Effects of Non-Equilibrium Plasma Discharge on Ignition and NTC

Chemistry of DME/O2/Ar Mixtures: A Numerical Investigation

Yao Zhang1, Suo Yang2, Vigor Yang3, and Wenting Sun4
School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA 30332

The effects of non-equilibrium plasma discharge on ignition characteristics and low-

temperature chemistry (LTC) of DME/O2/Ar mixture are numerically investigated through
a self-consistent simulation in a plane-to-plan geometry at reduced pressures of 76 Torr. One-
dimensional, nano-second plasma discharge model is used the first time to study the two-stage
ignition process and negative-temperature coefficient (NTC) behavior of DME mixtures. This
kinetic mechanism consists total number of 69 species, and 414 reactions, which is a
combination of plasma kinetic model and DME combustion kinetic model. The initial
temperatures are set to be 550K and 800K. With initial temperature of 550K, the negative
temperature coefficient (NTC) regime where LTC dominates, the ignition delay time for
plasma-assisted case is improved by ~250 times for the first-stage, and 15 times for the second-
stage or overall. With the initial temperature of 800K, the intermediate temperature regime,
the overall ignition delay time is shorten by ~75 times with plasma addition. Moreover, the
results suggest that, plasma not only enhance DME ignition characteristic time dramatically
by orders of magnitudes, but also alters reaction pathway and makes the disappeared two-
stage ignition behavior for 800K case reappeared. In addition, for 800K case, the intermediate
temperature regime, the enhancements bring by plasma addition on ignition delay time and
LTC are non-linear; there exists a range of optimized pulse number, 35-40 pulses in this work,
where accumulative energy input brings the most efficient enhancement.

I. Introduction
Dimethyl ether (DME), as a clean-burning, non-toxic, and potential diesel alternative for compression-ignition
engines, has drawn widely attention in recent years. Due to its lack of carbon-to-carbon bonds, DME produces very
low levels of soot emissions. Moreover, DME has a very high cetane number, which is a measure of the fuel’s
ignitibility in compression ignition engine; and exhibits a remarkable two-stage ignition mode due to the negative
temperature coefficient (NTC) effect, means that an increasing in temperature will result a decreasing in reactivity in
a specific temperature range[1]. Thus, the ignition property of DME-containing mixtures is an important problem in
applied combustion research.
Over the past decade, many studies have shown that the non-equilibrium plasma [2, 3] and addition of plasma-
generated species such as ozone [4-6] have a strong influence on the combustion process, which can help control
ignition and stabilize flame in engines [7-10]. Because the non-equilibrium plasma generates radials directly from
discharges, which has weak thermal dependence, several recent studies suggest that the non-equilibrium plasma
discharge can enhance the low-temperature chain branching and reduce the NTC behavior. Results from Filimonova
[11] showed that intensive plasma discharge stimulates the development and increase the intensity of cool flames for
propane-air mixtures, and greatly reduces the cool flame ignition delay times. Nagaraja et al. [12] studied the effect
of non-equilibrium plasma on the two-stage ignition of n-heptane, and concluded that plasma discharge reduces the
initiation time of the first-stage temperature rise of n-heptane. Although research efforts were being done for both
non-equilibrium plasma-assisted combustion, and for the ignition properties of DME, there are very few works about
the coupling effects between them. The experimental work from Sun et al. [13] studied the effects of non-equilibrium
plasma on the ignition and extinction of DME/O 2/He diffusion flame, which observed a strong low temperature
reactivity during ignition process. However, there are not many detailed numerical studies of non-equilibrium plasma
discharge on the ignition/extinction and NTC behavior of DME-containing mixtures. In addition to experimental
investigation, numerical simulations are needed to complement the experimental efforts and provide more detailed
insights into the plasma enhancement of the combustion process.

