PCCP

View Article Online

PAPER View Journal | View Issue
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Titanium nitride-based hyperbolic metamaterial

for near-infrared ultrasensitive sensing of
Cite this: Phys. Chem. Chem. Phys.,
2024, 26, 10273 microbes†
Dip Sarker and Ahmed Zubair *

We demonstrate an ultrasensitive microbe sensor for the first time using a titanium nitride (TiN)
nanowire-based hyperbolic metamaterial (HMM) structure. We detected the change in the resonance
wavelength shift due to the inclusion of microbes in a freshwater environment employing the finite-
difference time-domain (FDTD) method. Our proposed HMM sensor exhibits strong bulk plasmon
polariton (BPP) modes in the anisotropic hyperbolic regime (l Z 590 nm) and operates in the near-
infrared (NIR) wavelength region. We studied the impact of structural parameters on the resonance
wavelength shift, where our proposed HMM sensor structure exhibited an outstanding sensing capability
of 11 nm per bacteria. A limit of detection of 0.00008 RIU was achieved for our proposed HMM sensor
structure. Additionally, we verified our results theoretically to calculate mode frequency shift by solving
the effective medium theory (EMT). Our study revealed that HMM is the origin of highly sensitive BPP
modes. We obtained two BPP modes, where the BPP mode at a longer wavelength (q = 1) exhibited the
Received 23rd January 2024, highest resonance wavelength shift compared to the BPP mode at a shorter wavelength (q = 2). More
Accepted 1st March 2024 importantly, we demonstrated numerically the point-detection capability of our proposed HMM microbe
DOI: 10.1039/d4cp00302k sensor structure, which was unattainable in previously reported sensor work. Moreover, this sensor can
be adapted to detect different viruses and bacteria. Our proposed TiN-based HMM structure can
rsc.li/pccp potentially be an ultrasensitive and straightforward microbe sensor for label-free detection.

1 Introduction caused by E. coli can lead to potentially fatal issues such as

abdominal discomfort, diarrhoea, and vomiting. Existing plas-
Plasmonic sensors are conducive to biological investigation monic sensors show promising detection capability; however,
and disease diagnostics because they can rapidly and effectively tiny molecules of smaller weight, such as E. coli bacteria, are
detect biomolecular interactions. In particular, surface plas- challenging to detect in very diluted solutions because of their
mon resonance (SPR) can occur on a nanoparticle-like bound lower polarizability.6 Thus, simple, easy-to-operate, and real-
surface or at the interface between the metal and the dielectric. time detection kits capable of detecting single E. coli bacteria,
These oscillations yield an evanescent field highly susceptible are needed.
to variations in the refractive index of the biomolecules.1 Holowko et al. proposed a chemical biosensor that can
Researchers have recently shown enormous interest in harnes- detect and eliminate Vibrio cholerae;7 beneficial in preventing
sing their unique properties for diagnosing diseases. Numerous and treating cholera. Lubkowicz et al. designed a novel
works on various detection kits, such as tumour cells, cancer chemical biosensor to detect Staphylococcus aureus manipulat-
cells, Alzheimer’s disease biomarkers, and SARS-CoV-2 spike ing Lactobacillus reuteri.8 However, handling these chemical
protein, have been developed using SPR.2–4 Moreover, diseases biosensors requires expertise, and expensive chemicals are
from numerous water-borne bacteria, such as Escherichia coli sometimes unobtainable at point-of-care (POC). Mechanical
(E. coli), in drinking water are concerning due to pollution, as sensors can detect single biomolecules;9,10 however, it is hard
the majority of people live near freshwater sources.5 Infections to differentially fictionalize sensors that are close together (a
front-end problem) and multiplex the electrical readout of a
dense collection of devices (a back-end problem).11 Fu et al. and
Department of Electrical and Electronic Engineering, Bangladesh University of
Giden proposed one- and two-dimensional photonic crystal
Engineering and Technology, Dhaka 1205, Bangladesh.
E-mail: [email protected]
sensor structures.12,13 The quality factors were good enough
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/ to make a high-quality resonator; however, the sensitivity was
10.1039/d4cp00302k poor, which is essential for sensors. Moreover, the unit cell size

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10273
 View Article Online

Paper PCCP

of the photonic device is comparable to the wavelength of light; nanowire HMM sensors. Moreover, the impact of resonance
therefore, miniaturization is not possible. Several sensor shifts due to the incidence of microbes on different nanowire
structures, such as nanocomposites and 1D photonic crystals, regions has not been studied yet; therefore, there is vast scope
exhibit promising sensing capability.14,15 However, these in designing nanowire HMM sensor structures using different
proposed structures suffer from an inadequate limit of detec- materials and analyzing the impact of resonance shift when
tion (LOD) with small coverage of the light spectrum. On the biomolecules are incident on different nanowire regions.
contrary, as metamaterial-based artificially engineered materi- In this work, we proposed a TiN nanowire-based HMM
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

