Amorphous silicon-doped titania films for on-chip photonics
Thomas Kornher, Kangwei Xia, Roman Kolesov, Bruno Villa, Stefan Lasse, Cosmin S. Sandu, Estelle Wagner, Scott Harada, Giacomo Benvenuti, Hans-Werner Becker, Jörg Wrachtrup
aa r X i v : . [ c ond - m a t . m e s - h a ll ] F e b Amorphous silicon-doped titania films for on-chip photonics
Thomas Kornher, ∗ Kangwei Xia, Roman Kolesov, Bruno Villa, Stefan Lasse, Cosmin S. Sandu, Estelle Wagner, Scott Harada, Giacomo Benvenuti, Hans-Werner Becker, and J¨org Wrachtrup
3. Physikalisches Institut, Universit¨at Stuttgart, 70569 Stuttgart, Germany Semiconductor Physics Group,Cavendish Laboratory, JJ Thomson Avenue, Cambridge,CB3 0HE,UK RUBION, Ruhr-Universit¨at Bochum, 44780 Bochum, Germany
High quality optical thin film materials form a basis for on-chip photonic micro- and nano-devices,where several photonic elements form an optical circuit. Their realization generally requires thethin film to have a higher refractive index than the substrate material. Here, we demonstrate amethod of depositing amorphous 25% Si doped TiO films on various substrates, a way of shapingthese films into photonic elements, such as optical waveguides and resonators, and finally, theperformance of these elements. The quality of the film is estimated by measuring thin film cavityQ-factors in excess of 10 at a wavelength of 790 nm, corresponding to low propagation lossesof 5.1 db/cm. The fabricated photonic structures were used to optically address chromium ionsembedded in the substrate by evanescent coupling, therefore enabling it through film-substrateinteraction. Additional functionalization of the films by doping with optically active rare-earth ionssuch as erbium is also demonstrated. Thus, Si:TiO films allow for creation of high quality photonicelements, both passive and active and also provide access to a broad range of substrates and emittersembedded therein. λ≈
800 nmQ=146000
Si:TiO g l a ss / Y A G / s a pp h i r e Introduction
Nanoscale fabrication of optical waveguide structuresoffers a wide range of opportunities spanning from on-chip photonic devices for optical networks to sensor ap-plications. At the heart of such devices commonly lies acertain functionality that can be readily interfaced withother components, and be included into sophisticated ar-chitectures as it is standard with silicon-on-chip technol-ogy [1]. Material systems hardly compatible with siliconphotonics, such as YAG, YSO or sapphire crystals, canhost a large variety of emitters with applications rangingfrom quantum communication [2–4] and quantum memo-ries [5, 6] to lasing materials [7, 8]. To exploit their poten-tial in a scalable architechure, these substrates need to be ∗ [email protected] enabled by a waveguiding platform. Previously demon-strated photonic structures in these materials based onfemtosecond laser-writing share common drawbacks likea low refractive index contrast between waveguide andsubstrate [9]. The corresponding increase of the mini-mum feature size of photonic elements by up to two or-ders of magnitude makes this technology unpractical forthe field of cavity quantum electrodynamics (CQED) andnanoscale photonics in general.An alternative way to realize a waveguiding platformis the deposition of a high refractive index thin film ontop of these substrate crystals, which then provides on-chip access to these systems through evanescent lightfields [10, 11] coupling to embedded emitters. Moreover,this approach leaves the substrate crystal untouched fromprocessing and preserves the spectroscopic properties ofembedded emitters [12].In the visible range, titanium dioxide (TiO ) featuresthe highest refractive index out of a variety of transpar-ent thin film materials, thus allowing for waveguiding onthe majority of transparent substrates. Additionally, ithas a wide transparency range covering the whole visi-ble and near-infrared spectral regions. Shaping the de-posited TiO film into photonic elements can then beconveniently done by reactive ion etching (RIE) [13]. Thedeposited TiO films often tend to form nano-crystalliteswhich lead to significant scattering and, therefore, to sub-stantial optical losses [14, 15]. The best performance istypically shown by amorphous TiO films featuring a re-fractive index between 2.3 and 2.4 [15–19]. In order topreserve the amorphous composition of the film, specialprecautions have to be taken.In the present work, we report on Chemical Beam VapourDeposition (CBVD) of high optical quality Si:TiO filmwhose amorphous state is preserved by doping with sil-icon [20]. The amorphisation of deposited TiO with Sidoping was already reported [21], and a more recent de-tailed study of amorphisation of Nb:TiO thin films bydoping with Si is reported elsewhere [22]. Even thoughthis doping leads to a slight decrease in the refractive in-dex of the film, in agreement with results achieved withother Chemical Vapour deposition techniques [23], the in-dex is still high enough to support waveguiding on high-index optical crystals such as YAG and sapphire. On theother hand, Si doping keeps the TiO from forming nano-crystallites even at elevated temperatures up to 650 ◦ C,resolving the inherent thermal instability of amorphousTiO films to some extent [24].The optical performance of the film was assessed byfabricating whispering gallery mode (WGM) cavitiesand measuring their Q-factor. The latter was in excessof 10 . Evanescent coupling to fluorescent spots in thesubstrate material can be shown and also additionalfunctionality can be added to the Si:TiO film by dopingit with fluorescent rare-earth ions, such as erbium. Thisshows the potential of Si:TiO nanophotonic structuresto also act as active elements within photonic devices. Deposition of Si:TiO combinatorial film Oxide thin films were deposited by CBVD, as de-scribed in detail elsewhere [21]. The CBVD processmakes use of thermal decomposition of organometallicprecursors on a heated substrate in a high vacuum en-vironment ( ≤ − Pa). These precursors impinge uponthe substrate as molecular beams and do not undergoany gas phase reaction. Deposition was conducted onepi-polished YAG crystals (CrysTec) and quartz forfunctional characterization and on Si and glass wafersfor material characterization. The liquid organometallicprecursors used during the present investigation weretitanium tetraisopropoxide (Ti(OiPr) , CAS 546-68-9evaporated from a reservoir at 32 ◦ C) and tetrabutoxysi-lane (Si(OnBu) , CAS 4766-57-8, evaporated from areservoir at 65 ◦ C). The Ti and Si precursor flows werehomogeneously distributed on the substrate (the flowratio of precursors was estimated as Si/Ti=1.1). Thesubstrate temperature during the deposition was keptat 500 ◦ C. Under these conditions, the Si doping in thefilm (defined as Si/(Si+Ti) ratio) was about 25% (asmeasured by SEM-EDX) and the growth rate was about10 ± × − Pa.A liquid nitrogen-cooled cryo-panel helped to maintaina pressure below 2 × − Pa during the deposition. Themorphology and the chemical composition of the thinfilms were investigated by SEM-EDX using a MerlinSEM and in TEM cross-section using a Tecnai Osirismicroscope. From cross-sectional TEM images, we canestimate the average thickness of deposited films andtheir growth morphology. Typical images of character-ized films are presented in Figure 1 a) and b) for films ofthicknesses 800 ±
100 nm. A High Angle Annular DarkField image (Figure 1 a) shows a homogeneous compact a) b)c)
