IEEE SENSORS JOURNAL, VOL. 23, NO.

6, 15 MARCH 2023 5703

Dual Absorption Broadband Photoacoustic

Technique to Eliminate Interference
in Gas Mixtures
K. Saran Kumar , Y. Esther Blesso Vidhya, Ramya Selvaraj, Satyanarayanan Seshadri ,
S. M. Shiva Nagendra, and Nilesh J. Vasa

Abstract—This article presents a dual-wavelength band

broadband photoacoustic spectroscopy (BPAS) technique
ideally suited to low-energy broadband sources for industrial
and environmental multigas sensing applications. The broad
spectral range of the supercontinuum source is used to cover
a wide range of absorption bands of various gases. The
major limitation in most of the gas sensors with broadband
sources is the effect of interference between gases. This
article proposes a dual-wavelength technique to eliminate
the impact of cross sensitivity among gases. This sensor
exhibits a comprehensive concentration measurement from
parts per million (ppm) to % levels. The proposed technique
showed a ±1.2% error between the calculated and calibrated
concentration values. We achieved a minimum detection
limit (MDL) of 0.4 and 0.7 ppm for methane (CH4 ) with an integration time of 142 and 200 s using 2315- and 1650-nm
band, respectively, 0.3 and 1.8 ppm of ammonia (NH3 ) with an integration time of 100 and 48 s using 2315- and 2017-nm
band, respectively, and also 2.6 ppm of carbon monoxide (CO) with an integration time of 71 s using the 2315-nm band.
Index Terms— Broadband photoacoustic spectroscopy (BPAS), cross sensitivity, dual absorption, dual-wavelength
technique, multigas sensing.

I. I NTRODUCTION [3], [4]. Methane (CH4 ), which forms one of the primary
components of the hydrocarbon family present in natural
HE need for trace gas analysis is widespread and has
T found many applications that include industrial exhaust
sensing, breath gas analysis, outdoor and indoor air qual-
gas, is a significant contributor to climate change and global
warming. It is a harmful greenhouse gas whose impact is
28 times more than that of carbon dioxide (CO2 ) [5], and
ity monitoring, defense, and atmospheric sciences [1], [2],
in medicine, they are used as biomarkers for colon cancer and
bacterial growth; CH4 is most commonly produced in sewage
Manuscript received 18 December 2022; revised 18 January 2023; processing and biogas plants. Ammonia (NH3 ) is a highly
accepted 18 January 2023. Date of publication 2 February 2023;
date of current version 14 March 2023. This work was supported
toxic gas that is mostly common in fertilizer manufacturing
by the Research Grant from the Science and Engineering Research industries and biological processes. The NH3 concentrations
Board–Department of Science and Technology (SERB-DST), India, in the atmospheric boundary layer range from <1 parts per
under Project CRG/2020/003918. The work of K. Saran Kumar was
supported by the Prime Minister’s Research Fellowship (PMRF), India.
billion (ppb) by volume in the troposphere to parts per million
The associate editor coordinating the review of this article and approving (ppm) levels in animal waste, breading sites. A trace level
it for publication was Prof. Carlos Marques. (Corresponding author: of NH3 is present in exhale breath, which indicated liver
K. Saran Kumar.)
K. Saran Kumar, Y. Esther Blesso Vidhya, and Nilesh J. Vasa are with
and renal disorder in human beings [6]. Carbon monoxide
the Department of Engineering Design, Indian Institute of Technology (CO) is a fatal gas even if exposed to a small quality, and
Madras, Chennai 600036, India (e-mail: [email protected]; they are produced due to incomplete combustion of fuel in
[email protected]).
Ramya Selvaraj is with the Department of Electrical Engineering,
automobiles and mining industries. In mines, CO is present
National Institute of Technology at Raipur, Raipur, Chhattisgarh 492010, with H2 S, CH4 , and CO2 in a confined space,which builds up
India (e-mail: [email protected]). and the mine structure and thus makes it hard to disperse to
Satyanarayanan Seshadri is with the Department of Applied Mechan-
ics, Indian Institute of Technology Madras, Chennai 600036, India
the surrounding. This accumulated CO is a serious threat to
(e-mail: [email protected]). the workers in the mines [7]. Quantitative detection of gases
S. M. Shiva Nagendra is with the Department of Civil Engineering, is conventionally determined using laboratory-based analyti-
Indian Institute of Technology Madras, Chennai 600036, India (e-mail:
[email protected]). cal instruments such as gas chromatographs, semiconductor
Digital Object Identifier 10.1109/JSEN.2023.3239885 gas sensors, and electrochemical devices [8]. Semiconducting

1558-1748 © 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 5704 IEEE SENSORS JOURNAL, VOL. 23, NO. 6, 15 MARCH 2023

