GAIN STABILIZATION OF SCINTILLATION DETECTOR SYSTEMS

Abstract

A method and device are provided for obtaining the energy of nuclear radiation from a scintillation detector system for the measurement of nuclear radiation the device comprising a scintillation crystal, a light readout detector and a fast digital sampling analog to digital converter. The method comprises obtaining the anode current at the LRD for at least one scintillation event with N photo electron charges at the entrance of the LRD, sampling the measured anode current, obtaining the function of the scintillation pulse charges Q.sub.dint(N, G) at the anode of the LRD from said scintillation events, obtaining the RMS of the noise power charge Q.sub.drms(N, G), obtaining the function Q.sub.dSN(N) by calculating the ratio of Q.sub.dint(N, G) and Q.sub.drms(N, G), obtaining the constant gradient k from the function Q.sub.dSN(N)=Q.sub.dint(N, G)/Q.sub.drms(N, G)=k*N, and obtaining N.

Claims

1. A method for obtaining the energy of nuclear radiation from a scintillation detector system for the measurement of nuclear radiation, preferably gamma radiation, comprising a scintillation crystal, a light readout detector (LRD) and a fast digital sampling analog to digital converter (ADC), the scintillator being selected from a group of materials, having a light decay time of at least 1 ns, the method comprising the following steps: obtaining the anode current I.sub.A(t) at the LRD for at least one scintillation event with N photo electron charges q.sub.e at the entrance of the LRD, the LRD set to a suitable gain G, sampling the measured anode current I.sub.A(t) with a predefined sampling rate, obtaining the function of the scintillation pulse charges Q.sub.dint(N, G) at the anode of the LRD from said scintillation events by integrating the anode current I.sub.A(t) for a specific time period Δt, starting at a time t.sub.0, i.e. by summarizing the digitized anode current samples i.sub.τ for each scintillation event, obtaining the root mean square (RMS) function of the noise power charge Q.sub.drms(N, G) by calculating the square root of the summed squares of the second difference of digitized anode current samples i.sub.τ for each such scintillation event, obtaining the function Q.sub.dSN(N) by calculating the ratio of Q.sub.dint(N, G) and Q.sub.drms(N, G), that ratio being an almost linear function of the number N, obtaining the constant gradient k from the at least one function Q.sub.dSN(N)=Q.sub.dint(N, G)/Q.sub.drms(N, G)=k*N, obtaining the number N, said number N being equivalent to the energy E deposed in the scintillator by a nuclear radiation event, from the ratio of the measured values Q.sub.dint(N, G) and Q.sub.drms(N, G) and the gain factor k.

2. The method of claim 1, wherein the light readout detector (LRT) is selected from a group of detectors, comprising a photomultiplier tube (PMT) or a photo tube with a photo cathode as light detector, a semiconductor photomultiplier, and an Avalanche Photodiode

3. The method of claim 1, wherein the output signal of the LRT is sampled with a sampling rate, being at least 5, preferably at least 200 and even more preferred at least 1.000 times faster than the light decay constant of the scintillator selected.

4. The method of claim 1, wherein NaI(Tl) is selected as a scintillator material, where the output signal of the LRT is sampled with a sampling frequency of at least 20 MHz.

5. The method of claim 4, where the output signal of the LRT is sampled with a sampling frequency between 200 and 4.000 MHz.

6. A stabilized detector system for the measurement of nuclear radiation, preferably gamma radiation, comprising a scintillation crystal, a light readout detector (LRT) and a fast digital sampling analog to digital converter (ADC), the scintillator being selected from a group of materials, having a light decay time of at least 1 ns, the ADC being set to operate with a very high sampling rate of at least 5 MHz, wherein the detector system conducts the method steps of claim 1.

7. A Stabilized detector system for the measurement of nuclear radiation, preferably gamma radiation, comprising a scintillation crystal, a light readout detector (LRT) and an analog signal electronics for evaluation of the data, the scintillator being selected from a group of materials, having a light decay time of at least 1 ns, wherein the detector conducts the method steps of claim 1 using analog signal electronics in place of digital electronics.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0019] A specific example will be described in the following, referring to FIGS. 1 to 3, describing the principle set of the claimed system. It shows:

[0020] FIG. 1, showing a principle setup of the detector system,

[0021] FIG. 2, showing a digitized current signal of a measured 30 keV NaI(Tl) scintillation pulse,

[0022] FIG. 3, showing simulated noise charges vs photoelectron counts

DETAILED DESCRIPTION

[0023] FIG. 1 shows a general setup of a detector system, which is known in the art. Shown is a scintillator crystal 10 with a reflective coating 11, reflecting the light 12, emitted from the scintillator when a gamma ray 15 from a radiation source 16 interacts with the scintillator crystal 10. At one side of the scintillator crystal 10, a photocathode 20, connected to a PMT 50, is located. There is no reflective coating in a scintillator crystal 12 at the side of the photocathode 20.

