Accurate understanding of the behavior of commercial-off-the-shelf electrical devices is important in many applications. This paper discusses methods for the principled statistical analysis of electrical device data. We present several recent successful efforts and describe two current areas of research that we anticipate will produce widely applicable methods. Because much electrical device data is naturally treated as functional, and because such data introduces some complications in analysis, we focus on methods for functional data analysis.
7th IEEE Electron Devices Technology and Manufacturing Conference: Strengthen the Global Semiconductor Research Collaboration After the Covid-19 Pandemic, EDTM 2023
This paper presents an assessment of electrical device measurements using functional data analysis (FDA) on a test case of Zener diode devices. We employ three techniques from FDA to quantify the variability in device behavior, primarily due to production lot and demonstrate that this has a significant effect in our data set. We also argue for the expanded use of FDA methods in providing principled, quantitative analysis of electrical device data.
7th IEEE Electron Devices Technology and Manufacturing Conference: Strengthen the Global Semiconductor Research Collaboration After the Covid-19 Pandemic, EDTM 2023
Accurate characterization of electrical device behavior is a key component of developing accurate electrical models and assessing reliability. Measurements characterizing an electrical device can be produced from current-voltage (I-V) sweeps. We introduce the pairwise midpoint method (PMM) for estimating the mean of a functional data set and apply it to I-V sweeps from a Zener diode. Comparisons indicate that the PMM is a viable method for describing the mean behavior of a functional data set.
Measurements performed on a population of electronic devices reveal part-to-part variation due to manufacturing process variation. Corner models are a useful tool for the designers to bound the effect of this variation on circuit performance. To accurately simulate the circuit level behavior, compact model parameters for devices within a circuit must be calibrated to experimental data. However, determination of the bounding data for corner model calibration is difficult, primarily because available tolerance bound calculation methods only consider variability along one dimension and, do not adequately consider the variabilities across both the current and voltage axes. This paper presents the demonstration of a novel functional data analysis approach to generate tolerance bounds on these two types of variability separately and these bounds are then transformed to be used in corner model calibration.
Privat, Aymeric; Barnaby, Hugh J.; Adell, Phillipe C.; Tolleson, B.S.; Wang, Y.; Davis, P.; Buchheit, Thomas E.
A multiscale modeling platform that supports the “virtual” qualification of commercial-off-the-shelf parts is presented. The multiscale approach is divided into two modules. The first module generates information related to the bipolar junction transistor gain degradation that is a function of fabrication process, operational, and environmental inputs. The second uses this information as inputs for radiation-enabled circuit simulations. The prototype platform described in this paper estimates the total ionizing dose and dose rate responses of linear bipolar integrated circuits for different families of components. The simulation and experimental results show good correlation and suggest this platform to be a complementary tool within the radiation-hardness assurance flow. Finally, the platform may reduce some of the costly reliance on testing for all systems.
Nominal behavior selection of an electronic device from a measured dataset is often difficult. Device characteristics are rarely monotonic and choosing the single device measurement which best represents the center of a distribution across all regions of operation is neither obvious nor easy to interpret. Often, a device modeler uses a degree of subjectivity when selecting nominal device behavior from a dataset of measurements on a group of devices. This paper proposes applying a functional data approach to estimate the mean and nominal device of an experimental dataset. This approach was applied to a dataset of electrical measurements on a set of commercially available Zener diodes and proved to more accurately represent the average device characteristics than a point-wise calculation of the mean. It also enabled an objective method for selecting a nominal device from a dataset of device measurements taken across the full operating region of the Zener diode.
Nominal behavior selection of an electronic device from a measured dataset is often difficult. Device characteristics are rarely monotonic and choosing the single device measurement which best represents the center of a distribution across all regions of operation is neither obvious nor easy to interpret. Often, a device modeler uses a degree of subjectivity when selecting nominal device behavior from a dataset of measurements on a group of devices. This paper proposes applying a functional data approach to estimate the mean and nominal device of an experimental dataset. This approach was applied to a dataset of electrical measurements on a set of commercially available Zener diodes and proved to more accurately represent the average device characteristics than a point-wise calculation of the mean. It also enabled an objective method for selecting a nominal device from a dataset of device measurements taken across the full operating region of the Zener diode.
