Sensors and Actuators B 161 (2012) 235 243

Contents lists available at SciVerse ScienceDirect

Sensors and Actuators B: Chemical

journa l h o mepage: www.elsev ier .co

All pho m change

Hang QuGenie physique 37 3A

a r t i c l

Article history:Received 21 MReceived in reAccepted 12 OAvailable onlin

Keywords:Photonic crystBragg ber senFiber-optic sen

onstrx-conber co

inter per

of thhose bundomat, we dhold tted l

1. Introduction

Recently, liquid-core waveguides have drawn much attentionfor chemicacore wavegreection (Ttive index (suitable clathose of aqsor developthe leaky high-RI cladcore capillahowever, hloss, and aRIs.

Using se(PCFs) consPCF sensorswithin a ceber core. Tabsorption sors, HC-PCvirtually co

CorresponE-mail add

length, which makes HC-PCF sensors potentially more advanta-geous in detecting imaginary part of the analyte RI. Besides, HC-PCFsensors could be also used to detect real part of the analyte RI, as

0925-4005/$ doi:10.1016/j.l and biological sensing applications. A typical liquid-uide sensor operates on the principle of total internalIR) where liquid core is surrounded by the lower refrac-RI) cladding [1]. Unfortunately, the difculty in ndingdding material with the refractive index lower thanueous solutions (n 1.33) limits TIR liquid-core sen-ment. One way to circumvent this problem is to usemodes guided in the low-RI core surrounded by theding. This is an approach used by the liquid- and air-ry sensors reported in [2,3]. These capillary sensors,ave limited sensing length due to large propagationre mainly used to detect imaginary parts of analyte

nsors based on hollow-core (HC) photonic crystal berstitutes an alternative to capillary sensors. Most of HC-

guide by bandgap effect according to which the lightrtain frequency (reector bandgap) is conned in thehe propagation loss is determined by the core materialand ber radiation losses. Compared to capillary sen-F sensor enables longer sensing length, while havingmplete light-analyte overlap throughout the entire ber

ding author.ress: maksim.skorobogatiy@polymtl.ca (M. Skorobogatiy).

well as other measurands that could affect resonant conditions ofthe HC-PCFs. Up to date, HC-PCF based sensors have already beenused for sensing of real and imaginary parts of RIs of liquid or gasanalytes [420], as well as sensing of temperature, strain, pres-sure [2123], etc. These HC-PCF based sensors are summarized inTable 1.

Particularly, as to sensing of liquid analytes, most of the pre-viously reported HC-PCF sensors operate on the following threecongurations.

Firstly, as to detection of real parts of analyte RIs, most ofliquid-core PCF sensors use a resonant sensing mechanism accord-ing to which the transmission spectrum of the liquid-core PCFshifts in response to variations in RI of the liquid analyte llingthe ber core. For example, in [4,5] the liquid RI sensor based ona commercial glass PCF is reported with the sensitivity found tobe on order of 5 103 nm/RIU. A relative long time is neededto ll a liquid analyte into the micron-sized holes of the PCF(10 min/20 cm reported in [4]). Another example of the sensoruses the HC high-RI-contrast Bragg ber [6] which shows a tun-able bandgap when changing RI of the liquid analyte lling theber core. However, the high-RI-contrast Bragg ber sensor ismainly used to detect high-RI analytes (nanalyte > 1.40 in [6]), while itbecomes somewhat problematic when measuring low-RI analytesincluding aqueous solutions (nanalyte 1.33) as we will explain inSection 2.

see front matter 2011 Elsevier B.V. All rights reserved.snb.2011.10.025tonic bandgap ber spectroscopic systes in aqueous analytes

, Bora Ung, Mathieu Roze, Maksim Skorobogatiy

, Ecole Polytechnique de Montreal, C.P.6079, Succ, Centre-ville, Montreal, QC, Canada H

e i n f o

ay 2011vised form 4 October 2011ctober 2011e 3 November 2011

al bersorsor

a b s t r a c t

We propose and experimentally demon a hollow-core low-refractive-indelytes lling the hollow core of Bragg thus leading to both spectral shifts anddemonstration, we characterize sensotrations. The experimental sensitivityindex unit), which is comparable to tstrate the integration of a Bragg ber to replace the grating-based monochrincreased sensing speed. In summarythat uses a hollow core Bragg ber to spectral interpretation of the transmim/locate /snb

for detection of refractive index

7

ate a sensor of refractive index (RI) of aqueous analytes basedtrast Bragg ber. Variations in refractive index of liquid ana-uld modify the resonant condition of the modal connement,

nsity changes in the ber transmission. As a proof-of-principleformance using a set of NaCl solutions with different concen-e Bragg ber sensor is found to be 1400 nm/RIU (refractiveof surface plasmon resonance sensors. Moreover, we demon-le spectrometer in the liquid-core Bragg ber sensing systemor, which could potentially lead to signicant cost-saving andemonstrate an all-photonic-bandgap-ber RI sensing systemand probe the analyte, and a solid-core Bragg ber bundle foright.

2011 Elsevier B.V. All rights reserved.

236 H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243

Table 1HC-PCF based sensors.

HC-PCF based sensor Sensing mechanism Wavelength Sensitivity Ref.

Liquid analyte sensingLiquid RI se 600

400Bio-sensing 500

500Chemical se 500

600500600

Gaseous ana 71410001000155016301200500865

Temperatur 150012001500

Strain/press 150012001500

Note: The be ber hollow core su magin

Secondlywaveguide ciple [7,8]. cladding forrather thananalytes arethe cores is limited bcent eld inlong (severinto the mi

Thirdly, involves usipartially lltion techniqa PCF, whicthe increasethe air-lletherefore, gRaman signdetection liof 1010 M lling step sensor.

