Results in Physics 13 (2019) 102328

Contents lists available at ScienceDirect

Results in Physics
journal homepage: www.elsevier.com/locate/rinp

Highly sensitive open-channels based plasmonic biosensor in visible to near- T

infrared wavelength
⁎
Sanjida Akter , S.M. Abdur Razzak
Rajshahi University of Engineering & Technology, Department of Electrical & Electronic Engineering, Rajshahi 6204, Bangladesh

A R T I C LE I N FO A B S T R A C T

Keywords: In this paper, we numerically demonstrate a low-cost plasmonic refractive index sensor using two-sided open-
Finite element method channels that can operate in both visible and near-infrared region. Widely used finite element method (FEM)
Surface plasmon resonance software is employed to characterize the sensing properties. A thin coating of gold layer is used inside the open
Biosensors channels of the photonic crystal fiber (PCF) to create the plasmons. It is reported that the highest wavelength
Amplitude sensitivity
sensitivity of 5000 nm/RIU with a sensor resolution of 2.0 × 10−5 RIU can be achieved at an analyte refractive
Resolution
of 1.38. Besides, it is also achievable to have the highest amplitude sensitivity of 396 RIU−1. The proposed sensor
shows excellent linear characteristics with the highest figure of merit (FOM) of 47 RIU−1 between a refractive
index of 1.33 and 1.39. Due to structural simplicity and improved sensitivity, the proposed plasmonic sensor is
highly applicable in biological and biochemical analyte detections.

Introduction and permits lab-on-a-chip applications. Photonic crystal fibers (PCFs)

permit an excellent control of birefringence and leakage loss [11,12].
Surface plasmon resonance (SPR) is a distinct optical matter that By varying these guiding characteristics, modal properties can be ma-
can be potentially utilized to detect very small variation of refractive nipulated in such a way that eventually enables to control the sensing
index. Due to high sensitivity and level-free sensing properties, SPR performance effectively. Concerning structural simplicity and sensing
sensors have shown remarkable development in biological sample de- performance, PCF based sensors provide greater flexibility than prism
tection, medical diagnostics, environmental monitoring, food quality coupling SPR sensors. Moreover, by tuning the structural parameters of
control, organic-chemical sensing and antibody-antigen interactions the PCF, sensing performance can be improved significantly.
[1–7]. The idea of SPR was introduced by Zenneck [8] who solved Benefitting from the higher sensitive property of the PCF plasmonic
Maxwell equations to produce a wave equation that propagates along a sensor, a slight change of surrounding refractive index is possible to
surface and numerically demonstrated that electromagnetic waves in detect via large wavelength shift [13]. Sensing characteristics of sensors
the radio spectrum synthesize at the boundary of two media. Significant mostly depend on the choice of plasmonic materials. Several plasmonic
progress was not developed in SPR technology until 1957 when surface materials including gold, copper, silver, niobium, titanium dioxide, and
plasmon wave excitations were numerically proven to exist on the aluminum have been used in most of the existing works in the literature
metal surface by Ritchie [9]. Using prism coupling technique, SPR [14]. Although silver provides a sharp resonance peak, it deteriorates
sensor was proposed for the first time by Liedberg et al. [10] in 1983. the sensing performance due to chemical instability and oxidation issue
In order to create SPR effect, it is required to create an interaction of [12]. Graphene is a noble material that can be used as a protection layer
incident p-polarized wave with the plasmonic metal. Under the condi- of silver to remove the oxidization. Unfortunately, deposition of an
tion of wavelength matching of incident photons and surface electrons, additional layer on the silver increases the fabrication complexity as
stimulation of p-polarized wave occurs. As a result, free electrons are well as the overall cost. On the contrary, gold is an attractive plasmonic
emitted from the surface of the metal that eventually produce surface material that has high chemical stability and has no oxidation issue
plasmon wave and propagates along the metal–dielectric interface. [14].
Although prism coupling biosensors are commonly used, they are not Using the modal properties of PCFs, several noticeable SPR sensors
convenient for far distance sensing due to heavy configurations. Scaling have been reported. A triangular lattice-based plasmonic sensor was
down of the SPR sensors is essential since it allows low-cost fabrication proposed by An et al. [15] in order to detect unknown analytes having a

⁎
Corresponding author.
E-mail address: sanjida.ruet11@gmail.com (S. Akter).

https://doi.org/10.1016/j.rinp.2019.102328
Received 11 March 2019; Received in revised form 17 April 2019; Accepted 3 May 2019
Available online 08 May 2019
2211-3797/ © 2019 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

