plasma dissociation of hydrogen sulfide
TRANSCRIPT
Plasma Dissociation of Hydrogen Sulfide
A Thesis
Submitted to the Faculty
of
Drexel University
by
Kirill A. Gutsol
in partial fulfillment of the
requirements for the degree
of
Master of Science in Mechanical Engineering
September 2012
ii
ACKNOWLEDGMENTS
This work would not have been possible without the guidance of many people who
in one way or another contributed to the completion of my Master of Science degree.
First and foremost, I would like to express my deepest gratitude to my supervisor
and most knowledgeable teacher, Dr. Alexander Fridman for his guidance and support
throughout my studies. It was a great experience working with him and learning in the
process. I am grateful to Drs. Alexander Rabinovich and Andrey Starikovskiy for introducing
me and guiding me through the immense forest of experimental plasma science.
I would like to thank Chevron Energy Technology Company for continued support of
our project on Plasma Dissociation of Hydrogen Sulfide. I would like to acknowledge Drexel
University Machine Shop and Gary Nirenberg for help with manufacturing numerous parts
and for providing guidance.
With great pleasure I would like to acknowledge colleagues from Fuel Conversion
Lab of Drexel Plasma Institute, who worked with me in the lab: Michael Gallagher, Fela
Odeyemi, Robert Geiger, and Ryan Robinson. And special thanks to Thomas Nunnally, who
worked with me on Hydrogen Sulfide Dissociation project.
Next, I extend my gratitude to my family Natalia, Alexander, and Ksenia Gutsol for
their love, encouragement and invaluable support throughout my life. In addition, I want to
thank my mom for her endless patience and my dad for making science an exciting part of
my life.
Finally, I am grateful to my loving wife Krishna Priya Arjunan for being there for me
through successes and failures. This thesis would not have been possible without her help,
understanding, and support.
iii
TABLE OF CONTENTS
List of Tables ............................................................................................................................................................... v
List of Figures ........................................................................................................................................................... vi
Abstract ..................................................................................................................................................................... viii
Chapter 1: Introduction ......................................................................................................................................... 1
1.1 Background ......................................................................................................................................................... 1
1.2. Evaluation and comparison parameters................................................................................................ 3
1.2.1 Specific energy requirement (SER) ....................................................................................................... 3
1.2.2 Specific energy input (SEI) ....................................................................................................................... 4
1.2.3 SEI and SER term applied to plasma processes ............................................................................... 4
1.2.4 SEI and SER sensitivity ............................................................................................................................... 5
Chapter 2: Hydrogen sulfide dissociation modeling ................................................................................. 8
2.1 Motivation and terminology ........................................................................................................................ 8
2.1.1 Quenching ........................................................................................................................................................ 8
2.1.2 Absolute quenching ..................................................................................................................................... 9
2.1.3 Ideal quenching ............................................................................................................................................. 9
2.2 Results of the thermodynamic equilibrium modeling .................................................................... 10
2.2.1 Ideal vs. absolute quenching .................................................................................................................. 10
2.2.2 Effect of reduced pressure ...................................................................................................................... 12
2.3 Chemical kinetics of hydrogen sulfide dissociation ......................................................................... 14
2.3.1 Mechanism I .................................................................................................................................................. 14
2.3.2 Mechanism II ................................................................................................................................................ 17
2.4 Modeling summary ........................................................................................................................................ 18
Chapter 3: Experimental model verification .............................................................................................. 20
3.1 Motivation ......................................................................................................................................................... 20
3.2 Non-thermal plasma discharges .............................................................................................................. 20
iv
3.3 Materials and methods ................................................................................................................................. 21
3.3.1 Measurements .............................................................................................................................................. 22
3.3.2 Plasma discharge configuration ........................................................................................................... 24
3.4 Results ................................................................................................................................................................. 26
3.4.1 AC Corona Discharge ................................................................................................................................. 27
3.4.2 Dielectric Barrier Discharge ................................................................................................................... 30
3.4.3 Streamer Discharge ................................................................................................................................... 32
3.4.4 Contracted Normal Glow Discharge .................................................................................................... 34
3.5 Discussion .......................................................................................................................................................... 38
3.5.1 Corona Discharge ........................................................................................................................................ 38
3.5.2 Dielectric Barrier Discharge ................................................................................................................... 39
3.5.3 Streamer Discharge ................................................................................................................................... 39
3.5.4 Contracted Normal Glow Discharge .................................................................................................... 41
3.5.5 Summary of the Results ........................................................................................................................... 43
Chapter 4: Conclusions ........................................................................................................................................ 45
List of References ................................................................................................................................................... 47
v
LIST OF TABLES
Table 1: Mechanism I for chemical kinetic modeling of H2S dissociation. ..................................... 15
Table 2: The reactions amended to Mechanism I to form Mechanism II. ....................................... 18
Table 3: Different mechanisms for electronic impact initiated dissociation of hydrogen sulfide. ........................................................................................................................................................................ 40
vi
LIST OF FIGURES
Figure 1: SEI and SER sensitivity example. ................................................................................................... 5
Figure 2: Results of the modeling at atmospheric pressure, comparing ideal and absolute quenching concepts. ............................................................................................................................................. 11
Figure 3: Results of the modeling at atmospheric pressure, comparing SER for ideal and absolute quenching concepts. ........................................................................................................................... 12
Figure 4: Final hydrogen mole fraction, α as function of temperature and pressure (under assumption of ideal quenching). ..................................................................................................................... 13
Figure 5: SER as function of SEI and pressure (under assumption of ideal quenching). ......... 14
Figure 6: Comparison of SER of the kinetic (Mechanism I) and thermodynamic modeling. .. 16
Figure 7: Residence time required for kinetics model to reach equilibrium as function of reactor temperature. ............................................................................................................................................ 17
Figure 8: Simplified schematic of the experimental setup (NV –needle valve, BV – ball valve, PG – pressure gauge, and 3WV – 3-way ball valve). ................................................................................ 22
Figure 9: Schematic of the corona reactor (1 – power supply, 2 – silicone stopper, 3 – SS inlet tube, 4 – grounded furnace walls, 5 – molybdenum wire, 6 – SS spring and hook, and 7 – quartz tube). ......................................................................................................................................................... 24
Figure 10: Schematic of the DBD reactor (1 – power supply, 2 – silicone stopper, 3 – SS inlet tube, 4 – furnace walls, 5 – molybdenum wire, 6 – SS spring and hook, 7 – quartz tube, and 8 – DBD ground). ....................................................................................................................................................... 25
Figure 11: Schematic of the streamer and contracted normal glow discharge reactors (1 – power supply, 2 – silicone stopper, 3 – electrodes, 4 – quartz tube, and 5 – Teflon® nozzle). ....................................................................................................................................................................................... 26
Figure 12: Comparison of measured hydrogen mole fraction in experiment and in thermodynamic equilibrium modeling as function of temperature. ................................................ 28
Figure 13: Full and relative SER of hydrogen production (b) as a function of gas temperature. The modeling was performed for the pressure of 0.25 atm under assumption of absolute quenching. ......................................................................................................................................... 29
vii
Figure 14: Hydrogen production in DBD as a function of gas temperature and comparison with results of thermodynamic equilibrium modeling under the assumption of absolute quenching. The modeling was performed for the pressure of 0.25 atm. ........................................ 31
Figure 15: Average power of the streamer discharge as a function of the discharge frequency. ................................................................................................................................................................. 33
Figure 16: Measured hydrogen concentration in the streamer discharge as a function of discharge frequency. ............................................................................................................................................ 34
Figure 17: V-I characteristic of the contracted glow discharge. ......................................................... 35
Figure 18: Hydrogen concentration as function of discharge power in the contracted glow discharge reactor. .................................................................................................................................................. 36
Figure 19: Contracted normal glow discharge reactor in operation (less then 1 min after start). ........................................................................................................................................................................... 37
Figure 20: Contracted normal glow discharge reactor in operation (15 min after start). ...... 37
Figure 21: SER as function of SEI for all the discharge configurations (including MW [3, 4, 30] and GAT [6]) and thermodynamic equilibrium model under ideal quenching assumption. .............................................................................................................................................................. 43
viii
ABSTRACT
Plasma Dissociation of Hydrogen Sulfide
Kirill A. Gutsol
Supervisor: Alexander Fridman, Ph.D.
