Extracted Parameter Values Research Articles

An optimally self-biased threshold-voltage extractor

A novel threshold-voltage extractor architecture is presented. A differential-difference transconductor (DDT) loop automatically biases the device-under-test in continuous time around the inflection point of the /spl radic/I/sub D/ versus V/sub GS/ characteristics. Another DDT operates as an arithmetic processor to precisely implement multiplication by two and subtraction as needed for extrapolation. The extraction procedure thus complies entirely with all steps of the manual saturation method. With appropriate modifications, the architecture can also serve as an extractor implementing the linear method. The proposed architecture is applicable to both PMOS and NMOS on the same chip, and generates the value of V/sub T/ as a voltage with respect to the appropriate rail. It has been fabricated on silicon, and its accuracy has been experimentally verified by comparing automatically and manually extracted parameter values.

Analysis of cmos photodiodes- part I: quantum efficiency

An improved one-dimensional (1-D) analysis of the CMOS photodiode has been derived in which the effect of the substrate, which forms a high-low junction with the epitaxial layer, has been included. The analytical solution was verified with numerical simulations based on parameters extracted from a standard 0.35 /spl mu/m CMOS process. Two empirical parameters are suggested to offset the unavoidable inaccuracies in the extracted parameter values. The derived semiempirical expression exhibits a good agreement with the measured spectral response. In Part II of this paper, a three-dimensional (3-D) analysis of lateral photoresponse in CMOS photodiode arrays is presented along with an empirical modeling method utilizing test linear photodiode arrays.

Extraction of DFB laser rate equation parameters for system simulation purposes

Values for the parameters in a rate equation description of a distributed feedback (DFB) laser must be chosen appropriately in order to obtain agreement between calculated and measured results for system performance. A technique is described for readily extracting values of the rate equation parameters using measurements of the threshold current and the optical power, resonance frequency and damping factor for a bias current well above the threshold current. When used to estimate system performance, it is shown that different sets of physically reasonable parameter values which reproduce the measurements yield the same dependence of the receiver sensitivity on fiber length. Calculated and measured results for the receiver sensitivity and time resolved chirp exhibit good agreement when the extracted parameter values are used.

Parameter extraction for statistical IC modeling based on recursive inverse approximation

An accurate and efficient parameter extraction methodology, utilizing a new technique called recursive inverse approximation (RIA), is proposed for statistical modeling of integrated circuits. The main features of RIA are (1) linear approximation is used to obtain initial model parameter estimates, (2) reverse verification performs accuracy checking, and (3) error correction functions are constructed in the extracted parameter space to recursively refine the previously extracted parameter values. As a result, an approximate inverse mapping from the measured performance space to the model parameter space is established for statistical parameter extraction. Examples of parameter extraction for MOS transistors and IC multiplier block demonstrate high efficiency and accuracy of the new method.

Automatic learning of rules: A practical example of using artificial intelligence to improve computer-based detection of myocardial infarction and left ventricular hypertrophy in the 12-lead ECG

The authors developed a computer program that detects myocardial infarction (MI) and left ventricular hypertrophy (LVH) in two steps: (1) by extracting parameter values from a 10-second, 12-lead electrocardiogram, and (2) by classifying the extracted parameter values with rule sets. Every disease has its dedicated set of rules. Hence, there are separate rule sets for anterior MI, inferior MI, and LVH. If at least one rule is satisfied, the disease is said to be detected. The computer program automatically develops these rule sets. A database (learning set) of healthy subjects and patients with MI, LVH, and mixed MI + LVH was used. After defining the rule type, initial limits, and expected quality of the rules (positive predictive value, minimum number of patients), the program creates a set of rules by varying the limits. The general rule type is defined as: disease = lim 11 < p 1 ≤ lim 1u and lim 2l < p 2 ≤ lim 2u and … lim nl < p n ≤ lim nu. When defining the rule types, only the parameters (p 1 … p n) that are known as clinical electrocardiographic criteria (amplitudes [mV] of Q, R, and T waves and ST-segment; duration [ms] of Q wave; frontal angle [degrees]) were used. This allowed for submitting the learned rule sets to an independent investigator for medical verification. It also allowed the creation of explanatory texts with the rules. These advantages are not offered by the neurons of a neural network. The learned rules were checked against a test set and the following results were obtained: MI: sensitivity 76.2%, positive predictive value 98.6%; LVH: sensitivity 72.3%, positive predictive value 90.9%. The specificity ratings for MI are better than 98%; for LVH, better than 90%.

