Study of Plasma and Electrostatic Processes In

Home , Triboelectric effect

STUDY OF PLASMA AND ELECTROSTATIC

PROCESSES IN ENVIRONMENTALLY RELEVANT

PHENOMENA

JOSEPH ROBERT TOTH

Submitted in partial fulfillment of the degree requirements for the degree of

Doctor of Philosophy

Department of Chemical and Biomolecular Engineering

CASE WESTERN RESERVE UNIVERSITY

January, 2021 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis of

JOSEPH ROBERT TOTH

candidate for the degree of Doctor of Philosophy.*

Committee Chair

R. Mohan Sankaran

Committee Member

Julie Renner

Committee Member

Christian A. Zorman

Committee Member

Daniel Lacks

Date of Defence

August 3, 2020

*We also certify that written approval has been obtained for any proprietary material

contained therein. Table of Contents

Table of Contents ...... I

List of Tables ...... V

List of Figures ...... VI

Acknowledgements ...... XII

Abstract ...... 1

Chapter 1: Introduction ...... 2

1.1 Triboelectric charging of identical insulating particles ...... 2

1.2 Electrostatic effects on dust transport ...... 4

1.3 Plasma reactor design for methane conversion ...... 6

1.4 Hydrogen gas evolution at an electrified plasma-water interface ...... 9

1.5 Ammonia synthesis from a plasma-water system ...... 11

Chapter 2: Triboelectric charging of identical insulators ...... 14

2.1 Introduction ...... 14

2.2 Methods ...... 15

2.2.1 Experimental ...... 15

2.2.2 Modeling ...... 17

2.3 Results and discussion ...... 19

2.3.1 Experimental results ...... 19

2.3.2 Modeling results ...... 22

2.3.3 Discussion ...... 25

2.4 Conclusion ...... 27

Chapter 3: Electrostatic effects on dust transport ...... 28

3.1 Introduction ...... 28

3.2 Methods ...... 28

3.2.1 Experimental ...... 28

3.2.2 Modeling ...... 30

3.3 Results and discussion ...... 33

3.3.1 Laboratory-scale studies ...... 33

3.3.2 Modeling of field studies ...... 37

3.4 Conclusion ...... 43

Chapter 4: Direct, non-oxidative plasma conversion of methane ...... 45

4.1 Introduction ...... 45

4.2 Methods ...... 45

4.2.1 Experimental ...... 45

4.2.2 Modeling ...... 49

4.3 Experimental and plasma model characterization ...... 52

4.4 Results and discussion ...... 59

4.4.1 Results ...... 59

4.4.2 Discussion ...... 65

4.5 Conclusion ...... 67

Chapter 5: Hydrogen gas evolution at an electrified plasma-water interface ...... 69

5.1 Introduction ...... 69

5.2 Methods ...... 69

5.3 Results and discussion ...... 73

5.4 Conclusion ...... 85

Chapter 6: Continuous, process-intensified nitrogen fixation in a plasma-water droplet reactor ...... 86

6.1 Introduction ...... 86

6.2 Methods ...... 86

6.3 Results and discussion ...... 91

6.3.1 Reaction product characterization ...... 91

6.3.2 Control experiments for ammonia formation ...... 93

6.3.3 Energy Analysis ...... 96

6.3.4 Plasma and droplet characterization ...... 98

6.3.5 Insights into reaction chemistry ...... 105

6.4 Conclusions ...... 110

Chapter 7: Future work ...... 111

7.1 Triboelectric charging of identical insulators ...... 111

7.2 Electrostatic effects on dust transport ...... 111

III

7.3 Direct, non-oxidative plasma conversion of methane ...... 112

7.4 Hydrogen gas evolution at an electrified plasma-water interface ...... 112

7.5 Continuous, process-intensified nitrogen fixation in a plasma-water droplet

reactor ...... 112

References ...... 114

List of Tables

Table 5.1. Rate constants for the most favorable water vapor dissociation reactions by electron impact that produce H, presumably leading to the formation of H2 gas ...... 82

List of Figures

Figure 2.1. (a) Schematic illustration and (b) 3-D illustration of the humidity chamber containing the particle fountain, the electrostatic separator, and the bins ...... 16

Figure 2.3. Mass fraction of particles collected in the two side bins under the electrodes as a function of applied voltage to the electrodes for a low humidity of 5 % and a high humidity of 55 % ...... 21

Figure 2.4. Number fraction of large particles in sets that lose charge carriers and sets that gain charge carriers of the as a function of the number of collisions from a simulation of

1000 particles at the case of φΗ0= 0.015 ...... 23

Figure 2.5. Number fractions of large particles gaining and losing charge as a function of the fraction of initial carriers in H states, φΗ0 ...... 25

Figure 3.1. Experimental system schematic for studying dust transport through an electric

field ...... 29

Figure 3.2. Total fraction of particles collected in the top cup (FT) as a function of applied

electric field from the experimental system (black squares), the model (blue circles), and a

fit to the experimental results (red line) ...... 34

T T Figure 3.3. (a) fraction of total large particles, gL , and fraction of total small particles, gS ,

collected in the top cup as a function of applied electric field from the experimental system

(black squares), the model (blue circles), and a fit to the experimental results (red line) .35

T Figure 3.4. The mass fraction of the top cup that is large particles, fL , as a function of

applied electric field from the experimental system (grey squares), the model (blue circles), and a fit to the experimental results (red line) ...... 36

Figure 3.5. The fraction of particles removed by settling as they travel through the

atmosphere ...... 37

Figure 3.6. Average particle diameter as a function of elevation after 5.5 days of transport for various values of standard deviation of the product of ambient electric field and particle charge density, ...... 41

𝜎𝜎𝜎𝜎 Figure 3.7. Particle𝑠𝑠 size distribution within a 2000 m window after 5.5 days of transport

-3 with (red line, sσE=0) and without (black line, sσE=22 mC V m ) the involvement of electrostatic forces ...... 42

Figure 4.1. (a) Cross-sectional schematic of the confined dielectric barrier discharge reactor and (b) the reactor in the system used for characterization ...... 47

Figure 4.2. (a) A representative voltage (red) and charge (black) waveform for the confined

DBD formed in methane flow and (b) zoom-in of the charge waveform in (a) showing individual charge spikes ...... 54

Figure 4.3. (a) A representation of the real DBD reactor with periods of filamentary microdischarges that correspond to the spikes in the charge waveform and no microdischarges that correspond to the smooth region in the charge waveform and (b) a comparison following a volume element of gas (red outline) between the real reactor (left), a 0-D model from literature (center), and the steady-state model ...... 56

VII

Figure 4.4. A plasma reactor in an unconfined (a) and confined (b) system as power is

increased, and (c) power density versus power for the unconfined and confined

systems ...... 60

Figure 4.5. Simulation results for (a) conversion versus volume, (b) power versus volume,

and (c) conversion versus volume at equal powers ...... 61

Figure 4.6. Methane conversion versus power from experiments on confined DBD reactor

with volumes of 0.016 (black squares), 0.024 (red circles), 0.032 (green triangles), and

0.047 (blue diamonds) cm3, and from simulations with no fitting parameters (dashed black

line) and a scaled power density (solid red line) ...... 62

Figure 4.7. Plots of conversion versus power with linear fits for volumes of (a) 16 mm3,

(b) 24 mm3, (c) 32 mm3, and (d) 47 mm3 ...... 63

Figure 4.8. Methane conversion as a function of flow rate from experiments (black

squares) and simulations (red circles) results at ~1 W and 0.032 cm3 ...... 64

Figure 4.9. Methane conversion versus specific energy input from experiments on

confined DBD reactor with varying power and volumes of 0.016 (black squares), 0.024

(red circles), 0.032 (green triangles), and 0.047 (blue diamonds) cm3, constant power (1

W), and varying flow rate (open triangles), and from simulations with varying (scaled)

power (red line) and varying flow rate (open circles) ...... 67

Figure 5.1. Schematic diagram of experimental setup for measuring hydrogen production in plasma electrolytic reactor consisting of mass flow controllers (MFCs), direct current high voltage power supply, sealed reactor with plasma electrode and platinum counter-

electrode, and gas chromatograph...... 70

VIII

Figure 5.2. Representative current waveform for the plasma electrolytic reactor with a set

point current of 8 mA ...... 71

Figure 5.3. (a) Average H2 concentration exiting the reactor measured as a function of time for different currents in the plasma electrolytic reactor and (b) corresponding instantaneous faradaic efficiencies calculated at 30 minutes for different currents in both the plasma electrolytic and conventional electrolytic reactors ...... 74

Figure 5.4. Infrared images taken from the top view of the plasma electrolytic cell with source temperatures of (a) 80 °C and (b) 5 °C and solution surface temperatures as a function of (c) distance from the plasma electrode after operating for 30 min and (d) as a

function of time at the plasma electrode ...... 77

Figure 5.6. (a) Average H2 production as a function of time for different solution source

temperatures in the plasma electrolytic reactor and (b) corresponding instantaneous

faradaic efficiencies calculated at 30 minutes ...... 80

Figure 5.7. (a) Illustration of different regions in plasma-liquid system, including gas- phase, gas-liquid interface, and solution-phase, and key corresponding reactions that lead to formation of H2 gas, and (b) average H2 produced measured at different currents and

solution source temperatures at 30 minutes, separated into a faradaic and excess amount

that can be linked to the gas-phase (pink) and solution-phase (blue) reactions ...... 83

Figure 6.1. Schematic of the continuous, atmospheric-pressure plasma system studied for

reaction of nitrogen gas and water droplets ...... 87

- - Figure 6.2. Production rates of NH3, NO3 , and NO2 measured for the standard experiment at a power of 24 W ...... 92

Figure 6.3. Production rates of NH3 measured for the standard experiment (N2+H2O

droplets flowing through DBD reactor) and various controls ...... 93

Figure 6.4. Results from the thermodynamic equilibrium calculation for the reaction of

3 1 3 H O+ N → NH + O as a function of temperature ...... 94 2 2 2 2 3 4 2

Figure 6.5. Summary of production rates (black squares) and corresponding energy costs

(red circles) for NH3 synthesized from N2 and water droplets in atmospheric-pressure DBD

reactor as a function of power ...... 97

Figure 6.6. Representative (a) voltage (black), total charge (red), and (b) Lissajous waveforms for the DBD containing nitrogen and water droplets ...... 99

Figure 6.7. Optical emission spectroscopy (OES) of DBD with inlet feeds of (a) N2 and

water droplets and (b) Ar/N2 (90/10) and water droplets ...... 100

Figure 6.8. (a-d) Images of droplets exiting the reactor with (a) no plasma or heating, (b) pure N2 plasma, (c) 90 % Ar and 10 % N2 plasma and, (d) using the gas heater set to 40

°C. (e) NH3 production rate with (red) and without (blue) using the gas heater with both

pure N2 and 90 % Ar with 10 % N2 plasma ...... 101

Figure 6.9. Representative charge waveforms for DBD with inlet feeds of N2 + water droplets (black) and Ar/N2 (90/10) + water droplets (red) ...... 102

Figure 6.10. (a) Gas temperature as a function of power obtained from OES analysis of

nitrogen second positive system for DBD with inlet feeds of N2 + water droplets (black

X squares) and Ar/N2 + water droplets (red circles). Production rate of NH3 as a function of power for DBD with Ar/N2 (90/10) as inlet feed with (red circles) and without (black squares) preheating by gas heater off ...... 104

- Figure 6.11. Total production rates of fixed nitrogen consisting of NH3 (red), NO2 (green),

- and NO3 (blue) for N2 and H2O, H2, or O2 as the feeds ...... 106

Acknowledgements

I would like to thank the many collaborators that helped me with much of the work that went into this thesis: Mihai Bilici, Matti Murtomaa, Amber Phillips, Siddharth Rajupet,

Henry Squire, Blaire Volbers, Jun Zhou, Li Xie, Xiaozhou Shen, Ryan Hawtof, and Julie

Renner. I would also like to thank my advisors, Dan Lacks and Mohan Sankaran, for their extraordinary help and guidance.

XII

Study of Plasma and Electrostatic Processes in

Environmentally Relevant Phenomena

JOSEPH ROBERT TOTH

Abstract

Global climate change is becoming a critically important topic as temperatures continue to rise, reaching 1.5 °C over preindustrial levels within the next few decades, primarily due to anthropogenic activities.1 This drastic change will have devastating effects on the environment, including increasing temperature extremes, shifts in amounts of precipitation, raised sea levels, and ocean acidification.2 Thus, it is important to understand the various factors that contribute to climate change, as well as finding ways of reducing or eliminating emissions that are responsible for climate change. This thesis explores five areas where the environmental impact of a process is taken into consideration in order to understand and/or eliminate waste and emissions. In some of these areas, alternative solutions are offered in order to improve a process, while other areas look at the fundamental understanding behind different processes that contribute to climate change.

Chapter 1: Introduction

1.1 Triboelectric charging of identical insulating particles

In industrial processes, electrostatic charge on materials causes the materials to coat reactor

walls leading to inefficient operation of reactors and eventual shutdown.3 In dust storms,

generate electric fields greater than 100 kV m-1, which can in turn lift more particles into the atmosphere.4,5 These atmospheric particles interact with solar radiation, contributing to

global warming,6,7 and charge on the particles can lead to increases in their contribution

towards global warming.8 The process that causes this charge to build on particles occurs

when any two surfaces come into contact; the surfaces will exchange charged species and

become electrostatically charged in a process known as triboelectric charging. This process

can be amplified through frictional rubbing, by repeated contacts, or in particle systems

with a high surface area to volume ratio. While triboelectric charging itself is a poorly

understood process, the process has been observed extensively as in the areas described

above. The triboelectric effect is also a useful tool implemented in removing dust from

solar panels9–11 or in separating materials for recycling.12–14 In order to understand the full

impact that dust has on the atmosphere, how to prevent electrostatic shutdown of reactors,

or improve techniques utilizing triboelectric charging, a greater fundamental understanding

is necessary.

An interesting case of triboelectric charging is when the two materials that come

into contact are of the same chemical compositions and still develop electrostatic

charge.15,16 Without an apparent driving force, the surfaces would expectedly charge with either random polarities, or no net charge transferred. Contrary to this expectation, many

studies show that particles of the same composition will charge depending on their size,

with larger particles tending to charge positively, and small particles negatively.17–26

While the exact mechanism for triboelectric charging is still not well defined, a theory27,28 has been proposed to explain this size dependent charging, based on the model

of Lowell and Truscott.29 In this model, there is a distribution of mobile charged species

trapped in a higher energy states on the surface without a pathway to relax due to the

insulating properties of the material. When the material comes into contact with another

surface, this could provide a pathway for the carriers to find a lower energy state. If the

initial concentration of species trapped in high energy states is constant for a material, in a

group of particles, the larger particles will have more total of these species than the smaller particles, meaning that they can lose more charged species than smaller particles. Further, in a group of particles, larger particles will tend to lose charged species, and small particles will tend to gain charged species. If the charged species has a negative polarity (such as electrons or hydroxide ions), large particles will become more positively charged, and small particles will become more negatively charged.

Humidity has been shown to cause insulating particles to electrostatically act more like conducting particles30, but the role of humidity in triboelectric charging has been found

to both increase31,32 and decrease33,34 charging. Chapter 2 looks at supporting the non-

equilibrium effects and the role that humidity plays in the size-dependent charging of-

single component granular materials.35 Briefly, a fluidized bed charges particles through solely particle-to-particle contacts before they pass through an electrostatic separator, deflecting based on particle polarity. Experiments were performed at various humidifies showing that as humidity increased, the size dependence of charging decreased.

Simulations of particle interactions using the model from Lowell and Truscott29 shows that the change in initial concentration of species in high-energy states matches the results from the changing humidity. Understanding this model could be useful to improve triboelectric systems, such as more efficient electrostatic separation for recycling systems, and reduction of chemical waste and downtime in industrial processes.

1.2 Electrostatic effects on dust transport

The energy balance of incoming solar and outgoing thermal radiation is largely responsible for the climates across the Earth.36 Dust throughout the atmosphere can interact with both

the incoming and outgoing radiation directly through scattering and absorption6,7 or indirectly through cloud formation and biogeochemical feedback.37 The predominant

source of dust is typically more arid regions such as northern African and central Asia,

where dust is lifted into the atmosphere by wind.38,39 While dust is sourced in these regions,

it can travel far from the source, sometimes thousands of kilometers, affecting many

different regions on Earth.40 Anthropogenic factors account for 10 to 60 % of dust emission

primarily due to agricultural land use.41–43

The size distribution of airborne dust also has implications on the climate44 with

scattering and absorption of different radiation wavelengths depending on the size of

particles.45 The shorter wavelength light from solar radiation is primarily scattered away

from the earth by smaller particles, causing energy to be lost, while larger particles scatter

the longer wavelength thermal radiation back to the Earth.46 Larger particles are better at

absorbing both solar and thermal radiation, further trapping the energy on Earth.47 Thus, as

the distribution of airborne particles increases, more energy is trapped on earth, resulting

in warming of the climate.

Many previous studies looking at dust traveling through the atmosphere have found

interesting results for the particle size distribution of the dust.40,48–52 Typically, as dust

travels, gravity preferentially causes larger particles to settle out quicker, while smaller

particles remain suspended for longer periods of time. These studies have found larger

particles than expected from models being carried long distances. These larger particles

that remain lofted can have implications on the climate as discussed above. Several

mechanisms to explain these results have been proposed, including electrostatic forces

suspending particles.

As described in Section 1.1, particles in dust storms that come into contact with one another can become electrically charged, generating massive electric fields, even when the particles are of the same chemical composition.4,5 In dust systems, wind-blown dust

particles collide with one another, triboelectrically charging each other. While the total

charge of the particles may be small since there is little input of charge into the system,

individual dust particles may become highly charged, with large particles tending to charge

positively and small particles negatively.23 Coupling the size dependence of charging with

smaller particles being lofted higher than large particles, electric fields will be formed in

dust systems.

The electric field intensities found in dust storms are capable of lifting particles

from the ground.53,54 Field studies have shown a correlation between the density of dust

particles and the electric field strength present in a dust storm.55 This, supports the idea that

electric fields in dust storms lift particles, increasing the amount of dust a storm lofts into

the atmosphere.

The effects of electric fields on short range dust transport are fairly well

understood, but long range dust transport still remains a mystery. Since the particle density

is substantially lower, the electric fields generated from the dust may be much smaller.

Instead, background or fair weather electric fields may have an effect on the transport of

these particles. Chapter 3 looks at the effects of electric fields on particles that have already

been lofted.8 Experimentally, more particles are able to remain lofted when an electric field

is applied, and more larger particles are found at higher elevations. Simulations were

carried out incorporating the force on particles from the electric field, and they match the

experimental results. Further, this model is applied to previous long range dust transport

results to explain the presence of unexpected large particles.

1.3 Plasma reactor design for methane conversion

Methane is a widely used feedstock in the chemical industry for making hydrogen (H2) or

syngas in process known as steam methane reformation (SMR). H2 is typically used in the

Haber Bosch process to make ammonia (NH3) for fertilizer, while syngas, a mixture of H2

and carbon monoxide (CO), can be used in the Fischer-Tropsch process to make higher

order hydrocarbons such as ethylene and acetylene. In SMR, methane reacts with water,

56 yielding CO and H2 at high temperatures (800-1000 °C) over a nickel catalyst. Further

H2 is generated by reacting the CO with water in the water-gas shift reaction, giving H2 and

carbon dioxide (CO2), a greenhouse gas. SMR is also energy intensive, requiring 63 kJ/mol

56 of H2 produced typically obtained from burning natural gas. The oxidative conversion of

methane into hydrogen results in large amounts of carbon dioxide as a byproduct.

Instead of the oxidative conversion on methane, methane can be directly converted into hydrogen, carbon, and higher order hydrocarbons through the use of electrical discharges, or plasmas. Plasmas are formed when an electric field applied across a gas reaches the threshold to ionize the gas, forming a region of highly reactive and energetic species. These species are then able to undergo further chemical reactions, forming products that may not have been thermodynamically favorable at the gas temperature.

Plasmas have been used in many different gas conversion processes, including the conversion of ozone,57 the removal of unwanted pollutants,58 ammonia synthesis,59 and

methane conversion. There are many different types of plasma types and reactor designs

used for gas conversion; looking at direct non-oxidative methane conversion, different

types of plasmas used include thermal plasmas,60,61 coronas,62,63 and dielectric barrier

discharges (DBDs).64–68 Typically, in these reactors, the reactant gas flows through a

cylindrical vessel, occasionally filled with catalyst, reacting at steady state. Of the different

reactors, DBDs are typically used for their simple design, and the stability operating at

atmospheric pressure allowing higher throughput. DBDs consist of two electrodes, with at

least one covered with a dielectric layer and are formed when a high voltage is applied at

high frequency (~ 10 kHz) leading to cyclical microdischarges.

As is common in most steady state chemical systems,69,70 reactor parameters such

as the power, flow rate, and reactor volume have been varied in plasma system to find that

reactant conversion depends on reactor residence time.65,71–74 However, in DBD systems,

the relationship between power and residence time is complicated. Power and volume are

coupled, such that as the power is increased by increasing the applied voltage, the plasma

volume will also increase.75,76 Further, DBDs are highly non-uniform in both space and

time, with a wide range of length and time scales. The microdischarges that make up a

DBD have diameters on the order of 100 µm and lifetimes on the order of 10 ns, compared

to the residence time of reactors on the order of milliseconds to seconds and lengths of

millimeters to centimeters.77–80 These microdischarges occur stochastic positions, in both

space and time, that previous modeling approaches do not fully address.81–83 Due to all these complications, the independent of each volume, power, and flow rate on gas conversion in DBDs are not very well understood.

To understand the effects of each reactor parameter, care must be taken in setting up a system where each can be varied independently. Chapter 4 discuses a spatially- confined methane DBD reactor that decouples the effects of power and volume.84 Further,

it discusses a kinetic model that incorporates the spatial and temporal inhomogeneity of

the steady state reactor. From the experimental and modeled results, the reactant

conversion is found to be independent of the plasma volume, only depending on plasma

power and reactant flow rate, contrary to conventional thermal reactors. The conversion in

plasma systems is driven by the amount of energy added per amount of reactant, or

equivalently, the rate of energy (power) per the rate of reactant (flow rate). This relationship

is known, but the results found are contrary to previous findings that discuss the importance

of residence time to the conversion. These results have important implications for the

scaling of plasma processes, not only for methane conversion, but also for other areas

where plasmas may offer a key advantage.

1.4 Hydrogen gas evolution at an electrified plasma-water interface

Plasma electrolysis describes electrochemical systems where a gas discharge serves as at

least one of the electrodes and charge transfer reactions occur at the electrified plasma-

liquid interface. The discharge is formed either in the gas gap between a metal electrode

and liquid surface, referred to as glow discharge electrolysis (GDE), or inside the liquid

within a gas layer formed by electrolyte vaporization at the surface of a metal electrode,

referred to as contact glow discharge electrolysis (CGDE).85 These systems have found numerous applications, including: forming amino acids,86 nanomaterial synthesis,87–90

91 92 93 94 wastewater treatment, water disinfection, hydrogen generation, CO2 reduction, and ammonia synthesis.95 With such a broad range of applications, understanding the behaviors

and reactions of these systems is critical.

In aqueous electrochemical systems, the hydrogen evolution reaction (HER) is a

particularly important reduction reaction that competes with other desired reactions. One

measure of selectivity important for systems with charge-transfer reactions is the faradaic

efficiency, which has been assessed for the cathodic atmospheric-pressure microplasma-

96 + 90 94 95 driven reduction of ferricyanide, silver cations (Ag ), CO2, and N2 in aqueous

solution. The faradaic efficiency has been found to reach as high as 100% under certain

operating conditions, such as high reactant concentration and low current, suggesting that

the HER is negligible. On the other hand, at low reactant concentration and high current,

the HER has been found to become kinetically significant, as reflected by faradaic

+ efficiencies less than 10% for reduction of ferricyanide, Ag , CO2, and N2. While the

97 generation of hydrogen (H2) gas has been confirmed by isotope mass spectrometry, there have been no quantitative measurements of the HER to verify the mass balance.

In some CGDE98–102 and GDE103–105 systems, the faradaic efficiency has been shown to exceed 100%. In anodic GDE, gas phase ions bombard the liquid surface

103 dissociating water, leading to non-faradaic H2 generation. In cathode CGDE where the

electrode is in contact with the solution, excessive Joule heating leads the vaporization of

the electrolyte and secondary electrons emitted from the metal electrode that dissociate this

106,107 water vapor, resulting in non-faradaic H2 generation. The exact mechanisms of the non-faradaic H2 generation in cathodic GDE is still uncertain.

Chapter 5 discusses a cathodic GDE system, comprised of a DC atmospheric-

pressure microplasma formed in argon (Ar) contacting the surface of an aqueous solution

of dilute sulfuric acid.108 At the various currents studied, hydrogen in excess of that

predicted from charge transfer was observed, with the faradaic efficiencies greater than

100 %. A theory is proposed describing that there is evaporation of the electrolyte, leading

to water vapor appearing in the microplasma. This water vapor then engages in electron

impact dissociation, leading to the formation of hydrogen. When comparing the kinetics of

various plasma dissociation pathways, the predominant one is where the reaction does not

involve the consumption of electrons, meaning that the electrons are still available to

solvate into solution. Thus, hydrogen is generated in two mechanisms in this system: one

+ involving the faradaic reduction of hydronium (H3O ) in the solution phase, and the other, involving the non-faradaic electron-impact dissociation of water vapor in the plasma. As further support, varying the concentration of water vapor through heating and cooling the electrolyte shows that with higher water vapor contents, the amount of non-faradaic hydrogen increases. These results show that though electrolysis has been studied for

centuries, there is still room to improve or exceed the various limitations, such as the

faradaic efficiency.

1.5 Ammonia synthesis from a plasma-water system

As briefly mentioned in section 1.3, the primary use for H2 reacting with N2 for the

production of NH3 in the Haber-Bosch (HB) process at high temperatures, high pressures,

and with the use of solid metal catalysts.109,110 Ammonia is critically and most widely used for fertilizer but also used in pharmaceuticals, plastics, explosives, and fuels.111 HB is one

of the most energy intensive chemical processes, responsible for 1% of the entire global

energy consumption, and also environmentally impactful, responsible for 1 % of the global

112 CO2 emissions. The massive amount of CO2 released through this process is directly

linked to the generation of hydrogen that is used as a reactant in HB, typically through

56,113 SMR. From the reaction stoichiometry, for every ton of NH3 generated, one ton of

CO2 is also produced. Further, since natural gas is burned in order to provide the necessary

114 energy for the reaction, this number increases to 1.9 tons of CO2 per ton of ammonia.

Section 1.3 discusses using a non-oxidative process break ammonia, which could form H2

and Section 1.4 shows that electrolytic systems have generated hydrogen. While both of

these methods are able to bypass the CO2 made in SMR, HB is still a necessary process for

NH3 synthesis.

