Lecture 7 Plasma Modeling

Yiguang Ju, Princeton University

1.Plasma chemistry modeling

2.Macroscale modeling: 2D fluid model
3.Multiscale plasma catalysis modeling

Princeton Combustion Summer School

2021.6.21
 Multi-scale plasma modelling

25 Annemie Bogaerts, Online Catalysts Seminar, June 18, 2021

 7.1 Plasma chemistry: Electron impact reactions
• E/N (electric field / molecule number density):
Electron impact reactions depend on the electron energy
distribution function (or electron temperature) and the electron
collisional cross-section areas of a molecule. The electron energy
is controlled by the electric field strength, E/N.

electronic e
excitation
AB* +

CH4:O2:He=1:2:9 vibrational AB(v) + e

excitation

dissociation
e + AB A + B* + e

ionization
AB+ + 2e

CH4:O2:N2=1:2:3.76 ionization
dissociation
A+ + B + 2e
Energy loss fraction: Fraction for net energy loss per
unit time in each individual collision process k.
Energy loss coefficients(eV m3/s) attachment
 AB-
2me k T f 0
For elastic collisions: K k    [ k ( 2 f 0 + B )]d  ……
Mk 0 e 
For inelastic collisions: K k  U k kk   (2e / me )1/ 2 Mk: particle mass of target particles of collision process k.
𝑈𝑘 : threshold energy of inelastic collision process k.
Cross-sections database available for electron-molecule collisions

Andrey Starikovskiy and Nikolay Aleksandrov, AIAA paper-2017-1977

 Electron impact electronic excitation reactions

electronic
e + AB
excitation
AB* + e

e + N2 → e + N2(A, B, a’, C)

e + O2 → e + O2(𝑎1Δg),
O2(𝑏1Σg+),
O2 *

Reaction rate coefficients for each

individual collision process k (m3/s)

kk     k f 0 d
0

𝜀: electron energy.
f0: isotropic part of electron distribution
function, corresponding to zeroth-order term of
Threshold energy diagram
spherical harmonics expansion in velocity Impact on combustion
space.
𝜎 k: cross section areas of electron-neutral N2(A,B,a’,C) + O2 → N2 + 2O
collision process k. H + O2(𝑎1Δg) → OH + O
• Cross section area data comes from Lxcat database. (www.lxcat.net) H + O2 = OH + O
 Electron impact dissociation reactions
dissociation
e + AB A + B* + e

e + O2 → e + O + O Impact on combustion
e + O + O(1D)
e + O + O(1S)

e + CH4 → e + CH3 + H
e + CH2 + H2
e + CH + H2 + H
e + C + 2H2
New channel?

Threshold energy diagram

 Electron impact ionization reactions
ionization
e + AB AB+ + 2e

ionization
e + AB A+ + B + 2e
dissociation

e + O2 → 2e + O2+

e + N2 → 2e + N2+

e + CH4 → 2e + CH4+
2e + CH3+ + H
Electron and ion production

Impact on combustion: radical production and gas heating

Threshold energy diagram
dissociation recombination
N2++ CH4 → N2 + H + CH3+ e + O2 + → O + O
Electron attachment reactions
attachment
e + AB AB-

e + O2 → O 2 -

e + O2 → O + O-

Impact on combustion ~5 orders of

O2- + H → OH- +O magnitude
O2 + H = OH + O faster at 1000K
 Electron impact vibrational excitation reactions
vibrational
e + AB
excitation
AB(v) + e

e + N2 → e + N2(v) e + O2 → e + O2(v)
e + CH4 → e + CH4(v) e + CO2 → e + CO2(v)

Vibrational modes of CH4

Mode Energy(eV)
v1 Symmetric stretching 0.362
v2 Twisting 0.190
v3 Asymmetric stretching 0.374
v4 Scissoring 0.162

Vibrational modes of O2 Vibrational modes of N2 Vibrational modes of CO2

Energy(eV) Energy(eV)
v1 0.19 v1 0.29
𝜈 1:symmetric valence
v2 0.38 v2 0.59 vibrations;
v3 0.57 … … 𝜈 2: double degenerate
symmetric deformation
v4 0.75 v8 2.35 vibrations;
𝜈 3:asymmetric valence
vibrations.
1. Mi-Young Song, et al. J. Phys. Chem. Ref. Data, 44 (2015).
2. Tomas Kozak and Annemie Bogaerts. Plasma Sources Sci. Technol., 23 (2014).
3. Alexander Fridman. Plasma Chemistry, (2008).
 Reactions involving vibrationally excited species

V-T relaxation
AB(v=n) + M → AB(v=n-1) + M
e + AB
electron vibrational
excitation V-V relaxation.
AB(n) + C(m-1) → AB(n-1) + C(m)
M(v), M* + AB N2(v), O2(v),
excited CO2(v), RH(v),…
Chemical
species AB(v) + C → AC + B
reaction
excitation consumption

• VT relaxation: Energy transfer between vibrational and translational degrees of freedoms.

• VV relaxation: Energy transfer between vibrational states of different molecules.

• Vibrationally excited reaction: A reaction between a vibrational excited molecule with a

neutral molecule. It will increase the reaction rate, add new reaction pathways, and modify
heat release rate.
 Reaction rate of vibrationally excited molecules:
Fridman-Macheret α-Model
An effective decrease of activation energy
FA BC
ΔEa Ea  Ev
FA BC  FAB C
FA BC  1 Ea (1)

FAB C  2 Ea (2)

 1Ea (1) Ea (1)

 
 1Ea (1)   2 Ea (2) Ea (1)  ( 1 /  2 ) Ea (2)
Ea (1)
  (1) 1 /  2  1
Ea  Ea (2)

𝐹𝐴+𝐵𝐶 , 𝐹𝐴𝐵+𝐶 : are characteristic slopes of the terms A+BC and AB+C.
𝛾: reverse radii of corresponding exchange forces.
A + BC → AB + C 1: forward reaction direction.
2: reverse reaction direction.
A + BC*(Ev) → AB + C
Ea   Ev • The efficiency α of vibrational energy is the highest for
k ( Ev )  AT exp( n
) strongly endothermic reactions (small Ea(2)) with activation
T
energies closest to the reaction enthalpy (close to 100%).
If Ea>0, the overall
activation energy • The efficiency of vibrational energy is the lowest for
decreases exothermic reactions without activation energies (close to 0).
Alexander Fridman. Plasma Chemistry, (2008).
Vibrational-translational (V-T) relaxation for diatomic molecules

