Scintillation Light Collection Tests for the G0 Experiment (CEBAF E-91-017) Herbert Breuer Department of Physics University of Maryland, College Park, Maryland 20742 (September 16, 1998) PP# 99-021 Communications: Dr. Herbert Breuer Experimental Nuclear Physics Tel.: 301-405-6108 Email: BREUER@enp.umd.edu _________________________________________________________________________ _________________________________________________________________________ $PS newpage CONTENTS ======== (1) SUMMARY (2) MATERIALS AND METHODS USED FOR TESTS (3) TABLES OF RAW TEST RESULTS (3.1) Details of Hardware Used (3.2) Test results to obtain properties of light guide (3.3) Absolute normalization of photonelectrons seen by tube (3.4) Gain dependence on Photocathode-to-dynode HV (3.5) Gain dependence of tube-base combination on HV (3.6) Correlations between HV, signal hight, and maximum rate (4) DISCUSSION OF TEST RESULTS (4.1) Air versus optical grease joints (4.2) Effective attenuation length of the light guides (4.3) Attenuation on different surfaces and in the bulk material (4.4) How to minimize attenuation losses (4.5) Losses due to bending the light guide (4.6) Wrapping of the light guide (4.7) Wrapping of the scintillator (4.8) Photoelectron collection efficiency (4.9) Ideal yield of photoelectrons (4.10) Phototube gain and base choice (4.11) Improvement and sealing of machined surfaces by painting $PS newpage (1) SUMMARY =========== Tests have been performed to determine the performance of light guides and the absolute number of photoelectrons collected with hardware suggested to be used for the G0 experiment. The performance of the Philips XP2262 photomultiplier tube with a TRIUMF prototype base has been evaluated. Results are: -- The UV-transmitting acrylic plastic light guide material was found to be non-uniform in thickness and up to 20% thinner than the nominal value. -- Most of the light losses in the light guide are due to imperfections on the machined surfaces making the width of the light guide between the machined surfaces the critical quantity for light losses. A light guide with 2 x 0.32 inch**2 cross section (not ideally) machined on the narrow sides, showed an effective attenuation length of 185 cm, which is consistent with about 95-96% reflectivity of the machined surfaces. Thus it is suggested to use >= 44 mm wide light guides also for all narrower scintillators. -- Light losses in a 0.32 inch thick light guide around a bend with a radius of 2 cm and 5 cm were measured to be about 8 and 11%, respectively, consistent with MC predictions of 5 and 11%. -- Wrapping most of a 913 mm long light guide tightly with aluminized mylar resulted in a gain in light of 11%. Wrapping the same light guide instead with black Tetlar foil resulted in a loss of light of 10%. It is suggested to mount light guides and scintillators in air without any or with minimal wrapping in a light-tight box, resulting in little loss of light and a big gain in simplification. -- Minimum ionizing particles penetrating a 10.0 mm scintillator (n=1.58), coupled via two straight ideal light guides (n=1.49) to Philips XP2262 photomultiplier tubes will yield 440 photoelectrons in each tube. Bulk attenuation, imperfect surface preparation, and curved shapes will result in losses of typically 70% over two meters, resulting in about 130 detected photoelectrons. The exact losses due to those variables as well as due to possibly needed cross-section area mismatches can be determined by Monte Carlo simulations. -- A 1/4 inch (6.35 mm) thick scintillator rather than a 10 mm scintillator seems to be sufficient for the experiment and results in reduced neutron background sensitivity and reduced edge effects. -- The tube-base combination (XP2262 tube, TRIUMF base A, B prototype) used in these tests has to be operated at a rather low HV, in a range where the gain is extremely sensitive to small HV changes. As a consequence, the base was redesigned for lower gain, with information provided elsewhere. -- Light losses in the machined surfaces of the light guides can be strongly reduced by painting and sealing these surfaces with a solution of PMMA in Toluene. $PS newpage (2) MATERIALS AND METHODS USED FOR TESTS ======================================== (a) A block of scintillator was cut