metal oxide (SMO) sensors are dependent on the electrical microresonator (AmR) to detect trace gases. A QEPAS based
conductivity, and the gas species is detected based on the on the four-off-beam AmR is used to measure water vapor
oxidation and reduction taking place between the target gas in atmospheric air at room temperature (27 ◦ C) achieving a
and the sensor. The target gas is absorbed by the sensing minimum detection limit (MDL) of 278 ppb [21]. On the other
layer, which causes a change in the electrical resistance from hand, laser systems are made available by many commercial
which the concentrations can be obtained. However, the major industries such as Thorlabs, Leukos, and Photonics, to name
drawback of using SMO sensors is their poor selectivity, a few of them. Conventionally, PAS is studied with a narrow
operating temperature, susceptibility to degradation with time, linewidth source that can target a specific absorption line of a
and slow recovery rate. On the other hand, electrochemical given gas analyte. This technique can also achieve very high
sensors are mostly inexpensive, selective, sensitive, and best sensitivity up to parts per trillion (ppt), which serves best
suited for environmental monitoring in the order of ppm to for detecting trace gases, especially in breath gas analysis
ppb. The sensor consists of a working electrode where the for early and easy disease diagnosis applications. Optical
target gas undergoes oxidation, and a counter electrode is used sources with a wide tuning range can be employed to sense
to balance the current in the working electrode. A reference multiple gases with higher concentrations, achieving resolution
electrode is present with which the working electrode potential in ppm to ppb.
is measured. The current measured between the working and PAS is a well-established technique for gas sensing [22],
the counter electrode is proportional to the concentration of the [23], [24]. A distributed feedback laser (DFB) coupled with a
gas detected. The major drawbacks of electrochemical sensors Raman amplifier around 1651-nm wavelength detected 17 ppb
are the limited lifetime and also suffer from cross responses of CH4 [25]. Also, a tunable solid-state Cr2+ :ZnSe laser with
due to changing humidity levels [9]. Carbon nanomaterial 2200–2800 nm was used in the PAS system to achieve the
(CNM)-based surface plasmon resonance (SPR) is used to least detection limit of 0.8, 0.2, and 2.7 ppm for CH4 , CO2 ,
sense small amounts of low molecular weight compounds, and CO, respectively [26]. QCL and inter cascade laser (ICL)
including biomarkers, hormones, pesticides, and explosives. used around 3300-nm region are capable of detecting 1.5 ppb
The CNM has been used as an additional layer over the of C2 H6 [27], and the optical parametric oscillator (OPO)
plasmonic layer to enhance sensitivity and detect gases in around 3100–3500 nm was able to detect 32 ppb of CH4 [28].
lower concentrations [10]. Immunosensors based on plas- Most vibration bands are present in the infrared area, which
monic tilted fiber Bragg grating (TFBG) are developed and features more absorption lines. Hence, tuning the laser to a
reported for the detection of cortisol, which is a biomarker specific absorption line for trace gas detection is essential,
for metabolic syndrome, cardiovascular diseases, and anxiety. which is challenging due to its high operating temperature
The gold-coated TFBG is characterized by the change in dependence [29]. The other drawback of using a tunable laser
surrounding refractive index (SRI), and the sensor achieved source for trace gas detection is that only a few gases can
a detection limit of 0.275 ± 0.028 nm/mL with a total be measured using the same source. This drawback can be
wavelength shift of 3 nm [11]. Optical sensing methods are reduced by employing the simultaneous operation of several
mostly preferred due to their high sensitivity, selectivity, fast such sources. However, it increases the overall complexity and
response, and real-time monitoring capacity [12]. The recent cost of the system.
advancements in optics availability in the mid-IR region have On the other hand, the present work uses a broadband PAS
paved the way for ultrahigh-sensitive gas sensing systems [13], (BPAS) technique with a supercontinuum laser (SCL) source
[14]. Optical interferometry combined with molecular imprint to detect multiple gases in the NIR region [30], [31], [32].
polymer (MIP) is used as an ultrasensitive and selective remote Multiple gases with broadband absorption characteristics with
detection technique. A selective, online detection method by a wide dynamic range from sub-ppm to % level concentrations
combining a concatenated microfiber interferometer is used can be measured with the BPAS technique. BPAS considers
to precisely determine water pollutants [15]. The biosensors the summation of multiple absorption lines over a wavelength
based on SPR and a tapered-in-tapered fiber (TiTF) are band corresponding to the target analyte to produce the pho-
reported and used to measure the concentration of p-cresol. toacoustic (PA) signal. This technique can achieve a better
The refractive index change is measured due to the reaction dynamic range and sensitivity than conventional spectroscopic
of p-cresol with tyrosinase enzyme on the sensing region of sensors in the near IR range. However, interferences from other
the probe and resulting in the spectral variation [16]. The gas species in the same detection band must be studied in
photoacoustic spectroscopy (PAS) technique facilitated with detail, especially in cases where the absorption cross sections
a laser source in the near-infrared (NIR) region has opened of two or more targeted gas molecules in an analyte have over-
up a new regime of gas sensing. The mid-IR sources come in lapping absorption bands and equivalent absorption cross sec-
multiple different varieties, for quantitative measurements in tions. For example, in petrochemical industries, there are sev-
spectroscopic applications. The lasers offer sufficient optical eral combinations of hydrocarbons present in the gas species,
power and have a higher SNR, and the quantum cascade considering the most common and fundamental hydrocarbons
laser (QCL) provides narrow linewidth that helps in achiev- such as methane (CH4 ), ethane (C2 H6 ), propane (C3 H6 ), and
ing high selectivity and sensitivity, with rapid wavelength butane (C4 H8 ). The absorption spectra of these hydrocarbons
tuning for faster response and data acquisition [17], [18], show specific absorption wavelengths, which can be used to
[19], [20]. Quartz-enhanced PAS (QEPAS) uses an acoustic determine their concentrations without overlaps. Methane at

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: DUAL ABSORPTION BROADBAND PA TECHNIQUE TO ELIMINATE INTERFERENCE IN GAS MIXTURES 5705

