secondary electron energy deposition in thin polymeric films for neutron-photon discrimination
TRANSCRIPT
IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 3, JUNE 2014 1381
Secondary Electron Energy Deposition in ThinPolymeric Films for Neutron-Photon DiscriminationLaurence F. Miller, Matthew J. Urffer, Andrew Mabe, Rohit Uppal, Dayakar Penumadu, and George Schweitzer
Abstract—Thin polymeric films are evaluated in this researchas a potential replacement technology for radiation portal moni-tors where specific attention is given to the physical basis for neu-tron-photon discrimination. It is shown that the difference in theenergy deposition mechanics from charged particle reaction prod-ucts and from the Compton scattered electrons allows for the effec-tive discrimination between neutrons and gammas. One goal of thisstudy was to establish optimal thickness for polymeric films thatmaximize the neutron interactions and simultaneously minimizethe measured interaction of photons. Polymeric films ranging from
m to m containing were fabricated and tested fortheir capability to satisfy criteria established by the Department ofHomeland Security. Results from Monte Carlo simulations withthe GEANT4 code and data from measurements confirm the tech-nical basis for our proposed understanding of neutron-photon dis-crimination characteristics for thin films.
Index Terms—Detectors, energy deposition, GEANT4, MonteCarlo simulations.
I. INTRODUCTION
T HE Department of Homeland Security (DHS) in theUnited States continues funding research through the
Domestic Nuclear Detection Office (DNDO) to develop neu-tron detector technologies to support DHS objectives with aparticular focus on Radiation Portal Monitors (RPMs). TheseRPMs are expected to detect special nuclear material that emitneutrons while being insensitive to the gammas from medicalisotopes and other commercial and background sources. Mostof the RPMs currently deployed use a rapidly diminishingresource, , which should be replaced with more readilyavailable materials. As a result a number of alternative detec-tion systems continue to be investigated with the most viableincluding: boron trifluoride ( ) filled proportional detec-tors, boron-lined proportional detectors, lithium-6 ( ) loadedscintillation glass fiber detectors, and scintillator-coated
Manuscript received July 18, 2013; revised October 07, 2013; acceptedNovember 08, 2013. Date of publication May 19, 2014; date of current versionJune 12, 2014. This work was supported by the Domestic Nuclear DetectionOffice (DNDO) through award 003387891. Any opinions, findings, andconclusions or recommendations expressed in this material are those of theauthors and do not necessarily reflect the views of DNDO.L. F. Miller andM. J. Urffer are with the Department of Nuclear Engineering,
University of Tennessee, Knoxville, TN 37916 USA (email: [email protected],[email protected]).A.Mabe and G. Schweitzer are with the Department of Chemistry, University
of Tennessee, Knoxville, TN 37916 USA (email: [email protected]).R. Uppal and D. Penumadu are with the Department of Civil Engineering,
University of Tennessee, Knoxville, TN 37916 USA.Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TNS.2014.2312822
TABLE IREPLACEMENT CRITERIA RADIATION PORTAL MONITORS
Cost assumed to be a maximum
wavelength-shifting fiber detectors [1][2]. This research inves-tigates the application of -loaded thin polymeric films forneutron-photon discrimination. was chosen for the neutronabsorber because of its high thermal neutron absorption crosssection (940 barn) and a large Q-value (4.78 MeV) on theabsorption of a neutron.General specifications that replacement RPMs must satisfy
are listed in Table I [2][3]. In particular these include: 1) an ab-solute neutron detection efficiency greater than 2.5 counts persecond (cps) per ng for a lead-shielded and moderatedsource placed 2 m from the detector, 2) an intrinsic gamma-neu-tron detection efficiency of one in a million, and 3) a gammaabsolute rejection ratio for neutrons requiring that the perfor-mance of the detector should not change by more than 10% in a10 mR/hr gamma field. Additional details are published in Pa-cific Northwest National Laboratory (PNNL) reports [2][3].The absolute neutron detection efficiency, , is defined as
the number of neutron pulses recorded divided by the numberof neutrons emitted by the source (with only a neutron sourcepresent) as shown in Eq. (1).