Plasma dissociation of hydrogen sulfide has been studied by kinetic and
thermodynamic modeling and experimentally in four different discharges: AC corona,
dielectric barrier, streamer, and contracted glow discharge. The performance of corona
discharge and DBD was studied in the initial gas temperature range of 300 – 1200 K. A
robust reactor was created to perform the above studies in pure H2S with high degree of
reproducibly. A specific energy requirement (SER) was calculated as function of energy
input for each type of the discharge and compared with earlier experimental results and
modeling. The results showed that discharges at high 𝐸/𝑛 and low specific energy release
(corona, dielectric barrier, and streamer) perform significantly worse than contracted glow
discharge at low 𝐸/𝑛 where specific energy release was high and gas temperature was
elevated. The SER for non-thermal dissociation was 12 – 14 eV/molec and went down to 2.4
eV/molec in the case of the glow discharge with elevated temperature. Experiments
confirmed that an electrical discharge in H2S allows obtain SER close to the predicted by
thermodynamic equilibrium modeling if the discharge has gas temperature that is high
enough. Further reduction of SER in a plug flow reactor is not possible because of the
absence of chemical chain processes and energy recovery mechanisms.
1
CHAPTER 1: INTRODUCTION
1.1 Background
Hydrogen sulfide (H2S) processing is a necessity for the oil refining and gas
industries. Currently H2S, which is a byproduct of hydro-desulfurization and often
comprises a significant portion of natural gas deposits, is treated via the Claus process. A
gross reaction of the Claus process is partial oxidation of hydrogen sulfide to sulfur and
water. This process, even though well established, has the major drawback of oxidizing
valuable hydrogen, which could be used for hydro-desulfurization and, of course, for many
other applications including “green” energy generation. Hydrogen sulfide can be a good
source for hydrogen production since its dissociation enthalpy to hydrogen and solid sulfur
is only 0.2 eV/molec (compare with the dissociation enthalpy of liquid water that is 2.96
eV/molec). Unfortunately, a process that could accomplish the dissociation and is
economically feasible has not yet been implemented in a large scale.
Much research work has been done on hydrogen sulfide dissociation. Plasma
dissociation became very popular after the experiments in Kurchatov Institute in Soviet
Union demonstrated that dissociation could be achieved with the Specific Energy
Requirement (SER) less than 1 eV/molec. Review papers cover the research done in this
field in great detail [1, 2], and they discuss different methods for H2S dissociation including
plasmas.
Researchers from Kurchatov Institute indicated that their very promising results
were obtained under the conditions of thermal dissociation with additional energy recovery
inside the reactor due to the opposite flows of condensing sulfur clusters and incoming
hydrogen sulfide in a strong centrifugal field [3]. Another possible explanation they
2
provided is based on the dissociation process equilibrium shift due to the high diffusion
coefficient of hydrogen [4]. Thermodynamic equilibrium modeling shows that, at
atmospheric pressure dissociation of H2S to gaseous products (H2, HS, HS2, S2, …) has a
thermodynamic minimum of SER of 2.1 eV per molecule of H2 produced (assuming absolute
quenching), and 2.05 eV per H2 molecule produced (assuming ideal quenching when SH and
H radicals are also converted to H2 molecules). This 0-dimentional time-independent model
is a good first approximation of a plug flow reactor at high temperature, where all reactions
are fast enough to establish equilibrium, while all the reverse reactions are stopped by
quenching.
The best experimentally obtained SER is 0.76 eV/molec, which was reported for a
low pressure microwave plasma reactor [5]. Recently reported results of experiments in
Gliding Arc Tornado (GAT) reactor and Gliding Arc Plasmatron (GAP) at atmospheric
pressure show SER of 1.2 eV/molec [6]. As these results are significantly better than those
of thermodynamic equilibrium modeling, a possibility exists that the process is defined by
some non-equilibrium effects (for example on periphery of thermal plasma discharge) that
were not taken into consideration.
This uncertainty was the primary motivation for our research because it is
necessary to understand all the steps of the process to be able to design large (industrial)
scale installations. This work investigates the H2S dissociation process in different plasmas
and compares the obtained results to each other as well as to those of other researchers.
The experiments were designed so that as many parameters remained constant as possible
in between the different discharges such as reactor geometry, measurement and control
instrumentation.
3
Another motivation is that many researchers have performed experiments with
hydrogen sulfide in a number of different discharges, but almost all of these experiments
were done in mixtures where hydrogen sulfide was significantly diluted by a carrier gas.
The dilution of hydrogen sulfide by a carrier gas can make radical changes in the discharge
properties, which could, quite possibly, change the results of an otherwise very well
prepared and executed experiment.
1.2. Evaluation and comparison parameters
In the following sections many experimental and computational results will be
evaluated in terms of the specific energy requirement (SER), which is a measure of
effectiveness of a (plasma) system in production or destruction of a specific chemical
substance or a group of substances. Specifically most of the processes and systems
described in this article are evaluated by their SER of hydrogen production or hydrogen
sulfide dissociation.
1.2.1 Specific energy requirement (SER)
Specific energy requirement is an effective tool for the comparison of the systems,
which have very different underlying principles. Using SER, it is possible to compare
systems, which use thermal or non-thermal plasma, and even systems that do not use
plasma at all.
The common units used in the literature for SER are kilojoule per mole (kJ/mol),
kilowatt hour per normal cubic meter (kWh/Nm3), and electron volt per molecule
(eV/molec). The conversion factor is such, that
1𝑘𝑊ℎ𝑚3 = 0.84
𝑒𝑉𝑚𝑜𝑙𝑒𝑐
= 80.64𝑘𝐽𝑚𝑜𝑙
. (1)
4
The eV/molec unit is used exclusively throughout this thesis.
SER is not measured directly but calculated from the measured or computed
parameters. In general,
ε = 𝑊/𝑄𝑖 (2)
where 𝜀 is SER, W is total power input, and 𝑄𝑖 is a flow rate of the substance of interest in
the product mixture. When SER of destruction of a substance is sought then 𝑄𝑖 is the
difference of initial and final flow rate of the substance.
1.2.2 Specific energy input (SEI)
Another important parameter for comparison of different systems is specific energy
input (SEI), J, which is a measure of total power input to the source stream
𝐽 =𝑊𝑄0
(3)
and 𝑄0 is initial substance flow rate. In this research, we are focusing on hydrogen sulfide
dissociation, and thus, unless stated otherwise, SEI is a measure of energy input per
molecule of H2S, and it is commonly measured using the same units as SER. Calculation of
SEI is different for each system and model and will be explained where necessary.
1.2.3 SEI and SER term applied to plasma processes
It is obvious that SER and SEI are very important parameters from the application
standpoint; however SER has also very important physicochemical sense. Though SER is a
“macro”-parameter, usually it provides very significant information about the mechanism of
a plasma chemical process, especially if kinetic models are available and there is some
understanding about the plasma system applied (cold, thermal, or intermediate – “warm”).
For example, if SER in cold plasma is comparable with the energy necessary to produce one
electron-ion pair (more than 10 eV/molec) it means usually that the plasma chemical
5
process is controlled by a reaction that involves recombination of charged particles. If SER
in warm or thermal plasma is close to that predicted by thermodynamic equilibrium
modeling, it usually means that the process is controlled by temperature alone and there is
an effective quenching mechanism that suppresses the reverse reactions. If SER in cold
plasma is close to the reaction enthalpy, it means that some very special non-equilibrium
conditions are realized, like selective vibrational excitation (e.g. CO2 dissociation in
microwave plasma at low pressure [7]) or a chain process that allows multiple use of a
single charged or electronically excited particle for the chemical process propagation (for
example see the idea of non-equilibrium water dissociation [7]).
1.2.4 SEI and SER sensitivity
It is important, that SER is not very sensitive to system non-uniformity. To illustrate
this, let’s consider a rather extreme case. Let us assume that we have a gas (e.g. H2S) flow
through a very non-uniform discharge system (Figure 1).
Figure 1: SEI and SER sensitivity example.