1.6 µm, ethane at 4.2 µm, propane at 13 µm, and butane at TABLE I
3.2 µm can be measured without any effect of cross sensitivity. R ESONANCE F REQUENCY OF THE PA C ELL
However, if any other forms of hydrocarbons are identified
which may have any overlapping absorption bands, can be
measured by eliminating the interference using the proposed
technique. Also, in the NH3 synthesis process, NH3 , CO, CO2 ,
H2 S, CH4, and H2 O vapor have to be monitored systematically
to realize the efficiency of the process. With the developed
supercontinuum-based BPAS technique, the wavelength cov-
ered by the source is 800–2800 nm. NH3 has substantial
absorption bands at 2 and 2.3 µm, CH4 at 1.6 and 2.3 µm
and CO at 2.3 µm. More often, the presence of CO is masked
is used to calculate the resonance frequency ( f res ) of the PA
due to the higher concentration of NH3 in many synthetic
cell [33]
processes. Detection of CO even at lower concentrations is
very much essential as it is highly hazardous. A complimentary n ∗ Cs
f res = (1)
absorption band of CH4 and NH3 at 1.6- and 2-µm region 2 (L + δL + δ H )
was selected to retrieve the concentration of unknown CO. where n is the mode number representing a longitudinal mode
The target analyte concentrations were calculated using a dual- of vibration, the fundamental longitudinal mode is considered
wavelength technique. PAS has generally focused on tunable (n = 1), and the corresponding resonant frequency is calcu-
or single-line lasers to avoid cross sensitivity from other gases lated. Cs is the speed of sound and L is the length of the
present in the sample. resonant tube. Due to the boundary effects, two correction
The major motivation to carry out this work is to develop factors are included, namely, δL and δ H . The value of δ H
a broadband PA sensor that can be used for measurements is estimated to be ≈3r . The δL factor is accounted for the
of trace gases in process industries, sewage water treatment elongation of the resonator, estimated as ≈0.6r , where r is the
plants, mining plants, biosensing, and so on. The major radius of the resonant tube. Equation (2) is used to calculate
limitation of BPAS-based sensing for such applications is the speed of sound at standard temperature and pressure (STP)
the interference of gases. In this article, a dual-wavelength
γ RT
r
absorption technique is proposed and demonstrated with gas
mixtures of NH3 , CH4 , and CO to overcome the issue Cs = (2)
M
concerning cross-sensitivity. In Section II, an approach to
where γ is the adiabatic index of the gas, R is the gas constant
BPAS and experimental setup are described. In Section III,
(8.314 J/K · mol), T is the standard temperature (293.15 K),
results concerning direct absorption spectroscopy and BPAS
and M is the molar mass of the gas molecule in kg/mol.
are discussed. Finally, the results are concluded in Section IV.
For various combinations of gases, the cumulative adiabatic
The BPAS-based approach exhibits detection limits in sub-
index and the cumulative molar mass are calculated to obtain
ppm levels that are best suited for measuring gases present in
the speed of sound Cs in m/s. The corresponding resonant
trace concentrations.
frequency is calculated from (1) and is shown in Table I.
II. T HEORY AND E XPERIMENTAL P ROCEDURE A 0.05% error was observed between the calculated and the
experimental resonant frequency values. When the chopper
A. Broadband PAS
frequency was varied from 980 to 1010 Hz for different
In PAS, as the name suggests, the light energy is converted
concentrations of CH4 , NH3 , and CO in the range from a
to sound energy in a series of steps. Due to the modulation of
few thousand ppm to 10 ppm in N2 , the variation in the PA
the laser frequency, thermal waves are generated in the sample
signal amplitude was lesser than 0.01%. Hence, the chopper
when gas molecules are excited by the radiation, leading to
frequency was set near the resonance value of 100% N2
the formation of acoustic waves. The produced acoustic signal
at 982 Hz.
is detected by a suitable PA detector, usually a microphone,
and further processed for measurement. The laser light is
modulated at the resonance frequency of the PA cell. The PA B. Experimental Setup and Procedure
cell made from SS 312 material consists of three cylinders. We have developed a laboratory-based in-house BPAS
The central cylindrical tube or the resonator tube, having a setup, as shown in Fig. 1. A SCL source (Leukos, SM-250-IR)
length (L) of 160 mm and a diameter (dh ) of 10 mm, houses was used in this setup. A light chopper (NF instruments, 5584
the electric microphone at its center. The resonator tube opens A) modulates the light source with the resonant frequency of
into two larger cylinders at their respective ends called the the PA cell (982 Hz).
buffer cylinders: a length of 80 mm and a diameter of 60 mm. Two focusing lenses (L1 and L2 ) are used: L1 with a
The gas inlet and outlet ducts are housed in the buffer cells to focal length of 150 mm is placed before the PA cell, and
reduce the noise caused due to gas flow and any heat radiation L2 with a focal length of 100 mm is placed after the cell.
from the windows. Two quartz windows of diameter 20 mm Filters with a center wavelength of 1650 nm for detecting
are housed in slots and secured with “O” rings in the buffer CH4 , 2107 nm for detecting NH3 , and 2315 nm for detecting
cylinders for optical access through the PA cell. Equation (1) CH4 , NH3 , and CO were used. All the filters used have a

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 5706 IEEE SENSORS JOURNAL, VOL. 23, NO. 6, 15 MARCH 2023

Fig. 1. Experimental setup of broadband PAS.

Fig. 2. SCL source spectra (a) from 1000 to 2500 nm, (b) 1650-nm
bandwidth of 50 nm (FWHM). The transmitted optical power filter, (c) 2017-nm filter, and (d) 2315-nm filter.
was measured as P1650 nm = 23 mW, and P2017 nm = 11 mW
and P2315 nm = 3 mW. The PA signals are measured using uses 1650-, 2017-, and 2315-nm filters with 50-nm bandwidth,
an electric microphone (Ono Sokki, MI 1235) located in the as shown in Fig. 2(b)–(d).
middle of the PA resonator with a preamplifier circuit (Ono
Sokki, MI 3111). The signal from the preamplifier is then fed
C. PA Signal Generation
to a lock-in amplifier (Signal Recovery SR 7225) set at the
laser-modulating frequency. A lock-in amplifier is commonly The generated PA signal amplitude Spa (υ) is directly pro-
used to detect the signal of interest even if they are smaller portional to the incident optical power Po (υ) and the molecular
than the noise signal they accompany. The lock-in amplifier absorption coefficient α(υ). Contrary to conventional PAS in
used is capable of detecting very small ac signals down to broadband PAS, we use multiple wavelengths, and hence, the
nV or even smaller where the noise is several 100 times signal generated is cumulative and is calculated analytically
greater. The instrument uses phase-sensitive detection to single using (3), as shown in the following:
out the components of the signal at a specific frequency and
X X X
Spa (ν) = Ccell ∗ Po (ν) ∗ α (ν) (3)
phase. The noise signal at other frequencies or random phases
is rejected through the DSP lock-in filtering software. The where Ccell is the constant of proportionality [34]. In the
light source output power is measured using a power meter resonant operation, an acoustic standing wave oscillates within
(OPHIR NOVA 82495) positioned postcell to monitor beam the cavity, and the cell constant is calculated in (Pa · cm/W)
intensity changes with varying concentrations. The setup is and is given by the following equation:
aligned using an IR detector card to ensure that the laser beam
Q j (γ − 1) I j L c
passes through all the optics and then through the exit window Ccell = P j (rm , ωc ) (4)
of the PA cell. The SCL source spectrum is shown in Fig. 2(a). ω j Vc
The SCL source consisted of an optical fiber-based output where Q j is the quality factor, ω j is the resonance frequency,
with a broadband spectrum ranging from 0.8 to 2.5 µm. The γ is the adiabatic index, I j is the overlap integral between the
pulsewidth is approximately 1 ns with a pulse repetition rate of laser beam distribution and the acoustic mode of the cavity,
250 kHz. However, the pulsed SCL does not contribute to the L c is the length of the resonator, Vc is the volume of the
PA signal, as the pulse repetition rate is quite fast. The system cell, and P j (rm , ω j ) is the normalized acoustic mode at the
does not react to any changes caused due to the molecular position rm of the microphone. The Ccell value is generally
relaxation time as the intensity modulation frequency range ν M constant for a given gas, which contributes to the generation of
is limited between the molecular diffusion time (1/τd ) and the PA signal and is obtained experimentally using the following
molecular relaxation time (1/τr ) of the target gas molecules. equation:
Moreover, the calibrated gas samples (CH4 , NH3 , and CO
PA Signal
studied in the presented work) are buffered in N2 , which has Ccell = P . (5)
a longer relaxation rate when compared to pure CH4 , NH3 , Po ∗ σ ∗ N
or CO. Therefore, the laser intensity modulation frequency The PA signal is obtained in volts, with the microphone
of 982 Hz alone contributes to the PA effect. The super- responsivity (Smic ) being 36 mV/Pa. 6 Po is the incident
continuum light source has almost flat spectra from 0.8 to optical power in watts, σ is the absorption cross section
2.5 µm with a sharp peak around 1.55 µm, corresponding of the target gas molecule in cm2 /molecule, and N is the
to the fundamental laser wavelength. It is necessary to use number density in molecules/cm3 . On substituting the values
suitable filters to remove the other wavelengths. This study corresponding to a calibrated mixture of NH3 , experimentally,