An alternative to HB that does not rely on H2 as a source instead uses water to react with N2 to form ammonia, bypassing the production of H2. Using water actually gives a

thermodynamic advantage in the one-step conversion process rather than the two-step

115 process of making hydrogen and then forming NH3. Different techniques have explored

this one-step method, including electrochemical116 and photochemical117 conversion.

Similar to HB, these methods rely on solid catalysts for the conversion, but they have only

118 obtained low yields, or low selectivity for NH3.

Alternatively to these methods, another method of reacting N2 and water is through

the use of plasma conversion. Some systems using plasma jets at the surface of a bath of

+ 119–122 an aqueous solution have been able to fix nitrogen into NH3 (or ammonium, NH4 ).

Further, reactant excitation by ultraviolet radiation has been shown to enhance the

production of ammonia.120–122 Through the generation of solvated electrons in a plasma-

95 electrolytic system, NH3 was formed with a faradaic efficiency reaching up to 100 %.

The addition of water vapor into the nitrogen stream prior to entering a plasma jet over a

solution was shown to increase the production rate, indicating that water vapor can also

59 -1 produce NH3. This study also found the lowest yet energy consumption of 23 MJ mol

for a system forming NH3 from N2 and water, yet this value is much larger than that of HB

(~0.5 MJ mol-1).

These plasma jet systems are batch reactors that rely on limited aqueous volumes

that are not ideal for industrial use, which prefers scalable continuous flow systems. Such

systems would need to flow the water through the plasma reactor, similarly to the inclusion

of the reported water vapor added to the nitrogen stream.59 Another possibility is to flow

nebulized water droplets through the plasma, which has the advantage of higher surface- to-volume ratio over the plasma jets processes. Droplets have also been found to alter reaction kinetics, which could offer an advantage to reactions occurring in the droplet-

plasma interface.123,124

Chapter 6 discusses a nitrogen DBD system that continuously processes nebulized

water droplets to form ammonia. Several control experiments were used to confirm the

significant presence of ammonia in the product stream. Droplet stability was assessed as

they passed thought the plasma, and found that the droplets fully evaporated upon exiting

the reactor, likely due to the highly filamentary behavior of the nitrogen DBD. Though the

droplets evaporated, NH3 was still formed, indicating that the reactions likely occurred in

- - the gas phase. Other nitrogen fixation products, nitrite (NO2 ) and nitrate (NO3 ), were also

- detected, with the reactor having a high selectivity toward NO3 . To build a better

understanding of the reaction chemistry, water was replaced with either H2 or oxygen (O2) gas finding that NH3 production is comparable between water and H2, but water is the best reactant for N2 fixation, with O2 fixing almost no nitrogen. This system offers a viable

alternative to the formation of sustainable ammonia from readily available feedstock

without direct CO2 generation.

Chapter 2: Triboelectric charging of identical insulators

[The contents of this chapter have been published: Toth, J. R.; Phillips, A. K.; Rajupet, S.;

Sankaran, R. M.; Lacks, D. J. Particle-Size-Dependent Triboelectric Charging in Single-

Component Granular Materials: Role of Humidity. Ind. Eng. Chem. Res. 2017, 56 (35),

9839–9845. https://doi.org/10.1021/acs.iecr.7b02328.]

2.1 Introduction

As described in Section 1.1, triboelectric charging is a poorly understood process, especially in the case where identical materials contact, transferring charge without an apparent driving force. The proposed non-equilibrium model by Lowell and Truscott

describes that charged species trapped in high energy states can relax to low energy states

on another material, even of the same chemical composition.29 This model has used to

explain the size-dependent charging in particle systems where large particles have a tendency to charge positively, and small particles negatively.26–28 This chapter looks at the

role humidity plays in this model for the size-dependent charging of chemically-identical insulating particles. First, the effects of humidity on the size-dependence of charging are experimentally determined. Then, simulations of particle contacts are used to relate humidity to the non-equilibrium model.

2.2 Methods

2.2.1 Experimental

The triboelectric charging and collection of particles in the system described here is based on a previous methodology using a particle fountain and electrostatic separator.26 A 200 g bed of particles was placed in an open cylinder over top a gas distributor plate, with five holes in the middle, filling the cylinder. Gas flowed through the holes, through the middle of the bed, forming a fountain of particles on top of the bed. The fountain was kept small enough that only the particles in the center were perturbed, causing the particles to only collide with one another, and not the cylinder or the distributor plate. After a period of time to allow the particles to charge, gas was blown orthogonally across the top of the fountain, pushing particles from the fountain to fall into the electrostatic separator. The electrostatic separator consisted of two rectangular 1.2 m deep and 15 cm wide electrodes with one at positive 6 kV and the other at negative 6 kV. The polarities of the right and left electrodes were alternated through various trials to ensure that there was no bias from the positioning or alignment of the electrodes. The electrode spacing started with a narrower gap of 2.5 cm at the top to make sure only particles with a more downward trajectory enter between the electrodes increasing to 20 cm at the bottom to ensure that the particles are adequately separated. The particles fell through the separator from the gravitational force, and migrated to the side opposite of their polarity due to the Coulombic force. At the bottom of the electrodes were 3 bins to collect the particles: one bin at by the positive electrode to collect negative particles (negative sample), one bin at the negative electrode to collect positive particles (positive sample), and one bin (three boxes) in the middle collecting mostly neutral particles (neutral sample).

Figure 2.1. (a) Schematic illustration and (b) 3-D illustration of the humidity chamber containing the particle fountain, the electrostatic separator, and the bins.

This system was placed inside an environmentally controlled chamber in order to perform experiments at various values of relative humidity (RH), as shown in Fig. 2.1. The humidity was varied by flowing a stream of dry air (0 % RH) and a stream of wet air (>90

% RH) at various ratios through the chamber. The dry air was used from a tank (Airgas,

Industrial grade dry air) and the wet air came from bubbling a different stream of the dry air through heated water and a condenser at room temperature. The condenser was used to

prevent condensation of the wet stream inside the reactor. The humidity was determined

using a humidity meter (Fisher Scientific™ Traceable™ Humidity Meter), and a fan was placed inside the chamber to ensure mixing. Humidity was tested at various locations inside the chamber to ensure homogeneous mixing. The air used to form the fountain is at the

same humidity level as the chamber.

The particles in the fountain were composed of soda lime glass (Dragonite) and had

diameters between 300 and 800 µm. Glass particles were used due to their hydrophilic

nature, readily absorbing water from the atmosphere. Particles were charged for 5 minutes

in the fountain, until they reached steady state charge, before being blown into the

separator. After the particles are collected in the bins, each bin is sieved with a 600 µm

sieve to determine the amount of ‘large’ and ‘small’ particles in each bin.

2.2.2 Modeling

The basis of this model is that mobile charge carriers on the surface of an insulating

material can exist in various energy states.27–29 Typically, the carriers would like to relax

to the lowest state, but due to the insulating nature of the material, they are trapped in their

higher energy state. Instead of on the same material, when the surface contacts another

surface, there may be a new pathway to an unoccupied state on the new surface that could

let the charge carrier relax. This leads to size-dependent charging, since the larger particles

have a greater number of total charge carriers in high energy states on their surface

(assuming equal densities of species in each state across all surfaces). Since larger particles

have more total carriers, they will lose more carriers when they relax to lower level states.

While the exact species that constitutes the charge carrier is unknown,16 the model is still

applicable.125

This model consists of a collection of N particles with each ith particle having a

radius, Ri. Charge carriers on the surface of each particle resides in one of two states: high-

th energy (H), or low-energy (L). At any time, t, the i particle has niH(t) charge carries in

th high-energy states and niL(t) charge carriers in low-energy states. An i particle can collides

with a jth particle, with the probability of the collision proportional to the collisional cross

2 section, (Ri + Rj) . For each collision, there is a probability that a charge carrier transfers

from state α on the ith particle to state β on the jth given by,

= (2.1) 𝑛𝑛𝑖𝑖𝑖𝑖 2 𝑝𝑝𝑖𝑖𝑖𝑖→𝑗𝑗𝑗𝑗 𝑘𝑘𝛼𝛼𝛼𝛼 4𝜋𝜋𝑅𝑅𝑖𝑖 where kαβ is the state-to-state rate constant and α=H or L, β=L or H.

The rate constant is expected to be highest for transitions from H to L, while being completely negligible for transitions from L to H. There is a possibility that transfers could happen across the same state, and hypothesize that this rate constant is low, but non zero.

Thus, rate contants of kHL=0.5, kLH=0, kHH=kLL=0.05 were used for the transitions. Since the exact values of the rate constants were arbitrarily chosen, the results will not be quantitative, but the qualitative results should be independent of the exact values.

Monte Carlo simulations were carried out on this system to understand the behavior

via the following procedure repeated for Ncollisions times:

1. Particle i was randomly chosen from the entire set of particles.

2. Particle j was chosen from the remaining particles by choosing a random

number, r, that falls between 0 and + . If r is less than 2 ∑𝑗𝑗≠𝑖𝑖�𝑅𝑅𝑖𝑖 𝑅𝑅𝑗𝑗�

( + ) where k is each small particle, then the jth particle is also small, 2 𝑘𝑘≠𝑖𝑖 𝑖𝑖 𝑘𝑘 else,∑ the𝑅𝑅 jth 𝑅𝑅particle is large.

3. The probability of charge carrier transfer is calculated for each of the eight ways

that a transfer can occur is calculated, and a random number between 0 and 1

for each transfer, , is calculated. If the < , then a single 𝑖𝑖𝑖𝑖→𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖→𝑗𝑗𝑗𝑗 𝑖𝑖𝑖𝑖→𝑗𝑗𝑗𝑗 carrier is transferred𝑟𝑟 from state α on particle i to𝑟𝑟 state β on𝑝𝑝 particle j.

These simulations were carried out for N=1000 particles with half of the particles

6 having Ri=1 and the other half have Ri=0.5 for up to Ncollisions=50×10 . Initially, each

particle had identical surface densities of carriers in high-energy states, ρΗ0 and ρL0, giving

the number of initial charge carriers to be (0) = 4 . The sum of the initial surface 2 𝑖𝑖𝑖𝑖 𝑖𝑖 𝛼𝛼0 densities was held constant at 2000/π, while𝑛𝑛 the initial𝜋𝜋𝑅𝑅 fractional𝜌𝜌 density of carriers in the

high-energy state, φΗ0=ρΗ0/(ρΗ0 + ρL0), was varied.

2.3 Results and discussion

2.3.1 Experimental results

The RH in the chamber was varied from 2 to 58 % to determine the effects of humidity on

the size-dependence of charging from particle-to-particle interactions. Positive, negative, and neutral samples were collected in each trial and the fraction of large and small particles in each were determined. Figure 2.2a shows the results of each sample at 6 different bins of humidity. Each point refers to a mean of at least 10 data points, with the standard error.

At the lower humidity of 0-10 %, the positive sample is enriched in large particles, and the negative sample is enriched in small particles. This agrees with the previous research

discussed in Section 1.1, where large particles have a tendency to charge positively, and

small particles negatively.17–26

As the humidity increases though, the apparent size dependence of charging starts

to diminish. This is shown in Fig. 2.2b, with the difference between the large particle

fraction in the positive and negative samples. From 10 to 50 %, the difference dropped well

below the results from 0 to 10 %, but still shows significant size dependence of charging.

When the humidity goes even further up to 50-60 %, the difference is not significant (two sample t-test, P-56%), and thus, the size dependence of charging diminishes.

Figure 2.2. (a) Experimentally-determined mass fraction of large particles in each sample and (b) difference between positive and negative sample large particle mass fraction as a function of chamber relative humidity. The reference sample refers to the total mass fraction of large particles in all bins. Each data point refers to the average of numerous trials with the standard error reported.

Figure 2.3. Mass fraction of particles collected in the two side bins under the electrodes as

a function of applied voltage to the electrodes for a low humidity of 5 % and a high

humidity of 55 %.

There are two possible explanations for the results that show diminishing size-dependence of charging at high humidity: one, that the particles remain electrostatically charged, but with no size-dependent charging, or two, that the particles lose their charge in the high humidity. Thus, a set of experiments was performed to determine the explanation. These

experiments looked at the total mass deflected into the side bins as a function of applied

voltage. Figure 2.3 shows the results as a function of voltage applied to each electrode for

both high and low humidity. As the voltage increases, the total mass collected also

increases for both humidities. This behavior is due to the fact that the system is comprised

of a distribution of charged particles. The stronger electric fields are able to pull more

charged particles to the outside bins. Further, since these experimental results for both

humidities are roughly the same, this would indicate that the extent of charging is equal at

21 low and high humidity. Thus, while humidity does not have a strong effect on the size dependence of charging, this is not due to the humidity simply causing particles to lose their charge.

2.3.2 Modeling results

The simulation is first set up with the fraction of the density of charge carriers in H states,

φΗ0= 0.015 and the fraction φL0=0.985 charge carriers in the L state. Each of the particles start electrically neutral and have the same surface density of charge carriers. The larger particles will have a larger total number of charge carriers, both niH(0) and niL(0). The particles then collide with each other and exchange charge with the probabilities described by Equation 2.1. As the number of collisions increase, two sets of particles develop, those that have a net loss of charge carriers, and those that have a net gain of charge carriers. The results for these two sets as a function of the number of collisions is shown in Fig. 2.4. The particles that lose charge carriers are primarily composed of large particles, while the particles that gain charge carriers are mostly composed of small particles. Initially, the magnitude increases very rapidly, but reaches a peak before more slowly dropping back towards a fraction of 0.5.

Figure 2.4. Number fraction of large particles in sets that lose charge carriers and sets that

gain charge carriers of the as a function of the number of collisions from a simulation of

1000 particles at the case of φΗ0= 0.015.

These results can be understood by carefully considering the two different types of charge carrier transfers.

1. HL transitions. These transfers are entirely irreversible since the rate constant

for LH transfers is equal to zero. This transfer holds the key to the dynamics

in the system, but to understand its role, we first consider two particles, a large

and small particle, undergoing two collisions between each other. In the first

collision, Equation 1 gives equal probability for transfer to occur between each

particle, since , is merely the initial density of carriers in H states, which is 𝑛𝑛𝑖𝑖𝑖𝑖 𝑜𝑜 2 4𝜋𝜋𝑅𝑅𝑖𝑖 equal for all particles. If this transfer does occur for both particles, then the

second collision is where things start to change. Now, the density of carriers in

H states becomes , = , , which is greater for the large particle 𝑛𝑛𝑖𝑖𝑖𝑖 𝑜𝑜−1 𝑛𝑛𝑖𝑖𝑖𝑖 𝑜𝑜 1 2 2 2 4𝜋𝜋𝑅𝑅𝑖𝑖 4𝜋𝜋𝑅𝑅𝑖𝑖 − 4𝜋𝜋𝑅𝑅𝑖𝑖 than for the smaller particle. Thus, during the second collision, it is more likely

for the large particle than a small particle to lose a charge carrier in the H state.

As a result, the small particle will accumulate charge carriers while the large

particles will lose them.

2. HH and LL transitions. These transitions are reversible and will thus bring

the system back towards the equilibrium of equal charge carrier density.

The HL transitions dominate at lower collision numbers due the higher rate constant for transfer, but as the number of H states decreases, turning into L states, the LL transitions

start to take over. This is seen in Fig. 2.4, where initially HL transitions cause small

particles to predominantly lose carriers, thus making up the majority of particle that had

lost carriers, with analogous results for the large particles gaining carriers. Therefore, HL transitions are primarily responsible for the size dependence of particles losing or gaining charge. Once the LL transitions dominate, the system tends toward equilibrium, where all particles have the same carrier density, equal to the initial density, though all these carriers are in the L state. Triboelectric charging is a non-equilibrium phenomenon, where there is a separation of charge across the two surfaces; thus, it is likely that triboelectric charging typically exists at an intermediate point, where there is non-equilibrium in the densities of charge carriers.

Simulations were also carried out varying φΗ0. Figure 2.5 shows the simulated results of the number fraction of large particles that lose and gain species averaged between

40 to 50 million collisions from φΗ0=0.005 to zero. Initially, there is a significant size

dependence between particles losing and gaining species, but as φΗ0 decreases, the

difference between the fractions of particles losing and gaining carriers shrinks. Since the

HL transitions are responsible for the size dependence, it is logical that removing the

initial carriers in the H states would also diminish this size dependence.

Figure 2.5. Number fractions of large particles gaining and losing charge as a function of the fraction of initial carriers in H states, φΗ0. These values are averaged from results

between 40 and 50 million collisions with the standard errors. The x-axis is reversed to

emphasize the reduction of size-dependent charging as a function of φΗ0.

In this system, if the charge carrier is negative, these simulated results would

indicate that large particles tend to charge positively, and small particles tend to charge

negatively, consistent with the experimental results and previous work.17–26

2.3.3 Discussion

At low humidity, the experimental results show that particle systems exhibit a particle size

dependence in triboelectric charging such that small particles tend to charge negative and

large particles tend to charge positive. Further, this particle size dependence decreases as

humidity increases, such that at high humidity there is no difference in the charging

tendency of large and small particles.

The physical basis of this phenomena resides in the non-equilibrium behavior of

the charge carriers on the surface. Initially, the distribution of charge carriers in high energy

states are unable to equilibrate to low energy states on the surface due to the insulating

nature of the materials having low charge mobility. At high humidity, water absorbs on to

the surface of the material forming a conductive layer. A previous study has found that

nylon spheres sitting on electrodes will act as insulators at low humidity, but conductors at high humidity.30 This conducting layer on the surface can now allow some of the charge carries trapped in high energy states to relax to low energy states. Thus, higher humidity corresponds to a lower value of φΗ0 from the non-equilibrium model.

Since φΗ0 has a relationship to the humidity, the experimental and simulated results

can be compared. Figure 2.2 experimentally shows that at low humidity, small particles

charge negatively and large particles charge positively with this behavior disappearing at

high humidity. Similarly, the simulated results in Fig. 2.5 shows that at high φΗ0

(corresponding to low humidity), small particles accumulate charge and large particles lose

charge carriers. As φΗ0 decreases (corresponding to increasing humidity), this effect starts

to disappear. If the charge carrier has a negative polarity, the experiments and simulations

show identical trends.

2.4 Conclusion

This chapter provides support for the idea that non-equilibrium effects play an important

role in triboelectric charging of particle systems. The non-equilibrium effects, which are

greatest at low humidity, lead to the particle-size dependence of charge polarity. The non- equilibrium effects become smaller at high humidity because humidity creates a conductive water layer on the surface that enables charge carriers to relax to equilibrium states. Thus, as humidity increases, the particle-size dependence becomes smaller and ultimately becomes negligible. From these results, the humidity effect will be most significant for hydrophilic materials, which have significant water adsorption; for hydrophobic materials, this humidity effect will be smaller (less dependence of ϕH0 on humidity), and the particle-

size dependence will persist to high humidity.

Chapter 3: Electrostatic effects on dust transport

[The contents of this chapter have been published: Toth, J. R.; Rajupet, S.; Squire, H.;

Volbers, B.; Zhou, J.; Xie, L.; Sankaran, R. M.; Lacks, D. J. Electrostatic Forces Alter

Particle Size Distributions in Atmospheric Dust. Atmos. Chem. Phys. 2020, 20 (5), 3181–

3190. https://doi.org/10.5194/acp-20-3181-2020.]

3.1 Introduction

Section 1.2 discusses the effects that dust particles traveling through the atmosphere can have an impact on the global climate. These particles can be electrostatically charged through triboelectric charging at the source of the dust. Since electric fields are capable of lifting particles,54,55 not only can they lift more particles into the atmosphere at the source, but they can keep particles suspended as they are transported through the atmosphere. This chapter discusses the effects that electric field has on dust transport, primarily focusing on how the fields can affect the size distribution of dust. Further, the laboratory scale results are fit to a model that is then used to fit to previous field results. These field studies have found larger particle than predicted with models excluding electric field, but the inclusion of these fields may explain the discrepancy.52

3.2 Methods

3.2.1 Experimental

The system shown in Fig. 3.1 was constructed to study the effects of an electric field on dust transport after lift-off. The system consisted of two horizontal parallel electrodes (89 cm long, 15 cm wide) connected to a DC high-voltage power supply (HB-Z303-1AC)

oriented such that the electric field between the electrodes was perpendicular to the ground

surface. The distance between the electrodes was constant at 12 cm, and the electric field

was varied from 125 to -125 kV m-1. The polarity of the electric field was defined by the

polarity of the top electrode; i.e., a positive electric field had a positive top electrode and a

negative electric field had a negative top electrode. In order to prevent particle contacts with the electrodes and ensure that the sand particles contacted only other sand particles, the electrodes were covered with a thin layer of sand held in place by two-sided tape.

Figure 3.1. Experimental system schematic for studying dust transport through an electric field. The fan lifts dust from the sand bed and blows it between the electrodes, into the collection cups.

A 250 g sand bed that was 1.5 cm tall, 20 cm long, and 11 cm wide was positioned

4 cm upstream of the electrodes. The sand was polydisperse and characterized by a mean diameter of ~132 µm, 40% by mass with diameters smaller than 105 µm, and 60% by mass with diameters between 105 and 450 µm. Particles with diameters greater than 450 µm were removed by sieving the sand before placing into the sand bed. A fan placed 15 cm upstream of the sand bed was used to blow sand through the electrodes. The average air

speed over the sand bed was 6.7 m s-1, and decreased to 3.2 m s-1 at the end of the electrodes. As it is difficult to control exactly when particles become charged, the particles likely had an initial charge in the sand bed due to particle contacts during sample preparation and could acquire more charge during lift-off and transport between the electrodes. Each trial ran for 2 minutes, consuming approximately 80% by mass of the sand bed.

Two cups were placed 11 cm downstream the electrodes with slit openings 2.2 cm tall and 11 cm wide in order to collect the particles. The bottom cup collected particles between heights of 0 and 2.2 cm, and the top cup collected particles between heights of 8.8 and 11 cm. The total mass collected in both cups in the experiments ranged from 8 to 35 g, which represents approximately 21% of the mass passed through the system by the fan.

The particles collected in the two cups were sieved with a 105 µm mesh sieve with “small particles” defined as those that passed through the sieve and “large particles” were defined as those that did not pass through. The irregular shapes of the sand particles may cause some particles to be collected in a sieve size different than its nominal size; for example, a long thin particle may fall through though it is fairly large. After sieving the particles in each cup, the mass of the small and large particles collected in each cup was determined,

, where i designates the particle size (i=S for small particles, i=L for large particles) and 𝑗𝑗 𝑖𝑖 𝑚𝑚j designates the cup position (j=T for top, j=B for bottom).

3.2.2 Modeling

Monte Carlo simulations were used to address the effects of electric fields on dust transport in both the experimental system as well as previous field studies of dust

transported long distances in the atmosphere. The simulations addressed the final positions

of wind-blown particles under various conditions. The particles were assumed to be spheres

of diameter, D, with density, ρ, and surface charge density, σ and travel at the velocity of

the wind. The forces on the particles in the horizontal direction were not considered. Thus,

the only forces acting on the particles were assumed to be gravitational, electrostatic, and

drag forces in the vertical direction. The gravitational force, FG, is given by,

= 3 (3.1) 𝜋𝜋𝐷𝐷 𝐹𝐹𝐺𝐺 6 𝜌𝜌𝜌𝜌 where g is the acceleration due to gravity. The electrostatic force, FE, is given by,

= (3.2) 2 𝐸𝐸 where E is the electric field strength.𝐹𝐹 The 𝜋𝜋drag𝐷𝐷 𝜎𝜎𝜎𝜎force, which always acts in the direction

opposite to vertical velocity, is given by,

= (3.3) 1 2 2 𝐷𝐷 8 𝐷𝐷 𝑓𝑓 where is the density of air, v𝐹𝐹 is the− velocity𝐶𝐶 𝜌𝜌 𝜋𝜋 of𝐷𝐷 the𝑣𝑣 𝑣𝑣�particle in the vertical direction, and

𝑓𝑓 is the𝜌𝜌 direction of motion of the particle. For the drag coefficient, , the following

𝐷𝐷 correlation𝑣𝑣� was used,126 𝐶𝐶

. = (1 + 0.15 . ) + (3.4) 24 0 681 0 407 𝐷𝐷 8710 𝐶𝐶 𝑅𝑅𝑅𝑅 𝑅𝑅𝑒𝑒 1+ 𝑅𝑅𝑅𝑅 where Re is the Reynolds number,

| | = (3.5) 𝜌𝜌𝑓𝑓 𝑣𝑣 𝐷𝐷 𝑅𝑅𝑅𝑅 𝜂𝜂 and is the viscosity of air. The drag and electrostatic forces can act upwards or

downwards𝜂𝜂 depending on the direction of particle velocity and the sign of , respectively.

The net force on a particle, Fnet, is the sum of FG, FE, and FD. 𝜎𝜎𝜎𝜎

The equations governing the motion of the particle in the vertical direction are

= (3.6) 𝑑𝑑𝑑𝑑 6 3 𝑑𝑑𝑑𝑑 �𝜋𝜋𝐷𝐷 𝜌𝜌� 𝐹𝐹𝑛𝑛𝑛𝑛𝑛𝑛 and

= (3.7) 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 where y is the position of the particle in the vertical𝑣𝑣 direction.

Particle trajectories are determined by numerically integrating Eqns. (3.6) and (3.7).

At sufficiently long times, the system attains a constant terminal velocity with Fnet=0. Thus,

steady-state particle trajectories can be obtained by calculating the terminal velocity, rather

than by integrating the equations of motion.

Since particles traverse the electrodes more quickly (18 ms) than terminal velocity

can be reached in the experimental system, particle trajectories were obtained by numerical

integration of the equations of motion. In the horizontal direction, particles move at the

average wind speed in the system (4.95 m s-1). In the vertical direction, particles start with size-dependent initial vertical velocities, at an initial height of zero. A collection of particles with density, = 2600 kg m-3 (characteristic of quartz), had the particle diameter, charge, and initial vertical𝜌𝜌 velocity, v0, chosen randomly from normal distributions

characterized by respective means, µ, and standard deviations, s. For particle diameter,

= 132 µm and = 60 µm (as in the experiments). For surface charge density, =

𝐷𝐷 𝐷𝐷 𝜎𝜎 𝜇𝜇0 and = , where𝑠𝑠 α is a fitting parameter. For initial vertical velocity, = 0 and 𝜇𝜇 =

𝜎𝜎 𝑣𝑣0 𝑣𝑣0 / ,𝑠𝑠 which𝛼𝛼 depends on particle diameter, with β being a fitting parameter𝜇𝜇 (physically,𝑠𝑠 2 this𝛽𝛽 𝐷𝐷 relationship causes smaller particles to have higher initial velocities); only positive initial velocities were sampled from the distribution. Particle trajectories were calculated using Euler’s method with a time step of 18 µs. After traveling the length of the electrode

(0.89 m), particles with heights between 0 and 6 cm were considered to be collected in the

bottom cup, and those with heights greater than 6 cm were considered to be collected in

the top cup. Particles with heights less than 0 cm were not considered to be collected in the

collection cups.

