---
title: "Blue surface plasmon propagation along thin gold film - gas interface and its use for sensitive nitrogen dioxide detection"
authors: ["Elena V. Alieva", "Valery N. Konopsky", "Dmitry V. Basmanov", "Sergey K. Sekatskii", "Giovanni Dietler"]
journal: "Optics Communications"
year: 2013
volume: "309"
pages: "148-153"
doi: "10.1016/j.optcom.2013.05.058"
language: en
source: "anydoc from OpticsCommunications2013.pdf"
---

### This article appeared in a journal published by Elsevier. The attached

**copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution**

### and sharing with colleagues.

**Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party**

### websites are prohibited.

### In most cases authors are permitted to post their version of the

**article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information**

### regarding Elsevier’s archiving and manuscript policies are encouraged to visit:

**[http://www.elsevier.com/authorsrights**](http://www.elsevier.com/authorsrights**)

# Author's personal copy

Optics Communications 309 (2013) 148 –

Contents lists available at ScienceDirect

# Optics Communications

journal homepage: www.elsevier.com/locate/optcom

Blue surface plasmon propagation along thin gold fi lm – gas interface and its use for sensitive nitrogen dioxide detection

a a a, c b n, b

E.V. Alieva, V.N. Konopsky, D.V. Basmanov, S.K. Sekatskii, G. Dietler aInstitute of Spectroscopy, Russian Academy of Sciences, Fizicheskaya street 5, Troitsk, Moscow 142190, Russia bLaboratoire de Physique de la Matière Vivante, IPSB, BSP, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne-Dorigny, Switzerland cRussian Institute of Physical-Chemical Medicine, Malaya Pirogovskaya 1a, 119435 Moscow, Russia article info abstract Article history: We demonstrate effective surface plasmon propagation along thin gold fi lm in a blue spectral range: the Received 24 April 2013Received in revised form27 May 2013Accepted 29 May 2013 measured 0.6 1 angular width of the plasmon excitation peak corresponds to the propagation length
Available online 19 June 2013 around 7 μ m. For this end, we prepared a specially designed photonic crystal (PC; 18-layer 1D structure Keywords: of altering transparent dielectric layers with relatively large and low refraction indices deposited onto a Ultralong surface plasmon propagationAu thin fi lm silica support) coated by a 8 nm thick gold layer. For given blue light wavelength (405 nm) and certain Surface plasmon resonanceSensitive nitrogen dioxide detection light incidence angle this 1D PC structure supports ultralong surface plasmon propagation along the goldnano fi lm. This structure has been tested as a surface plasmon resonance sensor to detect small concentrations of nitrogen dioxide in air. This sensor exhibited the sensitivity similar to earlier works where surface plasmon resonance in red spectral range has been used for this purpose. & 2013 Elsevier B.V. All rights reserved.

The situation changes drastically if long range propagating

1. Introduction plasmons [6 – 8] are considered. These plasmons, existing for
symmetrical con fi gurations where a thin ( o 10 – 15 nm) metal Surface plasmon resonance (SPR) sensors nowadays is an layer is sandwiched between two dielectric transparent materials indispensable and broadly accepted research tool enabling detec-possessing the same refraction index at the wavelength used, are tion of different chemical and biochemical substances (species) characterized by much longer propagation lengths and thus can be with high sensitivity and precision. They can provide detailed excited for essentially broader range of materials and wavelengths. kinetic constants of interprotein and other practically important However, the requirement of the symmetry of the structure is interactions [1 – 3]. Some types of such sensors are commercially crucial, and this poses very serious experimental dif fi culties and available. However, still it should be noted that despite a very limitations: it is not easy to organize a symmetrical sandwich broad range of SPR sensor designs and applications, these same structure already when water (refraction index n around 1.33) is sensors almost exclusively exploit gold (or more rarely silver) as a involved and de fi nitely, apart from free standing ultrathin fi lm, no surface plasmon propagation supporting material, and red light(with the wavelength larger than ca. 600 nm) as a spectral range other simple symmetrical structure apparently can be prepared to where the surface plasmons are excited. Surely, there are just work in air or other gas when the refraction index is close to unity. reasons for this: amongst all existing materials, it is exactly this For the latter case, recently it has been proposed [9] and then combination of support and excitation wavelength which offers experimentally realized [10 – 12] to exploit specially designed 1D the best results excluding alkali metals with their too evident multilayer structures (also referred to as photonic crystals, PC) technical challenges (see e.g. [4, 5] devoted specially to comparison composed by alternating transparent dielectric layers of certain of the “ plasmon quality ” of different materials). Considering the thicknesses and refraction indices coated with thin metal layer. usual setting of the corresponding experiment, viz. thin gold fi lm The thicknesses of the dielectric layers are found from dispersion deposited onto some appropriate dielectric support, there is relation of 1D PC, as described in Ref. [27]. In a sense these PC simply no point to speak about surface plasmons at all if blue thicknesses is adjusted to “ equalize ” the EM fi eld distribution light (around 400 nm) is used: the propagation distance for such above and below the gold nano fi lm. In this case, zero fi eld “ plasmons ” is just of the order of the light wavelength. intensity occurs inside the nano fi lm for a certain wavelength and