als that can manipulate electromagnetic (EM) waves,16 optical sensor structure employing effective medium theory (EMT)
biosensor structures can bring photonics to nanophotonics; and finite-difference time-domain (FDTD) techniques. Firstly,
however, optical losses can cause a detrimental effect on these we solved the Maxwell Garnett equations and obtained the
devices. For these reasons, researchers have garnered interest hyperbolic dispersion region for the TiN nanowire HMM struc-
in engineering biosensor structures due to their unique EM ture. Secondly, we selected the structural parameters using a
characteristics. However, smaller, lighter biomolecules are not numerical sweep method. Thirdly, we examined the sensing
detectable because of their reduced polarizability. In addition, capability of our nanowire HMM structure with a slight change
the sensor figure of merit (FOM) is quite small. Hyperbolic in refractive index (RI) due to injecting microbes. In addition,
metamaterials (HMMs), artificially engineered anisotropic we compared the resonance shifts of BPP modes for freshwater
devices, can alleviate this issue and exhibit excellent sensitivity and injected microbes in the freshwater medium. Moreover, we
due to improved light–matter interactions. studied the mode frequency shift, a determining parameter for
HMMs exhibit hyperbolic dispersion, showing different the sensor’s sensitivity, solving EMT equations. Lastly, we
optical properties in different directions. In particular, they studied the impact of the position and coverage of E. coli on
exhibit a positive permittivity in one direction and a negative resonance shifts. Additionally, we provide a comprehensive
permittivity in another.17 Therefore, the performances of many comparative performance analysis of our sensor and previously
applications, such as negative refraction, hyperlens, and lasing reported biosensor structures.
emission, were enhanced.18–20 This anisotropic property allows
the sensor to detect a single biomolecule or tiny microbe with
high sensitivity. Compared to chemical, mechanical, metal, and
2 Methodology
photonic sensor structures, HMM-based sensor structures are A periodic nanowire array of TiN was designed to observe
easy to build, cost-effective, and have adjustable resonance ultrasensitive sensing performance in a broad optical wave-
characteristics. Multilayered and nanowire structures are the length range of the NIR region. Fig. 1(a) shows the proposed
two types of HMM structure. Researchers have reported numer- TiN nanowire HMM sensor structure on a thin TiN slab. The
ous multilayered HMM sensor structures with high sensitiv- proposed sensor structure was periodic in the x- and y-
ity;21,22 however, nanowire HMM sensor structures are more directions. Fig. 1(b) and (c) illustrate the xy and yz views of
suitable for detecting smaller and lighter biomolecules. Yan the sensor’s unit cell. We selected all the structural parameters
et al. reported a gold (Au) nanowire HMM sensor structure for of our proposed sensor structure using a parametric sweep. The
monitoring the binding of streptavidin, a protein purified for periodicity, P, of the proposed HMM structure was set to
bacteria.23 Vasilantonakis et al. and Yan et al. proposed Au 100 nm. We adopted a TiN thin layer, hbase, of 60 nm to support
nanowire HMM sensor structures with tuned refractive index the nanowire HMM structure. On top of the TiN slab, TiN
for enhancing light–matter interactions.24,25 These designed nanowires were designed to obtain hyperbolic dispersion and
nanowire HMM structures were limited to noble metals, such anisotropic properties. The height, hp, of the nanowire was set
as Au and Ag. However, some concerning issues with conven- to 400 nm for our proposed sensor structure. We adopted a
tional plasmonic materials, such as optical losses from limited radius, r, of the nanowire using,
intraband (Drude) and interband (Lorentz) transitions, moti-
pr2
vated researchers to find a more convenient alternative.26 The FF ¼ : (1)
P2
alternative plasmonic materials should have less optical loss,
be chemically stable, less prone to surface roughness, and have Here, FF denotes the fill factor of the nanowire. We used an FF
a higher value of real permittivity. Besides, their carrier concen- of 70% for our study. The detailed selection process is provided
tration should be easily tunable, and most importantly, in Section 3.3. We could employ electron beam lithography
their fabrication process should be compatible with growth (EBL) and electroplating techniques to fabricate the TiN nano-
techniques.27 TiN, an alternate plasmonic and gold luster wire HMM structure. EBL can achieve a well-structured periodic
material, can be considered a replacement for noble metals array of holes that can be removed using etchant. Yan et al.
because stoichiometric TiN films on sapphire or other trans- fabricated a Au nanorod-based HMM structure on the
parent substrates, exhibit lower imaginary permittivity profiles glass–metal hybrid substrate using EBL and electroplating
in visible and near-infrared regions than noble metals and methods.23 A detailed suggested fabrication technique is pro-
other transition metal nitrides.26 Moreover, the tunability of vided in Section S1 of the ESI.† Moreover, a high FF can be
plasmonic resonance can vary by changing the carrier concen- achieved using the electroplating technique.29 The complex
tration of TiN.28 TiN can play a pivotal role in designing refractive index of TiN was adopted from Naik et al.26 (see

10274 | Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 This journal is © the Owner Societies 2024
 View Article Online

PCCP Paper
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 1 (a) Schematic illustration of the periodic nanowire hyperbolic metamaterial structure. Here, y and d are the polarization and incidence angles of a
plane wave. (b) The xy- and (c) yz-view of the periodic nanowire hyperbolic metamaterial structure’s unit cell. Here, r, hp, hbase, and P denote the
nanowire radius, nanowire height, base height, and cell periodicity, respectively.

