Chemkin Theory
Chemkin Theory
Chemkin Theory
Release
4.0
Theory Manual
CHEMKIN Software
RD01400-C01-004-001A
June 2004Revision A
Licensing:
For licensing information, please contact Reaction Design at (858) 550-1920 (USA) or [email protected].
Technical Support:
Reaction Design provides an allotment of technical support to its Licensees free of charge. To request technical support, please include your license
number along with input or output files, and any error messages pertaining to your question or problem. Requests may be directed in the following
manner: E-mail: [email protected], Fax: (858) 550-1925, Phone: (858) 550-1920.
Additional technical support hours may also be purchased. Please contact Reaction Design for the hourly rates.
Copyright:
Copyright 2004 Reaction Design. All rights reserved. No part of this book may be reproduced in any form or by any means without express written
permission from Reaction Design.
Trademark:
CHEMKIN is a regsitered trademark of Reaction Design.
AURORA, CONP, CRESLAF, EQUIL, EQUILIB, KINetics, OPPDIF, OVEND, PASR, PLUG, PREMIX, Reaction Design, SENKIN, SHOCK, SPIN, SURFACE CHEMKIN,
SURFTHERM, TRANSPORT, TWAFER, TWOPNT are all trademarks of Reaction Design or Sandia National Laboratories.
Limitation of Warranty:
The software is provided as is by Reaction Design, without warranty of any kind including, without limitation, any warranty against infringement of
third party property rights, fitness or merchantability, or fitness for a particular purpose, even if Reaction Design has been informed of such purpose.
Furthermore, Reaction Design does not warrant, guarantee, or make any representations regarding the use or the results of the use, of the software or
documentation in terms of correctness, accuracy, reliability or otherwise. No agent of Reaction Design is authorized to alter or exceed the warranty
obligations of Reaction Design as set forth herein. Any liability of Reaction Design, its officers, agents or employees with respect to the software or the
performance thereof under any warranty, contract, negligence, strict liability, vicarious liability or other theory will be limited exclusively to product
replacement or, if replacement is inadequate as a remedy or in Reaction Designs opinion impractical, to a credit of amounts paid to Reaction Design
for the license of the software.
CHEMKIN 4.0
Contents
Table of Contents
Acknowledgments ....................................................................................................................................................11
1
Introduction................................................................................................................................................................13
1.1
1.2
Thermodynamic Expressions................................................................................................................................17
2.1
2.2
2.3
RD01400-C01-004-001A
CHEMKIN 4.0
3.6
3.7
3.8
3.9
4.9
4.10
4.11
7.3
Table of Contents
Shock-tube Experiments.............................................................................................................................99
Rankine-Hugoniot Relations for Normal Shocks ..................................................................................101
7.2.1
Shock Tube Laboratory Time and Gas-particle Time.........................................................101
7.2.2
Incident Shock Initial Conditions ............................................................................................103
7.2.3
Reflected Shock Initial Conditions .........................................................................................106
Downstream Model Equations.................................................................................................................110
7.3.1
Shock Tube Boundary Layer Effects.....................................................................................112
RD01400-C01-004-001A
Table of Contents
8.4
8.5
14
13
12
11
8.3.6
Heat Exchange Between Reactors in Reactor Clusters.....................................................129
8.3.7
Optional Wall Energy Balance and Heat Capacity Effects ................................................131
8.3.8
Treatment of Activities for Bulk Species ...............................................................................133
Internal Combustion Engine Model.........................................................................................................134
8.4.1
Heat-Transfer Options for the IC Engine Model ..................................................................137
Plasma Systems ........................................................................................................................................138
8.5.1
Electron Energy Equation for Plasma Systems...................................................................139
8.5.2
Gas Energy Equation Adjusted for Plasma Systems .........................................................142
8.5.3
Application of the Bohm Condition for Ion Fluxes to Surfaces..........................................142
8.5.4
Summary of Solution Variables for Homogeneous Systems.............................................144
10
Theory Manual
RD01400-C01-004-001A
CHEMKIN 4.0
15
15.2
15.3
16
Sensitivity Analysis................................................................................................................................................225
16.1
16.2
16.3
16.4
17
Table of Contents
Nomenclature ..........................................................................................................................................................237
Latin Equation Symbols ............................................................................................................................237
Greek Equation Symbols..........................................................................................................................252
Subscript Equation Symbols ....................................................................................................................257
Index ..........................................................................................................................................................................259
RD01400-C01-004-001A
CHEMKIN 4.0
Tables
List of Tables
Latin Symbols ..............................................................................................................................................................237
Greek Symbols ............................................................................................................................................................252
Subscript Symbols.......................................................................................................................................................257
RD01400-C01-004-001A
CHEMKIN 4.0
List of Tables
RD01400-C01-004-001A
CHEMKIN 4.0
Figures
List of Figures
3-1
Rate constant as a function of pressure at fixed temperature for a unimolecular fall-off reaction ................................41
3-2
3-3
Rate constant as a function of pressure at fixed temperature for a chemically activated reaction ...............................43
4-1
4-2
4-3
4-4
4-5
4-6
4-7
4-8
Sticking Coefficient for SiH2(g)Collision Probability with Si(s) and C(s) ....................................................................59
4-9
7-2
7-3
7-4
RD01400-C01-004-001A
CHEMKIN 4.0
List of Figures
8-1
8-2
8-3
8-4
Schematic of the gas phase control volume in the absence of the reactor wall. .........................................................132
8-5
Schematic of the gas-phase and wall control volumes for energy balance ................................................................132
8-6
9-1
Unmixedness vs. mixing frequency for PaSR of stoichiometric H2/air mixture with 1 ms residence time ..................151
10
RD01400-C01-004-001A
CHEMKIN 4.0
Acknowledgments
The content of this document is based, in part, on the following publications:
M. E. Coltrin, R. J. Kee, G. H. Evans, E. Meeks, F. M. Rupley, and J. F. Grcar, SPIN
(Version 3.83): A Fortran Program for Modeling One-Dimensional RotatingDisk/Stagnation-Flow Chemical Vapor Deposition Reactors, Sandia National
Laboratories Report SAND91-8003, 1991.
M. E. Coltrin, R. J. Kee, F. M. Rupley, and E. Meeks, SURFACE CHEMKIN III: A
Fortran Package for Analyzing Heterogeneous Chemical Kinetics at a SolidSurface - Gas-Phase Interface, Sandia National Laboratories Report SAND968217, 1996.
M. E. Coltrin and H. K. Moffat, SURFTHERM: A Program to Analyze
Thermochemical and Kinetic Data in Gas-phase and Surface Chemical Reaction
Mechanisms, Sandia National Laboratories Report SAND94-0219, 1994.
M. E. Coltrin, H. K. Moffat, R. J. Kee, and F. M. Rupley, CRESLAF (Version 4.0): A
Fortran Program for Modeling Laminar, Chemically Reacting, Boundary-Layer Flow
in Cylindrical or Planar Channels, Sandia National Laboratories Report SAND930478, 1993.
R. J. Kee, G. Dixon-Lewis, J. Warnatz, M. E. Coltrin, and J. A. Miller, A Fortran
Computer Code Package for the Evaluation of Gas-Phase Multicomponent
Transport Properties, Sandia National Laboratories Report SAND86-8246, (1986).
R. J. Kee, J. F. Grcar, M. D. Smooke, and J. A. Miller, A Fortran Program for
Modeling Steady Laminar One-Dimensional Premixed Flames, Sandia National
Laboratories Report SAND85-8240 (1985).
R. J. Kee, J. A. Miller, and T. H. Jefferson, Chemkin: A General-Purpose, ProblemIndependent, Transportable, Fortran Chemical Kinetics Code Package, Sandia
National Laboratories Report SAND80-8003, 1980.
R. J. Kee, F. M. Rupley, E. Meeks, and J. A. Miller, Chemkin-III: A Fortran Chemical
Kinetics Package for the Analysis of Gas-Phase Chemical and Plasma Kinetics,
Sandia National Laboratories Report SAND96-8216, 1996.
RD01400-C01-004-001A
11
CHEMKIN 4.0
Acknowledgments
12
RD01400-C01-004-001A
CHEMKIN 4.0
Introduction
Successful application of CHEMKIN to engineering and chemistry problems requires
some basic understanding of the theory and formulations behind chemically reacting
flow simulations. The CHEMKIN Theory Manual provides a broad overview of the
relationships and formulations used in calculations of chemical property and source
terms. It also provides brief derivations and explanations of the governing equations
solved by CHEMKIN Reactor Models, as well as a discussion of numerical solution
techniques and sensitivity analysis employed in the models. In most cases these
descriptions contain references to more rigorous treatments of the theoretical
background that are available in the open literature. For further background
information, we recommend a text book written by several of CHEMKINs original
authors: Chemically Reacting Flow: Theory and Practice.1
1.1
RD01400-C01-004-001A
13
CHEMKIN 4.0
Chapter 1: Introduction
In actuality there is no constraint that the surface must be only one atom thick. However,
defining a surface that is several monolayers thick may be conceptually much more difficult
to deal with.
Beyond the basic gas, surface, and bulk, we also use the notion of different physical
phases to group the chemical species in a problem. For example, there can be more
than one type of site on the surface. This may be used to specify that a surface
consists of ledge sites and plane sites, where the number of sites of each type
might be a characteristic of the solid crystal face. Alternatively, a particular site type
may consist of active sites vs. non-active sites, where different reaction rates are
defined on each type. There can be any number of site types. The user may define a
species that only resides on a certain type of site. For example, the thermodynamic
properties of a hydrogen atom on a ledge site might be different from a hydrogen on a
plane site, and they could be specified as different species (even though their
elemental composition is the same). The population of different species occupying a
given type of site is specified by site fractions. The sum of the site fractions of the
species on a given site is 1. (Thus an open site'' is considered as a distinct species.)
This general formalism provides the user extensive flexibility in defining surface
chemistry problems, whether they involve material deposition, surface catalysis, or
simple radical recombination.
In the bulk there can also be different types of bulk species. The simplest bulk phase
consists of a pure (single) species. There can be any number of pure bulk phases. It
is also possible to specify a bulk mixture with components A and B, for example. The
composition of the bulk phase may then be specified using the activities of each of
the bulk-phase components.
For gas-phase and surface site species, the mole fractions of the species correspond directly
to the species activities. In general this is not true for condensed-phase mixtures unless the
bulk phase is either an ideal mixture or a pure bulk species.
14
RD01400-C01-004-001A
Chapter 1: Introduction
Theory Manual
When defining problems in CHEMKIN, a user may define a chemistry set that is gasphase only. In all cases one or more gas-phase species must be defined, but
problems may be defined that include no gas-phase reactions. When surface
chemistry is included, it may include only surface species, only bulk species, or both
surface site and bulk species. Reactions may involve any combination of species from
different phases, as long as conservation rules are obeyed.
1.2
phase
In most cases, users do not need to be concerned with the internal indexing of
species within CHEMKIN. However, users who wish to create their own CHEMKIN-based
programs will need to be aware of the use of species and phase arrays in passing
information back and forth through the CHEMKIN/API. Also, these conventions are
used extensively in the mathematical descriptions of our governing equations and
rate formulations, so some familiarity is helpful in following the discussions here.
RD01400-C01-004-001A
15
CHEMKIN 4.0
Chapter 1: Introduction
16
RD01400-C01-004-001A
CHEMKIN 4.0
Thermodynamic Expressions
In defining the chemistry used in a CHEMKIN reactor simulation, users must first supply
the thermodynamic data for each species in the chemical system. These data are in
the form of polynomial fits to temperature, for species enthalpy, entropy, and specific
heat capacity. Sources of this data, as well as the utility that generates the needed
polynomial-fit parameters, are described in detail in the Getting Started with CHEMKIN.
Once these data are defined, they are used during a CHEMKIN simulation to determine
species thermodynamic properties, thermal transport properties, and reaction
equilibrium constants. This chapter describes the mathematical relationships,
formulations, and conversions used in such calculations.
These formulas are used internally within CHEMKIN program executables. In addition,
the CHEMKIN/API, allows users to call CHEMKIN subroutines to perform many of the
calculations described within their own programs. Such calls can be made from a
users C, C++, or FORTRAN program or to create interfaces to 3rd-party programs.
Details on the API for such calls are provided in the CHEMKIN Application
Programming Interface Manual. However, to aid in programming and formulation of a
problem, the descriptions of formulas and conversions in the following section contain
references to the CHEMKIN/API, giving the name of the subroutine that performs the
particular calculation.
Species can exist in the gas phase, on surface sites, or in bulk mixtures. In some
cases it is desirable to refer to information about species without regard to the
phases, and in other cases it is desirable to determine information about species in
one particular phase or group of phases. Therefore, before discussing surface state
variables, it is helpful to first introduce a nomenclature that facilitates the
mathematical expression of species information.
RD01400-C01-004-001A
17
CHEMKIN 4.0
2.1
State Variables
The formulation of any chemically reacting-flow problem requires selection of a set of
state variables. State variables define the thermodynamic and chemical state of the
fluid mixture. When surface chemistry is considered, the state variables may also
include the chemical state of a solid material and/or of the surface or interace
between the fluid and the solid.
2.1.1
P Tk Yk
Xk
[ Xk ]
In making these options available from among the many possible, we provide
combinations of variables that are natural ones for a wide class of chemically
reacting-flow problems. For example, pressure is a natural choice in situations where
pressure is fixed, and density is a natural variable where volume is fixed. Moreover,
density is a natural variable in many problems involving fluid mechanics because it is
determined directly from the mass continuity equation. Temperature is always taken
as a natural variable because the thermodynamic properties and the chemical rate
constants both depend directly on temperature. Mass fraction and mole fraction are
convenient variables for describing the composition of a gas. Molar concentration
( [ X k ] , in moles/cm3) is sometimes a convenient variable because the rate of progress
of chemical reactions depends directly on the molar concentration of the reactants
and products. In general, most CHEMKIN Reactor Models use species mass fractions
to track variations in chemical composition, because of the direct relation to mass
conservation principles. The formulas (and their related subroutines) in the following
sections show the calculations necessary to convert between state variables, which
are used in assembling reaction rates-of-production, thermodynamic property, and
other terms in the reactor governing equations.
18
RD01400-C01-004-001A
2.1.2
Theory Manual
Equation 2-1
l
K s( n)
f
l
( n = N s, , N s )
Zk ( n ) = 1
k = K s( n)
The sum in Equation 2-1 runs from the first species in phase n to the last species in
phase n . The surface molar concentration of a species is then
Equation 2-2
[ Xk ] = Zk ( n ) n k ( n )
where n is the density of sites of phase n (in moles/cm2) and k ( n ) is the number of
sites that each species k occupies. Note that Equation 2-2 assumes that the surface
site density is constant as a function of time. It is possible to override this requirement
in the surface chemistry reaction set, in which case use of this equation should
ensure that the correct value of n ( t ) is used. Non-conservation of sites is discussed
further in Section 4.11.
For the sake of parallelism, we adopt the nomenclature for bulk species:
Equation 2-3
[ Xk ] = ak
( k = K b, , K b )
SURFACE KINETICS takes the approach that the activity, a k , of bulk species k is used in
all chemical rate expressions. In the limiting case of an ideal solution, the activity of a
species is equal to its mole fraction. However, SURFACE KINETICS does not explicitly
evaluate the relationship between bulk mole fraction and the bulk activities. Instead, it
is up to the executable to specify the relationship between the two. In most existing
RD01400-C01-004-001A
19
CHEMKIN 4.0
CHEMKIN executables, ideal solutions are assumed when bulk mixtures are present,
although it is recognized that this approximation is often not a good one. Typically bulk
activities become dependent variables in the reacting-flow problem only if there are
more than one bulk species in any one bulk phase.
2.2
Equation 2-4
P =
[ Xk ]RTk
k=1
[ Xk ]Wk
k=1
CKMMWY
1
W = --------------------------------K
Yk Wk
k=1
or
Equation 2-7
CKMMWX
K
W =
Xk Wk
k=1
20
RD01400-C01-004-001A
Theory Manual
or
Equation 2-8
CKMMWC
K
k = 1 [ X k ] W k
W = ----------------------------------K
[ Xk ]
k=1
2.2.1
Equation 2-9
CKYTX
Yk
Yk W
X k = -------------------------------------- = ---------K
Wk
Wk
Yj Wj
j=1
2.2.2
Equation 2-10
CKYTCP
P ( Yk Wk )
[ X k ] = ---------------------------------------K
R
Yj Tj Wj
j=1
Equation 2-11
CKYTCR
Yk
[ X k ] = ------Wk
2.2.3
Equation 2-12
CKXTY
Xk Wk
Xk Wk
- = -----------Y k = --------------------------K
W
Xj Wj
j=1
RD01400-C01-004-001A
21
CHEMKIN 4.0
2.2.4
Equation 2-13
CKXTCP
P
[ X k ] = X k ---------------------R Xk Tk
Equation 2-14
CKXTCR
[ X k ] = X k ----W
2.2.5
Equation 2-15
CKCTY
[ X k ]W k
Y k = -------------------------------K
[ Xj ]Wj
j=1
2.2.6
Equation 2-16
CKCTX
[ Xk ]
X k = ------------------------K
[ Xj ]
j=1
2.3
22
RD01400-C01-004-001A
2.3.1
Theory Manual
Equation 2-17
M
C pk
-------- =
R
a mk T k
(m 1)
m=1
The superscript o refers to the standard-state. For gas-phase species, the standard
state is an ideal gas at 1 atmosphere. For perfect gases, however, the heat capacities
are independent of pressure, and the standard-state values become the actual
values.
For surface species the standard state of species k refers to the case of a chemical
potential for a surface of pure species k (i.e., Z k 1 ) with a fixed standard state site
o
density n . Moreover, a perfect solution (i.e., non-interacting) is assumed for the
surface phase, which is independent of the system pressure. Under these
assumptions the chemical potential for surface species k on surface site n may be
written as
Equation 2-18
o
k ( T, P, Z ) = k ( T ) + RT ln ( n Z k n )
The activity of a bulk species is defined in terms of the following equation for the
chemical potential:
Equation 2-19
o
k ( T, P, X ) = k ( T ) + RT ln ( a k ( T, P, X ) )
o
RD01400-C01-004-001A
23
CHEMKIN 4.0
Both conventions for the standard state work with SURFACE KINETICS, as do any other
o
definitions that conform to the formalism expressed by Equation 2-19 for . k ( T ) is
specified through the entry for species k in the thermodynamics data file. The value of
a k ( T, P, X ) is required as input to all SURFACE KINETICS subroutines that calculate bulk
phase thermodynamic quantities and reaction rates. Therefore, if desired, advanced
users can construct their own subroutines to calculate a k ( T, P, X ) , possibly
incorporating models for non-ideality of the bulk phase, and can have the
consequences properly incorporated into the surface kinetics mechanism. Although
the activities of all components of an ideal solution must sum to 1, this condition is not
enforced in SURFACE KINETICS. (It is, however, enforced in many of the CHEMKIN
program executables that employ SURFACE KINETICS.)
Other thermodynamic properties are given in terms of integrals of the molar heat
capacities.
2.3.2
Equation 2-20
o
Hk =
Tk
0 Cpk dT + Hk ( 0 )
so that
Equation 2-21
o
Hk
-------- =
RT k
m=1
(m 1)
a mk T k
a M + 1, k
---------------------------- + ----------------m
Tk
2.3.3
Equation 2-22
o
Sk
Cp k
o
= --------- dT + S k ( 0 )
298 T
Tk
24
RD01400-C01-004-001A
Theory Manual
so that
Equation 2-23
o
S
----k- = a 1k ln T k +
R
m=2
(m 1)
a mk T k
---------------------------- + a M + 2, k
(m 1)
2.3.4
Equation 2-24
CKCPOR, SKCPOR
o
C pk
2
3
4
-------- = a 1k + a 2k T k + a 3k T k + a 4k T k + a 5k T k
R
Equation 2-25
CKHORT, SKHORT
o
a 2k
a 3k 2 a 4k 3 a 5k 4 a 6k
Hk
--------- = a 1k + ------- T k + ------- T k + ------- T k + ------- T k + -----2
3
4
5
RT k
Tk
Equation 2-26
CKSOR, SKSOR
o
Sk
----- = a 1k ln T k
a 3k 2 a 4k 3 a 5k 4
+ a 2k T k + ------- T k + ------- T k + ------- T k + a 7k
2
3
4
The GAS-PHASE and SURFACE KINETICS Pre-processors allows for additional temperature
ranges. Details of the thermodynamic data format options are included in the .
3.S. Gordon and B. J. McBride, Computer Program for Calculation of Complex Chemical Equilibrium Compositions, Rocket Performance, Incident and Reflected Shocks and Chapman-Jouguet
Detonations, NASA Report SP-273, 1971.
RD01400-C01-004-001A
25
CHEMKIN 4.0
2.3.5
For gas-phase only, the specific heat capacity at constant volume C vk is:
Equation 2-27
CKCVML
o
C vk =
C pk R
o
CKUML, SKHML
o
o
U k = H k RT k
o
CKGML, SKGML
o
o
o
Gk = Hk Tk Sk
o
CKAML, SKAML
o
Ak = Uk Tk Sk
2.3.6
Equation 2-31
CKCPMS, SKCPMS
o
C pk
o
c pk = -------Wk
Equation 2-32
CKHMS, SKHMS
o
Hk
o
h k = ------Wk
Equation 2-33
CKSMS, SKSMS
o
Sk
o
s k = ------Wk
26
RD01400-C01-004-001A
Equation 2-34
CKCVMS
o
c vk
Equation 2-35
C vk
= ------Wk
CKUMS, SKUMS
o
uk
Equation 2-36
Theory Manual
U
= ------kWk
CKGMS, SKGMS
o
Gk
o
g k = ------Wk
and
Equation 2-37
CKCAMS, SKAMS
o
A
o
a k = ------kWk
2.3.7
Equation 2-38
CKCPBL
K
Cp =
Cpk Xk
k=1
Equation 2-39
CKCPBS
K
cp =
cpk Yk
= Cp W
k=1
RD01400-C01-004-001A
27
CHEMKIN 4.0
Equation 2-40
CKCVBL
K
Cvk Xk
Cv =
k=1
and
Equation 2-41
CKCVBS
K
cvk Yk
cv =
= Cv W
k=1
CKHBML
K
Hk Xk
H =
k=1
and
Equation 2-43
CKHBMS
K
hk Yk
h =
= HW
k=1
CKUBML
K
U =
Uk Xk
k=1
and
Equation 2-45
CKUBMS
K
u =
uk Yk
= UW
k=1
28
RD01400-C01-004-001A
Theory Manual
The mixture properties are more complex for the entropies and the Gibbs and
Helmholtz free energies. Here the actual values are not the same as the standardstate values and we must account for the appropriate pressure and entropy-of-mixing
terms. The entropy is then
Equation 2-46
o
S k = S k R ln X k R ln ( P P atm )
Thus the mixture-averaged entropies are
Equation 2-47
CKSBML
K
o
( Sk R ln Xk R ln ( P Patm ) )Xk
S =
k=1
and
Equation 2-48
CKSBMS
s = SW
Similarly, the mixture-averaged Gibbs and Helmholtz free energies are
Equation 2-49
CKGBML
K
o
[ Hk Tk ( Sk R ln Xk R ln ( P Patm ) ) ]Xk
G =
k=1
Equation 2-50
CKGBMS
g = GW
Equation 2-51
CKABML
K
o
[ Uk Tk ( Sk R ln Xk R ln ( P Patm ) ) ]Xk
A =
k=1
and
Equation 2-52
CKABMS
a = AW
RD01400-C01-004-001A
29
CHEMKIN 4.0
2.3.8
30
RD01400-C01-004-001A
CHEMKIN 4.0
3.1
Equation 3-1
K
ki k ki k
k=1
RD01400-C01-004-001A
( i = 1 ,, I )
k=1
31
CHEMKIN 4.0
Global reaction are sometimes stated with non-integer stoichiometric coefficients. CHEMKIN
can accommodate non-integer stoichiometric coefficients.
The production rate k of the kth species can be written as a summation of the rateof-progress variables for all reactions involving the kth species
Equation 3-2
k =
ki qi
( k = 1 ,, K )
i=1
where
Equation 3-3
CKNU
ki = ki ki
The rate of progress variable q i for the i th reaction is given by the difference of the
forward and reverse rates as
Equation 3-4
q i = k fi [ X k ]
k=1
ki
k ri [ X k ]
ki
k=1
where X k is the molar concentration of the kth species and k fi and k ri are the
forward and reverse rate constants of the i th reaction. As indicated in Equation 3-4,
the rate-of-progress of a reaction is evaluated, by default, using the concentration of
each reactant or product species raised to the power of its stoichiometric coefficient.
Thus, the rate-of-progress of a reaction that includes species A with a coefficient of 2
will be second-order with respect to the concentration of A . Equation 3-4 is always
valid when mass-action kinetics are obeyed, and when the mechanism is written in
32
RD01400-C01-004-001A
Theory Manual
Ei
k fi = A i T i exp ---------
R c T
where the pre-exponential factor A i , the temperature exponent i , and the activation
energy E i are specified. These three parameters are required input to the GAS-PHASE
KINETICS package for each reaction.
Two gas constants, R and R c are used throughout this report and within CHEMKIN programs.
R c is used only in conjunction with the activation energy E i and has compatible units. The
reason for the duality is that many users would rather use units of cal/mole for the activation
energies even though other energy units are used elsewhere.
In Equation 3-5 thru Equation 3-10, T refers to the gas temperature, unless auxiliary
reaction information is provided to indicate that the reaction depends on a
temperature associated with a particular species. Such information would be specified
using the auxiliary keyword, TDEP, which is described further in Section 3.4.2.8 of the
CHEMKIN Input Manual. In the case where the TDEP keyword is included for reaction i ,
T represents the temperature of the species whose name follows the TDEP keyword.
In thermal systems, the reverse rate constants k ri are related to the forward rate
constants through the equilibrium constants by
Equation 3-6
k fi
k ri = -----K ci
RD01400-C01-004-001A
33
CHEMKIN 4.0
Although K ci is given in concentration units, the equilibrium constants are more easily
determined from the thermodynamic properties in pressure units; they are related by
Equation 3-7
ki
P atm k = 1
= K pi ----------
RT
K ci
S H
K pi = exp --------i- ----------i
RT
R
The refers to the change that occurs in passing completely from reactants to
products in the i th reaction; specifically,
Equation 3-9
o
Si
--------- =
R
S
ki ----kR
k=1
Equation 3-10
o
---------- =
RT
Hi
H
ki ------kRT
k=1
34
RD01400-C01-004-001A
3.2
Theory Manual
Equation 3-11
3.3
RD01400-C01-004-001A
35
CHEMKIN 4.0
product in the reaction) raised to any specified power. To modify the reaction order for
the reaction in the forward or reverse direction, the user must declare the FORD or RORD
auxiliary keywords, respectively, in the Pre-processor input file. These keywords are
discussed in Reaction Order Parameters, Section 3.4.2.12 of the CHEMKIN Input
Manual. These options are available both for gas-phase and surface reactions.
When the reaction-order-dependence of reaction i is changed via the FORD or RORD
keywords, the rate-of-progress variable q i for the reaction is evaluated by
Equation 3-13
K
q i = k fi [ X k ]
F ki
k ri [ X k ]
k=1
R ki
k=1
where F ki is the reaction order specified through the FORD keyword and R ki is the
reaction order specified through the RORD keyword for species k . The default for
species participating in reaction i is the normal mass-action kinetics values
Equation 3-14
F ki = ki
and
Equation 3-15
R ki = ki
if an order-change parameter is not given for species k .
The user should exercise caution when specifying a change of reaction order, as such
a change may produce unexpected and unphysical results in a kinetic simulation. For
example, the user should consider the kinetics of the reverse reaction when changing
reaction-orders for the forward reaction. Such a reaction may no longer satisfy
microscopic reversibility. At equilibrium, elementary kinetics ensure that
Equation 3-16
K
k ri k fi =
[ Xk ]
k=1
ki
[ Xk ]
ki
[ Xk ]
ki ki
k=1
k=1
A reaction for which one has specified a change in reaction order will not have the
proper equilibrium behavior unless
36
RD01400-C01-004-001A
Theory Manual
Equation 3-17
F ki R ki = ki ki
( k = 1 ,, K )
The user specifying F ki may also wish to adjust R ki such that Equation 3-17 is
satisfied; GAS-PHASE KINETICS does not do this automatically. Another alternative
would be to simply specify that the reaction is irreversible. A user program can call
subroutine CKIORD for gas-phase reactions and SKIORD for surface reactions, to
determine if a reaction has user-specified orders and the values of those parameters.
3.4
Three-Body Reactions
In some reactions a third body is required for the reaction to proceed; this is often
the case in dissociation or recombination reactions, such as
H + O2 + M HO2 + M
When a third body is needed, the concentration of the effective third body must
appear in the expression for the rate-of-progress variable. Accordingly, the rate-ofprogress variable is different from Equation 3-4 by the first factor in the equation
Equation 3-18
ki
q i = ( a ki ) [ X k ] k fi [ X k ] k ri [ X k ] ki
k = 1
k = 1
k=1
If all species in the mixture contribute equally as third bodies, then a ki = 1 for all k ,
and the first factor is the total concentration of the mixture,
Equation 3-19
K
[M] =
[ Xk ]
k=1
However, it is often the case that some species act more efficiently as third bodies
than do others. The a ki that differ from 1 must be specified by auxiliary input to the
GAS-PHASE KINETICS Pre-processor, as described in Section 3.4.2.1 of the CHEMKIN
Input Manual.
RD01400-C01-004-001A
37
CHEMKIN 4.0
3.5
Pressure-Dependent Reactions
Under certain conditions, some reaction rate expressions depend on pressure as well
as temperature. GAS-PHASE KINETICS provides for two kinds of such reactions:
unimolecular/recombination fall-off reactions and chemically activated bimolecular
reactions. Generally speaking, the rate for unimolecular/recombination fall-off
reactions increases with increasing pressure, while the rate for chemically activated
bimolecular reactions decreases with increasing pressure. In both cases, GAS-PHASE
KINETICS makes available various expressions that blend smoothly between high- and
low-pressure limiting rate expressions.
3.5.1
38
RD01400-C01-004-001A
Theory Manual
Equation 3-20
k 0 = A 0 T exp ( E 0 R c T )
Equation 3-21
k = A T
exp ( E R c T )
Pr
k = k --------------- F
1 + P r
where the reduced pressure P r is given by
Equation 3-23
k0 [ M ]
P r = -------------k
and [ M ] is the concentration of the mixture, possibly including enhanced third-body
efficiencies.
It is also possible that the third body in the fall-off region could be a specific species rather
than the mixture as a whole. In such a case, the reaction could be written, for example, as
CH3 + CH3 (+N2) C2H6 (+N2). In this case, the concentration of Nitrogen [N2] would replace
the total concentration in the mixture [ M ] in these equations.
For this example, note that the units for k are 1/sec, k 0 are cm3/(mole sec), and k
are 1/sec. If the F in Equation 3-10 is unity, then this is the Lindemann form. The
other descriptions involve more complex expressions for the function F .
In the Troe form,6 F is given by
Equation 3-24
log P r + c
log F = 1 + ---------------------------------------n d ( log P r + c )
2 1
log F cent
6.R. G. Gilbert, K. Luther, and J. Troe, Ber. Bunsenges. Phys. Chem. 87:169 (1983).
RD01400-C01-004-001A
39
CHEMKIN 4.0
Equation 3-25
d = 0.14
and
Equation 3-28
F cent = ( 1 )exp ( T T
***
***
) + exp ( T T ) + exp ( T
*
**
T)
**
b
T
F = d aexp ------ + exp ------
T
c
X e
where
Equation 3-30
1
X = ----------------------------2
1 + ( logP r )
7.P. H. Stewart, C. W. Larson, and D. M. Golden, Combustion and Flame 75:25 (1989).
40
RD01400-C01-004-001A
Theory Manual
In addition to the six Arrhenius parametersthree each for the low-pressure limit k 0
and high-pressure limit k expressionsthe user must supply the parameters a , b ,
and c in the F expression. The parameters d and e were not discussed by Stewart,
et al., but we have included them as additional optional parameters to increase
flexibility. If one wishes, d and e can be considered parameters that define a weakcollision efficiency factor, in the event that one wants to compute strong-collision rate
parameters and correct them with such a factor.
Figure 3-1
Rate constant as a function of pressure at fixed temperature for a unimolecular fall-off reaction
Figure 3-1 illustrates the pressure dependence of rate expressions for the example
reaction, CH3 + CH3(+ M) C2H6(+ M), evaluated at a fixed temperature of 1000 K.
