Optical Calibration

The AARF System

The acrylic and aluminium reflector and fibre-optics (AARF) system is used to inject light into the detector during dedicated runs over the lifetime of the detector. The AARF system assembly is shown in Figure 2.20. Light from a 435 nm LED is directed along

2.7. CALIBRATION SYSTEMS CHAPTER 2. THE DEAP-3600 DETECTOR

Figure 2.20: A rendering showing the AARF system installed on a light guide. Large arrows in red show an example of a possible light path from the AARF acrylic stub, reflected from the PMT and down the light guide towards the AV. Rendered by Koby Dering.

fibre optic cables into 20 light guides uniformly distributed about the detector, and into 2 opposite sides of the neck. The fibre is bonded to an acrylic stub, which is bonded to the acrylic. In the case of a light guide 80% of LED light is observed in the AARF PMT, and the remaining 20% is reflected from the PMT glass into the detector.

The LED is driven by a pulse generator which pulses at a rate of 1 kHz. The data acquisition trigger module is synchronised with the pulse generator to ensure that data is read out as light pulses are generated and that each light pulse is a separate event. Only PMTs pulses that are observed between -24 ns and +44 ns relative to the AARF trigger are accepted.

The intensity of the AARF is described using the occupancy in the PMTs which are non-adjacent to the AARF light guide. The occupancy of a PMT is defined as the fraction of events for which 1 or more photoelectron is observed in that PMT. The AARF was operated at intensities which correspond to mean occupancies of 5% and 15% in non- AARF PMTs.

Figure 2.21: Prompt occupancy vs PMT angle from the AARF PMT. Occupancy is calculated as fraction of events for which a PMT sees 1 or more PE. Prompt occupancy only accepts PE from pulses detected within -24 ns and +44 ns of the DAQ trigger. PMTs are sorted in order or ascending angle to the AARF PMT. Two PMTs are disabled due to problems at time of data taking. Plot prepared by Berta Beltran.

a function of the angle between that PMT and the AARF PMT. Occupancy is observed to decrease in PMTs at higher angles away from the AARF PMT, reaching an approximately constant 5%. In 5% occupancy runs, on average only a single photoelectron per hit PMT is observed in PMTs furthest from the AARF.

Single PE Charge Calibration

The AARFs are used to record and model the single PE charge distributions of the PMTs, as described by the collaboration in Ref. [132]. The result of this calibration is sum- marised as follows.

PE multiplication at each dynode in the chain is a Poisson process. PE production from the first dynode is a sequence of Poisson processes with fluctuating rates, due to incomplete collection and multiplication of the primary PE produced by the photocathode. A sequence of Poisson processes is described by a Polya distribution, which approaches a Gamma distribution for many produced PE.

An example single PE charge spectrum as measured using the AARFs is shown in Figure 2.22. The SPE charge distribution is obtained by fitting to the data using the sum

2.7. CALIBRATION SYSTEMS CHAPTER 2. THE DEAP-3600 DETECTOR 0 10 20 30 40 50 1 10 2 10 3 10 4 10 PMTID 0_Data Full fit Pedestal 1 PE contribution 2 PE contribution 3 PE contribution Charge [pC] 0 10 20 30 40 50 χ 4 −2 −0 2 4

Figure 2.22: A single PE (SPE) charge spectrum in pC for pulses within a 68 ns window. Vertical axis shows the number of pulses observed with a charge on the horizontal axis. Produced using the AARF on a PMT on the top ring of PMTS nearest the neck. Full waveform data was taken (without ZLE) with 15% occupancy in the AARF PMT. The fitted function, shown in blue, is described by equations 2.3-2.5. The pedestal Gaussian (grey dotted), single (green dotted) and multiple (pink and purple) PE contributions to the fit are shown under data (black with error bars). Shown below is the difference in between function and data. The function fits the data with a χ2 per number of degrees of freedom at ∼1. Plot prepared by T. Pollmann, reproduced from Ref. [132].

of two Polya distributions and an exponential term. The first Polya distribution models charge produced by the primary PE reaching the first dynode. The second models charge due to the primary PE from the photocathode reaching the second dynode and produc- ing incomplete electron multiplication. An exponential term describes the photoelectron scattering on a dynode multiple times, such as in the double pulse as explained in Section 2.3.6. The total single PE charge distribution model is given by:

SPE(q) = η1Gamma(q; µ, b) + η2Gamma(q; µ fµ, b fb) +

     η3le−ql for (q< µ) 0 for (q> µ) (2.3)

where η describes the amplitude of each component such that the distribution is nor- malised to 1. The parameter µ is the mean, and b controls the width, of the first gamma distribution. The µ fµ and b fb in the second gamma distribution are relative to µ and

notation is given by: Gamma(q; u, b) = 1 bµΓ(1) ¯  q bµ 1_b−1 exp(−q bµ) (2.4)

The full fit also includes the noise pedestal: a Gaussian function to account for charge fluctuations from electronics noise in PMTs that see zero PE. The charge observed in a PMT that observes n PE is the sum of a charge drawn randomly from the pedestal and ncharges drawn randomly from SPE charge model SPE(q). Thus the charge distribution for n PE is given by SPE(q) convolved n times with the pedestal function Ped(q). The total charge distribution observed in a PMT that observed a mean λ PE over a set of AARF events is fitted using the sum of a set of n PE components weighted by the Poisson probability of seeing n PE given a mean PE λ observed in each hit PMT:

f(q) =B · [A · Ped(q) + Poisson(1, λ ) · Ped(q) ⊗ SPE(q) + Poisson(2, λ ) · Ped ⊗ SPE(q) ⊗ SPE(q) + . . . ]

(2.5)

where the noise pedestal normalisation A is floated in the fit alongside the parameters in SPE(q).