1
Ph.D. Candidate, yzhang401@gatech.edu, Student Member AIAA.
2
Ph.D. Student Member AIAA.
3
William R. T. Oakes Professor and Chair, Fellow AIAA.
4
Assistant Professor, Member AIAA.
 The recent work from Gururajan and Egolfopoulos [14] shows the transient plasma effects on the auto-ignition of
DME/O2/Ar by using a zero-dimensional model, and the conclusion suggested that, plasma addition can greatly reduce
the first-stage ignition time, however this conclusion is segmental. On the other hand, according to the work from
Yang et al. [3], after detailed comparison between numerical results from 1D model and 0D model of plasma-assisted
ignition, it shows that 1D model has a much better agreement with experimental results. In the 0D kinetic models,
plasma discharges are set to be uniform for the entire domain during each voltage pulse, without take into consideration
of cathode sheath formation, accumulation on the dielectric layers, and species/heat diffusion effects. In addition, the
reduced electric field (E/N) and electron density charge are also pre-specified, all these together create inaccuracies
which mismatched the experimental results.
In this work, a detailed numerical study on the effect of non-equilibrium plasma discharge on ignition and NTC
behavior of DME/O2/Ar mixtures is presented by using a one-dimensional model. A self-consistent, one-dimensional
numerical framework that is adopted from Nagaraja et al. [15] is used to simulate pulsed, nano-second discharges,
and calculated species and temperature evolutions. This model was well verified from previous works [3, 16, 17], and
was capable of resolving electric field transients during each nanosecond discharge pulse, as well as calculating the
cumulative effects of multiple pulses on fuel oxidation and combustion. Detailed analysis is conducted to numerically
investigate the effect of non-equilibrium plasma discharge on the ignition behavior and low-temperature chemistry
(LTC) of DME/O2/Ar mixtures.

II. Theoretical Framework

The based numerical model used for simulations of ignition of preheated DME/O2/Ar is adopted from Nagaraja et
al. [15] with some code modifications to fit conditions of the present work. Governing equations include the equation
of electric potential, equation of electron energy, transport equations of both charged and neutral species, and
conservation equations for mass, momentum, and energy of the gas mixture, and are solved simultaneously to model
the flow motions. The electron transport and reaction rate coefficients are expressed as functions of electron energy
by BOLSIG [18], and are updated through interpolation for every time-step. The plasma drift-diffusion fluid model
with the ‘local electron mean energy approximation’ is applied for the transport of plasma species and energy[19].

A. Physical configuration
The physical configuration of the plan-to-plan dielectric barrier discharge is shown in Fig. 1 with the 1D simulation
domain being marked as a dashed line. In this work, the computational domain is set to be 14 mm long from the left
electrode to the right electrode, with the right electrode connecting to a high voltage power supply, and the left
electrode being grounded. Each electrode is covered with a 1.6 mm quartz layer, along with 0.8 mm thickness of
dielectric layer of Kalrez. DME/O2/Ar mixture is filled in the gap between dielectric. The dielectric constants are fixed
for all simulations in the present work with the constant of quartz (εg) being 4.8, and the constant of Kalrez (εs) is 3.2.
Gaussian voltage pulses are applied with peak voltage of 12.5 kV and pulse duration of 60 ns. The repetition frequency
is fixed to be 30 kHz for the simulation.

Fig. 1 Schematic of physical configuration of the simulation

B. Numerical method
A non-uniform mesh consisting of 600 grid points is used. Only a brief summary of the numerical methods is
presented here, more details can be found in [3, 15, 16, 20]. The second order Strang type splitting scheme is used to
treat the convection-diffusion term and the kinetics source term separately [15]. Techniques of point-implicit stiff
ODE solver (ODEPIM), and correlated dynamic adaptive chemistry and transport (CoDACT) are applied to accelerate
the calculation of chemical source terms and the evaluation of transport properties [20, 21]. Due to the nature of
various physical phenomena and large timescale difference, an adaptive time-step approach is utilized to tackle the
large disparity in timescales. In order to capture the electric field transients and the electron-energy relaxation process,
the time-step during each discharge pulse is in the order of magnitude of picosecond (10-13 s - 10-12 s). In between
pulses, electric field is set to be zero, thus the order of magnitude of the time-step is adjusted to the nanosecond level.
After all plasma reactions are finished, an even larger time-step (10-7-10-6 s) is applied to only capture the chemical
reactions. Implicit time integration is applied for stiff chemical source term, while species and flow transports are
treated explicitly. To further improve computational efficiency, parallel computing over multiple processors is applied
with domain decomposition using message-passing interface (MPI).