Section S2 of ESI,† for details). Performing the sensing ability of wavelength span from 300 to 3000 nm for conducting our
our proposed sensor structure, we submerge the nanowire study, where the incidence angle, d, was set to 301. We
portion in a freshwater medium. We considered a freshwater employed power monitors to calculate our proposed sensor
RI of 1.333 RIU for our study. We collected the RI of single structure’s transmittance, reflectance, and absorbance. The
E. coli bacteria (difference of RI, D = 0.005 RIU) from an transmittance, reflectance, and absorbance were solved using,
experimental report of optofluidic immersion refractometry.30
We performed numerical and analytical calculations Pr
TðlÞ ¼ ; (2)
employing FDTD (Ansys Lumerical)31 and EMT24 techniques. Pi
More specifically, we used a 3D FDTD analysis method to
measure how much light was reflected by the proposed HMM Pt
RðlÞ ¼ ; (3)
sensor structure. As FDTD is a time-consuming technique, Pi
boundary conditions can save computational space and time.
Thus, we utilized periodic boundary conditions in the x- and y- and
directions and perfectly matched layers (PMLs) in the z-
A(l) = 1  T(l)  R(l). (4)
direction to absorb light, which was beyond the FDTD simula-
tion region. Moreover, we employed non-uniform mesh to Here, l denotes the wavelength of the plane wave. T(l), R(l),
conduct our numerical studies within limited memory storage and A(l) represent the transmittance, reflectance, and absor-
and time. We used a mesh-grid size of 1 nm in the x- and y- bance of the proposed sensor structure as a function of l,
directions and 3 nm in the z-direction. We stopped our numer- respectively. Pt, Pi, and Pr are the transmitted, incident, and
ical calculations when the energy level reached 105 of its reflected powers, respectively. We set the temperature to 300 K
initial value. As our nanowire HMM structure is symmetrical, in our study. The simulation region was equal to the structural
transverse electric (TE) and transverse magnetic (TM) polarized parameter P and 4500 nm in the z-axis. Additionally, we
plane waves will not show different results. Here, polarization calculated the hyperbolic dispersion region and iso-frequency
angle y = 01 and 901 denote the TM and TE polarized plane curve for our proposed structure. We analytically studied the
waves, respectively. We used a TM-polarized plane wave with a mode frequency shift for our proposed structure employing

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10275
 View Article Online

Paper PCCP
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 2 (a) The enumerated isofrequency surface curve for our proposed nanowire HMM sensor structure. Here, we adopted ed,eff > 0 and ez,eff o 0 for
our calculation. (b) The relative permittivity of our proposed TiN nanowire HMM sensor structure using effective medium theory. We obtained a
hyperbolic dispersion at l Z 590 nm (shaded portion). (c) Depiction of the ez,eff for varying FFs. A large FF provided a strong hyperbolic photonic density of
states (PDOS), resulting in strong hyperbolic properties. (d) Reflectance spectra of the proposed sensor structure with and without HMM. No BPP modes
were obtained without HMM. Here, q denotes the mode of our proposed HMM structure.

EMT equations. A more elaborate discussion can be found in constants were symmetrical in the x and y directions; however,
the later sections. the propagation constant in the z direction was asymmetrical,
which was the main reason for obtaining anisotropic properties
for our proposed HMM structure, as shown in Fig. 2(a). More-
3 Results and discussion over, we calculated the effective permittivity to obtain the
3.1 HMM properties hyperbolic dispersion region of our proposed HMM structure.
Most materials are isotropic, which means the optical proper- The effective perpendicular (ed,eff) and parallel (ez,eff) compo-
ties do not alter for different incident angles. Moreover, they nents of permittivity can be found using Maxwell-Garnet
exhibit spherical iso-frequency contours. Contrarily, the sphe- approximations,33
rical contour becomes an ellipsoid when objects show aniso-
ðFF þ 1Þem eh þ eh2 ð1  FFÞ
tropic properties. The dispersion relation can be defined by,32 ed;eff ¼ ; (6)
ðFF þ 1Þeh þ ð1  FFÞem
kx2 ky2 kz2
þ þ ¼ ko2 : (5) and
ex;eff ey;eff ez;eff
ez,eff = FFem + (1  FF)eh. (7)
Here, kx, ky, kz, and ko denote the wave vector in the x-, y-, z-
directions, and free space, respectively. ex,eff, ey,eff, and ez,eff are Here, em and eh are the permittivity of TiN and the surrounding
the permittivity of the x-, y-, z-directions, respectively. HMM has environment of TiN nanowire, respectively. We obtained a
two types of hyperbolic dispersion: when the HMM structure hyperbolic dispersion at l Z 590 nm where, ed,eff > 0 and
exhibits ed,eff > 0 and ez,eff o 0, we name it type 1 HMM (here, d ez,eff o 0, as shown in Fig. 2(b). A large FF provides a strong
denotes the x- and y-directions); contrarily, we denote HMM hyperbolic photonic density of states (PDOS), as shown in
structure as type 2 when the structure exhibits ed,eff o 0 and Fig. 2(c), which is conducive for detecting smaller and lighter
ez,eff > 0. We obtained ed,eff > 0 and ez,eff o 0 from our dispersion microbes. Therefore, we utilized an FF of 70% for our study.
relation calculation because our proposed structure exhibits Additionally, we studied the impact of HMM on reflectance
the type 1 HMM properties. Dispersion relations reveal that spectra. No BPP modes were found without HMM, as depicted
HMM structures exhibit anisotropic properties. Propagation in Fig. 2(d). Contrarily, we obtained two BPP modes for the

10276 | Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 This journal is © the Owner Societies 2024
 View Article Online

PCCP Paper
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 4 Illustration of the impacts of the geometric parameters (a) hbase, (b)
hp, and (c) FF on the reflectance spectra for the TM-polarized plane wave.