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: DUAL ABSORPTION BROADBAND PA TECHNIQUE TO ELIMINATE INTERFERENCE IN GAS MIXTURES 5707

Fig. 3. Variation in cell constant with calibrated NH3 , CH4 , and CO with
the mixture all the three gases.

Ccell was estimated to be 34.5 Pa · cm/W. However, when a

combination of gases contributes to the combined PA signal
generation, the adiabatic index of the gas varies as the compo-
sition between the target and the interfering analyte varies. The
Ccell values with calibrated CH4 , NH3 , and CO along with a
mixture of CH4 , NH3 , and CO in the 2315-nm band are shown
in Fig. 3. The standard deviation of Ccell values for CH4 ,
NH3 , and CO was within ±0.2 Pa · cm/W. In comparison, the
standard deviation increases to ±0.9 Pa · cm/W in the 2315-nm
band for a mixture of gases. However, this deviation would
also include the experimental repeatability of the system.
The absorption cross section from the HITRAN 2020
(high-resolution transmission molecular absorption) database
of CH4 , NH3 , and CO in the 2.3-µm range is in the order
of ≈10−21 cm2 /molecule, as shown in Fig. 4(a)(i)–(iii). Fig. 4. Absorption cross section of CH4 , NH3 , and CO in (a) 2.3-,
(b) 1.6-, and (c) 2-µm wavelength range.
While in the 1.6-µm range, only CH4 is present (as there is
no absorption of NH3 and CO) on the other hand in the 2-µm
range, and only NH3 is present (there is no absorption of CH4
and CO) and is also in the order of ≈10−21 cm2 /molecule,
as shown in Fig. 4(b)(i)–(iii) and (c)(i)–(iii), respectively.

III. R ESULTS AND D ISCUSSION

Direct absorption spectroscopy was conducted using a 1-m
path length gas cell and an FTIR spectrum analyzer OSA203C,
Thorlabs, to study the transmission spectra of CH4 , NH3 , and
CO in the chosen wavelength range.
Fig. 5(a)–(c) shows the absorption spectra of 10% of CH4 ,
5% of NH3 , and 0.2% of CO balanced with N2 , respectively.
The υ3 + υ4 absorption peaks of CH4 , υ2 + υ3 absorption
peaks of NH3 , and υ1 + υ2 absorption peaks of CO are
shown in the 2315-nm band. Fig. 5(d) and (e) shows the
absorption spectra of 10% of CH4 showing 2υ3 peaks in Fig. 5. Absorption peaks of (a) CH4 , (b) NH3 , and (c) CO using the
1650 nm and 5% of NH3 showing υ1 + υ2 peaks in 2017 nm 2315-nm wavelength filter. (d) CH4 using 1650-nm wavelength filter.
bands, respectively. When a mixture of 2% of CH4 , 2% of (e) NH3 using 2017-nm wavelength filter. (f) Mixture of CH4 , NH3 , and
CO in 2315-nm wavelength filter.
NH3 , and 0.2% of CO buffered with N2 was studied using
a 2315-nm bandpass filter, clearly indicating a substantial
interference between all the three gases, as shown in Fig. 5(f).
Homonulcear diatomic molecules, such as H2 , N2 , and O2 , chosen as a buffer gas in order to obtain various concentrations
have no dipole moment and thus are IR inactive. Hence, N2 is of CH4 , NH3 , and CO for calibration of the system. Also, the

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 5708 IEEE SENSORS JOURNAL, VOL. 23, NO. 6, 15 MARCH 2023

