Scintillators for the detection X-rays, gamma rays, and thermal neutrons Dr. P. Dorenbos Section Radiation Detection & Matter Scintillators are materials that convert the energy of ionizing radiation into a flash of light. Scintillation material can be gaseous, liquid, glass-like, organic (plastics), or inorganic. In each case the material should be transparent to its own scintillation light. Inorganic wide band gap ionic crystals are the most widely used scintillators for detection of X-rays, gamma rays, and thermal neutrons. Gamma rays interact with a scintillator by means of ) photoelectric interaction (dominant below 500 kev and interaction probability is proportional to Z eff - where Z eff is the effective atomic number of the atoms in the compound), ) Compton scattering (dominant around MeV and interaction probability proportional to the density), or ) pair creation (dominant well above the threshold at.0 MeV). For efficient detection of gamma rays of energies 00 kev to 0 MeV, a scintillator should contain high atomic number elements (e.g. Ba, La, I, Lu, Cs, Pb, Bi) and posses a high density. Depending on application crystals of sizes that may range from a few cm up to dm may be required. For the detection of thermal neutrons, thermal neutrons need to be captured by isotopes with high capture cross section. Most popular is Li with % natural abundance n + Li 7 Li H + α +.79 MeV () The reaction energy of.79 MeV is shared between the triton and alpha particle, and both particle create an ionization track. The history on inorganic scintillator discovery is shown in Figure. ZnS played an important role in the discovery of the alpha-particle by the famous experiments of Ernest Rutherford. Around 900 photon detectors were not yet available and for the detection of the scintillation flashes the eyes of young students were used. Matters changed after the development of the photomultiplier tube around 95. Very soon NaI activated with Tl + was discovered. Even today, is still the most widely applied scintillator. CsI:Tl has higher density than but it is much slower, LiI:Eu was developed for thermal neutron detection and BGO is popular for its very high density. New scintillation research activities arose after the discovery of a sub-nanosecond fast scintillation decay component in BaF in 98. The fast emission is caused by a then new luminescence phenomenon, i.e., core valence luminescence (CVL). PbWO was developed at CERN in Geneva. More than 75 thousand cm long PbWO scintillator crystals are required for the electromagnetic calorimeter of the CMS (Compact Muon Selenoid) detector at LHC (Large Hadron Collider). In terms of light output (00 photons/mev) it is a very poor scintillator. However, considering the very high energies of the gamma rays involved, the high density and the short decay time are much more important parameters. Past 5 years many Ce + activated scintillators have been developed. LuSiO5 combines a high density with a fast scintillation response and is used in scanners for medical diagnostics. The newest scintillators LaCl:0%Ce and LaBr:5%Ce+ were discovered at Delft University and provide record high energy resolution and ultrafast detection of gamma rays. It is available under the trade mark BriLanCe and generates much interest in the radiation detection world.
ZnS:Ag CaWO LaBr LaCl RbGdBr7 LuAlO LuSiO5 PbWO CeF (Y,Gd)O BaF( fast) YAlO BiGeO BaF (slow) CsI:Na CdS:In ZnO:Ga CaF:Eu silicate glass LiI:Eu CsI CsF CsI:Tl CdWO 900 90 90 90 980 000 00 year Fig,. The history of inorganic scintillator discovery (reproduced from M.J. Weber, J. Lumin 00 (00) p.5) Table : Compilation of scintillation properties (density ρ, refractive index n, scintillation decay τ, emission wavelength λ, photon yield Y, energy resolution (FEHM) at kev) of some well known scintillators. scintillator ρ (g/cm R @ kev ) n τ(ns) λ (nm) Y (ph/mev) (%) NaI(Tl).7.85 0 5 000.5 CsI(Tl).5.80 0 50 5000 7 BaF slow comp..89.5 0 0 9500 8 BaF fast comp, 0. 0 00 -- BiGeO (BGO) 7..5 00 80 800 7.7 PbWO 8.8.0 0 70 00 -- LuSiO5 (LSO) 7..8 7 0 5000 8 YAlO 5.7.95 7 70 8000 LaCl:0%Ce.8.9 50 000. LaBr:5%Ce 5.07.95 7 80 70000.8 Scintillation light output and scintillation mechanisms
