The simulation-based analyses of production yield losses are still in the early stage. With the possibility to use input parameters with a stochastic distribution, taking in account the manufacturing tolerances, and with the ability to calculate failure probabilities these analyses become nowadays possible.The approach shown here for a RF product can give useful insights for the reduction of the production yield loss, ideally already in the early design stage of the manufacturing process itself.

This paper deals with the estimation of a failure probability of an industrial product. To be more specific, it is defined as the probability that the output of a physical model, with random input variables, exceeds a threshold. The model corresponds with an expensive to evaluate black-box function, so that classical Monte Carlo simulation methods cannot be applied. Bayesian principles of the Kriging method are then used to design an estimator of the failure probability. From a numerical point of view, the practical use of this estimator is restricted. An alternative estimator is proposed, which is equivalent in term of bias. The main result of this paper concerns the existence of a convex order inequality between these two estimators. This inequality allows to compare their efficiency and to quantify the uncertainty on the results that these estimators provide. A sequential procedure for the construction of a design of computer experiments, based on the principle of the Stepwise Uncertainty Reduction strategies, also results of the convex order inequality. The interest of this approach is highlighted through the study of a real case from the company STMicroelectronics.

In competitive industries, a reliable yield forecasting is a prime factor to accurately determine the production costs and therefore ensure profitability. Indeed, quantifying the risks long before the effective manufacturing process enables fact-based decision-making. From the development stage, improvement efforts can be early identified and prioritized. In order to measure the impact of industrial process fluctuations on the product performances, the construction of a failure risk probability estimator is presented in this article. The complex relationship between the process technology and the product design (non linearities, multi-modal features...) is handled via random process regression. A random field encodes, for each product configuration, the available information regarding the risk of non-compliance. After a brief presentation of the Gaussian model approach, we describe a Bayesian reasoning avoiding a priori choices of location and scale parameters. The Gaussian mixture prior, conditioned by measured (or calculated) data, yields a posterior characterized by a multivariate Student distribution. The probabilistic nature of the model is then operated to derive a failure risk probability, defined as a random variable. To do this, our approach is to consider as random all unknown, inaccessible or fluctuating data. In order to propagate uncertainties, a fuzzy set approach provides an appropriate framework for the implementation of a Bayesian model mimicking expert elicitation. The underlying leitmotiv is to insert minimal a priori information in the failure risk model. The relevancy of this concept is illustrated with theoretical examples.

We propose a new tool of decision support in front of a globally unknown phenomenon which is modeled by a random field representing simultaneously our knowledge and our lack of information. This tool is the distribution of a random variable called failure risk probability. Before giving the precise definition of this object, we describe an industrial context in which the decision problem occurs and we discuss Bayesian random field model constructions.

The increasing expansion of telecommunication applications leads to the integration of complete system-on-chip associating analog and digital processing units. Besides, the passive elements occupy an increasing silicon footprint, compromising circuit scalability and cost. Moreover, passive components' performances are limited by the proximity of lossy Si substrate and surrounding metallization. Then, obviously, the characteristics of the substrate become crucial for monolithic radio frequency (RF) systems to reach high performances. So, looking for integrated circuit compatible processes, porous silicon (PS) seems to be a promising candidate as it can provide localized isolating regions from various silicon substrates. In this review, we first present all the possible porous silicon substrates, which can be used for RF devices. In particular, we put the emphasis on the etching conditions, leading to high thickness localized PS layers. The intrinsic electrical properties of porous silicon such as AC electrical conductivity or dielectric constant are also detailed, and the results extracted from the literature are commented. Then, we describe the performances of widespread RF devices, that is, inductors or coplanar waveguides. Finally, we describe methodologies used for predicting RF electrical responses of PS isolated devices, based on electromagnetic simulations. (C) 2014 AIP Publishing LLC.

In this paper, we present processes to etch high thickness porous silicon layers for RF device applications. Indeed, on-chip inductors realized on bulk silicon suffer from mediocre Q-factor values mainly because of electrical losses into the substrate and capacitive coupling with the silicon appearing beyond 1 GHz. We present a detailed study of etching parameters such as the current density or the HF concentation in HF: H2O: acetic acid based electrolytes. In addition, we propose a prospective study of integrated copper inductor performances on porous silicon substrates in the range of 100 MHz to 10 GHz.