(1)
where is the neutron count rate, and is the neutronemission rate from the source.The absolute neutron detection efficiency is the probability
of a neutron count per source neutron. In the case of radiationportal monitors the neutron source strength, , is defined tobe 1 ng of , which has a source strength of neu-trons per second, and the source is two meters from the detector.
0018-9499 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1382 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 3, JUNE 2014
The intrinsic gamma-neutron detection efficiency, , isdefined as the count rate of pulses that could be interpreted asoriginating from neutrons divided by the rate of photons en-tering the detector, as shown in Eq. (2).
(2)
where is the number counts classified as photons that areindistinguishable from neutrons per unit time, and is thenumber of photons incident on the detector per unit time.The intrinsic gamma-neutron detection efficiency may be
measured using either a , , or source placedat an appropriate distance so as to produce an exposure rateof 10 mR/hr at the detector [3]. The gamma absolute rejectionratio for neutrons (GARRn) is a parameter that characterizesthe detector response in the presence of both a gamma raysource of 10 mR/hr and of a neutron source configuredas it would be for an absolute neutron detection efficiencymeasurement. The GARRn is defined as the absolute neutrondetection efficiency in the presence of both sources, divided bythe absolute neutron detection efficiency [3].It is problematic for some of the technologies evaluated to
simultaneously satisfy the intrinsic gamma efficiency and theabsolute neutron detection efficiency [1]. It is shown in thispaper that this problem can be mitigated for thin films that con-tain a neutron absorber by taking advantage of differences inthe kinetic energy deposition characteristics of secondary elec-trons produced by photon interactions and by charged particlesslowing down.
II. THEORETICAL BASIS
In the case of reaction products, energy is dis-tributed between a triton of energy 2.73 MeV and an alpha ofenergy 2.05 MeV, and their energy loss produces many elec-trons with a median energy of about 0.8 keV. For photons withan energy range between 0.5 MeV to 5 MeV Compton scat-tering is the predominant interaction mechanism for low-Z ma-terials, and this interaction produces a single electron with anaverage energy much higher than those produced by chargedparticles with energies below 10 MeV slowing down. For ex-ample, the 1.173 MeV and 1.332 MeV photons from pro-duce Compton-scattered electrons with maximum energies of0.96 MeV and 1.12 MeV, and the median energies of theseCompton-scattered electrons are 860 keV and 1.07 MeV. Ifelastic scattering between the alpha, triton and electrons is as-sumed, the maximum kinetic energy of a scattered electron is1.097 keV for the alpha particle and 1.986 keV for the triton [4].The study of secondary electron ranges is of particular interestsince the energy deposition from electrons directly impacts thescintillation process.The ranges of the secondary electrons from charged particle
slowing down and from Compton-scattered electrons are listedin Table II. Since the relative ranges of the secondary electronscreated from charged particle slowing down differ from thosegenerated by Compton scattering by several orders of magni-tude, electrons generated by the alpha and the triton deposit sig-nificantly more of their energy in a thin film than the secondaryelectrons produced from Compton-scattered gammas. This is
TABLE IISECONDARY ELECTRON RANGE IN POLYSTYRENE [4]–[6]
The median values reported for the are averaged over both possiblegammas.
Fig. 1. GEANT4 simulation of cumulative energy deposition of electrons inpolystyrene.
also reflected in the stopping power, where the reaction productsecondary electrons have a stopping power of about 40 timesthat of the secondary electrons from a gamma [5]. The cumu-lative energy deposition as a function of depth in the materialfor electrons from 3 keV to 320 keV in polystyrene is shown inFig. 1. For electrons below 10 keV all of the energy is depositedwithin 5 microns of the material, but more energetic electronsdeposit their energy over a broader range.The difference in the transfer of kinetic energy from charged
particles to electrons and from photon interactions to electronsmay be exploited to maximize the discrimination betweenneutron and photon interactions in a detector. For a particularmaterial and neutron absorber, the detector geometry can beoptimized to maximize the energy deposited in scintillationmaterials by charged particles relative to the energy depositedby photon interactions. This in turn permits one to maximizethe recorded neutron interaction rate relative to the recordedphoton interaction rate by setting a lower level discriminator(LLD) to a particular pulse height of the signals generatedby the detector electronics. The LLD does not need to be aphysical voltage cutoff set by the instrumentation. Insteadpulse amplitudes below a particular level may be neglectedby integrating the pulse height spectrum above an arbitrarilyspecified discriminator, which is defined as a mathematicallower level discriminator (MLLD).
