Disordered, strongly scattering porous materials as miniature multipass gas cells
Tomas Svensson, Erik Adolfsson, Märta Lewander, Can T. Xu, Sune Svanberg
aa r X i v : . [ phy s i c s . op ti c s ] N ov Disordered, strongly scattering porous materials as miniature multipass gas cells
Tomas Svensson, Erik Adolfsson, M¨arta Lewander, Can T. Xu, and Sune Svanberg Division of Atomic Physics, Department of Physics,Lund University, Box 118, 221 00 Lund, Sweden Ceramic Materials, SWEREA IVF, 431 83 M¨olndal, Sweden
Spectroscopic gas sensing is both a commercial success and a rapidly advancing scientific field.Throughout the years, massive efforts have been directed towards improving detection limits byachieving long interaction pathlengths. Prominent examples include the use of conventional multi-pass gas cells [1–4], sophisticated high-finesse cavities [5–8], gas-filled holey fibers [9–11], integratingspheres [12–14], and diffusive reflectors [15, 16]. Despite this rich flora of approaches, there is acontinuous struggle to reduce size, gas volume, cost and alignment complexity. Here, we show thatextreme light scattering in porous materials can be used to realise miniature gas cells. Near-infraredtransmission through a 7 mm zirconia (ZrO ) sample with a 49% porosity and subwavelength porestructure (on the order of 100 nm) gives rise to an effective gas interaction pathlength above 5 meters,an enhancement corresponding to 750 passes through a conventional multipass cell. This essentiallydifferent approach to pathlength enhancement opens a new route to compact, alignment-free andlow-cost optical sensor systems. PACS numbers: 07.07.Df; 78.67.Rb; 42.25.Dd; 42.62.Fi; 33.70.-w
Disordered, strongly scattering materials are contin-uously finding new applications in photonics. Besidesconstituting the physical system for fundamental investi-gation of Anderson localization of light [17, 18] and ran-dom lasing [19, 20], strong turbidity can also be used for,e.g., light trapping in solar cells [21] and focussing be-yond the diffraction limit [22]. The use of porous andstrongly scattering media as efficient multipass gas cells,as described here, is an exciting addition to this diversefield. The main idea is to inject light into the porousmaterial and collect diffusively transmitted light at somedistance from the injection spot. Multiple scattering willforce photon pathlengths to greatly exceed the physicalsource-detector separation (Figure 1), and this randommultipass effect will result in enhanced gas absorptionimprints. As shown in previous work on spectroscopy ofgases in porous media [23], absorption related to the solidis easily distinguished from gas phase absorption due tothe great difference in spectral sharpness. Here, by uti-lizing disordered and highly porous materials based onlow-absorption and high-refractive index ceramic mate-rials such as alumina (Al O , refractive index n = 1 . , n = 2 .
14) and titania (TiO , n = 2 . s L tot FIG. 1: Illustration of porous media as gas cells (left), and ascanning electron microscope image of a utilized nanoporouszirconia with 115 nm pores (right). When light propagatesthrough a strongly scattering and porous material of thick-ness s , the average pathlength of transmitted light ( L tot ) willgreatly exceed s . The pathlength through gas ( L ) is approx-imately proportional to L tot and the porosity φ . The en-hancement of the gas interaction pathlength may be definedas N = L/s . size distributions. The titania is nanoporous (sinteredKronos 1001) and has a 42% porosity and pores in the79 ±
10 nm range (width here defined as the half widthat half max, HWHM, of the pore size distribution). Onealumina is nanoporous with 69 ± ± µ m pores and a 34% porosity(made to 92.5% from AA10, Sumitomo, the rest beinga 40 nm silica powder used as a binder). The zirconiamaterials are nanoporous and both have a 49% poros-ity, pores being 43 ± ±