Both the Lindemann and the Troe forms are shown, as well as the low- and highpressure limits. The specific constants in fits to the Troe form ( A 0 = 1.135E36 ,
0 = 5.246 , E 0 = 1704.8cal mole , A = 6.22E16 , 1.174 ,
***
*
= 1120K , T = 69.6K ) are taken from
E = 653.8cal mole , = 0.405 , T
Wagner and Wardlaw.8 For the relatively simple Lindemann case ( F = 1 ), the limiting
behavior is apparent. In the low-pressure limit, [ M ] 0 , the denominator in
Equation 3-22 approaches unity and the rate expression becomes k k 0 [ M ] . In the
high-pressure limit, [ M ] , the pressure-ratio factor approaches one, and the rate
expression becomes k k , i.e., a constant. For both the Troe and SRI forms, F
approaches unity for both high and low pressures. Thus, all expressions recover the
correct limiting behavior.
RD01400-C01-004-001A
41
CHEMKIN 4.0
3.5.2
Equation 3-31
1
k = k 0 --------------- F
1 + Pr
where 1 ( 1 + P r ) is analogous to the Lindemann form of Equation 3-22. Note that in
Equation 3-31, k 0 is the pressure-independent factor, whereas in Equation 3-22 it is
k . The three choices for the F function are exactly the same as for the unimolecular
fall-off reactions, i.e., the Lindemann ( F = 1 ), Troe, or SRI forms.
Figure 3-2
42
RD01400-C01-004-001A
Theory Manual
Figure 3-3 illustrates the rate-expression behavior for the example chemically
activated reaction, CH3 + CH3(+ M) C2H5 + H(+ M). Both the Lindemann and the
SRI formulations are shown, as well as the high- and low-pressure limiting cases. The
12.698
specific constants for the SRI form ( A 0 = 10
, 0 = 0.0099 ,
6.42
E 0 = 10600cal mole , A = 10
, = 4.838 , E = 7710cal mole , a = 1641 ,
b = 4334 , c = 2725 ) are taken from Stewart, Larson, and Golden.7, p. 40 For this
example, note that the units for k are cm3/(mole sec), k 0 are cm3/(mole sec), and
k are 1/sec. The limiting cases are recognized easily from the behavior of
Equation 3-31. In the low-pressure limit, [ M ] 0 , P r 1 , causing the pressure-ratio
factor in Equation 3-31 to approach unity. Hence, k k 0 , i.e., a pressureindependent function. In the high-pressure limit, [ M ] , P r 1 and
k k0 Pr k [ M ] .
Figure 3-3
Rate constant as a function of pressure at fixed temperature for a chemically activated reaction
3.5.3
RD01400-C01-004-001A
43
CHEMKIN 4.0
logP logP i
logk = logk i + ( logk i + l logk i ) -------------------------------------logP i + l logP i
If the rate of the reaction is desired for a pressure lower than any of those provided,
the rate parameters provided for the lowest pressure are used. Likewise, if rate of the
reaction is desired for a pressure higher than any of those provided, the rate
parameters provided for the highest pressure are used.
This logarithmic interpolation method can be used as an alternative approach to
describing any type of pressure dependence, including the multiple-well, multiplechannel reactions discussed in Section 3.5.4. It has the advantage of being
conceptually straightforward to implement. However, the resolution or accuracy of the
pressure dependence will depend on the number of pressure points included for each
reaction.
44
RD01400-C01-004-001A
3.5.4
Theory Manual
Instead of using the modified Arrhenius form for the rate coefficient, the Chebyshev
expansions approximate the logarithm of the rate coefficient directly as a truncated
bivariate Chebyshev series in the reverse temperature and logarithm of the pressure.
Since the Chebyshev polynomials are only defined in the interval of [-1, +1], the
temperature and pressure boundaries for the approximation must be established first,
i.e.,
T min T T max
and
P min P P max
RD01400-C01-004-001A
45
CHEMKIN 4.0
2T T min T max
T = -------------------------------------------1
1
T min T max
and
Equation 3-34
log k (T ,P )
a nm n ( T ) m ( P )
n = 1m = 1
where the Chebyshev polynomials of the first kind of degree n 1 are given by
Equation 3-36
1
( x ) = cos ( ( n 1 )cos ( x ) )
n = 1, 2,
with 1 x +1 . The integers N and M denote the number of basis functions along
the temperature and the pressure axis, respectively. The accuracy of the
approximates will increase monotonically with N and M . The N M coefficients,
a nm , of the Chebyshev expansions are determined from a least-squares fit to a set of
rate coefficient data points, k (T ,P ) , computed from a detailed theory such as the RiceRamsperger-Kassel-Marcus (RRKM) theory. To ensure the approximation is uniform
over the desired domain, the computed rate coefficient data must be on the GaussChebyshev grid. For example, if the computed rate coefficient data is on a GaussChebyshev grid, its coordinates, T i and P j , must be the roots of a high-order
Chebyshev polynomial of the first kind given by
Equation 3-37
2i 1
T i = cos --------------
2d T
46
RD01400-C01-004-001A
Theory Manual
Equation 3-38
2j 1
P j = cos --------------
2d P
where 1 i d T and 1 j d P . The integers d T and d P are the resolutions of the
computed rate coefficient data in the temperature and the pressure direction,
respectively. As a necessary condition, the number of basis functions in each
direction of the Chebyshev expansions should be no greater than the data resolution
of that direction, i.e., N d T and M d P .
To use the Chebyshev expansions to represent the rate coefficient of a reaction, the
user should provide the temperature and pressure limits of the expansion, T max ,
T min , (auxiliary keyword TCHEB) and P max , P min (auxilary keyword PCHEB) the number
of basis functions used in each direction, N and M , and the N M coefficients, a nm ,
(auxiliary keyword CHEB). See Section 3.4.2.3 and Figure 3-4 of the CHEMKIN Input
Manual for more information about the use of these keywords.
3.6
Equation 3-39
Bi
Ci
k fi = A i exp ---------- + ----------
T 1 3 T 2 3
In GAS-PHASE KINETICS, we have provided the possibility to blend the Arrhenius
expression with the Landau-Teller expression in the general expression, as follows:
Equation 3-40
E
Ci
Bi
i
- + ----------
k fi = A i T exp ---------i + --------- R c T T 1 3 T 2 3
Clearly, by setting B i and C i to zero, the Arrhenius expression is recovered, and by
setting i and E i to zero, the standard Landau-Teller expression is recovered. If
appropriate, however, all the parameters can be used together to provide more
flexibility in the reaction-rate expression than could be afforded by one of the forms
alone.
RD01400-C01-004-001A
47
CHEMKIN 4.0
3.7
Equation 3-41
9
E
n 1
i
k fi = A i T exp ----- + b ni ( ln T )
T
n=1
This form is consistent with the rate-constant fits for electron-hydrogen and electronhelium processes in a publication by Janev, Langer, Evans, and Post,13 when the
Arrhenius parameters i and E i are zero. The user can specify this rate-constant fit
option using the auxiliary keyword JAN, followed by the nine b ni parameters.
A second form introduces a power series within the exponential of a modified
Arrhenius expression, as follows:
Equation 3-42
k fi = A i T exp
b ni
-----n
T
n=1
The user may specify this rate-constant expression using the auxiliary keyword FIT1,
followed by the four b ni parameters. The use of auxiliary keywords are described in
more detail in the Section 3.4.2.6 of the CHEMKIN Input Manual.
3.8
13.R. K. Janev, W. D. Langer, J. K. Evans, and J. D. E. Post, Elementary Processes in HydrogenHelium Plasmas, Springer-Verlag, New York, 1987.
48
RD01400-C01-004-001A
Equation 3-43
Theory Manual
k = C k D k
where, for non-three-body reactions, the creation rate is
Equation 3-44
I
C k =
ki kri [ Xj ]
i=1
ji
j=1
ki kfi [ Xj ]
i=1
j=1
ji
D k =
ki kfi [ Xj ]
i=1
j=1
ji
ki kri [ Xj ]
i=1
ji
j=1
When third body reactions are involved, each sum in the above equations is multiplied
by the third-body concentration
K
[M] =
ki [ Xk ]
k=1
Another useful form for the chemical production rates is found by defining a creation
rate and characteristic time for the destruction rate, i.e.,
Equation 3-46
[ Xk ]
k = C k --------k
Here the characteristic time for destruction is given in terms of D k as
Equation 3-47
[X ]
k
k = -------- -
Dk
As a precaution against [ X k ] and D k simultaneously approaching zero, the GASPHASE KINETICS implementation of the destruction time is written as
RD01400-C01-004-001A
49
CHEMKIN 4.0
Equation 3-48
[X ]
k
k = -------------
Dk +
30
The computer-dependent small number used in Equation 3-48 is set in the GAS-PHASE
KINETICS Subroutine Library at the time that the library is created.
3.9
50
RD01400-C01-004-001A
CHEMKIN 4.0
4.1
RD01400-C01-004-001A
51
CHEMKIN 4.0
A chemical species on the top layer of the solid, i.e., a surface species, occupies a
site. For example, an arsine molecule adsorbed on a surface could occupy a site, and
might be denoted AsH3(s). Another example might be a bare gallium atom, Ga(s), on
top of a gallium arsenide (bulk) crystal. If another species, say a gas-phase AsH3(g),
lands on top of the Ga(s), it might stick or adsorb on that site, as shown in
Figure 4-4. In this case the gallium atom that was at the surface is now covered up,
such that it is no longer accessible to react with the gas and therefore no longer a
surface species. In our nomenclature it has become a bulk species. The adsorbed
AsH3 molecule now occupies the top-most layer at this site, so it has become the
surface species AsH3(s). In our formalism, we might write the adsorption reaction in
Figure 4-4 as
Figure 4-1
52
RD01400-C01-004-001A
Theory Manual
The Atomic Site and Open Site Formalisms are equally valid ways of stating these
surface reactions. Either is allowed by the SURFACE KINETICS Pre-processor. Personal
preference or, perhaps, the nature of a particular problem might dictate one over the
other. Note that an open site must be considered as a species and therefore must
have thermodynamic data specified (even if all the data for the species are zeroes).
Figure 4-4
Figure 4-5
RD01400-C01-004-001A
53
CHEMKIN 4.0
4.2
54
RD01400-C01-004-001A
Equation 4-2
Theory Manual
SKRAT
I
s k =
ki qi
( k = 1, , K )
i=1
RD01400-C01-004-001A
55
CHEMKIN 4.0
4.3
Equation 4-3
SKEQ
l
K s(n)
Kg
K ci
P atm k = 1
= ----------
RT
ki
Ns
o
( n)
f
k = K s(n)
ki
K s( n)
k ki K pi
k = K s(n)
n=Ns
where P atm denotes a pressure of 1 atm, and n is the standard-state surface site
density of site type n . The sum in the first exponent runs only over the gas-phase
species, and the sum in the second exponent runs only over surface species in
surface phase n . The refers to the change that occurs in passing completely from
reactants to products in the i th reaction.
4.4
56
RD01400-C01-004-001A
4.5
Theory Manual
Equation 4-4
l
Ei
k fi = A i T i exp --------
RT
K s(N s)
10
ki [ Zk ( n ) ]
k = K s(N s)
l
ki [ Z k ( n ) ]
[ Z k ( n ) ] ki exp ----------------------------
RT
K b(N b)
10
ki [ a k ( n ) ]
k = K b(N b)
ki [ a k ( n ) ]
-
[ a k ( n ) ] ki exp --------------------------RT
where the three coverage parameters are ki , ki , and ki for species k and
reaction i . The product in Equation 4-4 runs over only those surface species that are
specified as contributing to the coverage modification. Note that the surface site
fractions appear in Equation 4-4 rather than molar concentrations [ X k ] (mole/cm2) for
surface species, while bulk activities appear for bulk species. The term associated
with ki now makes it possible for the rate-of-progress of a reaction to be proportional
to any arbitrary power of a surface species concentration. Also, using this modified
expression for k fi , the net pre-exponential factor may be a function of coverage
Equation 4-5
l
K s( N s)
log 10 A = log 10 A i +
K b(N b)
ki [ Z k ( n ) ] +
f
k = K s(N s)
ki [ a k ( n ) ]
f
k = K b( N b)
Ks ( Ns )
E = Ei +
ki [ Z k ( n ) ] +
f
k = Ks ( Ns )
RD01400-C01-004-001A
Kb ( Nb )
ki [ a k ( n ) ]
f
k = Kb ( Nb )
57
CHEMKIN 4.0
For reactions with optional coverage dependence, the rate of progress is calculated
employing Equation 4-1, with the forward rate coefficient from Equation 4-4. The
reverse rate constant is calculated via Equation 4-5.
If the form of Equation 4-4 is not flexible enough to describe a certain coverage
behavior, one can repeat the same reaction several times with different values for the
coverage parameters such that the sum of the rate constants approximates the
desired form.
4.6
Sticking Coefficients
For some simple surface reaction mechanisms we have found it convenient to specify
the surface reaction rate constant in terms of a sticking coefficient (probability),
rather than an actual reaction rate. This approach is only allowed when there is
exactly one gas-phase species reacting with a surface. Sticking-coefficient reactions
may include any number of surface site or bulk-phase species as reactants, and any
number of species of any phase as products
In such cases, one might have a measurement or intuition about the probability that a
certain process takes place when a collision between a given gas-species occurs with
the surface. For consistency in expressing each surface reaction in terms of a rate
constant, we provide a conversion between this sticking coefficient form and the usual
rate expression. The actual reaction rate (in moles/cm2/sec) is derived from this
probability together with the physical gas-surface collision frequencies, as discussed
further below.
The unitless sticking coefficients functional form has an Arrhenius-like form as
follows:
Equation 4-7
In this case, i , a i and b i are unitless and c i has units compatible with R c , the realgas constant used for reaction activation energies. SURFACE KINETICS also allows for
surface-coverage modification of a sticking coefficient, analogous to Equation 4-4.
58
RD01400-C01-004-001A
Theory Manual
Because i is defined as a probability, it must lie between 0 and 1 to make physical sense.
Therefore, SURFACE KINETICS checks the value of i , and an unphysical sticking coefficient
greater than 1 is changed to the value 1. Some earlier versions of SURFACE KINETICS did not
truncate the values at 1.
SiH2(g) Si(b) + H2
In this example, we have not explicitly included the surface in writing Figure 4-6.
A somewhat more detailed way of using the sticking-coefficient specification would be
to say that SiH2(g) reacts with probability i upon each collision with a bare surface
silicon atom, Si(s):
Figure 4-7
RD01400-C01-004-001A
59
CHEMKIN 4.0
' ji
RT =1
- ------------k fi = i j-----------------m 2W
k
( tot )
Here, R is the universal gas constant, W k is the molecular weight of the gas-phase
species, tot is the total surface site concentration summed over all surface phases
(number of moles of surface sites per unit area), and m is the sum of all the
stoichiometric coefficients of reactants that are surface species. The term involving
tot raised to the m power is needed to convert from the unitless sticking coefficient
form to units appropriate for a rate constant, and the term in the square root accounts
for the gas/surface collision frequency. In the third example given above, Figure 4-8,
the value of m is 2, because there are two surface sites appearing as reactants, i.e.,
Si(s) and C(s). The product term in Equation 4-8 is the product of the site-species
occupancies, raised to a power equal to the reaction order for that species, for all site
species that are reactants. Here, j is the number of sites that the surface species
occupies, and j is the reaction order for that species. The product term will be equal
to one when there are unity site occupancies for all of the surface species in the
reaction.
Implicit in the sticking coefficient description just presented is an assumption that the
sticking coefficient is relatively small, i.e., much less than one. In this case the
molecular motion in the vicinity of the solid surface is random and the collision
frequency of gas-phase species with the surface is not affected by the surface itself.
However, when the sticking coefficient is large, i.e., close to one, then the velocity
distribution becomes skewed. Species whose random motion carries them close to
the surface have a high probability of staying there, which causes a non-Maxwellian
velocity distribution that, in turn, alters the net species flux near the surface. Motz and
Wise14 analyzed this situation and suggested a correction factor that modified
Equation 4-8 to become:
60
RD01400-C01-004-001A
Theory Manual
Equation 4-9
' ji
i
RT =1
- ------------k fi = ------------------- j-----------------m 2W
1 i 2
k
( tot )
Goodwin and Gavillet15 have incorporated this effect in their analysis of chemical
vapor deposition of diamond films. However, in most cases sticking coefficients are
derived from empirical data rather than theory, in which case it is usually inappropriate
to apply the Motz-Wise correction. Users may turn this option on by including MWON in
the REACTION line of their SURFACE KINETICS input file.
Early versions of SURFACE KINETICS always applied Equation 4-9. Later versions allow
optional use of Equation 4-8 to relate the sticking coefficient to rate constants through use of
the keyword MWOFF on the REACTION line. Currently, the default usage is Equation 4-8, or
MWOFF. Users can invoke
Using the kinetic rate constant derived from the sticking probability, the rate-ofprogress is calculated using Equation 3-4, as usual. The sticking coefficient
specification is only allowed for the forward reaction. If the reaction is written as
reversible, the reverse reaction rate constant would be calculated from Equation 3-6
to assure microscopic reversibility.
4.7
RD01400-C01-004-001A
61
CHEMKIN 4.0
isotherm, which assumes that the adsorption sites on the surface are independent
from each other (single site adsorption), the sites are equivalent, and the surface
coverage decreases the number of sites available for adsorption only, but does not
alter the energetics of adsorption/desorption.
The following example of a LH reaction illustrates its features. Species A and B
coadsorb onto the surface, react to products C and D, which can then desorb. The
reaction between adsorbed A and adsorbed B is assumed to be rate-limiting and
irreversible, while the adsorption/desorption processes are assumed to be in
equilibrium. In the LH formulation, the elementary chemical reactions shown in
Figure 4-9 would be replaced by the single overall reaction shown in Figure 4-10,
which does not explicitly include any surface species.
Figure 4-9
A + O(s) A(s)
B + O(s) B(s)
A(s) + B(s) C(s) + D(s)
C(s) C + O(s)
D(s) D + O(s)
Figure 4-10
A+BC+D
The effects of surface-sites being blocked by various species are included via the
adsorption/desorption equilibria. This "lumping" of a number of elementary steps
together results in a rate expression that differs substantially from a simple massaction rate expression. The rate of progress variable is given by:
Equation 4-10
kK A [ X A ]K B [ X B ]
q = --------------------------------------------------------------------------------------------------------------2
( 1 + KA [ XA ] + KB [ XB ] + KC [ XC ] + KD [ XD ] )
where the K s are the equilibrium constants for the adsorption/desorption steps and
[ X i ] s are the concentrations of the species. As product species, C and D do not
appear in the numerator, but as adsorbed species they can block surface sites, so
they do appear in the denominator. The k is expressed in terms of Arrhenius
parameters, as are the K s. The equilibrium constant is defined as
62
RD01400-C01-004-001A
Theory Manual
Equation 4-11
k [ X A ] [ X B ]
q = --------------------------------------------------------------------------------------------------------------2
( 1 + KA [ XA ] + KB [ XB ] + KC [ XC ] + KD [ XD ] )
The generalized form of the above expression is:
Equation 4-12
[ Xi ]
i
q = k --------------------------------------
2
1 + K i [ X ] i
where i represents gas-phase species in the reaction, and the exponent of 2 in the
denominator comes from the fact that the reaction rate is determined by the reaction
between two adsorbed species. In practice, this rate form is often used for empirical
parameter fitting, so we further generalize it to:
Equation 4-13
[ Xi ]
li
i
q = k ------------------------------------------m
n i
1 + Ki [ Xi ]
where:
1. the chemical species in the rate law are not limited to the reactants and
products in the reaction description,
2. the exponents ( l i ) for the chemical species concentrations in the numerator of
Equation 4-13 may differ from the stoichiometric coefficients ( i ) and may be
fractional,
3. the overall exponent in the denominator ( m ) of Equation 4-13 may differ from
2, and may be fractional,
4. the exponents ( n i ) for the concentrations of species in the denominator may
differ from 1 or the stoichiometric coefficients, and may be fractional.
RD01400-C01-004-001A
63
CHEMKIN 4.0
Eley-Rideal (also called Rideal-Eley) reactions are less common than LH reactions.
The following example illustrates its features. Species A adsorbs onto the surface,
then reacts with gas-phase species B to produce C, which can then desorb. The
reaction between adsorbed A and gas-phase B is assumed to be rate-limiting and
irreversible, while the adsorption/desorption processes are assumed to be in
equilibrium. In the ER formulation, the elementary chemical reactions shown in
Figure 4-11 would be replaced by the single overall reaction shown in Figure 4-12,
which does not explicitly include any surface species.
Figure 4-11
A + O(s) A(s)
A(s) + B(g) C(s)
C(s) C + O(s)
Figure 4-12
A+BC
In this case, the rate of progress variable is given by:
Equation 4-14
kK A [ X A ] [ X B ]
q = ---------------------------------------------------------( 1 + KA [ XA ] + KC [ XC ] )
or
Equation 4-15
k [ X A ] [ X B ]
q = ---------------------------------------------------------( 1 + KA [ XA ] + KC [ XC ] )
The generalized form of this is:
Equation 4-16
[ Xi ]
i
q = k -----------------------------------
1 + K i [ X ] i
which is the same as the Equation 4-12 above for LH kinetics, except that the
denominator has an overall exponent ( m ) of one rather than two. ER reactions are
thus treated as a special case of the LH rate law.
64
RD01400-C01-004-001A
Theory Manual
Using the LH option requires paying careful attention to the units of the reaction rates.
The discussion above assumes that the rate expressions are given in terms of gas
concentrations, which is the standard for GAS-PHASE KINETICS. However, literature
values for LH rate parameters, especially equilibrium constants, are often provided in
pressure units. To reduce the number of units conversions required of the user,
equilibrium constants may be input in either pressure units or concentration units. This
option is currently limited to the LH rate expression and only for the equilibrium
constants. Rate parameters still must be input in concentration units. In SURFACE
KINETICS, the default units, unless altered on the REACTION line, for the rate of a
reaction are moles cm-2sec-1. Rate parameters given in pressure units, for example in
atm sec-1, do not have the same dimensions as moles cm-2sec-1. Such a rate would
need to be divided both by RT and the surface-area to volume ratio ( AV ), before use.
Rates given in terms of weight of catalyst need to be converted to a rate expressed in
terms of the effective surface area of the catalyst via the surface area per unit weight
of catalyst and the dispersion. Rates given on a per site basis should also be
converted to a per area basis.
4.8
Plasma-Surface Interactions
For chemistry sets that involve electrons and ions, some special formulations of
surface reactions are available. These formulations allow modification of the ion flux
as well as the ion impact energy for ion-surface collisions.
Ion fluxes are often not determined by a thermal collision frequency (as used in our
sticking-coefficient formulations). Instead, ion fluxes are often limited by transport
from the plasma to the edge of the plasma sheath, which an electric field that exists
near surfaces adjacent to a plasma gas. For non-thermal plasmas, the limiting flux is
then defined by the Bohm criterion,16 which depends on the ion mass and the
electron temperature. CHEMKIN SURFACE KINETICS allows a user to indicate that
certain reactions are governed by the Bohm flux.
In addition, there are many examples in materials processing, where ions interact with
surfaces to alter the morphology, sputter material, or enhance heterogeneous
chemical reactions. Ions are often accelerated through a plasma sheath to grounded
or electrically biased materials. In this way, the directed energy of ions encountering a
surface may be significantly greater than that represented by the ion temperature in
the plasma gas. SURFACE KINETICS therefore makes the provision for a reaction-rate
constant to depend upon the energy of a positive ionic reactant species, E ion .
16.M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges and Materials Processing, John Wiley and Sons, New York (1994).
RD01400-C01-004-001A
65
CHEMKIN 4.0
Details of the ion flux and ion-energy options are described in this section.
4.8.1
Equation 4-17
' ji
b ci Rc T j = 1
------------------m
( tot )
k fi, Bohm = a i T i e
RT
----------e
W ion
In Equation 4-17, the reaction-rate expression is similar to that of Equation 4-8, used
to describe sticking coefficients. As with a sticking coefficient, Bohm reactions allow
the user to specify a probability for the reaction to occcur, rather than a reaction rate.
Also, the Bohm reaction formulation is only valid for reactions where there is exactly
one gas-phase ion that is a reactant. The units of the pre-exponential, temperature
exponent term, and activation energy correspond to the parameters in a sticking
coefficient. The main difference, however, is that the resulting probability is multiplied
by the Bohm velocity expression (the term in the square root in Equation 4-17), rather
than the thermal velocity of the ion species. The Bohm velocity is based on the
electron temperature, rather than the thermodynamic temperature of the gas.
For use of the Bohm reaction option, the electron must be declared as a gas-phase species in
the list of species names in the GAS-PHASE KINETICS Pre-processor input.
The molecular weight in the last term is that of the positive ion. Bohm reactions can be
declared through the Pre-processor input via the auxiliary keyword BOHM. An
executable can find out which reactions were declared as Bohm reactions by a call to
SKIBHM. Use of the BOHM keyword is only allowed for irreversible reactions.
4.8.2
Equation 4-18
f
i
i
k i ( E ion ) = k i ( thermal ) max 0, ( E ion
E ion
, 0)
66
gi
RD01400-C01-004-001A
Theory Manual
In this case, the reaction rate depends upon a threshold energy, E ion, 0 , and the
energy expressions can be raised to a specified power in two different ways through
the use of the parameters f i and g i . Ion-energy dependent reactions are declared in
the Pre-processor input via the auxiliary keyword ENRGDEP. An executable can find out
which reactions were declared as ion-energy-dependent reactions and get an array of
the parameters by a call to SKIENR. Because the subroutines that evaluate rate
constants in SURFACE KINETICS take temperature as an argument, and not species
energy, subroutine SKRPAR must be called to input an array of ion energies, ENRGI
before the rate constant routine is called. Use of the ENRGDEP keyword is only allowed
for irreversible reactions.
4.8.3
Figure 4-13
RD01400-C01-004-001A
67
CHEMKIN 4.0
consisting of all of the species in the original reaction that are multiplied by the
coefficient , must also satisfy mass, charge, and elemental balance. In addition,
unless the NONCON auxiliary keyword was declared on the REACTION line (described
later), the sub-reaction must also conserve the number of surface sites.
An example of a reaction using the ion-enhanced yield option in the form accepted by
the SURFACE KINETICS Pre-processor is
E + CL+ + #SICL3(S) + #SI(B) + SICL(S) => SICL2(S) + #SICL2 + #SICL(S)
The special character # identifies the energy-dependent multiplicative factor for the
stoichiometric coefficient. Notice that the sub-reaction consisting of every species
preceded by the # sign balances mass, elements, charge, and number of surface
sites. The yield of this reaction (per incident CL+ ion) depends upon the energy of the
ion.
We allow the following functional form for the yield enhancement:
Equation 4-19
t
i
i
( E ion ) = h yield max 0, ( E ion
E yield
, 0)
ui
The ion-enhanced yield can depend upon a threshold energy, E yield, 0 , and the
energy expressions can be raised to a specified power in two different ways through
the use of the parameters t i and u i . Ion-enhanced-yield reactions can be declared
through the Pre-processor input via the auxiliary keyword YIELD. An executable can
find out which reactions were declared as ion-enhanced yield reactions and get an
array of the parameters via a call to SKIYLD. Because the subroutines that evaluate
rate constants in SURFACE KINETICS take temperature as an argument, and not
species energy, subroutine SKRPAR must be called to input an array of ion energies,
ENRGI, before the rate constant routine is called. Use of the YIELD keyword is only
allowed for irreversible reactions.
4.9
68
RD01400-C01-004-001A
Theory Manual
s k [ X l ] l
ds
s
-------k- = -------k- + ------------- ------------- -------- [ X l ] l i
d i
i
l
where the components of are the mass fractions, site fractions, and activities for
gas-phase, surface, and bulk species, respectively. The term [ X l ] l converts
from concentration units to the units of :
Equation 4-21
2
P- W Y l W
f
l
------ ------------
Kg l Kg
RT ----2
Wl Wl
[ Xl ]
------------- =
f
f
l
l
l
n k ( n )
K s(N s) l K s(N s)
f
f
l
l
1
K b(N b) l K b(N b)
SKDRDA
s
-------k- = ki q i i
i
Note that subroutine SKDRDA calculates the derivative with respect to the Arrhenius preexponential if the reaction was originally stated in standard Arrhenius form, or with respect to
the sticking coefficient pre-exponential if a sticking coefficient was used.
RD01400-C01-004-001A
69
CHEMKIN 4.0
Equation 4-23
s k = ki q i
Therefore, the dependence of s k upon the concentration of some species l is
Equation 4-24
s k
f
------------- = ki q i
[ Xl ]
r
ki q i
li
li
k ( n )
---------------------- +
ln ( 10 ) + ---------- ---------
[ X l ] n li
[ Xl ] Rc T
' li
'' li k ( n )
li
li
--------- + -------------- li ln ( 10 ) + --------- --------
[ Xl ] n
[ Xl ] Rc T
The terms inside the curled braces will only be present if species l modifies the rate
of reaction i through coverage parameters, as in Equation 4-4. The partial of the
production rate of species k due to all reactions with respect to the concentration of
species l is then
Equation 4-25
SKDRDC
s k
------------=
[ Xl ]
i=1
s ki
------------ [ Xl ]
These terms can all be combined to calculate the desired s k i in the executable.
4.10
Equation 4-26
n [ Y k ( V k + u ) ] = s k W k
( k = 1, , K g )
where n is the unit inward-pointing normal vector to the surface, and the gas-phase
diffusion velocities are related to the species and temperature gradients by
70
RD01400-C01-004-001A
Theory Manual
Equation 4-27
Kg
Dk T
1
V k = ----------- W j D k, j X j --------- ------ Yk T
Xk W
( k = 1, , K g )
jk
Here the X k are the gas-phase mole fractions, the Y k are the gas-phase mass
fractions, W is the mean molecular weight, D k, j is the ordinary multicomponent
T
diffusion coefficient matrix, and the D k are the thermal diffusion coefficients. (Both
types of diffusion coefficients can be evaluated by calls to the TRANSPORT Package) In
nonreacting fluid mechanics the fluid velocity normal to a solid wall is zero. However,
if there are chemical reactions at the wall, then the velocity can be nonzero. This socalled Stefan flow velocity occurs when there is a net mass flux between the surface
and the gas. The induced Stefan velocity is given by
Equation 4-28
Kg
n u = --1-
s k Wk
k=1
The expression for the Stefan velocity is easily obtained from the interfacial mass
balance (Equation 4-26) by summing over all K species and noting that the mass
fractions must sum to one, i.e.,
Equation 4-29
Kg
Yk
= 1
k=1
and that the sum of the diffusion fluxes must be zero, i.e.,
Equation 4-30
Kg
Yk Vk
= 0
k=1
RD01400-C01-004-001A
71
CHEMKIN 4.0
n T gas +
K b(N b)
Kg
4
Y k ( V k + u )h k = ( T T o ) +
k=1
s k W k h k n bulk T bulk
k = K s(N s)
The summation on the right-hand side runs over all surface and bulk species. It is
interesting to note that by substituting Equation 4-26 into the flux term on the left-hand
side, the energy balance can be written in a more compact form as
Equation 4-32
K
n T
= ( T
gas
4
To )
s k Wk hk n bulk T bulk
k=1
Now the reaction-rate summation on the left-hand side runs over all species, including
the gas-phase species.
The SURFACE KINETICS package allows the user to specify mass densities k for the
bulk species. A possible use for the densities would be to convert surface reaction
rate-of-production of a bulk species (in mole/cm2/sec) into a growth rate G (in
cm/sec). The needed relationship is
Equation 4-33
l
K b(N b)
G =
s k W k
-----------
k = K b(N b)
4.11
72
RD01400-C01-004-001A
Theory Manual
system, we require that one acknowledge its use by adding a keyword on the
REACTION line (Section 4.5.1 of the CHEMKIN Input Manual). It is up to the user's
program to ensure that the current site concentrations are correct. Subroutines SKRAT,
SKRATI, which return an array of species production rates, also return an array of
surface phase production rates, which would all be zero if sites are conserved in
every reaction.
In the case where the user overrides the default of site conservation, the production
2
rate n (in moles cm sec ) for each surface phase is given by:
Equation 4-34
n =
( n, i )qi
( n = N s, , N s )
i=1
where
Equation 4-35
SKRAT, SKRATI
l
Ks ( n )
( n, i ) =
ki k ( n )
k = Ks ( n )
Here, the term ( n, i ) is the net change in the number of surface sites of type n for
surface reaction i .
RD01400-C01-004-001A
73
CHEMKIN 4.0
74
RD01400-C01-004-001A
CHEMKIN 4.0
RD01400-C01-004-001A
75
CHEMKIN 4.0
The multicomponent formulation has several important advantages over the relatively
simpler mixture formulas. The first advantage is accuracy. The mixture formulas are
only correct asymptotically in some special cases, such as in a binary mixture, or in
diffusion of trace amounts of species into a nearly pure species, or systems in which
all species except one move with nearly the same diffusion velocity.20 A second
deficiency of the mixture formulas is that overall mass conservation is not necessarily
preserved when solving the species continuity equations. To compensate for this
shortcoming one has to apply some ad hoc correction procedure.19, p. 76,21 The
multicomponent formulation guarantees mass conservation without any correction
factors, which is a clear advantage. The only real deficiency of the multicomponent
formulation is its computational expense. Evaluating the ordinary multicomponent
diffusion coefficients involves inverting a K K matrix, and evaluating the thermal
conductivity and thermal diffusion coefficients requires solving a 3K 3K system of
algebraic equations, where K is the number of species.
