Article

pubs.acs.org/ac

Photoacoustic Spectroscopy for the Quantification of N2O in the Off-

Gas of Wastewater Treatment Plants
Klemens M. Thaler,† Christoph Berger,† Carmen Leix,‡ Jörg Drewes,‡ Reinhard Niessner,†
and Christoph Haisch*,†
†
Chair of Analytical Chemistry, Institute of Hydrochemistry, Technical University of Munich, Marchioninistrasse 17, 81377 Munich,
Germany
‡
Chair of Urban Water Systems Engineering, Technical University of Munich, Am Coulombwall 3, 85748 Garching, Germany
*
S Supporting Information

ABSTRACT: Different configurations of photoacoustic (PA)

setups for the online-measurement of gaseous N2O, employing
semiconductor lasers at 2.9 and 4.5 μm, were developed and tested.
Their performance was assessed with respect to the analysis of N2O
emissions from wastewater treatment plants. For this purpose, the
local N2O emissions of a wastewater treatment bioreactor was
sampled by a dedicated mobile sampling device, and the total N2O
emissions were analyzed in the gastight headspace of the bioreactor.
We found that the use of a quantum-cascade laser emitting at about
4.53 μm, operated in a wavelength modulation mode, in
combination with a conventional longitudinal PA cell yielded the
highest sensitivity (<100 ppbv). However, we also observed a strong cross-sensitivity to humidity, which can be explained by
increased V−T relaxation. This observation in combination with the limited dynamic range (max conc. ∼ 3000 ppmv) led us to
the use of the less-sensitive but spectroscopically more robust 2.9 μm laser. A detection limit below 1 ppmv, a dynamic range of
more than 4 orders of magnitude, no influence of humidity or any other substance relevant to the off-gas analysis, as well as a
comparable low price of the laser source made it the ideal tool for N2O analyses of the off-gas of a wastewater treatment plant.
Such a system was implemented successfully in a full-scale wastewater treatment plant. The results regarding the comparison of
different PA setups can be transferred to other systems, and the optimum performance can be selected according to the specific
demands.

N itrous oxide (N2O) is a strong greenhouse gas with a

global warming potential 298 times higher than that of
CO2 over a 100-year time horizon.1 N2O is also an ozone-
depleting substance (ODS) with its annual emissions of 17.7
teragrams of nitrogen being the largest ODS emitted in the 21st
century.2 About 30% of these N2O emissions are related to
human activities.3 This leads to an increase in the atmospheric
N2O concentration of about 0.2−0.3% year−1 or an increase of
almost 20% since 1750.4 Anthropogenic emissions come from
agricultural activities, fossil fuel combustion, and wastewater
treatment. For the calculation of present and future N2O
emissions and the goal of reducing global warming, reliable data
on N2O production and emission are required, which
necessitates robust sampling and measuring strategies. Gas
chromatography equipped with electron capture detection is
the method of choice at World Meteorological Organization’s
Global Atmosphere Watch (GAW) stations, and it is also the
most widely used analytical method.5 However, it has some
disadvantages: frequent calibration is required and reference
gases have to be injected frequently to avoid signal drift. The
major disadvantage is its inability to measure continuously.6 By
contrast, optical techniques allow rapid and continuous
measurement. Fourier-transform infrared spectroscopy is used
in the mid-infrared (MIR) region to quantify N2O. A limit of
detection (LOD) of 1 ppbv and a precision of 0.1 ppbv can be
achieved.7,8 Instruments are commercially available and can be
applied to field measurements. However, the commercial
instruments are hardly suitable for routine online analysis due
to the rather high costs.9 Lasers in the MIR region are also used
for multielement absorption analyses.10,11 In this case, only one
absorption line is selected, corresponding to the laser
wavelength. Photoacoustic (PA) spectroscopy is a highly
sensitive method to measure optical absorption and can be
realized in small and robust setups, thus being particularly
suitable for field measurements.
