You are on page 1of 25

Progress in Energy and Combustion Science 73 (2019) 1–25

Contents lists available at ScienceDirect

Progress in Energy and Combustion Science


journal homepage: www.elsevier.com/locate/pecs

The effect of ozone addition on combustion: Kinetics and dynamics


Wenting Sun a,∗, Xiang Gao a, Bin Wu a, Timothy Ombrello b
a
School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United States
b
U.S. Air Force Research Laboratory, Aerospace Systems Directorate, Wright-Patterson AFB, OH 45433, United States

a r t i c l e i n f o a b s t r a c t

Article history: Ozone addition is a promising method to enhance and control combustion and ignition processes. It can
Received 29 May 2018 be produced efficiently at atmospheric or elevated pressure conditions and its lifetime is sufficiently long
Accepted 19 February 2019
to allow for remote production and transport to reaction zones. Over the past decades, the effect of ozone
addition on ignition and combustion processes has been extensively studied from bench top fundamental
Keywords: burners to practical internal combustion engines. Ozone has shown the capability to accelerate ignition
ozone assisted combustion and control ignition timing, enhance flame propagation, improve flame stabilization, pre-process fuel to
ozonolysis reaction modify emission and reactivity characteristics, and reduce certain pollutant formation. Such enhancement
ozone decomposition is closely related to ozone chemistry, especially the decomposition of ozone to produce atomic oxygen
and the rapid exothermic ozonolysis reactions with unsaturated hydrocarbons. The former requires el-
evated temperature to release atomic oxygen and initiate fuel/atomic oxygen reactions typically in the
preheat zone, while the latter initiates low temperature (even room temperature) direct fuel/ozone reac-
tions. These findings provide new opportunities in the development of strategies to enhance and control
combustion/ignition processes. This article provides a comprehensive review of the basic principles of
ozone enhanced reactive processes, including fundamental ozone chemistry, ozone generation and quan-
tification, and the progress in the study of ozone addition in combustion systems.
© 2019 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 2
1.1. Background and motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 2
1.2. Combustion enhancement: challenges and opportunities . . . . . . ................................................. 2
1.2.1. Conventional passive and active control . . . . . . . . . . . . . ................................................. 2
1.2.2. Reactive species addition and other additives . . . . . . . . ................................................. 2
1.2.3. Plasma assisted combustion. . . . . . . . . . . . . . . . . . . . . . . ................................................. 3
1.3. Ozone assisted combustion and history of ozone . . . . . . . . . . . . ................................................. 3
2. Review of ozone chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 3
2.1. Ozone decomposition reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 3
2.2. Reactions with radicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 4
2.3. Reactions of ozone with hydrocarbons. . . . . . . . . . . . . . . . . . . . . ................................................. 4
2.4. Reactions with pollutants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 6
3. Generation and measurement of ozone . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 6
3.1. Ozone generation techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 6
3.2. Ozone measurement techniques . . . . . . . . . . . . . . . . . . . . . . . . . . ................................................. 7
4. The effect of ozone addition on combustion . . . . . . . . . . . . . . . . . . . . . ................................................. 7
4.1. The effect of ozone on ignition: from benchtop to real engines ................................................. 7
4.2. The effect of ozone on laminar flames. . . . . . . . . . . . . . . . . . . . . ................................................. 8
4.2.1. Ozone decomposition flame . . . . . . . . . . . . . . . . . . . . . . . ................................................. 8


Corresponding author.
E-mail address: wenting.sun@ae.gatech.edu (W. Sun).

https://doi.org/10.1016/j.pecs.2019.02.002
0360-1285/© 2019 Elsevier Ltd. All rights reserved.
2 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

4.2.2. The effect of ozone addition on flame propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9


4.2.3. The effect of ozone on flame stabilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.3. The effect of ozone on turbulent flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.4. Ozone induced cool flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.5. The effect of ozone on combustion through ozonolysis reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5. Fuel processing using ozone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.1. Treatment of solid fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.2. Treatment of liquid fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6. Emission control using ozone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
7. Challenge and future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
8. Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Supplementary materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1. Introduction damper used to suppress acoustic fluctuations that feed in to com-


bustion instabilities [9]. Another approach is to maintain uniform
1.1. Background and motivation combustion processes by promoting small-scale turbulence and
hampering large-scale structures [8].
Combustion plays a vital role in civilian life, such as transporta- Active control involves an actuator which can affect the sys-
tion and power generation. However, concerns of efficiency, emis- tem dynamically through a control unit. It can be categorized as
sion, and operation at extreme conditions highlight the need to open-loop or closed-loop. While open-loop control can provide
enhance and control combustion processes. For example, to in- enhanced level of operability, closed-loop control utilizes infor-
crease efficiency and reduce emissions for ground transportation mation feedback, and generally allows a finer tuning of operat-
application, various new engine technologies have been developed, ing conditions. This concept has been used in automotive engines
such as Homogeneous Charge Compression Ignition (HCCI) engines [10] for decades. Modern gas turbines operate near the lean sta-
[1] and Partially Premixed Compression Ignition (PRCI) engines [2]. bility limit to reduce emission. However, under such conditions,
However, the control of ignition timing is extremely difficult, and a small change in operating condition (a few tens of Kelvin (K)
it becomes necessary to develop an alternative method to con- of temperature change) might be responsible for a dramatic in-
trol the ignition process. For air transportation, in order to meet crease in pollutant formation [11]. Active control techniques have
stringent emission standards, new lean burn aircraft engines have been developed to mitigate these types of variations by precisely
been developed, such as lean direct injection (LDI) [3] and trapped controlling/adjusting the equivalence ratio in gas turbines. How-
vortex lean combustors [4]. Therefore, flame stabilization strate- ever, the estimation of the air inflow to the combustor rate is diffi-
gies are necessary to achieve stable combustion at such ultra-lean cult [12] necessitating more robust active control methods. Besides
conditions. Another example is high-speed air-breathing propul- emission control, combustion instability is another issue where ac-
sion systems, such as supersonic combustion ramjet (scramjet) en- tive control techniques have been applied. For example, one tech-
gines. The short flow residence times in the engine requires en- nique summarized in Ref. [13] uses a microphone to monitor the
hancement and control of ignition and flame stabilization. If the pressure fluctuations, which informs a controller to activate an
reaction rates of limiting reaction steps can be enhanced, igni- acoustic driver to mitigate instabilities in the combustor. Regard-
tion and combustion processes could be dramatically accelerated. ing the control and enhancement of combustion process, various
For ground power generation and industrial burners, driven by the approaches have been proposed. Acoustic forcing was proposed
needs of NOx emission reduction and CO2 capture, new techniques to control [14] and stabilize [15] lifted non-premixed jet flames.
such as flameless combustion [5,6] and oxy-fuel [7] combustion Vorticity production was used to enhance the fuel/oxidizer mix-
have been developed. In these systems, combustion instability and ing, thus promoting more complete combustion [16]. Additives and
flame flashback are challenging issues, which also highlights the plasma assisted combustion are two other promising means of en-
need for novel combustion control techniques. hancing combustion [17].
All of the needs with regard to combustion control and en-
hancement elicit both challenges and opportunities. To provide
some context, recent developments on combustion control will be 1.2.2. Reactive species addition and other additives
briefly reviewed in the next section. Due to its low autoignition temperature (∼291 K) and sponta-
neous combustion when coming into contact with air (pyrophoric),
1.2. Combustion enhancement: challenges and opportunities silane has been used to promote ignition in scramjet engines. Mor-
ris et al. [18] experimentally studied silane as a fuel additive to
Facing the challenges discussed above, techniques for combus- promote ignition in hydrogen-fueled scramjets. The results indi-
tion enhancement and control, either passive or active, have be- cated that blending as low as 2.5% silane in hydrogen was effective
come necessary. A survey of different methods is briefly reviewed for scramjets to operate at conditions where hydrogen alone could
in this section. not combust. However, storage and handling of such pyrophoric
fluids are extremely difficult and raise significant safety concerns.
1.2.1. Conventional passive and active control Correspondingly, Gerstein and Choudhury [19] proposed a “pro-
Passive control methods are typically based on geometrical grammed ignition” strategy where an ignition promoter (silane)
modification of combustors. Such concepts have been extensively and volatile diluent (liquid methane) were mixed first, and the liq-
used to mitigate the combustion instabilities encountered in gas uid mixture was sprayed into the combustion chamber upstream
turbine combustors [8]. One typical approach is to decouple the of the main fuel (e.g. hydrogen). The mixture was initially non-
periodic heat release and acoustic pressure oscillations so that they flammable and then underwent rapid methane vaporization, after
are not in phase with each other. An example is the Helmholtz which the mixture became flammable.
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 3

Ignition promoters have also been applied in research using re- for plasma assisted flames when compared to the case without
newable fuels. Ethanol is a possible substitute for diesel fuels, how- plasma [53,54].
ever, it has a lower cetane number and high heat of vaporization.
The addition of an ignition promoter has been proposed as a po- 1.3. Ozone assisted combustion and history of ozone
tential solution. Cipolat and Bhana [20–22] performed a series of
studies using dimethyl ether (DME) as the ignition promoter and The difficulty of generating stable, uniform, non-equilibrium
concluded that an ethanol/DME mixture can allow for engines to plasmas at high pressures can be mitigated by the injection of
achieve similar power as diesel fuel, but with considerably reduced long-lived, plasma-produced species. One promising species is
emissions. Besides diesel engines, DME has also been applied in ozone. Ozone, or trioxygen (O3 ), is a pale blue gas with a distinc-
spark-ignited engines. Liang et al. [23] blended DME (1–2% volume tively pungent smell at room conditions. Properties of ozone have
fractions) with ethanol and supplied the fuel mixture to a gaso- been long studied [55] since it plays a vital role in our life in many
line engine. It was found that DME addition increased the thermal different ways. Ozone in the stratosphere (ozone layer) protects
efficiency and the unburned hydrocarbon emissions were reduced. earth from biologically-damaging ultraviolet radiation. On earth, it
However, one potential issue of the duel fuel strategy is the com- is used in many industrial processes and as a disinfectant. Because
plicated fueling system. of its vigorous oxidizing properties, ozone can be toxic and dam-
In recent years, the addition of reactive metallic nanoparticles age lungs and vital organs once inhaled [56], and is considered a
(NPs) to hydrocarbon fuels has drawn great interest for propulsion pollutant at ground level because of being involved in the forma-
systems. Liquid fuels with a stable suspension of NPs are some- tion of photochemical smog. Furthermore, the reactions between
times referred to as nanofluid fuels [24]. Such NPs significantly in- O3 and alkenes may form photochemical air pollution [57,58]. Due
crease reactive interfacial area, as well as thermal and mass trans- to these important issues, the kinetics of gas-phase reactions of O3
port properties. The energy stored in these metallic fuels can be has been extensively studied.
released rapidly by chemical reactions [24]. It has been shown that Ozone is one of the strongest oxidizers. It can be efficiently and
the addition of such NPs can increase the volumetric energy den- economically produced at high pressure conditions, which corre-
sity [25] and decrease ignition delay [26,27], increase the burning sponds to the operating condition of most practical combustion
rate [28], and catalyze fuel decomposition [29]. One drawback of devices. One advantage of O3 is that it is produced from oxygen
this method is the formation of metal particles after combustion. molecules which is the oxidizer of almost all combustion systems.
For more detailed review of related studies, the readers are re- Therefore, no additional storage of fluids or gases is needed. Ozone
ferred to Ref. [30]. has a very long lifetime of approximately 1500 min in quiescent air
at room temperature (298 K) and zero humidity [59]. Compared to
1.2.3. Plasma assisted combustion the time scales in most combustors, the lifetime of O3 could be
Non-equilibrium plasma assisted combustion (PAC) has shown long enough to be transported from remote production to the de-
promising potential for enhancing and controlling combustion pro- sired combustion region to modify fuel oxidization pathways. Pre-
cesses [17,31–33]. Plasma discharges may enhance the combustion vious studies have shown that O3 has the ability to accelerate ig-
process via thermal, kinetic, and/or transport effects. Investigations nition [60–62], enhance flame propagation [63–65] and flame sta-
have shown its ability to enhance flame stabilization [34,35] and bilization [66], process fuel to modify the emissions [67] and re-
ignition [31,32], as well as reduce soot [36] and NOx emissions activity characteristics [68], and reduce certain pollutant formation
[37,38]. Potential applications of plasma assisted combustion in- [62,69].
clude supersonic propulsion and scramjet engines [39–42], gaso-
line and diesel engines [43,44], pulse detonation engines [45], and 2. Review of ozone chemistry
gas turbine engines [46,47]. One basic principle of plasma assisted
combustion is that reactive species produced from plasma, such as An enormous number of reactions involving O3 have been stud-
atomic oxygen [48], excited O2 [49] and ozone (O3 ) [50], can en- ied. For example, searching O3 related reactions on the National
hance combustion processes. Besides the kinetic enhancement ef- Institute of Standards and Technology (NIST) Kinetic Database1 re-
fect, the Joule heating from plasma causes thermal enhancement turns 1974 records for 719 different reactions, predominantly con-
of combustion processes, and the change of molecular structure of tributed by the atmospheric chemistry community. However, O3
fuels owing to fast plasma kinetics and its induced subsequent re- related reactions (except O3 decomposition reactions) did not draw
actions also change the transport properties of the working fluid. much attention in the combustion community, as O3 rarely exists
All of these enhancement mechanisms could couple with aerody- in a conventional combustion system.
namic effects to influence the entire combustion system [17,31,32].
For more detailed reviews of plasma assisted combustion, the read- 2.1. Ozone decomposition reactions
ers are referred to Ref. [17,31,32].
One outstanding challenge of the plasma assisted combus- In the study of O3 assisted combustion [70–74], one of the most
tion technique is that the operating conditions of non-equilibrium important roles that O3 plays is through decomposition (reaction
plasma and combustion systems are not well aligned. Most com- (R1)) to release atomic oxygen in the preheating zone of a flame,
bustion systems operate at elevated pressure conditions, while therefore inducing early fuel oxidation. The reaction is
non-equilibrium plasmas favor reduced pressure conditions, typ- O3 +M =O2 +O + M (R1)
ically below atmospheric pressure [51]. High pressure conditions
where M stands for a third-body reactant representing all colli-
causes plasma thermal instability, non-uniform structure [51], and
sion species. The rate constant of this reaction has been exten-
collisional thermalization of the discharge [51]. Therefore, it is
sively studied. A summary of reaction rate constants of reaction
very challenging to generate a uniform large volume plasma at
(R1) from different studies is presented in Table S1 in the Sup-
high pressure conditions. At the same time, radicals have much
plementary Material, where A is the pre-exponential factor, n is a
shorter lifetime at high pressures [52], and thus they are generally
dimensionless parameter, T is the temperature measured in K, R is
quenched before being transported to the reaction zone (if radi-
the universal gas constant, and Ea is the activation energy.
cal generation is not in situ). Another issue which has not been
fully addressed in plasma assisted combustion systems is the po-
tential for increased NO concentration which has been observed 1
http://kinetics.nist.gov/kinetics/index.jsp, webpage visited on August 26, 2017.
4 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

reaction rate constants at 800 K calculated using different expres-


sions shows a standard deviation of 28% relative to their average
value.