C. Initial and boundary conditions

The simulation conditions for present work are set at constant pressure of 76 Torr, with two initial temperature
conditions of 550K and 800K. The mass fraction ratio of DME/O2/Ar is 6.6%/19.6%/73.8%.
A zero flux boundary condition is imposed for mass and momentum conservation equations, as well as neutral
species at the solid walls. The boundary flux conditions for electron, positive and negative ions, and electron energy
are calculated based on the sum of drift components due to electric field and thermal velocity components, with the
electron flux at the cathode boundary has contribution from secondary electron emission. In order to calculate the
boundary temperature, semi-infinite solid with constant heat flux is applied to the dielectric layer.

D. DME/O2/Ar plasma chemistry mechanism

The kinetic mechanism is generated by the combination of a plasma kinetic model and a DME combustion kinetic
model [22-24], with Ar and O2 plasma reactions taken from ZDPlasKin database [25]. This mechanism consists total
number of 69 species, and 414 reactions, including charged species: Ar+, O2+, O+, CH3+, CH3OCH2+, CH3OCH3+, O2-,
O-, and e; exited species: Ar*, O(D), O2 (a), O2(b), O2(c); and all other neutral species. Electron impact on electronic
excitation, dissociation, ionization, and dissociative attachment processes along with quenching, detachment, and
recombination reactions are considered. The reaction rate constants of electron impact related reactions are obtained
from solving Boltzmann equation with two-term expansion using the BOLSIG software [18]. Note that since there are
no cross-section data available for electron impact of DME, the cross-sections of reactions are approximated by
electron impact reactions of ethane based on the similarity in carbon number, single C-C/C-O/C-H bounds, and radical
formation. [13].

III. Results and Discussion

A. Electrical characteristics
All electrical properties depend on the reduced electric field E/N, where E is the electric field, and N is the number
density of neutral gas particles. Figure 2 shows time evolution of various electrical characteristics at the discharge
center for the 1st pulse and 150th pulse.
As shown in Fig. 2(a), for the first pulse applied, the breakdown happens at E/N value of ~400 Td and rapidly
drops down to ~5 Td within a few nanoseconds, and this causes electron density to increase. It can be observed that
the rapidly fall of E/N happens at ~15ns, and electron number density has a sharp rise at about the same time as shown
in Fig. 3(a). Compare with the 1st pulse, the 150th pulse is at the stage of quasi-equilibrium, which also shows a similar
manner but with an even higher E/N peak value. Note that the ~400 Td E/N value is strong enough to ionize the gases
in air, and the discharge can self-propagate in a self-sustained manner. In Fig. 2(b), the total energy input per pulse
reaches ~0.14 mJ for both the 1st pulse and the 150th pulse. At the time of ~15ns, where the rapid fall of E/N happens,
the input energy value is ~0.105 mJ, which suggests that the breakdown accounts for most part of the total energy
(~75%).
Fig. 2 (a) Time evolution of reduced electric field, applied voltage, and current during the 1st and the 150th pulse; (b)
time evolution of reduced electric field, E/N, and the energy input during the 1st and the 150 pulse.

Both temporal evolution and spatial distribution of electron number density are shown in Fig. 3. The electron
number density for temporal evolution in Fig. 3(a) is taken at the center of simulation domain. As discussed previously,
the peak of the electron number density happens at the rapid fall of the reduced electric field. The electric field
reduction is caused by charge separation which terminates electron impact ionization, and results a high level of
electron number density (~1012 cm-3), and is almost constant for each pulse. This large number of electron number
density will cause electron-impact dissociation and excited species, thus will enhance heat and radical production. The
spatial distribution of electron number density is shown in Fig. 3(b) for different pulses. All pulse numbers show a
similar trend, where the maximum electron density happens at two sides close to the electrodes, and a relatively
uniform density is in the middle part, which is the domain for uniform heat release and radical productions.
Fig. 3 (a) Time evolution of the electron number density at the center of the simulation domain; (b) spatial distribution
of electron number density for different pulses.

