Home Search Collections Journals About Contact us My IOPscience

Emission considering self-absorption of OH to simultaneously obtain the OH density and gas

temperature: validation, non-equilibrium effects and limitations

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2017 Plasma Sources Sci. Technol. 26 095007

(http://iopscience.iop.org/0963-0252/26/9/095007)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 132.174.250.220
This content was downloaded on 31/08/2017 at 14:41

Please note that terms and conditions apply.

 Plasma Sources Science and Technology

Plasma Sources Sci. Technol. 26 (2017) 095007 (11pp) https://doi.org/10.1088/1361-6595/aa8688

Emission considering self-absorption of OH

to simultaneously obtain the OH density and
gas temperature: validation, non-equilibrium
effects and limitations
Yanjun Du1,2, Gaurav Nayak1, Gaku Oinuma1,3, Yanjun Ding2,
Zhimin Peng2,4 and Peter J Bruggeman1,4
1
Department of Mechanical Engineering, University of Minnesota, 111 Church Street SE, Minneapolis,
MN 55455, United States of America
2
Department of Thermal Engineering, State Key Laboratory of Power Systems, Tsinghua University,
Beijing, 100084 People’s Republic of China
3
Advanced Technology R&D Center, Mitsubishi Electric Corporation, Hyogo 661-8661, Japan

E-mail: [email protected] and [email protected]

Received 10 June 2017, revised 27 July 2017

Accepted for publication 16 August 2017
Published 31 August 2017

Abstract
The measurement of absolute densities of ubiquitous OH radicals and gas temperatures in water
containing plasmas has recently drawn a lot of attention. In this paper, we extend the self-
absorption model introduced in Du et al 2016 Plasma Sources Sci. Technol. 25 04LT02 with a
description of the excited state by a superposition of two Boltzmann distributions to take the
non-thermal rotational distribution of the excited state into account. This technique is applied to
a diffuse He+H2O RF discharge and it is shown that in addition to the determination of the
ground state OH density and rotational temperature, the properties of the excited state OH(A) can
also be simultaneously determined. A model of the steady-state distribution of the hot and cold
group density of OH(A) is able to describe the dependence of the rotational population
distributions of the excited state as a function of the water concentration. The method is also
applied to a filamentary Ar+H2O DBD. While the non-homogeneous nature of the DBD leads
to complications, the production of OH(A) by multiple production mechanisms leading to a
complex nascent rotational population distribution causes the fitting procedure of the emission
spectrum (with self-absorption) to break down.

Keywords: self-absorption, optical emission spectroscopy, atmospheric pressure plasma, OH

radical densities, non-equilibrium

1. Introduction used to achieve this goal. However, in addition to the

requirement of complex collisional models and an absolute
In the last few decades, the measurement of absolute densities calibration for LIF [7], several of these methods require
of ubiquitous OH radicals has been intensively investigated, optical access through multiple windows and some even
as water containing atmospheric-pressure plasmas have con- expensive laser systems, which are not readily available in
siderable potential in environmental, bio-medical, and mat- every laboratory.
erial processing applications. Several methods, including Therefore, there has been an increasing interest in using
laser-induced fluorescence (LIF) [1, 2], cavity ring-down optical emission spectroscopy (OES) that has the advantage
spectroscopy [3] and broadband absorption [4–6], have been of being non-intrusive, inexpensive, and easy to implement
[8–10]. Specifically, the OH(A–X) emission band is also often
4
Authors to whom any correspondence should be addressed. used for gas temperature determination [10–13]. However,

0963-0252/17/095007+11$33.00 1 © 2017 IOP Publishing Ltd

Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

Figure 1. Schematic representation of the experimental setup. L1–L3 are three convex lenses with focal length of 10 cm, 10 cm and 5 cm
respectively. D1 is a diaphragm with a diameter of 1 mm. D2 is the entrance slit of the monochromator with a width of 50 μm. The top view
of the plasma reactor shows the two directions used for the optical detection.