In this system 16 of the flow, 𝑄, is passing throw a Zone #1 with optimal conditions
where SER has the lowest value, 𝜀1, and conversion degree is 𝛼1. In this case the power
released in the Zone #1 is
6
𝑊1 =16𝑄𝛼1𝜀1 (4)
Next, let us assume that another 16 of the flow 𝑄 is passing through Zone #2 with the
same power as Zone #1, but unfavorable conditions where SER has rather high value,
𝜀2 = 3𝜀1. Lastly let us assume that 23 of the flow 𝑄 is passing throw Zone #3, where there is
no power release, and therefore, there is no conversion (𝛼3 = 0). It is easy to see that in this
extreme case 𝛼2 = 13𝛼1, and total flow rate of the product (H2 for H2S dissociation with
sulfur condensation) is
𝑄𝑝 =16𝑄𝛼1 +
16𝑄𝛼2 =
29𝑄𝛼1, (5)
the total power is
𝑊 = 𝑊1 + 𝑊2 = 2𝑊1 =13𝑄𝛼1𝜀1. (6)
Also, in this case, the average SER is
ε =𝑊𝑄𝑝
=32𝜀1, (7)
which is rather close to the lowest value and is absolutely insensitive to the flow through
the plasma-free zone. SEI, on the other hand, is very sensitive to the flow through the
plasma-free zone. Thus, for the example above,
𝐽1 = 𝛼1𝜀1 = 𝐽2; 𝐽3 = 0, (8)
and the average SEI is
𝐽 =𝑊𝑄
=13𝛼1𝜀1. (9)
7
Though SEI has much less sense for process characterization than SER, presentation
of SER as a function of SEI is very convenient and will be used in this thesis because it
provides information about the conversion degree α = 𝐽/ε. Also, it is rather obvious, that
SEI has significant meaning in 0-D and 1-D kinetic and thermodynamic simulations. For
example, in these cases, process temperature is completely defined by SEI.
8
CHAPTER 2: HYDROGEN SULFIDE DISSOCIATION MODELING
2.1 Motivation and terminology
A number of research group have attempted research on plasma dissociation of
hydrogen sulfide. Usually each group is focusing on one specific plasma discharge with a
relatively fixed set of parameters. In our research we attempt to breach this limitation, and
we using modeling as starting point that will allow us to obtain a solid hypothesis for future
experimental studies.
2.1.1 Quenching
In the process of modeling we compute species concentration at certain
temperature, which is not the same that can be effectively reached in experimental and
production conditions. Most of the time the end products are collected at ambient
temperature, which in the lab environment corresponds to 300 K. So, we know
concentrations at elevated temperature (suitable for high degree of dissociation or low
SER), but we also need know how the drop in temperature to ambient will affect the
calculated concentrations. Thus, the resulting concentration depends on the cooling or
quenching speed of the process.
Let’s take hydrogen sulfide dissociation as an example. The dissociation causes H2S
to split into a number of species, which may include some of the following or even more
species depending on the temperature: H, H2, HS, H2S, S, and S2. Here S and S2 is a gaseous
because H2S dissociates at temperatures above 750K, and at these temperatures sulfur is
already evaporated and found in the forms from S2 to S8 (with S8 being present at lower
temperatures). At the room temperature conditions (300 K) only the following species
9
remain: H2, H2S, and S (solid sulfur). In the worst case scenario the quenching is so slow that
all the products recombine back into H2S.
Let’s say the main product of interest is hydrogen, and let’s say final hydrogen
concentration is [H2]f. Without performing detailed quenching kinetics modeling we can still
make meaningful estimations for the value of [H2]f, using absolute and ideal quenching
concepts [8].
2.1.2 Absolute quenching
One of the ways to estimate [H2]f is using the concept of absolute quenching.
Absolute quenching concept assumes that all stable products are preserved during
quenching, and the rest of the radicals convert to the initial substance. So, in our case:
𝑀H2𝑓 = 𝑀H2 , 𝑀SS
𝑓 = 2𝑀S2 (10)
where 𝑀𝑋 is the number of moles of specie X. Now taking to account that SS is solid and this
doesn’t contribute to gaseous concentration of the products then
[H2S]𝑓 = 1 − [H2]𝑓 . (11)
2.1.3 Ideal quenching
The concept of ideal quenching assumes that all the radicals are converted into
products rather than initial substance. So, in the case of hydrogen sulfide dissociation:
𝑀H2𝑓 = 𝑀H2 +
12𝑀H +
12𝑀HS, 𝑀SS
𝑓 = 2𝑀S2 + 𝑀S + 𝑀HS, 𝑀H2S𝑓 = 𝑀H2S (12)
The final concentration of hydrogen ([H2]𝑓) is calculated, similar to hydrogen sulfide final
concentration the case of absolute quenching, using the following equation:
10
[H2]𝑓 = 1 − [H2S]𝑓 . (13)
2.2 Results of the thermodynamic equilibrium modeling
All thermodynamic modeling was performed in CHEMKIN® 4.1.1 software, which
uses STANJAN libraries to determine equilibrium mixture composition. In general, the
method for determining equilibrium is based on minimizing Gibbs free energy.
The enthalpy of the mixture and mixture composition are calculated for each set
temperature in the model. This data is used to calculate corresponding SEI and SER of
species:
𝐽 = 𝐻𝑇 − 𝐻0 (14)
𝜀 =𝐽𝛼
(15)
where 𝐻𝑇 is enthalpy of the mixture at temperature 𝑇, 𝐻0 is the enthalpy of the H2S at room
temperature and α is the conversion degree.
2.2.1 Ideal vs. absolute quenching
Earlier given definitions of absolute and ideal quenching can be very well illustrated
with thermodynamic model of hydrogen sulfide dissociation. The results produced by such
modeling are shown in Figure 2 and Figure 3. Note the difference in both hydrogen
concentrations and SER of two different quenching mechanisms.
11
Figure 2: Results of the modeling at atmospheric pressure, comparing ideal and absolute quenching
concepts.
The Figure 3 clearly shows that the process following ideal quenching has lower SER
(2.03 eV/molec vs. 2.07 eV/molec), and therefore, it is of greater interest. Moreover, ideal
quenching result becomes even better at higher values of SEI. So, the following modeling
results are provided for ideal quenching conditions and following experimental sections
compare results with ideal quenching modeling results.
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
750 1250 1750 2250 2750
H 2 Mol
e Fr
actio
n, α
Temperature, T (K)
Absolute
Ideal
12
Figure 3: Results of the modeling at atmospheric pressure, comparing SER for ideal and absolute
quenching concepts.
2.2.2 Effect of reduced pressure
In agreement with Le Chatelier's principle, the reduction in pressure causes greater
dissociation and respectively lowers specific energy requirements. The results of the
modeling are show in Figure 4 and Figure 5.
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00
SER,
ε(e
V/m
olec
)
SEI, J (eV/molec)
Absolute
Ideal
13
Figure 4: Final hydrogen mole fraction, α as function of temperature and pressure (under assumption
of ideal quenching).
We can observe that thermodynamic equilibrium modeling even in ideal quenching
assumptions cannot explain the best results of experiments with microwave, radio
frequency, and gliding arc plasma discharges. Even at pressures of 0.001 atm (0.76 Torr)
the best modeling results for SER are 1.46 eV/molec. This observation leads to conclusion
that the earlier mentioned plasma systems must have noticeable degree of non-equilibrium,
which helps to reduce SER. If the system is non-equilibrium then it may be possible to
observe it in chemical kinetics modeling, which we discuss in greater detail in the following
sections.
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
750 1250 1750 2250 2750
H 2 Mol
e fr
actio
n, α
Temperature, T (K)
1.0 atm
0.1 atm
0.01 atm
0.001 atm
14
Figure 5: SER as function of SEI and pressure (under assumption of ideal quenching).
2.3 Chemical kinetics of hydrogen sulfide dissociation
Chemical kinetics modeling was performed using CHEMKIN ® 4.1.1 Software, using
plug-flow reactor approximation model. For each set of calculations the following
parameters were kept constant: reactor cross-section area, volumetric flow rate (1 L/min),
and pressure (1 atm). Dependence of dissociation effectiveness on temperature was
investigated by changing constant temperature of isothermal reactor in the increments of
100 K. SEI and SER of the model was calculated in the same way as it was done for
thermodynamic equilibrium models (see equations (14 and (15)
2.3.1 Mechanism I
The first kinetic mechanism that we considered was introduced by researcher from
the Kurchatov Institute in Moscow [3, 5, 9]. The complete list of reaction of this mechanism
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00
SER,
ε (e
V/m
olec
)
SEI, J (eV/molec)
1.0 atm
0.1 atm
0.01 atm
0.001 atm
15
is presented in Table 1. The mechanism is characterized by high activation energy of the
two initiating reactions (1 and 2 in Table 1). As a results despite of the fact that some small
dissociation is possible thermodynamically under 1000 K, but due the kinetic limitations
the process is very slow in this temperature range and thus, the same concentration of
hydrogen cannot be obtained keeping residence time within several seconds (technological
limitations for flow systems).