Fig. 3 (a) Reflectance spectra of our HMM sensor for various incidence
angles, d, of TM-polarized plane wave. The dips were blue-shifted with
complicated to fabricate; therefore, we utilized 60 nm for our
increasing incidence angle. The inset shows an illustration of the light’s
incidence angle. (b) Reflectance spectra of our HMM sensor for various study. Contrarily, the number of BPP modes increased and
polarization angles. We observed no shift in the reflectance spectra dips redshifted due to the reliance of mode frequency shift on hp
due to the symmetric nature of the nanowire HMM structure. Here, 01 and (see Section 3.4 for details), while we increased the hp of our
901 denote the TM and TE polarized plane wave. proposed structure, as shown in Fig. 4(b). The full-width half-
maximum (FWHM) increased with the inclination of hp and
reduced the quality factor; thus, we adopted a hp of 400 nm to
HMM structure. Therefore, it can be inferred that HMM is the
obtain a narrow FWHM for our work. Additionally, we varied
main reason to exhibit BPP modes. The BPP modes of our
the radius of the nanowire and studied the impact of FF on the
proposed sensor structure are denoted by q. As the relative
reflectance spectra. The sensing range of the proposed struc-
permittivity at the hyperbolic region is negative, the propaga-
pffiffi ture covered a wide range in the optical wavelength due to the
tion constant, k ¼ eo; will be imaginary, which is not possible
shift of the hyperbolic dispersion region, as illustrated in
to obtain in the real world. Thus, the transmittance is negli-
Fig. 4(c). Moreover, the sensitivity increased with increasing
gible for our structure, as shown in Fig. S3 of ESI.† Conse-
FF due to strong hyperbolic dispersion. However, the nanowire
quently, we achieved absorbance at resonance wavelengths of
of the HMM structure overlapped while we utilized higher FF
1267 and 1935 nm, as shown in Fig. S3 of ESI.†
beyond 70%, which diminished the BPP modes (see Fig. S4 of
3.2 Anisotropic properties ESI†). Therefore, we used an FF of 70% for further studies.
Our proposed nanowire HMM sensor structure exhibited ani- 3.4 Dependence of mode frequency shift on the RI of E. coli
sotropic properties at the hyperbolic dispersion region (l Z
We analytically calculated the mode frequency shift using the
590 nm) due to the change in the effective modal index, nmodal,
EMT method. The mode frequency shift revealed the shift of
of plasmonic guided modes at different incidence angles of TM-
the BPP modes and the sensitivity of the HMM structure. The
polarized plane light. The nmodal is defined by,34
mode frequency shift is defined by,24
lo "   2 #
nmodal ¼ no sin d: (8) @oq co2 kx 2 @ez;eff qp @ed;eff
P ¼ þ ; (9)
@eh;r 2oq ez;eff @eh;r hp ed;eff @eh;r
Here, lo, d, and no denote the resonance wavelength, inci-
dent angle of the plane wave, and refractive index of the and
incident medium, respectively. As evident from Fig. 3(a), reso- " 2  2 #
@oq co2 kx @ez;eff qp @ed;eff
nance wavelength shifts in reflectance spectra were blueshifted i ¼i þ : (10)
@eh;i 2oq ez;eff @eh;i hp ed;eff @eh;i
with increasing d of TM-polarized plane light, which corrobo-
rates with eqn (8). The inset of Fig. 3(a) depicts the incidence @o q @o q
angle of the TM-polarized plane wave. Additionally, we analyzed Here, and denote the real and imaginary part of mode
@eh;r @eh;i
the impact of y on the reflectance spectra of our proposed
frequency shift by changing the RI due to introduction of the
sensor structure. Our proposed nanowire HMM structure was
E. coli bacteria in the freshwater solutions. The oq represents
symmetrical; thus, we did not observe any resonance wave-
the mode frequency. The co and eh represent the speed of light
length shift with varying y, as can be seen in Fig. 3(b).
at vacuum and the RI of the surrounding environment of TiN
3.3 Impact of geometrical parameters on reflectance spectra nanowire, which was a single E. coli bacteria in the freshwater
solutions for our study, respectively. The partial derivatives of
We investigated the effect of structural parameters – hbase, hp, effective permittivities with respect to eh are provided by,
and FF – on the reflectance spectra and adopted parameters of
our proposed nanowire HMM structure. Fig. 4 illustrates the @ed;eff @ed;eff mem þ 2eh mem þ eh
¼ i ¼  meh ; (11)
impact of the geometrical parameters on reflectance spectra. @eh;r @eh;i meh þ em ðmeh þ em Þ2
Our proposed sensor structure exhibited no resonance wave-
length shifts while we changed the hbase of our nanowire HMM @ez;eff @ez;eff
¼ i ¼ 1  FF; (12)
structure, as depicted in Fig. 4(a). The thin TiN film could be @eh;r @eh;i

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10277
 View Article Online

Paper PCCP
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 5 The (a) fundamental (q = 1) and second (q = 2) mode frequency shifts with the permittivity of the surrounding medium, eh, for a FF of 70% and hp of
400 nm. (b) The calculated mode frequency shift of q = 1 BPP mode for different FF, where hp was set to 400 nm. (c) The calculated mode frequency shift
of fundamental and (d) second BPP modes for different nanowire heights. We enumerated all the mode frequency shifts utilizing the EMT equations.

