Geant4 electromagnetic physics for medical physics applications

Transcription

1 Geant4 electromagnetic physics for medical physics applications 1 st Geant4 User Workshop University of Wollongong April 2011 Vladimir Ivanchenko, CERN On behalf of Geant4 electromagnetic working groups

2 Outline Introduction Unification of standard and low-energy subpackages Selected topics on recent EM model developments Cross sections Energy loss and ranges Multiple scattering Ranges Upgrade of user interfaces Summary 4/14/2011 Geant4 EM physics, 2

3 Geant4 Electromagnetic Physics Significant permanent development in many aspects of EM physics from 1 st Geant4 release up to now Used for many years for production by large HEP experiments BaBar, SLAC (since 2000) LHC experiments ATLAS, CMS and LHCb (since 2004) Many common requirements for HEP, space, medical and other applications EM web page: 4/14/2011 Geant4 EM physics, 3

4 Located in $G4INSTALL/sources/processes/electromagnetic Geant4 EM packages Standard γ, e up to 100 TeV hadrons up to 100 TeV ions up to 100 TeV Muons up to 1 PeV energy loss propagator X-rays X-ray and optical photon production processes High-energy processes at high energy (E>10GeV) physics for exotic particles Polarisation simulation of polarised beams Optical optical photon interactions Low-energy Livermore library γ, e- from 250 ev up to 1 GeV Livermore library based polarized processes PENELOPE code rewrite, γ, e-, e+ from 250 ev up to 1 GeV (2001 version & 2008 version as beta) hadrons and ions up to 1 GeV atomic de-excitation (fluorescence + Auger) DNA microdosimetry models for radiobiology (Geant4-DNA project) from ev to 10 MeV Adjoint New sub-library for reverse Monte Carlo simulation from the detector of interest back to source of radiation Utils : general EM interfaces 4/14/2011 Geant4 EM physics, 4

5 Unification of standard and lowenergy sub-packages V.Ivanchenko et all., MC2010, accepted for publication in PNST 4/14/2011 Geant4 EM physics, 5

6 Why unification of EM physics? Standard EM developments was concentrated on HEP and in a great part to LHC experiments LHC experiments are successfully taking and analyzing data now For many years EM low-energy sub-package was developed separately Focused on medical and space science requirements The were many recommendation extend Geant4 EM physics using the best features of both packages Previously there were technical limitations to use in one run both standard and low-energy models Migration to common design for the low-energy package was done for Geant4 9.3 (December 2009) 4/14/2011 Geant4 EM physics, 6

7 Main benefits of the unification Possible to combine low-energy and high-energy models Optimal configuration per use-case The best choice precision versus CPU performance Code for each model is simpler Book keeping issued separated and provided Development of new models was facilitated Number of long-stand issues of the low-energy package were fixed by migration CPU performance of the low-energy package was improved User interfaces were improved Easy access to cross sections and stopping powers is provided All EM components can use any new features from Geant4 kernel in easy way 4/14/2011 Geant4 EM physics, 7

8 Selected topics on recent EM model developments 4/14/2011 Geant4 EM physics, 8

9 Spline method for interpolation of stopping powers and ranges improved the accuracy of proton transport in Lead New interpolation automatically available to all EM processes G points per decade EM default G4 9.3 G points per decade EM option3, low-energy G /14/2011 Geant4 EM physics, 9

10 Geant4 EM photon models validation G.A.P. Cirrone et al., NIMA 618 (2010) Systematic validation of cross-sections for electromagnetic photon models of migrated models Standard, Livermore, Penelope models (Geant4 9.3) Photoelectric, Compton, gamma conversion, Rayleigh models EDPL97, SANDIA, NIST data libraries 4/14/2011 Geant4 EM physics, 10

11 Energy loss and ranges 4/14/2011 Geant4 EM physics, 11

12 Proton stopping power comparison: G4 NIST/ G4 ICRU/ SRIM-06/ NIST PSTAR data ±5% 4/14/2011 Geant4 EM physics, 12

13 Alpha stopping power comparison: G4 NIST/ G4 ICRU/ SRIM-06/ NIST PSTAR data NIST and ICRU parameterisations are close SRIM is significantly different below 10 kev 4/14/2011 Geant4 EM physics, 13