This report presents a detailed process for compact model parameter extraction for DC circuit Zener diodes. Following the traditional approach of Zener diode parameter extraction, circuit model representation is defined and then used to capture the different operational regions of a real diode's electrical behavior. The circuit model contains 9 parameters represented by resistors and characteristic diodes as circuit model elements. The process of initial parameter extraction, the identification of parameter values for the circuit model elements, is presented in a way that isolates the dependencies between certain electrical parameters and highlights both the empirical nature of the extraction and portions of the real diode physical behavior which of the parameters are intended to represent. Optimization of the parameters, a necessary part of a robost parameter extraction process, is demonstrated using a 'Xyce-Dakota' workflow, discussed in more detail in the report. Among other realizations during this systematic approach of electrical model parameter extraction, non-physical solutions are possible and can be difficult to avoid because of the interdependencies between the different parameters. The process steps described are fairly general and can be leveraged for other types of semiconductor device model extractions. Also included in the report are recommendations for experiment setups for generating optimum dataset for model extraction and the Parameter Identification and Ranking Table (PIRT) for Zener diodes.
Sealing glasses are ubiquitous in high pressure and temperature engineering applications, such as hermetic feed-through electrical connectors. A common connector technology are glass-to-metal seals where a metal shell compresses a sealing glass to create a hermetic seal. Though finite-element analysis has been used to understand and design glass-to-metal seals for many years, there has been little validation of these models. An indentation technique was employed to measure the residual stress on the surface of a simple glass-to-metal seal. Recently developed rate- dependent material models of both Schott 8061 and 304L VAR stainless steel have been applied to a finite-element model of the simple glass-to-metal seal. Model predictions of residual stress based on the evolution of material models are shown. These model predictions are compared to measured data. Validity of the finite- element predictions is discussed. It will be shown that the finite-element model of the glass-to-metal seal accurately predicts the mean residual stress in the glass near the glass-to-metal interface and is valid for this quantity of interest.
Predicting the residual stress which develops during fabrication of a glass-to-metal compression seal requires material models that can accurately predict the effects of processing on the sealing glass. Validation of the predictions requires measurements on representative test geometries to accurately capture the interaction between the seal materials during a processing cycle required to form the seal, which consists of a temperature excursion through the glass transition temperature of the sealing glass. To this end, a concentric seal test geometry, referred to as a short cylinder seal, consisting of a stainless steel shell enveloping a commercial sealing glass disk has been designed, fabricated, and characterized as a model validation test geometry. To obtain data to test/validate finite element (FE) stress model predictions of this geometry, spatially-resolved residual stress was calculated from the measured lengths of the cracks emanating from radially positioned Vickers indents in the glass disk portion of the seal. The indentation crack length method is described, and the spatially-resolved residual stress determined experimentally are compared to FE stress predictions made using a nonlinear viscoelastic material model adapted to inorganic sealing glasses and an updated rate dependent material model for 304L stainless steel. The measurement method is a first to achieve a degree of success for measuring spatially resolved residual stress in a glass-bearing geometry and a favorable comparison between measurements and simulation was observed.
We present a new, non-destructive electrical technique, Power Spectrum Analysis (PSA). PSA as described here uses off-normal biasing, an unconventional way of powering microelectronics devices. PSA with off-normal biasing can be used to detect subtle differences between microelectronic devices. These differences, in many cases, cannot be detected by conventional electrical testing. In this paper, we highlight PSA applications related to aging and counterfeit detection.
In this work, a crystal plasticity-finite element (CP-FE) model is used to investigate the effects of microstructural variability at a notch tip in tantalum single crystals and polycrystals. It is shown that at the macroscopic scale, the mechanical response of single crystals is sensitive to the crystallographic orientation while the response of polycrystals shows relatively small susceptibility to it. However, at the microscopic scale, the local stress and strain fields in the vicinity of the crack tip are completely determined by the local crystallographic orientation at the crack tip for both single and polycrystalline specimens with similar mechanical field distributions. Variability in the local metrics used (maximum von Mises stress and equivalent plastic strain at 3% deformation) for 100 different realizations of polycrystals fluctuates by up to a factor of 2-7 depending on the local crystallographic texture. Comparison with experimental data shows that the CP model captures variability in stress-strain response of polycrystals that can be attributed to the grain-scale microstructural variability. This work provides a convenient approach to investigate fluctuations in the mechanical behavior of polycrystalline materials induced by grain morphology and crystallographic orientations.