Table 1: Mechanism I for chemical kinetic modeling of H2S dissociation.
# Reaction 𝐴,𝑐𝑚3
𝑚𝑜𝑙𝑒𝑐 ⋅ 𝑠 n 𝐸𝑎 ,
𝑘𝑐𝑎𝑙𝑚𝑜𝑙𝑒
1 H2S + M ↔ SH + H + M 2.92E-08 0.00 66.21 2 H2S ↔ H2 + S 3.16E-10 0.00 65.49 3 H2S + H ↔ H2 + SH 2.31E-07 1.94 0.90 4 SH + S ↔ H + S2 4.00E-11 0.00 0.00 5 SH + H ↔ H2 + S 3.01E-11 0.00 0.00 6 SH + SH ↔ H2 + S2 1.00E-14 0.00 0.00 7 SH + SH ↔ H2S + S 1.50E-11 0.00 0.00 8 S2 + M ↔ S + S + M 7.95E-11 0.00 76.96
10 H2 ↔ H + H 3.70E-10 0.00 96.08
The comparison of this kinetic model results with those of thermodynamic
equilibrium are shown in Figure 6. We can observe that kinetic modeling curve matches
thermodynamic equilibrium, but this is only true if either make the flow extremely slow or
reactor length extremely long. The time that it takes for kinetic system to reach equilibrium
is shown as function of reactor temperature in Figure 7.
16
Figure 6: Comparison of SER of the kinetic (Mechanism I) and thermodynamic modeling.
In general, industry acceptable time is no more than a few seconds. This time limit
comes from huge flow rates of gas that needs to be processed yet maintain reasonable size
of the installation. From the Figure 7 we can see that this time corresponds isothermic
reactor temperature of 1200 - 1300 K.
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
0 0.5 1 1.5 2 2.5 3
SER,
ε (e
V/m
olec
)
SEI, J (eV/molec)
Thermodynamic
Kinetic
17
Figure 7: Residence time required for kinetics model to reach equilibrium as function of reactor temperature.
2.3.2 Mechanism II
The later research by Binoist, et al. introduced and experimentally confirmed
mechanism that included additional species, H2S2 and HSS. These species are radicals and
not found in products, but they provide additional pathway for dissociation. The Mechanism
I was amended by the reactions involving H2S2 and HSS. Reactions amended to Mechanism I
are shown in Table 2.
1.0E-06
1.0E-05
1.0E-04
1.0E-03
1.0E-02
1.0E-01
1.0E+00
1.0E+01
1.0E+02
1.0E+03
1.0E+04
900 1400 1900 2400 2900
Tim
e, t
(s)
Temperature, T (K)
Feas
ible
tim
e fr
ame
18
Table 2: The reactions amended to Mechanism I to form Mechanism II.
# Reaction 𝐴,cm3
molec ⋅ s n 𝐸𝑎 ,
kcalmole
1 SH + H2S ↔ H2S2 + H 3.32E-10 0.50 27.00 2 H2S2 + M ↔ SH + SH + M 3.43E-07 1.00 57.12 3 HSS + HSS ↔ H2S2 + S2 3.46E-15 2.37 -1.67 4 HS + HSS ↔ H2S + S2 3.66E-13 3.05 -1.10 5 H + HSS ↔ S + H2S 7.32E-11 0.00 6.32 6 H + HSS ↔ H2 + S2 2.51E-12 1.65 -1.10 7 S + HSS ↔ HS + S2 2.00E-02 2.20 -0.60
Addition of the seven reactions from Table 2 does not change the outcome of the
calculated SER values. Still, close inspection of time resolved solutions and their comparison
with thermodynamic modeling of the mixture including H2S2 and HSS showed that there is
significant accumulation (~2%) of H2S2 at lower temperatures 900-1200 K.
Literature research into properties of H2S2 revealed that H2S2 belongs to a group
sulfur compounds known as sulfanes. So, H2S2 is not a radical it is a stable compound that
can be present as a product even at room temperature [10].
2.4 Modeling summary
Thermodynamic equilibrium modeling of hydrogen sulfide dissociation showed that
• Earlier reported minimum for SER of hydrogen production under
assumption of ideal quenching (2.03 eV/molec) remains valid after adding
new species (H2S2 and HSS) to list.
• The thermodynamic SER minimum is achieved at the temperature of 1875 K
at atmospheric pressure. The respective temperatures of SER minimum for
the pressures of 0.1 atm, 0.01 atm, and 0.001 atm are: 1625 K, 1425 K, and
1250 K.
19
• There is rather strong dependance of SER minimum on pressure, which
favors low pressure plasma system like microwave system used in
Kurchatove institute [5].
Chemical kinetic modeling of hydrogen sulfide dissociation sowed that
• Given sufficent time, both kinetic mechanism I and II produce SER of
hydrogen sulfide dissociation the same as the thermodynamic model.
• The time required for kinetic models to reach equilibrium is an exponential
function of reactor temperature. And even though, at SER favorable
temperatures of 1700 K or higher the equilibration time is less than 0.01 s,
the time is too long for any temperature lower than 1300 K.
In general, modeling shows that high-temperature low-pressure conditions are
favorable for efficient H2S dissociation. Nonetheless, none of the models (even with ideal
quenching assumption) were able to match the efficiency reported from the experiments
with gliding arc, microwave, and radio frequency plasma.
20
CHAPTER 3: EXPERIMENTAL MODEL VERIFICATION
3.1 Motivation
Earlier described modeling of hydrogen sulfide dissociation allowed us to formulate
a hypothesis for experimental part of our study. The hypothesis can be stated as follows:
there exists a plasma assisted mechanism of low temperature hydrogen sulfide dissociation, in
which non-equilibrium dissociation takes place. This allows for high concentration of
hydrogen to be produced at the temperatures as low as 800-1200 K. As a result of low energy
input and high conversion, the specific energy requirement of the dissociation process is
significantly lower than SER of equilibrium processes with ideal quenching (2.1 eV/molec).
The hypothesis clearly indicates that experiments should be performed at elevated
temperatures, but these temperatures should not be sufficient for thermal decomposition of
H2S. Thus, there is additional limitation for our experimental study – we must use non-
thermal plasma discharges to catalyze the dissociation.
3.2 Non-thermal plasma discharges
All plasma discharges are commonly divided into thermal and non-thermal. The
former ones are also known as quasi-equilibrium, and the later are non-equilibrium. The
primary distinction in between these two groups of discharges is that unlike thermal
plasma discharges the non-thermal plasmas have significant differences in translational,
rotational, vibrational, and electronic temperatures. The detailed discussion on all the non-
thermal plasmas is outside of the scope of this work, but it is covered in a number of books
on plasma discharge physics, chemistry, and applications [8, 11, 12]. The non-thermal
21
discharges that we chose for the experimental portion of this study are corona, dielectric
barrier (DBD), streamer, and glow.
3.3 Materials and methods
Throughout all the experiments, a common reactor design was used to minimize
potential errors and uncertainties related to changes in geometry of the reactor and
instrumentation. The body of the reactor was made out of a quartz tube with length of 0.6
or 1.2 m, internal diameter of 47 mm, and wall thickness of 1.5 mm. Reactor ends were
closed with high purity silicone stoppers, which are heat resistant up to 250 °C. All the
metal parts and connections, unless especially mentioned, were made out of type 316
stainless steel (SS). Gas transport outside of the high temperature zone was provided by
clear PVC tubing with 3.2 mm (inlet) and 9.5 mm (exhaust) internal diameter. Exhaust
mixture was filtered from sulfur via 10 μm polyester filter and scrubbed off hydrogen
sulfide via 3 L washing bottle of 20% water solution of sodium hydroxide (NaOH). A double-
stage diaphragm vacuum pump was used to maintain low pressure (usually below 200
Torr) in the discharge zone. Gas flow rate was measured and controlled with rotameters
(gas pressure in the flow rate measurement point was close to 1 atm).
