Calibration of X-Ray Source of a Powder Diffractometer and Radiation Test of Silicon Microstrip Detectors
Emmanuel Fokitis, Theodoros Geralis, Stavros Maltezos, Nikolaos Vodinas
CCalibration of X-Ray Source of a Powder Diffractometerand Radiation Test of Silicon Microstrip Detectors
E. Fokitis a , T. Geralis b , S. Maltezos a , and N. Vodinas ca National Technical University of Athens (NTUA), Physics Department b National Centre of Scientific Research “Demokritos” (NRCPS) c Ministry of Education and Religious Affairs, Greece
Abstract
Abstract.
A flexible apparatus for calibration of the absolute flux at the focal plane of the X-ray Source ofa Powder Diffractometer, based on a fast scintillator counter, is presented. The measured fluxes, dependingon the high voltage on the X-ray tube, were at the range 200 - 400 MHz, while an uncertainty in the fluxof the order of 5% has been estimated. We also applied this calibration for radiation hardness study of amultichannel silicon microstrip X-Ray detector.
Keywords—
Powder Diffractometer, X-Rays, Silicon microstrips
This work deals with the assembly of a detector prototype for the calibration of the intensity of a linear focusX-ray source of 1.5 kW electrical power, and the operational test of silicon microstrip detectors; the latter areused in the goniometric circle of a development diffractometer of Debye-Scherrer type. The beam is producedin a monochromatic mode after dispersion with monochromator of bent crystal. The calibration of the X-raysource will be necessary for the radiation tests of silicon detectors since they are expected to indicate (possible)changes due to radiation (Radiation Damage) and will give an estimate of their useful time. The knowledge ofthe relative peaks in the Debye-Scherrer spectra allows us to determine the relative strength of various crystalphases in each sample, but for an absolute estimation of these, the absolute flux of the X-ray source is necessaryand is usually required in a X-ray refractometer. The intrinsic calibration difficulty for the X-ray beam relatesto the counting rates which are of the order of 1 GHz, and pose well-known difficulties in coping with these. Inthe past, methods used for calibration employ absorbers with calibrated material named “Lupolen” (platelet ofPolyethylene), and recently use of single or double slits.The design presented in this work relates to a fast detector system with high efficiency in detecting X-raysin the energy range of 5-20 keV. We present here the counting rates for the spectral emission Cu-K a1 as afunction of the source current and high voltage. The appropriate correction coefficients for the dead time andgeometrical parameters of the detection system are given in this work.The experimental design and configuration is presented in Section 2. In Section 3, the experimental mea-surements and the data analysis methodology are analyzed. In Section 4, an analytical comparison of ourdetector with a commercial detector is performed. Finally, in Section 5, we present a radiation hardness test ofmicrostrip silicon detectors, while in Section 6 the conclusions and prospects are discussed. For the choice of the appropriate detector we studied the characteristics of detectors with high efficiency inX-rays in the range of 8 keV, such as silicon detectors of PIN or Avalanche Photodiodes, and organic orinorganic scintillators. The first category needs cooling and continuous temperature control due to the strongdependence of the signal and background on the temperature, and because of problems arising from the exposureto large radiation fluxes. The scintillators (organic and inorganic) are in general easier to use, with exceptionof the hygroscopic ones, and especially these having extremely small de-excitation time, (for example, the BaF scintillator has τ ≈ . − . τ ≈ − a r X i v : . [ phy s i c s . i n s - d e t ] A ug igure 1: The experimental setup of the cal-ibration apparatus. Figure 2: Schematics indicating the useof the X-ray detector prototype.Scintillators, such as Nai(Tl), which is customarily used for the X-ray detection, due to its high efficiencyand the large number of scintillation photons, is not recommended for large radiation fluxes because of its largede-excitation time. The BaF , among the fastest scintillators presents an additional requirement: the need ofan optical filter for isolating the fast component ( λ < λ >
220 nm. One of itsdisadvantages which appear also in certain fast scintillators is the small number of produced scintillation photonsin the visible range for each X-ray photon; this number is considerably smaller in comparison with the yield froma plastic scintillator. Thus, the scintillator of model BC452 of Bicron which was selected for our application, isdoped with Lead (5% Pb), has a small absorption length (large absorption coefficient: µ = 4 .
