Przemyslaw Jakub Gromala

Cure dependent characterisation of moulding compounds

Curing effects and difference in thermal contraction of components cause residual stresses and warpage during encapsulation of electronic packages. Residual stresses combined with thermal and mechanical loads influence package reliability and performance and may eventually lead to product failure. Comprehensive material characterisation is needed in order to perform numerical simulations which take into account the effect of shrinkage and stress development during the curing reaction of the moulding compound. With that, more reliable predictions can be made.

Comprehensive material characterization and method of its validation by means of FEM simulation

Numerical simulation plays an important role in product design. Its accuracy relays on a detailed description of geometry, material models, load and boundary conditions. This paper focuses on a new approach of FEM material modeling of three commercially available molding compounds. Curing shrinkage, modulus of elasticity and coefficient of thermal expansion were measured and implemented into commercially available FEM code Ansys. Fringe pattern technique has been used to measure warpage of bimaterial strips. Then FEM simulation of bimaterial strips were done and compared with experimental results. Curing shrinkage has been modeled in an effective way. Its accuracy has been checked on one of the materials by creating bimaterial strips with three different geometrical dimensions, that is varied thickness of mold and copper substrate.

Prognostic approaches for the wirebond failure prediction in power semiconductors: A case study using DPAK package

Reliability of the standard DPAK component under passive temperature cycling as well as combined passive and active temperature cycling is investigated. A special test vehicle is designed to mount six DPAK components with various orientations on a PCB. The stress state during the passive temperature cycling is monitored using three IForce piezoresistive stress sensors mounted on one side of the PCB. In addition, three temperature sensors are used for in-situ temperature measurement, which is critically required for active operation of the designed test vehicle during active power cycling and/or field condition tests. In this paper, only the results of the passive temperature cycling are presented. Based on the preliminary results, a relation between the damage evolution and the failure of the DPAK can be established, which offers a potential application of using the IForce sensor as a canary device for Prognostics and health management.

Simulation driven design of novel integrated circuits - Part 2: Constitutive material modeling of thermosets

This paper presents the comparison of the simulation capabilities of material behaviors based on the linear viscoelastic (LVE) model and non-linear viscoelastic (NLVE) model for a commercially available molding compound. Both modeling approaches do consider the effect of time and temperature dependency. However, it is found that the LVE modeling technique is not capable of reproducing the large strain, highly non-linear mechanical behavior of the tested thermosets. In order to model this highly non-linear behavior quantitatively, the Bergstrom-Boyce (BB), a non-linear hyperviscoelastic model, is used. When the NLVE model is used to describe the thermo-mechanical behavior of the material, significant improvement in the prediction of time and temperature dependent behavior, as well as in the strain dependent relaxation is achieved. In order to validate the accuracy of BB model, three point bending test and cyclic tensile tests with relaxation segments during loading and unloading to zero, which are not previously used for the calibration of the model, are used. It is also shown and discussed that the proposed methodology can be successfully used in the simulation driven design process in which development of the ECU is supported by quantitative numerical simulation methods.

Investigation of Uncertainty Sources of Piezoresistive Silicon Based Stress Sensor

The aim of this paper is to get insight into measurement uncertainties for thermomechanical measurements performed using a piezoresistive silicon-based stress sensor in a standard microelectronic package. All used sensors have the same construction, were produced in the same technological processes at the same time, yet the measurement results show significant distribution. The possible causes for this phenomenon are discussed in this paper. Additionally, Finite Element Method (FEM) model is created and validated, what enables a study of sensitive parameters influencing the measurement uncertainties.

Finite strain thermomechanical material characterization of adhesives used in automotive electronics for quantitative finite element simulations

