TCAD + Allpix² Simulation study of MALTA2, a Depleted Monolithic Active Pixel Sensor for future tracking
Pith reviewed 2026-05-25 02:19 UTC · model grok-4.3
The pith
A hybrid TCAD-Allpix² framework matches MALTA2 active depth to measured values within 2 percent using only generic doping profiles.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The authors show that a TCAD-Allpix² pipeline with generic doping profiles can be tuned via the N-type blanket concentration to reproduce the measured active depth of the MALTA2 sensor to within 2 percent at a 450 e− threshold, thereby demonstrating that sensor performance quantities such as detection efficiency and cluster size can be studied without access to proprietary process information.
What carries the argument
Hybrid TCAD-Allpix² simulation chain that extracts doping profiles and electric fields from 3D transient TCAD runs and feeds them into Allpix² Monte Carlo simulations for efficiency and cluster-size predictions.
If this is right
- Detection efficiency and cluster size in MALTA2 are predicted to vary sharply with N-type blanket doping concentration.
- The same optimization procedure can be applied to other depleted monolithic sensors provided the same generic modeling assumptions hold.
- Active depth extracted from simulation at the optimal doping matches the grazing-angle result obtained in both DUT-only and full-telescope configurations.
- The framework supplies a complete prediction chain from doping profile to hit-level observables without proprietary layout data.
Where Pith is reading between the lines
- The approach could be extended to predict radiation-damage effects by inserting time-dependent trap models into the same TCAD-Allpix² chain.
- If the 2-percent agreement persists across multiple sensor variants, the method could reduce reliance on dedicated test-beam campaigns for early sensor prototyping.
- The grazing-angle technique combined with the simulation may allow extraction of depth-dependent charge-collection efficiency maps that are difficult to obtain by other means.
Load-bearing premise
Generic doping profiles and simplified well structures in the TCAD model are close enough to the real MALTA2 fabrication process that tuning only the N-type blanket concentration can bring simulated active depth into agreement with measurement.
What would settle it
Repeat the grazing-angle active-depth measurement on a MALTA2 sensor whose N-type blanket doping is known to be the value identified as optimal in the paper; the simulated depth should deviate by no more than 2 percent from the new measurement at 450 e− threshold.
read the original abstract
In this work, a hybrid simulation framework combining TCAD and Allpi$\text{x}^2$ is presented to investigate the sensor properties of MALTA2, a depleted monolithic active pixel sensor designed for future tracking. The study starts from 3D modeling and transient simulations in TCAD, with generic doping profiles and simple well structures. The resulting doping profiles and electric field are extracted and fed into Allpi$\text{x}^2$ for high-statistics Monte Carlo simulations in both DUT-only and full-telescope mode. Simulations reveal a strong dependence of sensor performance, specifically the detection efficiency and cluster size, on the doping concentration of the N-type blanket at the sensor surface. The doping concentration is then optimized by comparing simulations with measurement data. The active depth of the depleted region of the MALTA2 sensor is estimated in both simulations and measurements using a grazing angle method, in which the sensor is positioned at various inclinations relative to the beam, covering angles from 0 to 60 degrees. Excellent agreement on active depth is obtained with the optimal doping concentration, showing a deviation of 2\% from the measured value at a threshold of 450\,$\text{e}^-$. Consequently, the framework offers a generic toolkit for sensor studies without requiring proprietary information.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper presents a hybrid TCAD + Allpix² simulation framework for the MALTA2 depleted monolithic active pixel sensor. 3D TCAD models with generic doping profiles and simplified wells generate doping and electric-field maps that are imported into Allpix² for high-statistics Monte Carlo studies in DUT-only and telescope configurations. The N-type blanket doping concentration is tuned to reproduce measured detection efficiency and cluster size; the active depth of the depleted region is then extracted in both simulation and data via a grazing-angle scan (0–60°) and reported to agree within 2 % at a 450 e⁻ threshold. The framework is positioned as a generic, non-proprietary toolkit for sensor performance studies.
Significance. If the reported agreement is shown to be independent of the tuning observables, the work supplies a reproducible simulation pipeline that can explore depleted monolithic sensor behavior without access to proprietary process details. This is potentially useful for design iterations in future tracking detectors, provided the model’s predictive power outside the tuned observables is demonstrated.
major comments (2)
- [Abstract] Abstract and the optimization/active-depth comparison section: the single free parameter (N-type blanket doping concentration) is adjusted to match measured efficiency and cluster size. Because depletion depth directly governs collected charge (and therefore efficiency at normal incidence) and the grazing-angle method extracts depth from the angular dependence of the same observables, the subsequent 2 % active-depth agreement is not demonstrably independent of the fit; the tuned value may simply reproduce the effective depth already implicit in the efficiency data.
- [Abstract] The manuscript provides no information on the fitting procedure, the number of data points or angular settings used for the optimization, the uncertainty on the resulting doping value, or whether the active-depth comparison employs held-out data or a different observable set. Without these details the robustness of the 2 % agreement cannot be assessed.
minor comments (2)
- Clarify the exact definition of “active depth” extracted from the grazing-angle scan and how it is computed from the simulated charge-collection maps.