an incidence angle of the light. The gas to be analyzed is located from the other side of thin metal layer; thus now we get the nCorresponding author. Tel.: + 41 21 693 0445; fax: + 41 21 693 0422.ultralong propagating surface plasmons not for a free standing and E-mail address: Serguei.Sekatski@ep fl.ch (S.K. Sekatskii). very fragile thin metal fi lm, but for such a fi lm safely supported by

0030-4018/$ - see front matter & 2013 Elsevier B.V. All rights reserved. [http://dx.doi.org/10.1016/j.optcom.2013.05.058](http://dx.doi.org/10.1016/j.optcom.2013.05.058)

E.V. Alieva et al. / Optics Communications 309 (2013) 148 –
a solid substratum. Earlier, this approach, while still using the red light, had been applied to realize the plasmon propagation for a poor-performing plasmon material such as palladium. Corre- sponding structures (photonic crystal + thin Pd layer) were used for ultrasensitive measurements of hydrogen concentration in a gas phase [10]. In the present note we report an observation of ultralong surface plasmon propagation in blue spectral region, viz. at 405 nm wavelength, rather than in red spectral region as this took place before. This observation constitutes the main result to be reported. Still, evidently, this would be unreasonable not to try to use the structure at hand for an analysis and detection of certain appropriate substances. For this end, we have selected nitrogen dioxide gas based on the earlier obtained by other authors, when the SPR technique has been used, promising results [13 – 15] which, however, remain somewhat controversial, and this despite the 2chemi- and physisorption circumstance that the question of NO on single crystal and polycrystalline gold has been the subject of rather intensive studies [16 – 23]. Besides, we were motivated by the practical importance of the problem of detection of nitrogen xgases dioxide as well as, more generally speaking, different NO and NO2- group containing gases. Nowadays it is well established 2and other NOxgases causes that even short term exposure to NO adverse respiratory effects including airway in fl ammation in healthy people and increases respiratory problem in asthma patients, while long-term exposure is correlated to pulmonary edema and death. Besides, these gases are considered as a major cause of photochemical smog, acid rains and depletion of the

Fig. 1. Schematic of an experimental setup.

protective stratospheric ozone layer, see e.g. [24] and references 2in the cited therein. As small concentrations as 0.5 – 5 ppm of NO interior are authorized by competent governmental and intergo-gaseous reaction products were mixed with pure nitrogen in a vernmental regulatory bodies [25]. Results and perspectives of desirable concentration using a custom-built simple gas mixture/ these experiments are also brie fl y discussed. inlet/outlet system. The total pressure in the system was nearlyequal to the atmospheric one and gas fl ow rate through the cuvette was equal to 70 ml/min.

2. Experimental 2.1. Long range propagating surface plasmons at 405 nm
The schematic of our experimental setup is presented in Fig. 1. The photonic crystal was: silica substrate/H ( LH) 8 /H ′ M/gas, This is essentially the same setup which has been used by us where2O₅ layer with the thickness H is a TaH¼ 57.8 nm and the d earlier for the detection of small hydrogen concentrations in a gas refraction index at the wavelength usedH¼ 2.2, L is a SiO2layer n phase using Pd nano fi lms [10] and (especially) Pd nanoparticle with the thicknessL¼ 84.4 nm and the refraction d index at the layers [11]; more details can be found in these papers. Very brie fl y, L¼ 1.47, H ′ is a wavelength tantalum pentoxide used layer with n p-polarized laser radiation is focused by a cylindrical lens via a the thickness of 55 nm and M is a 8 nm-thick gold layer. All layers silica prism (a droplet of phase matching oil is introduced between were prepared by magnetron sputtering (both tantalum pentoxide the contacting surfaces of the prism and structure) onto external and silicone oxide are standard materials widely used with such surface of the structure used (photonic crystal + thin gold layer) in technique) in one coating run. The calculated, similarly to Ref. [27], such conditions that surface plasmons are excited in the Kretsch- dispersion curve for this structure is presented in Fig. 2. The mann con fi guration [26].Re fl ected light is directed without any dielectric constant of the gold taken from the Palik book [28], i.e. additional optics onto the entrance screen of a CMOS camera ε ¼ − 1.1 + i 6.5 for 405 nm, was used during the calculations. placed 130 mm away from the prism. The changes of the light As one can see, this structure should support ultralong surface intensity distribution onto the detector caused by the changes of 2sorption by a thin gold plasmon propagation, and an electromagnetic fi eld intensity the plasmon excitation conditions (e.g. NO layer for the case at question) are measured and processed by a enhancement (logarithm of the ratio of the intensity of propagat- custom-written software; this procedure nowadays is often namedeto that of an incoming ing in the external medium (gas) light I

