Highspatialresolutionneutronsensingmicrochannelplatedetectors.pdf

nuclear instruments and methods in physics research a 576 (2007) 178182 high spatial resolution neutron sensing microchannel plate detectors oswald h. siegmunda,?, john v. vallergaa, anton s. tremsina, jason mcphatea, bruce fellerb aspace sciences laboratory, university of california, 7 gauss way, berkeley, ca 94720, usa b nova scientifi c inc., 10 picker road, sturbridge, ma 01566, usa abstract neutron sensing microchannel plates (mcps) developed by nova scientifi c inc. incorporate high-effi ciency neutron conversion materials in the mcp glass to provide a high neutron stopping power. we have developed and tested neutron event counting 2d-imaging detectors using these mcps with three different imaging systems, a cross delay line readout, a cross strip readout, and a medipix readout. tests at mnrc and nist with the cross delay line and cross strip readouts have established spatial resolution with neutrons as good as 17mm rms over 27mm, with event rates of 1mhz, event time tagging of 25ns and intrinsic background rates of o0.05 events cm?2s?1. medipix readouts for neutron imaging have poorer spatial resolution (55mm) and event timing, but are capable of even higher rates (ghz). r 2007 elsevier b.v. all rights reserved. pacs: 85.60.gz; 29.40.?n; 95.55.vj keywords: microchannel plate; medipix; cross delay line anode; neutron; imaging 1. introduction 10b and157gd have high interaction cross-sections for low-energy neutrons and thus have potential use in neutron detectors. microchannel plates (mcps) have been devel- oped by nova scientifi c inc. that incorporate these high- effi ciency neutron conversion materials into the mcp glass. their incorporation into mcp glass is a convenient way to make mcps effi cient for neutron counting with high spatial resolution detectors 13. neutron interaction with 10b produces7li (1.01mev) and a recoil alpha particle (1.77mev) (6% probability), or 7li (0.84mev) an alpha particle (1.46mev) and emitted photons (0.48mev) (94% probability). the alpha particles of both interaction channels have a high probability of initiating an electron avalanche and have a short range (?3.5mm 2) in the mcp glass. in radiative capture of neutrons, which is the dominant neutron interaction with gadolinium, gamma rays, x-rays and fast electrons are produced 4. the electrons also have a high probability of initiating an electron avalanche in the mcp. since the range of the alpha particle and electrons are less than 10 and 50mm, respectively in the mcp glass, the neutron events can be counted and imaged with a mcp detector with a two- dimensional readout system (figs. 1 and 2). we have used 10b, andnatgd-doped mcps in detectors with cross delay line 1, cross strip 1, and medipix 6 readouts to image neutrons with high spatial resolution at the mcclellan nuclear radiation center (mnrc) and the national center for neutron research (ncnr) at nist. typical detectors for neutron imaging include scintillator screens observed by imaging cameras 5, solid state devices coated with neutron conversion materials 6, and neutron sensitive solid state devices 7. the advent of neutron sensitive mcps allows us to utilize the extensive develop- ment history of this type of device to achieve excellent spatial and temporal resolving performance that is not usually obtained with other schemes. two types of neutron sensing mcps have been tested. the gadolinium-doped mcps have 8.5mm wide pores on 11mm centers giving a minimum wall thickness of 2.5mm, and a length to diameter (l/d) ratio of 75:1. the boron-doped mcps also have 8.5mm pores on 11mm centers. but a l/d ratio of article in press 0168-9002/$-see front matter r 2007 elsevier b.v. all rights reserved. doi:10.1016/j.nima.2007.01.148 ?corresponding author. tel.: fax: e-mail address: (o.h. siegmund). 100:1. both have a 33mm diameter with an active area of 26mm. 2. mcp neutron detector with medipix2 readout themedipix2isapixeldetector readout chip consisting of 256?256 identical 55?55-mm pixel elements, each working in single-photon counting mode 8. each pixel consists of a preamplifi er, a windowed discriminator, and a 14-bit pseudo-random counter. electron clouds from the mcp impinge onto the medipix pixel input pads and are counted as single events if the collected charge is above a (user selectable) threshold, typically 12 thousand electrons. single pixel rates can approach 100khz while a full array can handle ?1ghz count rate. the total active area of the chip is 14?14mm. readout of the medipix2 chip may be performed via either a serial or a parallel interface. both serial and parallel interfaces are clocked by a fast external clock (on the order of 100mhz). this fast readout is possible as the data are digital, unlike a charge integrating ccd type readout. a 100mhz clock speed results in a parallel full frame read time of 286ms which can support frame rates of 1khz. the advantage of the massively parallel medipix2 readout versus cross delayline or cross strip anodes is its ability to handle the high fl uxes that can be encountered in neutron tomography studies. not only is every pixel fully active, but also the preamps are very low noise (o 100e-) allowing low gain operation of the mcps. what are sacrifi ced in using the fi xed 55mm pixel size medipix2 readout are the spatial resolution and the sub-nanosecond single event timing resolution. we tested the medipix2 readouts with the natgd- doped mcps using a demountable test detector (fig. 1) fabricated for our optically sensitive imaging tube devel- opment for adaptive optics 9. we used two gd mcps back-to-backand operated atsingleeventgains of ?50,000. the detector, mounted on an 8in.0stainless steel fl ange attached to a portable vacuum chamber. 3. cross delay line detector a cross delay line detector was also used for neutron imaging, comprising a stack of three mcps positioned ?6mm in front of a cross delay line anode (fig. 2). detected neutrons are converted in the fi rst mcp to an electron avalanche giving overall event gain of a few 106e?. the cross delay line (xdl) anode consists of an upper and a lower serpentine pattern in orthogonal directions. the mcp fast charge pulse (?2ns) is divided between two conductive traces in the charge collection area. the difference in arrival times of the fast pulses at the opposite ends of each delay line allow the centroid position of the charge cloud to be measured, giving the x- and y-centroid positions of the event. this is accomplished by amplifi ca- tion of the signals followed by a start-stop time to amplitude/digital conversion (tdc) for each axis. the amplifi ers were mounted close to the detector, while the tdc nim module was outside the neutron beam room. the tdc transmits x,y,t data to a national instruments dio-32hs i/o card in a pc, recording the event data. resolution tests with a pinhole mask (10mm holes) showed spatial resolution of o9mm rms. the electronics also provides a time tag for each detected event with an article in press fig. 2. schematic of a mcp stack placed in front of a cross fi nger anode. the electron cloud resulting from amplifi cation of a neutron detection is spread over several periods of the anode pattern. resulting signals can be used to derive the centroid position of the original event. fig. 1. medipix readout chip mounted behind a mcp detector housing, showing part of the bias electronics. o.h. siegmund et al. / nuclear instruments and methods in physics research a 576 (2007) 178182179 accuracy of 25ns. the electronics dead time is about 450ns, allowing event rates of ?1mhz to be reached with acceptable dead time losses (?30%). neutron sensing mcps were used in a detector with a 32mm xdl anode. neutrons are incident on only a 12mm circle in center of image, with most of the background due to gamma ray fl ux. the dark shadow images are areas of lower gain due to prior high-fl uence tests causing reduction of gain in the last mcp of the stack (which would have been avoided had the stack been stabilized 10). the circular active area is 27mm diameter using a xdl anode. 4. neutron imaging tests imaging tests of the delay line and cross strip detectors were done at mnrc and ncnr facilities. to measure resolution cadmium masks were placed directly onto the top mcp. the cd masks provide an effective block for low-energy neutrons, regions thus masked will mostly contain only transmitted or generated gamma radiation. the achievable gain is in excess of 107for a gadolinium or boron-doped mcp when they are used as the top mcp in a stack with two glass mcps (l/d 80:1, 10mm pores). the background rate for a stack of three gadolinium-doped mcps was found to be o0.05 events cm?2s?1, which is as good as the lowest noise glass mcps 11. both the cross delay line and cross strip anode systems provide the pulse height (amplitude) of each detected event from the mcps. this signal is proportional to the total amplifi ed charge of each event detected by the mcps. the deeper the initial interaction is within the mcp stack, the less gain provided by the microchannels. uv events are detected at the mcp stack entrance whereas gammas are detected throughout the bulk of the mcp. thus, 60co gamma rays gave a negative exponential pulse amplitude distribution, compared with a peaked distribution for uv. the image (fig. 3) obtained with a boron-doped mcp with a star