evaluation of the evaluated data...2012/09/04 · equation) with bolsig+ which is based on the...
TRANSCRIPT
Y. Itikawa
At the IAEA Consultants Meeting
7-9 February, 2012, NIFS
my personal experience in data evaluation was reported.
Today
one step further !
There is no standard method of data evaluation.
→
The result of evaluation (i.e., the recommended data) may be different depending on who evaluates them.
We need
evaluation of the evaluated data
Examples of attempt:
(1)Consistency of the total scattering cross section
(2)Swarm parameters
(3)Modelling result
Consistency of
the total scattering
cross section
QT = Σ Qs S= elastic and all inelastic processes
QT can be measured directly.
We can test
QT (measured) = Σ Qs ?
0.01
0.1
1
10
100
cross
sec
tion (
10
-1
6 c
m2)
0.001 0.01 0.1 1 10 100 1000
electron energy (eV)
elas
tot
mom transf
rot J=0 2
v=0 1
ion(tot)
diss
ion(diss)
a
B
C
e + N2
vib v=0 1
e + N2 Itikawa, JPCRD 35, 31 (2006)
Energy (eV) 100 200 500 1000 uncertainty
elastic 5.6 3.5 1.85 1.0 20%
ionization 2.51 2.19 1.36 0.85 5%
dissociation 1.16 0.95 20%
electronic
exc
0.43 25%
sum 9.70 6.64 3.21 1.85
Total
(measured)
8.94 6.43 3.55 2.13 5%
e + N2 (cross sections in 10-16 cm2 )
e + O2 Itikawa, JPCRD 38, 1 (2009)
Energy (eV) 100 200 500 1000 uncertainty
elastic 4.78 3.15 1.72 1.10 20%
ionization 2.43 2.28 1.46 0.922 5%
dissociation 0.33 0.29 34%
electronic
exc
0.41 0.26 0.13 0.08 25%
sum 7.95 5.98 3.31 2.10
Total
(measured)
8.68 6.24 3.58 2.08 5%
e + O2 (cross sections in 10-16 cm2 )
This relation generally holds, when uncertainty of individual cross sections is considered.
Details
We have no data, or less reliable data, for excitation and dissociation.
Furthermore, a part of excitation
contributes to diss.
→ Double counting ?
But contribution of exc & diss to QT is small. It is not a serious problem.
The most serious problem is the uncertainty of elastic cross section.
Can we measure the elastic cross section within an uncertainty of 5% ?
Swarm parameters
Website for
collecting, displaying and
downloading
electron-scattering cross sections
and
swarm parameters
(mobility, diffusion coefficient,
reaction rate etc)
Developed and maintained by
the Toulouse group (L.C.Pitchford et al.)
See
S.Pancheshnyi et al.
Chem. Phys. 398, 148 (2012)
(1)Make an intercomparison of the cross section data sets
(2)Calculate swarm parameters and compare them with experimental values
(with solving the Boltzmann
equation)
with BOLSIG+
which is based on the two-term
approximation of the electron
distribution function
See
Hagelaar and Pitchford
Plasma Sources Sci. Technol. 14, 722 (2005)
All the following figures of swarm parameters are provided by Leanne Pitchford.
0.1 1 10 100
1x1024
1025
Experiments : Nakamura
calculations : Itikawa
µN
(m
. v.
s)-
1
E/N (Td)
N2
1 Td = 10-21 V m2
0.1 1 10 100
0.01
0.1
1
10
calculations
Itikawa data
Expriment
Crompton
Jory
Crompton 1963
townsend
Naidu
Warren
Fletcher
Wedding
DT
/µ (
eV
)
E/N (Td)
N2
0.01 0.1 1 10 100 1000 10000
1x1024
1025
1026
Experiments
Nelson
Pack
Herreng
Nielsen
Fleming
Doehring
Naidu
Fromhold
Schlumbohm
Roznerskhi(below 27.5Td)
Roznerskhi(taken 293K)
Mobili
ty (
/m/V
/s)
E/N (Td)
Calculations
Biagi LXCat and Bolsig+
Phelps LXCat and Bolsig+
Trinite LXCat and Bolsig+
Biagi Monte Carlo
Itikawa No Dissoc LXCat and Bolsig+
O2
0.1 1 10 100 1000
0.1
1
10
Calculations
Biagi LXCat and Bolsig+
Phelps LXCat and Bolsig+
Trinite LXCat and Bolsig+
Biagi Monte Carlo
Itikawa No Dissoc LXCat and Bolsig+
D/
(eV
)
E/N (Td)
experiments
Fleming
Rees
Huxley
Naidis
O2
Mobility and diffusion coeffcient calculated with Itikawa’s cross sections generally agree with the corresponding measured values, except for some discrepancy at high E/N.
A part of electronic-state excitation contributes to dissociation. Can we separate that part ?
In the present calculation, we simply ignore the dissociation process.
We found that
the dissociation has a significant effect in the case of O2,
but much less effect in the case of N2.
Chemical modelling
with reaction rate data
base
A wide variety of molecules have been found in interstellar space
→ (Interstellar) Molecular Cloud
→ Formation of stars
Calculation of synthesis and destruction of interstellar molecules
with solving time-dependent rate equations
→ Molecular abundance (time dependent)
Needs reaction rate coefficients
V. Wakelam et al.,
Space Sci. Rev. 156, 13 (2010)
Reaction Networks for Interstellar Chemical Modelling: Improvements and Challenges
udfa
(The UMIST Database for Astrochemistry)
420 species, 4573 reactions
www.udfa.net
OSU model
(Ohio State University, Astrophysical Chemistry
Group)
468 species, 6046 reactions
www.physics.ohio-state.du/~eric/research.html
KIDA
(Kinetic Database for Astrochemistry)
474 species, 6090 reactions
kida.obs.u-bordeaux1.fr/models
Comparison with observation
Example:
negative ions (recently observed)
using udfa
Observation vs model : Molecular abundance In the circumstellar region IRC+10216
Error bars of calculation are due to the uncertainties in the rate coefficients.
Discrepancy is caused probably by
incompleteness in the chemical network
or
wrong reaction rates of key reactions involving those species
Intercomparison of models
Example
abundances of H3O+, HCO+,
etc
Kida vs OSU vs udfa
Molecular abundances in a dense cloud (T=10 K)
Black dotted lines are 2σ error bars of kida results.
The three different chemical models are consistent with each other, if the uncertainties of the rate coefficients are considered.
But
Quite a small number of reaction processes have measured or calculated rate coefficients at low temperature.
Then many processes have rate coefficients obtained by a combination of experiment, theory and common sense (chemical intuition !).
Any attempt of application of the evaluated data can be a chance of the evaluation of them.
In this sense, comments of users are helpful for evaluation of the evaluated data.
Website of database should have a page for feedback where users can easily contact and send their comments.