from 10.0 mm thick Bicron BC408 plastic scintillator sheets. The block was machined to a size of 85 x 50.8 mm**2, with the edges diamond-cut and polished. (b) Light guides were cut from UV transparent acrylic plastic sheet, "CRYO, Acrylite GP, UVT", nominally 3/8 inch (9.5 mm) thick, however measured to be between 0.342 and 0.307 inch (8.7, 7.8 mm) thick. The acrylic plastic was cut, machined and polished to 50.8 mm or 15.5 mm, respecively, wide light guides of different lengths. (c) Light guides were also cut from non-UV transparent acrylic cast sheet, "Otto Hass, Plexiglass type G" nominal thickness 3/8 inch (measured: 0.322-0.361 inch), cut, machined, and polished the same way/dimensions as the UV transmitting acrylic light guides. (d) Cutting/machining was done with a diamond tipped tool, 3 cutting surfaces; rotational speed was at a maximum of 2720 rpm; translational speed of 1.125 inch/minute; cooling with water mixed with 2% cutting fluid "Cool Mist". (e) Machined surfaces were polished with "classic buffing compound, blue"; blue is for plastic. (f) A Philips XP2262 photomultiplier tube was used. A mu-metal shield was placed around the tube, extending typically about 2 inch beyond the tube face. The TRIUMF tube housing was not used. (g) Base prototypes A and B (L.Lee, G0 6/6/97 collaboration meeting proceedings) were tested and used. For most tests -1700 V (base A) and -1707 V (base B) were applied, which give nearly the same voltage distribution and the same gain for the two different bases. (h) Radiation sources: (a) cosmic rays, requiring a 90x90 mm**2 trigger scintillator approximately 11 cm from the test scintillator. (b) a rather tired 17 year old Ru-106 electron source; for this the detector assembly (i) was used. (i) Detector assembly for source operation: plastic scintillator, 4 mm thick, 20 mm wide, 40 mm high, coupled with optical grease to the 8-mm face of a 50.8x50.8x8 mm**2 block of UV-transparent light- guide material. The assembly was held together by aluminized mylar stretched across the scintillator. The mylar was taped to the block. (j) Electronics: The trigger scintillator (HV constant at -1600 V) signal was constant-fraction discriminated with a rather high threshold. The logic output signal provided the gate for a LeCroy QVt-module (LeCroy 3001), used as a charge integrating multi-channel analyser. The QVt's internal gate width was set to about 100 ns. The test scintillator signal travels first through two variable attenuators (0.1 to 1.0 each) (Phillips 5010) and two stages of variable amplifiers (1.0 to 10.0 each) (Phillips 771), and then through appropriate delay cables into the Q-input of the QVt. (k) An absolute photoelectron calibration was done before and after nearly every measurement. For this the charge for a single photoelectron was measured by taking a spectrum with the overall gain changed by a factor of 200 or 400 while also changing the Constant Fraction discriminator input from the trigger scintillator to the test scintillator. Differences in gain are due to tube and mu-metal shield orientation as well as temperature fluctuations. After HV-turn-on a waiting period of at least 10 minutes is required for the gain to stabilize. (l) Mounting of setup: normally all modules were set up in a dark room without any wrapping and with minimal contact with support structures. Contact between optical components was initially with minimal air gaps, later with optical grease. (m) Uncertainties: QVt-peak readout accuracy (including pedestal subtraction) was about 2-3%. Reproducibility of results was better than +-2%. Uncertainties between different pieces of hardware are estimated to be about 5%. The absolute calibration uncertainty between single photoelectron and multiple photoelectrons is (conservatively) estimated to be below 10%. Note 1: UVT acrylic plastic: the sheet used, with a 14 inch wide piece cut from the inner portion of the 4x6 foot original piece has LARGE VARIATIONS IN THICKNESS exceeding 10%, ranging from 8.69 to 7.80 mm for the few points measured. The sheet is thicker near the original edges than in the center. Note 2: The machined surfaces show easily visible tool marks. The buffing only polishes these tool marks, but leaves the main marks in place. (3) TABLES OF RAW TEST RESULTS ============================== (3.1) Details of Hardware Used ------------------------------ - Phototube: Philips XP2262, serial number 26146. - Bases: two bases of slightly different designs by TRIUMF. - Signal from Ru-106 electron source or cosmic rays. - Trigger from separate detector behind the "primary" scintillator. - "Primary" scintillator assembly: 4 mm thick, 2 cm wide by 4 cm high scintillator coupled with optical grease on 4 cm side to 2x2x3/8 inch**3 UV transm. acrylic; held together by a strip of aluminized mylar taped to acrylic piece. - All UV-transmitting light guides from 3/8 nominal (approx. 