Key Laboratory of Particle and Radiation Imaging of Ministry of Education, Department of Engineering Physics, Tsinghua University, Beijing 100084, China
Fund Project:Project supported by the Program of Ministry of Science and Technology of China (Grant Nos. 2015CB856905, 2008CB8177072, 2016 YFA0400100) and the National Natural Science Foundation of China (Grant Nos. 11420101004, 11461141011, 11275108, 11735009).
Received Date:13 December 2018
Accepted Date:25 March 2019
Available Online:01 May 2019
Published Online:20 May 2019
Abstract:Particle identification is very important in nuclear and particle physics experiments. Time of flight system (TOF) plays an important role in particle identification such as the separation of pion, kaon and proton. Multi-gap resistive plate chamber (MRPC) is a new kind of avalanche gas detector and it has excellent time resolution power. The intrinsic time resolution of narrow gap MRPC is less than 10 ps. So the MRPC technology TOF system is widely used in modern physics experiments for particle identification. With the increase of accelerator energy and luminosity, the TOF system is required to indentify definite particles precisely under high rate environment. The MRPC technology TOF system can be defined as three generations according to the timing and rate requirement. The first-generation TOF is based on the float glass MRPC and its time resolution is around 80 ps, but the rate is relatively low (typically lower than 100 Hz/cm2). The typical systems are TOF of RHIC-STAR, LHC-ALICE and BES III endcap. For the second-generation TOF, its time resolution has the same order as that for the first generation, but the rate capability is much higher. Its rate capability can reach 30 kHz/cm2. The typical experiment with this high rate TOF is FAIR-CBM. The biggest challenge is in the third-generation TOF. For example, the momentum upper limit of $ {\rm{K}}/{\text{π}}$ separation is around 7 GeV/c for JLab-SoLID TOF system under high particle rate as high as 20 kHz/cm2, and the time requirement is around 20 ps. The readout electronics of first two generations is based on time over threshold method, and pulse shape sampling technology will be used in the third-generation TOF. In the same time, the machine learning technology LSTM network is also used to analyze the time performance. As a very successful sample, MRPC barrel TOF has been used in RHIC-STAR for more than ten years and many important physics results have been obtained. A prominent result is the observation of antimatter helium-4 nucleus. This discovery proves the existence of antimatter in the early universe. In this paper, we will describe the evolution of MRPC TOF technology and key technology of each generation of TOFs including MRPC detector and related electronics. The industrial and medical usage of MRPC are also introduced in the work finally. Keywords:multi-gap resistive plate chamber/ time resolution/ slewing correction/ machine learning/ pulse shape sampling
其中${\sigma _{{\rm{TOF}}}}$为飞行谱仪系统的时间分辨. 图1显示了不同时间分辨的飞行时间谱仪系统的粒子鉴别能力. 可以得到, 系统时间分辨越高, 粒子鉴别能力越强. 在飞行距离为8 m时, 对于60 ps 的时间分辨, 对动量为4 GeV/c的$ {\text{π}}/{\rm{K}}$鉴别能力可达3σ, 如果要求对于7 GeV/c的$ {\text{π}}/{\rm{K}}$鉴别能力达到3σ, 则时间分辨需达到20 ps. 图 1 几种不同时间分辨飞行时间谱仪系统的$ {\text{π}}/{\rm{K}}$鉴别能力, 飞行距离L = 8 m Figure1.$ {\text{π}}/{\rm{K}}$ separation power of TOF system with different time resolution, flight distance L = 8 m.
表2低电阻玻璃性能 Table2.The performance of low resistive glass.
由于MRPC电极间的气隙窄, 气隙中的场强高, 因此对电极材料的厚度均匀性、表面光滑度均有很高要求, 我们研制的低电阻玻璃这些主要性能与浮法玻璃接近, 实验证明可以用作MRPC的电极材料. 另外玻璃高压测试累积电荷达1 C/cm2, 这相当于CBM-TOF最高计数率区域工作五年的累积电荷, 能够保证探测器的长期稳定工作. 采用此低电阻玻璃, 我们研制了读出块和读出条的高计数率MRPC原型, 并赴德国德累斯顿Helmholtz-ZentrumDresden-Rossendorf(HZDR)采用其强流电子束流测试了探测器在强束流下的性能, 结果如图9所示. 可以看出, 探测效率和时间分辨均受计数率的影响, 即使计数率达到70 kHz/cm2, MRPC探测器效率仍高于90%, 时间分辨优于80 ps. 原型探测器的性能大大超过了CBM-TOF的要求. 图 9 测试得到的MRPC探测效率和时间分别随粒子计数率的变化[15] Figure9. Measured efficiency and time resolution of MRPC change with particle rate.
CBM合作组已采用我们的技术建造飞行时间谱仪系统. 图10为CBM-TOF探测器结构图[16], 其中MRPC1, MRPC2和MRPC3a均采用低电阻玻璃制造, MRPC3b和MRPC4采用超薄浮法玻璃制造, 总面积约120 m2, 电子学道数达10万道. 前端电子学采用PADIX, 时间数字化采用GET4, 电子学的时间抖动小于30 ps. 图 10 CBM-TOF结构 Figure10. The structure of CBM-TOF.