limited research has been reported using OES to obtain 2. Experimental setup
absolute OH densities [14, 15]. The key reason is that the
complexity of emission spectra requires detailed collisional 2.1. Plasma setup
radiative models and a good knowledge of the plasma prop-
The detailed assessment of the self-absorption technique was
erties before reliable plasma parameters can be obtained.
performed in an atmospheric pressure He–H2O RF discharge.
Recently, Du et al [16] used a molecular absorption
The schematic of the setup is shown in figure 1. The dis-
model to obtain the absolute OH density in an atmospheric
charge is generated between two parallel copper electrodes
He+H2O RF discharge. Although accurate OH densities
with a length of 76 mm and a width of 3 mm. The inter-
have been obtained by this method, it was achieved with a
electrode distance is 2 mm. The edges of the electrodes are
very high-resolution (∼2.6 pm) spectrometer, which is not
rounded off to avoid high local electric fields. For the purpose
readily available in most labs. In addition, we assumed that
of conducting optical measurement, the plasma reactor has
the OH(A) state was well characterized by a Boltzmann dis-
ample optical access through four quartz windows. These
tribution. This was only possible when considering transitions
windows are sealed with O-rings and mounted on the main
with small rotational numbers (J′<6.5). Indeed, many body of the reactor. The discharge is intentionally much
OH(A) rotational population distributions in low temperature longer (along the gas flow direction) than wide. The optical
non-equilibrium plasmas show significant deviation from a access in directions parallel (direction A) and perpendicular
Boltzmann distribution [4, 16]. These non-equilibrium char- (direction B) to the gas flow, as shown in the top view in
acteristics can lead to significant limitations of the usability of figure 1, allows recording spectra with and without self-
the method as the accuracy of the OH density measurement is absorption (see further).
sensitive to the rotational distribution of the OH(A) state. An amplifier (E&I AB250) was used to amplify the RF
In this contribution, we present partially resolved OH(A–X) sinusoidal waveform at 13.56 MHz generated by a function
spectra that are fitted with the same technique enabling the generator (Tektronix AFG 3022B). The coupling of the RF
simultaneous determination of the properties of the ground state power to the plasma was achieved by a homemade π—type
(rotational temperature Tr(X), and OH density nOH) and the matching network consisting of two adjustable vacuum
excited state assuming a superposition of two Boltzmann dis- capacitors and an inductor. The current and voltage were
tributions with different temperature parameters (Tr H(A), measured by a current monitor (Pearson 2877) and a high
Tr C(A), R). Tr H(A) and Tr C(A) refer to the hot and cold group of voltage probe (Tektronix—P6015A).
the excited state and R is the ratio of hot to the total OH(A) The plasma-dissipated power was obtained in real time
population. By analyzing the parameters of the excited state, using the automatic integral function on the oscilloscope
Tr H(A), Tr C(A) and R, the production mechanism of the OH(A) (Teledyne Lecroy, WaveSurfer 10, 1 GHz, 10 GS s−1) on the
state is analyzed. The accuracy of the model is assessed by measured current and voltage signals in order to keep the
comparing spectra from the same discharge but different power constant. The measurements were performed at a fixed
directions of detection. This allows the comparison of spectra power of 7.8 W throughout the work.
with and without self-absorption. The results are also compared To achieve the mixture of water vapor and helium, dry
with broadband absorption measurements. The self-absorption helium was sent through a bubbler (Pyrex 500 ml), after
method is applied both on a diffuse He+H2O RF discharge which the helium can be assumed to be saturated with water
and a filamentary Ar+H2O DBD. The limitations of the vapor (2.6%) at room temperature (295 K). Two mass flow
method are discussed based on these results. controllers (MKS GE50A) were used to control the admixture