Thermoset-based adhesives are used as thermal and electrical interfaces. These adhesives are filled with different particles in order to meet the requirements of heat transfer and electrical properties. In automotive applications, they are required to have excellent adhesion since bulk cracking and/or delamination may precipitate other electrical, thermal or mechanical failure mechanisms. With the help of finite element analysis, it is possible to calculate the behavior of the joint and to locate regions of stress and strain concentration where failure is expected to initiate. However, the accuracy of numerical calculations is dependent on the validity of the material models used in the analysis to describe the deformation behavior of the adhesive and adherents. Linear elastic (LE), elastic-plastic (EP) and linear viscoelastic (LVE) material models are frequently used in microelectronics industry. However, up to now in microelectronics industry, there is no work where the limitations of these material models are discussed. The present paper addresses the above issue. We will show the limitations of LVE models and propose a nonlinear viscoelastic (NLVE) model which is capable to describe the large strain behavior of the observed material behavior. Although the NLVE model is illustrated for an adhesive, similar behavior is also observed at other organic materials such as molding compounds and lamination foils. Thus, the suggested NLVE material model has the potential to be applied to a very wide-range of materials. The authors present LVE (between -40°C and 200°C) and NLVE (at 25°C and 100°C) characterization and modelling of the adhesive. For LVE characterization, dynamic mechanical analysis (DMA) and pressure-volume-temperature (PVT) experiments are used. Results are combined to obtain a LVE model which is described by the Prony terms and shift function. Validation of the LVE model is performed at small and large strains with the help of a newly designed dogbone geometry, which is developed in the course of this work to eliminate the disadvantages of the existing DIN EN ISO 527-2 standard. For validation, static tensile tests (STT) and static tensile tests with stress relaxation segments (STSR) up to failure are used. It is found out that the LVE model is capable of predicting the mechanical behavior of the adhesive only at small strains and fails to represent the highly nonlinear mechanical behavior. As it is crucial to predict the adhesive strength at large strains, already obtained STT and STSR results are used to fit the Bergstrom-Boyce (BB) NLVE material model. It is shown that the BB model can accurately describe the material behavior which is observed from STT and STSR experiments. In order to validate the BB model, static tensile tests with creep segments (STCR), which are not previously used for the calibration of the model, are used. A comparison of LVE and NLVE material models is also presented for the STCR simulations. In order to check the behavior of the BB model at temperatures other than the material model input temperatures (25°C and 100°C), STCR experiments at 70°C are also performed and simulated. In all cases, when compared to the LVE material model, NLVE BB model is shown to improve the predictions of the experimental results. Thus, the BB model is shown to be useful for adhesives. This will allow designers to perform quantitative FE simulations of adhesive joints.

Smart Features Integrated for Prognostics Health Management Assure the Functional Safety of the Electronics Systems at the High Level Required in Fully Automated Vehicles

The current developments in automotive industry toward automated driving require a massive increase in functionality, number, and complexity of the electronic systems. At the same time, the functional safety of those electronic systems must be improved beyond the high requirements applied today already. Designing the systems for a guaranteed lifetime on statistical average will no longer suffice. Therefore, new methods in the design and reliability assessment toward maintainable or replaceable systems are required. Prognostics and health management (PHM) provides the way for this upgrade in reliability methodology. The paper introduces a multi-level PHM strategy based on smart sensors and detectors integrated into the functional electronic units so that maintenance can be triggered if needed yet always well before the actual failure occurs in the individual system.

Condition Monitoring Algorithm for Piezoresistive Silicon-Based Stress Sensor Data Obtained from Electronic Control Units

Recent advancements in automotive technologies, most notably autonomous driving, demand electronic systemsmuch more complex than realized in the past. The automotiveindustry has been forced to adopt advanced consumerelectronics to satisfy the demand, and thus it becomes morechallenging to assess system reliability while adopting the newtechnologies. The system level reliability can be enforced byimplementing a process called condition monitoring. In thispaper, a piezoresistive silicon based stress sensor isimplemented to detect physical damages in outer moldedelectronic control units (ECU) subjected to reliability testingconditions. The test vehicle consists of six DPAK powerpackages and three stress sensors mounted on a PrintedCircuit Board (PCB). A unique algorithm is proposed andimplemented to handle the data obtained from thepiezoresistive stress sensing cells. The accuracy of measureddata is examined by Finite Element method (FEM), and thephysical changes are validated with Scanning AcousticMicroscope (SAM).

Hybrid Approach to Conduct Failure Prognostics of Automotive Electronic Control Unit

A model/sensor hybrid approach is implemented to conduct failure prognostics of an automotive electronic control unit (ECU). A 3-D finite element model simulating a complex ECU is built, and its predictability is calibrated and verified by an optical displacement measurement technique called moiré interferometry. The stress states of the ECUs during thermal cyclic loadings are documented by piezoresistive-based stress sensors embedded in the test vehicles. The verified model is then utilized to develop a quantitative relationship between the stress senor data and the stress of the most critical locations in the ECU. The modeling and verification steps that lead to the predictive finite element model are described. The stress history of the critical locations obtained by the hybrid approach is presented.

Simulation Driven Design of Novel Integrated Circuits -- Physics of Failure Simulation of the Electronic Control Modules for Harsh Environment Application

Reduction of development time of an advanced Electronic Control Unit (ECU) requires optimization of the product design at the early stage of development. A simulation-driven design has been used effectively to define the ideal position of components for optimum reliability even before the first prototype is manufactured. Such methods are based on a virtual design of experiments that allows to investigate different layouts or to optimize the ECU. In this paper a physics of failure simulation of the ECU is presented for automotive applications. A test vehicle with power packages on a PCB is designed and tested to enable the analysis of system behavior with three different relative orientations of DPAKs (Discrete Package). The failure modes are identified by FMMEA, and risk assessment is conducted using a systematic simulation approach. The approach involves a sequential numerical simulation of electrical, thermal and thermo-mechanical analysis. The final results are analyzed with a special emphasis on the design elements, where the failure is expected.