absorption studies were carried out in open air and the system
response is identical to the one carried out with N2 buffer gas.
As stated earlier, the limitation of BPAS is its inability to
resolve the interference from other gas molecules present in
the sample. Generally, when multiple gases are present in
a mixture with overlapping absorption bands, the PA signal
amplitude would be the cumulative value of all the absorbing
gases. Thus, (3) can be modified and rewritten as follows:
X X
Spa (ν) = Ccell ∗ Po (ν)
∗(σ1 C1 + σ2 C2 + σ3 C3 + · · · ) (6)
where (σ1 C1 + σ2 C2 + σ3 C3 + · · · ) = 6α(υ), with σ1 ,
σ2 , σ3 ,... , and C1 , C2 , C3 ,... , are the absorption cross sections
and concentrations of different gases in the measurement
range. From Fig. 5(f), it is seen that in the 2315-nm wave-
length band, there is a strong interference between CH4 , NH3 ,
Fig. 6. Variation of PA signal amplitude as a function of concentration
and CO, and the absorption spectra are almost overlapping. (a-i and ii) NH3 , (b-i and ii) CH4 , and (c-i and ii) CO in the 2315-nm
The PA signal generated at this wavelength range would be due band.
to the mixture of gases, which makes it difficult to determine
the concentration of any of the gas. This challenge can be
out directly or stored in a storage tank for further use. The
overcome using the SCL with an appropriate filter to match
flow rate was maintained at 100 SCCM, taking approximately
with an absorption band that has no interference with other
1.5 min to fill the gas cell at 1 atm. The concentration of CH4
gases in the mixture. From Fig. 4(d) and (e), it is seen that the
was varied between 25 ppm to 10%, NH3 from 25 ppm to 5%,
1650-nm wavelength band has no interference of NH3 and CO
and CO from 10 ppm to 0.2%. The standard deviation was
on CH4 and the 2017-nm wavelength band has no interference
calculated for five repeated experimental values to determine
of CH4 and CO on NH3 .
the precision of the system.
The PA gas cell is made from SS 312, and it was found that
A. Calibration NH3 adhesion to the cell walls caused persistent background
Initially, to establish a baseline signal, separate calibrated signals even after vacuuming the cell. Atmospheric air was
gases (10% of CH4 balanced with N2 , 5% of NH3 balanced first used for flushing out NH3 in the cell, after which pure N2
with N2 , and 0.2% of CO balanced with N2 ) were used. gas was passed to check the background levels. High humidity
Calibration studies were performed at different wavelengths levels in the air (relative humidity ∼ 77% in Chennai) could
for CH4 , NH3 , and CO. be displacing NH3 from the cell walls [35].
A gas mixture system with mass flow controllers (MFCs) It was seen that after two rounds of this method, ammonia
capable of mixing various gases was used for calibrating the was expelled significantly from the cell. It was observed
BPAS setup. A custom-made gas mixing system is used to that the PA signal values are always within ±0.03%.
combine multiple gases in the required concentration. Alicat Fig. 6(a-i and ii)–(c-i and ii) shows a linear relationship
scientific MFC is used for mixing the gases; the system offers between the PA signal and the concentration with the coef-
mixing of three gases at a time. The first MFC is designed ficient of determination for the fit to be R 2 ∼ 0.98, 0.99,
exclusively for the use of corrosive gases. The physical time and 0.99 for NH3 , CH4 , and CO, respectively, in the 2315-nm
constant (s) of the MFCs response time is a measure of the band. Fig. 7(a-i and ii) and (b-i and ii) shows a linear rela-
amount of time that the controller takes to reach 63.2% of tionship between the PA signal and the concentration with
the commanded set point. The gas mixing system used in the the coefficient of determination for the fit to be R 2 ∼ 0.99,
experiment uses Alicat MFC, which takes 7.4 ms to reach 0.98 for CH4 and NH3 in 1650 and 2017 nm, respectively.
the 63.2% set point value. The MFC reaches 100% of the The usage of higher concentration in the calibration is
set point in 11.4 ms. The accuracy specifications of the MFC to show that the PA signal is linearly dependent on the
fall within the 0.5%–2% range, a common settling error band concentration irrespective of being very low or high. In cer-
equivalent to Alicat controllers is achieved after 27.6 ms. The tain situations, say breath gas sensing, the concentrations of
gas concentration remained precious as the gas mixture was the gas would be only in trace levels, and on the other
filled in the cell only after a 2-s time interval that was given hand, sensing gas concentrations in industrial exhausts would
to the system for it to reach its steady flow rate. be in higher levels ranging a few tens of percentages. The
The most commonly used gases are precalibrated and stored proposed system can be used to monitor both lower and
in the library that can be directly selected from the gas higher concentration measurements. The proposed system can
selection menu. The required concentration percentage is fed be used to monitor both lower and higher concentration
to the MFC using system-specific software (Flow Vision MX). measurements. The testing of CO at 0.2% was carried out
A digital pressure gauge reads out the pressure in which carefully in a laboratory environment with proper ventilation,
the gas flows. After mixing, the gas mixture is either taken and also, the exposure time for the experiment was monitored

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: DUAL ABSORPTION BROADBAND PA TECHNIQUE TO ELIMINATE INTERFERENCE IN GAS MIXTURES 5709

Fig. 7. Variation of PA signal amplitude as a function of concentration Fig. 9. Variation of PA signal amplitude as a function of a fixed concen-
(a-i and ii) CH4 in the 1650-nm band and (b-i and ii) NH3 in the 2017-nm tration of CH4 and CO (1000 ppm each) and varying concentrations of
band. NH3 (0–1000 ppm).

Fig. 10. Variation of PA signal amplitude as a function of a fixed concen-

Fig. 8. Variation of PA signal amplitude as a function of a fixed concen- tration of CH4 and NH3 (1000 ppm each) and varying concentrations of
tration of NH3 and CO (1000 ppm each) and varying concentrations of CO (0–1000 ppm).
CH4 (0–1000 ppm).

of NH3 and CO is continuously present as the CH4

to keep it well below the prescribed OSHA standards of CO concentration is further varied from 0 to 1000 ppm.
exposure. 3) A fixed concentration of CH4 and NH3 (1000 ppm each)
were mixed with varying concentrations of CO from 0 to
1000 ppm; the corresponding PA signal was acquired,
B. Gas Mixture of CH4 , NH3 , and CO as shown in Fig. 10. It can be seen that a background
A known mixture of CH4 , NH3 , and CO concentrations signal of 1.47 mV corresponding to 1000 ppm of CH4
were prepared and tested in the 2315-nm band. and NH3 were continuously present as the CO concen-
1) A fixed concentration of NH3 and CO (1000 ppm tration is further varied from 0 to 1000 ppm.
each) were mixed with varying concentrations of CH4 From Fig. 4(b) and (c), it is evident that there is no
from 0 to 1000 ppm; the corresponding PA signal was absorption of NH3 and CO in the 1650-nm band and CH4
acquired, as shown in Fig. 8. It can be seen that a back- and CO in the 2017-nm band, respectively. Hence, the PA
ground signal of 614 µV corresponding to 1000 ppm signal generated from the unknown gas mixture of CH4 , NH3 ,
of NH3 and CO is continuously present as the CH4 and CO at 1650 nm is mainly due to the presence of CH4 .
concentration is further varied from 0 to 1000 ppm. The concentration of unknown CH4 can be measured using
2) A fixed concentration of CH4 and CO (1000 ppm the calibration data as shown in Fig. 7(a-i and ii) and can be
each) were mixed with varying concentrations of NH3 theoretically described by the following equation:
from 0 to 1000 ppm; the corresponding PA signal was X
acquired, as shown in Fig. 9. It can be seen that a back- PA signal(CH4 ) = Ccell ∗ Po ∗ (σCH4,1650 nm ∗ CCH4 ).
ground signal of 1.27 mV corresponding to 1000 ppm (7)

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 5710 IEEE SENSORS JOURNAL, VOL. 23, NO. 6, 15 MARCH 2023

