&.,1785802,*5'72>.:;2<A=+527&.,1785802,*5'72>.:;2<A=+527
$$!(&'=+527$$!(&'=+527
:<2,5.;
%,18858/*7=/*,<=:270*7-.;207
7027..:270
201%.7;2<2>2<A$./:*,<2>.7-.@%.7;8:*;.-87*&*9.:.-201%.7;2<2>2<A$./:*,<2>.7-.@%.7;8:*;.-87*&*9.:.-
%6*558:.%2705.8-.2+.:%<:=,<=:.%6*558:.%2705.8-.2+.:%<:=,<=:.
.3=72=
&.,1785802,*5'72>.:;2<A=+527
-.3=752=<=-=+5272.
:=7*5524
&.,1785802,*5'72>.:;2<A=+527
*:=76*5524<=-=+5272.
271=2)=*7
.9*:<6.7<8/"1A;2,;*7-5.,<:2,*57027..:270 8:<1=6+:2*'72>.:;2<A'
%..7.@<9*0./8:*--2<287*5*=<18:;
8558?<12;*7-*--2<287*5?8:4;*<1<<9;*::8?<=-=+5272..70;,16*7*:<
"*:<8/<1.7027..:27086687;
$.,866.7-.-2<*<287$.,866.7-.-2<*<287
*5524201;.7;2<2>2<A:./:*,<2>.27-.@;.7;8:+*;.-87*<.9.:.-;6*55,8:.;2705.68-.C+.:
;<:=,<=:.
!9<2,;.<<.:;>857899-82!
&12;:<2,5.2;+:8=01<<8A8=/8:/:..*7-89.7*,,.;;+A
<1.%,18858/*7=/*,<=:270*7-.;2077027..:270*<
$$!(&'=+527<1*;+..7*,,.9<.-/8:27,5=;28727
:<2,5.;+A*7*=<18:2B.-*-6272;<:*<8:8/$$!(&'
=+5278:68:.27/8:6*<28795.*;.,87<*,<
*::8?*-627<=-=+5272.*2;5270,8A7.<=-=+5272.
&12;?8:42;52,.7;.-=7-.:*:.*<2>.86687;
<<:2+=<287 87,866.:,2*5%1*:.524.2,.7;.
High sensitivity refractive index sensor based on
a tapered small core single-mode fiber structure
DEJUN LIU,
1
ARUN KUMAR MALLIK,
1
JINHUI YUAN,
3
CHONGXIU YU,
3
GERALD FARRELL,
1
YULIYA SEMENOVA,
1
AND QIANG WU
1,2,3,
*
1
Photonics Research Center, Dublin Institute of Technology, Kevin Street, Dublin 8, Ireland
2
Department of Physics and Electrical Engineering, Northumbria University, Newcastle Upon Tyne, NE1 8ST, UK
3
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications,
Beijing 100876, China
*Corresponding author: qiang.wu@northumbria.ac.uk
Received 15 June 2015; revised 6 August 2015; accepted 11 August 2015; posted 11 August 2015 (Doc. ID 242978); published 1 September 2015
A high sensitivity refractive index (RI) sensor based on a
tapered small core single-mode fiber (SCSMF) structure
sandwiched between two traditional single-mode fibers
(SMF28) is reported . The microheater brushing technique
was employed to fabricate the tapered fiber structures with
different waist diameters of 12.5, 15.0, and 18.8 μm.
Experiments demonstrate that the fiber sensor with a waist
diameter of 12.5 μm offers the best sensitivity of
19212.5 nm∕RIU (RI unit) in the RI range of 1.4304 to
1.4320. All sensors fabricated in this Letter show good
linearity in terms of the spectral wavelength shift versus
changes in RI. Furthermore, the sensor with the best sen-
sitivity to RI was also used to measure relative humidity
(RH) without any coating materials applied to the fiber sur-
face. Experimental results show that the spectral wavelength
shift changes exponentially as the RH varies from 60% to
95%. A maximum sensitivity of 18.3 nm per relative humid-
ity unit (RHU) was achieved in the RH range of 90.4% to
94.5% RH.
