Optics Communications 530 (2023) 129063

Contents lists available at ScienceDirect

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

Design and demodulation stability analysis of cascaded photonic crystal

nanobeam cavities for dual-parameter detection
Zhe Han a , Chao Wang a , Jinzhi Wang a , Yuanyuan Liu b,c , Huiping Tian a ,∗
a
State Key Laboratory of Information Photonics and Optical Communications, Beijing Key Laboratory of Space-ground interconnection and Convergence, School of
Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China
b
Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing 100049, China
c
The Engineering Research Center for Semiconductor Integrated Technology, Institute of Semiconductors, CAS, Beijing 100083, China

ARTICLE INFO ABSTRACT

Keywords: The influence of dual-parameter sensor resolution on the demodulation results is usually ignored. We present
Dual-parameter sensor a cascaded photonic crystal nanobeam cavities sensor (CPCNCS) on the silicon-on-insulator (SOI) platform
Photonic crystal nanobeam that can effectively suppress demodulation deviations. The sensor consists of a slotted photonic crystal (PhC)
Demodulation stability factor
nanobeam cavity (𝑐𝑎𝑣1 ) and an unslotted PhC nanobeam cavity (𝑐𝑎𝑣2 ). To describe the ability for suppressing
Vectorization analysis method
demodulation deviations, we propose the concept of demodulation stability factor and perform a quantitative
analysis. The vectorization analysis method provides a clear design idea for dual-parameter sensors with high
demodulation stability factor. The demodulation stability factor of the proposed CPCNCS is as high as 2497 nm
RIU−1 K−1 , indicating that the proposed sensor has a superior performance in demodulating refractive index
and temperature.

1. Introduction form a two-dimensional sensing matrix. The premise of using a two-

dimensional sensing matrix is that the determinant of the matrix cannot
In recent years, optical sensors have attracted wide attention in the be zero. Although researchers take this premise into account, the
areas of medical diagnosis, food safety monitoring, chemical detection demodulation deviations caused by sensor resolution is usually ignored.
and environmental monitoring [1–9]. Among these optical sensors, Quantitative analysis of demodulation deviations is lacking.
sensors working in aqueous solutions are widely used. The refractive In this article, we present a cascaded PhC nanobeam cavities sensor
indexes (RIes) of aqueous solutions are not only affected by con- (CPCNCS) on the silicon-on-insulator (SOI) platform. The CPCNCS con-
centration, but also by temperature. Therefore, the single parameter sists of a slot width modulated PhC nanobeam cavity (𝑐𝑎𝑣1 ) and a lattice
sensors that detect wavelength shift of merely one resonant mode constant modulated PhC nanobeam cavity (𝑐𝑎𝑣2 ). The combination of
cannot avoid the effect of temperature fluctuation around the detection
two nanobeam cavities with significant differences in electric field
environment. Since the temperature fluctuation leads to deviations of
is proved to effectively suppress the demodulation deviations. The
detection results, it is crucial to eliminate the influence of temperature
demodulation deviations of RI and temperature caused by resolution
during the sensing process. The dual-parameter sensors, which can
are analyzed quantitatively. Demodulation stability factor is proposed
simultaneously detect RI and temperature, can effectively eliminate the
as a comprehensive indicator to describe the ability of dual-parameter
influence of temperature. Many structures can detect RI and tempera-
ture, such as photonic crystals (PhCs) [8,10–13], optical fibers [14–21] sensor to suppress demodulation deviations. The design method of
and metasurfaces [22–24], etc. dual-parameter sensor with high demodulation stability factor is sum-
Compared with other types of optical sensors, PhC sensors have marized by vectorization analysis. The demodulation stability factor of
several advantages such as ultra-small size, high Q-factor and small the proposed CPCNCS is as high as 2497 nm RIU−1 K−1 . Compared
mode volume [10–13]. The Q-factor and mode volume are the metrics with other dual-parameter sensors, the proposed sensor exhibits high
for spatial confinement and temporal enhancement, respectively [25]. demodulation stability (Table 1).
Therefore, the optical microcavities with high Q-factor and small
mode volume provide powerful means for modifying the interaction 2. Structure design and numerical results
between light and matter [26]. For PhC dual-parameter sensors, the
RI and temperature can be demodulated by the dual-wavelength ma- Fig. 1 illustrates the schematic of the proposed dual-parameter sen-
trix method [10,27]. The sensitivities of the two sensing modes sor based on two cascaded PhC nanobeam cavities. Two 220 nm-thick

∗ Corresponding author.
E-mail address: [email protected] (H. Tian).

https://doi.org/10.1016/j.optcom.2022.129063
Received 28 July 2022; Received in revised form 6 October 2022; Accepted 9 October 2022
Available online 18 October 2022
0030-4018/© 2022 Elsevier B.V. All rights reserved.
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

Fig. 1. Schematic of the proposed CPCNCS on the SOI platform. The thickness of silicon PhC nanobeam cavities and buried silica layer are 220 nm and 2 μm, respectively. The
slot width of 𝑐𝑎𝑣1 and the lattice constant of 𝑐𝑎𝑣2 are quadratically modulated from center to sides.