B. Temperature Analysis
The spatial distribution for initial temperature of 800K is shown in Fig. 4(a), and time evolution of temperature in
the mid-point of the simulation domain is shown in Fig. 4(b). Note that spatial distributions for both 550K and 800K
follow the same trend. From the spatial distributions of the temperature shown in Fig. 4(a), it is easy to observe that
more pulses will lead to a higher temperature, and the uniformity happens at the center of the domain. Temperature
rises rapidly at the near wall region due to the ion Joule heating during discharge pulse.
 (a)

(b)

Fig. 4 (a) Spatial distribution of gas temperature; (b) comparison of gas temperature evolutions for plasma-assisted
ignition and auto-ignition
Auto-ignition temperature profiles are also conducted from CHEMKIN-II PRO [26] to compare with plasma-
assisted ignition. As shown in Fig. 5(b), with initial temperature of 550K, both plasma-assisted ignition and auto-
ignition cases present the two-stage ignition property, which is an indication of NTC. The first-stage ignition occurs
at ~5.25 ms for the plasma-assisted case, and occurs at ~1.27 sec for the auto-ignition case, thus there is a ~250 times
improvement when adding plasma to the system. For the second-stage or overall ignition, the ignition delay is ~0.57
sec for plasma-assisted case, and is ~8.71 sec for auto-ignition case, which shows an improvement of ~15 times with
plasma addition. With the initial temperature of 550K, the LTC pathways present in both cases with and without
plasma addition, and the present of plasma has shortened the characteristic times of both the first- and the second-
stage ignition, with a significantly larger enhancement for the first-stage ignition. The reason is that plasma brings
kinetics enhancement for the first-stage ignition, and acts like the “catalyst” [12]: electron and ion impact dissociation
reactions help to produce exited species (Ar*, O(1D), etc.) and active radicals (e.g. O, H, and OH), which will trigger
the self-accelerating LTC feedback loop, and thus with a few pulses plasma burst, the first-stage ignition delay time
is greatly enhanced.
For initial temperature of 800K, as shown in Fig. 5(b), the ignition occurs at ~14.9 sec for auto-ignition case, and
at ~0.2 sec for plasma-assisted case. Ignition delay is improved by ~75 times with the present of non-equilibrium
plasma. More importantly, there is no two-stage ignition phenomenon for auto-ignition since the intermediate
temperature regime (Tinit = 800K) is too high for NTC; however, with plasma addition, the two-stage ignition
reappeared. Moreover, comparing with the case of initial temperate of 550K, which is in the low-temperature NTC
regime, plasma addition brings more significant change to the ignition process for the initial temperature of 800K. In
order to better understand this interesting two-stage temperature reappearance for, some key species for ignition and
LTC are investigated.

C. Key Species Analysis for intermediate temperature regime (Tinit = 800K)

Figure 5 shows the time evolution of key species’ number density over the simulation time. Both atomic O, and
atomic H follow the similar trend. The steep increase in number densities of atomic O and atomic H occurs at the
same time as the rapid increase in electron density (Fig. 3a), which indicates that electron impact dissociations during
the discharge pulses cause productions of atomic O and atomic H, and later react with hydrocarbon radicals to form
OH radicals, and lead to ignition. The number density of the key radical for first stage ignition of cool flame, HO 2,
gradually increases over period with plasma pulses applied, and triggers the first-stage ignition. Laster on, HO2 radicals
turn into H2O2, and eventually to OH radicals through chain branching reactions, which lead to the second-stage
ignition. Overall ignition mechanisms for the two-stage ignition can be briefly concluded as HO2  H2O2  2OH.