Figure demonstrates the general principle of scintillation. An ionizing particle creates an ionization track in the host crystal, and electrons from the filled valence band are excited to the empty conduction band (arrow ). The average energy to create one ionization is about.5e g where E g is the band gap of the scintillator. The value for β is larger than unity because in the electron-electron collisions during track creation momentum conservation does not allow for the creation of electron and holes of zero kinetic energy (=momentum). This means that for a wide band gap oxide crystal with E g =8 ev, approximately 50,000 free electrons and free holes are created upon total absorption of MeV gamma ray energy. L 5 Fig. Principle of scintillation in activated wide band gap materials. After ionization, the hot electrons relax to the bottom of the conduction band (arrow ) and the hot holes relax to the top of the uppermost valence band (arrow ). Next the free electrons should recombine with the hole to emit a photon. Figure illustrates the situation for an impurity activated scintillator. The activator ion creates energy levels within the forbidden gap of the host material, and the free electrons and holes recombine radiatively via an excited state of the impurity ion (arrow ). The total light output is given by Y ph 0 SQ = photons/mev () β E g where Y ph is the number of photons emitted by the scintillator per unit of energy absorbed (usually photons/mev). β is a constant that appears approximately.5. For the ideal situation, the transfer efficiency S and the quantum efficiency Q of the activator ion are 00% and then with E g =8 ev the light yield will be 50000 ph/mev. In the ideal situation also the transfer speed of free electrons and free holes to the impurity ion is instantaneous, i.e., faster than ns. In that case the rise time of the scintillation pulse is very short and the scintillation decay time τ s is determined by the life time τ ν of the activator excited state only. Figure shows the light output of known scintillators and luminescent phosphors as function of the band gap of the host material. The solid curve is the theoretical maximal output obtained when S=Q=. For the available oxide scintillators the yield is limited to below 0000 ph/mev whereas based on Eq. () values of 50,000 ph/mev should be possible.
The situation is much better for LaCl and LaBr. They appear, in terms of light output, ideal scintillators with yields close to the theoretical maximum. Proceeding to increasingly smaller band gap materials we arrive at the most popular scintillator. With 000 ph/mev, appears only 50% as efficient as the theoretical maximum. Pure NaI at liquid nitrogen temperature is known to yield much higher light output of 80000 ph/mev. The highest light outputs are reported for the sulfides which have the smallest band gap. ZnS:Ag, used for α particle detection by Rutherford, has a light yield of 90000 ph/mev, which is about the same as for CaS used in the first generation of cathode ray tubes for TV screens. 0 Yield (photons/kev) 0 00 80 0 0 0 ZnS:Ag Lu S LaBr NaI:(80K) CsI:Tl K LaCl 5 Lu SiO 5 LaCl YAlO BaF bromides oxides fluoride sulfides chlorides iodides β=.5 CaF :Eu 0 5 7 8 9 0 E gap Fig. Scintillator light output of various scintillators as function of the band gap of materials. The solid curve indicates the theoretical limit. Most scintillators and phosphors do not reveal the theoretical maximum light yield. There are many causes for electron-hole losses in scintillators. Figure shows that electrons and holes may recombine without emitting a photon (arrow 5). This may occur in the intrinsically pure lattice but also due to the presence of defects or unintended impurities that are sometimes called ``killer centers'' (arrows ). Other mechanisms competing with the wanted scintillation process are illustrated in Figure. It shows that a hole in the valence band is trapped in the activator ion (arrow ), but the electron is trapped somewhere else (arrow ). When the electron trap is shallow (< 0.5 ev), the trapping is not stable at room temperature. Thermal activation of the electron back to the conduction band and subsequent transfer to the hole trapped on the activator ion (arrow ) may lead to delayed luminescence (arrow ). Depending on the depth of the electron trap, this afterglow may last several ms or it may persist much longer. LuSiO5+, for example, shows a strong afterglow lasting for several hours. For materials with even large trapping depth the trapping is permanent at room temperature. Those types of materials can be used for dosimeters. The number of filled traps is proportional to the amount of radiation dose received. By heating the material the electrons can be liberated from their traps. Recombination of the electron with the hole then yields luminescence (thermo-luminescence). The TL intensity is a direct measure for the received dose. A hole in the valence band tends to be shared between two adjacent anions and a molecular like defect is created in the lattice. In alkali-halides the molecular complex is known as a V k center. By thermal activation the V k center may jump from one site to an adjacent site. It tends to trap an