The goacontrast Brreal part ofing principlling the Btral shifts inof sensor issensitivity odent on thean 80 cm lopared to thetime 10 mshow good high-RI-conbandgaps t

becast Br

proping ahis pundl. The

highm, w-corially us le

Besition

of thctronsing Resonant sensing (Re(na) sensing)

Evanescent eld sensing (Im(na) sensing)

nsing Raman scattering sensing

lyte sensing Gas specic absorption sensing (Im(na) sensing)

Raman scattering sensing

e sensing Resonant sensing: bandgap shift detection

ure sensing Resonant sensing: bandgap shift detection

r used in Refs. [6,13,14] is the HC high-RI-contrast Bragg ber. In other papers, therrounded by periodical microscopic holes. Re(na) and Im(na) are the real part and i

, the liquid-core PCFs can be used as multi-coresensors operating on the evanescent-eld sensing prin-While each node of the struts in the microstructuredms a core, the PCF is used as a multi-core waveguide

a photonic bandgap (PBG) ber. Fluorophore-labeled then probed by the evanescent eld penetrating frominto the liquid-lled holes. The sensitivity of the sensory the small overlap between analytes and the evanes-

the micron-sized holes. Reported response time is alsoal minutes) due to the slow process of lling analytescron-sized holes.another implementation of the liquid-core PCF sensorng normal or surface-enhanced Raman scattering in theed HC-PCFs [912]. In such a sensor, a selective inltra-ue is used to ll liquid analytes only in the center hole ofh will enhance the light connement of the ber due to

water contramodesof sens

In tber bsystemshownspectrua liquidpotentters, thspeed.resolulengthdle sped RI-contrast between the center liquid-lled core andd microstructured cladding. The selectively lled PCF,uides by a modied index-guiding mechanism, and theal from liquid analytes in the ber core is detected. Themit of the PCF based Raman sensor could be on order[12]. Main challenge of this technique is in the selectivewhich considerably complicates fabrication of the PCF

l of this paper is to demonstrate a liquid-core low-RI-agg ber sensor (Fig. 1) which is used for detection of

liquid RIs. The sensor operates on a resonant sens-le in which variations in real part of the analyte RIragg ber core lead to both intensity changes and spec-

the ber transmission [24]. Experimental sensitivity found to be 1400 nm/RIU. We also show that thef the liquid-core Bragg ber sensor is virtually indepen-

length of the ber. The response time of the sensor usingng ber is 1 s, which represents an advantage com-

previously reported liquid-core PCF sensors (responsein). Besides, we note that low-RI-contrast Bragg bersguidance when lled with aqueous solutions. However,trast Bragg bers have the TM (transverse magnetic)hat tend to collapse in the vicinity of the light line of

length. Thesensor couber bundlour knowleber-optic cell and thePBG bers.

2. Operati

The all-psor is fabricair core (dimethyl met1.487/1.581core of thewhich mighalso leads tresistance iincreases. Elong ber is

The guidber can b1700 nm 5000 nm/RIU [4,5]900 nm 300 nm/RIU [6]600 nm - [7]760 nm - [8]4000 cm1 1.7107M (molarity) [9]2000 cms1 104105M [10]2000 cm1 103M [11]1800 cm1 1010M [12]

m 30 ppb (parts per billion) [13]7000 cm1 - [14]1700 nm - [15]1640 nm 647 ppm (parts per million) [16]1680 nm 10 ppm [17]1600 nm 49 ppm [18]600 nm - [19]900 nm - [20]1750 nm 29 nm/C [21]1600 nm 3.97 pm/C [22]1660 nm 0.86 pm/C [23]1750 nm 0.7 pm/| [21]1600 nm 0.81 pm/ [22]1660 nm 0.6 pm/ [23]

used for sensing is the HC photonic bandgap ber featuring a centerary part of the analyte RI, respectively.

use of the Brewster angle phenomenon. Thus, high-RI-agg bers have relatively poor guidance of the hybridagating in an aqueous core, which cause the difcultyqueous analytes.aper, we also demonstrate the integration of a Bragge spectrometer in the liquid-core Bragg ber sensing

Bragg ber bundle spectrometer, reported in [25] has accuracy in detecting the center peak position of a testhich can be used to analyze transmission peak shifts of

e Bragg ber. Using the Bragg ber bundle spectrometerallows forgoing traditional grating-based spectrome-ading to signicant cost-saving and increased sensingdes, unlike the grating-based spectrometer with thedependent on spatial line density of the grating ande optical path inside the spectrometer, the ber bun-

meter has a resolution independent on the ber bundle

refore, a highly compact all-PBG-ber based liquid RIld be constructed by using a several-centimeter-longe connected to a coiled HC Bragg ber. To the best ofdge, this is the rst time when a complete liquid-core RIsensor has been demonstrated where both the sensing

spectrometer are realized using hollow- and solid-core

on principle of liquid-core Bragg ber sensor

olymer HC low-RI-contrast Bragg ber used in the sen-ated in our group [26]. This Bragg ber features a largeameter: 0.8 mm) surrounded by an alternating poly-hacrylate (PMMA)/polystyrene (PS) Bragg reector (RI:

@ 589 nm) and a PMMA cladding (Fig. 1(b)). The large Bragg ber facilitates lling it with aqueous solutionst even contain some large objects. Such a large coreo a short response time of the sensor, since the own the core decreases polynomially as the core radiusxperimentally, response time of a sensor using an 80 cm

found to be 1 s.ing and sensing properties of the low-RI-contrast Bragge elucidated from the Bragg reector band diagram

H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243 237

Fig. 1. (a) Setup of the Bragg ber sensing system. An 80 cm long liquid-core Bragg ber, coiled into a 15 cm diameter circle, is integrated in the setup by using twoopto-uidic coupling blocks. The beam from a supercontinuum source is coupled into the liquid-core Bragg ber, and the transmission spectrum of the liquid-core Braggber is then analyzed by a traditional grating monochromator. (b) Cross section of a HC Bragg ber; the inset is the graph of the Bragg reector taken by a scanning electronmicroscope (SEM). (c) Inner structure of the opto-uidic block. A tip of a liquid-core Bragg ber is sealed in the horizontal channel of the block lled with the liquid analyte.The extremity of the horizontal channel is sealed by a glass window through which the light is coupled into (or out of) the sensing system. In each block there is also a verticalchannel that connects to the horizontal channel to constitute the uidic path for uidic coupling of the Bragg ber. The colorful appearance of the Bragg ber is due to thereection of ambient lights from the Bragg reector. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

(frequency versus propagation constant). In Fig. 2, we show theband diagram of the TE (transverse electric) and TM polarizedmodes propagating in an innite planar Bragg reector made fromPMMA/PS multilayer. To make the band diagram, we measure therefractive indices of PMMA and PS with a VASE Ellipsometer (J.A.Woollam Coand PS layefrom the SEcalized oveirradiated oin the multphase spacBragg reecliquid analywill have efof water (bin the liquidlight lines oregions in F