91 cm − at 20 keV),and an extremely fast response (de-excitation time, τ ∼ . λ = 420 nm, and a rather narrow FWHM ≈
30 nm. The narrow spectral bandwidth ofour detection system indicates the small optical noise which would arise from small and ultimately un-avoidablelight leaks.
The scintillator and the PMT were optically coupled and the system was made light-tight. Both were surroundedby a metallic shield. A housing-cover was attached to the scintillator side in such a way as to allow the insertionof a variable width slit (see Fig. 1 and Fig. 2). The PMT was connected with an appropriate HV distributioncircuit which allowed the possibility of optimum control of the voltage between the first two dynodes. For theX-ray flux measurement, we selected the method of operation in a “photon counting mode”. Thus, the analoguepulse from the PMT is input to the discriminator, model 4806C of LeCroy, and this output is input to a NIMto TTL converter. Finally, the TTL pulse is recorded by a homemade fast (100
M Hz ) counter-timer [1]. Thelatter had the capability to be controlled by a PC computer, which also records the number of counts in apredefined time gate (t=10 s). Before the flux measurements, we studied the PMT plateau curve versus HighVoltage and the dark current behaviour seen in Fig. 3.The selected-optimum High Voltage was 2000 V and the threshold of the discriminator was set at -30 mV.Next, the detector was at a distance of 26.8 cm from the capillary tube and its axis was in the X-ray beamdirection. To reduce the detected rate of X-ray photons, a metallic slit of width d = 40 ± µ m was placed infront of the scintillator entrance. We next describe a set of measurements done to study, first the characteristicsof the detector, namely its efficiency, dead time, and secondly, the calibration of the X-ray tube flux for variousHigh Voltage values, and tube currents. Setting the High Voltage in the X-ray tube at 16, 20, 25 and 30 kV, the X-ray fluxes were recorded for X-ray tube currents varying from 5 mA up to the maximum allowed value for each High Voltage selected, i.e.2igure 3: The PMT XP2262 plateau using the X ray tube. We see measurements with the source (triangulardata points), without source , i.e. dark current (square black data points) and the difference of the two (squarewhite points).16,20, 25 and 30 mA, respectively. The step in the tube current increment was 1 mA. For each of the abovesettings the dark counts of the detector were recorded before and after each measurement. The values ofthe dark current corresponding counting rate were of the order of 2-5 kHz. This value is consistent with thePMT manufacturer’s specifications. For large X-ray flux values, the well-known pile-up effect causes significantdeviations of the measured rates from the true ones, and above certain limit, we have the phenomenon ofparalysis [3], [4], [5], [6], [7]. The true beam intensity R may be written as a linear function of the X-ray tubecurrent I : R ( I ) = aI + β (1)where, R ( I ) is the counting rate of X-rays (counts/s) and I is given in [mA]. The coefficients to be determined, a and β , are called the calibration parameters. Due to the finite dead time, let τ , the recorded rate of X-rays, m , is: m = R ( I ) e − Rτ = ( aI + β ) e − ( aI + β ) τ (2)The dead time of the detector is also considered as a parameter to be determined at the same time withthe parameters a and β by non-linear χ fitting method. In Fig. 4, we observe the four different sets ofmeasurements for X-ray tube HV: 16, 20, 25 and 30 kV. The continuous line is the result of the fit and thedotted line represents the actual counting rate R .We observe, as seen in this figure, the higher the counting rates the stronger is the pile-up effect. In addition,we should note that for the case HV=30 kV, the fit fails if we include more than the 15 first points on the curve.This effect is due to the loss of linearity in the PMT performance; this may be due to reasons such as large spacecharges appearing between its dynodes and the induced deviation from linear change of the dynodes potentials[8]. Despite these effects, and up to HV around 40 kV, a good number of data points, in small current valuescan be used in order to obtain calibration for all allowed HV values in the range of 10-40 kV. A criterion forthe correctness of this calibration method can be the agreement of the dead time values of the detector, as theyare extracted independently from our four data sets.We used four combinations of the fitting parameters, a , β and τ . In two cases, alternatively, we had setconstant values, β = 0 and τ = 24 . β was statistically equal to zero in all the cases, which shows a “healthy” behaviour of the fittingprocedure. The parameter a differed, in the four cases studied, from the average value at most by 1 . . . . . a HV for the different HV settings are given. Theobtained results were statistically consistent between each other. The value of the dead time of the detectorwhich we accept is the statistical weighted average, using as weight the inverse of the square of the error,thatis, τ = 24 . ± . a and β .High Voltage [kV] Dead time [ns] a HV [mA − s − ] × χ / dof16 29 . ± . . ± . ± . . ± . . ± . . ±