To maximize computational efficiency, TRANSPORT is structured to do a large portion of
the calculations in a Pre-processor that provides information to CHEMKIN through a
Linking File. Polynomial fits are thus computed a priori for the temperature-dependent
parts of the kinetic theory expressions for pure species viscosities and binary diffusion
coefficients. (The pure species thermal conductivities are also fit, but are only used in
the mixture-averaged formulation.) The coefficients from the fit are passed to
subroutines in the TRANSPORT Subroutine Library, which can be used to return either
mixture-averaged properties or multicomponent properties. With this fitting procedure,
expensive operations, such as evaluation of collision integrals, are only done once
and not every time a property value is needed.
5.1
19.R. J. Kee, J. Warnatz, and J. A. Miller, A Fortran Computer Code Package for the Evaluation of
Gas-Phase Viscosities, Conductivities, and Diffusion Coefficients, Sandia National Laboratories Report SAND83-8209, 1983.
20.R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena, John Wiley and Sons, New
York, 1960.
21.T. P. Coffee and J. M. Heimerl, Combustion and Flame 43:273 (1981)
22.J. O. Hirschfelder, C. F. Curtiss, and R. B. Bird, Molecular Theory of Gases and Liquids, John
Wiley and Sons, New York, 1954.
76
RD01400-C01-004-001A
Theory Manual
Equation 5-1
kB T
*
T k = --------
1 k
= --- ----------32
k k
In the above expression k is the Lennard-Jones potential well depth and k is the
dipole moment. The collision integral value is determined by a quadratic interpolation
of the tables based on Stockmayer potentials given by Monchick and Mason.23
The binary diffusion coefficients22, p. 76 are given in terms of pressure and
temperature as
Equation 5-4
3 3
D kj
3 2 k B T m jk
= ------ ---------------------------------16 P 2 (1,1)*
jk
mj mk
m jk = ----------------mj + mk
RD01400-C01-004-001A
77
CHEMKIN 4.0
(1,1)*
j k
---- ---- k B k B
jk
------ =
kB
Equation 5-7
1
jk = --- ( j + k )
2
Equation 5-8
2
jk = j k
For the case of a polar molecule interacting with a nonpolar molecule:
Equation 5-9
np
------- =
kB
n p
---- ---- k B k B
Equation 5-10
np
1
= --- ( n + p )
2
1--6
Equation 5-11
2
np = 0
where,
Equation 5-12
1 * *
= 1 + --- n p ----p4
n
78
RD01400-C01-004-001A
Theory Manual
In the above equations n is the reduced polarizability for the nonpolar molecule and
*
p is the reduced dipole moment for the polar molecule. The reduced values are
given by
Equation 5-13
*
n = -----n3
n
Equation 5-14
p
*
p = -------------3
p p
The table look-up evaluation of the collision integral
temperature
(1,1)*
Equation 5-15
kB T
*
T jk = --------
jk
1 *2
*
jk = --- jk
2
Although one could add a second-order correction factor to the binary diffusion
coefficients24 we have chosen to neglect this since, in the multicomponent case, we
specifically need only the first approximation to the diffusion coefficients. When higher
accuracy is required for the diffusion coefficients, we therefore recommend using the
full multicomponent option.
5.2
RD01400-C01-004-001A
79
CHEMKIN 4.0
Equation 5-17
where
Equation 5-18
5
2 C , rot. A
f trans. = --- 1 --- ------------------ ---
C , trans. B
2
Equation 5-19
D kk
2A
f rot. = ------------ 1 + --- ---
k
B
Equation 5-20
D kk
f vib. = ----------k
and,
Equation 5-21
D kk
A = 5--- ----------2
k
Equation 5-22
2 5 C , rot. D kk
B = Z rot. + --- --- --------------- + ---------- 3 R
k
The molar heat capacity C relationships are different depending on whether (or not)
the molecule is linear (or not). In the case of a linear molecule,
Equation 5-23
C , trans.
-----------------= 3--R
2
Equation 5-24
C , rot.
--------------- = 1
R
80
RD01400-C01-004-001A
Theory Manual
Equation 5-25
5
C , vib. = C --- R
2
In the above, C is the specific heat at constant volume of the molecule and R is the
universal gas constant. For the case of a nonlinear molecule,
Equation 5-26
C , trans.
------------------ = 3--R
2
Equation 5-27
C , rot.
-------------- = 3--R
2
Equation 5-28
C , vib. = C 3R
The translational part of C is always the same,
Equation 5-29
3
C , trans. = --- R
2
In the case of single atoms (H atoms, for example) there are no internal contributions
to C , and hence,
Equation 5-30
3
k = ------k- f trans. --- R
Wk
2
where f trans. = 5 2 . The self-diffusion coefficient comes from the following
expression,
Equation 5-31
3 3
D kk
The density comes from the equation of state for a perfect gas,
RD01400-C01-004-001A
81
CHEMKIN 4.0
Equation 5-32
PW
RT
F ( 298 )
Z rot. ( T ) = Z rot. ( 298 ) ----------------F(T)
where,
Equation 5-34
3--2
1---
k B 2
---------------
k B
3--2
3---
kB 2
F(T) = 1 +
+
----- + 2 ------------ + ------------
4
T
T
2 T
5.3
Equation 5-35
N
ln k =
an, k ( ln T )
n1
n=1
82
RD01400-C01-004-001A
Theory Manual
Equation 5-36
N
bn, k ( ln T )
ln k =
n1
n=1
The fits are done for each pair of binary diffusion coefficients in the system.
Equation 5-37
N
ln D kj =
dn, jk ( ln T )
n1
n=1
By default TRANSPORT uses third-order polynomial fits (i.e., N = 4 ) and we find that
the fitting errors are well within one percent. The fitting procedure must be carried out
for the particular system of gases that is present in a given problem. Therefore, the
fitting cannot be done once and for all, but must be done once at the beginning of
each new problem.
The viscosity and conductivity are independent of pressure, but the diffusion
coefficients depend inversely on pressure. The diffusion coefficient fits are computed
at unit pressure; the later evaluation of a diffusion coefficient is obtained by simply
dividing the diffusion coefficient as evaluated from the fit by the actual pressure.
Even though the single component conductivities are fit and passed to the TRANSPORT
Subroutine Library, they are not used in the computation of multicomponent thermal
conductivities; they are used only for the evaluation of the mixture-averaged
conductivities.
5.4
Equation 5-38
2
3
= ( + ( ) ) + --- ( )
T
RD01400-C01-004-001A
83
CHEMKIN 4.0
q =
RT
-D
jk hk 0 T -----------Wk Xk k
k=1
dk
k=1
where,
Equation 5-40
1
d k = X k + ( X k Y k ) --- P
P
The multicomponent species flux is given by
Equation 5-41
jk = Yk Vk
where Y k are the mass fractions and the diffusion velocities are given by
Equation 5-42
K
Dk 1
1
V k = ----------- W j D k, j d j --------- --- T
Yk T
Xk W
jk
The species molar masses are denoted by W k and the mean molar mass by W . D k, j
T
are the ordinary multicomponent diffusion coefficients, and D k are the thermal
diffusion coefficients.
By definition in the mixture-average formulations, the diffusion velocity is related to
the species gradients by a Fickian formula as,
Equation 5-43
T
k 1
1D d D
- --- T
V k = ----km k -------
Y
Xk
kT
28.R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena, John Wiley and Sons, New
York, 1960.
84
RD01400-C01-004-001A
Theory Manual
Equation 5-44
1 Yk
D km = ------------------------------K
Xj Djk
jk
A potential problem with this expression is that it is not mathematically well defined in
the limit of the mixture becoming a pure species. Even though diffusion itself has no
real meaning in the case of a pure species, the numerical implementation must
ensure that the diffusion coefficients behave reasonably and that the program does
not blow up when the pure species condition is reached. We circumvent these
problems by evaluating the diffusion coefficients in the following equivalent way.
Equation 5-45
K
j k X j W j
D km = ------------------------------------K
W X j D jk
jk
In this form the roundoff is accumulated in roughly the same way in both the
numerator and denominator, and thus the quotient is well behaved as the pure
species limit is approached. However, if the mixture is exactly a pure species, the
formula is still undefined.
To overcome this difficulty we always retain a small quantity of each species. In other
words, for the purposes of computing mixture diffusion coefficients, we simply do not
allow a pure species situation to occur; we always maintain a residual amount of each
species. Specifically, we assume in the above formulas that
Equation 5-46
X k = Xk +
where X k is the actual mole fraction and is a small number that is numerically
insignificant compared to any mole fraction of interest, yet which is large enough that
12
there is no trouble representing it on any computer. A value of 10
for works well.
In some cases (for example, Warnatz29 and Coltrin, et al.30) it can be useful to treat
multicomponent diffusion in terms of an equivalent Fickian diffusion process. This is
sometimes a programming convenience in that the computer data structure for the
multicomponent process can be made to look like a Fickian process. To do so
RD01400-C01-004-001A
85
CHEMKIN 4.0
supposes that a mixture diffusion coefficient can be defined in such a way that the
diffusion velocity is written as Equation 5-43 rather than Equation 5-42. This
equivalent Fickian diffusion coefficient is then derived by equating Equation 5-42 and
Equation 5-43 and solving for D km as
Equation 5-47
K
j k W j D k, j d j
D km = ------------------------------------Wd k
j k Wj Dkj dj +
D km = -------------------------------------------W ( dk + )
Furthermore, for the purposes of evaluating the multicomponent D km , it may be
advantageous to compute the d k in the denominator using the fact that
X k = K
X j . In this way the summations in the numerator and the
jk
denominator accumulate any rounding errors in roughly the same way, and thus the
quotient is more likely to be well behaved as the pure species limit is approached.
Since there is no diffusion due to species gradients in a pure species situation, the
exact value of the diffusion coefficient is not as important as the need for it to be well
defined, and thus not cause computational difficulties.
In practice we have found mixed results using the equivalent Fickian diffusion to
represent multicomponent processes. In some marching or parabolic problems, such
as boundary layer flow in channels,30, p. 86 we find that the equivalent Fickian
formulation is preferable. However, in some steady state boundary value problems,
we have found that the equivalent Fickian formulation fails to converge, whereas the
regular multicomponent formulation works quite well. Thus, we cannot confidently
recommend which formulation should be preferred for any given application.
86
RD01400-C01-004-001A
5.5
Theory Manual
Equation 5-49
K
Xk k
---------------------------- K
X j kj
k = 1
j=k
where
Equation 5-50
1---
2
1--2
1--- 2
4
W
W
1
kj = ------- 1 + ------k- 1 + -----k -------j
j Wk
Wj
8
1---
1
= X k k + -------------------------------
K
2
k = 1
Xk k
k=1
5.6
RD01400-C01-004-001A
87
CHEMKIN 4.0
A thermal diffusion ratio k can be defined such that the thermal diffusion velocity W k
is given by
Equation 5-52
D km k 1 T
W ki = ----------------- --- ------Xk T xi
where x i is a spatial coordinate. The mole fractions are given by X k , and the D km are
mixture diffusion coefficients Equation 5-43. In this form we only consider thermal
diffusion in the trace, light component limit (specifically, species k having molecular
mass less than 5). The thermal diffusion ratio33 is given by
Equation 5-53
K
k =
kj
jk
where
Equation 5-54
*
Wk Wj
15 ( 2A kj + 5 ) ( 6C kj 5 ) - -------------------X X
= ------ --------------------------------------------------------2 A * ( 16A * 12B * + 55 ) W k + W j k j
kj
kj
kj
kj
(2,2)
Equation 5-56
*
B ij
(1,2)
(1,3)
1 5 ij ij = --- ----------------------------------(1,1)
3
ij
Equation 5-57
*
C ij
(1,2)
33.S. Chapman and T. G. Cowling, The Mathematical Theory of Non-Uniform Gases Cambridge
University Press, Cambridge, 1970.
88
RD01400-C01-004-001A
Theory Manual
5.7
Equation 5-58
00, 00 00, 10
L
0
L
10, 00 10, 10 10, 01
L
L
L
01, 10 01, 01
L
0 L
a 1
00
1
a 10
1
a 01
= X
where right hand side vector is composed of the mole fraction vectors X k . The
00, 00
multicomponent diffusion coefficients are given in terms of the inverse of the L
block as
Equation 5-59
16T W
D i, j = X i ---------- ------ ( p ij p ii )
25P W j
where
Equation 5-60
(p) = (L
RD01400-C01-004-001A
00, 00
89
CHEMKIN 4.0
The thermal conductivities are given in terms of the solution to the system of
equations by
Equation 5-61
K
0, tr. = 4
Xk ak10
k=1
Equation 5-62
K
1
Xk ak01
0, int. = 4
k=1
Equation 5-63
0 = 0, tr. + 0, int.
and the thermal diffusion coefficients are given by
Equation 5-64
8m k X k 1
T
D k = ---------------- a k00
5R
The components of the L matrix are given by Dixon-Lewis,17, p. 75
Equation 5-65
00, 00
L ij
16T
= ---------25P
Xk
- { m X ( 1 ik ) m i X j ( ij jk ) }
------------m i D ik j j
k=1
Equation 5-66
00, 10
L ij
8T
= ------5P
k=1
m k ( 1.2C jk 1 )
X j X k ( ij ik ) -----------------------------------( m j + m k )D jk
Equation 5-67
10, 00
L ij
00, 10
= L ji
Equation 5-68
01, 00
L ij
00, 01
= L ji
= 0
90
RD01400-C01-004-001A
Theory Manual
Equation 5-69
10, 10
L ij
16T
= ---------25P
k=1
Xi Xk
m
2 *
15 2 25 2
-----i ----------------------------------- ( jk ij ) ------ m j + ------ m k 3m k B ik
2
2
4
mj ( m + m ) D
jk
*
5 c i, rot. c k, rot.
4m i m k A ik ( jk + ij ) 1 + ------ ------------ + --------------
3 k B ik k B ki
Equation 5-70
10, 10
L ii
Xi Xk
-
---------------------------------2
k i ( mi
+ m k ) D ik
15 2 25 2
2 *
*
5 c i, rot. c k, rot.
- m i + ------ m k 3m k B ik + 4m i m k A ik 1 + ------ ------------ + --------------
----3 k B ik k B ki
4
2
Equation 5-71
10, 01
L ij
32T
= ----------------------5 Pc j, int.
k=1
c j, rot.
m j A jk
------------------------------( ik + ij ) X j X k -----------( m j + m k )D jk
k B jk
Equation 5-72
10, 01
L ii
*
m i X i k B c i, int.
32Tk B
c i, rot.
m i A ik
16
= ------ --------------------- ------------ + ----------------------- ------------------------------- X i X k -----------k B ik
3 R i c i, int. k B ii 5 Pc i, int.
( m i + m k )D ik
ki
Equation 5-73
01, 10
L ij
10, 01
= L ji
Equation 5-74
01, 10
L ii
RD01400-C01-004-001A
K
K
2
*
k
8k B m i X i c i, rot. 4k B T
Xi Xk
12X i X k m i A ik c irot.
= --------------- ------------ ------------ ---------------- ---------------- + ----------------- ------ ------- ----------
2
c i, int. R i k B ii c i, int. P k = 1 D iint., k k i 5 c iint. m k D ik ii
91
CHEMKIN 4.0
Equation 5-75
01, 01
L ij
= 0(i j)
c k, rot.
------------- = 1
kB
and for a nonlinear molecule
Equation 5-77
c k, rot.
------------ = 3--2
kB
The internal component of heat capacity is computed by subtracting the translational
part from the full heat capacity as evaluated from the CHEMKIN Thermodynamic
Database.
Equation 5-78
c k, int.
c
------------- = ----p- 3--kB
kB 2
Following Dixon-Lewis,17, p. 75 we assume that the relaxation collision numbers ij
depend only on the species i , i.e., all ij = ii . The rotational relaxation collision
number at 298 K is one of the parameters in the TRANSPORT database, and its
temperature dependence was given in Equation 5-33 and Equation 5-34.
For non-polar gases the binary diffusion coefficients for internal energy D iint., k are
approximated by the ordinary binary diffusion coefficients. However, in the case of
collisions between polar molecules, where the exchange is energetically resonant, a
large correction of the following form is necessary,
Equation 5-79
D pp
D pint., p = ----------------------( 1 + ' pp )
92
RD01400-C01-004-001A
Theory Manual
Equation 5-80
' pp = 2985
-----------T
WY
X k = ----------k +
Wk
12
A value of = 10
works well; it is small enough to be numerically insignificant
compared to any mole fraction of interest, yet it is large enough to be represented on
nearly any computer.
5.8
Species Conservation
Some care needs to be taken in using the mixture-averaged diffusion coefficients as
described here. The mixture formula are approximations, and they are not
constrained to require that the net species diffusion flux is zero, i.e., the condition,
Equation 5-82
K
Vk Yk
= 0
k=1
RD01400-C01-004-001A
93
CHEMKIN 4.0
+V
Vk = V
k
c
Vc =
Yk Vk
k=1
94
RD01400-C01-004-001A
CHEMKIN 4.0
36.W. C. Reynolds, The Element Potential Method for Chemical Equilibrium Analysis: Implementation in the Interactive Program STANJAN, Department of Mechanical Engineering, Stanford University (1986).
RD01400-C01-004-001A
95
CHEMKIN 4.0
The user may specify the state parameters in a number of different ways, including
a. temperature and pressure;
b. pressure and entropy;
c. enthalpy and pressure; and
d. volume and entropy.
Species composition can be frozen in a given calculation, or the equilibrium
composition can be determined. Calculations may be linked through continuations,
such that the conditions calculated from a previous equilibrium case can be used as
the starting point for a subsequent case with different constraints. In this way, the user
can employ the Equilibrium Reactor Model to analyze stages in a thermodynamic
cycle.
The Equilibrium Reactor Model is also commonly used to determine adiabatic flame
temperatures for combustible gas mixtures. Such a simulation is performed by
specifying an initial (reagent) gas mixture and constraining equilibrium for constant
enthalpy (adiabatic) and constant pressure. The calculation can also be performed
using constant internal energy and constant volume. An initial temperature guess at
or above about 1000 K is usually needed to cause the equilibrium solver to find the
burned-gas solution. For accurate adiabatic-flame temperature calculations, it is
important to include all radical species that might occur in the flame, as well as stable
reactants and products.
In the remainder of this chapter, we describe the equations solved and the
methodology used for determining chemical and phase equilibria of arbitrary systems.
6.1
Equation 6-1
K
G =
gk Nk
k=1
where g k is the partial molal Gibbs function and N k is the number of moles of each
species k in the system. K is the total number of species.
96
RD01400-C01-004-001A
Theory Manual
For ideal-gas mixtures or ideal solutions, the partial molal Gibbs functions are given
by:
Equation 6-2
g k = g k (T,P) + RT ln X k
where g k (T,P) is the Gibbs function for the pure species k , evaluated at the system
temperature and pressure; R is the universal gas constant; and X k is the mole
fraction of the kth species.
The equilibrium solution at a given temperature and pressure is the distribution of N k
that minimizes the system Gibbs function, G , subject to atomic population constraints
(and non-negative N k ). The atomic population constraints are:
Equation 6-3
K
njk Nk
= pj
j = 1, , M
k=1
where n jk is the number of the j th atoms that appear in the kth molecule, is the total
population in moles of the j th atom in the system, and M is the total number of
different elements that are present in the system.
Details regarding the relationship between the partial molar Gibbs functions and the
elemental potentials for the atoms, as well as the explicit form of the equations solved
in the STANJAN library, are described in the STANJAN report.36, p. 95
RD01400-C01-004-001A
97
CHEMKIN 4.0
98
RD01400-C01-004-001A
CHEMKIN 4.0
7.1
Shock-tube Experiments
The shock tube has found widespread use as an experimental device in which to
investigate chemical kinetic behavior in reactive gas mixtures. Much can be learned
by experiment alone, however such investigations are enhanced considerably when
done in concert with computer simulations. To this end, the Shock Reactor Model
simulates the chemical changes that occur after the shock heating of a reactive gas
mixture. The Shock Reactor Model is designed to account for both incident and
reflected normal shock waves. It makes allowances for the non-perfect gas behavior,
boundary layer effects and detailed finite-rate chemistry.
RD01400-C01-004-001A
99
CHEMKIN 4.0
100
RD01400-C01-004-001A
Figure 7-1
Theory Manual
A distance-time diagram
shows the movements of
the shock front, contact
surface, rarefaction wave,
and reflected shock wave
7.2
7.2.1
RD01400-C01-004-001A
101
CHEMKIN 4.0
z = U s ( t l t p ) = U t p
Figure 7-2
Taking the limit as z approaches zero results in the following ordinary differential
equation relating gas particle and laboratory times:
Equation 7-2
dt
U-------l = 1 ----dt p
Us
For mass continuity across the shock wave, it can be shown that
Equation 7-3
1 A1 A = 1 U Us
and, therefore, Equation 7-2 can be written as
Equation 7-4
dt
-------l = 1 A 1 A
dt p
102
RD01400-C01-004-001A
Theory Manual
Thus, when rate processes are measured, the time as measured must be multiplied
by the density-area ratio across the shock to obtain the true rate referred to the test
gas. In the Shock Reactor Model, Equation 7-4 is integrated along with the other
ordinary differential equations (ODEs) presented in Section 7.3, so that laboratory
time as a function of gas-particle time is available. The Shock Reactor Model solves
the coupled set of ODEs for either an incident or reflected shock problem.
The initial time for a problem, t p = 0 , is taken as the time immediately after the shock
wave has elevated the test gas to new levels of temperature and pressure. The
incident shock wave is assumed to instantaneously raise the test gas from initial
conditions 1 to conditions 2; the reflected shock is assumed to instantaneously raise
the test gas from conditions 2 to conditions 5. Gas composition immediately after
passage of the shock is assumed to remain unchanged from the initial conditions.
Vibrational and rotational energy relaxation processes are neglected.
7.2.2
Equation 7-5
u1 = Us
Equation 7-6
u2 = Us U2
where U s is the shock velocity. Gas conditions associated with the incident shock in
the two coordinate systems are shown in Figure 7-3. The Rankine-Hugoniot relations
for properties across the incident shock front are
Equation 7-7
1 u1 = 2 u2
Equation 7-8
2
P1 + 1 u1 = P2 + 2 u2
RD01400-C01-004-001A
103
CHEMKIN 4.0
Equation 7-9
2
h1 + u1 2 = h2 + u2 2
Utilizing the equation of state and Equation 7-7 to eliminate the velocity u 2 from
Equation 7-8 and Equation 7-9 results in the following expressions for the pressure
and temperature ratios across the incident shock:
Equation 7-10
2
1 + 1 u1 P1 [ 1 ( T2 T1 ) ( P1 P2 ) ] P2 P1 = 0
Equation 7-11
2
h1 + u1 2 [ 1 ( P1 P2 ) ( T2 T1 ) ] h2 = 0
Figure 7-3
104
RD01400-C01-004-001A
Theory Manual
An iterative procedure is employed to solve Equation 7-10 and Equation 7-11 for P 2
and T 2 . Letting and be the temperature and pressure ratios, respectively, across
the shock, Equation 7-10 can be solved for in terms of ( 36 ) to yield
Equation 7-12
2
2 2
1 u 1
1 u 1 4 1 u 1
1
= --- 1 + ---------- + 1 + ---------- ------------------2
P1
P1
P1
This expression is then substituted into Equation 7-11 to yield one equation with one
unknown, . Within CHEMKIN, a routine called ZEROIN37, which finds the zeros of
functions, is employed to determine the value of that satisfies this equation. An
initial guess for is provided by assuming that the test gas is ideal ( c p and c v are
constant and independent of temperature). For ideal gases
Equation 7-13
1- 2
1- --------- M 2 ---------M1 + 1
T2 1
2 2
= ----- ------------------------------------------------------------------T1
+ 1 2 2
----------M
2 1
where is the specific heat ratio and M 1 is the Mach number of the incident shock.
Equation 7-14
1
M 1 = U s --------
P1
12
Many times the experimentalist reports the incident shock speed, U s , and
temperature and pressure behind the shock, T 2 and P 2 , respectively. Before the
experiment can be modeled, however, the gas velocity behind the shock must be
determined. Employing the equation of state in Equation 7-10 to eliminate 1 P 1
results in
Equation 7-15
2
2 u1
1 + ----------- ( T 2 T 1 ) [ 1 ( T 2 T 1 ) ( P 1 P 2 ) ] ( P 2 P 1 ) = 0
P2
37.L. F. Shampine and H. A. Watts, Zeroin, A Root-Solving Code, Sandia National Laboratories Report SC-TM-70-631, 1970.
RD01400-C01-004-001A
105
CHEMKIN 4.0
This equation and Equation 7-11 again represent two equations in two unknowns.
The solution gives T 1 and P 1 and from these the density in region 1 is determined
from the equation of state. The velocity behind the shock, u 2 , is determined from
Equation 7-7. The solution to Equation 7-15 and Equation 7-11 is analogous to that
already described for Equation 7-10 and Equation 7-11.
7.2.3
Equation 7-16
u' 2 = U rs + U 2
and
Equation 7-17
u 5 = U rs U 5
The gas velocity measured in shock-fixed coordinates with respect to the reflected
shock is u' . Gas conditions associated with the reflected shock in the two coordinate
systems are shown in Figure 7-4. The Rankine-Hugoniot relations for properties
across the reflected shock are
Equation 7-18
2 u' 2 = 5 u 5
Equation 7-19
2
P 2 + 2 ( u' 2 ) = P 5 + 5 ( u 5 )
and
Equation 7-20
2
h 2 + ( u' 2 ) 2 = h 5 + ( u 5 ) 2
106
RD01400-C01-004-001A
Theory Manual
By analogy with Equation 7-7, Equation 7-8 and Equation 7-9, the solution to the
above set of coupled equations is found by finding the values of T 5 and P 5 which
satisfy
Equation 7-21
2 ( u' 2 )
1 + -------------------- [ 1 ( T 5 T 2 ) ( P 2 P 5 ) ] ( P 5 P 2 ) = 0
P2
and
Equation 7-22
2
( u' 2 )
2
2
h 2 + ------------[ 1 ( P2 P5 ) ( T5 T2 ) ] h5 = 0
2
where
Equation 7-23
u' 2 = U rs + U 2 = U rs + U s u 2
When the gas is assumed to be at rest behind the reflected shock (i.e., U 5 = 0 ), then
the reflected shock velocity, U rs , is given by
Equation 7-24
( P2 P5 ) ( T5 T2 ) ( Us u2 )
U rs = ---------------------------------------------------------------[ 1 ( P2 P5 ) ( T5 T2 ) ]
Only in the ideal case is the gas behind the reflected shock at rest, however. Nonidealities cause the gas to move in the same direction as the incident shock and,
hence, at conditions different from those calculated assuming U 5 = 0 . Therefore,
often the velocity of the reflected shock, as well as that of the incident shock, is
measured. Then the values of T 5 and P 5 are computed so as to satisfy the RankineHugoniot relationships using these measured velocities. Having determined T 5 and
P 5 , the density of the gas at condition 5 can be determined from the equation of state
and u 5 , from Equation 7-18.
RD01400-C01-004-001A
107
CHEMKIN 4.0
Initial estimates for T 5 and P 5 can be obtained by again assuming ideal gas behavior.
The temperature ration across the incident shock when c p and c v are constant is
given by Equation 7-13 and the pressure ratio by
Equation 7-25
2
2 1 M1 ( 1 1 )
P 2 P 1 = ---------------------------------------( 1 + 1 )
Letting and be defined as the following density ratios:
Equation 7-26
= 2 1
Equation 7-27
= 5 1
then the equation of state, Equation 7-13 and Equation 7-21 to Equation 7-25 can be
combined to show that for temperature-independent specific heats
Equation 7-28
2
M1 ( 1 + 1 )
= ----------------------------------2
( 1 1 )M 1 + 2
Equation 7-29
2
[ M1 ( 1 ) 1 + ]
= -----------------------------------------------------2
M1 ( 1 ) ( 1 1 ) +
Equation 7-30
2
1 M1 ( 1 ) ( 1 )
P 5 P 1 = 1 + ----------------------------------------------( )
Equation 7-31
2
M1 ( 1 1 ) ( 1 ) ( 1 )
T 5 P 1 = 1 + -----------------------------------------------------------( )
108
RD01400-C01-004-001A
Theory Manual
and
Equation 7-32
Us ( 1 )
U rs = ----------------------( )
Even in the event that U rs is specified, and hence U rs is not necessarily zero, we still
assume that the gas is at rest when solving the conservation equations. Only the
initial state of the gas is modified by the non-ideal reflected shock velocity. Because of
this, for reflected shock problems, we find it more convenient to use laboratory-time
as the independent variable, since then the boundary layer effects are of no
consequence. For such cases, laboratory-time and gas-particle time are the same.
Figure 7-4
RD01400-C01-004-001A
109
CHEMKIN 4.0
7.3
Equation 7-33
Continuity
vA = constant
Equation 7-34
Momentum
dv
dz
v ------ + dP
------- = 0
Equation 7-35
dz
Energy
dv
dh
------ + v ------ = 0
dz
dz
Equation 7-36
Species
dY
v --------k = k W k
dz
Temperature is related to the specific enthalpy of the gas mixture through the
relations:
Equation 7-37
K
h =
hk Yk
k=1
and
Equation 7-38
T
h k = ( h k ) 0 + c pk dT
T0
110
RD01400-C01-004-001A
Theory Manual
The net molar production rate of each species due to chemical reaction is denoted by
k . A detailed description of this term is given in the Section 2.1 of the CHEMKIN Input
Manual. The equations of state relating the intensive thermodynamic properties is
given by:
Equation 7-39
PW = RT
where the mixture molecular weight is determined from the local gas concentration
via:
Equation 7-40
1
W = -------------------------K
Yk Wk
k=1
In the shock tube experiments, the usual measurable quantities are density, species
concentration, velocity and temperature as functions of time. It is therefore desirable
to have time as the independent variable and not distance. Employing the relation
Equation 7-41
dd- = v ------dz
dt
differentiating Equation 7-37, Equation 7-38, Equation 7-39, and Equation 7-40, and
combining the equations results in the following set of coupled, ordinary differential
equations:
Equation 7-42
R
1
d------ = ---------------------------------------------- ---------
dt
P + Pv 2 c p T Pv 2 Wc p
k=1
dA
Wc p T 2 v 3
R
k W k h k -------------- + ----------- 1 ---------- -------
A
Wk
c p W dz
Equation 7-43
dY k
k W
-------- = -------------k
dt
Equation 7-44
2
d v dA
dv
------ = --v- ------ ----- -------
dt
dt A dz
RD01400-C01-004-001A
111
CHEMKIN 4.0
Equation 7-45
2
v d
1
dT
------ = -------- ------ ------- c p dt c p
dt
k=1
v dA
h k k W k + --------- -------
Ac p dz
The time-histories of the measurable flow quantities should satisfy these relations.
The distance of a fluid element from the shock, z , follows from Equation 7-41 and is
given by
Equation 7-46
dz
----- = v
dt
These ODEs, Equation 7-42, Equation 7-43, Equation 7-44 and Equation 7-45, are
integrated along with Equation 7-46, for distance from the shock, and Equation 7-4 for
laboratory time, when gas-particle time is the independent variable. Values of the
pressure, mean molecular weight and area as a function of gas-particle time are also
determined.
7.3.1
Equation 7-47
v - = 1 ( z l )1 2
-----------m
( v )2
where l m is the distance between the shock and contact surface at infinite distance
from the diaphragm and the subscript 2 denotes conditions immediately behind the
shock. He then obtained an expression for l m by considering the simultaneous
boundary layer development and change in free-stream conditions external to the
boundary layer.
112
RD01400-C01-004-001A
Theory Manual
This expression is
Equation 7-48
lm
2
2 2 1
u2
d
= -----------2- ------ ----------------- -----
16 w ( w 1 ) v w
where
Equation 7-49
+ 0.802W-
= 1.59C 1 + 1.796
------------------------------------( Zw 1 )
with Z = ( + 1 ) ( 1 ) and w = u w u 2 . The effect of variable viscosity is
accounted for by C , where
Equation 7-50
2 2
C = ------ ------
w w
0.37
The wall is assumed to remain at its initial temperature, while the pressure at the wall
changes to P 2 after passage of the shock. The viscosity correction is based on
numerical solutions for air. For the purposes of the boundary layer corrections, we
take the viscosity to be that of the diluent gas.
Hirschfelder, Curtiss and Bird39 give the viscosity of a pure gas as
Equation 7-51
5
10
WT
= 2.6693
---------------------------------------------*
2 ( 2, 2 ) *
T
where is the viscosity in gm/cm sec; , the low-velocity collision cross-section for
the species of interest
in Angstroms; W , the molecular weight; T , the temperature in
*
( 2, 2 )
Kelvin; and
, the reduced collision integral, a function of the reduced
*
*
temperature T ( T = T ( k ) ) where ( k ) is the potential parameter for the
species of interest). The reduced collision integral represents an averaging of the
collision cross-section over all orientations and relative kinetic energies of colliding
molecules. Tabulated values of this integral at various reduced temperatures are
given by Camac and Feinberg.40
39.J. O. Hirschfelder, C. F. Curtiss, and R. B. Bird, Molecular Theory of Gases and Liquids, John
Wiley and Sons, Inc., New York, 1967.
