HERA-B RICH light collection system
Transcription
HERA-B RICH light collection system
Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 HERA-B RICH light collection system Daniel R. Broemmelsiek Department of Physics and Astronomy, Northwestern University, Evanston, IL 60208, USA Abstract The HERA-B RICH detector utilizes the recently developed Hamamatsu R5900 multianode photomultiplier. In order to increase the light collection e$ciency, a novel scheme to optically magnify the photocathode is implemented. The design and development of these optics are persented here. 1999 Elsevier Science B.V. All rights reserved. 1. Introduction The primary purpose of the HERA-B spectrometer is to measure CP violation in the B meson system. The HERA-B RICH detector, [1] and Fig. 1, is designed to provide excellent particle identi"cation over the momentum range 4}80 GeV/c. Flavor tagging of B (or BM ) mesons by e$cient kaon}pion separation for HERA-B is the speci"c role of this RICH detector. The photosensitve component of the photon detector consists of both the 4 and 16 anode versions of the Hamamatsu R5900 multianode photomultiplier arranged on 36 mm centers as shown in Fig. 2. A basic problem for any RICH detector is to collect Cherenkov light with high e$ciency while preserving the required position resolution. By fortuitous coincidence, the R5900 anode cell sizes, 9 mm;9 mm and 4.5 mm;4.5 mm for the 4 and 16 anode versions, respectively, are almost exactly 1/2 the cell size needed for su$cient position resolution required by kaon}pion separation. By E-mail address: daniel.broemmelsiek@desy.de (D.R. Broemmelsiek) magnifying the photocathode a factor 2, not only is the fraction of photosensitive area on the focal plane improved, 25}95%, but the cell sizes are then matched to the intrinsic angular resolution for a single photon in the di!erent regions shown in Fig. 2. An array of telescopes consisting of 2 acrylic lenses is mounted in front of the PMT array in order to give the correct magni"cation of the photocathodes. 2. Telescope design The primary consideration when designing such a telescope is the angular distribution of Cherenkov light at the focal plane surface. The focal plane consists of 10 planar surfaces which approximate a cylindrical surface. The cylindrical surface is chosen to remove the shift in the photon angular distribution at the focal plane due to the tilted spherical mirror and the de#ection of charged tracks by the 2 T m dipole magnet 7 m closer to the target than the RICH mirrors [2]. Care was taken that the focal plane remain within the mirror's depth of focus and not further increase the intrinsic single photon resolution. 0168-9002/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 9 9 ) 0 0 2 9 4 - 6 D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 137 Fig. 1. HERA-B RICH detector schematic [3]. As seen in Fig. 1, the radiator occupies the "nal 25% of the distance from the target to the mirrors. Therefore the Cherenkov photon emission angle and position would imply a distribution of photons in the range 0 40420 where 0 is the polar angle with respect to the focal surface normal and 0 "52 mrad. Studies [2] using realistic produc tion models and detector simulation indicate that Cherenkov photons will have an angular distribution at the chosen focal surface of 04100 mrad. 2.1. A two lens solution The basic design consists of a planoconvex "eld lens, the planar surface being placed at the chosen focal surface, and a biconvex circular lens as shown in Fig. 3. UV transmitting acrylic can be molded simply and e$ciently and makes an excellent choice of material for the lenses. By placing the biconvex lens at the focus of the "eld lens, the required magni"cation factor of 2 implies that f "2f where f and f are the focal lengths for the "eld and biconvex lenses, respectively. Equivalently, the total length from the focal plane to photocathode will be 1.5f in the thin lens approximation. A few initial test rays quickly show that lenses of reasonable thickness, F number (,f/lens diameter) &1.5, can be designed for the choice of f "95 mm. The diameter of the bicon vex lens is chosen to be 32 mm as a compromise SECTION III 138 D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 Fig. 2 . PMT map of the photon detector [3]. The R and f for each lens are taken to be G G correlated through the uncorrected Lens Maker's Formula. With the choice of f "95 mm the "ve remaining parameters of the telescope to be determined are, Fig. 3. A schematic of the proposed telescope solution. between aperture and mechanical assembly considerations. As previously noted, the Cherenkov light does not arrive at the telescope parallel to the optical axis. This causes image distortions at the photocathode. By using aspheric lenses, these distortions may be minimized. The aspheric surfaces may be parameterized by introducing a coe$cient, b where G i"1,2 for the planoconvex and biconvex lenses, respectively, to the second order term in the expansion of a sphere with radius R. Suppressing the constant term in such an expansion gives, b o o z"$ 1# G 2R 4 R G G as the equations for the curved surfaces in Fig. 3. The R represent the `nominala radii of curvature G for each lens surface. E b for i"1,2, i.e. the aspheric corrections for both G lenses, E positions of the biconvex lens and photocathode relative to the "eld lens, E and f . With the telescope parameterized in this fashion, all that is now needed to optimize the telescope is a reasonable measure of the imaging performance properties pertinant