[0024] When the light 12 hits the photocathode 20, photoelectrons 25 are emitted and directed to a dynode chain 30, hitting the first dynode D1. The number of electrons hitting the first dynode D1 is then multiplied by a factor δ from the first dynode, then hitting the second third and so on dynodes before leading to the anode 40. The resulting anode current I.sub.A(t) is digitized with an appropriate sampling rate, resulting in digitized anode current samples i.sub.τ. Individual pulse charges are collected by integrating the anode current starting at the trigger time t.sub.0 for a specific time period Δt. The current integration is realized by summing the digitized anode current samples i.sub.τ. They can be expressed in terms of the suppositional cathode current samples j.sub.τ and the electron multiplication gain G:

[00001]$Q_{\mathrm{int}}\left(N,G\right):=\underset{N}{.Math.}.Math..Math.q_e\left(t_n\right)={\int }_{t_0}^{t_0+\Delta .Math..Math.t}.Math.I_A\left(t\right).Math.\mathrm{dt}\cong .Math.Q_{\mathrm{dint}}\left(N,G\right).Math.\text{:}=\underset{{\tau }_0}{\overset{{\tau }_0+\mathrm{\Delta \tau }}{.Math.}}.Math..Math.i_{\tau }=.Math..Math.{\mathrm{Gj}}_{\tau }$

[0025] It is of course also possible to measure the charge signal with analogue electronics. Doing so, the current signal from anode 40 is directed to a preamplifier, a voltage amplifier and a discriminator before it is further processed by an analog to digital converter (ADC).

[0026] In any case, the resulting digital signal is stored in a memory, i.e. a memory of a multichannel analyzer. That spectrum—the “energy spectrum”—may be further evaluated with a computer.

[0027] In practice, only charge integrating measurements as described above can capture the total charge of the signal. The second charge measuring method, a direct count of the photoelectrons, might still be possible after applying advanced deconvolution techniques. However, deconvolving photoelectron shapes accurately enough for counting, is rendered impossible in real measurements due to noise and statistical photoelectron pulse shape variations.

[0028] FIG. 2 shows the digitized current signal of a 30 keV NaI(Tl) scintillation pulse. It has been acquired at 10 Gs/s at 1 GHz bandwidth and subsequently smoothed by a 10-tap moving average FIR filter. The fragmented trace illustrates a typical complex signal and the difficulty to distinguish photoelectrons from noise is obvious. There is no method to count photoelectrons directly in this case. The single electron pulse width and noise prohibit photoelectron counting even at otherwise sufficiently high sampling rates.

[0029] The invention for the first time makes use of the fact that there is a principal difference between the patterns of low energy and high energy pulses when studying those on an oscilloscope. The rough pattern of the visible single photoelectrons at low energies will smooth out significantly towards higher energies. According to the invention, this photoelectron-pattern is described as random noise, added to the spectrum. In the following, this noise pattern will be called photoelectron “noise charge”.

[0030] This so-called photoelectron noise charge is calculated in an RMS (root mean square) expression, equivalent to the electronic noise power calculation. It is also the same calculation as for the second moment of a noise distribution. The inventor found that the second moment scaled by the total charge is independent of the PMT electron gain. In other words, the pulse pattern itself—the photoelectron noise charge—always is a measure of the energy of a scintillation event, no matter which scale or amplification gain.

[0031] This follows from the definition of the noise (power) charge function Q.sub.drms for the digitized pulse current, which can be described as follows:

Q.sub.drms(N,G):=√{square root over (Σi.sub.τ.sup.2)}G√{square root over (Σj.sub.r.sup.2)}

[0032] The ratio Q.sub.dSN of the total charge Q.sub.dint(N,G), expressed in terms of the (virtual) cathode current samples, and the electron multiplication gain G as described above, and the scintillation pulse noise charge Q.sub.drms(N,G) reduces with respect to the gain G. The magnitude of j.sub.τ only depends on the elementary charge constant e and the number of elementary charges within the time interval Δτ. Thus, the ratio Q.sub.dSN is a function of N only:

[00002]$Q_{\mathrm{dSN}}\left(N\right):=\frac{Q_{\mathrm{dint}}}{Q_{\mathrm{drms}}}=\frac{.Math..Math.i_{\tau }}{\sqrt{.Math..Math.i_{\tau }^2}}=\frac{G.Math..Math..Math.j_{\tau }}{G.Math.\sqrt{.Math..Math.j_{\tau }^2}}=\frac{.Math..Math.j_{\tau }}{\sqrt{.Math..Math.j_{\tau }^2}}$