RD01400-C01-004-001A
113
CHEMKIN 4.0
The values can be fit to within 2% for T > 2.7 by the expression:
Equation 7-52
( 2, 2 )
* 0.1756
= 1.2516 ( T )
2.1327 10
W- 0.6756
= -----------------------------------------T
2
0.1756
( k)
Evaluating the above equation at 300 K and using this as a reference point results in
the following expression for the viscosity:
Equation 7-54
= 0 ( T 300 )
0.6756
A value for 0 for the diluent gas must be specified by the user when considering
boundary layer effects.
To derive an equation for the area variation, we first combine Equation 7-33 and
Equation 7-47 to yield
Equation 7-55
A2
12
------ = 1 ( z l m )
A
Then, the change in cross-sectional area with distance downstream of the shock is
given by
Equation 7-56
1 2
( z lm )
--1- dA
------- = --------------------------------------------12
A dz
2l m ( 1 ( z l m ) )
This expression is used in Equation 7-42, Equation 7-44 and Equation 7-45.
Equation 7-56 allows us to account for the boundary layer effects by computing the
effective area through which the gas must flow.
114
RD01400-C01-004-001A
CHEMKIN 4.0
RD01400-C01-004-001A
115
CHEMKIN 4.0
8.1
8.2
116
RD01400-C01-004-001A
Theory Manual
Further assumptions and limitations are described for specific model formulations in
the following sections.
8.3
General Equations
Conservation of mass, energy, and species for a well mixed reactor or homogeneous
system include net generation of chemical species within the reactor volume, and net
loss of species and mass to surfaces in the reactor. Homogeneous systems include
closed batch reactors, single-zone engine-cylinder models, perfectly stirred reactor
approximations, well mixed (low-pressure) plasma processing reactors, for example.
In closed batch reactors, there are no inlets or outlets flow during the period of
interest, although there may or may not be heat flow to the external environment.
Such batch systems are inherently transient, where the chemical state changes as
production and destruction of species progress through chemical reaction. A perfectly
stirred reactor (PSR) consists of a chamber, which may or may not allow heat loss,
having inlet and outlet ducts. There may be more than one inlet defined for each
reactor.
Figure 8-1 illustrates the conceptual representation of a generic reactor chamber. A
steady flow of reactants is introduced through the inlet with a given species
composition and temperature. For transient systems, there may be no flow, such that
the system is closed with respect to mass transfer other than surface losses or gans.
*
In some cases, the sum of the mass flow rates into the reactor, m i , may not be
equal to the mass flow rate out of the reactor, m , due to deposition on or etching of
surface materials. For open systems, the reactor pressure is specified, so that the
conservation equations determine the volume outflow. Although Figure 8-1 depicts a
single surface in the reactor, an actual reactor may contain many different material
surfaces, such as reactor walls, silicon wafer, substrate holder, etc. Each of these
materials may have a different set of reaction kinetics associated with it. For this
reason, we have included the capability of defining multiple surface materials that
represent different fractions of the total surface area, with corresponding surface
chemistry mechanisms.
RD01400-C01-004-001A
117
CHEMKIN 4.0
Figure 8-1
In addition to modeling a single well mixed reactor, CHEMKIN allows the user to build a
reactor cluster composed of many reactor modules connected in series, with the
possibility of recycling streams feeding back to upstream reactors or being directed
forward to downstream reactors. A schematic of such a reactor cluster is given in
Figure 8-2. This cluster is included for illustrative purposes only, to show the options
for connecting reactors. The symbols R11, R13, and R32 are examples of recycling
streams, defined as fractions of total mass flow out of a reactor.
Reactor clusters are created using the Graphical User Interface Diagramming tool, as
described in Getting Started with CHEMKIN. This allows users to drag and drop
reactors and connect them using mass-flow streams. Each inlet stream can be
assigned its own attributes for flow rates, gas temperature, and composition.
Recycling streams can be defined by specifying the fraction of the outflow from one
reactor that will flow into another reactor. By default the outflow of a reactor will flow
100 percent into the next (downstream) reactor in series. When recycling streams are
defined by the user, the mass flow from one PSR to the next in series will be one
minus the sum of the recycle flows defined from that reactor to all other reactors (e.g.
in Figure 8-2, R12 = 1 - R11 - R13).
118
RD01400-C01-004-001A
Theory Manual
As an example, the following set of keywords could be entered to define the reactor
network shown in Figure 8-2:
INLET
INLET
INLET
RECY
RECY
RECY
inletA 1 !
inletB 1
inletC 2
1 1 0.10
1 3 0.30
3 2 0.25
Figure 8-2
8.3.1
RD01400-C01-004-001A
119
CHEMKIN 4.0
conservation in the reactor volume, where the time-rate of change of the mass in the
reactor is equal to the difference between the mass flow in and the mass flow out,
plus any material that is added to or subtracted from the surfaces within the chamber.
This equations is stated as:
Equation 8-1
d(j)
---(V) =
dt
inlet
(j)
N PSR
*(j)
m i
i=1
Kg
(r)
R rj m
(j)
r=1
(j)
Am
m=1
(j)
s k, m Wk
j = 1, N PSR
k=1
*
Here j is the reactor number, is the mass density, V is the reactor volume, m is
the inlet mass flow rate, and m is the outlet mass flow rate. N
( j ) is the number of
inlet
inlets for each reactor j , while N PSR is the total number of reactor modules in the
reactor network. R rj is the fraction of the outflow of reactor r that is recycled into
reactor j . The outlet mass flow differs from the sum of the inlet and recycled mass
flow when deposition or etching of materials within the reactor occurs, as represented
by the last term on the right-hand side. In this term, A m is the surface area of the mth
material defined within the reactor, and s k, m is the molar surface production rate of
the kth species on the mth material per unit surface area. There are K g gas-phase
species and M materials.
( k V ) -------dt
(j)
inlet
(j)
N PSR
*(j)
*
m i ( Y k, i
Yk ) +
i=1
Kg
M
(j)
Yk
(j)
(j)
(r)
(j)
R rj ( Y k Y k )
r=1
M
Am s k, m Wk + ( k V )
m=1
(r)
m k
k=1
(j)
Wk +
(j) (j)
Am s k, m Wk
m=1
In Equation 8-2, Y k is the mass fraction of the kth species, W k is the molecular
weight of the kth species, and k is the molar rate-of-production of the kth species by
gas-phase chemical reaction per unit volume. The superscript * indicates inlet stream
quantities.
For steady-state conditions, the nominal residence time in the reactor can be
related to the reactor volume and the inlet mass flow rate as follows:
120
RD01400-C01-004-001A
Equation 8-3
Theory Manual
V
= -----------------------------------------------------------N
inlet
(j)
i=1
N PSR
*(j)
m i
(r)
R rj
r=1
where the mass density is related to the pressure, gas temperature and electron
temperature through the multi-fluid ideal gas equation of state (see Section 2.2). The
residence time is often used as a characteristic parameter of the reactor, rather than
*
the mass flow rate for steady-state flow. In this case, m can be calculated from a
specified residence time using Equation 8-3. Alternatively, the effective volume V can
be calculated given specified values of both the residence time and the mass flow
rates, also using Equation 8-3. If the mass flow rate is zero, then the reactor may not
be characterized by residence time.
8.3.2
Equation 8-4
d---( A c W ) = A m W k s k
dt m k k
k = K s ( m ), , K s ( m ) ;m = 1, , M
Here A m is the surface area of the mth material in the reactor and c k is the molar
concentration of the kth surface species (mole/cm2).
Within the CHEMKIN reactor models, we use a strict ordering of species, species types,
and surface materials that allows very general descriptions of the chemistry between
species of all types in the reactor. The subscript k is then used for all species,
whether gas-phase, surface-phase, or bulk-phase. For this reason, in Equation 8-4
and subsequent equations, we introduce somewhat complicated indices for the range
f
f
of surface species. Specifically, K s ( N s ( m ), m ) refers to the first surface-phase species
RD01400-C01-004-001A
121
CHEMKIN 4.0
in the first surface phase of the material m , while K s ( N s ( m ), m ) refers to the last
surface-phase species of the last surface phase of the mth material. The total number
of surface materials is M , with the total number of surface phases on each material
given by N s ( m ) and the total number of bulk phases by N b ( m ) . The surface phase
f
l
index range for a material is given by N s ( m ) n N s ( m ) . The first phase defined in
j
the system is always the gas-phase, with n = 1 , such that N s ( 1 ) = 2 and
f
K s ( 2, 1 ) = K g + 1 in all cases where surface species are defined. In each phase, n ,
there are K phase ( n, m ) species, whose indices are in the range,
f
l
K s ( n, m ) k K s ( n, m ) . When the species index range includes all surface species,
f
f
f
as in Equation 8-4, we introduce a shorthand notation, where K s ( m ) = K s (N s ( m ),m)
l
l
l
and K s ( m ) = K s (N s ( m ),m) . Although this nomenclature is somewhat confusing on
first glance, it need not unduly concern the user; all of this formalism is strictly
maintained internally in the CHEMKIN software structure. For further insight into this
nomenclature, see the Chapter 1 explanation.
The molar concentration of a surface species is related to the total site density of a
surface phase, n, m , to the site fraction of the kth surface species, and to the species
coverage factor, k , as follows:
Equation 8-5
n, m Z k
c k = --------------k
Assuming the surface areas of each material are constant, substituting Equation 8-5
into Equation 8-4, and expanding the derivatives, we obtain:
Equation 8-6
Z k d n, m
dZ k
s k, m
- ------------- ---------------- = k --------dt
n, m n, m dt
k = K s ( m ), , K s ( m ) ;m = 1, , M
K s ( n, m )
d n, m
k s k, m = -----------dt
n = N s ( m ), , N s ( m ) ;m = 1, , M
k = K s ( n, m )
Equation 8-7 is the surface site conservation equation that must hold true for each
surface phase in each material. Here we note that, for the steady-state case, applying
Equation 8-6 to all surface species will lead to a singular Jacobian matrix. To obtain a
well posed system of equations, one must introduce the additional requirement that
all surface site fractions, Z k , sum to one:
122
RD01400-C01-004-001A
Theory Manual
Equation 8-8
l
K s ( n, m )
1 =
Zk
k = K s ( n, m )
Equation 8-8 may then be solved directly in place of one of the surface-species
balances in Equation 8-6. However, this approach results in the numerical round-off
error in the calculation of the all of the s k s to be assigned to one equation
(Equation 8-8). This may cause problems in the case when that surface-site fraction is
small. Instead, we chose to apportion the additional constraint represented by
Equation 8-8 over all of the equations represented by Equation 8-6 for the surface
phase n according to the size of the surface site fraction, as follows:
Equation 8-9
s k, m
Z k d n, m
dZ
--------k- = k ----------- ----------- -------------- n, m n, m dt
dt
l
K s n, m
Zk
+ ----- 1 Z l
l = K fs n, m
k = K s ( m ), , K s ( m ) ;m = 1, , M
Note that the time constant for the last term on the right-hand-side of Equation 8-9 is
arbitrary, and we have chosen because it is dimensionally correct and physically
significant for the well mixed reactor model. For transient cases, the last term on the
right-hand side is set to zero and a true transient for all species is solved. If
Equation 8-9 is summed over all surface species in phase n and combined with
Equation 8-7, then the following relation results:
Equation 8-10
l
l = K s ( n, m )
1 l = K s ( n, m ) Z
1 l = K s ( n, m ) Z
d 1
Z l
l
f
f
l = K fs ( n, m ) l
d n, m
l = K s ( n, m )
l = K s ( n, m )
-------------------------------------------------------------- = ---------------------------------------------- ----------------------------------------------dt
dt
n, m
which is equivalent to the sum of the Z k s being equal to one, with a false transient.
8.3.3
RD01400-C01-004-001A
123
CHEMKIN 4.0
bulk phase, the composition of the bulk phase may change with time, requiring
solution of the bulk-species mole fraction. For bulk phases that are being deposited,
the mole fractions of the bulk-phase species are assumed to be initially unknown and
are determined through time-dependent species balance equations for each bulk
species:
Equation 8-11
db b
---( A m L n, m X k C n, m W k ) = A m W k s k, m
dt
or
Equation 8-12
d b
b
bd
b
A m L n, m C n, m W k ----- ( X k ) + A m W k X k ----- ( L n, m C n, m ) = A m W k s k, m
dt
dt
b
Here X k is the bulk-species mole fraction of the kth species in phase n of the mth
b
material. L n, m is the film thickness for the nth bulk phase of the mth material. C n, m is
the average molar concentration of the species in the nth bulk phase of the mth
material.
The molar growth rate of the bulk phase n , can be written as
Equation 8-13
l
b
d ( A m L n, m C n, m )
K b ( n, m )
------------------------------------- =
dt
A m s k, m
k = K b ( n, m )
n, m = C n, m W n, m
n = N b ( m ), , N b ( m ) ;m = 1, , M
b
W k dX k
b
-------A m L n, m n, m -----------b
W n, m dt
K b ( n, m )
= A m W k s k, m A m W k X k
s l, m
l = K b ( n, m )
124
RD01400-C01-004-001A
Theory Manual
b
dX k
--------- =
dt
K b ( n, m )
b
W n, m
-------------------b
L n, m n, m
b
s k, m X k
s l, m
l = K b ( n, m )
We define the film-thickness length scale, L n, m , somewhat arbitrarily. For steadystate, we define it as the linear growth rate of the bulk phase multiplied by the reactor
residence time, as given in Equation 8-17. For transient cases, we use Equation 8-17,
but replace the residence time with the time-step value.
Equation 8-17
l
L n, m =
b
W n, m
----------- b
n, m
K b ( n, m )
s k, m
k = K b ( n, m )
For deposition phases, then, Equation 8-16 and Equation 8-17 provide a false
transient equation that is employed for the bulk species:
Equation 8-18
b
dX k
s k, m
1
b
--------- = --- -------------------------- Xk
l
K b n, m
dt
s
l, m
l = K b n, m
Note that L n, m does not appear in Equation 8-18; CHEMKIN makes no assumption
about the total amount of bulk phase deposits, other than that their total amount is
small compared to the total volume of the reactor. Equation 8-18 is only used when
the right-hand side is greater than zero; i.e. when there is net deposition of the bulk
species. Under etching conditions, the bulk species equations must be treated
differently.
8.3.4
RD01400-C01-004-001A
125
CHEMKIN 4.0
Equation 8-19
b0
Xk
b
X k
K b ( n, m )
b
--------- = -------------------- + ----- 1
Xl
dt
f
l = K ( n, m )
b
dX k
b
Xk
b0
where X k is the user-supplied initial estimate for the mole fraction of species k in
bulk phase n , normalized so that their sum over all bulk-phase species is equal to
one. Here, the residence time in Equation 8-19 is replaced with the time-step in
transient simulations.
o not conserve surface sites, if this is a desired feature of the problem. In such cases,
the surface site density n, m , for every phase n whose total number of sites may not
be conserved, becomes an additional solution unknown. The total number of surface
phases on each material m , for which this condition may hold, is defined as equal to
the variable N ( m ) . For each material m , there are N ( m ) additional equations to be
solved, corresponding to N ( m ) additional unknowns. The equation for the time
derivative of surface site densities is given by
Equation 8-20
l
K s ( n, m )
d n, m
------------- =
dt
k s k, m
n = N ( m ) , , N ( m )
k = K s ( n, m )
8.3.5
Equation 8-21
------------- =
dt
inlet
(j)
i=1
N PSR
(j)
( j ) dV
(j)
(j)
Q loss + Q source P ------dt
Kg
Kg
(j)
(
r
)
(
r
)
*
(
j
)
*
*
m i ( Y k, i h k, i ) + m R rj ( Y k h k ) m Y k h k
k=1
k=1
r=1
k=1
Kg
(j)
j = 1, N PSR
126
RD01400-C01-004-001A
Theory Manual
The total internal energy U sys consists of the internal energy of the gas, surface
phases, deposited or etched solid phases, and walls. Q loss is the net heat flux
directed out of the reactor. Q loss can either be specified directly as a constant (QLOS
keyword) or can be specified in terms of a constant heat transfer coefficient, h t , and
ambient temperature, T o (HTRN keyword), as follows:
Equation 8-22
Q loss = Ah t ( T T o )
Q source refers to energy deposited into the gas in the reactor. The term P ( dV dt )
represents the work done by the control volume on the external world. For plasma
systems, this can represent the power deposited through Joule heating into the
plasma by acceleration of charged species along electric-fields. This term will be
discussed in more detail in the description of the electron energy equation below.
The time derivative of the internal energy can be equated with the time derivative of
the enthalpy, minus the time rate of change of the product of pressure and volume:
Equation 8-23
U sys = H sys PV
Equation 8-24
N
U sys = H gas +
l
b
m = 1n = N
( H n ( m ) PV n ) + H walls PV PV walls*
f
s
dH gas
dU sys
------------- = ------------+
dt
dt
RD01400-C01-004-001A
l
b
m = 1n = N
f
s
dH n ( m ) dH walls
dV
dP
------------------- + ---------------- P ------- V ------dt
dt
dt
dt
127
CHEMKIN 4.0
We neglect the term on the right-hand-side that represents the heat capacity
contribution from the walls. Expansion of the gas-phase contribution in Equation 8-25
yields a heat balance for each reactor module:
Equation 8-26
Kg
Kg
Kg dY
dT
dH gas
d( V)
d ( Vh )
------------- = ------------------ = V Y k c pk --------k + V h k --------k + Y k H k --------------dt
dt
dt
dt
dt
k = 1
k = 1
k=1
where h is the specific enthalpy of the gas mixture equal to the sum of the product of
the species mass fraction and the pure species specific enthalpy. Note that in a multitemperature system, the species enthalpies are evaluated at the species temperature
T k , which may differ from the mean gas temperature. In thermal systems, all T k equal
T , the gas temperature. c pk represents the species specific heat capacity at constant
pressure. Expansion of the bulk and surface contributions to Equation 8-25 yields:
Equation 8-27
l
N b(m)
m = 1 n = N f (m)
b
l
N b(m)
M
Hd n ( m )
------------------- =
dt
b
b
b
b
d
d
- ( A m L n, m C n, m ) + A m L n, m C n, m ----- ( h n ( m )W n, m ) +
h n ( m )W n, m ---dt
dt
m = 1 n = N f (m)
b
l
l
N s(m)
K s ( n, m )
M
m = 1 n = N f ( m ) k = K f ( n, m )
s
s
dh k
d ( Am ck Wk )
+ A m c k W k --------
h ---------------------------dt
dt
If we neglect the time dependence of the bulk- and surface-phase species enthalpies
and molecular weights, and make use of Equation 8-11, Equation 8-27 is greatly
simplified to:
Equation 8-28
M
N b(m)
m = 1 n = N f (m)
s
dH n ( m )
------------------- =
dt
K b(m)
A m s k, m W k h k
m = 1 n = K f (m)
s
128
RD01400-C01-004-001A
Theory Manual
Combining Equation 8-1, Equation 8-2, Equation 8-21, Equation 8-25, Equation 8-26,
and Equation 8-28 gives the transient energy equation for solving the gas
temperature, as follows:
Equation 8-29
Energy Equation
(V)
N
inlet
(j)
dT e ( j )
dT
=
c p ( 1 Y e ) ------ + Y e c pe --------dt
dt
(j)
(j)
m i
i=1
(j)
Kg
*
*
Y k, i ( h k, i
hk )
(j)
N PSR
k=1
Kg
(j)
( hk k ) Wk
m=1
k=1
(j)
Am
r=1
K tot
(r)
Kg
R rj
(r)
(r)
(j)
Yk ( hk hk )
k=1
(j)
(j)
(j)
( h k s k ) W k Q loss + Q source + V
k=1
( j ) dP ( j )
------dt
Here we define c p as the mean gas specific heat excluding the contribution of the
electrons, since we assume that the electron temperature may be significantly
different from the gas temperature. All other species are assumed to be in thermal
equilibrium at the gas temperature T . In other words,
Equation 8-30
Kg
cp =
Y k c pk
k = 1 ;k e
where the subscript, e , indicates the electron species. When no electrons are present
Equation 8-29 reverts back to thermal-equilibrium case and the mean specific heat is
merely the mass-averaged value of all species components. The actual form of the
gas energy equation solved is the result of subtracting the electron energy equation
from Equation 8-29. This form is presented after the introduction of the electron
energy equation below.
8.3.6
RD01400-C01-004-001A
129
CHEMKIN 4.0
Reactor M
Heat Flux
T = T(M)
Heat Flux
Reactor N
T = T(N)
Heat Flux
Surroundings
If heat transfer is allowed between reactor M and reactor N, (see Figure 8-3), the
transient energy balance equation for reactor M becomes:
Equation 8-31
(V)
N
inlet
(M)
dT e ( M )
dT
c p ( 1 Y e ) ------ + Y e c pe --------=
dt
dt
(M)
(M)
m i
i=1
(M)
Kg
*
*
Y k, i ( h k, i
k=1
Kg
(M)
( hk k ) Wk
k=1
M
Q loss
hk )
(M)
Q source
N PSR
(M)
N PSR
(r)
Kg
R rM
r=1
k=1
K tot
(M)
(M)
Am
( h k s k ) W k
m=1
k=1
(r)
(r)
(M)
Yk ( hk hk
+V
( M ) dP ( M )
------dt
(M)
Q N M
N=1
(M)
where Q N M is the heat flow coming from reactor N to reactor M and is given as:
Equation 8-32
130
RD01400-C01-004-001A
Theory Manual
The first term of Equation 8-32 describes the heat flow caused by localized heat
transfer modes such as conduction and convection and the second term is an
optically-thin model for thermal radiation heat transfer between the two reactors. The
area, A e , is the external surface area that is available for heat transfer, which may
be different from the internal surface area available for surface chemistry. Since
( hA e ) N M = ( hA e ) M N and ( A e ) N M = ( A e ) M N , the properties of a "heat
exchange matrix" can be obtained from Equation 8-32, with the additional
relationships:
Equation 8-33
(N)
(M)
Q M N = Q N M
and
Equation 8-34
(M)
Q M M = 0
In this way, the total energy among the reactors is conserved in the absence of any
external heat loss. The user input parameters that control this reactor-to-reactor heat
transfer are described in the CHEMKIN Input Manual (QXCO and QXRA keywords).
8.3.7
RD01400-C01-004-001A
131
CHEMKIN 4.0
Figure 8-4
Schematic of the gas phase control volume in the absence of the reactor wall.
Heat Generated by Surface
Reactions
However, when the reactor wall is included in the energy balance, an additional
equation is needed to solve for the wall temperature since the wall temperature can
be different from the temperature of the gas mixture. To this end, we consider a new
reactor-wall control volume, so that the wall energy equation can be established.
Furthermore, the original gas-phase control volume is modified to make the gas-wall
interface part of the reactor-wall control volume. The new configuration of the control
volumes and their thermal energy interactions are shown in Figure 8-5.
Figure 8-5
Schematic of the gas-phase and wall control volumes for energy balance
Heat Generated by Surface
Reactions
External
Heat Flux
Surroundings
Gas Phase
Control Volume
Wall/Surface
Control Volume
Accordingly, the wall energy-balance equation for reactor M is given in Equation 8-35.
Equation 8-35
m=1
(M)
Am
K tot
( h k s k )
(M)
Wk
M
Q gas wall
k=1
(M)
Q loss
N PSR
(M)
Q N M
N=1
132
RD01400-C01-004-001A
Theory Manual
Modified Gas Energy Equation for Reactor M Accounting for Wall-energy Balance
(V)
N
inlet
(M)
dT e ( M )
dT
=
c p ( 1 Y e ) ------ + Y e c pe --------dt
dt
(M)
i=1
(M)
Kg
(M)
m i
*
*
Y k, i ( h k, i
hk )
k=1
Kg
(M)
( hk k ) Wk
k=1
M
Q gas wall
(M)
Q source
N PSR
(r)
Kg
R rM
r=1
k=1
K tot
(M)
(M)
Am
( h k s k ) W k
m=1
k=1
(r)
(r)
(M)
Yk ( hk hk
N PSR
(M)
+V
( M ) dP ( M )
------dt
(M)
Q N M
N=1
In the above equations, the general term for heat-transfer to the external environment
has been replaced by a term accounting for heat-transfer between the gas in the
reactor and the wall material, where the wall material now interacts exclusively with
the external environment. Since convection is the dominant heat-transfer mechanism
between the gas mixture and the reactor wall, the heat flux can be cast into the
following form:
Equation 8-37
To enable Equation 8-36 and Equation 8-37, the user must supply the gas-to-wall
convective heat-transfer coefficient as well as the thermal mass and heat capacity of
the reactor wall. These input parameters are provided through the CHEMKIN User
Interface.
8.3.8
The net gas-phase chemical production rate of each species, k , results from the
competition between all the chemical reactions involving that species. For reactions
stated as reversible in the GAS-PHASE KINETICS and SURFACE KINETICS input files, the
reverse rate parameters can be determined from the forward rate coefficients and the
thermodynamic properties of the species involved in the reaction, via the law of mass
RD01400-C01-004-001A
133
CHEMKIN 4.0
action. In these cases, SURFACE KINETICS requires activities to be defined for bulkphase species, in order to determine the species thermochemistry. In these reactor
b
models, the program sets the bulk activities, a k , equal to their mole fraction in the
bulk phase:
Equation 8-38
b
a k ( T, P, X k ( n, m ) ) = X k ( n, m )
More details regarding chemical reaction specifications, reaction-rate determinations,
and thermochemical properties, are available in the Chapter 3 and Chapter 4.
8.4
41.J. B. Heywood, Internal Combustion Engines Fundamentals, McGraw-Hill Science/Engineering/Math, New York, 1988.
134
RD01400-C01-004-001A
Figure 8-6
Theory Manual
For the time-dependence of the engine cylinder volume, consider the diagram in
Figure 8-6. The connecting rod length is given by L C , while the crank arm radius is
given by L A . The volume swept by the piston (cross-hatched area) is represented by
the dashed rectangle above the piston. The clearance volume V c is represented by
the open areas above and below the swept volume. The maximum swept or displaced
volume is given by:
Equation 8-39
2
V s, max = ---- D L A
2
where D is the cylinder bore diameter as shown in Figure 8-6. The engine
compression ratio C is defined as the ratio of the maximum total volume to the
clearance volume,
RD01400-C01-004-001A
135
CHEMKIN 4.0
Equation 8-40
Compression Ratio
V s, max + V c
C = ------------------------Vc
Note that in some cases it is only necessary to define the compression ratio, and it is
not strictly necessary to define the clearance volume, since it is only used to scale the
calculated volume. However, if the user is interested in performing post-analyses for
engine efficiency, for example, or in the case that the IC Engine heat-transfer
coefficient correlation is used, the clearance volume should be specified with a
physical value. By default, a value of 1.0 is assumed. The other parameter required
for the IC engine problem is R , the ratio of the connecting rod length L C to the crankarm radius L A :
Equation 8-41
L
R = -----CLA
Finally, the user must specify the rotation rate of the crank arm, where
Equation 8-42
d
-----dt
With these definitions, one can derive the relationship between the total volume
available for combustion in the cylinder as a function of time, scaled by the clearance
volume:41, p. 134
Equation 8-43
V(t)
1- [ R + 1 cos R 2 sin 2 ]
---------- = 1 + C
-----------Vc
2
while the time derivative of the volume is:
Equation 8-44
d ( V Vc )
C1
1 + cos
--------------------- = ------------- sin --------------------------- 2
2
2
dt
R sin
Equation 8-39 through Equation 8-44) provide the volume and volume-derivative
functions of time, which allows solution of the general equations for species and
energy conservation discussed in Section 8.3.
136
RD01400-C01-004-001A
8.4.1
Theory Manual
Equation 8-45
Q wall = hA ( T T wall )
where the user specifies T wall and the heat transfer coefficient, h , is obtained from
the following generalized heat transfer correlation41, p. 134 based on user-specified
constants a , b , and c :
Equation 8-46
Nu h = aRe Pr
Nu h is the Nusselt number for heat transfer, Re is the Reynolds number, and Pr is
the Prandtl number. These are defined according to:
Equation 8-47
hD
Nu h -------
Equation 8-48
DS p
Re ------------
Equation 8-49
Cp
Pr ---------
where is the gas conductivity, S p is the mean piston speed ( = 2L A ), and is the
gas viscosity. In calculating the heat transfer coefficient using Equation 15-9, the gas
properties are assumed to be those of air at the specified initial conditions, since, for
typical operating conditions, the molar concentrations of fuel and its by-products are
relatively dilute. The area available for heat transfer, A , includes the cylinder walls
2
(time-varying) and end surfaces ( 2 D 4 ). Accordingly, when the IC Engine heattransfer correlation is invoked, the user must also specify the engine bore diameter,
D.
RD01400-C01-004-001A
137
CHEMKIN 4.0
8.4.1.1
Equation 8-50
Dw
Re ------------
To obtain the average cylinder gas velocity, Woschni proposed a correlation that
relates the gas velocity to the mean piston speed and to the pressure rise due to
combustion, ( P P motored ) , as given in Equation 8-51.
Equation 8-51
v swirl
Vd Ti
w = C 11 + C 12 ---------S p + C 2 ---------( P P motored )
Pi Vi
S
p
8.5
Plasma Systems
The treatment of plasmas with the Plasma Reactor or Closed Plasma Reactor model
requires that power is deposited uniformly into the plasma bulk or, alternatively, that
the electrons diffusion and thermal conduction transport processes are fast relative to
the rates of electron-driven chemical kinetics. In addition, we assume that the
chemical reaction-rate coefficients are independent of reactor conditions. This
assumption may be invalid in the case where the electron energy distribution function
deviates significantly from Maxwellian conditions.41, p. 134 However, modeling plasma
138
RD01400-C01-004-001A
Theory Manual
systems in this manner still provides important insight into the reactor
chemistry.42,43,44,45,46 Finally, the characterization of a quasineutral bulk plasma
depends on a thin-sheath approximation, where the sheath thickness is much smaller
than the reactor-chamber dimensions.
In this section we describe the equations and formulations specific to homogeneous
plasma reactor models.
8.5.1
Equation 8-52
d(j)
---( e Vu e ) =
dt
inlet
(j)
N PSR
*(j) *
*
m i Y e, i h e, i
i=1
(j) (j) (j)
m Y e h e + e W e h e V +
(r)
(r) (r)
R rj Y e h e
r=1
M
( s e, m he Am )
(j)
elas
inel
m=1
where e is the electron mass density (equal to the product of the electron number
density and the electron mass). h e refers to the electron enthalpy of newly created
electrons in the gas-phase; when electrons are formed from the ionization of a
relatively cold neutral, the electron is assumed to originate close to the neutral
temperature. The energy required to thermalize new electrons is therefore taken into
account. At the surface, electron losses are assumed to dominate electron emission,
so that no equivalent term is included to account for new electrons coming off the
surface with thermal energies equal to the surface temperature. The electron
enthalpy loss at the surface is therefore calculated from the net production rate of
electrons due to surface reactions on each material, s e, m , and the electron enthalpy,
42.M. A. Lieberman and R. A. Gottscho, in Physics of Thin Films, edited by M. Francombe and J.
Vossen Academic Press, New York, 1993.
43.C. Lee, D. B. Graves, M. A. Lieberman, and D. W. Hess, Journal of the Electrochemical Society
141:1546 (1993).
44.E. Meeks and J. W. Shon, IEEE Transactions on Plasma Science 23:539 (1995).
45.E. Meeks and J. W. Shon, Journal of Vacuum Science and Technology A 13:2884 (1995).
46.E. Meeks, R. S. Larson, S. R. Vosen, and J. W. Shon, Journal of the Electrochemical Society
144:358 (1997).
