TOF at 10ps with SiGe BJT Amplifiers Time resolution of solid state detectors and design of fast, low noise, charge amplifier Lorenzo Paolozzi Université de Genève Geneva - 17/06/2015

Time resolution of solid state detectors Geneva - 17/06/2015

Solid state detectors with planar geometries Why planar geometry with uniform field: Uniformity of electric field. Necessary to have same charge collection time disregarding particle hit position. Saturation of carrier drift velocity. Can be achieved more easily in the whole sensor with uniform field. Uniformity of Ramo Field. Necessary to have uniform output pulse rise time and to reduce time fluctuation due to charge collection noise. Geneva - 17/06/2015

Solid state detectors with planar geometries Intrinsic crystals used as particle detectors should have high band gap: Diamond Detectors Excellent carriers velocity. Excellent radiation hardness. High energy required to produce an electron-hole pair (13 eV). Detectors based on PN junction can have a lower band gap, since the free carriers concentration is reduced in the depletion region. Silicon Detectors Good carriers velocity. Good radiation hardness. Low energy required to produce an electron-hole pair (3.6 eV). Geneva - 17/06/2015

Solid state detectors with planar geometries Timing properties of the diamond and silicon crystals: For solid state detectors with planar geometries based on direct collection of the ionization charge time resolution is typically limited by the signal to noise ratio of the output pulse from the amplifier. 𝜎 𝑡 ≅ 𝑅𝑖𝑠𝑒 𝑇𝑖𝑚𝑒 𝑆𝑖𝑔𝑛𝑎𝑙 𝑡𝑜 𝑁𝑜𝑖𝑠𝑒 𝑟𝑎𝑡𝑖𝑜 Comparing the expected time resolution with an integrating front end electronics for silicon and diamond the result is: 𝑡 𝑟𝑖𝑠𝑒, 𝑠𝑖𝑙𝑖𝑐𝑜𝑛 ≅2 𝑡 𝑟𝑖𝑠𝑒, 𝑑𝑖𝑎𝑚𝑜𝑛𝑑 𝑄 𝑠𝑖𝑔𝑛𝑎𝑙, 𝑠𝑖𝑙𝑖𝑐𝑜𝑛 ≅2.4 𝑄 𝑠𝑖𝑔𝑛𝑎𝑙, 𝑑𝑖𝑎𝑚𝑜𝑛𝑑 𝜎 𝑡,𝑠𝑖𝑙𝑖𝑐𝑜𝑛 ≅0.83 𝜎 𝑡, 𝑑𝑖𝑎𝑚𝑜𝑛𝑑 Note: at this level time resolution does not depend on detector thickness for penetrating particles. Geneva - 17/06/2015

Charge collection noise Time resolution for solid state detectors depends on signal to noise ratio. The noise introduced by the source is typically considered negligible with respect to the one from the amplifier itself. For a time resolution below 100 ps, another source of noise should be introduced, that is charge collection noise1. When the ionizing particle traverses the detector, ionization occurs following Landau statistics. Most of the produced clusters have a small charge. Few events with very large transferred energy are possible. 1 L. Paolozzi, Development of particle detectors and related Front End electronics for sub-nanosecond time measurement in high radiation environment, PhD Thesis Geneva - 17/06/2015

Charge collection noise Being the induced current, from Shockley-Ramo’s theorem 𝑖 𝑖𝑛𝑑 =− 𝑞𝑣 𝐷 when the large clusters are absorbed at the electrodes, their contribution is removed from the induced current. The statistical origin of this variability of the induced current makes this effect irreducible, so that it can be considered as an equivalent noise current. Time jitter introduced by the charge collection noise for a silicon detector traversed by a Minimum Ionizing Particle (MIP). 50% Constant fraction threshold. Weightfield simulation2. 2 F. Cenna, N. Cartiglia, Weightfield2: A Simulation Program for silicon detectors, REMDD14, Firenze 2014. https://indico.cern.ch/event/313925/. Geneva - 17/06/2015

Sensor readout for a TOF PET scanner Strip readout to limit the number of readout channels: Sensor thickness =100 𝜇𝑚. Strip pitch =1 𝑚𝑚. Strip width ≅1 𝑚𝑚 to have uniform Ramo field. Double side readout with mean time measurement to obtain orthogonal coordinate and get rid of signal propagation time on the strip. Line impedance ~20 𝑂ℎ𝑚. Preamplifier must be adapted to the transmission line. Geneva - 17/06/2015

