CALCE EPSC Graduate Student Theses and Dissertation Abstracts (2009)

Askari, Alex F. (M.S. Mechanical Engineering)
Modeling and Simulation of Shock and Drop Loading for Complex Portable Electronic Systems
In this study, the transient response of electronic assemblies to mechanical loading encountered in drop and shock conditions are investigated. Many manufactures face design challenges when evolving new designs for high-strain rate life-cycle loading. Examples of high-strain rate loading include drop events, blast events, vibration, ultrasonic process steps, etc. New design iterations invariably bring new unexpected failure modes under such loading and costly trial-and-error design fixes are often necessary after the product is built. Electronics designers have long sought to address these effects during the design phase, with the aid of computational models. However, such efforts have been difficult because of the nonlinearities inherent in complex assemblies and complex dynamic material properties. Our goal in this study is to investigate the ability of finite element models to accurately capture the transient response of a complex portable electronic product under shock and drop loading. While many researchers have shown qualitative ability for such modeling, further work is still needed to demonstrate good quantitative agreement. The product consists of a circuit card assembly in a plastic housing. Dynamic loading, consisting of various shock profiles, is applied using an electrodynamic shaker. A Limited number of drop tests are also conducted on a drop tower. The modeling is conducted in ABAQUS/Explicit. In-plane strains and accelerations are the parameters that are compared to assess the agreement between the model and the experimental results. The long-term goal of this study is to demonstrate a systematic methodology to predict failure modes during the design phase of future products.


Brock, Garry (M.S. Mechanical Engineering)
Reliability of Multilayer Ceramic Capacitors
Flexible termination capacitors were designed to reduce stresses transmitted to the ceramic dielectric of a capacitor and prevent flex cracking. In my studies, a set of experiments was conducted to compare the effects of termination type (standard vs. flexible), presence of a conformal coating (acrylic coating vs. no coating), and voltage bias level on multilayer ceramic capacitors (MLCCs). Exposure to the temperature-humidity-bias (THB) conditions induced changes in the electrical parameters, with the insulation resistance experiencing the greatest change, dropping by two to three orders of magnitude when measured at the elevated temperature and humidity. Both permanent and intermittent failures were also observed during THB testing. The in-situ monitoring showed similar failure statistics between the flexible and standard termination capacitors, but little correlation to the presence of conformal coating or the voltage bias level. Recovery occurred only in the standard termination MLCCs upon removal from THB conditions. Flexible termination capacitors at the rated voltage bias were found to have more permanent failures after exposure to THB testing as compared to standard termination capacitors. Failure analysis indicated that silver and palladium migration between electrodes was the failure mechanism in the biased flexible termination capacitors.


Eaidgah, Youness (M.S. Mechanical Engineering)
Supply Chain Management and Production Engineering
Supply chain management traditionally is concerned about flow of material and information along the chain. In this two dimensional way of thinking, primary goals are improving service level and reducing cost. Even though backbone of companies is cash and the main principle is generating profit, most supply chain managers neglect impacts of SCM on financial performance of their company. Moving form a two dimensional way of thinking (flow of material, flow of information) toward a three dimensional way of thinking (flow of material, flow of information, and cash flow) would bring a significant advantage to entire the chain. My main research interest is “Investigating the impact of supply chain management policies on financial performance of companies” through qualitative and quantitative methods. By means of qualitative analysis, I would like to gain insight about the interaction between SCM policies and financial performances. Using quantitative methods, I would like to develop or select a set of metrics that have enough power to reflect supply chain performance. Putting this set of metrics beside financial ratios, and using mathematical models and simulation tools, it would allow me to reveal quantitative patterns. These quantitative patterns would yield in-depth information about the impact of supply chain policies on the financial performance of companies. These patterns would managers to align SCM policies with company financial goals. I will use the result of the above research to investigate effect of Counterfeit on both supply chain and financial performance of companies. However through acquiring counterfeit parts, companies would be able to shorten their lead-time and improve their order fulfillment rate; nevertheless, in long run, their service level will decline and they lose assets.