The variation with time during commissioning of the mean single PE charge µ in the AARF PMTs closest to the neck of the AV is shown in Figure 2.23. The mean single PE charge increased as the detector cooled during commissioning and filling and has since stabilised after the fill in the conditions under which physics data is taken.

AARF data is collected monthly and the parameters from the function fitted to the data each month are stored in a CouchDB database. The charge model in simulation is also updated to match the measurement from data. Charge-based position reconstruction algorithms can use single photoelectron charge distributions as part of their model as dis- cussed in Chapter 4. The position reconstruction automatically accounts for the variation of the parameters of the single photoelectron model with time by requesting them from the database.

2.7. CALIBRATION SYSTEMS CHAPTER 2. THE DEAP-3600 DETECTOR

Figure 2.23: The variation of mean single PE charge with time for a group of 8 PMTs at the top of the detector. One representative PMT (no. 6, in red) is shown with the mean SPE charge deter- mined from the AARF monitoring runs that occurred over this period, and error bars representing the parameter errors from the fit depicted in Figure 2.22. The time axis begins on 1st June 2016. A slight upward trend is observed for most PMTs over the course of the year depicted. Discontinu- ities in the mean SPE charge occur at times when the PMTs were powered down and back up. The environmental conditions changed a number of times, as depicted on the plot: Phase 1: AV under vacuum with TPB deposited and compensation coils on; Phase 2: in addition, water shielding tank is filled with chilled water; Phase 3: AV filled with argon gas at room temperature; Phase 4: Cool down phase with increasingly cold argon gas. Plot prepared by T. Pollmann.

The Laserball

The laserball was deployed once after the deposition of the TPB source. The laserball consists of a laser head attached to a fibre, which terminates at an acrylic stub light guide within a PerFluoroAlkoxy plastic flask containing 50 µm glass beads suspended in sil- icone gel. The flask as shown in Figure 2.24 is designed to emit pulses of UV light isotropically after repeated light scattering within the silicone gel. The laser is driven us- ing a Hamamatsu PLP-10 picosecond light pulse generator and a set of laser diode heads that emit at 375 nm and 445 nm. The data acquisition trigger module is again synchro- nised with the pulse generator to ensure that data is read out as light pulses are generated, and that each light pulse is a separate event. The distribution of peak times of pulses in PMTs relative to the trigger time is shown in Figure 2.25, with a total range of 3.5 ns.

During deployment the laserball was attached to a support assembly, and suspended at an adjustable height within the AV. The laserball was deployed within the AV after TPB deposition, with the AV filled with N2gas at 20.28 PSIA to prevent the propagation

Figure 2.24: The laserball driven by the 445 nm laser head, photographed during ex-situ testing by N. Fatemighomi. , ns) pulse Pulse time (t 6400 6402 6404 6406 6408 6410 6412 6414 ) / 0.4ns pulse Probability, P(t 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 DEAP 0 0 6 3 Commissioning Preliminary

Figure 2.25: Pulse time from every PMT measured relative to the start of the DAQ event waveform. Data was taken with the laserball at the centre of the detector, using the 445 nm laser diode. The pulse time has been corrected for SCB channel timing offsets, and timing offsets from varying PMT cable lengths for each channel. Plot prepared by F. La Zia.

2.7. CALIBRATION SYSTEMS CHAPTER 2. THE DEAP-3600 DETECTOR

Figure 2.26: Left: Relative efficiencies of the PMT array obtained by averaging the relative effi- ciencies obtained four data sets taken with the laserball at the centre of the detector, separated by rotation angle ∆φLB= π/2. Vertical axis is the number of PMTs with relative efficiency observed in each 0.02 wide bin. Right: A plot comparing the relative efficiencies obtained from the laserball data to those obtained from AARF data, where an efficiency of 1 is defined for a single PMT. Plots prepared by R. Mehdiyev.

material. Data was taken with the laserball placed at the centre of the x − y plane, at three elevations of z = 0 mm and z = ±550 mm relative to the equator of the detector, and rotated to four positions separated by rotation angle ∆φLB= π/2. The error on each

elevation is estimated at ±50 mm, and the error on each rotation is estimated at ±8◦. The laserball data was used to estimate and correct for timing offsets between recorded PMT pulse times. Channel-to-channel timing offsets are produced in 8 ns intervals by the SCB electronics. The variation in PMT cable length produced variable offsets for different PMTs around the detector. The offset was measured using the laserball at z = 0 mm such that the transit time, and distance to every PMT, of photons leaving the laserball surface is equal. The pulse time distribution that is corrected by subtracting the offsets is shown in Figure 2.25. The width of the pulse time distribution is 3.5 ns.

The laserball was also used to calculate the relative variation in PE production effi- ciency due to the combination of PMT collection efficiency and individual light guide optical effects. The laserball was placed at the centre of the detector. The variation in oc- cupancy with PMT ID was recorded, and fitted with a straight line, an example of which is shown in Figure 2.26. PMT ID’s are indexed from the PMT closest to the neck to the PMT furthest from the neck with increasing ID number. The efficiency is recorded as

the ratio of observed occupancy to the fitted line, and averaged over four rotated posi- tions of the laserball separated by rotation angle ∆φLB= π/2. Mean relative efficiencies

measured using the laserball are shown in Figure 2.26. These efficiencies are recorded and applied in simulation, and compensated for in position reconstruction as discussed in Chapter 4. Importantly the laserball is the only calibration source placed at a known location within the AV, so it can be used to demonstrate that the position reconstruction functions correctly, as discussed in Chapter 5.