Optik - International Journal for Light and Electron Optics 321 (2025) 172177

Contents lists available at ScienceDirect

Optik
journal homepage: www.elsevier.com/locate/ijleo

Ultra-high-sensitive SPR device composed of CaF2 prism, Ag,

graphene, and sensing medium for refractive index
detection-artificial neural network training
Imed Sassi *, Mariem Zbidi , Mounir Ben El Hadj Rhouma
Université de Monastir, Institut Préparatoire aux Études des Ingénieurs de Monastir, Laboratoire d’Études des Milieux Ionisés et Réactifs (EMIR),
Monastir 5019, Tunisia

A R T I C L E I N F O A B S T R A C T

Keywords: The current work focuses on a refractive index (RI) biosensor using prism configuration used for
CaF2 Prism various sensing applications. This design is suggested for evaluating the intensities of the
Graphene reflectance for visible and near infrared wavelength range at fixed incident direction (=80 de­
Refractive index sensitivity
gree). Different materials used to realize our biosensor, the CaF2 prism, the silver (Ag), and the
Temperature sensitivity
Wavelength interrogation
graphene. A numerical method based on the electromagnetic theory is employed to compute the
Artificial neural network reflectance property. Significant results concerning the sensitivity and reflection depth are
showed after optimizing some geometrical parameters. The parameters used to examine the
biosensor’s performance are the RI sensitivity (SRI), figure of merit (FoM), and quality factor (QF).
The higher values of SRI, FoM, and QF provided by our suggested structure are 54600 nm/RIU,
151 /RIU-1, and 6.29 respectively, when the RI is ranging from 1.33 to 1.39. In this research, for
input parameters of our sensor, we predict the sensitivity’s data by using the artificial neural
network technique. It is remarked that the latest approach shows a significant results compared to
that obtained by the theoretical method. In the context of temperature sensing, our device showed
an average temperature sensitivity of 6.99 nm/K, and a maximum, sensitivity of 17.68 nm/K.

1. Introduction

Surface plasmons (SPs) have emerged as an intriguing area of investigation, drawing considerable attention due to their unique
ability to manipulate interactions between light and matter at the nanoscale. SPs represent collective oscillations of electrons that
occur along the interface metal-dielectric. They are generated when incident photons interact with free electrons on a metal surface.
This interaction leads to the resonance condition, resulting in a sharp decrease in the intensity of reflected light at a specific wavelength
known as the resonance wavelength, influenced by the refractive index of surrounding medium. When a transverse magnetic (TM)
polarized electromagnetic wave illuminates the prism and reflects off the different layers deposited on it, surface plasmons are excited,
leading to electron oscillations. The surface plasmon wave (SPW) reacts to variations in the RI of the sensing medium. A structure based
on the SF11 prism, the silver (Ag), Zinc oxide (ZnO), antimonene, and sensing medium have been investigated for refractive index RI in
the range of 1.33–1.631 [1]. The Surface Plasmon Resonance (SPR) curve exhibits exceptional sensitivity to changes in the RI of the
sensor interface. High sensitivity is crucial for optimal parameters of an SPR biosensor. Moreover, SPs offer promising prospects for

* Corresponding author.
E-mail address: [email protected] (I. Sassi).

https://doi.org/10.1016/j.ijleo.2024.172177
Received 17 October 2024; Received in revised form 9 December 2024; Accepted 13 December 2024
Available online 15 December 2024
0030-4026/© 2024 Elsevier GmbH. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
I. Sassi et al. Optik 321 (2025) 172177

applications in sensing field [2–7].

In this study, SPs are excited using the Kretschmann configuration. In this configuration, a thin metal film is applied to the surface
of the prism. TM-polarized light is directed onto one side of the prism at a specific wavelength and at an angle exceeding the critical
angle of the interface between the prism and the metal coating. The Kretschmann configuration is a straightforward and easily
implemented setup that is widely employed for SPR experiments, wherein a thin layer of metal, typically gold or silver is coated onto
the prism. Using the angular interrogation method, many studies have proposed the SPR sensor using Kretschman configuration to
detect numerous biological and chemical entities. These include detecting urine glucose concentration [8], hemoglobin and glucose
present in blood samples [5], the COVID-19 virus [9–11], the liver mutations [12], the dengue virus [13], the NaCl concentration in
aqueous solutions [14].Other SPR sensor formed on BaF2 prism-Cu-Pt-BP and SM studied by Almawgani et al. [15] to identify levels of
hemoglobin in the bloodstream and glucose in the urine specimen at a specific wavelength (633 nm). A highly biosensor is designed to
detect the novel SARS COV-2 corona virus [16,17] using various prisms, including BaF2, BK7, CaF2, CsF, SF6, and SiO2. Recently, to
enhance the sensitivity of the traditional Kretschmann design based on (prism-metal) of the SPR sensor, Shanmuga Vadivu et al. [18]
proposed an SPR sensor based on prism-metal layer and several silicon(Si) layers to capture the Escherichia coli (E.C) bacteria, they
reported a sensitivity limit of 268.95◦ /RIU. Researchers have used different wavelengths; Verma et al. [19] have proposed a
Kretschmann configuration based on SF10 prism/Gold/Silicon/Graphene in order to detect biomolecules using three wavelengths.
Recently, by using the interrogation method, Sassi et al. [7] have proposed a silver-titanium dioxide biosensor for various sensing
applications; and Hma Salah et al. [20] have examined a design used for petrochemicals detection, such as diesel, petrol and kerosene.
In recent years, it has been shown that 2D nanomaterials (TMDCs, MXene, Graphene and BP), display numerous plasmonic re­
sponses, provides stability and boosts performance parameters of SPR biosensor. Uniyal et al. [21] have proposed a biosensor
composed of bimetallic layers and TMDCs materials (MoS2/MoSe2/WS2/WSe2), reporting a maximum sensitivity of 446◦ /RIU.
Similarly, Almawgani et al. [22] have added the MXene on the metallic layer for the detection of cancer cells, achieving a sensitivity of
232◦ /RIU. Furthermore, other researchers have integrated the Black Phosphorus to enhance sensitivity of the sensor [8,12].
Additionally, Graphene has ignited research interest to enhance sensitivity of various biosensors. It is a specific 2D material
characterized by a single layer of carbon atoms arranged in hexagonal lattice. Its remarkable properties [23–26] arouse the interest of
numerous researchers. Graphene, characterized as a conductor with zero band gap, extremely low resistivity (10− 6), has become an
optical material for Surface Plasmon Resonance (SPR) biosensing [27]. This is attributed to its rich π-conjugation structure and
extensive surface. Consequently, the incorporation of graphene layers on the sensor surface offers excellent conductivity and serves as
an exceptional nanomaterial for bimolecular adsorption.
Several researches have indicated that incorporating of a graphene layer enhances the biosensor’s sensitivity. Wu et al. [28] have
presented a highly sensitive SPR sensor based on graphene-Au metal. They showed that the sensitivity improves 25 % in comparison
with the SPR sensor based on gold. Using the angular interrogation method and for different wavelengths (600, 633 and 660 nm) of
light source, Verma et al. [19] have showed that the addition of graphene layers augments the biomolecules adsorption. Pradeep
Kummar et al. [29] have proposed a heterostructure consists on BK7 prism/graphene/silver (Ag) material, they showed that the
electric field enhancement is 30 % greater compared to the bimetallic Ag-Au configuration. The SnSe-Graphene hybrid structure is
proposed by Rahman et al. [30] for biosensing applications. The BK7 prism configuration and the grating structure using the Graphene
and Ag materials, have been exploredfor detection the chemical entities [31]. Ghodrati et al. [32] have investigated a highly biosensor
composed of Graphene layers for the detection of uric acid. El-Sayed et al. [33] have presented an experimental investigation of the
optical properties of Graphene grown through chemical vapor deposition on the ultraviolet, visible, and near-infrared spectral ranges.
Considering the temperature-dependent of Graphene, Ghafari et al. [34] have investigated a temperature sensor with Graphene and
InSb structure. They showed that the highest value of the sensitivity is equal to 160 (nm/K). Recently, researches have focused on
metal oxides materials to enhance the sensitivity of structures based on graphene layer. Aliqab et al. [35] explored the impact of the
TiO2 layer on the performance parameters of the graphene-based structure. A heterostructure composed of graphene, Au, and TiO2 has
been used [36] to enhance the sensing of the cancerous cells. A configuration consisting of (2S2G/BK7SF10/BP/MoS2/Graphene)
proposed by Aghazadeh et al. [37] to investigate the reflectance and sensitivity for various materials, utilizing an angular interrogation
technique at the specific wavelength of 633 nm. A hybrid biosensor consists of (BK7 prism/Ag/PtSe2/Graphene) [38] is proposed to
detect cancer cells using a fixed wavelength (= 633 nm). However, utilizing wavelength interrogation, a sensitive sensor is proposed,
based on fiber optic SPR configuration by Alagdar et al. [39], composed by Ag/Pt/ITO/Graphene layer and sensing medium, showed a
high sensitivity of 4150 nm/RIU. Recently, a highly sensitive Graphene-Au sensor is studied by Muheki et al. [40] for the Cu2+ and
Mg2+ detection of the drinking water. Shatrughna Kumar et al. [41] have integrated the Black Phosphorus and the Graphene layers to
enhance the urine glucose detection. A hybrid structure consists of CaF2 prism/TiO2/Ag/BP and Graphene layer [16] to detect the
novel SARS-COV-2 virus. Another multilayered structure of BK7, silver (Ag}, platinum (Pt), graphene, and magnesium fluoride (MgF2)
have been used to detect the pregnancy [42]. Notably, the Graphene layer on the surface of the SPR sensor exhibits exceptional
conductivity, excellent carrier mobility, leading to enhanced adsorption of biomolecules on the sensing surface.
In fields such as medical, agriculture, industry and environment; the artificial network is an important tool used by several re­
searchers to predict results for various nonlinear problems [43–46]. For instance, the ANN is used to predict the depth of scour
occurring around the pier of a bridge [47], and to manage simultaneous the water and wastewater [48]. E. Akbari et al. have
demonstrated that the results produced by the ANN simulation are more accurate than those by the analytical model [49]. A simple
modelling technique based on the ANN is proposed by J. C. Patra et al. to investigate a pressure sensor [50]. In addition, the ANN was
chosen for nonlinear environments to validate the results obtained by the fault sensing of wind turbine sensor [51]. Recently in the
context of SPR sensors, the optimization of the parameters of the fiber optic sensor by using the ANN [52], yielding a sensitivity
(=3890 nm/RIU) greater than that obtained by the finite element method (=3750 nm/RIU).