electron. If this occurs, a self trapped exciton (STE) is created. The STE is a neutral defect and may also migrate relatively easily by thermal activation through the lattice. The self trapped exciton can also decay under the emission of a photon. Fig.. The role of trapped holes in the scintillation process a) LaC 0. % b) LaC % c) LaC 0 % 00 K 00 K 00 K 00 K 75 K 5 K 50 00 50 00 50 500 550 wavelength (nm) 00 K 50 00 50 00 50 500 550 wavelength (nm) 00 K 50 00 50 00 50 500 550 wavelength (nm) Fig. 5. X -ray excited emission in LaCl with 0.%, %, and 0% Ce+ as function of temperature in K Figure 5 shows X-ray excited emission spectra of LaCl with Ce concentration of 0.%, %, and 0% Ce. At 5 K and 0.%, Ce + emission is observed as the double peaked emission at 7 nm and 58 nm. In addition a 0.70 ev broad band emission is observed peaking at 00 nm. This emission is caused by STEs. Upon heating to 00 K, the STE emission disappears and the Ce emission gains intensity. The explanation is as follows. At low temperature the free holes have two options: ) they self-trap to form a V k center or ) they are trapped by Ce+ to form Ce+. This all happens on the sub-nanosecond timescale. At 5 K, the V k center mobility is low and it 5
will trap the electron to from the STE that provides the broad band emission. Electrons trapped by Ce+ give the characteristic Ce+ emission doublet. Upon heating the crystal, the STE becomes mobile and transfers its energy to Ce+. Ce+ emission with an effective lifetime dictated by the transfer rate from the STE is observed. When the concentration increases the role of STEs becomes less and for 0% Ce+ almost all emission is as Ce+ emission. Figure (left panel) shows that the scintillation pulse of LaCl contains a fast component and a slow component. Scintillation decay curves of and LuSiO5+ (LSO) are also shown for comparison. 0000 Intensity (a.u.) LaCl :0%Ce + LaCl :0%Ce + Lu SiO 5 intensity (arb. units) 000 00 0 LaBr :0.5%Ce LaBr :%Ce 0 00 00 00 800 000 time (ns) 0 50 00 50 00 50 00 time (ns) Fig.. γ excited decay curves of various scintillators. Energy resolution and non-proportionality Figure 7 shows pulse height spectra of a 7 Cs source, emitting kev gamma rays and kev X-rays, measured with a scintillator and a LaBr :0.5% Ce+ on a standard photomultiplier tube. One of the most important properties of a scintillator applied for gamma ray spectroscopy is the resolution with which the energy of gamma rays can be determined. Figure 7 shows that it is much better for LaBr than for the traditional.. (5).0 () counts (arb. units) 0.8 0. 0. 0. () () a) b) () E C () 77 kev 0.0 0 00 00 00 00 500 00 700 800 energy (kev) Fig. 7. 7 Cs pulse height spectra with a) LaBr:0.5%Ce+ and b) with +
The energy resolution R is usually specified as E R = =.5σ ( E) () E where E is the full width at half maximum intensity (FWHM) of the total absorption peak at gamma energy E and σ(e) is the standard deviation in the pulse height. The energy resolution achievable with a photomultiplier combination is about.%. The scintillator LaBr:0.5% Ce+ reveals a much better resolution of.%. This together with the much faster decay time of the LaBr+ scintillator, makes LaBr+ and also LaCl:0% Ce+ a scintillator that is superior to +. Formally the energy resolution can be written as R = R + R + R + R () stat np inh det where R stat is the contribution from the statistics in the number N dph of detected photons. R np is a contribution connected with non-proportionality in the scintillation light yield with gamma ray or electron energy. R inh is a contribution from in-homogeneities or non-uniformities in the scintillator, the light reflector or the quantum efficiency of the photon detector. R det is a contribution from noise and variance in the gain of the photon detector. The last two contributions are related to crystal growth and detector technology. The first two are fundamental in nature and intrinsic to the scintillator. R stat follows Poisson statistics R stat ( ) + v M =. (5) N dph Where v(m) is the variance in the gain of the photomultiplier tube and is about 0.. Figure 8 shows the energy resolution of scintillators at kev as function of N dph. These are the number of generated photoelectrons in the case of PMT readout. The solid curve represents R stat given by Eq. (). As required by Eq. () energy resolutions are always larger than R stat. Data on YAlO+, LaCl:0% Ce+ and LaBr:0.5% Ce+ are very close to R stat indicating that the other three contributions in Eq. () are insignificant. The situation for the well known scintillators, CsI:Tl, and LuSiO5+ is much different. The observed resolution appears twice as large as R stat indicating important other contributions. The poor resolution of, CsI:Tl, and LuSiO5+ is caused by a response of the scintillator that is not proportional with the energy of the gamma ray. Figure 9 shows the proportionality curves for several scintillators. The light yield in photons/mev at gamma ray energy E γ relative to the light at energy kev is shown as function of E γ. For a proportional response, the curve should be a constant line at value. That of YAlO+ and LaBr+ are indeed close to one between 0 kev and MeV. On the other hand for LuSiO5+, 7