200 400 600 800 1000 1200 1400 16002.02.12.22.32.42.52.62.72.82.9 n (Si:TiO , thin film)n (Anatase TiO , bulk)k (Si:TiO , thin film) Wavelength [nm] n k FIG. 1. Cross-sectional view by TEM of a Si:TiO film de-posited on glass substrate. a) HAADF-STEM image andSAED pattern. b) HRTEM image with inserted FFT. c) Re-fractive index for an unpatterened Si:TiO film on silica sam-ple and for comparison the refractive index of bulk anataseTiO [25]. n and k denote the refractive index and the ex-tinction coefficient, respectively. film with relatively low roughness. The inserted SelectedArea Electron Diffraction pattern together with theHigh Resolution TEM image (Figure 1 b) confirm theamorphous phase of the film. Further characterizationof the unpatterned Si:TiO film of a thickness of 535 nmby ellipsometry yields the dispersion of the refractiveindex as shown in figure 1 c). Within the transparencywindow of the film starting roughly at 400 nm andextening all the way to the infrared, the refractive indexranges between 2.3 and 2.1. Structuring Si:TiO films In order to assess the optical quality of the film, we fab-ricated monolithic optical WGM resonators and testedthe width of their resonances. The resonators were mi-crorings evanescently coupled to a nearby optical waveg-uide through which excitation light was supplied.The patterning was done by standard RIE with a Nimask defined by e-beam lithography. RIE of Si:TiO was performed in an atmosphere of Ar/CF /O withthe respective flow rates 4/16/3 sccm and the RF power a) b) FIG. 2. a) SEM image of a Si:TiO WGM resonator evanes-cently coupled to a straight optical waveguide. b) Close-up ofthe coupling region between ring resonator and access waveg-uide. being 100 W [26]. The process pressure was 15 mTorrand the total time required to etch through 535 nm ofSi:TiO was 24 minutes. After etching, the nickel maskwas still present indicating that the etching selectivitywas better than 1:11. The residual nickel mask wasremoved by dissolving the metal in an aqueous 1Msolution of nitric acid. In order to remove damageintroduced into Si:TiO films by ion bombardmentduring plasma etching, the resulting structures wereannealed in air for 4 hours at 500 ◦ C. However, annealingthe film at temperatures above 800 ◦ C leads to a visualchange of the film, suggesting recrystallization. SampleSEM images of the resulting structures are shown inFigure 2. The radius of the WGM resonator was 8 µ mwhile the length of the waveguide was 25 µ m. Thedistance between the resonator and the waveguide wasaround 400 nm. Optical characterization of waveguides and res-onators on various substrates
In the following, waveguides and resonators were fab-ricated out of Si:TiO thin films on two different sub-strates, namely YAG and silica with a refractive indexof 1.82 and 1.45 at 790 nm, respectively. Studies of theoptical performance of waveguides and cavities were per-formed in a home-built confocal microscope with an addi-tional ability of scanning the detection point in the vicin-ity of the position of the excitation laser focus [27]. Itsschematic is shown in Figure 3 a). The excitation laserwas focused onto the sample with a high numerical aper-ture objective lens (Olympus 0.95 × Al O ) was doped with chromium byion implantation through a perforated copper mask withan average hole diameter of ≈
400 nm. With an energyof 100 keV and a dose of 10 cm − , Cr ions end up inthe YAG crystal in a depth of ≈ ±
22 nm according toSRIM simulations [28]. A Si:TiO film was subsequentlydeposited and shaped into waveguides in order to asseswaveguiding and evanescent coupling of waveguided lightto shallow implanted Cr-doped spots. Cr fluoresence ofimplanted spots could be detected close to 700 nm un- X-axis [µm] Y - a x i s [ µ m ] a)b)c) X-axis [µm] Y - a x i s [ µ m ] FIG. 3. a) Schematic diagram of the microscope. The sam-ple is mounted on the 3D nanopositioner and can be movedthrough the laser focus. Light emission is collected throughthe same objective lens and sent through a 2F-2F telescopeand a tip-tilt mirror onto the pinhole selecting the observationpoint on the sample. b) Standard confocal microscope scanwith overlapping excitation and observation point. Whitespots represent fluorescence of Cr implanted spots in YAG,two of them underneath a fabricated Si:TiO waveguide. c)Tip-tilt mirror scan with excitation point kept at a stationaryposition to couple light into the fabricated Si:TiO waveguide(left end). The two white spots at around 7 µ m and 18 µ mon the X-axis represent fluorescence of Cr implanted spots,evanescently excited by light traveling through