2
I. Sassi et al. Optik 321 (2025) 172177

In the present paper, to enhance the sensitivity of the metallic SPR sensor, we suggest the incorporation of Graphene layers on the
top of the Ag layer through the wavelengths interrogation technique. Thus, our proposed sensor structure is CaF2 Prism-Ag-Graphene-
sensing medium. The goal of this work is to investigate the reflectance and the performance parameters of our proposed sensor, when
varying the parameter wavelength at fixed direction for different values of refractive index.A comparison between the sensitivity
results obtained from the theoretical method and the artificial neural network (ANN) is presented in this work. In addition, as an
application, we use our proposed structure as a temperature sensor to control the temperature of ethanol.

2. Numerical modeling and designing

Our simple structure sensor is composed of three layers deposited on a CaF2 prism. The first silver layer is stacked on the CaF2 rism
base. The second graphene layer is made to contact the third sensing medium layer, where it is deposited on the Ag layer. The proposed
sensor is based on the Kretschmann design, with the research centered on calculating reflectance under wavelength interrogation. TM-
polarized light at a defined wavelength is directed at the CaF2 prism interface at a fixed angle, leading to light emerging from the
prism’s opposite side.
The RI of our suggested CaF2 prism is given by the following equation [53]:
√(̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)̅
O.5675888λ2 0.4710914λ2 3.8484723λ2
nCaF2 = 1+ 2 + + . (1)
λ − (0.050263605)2 λ2 − (0.1003909)2 λ2 − (34.649040)2

The formula of the silver RI can be determined by using the Drude-Lorentz model [54]:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)̅
√( 2
√ λ λ c
nAg = √ 1 − 2 (2)
λp (λc + iλ

whereλc = 1.7614 × 10− 5 m andλp = 1.4541 × 10− 7 m.

We also utilize the following relation to determine the complex RI of the Graphene layer [55]:
λ
nGr = 3 + (5.446 ∗ ) i (3)
3

where i2 = − 1.
We choose silver, CaF2 and graphene in our structure, because of its particularity properties. Silver has a strong plasmonic response
in the visible and near-infrared regions of the electromagnetic waves. The CaF₂ is a stable material in a wide range of conditions
material, used to enhance the interaction between matter and light, and facilitate the confinement of plasmonic waves near metallic
nanostructures. Graphene’s high conductivity enables efficient plasmonic excitations, making it an ideal material for the development
of highly sensitive plasmonic devices.
To construct the suggested sensor, the process begins in the laboratory with the preparation of the CaF2 glass. The latter is
meticulously cleaned to eliminate any contaminants that could affect the quality of the deposited layers. A thin layer of Ag is then
applied on the CaF2 prism by thermal evaporation technique, serving as a reflective layer necessary for plasmonic applications. On the
top of Ag layer, multiple layers of Graphene are synthesized using chemical vapor deposition (CVD) method [56] and finally trans­
ferred over the Ag layer through specific Graphene methods. The top layer comprises the sensing medium which contains the analytes
to be detected and interacts with Graphene to adjust the plasmonic properties of the sensor.
A numerical method, called the transfer matrix method (TMM), based on the electromagnetic theory.It is based on Maxwell’s
equation with appropriate boundary conditions at different interfaces. It is always convergent and gives results immediately. That
helps in the analysis of the sensor’s performance. We were used the TMM to compute the reflectance for the parameters: a fixed
incident angle (=80 degree), a wavelength λ varied from 400 nm to 2500 nm, a silver thickness layer t_Ag, a grapheme thickness
layert_Gr, and a sensing medium’s refractive index. The details of this transfer matrix method are described by several authors [7–9,28,
38,57].
In this research, the performance of our biosensor is evaluated through the performance parameters [7]: the refractive index
sensitivity (SRI), the quality factor (QF), and the figure of merit (FoM). Also the reflectance at resonance (RR(%)), which is defined as
the minimum reflectance at the SPR wavelength, is an important parameter used to investigate our sensor’s performance.A highly
sensitive sensor is beneficial for detecting very low concentrations or minor changes in composition.
The sensitivity (SRI) and the detection accuracy (DA) of the sensor device have been obtained by using the following equations:
ΔλRes
SRI = (4)
Δn

1
DA = (5)
FWHM

where ΔλRes ,Δn, and FWHM are the change in resonance wavelength, the change in RI of the sensing medium, and the full width at half
maximum, respectively.
Also, the quality factor (QF) and the figure of merit (FoM) can be calculated as follows:

3
I. Sassi et al. Optik 321 (2025) 172177

QF = λRes ∗ DA 6)

SRI
FoM = (7)
FWHM

Fig. 1. SRI and RR(%) vs. RI for t_Ag: (a) 37, (b) 39, (c) 41, (d) 43, (e) 45, (f) 47, (g) 49, and (h)51 nm.