Fig. 5 Time evolution of number densities of key species for ignition

Since the first-stage ignition is directly related to LTC pathways, and the one of the most important characteristic
radical formed through LTC is RO2 (CH3OCH2O2), the number density of RO2 over the whole simulation time for all
cases is investigated and shown in Fig. 6. When comparing time evolution of HO2 (Fig. 5) and time evolution of RO2
(Fig. 6), it is interesting to notice that the number density of RO2 rises up before HO2, thus the reappearance of the
two-stage ignition, and shorten of ignition delay time are related to the evolution of RO 2 number density.
 Fig.6 Time evolution of RO2 number density
The DME LTC reaction controls the production of RO2 is CH3OCH2 + O2 = CH3OCH2O2[10]. To understand the
reappearance of the two-stage ignition property with plasma addition, the production rate (𝜔̇ ) of CH3OCH2O2 (RO2)
is integrated over the time domain of initial temperature (550K or 800K) to 1000K for each case, and results are shown
in the following (Table 1). The production of RO2 from 800K auto-ignition case is about four orders of magnitude
smaller than 800K plasma-assisted case, which explains the loss of first-stage ignition from 800K auto-ignition
temperature profile in Fig. 7. However, when comparing 800K plasma-assisted ignition case and 550K auto-ignition
case, their orders of magnitude of production of RO2 are comparable, which indicates that adding plasma to the system
alters the reaction pathways, and re-establishes the missing LTC pathways for initial temperature of 800K.
Table 1. Comparison of RO2 production from different cases

𝑡(1000𝐾)
∫𝑡(𝑇 𝜔̇ 𝑑𝑡 (molecules/cm3)
Initial Temperature 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 )

Plasma Assisted Auto-ignition

550K 3.42E-3 1.73E-4
800K 5.19E-4 9.78E-8

D. Pathways Analysis for intermediate temperature regime (T init = 800K)

Pathway analysis is a strong tool to investigate the complicated effects of chemical kinetics in combustion systems.
It is well known that DME oxidization processes begin with H-abstraction reaction by removing an H atom from each
fuel specie, CH3OCH3, and produce R radicals, CH3OCH2 (RXN. 1). Then R radicals either decompose to produce
smaller hydrocarbon fragments (RXN.2) or combines with O 2 to produce alkylperoxy radical, RO2 (RXN.3). The
competition between decomposition and RO2 production will strongly influence the first-stage ignition phenomena.
CH3OCH3 + X  CH3OCH2 +HX, X = O, H, OH, HO2… (RXN.1)
CH3OCH2  CH2O + CH3 (RXN.2)
CH3OCH2 + O2  CH3OCH2O2 (RXN.3)
For the auto-ignition with initial temperature of 800K, the amount of R radicals produced from all H-abstraction
reactions (RXN.1) are within the same order of magnitude as the amount of R radicals consumed by decomposition
(RXN.2). On the other hand, both of the R radical production rate (RXN.1) and the R radical decomposition rate
(RXN.2) are orders of magnitude larger than RO2 formation rate (RXN.3). Therefore, all R radicals are quickly
decomposed through high temperature pathway (RXN.2) after formation.
 However, with plasma addition, the overall H-abstraction rate of 800K is one order of magnitude higher than the
800K auto-ignition. When adding non-equilibrium plasma discharges, electron impact dissociation generates radicals
and excited species (O, O(1D), Ar*, etc.), which significantly help on production of R radicals through H-abstraction
(RXN.1). Furthermore, energetic electrons also bring an additional reaction plasma pathway (RXN.4) to form R
radicals.
e + CH3OCH3  CH3OCH2 + H + e. (RXN.4)
The overall amount of R radical production from plasma-assisted ignition is ~35 times higher compare with the
auto-ignition, and more importantly ~5 times higher than the R radical consumption through decomposition reaction
(RXN.2). The high temperature decomposition reaction rate (RXN.2) is only twice of LTC or RO 2 formation rate
(RXN.3), thus due to the decomposition rate limitation, the excess amount of R radicals produced cannot be
completely decomposed via RXN.2 pathway, and approximately 30% of the leftover R radicals are forced to go to
through LTC pathways (RXN.3)

Fig. 7 Reaction pathway comparison diagram for DME with and without plasma addition

In general, as shown in Fig. 7, non-equilibrium plasma discharge provides new sources of radical production via
energetic electrons (RXN.4), as well as introduces new H-abstraction reaction pathways via electron-impact
dissociation reactions (RXN.1), all there help to produce large surplus of R radicals, which cannot be consumed totally
through high temperature decomposition (RXN.2). Combining all the effects together, the lost two-stage ignition
behavior from 800K auto-ignition is reinstated with plasma addition.