Emmel, Rachel (M.S. Mechanical Engineering)
Development of a Test Metohdology to Determine Dynamic Fracture Strength in Microfabricated MEMS Structures
As micro-electro-mechanical systems (MEMS) are becoming more and more common in both military and consumer electronics, virtual qualification of these devices remains an important design tool. To model MEMS devices subjected to high shock loading, the dynamic fracture strength of the most widely used MEMS material, single crystal silicon (SCSi), is needed. Minimal research has been performed to determine this material property and the research that has been completed suggests that fracture strength varies considerably with processing parameters. Therefore, an efficient and inexpensive testing method to determine the dynamic fracture strength of processed SCSi has been developed.

Experimentation with SCSi MEMS structures has also been carried out using this new testing method. A probabilistic Weibull distribution for bending of DRIE processed SCSi around the <110> directions was created as a design for reliability tool. Additional experiments demonstrated that the fracture strength for bending of DRIE processed SCSi around the <100> directions is greater than 1.1 GPa. Suggestions for subsequent work that focuses on the bending of SCSi around the <100> directions are also presented.


Zhou, Yuxun (Ph.D. Mechanical Engineering)
Harmonic and Random Vibration Durability Investigation for SAC305 (Sn3.0Ag0.5Cu) Solder Joint
Vibration loading is commonly encountered during the service life of electronic products. However, compared to thermal cycling durability, vibration durability is more complex and has been less investigated. In surface mount technology, solder joints are the primary mechanical, thermal and electrical interconnects between the component and the PWB. So the reliability of solder joints is very crucial for most electronic assemblies. The vibration durability of Pb-free solder joints is the focus of this dissertation.

The characteristics of the stress from vibration loading are low amplitude and high frequency, while those from cyclic thermal loading are high amplitude and low frequency. In this study, several exploratory vibration tests were conducted, using both narrow band and broad-band, step-stress excitation at several different isothermal and thermal cycling conditions. The effect of thermal pre-aging on solder joint vibration failures was also investigated. Some of the vibration durability results were analyzed in detail, to obtain quantitative insights into the vibration fatigue behavior of the SAC305 solder material. A time-domain approach was adopted to investigate the durability of solder interconnects under different kinds of vibration and quasi-static mechanical loading.

First, the solder interconnects were subjected to narrow-band (harmonic) vibration loading. The test were conducted at the first natural frequency of the test board using constant-amplitude excitation and solder fatigue properties were extracted with the help of a time-domain analysis that is based on quasi-static finite element simulation. Compared to broad-band step-stress vibration durability tests, the advantage of the harmonic constant-amplitude test is less complexity in the model extraction process, hence, less uncertainty in the desired fatigue constants. Generalized strain-based S-N curves have been obtained for both SAC305 and Sn37Pb solder materials. The strain-life model constants show that SAC305 solder material has superior fatigue properties compared to Sn37Pb solder material under low-cycle fatigue loading, while the reverse is true for high-cycle fatigue loading. These results are consistent with test results from other researchers. In actual application, SAC305 assemblies almost always fail before Sn37Pb assemblies under comparable vibration excitation because of (i) higher solder strain at a given excitation level; and (ii) multiple failure modes such as copper trace cracking.

Next, durability was investigated under step-stress, broad-band (random) excitation. These test results show that SAC305 interconnects are less durable than Sn37Pb interconnects under the random excitation used in this study, which agrees with the harmonic durability results. The random and harmonic durability results were quantitatively compared with each other in this study. Finite element simulation was used to investigate the stress-strain response in the interconnects. The output of this simulation is the strain transfer function due to the first flexural mode of the PWB. This transfer function is used to obtain the solder strain from the measured board strain. This fatigue assessment method demonstrated that the model constants obtained from the harmonic test overestimate the fatigue life under random excitation by an order of magnitude. The causes for this discrepancy were systematically explored in this study. The effects of cyclic loading and mean stress on the vibration durability were addressed and found to be minimal in this study. The stress-strain curves assumed for the solder material were found to have a very large effect on the durability constants, thus affecting the agreement between harmonic and random durability results. The transient response of the components on the test board under both harmonic and random excitation was also included in the strain transfer function with the help of dynamic implicit simulation, and found to have a much stronger effect on the vibration durability at the high frequencies used in broad-band excitation compared to the low frequency used in narrow-band test. Furthermore, the higher PWB vibration modes may play a strong role and may need to be included in the strain transfer-function. This study clearly reveals that the solder strain analysis for broad-band random excitation cannot be limited to the quasi-static strain transfer-function based on the first PWB flexural mode, that has been used in some earlier studies in the literature.

