Monte Carlo solutions for several problems in low level
underground gamma-ray spectrometryO. Sima and D. Arnold
Bucharest University and PTB Braunschweig
Bucharest Workshop 25 – 27 April 2007
Overview
1. Introduction - Low level gamma-ray spectrometry
2. Specific problems solved by Monte Carlo simulation
3. Validation
4. Conclusions
Bucharest Workshop 25 – 27 April 2007
1. Introduction
• Broad range of samples (type, volume, matrix) • High efficiency detectors, high efficiency measurement
geometries • Severe requirements on quality control and quality assurance,
on detection limit • Automatic analysis of spectra
Gamma-ray spectrometry today:
Bucharest Workshop 25 – 27 April 2007
Low level and ultra-low level gamma-ray spectrometry:
• purpose: the best detection limit
- very low background
- high efficiency detectors and measurement geometries
• Volume samples with various matrices: => self-attenuation effects=> sample specific efficiency
• High efficiency detectors and measurement geometries: => coincidence-summing effects=> nuclide specific efficiency
Experimental calibration possible in relatively few cases;expensive, problems with sources
Monte Carlo simulation able to cover all cases of interest in gamma ray spectrometryAfter validation, very useful to complement experimental
calibration
Difficulties in efficiency calibration:
Bucharest Workshop 25 – 27 April 2007
2. Specific problems solved by Monte Carlo simulation
Evaluation of efficiency correction factors:
- self-attenuation corrections
- coincidence-summing corrections
- geometry corrections
Direct evaluation of efficiency
Background under the peaks
Spectrum simulation
= > GESPECORBucharest Workshop 25 – 27 April 2007
GESPECOR• GERMANIUM SPECTROSCOPY CORRECTION
FACTORS, authors O. Sima, D. Arnold, C. DovleteRealistic Monte Carlo simulation program
- detailed description of the physics processes (photon interactions: XCOM; electron interactions; bremsstrahlung)
- detailed description of the measurement arrangement- nuclear decay data: NUCLEIDE, ENDSF (250
nuclides in GESPECOR data base)- efficient algorithms – variance reduction techniques- user friendly interfaces- thoroughly tested at PTB (D. Arnold)
Bucharest Workshop 25 – 27 April 2007
Coincidence summing:
- Summing out = Losses from peaks due to simultaneous detection of other photons (γ, X-ray) => apparent efficiency lower than corrected efficiency
- Summing in = additional counts in the peak no. 3 due to simultaneous complete energy absorption of photons 1 and 2 => apparent efficiency higher than corrected efficiency
Magnitude of coincidence-summing effects• Enhanced in high efficiency measurement conditions => important
in low level gamma-ray spectrometry• Depend on the detailed decay scheme of the nuclide (peak energy,
probability of emission of other photons in cascade, energy of the cascading photons)=> nuclide dependent efficiency
• Depend on the geometry (including surrounding materials), matrix⇒Difficult to evaluate: nuclear data + radiation transport
GESPECOR: automatic evaluation of decay data, variance reduction techniques implemented
Bucharest Workshop 25 – 27 April 2007
GESPECOR versus GEANT
-14
-12
-10
-8
-6
-4
-2
0
2
10 100 1000 10000
Energy (keV)
% D
iffer
ence G 1
G 2
G 3
Peak Efficiency
Bucharest Workshop 25 – 27 April 2007Bucharest Workshop 25 – 27 April 2007Bucharest Workshop 25 – 27 April 2007
Differences in photon cross sections in Ge
XCOM versus GEANT 3.21
-2
-1
0
1
2
3
4
5
0.00E+00 5.00E-01 1.00E+00 1.50E+00 2.00E+00 2.50E+00 3.00E+00 3.50E+00
Energy (MeV)
% D
iffer
ence Ge Photoeffect
Ge Compton
Ge Rayleigh
Ge Pair Production
Bucharest Workshop 25 – 27 April 2007
Bucharest Workshop 25 – 27 April 2007
GESPECOR: coincidence-summingSum peaks with X-rays (n-type HPGe)
Ba-133 spectrum - low energy part
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E+06
1.0E+07
0 50 100 150 200 250
Energy (keV)
Cou
nts
1
2
3
4
5
6
7
89
1011
12
13
14
20
1516
1718
19
21 22
24
23 25 26
• Ba-133 point source
• Corrections in good agreement with experimental data (PTB)
Bucharest Workshop 25 – 27 April 2007
GESPECOR: coincidence-summingSum peaks with X-rays (n-type HPGe)
• Ba-133 point source
• Corrections in good agreement with experimental data (PTB)
Ba-133 spectrum - high energy part
1.0E+00
1.0E+01
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E+06
250 300 350 400 450 500Energy (keV)
Coun
ts
27
29
2830 31
32
3
34
35
37
3
3
3940
41
4243
44
45
46
47 48
49
50
52
51
53
54
55
56
5758 59
60
61
62
Bucharest Workshop 25 – 27 April 2007
Are the small coincidence summing peaks present in the spectrum relevant?