4
I. Sassi et al. Optik 321 (2025) 172177

Fig. 2. SRI and RR (%) vs. RI for various t_Ag: (a, b)43 nm, (c, d)45 nm, (e, f)47 nm.

5
I. Sassi et al. Optik 321 (2025) 172177

The parameters: SRI (nm/RIU), RR (%), QF, and FoM (RIU− 1) have been used to characterize the sensor’s performance.

3. Results

3.1. The sensor’s performance by using the transfer matrix method

In this paper, to investigate our sensor’s performance parameters, we first optimize the Ag thickness layer for a fixed t_Gr
(=0.34 nm). Fig. 1 illustrates the variation in both SRI and RR (%) as a function of the RI, for various values of t_Ag ranging from 37 to
51 nm with a step of 2 nm. The RI of the considered sensing medium is taken from 1.330 to 1.390. The results indicate that when t_Ag
takes the values 37, 39, and 41 nm, the sensitivity reaches maximum values of 52500, 51240, and 49560 nm RIU− 1, respectively. The
corresponding RR takes the values 6.33, 3.07, and 0.91 %, respectively, when the RI varies from 1.385to 1.39. For these thicknesses,
the resonance reflectance exceeds 5 % over most RI intervals. However, for t_Ag values of 43, 45, and 47 nm, the reflectance at
resonance decreases significantly and takes the weaker minimum reflectance (not exceed 4 %), where the maximum values of
sensitivity are 48720, 48300, and 47460 nm RIU− 1, respectively. Although for 49 nm silver thickness layer, the RR exceeds 5 %, when
the RI varies from 1.385 to 1.39. The value 51 nm of silver thickness layer is not considered as an optimal value, because its RR exceeds
5 % (see Fig. 1(d)) for the most RI intervals. Consequently, it is clear that the optimal values for the Ag thickness layer are identified as
43, 45, and 47 nm, as they exhibit maximum sensitivity and significant minimum reflectance at resonance.
To study the effect of increasing graphene thickness layer on the sensitivity and reflectance, we selected Ag layer thicknesses of 43,
45, and 47 nm. Our goal in this subsection is to select the best values of silver and graphene thickness layers, which provide significant
values of RR and sensitivities across the RI range of 1.33–1.39. Fig. 2 presents the sensitivity and reflectance at resonance vs. the
refractive index of sensing medium for six Graphene layers (G=1, 2, 3, 4, 5, and 6) with thickness calculated as G*0.34. It is clear in
Fig. 2 that the sensitivity increases with the thickness of the Graphene layer. For 43 nm silver thickness layer, as shown in Fig. 2a the
sensitivity increases from 48720 nm/RIU (G=1) to 56400 nm/RIU (G=6), when the RI is varied from 1.385 to 1.390. Additionally, for
the greater values of the t_Gr (= 1.36, 1.7, 2.04 nm), the lowest values of the RR is obtained for T_Ag = 43 nm compared tot_Ag = 45 or
47 nmfor RI interval 1.33–1.335. Furthermore, the curves presented in Fig. 2 indicate that the highest sensitivity values are given for RI
interval 1.385–1.39. In addition, the lowest values of the RR, which favorite the sharpness of the curves is obtained for the lowest value
of t_Ag (= 43 nm). Consequently, we can select 43 nm thickness layer as an optimum value for silver compared to 45 and 47 values. We
summarize in Table 1 the results concerning the RR (%) and the SRI (nm/RIU) for different values of t_Ag and t_Gr for RI intervals
(1.385–1.39) and (1.33–1.335). The investigation of the Table 1 show clearly the optimum values of the t_Ag and t_Gr that we
recommend are the following: t_Ag = 43 nm and t_Gr = 1.7 nm. This choice can be explain by its significant values of the sensitivities
(4200–54600 nm/RIU) and reflectance at resonance (not exceed 6 %) for all considered RIs (1.33, 1.335, 1.340,., 1.39). Note that, for
t_Gr = 2.04 nm, t_Ag = 43 nm, we obtain the highest value of the sensitivity (= 56280 nm/RIU), however the RR, in the RI interval
1.33–1.335 exceed 7 % (= 8.419–7.397 %).
For optimal SPR sensor, characterized by t_Ag = 43 nm and t_Gr = 1.7 nm, three parameters are crucial: SRI, FoM), and QF. We
have calculated all these parameters, as shown in Table 2. The results show an increase in sensitivity and FoM, from 4680 to 54600 nm
RIU− 1, and from 46.76 to 151.19, respectively, when the RI is varied from 1.335 to 1.390, with a step of 0.005.However, the QF
decreases from 6.62 to 4.32 for the same variation of RI. The result can be explained by the FWHM, which is an important performance
parameter, influenced by the reflectance at resonance (RR) and the sharpness of the reflectance curve shape. The important parameters
are also computed and presented in Table 2, highlighting the significant results for RI range of 1.385–1.39. Within this range, the
sensor provides a highest value of sensitivity (= 56400 nm/RIU) and a highest value of the FoM (= 151.19 RIU− 1).
In Fig. 3, we present for RI varying from 1.33 to 1.39, the variation of the reflectance vs. wavelength, for optimal structure,
characterized by t_Ag = 43 nm and t_Gr = 1.7 nm. It is remarked that, the most reflectance curves exhibit sharp profiles, with
sharpness phenomena decreasing as the wavelength increases. Consequently an enhancement of the FWHM is the result of the
sharpness of the curves. To investigate, the influence of both the wavelength and the RI on the performance parameters, we have
presented in Table 2for the optimal structure, the parameters: SRI, RR, FWHM, FoM, and QF for RI ranging from 1.33 to 1.39. It is

Table 1
The RR (%) and SRI for different values of t_Ag= 43, 45, and 47 nm, and t_Gr= 0.34, 0.68, 1.02, 1.36, 1.7, and 2.04 nm, for RI: 1.385–1.39/
1.330–1.335.
t_Gr/t_Ag 0.34 0.68 1.02 1.36 1.7 2.04

RI: 1.3385–1.390 / RR (%) 43 0.252–0.019 0.129–0.136 0.041–0.679 0.475–1.531 01.27–02.59 02.329–03.821

1.330–1.335 1.497–1.789 0.055–0.024 1.083–0.693 3.163–2.512 05.69–04.87 08.419–07.397
45 0.014–0.474 0.258–1.413 1.108–2.675 2.350–4.144 03.837–05.748 05.479–07.430
0.095–0.175 0.916–0.623 3.559–2.923 6.837–5.931 10.248–09.142 13.572–12.323
47 0.558–2.263 1.844–3.972 3.556–5.876 5.513–7.884 07.596–09.936 09.726–11.987
0.402–2.462 3.368–2.806 7.440–6.547 11.672–10.541 15.721–14.426 19.468–18.068
SRI (nm/ 43 48720 49980 51660 52920 54600 56280
RIU) 04200 04200 04620 04620 04680 04620
45 48300 49140 50820 52080 53760 55020
04200 04200 04200 04200 04200 04620
47 47460 49140 50400 51660 52920 56400
04200 04200 04200 04200 04200 04620