© 2015 Optical Society of America
OCIS codes: (060.2280) Fiber design and fabrication; (060.2310)
Fiber optics; (060.2370) Fiber optics sensors; (060.4005)
Microstructured fibers.
http://dx.doi.org/10.1364/OL.40.004166
Optical fiber sensors show great potential for applications
involving measurements of gas and liquid concentrations, com-
pound materials solidification monitoring, and biomolecules
detection, where sensing of the surrounding refractive index
(SRI) is used as an underlying princ iple [1,2]. To date, a num-
ber of approaches have been proposed for refractive index (RI)
sensors, including fiber Bragg gratings [3], surface plasmon res-
onance [4], single-mode-multimode-single-mode (SMS) fiber
structures [5], optical ring resonators [6], and optical microfiber
couplers [7]. Among these structures, SMS-fiber-structure-
based sensors are attractive, given the additional advantages
of low cost and ease of fabrication. However, to fabricate an
SMS-based RI sensor, a chemical etching process is required
to remove the cladding from the multimode fiber (MMF) sec-
tion, which may introduce some problems with surface rough-
ness of the etched fiber, in addition to safety concerns [5].
A small core single-mode fiber (SCSMF) is a good candidate
to replace the MMF section in an SMS fiber structure, with the
advantage that etching is not required. In our previous work,
we have proved theoretically and experimentally that a
SCSMF-based fiber structure can act as a high sensitivity RI
sensor with a maximum sensitivity of 1808 nm∕RIU [8].
Furthermore, we have demonstrated as an application of the
structure, a humidity sensor utilizing a coating with a maxi-
mum sensitivity of 430 nm∕RHU [9]. Such sensitivities are
still inadequate in some applications, particularly in bio-sens-
ing, which requires detection of extremely small RI variations.
Recently, microf iber and nanofiber sensors have attracted sig-
nificant interest as they offer a range of advantages, such as large
evanescent fields, high nonlinearity, low loss connections to
standard fibers, and strong mode confinement [10]. For exam-
ple, Yadav et al. reported an RI sensor with a sensitivity of
1500 nm∕RIU based on a tapered single-mode fiber [11].
Most recently, Zhu et al. developed a highly sensitive RI sensor
with a sensitivity of 6008 nm∕RIU by coating a Al
2
O
3
nano-
film on a tapered fiber [12]. Xu et al. have demonstrated an RI
sensor with a sensitivity of 6523 nm∕RIU based on cascaded
microfiber knot resonators (CMKRs) with Vernier effect [13 ].
In this Letter , following on our previous work, we report that
by tapering the SCSMF structure, an improved RI sensor is pos-
sible with a sensitivity that is an order of magnitude higher than
the untapered device we reported in [8]. The maximum demon-
strated sensitivity was estimated as 19212.5 nm∕RIU in the RI
range of 1.4304–1.4320. All the sensor samples studied in this
Letter show good linearity of the wavelength shift versus surround-
ing RI change. Furthermore, one of the sensors with the highest
RI sensitivity is also chosen to show that the tangible benefit of the
much improved RI sensitivity is that we show it is possible to
implement a relative humidity (RH) sensor without the need for
any additional humidity sensitivecoatingsonthefibersurface.
In a typical SMF28-SCSMF-SMF28 fiber structure, clad-
ding modes are excited within the cladding of the SCSMF
4166
Vol. 40, No. 17 / September 1 2015 / Optics Letters
Letter
0146-9592/15/174166-04$15/0$15.00 © 2015 Optical Society of America
section because of the core diameter mismatch between SMF28
and SCSMF. Multimode interference for these cladding modes
occurs within the SCSMF section. It is well known that the
propagation constant of the cladding mode (corresponding to
effective RI) is influenced by the RI of the surrounding envi-
ronment. Therefore, a change in the SRI for the sensor will
affect the multimode interference, and, in turn, this will result
in changes in the spectral response which can be monitored by
an optical spectrum analyzer (OSA). Thus, the SRI can be mea-
sured by monitoring the variations of the spectral response,
assuming a suitable calibration has taken place. This type of
sensor is often referred to as an “evanescent sensor.”
Typically an increase in the portion of the evanescent field
exposed to the surrounding environment results in a higher
sensitivity for the sensor [10]. A simple, non-tapered SCSMF
structure has a relatively large diameter (typically 125 μm), re-
sulting in a relatively small portion of the evanescent field being
in contact with the surrounding environment, and, hence, such
a sensor has relatively low sensitivity. One of the solutions to
improve the sensitivity is to use a tapered SCSMF which has a
smaller waist diameter, as rep orted in our previous work [14].