E. Optimization of Accumulative Energy Input for intermediate temperature regime (T init = 800K)
As shown in Fig. 4(b) from Section B, the missing of two-stage ignition from auto-ignition case recurred after
adding plasma discharges to the system. Also with the same number of 150 pulses applied, the overall or second-stage
ignition delay time is improved ~75 times for initial temperature of 800K, which is larger comparing with ~15 times
improvement for initial temperature of 550K. This larger enhancement from 800K case is related to the LTC pathways
re-establishment, including O2 additions to R and to QOOH, as well as the isomerization and decomposition of
O2QOOH, all of which helped to produce OH radicals, therefore enhance heat release and ignition. The question then
arises: what is the relationship between accumulative energy input to the system and its level of enhancement on LTC
pathway as well as ignition delay time in intermediate temperature regime? In order to answer this, different cases
with various accumulative input energy to the system are investigated.
With initial temperature of 800K, the production from both high temperature decomposition (RXN.2) and low
temperature oxidation (RXN.3) with respect to different accumulative input energy is shown in Fig. 8. Note that the
accumulative energy input is directly proportional to pulses applied to the system, with each pulse supplies ~0.14mJ
energy (Fig. 2b).
As shown in Fig. 8, the low-temperature oxidation production curve initially shows a linear growth, and then
followed by a logarithmic growth. The high-temperature decomposition production curve shows a power-law growth
as more pulses applied to the system. This suggests that, adding plasma discharges to the system can always help to
re-establish the LTC pathways. Detailed reasons are explained in Section D, which can be summarized as the following:
plasma introduces new sources of radicals/excited species to bring ultra-fast H-abstraction reactions, as well as an
additional H-abstraction reaction via energetic electrons, all these help to produce an excess amount of R radicals,
which cannot be completely decomposed through high-temperature decomposition pathways, thus part of the R
radicals are forces to go through the LTC pathways. Note that from Fig. 8, as more pulses applied, the low temperature
production curve is flatten out but high temperature production curve keeps increasing, this is because as more and
more pulses applied to the system, thermal effect from pulses eventually became dominate.

Fig. 8 Comparison of LTC and HTC production as a function of accumulative energy input

To investigate the ignition delay enhancement, the parameter of normalized characteristic ignition delay time 𝜏 ∗ is
𝜏𝑝𝑙𝑎𝑠𝑚𝑎
defined: 𝜏 ∗ = . Figure 9 shows the results of 𝜏 ∗ with respect to accumulative energy input. The curve appears
𝜏𝑎𝑢𝑡𝑜
to be exponential decay. In this work, within ~35 pulses applied, the normalized ignition delay time decreased to ~20%
of its initial value. This suggests that, there is a saturation point where additional accumulative energy input will no
longer help much on reducing the ignition delay. This saturation phenomenon is even more obvious from the derivative
curve (dτ/dN), where there is a clear trend that the curve decreases very rapidly with the first ~15 pulses, then level
off to become asymptotic towards the x-axis, and with less than ~50 pulses, the derivative curve reaches 0.

𝑑𝜏∗
Fig. 9 𝜏 ∗ , as a function of accumulative energy input
𝑑𝑁

From all above analysis, in the intermediate temperature regime, the capacity of plasma-assisted ignition or re-
active of LTC chain branching of a system under a certain condition are limited and non-linear. In this work, with
initial temperature of 800K, if considering energy input to be ~10% of the total heat release of the system and the
asymptotic behavior, the optimized pulse number falls in the range of 35-40 pulses.