2.2. Reactions with radicals

Besides the decomposition of O3 , there are several O3 reactions


with radicals that are expected to be important in combustion sys-
tems involving O3 .
O3 +H =O2 +OH (R2)

O3 +H = O + HO2 (R3)

O3 +OH =O2 +HO2 (R4)

O3 +HO2 = OH+O2 +O2 (R5)


Fig. 1. Pressure dependence of reaction rate coefficient of O3 decomposition reac-
tion.
O3 +O =O2 +O2 (R6)
For reaction (R2), the earliest research may date back to 1962,
It is noteworthy that the O3 decomposition reaction is a bi- which only reported a kinetic rate. More recent works included ac-
molecular reaction at its low pressure limit (expressed as (R1)) tivation energy [77]. Recent works on reaction (R2), and its tem-
and a unimolecular reaction only at its high pressure limit. So the perature dependent reaction rate constants are summarized in Ta-
reaction rate constant of O3 decomposition reaction depends on ble S2 and plotted in Fig. S2 in the Supplementary Material.
both temperature and pressure (away from its high pressure limit). For reaction (R3), fewer studies have been conducted. The only
Starting from the high pressure limit, the rate coefficient decreases measurement reported was in 1980 by Howard and Finlayson-Pitts
with decreasing third body concentration [M] and the correspond- [78], with a value of 4.52 × 1011 at 298 K. The temperature de-
ing representation of the reaction rate constant as a function of pendence was not investigated. Reactions (R4)–(R6) are relatively
[M] is termed as the “falloff” curve of the reaction. The pressure well studied, and their reaction rate constants from past studies
dependence of the rate constant of O3 decomposition could be de- are summarized in Table S3, S4, and S5, respectively, in the Sup-
scribed by the Lindemann-Hinshelwood reaction scheme [75] and plementary Material. The temperature range at which reaction (R4)
calculated using the classical Troe formalism [75]. A Troe formal- was investigated is relatively narrow, except the work conducted
ism obtained from theoretical model (RRKM model) was presented by Ju et al. [79]. The results are visualized in Fig. S3 for (R4), Fig.
by Peukert et al. [76], which is valid in a temperature range from S4 for (R5) and Fig. S5 for (R6) in the Supplementary Material. The
70 0 K to 140 0 K and pressure range from 5 mbar to 3 bar. The deviations among existing results for (R4)–(R6) are relatively large.
pressure dependence of O3 decomposition reaction rate constant
is calculated at 900 K using this Troe formalism and is presented 2.3. Reactions of ozone with hydrocarbons
in Fig. 1. Experimental measurements of O3 decomposition reac-
tion rate constants (T = 887 K to 907 K; P = 0.097 bar to 0.49 bar) In the reactions between O3 and hydrocarbons, there is a dra-
from Peukert et al. [76] are also shown in Fig. 1 for comparison. It matic difference in the behavior of saturated hydrocarbons (alka-
can be seen that the experiments agree with the RRKM model rea- nes) and unsaturated hydrocarbons (alkenes) [80]. While the di-
sonably well. The results in Fig. 1 at pressures greater than 3 bar rect reactions between O3 and alkanes is immeasurably slow and
are extrapolation and for demonstration only as the Troe formal- therefore of little importance in many systems, reactions of alkenes
ism is invalid beyond 3 bar according to Peukert et al. [76]. It can with O3 can be fast. Alkenes undergo a cycloaddition of ozone,
be seen that the high pressure limit of O3 decomposition occurs at and this exothermic class of reactions is generally referred to as
very high pressure (above 10 0 0 bar). Expression of the reaction rate “ozonolysis reactions” (R7) where Q means heat release. While the
at the low pressure limit works well below approximately 5 bar. rate constants for these reactions have been extensively studied ei-
However, if the working pressure is greater than 5 bar, Troe for- ther experimentally or computationally [81,82], significant uncer-
malism should be used to calculate the reaction rate constant for tainties/unknowns still exist in detailed reaction pathways. Such
O3 decomposition. The Troe formalism for O3 decomposition reac- information could be critical for its application in combustion sys-
tion adapted from Ref. [76] is included in an O3 sub-mechanism in tems, especially for the ozonolysis reaction.
the Supplementary Material. As a majority of studies/applications
O3 +Fuel = Products + Q (R7)
of ozone occur well below the high pressure limit of O3 decom-
position reaction owing to the short lifetime of O3 at those condi- The reaction between O3 and CH4 (an alkane) is very slow (the
tions, the values presented in Table S1 focus on rates well below its reaction rate constant at 300 K is 0.99 cm3 /mol s [83]), so it is
high pressure limit. The O3 decomposition reaction rate constant at normally ignored by the combustion community. However, the di-
the high pressure limit is provided as the last entry of Table S1. rect reaction between O3 and ethylene (C2 H4 ) (an alkene) is or-
The reaction rate of (R1) may depend on the third-body col- ders of magnitude faster compared to that with CH4 [80,84] (reac-
liding partner, M. The results listed in Table S1 are visualized in tion rate constant at 300 K is 1.12 × 106 cm3 /mol s [85]). The reac-
Fig. S1 as a function of temperature in the Supplementary Mate- tion rate constants of these reactions between fuel and O3 have
rial. It shows that the overall deviation among different studies is been studied for several different hydrocarbons, as summarized
reasonably small, regardless of the different rate expressions. For in Table 1. For these components, the reaction rate constants for
the results reported using N2 (relevant to most combustion sys- alkenes are approximately a million times larger than those for
tems since it is the major component of air) as the base gas, the alkanes. For these alkenes, the molecular structures significantly
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 5

Fig. 2. Criegee mechanism of the ozonolysis reaction.

Table 1 [91] for alkenes. The first step is a 1,3-dipolar cycloaddition of O3


Reaction rate constants between O3 and hydrocarbons, k, at
to the alkene. In this step, combining with O3 , the double carbon-
300 K for different hydrocarbons.
carbon bond is broken and a 5-member cyclic structure is formed.
k (cm3 /mol s) Reference This product is called primary ozonide (POZ), which is typically
Alkanes CH4 0.99 [83] not collisionally stabilized under atmospheric conditions [92] and
C2 H 6 8.95 [86] subject to decomposition. The decomposition of POZ produces a
C3 H 8 4.19 [86] Criegee intermediate (a carbonyl oxide) and a carbonyl compound.
C4 H10 6.05 [83]
The Criegee intermediate is in an excited state. In gas phase, ap-
Alkenes C2 H4 1.12 × 106 [85]
C3 H 6 6.29 × 106 [84] proximately 50% of Criegee intermediates can be stabilized through
iso-C4 H8 6.68 × 106 [87] collisions and approximately 50% of Criegee intermediates are sub-
trans-2-C4 H8 1.43 × 108 [87] ject to decomposition and reaction with other hydrocarbon species,
cis-2-C4 H8 7.77 × 107 [87]
for example in the ozonolysis process of C2 H4 [93]. The process
1-C4 H8 5.82 × 106 [87]
mentioned above is summarized in Fig. 2.
In the case of C2 H4 , the carbonyl compound is CH2 O, and the
excited Criegee intermediate is CH2 OO∗ [94], which is the sim-
affect the rate constants. For example, at 298 K, the reaction rates plest Criegee intermediate and can be collisionally stabilized. Di-
for trans-2-butene (trans-2-C4 H8 ) and cis-2-butene (cis-2-C4 H8 ) rect detection of Criegee intermediate is very challenging, espe-
are much higher than that for iso-C4 H8 and 1-butene (1-C4 H8 ). It cially at elevated pressure conditions. Recently, there has been
is also interesting to see the noticeable difference between the re- significant progress in the measurement of Criegee intermediates
action rate constants of trans-2-butene and cis-2-butene ozonolysis at reduced pressure conditions. In 2008, Taatjes et al. [95] re-
reactions. ported direct observation of Criegee intermediates in the oxidation
The parameters showing the temperature dependence of the re- process of dimethyl sulfoxide (CH2 S(O)CH3 ) using photoionization
actions for alkanes are listed in Table S6 and the corresponding time-of-flight mass spectrometry. The oxidation process was pho-
reaction rate constants are plotted in Fig. S6 in the Supplemen- tolytically initiated by a pulsed laser to produce Chlorine atoms.
tary Material. The measurements were typically conducted within In a later work, Welz et al. [96] used photolysis of CH2 I2 in O2
a very narrow temperature range. For CH4 , the results reported by to generate CH2 OO and measured CH2 OO directly using photoion-
Stedman and Niki [88] is approximately 10 0 0 times faster than the ization time-of-flight mass spectrometry. Su et al. [97] presented
results by Dillemuth et al. [89] or Schubert et al. [83]. In a recent the infrared absorption spectrum of CH2 OO produced from photol-
experiment conducted [64], the smaller reaction rate constant was ysis of CH2 I2 in O2 at 12.5 kPa. Recently, a method has been ex-
indirectly supported. perimentally demonstrated by Berndt et al. [98] to detect Criegee
The reactions rate parameters for alkenes are listed in Tables intermediates directly from ozonolysis reactions at near atmo-
S7–S9 and the corresponding reaction rate constants are plotted in spheric conditions using chemical ionization mass spectrometry.
Figs. S7–S9, respectively, in the Supplementary Material. For C2 H4 , Elsamra et al. [99] experimentally measured reaction rate constants
the averaged value of reaction rate constant at 298 K using the re- of CH2 OO+CH3 COCH3 reaction and CH2 OO+CH3 CHO reaction from
ported results/expressions is 1.1 × 106 cm3 /mol s, and the stan- 298–500 K and 0.5–6.7 kPa in a photolysis reactor. In the same set
dard deviation among different studies at this temperature is 28%. up, Buras et al. [100] measured self-reaction of CH2 OO in a follow
If only results from studies after 1980 are considered, the deviation up work. The readers are referred to a recent review [101] on the
decreases to 11% and the average value is 1.0 × 106 cm3 /mol s. For kinetics of Criegee intermediates.
C3 H6 , the averaged value is 7.1 × 106 cm3 /mol s at 298 K, and the The subsequent reactions after the decomposition of POZ are
standard deviation is 15.6%. The result from Adeniji et al. [90] de- still not very clear because of their complexity even though various
viated from the average value the most. If this result is not consid- pathways have been proposed. Criegee suggests that the Criegee
ered, the deviation decreases to 9.5% of the average value, which is intermediate and the carbonyl compound could recombine and
6.7 × 106 cm3 /mol s at 298 K. As the number of carbon atoms fur- yield a secondary ozonide (SOZ), with a different molecular struc-
ther increases, the ozonolysis reactions become very complex and ture than POZ. More recent investigations suggest that the excited
the reaction rate constants from different studies also deviate more state Criegee intermediate formed in the POZ cleavage may de-
significantly. For butene, the molecular structure becomes an im- compose unimolecularly [92,94,102] in addition to collisional sta-
portant factor. The ozonolysis reaction rate constants of iso-C4 H8 bilization. Several pathways for this unimolecular decomposition
and 1-C4 H8 are similar while those for trans-2-C4 H8 and cis-2- reaction of the Criegee intermediate have been proposed as sum-
C4 H8 ozonolysis reactions are approximately 10 times faster (refer marized in Ref. [94]: (1) the ester channel, which forms dioxirane
to Table S9 in the Supplementary Material). and subsequently decomposes into simpler species, (2) the forma-
As the direct reactions between O3 and alkanes are generally tion of an OH radical via H migration, and (3) the split-off of O(3 P)
very slow at room temperature and can be neglected, only the atoms. The relative branching fractions of decomposition of Criegee
ozonolysis reactions of alkenes will be discussed here. A well- intermediates are predicted to be pressure dependent [92]. As the
known ozonolysis mechanism was proposed in 1975 by Criegee products of these channels are highly reactive, secondary reactions
6 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

can be initiated and large uncertainties can be introduced, which 3. Generation and measurement of ozone
complicates the investigation of ozonolysis. Horie and Moortgat
[103] conducted experiments at 101.1 kPa (∼1 bar) and 297 K using 3.1. Ozone generation techniques
a continuously-stirred tank-reactor and determined the branching
ratios of the decomposition channels for some types of Criegee in- One technique for O3 production is through the interaction of
termediates. For Criegee intermediates produced from ozonolysis O2 with light in the ultraviolet (UV) wavelength range. The mech-
of C2 H4 , which is CH2 OO, the branching ratio of CH2 OO decom- anism is responsible for the formation of the O3 layer in the at-
position is: (CO+H2 O):(CO2 +H2 ):(CO2 +2H) = 58%:24%:18%. Later mosphere. Conventionally, a UV lamp (∼100 nm to 260 nm) can be
work showed and considered reaction pathways of the yield of OH used in such O3 generators.
radicals [104,105]. In a work conducted by Neeb et al. [93], the O3 can also be produced by electrical discharge, for example
ozonolysis reaction pathways of C2 H4 is summarized as from lightning during a thunderstorm. Most commercial O3 gen-
erators are based on electrical discharges. An electrical discharge
O3 +C2 H4 = CH2 O + CH2 OO∗ (R8)
(typically corona discharge or dielectric barrier discharge to main-
tain low gas temperatures and minimize O3 decomposition) is used
CH2 OO∗ =H2 +CO2 (R8a) to break the molecular bond of O2 , therefore, producing atomic
oxygen, which then recombines with O2 to form O3 via the reac-
tion O + O2 + M = O3 + M. A relatively high pressure is preferred
CH2 OO∗ = 2H + CO2 (R8b)
for O3 formation as it is a three-body reaction. Modern high-
power O3 generators take advantage of semiconductors to generate
CH2 OO∗ =H2 O + CO (R8c) square-wave currents or specially formed wave trains with a rep-
etition rate of 0.5–5 kHz. Ozone concentrations up to 5 wt% from
air and up to 18 wt% from oxygen can be reached [112]. Large scale
CH2 OO∗ = OH + HCO (R8d) O3 generators with production up to 100 kg/h have been built for
industrial applications. With modern technology, ozone can be pro-
CH2 OO∗ = HCOOH (R8e) duced for less than $2/kg [112].
Simek and Clupek [113] experimentally studied the efficiency of
O3 production by a pulsed discharge in air. The energy needed to
CH2 OO ∗ +M = CH2 OO + M (R8f) produce a given quantity of O3 (O3 yield, g/kWh) was measured
at different air flow rates and O3 concentrations. Ozone yield var-
The branching ratio of (R8a + R8b):(R8c + R8d):(R8e):(R8f)
ied in a range of 15–55 g/kWh. Samaranayake et al. [114] reported
=0.23:0.23:0.04:0.5 [93], however, the reaction rate constants were
O3 yield of 10–110 g/kWh, with a non-monotonic trend depend-
not reported. A recent work by Rousso et al. [106] revealed the
ing on O3 concentration. Based on the authors’ estimation from
complexity of products from the simplest ozonolysis reaction be-
the specification data available [115], the O3 yield for some com-
tween C2 H4 and O3 . They identified the products from C2 H4 /O3
mercial O3 generators based on pulsed discharges is approximately
reaction in a jet-stirred reactor including C2 H4 O3 , H2 O2 , CH3 OOH
67–125 g/kWh, on a similar order of the values reported in Refs.
and C2 H5 OOH using photoionization time-of-flight mass spec-
[113,114]. These measurements of O3 yield (mass per energy input)
trometry. They also showed experimental evidence for additional
suggest that approximately 3 × 107 –3 × 108 J of energy is needed
oxygenated species such as methanol, ketene, acetaldehyde, and
for the production of 1 kg of O3 . For comparison, the difference be-
hydroxy-acetaldehyde suggesting multiple active oxidation routes.
tween the enthalpy of 1 kg of O2 and 1 kg of O3 is 3.0 × 106 J at
Much more work is required in the future to fully understand
300 K and 101 kPa, which means only 1–10% of the total consumed
ozonolysis reactions.
energy is actually used to produce O3 . To improve the efficiency
of O3 production, experimental studies have been conducted con-
2.4. Reactions with pollutants
cerning the electrode arrangements [116–118], dielectric materials
[119], structure of discharge tube, feed-gas type (dried air or O2 )
Ozone has been investigated as a reagent to remove NOx [107–
[119], feed-gas flow rate [120], operating pressure [120,121], power
109]. In most practical flue gas, NO is the major NOx species. A
supply frequency [121], and humidity [121].
general method is to convert NO, which has low solubility, to a
The recombination reaction O + O2 + M = O3 + M, is the dom-
species that is highly soluble in water, such as NO2 , NO3 , or N2 O5 .
inant channel to consume atomic oxygen and produce O3 . If air is
The water soluble species can then be captured later downstream.
used as the gas through the discharge, NOx can also be produced
The related reactions are listed as followed:
and its collision with O3 will consume and significantly decrease
NO+O3 = NO2 +O2 (R9) the O3 yield through the reactions of NO + O3 = NO2 + O2 and
NO2 + O3 = NO3 + O2 . The production of NOx simultaneously with
O3 arises because the recombination reaction takes approximately
NO2 +O3 = NO3 +O2 (R10)
10 μs in O2 and 100 μs in air, allowing for a competition of differ-
The reaction rate constants of (R9) and (R10) have been ex- ent pathways for O consumption [122]. The production of N atoms
tensively studied over the past several decades. The reaction be- from the discharge is potentially more critical for NOx production
tween O3 and other common pollutants are relatively less studied. as the N + O2 = NO + O reaction is orders of magnitude faster
The only study on the reaction CO + O3 = CO2 + O2 was con- than the O + N2 = N + NO reaction if N2 is in the ground state.
ducted in 1972 by Arin and Warneck [110]. A very low reaction Therefore, the rate controlling step of NOx production in electrical
rate, 0.24 cm3 /mol s, is reported at 296 K. As for the reaction with discharges is the production of N atoms. If N2 is in the discharge
SO2 , SO2 + O3 = SO3 + O2 , the reaction rate is higher, of the order gas, the production of NOx cannot be avoided [123]. The trace com-
of 100 cm3 /mol s at 300 K [111]. These reactions are summarized pounds can interfere with O3 formation and when they reach a
in Table S10 in the Supplementary Material with their measured certain concentration, O3 generation completely breaks down. This
reaction rate parameters. The temperature dependence is visual- “ozoneless” mode of discharge is referred to as the state of dis-
ized in Fig. S10 in the Supplementary Material, and relatively small charge poisoning [124]. The reaction between O3 and NOx is sensi-
variability is observed. tive to temperature, and therefore such effects can be reduced by
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 7

Fig. 3. Schlieren images of ignition process, 300 K and 0.1 MPa. Case 1: stoichiometric H2 /O2 /O3 /Ar mixture; Case 2: stoichiometric H2 /O2 /Ar mixture. (Reprinted from Liu
et al. [137] with permission of J-STAGE.)

