Olivier Saint-Pé

Displacement Damage Effects Due to Neutron and Proton Irradiations on CMOS Image Sensors Manufactured in Deep Submicron Technology

Displacement damage effects due to proton and neutron irradiations of CMOS image sensors dedicated to imaging are presented through the analysis of the dark current behavior in pixel arrays and isolated photodiodes. The mean dark current increase and the dark current nonuniformity are investigated. Dark current histogram observations are compared to damage energy distributions based on GEANT 4 calculations. We also discuss, through annealing analysis, which defects could be responsible for the dark current in CMOS image sensors.

Multilevel RTS in Proton Irradiated CMOS Image Sensors Manufactured in a Deep Submicron Technology

A new automated method able to detect multilevel random telegraph signals (RTS) in pixel arrays and to extract their main characteristics is presented. The proposed method is applied to several proton irradiated pixel arrays manufactured using a 0.18 mum CMOS process dedicated to imaging. Despite the large proton energy range and the large fluence range used, similar exponential RTS amplitude distributions are observed. A mean maximum amplitude independent of displacement damage dose is extracted from these distributions and the number of RTS defects appears to scale well with total nonionizing energy loss. These conclusions allow the prediction of RTS amplitude distributions. The effect of electric field on RTS amplitude is also studied and no significant relation between applied bias and RTS amplitude is observed.

Similarities Between Proton and Neutron Induced Dark Current Distribution in CMOS Image Sensors

Several CMOS image sensors were exposed to neutron or proton beams (displacement damage dose range from 4 TeV/g to 1825 TeV/g) and their radiation-induced dark current distributions are compared. It appears that for a given displacement damage dose, the hot pixel tail distributions are very similar, if normalized properly. This behavior is observed on all the tested CIS designs (4 designs, 2 technologies) and all the tested particles (protons from 50 MeV to 500 MeV and neutrons from 14 MeV to 22 MeV). Thanks to this result, all the dark current distribution presented in this paper can be fitted by a simple model with a unique set of two factors (not varying from one experimental condition to another). The proposed normalization method of the dark current histogram can be used to compare any dark current distribution to the distributions observed in this work. This paper suggests that this model could be applied to other devices and/or irradiation conditions.

Total Dose Evaluation of Deep Submicron CMOS Imaging Technology Through Elementary Device and Pixel Array Behavior Analysis

Ionizing radiation effects on CMOS image sensors (CIS) manufactured using a 0.18 mum imaging technology are presented through the behavior analysis of elementary structures, such as field oxide FET, gated diodes, photodiodes and MOSFETs. Oxide characterizations appear necessary to understand ionizing dose effects on devices and then on image sensors. The main degradations observed are photodiode dark current increases (caused by a generation current enhancement), minimum size NMOSFET off-state current rises and minimum size PMOSFET radiation induced narrow channel effects. All these effects are attributed to the shallow trench isolation degradation which appears much more sensitive to ionizing radiation than inter layer dielectrics. Unusual post annealing effects are reported in these thick oxides. Finally, the consequences on sensor design are discussed thanks to an irradiated pixel array and a comparison with previous work is discussed.

Theoretical Models of Modulation Transfer Function, Quantum Efficiency, and Crosstalk for CCD and CMOS Image Sensors

This paper proposes analytical models of modulation transfer function (MTF), quantum efficiency (QE), and crosstalk for charge-coupled device (CCD) and CMOS image sensors. A unified MTF model for a CCD sensor built on an epitaxial layer deposited on a highly doped substrate was developed by Stevens. The Stevens model uses sinusoidal illumination to calculate the sensor MTF degradation due to charge diffusion and sampling aperture as a function of spatial frequency. The drawback of this approach is the difficulty to evaluate analytically the electrical crosstalk distribution, which can be a good tool for predicting the detector performances, particularly for smaller pixels. In this paper, we use point-source illumination to evaluate the pixel response function (PRF). This approach is applied to the case of CMOS sensors and buried channel CCD sensors. The MTF model includes the impact of pixel size and charge diffusion. The QE model and crosstalk distribution are directly derived from the PRF expression. The models can take into account an electric field induced by a doping gradient.

Total Ionizing Dose Versus Displacement Damage Dose Induced Dark Current Random Telegraph Signals in CMOS Image Sensors

Dark current Random Telegraph Signals due to total ionizing dose (TID) and displacement damage dose (DDD) are investigated in CMOS image sensors. Discrepancies between both RTS are emphasised to better understand the microscopic origins of the phenomena.

