You are on page 1of 17

pubs.acs.

org/JPCC Article

Time Evolution of the Dissociation Fraction in rf CO2 Plasmas:


Impact and Nature of Back-Reaction Mechanisms
Ana Sofia Morillo-Candas,* Vasco Guerra, and Olivier Guaitella

Cite This: J. Phys. Chem. C 2020, 124, 17459−17475 Read Online

ACCESS Metrics & More Article Recommendations


See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: The time evolution of the dissociation fraction in a


pulsed radio frequency (rf) CO2 discharge is studied by infrared
absorption. A large parametric study performed in a closed reactor
Downloaded via UTRECHT UNIV on October 8, 2020 at 06:57:38 (UTC).

brings valuable information about both dissociation and recombi-


nation processes. The CO2 conversion shows a time evolution
initially controlled by electron impact dissociation. For longer
plasma-on times, the dissociation fraction reaches a steady-state
that corresponds to a balance between dissociation processes and
back-reaction mechanisms. The characteristic times of vibrational
and rotational excitation of CO and CO2 are measured during a
single plasma pulse. The dependence of the CO2 conversion as a function of pulse duration and frequency is then analyzed, showing
the influence of the plasma excitation conditions on the back-reaction mechanisms. The study of CO2−O2 and CO−O2 gas mixtures
gives further insight into the impact of the oxygen content in these mechanisms. The global back-reaction rate observed in these
experiments is compared with calculated values from rate coefficients available in the literature. The experimental results and
preliminary calculations reveal a key role of molecular oxygen and of the metastable electronically excited state CO(a3Πr) in the
back-reaction. Competing processes involving vibrational excited CO are not dominant in our discharge conditions but may become
relevant at slightly higher vibrational temperatures.

1. INTRODUCTION and continuous gas flow may be interesting in order to develop


Low-temperature CO2 plasmas have been studied because of industrial solutions, but it does not facilitate the understanding
their importance in the field of CO2 lasers1 and, more recently, of the fundamental mechanisms involved.19,20 At high
because of the numerous applications related to surface pressures, the short characteristic times of the evolution of
treatment,2−5 study of spacecraft entry in planetary atmos- main neutral gas phase species make time-resolved measure-
pheres,6−8 oxygen extraction from the atmosphere of Mars,9,10 ments challenging, and therefore, these are rarely performed.21
and particularly for CO2 recycling purposes. The current At lower pressures, the amount of CO created in a single
context of global warming and increase of anthropogenic plasma pulse in the range of a few milliseconds or nanoseconds
greenhouse gas emissions,11 along with the necessity of finding is small; therefore, the CO fraction during a short pulse
new sources of fuels and chemicals, has put the focus on the remains almost constant.22,23 To detect significant amounts of
development of efficient CO2 recycling technologies. CO2 dissociation products it is necessary either to use long plasma
could be used as a raw material from which to generate pulses or to accumulate several plasma pulses. However, in that
more complex organic compounds, such as energy-dense case, in situ time-resolved measurements would be influenced
hydrocarbon fuels. The first step in this process is to dissociate by the renewal of the gas in the reactor due to the continuous
CO2 to CO. In this context, CO2 plasmas present several gas flow and the main gas phase species would be in pseudo
advantages compared with traditional dissociation methods,12 steady-state conditions.
in particular with regard to the energy efficiency.13 Another The lack of knowledge with regard to some fundamental
major advantage is the possibility to be integrated into the mechanisms brings difficulty to the formulation of the
electrical network and to adapt to the availability of
intermittent renewable energy sources.14,15
Many types of discharges in different experimental Received: April 15, 2020
configurations have been studied trying to increase both the Revised: July 11, 2020
energy efficiency and the conversion.16,17 In spite of the Published: July 13, 2020
extensive literature, several basic processes essential for
describing the CO2 plasma kinetics are not yet well
understood.18 Working with complex reactor configurations

© 2020 American Chemical Society https://dx.doi.org/10.1021/acs.jpcc.0c03354


17459 J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 1. Schematic representation of the experimental setup used for the experiments. SG: Signal generator.

computational models, resulting in discrepancies between was used to establish experimentally the rate coefficient for the
experimental data and model predictions, and makes it difficult electron impact dissociation of CO2 in a large range of reduced
to conceive more efficient conversion systems. Among these electric fields, and to validate the corresponding cross
processes still surrounded by large uncertainties, one of the section,25 illustrating the interest of this approach for
most important is the so-called “back-reaction”, i.e., any fundamental studies.
reaction mechanism involving the oxidation of CO back to This work focuses on the investigation of the back-reaction
CO2. Evidently, back-reaction mechanisms must be prevented processes and their time scales and discusses the main possible
for an efficient CO2 conversion process. underlying mechanisms. Knowledge of the characteristic times
The reactivity of ground-state species is generally well- of the gas heating and vibrational excitation is essential for the
known and the rate coefficients for most of the reactions, analysis of back-reaction mechanisms. The pulsed capacitively-
usually temperature-dependent, are for instance available in the coupled radio frequency (rf) discharge studied in this work was
NIST database. However, under plasma conditions, atoms and ignited under a low-power and low-pressure regime which, in
molecules in vibrationally and/or electronically excited states addition to its relevance for some applications, has the
may have a significant impact on the gas kinetics due to their advantage of slowing down characteristic times and eases the
fast reaction rates or large amount of stored energy. The monitoring of the time evolution of various parameters. The
required vibrational or electronically state-dependent rate time evolution of the rotational and vibrational temperatures of
coefficients of chemical reactions are rarely known. The CO2 and CO can therefore be monitored during a plasma
information available about rate coefficients involving elec- pulse by time-resolved step-scan FTIR spectroscopy, following
tronically excited states is scarce, and their role is often the experimental approach described in refs 22 and 26. In
neglected. The vibrational excitation of CO and CO2 is order to estimate the contribution of different mechanisms to
frequently assumed to lower the activation energy for several the overall back-reaction, the electron densities in the
reactions. This effect may be taken into account by subtracting discharge were determined by hairpin probe experiments.27,28
a certain contribution to the activation energy in an Arrhenius- A careful experimental procedure was developed to obtain
type reaction rate equation.13,24 The value of that contribution reproducible results, described in section 2 along with the
must be validated. Therefore, more work is needed to identify setup details. The experimental results in CO2 plasmas are
the most relevant mechanisms, to validate rate coefficients, and presented in section 3. We follow a step-by-step approach: (1)
to verify commonly used assumptions. description of the general characteristics of the time evolution
In order to investigate fundamental mechanisms in CO2 of the dissociation fraction comparing pulsed and continuous
plasmas we have designed an experiment, so-called “building- plasma; (2) study of the effect of pressure and power; (3)
up”, performed under static conditions (closed reactor, without characterization of the time evolution of the vibrational and
gas flow), which allows us to follow the time evolution of a rotational temperatures during a plasma pulse; and (4)
fixed number of CO2 molecules, their dissociation into CO, parametric study of the effect of pulse duration and pulse
and their recombination in back-reactions. In this experiment, frequency. In order to study the effect of oxygen in the
consecutive trains of plasma pulses are ignited, while the dissociation fraction and on the back-reaction mechanisms,
densities of CO and CO2 are monitored over time by infrared experimental results in CO2−O2 and CO−O2 mixtures are
absorption, until the dissociation fraction reaches the steady- presented in section 4. The main back-reaction mechanisms in
state value. This experiment has several advantages: (1) our experimental conditions are discussed in section 5, before
Working in static conditions, without renewing the gas, allows finishing with the conclusions of this work and perspectives in
us to remove the possible influence of the residence time. (2) section 6.
Playing with the pulse duration and frequency allows us to
investigate the influence of processes with different character-
istic times during the active phase and in the postdischarge. (3) 2. EXPERIMENTAL DETAILS
The cumulative effect of multiple pulses in a closed system 2.1. Building-Up Experimental Setup. The experimental
reveals the impact of subtle phenomena on the final setup used for the building-up experiments is shown
dissociation fraction. In a previous work, a similar experimental schematically in Figure 1. The rf discharge was generated in
configuration with a pulsed direct current (dc) glow discharge a cylindrical reactor made of Pyrex (2 cm inner diameter and
17460 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

23 cm length), positioned in the sample compartment of a


FTIR spectrometer (Bruker, V70). A buffer volume including a
bypass valve allows a fine adjustment of the pressure in the
reactor and is connected to the reactor through two Teflon
valves. The pressure was varied between 0.5 and 5 Torr. After
setting the pressure in the buffer + reactor volume and before
starting the experiment, these valves are closed, isolating the
gas contained in the plasma reactor from the rest of the gas
line. The gas flow to fill the reactor was set by Bronkhorst (F-
201CV) mass flow controllers using gas from bottles of CO2,
CO, and O2 (Air Liquide, Alphagaz for CO2 (>99.998%),
Alphagaz 2 for O 2 (>99.9995%), and N47 for CO
(>99.997%)). The pressure was measured with a capacitance
manometer (Pfeiffer, CMR263) placed in the upstream part of
the bypass valve in the buffer volume. The gas was evacuated
by a scroll pump (Edwards, XDS35).
The plasma reactor is powered by a rf generator (Advanced
Energy, RFX 600) through a match box (Solayl SAS) at 13.6
MHz. The transmitted and reflected power were measured
with a rf probe (Vigilant Power Monitor, Solayl SAS) placed
between the rf generator and the match box. The rf power is
transmitted to the plasma by a three-electrode system
consisting in three copper sheets rolled around the Pyrex
tube, without direct contact between the metal electrodes and
the plasma. Two grounded electrodes are located in the
extremes of the reactor (15 cm apart) and one high power
electrode is placed in the middle. This configuration allows for Figure 2. (a) Experimental protocol and (b) detail of the FTIR
an adequate calibration of the power dissipated in the plasma measurement.
via the subtraction method.29 The current was measured by a
current probe (Pearson, 2868) placed around the high-power
electrode. The imposed power was varied typically between 40 steady-state dissociation value, probably due to slow chemical
and 80 W, which correspond to real powers dissipated in the reactions at the surface, leading to non-reproducibility issues.
discharge ∼27.7 and 44.2 W. For simplicity, we keep the The choices made for the two steps of the pretreatment can be
notation of the imposed power in the remainder of the text. debatable and have consequences that are briefly discussed in
The error is estimated to be <8%, based on the reproducibility section 4.2, but allow us to start always from the same surface
of the power measured for different pulse duration, pressure, or conditions.
gas mixture conditions and from pulse to pulse within the same After the pretreatment, the reactor is pumped and filled with
measurement. Signal generators (TTi, TGP110) were used to the proper gas mixture at the targeted pressure and closed. The
produce square pulses with rise/fall times in the order of a few measurement proceeds following the method defined to
microseconds and to trigger the FTIR or the rf generator, control the FTIR working in rapid-scan mode (see Figure
depending on the experiment. The trigger signals and the rf 2a). Before any plasma, an infrared absorption spectrum is
current were monitored with an oscilloscope (LeCroy, acquired. Subsequently, the FTIR spectroscope triggers a train
Waverunner LT584M). of plasma pulses. The characteristics of the train are defined by
2.2. Measurement Protocol for Building-Up Experi- two signal generators (SG): SG1, triggered by the FTIR,
ments. The experimental procedure is schematically pre- defines the number of pulses within a train by adjusting the
sented in Figure 2. In order to obtain reproducible results in gate duration; SG2, gated by SG1, controls the pulse duration
static conditions, two preliminary steps were followed before and the delay between pulses within a train and triggers the rf
the building-up measurement: (1) Adjustment of the rf power: generator. After each train of plasma pulses, an IR absorption
The optimal functioning of the match box varies with the measurement with the plasma OFF (gas in thermal
pressure and the gas mixture. The reactor is filled and closed at equilibrium) is acquired with a mirror scan speed of 80 kHz
the measurement conditions. Under continuous plasma at the (see Figure 2b). This procedure is repeated for 500 trains of
targeted power, the matching is adjusted by tuning the load plasma pulses until the total accumulated plasma ON time TON
and tune capacitors of the match box until the reflected power is around 25 s, long enough to reach the steady-state
measured by the rf probe is zero. After the adjustment step, the equilibrium for all the conditions studied.
reactor is pumped and the match box settings remained The configuration of the series of trains of plasma pulses,
untouched. (2) Pretreatment of the reactor wall: This involves hereinafter called “series configuration”, is defined by a set of
two steps: 15 min of O2 continuous plasma at 1.5 Torr (where parameters detailed in Table 1. The reference series
a maximum of the O atom density in similar experimental configuration is Ntr = 500 trains × Np = 10 pulses of tON p =
conditions was observed30), 40 W, in flowing conditions (7.4 5 ms plasma ON and tOFF p = 10 ms plasma OFF per train,
sccm gas flow), followed by 10 min of CO2 gas flow (no which corresponds to a total plasma ON time per train of tON tr =
plasma) at similar pressure and flow. In the range of pressures ∑ tON
p = 50 ms and a total plasma ON per experiment of TON
studied and working in static conditions, the initial surface = ∑ ttr = 25 s. Other series configurations are studied and
ON