Yes, especially in low background measurements:• Corrections factors for the main peaks required for activity
determination, but:• All features of the spectra should be correctly described
(including small peaks and pure sum peaks)- peak interference- nuclide identification problems=> more important in low background measurements!
• Present day tendency: completely automatic analysis of spectra => requires improvement [Arnold, Blaauw, Fazinic, Kolotov, Nucl. Instrum. Meth. A 536 (2005) 196 ]
Bucharest Workshop 25 – 27 April 2007
Efficiency transfer for volume sources with slightly different geometry
• actual geometry slightly different from the reference geometry– identical containers but different filling height
• Transfer factor: – Monte Carlo simulation based on a correlated sampling technique– simultaneous computation of each efficiency
Actual geometry Reference SimulationBucharest Workshop 25 – 27 April 2007
- at low energy, weak dependence of scattered photon energy on angle
e.g. 45 keV: all cases with scattering angle <30° have E>44.5 keV
⇒ Should it be included in the peak efficiency definition? (in programs like GEANT the efficiency is usually defined by the number of counts in a given region of the peak)
⇒ it might be also influenced by the environment of the
measurement, not only by the source, detector and media
in betweenBucharest Workshop 25 – 27 April 2007
Background under the peaks: Small angle scattering in the source and other media
3. Validation
Monte Carlo techniques are well established, but
- are the detector data perfectly known ?
- are the interaction data absolutely correct ?
- is the calculation model perfect ?
⇒Validation required
⇒ Comparison of selected experimental results with calculation,
adjustment of some parameters in case of discrepancy
Bucharest Workshop 25 – 27 April 2007
Comparison between GESPECOR and GEANT 3.21
ICRM Gamma-ray spectrometry Working Group action:
Comparison of Monte Carlo simulation evaluation of the efficiency for 3 geometries:
1. Point source and a bare crystal;
2. Point source and a realistic Ge detector
3. Volume source and a realistic Ge detector
Bucharest Workshop 25 – 27 April 2007
Results
Comparison between our results obtained with GESPECOR and with GEANT:
- in geometry 1 reasonable agreement- in geometry 2 and 3 higher discrepancies, especially at low energies
Agreement of the results for geometry 1 implies that cross sections are in accord?
- no, in geometry 1 at low energy the detector is practically an „infinitely thick“ detector, irrespective to the exact values of the cross sections.- in fact, important differences between XCOM and GEANT cross sections, especially in photoelectric cross section (XCOM – from NIST, Berger and Hubbell)
Bucharest Workshop 25 – 27 April 2007
Differences in photon cross sections in Ge
XCOM versus GEANT 3.21
-2
-1
0
1
2
3
4
5
0.00E+00 5.00E-01 1.00E+00 1.50E+00 2.00E+00 2.50E+00 3.00E+00 3.50E+00
Energy (MeV)
% D
iffer
ence Ge Photoeffect
Ge Compton
Ge Rayleigh
Ge Pair Production
Bucharest Workshop 25 – 27 April 2007
Differences in photon cross sections in Al and H2O
XCOM versus GEANT
0
1
2
3
4
5
6
7
8
9
10
0.00E+00 5.00E-01 1.00E+00 1.50E+00 2.00E+00 2.50E+00 3.00E+00 3.50E+00
Energy (MeV)
% D
iffer
ence
Al Photoeffect
H2O Photoeffect
Bucharest Workshop 25 – 27 April 2007
Effect of the differences in cross sections in the case of geometry 1:- at 45 keV, peak efficiency is changed by 0.08% and totalefficiency is changed by 0.07% if artificial Ge density equal to 6 g/cm^3 (instead of 5.323) is used in GEANT calculations
Effect of differences in cross sections in the case of other geometries:- at 45 keV, if in GEANT the dead layer density is changedfrom 5.323 g/cm^3 to 5.565 and the Al density is changed from 2.7 g/cm^3 to 2.828, the XCOM and GEANT 3.21 cross sections become equal
=> change in peak efficiency –12.6% (G2)=> change in total efficiency –12.2% (G2)
- at 45 keV if in GESPECOR the cross sections are modifed artificially to be equal to the GEANT cross sections, the differences between GESPECOR and GEANT decreaseto 0.5% (G2) and 1.2% (G3) (Ge X-Ray escape excluded)
Bucharest Workshop 25 – 27 April 2007
⇒At low energy photon attenuation in the media between the emission point and the sensitive volume of the detector aremost important
=> dependence on the photon cross sections (especially for photoeffect)
⇒Which cross sections are the best?==========================
Effect of the low energy cut-off:- in GESPECOR 1.9 keV for photons, 10 keV for
electrons- in GEANT 3.21 10 keV for photons, 10 keV for
electronsComputations with GEANT also with 15 and 20 keV low energy