Fig. 11. Comparison of calibrated CO and calculated CO

concentrations. Fig. 12. Allen deviation plot for 25 ppm of (a) CH4 , (b) NH3 , and
(c) CO with 2315-nm filter; (d) CH4 with 1650-nm filter; and (e) NH3
with 2017-nm filter.
Similarly, using a 2017-nm filter generates the PA signal TABLE II
that is mainly due to the presence of NH3 , which can be MDL S
used to determine the unknown concentration of NH3 from
the calibration data as shown in Fig. 7(b-i and ii) and can be
theoretically described by the following equation:
X
PA signal(NH3 ) = Ccell ∗ Po ∗ (σNH3,2017nm ∗ CNH3 ).
(8)
The signal corresponds to the identified concentrations of
CH4 from the 1650-nm filter and NH3 from the 2017-nm filter
in the 2315-nm filter band. The PA signal with CH4 at 1650 nm Fig. 12. A 25-ppm CH4 , NH3 , and CO sample was held in the
was used to estimate the concentration of CH4. Subsequently, resonant PA gas cell. The PA signal amplitude was acquired
the PA signal at 2315 nm was estimated. Similarly, the value of over 1000 s at an interval of 0.5 s. A similar procedure was
the PA signal for NH3 was estimated. These values were used performed for CH4 at 1650-nm bands and NH3 in 2017-nm
to estimate the concentration of CO. The PA signal obtained bands. The Allan deviation plots in Fig. 12(a)–(c) show that
at the 2315-nm band is the cumulative sum of all the three the MDL for the CH4 , NH3 , and CO in the 2315-nm band
gases (CH4 , NH3 , and CO) as expressed in the following: is 0.48, 0.36, and 2.59 ppm with integration time (τ ) of 142,
100, and 71 s, respectively, whereas the MDL for CH4 at
PA signal(CH4 +NH3 +CO@2315) the 1650-nm band and NH3 at the 2017-nm band is 0.75 and
= PA (CH4 @2315) 1.8 ppm with an integration time of 200 and 48 s, respectively.
+ PA (NH3 @2315) + PA (CO@2315) . The Integration time allows the determination of how long the
PA sensor signals can be averaged in order to increase the
The unknown concentration of CO from the mixture can be sensitivity of the system before the other noise sources, such
obtained by removing the sum of the signal obtained due to as temperature, laser instability, and mechanical drifts, begin
CH4 and NH3 in the 2315-nm band as shown in the following: to dominate it. The response time and the integration time are
PA (CO@2315) = {PA (CH4 +NH3 +CO@2315) } not the same as the system starts to respond immediately when
the modulated laser intensity falls on the gas molecules. Laser
− {PA (CH4 @2315) + PA (NH3 @2315) }. power instabilities along with the photothermal effect induce
Fig. 11 shows the concentration of the calibrated CO and noise that is the main cause for the increase in the overall
calculated CO from the mixture by substituting the PA signal PAS cell noise levels. To reduce the integration time, the laser
amplitudes, the input power, and absorption cross sections. power and the chopper fluctuations are to be reduced, and
The calibrated and calculated values of CO concentrations fall this can be done by implementing highly stable drivers. The
within an error percentage of ±1%. The error is attributed to corresponding integration time and MDL are obtained from
the experimentally determined Ccell values encompassed in the the Allen plot and are shown in Table II.
system repeatability.
IV. C ONCLUSION
C. Allan Deviation Analysis A dual absorption band multiwavelength technique for
The MDL of the system was determined using the Allan removing interference using a BPAS technique is proposed
deviation plots [36] obtained experimentally, as shown in and demonstrated successfully. This approach is developed to

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: DUAL ABSORPTION BROADBAND PA TECHNIQUE TO ELIMINATE INTERFERENCE IN GAS MIXTURES 5711