AB(v=n) + M → AB(v=n-1) + M
AB(v=1) + M → AB + M
Values of parameters for rate coefficients of the
processes of N2(v1) + M → N2 + M
 V-T relaxation rate constants of diatomic
molecules M n m A B C D
N2 1 1 7.8x10-12 218 690 1
k10 (cm3 / s )
H2 1 2/3 4.9x10-12 167.1 394 1
1
B C  E10 
 AT n exp(  )  1  D  exp(  ) H2O 1 0 2.5x10-15 21.18 0 0
 T 
1/3 m
T T CO2 1 1 1.1x10-12 218 690 1

The probability of the vibrational transition n in one collision

Em  En
Pnm  Pmn exp( )
k BT Values of parameters for rate coefficients of the
Em  En processes of O2(v1) + M → O2 + M
knm  kmn exp( )
k BT M n m A B C D
Single-quantum transitions with probabilities O2 1 0 1.35x10-12 137.9 0 1

Pn 1, n  (n  1) P10 H2 1 0 2.69x10-12 91.5 0 1

He 1 0 4.54x10-15 60.85 0 1
kn 1,n  (n  1)k10 Ar 1 2 3.14x10-12 173.1 6.2x105 1

M. Capitelli, et al. Plasms Kinetics in Atmospheric Gases, (2000).

V-T relaxation and V-V relaxation for non-diatomic molecules
 V-T relaxation rate constants calculation  V-V exchange rate constants
by Schwartz, Slawsky, and Herzfeld (SSH
theory) AB(n) + CD(m-1) → AB(n-1) + CD(m)
F ( n ) F ( nm )
kn ,n 1  k1,0 Z n
F ( 1 ) knm,n1,1m  k1,0
0,1
Zn Zm
F ( 11 )
1  xe
Zn  n Vibrational energy transfer reactions of CO2
1  nxe
Reaction xe(10-3)
1 2  2
F ( )   3  exp(   )  exp(  ) CO2(va) + M → CO2 + M
2 3  3 0.0 a

0.32E  1/2 CO2(v1) + M → CO2(va) + M 3.7 b

n  ( )
 Tg CO2(v1) + M → CO2(vb) + M 1.0 b
E  En  En 1 CO2(v1) + M → CO2(vc) + M -15.6 b
  17.5 / r0 CO2(v1) + CO2 → CO2(vb) + CO2(va) 2.8
𝑍𝑛 : scaling factor.
CO2(v1) + CO2 → CO2(va) + CO2(vb) 17.6
𝑥𝑒 : the anharmonicity of the energy levels.
𝛾𝑛 : a parameter which is a measure of adiabaticity of the
CO2(v1) + CO → CO2 + CO(v1) 5.25;6.13
reaction.
𝛼 : a parameter of the exponential repulsive potential
between the colliding species. a. Multiply by 1.0, 0.7 and 0.7 for M=CO2, CO and O2
𝜇 : reduced mass of the collision species. b. Multiply by 1.0, 0.3 and 0.4 for M=CO2, CO and O2
𝑟0 : the
。 radius parameter of the Lennard-Jones Potential. 1. M. Capitelli, et al. Plasms Kinetics in Atmospheric Gases, (2000).
3.94 A for CO2. 2. Tomas Kozak and Annemie Bogaerts. Plasma Sources Sci. Technol., 23 (2014).
3. Alexander Fridman. Plasma Chemistry, (2008).
4. M. Capitelli. Nonequilibrium Vibrational Kinetics,(1986).
7.2 Macroscale simulation of plasma assisted combustion
Challenges:
• Multi-physics problems: photons, electrons, electronic and vibrational
excitations
• Electromagnetic field, acoustic waves, shockwaves, ignition and
combustion waves
• Multi-length scale: Sheath (Debye length), diffusion & reaction zones,
mixing layer, far field effects
• Multi-timescale: from plasma, reactive flow, to combustion (ps-ms)
• Multi-species and stiffness of reactions
• Multi-dimension and far non-equilibrium (not Maxwellian distribution)
• …

Kinetic description of plasma

– distribution function
– kinetic equation
– collisional energy transfer

Multi-fluid description of plasma

– Fluid conservation equations
– Input of transport coefficients and rate coefficients, e.g. fitted as functions of E/N
– Different energy modes (electrons, vibration, neutral temperature)
– Coupling between neutral and charged particles
 Plasma Modeling: A Hybrid ZDPlasKin-CHEMKIN Model (0D)
 BOLSIG+: incorporated in ZDPlasKin[1], a
computer program for the numerical solution of
the Boltzmann equation for electrons in weakly
ionized gases in uniform electric field.
Boltzmann equation for electron energy
distribution function (EEDF) in a plasma is:
f e
   f  E   v f  C[ f ]
t m
𝑓 : electron distribution in six-dimensional phase space.
𝜐 : velocity coordinates.
𝛻v : the velocity-gradient operator.
C : the rate of change in 𝑓 due to collisons.
Assume: a steady state problem and the electric field and
the collision probabilities are spatially uniform, on the
scale of the collisional mean free path (elastic collision),
using the two-term approximation expansion in spherical
coordinates
f (v,cos , z , t )  f 0 (v, z , t )  f1 (v, z , t )cos
 ZDPlasKin[1]: A zero-dimensional Plasma Kinetics
solver which is a Fortran 90 module designed to follow
isotropic part anisotropic part
the time evolution of the species densities and gas
(temporal dependence) (spatial dependence)
temperature in a non-thermal plasma with an arbitrarily
complex chemistry. 𝑣 : the magnitude of the velocity.
𝜃 : the angle between the velocity and the field direction.
[1].S. Pancheshnyi, B. Eismann, G.J.M. Hagelaar, L.C. Pitchford, Computer code ZDPlasKin,
http://www.zdplaskin.laplace.univ-tlse.fr (University of Toulouse, LAPLACE, CNRS-UPS-INP, z : the position along this direction.
Toulouse, France, 2008).
[2]. G.J.M. Hagelaar, et al. Plasma Sources Sci. Technol., 14 (2005). BOLSIG+ solver: https://www.bolsig.laplace.univ-tlse.fr
 Plasma Modeling: A Hybrid ZDPlasKin-CHEMKIN Model
 Time evolutions of the species and temperature
of plasma reactions using ZDPlasKin
dN i jmax
  Qij (t )
dt j 1