Ionizing radiation effects on CMOS imagers manufactured in deep submicron process

We present here a study on both CMOS sensors and elementary structures (photodiodes and in-pixel MOSFETs) manufactured in a deep submicron process dedicated to imaging. We designed a test chip made of one 128×128-3T-pixel array with 10 μm pitch and more than 120 isolated test structures including photodiodes and MOSFETs with various implants and different sizes. All these devices were exposed to ionizing radiation up to 100 krad and their responses were correlated to identify the CMOS sensor weaknesses. Characterizations in darkness and under illumination demonstrated that dark current increase is the major sensor degradation. Shallow trench isolation was identified to be responsible for this degradation as it increases the number of generation centers in photodiode depletion regions. Consequences on hardness assurance and hardening-by-design are discussed.

Theoretical evaluation of MTF and charge collection efficiency in CCD and CMOS image sensors

Classical models used to calculate the Modulation Transfer function (MTF) of a solid-state image sensor generally use a sinusoidal type of illumination. The approach, described in this paper, consists in considering a point-source illumination to built a theoretical three-dimensional model of the diffusion and the collection of photo-carriers created within the image sensor array. Fourier transform formalism is used for this type of illumination. Solutions allow to evaluate the spatial repartition of the charge density collected in the space charge region, i.e. to get the Pixel Response Function (PRF) formulation. PRF enables to calculate analytically both MTF and crosstalk at every needed wavelengths. The model can take into account a uniformly doped substrate and an epitaxial layer grown on a highly doped substrate. The built-in electric field induced by the EPI/Substrate doping gradient is also taken into account. For these configurations, MTF, charge collection efficiency and crosstalk proportion are calculated. The study is established in the case of photodiode pixel but it can be easily extended to pinned photodiode pixels and photogate pixels.

Temperature Dependence and Dynamic Behavior of Full Well Capacity in Pinned Photodiode CMOS Image Sensors

This paper presents an analytical model of the full well capacity (FWC) in pinned photodiode (PPD) CMOS image sensors. By introducing the temperature dependence of the PPD pinning voltage, the existing model is extended (with respect to previous works) to consider the effect of temperature on the FWC. It is shown, with the support of experimental data, that whereas in dark conditions the FWC increases with temperature, a decrease is observed if FWC measurements are performed under illumination. This paper also shows that after a light pulse, the charge stored in the PPD drops as the PPD tends toward equilibrium. On the basis of these observations, an analytical model of the dynamic behavior of the FWC in noncontinuous illumination conditions is proposed. The model is able to reproduce experimental data over six orders of magnitude of time. Both the static and dynamic models can be useful tools to correctly interpret FWC changes following design variations and to accurately define the operating conditions during device characterizations.

CMOS detectors for space applications: from R&D to operational program with large volume foundry

Nowadays, CMOS image sensors are widely considered for space applications. The use of CIS (CMOS Image sensor) processes has significantly enhanced their performances such as dark current, quantum efficiency and conversion gain. However, in order to fulfil specific space mission requirements, dedicated research and development work has to be performed to address specific detector performance issues. This is especially the case for dynamic range improvement through output voltage swing optimisation, control of conversion gain and noise reduction. These issues have been addressed in a 0.35μm CIS process, based on a large volume CMOS foundry, by several joint ISAE- EADS Astrium R&D programs. These results have been applied to the development of the visible and near-infrared multi-linear imager for the SENTINEL 2 mission (LEO Earth observation mission for the Global Measurement Environment and Security program). For this high performance multi-linear device, output voltage swing improvement is achieved by process optimisation done in collaboration with foundry. Conversion gain control is also achieved for each spectral band by managing photodiode capacitance. A low noise level at sensor output is reached by the use of an architecture allowing Correlated Double Sampling readout in order to eliminate reset noise (KTC noise). KTC noise elimination reveals noisy pixels due to RTS noise. Optimisation of transistorss dimensions, taking into account conversion gain constraints, is done to minimise these noisy pixels. Additional features have been also designed: 1) Due to different integration times between spectral bands required by mission, a specific readout mode was developed in order to avoid electrical perturbations during the integration time and readout. This readout mode leads to specific power supply architecture. 2)Post processing steps can be achieved by alignment marks design allowing a very good accuracy. These alignment marks can be used for a black coating deposition between spectral bands (pixel line) in order to minimise straight light effects. In conclusion a review of design improvements and performances of the final component is performed.