6
I. Sassi et al. Optik 321 (2025) 172177

Table 2
Performance parameters.
RI RW (nm) SRI RR (%) FWHM FOM QF
(nm/RIU)

1.330 631.4 - 5.69 100.27 - 6.29

1.335 654.8 04680 4.87 098.80 46.76 6.62
1.340 678.2 04680 4.09 104.79 48.09 6.48
1.345 706.8 05720 3.38 111.71 48.87 6.32
1.350 738 06240 2.77 119.79 52.59 6.16
1.355 777 07800 2.22 129.33 58.45 5.99
1.360 821.2 08840 1.77 140.88 62.60 5.82
1.365 873.2 10400 1.39 155.18 68.56 5.63
1.370 935.6 12480 1.13 173.44 71.95 5.39
1.375 1018.8 16640 0.96 197.57 84.22 5.15
1.380 1128 21840 0.96 230.97 94.55 4.88
1.385 1286.6 31720 1.27 280.91 112.92 4.58
1.390 1559.6 54600 2.59 361.13 151.19 4.32

1,0

RI
0,8
1.33
1.335
1.340
0,6
Reflectance

1.345
1.350
1.355
0,4 1.360
1.365
1.370
1.375
0,2
1.380
1.385
1.390
0,0
0,5 1,0 1,5 2,0 2,5 3,0

Wavelength (µm)
Fig. 3. Reflectance spectra.

remarked that, an increasing of RI yield only an improvement of the SRI and the FoM. However the increasing/decreasing of the
FWHM/QF when the RI increase unenhanced the performance of our biosensor. It is seen that the best values of the pair (SRI, FoM) are
the values 54600 nm/RIU, and 151.19, respectively, obtained when the RI is ranged from 1.385–1.39. However, the best values of the
pair (FWHM, QF) are the values 98.8 nm, and 6.62, respectively, obtained at RI = 1.335. Note that significant values are written in
bold.

3.2. Prediction of the sensitivity by using the artificial neural network (ANN)

ANN is a tool used to solve nonlinear functions related to various input and output datasets and is capable to provide the output
data. The output values at this stage are compared to the actual data (computed by using a numerical code) and the deviation is
computed. If the deviation value is not small enough, the weights are updated and a new cycle is initialized. This procedure is repeated
several times until the deviation between the actual data and the output data is weak. To improve the ANN training, we choose the best
values of the follows parameters: the number of hidden layers (N_HL), the number of neuron in hidden layer (N_NHL), the number of

Table 3
The ANN properties.
Network type N_HL N_NHL AF TF ALF PF

Feed-forward backprop 2 Layer1 Layer2 Layer 1 Layer 2 TRAINLM LEARNGDM MSE

20 1 TAMSIG PURELIN

7
I. Sassi et al. Optik 321 (2025) 172177

iterations, the transfer function (AF), the training function (TF), the adapting function (ALF), and the performance function (PF) is the
mean square error (MSE). The selected ANN properties are presented in Table 3.
In order to test the ability of the ANN to predict the values of the sensitivity with negligible error, we calculate the sensitivity for the
same input parameters. At fixed incident angle (= 80 deg), the sensor performance depends on 3 variables: the refractive index (RI) of
sensing medium, the silver thickness layer (Ag_th), and the graphene thickness layer (Graphene_th). The input layer comprises 3 in­
puts: RI, Ag_th, and Graphene_th. It consists of 9 different RI values ranging from 1.35 to 1.39 with a step of 0.005, 9 different layers
Ag_th from 35 to 51 nm, with step of 2 nm, and 5 different values of Graphene_th: 0.34, 0.68, 1.02, 1.36, and 2.04 nm. The output layer
corresponds to the sensitivity: S (nm/RIU). The architecture of the NN used in this study is presented in Fig. 4.
The MATLAB R2014b with its Neural Network Toolbox was used to train the network. We train the network several times until we
reach the best results concerning the regression curves. In the Table 4, we present the regression results predicted by using the ANN.
It is remarked that for the same number of the iterations (= 6), an increasing of the number of training (10, 38, 43, 67, 87, 92, and
99), leads an enhancement of the best MSE validation. The more significant value of the best MSE validation (=5.310− 23) is obtained
for the 67thtraining, characterized by the R (training) = 0.99989, R (validation) = 1, R (test) = 1, and R (All) = 0.99992. We obtain
the same latest values of regression for the 87th, the 92th, and the 99th training.
For the latest number of training (= 99), we present in Fig. 5, the best results of the regression, corresponded to the best values of
the sensitivities, predicted with the ANN technique. More clearly, Fig. 5 shows the relationship between the output of the networks and
the targeted values obtained from numerical results, as visualized through regression charts during the training, validation, and testing
phases.
In order to test the ability of the ANN to predict the output values of the sensitivity with negligible error, we present in Table 5 the
results obtained through the TMM approach and those of ANN. A comparison between the two approaches shows a great agreement
between several sensitivity values (presented in italic form). For the comparison, we define the absolute error (AE) and the relative
error (RE), presented as follows:
AE = |SRI (ANN) − SRI |, (8a)

RE(%) = AE/SRI . (8b)

− 2
It is remarked that the most values of RE (%) are limited by the negligible value 7 10 , for the most considered parameters of our
sensor. This implies that the ANN is a powerful tool for predicting accurate results.

3.3. Performance comparison with other refractive index sensors

Table 6 shows a general comparative study for performance parameters sensitivity and figure of merit. Each reference has a specific
refractive index range and a specific structure. Our results are the best, compared to that offered by the all references: [6,58–68]. An
ultra high sensitivity of 54600 nm/RIU is showed for our work, greater than that of all the sensitivity’s values presented in Table 6.

3.4. Temperature sensing

Temperature sensors have been used in various fields, in industry, medical, environmental monitoring, aerospace, and automotive.
For example, in the medical field, temperature sensors are used in thermometers [69], and incubators [70]. The temperature sensitivity
can be calculated by using the following equation [71].
Δλ
ST = , (9)
ΔT

Where, Δλ is the change in resonant wavelength, and ΔT is the change in temperature.

As an application, we can use our device to control the temperature of the ethanol. The sensors called refractive index sensors and
the temperature sensors are related by the relationship between the RI and the temperature. As an example, the expression of the
ethanol refractive index at temperature T is as follows [63]:

nEt = 1.36048 − 3.94 × 10− 4 (T − T0 ), (10)

Fig. 4. The architecture of the neural network (NN).

8
I. Sassi et al. Optik 321 (2025) 172177

Table 4
The enhancement of the regression (R) results by the increasing of the number of training.
Number of Training 1 10 38 43 67 87 92 99

Number of iterations 9 6 6 6 6 6 6 6
Best MSE validation 2153858 72004 2.56 2343 5.310− 23 249.95 1.656 0.27302
R(Training) 1 0.99973 0.99981 1 0.99989 0.99989 0.99989 0.99989
R(Validation) 0.99681 0.99996 1 1 1 1 1 1
R(Test) 0.99645 0.9999 1 0.99997 1 1 1 1
R(All) 0.99769 0.99985 0.99984 0.9999 0.99992 0.99992 0.99992 0.99992

Fig. 5. Training, testing, and validation performance of NN.

where T0 is the room temperature ( = 293.15K). Notably, the melting and the boiling points of the ethanol are 158.85 K and 351.55 K,
respectively [63].
As shown in Fig. 6, the variation of the reflectance vs. the wavelength for different values of the temperature T, ranging from
223.15 K to 323.15 K, with a step of 10 K. As Temperature T increases, the resonant wavelength increases.
In Table 7, we present for different values of T, the corresponding values of the resonant wavelengths. For each interval of the
temperature, we have its corresponded value of the temperature sensitivity. The proposed sensor achieves maximum temperature
sensitivity of 17.68 nm/K, and minimum temperature sensitivity of 2.86 nm/K, showed in Table 7.
The average sensitivity ( = 6.99 nm/K) is obtained by adding the different values of the sensitivity and divided by 10 (number’s
sensitivity values). For the same sensing medium (ethanol), our temperature sensor has higher temperature sensitivity than several
temperature sensors presented in Table 8.