Deposited energy: Pext  Pgas  Pelec  Pchem

3 d( N eTe )
Translational degree of Pelec  kB
freedom of electrons: 2 dt
1 d ( NT )
Translational degree of Pgas  kB
freedom of the gas:  1 dt
dNi
Internal degree of Pchem    i
freedom of the gas: i dt
 Time evolutions of the species and temperature
in combustion using CHEMKIN
• A combined ns-plasma and DC discharge to 
dYi
 iWi
dt
control electronic and vibrational excitation and
i max
dT
selective reactivity in plasma. C p    eiiWi
dt i 1

 Coupling of ZDPlasKin and CHEMKIN

in 1 -in Tn 1  Tn
 icombustion  iplasma  Tcombustion  Tplasma
t t

i* -in T *  Tn
 Tcombustion
 icombustion t
t
Tn 1  T *
in 1 -i*  Tplasma
 plasma
t
t
i

1. A.E. Lutz, et al. SEKIN, (1998).

2. P.N. Brown, et al. J Sci. Stat. Comput., 10 (1989). 𝜌𝑖∗ and 𝑇 ∗ are the intermediated density and temperature between
3. J.K. Lefkowitz, Y. Ju, et al. Phil. Trans. Soc. A, (2015).
the calculation of CHEMKIN and ZDPlasKin.
0-Dimensional model approximation
• Plane to plane
• Homogenous
• Square wave of the E/N

Kinetics model
• Species 100

• Reactions
Plasma kinetics: 541(ZDPlasKin)
HP-mech: 595 (C1-C2) (CHEMKIN) Mean electron energy

Modelling conditions
• Pressure 60torr
• Temperature 373K
• Frequency 30kHz
• Pulse number 300
• NSD 180Td
• DC 1Td, 5Td, 10Td, 20Td
• Mixture 8.333% CH4, 16.667% O2, 75% He

http://engine.princeton.edu/mechanism/HP-Mech.html Electron number density

Time evolution of the vibrationally excited species
Reaction rates in the first 20 pulses at 20Td
 1D Plasma Modeling
 Kinetic model of Nagaraja et al. (2015)
 𝛁 ⋅ 𝝐𝛁𝜙 = −𝑒 𝑛+ − 𝑛− − 𝑛𝑒
𝜕𝑛𝜖
 +𝛁 ⋅ 𝑱𝜖 = 𝑄𝜖
𝜕𝑡
𝜕𝑛𝑘
 +𝛁 ⋅ 𝑱𝑘 = 𝜔𝑘
𝜕𝑡
𝜕𝜌 𝜕𝜌𝑢𝑖
 + = 0
𝜕𝑡 𝜕𝑥𝑖
𝜕𝜌𝑢𝑖 𝜕(𝜌𝑢𝑖 𝑢𝑗 ) 𝜕𝑝 𝜕𝜏𝑖𝑗
 + =− + + 𝐹𝑖𝐸𝐻𝐷
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑖 𝜕𝑥𝑗
𝜕𝜌𝐸 𝜕[ 𝜌𝐸+𝑝 𝑢𝑖 ] 𝜕𝑞𝑖 𝜕 𝑢𝑖 𝜏𝑖𝑗
 + =− + + 𝑄 𝐽𝐻
𝜕𝑡 𝜕𝑥𝑖 𝜕𝑥𝑖 𝜕𝑥𝑗

 Transport coefficients of electron and rate

coefficients of electron impact reactions: BOLSIG
𝜙: potential (voltage) 𝑢𝑖 : 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦
𝑛𝑘 : 𝑛𝑢𝑚𝑏𝑒𝑟 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑜𝑓 𝑘𝑡ℎ 𝑠𝑝𝑒𝑐𝑖𝑒𝑠 𝐸: 𝑇𝑜𝑡𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦
𝑛𝜖 : electron energy density, 𝑛𝑒 *E𝑒

Multi-scale modeling
 Adaptive time-stepping: small (10-13 - 10-12 s)
during each discharge pulse; larger (10-10 s) in the Fig. Time evolution of short lived
gap between 2 consecutive pulses. electronically excited species after a
Nagaraja, S.,Yang, V. and Adamovich, I., 2013.. Journal of single nanosecond pulse in air at 60 Torr
Physics D: Applied Physics, 46(15), p.155205. and 300 K
 7.3 Multi-timescale modeling method

Multi-timescale (MTS) Method

t
1 
   dYi  k
 i    
 Yi  dt 
Yk  K k e
40
t = 0.1 ms
35 t = 0.2 ms
30 t = 0.3 ms

Number of species
ΔtF 25
Fastest
Group
Fast species 20
ΔtM 15
Medial
Groups 10
5
0
-1 -2 -3 -4 -5 -6 -7 -8 -9 -10 -11 -12 -13
ΔtS
Slowest Log10(characteristic time / s)
Group

0 1 2 Δt
Time
Diagram of multi time scale scheme
ΔtF is the time step of the fastest group, ΔtM is the time step of the medial
group, and ΔtS is the time step of the slowest group

Gou, Sun, Chen and Ju., Combust. Flame, 2010, 2013

 Hybrid Multi-Time Scale (HMTS) Method

• HMTS method: Initial calculation

Calculate characteristic
times

Calculate group number

and time steps

Group calculation

Parameters update

No Yes Output
End ? results

X.L. Gou, W.T. Sun, Z. Chen, Y.G. Ju, Combust. Flame 157 (6) (2010) 1111–1121.
22
 HTMS Validation: Homogeneous ignition
Ignition in
homogeneous
n-decane/Air 121 species (M. Chaos, IJCK,2007) mixture