For long-range dust transport, particles reach terminal velocity, and thus a simpler

method can be used to determine particle trajectories. The dust was assumed to be already

lofted and following the model of Maring et al.,48 particles were uniformly distributed in a vertical 2000 m window, with initial velocities equal to their terminal velocities. A

collection of particles of density = 2600 kg m-3 was used, and the particle diameter and

charge were chosen based on the𝜌𝜌 study being modeled. The terminal velocity was found

by numerically solving for the velocity at which = 0. The final height of each particle

𝑛𝑛𝑛𝑛𝑛𝑛 after traveling for a given amount of time was obtained𝐹𝐹 simply by calculating the initial

height plus the terminal velocity multiplied by time.

3.3 Results and discussion

3.3.1 Laboratory-scale studies

In the experimental system, as particles move through the electric field, their final elevation

can be influenced by the applied electric field. This effect is quantified using the fraction

of total collected particles in the top cup, , where 𝑇𝑇 𝐹𝐹 = 𝑇𝑇 𝑇𝑇 . (3.8) 𝑇𝑇 𝑚𝑚𝐿𝐿 +𝑚𝑚𝑠𝑠 𝑇𝑇 𝑇𝑇 𝐵𝐵 𝐵𝐵 𝐹𝐹 𝑚𝑚𝐿𝐿 +𝑚𝑚𝑠𝑠 +𝑚𝑚𝐿𝐿 +𝑚𝑚𝑠𝑠 Figure 3.2 shows the results for as a function of applied electric field. As the magnitude 𝑇𝑇 of the electric field increases, 𝐹𝐹more of the airborne dust remains suspended higher, as indicated by the increase in the fraction of particles collected in the top cup. Note that this enhancement of dust elevation occurs with either polarity of electric field.

Figure 3.2. Total fraction of particles collected in the top cup (FT) as a function of applied

electric field from the experimental system (black squares), the model (blue circles), and a

fit to the experimental results (red line).

As described in Chapter 2, particle charging can have a size dependence, with larger

particles charging positively and smaller particles negatively. Quantifying this effect in this system is more complicated, since gravity is not symmetric here, causing large particles to preferentially stay toward the bottom of the system. In order to understand the effect of the electric field, we first need to look at the particle size dependence of the electric field effect described above. Thus, Fig. 3.3a shows results for the fraction of large particles that are collected in the top cup rather than the bottom cup, , where 𝑇𝑇 𝑔𝑔𝐿𝐿 = 𝑇𝑇 (3.9) 𝑚𝑚𝐿𝐿 𝑇𝑇 𝑇𝑇 𝐵𝐵 𝑔𝑔𝐿𝐿 𝑚𝑚𝐿𝐿 +𝑚𝑚𝐿𝐿 as a function of applied electric field. Likewise, Fig. 3.3b shows results for the fraction of

small particles that are collected in the top cup rather than the bottom cup, , where 𝑇𝑇 𝑔𝑔𝑆𝑆

T T Figure 3.3. (a) fraction of total large particles, gL , and fraction of total small particles, gS , collected in the top cup as a function of applied electric field from the experimental system

(black squares), the model (blue circles), and a fit to the experimental results (red line).

= 𝑇𝑇 . (3.10) 𝑚𝑚𝑆𝑆 𝑇𝑇 𝑇𝑇 𝐵𝐵 𝑔𝑔𝑆𝑆 𝑚𝑚𝑆𝑆 +𝑚𝑚𝑆𝑆 The values and show the fraction of large and small particles, respectively, 𝑇𝑇 𝑇𝑇 𝐿𝐿 𝑆𝑆 in the system that 𝑔𝑔are collected𝑔𝑔 in the top cup. When no electric field is applied, is 𝑇𝑇 𝑆𝑆 greater than meaning that a bigger fraction of total small particles are in the top𝑔𝑔 cup 𝑇𝑇 𝐿𝐿 than of the large𝑔𝑔 particles in the top cup. This is due to gravitational effects which cause smaller particles to remain suspended higher than larger particles. As the magnitude of the electric field increases, large and small particles both remain suspended higher, as shown by the increase in the fractions of large and small particles in the top cup.

Now, looking at the shift in the size distribution in the cups, the fraction of particles in the top cup that are large, , is defined as 𝑇𝑇 𝑓𝑓𝐿𝐿

T Figure 3.4. The mass fraction of the top cup that is large particles, fL , as a function of

applied electric field from the experimental system (grey squares), the model (blue circles),

and a fit to the experimental results (red line).

= 𝑇𝑇 . (3.11) 𝑚𝑚𝐿𝐿 𝑇𝑇 𝑇𝑇 𝑇𝑇 𝑓𝑓𝐿𝐿 𝑚𝑚𝐿𝐿 +𝑚𝑚𝑆𝑆 While was obtained directly from the experimental data, these results are especially 𝑇𝑇 𝐿𝐿 susceptible𝑓𝑓 to noise, as the overall ratio of large to small particles collected varies between trials. This could be due to the fact that the initial particle size distribution could change over time as small particles are preferentially lost from the system. A better estimate of 𝑇𝑇 𝐿𝐿 can be obtained by calculating it independently, using variables that are less susceptible𝑓𝑓 to

noise, with

= 𝑇𝑇 , (3.12) 𝑇𝑇 𝑔𝑔𝐿𝐿 𝐹𝐹𝐿𝐿 𝑇𝑇 𝑓𝑓𝐿𝐿 𝐹𝐹 where FL, the fraction of particles in both cups that are large, is defined as

= 𝑇𝑇 𝐵𝐵 , (3.14) 𝑚𝑚𝐿𝐿 +𝑚𝑚𝐿𝐿 𝑇𝑇 𝑇𝑇 𝐵𝐵 𝐵𝐵 𝐹𝐹𝐿𝐿 𝑚𝑚𝐿𝐿 +𝑚𝑚𝑠𝑠 +𝑚𝑚𝐿𝐿 +𝑚𝑚𝑠𝑠 36

and come from fits to the experimental data, and the average value of across all 𝑇𝑇 𝑇𝑇 𝐿𝐿 𝐿𝐿 trials𝑔𝑔 is 𝐹𝐹0.63 ± 0.01. The results for as a function of electric field are shown𝐹𝐹 in Fig. 𝑇𝑇 𝐿𝐿 3.4, showing that increasing the electric𝑓𝑓 field causes the particle size distribution at higher

elevations to be enriched in larger particles.

Simulations were carried out on the model of the experimental system. The fitting

parameters were varied until the model showed a reasonable fit to the experimental results,

when = 1.4 µC m-2 and = 11000 µm2. The surface charge densities characterized by

= 1𝛼𝛼.4 µC m-2 are reasonable,𝛽𝛽 as they are similar to results found experimentally in glass

𝛼𝛼particle systems,24 and they are significantly smaller than surface charge densities

measured for triboelectrically charged quartz.127,128 Results were obtained for trajectories

of 106 particles. As seen in Figs. 3.2-3.4, the model results match the experimental results

reasonably well. In both the model and experiments, the electric field maintains particles at higher elevations and shifts the size distribution at higher elevations towards large particles. While the distributions for particle size and charge density are not perfectly representative of the particles in the experimental system, the trends in the model will be the same regardless of what distributions are used.

3.3.2 Modeling of field studies

Since the effect of dust on the climate is dependent on particle size, to accurately gauge the role of atmospheric dust on the climate, the size distribution of airborne dust must be known. The results from the laboratory system indicate that electric fields can alter the size distribution of transported dust. As mentioned in the Introduction in Section 1.2, previous field studies have found surprising results regarding particle size distributions of dust

transported far from the source. Here, the model is applied to address whether these effects

might be due to electrostatic forces on long-range dust transport.

Maring et al. compared the size distribution of aerosol samples collected in Izaña

near the dust emission source with Puerto Rico after a transport time of 5.5 days from the

source.48 They modeled a 2000 m tall column of air typical of the Saharan Air Layer over

the North Atlantic and Caribbean that has no vertical mixing and an initially uniform size

distribution. When only taking into account gravitational and Stokes drag forces, their

model predicted more large particles settled out than found in the transported aerosol

samples. In contrast, their model showed agreement when they modified the terminal

velocity with an arbitrary upward-contribution for all particles.

Using the model, the displacement of different-sized particles after transport for 5.5 days at terminal velocity was calculated to determine the fraction of particles removed due to settling using similar assumptions as Maring et al.,48 but explicitly including electrostatic forces rather than an arbitrary upward velocity component. To simplify, this model

assumed that all the particles had the same polarity and magnitude of charge. As shown in

Eq. 3.2, the electrostatic force on each particle depends on the particle surface charge density and ambient electric field strength. The actual surface charge density of lofted dust

particles is unknown, but typical triboelectric charging values range over several orders of

magnitude, from ~1 µC m-2,129–131 to ~10 µC m-2,128 and up to ~100 µC m-2.127,132–136 The maximum electrostatic charge on the surface of a material prior to gas breakdown was found by Matsuyama to be approximately 400 µC m-2 for a 10 µm particle.137 The

magnitude of the electric field is also difficult to determine and can range over multiple

orders of magnitude, with values up to 200 V m-1 in fair weather electric fields,138–140 up to

15 to 150 kV m-1 in dust storms,5,141–144 and up to 500 kV m-1 during thunderstorms.145

Since both the charge density and the electric field can vary by orders of magnitude, the product between the two, , is considered to be the important parameter for determining the effects of the electrostatic𝜎𝜎𝜎𝜎 forces. Given the range of possible electric fields and charge densities discussed above, the value of can range from 0.2 mC V m-3 to 200 C V m-3; however, particles can only sustain a 𝜎𝜎𝜎𝜎 on the lower end of this range where there are more consistent electric fields and more𝜎𝜎𝜎𝜎 prevalent charge densities. As shown in Fig. 3.5, when = 38 mC V m-3, the model results match well with the experimental data of

Maring𝜎𝜎𝜎𝜎 et al.48

Figure 3.5. The fraction of particles removed by settling as they travel through the atmosphere. The dashed blue line takes into account gravitational and drag forces, the red line takes into account gravitational, drag, and upward electrostatic forces, and the black points are field data collected from Maring et al.48

Other field studies also found large particles to be enriched at higher elevations in

comparison to models. Reid et al. used a particle measuring system mounted to an aircraft to measure dust size distributions in Puerto Rico and found that the ratio of large to small particles did not depend strongly on elevation.49 Furthermore, lidar studies of Saharan dust

over Barbados (i.e., far from the source) suggested little variation in the particle size

distribution between 1 and 4 km elevation, based on the near-constant value of the depolarization ratio and the similarity of the 355 nm, 534 nm, and 1085 nm laser results in this elevation range.50

Further simulations were carried out to investigate whether electric fields and the

resulting electrostatic force on particles can lead to this enrichment of large particles at

higher elevations. Particle diameters were based on a lognormal size distribution

characteristic of atmospheric dust with = 1.5 µm and = 0.75 µm.49 The products of

𝐷𝐷 𝐷𝐷 electric field and particle charge density𝜇𝜇 were based on a normal𝑠𝑠 distribution with = 0

𝜎𝜎𝜎𝜎 and = 0, 11, and 22 mC V m-3 such that half the particles experienced an electrostatic𝜇𝜇

𝜎𝜎𝜎𝜎 force𝑠𝑠 in the upward direction, and half in a downward direction. Simulations were carried

out for 105 particles. As shown in Fig. 3.6, in the case of no electrostatic forces, the average

particle diameter is significantly larger at low elevations than at high elevations, in

agreement with current models. In comparison, with = 22 mC V m-3, the average

𝜎𝜎𝜎𝜎 particle diameter becomes more constant with changing𝑠𝑠 elevation.

Figure 3.6. Average particle diameter as a function of elevation after 5.5 days of transport

for various values of standard deviation of the product of ambient electric field and particle

charge density, .

𝜎𝜎𝜎𝜎 𝑠𝑠

Electrostatic forces act on charged particles such that those of one polarity are lifted

to higher elevations while those of the opposite polarity fall to lower elevations. As a result,

the elevation distribution of particles is stretched out and becomes more uniform. This

leveling of the elevation distribution occurs for all sized particles, such that both large and

small particles are more uniformly vertically distributed, causing the size distribution of

particles to become more constant with changing elevation.

Several field studies have found larger particles transported than predicted when

only taking into account gravitational and drag forces.40,51,52 Some studies have found

particles with diameters greater than 40 µm and ranging to 450 µm transported long

distances.51,52 In the modeled system, if electrostatic forces are neglected, any particle with a diameter greater than 7.3 µm would settle out of the 2000 m column within 5.5 days.

However, from Fig. 3.7, which represents the size distribution of lofted particles,

electrostatic forces cause several particles larger than 7.3 µm to remain lofted, up to 17.6

µm. Even with the consideration of electrostatic forces, it is not expected that these much

larger particles (40 to 450 µm) would be lofted. Thus, it is likely that other forces contribute

to the transport of large particles such as fast horizontal wind speeds, turbulence, and uplift

in convective systems.52

Figure 3.7. Particle size distribution within a 2000 m window after 5.5 days of transport

-3 with (red line, sσE=0) and without (black line, sσE=22 mC V m ) the involvement of electrostatic forces.

For electrostatic forces to account for these deviations in size distributions of atmospheric dust particles, must be on the order of ~10 mC V m-3. Field studies of

atmospheric dust have not measured𝜎𝜎𝜎𝜎 surface charge densities of dust particles to confirm

whether particles have sufficient charge for electrostatic effects to be significant; though

volumetric charge densities have been measured,146 but cannot be translated to surface

charge densities, which may be much larger than volumetric charge since it is the net charge

of a distribution of negatively and positively charged particles. However, surface charge

densities on the order of ~10 to ~100 µC m-2, which could cause significant electrostatic

effects on transport, are routinely observed in laboratory settings.127,128,132–136 While

charges on particles may decay through gas neutralization, experiments have found

surfaces retaining charge even after days of exposure to ambient conditions147–149 and

charged dust has been observed in the atmosphere even after being transported for long

times.51,146,150 Moreover, subsequent particle collisions during transport could further

charge particles through the triboelectric effect.

Additionally, while electric fields up to 200 V m-1 are naturally occurring in the

atmosphere138–140 and likely persist throughout transport, local electric fields within dust

layers may be much higher due to the presence of dust particles and separation of positively

and negatively charged particles.151,152 Thus, to better gauge the extent electrostatic forces influence dust transport, future field studies should investigate characteristic particle surface charge densities and local electric fields strengths within dust layers which can persist throughout long-distance dust transport.

3.4 Conclusion

This chapter discussed experiments that characterized the size distribution as a function of

height for sand blown between electrodes with vertically oriented electric fields and a

model for this system considering the gravitational, electrostatic, and drag forces on particles. The experimental and modeling results indicate that electrostatic forces maintain

particles at higher elevations and increase the concentration of larger particles at higher

elevations. The model was extended to long-distance dust transport and found that sufficient electrostatic forces suspend large particles that would otherwise settle out during transport, thereby increasing the concentration of large particles in atmospheric dust. Since large particles have a heating effect due to absorption and scattering of radiation, the results suggest that electrostatic forces could contribute to the warming of the climate. In addition, sufficient electrostatic forces may explain unexpected size and elevation distributions of atmospheric dust.

Chapter 4: Direct, non-oxidative plasma conversion of methane

[The contents of this chapter have been published: Toth, J. R.; Shen, X.; Lacks, D. J.;

Sankaran, R. M. Reaction Conversion for a Plasma-Based Steady-State Flow Process Is

Independent of Reactor Volume. Ind. Eng. Chem. Res. 2018, 57 (18), 6048–6056.

https://doi.org/10.1021/acs.iecr.7b05091.]

4.1 Introduction

An alternative to the high temperature oxidative conversion of methane is the use of a DBD

to directly convert methane into hydrogen and other higher order hydrocarbons.64–68 In

order to utilize DBDs for this industrial process, it is necessary to scale up the reactors

through common scaling factors, such as residence time. As discussed in section 1.3, the

relationship between power and residence time is complicated in plasma processes, where

the power applied can change the plasma volume. Thus, this chapter uses an experimental

system that has decoupled power and volume in order to study each independently. Further,

the complicated behavior of DBDs is accounted for in a steady-state kinetic model in order to verify the experimental results.

4.2 Methods

4.2.1 Experimental

In order to independently study power, volume, and flow rate, a novel spatially-confined

DBD plasma reactor was designed that eliminates plasma expansion with increased power, schematically shown in Fig. 4.1a. The reactor converts methane by a plasma-assisted process in the gas phase, in the absence of catalyst or gas heating and is characterized by

the system shown in Fig. 4.1b. The reactor was constructed from a quartz tube with an outer diameter (OD) of 6.3 mm and an inner diameter (ID) of 2.0 mm. Copper wires were

wrapped around the outer surface of the quartz tube and electrically-biased as the high

voltage electrode, a tungsten wire (OD= 0.25 mm) attached to a rod inside the quartz tube

was the ground electrode, and the quartz tube served as the dielectric barrier. The plasma

formed between the tungsten wire and the quartz tube wall. To prevent discharges on the

outer surface of the quartz tube, the outer electrode was covered with a dielectric sealant.

The plasma was confined by two additional grounded stainless steel (SS)

components: a SS tube (OD= 2.0 mm, ID= 1.3 mm) fit snugly within the quartz tube on

the upstream side, and a SS rod (OD= 1.98 mm) with a small gap between the quartz tube

on the downstream side. The plasma volume was adjusted by moving the SS rod using a

micrometer from 0 to 20 mm allowing volume control from 0 to 50 mm3. The gas flow

passed through the SS tube into the plasma region, and then out of plasma region through

the gap between the SS rod and the quartz tube.

Figure 4.1. (a) Cross-sectional schematic of the confined dielectric barrier discharge reactor and (b) the reactor in the system used for characterization. Gas flows from the mass flow controller to the reactor, in through the grounded tube, through the plasma region (blue), out around the grounded rod, and then into the gas chromatograph for characterization. The oscilloscope measures the voltage (V) and the charge (Q).

To create the plasma, a sinusoidally changing voltage was applied across the electrodes, with a frequency of 34 kHz and an amplitude up to 25 kV (Information

Unlimited, Model PVM500). A capacitor (45.71 nF) placed in series between the ground electrode and the supply ground was used to determine the total charge passing through the plasma, qt. The charge and voltage waveforms were measured by an oscilloscope (Hewlett

Packard 54616C) and recorded using a LabVIEW virtual instrument. The displacement

charge, qd, was estimated as the total charge at voltages sufficiently low to avoid plasma formation, fitting a line to the data, and extrapolating the line to the voltage where the plasma was ignited (this linear trend can be assumed since a DBD can be approximated with an electrical circuit composed of resistors and capacitors153). The charge consumed by the plasma, qp, was then obtained from the difference between qt and qd.

The plasma power, Pplasma, was varied by changing the applied voltage, and using

standard Lissajous analysis, which plots qt and qd versus voltage to determine the value of

Pplasma. The qt and qd values form loops over a voltage cycle, and the energy per cycle

corresponds to the area of the qt loop minus the area of the qd loop. The plasma power was

obtained by dividing the energy per cycle by the time of a cycle. Real-time construction

and analysis of the Lissajous plots was carried out using the LabVIEW interface, enabling

a priori setting and measuring the power for a given run. Calculating power during plasma

operation gave the ability to average the power over many cycles (approximately 1600), resulting in a standard error of approximately 0.1 W for a typical trial. The confined reactor was able to operate with powers ranging from 0.2-3 W, corresponding to average power densities in the reactor of 2-100 W/cm3 (due to the inhomogeneous nature of the plasma,

the power density in the localized microdischarges will be higher). The plasma became less

stable at larger volumes due to the increased frequency of microdischarges, leading to a

higher uncertainty in Pplasma.

The reactor was run at steady-state by introducing a constant flow rate of pure

methane (4.3 ml/min) through the confined plasma volume and applying a constant Pplasma.

The reactor effluent was analyzed by injecting a 100 µL sample from a continuously filled

sample loop into a gas chromatograph (GC, Shimadzu, Model GC-2014) equipped with a

flame ionization detector (FID) and a thermal conductivity detector (TCD). The sample

was injected after 30 minutes of plasma operation in order to ensure a steady state

concentration from the reactor within the sample loop. The concentration of the methane

exiting the reactor was determined from calibration curves for methane in the GC. The methane conversion in the system is then defined as,

= 1 , (4.1) 𝑛𝑛̇ 𝐶𝐶𝐻𝐻4 𝑜𝑜𝑜𝑜𝑜𝑜, 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 − 𝑛𝑛̇ 𝐶𝐶𝐻𝐻4 𝑖𝑖𝑖𝑖 where is the molar flow rate.

𝑛𝑛̇

4.2.2 Modeling

Kinetic simulations of methane conversion in the DBD reactor were carried out using the

same conditions as the experiments and assuming a tubular plug flow reactor (PFR). For

an ideal PFR, the equation that governs the molar concentration of species i as a function

of position and time, ( , ), is

𝑛𝑛�𝑖𝑖 𝑧𝑧 𝑡𝑡 ( , ) ( , ) + + ( , ) = (4.2) 𝑑𝑑𝑛𝑛�𝑖𝑖 𝑧𝑧 𝑡𝑡 𝑉𝑉̇ 𝑑𝑑𝑛𝑛�𝑖𝑖 𝑧𝑧 𝑡𝑡 1 𝑑𝑑𝑉𝑉̇ 𝑑𝑑𝑑𝑑 𝐴𝐴 𝑑𝑑𝑑𝑑 𝐴𝐴 𝑛𝑛�𝑖𝑖 𝑧𝑧 𝑡𝑡 𝑑𝑑𝑑𝑑 ∑𝑗𝑗 𝑟𝑟𝑖𝑖𝑖𝑖 where t is time, z is the axial distance along the reactor, is the gas volumetric flow rate,

A is the axial cross sectional area of the reactor and rij is𝑉𝑉̇ the molar rate of production of

species i in elementary reaction j. The rij depend on rate coefficients, ( ), as well as the

𝑗𝑗 concentrations of the species involved in the reactions, and , and 𝑘𝑘are𝑇𝑇 of the form

𝑘𝑘 𝑙𝑙 = ( ) 𝑐𝑐 𝑐𝑐 (4.3)

𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑗𝑗 𝑘𝑘 𝑙𝑙 where vji is the stoichiometric coefficient𝑟𝑟 for𝜐𝜐 𝑘𝑘species𝑇𝑇 𝑐𝑐 i𝑐𝑐 in reaction j (the example given here is for a second order reaction where component i is a product). The reaction network from a previous study by De Bie et al. was used, which included 36 different species (9 neutral

molecules, 17 ionic molecules, and 10 radicals), 135 electron–molecule reactions and 232

molecule–molecule reactions.154

Rate coefficients for the molecule-molecule reactions were obtained from De Bie

et al.,154 which were in turn compiled from other references.155–164 Since the background

gas temperature was kept constant at room temperature, there was no temperature

dependence included for the molecule-molecule reaction rates; previous studies found that

the temperature change was minimal,79 which was also supported by experimental

measurements.

Rate coefficients for electron-molecule reactions were calculated as follows. First,

the electron behavior was simulated in a given reduced electric field, E/N (where E is the

electric field and N is the concentration of neutral particles or in this case molecules), to determine the electron energy distribution function (EEDF). From the EEDF, average

electron temperature and rate constants for the electron-molecule reactions were obtained.

This analysis was repeated at different values of E/N varied from 0 to 300 Td to obtain a

correlation between the average electron temperature and the rate constants. For the reactor

simulation, instead of explicitly considering E/N, the average plasma power input to the system was specified, which alters the EEDF (characterized by the average electron temperature) via an energy balance. In this way, the rate constants were obtained for the electron-molecule reactions from the applied plasma power (as measured experimentally).

This procedure is described in more detail below.

Rate coefficients for electron-molecule reactions were determined as follows,

( ) = ( ) ( ) (4.4) 𝑒𝑒 ∞ 𝑘𝑘𝑗𝑗 𝑇𝑇𝑒𝑒 �2𝑚𝑚 ∫0 𝑓𝑓 𝜀𝜀 𝜎𝜎 𝜀𝜀 𝑑𝑑𝑑𝑑

where Te is the average electron temperature, e/m is the charge to mass ratio of an electron,

ε is the electron energy, f(ε) is the electron energy distribution function (EEDF), and σ(ε)

is the energy-dependent collisional cross section. The EEDF and average electron

temperature were obtained at various values of E/N by solving the Boltzmann equation, which takes the relevant collision cross-sections as input. The results for the electron temperature and reaction rates at various E/N were used to fit the reaction rates as functions of electron temperature. The solution of the Boltzmann equation, electron temperature, and reaction rates were carried out with the BOLSIG+ software package (Version 12/2017).165

Most of the energy-dependent cross sections were obtained from Janev and coworkers,166–

170 but momentum transfer and electron-hydrogen collisions were obtained from LXcat

databases.171

The electron density was analyzed as follows. The molar concentration of electrons as a function of position and time, ne(z,t), was determined from Equation 4.2, in the same

way that the molar concentrations of molecular species were determined. The initial (pre-

plasma) electron density is small but non-zero, as is the case in real systems where there is

a small, but non-zero concentration of free electrons initially in the gas phase; the non-zero

concentration of free electrons in an ambient gas is a necessary factor for forming a plasma

(both in the model and in real systems). The average electron temperature, Te, is a function

of position and time, and was determined from an energy balance,

( , ) ( , ) = + (4.5) 𝑑𝑑𝑇𝑇𝑒𝑒 𝑧𝑧 𝑡𝑡 𝑛𝑛𝑒𝑒 𝑧𝑧 𝑡𝑡 𝐶𝐶𝑒𝑒 𝑑𝑑𝑑𝑑 𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑜𝑜𝑜𝑜ℎ𝑒𝑒𝑒𝑒 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 where Ce is the molar heat capacity of the electrons ( = , where R is the gas constant). 3 𝑒𝑒 2 The other terms (omitted in equation for simplicity)𝐶𝐶 include𝑅𝑅 the energy needed to

thermalize newly formed electrons to the electron temperature, elastic collisional losses,

inelastic collisional losses, etc.

Overall, the simulations were carried out by first assuming a population of

electrons, described by energies with a correspondingly low rate constant. Applying and

increasing power to the system resulted in a shift of the initial EEDF and an increase in Te

(Equation 4.5). Formation of a discharge further shifted the EEDF, in particular increasing

the electron density. Ultimately, the EEDF impacts the electron impact rate constants

(Equation 4.4), which in turn increases the molar rates of reaction (Equation 4.3). Many of these reactions produce additional electrons; thus, this series of events initiates a chain reaction that leads to a steady-state plasma with a relatively high concentration of electrons.