MILLER et al.: SECONDARY ELECTRON ENERGY DEPOSITION IN THIN POLYMERIC FILMS 1383
Fig. 2. Pulse height measurement electronics.
Fig. 3. MCNPX Rendering of Neutron Irradiator (x-z view and x-y view). Thesmall cylinder shown inside is the source, and the two detector wells are shownas the cylinders in the top view on the right.
III. EXPERIMENTAL METHODS
The energy deposition of neutrons and gammas was investi-gated for polymeric films containing by obtaining neutronand gamma-ray spectra for polystyrene film thickness rangingfrom m to m. The reader is referred to the work ofMabe et al. for the sample preparation protocols [7]. Fig. 2 pro-vides a diagram of the equipment used to measure the pulseheight spectra from amoderated neutron source and from asource. Measured data are compared with calculations to con-firm modeling and to establish rates of particles crossing thedetector.Fig. 3 is a rendering (from a Monte Carlo transport code,
MCNPX [10]) of the neutron irradiator used as the neutronsource in this work. It is a custom built facility that contains a
g (as of July 2009) source encased in high densitypolyethylene (HDPE) for moderating the spectrum. TheHDPE box is approximately 50.8 cm long, 30.48 cm wide, and35.56 cm tall. There are two 0.16 cm thick acrylic detectorswells, one surrounded by a 0.16 cm cadmium to shield outthermal neutrons, and the other surrounded by 0.16 cm of leadto shield out a similar amount of gammas as the cadmium well.The source is surrounded by stainless steel, which in turnis contained within a 5.08 cm diameter, 1.27 cm thick, 12.7 cmtall lead vessel.The lead well is used to measure the responses of all neutron
energies present, while the cadmium well is used to measurethe response of neutrons above the cadmium cutoff. Thetwo responses may be subtracted to yield a good estimateof the thermal neutron response. An implemented radiation
Fig. 4. MCNPX Rendering of Gamma Irradiator.
portal monitor would most likely use entire neutron spectrumfor counting, rather than the part of the spectrum below thecadmium cutoff (0.6 eV). MCNPX modeling of the neutronirradiator obtains an interaction rate for a 2.54 cm diameter2 mm thick glass (GS20) of 423 interactions per secondas of December 1st, 2012. The observed count rate of GS20measured on December 1st, 2012 was 428 cps. Polymeric filmshad a simulated neutron interaction rate within 15% of the ob-served count rate; however, the polymeric films are not as wellcharacterized as the GS20, since the composition is determinedbefore casting the films. Therefore it is assumed that simula-tions with MCNPX can be used obtain the interaction rate for adetector mocked up according the PNNL specifications encasedin an existing RPM cabinet ( cm cm cm).MCNPX was used to obtain the number of particles incident onthe detector for the determination of the intrinsic efficiency.An MCNPX rendering of the gamma irradiator fabricated for
our research is shown in Fig. 4. It contains Ci sourceencased in 5.08 cm of steel, with a 0.32 cm thick steel cap. Thedetector well is a 10.16 cm outer diameter 35.56 cm pipe thatis 0.63 cm thick, and 7 cm of foam is used to displace the de-tector to obtain a 10mR/hr field. The detector well is surroundedby two levels of cm cm cm lead bricksand one level of HDPE blocks, which are contained in an outersteel outer box cm cm cm. The mea-sured gamma intrinsic efficiency of the sample is calculated byintegrating the measured spectra, , as a function of a math-ematical lower level channel discriminator MLLD setting anddividing by the incident photon fluence, , as shown in Eq. (3).
(3)
1384 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 3, JUNE 2014
Fig. 5. GEANT4 Rendering of the Simulation for a Thin Film.
The number of incident photons was calculated with MCNPXand was confirmed by measurements in the gamma irradiator.At a distance of 10.2 cm from the source the dose rate wasmeasured to be 10 mrem/hr and the calculated dose rate was10.3 mrem/hr.