0.32 inch actual) thickness UV transmitting acrylic; the non-UV transmitting light guides are from 3/8 nominal (0.35 inch thick) sheets. The following light guides are 2.00 inch wide: (A) 2x36 inch**2 straight, UV transmitting; (B) 2x36 inch**2; 6 inch straight, 2 cm radius bend, rest straight, UV; (C) 2x36 inch**2; 6 inch straight, 5 cm radius bend, rest straight, UV; (D) 2x11.5 inch**2 straight, UV; (E) 2x2 inch**2 straight, UV; (F) 2x36 inch**2 straight, non-UV transmitting, xx thick. The following narrow guides are 0.610 +- 0.010 inch wide: (g) .6x36 inch**2 straight, UV transmitting; (h) .6x10 inch**2 straight, UV transmitting; (i) .6x36 inch**2 straight, non-UV tr.; (j) .6x10 inch**2 straight, non-UV tr.. - Unless otherwise indicated, the light guides are in air with minimal support and without covering or wrapping. - Coupling is air or optical grease; the first value in the table below is between scintillator assemby and light guide, the second refers to the coupling between light guide and phototube. - Net channel: from MCA, 1 channel is 1 pC. Net channel number uncertainties are typically 1.0 or 2.0. - Calibration: from single photoelectron peak, amplified 200 or 400 times as compared to source signal. - Renormalized: according to measured light transmission before bending A: factor 1.00, B: factor: 0.9433, C: factor 0.9779; each +- 0.04. $PS newpage (3.2) Test results to obtain properties of light guide ------------------------------------------------------ For details of table entries see section (3.1) above. ___________________________________________________________________________ test logb. net # of ph. renor- # light guides coupling page chan. calib. electr. malized --- ------------ -------- ----- ----- ------- -------- ------- [using base A (300V fixed) at 1700 V] 1 E: 2x2 ...... air/air 34 116.0 0.4625 251 2 D: 2x11.5 ...... air/air 33 99.6 0.4650 214 3 A: 2x36-straight .. air/air 33,32 77.5 0.4696 165 165 4 C: 2x36-5cm ...... air/air 31 79.9 0.4630 172 168 5 B: 2x36-2cm ...... air/air 31 78.4 ~0.4575 171 162 6 E: 2x2 ...... gre/air 34 134.4 0.4638 290 7 D: 2x11.5 ...... gre/air 37 112.0 0.4563 245 8 A: 2x36-str. ...... gre/air 36,37 86.3 0.4634 186 9 D: 2x11.5 ...... gre/gre 38 161.1 0.4438 363 10 A: 2x36-str. ...... gre/gre 39 119.8 0.4575 262 262 11 B: 2x36-2cm ...... gre/gre 40 114.7 0.4575 251 236 [using base B (200V fixed) at 1707 V] 12 E: 2x2 ...... gre/gre 46 179.5 0.4488 428 13 A: 2x36-str. ...... gre/gre 46,47 119.2 0.4546 262 262 14 B: 2x36-2cm ...... gre/gre 44 110.1 0.4456 247 233 15 C: 2x36-5cm ...... gre/gre 44 109.5 0.4444 246 241 16 B+C both coupled .. gre/gre/gre 43 63.3 0.4563 139 128 17 A: 2x36-str,full .. gre/gre 47 137.5 0.4463 308 aluminized mylar 18 A: 2x36-str, 3 inch gre/gre 48 128 0.4625 277 aluminized mylar 19 A: 2x36-str, 3 inch gre/gre 48 107.5 0.4338 249 al.m.+black Tetlar [scintillator assembly cleaned and redone, now less light output] 31 g:.6x36, UV gre/gre 54,55 96.5 1.9175 50.3 31a g:.6x36, UV gre/gre 58 98.0 1.9100 51.3 32 h:.6x10, UV gre/gre 57 104.5 0.9538 109.6 33 i:.6x36, non-UV gre/gre 55 70.6 1.9275 36.6 34 j:.6x10, non-UV gre/gre 57 98.0 0.9413 104.1 36 i':.6x36, non-UV,polished gre/gre 64 120.5 1.9288 62.5 41 A: 2x36-str UV...... gre/gre 61 106.1 0.4647 228.2 42 F: 2x36-str non-UV.. gre/gre 59 90.0 0.4575 196.7 ___________________________________________________________________________ $PS newpage (3.3) Absolute normalization of photonelectrons seen by tube ------------------------------------------------------------ For details of table entries see section (3.1) above. ___________________________________________________________________________ test logb. net # of