and frequency shift for the two BPP modes, where we considered
a single E. coli bacteria as the surrounding medium (n = 1.338).
1 þ FF
m¼ : (13) We observed a large increase in the mode frequency shift for
1  FF
longer wavelength BPP mode to the host medium RI variations
The propagation constant, kx, in the x-direction and oq are compared to shorter wavelength BPP mode, which resulted in
defined by, high sensitivity for q = 1 BPP mode, as shown in Fig. 5(a).
 2   Similar findings were obtained using the FDTD method (see in
qp ez;eff the later section). The mode frequency shift relies on the BPP
kx2 ¼ ez;eff ko2  ; (14)
hp ed;eff mode’s spatial wavelength hp/qp, increasing with the increase
of hp and the decrease of BPP mode. Interestingly, this leads to
and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi an increase in the mode sensitivity, a trend similar to that
 2 observed with conventional SPR or LSPR transducers. More-
kx2 qp 1
oq ¼ co þ : (15) over, we calculated the mode frequency shift of the fundamen-
ez;eff hp ed;eff
tal (q = 1) BPP mode by varying the FF of the nanowire, where
Here, ko represents the propagation constant in the vacuum, other parameters were kept the same. A higher mode frequency
which was enumerated by o/co, and o is the frequency of the shift was obtained when we increased the FF, as shown in
incident plane wave. The mode position dependence to the real Fig. 5(b). Meanwhile, this happened due to the shift of BPP
part of the constituents’ permittivities is plotted in Fig. 5 within mode in the longer wavelength. The calculated mode frequency
the 1st Brillouin zone to ensure EMT validity. The mode shift of the second (q = 2) BPP mode enhanced with increasing
frequency shift relies on the real part of the eh’s permittivity, FF of the TiN nanowire, as depicted in Fig. S5 of ESI.†
 
@on Furthermore, we analyzed the impact of hp on the mode
where the mode frequency shift of the host is stronger
@e frequency shift and the sensitivity of the proposed sensor
  h
@on structure analytically. The mode frequency shift and sensitivity
than the mode frequency shift of TiN . Therefore, we
@em increased with the increase of hp, as depicted in Fig. 5(c) and
@on (d). The q = 1 BPP mode exhibited the highest mode frequency
enumerated for various cases within the 1st Brillouin zone
@eh shift compared to the q = 2 BPP mode due to the strong
to maintain the EMT validation. Fig. 5(a) shows the mode hyperbolic dispersion.

10278 | Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 This journal is © the Owner Societies 2024
 View Article Online

PCCP Paper
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 6 (a) Reflectance spectra of the proposed nanowire HMM sensor structure after adding a single E. coli bacteria into fresh water. Two BPP modes
were redshifted due to the addition of E. coli bacteria in fresh water. Red-shifted reflectance spectra of HMM sensor structure for (b) q = 1 and (c) q = 2
modes. The proposed structure provided 11 and 5.41 nm per bacteria shifts after injection of a single E. coli bacteria.

3.5 Sensing properties can play a vital role in biomedical applications by changing the
Our proposed sensor structure exhibited two BPP modes in the functionality group of microbes. Thus, we studied the impact of
hyperbolic region, as illustrated in Fig. 6(a), where we consid- different viruses and bacteria on resonance wavelength shifts.
ered the freshwater as our simulation surrounding medium. We obtained resonance wavelength shifts in the BPP modes of
Due to HMM’s anisotropic properties, these modes were highly our proposed structure for various microbes (see Section S6 of
sensitive towards smaller and lighter microbes. Therefore, we ESI,† for details).
designed our highly sensitive E. coli sensor using these modes. 3.6 Impact of multiple E. coli bacteria on the resonance shift
To observe the sensing performance of our proposed sensor
structure, we varied a slight change in the surrounding med- Multiple E. coli bacteria could be cultured in a small place. To
ium’s RI due to injecting single E. coli bacteria into freshwater consider the practical case, we varied the RI of the E. coli
solutions. Thus, the proposed HMM sensor structure red- bacteria and analyzed the resonance shift in the reflectance
shifted the resonance dips in the reflectance spectra, as spectra. Various RI values reflected the concentration of E. coli
depicted in Fig. 6(a). In addition, we compared the reflectance bacteria existing in the aqueous solution. Fig. 7(a) depicts the
spectra of freshwater and the aqueous solution containing reflectance spectra of our proposed nanowire HMM sensor
single E. coli, and calculated the resonance wavelength shift. structure with varying amounts of E. coli bacteria. Resonant
We obtained resonance wavelength shifts of 11 and 5.41 nm per dips redshifted with increasing number of E. coli bacteria. We
bacteria for q = 1 and q = 2 BPP modes, respectively, as shown in obtained the highest resonant shift for our proposed sensor
Fig. 6(b) and (c). The sensitivity, S, which is a performance structure’s q = 1 BPP mode because the longer wavelength of
parameter of the sensor, was calculated using,21 the HMM structure exhibited the highest hyperbolic disper-
sion, as shown in Fig. 7(b). We achieved the highest resonant
Dl
S¼ : (16) shifts of B127 and B75 nm for q = 1 and q = 2 BPP modes,
Dn respectively, where we utilized 17 E. coli bacteria on the
Here, Dl and Dn denote the resonant wavelength shift for fresh proposed sensor structure. We formulated empirical equations
water and the difference in the refractive index of the surround- for the BPP mode for the RI dependency of resonance wave-
ing medium, respectively. We achieved an S of 2200 and length shift given by,
1082 nm per bacteria for q = 1 and q = 2 BPP modes,
dlr,1 = 0.2 + 7.77NE.coli  0.22NE.coli2 + 0.01NE.coli3, (17)
respectively. A multitasking detecting kit with a single design
and

dlr,2 = 0.28 + 4.6NE.coli  0.13NE.coli2 + 0.01NE.coli3. (18)

Here, dlr,n denotes the resonance wavelength shift of the n-th

BPP mode. NE.coli is the number of E. coli bacteria. The
coefficient of determination, R2, is a statistical measure of
how well the regression predictions approximate our real data
points. Here, the R2 for these empirical equations was >0.995.
Additionally, we studied the LOD of our proposed TiN nanowire
HMM sensor structure. An LOD of 0.00008 RIU was obtained
for our proposed sensor structure, which was substantially
Fig. 7 (a) Reflectance spectra of our proposed TiN nanowire HMM sensor
structure for varying the number of E. coli bacteria in fresh water.
lower compared to previously reported works. We obtained a
(b) Calculated resonance shift of dips for injecting multiple E. coli bacteria resonance wavelength shift of 0.27 nm for both BPP modes in
in fresh water. the LOD case (see Section S7 of ESI,† for details).