state of the reactor walls was found to influence the final detailed in the following sections. The “measurement” time per
17461 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Table 1. Parameters Defining the Series of Trains of Plasma where CO and CO2 denote the gas phase concentrations of the
Pulses two molecules. The algorithm assumes the relation CO2 →
CO + 1/2O2, i.e., for every two CO molecules one O2 is
symbol definition
produced. The dissociation of CO and presence of free carbon
Np number of pulses per train of plasma pulses atoms are neglected, justified by the high energy threshold for
Ntr total number of trains of plasma pulses the electron impact dissociation of CO, above 11 eV31 and the
tON
p pulse duration (plasma ON) fast oxidation rates of carbon atoms.32 The spectra measured in
tON
tr total plasma ON time per train of plasma pulses the postdischarge of a train of plasma pulses in the building-up
tr = ∑ tp = Np·tp
tON ON ON

tOFF delay between consecutive plasma pulses (plasma OFF)


experiment are fitted with a similar script, but that considers
p
tOFF total plasma OFF time per train of plasma pulses
only gas mixtures in thermal equilibrium, since the character-
tr
tOFF
tr = ∑ tOFF
p = Np·tOFF
p
istic time of the relaxation of the vibrational excitation is
TON total plasma ON time in the experiment TON = ∑ tON
tr = Ntr·ttr
ON significantly shorter than the time scales attainable with the
rapid-scan mode of the FTIR. The determination of the
molecular densities of the different species in thermal
equilibrium is more accurate, as there are less fitting
train of plasma pulses is around 2.08 s, which includes the time parameters. In this case, the number densities of CO and
the FTIR waits for the ignition of the train of plasma pulses CO2 are fitted independently. The precision on the CO2 and
(500 ms), the “settings” time of the FTIR and the time CO densities is in the order of 1019 m−3, to be compared with
required by the mirror of the FTIR to scan the whole absolute densities in the order of 1022−1023 m−3. For
wavelength range, ∼1 s at 80 kHz. Therefore, the OFF time in consistency, the same definition of the dissociation fraction,
between consecutive trains is roughly 1.5 s. The total α, is kept. The sum of the fitted CO and CO2 densities is
experiment time for the 500 trains of pulses is around 17 constant as a function of TON (normalized standard deviation
min for each plasma condition. The influence of the equal to 0.67%, with no visible trend), fulfilling the carbon
measurement time and the choice of the series configuration balance and confirming that possible carbon deposition, not
are briefly discussed in section 3.1. observed by eye, is indeed negligible. The fitted gas
2.3. Measurements under nonequilibrium condi- temperature, measured during the postdischarge, remains
tions. Two additional types of measurements were performed close to room temperature. The script accounts for the
in the active plasma phase, under nonequilibrium conditions: possibility of gas mixtures with gases that are not infrared
(1) Continuous plasma ON: Under static conditions, the active, such as O2, by introducing a dilution factor. The
plasma is switched ON after the initial trigger signal, and it is increase of pressure due to leaks, unavoidable in any vacuum
kept ON while consecutive absorption measurements are taken system is also included in the script. The leak, verified to be
until the end of the experiment (25 s), without switching OFF ∼0.5 Torr/hour is mostly from the N2 purge chamber around
the discharge. The scan speed of the IR mirror was set at the the IR beam and the reactor windows, set to avoid absorption
fastest speed possible 160 kHz. In these conditions the scan of the IR beam by CO2 from the room.
time is ∼0.5 s and the measurement time is around 1 s. The The oxidation of the Teflon valves closing the plasma
total experiment time is significantly shorter than in the reactor or the Teflon rings used for the mounting of the CaF2
reference experiment, ∼25 s. A similar measurement with the IR windows can be a source of CO2 or CO molecules in static
IR source of the FTIR OFF is required to subtract the conditions. To quantify this pollution and compare it to the
contribution of the spectrally resolved emission from the measured CO2 and CO densities, a full building-up measure-
plasma, as described in refs 22 and 26. (2) Step-scan time- ment was done in pure O2 at 1.5 Torr and 40 W, to maximize
resolved measurements during a plasma pulse: These measure- the possible oxidation of the Teflon keys and rings. This test
ments allow the study of the time evolution of the vibrational showed that the sum of CO and CO2 molecules created after
temperatures of CO2 and CO during a single plasma pulse. the O2 plasma represented only 0.47% of the total gas density.
The FTIR is operated in step-scan mode, and the measure- Hence, the contribution of possible pollution can be
ments are done with a continuous gas flow of 7.4 sccm since considered negligible.
they require repetitive conditions during the experiment (∼2.5 2.5. Hairpin Probe Measurements. A floating induc-
h). The experimental details are thoroughly described in refs tively coupled hairpin probe (tungsten wire of 0.1 mm radius,
22 and 26. The purpose of these measurements is to check that tines ∼25 mm long, f 0 ∼2.903 GHz, separated by 1.8 mm) was
the vibrational kinetics, particularly their temporal evolution, in attached to a glass tube containing a rigid coaxial cable
the pulsed rf discharge are similar to those measured and terminated in a loop to couple the microwaves. The probe was
analyzed in detail for the case of a CO2 pulsed dc glow situated at, and aligned with, the axis of the discharge tube and
discharge.22,26 could be translated longitudinally. A microwave sweep
2.4. Data Treatment. The infrared spectra measured oscillator (HP, 8350B) provides tunable microwave signal,
under nonequilibrium conditions are fitted with a Matlab script repetitively swept over a small frequency range. The reflected
described in detail in ref 22. The fitting provides the vibrational microwave amplitude was measured using a Shottky diode,
temperatures of CO2 corresponding to the bending and allowing the reflectance spectrum to be recorded with an
symmetric vibrational modes, T1,2, asymmetric stretch mode, oscilloscope (LeCroy, Waverunner LT584M). The electron
T3, and the vibrational temperature of CO, TCO, in addition to densities under plasma conditions were calculated from the
the rotational temperature, Trot, the pressure and the shift in frequency of the resonance peak, using the iterative
dissociation fraction, represented by the parameter α, method described by Piejak et al.33 Corrections for the sheath
[CO] and for electron collisions were made using the formulas
α= proposed by Sands et al.28 and the step-front sheath model
[CO] + [CO2 ] (1) from ref 33 to estimate the sheath width. The electron-neutral
17462 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

collision frequency was calculated with the method proposed trains. The pulse duration (tONp = 5 ms) and delay (tp
OFF
= 10
by Peterson et al.34 The collision frequencies derived from the ON
ms), and the total plasma ON time (T = 25 s) were kept
fitting of the Lorentzian profile of the resonance peak are in fair constant for all the pulsed plasma experiments. The parameters
agreement with tabulated values for different gases given in ref defining the pulsed experiments are summarized in Table 2.
35. The general trend for all the experimental conditions is
similar. It presents an initial fast increase followed by a
3. CO2 PLASMA EXPERIMENTS saturation toward longer total plasma ON times. We can
This section presents the experimental results obtained when distinguish three time “regions” schematically pointed out in
the experiments are initiated in pure CO2. We first compare Figure 3: an “initial slope” for short plasma ON times, until
the time evolution of the dissociation fraction in continuous TON ∼ 0.3−1 s, where α follows a roughly linear evolution with
and in pulsed plasma, followed by the study of the effect of TON; a “turning region”, showing an exponential-like variation,
pressure and power. Last, the pulse frequency and duration followed by a transition to the next region; and the final steady-
within a train of plasma pulses are varied to assess the effect of state equilibrium (“plateau”), where the dissociation fraction
the vibrational excitation. remains constant as a function of TON.
3.1. Pulsed versus Continuous Plasma. Figure 3 plots The figure shows that the dissociation curves saturate in all
the time evolution of the dissociation fraction, α, as a function the experiments. If only dissociation mechanisms were playing
a role in the time evolution of the gas mixture, then we would
expect α to keep increasing up to 100%. The increase would be
slower as TON increases, simply due to the energy spent in the
dissociation or excitation of the accumulated dissociation
products, such as CO, O2, O atoms and other minor species,
instead of with CO2. However, a steady-state (“plateau”) is
reached. The chemical composition does not evolve any more
even though the gas continues receiving plasma pulses.
Dissociation and recombination processes are thus balanced,
demonstrating that back-reaction mechanisms must be taking
place during the ON phase.
Two other points are worth noting. First, the initial slope
remains the same for all the experiments (the different traces
diverge only after TON ∼2 s), and second, the steady-state
dissociation fraction in continuous plasma is larger than that in
any of the pulsed plasma experiments. This effect suggests the
occurrence of back-reactions also during the OFF phase,
between plasma pulses. Introducing a delay between trains,
equivalent to enlarging the measurement time of the FTIR,
does not affect the time evolution of α. The reference
Figure 3. Time evolution of the dissociation fraction α as a function
condition, which requires 2.08 s per measurement and a total
of the total plasma ON time, TON (plotted only up to TON = 15 s),
measured at 2 Torr and 40 W, for the continuous plasma ON experiment time of 17.34 min (blue curve in Figure 3),
experiment and for the pulsed building-up experiments described in perfectly overlaps with two experiments where the FTIR was
Table 2. The notation in the legend for the pulsed experiments forced to wait an extra 2 or 4 seconds, increasing the total
represents Ntr (number of trains) × Np (number of pulses) × tON p − experiment time to 34.67 and 52.01 min, respectively (red and
tOFF
p (ms ON−OFF per pulse). yellow curves). Therefore, long OFF times with the gas in
thermal equilibrium do not have any impact on the
of the total plasma ON time, TON, for continuous and pulsed dissociation fraction. On the contrary, varying the series
plasma with the reference configuration (Ntr = 500 trains × Np configuration (Ntr, Np, and tONtr ) does have a small effect in the
= 10 pulses of tONp − tp
OFF
= 5−10 ms ON−OFF per train, tON tr final dissociation value. Lower number of trains Ntr (higher
= 50 ms) measured at 2 Torr, 40 W. Other pulsed plasma Np) leads to a slightly higher dissociation fraction. This effect
measurements exemplify the effect of varying the number of could suggest a certain CO destruction after the end of each
trains, Ntr, the plasma ON per train, tON
tr , and the delay between train, which should happen at a shorter time scale. It could be