PA Detection of N2O. The PA effect is based on the
conversion of absorbed optical energy into heat, which again
leads to thermal expansion.12 For gas-phase PA spectroscopy,
usually a narrowband light source is intensity-modulated with a
frequency in the acoustic range (i.e., some kHz), which
generates a modulated acoustic expansion (i.e., an acoustic
wave). Alternatively, wavelength scanning over a specific
Received: February 9, 2017
Accepted: February 24, 2017
Published: February 24, 2017

© XXXX American Chemical Society A DOI: 10.1021/acs.analchem.7b00491

Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
absorption line of the target substance is possible. PA
spectroscopy has been applied in a surprisingly large variety
of configurations. In 1993, Yang et al. detected N2O in the
near-infrared region at 775.25 nm, generated by a titanium−
sapphire laser.13 Costopoulos et al. used a pulsed optical
parametric oscillator emitting at 2.86 μm, yielding an
impressive LOD of 60 ppbv. Others used a mechanically
modulated continuous-wave CO2-laser emitting at 10.532
μm,14 a semiconductor laser emitting around the N2O
absorption band at 2.292 μm,15 or a quantum cascade laser
(QCL).16,17
Lasers with a rapid wavelength tuning capability, such as
QCLs, allow for PA spectroscopy by wavelength modulation, as
first proposed by Castleden et al.,18 instead of the common
intensity modulation. Modulation of the wavelength over a
small range around a narrow absorption peak, typically in the
range of less than the fwhm of the absorption line, results in a
PA signal with twice the modulation frequency, allowing for
second-order detection. This signal is essentially baseline-free as
it is influenced neither by incoherent noise (acoustic noise from
outside the PA cell) nor by coherent noise, as caused by
wideband window absorption.19,20 Detection methods with
QCLs and multi pass sample cells have been developed.21 Only
a few examples on the combination of QCLs with photo-
acoustic spectroscopy (PAS) are described in the litera-
ture.16,17,22,23
In this paper, we present two new PA instruments for the
detection of N2O emissions from wastewater treatment plants
(WWTPs). The instruments were combined with a dedicated
sampling system to collect gaseous emissions from WWTP
reactors. The intended use of this system is a routine
application in WWTPs and similar applications to monitor
N2O emissions for a better process understanding and
eventually for the development of N2O mitigation strategies.
Hence, the price and robustness are key parameters for the
assessment of such a system. Consequently, we tested two
different PA systems, one employing a QCL source at 4.5 μm
excitation wavelength and one using a distributed feedback
(DFB) laser source at 2.9 μm. At 4.5 μm, a higher sensitivity is
expected because N2O has its strongest optical absorption band
at this wavelength (see Figure S1, Supporting Information
(SI)), whereas the DFB laser source is significantly more cost-
efficient; a QCL system is typically 3 to 5 times costlier than a
conventional diode laser of a comparable emission power.
■ EXPERIMENTAL SETUP
The aim of the presented work was the optimization of the
analytical performance of a PA system with a clear focus on the
routine application in the rough environment of WWTPs with
respect to sensitivity and selectivity as well as robustness and
costs. The complete analytical system for the quantification of
N2O emissions from WWTPs comprises a gas sampling hood, a
PA system, and a pump providing a continuous gas flow from
the hood to the detector. We compared two different laser
systems, operating at two different emission wavelengths,
different PA cell designs, and different signal detection and
demodulation systems.
Laser Systems. The PA setup basically comprises laser
source including the corresponding driver and thermostatic
control and a PA cell with an acoustic resonator (see Figure 1).
One laser source is a QCL (AdTechoptics, U.S.A.) purchased
from EQ Photonics (Eching, Germany) with an emission band
ranging from 4.53 to 4.54 μm and a maximum output power of
B DOI: 10.1021/acs.analchem.7b00491
Article
Figure 1. Fundamental experimental PAS setup.
134 mW. The wavelength was chosen to fit the absorption peak
of N2O in the IR. Wavelength tuning of the QCL is performed
by changing the laser temperature by means of a temperature
control unit (TEC-1089, Meerstetter Engineering GmbH,
Switzerland). The laser-diode driver (LDD; LDD-1124,
Meerstetter Engineering GmbH, Switzerland) is connected to
an Arduino Uno (Arduino.cc, Italy), which triggers the internal
modulation of the LDD. This is necessary because the LDD has
no reference output signal as required for lock-in detection.
The laser is modulated by a sinusoidal current (Imax = 320 mA)
with a frequency of 2 kHz (modulation depth = 30 mA).
Temperature of the QCL is set to 18.3 °C, corresponding to an
emission wavelength at the absorption peak of N2O. The laser
current modulation results in a laser wavelength modulation
around the center of the N2O absorption peak, which in turn
leads to a 2f PA signal. Thus, the first longitudinal acoustic
mode of the resonant PA cell at 4 kHz is excited. The QCL
light beam, collimated by a lens directly built into the laser unit
and then guided through two identical PA cells by two gold
mirrors. All optical components were purchased from Thorlabs
(Dachau, Germany). The second PA cell serves as reference
cell, which is sealed and filled with 1000 ppmv N2O. This cell is
built identically to the first PA cell and refilled every six months.