IV. Conclusions
Plasma-assisted ignition characteristics and LTC of DME/O2/Ar mixture are investigated numerically for two
initial temperature conditions of 550K and 800K. The results show that non-equilibrium plasma discharges can greatly
reduce the ignition delays for both cases.
With initial temperature of 550K, which is in the NTC regime with LTC dominates, when applying 150 pulses to
the system, the ignition characteristic time is shorten by ~250 times for the first-stage ignition, and ~15 times for the
second-stage or overall ignition. Due to the reason that plasma brings mainly kinetic enhancement for the first-stage
ignition by initiating a self-accelerating LTC feedback loop, but mainly thermal enhancement after the first-stage, the
first-stage ignition enhancement is more significant in this temperature regime, and it requires a large amount of energy
input to improve the second-stage or overall ignition delay time.
With initial temperature of 800K, which is in the intermediate temperature regime, plasma not only accelerated
the overall ignition by ~75 times, but also alters the reaction pathways and brings the missing two-stage ignition
phenomenon back from the auto-ignition. Plasma introduces additional electron impact dissociation processes for H-
abstraction reaction, and generates active radicals/excited species to produce orders of magnitude larger amount of R
radicals than auto-ignition. The significant amount of excess R radicals cannot be completely consumed through high
temperature decomposition and are forced to proceed through LTC pathways; hence, the two-stage ignition reappeared.
Moreover, the enhancements on ignition delay time and LTC from plasma addition are non-linear; there exists a range
of optimized pulse number, 35-40 pulses in this work, where accumulative energy input brings the most efficient
enhancement.
In general, whether brings enhancement on ignition delays in the low temperature NTC regime, especially for the
first-stage ignition, or brings new H-abstraction reaction pathways to alter fuel kinetics and re-active the LTC
pathways in the intermediate temperature regime, with the addition of non-equilibrium plasma, ignition delays of
DME/O2/Ar mixture are dramatically reduced.