2.8 2.0
temperature VODE
5 2.6

Ignition delay time (ms)

10 CO2 MTS
1.5 1atm

Temperature (1000K)
0
2.4 HMTS
10 OH
Mass fraction

2.2 1.0
-5
10 ODE 2.0
0.5
10
-10 MTS 1.8 20atm

-15
HMTS 1.6 0.0
10
1.4
C10H22 -0.5
-20
10 1.2 0.5 0.6 0.7 0.8 0.9
0 1 2 3 4 5
1000/Initial temperature (1/K)
Time (0.1 ms)

Temperature and species profiles Ignition delay time for n-decane-air

Comparison of computation efficiency: same mechanism
with HMTS, HMTS/DAC vs. VODE, VODE/DAC
120
Real Fuel 2_Reduced_425
P = 1.0 atm; Φ = 1.0; T0 = 400 K
100
PFA Time w. DAC
CPU Time

80

Solver's Time (VODE

60
or HMTS)

40 Other Time

20
?
0

1D unsteady Flame propagation

 Methodology
Correlated DAC (CO-DAC) method

• Correlated groups Time correlation

Space correlation

tn+1

tn
How to choose
criteria?

Sun, W., Gou, X., El-Asrag, H.A., Chen, Z. and Ju, Y., 2015. Multi-timescale and correlated dynamic adaptive chemistry modeling of
ignition and flame propagation using a real jet fuel surrogate model. Combustion and Flame, 162(4), pp.1530-1539.

25
Computation efficiency: Spherically Propagating Flame

• Computation time

Transport

~50%

Sun, W., Gou, X., El-Asrag, H.A., Chen, Z. and Ju, Y., 2015. Multi-timescale and correlated dynamic adaptive chemistry modeling of
ignition and flame propagation using a real jet fuel surrogate model. Combustion and Flame, 162(4), pp.1530-1539. 26
 7.4 2D fluid model for plasma discharge

Input of transport coefficients

and rate coefficients, e.g.
fitted as functions of E/N

Anne Bourdon, Sumire Kobayashi, Zdenek Bonaventura,Fabien Tholin and Nikolay Popov, Kaust
Research Conference: New Combustion Concepts, March 6-8, 2017, KAUST
Anne Bourdon, Sumire Kobayashi, Zdenek Bonaventura,Fabien Tholin and Nikolay Popov,
Kaust Research Conference: New Combustion Concepts, March 6-8, 2017, KAUST
Time histories of species and temperature

Key reactions for gas heating

• Rapid decrease of ne at the end of the voltage pulse, then a

much slower decrease up to 100 ns
• O(P), H and OH are mainly produced in the post-discharge by
dissociative quenching reactions

Anne Bourdon, Sumire Kobayashi, Zdenek Bonaventura,Fabien Tholin and Nikolay Popov,
Kaust Research Conference: New Combustion Concepts, March 6-8, 2017, KAUST
Yifei Zhu et al 2017 Plasma Sources Sci. Technol. 26 125004
Yifei Zhu et al 2017 Plasma Sources Sci. Technol. 26 125004
2D fluid model for plasma assisted combustion

A multiscale adaptive reduced chemistry simulation of plasma assisted combustion

(MARCS-PAC) solver: PASSKEy + ASURP + MTS/CO-DAC

Governing equations
 PASSKEy  ASURF+
∂nk ∂ρ
Species: + ∇∙Jk = Sk + Sph Mass: + ∇∙(ρV) = 0
∂t ∂t
∂(ρYk )
Jk = zk μk nk E−Dk ∇nk Species: + ∇∙[ρ(V+V'k )Yk ] = ωk
∂t
(For charged species, the flux is calculated
∂(ρu) ∂p
based on drift-diffusion approximation) Momentum: + ∇∙(ρuV) = − + ∇∙ 𝜏
∂t ∂R
∂(ne ϵe ) ∂(ρv) ∂p
Electron energy equation: + ∇∙Jϵ = − E∙ Je − ϴe + ∇∙(ρvV) = − + ∇∙𝜏
∂t ∂t ∂Z
Poisson’s equation: ∇∙( − ε0 ε∇φ) = qe zk nk Energy: ∂e
+ ∇∙((e + p)V) = − ∇∙q + ∇∙(𝜏V) + Pgas
k ∂t
Energy: Pgas = Pext − Pelec − Pinter

 Coupling of PASSKEy and ASURF+

plasma
Species: ωk = ωk + ωcombustion
k

plasma
ωk = Sk + Sph − ∇∙Jk

Xingqiang Mao et al., Combustion and Flame, submitted, 2021 33

 2D effects of plasma discharge on ignition
 Electrode structure  Input voltage and calculated current

D = 2.0 mm
Cylindrical electrode

Xingqiang Mao et al., Combustion and Flame, submitted, 2021 34

 2D effects of plasma discharge on ignition
 2D modeling results

(a) E/N

(b) Te

(c) electron

Xingqiang Mao et al., Combustion and Flame, submitted, 2021 35

 2D effects of plasma discharge on ignition
 2D modeling results

(d) O(1D)

(e) O

(f) T

Xingqiang Mao et al., Combustion and Flame, submitted, 2021 36

 2D effects of plasma discharge on ignition
 Time dependent results near anode (Point 1)

During the discharge After discharge (~ 10ns) After discharge (10ns ~ 1μs)
e + N2 → e + N2(v1) N2(B) + O2 → N2 + O + O N2(B) + N2/H2 → N2(A) + N2/H2
e + N2 → e + N2(A)/N2(B)/N2(a’)/N2(C) N2(C) + O2 → N2 + O + O(1D) N2(A) + H2 → N2 + H + H
e + H2 → e + H + H N2(a’)/N2(C) + H2 → N2 + H + H N2(A) + O2 → N2 + O + O
e + O2 → e + O + O/O(1D) O(1D) + H2 → H + OH N2(v1) + H2 → N2 + H2
e + N2 → 2e + N2+ N(2D) + O2 → NO + O/O(1D)
…

