480 likes | 736 Views
江门中微子实验的探测器模拟. 高能所 邓子艳. 2014 年 4 月 21 日 湖北 武汉 中国物理学会高能物理分会第九届全国会员代表大会暨学术年会. 江门中微子实验( JUNO ). 顶部探测器. 中心探测器. 水切伦科夫探测器. JUNO 离线软件系统. 模拟软件: 基于 Geant4 的探测器模拟 读出电子学模拟 本底混合. DetSim. ElecSim. TriggerSim. ReadoutSim. 探测器模拟软件. 在探测器设计阶段,探测器模拟软件的必要性
E N D
江门中微子实验的探测器模拟 高能所 邓子艳 2014年4月21日 湖北 武汉 中国物理学会高能物理分会第九届全国会员代表大会暨学术年会
江门中微子实验(JUNO) 顶部探测器 中心探测器 水切伦科夫探测器
JUNO离线软件系统 • 模拟软件: • 基于Geant4的探测器模拟 • 读出电子学模拟 • 本底混合 DetSim ElecSim TriggerSim ReadoutSim
探测器模拟软件 • 在探测器设计阶段,探测器模拟软件的必要性 • 针对不同探测器设计方案的模拟,为探测器设计提供参考依据 • 能量分辨率 • 本底事例率 • 模拟软件可用于顶点和能量重建算法的研究 • 模拟软件用于刻度研究 • 一方面,要保证模拟软件的可用性 • 为探测器设计服务 • 及时给出模拟结果 • 另一方面,在整个离线软件框架下,软件功能的逐步完善 • 实现完整的模拟流程 • 产生子->探测器模拟->电子学模拟->触发模拟->读出系统模拟 • 统一的几何管理 • 良好的用户接口 • 速度和性能的优化
Generator && Interfaces • Generator • Inverse beta decay • Written in C++, imported from Dayabay • Radioactivity • Uranium(238U), Thorium(232Th), Potassium(40K) • AP generator,Written in Fortran by Andreas Piepke • GenDecay (in C++) • Generator interfaces • Support to generate: • Particles with specified name and momentum • At fixed position (x0,y0,z0) • Randomly on a surface R=R0 • Randomly inside a spherical volume R<R0,R1<R<R2 • Randomly in specified material, for example, PMT glass
GenDecay 1.数据来源:Evaluated Nuclear Structure Data File (ENSDF) Data Sets (记录各种原子核的可能衰变,及其分支比) 2.通过ENSDF的数据来构建衰变链 (ENSDF的数据需要使用libmore来转换) NucState: 描述原子核的类,存储了Z/A,半衰期,能级,以及可能发生的衰变(NucDecay) NucDecay: 描述一个衰变,存储了指向母核和子核的指针(NucState*),衰变类型(α,β,γ),衰变释放的能量,以及分支比
An example:U-238 to Th-234 U-238 level=0 keV[0] A DECAY E=4.038 MeV br=0.0780172% Th-234 level=163 keV[0] Gamma E=0.1135 MeV br=100% A DECAY E=4.151 MeV br=20.9046% Th-234 level=49.55 keV[0] Gamma E=0.04955 MeV br=100% A DECAY E=4.198 MeV br=79.0174% Th-234 level=0 keV[0]
Optical parameters && Geometry • Optical parameters • Based on DYB (tuned to data), except QE and absorption length • maxQE • from 25% -> 35% • LS absorption length • From 25m -> 40m • LS attenuation length • from 15m = absorption 25m + Rayleigh scattering 40m • to 20 m = absorption 40m + Rayleigh scattering 40m • Geometry • 20k ton liquid scintillator • Buffer thickness in front of PMT is 0.8m Acrylic tank:F35.4m Stainless Steel tank :F38.4m F=0.8m, G=0.7m A=17.7m, B=1.05m, C=0.45m A+B=18.75m (position of PMT sphere center) B+C=1.5m, Radius of PMT: 0.254m
Detector design options DetSim1 有机玻璃罐方案 LS/sphericalacrylic vessel/water/ steel truss DetSim2: 模块方案 LS/acrylic module/steel tank DetSim3: 气球方案 LS/balloon/steel tank buffer: mineral oil Triangle module or single PMT module LAB inside modules, With LS in the gap between modules
Geometry DetSim1 && DetSim3 && DetSim2(single PMT module) DetSim2 (Triangle Module) PMT sphere center: @18.75m Total: 17520 PMTs Coverage: ~77% PMT sphere center: @18.75m Total: 16720 PMTs Coverage: ~76.7% 。
Compare between different detector options • (1)Radioactivity background event rate • From PMT glass • From LS • From Acrylic • From balloon • From dust on balloon • From steel
14 radioactive decays in the chain 10 radioactive decays in the chain
Generator BG event rate from PMT glass (estimated with Schott glass) Mass of glass per PMT = 10kg 238U && 232Thgenerator: Written in Fortran, by Andreas Piepke If estimate with Schott glass, the generator event rate is ~1MHz If estimate with DYB PMT glass, the generator event rate is ~8MHz
模块方案 vs 气球方案 vs 有机玻璃罐方案 F=80cm 238U background from PMT glass
本底事例率随屏蔽层厚度的变化 (~0.5MeV) F=30/60/80cm (238U+232Th+40K Background from PMT glass)
Background from dust/balloon/LS Background from dust on balloon surface Background from balloon
Summary of single rates (0.8m buffer) • 对有机玻璃罐方案和气球方案,增加屏蔽层厚度可有效减小single rates • DetSim1(Schott glass): 1.1m(30Hz), 1.4m(1Hz) • DetSim3(Schott glass): 1.1m(100Hz), 1.4m(20Hz) • 而对模块方案则不然
Compare between different detector options (2)Energy resolution
Non-uniformity • 1MeV gamma generated at R<17.7m, energy not deposited totally at R>17m • 有机玻璃罐方案: optical photons total reflected at large R because of large difference of the refraction index between water and acrylic Refraction index Non-uniformity R=15.7m R=17m