RD01400-C01-004-001A
139
CHEMKIN 4.0
h e . The second- and third-to-last terms on the right-hand-side refer to the collision
energy lost by the electrons both from elastic, momentum-transfer collisions, and from
inelastic collisional processes. The inelastic collisions may include both excitation
reactions, as well as chemical reactions resulting from electron-impact collisions. The
source term Q source differs from Q source in Equation 8-29 in that it represents only
that power deposited to the electrons, rather than to the plasma as a whole. In
particular, some of the deposited power may contribute to heating of ions in the
plasma bulk, or accelerating ions through the plasma sheath. We therefore define the
electron-energy source term as:
Equation 8-53
sheath
plasma
The internal energy of the electron and electron specific heats are defined by:
Equation 8-54
P
RT
5R
3R
3R
5R
u e = h e -----e = --- ------- T e ---------e = --- ------- T e ;c pe = --- ------- ;c ve = --- ------2 We
2 We
2 We
2 We
e
We
If we assume h e = ( 5 2 )RT W e , substitute Equation 8-53, and Equation 8-54 into
Equation 8-52, and subtract Equation 8-2 multiplied by c pe T e , we arrive at:
Equation 8-55
N
(V)
(j)
N PSR
(r)
(j)
dT
R dY
Y e c ve --------e ------- T e --------e
dt W e dt
inlet
(j)
*(j) *
*
m i Y e, i c pe ( T e T e ) +
i=1
(j)
(r)
(r)
(j)
elas
inel
r=1
Here the second term on the right-hand-side represents the thermalization energy
required for newly created electrons. The loss and source terms in Equation 8-55 and
Equation 8-53 are defined as follows:
Equation 8-56
elas
Q loss
3VR
= ----------------e ( T e T )
We
Kg
k = 1 ;k e
We
------- ek
Wk
140
RD01400-C01-004-001A
Theory Manual
Equation 8-57
I ei
inel
Q loss = V H r q r + Q other ( T e )
r=1
Equation 8-58
Ki
M
sheath
Q ions
s i, m i Am
m = 1i = 1
and
Equation 8-59
Ki
plasma
Q ions
i ( hi ( Tion ) hi ( T ) )V
i=1
RD01400-C01-004-001A
141
CHEMKIN 4.0
Equation 8-59, we introduce an ion temperature T ion to capture the source energy
that is deposited into the ions, although we are not solving an ion energy equation
explicitly. The ion temperature in the plasma bulk is specified directly by the user and
assumed to be constant.
8.5.2
Equation 8-60
(V)
(j)
dT R d ( Y e T e -)
c p ( 1 Y e ) ------ + ------- -----------------dt W e dt
(r)
m=1
8.5.3
i=1
(j)
hk )
(j)
Kg
(j)
m i
*(j)
Y k ( h k, i h k )
(j)
k = 1 ;k e
(j)
( h e e ) W e
k = 1 ;k e
K tot
(j)
Kg
(r)
(r)
Yk ( hk
k = 1 ;k e
r=1
R rj
inlet
Kg
N PSR
(j)
(j)
Am
( s k, m hk )
(j)
(j)
elas ( j )
W k Q loss + Q loss
inel ( j )
plasma ( j )
+ Q loss + Q ions
sheath ( j )
+ Q ions
k=1
Equation 8-61
RT 1 2
U Bohm = ---------e
Wi
For an electronegative gas, this expression must be modified to account for the
presence of negative ions and their effect on the plasma sheath behavior. In the limit
of a purely electronegative gas, the ion flux to a surface would be limited by the
thermal speed of the ion.
142
RD01400-C01-004-001A
Theory Manual
To accommodate a large range of conditions, then, we use the correction to the Bohm
velocity derived by Braithewaite and Allen,47 as follows:
Equation 8-62
RT 1 2 ( c e + c n, tot )T i
------------------------------------U Bohm, mod = ---------e
Wi
( c n, tot T e + c e T i )
12
where c n, tot is the sum over all negative ions of the product of the ion species
concentration and its integer electronic charge.
The Plasma Reactor Reactor Models allow the user to specify this constraint in one of
two ways. The first way is to use the Plasma Reactor Bohm factor setting. This
keyword includes the input of a correction factor to the above Bohm velocity. When
this option is included, the production rate of each ion by each surface reaction will be
scaled, such that the net production rate of the ion on each material is given by
Equation 8-63
12
( c e + c n, tot )T i 1 2
RT e
s
The second way to apply the modified Bohm criterion, is to use the SURFACE KINETICS
BOHM auxiliary keyword. In this case, the individual reaction for which the auxiliary BOHM
keyword is included, will have a rate of progress calculated in the SURFACE KINETICS
routines as
Equation 8-64
RT 1 2
q = c i U Bohm, mod = c i ---------e
Wi
where, in this case, the correction factor is the first reaction-rate coefficient
specified on the reaction line in the SURFACE KINETICS input file. This rate-of-progress
of the Bohm reaction will then be modified within the Plasma Reactor Reactor Model
to account for the presence of any negative ions in the plasma; i.e. each reaction with
a BOHM auxiliary keyword ultimately has a rate of progress defined as:
Equation 8-65
RT 1 2 ( c e + c n, tot )T i
------------------------------------q mod = c i ---------e
Wi
( c n, tot T e + c e T i )
12
RD01400-C01-004-001A
143
CHEMKIN 4.0
The main difference between these two approaches arises when an ion participates in
more than one surface reaction subject to the Bohm criterion. In the first approach,
the net ion flux to the surface will be automatically scaled to the Bohm-limiting flux
modified for electronegative gases and the user-defined correction factor . In the
latter approach, each reaction will be subject to the Bohm limit independently. In that
case, it is up to the user to make sure that the reaction coefficients add up to the
desired overall correction factor, for all the reactions involving a particular ion. This
overall correction factor is often used to account for spatial variations in ion density or
transport limitations in the reactor being modeled. For example, the correction factor
may be set equal to an estimation of the ratio of the ion density at the sheath edge to
the ion density in the bulk of the plasma.43,44,45
8.5.4
Equation 8-66
M
K = Kg +
[ Ks ( m ) + Kb ( m ) ]
m=1
J = L+K+
N ( m )
N PSR
m=1
144
RD01400-C01-004-001A
CHEMKIN 4.0
RD01400-C01-004-001A
145
CHEMKIN 4.0
The composition and temperature in the PaSR are described by a probability density
function (PDF). This composition PDF is a subset of the joint velocity-composition
PDF, because the flow field in the PaSR is assumed to be spatially homogeneous.
Velocity fluctuations are also ignored; that is, the PDF is over scalars only, but is not a
delta-function in scalar space because reactants, intermediates, and products are not
mixed at the molecular level.
The PaSR is related to and bounded by other models commonly used in combustion.
When the mixing time scale approaches zero, the mixing process becomes fast
enough that the properties inside the PASR are homogeneously mixed at the
molecular level. In this limit, the PaSR becomes a PSR, for which the joint scalar PDF
degenerates to a delta function in the composition space and the mean residence
time is the sole controlling time scale. In the other extreme limit (large mixing time),
there is no mixing among the pockets of gas in the PaSR; consequently, the PaSR
consists of segregated reactive mixtures. The average statistics are the sum of the
properties of the reactive mixture pockets weighted by the PDF of their ages inside
the PaSR. In this case the PaSR acts like a plug-flow reactor, for limit of no mixing
and relatively large mean residence times (closed system).
The PaSR may be used as a stand-alone model for studying turbulent combustion or
other reactor systems where mass transport may be a rate-limiting factor. Or a PaSR
can be used to simulate the sub-grid turbulent mixing and chemical reactions in a
computational fluid dynamic (CFD) cell.
One of the important concerns of turbulent reactive flows, especially turbulent flames,
is the coupling between chemical reactions and turbulence. The interaction between
chemical reactions and fluid dynamics is best described by the Damkhler number,
which is defined as the ratio of characteristic flow time and the characteristic chemical
reaction time. Due to the large spectrum of chemical times in a multicomponent
chemical system, the Damkhler number can also span a large spectrum. As the
Damkhler number corresponding to a specific chemical reaction approaches infinity,
the reaction responds much faster to the flow so that it approaches equilibrium
conditions. On the other hand, if the Damkhler number of a reaction is small, the
reaction is considered frozen. Only when the Damkhler number is of the order of
unity, are the interactions between the reaction and the fluid dynamics strong. In this
case, the reaction becomes one of the "controlling" steps of the process. Depending
on the flow time, the same chemical process can be controlled by different sets of
reactions.
146
RD01400-C01-004-001A
Theory Manual
mix = C D --
where C D is usually treated as a constant but its value varies for different flow
configurations. The ratio of turbulent kinetic energy to its dissipation rate, ,
represents the time scale of the energy-containing eddies, which characterize the
turbulent mixing action.
9.1
RD01400-C01-004-001A
147
CHEMKIN 4.0
equation are impractical because of the large . Pope50 developed a Monte Carlo
algorithm that makes solving the PDF transport equation practical for general
turbulent reactive flows. Rather than considering P ( ;x,t) explicitly, the dependent
variable in the simulation is represented by an N-member ensemble:
Equation 9-2
(1)
(2)
(n)
, , , , ,
(N)
1
Q ( ) ---N
Q(
(n)
n=1
In the limit of large N , Pope50, p. 148 showed that the ensemble average Q ( )
converges to the corresponding density-weighted average, i.e.,
Equation 9-4
() =
Q ( ) Q
Q ( )P ( ) d( )
For the general multiple reactive scalars, the transport equation for the joint PDF in
the PaSR is derived by integrating the governing equation of the single-point joint
scalar PDF over the reactor volume. The resulting PDF transport equation for the
PaSR is
Equation 9-5
P (,t)
-------------------- =
t
K tot
= 1, = 1
K tot
=1
{ P , i (,t) P (,t) }
i=1
-------------------{ = P (,t) }
148
RD01400-C01-004-001A
Theory Manual
The first two terms on the right hand side of Equation 9-5 represent the effects of
chemical reaction and the through-flow on the joint PDF, respectively. The last term
represents the effect of micro-scale mixing on the PDF, which requires the use of a
mixing model. The mixing model mimics the finite rate mixing of particles in the
stochastic simulations. Two widely used mixing models are employed as options in
the current PaSR model.
The simplicity of using a Monte Carlo method and a scalar PDF permits us to carry
out simulations with detailed chemistry without significant computing costs.
9.2
Equation 9-6
= 1, = 1
--------------------- { = P (,t) } =
1
----------
P (',t)P ('',t)H ( ', ''' ) d 'd '' P (,t)
mix ', ''
where H is the transitional probability defined as H ( ', '' ) = { 1 '' ' for
[ ', '' ] otherwise 0 .
The Interaction-by-Exchange-with-the-Mean (IEM) or the Linear-Mean-SquareEstimation (LMSE) model52 is shown in Equation 9-7
Equation 9-7
= 1, = 1
2
C
-------------------{ = P (,t) } = ------------- ---------- { ( )P (,t) }
2 mix
RD01400-C01-004-001A
149
CHEMKIN 4.0
1
unmixedness = --------------------------------1 + res 3 mix
Equation 9-9
1
unmixedness = ------------------------------------1 + C res mix
When C = 1 3 , the two mixing models should produce identical levels of
unmixedness for a given mixing time scale. Figure 9-1 shows the unmixedness
versus mixing frequency from simulations of a pure mixing problem for R = 1ms .
The numerical results of CHEMKIN show excellent agreement with the theoretical
values.
150
RD01400-C01-004-001A
Figure 9-1
9.3
Theory Manual
Unmixedness vs. mixing frequency for PaSR of stoichiometric H2/air mixture with 1 ms residence time
Reactor Equations
The PaSR consists of an adiabatic chamber having M inlet streams and one outlet.
Steady flows of reactants are introduced through the inlets with given gas
compositions and temperatures. The reactor pressure is assumed to be constant.
Since there is no surface reaction, the mass flow rate at the outlet must be equal to
the sum of the mass flow rates of all inlets so that the mass is conserved. In order to
represent the evolution of the PDF properly by a stochastic scheme, PaSR addresses
all problems in a transient manner. The overall mass balance for the gas mixture
inside the PaSR is
Equation 9-10
d------------ M = d-------------------( V )- =
dt
dt
m i m o
= 0
i=1
where m i is the mass flow rate of the i th inlet and m o is the through-flow mass flow
rate. The average properties of the PaSR are obtained from the ensemble of particles
inside the reactor. Each particle is regarded as an independent PSR and interacts
with others only through the molecular mixing process. Therefore, the conservation of
energy and species is applied to an individual particle rather than to the reactor.
RD01400-C01-004-001A
151
CHEMKIN 4.0
k
1
k
- { m i ( Y i, k Y k ) } + ---------------------------- = ----------
(
n
)
mo R
dt
i=1
(n)
Wk k hk
1
dT
- m i Y i, k ( h i, k h k ) -------------------------------------- = ---------------------(n)
(n) (n)
dt
C p m o R
k = 1
k = 1 Cp
i=1
In the above equations, the angled bracket ( ) indicates the ensemble average that
we use to approximate the density-weighted average in the simulation. The average
residence time of the reactor, R , is calculated as
Equation 9-13
V
R = -----------m
o
9.4
Stochastic Simulation
A time marching scheme with a time-step size of t is used to solve Equation 9-13
and the stochastic simulation is carried out by the following sequential procedures
with N statistical particles:
9.4.1
Through-flow (Convection)
The number of correct particles in a time step is chosen as:
Equation 9-14
Nc = N t R
In the first procedure, we set the properties of these N c particles from the stochastic
ensemble, the properties of the inlet mixture. Property changes of the statistics due to
mixing and chemical reaction will be computed by the next two procedures.
152
RD01400-C01-004-001A
9.4.2
Theory Manual
Molecular Mixing
With the modified Curl's mixing model,
Equation 9-15
N m = C m N t mix
particles are chosen to mix with each other. C m in Equation 9-15 is a parameter for
the modified Curl's model. If the IEM (LMSE) model is used, Equation 9-9 is solved to
determine the statistics over a period of t .
9.4.3
Chemical Reaction
Next, we compute chemical kinetics for each particle by integrating the species and
energy equations (Equation 9-11 and Equation 9-12) over a period of t .
These same procedures (Section 9.4.1 through Section 9.4.3) are repeated for the
next time step until the end of the simulation time is reached.
RD01400-C01-004-001A
153
CHEMKIN 4.0
154
RD01400-C01-004-001A
CHEMKIN 4.0
10
10
RD01400-C01-004-001A
155
CHEMKIN 4.0
transverse gradients implies that mass-transfer limitations are absent, once again
enhancing the reactor performance. Along with these practical advantages, the plug
flow reactor is computationally efficient since it is modeled using first-order ordinary
differential equations (ODEs), and no transport properties are needed.
The equations governing the behavior of a plug-flow reactor are simplified versions of
the general relations for conservation of mass, energy, and momentum.54 They can
be derived most easily by writing balances over a differential slice in the flow direction
x , with the stipulations that (a) there are no variations in the transverse direction, and
(b) axial diffusion of any quantity is negligible relative to the corresponding convective
term. In this way the overall mass balance (continuity equation) for the gas is found to
be
Equation 10-1
dA
du
d
u ------- + A ------ + uA ------ =
dx
dx
dx
a i, m
m=1
s k, m Wk
k=1
Here is the (mass) density and u the axial velocity of the gas, which consists of K g
species; W k is the molecular weight of species k , and s k is the molar production rate
of this species by all surface reactions. The quantities A and a i, m are the crosssectional (flow) area and the effective internal surface area per unit length of material
um, respectively, in the reactor. Both A and a i, m can change as arbitrary functions of
x . Equation 10-1 simply states that the mass flow rate of the gas can change as a
result of generation or consumption by surface reactions on all materials in the
reactor. A similar equation can be written for each species individually:
Equation 10-2
dY
uA --------k + Y k
dx
Kg
k=1
m=1
M
Here Y k is the mass fraction of species k and k is its molar rate of production by
homogeneous gas reactions. Such reactions cannot change the total mass of the gas,
but they can alter its composition.
54.R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena, John Wiley and Sons, New
York, 1960.
156
RD01400-C01-004-001A
Theory Manual
Energy Equation
M
g
b
g
g
M
dY k
du
dT
1 2
where h k is the specific enthalpy of species k , C p is the mean heat capacity per unit
mass of the gas, T is the (absolute) gas temperature. In the right-hand summation,
s k, m is the molar production rate of bulk solid species k by surface reactions on
material m . The distinction between bulk and surface species is discussed in
Section 1.1 and Chapter 4 of this manual. Equation 10-3 states that the total energy
(enthalpy plus kinetic) of the flowing gas changes due to the heat flux Q e from the
surroundings to the outer tube wall (whose surface area per unit length is a e ) and
also due to accumulation of enthalpy in the bulk solid. It is worth noting that
Equation 10-3 does not involve the enthalpies of the surface site species.
The momentum equation for the gas expresses the balance between pressure forces,
inertia, viscous drag, and momentum added to the flow by surface reactions. Thus,
Equation 10-4
Momentum Equation
dP
du
A ------- + uA ------ + dF
------- + u
dx
dx dx
Kg
m=1
s k, m Wk
a i, m
= 0
k=1
where P is the absolute pressure and F is the drag force exerted on the gas by the
tube wall, to be discussed below. The pressure is related to the density via the idealgas equation of state, as given in Equation 2-4.
Since the heterogeneous production rates s k, m will depend, in general, on the
composition of the surface as well as that of the gas, equations determining the site
fractions Z k of the K s surface species are now needed. Assuming that these species
are immobile, the steady-state conservation equations are simply stated in
Equation 10-5. The surface species conservation equation is applied to every species
in each surface phase n contained on each surface material m .
Equation 10-5
s k = 0
RD01400-C01-004-001A
k = K s, , K s
157
CHEMKIN 4.0
i.e., the net production rate of each surface species by heterogeneous reactions must
be zero. However, we also assume that the total site density for each surface phase is
a constant. As a result, the algebraic equations represented by Equation 10-5 are not
all independent, and for each phase on each material, one of the equations must be
replaced by the condition,
Equation 10-6
Zk
= 1
phase
In order to minimize errors, Equation 10-6 is used to replace Equation 10-5 for the
species having the largest site fraction.
The system of governing equations for the reactor is now mathematically closed.
However, because the residence time of the gas is often a quantity of interest, it is
useful to include an equation that computes it automatically. This is simply
Equation 10-7
d------ = 1--dx
u
Equation 10-1 through Equation 10-7 provide a total of 5 + K g + K s
differential/algebraic relations involving the dependent variables , u , T , P , , Y k
158
RD01400-C01-004-001A
Theory Manual
Qe = U ( T T )
Both T and U must be supplied by the user.
The viscous drag force F is written in terms of a friction factor f as follows:
Equation 10-9
1 2
dF
------- = a i --- u f
2
dx
The friction factor can in turn be expressed as a function of the local Reynolds
number
Equation 10-10
Du Re = ----------
where D is the tube diameter (or the mean hydraulic diameter for a conduit with a
noncircular cross section) and is the gas viscosity. For laminar flow ( Re = 2100 )
the analytical result for round tubes is
Equation 10-11
16
f = -----Re
while for turbulent flow one can use the approximate Blasius formula,54
Equation 10-12
f = 0.0791Re
0.25
This approach is only approximate, especially for noncircular conduits, but viscous
drag is usually of very minor importance in gas-phase reactors. In keeping with this,
and in order to avoid having to calculate transport properties, the gas viscosity is
0.5
computed by scaling the inlet value (supplied by the user) by ( T T in )
and ignoring
the composition dependence.
It remains to specify the initial (inlet) conditions for the reactor. Clearly, values for ,
u , T , P and Y k at x = 0 should be known or easily obtainable from the problem
statement, the ideal gas law, and the reactor geometry, and of course = 0 at this
point. Since there are no derivatives of the Z k in the governing equations, it might
appear that no initial conditions are needed for them. However, the transient solver
RD01400-C01-004-001A
159
CHEMKIN 4.0
dZ k
s k k
-------- = --------dt
k = K s, , K s
Here k is the site occupancy number for species k and is the total site density of
the phase in question. The initial values of Z k for Equation 10-13 are essentially
arbitrary (unless there are multiple steady states), although better guesses will lead to
faster convergence.
10.1
Equation 10-14
dT e R dY e
elas
inel
3AR e
= ---------------- ( T e T )
We
Kg
k = 1 ;k e
We
------- ek
Wk
Equation 10-16
I ei
inel
Q loss
= A
Hr qr + Qother ( Te )
r=1
160
RD01400-C01-004-001A
Theory Manual
Equation 10-17
M
sheath
Q ions
Ki
s j, m j ai, m
m = 1j = 1
and
Equation 10-18
Ki
plasma
Q ions
i ( hi ( Tion ) hi ( T ) )A
i=1
RD01400-C01-004-001A
161
CHEMKIN 4.0
162
RD01400-C01-004-001A
CHEMKIN 4.0
11
11
55.M. E. Coltrin, R. J. Kee, and J. A. Miller, Journal of the Electrochemical Society 131:425 (1984).
56.M. E. Coltrin, R. J. Kee, and J. A. Miller, Journal of the Electrochemical Society 133:1206
(1986).
RD01400-C01-004-001A
163
CHEMKIN 4.0
conditions or boundary conditions, such as surface temperature, flow rate, inlet partial
pressure of the reactants, total pressure, and reactor dimensions. The models are
restricted to two-dimensional geometric representations, using either planar or radial
coordinates for Cartesian or axisymmetric flows, respectively.
The Shear-layer Flow Reactor Models require gas transport properties. These are
determined using the TRANSPORT Pre-processor and subroutine libraries for
calculating thermal diffusion coefficients and for the rigorous calculation of ordinary
multicomponent transport properties. The effects of thermal diffusion, which is the
separation of species of differing size in a temperature gradient may be included
when requrested by the user. Thermal diffusion can have an important effect on
predicted concentration profiles.56 The boundary conditions describing chemical
reactions at the surface are formulated using the SURFACE KINETICS Pre-processor,
while the gas-phase kinetics calculations employ the GAS-PHASE KINETICS Preprocessor.
11.1
Boundary-layer Equations
The boundary-layer equations can again be derived from general conservation
equations that govern the fluid flow coupled with species and energy transport. These
equations describe chemical species production and destruction, and both convective
and diffusive transport. Details of the formulation have been published by Coltrin, et
al.55, p. 163,56, p. 163 The applicability of the equations relies on the existence of a
principal flow direction in which diffusive transport is negligible compared to
convective transport. To simplify the numerical procedure somewhat, we recast the
equations using the Von Mises transformation,57 in which the cross-stream coordinate
is replaced by the stream function as an independent variable. We use an additional
transformation of the stream function56, p. 163 that accounts for possible mass loss (or
gain) in the gas due to deposition (or etching) at the reacting surfaces.
The set of equations describing the channel-flow models are given as follows:
Equation 11-1
Momentum
dm u
u u
u
2 u
u ------ ----- - dm
------- ---------l ------ + dP
------- = ------2 ------ u y ------ + g ( i )
x m dx dx dx
57.F. K. Moore, in High Speed Aerodynamics and Jet Propulsion (Princeton University Press, Princeton, NJ, 1964), Vol. IV.
164
RD01400-C01-004-001A
Equation 11-2
Theory Manual
Species
Y u
dm Y k
u- -----
u --------k ----- - dm
------- ---------l -------- = k W k ----( y Y k V k, y )
x m dx dx
m
Equation 11-3
k = 1, , K g
Energy
2 T
T uc p- dm dm l -----Tu
= ------- ------ u y ------
uc p ------ ---------- ------- ---------
2
x
m
dx dx
m
Equation 11-4
Kg
k=1
k Wk hk
uy -------------m
Kg
k=1
T
Y k V k, y c pk -----
State
RTP =
---------W
We provide the option of choosing from two different transport models. In these
equations when multicomponent transport is used, the diffusion velocity V k, y is given
by
Equation 11-5
V k, y
Kg
X k D k uy T
- --------- ------------ -----=
Wj Dk, j ------- Y k Tm
X k Wm
uy
---------------
jk
V k, y
D km uy X k D k uy T
- --------- --------- ----------- -----= ---------------------X k m
Y k Tm
RD01400-C01-004-001A
165
CHEMKIN 4.0
1
+1
= ------------- u dy
+1 0
The stream function is defined such that there is an equal mass flow rate between two
lines of constant stream function value (between streamlines, when there is no mass
loss). If there is no mass loss to the walls, the reactor walls themselves are
streamlines, i.e., lines of constant streamfunction. The independent variable y then
ranges from zero at one boundary to the total mass flux m at the other. If there is no
mass loss at the surfaces, then the total mass flux is evaluated at the initial condition
and is constant throughout the computation. In this case, the numerical method can
use a mesh in which each mesh point has a specified value of stream function.
However, if mass is lost from the gas then the total mass flux changes along the flow
direction, and the independent variable changes at each mesh point (i.e., a moving
coordinate system). In order to make a new independent variable whose total
magnitude is fixed for the entire problem, we define a new stream function that is
normalized by the local total mass flux
Equation 11-8
= --m
where m is the local value of the total mass flux. Therefore, ranges between 0 and
1, and is not dependent on the total mass previously lost (or gained) at the walls.
The relationships between the physical coordinates ( y and x ) and the transformed
coordinates ( , and x ) are stated in the following equations that define the Von
Mises transformation.
Equation 11-9
dm
---- ----------- + ---------l -----=
vy
x dx x
x y
x
Equation 11-10
dm 1 dm l
1
----= ----- ---- ------- ---- --------- ------ vy ---- ------
x y
x m dx m dx x
m x
166
RD01400-C01-004-001A
Theory Manual
Equation 11-11
1
----= uy ------- = uy ---- ------
x
y x
m x
The total local mass flux m is computed from an equation that accounts for the mass
deposited on each boundary
Equation 11-12
dm
dm
dm
------- = ---------l + ---------udx
dx
dx
The mass flux at the lower boundary (in the asymmetric planar case) is determined
from the convective (Stefan) mass flux to the boundary
Equation 11-13
dm l
--------- = v
dx
y=0
dm u
---------- = vy
dx
y = y max
max
m 0 =
uy dy
0
0
+1
1- --------------y + 1--=
-----------
m
u
RD01400-C01-004-001A
167
CHEMKIN 4.0
s k ( n ) = 0
k = K s ( n ), , K s ( n ) 1
n = 1, , N s
Equation 11-18
l
K s(n)
1 =
zk ( n )
n = 1, , N s
K s(n)
f
Here, K s ( n ) and K s ( n ) represent the first and last species for surface phase n . This
nomenclature is described in more detail in Section 1.2. For one species in each
surface phase, we do not solve the steady state condition given by Equation 11-17,
but instead use Equation 11-18, which requires that the site fractions sum to unity in
each phase.
To summarize the system of equations considered for channel-flow, we solve
Equation 11-1 through Equation 11-4, Equation 11-13, Equation 11-14, and
Equation 11-6 at every gas-phase mesh point. In planar coordinates, we also solve
Equation 11-17 and Equation 11-18 at the upper and lower boundaries. In cylindrical
coordinates (or for a symmetric channel in planar coordinates) we solve
Equation 11-17 and Equation 11-18 only at the upper boundary. This is a parabolic
+1
system of equations, where the dependent variables are: P , y
, u , T , m l , m u ,
and Y k . The equations are subject to a set of algebraic constraints, which are the
equations for the surface site fractions, z k . This system of equations is solved using
the method of lines. We treat the equations as a set of Differential/Algebraic
Equations (DAEs),58 which we solve using the numerical software DASSL.59, 60
58.R. J. Kee and L. R. Petzold, Sandia National Laboratories Report SAND86-8893, 1982.
59.L. R. Petzold, A Description of DASSL, Sandia National Laboratories Report SAND82-8637,
1982.
60.K. E. Brenan, S. L. Campbell, and L. R. Petzold, Numerical Solution of Initial-Value Problems
in Differential-Algebraic Equations North-Holland, New York, (1989).
168
RD01400-C01-004-001A
Theory Manual
At the entrance to the reactor channel, the initial profiles of u , T , and Y k , the
pressure, and the surface site fractions z k must be specified (see Section 11.3 on
species concentration at boundaries). The velocity profile can be either a fully
developed parabolic profile or a flat velocity profile, with an optionally specified
boundary-layer thickness. If the boundary-layer thickness is specified, then a
parabolic profile is assumed only within the boundary layer. The initial gas
temperature across the channel is usually set equal to the initial surface temperature.
However, the user may also specify different gas and surface temperatures. As with
the velocity profile, if a boundary-layer thickness is specified, the application linearly
interpolates the gas-phase temperature between the surface temperature and the gas
temperature over the boundary-layer thickness. The initial gas-phase species mass
fractions Y k are taken to be uniform across the channel, with the exception of the
mass fractions at reactive walls. These are calculated by the procedure described in
the Section 11.3 on species concentrations at boundaries. From the initial profiles we
compute the local mass flux m and the physical locations of all the mesh points, i.e., a
profile of y .
11.2
Boundary Conditions
For the energy equation, either the temperature, a zero-heat-flux (adiabatic), or a
specified heat flux condition is specified at the solid walls. In the transformed
equations, is the independent variable and the physical coordinate y is a
dependent variable. For the evaluation of y , then, we specify as boundary conditions
that y = 0 at the lower boundary and y = y max at the upper boundary (the channel
radius in the case of cylindrical coordinates). Notice that there is no explicit equation
or boundary condition for the pressure P even though it is a dependent variable. Note
also that a boundary value of y is specified at both boundaries even though
Equation 11-16 is only a first-order equation. This apparent over-specification is
resolved by the fact that there is no boundary condition for pressure.55, p. 163
The boundary conditions for the species involve the heterogeneous reactions. The
convective and diffusive mass fluxes of gas-phase species at the surface are
balanced by the production (or depletion) rates of gas-phase species by surface
reactions. This relationship is
Equation 11-19
Y k ( V k, y + v ) = s k W k
( k = 1, , K g )
where the gas-phase diffusion velocities are given by Equation 11-5 or Equation 11-6.
RD01400-C01-004-001A
169
CHEMKIN 4.0
In nonreacting flows the fluid velocity normal and tangential to a solid wall is zero.
However, if there are chemical reactions at the wall, then the normal velocity can be
nonzero. This so-called Stefan flow velocity occurs when there is a net mass flux
between the surface and the gas. The Stefan velocity is given by
Equation 11-20
Kg
1
v = ---
s k Wk
k=1
This expression is easily obtained from the interfacial mass balance, Equation 11-19,
by summing over all K g gas-phase species and using the requirement that the mass
fractions must sum to one, i.e.,
Equation 11-21
Kg
Yk
= 1
k=1
s k W k
G k = -----------
11.3
( k = K b, , K b )
170
RD01400-C01-004-001A
Theory Manual
numerical software TWOPNT to solve for the self-consistent set of gas-phase and
surface concentrations at each wall. The user can optionally give the TWOPNT
procedure initial guesses for the gas-phase and surface concentrations to aid in the
convergence. The initial TWOPNT problem can also be bypassed altogether if the selfconsistent values for the z k and Y k are supplied by the user.
11.4
Equation 11-23
Kg
j k Wj Dk, j ( Xj )
D km = -------------------------------------------------------W
Kg
jk
( X j )
Xj
Xk
--------- = -------
jk
Here, we implement multicomponent transport using the diffusion velocity of the form
in Equation 11-6, with D km calculated using Equation 11-23.
RD01400-C01-004-001A
171
CHEMKIN 4.0
Mass conservation requires that the diffusive mass fluxes sum to zero
Equation 11-25
Kg
Yk Vk
= 0
k=1
Vc =
Yk Vk
k=1
When this correction velocity (independent of species, k ) is added to all the species
diffusion velocities as computed from Equation 11-6, diffusional mass conservation is
assured. This Correction Velocity option can be specified in the User Interface.
Another option to account for the deficiencies of the mixture-averaged closure of the
transport problem and to assure mass conservation is to solve only K g 1 gas-phase
species conservation equations and to determine the remaining mass fraction by
K
requiring g Y k = 1 . (The mixture-averaged transport closure is asymptotically
k=1
correct in the trace-species limit.) In cases where one species is present in large
excess (such as a carrier gas), this is a reasonable option. The carrier-gas
composition is conveniently determined as
Equation 11-27
Kg
Y km = 1
Yk
k = 1
k km
172
RD01400-C01-004-001A
Theory Manual
Where k m is the species for which we have chosen to apply the closure constraint.
The program will determine k m as the index of the species with the highest
concentration. If the user does not specify use of a correction velocity, it is assumed
that the species k m is the carrier gas and thus a corresponding conservation equation
(Equation 11-2) for that species is not solved. Instead, for that species,
Equation 11-27 applies.
11.5
Thermal Diffusion
Thermal diffusion is the separation of two species of differing size in the presence of a
temperature gradient. Because there can be strong temperature gradients in a
reactor, thermal diffusion can significantly influence deposition rates56, p. 163,61,62 as
well as density profiles, as observed by in situ measurements.63,64,65,66 The effect of
thermal diffusion is included in the diffusion velocity as the second term on the right
side of Equation 11-5 or Equation 11-6. Note that a full multicomponent model is used
for the computation of thermal diffusion coefficients, regardless of whether the user
has specified the mixture-averaged or the multicomponent option for the calculation of
the diffusion velocity. See Chapter 5 for more details about this formulation.
11.6
61.J. P. Jenkinson and R. Pollard, Journal of the Electrochemical Society 131:2911 (1984).
62.J. Juza and J. Cermak, Journal of the Electrochemical Society 129:1627 (1982).
63.W. G. Breiland and M. J. Kushner, Applied Physics Letters 42:395 (1983).
64.P. Ho and W. G. Breiland, Applied Physics Letters 43:125 (1983).
65.P. Ho and W. G. Breiland, Applied Physics Letters 44:51 (1984).
66.W. G. Breiland and P. Ho, in The Electrochemical Society Softbound Proceedings Series, edited
by M. Robinson, C. H. J. v. d. Brekel, G. W. Cullen, J. M. Blocher and P. Rai-Choudhury The Electrochemical Society, New York, 1984.