15 nm, respectively (TZ3YSBE, Tosho, Japan).Optical properties (scattering and absorption) of thesematerials have been measured in the 700-1400 nmrange by employing photon time-of-flight spectroscopy(PTOFS) [26]. The measured reduced scattering coeffi-cients ( µ ′ s ) are presented in Figure 2a, revealing a strongdependence on refractive index and pore size. The scat- R e du ce d s c a tt e r i n g , µ ′ s ( c m − ) TiO ZrO Al O Al O ZrO (79 nm)(115 nm)(69 nm)(3.7 µ m)(43 nm)(a)700 800 900 1000 1100 1200 1300 14000.00.10.20.3 A b s o r p t i o n , µ a ( c m − ) Wavelength, λ (nm)(b) FIG. 2: Optical properties of the ceramic materials (measuredusing PTOFS). Reduced scattering coefficients are shown in(a), and absorption coefficients in (b). Solid lines in (a) showfitted a × ( λ/µm ) − b curves and give information on scatteringdecay, being 4045 × λ − . for the titania, 537 × λ − . and1187 × λ − . for the aluminas, and 3167 × λ − . and 480 × λ − . for the zirconias, respectively. tering of the two strongest scatterers even approach thatof the materials used in the quest for Anderson localiza-tion of light [17, 18]. At 700 nm, the transport mean freepath of photons ( l ∗ = 1 /µ ′ s ) are about 0.7 µ m and 1.0 µ m for the titania and 115 nm zirconia, respectively (theshortest photon mean free path observed, published inconnection with work on Anderson localization, is about0.2 µ m [17]). It is also interesting to note that the mate-rials with pore size smaller than the wavelength exhibita λ − Rayleigh-type scattering dependency. The 40 nmzirconia is partly an exception, indicating that scatteringdue to collective heterogeneity appears to have a strengthcomparable to that of individual pores. The macroporousalumina, on the other hand, shows that a heterogeneitylarger than the wavelength can be used to maintain highscattering over a large spectral range.The estimated absorption of the materials are shown inFigure 2b. As expected for these ceramic materials, theabsorption is low in the near-infrared range. The titania,however, exhibit a fairly strong absorption in the shortwavelength region, being a tail of strong absorption inthe ultraviolet. It should also be noted that all materi-als, except the macroporous alumina, exhibit significantabsorption at 1400 nm related to adsorbed water.The potential of the materials as spectroscopic multi-pass gas cells are demonstrated in high-resolution near-infrared (760.654 nm, vacuum wavelength of transition) -15 -10 -5 0 5 10 150.900.920.940.960.981.00 T r a n s m i tt a n ce ( - ) Relative optical frequency (GHz)
ZrO (115 nm) Γ fit = 2 .
22 GHz L fit = 541 cm FIG. 3: Oxygen (O ) absorption for light transmitted throughthe zirconia material with 115 nm pores. The baseline-corrected experimental gas spectrum (black, noisy) is showntogether with a fitted Lorentzian lineshape (red, smooth).Fitted pathlength through gas and line HWHM are stated.The acquisition time was 10 s, and the detected power wasabout 10 nW (0.003% transmission). laser absorption spectroscopy of molecular oxygen. Atatmospheric conditions, the line has a peak absorptioncoefficient of 2 . × − cm − and a linewidth of 1.6 GHzHWHM. Spectroscopic results are presented in Table 1,and the experimental gas spectrum from the most potentmaterial (the zirconia with 115 nm pores) is shown inFigure 3. The pathlength enhancement of this zirconia isimpressive, and correspond to 750 passes through a con-ventional multipass cell. It is, however, important not tolook only at the achieved pathlength enhancement. Theaverage pathlength of light transmitted through a turbidmaterial is proportional to the square of the thickness.This holds also for the pathlength through gas [27], i.e. L ∝ s , and the pathlength enhancement is therefore pro-portional to the material thickness, N ∝ s . Instead, thesuitability of a porous material as a random multipassgas cell is jointly determined by the scattering efficiency,the absorption of the solid, and the material porosity.Strong scattering and high porosity being ideal proper-ties for this purpose, while absorption is detrimental bothin terms of pathlength through gas and total transmis-sion. Although the titania material exhibits the strongestscattering, the rather strong absorption at 760 nm makesit a less successful gas cell candidate than the zirconia(at this wavelength). It is also important to realise thatlosses due to scattering and absorption sets a limit to themaximum possible thickness. Diffuse light propagationcause transmitted intensity to decrease with 1 /s , even ifthe absorption would be zero. For example, only 0.003%of the light incident on the 7.2 mm thick zirconia reachedthe detector. For comparison, the strong scattering com-bined with absorption of the titania (at 760 nm), makesit difficult to boost the pathlength through gas by in-creasing sample thickness (only 0.001% reaches throughthe 1.4 mm sample).Another interesting aspect of the results in Table 1, not TABLE 1: Results from near-infrared (760.654 nm) spectroscopy of molecular oxygen in sintered ceramic materials. The tablestates material thickness ( s ), photon transport mean free path ( l ∗ ), porosity ( φ ), pore size ( d ), detected transmission ( T ),spectroscopic linewidth (Γ), pathlength through gas ( L ), and pathlength enhancement ( N = L/s ). Sample diameters were inall cases 12-14 mm.Material s (mm) l ∗ ( µ m) φ (%) d (nm) T (%) Γ (GHz) L (cm) N = L/s (-)ZrO ±
