer in thermally poled fused silica glass,” Appl.Phys. Lett. 85, 5819–5821 (2004).
14. A. Kudlinski, Y. Quiquempois, M. Lelek, H. Zeghlache, and G.Martinelli, “Complete characterization of the nonlinear spatialdistribution induced in poled silica glass with a submicronresolution,” Appl. Phys. Lett. 83, 3623–3625 (2003).
15. A. Ozcan, M. J. F. Digonnet, and G. S. Kino, “Iterative pro-cessing of second-order optical nonlinearity depth profiles,”Opt. Express 12, 3367–3376 (2004).
16. H. An and S. Fleming, “Hindering effect of the core-claddinginterface on the progression of the second-order nonlinearitylayer in thermally poled optical fibers,” Appl. Phys. Lett. 87,101108 (2005).
17. A. Feltri, S. Grandi, P. Mustarelli, M. Cutroni, and A. Man-danici, “GeO2-doped silica glasses: an ac conductivity study,”Solid State Ionics 154–155, 217–221 (2002).
18. O. Deparis, C. Corbari, P. G. Kazansky, and K. Sakaguchi,“Enhanced stability of the second-order optical nonlinearity inpoled glasses,” Appl. Phys. Lett. 84, 4857–4859 (2004).