to the RICH. 2.2. Optimization of parameters Ray tracing is used to determine the focal properties of the telescope. The rays are generated on a 2 mm grid at the planar surface of the "eld lens according to a #at angular distribution in the range 0 40420 where 0 is the polar angle with re spect to the optical axis of the telescope. Chromatic dispersion is a negligible contribution to the focal properties of the lenses. The spectral sensitivity of the PMTs is averaged and monochromatic rays are generated. The refractive index of the acrylic material for this average wavelength is n"1.51. D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 139 Fig. 4 . RICH photon detector module [3]. Table 1 The telescope parameters found by the optimization scheme. The distances d and d are shown in Fig. 3 b b f d d !1.32 !1.52 30.27 mm 101.67 mm 47.85 mm The focal point of the "eld lens and the "nal focus at the PMT photocathode are of interest when developing a reasonable measure of the imag- ing performance. `Focusa is de"ned as "q (z)!q( (z)" G N where z is measured along the optical axis, q is the transverse vector of a ray and q( is the average transverse vector for the N generated rays. A set of rays with the same incidence angle at each grid point provides a measure of the "eld lens focal point. A set of rays in the range 0 40420 at SECTION III 140 D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 Fig. 5. Transmission of injection molded lenses. a single grid point provides a measure of the "nal focus at the PMT photocathode. Clearly, the size of these focal points is a minimum for the optimal telescope. Aperture losses at the biconvex lens and the photocathode are also minimized to keep the photon collection e$ciency high. Minimization of the focal points and aperture losses does not necessarily imply a telescope with a magni"cation of 2. A Gaussian term to regulate the magni"cation of the telescope to 1% is needed to ensure an appropriate solution. A simple summation of these positive de"nite considerations on the telescope imaging performance is taken as the appropriate measure to minimize. The chosen measure re#ects a desire for large collection e$ciencies and the lack of any a priori knowledge of correlations. Table 1 gives the parameters which de"ne the telescope design in use at the HERA-B RICH. The "eld and biconvex lenses are 6.78 and 9.45 mm thick with minimum thicknesses of 1 and 2 mm, respectively. Minimum thicknesses were adopted to simplify the lens molds. 2.3. Telescope performance Performance of the optimized telescope can also be estimated by ray tracing. Rastor plots are used to qualitatively assess the focusing aberrations of the telescope. The path length of simulated Cherenkov rays and the re#ection losses at the lens surfaces are estimated to be 415 mm and 415%, respectively. Manufacturer's data for acrylic D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 141 Fig. 6 . 12;12 simulated photocathode array as viewed from 3 m. indicates there is &1%/mm loss in total collection e$ciency due to the absorption of detectable Cherenkov photons. 2.4. Telescope assembly The telescope assembly is shown in Fig. 4 mounted in front of photomultipliers. The photomultipliers are mounted in an iron frame set in molded plastic which provides a rigid support frame and magnetic shielding. The lenses are mounted to an aluminum frame made from 1 mm thick "ns which are also imbedded in molded plastic. At the position of the "eld lens, the aluminum "ns are scalloped to allow the lenses to be glued to both the surrounding lenses and the framing. The biconvex lenses were placed and glued directly onto the plastic molding. Positioning jigs were used to accurately place the lenses for gluing to the frame. The result is a rigid assembly of 2;12 telescopes weighing &300 g. 3. Results The designed lenses were injection molded by Wahl Kunststo!optik, GmbH. The measured surfaces of the delivered lenses are well within the 0.01 mm tolerances given by the manufacturer. The acrylic remembers the mold stresses as indicated by an orientation dependent transmission, Fig. 5. However, the transmission of the acrylic compares well with the quantum e$ciency of the normal SECTION III 142 D.R. Broemmelsiek / Nuclear Instruments and Methods in Physics Research A 433 (1999) 136}142 Fig. 7 . 12;12 simulated photocathode array as viewed from 1.5 m. bialkali Hamamatsu photocathode with a borosilicate window. To test the "nal quality of the telescope assembly, a 12;12 array of simulated photocathodes was backlit and viewed from 3 and 1.5 m, Figs. 6 and 7, respectively. From Fig. 6, the simulated photocathode array is seen to be magni"ed with high "delity and uniformity of cellularization. The framing losses are clearly seen to be at the 5% level. Additionally, the inversion properties of the telescope are explicitely shown by the lettering. Each telescope is on 36 mm centers so that Fig. 7 explicitely shows the aperture cuto! at 140 mrad ("216 mm/1500 mm) as expected from the ray tracing simulations. The losses are due primarily to the aperture of the biconvex lens and do not represent collection losses at the photocathode. 4. Conclusion The timely development of the Hamamatsu multianode photomultipliers has enabled the HERA-B RICH to deploy a large area, 3.25 m, photocathode. An optical telescope, optimized for the RICH, is an essential component. References [1] Samo Korpar, Nucl. Instr. and Meth. A 433 (1999) 128. [2] Marco Staric, HERA-B Note 96-290. [3] JoK rg Pyrlik provided these technically accurate drawings.