A series of Ti-rich Ni-Ti-Pt ternary alloys with 13 to 18 at. pct Pt were processed by vacuum arc melting and characterized for their transformation behavior to identify shape memory alloys (SMA) that undergo transformation between 448 K and 498 K (175 °C and 225 °C) and achieve recoverable strain exceeding 2 pct. From this broader set of compositions, three alloys containing 15.5 to 16.5 at. pct Pt exhibited transformation temperatures in the vicinity of 473 K (200 °C), thus were targeted for more detailed characterization. Preliminary microstructural evaluation of these three compositions revealed a martensitic microstructure with small amounts of Ti2(Ni,Pt) particles. Room temperature mechanical testing gave a response characteristic of martensitic de-twinning followed by a typical work-hardening behavior to failure. Elevated mechanical testing, performed while the materials were in the austenitic state, revealed yield stresses of approximately 500 MPa and 3.5 pct elongation to failure. Thermal strain recovery characteristics were more carefully investigated with unbiased incremental strain-temperature tests across the 1 to 5 pct strain range, as well as cyclic strain-temperature tests at 3 pct strain. The unbiased shape recovery results indicated a complicated strain recovery path, dependent on prestrain level, but overall acceptable SMA behavior within the targeted temperature and recoverable strain range.
In this report, we present a multi-scale computational model to simulate plastic deformation of tantalum and validating experiments. In atomistic/ dislocation level, dislocation kink- pair theory is used to formulate temperature and strain rate dependent constitutive equations. The kink-pair theory is calibrated to available data from single crystal experiments to produce accurate and convenient constitutive laws. The model is then implemented into a BCC crystal plasticity finite element method (CP-FEM) model to predict temperature and strain rate dependent yield stresses of single and polycrystalline tantalum and compared with existing experimental data from the literature. Furthermore, classical continuum constitutive models describing temperature and strain rate dependent flow behaviors are fit to the yield stresses obtained from the CP-FEM polycrystal predictions. The model is then used to conduct hydro- dynamic simulations of Taylor cylinder impact test and compared with experiments. In order to validate the proposed tantalum CP-FEM model with experiments, we introduce a method for quantitative comparison of CP-FEM models with various experimental techniques. To mitigate the effects of unknown subsurface microstructure, tantalum tensile specimens with a pseudo-two-dimensional grain structure and grain sizes on the order of millimeters are used. A technique combining an electron back scatter diffraction (EBSD) and high resolution digital image correlation (HR-DIC) is used to measure the texture and sub-grain strain fields upon uniaxial tensile loading at various applied strains. Deformed specimens are also analyzed with optical profilometry measurements to obtain out-of- plane strain fields. These high resolution measurements are directly compared with large-scale CP-FEM predictions. This computational method directly links fundamental dislocation physics to plastic deformations in the grain-scale and to the engineering-scale applications. Furthermore, direct and quantitative comparisons between experimental measurements and simulation show that the proposed model accurately captures plasticity in deformation of polycrystalline tantalum.
This report provides a detailed characterization Tri-lab Tantalum (Ta) plate jointly purchased from HCStark Inc. by Sandia, Los Alamos and Lawrence Livermore National Laboratories. Data in this report was compiled from series of material and properties characterization experiments carried out at Sandia (SNL) and Los Alamos (LANL) Laboratories through a leveraged effort funded by the C2 campaign. Results include microstructure characterization detailing the crystallographic texture of the material and an increase in grain size near the end of the rolled plate. Mechanical properties evaluations include, compression cylinder, sub-scale tension specimen, micohardness and instrumented indentation testing. The plate was found to have vastly superior uniformity when compare with previously characterized wrought Ta material. Small but measurable variations in microstructure and properties were noted at the end, and at the top and bottom edges of the plate.