2
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

concentration by varying the fraction of humid helium with 3. Method

respect to the dry helium. The used water concentrations
ranged from 0.1% to 1% while the total gas flow rate fed into 3.1. Absorption, emission and self-absorbed emission
the reactor was kept constant at 2 standard liters per minute
3.1.1. Broadband absorption spectrum. The detailed procedure
for all conditions.
of broadband absorbance is reported in [4, 16]. In brief, the
Similar measurements were also conducted for an
absorbance is calculated as follows:
Ar+H2O DBD at atmospheric pressure. The used DBD
setup is identical to the setup described in [17]. In brief, the ⎛ Il + p (v) - Ip (v) ⎞
A exp (v) = - ln ⎜ ⎟, (1 )
plasma is generated between two electrodes with a length of ⎝ Il (v) - Ib (v) ⎠
8 cm and a width of 3 mm. The optical access also enables to
record spectra from both directions A and B as shown in where Ip and Il are the emission spectra of the plasma and light
figure 1. In the underlying work, we only consider a fila- source, respectively; Il+p is the spectrum of plasma
with lamp on; Ib is the recorded background spectrum (both
mentary DBD in Ar with 1% water and a power of 8 W.
plasma and lamp off) and v is the frequency. The theoretical
Detailed results and discussion of the plasma parameters for a
absorbance is calculated as follows,
broader operational range have been reported in [17].
⎛ I cal (v) Ä Vins ⎞
A cal (v) = - ln ⎜ t ⎟, (2 )
⎝ I0 ⎠
2.2. Optical measurement
where Vins is the Voigt profile of the instrumental broadening
The spectrally resolved measurements are performed with a function. Itcal (v ) is the transmitted light intensity calculated
monochromator (ACTON AM510) with a focal length of based on the Beer–Lambert law [4, 16] and the incident light I0.
1 meter equipped with a 1200 grove mm−1 grating. The The absorption spectrum can be described as a function with the
spectrometer is backed by a 26 μm pitch CCD camera (Andor following parameters:
DU420A-OE) to record the spectra. The spectral resolution of
A cal (v) ~ f (Tr (X ) , n OH , v). (3 )
this system is 25 pm FWHM at 310 nm.
Figure 1 shows the optical configuration for recording the
absorption and emission spectra of OH. The light from the
reactor is focused by the lens L2 ( f=10 cm) onto the dia- 3.1.2. Emission spectrum. The rotational emission spectra
phragm D1 and then imaged onto the entrance slit of the exhibit a strong non-equilibrium of the rotational population
monochromator by lens L3 ( f=5 cm). Diaphragm D1 has a distribution of the excited state OH(A) [11–13]. To model this
diameter of 1 mm and is situated at the focal point of lens L2. rotational distribution, the number density of the excited state is
Diaphragm D1 and entrance slit D2 are located at a distance represented by a superposition of two Boltzmann distributions
2f from the lens L3. This configuration guarantees the same characterized by two different temperature parameters as [18–20]
solid angle of collection (∼0.005) for the emission of each NJ ¢ µ (1 - R)(2J ¢ + 1) QrC ⋅ exp ( - FJ ¢ kTrC (A))
segment of the plasma along the detection direction (direction , (4 )
+ R (2J ¢ + 1) QrH ⋅ exp ( - FJ ¢ kTrH (A))
A). As we assume that the detector observes the same amount
of emission from each plasma segment in the self-absorption with FJ′ the rotational energy of rotational level J′ and k the
model (without considering absorption), such an optical Boltzmann constant. Tr C(A), Tr H(A) and Qr C, Qr H are the
configuration or the use of two pinholes is required [16]. temperature parameters and partition functions corresponding to
To perform absorption measurements, a laser stabilized the Boltzmann distribution of the hot (H) and cold (C) groups. R
broadband lamp (Energetiq EQ99) is used as the UV light is the fraction of molecules in the hot group compared to the total
source. The light produced by the lamp is first focused and number of excited molecules. The OH (A–X) emission spectrum
shaped into a close to parallel light bundle through the RF can be written as
reactor by lens L1. After passing the reactor the light is hcv 3
emi
Ical (v ) = ⋅ A vv¢JJ¢ ⋅ NJ ¢ Ä Vins, (5 )
immediately focused onto the entrance slit of the mono- 4p
chromator by the same lens system used for the emission
where h is the Planck constant and c is the speed of light. Avv¢JJ¢
measurement. is the Einstein emission coefficient. The emission spectrum can
The spectra are recorded similarly in direction B by be written as a function with the following parameters:
rotating the reactor 90°. We checked that the absorption in
direction B is negligibly small, and hence the emission spectra
emi
Ical (v) ~ g (TrH (A) , TrC (A) , R , v). (6 )
unaffected by self-absorption can be detected. To increase the
signal-to-noise ratio, only lens L3 was used to focus the
emission on the entrance slit of the spectrometer. A 2 mm 3.1.3. Self-absorbed emission spectrum. Assuming a
wide diaphragm was included at the position of the lens to homogeneous plasma in which the OH molecules in the
avoid stray light going into the monochromator and to ensure excited state A and ground state X co-exist, the effect of self-
a spatial resolution of 2 mm. absorption on the emission intensity can be included in the