overcome the problem of cross sensitivity among multiple [10] V. Kumar, S. K. Raghuwanshi, and S. Kumar, “Recent advances
gases when using the BPAS technique. The PA signal is in carbon nanomaterials based SPR sensor for biomolecules and
gas detection—A review,” IEEE Sensors J., vol. 22, no. 16,
directly dependent on the Cell constant (Ccell ), input optical pp. 15661–15672, Aug. 2022.
power (Po ), and the molecular absorption coefficient (α). Ccell [11] C. Leitão et al., “Cortisol in-fiber ultrasensitive plasmonic immunosens-
depends on the quality factor of and the design of the cell, and ing,” IEEE Sensors J., vol. 21, no. 3, pp. 3028–3034, Feb. 2021.
Ccell can be improved by employing a multipass gas cell and [12] J. Hodgkinson and R. P. Tatam, “Optical gas sensing: A review,” Meas.
Sci. Technol., vol. 24, no. 1, 2013, Art. no. 012004.
using a high-sensitivity microphone. On the other hand, Po can [13] Y. Wang, K. Zheng, F. Song, K. F. Tittel, and C. Zheng, “Mid-infrared
be improved by using a narrowband, high-power laser source; absorption spectroscopy for gas sensing and application,” in Proc. IEEE
in this work a low-power and a cost-effective broadband laser 5th Optoelectron. Global Conf. (OGC), Sep. 2020, pp. 80–83.
[14] A. Lambrecht and K. Schmitt, “Mid-infrared gas-sensing systems and
is used in view of sensing multiple gases using a single applications,” in Woodhead Publishing Series in Electronic and Optical
source. This study demonstrated two wavelength bands, for Materials, Mid-infrared Optoelectronics. Sawston, U.K.: Woodhead,
CH4 at 1650 and 2315 nm, NH3 at 2017 and 2315 nm, and a 2020, pp. 661–715.
single band for CO at 2315 nm. Calibration of all three gases [15] P. Gorai, S. Kumar, C. Marques, P. K. Singh, and R. Jha, “Imprinted
polymer functionalized concatenated optical microfiber: Hypersensitive
was performed and linearity has been established between and selective,” IEEE Sensors J., vol. 23, no. 1, pp. 329–336, Jan. 2023.
the increased PA signal amplitude due to interfering gases in [16] S. Kumar et al., “Plasmon-based tapered-in-tapered fiber structure for
the target gas range. The concentration of unknown CO in the p-cresol detection: From human healthcare to aquaculture application,”
IEEE Sensors J., vol. 22, no. 19, pp. 18493–18500, Oct. 2022.
mixture of CH4 , NH3 , and CO was identified by successfully
[17] G. A. West, J. J. Barrett, and D. R. Siebert, “Photoacoustic spec-
eliminating the interference of CH4 and NH3 in the mixture. troscopy,” Rev. Sci. Instrum., vol. 54, no. 7, pp. 797–817, 1983.
The calibrated and calculated values of CO concentrations [18] A. Fathy, M. Y. Sabry, W. I. Hunter, D. Khalil, and T. Bourouina, “Direct
fall within an error percentage of ± 1%. The Allan deviation absorption and photoacoustic spectroscopy for gas sensing and analysis:
A critical review,” Laser Photon. Rev., vol. 16, no. 8, Aug. 2022,
analysis was performed to identify the MDL for CH4 , NH3 , Art. no. 2100556.
and CO. MDLs of 0.48 and 0.75 ppm for CH4 with an [19] K. S. Gadedjisso-Tossou et al., “Cavity ring-down spectroscopy for
integration time of 142 and 200 s using 1650- and 2315-nm molecular trace gas detection using a pulsed DFB QCL emitting at 6.8
µm,” Photonics, vol. 7, no. 3, p. 74, 2020.
filters, 0.36 and 1.8 ppm for NH3 with an integration time of
[20] K. Zheng, C. Zheng, Y. Zhang, Y. Wang, and F. K. Tittel, “Review
100 and 48 s using the 2017- and 2315-nm filter, and 2.59 ppm of incoherent broadband cavity-enhanced absorption spectroscopy
for CO with an integration time of 71 s using 2315-nm filter (IBBCEAS) for gas sensing,” Sensors, vol. 18, no. 11, p. 3646, 2018.
were achieved. This technique of dual absorption BPAS can [21] Z. Wang et al., “Quartz-enhanced photoacoustic spectroscopy based on
the four-off-beam acoustic micro-resonator,” J. Lightw. Technol., vol. 38,
be used to eliminate the major drawback of interference of no. 18, pp. 5212–5218, Sep. 15, 2020.
gases in various gas mixtures. [22] R. Selvaraj, N. J. Vasa, and S. M. Shiva Nagendra, “Off-resonance
photoacoustic spectroscopy technique for multi-gas sensing in biogas
plants,” Anal. Chem., vol. 91, no. 22, pp. 14239–14246, Nov. 2019.
ACKNOWLEDGMENT [23] R. Bauer et al., “Miniaturized photoacoustic trace gas sensing using
a Raman fiber amplifier,” J. Lightw. Technol., vol. 33, no. 18,
The authors are thankful to Mohd Zaid and B. Ashok Kumar pp. 3773–3780, Sep. 15, 2015.
for their help in the experiments. [24] K. Sulochana, K. Akash, M. Kumaravel, and N. J. Vasa, “Development
of super luminescent diode based sensing technique for multiple-gas
monitoring,” Int. J. Adv. Eng. Sci. Appl. Math., vol. 6, nos. 1–2,
R EFERENCES pp. 117–124, Jun. 2014.
[25] R. Selvaraj, N. J. Vasa, and S. M. Shiva Nagendra, “Off-resonant pho-
[1] A. Schmohl, A. Miklos, and P. Hess, “Detection of ammonia by pho- toacoustic spectroscopy for analysis of multicomponent gas mixtures at
toacoustic spectroscopy with semiconductor lasers,” Appl. Opt., vol. 41, high concentrations using broadband vibrational overtones of individual
no. 2, pp. 1815–1823, Mar. 2002. gas species,” Appl. Opt., vol. 58, no. 15, pp. 4118–4126, 2019.
[2] S. Das, S. Pal, and M. Mitra, “Significance of exhaled breath test in [26] C. Fischer, E. Sorokin, I. T. Sorokina, and M. W. Sigrist, “Photoacoustic
clinical diagnosis: A special focus on the detection of diabetes mellitus,” monitoring of gases using a novel laser source tunable around 2.5 µm,”
J. Med. Biol. Eng., vol. 36, no. 5, pp. 605–624, Oct. 2016. Opt. Lasers Eng., vol. 43, pp. 573–582, Mar./May 2005.
[3] L. S. K. Achary et al., “Reduced graphene oxide-CuFe2 O4 nanocom- [27] M. Kim et al., “Room temperature interband cascade lasers,” Phys. Proc.,
posite: A highly sensitive room temperature NH3 gas sensor,” Sens. vol. 3, pp. 1195–1200, Jan. 2010.
Actuators B, Chem., vol. 272, pp. 100–109, Nov. 2018. [28] P. Werle, “Tunable diode laser absorption spectroscopy: Recent findings
[4] R. Selvaraj, N. J. Vasa, S. M. S. Nagendra, and B. Mizaiko, “Advances in and novel approaches,” Infr. Phys. Technol., vol. 37, no. 1, pp. 59–66,
mid-infrared spectroscopy-based sensing techniques for exhaled breath Feb. 1996.
diagnostics,” Molecules, vol. 25, no. 9, p. 2227, 2020. [29] M. Lassen, L. Lamard, Y. Feng, and J. C. Petersen, “Off-axis quartz-
[5] T. Aldhafeeri, M. K. Tran, R. Vrolyk, M. Pope, and M. Fowler, enhanced photoacoustic spectroscopy using a pulsed nanosecond mid-
“A review of methane gas detection sensors: Recent developments and infrared optical parametric oscillator,” Opt. Lett., vol. 41, no. 17,
future perspectives,” Inventions, vol. 5, vol. 3, p. 28, 2020. pp. 4118–4121, 2016.
[6] Z. Bielecki, T. Stacewicz, J. Smulko, and J. Wojtas, “Ammonia gas [30] R. Selvaraj, K. T. Abdul Nasir, N. J. Vasa, and S. M. Shiva Nagendra,
sensors: Comparison of solid-state and optical methods,” Appl. Sci., “Monitoring of CO2 and CH4 composition in a biogas matrix from
vol. 10, no. 15, p. 5111, 2020. different biomass structures,” Sens. Actuators B, Chem., vol. 249,
[7] J. Hansen et al., “Climate impact of increasing atmospheric carbon pp. 378–385, Oct. 2017.
dioxide,” Science, vol. 213, no. 4511, pp. 957–966, 1981. [31] R. Selvaraj, K. Sulochana, N. J. Vasa, and S. M. Shiva Nagendra,
[8] S. Qiao, Y. Ma, Y. He, X. Yu, Z. Zhang, and F. K. Tittel, “A sensitive “Superluminescent diode coupled with etalon for measuring CH4 and
carbon monoxide sensor based on photoacoustic spectroscopy with a CO2 in Biogas,” in Proc. Frontiers Opt., San Jose, CA, USA, 2017,
2.3 µm mid-infrared high-power laser and enhanced gas absorption,” pp. 1–2.
Sensors, vol. 19, no. 14, p. 3202, 2019. [32] C.-S. Liao, R. Blanchard, C. Pfluegl, M. Azimi, F. Huettig, and
[9] N. Mitrayana et al., “CO2 laser photoacoustic spectrometer for measur- D. Vakhshoori, “Portable broadband photoacoustic spectroscopy for
ing ethylene, acetone, and ammonia in the breath of patients with renal trace gas detection by quantum cascade laser arrays,” Opt. Lett., vol. 45,
disease,” Sens. Bio-Sens. Res., vol. 30, Mar. 2020, Art. no. 100387. no. 12, pp. 3248–3251, 2020.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.
 5712 IEEE SENSORS JOURNAL, VOL. 23, NO. 6, 15 MARCH 2023