silicon PhC nanobeam cavities are insulated into a 2 μm buried silica where 𝜔1 and 𝜔2 are the upper and lower band edge frequency,
layer. The RIes of silicon core, silica buffer layer and water are 3.46, respectively. 𝜔𝑟 is the target resonant frequency, and 𝜔𝑚 is the midgap
1.44 and 1.33, respectively [13]. The TO coefficients of the silicon core, frequency. The electric field distribution of fundamental mode for
the silica buffer layer and water are 1.8 × 10−4 K−1 , 1 × 10−5 K−1 and 𝑐𝑎𝑣1 in the x-y plane with z-axis corresponding to the half-height
−0.8 × 10−4 K−1 , respectively. Since the RI sensitivity and temperature of the silicon waveguide is shown in Fig. 2(d). The fundamental
sensitivity of 𝑐𝑎𝑣1 should be significantly different from 𝑐𝑎𝑣2 , the design mode is well confined within the slot region, which indicates that a
idea is to localize the light field in the low RI region of 𝑐𝑎𝑣1 and high RI strong interaction between the cavity mode and the analytes can be
region of 𝑐𝑎𝑣2 , respectively. The analysis process in Section 3 describes achieved.
the reason for this design in detail. The optical sensor with excellent performance should have high
The slotted cavity tightly confines light to the slot region. The Q-factor, high sensitivity and acceptable transmittance. Therefore, the
confinement enhances the light–matter interaction and thus results in structural parameters (𝑁𝑡 , 𝑁𝑚 and 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 ) closely related to these three
higher sensitivity. Therefore, 𝑐𝑎𝑣1 adopts a slot structure. The geometri- performance indicators should be optimized. Since the optimization
cal parameters of 𝑐𝑎𝑣1 are defined in Fig. 2(a). 𝑐𝑎𝑣1 consists of an array process of the structural parameters is similar to the previous work [10,
of rectangular holes in the same dimension and a nano-slot etched into 27], it will not be iterated here. The optimized parameters for 𝑐𝑎𝑣1 are:
the silicon strip waveguide with a width (𝑤𝑛𝑏 ) of 650 nm. The width 𝑁𝑡 = 20, 𝑁𝑚 = 5, 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 = 170 nm. The Q-factor of 𝑐𝑎𝑣1 is 1.3 × 104 .
(𝑤𝑥 ) and length (𝑤𝑦 ) of the rectangular holes are 450 nm and 150 nm, 𝑐𝑎𝑣2 is designed to localize the light field mainly in the high
respectively. The lattice constant (a) is 490 nm. The whole nanobeam RI region, so the unslotted dielectric-mode structure is adopted. The
cavity is symmetric with respect to the red dashed line. According to geometrical parameters of 𝑐𝑎𝑣2 are defined in Fig. 3(a). 𝑐𝑎𝑣2 consists of
the deterministic cavity design approach [28,29], we design 𝑐𝑎𝑣1 by an array of elliptical holes with the same dimension. Here, the elliptical
tapering the width of the slot from the center to both ends quadratically hole is chosen to achieve wider photonic band gap (PBG) [31–33].
to create a Gaussian mirror, while other structural parameters remain The minor semi-axis (𝑟𝑥 ) and the major semi-axis (𝑟𝑦 ) of the elliptical
unchanged. The variation of the slot width is as follows [30,31]: holes are 225 nm and 100 nm, respectively. The width of silicon strip
(𝑚 − 1)2 waveguide (𝑤𝑛𝑏 ) is 650 nm, and the structure is symmetric with respect
𝑤𝑚 = 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 − ( )2 ( ) (1) to the red dashed line. To create a Gaussian mirror, the lattice constant
𝑁𝑡 − 1 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 − 𝑤𝑒𝑛𝑑
a is tapered from 𝑎𝑐𝑒𝑛𝑡𝑒𝑟 to 𝑎𝑒𝑛𝑑 quadratically while other geometric
where 𝑤𝑚 represents the slot width and m increases from 1 to 𝑁𝑡 . 𝑁𝑡 parameters remain unchanged. The optimized structural parameters are
represents the number of rectangle holes in the taper region. 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 as follows: 𝑎𝑐𝑒𝑛𝑡𝑒𝑟 = 335 nm, 𝑎𝑒𝑛𝑑 = 374 nm, 𝑁𝑡 = 8, 𝑁𝑚 = 2. The
and 𝑤𝑒𝑛𝑑 are the widths of the slot in the center and the slot at calculated steady-state electric field distribution in the x-y plane with z-
both ends, respectively. In previous studies, nanobeam cavities are axis corresponding to the half-height of the silicon waveguide is shown
mainly based on the modulated lattice constant and nanohole size [10– in Fig. 3(b). The electric field is mainly localized in the dielectric region
12]. 𝑐𝑎𝑣1 is formed by modulating the width of slot, which enriches and the Q-factor 𝑐𝑎𝑣2 is 6.1 × 104 .
the design method of the nanobeam cavity. The three-dimensional The two cavities are connected in parallel by using a 1 × 2 power
finite-difference-time-domain (3D-FDTD) method is utilized for the splitter and a 2 × 1 power combiner in the input and output port,
simulations. The band diagrams of 𝑐𝑎𝑣1 with slot width 𝑤𝑐𝑒𝑛𝑡𝑒𝑟 = 170 nm respectively. The total transmission spectrum of the proposed sensor
and 𝑤𝑒𝑛𝑑 = 120 nm for TE polarization are given in Fig. 2(b). The is shown in Fig. 4, where the fundamental mode resonant wavelengths
frequency marked by the black solid circle is the resonant frequency. of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 are 1540.4 nm and 1517.9 nm, respectively.
The width of the edge slot is smaller than the center slot, which means Fig. 5(a) shows the transmission spectra of the proposed sensor
that the effective RI of the guiding structure increases gradually from with the temperature varying from 300 K to 330 K, while RI remains
the center to both sides, thus the cutoff frequency becomes lower. The unchanged at 1.33. Due to the different electric field distributions of
value of 𝑤𝑒𝑛𝑑 is determined by calculating the mirror strength of several 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , the resonant wavelengths experience a blue shift for
slot widths, and 𝑤𝑒𝑛𝑑 = 120 nm is chosen to realize maximum mirror 𝑐𝑎𝑣1 and a red shift for 𝑐𝑎𝑣2 with the increase of temperature. A linear
strength, as shown in Fig. 2(c). The mirror strength 𝛾 is calculated fit of the resonant wavelength shift as a function of the temperature
by [28]: is depicted in Fig. 5(b). The temperature sensitivities of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2
√
√(
√ 𝜔 − 𝜔 )2 (𝜔 − 𝜔 )2 are −29 pm/K and 58 pm/K, respectively. The calculated transmission
√ 1 2 𝑟 𝑚
𝛾 = √( )2 − (2) spectra with varying RI are illustrated in Fig. 5(c). The transmission
𝜔1 + 𝜔2 𝜔𝑚 2 spectra experience a red shift in their resonance response due to the