There was no carrier gas used, and all experiments were performed with 100%
research grade hydrogen sulfide. Ultra-high purity nitrogen was used to purge H2S out of
the reactor at the end of each experiment. A simplified schematic of the installation is show
in Figure 8.
22
Figure 8: Simplified schematic of the experimental setup (NV –needle valve, BV – ball valve, PG –
pressure gauge, and 3WV – 3-way ball valve).
3.3.1 Measurements
Exhaust gas mixture was sampled with vacuum tight disposable 60 mL syringe. The
samples were taken from a low pressure stage of the experimental system, and
consequentially, the size of the sample at normal conditions was 6 – 15 mL depending on
the system pressure. The samples were analyzed with a gas chromatograph (Agilent Micro
GC 3000). The mole fractions of H2, H2S, N2, and O2 were measured. Measures were taken to
ensure that small amount of nitrogen and oxygen were the result of a sampling procedure
only and were not a part of the exhaust gas. This parasitic nitrogen and oxygen were always
present in the same relative concentrations as in ambient air and the overall molar fraction
did not exceed 4%.
The mole fractions of H2S and H2 were normalized to 100%. This is acceptable under
assumption that the only gaseous product that is produced in H2S dissociation is hydrogen
or, in other words, if
23
H2S → αH2 + (1 − α)H2S + Ssolid (16)
where α is molar fraction of H2 in the product stream that is equal to the conversion degree
for this particular process. In this case the SER of H2 production and SER of H2S dissociation
are the same.
Plasma power input was calculated from measurements of voltage and current
obtained from an oscilloscope (Tektronics TDS5052B), using a high voltage probe and
either probe or shunt for current measurement depending on the type of the discharge.
Voltage-current product was integrated over time to calculate average power.
The flow rate of the hydrogen sulfide before the reactor was measured using
rotameters with 2% full scale accuracy. The rotameters were calibrated for air and the
conversion factor 𝑓𝑚 was used to adjust for H2S:
𝑄H2S = 𝑓𝑚𝑄𝐴𝑖𝑟 (17)
𝑓𝑚 = �𝑀Air
𝑀H2S (18)
where M is molecular weight. The use of such a conversion factor makes flow rate
measurement precision rather low with error on the level of 10%. In the general case,
additional corrections for pressure and temperature would have to be made; however, the
inlet gas pressure was maintained at 1 atm and temperature at 25 °C throughout the
experiments, thus eliminating the need for such correction.
SEI was calculated from H2S flow rate 𝑄, set at the inlet of the reactor, and from
voltage and current measurements as follows:
𝐽 =∫ 𝑈𝐼d𝑡𝜏0𝑄τ
(19)
24
where J is specific energy input, U and I are voltage and current respectively, and τ is a
length of a single period for high frequency corona and DBD discharges. In the case of a DC
discharge the above formula reduces to 𝐽 = 𝑈𝐼/𝑄. And, SER of hydrogen production is
ε =𝐽α
(20)
where α is the conversion degree (equal to the mole fraction of hydrogen in the exhaust
mixture after cooling and sulfur separation).
3.3.2 Plasma discharge configuration
Four different types of discharges were used in the experiments: AC corona,
dielectric barrier discharge (DBD), streamer discharge, and contracted normal glow
discharge.
The AC corona discharge was arranged along the reactor axis with 0.25 mm
molybdenum wire serving as a high voltage electrode and furnace casing serving as a
ground. A custom-built high-voltage high-frequency AC power supply was used. The
schematic of the corona reactor is shown in Figure 9.
Figure 9: Schematic of the corona reactor (1 – power supply, 2 – silicone stopper, 3 – SS inlet tube, 4 –
grounded furnace walls, 5 – molybdenum wire, 6 – SS spring and hook, and 7 – quartz tube).
25
The DBD was arranged similar to the corona but with the ground electrode made
out of a thin (~0.5 𝑚𝑚) aluminum 20 mm wide stripe wrapped tightly around the quartz
tube. Two series of experiments were performed, where DBD was placed before and after
the furnace. The power supply that was used in AC corona discharge experiments was used
for DBD experiments as well. The schematic of the DBD reactor is shown in Figure 10.The
length of the furnace and flow rate of the H2S was chosen in corona and DBD experiments so
that the gas would heat up to the set furnace temperature. The estimation showed that the
expected residence time within the furnace can be up to 1 minute, which would guarantee
gas preheating.
Figure 10: Schematic of the DBD reactor (1 – power supply, 2 – silicone stopper, 3 – SS inlet tube, 4 – furnace walls, 5 – molybdenum wire, 6 – SS spring and hook, 7 – quartz tube, and 8 – DBD ground).
Both streamer and contracted normal glow discharges were organized between two
pointed electrodes at 20 cm distance from the inlet. The furnace was not used for the
streamer discharge and contracted normal glow discharge experiments. A high voltage
nanosecond power supply (FID FPG 20-01NK10) was used as a power source for the
streamer discharge.
The contracted normal glow discharge geometry was similar to the streamer
geometry with the addition of a nozzle located 2 cm upstream from the tips of the
26
electrodes. The nozzle concentrated the H2S flow and provided better electrode cooling. The
glow discharge was powered by a high-voltage DC power supply (Universal Voltronics BRC
10,000). The schematic of streamer and glow discharge reactors is shown in Figure 11.
Figure 11: Schematic of the streamer and contracted normal glow discharge reactors (1 – power
supply, 2 – silicone stopper, 3 – electrodes, 4 – quartz tube, and 5 – Teflon® nozzle).
3.4 Results
The experimental results were obtained at varying flow rate, pressure, and
temperature conditions in the dissociation reactor. The experiments with corona and DBD
were focused on identifying and quantifying the effect that low power discharges with high
E/n have on H2S dissociation when these discharges are combined with elevated gas
temperatures (as usually in discharge physics, E/n is the ratio of electric field to gas
density). These experiments were designed to perform in such a way that the discharges by
themselves would cause measurable dissociation, but would not increase the gas
temperature considerably. The streamer and contracted normal glow discharge
experiments were focused on obtaining SER of the process where plasma is the only energy
source, and a wide range of power and flow rates were used to obtain a correspondingly
wide range of SEI. An attempt to find the optimal conditions for SER minimization was
made.
27
3.4.1 AC Corona Discharge
The experiments with corona discharge were performed at constant flow rate,
pressure, and power supply settings. The flow rate was set to 0.22 L/min (here as well as in
the following sections L/min refer to normal liters per minute at temperature 298 K and
pressure 1 atm). The power supply was set to its maximum frequency and voltage settings,
which corresponded to the discharge power of 2.1 W. The pressure was set to 265 Torr.
This regime was chosen in order to obtain noticeable and reproducible H2S dissociation of
0.5% at room temperature. The corresponding SEI was equal to 0.13 eV/molec, while SER
was 26.6 eV/molec.
The corona discharge operating in this regime did not provide noticeable gas
heating. Consequently, the furnace was used to heat the gas in the reactor to the desired
temperature. The furnace (Carbolite STF 16/610) allowed temperature variation in the
range 300 – 1200 K. Temperatures beyond 1200 K compromised reactor integrity by
melting the silicone stoppers and plastic tubing. The furnace temperature controller
provided stable operation at high temperatures (above 700 K) where the error would only
be on the level of 10 K or about 1%. At the lower temperatures the error was noticeably
higher, 25 – 50 K or up to 10%. Dependence of hydrogen concentration and SER of
hydrogen production on temperature in corona discharge and results of thermodynamic
equilibrium modeling for H2 concentration are shown on Figure 12 and Figure 13.
28
Figure 12: Comparison of measured hydrogen mole fraction in experiment and in thermodynamic
equilibrium modeling as function of temperature.
0
2
4
6
8
10
12
14
16
18
20
800 850 900 950 1000 1050 1100 1150 1200
H 2 Con
cent
ratio
n, [H
2] (m
ole
%)
Temperature, T (K)
w/ Corona
w/out Corona
Modeling
29
Figure 13: Full and relative SER of hydrogen production (b) as a function of gas temperature. The modeling was performed for the pressure of 0.25 atm under assumption of absolute quenching.
In the case of the corona experiment SER was calculated in two ways: relative
hydrogen production between the experiments with and without plasma, and full SER that
includes the heating energy of the furnace.