Xingqiang Mao et al., Combustion and Flame, submitted, 2021 37

 7.5 Plasma catalysis Modeling: packed bed DBD
[1]
DBD reactor for CO2 splitting

Dielectric packing

[1] Van Laer and Bogaerts, Energy Technology, 3 (2015) 1038

13 (ii) Plasma reactor: packed bed DBD

 Polarization of packing beads
Polarization of packing beads (4 kV, 23.5 kHz)
=> enhanced electric field
=> enhanced electron temperature

Power more efficiently used for electron heating

 more e-impact ionization + excitation-dissociation of CO2  E-efficiency 
(ii) Plasma reactor: packed bed DBD
 Streamer propagation (dry air)

2 main types of discharge behavior:

- filamentary microdischarges
(localized between packing beads)
- surface ionization waves
(surface charging)

r = 5: surface ionization waves

Wang, et al., Chem. Eng. J, 334, 2467 (2018)

(ii) Plasma reactor: packed bed DBD
 PIC-MCC model for streamer propagation

Other packing geometries: honeycomb + 3DFD structures

Honeycomb structure: 3DFD structure:

Streamers limited to 1 channel Streamers distribute over channels
 Discharge enhancement
Channels // electrodes: Channels  electrodes:  Broader plasma distribution
Low plasma density Higher plasma density  Beneficial for plasma catalysis
 Larger catalyst surface area
exposed to plasma

Need for experimental validation !

Zhang and Bogaerts, PSST 27, 105013 (2018)

(ii) Plasma reactor: packed bed DBD
 8. Thermal-Chemical Instability in
Plasma Assisted Combustion and Chemical Reforming

Yiguang Ju
Princeton University

Princeton Combustion Summer School

2021.6.21
 Why does lightening love hydrocarbons
Lightening captures rocket exhaust trail?
Air pollution helps wildfires create their own lightning

Nikk Ogasa Science, 2021 doi:10.1126/science.abj6782

https://www.space.com/russian-rocket-
launch-lightning-strike.html, May 27, 2019
 Questions: when will plasma contract to an arc?
How can we control plasma contraction to enable volumetric chemical enhancement?

Plasma assisted ignition for HCCI to remedy misfire

Fuel/NH3 synthesis

https://insidemazda.mazdausa.com/press-release/mazda-next-
generation-technology-press-information/, October 24, 2017
What is the mechanism of plasma instability
for a reactive mixture?

4
 Classical Plasma Instability Theory: Joule heating

Cathode volumetric Anode Icr

(upper lower)

contracted
Icr
(lower upper)

Figure. Diffuse (top) and contracted (bottom) discharges.

Pictures from J. Stark’s book (1902)
Shneider, M. N, et al. 2014

Positive feedback between Joule heating and electric field.

δne ↑ δ(jE) ↑ δT↑ T: gas temperature

N: gas number density Streamer-leader transition
E/N: reduced electric field
Te: electron temperature Gallimberti [1979, 2002]
ne: electron number density Bazelyan et al. [2007]
δTe ↑ δ(E/N) ↑ δN ↓ jE: Joule heating 5
Examples of Coexistence of Constricted and Diffuse Regions at const. pressure 6

Experiments by Ionikh et al. [2008]: glow discharge in tube

In a steady state partially constricted discharge one part of the column was
constricted while the other part remained diffuse in Ar:N2 mixture

Glow Discharge with free boundaries

(Yatsenko, 1995)
The discharge occurred in Ar at 185 Torr. The
gap between electrodes is 6 cm.
Left panel U=450 V, I=130 mA. Right panel
U=500 V, I=115 A

Courtesy of M. Shneider
How does combustion change plasma instability/contraction?
Hypothesis: Plasma Chemistry Instability?

Fuel/air mixture? Plasma thermal instability

 ne  jE  T   N    i ( E / N )   ne 
Plasma-chemistry instability
(1) Fuel ionization & electron increase  ne 
Plasma-combustion chemistry coupling Electron impact fuel pyrolysis
(2) → T decrease  ne 
Fuel(RH) (E / N ) 

e + RH = R+ + H + 2e
+OH
(3) Vibrational energy transfer to fuel
& oxidized species→ T decrease
 ne 
e + N2 = N2(A,B,C) + e RO2
+O2 Small C2H3/CH2O (E / N ) 
R alkene
e + O2 = O2-
 ne 
+O2 Oxygen consumption by fuel
e +O2=O+O(1D) + e (4) → Reduced electron attachment
QOOH HO2 H/HCO
H+O2(1Δg) =O+OH +O2+(M)
O(1D)+RH =OH+R
N2(A,B,C)+O2=O+O+N2 O2QOOH H2O2
+O2 CO/CO2
(5)
Plasma assisted fast cool flame
heat release → T increase
 ne 
N2(v)+HO2 =OH+O+N2 (E / N ) 
R(v,*)+O2 =RO2
R + O2(v) = RO2
Slow
2OH
Plasma (6) Plasma assisted hot flame heat
release → T increase
 ne 
O3+M =O+O2+M e, R*, N2*, O2* (E / N ) 
R(*), R(v), N2(v), O2(v)
Zhong, H., Shneider, M.N., Mokrov, M.S. and Ju, Y., 2019.
Ju and Sun, PECS, 2015 Journal of Physics D: Applied Physics, 52(48), p.484001. 7
 Objective
• Experimental demonstration of the existence of
plasma chemistry instability