From the basic theory [27] of low-RI-contrast Bragg bers, thecenter wavelength, c, of the fundament reector bandgap can beapproximately calculated by:

c = dh(n2h n2)1/2 + d (n2 n2)1/2 (1)

dl, dtivelyis reanaly) of nt win se

noteor liqrparr sens wi

high

Fig. 2. Band dBragg reectoline of distilleinterpretation., Inc.). The average thicknesses of the individual PMMArs are 0.37 m and 0.13 m, respectively, estimatedM graphs. Grey regions in Fig. 2 indicate states delo-r the whole Bragg reector. Such states are efcientlyut of the ber due to scattering on the imperfectionsilayer. Clear regions (bandgaps) dene the parts of thee where light is forbidden to propagate inside of thetor. The black thick curves represent the light line of thete, i.e. distilled water. Modes guided in the hollow corefective RIs close to, while somewhat smaller than thatlack thick curves in Fig. 2). Therefore, a mode conned

core will exist in the regions of a band diagram wheref water intersect the reector bandgap (horizontal greenig. 2)

2

whererespecers; ncof the tion (1resonathe ma

Wetages fcounteused foanalytefor theiagram of (a) TE and (b) TM polarized modes of a PMMA/PS Bragg reector. The grey regr. The clear regions correspond to the parts of phase space where light is forbidden to prd water. Transmission bands (green) of the Bragg ber can be estimated from the inters

of the references to color in this gure legend, the reader is referred to the web version oc l l c

h are thicknesses of the low- and high-index layer,; nl, nh are real parts of RIs of the corresponding lay-

al part of RI of the core material. Variations in real partte RI lling the core could modify the resonant condi-the Bragg ber, thus resulting in spectral shifts of theavelength in the ber transmission, which constitutesnsing principle of the Bragg ber sensor.

that low-RI-contrast Bragg bers have certain advan-uid analyte sensing compared to their high-RI-contrastts. Previously, high-RI-contrast Bragg bers have beensing of gas analytes (nanalyte 1) [13], as well as liquidth high refractive indices (nanalyte > 1.40) [6]. However,-RI-contrast Bragg bers used in these works, the TMions corresponds to (, ) for which light can propagate within theopagate in the Bragg reector. Thick black curves represent the lightection of light line of water with the Bragg reector bandgaps. (Forf this article.)

238 H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243

Fig. 3. (a) Sim erent NaCl solutions. (b) Experimental transmission spectra of the 40 cmlong Bragg be rst experiment after several hours, which demonstrate a good repeatabilityof the measure in both gures as insets.

bandgaps othe aqueousnomenon, tpropagatingbers showwater, thusMoreover, wsitive to chcompared t(1), we deri

S = dcdnc

= 2

Accordinof the indivthe Bragg core, whichsensor.

3. Simulat

To verifyulate the loliquid-core [28]. The stthose used solutions wa 5% incremshown in Fspectral ranthat of purea propagatiloss of watereector. Thband of thethe liquid a

In our exto simultan(Fig. 1). A circle, is intBragg ber izontal chaof its extremIn each blothe horizonpling of thein the sensimission. Af

m from a supercontinuum source into the ber using a 10ve, and the transmission spectrum of the liquid-core Bragg

then analyzed by the Newport grating monochromator. Tothe repeatability of the sensor, we have repeated the samement after 3 h by rst purging the setup with distilled water.) shows the result of two consecutive experiments. The solidepresents transmission spectrum of the rst measurement,comp

curvectetays specn spen llnsmiive in

noten peay duectraticalighe

the , we an peaode

perimrder l speulated loss of the fundamental mode (HE11 mode) of the Bragg ber lled with diffr lled with NaCl solutions (solid curves). The dotted curves indicate a repeat of the ment. The weight concentrations (wt.%) and the RIs of the NaCl solutions are listed

f the Bragg reector tend to collapse near the light line of material (nanalyte 1.33) due to the Brewster angle phe-hus leading to high loss for the hybrid (HE/EH) modes

in the liquid cores. In contrast, low-RI-contrast Bragg large TM bandgaps in the vicinity of the light line of

resulting in good guidance of HE/EH modes (Fig. 2).e note that low-RI-contrast Bragg bers are more sen-

anges in RI of liquid analytes lling the ber core, aso high-RI-contrast Bragg bers. Particularly, from Eq.ve the sensitivity of the sensor, S, as:[

dh

(n2

h

n2c 1)1/2

+ dl(

n2l

n2c 1)1/2]

(2)

g to Eq. (2), the closer is the value of the core RI to thoseidual layers of the Bragg reector; the more sensitiveber sensor will be to variations in RI of the analyte-lled

is exactly the case for our low-RI-contrast Bragg ber

ion and experiments

the resonant sensing mechanism theoretically, we sim-ss spectra of the fundamental mode (HE11 mode) of theBragg ber based on the Transfer Matrix Method (TMM)ructural parameters of the HC Bragg ber are same asin Section 2. As liquid analytes, we choose a set of NaClith the weight concentration ranging from 0 to 25% withent step. The corresponding RIs of NaCl solutions are

ig. 3 [29]. The bulk absorption of NaCl solutions in thege of interest was shown to be virtually identical with

water [30]. In our simulations, we, therefore, computeon loss of an HE11 mode taking into account absorptionr, as well as dispersion of water and plastics in the Bragge simulated loss spectra suggest that the transmission

liquid-core Bragg ber shows a blue-shift as the RI of

the beaobjectiber isverify experiFig. 3(bcurve rand it dottedthe detpeak sof our missiosolutiothe trarefract

Wemissiois likelthe spRI idenmost hthat ofEq. (2)missioHE11 mthe exhigh-oimentanalyte increases (Fig. 3(a)).periments, we employ two opto-uidic blocks designedeously enable optical and uidic coupling of the HC ber40 cm long Bragg ber, coiled into a 10 cm diameteregrated into the sensing system. On each side, the HCtip is inserted hermetically into the analyte-lled hor-nnel that has a thin glass window attached at the one

ities for optical coupling of the Bragg ber (Fig. 1(c)).ck there is also a vertical channel which connects total channel to constitute the uidic path for uidic cou-

Bragg ber. This design avoids formation of air bubblesng system which would strongly suppress ber trans-ter pumping a liquid analyte into the ber, we couple

