Assembling and Testing a Neutron Detector

Assembling and Testing a Neutron Detector

Dominik Wermus

Office of Science, Science Undergraduate Laboratory Internship (SULI)

Virginia Military Institute

Thomas Jefferson National Accelerator Facility

Newport News, Virginia

July 31, 2009

Prepared in partial fulfillment of the requirements of the Office of Science, Department of Energy’s Science Undergraduate Laboratory Internship under the direction of Dr. Douglas W. Higinbotham in the Hall A Division at Thomas Jefferson National Accelerator Facility.

Participant: ______

Signature

Research Advisor:______

Signature

Assembling and Testing a Neutron Detector. DOMINIK WERMUS (Virginia Military Institute, Lexington, VA 24450), DOUG HIGINBOTHAM (Thomas Jefferson National Accelerator Facility, Newport News, VA, 23606).

ABSTRACT

Detecting neutrons is of high interest in experiments at Thomas Jefferson National Accelerator Facility. Scientists use neutron detectors made of rectangular bars of scintillating plastic with PMTs (photomultiplier tubes) attached to each end. When charged particles pass through, scintillator plastic releases photons, which activate the light-sensitive PMTs. Neutrons have a small but known probability of colliding with nuclei in the plastic and releasing protons, which in turn produce light and are detected. A two-inch wall of lead in front blocks out nearly all charged particles, making the array of scintillator bars a neutron detector. For several upcoming experiments in Hall A, scientists wished to add twenty-four one-meter-long bars to an existing detector. There were three stages to the project: physically assembling the bars, setting up the data acquisition system, and testing the detector with cosmic rays (high-energy particles from deep space). PMTs were taken from old bars, then were cleaned, tested and glued onto new scintillator bars. A data acquisition system composed of an ADC (analog-to-digital converter) and TDC (time-to-digital converter) was assembled, allowing for calibration and testing with cosmic rays. The energy deposited by the passage of cosmic rays was used to set the PMT high voltages. Once set, the known cosmic ray rate of 100 rays/m2·s was verified. These detectors will be used for experiments such as E07-006 (Studying Short-Range Correlations via the Triple Coincidence (e, e’pn) Reaction) for detecting neutrons released from a correlated pair of nucleons.

INTRODUCTION

Thomas Jefferson National Accelerator Facility’s electron beam can reach energies of up to 6 GeV, easily knocking out nucleons from targets. Detecting neutrons has been of high interest in many experiments, including past and future short-range correlation experiments [1-3]. Scientists at Jefferson Lab’s Hall A wanted to be able to detect neutrons more efficiently, so they decided to expand the current detector, HAND (Hall A Neutron Detector) by adding two layers of scintillating plastic bars. These bars could either be added behind the existing detector or be given its own trigger layer and act as a separate detector.

Scintillating plastic gives off a constant glow: this is because charged particles passing through the material cause the electrons to emit photons in the visible spectrum. The more energetic the charged particle, the more photons are released. A particle detector can be made as simply as gluing a light detector to a bar of scintillator and wrapping it in masking tape (blocking out natural light).Specifically, the HAND is made up of meter-long rectangular bars covered in black masking tape with one photomultiplier tube (PMT) attached to each end. There are eighty-eight bars in total, standing tall in five rows back-to-back, with a “veto” layer making the front (Figure 1).

Since neutrons have no net charge, they are not explicitly detected by scintillators. Instead, a neutron has a small probability of hitting a nucleus and releasing a proton, which then releases photons in the plastic. 10 cm-thick bars will detect about 10% of the neutrons passing through. Two bars back-to-back will detect another 10% of that (19% total), and so on. Unwanted charged particles such as protons and electrons must be filtered out; this is done with a thin layer of “veto” bars which have very little chance of detecting neutrons, but almost certainlydetect charged particles.Signals given in the veto layer areused to remove charged particle events off-line. A two-inch lead wall in the hall can also be used to filter out charged particles before they ever reach the detector.

A neutron detector is simply a charged-particle detector that uses electronics and shielding to filter out unwanted particles. Neutrons release charged particles through nuclear reactions inside the detector itself. The neutron detector was assembled with this approach. There were three phases to the project: physically assembling the bars, hooking up the data acquisition system (DAQ) and hardware systems, and performing the cosmic ray tests.

PHYSICAL ASSEMBLY

The bars,ordered from Kent State University, are made of 100 cm long, 10 x 25 cm rectangular clear scintillating plastic (Figure 2).Two triangular “light guides” are fixed at each end, narrowed to allow the round 5” diameter PMTs to be attached. The entire body of the bar is wrapped in black masking tape toshield out natural light. Twenty-four new bars were ordered, so forty-eight PMTs would be needed.