These equations were solved to obtain the molar concentrations of all species using the

CHEMKIN software package.172

A DBD plasma occurs as a collection of localized and transient microdischarges

rather than as a continuous and uniform plasma. The strategy for including the

microdischarges in the simulations was guided by the experimental characterization, and

thus, discussed below.

4.3 Experimental and plasma model characterization

Power in a DBD plasma is often dissipated through filamentary microdischarges that occur

only during certain windows of the voltage cycle and do not fill the entire volume. This

process is apparent in the voltage and charge waveform results, shown in Fig. 4.2a, where

the spikes in the charge waveform correspond to microdischarge and occur only at certain

times in the AC cycle – specifically, at the rising and falling edges of the cycle. A close-

up, shown in Fig. 4.2b, allows closer examination of the charge spikes. The length of time of a single microdischarge, τmicrodischarge, is estimated from the width of the spikes to be 20-

30 ns, which agrees with literature values of ~30 ns.76Error! Bookmark not defined. The frequency

of microdischarges per unit area, fmicrodischarge, is estimated by counting the number of

discharges in a single voltage cycle and dividing by the period of the voltage cycle and the

surface area of the dielectric exposed to the plasma. For this system, fmicrodischarge ≈ 12

million microdischarges s-1 cm-2, but note that this method likely underestimates the true

value because the minimum time resolution of the oscilloscope (10 ns) makes it difficult

to distinguish multiple simultaneous discharges. A representative Lissajous plot of qt

(black) and qd (red) versus voltage is shown in Fig 4.2c; as discussed above, the plasma power is obtained as the area of the qt loop minus the area of the qd loop, all divided by the period of the voltage cycle.

Figure 4.2. (a) A representative voltage (red) and charge (black) waveform for the confined

DBD formed in methane flow and (b) zoom-in of the charge waveform in (a) showing individual charge spikes.

The spatiotemporal dynamics of the DBD that underlies the measured voltage and charge waveform is schematically depicted in Fig. 4.3a, where on the left, for

corresponding points in the voltage cycle, on the right, white squares represent regions in

the reactor where an active filamentary microdischarge is occurring and filled squares

represent regions where a microdischarge is not occurring. At certain points in the voltage

cycle – just before the voltage extrema are reached – spatially localized microdischarges

occur at different locations in the reactor volume (but do not fill the entire reactor volume).

This process involves three relevant, but highly disparate timescales: the duration of a

single microdischarge (τmicrodischarge, on the order of tens of nanoseconds), the period of the

voltage cycle (τAC, on the order of tens of microseconds), and the residence time of the gas

in the reactor (τres, on the order of seconds).

This process can be observed from the point-of-view of a volume element of gas moving through the reactor. The behavior is shown schematically in the left of Figure 3b, where the red squares denote a particular volume element at different space times in the

DBD. Localized microdischarges form inhomogenously and sporadically (during certain parts of the voltage cycle), and the volume element is only sometimes at the same location as a microdischarge. Even during the parts of the voltage cycle where microdischarges occur somewhere in the reactor, the volume element is not necessarily within a microdischarge, because the microdischarges do not cover the entire reactor. Thus, a fourth timescale is relevant – the average time between instances of a particular volume element being within a microdischarge, which is denoted τafterglow. Note that τafterglow is distinct from

τAC because the microdischarges do not fill the entire reactor volume during the parts of the

voltage cycle when discharges occur; τafterglow is necessarily larger than τAC.

b) Real reactor

Figure 4.3. (a) A representation of the real DBD reactor with periods of filamentary

microdischarges that correspond to the spikes in the charge waveform and no

microdischarges that correspond to the smooth region in the charge waveform and (b) a

comparison following a volume element of gas (red outline) between the real reactor (left),

a 0-D model from literature (center),82 and the steady-state model. The white regions correspond to microdischarges, where the power is primarily dissipated.

It is overly challenging to model this real process, with the inhomogeneity in space

and time, and approximate methods have been developed. Previously reported models of plasma reactors have used a “0-D” reactor model where the plasma reactor volume is considered to be homogeneous in concentration and the conversion depends on time.Error!

Bookmark not defined.,Error! Bookmark not defined.,Error! Bookmark not defined. This approach models the

process as spatially delocalized discharges encompassing the entire reactor, which have

durations τmicrodischarge, and occur at perfectly periodic times τafterglow (although some studies

have neglected the difference between τafterglow and τAC for simplicity), as schematically

depicted in the center of Fig. 4.3b. From the point-of-view of a particular volume element of gas, the 0-D model gives equivalent average behavior as the real system – as the volume element of gas moves through the reactor, it experiences microdischarges with duration of

τmicrodischarge at periodicity of τafterglow.

An alternative approach enables a steady-state solution to the 1-D PFR while giving

the same average behavior as the real process and the previously used 0-D model (note that

the steady-state nature of this model is not an approximation, as the experiments are carried

out at steady state as well). Note that in the 1-D model, the concentration depends on position in the reactor, but in the 0-D model, it does not. A volume element of gas flowing through the reactor experiences (on average) discharge events lasting τmicrodischarge at time

intervals of τafterglow. Therefore, the full reactor is modeled as multiple reactor segments connected in series, as schematically shown in the right of Fig. 4.3b, where the reactor segments alternate between a reactor segment with constant microdischarge and a residence time of τmicrodischarge, and a reactor segment with no microdischarges (only

afterglow) and a residence time of τafterglow. These segments are modeled as separate PFR

reactors with volumes determined by the residence time, τ, and flow rate, :

= 𝑉𝑉̇ (4.6)

𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 and 𝑉𝑉 𝑉𝑉̇ 𝜏𝜏

= . (4.7)

𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 This alternation of microdischarge𝑉𝑉 and𝑉𝑉̇ 𝜏𝜏 afterglow segments continues until the

volume element exits the reactor. The advantage of this approach is that it eliminates the

temporal inhomogeneity of the plasma in the reactor, making it a steady-state system. Thus,

in this approach, Equation 1 simplifies to

( ) + = . (4.8) 𝑉𝑉̇ 𝑑𝑑𝑛𝑛𝑖𝑖 𝑧𝑧 1 𝑑𝑑𝑉𝑉̇ 𝐴𝐴 𝑑𝑑𝑑𝑑 𝐴𝐴 𝑖𝑖 𝑑𝑑𝑑𝑑 𝑗𝑗 𝑖𝑖𝑖𝑖 This steady-state equation was applied to each𝑛𝑛 of the∑ reactor𝑟𝑟 segments in series, where the

output in one segment serves as the input to the next segment, and the molar concentrations

of all the reaction species was obtained along the length of the reactor.

Model parameters were obtained from the experimental characterization of the

DBD. Each microdischarge occurs in a (seemingly) random position in the reactor. After a

microdischarge extinguishes, this region of the reactor is in an afterglow plasma state,

characterized by a decaying electron temperature and electron density.83 The average duration of the afterglow state, τafterglow, was estimated to be 1.1 ms. from the following

relation:

+ = (4.9) 1 𝜏𝜏𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝜏𝜏𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑓𝑓𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝐴𝐴𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 where τmicrodischarge and fmicrodischarge were obtained from experiments and Amicrodischarge was

calculated based on literature values which have shown microdischarges to have a

173 cylindrical cross-section with a diameter, Dfilament~100 µm. The afterglow volume 58

segment was assumed to consume no plasma power, and the microdischarge volume

segment was assumed to consume all the plasma power corresponding to experiments, i.e.,

Pplasma = P’plasma + . A constant plasma power density, P’plasma

𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 (rather than plasma�𝑉𝑉 power) was used𝑉𝑉 in simulations.�

4.4 Results and discussion

4.4.1 Results

The confined configuration was compared to the more typical unconfined version with no

additional grounded SS components in Fig. 4.4. In the unconfined case in Fig 4.4a, the

plasma volume expanded as power was increased, but in the confined case in Fig 4.4b, the

volume remained constant. The importance of confining the volume is further shown in

Fig 4.4c, where the plasma power density changes as a function of applied plasma power

in the confined case, but remains relatively constant in the unconfined case. Thus, in order

to vary the power density, the volume must be confined, and all results presented in this

chapter are from using the confined reactor.

Figure 4.5a shows simulation results for methane conversion, as a function of

volume, at constant power densities of 2.75 W/cm3 and 11 W/cm3. The conversion appears to depend strongly on volume, but because the power density is kept constant in the simulations, Pplasma increases as a function of volume, as shown in Figure 4.5b. Results

from different power densities and volumes enable comparing the conversion at different

volumes with constant power. Figure 4.5c shows that at constant power, conversion is

relatively insensitive to changes in volume. The inset to Figure 4.5c shows that when the

59 simulation results for conversion are plotted as a function of Pplasma, the conversion results fall on a single curve independent of volume.

Figure 4.4. A plasma reactor in an unconfined (a) and confined (b) system as power is increased, and (c) power density versus power for the unconfined and confined systems.

This representation used helium to show the expansion to avoid external discharges.

Figure 4.5. Simulation results for (a) conversion versus volume, (b) power versus volume, and (c) conversion versus volume at equal powers. Inset shows conversion versus power at a power density of 11 W/cm3. These results are at a constant flow rate of 4.3 ml/min.

Figure 4.6 shows experimental results for methane conversion in the confined reactor as a function of power at different volumes and constant flow rate. Due to the scatter in the data, it is not clear whether there is a dependence of the conversion on volume. There appears to be an increase in conversion with increasing volume, but a statistical analysis

shows that the slopes of the linear regression are, within error, the same for all volumes

(Fig. 4.7).

Figure 4.6. Methane conversion versus power from experiments on confined DBD reactor with volumes of 0.016 (black squares), 0.024 (red circles), 0.032 (green triangles), and

0.047 (blue diamonds) cm3, and from simulations with no fitting parameters (dashed black

line) and a scaled power density (solid red line). These results are at a constant flow rate of

4.3 ml/min.

Figure 4.7. Plots of conversion versus power with linear fits for volumes of (a) 16 mm3,

(b) 24 mm3, (c) 32 mm3, and (d) 47 mm3. Multiple two sample t-tests were performed between all of the slopes. Since all p-values are larger than the cutoff (0.05), there is no indication that the slopes are statistically different.

The simulations and experiments thus agree that methane conversion depends on power, but not volume. However, quantitatively, the absolute conversion is ~6-7 times

higher in the simulations than in the experiments. It is possible that not all of this power

goes towards the reaction chemistry as assumed by simulations. In simulations, the model assumes a perfect transfer of power to the electrons that, in turn, transfer their energy through molecular excitation and bond breaking, which ultimately leads to conversion.

This energy transfer picture is idealized and in experiments, some power can be lost to processes that do not produce conversion. This effect can be accounted for in a simple way, by introducing a scaling factor that reduces the power; a scaling factor of 15% gives a quantitative fit between the simulation and experimental results, as shown in Fig. 4.6.

Figure 4.8. Methane conversion as a function of flow rate from experiments (black squares) and simulations (red circles) results at ~1 W and 0.032 cm3.

Results for methane conversion as a function of flow rate, at constant plasma power

and volume, are shown in Fig. 4.8. The experimental and simulation results follow the same trend, in that the conversion decreases as the flow rate increases, and the slope of the change is in good agreement. Again, the absolute conversion obtained by modeling is

higher than found experimentally, which is likely due to inefficient transfer of power

measured in experiments to electrons and the reactant molecules.

4.4.2 Discussion

Many studies of gas conversion using a plasma still consider residence time, and,

thus, reactor volume, to be an important parameter for reactant conversion.65,174–176 For

classical steady-state chemical reactors, where reactions are carried out thermally and/or

catalytically, the conversion does depend strongly on volume. For example, in a PFR, the

conversion of species A is related to the volume by

= (4.10) 𝑉𝑉 𝑓𝑓𝐴𝐴 𝑑𝑑𝑓𝑓𝐴𝐴 𝑉𝑉̇ 𝐶𝐶𝐴𝐴0 ∫0 −𝑟𝑟𝐴𝐴 where fA is the fractional conversion, CA0 is the initial concentration of A, and rA is the rate

of A consumption.69,70 Assuming a zero-order homogeneous reaction ( = ), solving

𝐴𝐴 𝐴𝐴 Equation 8 gives the relation 𝑟𝑟 −𝑘𝑘

= (4.11) 𝑘𝑘𝐴𝐴𝑉𝑉 𝑓𝑓𝐴𝐴 𝐶𝐶𝐴𝐴0𝑉𝑉̇ which shows that the conversion is a strong function of the reactor volume.

This chapter shows, by systematically changing power, volume, and flow rate, that

for plasma reactors, the conversion depends on power and flow rate, but is insensitive to

changes in volume or power density. These results are for changes in one parameter while

the other parameters are held constant; for example, the power was changed while keeping

both volume and flow rate constant. This volume-invariant behavior becomes even more

evident if the results for conversion at different powers, volumes, and flow rate are

combined. Figure 4.9 shows the conversion as a function of which is

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 fundamentally equivalent to since 𝐸𝐸 ⁄𝑉𝑉

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑃𝑃 ⁄𝑉𝑉̇ 65

= , and (4.12)

𝐸𝐸𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ∗ 𝜏𝜏 = . (4.13) 𝑉𝑉 𝜏𝜏 𝑉𝑉̇ All of the conversion results are found to follow this single master curve of ,

𝑝𝑝𝑝𝑝𝑝𝑝𝑠𝑠𝑠𝑠𝑠𝑠 which again demonstrates that conversion does not depend on volume (at constant𝑃𝑃 power)⁄𝑉𝑉̇

or power density.

Thus, the key finding is that the conversion is determined by , which is

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 in essence the energy input per reactant molecule. Individual process𝐸𝐸 parameters⁄𝑉𝑉 such as

power density, residence time, and volume will affect conversion only if they affect this

combined parameter, . For example, a change in volume not only changes the

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 denominator of this ratio,𝐸𝐸 but ⁄the𝑉𝑉 numerator as well and, at constant power, a change in

volume leads to a proportional change in such that remains constant

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 and the conversion is unaffected by the changes𝐸𝐸 in volume and𝐸𝐸 power⁄ density.𝑉𝑉

The idea of being important in a plasma process is not new and several

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 previous studies have𝐸𝐸 shown⁄𝑉𝑉 that conversion depends on ,77,177 or equivalently

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 .57,178–180 However, this study is the first 𝐸𝐸to show⁄ 𝑉𝑉this relationship by

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑃𝑃independently⁄𝑉𝑉̇ varying power, volume, and flow rate, and that only is important.

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 Baylet et al. changed power, but did not vary volume or flow rate.181𝐸𝐸 Nozaki⁄𝑉𝑉 et al. changed

power and flow rate, but did not vary volume.182 Aerts et al. did change power, volume,

and flow rate, but the study varied volume by changing the electrode gap distance; in

contrast to the method described in this chapter, changing the electrode gap distance would

affect the nature of the discharge in addition to the volume as was noted in their paper.174

Figure 4.9. Methane conversion versus specific energy input from experiments on

confined DBD reactor with varying power and volumes of 0.016 (black squares), 0.024

(red circles), 0.032 (green triangles), and 0.047 (blue diamonds) cm3, constant power (1

W), and varying flow rate (open triangles), and from simulations with varying (scaled)

power (red line) and varying flow rate (open circles).

4.5 Conclusion

A combined experimental and simulation methodology showed that conversion in a plasma reactor depends on power and flow rate, but not volume. Experiments were carried out with a confined DBD plasma reactor so that the power and volume could be separately controlled. Simulations were performed under the same process conditions with a steady- state segmented reactor model. This unique combination of controlled experiments and closely connected simulations helped to develop an understanding of the physical behavior of the plasma reactor, which can be generalized to other reactions besides methane conversion. This could help to allow plasmas to be industrially scaled in order to replace

67 certain processes with more environmentally friendly options, such as the direct conversion of methane.

Chapter 5: Hydrogen gas evolution at an electrified plasma-water interface

[The contents of this chapter have been published: Toth, J. R.; Hawtof, R.; Matthiesen, D.;

Renner, J. N.; Sankaran, R. M. On the Non-Faradaic Hydrogen Gas Evolution from

Electrolytic Reactions at the Interface of a Cathodic Atmospheric-Pressure Microplasma and Liquid Water Surface. J. Electrochem. Soc. 2020, 167 (11), 116504. https://doi.org/10.1149/1945-7111/aba15c.]

5.1 Introduction

As discussed in Section 1.4, there are many applications for plasma electrolysis with various reactions. In conventional electrolysis, the reactions are typically limited to the solution, at the electrode interface. With a plasma, not only are reactions occurring in the solution, but they can occur at the interface of the plasma and the liquid as well as in the gas phase itself (within the plasma). While these extra reactions have been documented for CDGE98–102 and anodic GDE leading to non-faradaic reactions,103–105 there is no

discussion of cathodic GDE, which may be advantageous for certain applications. Thus,

this chapter investigates the non-faradaic behavior of cathodic GDE pertaining to the

hydrogen evolution reaction. Various tests are used to determine exactly where the excess

hydrogen originates to build a mechanism for non-faradaic hydrogen evolution.

5.2 Methods

Figure 5.1 illustrates the experimental setup used to measure H2 gas production from a

direct-current (DC), atmospheric-pressure, cathodic microplasma contacting an aqueous

solution surface, consisting of the reactor, mass flow controllers (MFCs), DC high voltage

power supply, gas chromatograph (GC), and optical emission spectrometer (OES). Details

of this reactor have been previously reported.90,96,97,183 A quartz cell sealed with a

polytetrafluoroethylene (PTFE) lid housed two electrodes: a stainless steel capillary tube

(outer diameter of 1.6 mm) separated from the surface of the solution by 1 mm, and a

platinum (Pt, area ~50 mm2) foil in contact with the solution. The capillary tube was

electrically connected with a ballast resistor inline to the high voltage side and the Pt foil

was electrically connected to the ground side of the DC power supply (max voltage of 3

kV and current of 20 mA).

electrode, and gas chromatograph.

A microplasma was ignited in the gas gap between the end of the capillary tube cathode and the solution surface by increasing the voltage to ~2 kV after which the desired current, between 2 and 8 mA, was set to operate under galvanostatic control. The discharge voltage was measured by a multimeter and the current was measured by the voltage drop across a resistor in series on the ground side and recorded by a computer with a custom

LabVIEW virtual interface. The discharge and supply voltages were measured to be 360 and 730 V at 2 mA, 380 and 1160 V at 4 mA, and 470 and 2060 V at 8 mA, respectively.

Current fluctuations are shown in Fig. 5.2 and were measured to be less than 0.05 mA after an initial spike resulting from gas breakdown. Thus, a constant current was assumed.

Figure 5.2. Representative current waveform for the plasma electrolytic reactor with a set point current of 8 mA. There is an initial large spike resulting from gas breakdown, but the current quickly stabilizes at 8 mA, with a calculated variance of 0.05 mA.

Because the focus of this study was H2 evolution and to minimize any other potential chemical reactions, an inert gas, Ar, was chosen as the supply gas and the solution

was a dilute mixture of sulfuric acid and water (pH=3.5). The gas flow rate through the

plasma was 50 sccm (standard cubic centimeters per minute) and the solution volume was

20 ml. Before each experiment, the solution was sparged with Ar at 75 sccm for at least 30

min. During the experiment, the system was continuously purged with 75 sccm; the

combined flow from the purge and plasma carried the H2 evolved out of the reactor for analysis.

To benchmark the microplasma experiments, a conventional electrolysis configuration was also characterized by replacing the stainless steel capillary tube with a

Pt foil in contact with the solution to serve as the cathode. The system was again galvanostatically controlled and the voltage was applied by a HP E3610A DC power supply

(max 15 V and 3 A) at currents between 2 and 8 mA, without the formation of a plasma.

The temperature of the solution with and without external heating or cooling was

measured by a FLIR i3 infrared camera. Optical access for the camera required that the

quartz cell was open (without the PTFE lid) and the solution temperature was measured

from the top over the surface. An emissivity for water, ε, of 0.95, was assumed and the

accuracy of the temperatures was independently verified by a thermometer to be ±2 oC.

The amount of H2 gas produced was measured using a Shimadzu GC-2014 GC with a thermal conductivity detector and a Restek ShinCarbon ST 80/100 mesh 2 m x 2 mm column. The gas analysis was carried out by injecting a 100 μl sample into the GC at various times. Nitrogen was used as the carrier gas to allow identification of the Ar and H2

from the reactor. Confirmation of the retention times for H2 and Ar peaks and calibration

of the peak areas was carried out by analyzing reference gas mixtures of H2 and Ar. Two

separate MFCs were used to obtain controlled ratios and a calibration curve was

constructed relating the GC peak area ratio to the H2 concentration with a lower detection

limit of 15 ppm. Experiments were repeated with a minimum of three trials to obtain the

average H2 concentration at each time and carry out error analysis. All data are represented as the mean and standard error.

OES was carried out on the microplasma using an Ocean Optics USB4000. Similar

to thermal camera measurements, the quartz cell was open, in this case, to avoid

condensation of water vapor on the reactor wall which would affect light collection.

Spectra from 200 to 900 nm were collected every second for 1000 seconds during each

trial.

5.3 Results and discussion

Figure 5.3a shows the average production of H2 measured as a function of time in the

plasma electrolytic reactor at different currents. The relatively constant production with

time indicates that the system operates close to steady state at all currents over the time

periods studied. The H2 production increases with current, but not by double the amount

between each current increase as would be expected from Faraday’s law.

The faradaic efficiency corresponding to the H2 produced was calculated by the

following equation based on Faraday’s law and the hydrogen evolution reaction (HER)

which is a two electron reaction:

(%) = (5.1) / 𝑛𝑛̇ 𝐻𝐻2 𝐼𝐼 2𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 -1 where is the molar production rate of H2 in mol s , t is the time in s, I is the average

𝐻𝐻2 rate of 𝑛𝑛chargė passed over time (current) through the electrolytic system in C s-1 (A), and

-1 F is Faraday’s constant and equal to 96,485 C mol . The H2 molar production rate was calculated from the instantaneous concentration measured by the GC. Because the faradaic

efficiency was calculated based on measuring H2 at a given time, as opposed to an

accumulation over time, it corresponds to an instantaneous value.

Figure 5.3b shows the average HER faradaic efficiencies calculated for the plasma

electrolytic reactor at the respective currents studied. Results for the conventional

electrolytic reactor with a Pt cathode are also shown for comparison. Due to the weak

dependence on time, the time of 30 min was chosen for comparison. For the conventional

electrolytic reactor, the faradaic efficiencies are ~100% at all currents, which is consistent

with only having water electrolysis possible in the simple acid electrolyte. This confirms

that the H2 measurement is reliable, including effective collection of the gas evolved in the reactor, detection, and quantitative analysis. In comparison, the faradaic efficiencies are substantially higher and exceed 100% at all currents for the plasma electrolytic reactor, slightly decreasing from 200% to 140% between 2 and 8 mA, respectively. The voltage required to operate the plasma is much larger than the conventional process and therefore, the energy consumption will be much higher as well. As an example, at 8 mA, the energy

consumption for the plasma electrolytic reactor is estimated to be 12.5 MWh/kg (0.3%

efficiency on a HHV basis), while typical values for conventional water electrolysis are

~57 kWh/kg (70% efficiency).184

The faradaic efficiency calculations are a construct that relates the amount of

reaction product to the charge transferred. Faradaic efficiencies greater than 100% suggest that there is a reaction mechanism for the same product that does not involve charge

transfer. Such a non-faradaic mechanism is possible in a plasma electrolytic reactor due to

gas-phase excitation, which could produce neutrally charged radicals that recombine in the

gas phase or dissolve and participate in solution-phase reactions. As Ar is an inert gas, a

potential source of gas-phase chemistry is water vapor from evaporation of the liquid phase

electrolyte. The water vapor concentration is related to the equilibrium vapor pressure

which in turn depends on the system temperature. For this reason, the temperature was

varied to study the effect of water vapor on the H2 production.

The solution temperature was varied by externally heating or cooling the cell with

a hot plate or an ice bath at a source temperature of 85 and 5 oC, respectively. Controlling

the system temperature is quite difficult, since the plasma also heats the solution surface at

the point of impingement. Figure 5.4a and 5.4b show representative IR images of the after

30 min with heating and cooling, respectively. The plasma is the hottest point in the system,

20 to 40 °C hotter than the bulk of the electrolyte. Further, Fig. 5.4c shows the solution surface temperature as a function of distance from the plasma, determined from the infrared camera images. A gradient exists from the solution bulk, which corresponds to the source temperature, to the plasma-solution interface. The temperature at the plasma-solution

interface also varies with time, as shown in Fig. 5.4d, eventually reaching a steady state.

Even though these spatiotemporal gradients make the actual temperature of the system at

the solution surface different than the source temperature (i.e., 5 °C, ambient, 85 °C) and do not readily allow a singular temperature to be assigned, there is still a significant temperature difference of approximately 40 °C between each of the conditions that provides a sufficient perturbation. For example, the partial pressures of water at the maximum plasma-solution surface temperatures of 103 °C, 58 °C, and 18 °C are 113 kPa,

18 kPa, and 2 kPa, respectively, giving a difference in the water vapor content of almost

10 times between ambient temperature and 85 °C and between ambient temperature and 5

°C.

function of time at the plasma electrode. Ambient refers to no external heating or cooling.

The current in all cases was 8 mA.

The local presence of water vapor was corroborated in the plasma phase above the

solution surface as a function of the solution source temperature by OES. Water vapor is excited and/or dissociated in the plasma and OES provides direct evidence of the resulting

77 excited states by their radiative transitions. Figure 5.5a shows emission spectra collected from the plasma-solution interface at different bulk solution temperatures for a constant current of 8 mA. To allow direct comparison, the spectra were normalized to the Ar line at

751 nm. The peak at 310 nm corresponds to an OH state (A2Σ+(v=0)→X2П(v=0)) and supports the presence of water vapor. At longer wavelengths (312-380 nm), peaks

3 3 corresponding to the second positive system of N2 (C Пu→B Пg) are detected. While Ar was the supply gas for the plasma, there is N2 contamination from air in the OES measurements because the cell was kept open to air to have optical access. As the temperature increases, the OH emission line intensity increases and concomitantly, the N2 emission line intensities decrease, indicating that the water vapor concentration increases as expected from equilibrium vapor pressure. The dependence of the line intensities is shown more clearly in Fig. 5.5b as a function of time at different temperatures. Overall, the line intensity increases with temperature, but there is some variation with time. The open cell caused the solution to evaporate quickly, especially at higher temperatures, and as the gap between the plasma electrode and solution surface increased, the plasma increased in volume which increased the emission intensity. To address the evaporation, more solution was added, and as the plasma volume decreased, the emission intensity decreased back to its original value. Despite these effects from the modified setup, the OH peak intensity was clearly found to be higher as the solution temperature increased, indicating that there are more water molecules in the plasma able to engage in chemical reactions in the gas phase.

Figure. 5.5. Optical emission spectroscopy (OES) characterization of plasma electrolytic reactor at different solution source temperatures showing (a) spectra exhibiting lines corresponding to OH and (b) normalized intensity of OH line at 310 nm as a function of time. The current was 8 mA in all cases. At high solution source temperature, the solution evaporated which increased the plasma volume and subsequently, the emission intensity and water was added during OES characterization to compensate.