The PA signal of this cell is demodulated at 3f, resulting in a
zero PA signal at the absorption maximum, which again allows
for a precise identification of the maximum. This information is
used to control the QCL wavelength via the QCL temperature
and the in-house made control software written in LabView
2013 (National Instruments).
The second laser system consists of a 2.9 μm DFB laser
diode (Nanoplus, Germany) with an optical power of 5.5 mW.
An integrated thermoelectric cooling device and a negative-
temperature-coefficient thermistor was attached to the laser
diode (LD). Temperature and current were controlled by a
LDC1000 (Profile, U.S.A.), leading to a variable emission from
2893 to 2900 nm. The laser is modulated with a simple
transistor-to-transistor logic signal at 4 kHz. The laser beam is
focused into the PA cell described above by an aspherical lens
( f = 25.4 mm). For this setup, a second PA cell was not
necessary. Temperature regulation was precise enough to
ensure a high stability and therefore no wavelengths drift.
Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
PA Cells. The PA cells are conventional PA cells featuring
longitudinal acoustic resonances in a central optical resonator
formed by a cylindrical glass tube (length 40 mm, inner
diameter 4 mm) equipped in its middle with a condenser
microphone (EK-23029, Knowles, U.S.A.) for the PA signal
detection and a suitable preamplifier, as described elsewhere.24
In earlier experiments, we successfully used a Helmholtz type
resonator as the PA detection cell in combination with a
QCL.23 This cell type is particularly suitable for laser sources
with limited beam quality. The system essentially consists of
two coaxial cylinders (diameter 22 mm, length 2 mm),
combined by a connecting duct (diameter 4 mm, length 2
mm). The laser source is positioned directly at the center of
one of the cell inlets without any focusing element. Aluminum
was chosen as the cell material due to its high reflectivity for the
laser light; additionally, the inner surface of the cell was
polished. A gas inlet is positioned laterally into the same cell,
while a microphone sits in the other cell right opposite to the
connecting tube. Finite element simulation (Comsol Multi-
physics, Germany) was performed to estimate the eigenfre-
quency of the cell (see Figure S2, SI).
As an alternative to the microphone described above, which
was used in longitudinal PA cells, a microelectro-mechanical
system (MEMS) microphone from Knowles (SPM0408LE5H-
TB, Knowles, U.S.A.) was tested. The advantage of this
microphone is its high sensitivity of 126 mV/Pa and an output
impedance of 200 Ω, which makes higher amplification
possible. As an all solid-state device, it is particularly robust,
chemically as well as mechanically.
Signal Recording and Demodulation. Three different
approaches for signal recording and demodulation were
compared. The most conventional system uses a commercial
lock-in amplifier (SR 830, Stanford Research, U.S.A.) to detect
the preamplified PA signal. In the second approach, the signals
are recorded using an analogue to digital converter (ADC) (NI-
PCI 4474 dynamic signal acquisition card, National Instru-
ments, U.S.A.) recording four channels at 102.4 kS/S and a 24-
bit resolution. The signal lock-in is performed by an in-house
software, also programmed in LabView 2013. The third
approach employs a simple PC soundcard (24-bit resolution,
Xonar DG, ASUS, Taiwan) instead of the ADC for the signal
recording, but the same software-based demodulation. Temper-
ature and humidity of the gas are measured by analogue sensors
(HIH4010−004, Honeywell, U.S.A. and TMP36GZ, Analog
Devices, Inc. U.S.A.) connected to the measurement computer
via the Arduino. The whole setup is integrated in a mobile
housing, allowing field-measurements.
Gas Sampling System for WWTPs. For the collection of
the emitted off-gas at a specific spot in a full-scale wastewater
treatment plant (more specifically in the nitritation step of a
two-stage deammonification process being intermittently
aerated by aeration membranes installed at the bottom of the
reactor with an aeration duration of 45 min followed by an
unaerated phase of 15 min), a floating sampling hood was
placed in the reactor (see Figure S3, SI). The square-shaped
hood is made up of stainless steel (V2A) with floating bodies
on each side and a cross-sectional surface of 0.5 m2. A gas-inlet
and a gas-outlet are located on two opposite corners of the
hood. The gas-outlet is connected via gastight tubing to the PA-
detection system, which actively sucks in the sample air. Short
distances between the hood and the PAS system were
emphasized to enable a quick response time. A ball check
C DOI: 10.1021/acs.analchem.7b00491
Article
valve is connected to the top of the floating hood to prevent air
from entering.