The time-domain approach used in this study provided fundamental and comprehensive insights into the key factors that affect vibration durability under different types of excitation, thus leading to a generalized S-N modeling approach that works for both harmonic and random vibration loading.


Gu, Jie (Ph.D. Mechanical Engineering)
Prognostics of Solder Joint Reliability Under Vibration Loading Using Physics of Failure Approach
Physics-of-failure (PoF) is an approach that utilizes knowledge of a product's life cycle loading and failure mechanisms to perform reliability modeling, design, and assessment. Prognostics is the process of predicting the future reliability of a system by assessing the extent of deviation or degradation of a product from its expected normal operating states. When prognostics is combined with physics-of-failure models, it is possible to make continuously updated reliability predictions based on the monitoring of the actual environmental and operational conditions of each individual product.

A literature review showed that the research on prognostics of solder joint reliability under vibration loading is very limited. However, personal portable electronic products are no longer used exclusively in a benign office environment. For example, any electronic component (throttles, brakes, or steering) in an automobile should be able to survive in a vibration environment.

In this thesis, a methodology was developed for monitoring, recording, and analyzing the life-cycle vibration loads for remaining-life prognostics of solder joints. The responses of printed circuit boards (PCB) to vibration loading were monitored using strain gauges and accelerometers, and they were further transferred to solder strain and stress for damage assessment using a failure fatigue model. Damage estimates were accumulated using Miner's rule after every mission and then used to predict the life consumed and the remaining life. The results were verified by experimentally measuring component lives through real-time daisy-chain resistance measurements.

This thesis also presents an uncertainty assessment method for remaining life prognostics of solder joints under vibration loading. Basic steps include uncertainty source categorization, sensitivity analysis, uncertainty propagation, and remaining life probability calculation. Five types of uncertainties were categorized, including measurement uncertainty, parameter uncertainty, model uncertainty, failure criteria uncertainty, and future usage uncertainty. Sensitivity analysis was then used to identify the dominant input variables that influence model output. After that, a Monte Carlo simulation was used for uncertainty propagation and to provide a distribution of accumulated damage. From the accumulated damage distributions, the remaining life was then able to be predicted with confidence intervals. The results showed that the experimentally measured failure time was within the bounds of the uncertainty analysis prediction.


Khuu, Vinh (Ph.D. Mechanical Engineering)
Evaluation of Thermal Interface Materials and the Laser Flash Method
Thermal interface materials (TIMs) are used to reduce the interfacial thermal resistance between the chip and the heat sink, which has become a bottleneck to heat removal in a variety of electronic applications. Degradation in thermal performance of the TIM can contribute to unacceptably high chip temperatures, which can significantly impact device or system performance during operation. While progress has been made in recent years in the development of tools to measure beginning-of-life thermal performance, characterizing the long-term performance of the TIM can be crucial from a life cycle stand point since TIMs may experience harsh operating conditions, including high temperature and high humidity, for extended periods of time in typical applications.

The laser flash method is one approach for measuring thermal conductivity that has an advantage over more commonly used techniques because of the non-contact nature of the measurement. This technique was applied to 3-layer structures to investigate the effects of thermal cycling and elevated temperature/humidity on the thermal performance of select polymer TIMs in pad form, as well as an adhesive and a gel. While most samples showed little change (less than 10% in thermal resistance) or slight improvement in the thermal performance, one thermal putty material showed degradation due to temperature cycling resulting from bulk material changes near the glass transition temperature. Scanning acoustic microscope images revealed delamination in one group of gap pad samples and cracking in some putty samples due to temperature cycling.