In Hall A’s storage shack there contained an old detecting array whose scintillator bars had yellowed from use. These PMTs, however, were still good, and as the type used (Photonis XP4578/B) cost around $3000 each, it was budget-wise to reuse them. A total of forty-two were salvaged (saving $136,000), removed by cutting through the silicon gluewhile being careful not to cause any shock or heavy vibrations that would break the sensitive vacuum-sealed chambers. The PMTs were returned to the Test Lab, where the surface of each PMT was cleaned with alcohol, removingany smudges which would block out light.

The PMTlooks and behaves like a reverse flashlight – when a single photon makes contact with its thin metal surface, a single electron is released, which then makes contact with other cathodes, causing a cascade of more and more electrons until a noticeable signal is made. PMTs are extremely sensitive – with one photon releasing more than a million electrons,they are overwhelmed if exposed to normal light.

The PMT has two parts: the vacuum-sealed, 5”-diameter glass-surface photon receptor and the electronic base attached to the other end. The base has two cable plugs: one high-voltage for powering the device and one low-voltage for delivering a signal to a computer. After being cleaned, each individual PMT was hooked up within a“blackbox” – a modified toolbox made to seal out all natural light when shut.The PMT was connected inside the box, and cables connected from the box to a power source and an oscilloscope (Figure 3). The power source delivered well over -2000 volts, though the voltage to each PMT was later adjusted. The oscilloscope gave a visual reading of the PMT’s output signal.

The PMTs were first turned on and assessed for “noise” – if not receiving any light, the PMT should produce only a very small signal coming from thermal electrons within. Good PMTs would fluctuate in the five millivolt region, while others would give outputs in the dozens of millivolts region, rendering them unusable. However, this often signaled a mere problem with the PMT’s base rather than the complex cavity.Bases only cost around $300, so if the vacuum cavity still worked (checked by switching the noisy base with a good base), it still saved a great deal of money. Of the forty-two PMTs, thirty-seven had good bases, and five needed replacement, which were taken from other sparesat the lab. One Photonis XP4572/B was used, and five more EMI 9823KB PMT bases were ordered, to meet the total forty-eight needed.

In testing each PMT, a small, rectangular piece of scintillating plastic was pasted onto the surface with a clear, sticky, non-drying glue. The PMT-scintillator combination was placed again in the blackbox, powered and attached to an oscilloscope. Though the blackbox would shut out natural light and charge, the scintillator would still give a faint amount of light from the cosmic rays passing through.Cosmic rays are highly energetic particles which come from deep space and pass through the Earth at a constant rate. Primary cosmic rays are usually highly relativistic protons. These collide with cosmic dust or ozone in the atmosphere, breaking up into secondary cosmic rays, usually relativistic mesons and other exotic particles. Approximately 75% of what finally reaches the Earth’s surface are muons, or heavy electrons. Through any surface, it is expected that one hundred of these secondary cosmic rays pass through a square meter every second [4]. For this small piece of plastic, about 8 x 20 centimeters and half a centimeter thin, we expected to get about, which corresponded to what was observed on the oscilloscope – about three hits for every two seconds, oftenunsteadily. A “hit” could be seen as a strong spike in negative voltage, happening in the course of only a few nanoseconds.

Once the PMTs were determined ready for use, the next step was to attach them to the bars. A clear adhesive called Elastosil was used to glue the surface of each PMT to the unwrapped end of each bar. The chemical comes in as two parts, Elastosil A and B, and must be combined prior to use. Elastosil is a silicon glue, useful because it is both hard to break physically, yet easy to break chemically (with most solvents). The compound is first mixed (90% part A, 10% part B), with 200 grams enough to use for four or five PMTs. Air bubbles from the mixture are removed using a blow drier,and thena thin layer is applied to the surface of the bar.

Because the glue takes twenty-four hours to dry, the bars needed to be stood upright and PMTs balanced on top (Figure 4). Five bars at a time were strapped upright to an A-frame and balanced, the bottoms held in unique stands given by Kent State University for this process. The glue was applied, and each PMT was attached and taped on sides to the bar for extra support. This was a delicate process, as the bars weigh close to one hundred pounds each and the PMTs can easily shatter if a moderate force acts on it. Harder still was placing the second PMT on the bar’s other end – the bars had to be suspended an extra foot off the ground to keep the PMT on the bottom from touching anything.

A bar was considered complete when it had two PMTs attached, and all its remaining exposed areas were wrapped in at least two layers of masking tape. Each bar was then inspected for light leaks (holes in the tape which would allow natural light in) by having each PMT connected and the bar covered with a black wool blanket. Parts of the blanket would be lifted, and if a signal began to appear, a flashlight would be used to pinpoint the specific area where the leak was present. Once all the bars were tested, they were neatly stacked in a way that the PMTs would not touch each other.