broad refractive indices. Using gold nanowires inside the cladding area,
it was possible to obtain low average wavelength sensitivity of 100 nm/
RIU over refractive index between 1.30 and 1.79. A novel hollow-core
PCF based refractive index sensor was numerically demonstrated by
Duan et al. [16], where six inner air-holes were coated using gold. This
sensor exhibited a spectral wavelength sensitivity of 5653 nm/RIU over
analyte refractive index from 1.27 to 1.45. Very recently, a new sensor
has been introduced by Rifat et al. [17], where a large cavity was in-
tentionally made inside the first ring of the PCF. Facilitated by stronger
interaction of guided light with a metallic cavity, this structure showed
the maximum wavelength sensitivity of 11,000 nm/RIU. In terms of
practical fabrication, getting a smooth metallic surface inside the inner
air-hole is relatively challenging. Recently, Liu et al. [18] proposed a
symmetrical dual D-shaped PCF sensor, where average spectral sensi-
tivity of 14,600 nm/RIU was reported. Recently, D-shaped PCF sensor
has been experimentally investigated [19], where effects of polishing
depth were demonstrated. In general, higher sensing performance can
be obtained from D-shaped PCF sensors. However, due to the require- Fig. 1. Representation of the open-channels PCF sensor by showing different
ment of an outer flat surface, D-shaped PCFs require additional pol- layers.
ishing, which is a time-consuming process. Recently, Hasan et al. pro-
posed a dual-polarized plasmonic sensor [20] and an external metal-
other parameters remained constant. After the optimization, we got the
coated sensor [21] using plasmonic gold and niobium, where the re-
following design parameters: Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ,
ported maximum wavelength sensitivities were 4600 and 8000 nm/
d1 = 0.5Λ, tg = 40 nm and r = 0.6Λ. To facilitate strong interaction
RIU, respectively. Very recently, the PCF sensor proposed in [22] was
with guided waves, plasmonic gold is placed on the inner surface of the
based on dual-core PCF. Using the facility of a single open channel, the
open channels. As illustrated in Fig. 1, two open-channels are made by
authors were able to obtain the highest spectral sensitivity of 4900 nm/
cutting the PCF in the left and rightmost direction. Gold is deposited
RIU. However, current fabrication technology does not advance enough
only the curvature of the channel, which is indicated by solid black
to fabricate such irregular PCF structures without extreme complexity.
color. Unlike an external metal coating sensor, it needs only fewer
Two open-ring channels based plasmonic sensor has been reported
amount of gold. This requirement has made this sensor cost-effective.
numerically having average wavelength sensitivity of 5500 nm/RIU
Fused silica is used as the background material. The wavelength
between analyte refractive index of 1.23 and 1.29 [23]. Recently, apart
dependent refractive index (RI) can be obtained by using the Sellmeier
from PCF based SPR sensors several promising plasmonic sensors have
equation, which is given by the following formula [28],
been reported [24–27].
Herein, we introduce a plasmonic sensor having two open-channels B1 λ2 B λ2 B λ2
n2 (λ ) = 1 + + 2 2 + 2 3
in the PCF structure. Comparing with existing PCF sensors, this sensor λ2 − C1 λ − C2 λ − C3 (1)
can act as a filter to detect the concentration of organic molecules. This
can be done by allowing the molecules having smaller concentration where n is the wavelength dependent refractive index of fused silica, λ
than the incision of the open channels. Moreover, noble plasmonic gold is the wavelength in µm. B1, B2, B3, C1, C2, and C3 are the Sellmeier
is placed inside the open channels of the structure rather than the entire constants. For fused silica, the constants are 0.69616300, 0.407942600,
outer surface, which requires less amount of gold. It is shown that the 0.897479400, 0.00467914826, 0.0135120631 and 97.9340025, re-
radius of the open-channel can be adjusted according to the require- spectively.
ment of analyte volume. We also investigate the effects of varying the The complex RI of gold is taken from the following Drude-Lorentz
thickness of gold, the radius of open channels, the diameter of air-holes model [29],
and pitch. The proposed plasmonic sensor can be implemented in ωD2 Δε ·Ω2L
visible and near-infrared (NIR) wavelength. Finally, a typical experi- εAu = ε∞ − −
ω (ω + jγD ) (ω − Ω2L) + jΓL ω
2
(2)
mental setup is considered using the proposed sensor.
where εAu is the permittivity of gold, ε∞ is the permittivity at high
Design and theoretical model frequency with the value of 5.9673, ω is the angular frequency, which is
given by ω = 2πc/λ, c is the velocity of light in vacuum, ωD is the
Fig. 1 presents the two dimensional (2D) view of the open-channels plasma frequency and γD is the damping frequency. Here, ωD/
plasmonic sensor consisting of hexagonal lattice air-holes with three 2π = 2113.6 THz and γD/2π = 15.92 THz and the weighting factor
rings (Nr = 3). Four air-holes in the second (two air-holes) and third Δε = 1.09. The spectral width and oscillator strength of the Lorentz
ring (two air-holes) are omitted intentionally. Missing air-holes are used oscillators are given by ΓL/2π = 104.86 THz and ΩL/2π = 650.07 THz,
to create birefringence that assists to enhance the sensitivity of a par- respectively.
ticular core-guided mode. The central air-hole is used in order to con- One of the challenging tasks is to deposit a uniform gold layer on the
trol the evanescent field effectively. Moreover, four air-holes near the inner open-channels of the PCF. Several deposition methods such as
open-channels are kept relatively smaller in order to keep sufficient sputtering [30], and wet chemistry deposition [31] are available for the
space for enlarging the radius of the open-channels. If the dimension of deposition of gold. However, accuracy and precision of these conven-
such air-holes is remained same as other air-holes (except central air- tional coating methods are limited due to excessive surface roughness.
hole), they would touch the open-channels for an enlarged radius of the Due to small radius of the open-channels, chemical vapor deposition
open-channels. (CVD) may not be used effectively to achieve a homogeneous coating
We define the central distance between air-holes of the adjacent ring [32]. However, the atomic layer deposition (ALD) can be employed to
by Λ, central air-hole diameter by dc, smaller air-hole diameter by ds, get smooth surface of the gold [33]. Another critical issue of the pro-
gold layer thickness by tg, large air-hole diameter by d1 and open- posed sensor is the complexity during practical fabrication. It is possible
channel radius by r. To find the optimized value of a particular para- to fabricate hexagonal-lattice PCF using the conventional stack-and-
meter, we simulated the structure with several values while keeping the draw technique [34]. The open-channels can be obtained by first