Fig. 4. Spectraulation (blackinterpretationto the web verares well with the second measurement shown as thee. We observe up to no more than 2.51% uctuation ind intensity, while the center position of the transmissionat the same position up to the resolution limit (1 nm)trometer. As seen in Fig. 3(b), the experimental trans-ctrum also features a blue-shift as the RI of the NaCling the core increases. Moreover, the spectral shifts ofssion peak show a linear dependence on the increasingdex of the analyte lling the core (Fig. 4).

that the experimental spectral shifts of a ber trans-k are somewhat smaller than the simulated ones. Thise to the fact that in our simulation we only calculatel shifts of the fundamental (HE11) mode with effective

to that of the liquid analyte lling the core. Virtually,r-order core modes have effective RIs (neff) lower thanHE11 mode. Therefore, by substitution of nc by neff inrrive to the conclusion that spectral shifts of the trans-ks of higher-order modes are smaller than those of the. Due to large diameter of the Bragg ber core used inents, many higher-order modes are excited. As these

modes are less sensitive than the HE11 mode, the exper-ctral shifts are, therefore, smaller than those of the HE11l shifts of the ber transmission peak obtained from the TMM sim- solid line) and the experimental measurements (red dash line). (For

of the references to color in this gure legend, the reader is referredsion of this article.)

H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243 239

mode. Finally, from Fig. 4, we conclude that the experimental sen-sitivity of the liquid-core Bragg ber sensor is 1400 nm/RIU. Sucha sensitivity is comparable to those of previous MOF based sen-sors [4,5,31,32], but is an order smaller than that of the dual-coreMOF sensorwe note thsensing liquto other MOin its short ber.

4. Discussi

4.1. Dynam

The chomined by twe routinelin 500800conducted ivarying thealready spaband diagraoperation isRI analytes analytes, ththe wavelenthe sensor lytes the bthere is no fanalytes.

4.2. Insertio

Due to thcoupling efof the well the sensor cwindow covand out-cou

4.3. Depend

To studyber, we rlong Bragg tions. Thenlength we in responseber transmlengths.

The tranclear spectrThe sensitiwhich is esmission peaof the 30 cmtransmissiobecome broedges becauof the 30 cmFurther redsmaller prober loss cafeatures in tdifcult to d

Therefore, we conclude that the liquid-core Bragg ber shouldhave a minimal (threshold) length to ensure sufcient attenuationat the wavelengths in the vicinity of bandgap edges in order to allowformation of spectral features (such as transmission peaks) in the

ansm cm fngthberis coo a st

high samease

to ssing

thaton oftiong bne cithog a spber

e ma

uid-cundl

agg

ently impore Baditi

to te den

specularan beindivolut mebern of rge restinased ectro, oneric men

ng syematThe bersre mng ace pro ide cor

of t posiore Bhe B with a sensitivity of 30,000 nm/RIU [33]. However,at the dual-core MOF sensor is limited to be used forid analytes with nanalyte > nsilica. In addition, comparedF based sensors, our Bragg ber sensor is advantageousresponse time (1 s) due to the large core of the Bragg

on of factors inuencing sensor performance

ic range of the sensor

ice of the operation frequency range is mainly deter-he position of the Bragg reector bandgap. Currently,y produce Bragg bers with primary bandgaps located

nm spectral range; therefore, all our experiments aren this range. Sensitivity of the sensor is so high that with

liquid-core RI from 1.333 to 1.378, the ber bandgapns this whole operational window (see Fig. 3). Fromm of Fig. 2, it follows that the largest wavelength of

1300 nm. That is the region where sensing of low-with neff 0 can be achieved. In application to aqueouse bulk absorption is generally larger than 50 dB/m atgth above 1 m, thus limiting the operation range of

in the near-infrared range. Finally, for the high RI ana-er bandgap will shift into blue spectral region; however,undamental limit for the use of these sensors for high-RI

n and coupling loss

e large diameter of the Bragg ber core (over 0.5 mm),ciency of the ber is superb (above 90%), which is one

known advantages of such bers. Main coupling loss ofomes from the reection between the air and the glassering the uidic channel (Fig. 1(c)). The total couplingpling loss is estimated to be 10%.

ence of sensitivity on ber length

the dependency of the sensitivity on the length of Braggst measure the transmission spectral shifts of a 50 cmber lled with NaCl solutions of different concentra-, we cut the ber into 30 cm and 20 cm, and for eachrepeat the experiment to measure the spectral shifts

to variations in the liquid-core RI. In Fig. 5, we plotission spectra of liquid-core Bragg ber with different

smission spectra of the 50 cm long Bragg ber showal shifts caused by variations in the core RI (Fig. 5(a)).vity of the 50 cm long ber sensor is 1300 nm/RIUtimated from the spectral shifts of the specic trans-k marked by white arrows. In the transmission spectra

long Bragg ber, we still observe spectral shifts of then peak (Fig. 5(b)); however, the transmission spectraader due to reduced ber attenuation at the bandgapse of shorter ber length. The experimental sensitivity

long Bragg ber sensor is 1270 nm/RIU (Fig. 5(d)).uction of the Bragg ber length to 20 cm leads to evenpagation losses at all frequencies. However, the reduceduses the difculty in differentiating specic resonanthe ber transmission spectra, and as a result it becomesetect spectral shifts.

ber trbe 10ber leBragg which leads ting as aAt theto incrleadingfor senfactorsmizatiapplicaof Brag1 m, ober wsensinBragg and th

5. Liqber b

5.1. Br

Reccan besolid-cof a trrelatedthe linbundlea partitrum cin the deconvmatrixBragg positiofor a lais interber band spIdeallytrometrequirereduci

SchFig. 6. Bragg eter cofeaturiguidanare alsthat thnessescentersolid-cduce tission. We experimentally nd the threshold length toor most Bragg bers used in our experiments. Once the

is longer than the threshold value, the sensitivity of a sensor does not strongly depends on the ber length,nsistent with the prediction in [24]. Longer ber alsoronger attenuation of the higher-order modes, thus act--order mode striper, which is also benecial for sensing.e time, signal strength in longer bers decreases dued material absorption and radiation losses, eventuallyignal-to-noise degradation, which is disadvantageous. It is the trade-off between the two above-mentioned

determine the optimal ber length. Note that the opti- the sensing length is not strictly required for sensings. Experimentally, we found the maximal sensing lengther to be 1 m. With the sensing length smaller thanan always measure the transmission spectrum of Braggut much signal-to-noise degradation. Therefore, whenecic liquid analyte, one, in practice, could choose thewith any length between the threshold value (10 cm)ximal length (1 m).