To correlate the presence of water vapor to H2 production, the H2 evolved was measured at different solution source temperatures. The current was kept constant at 8 mA which was the highest current studied and should lead to the largest thermal loading and

therefore, have the strongest effect on H2 production. Figure 5.6a shows the average production of H2 measured as a function of time at the different solution source

temperatures. The ambient temperature result refers to the lack of any external heating or

cooling and the bulk solution was measured to be approximately 30-40 °C. When the

solution was externally cooled to a bulk temperature of 5 °C, the H2 production was found

to decrease, more pronounced at early times and then less substantially at longer times.

Externally heating the solution to a bulk temperature of 85 °C was found to greatly increase

the H2 production at all times. The effect of solution temperature on H2 production was

further analyzed by calculating the corresponding faradaic efficiencies. As shown in Fig.

5.6b, in comparison to the room temperature faradaic efficiency of 140%, cooling the

solution to 5 °C was found to decrease the faradaic efficiency to 110%, while heating the

solution to 85 °C was found to increase the faradaic efficiency to 220%.

Figure 5.6. (a) Average H2 production as a function of time for different solution source

temperatures in the plasma electrolytic reactor and (b) corresponding instantaneous

faradaic efficiencies calculated at 30 minutes. The current was 8 mA in all cases.

These anomalous faradaic efficiencies exceeding 100% can be explained by

analyzing the different possible reaction mechanisms for H2 evolution in the plasma

+ electrolytic system. The first mechanism is similar to conventional electrolysis where H3O

in an acidic solution is reduced by electrons in solution. In the case of a cathodic plasma,

electrons are injected from the plasma across the plasma-liquid interface and solvate.

- Reactions involving plasma-injected solvated electrons, e(aq), have been shown to closely

follow radiolytic chemistry,94 with the most kinetically relevant being second order recombination,

2e( ) + 2H O( ) H ( ) + 2OH( ) (5.2) − − aq 2 l 2 g aq + and the reduction of H3O , ⟶

2e( ) + 2H O( ) H ( ) + 2H O( ) (5.3) − + aq 3 aq 2 g 2 l In acidic solutions, Eqn. 5.3 will be kinetically⟶ favored over Eqn. 5.2, and as it is a charge-

transfer reaction, the H2 generated is directly proportional to the current, reaching as high

as 100% faradaic efficiency in the absence of kinetically significant competing reactions.

Distinct from conventional electrolysis, a second mechanism is comprised of gas-

phase reactions in the plasma. Although the supply gas, Ar, is inert, the equilibrium vapor

pressure of water, enhanced by heating at the plasma-liquid interface, introduces water vapor as the potential source of H2. Electron-impact dissociation of water vapor could lead to the formation of H2. There are numerous possible reactions that describe this

dissociation, and a simple kinetic analysis was carried out to compare their rates and

identify the kinetically most significant.

The rate constants were calculated using BOLSIG+ (version 12/2017)165 from the

reaction cross sections reported by Itakawa et al.171, assuming a mean electron temperature

of 1 eV, which is reasonable for an atmospheric-pressure DC microplasma.185 A summary

of the rate constants for the various electron impact reactions is shown in Table 5.1 showing

that the highest rate constant among the reactions considered is Reaction 4. This reaction

would quickly be followed by recombination of H to generate H2:

2H H (5.4)

2 This reaction pathway does not involve charge⟶ transfer and is not faradaic (generally

referred to as non-faradaic). There could be even more pathways that similarly produce H2

without charge transfer shown in Table 1, as well as others such as the dissociation of water

by Ar metastables to produce H.186

Table 5.1. Rate constants for the most favorable water vapor dissociation reactions by

electron impact that produce H, presumably leading to the formation of H2 gas. All rate

constants were calculated assuming a mean electron temperature of 1 eV with BOLSIG+165

using the reaction cross sections reported by Itakawa et al.171

Reaction Rate constant (m3/s) - - -19 1. e + H2O → H + OH 5.69×10 - - -18 2. e + H2O → 2H + O 2.03×10 - - -17 3. e + H2O → H + OH 1.65×10 - - -16 4. e + H2O → e + H + OH 7.27×10 - - -17 5. e + H2O → e + 2H + O 2.05×10 - - + -20 6. e + H2O → 2e + H + OH 2.89×10 - - + -19 7. e + H2O → 2e + H + OH 1.22×10 - - + -21 8. e + H2O → 2e + 2H + O 5.34×10

The two reaction mechanisms, one in solution and another in the gas (plasma)

phase, would produce H2 in parallel without necessarily affecting one another. The H2

evolved in the gas phase involves electrons, but without charge transfer, the inelastically

scattered electrons would be preserved to subsequently enter the solution, solvate, and react. The charge transfer (faradaic) efficiency calculation does not account for the H2

produced by the gas-phase mechanism. The gas-phase reactions depend on the presence

and amount of water vapor, controlled by the solution temperature, which is the source of

hydrogen species. In the solution phase, the reactions again involve electrons, but with

charge transfer, they are captured by the faradaic efficiency calculation.

As illustrated in Fig. 5.7a, this plasma electrolytic system is characterized by a

multiphase environment and reactions occur at the plasma-liquid interface in both the gas

phase, non-faradaically, and liquid phase, faradaically. Specifically, H2 is formed in the gas phase via impact dissociation of water vapor, and simultaneously in the liquid phase

+ via solvated electron reduction of water or H3O in the case of an acidic electrolyte. The

water vapor concentration that drives the gas-phase chemistry depends on the equilibrium

vapor pressure of water, which in turn is a function of the system temperature. The system

temperature is complex because the solution surface is heated locally by the impinging

plasma, and is a function of current and time, leading to spatiotemporal gradients in the

solution phase and concentration gradients in the gas phase.

Figure 5.7. (a) Illustration of different regions in plasma-liquid system, including gas-

phase, gas-liquid interface, and solution-phase, and key corresponding reactions that lead

to formation of H2 gas, and (b) average H2 produced measured at different currents and

solution source temperatures at 30 minutes, separated into a faradaic and excess amount

that can be linked to the gas-phase (pink) and solution-phase (blue) reactions.

To gain further insight, it is assumed that the solution-phase HER proceeds at 100%

faradaic efficiency and therefore, the gas-phase reactions are the contributing factor for the

faradaic efficiency in excess of 100%. This assumption is supported by the absence of any

competing reactions either from dissolution of gas-phase species since an inert gas, Ar,

serves as the background, or from the electrolyte, which is only composed of dilute acid in

water.187 Additional evidence is provided by the results with cooling of the solution where the water vapor is lessened and the faradaic efficiency approaches 100%, shown in Fig.

5.5b. The quantitative measurements of H2 produced at different currents and solution

temperatures are then separable into a gas-phase (non-faradaic) and liquid-phase (faradaic)

component, as shown in Fig. 5.7b. Cooling the solution, which diminishes water vapor

formation, decreases the amount of H2 produced in the gas phase, while heating the solution, which enhances water vapor formation, increases the amount of H2 produced in

the gas phase. Since the amount of H2 produced in the solution phase only depends on

current and is unchanged, the apparent faradaic efficiency correspondingly increases with

temperature, as shown in Fig. 5.5b. The effect of current is not as straightforward because

of multiple effects. In the solution phase, the amount of H2 produced doubles as the current is increased from 2 to 4 to 8 mA. The excess measured H2 suggests that in the gas phase,

the amount of H2 produced increases slightly from 2 to 4 mA, but then decreases slightly

from 4 to 8 mA. As the current increases, more Joule heating is expected, which should

enhance water vapor formation and H2 production. However, increasing the power also has

effects on the gas-phase chemistry, potentially influencing plasma parameters such as the

electron density, electron energy distribution, and neutral gas temperature that together will

influence the reaction kinetics. This analysis suggests that the effect of current on gas-

phase chemistry counteracts that on water vapor concentration such that the amount of H2

produced is relatively constant with increasing current. Consequentially, the apparent faradaic efficiency actually decreases with increasing current, as shown in Fig. 5.2b, because the gas-phase contribution to H2 production is more significant at lower current

where the solution-phase contribution is smaller.

5.4 Conclusion

Quantitative measurements show that the H2 evolved in a plasma electrolytic system

characterized by an atmospheric-pressure microplasma formed at the surface of liquid

water can be more than the faradaic equivalent, and depends on solution temperature.

Based on a kinetic analysis, the cumulative reactions in the gas and solution phases offer a

picture for how this is possible, with a non-faradaic contribution from the gas phase and a faradaic one from the solution phase. The gas-phase reactions make these systems distinct from more conventional electrolytic systems and could be increasingly complex as more volatile solvents such as organics and reactive gases such as air are employed. The mechanisms related to the excess hydrogen evolution are an important for applications where cathodic GDE is desired.

Chapter 6: Continuous, process-intensified nitrogen fixation in a plasma-water

droplet reactor

[The contents of this chapter are under review to be published: Toth, J. R.; Abuyazid, N.

H.; Lacks, D. J.; Renner, J. N.; Sankaran, R. M. A Plasma-Water Droplet Reactor for

Process-Intensified Continuous Nitrogen Fixation at Atmospheric Pressure. ACS Sustain.

Chem. Eng. 2020, under review.]

6.1 Introduction

The typical process to make ammonia is carried out at high temperature and high pressure

process reacting hydrogen and nitrogen. The hydrogen source typically results in the formation of large amounts of carbon dioxide. An alternative option for ammonia generation is reacting water and nitrogen gas directly through the use of a plasma at atmospheric pressure, without external heating. This chapter discusses a dielectric barrier discharge reactor that fixes nitrogen using nebulized water droplets. Various nitrogen products are characterized with controls to verify the ammonia formation. As plasmas can often get hot, droplet stability is also characterized in the plasma. Further, a mechanism for the formation of ammonia in this novel system is discussed.

6.2 Methods

Figure 6.1 shows the continuous, atmospheric-pressure plasma-water droplet reactor that

was constructed and used to study N2 activation. The system is composed of four main

components: 1) an atomizer to continuously generate liquid H2O aerosol droplets; 2) a gas

heater to heat the gas flowing out of the nebulizer; 3) a dielectric barrier discharge (DBD)

to generate a non-thermal plasma at atmospheric pressure; and 4) a liquid trap to collect

the reaction products by dissolution.

Atomization was performed with a medical nebulizer (Philips Respironics HS860

SideStream Reusable Nebulizer) in which liquid H2O is pulled into a gas flow by the

Venturi effect to generate small liquid aerosol droplets (~1 µm). In a typical experiment,

-1 the atomizer was filled with 10 mL of deionized H2O and the gas flow rate was 5 L min .

In some experimental runs, a gas heater (OMEGA T-Type AHP-5052) was used to evaporate the droplets and heat the gas stream entering the DBD reactor. The volume of the gas heater is 15 mL. The temperature of the heater was varied by changing the applied voltage to the heater, allowing control from room temperature to up to 600 °C.

Figure 6.1. Schematic of the continuous, atmospheric-pressure plasma system studied for

reaction of nitrogen gas and water droplets. Water droplets were generated by atomization

and passed through a gas heater that was used in some experiments to evaporate them

before entering the plasma. The plasma was a coaxial DBD with an outer copper ring and

an inner tungsten wire that served as the high voltage and electrically grounded electrode,

respectively. The reacted products were collected by two stages of trapping, bubbling in

an acid solution at 5 °C and condensing at -39 °C.

The DBD reactor was homemade using a quartz tube with an inner and outer

diameter of 4.0 and 6.3 mm, respectively, wrapping copper foil with a width of 10 mm

around the outside to serve as the high-voltage electrode, and inserting a tungsten wire with a diameter of 0.25 mm in the center to serve as the counter electrode. The total volume of the plasma zone was 0.13 mL. A DBD was ignited and sustained by a high-voltage alternating current (AC) power supply (Information Unlimited Model PVM500). The voltage and charge waveforms were continuously monitored by a digital oscilloscope

(Hewlett Packard 54616C) and recorded using a LabVIEW virtual instrument. The charge was measured by the voltage across a capacitor (44.5 nF) in series between the counter electrode and electrical ground and is made up of both the charge passing through the DBD and the displacement charge. The displacement charge was estimated by measuring the charge at voltages below the threshold for gas breakdown and linearly extrapolating to the voltage at which the DBD operated (~9 kV); linear dependence for the displacement charge was assumed since a DBD can be approximated by an electrical circuit composed of resistors and capacitors.153 The areas enclosed by Lissajous plots of total charge and

displacement charge versus voltage correspond to the total and displacement energies

during a single cycle. The energy consumed by the DBD was estimated by subtracting the

displacement energy from the total energy, and the power was determined by dividing the

energy by the period of the cycle. The gas temperature in the DBD was determined by

analysis of the nitrogen molecular band emission measured by a spectrometer (Teledyne

Princeton Instruments HRS 500 with PIXIS camera).

Reaction products were characterized in the solution phase by collecting them with

two stages of liquid traps. In the first stage, the reactor effluent was passed through a

bubbler filled with 10 mL of concentrated sulfuric acid (pH 2.0). To minimize evaporation

of the solution in the bubbler and enhance condensation of vapor products, the trap was

kept at a temperature of approximately 5 °C using an ice bath. In the second stage, the

bubbler effluent was passed through a condenser kept at -39 °C using an oil bath with an

immersion chiller. The two stages were required to ensure that as much of the reaction

products as possible were dissolved, especially given the very high gas flow rate through

the system. The two liquid traps were combined and characterized together. In a typical

experiment, 6.7 mL of deionized H2O was atomized, 13.2 mL of solution was collected in

the first stage, and 1.1 mL was collected in the second stage. Thus, out of the possible 16.7

mL combined between the amount of H2O passed through the system and originally contained in the bubbler, 14.3 mL was recovered; the remaining solution was likely lost as vapor or converted to gaseous products.

+ The NH4 concentration collected in solution was first determined by a colorimetric

method using the o-phthalaldehyde (OPA) fluorometric assay test (QuantiFluo™ ammonia assay kit). Samples were diluted with deionized H2O in order to be within the linear range

of the calibration curve. Different dilutions were used with a typical one being 1:1 with a

100 µL aliquot added to 100 µl of deionized H2O. After dilution, the sample was reacted

with 90 µL of the reagent for 15 min in the dark at room temperature and the fluorescence

was measured (λexcitation=360 nm, λemission=450 nm) with a spectrometer (Molecular Devices

Spectramax M2). A calibration curve was created using aqueous solutions of ammonium

chloride between concentrations of ~0.02 to 1 mM.

+ The NH4 concentration was also independently determined by an L-glutamate dehydrogenase (GDH) enzymatic assay test (Sigma-Aldrich). Before testing, the pH of the samples was brought up to ~7.0 by adding concentrated sodium hydroxide (~1 mL at pH

+ of ~14). Then, we followed the standard operating procedure. Briefly, the NH4 reacts with a-ketoglutaric acid (KGA) and reduced nicotinamide adenine dinucleotide phosphate

(NADPH) in the presence of GDH to form L-glutamate, oxidized nicotinamide adenine

+ dinucleotide phosphate (NADP ), and H2O. The absorbance was measured at 340 nm, where NADPH absorbs, before and after adding 10 µL of GDH and reacting for 5 min at

+ room temperature. The NH4 concentration was calculated from a formula provided by the

assay kit. The equation was confirmed using an aqueous solution of ammonium sulfate at

a similar concentration to the samples.

- - The NO3 and NO2 concentrations collected in solution were determined by a

colorimetric method using the Griess absorbance assay test (Sigma Aldrich nitrite/nitrate

- - assay kit). To detect the total concentration of NO3 and NO2 , samples were diluted by

adding a 10 µL aliquot to 490 µl of deionized H2O, followed by the addition of nitrate

- - reductase to reduce NO3 to NO2 , which incubated for 2 hours. Separately, to detect the

- concentration of only NO2 , samples were diluted by adding a 10 µL aliquot to 90 µl of

deionized H2O. In both cases, after dilution and incubation of the enzyme, the Griess reagents were added and reacted for 15 min at room temperature and the absorbance was measured at 540 nm with a spectrometer (Shimadzu UV-1800). Calibration curves were

created using aqueous solutions of sodium nitrate and sodium nitrite between

concentrations of 0 and 80 µM.

Experiments were repeated at a given process condition a minimum of three trials

and up to >20 trials, and the generated datasets are available in the Supplementary

Information. For all repeated trials, we calculated the mean and standard error which

corresponds to the variance within one standard deviation of the mean (approximately 67%

confidence interval). Significance was determined by t-tests using the Minitab 2017

Statistical Software. The statistical differences between zero and the mean of a dataset

were determined by a one-sample t-test, and between the means of two datasets by a two-

sample t-test, with p-values less than 0.05 considered to be significantly different (α=0.05).

Constant variances were not assumed.

6.3 Results and discussion

6.3.1 Reaction product characterization

The standard experiment consisted of an inlet stream of N2 and H2O droplets flowing

through a DBD reactor operated at a plasma power of 24 W. From OPA colorimetric assay

measurements, the NH3 production rate at the standard experiment was found to be 184 ±

17 µg h-1 (10 ± 1 µmol h-1), as shown in Fig. 6.2. The results of the OPA method were corroborated by the GDH enzymatic assay; with its accuracy confirmed, the OPA method

+ was used to measure the NH4 concentration in all other experiments. Based on mass

balances for nitrogen and hydrogen, the conversion of input N2 and H2O to NH3 were

estimated to be approximately 0.4×10-4 and 40×10-4 %, respectively.

In addition to reduction to form NH3, the other reaction pathway expected for N2 with

59,95,115,119–122 H2O as a feedstock is oxidation to form nitrogen oxides (NOx). The initial

step for N2 oxidation, similar to reactions of N2 and O2 in air plasmas, is the formation of

nitric oxide (NO) and nitrogen dioxide (NO2). If liquid H2O is present, these species can

- - solubilize to form NO2 and NO3 . In the presence of H2O vapor, NO and NO2 can further

react in the gas phase to form nitrous acid (HNO2) and nitric acid (HNO3) vapor which,

- - again in the presence of liquid H2O, can subsequently solubilize to form NO2 and NO3 .

- - By measuring the NO2 and NO3 collected in solution, we characterized nitrogen oxidation

- - by either of these pathways. Figure 6.2 shows that both NO2 and NO3 ions were produced, with rates of 10 ± 2 µmol h-1 and 39 ± 3 µmol h-1, respectively, corresponding to

-4 -4 conversions of approximately 2×10 and 400×10 % of the input N2 and H2O,

- respectively, into these two products; the production rate of NO3 was substantial, greater

than that of NH3.

All the products generated may not have been trapped for characterization. Previous studies have found that using a solution with a different pH would have a different affinity

188 for trapping the various products. While low pH was more efficient in trapping NH3 and

- - NO3 , while NO2 was better trapped at a higher pH.

- - Figure 6.2. Production rates of NH3, NO3 , and NO2 measured for the standard experiment at a power of 24 W.

6.3.2 Control experiments for ammonia formation

Control experiments have recently been shown to be critical in electrosynthesis of NH3 due

to the possibility of false positives from background NH3 in the system or contamination

from various sources.189,190 A particularly useful control is isotopic labelling which can

directly confirm whether the NH3 detected originates from the N2 gas. In plasma-based

approaches, the NH3 produced is typically substantially higher than the background

contamination,59,95,122 and more cost-effective controls can be applied. Thus, several

different controls were carried out, focusing on the NH3 detected in the standard experiment, and the NH3 production rates measured are summarized in Fig. 6.3.

Figure 6.3. Production rates of NH3 measured for the standard experiment (N2+H2O

droplets flowing through DBD reactor) and various controls. All experiments with a plasma on were conducted at a plasma power of 24 W.

Figure 6.4. Results from the thermodynamic equilibrium calculation for the reaction of

3 1 3 H O+ N → NH + O as a function of temperature. The equilibrium constant, was 2 2 2 2 3 4 2

calculated from correlations for the enthalpy and entropy.191 The values are very small,

even at higher temperatures, indicating that ammonia is not thermodynamically favored.

The first control was N2 gas and H2O droplets in the inlet stream without any power

-1 applied (i.e., no plasma). The NH3 production rate was measured to be 7 ± 3 µg h (or

0.4 ± 0.2 µmol h-1) at room temperature, which was insignificant from zero (one sample t- test, p-value = 0.08). Because the plasma heats the gas, this control was also extended to

higher temperatures and measured a production rate of 21 ± 3 µg h-1 at 590 oC, the

maximum that could be reached by the gas heater. However, a similar production rate of

-1 19 ± 9 µg h was measured even when the N2 was replaced by Ar. No NH3 is expected to

be synthesized in this case, as confirmed by thermodynamic equilibrium calculations as

shown in Fig. 6.4, and the detected NH3 is attributed to background contamination. A

potential source of contamination is NH3 produced in previous experiments that is adsorbed 94

to the reactor walls. The result of the standard experiment was significantly larger than the

background measured at room temperature (two sample t-test, p-value = 10-9) and 590 oC

(two sample t-test, p-value = 0.01).

The second control was Ar gas and H2O droplets as the inlet stream, with the plasma

-1 at a power of 24 W. The NH3 production rate was measured to be 4.1 ± 1.4 µg h , which

is not significantly different than the background (two sample t-test, p-value = 0.45). The

Ar control was also used to measure the background production rate of NOx, finding the

-1 - - production rates to be 0.8 ± 0.2 µmol h and zero for NO3 and NO2 , respectively.

The next control was N2 gas only as the inlet stream (no H2O droplets), with the

-1 plasma at a power of 24 W. The NH3 production rate was measured to be 41 ± 7 µg h , which is significantly larger than the background (two sample t-test, p-value = 10-3).

However, the gases for these experiments were relatively low purity (99.5%), and we

suspect there was background H2O vapor that enabled the formation of NH3. For this

reason, a similar control experiment was carried out using a higher purity N2 gas feed

(99.999%); in this case, a production rate of 8.6 ± 3.5 µg/h was measured, which is

insignificantly different from the background (two sample t-test, p-value = 0.75). This test suggests that the largest source of contamination in our experiments is from the gas feed.

Although the system operates at atmospheric pressure and is not vacuum tight or evacuated before experiments by a vacuum pump, the contamination from the ambient air, as a source of N2 and H2O vapor, is not significant (in comparison to the background).

6.3.3 Energy Analysis

The standard experiment was extended to different plasma powers. Increasing the plasma

power was found to increase the production rate, from a minimum of 120 µg h-1 at 15 W to

-1 a maximum of 560 µg h at 72 W as shown in Fig. 6.5. Since the production rate of NH3

increases nearly linearly with an increase in plasma power, the energy consumption should

be nearly constant at 9500 ± 700 MJ mol-1. The energy consumption varied somewhat with

plasma power ranging from 10000 to 7000 MJ mol-1, with the minimum value occurring at a plasma power of 50 W. By including all fixed forms of nitrogen measured, the production rate for our plasma-water droplet system improves to 60 µmol h-1 with an

associated energy consumption of 1900 MJ mol-1. While these energy consumptions are

Error! substantially larger than those reported for conversion of N2 and H2O by plasma jets,

Bookmark not defined.,Error! Bookmark not defined.,Error! Bookmark not defined.,Error! Bookmark not defined. we

expect that optimization of the reactor conditions will lead to dramatic improvements. The

experiments were carried out with relatively low concentrations of H2O (3 %), limited by

the atomization which requires large gas flows, and a significant amount of energy could

be “wasted” in exciting N2 molecules that do not necessarily contribute to NH3 formation.

In N2/H2 plasma systems, the optimal feed is a stoichiometric ratio of H2 to N2 (1:3), and a

192 similar ratio might be expected for H2O as a feedstock.

Figure 6.5. Summary of production rates (black squares) and corresponding energy costs

(red circles) for NH3 synthesized from N2 and water droplets in atmospheric-pressure DBD reactor as a function of power. The production rate is found to increase nearly linearly over the range of tested powers and thus, the energy cost remains approximately constant.

Further, catalysts could be used to significantly improve the NH3 production rate; in N2/H2 plasma systems, catalysts have been found to double the NH3 production rate over a purely plasma process.193 While the tungsten wire in our system may have had a catalytic effect, we replaced it with an identically-sized copper wire, since copper has been found to be a much more effective catalyst than tungsten for ammonia synthesis.194 Compared to the standard experiment, the production rate of NH3 was found to have approximately

-1 -1 - - doubled, to 400 ± 50 µg hr (4000 ± 500 MJ mol ), while NO3 and NO2 remained relatively unchanged at 65 ± 6 and 8.1 ± 0.2 µmol h-1, respectively. This indicates that catalysis can improve our process, and that tungsten had a very weak catalytic effect if any.

Further improvements can be used if a nano-structured catalyst was used in place of the

small surface area wire.

6.3.4 Plasma and droplet characterization

To understand the role of the droplets in NH3 formation, both the plasma and the droplets

were characterized. For the confined plasma reactor dimensions and atmospheric-pressure operation, it was necessary to apply non-intrusive diagnostics such as electrical and optical methods. Figure 6.6a shows representative voltage and charge waveforms measured in the

DBD reactor with N2 gas and H2O droplets as the input stream. The charge waveform exhibits frequent spikes along the falling edge of the AC charge cycle, and sporadic spikes along the rising edge of the AC waveform. These spikes correspond to filamentary discharges that form locally and intermittently, with dimensions of hundreds of microns and timescales of tens of nanoseconds.78–80,173 The filaments carry most of the charge

passing through the plasma. The number and frequency of the filaments or charge spikes

will depend on many factors in the DBD, one of which is the supply gas. For example,

195 more filaments occur in N2 than Ar because of its higher electric breakdown strength. A

representative Lissajous plot of the total charge is shown in Fig. 6.6b which, after

subtracting the energy of the displacement charge, corresponds to a plasma power of 24

W. In general, the plasma power depended on applied voltage, and increased from 15 to

70 W ± 5% as the applied voltage increased from 6 to 12 kV.

Figure 6.6. Representative (a) voltage (black), total charge (red), and (b) Lissajous waveforms for the DBD containing nitrogen and water droplets. The area of the Lissajous plot corresponds to the total energy passing through the DBD in a single cycle and from the displacement energy and period of the cycle, was used to find the plasma power, which in this case was 24 W.

The gas temperature was estimated by the well-known technique of analyzing the

196 emission spectrum corresponding to the second positive system of N2. Optical emission

spectroscopy (OES) spectra are shown in Fig. 6.7. The analysis provides the rotational temperature of the molecule, which is expected to be very similar to the translational temperature due to the rapid energy transfer between these excitation modes.197 To obtain

the rotational temperature, a Boltzmann plot was constructed from spectral line intensities

of different nitrogen states making up the second positive system using the published software massiveOES.198,199 The linearity of the Boltzmann plot indicated that the state

population was in thermal equilibrium, supporting the validity of our analysis. At a power of 24 W, the gas temperature was estimated to be 650 ± 70 °C.