0) equal to 45 is anticipated for the propagation parameter light I “ an angular interrogation ”.
icosi¼ϑ 1 : 0066. (Hereiis a refraction index of some layer or n ρ ≡ n The main difference with respect to our aforementioned earlier iis an angle between the direction of an external medium and ϑ works, besides the use of a blue diode laser at 405 nm (Qphotonics the propagation of light inside this layer/medium and normal to LLC, Ann Arbor, MI, USA; 1 – 10 mW cw laser power, spectral line the interlayer interface. The Snell refraction law guaranties the width about 2 nm, fi ber-coupled version is used to improve the constancy of this parameter throughout the whole structure thusmaking it a very useful apparatus to describe the light propaga-spatial quality of the beam), is the addition of a simple custom-made reaction chamber where the necessary quantities of NO 2gas tion; see [27] for further details.) Calculated width of resonance were prepared exploiting the reaction Cu3¼ Cu(NO3)2+ + 4HNO dip for the presented structure is 0.4 1, and, correspondingly, the 2H2O + NO2↑. A few milligrams of pure copper in small pieces plasmon propagation distance (such a distance for which the (shavings) fi rst react with concentrated nitric acid (taken with a intensity of the plasmonic fi eld becomes e -times smaller) equal large excess) inside a hermetically closed air-fi lled chamber with to m m may be anticipated. For comparison, let us note that the volume around 1 l, and then, after the reaction is fi nished, the

log₁₀( I/ I)e

0.35
1.4
1.2 1
0.8
m 0.4

0.6
λ **=405 nm**, μ 0.4 λ

0.2 0
0.45 ρ **=1.0066** −0.2
−0.4 −0.6

0.95 1 1.05 1.1 1.15 1.2
ρ =nosin(o) θ

Fig. 4. Sensogram showing the reaction of the sensor on the introduction of

Fig. 2. The calculated dispersion diagram of the 1D photonic crystal structure used

2gas with the concentrations of nitrogen dioxide equal to 35 ppmnitrogen:NO 10ð eI =0IÞ to support blue surface plasmon propagation. Optical enhancement log is (arrows pointing up), and pure nitrogen (arrows pointing down). presented in color scale (index e is used for an external medium (air) and index 0 for an incidence medium (glass prism)). See text and Ref. [27] for further details.

2.2. Detection of nitrogen dioxide
Typical experimental results (sensogram) attesting the changes of the light intensity distribution on the CMOS matrix, expressed as changes of the surface plasmon excitation peak, and resulting from introduction of 35 ppm nitrogen dioxide into the sensor, are 2introduction is very well presented in Fig. 4. The effect of NO observable — the propagation parameter ρ rises very fast (3 – 5s) 2introduction and continues to rise slowly till pure N₂ after NO introduction. At the period from 1700 s to 2300 s NO₂ was fl owing throw the sensor and it is seen that the propagation parameter 2inlet continues to rise slowly all this time. The interruption of NO and switch to the pure N2inlet leads to the signal decrease with a characteristic time scale around 20 s. Based on the data pertinent for the smallest nitrogen dioxide concentration tested in our experiments, we estimate the attained limit of the reliable detection of nitrogen dioxide as ca. 3 – 4 ppm. Control experiments performed using the same photonic crystal on the same silica support but without any gold fi lm deposited on top of it, do not reveal any noticeable signal in the concentration range up to 500 ppm. Aforementioned results, including the detection limit, are

Fig. 3. Surface plasmon excitation peak. rather close to those presented in Ashwell and Roberts paper