the waveguide. der excitation with 600 nm light as shown in the confocalscan in Figure 3 b). Since the refractive index of YAGis smaller than the index of the film, waveguiding couldbe observed. Incoupled light traveling within the waveg-uide was able to evanescently excite Cr implanted spotsas shown in the tilting mirror scan in Figure 3 c). Here,the excitation laser position was kept stationary couplinglight into the left end of the waveguide. By scanning thepoint of detection with the tilting mirror, not only theexcitation laser position yields signal, but also implantedCr spots light up, which are embedded below the waveg-uide extending to the right. This confirms the evanescentcoupling of light between fabricated waveguide and emit-ters in the substrate. In combination with high qualityresonators, this film-substrate interaction has the poten-tial to facilitate CQED experiments with rare-earth iondoped crystals [12].For characterizing fabricated Si:TiO resonators onglass substrates, the widths of the cavity resonances werestudied with a single frequency tunable diode laser (Top- a) λ =787.98 nm b) Laser Detuning (GHz) I n t e n s i t y ( A r b . U n i t s ) I n t e n s i t y ( A r b . U n i t s ) Wavelength (nm)780 785 790 795 80004812 c) Incoupling Point
FIG. 4. a) Camera image of a structure excited on resonance.The left end of the waveguide was used as input and the cavitymode is visualized by the residual scattering of the laser beingin resonance. Tunable blue diode laser was used for visualiza-tion. b) The spectral shape of the cavity mode was obtainedby sweeping a single-mode laser through the cavity resonance.c) Typical mode spectrum of the studied resonator geometrybetween 780 nm and 800 nm obtained with a broadband lightsource. Mode width measurement is limited by the resolutionof the spectrometer for the fundamental mode. tica DL Pro). The laser could be tuned coarsely overthe range of 770-800 nm with a mode-hop-free fine tun-ing range of 30 GHz. The spectral width of the laserwas below 1 MHz. Laser output was inserted into one ofthe ends of the optical waveguide and its frequency wasswept while monitoring the scattered emission at the rimof the cavity. The rim lit up once the laser was in reso-nance with one of the cavity modes due to residual scat-tering on the imperfections of the structured film (seeFigure 4 a)). This scattering was detected as a func-tion of laser frequency to estimate the spectral width ofthe mode. The result of the spectral measurements isshown in Figure 4 b). The estimated mode width was2.6 GHz as the laser was finely swept around 787.89 nmwavelength. This value corresponds to the quality factorof Q ≈ R = 8 µ m, a film thickness of535 nm and a rim width of 2 µ m, the mode spectrum wasmeasured with a broadband light source in a wavelengthregion between 780 nm and 800 nm in order to extractthe free spectral range (FSR) of ∆ λ FSR = 5 .
60 nm for thefundamental mode. Including the refractive index mea-surement, our corresponding finite element method basedsimulation can confirm the measured FSR for the funda-mental mode in this resonator geometry. The respec-tive resonator sketch and the electric field profile of themode is shown in Figure 5. With the FSR measurementaround 790 nm, the group index of 2.18 was estimated by n g ≈ λ / (2 πR · ∆ λ FSR ). Based on the quality factormeasurement and the group index estimation, propaga-tion losses α in fabricated waveguides of 5.1 dB/cm wereestimated by α = 2 πn g /Qλ .Table I compares different TiO -based thin film waveg-uiding structures based on their propagation loss. Thedoped Si:TiO film reaches benchmark propagation lossesin the visible in exchange for a doping dependent decreaseof the refractive index. TABLE I. Comparison of propagation loss in dB/cm in TiO -based thin film waveguiding structures.Film Loss at Loss at Referencedetails 600-800 nm 1550 nmRF magnetron 9 4 [17]sputteringlaser molecular 57 - [14]beam epitaxyatomic layer - 2.4 [16]depositionRF magnetron 9.7 - [19]sputteringCBVD 5.1 - this workFIG. 