To prevent foam, which can cover the reactor’s liquid surface
as high as 20 cm, from entering the sampling line, a sieve cloth
(60 μm mesh) is placed between the gas-outlet and the tubing.
A pump (VDE0530, KNF Neuberger GmbH, Germany)
provides a continuous gas-flow of 700 mL/min through the
whole system of the sampling hood and PA analyzer. During
the aerated phases of the process, excess air can escape from the
ball check valve, which is connected to a mass flow meter (red-
y, Vögtlin Instruments AG, Switzerland). The concentration
measured by our PA device, in combination with the total
emitted volume, which is the sum of the flow through our
instrument and the volume emitted through the ball check
valve, allows for a precise calculation of the total N2O
emissions. During unaerated phases, ambient air is aspirated
through the gas-inlet for keeping the system at ambient
pressure. The low flow rate in the sampling hood allows for
equilibration between the liquid-phase and the gas phase N2O
concentration. The sampling hood is held in position by an
aluminum bar of 4 m length, connected to the hood by a hinge
joint enabling us to follow the water surface even during
turbulent aeration phases. Simultaneous determination of the
dissolved N2O concentration is performed using an electro-
chemical N2O sensor with a detection limit of 0.5 μM, a
response time below 45 s and a working range up to 300 μM
(Unisense Environment A/S, Denmark), fixed in the center
part of the hood approximately 20 cm below the liquid surface.
As an alternative to the measurement using the sampling hood,
the total N2O emission of the reactor could be sampled via the
headspace, as this reactor is covered by a nearly gastight hatch.
An exhaust tube on its top allows for a time-resolved, spatially
integrating analysis of the gas emission.
Calibration Gases. For the calibration of the PA-systems,
commercial test gases of different concentrations were used:
98%, 1%, 200 ppmv and 20 ppmv N2O in nitrogen (Westfalen
AG, Germany). The test gases were diluted with pure nitrogen
(99.99% N2) using rotameters. To assess the influence of
humidity on the PA signal, water-saturated test gases were
achieved by bubbling the gases through a washing bottle and
then diluted to the required humidity.
■
RESULTS AND DISCUSSION
Comparison of the Recording and Demodulation
Systems. A comparison of the three data recording systems
was performed using the 2.9 μm DFB laser source. It was found
that the Stanford Research SR830 lock-in amplifier yields an
approximately 10% lower standard deviation, resulting in a
slightly lower LOD than the combination of the ADC with the
software demodulator (see Figure S4, SI).
The comparison of the ADC with the soundcard yielded the
surprising result that there was no significant difference
between the two with respect to the LOD and the relative
standard deviation of the signals. A major limitation of the
soundcard approach is that it has only two channels, one for the
PA signal and one for the modulation reference required for the
lock-in, meaning that the detection of two independent PA
channels is not possible (i.e., the signal from two PA cells
cannot be recorded simultaneously). The SR830 has two input
channels plus a separate channel for the modulation reference,
but these two channels can only be used in a differential way,
not for independent measurements. Only the ADC has four
independent channels, of which one is required for the
Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
modulation reference. As mentioned above, for the 4.5 μm
QCL laser source a second PA cell with a corresponding
detection channel is required. Hence, we used the combination
of the ADC with a software lock-in throughout the following
experiments.
Comparison of the Laser Sources. The combination of
the intensity modulated (1f) 2.9 μm DFB laser system with the
conventional longitudinal PA cell and the ADC detection
system demonstrated to be a good compromise between
sensitivity, robustness, and cost efficiency. Figure 2 depicts a
Figure 2. Simulated spectrum of 70 ppmv N2O + 1% H2O in air, on
the basis of absorption data from the HITRAN database (red line),
compared with the measured spectrum with a DFB laser (blue line).
scan over the wavelength range accessible with this laser source,
measured in air containing 70 ppmv N2O in comparison to the
theoretical absorption values of N2 containing N2O and H2O as
extracted from the HITRAN database.25 While the overall
correlation between theory and measured data is satisfying, the
divergence between the two increases at the high wavelength
range of the covered spectrum. The reason is the lower laser
emission intensity of the laser at higher temperatures. A
temperature ramp is required to tune the laser emission
wavelength. For further measurements, 2894.7 nm was chosen
as the signal wavelength. The H2O influence should be
negligible in this region of the spectrum. For an application in
WWTPs cross-sensitivity to other gases was assessed. The
following substances were reported to be emitted from typical
reactors: methane, ammonia, carbon dioxide, hydrogen sulfide,
and methyl mercaptan.26,27 Absorption coefficients of the
mentioned gases are significantly lower than the absorption
coefficient of N2O in the two wavelength ranges of the lasers as
it can be appreciated in S5 (SI).28 The absorption spectra
presented there are calculated with the concentration listed in
the reference by using the HITRAN data.25 This low spectral
overlap in combination with the wavelength tuning approach
rules out the risk of spectral interferences.