[13] as well as in Nikitin et al. papers [14, 15] where SPR on without the photonic crystal, i.e. for a similar gold layer deposited 40 – 50 nm-thick gold fi lms deposited onto glass or silicon surface directly onto silica substratum, as small propagation length as and red light (633, 647 or 670 nm) were used to detect nitrogen

0.7 m m is attained: the value where there is no point to speak dioxide. In particular, similarly to [15], we also observed fast and
2introduc-about surface plasmons at all. much more slow components of gold response to NO tion. However, we have already mentioned important differences Of course, the above values are just theoretical ones calculated in observations reported in [13 – 15]. While an excellent reversi-supposing an ideal structure where all layer thicknesses and bility of the process of absorption/desorption cycles (1500 cycles at refraction indices are equal exactly to those calculated, and no 100 ppm showed only 10% signal deterioration) was observed in one loss mechanism apart from that related with the imaginary [13], quite a long time necessary to approach a saturation in part of the metal dielectric constant is present. In practice, only nitrogen dioxide absorption (1 – 1.5 h) and very long recovery time smaller, and probably even essentially smaller, values can be (6 – 8 h) were attested by Nikitin et al. [14, 15]. Our experimental realized due to technical (deviation of the structure properties data side with those of Nikitin and co-authors, and aforemen-from optimal, scattering at interlayer interfaces and at defects in tioned signal rise and decrease time components are to be under-the bulk, etc.) and fundamental ( fi rst of all scattering of light on stood as fast components superimposed onto much more slow the boundaries of thin metal layer, sf. [27]) limitations, and it is saturation/recovery dynamics; see Fig. 4. very dif fi cult to say in advance how important all these factors are. The following interpretation of the experimental results can be Our experiments revealed that the structure at hand does suggested following the reasoning of Nikitin et al. [14, 15] and support ultralong surface plasmon propagation, and the angular 2adsorption on the gold based also on newer data pertinent to NO width of the plasmon excitation peak about 0.6 1, thus correspond- surface [19 – 23]. It is long known that quite ef fi cient chemisorption ing to the plasmon propagation distance as long as 7 m m, has been of the nitrogen dioxide onto Au surface as a gold O – O ′ nitrito-observed experimentally; see Fig. 3. The experimental propagation chelate at room temperatures (saturation coverage ≅ : 4ML) is length of surface wave is somewhat less than the theoretical one followed, provided the exposure is large enough, by a relatively due to plasmon scattering by the fi lm roughness.