So, for each temperature setting relative SER, 𝜀𝑟:
And full SER, εf:
0
10
20
30
40
50
60
70
80
250 350 450 550 650 750 850 950 1050 1150
SER
(eV/
mol
ec)
Temperature (K)
Relative
Full
εr =𝐽
𝛼𝑝 − 𝛼𝑡 (21)
εf =Δ𝐻 + 𝐽𝛼𝑝
(22)
30
Here α is hydrogen concentration in a product flow for the case with plasma (αp) or
with the furnace only (αt), and Δ𝐻 is difference between gas enthalpies at the furnace and
room temperatures.
Summarizing results for AC corona discharge, at high temperatures hydrogen
production rate was below thermodynamic equilibrium; however, the experimental data
followed the thermodynamic equilibrium model under assumption of absolute quenching.
Hydrogen concentration difference between the experiments with corona and without
varied in the range from 0.2% to 1.7%, and corresponding SER of relative hydrogen
production in AC corona discharge also varied in a wide range 7.9 – 68.9 eV/molec at the
constant SEI of 0.13 eV/molec. A significant minimum of SER for the relative hydrogen
production is observed at the temperatures close to 1000 K. At this temperature range
corona accelerates the approach to chemical equilibrium, which may indicate a plasma
catalytic effect. It is worthwhile to study this effect in more detail in the future.
3.4.2 Dielectric Barrier Discharge
The experiments with DBD were performed at constant pressure, flow rate, and
power supply settings. The pressure was 187 Torr, and flow rate was 0.09 L/min. A
noticeable difference was observed in the results of DBD and corona experiments. The
hydrogen production in the DBD reactor was much higher at room temperature (18.5%).
The discharge filaments were few and probably hot. The exact temperature was not
measured, but the discharge heated up the surface of the reactor, which at that time was not
heated by the furnace, to approximately 60 – 70 °C, and inside the reactor within the area
covered by the outer electrode, the sulfur remained liquid (sulfur melting temperature is
388 K and boiling temperature is 717 K). The observed heating corresponded to the
significant increase of SEI (in comparison with corona), which was found to be 1.68
31
eV/molec. Dependence of hydrogen production in DBD on temperature and its comparison
with results of thermodynamic equilibrium modeling are shown on Figure 14. The
experiments with DBD positioned after the furnace did not differ significantly from those
obtained with DBD positioned in front of the furnace, and therefore, only the data from the
DBD positioned in front is presented.
Figure 14: Hydrogen production in DBD as a function of gas temperature and comparison with
results of thermodynamic equilibrium modeling under the assumption of absolute quenching. The modeling was performed for the pressure of 0.25 atm.
Summarizing DBD experimental results, dissociation of 18.5% of H2S was observed
at the room temperature. This level of dissociation, which is significantly above the
equilibrium value, remained virtually constant in the temperature range 300 – 700 K. A
reduction of dissociation was observed at temperatures above 700 K, which was likely
caused by the accelerated reverse reactions at higher temperatures. The SER of hydrogen
0
5
10
15
20
250 500 750 1000
H 2 Con
cent
ratio
n (m
ole
%)
Temperature (K)
Experiment
Modeling
32
production without additional gas heating was found to be 12 eV/molec at SEI of 1.68
eV/molec.
3.4.3 Streamer Discharge
Corona and DBD discharge, though different in their organization and SEI value,
have an important common feature that the following experiment with nanosecond
streamer discharge had to highlight. This common feature is low current and high 𝐸/𝑛 in
combination with overall low gas heating.
The streamer discharge was generated at different frequencies from 10 to 160 Hz
with 10 ns pulse of voltage (30 kV). The experiment was performed at constant flow rate of
0.15 L/min and constant pressure of 150 Torr. It was found that hydrogen production and
discharge power increased proportionally with the frequency (Figure 15) because energy of
each streamer was approximately constant, and as a result the SER remained virtually
constant and equal to 14 eV/molec.
33
Figure 15: Average power of the streamer discharge as a function of the discharge frequency.
0
2
4
6
8
10
0 50 100 150
Pow
er (W
)
Frequency (Hz)
34
Figure 16: Measured hydrogen concentration in the streamer discharge as a function of discharge
frequency.
3.4.4 Contracted Normal Glow Discharge
Completion of the experiments with DBD, corona, and streamer showed that the use
of discharges with high 𝐸/𝑛 and low (room) gas temperature is not an efficient approach for
hydrogen sulfide dissociation. The following experiments were designed to show how a
warm discharge [13-15] would perform at this task. This kind of discharge is much closer in
its properties to those that showed high efficiency of H2S dissociation [3, 6]. A warm
discharge is a non-equilibrium discharge with stepwise ionization, in which stable
concentration of electronically excited molecules is supported not only by electrons, but
also by gas temperature and/or by chemical processes. On the other hand, the gas
temperature of warm discharge is below the threshold of thermal ionization, where the role
of electric field in ionization becomes insignificant. Ionization in warm discharges has
0
1
2
3
4
5
6
0 50 100 150
H 2 Con
cent
ratio
n (m
ole
%)
Frequency (Hz)
35
comparable sensitivity to gas temperature as well as to 𝐸/𝑛 parameter. For H2S, the warm
discharge temperature range is between 1000 K and 3700 K.
The experiments with contracted normal glow discharge [13] were performed at
two flow rates of 1 and 2 L/min and constant pressure of 150 Torr. SEI of the process was
varied by varying the discharge current from 40 to 200 mA. The current remained stable
only above 40 mA. At the current of 200 mA and above the electrodes deteriorated quickly
and reactor parts overheated. The dependence of discharge voltage on current is presented
on Figure 17.
Figure 17: V-I characteristic of the contracted glow discharge.
Growing voltage-current characteristic is not typical for contracted glow discharges
[13], and in this particular case, it can be related to the growth of hydrogen concentration
and gas thermal conductivity. The gas temperature in the discharge channel was not
measured, and as it is one of the most important parameters, which would allow direct
500
600
700
800
0.04 0.05 0.06 0.07 0.08 0.09
Volta
ge (V
)
Current (A)
36
comparison with modeling results, it becomes a high priority objective for future
experiments. Taking into account discharge current and pressure, and comparing this
discharge with contracted glow discharges in different gases [15], we made preliminary
estimation that the gas temperature in the discharge should be in the range 2000 – 4000 K.
The elevated temperature in the discharge causes significant thermal dissociation,
but in this case hydrogen production is limited by slow quenching. The inefficient
quenching is illustrated on the Figure 18. The sufficiently fast quenching would maintain
the linearity of the hydrogen production curve, but instead it is leaning towards saturation.
Figure 18: Hydrogen concentration as function of discharge power in the contracted glow discharge
reactor.
The contracted glow discharge reactor in operation is shown in Figure 19 and
Figure 20. The figures illustrate intensive deposition of sulfur on cold walls of the reactor,
right outside of the discharge zone.
0
5
10
15
20
0 10 20 30 40 50 60
H 2 Con
cent
ratio
n (m
ole
%)
Power (W)
37
Figure 19: Contracted normal glow discharge reactor in operation (less then 1 min after start).
Figure 20: Contracted normal glow discharge reactor in operation (15 min after start).
A number of setting configurations were tested in experiments, and the
corresponding optimum setting was found at flow rate of 1 L/min and current of 50 mA.
The dependence of hydrogen production on discharge power at 1 L/min flow rate is shown
on Figure 18. The minimum SER corresponding to 50 mA current was 2.35 eV/molec and
the hydrogen production was 15.4%.
An attempt to increase H2S conversion was made and the reactor was modified to
include a quartz tunnel enclosing the plasma so that almost all hydrogen sulfide flow came
in contact with the discharge. Conversion rates of up to 60% were obtained but the SEI in
38
this system was rather high and the corresponding SER was not lower than in the system
without the tunnel.
3.5 Discussion
The experiments presented in this paper were designed to compare the influence of
different types of plasma discharges on dissociation of hydrogen sulfide, and they were not
intended to obtain the absolute minimum SER of hydrogen sulfide dissociation process. The
research goal was to obtain reproducible SEI and SER measurements using a variety of
plasmas in highly controllable conditions close to those in a plug flow reactor.
3.5.1 Corona Discharge
Literature research showed that there were no experiments done in the past with
AC corona discharge. Some work was done with pulsed corona, but if implemented
correctly, this discharge is the same as multiple streamer discharges, and therefore, these
works are compared with our streamer discharge experiments.