5. Resonator sketch with calculated electric field profileof the fundamental 790-nm mode. Doping of Si:TiO resonators with erbium Another way of adding functionality to the thin film,specifically to the fabricated thin film resonators, is byoptical activation with fluorescent species. We have cho-sen rare-earth (RE) doping due to robust optical prop-erties of RE ions in most crystalline and glassy trans-parent media. RE doping can be performed by ion im-plantation with very high yield of fluorescent species [29].In addition, optical materials doped with erbium exhibitstrong upconverted fluorescence in the visible once ex-cited in the infrared. This makes erbium doping conve-niently detectable. The Si:TiO film on glass was im-planted with erbium ions of 2 MeV energy and with adose of 10 cm − . According to SRIM simulations, this a) b)c) d)
25 µm
525 530 535 540 545 5555500841216
Wavelength (nm) U p c o n v e r t e d F l u o r e s c e n c e I n t e n s i t y ( A r b . U n i t s )
525 530 535 540 545 555550
Wavelength (nm) U p c o n v e r t e d F l u o r e s c e n c e I n t e n s i t y ( A r b . U n i t s ) FIG. 6. a) Spectrum of upconverted fluorescence of erbium inSi:TiO film. Excitation wavelength is 785 nm. b) Microscopeimage of the WGM resonators manufactured out of Er -implanted film. Scale bar is 25 µ m long. c) Cavity modevisualized by upconverted fluorescence. The excitation laseris in resonance with one of the infrared cavity modes. d)Spectrum of erbium upconverted fluorescence collected at theend of the waveguide. The smooth background is due to thefluorescence of the waveguide. The sharp peaks correspondto resonances of the cavity. leads to an erbium doping inside the Si:TiO film in adepth of ≈ ±
78 nm, corresponding to a maximumlocal density of 5 · cm − [28]. Immediately after theimplantation, the appearance of the film was changedfrom pink to grey, probably, due to implantation-induceddamage. At this point no upconverted fluorescence fromEr ions was detected. Post-implantation annealing inair at 500 ◦ C for 4 hour heals out the implantation dam-age and, at the same time, activates erbium emission.After annealing, the film restored its original color. Atthe same time, strong green upconverted fluorescence oferbium ions could be detected under the excitation withred (650 nm) and infrared (800 nm) laser light. The spec-trum of the upconverted emission is shown in Figure 6 a),in good agreement with other works on erbium in glassyenvironment [30, 31]. The film was shaped to form whispering gallerymode cavities with 5 µ m radius coupled to straightwaveguides as described above (see Figure 6 b). Theupconversion resonances of Er in glassy environmentare broad ( ≈
10 nm), therefore, several infrared cavitymodes could lead to upconverted fluorescence. Oncethe infrared excitation laser is tuned in resonancewith one of such modes, green fluorescence on therim of the WGM cavity lights up (see Figure 6 c,the excitation laser is filtered out). The spectrumof the fluorescence collected at one of the waveguideends shows its mode structure as indicated in Figure 6 d).
Conclusion
We have demonstrated a method of depositing lowloss high index Si:TiO films on different substrates suchas glass, sapphire, and YAG. For most substrates, therefractive index of the film is high enough to supportwaveguiding and also evanescently excite shallow fluo-rescent centers within the substrate material. We havealso shown a way of structuring the film to form on-chipphotonic elements such as waveguides and resonators.The low propagation loss of 5.1 db/cm results in a highQ-factor of the resonators (1.5 × at 800 nm) andunderlines the potential for CQED application in con-nection with rare-earth ion doped crystals for example.Finally, due to the increased thermal stability of thisfilm when compared to TiO , we could demonstratefurther optical functionalization of the film by dopingwith fluorescent rare-earth species (erbium). Theseresults show how the silicon doped titania film can beapplied to on-chip photonics in various ways. Acknowledgements