gamma rays or X-rays of 0 kev are 0% less efficient in producing scintillation light than at kev energy. For 0 kev gammas are 0% more efficient than at MeV. Energy resolution at kev 9 8 7 5.. BaF Lu SiO 5 YAlO.0 CsI:Tl 0.9 LaBr YAlO 0.8 LaCl LaBr 0.7 Lu SiO 5 CdZnTe 0. Ge 0.5 000 0000 00000 0 00 000 N dq energy (kev) relative yield Fig. 8. Left panel. The energy resolution (FWHM) of scintillators at kev. The solid curve is the calculated Poisson statistical contribution. Ge and CdZnTe are semiconductor detectors. For Ge the energy resolution of 0.% at kev is 5 times better than expectations from Poisson statistics. The is because the variance in the number of generated and detected electron hole pairs in the ionization track does not follow Poisson statistics. Fig. 9 Right panel. Proportionality curves of scintillators, i.e., normalized scintillation yield as function of gamma ray energy. The non-proportionality with gamma ray energy is directly related with non-proportionality with electron energy. Suppose Y(E e ) is the light output of a scintillator as function of electron energy E e. After the absorption of a gamma ray with energy kev, a cascade of events takes place both in the atom that the gamma particle interacted with and in the ionization track formed by the primary electron. For a gamma particle labeled, the cascade eventually results into a collection of n secondary electrons (also called δ-rays) with energies E, E,.., E n that each create a branch n of the ionization track with. E = kev. Another gamma particle labeled creates another i i m collection of m secondary electrons E, E,.., E m. Again the sum E = kev. Since the light yield is not proportional with electron energy, the two gamma s of the same energy do not produce equal amount of photons. The creation of energetic secondary electrons is a statistical process and leads to fluctuating light output and therefore to a contribution R np in Eq. (). Only when a scintillator is proportional the light output does not depend on secondary electron distribution. This is the situation for YAlO+, LaCl+, and LaBr+ at energies above 0 kev. Scintillation decay time and the luminescence centers Decay time of the scintillation pulse is the most important parameter when a scintillator is used for fast timing. It depends on the transfer speed of charge carriers from the ionization track to the luminescence center and by the lifetime of its excited state. In the ideal case of infinitely fast (i.e. i i 8
faster than ns) energy transfer to the luminescence center, the scintillation decay is determined by the decay rate Γ ν =/τ ν of the excited state. The decay rate is given by ( + ) n n Γ ν = f µ i () τ λ ν The decay rate decreases with the third power of the wavelength λ of emission and increases with the refractive index n of the host material. The last factor in Eq. () is the matrix element that connects the initial state (this is the excited state of the activator ion) with the final state via the electric dipole operator µ. This is non-zero only for electric dipole allowed transitions. The number of activator ions that show electric dipole allowed transitions suitable for scintillation applications is limited. These are the transitions between the 5d and the f orbitals in the lanthanides, Ce+, Pr+, and Eu+ and the transitions between the sp and s configurations in the so-called s -elements Tl+, Pb+, and Bi+. These are precisely the activator ions encountered in applied scintillators. Transitions in Tl+, Pb+, and Bi+ are from a sp spin triplet P J state to the s spin singlet S 0 ground state. Those transitions are spin forbidden, yet partly relaxed by the spin-orbit interaction, leading to relatively long scintillation decay times. The scintillation decay time in is 0 ns (see Figure ) and 00 ns for BGO. For fast timing purposes the s -elements are not suitable as activator. Eu+ is also relatively slow (~ µs). Figure 0 shows X-ray excited emission spectra of Ce+, Pr+, and Nd+ in various compounds. Ce+ emits in a characteristic doublet at wavelengths depending on the type of compound. It is usually around 00 nm in fluorides and it tends to shift to longer wavelengths with smaller value for the band gap of the host crystal. For LuS+ the emission is in the red at 00 nm. Pr+ shows a more complex emission with four main bands. When in the same compound as Ce+ the emission to the ground state is always at.5 ev higher energy than Ce+. That of Nd+ emits at.8 ev higher energy. wavelength [nm] 800 00 00 00 yield [arb. units] Ce + 0.8 Lu S LiYF 0. 