9
I. Sassi et al. Optik 321 (2025) 172177

Table 5
The comparison between the TMM and the ANN sensitivity results for the same parameters of the proposed sensor structure.
RI T_Ag T_Gr SRI SRI(ANN) AE RE (%)
07
1.350 43 0.34 5880 5880 7.07 10− 1.2 10− 8
6
1.355 43 0.34 7140 7140 3.00 10− 4.2 10− 8
1.360 43 0.34 8400 8292.6202 107.38 1.2
6
1.365 43 0.34 9660 9660 4.49 10− 4.64 10− 8
1.370 43 0.34 12180 12148.3799 31.62 0.26
7
1.375 43 0.34 15120 15120 9.95 10− 0.53 10− 8
1.380 43 0.34 20580 20594.5073 14.51 7 10¡2
8
1.385 43 0.34 29400 29400 1.16 10− 3.94 10− 11
7
1.390 43 0.34 48720 48720 1.21 10− 2.48 10− 10
1.390 35 0.34 54600 54598.955 1.045 1.91 10− 3
7
1.39 37 0.34 52500 52500 1.01 10− 1.92 10− 10
8
1.39 39 0.34 51240 51240 7.44 10− 1.45 10− 10
1.39 41 0.34 49560 49557.4263 2.57 5.18 10− 3
7
1.39 43 0.34 48720 48720 0.21 10− 4.31 10− 11
7
1.39 45 0.34 49140 49140 1.82 10− 3.70 10− 10
7
1.39 47 0.34 47460 47460 2.40 10− 5.05 10− 10
7
1.39 49 0.34 47040 47040 4.01 10− 8.52 10− 10
7
1.39 51 0.34 46620 46620 3.48 10− 7.46 10− 10
7
1.39 43 0.34 48720 48720 1.21 10− 2.48 10− 10
1.39 43 0.68 49980 48837.492 1142.51 2.28
8
1.39 43 1.02 51560 51560 2.04 10− 3.95 10− 11
7
1.39 43 1.36 52920 52920 4.23 10− 7.99 10− 10
7
1.39 43 1.7 54600 54600 6.62 10− 1.21 10− 9
7
1.39 43 2.04 56280 56280 4.37 10− 7.76 10− 10

Table 6
General comparative with literature.
Ref/Year RIR WR(nm) Structure SRI (range) FoM (range) QF (range)

This work 1.33–1.39 400–2500 Prism (CaF2)-Ag-Graphene 4620–54600 46.76–151.19 6.29–4.32

[7]/2024 1.33–1.41 400–1500 Prism (Bk7)-Ag-TiO2 2000–6300 28–52.24 6.7–11.83
[58]/2023 1.35–1.39 500–750 Ag-grated fiber optic 1350–3775 - -
[59]/2022 1.32–1.4 400–800 Ag-Au alloy -graphene fiber optic 2000–7100 34–38.88 -
[60]/2019 1.33–1.4 1200–1500 Au-MgF2-Au grating structure 625–1125 7.5–87.5 -
[61]/2022 1.33–1.36 1200–2500 ITO-Ag-ITO fiber optic 4000–14000 12–60 -
[62]/2024 1.33303–1.37340 ​ GaSe-Au fiber optic ∼2900–8598 ∼46–69.57 -
1.3–1.4 ​ TiN-Insulator-TiN waveguide …− 913 . − 36.52 … − 46.35
[63]/2024
[64]/2019 1.22–1.36 1700–2400 Au-TiO2-Silica-PML Photonic crystal fiber 1000–51000 6–566 -
[65]/2023 1.32–139 1300–1500 Au-SiO2-Graphene-Ag Grating structure 400–760 69–148 138–398
[66]/2019 1.33–1.41 800–1200 SiO2-Au-(a-Si) grating structure 300–350 - -
[67]/2020 1.33–1.4 600–2000 Ag- Metal-Ag grating structure 2471–4000 49.11–52.71 -
[68]/2021 1.33–1.4 1050–1500 Circular rings MIM waveguide 1021–1054 - -

4. Conclusion

This article was devoted to the study of the performance of a refractive index sensor with wavelength interrogation method,
involving the effects of the thickness layers of silver or graphene on the performance of our sensor device. The optimization of silver
and graphene thickness was performed to get a significant minimum reflection at resonance (not exceed 5 % for all RIs varied from
1.335 to 1.39), and ultra high sensitivity. Our optimized structure demonstrated high values of sensitivity (= 54600 nm/RIU, quality
factor (= 6.62), and figure of merit (= 151.19 RIU− 1) when the RI of the sensing medium is varied from 1.33 to 1.39 with a step of
0.005. In this research, we predict the sensitivity data of our sensor using an artificial neural network (ANN) technique. Notably, this
approach yields significantly improved results compared to those obtained through theoretical model. In the context of temperature
sensing, our device exhibited an average temperature sensitivity of 6.99 nm/K, and a maximum temperature sensitivity of 17.68 nm/
K.

CRediT authorship contribution statement

Mariem Zbidi: Writing – review & editing, Investigation. Mounir Ben El Hadj Rhouma: Investigation. Imed Sassi: Writing –
review & editing, Writing – original draft, Visualization, Validation, Investigation.

10
I. Sassi et al. Optik 321 (2025) 172177

1,0 -2
(a)
(b)
Temperature
-4
(k)
0,8
-6
223.15 Sensitivity (nm/K)

Sensitivity (nm/K)
-8
0,6 233.15
Reflectance

243.15
-10
253.15
0,4 263.15 -12
273.15
283.15 -14

0,2
293.15
-16
303.15
313.15
-18
0,0
323.15
0,5 1,0 1,5 2,0 2,5 3,0 220 240 260 280 300 320

Wavelength (nm) Tempertaure (K)

Fig. 6. Temperature sensing, (a) reflectance spectrum, (b) Sensitivity vs. temperature.

Table 7
The temperature sensitivity for temperature varying from 223.15 to 323.15 K.
Temperature(K) RW (nm) Sensitivity (nm/K)

223.15 1429.6
233.15 1252.8 ¡17.68
243.15 1133.2 − 11.96
253.15 1042.2 − 09.10
26315 972 − 07.02
273.15 9148 − 05.72
283.15 865.4 − 04.94
293.15 823.8 − 04.16
303.15 790 − 03.38
313.15 758.8 − 03.12
323.15 730.2 − 02.86

Table 8
Comparison of temperature sensitivity with literature.
Reference ST,max (nm/K) Year

[72] 0.62 2014

[73] 0.84 2017
[74] 2.05 2019
[75] 2.60 2020
[76] 2.36 2021
[63] ∼ 0.39 2024
This paper 17.68 2024

Author Contributions

The manuscript was written with contributions from all authors. Imed Sassi wrote the first draft of the manuscript and all authors
commented on previous versions. Performed the numerical simulation: Imed Sassi. Analyzed the data: Imed Sassi, and Meriem Zbidi.
Revised the manuscript: Imed Sassi, Mariem Zbidi, and Mounir Ben El Hadj Rhouma.