• Theoretical analysis of the kinetics of plasma

chemistry instability

• Computational modeling of thermal-chemical

instability

8
8.1 Experimental Study
 Reactor
 Pressure: 50 Torr
 Initial Temperature: 293 K
 Flow speed: 70 cm/s
 Gap distance: 14 mm
 Nanosecond repetitively pulsed discharge: FID
GmbH FPG 30-50MC4
 Peak Voltage: ~7 kV
 Pulse Duration: ~12 ns FWHM
 Pulse burst mode: 150-1500 pulses, 30 kHz
 Quartz double dielectric barrier: 1.6 mm thickness
 Diagnostics:
 ICCD 1 ns gate imaging
 CMOS 30 kHz high speed imaging
 E-FISH: electric field
 Rayleigh Scattering: number density
99
Rousso, A.C., Goldberg, B.M., Chen, T.Y., Wu, S., Dogariu, A., Miles, R.B., Kolemen, E. and Ju, Y., 2020. Plasma Sources Science and Technology, 29(10), p.105012.
Electric field measurement: E-FISH
 Electric Field Induced Second Harmonic Generation (E-FISH)
 Externally applied electric field induces a dipole in a medium, which
destroys inversion symmetry
 Third order nonlinear interactions cause second harmonic
generation
2
 𝐼 2𝜔 = 𝐴 ∙ 𝑁 2 𝐸𝐸𝑥𝑡 2 𝐼𝑃𝑢𝑚𝑝
 Measurement length scale determined by beam focusing
 Laser: Spectra-Physics Solstice Ace Femtosecond laser
 Peak energy: 6 mJ
 Pulse duration 60 fs
 Measurement volume:
 25µm x 1cm x 2 cm
 PIMax4 1024i ICCD

10
10 A Dogariu et al., Phys. Rev. Applied. (2017).
B.M. Goldberg et al., Appl. Phys. Lett 112 (2018) 064102.
Nanosecond Imaging of pulses
6.6% CH4 13.4% O2 80% Ar
Pulse 25 Pulse 100 Pulse 150

Scale/5 Scale/5

 Nanosecond resolution of single pulses

 1 ns gate, but only 10 Hz camera repetition rate
 Images not from the same burst: uncorrelated shot to shot
 Shows evidence of significant streamers by pulse 100
 Non-uniformity moves from burst to burst
11
11
High Speed Imaging 6.6% CH4 13.4% O2 80% Ar
End View: Streamwise direction
 Using high speed CMOS cameras, image
every pulse in a burst
 10 us gate
 Filaments start forming around pulse 100
 Within a given burst, filaments are stationary
 Convection slows uniformity formation in the
orthogonal direction
End View: Streamwise direction 13.4% O2 86.6% Ar, no combustion Side View: Orthogonal direction
 Average Electric Field: Pulse 100
 Field follows applied voltage until breakdown
Voltage pulse we send in
 Sharp drop due to plasma formation Sub-breakdown
 Lack of additional lobes likely due to higher electron
concentrations
 Highly mobile electrons within discharge move to shield plasma volume from applied
voltages
 Low field after main pulse indicates a diffuse discharge

6.6% CH4 93.3% Ar 13.4% O2 86.6% Ar 6.6% CH4 13.4% O2 80% Ar

8000 8000 8000
Applied Electric Field (V/cm)

Applied Electric Field (V/cm)

Applied Electric Field (V/cm)

Applied Electric Field Applied Electric Field Applied Electric Field
Pulse 100 Pulse 100 Pulse 100
6000 6000 6000

CH4/Ar O2/Ar CH4/O2/Ar

4000 4000 4000

2000 2000 2000

0 0 0

13
13 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Time (ns) Time (ns) Time (ns)
Single-Shot E-FISH: Pulse 150
6.6% CH4 13.4% O2 80% Ar
 Very non-uniform E-filed distribution 6750
Time Bin Average
 Still raw counts, not electric field 6000 Shot 30
Shot 33
 Signal is proportional to ~E2 5250 Shot 48
4500 Shot 56
Shot 64

Counts
3750 Shot 66
Shot 76
3000
2250
1500
750
0
0 2 4 6 8 10 12
E≠V/d Measurement Length (mm)
14
14
Filaments cause a much stronger local field than the average
Time-dependent 1-D Single shot E-field
 Single shot data across pulse 150
 ~1 cm spatial resolution 6.6% CH4 13.4% O2 80% Ar
Pulse 150
 25 µm per pixel
 Preliminary raw data – signal counts
only
 Significant spikes of signal
 As expected, these strong local e-
fields are prior to breakdown, but
also during secondary pulses
13.4% O2 86.6% Ar
Pulse 150

Measurement
15
15
Region
 PRINCETON
Mechanical
and Aerospace

Pentane/O2 mixtures
Engineering

Stoichiometric 2.8% C5H12, 22.2% O2, 75% Ar Dilution

Plasma Pulses: Burst, 30 KHz,
Pulse 600 pulses per burst
50 300

Without Pentane
150 450

200 600

With Pentane
Pulse
50 300

150 450
OH Fluorescence
100 shot exposure
200 600

Images of single plasma pulses, 10 shots averaged

8.2 Theoretical analysis of Plasma Thermal-Chemical Instability

ne, nn, np, T: number density of electron, negative,

and positive particles, and temperature
 𝒅𝑬
=𝟎
𝒅𝑰

Fig.2 Comparison of the homogeneous branch of the

current–voltage characteristics using zero-dimensional and one-
dimensional simulation under the same condition

Zhong, H., Shneider, M. N., Mao, X., & Ju, Y. (2020),vPlasma Sources Science and Technology, 2020, Submitted.
 Plasma thermal-chemical instability criterion
=0

Δ represents interactions between ionization and other plasma kinetics

Combustion heat release rate, transport, vibrational relaxation, plasma and

combustion chemistry, and Joule heating all affect the onset of instability.

Zhong, H., Shneider, M. N., Mao, X., & Ju, Y. (2020),vPlasma Sources Science and Technology, 2020, Submitted.
0D Plasma Model with Detailed Chemical Kinetics

Only consider the homogeneous non-equilibrium plasma volume (Te > Tv > Tg)

𝑑𝑁𝑖 𝑁𝑖 Auxiliary Relations

= 𝜔𝑖𝑗 𝑡 −
𝑑𝑡 𝜏 𝐼
species 𝑗 chemical convective 𝐸 𝑡 = current-voltage
number reaction removal 𝜎 𝑡 𝑆 characteristic
density
𝑝 = 𝑁 𝑘 𝑇𝑔 = Const isobaric
𝛾 𝑑(𝑁𝑇𝑔 ) 3 𝑑 𝑛𝑒 𝑇𝑒 𝑑𝑁𝑖 𝐻 − 𝐻0 condition
𝑘 = 𝑗𝑐 𝐸 − 𝑘 − ℎ𝑖 −
𝛾−1 𝑑𝑡 2 𝑑𝑡 𝑑𝑡 𝜏
𝑖
gas translational Joule energy transfer chemical convective
energy heating with electrons reaction removal

• Implemented by integrating the CHEMKIN [1] and ZDPlaskin [2].