A calibration with this setting and the resonant PA cell was
performed (see Figure 3). From this data, an LOD of 2.0 ppmv
can be calculated (3·sb-criterion). Further improvement to an
LOD of 0.82 ppmv is possible by switching to 2f modulation; at
the same time, the calibration is linear up to a concentration of
1%. This dynamic range of more than 4 orders of magnitude is
certainly one of the most impressive figures of merit of this PA
detection system. In industrial process monitoring like the
intended use at WWTPs, such a dynamic range is essential, as
the N2O concentrations can vary greatly depending on diverse
influencing factors (e.g., the aeration regime). The 2.9 μm DFB
D DOI: 10.1021/acs.analchem.7b00491
Article
Figure 3. Correlation of the PA signal excited at 2.9 μm to the N2O
concentration (T = 295 K, cwater = 6.83 g/m3).
laser system was tested also for long-term stability. After a
continuous use of 24 h/day for 3 months, a repeated calibration
was within the standard deviation of the previous signals.
With the use of the MEMS-microphone instead of the
electret microphone, a significantly higher sensitivity was
expected. Indeed, experimental results show signals that are
doubled. However, the noise also more than doubled.
Additionally, the MEMS microphone we used features a nearly
constant frequency response down to 100 Hz, making the
system very sensitive to the ubiquitous 1/f background noise.
Thus, continuous flow measurements were impossible. The
LOD of this configuration was 3.85 ppmv. The approach of
combining two of these microphones, one in each of the two
cells of the Helmholtz resonator, operated in a differential
mode, was not successful.
The N2O absorption at 4533.7 nm is the highest of the
complete IR spectrum and it is not cross-sensitive to any other
relevant component (CH4, NH3, CO2, CH3SH). Consequently,
the highest sensitivity of the PA analysis is to be expected at this
wavelength. The selected QCL features its central wavelength
at exactly this wavelength. Figure S6 (SI) displays a calibration
of the PA system at this wavelength, produced with 2f
wavelength-modulation schema and the conventional longi-
tudinal PA cell. The LOD here was found to be 77 ppbv, thus
well below the natural atmospheric background concentration
and more than sufficient for our application. However, the
dynamic range of the system when using the QCL is limited to
a maximum of about 3000 ppmv; at higher concentrations, the
PA signal saturates, and above 4000 ppmv it even decreases.
We assume that this behavior is due to the high optical
absorption at the selected wavelength, which leads to a
significant loss of optical energy along the optical axis before
reaching the center of the PA cell. This effect is not observed
for the 2.9 μm PA setup because of the lower absorption
coefficient of N2O at 2.9 μm.
Another important feature is shown in Figure 4, where the
PA signal at different constant N2O concentrations is plotted
against the humidity. A spectral interference of the H2O
absorption with the N2O measurement can be excluded. For
the 2.9 μm DFB laser, no cross-sensitivity to H2O is observed.
The QCL PA signals of N2O scale linearly with the humidity.
Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Figure 4. QCL PA signal at different humidities and concentrations
(no. of repeated measurements: 5; no. of data points: 9).
At a temperature of 22 °C and a relative humidity of 100%, the
PA signals of N2O is about 7 times higher than at 0% humidity
for all N2O concentrations in the linear calibration range. A
possible reason for this behavior is vibrational−translational
(V−T) relaxation of N2O molecules. Briefly, collisions with
H2O molecules lead to a reduced lifetime of the state excited by
the absorption, which again lead to a more intense, in-phase PA
signal.29−32
Obviously, for a practical PA analysis at this wavelength of
4.5 μm, either the humidity should be kept constant or the
influence of humidity should be eliminated. We chose the latter
option: the humidity sensor mentioned above is used to
measure simultaneously temperature and humidity of the gas
flowing through the PA cell. Our LabView control program
implements a routine that corrects the N2O concentration
using the measured values of the sensor and the slope of the
linear regressions for constant N2O concentrations at different
humidities measured in the laboratory. A disadvantage of this
procedure is the spatial separation of humidity determination
and PA measurement, leading to a temporal offset that has to
be corrected for. For our specific application at WWTPs, we
found that the relative humidity (100%) as well as the
temperature (22 °C) were constant throughout all our
experiments.