ore Bragg ber sensor integrated with Bragge spectrometer

ber bundle spectrometer

, we demonstrated that the function of a spectrometerlemented in a solid-core Bragg ber bundle [25]. Theragg ber bundle spectrometer breaks the paradigm

onal spectrometer in which the resolution is directlyhe length of the light path inside the spectrometer andsity of the Bragg grating used. Instead, the Bragg ber

ctrometer relies on the ability of Bragg bers to lter spectral window of the incoming light. The test spec-

reconstructed by interrogating transmitted intensitiesidual Bragg bers in the bundle and then by applying aion algorithm to these intensities with a transmissionthod. We have experimentally demonstrated that the

bundle spectrometer can reconstruct the center peaka test spectrum within several percent of its true valueange of the peak widths and positions. In this context, itg to investigate the possibility of designing an all-PBGspectroscopic system where the function of sensing cellmeter are implemented in the all ber conguration.

would splice the hollow core Bragg ber with a spec-ber to implement an integrated system with minimalts on the processing and acquisition equipment, thusstem cost and simplifying its maintenance.ic of the Bragg ber bundle spectrometer is shown inBragg ber bundle is fabricated using 100 solid-core. Individual Bragg bers have a large 300800 m diam-ade of PMMA. The core is surrounded by a Bragg reector

submicrometer-thick PMMA/PS multilayer. Note thatinciples of the liquid-core Bragg ber described aboveentical to those of the solid-core Bragg ber, exceptes RI of the latter is xed. According to Eq. (1), thick-he individual layers in the Bragg reector determinetion of the transmission of a Bragg ber. By choosingragg bers with different transmission bands, we pro-ragg ber bundle which at the output has a mosaic

240 H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243

Fig. 5. Transmission spectra of the liquid-core Bragg ber with different lengths: (a) 50 cm; (b) 30 cm; (c) 20 cm. The RIs of the analytes lling the ber core are also listedin the inset of (c). The white arrows in (a) and (b) mark the resonant peak positions which we use to measure spectral shifts. A linear dependence of the spectral shifts onchanges in RI of the ber core is shown in (d). Black solid line represents the spectral shifts of a 50 cm long Bragg ber and red dashed line represents the spectral shifts of a30 cm long Bragg ber. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

Fig. 6. Fiber bundle spectrometer [25]. Top, schematic of the spectrometer. The test spectrum is launched into the ber bundle; output image of the ber bundle is taken bya monochrome CCD camera. Bottom, when the broadband light is launched into the ber bundle, the output is a mosaic of colors selected by the individual Bragg bers. Abeam compressor is used to ensure that the CCD camera accommodates the whole image of the output of Bragg ber bundle. (For interpretation of the references to color inthis gure legend, the reader is referred to the web version of this article.)

H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243 241

Fig. 7. Transm ber bThe weight co set. (cmonochromat For intis referred to t

of colors (light.

An impober spectrpropagationIn practice,the 30 cm lois 20 dB of bands of thwide (601these factoris typically

The operdescribed abers are spintensity ofing spectrain our specoverlappingstruct the tBragg berstime of a CCof a CCD campresented a

Cn = max

min

where I() iinput end; Athe transmisitivity of atransmissiosor). In Eq. (the transmcoupled ligdene the i(bins) with

et th

=

vectoividuents

issiectruasur

ratioq. (4ng Cberissioission spectra of the 40 cm long liquid-core Bragg ber measured by (a) the Bragg ncentrations and the corresponding RIs of the NaCl solutions are listed in the inor (dashed red line) and by the Bragg ber bundle spectrometer (solid black line). (he web version of this article.)

bottom part of Fig. 6), when illuminated by white

rtant factor that inuences the performance of a Braggometer is the intensity throughput which relates to the

losses of individual Bragg bers in the ber bundle. the typical propagation loss of an individual ber inng bundle is 3 dB of the light inside the bandgap andthe light outside the bandgap. Besides, transmissione individual Bragg bers in the bundle are relatively80 nm FWHM) and partially overlapping. Because ofs, the intensity throughput of an individual Bragg beras much as 20% of the in-coupled light.ation principle of a Bragg ber bundle spectrometer iss follows. If transmission bands of the individual Braggectrally narrow and non-overlapping, the transmitted

each ber would be directly related to the correspond-l component of an incoming light. As a matter of fact,trometer the individual ber bandgaps are strongly

interpras C1...

C100

wherethe indreprestransmtest sptor) mea calibFrom EspondiBragg transm; we, therefore, use a deconvolution algorithm to recon-est spectrum from the transmitted intensities of the

in the ber bundle. Particularly, for a xed exposureD array, intensity Cn registered by the assigned regionera corresponding to the output of the nth ber can be

s:

I()AnFn()S()On()d (3)

s the incoming spectral ux density at the ber bundlen is the area of the nth ber input cross section; Fn() isssion function of the nth ber, S() is the spectral sen-

CCD sensor; and On() is a ber-position-dependentn function of various optics (diffuser, beam compres-3), we can dene a Tn() = AnFn()S()On() to representission efciency of the nth ber to transmit the in-ht to the CCD camera. In addition, we experimentallyncoming spectrum in terms of small spectral intervalsequivalent width. We, then, use a matrix formalism to

the matrix Temploy a sto calculatespectral rec

5.2. Calibrabundle spec

To calcucalibration We couple monochromof the monber (fromwith distillis relative wing calibratloss, thus ment. Ther(SMBF) sououtput is cBragg ber undle spectrometer and (b) a conventional grating monochromator.) Spectral shifts of the transmission peaks measured by the gratingerpretation of the references to color in this gure legend, the reader

e integral in (3) for all the Bragg bers of the ber bundle

T1,1 T1,N... . . . ...

T100,1 T100,N

I1...