2
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

fabricating the PCF having the complete structure of open-channel air evanescent wave near the plasmonic metal than the fundamental x-
holes and then polish out the left and right sides of the PCF. The polarized mode. As a result, it can strongly interact with the gold sur-
electromagnetic modes have been calculated using the full-vector finite face that enhances the modal loss. Fig. 2(d) illustrates the situation of
element method (FV-FEM) based COMSOL Multiphysics v.5.0 software. the occurring resonance condition. According to this figure, the fun-
We set appropriate boundary conditions to prevent the backscattering. damental y-polarized mode presents higher and narrow loss spectrum
In our simulation, a perfectly matched layer (PML) having thickness of compared with the x-polarized mode. Due to the highly sensitive
3 µm was implemented which was taken after the convergence test. An characteristic of y-polarized mode, we only consider this in our whole
analyte layer of thickness 2.5 µm was also included during the simu- analysis. According to Fig. 2(d), neff of the fundamental y-polarized
lation. We have observed that meshing elements have a significant ef- mode and the y-polarized plasmonic mode is identical at the wave-
fect on the guided modes. Bigger mesh element size results in a reduced length of 0.70 μm. This wavelength is called the resonant wavelength. A
number of triangular elements, which require less computational time; sharp peak of the confinement loss is appeared at this wavelength.
however, it may provide inaccurate results. Considering this fact, we In general, the confinement loss peak method is used to assess the
used finer element mesh size so that it can map smaller air-holes and performance of a plasmonic sensor. More confinement loss means more
provide reliable results. evanescent field in the cladding region that improves the sensitivity. We
can calculate the confinement loss from the equation below [14],
Simulation results and discussion
α (dB / cm) = 8.686 × ko Im(neff ) × 10 4 (3)
2π
Generation of SPR effect in a PCF relies on the strength of the in- where Im(neff ) is the imaginary part of effective RI and ko = is the
λ
cident evanescent fields and amount of emitted surface electrons. When wave number.
fundamental mode propagates in the core region, evanescent waves A highly sensitive sensor should detect a very small change of
spread in the cladding of the PCF. Under the resonance condition, i.e. analyte RI. To get this property, neff of the fundamental mode should be
when the real part of the effective refractive index (neff) of the core- extremely influenced by the analyte RI (na). This particular case is il-
guided mode and surface plasmon polariton (SPP) mode is equal, sur- lustrated in Fig. 3. It illustrates shifting of resonance wavelength when
face plasmon wave (SPW) is created. At the resonance wavelength, the na is varied from 1.33 to 1.39. It can be evident that increasing value of
generation of SPW is maximized which can be observed by the sharp na shifts the real part of neff in Fig. 3(a) towards higher value. For
peak in the loss spectrum of the fundamental mode. A slight change of changing the value of na from 1.33 to 1.34, 1.34 to 1.35, 1.35 to 1.36,
refractive index of the sensing medium can result in a significant shift of 1.36 to 1.37, 1.37 to 1.38, 1.38 to 1.39, resonance wavelength shifts to
resonance wavelength. This property is utilized to detect an analyte 0.59 to 0.61 μm, 0.61 to 0.63 μm, 0.63 to 0.65 μm, 0.65 to 0.67 μm,
sample. 0.67 to 0.70 μm and 0.70 to 0.75 μm, respectively. As observed in
Fig. 2(a)–(c) show distribution of electric field in the fundamental x- Fig. 3(a), loss peak is increased with higher na. We found the minimum
polarized mode, y-polarized mode and y-polarized plasmonic mode, confinement loss of 103.45 dB/cm at 0.59 μm for na of 1.33. Besides, the
respectively. As seen in the design of the PCF, there are four missing air- highest peak loss appeared at 0.75 μm which is 375.85 dB/cm when na
holes in the core that induce birefringence. In practice, birefringence in of 1.39.
a PCF allows a specific fundamental mode to interact strongly with In general, the confinement loss peak method uses wavelength in-
metal. This particular case can be illustrated using Fig. 2(a) and (b). It terrogation technique to calculate the sensitivity. Using this technique,
can be evident that the fundamental y-polarized mode has higher wavelength sensitivity can be expressed as [20],
Sλ (nm / RIU ) = Δλpeak /Δna (4)
where Δλpeak denotes the difference between two successive resonance
wavelength and Δna denotes the variation of na. In this plasmonic
sensor, we observed Δλpeak of 20 nm, 20 nm, 20 nm, 20 nm, 30 nm,
50 nm when na is varied from 1.33 to 1.34, 1.34 to 1.35, 1.35 to 1.36,
1.36 to 1.37, 1.37 to 1.38 and 1.38 to 1.39, respectively. Using Eq. (4),
we found the corresponding wavelength sensitivities of 2000, 2000,
2000, 2000, 3000 and 5000 nm/RIU. Additionally, an average wave-
length sensitivity of 2666.67 can be achieved between na of 1.33 and
1.39. This sensitivity is comparable with previously proposed results
[12,20,35–43]. Another key parameter of a sensor is the resolution that
can be calculated by the following formula [20],
R (RIU ) = Δna × Δλ min /Δλpeak (5)
In our simulation, we varied na from 1.33 to 1.34, 1.34 to 1.35 and
so on, which indicate Δna of 0.01. With Δλmin = 0.1 nm and
Δλpeak = 50 nm, the maximum sensor resolution (wavelength inter-
rogation) of 2.0 × 10−5 RIU can be found for analyte RI at 1.38.
Wavelength interrogation method associates with spectral manip-
ulation, which is costly and challenging to implement. On the contrary,
amplitude sensitivity can be calculated at a particular wavelength,
which is expressed by [21],
1 ∂α (λ, na )
SA = −
α (λ, na ) ∂na (6)
Fig. 2. Distribution of electric field in (a) the fundamental x-polarized mode,
(b) the fundamental y-polarized mode, (c) y-polarized plasmonic mode at where α (λ,na) is the overall propagation loss and ∂α (λ, na) is the loss
0.70 μm and (d) illustration of the phenomena of resonance condition. Other difference. Fig. 3(b) illustrates the dependence on amplitude sensitivity
parameters are fixed at Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ, when na is changed from 1.33 to 1.38. Using optimum design para-
na = 1.38 and r = 0.6Λ. meters, we observed the maximum sensitivity of 396 RIU−1 at 0.70 µm