Over the last few years, thin films of PbZrxTi1 − xO3 (PZT) have been the focus of extensive researches for high-k capacitor applications. However, we believe that the reliability properties and the degradation mechanisms of PZT capacitors need to be better understood. A good way to learn about failure mechanisms is to investigate the characteristics of leakage current conduction. In this paper we propose a model for current density evolution of IrO2/PZT/Pt structures as a function of time, voltage and temperature. The voltage and temperature evolution of leakage current is interpreted as an interface controlled thermoionic injection of carriers over a potential barrier at the cathode/PZT contact. The time evolution of the leakage current is mainly characterized by the resistance degradation phenomenon which results in a large increase in current density. A quantitative analytical model based on the redistribution of oxygen vacancies near the cathode interface has already been developed to account for this effect [1]. We propose a more complete model that also includes the role of oxygen vacancies on dielectric relaxation and trapping phenomena. The contributions of Pt and IrO2 electrodes on leakage current evolution are also discussed.

Over the last few years, thin films of PbZx(x)Ti(1-x)O(3) (PZT) have been the focus of extensive researches for high-k capacitor applications. However, some electrical properties such as leakage current conduction and degradation mechanisms need to be better understood.From Constant Voltage Stress experiments, we identified two distinct failure mechanisms depending on the applied voltage levels. The existence of these two failure mechanisms makes it impossible to extrapolate lifetime results from high voltage to low voltage. Since the typical operating voltage for decoupling capacitors is around 3 V the reliability study has to be focused on the low voltage breakdown. A good opportunity to learn about the low voltage failure mechanisms is to investigate the characteristics of leakage current. The time evolution of leakage current is mainly controlled by the resistance degradation phenomenon. A quantitative analytical model has already been developed to account for this effect. We propose a more complete model that also includes dielectric relaxation and trapping effects. The resulting model is combined to a charge-influenced thermoionic emission model that fits fairly well the voltage and temperature dependence of leakage current. The static and dynamic parts of our model are found to be consistent, especially in terms of barrier lowering effect induced by the resistance degradation process. We believe that the low voltage breakdown is related to a trapping-induced creation of defects in the film. (c) 2005 Elsevier B.V. All rights reserved.

Today, an exhaustive library of models describing the electrical behavior of integrated passive components in the radio-frequency range is essential for the simulation and optimization of complex circuits. A rigorous and efficient procedure is presented, based on design of experiments, that is set up in order to extract, for planar spiral inductors, an analytical model parameterized as a function of the inductor geometry. This model can be deduced from either simulations or measurement results and we will focus in this paper on electromagnetic simulations based on a modified formulation of the PEEC method.

Today, an exhaustive library of models describing the electrical behavior of integrated passive components in the radio-frequency range is essential for the simulation and optimization of complex circuits. In this work, a physical model for planar spiral inductors is introduced, based on a circular PEEC method. A mutual inductance formula is derived to include a coplanar ground plane and extend the model to inductors with an incomplete outer turn. The numerical solution is also discussed, including an efficient inversion procedure.

Today, an exhaustive library of models describing the electrical behavior of integrated passive components in the radio-frequency range is essential for the simulation and optimization of complex circuits. In this work, a preliminary study has been done on Tantalum Nitride (TaN) resistors integrated on silicon, and this leads to a single p-type lumped-element circuit. An efficient extraction technique will be presented to provide a computer-driven optimizer with relevant initial model parameter values (the "guess-timate"). The results show the unicity in most cases of the lumped element determination, which leads to a precise simulation of self-resonant frequencies.

Since the first electrochemical reduction step of NAD÷ and the electrochemical oxidation step of NADH lie respectively at —0.69 and 0.5–0.9 V (N.H.E.). no direct measure of the formal potential of NAD÷-NADH system may be attained potentiometrically. We examine the significance of the previous potentiometric studies realized in the presence of a mediator and an enzyme. A reliable zer-current potential of NAD÷ and NADH solution is only obtained when small amounts of benzyl-viologen (BV2÷) and xanthine oxidase (XO) are added. The various electrochemical reduction steps of 10 to 1000 μM BV2÷ aqueous solutions are studied at pH 9.8 on dropping mercury and rotating platinum disk electrodes. The formal potential of the BV2÷-BV− system, equal to —0.360 V (N.H.E.), may be measured on a platinum electrode, while on a mercury electrode a strong adsorption interferes and gives a polarographic prewave. When XO is present, the addition of NADH decreases the BV2÷ reduction waves: an anodic wave corresponding to BV2÷ oxidation appears at the same potential as the first BV2÷ reduction wave. The zero-current potential measured in NAD÷, NADH, BV2÷, BV2÷ and XO solutions is actually fixed by the BV2÷-BV÷ system, which equilibrates through chemical oxido-reduction reactions with the NAD÷-NADH system.