Figure 6.7. Optical emission spectroscopy (OES) of DBD with inlet feeds of (a) N2 and

water droplets and (b) Ar/N2 (90/10) and water droplets. The spectra show the N2 second

positive system which was analyzed to obtain the gas temperatures.

A key question in understanding the system is the stability of the liquid H2O

droplets as they are transported through the reactor, especially since the gas temperature in the plasma is substantially higher than the vaporization temperature of water at atmospheric pressure. The atomizer generates liquid H2O droplets with diameters ~ 1 µm, but the

droplets could be converted to H2O vapor by several mechanisms. The N2 carrier gas is

essentially dry and the H2O droplets could evaporate because of the lower H2O vapor

pressure in the surrounding gas. Additionally, the DBD heats the gas to ~650 °C, which

will increase the rate of evaporation. The direct interaction of droplets with filamentary

100 microdischarges further enhances evaporation, as the microdischarges are substantially hotter than the background gas and contain high-energy ions that can sputter the droplets.200

For our DBD reactor, electrical characterization indicates that there are ~6 million microdischarges per second at a power of 24 W. Assuming all the plasma power passes through these discharges, each microdischarge contains ~4.2 µJ of energy. From the microdischarge frequency and diameter, assumed to be ~100 µm, we can compare the number of microdischarges filling the DBD reactor to the reactor residence time to estimate that each droplet interacts with on average one microdischarge. Based on the latent heat of vaporization for a 1 µm droplet of ~1 nJ, this is more than sufficient for a droplet to be evaporated.

No plasma Heating or heating N2 plasma Ar/N2 plasma to 40 °C (a) (b) (c) (d) (e)

Figure 6.8. (a-d) Images of droplets exiting the reactor with (a) no plasma or heating, (b) pure N2 plasma, (c) 90 % Ar and 10 % N2 plasma and, (d) using the gas heater set to 40

°C. (e) NH3 production rate with (red) and without (blue) using the gas heater with both pure N2 and 90 % Ar with 10 % N2 plasma. The plasma power in each of these conditions is 24 W.

101

The stability of the droplets was assessed by visual analysis at the exit of the reactor.

Figure 6.8a shows an optical image of the droplets without any power applied, i.e., no plasma. The image shows that the droplets are not completely evaporated by the N2 gas

flow alone. In comparison, when the plasma is on, no droplets are visible in the reactor

exit stream (Fig. 6.8b). The water was not completely converted by the reactions, indicating

that gas heating and/or the filaments in the plasma removes the droplets.

We addressed the effect of the DBD on droplet stability by introducing Ar in the

gas feed, which was found to produce a far less filamentary discharge as shown in Fig. 6.9.

Figure 6.8c shows that the H2O droplets are mostly preserved with Ar in the feed even though the gas temperature is 760 ± 80 °C. This result suggests that the filaments in the N2

plasma are responsible for evaporating the droplets.

Figure 6.9. Representative charge waveforms for DBD with inlet feeds of N2 + water droplets (black) and Ar/N2 (90/10) + water droplets (red). The addition of Ar is found to

102

produce much fewer spikes in the charge waveform which originate from filamentary

microdischarges.

If the H2O droplets are completely evaporated at some point during their transport

through the DBD, then it is not clear that liquid-phase H2O plays a role in the reaction, as

has been previously reported for other reaction chemistries.123,124 As a control, the atomized

H2O water droplets were preheated using a gas heater to 40 °C before the plasma; their

complete evaporation was confirmed again by our imaging (Fig. 6.8d). Even though the

gas heater was substantially cooler than the Ar/N2 plasma, the residence time (200 ms) was

two orders of magnitude larger than the reactor, which was sufficient for heating alone to

cause complete evaporation. The NH3 production rates under the different scenarios are

compared in Fig. 6.8e. For N2, a slightly higher NH3 production is found for the evaporated

H2O droplets, i.e., H2O vapor, in comparison to H2O droplets in the inlet stream (two

-3 sample t-test, p-value = 10 ). For a mixture of Ar/N2 (90:10), where the droplets were

found to mostly survive through the DBD, there is no significant difference between the

H2O vapor and H2O droplets (two sample t-test, p-value = 0.24). The power for the Ar/N2

plasma was further decreased, lowering the temperature, to ensure droplet survival, and

still no difference was found in the NH3 production rates between the H2O vapor and H2O,

as shown in figure 6.10. These results suggest that NH3 is formed from gas-phase chemistry

in our reactor and the liquid-phase plays little or no role, which is consistent with several

59,201,202 recent studies of NH3 production from N2 and water. In fact, the slightly higher rate with H2O vapor suggests that the liquid-phase may slow down NH3 formation, either by decreasing the gas temperature or affecting other plasma properties (electron energy

103 distribution, EED, electron density, etc.), or by having fewer H2O molecules accessible for reaction in the gas phase. It is difficult to draw definitive conclusions between the N2 and

Ar/N2 experiments because the addition of Ar changes many properties of the plasma – the previously discussed filament formation, as well as the EED and electron density, concentration of various molecular and atomic states of Ar and N2, and reaction chemistry with H2O.

Figure 6.10. (a) Gas temperature as a function of power obtained from OES analysis of nitrogen second positive system for DBD with inlet feeds of N2 + water droplets (black squares) and Ar/N2 + water droplets (red circles). Production rate of NH3 as a function of power for DBD with Ar/N2 (90/10) as inlet feed with (red circles) and without (black squares) preheating by gas heater off. The open symbols indicate single measurements and closed symbols indicate an average of multiple measurements with error bars corresponding to the standard error.

104

While H2O droplets do not appear to enhance the reaction chemistry, from a

practical point of view, atomizing is a simple approach to introducing H2O into a gas-phase process and our results show that the NH3 production is not substantially affected by the

physical state of the H2O fed into the reactor. In this system, the H2O droplets were dilute because the atomization required a relatively high N2 gas flow rate, but other approaches

to generating H2O droplets such as heating or piezoelectric could be operated independent

of gas flow rate and allow the concentration of H2O droplets in the gas to be further

increased. In contrast, H2O vapor is limited by liquid-vapor equilibrium; for example, at

25 °C, the maximum fraction of H2O would be 3%. Increasing the temperature can drive the equilibrium towards higher vapor concentrations, but would require heating of the reactor to avoid condensation.

6.3.5 Insights into reaction chemistry

The reaction between N2 and H2O is complex because of the simultaneous reduction and

oxidation of N2 to NH3 and NOx species, respectively. While all of these forms of fixed N2

are typically observed in N2 plasmas reacting with H2O, a mechanistic understanding

remains unknown. Some studies suggest that the nitrogen species are first formed in the

201,202 gas phase, which contains H2O vapor, and then subsequently dissolve in the liquid

59 (H2O)-phase where they are deteced. Other studies have shown evidence for NH3

formation in the liquid phase95 or combinations of both the liquid and gas phase.120,122

Almost all previous studies have focused on either a N2 plasma jet contacting a liquid H2O

bath or humid air plasmas. Studies of N2-H2O reactive systems, with liquid aerosols or H2O

vapor, are much fewer. To provide insight into the reaction chemistry for this type of

105

system, control experiments were performed in the reactor with inlet streams of N2 and H2

that should only lead to reduction, and N2 and O2 that should only lead to oxidation. In both cases, the H2 and O2 were added at the same atomic composition as hydrogen and

oxygen in water.

- - Figure 6.11 shows a summary of the NH3, NO2 , and NO3 production rates

measured for either N2/H2 or N2/O2 as the inlet stream in the DBD reactor. It is expected

- - that only NH3 would be produced for N2 and H2, and only NO2 and NO3 would be

- produced for N2 and O2. In agreement, only NH3 was measured for N2 and H2 and no NO2

- -1 /NO3 was detected. The NH3 production rate of 284 ± 46 µg h was not significantly

larger than that for N2 and H2O (two sample t-test, p-value = 0.08). However, for N2 and

- - -1 O2, very little NO2 /NO3 was detected (note the NH3 production rate of 16 ± 4 µg h was

not significantly different than background (two sample t-test, p-value = 0.11)).

- Figure 6.11. Total production rates of fixed nitrogen consisting of NH3 (red), NO2 (green),

- and NO3 (blue) for N2 and H2O, H2, or O2 as the feeds. All experiments were conducted at

a plasma power of 24 W.

106

The detailed mechanisms for plasma activation of N2 are not known and there are very few studies focusing on N2 + H2O gas phase chemistry, as opposed to N2 + H2, air (N2

203,204 + O2), and humid air (N2 + O2 + H2O). Previous mechanistic studies with H2O systems

focused on the reactions in the liquid phase with the plasma only exciting nitrogen.188,205,206

For N2 reacting with H2, the formation of NH3 is believed to occur by dissociation of N2

203,207 and/or H2 followed by additive reactions At atmospheric pressure where the energies

of plasma species such as electrons are reduced by collisions, the concentrations of atomic

nitrogen (N) and hydrogen (H) will be much lower than those of N2 and H2. Furthermore,

the concentration of N is expected to be much lower than that of H because of the higher

bond strength of N2 as compared to H2. For this reason, it is likely that NH3 formation

proceeds by the following:

H2 → 2H (6.1)

N2 + H → NH + N (6.2)

NH + H2 → NH2 + H (6.3)

NH2 + H2 → NH3 + H (6.4)

The importance of N2 for ammonia synthesis was confirmed by previous results that

showed ammonia production remained unchanged when excited states of N were depleted,

205,206 but states of N2. Note that the N2 and H2 in the plasma will often be excited by a

number of modes with the most likely being vibrational.

Based on the similar production rates for NH3 with H2O as a reactant, we suggest

that NH3 is formed by a similar mechanism after H2O vapor is dissociated to H and

hydroxide (OH):

H2O → H + OH (6.5)

107

After formation of H, the mechanism follows the steps in Equation 6.2-6.4. Differences in

NH3 production between H2 and H2O may be correlated with the bond strength of H2 vs.

H2O in the plasma and as previously noted, the competitive oxidative pathway.

For N2 reacting with O2, the well-known Zeldovich mechanism from combustion

likely applies:177,208,209

O2 → 2O (6.6)

N + O → NO + O (6.7)

2 N2 + O → NO + N (6.8)

At atmospheric pressure, Equation 6.8 is more likely than Equation 6.7 because of the

higher bond strength of N2. The NO produced can then be oxidized further by reacting

with O2 or reactive O:

NO + O → NO2 (6.9) 1 2 2 NO + O + M → NO2 + M (6.10)

where M is a third body, most likely N2 in our reaction, required to remove excess energy.

- - The nitrogen oxides can dissolve in H2O to form NO2 and NO3 by the following reaction

pathways:210

NO2 → NO2,(aq) (6.11)

- - + 2NO2,(aq)+H2O → NO2 + NO3 + 2H (6.12)

- - The experiments with N2 and O2 as the inlet stream produced no NO2 and NO3 and the

NO2 produced in the DBD was likely not effectively solubilized in the trap. The gas flow

rates were relatively high and the resulting short contact time in the trap may not have been

sufficient to dissolve the gas, and the pH may not have been optimal for trapping these

species.188

108

In contrast, in N2 and H2O systems, OH is also formed, as shown in equation 6.5,

and can still lead to the oxidation of nitrogen into NO:211,212

N + OH → NO + H (6.13)

Further, plasma studies of humid air show that nitrogen oxides could react with OH to form

177,211–213 HNO2 and HNO3 in the gas phase:

NO + OH → HNO2 (6.14)

NO2 + OH → HNO3 (6.15)

- - These vapors can subsequently dissolve in H2O from a trap to form NO2 and NO3 :

HNO2 → HNO2,(aq) (6.16)

HNO3 → HNO3(aq) (6.17)

- + HNO2,(aq)→ NO2 + H (6.18)

- + HNO3,(aq) → NO3 + H (6.19)

The results with the N2 and H2O inlet streams indicate that HNO2 and HNO3 vapors are

- - effectively solubilized in the trap leading to the formation of NO2 and NO3 .

Several alternative pathways exist for the various products in this system. For

example, it is possible that NH3 or NOx can decompose into the other product. In order to

- test this, either NH3 (100 mM (NH4)2SO4) or NO3 (100 mM NaNO3) were added to an Ar plasma, measuring either NOx or NH3, respectively. Adding NH3 did not result in the

- - production of either NO2 or NO3 , showing that there is not pathway from NH3 to NOx in

- -1 this system. However, when NO3 was added, ammonia was produced (79 ± 22 µg h ),

indicating that there is a pathway decomposing NOx into NH3. Even with this additional pathway, there still must be a predominate pathway from H2O and N2, as NH3 is formed

using H2, where there can be no NOx present.

109

This study demonstrates that H2O is capable of replacing H2 as a feedstock for N2

activation in a plasma process which is environmentally friendly, safe, and inexpensive.

For the DBD reactor, the NH3 production rate with H2O in the feed is similar to that with

H2. Other studies have reported higher NH3 production from N2 and H2 using different plasma reactors or incorporating catalysts, but with optimization and/or the use of catalysts, the NH3 production rate could be further increased from N2 and H2O in this system. An additional benefit of using H2O feed, rather than H2, is that other fixed and oxidized forms

- of nitrogen are also produced. These products are also valuable chemicals as NO3 can be

214 - used to make fertilizer (usually as ammonium nitrate) and NO2 is a preservative and chemical precursor for dyes.215 For the DBD reactor, these oxidized products are not

formed or collected from plasma activation of N2 and O2, which shows the unique chemistry

possible with H2O.

6.4 Conclusions

This chapter has successfully demonstrated that N2 and water droplets can be directly

reacted in a DBD reactor to continuously produce NH3 at atmospheric pressure. The

droplets are evaporated by the filaments in the DBD, which appears to be critical to

generating the water vapor necessary for the gas-phase chemistry. Interestingly, the NH3

production rate with H2O as the feed is not that different from H2 in the same reactor.

Further improvements may be achieved by optimizing the plasma design or introducing

catalysts as in the case of other N2/H2 plasma systems. In contrast to H2, oxidized forms

- of nitrogen such as NO3 are also generated when water is used as the feed.

110

Chapter 7: Future work

This chapter aims to discuss possible future work for each of the areas of research discussed thus far in this thesis.

7.1 Triboelectric charging of identical insulators

Chapter 2 discussed how humidity affects the size-dependence of charging for insulating particles of identical chemical composition. The results were also fit into a non-equilibrium model to help build an understanding of how these particles can charge. These results focused on glass beads, but can be expanded with other materials. For example, nylon particles are more hydrophilic and may be more susceptible to the effects of humidity, whereas polytetrafluoroethylene (PTFE) is much more hydrophobic. The different interactions of these materials with the water on their surface may lead to different results than observed with glass, based on the non-equilibrium model.

7.2 Electrostatic effects on dust transport

Chapter 3 shows how electric fields can affect the transport of particles through the atmosphere by causing larger particles to remain lofted that would otherwise fall out of the flow. The exact electrostatic force on the particles is still unknown, without exact measurements of the electric fields or particle charge. To build a more robust model, the electric field strength throughout the region that the particles travel must be measured, as well as the charge on the particles themselves. The difficulty here lies in measuring the charge on individual particles, rather than the net charge of large groups of particles.

111

Simply separating the particles first based on charge or size may be enough to adequately

characterize the charge distribution of the particles.

7.3 Direct, non-oxidative plasma conversion of methane

Chapter 4 shows that for methane conversion in a dielectric barrier discharge, the rate of

conversion depends on the relationship between power and flow rate, but not on reactor

volume. This is believed to be generally applicable to other systems and reactions, but

would require further results to verify this effect. Further, as this relationship is useful for

scaling the system, larger scale reactors can be tested to determine the effectivity of this

relationship as a scaling factor.

7.4 Hydrogen gas evolution at an electrified plasma-water interface

Chapter 5 shows that hydrogen evolution in cathodic glow discharge electrolysis behaves

non-faradaically due to reactions in the plasma phase on the electrolyte leading to hydrogen generation. This study can be expanded by measuring the reactive species, perhaps by using spectroscopic techniques. This would give useful information towards building a full model of the process, which can enable a deeper understanding of the non-faradaic behavior.

7.5 Continuous, process-intensified nitrogen fixation in a plasma-water droplet reactor

Chapter 6 shows a system that uses a plasma to fix nitrogen by using water as a reactant.

Currently, this process is highly energy intensive, but still has significant room to improve.

As mentioned in Section 6.3.3, the ratio of water to nitrogen has a significant amount of room for improvement. Further, preliminary results with catalysts show that significant

112

improvements can be made in both production rate, and ammonia selectivity. Catalysts optimization could be used to make this system potentially viable for industrial applications.

113

References

(1) IPCC, 2018: Global Warming of 1.5°C An IPCC Special Report on the Impacts of

Global Warming of 1.5°C above Pre-Industrial Levels and Related Global

Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global

Response to the Threat of Clim; Masson-Delmotte, V., Zhai, P., Pörtner, H.-O.,

Roberts, D., Skea, J., Shukla, P. R., Pirani, A., Moufouma-Okia, W., Péan, C.,

Pidcock, R., et al., Eds.; 2019.

(2) IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working

Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel

on Climate Change; Pachauri, R. K., Meyer, L. A., Eds.; Cambridge University

Press: Geneva, Switzerland, 2015.

(3) Hendrickson, G. Electrostatics and Gas Phase Fluidized Bed Polymerization

Reactor Wall Sheeting. Chem. Eng. Sci. 2006, 61 (4), 1041–1064.

https://doi.org/10.1016/j.ces.2005.07.029.

(4) Rudge, W. A. D. Atmospheric Electrification during South African Dust Storms.

Nature 1913, 91, 31–32.

(5) Schmidt, D. S.; Schmidt, R. A.; Dent, J. D. Electrostatic Force on Saltating Sand.

J. Geophys. Res. Atmos. 1998, 103 (D8), 8997–9001.

https://doi.org/10.1029/98JD00278.

(6) Tegen, I. Modeling the Mineral Dust Aerosol Cycle in the Climate System. Quat.

Sci. Rev. 2003, 22, 1821–1834. https://doi.org/10.1016/S0277-3791(03)00163-X.

(7) Kok, J. F.; Ward, D. S.; Mahowald, N. M.; Evan, A. T. Global and Regional

Importance of the Direct Dust-Climate Feedback. Nat. Commun. 2018, 9.

114

https://doi.org/10.1038/s41467-017-02620-y.

(8) Toth, J. R.; Rajupet, S.; Squire, H.; Volbers, B.; Zhou, J.; Xie, L.; Sankaran, R. M.;

Lacks, D. J. Electrostatic Forces Alter Particle Size Distributions in Atmospheric

Dust. Atmos. Chem. Phys. 2020, 20 (5), 3181–3190. https://doi.org/10.5194/acp-

20-3181-2020.

(9) Sims, R. A.; Biris, A. S.; Wilson, J. D.; Yurteri, C. U.; Mazumder, M. K.; Calle, C.

I.; Buhler, C. R. Development of a Transparent Self-Cleaning Dust Shield for

Solar Panels. Proc. ESA-IEEE Jt. Meet. Electrost. 2003, No. 1, 814–821.

(10) Calle, C. I.; Buhler, C. R.; McFall, J. L.; Snyder, S. J. Particle Removal by

Electrostatic and Dielectrophoretic Forces for Dust Control during Lunar

Exploration Missions. J. Electrostat. 2009, 67 (2–3), 89–92.

https://doi.org/10.1016/j.elstat.2009.02.012.

(11) Sharma, R.; Wyatt, C. A.; Zhang, J.; Calle, C. I.; Mardesich, N.; Mazumder, M. K.

Experimental Evaluation and Analysis of Electrodynamic Screen as Dust

Mitigation Technology for Future Mars Missions. IEEE Trans. Ind. Appl. 2009, 45

(2), 591–596. https://doi.org/10.1109/TIA.2009.2013542.

(12) Bilici, M.; Dascalescu, L.; Dragan, C.; Fati, O.; Iuga, A.; Samuila, A.

Tribocharging and Electrostatic Separation of Mixed Granular Solids in Fluidized

Bed Devices. IEEE Trans. Dielectr. Electr. Insul. 2011, 18 (5), 1476–1483.

https://doi.org/10.1109/TDEI.2011.6032818.

(13) Wu, G.; Li, J.; Xu, Z. Triboelectrostatic Separation for Granular Plastic Waste

Recycling: A Review. Waste Manag. 2013, 33 (3), 585–597.

https://doi.org/10.1016/j.wasman.2012.10.014.

115

(14) Inculet, I. I.; Castle, G. S. P.; Brown, J. D. Electrostatic Separation of Plastics for

Recycling. Part. Sci. Technol. 1998, 16 (1), 91–100.

https://doi.org/10.1080/02726359808906787.

(15) Lowell, J.; Rose-Innes, A. C. Contact Electrification. Adv. Phys. 1980, 29 (6),

947–1023. https://doi.org/10.1080/00018738000101466.

(16) Lacks, D. J.; Sankaran, R. M. Contact Electrification of Insulating Materials. J.

Phys. D. Appl. Phys. 2011, 44 (45), 453001. https://doi.org/10.1088/0022-

3727/44/45/453001.

(17) Zhao, H.; Castle, G. S. P.; Inculet, I. I. The Measurement of Bipolar Charge in

Polydisperse Powders Using a Vertical Array of Faraday Pail Sensors. J.

Electrostat. 2002, 55 (3–4), 261–278. https://doi.org/10.1016/S0304-

3886(01)00209-1.

(18) Huiliang Zhao; Castle, G. S. P.; Inculet, I. I.; Bailey, A. G. Bipolar Charging of

Poly-Disperse Polymer Powders in Fluidized Beds. IEEE Trans. Ind. Appl. 2003,

39 (3), 612–618. https://doi.org/10.1109/TIA.2003.810663.

(19) Trigwell, S.; Grable, N.; Yurteri, C. U.; Sharma, R.; Mazumder, M. K. Effects of

Surface Properties on the Tribocharging Characteristics of Polymer Powder as

Applied to Industrial Processes. IEEE Trans. Ind. Appl. 2003, 39 (1), 79–86.

https://doi.org/10.1109/TIA.2002.807228.

(20) Forward, K. M.; Lacks, D. J.; Sankaran, R. M. Charge Segregation Depends on

Particle Size in Triboelectrically Charged Granular Materials. Phys. Rev. Lett.

2009, 102 (2), 1–4. https://doi.org/10.1103/PhysRevLett.102.028001.

(21) Forward, K. M.; Lacks, D. J.; Sankaran, R. M. Triboelectric Charging of Granular

116

Insulator Mixtures Due Solely to Particle - Particle Interactions. Ind. Eng. Chem.

Res. 2009, 48 (5), 2309–2314. https://doi.org/10.1021/ie8004786.

(22) Forward, K. M.; Lacks, D. J.; Sankaran, R. M. Particle-Size Dependent Bipolar

Charging of Martian Regolith Simulant. Geophys. Res. Lett. 2009, 36 (13), 1–5.

https://doi.org/10.1029/2009GL038589.

(23) Merrison, J. P.; Gunnlaugsson, H. P.; Hogg, M. R.; Jensen, M.; Lykke, J. M.;

Madsen, M. B.; Nielsen, M. B.; Nørnberg, P.; Ottosen, T. A.; Pedersen, R. T.; et

al. Factors Affecting the Electrification of Wind-Driven Dust Studied with

Laboratory Simulations. Planet. Space Sci. 2012, 60 (1), 328–335.

https://doi.org/10.1016/j.pss.2011.10.008.

(24) Waitukaitis, S. R.; Lee, V.; Pierson, J. M.; Forman, S. L.; Jaeger, H. M. Size-

Dependent Same-Material Tribocharging in Insulating Grains. Phys. Rev. Lett.

2014, 112 (21), 218001. https://doi.org/10.1103/PhysRevLett.112.218001.

(25) Lee, V.; Waitukaitis, S. R.; Miskin, M. Z.; Jaeger, H. M. Direct Observation of

Particle Interactions and Clustering in Charged Granular Streams. Nat. Phys. 2015,

11 (9), 733–737. https://doi.org/10.1038/nphys3396.

(26) Bilici, M. A.; Toth, J. R.; Sankaran, R. M.; Lacks, D. J. Particle Size Effects in

Particle-Particle Triboelectric Charging Studied with an Integrated Fluidized Bed

and Electrostatic Separator System. Rev. Sci. Instrum. 2014, 85 (10), 103903.

https://doi.org/10.1063/1.4897182.

(27) Lacks, D. J.; Levandovsky, A. Effect of Particle Size Distribution on the Polarity

of Triboelectric Charging in Granular Insulator Systems. J. Electrostat. 2007, 65

(2), 107–112. https://doi.org/10.1016/j.elstat.2006.07.010.

117

(28) Lacks, D. J.; Duff, N.; Kumar, S. K. Nonequilibrium Accumulation of Surface

Species and Triboelectric Charging in Single Component Particulate Systems.

Phys. Rev. Lett. 2008, 100 (18), 188305.

https://doi.org/10.1103/PhysRevLett.100.188305.

(29) Lowell, J.; Truscott, W. S. Triboelectrification of Identical Insulators : 11 . Theory

and Further Experiments. 1986, 19, 1281–1298.

(30) Sow, M.; Widenor, R.; Akande, A. R.; Robinson, K. S.; Sankaran, R. M.; Lacks,

D. J. The Role of Humidity on the Lift-off of Particles in Electric Fields. J. Braz.

Chem. Soc. 2013, 24 (2), 273–279. https://doi.org/10.5935/0103-5053.20130035.

(31) Wiles, J. A.; Fialkowski, M.; Radowski, M. R.; Whitesides, G. M.; Grzybowski, B.

A. Effects of Surface Modification and Moisture on the Rates of Charge Transfer

between Metals and Organic Materials. J. Phys. Chem. B 2004, 108 (52), 20296–

20302. https://doi.org/10.1021/jp0457904.

(32) Pence, S.; Novotny, V. J.; Diaz, A. F. Effect of Surface Moisture on Contact

Charge of Polymers Containing Ions. Langmuir 1994, 10 (2), 592–596.

https://doi.org/10.1021/la00014a042.

(33) Nomura, T.; Satoh, T.; Masuda, H. The Environment Humidity Effect on the

Tribo-Charge of Powder. Powder Technol. 2003, 135–136, 43–49.

https://doi.org/10.1016/S0032-5910(03)00157-8.

(34) Schella, A.; Herminghaus, S.; Schröter, M. Influence of Humidity on Tribo-

Electric Charging and Segregation in Shaken Granular Media. Soft Matter 2017,

13 (2), 394–401. https://doi.org/10.1039/c6sm02041k.

(35) Toth, J. R.; Phillips, A. K.; Rajupet, S.; Sankaran, R. M.; Lacks, D. J. Particle-

118

Size-Dependent Triboelectric Charging in Single-Component Granular Materials:

Role of Humidity. Ind. Eng. Chem. Res. 2017, 56 (35), 9839–9845.

https://doi.org/10.1021/acs.iecr.7b02328.

