The impact of X-ray and cobalt 60 irradiations on the DC characteristics of two advanced BiCMOS technologies, Si/SiGe:C HBTs, is compared and analyzed. The effects of geometry, post-irradiation time, and annealing are investigated.

The DC and low-frequency noise (LFN) study of pre-rad and post-rad X-ray irradiation of SiGe:C HBTs based on a 55 nm BiCMOS technology is presented. HBTs are developed with three different levels of doping in the collector region for applications requiring high speed (HS), medium voltage (MV), and high voltage (HV) operations. This work presents excess DC base current, the 1/f noise model using the SPICE modeling and provides the descriptive behavior of G-R noise components for three different HBTs (HS, MV, and HV) before and after irradiation (as a function of the total ionizing dose (TID)). By comparing the DC and low- frequency noise (LFN) results with respect to TID, the discussion deals with the location of the noise sources and related traps. Particularly, it underlines the role of the SiO2/Si E-B spacer interface.

Gamma irradiation effects are investigated on Si/SiGe:C HBTs developed with the latest BiCMOS technologies. Unbiased HBTs are irradiated with a Co60source until in a Total Ionizing Dose of 330 krad. Irradiation effects are evaluated by measuring the excess base current from DC characteristics (Gummel plot) and the base current spectral density from Low Frequency Noise measurements, mainly by analyzing the 1/f noise level. Degradation comparison on two advanced BiCMOS technologies is done. The recent generation of the technology presents the highest robustness. Moreover, a comparison of the degradation induced by the Gamma source with an earlier study using an X-ray source is held.

This paper proposes a strategy to handle missing data for the classification of electroencephalograms using covariance matrices. It relies on the observed-data likelihood within an expectation-maximization algorithm. This approach is compared to two existing state-of-the-art methods: (i) covariance matrices computed with imputed data; (ii) Riemannian averages of partially observed covariance matrix. All approaches are combined with the minimum distance to Riemannian mean classifier and applied to a classification task of two widely known paradigms of brain-computer interfaces. In addition to be applicable for a wider range of missing data scenarios, the proposed strategy generally performs better than other methods on the considered real EEG data.

Linear and Quadratic Discriminant Analysis are well-known classical methods but can heavily suffer from non-Gaussian distributions and/or contaminated datasets, mainly because of the underlying Gaussian assumption that is not robust. To fill this gap, this paper presents a new robust discriminant analysis where each data point is drawn by its own arbitrary Elliptically Symmetrical (ES) distribution and its own arbitrary scale parameter. Such a model allows for possibly very heterogeneous, independent but non-identically distributed samples. After deriving a new decision rule, it is shown that maximum-likelihood parameter estimation and classification are very simple, fast and robust compared to state-of-the-art methods.

Change detection for radar image time series is an important task that can help to monitor deforestation and global warming consequences. We present a method to detect changes in time for Polarimetric SAR images based on a clustering approach. The first step provides a segmentation for each image and then one detects changes by monitoring the resulting labels. This work is based on a robust clustering algorithm that increases the flexibility in the segmentation stage. We report the outcome of our method when it is tested on simulated and real Polarimetric SAR data. The change detection results are promising when comparing our performance to that of other standard methods.

We consider the problem of detecting a known M-dimensional target signature vector from an observation corrupted by an additive noise with unknown covariance matrix. In that case, standard statistical methods of detection usually assume that N - “target free” observations are available to perform estimation of the noise covariance matrix. However, in several applications, the target signal may contaminate the training data, resulting in a deviation of the expected performance of the detectors. In this paper, we consider the performance analysis of two low-rank detectors under the assumption that Nc elements of the training data are contaminated by the target signal. More precisely, we derive the asymptotic false alarm and detection probabilities in the high dimensional regime in which both the dimension M, the number of training data N and contaminated data Nc converge to infinity at the same rate. Numerical simulations illustrate the fact that, despite the asymptotic nature of the analysis, the results obtained are accurate for reasonable values of M, N and Nc.

This work presents Low Frequency Noise (LFNoise) characterization and modeling performed on DPSASEG SiGe HBT integrated in a 55-nm CMOS node. The aim of this study is to evaluate the advantage brought by the implementation of a Dynamic Surface Annealing (DSA) in addition to the well-known Spike Annealing process. The HBTs are supplied by STMicroelectronics Crolles and present transit (fT) and maximum oscillation (fMAX) frequencies in the 320-370 GHz range. Spectra can be affected by the presence of generation-recombination (GR) components. The 1/f noise amplitude is modeled following the SPICE compact model, and the 1/f parameters KF and AF are calculated. The extracted figure of merit KB = KFAe has a very good value of 6.8 10-10 μm² for transistors processed using the DSA technique.

This study presents an investigation of the dose response of Si/SiGe HBTs developed with the last generation of BiCMOS technologies. DC electrical characterization and low-frequency noise measurements are carried out to evaluate the post-radiation degradations.