optimizing the dielectric material, which affects the gas temper- measurement is being performed in order to provide the most pre-
ature of the discharge [119]. The plasma chemistry of O3 forma- cise concentration measurement. An uncertainty of approximately
tion can be optimized in electrical discharge by providing electrons 2% and a minimum detectable threshold of approximately 15 ppm
in a suitable energy range of a few electronvolts to reduce energy were reported in Ref. [63]. Commercial O3 monitors applying the
losses to ionic species and minimize recombination reaction of O same principle to measure O3 concentration are available on the
to form O2 [112]. It was reported that the optimal reduced electric market, as employed in the works [64,65,135].
field is approximately 140 Td in O2 and 200 Td in air [123,125] for
O3 production. Comprehensive kinetic models in ozone generation 4. The effect of ozone addition on combustion
can be found in Refs. [121,126,127].
4.1. The effect of ozone on ignition: from benchtop to real engines
3.2. Ozone measurement techniques
Initial studies on O3 assisted combustion focused on ignition.
Ozone has strong absorption bands from UV to the IR region Photolysis of O3 could yield atomic oxygen and therefore was pro-
[128–130]. Therefore, a typical technique for ozone quantification posed to enhance ignition. In 1987, Lucas et al. [136] conducted
is the absorption method. In Ref. [63], a mercury lamp was used experiments in closed aluminum cells with Plexiglas side windows
to provide UV at the wavelength of 253.7 nm where O3 has a using a mixture of H2 /O2 /O3 with KrF excimer laser (248 nm) to
peak absorption cross section of 1.137 × 1017 cm2 at 300 K [131], photo dissociate O3 . The photolysis of one O3 molecule could pro-
then the O3 concentration was determined using the Beer–Lambert duce one oxygen atom and one O2 molecule with 90% of atomic
law [132]. A continuous wave deuterium lamp together with a oxygen in the electronic excited (1 D) state and O2 molecules in
monochromator can be used to perform O3 measurement at dif- the electronic (1 ) state. O(1 D) is extremely reactive, especially
ferent wavelengths (195–345 nm across the Hartley and Huggins with species like H2 with zero activation energy. The effects of
bands) [128]. Absorption cross-sections at different wavelengths equivalence ratio, pressure, and initial gas temperature on the min-
have been well characterized at different temperature conditions imum ozone concentration needed to produce ignition were in-
[128,129,131,133]. Reuter et al. [130] conducted O3 measurement vestigated. All the experiments were conducted at reduced pres-
using absorption spectroscopy in both the UV and IR wavelengths. sure conditions below 13.3 kPa and it was found that only the
In the first method, a deuterium lamp which emits a broadband initial temperature had a significant effect on the required mini-
spectrum of UV light in the range from 200 to 300 nm was used mum ozone concentration for ignition. With the increase of initial
as background source. An interference filter with a central trans- temperature, the required minimum ozone concentration for igni-
mission wavelength of 254 nm with full width at half maximum tion decreased. Numerical modeling also showed that ignition was
(FWHM) of 10 nm was used to produce light around 254 nm. An not due solely to thermal effects, but was strongly dependent on
ICCD camera connected to a spectrometer served as the detec- the number and type of radicals present initially after photolysis
tor. Ozone concentration around a plasma jet was then calcu- [136]. Similar photolysis experiment was conducted by Liu et al.
lated based on the absorption spectrum. In the second method, [137] in 1998 under atmospheric pressure using a pipe burner
a quantum cascade laser in the mid-IR region (center frequency with 6 mm inner diameter. The minimum incident laser energy in-
1027 cm−1 = 9737.1 nm) was used as the light source. Ozone mea- tensity (MILEI) from a KrF excimer laser (248 nm) was measured
surement was demonstrated in a 60 cm multipass white cell. Since for the ignition of H2 /O2 /O3 /Ar mixtures. An ozone generation de-
the quantum cascade laser has fine spectral resolution, a more ac- vice was placed downstream of the O2 flow controller and it was
curate measurement of O3 concentration was obtained using mid- found that the MILEI decreased with the increase of O3 concen-
IR absorption spectroscopy. Teranishi et al. [134] investigated mea- tration. The authors further conducted photolysis ignition experi-
surement of O3 concentration using absorption spectroscopy in the ments using ArF laser (193 nm) to dissociate O2 for H2 /O2 /Ar mix-
visible light region. Ozone has photon absorption band in the vis- tures and compared schlieren photographs of ignition processes
ible wavelength region around 450–850 nm, which is referred to with those from H2 /O2 /O3 /Ar mixtures (illustrated in Fig. 3). A
as the Chappuis band, the peak cross-section of which is approxi- large difference was observed between their flame kernel growth
mately 20 0 0 times smaller than that of the Hartley band. In their processes. For H2 /O2 /O3 /Ar mixtures (case 1 in Fig. 3), the flame
work, an LED having an emission peak at 609 nm and a photodiode propagated along the laser path uniformly compared to the non-
sensitive at visible light frequencies was used as the light source uniform growth pattern for H2 /O2 /Ar mixtures (case 2 in Fig. 3).
and the photodetector, respectively. The measurement was con- Correspondingly, the growth speed of the flame kernel for case
ducted in a stainless-steel absorption optical cell. Good agreement 1 was much faster than that in case 2. The growth speed of the
between measurement based on the visible absorption method and flame kernel was estimated to be 700 m/s, which was two order
UV ozone monitor was reported. of magnitude faster than the 15 m/s in case 2. Moreover, the ig-
It is noteworthy that most absorption spectroscopy measure- nition of H2 /O2 /O3 /Ar mixtures was accompanied by the forma-
ments of ozone concentration using broadband UV light sources tion of shock waves which did not occur in case 2. The authors
has the potential to simultaneously photolytically produce ozone explained this observation by the different products from photol-
because of several strong UV absorption bands. When using ab- ysis reactions in the two cases. For H2 /O2 /O3 /Ar mixtures, photol-
sorption spectroscopy, care has to be taken to minimize wave- ysis of O3 (O3 +hv = O(1 D) + O2 (1 )) yields electronic excited
lengths other than the wavelength range where the absorption atomic oxygen (O(1 D)) which reacts with H2 on the order of 0.5 ns.
8 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

ether (DME) as the fuel. While methanol was confirmed to retard


ignition, attributed to CH3 OH + OH → CH2 OH/CH3 O + H2 O reac-
tion in which methanol is a radical scavenger and terminates DME
chain reaction system (more than one OH can be produced in the
DME + OH reaction chain) in low temperature oxidation, O3 ad-
vanced the ignition timing up to 20 CAD at an estimated addition
of only 0.015% (15 ppm). Yamada et al. [142] inferred that the early
radical supply from O3 reduced the cool flame onset temperature.
Since n-heptane is a fuel that has a cetane number similar to that
of a commercial diesel fuel, Foucher et al. [143] investigated the
effects of O3 addition on HCCI combustion of n-heptane. It was
observed that 50 ppm of O3 addition is able to advance the igni-
tion by approximately 15 CAD. Based on simplified simulation of
constant volume combustion, they concluded that the O3 addition
(which releases atomic oxygen) promoted the combustion via n-
C7 H16 + O = C7 H15 + OH, and the early production of OH acceler-
ated both cool and hot flames. Masurier et al. [144] further inves-
tigated O3 addition as a way to control the ignition timing of an
HCCI engine fueled with primary reference fuels (PRF). Similar to
previous studies, injection of low concentration O3 showed signifi-
cant impact on the flame phasing. Similar results were found when
Fig. 4. Calculated ignition process in HCCI for different O3 concentrations.
(Reprinted from Nishida and Tachibana [140] with permission of American Institute
the fuel was replaced by alcohol [61]. Masurier et al. [145] later
of Aeronautics and Astronautics, Inc.) compared the effects of O3 addition with the addition of NO and
NO2 , respectively, on the enhancement of ignition in an HCCI en-
gine fueled by iso-octane. It was found that the highest effect is
Correspondingly, photolysis of O2 (O2 + hv = 2O(3 P)) yields ground achieved by O3 , and the lowest was for NO2 .
state O atoms (O(3 P)), that reacts with H2 much slower with a
characteristic time of 7.5 ms [137]. 4.2. The effect of ozone on laminar flames
In 1988, Nomaguchi and Koda [138] investigated spark ignition
using electric pulses in a closed vessel at reduced pressure for pre- 4.2.1. Ozone decomposition flame
mixed fuel (CH4 or CH3 OH) and ozonized air. It was found that The ozone decomposition laminar flame was intensively in-
the minimum spark pulse duration for successful ignition could be vestigated and has made a great contribution to the develop-
shortened considerably by adding 0.2% O3 in O2 . Simulations con- ment of combustion theory. Owing to the overall exothermic
ducted at an initial temperature of 1100 K, and a constant pres- process of ozone decomposition and subsequent O2 formation
sure of 20.0 kPa corresponding to their experiments showed that (O3 = 1.5O2 + 34 kcal/mol), O3 is a monopropellant. Ozone de-
the acceleration of ignition was due to the atomic oxygen (from composition flames can be created using O2 /O3 mixtures or even
O3 decomposition) attacking fuel molecules. Later, Tachibana et al. pure O3 with only three species (O, O2 and O3 ) involved in the
[62] used a diesel CFR engine to examine the effect of O3 addi- reaction. Since the reaction mechanism of an ozone decomposi-
tion on compression ignition. O3 addition accelerated ignition, in- tion flame is the simplest real one having all requisite features of
creased cetane number, and lowered the compression ratio of the a truly detailed mechanism, it is suitable for a systematic varia-
ignition limit. These observations were consistent with the work tion of the transport models and kinetic models used to investi-
by Nomaguchi and Koda [138]. The findings on ignition inspired gate/validate flame theoretical models [74,146]. The calculated lam-
researchers to develop O3 addition as an ignition control tech- inar flame speeds can also be compared with experimental mea-
nique. Such an attempt was proposed in a patent [139] filed in surements.
1997. Several ways of using O3 were proposed in this patent: (i) Early work on ozone decomposition flames include those from
by adjusting the amount of O3 added, O3 can be used to control Lewis and von Elbe [147], Hirschfelder et al. [148], von Karman and
the start of combustion; (ii) O3 can be used as a cold starting aid; Penner [149] in the 1930s–50s by employing steady state approx-
and (iii) O3 can be used to reduce cold start emissions. The use imations of atomic oxygen to solve the governing equations. More
of O3 addition as a way to control ignition was later investigated detailed studies were performed in the 1970s. Wilde [150] con-
by Nishida and Tachibana [140] in 2006 on an HCCI system us- ducted laminar flame speed calculations using three different O3
ing natural gas as the fuel. Ignition timing and combustion du- decomposition kinetic models and compared with experimental
ration in HCCI engines are governed by chemical kinetics, not by measurements. He showed that the flame speeds calculated us-
spark or injection, and thus techniques to control ignition timing ing different kinetic models varied significantly from 60 cm/s to
are essential. Nishida and Tachibana [140] carried out a numeri- 152 cm/s for an O3 /O2 (mole ratio 28/72) mixture while the exper-
cal analysis, showing the feasibility of controlling ignition timing imental measurement was 52 cm/s at atmospheric pressure. Cra-
by varying the amount of O3 added, as illustrated in Fig. 4. They marossa and Dixon-Lewis [72] solved time-dependent heat con-
also experimentally showed that addition of ∼1200 ppm O3 could duction and diffusion equations for a number of ozone decom-
advance the main heat release by 18 crank angle degrees (CAD). position flames with different initial ozone mole fractions. Signif-
They also found that adding O3 was more effective than the addi- icant deviation between computations and experiments was ob-
tion of OH radicals, and was as effective as injection of atomic oxy- served. The authors suspected both deficiencies in the reaction ki-
gen radicals. Later Mohammadi et al. [141] investigated O3 addition netic model and the quenching effects in the ozone flame experi-
as a way to enhance ignition in a Premixed Charge Compression ments because of the small diameter of the experimental appara-
Ignition (PCCI) natural-gas engine. Besides natural gas, other fuels tus. In a later work by Warnatz [146], time-dependent conservation
were also investigated using ozone as a potential ignition timing equations of a laminar flat ozone decomposition flame was solved
control agent. Yamada et al. [142] explored the effect of different including complete multicomponent formulations of diffusion and
additives on the ignition timing of an HCCI engine with dimethyl heat conduction. In the kinetic model, an updated rate constant
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 9

Fig. 5. Enhancement of CH4 /air flame speeds as a function of O3 concentration, where φ is the equivalence ratio. (Reprinted from Wang et al. [65] with permission of
Elsevier.)

of the O3 decomposition reaction (O3 + M = O2 + O + M), which equivalence ratios slightly and O3 concentrations for different
was approximately 10 times smaller at temperatures above 20 0 0 K, calculations. They found that O3 addition could still enhance
was employed. The calculated flame speeds agreed well with ex- flame speeds significantly at constant adiabatic flame temperature
perimentally measured values. conditions.
As already mentioned, during the same period of time as the Besides their work on ignition in 1989, Nomaguchi and Koda
calculations of ozone decomposition flame speeds, multiple ex- [138] also measured the burning velocity of CH4 /air at room
periments were performed. Lewis and von Elbe [151] conducted temperature and atmospheric pressure. Their data showed that
the first measurement of ozone decomposition flame speed us- 50 0 0 ppm O3 addition increased the burning velocity by 5% at sto-
ing a spherical bomb. No experimental details were presented ichiometric conditions and that the enhancement of flame speed
as the focus of that work was the development of the method- was greater at fuel-lean and fuel-rich conditions than that at sto-
ology of a spherical bomb for flame speed measurement. In a ichiometric conditions. Ombrello et al. [52,63] studied O3 addition
later work by Streng and Grosse [152], flame speeds of O3 /O2 on propane (C3 H8 ) flame propagation speed (SL ) using lifted lami-
mixtures were measured with different O3 concentrations rang- nar flames, in which liftoff height is very sensitive to SL . O3 addi-
ing from 17% to 100% O3 mole fraction in the mixture. In their tion was observed to decrease the liftoff height, indicating an in-
experiment, ozone was generated by a standard ozone generator crease in SL . When accounting for the kinetic and hydrodynamic
using an electrical discharge. The O3 /O2 stream was then passed effects, an approximately 4% enhancement of SL at a stoichiomet-
through a dry-ice trap to condense O3 . Small quantities of liquid ric condition was reported with 1260 ppm O3 addition. These ob-
oxygen were removed by using standard vacuum techniques owing servations were later supported by Halter et al. [153] with simula-
to the different boiling points of O3 (−112 °C) and O2 (−183 °C). tions and a similar experiment. A more recent study by Wang et al.
Pure ozone was then used to prepare test mixtures with different [65] using a heat-flux burner showed a 3.5% increase of CH4 /air
ozone concentrations. Two methods were used to determine the laminar flame speed with 3730 ppm O3 addition at stoichiomet-
flame speeds of O3 /O2 mixture. One was the open tube method ric conditions, and 9% increase for 70 0 0 ppm O3 . Generally, the
where a glass tube of approximately 1 m length was filled with enhancement increases with the increase of O3 concentration, as
a test mixture. The mixture was burned in the tube in a hori- shown in Fig. 5. Sensitivity analysis was performed for cases with
zontal position in a dark room and the propagation of the flame and without O3 addition and showed very similar results. For a
front was recorded to obtain the flame propagation speed. The ob- CH4 /air/O3 system, O3 addition did not significantly change the
served flame speed was then converted into normal flame speed combustion kinetics in the flame, with the primary mechanism be-
by assuming a hemispheric flame front. Another technique used ing the release of atomic oxygen in the preheat zone of the flame
the burner tip method similar to the Bunsen flame method. It to promote early fuel oxidation. The addition of atomic oxygen in
was found that the measured flame speed was a linear function the preheat zone did not alter the reaction pathways of CH4 . An
of O3 mole fraction in the mixture. In the same work, the au- O3 reaction mechanism was also provided in this work [65], con-
thors also tested the stability of O3 /O2 mixtures. It was reported sidering the duplicated and missed reactions in previous works. An
that the decomposition rate of pure O3 at 296 K and 1 bar was approximately 9% enhancement of SL was reported with 8500 ppm
0.9% per day. O3 addition at stoichiometric conditions for an H2 /CO/N2 /air mix-
ture by Liang et al. [135], which was also accomplished using a
4.2.2. The effect of ozone addition on flame propagation heat-flux burner.
The effects of O3 addition on laminar flames have also drawn Gao et al. [64,154] further studied the effects of pressure and
significant attention. It is not surprising that O3 addition could fuel kinetics on CH4 /air/O3 flames and C2 H4 /air/O3 flames using a
enhance flame propagation speeds as the addition causes an in- Bunsen burner. A significant increase of SL enhancement was ob-
crease in enthalpy of the combustible mixture. Ehn et al. [50] cal- served at elevated pressures: the enhancement on the measured
culated CH4 /air flame speeds with and without O3 addition at SL for a stoichiometric CH4 /air mixture with 6334 ppm O3 addi-
constant adiabatic flame temperature conditions by adjusting the tion increased from 7.7% at atmospheric pressure to 11% at 2.5 atm.
10 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

Fig. 6. Contribution of major O3 consumption pathways in stoichiometric CH4 /air/O3 flames at atmospheric pressure. T0 = 300 K. (Reprinted from Gao et al. [64] with per-
mission of Elsevier.)