RD01400-C01-004-001A
173
CHEMKIN 4.0
Equation 11-28
fj + 1 fj
fj fj 1
- ---- f- --------------------------2
------------------------ j + 1 2 --------------------
2
j + 1 j
j j 1
j + 1 j 1
Here the subscript j denotes the jth grid point. We approximate the first derivatives,
as needed in Equation 11-3, by central differences as
Equation 11-29
-----T- T
j + 1 Tj 1
---------------------------- j + 1 j 1
We evaluate terms with no derivatives, such as the chemical production rate in
Equation 11-2, using the conditions existing at j . Likewise, the coefficients of
derivatives, such as u in Equation 11-1, are also evaluated at j .
First-order ODEs, such as Equation 11-16, are differenced according to the
trapezoidal rule as
Equation 11-30
+1
+1
yj 1
1- y-----------------------------2
j
--- = ( + 1 ) -------------------------------------
m j j 1
j uj + j 1 uj 1
It is important to represent the integral equations as first-order differential equations
+1
and include the variables such as y
in the dependent variable vector. The reason
for this choice is associated with the structure of the Jacobian matrix, which is needed
to solve the problem. When Equation 11-30 is used, the number of dependent
variables increases, but the Jacobian remains banded (a very desirable feature). On
+1
the other hand, if y
were to be considered as a coefficient in the transport
equations, as defined by the integral of the stream function, then the Jacobian loses
its banded property and the required computer storage would increase enormously.
11.7
Non-Uniform Grid
In many reacting flow problems a thin reactive boundary layer forms near a surface.
Many grid mesh points may be needed to resolve the important chemical species
concentration profiles. However, further away from the surface there may be no need
for such a finely resolved mesh. A non-uniform grid may be specified by the user,
through the Stretch parameter, s . This parameter can be used to includea mesh that
is finer near a surface and more widely spaced as distance from the surface
increases.
174
RD01400-C01-004-001A
Theory Manual
For cartesian coordinates, if the user has specified N mesh points, then the jth grid
point will be placed at a distance y ( j ) from the lower wall
Equation 11-31
H
s
y ( j ) = -------------------s ( j 1 )
(N 1)
where H is the channel height. If s = 1 , a uniform grid is produced. For s > 1 , the grid
is more tightly spaced at the lower boundary. In cylindrical coordinates, the grid will be
made finer at the outer boundary through use of an analogous formula.
11.8
Jacobian Matrix
The finite-difference representation of the defining equations forms a set of
differential/algebraic equations (DAEs), which we solve numerically using the
software package, DASSL.59, p. 168, 60, p. 168 The DAEs are written in the general form
g ( y, y, t ) = 0 , where y and y represent the components of the solution and their
time derivatives, and t is time. As part of the solution to the equations, a numerical
Jacobian is formed, i.e., the partial derivative of the residual g with respect to each
component of the solution. A Jacobian is needed because DASSL employs an implicit
time stepping method. Implicit methods are most efficient for solving the stiff
equations usually found in chemical kinetics problems.
Calculating the Jacobian is expensive and the most computationally expensive aspect
of evaluating it is calculation of the multicomponent transport properties. The
computer time required to solve the set of DAEs can be reduced significantly by
holding the transport properties fixed when calculating the numerical Jacobian.
Although this introduces some error into the Jacobian, DASSL requires only an
approximate Jacobian for the purpose of iterating to a converged solution.59, p. 168,
60, p. 168 No error is introduced into the solution by making this simplification in the
Jacobian evaluation. However, the computer time required by the model can be
decreased by a factor of ten or more.
11.9
RD01400-C01-004-001A
175
CHEMKIN 4.0
We have also added options to DASSL for printing (sometimes copious amounts of)
information when integration difficulties are encountered. The user can turn on these
error diagnostics in the Advanced Solver-panel tab of the User Interface. The default
level is 0, which results in no reporting. Levels 1 and 2 (the maximum) provide
information on the step size, order of integration, time-step truncation error, and
solution components that most significantly contributed to a convergence test failure.
Level 1 prints this information only upon a convergence test failure. Level 2 prints this
information at every time-step. This diagnostic information can be helpful, but requires
thorough knowledge of DASSLs solution algorithm to be useful. The reader is referred
to the DASSL citations for further documentation.59, p. 168, 60, p. 168
176
RD01400-C01-004-001A
CHEMKIN 4.0
12
12
67.J. A. Miller, R. E. Mitchell, M. D. Smooke, and R. J. Kee, in Proceedings of the Nineteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1982,
p. 181.
68.J. A. Miller, M. D. Smooke, R. M. Green, and R. J. Kee, Combustion Science and Technology
34:149 (1983).
69.J. A. Miller, M. C. Branch, W. J. McLean, D. W. Chandler, M. D. Smooke, and R. J. Kee, in Proceedings of the Twentieth Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, Pennsylvania, 1985, p. 673.
RD01400-C01-004-001A
177
CHEMKIN 4.0
The Premixed Flame Models solve the set of governing differential equations that
describe the flame dynamics using implicit finite difference methods, as well as, a
combination of time-dependent and steady-state methods. The solver algorithm
employed automates coarse-to-fine grid refinement as a means to enhance the
convergence properties of the steady-state approach and as a means to provide
optimal mesh placement.
The Burner-stabilized Flame Model is the one most often used for analyzing species
profiles in flame experiments, where the mass flow rate through the burner is known.
The user has two options for the burner-stabilized flameone where the temperature
profile is known and one in which the temperature profile is determined by the energy
conservation equation. Often the temperatures are obtained from experiment. In this
case, only the species transport equations are solved. In many flames there can be
significant heat losses to the external environment, which are of unknown or
questionable origin and thus are difficult to model. However, since the chemistry
depends strongly on temperature, it is essential to know the temperatures accurately
in order to draw conclusions about the chemical kinetics behavior. If a temperature
profile can be measured accurately, it is often better to use this measurement than the
temperature profile obtained by solving an energy conservation equation. For cases
where the heat losses are known to be negligible, the application can solve a burnerstabilized flame problem in which the temperatures are determined from the energy
conservation equation. Comparing these two problem types for the burner-stabilized
model may also provide some indication of the heat losses.
The Flame-speed Calculation Model involves a freely propagating flame. This
configuration is used to determine the characteristic flame speed of the gas mixture at
specified pressure and inlet temperature. In this case there are no heat losses (by
definition) and thus the temperatures should be computed from the energy equation.
Flame speed depends, in part, on the transport of heat, and predicting the
temperature distribution is an integral part of the flame speed calculation.
12.1
Equation 12-1
Continuity
M = uA
178
RD01400-C01-004-001A
Equation 12-2
Energy
dT 1 d
dT
A
M ------ ----- ------ A ------ + ----dx c p dx dx c p
Equation 12-3
Theory Manual
k hk Wk
k=1
Species
dY k d
M -------- + ------ ( AY k V k ) A k W k = 0
dx dx
Equation 12-4
= 0
( k = 1, , K g )
Equation of State
= PW
--------RT
In these equations x denotes the spatial coordinate; M the mass flow rate (which is
independent of x ); T the temperature; Y k the mass fraction of the kth species (there
are K g species); P the pressure; u the velocity of the fluid mixture; the mass
density; W k the molecular weight of the kth species; W the mean molecular weight of
the mixture; R the universal gas constant; the thermal conductivity of the mixture;
c p the constant-pressure heat capacity of the mixture; c pk the constant pressure heat
capacity of the kth species; k the molar rate of production by chemical reaction of
the kth species per unit volume; h k the specific enthalpy of the kth species; V k the
diffusion velocity of the kth species; and A the cross-sectional area of the stream
tube encompassing the flame (normally increasing due to thermal expansion)
normalized by the burner area. The user may provide an area profile (APRO) or
alternativelya subroutine to specify the area as a function of the spatial coordinate. By
default, the stream tube area is taken to be constant and equal to unity.
The net chemical production rate k of each species results from a competition
between all the chemical reactions involving that species. We presume that each
reaction proceeds according to the law of mass action and the forward rate
coefficients are in the modified Arrhenius form,
Equation 12-5
EA
k f = AT exp ---------
RT
The details of the chemical reaction equations and the thermochemical properties are
found in Chapter 3 and Chapter 2, which discuss the evaluation of these expressions.
RD01400-C01-004-001A
179
CHEMKIN 4.0
In addition to chemical reaction rates, we must also be concerned with the transport
properties of the species, i.e., thermal conductivities and diffusion coefficients.
Stockmayer potentials are used throughout in evaluating transport properties, as
described in Chapter 5. The user has the option of evaluating transport properties
using mixture-averaged formulas or a multicomponent diffusion model. Although
details of the calculation of transport properties are available in Chapter 5, a brief
description is also provided here.
12.2
Equation 12-6
Vk = Vk + Wk + Vc
V k is the ordinary diffusion velocity and is given in the Curtiss-Hirschfelder22, p. 76
approximation by
Equation 12-7
1 dX k
V k = D km ----- --------X k dx
where X k is the mole fraction, and where the mixture-averaged diffusion coefficient
D km is given explicitly in terms of the binary diffusion coefficients D kj
Equation 12-8
1 Yk
D km = -----------------------K
Xj Dkj
jk
A non-zero thermal diffusion velocity is included only for the low molecular weight
species H , H 2 , and He . The trace, light-component limit is employed in determining
W k , i.e.,
Equation 12-9
D km k 1 dT
W k = ----------------- --- -----X k T dx
180
RD01400-C01-004-001A
Theory Manual
where k is the thermal diffusion ratio.33, p. 88 The sign of k makes the lower
molecular weight species diffuse from low to high temperature regions.
The correction velocity V c (independent of species but a function of the distance x ) is
included to insure that the mass fractions sum to unity or equivalently
Equation 12-10
K
Yk Vk
= 0
k=1
The formulation of the correction velocity is the one recommended by Coffee and
Heimerl21, p. 76, 70 in their extensive investigation of approximate transport models in
hydrogen and methane flames and discussed further in Section 5.5 (Equation 5-84).
12.3
Equation 12-11
Vk = Vk + Wk
Now, the ordinary diffusion velocity term is given in Equation 12-12.
70.T. P. Coffee and J. M. Heimerl, Combustion Science and Technology 34:31 (1983).
RD01400-C01-004-001A
181
CHEMKIN 4.0
Equation 12-12
K
1
V k = ----------- W j D k, j d j
Xk W
jk
Here W is the mean molar mass, W j is the molar mass of species j , and d j is
defined as:
1
d j = X k + ( X k Y k ) --- P
P
The thermal diffusion velocity is given as:
Equation 12-13
T
Dk 1
W k = --------- --- T
Yk T
T
12.4
Boundary Conditions
We consider two different types of flames: burner-stabilized flames and adiabatic,
freely propagating flames. The conservation equations governing the two are the
same, but the boundary conditions differ. In both cases the appropriate boundary
conditions may be deduced from the early work of Curtiss and Hirschfelder.22, p. 76
For burner-stabilized flames M is a known constant, the temperature and mass flux
fractions ( k = Y k + Y k V k A M ) are specified at the cold boundary, and vanishing
gradients are imposed at the hot boundary.
For freely propagating flames M is an eigenvalue and must be determined as part of
the solution.71 Therefore, an additional constraint is required, or alternatively one
degree of freedom must be removed from the problem. We choose to fix the location
of the flame by specifying and fixing the temperature at one point. This is sufficient to
allow for the solution of the flame speed eigenvalue M . The user must select this
point in such a way as to insure that the temperature and species gradients nearly
vanish at the cold boundary. If this condition is not met then the resultant M will be too
low because some heat will be lost through the cold boundary. More details on
boundary conditions are described in Section 12.6.
71.M. D. Smooke, J. A. Miller, and R. J. Kee, Combustion Science and Technology 34:79 (1983).
182
RD01400-C01-004-001A
12.5
Theory Manual
Equation 12-14
T j T j 1
M dT
------ M j -------------------- dx j
xj xj 1
where the index j refers to the mesh point. The central difference formula is
Equation 12-15
hj 1
hj hj 1
hj
M dT
------ M j ------------------------------T j + 1 + --------------------- T j -------------------------------------- T j 1
dx j
hj ( hj + hj 1 )
hj hj 1
hj 1 ( hj + hj 1 )
where h j = x j + 1 x j . The windward difference formula introduce artificial diffusion
on a coarse mesh; this has the effect of spreading out the solution and making the
convergence of Newtons method less sensitive to the starting estimate. However,
because the mesh is refined in regions of high gradient, the artificial diffusion
becomes relatively unimportant after the solution has progressed to the fine meshes.
Nevertheless, for a given mesh, the windward difference approximation is less
accurate than the central difference formula. Therefore, the user may want to select
the central difference formula on finer meshes or in cases where the solution is
converging without difficulty.
The first derivative in the summation term in the energy Equation 12-2 is always
approximated by a central difference formula,
Equation 12-16
hj 1
hj hj 1
hj
dT
------ ------------------------------ T j ------------------------------------- T j + 1 + --------------------T
dx j h j ( h j + h j 1 )
hj hj 1
h j 1 ( h j + h j 1 ) j 1
and the coefficients in the summation are evaluated at j .
RD01400-C01-004-001A
183
CHEMKIN 4.0
The second derivative term in the energy equation is approximated by the following
second order central difference:
Equation 12-17
T j T j 1
T j + 1 T j
d-
dT
2
- ( A ) 1 --------------------- + ( A ) 1 ------------------------------ --------------------------(
A
)
dx
dx j x j + 1 x j 1
j + --- x j + 1 x j
j + --- x j x j 1
2
The coefficients in this formula (at j 1 2 ) are evaluated using the averages of the
dependent variables between mesh points.
The diffusive term in the species conservation equation is approximated in a similar
way, but it appears to be different because we have written it using diffusion
velocities. The ordinary (Equation 12-7) and thermal (Equation 12-9) diffusion
velocities are approximated at the j 1 2 positions as illustrated by the following
mixture-averaged evaluation:
Equation 12-18
X k, j + 1 X k, j
W k D km
--------------- -------------------------------
( Yk Vk )j + 1 2
W j + 1 2 xj + 1 xj
and
Equation 12-19
W k D km k
T j + 1 T j
( Y k W k ) j + 1 2 ------------------------
----------------------
WT j + 1 2 x j + 1 x j
Since the mole fraction of a species can be zero, we avoid difficulties by forming
Y k V k , which is the expression needed in Equation 12-3, rather than V k itself
( Y k = X k W k W ). After the diffusion velocities are computed at all the mesh midpoints,
the correction velocity V c is computed at the midpoints from
Equation 12-20
K
Vc =
( Vk + Wk )
k=1
184
RD01400-C01-004-001A
Theory Manual
( AY k V k ) j + 1 2 ( AY k V k ) j 1 2
d
------ ( AY k V k ) j --------------------------------------------------------------------------------dx
xj + 1 2 xj 1 2
All the non-differentiated terms, such as the chemical production rate terms, are
evaluated at the mesh points j . Coefficients not appearing within derivatives are also
evaluated at the mesh points.
12.6
Boundary Conditions
The boundary conditions are relatively easily implemented. At the cold boundary we
specify the mass flux fractions and the temperature, i.e. we solve
Equation 12-22
AY k V k
k, 1 Y k, 1 ----------------- -
M
j = 11 2
= 0
and
Equation 12-23
T1 Tb = 0
where k, 1 is the inlet reactant fraction of the kth species and T b is the specified
burner temperature. At the hot boundary we specify that all gradients vanish, i.e.,
Equation 12-24
Y k, J Y k, J 1
-------------------------------- = 0
xJ xJ 1
and
Equation 12-25
TJ YJ 1
----------------------- = 0
xJ xJ 1
RD01400-C01-004-001A
185
CHEMKIN 4.0
The boundary conditions for M depend on whether the given problem is a burnerstabilized or a freely propagating flame. The boundary conditions for each case are
discussed in Section 12.4.
12.7
Equation 12-26
T
T 1
T
A
A ------ + M ------ ----- ----- A ------ + ----t
x cp x
x
cp
k=1
T A
Y k V k c pk ------ + ---- x cp
k hk Wk
= 0
k=1
and
Equation 12-27
Y
Y
( k = 1, , K )
The full system now becomes a system of parabolic partial differential equations,
rather than an ordinary differential equation boundary value system. Solution is
obtained via the backward Euler method. In this method, the time derivatives are
approximated by finite differences as illustrated by
Equation 12-28
n+1
Tj
T
n + 1 Tj
-
A ------ j A j ----------------------t
h
where the superscript n indicates the time level and h represents the size of the time
step. All other terms are approximated with finite differences as before, but at time
level n + 1 . Since all variables are known at time level n , the discretized problem is
just a system of nonlinear algebraic equations for the dependent variable vector at
time level n + 1 .
186
RD01400-C01-004-001A
CHEMKIN 4.0
13
13
13.1
RD01400-C01-004-001A
187
CHEMKIN 4.0
oxidizer, a diffusion flame is established. Since most fuels require more air than fuel
by mass, the diffusion flame usually sits on the oxidizer side of the stagnation plane;
fuel diffuses through the stagnation plane to establish the flame in a stoichiometric
mixture.
Our Opposed-flow Flame Model is derived from a model that was originally developed
by Kee, et al.72 for premixed opposed-flow flames. The reduction of the threedimensional stagnation flow is based upon similarity solutions advanced for
incompressible flows by von Karman,73 which are more easily available in
Schlichting.74 The use of the similarity transformation is described in more detail in
Chapter 14. Note that the CHEMKIN impinging and stagnation-flow models are based
on a finite domain, where the user specifies the nozzle separation. For this approach,
an eigenvalue must be included in the solution of the equations and the strain rate
varies, such that a characteristic strain rate must be determined from the velocity
profile. Following the analysis of Evans and Grief,75 Kee, et al.72 showed that this
formulation allowed more accurate predictions of the extinction limits for premixed
flames than other approaches.
The geometry and axes for the axisymmetric and planar configurations are sketched
in Figure 13-1 and Figure 13-2, respectively. In the following equations, represents
either the radial direction r for the axisymmetric case, or the perpendicular direction y
for the planar case. The coordinate parameter n allows us to present one set of
equations for both cases, with n = 3 for the 3-D axisymmetric flow and n = 2 for the
2-D planar case. A more detailed derivation of the governing equations for the
opposed-flow geometry is provided by Kee, et al.72, p. 188
At steady-state, conservation of mass in cylindrical or planar coordinates is
Equation 13-1
n2
)
1 ( v r
-------------( u ) ----------+ n 2- ----------------------------- = 0
where u and v r are the axial and radial velocity components and is the mass
density.
72.R. J. Kee, J. A. Miller, and G. H. Evans, in Proceedings of the Twenty-Second Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, Pennsylvania, 1988, p. 1479.
73.T. V. Karman, A. Angew. Math. Mech. 1:233 (1921).
74.H. Schlichting, Boundary Layer Theory, McGraw-Hill, New York, 1979.
75.G. H. Evans and R. Greif, Numerical Heat Transfer 14:373 (1988).
188
RD01400-C01-004-001A
Figure 13-1
Theory Manual
Figure 13-2
RD01400-C01-004-001A
189
CHEMKIN 4.0
Following von Karman,73, p. 188 who recognized that v r and other variables should
be functions of x only, we define
G(x) =
-----v-
u
F ( x ) = ---------------(n 1)
(x)
G ( x ) = dF
-------------dx
for the axial velocity u . Since F and G are functions of x only, so are , u , T and
Yk .
The perpendicular momentum equation is satisfied by the eigenvalue
Equation 13-3
1 - -----p= constant
H = ----------n2
= 0
dT 1 d dT
dT 1
u ------ ----- ------ ------ + ----- c pk Y k V k ------ + ----- h k k = 0
dx
c p dx
dY
dx
dx
dx
cp
dx
cp
Equation 13-6
d- ( Y V ) W = 0
u --------k + ----k k
k k
k = 1, , K
where the diffusion velocities are given by either the multicomponent formulation
Equation 13-7
K
dX j D k 1 dT
1
V k = ----------- W j D k, j -------- --------- --- -----dx Y k T dx
Xk W
jk
190
RD01400-C01-004-001A
Theory Manual
dX k D k 1 dT
1 D -------- --- -----V k = ----km dx- -------Y k T dx
Xk
where
1 Yk
D km = --------------K
Xj
------D jk
jk
x = 0:
F uF
F = ---------------- ; G = 0; T = TF ; uY k + Y k V k = ( uY k ) F
(n 1)
x = L:
O uO
F = ---------------- ; G = 0; T = T O ; uY k + Y k V k = ( uY k ) O
(n 1)
Note that the inflow boundary condition (Equation 13-9) specifies the total mass flux,
including diffusion and convection, rather than the fixing species mass fraction
Y k = Y k, F . If gradients exist at the boundary, these conditions allow diffusion into the
nozzle.
The differential Equation 13-2 through Equation 13-6 and boundary conditions
Equation 13-9 form a boundary value problem for the dependent variables ( F , G , H ,
T , Y k ). The GAS-PHASE KINETICS Subroutine Library provides the reaction rates and
thermodynamic properties, while the TRANSPORT package evaluates the transport
properties for these equations.
RD01400-C01-004-001A
191
CHEMKIN 4.0
13.2
Finite-Difference Approximations
Discretization of the differential equations uses conventional finite differencing
techniques for non-uniform mesh spacing. Diffusive terms use central differences,
with truncation error that is second-order in the mesh spacing. For better
convergence, convective terms use upwind differencing, which uses the sign of the
velocity to choose which direction the spatial difference will go. If u j = 0 , for example,
then the convective term in the energy equation is differenced as:
Equation 13-10
dT
dx
T T
xj xj 1
j
j 1
u ------ i u j --------------------
The truncation error of this approximation is first-order in the mesh spacing, leading to
what is often called artificial diffusion, but this form avoids unwanted oscillations
during the solution on a coarse mesh. Alternatively, the convective terms can be
centrally differenced, but the default windward differencing is recommended.
13.3
Regrid Operation
A Regrid operation is specified by supplying a new number of grid points during a
restart or continuation, which allow a new flame solution to begin from an initial guess
based on the solution of a previous flame.
The steady-state solver, TWOPNT, automatically refines the grid by adding points in
regions where they are needed to resolve the first and second derivatives of the
solution, using criteria controlled by the Gradient and Curvature grid parameters.
However, TWOPNT does not move or remove points. If it reaches a maximum number
of points (internally defined by the dimensions), a warning message is printed and the
adaptation is terminated. In some cases, then, it may be necessary to reduce the
number of points when starting a new solution from a previous result. The Regrid
operation redefines the solution guess on the user-specified number of mesh points.
The Regrid operation is different from the grid-point insertion operation performed by
TWOPNT. Both operations attempt to resolve the gradient and curvature in the solution,
except that TWOPNT considers all solution components, whereas Regrid only
considers the temperature profile. TWOPNT only adds points, leaving the old points as
they were, but Regrid alters the location and solution of all the points interior to the
boundaries. Regrid computes new locations for exactly the given number of points,
192
RD01400-C01-004-001A
Theory Manual
and then interpolates the solution from the previous grid to obtain a new
approximation of the solution. Regrid does not conserve any properties of the
solution; in fact, it tends to smooth the solution by the error inherent in the
interpolation.
Regrid redistributes a weighting function of the first and second derivatives of the
temperature. The profiles of the other dependent variables are ignored on the
assumption that the temperature profile defines the flame location well enough for the
purposes of realigning the mesh for an initial condition. The redistribution uses a
transformation from the physical coordinate x to a new coordinate
Equation 13-11
dx
-----W ( x, T ) = C
d
with the weighing function,
Equation 13-12
2
TW ( x, T ) = 1 + b 1 dT
------ + b 2 d-------2
dx
dx
Integration over the entire domain defines the constant
Equation 13-13
L
1
C = ------------- W ( x, T ) dx
N1
O
Integrating over a portion of the domain gives an expression for the point locations in
-space
Equation 13-14
x
1- W ( x, T ) dx
= 1 + --C
0
The new grid locations x come by interpolation between the computed values of
defined using the old mesh, onto a uniform mesh in -space. Since d is constant on
this uniform mesh, the solution to Equation 13-11 states that W ( x, T ) dx is constant, so
the new values of x will be concentrated where the weighting function is large.
RD01400-C01-004-001A
193
CHEMKIN 4.0
Below is a sample set of parameters that would direct the Opposed-flow Flame Model
to perform the Regrid operation during a restart:
Number of Grid Points for Regrid
20
0.6
1.0
These parameters are specified in the Cluster Properties Panel, when Restart
problem-type is chosen. This sequence will create a new solution guess on 20 points,
devoting 60 percent of the points to resolving gradients, with equal weighting of
gradient and curvature in the temperature profile. From experience, we recommend a
value greater than or equal to 1 for the gradient to curvature adaptation ratio.
Depending on the resolution of the existing solution, the percent of grids used in the
regrid process should be in the neighborhood of 1/2. Note that using 0% grids for
regrid results in a uniform mesh.
194
RD01400-C01-004-001A
CHEMKIN 4.0
14
14
RD01400-C01-004-001A
195
CHEMKIN 4.0
Fortunately, however, the transformed equations still provide a very good practical
approximation to the flow in a finite-radius reactor over a large fraction of the disk (up
to ~90% of the disk radius) when the reactor operating parameters are properly
chosen, i.e., high rotation rates.76, p. 195
In the limit of zero rotation rate, the rotating disk flow reduces to a stagnation-point
flow, for which a similar separation-of-variables transformation is also available. Such
flow configurations (pedestal reactors) also find use in CVD reactors.
This chapter reviews the rotating-disk/stagnation-point flow equations. An infiniteradius disk rotating below a fluid medium is a classic problem in fluid
mechanics.73, p. 188,74, p. 188,80 Under these ideal conditions this problem has a
solution that is an exact solution of the Navier-Stokes equations. Consequently, the
heat and mass transfer near an infinite-radius rotating disk have been extensively
studied.81,82 For CVD applications, Olander83 used a rotating disk to study deposition
in the germanium-iodide system. Pollard and Newman79, p. 195 performed a
theoretical study of the deposition of Si from SiCI4 on a rotating-disk susceptor. They
extended the von Karman similarity solution for isothermal flow73, p. 188 by adding
energy and species equations and incorporating temperature-dependent fluid
properties to obtain an ordinary differential equation boundary-value problem for the
heat, mass and momentum transfer. Hitchman, et al.84 studied epitaxial Si deposition
from SiCI4 in a rotating-disk reactor and analyzed their results in terms of the infinitedisk solution.
Consider a solid rotating surface of infinite extent in the r, plane (Figure 14-1)
separated from a facing, parallel, porous, non-rotating surface by a distance x = L . A
forced flow with purely axial velocity emerges from the porous surface and is directed
toward the rotating one. The flow at x = L approximates the inlet flow conditions in a
cylindrical, rotating-disk CVD reactor. The finite x domain in the present case results
in a nonzero value of the radial pressure gradient.
Analysis by Evans and Greif76, p. 195 combined the stagnation and rotating disk flows
for the situation of an infinite-radius, porous, non-rotating disk separated by a distance
L from an infinite-radius, nonporous, heated, rotating disk. This combination of flows
provides a good approximation to the flow field in a rotating disk reactor. The Evans
and Greif analysis was used as a foundation for the CHEMKIN stagnation-flow and
80.F. M. White, Viscous Fluid Flows, McGraw-Hill, New York, 1974.
81.W. G. Cochran, Proc. Cambridge Phil. Soc. 30:365 (1934).
82.E. M. Sparrow and J. L. Gregg, Journal of Heat Transfer ASME 82:294 (1960).
83.D. R. Olander, I. and E. C. Fund. 6:178 (1967)
84.M. L. Hitchman, B. J. Curtis, H. R. Brunner, and V. Eichenberger, in Physicochemical Hydrodynamics, edited by D. B. Spalding Advanced Publications, London, 1977, Vol. 2
196
RD01400-C01-004-001A
Theory Manual
14.1
Equation 14-1
Mixture continuity
u- 2V --u-
--1-
------ = ---------- = 0
t
x
x
Equation 14-2
Radial momentum
V
V
V
1 dp m
2
2
= 0
------ = ----- ------ u ------ ( V W ) --- --------t
x x
x
r dr
Equation 14-3
Circumferential momentum
W
W
W
-------- = ----- -------- u -------- 2 VW = 0
x x
t
x
Equation 14-4
Thermal energy
T
T
T
c p ------ = ----- ------ c p ------
t
t
x x
Kg
- + k h k M k + S q ( x )
cpk Yk Vk ----
x
= 0
k=1
RD01400-C01-004-001A
197
CHEMKIN 4.0
Figure 14-1
Equation 14-5
Species continuity
Y
Yk
Y k V k
( k = 1, K, K g )
Equation of state
RTP =
---------M
Equation 14-7
Surface-species conservation
dZ k
s
-------- = -----k- = 0
dt
n
( k = 1, K, K s )
198
RD01400-C01-004-001A
Theory Manual
D k 1 dT
1 D dX
- --- ------j -------V k = ----km ------dx Y k T dx
Xk
where,
Equation 14-9
1 Yk
D km = ------------------------------K
X
D
j kj
jk
RD01400-C01-004-001A
199
CHEMKIN 4.0
dX j D k 1 dT
1
V k = ----------- M j D k, j -------- --------- --- -----dx Y k T dx
Xk M
j=1
Both formulations have an ordinary diffusion component and may have a thermal
diffusion component (Soret effect). In these expressions, X k is the mole fraction for
the kth species, D kj is the binary diffusion coefficient matrix, D k, j is the matrix of
T
ordinary multicomponent diffusion coefficients, and D k is the thermal diffusion
coefficient for the kth species. Thermal diffusion often plays an important role in CVD
problems. In the presence of strong temperature gradients, thermal diffusion causes
high molecular-weight species in a low molecular-weight carrier to diffuse rapidly
toward the low-temperature region.22, p. 76 The multicomponent and mixture transport
properties are evaluated from the pure species properties using the averaging
procedures as discussed in Chapter 5.
K
Vc =
Yk Vk
k=1
When this correction velocity (independent of species k ) is added to all the species
diffusion velocities as computed from Equation 14-8, mass conservation is assured.
Another option is to account for the deficiencies of the mixture-averaged closure of
the multicomponent transport problem and to assure mass conservation is to solve
only K g 1 gas-phase species conservation equations and determine the remaining
K
mass fraction by requiring g Y k = 1 (Trace option). The mixture-averaged
k=1
transport closure is asymptotically correct in the trace-species limit. In cases where
one species is present in large excess (such as a carrier gas in a CVD reactor), this is
a reasonable option.
200
RD01400-C01-004-001A
Theory Manual
Y kg = 1
Yk
k=1
The default for this option is to consider the last-named gas-phase species in the
GAS-PHASE KINETICS input ( K g ) as the species for which a conservation equation is
not solved. Since the last species may not always be the most abundant species, a
further option provides for dynamically determining the largest species concentration
at each mesh point and removing its conservation equation from the system of
equations (Reorder option in the User Interface).
The source term in the thermal energy equation S q ( x ) is a spatially distributed
thermal energy source that we assume is in the form of a Gaussian:
Equation 14-13
2
3 ( x xs )
1
S q ( x ) = q ----- --3- exp -------------------------2
ws
w
s
To use this expression, the user is expected to specify three parameters: q , w s and
x s . The parameter q is the total energy integrated over its full spatial extent. Implicit in
Equation 14-13 is the fact that
Equation 14-14
q =
Sq ( x ) dx
j k = s k M k
RD01400-C01-004-001A
( k = 1, , K g )
201
CHEMKIN 4.0
The gas-phase mass flux of species k at the surface is a combination of diffusive and
convective processes,
Equation 14-16
jk = Yk u + Yk Vk
( k = 1, , K g )
where u is the bulk normal fluid velocity at the surface and V k is the diffusion velocity
of the kth species. The bulk normal fluid velocity at the surface is computed from the
surface reaction rates summed over all the gas-phase species K g ,
Equation 14-17
Kg
u =
s k M k
------------
k=1
Even though the susceptor surface is solid, there is a bulk fluid velocity into the
surface (the Stefan velocity) that accounts for the mass of solids deposited. This bulk
velocity at the surface is usually small, and thus the boundary movement due to the
deposition is neglected. That is, the problem is solved in a fixed spatial domain. While
the surface growth rate is predicted, the computational domain is not adjusted to
account for small changes resulting from surface growth.
There are two options for treating the thermal-energy boundary condition on the
deposition surface. The first is to simply specify the surface temperature. If the
temperature is controlled or measured directly, this option is usually the one of choice.
However, some problems require that the surface temperature be predicted as part of
the solution. The appropriate boundary condition is derived from a surface energy
balance. Exothermicity (or endothermicity) of surface reactions contributes to the
energy balance at an interface. Diffusive and convective fluxes in the gas-phase are
balanced by thermal radiative and chemical heat release at the surface. This balance
is stated as
Equation 14-18
T
------
x
Kb
Kg
Yk ( Vk + u )hk
k=1
= ( T T w ) +
s k M k h k + P
k=Ks
202
RD01400-C01-004-001A
Theory Manual
T
4
4
------ = ( T T w ) =
x
s k Mk hk + P
k=1
The reaction-rate summation on the right-hand side runs over all species, including
the gas-phase species. The term P presents an energy source in the surface itself,
such as might be generated by resistance heating.