3
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

Fig. 3. (a) Dependence of loss spectrum when na is varied from 1.33 to 1.39 and (b) variation of amplitude sensitivity when na is changed from 1.33 to 1.38. Other
parameters are set to Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ, tg = 40 nm, and r = 0.6Λ.

Table 1 5000 nm/RIU for tg of 30, 40 and 50 nm, respectively. Variation of

Performance comparison for varying analyte RI from 1.33 to 1.39. amplitude sensitivity for different tg is shown in Fig. 4(b). Amplitude
na Δna Peak loss λpeak Δλpeak Sλ R (RIU) SA (RIU−1)
sensitivities about 448, 396 and 296 RIU−1 can be obtained for the tg of
(dB/cm) (nm) (nm) (nm/ 30 nm, 40 nm, and 50 nm, respectively.
RIU) We now analyze the effect of changing pitch values on confinement
loss and amplitude sensitivity. Variations of loss spectrum and ampli-
1.33 0.01 103.45 590 20 2000 5.0 × 10−5 60
tude sensitivity for scaling up of pitch from 1.9 to 2.1 μm are shown in
1.34 0.01 126.75 610 20 2000 5.0 × 10−5 74
1.35 0.01 128.59 630 20 2000 5.0 × 10−5 95 Fig. 5(a) and (b). It can be seen that although confinement loss is re-
1.36 0.01 182.11 650 20 2000 5.0 × 10−5 130 duced significantly due to the scaling up of the pitch values, there is no
1.37 0.01 232.59 670 30 3000 3.33 × 10−5 185 noticeable variation of the amplitude sensitivity. Extending the pitch
1.38 0.01 299.02 700 50 5000 2.0 × 10−5 396
value from 1.9 to 2.1 μm causes a reduction of peak loss from 332.38 to
1.39 0.01 375.85 750 – – – –
267.79 dB/cm at na of 1.38. Besides, a slight red shift (resonant wave-
length shifts towards longer wavelength) can be observed when pitch
when na of 1.38. Besides, with na of 1.33, 1.34, 1.35, 1.36 and 1.37 the value is enlarged. For the pitch value of 1.9, 2.0 and 2.1 μm, we found
corresponding amplitude sensitivities are 60, 74, 95, 130 and amplitude sensitivities of 402, 396 and 387 RIU−1 respectively. Be-
185 RIU−1, respectively. Moreover, assuming 1% change of transmitted sides, we observed wavelength sensitivities of 5000, 5000 and
intensity a maximum sensor resolution (amplitude interrogation) of 5000 nm/RIU when pitch values are 1.9, 2.0 and 2.1 μm, respectively.
2.53 × 10−5 RIU is achievable at analyte RI of 1.38. Table 1 shows the We also demonstrate the effects of the varying diameter of the
summary of several sensing parameters including wavelength sensi- central air-hole (dc) and large air-holes of the PCF. Loss spectrum and
tivity, amplitude sensitivity, and sensor resolution at different analyte amplitude sensitivity for dc of 0.10Λ, 0.15Λ, 0.20Λare shown in
RI. Fig. 6(a) and (b). As illustrated in Fig. 6(a), bigger diameter of central
Fig. 4(a) depicts the dependence on loss spectrum for different gold air-hole results in an increase of the confinement loss. Moreover, we
layer thickness (tg = 30, 40 and 50 nm) with na of 1.38 and 1.39. As found a tiny blue shift when dc is enlarged. However, changing dc has a
illustrated in the figure, loss depth notably reduces with thicker gold negligible effect on the amplitude sensitivity. The calculated amplitude
layer while loss depth increases with the thinner gold layer. According sensitivities are 388, 396 and 398 RIU−1 for the air-hole diameter of
to Fig. 4(a), we found maximum peak loss depth about 642.56 and 0.10Λ, 0.15Λ, 0.20Λ, respectively.
174.01 dB/cm for tg of 30 and 50 nm with na of 1.39. The reduction of Similarly, we increased the diameter of large air-holes (d1) from
peak loss is due to the higher damping of a thicker gold layer. With an 0.45Λ to 0.55Λ and plotted the respective loss spectrum and amplitude
analyte RI of 1.38, we found wavelength sensitivities of 4000, 5000 and sensitivity. It can be seen that there is insignificant change of confine-
ment loss due to the change of d1 from 0.45Λ to 0.55Λ, which is shown