Funding

Not applicable

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to

11
I. Sassi et al. Optik 321 (2025) 172177

influence the work reported in this paper.

Data availability

No data was used for the research described in the article.

References

[1] Mahmud Zaman Uz, et al., ZnO and antimonene-based surface plasmon resonance sensor for enamel, dentin, and cementum layer detection in human teeth,
J. Mater. Sci.: Mater. Electron. 35 (27) (2024) 1831, https://doi.org/10.1007/s10854-024-13471-x.
[2] B. Karki, A. Jha, A. Pal, V. Srivastava, Sensitivity enhancement of refractive index-based surface plasmon resonance sensor for glucose detection, Opt. Quant.
Electron 54 (9) (2022) 595, https://doi.org/10.1007/s11082-022-04004-z.
[3] A. Kumar, A. Kumar, S.K. Srivastava, Silicon nitride-BP-based surface plasmon resonance highly sensitive biosensor for virus SARS-CoV-2 detection, Plasmonics
17 (3) (2022) 1065–1077, https://doi.org/10.1007/s11468-021-01589-1.
[4] I.A. Sassi, M. Ben El Hadj Rhouma, M.G. Daher, Highly sensitive refractive index gas sensor using two-dimensional silicon carbide grating based on surface
plasmon resonance, Opt. Quant. Electron 55 (5) (2023) 402, https://doi.org/10.1007/s11082-023-04682-3.
[5] I. Sassi, M. Ben El Hadj Rhouma, Sensitivity enhancement of SPR sensor assisted by Ag, TiO2 and BP for glucose and hemoglobin detection: numerical analysis,
Opt. Quant. Electron 55 (11) (2023) 959, https://doi.org/10.1007/s11082-023-05217-6.
[6] B. Karki, A. Uniyal, P. Sarkar, A. Pal, R.B. Yadav, Sensitivity improvement of surface plasmon resonance sensor for glucose detection in urine samples using
heterogeneous layers: an analytical perspective, J. Opt. 53 (2024) 2567–2577, https://doi.org/10.1007/s12596-023-01418-0.
[7] I. Sassi, M.Ben El. Hadj Rhouma, M. Zbid, J.W. Simatupang, A highly sensitive structure based on prism, silver, and titanium dioxide for biochemical sensing
applications, Plasmonics (2024) 1–10, https://doi.org/10.1007/s11468-024-02271-y.
[8] A. Yadav, A. Kumar, P. Sharan, Sensitivity enhancement of a plasmonic biosensor for urine glucose detection by embloying Black Phosphorous, J. Opt. Soc. Am.
B 39 (1) (2022) 200–206, https://doi.org/10.1364/JOSAB.444838.
[9] C.M. Das C, Y. Guo, G. Yang, L. Kang, G. Xu, H.P. Ho, K.T. Yong, Gold nanorod assisted enhanced plasmonic detection scheme of COVID-19 SARS-COV-2 spike
protein, Adv. Theory Simul. 3 (11) (2020) 2000185, https://doi.org/10.1002/adts.202000185.
[10] P.S. Pandey, S.K. Raghuwanshi, A. Shadab, M.T.I. Ansari, U.K. Tiwari, S. Kumar, SPR based biosensing chip for COVID-19 diagnosis-a review, IEEE Sens. J. 22
(14) (2022) 13800–13810, https://doi.org/10.1109/JSEN.2022.3181423.
[11] Arun Uniyal, et al., Fluorinated graphene and CNT-based surface plasmon resonance sensor for detecting the viral particles of SARS-CoV-2, Phys. B: Condens.
Matter 669 (2023) 415282, https://doi.org/10.1016/j.physb.2023.415282.
[12] S. Srivastava, S. Yadav, A.C. Mishra, S. Singh, P. Lohia, D.K. Dwivedi, R.K. Yadav, M.K. Hossain, Ultra-sensitive surface plasmon resonance biosensor for liver
metastases and hepatocellular carcinoma detection using silicon nitride and black phosphorus nanomaterial, Plasmonics 19 (2) (2024) 1031–1041, https://doi.
org/10.1007/s11468-023-02059-6.
[13] B. Karki, N.H. Salah, G. Srivastava, A. Muduli, R.B. Yadav, A simulation study for dengue virus detection using surface plasmon resonance sensor heterostructure
of silver, barium titanate, and cerium oxide, Plasmonics 18 (2023) 2031–2040, https://doi.org/10.1007/s11468-023-01907-9.
[14] J. Chylek, D. Ciprian, P. Hlubina, Optimized film thicknesses for maximum refractive index sensitivity and figure of merit of a bimetallic film surface plasmon
resonance sensor, Eur. Phys. J. 139 (1) (2024) 1–11, https://doi.org/10.1140/epjp/s13360-023-04798-1.
[15] A.H.M. Almawgani, A. Uniyal, P. Sarkar, G. Srivastava, A. Pal, A.R.M. Alhawari, S. A. Taya, A. Muduli, Sensitivity enhancement of optical plasmon-based sensor
for detection of the hemoglobin and glucose: a numerical approach, Opt. Quantum Electron. 55 (11) (2023) 963, https://doi.org/10.1007/s11082-023-05219-4.
[16] T.B.A. Akib, S. Mostufa, M.M. Rana, M.B. Hossain, M.R. Islam, A performance comparison of heterostructure surface plasmon resonance biosensor for diagnosis
of novel coronavirus SARS-CoV-2, Opt. Quantum Electron. 55 (5) (2023) 448, https://doi.org/10.1007/s11082-023-04700-4.
[17] Amrindra Pal, et al., Tuning sensitivity of surface plasmon resonance sensor based on ZnO layer and CaF2 prism for the recognition of SARS-CoV-2, J. Mater.
Sci.: Mater. Electron. 35 (22) (2024) 1523, https://doi.org/10.1007/s10854-024-13287-9.
[18] S. Lavadiya, et al., Sensitivity enhancement of Silicon layer- based surface plasmon resonance sensor for detection of Escherichia coli bacteria, Plasmonics
(2024) 2266, https://doi.org/10.1007/s11468-024-02266-9.
[19] R. Verma, B.D. Gupta, R. Jha, Sensitivity enhancement of a surface plasmon resonance based biomolecules sensor using graphene and silicon layers, Sens.
Actuators B (2011) 623–631, https://doi.org/10.1016/j.snb.2011.08.039.
[20] N. Hma Salah, A. Pal, A. Uniyal, Enhancing precision in fuel adulteration detection: utilizing a wavelength interrogation surface plasmon resonance approach,
Plasmonics (2024) 1–10, https://doi.org/10.1007/s11468-024-02340-2.
[21] A. Uniyal, A. Pal, G. Srivastava, M.M. Rana, S.A. Taya, A. Sharma, A.N.Z. Rashed, Surafce plasmon resonance biosensor sensitivity improvement employing of
2D materials and BaTiO3 with bimetallic layers of Silver, J. Mater. Sci: Mater. Electron 34 (6) (2023) 466, https://doi.org/10.1007/s10854-023-09821-w.
[22] A.H. Almawgani, M.G. Daher, S.A. Taya, A.T. Hindi, I. Colak, A. Pal, Detection of blood cancer using sensitive surface plasmon resonance sensor based on MXene
2D nanomaterial, Diam. Relat. Mater. 137 (27) (2023) 110142, https://doi.org/10.1016/j.diamond.2023.110142.
[23] A.H.C. Neto, F. Guinea, N.M. Peres, K.S. Novoselov, A.K. Geim, The electronic properties of graphene, Rev. Mod. Phys. 81 (1) (2009) 109–162, https://doi.org/
10.1103/RevModPhys.81.109.
[24] A. Vakil, N. Engheta, Transformation optics using graphene, Science 332 (6035) (2011) 1291–1294, https://doi.org/10.1126/science.1202691.
[25] A.A. Balandin, Thermal properties of graphene and nanostructured carbon materials, Nat. Mater. 10 (8) (2011) 569–581, https://doi.org/10.1038/nmat3064.
[26] C. Lee, W. XIAODING, J.W. Kysar, J. Hone, Measurement of the elastic properties and intrinsic strength of monolayer graphene, Science 321 (5887) (2008)
385–388, https://doi.org/10.1126/science.1157996.
[27] W. Wei, J. Nong, G. Zhang, L. Tang, X. Jiang, N. Chen, S. Luo, G. Lan, Y. Zhu, Graphene-based long-period fiber grating surface plasmon resonance sensor for
high-sensitivity gas sensing, Sensors 17 (1) (2016) 2, https://doi.org/10.3390/s17010002.
[28] L. Wu, H.S. Chu, W.S. Koh, E.P. Li, Highly sensitive graphene biosensors based on surface plasmon resonance, Opt. Express 18 (14) (2010) 14395–14400,
https://doi.org/10.1364/OE.18.014395.
[29] P.K. Maharana, P. Padhy, R. Jha, On the field enhancement and performance of an ultra-stable SPR biosensor based on graphene, IEEE Photonics Technol. Lett.
25 (22) (2013) 2156–2159, https://doi.org/10.1109/LPT.2013.2281453.
[30] M.S. Rahman, K. Rikta, L.F. Abdulrazak, S. Anower, Enhanced performance of SnSe-Graphene hybrid photonic surface plasmon refractive sensor for biosensing
applications, Photon. Nanostruct. 39 (2020) 100779, https://doi.org/10.1016/j.photonics.2020.100779.
[31] L. Kong, G. Qiongchan, Y. Yu, J. Xiaoxiao, S. Guangyuan, Sensitivity-enhanced SPR sensor based on graphene and subwavelength silver gratings, Nanomaterials
10 (11) (2020) 2125, https://doi.org/10.3390/nano10112125.
[32] M. Ghodrati, A. Mir, A. Farmani, Proposing of SPR biosensor based on 2D Ti3C2Tx MXene for uric acid detection immobilized by uricase enzyme, J. Comput.
Electron 22 (1) (2023) 560–569, https://doi.org/10.1007/s10825-022-01959-w.
[33] M.A. El-Sayed, G.A. Ermolaev, K.V. Voronin, R.I. Romanov, G.I. Tselikov, D.I. Yakubovsky, N.V. Doroshina, V.S. Volkov, Optical constants of chemical vapor
deposited Graphene for photonic applications, Nanomaterials 11 (5) (2021) 1230, https://doi.org/10.3390/nano11051230.
[34] S. Ghafari, M.R. Forouzeshfard, Z. Vafapour, Thermo optical switching and sensing applications of an infrared metamaterial, IEEE Sens. J. 20 (6) (2019)
3235–3241, https://doi.org/10.1109/JSEN.2019.2955672.