Numerical simulations and kinetics pathway analysis showed that the progress of ozonolysis reaction was not significant so only an
O3 is consumed in the flame preheat zone by primarily two reac- insignificant quantity of C2 H4 reacted with O3 inside the burner to
tions: O3 decomposition and the reaction O3 + H = O2 + OH. The release a small amount of heat release. Therefore, O3 decomposi-
first reaction provides atomic oxygen, which attacks fuel molecules tion remained the dominant effect to enhance C2 H4 flame speeds
and promotes radical production. In contrast, sensitivity analysis at reduced pressure conditions. The pressure effect was confirmed
showed that SL has a negative dependence [66] on the reaction experimentally by conducting experiments at different pressure
O3 + H = O2 + OH. Elevated pressure suppresses the dissocia- conditions by Gao et al. [154]. They performed further experiments
tion of H-containing species and diffusion of H atoms from the at cryogenic conditions to suppress the ozonolysis reactions at at-
flame zone, thus lowering the H atom concentration at elevated mospheric pressure. In the experiment, the entire burner system
pressure conditions in the preheat zone. As a result, the reaction was cooled to approximately 200 K using liquid N2 to avoid con-
O3 + H = O2 + OH is suppressed and the competing pathway, densation of reactants but assure insignificant ozonolysis reactions
O3 decomposition, became a more dominant pathway at elevated inside the burner. At such temperature conditions, less than 10%
pressure conditions. As demonstrated in Fig. 6 at 1 atm O3 decom- of O3 was consumed by ozonolysis reactions prior to arriving at
position only accounts for 37.5% of total O3 consumption; while the burner exit. The measured flame speeds with reactant temper-
at 2 atm, 57.6% of O3 is consumed by the decomposition reaction. atures at 200 K with and without O3 addition are shown in Fig. 7
This trend of increasing contribution of this pathway continues if compared with measurements from existing reference [157] and
the pressure is increased further. A crossover pressure, at which simulation. It can be seen that experiments from both Gao et al.
O3 decomposition starts to dominate, was observed with a value [64] and Dugger [157] deviated from simulations by approximately
of approximately 1.5 atm. 10%–15%. However, the trend from the experiments by Gao et al.
As SL positively depends on O3 decomposition, which provides [64] is consistent in that O3 addition enhanced flame propagation
atomic oxygen, the enhancement of SL due to O3 addition be- speeds of C2 H4 at cryogenic conditions in which the ozonolysis re-
comes more significant at elevated pressure, as observed in the action was suppressed at atmospheric pressure.
experiment.
In contrast to the results for saturated hydrocarbons, inconsis- 4.2.3. The effect of ozone on flame stabilization
tencies (both detrimental and beneficial effects due to O3 addi- O3 addition can also improve flame stability of laminar flames.
tion) exist in previous studies on the measurement of C2 H4 flame Vu et al. [66] demonstrated increased blowoff velocity and ex-
speeds with O3 addition. Gluckstein et al. [155] observed decreased tended flammability limits in a Bunsen burner at atmospheric
C2 H4 laminar flame speeds after O3 addition at room temperature pressure using CH4 and C3 H8 as the fuels. They observed that an
and atmospheric pressure in a Bunsen burner. They suspected that initially unstable flame can be stabilized by O3 addition. As illus-
this was due to production of CH2 O from low temperature reaction trated in Fig. 8, a C3 H8 /air premixed flame was unstable at fuel
between O3 and C2 H4 . This explanation is not adequate because rich extinction limit conditions (φ = 1.5) and the flame fluctuated.
CH2 O actually has greater laminar flame speed than that of C2 H4 . If a portion of the O2 in air was converted to O3 , the initially un-
While another work at sub-atmospheric pressure by Pinchak et al. stable flame could then be stabilized without changing the equiva-
[156] showed enhanced C2 H4 SL by O3 addition, and explained this lence ratio. The improved flame stability was attributed to the en-
with the early heat release and radical production due to O3 de- hancement of laminar flame speeds from O3 addition. Regarding
composition which produces atomic oxygen. flame speed, they reported a positive sensitivity coefficient of re-
Gao et al. [64,154] systematically investigated the effect of O3 action O3 + N2 = O2 + O + N2 on flame speed, consistent with
addition on C2 H4 laminar flame speeds and showed that the in- the fact that this reaction produces reactive atomic oxygen to en-
consistencies previously observed were due to the rapid ozonoly- hance flame propagation, and a negative sensitivity coefficient of
sis reactions of C2 H4 at 1 atm and 298 K. At atmospheric pressure the reaction O3 + H = OH + O2 . Such sensitivity analysis supports
and 298 K, the ozonolysis reaction of C2 H4 is very fast compared their hypothesis that O3 + N2 = O2 + O + N2 , a chain initiation
to the reactions between CH4 and O3 , therefore, significant quanti- reaction, is more beneficial to enhance combustion. The increase
ties of C2 H4 reacted with O3 while mixing inside the burner. Heat in flame speed is used to explain the increase in blowoff veloc-
release from ozonolysis reactions was then lost to the ambient, ity and improved flame stability. With 3810 ppm O3 addition, the
which caused decreased SL to be observed in the experiments con- extension of blowoff velocity at stoichiometric conditions was ap-
ducted by Gluckstein et al. [155]. While at sub-atmospheric pres- proximately 9% for a CH4 flame and 11% for a C3 H8 flame [66].
sure conditions, as in the work conducted by Pinchak et al. [156], Consistently, Zhang et al. [158] found that the flammability limits
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 11

Fig. 7. Effect of O3 addition on C2 H4 SL at atmospheric pressure and T0 = 200 K. (Reprinted from Gao et al. [64] with permission of Elsevier.)

Fig. 8. Enhancement of the stability of a C3 H8 /air flame at atmospheric pressure


and room temperature by O3 addition. (Reprinted from Vu et al. [66] with permis-
sion of Elsevier.)

of syngas were substantially extended by O3 addition. Furthermore,


O3 could sustain syngas flames at higher dilution ratios.
The kinetic effect of O3 addition on flames was further inves-
tigated by Weng et al. [159]. A significant increase of CH2 O con- Fig. 9. CH2 O-PLIF for a turbulent CH4 /air flame (a) without O3 and (b) with
centration due to O3 addition was detected in a laminar CH4 /air 4500 ppm O3 where φ is the equivalence ratio. (Reprinted from Weng et al.
[159] with permission of Elsevier.)
Bunsen flame using CH2 O-PLIF (planar laser-induced fluorescence),
with effect greater at rich conditions compared to the stoichio-
metric case. With 4500 ppm O3 addition, the CH2 O concentration
kinetic process changes in turbulent flames with O3 addition owing
was enhanced by 58.5% at a fuel rich condition (φ = 1.4) and
to the expansion of the preheating zone and turbulent transport.
15.5% at a stoichiometric condition. The experimental measure-
ments agreed well with simulation results. To isolate the effects
of the flame, they measured the CH2 O production in heated reac- 4.3. The effect of ozone on turbulent flames
tants (no flame), and found that the temperature needed to pro-
duce CH2 O was significantly lowered by O3 addition. Their sim- Only a few works have been devoted to the effect of O3 addi-
ulation showed that the methoxy radical (CH3 O) was the key tion on turbulent flames. Weng et al. [159] measured the concen-
species for the production of CH2 O at lower temperature condi- tration of CH2 O in turbulent CH4 /air flames with or without O3 ad-
tions. In laminar flames, CH2 O was formed via decomposition of dition, using a water-cooled McKenna burner with a jet tube along
CH3 O (CH3 O + M = CH2 O + H + M) and CH3 O was produced via the central axis. The Reynolds number of the turbulent flame was
reactions CH3 + HO2 = CH3 O + OH and CH3 + O3 = O2 + CH3 O. maintained at 6110 with the jet exit flow velocity of 4 m/s. Consis-
However, in the pre-heated gas case, CH2 O was formed owing to tent with observations for laminar flames conducted in the same
the reaction CH3 O + O2 = CH2 O + HO2 and CH3 O comes from work, they observed increased CH2 O concentration in turbulent
reaction CH3 + O3 = O2 + CH3 O only as the production of HO2 flames. Typical CH2 O PLIF images of turbulent CH4 /air premixed
is very low without the presence of a flame. Based on the results flames are presented in Fig. 9. It was found that such enhance-
presented by Weng et al. [159], it is very interesting to see how the ment on CH2 O production was greater than in laminar cases with
12 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

the same quantity of O3 addition (4500 ppm). It was also observed


that there was increased CH2 O concentration in the downstream
region of the flame. The authors explained this phenomenon by
the turbulence-enhanced mixing of unburned and hot gases. No
detailed kinetic analysis was conducted for the turbulent flow case.
Ehn et al. [50] investigated CH4 /air flames in a model gas
turbine swirl stabilized combustor with O3 addition. PLIF mea-
surement showed increased CH2 O concentration as small quanti-
ties of O3 were added owing to the acceleration of the reaction
CH3 + O = CH2 O + H. Their simulation revealed that O3 enrich-
ment could increase laminar flame speed (∼10%) and extend the
extinction strain rate (∼20%) with 0.57% (by volume) O3 addition.
Sensitivity analysis showed that the extension of extinction strain
rate was primarily due to the acceleration of chain branching reac-
tions related with atomic oxygen produced from O3 . The increased
extinction strain rate enabled the flame with O3 addition to burn
under more turbulent conditions than would be possible without
O3 addition. It is worthy to note that the changes of kinetic pro-
cesses presented by Weng et al. [159] and Ehn et al. [50] were not
identical for laminar flames and turbulent flames. More work is re-
quired to understand the interaction between O3 kinetics and flow
field effects to understand the turbulence/chemistry coupling ef- Fig. 10. Direct photos of n-heptane (a) O3 -activated cool diffusion flame and (b)
fect. normal high temperature diffusion flame. (Reprinted from Won et al. [163] with
A very important phenomenon which has not been studied un- permission of Elsevier.)
der turbulent conditions is the effect of O3 addition on hydrody-
namics. In turbulent flames, the flame is stretched and has cur-
vature and equivalence ratio fluctuations. Since flame speed en- both had a 13 mm inner diameter at the exit. The fuel (n-heptane)
hancement by O3 addition can be very sensitive to small differ- was diluted by N2 and heated to 550 K. The mildly heated fuel
ences in the flow field interaction with the flame, the coupling be- promoted O3 decomposition to release atomic oxygen, which in-
tween kinetics and hydrodynamics changes. In the work by Om- duced the initial fuel oxidation. The early oxidation allowed for
brello et al. [63], the change of kinetic and hydrodynamics owing subsequent low temperature chain-branching reactions and cool
to O3 addition was discussed using a laminar lifted jet diffusion flames were observed as shown in Fig. 10(a). Numerical simula-
flame. The laminar lifted jet diffusion flame featured a premixed tions showed that the cool flame extinction limit was substantially
curved flame head at the base to anchor the jet flame, followed by extended by a factor of 3 with 3% (by volume) O3 addition, sug-
a diffusion flame tail. With O3 addition, the non-uniform enhance- gesting that the low temperature chemistry timescale was signifi-
ment of laminar flame speed with the equivalence ratio induced cantly shortened by adding O3 [163]. Ozone addition promotes low
an increase in the radius of the triple flame front since the lean temperature oxidation processes by producing atomic oxygen prior
and rich premixed flames were enhanced more. The larger radius to the reaction zone. Atomic oxygen then reacts with fuel to gen-
of the flame led to more significant flow redirection upstream of erate OH radicals, which are critical to activate low temperature
the flame. Therefore, the local flow velocity at the premixed flame chemistry through the pathways of C7 H16 + OH = C7 H15 + H2 O
head decreased and allowed for enhanced lifted flame propagation and C7 H15 + O2 = C7 H15 O2 reactions. Further sensitivity analysis
speeds. In the work by Pinchak et al. [156], it was shown that ax- showed that the cool flame was governed predominantly by low
ial stretch rate affected flame speed enhancement in a nearly one- temperature reactions (relevant to formation of RO2 and QOOH)
dimensional laminar flame from a Hencken burner. This is owing and O3 decomposition as shown in Fig. 11(a) rather than those
to the increased flux of H + O3 reaction at higher flame stretch well-known chain branching/propagation reactions for hot flames,
rate to provide increased enhancement. These works set the stage such as H + O2 = O + OH and CO + OH = CO2 + H. Sensitivity
to help explain the more profound effect of O3 addition under tur- analysis on the cool flame extinction strain rates on diffusive trans-
bulent flow conditions. port was also conducted and the results are shown in Fig. 11(b). It
is clear that the diffusion of O3 and intermediate species relevant
4.4. Ozone induced cool flames to low temperature chemistry, such as acetaldehyde (CH3 CHO) and
CH2 O, are critical to cool flames. Cool flames become unstable at
Ozone was used to induce the formation of cool flames [160– high strain rate (defined as the gradient of axial flow velocities ac-
164]. Cool flames were observed accidentally more than 100 years counting for the density difference between the fuel and oxidizer
ago [165]. Unlike high-temperature flames, the temperature of cool streams [163]) or high fuel dilution levels. At low strain rate or
flames is generally around or below 10 0 0 K. These types of flames low fuel dilution levels, normal high-temperature flames were ob-
receive attention since low temperature chemistry, which plays a served instead of cool flames as shown in Fig. 10(b).
key role in cool flames, is relevant to practical problems such as In a work that followed, Ju et al. conducted numerical simula-
engine knock and fire safety. However, it is challenging to obtain tions on premixed dimethyl ether (DME) flames [160]. Three dif-
stable self-sustained cool flames since the low temperature oxida- ferent branches were observed: normal high-temperature flames,
tion reactions have long induction times preventing radical chain double flames, and cool flames, as shown in Fig. 12. It was ob-
branching [163]. After initiation, the cool flame either quenches or served that cool flames can exist at the conditions beyond the
transits to a hot flame. Therefore, addition of O3 , which could as- flammability of conventional high temperature flames, thus signif-
sist the initiation and branching of radicals, has been proposed as a icantly extending the flammability of the DME mixture. However,
method to induce cool flames. Won et al. [163] demonstrated sta- for freely propagating flames at an initial temperature 300 K, only
ble cool flames experimentally using a counterflow burner with O3 a normal high temperature flame branch was observed. To observe
added to the oxidizer side initially. The upper and lower burners cool flames, the temperature of the reactants needed to be raised
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 13

Fig. 11. Sensitivity analysis of extinction strain rate to (a) reaction rate constants; (b) the species diffusivity, for n-heptane cool diffusion flame. X refers the mole fraction, Tf
refers the initial fuel stream temperature and T0 refers the initial oxidizer temperature. (Reprinted from Won et al. [163] with permission of Elsevier.)

creased the flame reactivity more significantly compared to merely


heating the reactants. Therefore, O3 kinetically promoted the for-
mation of cool flames as O3 decomposition releases atomic oxygen
in the preheating zone of the cool flame, thus accelerating chain-
branching and stabilizing cool flames at low temperature. Similarly,
for stretched counterflow flames, it was reported that O3 addition
could extend the range of stretch rates at which cool flames can
exist [160]. Sensitivity analysis of the flame speed to reaction rates
and mass diffusivity was performed, as illustrated in Fig. 13, which
shows the cool flame is sensitive to typical low temperature re-
actions and diffusivities of intermediate species. However, for high
temperature flames, it is more sensitive to the reactants and prod-
ucts.
Fig. 12. Dependence of DME/O2 flame temperature on equivalence ratio for The numerical prediction of the DME premixed cool flame
DME/O2 freely propagating premixed flames. (Reprinted from Ju et al. [160] with activated by O3 addition was later experimentally demonstrated
permission of Elsevier.)
by Reuter et al. [161]. The experiment was performed at atmo-
spheric pressure in a counterflow burner with 13 mm diameter,
with very lean premixed reactants at 300 K, and the counterflow
to 530 K, or O3 needed to be added in the reactants. It is true being heated N2 (600 K). Cool flames were observed when O3 was
that the enthalpy was increased as part of O2 was replaced by O3 , added and different flame dynamics compared to normal high-
but the authors [160] explained that the formation of cool flames temperature flames was exhibited. For example, the location of the
at 300 K with O3 addition was not solely a thermal effect. They cool flame showed little dependence on equivalence ratio, indicat-
showed that even at a lower mixture enthalpy, O3 addition in- ing the weak dependence of its flame speed on equivalence ratio.

Fig. 13. The normalized DME/O2 flame speed sensitivities to (a) the reaction rates for cool flames, and (b) the mixture averaged mass diffusivity for both high temperature
flames (HTF) and cool flames. (Reprinted from Ju et al. [160] with permission of Elsevier.)
14 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

Fig. 14. Raw CH2 O PLIF images of (a) a O3 -activated DME/O2 cool flame and (b) a DME/O2 normal high-temperature flame at the same flow conditions, (c) the measurement
along the centerline for (a) and (b), and (d) the calculated CH2 O mole fraction and heat release rate (HRR). (Reprinted from Reuter et al. [161] with permission of Elsevier.)