IN

(4)

r (C)100 1 represents the intensities of light coming outal Bragg bers as measured by the CCD camera; (I)n 1

the discretized spectrum of the illuminant; T(100 n) is aon matrix which denotes a linear conversion from them (I vector) to the ber transmitted intensities (C vec-ed by the CCD sensor. We calculate the matrix T throughn measurement demonstrated in the following section.) the test spectrum can be reconstructed from the corre-CD image by inverting the transmission matrix of the

spectrometer. Note that due to strong overlap of then spectra of the individual Bragg bers in the bundle, is ill-conditioned (even if made square). We, therefore,tandard singular value decomposition (SVD) method

the pseudoinverse of the response matrix T to enableonstruction.

tion of the sensor integrated with the Bragg bertrometer

late the response matrix T in Eq. (4), we carry out ameasurement for the Bragg ber bundle spectrometer.the beam of a supercontinuum source into a tunableator (2 nm FWHM). Using a 10 objective, the outputochromator is then coupled into a 40 cm long Bragg

the same perform of the ber used in Fig. 3) lleded water. Since the output from the monochromatoreak, the water-lled Bragg ber is kept straight dur-

ion measurement in order to reduce the propagationincreasing the signal-to-noise ratio of the measure-efore, a supercontinuum-monochromator-Bragg-berrce which has a tunable and quasi-monochromaticonstructed. The SMBF source is then coupled to thebundle spectrometer using another 10 objective. We

242 H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243

Fig. 8. Cross s lluminmosaic of colo etatioto the web ver

vary the SMeffectively sideration iintensity, Iipowermetewe also acqcolumn of teter. By divtransmissioT, we can rEq. (4). We setup and dcore Bragg of the liquiber bundlenation condbe identicalis critical tprocess [25

5.3. Liquid-bundle spec

The NaCSection 3 ara ber bundcore Bragg is shown in Bragg ber grating monspectra meaby the gratitra reconstrby the gratcalculate ththe correspare above 0

In Fig. 7sured by bomonochromdependenceBragg ber.sor is 110and is 105Note that itra shifts (s

4. T (incber liquimeas senthrouer, iationpixetectl thep ou

wing

ally, r ind, thisg b

the Secoaw trawiundlplim

tly win orters, ections of ber bundles drawn by the two-stage drawing technique. The bundle is irs is visible as the white light is ltered by each ber inside the bundle. (For interprsion of this article.)

BF source center wavelength in 2 nm increments, thussubdividing 560750 nm spectral interval under con-nto m = 96 equivalent 2 nm wide bins. The SMBF source, at each wavelength is also measured by the calibratedr. For every position of the source center wavelength,uire a C vector using the CCD array and consider it as ahe response matrix of the Bragg ber bundle spectrom-iding every C vector by the corresponding Ii value, then matrix T is constructed. After we calculate the matrixeconstruct a test spectrum by inverting T according tothen remove the monochromator from the calibrationirectly couple the supercontinuum beam to the liquid-ber with a 10 objective. The transmission spectrad-core Bragg ber can be then analyzed by the Bragg

spectrometer. Note that in the calibration, the illumi-ition on the input-end facet of the ber bundle should

to that in the following spectral characterization, whicho minimize the error in the spectral characterization].

core Bragg ber transmission measured by Bragg bertrometer

l solutions of the same concentrations as discussed ine used to characterize the sensing system that includesle spectrometer. The transmission spectra of the liquid-ber measured by the Bragg ber bundle spectrometerFig. 7(a). As a reference, we also measure the liquid-coretransmission by directly coupling it to a conventionalochromator (Fig. 7(b)). For the ease of comparison, the

in Fig.mentsBragg of the of the are lesagate Howevdegradsingle-CCD dewe coito strishifts.

6. Dra

Finable foBrieya Bragit into2 mm. and drstage dber bing comcurrennique diamesured both by the Bragg ber bundle spectrometer andng monochromator are normalized to 1. Since the spec-ucted by the ber bundle spectrometer and measureding monochromator are relatively wide, we, therefore,e center position of a transmission peak by averagingonding wavelengths where the normalized intensities.95.(c), we compare the shifts of transmission peaks mea-th the Bragg ber bundle spectrometer and the gratingator. The transmission peak shifts also feature linear

with respect to the variations in the core RI of the Experimental sensitivity of the liquid Bragg ber sen-0 nm/RIU measured by the grating monochromator0 nm/RIU measured by the Bragg ber spectrometer.n both measurements presented in Fig. 7, the spec-ee Fig. 7(c)) are marginally smaller than those shown

spliced wit

7. Conclus

We repoliquid RI dmechanismshifts in resber core. Bments are cThe experimis comparaand other Msensitivity odent on theated with a broadband halogen lamp. At the output of ber bundle, an of the references to color in this gure legend, the reader is referred

his is likely due to the fact that during the measure-luding the calibration measurements) for Fig. 7, theis kept straight in order to reduce the propagation lossd-core ber and then increase the signal-to-noise ratiosurements. Therefore, many high-order modes, whichsitive to variations in RI of the ber core, can prop-gh the ber, thus resulting in smaller spectral shifts.n the measurements for Figs. 3 and 4, signal-to-noise

is not a critical issue, since we use a more sensitivel detector for spectral interrogation as compared to theor used in the Bragg ber spectrometer. Consequently,

Bragg ber piece as a 10 cm diameter circle in ordert high-order modes, thus leading to stronger spectral

of PBG ber bundles

we demonstrate a two-stage drawing technique suit-ustrial-strength production of the Bragg ber bundles.

technique comprises two steps. Firstly, we fabricateer perform by the co-rolling method [34] and drawcanes of different diameters ranging between 1 andndly, we bundle these canes inside a supercladding tube,he structure into the nal Bragg ber bundle. The two-ng technique, in principle, enables fabrication of smalleres that include a large number of Bragg bers featur-entary spectral transmission windows (Fig. 8). We are

orking on the perfection of the two-stage drawing tech-der to draw Bragg ber bundles with reduced outerwhich eventually enables a ber bundle to be directly

h a single HC Bragg ber.

ions

rt a liquid-core low-RI-contrast Bragg ber sensor foretection. The sensor operates on a resonant sensing

in which the transmission spectrum of a Bragg berponse to changes in RIs of liquid analytes lling in theoth theoretical simulations and experimental measure-arried out to verify the resonant sensing mechanism.ental sensitivity of the sensor is 1400 nm/RIU, which

ble to those of surface plasmon resonance sensors [35]OF-based sensors. Besides, we also conclude that thef a liquid-core Bragg ber sensor is generally indepen-

ber length.