pattern cd mask has good contrast and resolution. other features seen include a 12mm wide bright area that corresponds with the hole in a li-doped blocker placed in front of the detector. the pulse height distribu- tion for neutron and gamma events within the star pattern is peaked (fig. 4). however, the gamma background where neutrons are completely stopped with the li-doped blocker has a negative exponential distribution. the peaked pulse amplitude spectrum for boron-doped mcps with an additional higher amplitude lobe is a result of the detection of alpha particles produced by the neutron interactions. outside the 12mm area the background is dominated by detection of incident gamma rays produced from the reactor and materials around the detector. inside the neutron illuminated circle the background under the cd mask is slightly higher (fig. 5) due to detection of betas and gammas generated by neutrons interacting with the cd mask. background from gammas produced by neutron interactions in the mcps is comparatively small due to the low (?2% 12) gamma detection effi ciency of mcps. dark image features in fig. 3 are burn in marks from other (8arm) mask pattern exposures, that are avoidable by stabilizing the mcps 10. the effi ciency of neutron detection for the gadolinium and boron mcps has been estimated from data taken on the bt6 beam at ncnr 13. the events for a section in the center of a star pattern (8arm) image were counted, article in press fig. 3. neutron image of a 6arm star mask pattern using a boron-doped mcp above 2 standard mcps (l/d 80:1, 10mm pores). the neutron mask is cut into a 1-mm-thick cd foil with 300mm wide arms and radial length of 6.4mm. 0 500 1000 1500 2000 2500 01 1061.5 1062 1062.5 1063 106 gamma background neutrons and gammas counts pulse amplitude (e) 5 105 fig. 4. pulse amplitude spectra for gamma ray background with the detector shielded by a li-doped blocking plate, and the direct neutron and gamma ray fl ux detected by a mcp stack comprising one boron and two conventional mcps. o.h. siegmund et al. / nuclear instruments and methods in physics research a 576 (2007) 178182180 and the event count in a neighboring equivalent area under the cadmium mask were subtracted. to estimate the overall neutron detection effi ciency resulting number was com- pared to the expected beam neutron fl ux. for the gadolinium-doped mcps the effi ciency is ?18%, and for the boron-doped mcps the estimated effi ciency is ?21%. the spatial resolution has been estimated by measuring the sharpness of the edges of the star pattern arm images. the dominant factor determining the spatial resolution is neutrons intersecting several neighboring mcp pores due to the geometrical effect of the 81 pore bias angle. high- resolution histogram cuts (fig. 6) of an 8arm star pattern were assessed by fi tting the bar image edge to determine spatial resolution. on the ncnr ng6-m beam we fi nd values as good as 17mm rms. further tests with a cross strip detector 1 with o5mm rms resolution achieved neutron imaging spatial resolution of ?15.5mm rms. this is in accord with our expectations of a resolution limit of 23 mcp pores for the geometry that we have used. medipix imaging tests were accomplished at mnrc. we mounted a cd mask with rectangular slits 0.630mm width on 1.26mm centers on the outside the vacuum chamber approximately 4cm in front of the detector. because of the materials surrounding detector the gamma ray background was quite substantial. however, we were able to see the mask pattern (fig. 7) illuminated with neutrons even with this non-optimal test setup. with the neutron sensing capability of natgd/b-doped mcps the medipix readout performance (high-input fl ux, fast readout, 55mm pixel resolution) should be fully achievable. acknowledgments the contributions of paul and brian white at nova scientifi c are acknowledged. nova scientifi c received us government support through contract dasg60-98-c- 004. the medipix readout development was supported by nsf under aura # ast-0132798-spo#6 (ast- 0336888). we also thank dr. a. muhammad, dr. d. hussey, and dr. d. jakobson of nist for their support of the tests at ncnr. references 1 o.h.w. siegmund, spie 4854 (2003) 181. 2 a.s. tremsin, w.b. feller, r.g. downing, nucl. instr. and meth. a 539 (12) (2005) 278. 3 a.s. tremsin, w.b. feller, r.g. downing, d.f.r. mildner, ieee trans. nucl. sci. ns-52 (n5) (2005) 1739. 4 g.w. fraser, j.f. pearson, nucl. instr. and meth. a 293 (1990) 569. 5 b.schillinger,h.abele,j.brunner,g.frei,r.gahler, a. gildemeister, a. hillenbach, e. lehmann, p. vontobel, nucl. instr. and meth. a 542 (2005) 142. article in press 40 20 intensity (relative) 0900 position (pixels) fig