Nuclear Instruments and Methods in Physics Research A 424 (1999) 183 189 Position sensitive detection of thermal neutrons with solid state detectors (Gd Si planar detectors) G. Bruckner*, A. Czermak, H. Rauch, P. Weilhammer Institut Laue Langevin, B.P. 156, F-38042 Grenoble, France Atominstitut der O$ sterreichischen Universita( ten, Stadionallee 2, A-1020 Wien, Austria Institute of Nuclear Physics, PL-31-342 Krakow, Poland CERN, CH-1211 Geneva 23, Switzerland Abstract Recent advances in the technology of position sensitive silicon detectors and the corresponding electronics allow the construction of fast time response thermal neutron detectors. These detectors also exhibit excellent position resolution by combination of silicon detectors with thin Gd converter foils. We constructed several one- and two-dimensional prototype detectors, using DC and AC coupled silicon strip detectors, pad detectors and different VLSI readout electronics. The position resolution and the detector efficiency for different converters at wavelengths from 1.1 to 3.3 A were determined at the TRIGA reactor in Vienna and at the ILL in Grenoble. Spatial resolutions of less than 100 μm and efficiencies up to 40% have been achieved. The results of these measurements are compared with a Monte Carlo simulation of the detector operation. These detectors can also be used for phase topography experiments using perfect crystal neutron interferometers. In certain cases an increase of the sensitivity in the order of 100 can be anticipated compared to beam attenuation radiography. 1999 Elsevier Science B.V. All rights reserved. Keywords: Position sensitive neutron detector; Silicon; Gadolinium; Image 1. Introduction The central concept of these new detectors is to combine position sensitive Si planar sensors with external Gd converter foils [1 3]. The thermal neutrons are absorbed in the Gd foil and the resulting conversion electrons are detected in the silicon devices. Fig. 1 shows a schematic drawing of * Corresponding author: Tel: #33 4 76 20 7523; fax: #33 4 76 20 7777; e-mail: bruckner@ill.fr. a Gd Si neutron detector consisting of one Gd converter foil and two Si detectors. A position sensitive Si planar detector consists of an array of many pn-diodes on one single substrate. Each pn-junction works as a detector for ionizing particles. The position of the detected particle is given by the individual address of the diode that collects the charge. Each diode is equipped with an individual amplifier chain to ensure real time readout. Typical wafers have a diameter of 4 6 in. and a thickness of about 300 μm. The single structures can be as small as 10 μm and are limited in their 0168-9002/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8-9 0 0 2 ( 9 8 ) 0 1 2 8 3-2
184 G. Bruckner et al./nuclear Instruments and Methods in Physics Research A 424 (1999) 183 189 Fig. 2. Spectra of the reaction products of the neutron capture in Gd for different neutron wavelengths, as seen in a Siplanar detector in backward direction. Fig. 1. Schematic of a 4π Gd Si-neutron detector consisting of one converter foil and two Si-detectors with corresponding electronics. extent only by the size of the wafer. A position resolution of less than 10 μm has been achieved with these devices [4], as well as the construction of large detectors with areas up to m, for several high-energy physics experiments [5 7]. 2. External converter Considering only one converter layer (no sandwich detectors), the (n, γ) reaction in Gd, which produces also a reasonable number of conversion electrons, is the only suitable one for high detector efficiency. The proton, α-particle and triton from the Li and B reactions have a very short range. Therefore, these converters can only be used in situations, which require a low detection efficiency, e.g. beam monitors, or for the detection of cold or ultra cold neutrons. One can also think of doping the Si detectors with Li, but the usual doping concentrations of about 10/cm are much too low for efficient absorption of thermal neutrons. The (n, γ) reaction in Gd produces a quite complex conversion electron spectra (Figs. 2 and 3) Fig. 3. Spectra of the reaction products of the neutron capture in natural Gd for different neutron wavelengths, as seen in a Si-planar detector in backward direction. with significant energy lines between 29 and about 200 kev. Due to the high absorption cross section of Gd (σ "61 100 b for natural Gd, 259 000 b for Gd) the converter can be made very thin, so that the range of the conversion electrons is sufficient to reach the Si-detector. The band-gap in Si of 1.11 ev results in an average energy of about 3.6 kev for the electron hole creation. Therefore, the device must be capable to measure a charge of less than 5000 e to detect the low-energy conversion electrons,