IV. SIMULATIONS FOR ELECTRON TRANSPORT AND DATAANALYSIS
MCNPX is used for neutral particle transport since it is auser friendly Monte Carlo code widely used in the nuclearindustry. However, MCNPX does not track individual sec-ondary charged particles generated by charged particle slowingdown. Fig. 5 illustrates a rendering of the thin film detectorgeometry used for simulating secondary electron transport withGEANT4 [8], and Fig. 5 is an image of one of the polymericfilms, which is mounted on a non-scintillating acrylic disc.Initial events for the GEANT4 (v9.6 p01) simulation weregenerated by using the general particle source for pho-tons and a particle gun for thermal (0.025 eV) neutrons. Ahigh performance, data-driven hadronic modular physics list(HadronPhysicsQGSP_BERT_HP) was used to simulate theneutron interactions. The subsequent charged reaction prod-ucts (as well as photon interactions and its secondaries) weresimulated with the detailed G4EMStandardPhysics (option3). A range cut was implemented at 10 nm. The calculationframework was verified by simulating the single collisionenergy loss in water; for this simulation G4DNAPhysics wereemployed along with G4Water. Note that the values presentedin Table II are based on calculations preformed by [4]–[6], andare not simulated values.The simulation physics were first tested by comparing the
single collision energy loss spectra for water with previously
Fig. 6. Image of a loaded polystyrene thin film mounted on an acrylicdisc.
Fig. 7. Simulated Single Collision Energy Loss ofWater. The simulated energyspectrum matches that of Turner (shown in the insert) [11].
published measured spectra [11]. Results of simulations forthe single collision energy loss spectra for water are shown inFig. 7. The simulated spectra essentially overlap the referencemeasurement.The validity of the GEANT4 simulation for the thin films
is evaluated by comparing the spectral shapes of measuredgamma spectra to their corresponding simulated energy deposi-tions. Fig. 8 illustrates the GEANT4 simulated gamma intrinsicefficiencies for several thin film thicknesses, and Fig. 9 showsthe measured gamma intrinsic efficiencies.
V. RESULTS AND ANALYSIS
Simulated results shown in Fig. 8 are based on gammarays directly incident on the thin film, and the measured spec-trum includes direct and scattered photons from . For thesimulated case shown in Fig. 8, only 1.17 and 1.33MeV photonsare incident on the detector in a single direction, whereas in theexperiment setup photons scattered off the irradiator assemblymay interact with the film so some discrepancies between the
MILLER et al.: SECONDARY ELECTRON ENERGY DEPOSITION IN THIN POLYMERIC FILMS 1385
Fig. 8. GEANT4 Simulation of Energy Deposition by Gamma in Poly-styrene Thin Films. The two Compton edges and the photo peak from areobserved clearly in the 1 cm film simulation. These spectra are not resolutionbroadened.
Fig. 9. Measured Energy Deposition by Gamma in Polystyrene ThinFilms.
measured and simulated spectra should be expected. The en-ergy scale of Fig. 9 based on the Compton edge of a 1 cm thicksample, and the energy calibration was then applied to all of theother measured spectra.Fig. 10 illustrates the simulated and measured energy deposi-
tion by photons as a function of film thickness. The continuousincrease in the average energy deposition and subsequent spec-tral average light yield (SALY – the average of the distributionof light yield of the measured pulse height spectra comparedto a reference) is due to the range of the photons and theCompton scattered secondary electrons being several orders ofmagnitude greater than the thickness of the film. For the caseof charged particles, as shown in Fig. 11, the average energyabsorbed levels off as the film thickness becomes greater thanabout 150 microns.
Fig. 10. Gamma ( ) average energy deposition and light yield. The simu-lated average energy deposition are shown as square dots, while the measuredspectra average light yield is shown as circular dots. The light yield is based onthe Compton edge of GS20 as an energy calibration feature with a light yield of3,800 Photons per MeVee.
Fig. 11. Neutron average energy deposition and light yield. The simulated av-erage energy deposition are shown as square dots, while the measured spectraaverage light yield is shown as circular dots. The light yield of the films are es-timated based on GS20 emitting 6,200 photons per neutron.