Table 2. Pulse Parameters Used in the Experiments Presented in Figure 3


notation Ntr Np tON
p (ms) tOFF
p (ms) tON
tr (ms) tOFF
tr (ms) TON (s) meas (s)a exp (min)b
Varying Ntr and tON
tr
100 × 50 × 5−10 100 50 5 10 250 500 25 2.08 3.47
250 × 20 × 5−10 250 20 5 10 100 200 25 2.08 8.67
500 × 10 × 5−10c 500 10 5 10 50 100 25 2.08 17.34
1000 × 5 × 5−10 1000 5 5 10 25 50 25 2.08 34.67
Varying Delay between Trains
500 × 10 × 5−10 + 2 s 500 10 5 10 50 100 25 4.161 34.67
500 × 10 × 5−10 + 4 s 500 10 5 10 50 100 25 6.241 52.01

a
Measurement time per train. bTotal experiment time. cReference series configuration.

17463 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 4. Time evolution of the dissociation fraction α as a function of the total plasma ON time TON for (a) different initial pressures from 0.5 to 5
Torr for the same imposed rf power, 40 W, and (b) for different rf 40, 60, and 80 W, corresponding to real transmitted power of ∼27.7, 36.3, and
44.2 W respectively, for two pressures, 2 and 5 Torr.

caused for instance by long-lived species such as oxygen atoms controlled by the electron kinetics through electron impact
or metastable states, or slow chemical reactions on the surface. dissociation, which in any case should be one of the main
Pulse frequency and duration are varied in section 3.3 to dissociation mechanisms. The possible contribution of the
provide insight into this question. For the building-up vibrational excitation of CO2 molecules to the measured
experiments in the following sections, we keep constant the dissociation fraction is studied in the next section.
number of trains Ntr = 500, the total plasma ON per train 3.3. Pulse Duration (tON p ) and Delay between Pulses
tON
tr =50 ms, and the total plasma ON time T
ON
= 25 s, varying (tp ) Variation. 3.3.1. Characterization of a Plasma Pulse.
OFF

only the series configuration within a train of pulses, i.e., the The time evolution of Trot, T1,2, T3, and TCO during a plasma
number of pulses Np, pulse duration tONp , delay between pulses pulse of 5−10 ms ON−OFF (the reference pulse) is studied in
tOFF
p , and the ratio between them tON OFF
p /tp . the same discharge but working in continuous gas flow to
3.2. Pressure and Power Variation. Figure 4 presents the provide the repetitive conditions required by the step-scan
temporal evolution of the dissociation fraction, α, for the FTIR measurements. The obtained temperatures for 2 and 5
reference pulse configuration (Ntr = 500 trains × Np = 10 Torr at 40 W are presented in Figure 5. The right axis plots the
pulses of tONp − tOFF
p = 5−10 ms ON−OFF per train) for vibrational excitation of T3, [T3 − Trot]. The vertical bars show
different pressures and powers. Figure 4a shows the effect of some of the values of pulse duration tON p (in red) or delay
the initial pressure for the same imposed rf power, 40 W, between pulses tOFFp (in blue) that are investigated in the
whereas Figure 4b shows the effect of varying the rf power for following section. The dissociation fractions in both conditions
two pressures, 2 and 5 Torr. Pressure and power have a clear
impact on the time evolution of the dissociation fraction, both
in the kinetic evolution at the beginning of the experiment and
in the final dissociation value. In general, lower pressures and
higher powers lead to a faster evolution of the gas mixture: The
initial slope is steeper, and the gas composition reaches the
final steady-state for shorter TON. At higher pressures a
significantly slower evolution of the gas mixture is observed. As
an example, at 0.5 Torr, the lowest pressure studied, the gas
reaches the chemical equilibrium after TON ∼2.5 s, while at 5
Torr it requires up to TON ∼25 s (not shown) to reach the
steady-state.
The final steady-state value of α decreases significantly as a
function of pressure, more than a factor of 2 between 0.5 and 5
Torr, and increases with power. These measurements taken in
an rf discharge are qualitatively in good agreement with the
trends observed in a glow discharge measured in similar
experimental configuration.25,36 The dependence of α as a
function of pressure can be partly explained by an expected
decrease of the reduced electric field and by the decrease of the
specific energy input per molecule (SEI). The total gas density
increases with pressure, and consequently, for similar power Figure 5. Time evolution of the rotational temperature Trot and the
vibrational temperatures of CO2 and CO, T1,2, T3, and TCO, during a
and plasma ON time, the SEI decreases. The relation between
plasma pulse of 5−10 ms ON−OFF for 2 Torr (panel (a)), and 5
the dissociation fraction and the power is basically linear with a Torr (panel (b)), 40 W. The right axis gives the scale for the
similar effect for both pressures studied, as expected from the vibrational excitation of T3, [T3 − Trot], represented by a dotted gray
increase of electron density as a function of power.36 line in both panels. Vertical lines indicate relevant time points for the
The variation of the dissociation fraction with pressure and pulse duration (red lines) and delay between pulses (blue lines)
power suggests that the dissociation process is strongly variation experiments.

17464 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 6. Time evolution of the dissociation fraction α as a function of the total plasma ON time TON in linear time scale (graphs (a) and (c)) and
logarithmic time scale (graphs (b) and (d)) measured at 2 Torr (top panel) and 5 Torr (bottom panel), 40 W. Graphs (a) and (b) present data for
different pulse duration (tON OFF
p ) but same delay between pulses (tp ). The number of pulses Np is varied to keep always the same plasma ON per
train (tON
tr ). Panels (c) and (d) present data for different delay times between pulses (tOFF ON
p ) but the same pulse duration (tp ). The notation in the
legend represents Np × tp − tp .
ON OFF

are approximately constant during the plasma pulse; α = 0.1 T1,2 increase during the pulse and are higher at 5 Torr. Longer
and 0.08 at 2 and 5 Torr, respectively. pulses have higher average Trot values. (iii) After the end of the
The vibrational and rotational temperatures follow qual- pulse, for short delay times (1 and 2.5 ms), Trot and Tvib do not
itatively a similar evolution as that measured in a pulsed dc relax completely before the next pulse starts. When the plasma
glow discharge in the same pressure range.22,26 The purpose of is switched ON again it restarts in a gas that is at higher
these measurements is not to investigate in detail the temperature than room temperature for both pressures and out
vibrational kinetics in a rf discharge. Nevertheless, they allows of equilibrium at 2 Torr. (iv) For delay times of 5 ms or longer,
us, on the one hand, to verify that we are in similar excitation all the temperatures have relaxed completely down to room
conditions as in refs 25 and 26 and therefore already suggest temperature before the next pulse.
that the contribution of the vibrational excitation to the All the temperatures reach a steady state value toward the
dissociation could be negligible (discussed in more detail in the end of the 5 ms pulse. For longer pulse durations, the
next section). On the other hand, these experimental data in rotational and vibrational temperatures are expected to follow a
the rf remind and confirm several key points to have in mind constant time evolution after tON = 5 ms and similar relaxation
for the analysis of the influence of pulsing and of Tvib and Trot after the end of the pulse. The similarities in the behavior of
on the gas mixture evolution. These points can be summarized the dc glow and the rf discharges under similar experimental
as follows: (i) The vibrational excitation of T3 and TCO shows a conditions also allow us to assume that the influence of short
maximum around 1 ms after the beginning of the pulse. The pulses or short relaxation times after the pulse are similar to
average vibrational excitation [T3,CO − Trot] during the plasma those described in ref 22. The building-up experiment is
pulse is higher for short pulses than for longer pulses. T3,CO and performed in static conditions; thus, the gas mixture and the
[T3,CO − Trot] are higher at 2 Torr than at 5 Torr. (ii) Trot and total pressure evolve during the experiment. In addition, the
17465 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