(36) Hansen, J. Earth’s Energy Imbalance: Confirmation and Implications. Science (80-

. ). 2005, 308 (5727), 1431–1435. https://doi.org/10.1126/science.1110252.

(37) Mahowald, N. Aerosol Indirect Effect on Biogeochemical Cycles and Climate.

Science (80-. ). 2011, 334 (6057), 794–796.

https://doi.org/10.1126/science.1207374.

(38) Bagnold, R. A. A. The Physics of Blown Sand and Desert Dunes; Springer

Netherlands: Dordrecht, 1941; Vol. 148. https://doi.org/10.1038/148480a0.

(39) Kok, J. F.; Parteli, E. J. R.; Michaels, T. I.; Karam, D. B. The Physics of Wind-

Blown Sand and Dust. Reports Prog. Phys. 2012, 75 (10), 106901.

https://doi.org/10.1088/0034-4885/75/10/106901.

(40) Betzer, P. R.; Carder, K. L.; Duce, R. A.; Merrill, J. T.; Tindale, N. W.; Uematsu,

M.; Costello, D. K.; Young, R. W.; Feely, R. A.; Breland, J. A.; et al. Long–Range

Transport of Giant Mineral Aerosol Particles. Nature 1988, 336, 568–571.

https://doi.org/10.1038/336568a0.

(41) Paul, G.; M., P. J.; E., G. T.; Christina, H. N.; Ming, Z. Global‐scale Attribution of

Anthropogenic and Natural Dust Sources and Their Emission Rates Based on

MODIS Deep Blue Aerosol Products. Rev. Geophys. 2012, 50, 1–36.

https://doi.org/10.1029/2012RG000388.

(42) Xi, X.; Sokolik, I. N. Quantifying the Anthropogenic Dust Emission from

Agricultural Land Use and Desiccation of the Aral Sea in Central Asia. J.

119

Geophys. Res. Atmos. 2016, 121 (20), 12,270-12,281.

https://doi.org/10.1002/2016JD025556.

(43) Webb, N. P.; Pierre, C. Quantifying Anthropogenic Dust Emissions. Earth’s

Futur. 2018, 6 (2), 286–295. https://doi.org/10.1002/2017EF000766.

(44) Mahowald, N.; Albani, S.; Kok, J. F.; Engelstaeder, S.; Scanza, R.; Ward, D. S.;

Flanner, M. G. The Size Distribution of Desert Dust Aerosols and Its Impact on

the Earth System. Aeolian Res. 2014, 15, 53–71.

https://doi.org/10.1016/j.aeolia.2013.09.002.

(45) Tegen, I.; Lacis, A. A. Modeling of Particle Size Distribution and Its Influence on

the Radiative Properties of Mineral Dust Aerosol. J. Geophys. Res. Atmos. 1996,

101, 19237–19244. https://doi.org/10.1029/95jd03610.

(46) Kok, J. F.; Ridley, D. A.; Zhou, Q.; Miller, R. L.; Zhao, C.; Heald, C. L.; Ward, D.

S.; Albani, S.; Haustein, K. Smaller Desert Dust Cooling Effect Estimated from

Analysis of Dust Size and Abundance. Nat. Geosci. 2017, 10, 274–278.

https://doi.org/10.1038/ngeo2912.

(47) Otto, S.; De Reus, M.; Trautmann, T.; Thomas, A.; Wendisch, M.; Borrmann, S.

Atmospheric Radiative Effects of an in Situ Measured Saharan Dust Plume and the

Role of Large Particles. Atmos. Chem. Phys. 2007, 7, 4887–4903.

https://doi.org/10.5194/acp-7-4887-2007.

(48) Maring, H. Mineral Dust Aerosol Size Distribution Change during Atmospheric

Transport. J. Geophys. Res. 2003, 108, 8592.

https://doi.org/10.1029/2002JD002536.

(49) Reid, J. S. Analysis of Measurements of Saharan Dust by Airborne and Ground-

120

Based Remote Sensing Methods during the Puerto Rico Dust Experiment

(PRIDE). J. Geophys. Res. 2003, 108, 8586.

https://doi.org/10.1029/2002JD002493.

(50) Haarig, M.; Ansmann, A.; Althausen, D.; Klepel, A.; Groß, S.; Freudenthaler, V.;

Toledano, C.; Mamouri, R.-E. E.; Farrell, D. A.; Prescod, D. A.; et al. Triple-

Wavelength Depolarization-Ratio Profiling of Saharan Dust over Barbados during

SALTRACE in 2013 and 2014. Atmos. Chem. Phys. 2017, 17 (17), 10767–10794.

https://doi.org/10.5194/acp-17-10767-2017.

(51) Renard, J.-B. B.; Dulac, F.; Durand, P.; Bourgeois, Q.; Denjean, C.; Vignelles, D.;

Couté, B.; Jeannot, M.; Verdier, N.; Mallet, M. In Situ Measurements of Desert

Dust Particles above the Western Mediterranean Sea with the Balloon-Borne Light

Optical Aerosol Counter/Sizer (LOAC) during the ChArMEx Campaign of

Summer 2013. Atmos. Chem. Phys. 2018, 18 (5), 3677–3699.

https://doi.org/10.5194/acp-18-3677-2018.

(52) van der Does, M.; Knippertz, P.; Zschenderlein, P.; Giles Harrison, R.; Stuut, J.-B.

W. The Mysterious Long-Range Transport of Giant Mineral Dust Particles. Sci.

Adv. 2018, 4 (12), 1–8. https://doi.org/10.1126/sciadv.aau2768.

(53) Kok, J. F.; Renno, N. O. Enhancement of the Emission of Mineral Dust Aerosols

by Electric Forces. Geophys. Res. Lett. 2006, 33 (19), L19S10.

https://doi.org/10.1029/2006GL026284.

(54) Kok, J. F.; Renno, N. O. The Effects of Electric Forces on Dust Lifting:

Preliminary Studies with a Numerical Model. J. Phys. Conf. Ser. 2008, 142,

012047. https://doi.org/10.1088/1742-6596/142/1/012047.

121

(55) Esposito, F.; Molinaro, R.; Popa, C. I.; Molfese, C.; Cozzolino, F.; Marty, L.; Taj-

Eddine, K.; Di Achille, G.; Franzese, G.; Silvestro, S.; et al. The Role of the

Atmospheric Electric Field in the Dust-Lifting Process. Geophys. Res. Lett. 2016,

43 (10), 5501–5508. https://doi.org/10.1002/2016GL068463.

(56) Abbas, H. F.; Wan Daud, W. M. A. Hydrogen Production by Methane

Decomposition: A Review. Int. J. Hydrogen Energy 2010, 35 (3), 1160–1190.

https://doi.org/10.1016/j.ijhydene.2009.11.036.

(57) Šimek, M.; Pekárek, S.; Prukner, V. Influence of Power Modulation on Ozone

Production Using an AC Surface Dielectric Barrier Discharge in Oxygen. Plasma

Chem. Plasma Process. 2010, 30 (5), 607–617. https://doi.org/10.1007/s11090-

010-9245-4.

(58) Kim, H. H.; Teramoto, Y.; Ogata, A.; Takagi, H.; Nanba, T. Plasma Catalysis for

Environmental Treatment and Energy Applications. Plasma Chem. Plasma

Process. 2016, 36 (1), 45–72. https://doi.org/10.1007/s11090-015-9652-7.

(59) Gorbanev, Y.; Vervloessem, E.; Nikiforov, A.; Bogaerts, A. Nitrogen Fixation

with Water Vapor by Nonequilibrium Plasma: Toward Sustainable Ammonia

Production. ACS Sustain. Chem. Eng. 2020, 8, 2996–3004.

https://doi.org/10.1021/acssuschemeng.9b07849.

(60) Fincke, J. R.; Anderson, R. P.; Hyde, T. A.; Detering, B. A. Plasma Pyrolysis of

Methane to Hydrogen and Carbon Black. Ind. Eng. Chem. Res. 2002, 41 (6),

1425–1435. https://doi.org/10.1021/ie010722e.

(61) Fincke, J. R.; Anderson, R. P.; Hyde, T.; Detering, B. A.; Wright, R.; Bewley, R.

L.; Haggard, D. C.; Swank, W. D. Plasma Thermal Conversion of Methane to

122

Acetylene. Plasma Chem. Plasma Process. 2002, 22 (1), 107–138.

https://doi.org/10.1023/A:1012944615974.

(62) Zhao, G.-B.; John, S.; Zhang, J.-J.; Wang, L.; Muknahallipatna, S.; Hamann, J. C.;

Ackerman, J. F.; Argyle, M. D.; Plumb, O. a. Methane Conversion in Pulsed

Corona Discharge Reactors. Chem. Eng. J. 2006, 125 (2), 67–79.

https://doi.org/10.1016/j.cej.2006.08.008.

(63) Yao, S. L.; Suzuki, E.; Meng, N.; Nakayama, A. A High-Efficiency Reactor for the

Pulsed Plasma Conversion of Methane. Plasma Chem. Plasma Process. 2002, 22

(2), 225–237. https://doi.org/10.1023/A:1014843425384.

(64) Indarto, A. Hydrogen Production from Methane in a Dielectric Barrier Discharge

Using Oxide Zinc and Chromium as Catalyst. J. Chinese Inst. Chem. Eng. 2008,

39 (1), 23–28. https://doi.org/10.1016/j.jcice.2007.10.001.

(65) Liu, S. Y.; Mei, D. H.; Shen, Z.; Tu, X. Nonoxidative Conversion of Methane in a

Dielectric Barrier Discharge Reactor: Prediction of Reaction Performance Based

on Neural Network Model. J. Phys. Chem. C 2014, 118 (20), 10686–10693.

https://doi.org/10.1021/jp502557s.

(66) Yang, Y. Direct Non-Oxidative Methane Conversion by Non-Thermal Plasma:

Experimental Study. Plasma Chem. Plasma Process. 2003, 23 (2), 283–296.

https://doi.org/10.1023/A:1022968002315.

(67) Scapinello, M.; Delikonstantis, E.; Stefanidis, G. D. The Panorama of Plasma-

Assisted Non-Oxidative Methane Reforming. Chem. Eng. Process. Process

Intensif. 2017, 117 (February), 120–140.

https://doi.org/10.1016/j.cep.2017.03.024.

123

(68) Gallon, H. J.; Tu, X.; Twigg, M. V.; Whitehead, J. C. Plasma-Assisted Methane

Reduction of a NiO Catalyst-Low Temperature Activation of Methane and

Formation of Carbon Nanofibres. Appl. Catal. B Environ. 2011, 106 (3–4), 616–

620. https://doi.org/10.1016/j.apcatb.2011.06.023.

(69) Levenspiel, O. Chemical Reaction Engineering, Third Edit.; John Wiley & Sons:

Hoboken, 1999.

(70) Fogler, H. S. Essentials of Chemical Reaction Engineering; Pearson Education,

Inc.: Boston, 2011.

(71) Ma, Y.; Wang, X.; Ning, P.; Cheng, C.; Xu, K.; Wang, F.; Bian, Z.; Yan, S.

Conversion of COS by Corona Plasma and the Effect of Simultaneous Removal of

COS and Dust. Chem. Eng. J. 2016, 290, 328–334.

https://doi.org/10.1016/j.cej.2016.01.052.

(72) Wang, H. Y.; Lua, A. C. Methane Decomposition Using Ni-Cu Alloy Nano-

Particle Catalysts and Catalyst Deactivation Studies. Chem. Eng. J. 2015, 262,

1077–1089. https://doi.org/10.1016/j.cej.2014.10.063.

(73) Tu, X.; Gallon, H. J.; Twigg, M. V; Gorry, P. A.; Whitehead, J. C. Dry Reforming

of Methane over a Ni/Al2O3 Catalyst in a Coaxial Dielectric Barrier Discharge

Reactor. J. Phys. D. Appl. Phys. 2011, 44 (27), 274007.

https://doi.org/10.1088/0022-3727/44/27/274007.

(74) Kado, S.; Sekine, Y.; Nozaki, T.; Okazaki, K. Diagnosis of Atmospheric Pressure

Low Temperature Plasma and Application to High Efficient Methane Conversion.

Catal. Today 2004, 89 (1–2), 47–55. https://doi.org/10.1016/j.cattod.2003.11.036.

(75) Gibalov, V.; Pietsch, G. The Development of Dielectric Barrier Discharges in Gas

124

Gaps and on Surfaces. J. Phys. D. Appl. Phys. 2000, 33 (20), 2618–2636.

https://doi.org/10.1088/0022-3727/33/20/315.

(76) Hoskinson, A. R.; Oksuz, L.; Hershkowitz, N. Microdischarge Propagation and

Expansion in a Surface Dielectric Barrier Discharge. Appl. Phys. Lett. 2008, 93

(22), 221501. https://doi.org/10.1063/1.3039805.

(77) Eliasson, B.; Kogelschatz, U. Modeling and Applications of Silent Discharge

Plasmas. IEEE Trans. Plasma Sci. 1991, 19 (2), 309–323.

https://doi.org/10.1109/27.106829.

(78) Kogelschatz, U.; Eliasson, B.; Egli, W. Dielectric-Barrier Discharges. Principle

and Applications. Le J. Phys. IV 1997, 07 (C4), C4-47-C4-66.

https://doi.org/10.1051/jp4:1997405.

(79) Fridman, A.; Chirokov, A.; Gutsol, A. Non-Thermal Atmospheric Pressure

Discharges. J. Phys. D. Appl. Phys. 2005, 38 (2), R1–R24.

https://doi.org/10.1088/0022-3727/38/2/R01.

(80) Xu, X. Dielectric Barrier Discharge — Properties and Applications. Thin Solid

Films 2001, 390 (1–2), 237–242. https://doi.org/10.1016/S0040-6090(01)00956-7.

(81) Gentile, A. C.; Kushner, M. J. Reaction Chemistry and Optimization of Plasma

Remediation of N x O y from Gas Streams. J. Appl. Phys. 1995, 78 (3), 2074–

2085. https://doi.org/10.1063/1.360185.

(82) Dorai, R.; Kushner, M. J. Repetitively Pulsed Plasma Remediation of NO x in Soot

Laden Exhaust Using Dielectric Barrier Discharges. J. Phys. D. Appl. Phys. 2002,

35 (22), 2954–2968. https://doi.org/10.1088/0022-3727/35/22/310.

(83) Snoeckx, R.; Aerts, R.; Tu, X.; Bogaerts, A. Plasma-Based Dry Reforming: A

125

Computational Study Ranging from the Nanoseconds to Seconds Time Scale. J.

Phys. Chem. C 2013, 4 (10), 4957–4970. https://doi.org/10.1021/jp311912b.

(84) Toth, J. R.; Shen, X.; Lacks, D. J.; Sankaran, R. M. Reaction Conversion for a

Plasma-Based Steady-State Flow Process Is Independent of Reactor Volume. Ind.

Eng. Chem. Res. 2018, 57 (18), 6048–6056.

https://doi.org/10.1021/acs.iecr.7b05091.

(85) Akolkar, R.; Sankaran, R. M. Charge Transfer Processes at the Interface between

Plasmas and Liquids. J. Vac. Sci. Technol. A Vacuum, Surfaces, Film. 2013, 31

(5), 050811. https://doi.org/10.1116/1.4810786.

(86) Harada, K.; Suzuki, S. Formation of Amino Acids from Elemental Carbon by

Contact Glow Discharge Electrolysis. Nature 1977, 266 (5599), 275–276.

https://doi.org/10.1038/266275a0.

(87) Toriyabe, Y.; Watanabe, S.; Yatsu, S.; Shibayama, T.; Mizuno, T. Controlled

Formation of Metallic Nanoballs during Plasma Electrolysis. Appl. Phys. Lett.

2007, 91 (4), 1–4. https://doi.org/10.1063/1.2760042.

(88) Chen, Q.; Kaneko, T.; Hatakeyama, R. Rapid Synthesis of Water-Soluble Gold

Nanoparticles with Control of Size and Assembly Using Gas-Liquid Interfacial

Discharge Plasma. Chem. Phys. Lett. 2012, 521, 113–117.

https://doi.org/10.1016/j.cplett.2011.11.065.

(89) Kareem, T. A.; Kaliani, A. A. Glow Discharge Plasma Electrolysis for

Nanoparticles Synthesis. Ionics (Kiel). 2012, 18 (3), 315–327.

https://doi.org/10.1007/s11581-011-0639-y.

(90) Ghosh, S.; Hawtof, R.; Rumbach, P.; Go, D. B.; Akolkar, R.; Sankaran, R. M.

126

Quantitative Study of Electrochemical Reduction of Ag + to Ag Nanoparticles in

Aqueous Solutions by a Plasma Cathode . J. Electrochem. Soc. 2017, 164 (13),

D818–D824. https://doi.org/10.1149/2.0541713jes.

(91) Wang, X.; Zhou, M.; Jin, X. Application of Glow Discharge Plasma for

Wastewater Treatment. Electrochim. Acta 2012, 83, 501–512.

https://doi.org/10.1016/j.electacta.2012.06.131.

(92) Sakiyama, Y.; Tomai, T.; Miyano, M.; Graves, D. B. Disinfection of E. Coli by

Nonthermal Microplasma Electrolysis in Normal Saline Solution. Appl. Phys. Lett.

2009, 94 (16). https://doi.org/10.1063/1.3122148.

(93) Yan, Z.; Chen, L.; Wang, H. Hydrogen Generation by Glow Discharge Plasma

Electrolysis of Ethanol Solutions. J. Phys. D. Appl. Phys. 2008, 41 (15).

https://doi.org/10.1088/0022-3727/41/15/155205.

(94) Rumbach, P.; Xu, R.; Go, D. B. Electrochemical Production of Oxalate and

Formate from CO 2 by Solvated Electrons Produced Using an Atmospheric-

Pressure Plasma. J. Electrochem. Soc. 2016, 163 (10), F1157–F1161.

https://doi.org/10.1149/2.0521610jes.

(95) Hawtof, R.; Ghosh, S.; Guarr, E.; Xu, C.; Mohan Sankaran, R.; Renner, J. N.

Catalyst-Free, Highly Selective Synthesis of Ammonia from Nitrogen and Water

by a Plasma Electrolytic System. Sci. Adv. 2019, 5 (1), eaat5778.

https://doi.org/10.1126/sciadv.aat5778.

(96) Richmonds, C.; Witzke, M.; Bartling, B.; Lee, S. W.; Wainright, J.; Liu, C.;

Sankaran, R. M. Electron-Transfer Reactions at the Plasma–Liquid Interface. J.

Am. Chem. Soc. 2011, 133 (44), 17582–17585. https://doi.org/10.1021/ja207547b.

127

(97) Witzke, M.; Rumbach, P.; Go, D. B.; Sankaran, R. M. Evidence for the

Electrolysis of Water by Atmospheric-Pressure Plasmas Formed at the Surface of

Aqueous Solutions. J. Phys. D. Appl. Phys. 2012, 45 (44), 442001.

https://doi.org/10.1088/0022-3727/45/44/442001.

(98) Hickling, A.; Ingram, M. D. Contact Glow-Discharge Electrolysis. Trans. Faraday

Soc. 1964, 60, 783. https://doi.org/10.1039/tf9646000783.

(99) Palit, S. R. A Simple Experimental Demonstration of the Breakdown of Faraday’s

Law of Electrolysis. Indian J. Phys. 1967, 41, 782–784.

(100) Palit, S. R. Liberation of Hydrogen and Oxygen Together on the Electrodes during

Electrolysis Accompanied by Electrode Glow. Indian J. Phys. 1968, 42, 414–418.

(101) Sengupta, S. K.; Singh, O. P. Contact Glow Discharge Electrolysis: A Study of Its

Chemical Yields in Aqueous Inert-Type Electrolytes. J. Electroanal. Chem. 1994,

369 (1–2), 113–120. https://doi.org/10.1016/0022-0728(94)87089-6.

(102) Mizuno, T.; Akimoto, T.; Azumi, K.; Ohmori, T.; Aoki, Y.; Takahashi, A.

Hydrogen Evolution by Plasma Electrolysis in Aqueous Solution. Jpn. J. Appl.

Phys. 2005, 44 (1A), 396–401. https://doi.org/10.1143/JJAP.44.396.

(103) Hickling, A.; Ingram, M. D. Glow-Discharge Electrolysis. J. Electroanal. Chem.

1964, 8 (1), 65–81. https://doi.org/10.1016/0022-0728(64)80039-5.

(104) Hickling, A. Electrochemical Processes in Glow Discharge at the Gas-Solution

Interface. In Modern Aspects of Electrochemistry No. 6; Springer US: Boston,

MA, 1971; pp 329–373. https://doi.org/10.1007/978-1-4684-3000-4_5.

(105) Denaro, A. R.; Hough, K. O. Glow-Discharge Electrolysis of Sulphuric Acid

Solutions. Electrochim. Acta 1972, 17 (3), 549–559. https://doi.org/10.1016/0013-

128

4686(72)80054-9.

(106) Sen Gupta, S. K. Contact Glow Discharge Electrolysis: Its Origin, Plasma

Diagnostics and Non-Faradaic Chemical Effects. Plasma Sources Sci. Technol.

2015, 24 (6), 63001. https://doi.org/10.1088/0963-0252/24/6/063001.

(107) Sen Gupta, S. K.; Singh, R. Cathodic Contact Glow Discharge Electrolysis: Its

Origin and Non-Faradaic Chemical Effects. Plasma Sources Sci. Technol. 2016, 26

(1), 015005. https://doi.org/10.1088/0963-0252/26/1/015005.

(108) Toth, J. R.; Hawtof, R.; Matthiesen, D.; Renner, J. N.; Sankaran, R. M. On the

Non-Faradaic Hydrogen Gas Evolution from Electrolytic Reactions at the Interface

of a Cathodic Atmospheric-Pressure Microplasma and Liquid Water Surface. J.

Electrochem. Soc. 2020, 167 (11), 116504. https://doi.org/10.1149/1945-

7111/aba15c.

(109) Haber, F.; van Oordt, G. Über Die Bildung von Ammoniak Den Elementen.

Zeitschrift für Anorg. Chemie 1905, 44 (1), 341–378.

https://doi.org/10.1002/zaac.19050440122.

(110) Tamaru, K. The History of the Development of Ammonia Synthesis. In Catalytic

Ammonia Synthesis; Jennings, J. R., Ed.; Springer US, 1991; pp 1–18.

https://doi.org/10.1007/978-1-4757-9592-9.

(111) Erisman, J. W.; Sutton, M. A.; Galloway, J.; Klimont, Z.; Winiwarter, W. How a

Century of Ammonia Synthesis Changed the World. Nat. Geosci. 2008, 1 (10),

636–639. https://doi.org/10.1038/ngeo325.

(112) IEA. Technology Roadmap - Energy and GHG Reductions in the Chemical

Industry via Catalytic Processes; Paris, 2013.

129

(113) Schlögl, R. Catalytic Synthesis of Ammonia—A “Never-Ending Story”? Angew.

Chemie Int. Ed. 2003, 42 (18), 2004–2008.

https://doi.org/10.1002/anie.200301553.

(114) Rafiqul, I.; Weber, C.; Lehmann, B.; Voss, A. Energy Efficiency Improvements in

Ammonia Production - Perspectives and Uncertainties. Energy 2005, 30 (13),

2487–2504. https://doi.org/10.1016/j.energy.2004.12.004.

(115) Chen, J. G.; Crooks, R. M.; Seefeldt, L. C.; Bren, K. L.; Bullock, R. M.;

Darensbourg, M. Y.; Holland, P. L.; Hoffman, B.; Janik, M. J.; Jones, A. K.; et al.

Beyond Fossil Fuel–Driven Nitrogen Transformations. Science (80-. ). 2018, 360

(6391), eaar6611. https://doi.org/10.1126/science.aar6611.

(116) Kyriakou, V.; Garagounis, I.; Vasileiou, E.; Vourros, A.; Stoukides, M. Progress in

the Electrochemical Synthesis of Ammonia. Catal. Today 2017, 286 (June 2016),

2–13. https://doi.org/10.1016/j.cattod.2016.06.014.

(117) Medford, A. J.; Hatzell, M. C. Photon-Driven Nitrogen Fixation: Current Progress,

Thermodynamic Considerations, and Future Outlook. ACS Catal. 2017, 7 (4),

2624–2643. https://doi.org/10.1021/acscatal.7b00439.

(118) Foster, S. L.; Bakovic, S. I. P.; Duda, R. D.; Maheshwari, S.; Milton, R. D.;

Minteer, S. D.; Janik, M. J.; Renner, J. N.; Greenlee, L. F. Catalysts for Nitrogen

Reduction to Ammonia. Nat. Catal. 2018, 1 (7), 490–500.

https://doi.org/10.1038/s41929-018-0092-7.

(119) Kubota, Y.; Koga, K.; Ohno, M.; Hara, T. Synthesis of Ammonia through Direct

Chemical Reactions between an Atmospheric Nitrogen Plasma Jet and a Liquid.

Plasma Fusion Res. 2010, 5, 042–042. https://doi.org/10.1585/pfr.5.042.

130

(120) Peng, P.; Chen, P.; Addy, M.; Cheng, Y.; Zhang, Y.; Anderson, E.; Zhou, N.;

Schiappacasse, C.; Hatzenbeller, R.; Fan, L.; et al. In Situ Plasma-Assisted

Atmospheric Nitrogen Fixation Using Water and Spray-Type Jet Plasma. Chem.

Commun. 2018, 54 (23), 2886–2889. https://doi.org/10.1039/C8CC00697K.

(121) Sakakura, T.; Uemura, S.; Hino, M.; Kiyomatsu, S.; Takatsuji, Y.; Yamasaki, R.;

Morimoto, M.; Haruyama, T. Excitation of H 2 O at the Plasma/Water Interface by

UV Irradiation for the Elevation of Ammonia Production. Green Chem. 2018, 20

(3), 627–633. https://doi.org/10.1039/C7GC03007J.

(122) Haruyama, T.; Namise, T.; Shimoshimizu, N.; Uemura, S.; Takatsuji, Y.; Hino,

M.; Yamasaki, R.; Kamachi, T.; Kohno, M. Non-Catalyzed One-Step Synthesis of

Ammonia from Atmospheric Air and Water. Green Chem. 2016, 18 (16), 4536–

4541. https://doi.org/10.1039/C6GC01560C.

(123) Lee, J. K.; Banerjee, S.; Nam, H. G.; Zare, R. N. Acceleration of Reaction in

Charged Microdroplets. Q. Rev. Biophys. 2015, 48 (4), 437–444.

https://doi.org/10.1017/S0033583515000086.

(124) Bain, R. M.; Pulliam, C. J.; Cooks, R. G. Accelerated Hantzsch Electrospray

Synthesis with Temporal Control of Reaction Intermediates. Chem. Sci. 2015, 6

(1), 397–401. https://doi.org/10.1039/c4sc02436b.

(125) Castle, G. S. P. Contact Charging Between Particles; Some Current

Understanding. In ESA Annual Meeting on Electrostatics; 2008; p Paper M1.