Aqueous solutions of NADH and of model compounds such as I-4,dihydro-N1-propyl- and 1–4,dihydro-N1-benzyl-nicotinamide are oxidized on rotating Pt disk electrode at +0.66 ± 0.02 V (N.H.E.). This two-electron wave is pH independent in the range of 7–13. Both macroelectrolysis on Pt at +0.85 V (N.H.E.) and chemical oxidation by [Fe (CN6]3− lead to NAD+ or model compounds.

The electrochemical oxidation of NADH and its model compound, N-benzyl-1,4-dihydronicotinamide (DHN), has been studied at gold electrode modified with self-assembled monolayer of terminally substituted thiols/disulfide, i.e., cystamine (CYST), mercaptopropionic acid (MPA) and mercaptoethanol (ME). A substantial decrease in the overpotential (∼250 mV) when compared to the bare electrode has been observed for the oxidation of NADH at the monolayer-modified electrodes, containing no so-called redox mediator. The bare electrode shows an ill-defined voltammetric peak for the oxidation of DHN, whereas the monolayer-modified electrodes showed a well-defined voltammetric peak. The monolayer assembly on the gold electrode prevents the fouling of electrode surface by the oxidation products, which favors the oxidation at the less positive potential. The square-wave voltammograms showed a sharp voltammetric signal for the oxidation of NADH at all the monolayer-modified electrodes. All the monolayer-modified electrodes showed a linear current response to change in the NADH concentration in its range of 25–300 μM and their sensitivities were found to be 0.005±0.0003, 0.0063±0.0002 and 0.0052±0.0003 μA/μM for CYST–Au, ME–Au and MPA–Au electrodes, respectively. The hydrodynamic voltammograms obtained at the rotating CYST–Au electrode for the oxidation of NADH and DHN were used to estimate the diffusion coefficient of DHN, and the number of electrons involved in the oxidation process of NADH.

The electrochemical reduction of nicotinamide—adenine-dinucleotide (NAD+) and model compounds has been extensively studied1-3 ; on the other hand very few papers have been published on the electrochemical oxidation of nicotinamide—adenine-dinucleotide reduced (NADH) in aqueous solution. Haas4 reported the occurrence of an oxidation wave of NADH at pH 8.6 at rotating platinum and glassy carbon disk electrodes (with respective half-wave potentials 0.90 ± 0.02 and 0.67 ± 0.03 V vs. NHE); macroscale electrolyses lead to NAD+ by a two-electron oxidation4'5. In this paper we shall present the features of the electrochemical oxidation of model compounds of NADH, namely Ni-benzyl-and NL-propyl-l,4-dihydronicotinamide (NBzNH and NPrNH) in aqueous solution, buffered within a large pH range; a possible mechanism is deduced from these data. Extensive studies have shown the occurrence of a primary acid modification of aqueous NADH solutions or its models below pH 7 (ref. 6); the common explanation put forward is a saturation of the 5-6 double bond in the pyridinic ring, the u.v. absorption band being shifted from 355 to 290 nm. Therefore the results and discussions presented here only concern the pH range 7 to 13. Experimental All products used were prepared in this laboratory, according to published methods, namely NBzNTP (ref. 7), NBzNH and NPrN+Br~ (ref. 8), NPrNH (ref. 9). The purity and concentration of the freshly prepared aqueous solutions were checked by u.v. spectrophotometry with a Cary 14 apparatus. Drop-time controlled portography and voltammetry were carried out with Tacussel equipment (PRT 500 LC Servovit 9 B, S6 RZ millivoltmeter) and a Sefram Luxy trace bigalvanometric recorder. A Tacussel IG 4-100 electronic integrator was used for coulometry. At the beginning of all macroelectrolyses a Fontaine A 60 power-supply was connected between the counter-electrode and the generator. All aqueous solutions were thermostated at 25.0 ± 0.1°C during the experiments.