3
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

emission intensity as follows [16, 21] allows us to accurately determine the rotational temperature
and the density of ground state OH(X). Because of the very
e (v )
I (l , v ) = ⋅ (1 - exp ( - k (v) ⋅ L )). (7 ) large lifetime of OH(X) compared to the time between colli-
k (v ) sions with He atoms at atmospheric pressure, the ground state
While the formula is identical as in [16, 21], the self-absorbed rotational temperature is a good representation of the gas
emission spectrum, when including non-equilibrium effects, temperature [8]. The obtained values of the gas temperature
is not only a function of the ground state rotational and OH(X) density are used to assess the accuracy of the self-
temperature and OH density, but also a function of the absorption analysis.
temperature parameters and the density fraction R of the Figure 2(b) shows the recorded emission spectrum from
excited state. direction B and its best fit assuming that the excited state is
In summary, the self-absorbed emission spectra depend composed of a hot and a cold group that can be described
on the following parameters: with a superposition of two Boltzmann distributions, each
with a given temperature parameter. The excellent agreement
self
Ical (v) ~ h (TrH (A) , TrC (A) , R , Tr (X ) , n OH , v). (8 ) between measured and simulated spectra proves the validity
of this description. A significant non-equilibrium condition
exists as 67% (R) of the excited state molecules are in the hot
group. This non-equilibrium is expressed in a significant
3.2. Fitting method overpopulation of the high rotational states. The temperature
We perform a least-square fit of the experimental data with parameter of the cold group Tr C(A) is 348±5 K and matches
Matlab, in particular the function fit which is part of the curve well with the ground state rotational temperature Tr(X) from
fitting toolbox. The reported uncertainty of each fitting figure 2(a). Hence, in spite of the non-equilibrium distribu-
parameter is a 95% confidence interval as determined by the tion, the excited rotational levels with low rotational numbers
fitting procedure. The numerical methods used to fit the are thermalized. The two-temperature distribution, as shown
recorded spectra are identical as in [16, 22]. We did not later, is due to two production mechanisms of OH(A), one
perform a fitting of the wavelength as e.g. in [8], however a producing thermalized OH(A). The temperature parameter of
wavelength calibration was performed and required in order the hot group Tr H(A) reaches a value of 14 900±1800 K and
to obtain accurate fitting results. is a remnant of the formation process of OH(A) radicals that
causes high rotational excitation. Tr H(A) shows excellent
agreement with the temperature parameter of the hot group of
4. Results and discussions OH(A) produced by electron dissociative excitation of water
by mono-energetic electrons as reported by Mohlmann et al
First, we discuss in detail the fitting results and accuracy for [23]. This suggests the importance of electron dissociative
the He RF discharge with 6000 ppm water. Next, we present excitation of water in the production of OH(A). The produc-
the dependence of the fitting parameters on water con- tion processes of OH(A) will be discussed in more detail
centration and discuss the derived plasma parameters. The later.
obtained information allows us to discuss the formation of the Figures 2(c)–(e) show the comparison of the fitted
OH(A) state and related plasma kinetics in detail. Finally, we synthetic spectrum with the recorded emission from direction
apply the same method on a filamentary Ar+H2O DBD to A. Three different assumptions are made for the fit.
discuss the limitations of the technique and the underpinning Figures 2(c) and (d) both consider the self-absorption effect,
plasma properties that can cause the breakdown of the self- although with different restrictions on the fitting parameters.
absorption method. The SSEs of figures 2(c) and (d) are almost the same and are
much smaller than the SSE of the fit in figure 2(e) that does
4.1. Fitting results not take self-absorption into account. This confirms the pre-
sence of non-negligible self-absorption in the emission
Figure 2 shows the best fits of the experimental and calculated spectra in direction A. The fitting residue of figure 2(e) shows
emission and absorbance spectra for the He–H2O RF plasma important discrepancies between the fit and the recorded
with 6000 ppm water. The spectra are recorded from two spectrum for a few rotational lines with low rotational num-
different directions as indicated in the figure. Direction A is bers. Nonetheless the effect remains small and can only be
along the plasma axis with a length of 76 mm while direction observed for accurately recorded spectra with sufficiently
B has only a plasma thickness of 3 mm with absorption (and high spectral resolution.
self-absorption) below the detection limit. The corresponding For the fit shown in figure 2(c) no restrictions are
fitting parameters for all cases are tabulated for further com- imposed on any fitting parameters (Tr(X), nOH, Tr H(A), Tr C(A)
parison. The sum of the square error (SSE), an overall num- and R). This means that the spectrum depends on all fitting
erical indicator used to assess the accuracy of the fit, is also parameters for the case of pure emission (figure 2(b)) and the
included in the tables. properties of the ground state OH(X) as well. In spite of these
Figure 2(a) shows the best fit of the recorded broadband two additional degrees of freedom, the parameters of the
absorbance spectrum. The small SSE reflects the high acc- OH(A) state (Tr H(A), Tr C(A) and R) obtained from self-
uracy of the broadband absorption method. This method absorption show excellent agreement with those deduced

4
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

Figure 2. Best fits of the recorded absorption and emission spectra for the RF He–H2O discharge with a water concentration of 6000 ppm.
The corresponding fitting parameters with 95% confidence interval are listed in the separate tables at the right of each spectrum. The residuals
of the fit are also shown. (a) Absorbance spectrum recorded in direction A and its best fit; (b) emission spectrum recorded in direction B
(without self-absorption) and its best fit; (c) emission spectrum detected from direction A and its best fit considering self-absorption.
(d) Emission spectrum detected from direction A and its best fit considering self-absorption without fitting parameters obtained by emission;
(e) emission spectrum detected from direction A and its best fit without considering self-absorption.

from the emission in direction B. In addition, the deduced the emission data in figure 2(b), the same results are found for
rotational temperature and density of the OH(X) state from the the remaining parameters. This strengthens the conclusion
emission with self-absorption correspond within the exper- that it is possible to simultaneously and accurately fit the
imental uncertainty with the values obtained from broadband spectrum considering self-absorption and non-equilibrium
absorption. When the same fitting is performed, although properties of the excited state without a significant increase in
imposing the values of Tr H(A), Tr C(A) and R as obtained from the uncertainty of the fitting parameters.