Fig. 4. (a) Dependence on loss for different tg and (b) variation of amplitude sensitivity with wavelength for various tg. Other parameters are set to Λ = 2.0 μm,
dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ, na = 1.38, and r = 0.6Λ.

4
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

Fig. 5. (a) Effect on the loss depth due to scaling up of the pitch from 1.9 to 2.1 µm and (b) variation of amplitude sensitivity with wavelength for analyte RI of 1.38.

Fig. 6. (a) Dependence of confinement loss due to enlarging the diameter of central air-hole from 0.10Λ to 0.20Λ and (b) amplitude sensitivity for changing the
diameter of central air-hole with an analyte RI of 1.38.

Fig. 7. (a) Loss spectrum for enlarging the large air-hole diameter from 0.45Λ to 0.55Λ and (b) amplitude sensitivity for changing diameter of large air-hole with an
analyte RI of 1.38.

in Fig. 7(a). The variation of amplitude sensitivity for varying diameter and 0.65Λ, respectively. Since amplitude sensitivity is slightly depen-
of large air-holes is shown in Fig. 7(b). The calculated sensitivities are dent on channel radius, the volume of an analyte can be adjusted by
378, 396 and 411 RIU−1 for the air-hole diameter of 0.45Λ, 0.50Λ and simply varying the channel radius. In this case, we found wavelength
0.55Λ, respectively. Moreover, wavelength sensitivities of 4000, 5000 sensitivities of 5000, 5000 and 5000 nm/RIU for open-channel radius of
and 4000 nm/RIU are obtained when d1 are 0.45Λ, 0.50Λ and 0.55Λ, 0.55Λ, 0.60Λ and 0.65Λ, respectively.
respectively. In practice, the value of the structural parameters can be deviated
Likewise, we consider the effect of changing the radius of the open- up ± 2% to their optimum values during fabrication [44]. In this study,
channels (r) on the loss spectrum which is depicted in Fig. 8(a). As seen we consider fabrication tolerance of ± 5% and ± 10% to their optimum
in the figure, when the radius of open channels is increased from 0.55Λ values. The effects of ± 5% and ± 10% deviation of tg and pitch value
to 0.65Λ, peak loss reduces from 309.67 to 291.30 dB/cm. Variation of on the sensitivity are shown in Fig. 9(a) and (b). As clearly seen in
the open-channel radius has a negligible effect on the amplitude sen- Fig. 9(a), deviation of gold layer thickness by ± 5% and ± 10% results
sitivity, which is depicted in Fig. 8(b). Amplitude sensitivities of 384, in a significant variation of amplitude sensitivity. Therefore, careful
396 and 400 RIU−1 can be found for channel radius of 0.55Λ, 0.60Λ attention should be paid during the deposition of the gold layer on the

5
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

Fig. 8. (a) Loss spectrum for enlarging the radius of the open channel 0.55Λ to 0.65Λ and (b) amplitude sensitivity for changing the radius of the open channel with
an analyte RI of 1.38.

Fig. 9. (a) Fabrication tolerance due to the deviation of gold layer thickness by ± 5% and ± 10% of their optimum value and (b) fabrication tolerance due to the
deviation of pitch value by ± 5% and ± 10% of their optimum value.

