Development of a Neutron Imaging Sensor using INTPIX4-SOI Pixelated Silicon Devices
Y. Kamiya, T. Miyoshi, H. Iwase, T. Inada, A. Mizushima, Y. Mita, K. Shimazoe, H. Tanaka, I. Kurachi, Y. Arai
DDevelopment of a Neutron Imaging Sensor using INTPIX4-SOI PixelatedSilicon Devices
Y. Kamiya a, ∗ , T. Miyoshi b , H. Iwase c , T. Inada a , A. Mizushima d , Y. Mita d , K. Shimazoe e , H. Tanaka f , I. Kurachi g , Y.Arai b a Department of Physics and International Center for Elementary Particle Physics, The University of Tokyo, Tokyo 113-0033, Japan b Institute of Particle and Nuclear Studies, High Energy Accelerator Research Organization (KEK), Ibaraki 305-0801, Japan c Radiation Science Center, High Energy Accelerator Research Organization (KEK), Ibaraki 305-0801, Japan d Department of Electrical Engineering and Information Systems, The University of Tokyo, Tokyo 113-8656, Japan e Department of Nuclear Engineering and Management, The University of Tokyo, Tokyo 113-0033, Japan f Institute for Integrated Radiation and Nuclear Science, Kyoto University, Osaka 590-0458, Japan g Department of Advanced Accelerator Technologies, High Energy Accelerator Research Organization (KEK), Ibaraki 305-0801, Japan
Abstract
We have developed a neutron imaging sensor based on an INTPIX4-SOI pixelated silicon device. Neutron irradiationtests are performed at several neutron facilities to investigate sensor’s responses for neutrons. Detection e ffi ciency ismeasured to be around 1 .
5% for thermal neutrons. Upper bound of spatial resolution is evaluated to be 4 . ± . µ min terms of a standard deviation of the line spread function. Keywords: imaging sensor, semiconductor, neutron, silicon on insulator
1. Introduction
Imaging sensors based on semiconductor technolo-gies are widely utilized for optical and quantum beamimaging, because of their superior spatial and tempo-ral resolutions, and their high degree of freedom inhandling due to availability of electrical control. Re-cently, several developments of neutron imaging sen-sors with pixelated devices have been reported and areaimed at imaging ultra-cold neutrons (UCNs). The re-ported sensors are the Timepix-based (55 µ m pixel size)with LiF coating [1, 2], Hamamatsu back-illuminatingCCD-based (24 µ m pixel size) with B coating [3],Logitech Webcam CMOS-based (3 µ m pixel size) with LiF and B layers [4], Hamamatsu back-illuminatingCCD-based (14 µ m pixel size) with B coating [5],and the DECam CCD-based (15 µ m pixel size) with B coating [6]. Those technologies are expected tobe useful also for industrial applications such as a non-destructive imaging of light elements [7, 8].Spatial resolution for the Timepix-based and theCCD-based sensors were evaluated to be 2 . µ m in ∗ Corresponding author
Email address: [email protected] (Y.Kamiya) terms of a standard deviation of the point spread func-tion (PSF) [1] and 3 . µ m in the line spread function(LSF) [3], respectively. For the Webcam CMOS-basedsensor, upper bound of the resolution was estimated tobe 60 µ m. These sensors have a detection e ffi ciencyof several tens percents and more for UCNs, however,it is only a few percents or less for cold and thermalneutrons. Therefore, even at the expense of its spatialresolution, neutron sensors with a micro-structured con-verter aiming for higher detection e ffi ciency have beendeveloped [9, 10].In this paper, we report developments of new neu-tron imaging sensors that are based on the INTPIX4-SOI pixelated silicon device [11, 12], aiming at a finespatial resolution with time-resolving readout. In sec-tion 2 the INTPIX4 chip is introduced, followed by theprincipal of operation and sensor design in section 3.Based on measurements from thermal neutron facili-ties, event identification processes using the shapes of acharge cluster are discussed in section 4. Detection e ffi -ciency was estimated under the criteria that we decidedto use. We also evaluated the spatial resolution by fittingedges of shadow of a neutron mask, which has a finestructure with sub-micron accuracy. The detail of themeasurement is presented in section 5. In contrast withour previously developed sensor [3], its faster readout Preprint submitted to Journal of L A TEX Templates June 11, 2020 a r X i v : . [ phy s i c s . i n s - d e t ] J un ime helps us to measure a neutron event with resolu-tion in O (10 − ) seconds or less. The SOI-CMOS archi-tecture with on-chip-circuit-layers is attractive, becauseof its availability of integrations of some rich functionson chip. We are planning to integrate, for example, adedicated self-triggering circuit with a cluster identifi-cation engine. The discussion here will be the first stepin the development of these new sensors.