pressure evolution during a plasma pulse can be slightly detailed list of experiments and related times involved in each
different in a closed reactor compared to that under flowing experiment is given in Table 3. Extra measurements varying the
conditions. However, the short characteristic times for pulse duration, tON
p , but keeping constant the duty cycle ratio,
reaching the steady-state of rotational and vibrational temper- tON
p /tp
OFF
(Table 3, section 3), are not shown but are discussed
atures and the gas flow experiments done in a regime below.
dominated by diffusion over the transport due to the gas Several general observations can be made about these
flow37 suggest that the main difference between both results: (i) The initial slope (TON < ∼0.5 s) is the same for any
experimental configurations would be given by the different pulse configuration. (ii) Both tONp and tp
OFF
have a strong effect
gas composition. This composition could affect the absolute on the “turning region” and, consequently, on the steady-state
temperature values but not significantly the characteristic times dissociation value. (iii) An increase of the pulse duration tON p
involved.22 Therefore, these measurements, obtained in leads to a significant increase of the steady-state dissociation
flowing conditions and for a certain dissociation fraction, do fraction, α(plateau). (iv) An increase of tOFF
p , i.e., longer delay
not reproduce the exact behavior of every single pulse of the between pulses, leads to a decrease of α(plateau). (v) The time
experiment but provide information on the time scales for the TON at which the curves diverge in tONp variation experiments is
heating and vibrational excitation and a quantitative estimation similar for both pressures TON ∼0.5 s. The effect of varying
of the temperature values in our experimental conditions. tOFF
p seems to happen earlier at 5 Torr TON ∼ 0.5−1 s than at 2
3.3.2. Pulse Duration Experiments. According to the Torr.
observed evolution of the rotational and vibrational temper- Concerning the initial slope, it is noticeable that for short
atures, varying the pulse duration or the delay between pulses total plasma ON times (typically below TON ∼0.5 s), the
in the range of a few milliseconds allows us to play with the evolution of the dissociation fraction with time is quasi-linear
average level of vibrational excitation and gas temperature to and is the same for any pulse configuration, only the initial
study their effect on the dissociation fraction. Figure 6 presents pressure and power affect the initial slope. This behavior is the
in linear time scale (plots (a) and (c)) and logarithmic scale signature of a CO2 first-order dissociation process dominated
(plots (b) and (d)) the time evolution of the dissociation by electron impact: e− + CO2 → e− CO + O.25 Similar
fraction α as a function of TON for 2 and 5 Torr (top and measurements done in a dc glow discharge were used to
bottom panels, respectively) at 40 W, in two cases: varying the validate the electron impact dissociation cross section of CO2
pulse duration, tON in.25 The vibrational excitation of CO2 molecules does not
p , while keeping constant the plasma OFF
time in between pulses, tOFF seem to contribute to the dissociation or to back-reaction
p =10 ms (Table 3, section 1) and
varying the plasma OFF time (delay) in between pulses, tOFF mechanisms in our experimental conditions for short TON.
p ,
for a fixed pulse duration tON In principle, no effect of tON
p or tp
OFF
would be expected in a
p = 5 ms (Table 3, section 2). A
time evolution purely driven by electron impact dissociation of
Table 3. Pulse Parameters Used in the Experiments CO2 with no additional chemical reactions. Nevertheless, a
Presented in Figure 6a clear effect of the pulsing starts from TON ∼0.5 s, when the
amounts of CO and O2 in the total gas mixture become
1: Varying tON OFF
p , Constant tp significant. The differences observed for different pulse
notation Np tON
p tOFF
p tOFF
tr configurations provide an evidence of the influence of chemical
1 × 50−10 1 50 10 10 reactions between heavy species in ground or electrically/
5 × 10−10 5 10 10 50 vibrationally excited states producing back CO2, induced by
10 × 5−10 10 5 10 100 temperature and/or by the excitation degree. In order to
20 × 2.5−10 20 2.5 10 200 discuss the effect of the pulse in the final steady-state, Figure 7
50 × 1−10 50 1 10 500 summarizes the results from Figure 6 by plotting the values of
100 × 0.5−10 100 0.5 10 1000 α at the “plateau” (average of the last 5 s of TON, 20−25 s)
2: Varying tOFF ON
p , Constant tp either as a function of the delay between pulses, tOFF p (panel
notation Np tON
p tOFF
p tOFF
tr
(a)), or as a function of the pulse duration for the tON
p variation
10 × 5−1 10 5 1 10 experiments (panel (b)). Panel (b) includes data taken keeping
10 × 5−2.5 10 5 2.5 25 the duty cycle ratio tON
p /tp
OFF
constant (see Table 3, section 3).
OFF
10 × 5−5 10 5 5 50 3.3.2.1. tp Variation Experiments. Figure 7a shows that
10 × 5−10 10 5 10 100 the dissociation fraction α(plateau) decreases as a function of
10 × 5−20 10 5 20 200 tOFF
p . The decrease is faster for short tp
OFF
values (≤10 ms),
10 × 5−50 10 5 50 1000 following an exponential-like variation with tOFF p . For longer
3: Varying tON
p and t OFF
p , Constant tON
p /tp
OFF tOFF
p , the effect tends to become negligible. The inner graph in
notation Np tON
p tOFF
p tOFF
tr
Figure 7a plots the same data normalized to the shortest tOFF p

1 × 50−100 1 50 100 100


(1 ms) and shows that for higher pressures (5 Torr) α
2 × 25−50 2 25 50 100
decreases more and faster than for lower pressures (2 Torr).
5 × 10−20 5 10 20 100
The variation of α(plateau) with tOFF
p suggest the occurrence of
10 × 5−10 10 5 10 100
“back-reaction” in between pulses, more efficient during the
20 × 2.5−5 25 2.5 5 100 first milliseconds of the OFF phase. The smaller effect for long
50 × 1−2 50 1 2 100 tOFF
p could be caused by a back-reaction mechanism with a
100 × 0.5−1 100 0.5 1 100 longer characteristic time, compatible with the effect of varying
the number of trains observed in section 3.1. The back-
a reaction mechanism either in gas phase and/or on the surface,
Ntr = 500, tON
tr = 50 ms, and T
ON
= 25 s are constant for all the
experiments. The time parameters are in units of milliseconds. could be related with the gas temperature, or with the presence
17466 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 7. Variation of α at the plateau (average of TON = 20−25 s) as a function of (a) delay time between pulses, tOFF ON
p , and (b) pulse duration, tp ,
OFF ON
for 2 and 5 Torr. The corresponding time evolution data for tp and tp variation experiments can be seen in Figure 6. The experimental
parameters corresponding to the constant tON OFF
p /tp (constant duty cycle ratio) data, included in panel (b) and not shown in Figure 6, are detailed in
Table 3, section 3. The inner graph in panel (a) plots the same data as the outer graph but normalized to the shortest tOFF p (1 ms).

of exited species or radicals created during the pulse, further the short pulse of tON p = 0.5 or 1 ms leads to a lower final
discussed in sections 3.4 and 5. dissociation fraction (10 ms versus 1 or 2 ms). The most
3.3.2.2. tON
p Variation Experiments. Figure 7b shows a probable reason is that after 1−2 ms the gas is not yet
significant increase of α(plateau) as a function of tON p in both completely relaxed, particularly at 2 Torr, as shown in Figure 5.
tON
p variation experiments, with variable or constant duty cycle A shorter tOFF p (constant DCR experiment) would simply
ratio (DCR). Longer plasma pulses lead to higher steady-state reduce the time available for back-reaction in the OFF phase
dissociation values for the same total accumulated plasma ON leading to higher dissociation values.
time, TON. For the longest pulses tON p ≥ 2.5 ms (tp ≥ 5 or 10
OFF
3.4. Hypothesis for Back-Reaction Mechanisms. We
ms), both experiments give the same dissociation fraction have observed evidence of back-reaction mechanisms. The
within the reproducibility error. The difference between curve of the dissociation fraction as a function of time saturates
keeping constant or not the DCR is noticeable only for very for both continuous and pulsed plasma experiments. This
short tON
p (0.5, 1 ms). effect suggests the occurrence of back-reactions in the ON
In the literature, pulsing is claimed to be beneficial for an phase, enough to eventually balance the CO2 dissociation
efficient CO2 dissociation due to the lower average gas processes. tOFF
p experiments provide evidence of back-reaction
temperature compared with continuous plasma38,39 and mechanisms in the early postdischarge (OFF phase). Finally,
because the accumulation of reactive species, such as oxygen tON
p variation experiments are consistent with tOFFp and both
atoms, is limited. However, this is only true if several show that the characteristic time for the main back-reaction in
conditions are met: (1) if the vibrational excitation contributes postdischarge is below 10 ms. tOFF
p experiments with long delay
significantly to the CO2 dissociation because the vibrational times and Ntr variation experiments (section 3.1) show a minor
excitation decreases when the gas temperature increases;40,41 back-reaction mechanism happening with a longer character-
(2) if there is no significant back-reaction in the OFF phase; istic time.
and (3) if the main CO recombination reactions are only Different possible mechanisms could be responsible for
dependent on Tgas, such as those involving CO and O in these observations: (i) Direct reaction between CO and O
ground state. In such a case, the gas temperature must remain atoms in a 3-body recombination CO + O + M → CO2 + M or
as low as possible and short pulses are expected to have a ground state CO + O2 reaction in gas phase CO + O2 → CO2
positive effect on the dissociation. + O.42 (ii) Reactions involving two vibrationally excited CO
In our system the dissociation is dominated by electron molecules, such as CO(ν) + CO(ω) → CO2 + C,18,43−45 or
impact, with no significant contribution of the vibrational the associative ionization reaction CO(ν) + CO(ω) → CO2+ +
excitation and pulsing seems to be detrimental for the CO2 C + e.18 (iii) Reactions involving vibrationally excited CO in a
conversion. tOFF
p variation experiments already point toward a 2-body recombination with O2, CO(ν) + O2 → CO2 + O for
“back-reaction” mechanism in the early postdischarge. In the which the vibrational excitation of CO would reduce the
tON
p variation experiments, with constant or variable DCR, and energy barrier.13,24 (iv) Reactions involving electronically
focusing in the common trend for both experiments (tON p ≥ 2.5 excited states of CO, such as CO(a3Πr).32,46 (v) Reactions
ms, tOFF
p ≥ 5 ms), we observe that the total OFF time per train involving, electronically excited states of O2 or atomic oxygen.
tOFF
tr , different in both cases, does not relevantly affect the (vi) Surface back-reaction between O atoms and CO adsorbed
results when tOFF p is long enough (≥5 ms). However, to in the reactor walls.7
decrease the pulse duration but keeping the same TON and tON tr , We can differentiate two groups of processes based on their
we increase the number of pulses Np, therefore increasing the characteristic times: slow, with characteristic times ≥10 ms,
number of “early postdischarge events” (see Table 3). The and relatively fast, ≤5−10 ms. Processes with slow character-
cumulative effect of the back-reaction over a large number of istic times could involve atomic oxygen or long-lived
early postdischarge events explains the decrease of the final metastable states, such as O2(a1Δg), both in gas phase or at
dissociation fraction when the pulse duration decreases. the surface, because of their relatively long lifetime. Short
Differences between both tON p variation experiments appear processes would involve species with shorter lifetimes, such as
for short tON
p (0.5 or 1 ms). In these cases, a longer tp
OFF
after vibrationally or electronically excited states.
17467 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 8. Time evolution of the dissociation fraction α for different initial CO2−O2 mixtures, measured at 2 Torr ((a), in linear time scale, and (b),
in logarithmic time scale) and 5 Torr ((c) and (d), in linear and logarithmic time scales), 40 W.