Acknowledgement
This work was sponsored by the William R.T Oakes Endowment of the Georgia Institute of Technology.
 References
[1] Compton, R., Hancock, G., and Pilling, M., "Comprehensive Chemical Kinetics: Low-Temperature Combustion
and Autoignition." Elsevier Science, Amsterdam, 1997.
[2] Sun, W., and Ju, Y., "Nonequilibrium Plasma-Assisted Combustion: A Review of Recent Progress," J. Plasma
Fusion Res, Vol. 89, No. 4, 2013, pp. 208-219.
[3] Yang, S., Gao, X., Yang, V., Sun, W., Nagaraja, S., Lefkowitz, J. K., and Ju, Y., "Nanosecond Pulsed Plasma
Activated C2H4/O2/Ar Mixtures in a Flow Reactor," Journal of Propulsion and Power, Vol. 32, No. 5, 2016, pp.
1240-1252. doi: 10.2514/1.B36060
[4] Gao, X., Zhang, Y., Adusumilli, S., Seitzman, J., Sun, W., Ombrello, T., and Carter, C., "The effect of ozone
addition on laminar flame speed," Combustion and Flame, Vol. 162, No. 10, 2015, pp. 3914-3924. doi:
10.1016/j.combustflame.2015.07.028
[5] Gao, X., Zhang, Y., Adusumilli, S., Seitzman, J., Sun, W., Ombrello, T., and Carter, C., "The Effect of Ozone
Addition on Flame Propagation," 53rd AIAA Aerospace Sciences Meeting, 2015.
[6] Gao, X., Zhai, J., Sun, W., Ombrello, T., and Carter, C., "The Effect of Ozone Addition on Autoignition and Flame
Stabilization," 54th AIAA Aerospace Sciences Meeting, 2016, pp. AIAA 2016-0960.
[7] Starikovskaia, S., "Plasma assisted ignition and combustion," Journal of Physics D: Applied Physics, Vol. 39, No.
16, 2006, p. R265.
[8] Starikovskiy, A., and Aleksandrov, N., "Plasma-assisted ignition and combustion," Progress in Energy and
Combustion Science, Vol. 39, No. 1, 2013, pp. 61-110.
[9] Ju, Y., and Sun, W., "Plasma assisted combustion: Dynamics and chemistry," Progress in Energy and Combustion
Science, Vol. 48, 2015, pp. 21-83. doi: 10.1016/j.pecs.2014.12.002
[10] Ju, Y., Lefkowitz, J. K., Reuter, C. B., Won, S. H., Yang, X., Yang, S., Sun, W., Jiang, Z., and Chen, Q., "Plasma
Assisted Low Temperature Combustion," Plasma Chemistry and Plasma Processing, Vol. 36, No. 1, 2016, pp. 85–
105. doi: 10.1007/s11090-015-9657-2
[11] Filimonova, E., "Discharge effect on the negative temperature coefficient behaviour and multistage ignition in
C3H8-air mixture," Journal of Physics D: Applied Physics, Vol. 48, No. 1, 2014, p. 015201.
[12] Nagaraja, S., Sun, W., and Yang, V., "Effect of non-equilibrium plasma on two-stage ignition of n-heptane,"
Proceedings of the Combustion Institute, Vol. 35, No. 3, 2015, pp. 3497-3504.
[13] Sun, W., Won, S. H., and Ju, Y., "In situ plasma activated low temperature chemistry and the S-curve transition
in DME/oxygen/helium mixture," Combustion and Flame, Vol. 161, No. 8, 2014, pp. 2054-2063.
[14] Gururajan, V., and Egolfopoulos, F. N., "Transient plasma effects on the autoignition of DME/O 2/Ar and C 3 H
8/O 2/Ar mixtures," Proceedings of the Combustion Institute, 2016.
[15] Nagaraja, S., Yang, V., and Adamovich, I., "Multi-scale modelling of pulsed nanosecond dielectric barrier plasma
discharges in plane-to-plane geometry," Journal of Physics D: Applied Physics, Vol. 46, No. 15, 2013, p. 155205. doi:
10.1088/0022-3727/46/15/155205
[16] Nagaraja, S., Yang, V., Yin, Z., and Adamovich, I., "Ignition of hydrogen–air mixtures using pulsed nanosecond
dielectric barrier plasma discharges in plane-to-plane geometry," Combustion and Flame, Vol. 161, No. 4, 2014, pp.
1026-1037. doi: 10.1016/j.combustflame.2013.10.007
[17] Rousso, A., Yang, S., Lefkowitz, J., Sun, W., and Ju, Y., "Low Temperature Oxidation and Pyrolysis of n-Heptane
in Nanosecond-pulsed Plasma Discharges," Proceedings of the Combustion Institute, Vol. 36, No. 3, 2017, pp. 4105–
4112. doi: 10.1016/j.proci.2016.08.084
[18] Hagelaar, G., and Pitchford, L., "Solving the Boltzmann equation to obtain electron transport coefficients and
rate coefficients for fluid models," Plasma Sources Science and Technology, Vol. 14, No. 4, 2005, pp. 722-733.
[19] Kulikovsky, A., "A more accurate Scharfetter-Gummel algorithm of electron transport for semiconductor and
gas discharge simulation," Journal of computational physics, Vol. 119, No. 1, 1995, pp. 149-155.
[20] Yang, S., Yang, V., Sun, W., Nagaraja, S., Sun, W., Ju, Y., and Gou, X., "Parallel On-the-fly Adaptive Kinetics
for Non-equilibrium Plasma Discharges of C2H4/O2/Ar Mixture," 54th AIAA Aerospace Sciences Meeting, AIAA
Paper 2016-0195, Jan. 2016. doi: 10.2514/6.2016-0195
[21] Yang, S., "Effects of Detailed Finite Rate Chemistry in Turbulent Combustion," Aerospace Engineering. Vol.
Ph.D., Georgia Institute of Technology, 2017, pp. 1-229.
[22] Uddi, M., Jiang, N., Mintusov, E., Adamovich, I. V., and Lempert, W. R., "Atomic oxygen measurements in air
and air/fuel nanosecond pulse discharges by two photon laser induced fluorescence," Proceedings of the Combustion
Institute, Vol. 32, No. 1, 2009, pp. 929-936. doi: 10.1016/j.proci.2008.06.049
[23] Zhao, Z., Chaos, M., Kazakov, A., and Dryer, F. L., "Thermal decomposition reaction and a comprehensive
kinetic model of dimethyl ether," International Journal of Chemical Kinetics, Vol. 40, No. 1, 2008, pp. 1-18.
[24] Sun, W., Uddi, M., Won, S. H., Ombrello, T., Carter, C., and Ju, Y., "Kinetic effects of non-equilibrium plasma-
assisted methane oxidation on diffusion flame extinction limits," Combustion and Flame, Vol. 159, No. 1, 2012, pp.
221-229.
[25] Pancheshnyi, S., Eismann, B., Hagelaar, G., and Pitchford, L., "Computer code ZDPlasKin." University of
Toulouse, LAPLACE, CNRS-UPS-INP, Toulouse, France, 2008.
[26] "CHEMKIN-PRO 15131." Reaction Design: San Diego, 2013.