15 0.003 2.218 ± ± ≃ O ± ± ± ≃ ±
10 0.001 2.17 ± ± ≃ ± ± ± ≃ O ±
400 0.02 1.616 ± ± ≃ yet mentioned, is the observed linewidths. The linewidthof oxygen within the macroporous alumina agrees wellwith the width expected for oxygen at atmospheric pres-sure, i.e. µ m, and by accompanying alonger wavelength with a scaling of pore size and sam-ple thickness, the scattering efficiency and pathlength en-hancement can be maintained. This means that porousceramics may be used as spectroscopic gas cells for nu-merous important gases, including methane (CH ), car-bon dioxide (CO ), carbon monoxide (CO), nitrous oxide(NO), nitrogen dioxide (NO ), ammonia (NH ) and wa-ter (H O). An issue that requires further investigation is,however, how the response time of porous gas cells variesbetween gases, as well with pore size and cell geometry.While relatively inert gases, such as oxygen, have beenfound to move rapidly in and out of nanoporous alumina[24, 28], the exchange of water vapor (known to be a”sticky” gas) is fairly slow [28].To summarize, we have shown that strongly scatter-ing porous materials can function as miniature gas cells.Important advantages of such gas cells are their smallsize, the small gas volume needed (despite strong path-length enhancement), and their alignment-free nature.The most obvious disadvantage is that strong scatteringresults in only a small fraction of incident light reachingthe detector. In the case of laser spectroscopy, anotherproblem is speckle interference noise [28, 29]. However,the topic is new and significant improvements regard- ing detection schemes and gas cell design can be antici-pated. For example, other configurations than the puretransmission geometry (as adopted here) can be explored.This includes designs where lost light is reinjected intothe material, where light is launched inside the porousgas cell, or where a reflection-type geometry is employed.Porous gas cells can also be manufactured from othercrystal forms of the ceramics or other low-absorption ma-terials. Finally, the pore structure (porosity and poresize) can be optimized to yield stronger scattering [30],allowing smaller gas cells while maintaining or increasing L . In particular, materials with pore sizes in the 100-3000nm range need to be explored. METHODS
Sintering
Titania and zirconia materials were sinteredfor 2 hours at 900 ◦ C. The nanoporous alumina was sin-tered for 10 minutes at 1000 ◦ C, and the macroporousalumina for 45 minutes at 1400 ◦ C. High-resolution spectroscopy is based on tunablediode laser absorption spectroscopy (TDLAS), and theutilized instrument is described in detail in Ref. [31]. Op-tical interference noise, an important limitation in laserspectroscopy of gases in turbid porous media, is sup-pressed by means of laser beam dithering [29]. Briefly,light from a 0.3 mW VCSEL diode laser is injected intothe pillbox-shaped ceramic samples, and diffuse trans-mission is detected by a 5 . × . large-area photo-diode (no alignment needed). The laser is tuned over theR9Q10 absorption line at 760.654 nm. This transitionhas a line strength of about 2 . × − cm Hz, which atatmospheric conditions (1 atm, 293 K, 20.9% O ) givesrise to a peak absorption coefficient of about 2 . × − cm − . The corresponding linewidth is about 1.6 GHzHWHM, originating from a 1.4 GHz pressure broadeningcombined with a 0.4 GHz Doppler broadening. Photon time-of-flight spectroscopy is conductedusing a tunable system based on a fiber laser combinedwith super-continuum generation, acousto-optical tun-able filters, and time-correlated single photon count-ing [26]. Picosecond light pulses are injected into thin,polished-down versions of the ceramic materials, and thetemporal shape of the transmitted pulses allows, via dif-fusion theory [32], assessment of absorption and reducedscattering coefficients. Since effective medium theoriesfor strongly scattering materials are not fully developed[33], the evaluation is based on a volume-averaged re-fractive index of the porous materials (using appropriateSellmeier equations for the solids). The volume-averagingapproach limits accuracy, since light in porous mediatends to interact predominantly with the solid [34].