Figs. 12(a)–(d) illustrates the effect of the MLLD on severaldifferent detector materials. As the MLLD is increased, the ef-ficiency for detecting neutrons is diminished; however, the in-trinsic efficiency for detecting neutrons relative to photons isdramatically increased. Fig. 12(a) illustrates measured intrinsicgamma efficiency as a function of MLLD by the dotted lines forseveral sample thicknesses with the corresponding scale on theleft. Measured net neutron count rate (count rate in the lead-cov-ered tube minus the count rate in the cadmium-covered tube) isshown as a function of channel number with its scale on theright. Note that the response for the 50 micron film is plotted asa solid line and as a dashed line for the 150 micron film. Thesemeasurements are obtained from the gamma and neutron irradi-ators previously described. Since DHS criteria require that theintrinsic efficiency be less than , only the fraction of the
1386 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 3, JUNE 2014
Fig. 12. Gamma intrinsic efficiency (dashed lines on the left axis) plotted against neutron count rates (solid axis on the right). The neutron performance of asample above a given intrinsic efficiency can be found by noting at what MLLD the gamma intrinsic efficiency falls below the desired value and then reading offthe neutron count rate on the right axis. Some data were obtained at different measurement settings since the light output of LiF:ZnS(Ag) is over a factor of 10more than GS20. (a) 10% LiF Polystyrene Based Film (b) 25% LiF Composite Polyethylene Naphthalene Film (c) LiF:ZnS (EJ-426HD2-PE) (d) GS20.
neutron spectrum above the corresponding channel number maybe recorded as neutron counts. Counts in lower channel numbersmust be discarded. Note also that layering thin films can satisfythe absolute count rate criterion and that the fraction of countsin the neutron spectrum above is strongly thickness dependent.These results are for polystyrene (PS) based film loaded with10% LiF by weight. Other materials exhibit different spectralshapes; thus, materials, clarity of the film, and thickness all im-pact the efficacy of the films for meeting DHS criteria.Fig. 12(b) illustrates intrinsic efficiency and neutron count
rate for a composite polyethylene naphthalene (PEN) film.Fig. 12(c) provides measurement data for a 100 micronLiF:ZnS (EJ-416HD2-PE) film, and Fig. 12(d) shows resultsfor a 2 mm thick GS20 sample. Amplifier gain adjustmentswere made to satisfy the dynamic range limitations of the dataacquisition instrumentation. The necessary mathematical lowerlevel discriminator settings necessary to achieve a given levelof gamma intrinsic efficiency for PS films are shown in Fig. 13for the various film thicknesses.Table III shows the fraction of neutron count rate that is above
the MLLD necessary for . Polystyrene films from15 to 50 microns have over 15% of the counts above the
Fig. 13. Gamma intrinsic efficiency versus the pulse height discriminator set-ting (MLLD) to achieve the corresponding level of discrimination. It is observedthat there is a change in the shape of the curves reflecting the increased fractionof energy deposition in thicker films relative to the thinner films.
MILLER et al.: SECONDARY ELECTRON ENERGY DEPOSITION IN THIN POLYMERIC FILMS 1387
TABLE IIINEUTRON PERFORMANCE ABOVE DISCRIMINATOR SETTING
TABLE IVFRACTIONAL ENERGY DEPOSITION FOR VARIOUS THICKNESSES
Fig. 14. Simulated Kinetic Energy of Secondary Electrons from ComptonScattering and from reaction products.
gamma intrinsic efficiency discriminator setting, while thickerfilms have a much lower fraction above the MLLD.In addition it is observed in Fig. 11 that thicker films enhance
the neutron count rate and resolution of the film but do little toincrease the light yield, as most of energy from a neutron eventis captured in the film. The average energy deposited was com-puted for each thickness and normalized by the incident energy
Fig. 15. Alpha and Triton Secondary Electron Kinetic Energy Distribution.