• EEDF is calculated by BOLSIG+ [3].
• Example: N2-H2-O2 mechanism
• 62 species, 151 elementary reactions
• 666 plasma reactions
20
[1] Kee, R. J. et al., (1989) [2] Pancheshnyi, S., et al., (2008). [3] Hagelaar, G. J. M., et al., (2005)
Thermal-Chemical Mode Analysis (TCMA)
𝑑𝒚
0D plasma model in generic form: = 𝒈(𝒚)
𝑑𝑡

y: time-dependent column vector consisting species and temperature.

g: source terms containing the thermal-chemical property of the system.

Eigen-decomposition of thermal-chemical Jacobian Matrix 𝑱 𝑡 :

𝑑𝒈
𝑱 𝑡 = = 𝑨𝜦𝑩, 𝜦 = 𝐷𝑖𝑎𝑔(𝜆1 , 𝜆2 , 𝜆3 , … 𝜆𝑁 )
𝑑𝒚
𝑑𝒇
Conversion from column vector y to thermal-chemical modes 𝒇: = 𝜦𝒇, 𝒇 = 𝑩𝒈
𝑑𝑡
Why thermal-chemical modes 𝒇?
• 𝑓 = 𝑓0 𝑒 Λ𝑡 : the sign of Re(Λ) determines the stability of the system.
• 𝑅𝑒(Λ) < 0: system stable.
• 𝑅𝑒 Λ𝑖 > 0: the thermal-chemical mode 𝑓𝑖 corresponding to Λ𝑖 is unstable.

21
[1] Zhong, H., et al., PSST, 2020, Under Review. [2] Lam, S. H., et al., Symposium (International) on Combustion, 1989.
 Explosive mode species/temperature pointer (SP)

• Thermal-chemical modes (TCMs) are a bridge between original time dynamics and stability.

original time dynamics thermal-chemical modes system stability

eigen- 𝒇 = 𝒇𝟎 𝒆𝜦𝒕
𝑑𝒚 𝑑𝒇
= 𝜦𝒇, 𝒇 = 𝑩𝒈 sign(𝑹𝒆(𝜦))
= 𝒈(𝒚) decomposition 𝑑𝑡
𝑑𝑡

• How to directly connect system stability and

species/temperature? 𝒚𝟏 = [OH]
TCM 𝒇
• SPi = 𝐷𝑖𝑎𝑔(𝑎𝑖 ∘ 𝑏𝑖 ) for ith mode
𝒚𝟐 = [e]
• SP: “coordinate” of a mode 𝒇 in … 𝒚𝟑 = [N2*]
species/temperature “axis”. 𝒚𝟒 = [N2(v=1)]
𝒚𝑵 = 𝑇𝑔
[1] Zhong, H., et al., PSST, 2020, Under Review.
[2] Lam, S. H., et al., Symposium (International) on Combustion, 1989.
Comparisons among Thermal Instability, Thermal-
Chemical Instability and Ignition

Initial Composition 𝑇0 , K p, Torr 𝜏, ms I, mA Comment

Thermal-ionization
Case 1 N2 :O2 = 80:20 800 200 5 0.5
instability
N2 :O2: H2 = Thermal-chemical
Case 2 800 200 5 0.5
80:6.7:13.3 instability

N2 :O2: H2 =
Case 3 825 200 5 0 Homogeneous ignition
80:6.7:13.3

23
[1] Zhong, H., et al., PSST, 2020, Under Review.
Dynamics and Unstable Thermal-Chemical Modes

Time, ms Time, ms Time, ms

• Thermal-chemical modes of plasma thermal-chemical instability are different from other two
cases.
24
[1] Zhong, H., et al., PSST, 2020, Under Review.
 Thermal-Chemical Mode Analysis (TCMA) in
Plasma Thermal-Chemical Instability of H2/air ignition

(unit: s-1)

• TCMA allows for a clear separation of

different kinetics.

• TCMA provides an overview for

reaction dynamics and active species.

25
[1] Zhong, H., et al., PSST, 2020, Under Review.
8.3 1D modeling of plasma thermal-chemical instability

• Quasi-neutrality of plasma. (H2/N2/O2) p=Const=50 Torr

Neglect near-electrode sheath dynamics.
Convection is “zero-dimensional”.

• Max reduced electric field < 80 Td τconv E

Electron energy 0.8-1.5 eV

• Only consider three-body attachment to O2 as

e + O2 + M -> O2- + M for electron attachment.

• In view of rapid vibrational-translational (VT) • DC discharge.

relaxation of oxygen molecules, it’s assumed that
the vibrational energy is stored in the nitrogen • 1D configuration (radial dependent)
molecules.

26
Numerical Model

1D cylindrical governing equations Charge conservation

Two possible solutions:

ne , n- , n+ , T, Tv ne , n- , n+ , T, Tv

r r
contraction homogeneous

27
Plasma and Combustion Kinetics

HP Mech
Simplified Electron/Ion Kinetics [4,5,6]
(H2-O2 reactions)1,2,3
e + N2 => e + e + N2+
Chain Reactions e + O2 => e + e + O2+
H+O2 <=> O+OH e + H2 => e + e + H2+
O+H2 <=> H+OH e + O2 + H2O => O2- + H2O
H2+OH <=>H2O+H e + O2 + O2 => O2- + O2
Dissociation Reactions O2- + O2 => O2 + O2 + e
H2+M<=>H+H+M ……
O2+M <=> O+O+M
H2O+M <=> H+OH+M
Formation and Consumption Simplified Excited Species Kinetics [6]
of HO2 and H2O2 e + N2 => N2(v) + e
…… N2(v) + O2 => 2O + N2
……

[1] X. Shen, et al., 2015 [4] A.V. Phelps, ftp://jila.colorado.edu/collision data/eletrans.txt

[2] M.P. Burke, et al., 2012 [5] BOLSIG+, https://www.bolsig.laplace.univ-tlse.fr/
[3] http://engine.princeton.edu/mechanism/HP-Mech.html [6] M. N. Shneider, et al. 2014 28
Case studies

Initial
Initial Composition Comments
Temperature, K
Baseline case.
Case 1 N2 :O2 =80:20 293
Results matched well with [1].
Case1 vs. Case2:
Case 2 N2 :H2 =80:20 293
effect of fuel ionization
Case1 vs. Case3.a:
Case
N2 :O2: H2=80:10:10 293 combustion initiation in the hot
3.a
contracted channel

Case1 vs. Case3.b:

Case
N2 :O2: H2=80:10:10 800 chemical kinetics triggering
3.b
the contraction

29
[1] Shneider, M. N., M. S. Mokrov, and G. M. Milikh. Physics of Plasmas 21.3 (2014): 032122.
 Effect of Electron Attachment to Oxygen w. H2O

How electron attachment would be different for combustible mixtures?