ACKNOWLEDGEMENTS
This work was funded by the Swedish Research Coun-cil. T.S. gratefully acknowledge MSc. Erik Alerstamfor important assistance in PTOFS experiments, formercolleagues Dr. Mats Andersson and Dr. Lars Rippe for fruitful cooperation in development of the TDLASinstrumentation, as well as Dr. Dmitry Khoptyar andProf. Stefan Andersson-Engels for general collaborationon PTOFS. Karin Lindqvist at SWEREA IVF is ac-knowledged for manufacturing the alumina materials.
AUTHOR CONTRIBUTIONS
T.S. initiated the project, was the principal investi-gator, and drafted the manuscript. E.A. manufacturedzirconia and titania materials. M.L. and C.X. assistedin TDLAS measurements. S.S. was instrumental in earlyconcept developments. All authors approved the finalmanuscript version. Correspondence should be addressedto T.S. (email: [email protected]). [1] White, J. Long optical paths of large aperture.
J. Opt.Soc. Am. , 285–288 (1942).[2] Herriott, D., Kompfner, R. & Kogelnik, H. Off-axis pathsin spherical mirror interferometers. Appl. Opt. , 523–526 (1964).[3] McManus, J. B., Kebabian, P. L. & Zahniser, W. S.Astigmatic mirror multipass absorption cells for long-path-length spectroscopy. Appl. Opt. , 3336–3348(1995).[4] Chernin, S. M. & Barskaya, E. G. Optical multipassmatrix systems. Appl. Opt. , 51–58 (1991).[5] O’Keefe, A. & Deacon, D. A. G. Cavity ring-down opticalspectrometer for absorption-measurements using pulsedlaser sources. Rev. Sci. Instrum. , 2544–2551 (1988).[6] Engeln, R., Berden, G., Peeters, R. & Meijer, G. Cav-ity enhanced absorption and cavity enhanced magneticrotation spectroscopy. Rev. Sci. Instrum. , 3763–3769(1998).[7] Ye, J., Ma, L. S. & Hall, J. L. Ultrasensitive detections inatomic and molecular physics: demonstration in molecu-lar overtone spectroscopy. J. Opt. Soc. Am. B , 6–15(1998).[8] Bernhardt, B., Ozawa, A., Jacquet, P., Jacquey, M.,Kobayashi, Y., Udem, T., Holzwarth, R., Guelachvili,G., Hansch, T. W. & Picque, N. Cavity-enhanced dual-comb spectroscopy. Nat. Photon. , 55–57 (2010).[9] Ritari, T., Tuominen, J., Ludvigsen, H., Petersen, J. C.,Sorensen, T., Hansen, T. P. & Simonsen, H. R. Gassensing using air-guiding photonic bandgap fibers. Opt.Express , 4080–4087 (2004).[10] Ghosh, S., Sharping, J. E., Ouzounov, D. G. & Gaeta,A. L. Resonant optical interactions with molecules con-fined in photonic band-gap fibers. Phys. Rev. Lett. ,093902 (2005).[11] Hoo, Y. L., Liu, S. J., Ho, H. L. & Jin, W. Fast responsemicrostructured optical fiber methane sensor with multi-ple side-openings. IEEE Photonic. Tech. L. , 296–298(2010).[12] Tranchart, S., Bachir, I. & Destombes, J. Sensitive tracegas detection with near-infrared laser diodes and an in-tegrating sphere. Appl. Opt. , 7070–7074 (1996).[13] Hawe, E., Chambers, P., Fitzpatrick, C. & Lewis, E. CO monitoring and detection using an integrating sphere as amultipass absorption cell. Meas. Sci. Technol. , 3187–3194 (2007).[14] Masiyano, D., Hodgkinson, J. & Tatam, R. Gas cells fortunable diode laser absorption spectroscopy employingoptical diffusers. Part 2: Integrating spheres. Appl. Phys.B , 303–312 (2010).[15] Hodgkinson, J., Masiyano, D. & Tatam, R. Gas cells fortunable diode laser absorption spectroscopy employingoptical diffusers. Part 1: Single and dual pass cells.