(126) Brown, P. P.; Lawler, D. F. Sphere Drag and Settling Velocity Revisited. J.

Environ. Eng. 2003, 129, 222–231. https://doi.org/10.1061/(ASCE)0733-

9372(2003)129:3(222).

131

(127) Miura, T.; Arakawa, I. Gas Discharge Caused by Triboelectricity around a Contact

during Friction between Insulators. IEEE Trans. Dielectr. Electr. Insul. 2007, 14,

560–565. https://doi.org/10.1109/TDEI.2007.369513.

(128) Shen, X.; Wang, A. E.; Sankaran, R. M.; Lacks, D. J. First-Principles Calculation

of Contact Electrification and Validation by Experiment. J. Electrostat. 2016, 82,

11–16. https://doi.org/10.1016/j.elstat.2016.04.006.

(129) Waitukaitis, S.; Lee, V.; Jaeger, H. M. Tribocharging and Charged Interaction in

Same-Material, Microscopic Grains. In European Planetary Science Congress;

2015; Vol. 10.

(130) Lee, V.; James, N. M.; Waitukaitis, S. R.; Jaeger, H. M. Collisional Charging of

Individual Submillimeter Particles: Using Ultrasonic Levitation to Initiate and

Track Charge Transfer. Phys. Rev. Mater. 2018, 2 (3), 35602.

https://doi.org/10.1103/PhysRevMaterials.2.035602.

(131) Wang, A. E.; Greber, I.; Angus, J. C. Contact Charge Transfer between Inorganic

Dielectric Solids of Different Surface Roughness. J. Electrostat. 2019, 101 (April),

103359. https://doi.org/10.1016/j.elstat.2019.103359.

(132) Horn, R. G.; Smith, D. T. Contact Electrification and Adhesion Between

Dissimilar Materials. Science (80-. ). 1992, 256 (5055), 362–364.

https://doi.org/10.1126/science.256.5055.362.

(133) Matsuyama, T.; Ogu, M.; Yamamoto, H.; Marijnissen, J. C. M.; Scarlett, B. Impact

Charging Experiments with Single Particles of Hundred Micrometre Size. Powder

Technol. 2003, 135–136, 14–22. https://doi.org/10.1016/S0032-5910(03)00154-2.

(134) Cottrell, G. A. The Measurement of True Contact Charge Density Using Soft

132

Rubber. J. Phys. D. Appl. Phys. 1978, 11 (5), 681–687.

https://doi.org/10.1088/0022-3727/11/5/011.

(135) Nordhage, F.; Bäckström, G. Sliding and Rolling Electrification of a NaCl-Metal

System. J. Electrostat. 1977, 3 (4), 371–383. https://doi.org/10.1016/0304-

3886(77)90117-6.

(136) Donald, D. K. Contact Electrification of Insulators and Its Relevance to Electrets.

J. Electrochem. Soc. 1968, 115 (3), 270–272. https://doi.org/10.1149/1.2411128.

(137) Matsuyama, T. A Discussion on Maximum Charge Held by a Single Particle Due

to Gas Discharge Limitation. In AIP Conference Proceedings; American Institute

of Physics Inc., 2018; Vol. 1927. https://doi.org/10.1063/1.5021189.

(138) Harrison, R. G. Fair Weather Atmospheric Electricity. J. Phys. Conf. Ser. 2011,

301, 012001. https://doi.org/10.1088/1742-6596/301/1/012001.

(139) Bennett, A. J.; Harrison, R. G. Atmospheric Electricity in Different Weather

Conditions. Weather 2007, 62 (10), 277–283. https://doi.org/10.1002/wea.97.

(140) Adlerman, E. J.; Williams, E. R. Seasonal Variation of the Global Electrical

Circuit. J. Geophys. Res. Atmos. 1996, 101 (23), 29679–29688.

https://doi.org/10.1029/96jd01547.

(141) Bo, T. L.; Zheng, X. J. A Field Observational Study of Electrification within a

Dust Storm in Minqin, China. Aeolian Res. 2013, 8, 39–47.

https://doi.org/10.1016/j.aeolia.2012.11.001.

(142) Zhang, X.; Li, D.; Bo, T. The Variation of the Vertical Electric Field (

$${{\varvec{E}}}_{\mathrm{z}}$$ E z ) with Height during Dust Storms and the

Effects of Environmental Variables on $${{\varvec{E}}}_{\mathrm{z}}$$ E Z.

133

Granul. Matter 2018, 20 (3), 33. https://doi.org/10.1007/s10035-018-0801-6.

(143) Jackson, T. L.; Farrell, W. M. Electrostatic Fields in Dust Devils: An Analog to

Mars. IEEE Trans. Geosci. Remote Sens. 2006, 44 (10), 2942–2949.

https://doi.org/10.1109/TGRS.2006.875785.

(144) Harrison, R. G.; Barth, E.; Esposito, F.; Merrison, J.; Montmessin, F.; Aplin, K. L.;

Borlina, C.; Berthelier, J. J.; Déprez, G.; Farrell, W. M.; et al. Applications of

Electrified Dust and Dust Devil Electrodynamics to Martian Atmospheric

Electricity. Space Sci. Rev. 2016, 203, 299–345. https://doi.org/10.1007/s11214-

016-0241-8.

(145) Stolzenburg, M.; Marshall, T. C.; Rust, W. D.; Bruning, E.; MacGorman, D. R.;

Hamlin, T. Electric Field Values Observed near Lightning Flash Initiations.

Geophys. Res. Lett. 2007, 34 (4), 1–7. https://doi.org/10.1029/2006GL028777.

(146) Nicoll, K. A.; Harrison, R. G.; Ulanowski, Z. Observations of Saharan Dust Layer

Electrification. Environ. Res. Lett. 2011, 6, 014001. https://doi.org/10.1088/1748-

9326/6/1/014001.

(147) Olthuis, W.; Bergveld, P. On the Charge Storage and Decay Mechanism in Silicon

Dioxide Electrets. IEEE Trans. Electr. Insul. 1992, 27, 691–697.

https://doi.org/10.1109/14.155784.

(148) Yuan, N.; Li, J. SiO 2 Film Electret with High Surface Potential Stability. Appl.

Surf. Sci. 2005, 252, 455–460. https://doi.org/10.1016/j.apsusc.2005.01.025.

(149) Leonov; Fiorini; Van Hoof. Stabilization of Positive Charge in SiO/Sub 2//Si/Sub

3/N/Sub 4/ Electrets. IEEE Trans. Dielectr. Electr. Insul. 2006, 13, 1049–1056.

https://doi.org/10.1109/TDEI.2006.247831.

134

(150) Harrison, R. G.; Nicoll, K. A.; Marlton, G. J.; Ryder, C. L.; Bennett, A. J. Saharan

Dust Plume Charging Observed over the UK. Environ. Res. Lett. 2018, 13,

054018. https://doi.org/10.1088/1748-9326/aabcd9.

(151) Brazenor, T. J.; Harrison, R. G. Aerosol Modulation of the Optical and Electrical

Properties of Urban Air. Atmos. Environ. 2005, 39, 5205–5212.

https://doi.org/10.1016/j.atmosenv.2005.05.022.

(152) Ulanowski, Z.; Bailey, J.; Lucas, P. W.; Hough, J. H.; Hirst, E. Alignment of

Atmospheric Mineral Dust Due to Electric Field. Atmos. Chem. Phys. 2007, 7,

6161–6173. https://doi.org/10.5194/acp-7-6161-2007.

(153) Pipa, A. V.; Koskulics, J.; Brandenburg, R.; Hoder, T. The Simplest Equivalent

Circuit of a Pulsed Dielectric Barrier Discharge and the Determination of the Gas

Gap Charge Transfer. Rev. Sci. Instrum. 2012, 83 (11), 115112.

https://doi.org/10.1063/1.4767637.

(154) De Bie, C.; Verheyde, B.; Martens, T.; van Dijk, J.; Paulussen, S.; Bogaerts, A.

Fluid Modeling of the Conversion of Methane into Higher Hydrocarbons in an

Atmospheric Pressure Dielectric Barrier Discharge. Plasma Process. Polym. 2011,

8 (11), 1033–1058. https://doi.org/10.1002/ppap.201100027.

(155) Tsang, W.; Hampson, R. F. Chemical Kinetic Data Base for Combustion

Chemistry. Part I. Methane and Related Compounds. J. Phys. Chem. Ref. Data

1986, 15 (3), 1087–1279. https://doi.org/10.1063/1.555759.

(156) Tsang, W. Chemical Kinetic Data Base for Combustion Chemistry. Part 3:

Propane. J. Phys. Chem. Ref. Data 1988, 17 (2), 887–951.

https://doi.org/10.1063/1.555806.

135

(157) Stewart, P. H.; Larson, C. W.; Golden, D. M. Pressure and Temperature

Dependence of Reactions Proceeding via a Bound Complex. 2. Application to

2CH3 → C2H5 + H. Combust. Flame 1989, 75 (1), 25–31.

https://doi.org/10.1016/0010-2180(89)90084-9.

(158) Tsang, W. Chemical Kinetic Data Base for Combustion Chemistry Part V.

Propene. J. Phys. Chem. Ref. Data 1991, 20 (2), 221–273.

https://doi.org/10.1063/1.555880.

(159) Baulch, D. L.; Cobos, C. J.; Cox, R. A.; Esser, C.; Frank, P.; Just, T.; Kerr, J. A.;

Pilling, M. J.; Troe, J.; Walker, R. W.; et al. Evaluated Kinetic Data for

Combustion Modelling. J. Phys. Chem. Ref. Data 1992, 21 (3), 411–734.

https://doi.org/10.1063/1.555908.

(160) Harding, L. B.; Guadagnini, R.; Schatz, G. C. Theoretical Studies of the Reactions

Hydrogen Atom + Methylidyne .Fwdarw. Carbon + Hydrogen and Carbon +

Hydrogen .Fwdarw. Methylene Using an Ab Initio Global Ground-State Potential

Surface for Methylene. J. Phys. Chem. 1993, 97 (21), 5472–5481.

https://doi.org/10.1021/j100123a005.

(161) Laufer, A. H.; Fahr, A. Reactions and Kinetics of Unsaturated C 2 Hydrocarbon

Radicals. Chem. Rev. 2004, 104 (6), 2813–2832.

https://doi.org/10.1021/cr030039x.

(162) Lin, S. Y.; Guo, H. Case Study of a Prototypical Elementary Insertion Reaction: C(

1 D ) + H 2 → CH + H. J. Phys. Chem. A 2004, 108 (46), 10066–10071.

https://doi.org/10.1021/jp046039y.

(163) Harding, L. B.; Georgievskii, Y.; Klippenstein, S. J. Predictive Theory for

136

Hydrogen Atom−Hydrocarbon Radical Association Kinetics. J. Phys. Chem. A

2005, 109 (21), 4646–4656. https://doi.org/10.1021/jp0508608.

(164) Harding, L. B.; Klippenstein, S. J.; Georgievskii, Y. On the Combination

Reactions of Hydrogen Atoms with Resonance-Stabilized Hydrocarbon Radicals

†. J. Phys. Chem. A 2007, 111 (19), 3789–3801.

https://doi.org/10.1021/jp0682309.

(165) Hagelaar, G. J. M.; Pitchford, L. C. Solving the Boltzmann Equation to Obtain

Electron Transport Coefficients and Rate Coefficients for Fluid Models. Plasma

Sources Sci. Technol. 2005, 14 (4), 722–733. https://doi.org/10.1088/0963-

0252/14/4/011.

(166) Janev, R. K.; Wang, J. G.; Murakami, I.; Kato, T. Cross Sections and Rate

Coefficient for Electron Impact Ionization of Hydrocarbon Molecules. 2001.

(167) Janev, R. K.; Reiter, D. Collision Processes of CHy and CHy+ Hydrocarbons with

Plasma Electrons and Protons. Phys. Plasmas 2002, 9 (9), 4071.

https://doi.org/10.1063/1.1500735.

(168) Janev, R. K.; Reiter, D. Collision Processes of Hydride Species in Hydrogen

Plasmas: III. The Silane Family. Contrib. to Plasma Phys. 2003, 43 (7), 401–417.

https://doi.org/10.1002/ctpp.200310055.

(169) Janev, R. K.; Reiter, D. Collision Processes of Hydrocarbon Species in Hydrogen

Plasmas. Part 2. The Ethane and Propane Families. ChemInform 2003, 34 (37), 1–

124.

(170) Janev, R. K.; Reiter, D. Collision Processes of C2,3Hy and C2,3Hy+

Hydrocarbons with Electrons and Protons. Phys. Plasmas 2004, 11 (2), 780–829.

137

https://doi.org/10.1063/1.1630794.

(171) Itikawa, Y.; Mason, N. Cross Sections for Electron Collisions with Water

Molecules. J. Phys. Chem. Ref. Data 2005, 34 (1), 1–22.

https://doi.org/10.1063/1.1799251.

(172) ANSYS Reaction Design. Chemkin. San Diego 2016.

(173) Eliasson, B.; Kogelschatz, U. Nonequilibrium Volume Plasma Chemical

Processing. IEEE Trans. Plasma Sci. 1991, 19 (6), 1063–1077.

https://doi.org/10.1109/27.125031.

(174) Aerts, R.; Somers, W.; Bogaerts, A. Carbon Dioxide Splitting in a Dielectric

Barrier Discharge Plasma: A Combined Experimental and Computational Study.

ChemSusChem 2015, 8 (4), 702–716. https://doi.org/10.1002/cssc.201402818.

(175) Whitehead, J. C. Plasma–Catalysis: The Known Knowns, the Known Unknowns

and the Unknown Unknowns. J. Phys. D. Appl. Phys. 2016, 49 (24), 243001.

https://doi.org/10.1088/0022-3727/49/24/243001.

(176) Mei, D.; Zhu, X.; He, Y.-L.; Yan, J. D.; Tu, X. Plasma-Assisted Conversion of CO

2 in a Dielectric Barrier Discharge Reactor: Understanding the Effect of Packing

Materials. Plasma Sources Sci. Technol. 2014, 24 (1), 015011.

https://doi.org/10.1088/0963-0252/24/1/015011.

(177) Fridman, A. Plasma Chemistry; Cambridge University Press: Cambridge, 2008.

https://doi.org/10.1017/CBO9780511546075.

(178) Yasuda, H.; Wang, C. R. Plasma Polymerization Investigated by the Substrate

Temperature Dependence. J. Polym. Sci. Polym. Chem. Ed. 1985, 23 (1), 87–106.

https://doi.org/10.1002/pol.1985.170230110.

138

(179) Hegemann, D.; Nisol, B.; Watson, S.; Wertheimer, M. R. Energy Conversion

Efficiency in Low- and Atmospheric-Pressure Plasma Polymerization Processes,

Part II: HMDSO. Plasma Chem. Plasma Process. 2017, 37 (1), 257–271.

https://doi.org/10.1007/s11090-016-9754-x.

(180) Penetrante, B. M.; Hsiao, M. C.; Merritt, B. T.; Vogtlh, G. E.; Wallman, P. H.

Techniques for Nonthermal Plasma Processing of NO in N2. IEEE Trans. Plasma

Sci. 1995, 23 (4), 679–687.

(181) Baylet, A.; Marécot, P.; Duprez, D.; Jeandel, X.; Lombaert, K.; Tatibouët, J. M.

Synergetic Effect of Plasma/Catalysis Hybrid System for CH4 Removal. Appl.

Catal. B Environ. 2012, 113–114, 31–36.

https://doi.org/10.1016/j.apcatb.2011.10.026.

(182) Nozaki, T.; Abe, S.; Moriyama, S.; Kameshima, S.; Okazaki, K.; Goujard, V.;

Aʇiral, A. One Step Methane Conversion to Syngas by Dielectric Barrier

Discharge. Jpn. J. Appl. Phys. 2015, 54 (1), 1–6.

https://doi.org/10.7567/JJAP.54.01AG01.

(183) Richmonds, C.; Sankaran, R. M. Plasma-Liquid Electrochemistry: Rapid Synthesis

of Colloidal Metal Nanoparticles by Microplasma Reduction of Aqueous Cations.

Appl. Phys. Lett. 2008, 93 (13), 131501. https://doi.org/10.1063/1.2988283.

(184) Sapountzi, F. M.; Gracia, J. M.; Weststrate, C. J. (Kee. J.; Fredriksson, H. O. A.;

Niemantsverdriet, J. W. (Hans. Electrocatalysts for the Generation of Hydrogen,

Oxygen and Synthesis Gas. Prog. Energy Combust. Sci. 2017, 58, 1–35.

https://doi.org/10.1016/j.pecs.2016.09.001.

(185) Tendero, C.; Tixier, C.; Tristant, P.; Desmaison, J.; Leprince, P. Atmospheric

139

Pressure Plasmas: A Review. Spectrochim. Acta Part B At. Spectrosc. 2006, 61

(1), 2–30. https://doi.org/10.1016/j.sab.2005.10.003.

(186) Hibert, C.; Gaurand, I.; Motret, O.; Pouvesle, J. M. [OH(X)] Measurements by

Resonant Absorption Spectroscopy in a Pulsed Dielectric Barrier Discharge. J.

Appl. Phys. 1999, 85 (10), 7070–7075. https://doi.org/10.1063/1.370514.

(187) Mota-Lima, A.; do Nascimento, J. F.; Chiavone-Filho, O.; Nascimento, C. A. O.

Electrosynthesis via Plasma Electrochemistry: Generalist Dynamical Model To

Explain Hydrogen Production Induced by a Discharge over Water. J. Phys. Chem.

C 2019, 123 (36), 21896–21912. https://doi.org/10.1021/acs.jpcc.9b04777.

(188) Sakakura, T.; Takatsuji, Y.; Morimoto, M.; Haruyama, T. Nitrogen Fixation

through the Plasma/Liquid Interfacial Reaction with Controlled Conditions of

Each Phase as the Reaction Locus. Electrochemistry 2020, 88 (3), 190–194.

https://doi.org/10.5796/electrochemistry.19-00080.

(189) Greenlee, L. F.; Renner, J. N.; Foster, S. L. The Use of Controls for Consistent and

Accurate Measurements of Electrocatalytic Ammonia Synthesis from Dinitrogen.

ACS Catal. 2018, 8 (9), 7820–7827. https://doi.org/10.1021/acscatal.8b02120.

(190) Suryanto, B. H. R.; Du, H.-L.; Wang, D.; Chen, J.; Simonov, A. N.; MacFarlane,

D. R. Challenges and Prospects in the Catalysis of Electroreduction of Nitrogen to

Ammonia. Nat. Catal. 2019, 2 (4), 290–296. https://doi.org/10.1038/s41929-019-

0252-4.

(191) M.W. Chase; Davies, C. A.; Downey, J. R.; Frurip, D. J.; McDonald, R. A.;

Syverud, A. N. NIST-JANAF Thermochemical Tables; Gaithersburg, MD, 1985.

https://doi.org/10.18434/T42S31.

140

(192) Mehta, P.; Barboun, P.; Herrera, F. A.; Kim, J.; Rumbach, P.; Go, D. B.; Hicks, J.

C.; Schneider, W. F. Overcoming Ammonia Synthesis Scaling Relations with

Plasma-Enabled Catalysis. Nat. Catal. 2018, 1 (4), 269–275.

https://doi.org/10.1038/s41929-018-0045-1.

(193) Barboun, P.; Mehta, P.; Herrera, F. A.; Go, D. B.; Schneider, W. F.; Hicks, J. C.

Distinguishing Plasma Contributions to Catalyst Performance in Plasma-Assisted

Ammonia Synthesis. ACS Sustain. Chem. Eng. 2019, 7 (9), 8621–8630.

https://doi.org/10.1021/acssuschemeng.9b00406.

(194) Iwamoto, M.; Akiyama, M.; Aihara, K.; Deguchi, T. Ammonia Synthesis on

Wool-like Au, Pt, Pd, Ag, or Cu Electrode Catalysts in Nonthermal Atmospheric-

Pressure Plasma of N2 and H2. ACS Catal. 2017, 7 (10), 6924–6929.

https://doi.org/10.1021/acscatal.7b01624.

(195) Tschiersch, R.; Bogaczyk, M.; Wagner, H. E. Systematic Investigation of the

Barrier Discharge Operation in Helium, Nitrogen, and Mixtures: Discharge

Development, Formation and Decay of Surface Charges. J. Phys. D. Appl. Phys.

2014, 47 (36). https://doi.org/10.1088/0022-3727/47/36/365204.

(196) Laux, C. O.; Spence, T. G.; Kruger, C. H.; Zare, R. N. Optical Diagnostics of

Atmospheric Pressure Air Plasmas. Plasma Sources Sci. Technol. 2003, 12 (2),

125–138. https://doi.org/10.1088/0963-0252/12/2/301.

(197) Bruggeman, P. J.; Sadeghi, N.; Schram, D. C.; Linss, V. Gas Temperature

Determination from Rotational Lines in Non-Equilibrium Plasmas: A Review.

Plasma Sources Sci. Technol. 2014, 23 (2), 023001. https://doi.org/10.1088/0963-

0252/23/2/023001.

141

(198) Voráč, J.; Synek, P.; Procházka, V.; Hoder, T. State-by-State Emission Spectra

Fitting for Non-Equilibrium Plasmas: OH Spectra of Surface Barrier Discharge at

Argon/Water Interface. J. Phys. D. Appl. Phys. 2017, 50 (29), 294002.

https://doi.org/10.1088/1361-6463/aa7570.

(199) Voráč, J.; Synek, P.; Potočňáková, L.; Hnilica, J.; Kudrle, V. Batch Processing of

Overlapping Molecular Spectra as a Tool for Spatio-Temporal Diagnostics of

Power Modulated Microwave Plasma Jet. Plasma Sources Sci. Technol. 2017, 26

(2), 025010. https://doi.org/10.1088/1361-6595/aa51f0.

(200) Nozaki, T.; Unno, Y.; Miyazaki, Y.; Okazaki, K. Optical Diagnostics for

Determining Gas Temperature of Reactive Microdischarges in a Methane-Fed

Dielectric Barrier Discharge. J. Phys. D. Appl. Phys. 2001, 34 (16), 2504–2511.

https://doi.org/10.1088/0022-3727/34/16/316.

(201) Gorbanev, Y.; O’Connell, D.; Chechik, V. Non-Thermal Plasma in Contact with

Water: The Origin of Species. Chem. - A Eur. J. 2016, 22 (10), 3496–3505.

https://doi.org/10.1002/chem.201503771.

(202) Gorbanev, Y.; Verlackt, C. C. W.; Tinck, S.; Tuenter, E.; Foubert, K.; Cos, P.;

Bogaerts, A. Combining Experimental and Modelling Approaches to Study the

Sources of Reactive Species Induced in Water by the COST RF Plasma Jet. Phys.

Chem. Chem. Phys. 2018, 20 (4), 2797–2808.

https://doi.org/10.1039/C7CP07616A.

(203) Patil, B. S.; Wang, Q.; Hessel, V.; Lang, J. Plasma N2-Fixation: 1900–2014.

Catal. Today 2015, 256, 49–66. https://doi.org/10.1016/j.cattod.2015.05.005.

(204) Peng, P.; Chen, P.; Schiappacasse, C.; Zhou, N.; Anderson, E.; Chen, D.; Liu, J.;

142

Cheng, Y.; Hatzenbeller, R.; Addy, M.; et al. A Review on the Non-Thermal

Plasma-Assisted Ammonia Synthesis Technologies. J. Clean. Prod. 2018, 177,

597–609. https://doi.org/10.1016/j.jclepro.2017.12.229.

(205) Sakakura, T.; Murakami, N.; Takatsuji, Y.; Morimoto, M.; Haruyama, T.

Contribution of Discharge Excited Atomic N, N2*, and N2+ to a Plasma/Liquid

Interfacial Reaction as Suggested by Quantitative Analysis. ChemPhysChem 2019,

20 (11), 1467–1474. https://doi.org/10.1002/cphc.201900212.

(206) Sakakura, T.; Murakami, N.; Takatsuji, Y.; Haruyama, T. Nitrogen Fixation in a

Plasma/Liquid Interfacial Reaction and Its Switching between Reduction and

Oxidation. J. Phys. Chem. C 2020, 124 (17), 9401–9408.

https://doi.org/10.1021/acs.jpcc.0c02392.

(207) Catalytic Ammonia Synthesis; Jennings, J. R., Ed.; Fundamental and Applied

Catalysis; Springer US: Boston, MA, 1991. https://doi.org/10.1007/978-1-4757-

9592-9.

(208) Warnatz, J. Combustion Chemistry; Gardiner, W. C., Ed.; Springer New York:

New York, NY, 1984. https://doi.org/10.1007/978-1-4684-0186-8.

(209) Vervloessem, E.; Aghaei, M.; F., J.; Hafezkhiabani, N.; Bogaerts, A. Plasma-

Based N2 Fixation into NOx : Insights from Modelling toward Optimum Yields

and Energy Costs in a Gliding Arc Plasmatron. ACS Sustain. Chem. Eng 2020.

https://doi.org/10.1021/acssuschemeng.0c01815.

(210) Tan, S. P.; Piri, M. Modeling the Solubility of Nitrogen Dioxide in Water Using

Perturbed-Chain Statistical Associating Fluid Theory. Ind. Eng. Chem. Res. 2013,

52 (45), 16032–16043. https://doi.org/10.1021/ie402417p.

143

(211) Abdelaziz, A. A.; Kim, H. H. Temperature-Dependent Behavior of Nitrogen

Fixation in Nanopulsed Dielectric Barrier Discharge Operated at Different

Humidity Levels and Oxygen Contents. J. Phys. D. Appl. Phys. 2020, 53 (11),

ab5c78. https://doi.org/10.1088/1361-6463/ab5c78.

(212) Abdelaziz, A. A.; Ishijima, T.; Osawa, N.; Seto, T. Quantitative Analysis of Ozone

and Nitrogen Oxides Produced by a Low Power Miniaturized Surface Dielectric

Barrier Discharge: Effect of Oxygen Content and Humidity Level. Plasma Chem.

Plasma Process. 2019, 39 (1), 165–185. https://doi.org/10.1007/s11090-018-9942-

(213) Locke, B. R.; Lukes, P.; Brisset, J.-L. Elementary Chemical and Physical

Phenomena in Electrical Discharge Plasma in Gas-Liquid Environments and in

Liquids. In Plasma Chemistry and Catalysis in Gases and Liquids; Wiley-VCH

Verlag GmbH & Co. KGaA: Weinheim, Germany, 2012; pp 185–241.

https://doi.org/10.1002/9783527649525.ch6.

(214) Lazzo, D. W. Alternative Control Techniques Document — Nitric and Adipic Acid

Manufacturing Plants; 1991.

(215) Aljamali, N. M. Review in Azo Compounds and Its Biological Activity. Biochem.

Anal. Biochem. 2015, 04 (02), 2–5. https://doi.org/10.4172/2161-1009.1000169.

144