G. Bruckner et al./nuclear Instrument and Methods in Physics Research A 424 (1999) 183 189 185 which have already lost part of their energy when escaping the converter foil. 3. Planar diode detectors and VLSI readout electronics The detectors are made of high resistivity silicon in planar technology and operated in fully depleted mode [8,9]. For a typical detector less than 100 V applied to the pn-junction ensures, that the whole detector works as one large depletion region. This mode decreases the detector noise, because it reduces the capacity between strip and backplane, and the thermally generated leakage currents. In addition, it guarantees that the entrance window on the rear side is as thin as the one on the front side and that there are no dead sections between the strips. While CCDs are made of only partly sensitive shift registers and contain inefficient channel stops to separate the colums, Si planar detectors are continuously efficient devices. Another important distinction between these detectors and CCDs concerns their readout. CCDs are integrating devices with charge transfer, that means serial readout. Therefore, a high-resolution time information is very difficult to obtain. Contrary Si detectors are constructed to measure single events with parallel readout electronics. Provided the lower level discriminator is set sufficiently high, the energy distribution of the conversion electrons does not effect the detector response, as in the case when the spectra would be detected by an integrating device such as film or CCD. The theoretical time resolution of the Gd Si detector is given by the charge collection time in Si, the transit time of the thermal neutrons through the converter foil and the lifetime of the excited Gd atoms after the neutron capture. None of these intervals is longer than some nanoseconds, which allows the construction of a very fast detector. For parallel readout it is necessary to equip each detector diode with the full amplification and readout electronics (more sophisticated readout techniques can reduce the number of channels). For many channels and/or a high position resolution this can only be done by the use of very large-scale integrated electronics (VLSI) [10]. Although the technique used for the detector fabrication is from a general point of view similar to the one used for electronic circuits, it differs in some essential aspects which prevent so far the construction of detector and electronics on the same wafer. 4. Efficiency The probability P (x) that the neutron will be absorbed in the converter between x and x#dx after surviving the distance x is given by the properties of the converter material namely the absorption cross section σ and number of converter atoms per unit volume N P (x)"(1!e ) Nσ dx. (1) For every stopped neutron the probability, that an electron will be emitted is given by the conversion factors of the corresponding (n, γ) reaction. As the decay pattern of Gd is quite complex some simplifications have to be made for the calculations of the detector efficiency. Taking the values of the dominant isotopes Gd and Gd from the nuclear data sheets, we used only the lowest energy levels which have the highest conversion factors, E (kev) "88.97, 199.21, 296.53, 263.58 for Gd and E (kev)"79.51, 181.93, 277.54, 255.66 for Gd [11]. The energy distribution of the conversion electrons was calculated, by subtracting the binding energies of the K, L and M shells. Although each conversion electron is accompanied by several γ-rays as well as X-rays and Auger electrons, which follow the ionization of the atom, only the conversion electrons, and the Auger electrons with an energy of more than 20 kev have been taken into account. To each electron created between x and x#dx, an energy E and a creation probability can be assigned according to the energy distribution, together with an isotropically distributed emission angle. From these values the probability for the electrons to reach the detector volume has been calculated by evaluating the escape probability successively for every layer the electron has to penetrate (Gd-converter, Al-contacts, p-doping) using the following semiempirical equations [12]. The range R in (μm) is a function of the initial