The lifetime of oxygen atoms, measured in similar varying in the range of ∼2.5−7.6 ms.52−55 Although we cannot
experimental conditions is between 30 and 140 ms. In our completely discard yet an effect of the gas temperature that
range of pressures and gas temperatures the O atom density is would enhance the efficiency of reactions involving atomic
controlled by surface recombination into O2.47 Surface oxygen, or less probably, O2,24 such process would not be
mechanisms involving CO are even slower, in the order of consistent with the long lifetime of O atoms. In addition, the
0.5−1.3 s.32,47 Gas-phase recombination of oxygen atoms into stronger effect of tON p variation at low pressure and the
O2 is less relevant than on the surface, but it is still more decreasing difference between variable and constant tON OFF
p /tp
probable than the reaction between CO and O to produce from 2 to 5 Torr supports an effect of vibrational or electronic
back CO2 in our range of gas temperatures. The rate for this excitation rather than a gas temperature effect. It could be
reaction with both CO and O in ground state is rather low,47,48 related either to a larger reduced electric field at low pressure
as expected for a spin-forbidden recombination reaction and/or to a higher vibrational excitation (see Figure 5). Higher
(CO(X1Σ+) + O(3P) → CO2(X1Σ+g ).49 The radiative lifetime E/N may cause, for instance, a larger density of electronically
of O2(a1Δg) is very long, in the order of ∼72 min.50 Under excited states, that are in addition less efficiently quenched at
plasma exposure, for an O2 plasma under similar experimental low pressure. At higher pressures, the larger effect of tOFF p
conditions, the lifetime of this state was found to be in the variation, where excitation mechanisms do not play a role any
order of 150 ms.51 The long lifetimes of oxygen atoms and more, may be caused by the larger collision frequency.
O2(a1Δg) imply that a buildup of their densities during a train In summary, oxygen atoms or other long-lived species and
of plasma pulses, particularly for short tOFF surface processes could be responsible for the small back-
p , can not be
discarded. However, they certainly de-excite/recombine in reaction effect for long tOFF
p or for different Ntr (see Figures 3
between trains (∼1.5 s). When the next train begins, we restart and 7a), but the strong effect clearly observed right after the
from a gas mixture of CO2 + CO + O2, all in ground state and end of the pulse suggests processes involving vibrationally and/
or electronically excited states. These hypotheses are further
at room temperature. The configuration in series of trains of
discussed in section 5.
plasma pulses limits therefore the accumulation of oxygen
atoms and O2(a1Δg).
Gas temperature and vibrational excitation show faster 4. CO2−O2 AND CO−O2 MIXTURES
relaxation times, in the order of 5 ms, as shown in Figure 5. In principle any back-reaction mechanism leading to the
The lifetime of CO(a3Πr), the first electronically excited state conversion of CO back to CO2 necessarily involves oxygen
of CO, is also in a similar range. CO(a3Πr) is a metastable either in atomic or in molecular form. The long characteristic
state, since the transition to the ground state CO(X1Σ+) is lifetime of oxygen atoms and the associated slow reaction rates
spin-forbidden. Its radiative lifetime depends on the rotational do not support a relevant role of atomic oxygen. In order to get
and vibrational quantum number and the electronic sublevel,52 more insight into back-reaction mechanisms and study the
17468 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 9. Time evolution of the dissociation fraction α for different CO2/CO2−O2 and CO−O2 mixtures with two carbon vs oxygen ratios, 1/4
and 1/2, measured at 2 Torr ((a) in linear time scale and (b) in logarithmic time scale) and 5 Torr ((c) and (d) linear and logarithmic time scales),
40 W.

possible role of O2, CO2−O2, and CO−O2 mixtures are corresponding to different gas mixtures start diverging. When
investigated in this section. increasing the O2 content, the steady-state region is reached for
4.1. CO2−O2 Mixtures. The time evolution of the CO2 a shorter TON. For example, at 2 Torr the steady-state is
conversion for different CO2−O2 gas mixtures, ranging from reached at TON ∼3 s and ∼8−10 s for 0.25CO2−0.75O2 and
100% CO2 to 25%CO2−75%O2, was investigated with the 100% CO2, respectively. The case of 5 Torr is even more
reference series configuration (500 trains ×10 pulses ×5−10 evident: For 0.25CO2−0.75O2, the steady-state is reached at
ms ON−OFF) and is presented in Figure 8. Panels (a) and (b) TON ∼5 s, whereas at 100% CO2 it is barely reached at the end
show the time evolution for 2 Torr up to TON = 25 s in linear of the experiment, TON = 25 s.
(a) and logarithmic scale (b) to focus on the initial rising of A clear threshold at which the “back-reaction” becomes
the dissociation fraction. Panels (c) and (d) show similar data prominent (around TON ∼0.5 s, qualitatively indicated by
but for 5 Torr (40 W). Two main observations can be noticed arrows) is noticeable in Figure 8. It is remarkable that the
from these graphs: (i) Initial rise-up until TON ∼0.5 s not threshold is independent of the percent of O2 in the initial gas
affected, within the experimental error, by the O2 fraction in mixture. Although we could think that the threshold is related
the initial gas mixture for both pressures studied. (ii) Steady- to time, the experiment configuration based on series of trains
state value of the dissociation fraction α strongly affected by of pulses does not support such an idea. There is no
the initial O2 content. Increasing the amount of O2 in the cumulative effect on the gas temperature or on the O atom
initial gas mixture leads to a clear decrease in the final density when progressing along the TON axis. However, the gas
dissociation value. mixture at the beginning of each train does evolve from pure
The similar initial slope agrees with the previous CO2/CO2−O2 toward a mixture of CO2−CO−O2 with
observations made when varying pulse duration or delay increasing CO and O2 density. It is therefore more probably
between pulses, reinforcing the conclusion that these first steps a threshold in concentration. Notably this threshold is similar
in the dissociation are only controlled by the electron impact to that found in the pulse variation experiments (see Figure 6).
dissociation of CO2. The presence of O2 reduces the reduced The lower α value for which the threshold occurs at 5 Torr (α
electric field and therefore the value of the dissociation rate ∼ 0.04) compared to that at 2 Torr (α ∼ 0.1) could then be
coefficient, Kdiss,56 but increases the electron density. For related to the different total gas density.
instance, the electron density increases by ∼16% between pure 4.2. CO−O2 Mixtures. Mixtures of CO + O2 represent the
CO2 and 0.5CO2−0.5O2. Both effects seem to compensate for final “ideal” gas mixture if we could dissociate all the CO2
each other, implying that the population of electrons with available and no recombination/oxidation of the CO produced
enough energy for the dissociation of CO2 is not significantly happened. Two gas mixtures were investigated: 0.4CO +
changed by the addition of O2. After ∼ TON = 0.5 s, the curves 0.6O2, with 1 carbon atom for 4 oxygen atoms, a similar
17469 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Table 4. Set of Reactions Used for the Calculations in Figure 10


reaction no. reaction rate coefficient (cm3 s−1)a ref
− −
1 e + O2 → e + O( P) + O( P)/O( D)
3 3 1
EEDF b
59
1 −3
2 O + wall → O2 γOυth/2R = 1.8 × 10 exp(−948/Trot)υth/2R c
47
2
3 e− + CO ↔ e− + CO(a3Πr) EEDF 60
4 CO(a3Πr) + O2 → CO + O2 5−17 × 10−11 32, 52, 61
5 CO(a3Πr) + CO → CO + CO 5.7−14 × 10−11 32, 52, 61
6 CO(a3Πr) + O2 → CO2 + O 3 × 10−11 32
7 CO(a3Πr) + CO → CO2 + C 1.2−2 × 10−12 32, 45, 62, 63
8 C + O2 → CO + O 3 × 10−11 32
9 CO(ν) + O2 → CO2 + O 4.2 × 10−12 exp((24000 − ανEν)/Tgas) 24, 42
a
In s−1 for reaction 2. bEEDF: electron energy distribution function. cυth is the thermal velocity of the oxygen atoms υth = 8kBT /πmO .

proportion C/O as in a mixture of 0.5CO2 + 0.5O2; and O2 and CO2 plasmas gave similar final α(plateau), but it was
0.666CO + 0.333O2, with 1 carbon atom for 2 oxygen atoms, slightly higher than that with an argon plasma pretreatment
similar to a 100% CO2 plasma. (∼6% higher). In the second step, O2 gas flows seems to give
Figure 9 presents data for 2 and 5 Torr the time evolution of higher steady-state α than CO2 flows (∼8% higher). These
α for these two gas mixtures compared with the equivalent pretreatment tests show the same initial slope but suggest that
CO2−O2 measurements in terms of the C/O ratio, shown in surface reactions may play a role for long TON.
Figure 8. Similar to the experiments in CO2 or CO2−O2
mixtures, we observe three regions: an initial fast evolution, in 5. DISCUSSION
this case decreasing the value of α; a “turning region”; and the
final steady-state. These experimental results confirm a We have observed evidence of back-reaction mechanisms
significant “back-reaction” in our plasma. Some additional during the ON phase and in postdischarge with two
observations can be pointed out: (i) The evolution of α for characteristic times: the main effect within a few milliseconds
both CO−O2 mixtures diverges from the beginning of the and a minor effect in the order of hundreds of milliseconds.
experiment, contrary to the common threshold observed in The final steady state is significantly affected by the oxygen
CO2−O2 mixtures. (ii) The decrease of α is faster for larger content, which in principle could participate in back-reactions
amounts of O2 in the initial gas mixture. (iii) The evolution of in the form of atomic oxygen or in the form of O2. However, as
the gas mixture is faster when starting from CO−O2 than that discussed in section 3.4, several arguments discard a dominant
when the initial gas mixture is CO2−O2. Consequently, the effect of oxygen atoms in the main back-reaction mechanism:
steady state is reached for shorter TON. For example, at 2 Torr (1) the long lifetime compared with the characteristic time of
the measurement of 0.5CO2−0.5O2 reaches steady state the main back-reaction in the postdischarge; (2) O atom loss
approximately at TON ∼8−9 s, whereas 0.4CO−0.6O2 is processes dominated by O atom recombination into O2 in the
already stable at TON ∼5−6 s. (iv) The steady-state walls in our range of gas temperatures;47 (3) slow reaction rate
dissociation value is slightly higher for CO−O2 mixtures between CO and O in gas phase or on the surface; (4) limited
than for CO2−O2. accumulation of oxygen atoms due to the experimental
The absence of threshold in these experiments supports the configuration in series of trains of pulses; (5) more significant
conclusion of a threshold related to concentration of reactants back-reaction in pulsed plasma experiments compared with
and not time-related. Assuming that the molecules involved in continuous plasma, where the average O atom density is larger.
the back-reaction mechanism are CO and O2 (or even O), the In CO2−O2 and CO−O2 mixtures, the O2 content could in
larger amount of reactants in the initial gas mixture induces a principle increase the O atom density. However, even in
faster overall evolution and explains the shorter TON required continuous pure O2 plasma the maximum of oxygen atom
to reach the steady-state. density measured in a similar reactor configuration was only
Measurements done starting from a CO−O2 mixture give between 17 and 8% of the total gas density at 2 and 5 Torr,
slightly higher steady-state dissociation fractions. The differ- respectively, for a similar deposited power.30 In conclusion, the
ence between α(CO2−O2) and α(CO−O2) seems to decrease main back-reaction mechanism should involve molecular
with the initial amount of O2 for both pressures, i.e., oxygen and not atomic oxygen.
[α(0.666CO−0.333O2) − α(1CO2)] > [α(0.4CO−0.6O2) − There are several possibilities for back-reaction mechanisms
α(0.5CO2 − 0.5O2)], and with pressure. The reason for this involving CO and O2: reactions between ground-state
minor effect is still unclear. A plausible hypothesis would be molecules controlled by the gas temperature, reactions
the influence of surface processes that can be relevant when involving vibrationally excited CO, or reactions involving
working in static conditions. This could be caused for instance electronically excited states. In order to do a preliminary
by different surface reactivity (and/or adsorption) of CO2 and analysis of the role of the different possible back-reaction
CO on the Pyrex surface. The presence of carbonated species, mechanisms we have done a exploratory calculation focusing
particularly CO, has already been shown to affect the surface on CO−O2 mixtures. Only the first 300 ms are considered,
recombination of atomic oxygen in similar experimental since the density of CO2 is still low and can be neglected. A
conditions.47 Although more experiments are required to reduced set of reactions, detailed in Table 4, is used to
confirm this hypothesis, an indication of the importance of generate a system of rate-balance equations, solved by a
surface processes was already observed when modifying the Runge−Kutta method. Two independent calculations were
pretreatment procedure. In the first step of the pretreatment, done, considering either (1) reactions involving CO(a3Πr)
17470 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