Fig. 10. (a) Effect of varying the PML thickness on the real part of the effective refractive and confinement loss and (b) variation of amplitude sensitivity with PML
thickness. PCF parameters are set to Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ, na = 1.38, r = 0.6Λ and λ = 0.6 μm.

inner surface of the open channels. On the other hand, ± 5% and ± larger PML thickness could be used, it would unexpectedly lengthen the
10% deviation of pitch value has no noticeable effect on the amplitude simulation time. Therefore, optimal PML thickness of 3 μm was used in
sensitivities. order to keep a balance between simulation speed and accuracy of the
In order to observe the effect of changing the PML thickness on the simulation results.
real part of the effective refractive index (Real (neff)), confinement loss A good sensor should represent high linear characteristics. Fig. 11
and amplitude sensitivity, we have changed the PML thickness from 0.5 shows the linear curve fitting by varying na from 1.33 to 1.39. The
to 5 μm. The results are shown in Fig. 10. We found that increasing the progressive equation for linear fitting can be written as y = 2.142x –
PML thickness results in an extremely small change of Real (neff). As 2.257, where resonance wavelength is given by y and RI of the analyte
illustrated in Fig. 10(a), the variation of Real (neff) is about order is given by x. Considering the entire RI range, the R–square value is
of ± 10−6. Besides, confinement loss does not vary significantly due to observed about 0.9780 which indicates improved linearity. It is ex-
negligible variation of Im (neff). Additionally, amplitude sensitivity as a pected that such high linear characteristics can be used for biochemical
function of PML thickness is depicted in Fig. 10(b). It can be evident detection.
from the figure that amplitude sensitivity remains almost constant For an improved sensing performance, the signal-to-noise ratio of a
when PML thickness is sufficiently varied. Such insignificant change of sensor (SNR) is a crucial factor. In practice, higher SNR results in a
the sensitivity does not affect the performance of the sensor. Although small standard deviation to the change of spectra. Moreover, the

6
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

denotes the full-width at half-maximum. Fig. 12 shows FOM of the

proposed sensor for the variation of na with the gold layer thickness of
40 nm. We found that FOM increases from 29 to 47 RIU−1 when na is
varied from 1.33 to 1.39.
Table 2 shows a comparison of the proposed sensor with existing
plasmonic sensors in the literature. We compare several aspects such as
structure of the sensors, figure of merit, RI range, wavelength sensi-
tivity and wavelength resolution. It can be concluded from the table
that the performance of the proposed sensor is comparable with the
Refs. [12,13,16,20–22,41], however, it shows lower FOM then existing
sensors. Moreover, wavelength sensitivity of nanoscale hyperbolic
metamaterial [46] is comparable with the proposed sensor. The de-
signed sensor exhibits lower wavelength sensitivity and FOM than hy-
perbolic metamaterials based plasmonic sensors [47,48].

Experimental consideration
Fig. 11. Linear fitting characteristics of the proposed sensor with na from 1.33
to 1.39. Other parameters are Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ,
Fig. 13 illustrates the typical experimental setup of a sensing system
tg = 40 nm and r = 0.6Λ.
using the proposed sensor. Initially, a broadband source is used to
produce the incident light, which is transmitted through the polarizer.
The polarizer controller converts the broadband light into a linearly
polarized light (i.e., p-polarized). The output of polarizer controller is
then coupled with a single mode fiber followed by the proposed sensor.
After interacting with the sensing medium, the output is sent through
another single mode fiber. It is worthy to mention that IN and OUT of
the analyte sample can be maintained through a continuous pump. The
transmission spectrum is measured by the optical spectrum analyzer
(OSA). The recorded data are analyzed by the computer to measure the
sensing performance. In the wavelength interrogation method, by
measuring the variation of resonance wavelength with the analyte RI is
used to measure the sensing performance. In intensity-based method,
sensitivity can be obtained by measuring the change of resonance in-
tensity with the analyte RI. In this case, when the analyte RI changes
from na to na + Δna, the resonance amplitude changes accordingly.

Conclusion
Fig. 12. FOM of the proposed sensor with the variation of na. Other parameters
are kept constant as Λ = 2.0 μm, dc = 0.15Λ, ds = 0.25Λ, d1 = 0.5Λ, Two open-channels based SPR biosensor has been numerically stu-
tg = 40 nm and r = 0.6Λ. died. The open-channels can contain a sufficient volume of the analyte,
which is much easier than pouring analyte on the whole exterior sur-
face of the sensor. Besides, it requires less amount of gold layer coating,
improvement of the SNR and spectral width will promote to a better
which reduces the overall fabrication cost. FEM based analysis showed
detection limit, which can be evaluated by the figure of merit (FOM).
that maximum spectral sensitivity of 5000 nm/RIU and amplitude
The FOM can be expressed by [45],
sensitivity 396 RIU−1 can be attained at an analyte RI of 1.38.
S (nm / RIU ) Additionally, it exhibited excellent linear characteristics and improved
FOM = FOM within RI ranging from 1.33 to 1.39. Dependence of sensing
FWHM (nm) (7)
performance on the variation of all design parameters has been also
where S denotes the slops of the resonant wavelength and FWHM investigated. It is anticipated that the proposed plasmonic sensor can be

Table 2
Comparison of the proposed sensor with existing plasmonic sensors in the literature.
Ref. Structure of the SPR sensor Figure of merit RI range Wavelength sensitivity (nm/ Resolution (RIU)
(RIU−1) RIU)