In the end, there were nineteen complete bars. Five more had EMI 9823KB PMTs but without bases, and could not yet be tested. Five complete bars were then stacked together to mimic the final detector product. Four bars were laid flat side-by side, and another was placed on top, intersecting them(Figure 5). Their ten PMTs were connected to a high voltage power supply and given negative 2000 volts each. The low-voltage data cableswere connected to the rest of the data acquisition system (DAQ), which is described in detail below.

SETTING UP THE DAQ

The DAQ was composed of two parts: the analog-to-digital converter (ADC), which measured the strength of the signal, and time-to-digital converter (TDC), which measured when the signal occurred.Because a constant collection of events would overwhelm the computer, windows of time or “gates” were used to determine when data would be recorded. With four bars on the floor and one spread on top, the top bar acted as a “trigger”. From above, a cosmic ray would go through the trigger,starting a timing “gate” for data to be recorded in the other bars. In a matter of nanoseconds, the ray would continue, activatethe other bars it passed through, and a signal would be recorded and sent to the computer. By taking the timing and strength of the signal of all the PMTs on two bars, the path of the cosmic ray could be determined. With a large collection of data, both the rate and angular distribution of cosmic rays passing through the Earth could be calculated.

A window of about 200 nanoseconds was needed to record the energy deposited by a cosmic ray. First, when a ray would first pass through the trigger bar, the trigger PMTs would activate, sending their two signals to a coincidence unit. If they occurred simultaneously, the unit would activate and send a signal to a gate generator, which started the gate. One copy of this signal was sent through about three and a half-meters of delay cables (a signal travels through cable at 17 centimeters per nanosecond), back to the gate generator which closed the gate, and another to a timing device explained below (Figure 6).

Therewere ten channels for the ten different PMTs. Each PMT had a cable which sent a signal to a 10x amplifier, allowing for better distinguishing of the different signals, and then to adiscriminator. The discriminator was set to accept signals only above a certain voltage, blocking noise and low-energy particles. If accepted, two copies were made: the first signals went to the computer as the ADCs. The second signals went to a timing device which output the TDCs. The ADC signal was the integral of the voltage from the opening to the closing of the timing gate, with a reference voltage subtracted to reduce the effects of noise. The TDCs were the time the signal arrived minus the time the gate began.

The computer was able to collect cosmic ray readings with CODA software and process them using ROOT. CODA (Common DAQ) is a software system developed at Jefferson Lab for the collection of data to be used in conjunction with ROOT. ROOT was developed at CERN by physicists working with high-energy experiments where the number of events in one run can be in the millions, and with hundreds of runs in one experiment. ROOT is useful for its processing power and graphing capabilities [5].

TESTING WITH COSMIC RAYS

With the setup complete, data was nearly ready to be recorded. However, because each PMT’s base is slightly unique, each PMT would need a slightly different “gain”, or amount of power to give the same reading for the same energy particle.Quickruns were taken to check the ADCs coming from each PMT. If the energies detected were too high, the power going to the PMT would be lowered, and vice-versa. The gains on the PMTs were adjusted so they would all give readings in the same range (Table 1).

Charged particles lose energy in the plastic at 2 MeV/cm. With most of the cosmic rays coming from the atmosphere above, there would be a mean energy deposit of 50 MeV per cosmic ray, corresponding to about 200 mV in a signal. To evaluate the rate of events detected, the discriminator was tested on a range of settings (Figure 7). The rate of events should have been, reduced to 20 hits/s since not all events would be detected by the PMTs. This corresponded to a setting of 150 mV.

The final phase of testing would be to measure the paths of the cosmic rays.A new arrangement was set up: five bars were arranged so that one trigger bar laid flat on top of two stacks of two bars. Large recordings were taken (two to three million events), and an evaluation program written in ROOT displayed ADC and TDC data from each PMT in histograms (Figure 8). The difference in TDCs on a single bar could give the point of entry of the cosmic ray (Figure 9). By comparing two bars, the complete path (including angle) of a cosmic ray could be determined. Total spatial and angular distribution of cosmic rays was calculated by adding all of these events (Figure 10).

By examining the TDC data, there seemed to be a strong bias towards one side on each of the bars. The reason for this is still undetermined – either there is some source giving off strong radiation (which in the Test Lab is very possible), or the gains on the PMTs are imbalanced. Either way, based on these results, the bars were determined ready for use. The detector will be used for experiments such as E07-006 (Studying Short Range Correlations via the Triple Coincidence (e, e’pn) Reaction), where an electron beam scatters paired nucleons from a target. For the future, the bars still need to be arrayed in a metal rack just as with the original detector. Further, a “veto” layer may still be assembled if this detector is to be used independently from the original. A 2-inch lead wall will absorb almost all charged particles, but a thin layer of scintillator poorly absorbs neutrons and can signal “false hits” for unwanted particles.