At room temperature [1]
e + O2 + N2 -> O2- + N2 ka=1.6× 10−31 cm6/s
e + O2 + O2 -> O2- + O2 ka=2.5× 10−30 cm6/s
e + O2 + H2O -> O2- + H2O ka=1.4× 10−29 cm6/s
O2 H2O
Icr Icr
(lower→upper) (upper→lower)

Case 1 N2 (80%) + O2 (20%)

5 mA 30 mA
ka ~ 10-32 cm6 s-1

Case 4 Perturb ka
6 mA 36 mA
ka ~ 10-31 cm6 s-1 Composition: N2 (80%) + O2 (20%)
Pressure: 50 Torr

Electron attachment to oxygen in combustion will affect the onset of plasma

combustion instability (PCI).
30
[1] Data source: Fridman, Alexander. Plasma Chemistry. Cambridge University Press, 2008.
Effect of Fuel Ionization

Case 1: N2 (80%) + O2 (20%) • Ionization potential

Case 2: N2 (80%) + H2 (20%) H2 : 15.40 eV, O2 12.06 eV

Icr (lower→upper) Icr (upper→lower) O2 H2

Case 1 O2 5 mA 30 mA
Case 2 H2 14 mA 120 mA

Fuel ionization potential extends the onset of plasma

instability to a higher current. 31
 Effect of Plasma Assisted Exothermic Fuel Oxidation

Case 1: N2 (80%) + O2 (20%)

Case 3a: Initial: N2 (80%) + O2 (20%)
Injecting stoichiometric H2/O2/N2 mixtures Uniform discharge

into the system Onset of instability Onset of

w. combustion instability
Without
combustion
H2/O2/N2
N2/O2
80:20
4 × 10−2 mol/s Contracted arcing
φ=1

Fuel oxidation shifts the stability boundary, making the system more unstable.
• Fuel oxidation not only releases heat, but also updates the
composition and therefore mixture kinetic properties. 32
Effect of Exothermic Fuel Oxidation

Case 1: N2 (80%) + O2 (20%)

Case 3a: N2:O2: H2 = 80:10:10 at 293 K

• The heat generated from the contraction

initiates combustion in the flow channel.

• Combustion not only releases heat, but

also changes the plasma chemical
properties due to composition changes.

33
Effect of Exothermic Fuel Oxidation w. Rapid Ignition

Case 1: N2 (80%) + O2 (20%)

Case 3b: N2:O2: H2 = 80:10:10 at 800 K

• At 10-2 s, chemical reactions were

triggered homogeneously first.

• At 10-1 s, the weakly ionized plasma

began to contract.

34
 Conclusion

• A new thermal-chemical mechanism for plasma-chemistry instability (PCI) is

proposed and experimentally demonstrated.

• E-FISH shows PCI have drastic effect on the sharp electric field distribution.

• An theoretical model for plasma chemistry instability is developed.

• Plasma assisted reaction pathways in a combustible mixtures can enhance PCI

by releasing heat, changing electron attachment and plasma chemistry.

• Plasma chemistry instability can be used to control plasma assisted volumetric

ignition and fuel reforming.

35
9. Future research of plasma assisted reactive flow

Yiguang Ju

Princeton University

2021. 6. 21

Princeton Combustion Summer School

What are our questions?

Chemistry Combustion
Plasma
Physics
Instability Catalysis

2
 9.1 Future research of plasma research and applications
1. Game changers in non-equilibrium plasma applications
• Engines (ICEs: lean burn, plasma assisted HCCI, Turbine engine: relight, ignition, NH3 combustion)
• Chemical synthesis and novel plasma catalysts: CO2 capture and utilization , H2, NH3, chemicals
• Materials synthesis
• Bio-medicine
• Electronics
• …

2. Game changers in plasma control

• High pressure volumetric discharge and plasma instability control
• Selective excitation (electronic and vibrational)
• Selective active species/radicals production
• Low cost and low electronic noise plasma resources
• …

3. Fundamental Understanding of PAC

• Diagnostics of plasma properties: Electric field, electron number density, excitation states, non-equilibrium
• Plasma physics: Energy transfer processes between different excited states and plasma instability
• Plasma chemistry: reactions involving excited states, non-equilibrium kinetics, and plasma catalysis
• Kinetic models: Key species, reaction rates, and cross-section areas for large fuel molecules
• Multi-scale plasma modeling tools: atomic scale to macroscale
 9.2 Research questions in plasma catalysis

How does the catalyst affect the plasma?

• Electric field enhancement ?
• Change in discharge type ?
• Can plasma penetrate in catalyst pores ?
• New products formed at catalyst surface ?
• Effect of adsorption at catalyst on plasma species densities ?
Finding answers to these Needed:
crucial questions Multi-scale modeling

How does the plasma affect the catalyst ?

• Interaction radicals, excited species with catalyst ?
• Effect electric field on surface reactions ?
• Effect surface morphology on behavior of species at surface ?
Combination of:
(i) Plasma chemistry modeling (0D scale), (ii) Plasma reactor modeling (macro-scale),
(iii) Modeling in catalyst pores (micro-scale), (iv) Surface chemistry (atomic / 0D-scale):
Challenging…  Long-term perspective !
3 Annemie Bogaerts, Online Catalysts Seminar, June 18, 2021