[12] Two layer hexagonal lattice PCF with external plasmonic gold – 1.33–1.36 4000 2.5 × 10−5
[13] Two layer circular lattice PCF with external plasmonic silver – 1.33–1.37 4200 2.38 × 10−5
[16] Hollow-fiber based sensor – 1.27–1.45 5653.57 –
[18] Symmetrical dual D-shape PCFs 250 1.36–1.41 14,660 6.82 × 10−6
[20] Dual-polarized spiral PCF with external plasmonic gold – 1.33–1.38 4600 2.17 × 10−5
[21] Two layer hexagonal lattice PCF with external plasmonic 171.5 1.36–1.40 8000 1.25 × 10−5
niobium
[22] Dual-core microstructured optical fiber with plasmonic gold – 1.44–1.46 4900 –
[41] Two layer hexagonal lattice PCF with external plasmonic gold – 1.33–1.36 4000 2.5 × 10−5
[47] Hyperbolic metamaterials 590 1.333–1.3336 30,000 –
[46] Nanoscale hyperbolic metamaterial – 1.33–1.364 3450 –
[48] Hyperbolic metamaterials – 1.333–1.3336 30,000 –
Proposed work Two open-channels with plasmonic gold 47 1.33–1.38 5000 2.0 × 10−5

7
S. Akter and S.M. Abdur Razzak Results in Physics 13 (2019) 102328

Fig. 13. Schematic of the sensing system in the experiment.