5
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

Figure 4. Comparison of Tr H(A) and the density fraction R as a

Figure 3. Rotational temperature of the ground state OH(X) and function of water concentration.
temperature parameter of the cold group of OH(A) as a function of
water concentration. The parameters have been obtained by broad- concentration, although seems to decrease with increasing
band absorption, emission (direction B) and emission with self-
absorption (direction A) as indicated. water vapor. The observed change in R is due to a strongly
reducing effective lifetime of OH(A) with increasing water
concentration [19]. The effective lifetime of OH(A) is mainly
The excellent agreements between emission (with and
determined by water vapor [23–26], and reduces from 59 ns at
without self-absorption) and broadband absorption validate
the lowest water concentration to 6 ns at the highest water
the use of self-absorption to simultaneously obtain the gas
concentration. The rotational energy thermalization time is
temperature, OH ground state density and the properties of
however mainly determined by helium and does not change
the non-equilibrium excited OH(A) state. The validity of the
significantly for the H2O concentration range used in this
approach however dependents on the experimental conditions
study. Unfortunately, the thermalization time of OH(A) with
and production mechanism of the OH(A) state. These lim-
He is, to our knowledge, not reported. However, an estima-
itations will be discussed later in this work in the context of
tion can be made by comparing with the Ar thermalization
an Ar DBD plasma.
time and considering that the RET cross sections of OH(A)
with Ar is in good approximation 3 times faster than with
4.2. OH(X) and OH(A) rotational temperatures and density He leading to a tripling of the thermalization time to 63 ns
fraction [27–29].
The similar or smaller value of the effective lifetime
Figure 3 shows an overview of the rotational temperature of compared to the thermalization time does not allow the high
ground state, Tr(X), and the temperature parameter of the cold rotational levels to equilibrate with the gas temperature. The
group of OH(A), Tr C(A), for different water concentrations. thermalization time was determined for Ar for N„13 and
As the rotational temperature of the ground state, Tr(X), is a will increase for increasing number of rotational levels [27].
good representation of the gas temperature, we can conclude Simulation results in [8] show that the thermalization time of
that the gas temperature does not vary with the water con- OH(A) for rotational levels at N=20 is ten times larger than
centration at a fixed plasma power. that for N=7, although in water vapor. Hence, if a strongly
The rotational temperature of the ground state and the non-thermal nascent rotational distribution is produced, it is
temperature parameter of the cold group of the excited state visible in the emission spectrum. For smaller concentrations
obtained by the emission and self-absorbed emission corre- of water, the effective lifetime increases and part of the highly
spond within the experimental precision. Hence, the excited rotational excited molecules will be transferred through col-
state rotational levels with low rotational number are ther- lisions to lower rotational levels. This leads to the reduction
malized even for water concentrations up to 1%. of the density fraction R for smaller water concentrations as
The rotational temperature parameter of the hot group, seen in figure 4.
Tr H(A), and the density fraction of the hot group over the total
is shown as a function of water concentration in figure 4. The
4.3. OH density
results obtained from emission in direction B and self-
absorbed emission from direction A are in excellent agree- The absolute OH densities deduced from the fit of the
ment with each other. recorded self-absorbed emission and broadband absorption
The ratio of the number of molecules in the hot group to spectra are compared in figure 5 for different water con-
the total amount of excited states molecules increases with centrations. The densities obtained from the self-absorption
increasing water concentration. The rotational temperature of method correspond well with those from broadband absorp-
the hot group Tr H(A) does not strongly depend on the water tion with a maximum deviation smaller than a factor 1.3,

6
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

with the same method although considering non-equilibrium

plasma conditions [34]. The reaction rate coefficient of
reaction 1 contains the highest uncertainty of all considered
reactions.
Comparing all the production rates in table 1 shows that
electron excitation of OH(X) (reaction 1) and direct electron
dissociative excitation of H2O (reaction 2) are the two
dominant production mechanisms of OH(A) for the investi-
gated experimental conditions.
While reaction 2 causes strong rotational excitation of the
OH(A) products, the OH(A) radicals produced by reaction 1
obey a Boltzmann distribution with the rotational temperature
equal to the gas temperature. This is because electrons map
the thermalized rotational distribution of the ground state on
the rotational distribution of the excited state. In addition,
the rotational constants of the X and A states are similar
Figure 5. OH(X) density as a function of water concentration
(BX=18.90 cm−1, BA=17.39 cm−1) causing a maximum
obtained by fitting the self-absorbed emission and broadband
absorption spectra. The red curve is a fit of the data with a square deviation in rotational temperature of 8% [8].
root dependence of the water concentration. Considering these two dominant production mechanisms,
a model of the population of the hot and cold groups of the
OH(A) radical can be constructed. Direct electron excitation
except for the lowest water concentration. This condition leads to the population of the cold group
corresponds to a line-integrated density of 1.43×1019 m−2.
This value is smaller than the reported detection limit of the k1
self-absorption method 2×1019 m−2 in [16]. The OH den- e- + OH (X 2P) ⟶ e- + OH (A2 S)cold , (10)
sity increases in first approximation with the square root of
the water concentration as can be seen by the red curve in while the hot group of OH(A) is produced by the electron
figure 5 corresponding to the formula: dissociative excitation of H2O