2. INTPIX4
INTPIX4 is an charge integrating device of 17 µ m × µ m pixels with circuit matrix on a “silicon on insula-tor” (SOI) wafer [13], in which correlated double sam-pling circuits and storage capacitors are implementedfor synchronized operation with external triggers. Fig-ure 1 shows the front-end circuit, which is embeddedon the circuit layer of SOI. The number of pixels is 832 ×
512 and active area is 14 . × . ×
512 pixels and the chargescan be read out parallelly for each block. Readout timeis around 280 ns / pixel, which allows a maximum framerate to be around 100 Hz. Float-Zone (FZ) silicon waferof 300 µ m thickness is used for the sensor’s substrate.Its resistivity is a few kOhm. The circuit gain was evalu-ated to be 13 . µ V / e − [14]. Al layer of 200 nm is coatedon back side of the sensor in the fabrication process. P2P1 P3 P4+V_detV_sense LOADV_ddRST V_RST C_store RST_CDS+V_RST_CDSV_cC_cdsSTORE READ_x COL_OUT
Figure 1: Front-end circuit embedded on a SOI wafer.
The INTPIX4 chip is wire-bonded to a sensor boardand the sensor board is connected to a SEABAS2-board,a general-purpose readout platform, developed in HighEnergy Accelerator Research Organization (KEK) [12].The SEABAS2-board has a Virtex5 FPGA which con-trols the sensor circuit. It also has a Virtex4 FPGA [15]that allows data transfers over the TCP / IP protocol.
3. Fabrication of Neutron-sensitive Sensor
Recently, semiconductors containing Li such as LiInSe or LiInP Se have been reported [16, 17, 18]. Pixelated read-out feature is not implemented inthese materials yet, therefore, the conventional scheme,making a neutron-to-charged-particle conversion layersformed in front of the pixelated devices, is currently thebest choice for achieving neutron-sensitive imaging sen-sors.The following nuclear reactions for the neutron con-version processes are utilized, n + B → α (1 .
47 MeV) + Li (0 .
84 MeV) + γ (0 .
48 MeV) , (1) n + B → α (1 .
78 MeV) + Li (1 .
01 MeV) , (2)where the secondary particles are emitted in back-to-back directions. Branching ratios are 94 % and 6 %,respectively. One of the emitted particles enters the ac-tive volume of the sensor, and stops with correspondingenergy loss rate. O (10 ) electron-hole pairs are generated along thetraveling pass of the secondary particle and the holesstart to drift to the electrodes. In the very early phaseduring the drifting, charges are repulsed each other bythe Coulomb force due to its relatively high chargedensity, and then the natural di ff usion process fol-lows. Because of these di ff usion processes, the chargesare shared by multiple pixels and the cluster’s shapebecomes a symmetrical two-dimensional Gaussian orslightly bell-shaped distribution.Figure 2 shows schematic drawing of the sensor andthe conversion layer in the cross-section view. The Bconversion layer was formed directly on the backsideAl layer using the argon RF sputtering technique. Thethickness of this layer was adjusted to be 200 nm. The B layer was covered with thin Ti film to prevent oxi-dation, and degradation by chipping.We used the Shibaura CFS-4EP-LL sputtering ma-chine at the University of Tokyo, maintained under theframework of the Nanotechnology Platform program inJapan. Sputtering sources are 3-inch disks, located at110 mm distance from the samples. Target and sub-strate are held vertically and side-by-side aligned toavoid contamination from residual particles which arefalling on. There are a pre-vacuum chamber and a loadlock system, which can help us to keep the main cham-ber vacuum even when changing sample configurations.During the sputtering, the sample holder disk rotateswith 20 rpm speed. Sputtering rate for B is 0.03 nm / swith 400 W RF power in 470 mPa Ar, then the B layerof 200 nm thickness is formed in around 2 hours.