used in biological sample detection in a lab-on-a-chip platform. plasmon polariton and dielectric waveguide modes. Photonics Res
2016;4(6):262–6.
[26] Wu L, You Q, Shan Y, et al. Few-layer Ti3C2Tx MXene: a promising surface plasmon
References resonance biosensing material to enhance the sensitivity. Sens Actuat B
2018;277:210–5.
[1] Berger CE, Greve J. Differential SPR immunosensing. Sens Actuat B [27] Ruan B, Guo J, Wu L, et al. Ultrasensitive terahertz biosensors based on Fano re-
2000;63(1):103–8. sonance of a graphene/waveguide hybrid structure. Sensors 2017;17(8):1924.
[2] Zhao Y, Deng Z-Q, Li J. Photonic crystal fiber based surface plasmon resonance [28] Sellmeier W. Zur Erkarung der abnormen Farbenfolge im Spectrum einiger.
chemical sensors. Sens Actuat B 2014;202:557–67. Substanzen. Annalen der Physik und Chemie 1871;219:272–82.
[3] Su Y-D, Chen S-J, Yeh T-L. Common-path phase-shift interferometry surface [29] Vial A, Grimault A-S, Macías D, et al. Improved analytical fit of gold dispersion:
plasmon resonance imaging system. Opt Lett 2005;30(12):1488–90. application to the modeling of extinction spectra with a finite-difference time-do-
[4] Jorgenson R, Yee S. A fiber-optic chemical sensor based on surface plasmon re- main method. Phys Rev B 2005;71(8):085416.
sonance. Sens Actuat B 1993;12(3):213–20. [30] Armelao L, Barreca D, Bottaro G, et al. RF-sputtering of gold on silica surfaces:
[5] Fang Y. Label-free cell-based assays with optical biosensors in drug discovery. Assay evolution from clusters to continuous films. Mater Sci Eng C
Drug Dev Technol 2006;4(5):583–95. 2005;25(5–8):599–603.
[6] Fang S, Lee HJ, Wark AW, et al. Attomole microarray detection of microRNAs by [31] Sioss JA, Keating CD. Batch preparation of linear Au and Ag nanoparticle chains via
nanoparticle-amplified SPR imaging measurements of surface polyadenylation re- wet chemistry. Nano Lett 2005;5(9):1779–83.
actions. J Am Chem Soc 2006;128(43):14044–6. [32] Sazio PJ, Amezcua-Correa A, Finlayson CE, et al. Microstructured optical fibers as
[7] Homola J. Surface plasmon resonance sensors for detection of chemical and bio- high-pressure microfluidic reactors. Science 2006;311(5767):1583–6.
logical species. Chem Rev 2008;108(2):462–93. [33] Griffiths MB, Pallister PJ, Mandia DJ, et al. Atomic layer deposition of gold metal.
[8] Zenneck J. Über die Fortpflanzung ebener elektromagnetischer Wellen längs einer Chem Mater 2015;28(1):44–6.
ebenen Leiterfläche und ihre Beziehung zur drahtlosen Telegraphie. Ann Phys [34] Knight J, Birks T, Russell PSJ, et al. All-silica single-mode optical fiber with pho-
1907;328(10):846–66. tonic crystal cladding. Opt Lett 1996;21(19):1547–9.
[9] Ritchie R. Plasma losses by fast electrons in thin films. Phys Rev 1957;106(5):874. [35] Hameed MFO, Alrayk YK, Shaalan AA, et al. Design of highly sensitive multichannel
[10] Liedberg B, Nylander C, Lunström I. Surface plasmon resonance for gas detection bimetallic photonic crystal fiber biosensor. J Nanophotonics 2016;10(4):046016.
and biosensing. Sensors Actuat 1983;4:299–304. 046016.
[11] Rifat AA, Ahmed R, Yetisen AK, et al. Photonic crystal fiber based plasmonic sen- [36] Li D, Zhang W, Liu H, et al. High sensitivity refractive index sensor based on
sors. Sens Actuat B 2017;243:311–25. multicoating photonic crystal fiber with surface plasmon resonance at near-infrared
[12] Rifat AA, Hasan MR, Ahmed R, et al. Photonic crystal fiber-based plasmonic bio- wavelength. IEEE Photonics J 2017;9(2):1–8.
sensor with external sensing approach. J Nanophotonics 2018;12503:1. [37] Liu C, Wang F, Lv J, et al. A highly temperature-sensitive photonic crystal fiber
[13] Momota MR, Hasan MR. Hollow-core silver coated photonic crystal fiber plasmonic based on surface plasmon resonance. Opt Commun 2016;359:378–82.
sensor. Opt Mater 2018;76:287–94. [38] Luan N, Yao J. A hollow-core photonic crystal fiber-based SPR sensor with large
[14] Rifat AA, Hasan MR, Ahmed R, et al. Microstructured optical fiber-based plasmonic detection range. IEEE Photonics J 2017.
sensors. Computational photonic sensors. Springer; 2019. p. 203–32. [39] Otupiri R, Akowuah E, Haxha S, et al. A novel birefrigent photonic crystal fiber
[15] An G, Li S, Yan X, et al. Extra-broad photonic crystal fiber refractive index sensor surface plasmon resonance biosensor. IEEE Photonics J 2014;6(4):1–11.
based on surface plasmon resonance. Plasmonics 2017;12(2):465–71. [40] Otupiri R, Akowuah EK, Haxha S. Multi-channel SPR biosensor based on PCF for
[16] Duan L, Yang X, Lu Y, et al. Hollow-fiber-based surface plasmon resonance sensor multi-analyte sensing applications. Opt Express 2015;23(12):15716–27.
with large refractive index detection range and high linearity. Appl Opt [41] Rifat A, Mahdiraji GA, Sua Y, et al. Surface plasmon resonance photonic crystal
2017;56(36):9907–12. fiber biosensor: a practical sensing approach. IEEE Photon Technol Lett
[17] Rifat AA, Ahmed R, Mahdiraji GA, et al. Highly sensitive selectively coated photonic 2015;27(15):1628–31.
crystal fiber-based plasmonic sensor. Opt Lett 2018;43(4):891–4. [42] Rifat AA, Mahdiraji GA, Ahmed R, et al. Copper-graphene-based photonic crystal
[18] Liu C, Su W, Liu Q, et al. Symmetrical dual D-shape photonic crystal fibers for fiber plasmonic biosensor. IEEE Photonics J 2016;8(1):1–8.
surface plasmon resonance sensing. Opt Express 2018;26(7):9039–49. [43] Rifat AA, Mahdiraji GA, Chow DM, et al. Photonic crystal fiber-based surface
[19] Xie Q, Chen Y, Li X, et al. Characteristics of D-shaped photonic crystal fiber surface plasmon resonance sensor with selective analyte channels and graphene-silver de-
plasmon resonance sensors with different side-polished lengths. Appl Opt posited core. Sensors 2015;15(5):11499–510.
2017;56(5):1550–5. [44] Hasan MR, Hasan MI, Anower MS. Tellurite glass defect-core spiral photonic crystal
[20] Hasan MR, Akter S, Rifat AA, et al. Spiral photonic crystal fiber-based dual-polar- fiber with low loss and large negative flattened dispersion over S+ C+ L+ U
ized surface plasmon resonance biosensor. IEEE Sens J 2018;18(1):133–40. wavelength bands. Appl Opt 2015;54(32):9456–61.
[21] Hasan MR, Akter S, Ahmed K, et al. Plasmonic refractive index sensor employing [45] Fan Z, Li S, Liu Q, et al. High sensitivity of refractive index sensor based on analyte-
niobium nanofilm on photonic crystal fiber. IEEE Photonics Technol Lett filled photonic crystal fiber with surface plasmon resonance. IEEE Photonics J
2018;30(4):315–8. 2015;7(3):1–9.
[22] Luan N, Han H, Zhao L, et al. Opening up dual-core microstructured optical fiber- [46] Baqir M, Farmani A, Fatima T, et al. Nanoscale, tunable, and highly sensitive bio-
based plasmonic sensor with large detection range and linear sensitivity. Opt Mater sensor utilizing hyperbolic metamaterials in the near-infrared range. Appl Opt
Express 2019;9(2):819–25. 2018;57(31):9447–54.
[23] Liu C, Yang L, Lu X, et al. Mid-infrared surface plasmon resonance sensor based on [47] Sreekanth KV, Alapan Y, ElKabbash M, et al. Extreme sensitivity biosensing plat-
photonic crystal fibers. Opt Express 2017;25(13):14227–37. form based on hyperbolic metamaterials. Nat Mater 2016;15(6):621.
[24] Wu L, Guo J, Wang Q, et al. Sensitivity enhancement by using few-layer black [48] Sreekanth KV, ElKabbash M, Alapan Y, et al. Hyperbolic metamaterials-based
phosphorus-graphene/TMDCs heterostructure in surface plasmon resonance bio- plasmonic biosensor for fluid biopsy with single molecule sensitivity. EPJ Appl
chemical sensor. Sens Actuat B 2017;249:542–8. Metamater 2017;4:1.
[25] Wu L, Guo J, Xu H, et al. Ultrasensitive biosensors based on long-range surface