186 G. Bruckner et al./nuclear Instruments and Methods in Physics Research A 424 (1999) 183 189 energy E (kev), the atomic weight A (g/mol), the atomic number Z, and the density ρ (g/cm) R (μm)" 0.0276 ) A ) E (kev). (2) Z ) ρ (g/cm) The residual energy E after penetrating a distance s in a given material can be expressed in terms of the reduced range y"s/r E "(1!y) ) E. (3) The electron transmission in forward direction is also a function of the reduced range and a parameter γ, which accounts for multiple scattering and diffusion losses. I "exp γ ) y I! γ"0.187 ) Z. (4) 1!y For all electrons, which enter the Si detector with a remaining energy of at least 15 kev, the efficiency of the Si detector can be assumed to be 100%. Our calculations show, that the detector efficiency depends critically on the cross section of the element and is therefore a function of the neutron energy (Fig. 5). The extremely high cross section of Gd results in different characteristic of the efficiency in forward and backward direction, as can be seen in Fig. 4. This has to be kept in mind when comparing Gd with other converters as for example Li. The efficiency, defined as the number of registered neutrons divided by the number of incident neutrons, was measured in backward direction for six different wavelengths in comparison to a calibrated He counter. The results for natural Gd and an enriched (90.5%) Gd converter are shown in Fig. 5. Figs. 2 and 3 show the spectra of the reaction products of the neutron capture in Gd and nat. Gd as seen by the Si-planar detector for the measurements made at the ILL. For thermal neutrons we can conclude an efficiency of more than 35% for the enriched Gd converter and about half of this value for natural Gd. One can clearly see the increase towards higher wavelengths. The measured values are constantly higher than the calculated ones. The dashed line in Fig. 5 gives the calculated values, but with a constant offset. From the spectra in Figs. 2 and 3 one can assume, Fig. 4. Calculated detection efficiency for neutrons with 1.8 A as a function of the converter thickness for different converters. Fig. 5. Detector efficiencies in backward direction for different wavelengths, measured at the Atominstitut in Vienna (ATI) and the ILL in Grenoble.
G. Bruckner et al./nuclear Instrument and Methods in Physics Research A 424 (1999) 183 189 187 Fig. 6. Thermal neutron images and histograms for simply formed absorbers as seen by the pad detector. Original size 1616 pads with 300300 μm each, respectively, 4.84.8 mm. that this is caused by X-rays and γ-rays following the neutron capture, which are partly detected. The X-ray lines of Gd can be seen in both spectra, although, it is more dominant for natural Gd. Conversion electrons, γ-rays and X-rays are emitted simultaneously compared to the time resolution of the amplifier of about 1 μs, therefore, one captured neutron can cause only one signal in the detector. Nevertheless, the electromagnetic radiation can increase the overall efficiency in the case when the electron is not detected. Otherwise, the detection of γ-rays and X-rays causes only a spread of the energy spectra. 5. Position sensitive detectors With our first measurements about three years ago, we could demonstrate that a one-dimensional neutron detector can be constructed from a single-sided strip detector [2]. The position resolution achieved with this early system is compared with the results of a pad detector and the latest measurements with a double-sided strip detector in Fig. 7. Fig. 7. Comparison of the position resolution, defined by an Lorentzian edge spread function, for three different detectors. A pad detector offers the advantage of a twodimensional position information on a single-sided detector design [13]. The sensor used for neutron detection has 256 pads with pitch of 300 μm, corresponding to an area of 4.84.8 mm. For the readout we used two VA2 chips, each consisting of 128 channels with preamplifier, shaping amplifier and output multiplexer. The external trigger for this electronic system was provided by processing the backplane signal of the detector with a shorter shaping time (500 ns) than that of the VA2 amplifiers (1.2 μs) [13,14]. Fig. 6 shows the images and histograms of some simply shaped neutron absorbers taken with the pad detector, in comparison with a negative film pattern and a scanner image, respectively. The almost triangular shaped hole was VLSI chip produced by Integrated Detector and Electronics AS (IDE AS), Veritasveien 9, N-1322 Hovik, Norway.