ph. renor- # light guides coupling page chan. calib. electr. malized --- ------------ -------- ----- ----- ------- -------- ------- Light propagating into the direction of photo-tube: - - - - - - - - - - - - - - - - - - - - - - - - - - [same base/HV (Base B, 1707 V), but COSMICS and 85x51x10 mm**3 scintillator as source; vertical orientation of scintillator and light guide; electronic gain has been changed] 101 D:2x11.5 gre/gre 51 129.5 0.3244 375 102 D:2x11.5,black far end gre/gre 52 90.5 0.3225 281 103 D:2x11.5,AlMyl far end gre/gre 53 108.5 0.2252 482 Reabsorption of light leaving scintillator: - - - - - - - - - - - - - - - - - - - - - - [horizontal orientation of scintillator and light guide] [check effect of aluminized mylar on scintillator only: loosely cover one side or both with al. mylar] 104 D:2x11.5, open gre/gre 74 105 0.2931 351+-9 105 D:2x11.5, top covered gre/gre 75 102.5 0.2913 352+-10 106 D:2x11.5, top+bott cov.gre/gre 76 112.5 0.2913 386+-9 107 D:2x11.5, bottom cov. gre/gre 76 105 0.2994 351+-13 108 D:2x11.5, open gre/gre 77 105 0.3038 346+-10 ___________________________________________________________________________ (3.4) Gain dependence on Photocathode-to-dynode HV -------------------------------------------------- Previous base (B) @ 1707 had 306 V cathode-to-dynode; now use modified base A with 582 V cathode-to-dynode HV (fixed nominal 600 V); adjusted HV for same gain (i.e. 1960 V applied). ___________________________________________________________________________ test logb. net # of ph. renor- # light guides coupling page chan. calib. electr. malized --- ------------ -------- ----- ----- ------- -------- ------- 109 D:2x11.5, open gre/gre 79 116.7 0.3013 387+-11 110 same as 109 79 117 0.2981 393+-14 111 D:2x11.5, top+bott cov.gre/gre 80 123.4 0.2975 418+-13 112 same as 111 81 511 1.2000 426+-10 113 same as 111,112 81 516 1.2100 426+-7 ___________________________________________________________________________ $PS newpage (3.5) Gain dependence of tube-base combination on HV ---------------------------------------------------- Using a Phillips XP2262 tube with TRIUMF base with a fixed cathode to first-dynode ("acceleration") voltage. With nominally 600 V (measured 582 V) acceleration voltage and 1960 V applied: gain = 1.01pC / photoelectron With nominally 300 V (measured 294 V) acceleration voltage: HV HV/stage SPE(noise) SPE(cosmic) factor (V) (V) (pC) (pC) /100V -1400 58 0.0115 -1500 69 0.0632 5.50 -1550 0.123 -1600 80 0.238 0.250 * 3.95 -1650 0.457 -1700 91 0.833 0.793 * 3.50/3.17 -1750 1.34 -1800 102 2.17 2.16 * 2.61/2.72 -1850 3.34 3.23 -1900 113 5.10 4.36 2.35/2.02 -2000 124 10.7 8.23 2.10/1.89 = HV is the overall applied High Voltage. = HV/stage is the HV at each of the nine variable stages, after subtracting the nominal 300+180+200+200V for the Zener-Diode stabilized stages. = Results are measured with a charge integrating multi-channel analyser = SPE(noise) are single photons from light leaks. = SPE(cosmics) are results from cosmic rays. = *: calibration of SPE(cosmic) from average of 1600,1700,1800 V values. (3.6) Correlations between HV, signal hight, and maximum rate ------------------------------------------------------------- = Assuming a 10 mm scintillator; = 150 photoelectrons is a realistic estimate for electrons (2 MeV); = 1050 photoelectrons is a realistic estimate for protons (14 MeV); = signal hight for 10 ns FWHM triangular signal into 50 Ohm; = For rate limit: the current limit of the tube is 0.2 mA average current; using 0.05 mA for signals of interest, which allows for the same additional current for background events and a safety factor of two. HV signal signal rate rate rate (V) (mV) (mV) (MHz) (MHz) (MHz) 2 MeV 15 MeV photo- 2 MeV 15 MeV electr -1400 9 60 4400 29. 