2
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

Fig. 2. (a) Top view and geometrical parameters of 𝑐𝑎𝑣1 . (b) TE band diagrams of 𝑐𝑎𝑣1 with 𝑤 = 170 nm (red line) and 𝑤 = 120 nm (blue line). The black solid circle indicates
the target resonant frequency. (c) Mirror strengths at different slot widths from the band diagram simulations. (d) The field distribution of fundamental mode in the x-y plane
with z-axis corresponding to the half-height of the silicon waveguide.

Fig. 3. (a) Top view and geometrical parameters of 𝑐𝑎𝑣2 . (b) The field distribution of fundamental mode in the x-y plane with z-axis corresponding to the half-height of the
silicon waveguide.

increase in RI from 1.33 to 1.345, while the temperature remains at and RI (both of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 ) are depicted in Fig. 5(d). The RI sensitiv-
300 K. The influence of RI on resonant wavelength can be explained ities 𝑆𝑛1 = 728 nm/RIU and 𝑆𝑛2 = 146 nm/RIU are calculated for 𝑐𝑎𝑣1
by perturbation theory. The relationship between frequency shift 𝛥𝜔 and 𝑐𝑎𝑣2 , respectively. Here, the sensitivities of temperature and RI of
CPCNCS have a great difference, which meets our design requirements.
and small perturbation 𝛥𝑛 of the material RI can be expressed as [34]:
2 3. Dual-parameter sensing and demodulation stability
𝛥𝑛 ∫ 𝑉𝑝 𝑑 𝐫𝜀 (𝐫) |𝐄 (𝐫)|
3
𝛥𝜔
≈− (3)
𝜔 𝑛 ∫ 𝑑 𝐫𝜀 (𝐫) |𝐄 (𝐫)|
3 2
3.1. Analysis of dual-parameter sensing
Herein, 𝜔 and n are the original resonance frequency and material RI.
The dual-wavelength matrix can demodulate the temperature and
𝑉𝑝 and 𝜖 is the perturbed region and dielectric function. E is the mode
RI, which has been applied in previous works [11,27]. In order to
profile of the resonant mode. According to Eq. (3), we can conclude that simultaneously detect temperature and RI, a sensitivity matrix M is
the decrease of the RI causes the increase of the resonant frequency of defined as follows:
the sensing mode. The results shown in Figs. 5(a) and 5(c) are consis- ( )
𝑆𝑛1 𝑆𝑇 1
tent with Eq. (3). The linear fittings between resonant wavelength shift 𝑀= (4)
𝑆𝑛2 𝑆𝑇 2

3
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

where 𝑀 −1 is the inverse matrix of M. To make sure the existence

of 𝑀 −1 , the determinant of M (lM l) cannot be zero. Substitute the
sensitivities above into Eq. (6). Once the wavelength shifts of 𝑐𝑎𝑣1 and
𝑐𝑎𝑣2 are measured, 𝛥𝑛 and 𝛥𝑇 can be mathematically estimated by the
following formula:
( ) ( )−1 ( )
𝛥𝑛 728𝑛𝑚∕𝑅𝐼𝑈 − 29𝑝𝑚∕𝐾 𝛥𝜆1
= (7)
𝛥𝑇 146𝑛𝑚∕𝑅𝐼𝑈 58𝑝𝑚∕𝐾 𝛥𝜆2