n OH [m-3] = 5.45 ´ 1020 ´ n H2 O [%] . (9 ) k2

e- + H2 O (X1A1) ⟶ H + e- + OH (A2 S)hot , (11)
This dependence can be obtained by assuming a balance
between production and destruction of OH as in [4]. The where k1 and k2 are the direct excitation and dissociative
difference in the constant factors in equation (9) and [4] is excitation rate coefficients, respectively.
only 20% when considering that the OH density is actually a The balanced equations for the hot and cold group den-
factor 2 higher than reported in [4] as the effect of Λ-doubling
sities can be written as
was not considered [30]. The remaining difference is poten-
tially caused by the different electron temperatures. While the ⎧ dN cold
fit is a good first-order approximation of the OH density ⎪ OH = n e ⋅ n OH ⋅ k1 - (A C + QC ) NOH cold
+ RH ⋅ NOH
hot
=0
⎪ dt
dependence on the water concentration, the dependence is ⎨ .
only obtained with simplified assumptions. One of these ⎪ dNOH
hot
⎪ = n e ⋅ n H2O ⋅ k2 - (AH + QH + RH ) ⋅ NOH = 0
hot
assumptions is a small variation in the electron temperature ⎩ dt
that will only be valid in a limited parameter range. This (12)
limitation could explain why the observed trend starts to
deviate at the highest water concentration. In this equation, NOH cold hot
and NOH are the number densities of
the cold and hot groups of OH(A). ne represents the electron
4.4. Formation of OH(A) density and n H2 O , the water density. A and Q are the Einstein
emission and the collisional quenching coefficient, respec-
Table 1 summarizes several reactions that lead to the for-
mation of OH(A) in non-equilibrium plasmas. The production tively. RH is an effective rate at which the hot group reaches
rates reported in table 1 are estimated based on the reaction thermal equilibrium (and becomes part of the cold group).
rate coefficients and the species densities obtained from a Assuming steady-state and imposing that in good approx-
global model calculation for the same plasma and similar imation AC+QC≈AH+QH the following simplified
conditions [31]. The reported electron temperature for the expression is obtained for the ratio of hot to total OH(A)
studied plasma is between 2 and 3 eV [32]. The rate coeffi- density:
cient of reaction 1 is approximately 6–10 times smaller than
reported in [33]. This paper assumed thermal plasma condi- hot
NOH 1 1
R= = ⋅ . (13)
tions (electron temperature equals gas and vibrational temp- total
NOH R
1 + A +HQ n OH
⋅
k1
+1
erature). The results presented in table 1 have been calculated C C n H2 O k2

7
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

Table 1. Production reactions for OH(A) and the corresponding rate coefficients quoted from the literature.

No. Reaction Rate coefficient k (m3 s−1) Remark References R=k[x][y] (m−3 s−1)
1 e- + OH (X )  OH (A) + e- 3.1 ´ 10-14 - 2.0 ´ 10-13 Te = 2-4 eV [33, 34] 3.1 ´ 10 23 - 2 ´ 10 24
2 e- + H2 O  OH (A) + H + e- 2.3 ´ 10-17 - 2.8 ´ 10-16 Te = 2-4 eV [35] 4.6 ´ 10 23 - 5.6 ´ 10 24
3 e- + H2 O+  OH (A X ) + H 5.1 ´ 10-14Te-1/2 [36] <10 20
4 e- + H3O+  OH (A X ) + 2H 1.05 ´ 10-13Te-1/2 [37] <1019
5 H- + H3O+  OH (A X ) + H + H2 2.3 ´ 10-13 Tg » 300 K [37] ~1017
6 He+2 + H2 O  OH (A) + He + HeH
+
1.3 ´ 10-16 H2O: 0.1–0.3% [38] < 5 ´ 10 21 - 1.5 ´ 10 22

also suggests that the electron temperature does not vary

significantly in the investigated range of water concentrations
otherwise the trend in figure 6 would not be linear.
Note that all plasma parameters needed to calculate the ratio
of k1/k2 from equation (14) can be deduced from fitting the
emission spectra with the inclusion of self-absorption. If reaction
rate coefficients of all involved processes would be accurately
known, an extended model that contains state-to-state rotational
transfer could be used to obtain an estimate of the electron
temperature. This would be easier for plasma conditions at
reduced pressure and/or high water concentrations when the
effect of thermalization of the rotational population distribution
is smaller.

4.5. Limitations of self-absorption

Figure 6. Relationship between the experimental ratio of hot to total
OH(A) concentration (R) and water concentration. To assess the limitations of the self-absorption analysis, we
present the same spectra as shown in figure 2 but recorded for
a filamentary DBD in Ar with 1% water and a power of 8 W
The thermalization time is taken to be 63 ns as before in figure 7. These results include fitting of the broadband
leading to a thermalization rate RH of 1.6×107 s−1. The absorption, pure emission and emission with self-absorption.
spontaneous emission coefficient AC and quenching rate While the broadband absorption method still yields
coefficient are taken from [25, 39], respectively. reliable Tr(X) and nOH results for water containing DBD, the
By substituting equation (9) into equation (13), we obtain SSE is larger in spite the significantly smaller absolute
absorbance. We also experienced that the alignment of the
⎛ RH ⎞ k1 5.45 broadband light source for the Ar+H2O filamentary DBD
R - 1 = ⎜1 + ⎟⋅ ⋅ is more difficult than for the He+H2O diffuse RF dis-
⎝ A C + QC ⎠ k 2 2.69 ´ 10 3
charge. This is caused by a shift of the incident light due to
1 ⎛ RH ⎞ the local change in refracting index by the hot filaments
⋅ + ⎜1 + ⎟. (14)
n H2 O ⎝ A C + QC ⎠ when the plasma is on. This inhomogeneity intrinsically
leads to a breakdown of the assumptions made in the self-
The experimentally obtained R is plotted in figure 6 using absorption model [16].
the linear relation of R−1 as a function of n H2 O-1/2 according The fit of the emission spectrum shown in figure 7(b)
to equation (14). It results in a linear regression with an yields a rotational temperature of the cold group Tr C(A) which
R-squared value of 0.972, as shown in figure 6. The calcu- is lower than the rotational temperature of the ground state, this
lated values of the intercept (that depends on the water con- suggests an issue with the fitting approach. The thermalization
centration because of Qc) are within a factor 2 of the obtained time for N„13 is 21 ns and the effective lifetime of OH(A)
fitted value. The slope of the fit yields k1/k2∼250. The for 1% water is approximately 5 ns [8, 28, 29]. Hence the
minimum ratio from reported rate coefficients is however 3 rotational distribution of the hot group is not strongly impacted
times higher (at Te=4 eV). This seemingly large difference by rotational energy transfer. The temperature parameter of the
is due to the factor 2 uncertainty of the pre-factor and the hot group Tr H(A) is only 6800 K compared to 15 000 K for the
uncertainty in the calculated value of k1 and the required He RF discharge. This value suggests that the hot group of the
extrapolation as only rate coefficients for electron tempera- OH(A) is not dominantly produced by electron dissociative
tures up to 1.5 eV have been reported. excitation of water (as in the He RF discharge), but rather by
The excellent agreement between the experimental and electron H2O+/H3O+ ion dissociative recombination [40] and
modeled dependence of R on the water concentration supports dissociative excitation of H2O by Ar metastable atoms [41].
our analysis of the OH excitation and de-excitation balance. It Plasma conditions that identified these processes as dominant