0.8 0. Nd + 0.8 LaF 0. Pr + f -f Y Al 5 O f -f LiLuF quantum efficiency [%] 80 0 0 0 TMAE Nd + Pr + photomultiplier tube Ce + silicon photodiode 0 0 0 0 50 0 wavenumber [0 cm - ] 0 00 00 00 800 000 wavelength [nm] Figure 0. Left panel. X-ray excited luminescence in Ce+, Pr+, and Nd+ doped compounds. Figure Right panel. Wavelength range of Ce+, Pr+, and Nd+ 5d-f emission in compounds compared with typical quantum efficiency curves of a photomultiplier tube, a photodiode, and the photosensitive gas TMAE. 9
In addition to the relatively broad 5d-f emissions in Pr+ and Nd+, these ions also reveal narrow emission bands caused by transitions between f levels. These emissions are dipole forbidden and very slow (ms). The life time of the Ce+ 5d state depends on the type of compound and is found between 5 and 0 ns. Because of the shorter wavelength of emission, see Eq. (), the lifetime of Pr+ is roughly two times shorter and that of Nd+ is four times shorter than that of Ce+. Figure shows the observed range of values for the 5d-f emission of Ce+, Pr+, and Nd+ in compounds together with the quantum efficiency curves of a photosensitive gas TMAE, a bialkali photomultiplier and an photodiode (PD). Depending of the type of compound the emission of Ce+ may match nicely with the maximum quantum efficiency of a PMT or of a (A)PD. Altogether Ce+ is an unique activator ion. It combines: ) a fast 5d-f emission, ) absence of slow f-f emission, ) an almost unity Q= internal quantum efficiency, ) the proper wavelength of emission, 5) it is also an excellent hole trap, and finally ) it can be substituted on La+, Gd+, and Lu+ sites that are constituents of high density host crystals. Anatomy of a pulse height spectrum Figure shows the pulse height spectrum of Na measured with LaBr, the highest energy resolution scintillation available today, coupled to a photomultiplier tube (PMT). Although the source emits only gamma photons of.75 MeV and.7 MeV, the spectrum is very rich in features that are, in order of decreasing energy, numbered to 5. Starting on the high energy side we first observe the total absorption peak () at.75 MeV caused by a) photoelectric absorption, b) Compton scattering followed by photoelectric absorption of the scattered gamma ray, c) pair creation followed by absorption of the two 5 kev annihilation quanta. The maximum energy E C (E γ ) transferred to the electron in Compton scattering is E C Eγ = 5+ E γ (7) and it gives the Compton edge at E C (.75)=.5 MeV which is denoted by feature () in Fig.. When a gamma ray undergoes multiple Compton scattering interaction in the scintillator it contributes to the counts in the tail at (). Part of the.75 MeV gamma rays is absorbed by pair creation. The created positron looses its kinetic energy and subsequently annihilates with an electron creating two 5 kev gamma rays that may or may not escape from the crystal. Features (7) and (5) at 7 kev and kev are the double and single 5 kev escape peaks. 5 kev annihilation quanta Compton scattered in the scintillator with escape of the scattered gamma ray lead to features () and () on top of the Compton background from.75 MeV gamma rays. Those features extend to 07 kev, i.e., 70 kev below the single 5 kev escape peak (5), and to 58 kev, i.e., 70 kev below the total absorption peak (). The pulse height spectrum is flat between 50 kev and 50 kev (8). This is the only part of the spectrum that is composed of one single contribution, i.e., from Compton scattering of.75 MeV gamma rays only. Peak (9) is the total absorption peak of.7 MeV gamma rays. Pair creation leads to the faint double () and single () 5 kev escape peaks. 0
9 counts (arb.units) 5 7 0 8 5 0 0.0 0.5.0.5.0.5.0 Energy (MeV) Fig.. The Na gamma ray pulse height spectrum measured with LaBr. Features to 5 are discussed in the text. The horizontal dashed line indicates the Compton background from.75 MeV gamma rays. Dashed vertical lines indicate the location of Compton edges. The Compton edge () starts below at E C (.7)=. MeV with a tail (0) due to multiple Compton scattering. Absorption of.7 and.75 MeV gamma rays outside the scintillator either by pair creation or Compton scattering and subsequent detection of 5 kev annihilation gamma ray or the Compton scattered gamma ray leads to the 5 kev back scatter peak () and the Compton back scatter events (5) at around 50 kev.