3.2. Vectorization analysis method of demodulation stability

In actual sensing scenarios, the elements in the dual-wavelength

matrix have deviations due to the resolution of the sensor (i.e., the
minimum wavelength shift that can be detected by the sensor). Here,
we propose the concept of demodulation stability factor to characterize
the ability of the dual-parameter sensor to suppress the influence of the
wavelength deviations on the detection results.
Fig. 4. The transmission spectrum of the proposed CPCNCS. When considering the influence of wavelength deviations, the de-
modulation process of RI and temperature can be expressed as:
( ) ( )( )
where 𝑆𝑛1 and 𝑆𝑇 1 are the sensitivities of RI and temperature for 𝑐𝑎𝑣1 , 𝛥𝑛 + 𝛿𝑛 1 𝑆𝑇 2 + 𝛿𝑆𝑇 2 −𝑆𝑇 1 − 𝛿𝑆𝑇 1 𝛥𝜆1 + 𝛿𝜆1
=
respectively. 𝑆𝑛2 and 𝑆𝑇 2 are RI sensitivity and temperature sensitivity 𝛥𝑇 + 𝛿𝑇 |𝑀| + 𝛿 |𝑀| −𝑆𝑛2 − 𝛿𝑆𝑛2 𝑆𝑛1 + 𝛿𝑆𝑛1 𝛥𝜆2 + 𝛿𝜆2
for 𝑐𝑎𝑣2 , respectively. The resonant wavelength shifts of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 (8)
can be expressed as:
( ) ( ) where 𝛿n and 𝛿T are the deviations of 𝛥𝑛 and 𝛥𝑇 , respectively. 𝛿|𝑀| is
𝛥𝜆1 𝛥𝑛
=𝑀 (5) the deviation of |𝑀|. Similarity, 𝛿𝑆𝑇 𝑖 and 𝛿𝑆𝑛𝑖 are the deviations of 𝑆𝑇 𝑖
𝛥𝜆2 𝛥𝑇
and 𝑆𝑛𝑖 , respectively. Sensitivity is the ratio of the resonant wavelength
where 𝛥𝜆1 and 𝛥𝜆2 are the resonant wavelength shifts of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 shift to the corresponding temperature or RI change, so there is also a
as responses to the variations of RI (𝛥𝑛) and temperature (𝛥𝑇 ). deviation in sensitivity. 𝛿𝜆1 and 𝛿𝜆2 are the wavelength deviations of
After detecting the resonant wavelength shifts of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 in 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , respectively, which do not exceed the resolution of the
the output spectrum, 𝛥𝑛 and 𝛥𝑇 can be calculated as: sensor [35]:
( ) ( ) ( )( )
𝛥𝑛 𝛥𝜆1 𝑆𝑇 2 −𝑆𝑇 1 𝛥𝜆1
= 𝑀 −1 =
1
(6) |𝛿𝜆1 | ≤ 𝑅1 = 𝜆1 (9)
𝛥𝑇 𝛥𝜆2 |𝑀| −𝑆𝑛2 𝑆𝑛1 𝛥𝜆2 | | 10𝑄1

Fig. 5. (a) The transmission spectra of the proposed CPCNCS when temperature changes from 300 K to 330 K. (b) Linear fitting of the corresponding resonant wavelength shifts
via varying different temperature changes. (c) The transmission spectra of the proposed CPCNCS when RI changes from 1.330 to 1.345. (d) Linear fitting of the corresponding
resonant wavelength shifts via varying different RI changes.