The result is in contrast to normal high-temperature flames where ozone-less cool flames increases significantly with the chain length
there is a well-known strong correlation between the flame speed of n-alkane. In contrast, the extinction strain rates of normal high-
and equivalence ratio. Difference in flame structure was also ob- temperature flames were not very sensitive to the chain length of
served. CH2 O PLIF measurements showed that the CH2 O profile for n-alkane molecules. However, for the case with O3 , as the low-
the cool flame was much broader than that of its normal high- temperature reactivity was significantly promoted, the reactivity of
temperature flame counterpart, as compared in Fig. 14. the cool flames could be essentially independent of the size of n-
Hajilou et al. [162] further investigated O3 activated DME pre- alkane molecules.
mixed cool flames using a Hencken burner at sub-atmospheric
pressures. Such a platform has been used to study freely prop- 4.5. The effect of ozone on combustion through ozonolysis reactions
agating flames [166], and was used to measure the propagation
speeds of cool flames [162]. Consistent with previous work [161], While the majority of fuel oxidation reactions require high-
the flame speed was found to be insensitive to equivalence ratio at temperature conditions, those involving O3 and alkenes (ozonol-
very lean conditions. However, it was also reported that the propa- ysis), occur very rapidly at low temperature conditions, e.g., 300 K,
gation speed decreases as the equivalence ratio increases from 0.6 as indicated by the reaction rate constants in Table 1. The sub-
to 1.4, shown in Fig. 15. The peak flame temperatures decreased as sequent radical production and heat release could induce in situ
equivalence ratio increased in the reported range, which is in stark fuel reforming, or even autoignition in a fuel/air mixture. It is
contrast to normal high-temperature flames. anticipated that ozonolysis could significantly enhance combus-
As a continuation of the previous work [161], Reuter et al. tion processes. For mixtures containing O3 and alkenes, ozonoly-
[164] investigated cool diffusion flames for large n-alkanes. As the sis is dominant at low temperature conditions (approximately be-
presence of O3 slightly complicates the chemical kinetics, they gen- low 700 K) and O3 decomposition is only important at higher tem-
erated “ozone-less” cool flames by turning off the O3 generator perature. A simulation of autoignition in a closed constant pres-
after cool flames were established. Without O3 , the cool flames sure homogeneous reactor at 1 atm was performed by the au-
were much dimmer in terms of the chemiluminescence intensity, thors of this paper and is shown in Fig. 17 for a mixture of
as shown in Fig. 16, but remained self-sustained. “Ozone-less” cool C2 H4 /O2 /N2 /O3 = 6.3:76.2:13.1:4.4. The simulation was conducted
flames were generated using this method for large n-alkanes rang- using USC Mech II [167] combined with a sub O3 mechanism
ing from n-heptane to n-tetradecane. Similar to previous work, [65] including simplified ozonolysis reactions [64], which assume
the cool flames showed different behavior compared to normal four branches equally for the decomposition of Criegee intermedi-
high-temperature flames. It was found that the chain length of ate produced from C2 H4 ozonolysis: (1) CH2 O + OH + HCO, (2)
n-alkanes significantly affected the ozone-less cool flame reactiv- CH2 O + 2H + CO2 , (3) CH2 O + H2 + CO2 , (4) CH2 O + H2 O + CO.
ity. Given the same fuel mole fraction, the extinction strain rate of The sub-O3 kinetic model including simplified ozonolysis reactions
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 15

Fig. 15. O3 -activated Cool DME flame propagation speeds derived from lift-off height analyses at sub-atmospheric pressures. (Reprinted from Hajilou et al. [162] with
permission of Elsevier.)

tively slow. The effects of ozonolysis become less significant as ini-


tial temperature increases. As shown in Fig. 17, the simulation re-
sults using the models with and without ozonolysis (red and blue,
respectively) gradually converge as initial temperature increases.
These simulation results imply that, at low temperature, ozonol-
ysis of alkenes becomes the dominant O3 consumption pathway
and can significantly promote the combustion to the extent that
O3 decomposition cannot achieve. At temperatures above approxi-
mately 800 K, the effect of ozonolysis reactions becomes negligible
as O3 decomposition becomes very fast. Regardless, the simulation
Fig. 16. Direct photographs of n-dodecane cool flames with (a) and without (b)
results show that O3 can dramatically change the autoignition de-
ozone addition. (Reprinted from Reuter et al. [164] with permission of Elsevier.) lays at any temperatures. As shown in the simulation, autoignition
can be easily activated by ozonolysis reactions even at room tem-
perature conditions. Following this idea, ozonolysis reactions have
been proposed as a new way to enhance combustion processes
[168–172].
By adding O3 into a C2 H4 /O2 /N2 system, different phenomena
were observed [168–172]. In those studies, non-premixed burners
were used to avoid the heat released from ozonolysis being lost to
the ambient as the time scale of ozonolysis reactions is in the or-
der of milliseconds. An experimental setup used in these studies
is shown in Fig. 18. To minimize the fluctuation of O3 concentra-
tion, a mixing buffer was added between the O3 generators and
the O3 monitor. During the experiment, initially the oxidizer and
O3 generators were turned on, but the fuel stream was turned off.
When the generation of O3 reached steady state, fuel was suddenly
turned on, and autoignition may occur, which was recorded by a
high-frame-rate camera.
Fig. 17. Simulation of autoignition delay for C2 H4 /O2 /N2 /O3 = 6.3:76.2:13.1:4.4 It was observed that ozonolysis reactions initiated autoigni-
(φ = 1) at 1 atm as a function of initial temperature. (For interpretation of the
tion events, as illustrated in Fig. 19. Before the autoignition ker-
references to colour in this figure legend, the reader is referred to the web version
of this article.) nel was formed, a region with weak filtered chemiluminescence
centered at 430 nm with ±10 nm bandwidth was observed, as in-
dicated for t = 1.6–2 ms in Fig. 19. While further investigations
of C2 H4 is provided as Supplementary Material for the readers’ ref- are necessary to identify the emitters of the chemiluminescence
erence. In this simulation, reactions between Criegee intermedi- observed around 430 nm before autoignition, one possible emit-
ates and other hydrocarbons are omitted because they are largely ter is CH2 O∗ (electronically excited CH2 O) whose emission cov-
unknown. The simplification artificially increases the reactivity of ers a wide band (centered at 420–430 nm [174]). The chemilu-
the mixture because the decomposition of the Criegee intermedi- minescence could not arise from soot, as there is no combustion
ates generates radicals and H2 and is faster than reactions between at this moment and therefore no soot. Typically, CH2 O∗ chemilu-
Criegee intermediates and hydrocarbons. Therefore, this simulation minescence is not used to study conventional high temperature
is for demonstration only. Without O3 addition, such a mixture (>10 0 0 K) flames because at such condition CH∗ chemilumines-
does not autoignite at low temperature conditions, but with O3 cence (which is centered at 431.5 nm [175,176]) from combustion
addition, it only takes 10 ms to autoignite at 300 K, as shown in is significantly stronger than CH2 O∗ chemiluminescence. However,
Fig. 17. The result is primarily due to ozonolysis, without which it it is possible that at low temperatures CH2 O∗ becomes the ma-
takes more than 10 0 0 ms to autoignite as O3 decomposition is rela- jor chemiluminescence emitter. In an early work, Sheinson and
16 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

Fig. 20. Simulation results of plug flow reactor model at 101 kPa. Inlet conditions:
U0 = 2.37 m/s, T0 = 300 K, C2 H4 /O2 /O3 /N2 = 9.6/66.5/3.7/20.2. (Reprinted from Gao
et al. [172] with permission of IOP Publishing.)

Fig. 18. Experimental setup of non-premixed jet flame for the study of ozonolysis
activated autoignition. (Reprinted from Gao [173].)
from CH∗ . At intermediate times (t = 1.6–2.4 ms) chemilumines-
cence from both CH2 O∗ and CH∗ existed, as is evident in Fig. 19 at
2 ms where relatively weak chemiluminescence from CH2 O∗ (pre-
Williams [174] obtained resolved chemiluminescence spectra from sumed emitter) and strong emission from the ignition kernel from
cool flames (473–673 K) and concluded that the primary emission CH∗ were observed. At highly diluted conditions without a flame,
source was CH2 O∗ . Other possible emission sources such as CO∗ , gas chromatography (Inficon Fusion μGC) sampling confirmed the
CH∗ , C2 ∗ , OH∗ , and HCO∗ were shown to be insignificant at such existence of CH2 O, CO, H2 , CO2 , which are products of ozonolysis
conditions. Particularly, CH∗ emission only becomes comparable reactions. Although OH could be produced directly from ozonolysis
with the CH2 O∗ signal at higher temperature (∼1073 K). Therefore, reactions, its chemiluminescence was not observed probably due to
CH2 O∗ can be used to identify cool flames owing to low temper- its extreme low concentration at low temperature conditions. The
ature chemistry, such as demonstrated in Ref. [177]. Chemilumi- findings implied that autoignition occurred in such a sequence: ox-
nescence from CH2 O∗ was also reported in ozonolysis reactions idizer and fuel were mixed in the shear layer; at the same time,
of C3 H4 (propadiene) at room temperature conditions (no flame) ozonolysis reactions occurred and induced heat release and forma-
[178]. A simulation conducted using a PLUG flow reactor [179] with tion of chemiluminescence emitters (presumed CH2 O∗ as discussed
a sub-mechanism of OH∗ , CH∗ , and CO2 ∗ [180,181] shows that at above); and finally autoignition was induced by the combined ef-
the experimentally interrogated condition the mole fraction of CH∗ fect of reactive species and heat production from ozonolysis and
is less than 10−18 so its effect is negligible before hot ignition. its subsequent reactions.
Thus, the authors of Ref. [172] presume the filtered chemilumines- The role of ozonolysis at low temperature can be further illus-
cence before hot ignition (t ≤ 1.6 ms) was from CH2 O∗ and the fil- trated using simulations in a PLUG flow reactor, and the results are
tered chemiluminescence after hot ignition (t = 2.4–4.0 ms) was shown in Fig. 20. The kinetic model used in this simulation is the

Fig. 19. Filtered chemiluminescence images centered at 430 nm of ozonolysis activated autoignition, Re = 1317. (Reprinted from Gao et al. [172] with permission of IOP
Publishing.)
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 17

Table 2 nation for the flame stabilization and propagation mechanism cor-
Summary of PLUG products and corresponding burning velocity.
respondingly. Fuel was “preprocessed” in situ owing to ozonolysis
Reactor length (cm) 0 2 3.5 3.8 4 reactions upstream of the flame front, which released heat and
Amount of fuel reacted 0% 3% 11% 13% 15% produced reactive species. For large hydrocarbon molecules, the
Exit temperature (K) 300 372 538 594 641
transport properties were also modified. Therefore, the reactivity
Flame speed (m/s) 2.2 2.8 4.3 5.7 10.5
Burning rate (g/s cm2 ) 0.28 0.29 0.30 0.36 0.62 of the fuel-oxidizer mixture increased, and the flame propagation
Increase of burning rate 0% 1% 6% 27% 115% speed was significantly increased. During the interval of t = 2.0–
Mole fraction∗ O2 67% 66.4% 66.0% 65.8% 65.6% 3.6 ms in Fig. 19, the front of the reaction zone propagated approx-
N2 20% 20.1% 20.0% 19.9% 19.9% imately 1.8 cm, corresponding to a very high propagation speed
C2 H 4 10% 9.3% 8.6% 8.3% 8.1%
(∼11 m/s) relative to the burner. For comparison, the laminar flame
O3 4% 3.3% 2.3% 2.0% 1.8%
CO 0 0.2% 0.9% 1.1% 1.3% speed calculated by PREMIX [182] is 2.0 m/s at T0 = 300 K without
H2 O 0 0.2% 0.8% 1.0% 1.2% including ozonolysis reactions in the kinetic model. An extreme
CH2 O 0 0.2% 0.5% 0.6% 0.7% event of fuel preprocessing is an autoignition kernel formed up-
CO2 0 0.1% 0.4% 0.5% 0.6%
stream of an existing combustion zone. One example is given in
H (ppm) 0 0.0 0.0 0.1 0.1
O (ppm) 0 0.0 0.0 0.1 0.2 Fig. 22. At t = 0, the first detectable autoignition kernel (Kernel-1)
OH (ppm) 0 0.8 1.3 1.4 1.5 was generated at approximately 11.5 cm downstream of the burner

exit. For a time 0.25 ms later, a second autoignition event (Kernel-
Only O, H, OH and species with mole fraction greater than 0.1% are listed.
2) was generated upstream of Kernel-1. Both kernels started to
grow and finally merged. The authors also noticed that once the
flame propagated to the lower boundary of the CH2 O zone (below
same as the one used to produce the results in Fig. 17. Simula- which CH2 O concentration was negligible), the propagation speed
tion results show that autoignition is a two-stage process. Start- of the flame was significantly reduced. The result indicated that
ing from the inlet of the reactor, temperature starts to increase the abnormally high flame propagation speed was a result of high
gradually owing to ozonolysis reactions and induced subsequent CH2 O concentration and elevated temperature.
reactions. At approximately 5.5 cm downstream, the temperature The autoignitive environment also allows for the interaction be-
reaches a plateau at approximately 800 K. In the meantime, both tween autoignition kernels and the flame front. Gao et al. [184] ob-
O3 and C2 H4 concentrations continue to decrease (from the inlet served the oscillation of the flame front due to the formation and
of the reactor) due to the production of CH2 O and other ozonolysis blow-out of the autoignition kernel upstream of the flame front.
products. Once O3 is consumed, ozonolysis reactions cease and the Fig. 23 (broadband line-of-sight chemiluminescence) showed an
temperature reaches a plateau. Because the dominant product from example of this process. As shown in Fig. 23(a)–(d), autoignition
ozonolysis reactions, CH2 O, is relatively stable at this temperature, kernels formed upstream of the main flame front. This effectively
reactions slow and the temperature does not change substantially. decreased the lift off height from 2.55 cm in Fig. 23(a) to 1.21 cm in
Following the temperature plateau, complete combustion occurs. It Fig. 23(f). However, the new flame front formed by the autoignition
was observed that ozone decomposition only starts to dominate kernels failed to propagate downstream further owing to the large
when the temperature is above 800 K. It is worth noting that the jet velocity and decreased radical concentrations and temperature
reaction O3 + H = OH + O2 also plays an important role for O3 owing to the decreased level of ozonolysis closer to the nozzle exit:
consumption. If ozonolysis reactions are turned off in the simula- once the kernels extinguished, as shown in Fig. 23(f)–(j), and the
tion, no intermediate temperatures or CH2 O can be observed; O3 liftoff height of the flame returned to a similar level as Fig. 23(a).
decomposition only occurred in the preheat zone very close to the The results discussed above illustrate that it is possible to modify
reaction zone. The results of the simulation also indicate that be- the flame propagation and stabilization mechanism using the au-
fore hot ignition occurred and in a region away from the reaction toignitive environment created using rapid exothermic ozonolysis
zone, ozonolysis is the dominant O3 consumption pathway and is reactions.
responsible for the induced autoignition phenomenon. The inter-
mediate products and temperature are listed in Table 2, together 5. Fuel processing using ozone
with the corresponding simulated burning velocities. Besides heat
released from ozonolysis reactions, reactive species and radicals 5.1. Treatment of solid fuels
are produced, supporting the claim that ozonolysis reactions en-
hance the combustion via kinetic effects. Ozone treatment for coal desulfurization was attempted by
To better understand the experiment, a 3D FLUENT transient Steinberg et al. [67]. In their work, enrichment of the sulfur (SO2 )
simulation was also conducted. The kinetic model employed in this in the effluent gas was detected after O3 treatment of solid coal
simulation was USC Mech II [167] combined with a sub-O3 mech- in a glass tube, indicating the possibility of desulfurization. It was
anism [65] including simplified ozonolysis reactions [64], as pre- reported that the organic sulfur was harder to remove than the
sented in the Supplementary Material. The simulation was able to pyritic sulfur. A sulfur balance for the reaction system indicated
predict similar transient phenomena observed in the experiment. that the sulfur content of the coal sample was reduced by approx-
As illustrated in Fig. 21, at t = 0, the maximum temperature was imately 0.1–1.3%. The authors concluded that such sulfur enrich-
only 831 K with incoming reactants at 300 K and a CH2 O region ment could not indicate a practical desulfurization process unless
downstream was formed with peak mole fraction of ∼3%. Approxi- the enrichment can be significantly increased. No further discus-
mately 2 ms later, a small elevated temperature region was formed sion on the kinetic process was presented.
inside the CH2 O region and the maximum temperature increased Because O3 can react with almost all functional groups of lignite
to 2068 K, indicating the formation of a hot ignition kernel. The ac- organic matter, even with polyaromatic fragments at room tem-
cumulation of CH2 O before autoignition upstream of the localized perature [185], O3 has been used as one reagent to treat lignite.
elevated temperature region was consistent with the experimen- Sharypov et al. [186] investigated the influence of ozonization of
tally observed process shown in Fig. 19. Kansk-Achinsk lignite on the chemical composition, structure, and
With such an autoignitive environment, flame propagation modified lignite reactivity. Lignite particles of size less than 0.1 mm
showed different characteristics. Based on simulation results and were treated by a flowing mixture containing O2 and O3 (2% mole
experimental observations, the authors [172] proposed an expla- fraction) at room temperature in a quartz reactor. The ozonized
18 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

Fig. 21. Cross-sectional plane of 3D FLUENT simulation of the generation of autoignition kernels in a coflow burner at atmospheric pressure and T0 = 300 K. Center flow is
C2 H4 and coflow is O2 /N2 /O3 , (a, b) heat release rate; (c, d) CH2 O mole fraction; (e, f) temperature profile. (Reprinted from Gao et al. [172] with permission of IOP Publishing.)