The SURFACE KINETICS Pre-processor requires as input the mass densities k of the
bulk species. These densities are used to convert the rate of production of a bulk
species (in moles/cm2/sec) into a thickness growth rate G (in cm/sec). The needed
relationship is
Equation 14-20
l
Kb
G =
s k M k
-----------
k=Ks
When solving for the flow induced by the rotation of a disk in an infinite, otherwise
quiescent fluid, the axial velocity u at x = L is part of the solution. However, for the
case corresponding to injection of the gas through a non-rotating porous surface, u is
the specified inlet velocity at height x = L . This gives us the flexibility of either
forcing or starving the inlet flow compared to the natural flow induced by the
spinning disk itself. It is always necessary to specify the inlet velocity in the case of a
stagnation-point flow.
The other boundary conditions on the fluid flow fields are relatively simple. The
temperature at x = L (the reactor inlet) is specified. Normally, the radial and
circumferential velocities are zero at x = L . A linearly varying radial velocity or a
specified spin rate may also be specified at x = L . In these cases, V = v r = a i or
RD01400-C01-004-001A
203
CHEMKIN 4.0
u k = uY k + Y k V k
where k is the species mass fraction specified for the inflow. The user may also opt
to fix the species composition (i.e., Y k = k ), by specifying the option in the User
Interface to fix the inlet composition rather than the inlet flux.
14.2
Equation 14-22
V j + 1 V j
V
u ------ j u j ---------------------x
xj + 1 xj
The u velocity is always negative in these problems (flowing from the inlet at large
values of x toward the substrate at x = 0 ), so the upwind differencing calls for the
derivative to be formed between j + 1 and j . Upwind differences cause substantial
artificial diffusion, especially on coarse mesh networks. However, we have found that
they lead to much more reliable convergence on coarse meshes early in the iteration
procedure. Ultimately, because the meshes are adaptively refined, the artificial
diffusion is very small in the final solution.
The first derivative in the summation term in the energy Equation 14-4 is always
approximated by a central difference formula,
Equation 14-23
hj 1
hj hj 1
hj
-----T- ------------------------------T j + 1 + --------------------- T j -------------------------------------- T j 1
x j hj ( hj + hj 1 )
hj hj 1
hj 1 ( hj + hj 1 )
204
RD01400-C01-004-001A
Theory Manual
T j + 1 T j
T j T j 1
---- T- --------------------------2
----- --------------------- --------------------
xj xj 1
x x j x j + 1 x j 1 x j + 1 x j
The coefficient in this formula (at j 1 2 ) are evaluated using the averages of the
dependent variables between mesh points.
The diffusive terms in the species conservation equations are analogous to the
diffusive term in the energy equation. However, because we express them in terms of
diffusion velocities, they at first appear different. In Equation 14-5, the diffusive term is
approximated as
Equation 14-25
Y k V k j + 1 2 Y k, j + 1 2 V k, j + 1 2 j 1 2 Y k, j 1 2 V k, j 1 2
------------------ -----------------------------------------------------------------------------------------------------------------------------x
xj + 1 2 xj 1 2
The ordinary multicomponent diffusion and thermal diffusion fluxes (Equation 14-10)
are approximated at the j 1 2 positions as illustrated by
Equation 14-26
m k
( Y k V k ) 1 = ---------- 1
j + -- M 2 j + --22
Kg
( m l D k, l )
l=1
X l, j + 1 X l, j D k
T j + 1 T j
------------------------------- + ------------- j + 1--- -----------------------
1
x
xj
j + --- x j + 1 x j Y k T
2 j + 1
2
RD01400-C01-004-001A
205
CHEMKIN 4.0
206
RD01400-C01-004-001A
CHEMKIN 4.0
15
15
15.1
RD01400-C01-004-001A
207
CHEMKIN 4.0
likelihood of convergence for the Newton method. After time-stepping, the Newton
algorithm tries again for convergence and, if this fails, TWOPNT takes additional time
steps on the transient solution to further improve the initial iterate. Ultimately, the
Newton iteration converges on the steady-state solution.
15.1.1
Starting Estimates
The use of a Newton iteration algorithm for solution of the governing equations
requires the user to provide initial estimates of the solution variables. It may
sometimes be difficult to obtain good initial estimates of species composition,
especially when one is not familiar with the chemistry system being studied. In such
cases, it may be necessary to approach the problem from points that are more
intuitive. For example, in a thermal problem, a good initial estimate of gas species is
often the thermodynamic equilibrium composition at the initial temperature. For this
purpose, an optional equilibrium calculation (by default if no initial guess is provided),
minimizes the Gibbs free energy of the system in order to determine the equilibrium
gas mole fractions. This method of determining the initial estimate of the gas-phase
composition is automatically initiated when the user does not supply any initial
estimates for the gas-phase species mole fractions. If equilibrium composition does
not provide immediate convergence for the reactor conditions of interest, the user
may increase the residence time (thereby driving the conditions closer to equilibrium)
from the target conditions until a solution is reached. This solution can then be used
as the initial estimate for the desired residence time. Such gradual approach to a
desired solution is facilitated by the Continuation capability, and generally requires
very little computational time. For a plasma system, a similar approach could be used
to obtain a solution where the initial guess is difficult to obtain. In this case, the user
might assume an initial guess for the gas-phase composition close to that of the
reagent gases entering the reactor, and set the power deposition level very low. Then
gradually increasing the reactor power using the continuation capability can lead to
the desired solution.
Initial estimates of surface site fractions are often more difficult, but also have less
impact on the ability of the Newton algorithm to reach a solution. A general rule of
thumb for all species types is to provide an estimate for every species in the system,
even if these numbers are small. An initial guess of 1 X 10-7, for example, is generally
much better than allowing the application to set the initial fraction to zero. All initial
estimates will be normalized, such that the sums of gas mole fractions, surface site
fractions, and bulk species fractions are all equal to one.
208
RD01400-C01-004-001A
Theory Manual
The user has an option to either solve the gas energy equation or to keep the
temperature fixed at the initial specified value. There are some cases when the user
may choose not to solve the gas energy equation. For example, the reactor
temperature may have been accurately measured, while heat losses are difficult to
estimate. In cases where the energy equation is solved, the user has an option to
solve the system of equations in one or two steps. The default is to use two steps: first
solve for the species composition at the fixed temperature estimate provided by the
user, and then solve simultaneously the energy equation and the species composition
using the first solution as the initial guess. This two-stage method provides more
robust convergence for thermal systems, where the reaction rates exponential
dependence on the gas temperature is the primary source of equation stiffness and
nonlinearities. For plasma systems, when one is solving for the electron energy
equation, convergence is usually most expedient with the one-step option with no
fixed-temperature iteration.
15.1.2
Equation 15-1
F() = 0
For 0-D homogeneous reactors, this vector may be, for example:
Equation 15-2
= { j, , N
PSR
where:
Equation 15-3
j = T, T e, Y 1, , Y K , Z f , , Z l , X f , , X l , f , , l , Z f
g
K (1)
K (1) K (1)
K (1) N (1)
N (1) K (M)
RD01400-C01-004-001A
K s(M)
, , Z
K s(M)
,X
b
f
, , X
K b(M)
b
l
K b(M)
209
N (M)
, ,
l
N (M)
CHEMKIN 4.0
where T and T e are the gas and electron temperatures, the Y s are the gas species
mass fractions, the Z s are the surface site fractions within each surface phase, the
X s are the mole fractions of species in the bulk phases, and the s are the surface
site densities of those surface phases whose site densities are allowed to change.
The exact components of the solution vector may vary depending on problem type
and reactor model. The total number of unknowns, J , is defined by Equation 8-66.
The corresponding vector F is composed of a corresponding set of residuals of the
gas energy equation, the electron energy equation, the species equations, and the
continuity equation for the surface site densities.
(0)
(n + 1)
(n)
F 1
(n)
------ ( n ) F ( )
(n + 1)
where 0 <
(n)
(n)
(n)
(J
( n ) 1
) F(
(n)
1 , and
Equation 15-6
(n)
F
= ------ ( m )
1<mn
Internally, TWOPNT will generate a new Jacobian whenever convergence with the old
Jacobian fails. While Equation 15-11 correctly indicates the relation between the new
and the old iterate, TWOPNT does not compute the inverse of the Jacobian matrix, but
(n)
(n)
(n)
rather solves a system of linear equations, J
= F ( ) , for the undamped
(n)
vector, .
210
RD01400-C01-004-001A
Theory Manual
(n)
(J
( n ) 1
) F(
(n + 1)
) < (J
( n ) 1
) F(
(n)
(n + 1)
If the solution
fails this criterion, TWOPNT rejects it and retries the step with half
the damping parameter or a new Jacobian matrix. TWOPNT adjusts the damping
parameter to ensure that the evolving solution always remains within the solution
bounds. Examples of physical bounds that are imposed on the solution are: the
temperature(s) must be positive and the species mass fractions must be between
zero and one. For the steady-state solver, it is actually possible to modify the lower
bounds placed on the species mass fractions to be slightly negative (species floor
value). Allowing some species mole fractions to become slightly negative during
iteration sometimes enhances the convergence rate, especially when the solution
composition has species mass fractions that vary over many orders of magnitude.
The Newton iteration procedure continues along these lines until the user-defined
convergence criteria are met. Convergence requires that the maximum norm of the
(n)
undamped correction vector
has been reduced to within user-defined absolute
and relative tolerances, i.e.,
Equation 15-8
max ( A, R )
where A is the absolute tolerance and R is the relative tolerance. The relative
tolerance roughly indicates the number of significant digits in the converged solution,
and the absolute tolerance serves to exclude solution components smaller than A
from the relative convergence criterion. Typically, the absolute tolerance A should be
smaller than the maximum mass fraction of any species of interest. The user should
be particularly careful in specifying A for plasma solutions, since the small electron
mass will result in a much smaller mass fraction for electrons than for other important
species. The relative tolerance should generally be in the range of 10-3 to 10-4.
RD01400-C01-004-001A
211
CHEMKIN 4.0
If damping does not produce a suitable correction vector, TWOPNT computes a new
Jacobian. In the case when a new Jacobian has just been computed, and a damped
Newton step still cannot produce a suitable correction vector, TWOPNT begins to take
time steps. The time-stepping procedure is described in the section following the
description of the Jacobian Matrix below.
15.1.3
Jacobian Matrix
The Jacobian Matrix is a dense J J matrix. Rather than derive and evaluate analytic
expressions for the Jacobian elements, we form the elements of the Jacobian
numerically, through finite-difference perturbations. This approach is justified since
the accuracy of analytic Jacobians is not required for the modified Newton method
described above. This is demonstrated by the successful use of old (and therefore
inaccurate) Jacobians. We evaluate the numerical Jacobian elements from a onesided finite-difference formula as follows:
Equation 15-9
Fi ( j + ) Fi ( j )
J i, j ---------------------------------------------
where
Equation 15-10
= r j + a
We choose the relative and absolute perturbations, r and a , to be the square root of
the computers unit roundoff.
15.1.4
212
RD01400-C01-004-001A
Theory Manual
The transient equations are given in Chapter 8. We solve this system of ordinary
differential equations using the backward-Euler method. In this method, the time
derivatives are approximated by first-order, backwards finite differences. For
example, dT dt is approximated as
Equation 15-11
n+1
Tj
Tj
dT
------ ----------------------dt
t
where here the subscript n indicates the time step index and t represents the size
of the time step. All other terms in the governing equation are evaluated at time level
n . The time-discretized problem is then another system of nonlinear algebraic
equations for the dependent variable vector at time level n + 1 .
To solve the system of equations for each time step we use the same Newton method
as we employ in the steady-state problem. However, the transient problem is much
more likely to converge; it should always converge for a sufficiently small time step.
The objective, then, is to choose a time step that is sufficiently large to make progress
toward the steady solution, yet not so large as to introduce convergence difficulties in
the transient solution method. Typical time steps range from 1-10 microseconds,
although in plasma systems it is often necessary to choose time steps much smaller
than this (10-100 nanoseconds).
After solving the specified number of time steps, TWOPNT again attempts to solve the
steady problem by Newtons method. If the steady solution fails again the application
reverts to time stepping, beginning where it left off at the last time-stepping procedure.
Clearly the better the users initial estimate of the solution, the less likely the
application will have to resort to time stepping. If the application fails to converge in
time-stepping, then, the user may either choose smaller time steps, or try a new
starting estimate.
The time-stepping solution procedure described here is not particularly sophisticated,
since we are neither interested in the accuracy of the transient path nor in solving the
transient problem all the way to steady state. We chose to use a method that is
relatively inexpensive per step and employs the same Newton algorithm that is used
in the steady-state solution. The transient solution therefore serves only to condition
the starting estimates for the Newton algorithm, but does not provide accurate
solution of the transient process.
RD01400-C01-004-001A
213
CHEMKIN 4.0
15.2
15.2.1
Starting Estimates
The 1-D steady-state reactor models require a starting estimate of the solution from
which to begin its iteration. The general form of this estimate is shown in Figure 15-1.
For this estimate, we presume that there is a reaction zone in which reactants are
specified at one end of the reaction zone and the reactants change from their
unreacted values to the products that are found on the far-end of the flow domain.
The user is asked to provide estimates for the location and thickness of this reaction
zone. The user also needs to estimate the product species fractions, or the species
composition that is expected at the far-end of the domain.
214
RD01400-C01-004-001A
Theory Manual
The shape of the assumed species profiles is not too important. Smoother functions,
such as cubic polynomials and hyperbolic tangents for the reactant and product
species, have no apparent effect on the convergence properties of the method. Since
the starting profiles are typically evaluated on a very coarse mesh, it makes little
difference whether the underlying function is smooth or not. Therefore, in simple
linear starting profiles are used.
RD01400-C01-004-001A
215
CHEMKIN 4.0
15.2.2
15.2.3
216
RD01400-C01-004-001A
Theory Manual
approximate solution vectors . When any arbitrary is substituted into the finite
difference analog of Equation 12-2 and Equation 12-3, they do not equal zero as they
would if the true solution were substituted; in general, they equal a residual vector that
we will call F . The objective is to find a vector that satisfies
Equation 15-12
F( ) = 0
In our case the vector is composed as follows:
Equation 15-13
= ( T 1, Y 1, 1, , Y K, 1, M 1, , T j, , Y k, j, , M j, , T J, Y 1, J, , Y K, J, M J )
The corresponding F vector is composed of the residuals of the energy equation, the
species equation, and an equation specifying that dM dx = 0 . The final equation is
included for purposes of maintaining a banded Jacobian structure; this is discussed in
more detail in Section 15.2.4. The ordering of the F vector corresponds to the order
of the vector; it begins with the residuals of the left boundary condition j = 1 ,
followed by the residuals at the interior mesh points, and finally the residuals at the
right boundary, j = J .
(0)
(n + 1)
(n)
F 1
(n)
------ ( n ) F ( )
is too expensive and delicate to be used in practice. On the one hand, evaluation of
the Jacobian matrices F by numerical or other means is time consuming, and on
(0)
the other hand, convergence usually requires a very good initial estimate .
CHEMKIN employs the usual remedies. First, the Jacobian matrix is replaced by one,
(n)
(n)
J , inherited from a previous step of the algorithm. Second, the full step from
to
(n + 1)
(n)
RD01400-C01-004-001A
(n + 1)
(n)
(J
( n ) 1
) F(
(n)
217
CHEMKIN 4.0
where, 0 <
(n)
1 , and
Equation 15-16
J
J
(n)
(n)
= J
(n 1)
F
= ------ ( n )
The inverse Jacobian matrix in Equation 15-15 is not actually computed; instead a
(n)
system of linear equations is solved for the undamped correction vector .
(n)
(n)
(J
( n ) 1
) F(
(n + 1)
) < (J
( n ) 1
) F(
(n)
This prevents the iteration from stepping away from the region where there is good
(n + 1)
reason to believe a solution lies. Should
fail this criterion, the application
rejects it and retries the step with a halved damping parameter or a new Jacobian
(n)
matrix. The damping parameter
is initially chosen to be as large as possible so
(n + 1)
that
does not violate various bounds that are set on the solution variables. We
know, for example, that the temperature and mass flow rate must be positive, and that
the species mass fractions must be between zero and one. The concentrations of
many species, such as fuels downwind of a flame, are close to zero and frequently
threaten to place the solution out of bounds.
The Newton iteration continues until the maximum norm of the undamped correction
vector is reduced to within a user-specified tolerance. Specifically, the criteria for
Newton iteration convergence is when the solution correction vector satisfies
Equation 15-18
max ( A, R )
The relative R and absolute A tolerances are parameters that govern the
convergence criteria for the Newton iteration. Roughly speaking the relative tolerance
indicates how many significant digits the converged solution should contain, and the
absolute tolerance serves to exclude solution components smaller than A from the
218
RD01400-C01-004-001A
Theory Manual
15.2.4
Jacobian Matrix
The Jacobian is a large, banded (actually block tridiagonal) matrix. In principle, we
could derive and evaluate analytic expressions for the Jacobian elements. However,
because this is a difficult and error-prone task, and because the modified Newton
method works well with old (and hence inaccurate) Jacobians, the effort to develop
analytic Jacobian expressions is not warranted. Instead, we form the elements of the
Jacobian by finite difference perturbations in the manner suggested by Curtis, et al.87
Specifically, we evaluate a one-sided finite difference formula,
Equation 15-19
Fi ( j + ) Fi ( j )
J i, j ---------------------------------------------
where
Equation 15-20
= r j + a
We choose the relative and absolute perturbations, r and a , to be the square root of
the computers unit roundoff.
It can be seen from Equation 15-19 that with a single perturbation of j , an entire
column i of J can be computed. Moreover, because the Jacobian is banded, several
of its columns can be evaluated at once. When two or more columns share no nonzero rows, then they can be evaluated numerically by a single perturbation of the
vector. Since evaluating the Jacobian and computing its LU decomposition are by far
the most expensive parts of the flame computation, taking advantage of the
Jacobians structure is essential to producing an efficient application.
87.A. R. Curtis, M. J. Powell, and J. K. Reid, J. Inst. of Math. Appl. 13:117 (1974).
RD01400-C01-004-001A
219
CHEMKIN 4.0
dM
-------- = 0
dx
Equation 15-21 represents the introduction of a new first-order differential equation to
the system and thus requires the introduction of an additional boundary condition.
For the Flame-speed Calculation, the introduction of Equation 15-21 is essential to an
efficient solution procedure. Recall that for the freely propagating flame configuration
we added an interior boundary condition that anchored the temperature at one mesh
point and thus defined a flame-fixed coordinate system. This is the additional
boundary condition that is needed to compute the flame speed M , which is now a
dependent variable. For the sake of notation, let us say that we fix a temperature T f at
mesh point j f . Then for j < j f , the forward-differenced finite difference equation to be
solved is
Equation 15-22
M j M j 1
------------------------- = 0
xj xj 1
At j < j f we solve
Equation 15-23
T jf T f = 0
220
RD01400-C01-004-001A
Theory Manual
M j + 1 M j
-------------------------- = 0
xj + 1 xj
15.2.5
Adaptation
We have found that starting the iteration on a coarse mesh has several important
advantages. One is that the Newton iteration is more likely to converge on a coarse
mesh than on a fine mesh. Moreover, the number of variables is small on a coarse
mesh and thus the cost per iteration is relatively small. Since the iteration begins from
a user-specified guess at the solution, it is likely that many iterations will be required.
Ultimately, of course, to be accurate, the solution must be obtained on a fine mesh.
However, as the solution is computed on each successively finer mesh, the starting
estimates are better, since they come from the converged solution on the previous
coarse mesh. In general, the solution on one mesh lies within the domain of
convergence of Newtons method on the next finer mesh.88 Thus, even though the
cost per iteration is increasing, the number of required iterations is decreasing.
The adaptive placement of the mesh points to form the finer meshes is done in such a
way that the total number of mesh points needed to represent the solution accurately
is minimized. Specifically, we place the mesh points in the following way. To resolve
the gradients we bound the variation in the solution components between mesh points
as
Equation 15-25
n, j n, j 1 ( max n min n )
and to resolve the curvature in the solution we bound the variation in the solutions
derivatives between mesh points by
Equation 15-26
n d n
d n
d n
d------- ------- max -------- min --------
dx j dx j 1
dx
dx
We evaluate the above expressions between each of the mesh points. In each of the
subintervals where the inequality is not satisfied, a new mesh point is placed at the
midpoint of the subinterval. The parameters and are user-specified input
parameters (Gradient and Curvature ratios, respectively). We exclude variables that
88.M. D. Smooke and R. M. M. Mattheij, Appl. Numer. Math. 1:463 (1985)
RD01400-C01-004-001A
221
CHEMKIN 4.0
are below a certain floor value from consideration in the adaptation. This avoids
adapting on variables that are essentially zero, but due to rounding errors may show
locally high derivatives. The starting estimate for the dependent variable vector is
determined by a linear interpolation of the coarse mesh solution onto the new finer
mesh. After determining a converged solution on this new fine mesh, the adaptation
procedure is performed once again. A sequence of solutions on successively finer
meshes is computed until the inequalities in Equation 15-25 and Equation 15-26 are
satisfied between all mesh points.
15.3
222
RD01400-C01-004-001A
Theory Manual
RD01400-C01-004-001A
223
CHEMKIN 4.0
224
RD01400-C01-004-001A
CHEMKIN 4.0
16
16
Sensitivity Analysis
Sensitivity analysis of a problem solution allows quantitative understanding of how the
solution depends on the various parameters contained in a model.91 Sensitivity
analysis is often an invaluable tool in interpreting the results of stirred reactor92 and
flame experiments.93,21, p. 76,94
In CHEMKIN, we consider the first-order sensitivity coefficients of the gas temperature,
species fractions, and, where appropriate, the bulk-phase growth rates, with respect
to the reaction-rate coefficients. The computationally efficient sensitivity analysis
methods exploit the fact that the differential equations describing the sensitivity
coefficients are linear, regardless of any non-linearities in the model problem itself.
Rate-of-production analysis provides complementary information on the direct
contributions of individual reactions to species net production rates.
Sensitivity analysis results are normalized and included in the XML Solution File (e.g.
XMLdata.zip) for post-processing. In addition, results may be printed to the diagnostic
output file. Here we describe the methods used for calculating sensitivity coefficients,
for both steady-state and transient solutions, and for normalizing the resulting
quantities.
91.J. A. Miller, M. C. Branch, W. J. McLean, D. W. Chandler, M. D. Smooke, and R. J. Kee, in Proceedings of the Twentieth Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, Pennsylvania, 1985, p. 673
92.P. Glarborg, R. J. Kee, and J. A. Miller, Combustion and Flame 65:177 (1986).
93.J. A. Miller, M. C. Branch, W. J. McLean, D. W. Chandler, M. D. Smooke, and R. J. Kee, in Proceedings of the Twentieth Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, Pennsylvania, 1985, p. 673.
94.Y. Reuven, M. D. Smooke, and H. Rabitz, Sensitivity Analysis of Boundary Value Problems: Application to Nonlinear Reaction-Diffusion Systems, Yale University Mechanical Engineering Department Report ME-104-85, 1985.
RD01400-C01-004-001A
225
CHEMKIN 4.0
16.1
Equation 16-1
F ( ( ) ; ) = 0
Here, we have introduced the idea that the equation may depend not only on the
solution variables but also on a set of model parameters, . The residual vector F
depends on a both explicitly and implicitly through the solution vector . For reactionrate sensitivity, we consider the s to represent the pre-exponential A-factors in the
Arrhenius reaction-rate expressions. By differentiating Equation 16-1 with respect to a
we obtain a matrix equation for the sensitivity coefficients:
Equation 16-2
-----F -----
F- = 0
+ -----F
The matrix F is the Jacobian of the original system and F is the matrix of
partial derivatives of F with respect to the parameters. The sensitivity coefficients are
defined as F . It is helpful to think of the F matrix column by column, with
each column indicating the dependence of the residual vector F on each parameter.
There are as many columns as there are parameters, i.e. the number of gas-phase
and surface reactions. The sensitivity coefficient matrix F contains quantitative
information on how each reaction-rate coefficient affects the temperature(s) and
species fractions. The sensitivity matrix has a structure similar to that of the F
matrix. That is, each column contains the dependence of the solution vector on a
particular chemical reaction.
226
RD01400-C01-004-001A
Theory Manual
The Jacobian and its LU factorization are already available from the solution of the
original system by Newtons method. Parameter derivatives are computed by finite
differences in a manner similar to the computation of the Jacobian. We can therefore
readily solve the linear system of equations represented by Equation 16-2 for each
column of the sensitivity matrix corresponding to the sensitivities of the solution vector
to each of the reaction rates in the GAS-PHASE KINETICS and SURFACE KINETICS
mechanisms. TWOPNT employs the LINPACK98 software to perform these
computations. We factor the Jacobian only once, and compute each column of the
sensitivity matrix by back substitution, such that the calculation is relatively
inexpensive computationally.
16.2
Equation 16-3
d
------ = F ( , t ;a )
dt
where, in our case, is the vector of temperature(s), mass fractions, surface site
fractions, and bulk activities, as given for example in Equation 15-3 of Section 15.1.2.
Here, the parameter vector a may represent the rate expressions for the gas-phase
and/or surface reactions. For heat-of-formation sensitivity, a represents the vector of
heats of formation for all the species in the system (available for Premixed Laminar
Flame, Opposed-flow Flame, Stagnation-flow and Rotating-disk CVD reactor models).
The first-order sensitivity coefficient matrix is defined as
98.J. J. Dongarra, C. B. Moler, J. R. Bunch, and G. W. Stewart, "LINPACK Users' Guide", Society
of Industrial and Applied Mathematics, 1979.
RD01400-C01-004-001A
227
CHEMKIN 4.0
Equation 16-4
w j, i = ------ i
where the indices j and i refer to the dependent variables and reactions,
respectively. Differentiating Equation 16-4 with respect to the parameters A i yields
Equation 16-5
dw j, i
F
F
------------ = ------ w j, i + --------j
dt
i
Note that this equation for the sensitivity coefficients is linear, even though the model
problem may be nonlinear. Of course, when coupled with a nonlinear model problem,
the whole system is still nonlinear. Nevertheless, when solved via the same
backwards-differentiation formula method as the model problem, the sensitivity
solution is efficient because of the linearity. The Newton iteration for the corrector step
converges in one iteration.
The Jacobian matrix F that appears in Equation 16-5 is exactly the one that is
required by the backwards-differentiation formula method in solving the original model
problem, so it is readily available for the sensitivity computation. Each column
corresponds to the sensitivities with respect to one of the reaction pre-exponential
constants. The solution proceeds column by column. Note that the Jacobian matrix is
the same for each column of the w j, i . However, since the F j i matrix describes
the explicit dependence F on each of the reaction parameters i , each of its columns
must be formed prior to solving for a column of w j, i .
The main weakness with the method as implemented is in the computation of the
F and F j i matrices. The Jacobian, as used for solving the model problem
by the BDF method, is only an iteration matrix, and thus even though its accuracy
may effect the convergence rate, its accuracy has no effect on the accuracy of the
solution. The situation is quite different in the sensitivity equation, where these two
matrices appear explicitly. We use finite difference methods to approximate the
matrices, rather than derive analytic expressions for them. It is possible to derive the
exact expressions, but it is a tedious and time-consuming job. Thus, since the
sensitivity equations themselves are defined in terms of approximate finite difference
representations of the two coefficient matrices, we recognize that the accuracy in the
predicted sensitivity coefficients is somewhat compromised. Nevertheless, since our
principal application of the sensitivity coefficients is to provide insight about the
behavior of the model, we believe that the approximations are justified.
228
RD01400-C01-004-001A
Theory Manual
In addition to determining time-integrated sensitivity of solution variables to reactionrate coefficients, derived sensitivity of growth or etch-rates to the reaction parameters
can also be calculated. Growth-rate sensitivity analysis for the transient calculations
uses the same local-sensitivity calculations described for the steady-state
calculations, as described in Section 16.1. In this case, the sensitivities are not
integrated over time but are instead based only on local conditions at each time-step.
16.3
Equation 16-6
ln Y k
------------- ln i
i Yk
= ----- -------Y k i
or
Equation 16-7
ln T------------ ln i
i T
= ----- -------T i
Second, although the gas-phase species solution variables are mass fractions, the
sensitivity coefficients are computed in terms of mole fractions, as follows:
Equation 16-8
X
-----i --------kX k i
i Yk
= ----- -------Y k i
Kg
1 Yk
i W ------ -------W j i
j=1
where X k are the gas-phase mole fractions, W j are the species molecular weights,
and W is the mean molecular weight of the gas-phase mixture.
Normalized sensitivity coefficients for the surface-site fractions and bulk-phase mole
fractions are computed as:
Equation 16-9
ln Z
-------------k ln i
RD01400-C01-004-001A
i Zk
= ----- -------Z k i
229
CHEMKIN 4.0
and
Equation 16-10
b
ln X k
-------------- ln i
i Xk
= -----b- --------X i
k
F
b
where Z k is the site fraction of surface species k , and X k is the mole fraction of the
bulk species k .
In some cases, it can be argued that sensitivity coefficients are more useful when they
are normalized by the maximum value of each dependent variable (e.g., the
maximum value over the solution time or the maximum value over a spatial profile).
For such cases, CHEMKIN allows the option for users to obtain sensitivity coefficients
normalized as:
Equation 16-11
i Yk
---------- -------max
Y k i
max
where Y k is the maximum mass fraction of the species in the relevant domain. This
normalization avoids artificially high sensitivity coefficients in regions where the mass
fractions are approaching zero, and thus subject to numerical errors. The
implementation of this maximum normalization option is performed during postprocessing as a units option in CHEMKIN Graphical Post-processor.
16.4
230
RD01400-C01-004-001A
Theory Manual
First, the linear growth rate (units of cm/sec) of the bulk phase n can be expressed
by:
Equation 16-12
l
K b ( n, m )
Wk
------- s k, m
G n, m =
k = K b ( n, m )
dG n, m
-------------d i
K b ( n, m )
=
F
k = K b ( n, m )
W k ds k, m
------- ------------k d i
Equation 16-14
l
K b ( n, m )
Wk
-------
k=K
f
b ( n,
m)
l=1
s k, m
------------ l
s k, m
+ ------------
--------l
i, p l
ln G n, m
------------------- ln i
i G n, m
= --------------- --------------G n, m i
RD01400-C01-004-001A
231
CHEMKIN 4.0
232
RD01400-C01-004-001A
CHEMKIN 4.0
17
17
Rate-of-production Analysis
Another useful tool in understanding reacting-flow calculations is rate-of-production
analysis. Rate-of-production analysis determines the contribution of each reaction to
the net production or destruction rates of a species. Rate-of-production analysis is
particularly useful for 0-D and plug-flow systems, where the computational expense
for the added calculations is small and it is possible to consider data from a large
reaction set. For this reason, rate-of-production tables are printed to the diagnostic
output file for these types of models (based on user requests), to allow quick
identification of dominant reaction paths. For all other reactor models in CHEMKIN, rate
of production data can be obtained through the Graphical Post-processor. This data
option can be selected at the time that the solution file data is being imported into the
Post-processor.
17.1
Equation 17-1
M
Pk = k +
Am
------- s k, m =
V
m=1
ki qi +
i=1
m=1
Is
Am
------- ski q si
V
i=1
where ki and ki are the stoichiometric coefficients for the gas and surface
s
reactions, respectively, and q i and q i are the rate of progress of the I gas-phase
reactions and the I s surface reactions. The contribution to the rate of production of
species k from gas-phase reaction i is therefore
Equation 17-2
C ki = ki q i
RD01400-C01-004-001A
233
CHEMKIN 4.0
Am s s
s
C ki = ------- ki q i
V
where A m is the surface area of the material m with which the reaction i is
associated. Further details about how the reactions contribute to net species
production rates are found in Chapter 3 and Chapter 4. The reactor model computes
normalized values of the reaction contributions to the species production and
destruction rates. The normalized production-contributions for gas-phase reactions
are given by:
Equation 17-4
max ( ki, 0 )q i
p
C ki = -----------------------------------------------------------------------------------------------------------------------------------I
M
Am I s
s
s
-
max ( ki, 0 )q i
i = 1 max ( ki, 0 )qi + m = 1 -----V i=1
and the normalized destruction values for the gas-phase reactions is given by:
Equation 17-5
min ( ki, 0 )q i
d
C ki = ---------------------------------------------------------------------------------------------------------------------------------M
I
Am I s
s
s
-
min ( ki, 0 )q i
i = 1 min ( ki, 0 )qi + m = 1 -----V i=1
The normalized production values for the surface reactions are given by:
Equation 17-6
p
C kis
Am
------- max ( ski, 0 )q si
V
= -----------------------------------------------------------------------------------------------------------------------------------I
M
Am I s
s
s
max ( ki, 0 )q i
i = 1 max ( ki, 0 )qi + m = 1 -----V i = 1
and the normalized destruction values for surface reactions are given by:
Equation 17-7
d
C kis
Am
------ min ( ski, 0 )q si
V
= ---------------------------------------------------------------------------------------------------------------------------------I
M
Am I s
s
s
-------
min
(
,
0
)q
+
min ( ki, 0 )q i
ki
i
i = 1
V
m=1
i=1
234
RD01400-C01-004-001A
Theory Manual
C ki + C ki
= 1
= 1
i=1
i=1
and
Equation 17-9
I
Cki + Cki
i=1
i=1
RD01400-C01-004-001A
235
CHEMKIN 4.0
236
RD01400-C01-004-001A
CHEMKIN 4.0
Nomenclature
Latin Equation Symbols
The table below shows the conventions used in this document for latin equation
symbols.