(reactions 3−7) or (2) reactions involving vibrationally excited


CO (reaction 9). Two reactions accounting for O2 dissociation
(producing either 2O(3P) or O(3P) and O(1D)) and O atom
recombination (reactions 1 and 2) are also included for both
sets, and a recombination reaction of carbon atoms (reaction
8) is included for the first set. The density of the vibrational
levels of CO is calculated assuming a Treanor−Gordiets
distribution57,58 at the vibrational temperatures shown in
Figure 5. The rate coefficient from ref 42 is modified to include
the effect of the vibrational excitation on lowering the reaction
barrier, by adding the term −ανEν where Eν is the energy of the
corresponding vibrational level and αν = 0.5 according to the
Fridman−Macheret αν model.13,24 For vibrational levels for
which the activation energy for the reaction is lower than
−ανEν, we impose Eact − ανEnν = 0 excluding the possibility of
a negative activation energy.18
The electron density is estimated by hairpin probe
experiments, as described in section 2.5. Axial measurements
along the length of the reactor axis show that the electron
density is not homogeneous, with a maximum under the high-
voltage electrode and a minimum under the grounded
electrodes. The electron densities considered for the
calculation are an average of the densities measured along
the axis. The obtained values are 1.15 × 1016 and 1.25 × 1016
m−3 for 2 and 5 Torr, respectively, for the 0.667CO−0.333O2 Figure 10. Time evolution of the dissociation fraction α for CO−O2
mixture, and 1.37 × 1016 and 1.49 × 1016 m−3 for 2 and 5 Torr, mixtures (top panel for 0.667CO−0.333O2 and bottom panel for
respectively, for 0.4CO−0.6O2. The electron energy distribu- 0.4CO−0.6O2) for 2 and 5 Torr. The experimental data points are
tion functions (EEDF) for the distinct experimental conditions represented by blue squares for 2 Torr and by red diamonds for 5
are calculated using the Boltzmann solver LoKI,64 with the Torr. The model results are represented by a shaded area for the
cross section set from IST-Lisbon database in LxCat.65 Details calculations accounting for CO(a3Πr) reactions and with circles for
the reactions involving vibrationally excited CO (CO(ν)).
on the processes considered in the calculation of the EEDF can
be found in refs 56, 66, and 67. The rate coefficients for
electron excitation of CO and for the electron impact
dissociation of O2 are calculated by integration of the Both panels in Figure 10 also include calculations accounting
corresponding cross section over the EEDF. The cross section for the back-reaction with vibrationally excited CO (reaction
for electron impact excitation of CO(X1Σ+) to the CO(a3Πr) 9), represented by circles. For the vibrational temperatures
state is taken from Sawada et al.60 The rate coefficients for shown in Figure 5, considering either the excitation peak or the
reactions involving CO(a3Πr) and ground state species are average TCO values, no decay in the dissociation fraction α is
scarce and the values available in the literature exhibit observed in the calculations for any experimental condition
significant discrepancies; see refs 32, 45, 52, and 61−63 and (the calculations for all the conditions overlap and are
references therein. Note that some refs 52 and 61 provide rate indistinguishable in the figure). Higher vibrational temper-
coefficients for the disappearance of the CO(a3Πr) state atures are required to see an effect. As an example, at the
without specifying the path. Table 4 gives the range of values average Tgas during the pulse for 2 Torr, ∼530 K, an effect of
explored in the present calculations. back-reaction through vibrationally excited CO starts to be
Figure 10 shows a zoom on the CO−O2 experiments for noticeable for a TCO value ∼1900 K, higher than the measured
both 2 and 5 Torr up to TON = 350 ms. The top panel shows average value, 1350 K, and starts to be significant at 2000 K.
the results for 0.667CO−0.333O2 mixture, and the bottom To illustrate this effect the calculated α values for 2 Torr with
panel shows the results for 0.4CO−0.6O2 mixture. In addition the experimental Tgas and TCO = 2000 K are shown in both
to the experimental data points, the shaded areas represent the panels of Figure 10 for the corresponding gas mixtures.
range of α values obtained from the calculations involving Obviously, the back-reaction involving O2 molecules and CO
CO(a3Πr) (reactions 1−8) with the maximum and minimum in ground state is even less relevant, and shows no visible
reaction rate coefficients for the quenching (reactions 4 and 5) decay. The three body recombination between CO and O
and for back-reaction 7 in Table 4. The calculated values atoms does not lead to any significant effect either. Reactions
assume a continuous plasma, do not account for the trains of involving two vibrationally excited CO molecules CO(ν) +
pulses and are done for the rotational and vibrational CO(ω) → CO2 + C18,43 could be possible. There are
temperatures measured in a gas mixture that is roughly significant discrepancies in the literature about the activation
0.85CO2−0.10CO−0.05O2 (at 2 Torr) and for an average energy for this reaction.45,63,68 However, calculations consid-
electron density. Therefore, they provide only a first estimation ering the lowest energy reported (∼6 eV), completed with
of the possible role of the different mechanisms. It is however reactions 1, 2, and 8, show no significant decay of α in our
clear from Figure 10 that the calculated values are in the same conditions. Similarly, the associative ionization reaction
order of magnitude and exhibit the same trend as the (CO(ν) + CO(ω) → CO2+ + C + e) is not significant either
experimental data points. because of the large ionization energy.44
17471 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

We can therefore conclude that the main back-reaction 6. CONCLUSIONS AND PERSPECTIVES
mechanism in our experimental conditions is given by the
In this work,we have studied the time evolution of the
following reactions:32,68,69
dissociation fraction as a function of the plasma ON time in a
CO(a3Π r) + O2 → CO2 + O rf CO2 discharge at pressures between 0.5 and 5 Torr.
Dedicated experiments done under static conditions allowed to
play with the characteristic times of different phenomena. The
CO(a3Π r) + CO → CO2 + C large parametric study shows an initial region dominated by
direct electron impact dissociation. When the CO density
These reactions would be followed by the recombination of becomes large enough, an important contribution of back-
the O atoms into O2 at the walls and the carbon atom reaction mechanisms controlled by electronically excited CO
oxidation through different mechanisms, more probably have been demonstrated. The characteristic time of this
involving O2 but also possibly O, C2O, etc.24,32 Reactions mechanism, observed to influence the early postdischarge (a
between CO(a3Π r) and CO producing C2O are less few milliseconds), and the rate coefficients found in the
probable.52,63 The CO(a3Πr) state is populated by electron literature strongly suggest an important role of the reaction
impact at 6 eV,60,67 radiative cascade transitions, and CO(a3Πr) + O2 → CO2 + O. These results put forward two
dissociative ionization recombination,18 or is created from more general considerations: (i) We note the importance of
the direct electron impact dissociation of CO2 by electrons electronically excited states in the chemistry of neutral species
with energies above 10 eV.70,71 Considering that in our under discharge conditions. These states are often neglected in
discharge the dominant CO2 dissociation mechanism is the study of plasma chemistry. The fast reaction and quenching
electron impact dissociation at ∼7 eV, this state can be rates lead to a low density and make them very challenging to
significantly populated. For gas mixtures with large amount of measure. However, they carry a significant amount of energy
CO2 but low CO density, the reaction CO(a3Πr) + CO2 →
which may influence relevantly the chemistry in the discharge.
2CO + O could contribute to enhance the dissociation.32,68
(ii) The main back-reaction mechanism under our conditions
Reactions involving CO(a3Πr) may therefore play a critical
involves O2 and not atomic oxygen. Atomic oxygen, as a very
role in the “turning region” and in the overall time evolution of
reactive species, is often believed to be the main contributor to
the dissociation fraction. From the calculations, the rate
the back-reaction, and several strategies have been developed
coefficients for quenching reactions must be larger than for
to induce the recombination of O atoms into O2. However, O2
back-reactions; otherwise, the decrease of α is clearly
overestimated. The rate coefficients that provide a closer may also behave as a relevant oxidation species in conjunction
result to the experimental values are from ref 32, particularly with excited reaction partners.
for gas mixtures of 0.4CO−0.6O2. Rate coefficients from ref 62 The information and rate coefficients for reactions involving
clearly underestimate the decay of α. excited states are very scarce or nonexistent. Electronically
The contribution of reactions involving electronically excited excited states, such as CO(a3Πr), can also be vibrationally
states of O2 cannot be discarded. Particularly, the density of excited, but the information available is even less abundant.
O2(a1Δg) may be significant due its low energy and long Quenching reactions of electronically excited states may also
lifetime.50 No rate coefficient was found in the literature for have a strong impact in the vibrational kinetics of CO2 and
reactions involving O2(a1Δg) in CO2−CO−O2 plasmas, nor CO, such as between O(1D) and CO2 or CO78,79 or between
did we find information about densities or lifetimes, except CO(a3Πr) and CO(X1Σ+).43,63
about quenching of this state by CO2, which seems to be in a Back-reaction mechanisms involving vibrationally excited
range similar to that by O2.72 Modeling and experimental CO are possible, but they are not dominant under our
results in N2−O2 mixtures show that its density can reach discharge conditions. It is important to notice that there is
values around 10−15% of the total O2 density and also that the always a back-reaction mechanism with an energy threshold
addition of N2 does not affect the kinetics of this state.73 close to the dissociation threshold. In plasma regimes with
Additionally, the reaction CO(X1Σ+) + O2(a1Δg) producing large reduced electric fields, such as those in glow discharges,
ground-state CO2(X1Σ+g ) and O(3Pg) is spin-forbidden and is dielectric-barrier discharges (DBD), or nanosecond discharges,
not likely to be very significant. The reaction producing atomic the main CO2 dissociation mechanism is through electron
oxygen in the O(1Dg) state is not forbidden but is endothermic impact at ∼7 eV, whereas the back-reaction could be
and would require en extra energy input,52 which should/could controlled by electronically excited states of CO at ∼6 eV.
be partially provided by vibrational excitation of CO. In regimes where the vibrational excitation is more
Finally, it is worth mentioning that both mechanisms, important, such as microwave discharges, back-reactions
involving CO(a3Π r) or CO(ν), are strongly coupled. based on the vibrational excitation of CO may become
Efficiencies as high as 89% for the energy transfer of electronic dominant. The low vibrational energy levels of CO are close in
to vibrational energy in the quenching of CO(a3Πr) by CO energy to the vibrations of the asymmetric stretch mode of
have been reported.52,74 The rate coefficients for this process CO2, and quasi-resonant energy transitions between them are
are large (∼10−16 m3 s−1).45,63 It is also possible that the possible. However, the relaxation mechanisms are stronger for
CO(a3Πr) state gets populated by the reaction CO(ν) + CO2 due to inter mode vibration−vibration relaxation (V−V′),
CO(ω) → CO(a3Πr) + CO.43,75−77 The efficiency approaches i.e., energy transfer from asymmetric stretch mode to
the gas kinetic rate when the vibrational energy of the symmetric or bending vibrational modes. Consequently, CO
molecules is similar to the energy required for the formation of always has a higher degree of vibrational excitation, as observed
CO(a3Πr).68 Even under conditions of large vibrational in the time evolution of the vibrational temperatures during a
excitation and low electron energies, CO(a3Πr) can be plasma pulse. In the eventual case of a main dissociation
efficiently populated by this process and undergo the discussed mechanism based on vibrational ladder climbing, the vibra-
back-reactions mechanisms. tional up-pumping between two vibrationally excited CO could
17472 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