4.1 -1500 47 330 790 5.3 0.75 -1600 180 1250 210 1.4 0.20 -1700 620 4400 60 0.40 0.06 -1800 1600 11000 23 0.15 0.02 $PS newpage (4) DISCUSSION OF TEST RESULTS ============================== The numbers in the following sections refer to the test # in the tables. The term "large angle" refers to photons travelling ("rays") at angles close to the angle of total internal reflection, while "small angles" are close to perpendicular to the face of the photo tube and thus close to the average light propagation direction in the light guide. (4.1) Air versus optical grease joints -------------------------------------- Comparing tests #10 and 8, where the difference is that the joint between light guide and phototube is either air or optical grease, shows a loss of 29% due to the air gap. A second air gap between scintillator and light guide only reduces the light output by an additional 11% (comparing #8 with #3). The much smaller effect of the second air gap can be explained by observing that the rays with angles furthest away from normal incidence on the plastic-air-plastic/glass interface are most strongly reflected (lost). These rays thus already have a smaller probability of reaching a second plastic-air-plastic/glass interface. For the shorter light guides the percentage of light loss at the air gaps are larger. This can be explained that the same rays which have a high probability to reflect at the airgap also have a higher probability to leave the light guide at surface imperfections on the longer light guide, being closer to the angle of total internal reflection. (4.2) Effective attenuation length of the light guides ------------------------------------------------------ We extract the attenuation length d0 from the measured intensity difference I and I0 between two different lengths of light guides, I = I0*exp(-d/d0), with the length difference d. This assumes that the different pieces of light guide are of identical quality. We use a short length of "reference" light guide to let light beyond the angle of total internal reflection escape and to minimize the effect that light escaping from the light-guide may strike the open photo-tube through air. Results are: (a) double airgap 36/11.5 inch (#3,2): d=0.622 m; d0 = 2.4 m (b) double airgap 36/2 inch (#3,1): d=0.864 m; d0 = 2.1 m (c) single airgap 36/11.5 inch (#8,7): d=0.622 m; d0 = 2.3 m (d) single airgap 36/2 inch (#8,6): d=0.864 m; d0 = 1.9 m (e) no airgap 36/2 inch (#13,12): d=0.862 m; d0 = 1.8 m The different results for d0 are consistent with the uncertainties in the measurements; the apparently systematic deviations between 11.5 and 2 inch "reference" light guides may be due to differences in machined surface quality as well as to double reflections, recovering some light with the 2-inch long light guide. Light rays with large angles have longer paths through the light guide ("bulk"-attenuation is effective) as well as more reflections on surfaces ("surface imperfection" attenuation), thus being attenuated stronger than small angle rays. There are fewer large angle rays present for the air-gap examples. (Small, large angle defition: see section (4)). From the above one could assume that, once the large angle rays are removed, additional lengths of light guide will have a lower effective attenuation length. #16 tests this idea. (f) B,E ...................... (#14,12): d=0.864 m; d0 = 1.42 m (g) C,E .......................(#15,12): d=0.864 m; d0 = 1.50 m (h) B+C,E .....................(#16,12): d=1.778 m; d0 = 1.47 m The result is that joining two long pieces of light guide leaves the effective attenuation length unaffected: (h) lies between (f) and (g). The reason for this is that the machining imperfections scatter the light at each reflection away from regular reflection between perfect parallel surfaces and sometimes also beyond the angle of total internal reflection. (4.3) Attenuation on different surfaces and in the bulk material ---------------------------------------------------------------- A laser beam provided qualitative evidence that the machined surfaces are the major source of light loss. While the beam within the light guide is faintly visible (a sign of scattering and thus attenuation within the bulk material), the machined surface is brightly visible. The cast surfaces, in contrast, have nearly no light output, as long as they are not damaged in any way. An attempt to "polish" a cast surface with 0.3 micron polishing compound resulted in an exit of laser light through the polished area. Based on Bicron test results, the cast surfaces have a reflectivity of about 99.8%. Based on a comparision with Monte Carlo simulations, we find that our measured attenuation lengths are consistent with reflectivities of the machined surfaces of about 95-96%. (4.4) How to minimize attenuation