4. Neutron Events A B coated INTPIX4 ( B-INTPIX4) sensor wastested with thermal neutrons supplied from the Standard2 eutron charge cluster (10B 200 nm) conversion layersecondary particles (Al 200 nm)(Ti 20 nm) charge diffusion
Figure 2: Cross-section view of the sensor. B and Ti layers were de-posited by physical vapor deposition processes. B nucleus absorbsa neutron and emits charged secondary particles in back-to-back di-rections. One of them deposits its energy in the active volume of theINTPIX4 device. Generated holes drift to pixelated electrodes anddi ff use two-dimensionally to the x-y plane in the process. Thermal Neutron Irradiation Laboratory at KEK. Neu-trons are emitted from a
Am-Be neutron source with37 GBq intensity, which is located in the center of a 1 .
9m (width) × . × . / cm ). Neutron intensity is measuredto be 20 / cm / s at the point of our testing sensors thatis 1.25 m horizontally away from the neutron source.The expected energy distribution is simulated by PHITS[19], as shown in Fig. 3. Totally, 2 . × imageswere taken. I n t e n s it y [ a ub . un it ] Figure 3: Expected energy distribution simulated by PHITS [19].Most of neutrons are in the range of 10 − eV to 10 − eV. Events are selected with the following scheme.1. estimate pedestals of each pixel on each image byaveraging successive 10 images before and afterthe image to be evaluated, and make the pedestalcorrections.2. find the pixel which has higher charge than thethreshold level which corresponds to a 34 keV en-ergy deposit. 3. check the adjacent pixels inside 7 × × / cluster.5. if you find the only one pixel, which have chargehigher than 3.4 keV (10% of the threshold), in theframe, this event is due to the noisy pixel, then re-ject it.The sum of the pixels’ charges in the frame ( Q ) cor-responds to the total energy deposit of the secondaryparticle. The charge weighted mean of the positions ofeach pixel ( (cid:126)µ = ( µ x , µ y )) represents an estimate of neu-tron incident position. To identify neutron events fromelectron-like non-symmetric events or noisy images af-fected by a common noise, we use the 2nd order centermoment on x-axis and y-axis ( (cid:126)σ = ( σ x , σ y )) as addi-tional identifiers. They are written as Q = i = (cid:88) i = q i , (3) (cid:126)µ = Q i = (cid:88) i = q i (cid:126) r i (4) (cid:126)σ k = Q i = (cid:88) i = q i ( (cid:126) r i − (cid:126)µ ) , (5)here q i is a charge of i -th pixel in the frame, and (cid:126) r i = ( x i , y i ) is a pixel position.Fig. 4(a) shows a scatter plot for σ y vs Q (Otheridentifier σ x shows similar distribution to the σ y ). Theprofile of these chained mountains in Q > .
25 MeVrange was evaluated by fitting with a linear function σ y = aQ + b (call it as line L, here after). The bestestimated values were ˆ a = . × − (pixel) / eV andˆ b = .
39 (pixel) Projecting data (again in Q > . σ y distribution around the line L,as shown in the figure 4(b). Here, we determined theevents inside a region | σ y − ˆ aQ − ˆ b | < .
16 as neutron-origin events, in which clusters show well symmetricdistributions. The criterion parameter 0 .