H. Qu et al. / Sensors and Actuators B 161 (2012) 235 243 243

Moreover, we demonstrate the integration of a Bragg ber bun-dle spectrometer in the liquid-core Bragg ber sensor system forspectral acquisition, which could potentially lead to signicantcost-savings and increased sensing speed by forgoing a traditionalgrating-based spectrometer. To the best of our knowledge, it is therst time that a ber-optic sensor is proposed that simultaneouslyuse all hollow- and solid-core PBG bers to physically hold andspectrally probe liquid analytes. Finally, we propose a two-stagedrawing technique for the fabrication of solid-core Bragg ber bun-dle spectrometers. This technique results in integrated Bragg berbundles with smaller diameters, which could also allow their directsplicing witsensitive, lo

References

[1] R. Manor,tion and cTeon AF

[2] C. Ma, B.multigas

[3] V. Benoit,ment in c

[4] J. Sun, C.Sens. Actu

[5] H.F. Xuanbandgap European

[6] K.J. RowpropertieOptics/In

[7] L. Rindorbiochip u(2006) 13

[8] J.B. JensenJ. Riishedber basesolutions

[9] Y. Han, Mliquid-cosurface-e

[10] Y. Zhang,sensor ba193504.

[11] J. Irizsar, Helmy, Ratal bers,

[12] X. Yang, sensitivitsurface-e

[13] C. Charltomeet nanphotonic

[14] L.C. Shi, Wapplicatioceedings 2011, p. 7

[15] T. Ritari, JSimonsen12 (2004)

[16] Y.L. Hoo,methane (2010) 29

[17] A.M. CubLopez-Hitonic ban

[18] A.M. CubMethane gas cell, E

[19] M.P. Burictering an47 (2008)

[20] M.P. Buric, K.P. Chen, J. Falk, S.D. Woodruff, Improved sensitivity spontaneousRaman scattering multi-gas sensor, CLEO/IQEC 2009, CThI6.

[21] S.H. Aref, R. Amezcua-Correac, J.P. Carvalho, O. Frazao, J.L. Santos, F.M. Araujo, H.Lati, F. Farahi, L.A. Ferreira, J.C. Knight, Spectral characterization of a photonicbandgap ber for sensing applications, Appl. Opt. 49 (2010) 1870.

[22] G. Kim, T. Cho, K. Hwang, K. Lee, K.S. Lee, Y.G. Han, S.B. Lee, Strain and temper-ature sensitivities of an elliptical hollow-core photonic bandgap ber based onSagnac interferometer, Opt. Express 17 (2009) 2481.

[23] H. Xuan, W. Jin, M. Zhang, J. Ju, Y. Liao, In-ber polarimeters based on hollow-core photonic bandgap bers, Opt. Express 17 (2009) 13246.

[24] M. Skorobogatiy, Resonant bio-chemical sensors based on Photonic Bandgapwaveguides and bers, in: M. Zourob, L. Akhlesh (Eds.), Optical Guided-waveChemical and Biosensors II, Springer, Berlin, Heidelberg, 2010, pp. 4372.

[25] H. Qu, B. Ung, I. Syed, N. Guo, M. Skorobogatiy, Photonic bandgap ber bundlectrometer, Appl. Opt. 49 (2010) 4791.one, C. Dubois, N. Guo, A. Dupuis, F. Boismenu, S. Lacroix, M. Skorobogatiy,wing Skorobe wav(2005. Johnjacic, Sde pro. Ha

a Rato. Sulliaghaorptiot. 45 (2. Fros

indeing, Oindorfg-peri.C. Wactiveao, Nartin

rollingater. assaned sur

phie

has rspecti

Polyte ber

g has and 20ics athip inics anhertz

ieu R work10. His proc

sim S of Tecntralssachcs alsoons grzationser beMicroral, Cc cryst

plasmh a HC Bragg ber in order to enable a highly compact,w-cost, all-ber solution of liquid RI sensing.

A. Datta, I. Ahmad, M. Holtz, S. Gangopadhyay, T. Dalls, Microfabrica-haracterization of liquid core waveguide glass channels coated with, IEEE Sens. J. 3 (2003) 687.

Scott, G. Pickrell, A. Wang, Porous capillary tubing waveguide forsensing, Opt. Lett. 35 (2010) 315.

M.C. Yappert, Effect of capillary properties on the sensitivity enhance-apillary/ber optical sensors, Anal. Chem. 68 (1996) 183188.C. Chan, Photonic bandgap ber for refractive index measurement,ators B 128 (2007) 4650., W. Jin, J. Ju, H.L. Ho, M. Zhang, Y.B. Liao, Low-contrast photonicbers and their potential applications in liquid-base sensors, in: Third

Workshop on Optical Fibre Sensors, Proc. SPIE 6619 (2007) 661936.land, S. Afshar, V.A. Stolyarov, Y. Fink, T.M. Monro, Spectrals of liquid-core Bragg bers, in: Conference on Lasers and Electro-ternational Quantum Electronics Conference, Maryland, CThE2, 2009.f, P.E. Hoiby, J.B. Jensen, L.H. Pedersen, O. Bang, O. Geschke, Towardssing microstructured optical ber sensors, Anal. Bioanal. Chem. 385701375., L.H. Pedersen, P.E. Hoiby, L.B. Nielsen, T.P. Hansen, J.R. Folkenberg,

e, D. Noordegraaf, K. Nielsen, A. Carlsen, A. Bjarklev, Photonic crystald evanescent-wave sensor for detection of biomolecules in aqueous, Opt. Lett. 29 (2004) 1974..K. Khaing, Y. Zhu, L. Xiao, M.S. Demohan, W. Jin, H. Du, Index-guidingre photonic crystal ber for solution measurement using normal andnhanced Raman scattering, Opt. Eng. 47 (2008) 040502.