A new large signal HBT model

Several effects important for large signal operations of heterojunction bipolar transistor (HBTs) were not included in the previous HBT models used in most commercial circuit simulators. Exclusion of these effects resulted in large discrepancies between modeled and measured device characteristics. This paper presents a new large signal HBT model which takes into account those important effects for the device operation. The effects have been identified from measured device characteristics and can be justified from first principles. To make it easy to use, the model is made up of the elements available from SPICE. During the course of the model development, an extraction procedure for the model parameters has been established to minimize the uncertainty of the extracted parameter values. The new model has been applied to HBTs with various emitter sizes and excellent agreement has been achieved between modeled and measured data over a wide range of bias conditions and signal frequencies.

Effective parameter extraction using multiple-objective function for VLSI circuits

Effective parameter extraction is a crucial step in accurate simulation of microelectronic circuits and systems. A parameter extraction program, which is based on a global optimization algorithm, uses a single objective function that minimizes the drain current error between the predicted and measured data. Parameter sets extracted on this basis are adequate for the purpose of digital circuit simulation, but fall short of simulation requirements for analog circuits. The use of a multiple-objective function is proposed, which simultaneously optimizes several critical electrical quantities including the drain current, output conductance, and transconductance. Experimental results with the multiple-objective function are presented, to show the improvement in extracted parameters. The recently developed BSIM_plus MOS transistor model for sub-half-micron integrated circuits uses a compact set of parameters, which greatly enhances the ability to accurately extract parameter values. This model was implemented into the parameter extraction program and some extraction results are presented. The parameter space has several local minima within which a gradient descent method may be trapped. Simulated annealing techniques can be applied to find near-optimal solutions of problems containing multiple local minima in their solution spaces. Experimental results showing the optimization of drain current error using simulation annealing techniques are presented.

Analytical extraction method for submicron MOS transistor model parameters in the linear region

A novel four-point technique for direct extraction of model parameters of submicron transistors in the linear region is presented. Detailed analytical expressions are given for the extraction procedure. Theoretical sensitivity analysis shows that the influence of measurement noise can be reduced by proper choice of the applied voltages for which the data points are measured. As an example, a sensitivity of only 2% in the threshold voltage was obtained for 1% noise in the four data points. Experiments show that the extracted parameter values are constant within large intervals of applied voltages. Finally, as a generalisation of the direct extraction method, a least-square fitting technique is suggested which offers the user full freedom concerning the number of data points to be used.

OPTIMA: A nonlinear model parameter extraction program with statistical confidence region algorithms

A device model parameter optimization program, OPTIMA, for extracting parameter values of empirical and/or analytical models most commonly used for VLSI circuit simulation, is described. Such device models often have parameters which are correlated, and some of them could be redundant. OPTIMA can automatically detect redundant model parameter combinations. This not only allows one to extract more meaningful parameters, but also helps in the development of improved physical models with a minimum number of empirical parameters. The parameter redundancy is detected using a statistical confidence region algorithm which can be implemented as a postprocessor to any gradient-based least-squares optimization method. The advantage of the statistical confidence region algorithm in OPTIMA, as applied to MOSFET model parameter extraction, is discussed, using examples from drain and substrate current modeling. >

Application of permittivity tensor for accurate interpretation of magneto-optical spectra

The inaccuracies of widely used approximations to the theory of the off-diagonal elements ε1= ε′1 + iε″1 of the permittivity tensor are examined. Errors occurring in the ‘‘diamagnetic’’ case can become important when curve fitting is employed to extract parameter values from Bi-substituted magnetic garnets. A discrepancy between the exact and approximate forms of the ε′1 relation (which is responsible for Faraday rotation and Kerr ellipticity at lower energies) as a function of transition energy is found in the low-energy (wavelength λ∼1 μm) tails of the lines and originates from the dropping of a nonresonant term from the complete theory. Errors also occur in the line shapes near resonance where the ratio of the excited-state splitting to the Lorentzian linewidth approaches unity (Δ/Γ→1). A family of curves with Δ/Γ as a parameter is presented, and an exact expression relating Δ/Γ to the resonance value of ε′1/ε″0 (where ε″0 is the imaginary part of the diagonal element ε0) is derived.