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10279
 View Article Online

Paper PCCP
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

Fig. 8 Numerically calculated resonance shift of dips with varying FF for (a) q = 1 and (b) q = 2. Numerically calculated resonance shift of dips with
varying hp for (c) q = 1 and (d) q = 2.

3.7 Dependency of geometrical parameter on resonance shift adopted the hp of 400 nm to maintain a narrow FWHM with
We varied the geometrical parameters, such as FF and hp, and high sensitivity.
NE.coli to analyze the resonance wavelength shift in the reflec-
tance spectra of our proposed nanowire HMM sensor structure. 3.8 Point detection capability
We varied the FF from 0.5 to 0.7 and calculated the resonance 3.8.1 Dependency of E. coli’s surface coverage on reso-
wavelength shift. A higher FF provides strong hyperbolic dis- nance shift. E. coli bacteria cover only a small portion of the
persion properties, which resulted in highly sensitive BPP nanowire HMM structure rather than covering the full nano-
modes, as shown in Fig. 2(c). Thus, we obtained the highest wire. We call it a point detection scheme of E. coli bacteria. This
resonance wavelength shift for the FF of 0.7, as depicted in phenomenon is relevant because the specific binding of the
Fig. 8(a) and (b). Additionally, we can infer that the longer
wavelength BPP mode (q = 1) exhibited the highest resonance
wavelength shift compared to the shorter wavelength BPP mode
(q = 2) due to the strong hyperbolic dispersion. Moreover, we
calculated the resonance wavelength shift of NE.coli by varying
hp. The resonance wavelength shift of NE.coli increased with the
inclination of hp, as illustrated in Fig. 8(c) and (d). We varied
the hp from 300 to 500 nm for our study. We obtained the
resonance wavelength shifts of B15 and B8 nm per bacteria
for a single E. coli bacteria, while we used the hp of 500 nm. The
resonance wavelength shift varied linearly with changing FF
and hp, which corroborates with eqn (17) and (18). Conse-
quently, we achieved an S of 3000 and 2200 nm per bacteria
for the hp of 500 and 400 nm, respectively, in the longer
wavelength BPP mode (q = 1) and 1600 and 1082 nm per Fig. 9 (a) Schematic illustration of the simulated structures with different
surface coverages of E. coli RI for hp = 400 nm. The bacterium in the
bacteria for the hp of 500 and 400 nm in the shorter wavelength illustration denote the collective RI of multiple microbes, which created a
BPP mode (q = 2), respectively. The hp of 500 nm exhibited 0.005 RIU change. (b) Simulated resonance shift of dips in reflectance
higher sensitivity compared to the hp of 400 nm; however, we spectra for different surface coverage.

10280 | Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 This journal is © the Owner Societies 2024
 View Article Online

PCCP Paper

resonance shift of the proposed nanowire HMM sensor struc-

ture. The illustration of the E. coli bacteria position on the
HMM’s nanowire is provided in Fig. 10(a). We obtained differ-
ent resonance shifts for both BPP modes while we varied the D/
hp, as shown in Fig. 10(b). Meanwhile, the binding of E. coli
bacteria with nanowire altered with varying D/hp. We studied
the electric field distribution around the nanowire, which is
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