[0033] It follows that the normalized noise charge E.sub.dSN for small N is proportional to N:

E.sub.dSN:=Q.sub.dSN.sup.2.fwdarw.N

[0034] The function E.sub.dSN can be linearized with a modification in the moment calculation by using derivatives. More specifically, N-th order derivatives correspond to n-th differences for discrete samples. With second differences, denoted as

[00003]$\frac{d^2.Math.i_{\tau }}{\mathrm{d\tau }},$

in the square root expression instead of the unmodified current value i.sub.τ, the ratio remains still independent of G and the function Q.sub.dN approaches const√{square root over (N)}:

[00004]$Q_{\mathrm{dN}}\left(N\right):=\frac{Q_{\mathrm{dint}}}{Q_{\mathrm{drms}}}=\frac{.Math..Math.i_{\tau }}{\sqrt{.Math..Math.{\left(\frac{d^2.Math.i_{\tau }}{\mathrm{d\tau }}\right)}^2}}~\sqrt{N}$

[0035] The noise charge E.sub.N(N) is then defined as

E.sub.N(N):=Q.sub.dN(N).sup.2

[0036] It has to be noted that even if the above example does show the 2.sup.nd difference, other, especially higher differences can be used also. Experiments show that the use of the 4.sup.th difference show very good results also. In general, the higher the difference used is, the better the result will be. As the resulting noise does increase with higher differences also, there is a practical limit to this increase.

[0037] In FIG. 3 simulated noise charges E.sub.N(N) are plotted versus photoelectron counts from 0 to 3 MeV, which corresponds to 0 to 30.000 photoelectrons when QE is set to 10 for this simulation. E.sub.N(N) exhibits a strong statistical variance. It increases with higher energies. The variance is the “price” for the statistical method of deriving the underlying counts. It stems from the deconvolution of the random photoelectron pileup by double differentiation and the consecutive variance calculation. Scintillation pulses at the very same energy, the same total charge and the same number of photoelectrons will still have different, randomly distributed current trajectories.

[0038] The photoelectron noise charge may be used for gain stabilization of the detector system as follows: the noise charge of a few hundred pulses using an arbitrary selected gamma source or the natural background for that matter is fitted to a straight line. The gradient of the straight is a direct measure of the PMT gain. In order to stabilize the PMT photoelectron multiplication, the gain setting has to be adjusted to keep this gradient constant:

[00005]$E_N\left(N\right)=\frac{{\left(.Math..Math.i_{\tau }\right)}^2}{.Math..Math.{\left(\frac{d^2.Math.i_{\tau }}{\mathrm{d\tau }}\right)}^2}=\mathrm{const}.Math..Math.N$

[0039] It is obvious from the above, that this method for gain stabilization of a detector system can be performed in parallel to a running measurement, as there is no need for a calibration source or any other signal. As long as the sampling rate and the speed of the PMT is fast enough in relation to the scintillator crystal, the proposed utilization of a statistical photon count in combination with a classical charge integration provides a simple method for gain stabilization of the detector, as no additional hardware is necessary. Typically, the sampling rate is 5 to 1.000 times faster than the light decay constant of the scintillator, preferably between 50 and 1.000 times faster. Therefore, the sampling rate is typically between 5 and 4.000 MHz, but sampling rates of 10.000 MHz or even more may also be used. When using NaI(Tl) as a scintillator material, this would result in preferred sampling rates between 20 and 4.000 MHz, preferably between 20 and 4.000 MHz, as the light decay constant τ of NaI(Tl) is in the order of 250 ns. This makes both, the method and the detector, making use of said method, simple, reliable and comparably cheap.

[0040] It has to be noted that all functions described above in a digital setup, can also be realized in analog electronics. Therefore, the invention is not limited to a method and a device using digital electronics, even if the method is explained in the context of such a digital setup.

LIST OF ABBREVIATIONS

[0041] δ overall multiplication factor for one dynode of a PMT [0042] e elementary charge [0043] E.sub.dSN normalized noise charge [0044] eV electron volt [0045] G (total) gain of a PMT [0046] I.sub.A(t) anode current at the PMT [0047] i.sub.τ anode current sample [0048] j.sub.τ cathode current sample [0049] N number of dynodes in the PMT [0050] q.sub.e photo electron charge [0051] QE quantum efficiency [0052] Q.sub.rms/Q.sub.drms noise (power) charge function [0053] Q.sub.erms RMS scintillation pulse noise charge [0054] Q.sub.int/Q.sub.dint total charge of a scintillation pulse measured at the anode [0055] Q.sub.SN/Q.sub.dSN ratio of Q.sub.int/Q.sub.dint and Q.sub.rms/Q.sub.drms [0056] PMT photo multiplier tube [0057] RMS root mean square