Concurrent global optimisation for microwave-device parameter estimation

A novel concurrent implementation of a global-optimisation algorithm is used to extract parameter values for MESFET equivalent-circuit models. The problem of the location of the global minimum of an error function is addressed by the use of an efficient stochastic global-optimisation program which does not require accurate parameter estimates and which operates on a multiprocessor array. The program has been used to estimate the parameter values of a process-based physical equivalent model of a MESFET under multibias small-signal operation. These results are compared with those derived from an electrical equivalent-circuit model.

Variance study of burette aliquot volumes

An experimental study using gravimetry was carried out in order to measure the variances observed in aliquot volumes delivered by graduated burettes operating with various flow-rates and surface tensions and with the burette tip immersed and not immersed in the receiving liquid. Aliquot mass versus meniscus level data were analysed for the aliquot volume variance by a maximum likelihood calculation that corrected for random and systematic mass errors including liquid evaporation. The observed variances provide estimates of titrant volume standard deviations for analytical chemists using non-linear regression analysis to extract parameter values and standard errors from potentiometric titration data.

Nucleation calculations in a pair-binding model.

Calculations of the nucleation and growth of thin films are presented. These atomistic calculations depend on adsorption (${E}_{a}$), diffusion (${E}_{d}$), and lateral binding (${E}_{b}$) energies. A simplified pair binding model of small two-dimensional clusters is used to make the calculations explicit for layer-plus-island (or Stranski-Krastanov) growth systems. Within such a model it is found that at least a crude (Einstein) representation of surface vibrations is needed to make reasonable predictions at low supersaturation. The calculations are applied to extract parameter values from nucleation and growth experiments on Ag/W(110), Ag/Mo(100), Ag/Si(111), and Ag/Si(100), and for rare gases onto various (plated) substrates. Comments are made about the parameters obtained for these systems, and about the role of surface crystallography and defects.

The resistivity, temperature coefficient of resistivity and thermoelectric power of thin continuous metal films I: A survey and critical appraisal of the application of processing methods to experimental data

In this paper the results are reported of an investigation into the accuracy and applicability of the methods which have been used to extract parameter values from sets of experimental data points. Theoretical models describing resistivity, temperature coefficient of resistivity and thermoelectric power of thin continuous metal films and experimental processing methods are identified and categorized. A critical appraisal is given of experimental work reported in papers published between 1961 and 1984. This is followed by details of investigations into the methods commonly used for fitting data to exact and thick film relationships. These investigations are presented within the framework of the Fuchs-Sondheimer relationships. However, the major problem areas that are identified are of a general nature and do not appear to be limited to this model. The problems with the methods investigated, especially those associated with thick film approximations, are shown to be so great as to cast doubt on many published values.

An Efficient MOS Transistor Model for Computer-Aided Design

This paper describes a dc model for MOS enhancement-mode transistors. The transport current and threshold voltage equations are simpler in form than many other models that are currently in use. This greatly eases the problem of extracting parameter values from measured data, since major parameters are directly related to process constants and secondary parameters follow from realistic approximations. The model has been successfully fitted to a high degree of accuracy to several NMOS and CMOS processes. When installed in a Computer-Aided Design (CAD) program, the model can substantially improve the execution time for circuit simulations.

Impedance analysis of a tight epithelium using a distributed resistance model

This paper develops techniques for equivalent circuit analysis of tight epithelia by alternating-current impedance measurements, and tests these techniques on rabbit urinary bladder. Our approach consists of measuring transepithelial impedance, also measuring the DC voltage-divider ratio with a microelectrode, and extracting values of circuit parameters by computer fit of the data to an equivalent circuit model. We show that the commonly used equivalent circuit models of epithelia give significant misfits to the impedance data, because these models (so-called "lumped models") improperly represent the distributed resistors associated with long and narrow spaces such as lateral intercellular spaces (LIS). We develop a new "distributed model" of an epithelium to take account of these structures and thereby obtain much better fits to the data. The extracted parameters include the resistance and capacitance of the apical and basolateral cell membranes, the series resistance, and the ratio of the cross-sectional area to the length of the LIS. The capacitance values yield estimates of real area of the apical and basolateral membranes. Thus, impedance analysis can yield morphological information (configuration of the LIS, and real membrane areas) about a living tissue, independently of electron microscopy. The effects of transport-modifying agents such as amiloride and nystatin can be related to their effects on particular circuit elements by extracting parameter values from impedance runs before and during application of the agent. Calculated parameter values have been validated by independent electrophysiological and morphological measurements.