provided in Section S8 of ESI.† The resonance wavelength shift

exhibited fluctuations with varying D/hp due to the fluctuations
of the electric field around the nanowire. Interestingly, our
proposed sensor structure exhibited resonance wavelength
shifts even for the worst E. coli position. Here, we utilized
eleven E. coli bacteria. Additionally, we provided the effect of a
Fig. 10 (a) Schematic illustration of the simulated unit structures for single E. coli’s position in Fig. S9 of ESI.†
different D/hp where D is the height of the E. coli bacteria on the nanowire.
(b) Numerically calculated resonance shift of dips in reflectance spectra for
varying D/hp.
4 Comparative analysis
We compared the performance parameters of our proposed
analytes can act differently on the entire nanowire’s surface.
sensor with those of previously reported sensors, in Table 1.
However, most of the previously reported sensor structures
Chowdhury and Zubair introduced triangular Au nanoparticles
cannot detect the point detection phenomena. Contrarily, we
in a dielectric slab; however, the sensitivity was poor, and the
report this capability in our sensor. Fig. 9(a) shows an illustra-
LOD was large, which is not good for detecting miniature
tion of E. coli’s surface coverage on TiN nanowire. We studied
microbes.6 Though 1D photonic crystal and split-ring resonator
the impact of E. coli bacteria’s out-of-plane surface coverage, Sc,
structures improved the sensitivity compared to nanoparticle-
on the resonance wavelength shift, as shown in Fig. 9(b).
comprised structures,35,36 the sensitivity and LOD of these
Interestingly, the resonance shift increased with the increase
sensors were not adequate for small microbes detection.
in surface coverage area. Meanwhile, E. coli bacteria bind to a
HMM sensors can resolve these issues. Silver (Ag)-based multi-
large portion of the nanowire. Thus, the sensitivity of our
layered HMM structures can enhance the performance of
proposed HMM sensor structure is elevated with the increase
LOD;37 however, sensitivity was reduced. Multilayered HMM
of Sc. We used the RI of a single E. coli bacteria for our study.
structures can provide excellent sensitivity;22 however, nano-
We extracted the empirical equations of resonance wavelength
wire HMM structures are best for single microbe detection.
shift for varying Sc of a single E. coli bacteria, which is given by,
Several nanowire HMM structures were reported for smaller
dlr,1 = 0.02 + 26.04Sc  34.01Sc2 + 17.29Sc3, (19) and lighter microbes detection.23–25 However, all the demon-
strated structures were designed using Au, whereas we
and
designed a nanowire HMM sensor structure utilizing TiN,
dlr,2 = 0.07 + 4.55Sc  0.5Sc2. (20) which has not been studied before. Moreover, our proposed
HMM sensor structure exhibited outstanding sensitivity of 11
and 15 nm per bacteria for the TiN nanowire heights of 500 and
Due to strong hyperbolic dispersion, the q = 1 BPP mode
400 nm, respectively, with the capability of detecting a single
exhibited a significant resonance shift for a small change in Sc.
microbe. Furthermore, we obtained an outstanding LOD of
The cubic polynomial term is due to the different binding
0.00008 RIU for our proposed HMM sensor structure.
between E. coli bacteria and HMM structure. No alteration
was found for the q = 2 BPP mode due to weak hyperbolic
dispersion. Thus, a polynomial term was reduced for the q = 2 5 Conclusion
BPP mode, as shown in eqn (20).
3.8.2 Dependency of the position of E. coli on resonance We proposed an ultrasensitive TiN nanowire-based HMM
shift. We studied the impact of E. coli’s position, D/hp on the sensor structure employing FDTD and EMT methods. We

Table 1 Comparative performance analysis of sensors

Structure Wavelength (mm) Sensitivity LOD Ref.

Triangular gold nanoparticles 0.5–0.7 o2.92 0.3 mM 6
Au/1D photonic crystal 0.6–0.8 53.5 nm RIU1 0.009 g dl1 35
Double spilt-ring resonator 150–300 387 GHz RIU1 — 36
Ag/TiNxOy multilayered HMM 0.85 2475.20 nm RIU1 0.00635 RIU 37
Ag/Si multilayered HMM 0.8–2 3450 nm RIU1 — 38
TiN nanowire HMM sensor 1–3 11 nm per bacterium 0.00008 RIU This work

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10281
 View Article Online

Paper PCCP

calculated the hyperbolic region using the Maxwell-Garnett 2 S. C. Jena, S. Shrivastava, S. Saxena, N. Kumar, S. K. Maiti,
equations, where the components of incident light altered. To B. P. Mishra and R. K. Singh, Sci. Rep., 2019, 9, 13485.
evaluate the sensing performance, we varied the refractive 3 M. Yang, X. Yi, J. Wang and F. Zhou, Analyst, 2014, 139,
index by injecting microbes in the aqueous solution, which 1814–1825.
resulted in resonance wavelength shifts at resonant wave- 4 Q. Wu, W. Wu, F. Chen and P. Ren, Analyst, 2022, 147,
lengths. We calculated the resonance wavelength shift and 2809–2818.
enumerated the sensitivity of our proposed HMM sensor struc- 5 L. Fewtrell, R. B. Kaufmann, D. Kay, W. Enanoria, L. Haller
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