Alternative Cell Design and Setups. Finite element
simulations with COMSOL Multiphysics 5.0 (Comsol A.B.,
Sweden) determined the resonance frequency of the Helmholtz
resonator to be 3506 Hz, while experimentally, we found a
resonance maximum at 3640 Hz, which is a satisfying result
considering the limitations of such static simulation. The
determined resonance frequency fits into the frequency range
of maximum sensitivity for both microphones.
The use of intensity modulation (on/off) was not possible
for this cell type because absorption of the laser at the cell wall
results in a very high background signal. Therefore, 2f
frequency modulation was used. Measurements with this cell
type show a high standard deviation for concentrations below 1
ppmv. The detection limit was determined to be 1.98 ppmv,
which is 2 times higher than that for the longitudinal PA cell at
2f modulation.
Comparison of the Two PA Setups. The highest
sensitivity is achieved for the QCL PA system. The use of
E DOI: 10.1021/acs.analchem.7b00491
Article
software lock-in amplifier and OEM-laser drivers makes it
possible to integrate the whole setup in a small housing of 30
cm × 40 cm × 22 cm (see Figure S7, SI), making the system
highly mobile. A disadvantage of this system is the high
dependency of the measured signal on the humidity. Even
though calibrations for different humidities have been
performed, a complete correction, particularly for varying
humidities, is difficult and reduces the precision. Possible
humidification of the incoming sampling gas may result in
higher signals and controlled conditions for quantification. A
desiccation of the sampling gas may result in analyte loss.
Additionally, the lower signals are disadvantageous.
The LOD of the 2.9 μm PA setup is more than 1 order of
magnitude lower than that of the QCL PA system. The reason
is the lower power of the laser diode and the lower absorption
coefficient of N2O at 2.9 μm. The dynamic range of this system
is better than that of the QCL PA setup. A major advantage of
this setup is the humidity independence. Table 1 compares the
tested PA systems.
Table 1. Comparison of the Different PA Systems
Investigated
QCL system
(4.5 μm) DFB system (2.9 μm)
longitud. PA Helmholtz PA
longitud. PA cell cell cell
sensitivity 77 ppbv 0.82 ppmv 1.98 ppmv
dynamic range 104 104 n.d.
H2O influence strong − −
mech. stability + +++ ++
mobility ++ + ++
In summary, we decided to use the 2.9 μm DFB laser system
with 2f modulation and ADC signal recording for the field
applications described below. It provides the best compromise
between sensitivity on the one hand and robustness,
spectroscopical as well as mechanical, on the other hand.
These stability considerations include the long-time stability of
the calibration, which did not feature any significant changes
throughout our field measurements (i.e., for at least 3 months).
The limited dynamic range and the cross sensitivity of the 4.5
μm QCL system to water plus the significantly higher price
make this approach less suitable for the specific application,
particularly as the sensitivity of the 2.9 μm DFB system is more
than sufficient for this task.
Field Application on a WWTP. As an example application
we tested the 2.9 μm PA system for its practical applicability.
Two sampling approaches were tested: using the gas sampling
hood designed for spatially resolved sampling and total
emission analysis over the entire reactor headspace. The
temporal behavior of the N2O concentration in the liquid-phase
and in the PA-measured gas-phase over three consecutive
aeration cycles is shown in Figure 5. It can be appreciated that
the liquid-phase concentration is increasing over a period of 15
min. Then aeration starts, and the liquid-phase concentration
decreases quickly throughout the full aeration cycle of 45 min,
as N2O is stripped out of the liquid. With the stripping gas,
dissolved N2O is transferred into the gas phase, as the PA data
reveal. When the aeration ends, the liquid-phase concentration
starts to build up again. The gas-phase concentration increases
at the beginning of the aeration, reaches a maximum after about
15 min, and then decreases constantly. The reason is that the
Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Figure 5. N2O concentration in the gas phase (blue) measured by our
PA system and in the liquid phase (red), monitored at a WWTP.
liquid-phase concentration is already significantly reduced by
the stripping, and thus, only a small fraction of N2O can be
transferred into the gas phase.
The data shown in Figure 5 reveal that the PA setup has a
higher temporal resolution than the electrochemical sensor.