清华大学负责了高计数率MRPC3a的设计与制造, 中国科学技术大学负责MRPC3b和MRPC4的设计建造. 这三种探测器的结构类似, 只是电极材料不同. 图11所示为MRPC3a的照片. 该探测器采用两层结构, 每层4个气隙, 共8气隙, 气隙宽度为0.25 mm. 探测器共有32个信号读出条, 读出条尺寸为27 cm × 1 cm. CBM飞行时间探测器模块由五个MRPC组成, 为了减小噪声, 前放PADIX也放置在气盒中, 如图12所示. 图 11 MRPC3a探测器照片 Figure11. Picture of MRPC3a.
图 12 由5个MRPC和相应电子学组成的飞行时间探测器模块 Figure12. CBM-TOF module is consisted of 5 MRPC counters and related electronics.
采用束流测试MRPC3a探测器的性能, 结果如图13所示[17]. 可以看出, 探测器时间分辨达50 ps, 效率达97%, 簇大小为1.6. 这些性能均达到或超过CBM-TOF的要求, 可以用于建造CBM-TOF系统. 图 13 不同PADI阈值下, MRPC3a探测器的时间分辨, 探测效率和簇大小 Figure13. Time resolution, efficiency and cluster size of MRPC3a at different threshold of PADI.
CBM-TOF的电子学由德国GSI实验室研发, ASIC放大器为PADIX[18], TDC为GET4[19]. 二者组成系统的时间抖动小于30 ps, 保证了CBM-TOF系统的高分辨时间性能. 目前我们已经开始高计数率MRPC的批量生产, 图14显示了在同方威视公司密云生产基地批量生产的照片. 图 14 同方威视公司密云生产车间正在进行高计数率MRPC的批量生产 Figure14. High rate MRPC were produced at Miyun manufacture base of NUCTECH Ltd.
要使 σTOF小于20 ps,则MRPC的时间抖动σMRPC和电子学系统的时间抖动σelectronics都必须小于 14 ps. 我们知道, 窄气隙MRPC的本征时间分辨可达10 ps, 但是第一\二代飞行时间谱仪所用的电子学 NINO (PADIX) + HPTDC (GET4)的时间抖动一般大于20 ps. 这样必须采用高速波形采样技术如高速开关电容阵列SCA或者高速FADC. 这种技术路线如图16所示. MRPC包含32个气隙, 气隙宽度为104 μm. 高速电流放大器需采用差分输入, 带宽大于350 MHz, 高速波形采样可以采用DRS4-V5芯片, 其采样率可达5 GHz. 图 16 高时间分辨MRPC及读出电子学 Figure16. High resolution MRPC and read out electronics.
一般地, 根据得到的输出波形, 可以采用常规的过阈定时和时幅校正技术分析MRPC的时间性能. 由于上述方法只利用了波形的过阈时间点和波形积分信息, 忽略了波形上升沿、达峰时间点等关键信息, 因此常规分析方法存在一定局限性. 因此可以采用先进的深度学习方法来得到入射粒子到达MRPC的时间点, 准确地说是入射粒子在MRPC中发生初始电离的时间点. 如图17所示, 可以采用机器学习方法, 从信号波形得到粒子到达MRPC的精确时间点ta. 通过构建深度神经网络, 搭建完整的MRPC蒙特卡罗模拟系统, 为神经网络提供训练样本, 得到粒子入射到MRPC的精确时间点[22,23]. 图 17 粒子到达MRPC的时间点${t_a}$可以由信号波形前沿得到 Figure17. The time point ${t_a}$ of particle arriving at MRPC can be obtained from pulse shape.
为此, 我们建立一套完整的MRPC模拟系统, 从模拟工作气体参数开始, 综合考虑初级电离能量沉积、电离位置分布、电离的雪崩倍增、信号感应与成型以及电子学响应等过程, 模拟得到MRPC探测器的输出信号, 以此作为深度学习的样本. 采用长短期记忆网络(LSTM)进行学习, 如图18所示. 其输入为信号前沿各时间点, 输出为粒子到达时间点${t_a}$. 图 18 用于MRPC时间重建的LSTM网络架构 Figure18. The structure diagram of LSTM network used for time reconstruction of MRPC.
对图16中提出的32气隙MRPC进行模拟分析, 结果如图19所示. 分别使用时幅校正和神经网络LSTM进行探测效率和时间分辨率的分析, 可以看出, 当使用LSTM方法时, 效率一样, 但时间分辨较好. 两种分析方法得到的效率坪区的时间分辨均优于20 ps. 这证明了设计方案的可行性. 图 19 模拟得到MRPC探测效率和时间分辨随气隙场强的变化, 可以看出, 采用LSTM网络法重建出的时间分辨比时幅校正得到结果要好 Figure19. Simulated efficiency and time resolution of MRPC change with electric field in the gas gap. It can be seen the time resolution reconstructed with LSTM network is better than with slewing correction.
同时也进行了实验验证. 研制出图16所示两个结构相同的MRPC探测器, 电子学采用高速前放和基于DRS4[24]的波形采样电路, 采用宇宙射线进行了测试. 工作气体为90%的氟利昂, 5%的异丁烷和5% SF6的混合气体. 气隙中工作场强为150 kV/cm, 达到了效率坪场强. 分别采用时幅校正和LSTM网络方法分析MRPC的时间分辨. 采用时幅校正得到的时间分辨是19.8 ps. 图20是采用神经网络分析的结果, 可以看出, 神经网络分析结果较好, 达到16.7 ps. 图 20 采用LSTM网络方法分析得到MRPC的测试时间谱 Figure20. Time spectrum of MRPC in cosmic test analyzed with LSTM network.