The dependence of H2 concentration on temperature, which is presented in Figure
12, can be explained, by low thermal dissociation kinetics at relatively low temperatures
(below 1100 K). However, in the case when the gas is heated to high temperature (close to
1200 K), it does not cool fast enough to prevent reverse reactions. An interesting
observation can be made from the Figure 13. At gas temperature of 970 K, SER of relative
hydrogen production is several times lower than at any other point. This result indicates
that corona discharge may have a catalytic effect on thermal dissociation, which is expected
in this temperature range. More experimental data is necessary to test this hypothesis and
to quantify this effect. It is also necessary to look at the chemical kinetics at this
temperature, and it would be advantageous to use some of the kinetic mechanisms
39
developed recently that have good agreement with experimental data in a close
temperature range [16-18].
3.5.2 Dielectric Barrier Discharge
There are few publications reporting the use of dielectric barrier discharge for H2S
dissociation, but nonetheless, one reported experiment had conditions that are very similar
to those presented in this paper [19]. The results that were reported in the literature are all
in the higher range of SEI 1 – 15 eV/molec, and corresponding SER was also quite high 37 –
106 eV/molec. The experimental results presented here were also obtained at rather high
SEI of 1.68 eV/molec, but the corresponding SER of 12 eV/molec was much lower than in
the experiments of Traus and Suhr [19]. This can be explained by the difference in pressure,
temperature and discharge frequency conditions in the experiments presented here and in
those presented by Traus and Suhr. The discharge of Traus and Suhr was arranged in a high
temperature zone with a discharge gap of only 1.5 mm, and in that case a high temperature
wall could catalyze the reverse reactions (compare with Figure 10).
The overall performance of the DBD discharge does not seem to be sufficient for
economically reasonable hydrogen sulfide dissociation, and the following should be noted:
DBD as well as corona has high E/n and low gas temperature. Even though it may be
possible to reduce SER by reducing SEI in the DBD reactor it still seems unlikely that this
discharge configuration could perform as well as higher temperature plasma discharges.
3.5.3 Streamer Discharge
Experiments with the streamer discharge were designed to test if cold discharges
with high E/n in a wide range of SEI will still have high SER until they transform into the
discharges with high gas temperature. It can be observed from Figure 16 that hydrogen
production grows linearly with specific energy input in the streamer discharge. This
40
indicates that electronic impact is the primary source of dissociation. A number of
mechanisms for hydrogen sulfide dissociation initiated by electronic impact have been
proposed, and some of them are presented in Table 1.
Table 3: Different mechanisms for electronic impact initiated dissociation of hydrogen sulfide.
Mechanism Source
Direct ionization of H2S followed by dissociative neutralization:
e + H2S → H2S+ + 2e H2S+ + e → HS + H
Ma et al., 2001 Helfritch et al., 1993
Ionization of the balance gas (M), leading to charge transfer reaction, and subsequent dissociative neutralization:
e + M → M+ + 2e M+ + H2S → H2S+ + M H2S+ + e → HS + H
Abolentsev et al., 1995
Direct electron collision dissociation of H2S:
e + H2S → HS + H + e
Electron collision dissociation or excitation of the balance gas, which produces active species that contribute to H2S dissociation:
e + M → M∗ + e M∗ + H2S → H + HS + M
Zhao et al., 2007
In general, the results of our experiments with the streamer discharge are
comparable with the SER of pulsed corona discharge reported by other research groups [20,
21]. No other research works were found discussing a single streamer for H2S dissociation,
which is mostly due to the fact that a single streamer system is not practical for any
applications. Only recent reports showed some pulsed corona results that are significantly
better than any other previously reported results for corona as well as many other
discharges at pressures of 1 atm and higher [22]. It is important to note that high
41
frequencies used in the experiments by John et al. [22] (when the best results were
obtained) may have led to formation of sparks or just heating of the discharge channels.
These assumptions are based on the fact that the discharge was visually much brighter than
more powerful corona units operating at atmospheric pressure [23]. The authors reported
that the discharge was not uniformly distributed along the high voltage wire anode, but was
concentrated in different places depending on the gas nature. Existence of these
“preferable” places means that the conductance did not drop completely between voltage
pulses, and this makes this discharge closer to a filamentary DBD or spark discharge. The
reported energy input could heat the gas in the reactor by 450 K, thus making it 750 K total.
This temperature alone cannot be responsible for the SER of 4.9 eV/molec (the best results
reported in [22]), however, if the energy input was not uniform, local temperature increase
could be much higher and in that case the reported low SER can be a result of local thermal
dissociation.
3.5.4 Contracted Normal Glow Discharge
The switch from streamer to a contracted glow discharge was performed without
changing reactor geometry or significant changes in reactor pressure, but the results were
very different. The reason for this is a change in gas temperature. Comparing the results
with thermodynamic expectations (see section 1.1), it can be concluded that in the hot
plasma channel hydrogen sulfide dissociates to the equilibrium concentration
corresponding to 2000 – 3000 K, and then once the mixture leaves the plasma zone, it starts
to recombine to hydrogen sulfide. There was no special cooling used to increase quenching
speed, and therefore the resulting concentration of hydrogen is noticeably lower than
expected in the case of absolute quenching. It is important to note that this experiment was
performed in the simplest geometry, which can be viewed as a plug flow reactor where the
42
heat is released in very short zone across the flow. This zone also does not cover the reactor
cross section fully but approximately 30% of it.
There was not much work done on H2S dissociation in glow discharges and the
results that were reported [24] are not particularly impressive. Traus et al. reported SER of
25 – 42 eV/molec with 23 – 32% conversion of H2S at low pressures and SER 25 – 36
eV/molec with 25 – 35% conversion of H2S atmospheric pressure. What is important is that
these reported values correspond to a very high SEI of more than 7 eV/molec. Most likely,
there were very high discharge energy losses to the reactor wall because of very short
discharge length.
A discharge that usually can be considered “warm,” and therefore is comparable
with the contracted normal glow discharge, is gliding arc. This type of discharge has been
successfully used for different kinds of fuel conversion and waste treatment [25, 26], and
has been used for H2S conversion as well. The SER results that were reported for gliding
arcs range from 500 eV/molec to 1.2 eV/molec [6, 27-29]. Gliding arc discharge is sensitive
to flow organization, and a review of the literature shows that the researchers often use
very different gas-dynamic approaches.
Assuming that the main dissociation path in warm discharges is thermal, then, the
SER of hydrogen production will largely depend on reactor configuration. Reactor
configuration can, in general, allow SER reduction because of energy recovery or
equilibrium shift due to high diffusion coefficient of hydrogen. It is necessary to remember
that our system was designed to imitate a plug flow reactor, which allows to separate the
effect of the process itself and possible effects related to gas-dynamic or diffusion effects.
43
3.5.5 Summary of the Results
The overview of the experimental results presented in this paper is summarized in
the form of SER as function of SEI (Figure 21).
Figure 21: SER as function of SEI for all the discharge configurations (including MW [3, 4, 30] and
GAT [6]) and thermodynamic equilibrium model under ideal quenching assumption.
The results are compared with thermodynamic equilibrium modeling and two of the
best sets of results reported earlier [5, 6]. Since all experiments were performed at low
pressure in the range from 150 Torr (contracted glow) to 265 Torr (corona), both GAT and
0.1
1
10
100
0 0.5 1 1.5 2 2.5
SER,
ε (e
V/m
olec
)
SEI, J (eV/molec)
Model @ 1 atm
Model @ 0.25 atm
Model @ 0.25 atm, 0.3 SEI
GAT @ 0.3 atm
MW @ 0.1 atm
Corona
Streamer
DBD
Glow
44
Microwave results were also reported for low pressure, the modeling curves are given for
low pressure as well. The second modeling curve corresponds to the low pressure and a
case when only 30% of the gas is treated by the plasma, which is probably a better
approximation of the experimental conditions in the contracted glow discharge.
The experimental points for corona, dielectric barrier, streamer, and glow
discharges all have 10% error for both SEI and SER. The major contribution to this error is
flow rate measurements. This error is avoidable and will be minimized in the future work.