- Specify the TCAD mesh settings, boundary conditions, and the precise mapping procedure used to import doping and field profiles into Allpix².
Simulated Author's Rebuttal
We thank the referee for the careful reading and constructive comments on the independence of the active-depth validation and the need for methodological details. We address each point below and will revise the manuscript to improve clarity and robustness.
read point-by-point responses
-
Referee: [Abstract] Abstract and the optimization/active-depth comparison section: the single free parameter (N-type blanket doping concentration) is adjusted to match measured efficiency and cluster size. Because depletion depth directly governs collected charge (and therefore efficiency at normal incidence) and the grazing-angle method extracts depth from the angular dependence of the same observables, the subsequent 2 % active-depth agreement is not demonstrably independent of the fit; the tuned value may simply reproduce the effective depth already implicit in the efficiency data.
Authors: We agree that depletion depth is a primary driver of collected charge and normal-incidence efficiency. However, the optimization simultaneously matches both efficiency and cluster size at normal incidence. Cluster size is additionally sensitive to lateral charge sharing and the in-plane electric-field components, which depend on the full doping profile rather than vertical depth alone. The grazing-angle extraction (0–60°) determines depth from the geometric path length and the angular dependence of efficiency and cluster size, providing a largely orthogonal observable. In the revision we will add explicit discussion of this separation and, if feasible with existing data, show the variation of extracted depth with doping to quantify residual correlation. revision: yes
-
Referee: [Abstract] The manuscript provides no information on the fitting procedure, the number of data points or angular settings used for the optimization, the uncertainty on the resulting doping value, or whether the active-depth comparison employs held-out data or a different observable set. Without these details the robustness of the 2 % agreement cannot be assessed.
Authors: The current manuscript indeed omits a detailed description of the fitting procedure, the exact number of angular settings and data points, the uncertainty on the optimized doping concentration, and whether the depth comparison uses held-out angles or observables. We will add a dedicated subsection in the revised version that specifies the optimization method, lists the angular settings and data points employed, reports the uncertainty on the doping value, and clarifies the relationship between the optimization data set and the grazing-angle validation set. revision: yes
Circularity Check
N-blanket doping tuned to efficiency/cluster-size data; 2% active-depth agreement follows by construction from same angular scans
specific steps
-
fitted input called prediction
[Abstract]
"Simulations reveal a strong dependence of sensor performance, specifically the detection efficiency and cluster size, on the doping concentration of the N-type blanket at the sensor surface. The doping concentration is then optimized by comparing simulations with measurement data. The active depth of the depleted region of the MALTA2 sensor is estimated in both simulations and measurements using a grazing angle method, in which the sensor is positioned at various inclinations relative to the beam, covering angles from 0 to 60 degrees. Excellent agreement on active depth is obtained with the 2%"
Doping concentration is fitted to reproduce measured efficiency and cluster size. Active depth is extracted from the identical grazing-angle scans of efficiency and cluster size (angular dependence directly encodes the depth that sets collected charge). Matching the input data therefore forces the extracted depths to agree, rendering the 2% deviation non-independent.
full rationale
The paper optimizes the single free parameter (N-type blanket doping) to match measured detection efficiency and cluster size, then reports 2% agreement on active depth extracted via the grazing-angle method. Because the grazing-angle extraction itself derives depth from the angular dependence of the same efficiency and cluster-size observables, the reported agreement is not an independent check but a direct consequence of the fit. This is the fitted_input_called_prediction pattern with no additional independent content in the validation step.