Finite element simulations and laser flash measurements performed to validate the laser flash data indicated that sample holder plate heating, an effect previously unexamined in the literature, can lead to inaccurately high TIM thermal conductivity values due to suppression of the sample temperature rise during the laser flash measurement. This study proposed a semi-empirical methodology to correct for these effects. Simulated laser flash test specimens had bondlines that showed little thickness variation (usually within the measurement error) due to clamping by the sample holder plates. Future work was proposed to refine the laser flash sample holder design and perform additional validation studies using thermal test vehicles based on nonfunctional packages.


Kumar, Sachin (Ph.D. Mechanical Engineering)
Prognostic and Health Management of Electronic Products
The research work involves development of methodologies for detecting and isolating faults, identifying degradation in product’s health, and estimating probable time to failure. These methodologies are un-supervised and make use of a multivariate distance measure; the Mahalanobis distance (MD). An electronic product was subjected to various environmental and usage conditions in order to collect its performance data, which was used to characterize the product and the data was also used for training these methodologies. The effect of different environmental and uses conditions on the product’s performance parameters was studied and some unique observations were made, and reported. For the fault detection, a probabilistic approach was developed to establish a threshold MD value as opposed to classical approach that was based on an expert opinion or a MD corresponding to known fault. To isolate faults, a residual MD based approach was developed to identify faulty parameters in order to locate fault in a product. For identifying product’s health degradation, a health indicator was defined from observations (i.e., MD) made during a time interval. In each time interval, a histogram of MD values was constructed, and the health indicator was calculated by taking the weighted sum of a bin's fractional contribution. Establishment of a threshold health indicator was based on the probabilistic measure of product’s health. For the probabilistic estimation of time to failure, a non-linear dynamic Markov model was developed using the symbolic representation of product’s health. A symbolic time series was created by applying wavelet transform in order to reduce noise and retain time information associated with original health signal (i.e., MD). Few anomaly measures were defined for detecting existing and emerging faults or failures. The applicability of these methodologies was evaluated for real time health assessment of an electronic product. This research expanded the applicability of Mahalanobis distance from fault detection to fault isolation, to degradation identification, and to prognostics.


Lopez, Leoncio (Ph.D. Mechanical Engineering)
Quality and Reliability of Land Grid Array Sockets
Integrated Circuit (IC) sockets provide hundreds to thousands of electrical interconnects in enterprise servers, where quality and reliability are critical for customer applications. The evaluation of IC sockets, according to current industry practices, relies on the execution of stress loads and stress levels that are defined by standards, having no consideration to the physics of failure (PoF), target operating environment, or contact resistance behavior over time. In a similar manner, monitoring of contact resistance during system operation has no considerations to the PoF or environmental conditions. In this dissertation a physics of failure approach was developed to model the reliability of elastomer sockets that are used in an enterprise server application. The temperature and relative humidity environment, at the IC socket contact interface, were characterized as a function of external environmental conditions and microprocessor activity. The study applied state-of-the-art health monitoring techniques to assess thermal gradients on the IC socket assembly and to establish an operating profile that could be used for the development of a PoF model.

A methodology was invented for modeling and monitoring contact resistance of electrical interconnects. The technique, called M-SPRT, combines a PoF model with the Sequential Probability Ratio Test (SPRT). In the methodology the resistance behavior is characterized as a function of temperature. The effective a-spot radius was extracted from the characterization data and modeled with a power law. A PoF model was developed to estimate the resistance of an elastomer contact, based on the effective a-spot radius and the ambient temperature. The methodology was experimentally demonstrated with a temperature cycle test of the elastomer socket. During the evaluation the difference between estimated and observed resistance values were tested with the SPRT. The technique was shown to be very accurate for modeling contact resistance and to be highly sensitive for the detection of resistance degradation. A model was developed to estimate the reliability of an elastomer contact during the first year of life. To derive the model, the resistance behavior of contacts under nominal mechanical load was studied as a function of time and temperature. The elastomer contact was shown to have a very complex resistance behavior, which was modeled by multiple statistical distributions. The reliability of the elastomer contact at normal operating temperature was estimated with a log-normal distribution and a PoF model of the mean contact resistance.