Appl.Phys. B , 291–302 (2010).[16] Chen, J., Hangauer, A., Strzoda, R. & Amann, M.-C.Laser spectroscopic oxygen sensor using diffuse reflectorbased optical cell and advanced signal processing.
Appl.Phys. B , 417–425 (2010).[17] Wiersma, D. S., Bartolini, P., Lagendijk, A. & Righini,R. Localization of light in a disordered medium.
Nature , 671–673 (1997).[18] St¨orzer, M., Gross, P., Aegerter, C. M. & Maret, G. Ob-servation of the critical regime near Anderson localizationof light.
Phys. Rev. Lett. , 063904 (2006).[19] Cao, H., Zhao, Y. G., Ho, S. T., Seelig, E. W., Wang,Q. H. & Chang, R. P. H. Random laser action in semicon-ductor powder. Phys. Rev. Lett. , 2278–2281 (1999).[20] Wiersma, D. S. The physics and applications of randomlasers. Nature Physics , 359–367 (2008).[21] Garnett, E. & Yang, P. D. Light trapping in siliconnanowire solar cells. Nano Lett. , 1082–1087 (2010).[22] Vellekoop, I., Lagendijk, A. & Mosk, A. P. Exploitingdisorder for perfect focusing. Nat. Photon. , 320–322(2010).[23] Sj¨oholm, M., Somesfalean, G., Alnis, J., Andersson-Engels, S. & Svanberg, S. Analysis of gas dispersed inscattering media. Opt. Lett. , 16–18 (2001).[24] Svensson, T. & Shen, Z. Laser spectroscopy of gas con-fined in nanoporous materials. Appl. Phys. Lett. ,021107 (2010).[25] Rouquerol, J., Avnir, D., Fairbridge, C. W., Everett,D. H., Haynes, J. H., Pernicone, N., Ramsay, J. D. F.,Sing, K. S. W. & Unger, K. K. Recommendations for thecharacterization of porous solids. Pure Appl. Chem. ,1739–1758 (1994). [26] Svensson, T., Alerstam, E., Khoptyar, D., Johansson, J.,Folestad, S. & Andersson-Engels, S. Near infrared photontime-of-flight spectroscopy of turbid materials up to 1400nm. Rev. Sci. Instrum. , 063105 (2009).[27] Somesfalean, G., Sj¨oholm, M., Alnis, J., af Klinteberg,C., Andersson-Engels, S. & Svanberg, S. Concentrationmeasurement of gas embedded in scattering media by em-ploying absorption and time-resolved laser spectroscopy. Appl. Opt. , 3538–3544 (2002).[28] Svensson, T., Lewander, M. & Svanberg, S. Laser spec-troscopy of water vapor confined in nanoporous alumina:wall collision line broadening and gas diffusion dynamics. Opt. Express (2010).[29] Svensson, T., Andersson, M., Rippe, L., Johansson, J.,Folestad, S. & Andersson-Engels, S. High sensitivity gasspectroscopy of porous, highly scattering solids.
Opt.Lett. , 80–82 (2008).[30] Rivas, J. G., Lagendijk, A, Tjerkstra, R. W., Vanmaekel-berg, D & Kelly, J. J. Tunable photonic strength in porous GaP. Appl. Phys. Lett. , 4498–4500 (2002).[31] Svensson, T., Andersson, M., Rippe, L., Svanberg,S., Andersson-Engels, S., Johansson, J. & Folestad, S.VCSEL-based oxygen spectroscopy for structural analy-sis of pharmaceutical solids. Appl. Phys. B , 345–354(2008).[32] Contini, D., Martelli, F. & Zaccanti, G. Photon migra-tion through a turbid slab described by a model basedon diffusion approximation: I. Theory. Appl. Opt. ,4587–4599 (1997).[33] Schuurmans, F. J. P., Vanmaekelbergh, D., van de Lage-maat, J. & Lagendijk, A. Strongly photonic macrop-orous gallium phosphide networks. Science , 141–143(1999).[34] Svensson, T., Alerstam, E., Johansson, J. & Andersson-Engels, S. Optical porosimetry and investigations of theporosity experienced by light interacting with porous me-dia.
Opt. Lett.35