18 24/1830 24 . ± . ±
27 17/12Table 1: The dead time and the parameter a HV , obtained by the fitting procedure, for the four different valuesof X-ray tube High-Voltage. The above methodology was applied also for a similar commercial detector system, model SZINTIX [9], which isbased in a NaI(Tl) scintillator and has quite slower electronic units (amplifier, discriminator etc). The reasonswhich have led us to this additional test were, on the one hand, to confirm the validity of the method, and,on the other hand, to determine the efficiency of the detector “XP2262/BC452” in lower photon rates, as itis given that the SZINTIX detector has efficiency 100% at 8 keV. The dead time of the detector SZINTIX,for the corresponding cases 16, 20, 25, and 30 kV, with use of Aluminum absorber of thicknesses (2 . ± . . ± . . ± . . ± . µ m, giving an average value τ = (5 . ± . µ s. The detection efficiencyof XP2262/BC452 at 8 keV, using similar geometry for the two detectors, was found to be: A det = 1 . ± . Cu − K a X-rays
The efficiency of the detector is defined as the ratio of the experimentally recorded number of the X-rays to theactually incident ones. The inefficiency of the detector is related to the number of expected photons, within the4pectral window of sensitivity of the PMT, produced by each X-ray. According to our estimations the numberof such photons at 8 keV is around 46 for the scintillator BC452. The number of photons incident on thePMT depends on the angular distribution of the scintillation and the details of the geometry of the scintillator.Therefore, there should be an appropriate simulation in order to get an accurate estimation of the efficiency.However, assuming that around 15% of the scintillation photons are incident on the PMT photocathode, weobtain that around 7 photons per incident X-ray are recorded. This should be multiplied by the averagecollection efficiency ( ∼ .
95) and quantum efficiency of the PMT ( ∼ . Using the above data, and the corresponding analysis, it was possible to determine the dead time of our detectorsystem, which was τ = (24 . ± .
1) ns. Furthermore, by solving Eq. 2 for R , we may determine the true X-raysrate. The solution of this transcendental equation is done numerically, by the Newton-Raphson method. Thecorrection factor, f ( m ), as a non linear function of the measured counting rate, m , and its parameterization is, f ( m ) = 4 . × − m − . × − m + 6 . × − m + 1 . × − m + 1. Its value ranges from 1.0 to1.9. The real rate is then calculated by the formula, R = f ( m ) m .Figure 5: True rate of X-ray photons (N real ) as a function of the recorded rate (N counted ). It must be noted that the parameter a corresponds to X-rays rate for a slit of width d = 40 µ m at a distanceof r = 26 . . σ from the distribution peak). The parameter a suffices for the determination of the real X-rays photon rate, when the HV and the current I of the X-ray tubeis known. At the present time, the value of a was determined only in the four cases discussed. In the future,on systematic study is expected to give the value of a for continuous variation of HV in the range of 10-40 kV.In such a case, an additional correction must be applied, regarding the detection efficiency by dividing R with A det . Then the intensity of the beam, in [Hz], passing through the slit, with I expressed in [mA], is: I slit = aIA det (3)For the determination of the absolute beam intensity we applied the following method: keeping the valuesof X-ray tube HV and current constant (25 kV and 10 mA, respectively) we measured the flux, at variouspositions, of the beam intensity profile with an angular step of 0 . o , which corresponds to a length variation of93.5 µ m at distance 26.8 cm from the bent crystal to the detector slit of width 40 µ m. In Fig.6, we observe the5esults of this scanning. By performing a χ fit with a Gaussian distribution, we find that the mean value whichcorresponds to angle < φ > = 3 . and standard deviation, σ φ = 0 . o or FWHM= 0 . o . As a result, theabsolute beam intensity in [Hz] is given by: I tot = (cid:80) N i N i max D step d slit A det ε corr N (cid:48) i max = (cid:80) N i N i max D step d slit A det aI (mA) (4)obtaining, I tot = 12 .