Fig. 22. Chemiluminescence of observed autoignition kernels in a coflow burner at atmospheric pressure and T0 = 300 K. Center flow is C2 H4 and coflow is O2 /N2 /O3 .
(Reprinted from Gao et al. [183].)

lignite was compared with those treated by air or helium at 100 °C. able to reduce the surfactant absorptivity of fly ash carbon. A re-
Chemical analysis indicated the incorporation of oxygen into the cent work by Patrakov et al. [188] studied the treatment of coal
organic matter of lignite during ozonization. The authors suspected organic matter (COM) by O3 . Redistribution of atomic oxygen in
that the oxygen might be attributed to simple and complex ethers the functional groups was detected after ozonization because of
and oxygen heterocycles. It was also found that the oxygen in car- changes in the macromolecular organization. Covalent bonds in-
bonyl groups (>C=O, –COOH) increased while the concentration side large macromolecular fragments are destroyed by ozonolysis
of hydroxyl groups was reduced after ozone treatment. Ozoniza- reactions, which in turn decreased the asphaltenes under thermal
tion also resulted in loosening of the overall molecular structure liquefaction and increased the low weight compound yields. One
of lignite based on results from X-ray diffraction measurement. important kinetic effect is the decreased effective activation energy
Ozonized lignite was also reported to have higher reactivity in hy- of thermal liquefaction after O3 treatment, as shown in Fig. 24. It
drogenation processes in the presence of hydrogen–donor solvents was shown that C–O bonds were formed in the COM structure of
and pyrite catalysts. Gao et al. [187] demonstrated that O3 was ozonized coals. These bonds have a lower dissociation energy value
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 19

Fig. 23. The instantaneous images of the C2 H4 jet with O2 /O3 /N2 coflow at atmospheric pressure and T0 = 300 K showing oscillatory flame liftoff heights. The time differences
between frame (a) and the remaining frames are (b) 0.6 ms, (c) 1.0 ms, (d) 2.0 ms, (e) 2.8 ms, (f) 3.0 ms, (g) 4.2 ms, (h) 7.4 ms, (i) 9.4 ms, (j) 10.4 ms, (k) 11.4 ms, (l) 11.8 ms,
(m) 12.2 ms, (n) 13.0 ms, (o) 14.0 ms, (p) 15.0 ms, (q) 16.0 ms, (r) 17.0 ms, (s) 18.0 ms, and (t) 19.0 ms. (Reprinted from Gao et al. [184].)

than C–C bonds, which explained the decreased effective activation typical molecular structure of pour point depressants for conven-
energy. tional diesel, which act on the growing sites of the crystals during
cooling. Complete solidification of sunflower oil biodiesel was con-
5.2. Treatment of liquid fuels ducted. Treatment with 1% ozonized vegetable oils minimized the
solidification. Microscopic analysis revealed that neat sunflower oil
Ozone has also been applied to treat renewable fuels. The crys- biodiesel fused together into large irregular-shaped agglomerates
tallization of saturated fatty acid methyl ester (FAME) during cold after subjecting it to temperatures of 3 °C for 20 min as shown in
weather seasons can cause fuel starvation and operability prob- Fig. 25. However, the ozonized sunflower oil impeded the agglom-
lems. For this reason, the pour point needs to be depressed for eration, giving smaller crystals with sizes of approximately 10 μm
neat biodiesel derived from vegetable oils. Ozone was investi- which maintained the fluid flow properties and improved the pour
gated as a pour point depressant by Soriano et al. [189]. In their point of biodiesel.
work, ozonized vegetable oil was prepared by bubbling an O3 /O2 Ozone treatment has also been investigated to modify fuel
mixture into vegetable oil. Ozonized vegetable oils (1%–1.5% by reactivity adaptively to control ignition timing. Schönborn et al.
weight) were reported to be effective in reducing the pour point [68] demonstrated a method to control ignition timing under HCCI
of biodiesel prepared from sunflower oil, soybean oil and rape- combustion conditions by modifying the molecular structure of
seed oil to −24, −12 and −30 °C, respectively. The effect is due the fuel using ozone. The result was achieved by bubbling air
to the ozonolysis reaction of alkenes in vegetable oil with for- that contained O3 (4% by volume) through a variable proportion
mation of 1,2,4-trioxolane polar rings. The polar groups are the of the fuel (1,1-Diethoxyethane) prior to injecting into the engine
20 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

treated fuel were identified, which indicates the complexity of the


underlying chemical reactions.

6. Emission control using ozone

The effect of O3 addition on pollutants from combustion sys-


tems has been investigated as well. In 1980, Slater and Rizzone
[69] conducted experiments to investigate the removal of sulfur
oxide by ozone. Ozonated O2 was mixed with SO2 , NO, and Ar in
a heated stainless-steel reactor. A cold trap maintained at −80 °C
was used downstream to condense SO3 and NO2 formed in the re-
actor. Without NO addition, O3 decomposed to form atomic oxygen
and then reaction SO2 + O + M = SO3 + M converted SO2 to SO3
Fig. 24. The effective activation energy of thermal liquefaction process of (1) initial inside the reactor. The SO2 conversion ratio was found to be non-
and (2) ozonized coal vitrinites. (Reprinted from Patrakov et al. [188] with permis-
monotonic with increased reactor temperature, initially increas-
sion of Elsevier.)
ing and then decreasing. The SO2 conversion ratio was reduced
in the presence of NO, which consumed O3 quickly through reac-
tions NO + O3 = NO2 + O2 and NO2 + O = NO + O2 . Tachibana
et al. [62] measured the concentration of CO, unburned hydrocar-
(single-cylinder direct injection diesel engine). The bubbling pro- bons, and NOx in the exhaust gas of a diesel CFR (Cooperative Fuel
cess lasted approximately 4 h at 293 K and 1 atm. The contact Research) engine with and without O3 addition. Ozone produced
of fuel with ozone resulted in partial oxidation of the fuel. Even by a commercial ozonizer was introduced through the inlet port.
though the change of fuel composition was not clear, elemental In the case without combustion, the authors measured the con-
analysis carried out before and after O3 treatment suggested an centration of O3 in the intake and exhaust of the engine and no
increase in oxygen content occurred during the oxidation of the change in concentration was found. The result indicated that the
fuel with O3 . The authors suspected that the reactions between lifetime of O3 was quite long and therefore all of the O3 that was
fuel and O3 -containing air converted a portion of the fuel to per- produced would be available for reaction in the engine. In case
oxide (e.g., 1-ethoxy-1-(ethylperoxy)ethane) based on nuclear mag- of combustion, the advancement of crank angle and improvement
netic resonance (NMR) spectra and standard iodometric titration on ignition (decrease ignition delay and increase cetane number)
analysis. The hydroperoxides in the fuel induced by reactions with were observed with ozone addition. With O3 addition, the concen-
ozone could promote autoignition through decomposition to pro- trations of CO and unburned hydrocarbons decreased with the in-
duce OH. It was shown that as the proportion of oxidized fuel in- crease of O3 concentration. Regarding NOx , the emission slightly
creased, the ignition timing was advanced more in the engine test. decreased and then slightly increased with the increase of O3 con-
As a continuation of the work, Schönborn et al. [60] further investi- centration. The authors suspected that the increased NOx concen-
gated the possibility of controlling the ignition timing with another tration was caused by the changes of combustion temperature as
fuel, 1-hexene, by altering its molecular structure using prior reac- the enthalpy increased after adding O3 . Filter paper was placed in
tion with O3 . NMR was used to analyze the fuel before and after its the exhaust to trap particulates, and the contamination level was
reaction with O3 . Several oxygenated products were identified in estimated using the light intensity passing through the filter pa-
the fuel treated by O3 , which resulted in the formation of ozonide per. The level of particulates decreased as well, though not to the
molecules with a peroxidic structure. Correspondingly, an engine same extent as CO and unburned hydrocarbons decreased. Simi-
system utilizing such “smart fuel” was proposed. A portion of fuel lar results were observed by Wilk and Magdziarz [190], who sup-
was pre-processed by O3 and then mixed with the non-processed plied natural gas and air mixtures to a modified Mecker burner.
fuel. The ignition timing was therefore advanced, likely due to the The concentrations of CO, unburned hydrocarbons, NO, and NO2
early decomposition of peroxidic molecules (1-hydroperoxyhexane) were measured for cases with and without O3 . Consistent with
owing to the O3 /fuel reaction, as suspected by the authors. There- results from Ref. [62], O3 addition reduced concentrations of CO
fore, the ignition timing could be controlled via the fuel mix- and unburned hydrocarbons, and slightly increased NO concentra-
ture composition. Still, no definite composition changes of ozone tion. No effects on NO2 were detected. Besides the addition in the

Fig. 25. Sunflower oil biodiesel viewed under the microscope (200×) at 3 °C for 20 min ((A) neat sample; (B) with 1% ozonized sunflower oil). (Reprinted from Soriano Jr.
et al. [189] with permission of Elsevier.)
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 21

tions for alkenes have shown potential to enhance flame propa-


gation and stabilization, but only a limited number of studies have
been conducted on this topic. Since O3 can potentially be deployed
in situ to promote the reactivity of a system, such research is of
relevance to a number of practical devices.
Regarding the effects of O3 addition on fundamental combus-
tion processes, only several small molecule fuels have been inves-
tigated (syngas, CH4 , C3 H8 , and C2 H4 ). It is not clear how conclu-
sions change if fuels with larger molecules, or fuels with differ-
ent functional groups such as aromatics, are applied. For exam-
ple, large molecules generally break down into smaller molecules
before entering into the rapid reacting/high temperature burning
state of a flame. However, when O3 is added, ozonolysis or ozone
decomposition reactions may fundamentally change reaction path-
ways by changing both kinetic and transport properties of the mix-
tures. Furthermore, the effects of O3 addition on multi-component
fuels are also not clear.
Ozone can be economically produced at a large scale and in sig-
nificant quantities at elevated pressure conditions which is benefi-
cial to combustion systems. In most experiments, ozone was pro-
duced by commercial ozone generators typically based on corona
Fig. 26. NO conversion property with O3 added at different temperatures.
(Reprinted from Wang et al. [107] with permission of Elsevier.)
discharge which used pure oxygen as a working fluid. For most
practical applications, air has to be the working fluid, which poses
challenges on ozone production and lifetime for transport owing
oxidizer, O3 injection has also been used in processing flue gas. to the existence of nitrogen and water vapor in air. More stud-
Commercially, Cannon Technology Inc. and BOC Gases developed a ies are needed to optimize ozone generation using different exci-
low temperature oxidation (LTO) process for removing NOx emis- tation means and under different flow conditions. For example, if
sions using O3 injection in 2003 [191]. O3 was injected into the the ozone generator is based on absorption of UV radiation, selec-
cooled exhaust flue gas from a boiler. NOx was then oxidized in the tive wavelengths could be used for optimizing production. If the
oxidation chamber to N2 O5 (via reactions: NO + O3 = NO2 + O2 ; ozone generator is based on an electrical discharge, optimizing the
NO2 + O3 = NO3 + O2 ; NO2 + NO3 = N2 O5 ), with a portion of reduced electric field of the discharge can potentially maximize the
CO oxidized to CO2 , and SO2 oxidized to SO3 . Later, a two-step production of select products like ozone [193].
technique was proposed by Mok and Lee [109] to remove NOx and
SO2 simultaneously using O3 . In this technique, NO was first oxi- 8. Conclusion
dized to NO2 in the ozonizing chamber and an absorber contain-
ing a reducing agent solution (Na2 S) was used downstream to re- With our increasingly growing appetite for creative energy con-
move NO2 and SO2 . It was reported that NOx removal efficiency version means by pushing the limits of traditional combustion
was approximately 95% and SO2 removal efficiency was 100%. A processes, O3 has drawn significant attention as a means to en-
detailed kinetic model of ozone–nitrogen oxides summarized from hance and control combustion/ignition processes. Compared to
the NIST Chemical Reactions Database was also presented in this other combustion enhancement techniques, potential advantages
work [109]. Wang et al. [107] investigated the possibility of simul- of O3 addition are that it can be efficiently and economically gen-
taneously removing NOx , SO2 , and Hg using O3 . Another process erated at elevated pressures and it has a relatively long lifetime to
simultaneously removing NOx and SO2 by O3 was also investigated be transported. Significant advancements in ozone assisted com-
by Sun et al. [192]. A general observation from these works is that bustion have been made in the past decade, where it has been
NO contained in flue gas can be effectively oxidized by O3 owing shown to process fuels, control emissions, and enhance ignition,
to the fast NO + O3 = O2 + NO2 reaction. However, as O3 decom- flame propagation, and improve flame stabilization.
poses at high temperature, the effect of O3 injection on pollutant Ozone decomposition at relatively low temperature is responsi-
oxidization may be reduced. For example, as shown in Fig. 26, at ble for the majority of combustion enhancements achieved with
373 K, adding 200 ppm O3 oxidized more than 80% of NO, but at O3 addition. Such reactions release reactive atomic oxygen and
673 K, adding O3 essentially did not oxidize NO as O3 decomposed accelerate the chain branching process. However, in the reaction
and recombined to O2 before reacting with NO. zone of flames, where a radical pool has already been established,
O3 decomposition may not affect the combustion process signifi-
7. Challenge and future research cantly. For example, every ∼10 0 0 ppm O3 addition could enhance a
CH4 /air laminar flame propagation rate by approximately 1%, while
While many investigations have contributed significantly to an the autoignition delay of a CH4 /air mixture could be dramatically
understanding of ozone related reactions, inconsistency remains, shortened (by several orders of magnitude) with the same quantity
which is hampering accurate kinetic modeling of O3 assisted com- of O3 addition. Another potential mechanism to enhance and con-
bustion. Previous studies suggest that several kinetic pathways trol combustion processes is via low-temperature reactions related
may exist in parallel, including direct reactions with fuel (ozonoly- to O3 , such as ozonolysis. Such reactions could initiate fuel oxida-
sis), as well as with intermediate/radical species. However, it is still tion/dissociation far upstream of the flame or combustion zone to
not clear how and to what extent the different O3 reaction path- change the molecular structures (compositions) of reactants. It can
ways affect combustion processes. It is possible that such effects potentially change both kinetic and transport properties of the re-
are dependent on local conditions, including temperature, pres- actants, therefore affecting flame/ignition characteristics. However,
sure, and chemical composition. Therefore, a systematic study con- large knowledge gaps remain with regard to a fundamental un-
ducted over a wide range of operating conditions relevant to prac- derstanding of O3 assisted combustion, especially when ozonoly-
tical combustion systems is needed. Furthermore, ozonolysis reac- sis reactions are the dominant pathways. The detailed pathways of
22 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