produce a CO2 molecule, produce CO(a3Πr), or lead to an of catalytic reactions on the surfaces of heat shield materials in
increase of the reaction rate in reactions involving molecular dissociated carbon dioxide flows. Fluid Dyn. 1997, 32, 876−886.
oxygen and CO. (7) Kovalev, V.; Afonina, N.; Gromov, V. Shock Waves; Springer,
The similar thresholds for CO2 dissociation and back- 2005; pp 597−602.
(8) Marieu, V.; Reynier, P.; Marraffa, L.; Vennemann, D.; De
reaction mechanisms suggest three paths to be pursued in
Filippis, F.; Caristia, S. Evaluation of SCIROCCO plasma wind-tunnel
order to increase the CO2 conversion efficiency: effective capabilities for entry simulations in CO2 atmospheres. Acta Astronaut.
separation of the dissociation products, re-utilization of the 2007, 61, 604−616.
produced oxygen atoms in secondary reactions, or fine-tuning (9) Guerra, V.; Silva, T.; Ogloblina, P.; Grofulović, M.; Terraz, L.; da
of the electron energy to enhance dissociation paths but not Silva, M. L.; Pintassilgo, C. D.; Alves, L. L.; Guaitella, O. The case for
recombination mechanisms. in situ resource utilisation for oxygen production on Mars by non-


equilibrium plasmas. Plasma Sources Sci. Technol. 2017, 26, 11LT01.
AUTHOR INFORMATION (10) Premathilake, D.; Outlaw, R. A.; Quinlan, R. A.; Byvik, C. E.
Oxygen Generation by Carbon Dioxide Glow Discharge and
Corresponding Author Separation by Permeation Through Ultrathin Silver Membranes.
Ana Sofia Morillo-Candas − Laboratoire de Physique des Earth and Space Science 2019, 6, 557−564.
Plasmas (UMR 7648), CNRS-Univ. Paris Sud-Sorbonne (11) Stocker, T., Qin, D., Plattner, G., Tignor, M., Allen, S.,
Université-É cole Polytechnique, 91128 Palaiseau, France; Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P. M., Eds. Climate
orcid.org/0000-0002-6974-1240; Email: ana- Change 2013: The Physical Science Basis. Contribution of Working
sofia.morillo-candas@lpp.polytechnique.fr Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change (IPCC); Cambridge University Press: Cambridge,
Authors 2013. https://www.ipcc.ch/report/ar5/wg1/.
Vasco Guerra − Instituto de Plasmas e Fusão Nuclear, Instituto (12) Snoeckx, R.; Bogaerts, A. Plasma technology − a novel solution
for CO2 conversion? Chem. Soc. Rev. 2017, 46, 5805−5863.
Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa,
(13) Fridman, A. Plasma Chemistry; Cambridge University Press,
Portugal 2008.
Olivier Guaitella − Laboratoire de Physique des Plasmas (UMR (14) van Rooij, G. J.; Akse, H. N.; Bongers, W. A.; van de Sanden,
7648), CNRS-Univ. Paris Sud-Sorbonne Université-É cole M. C. M. Plasma for electrification of chemical industry: a case study
Polytechnique, 91128 Palaiseau, France on CO2 reduction. Plasma Phys. Controlled Fusion 2018, 60, 014019.
Complete contact information is available at: (15) Bogaerts, A.; Neyts, E. C. Plasma Technology: An Emerging
https://pubs.acs.org/10.1021/acs.jpcc.0c03354 Technology for Energy Storage. ACS Energy Letters 2018, 3, 1013−
1027.
(16) Liu, C.; Xu, G.; Wang, T. Non-thermal plasma approaches in
Notes CO2 utilization. Fuel Process. Technol. 1999, 58, 119−134.
The authors declare no competing financial interest. (17) Britun, N.; Silva, T.; Chen, G.; Godfroid, T.; van der Mullen, J.;

■ ACKNOWLEDGMENTS
This work was funded by LabEx Plas@par receiving financial
Snyders, R. Plasma-assisted CO2 conversion: Optimizing performance
via microwave power modulation. J. Phys. D: Appl. Phys. 2018, 51,
144002.
(18) Capitelli, M.; Colonna, G.; D’Ammando, G.; Pietanza, L. D.
aid from the French National Research Agency (ANR) under Self-consistent time dependent vibrational and free electron kinetics
project SYCAMORE, reference ANR-16-CE06-0005-01. V.G. for CO2 dissociation and ionization in cold plasmas. Plasma Sources
was partially funded by the Portuguese FCT (Fundaçaõ para a Sci. Technol. 2017, 26, 055009.
Ciência e a Tecnologia) under projects UIDB/50010/2020 (19) Nunnally, T.; Gutsol, K.; Rabinovich, A.; Fridman, A.; Gutsol,
and UIDP/50010/2020. The authors also thank S. Dine (from A.; Kemoun, A. Dissociation of CO2 in a low current gliding arc
Solayl SAS) for his advice and help with the rf discharge setup, plasmatron. J. Phys. D: Appl. Phys. 2011, 44, 274009.
B. Klarenaar for his help with the data treatment, and J.-P. (20) Bongers, W.; Bouwmeester, H.; Wolf, B.; Peeters, F.; Welzel, S.;
Booth, G. Curley, and D. J. Peterson for their help with the van den Bekerom, D.; den Harder, N.; Goede, A.; Graswinckel, M.;
hairpin probe measurements and electron density calculations. Groen, P. W.; et al. Plasma-driven dissociation of CO2 for fuel


synthesis. Plasma Processes Polym. 2017, 14, 1600126.
(21) Martini, L. M.; Lovascio, S.; Dilecce, G.; Tosi, P. Time-
REFERENCES Resolved CO2 Dissociation in a Nanosecond Pulsed Discharge.
(1) Witteman, W. J. The CO2 Laser; Springer, 2013; Vol. 53. Plasma Chem. Plasma Process. 2018, 38, 707−718.
(2) Pérez-Mendoza, M.; Domingo-García, M.; López-Garzón, F. (22) Klarenaar, B. L. M.; Engeln, R.; van den Bekerom, D. C. M.;
Modifications produced by O2 and CO2 plasma treatments on a glassy van de Sanden, M. C. M.; Morillo-Candas, A. S.; Guaitella, O. Time
carbon: comparison with molecular gases. Carbon 1999, 37, 1463− evolution of vibrational temperatures in a CO2 glow discharge
1474. measured with infrared absorption spectroscopy. Plasma Sources Sci.
(3) Wavhal, D. S.; Fisher, E. R. Modification of polysulfone Technol. 2017, 26, 115008.
ultrafiltration membranes by CO2 plasma treatment. Desalination (23) Damen, M. A.; Hage, D. A. C. M.; van de Steeg, A. W.; Martini,
2005, 172, 189−205. L. M.; Engeln, R. Absolute CO number densities measured using
(4) Bousquet, A.; Cartry, G.; Granier, A. Investigation of O-atom TALIF in a non-thermal plasma environment. Plasma Sources Sci.
kinetics in O2, CO2, H2O and O2/HMDSO low pressure radio- Technol. 2019, 28, 115006.
frequency pulsed plasmas by time-resolved optical emission spectros- (24) Kozák, T.; Bogaerts, A. Splitting of CO2 by vibrational
copy. Plasma Sources Sci. Technol. 2007, 16, 597−605. excitation in non-equilibrium plasmas: a reaction kinetics model.
(5) Babu, D. J.; Yadav, S.; Heinlein, T.; Cherkashinin, G.; Schneider, Plasma Sources Sci. Technol. 2014, 23, 045004.
J. J. Carbon Dioxide Plasma as a Versatile Medium for Purification (25) Morillo-Candas, A. S.; Silva, T.; Klarenaar, B. L. M.; Grofulović,
and Functionalization of Vertically Aligned Carbon Nanotubes. J. M.; Guerra, V.; Guaitella, O. Electron impact dissociation of CO2.
Phys. Chem. C 2014, 118, 12028−12034. Plasma Sources Sci. Technol. 2020, 29, 01LT01.
(6) Bykova, N.; Vasil’evskii, S.; Gordeev, A.; Kolesnikov, A. F.; (26) Klarenaar, B. L. M.; Morillo-Candas, A. S.; Grofulović, M.; van
Fershin, I.; Yakushin, M. Determination of the effective probabilities de Sanden, M. C. M.; Engeln, R.; Guaitella, O. Excitation and