47 1 A det ε corr N (cid:48) i max = 12 .
47 1 A det aI (mA) (5)where N i is the corrected recorded rate of the i th element, is the recorded rate at the element of the maximum, D strip =93.5 µ m is the spatial step; D strip = 40 µ m is the slit width, ε corr is the correction factor and N (cid:48) i max isthe recorded rate at the maximum of the combination of HV and current to the evaluated one (for example,30 kV, 15 mA). The overall relative error of I tot is: σ I tot I tot = σ d slit ⊕ σ A det ⊕ σ N I tot = 0 .
06 (6)The main contribution in the overall error comes from the measurement of the slit width. More accuratemeasurement of this parameter with metrological method is expected to give significant improvement to thebeam’s absolute intensity accuracy.Two examples of computation of the total beam intensity follow: I tot (30 kV , . × . × . × . × ±
12) MHz I tot (30 kV , . × . × . × ×
30 = (402 ±
24) MHzFor the radiation hardness test we selected the settings HV=25 kV and I = 10 mA. The measured rate was6.045 MHz while the corrected one was 7.250 MHz at 0 . σ and on the peak was 7.800 MHz.Figure 6: The recorded X-ray beam profile in [counts/s] as a function angle φ (dashed line obtained from thefitting of the data shown with dark circles), while the solid line represents the data after dead time corrections. For the radiation hardness studies we used a microstrip detector, which comprises of 2000 microstrip elementsof width 30 µ m and heigth 1 cm, while the distance of the centers of two successive strips is 50 µ m. We placedthe center of the microstrip detector at the position corresponding to the peak of the intensity of the X-Raybeam as determined above. According to the results of the previous sections, based on the slit used, the rate6s 7.8 MHz. Converting this rate, which corresponds to an irradiated area 40 µ m by 1.5 cm, to the microstripdetector pixel size of 30 µ m by 1 cm, we obtain an irradiation rate of 3.9 MHz.Additionally, based on the FWHM of the beam, we can find a corresponding spatial beam width equal to420 µ m (for a distance of 269 mm from the capillary tube - center of circumference of sweeping of beam). Sincethe distribution of the beam intensity is nearly Gaussian, the width of the detector which accepts high X-rayflux is expected to extend up to ± σ or about 4 FWHM we get 1680 µ m. In this range, about 34 siliconmicrostrips are exposed or 17 microstrips on each side of the central microstrip. In addition, the exposuretime in this photoflux was 12 hours. After the irradiation of the microstrip detectors, measurements of leakagecurrents were carried out (Institute of Microelectronics, Demokritos) in order to investigate if there are agingeffects which could be due to creation of surface states in the interface Si-SiO induced by the X-rays irradiation.In the center of the detector, the leakage current at potential -100 V (full depletion) was found to be on theaverage around two or three times larger in comparison to the current before irradiation. Three microstrips onthe left of the 100 th (central) microstrip had large leakage current (larger than 20 nA). As we are departingfrom the center, the leakage current tends to the values recorded before irradiation, and becomes equal to thecurrent before irradiation after 50 microstrips from the center.The observed deviation from the affected, due to irradiation, width of 17 microstrips, according to the beamprofile, may be due to poor geometrical alignment of the detector or to other reason. This effect should beinvestigated. The leakage current at the central microstrip, with the exception of the three microstrips withhigh leakage currents, is at acceptable levels, and should have no direct effect in the pre-amplifier noise. Thebehaviour of the three microstrips with the high leakage current is not understood since the neighbouring stripshad almost regular current. It is probable that their high current is due to combination of photoelectron inducedageing, and to mechanical stress during their mounting in the irradiation procedure. Conclusions and prospects
We have been able to assemble and use an experimental setup to determine the X ray flux at the focal pointof an X-ray diffractometer . The measured fluxes were at ranges of 200 to 400 MHz, and an error in the fluxof the order of 6% was estimated. We have been able to use this flux of X-rays in order to study the aging ofmicrostrip detectors for X-rays. We are investigating a method to be able to monitor even higher dosages ofX-rays in facilities such as in Synchrotron radiation.
Acknowledgements
We acknowledge the financial support of GSRT through the project EPET II.