ozonolysis reactions remain unclear, which brings large uncertainty [22] Cipolat D. Methanol/dimethyl ether fueling of a compression ignition engine.
in simulation and the interpretation of the kinetic effects. Further- In: Proc 11th ISAF; 1991. p. 411–16.
[23] Liang C, Ji C, Gao B, Liu X, Zhu Y. Investigation on the performance of a
more, the available studies focus on small molecule fuels and re- spark-ignited ethanol engine with DME enrichment. Energy Convers Manage
search using real fuels are limited. It is not clear if or how the con- 2012;58:19–25.
clusions change when O3 addition is applied to, for example, liquid [24] Gan Y, Lim YS, Qiao L. Combustion of nanofluid fuels with the addition of
boron and iron particles at dilute and dense concentrations. Combust Flame
fuels. With the potential for many pathways for O3 to couple into 2012;159:1732–40.
the kinetics of combustion systems, there is a future full of pos- [25] Yetter RA, Risha GA, Son SF. Metal particle combustion and nanotechnology.
sibilities, from simple burners to high-speed air-breathing propul- Proc Combust Inst 2009;32:1819–38.
[26] Allen C, Mittal G, Sung C-J, Toulson E, Lee T. An aerosol rapid compression
sion systems.
machine for studying energetic-nanoparticle-enhanced combustion of liquid
fuels. Proc Combust Inst 2011;33:3367–74.
[27] Javed I, Baek SW, Waheed K. Autoignition and combustion characteristics of
Acknowledgments heptane droplets with the addition of aluminium nanoparticles at elevated
temperatures. Combust Flame 2015;162:191–206.
This work is supported by the Air Force Office of Scientific Re- [28] Tanvir S, Qiao L. Droplet burning rate enhancement of ethanol with the ad-
dition of graphite nanoparticles: influence of radiation absorption. Combust
search (FA9550-16-1-0441), with Dr. Chiping Li as technical mon-
Flame 2016;166:34–44.
itor. The authors acknowledge thorough reviews and insightful [29] Van Devener B, Anderson SL. Breakdown and combustion of JP-10 fuel cat-
feedback from both anonymous reviewers and the editor (Prof. Hai alyzed by nanoparticulate CeO2 and Fe2 O3 . Energy Fuels 2006;20:1886–94.
Wang). [30] Basu S, Miglani A. Combustion and heat transfer characteristics of nanofluid
fuel droplets: a short review. Int J Heat Mass Transfer 2016;96:482–503.
[31] Starikovskiy A, Aleksandrov N. Plasma-assisted ignition and combustion. Prog
Energy Combust Sci 2013;39:61–110.
Supplementary materials
[32] Starikovskaia S. Plasma assisted ignition and combustion. J Phys D Appl Phys
2006;39:R265.
Supplementary material associated with this article can be [33] Ju Y, Lefkowitz JK, Reuter CB, Won SH, Yang X, Yang S, et al. Plasma assisted
found, in the online version, at doi:10.1016/j.pecs.2019.02.002. low temperature combustion. Plasma Chem Plasma Process 2016;36:85–105.
[34] Serbin S, Mostipanenko A, Matveev I, Topina A. Improvement of the gas tur-
bine plasma assisted combustor characteristics. In: 49th AIAA aerospace sci-
References ences meeting including the new horizons forum and aerospace exposition; 2011.
p. 61.
[1] Christensen M, Hultqvist A, Johansson B. Demonstrating the multi fuel ca- [35] Moeck J, Lacoste D, Laux C, Paschereit C. Control of combustion dynamics in
pability of a homogeneous charge compression ignition engine with variable a swirl-stabilized combustor with nanosecond repetitively pulsed discharges.
compression ratio. SAE technical paper; 1999. In: 51st AIAA aerospace sciences meeting including the new horizons forum and
[2] Manente V, Johansson B, Cannella W. Gasoline partially premixed com- aerospace exposition; 2013. p. 565.
bustion, the future of internal combustion engines? Int J Engine Res [36] Cha M, Lee S, Kim K, Chung S. Soot suppression by nonthermal plasma in
2011;12:194–208. coflow jet diffusion flames using a dielectric barrier discharge. Combust Flame
[3] Tacina R, Mao C-P, Wey C. Experimental investigation of a multiplex fuel in- 2005;141:438–47.
jector module with discrete jet swirlers for low emission combustors. In: AIAA [37] Lee DH, Kim K-T, Kang HS, Song Y-H, Park JE. Plasma-assisted combus-
paper; 2004. p. 135. tion technology for NOx reduction in industrial burners. Environ Sci Technol
[4] Hsu K-Y, Goss L, Roquemore W. Characteristics of a trapped-vortex combustor. 2013;47:10964–70.
J Propul Power 1998;14:57–65. [38] Song C-L, Bin F, Tao Z-M, Li F-C, Huang Q-F. Simultaneous removals of NOx,
[5] Cavaliere A, de Joannon M. Mild combustion. Prog Energy Combust Sci HC and PM from diesel exhaust emissions by dielectric barrier discharges. J
2004;30:329–66. Hazard Mater 2009;166:523–30.
[6] Katsuki M, Hasegawa T. The science and technology of combustion in highly [39] Do H, Cappelli MA, Mungal MG. Plasma assisted cavity flame ignition in su-
preheated air. Symp (Int) Combust 1998;27:3135–46 Elsevier. personic flows. Combust Flame 2010;157:1783–94.
[7] Buhre BJ, Elliott LK, Sheng C, Gupta RP, Wall TF. Oxy-fuel combus- [40] Leonov SB, Kochetov IV, Napartovich AP, Sabel’Nikov VA, Yarantsev DA. Plas-
tion technology for coal-fired power generation. Prog Energy Combust Sci ma-induced ethylene ignition and flameholding in confined supersonic air
2005;31:283–307. flow at low temperatures. IEEE Trans Plasma Sci 2011;39:781–7.
[8] Schadow K, Gutmark E. Combustion instability related to vortex shedding [41] Leonov A, Yarantsev DA, Napartovich AP, Kochetov IV. Plasma-assisted igni-
in dump combustors and their passive control. Prog Energy Combust Sci tion and flameholding in high-speed flow. 44th AIAA aerospace sciences meet-
1992;18:117–32. ing, Reno, NV; 2006.
[9] Bellucci V, Paschereit CO, Flohr P, Magni F. On the use of Helmholtz res- [42] Esakov I, Grachev L, Khodataev K, Vinogradov V, Van Wie D. Efficiency of
onators for damping acoustic pulsations in industrial gas turbines. In: ASME propane-air mixture combustion assisted by deeply undercritical MW dis-
turbo expo 2001: power for land, sea, and air: American Society of Mechanical charge in cold high-speed airflow. In: 44th AIAA aerospace sciences meeting
Engineers; 2001. p. V002T02A6–V00VT02A6. and exhibit; 2006. p. 1212.
[10] Woestman J, Logothetis E. Controlling automotive emissions. Ind Physicist [43] Ikeda Y, Nishiyama A, Kaneko M. Microwave enhanced ignition process for
1995;1:20. fuel mixture at elevated pressure of 1 MPa. In: 47th AIAA aerospace sci-
[11] Docquier N, Candel S. Combustion control and sensors: a review. Prog Energy ences meeting including the new horizons forum and aerospace exposition; 2009.
Combust Sci 2002;28:107–50. p. 223.
[12] Williams JG, Steenken WG, Yuhas AJ. Estimating engine airflow in gas-turbine [44] Cathey CD, Tang T, Shiraishi T, Urushihara T, Kuthi A, Gundersen MA.
powered aircraft with clean and distorted inlet flows; 1996. Nanosecond plasma ignition for improved performance of an internal com-
[13] McManus K, Poinsot T, Candel S. A review of active control of combustion bustion engine. IEEE Trans Plasma Sci 2007;35:1664–8.
instabilities. Prog Energy Combust Sci 1993;19:1–29. [45] Lefkowitz JK, Guo P, Ombrello T, Won SH, Stevens CA, Hoke JL, Schauer F,
[14] Demare D, Baillot F. Acoustic enhancement of combustion in lifted non- Ju Y. Schlieren imaging and pulsed detonation engine testing of ignition by a
premixed jet flames. Combust Flame 2004;139:312–28. nanosecond repetitively pulsed discharge. Combust Flame 2015;162:2496–507.
[15] Baillot F, Demare D. Physical mechanisms of a lifted nonpremixed flame sta- [46] Serbin S, Mostipanenko A, Matveev I, Tropina A. Improvement of the gas tur-
bilized in an acoustic field. Combust Sci Technol 2002;174:73–98. bine plasma assisted combustor characteristics. In: 49th AIAA aerospace sci-
[16] Gerlinger P, Stoll P, Kindler M, Schneider F, Aigner M. Numerical investigation ences meeting including the new horizons forum and aerospace exposition; 2011.
of mixing and combustion enhancement in supersonic combustors by strut p. 4–7.
induced streamwise vorticity. Aerosp Sci Technol 2008;12:159–68. [47] Lacoste D, Moeck J, Durox D, Laux C, Schuller T. Effect of nanosecond repeti-
[17] Ju Y, Sun W. Plasma assisted combustion: dynamics and chemistry. Prog En- tively pulsed discharges on the dynamics of a swirl-stabilized lean premixed
ergy Combust Sci 2015;48:21–83. flame. J Eng Gas Turbines Power 2013;135:101501.
[18] Morris N, Morgan R, Paull A, Stalker R. Silane as an ignition aid in scramjets. [48] Sun W, Uddi M, Ombrello T, Won SH, Carter C, Ju Y. Effects of non-equilib-
In: 22nd thermophysics conference; 1987. p. 1636. rium plasma discharge on counterflow diffusion flame extinction. Proc Com-
[19] Gerstein M, Choudhury P. Use of silane-methane mixtures for scramjet igni- bust Inst 2011;33:3211–18.
tion. J Propul Power 1985;1:399–402. [49] Ehn A, Petersson P, Zhu J, Li Z, Aldén M, Nilsson E, et al. Investigations
[20] Cipolat D, Bhana N. Fuelling of a compression ignition engine on ethanol with of microwave stimulation of a turbulent low-swirl flame. Proc Combust Inst
DME as ignition promoter: effect of injector configuration. Fuel Process Technol 2017;36:4121–8.
2009;90:1107–13. [50] Ehn A, Zhu J, Petersson P, Li Z, Aldén M, Fureby C, et al. Plasma assisted com-
[21] Cipolat D. Emission and combustion characteristics of a methanol/dimethyl bustion: effects of O3 on large scale turbulent combustion studied with laser
ether dual-fuelled compression ignition engine. In: Proceedings of 3rd inter- diagnostics and large Eddy simulations. Proc Combust Inst 2015;35:3487–95.
national conference on combustion technologies for a clean environment, Lis- [51] Sun W, Ju Y. Nonequilibrium plasma-assisted combustion: a review of recent
bon; 1995 Section. progress. J Plasma Fusion Res 2013;89:208–19.
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 23

[52] Ombrello T, Won SH, Ju Y, Williams S. Flame propagation enhancement by ing: evaluation no. 11 of the NASA panel for data evaluation. JPL Publication;
plasma excitation of oxygen. Part II: effects of O2(a1g). Combust Flame 1994. 94-26.
2010;157:1916–28. [86] Morrissey RJ, Schubert C. The reactions of ozone with propane and ethane.
[53] Stancu G, Simeni M, Laux C. Study of nitric oxide and carbon monoxide pro- Combust Flame 1963;7:263–8.
duction in plasma assisted combustion by quantum cascade laser absorption [87] Wegener R, Brauers T, Koppmann R, Rodríguez Bares S, Rohrer F, Tillmann R,
spectroscopy. In: 21st international symposium on plasma chemistry, Cairns, et al. Simulation chamber investigation of the reactions of ozone with short-
Australia; August 2013. p. 4–9. -chained alkenes. J Geophys Res Atmos 2007:112.
[54] Choe J, Sun W. Blowoff hysteresis, flame morphology and the effect of plasma [88] Stedman D, Niki H. Ozonolysis rates of some atmospheric gases. Environ Lett
in a swirling flow. J Phys D Appl Phys 2018;51:365201. 1973;4:303–10.
[55] Streng A. Tables of ozone properties. J Chem Eng Data 1961;6:431–6. [89] Dillemuth FJ, Skidmore DR, Schubert CC. The reaction of ozone with methane.
[56] Menzel DB. Ozone: an overview of its toxicity in man and animals. Journal of J Phys Chem 1960;64:1496–9.
Toxicology and Environmental Health 1984;13:181–204. [90] Adeniji S, Kerr J, Williams M. Rate constants for ozone–alkene reactions under
[57] Seinfeld JH. Urban air pollution: state of the science. Science atmospheric conditions. Int J Chem Kinet 1981;13:209–17.
1989;243:745–52. [91] Criegee R. Mechanism of ozonolysis. Angew Chem Int Ed 1975;14:745–52.
[58] Council NR. Rethinking the ozone problem in urban and regional air pollution. [92] Olzmann M, Kraka E, Cremer D, Gutbrod R, Andersson S. Energetics, kinet-
National Academies Press; 1992. ics, and product distributions of the reactions of ozone with ethene and 2,
[59] McClurkin J, Maier D. Half-life time of ozone as a function of air conditions 3-dimethyl-2-butene. J Phys Chem A 1997;101:9421–9.
and movement. Julius-Kühn-Archiv 2010;425:381. [93] Neeb P, Horie O, Moortgat GK. The ethene–ozone reaction in the gas phase. J
[60] Schönborn A, Hellier P, Ladommatos N, Hulteberg CP, Carlström G, Sayad P, Phys Chem A 1998;102:6778–85.
et al. 1-hexene autoignition control by prior reaction with ozone. Fuel Process [94] Anglada JM, Crehuet R, Bofill JM. The ozonolysis of ethylene: a theoretical
Technol 2016;145:90–5. study of the gas-phase reaction mechanism. Chemistry 1999;5:1809–22.
[61] Masurier J-B, Foucher F, Dayma G, Dagaut P. Ozone applied to the homoge- [95] Taatjes CA, Meloni G, Selby TM, Trevitt AJ, Osborn DL, Percival CJ, et al. Direct
neous charge compression ignition engine to control alcohol fuels combus- observation of the gas-phase Criegee intermediate (CH2 OO). J Am Chem Soc
tion. Appl Energy 2015;160:566–80. 2008;130:11883–5.
[62] Tachibana T, Hirata K, Nishida H, Osada H. Effect of ozone on combustion of [96] Welz O, Savee JD, Osborn DL, Vasu SS, Percival CJ, Shallcross DE, et al. Direct
compression ignition engines. Combust Flame 1991;85:515–19. kinetic measurements of Criegee intermediate (CH2 OO) formed by reaction of
[63] Ombrello T, Won SH, Ju Y, Williams S. Flame propagation enhancement CH2 I with O2 . Science 2012;335:204–7.
by plasma excitation of oxygen. Part I: effects of O3 . Combust Flame [97] Su Y-T, Huang Y-H, Witek HA, Lee Y-P. Infrared absorption spectrum of the
2010;157:1906–15. simplest Criegee intermediate CH2 OO. Science 2013;340:174–6.
[64] Gao X, Zhang Y, Adusumilli S, Seitzman J, Sun W, Ombrello T, et al. The effect [98] Berndt T, Herrmann H, Kurtén T. Direct probing of Criegee intermediates
of ozone addition on laminar flame speed. Combust Flame 2015;162:3914–24. from gas-phase ozonolysis using chemical ionization mass spectrometry. J Am
[65] Wang ZH, Yang L, Li B, Li ZS, Sun ZW, Aldén M, et al. Investigation of com- Chem Soc 2017;139(38):13387–92.
bustion enhancement by ozone additive in CH4 /air flames using direct lam- [99] Elsamra RM, Jalan A, Buras ZJ, Middaugh JE, Green WH. Temper-
inar burning velocity measurements and kinetic simulations. Combust Flame ature- and pressure-dependent kinetics of CH2 OO + CH3 COCH3 and
2012;159:120–9. CH2 OO + CH3 CHO: direct measurements and theoretical analysis. Int J
[66] Vu TM, Won SH, Ombrello T, Cha MS. Stability enhancement of ozone-as- Chem Kinet 2016;48:474–88.
sisted laminar premixed Bunsen flames in nitrogen co-flow. Combust Flame [100] Buras ZJ, Elsamra RM, Green WH. Direct determination of the sim-
2014;161:917–26. plest Criegee intermediate (CH2 OO) self reaction rate. J Phys Chem Lett
[67] Steinberg M, Yang RT, Hom TK, Berlad AL. Desulphurization of coal with 2014;5:2224–8.
ozone: an attempt. Fuel 1977;56:227–8. [101] Lee Y-P. Perspective: spectroscopy and kinetics of small gaseous Criegee in-
[68] Schönborn A, Hellier P, Aliev AE, Ladommatos N. Ignition control of homoge- termediates. J Chem Phys 2015;143:020901.
neous-charge compression ignition (HCCI) combustion through adaptation of [102] Anglada JM, Bofill JM, Olivella S, Solé A. Unimolecular isomerizations and oxy-
the fuel molecular structure by reaction with ozone. Fuel 2010;89:3178–84. gen atom loss in formaldehyde and acetaldehyde carbonyl oxides. A theoret-
[69] Slater SM, Rizzone MS. Simultaneous oxidation of SO2 and NO in flue gas by ical investigation. J Am Chem Soc 1996;118:4636–47.
ozone injection. Fuel 1980;59:897–9. [103] Horie O, Moortgat G. Decomposition pathways of the excited Criegee in-
[70] Sandri R. On the decomposition flame of liquid ozone-oxygen mixtures in a termediates in the ozonolysis of simple alkenes. Atmos Environ Part A
tube. Combust Flame 1958;2:348–52. 1991;25:1881–96.
[71] Streng A, Grosse A. The quenching diameter of ozone flames. Combust Flame [104] Fenske JD, Hasson AS, Paulson SE, Kuwata KT, Ho A, Houk K. The pressure
1961;5:81–6. dependence of the OH radical yield from ozone-alkene reactions. J Phys Chem
[72] Cramarossa F, Dixon-Lewis G. Ozone decomposition in relation to the prob- A 20 0 0;104:7821–33.
lem of the existence of steady-state flames. Combust Flame 1971;16:243–51. [105] Paulson SE, Fenske JD, Sen AD, Callahan TW. A novel small-ratio relative-rate
[73] Heimerl J, Coffee T. The detailed modeling of premixed, laminar steady-state technique for measuring OH formation yields from the reactions of O3 with
flames. I. Ozone. Combust Flame 1980;39:301–15. alkenes in the gas phase, and its application to the reactions of ethene and
[74] Rogg B, Linan A, Williams F. Deflagration regimes of laminar flames modeled propene. J Phys Chem A 1999;103:2050–9.
after the ozone decomposition flame. Combust flame 1986;65:79–101. [106] Rousso AC, Hansen N, Jasper AW, Ju Y. Low-temperature oxidation of ethylene
[75] Atkinson RB, Baulch DL, Cox RA, Crowley JN, Hampson RF, Hynes RG, by ozone in a jet-stirred reactor. J Phys Chem A 2018;122(43):8674–85.
Jenkin ME, Rossi MJ, Troe J. Evaluated kinetic and photochemical data for at- [107] Wang Z, Zhou J, Zhu Y, Wen Z, Liu J, Cen K. Simultaneous removal of NOx, SO2
mospheric chemistry: volume I - gas phase reactions of Ox, HOx, NOx and and Hg in nitrogen flow in a narrow reactor by ozone injection: experimental
SOx species. Atmos Chem Phys 2004;4:1461–738. results. Fuel Process Technol 2007;88:817–23.
[76] Peukert S, Sivaramakrishnan R, Michael J. High temperature shock tube stud- [108] Yoshioka Y, Sano K, Teshima K. NOx removal from diesel engine exhaust by
ies on the thermal decomposition of O3 and the reaction of dimethyl carbon- ozone injection method. J Adv Oxid Technol 2003;6:143–9.
ate with O-atoms. J Phys Chem A 2013;117:3729–38. [109] Mok YS, Lee H-J. Removal of sulfur dioxide and nitrogen oxides by us-
[77] Ibraguimova L, Smekhov G, Shatalov O. Recommended rate constants of ing ozone injection and absorption–reduction technique. Fuel Process Technol
chemical reactions in an H2-O2 gas mixture with electronically excited 2006;87:591–7.
species O2(࢞), O(D), OH(2σ ) involved, Institute of Mechanics of Lomonosov [110] Arin LM, Warneck P. Reaction of ozone with carbon monoxide. J Phys Chem
Moscow State University, 2003. 1972;76:1514–16.
[78] Howard CJ, Finlayson-Pitts BJ. Yields of HO2 in the reaction of hydrogen atoms [111] DeMore WB, Sander SP, Golden D, Hampson RF, Kurylo MJ, Howard CJ,
with ozone. J Chem Phys 1980;72:3842–3. et al. Chemical kinetics and photochemical data for use in stratospheric mod-
[79] Ju LP, Han KL, Varandas AJ. Variational transition-state theory study of the eling; 1997. Evaluation No. 12.
atmospheric reaction OH + O3 → HO2 + O2 . Int J Chem Kinet 2007;39:148–53. [112] Kogelschatz U. Atmospheric-pressure plasma technology. Plasma Phys Con-
[80] Drugman J. The oxidation of hydrocarbons by ozone at low temperatures. J trolled Fusion 2004;46:B63.
Chem Soc Trans 1906;89:939–45. [113] Simek M, Clupek M. Efficiency of ozone production by pulsed positive corona
[81] Greene CR, Atkinson R. Rate constants for the gas-phase reactions of O3 with discharge in synthetic air. J Phys D Appl Phys 2002;35:1171.
a series of alkenes at 296 ± 2 K. Int J Chem Kinet 1992;24:803–11. [114] Samaranayake W, Miyahara Y, Namihira T, Katsuki S, Sakugawa T, Hackam R,
[82] Atkinson R, Carter WP. Kinetics and mechanisms of the gas-phase reactions et al. Pulsed streamer discharge characteristics of ozone production in dry air.
of ozone with organic compounds under atmospheric conditions. Chem Rev IEEE Trans Dielectr Electr Insul 20 0 0;7:254–60.
1984;84:437–70. [115] Atkinson R, Baulch D, Cox R, Hampson R, Kerr J, Troe J. Evaluated kinetic
[83] Schubert CC, Schubert S, Pease RN. The Oxidation of lower paraffin hydro- and photochemical data for atmospheric chemistry: supplement III. Int J Chem
carbons. I. Room temperature reaction of methane, propane, n-butane and Kinet 1989;21:115–50.
isobutane with ozonized oxygen. J Am Chem Soc 1956;78:2044–8. [116] Fang Z, Qiu Y, Sun Y, Wang H, Edmund K. Experimental study on discharge
[84] Atkinson R, Baulch DL, Cox RA, Crowley JN, Hampson RF Jr, Kerr JA, characteristics and ozone generation of dielectric barrier discharge in a
et al. Summary of evaluated kinetic and photochemical data for atmospheric cylinder–cylinder reactor and a wire–cylinder reactor. J Electrostat
chemistry IUPAC subcommittee on gas kinetic data evaluation for atmospheric 2008;66:421–6.
chemistry; 2001. Web Version December 2001. [117] Takaki K, Hatanaka Y, Arima K, Mukaigawa S, Fujiwara T. Influence of elec-
[85] DeMore WB, Sander SP, Golden DM, Hampson RF, Kurylo MJ, Howard CJ, trode configuration on ozone synthesis and microdischarge property in di-
et al. Chemical kinetic and photochemical data for use in stratospheric model- electric barrier discharge reactor. Vacuum 2008;83:128–32.
24 W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25