有机玻璃罐方案的能量重建 • Strategy I: total charge corrected with calibration curve
有机玻璃罐方案的能量重建 • Strategy II: charge likelihood, consider the contribution of total reflection in the reconstruction
有机玻璃罐方案的能量重建 • Strategy III: charge likelihood, doesn’t consider the total reflection at first, and then correct the non-uniformity through calibration
气球方案的能量分辨率 • LS+Oil: using IBD events, applying 17m vertex cut, and correcting the non-linearity Charge Likelihood shows better performance than total PE
有机玻璃罐方案的能量分辨率 • LS+Water: Total Charge/Charge Likelihood (Strategy III)
有机玻璃罐方案 vs 气球方案 For total charge, there is little difference in energy resolution for LS+water and LS+Oil • Total Charge: comparison between LS+Oil and LS+Water
有机玻璃罐方案 vs 气球方案 For charge likelihood method, the performance of LS+Oil is slightly better than LS+Water • Charge Likelihood: comparison between LS+Oil and LS+Water (with 17m cut)
Background Mixing • Generate U/Th/K events randomly in PMT glass • Generate gamma/IBD events in LS • Mixing without electronics simulation • Using the generator event rate to sample the number of background events in one time window (for example: 300ns) • Sum the total p.e. of signal and background • Mixing during electronics simulation • Sample the BK event start time in one time window • Mix the hits from signal and background for each PMT • Pulse integration for each PMT
Mixing BG at different generator levels , without electronics simulation Detector type: DetSim2 Single PMT module, 30cm LAB E_true vs E_rec for e+ from IBD
Mixing BG at different generator levels , without electronics simulation Detector type: DetSim2 Single PMT module + balloon, 30cm LAB E_true vs E_rec for e+ from IBD
Mixing BG at different generator levels , without electronics simulation Detector type: DetSim2 Single PMT module + balloon, 80cm LAB E_true vs E_rec for e+ from IBD
Background mixing during ElecSim • A simple NHit trigger is implemented in ElecSim • Signal and background(from U/Th/K) are accidental coincidence in one readout window Single gamma energy spectrum after BG mixing DetSim2: single PMT module + balloon, 30cmLAB, mix 10 MHz BG
现阶段探测器设计方案 经过3月份的中心探测器方案评审之后,排除了本底事例率高的模块方案,现阶段中心探测器的首选方案是有机玻璃罐方案,备选方案是气球方案。
New PMT个数18306个 覆盖率77.65% PMT前端屏蔽层厚度: 1.4~1.5m Acrylic tank:F35.4m Stainless Steel tank :F39.9m A=17.7m (LS) B=(0.12+1.426+0.254)m=1.8m C=0.45m A+B=19.5m (PMT place) A+B+C=19.95m
光学参数 吸收长度 发射谱 瑞利散射长度 吸收重发射几率 量子效率 折射率 光产额:10400/MeV 液闪的瑞利散射长度 30 m @430nm 液闪的衰减长度 20 m @430 nm 液闪的吸收长度为 60 m @430 nm
PMT保护罩 PMT防爆罩形状与PMT形相同, 厚度为9mm,与PMT有1mm的空隙,空隙内填充水。 有机玻璃罐方案 气球方案 加或不加PMT保护罩的TotalPE分布
Muon simulation • 高能muon穿过大体积的液闪探测器,光学模拟过程速度非常慢,内存占用量大 • 可以通过对muon事例的光产额进行scale,达到提高模拟速度的目的 • 对muon事例设计新的数据结构,减小内存占用量 Generate Mu- at (0, 0, 20m) Momentum = (0, 0, -P)
Data Structure for muon events Only save firstHitTime and nPE (Merge the hits in a time window) Memory Usage: < 1GB
Summary • A simulation tool for preliminary study of detector design, calibration, and reconstruction has been developed for JUNO • Performance of different detector design options have been compared • DetSim algorithm has been integrated into JUNO offline framework • Detailed detector simulation for two options is still on-going
模块方案,来自PMT玻璃的天然放射性本底事例率模块方案,来自PMT玻璃的天然放射性本底事例率 屏蔽层厚度 F=30cm/60cm/80cm 238U+232Th+40K
Triangle Modules 5/21 PMT 5/26 PMT 5/17 PMT 5/21 PMT
Triangle Modules 0/28 PMT 5/26 PMT 5/21 PMT 5/26 PMT
Triangle Modules 5/26 PMT 9/26 PMT The PMTs marked with red numbers are in down layer, the other PMTs in up layer. Triangle side length is about (3-4)m. The buffer layer (LAB) in front of PMT in module is 80cm , the module length is about 1.919m. Compared with single PMT module: Less LS between modules. Longer module length because two layers of PMT in modules.
Event rate (Schott Glass) 如果为dyb玻璃本底水平大约是schott玻璃的10倍:19.5Hz。