4
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

|𝛿𝜆2 | ≤ 𝑅2 = 𝜆2 (10) vectors in the case of deviations. ‖𝑀‖ corresponds to 𝐴𝑒 , which can
| | 10𝑄2 be calculated as:
where 𝑅1 and 𝑅2 are the resolutions of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , respectively. 𝜆1
‖[ ]( )
and 𝜆2 are the resonant wavelengths of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , respectively. 𝑄1 𝐴𝑒 = |𝛿|𝑀|| = ‖ ||𝑣2 || cos (𝛼 + 𝜃) + ||𝑣𝑒2 || cos 𝛽2 ||𝑣1 || sin 𝜃 + ||𝑣𝑒1 || sin 𝛽1
( ‖ )[ ]|
and 𝑄2 are the Q-factors of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , respectively. According to − ||𝑣1 || cos 𝜃 + ||𝑣𝑒1 || cos 𝛽1 ||𝑣2 || sin (𝛼 + 𝜃) + ||𝑣𝑒2 || sin 𝛽2 |
|
Eq. (8), 𝛥𝑛 + 𝛿𝑛 can be calculated as: −||𝑣1 ||||𝑣2 || sin 𝛼 ||
( )( ) ( )( )
𝑆𝑇 2 + 𝛿𝑆𝑇 2 𝛥𝜆1 + 𝛿𝜆1 + −𝑆𝑇 1 − 𝛿𝑆𝑇 1 𝛥𝜆2 + 𝛿𝜆2 (19)
𝛥𝑛 + 𝛿𝑛 = (11)
|𝑀| + 𝛿 |𝑀|
where 𝜃 is the angle between 𝒗𝟏 and the horizontal axis. 𝛽1 (𝛽2 ) is the
In the absence of any error, 𝛥𝑛 can be calculated as:
angle between 𝒗𝒆𝟏 (𝒗𝒆𝟐 ) and the horizontal axis. The upper bounds for
𝑆𝑇 2 𝛥𝜆1 − 𝑆𝑇 1 𝛥𝜆2 |𝛿𝑛| and |𝛿𝑇 | alternatively can be written as:
𝛥𝑛 = (12)
|𝑀|
2𝑆𝑇 𝑚 𝑅𝑚 + 2𝛥𝜆𝛿𝑆 2𝐴𝑒 𝑆𝑇 𝑚 𝛥𝜆
and thus by Eqs. (11) and (12) we obtain: |𝛿𝑛| ≤ + (20)
𝐴 𝐴2
(𝑆𝑇 2 +𝛿𝑆𝑇 2 )(𝛥𝜆1 +𝛿𝜆1 )+(−𝑆𝑇 1 −𝛿𝑆𝑇 1 )(𝛥𝜆2 +𝛿𝜆2 ) 2𝑆𝑛𝑚 𝑅𝑚 + 2𝛥𝜆𝛿𝑆 2𝐴𝑒 𝑆𝑛𝑚 𝛥𝜆
𝛿𝑛 = |𝑀|+𝛿|𝑀| |𝛿𝑇 | ≤ + (21)
𝑆𝑇 2 𝛥𝜆1 −𝑆𝑇 1 𝛥𝜆2 (13) 𝐴 𝐴2
− |𝑀| Obviously, increasing A is beneficial to reduce the upper bounds of |𝛿𝑛|
To further simplify Eq. (13), Taylor series is applied to expand the and |𝛿𝑇 |. As shown in Fig. 6(a), the longer the length of 𝒗𝟏 and 𝒗𝟐
following fraction and the higher-order components are ignored: and the closer 𝛼 is to 90 degrees, the greater the value of A. However,
( ) the ideal state of orthogonality is difficult to achieve in the actual
1 1 𝛿 |𝑀|
≈ 1− (14) silicon PhC sensing system. According to the sensing characteristics of
|𝑀| + 𝛿 |𝑀| |𝑀| |𝑀|
silicon PhC microcavity, it is an effective solution to increase the angle
By substituting Eq. (14) into Eq. (13) and ignoring higher-order small between two vectors by designing two modes with opposite polarity
quantities, we can get an approximate expression for 𝛿𝑛: of temperature sensitivities. In addition, decreasing 𝐴𝑒 is beneficial to
𝛿𝑛 ≈
𝑆𝑇 2 𝛿𝜆1 +𝛥𝜆1 𝛿𝑆𝑇 2 −𝑆𝑇 1 𝛿𝜆2 −𝛥𝜆2 𝛿𝑆𝑇 1 reduce the upper bounds of |𝛿𝑛| and |𝛿𝑇 |. To simplify the calculation
|𝑀|
𝛿|𝑀|(𝑆𝑇 2 𝛥𝜆1 −𝑆𝑇 1 𝛥𝜆2 ) (15) of 𝐴𝑒 , the deviations of sensitivity vectors are simplified. Suppose that
−
|𝑀|2 the deviations of the two sensitivity vectors are the same (| 𝒗𝒆𝟏 | = | 𝒗𝒆𝟐 |
Obviously, = | 𝒗𝒆 |). Then we have:
|𝑆𝑇 2 ||𝛿𝜆1 |+|𝛥𝜆1 ||𝛿𝑆𝑇 2 |+|𝑆𝑇 1 ||𝛿𝜆2 |+|𝛥𝜆2 ||𝛿𝑆𝑇 1 | | ( ) ( ) ( )|
𝐴𝑒 = |||𝑣𝑒 || sin 𝛽1 − 𝛽2 + ||𝑣1 ||||𝑣𝑒 || sin 𝜃 − 𝛽2 + ||𝑣2 ||||𝑣𝑒 || sin 𝛽1 − 𝛼 − 𝜃 |
2
|𝛿𝑛| ≤ ‖𝑀‖ | ( ) ( ) ( ) |
|𝛿|𝑀||(|𝑆𝑇 2 ||𝛥𝜆1 |+|𝑆𝑇 1 ||𝛥𝜆2 |) 2| | | | | |
+ (16) ≤ ||𝑣𝑒 || |sin 𝛽1 − 𝛽2 | + ||𝑣1 ||||𝑣𝑒 |||sin 𝜃 − 𝛽2 | + ||𝑣2 ||||𝑣𝑒 |||sin 𝛽1 − 𝛼 − 𝜃 |
|𝑀|2 | | |( |) | |
≤ ||𝑣𝑒 || + ||𝑣1 || + ||𝑣2 || ||𝑣𝑒 ||
2𝑆𝑇 𝑚 𝑅𝑚 +2𝛥𝜆𝛿𝑆 2|𝛿|𝑀||𝑆𝑇 𝑚 𝛥𝜆 2
≤ ‖𝑀‖
+
|𝑀|2
(22)
where 𝑆𝑇 𝑚 represents the maximum of absolute values of temperature
sensitivities. 𝑅𝑚 is the maximum of 𝑅1 and 𝑅2 . 𝛥𝜆 is the maximum of Eq. (22) shows that the value of 𝐴𝑒 is related to the sum of | 𝒗𝟏 | and
absolute values of 𝛥𝜆1 and 𝛥𝜆2 . 𝛿𝑆 is the maximum of absolute values | 𝒗𝟐 |. According to the arithmetic–geometric average inequality [36],
of 𝛿𝑆𝑇 𝑖 and 𝛿𝑆𝑛𝑖 . under the constraint that the product of | 𝒗𝟏 | and | 𝒗𝟐 | is constant, the
Similarly, sum of | 𝒗𝟏 | and | 𝒗𝟐 | takes the minimum value when | 𝒗𝟏 | and | 𝒗𝟐 |
2𝑆𝑛𝑚 𝑅𝑚 + 2𝛥𝜆𝛿𝑆 2|𝛿|𝑀||𝑆𝑛𝑚 𝛥𝜆 are equal.
|𝛿𝑇 | ≤ + (17)
‖𝑀‖ |𝑀|2 In summary, in order to effectively suppress the influence of the
deviations on the detection results, the vectors 𝒗𝟏 and 𝒗𝟐 should
where 𝑆𝑛𝑚 represents the maximum of absolute values of RI sensitivi-
be orthogonal and equal in length as much as possible when de-
ties.
signing a dual-parameter sensor. And the RI sensitivities and tem-
The above analysis process is a quantitative analysis of the de-
perature sensitivities of the sensor should be enhanced as much as
viations. In order to analyze the demodulation stability of the dual-
possible.
parameter sensor more intuitively and make the above analysis results
Although the demodulation deviations of RI and temperature can
guide the design of the sensor more clearly, the vectorization method
be calculated by Eqs. (20) and (21) respectively, the comprehensive
is adopted in the following analysis process.
indicator for the two kinds of demodulation deviations is lacking.
First, we define 𝒗𝟏 (𝒗𝟏 = (𝑆𝑛1 , 𝑆𝑇 1 )) and 𝒗𝟐 (𝒗𝟐 = (𝑆𝑛2 , 𝑆𝑇 2 )) as the
According to Eqs. (20) and (21), both 𝛿𝑆 and 𝐴𝑒 are positively cor-
sensitivity vectors of 𝑐𝑎𝑣1 and 𝑐𝑎𝑣2 , as shown in Fig. 6(a). The lengths
related with 𝑅𝑚 . 𝛥𝜆 is caused by changes in temperature or solution
of the sensitivity vectors are determined by the magnitudes of RI
concentration, independent of the sensor itself. Therefore, the common
sensitivities and temperature sensitivities. The direction of sensitivity
point of the two deviation calculation formulas is that the numerator
vectors is determined by the magnitudes and polarity of RI sensitivities
is positively correlated with 𝑅𝑚 , and the denominator is positively
and temperature sensitivities. From a vector perspective, ‖𝑀‖ is the
correlated with A. The ratio of A and 𝑅𝑚 can be used as a kind of
absolute value of the vector product of 𝒗𝟏 and 𝒗𝟐 , and corresponds to
‘‘Figure of Merit’’ to describe the ability of the sensor to suppress the
the area of the shaded portion in Fig. 6(a). The calculation formula for
demodulation deviations. Here, the demodulation stability factor (𝐷𝑠 )
area (A) is as follows:
is proposed as the comprehensive indicator, which can be expressed
𝐴 = ‖𝑀‖ = ||𝑣1 || ||𝑣2 || sin 𝛼 (18) as:

where 𝛼 is the angle of the two vectors. Fig. 6(b) shows a situation 𝐴
𝐷𝑆 = (23)
where there are deviations in sensitivities. The area difference (𝐴𝑒 ) 𝑅𝑚
between red parallelogram and yellow parallelogram represents the Increasing A by ingenious structural design or decreasing 𝑅𝑚 by opti-
effect of sensitivity vector deviations (𝒗𝒆𝟏 and 𝒗𝒆𝟐 ) on A. The green mizing the performance of the resonator is beneficial to enhancing 𝐷𝑠 .
arrow indicates the deviation of the sensitivity vector while the black The larger the value of 𝐷𝑠 , the smaller the demodulation deviations of
arrow represents the sum of the sensitivity vector and the deviation RI and temperature. According to Eq. (23), the demodulation stability
vector. The two circles represent the value ranges of the sensitivity factor of our proposed sensor is 2497 nm RIU−1 K−1 .

5
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

Fig. 6. (a). The schematic of the sensitivity vectors. The horizontal axis represents temperature sensitivity, and the vertical axis represents RI sensitivity. 𝛼 is the angle between v 𝟏
and v 𝟐 . A represents the area of the shaded parallelogram enclosed by the two vectors. (b) The influence of the deviations of sensitivity vectors on A. The green arrows represent
the deviations of the sensitivity vectors in the actual sensing scene. The red parallelogram and the yellow parallelogram represent the case with and without deviation, respectively.
𝜃 is the angle between v 𝟏 and the horizontal axis. 𝛽 1 (𝛽 2 ) is the angle between v 𝒆𝟏 (v 𝒆𝟐 ) and the horizontal axis. The two gray circles represent the ranges of values for v 𝟏 and v 𝟐
in the case of deviations.