The authors wish to thank the CIME-EPFL teamfor their TEM investigation facilities. The authorswish to acknowledge the FEDER (Fonds Europ´een deD´eveloppement Economique et R´egional) for financingthe Nanobium project through the Interreg IVA pro-gramme. The work was also financed by ERC SQUTEC,EU-SIQS SFB TR21, and DFG KO4999/1-1. [1] J. Hulme, J. Doylend, M. Heck, J. Peters, M. Davenport,J. Bovington, L. Coldren, and J. Bowers, Opt. Express , 5861 (2015).[2] R. Kolesov, K. Xia, R. Reuter, R. St¨ohr, A. Zappe,J. Meijer, P. Hemmer, and J. Wrachtrup, Nat. Com-mun. , 1029 (2012).[3] T. Utikal, E. Eichhammer, L. Petersen, A. Renn,S. Gtzinger, and V. Sandoghdar, Nat. Commun. , 3627(2014).[4] R. Kolesov, K. Xia, R. Reuter, M. Jamali, R. St¨ohr,T. Inal, P. Siyushev, and J. Wrachtrup, Phys. Rev. Lett. , 120502 (2013). [5] M. Zhong, M. P. Hedges, R. L. Ahlefeldt, J. G.Bartholomew, S. E. Beavan, S. M. Wittig, J. J. Longdell,and M. J. Sellars, Nature , 177 (2015).[6] C. Clausen, I. Usmani, F. Bussi`eres, N. Sangouard,M. Afzelius, H. de Riedmatten, and N. Gisin, Nature , 508 (2011).[7] P. Moulton, Optics News , 9 (1982).[8] T. H. Maiman, Nature , 493 (1960).[9] T. Calmano, J. Siebenmorgen, O. Hellmig, K. Peter-mann, and G. Huber, Appl. Phys. B , 131 (2010).[10] E. Vetsch, D. Reitz, G. Sagu´e, R. Schmidt, S. Dawkins,and A. Rauschenbeutel, Phys. Rev. Lett. , 203603 (2010).[11] L. Liebermeister, F. Petersen, A. v. M¨unchow, D. Bur-chardt, J. Hermelbracht, T. Tashima, A. W. Schell,O. Benson, T. Meinhardt, A. Krueger, A. Stiebeiner,A. Rauschenbeutel, H. Weinfurter, and M. Weber, Appl.Phys. Lett. , 031101 (2014).[12] S. Marzban, J. G. Bartholomew, S. Madden, K. Vu, andM. J. Sellars, Phys. Rev. Lett. , 013601 (2015).[13] R. Quidant, J.-C. Weeber, A. Dereux, G. L´evˆeque,J. Weiner, and C. Girard, Phys. Rev. B , 081402(2004).[14] K. Abe, E. Y. M. Teraoka, T. Kita, and H. Yamada,Proc. SPIE , 79401 (2011).[15] T. Alasaarela, L. Karvonen, H. Jussila, A. S¨ayn¨atjoki,S. Mehravar, R. A. Norwood, N. Peyghambarian,K. Kieu, I. Tittonen, and H. Lipsanen, Opt. Lett. ,3980 (2013).[16] M. H¨ayrinen, M. Roussey, V. Gandhi, P. Stenberg,A. S¨ayn¨atjoki, L. Karvonen, M. Kuittinen, andS. Honkanen, J. Lightwave Technol. , 208 (2014).[17] J. D. Bradley, C. C. Evans, J. T. Choy, O. Reshef,P. B. Deotare, F. Parsy, K. C. Phillips, M. Lonˇcar, andE. Mazur, Opt. Express , 23821 (2012).[18] J. T. Choy, J. D. Bradley, P. B. Deotare, I. B. Burgess,C. C. Evans, E. Mazur, and M. Lonˇcar, Opt. Lett. ,539 (2012).[19] M. Furuhashi, M. Fujiwara, T. Ohshiro, M. Tsutsui,K. Matsubara, M. Taniguchi, S. Takeuchi, and T. Kawai,AIP Adv. , 032102 (2011). [20] P. Karasi´nski, C. Tyszkiewicz, R. Rogozi´nski, J. Jaglarz,and J. Mazur, Thin Solid Films , 5544 (2011).[21] E. Wagner, C. S. Sandu, S. Harada, C. Pellodi, M. Jobin,P. Muralt, and G. Benvenuti, ACS. Comb. Sci. , 154(2016).[22] C. Sandu, E. Wagner, S. Harada, G. Benvenuti,W. Maudez, M. Jobin, C. Pellodi, and P. Muralt, ThinSolid Films , 265 (2016).[23] S.-M. Lee, J.-H. Park, K.-S. Hong, W.-J. Cho, and D.-L.Kim, J. Vac. Sci. Technol., A , 2384 (2000).[24] N. Martin, C. Rousselot, D. Rondot, F. Palmino, andR. Mercier, Thin solid films , 113 (1997).[25] J. R. DeVore, J. Opt. Soc. Am. , 416 (1951).[26] K.-R. Choi, J.-C. Woo, Y.-H. Joo, Y.-S. Chun, and C.-I.Kim, Vacuum , 85 (2013).[27] R. Kolesov, B. Grotz, G. Balasubramanian, R. J.St¨ohr, A. A. Nicolet, P. R. Hemmer, F. Jelezko, andJ. Wrachtrup, Nat. Phys. , 470 (2009).[28] J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, Nucl.Instrum. Methods Phys. Res., Sect. B , 1818 (2010).[29] T. Kornher, K. Xia, R. Kolesov, N. Kukharchyk,R. Reuter, P. Siyushev, R. St¨ohr, M. Schreck, H.-W.Becker, B. Villa, A. D. Wieck, and J. Wrachtrup, Appl.Phys. Lett. , 053108 (2016).[30] F. Vetrone, J.-C. Boyer, J. A. Capobianco, A. Speghini,and M. Bettinelli, Appl. Phys. Lett. , 1752 (2002).[31] F. Song, G. Zhang, M. Shang, H. Tan, J. Yang, andF. Meng, Appl. Phys. Lett.79