16 is 3 σ of afitted Gausian in the figure 4(b). The typical clustershape for neutron candidate event (around the regionA) is shown in the figure 5(a). Fig. 5(b) shows theevent around the region B, which is expected to rep-resent a traveling electron via Compton scattering withbackground γ -rays.Fig. 4(c) shows a cluster charge distribution insidethe candidate region. Two large peaks show the α and3 y ( p r o j ec t e d )[ p i x e l s ] Region ARegion B0 0.4 0.8 1.2 1.6 e n t r i e s cluster charge Q [MeV] (c)(a)0
500 0 entries−0.40.400.8 line L ( b ) Figure 4: (a) Scatter plot for events in σ y vs Q plane. A dashed line(the Line L) is evaluated to be σ y = ˆ aQ + ˆ b , where ˆ a = . × − (pixel) / eV and ˆ b = .
39 (pixel) , by fitting the distribution in Q > .
25 MeV. (b) Projected σ y distribution in Q > .
25 MeV range alongthe line L. Mean and a standard deviation of the best fit Gaussian are0.39 and 0.053, respectively. (c) Cluster charge Q distribution in thecriteria of neutron events, | σ y − ˆ aQ − ˆ b | < .
16. Dashed histogramshows the best fit distribution with an inactive layer just under theAl skin of the INTPIX4 device. A thickness of the inactive layer isestimated to be 220 ±
20 nm. Li events from the main branch. Right side edges ofthe peaks are shifted from where they are supposed tobe due solely to the released energies during the nu-clear reactions, because the secondary particle travelsthrough an inactive volume with a finite length. Thetails in the left sides represent di ff erences in the passinglengths in the inactive volume, in which the measuredenergy varies with the emission angle with respect tothe normal incidence. We fit it to templates of the en-ergy distribution, which made by the Geant4 simulationframework for the geometry shown in the figure 2, withan additional inactive layer on the backside of the sen-sor. The thickness of the inactive layer was estimated tobe 220 ±
20 nm when one assume that the silicon is adominant material of it. The best fit template is shownin the figure 4(c) as a dashed histogram. Total numberof events measured in this criteria is 4038 for the netmeasurement time of 7.5 hours, which corresponds to adetection e ffi ciency of 1.5 % for the thermal-range neu-trons.
5. Spatial Resolution
Spatial resolution of the B-INTPIX4 sensor wasmeasured by evaluating a neutron’s shadow, which is a (a)(b) pixelspixelspixelspixels c h a r g e [ a r b . un it ] c h a r g e [ a r b . un it ]
565 570 575 580 585232236240244248252050100150200250 165 170 175 180 1854324364404444484520100200300400500600700
Figure 5: (a) Typical shape of cluster due to heavy charged particle ( α or Li). Symmetric cluster is observed. (b) Typical image of travelingelectron in the active volume. These events are expected to be madeby the Compton electrons from the γ -ray background. These clustersshow relatively larger σ y values. projection of the neutron mask made by well-collimatedneutron beams, supplied at the BL10 beam line ofthe Materials and Life Science Experimental Facility(MLF), J-PARC center.Fig. 6(a) shows schematic drawing of the BL10 beamline. Two B C slits are located at 7.050 m and 12.755m distances from a surface of a moderator of the MLF,respectively. We set their opening apertures to be 10mm squares and selected a small divergence beam ofless than 0.1 mrad. Pb neutron filter with 75 mm thick-ness was inserted in the beam line to reduce fast neu-tron components. Measured transmission of the filter isfound in elsewhere [20].The sensor and the neutron mask is located around20 cm downstream from the second slits. Fig. 6(b)is a close view of the sensor position. The neutronmask has a fine structure of the line / space arrangementwith 100 µ m pitch which is made by chemical etch-ing processes. Depth of the trenches were 100 µ m andsilicone rubber containing Gd was poured into them,in which the neutron beams are partially stopped andmakes the shadow on the detector. Distance betweensurfaces of