14 Electron ranges Geant4 (9.3beta) versus NIST ESTAR data All models are in good agreement with NIST Penelope results are more close to NIST the difference <5% in the energy range 0.1 MeV 1 GeV except for Pb below 0.1 MeV, where EM standard is best 1.03 RCSDA,G4 / RCSDA,NIST Pb E (MeV) 1.03 RCSDA,G4 / RCSDA,NIST Standard Standard Livermore Livermore Penelope Penelope E (MeV) Water RCSDA,G4 / RCSDA,NIST Al E (MeV) 4/14/2011 Geant4 EM physics, 14

15 Proton ranges Geant4 (9.3beta) versus NIST PSTAR data for water RCSDA,G4 / RCSDA,NIST RCSDA,G4 - RCSDA,NIST (mm) I=75 ev I=78 ev E (MeV) E (MeV) Spline interpolation improve stability of ranges for cut, and step limit modification Carbon ion Bragg peak results support change of the mean ionisation potential for water from 75 ev to 78 ev This providing 0.5% difference between Geant4 and NIST for protons in water 0.1 mm shift of the Bragg peak 4/14/2011 Geant4 EM physics, 15

16 G.A.P.Cirrone, G.Cuttone, F.Di Rosa, Z.Quiwei, F.Romano VALIDATION ACTIVITY AT INFN - LNS Depth dose in water for proton and carbon beams for 62 MeV case 4/14/2011 Geant4 EM physics, 16

17 Multiple scattering developments 4/14/2011 Geant4 EM physics, 17

18 Urban multiple scattering models L..Urban, CERN-OPEN The model of multiple scattering of L.Urban was the only model since Geant4 4.0 Provided accuracy to meet the requirements for LHC and other HEP applications Flexible step limitation algorithm Applicable for tracking in field Utilizes separate parameterisations for the central part and the tail of scattering angle distributions Keeps two first Lewis momenta Balanced CPU cost and HEP accuracy needs 4/14/2011 Geant4 EM physics, 18

19 Backscattering simulation with L.Urban model (Geant4 9.3) Electron Energy and Charge Albedos SANDIA Report SAND (1984) Electron energy MeV 4/14/2011 Geant4 EM physics, 19

20 Alternative models J. Phys: Conf. Ser. 219 (2010) There are natural limitations of the accuracy of the L.Urban model Parameterisations used available data on electrons scattering, do not cover all energies and media Optimisation of CPU performance for HEP applications has meant also some simplification Geant4 design allowing to have alternative models Specialisation per particle type and use-case Recently number of new models become available: Goudsmit-Saunderson fully theory based (e ± ) Single Coulomb scattering model and WentzelVI multiple scattering (µ ±, hadrons) 4/14/2011 Geant4 EM physics, 20

21 Goudsmit-Saunderson model O.Kadri et al., NIM B267 (2009) 3624 Angular distribution from Goudsmit-Saunderson theory Thanks to F.Salvat who provided ELSEPA code Lewis moments and displacement sampling according to EGSnrc prescriptions Thanks to I.Kawrakov Step limitation and path length corrections from L.Urban model The goal of the model is achieve maximum precision for electron transport Today it is still significantly slower than the L.Urban model (about factor 2) 4/14/2011 Geant4 EM physics, 21

22 New WentzelVI model J. Phys: Conf. Ser. 219 (2010) Is much simpler, but fully theory based Wentzel differential cross section with mass, spin and form-factor corrections Separate, original step limitation Angular limit between the single and multiple scattering is selected dynamically, depending on momentum and step size May be applied for transportation in vacuum or low-density media Can be used together with the hadron elastic scattering process 4/14/2011 Geant4 EM physics, 22

23 Recent results of Fano cavity test for water (Geant4 9.4beta) gamma beam 1 MeV Default value of step limit parameter Urban model Alternative models Dependence of ionisation dose inside the cavity demonstrates precision of MeV electron transport S.Elles et al., J. Phys: Conf. Ser. 102 (2008) /14/2011 Geant4 EM physics, 23