8
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

Figure 7. Best fits of the experimental absorption and emission spectra from an Ar+H2O DBD. The plasma operates at 8 W and with a
water concentration of 1%. The conditions for the spectra recording and fits are identical as in figure 2.

have reported temperature parameters of OH(A) between 4000 quality fit of the emission spectra and unrealistic value of the
and 4500 K [8]. These mechanisms are also consistent with temperature parameter of the thermalized cold group. These
the findings of metastable induced production of OH(A) issues propagate in the fitting of the recorded self-absorbed
for low pressure UV sources operating in Ar+H2O mix- spectrum shown in figures 7(c)–(e). The large uncertainties in
tures [42]. the obtained OH density and larger SSEs than in the case of
The observations suggest that the excited state of OH(A) the diffuse RF glow discharge underline this issue. It is also
in Ar+H2O DBD is produced by multiple reactions leading illustrative that in this case the quality of the fitting with and
to a complicated nascent rotational population distribution without self-absorption is similar.
that cannot be easily modeled by a superposition of two While the low OH density approaching the detection
Boltzmann distributions; hence, the fitting results in a lower limit of the method contributes to the uncertainty of the OH

9
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

density and gas temperature, the key reasons for the method University of Minnesota. The authors thank F Iza for pointing
to breakdown are: out the unpublished data of Riahi and Teulet.
(1) the inaccurate description of the excited state and the
sensitivity of self-absorption method to the excited state
information;
References
(2) the non-uniform distribution of the excited state OH(A)
and ground state OH(X) due to the filamentary
inhomogeneous discharge. [1] Dilecce G, Ambrico P F, Simek M and De Benedictis S 2012
Chem. Phys. 398 142–7
Although the latter limitation is straightforward to assess, [2] Ono R and Oda T 2001 IEEE Trans. Ind. Appl. 37 709–14
the former limitation requires knowledge of the plasma [3] Wang C, Srivastava N, Scherrer S, Jang P, Dibble T S and
properties or at least a good ‘guess’ of the production Duan Y 2009 Plasma Sources Sci. Technol. 18 025030
[4] Bruggeman P, Cunge G and Sadeghi N 2012 Plasma Sources
mechanisms of OH(A) before the technique can be used to Sci. Technol. 21 035019
determine accurate OH densities and gas temperatures. [5] Dilecce G, Ambrico P F, Simek M and De Benedictis S 2012
J. Phys. D: Appl. Phys. 45 125203
[6] Xiong Q, Yang Z and Bruggeman P 2015 J. Phys. D: Appl.
Phys. 48 424008
5. Conclusion [7] Luque J, Klein-Douwel R, Jeffries J B, Smith P and
Crosley D R 2002 Appl. Phys. B 75 779–90
[8] Bruggeman P, Sadeghi N, Schram D C and Linss V 2014
The fitting of the emission spectrum considering the effect of Plasma Sources Sci. Technol. 23 023001
self-absorption was used to simultaneously determine the abso- [9] Moon S Y, Kim D B, Gweon B and Choe W 2008 Phys.
lute OH density and gas temperature in a water containing He Plasmas 15 103504
RF glow discharge. A superposition of two Boltzmann dis- [10] Laux C O, Spence T G, Kruger C H and Zare R N 2003
Plasma Sources Sci. Technol. 12 125–38
tributions was shown to be a good description of the excited [11] Bruggeman P, Schram D, González M Á, Rego R,
state OH(A). Information of the excited state (Tr H(A), Tr C(A), R) Kong M G and Leys C 2009 Plasma Sources Sci. Technol.
and ground state (Tr(X), nOH) can be simultaneously determined. 18 025017
Excellent agreement of the results from self-absorption and [12] Bruggeman P, Verreycken T, González M Á, Walsh J L,
broadband absorption validates that the method can be used Kong M G, Leys C and Schram D 2010 J. Phys. D: Appl.
Phys. 43 124005
even in the case of non-thermalized rotational distributions of the
[13] Bruggeman P, Schram D C, Kong M G and Leys C 2009