Fig. 16. Number of Secondary Electrons Produced Per Neutron Interaction.
for gammas by the Q-value of the reaction for neutrons, and ispresented in Table IV. For thickness greater than m thereis little benefit in increasing the thickness of the film in terms ofenergy deposition by neutrons, since over 90% of the energy isbeing deposited in the film.Figs. 14 and 15 illustrate the simulated kinetic energy of
secondary electrons from Compton scattering and from alphaand triton interactions. It is observed that kinetic energy ofthe secondary electrons from the neutron reaction productshave predominately energies in the kilovolt range, while theCompton scattering electrons have energies about 100 timeshigher. However, it should be noted that there is only one sec-ondary electron from a Compton scattering event and multiplesecondary electrons from the reaction products. This differencein the energy distribution in the resulting secondary electronsbetween charged particle and photon interactions is the basisfor the enhanced discrimination for thin films relative to thickfilms. Fig. 16 shows the distribution in the average number ofsecondary electrons produced by the alpha and triton.
1388 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 3, JUNE 2014
VI. CONCLUSIONS
Analysis of data from measurements and results from cal-culations confirm that the difference in energy deposition be-tween gamma and neutron events in a thin film allows foreffective pulse height discrimination between neutrons andgammas. GEANT4 Monte Carlo simulations of the energy de-position show that for a m film over 90% of the reactionproduct energies from a are deposited in the film,but only 3.2% of the gamma energies from a decayis deposited. The optimal thickness films (for minimizing in-trinsic efficiency of photons while maximizing neutron countrate) may be between 25 to 50 microns. However, thinnerfilms will require more layers to satisfy the DHS count ratecriterion, but this may increase the cost of the detector system.Future work will focus on the optimization of a layered de-tector for candidate RPM designs in order to increase theneutron efficiency.
REFERENCES
[1] R. Van Ginhoven, R. Kouzes, and D. Stephens, “Alternative neutrondetector technologies for homeland security,” PNNL, Richland, WA,USA, Rep. 18471, 1999.
[2] R. T. Kouzes, J. H. Ely, L. E. Erikson, W. J. Kernan, A. T. Lintereur,E. R. Siciliano, D. L. Stephens, D. C. Stromswold, R. M. Van Gin-hoven, and M. L. Woodring, “Neutron detection alternatives to 3Hefor national security applications,” Nucl. Instrum. Methods Phys. Res.A, vol. 623, pp. 1035–1045, Nov. 2010.
[3] R. Kouzes, J. Ely, A. Lintereur, and D. Stephens, “Neutron detectorgamma insensitivity criteria,” PNNL, Richland, WA, USA, Rep.18903, 1999.
[4] J. E. Turner, “Interaction of electronswithmatter,” inAtoms, Radiation,and Radiation Protection. Hoboken, NJ, USA: Wiley, 2008.
[5] J. C. Ashley, C. J. Tung, and R. H. Ritchie, “Inelastic interaction ofelectrons with polystyrene: Calculations of mean free paths, stoppingpowers, and CSDA ranges,” IEEE Trans. Nucl. Sci., vol. NS-25, no. 6,pp. 1556–1570, Dec. 1978.
[6] M. Berger, J. Coursey, M. Zucker, and J. Chang, ESTAR, PSTAR andASTAR: computer programs for calculating stopping-power and rangetables for electrons, protons, and helium ions [Online]. Available:http://www.nist.gov/pml/data/star/index.cfm
[7] A. N. Mabe, J. D. Auxier, II, and M. J. Urffer et al., “Thin film polymercomposite scintillators for thermal neutron detection,” J. Composites,vol. 2013 [Online]. Available: http://dx.doi.org/10.1155/2013/539060
[8] S. Agostinelli, J. Allison, and K. Amako et al., “Geant4 - a simulationtoolkit,” Nucl. Instrum. Methods Phys. Res. A, vol. 506, pp. 250–303,Jul. 2003.
[9] J. Allison, K. Amako, and J. Apostolakis et al., “Geant4 developmentsand applications,” IEEE Trans. Sci. Nucl. Sci, vol. 53, no. 1, pp.270–278, Feb. 2006.
[10] D. Pelowitz, “MCNPX user’s manual 2.6.0,” Los Alamos NationalLaboratory, Los Alamos NM, USA, Rep. LA-CP-07-1473.
[11] J. E. Turner, H. G. Paretzke, R. N. Hamm, H. A. Wright, and R. H.Ritchie, “Comparative study of electron energy deposition and yields inwater in the liquid and vapor phases,” Radiat. Res., vol. 92, pp. 47–60,Oct. 1982.