Table 1 a dual-wavelength matrix. For 𝑐𝑎𝑣1 , the RI sensitivity and temperature

Parameter comparison of some published papers.
sensitivity are 728 nm/RIU and −29 pm/K, respectively. For 𝑐𝑎𝑣2 ,
Reference Structure type 𝑆𝑛 𝑆𝑇 (nm/K) 𝐷𝑆 (nm the RI sensitivity and temperature sensitivity are 146 nm/RIU and 58
(nm/RIU) RIU−1 K−1 )
pm/K, respectively. We propose the concept of demodulation stability
[11] PhC 355;0 0.047;−0.108 1708
factor for the dual-parameter sensing system. Through the vectorized
[12] PhC 186;135 0.158;0.147 407
[13] PhC 523;145 0.003;0.06 164
analysis method and combined with the Taylor series and arithmetic–
[15] Optical fiber 56;55 0.143;0.045 < 10 geometric average inequality, we analyze the demodulation stability
[16] Optical fiber −108;−77 0.12;0.08 15 factor of the sensor and give a calculation method. The demodulation
[17] Optical fiber 141;101 −0.036;−0.024 27 stability factor of CPCNCS is up to 2497 nm RIU−1 K−1 , which is
[21] Optical fiber 167;126 0.03;0.035 59
higher than previously reported dual-parameter sensors. We believe
[22] Metasurface 307;204 0.051;0.075 < 10
Our work PhC 728;146 −0.029;0.058 2497 that the proposed sensor can perform superior performance for lab-on-
chip dual-parameter sensing and the concept of demodulation stability
factor will provide a useful reference for the design of dual-parameter
sensors.
Ultimately, a comparison of the performance of our proposed sensor
to those of optical dual-parameter sensors for simultaneous measure-
Declaration of competing interest
ment of RI and temperature is presented in Table 1. 𝑆𝑛 and 𝑆𝑇 are
RI sensitivity and temperature sensitivity, respectively. Compared with
PhC structures in [12,13], the proposed CPCNCS has significant dif- The authors declare that they have no known competing finan-
ference in RI sensitivity, and the polarity of temperature sensitivity cial interests or personal relationships that could have appeared to
is opposite. Although the temperature sensitivity difference of PhC influence the work reported in this paper.
structure in [11] is more significant, coating SU-8 material on the
surface of nanobeam cavity increases the fabrication complexity and Data availability
cost. And the nanobeam cavity covered with SU-8 is isolated from
the analyte. The demodulation stability factors of optical fiber sensors Data will be made available on request.
[15–17,21] are generally 1–2 orders of magnitude lower than our
CPCNCS. In addition, PhC nanobeam sensors are easier to integrate Acknowledgments
on-chip than optical fiber sensors. Compared with the metasurface
structure in [22], the sensing performance of the two sensing modes This research was supported by National Natural Science Founda-
of the nanobeam cascade structure is easier to be independently and tion of China (62275026, 61431003), Key Technologies Research and
flexibly optimized. Therefore, the combination of the slotted cavity and Development Program (2016YFB0402405,2017YFA0205903), BUPT
unslotted cavity has unique advantages in enhancing the demodulation Excellent Ph.D. Students Foundation (CX2022206).
stability of the silicon PhC dual-parameter sensor, and our structure
realizes a balance among demodulation stability factor, sensitivity, and
References
fabrication complexity.
[1] D. Conteduca, I. Barth, G. Pitruzzello, C.P. Reardon, E.R. Martins, T.F. Krauss,
4. Conclusions Dielectric nanohole array metasurface for high-resolution near-field sensing and
imaging, Nature Commun. 12 (2021) 3293.
We propose a high demodulation stability factor dual-parameter [2] W.J. Westerveld, M. Mahmud-Ul-Hasan, R. Shnaiderman, V. Ntziachristos, X.
sensor for simultaneous measurement of RI and temperature. By elabo- Rottenberg, S. Severi, V. Rochus, Sensitive, small, broadband and scalable
optomechanical ultrasound sensor in silicon photonics, Nat. Photon. 15 (2021)
rately designing a slotted nanobeam cavity and an unslotted nanobeam
341–345.
cavity, the electric fields of the two resonant cavities are localized [3] Y.L. Wang, Z.H. Han, Y. Du, J.Y. Qin, Ultrasensitive terahertz sensing with
in different regions, thereby forming distinctive RI sensitivities and high-q toroidal dipole resonance governed by bound states in the continuum
temperature sensitivities. The RI and temperature are demodulated by in all-dielectric metasurface, Nanophotonics 10 (2021) 1295–1307.