cut-off=> small change in peak efficiency, practically no changein total efficiency Bucharest Workshop 25 – 27 April 2007
Is the 10 keV energy cut-off of no importance?Ge K X-Rays: Kα2 9.85543 (51.49%),
Kα1 9.88653 (100%), Kβ3 10.9781 Kβ1 10.9822
Kβ5” 11.0748 (22.37), Kβ2 11.101 (0.49%)
=> 6.63 times higher probability of X Rays with E < 10 keVthan with E > 10 keV
If in GESPECOR the simulation of Ge X-Rays is prohibitted, the discrepancy in geometry 1 at low energies between GESPECOR and GEANT is much reduced
Bucharest Workshop 25 – 27 April 2007
4. ConclusionsMonte Carlo methods are very useful for assisting in gamma ray spectrometry calibration problemsBest results in the case of the evaluation of efficiency correction factors => robust results, not sensitive to detector data and other imperfections of the model and data
- self-attenuation corrections- coincidence summing corrections- small geometry differences
Direct evaluation of the efficiency => less robust- good detector data are required;- good cross sections ( most important: photoelectric cross sections)- proper values of the simulation parameters (energy cut)
-Validation requiredBucharest Workshop 25 – 27 April 2007
M. C. Simulations of detector efficiencies
GESPECOR and GEANT 3.21-2 sets of results:
- first set: peak efficiency defined as in experimental work(1/10 from the peak height: 96.82 % from Gaussian)
- second set: ideal peak efficiency (the complete Gaussian)
GESPECOR-User friendly MC software for solving problems in gamma ray spectrometry with Ge detectors:
- self-attenuation- coincidence summing- peak and total efficiency
Bucharest Workshop 25 – 27 April 2007
- in measurement: based on a region of interest from the peak,e.g. between 1/10 from the peak height (for a Guassian shape, 96.82% from the ideal peak)
- in simulations peak width is not normally reproduced – it depends on charge production and collection etc. Possible to apply a Gaussian spread of the energy deposited in the sensitivevolume of the detector, assuming that the peak has a Gaussianshape or simply to take 0.9682 from the ideal efficiency.
- possible practical definitions in simulations based on:- interactions in the sensitive volume without any energy lost
outside the sensitive volume- the number of counts in a suitably defined energy bin
- important to have a consistent use of the definition of the peak,especially when coincidence summing effects are present
Peak efficiency definition
Bucharest Workshop 25 – 27 April 2007
Monte Carlo simulation is very useful to complement experimental calibrations- experimental calibration possible in relatively few cases;
expensive, problems with sources - Monte Carlo simulation capable to cover all cases of interest
in gamma ray spectrometry- Problems: requires good knowledge of the experimental setup,
especially detector data, good cross sections and appropriatecorrespondence between the simulated quantities and the quantities applied in experimental data
- Flexibility, user friendliness and short computing times
ICRM Gamma WG Meeting Paris Nov. 2006
GESPECOR physicsPhoton cross sections:
- for each material 100 points between 1.9 keV and 4 MeV; in addition, values before and after each X-Ray absorption edge, on the basis of XCOM- log-log interpolation- very good accord with XCOM (also in the close vicinity of absorption edges)
Electron processes:- multiple scattering (Moliere, third function included), orfaster semiempirical method- bremsstrahlung: fast algorithm, sampling using Walker algorithm- energy loss fluctuations neglected- delta electron production neglected- energy cut: 10 keV Bucharest Workshop 25 – 27 April 2007
GESPECOR – variance reduction- Peak and total efficiency evaluated separately:
-peak efficiency: photon history stopped when energy lost outsidethe sensitive volume of the detector-materials between the point of emission and the sensitivevolume of the detector : attenuating media-emission from source: focalized towards the detector
-total efficiency: photon history stopped at the first interaction inthe sensitive volume of the detector
- Always force the first interaction in the detector- Whenever possible use mean values instead of random sampling
(e.g. use probability of the emission of groups of photons instead of random simulation of decay cascades in coincidence summing computations)
- Use efficient sampling (Walker, interpolation)Bucharest Workshop 25 – 27 April 2007
Computing times: ACER Notebook, Pentium M Centrino 1.4 GHz
GESPECOR GEANTGeometry 1 Peak efficiency: 3 min 1100 minTotal efficiency: 3 min Geometry 2 Peak efficiency: 4 min 2150 min Total efficiency: 15 min Geometry 3 Peak efficiency: 8 min 17000 minTotal efficiency: 60 min
Bucharest Workshop 25 – 27 April 2007