excited OH(A) state. Furthermore, based on the information of Plasma Process. Polym. 6 751–62
the excited state OH(A) from the self-absorption fitting, the [14] Benstaali B, Boubert P, Cheron B G, Addou A and Brisset J L
OH(A) state is shown to be mainly produced by the electron 2002 Plasma Chem. Plasma Process. 22 553–71
excitation of OH(X) and the direct electron dissociative excita- [15] Machala Z, Janda M, Hensel K, Jedlovský I, Leštinská L,
Foltin V, Martišovitš V and Morvová M 2007 J. Mol.
tion of H2O for the He+H2O RF discharge. A model of the
Spectrosc. 243 194–201
steady-state distribution of the hot and cold group densities of [16] Du Y, Peng Z, Ding Y, Sadeghi N and Bruggeman P 2016
OH(A) is able to describe the dependence of the rotational Plasma Sources Sci. Technol. 25 04LT02
population distribution as a function of the water concentration. [17] Du Y, Nayak G, Oinuma G, Peng Z and Bruggeman P 2017
This method has also been applied to a filamentary J. Phys. D: Appl. Phys. 50 145201
[18] Wang Q, Doll F, Donnelly V M, Economou D J,
Ar+H2O DBD. While the inhomogeneous nature of the
Sadeghi N and Franz G F 2007 J. Phys. D: Appl. Phys. 40
DBD leads to complications, a more complex production of 4202–11
the OH(A) by multiple production mechanisms leading to a [19] Bruggeman P, Iza F, Guns P, Lauwers D, Kong M G,
complex nascent rotational population distribution causes the Gonzalvo Y A, Leys C and Schram D 2010 Plasma Sources
fitting procedure of the emission spectrum to break down. Sci. Technol. 19 015016
[20] Brennen W and Carrington T 1967 J. Chem. Phys. 46
Consequently, the self-absorption method also does not work.
7–18
To ensure that accurate gas temperatures and OH densities are [21] Du Y, Ding Y, Liu Y, Lan L and Peng Z 2014 Appl. Opt. 53
obtained by the method, one requires not only a homogeneous 4922–9
discharge but also knowledge of the plasma properties or at [22] van Gessel A F H, Hrycak B, Jasiński M, Mizeraczyk J,
least a good ‘guess’ of the production mechanisms of OH(A). van der Mullen J J A M and Bruggeman P J 2012 J. Instrum.
7 C02054
[23] Mohlmann G R, Beenakker C and Deheer F J 1976 Chem.
Phys. 13 375–85
[24] Copeland R A and Crosley D R 1984 Chem. Phys. Lett. 107
Acknowledgments 295–300
[25] Copeland R A, Dyer M J and Crosley D R 1985 J. Chem. Phys.
This work was in part funded by the Department of Energy 82 4022–32
Plasma Science Center through the US Department of Energy, [26] German K R 1975 J. Chem. Phys. 62 2584–7
[27] Kienle R, Lee M P and Kohse- Höinghaus K 1996 Appl. Phys.
Office of Fusion Energy Sciences, contract: DE-SC0001939 and B 63 403–18
the University of Minnesota. YD acknowledges the Chinese [28] Jörg A, Meier U and Kohse-Höinghaus K 1990 J. Chem. Phys.
Scholarship Council that supported her research stay at the 93 6453–62

10
Plasma Sources Sci. Technol. 26 (2017) 095007 Y Du et al

[29] Kliner D A V and Farrow R L 1999 J. Chem. Phys. 110 [36] Rowe B R, Vallee F, Queffelec J L, Gomet J C and Morlais M
412–22 1988 J. Chem. Phys. 88 845–50
[30] Verreycken T, Sadeghi N and Bruggeman P J 2014 Plasma [37] Millar T J, Farquhar P R A and Willacy K 1997 Astron.
Sources Sci. Technol. 23 045005 Astrophys. Suppl. Ser. 121 139–85
[31] Liu D X, Bruggeman P, Iza F, Rong M Z and Kong M G 2010 [38] Binns W R and Ahl J L 1978 J. Chem. Phys. 68
Plasma Sources Sci. Technol. 19 025018 538–46
[32] Vasko C A, Liu D X, van Veldhuizen E M, Iza F and [39] Dimpfl W L and Kinsey J L 1979 J. Quant. Spectrosc. Radiat.
Bruggeman P J 2014 Plasma Chem. Plasma Process. 34 Transfer 21 233–41
1081–99 [40] Sonnenfroh D M, Caledonia G E and Lurie J 1993 J. Chem.
[33] Riahi R, Teulet P, Ben Lakhdar Z and Gleizes A 2006 Eur. Phys. 98 2872–81
Phys. J. D 40 223–30 [41] Gardner J A, Dressler R A, Salter R H and Murad E 1992
[34] Riahi R and Teulet P Unpublished data J. Chem. Phys. 97 2473–80
[35] Itikawa Y and Mason N 2005 J. Phys. Chem. Ref. Data 35 [42] Artamonova E et al 2008 J. Phys. D: Appl. Phys. 41
1–22 155206

11