Latin Symbols
Symbol
Description
Chapter
CGS Units
ai
Ch. 4
none
Ch. 4, Ch. 8
none
Ch. 3, Ch. 4
ergs/g
Ch. 3, Ch. 4
depends on
Ch. 3
ergs/g
Ch. 8
cm2
Ch. 12
cm2
Ch. 7
cm2
Ch. 4
cm2
coefficient expression
ak
o
ak
a nk
data
RD01400-C01-004-001A
237
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
Ch. 3
ergs/mole
Ch. 3, Ch. 4
ergs/mole
depends on reaction
mixture
o
Ak
Ai
Am
Ch. 8
cm2
bi
Ch. 4
none
Ch. 8
cm
Ch. 8
mole/cm3
Ch. 8
mole/cm2
Ch. 4
cal/mole
ergs/(g K)
ergs/(g K)
Ch. 5
ergs/(g K)
Ch. 5
ergs/(g K)
coefficient expression
ck
ci
cp
c pk
c pk, int.
c pk, rot.
238
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
c vk
Ch. 3
ergs/(g K)
Ch. 3
ergs/(g K)
Ch. 3
ergs/(g K)
Ch. 8
none
Ch. 7
none
Ch. 3
ergs/(mole K)
Ch. 9
none
ergs/(mole K)
Ch. 5
ergs/(mole K)
Ch. 5
ergs/(mole K)
Ch. 5
ergs/(mole K)
Ch. 3
ergs/(mole K)
cp
cv
C pk
CD
C pk
C pk, trans.
C pk, vib.
C pk, rot.
C vk
Ch. 9
none
Cp
Ch. 3, Ch. 9
ergs/(mole K)
Ch. 3
ergs/(mole K)
pressure
Cv
RD01400-C01-004-001A
239
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
C k
Ch. 3
mole/(cm3 sec)
Ch. 7
cm
g/(cm sec)
cm2/sec
cm2/sec
cm2/sec
Ch. 3
mole/(cm3 sec)
Ch. 3, Ch. 4
[cal/mole]*
ergs/mole
species
Dh
Dk
D k, j
D kj
D km
in species j
D k
Ei
E ion
Ch. 4
[cal/mole]*
E ion, 0
Ch. 4
[cal/mole]*
Ch. 4
[cal/mole]*
Ch. 4
none
Ch. 4
none
dependent reactions
E yield, 0
fi
F ki
240
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
Acceleration of gravity
Ch. 11
cm/sec2
gi
Ch. 4
none
Ch. 3, Ch. 4
ergs/g
Ch. 3
ergs/g
dependent reactions
o
gk
Ch. 5
none
Gk
cm/sec
Ch. 3, Ch. 4
ergs/mole
Ch. 3
ergs/mole
ergs/g
Ch. 9
ergs/mole
ht
Ch. 8
ergs/(sec K)
h e
Ch. 8
ergs/g
species
o
Gk
hk
Ch. 3, Ch. 7
ergs/g
h yield
Ch. 4
depends on reaction
Enthalpy of formation
Ch. 4
ergs/mole
Ch. 3, Ch. 4
ergs/mole
Ch. 3, Ch. 4
ergs/mole
expression
Hf
o
Hk
species
Hk
RD01400-C01-004-001A
241
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
Ch. 3
ergs/mole
none
jk
Ch. 14
mole/(cm2 sec)
surface
jk
Ch. 5
ergs/g
Jacobian matrix
Ch. 12
none
Ch. 14
none
kB
Boltzmann Constant
Ch. 5
ergs/K
k fi
depends on reaction
k ri
Ch. 3, Ch. 4
depends on reaction
none
Kb
none
Kb ( m )
Ch. 8
none
Ch. 14
none
Ch. 11
none
Ch. 4
none
Ch. 8
none
Ch. 14
none
material.
f
Kb
K b(n)
phase n
f
Kb
242
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
Ch. 4, Ch. 11
none
Ch. 8
none
depends on reaction
none
none
Kg
none
K pi
depends on reaction
K b(n)
l
K b (n,m)
phase n
Index of the last bulk-phase species in
phase of the mth material.
K ci
Kg
f
Kg
l
Ks
none
Ks ( m )
Ch. 8
none
Ch. 14
none
Ch. 4, Ch. 11
none
Ch. 8
none
Ch. 14
none
Ch. 4, Ch. 11
none
Ch. 8
none
mth material
f
Ks
K s(n)
f
K s (n,m)
l
Ks
K s(n)
l
K s (n,m)
RD01400-C01-004-001A
243
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
K phase
Ch. 4, Ch. 14
none
Ch. 8
none
lm
Ch. 7
cm
Ch. 8
cm
Ch. 8
cm
Ch. 5
in each phase
LA
LC
m jk
mk
Ch. 4, Ch. 5
m i
Ch. 9
g/sec
m l
Ch. 11
g/(cm(1-) sec)
m o
Ch. 9
g/sec
m u
Ch. 11
g/(cm(1-) sec)
m 0
Ch. 11
g/(cm(1-) sec)
Mass flux
Ch. 11
g/(cm(1-) sec)
Ch. 12
g/sec
Ch. 8
none
Reactor mass
Ch. 9
Ch. 14
g/mole
reaction mechanism
Mk
replaces W k .
244
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
Ch. 14
g/mole
replace M .
[M]
Ch. 3
mole/cm3
Ch. 4
cm
none
Ch. 9
none
Ch. 4
none
Ch. 8
none
to C p R
Total number of event (statistical)
particles
N(m)
material
NA
Avogadros number
Ch. 4
none
Nb
Ch. 4, Ch. 14
none
Nc
Ch. 9
none
Ch. 8
none
Ch. 4, Ch. 14
none
Ch. 8
none
Ch. 4, Ch. 14
none
Ch. 8
none
Ch. 4, Ch. 8
none
be effected by convection
Nb ( m )
Nb
f
N b(m)
material
l
Nb
l
N b(m)
material
Ng
RD01400-C01-004-001A
245
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
Nm
Ch. 9
none
none
Ch. 8
none
Ch. 4, Ch. 14
none
Ch. 8
none
Ch. 4, Ch. 14
none
N s(m)
Ch. 8
none
Nu h
Ch. 8
none
Ch. 14
1/sec
Ns
Ns ( m )
Ns
f
N s(m)
l
Ns
l
oi
Pressure
dynes/cm2
Pm
Ch. 14
dynes/cm2
P atm
dynes/cm2
Ch. 9
none
Ch. 14
ergs/(cm2 sec)
Ch. 8
none
Ch. 14
ergs/sec
surface
Pr
246
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
q''
Ch. 14
ergs/(cm2 sec)
Heat flux
Ch. 5
ergs/(cm2 sec)
Ch. 14
mole/(cm3 sec)
Ch. 4
mole/(cm2 sec)
Ch. 8
ergs/sec
a heat source
qi
reaction.
q wall
Radial coordinate
Ch. 14
cm
ergs/(mole K)
Rc
Ch. 3, Ch. 4
[cal/(mole K)]*
Ch. 4
none
R ki
R CA
Ch. 8
none
Re
Ch. 8
none
Ch. 3, Ch. 4
ergs/(g K)
sk
Ch. 3
ergs/(g K)
s k
mole/(cm2 sec)
Ch. 3, Ch. 4
ergs/(mole K)
Ch. 3
ergs/(mole K)
surface reactions
o
Sk
Sk
RD01400-C01-004-001A
247
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
Sq
Ch. 3
ergs/(cm3 sec)
source
Ch. 3
ergs/(mole K)
Sp
Ch. 8
cm/sec
Time
sec
ti
Ch. 4
none
expression
tl
Laboratory time
Ch. 7
sec
tp
Gas-particle time
Ch. 7
sec
Temperature
Ch. 5
none
Tk
Ch. 5
none
To
Ambient temperature
Te
Electron temperature
Ch. 4, Ch. 8
T ion
Ch. 8
T surf
Surface temperature
Ch. 8, Ch. 14
T wall
Ch. 8
Ch. 14
T ik
*
248
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
cm/sec
Ch. 7
cm/sec
Ch. 4
none
Ch. 3, Ch. 4
ergs/g
Ch. 3
ergs/g
Ch. 7
cm/sec
Ch. 3
ergs/mole
Ch. 4
ergs/mole
U rs
Ch. 7
cm/sec
Us
Ch. 7
cm/sec
Ch. 4
ergs/mole
direction)
ui
expression
uk
Uk
Uk
species
Ch. 3
ergs/mole
Ch. 11
cm/sec
Ch. 5, Ch. 7
cm/sec
Volume
cm3
Ch. 14
1/sec
Ch. 8
cm3
cm/sec
Vc
RD01400-C01-004-001A
249
CHEMKIN 4.0
Nomenclature
Latin Symbols
Symbol
Description
Chapter
CGS Units
Vk
cm/sec
V k, y
Ch. 11
cm/sec
Ch. 5, Ch. 12
cm/sec
Ch. 14
cm
Ch. 7
none
Ch. 14
1/sec
g/mole
Wk
g/mole
Wk
Ch. 5, Ch. 12
cm/sec
Ch. 4
g/mole
cm
Ch. 14
cm
Vk
the y division
Ordinary diffusion velocity of the kth
species
ws
species
W ion
xs
Ch. 4
none
Xk
none
[ Xk ]
mole/cm3
250
RD01400-C01-004-001A
Nomenclature
Theory Manual
Latin Symbols
Symbol
Description
Chapter
CGS Units
Cross-stream coordinate
Ch. 11
cm
y max
Ch. 11
cm
Yk
none
Distance
Ch. 7
cm
Ch. 4
none
Ch. 7
none
Reactor.
Zk
Ch. 14
none
Zk ( n )
Ch. 4, Ch. 11
none
Z k (n,m)
Ch. 8
none
surface phase n
The energy CGS units that are noted by an asterisk (*) depend upon user input at the
REACTION line or in the UNITS auxiliary keyword for chemistry input files. The default unit is
calories.
RD01400-C01-004-001A
251
CHEMKIN 4.0
Nomenclature
Description
Chapter
CGS Units
Ch. 11
none
Ch. 7
none
Ch. 7
none
radial coordinates
'
Parameter in mechanism
Ch. 4
none
Ch. 5
Ch. 5
none
Ch. 4
none
Ch. 7
none
none
Ch. 7
none
Ch. 4
mole/cm2
Ch. 4
mole/cm2
tot
Ch. 4
mole/cm2
Ch. 9
sec
species.
ki
n
o
phases
mix
252
RD01400-C01-004-001A
Nomenclature
Theory Manual
Greek Symbols
Symbol
Description
Chapter
CGS Units
res
Ch. 9
sec
Ch. 4
mole/(cm2 sec)
Ch. 12
none
Ch. 7
none
Ch. 4
none
curvature resolution
reaction
Ch. 9
sec
Ch. 12
none
Unit tensor
Ch. 5
none
Ch. 5
none
Ch. 5
none
Ch. 5
none
Emissivity
Ch. 4
none
Ch. 9
cm2/sec3
Ch. 5
ergs
Ch. 12
none
Ch. 5
ergs
Ch. 4
[cal/mole]*
gradient
jk
species
*
dissipation rate
jk
ki
RD01400-C01-004-001A
Coverage parameter
253
CHEMKIN 4.0
Nomenclature
Greek Symbols
Symbol
Description
Chapter
CGS Units
Mixture viscosity
Ch. 5
g/(cm sec)
Ch. 7
none
2 1
k
Ch. 5
g/(cm sec)
ki
Coverage parameter
Ch. 4
none
air
Ch. 8
g/(cm sec)
Ch. 5, Ch. 12
none
Ch. 8
degrees
Ch. 5
g/(cm sec)
Ch. 9
cm2/sec2
ergs/cm sec K
Ch. 12
none
Ch. 5
ergs/cm sec K
(n)
Ch. 5
ergs/cm sec K
air
Ch. 8
ergs/cm sec K
Mixture viscosity
Ch. 7, Ch. 11
g/(cm sec)
jk
Ch. 5
Debye
Ch. 4
none
Ch. 5
Debye
254
RD01400-C01-004-001A
Nomenclature
Theory Manual
Greek Symbols
Symbol
o
Description
Chapter
CGS Units
Ch. 4
none
Ch. 5
none
species
ki
Coverage parameter
Ch. 4
none
Kinematic viscosity
Ch. 7
cm2/sec
Ch. 11
none
Density ratio, 5 1
Ch. 7
none
ij
Ch. 5
none
g/cm3
Ch. 11
g/cm3
g/cm3
n, m
Ch. 8
mole/cm2
Ch. 8
g/cm3
Ch. 4
none
Stefan-Boltzmann constant
Ch. 14
none
Stefan-Boltzmann constant
Ch. 4, Ch. 14
Ch. 5
Ch. 5
mth material.
air
species
jk
RD01400-C01-004-001A
255
CHEMKIN 4.0
Nomenclature
Greek Symbols
Symbol
Description
Chapter
CGS Units
Ch. 14
sec
Momentum flux
Ch. 5
g/(cm sec2)
Ch. 3
sec
none
none
none
Ch. 4, Ch. 12
none
ki
ki = ki ki
ki
ki
Ch. 4, Ch. 8
none
Stream function
Ch. 11
g/(cm sec2)
Ch. 4
none
Ch. 8
1/sec
Collision integral
Ch. 5
none
mole/(cm3 sec)
yield reaction)
The energy CGS units that are noted by an asterisk (*) depend upon user input at the
REACTION line or in the UNITS auxiliary keyword for chemistry input files. The default unit is
calories.
256
RD01400-C01-004-001A
Nomenclature
Theory Manual
Description
Chapter
Ch. 7
Ch. 7
shock
Ch. 7
incident shock
Ch. 7
reflected shock
RD01400-C01-004-001A
Grid-point index
Ch. 12
Species index
Ch. 8
Phases index
Ch. 4, Ch. 8,
Ch. 5
Ch. 5
Ch. 7
257
CHEMKIN 4.0
Nomenclature
258
RD01400-C01-004-001A
CHEMKIN 4.0
Index
Index
A
absolute
tolerance 211
absolute tolerance 218
acceleration, of charged species 127
accuracy 79, 87
activation energy 57
activities
bulk 19
bulk species 19
activities for bulk species, treatment of 133
adaptation 221, 222
adiabatic 158, 169
adiabatic, freely propagating
flame 182
adsorption 61
reaction 52, 53
algebraic equations 168, 170, 171, 175, 207
algorithm
Newton 207, 210, 213
allowable rate constant fitting options 48
ambient temperature 127, 159
analysis
sensitivity 68, 222, 227, 229
Application 19
CHEMKIN 24
applications
BOHM condition for ion fluxes to surfaces 142
approximation
approximation 139
approximations, finite difference 183, 204
arbitrary reaction order 35
area
surface 116, 117, 120, 121, 122, 123, 234
Arrhenius
coefficients 55
temperature dependence 55
Arrhenius form 179
RD01400-C01-004-001A
259
atomic
site formalism 52, 53
atoms 81
AURORA
assumptions and limitations 116
auxiliary
keywords
BOHM 66
COV 57
ENRGDEP 67
MWOFF 61
NONCON 68
REV 55
YIELD 68
average
molar concentration 124
axial
diffusion 156
mixing 155
velocity 156
axial velocity 188
B
back substitution 227
backward
difference 221
Euler method 186
backward-Euler method 213
backwards finite differences 213
backwards-differencing methods 227
backwards-differentiation formula 228
balance
elemental 52, 67, 68
energy 70, 71, 72, 126, 139
mass 70
site 52
surface-species 123
banded structure 220
CHEMKIN 4.0
Index
C
calculation
transient 229
capability, continuation 208
capacities, heat 92
central difference 183
260
RD01400-C01-004-001A
Index
Theory Manual
RD01400-C01-004-001A
continuation
capability 208
continuation start-up procedure 216
continuity equation 156, 178, 210
control
volume 126
convection 152
convective
flux 71
mass flux 70
convective heat
loss 137
convergence 171, 176, 208, 214, 215, 221
conversion
rate of 116
conversion formulas 20
coordinates
cylindrical 165, 167, 168, 175, 188
laboratory 109
planar 165, 167, 168, 188
shock-fixed 109
transformed 166
correction
factor 60
correction velocity 94, 172, 181, 184
coverage
parameters 57, 58, 70
crank arm
speed of revolution of the 134
criterion
Bohm 142, 143
cross-sectional
(flow) area 156
cross-sectional area 114
Curtiss-Hirschfelder approximation 180
cylinder
bore diameter 135
combustion 134
cylindrical coordinates 165, 167, 168, 175, 188
D
damping parameter 211, 217
DASPK 222, 227
DASSL 160, 168, 175, 176
modified version of 175
data
thermodynamic 25
database
thermodynamic 75, 92
defining equations 175
energy 165
momentum 164
species 165
261
CHEMKIN 4.0
Index
state 165
density
electron number 139
ion 142, 144
mass 72, 120, 121, 124, 139, 231
sites of phase 19
surface site 56
surface-phase 144
Deposition 123
deposition 117, 119, 125, 139, 208
chemical vapor 61
level, power 208
deposits
bulk phase 125
derivatives
time 213
derived sensitivity 229
desorption 52, 61
destruction rates 233
diagnostics
error 176
diameter
collision, reduced 78
cylinder bore 135
engine bore 137
tube 159
difference
finite 173
differential equations
ordinary 213, 222, 227
differential, ordinary 170
diffusion
coefficients 75
binary 191
mixture-averaged 191
multicomponent 191
thermal 71, 191
coefficients, thermal 75, 84, 87, 89, 90
Fickian 85
ratios, thermal 87, 89
thermal 173
velocities 70, 84, 190
velocity 165, 171, 172
diffusion rates 116
diffusion velocity 179, 180, 181
diffusive
flux 71
mass flux 70
dilution 23
dipole
moment 78
disk, rotating 196
displaced volume 134
distance-time diagram 101
distribution
non-Maxwellian velocity 60
velocity 60
distribution function
electron energy 138
drag force 157, 159
driver gas 100
dyadic product 83
E
efficiency
engine 136
eigenvalue 182, 190
elastic, momentum-transfer collisions 140
electric-fields 127
electromagnetic fields
externally applied 139
electron
energy
distribution function 138
equation 210
equation for plasma systems 139
loss per electron-impact reactions 141
enthalpy 139
mass 139, 211
temperature 66, 119, 121, 129, 141, 144, 210
electron number
density 139
electron-driven chemical kinetics 138
electronegative 142
gas 142, 144
electronic charge
integer 143
electron-impact
collisions 140
reactions 141
elemental
balance 52, 67, 68
composition 14
elementary reactions 14, 35
elements 52
Eley-Rideal 61
endothermicity 71
energy
activation 57
balance 70, 71, 72, 126, 139
conservation 70
defining equations 165
equation
electron, for plasma systems 139
transient 129
Gibbs free 208
Gibbs free 56
262
RD01400-C01-004-001A
Index
Theory Manual
RD01400-C01-004-001A
F
factor
correction 60
energy-dependent multiplicative 68
pre-exponential 57
factorization, LU 227
Fickian
diffusion 85
formula 84
fictitious transient equations 160
film thickness 124
finite difference 173, 186, 214, 217, 219, 220
approximation 173
approximations 183
finite difference approximations 204
finite differences
backwards 213
finite-differences
perturbation 212
fitting
errors 83
procedure 76, 82
fitting options
allowable rate constant 48
flame experiments 225
263
CHEMKIN 4.0
Index
264
G
gas
conductivity 137
electronegative 142, 144
energy equation 126, 210
adjusted for plasma systems 142
phase 15, 17, 71
temperature 119, 121, 126, 128, 129, 142, 144, 209,
225
viscosity 137
gas temperature 157
gas-particle time 101
gas-phase
species 15, 56, 60, 69, 70, 72
indices 15
species equation 119
GAS-PHASE KINETICS 75
multi-fluid 119
Pre-processor 35, 66
subroutine library 191
gas-surface interface
flux-mathing conditions 70
Gauss-Chebyshev grid 46
Gaussian 215
Gibbs free energy 208
Gibbs free energy 56, 96
minimization of 96
of reaction 56
Gibbs function 96
global
expressions 35
reaction 32
governing equations 72, 158, 159
Greek equation symbols 252
grid 178, 183
non-uniform 174
growth rate 72
bulk phase 225
linear 125, 231
sensitivity analysis 229
H
heat
capacity 157
flux 127, 157, 158
loss
convective 137
of formation 54
radiative 71
release, chemical 71
transfer 127, 137
coefficient 127, 137
RD01400-C01-004-001A
Index
Theory Manual
heat capacities 92
heat capacity 179
heat exchange
reactor to reactor 129
heat transfer 130
heat-transfer
coefficient correlation 136
Henrys law 23
heterogeneous production rates 157
holder, substrate 117
homogeneous reactions 155
hydraulic diameter profile
mean 159
I
ICEN model 135
ideal
solution 24
ideal gas 81
ideal-gas equation of state 157
implementation of multicomponent TRANSPORT 171
index
bulk species 15
first gas-phase species 15
last gas-phase species 15
surface phase 122
surface species 15
inelastic collisions 140
inertia 157
initial
conditions
on species concentrations at boundaries 170
profile 169
initial conditions
reflected SHOCK 103
inlet
mass flow rate 120
installation 17
integer
stoichiometric coefficients 35
integer electronic charge 143
integrals
collision 76, 77, 78, 79, 88
Interaction-by-Exchange-with-the-Mean 149
internal
combustion engine model 114
energy 126, 127, 139, 140
parts 92
internal combustion engines 177
internal surface area 156
ion
accelerated 65
bombardment 65
RD01400-C01-004-001A
265
J
Jacobian 171, 174, 175
matrix 175, 217, 219
Jacobian matrix 122, 210, 212, 228
joint PDF of scalars 147
K
keywords
auxiliary
BOHM 66
COV 57
ENRGDEP 67
MWOFF 61
NONCON 68
REV 55
YIELD 68
kinetic theory 75
kinetics, mass-action 35
L
L matrix 89
laboratory
fixed coordinates 109
time 101, 102, 103
laminar flame speed 177
laminar flow 159
Landau-Teller
formulation of the rate expressions 47
Langmuir-Hinshelwood 61
Latin equation symbols 237
Lennard-Jones
collision diameter 77
potential well depth 77
light-component limit 180
CHEMKIN 4.0
Index
limit
Bohm 144
pure species 85
linear growth rate 72, 125, 231
of the bulk phase 125
linear molecule 92
Linear-Mean-Square-Estimation 149
lines
method of 168
LINPACK 227
local-sensitivity 229
losses
collisional 139
surface 139
low-pressure
plasmas 142
processes 116
LU
decomposition 219
LU factorization 227
M
Mach number 105
manipulation of chemical rate sensitivity
coefficients 68
mass
balance 70, 156
conservation 76, 93, 172
conservation equation 119
density 72, 120, 121, 124, 139, 156, 179, 231
bulk species 72
density, bulk 124
electron 139, 211
flow rate 117, 120, 121, 156, 178, 179, 218, 220
inlet 120
outlet 120
flux 83, 166, 167, 169, 170, 172
convective 70
diffusive 70
flux fractions 185
fraction 120, 128, 156, 179, 181, 210, 211, 219, 227,
229, 230
266
RD01400-C01-004-001A
Index
Theory Manual
N
net chemical production rates 139
net production rate 54
network, reactor 116
Newton
algorithm 207, 210, 213, 214
iteration 208, 211, 218, 221
Newtons method 209, 210, 212, 213, 214, 216, 217, 221,
RD01400-C01-004-001A
227
267
CHEMKIN 4.0
Index
equations 170
ordinary differential equations 156, 213, 222, 227
ordinary diffusion velocity 180, 181
ordinary multicomponent diffusion coefficient 71
outlet mass flow rate 120
P
packed beds 155
parabolic profile 169
parameter
characteristic 121
damping 211
parameters
coverage 57, 58, 70
perfect
solution 23
perfect gas 81
perturbation
finite-difference 212
PFR 155
phase 14, 15
bulk 14, 119, 122, 123, 124, 125, 134, 210, 230
density of sites 19
gas 15, 17, 71
indices 15
surface 23, 56, 60, 72, 73, 119, 121, 122, 123, 126, 127,
210
268
polynomials
Chebyshev 45, 46
potential, chemical 23
power
bias 141
deposition 139
level 208
Prandtl number 137
pre-exponential
constant 69
factor 57
premixed flame 177
equations
boundary conditions 185
Pre-processor
GAS-PHASE KINETICS 35, 66
SURFACE KINETICS 35, 53, 55, 68
TRANSPORT 89
pressure 23
units 56
pressure forces 157
pressure-dependent reactions 38
probability 58
procedure
fitting 76, 82
pure-species fitting 82
time-stepping 207, 212
procedure, continuation start-up 216
processes
low-pressure 116
product
dyadic 83
production
molar rate of 179
rate 70, 73
rates, species 69, 73
rates, surface phase 73
production rates
bulk species 231
molar surface 120
net chemical 139
products 54, 55
profile
initial 169
parabolic 169
velocity 169
properties
mixture 75
mixture-averaged 76, 87
mixture-averaged TRANSPORT 180
multicomponent 76, 87, 89
multicomponent TRANSPORT 181
pure species 75
thermodynamic 14, 56
RD01400-C01-004-001A
Index
Theory Manual
variables 55
rates
species production 69, 73
surface phase production 73
ratio
compression 134
surface-to-volume 116
ratios
thermal diffusion 87, 89
raw sensitivity coefficients 229
reactants 54, 55
reacting flow 233
reaction
adsorption 52, 53
Bohm 143
chemical 35, 51, 65, 71
chemically activated bimolecular 42
electron-impact 141
elementary 14, 35
excitation 140
global 32
ion-energy-dependent 67
ion-enhanced yield 68
ionic 66
irreversible 66, 67, 68
non-site-conserving 144
orders 60
pressure-dependent 38
rate constants, surface 58
rate expressions, chemical 54
rate of progress 62
rate-of-progress 35, 57
reversible 55, 133
surface 53, 54, 58, 67, 69, 70, 71, 72, 73
three-body 37
unimolecular/recombination fall-off 38
yield, ion-enhanced 67
reaction-orders, changing 56
reaction-rate
coefficients 138, 225, 229
reactor
cluster 116
geometry 158, 159
network 116
plasma-etch 116
volume 116, 117, 120, 123
walls 116, 117, 126
well mixed 117, 118, 119, 123
Reactor Clusters 116
reactor equations 151
Reactor Networks 116
reactor networks 116
real
numbers 35
Q
quasineutral bulk plasma 139
R
radial velocity 188
radiative heat 71
Rankine-Hugoniot 100, 103, 106, 107, 110
Raoults law 23
rarefaction wave 100
rate
coefficients, forward 133
constants 60
forward 55
mass-action kinetics 60
reverse 55, 58
units 55
destruction 233
diffusion 116
expression
ion-energy-dependent 66
surface-coverage modification of 57
expressions
chemical reaction 54
flow 116, 117, 120, 121
growth 72
inlet mass flow 120
mass flow 117, 120, 121
molar surface production 120
net production 54
of conversion 116
outlet mass flow 120
production 70, 73
sensitivity coefficients, chemical 68
rate coefficients 179
rate expression
Arrhenius 62
Langmuir-Hinshelwood and Eley-Rideal 61
rate of production 233
rate-of-production 54, 72
molar 120
rate-of-progress 143
calculated 61
reactions 35, 57
RD01400-C01-004-001A
269
CHEMKIN 4.0
Index
raw 229
derived 229
of bulk growth or etch rates 230
time-integrated 229
sensitivity analysis 68
sheath
plasma 140, 142
sheath, plasma 65
shock
front 100, 103, 106
tube 99, 100, 103, 111
velocity, reflected 99, 107, 109
wave 99, 100, 101, 103
wave, reflected 100
shock-fixed coordinates 109
silicon wafer 117
site
balance 52
densities, non-constant surface-phase 126
densities, surface 56
density 158, 160
fractions 14, 19, 57, 59, 69, 72, 121, 122, 144, 157,
208, 210
of phase, density 19
open 14, 52
type 14
software package, numerical 207
solution
ideal 24
perfect 23
variables, summary of 144
special forms of the rate expressions 48
species
bulk 14, 15, 19, 51, 52, 69, 72
chemisorbed 121
concentration 35
concentrations at boundaries, initial
conditions 170
conservation 93
defining equations 165
energy 67, 68
equation 179
equations 210
gas-phase 15, 56, 60, 69, 70, 72
indices 15
physisorbed 121
production rates 69, 73
surface 51, 69, 72
transport equations 178
types
bulk 13
gas-phase 13
surface 13
species conservation 190
S
scalar mixing time 147
self-diffusion coefficient 81
sensitivity
analysis 222, 227, 229
for steady-state solutions 226
for transient solutions 227
growth rate 229
coefficient matrix 226, 227
coefficients 225, 226, 228, 229, 230, 231
normalized 229
270
RD01400-C01-004-001A
Index
Theory Manual
specific
enthalpy 128
heat, mean gas 129
specific enthalpy 157
speed
of revolution, of the crank arm 134
piston 137
thermal 142
sputter 65
stagnation-point flow equations 196
standard
state 23
STANJAN 12, 95, 97
starting
estimates 208
state
defining equations 165
equation, of 81
standard 23
state equation 179
steady state 155, 160
steady-state 115, 119, 120, 122, 125, 212, 213, 225, 226,
229, 230
RD01400-C01-004-001A
phase 23, 56, 60, 72, 73, 119, 121, 122, 123, 126, 127,
210
index 122
production rates 73
reaction rate constant 58
reactions 53, 54, 58, 67, 69, 70, 71, 72, 73
site
concentration 60
density 56
open 121
species 51, 69, 72
conservation equation 121, 157
equations 121
indices 15
SURFACE KINETICS
Pre-processor 35, 53, 55, 68
subroutine library 69
surface site fractions 168, 170
surface-coverage modification 58
of rate expression 57
surface-phase
densities 144
site densities, non-constant 126
surface-species
balances 123
surface-to-volume ratios 116
systems
multi-temperature 128
systems, plasma 67
T
temperature
ambient 127, 159
dependence, Arrhenius 55
electron 66, 119, 121, 129, 141, 144, 210
exponent 66
gas 119, 121, 126, 128, 129, 142, 144, 157, 209, 225
ion 119, 142
profile 178, 216
user-specified 216
reduced 77, 78, 79
surface 144
thermodynamic 66
test gas 100, 101, 103, 105
theory 96
theory, kinetic 75
thermal
conduction 138
conductivities 75, 82, 84, 90
mixture-averaged 87
pure species 76, 79
conductivity 179, 180, 181
diffusion 181
271
CHEMKIN 4.0
Index
transport
constraints 116
mass 116
properties 180, 181
transport equation for the joint PDF 148
transport properties 156, 159
trapezoidal rule 174
treatment of
activities for bulk species 133
tube
diameter 159
flow configuration 155
turbulent flow 159
turbulent mixing 116
turbulent reactive flows 146
TWOPNT 171, 210, 212, 213
type of site 14
U
undamped correction vector 211
unimolecular/recombination fall-off reactions 38
units
concentration 56
pressure 56
rate constants 55
universal gas constant 60, 179
unmixedness 150
unphysical results 56
V
value
mass-averaged 129
variables
rate-of-progess 55
vector
residual 209
undamped correction 211
velocity 83, 94
velocity
axial 188
Bohm 66, 142
conservation diffusion 94
correction 94, 172
diffusion 70, 84, 165, 171, 172, 190
distribution 60
distribution, non-Maxwellian 60
profile 169
flat 169
radial 188
Stefan flow 71
thermal diffusion 88
vector 83, 94
272
RD01400-C01-004-001A
Index
Theory Manual
version
of DASSL, modified 175
vibrational contributions 79
viscosity 75, 76, 82, 83, 87, 159
gas 137
mixture 83, 87
pure species 76
viscosity correction 113
viscous drag 157, 159
volume
clearance 134
control 126
displaced or swept 134, 135
engine cylinder 135
reactor 116, 117, 120, 123
Von Mises transformation 164
W
wafer, silicon 117
Wall energy balance 131
walls
reactor 116, 117, 126
weight
molecular 60, 66, 71, 120, 124, 128, 229, 231
well mixed reactor 117, 118, 119, 123
Wilke formula 87
windward difference 183
X
XMLdata.zip 225
Y
yield 68
enhancement 67, 68
ion-enhanced 68
ion-enhanced reaction 67
Z
ZEROIN 105
RD01400-C01-004-001A
273