In this Letter, we have systematically investigated the tapered
SCSMF structure for RI sensing, where the diameters of the
SCSMF are tapered down to 12.5, 15.0, and 18.8 μm.
In our experiments, a section of SCSMF (SM450) with a
length of circa 22.5 mm was fusion spliced between two
SMF-28 fibers. The SCSMF section was then tapered using
a microheater brushing technique as shown in Fig. 1(a)
[15]. Three SCSMF sensor samples were fabricated with differ-
ent tapered waist diameters (18.8, 15.0, and 12.5 μm). The
waist diameter was measured by a microscope. The prepared
sensor samples are referred hereinafter to as S-18.8, S-15.0,
and S-12.5. For each taper, the taper waist length is the same
with a length of approximately 3.5 mm. The taper transition
lengths for S-18.8, S-15.0, and S-12.5 are circa 8 mm, 8.5 mm,
and 9 mm, respectively. In each case the tapered SCSMF struc-
ture was fixed on a glass slide ensuring that the fiber sensor was
always straight and the sensing section was slightly above the
slide to avoid any physical contact with the glass surface.
Figure 1(b) illustrates a schematic diagram of the RI meas-
uring system and a tapered fiber structure, also showing a
microscopic image of the taper waist for S-12.5. Light from a
broadband light source (Thorlabs S5FC1005S) 1450–1650 nm
is launched into the tapered SCSMF structure, while the trans-
mitted light is measured by an OSA (Agilent 86142B). The RI
liquid can be placed on the glass slide so that the tapered SCSMF
fiber section is immersed in the liquid. All tests were conducted
at room temperature.
Figure 2 shows three examples of the measured spectral re-
sponses for the S-12.5 sensor immersed in various liquids with
calibrated RI values. The RI value ranges are (a) 1.3405–
1.3463, (b) 1.3748–1.3837, and (c) 1.4304–1.4311. The spec-
tral dip moves monotonically toward longer wavelengths as the
SRI increases in every case. The S-18.8 and S-15.0 sensors have
different spectral responses, but the direction of the wavelength
shift with increasing SRI is the same. The spectral responses for
the S-18.8 and S-15.0 sensors are not shown here for the sake
of brevity.
The dip wavelength shifts versus RI change for all the sen-
sors are presented in Fig. 3, and summarized in detail in
Table 1. As expected, RI sensitivity increases as the tapered
waist diameter decreases. This is likely because of the fact that
smaller waist diameter sensors have a larger portion of the evan-
escent field exposed to the surrounding environment. Figure 3
also shows that in the higher RI range, all three samp les have
higher sensitivity. The likely explanation for this is that, as light
propagates in the fiber, it experiences a larger penetration depth
Fig. 1. Schematic diagram of (a) the microfiber tapering setup and
(b) the RI sensing system setup.
Fig. 2. Measured spectral responses of S-12.5 in different RI ranges:
(a) 1.3405–1.3463, (b) 1.3748–1.3837, and (c) 1.4304–1.4320.
Letter
Vol. 40, No. 17 / September 1 2015 / Optics Letters 4167
in a higher RI liquid, which again also increases the portion of
the evanescent wave exposed to the local environment.
As shown in Fig. 3 and Table 1, for all three samples, the
linear correlative coefficients R
2
are greater than 0.992,
which indicates that the wavelength shift exhibits good linear
relationship with the RI change over a small RI range. The
maximum sensitivities for S-18.8, S-15.0, and S-12.5 are
4722.9 nm∕RIU (RI 1.4249–1.4319), 8353.6 nm∕RIU
(RI 1.4256–1.4312), and 19212.5 nm∕ RIU (RI
1.4304–1.4320), respectively. Taking into account that the
OSA has a wavelength resolution of 0.01 nm, the RI sensor
based on S-12.5 has an RI resolution of 5.025 × 10
−7
which,
to the best of our knowledge, is significantl y higher than pre-
viously reported [16–20]. It is noted that, in this Letter, the
central wavelength of the spectral dip is determined as the 3 dB
mean wavelength, which is a more reliable parameter compared
to the central peak wavelength.
Although the proposed sensor has very high RI sensitivity,
this sensor suffers from the disadvantage of a narrow RI
measurement range because of its limited free spectral range
(typically 15 nm, depending on the RI measurement range).
One possible solution to overcome this problem is to combine
a relatively low sensitivity RI sensor (for example, an SCSMF
sensor without tapering) with the proposed tapered SCSMF.
