From 3D to 5D tracking: SMX ASIC-based double-sided micro-strip detectors for comprehensive space, time, and energy measurements

We present the recent development of a lightweight detector capable of accurate spatial, timing, and amplitude resolution of charged particles. The technology is based on double-sided double-metal p+ – n – n+ micro-strip silicon sensors, ultra-light long aluminum-polyimide micro-cables for the analogue signal transfer, and a custom-developed SMX read-out ASIC capable of measurement of the time (Δt ≲ 5 ns) and amplitude. Dense detector integration enables a material budget > 0.3 % X 0. A sophisticated powering and grounding scheme keeps the noise under control. In addition to its primary application in Silicon Tracking System of the future CBM experiment in Darmstadt, our detector will be utilized in other research applications.


Introduction: Silicon Tracking System of the CBM experiment
The Silicon Tracking System (STS) of the Compressed Baryonic Mater (CBM) experiment is the core application and main motivation for the detector technology described in this paper.
CBM is a fixed-target heavy-ion experiment of the future Facility for Antiproton and Ion Research (FAIR) complex in Darmstadt, Germany; it is dedicated to study the strongly interacting matter under extreme conditions [1].
The CBM detector is a single-arm forward spectrometer capable of collecting data in a freestreaming mode to process the unprecedented beam-target interaction rates of up to 10 MHz.The detector will utilise the heavy-ion and proton beams from the SIS-100 accelerator at energies of up to 11 GeV and 29 GeV [2].
STS is the core tracking detector of CBM.Its 8 tracking stations (876 detector modules, total silicon area about 4 m 2 ) will be placed 30 − 100 cm downstream of the target in the aperture of the 1 T•m superconductive dipole.The primary goals of the STS are tracking of charged particles (≲ 700 tracks in the Au+Au central collision), momentum determination with  / ≲ 1.5% and secondary vertex reconstruction.For these tasks STS requires a position resolution better than 30 m in the bending plane, a good time resolution (in the order of 5 − 10 ns) and a material budget within 0.3% − 1.4% 0 per tracking station [3][4][5][6].

Micro-strip Silicon module: the fundamental functioning block of STS
A key component of the STS detector module is the custom STS-MUCH-XYTER (SMX) Application-Specific Integrated Circuit (ASIC) developed for the front-end electronics of the STS and Muon Chamber (MUCH) detectors of CBM.It features 128 channels each capable of simultaneous measurement of the signal amplitude and time with 5-bit ADC (dynamic range < 15 fC for the STS and < 100 fC for MUCH modes) and 14-bit TDC (Δ LSB = 3.125 ns) in a free-streaming mode [7].Picture of the partially assembled E16-104 module with 62 × 62 mm 2 sensor and 121 mm micro-cable.Micro-cable shielding, aluminium cooling fin and return path circuit wire are missing.
An STS detector module consists of a 320 m thick, 62 mm wide p+ -n -n+ double-sided double-metal (DSDM) silicon microstrip sensor of four versions: 22 mm, 42 mm, 62 mm and 124 mm long.The sensors are manufactured by Hamamatsu Photonics [8] using high-resistance silicon wafers: typical dark current is   < 40 nA/cm 2 at 20 • C [9][10][11].All sensors underwent an automatised optical inspection [12].There are 1024 strips with a pitch of 58 m on each side; the p-side strips are inclined with respect to the side edge at an angle of 7.5 • .The AC readout pads are located in two rows on the top and bottom of the sensor.Each sensor is connected to two front-end boards (FEBs), each containing eight SMX ASICs, via 32 custom ultra-lightweight aluminiumpolyimide microcables up to 500 mm long with the contribution to the material budget ≲ 0.3% 0 per module [13].A picture of the module during its assembling at GSI Detector Laboratory is shown in figure 1.More details concerning the module structure, construction, calibration and operation can be found in ref. [14].
Several versions of the detector Front-End Boards (FEB) were used during the setup developments: from the single-ASIC FEB-C (visible as a part of the prototype module in figure 2, left) to the present date FEB8 (part of the E16 STS module in figure 1).An alternative implementation of the SMX-based silicon micro-strip detector with few alternative FEB form-factors was done for the Strasse tracker [15].
To achieve the expected noise level during operation, the STS module implements a sophisticated ground and powering scheme.This is an important feature addressed in ref. [16].The HV and ground stability issues are also studied in ref. [17].
The Data Acquisition (DAQ) system of the STS detector together with other CBM subsystems feature radiation hard GBTx and Versatile Link based readout: details on the system architecture, protocol and particular implementations are provided in refs.[18] and [19].An alternative DAQ chain with commercial off-the-shelf components based on the GBTx-EMU is described in ref. [20] 3 Present and future detector applications A few dozens of prototype and pre-series modules has been assembled since 2018 [14].They were involved in multiple laboratory studies and beam tests, primarily with protons at COSY synchrotron (see dedicated chapters of refs.[21,22]) and heavy-ion collisions in mCBM at SIS18 [23].The mSTS detector is a fully-integrated functional STS prototype featuring eleven DSDM silicon strip modules of different sensor sizes (see figure 2, center); it is a part of the mCBM test setup [25].The mSTS detector was first proposed in 2017 as a part of the installation of the FAIR Phase 0 mCBM experiment which aim to prove the free-streaming DAQ and online reconstruction concept of CBM using high rate nucleus-nucleus collisions at SIS18 synchrotron.Since that time the mSTS setup went through two successful beam campaigns in 2018-2019 and 2020-2022 [26].
The E16 experiment at J-PARC aims to study in-medium modification of the vector mesons , , and , decaying via  +  − channel, with a high-intensity 30 GeV proton beam interacting with a C target and Cu targets at rates up to 40 MHz.It went through three commissioning runs in 2020-2021.Recently, the silicon strip detector was upgraded by replacing older silicon sensors with new ten STS modules built with 62 × 62 mm 2 sensors (see figure 2, right) [27]; it features the GBTx-EMU based DAQ chain [28].