[118] Jenei I, Kiss E. Development of the ozone generation by the variation of aux- [153] Halter F, Higelin P, Dagaut P. Experimental and detailed kinetic modeling
iliary electrodes. J Electrostat 2005;63:985–91. study of the effect of ozone on the combustion of methane. Energy Fuels
[119] Sung Y-M, Sakoda T. Optimum conditions for ozone formation in a micro di- 2011;25:2909–16.
electric barrier discharge. Surf Coat Technol 2005;197:148–53. [154] Gao X, Zhang Y, Adusumilli S, Seitzman J, Sun W, Ombrello T, et al. The effect
[120] Alsheyab MA, Muñoz AH. Optimisation of ozone production for water and of ozone addition on flame propagation 53rd AIAA aerospace sciences meeting;
wastewater treatment. Desalination 2007;217:1–7. 2015.
[121] Samoilovič VG, Gibalov VI, Kozlov KV. Physical chemistry of the barrier dis- [155] Gluckstein ME, Morrison RB, Khammash TB. Combustion with ozone-modifi-
charge. Deutscher Verlag für Schweisstechnik; 1997. cation of flame speeds C2 hydrocarbon-air mixtures. University of Michigan;
[122] Kogelschatz U, Eliasson B, Egli W. Dielectric-barrier discharges. Principle and 1955.
applications. J Phys IV 1997;7 C4-47–C4-66. [156] Pinchak M, Ombrello T, Carter C, Gutmark E, Katta V. The effect of hydrody-
[123] Eliasson B, Kogelschatz U, Baessler P. Dissociation of O2 in N2 /O2 mixtures. J namic stretch on the flame propagation enhancement of ethylene by addition
Phys B At Mol Phys 1984;17:L797. of ozone. Philos Trans R Soc A 2015;373:20140339.
[124] Kogelschatz U, Eliasson B, Egli W. From ozone generators to flat television [157] Dugger GL. Effect of initial mixture temperature on flame speed of methane-
screens: history and future potential of dielectric-barrier discharges. Pure Appl air, propane-air, and ethylene-air mixtures, NACA-TR-1061, 1952.
Chem 1999;71:1819–28. [158] Zhang Y, Zhu M, Zhang Z, Shang R, Zhang D. Ozone effect on the flammability
[125] Braun D, Pietsch G. Microdischarges in air-fed ozonizers. J Phys D Appl Phys limit and near-limit combustion of syngas/air flames with N2 , CO2 , and H2 O
1991;24:564. dilutions. Fuel 2016;186:414–21.
[126] Eliasson B, Hirth M, Kogelschatz U. Ozone synthesis from oxygen in dielectric [159] Weng W, Nilsson E, Ehn A, Zhu J, Zhou Y, Wang Z, et al. Investigation of
barrier discharges. J Phys D Appl Phys 1987;20:1421. formaldehyde enhancement by ozone addition in CH4 /air premixed flames.
[127] Yanallah K, Hadj Ziane S, Belasri A. Ozone decomposition on walls. Plasma Combust Flame 2015;162:1284–93.
Devices Oper 2006;14:215–22. [160] Ju Y, Reuter CB, Won SH. Numerical simulations of premixed cool flames of
[128] Daumont D, Brion J, Charbonnier J, Malicet J. Ozone UV spectroscopy I: ab- dimethyl ether/oxygen mixtures. Combust Flame 2015;162:3580–8.
sorption cross-sections at room temperature. J Atmos Chem 1992;15:145–55. [161] Reuter CB, Won SH, Ju Y. Experimental study of the dynamics and structure
[129] Gorshelev V, Serdyuchenko A, Weber M, Chehade W, Burrows J. High spectral of self-sustaining premixed cool flames using a counterflow burner. Combust
resolution ozone absorption cross-sections—part 1: measurements, data anal- Flame 2016;166:125–32.
ysis and comparison with previous measurements around 293 K. Atmos Meas [162] Hajilou M, Ombrello T, Won SH, Belmont E. Experimental and numerical char-
Tech 2014;7:609–24. acterization of freely propagating ozone-activated dimethyl ether cool flames.
[130] Reuter S, Winter J, Iseni S, Peters S, Schmidt-Bleker A, Dünnbier M, et al. De- Combust Flame 2017;176:326–33.
tection of ozone in a MHz argon plasma bullet jet. Plasma Sources Sci Technol [163] Won SH, Jiang B, Diévart P, Sohn CH, Ju Y. Self-sustaining n-heptane cool dif-
2012;21:034015. fusion flames activated by ozone. Proc Combust Inst 2015;35(1):881–8.
[131] Malicet J, Daumont D, Charbonnier J, Parisse C, Chakir A, Brion J. Ozone UV [164] Reuter CB, Lee M, Won SH, Ju Y. Study of the low-temperature reactivity
spectroscopy. II. Absorption cross-sections and temperature dependence. J At- of large n-alkanes through cool diffusion flame extinction. Combust Flame
mos Chem 1995;21:263–73. 2017;179:23–32.
[132] Swinehart D. The Beer-Lambert law. J Chem Educ 1962;39:333. [165] Perkin W. LVII.—some observations on the luminous incomplete combustion
[133] Serdyuchenko A, Gorshelev V, Weber M, Chehade W, Burrows JP. High spec- of ether and other organic bodies. J Chem Soc Trans 1882;41:363–7.
tral resolution ozone absorption cross-sections —part 2: temperature depen- [166] Ombrello T, Carter C, Katta V. Burner platform for sub-atmospheric pressure
dence. Atmos Meas Tech 2014;7:625–36. flame studies. Combust Flame 2012;159:2363–73.
[134] Teranishi K, Shimada Y, Shimomura N, Itoh H. Investigation of ozone con- [167] Wang H, You X, Joshi AV, Davis SG, Laskin A, Egolfopoulos F, Law CK. USC
centration measurement by visible photo absorption method. Ozone Sci Eng Mech Version II. High-Temperature Combustion Reaction Model of H2/CO/C1-
2013;35:229–39. C4 Compounds. http://ignis.usc.edu/USC_Mech_II.htm, May 2007.
[135] Liang X, Wang Z, Weng W, Zhou Z, Huang Z, Zhou J, et al. Study of [168] Gao X, Zhai J, Sun W, Ombrello T, Carter C. The effect of ozone addition
ozone-enhanced combustion in H2 /CO/N2 /air premixed flames by laminar on autoignition and flame stabilization. 54th AIAA aerospace sciences meeting;
burning velocity measurements and kinetic modeling. Int J Hydrogen Energy 2016. AIAA 2016-0960.
2013;38:1177–88. [169] Gao X, Sun W, Ombrello T, Carter C. The effect of ozonolysis activated au-
[136] Lucas D, Dunnrankin D, Hom K, Brown NJ. Ignition by excimer laser photoly- toignition on jet flame dynamics. 55th AIAA aerospace sciences meeting; 2017.
sis of ozone. Combust Flame 1987;69:171–84. AIAA 2017-1776.
[137] Liu F, Furutani H, Hama J, Takahashi S. The ignition of H2 -O2 -O3 /H2 -O2 -O3 -Ar [170] Gao X, Yang S, Wu B, Sun W. The effects of ozonolysis activated autoignition
mixture induced by the photolysis of ozone. JSME Int J Ser B 1998;41:951–8. on non-premixed jet flame dynamics: a numerical and experimental study.
[138] Nomaguchi T, Koda S. Spark ignition of methane and methanol in ozonized 53rd AIAA/SAE/ASEE joint propulsion conference; 2017.
air. Symp (Int) Combust 1989;22:1677–82. [171] Gao X, Wu B, Sun W, Ombrello T, Carter CD. The effects of ozonolysis ac-
[139] Flynn PF, Hunter GL, Loye AO, Akinyemi OC, Durrett RP, Moore GA, Muntean tivated autoignition on non-premixed jet flames. In: AIAA aerospace sciences
GG, Peters LL, Pierz PM, Wagner JA, Wright JF. Premixed charge compres- meeting; 2018. p. 1412.
sion ignition engine with optimal combustion control, U.S. Patent 6,276,334, [172] Gao X, Wu B, Sun W, Ombrello T, Carter C. Ozonolysis activated autoignition
2001. in non-premixed coflow. J Phys D Appl Phys 2019;52:105201.
[140] Nishida H, Tachibana T. Homogeneous charge compression ignition of natural [173] Gao X. The effects of ozone addition on flame propagation and stabilization.
gas/air mixture with ozone addition. J Propul Power 2006;22:151–7. Georgia Institute of Technology; 2017.
[141] Mohammadi A, Kawanabe H, Ishiyama T, Shioji M, Komada A. Study on com- [174] Sheinson RS, Williams FW. Chemiluminescence spectra from cool and blue
bustion control in natural-gas PCCI engines with ozone addition into intake flames: electronically excited formaldehyde. Combust Flame 1973;21:221–30.
gas. SAE technical paper; 2006. [175] Kojima J, Ikeda Y, Nakajima T. Spatially resolved measurement of OH∗ , CH∗ ,
[142] Yamada H, Yoshii M, Tezaki A. Chemical mechanistic analysis of additive ef- and C2 ∗ chemiluminescence in the reaction zone of laminar methane/air pre-
fects in homogeneous charge compression ignition of dimethyl ether. Proc mixed flames. Proc Combust Inst 20 0 0;28:1757–64.
Combust Inst 2005;30:2773–80. [176] Gaydon AG. The spectroscopy of flames. Chapman and Hall Ltd; 1957.
[143] Foucher F, Higelin P, Mounaїm-Rousselle C, Dagaut P. Influence of ozone [177] Deng S, Zhao P, Zhu D, Law CK. NTC-affected ignition and low-tem-
on the combustion of n-heptane in a HCCI engine. Proc Combust Inst perature flames in nonpremixed DME/air counterflow. Combust Flame
2013;34:3005–12. 2014;161:1993–7.
[144] Masurier J-B, Foucher F, Dayma G, Dagaut P. Homogeneous charge compres- [178] Toby S. Chemiluminescence in the gas-phase reaction between ozone and al-
sion ignition combustion of primary reference fuels influenced by ozone ad- lene. J Lumin 1973;8:94–6.
dition. Energy Fuels 2013;27:5495–505. [179] Larson RS. PLUG: a Fortran program for the analysis of plug flow reactors with
[145] Masurier J-B, Foucher F, Dayma G, Dagaut P. Investigation of iso-octane gas-phase and surface chemistry. Livermore, CAUnited States: Sandia Labs;
combustion in a homogeneous charge compression ignition engine seeded 1996.
by ozone, nitric oxide and nitrogen dioxide. Proc Combust Inst 2015;35: [180] Guiberti T, Durox D, Schuller T. Flame chemiluminescence from CO2 - and
3125–3132. N2 -diluted laminar CH4 /air premixed flames. Combust Flame 2017;181:110–22.
[146] Warnatz J. Calculation of structure of laminar flat flames I: flame velocity [181] Luque J, Jeffries J, Smith G, Crosley D, Walsh K, Long M, et al. CH (AX) and OH
of freely propagating ozone decomposition flames. Ber Bunsen Phys Chem. (AX) optical emission in an axisymmetric laminar diffusion flame. Combust
1978;82:193–200. Flame 20 0 0;122:172–5.
[147] Lewis B, Von Elbe G. Theory of flame propagation. Chem Rev 1937;21:347–58. [182] Kee RJ, Grcar JF, Smooke M, Miller J, Meeks E. A Fortran program for modeling
[148] Hirschfelder JO, Curtiss CF, Campbell DE. The theory of flame propagation. J steady laminar one-dimensional premixed flames. Sandia Laboratories Report
Phys Chem 1953;57:403–14. No. SAND 85-8240, 1985.
[149] Von Karman T, Penner S. Selected combustion problems. AGARD 1954;II:167. [183] Gao X, Wu B, Sun W, Ombrello T, Carter C. The effects of ozonolysis acti-
[150] Wilde KA. Boundary-value solutions of one-dimensional laminar flame prop- vated autoignition on non-premixed jet flames. 55th AIAA aerospace sciences
agation equations. Combust Flame 1972;18:43. meeting. Gaylord Palms, Kissimmee, Florida; 2018.
[151] Lewis B, von Elbe G. Determination of the speed of flames and the tempera- [184] Gao X, Zhai J, Sun W, Ombrello T, Carter C. The effect of ozone addition on
ture distribution in a spherical bomb from time-pressure explosion records. J autoignition and flame stabilization. In: 54th AIAA aerospace sciences meeting;
Chem Phys 1934;2:283–90. 2016. p. 0960.
[152] Streng AG, Grosse AV. The pure ozone to oxygen flame. J Am Chem Soc [185] Razumovskii S, Zaikov GE. Ozone and its reactions with organic compounds.
1957;79:1517–18. Elsevier; 1984.
W. Sun, X. Gao and B. Wu et al. / Progress in Energy and Combustion Science 73 (2019) 1–25 25

Mr. Bin Wu received his B.E. degree from University of


[186] Sharypov V, Kuznetsov B, Baryshnikov S, Beregovtsova N, Selyutin G, Chu-
Science and Technology of China in 2014, and is currently
makov V, et al. Some features of chemical composition, structure and re-
a Ph.D. candidate at Georgia Institute of Technology. He
active ability of Kansk-Achinsk lignite modified by ozone treatment. Fuel
has been working on the ozone-assisted combustion since
1999;78:663–6.
2016, focusing on ozone related chemistry.
[187] Gao Y, Külaots I, Chen X, Aggarwal R, Mehta A, Suuberg E, et al. Ozonation for
the chemical modification of carbon surfaces in fly ash. Fuel 2001;80:765–8.
[188] Patrakov YF, Fedyaeva O, Semenova S, Fedorova N, Gorbunova L. Influence
of ozone treatment on change of structural–chemical parameters of coal
vitrinites and their reactivity during the thermal liquefaction process. Fuel
2006;85:1264–72.
[189] Soriano NU, Migo VP, Matsumura M. Ozonized vegetable oil as pour point
depressant for neat biodiesel. Fuel 2006;85:25–31.
[190] Wilk M, Magdziarz A. Ozone effects on the emission of pollutants coming
from natural gas combustion. Polish J Environ Stud 2010;19:1331–6. Timothy Ombrello earned his Ph.D. in Mechanical and
[191] Jarvis JB, Day A, Suchak N. LoTOxTM process flexibility and multi-pollutant Aerospace Engineering from Princeton University in 2009
control capability. In: Proc combined power plant air pollution control mega where he worked towards developing an understanding
symposium, Washington, USA; 2003. p. 19–22. of the kinetic effects of long lifetime plasma-produced
[192] Sun W-Y, Ding S-L, Zeng S-S, Su S-J, Jiang W-J. Simultaneous absorption of species. Upon completing his graduate degree, he went
NOx and SO2 from flue gas with pyrolusite slurry combined with gas-phase on to work at the Air Force Research Laboratory, ini-
oxidation of NO using ozone. J Hazard Mater 2011;192:124–30. tially as a National Research Council Research Associate
[193] Li H-P, Ostrikov KK, Sun W. The energy tree: non-equilibrium energy transfer for one year before joining civil service as a Research
in collision-dominated plasmas. Phys Rep 2018;770-772:1–45. Aerospace Engineer in the Aerospace Systems Directorate.
He works predominately on research related to high-
Dr. Wenting Sun is currently an assistant professor in speed air-breathing propulsions systems, specifically su-
the School of Aerospace Engineering, Georgia Institute of personic combustion ramjets, covering a range of fluid dy-
Technology. He received his B.E. and M.E. degrees from namic, combustion, and diagnostic challenges. His inter-
Tsinghua University in 2005 and 2007, respectively and ests lie in performing research and crafting techniques to enhance reactivity for
received his Ph.D. degree from Princeton University in more rapid ignition and more robust flame propagation and stabilization, from fun-
2013. He was selected for the U.S. AFOSR Young Inves- damental bench-top to supersonic wind tunnel experiments. For his contributions
tigator Program in 2016 to study ozone chemistry and thus far, he has been given the Early Career Award from the Air Force Research
the effect of ozone on combustion dynamics. He has been Laboratory, and the Presidential Early Career Award for Scientists and Engineers
working in the field of plasma-assisted combustion since (PECASE) from the White House.
2007. His work on plasma/ozone assisted combustion is
to induce plasma generated species into combustion sys-
tem to enable combustion at extreme conditions.

Dr. Xiang Gao received his B.E. degree from Beijing Uni-
versity of Aeronautics and Astronautics in 2012, and his
Ph.D. degree from Georgia Institute of Technology in 2017.
He has been working on the ozone-assisted combustion
since 2014, covering the kinetic and dynamic parts. He
is currently a Data & Applied Scientist at Microsoft Re-
search.

You might also like