The short-term increases in the decreasing gas-phase
concentration (e.g., at 12.10 h) is due to local fluctuations.
The bioreactor content is not homogeneous and has strong
spatial variations. A mixing device installed at medium height of
the reactor leads to local fluctuations. The total emission
sampling at the top of the reactor cover avoids this problem. An
overnight analysis is plotted in Figure S8 (SI). Qualitatively, the
behaviors of both, gas-phase and liquid-phase N2O concen-
trations are identical. The variations over the whole analysis
time can be attributed to spatially variable conditions in the
reactor, such as different aeration intensities causing not only
differential stripping effects, but also varying dissolved oxygen
concentration affecting the biological N2O production.
■
CONCLUSIONS
For the detection and quantification of gaseous N2O
concentration, different PA setups were tested and compared,
namely, a QCL system at 4.5 μm and a DFB system at 2.9 μm.
The LOD was improved by the use of 2f modulation. The costs
of the setups were reduced using a PC soundcard as an analog-
to-digital converter and a software lock-in amplifier instead of a
much larger instrumental lock-in amplifier. This also allows for
the development of a robust and highly mobile setup, which
was tested at a WWTP. The results regarding the comparison
of different PA setups is valid for other analyst accessible in the
IR range and can thus be transferred to other applications, such
as automotive exhaust gas monitoring or industrial process
monitoring. The specific optimum performance can be selected
according to the specific demands.
■
ASSOCIATED CONTENT
*
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.anal-
chem.7b00491.
Additional data including absorption coefficients of N2O,
finite element simulation of acoustic pressure and sound
pressure, image of gas sampling hood, relative standard
deviation of a commercial lock-in amplifier compared to
a software lock-in amplifier, simulated absorption spectra
of the relevant biogas components, calibration of the
QCL-PA-system, image of PA prototype instrument, and
N2O concentration in WWTP (PDF)
F DOI: 10.1021/acs.analchem.7b00491
Article
■ AUTHOR INFORMATION
Corresponding Author
*E-mail: Haisch@tum.de. Tel. +49 89 2180-78242. Fax -99
78242.
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
The authors thank S. Wiesemann and R. Hoppe of the
institute’s mechanical workshop for the construction of the
mobile PA-system and the development of the gas sampling
device. We thank also Prof. Frank K. Tittel for his support.
Financial support of the project and particularly of K.T., C.B.,
and C.L. by the International Graduate School of Science and
Engineering IGSSE of the Technical University of Munich is
gratefully acknowledged.
■
REFERENCES
(1) Myhre, G., Shindell, D.; Bréon, F.-M., Collins, W.; Fuglestvedt, J.;
Huang, J.; Koch, D.; Lamarque, J.-F.; Lee, D.; Mendoza, B.; Nakajima,
T.; Stephens, G.; Takemura, T.; Zhang, H.. In Climate Change 2013:
The Physical Science Basis. Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change;
Stocker, T. F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K.,
Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P. M., Eds.;
Cambridge University Press: Cambridge, U.K., 2013.
(2) Ravishankara, A. R.; Daniel, J. S.; Portmann, R. W. Science 2009,
326, 123−125.
(3) Wuebbles, D. J. Science 2009, 326, 56−57.
(4) Machida, T.; Nakazawa, T.; Fujii, Y.; Aoki, S.; Watanabe, O.
Geophys. Res. Lett. 1995, 22, 2921−2924.
(5) World Meteorological Organization (WMO). GAW Report No.
185. Guidelines for the Measurement of Methane and Nitrous Oxide and
their Quality Assurance; WMO: Geneva, Switzerland, 2009.
(6) Hamilton, J. F.; Lewis, A. C. In Analytical Techniques for
Atmospheric Measurement; Dwayne, H., Ed.; Blackwell Publishing:
Oxford, 2006.
(7) Hammer, S.; Griffith, D. W. T.; Konrad, G.; Vardag, S.; Caldow,
C.; Levin, I. Atmos. Meas. Tech. 2013, 6, 1153−1170.
(8) Griffith, D. W. T.; Deutscher, N. M.; Caldow, C.; Kettlewell, G.;
Riggenbach, M.; Hammer, S. Atmos. Meas. Tech. 2012, 5, 2481−2498.
(9) Rapson, T. D.; Dacres, H. TrAC, Trends Anal. Chem. 2014, 54,
65−74.
(10) Yu, Y.; Sanchez, N. P.; Griffin, R. J.; Tittel, F. K. Opt. Express
2016, 24, 10391−10401.