Axiom & Free-Parameter Ledger
free parameters (1)
- N-type blanket doping concentration
axioms (1)
- domain assumption Generic doping profiles and simple well structures in TCAD accurately represent the real MALTA2 sensor
Reference graph
Works this paper leans on
-
[1]
Piro et al.,A 1𝜇W Radiation-Hard Front-End in a 0.18-𝜇m CMOS Process for the MALTA2 Monolithic Sensor,IEEE Trans
F. Piro et al.,A 1𝜇W Radiation-Hard Front-End in a 0.18-𝜇m CMOS Process for the MALTA2 Monolithic Sensor,IEEE Trans. Nucl. Sci.69(2022) 1299
2022
-
[2]
M. van Rijnbach et al.,Radiation hardness of MALTA2 monolithic CMOS imaging sensors on Czochralski substrates,Eur. Phys. J. C84(2024) 251 [2308.13231]
-
[3]
Pernegger,Monolithic pixel development in towerjazz 180 nm CMOS for the outer pixel layers in the ATLAS experiment,Nucl
H. Pernegger,Monolithic pixel development in towerjazz 180 nm CMOS for the outer pixel layers in the ATLAS experiment,Nucl. Instrum. Meth. A924(2019) 92
2019
-
[4]
M. Munker et al.,Simulations of CMOS pixel sensors with a small collection electrode, improved for a faster charge collection and increased radiation tolerance,Journal of Instrumentation14(2019) C05013
2019
-
[5]
L. Li et al.,Study of MALTA2, a Depleted Monolithic Active Pixel Sensor, with grazing angles at CERN SPS 180 GeV/c hadron beam,JINST20(2025) C06051 [2502.13590]
-
[6]
Solans Sánchez et al.,MALTA monolithic pixel sensors in TowerJazz 180 nm technology,Nucl
C. Solans Sánchez et al.,MALTA monolithic pixel sensors in TowerJazz 180 nm technology,Nucl. Instrum. Meth. A1057(2023) 168787
2023
-
[7]
Allpix$^2$: A Modular Simulation Framework for Silicon Detectors
S. Spannagel, K. Wolters, D. Hynds, N. Alipour Tehrani, M. Benoit, D. Dannheim et al.,Allpix2: A Modular Simulation Framework for Silicon Detectors,Nucl. Instrum. Meth. A901(2018) 164 [1806.05813]
work page internal anchor Pith review Pith/arXiv arXiv 2018
-
[8]
TCAD - Technology Computer Aided Design
Synopsys, “TCAD - Technology Computer Aided Design.” https://www.synopsys.com/manufacturing/tcad.html, 2026
2026
-
[9]
van Rijnbach et al.,Performance of the MALTA telescope,Eur
M. van Rijnbach et al.,Performance of the MALTA telescope,Eur. Phys. J. C83(2023) 581 [2304.01104]
-
[10]
Cardella et al.,MALTA: an asynchronous readout CMOS monolithic pixel detector for the ATLAS High-Luminosity upgrade,JINST14(2019) C06019
R. Cardella et al.,MALTA: an asynchronous readout CMOS monolithic pixel detector for the ATLAS High-Luminosity upgrade,JINST14(2019) C06019
2019
-
[11]
H. Wennlöf et al.,Simulating monolithic active pixel sensors: A technology-independent approach using generic doping profiles,Nucl. Instrum. Meth. A1073(2025) 170227 [2408.00027]
-
[12]
Shockley and W.T
W. Shockley and W.T. Read,Statistics of the recombinations of holes and electrons,Phys. Rev.87 (1952) 835
1952
-
[13]
Hall,Electron-Hole Recombination in Germanium,Phys
R.N. Hall,Electron-Hole Recombination in Germanium,Phys. Rev.87(1952) 387. [14]Particle Data Groupcollaboration,Review of Particle Physics,PTEP2020(2020) 083C01
1952
-
[14]
S. Agostinelli et al.,Geant4—a simulation toolkit,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment506(2003) 250
2003
-
[15]
Allison et al.,Geant4 developments and applications,IEEE Transactions on Nuclear Science53 (2006) 270
J. Allison et al.,Geant4 developments and applications,IEEE Transactions on Nuclear Science53 (2006) 270
2006
-
[16]
J. Allison et al.,Recent developments in geant4,Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment835(2016) 186. – 20 –
2016
-
[17]
L. Fasselt et al.,Charge calibration of MALTA2, a radiation hard depleted monolithic active pixel sensor,Nucl. Instrum. Meth. A1082(2026) 170972 [2501.13562]
-
[18]
Meroli et al.,A grazing angle technique to measure the charge collection efficiency for CMOS active pixel sensors,Nucl
S. Meroli et al.,A grazing angle technique to measure the charge collection efficiency for CMOS active pixel sensors,Nucl. Instrum. Meth. A650(2011) 230
2011
-
[19]
Smithrick and I.T
J.J. Smithrick and I.T. Myers,Average triton energy deposited in silicon per electron-hole pair produced,Phys. Rev. B1(1970) 2945
1970
-
[20]
R.C. Alig, S. Bloom and C.W. Struck,Scattering by ionization and phonon emission in semiconductors,Phys. Rev. B22(1980) 5565
1980
-
[21]
Masetti et al.,Modeling of carrier mobility against carrier concentration in arsenic-, phosphorus-, and boron-doped silicon,IEEE Transactions on Electron DevicesED-30(1983) 764–9
G. Masetti et al.,Modeling of carrier mobility against carrier concentration in arsenic-, phosphorus-, and boron-doped silicon,IEEE Transactions on Electron DevicesED-30(1983) 764–9
1983
-
[22]
Kerr and A
M.J. Kerr and A. Cuevas,General parameterization of auger recombination in crystalline silicon, Journal of Applied Physics91(2002) 2473
2002
-
[23]
Fossum and D
J. Fossum and D. Lee,A physical model for the dependence of carrier lifetime on doping density in nondegenerate silicon,Solid-State Electronics25(1982) 741
1982
-
[24]
A Data-Driven Fast Simulation Approach for MAPS-based Detectors and their Optimization
D.V. Berlea et al.,A Data-Driven Fast Simulation Approach for MAPS-based Detectors and their Optimization,2604.05893. – 21 –
work page internal anchor Pith review Pith/arXiv arXiv
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.