Precision Electromagnetic Physics in Geant4: the Atomic Relaxation Models

1
Precision Electromagnetic Physics in Geant4 the
Atomic Relaxation Models

• A. Mantero, B. Mascialino, Maria Grazia Pia, S.
  Saliceti
• INFN Genova, Italy

CHEP, Interlaken, 27-30 September 2004
http//www.ge.infn.it/geant4/lowE/index.html
2
Geant4 Low Energy Electromagnetic Physics

• Geant4 provides a specialised package to handle
  electromagnetic interactions down to low energy
• Low means up to 100 GeV

Negative charged hadrons
Positive charged hadrons and ions
Electrons and photons
Bethe-Bloch
Models based on Livermore Library (EEDL, EPDL)
Quantum Harmonic Oscillator
high energy
Ziegler/ICRU Parameterisations
low energy (lt 1 keV)
down to 250 eV (lower in principle)
MeV region
Penelope re-engineering
same as positive hadrons
Free electron gas
down to 100 eV
low energy (down to ionisation potential)
3
Vision

• Precise process modeling
• Cross sections, angular distributions
• Charge dependence
• Relevant at low energies
• Take into account the atomic structure of matter
• Detailed description of atoms (shells)
• Secondary effects after the primary process
• De-excitation of the atom after the creation of a
  vacancy

• X-ray fluorescence
• Auger electron emission
• PIXE (Particle Induced X-ray Emission)

Atomic Relaxation
following the creation of a vacancy by
photoelectric effect, Compton effect and
ionisation
4
The process in a nutshell

• Rigorous software process
• Iterative and incremental model
• Based on the Unified Process bidimensional,
  static dynamic dimension
• Use case driven, architecture centric
• Continuous software improvement process
• User Requirements Document
• Updated with regular contacts with users
• Analysis and design
• Design validated against use cases
• Unit, package integration, system tests physics
  validation
• We do a lot but we would like to do more
• Limited by availability of resources for core
  testing
• Rigorous quantitative tests, applying statistical
  methods
• Peer design and code reviews
• We would like to do more main problem
  geographical spread overwork
• Close collaboration with users

5
Use case fluorescence emission
Original motivation from astrophysics requirements
Cosmic rays, jovian electrons
X-Ray Surveys of Asteroids and Moons
Solar X-rays, e, p
Geant3.21
ITS3.0, EGS4
Courtesy SOHO EIT
Geant4
Induced X-ray line emission indicator of target
composition (100 mm surface layer)
C, N, O line emissions included
Wide field of applications beyond astrophysics
Courtesy ESA Space Environment Effects Analysis
Section
6
Design
Used by processes
7
Implementation

• Two steps
• Identification of the atomic shell where a
  vacancy is created by a primary process
  (photoelectric, Compton, ionisation), based on
  the calculation of cross sections at the shell
  level
• Cross section modeling and calculation specific
  to each process
• Generation of the de-excitation chain and its
  products
• Common package, used by all vacancy-creating
  processes
• Also used by Geant4 hadronic package, at the end
  of the nuclear de-excitation chain (e.g.
  radioactive decay)

8
X-ray fluorescence and Auger effect

• Calculation of shell cross sections
• Based on Livermore (EPDL) Library for
  photoelectric effect
• Based on Livermore (EEDL) Library for electron
  ionisation
• Based on Penelope model for Compton scattering
• Detailed atom description and calculation of the
  energy of generated photons/electrons
• Based on Livermore EADL Library
• Production threshold as in all other Geant4
  processes, no photon/electrons generated and
  local energy deposit if the transition predicts a
  particle below threshold

9
Test process
Test Plan Test Guidelines Test Automation
Architecture Test Cases Test Data Test Results

• Unit, integration and system tests
• Verification of direct physics results against
  established references
• Comparison of simulation results to experimental
  data from test beams
• Pure materials
• Complex composite materials
• Quantitative comparison of simulation/experimental
  distributions with rigorous statistical methods
• Parametric and non-parametric analysis

10
Verification X-ray fluorescence
Comparison of monocromatic photon lines generated
by Geant4 Atomic Relaxation w.r.t. reference
tables (NIST)
Transitions (Fe)
Transition Probability Energy
(eV) K L2 1.01391 -1 6349.85 K
L3 1.98621 -1 6362.71 K
M2 1.22111 -2 7015.36 K M3
2.40042 -2 7016.95 L2 M1
4.03768 -3 632.540 L2 M4
1.40199 -3 720.640 L3 M1
3.75953 -3 619.680 L3 M5
1.28521 -3 707.950
11
Verification Auger effect
Auger electron lines from various materials
w.r.t. published experimental results
Precision 0.74 0.07
Cu Auger spectrum
12
Test beam at Bessy - 1
Advanced Concepts and Science Payloads
A. Owens, A. Peacock
Monocromatic photon beam
HpGe detector
13
Comparison with experimental data
Photon energy
Experimental data Simulation
Parametric analysis fit to a gaussian Compare
experimental and simulated distributions Detector
effects! (resolution, efficiency)
difference of photon energies
Precision better than 1
14
Test beam at Bessy - 2
Advanced Concepts and Science Payloads
A. Owens, A. Peacock
Complex geological materials
Hawaiian basalt Icelandic basalt Anorthosite Doler
ite Gabbro Hematite
15
Comparison with experimental data
Pearson correlation analysis rgt0.93
plt0.0001
Effects of detector response function presence
of trace elements
Experimental and simulated X-ray spectra are
statistically compatible at 95 C.L.
16
PIXE

• Calculation of cross sections for shell
  ionisation induced by protons or ions
• Two models available in Geant4
• Theoretical model by Grizsinsky intrinsically
  inadequate
• Data-driven model, based on evaluated data
  library by Paul Sacher (compilation of
  experimental data complemented by calculations
  from EPCSSR model by Brandt Lapicki)
• Generation of X-ray spectrum based on EADL
• Uses the common de-excitation package

17
PIXE Cross section model

• Fit to Paul Sacher data library results of the
  fit are used to predict the value of a cross
  section at a given proton energy
• allow extrapolations to lower/higher E than data
  compilation
• First iteration, Geant4 6.2 (June 2004)
• The best fit is with three parametric functions
  for different groups of elements
• 6 Z 25
• 26 Z 65
• 66 Z 99

• Second iteration, Geant4 7.0 (December 2004)
• Refined grouping of elements and parametric
  functions, to improve the model at low energies

Next protons, L shell ions, K shell
18
Quality of the PIXE model

• How good is the regression model adopted w.r.t.
  the data library?
• Goodness of model verified with analysis of
  residuals and of regression deviation
• Multiple regression index R2
• ANOVA
• Fishers test
• Results (from a set of elements covering the
  periodic table)
• 1st version (Geant4 6.2) average R2 99.8
• 2nd version (Geant4 7.0) average R2 improved to
  99.9 at low energies
• p-value from test on the F statistics lt 0.001 in
  all cases

Test statistics
Fisher distribution
19
Bepi Colombo Mission to Mercury
Study of the elemental composition of Mercury by
means of X-ray fluorescence and PIXE Insight
into the formation of the Solar System
(discrimination among various models)
20
Summary

• Geant4 provides precise models for detailed
  processes at the level of atomic substructure
  (shells)
• X-ray fluorescence, Auger electron emission and
  PIXE are accurately simulated
• Rigorous test process and quantitative
  statistical analysis for software and physics
  validation
• Beware intrinsic precision of physics modeling
  and comparison with test beam results are two
  different aspects
• both must be verified
• Thanks to ESA for the support and collaboration
  to development and physics validation

Dont worry it is not just for space
science (also used at LHC!)