(11) Dong, L.; Tittel, F. K.; Li, C. G.; Sanchez, N. P.; Wu, H. P.;
Zheng, C. T.; Yu, Y. J.; Sampaolo, A.; Griffin, R. J. Opt. Express 2016,
24, A528−A535.
(12) Haisch, C. Meas. Sci. Technol. 2012, 23, Article No. 012001.
(13) Yang, X. K.; Petrillo, C. J.; Noda, C. Chem. Phys. Lett. 1993, 214,
536−540.
(14) Schramm, D. U.; Sthel, M. S.; da Silva, M. G.; Carneiro, L. O.;
Souza, A. P.; Vargas, H. Rev. Sci. Instrum. 2003, 74, 513−515.
(15) Grossel, A.; Zeninari, V.; Parvitte, B.; Joly, L.; Courtois, D. Appl.
Phys. B: Lasers Opt. 2007, 88, 483−492.
(16) Rocha, A. M.; Sthel, M. S.; de Castro, M. P. P.; Mothe, G. A.;
Silva, W. C.; Perez, V. H.; da Silva, M. G.; Miklos, A.; Vargas, H. Energy
Fuels 2014, 28, 4028−4032.
(17) Lima, J. P.; Vargas, H.; Miklos, A.; Angelmahr, M.; Hess, P.
Appl. Phys. B: Lasers Opt. 2006, 85, 279−284.
(18) Castleden, S. L.; Kirkbright, G. F.; Spillane, D. E. M. Anal. Chem.
1981, 53, 2228−2231.
Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry Article

(19) Peng, Y.; Zhang, W.; Li, L.; Yu, Q. X. Spectrochim. Acta, Part A
2009, 74, 924−927.
(20) Schilt, S.; Kosterev, A. A.; Tittel, F. K. Appl. Phys. B: Lasers Opt.
2009, 95, 813−824.
(21) Santoni, G. W.; Daube, B. C.; Kort, E. A.; Jiménez, R.; Park, S.;
Pittman, J. V.; Gottlieb, E.; Xiang, B.; Zahniser, M. S.; Nelson, D. D.;
McManus, J. B.; Peischl, J.; Ryerson, T. B.; Holloway, J. S.; Andrews,
A. E.; Sweeney, C.; Hall, B.; Hintsa, E. J.; Moore, F. L.; Elkins, J. W.;
Hurst, D. F.; Stephens, B. B.; Bent, J.; Wofsy, S. C. Atmos. Meas. Tech.
2014, 7, 1509−1526.
(22) Couto, F. M.; Sthel, M. S.; Castro, M. P. P.; da Silva, M. G.;
Rocha, M. V.; Tavares, J. R.; Veiga, C. F. M.; Vargas, H. Appl. Phys. B:
Lasers Opt. 2014, 117, 897−903.
(23) Lendl, B.; Ritter, W.; Harsek, M.; Niessner, R.; Haisch, C.
Proceedings of IEEE Sensors 2006, EXCO; Daegu, Korea, October 22−
25, 2006; pp 338−341.
(24) Beck, H. A.; Niessner, R.; Haisch, C. Anal. Bioanal. Chem. 2003,
375, 1136−1143.
(25) The HITRAN Database. http://www.cfa.harvard.edu/
HITRAN/.
(26) Frison, N.; Chiumenti, A.; Katsou, E.; Malamis, S.; Bolzonella,
D.; Fatone, F. J. Cleaner Prod. 2015, 93, 126−133.
(27) Lackey, J. C.; Peppley, B.; Champagne, P.; Maier, A. Waste
Manage. Res. 2015, 33, 767−771.
(28) Du, Z.; Zhen, W.; Zhang, Z.; Li, J.; Gao, N. Appl. Phys. B: Lasers
Opt. 2016, 122, 100.
(29) Rooth, R. A.; Verhage, A. J. L.; Wouters, L. W. Appl. Opt. 1990,
29, 3643−3653.
(30) Kosterev, A. A.; Mosely, T. S.; Tittel, F. K. Appl. Phys. B: Lasers
Opt. 2006, 85, 295−300.
(31) Avramides, E.; Hunter, T. F. J. Chem. Soc., Faraday Trans. 2
1979, 75, 515−527.
(32) Avramides, E.; Hunter, T. F. Chem. Phys. 1981, 57, 441−451.

G DOI: 10.1021/acs.analchem.7b00491
Anal. Chem. XXXX, XXX, XXX−XXX