24 Recent improvements of ion transport in Geant4 (Anton Lechner thesis) 4/14/2011 Geant4 EM physics, 24

25 ICRU 73 stopping powers in new Geant4 model The ICRU 73 report: Presents tabulated stopping powers of Li-Ar, Fe ions for many elemental media and compounds. Energy range: up to 1 GeV/nucleon. Stopping power calculations based on PASS code and binary theory P. Sigmund and A. Schinner, Nucl. Instr. and Meth. B 195 (2002) 64 P. Sigmund and A. Schinner, Eur. Phys. J. D 12 (2000) 425 ICRU 73 (PASS) stopping powers in Geant4: Since release 9.3, all ICRU 73 tables are available for ion transport simulations in Geant4 (tables incorporated in Low-Energy Electromagnetic Package). For a few materials revised tables were included (tables provided by P. Sigmund), which replace original ICRU 73 tables. P. Sigmund, A. Schinner and H. Paul, Errata and Addenda: ICRU Report 73, 2009 In addition PASS tables for extended set of projectile and target have been included in G4 9.4, which are not included in ICRU 73 report, were incorporated in Gean4 (tables provided by P. Sigmund and A. Schinner). 4/14/2011 Geant4 EM physics, 25

26 Validation of ICRU 73-based ion model in Geant4 A. Lechner et al., NIM B268 (2010) 2343 ICRU73 based model for ion ionisation is available since Geant4 9.3 Validation of 12 C Bragg peak simulation in water and polyethylene ( MeV/amu) for G4ionIonisation process Experimental Bragg peak position can be reproduced within 0.2% of ion range in case of water, and within 0.9% in case of polyethylene. Water Polyethylene See paper for details. Experimental data by courtesy of D. Schardt and U. Weber 4/14/2011 Geant4 EM physics, 26

27 Range stability in Geant4 9.3 Stability of 12 C Bragg peak position with respect to simulation parameters (production cut, step size limiter) was investigated. Ionization step function: same parameters used as in LowE builders (droverrange=0.1, finalrange=100 μm) Figure (a) shows the difference between simulated and measured Bragg peak position of a 400 MeV/u 12 C beam in water as a function of the δ-production cut. Figure (b) presents the average CPU time per event of corresponding simulation runs. Geant4 results were obtained using the low-energy ICRU 73-based parametrization model. For simplicity, no hadronic processes were activated. Simulations without user-defined step size limitation (diamonds) are compared against calculations with a maximum step size of 50 μm (circles). Experimental data by courtesy of D. Schardt/GSI (personal communication). The dashed lines indicate the experimental uncertainty. Error bars account for the uncertainty of simulated peak positions. 4/14/2011 Geant4 EM physics, 27

28 Validation of 12 C CSDA ranges in elemental media and compounds Range validation: Comparison of Geant4 (9.3) ranges, computed with ICRU 73-based model, against experimental data from the literature (Bichsel et al., Radiat. Res. 153 (2000), 208). Bichsel et al. report residual CSDA range of 275 MeV/u 12 C ions in water after traversing a solid absorber of thickness t. Beam energy in simulation was calibrated (275.5 MeV/u) aiming to reproduce the experimental range in water for the case t=0. 4/14/2011 Geant4 EM physics, 28

29 Validation of 12 C CSDA ranges in elemental media and compounds Results: Calculation of absorber thickness (t sim ) required in the simulation to achieve the same residual CSDA range in water as in the experiment. Figures: Percentage difference of t sim and t in elemental (left) and compound (right) targets. Data H.Bichsel et al., Radiat. Res. 153 (2000) 208 4/14/2011 Geant4 EM physics, 29