[33] E. Oscar Bonilla-Manrique, E. Julio Posada-Roman, Satyanarayanan Seshadri received the M.S.
A. Jose Garcia-Souto, and M. Ruiz-Llata, “Sub-ppm-level ammonia and Ph.D. degrees from Texas A&M University,
detection using photoacoustic spectroscopy with an optical microphone College Station, TX, USA, in 2004 and 2007,
based on a phase interferometer,” Sensors, vol. 19, no. 13, p. 2890, respectively.
2019. After completing his M.S. and Ph.D. degrees,
[34] F. Yehya and A. K. Chaudhary, “Designing and modeling of efficient he has worked as a Postdoctoral Research
resonant photo acoustic sensors for spectroscopic applications,” J. Mod. Associate with Rutgers University, New Jersey,
Phys., vol. 2, no. 4, pp. 200–209, 2011. USA, in the field of bioaerosols. He is currently
[35] R. A. Rooth, A. J. L. Verhage, and L. W. Wouters, “Photoacoustic the Principal Investigator of the Energy and
measurement of ammonia in the atmosphere: Influence of water vapor Emissions Laboratory, Indian Institute of Tech-
and carbon dioxide,” Appl. Opt., vol. 29, no. 25, pp. 3643–3653, 1990. nology Madras, Chennai, India. He is also an
[36] M. Giglio, P. Patimisco, A. Sampaolo, G. Scamarcio, K. F. Tittel, and Associate Professor with the Department of Applied Mechanics, Indian
V. Spagnolo, “Allan deviation plot as a tool for quartz-enhanced pho- Institute of Technology Madras. He was actively involved in guiding
toacoustic sensors noise analysis,” IEEE Trans. Ultrason., Ferroelectr., product and technology development activities for Forbes Marshall,
Freq. Control, vol. 63, no. 4, pp. 555–560, Apr. 2016. Chennai, India, in the areas of combustion, steam instrumentation, and
control systems. His research interests include waste heat recovery,
emission monitoring, and optics-based instrumentation.

K. Saran Kumar received the B.E. (Hons.) S. M. Shiva Nagendra is presently working as
degree in mechatronics engineering from Anna a Professor with the Department of Civil Engi-
University, Coimbatore, India, in 2011, and the neering, Indian Institute of Technology Madras
M.E. degree in computer integrated manufac- (IITM), Chennai, India. He has more than
turing engineering from the College of Engi- 24 years’ experience in research, teaching, con-
neering Guindy, Anna University, Chennai, India, sultancy, and community development. He has
in 2013. He is pursuing the Ph.D. degree in published more than 100 research publications
the area of optical trace gas sensing with the in international and national refereed journals,
Department of Engineering Design, IIT Madras, two reference books, more than 100 papers
Chennai. in conferences, and three patents. He is the
His research interests include photoacous- author of books titled Urban Air Quality Moni-
tic spectroscopy, hollow core-based photothermal interferometry, and toring, Modeling and Human Exposure Assessment (Springer, ISBN:
mechatronics system design. 978-981-15-5511-4) and Artificial Neural Networks in Vehicular Pollution
Modelling (Springer, SCI-41, ISBN-10: 3-540-37417-5). He has also
edited seven conference proceedings. His research interests focus on
air quality management which includes monitoring, source apportion-
ment, modeling, design and development of emission control systems,
development of air quality management systems, personal exposure
monitoring, environmental impact assessment, outdoor–indoor air pol-
lution relationships, and indoor air quality management.
Dr. Nagendra is a professional member of several technical institutions
Y. Esther Blesso Vidhya received the Ph.D. and organizations of India. He has been part of numerous grants, involv-
degree in optomechatronics from the Indian ing multiple national and international partners. He is a WHO Subject
Institute of Technology Madras, Chennai, India, Matter Expert in a global technical consultation on the transmission of
in 2018. respiratory pathogens through the air. He is the Vice Chairperson of
She was a Postdoctoral Researcher with the GATE, JAM 2023, IIT Madras. He is the Director of Clean Environment
Department of Engineering Design, Indian Insti- for Planetary Health in Asia (CEPHA) Network supported by UKRI. He is
tute of Technology Madras, from 2018 to 2022. the Founder Chairperson of the Indian International Conference on Air
She is currently working as a Research Engineer Quality Management (IICAQM) Series and the Founder President of
with Saint Gobain Research India, Chennai. Air Quality Management Association (AQMA). He is also an Associate
Her research interests include laser material Editor of the journal Frontiers in Sustainable Cities (Frontiers), and
processing and characterization, phase change Journal of the Institute of Engineers (India): Series A, Springer.
materials for photonics, and laser-assisted sensing.

Nilesh J. Vasa received the B.E. degree in pro-

duction engineering from the Veermata Jijabai
Ramya Selvaraj received the B.E. degree in Technological Institute (VJTI), Mumbai Univer-
electrical and electronics engineering from Anna sity, Mumbai, India, in 1988, the M.Tech. degree
University, Chennai, India, in 2011, the master’s in mechanical engineering from the Indian Insti-
degree in power electronics and drives from the tute of Technology Madras, Chennai, India,
PSG College of Technology, Coimbatore, India, in 1990, and the Doctorate of Engineering
in 2013, and the Ph.D. degree in optical sensors degree in electronic device engineering from
from the Department of Engineering Design, Kyushu University, Fukuoka, Japan, in 1997.
Indian Institute of Technology Madras, Chennai, From 1997 to 2005, he was a Faculty Member
in 2019. with Kyushu University. Since 2006, he has been
Her research interests include wavelength- a Faculty Member with the Department of Engineering Design, IIT
modulated photoacoustic spectroscopy, chemi- Madras. His research interests include laser micromanufacturing, laser-
cal and biomedical sensing, and hollow core fiber-assisted sensing. assisted sensing, and biophotonics devices.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 27,2023 at 06:33:27 UTC from IEEE Xplore. Restrictions apply.