The lower sensitivity RI sensor would then be used to deter-
mine the approximate RI range of the analyte. Then the appro-
priate sensor would be applied to measuring a highly accurate
RI value for the analyte.
It is noted that reducing the tapered waist diameter of such a
fiber sensor mi ght result in a higher RI sensitivity. However, a
tapered fiber with a smaller waist diameter is more fragile com-
pared to that of a larger waist diameter. Moreover, the tapered
SCSMF sensor with a smaller waist diameter has a smaller free
spectral range which results in a smaller RI measurement range.
By considering the trade-off between the measurement range,
sensitivity, and mechanical stability of the sensor, a minimum
tapered waist diameter of 12.5 μm was selected for our
experiments.
Most of the evanescent RH sensors based on optical fibers
reported to date require the coating of an additional humidity
sensitive (or hygroscopic) material on the surface of the fiber
sensor. Such humidity sensitive materials include polyvinyl
alcohol (PVA), polyimide (PI), poly(methyl methacrylate)
(PMMA), and nanoporous TiO
2
and SiO
2
films [21–23]. The
underlying operating principle of these types of RH sensors is
that the RI of the hygroscopic material coated on the surface of
the sensor changes in response to changes in the humidity.
The RI change results in a sensor spectral response variation,
and, hence, the RH can be determined. However, sensors
coated with humidity sensitive materials suffer from two main
disadvantages: (1) the coating of the fiber with a humidity
sensitive mat erial requires an additiona l fabrication step which
is difficult to control (usually because of the layer nonuniform-
ity, etc.); and (2) the coating materials have limited lifetimes
and are subject to contamination.
Dispensing with the need for a coating for an RH sensor is a
very useful development, and, given the increase in RI sensi-
tivity of the underlying sensor demonstrated already, here we
explore whether it is possible to implement an RH sensor with-
out the need for any additional humidity sensitive coatings on
the fiber surface.
A schematic diagram of the RH testing system is shown in
Fig. 4. In the experiment, an RI sensor with a tapered waist
diameter of 12.5 μm was placed in an RH chamber with nor-
mal air pressure (Electrotech Systems Inc., Model 5503-00
with Package F). Both the humidity and temperature can be
controlled by using this system. The maximum available reso-
lution of this RH control system is 0.1 RH%, and all tests were
carried out at a fixed temperature of 20.5 0.5°C.
As shown above, sample S-12.5 has the highest RI sensitivity,
and, hence, it was selected for this RH sensing demonstration.
In our experiments, the RH within the chamber was increased
gradually from 60% to 95% RH. Figure 5(a) shows the spectral
response of the sensor at different RH values. It is observed that,
as the RH increases from 60% to 95%, the central wavelength
of the spectral dip shifts to a longer wavelength monotonically.
The wavelength shift versus RH change is plotted in Fig. 5(b).
It is clear that the wavelength shift changes exponentially with
the increase of RH in the RH range, from 60.4% to 94.5%. The
total wavelength shift from 60.4% to 94.5% is circa 2.0 nm,
and the maximum sensitivity of 18.3 nm∕RHU was achieved
in the RH range of 90.4% to 94.5%. The RH sensitivity
achieved in this experiment is over four times higher than that
of the previously reported RH sensor, based on a bare fiber
structure without a hygroscopic coating layer [24].
As to the source of the RH-induced spectral shift, it is
known that water vapor can be adsorbed on a silica surface
Table 1. Sensor Sensitivities and Linear Fit Correlative Coefficients for Different Tapered Waist Diameters
Sensors S-18.8 S-15.0 S-12.5
RI Range 1.3333–
1.3468
1.3775–
1.3836
1.4249–
1.4319
1.3334–
1.3469
1.3749–
1.3835
1.4256–
1.4312
1.3405–
1.3463
1.3748–
1.3837
1.4304–
1.4320
Linear Fit Correlative
Coefficient
0.9956 0.9924 0.9968 0.9958 0.9989 0.9940 0.9949 0.9992 0.9990
Sensitivity (nm/RIU) 634.8 1313.1 4722.9 1038.5 1496.5 8353.6 1198.3 2123.6 19212.5
Fig. 3. Measured spectral wavelength shift versus RI for three differ-
ent tapered SCSMF-structure-based sensors: S-18.8, S-15.0, and
S-12.5.
4168 Vol. 40, No. 17 / September 1 2015 / Optics Letters
Letter