Analysis showed that error of GAT experiments was larger than it was reported earlier [6],
and the lowest SER of hydrogen production in GAT reactor was 1.5 ± 0.3 eV/molec. The
error of the experiments with microwave discharge is unknown, but more information
about the experiments can found in papers [3, 4, 30, 31]. Flow organizations in the GAT and
microwave plasma reactors were rather complex. These flow organizations could result in
partial thermal energy recovery in the systems, which results in SER reduction below the
thermodynamic equilibrium level.
45
CHAPTER 4: CONCLUSIONS
The experiments on dissociation of hydrogen sulfide in a plug flow reactor with
assistance of corona, dielectric barrier, and streamer discharges showed that low
temperature discharges with high 𝐸/𝑛 and low gas temperature produce results
characterized by a common SER higher than 10 eV/molec. This corresponds to the reaction
mechanisms that are initiated by direct electronic impact. Significant reduction of SER at
low temperature conditions does not look possible because of the absence of chemical chain
processes.
The results obtained in contracted glow discharge at lower 𝐸/𝑛 have the lowest SER
of 2.35 eV/molec, which is close to the thermodynamic equilibrium value in the assumption
of absolute quenching. The SER is slightly higher than the equilibrium one and this can be
explained by multiple factors like discharge non-uniformity (see section 1.1), non-optimal
temperature, and reverse reactions taking place downstream of the discharge zone.
Experiments confirmed that an electrical discharge in H2S allows obtaining SER
close to that predicted by thermodynamic equilibrium modeling if the discharge has gas
temperature that is high enough for thermal dissociation. Further reduction of SER in a plug
flow reactor with thermal dissociation is not possible because of the absence of energy
recovery mechanisms.
Contracted glow discharge experiments led us to the conclusion that thermal
dissociation is the key factor in obtaining reasonably low SER for hydrogen production from
hydrogen sulfide. Electrical discharges, operating at gas temperature of 2000 – 4000 K,
which is optimal for hydrogen sulfide pyrolysis, create a preferred medium for plasma
dissociation of hydrogen sulfide. These conditions can be produced in the following
46
discharges: contracted normal glow, gliding arc, microwave, and low pressure radio
frequency discharge (RF) [30].
47
LIST OF REFERENCES
[1] Luinstra EA. Hydrogen from H2S: Technologies and Economics. Calgary: Sulfotech Research; 1995.
[2] Zaman J, Chakma A. Production of hydrogen and sulfur from hydrogen sulfide. Fuel Processing Technology. 1995;41:159-98.
[3] Bagautdinov AZ, Jivotov VK, Eremenko JI, Kalachev IA, Musinov SA, Potapkin BV, et al. Plasma chemical production of hydrogen from H2S-containing gases in MCW discharge. International Journal of Hydrogen Energy. 1995;20:193-5.
[4] Kasharin AV, Potapkin BV, Rusanov VD, Fridman AA. Energetics of plasma-chemical systems in selective transfer processes. Journal of Engineering Physics and Thermophysics. 1989;57:1335-43.
[5] Balebanov AV. Dissociation of Hydrogen Sulfide in a Plasma. Doklady Physical Chemistry, Proceedings of the Academy of Sciences of the USSR. 1986;283:709-12.
[6] Nunnally T, Gutsol KA, Rabinovich A, Fridman AA, Starikovsky A, Gutsol AF, et al. Dissociation of H2S in non-equilibrium gliding arc "tornado" discharge. International Journal of Hydrogen Energy. 2009;34:7618-25.
[7] Asisov RI, Vakar AK, Gutsol AF, Zhivotov VK, Krasheninnikov EG, Rusanov VD, et al. Plasma chemical methods of energy carrier production. International Journal of Hydrogen Energy. 1985;10:475-7.
[8] Fridman AA. Plasma chemistry: Cambridge University Press; 2008.
[9] Kasharin AB, Potapkin BV, Rusanov VD, Fridman AA. Obout energetics of of plasmachemical systems in selective nature of transport process. 1988.
48
[10] Steudel R. Topics in current chemistry. In: Steudel R, editor. Elemental sulfur and sulfur-rich compounds. Berlin: Springer-Verlag; 2003.
[11] Fridman AA, Kennedy LA. Plasma physics and engineering. New York, London: Taylor and Francis; 2004.
[12] Raizer YP. Gas discharge physics. Berlin Heidelberg: Springer-Verlag; 1991.
[13] Staack D, Farouk B, Gutsol AF, Fridman AA. Characterization of a DC atmospheric pressure normal glow discharge. Plasma Sources Science and Technology. 2005;14:700.
[14] Staack D, Farouk B, Gutsol AF, Fridman AA. Spectroscopic studies and rotational and vibrational temperature measurements of atmospheric pressure normal glow plasma discharges in air. Plasma sources science and technology. 2006;15:818–27.
[15] Staack D, Farouk B, Gutsol AF, Fridman AA. DC normal glow discharges in atmospheric pressure atomic and molecular gases. Plasma Sources Science and Technology. 2008;17:025013.
[16] Binoist M, Labégorre B, Monnet F, Clark PD, Dowling NI, Huang M, et al. Kinetic Study of the Pyrolysis of H2S. Industrial & Engineering Chemistry Research. 2003;42:3943-51.
[17] Shiina H, Oya M, Yamashita K, Miyoshi A, Matsui H. Kinetic Studies on the Pyrolysis of H2S. The Journal of Physical Chemistry. 1996;100:2136-40.
[18] Woiki D, Roth P. Kinetics of the High-Temperature H2S Decomposition. The Journal of Physical Chemistry. 1994;98:12958-63.
[19] Traus I, Suhr H. Hydrogen sulfide dissociation in ozonizer discharges and operation of ozonizers at elevated temperatures. Plasma Chemistry and Plasma Processing. 1992;12:275-85.
[20] Helfritch DJ. Pulsed corona discharge for hydrogen sulfide decomposition. Industry Applications, IEEE Transactions on. 1993;29:882-6.
[21] Zhao GB, John S, Zhang JJ, Hamann JC, Muknahallipatna SS, Legowski S, et al. Production of hydrogen and sulfur from hydrogen sulfide in a nonthermal-plasma pulsed corona discharge reactor. Chemical Engineering Science. 2007;62:2216-27.
49
[22] John S, Hamann JC, Muknahallipatna SS, Legowski S, Ackerman JF, Argyle MD. Energy efficiency of hydrogen sulfide decomposition in a pulsed corona discharge reactor. Chemical Engineering Science. 2009;64:4826-34.
[23] Tak G, Gutsol AF, Fridman AA. Pulsed Corona Plasma Pilot Plant for VOC Abatement in Industrial Streams as Mobile and Educational Laboratory. 17th International Symposium on Plasma Chemistry. Toronto, Canada2005. p. 1128-9.
[24] Traus I, Suhr H, Harry JE, Evans DR. Application of a rotating high-pressure glow discharge for the dissociation of hydrogen sulfide. Plasma Chemistry and Plasma Processing. 1993;13:77-91.
[25] Gallagher MJ, Geiger R, Polevich A, Rabinovich A, Gutsol A, Fridman A. On-board plasma-assisted conversion of heavy hydrocarbons into synthesis gas. Fuel. 2010;89:1187-92.
[26] Kalra CS, Gutsol AF, Fridman AA. Gliding arc discharges as a source of intermediate plasma for methane partial oxidation. IEEE transactions on plasma science. 2005;33.
[27] Czernichowski A. Gliding arc. Applications to engineering and environment control. 11th International Symposium on Plasma Chemistry. Loughborough, UK: Pure and Applied Chemistry; 1994. p. 1301-10.
[28] Czernichowski A, Czernichowski M, Wesolowska K. Glidarc-Plasma Reactors for Hydrogen Recovery from Waste H2S. HYPOTHESIS V. Porto Conte, Italy2003.
[29] Dalaine V, Cormier JM, Lefaucheux P. A gliding discharge applied to H2S destruction. Journal of Applied Physics. 1998;83:2435-41.
[30] Krasheninnikov EG, Rusanov VD, Sanyuk SV, Fridman AA. Dissociation of hydrogen sulfide in an RF discharge. Soviet Physics - Technical Physics. 1986;31:645-8.
[31] Bagautdinov AZ, Jivotov VK, Eremenko JI, Kalachev IA, Kozbagarov AI, Konstantinov EI, et al. Plasmachemical Conversion of Hydrogen Sulfide. 2nd International Symposium on Heat and Mass Transfer under Plasma Conditions. Antalya, Turkey1999.