12
I. Sassi et al. Optik 321 (2025) 172177

[35] K. Aliqab, A. Uniyal, G. Srivastava, A. Muduli, M. Alsharari, A. Armghan, A theoretical analysis of refractive index sensor with improved sensitivity using
titanium dioxide, graphene, and antimonene grating: Pseudomonas bacteria detection, Measurement 216 (2023) 112957, https://doi.org/10.1016/j.
measurement.2023.112957.
[36] K.M. Ibrahimi, R. Kumar, W. Pakhira, A graphene/Au/TiO2 coated dual-core PCF SPR biosensor with improved design and performance for early cancer cell
detection of with high sensitivity, Optik 288 (2023) 171186, https://doi.org/10.1016/j.ijleo.2023.171186.
[37] S. Aghazadeh, H. Vahed, Sensitivity enhancement of optical long-range SPR sensor based on new 2-D materails: BP/MOS2/graphene, Plasmonics 18 (6) (2023)
2247–2261, https://doi.org/10.1007/s11468-023-01948-0.
[38] B. Karki, P. Sarkar, G. Dhiman, G. Srivastava, M. Kumar, Platinum diselenide and graphene-based refractive index sensor for cancer detection, Plasmonics 19
(2023) 953–962, https://doi.org/10.1007/s11468-023-02051-0.
[39] M. Alagdar, B. Yousif, N.F. Areed, M. Elzalabani, Highly sensitive fiber optic surface plasmon resonance sensor employing 2D nanomaterials, Appl. Phys. A
Mater. Sci. Process 126 (7) (2020) 522, https://doi.org/10.1007/s00339-020-03712-1.
[40] J. Muheki, J. Wekalao, H.B. Albargi, M. Jalah, A.H.M. Almawgani, S. Patel, A graphene gold metasurface inspired surface plasmon resonance sensor designed
for terahertz applications in sensing and detection of heavy metals in water, Plasmonics (2024) 1–15, https://doi.org/10.1007/s11468-024-02273-w.
[41] S. Kumar, A. Yadav, B.A. Malomed, Bimetal thin film, Semiconductors, and 2D Nanomaterials in SPR biosensors: an approach to enhanced urine glucose sensing,
IEEE Trans. NanoBioscience 336 (23) (2024) 343, https://doi.org/10.1109/TNB.2024.3354571.
[42] A. Pal, Y. Trabelsi, P. Sarkar, et al., Plasmonic pregnancy detector: enhancing sensitivity with SPR sensor, Opt. Quant. Electron 56 (2024) 1404, https://doi.org/
10.1007/s11082-024-07342-2.
[43] F. Amato, A. López, E.M. Peña-Méndez, P. Vaňhara, A. Hampl, J. Havel, Artificial neural networks in medical diagnosis, J. Appl. Biomed. 11 (2) (2013) 47–58,
https://doi.org/10.2478/v10136-012-0031-x.
[44] J.M. Haglin, G. Jimenez, A. E. M, Eltorai, Artificial neural networks in medicine, Health Technol. 9 (2019) 1–6, https://doi.org/10.1007/s12553-018-0244-4.
[45] S. Kujawa, G. Niedbała, Artificial neural networks in agriculture, Agriculture 11 (6) (2021) 497, https://doi.org/10.3390/agriculture11060497.
[46] F. Mumali, Artificial neural network-based decision support systems in manufacturing processes: a systematic literature review, Comput. Ind. Eng. 165 (2022)
107964, https://doi.org/10.1016/j.cie.2022.107964.
[47] W.H. Hassan, H.H. Hussein, M.H. Alshammari, H.K. Jalal, S.E. Rasheed, Evaluation of gene expression programming and artificial neural networks in PyTorch
for the prediction of local scour depth around a bridge pier, Results Eng. 13 (2022) 100353, https://doi.org/10.1016/j.rineng.2022.100353.
[48] F. Rastegaripour, M.S. Saboni, S. Shojaei, et al., Simultaneous management of water and wastewater using ant and artificial neural network (ANN) algorithms,
Int. J. Environ. Sci. Technol. 16 (2019) 5835–5856, https://doi.org/10.1007/s13762-018-1943-0.
[49] E. Akbari, Z. Buntat, A. Enzevaee, S.J. Mirazimiabarghouei, M. Bahadoran, A. Shahidi, A. Nikoukar, An analytical model and ANN simulation for carbon
nanotube based ammonium gas sensors, RSC Adv. 4 (69) (2014) 36896–36904, https://doi.org/10.1039/C4RA06291D.
[50] J.C. Patra, G. Panda, R. Baliarsingh, Artificial neural network-based nonlinearity estimation of pressure sensors, IEEE Trans. Instrum. Meas. 43 (6) (1994)
874–881, https://doi.org/10.1109/19.368082.
[51] A.G. Kavaz, B. Barutcu, Fault detection of wind turbine sensors using artificial neural networks, J. Sens. 1 (2018) 5628429, https://doi.org/10.1155/2018/
5628429.
[52] Y. Dogan, R. Katirci, I. Erdogan, E. Yartasi, Artificial neural network based optimization for Ag grated D-shaped optical fiber surface plasmon resonance
refractive index sensor, Opt. Commun. 534 (2023) 129332, https://doi.org/10.1016/j.optcom.2023.129332.
[53] I.H. Malitson, A redetermination of some optical properties of calcium fluoride, Appl. Opt. (1963) 1103–1107. https://opg.optica.org/ao/abstract.cfm?URI=ao-
2-11-1103.
[54] L. Zhao, D. Wu, Q. Chen, MXene-based Kretschmann configured surface plasmon resonance sensor in visible regime, Plasmonics 19 (2024) 1679–1685, https://
doi.org/10.1007/s11468-023-02097-0.
[55] M. Bruna, S. Borini, Optical constants of graphene layers in the visible range, Appl. Phys. Lett. 94 (2009) 03190, https://doi.org/10.1063/1.3073717.