slow but still rather ef fi cient process of the formation of suf fi-paper [31] and references therein. In any case these questions O₄ multilayer [16, 17]. Clearly, exactly these features ciently thick N deserve a further study which is currently planned. lie beyond the large sensitivity of Au-fi lm based SPR detection of (and other NO₂ group-containing gases, cf. [29]) as opposed to NO a quite low sensitivity for the detection of NH3,H2, CO, CO₂,SO2, 3. Conclusions HCl, H₂S and Cl₂ [13] as well as dichloromethane and ethanol [14] in the same sensor. The results presented here unambiguously attest an experi- As for the discrepancies in the results, recent investigations mental realization of effective surface plasmon propagation in blue 2-[18 – 22] point out to the importance of the water content in a NO spectral range. For us, it seems evident that as a result of this containing medium to be analyzed for the full understanding of realization, broad new perspectives related with the now possible the process dynamics (although already the authors of [14, 15] effective excitation of fl uorescence of numerous fl uorophores and reported an order of magnitude decrease of the sensor recovery plasmon-assisted chemical reactions (surely, blue light is muchmore fi t for these purposes than a red one) might appear in the 2air is humid enough). As it is time if analyzing containing NO nearest future, but this would be premature to discuss them here boldly stated in2and H[18]2O “ even though the adsorption of NO at length. are completely reversible when adsorbed separately on Au(111), Application of the corresponding SPR detector for detection of coadsorption leads to reaction ”. Detailed characterization of sur-NO 2traces in nitrogen con fi rmed theface chemical reactions involved is complicated, and not all details previous results obtained when using SPR devices in red spectral region and highlighted an are fully understood yet. It is generally believed that amorphous 2O₄ layers undergo subsequent transformation to importance of the careful control of experimental conditions, physisorbed N fi rst + − 2(ONO2O₄–) and nitrosonium nitrate NO N nitrite – NO3– NO with of all the water content and temperature in analyzing substances, subsequent partly reversible decomposition into NO and NO2(may for practical reliable detection of nitrogen dioxide. Detection limit, be with a possible partial conversion to N2O as well), and, in some attained with our approach, is roughly similar to that reported in circumstances, also cause the formation of adsorbed oxygen atoms the corresponding experiments earlier [13 – 15], which is not and desorbed oxygen gas [19, 22]. Note also a series of the papers surprising given the not so big difference in plasmon propagation 2O₄ where strong dependence of the order and composition of N constants for our case of blue light (where plasmon propagation is layers onto different substrata on NO₂ deposition conditions was supported by a specially designed photonic crystal) and that of the reported [30]. red light and not so thin gold fi lms simply deposited onto non- The reactions we are speaking about are to be clearly distin-structured dielectric support. 2and possible guished from chemical reactions between NO What concerns further perspectives of the use of our approach impurities in gold. The latter would lead to the irreversibility of for the problem of detection of nitrogen dioxide, we would like to the studied interaction which was not observed in all earlier works underline the following aspect. Recently, in experiments with the as well as in the present study and thus should be excluded. At the SPR detected using the cavity ring-down technique, a 35-fold same time the aforementioned chemical reactions, being verydependent on the analyte composition and other experimental increase in the sensitivity limit of nitrogen dioxide was reported conditions, are the reactions within the physically adsorbed layer when using gold nanoparticles with the average diameter of which do not change the very nature of physisorption. Corre-4.5 nm as compared with Au fi lms [29] (the authors insist that spondingly, the whole process remains reversible but, in many this increase is really a nanoparticles-based effect and is not due to circumstances, reversible on rather long time scale as this has the exploitation of the cavity ring-down technique). The detection been seen in experiments. limit of 29 ppb reported in that paper transforms the application We also agree with Nikitin et al. suggestion [15] that the nature of similar SPR-based approaches into the realm of the practical use 2-of NO [24, 25], and the same is valid for numerous other different NO 2as electroactive gas, in other words its capability to group containing gases, among which 2,4-dinitrotoluene (DNT), in fl uence the electronic structure and resistivity of the gold layer being the major impurity and decomposition product of 1,3,5- (or its most external part) upon adsorption, is also rather impor-trinitrotoluene (TNT) and characterized by larger than TNT vapor tant for the understanding of the experimental results; clearly, pressures [29], is, by evident reasons, probably the most important surface plasmon excitation conditions are very sensitive to such example. And, in our opinion, exactly here our approach mighthave important advantages: structures based on specially changes due to the very nature of this type of electromagnetic designed photonic crystals well support ultralong plasmon (here waves. (In relation with this, it looks instructive to note that in our better to say surface electromagnetic waves) propagation along recent surface plasmon experiment [11] even the signs of the the surfaces coated with (mono)layers of metal nanoparticles; see effect observed for 2- and 6-nm-diameter Pd nanoparticles were e.g. our recent work with Pd nanoparticles [11]. Moreover, besides different due to the “ too effective ” modi fi cation of the electronic larger sensitivity, better reproducibility and faster reaction/recov- structure of palladium upon hydrogen uptake.) In our case, where ery time with the eventual disappearance of very slow saturation the gold fi lm with the thickness of only 8 nm is used, this effect dynamics are to be anticipated for such devices. should be even more prominent than in experiments [13 – 15] ful fi lled with 40 – 50 nm-thick gold fi lms. From the practical point of view, the most important circum-Acknowledgments stance is that apparently either quite good reversibility of the The research has been partly supported by the Swiss National adsorption process or, other way around, long signal accumulation Science Foundation, Grant no. 200021-137711. and recovery times can be purposefully realized by changing and careful control of the detection conditions, fi rst of all the water content in an analyzed gas and the temperature. We also would References like to note that the rather long adsorption saturation time, observed by us as well as by Nikitin et al. [14, 15], is not necessarily a disadvantage. Quite the opposite, in some circumstances it even [1] J. Homola, Chemical Reviews 108 (2008) 462. can be an advantage — viz., when the so called accumulating [2] J. Homola, Surface Plasmon Resonance Based Sensors, Springer, Berlin, 2006. [3] X. Guo, Journal of Biophotonics 5 (2012) 483.sensing is required as this often takes place for different nitrogen [4] M.D. Arnold, M.G. Blaber, Optics Express 17 (2009) 3835. oxides sensing in the car industry and elsewhere; see e.g. recent

[5] M.G. Blaber, M.D. Arnold, M.J. Ford, Journal of Physics: Condensed Matter 22 (2010) 143201. [6] D. Sarid, Physical Review Letters 47 (1981) 1927 [7]

. [29] A.C.R. Pipino, V. Silin, Chemical Physics Letters 404 (2005) 361 [30]