C. Shi, C. Gu, L. Seballos, J.Z. Zhang, Liquid core photonic crystal bersed on surface enhanced Raman scattering, Appl. Phys. Lett. 90 (2007)

J. Dinglasan, J.B. Goh, A. Khetani, H. Anis, D. Anderson, C. Goh, A.S.man spectroscopy of nanoparticles using hollow-core photonic crys-

IEEE J. Select. Top. Quant. Electron. 14 (2008) 1214.C. Shi, D. Wheeler, R. Newhouse, B. Chen, J.Z. Zhang, C. Gu, High-y molecular sensing using hollow-core photonic crystal ber andnhanced Raman scattering, J. Opt. Soc. Am. A 27 (2010) 977.n, B. Temelkuran, G. Dellemann, B. Mizaikoff, Midinfrared sensorsotechnology: trace gas sensing with quantum cascade lasers insideband-gap hollow waveguides, Appl. Phys. Lett. 86 (2005) 194102.. Zhang, J. Jin, Y. Huang, J.D. Peng, Hollow-core Bragg ber and its

n in trace gas sensing, optical sensors and biophotonics II, in: Pro-of SPIE-OSA-IEEE Asia Communications and Photonics, SPIE, vol. 7990,99008.. Tuominen, H. Ludvigsen, J.C. Petersen, T. Sorensen, T.P. Hansen, H.R., Gas sensing using air-guiding photonic bandgap bers, Opt. Express

4080. S. Liu, H.L. Ho, W. Jin, Fast response microstructured optical bersensor with multiple side-openings, IEEE Photonic Technol. Lett. 226.illas, M. Silva-lopez, J.M. Lazaro, O.M. Conde, M.N. Petrovich, J.M.guera, Methane detection at 1670-nm band using a hollow-core pho-dgap ber and a multiline algorithm, Opt. Express 15 (2007) 17570.illas, J.M. Lazaro, M. Silva-Lopez, M.N. Petrovich, J.M. Lopez-Higuera,sensing at 1300 nm band with hollow-core photonic bandgap bre aslectron. Lett. 44 (2008) 403., K.P. Chen, J. Falk, S.D. Woodruff, Enhanced spontaneous Raman scat-d gas composition analysis using a photonic crystal ber, Appl. Opt.

4255.

spe[26] E. P

Dra[27] M.

cor30

[28] S.GSolmo

[29] W.MBoc

[30] J.MDonabsOp

[31] M.Htivemix

[32] L. Rlon

[33] D.Krefr

[34] Y. GL. Mco-J. M

[35] A. Hbas

Biogra

Hang Qu2007, reat Ecolebandgap

Bora Unin 2004 ing physScholarsber optand tera

Dr. Math2009. Hesince 20soft glas

Dr. MakInstitutesity, Mofrom Main physisimulatimercialipower laChair in de Montphotonields ofof the hollow all-polymer Bragg bers, Opt. Express 14 (2006) 5838.ogatiy, Efcient antiguiding of TE and TM polarizations in low-index

eguides without the need for an omnidirectional reector, Opt. Lett.) 2991.son, M. Ibanescu, M. Skorobogatiy, O. Weiseberg, T.D. Engeness, M..A. Jacobs, J.D. Joannopoulos, Y. Fink, Low-loss asymptotically singe-pagation in large core omniguide bers, Opt. Express 9 (2001) 748.ynes, CRC Handbook of Chemistry and Physics, 91st ed., CRC Press,n, FL, 2010, pp. 871.

van, M.S. Twardowski, J.R.V. Zaneveld, C.M. Moore, A.H. Barnard, P.L.y, B. Rhoades, Hyperspectral temperature and salt dependence ofn by water and heavy water in the 400750 nm spectral range, Appl.006).z, A. Stefani, O. Bang, Highly sensitive and simple method for refrac-x sensing of liquid in microstructured optical bers using four-wavept. Express 19 (2011) 10471., O. Bang, Highly sensitive refractometer with a photonic-crystal-berod grating, Opt. Lett. 33 (2008) 563.u, B.T. Kuhlmey, B.J. Eggleton, Ultrasensitive photonic crystal ber

index sensor, Opt. Lett. 34 (2009) 322.. Guo, B. Gauvreau, M. Rajabian, O. Skorobogata, E. Pone, O. Zabeida,u, C. Dubois, M. Skorobogatiy, Consecutive solvent evaportation and

technique for polymer multilayer hollow ber perform fabrication,Res. 21 (2006) 2246.i, M. Skorobogatiy, Design criteria for microstructured-optical-ber-face-plasmon-resonance sensors, J. Opt. Soc. Am. B 24 (2007) 1423.

s

eceived his B.Sc. and M.Sc. in physics from Jilin University in 2005 andvely. He is currently working towards a Ph.D. in engineering physicschnique de Montreal. His current elds of interest include photonic

sensing, capillary ber sensing, microstructured-ber-based devices.

received his BE and M.Sc. in engineering physics from Laval University07, respectively. He is currently working towards a Ph.D. in engineer-

Ecole Polytechnique de Montreal. He is the recipient of a SPIE 2011 Optics & Photonics. His current elds of interest include nonlineard waveguides, ber-based devices, nano-optics, plasmonics, infraredscience.

oze has received his Ph.D. in chemistry from University of Rennes 1 ins as a postdoctoral research fellow at Ecole Polytechnique de Montreals elds of interest include fabrication of microstructured optical ber,essing.

korobogatiy has received his B.Sc. degree in physics from Rochesterhnology, Rochester, USA (1995); M.Sc. in physics from McGill Univer-, Canada (1997); M.Sc. in electrical engineering and computer scienceusetts Institute of Technology (MIT), Cambridge, USA (2000); Ph.D.

from MIT in 2001. During 20012003 he was heading a theory andoup at MIT startup OmniGuide Inc., where he was working on com-

of the hollow photonic bandgap ber technology for delivery of highams. Since 2003, he is an associate professor and a Canada Research- and Nano-Photonics at Ecole Polytechnique de Montral/Universityanada. His current elds of interest include design and fabrication ofal bers for sensing and telecommunications, as well as the generalonics, and nano-photonics.

All photonic bandgap fiber spectroscopic system for detection of refractive index changes in aqueous analytes1 Introduction2 Operation principle of liquid-core Bragg fiber sensor3 Simulation and experiments4 Discussion of factors influencing sensor performance4.1 Dynamic range of the sensor4.2 Insertion and coupling loss4.3 Dependence of sensitivity on fiber length

5 Liquid-core Bragg fiber sensor integrated with Bragg fiber bundle spectrometer5.1 Bragg fiber bundle spectrometer5.2 Calibration of the sensor integrated with the Bragg fiber bundle spectrometer5.3 Liquid-core Bragg fiber transmission measured by Bragg fiber bundle spectrometer

6 Drawing of PBG fiber bundles7 ConclusionsReferences