188 G. Bruckner et al./nuclear Instruments and Methods in Physics Research A 424 (1999) 183 189 Fig. 8. Transmission image of a micro switch next to a nut as seen by the double-sided strip detector. Original size 23.4 30.0 mm. prepared by cutting an absorber into two pieces and making a notch in one side. The cut does not shield the neutrons completely, as can be seen in the histogram. As for the previous setup the position resolution was determined by measuring the detector response to an edge. An Lorentzian edge spread function was fitted to the data near the edge, which gives a position resolution of FWHM "210 μm (see Fig. 7). Figs. 8 and 9 show some images obtained with our latest system with neutrons of 2.1 A. It is based on a double-sided strip detector with a pitch of 50 μm. 640512 strips were read out by five plus four 128 channel CMOS VLSI chips developed by IDE AS, which contain a low-noise charge sensitive preamplifier, a shaping amplifier and a discriminator for each channel. The self-triggering capability of this circuit is essential for neutron experiments, where the trigger cannot be provided by an external source. The system was setup in a way, so that only single hits were accepted with a shaping time of 1.2 μs and a threshold of 20 kev. The detector area of 32.0 25.6 mm was only partly covered by the natural Gd converter, resulting in an effective area of 23.430.0 mm. The measurements confirmed, that the distance between Gd converter foil and detector is the critical parameter for the position resolution. To take advantage of the extremely high resolution of the strip detector, the Gd converter has to be placed onto the detector surface or deposited directly on the Si wafer. The position resolution for neutrons. FWMH "95 μm has been obtained by shielding about half of the detector with a Li containing absorber and evaluating the data near to the edge. This value includes not only the variances produced in the detector but also the divergence of the neutron beam, the imperfection of the edge, the scattering of the neutrons at the absorber and geometric misalignments. We therefore expect a further improvement of this value for the near future.
G. Bruckner et al./nuclear Instrument and Methods in Physics Research A 424 (1999) 183 189 189 References [1] B. Feigl, H. Rauch, Nucl. Instr. and Meth. 61 (3) (1968) 349. [2] G. Bruckner, H. Rauch, J. Neutron Res. 4 (1996) 141. [3] C. Petrillo, F. Sacchetti, O. Toker, N.J. Rhodes, Nucl. Instr. and Meth. A 378 (1996) 54. [4] J. Straver et al., Nucl. Instr. and Meth A 348 (1994) 485. [5] K. Borer et al., Nucl. Instr. and Meth A 253 (1987) 548. [6] Proc. 5th Int. Workshop on Vertex Detectors, Nucl. Instr. and Meth A 386 (1) (1997). [7] Proc. 2nd Int. Symp. on Development and Application of Semiconductor Tracking Detectors, Nucl. Instr. and Meth. A 383 (1) (1996). [8] Proc. 7th Symp. on Semiconductor Detectors, Nucl. Instr. and Meth. A 377 (1994). [9] G. Lutz, Nucl. Instr. and Meth 367 (1995) 21. [10] E. Nyga rd et al., Nucl. Instr. and Meth. A 301(1991) 506.P.Aspell et al. Nucl. Instr. and Meth. A 315 (1992) 425. [11] Nuclear Data Sheets 49 (2) (1986) and Nuclear Data Sheets 56 (2) (1989). [12] K. Kanaya, S. Okoyama, J.Phys. D 5 (1972). [13] P. Weilhammer et al., Nucl. Instr. and Meth. A 383 (1996) 89 97. [14] G. Bruckner, H. Rauch, A. Rudge, P. Weilhammer, W. Dulinsky, SPIE 2867 (1996) 554. Fig. 9. Transmission image of a relay, 23.430.0 mm, taken at a wavelength of 2.1 A with the double-sided strip detector. Acknowledgements This work has partly been supported by the EURATOM fusion program.