ture. The BPP mode at the longer wavelength exhibited the and J. M. Colford, Lancet Infect. Dis., 2005, 5, 42–52.
largest resonance wavelength shift due to the strong hyperbolic 6 E. Chowdhury and A. Zubair, Opt. Mater. Express, 2022, 12,
dispersion. Our proposed HMM sensor structure operated in a 2212–2225.
wide wavelength range from 1 to 3 mm with sensitivities of 7 M. B. Holowko, H. Wang, P. Jayaraman and C. L. Poh, ACS
3000 nm RIU1 and 2200 nm RIU1 for single bacterium with Synth. Biol., 2016, 5, 1275–1283.
TiN nanowire heights of 500 and 400 nm, respectively. Our 8 D. Lubkowicz, C. L. Ho, I. Y. Hwang, W. S. Yew, Y. S. Lee and
study revealed that the highly sensitive BPP modes originated M. W. Chang, ACS Synth. Biol., 2018, 7, 1229–1237.
from TiN-based HMM. We obtained a LOD of 0.00008 RIU for 9 D. Zhao, P. Xiao, X. Dong, Y. Ge, X. Guo, J. Ji, Y. Cheng and
our proposed sensor structure, which enabled the capability of S. Sang, J. Mater. Chem. B, 2023, 11, 9658–9665.
low-concentration microbe detection with high sensitivity. 10 M. SoltanRezaee and M. Bodaghi, Sci. Rep., 2020, 10, 3385.
Moreover, the calculated mode frequency shift suggested that 11 J. L. Arlett, E. B. Myers and M. L. Roukes, Nat. Nanotechnol.,
the increase in the nanowire’s height and radius enhanced the 2011, 6, 203–215.
sensitivity due to strong hyperbolic dispersion. Furthermore, 12 Y.-L. Fu, C.-S. Deng and S.-S. Ma, Appl. Opt., 2020, 59,
the point-detection capability differentiated our work from 896–903.
previously reported sensors. Our study paves the way for 13 I. H. Giden, Opt. Express, 2022, 30, 21679–21686.
designing and fabricating alternative material-based sensors 14 S. Sharma, A. M. Shrivastav and B. D. Gupta, IEEE Sens. J.,
for label-free detection of microbes. 2020, 20, 4251–4259.
15 J.-Y. Sui, S. Yuan Liao, B. Li and H.-F. Zhang, Opt. Lett., 2022,
47, 6065–6068.
Author contributions 16 S. Liao, J. Sui and H. Zhang, Opt. Express, 2022, 30,
Dip Sarker: conceptualization, methodology, visualization, soft- 34172–34187.
ware, investigation, writing – original draft, writing – review & 17 L. Ferrari, J. S. T. Smalley, Y. Fainman and Z. Liu, Nanoscale,
editing. Ahmed Zubair: conceptualization, methodology, 2017, 9, 9034–9048.
visualization, resources, writing – original draft, writing – 18 H. Cho, Y. Yang, D. Lee, S. So and J. Rho, Nanophotonics,
review & editing, supervision. 2021, 10, 3871–3877.
19 Z. Liu, H. Lee, Y. Xiong, C. Sun and X. Zhang, Science, 2007,
315, 1686.
Data availability statement 20 D. Lu, J. J. Kan, E. E. Fullerton and Z. Liu, Nat. Nanotechnol.,
2014, 9, 48–53.
The data supporting the findings detailed in this article are
21 D. Sarker and A. Zubair, TENCON 2023 – 2023 IEEE Region
currently not accessible to the public. However, they are
10 Conference (TENCON), 2023, pp. 1–6.
obtained upon reasonable request from the authors.
22 K. V. Sreekanth, Y. Alapan, M. ElKabbash, E. Ilker,
M. Hinczewski, U. A. Gurkan, A. De Luca and G. Strangi,
Conflicts of interest Nat. Mater., 2016, 15, 621–627.
23 R. Yan, T. Wang, X. Yue, H. Wang, Y.-H. Zhang, P. Xu, L.
The authors declare no conflict of interest. Wang, Y. Wang and J. Zhang, Photon. Res., 2022, 10, 84–95.
24 N. Vasilantonakis, G. A. Wurtz, V. A. Podolskiy and
Acknowledgements A. V. Zayats, Opt. Express, 2015, 23, 14329–14343.
25 R. Yan, T. Wang, H. Wang, X. Yue, L. Wang, Y. Wang and
D. Sarker and A. Zubair thank the Department of Electrical and J. Zhang, J. Mater. Chem. C, 2022, 10, 5200–5209.
Electronic Engineering, Bangladesh University of Engineering 26 G. V. Naik, B. Saha, J. Liu, S. M. Saber, E. A. Stach,
and Technology (BUET) for providing necessary support. D. J. M. K. Irudayaraj, T. D. Sands, V. M. Shalaev and
Sarker acknowledges the financial support from BUET through A. Boltasseva, Proc. Natl. Acad. Sci. U. S. A., 2014, 111,
its Postgraduate Fellowship program. 7546–7551.
27 J. A. Briggs, G. V. Naik, T. A. Petach, B. K. Baum,
D. Goldhaber-Gordon and J. A. Dionne, Appl. Phys. Lett.,
Notes and references
2016, 108, 051110.
1 R. B. M. Schasfoort, Handbook of Surface Plasmon Resonance, 28 M. S. Islam and A. Zubair, Opt. Continuum, 2023, 2,
The Royal Society of Chemistry, 2017. 1688–1700.

10282 | Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 This journal is © the Owner Societies 2024
 View Article Online

PCCP Paper

29 M. K. Sahoo, A. A. VS and A. Kumar, Nanotechnology, 2023, 34 K. V. Sreekanth, M. ElKabbash, Y. Alapan, A. R. Rashed,

34, 195301. U. A. Gurkan and G. Strangi, Sci. Rep., 2016, 6, 26272.
30 P. Liu, L. Chin, W. Ser, T. Ayi, P. Yap, T. Bourouina and 35 M. Shaban, A. M. Ahmed, E. Abdel-Rahman and H. Hamdy,
Y. Leprince-Wang, Procedia Eng., 2014, 87, 356–359. Sci. Rep., 2017, 7, 41983.
31 Ansys Lumerical FDTD| Simulation of Photonic Compo- 36 N. Cui, M. Guan, M. Xu, W. Fang, Y. Zhang, C. Zhao and
nents, 2024, https://www.ansys.com/products/optics/fdtd. Y. Zeng, Opt. Express, 2020, 28, 16834–16844.
32 I. I. Smolyaninov, Hyperbolic Metamaterials, Morgan & Clay- 37 H.-Z. Gao, W.-R. Xu, M.-C. Li, N. Ilyas, J.-M. Wang, W. Li and
Published on 07 March 2024. Downloaded by University of California - San Diego on 6/22/2024 6:09:59 PM.

pool Publishers, 2018. X.-D. Jiang, Opt. Laser Technol., 2022, 151, 108034.
33 J. Elser, R. Wangberg, V. A. Podolskiy and E. E. Narimanov, 38 M. A. Baqir, A. Farmani, T. Fatima, M. R. Raza, S. F. Shaukat
Appl. Phys. Lett., 2006, 89, 261102. and A. Mir, Appl. Opt., 2018, 57, 9447–9454.

This journal is © the Owner Societies 2024 Phys. Chem. Chem. Phys., 2024, 26, 10273–10283 | 10283