Spatial resolution
High spacial resolution and detection efficiency are prime objectives of the tracking detectors.
According to simulations performed in ref. [29], Δ ≲ 15 m and single-hit efficiency  hit ≈ 97% were obtained.Preliminary analysis of the mSTS data supports these estimates.

Timing and rate capabilities
Timing performance of the STS detector modules was measured in several beam tests with relativistic particles.Time resolution of Δ = 6.9 ns was obtained with a proton beam of 1.7 GeV at COSY synchrotron (see figure 3, left) [22].Comparable results were obtained with the mCBM data using the prototype Time-of-Flight detector (TOF) as a reference (see figure 3, right) [23].Several independent studies were conducted to establish the signal rate capabilities of the STS modules: beam tests with mSTS so far showed rate capabilities up to (128 × 40) kHz/uplink = 5.1 MHz/uplink (figure 4, left); meantime, E16 team showed stable hit transmission up to 9 MHz/uplink with the digitally periodically generated SMX hits (figure 4, right).

Energy measurements
Energy loss by a relativistic charge particle penetrating a thin detector follows Landau-Vavilov distribution [30,31].The width factors  and √  2 , and most probable value Δ  are usually used to describe the width and the position of Landau-Vavilov distribution; they both depend on the thickness  and material of the detector [32].With the mean energy loss of a minimum ionizing particle (MIP) being ⟨Δ⟩ = 388 eV/m we estimate Δ  ≈ 84 keV [32],  = 5.7 keV [33], and √  2 = 7.6 keV [34] for MIP in 320 m silicon.With the mean energy for electron-hole pair production in silicon  Si = 3.64 eV [33], the most-probable charge deposition is   = Δ    Si = 23.1 × 10 3 e = 3.7 fC.Taking into account the Poisson distribution for the electron-hole creation and a Fano factor for silicon  = 0.11 [35], we can finally estimate a typical variation of the charge deposit in the detector: The SMX chip allows us to measure the charge deposited in the micro-strip silicon sensor with its slow channel and 5-bit flash ADC: thus, assuming only perpendicular tracks resulting in uniquely one-strip clusters and dynamic range of [0.8, 14.0] fC the charge uncertainty resulting from the amplitude discretization is about: = 0.12 fC 3.7 fC = 3.3%, ( which is smaller than uncertainty resulting from the conservative estimate of the effective noise charge (ENC) of  ENC = 1200 e ≈ 0.2 fC from ref. [14]: Comparing equations 6.2 and 6.3 to equation 6.1 one can see that the measurement precision is dominated by the nature of the interaction rather than by the read-out electronics.Particle Identification (PID) using the charged particle energy deposition per unit length of detector material (specific energy loss) is a widely used approach in particle and nuclear physics.Heavy-ion experiments typically use information about the continuous energy loss in their gaseous trackers [36,37].To some extent, this approach is also applicable to silicon tracking detectors, as has been shown with data from the CMS Tracker [38] and ALICE ITS (see figure 5, left) [37,39].Independent measurements of the energy deposit in multiple tracking stations require averaging.As the arithmetic mean does not reduce the uncertainty, CMS Tracker team uses ⟨⟩ = Σ  1/ 2  −1/2 , while ALICE ITC team uses a truncated mean.We obtained the best results using the median energy value; the bands of the median energy deposits for various particles are shown in figure 5, right.
Applying a simple threshold of Δ/Δ > 80 × 10 3 e/300m on the median charge deposited in STS detectors results in almost complete separation of double-charged particles from singlecharged particles in the STS [41].According to CBM simulations, in the case of hyper tritium decay 3  Λ H → 3 He  − it helped to increase the signal purity by a factor of 50, as shown in Fig. 6: the main contamination in this channel are misidentified protons and deuterons, which are overlapped with helium isotopes in TOF detector data [40].

Figure 1 .
Figure 1.Picture of the partially assembled E16-104 module with 62 × 62 mm 2 sensor and 121 mm micro-cable.Micro-cable shielding, aluminium cooling fin and return path circuit wire are missing.

Figure 2 .
Figure 2. Left to right: an early prototype module with two FEB-C prepared for the beam tests at COSY in 2018; mSTS detector during the upgrade in 2021; E16 STS at J-PARC in 2022 (picture from ref. [24]).

Figure 3 .
Figure 3. Measured detector time resolution in the mSTS setup with the time-walk correction applied [23].
Beam intensity [ions/sec] Hit rate per channel [hits/sec]

Figure 4 .
Figure 4. Typical hit rate per SMX channel during the mCBM beam test from ref. [6], corrected (left).Detection efficiency per chip measured with ASIC hit generator with E16 STS module; the upper data transfer rate limit of 9.41 MHz for one uplink (corresponding to one chip) is indicated by the vertical line (right) [28].

Figure 5 .
Figure 5. Performance studies of the particle identification in the ALICE ITS with Pb+Pb at √  NN = 2.76 TeV (left) [37].PID in CBM STS with separately simulated thermalized reaction products from Au+Au at √  NN = 4.3 GeV (right) [40].