6
Z. Han, C. Wang, J. Wang et al. Optics Communications 530 (2023) 129063

[4] Y. Zhao, Q.L. Wu, Y.N. Zhang, Simultaneous measurement of salinity temperature [20] Y. Li, G.F. Yan, S.L. He, Thin-core fiber sandwiched photonic crystal fiber modal
and pressure in seawater using optical fiber SPR sensor, Measurement 148 (2019) interferometer for temperature and refractive index sensing, IEEE Sens. J. 18
106792. (2018) 6627–6632.
[5] M.I. Zibaii, H. Latifi, A. Asadollahi, A.H. Bayat, L. Dargahi, A. Haghparast, Label [21] J.F. Zhao, P.P. Niu, C. Zhang, H. Bai, X.D. Sun, Z.B. Han, Simultaneous refractive
free fiber optic apta-biosensor for in-vitro detection of dopamine, J. Lightw. index and temperature measurement using nested fiber balloon rings, Appl. Opt.
Technol. 34 (2016) 4516–4524. 57 (2018) 6835–6839.
[6] D. Rho, S. Kim, Label-free real-time detection of biotinylated bovine serum [22] J. Hu, T.T. Lang, G.H. Shi, Simultaneous measurement of refractive index
albumin using a low-cost optical cavity-based biosensor, Opt. Express 26 (2018)
and temperature based on all-dielectric metasurface, Opt. Express 25 (2017)
18982–18989.
15241–15251.
[7] D. Kalantarov, C.P. Search, Highly sensitive label-free coupled resonator Fabry–
[23] Y.Z. Cheng, Z.R. Li, Z.Z. Cheng, Terahertz perfect absorber based on InSb
Perot self-referencing photonic biosensor, J. Opt. Soc. Amer. B 34 (2017)
metasurface for both temperature and refractive index sensing, Opt. Mater. 117
968–975.
[8] Y. Yang, Z. Wang, F. Sun, H. Tian, Modeling and design of a coupled PhC slab (2021) 111129.
sensor for simultaneous detection of refractive index and temperature with strong [24] F. Chen, Y.Z. Cheng, H. Luo, Temperature tunable narrow-band Terahertz
anti-interference ability, Opt. Express 28 (2020) 22151–22164. metasurface absorber based on InSb micro-cylinder arrays for enhanced sensing
[9] X. Zhang, G. Zhou, P. Shi, H. Du, T. Lin, J. Teng, F.S. Chau, On-chip integrated application, IEEE Access 8 (2020) 82981–82988.
optofluidic complex refractive index sensing using silicon photonic crystal [25] X.Y. Lu, A. McClung, K. Srinivasan, High-q slow light and its localization in a
nanobeam cavities, Opt. Lett. 41 (2016) 1197–1200. photonic crystal microring, Nat. Photonics 16 (2022) 66–71.
[10] .J. Wang, Z. Fu, F. Sun, Z. Wang, C. Wang, H. Tian, Multiplexing dual-parameter [26] Q. Quan, M. Loncar, Deterministic design of wavelength scale ultra-high Q
sensor using photonic crystal multimode nanobeam cavities, Opt. Commun. 427 photonic crystal nanobeam cavities, Opt. Express 19 (2011) 18529–18542.
(2018) 382–389. [27] Z. Han, C. Wang, Y.Y. Liu, H.P. Tian, Simultaneous detection of complex
[11] X.P. Li, C. Wang, Z. Wang, Z.Y. Fu, F.J. Sun, H.P. Tian, Anti-external interference refractive index and temperature using a compact side-coupled photonic crystal
sensor based on cascaded photonic crystal nanobeam cavities for simultaneous nanobeam cavity, J. Opt. Soc. Amer. B 38 (2021) 2765–2774.
detection of refractive index and temperature, J. Lightw. Technol. 37 (2019) [28] Q. Quan, P.B. Deotare, M. Loncar, Photonic crystal nanobeam cavity strongly
2209–2216. coupled to the feeding waveguide, Appl. Phys. Lett. 96 (2010) 203102.
[12] F. Sun, J. Wei, B. Dong, Y. Ma, Y. Chang, H. Tian, C. Lee, Coexistence of air and
[29] J. Zhou, H.P. Tian, L.J. Huang, Z.Y. Fu, F.J. Sun, Y.F. Ji, Parabolic tapered
dielectric modes in single nanocavity, Opt. Express 27 (2019) 14085–14098.
coupled two photonic crystal nanobeam slot cavities for high-FOM biosensing,
[13] Z. Wang, Z. Fu, F. Sun, C. Wang, W. Zhou, H. Tian, Simultaneous sensing
IEEE Photonics Technol. Lett. 29 (2017) 1281–1284.
of refractive index and temperature based on a three-cavity-coupling photonic
[30] P. Saha, M. Sen, Ultrahigh Q-factor and ultrasensitive refractive index sensor
crystal sensor, Opt. Express 27 (2019) 26471–26482.
based on a multiple-slot photonic crystal cavity, IEEE Trans. Instrum. Meas. 70
[14] F.D. Yu, P. Xue, X.W. Zhao, J. Zheng, Simultaneous measurement of refrac-
tive index and temperature based on a peanut-shape structure in-line fiber (2021) 9504509.
Mach–Zehnder interferometer, IEEE Sens. J. 19 (2019) 950–955. [31] Q.M. Quan, I.B. Burgess, S.K.Y. Tang, D.L. Floyd, M. Loncar, High-q, low index-
[15] X.Q. Ni, M. Wang, D.M. Guo, H. Hao, J.L. Zhu, A hybrid mach–zehnder contrast polymeric photonic crystal nanobeam cavities, Opt. Express 19 (2011)
interferometer for refractive index and temperature measurement, IEEE Photonics 22191–22197.
Technol. Lett. 28 (2016) 1850–1853. [32] L.J. Huang, J. Zhou, F.J. Sun, Z.Y. Fu, H.P. Tian, Optimization of one dimensional
[16] M.L. Xiong, H.P. Gong, Z.P. Wang, C.L. Zhao, X.Y. Dong, Simultaneous refractive photonic crystal elliptical-hole low-index mode nanobeam cavities for on-chip
index and temperature measurement based on Mach–Zehnder interferometer sensing, J. Lightwave Technol. 34 (2016) 3496–3502.
concatenating two Bi-tapers and a long-period grating, IEEE Sens. J. 16 (2016) [33] L.J. Huang, D.J. He, X.W. Mi, J.Q. Ding, S.H. Chen, X.L. Peng, Photonic crystal
4295–4299. elliptical-hole tapered low-index-mode nanobeam cavities for sensing, Appl. Opt.
[17] S.H. Liu, H.L. Zhang, L.T. Li, L.M. Xiong, P.P. Shum, Liquid core fiber interfer- 57 (2018) 9822–9827.
ometer for simultaneous measurement of refractive index and temperature, IEEE [34] J.D. Joannopoulos, S.G. Johnson, J.N. Winn, R.D. Meade, Photonic Crystals:
Photonics Technol. Lett. 31 (2019) 189–192. Molding the Flow of Light, second ed., Cambridge University Press, 2007.
[18] X. Li, L.V. Nguyen, M. Becker, H. Ebendorff-Heidepriem, D. Pham, S.C. Warren-
[35] M. Mohammadi, M. Seifouri, Numerical investigation of photonic crystal ring
Smith, Simultaneous measurement of temperature and refractive index using an
resonators coupled bus waveguide as a highly sensitive platform, Photon.
exposed core microstructured optical fiber, IEEE J. Sel. Top. Quantum 26 (2020)
Nanostruct. 34 (2019) 11–18.
5600107.
[36] M.R. Spiegel, S. Lipschutz, J. Liu, Schaum’s outline of theory and problems
[19] B. Yin, S.H. Wu, M.G. Wang, W.Q. Liu, H.S. Li, B.L. Wu, Q.C. Wang, High-
of mathematical handbook of formulas and tables, 3nd ed., McGraw-Hill Book
sensitivity refractive index and temperature sensor based on cascaded dual-
wavelength fiber laser and SNHNS interferometer, Opt. Express 27 (2019) Company, New York, 2009.
252–264.