the neutron mask and the detector was set tobe 200 ± µ m.4 layerwith Gdsilicone rubber leveling mount (L = 7.050 m) slits 1 (L = 12.755 m) slits 2neutron guide (11.2 cm square) concrete shieldneutron modelatorMLF BL10 (NOBORU), J−PARC neutrons ~ 20 cm (b) neutron mask and sensor(a) B−INTPIX4 neutron mask Figure 6: (a) Experimental setup for the measurements of spatial res-olutions at the BL10 beam line. Two B C collimators (thicknessis 5 mm) were set to be 10 mm apertures in vertical and horizon-tal. (b) Closed view of the neutron mask and the B-INTPIX4 sen-sor. Distance of the silicone rubber and the B layer was set to be200 ± µ m. p i x e l s e n t r i e s pixels (b)masked region (a) Figure 7: (a) Image of the all area of the sensor (density of dots isreduced for visualization). Edges of the B C slits and shadow of thefine neutron mask are observed. (b) Profile of an accumulated edgeshadow. It is made by superimposing the all edge shadows of theneutron mask. Spatial resolution is evaluated to be 4 . ± . µ m, byfitting with an error function. Figure 7(a) shows a measured image (density of dotsare reduced). The vertical lines shown at x =
160 and x =
790 pixels are due to the beam ends shaped by theslits. The shadow of the fine neutron mask is imagedaround the center. We chose a region of 180 < x < < y <
400 for the following analysis. The im-ages in this area of interest is projected to a vertical axisand all edge images are superimposed according to therepeated pattern. Then the histogram of accumulatededges is fitted by an error function A erf( t ) + B , where t = y − y √ σ LS F (6)with fitting parameters, A , B , y , and σ LS F (the spatialresolution of interest). The line / space structure is notplaced perfectly parallel to the projection direction (hor- izontal direction in this analysis), therefore, we rotatethe measured image in the x-y plane before the projec-tion, and find the minimum σ LS F as a function of the ro-tation angle. Figure 7(b) shows the accumulated shadowedge, with the best rotation angle. We estimate that theupper bound of spatial resolution is to be 4 . ± . µ m,here the uncertainty indicates the fitting error.
6. Summary
We have succeeded to measure neutron events by the B-INTPIX4 new neutron imaging sensor. γ -ray back-ground in low energy can be rejected by the clustercharge distribution. Electron events from Compton scat-tering processes of the higher energy γ -rays (and cosmicmuon events) are able to be rejected by the additionalidentifier, the 2nd order center moment e ff ectively. De-tection e ffi ciency with the event criteria is evaluated tobe 1.5% for thermal neutrons. Upper bound of the spa-tial resolution is estimated to be 4 . ± . µ m as a stan-dard deviation of the line spread function. Acknowledgements
We wish to thank Prof. Takeshi Go Tsuru of KyotoUniversity and all of the members of the SOI group formany supports in the sensor development. Especially,we are indebted to Dr. Ryutaro Nishimura of KEK forcomprehensive developments of several tools for read-out systems for INTPIX4. We are deeply grateful to themembers of the Takeda Sentanchi super clean room ofthe Nanotechnology Platform Program for maintainingthe laboratory.This work was partially supported by JSPS KAK-ENHI Grant Number 17H05397, 18H04343, and18H01226. A part of this work was conducted atTakeda Sentanchi Supercleanroom, The University ofTokyo, supported by “Nanotechnology Platform Pro-gram” of the Ministry of Education, Culture, Sports,Science and Technology (MEXT), Japan, Grant Num-ber JPMXP09F19UT0065.
References [1] J. Jakubek, P. Schmidt-Wellenburg, P. Geltenbort, M. Platke-vic, C. Plonka-Spehr, J. Solc, T. Soldner, A coated pixel deviceTimePix with micron spatial resolution for UCN detection, NIMA 600 (2009) 651–656.[2] J. Jakubek, M. Platkevic, P. Schmidt-Wellenburg, P. Geltenbort,C. Plonka-Spehr, M. Daum, Position-sensitive spectroscopy ofultra-cold neutrons with Timepix pixel detector, NIM A 607(2009) 45–47.