The purpose of this paper is to derive new asymptotic properties of the robust adaptive normalized matched filter (ANMF). More precisely, the ANMF built with Tyler estimator (TyE-ANMF) is analyzed under the framework of complex elliptically symmetric (CES) distributions. We show that the distribution of TyE-ANMF can be accurately approximated by the well-known distribution of the ANMF built with the sample covariance matrix (SCM-ANMF) under the Gaussian assumption. To that end, the asymptotic properties of the difference between both ANMF detectors are derived. By comparison with the state of the art, the asymptotic properties of the TyE-ANMF are shown to be better approximated by the SCM-ANMF rather than using the NMF (test built with the true CM). Some Monte-Carlo simulations support that claim and demonstrate the interest of this theoretical result.

The main objective of the development of new BiCMOS technologies is to enhance the high frequency performances of the devices. Nevertheless, Low Frequency Noise (LFN) analysis, in particular the 1/f noise, is a very sensitive tool to evaluate a technology. In this work we present a complete characterization and modeling of the 1/f noise in Si/SiGe:C HBTs issued from three BiCMOS technologies (130 nm, shrinked 130 nm and 55 nm). The measured base current spectral density SIB, in the HBTs issued from a 130 nm BiCMOS technology, presented a typical behavior of 1/f noise. The extracted SPICE figure of merit Kb has an excellent value of 1.4 10 -10 μm 2 . For HBTs issued from the two other technologies (shrinked 130 nm and recent under development 55 nm) the low frequency noise spectra are often disrupted by the presence of more or less pronounced Lorentzian shape leading to noise dispersion. In HBTs where a typical behavior was observed, the 1/f noise level was found to be proportional to the square of IB, and inversely proportional to the emitter area Ae. The extracted Kb value was equal to 6 10 -10 for the shrinked 130 nm technology and approximately 10 -9 μ 2 for the 55 nm technology. In order to take into account the HF performances, we have also studied the ratio fc/ft. This figure of merit combines LF Noise and HF characteristics.

In this work we present the first Low Frequency Noise results obtained on the new 55 nm BiCMOS technology developed by STMicroelectronics. With this improved technology at higher integration level, SiGe:C Heterojunction Bipolar Transistors will address High Speed & High Data Rate communication systems and smart mobility integrated circuits involved in the future fully automated transportation systems. The LF Noise measurements are performed in the 1Hz-100 kHz frequency range as a function of the base bias current and of the emitter area A E . DC and Low Frequency Noise characteristics are presented on three collector HBTs architectures. In these structures, the collector doping varies in order to propose different output powers. The best 1/f noise figure of merit K B is found to be close to 4 10-10 μm 2 . This value represents a very good one but do not reach the best results obtained in the mature previous 0.13μm BiCMOS technology.

An original estimator of the orthogonal projector onto the signal subspace is proposed. This estimator is derived as the maximum likelihood estimator for a model of sources plus orthogonal outliers, both with varying power (modeled by Compound Gaussians process), embedded in a white Gaussian noise. Validity and interest - in terms of performance and robustness - of this estimator is illustrated through simulation results on a low rank STAP filtering application.

We study the room-temperature performance of micro-Hall magnetic sensors based on pseudomorphic InGaAs quantum wells. Active areas of our sensors range from 1 to 80 μm. We focus on the smallest detectable magnetic fields in small sensors and perform a systematic study of noise at room temperature in the frequency range between 1 Hz and 100 kHz. Our data are interpreted by the mobility fluctuation model. The Hooge parameter is determined for the applied technology. We show that, independently of the experimental frequency, the ratio of sensitivity to noise is proportional to characteristic length of the sensor. The resolution of 1 mG/Hz is achievable in a 3μm sensor at room temperature.

In this paper, we present the manufacture and characterization of a screen-printed, acoustic PZT component with enhancement of its properties by optimizing the poling and densification process. We discuss results obtained on ferroelectric properties using remnant hysteresis cycle measurements, piezoelectric properties by evaluation of the d <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">33</sub> coefficients and electrical properties by measuring the capacitance. In addition, the first acoustic characterization by spectrum analysis is presented.

This work presents an improved measurement setup to directly measure the collector low frequency current spectral density S IC of SiGe:C Heterojunction Bipolar Transistors (HBTs) when the base is AC short-circuited. A comprehensive analysis of the obtained results is also provided as well as a SPICE based electrical model. The additional white noise observed at high collector biases is due to the contribution of the emitter/base internal resistance. The 1/f noise exhibits a linear evolution with a 1.7 slope versus I C and is proportional to 1/√A e . No evolution of S IC with emitter periphery P e was observed. From a comparative study of the different 1/f noise term of S IC we found that S IC is mainly associated with the spontaneous fluctuations of the intrinsic collector current.

This work presents an improved measurement setup to directly measure the collector low frequency current spectral density S IC of SiGe:C Heterojunction Bipolar Transistors (HBTs) when the base is AC short-circuited. A comprehensive analysis of the obtained results is also provided as well as a SPICE based electrical model. The additional white noise observed at high collector biases is due to the contribution of the emitter/base internal resistance. The 1/f noise exhibits a linear evolution with a 1.7 slope versus I C and is proportional to 1/√A e . No evolution of S IC with emitter periphery P e was observed. From a comparative study of the different 1/f noise term of S IC we found that S IC is mainly associated with the spontaneous fluctuations of the intrinsic collector current.