Nie, Lei (Ph.D. Mechanical Engineering)
Reliability of Reballed and Reworked Ball Grid Array Packages in SNPB and SAC Assembly
In recent years, many countries banned the use of lead in select high volume electronic equipment. However, exemptions from lead-free legislation have been granted for certain products, especially those intended for high-reliability applications. Manufacturers with exemption are facing dwindling supply of lead based parts for their products. This change has left many high reliability electronic equipment manufacturers with the choices of, mixing lead-free components in tin-lead assembly process, converting products to lead-free, or reprocessing lead-free components to comply with the tin-lead assembly process.

Reballing has been used for component reclamation, but right now it offers a way to reprocess the ball grid array package. The reliability of reballed BGA assembly needs to be determined before the implementation. Mixing lead-free ball grid array packages with eutectic tin-lead solder paste bring new challenges to the current electronic industry. The long term reliability of mixed assembly needs to be investigated. Although rework is not a new technology, the impact of multiple rework process on the reliability of lead-free and mixed assemblies is still unknown.

Lead-free ball grid array (BGA) packages with Sn3.0Ag0.5Cu solder balls were subjected to both reballing process. Ball shear test and cold bump test were used to investigate the solder ball attachment strength of the reballed BGAs. Temperature cycling test was used to evaluate the reliability of reballed tin-lead, lead-free and mixed assemblies. The solder ball strength and the reliability of reballed BGAs were independent of the reballing method. The reliability of mixed assemblies was equivalent to that of lead-free assemblies. Microstructure differences in lead-free, mixed and reballed SnPb assemblies were investigated to explain the reliability results.

Lead-free and mixed assemblies were subjected to the rework process. Temperature cycling test was used to evaluate the reliability of reworked assemblies. Copper over-consumption, copper pad dissolution and thick interfacial intermetallic layer were found in the reworked assemblies. Microstructural investigation and geometry analysis were used to analyze the reliability degradation in the reworked assemblies after multiples rework processes.


Panashchenko, Lyudmyla (M.S. Mechanical Engineering)
Evaluation of Environmental Tests for Tin Whisker Assessment
Tin whiskers are electrically conductive crystalline structures of tin that over time may grow outward from tin-rich surfaces and present a reliability hazard to electronic systems. While the problem has been known for decades, no satisfactory explanation of whisker growth mechanisms exists, leaving the industry to create whisker-assessment tests based on empirical data gathered under various environmental storage conditions controlled for temperature, humidity and temperature cycling. The long-term predictability of these environmental storage tests has not been addressed and the accuracy of these tests in foreseeing whisker growth is unclear.

In this thesis, different tin finishes are assessed for whisker growth in accordance with existing environmental test standards and compared to growth seen in ambient storage conditions. The results indicate that environmental tests may over-predict, under-predict, or show little distinguishable growth as compared to ambient-stored tin finishes. In conclusion, environmental tests are not a reliable method of assessing future whisker growth.


 

Roettele, Shaughn M (M.S. Mechanical Engineering)
Probabilistic Physics of Failure Assessment of Thermomechanical Fatigue ofHigh-I/O Area-Array Interconnects
Thermal cycling durability of Plastic ball grid array (PBGA) interconnects is known to decrease as I/O count increases. This is due, in part, to deterministic effects; increasing package size results in increasing CTE mismatches between component and PWB. Deterministic failure predictions are based on the load level found in the critical joint (joint with the most severe loading). However, due to probabilistic effects, for example manufacturing variabilities, premature failure may result in one of several joints in the neighborhood of the critical one. Failure probability increases as the number of joints in this critical region increases. Thus, observed failure rates are due to a convolution of deterministic and probabilistic effects. In effect, for large BGAs, deterministic predictions may overestimate interconnect durability. This thesis uses thermal cycling experiments and detailed mechanistic modeling to present a methodology for adjusting deterministic predictions of solder joint failure with a suitable probabilistic correction factor.





Copyright © 2008 by CALCE and the University of Maryland, All Rights Reserved