losses ---------------------------------------- As a consequence of the dominance of light loss on machined surfaces, these surfaces should be kept as far apart as possible, to minimize the number of reflections. In this model, the thickness of the light guide, between cast surfaces may be very thin. For a fish-tail light-guide design (cutting strips and joining them repackaged) for matching a photo-tube surface, the fish tail should be close to the tube; of course this is the usually adopted design. (4.5) Losses due to bending the light guide ------------------------------------------- After measuring the light transport properties of tree 36-inch long straight light guides, two of these light guides were heated to about 130-150 degree C and bent by 90 degree around pipes of 2 cm and 5 cm radius, respectively. The quality of the bends is not perfect, but demonstrates that even imperfect bends result in low light loss. In both cases, a six-inch straight section was followed by the bend, which was followed by a straight section for the remainder of the 36 inch light guide. The numbers in the "renormalized" column of the tables are the relevant quantities here. These values take into account differences in the light guide performances observed before bending. Comparing #15,13 and #14,13, the loss of light is 8% in the 5 cm bend and 11% in the 2 cm bend. These losses have a 5 percentage point estimated uncertainty due to uncertainties between different pieces of hardware. Monte Carlo calculations indicate losses of 4.7% and 11.2% for perfectly bent light-guide pieces with 2 and 5 cm radius, respectively, in good agreement with the experimental results. (4.6) Wrapping of the light guide --------------------------------- Three setups were compared to observe the effect of wrapping the light guide. Contact with mounting hardware is estimated to be less than 1 cm**2. (A) The source assembly and the first three inches of light-guide were wrapped in aluminized mylar (#18). (B) In addition to the (A) configuration, the full length, excluding the last 5.5 mm near the photo-tube, were tightly wrapped with several layers of aluminized mylar (#17). (C) The aluminized mylar in (B) was replaced by black Tetlar foil (#19). Wrapping with aluminized mylar results in a gain of photons of 11%. Wrapping with black foil results in a loss of 10% of the light. Presumably the quality of the machined surfaces as well as the amount of surface contact will have an impact on the result. The black-foil-wrapping loss can be taken as a measure of surface contact due to the wrapping. The contact area, resulting in light absorption, seems to be rather substantial. The gain by wrapping with reflecting foil is very modest. Very tight wrapping or pressure from mounting hardware over extensive surface area may even cause light losses due to the limited reflectivity of Aluminum. (4.7) Wrapping of the scintillator ---------------------------------- Cosmic rays and the 81x51x10 mm**3 piece of scintillator, coupled to the open 11.5 inch light guide were used to test the effect of wrapping the scintillator. It was expected that UV light leaving the scintillator may be reabsorbed/scattered upon reflection back into the scintillator. Tests 104 to 108 and 109 to 112 test this feature. The scintillator end opposite from the light guide was always left uncovered in all of these tests. Placing aluminized mylar only on one face (105,107; compared to 104, 108) does not have any measurable effect. Covering all four faces of the scintillator which are parallel to the main light guide faces results in an increase in photoelectrons recorded of 8 to 10% (test 106 compared to 104,105,107,108; tests 111,112 compared to 109,110). It appears that multiple reflections are needed to create a measurable effect. It is possible that the increased yield is not due to absorption/reemission of UV but rather rescattering of light at imperfections of the machined surfaces. Thus the increased photon yield from covering the scintillator may be less pronounced when the machined surfaces have a better quality. The increased yield for the scintillator is very similar to the increase when covering the light guide; both have been mashined with similar quality. It is suggested NOT to wrap any of the light-guides