17473 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

relaxation of the asymmetric stretch mode of CO2 in a pulsed glow (47) Morillo-Candas, A. S.; Drag, C.; Booth, J.-P.; Dias, T. C.;
discharge. Plasma Sources Sci. Technol. 2019, 28, 035011. Guerra, V.; Guaitella, O. Oxygen atom kinetics in CO2 plasmas
(27) Stenzel, R. L. Microwave resonator probe for localized density ignited in a DC glow discharge. Plasma Sources Sci. Technol. 2019, 28,
measurements in weakly magnetized plasmas. Rev. Sci. Instrum. 1976, 075010.
47, 603−607. (48) Manion, J. A.; Huie, R. E.; Levin, R. D.; Burgess, D. R.; Orkin,
(28) Sands, B. L.; Siefert, N. S.; Ganguly, B. N. Design and V. L.; Tsang, W.; McGivern, W. S.; Hudgens, J. W.; Knyazev, V. D.;
measurement considerations of hairpin resonator probes for Atkinson, D. B. et al. NIST Chemical Kinetics Database, NIST Standard
determining electron number density in collisional plasmas. Plasma Reference Database 17, version 7.0 (web version), release 1.6.8, data
Sources Sci. Technol. 2007, 16, 716−725. version 2015.09; NIST, 2015.
(29) Godyak, V. A.; Piejak, R. B. In situ simultaneous radio (49) Xu, L. T.; Jaffe, R. L.; Schwenke, D. W.; Panesi, M. The Effect
frequency discharge power measurements. J. Vac. Sci. Technol., A of the Spin-forbidden CO(1Σ+)+O(3P) → CO2(1Σg+) Recombination
1990, 8, 3833−3837. Reaction on Afterbody Heating of Mars Entry Vehicles. AIAA
(30) Booth, J. P.; Guaitella, O.; Chatterjee, A.; Drag, C.; Guerra, V.; Thermophys. Conf. 2017, 3486.
Lopaev, D.; Zyryanov, S.; Rakhimova, T.; Voloshin, D.; Mankelevich, (50) Schweitzer, C.; Schmidt, R. Physical Mechanisms of Generation
Y. Oxygen (3P) atom recombination on a Pyrex surface in an O2 and Deactivation of Singlet Oxygen. Chem. Rev. 2003, 103, 1685−
plasma. Plasma Sources Sci. Technol. 2019, 28, 055005. 1758.
(31) Cosby, P. C. ElectronâĂ Ř impact dissociation of carbon (51) Chaterjee, A. O2 DC discharge: Study of O(3P) atoms, O2
monoxide. J. Chem. Phys. 1993, 98, 7804−7818. ground(X) and metastable (a, b) molecules. Ph.D. thesis, Laboratoire
(32) Cenian, A.; Chernukho, A.; Borodin, V.; Śliwiński, G. Modeling de Physique des Plasmas (LPP), Ecole Polytechnique, 2019.
of Plasma-Chemical Reactions in Gas Mixture of CO2 Lasers I. Gas (52) Schofield, K. Critically evaluated rate constants for gaseous
Decomposition in Pure CO2 Glow Discharge. Contrib. Plasma Phys. reactions of several electronically excited species. J. Phys. Chem. Ref.
1994, 34, 25−37. Data 1979, 8, 723−798.
(33) Piejak, R. B.; Godyak, V. A.; Garner, R.; Alexandrovich, B. M.; (53) Jongma, R. T.; Berden, G.; Meijer, G. State-specific lifetime
Sternberg, N. The hairpin resonator: A plasma density measuring determination of the a3Π state in CO. J. Chem. Phys. 1997, 107,
technique revisited. J. Appl. Phys. 2004, 95, 3785−3791. 7034−7040.
(34) Peterson, D. J.; Kraus, P.; Chua, T. C.; Larson, L.; Shannon, S. (54) Mori, S.; Akatsuka, H.; Suzuki, M. Numerical Analysis of
C. Electron neutral collision frequency measurement with the hairpin Carbon Isotope Separation by Plasma Chemical Reactions in Carbon
resonator probe. Plasma Sources Sci. Technol. 2017, 26, 095002. Monoxide Glow Discharge. J. Nucl. Sci. Technol. 2002, 39, 637−646.
(35) Raizer, Y. P. Gas Discharge Physics; Springer, 1991. (55) Gilijamse, J. J.; Hoekstra, S.; Meek, S. A.; Metsälä, M.; van de
(36) Hempel, F.; Röpcke, J.; Miethke, F.; Wagner, H.-E. Absorption Meerakker, S. Y. T.; Meijer, G.; Groenenboom, G. C. The radiative
spectroscopic studies of carbon dioxide conversion in a low pressure lifetime of metastable CO(a3Π, ν = 0). J. Chem. Phys. 2007, 127,
glow discharge using tunable infrared diode lasers. Plasma Sources 221102.
(56) Grofulović, M.; Alves, L. L.; Guerra, V. Electron-neutral
Science and Technology 2002, 11, 266−272.
scattering cross sections for CO2: A complete and consistent set and
(37) Klarenaar, B. L. M.; Grofulović, M.; Morillo-Candas, A. S.; van
an assessment of dissociation. J. Phys. D: Appl. Phys. 2016, 49, 395207.
den Bekerom, D. C. M.; Damen, M. A.; van de Sanden, M. C. M.;
(57) Treanor, C. E.; Rich, J. W.; Rehm, R. G. Vibrational Relaxation
Guaitella, O.; Engeln, R. A rotational Raman study under non-thermal
of Anharmonic Oscillators with Exchange−Dominated Collisions. J.
conditions in a pulsed CO2 glow discharge. Plasma Sources Sci.
Chem. Phys. 1968, 48, 1798−1807.
Technol. 2018, 27, 045009. (58) Gordiets, B. F.; Osipov, A. I.; Stupochenko, E. V.; Shelepin, L.
(38) Britun, N.; Chen, G.; Silva, T.; Godfroid, T.; Delplancke-
A. Vibrational relaxation in gases and molecular lasers. Soviet Physics
Ogletree, M.-P.; Snyders, R. In Green Chemical Processing and Uspekhi 1973, 15, 759−785.
Synthesis; Karame, I., Srour, H., Eds.; IntechOpen, 2017; Chapter 1. (59) Phelps, A. Tabulations of Collision Cross Sections and Calculated
(39) Vermeiren, V.; Bogaerts, A. Improving the Energy Efficiency of Transport and Reaction Coefficients for Electron Collisions with O2; JILA
CO2 Conversion in Nonequilibrium Plasmas through Pulsing. J. Phys. Information Center Report Univ. Colorado, 1985; p 28.
Chem. C 2019, 123, 17650−17665. (60) Sawada, T.; Sellin, D.; Green, A. Electron impact excitation
(40) Blauer, J.; Nickerson, G. A survey of Vibrational Relaxation cross sections and energy degradation in CO. Journal of Geophysical
Rate Data for Processes Important to CO2-N2-H2O Infrared Plume Research 1972, 77, 4819−4828.
Radiation Ultrasystems. 7th Fluid Plasma Dynamics Conf. 1974, 536. (61) Wysong, I. J. Measurement of quenching rates of CO(a3Π, ν =
(41) Grofulović, M.; Silva, T.; Klarenaar, B. L. M.; Morillo-Candas, 0) using laser pump-and-probe technique. Chem. Phys. Lett. 2000,
A. S.; Guaitella, O.; Engeln, R.; Pintassilgo, C. D.; Guerra, V. Kinetic 329, 42−46.
study of CO2 plasmas under non-equilibrium conditions. II. Input of (62) Cenian, A.; Chernukho, A.; Borodin, V. Modeling of Plasma-
vibrational energy. Plasma Sources Sci. Technol. 2018, 27, 115009. Chemical Reactions in Gas Mixture of CO2 lasers. II. Theoretical
(42) Beuthe, T. G.; Chang, J.-S. Chemical Kinetic Modelling of Model and its Verification. Contrib. Plasma Phys. 1995, 35, 273−296.
Non-Equilibrium Ar-CO2 Thermal Plasmas. Jpn. J. Appl. Phys. 1997, (63) Barreto, P. R. P.; Euclides, H. d. O.; Albernaz, A. F.; Aquilanti,
36, 4997−5002. V.; Capitelli, M.; Grossi, G.; Lombardi, A.; Macheret, S.; Palazzetti, F.
(43) Gorse, C.; Capitelli, M. Kinetic processes in non-equilibrium Gas phase Boudouard reactions involving singlet−singlet and singlet−
carbon monoxide discharges. II. Self-consistent electron energy triplet CO vibrationally excited states: implications for the non-
distribution functions. Chem. Phys. 1984, 85, 177−187. equilibrium vibrational kinetics of CO/CO2 plasmas. Eur. Phys. J. D
(44) Adamovich, I.; Saupe, S.; Grassi, M.; Schulz, O.; Macheret, S.; 2017, 71, 1434−6079.
Rich, J. Vibrationally stimulated ionization of carbon monoxide in (64) Tejero-del Caz, A.; Guerra, V.; Gonçalves, D.; da Silva, M. L.;
optical pumping experiments. Chem. Phys. 1993, 173, 491−504. Marques, L.; Pinhão, N.; Pintassilgo, C. D.; Alves, L. L. The LisbOn
(45) Essenhigh, K. A.; Utkin, Y. G.; Bernard, C.; Adamovich, I. V.; KInetics Boltzmann solver. Plasma Sources Sci. Technol. 2019, 28,
Rich, J. W. Gas-phase Boudouard disproportionation reaction 043001.
between highly vibrationally excited CO molecules. Chem. Phys. (65) IST-LXCat, IST-Lisbon database. www.lxcat.net (retrieved on 16
2006, 330, 506−514. August 2018).
(46) Pietanza, L.; Colonna, G.; D’Ammando, G.; Laricchiuta, A.; (66) Annušová, A.; Marinov, D.; Booth, J.-P.; Sirse, N.; da Silva, M.
Capitelli, M. Non equilibrium vibrational assisted dissociation and L.; Lopez, B.; Guerra, V. Kinetics of highly vibrationally excited
ionization mechanisms in cold CO2 plasmas. Chem. Phys. 2016, 468, O2(X) molecules in inductively-coupled oxygen plasmas. Plasma
44−52. Sources Sci. Technol. 2018, 27, 045006.

17474 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

(67) Ogloblina, P.; Tejero-del-Caz, A. T.; Guerra, V.; Alves, L. L.


Electron impact cross sections for carbon monoxide and their
importance in the electron kinetics of CO2−CO mixtures. Plasma
Sources Sci. Technol. 2020, 29, 015002.
(68) Gorse, C.; Cacciatore, M.; Capitelli, M. Kinetic processes in
non-equilibrium carbon monoxide discharges. I. Vibrational kinetics
and dissociation rates. Chem. Phys. 1984, 85, 165−176.
(69) Liuti, G.; Dondes, S.; Harteck, P. Photochemical Production of
C3O2 from CO. J. Chem. Phys. 1966, 44, 4051−4052.
(70) Cosby, P. C.; Helm, H. Dissociation Rates of Diatomic Molecules;
Aero Propulsion and Power Directorate, Wright Laboratory, 1992.
(71) Itikawa, Y.; Mason, N. Cross Sections for Electron Collisions
with Water Molecules. J. Phys. Chem. Ref. Data 2005, 34, 1−22.
(72) McLaren, I.; Morris, N.; Wayne, R. Is CO2 a good quencher of
O2(1Δg)? A kinetic reappraisal. J. Photochem. 1981, 16, 311−319.
(73) Guerra, V.; Loureiro, J. Kinetic model of a low-pressure
microwave discharge in O2-N2 including the effects of O− ions on the
characteristics for plasma maintenance. Plasma Sources Sci. Technol.
1999, 8, 110−124.
(74) Slanger, T.; Black, G.; Fournier, J. Electronic-to-vibrational
energy transfer between molecules. J. Photochem. 1975, 4, 329−339.
(75) Dunn, O.; Harteck, P.; Dondes, S. Isotopic enrichment of
carbon-13 and oxygen-18 in the ultraviolet photolysis of carbon
monoxide. J. Phys. Chem. 1973, 77, 878−883.
(76) Farrenq, R.; Rossetti, C.; Guelachvili, G.; Urban, W.
Experimental ro-vibrational populations of CO up to ν = 40 from
Doppler-limited Fourier spectra of the sequences Δν = 1, 2 and 3
emitted by a laser type source. Chem. Phys. 1985, 92, 389−399.
(77) Porshnev, P. I.; Wallaart, H. L.; Perrin, M.-Y.; Martin, J.-P.
Modeling of optical pumping experiments in CO. I. Time-resolved
experiments. Chem. Phys. 1996, 213, 111−122.
(78) Huestis, D.; Marschall, J.; Billing, G.; Maclagan, R. Theoretical
and Experimental Studies of O-CO2 Collisions. Eos Trans. AGU 2002,
83, SA72A-0514.
(79) Chen, H.-F.; Chiang, H.-C.; Matsui, H.; Tsuchiya, S.; Lee, Y.-P.
Distribution of Vibrational States of CO2 in the Reaction O(1D) +
CO2 from Time-Resolved Fourier Transform Infrared Emission
Spectra. J. Phys. Chem. A 2009, 113, 3431−3437.

17475 https://dx.doi.org/10.1021/acs.jpcc.0c03354
J. Phys. Chem. C 2020, 124, 17459−17475

You might also like