30 Upgrade of user interfaces 4/14/2011 Geant4 EM physics, 30

31 EM Physics List constructors for HEP Main user interface version for g4 9.3 described below Used by Geant4 validation suites Are robust due to intensive tests by Geant4 team well known precision and limitations Providing several alternatives focused to different application domain Constructor Components Comments G4EmStandardPhysics G4EmStandardPhysics_option1 G4EmStandardPhysics_option2 Default (QGSP_BERT, FTFP_BERT ) Fast due to simple step limitation, cuts used by photon processes (QGSP_BERT_EMV, ) Experimental:WentzelVI model of multiple scattering (QBBC, ) ATLAS, LHCb and other HEP productions, other applications CMS production, good for crystals not good for sampling EM calorimeters Used for testing of new models 4/14/2011 Geant4 EM physics, 31

32 Combined EM Physics List constructors Are available after migration to common EM design (g4 9.3) For today focus more to precision than to maximum simulation speed Ion stopping model based on the ICRU 73 data Step limitation for multiple scattering using distance to boundary Strong step limitation by the ionisation process defined per particle type Recommended for hadron/ion therapy, space applications Constructor Components Comments G4EmStandardPhysics_option3 Urban MSC model (QGSP_BIC_EMY, Shielding) Proton/ion therapy G4EmLivermorePhysics GodsmitSaunderson MSC model Livermore models for γ, e - below 1 GeV, Standard models above 1 GeV Livermore low-energy electron and gamma transport G4EmPenelopePhysics GodsmitSaunderson MSC model Livermore models for γ, e ± below 1 GeV, Standard models above 1 GeV Penelope low-energy e ± and gamma transport 4/14/2011 Geant4 EM physics, 32

33 User interfaces and helper classes G4EmCalculator easy access to cross sections and stopping powers (TestEm0) G4EmProcessOptions c++ interface to EM options alternative to UI commands G4EmSaturation Birks effect G4ElectronIonPair sampling of ionisation clusters in gaseous or silicon detectors G4EmConfigurator add models per energy range and geometry region 4/14/2011 Geant4 EM physics, 33

34 Specialized models per G4Region: example of DNA physics Standard EM physics constructor as a base G4EmConfigurator is used to add DNA models DNA models are enabled only in the small G4Region for energy below 10 MeV CPU performance optimisation 4/14/2011 Geant4 EM physics, 34

35 Summary Unification of Geant4 EM physics was achieved for the version 9.3 Physics List constructors combining standard and low-energy models are available since Geant4 9.3 DNA processes are working together with standard Alternative scattering models have been developed to control Urban model, which is currently used in many Monte Carlo productions (LHC, others) Improved accuracy is achived forcing more steps New model for ion ionisation based on ICRU73 report is available Accuracy of 0.1 mm is achieved for range of carbon ions 4/14/2011 Geant4 EM physics, 35

36 Thank you for your attention! 4/14/2011 Geant4 EM physics, 36

37 Complete Author List Vladimir IVANCHENKO, John APOSTOLAKIS, Alexander BAGULYA, Haifa Ben ABDELOUAHED, Rachel BLACK, Aleksey BOGDANOV, Helmut BURKHARD, Stéphane CHAUVIE, Pablo CIRRONE, Giacomo CUTTONE, Gerardo DEPAOLA, Francesco Di ROSA, Sabine ELLES, Ziad FRANCIS, Vladimir GRICHINE, Peter GUMPLINGER, Paul GUEYE, Sebastien INCERTI, Anton IVANCHENKO, Jean JACQUEMIER, Anton LECHNER, Francesco LONGO, Omrane KADRI, Nicolas KARAKATSANIS, Mathieu KARAMITROS, Rostislav KOKOULIN, Hisaya KURASHIGE, Michel MAIRE, Alfonso MANTERO, Barbara MASCIALINO, Jakub MOSCICKI, Luciano PANDOLA, Joseph PERL, Ivan PETROVIC, Aleksandra RISTIC-FIRA, Francesco ROMANO, Giorgio RUSSO, Giovanni SANTIN, Andreas SCHAELICKE, Toshiyuki TOSHITO, Hoang TRAN, Laszlo URBAN, Tomohiro YAMASHITA, Christina ZACHARATOU 4/13/2011 Geant4 EM physics, 37