[56] J.B. Smith, D. Hagaman, H.F. Ji, Growth of 2D black phosphorus film from chemical vapor deposition, Nanotechnology 27 (21) (2016) 215602, https://doi.org/
10.1088/0957-4484/27/21/215602.
[57] G.A. Vibisha, M.G. Daher, S.H. Rahman, Z. Jaroszewicz, K.B. Rajesh, R. Jha, Designing high sensitivity and high figure of merit spr biosensor using copper and
2d material on Caf2 prism, Results Opt. 11 (2023) 100407, https://doi.org/10.1016/j.rio.2023.100407.
[58] Y. Dogan, R. Katirci, I. Erdogan, E. Yartasi, Artificial neural network based optimization for Ag grated D-shaped optical fiber surface plasmon resonance
refractive index sensor, Opt. Commun. 534 (2023) 129332, https://doi.org/10.1016/j.optcom.2023.129332.
[59] Y. Saad, M.H. Gazzah, K. Mougin, M. Selmi, H. Belmabrouk, Sensitive detection of SARS-CoV-2 using a novel plasmonic fiber optic biosensor design, Plasmonics
17 (2022) 1489–1500, https://doi.org/10.1007/s11468-022-01639-2.
[60] F. Baranzadeh, N. Nozhat, Tunable metasurface refractive index plasmonic nano-sensor utilizing an ITO thin layer in the near-infrared region, Appl. Opt. 58
(2019) 2616–2623, https://doi.org/10.1364/AO.58.002616.
[61] D.S. Gaur, A. Purohit, S.K. Mishra, A.K. Mishra, An interplay between lossy mode resonance and surface plasmon resonance and their sensing applications,
Biosensors 12 (9) (2022) 721, https://doi.org/10.3390/bios12090721.
[62] J.D. Liu, N. Kari, H.S. Liu, W.S. Liu, W.S. Wang, Z.M. Xia, Q. Wang, Highly sensitive multimode-single-mode-multimode optical fiber SPR refractive index sensor
with GaSe nanosheets, Plasmonics (2024) 1–12, https://doi.org/10.1007/s11468-024-02252-1.
[63] R. Rahad, M.M. Sobhani, M.J.H. Emon, S.M.T.S. Afrid, M.K. Mahadi, A.S. Mohsin, R.H. Sagor, An alternative plasmonic material-based CMOS-compatible
temperature sensor, Opt. Commun. (2024) 130749, https://doi.org/10.1016/j.optcom.2024.130749.
[64] E. Haque, M. Anwar Hossain, Y. Namihira, F. Ahmed, Microchannel-based plasmonic refractive index sensor for low refractive index detection, Appl. Opt. 58
(2019) 1547–1554, https://doi.org/10.1364/AO.58.001547.
[65] A. Vajdi, M. Sadeghi, Z. Adelpour, Design of an optofluidic sensor based on metamaterial absorbers for detection of organic materials, Opt. Quant. Electron 55
(2023) 601, https://doi.org/10.1007/s11082-023-04785-x.
[66] A.A. Rifat, M. Rahmani, L. Xu, A.E. Miroshnichenko, Hybrid metasurface based tunable near-perfect absorber and plasmonic sensor, Materials 11 (7) (2018)
1091, https://doi.org/10.3390/ma11071091.
[67] I. Sassi, R. Mghaieth, Infrared thermal source or perfect absorber sensor based on silver 2D grating, Appl. Phys. A 126 (2020) 675, https://doi.org/10.1007/
s00339-020-03854-2.
[68] K.S. Rashid, M.F. Hassan, A.A. Yaseer, I. Tathfif, R.H. Sagor, Gas-sensing and label-free detection of biomaterials employing multiple rings structured plasmonic
nanosensor, Sens. Bio-Sens. Res. 33 (2021) 100440, https://doi.org/10.1016/j.sbsr.2021.100440.
[69] Yi Su, et al., Printable, highly sensitive flexible temperature sensors for human body temperature monitoring: a review, Nanoscale Res. Lett. 15 (2020) 1–34,
https://doi.org/10.1186/s11671-020-03428-4.
[70] Mustafa Burunkaya, Murat Yucel, Measurement and control of an incubator temperature by using conventional methods and fiber Bragg grating (FBG) based
temperature sensors, J. Med. Syst. 44 (10) (2020) 178, https://doi.org/10.1007/s10916-020-01650-2.
[71] Md. Farhad Hassan, et al., An optimized dielectric-metal-dielectric refractive index nanosensor, IEEE Sens. J. 21 (2) (2020) 1461–1469, https://doi.org/
10.1109/JSEN.2020.3016570.
[72] Shiwei Zou, et al., A nanoscale refractive index sensor based on asymmetric plasmonic waveguide with a ring resonator: a review, IEEE Sens. J. 15 (2) (2014)
646–650, https://doi.org/10.1109/JSEN.2014.2364251.
[73] Mohammad Reza Rakhshani, Mohammad Ali Mansouri-Birjandi, High sensitivity plasmonic refractive index sensing and its application for human blood group
identification, Sens. Actuators B: Chem. 249 (2017) 168–176, https://doi.org/10.1016/j.snb.2017.04.064.

13
I. Sassi et al. Optik 321 (2025) 172177

[74] Yuan-Fong Chau, Chou, Mid-infrared sensing properties of a plasmonic metal–insulator–metal waveguide with a single stub including defects, J. Phys. D: Appl.
Phys. 53 (11) (2020) 115401, https://doi.org/10.1088/1361-6463/ab5ec3.
[75] Rakibul Hasan Sagor, et al., Highly sensitive refractive index sensor optimized for blood group sensing utilizing the Fano resonance, Appl. Nanosci. 11 (2021)
521–534, https://doi.org/10.1007/s13204-020-01622-5.
[76] Kazi Sharmeen Rashid, et al., Cog-shaped refractive index sensor embedded with gold nanorods for temperature sensing of multiple analytes, Opt. Express 29
(23) (2021) 37541–37554.

14