Design of fast, low noise, charge amplifier. Geneva - 17/06/2015

Target performance of the new preamplifier Target time resolution is 30 ps for the MPV of the deposited charge in the sensor. If the 100 𝜇𝑚 sensor is operated at 2−3 𝑉/𝜇𝑚, it can be demonstrated that the expected time resolution with a charge amplifier operating as an ideal integrator is 𝜎 𝑡 [𝑝𝑠]= 5.1 Δ𝐸[𝑘𝑒𝑉] 𝐸.𝑁.𝐶. A target Equivalent Noise Charge of 250 electrons or better is required. The amplifier should operate as an ideal integrator for pulses with duration well below 10 ns. Geneva - 17/06/2015

Studies on fast, low noise, charge amplifier. BJT technology has the best performance in terms of noise when a fast charge integration is operated on input pulses. 3 E. Gatti, P. F. Manfredi, Processing the Signals from Solid-State Detectors in Elementary-Particle Physics, rivista del Nuovo Cimento Vol. 9, No. 1 (1986). Geneva - 17/06/2015

Studies on fast, low noise, charge amplifier. The principle of custom charge amplifier in SiGe - BJT technology4: Example of Input pulse (above) Working principle and relative output pulse (below) Fast pulse integration Possibility to match the impedance of a transmission line 4 R. Cardarelli, A. Di Ciaccio, L. Paolozzi, Development of Multi-Layer Crystal Detector and related Front End electronics, Nuclear Instruments and Methods in Physics Research A 745 (2014) 82–87. Geneva - 17/06/2015

SiGe technology for very low noise fast amplifiers Charge gain of the charge amplifier: 𝐴 𝑄 = 𝑑 𝑉 𝑜𝑢𝑡 𝑑 𝑄 𝑠 =− 1 𝐶 𝑓 + 𝐶 𝑑𝑒𝑡 𝐴 𝑣 The equivalent noise charge, for fast integration time, is dominated by the series noise: 𝐸𝑁𝐶 𝑠𝑒𝑟𝑖𝑒𝑠 𝑛𝑜𝑖𝑠𝑒 ∝ 2𝑘𝑇 𝑆𝑁𝐼 𝐶 𝑑𝑒𝑡 + 𝐶 𝑖𝑛 2 ℎ 𝑖𝑒 𝛽 + 𝑅 𝑏𝑏 𝐶 𝑑𝑒𝑡 2 In order to reduce noise, high value of current gain and small base spreading resistance are necessary. Geneva - 17/06/2015

SiGe technology for very low noise fast amplifiers Amplifier current gain can be expressed as (NPN BJT) 𝛽= 𝑖 𝐶 𝑖 𝐵 = 𝜏 𝑝 𝜏 𝑡 𝜏 𝑝 =hole recombination time in base 𝝉 𝒕 =𝐞𝐥𝐞𝐜𝐭𝐫𝐨𝐧 𝐭𝐫𝐚𝐧𝐬𝐢𝐭 𝐭𝐢𝐦𝐞 (𝐄 𝐭𝐨 𝐂) Increase gain Reduce base width Reducing base doping Spreading resistance increases! Geneva - 17/06/2015

SiGe technology for very low noise fast amplifiers A possible approach: changing the charge transport mechanisms in the base from diffusion to drift. SiGe heterojunction bipolar transistor technology. Introduces an electric field in the base. Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology The technology chosen for the Full Custom chip design was the SG25H3 SiGe HBT from IHP. Transition frequency of the order of 𝑓 𝑡 =110 𝐺𝐻𝑧 with a 𝛽 value of 150 are available. The production consists of two multi-project test runs and is done via Europractice-ic service. Layout design and simulation for the first test run was carried out with Cadence. Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Chip schematic layout Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Input impedance for amplifier channel 1 (red) and 3 (blue) vs Supply Voltage Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Output pulse rise time vs input pulse duration for amplifier channel 1 (red) and 3 (blue) Both channels operate as ideal integrators for pulses with nanosecond and sub- nanosecond duration. Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Equivalent noise charge vs detector capacitance for amplifier channel 1 (red) and 3 (blue) Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Calibration curve for amplifier channel 1 coupled to post-amplifier channel 2. Ideal schematics (blue) and Extracted view (red) ( 𝑪 𝒅𝒆𝒕 =𝟏𝒑𝑭) Geneva - 17/06/2015

Development of the new full custom Front End in SiGe technology Discriminator pulse duration vs input pulse charge. The compression permits to extend the dynamics of the discriminator up to a factor 200 at constant relative error on the input charge. Geneva - 17/06/2015

The test preamplifier ASIC performance has been measured Geneva - 17/06/2015

The test preamplifier ASIC performance has been measured Geneva - 17/06/2015