nor any of the scintillators, except at points where mounting hardware touches significant areas of the surface. Assuming well-machined rectangular surfaces, cross talk between detectors is nearly impossible. The whole assembly for one octant should be placed into a light tight box, with all inner surfaces of the box painted black and some strategically positioned Tetlar sheets for optical separation of detector elements. (4.8) Photoelectron collection efficiency ----------------------------------------- The photo-tube's photoelectron collection efficiency depends on the photo-cathode to first-dynode potential. For the Philips XP2262B tube used this voltage may be between 300 and 800 V. An increase of this voltage from nominally 300 V (tests 104 ff.) to 600 V (tests 109 ff.) results in about 11% increased photoelectron yield. From previous similar tests on 5-inch tubes it was found that the increase in yield decreases with increased photo-cathode to first-dynode voltage; thus the first few 100 V above 300 V are expected to provide the largest benefit. This test was performed with a 50.8 x 8.3 mm light guide on a nominally >44 mm diameter useful photocathode. A different light guide geometry may yield different results. It is recommended to increase the cathode to dynode-1 voltage to between 600 and 800 V by using Zener-diodes. Appropriate parallel resistors should be installed to protect the tube in case a diode malfunctions. The recommended interdynode voltage between dynode-1 and 2 and between dynode-2 and 3 should be used for best resolution (1.1*Vd and 0.9*Vd, respectively). (4.9) Ideal yield of photoelectrons ----------------------------------- We define the "ideal yield of photoelectrons", or the maximum possible yield, as the number of photoelectrons detected for a scintillator- light guide combination with perfect surfaces and infinite attenuation length, but with no reflection from the far end opposite of the phototube. Thus this ideal yield is the + number of photoelectrons detected, + for minimum ionizing particles in 10 mm of scintillator (2.0 MeV), + for a straight, rectangular geometry, with the same cross section for scintillator and light-guide, + with light only emitted towards the direction of the photo-tube, + accounting for all losses due to differences of index of refraction [n(scintillator)=1.58, n(light guide)=1.49, n(air), n(tube-glass)], + with no wrapping of any of the elements, + accounting for average photoelectron conversion efficiency. Using tests 101-103, the known effective attenuation length for the scintillator and light-guide (including the machining imperfections) and Monte Carlo simultions to check self-consistency, we find this "ideal yield" directly from tests 101 to 103 to be 360 +- 30 photoelectrons. Increasing the cathode to first-dynode voltage adds about 11%, restricting the cross section to the approximately 45 mm active diameter of the tube adds another 10%. Thus, with ideal scintillator and light guide we can expect about: 440 +- 40 photoelectrons per 2 MeV of energy-loss for each phototube at both ends of a tube-lightguide-scintillator- lightguide-tube assembly. Based on about 16000 photons per two-MeV energy-loss emitted, each tube at each end of an ideal scintillator-lightguide assembly would convert 2.75% of the photons created into photoelectrons. As a currently achievable example, we assume a - 45 mm wide light guide and a <= 45 mm wide scintillator, - a 200 cm long scintillator plus light-guide combination, - a 1.8 m effective attenuation length, - a 10% loss of light in a 90 degree bend. This results in an overall attenuation during transport to 0.296 * 440 = 130 photoelectrons for a minimum-ionizing particle crossing a 10 mm scintillator. We expect that the effective attenuation length will be larger with better surface treatment. The yield will be smaller if the light-guide is wider than the active tube surface. The effect of curved, non-uniform-width scintillators still needs to be determined. All these contributions to transport losses can be determined with Monte-Carlo simulations. From these results we conclude that it it is not necessary to use 10 mm of scintillator material; rather 1/4 inch (6.35 mm) or 5 mm thick scintillators should be considered. Thinner scintillators will have - less edge effects and - lower trigger rates from neutrons and high-energy (>1MeV) gammas. The maximum safe anode current of the phototube is probably limiting the count rate capabilities for each tube. Note that the effect of low energy gammas and neutrons, which may be the dominant contribution to the photo-tube anode current, is not reduced by a reduced scintillator thickness. This is due to the fact that the reduced number of photoelectrons per event is compensated by a needed increase in gain (higher voltage) of the phototube. Thus any linear decrease in rate is compensated by the same increase in tube gain. (4.10) Phototube gain and base choice ------------------------------------- The choice of photo-tube gain is critically limited between -- for this experiment -- two rather close extremes (see also section (3.6)). A minimum signal hight is required for discriminator settings and ADC charge input, after signal losses in cables, possibly back-termination, and line-noise pickup contributions. The minimum gain estimates have to take into account the dynamic range due to different energy losses as well as attenuation differences between both ends of the scintillator. The maximum gain is restricted by the maximum current through the tube and the tube degradation resulting from the long-term integrated anode current. This current depends on the expected signal rate from signals of interest (electrons, protons) of about 1 MHz and additional high energy (neutrons, gammas, stray charged particels) background as well as low energy "grass" signals (low energy photons, neutrons, activation). The experiment is to run over several months with continuous beam. An amplifier after the photo-tube, which would widen the tube's gain-limits separation, is currently not being considered as an option. The TRIUMF Bases A and B, combined with the 12-stage XP2262 tube provide a gain which is too high to run comfortably with the expected photoelectron yields of 100-200 for electrons and up to more than a 1000 for protons. The bases had been optimized for photo-electron yields much below the ones expected. The option of running these bases at much reduced high voltages results in very steep gain-HV characteristic curves and rather strong temperature dependencies. As a consequence of the present as well as additional tests, the bases have been modified to reduce the gain so that the nominal operating voltage will be between about 1700 V (low energy protons) and 2100 V (electrons) with a factor of 10 in gain safety-margin available above and below the nominal operating voltage range. The main modifications are an indroduction of strong modulations in the divider-chain voltages and an increase of the acceleration voltage to 450 V. Instead of changing the base one could have chosen a tube with a lower gain, e.g., a similar 8-stage tube (XP22082). Unfavorable experience with this model in Hall C as well as cost and ordering considerations resulted in not considering this option seriously. (4.11) Improvement and sealing of machined surfaces by painting --------------------------------------------------------------- It appears that a large fraction of the light losses on the machined surfaces originate from random scattering at microscopic (wave-length size) structures (cracks?) in the machined surfaces. To seal theses small imperfections, a solution of 5% PMMA in Toluene was prepared. The machined surfaces of the light guides were painted with this solution with the expectation that the small imperfections would be sealed after evaporation of the Toluene. The prodedure nearly halfed the reflection losses in some tests and in one extreme case even doubled the light output. Further improvement may be achieved by optimizing the index of refraction of the sealant. This solution, unmodified, cannot be used to paint scintillator material, since it dissolves the scintillator rapidly and extracts and deposits crystals on the surface after evaporation. While this method can be refined and may prove useful to help with light collection, at the present time we hope to achieve similar results by improved machining and polishing. It is anticipated that eventually a report about the method will be written. In the meantime the author can be asked for details.