Abstract: Stress relaxation in stamped metal LGA sockets can result in a loss of normal force and an increase in contact resistance, potentially leading to a failure. This paper describes an approach for determining the risks from stress relaxation in stamped metal LGA sockets by taking into account the effect of Joule heating of the socket contacts. Stress relaxation data were obtained at different temperatures and strain values, representative of both operating and overload conditions. Contact resistance measurements were conducted as the force was varied to determine the minimum force below which stress relaxation is likely to cause failure. Joule heating of the socket contacts was found to cause a measurable rise in temperature with a typical value of ~40°C above the unpowered state of 90°C at the maximum rated current of 3 A for the socket being studied. This temperature rise was determined to be a significant factor in stress relaxation and was found to cause an average reduction in normal force by ~26%. The properties of the polymer housing were found to be sensitive to Joule heating effects and to have a significant influence on the stress relaxation behavior of the socket.
Abstract: This paper presents a study on the contact resistance behavior of elastomer sockets used to interconnect microprocessors and printed circuit boards in enterprise servers. The integrated circuit sockets, installed in production representative assemblies, were evaluated at 25degC, 55degC, and 75degC for 2000 h. A sample subset was evaluated up to 16 500 h at 25degC and up to 4500 h at 55degC. The test results show that contact resistance decreases over time for all test conditions, as much as 50% from their initial values. Elastomer contact behavior is strongly dependent on temperature and time. The resistance behavior over temperature is modeled with multiple statistical distributions. The mean contact resistance is represented with a physics-of-failure model, and the elastomer contact reliability is estimated using a log-normal distribution.
Abstract: We present a methodology based on the physics of failure, and the sequential probability ratio test, for modeling and monitoring electrical interconnects in health monitoring, and electronic prognostic applications. The resistance behavior of an electrical contact was characterized as a function of temperature. The physics of failure of the contact technology were analysed. A contact resistance model was selected, and its parameters were fitted using the temperature characterization data. The physics of failure model was evaluated with a reliability application (temperature cycle test), and was found to produce estimation errors of < 1 mOmega of during a training period. The temperature and resistance of ten sample contacts were continuously monitored during the temperature cycle test, identifying the maximum temperature and resistances for each cycle. Using the physics of failure model, maximum resistance estimates were generated for each test sample. The residual between the monitored and estimated resistance values was evaluated with the sequential probability ratio test. The method was shown to overcome the issues of traditional threshold-based monitoring approaches, providing accurate resistance estimates, and allowing the detection of abnormal resistance behavior with low false alarm and missed alarm probabilities.
Abstract: Electronic Prognostics (EP) is a technique used in high-reliability and high-availability systems to actively and proactively detect faults, allowing the reduction of system downtime and unplanned repairs. The approach of Sun Microsystems to EP consists of a Continuous System Telemetry Harness (CSTH) that is coupled with Sequential Probability Ratio Test (SPRT) and Multivariate State Estimation Technique (MSET) algorithms. This approach provides a unique and complete EP solution, harnessing the rich information from sensors and system variables, and providing means for their storage and analysis. The background theory behind SPRT and MSET techniques as well as their implementation for advanced EP in enterprise servers is presented in this paper.
Abstract: The phenomenon of reverse-bias electroluminescence (EL) has been exploited extensively to elucidate the physics of failure of Si-based semiconductor devices. In this article, we study the reverse-bias emission of vertical-cavity surface-emitting lasers (VCSELs) that have been stressed under accelerated temperature, humidity, and bias. Using quantitative emission microscopy in conjunction with other electrical and optical measurements, we show that the dark line defect dislocation network responsible for VCSEL degradation can be imaged directly as reverse-bias EL and that the spectra of the reverse-bias EL provide further insight into the failure mechanism.
Abstract: We evaluate vertical-cavity surface-emitting lasers (VCSELs) reliability for next-generation high-productivity computers wherein several hundreds of terabits of bandwidth are envisioned. VCSEL failure rates are modeled, and an empirical relationship for VCSEL scaling versus bit rate and aperture is presented to explore the reliability of VCSEL-based links. The effects of VSCEL sparing, water cooling, and redundancy at the system level are analyzed.
Abstract: Many decisions regarding the design with a system using sockets are made using a point estimate (often from a standard or accepted practice) of the compressive load. The components of a CLGA assembly were analyzed using Monte Carlo simulation methods to determine the distribution of the compressive loads that are applied to the component socket. The components of the assembly that impact the resulting load were identified and their individual distributions defined. The relationship between the height of the components and the compression provided by the springs was determined and used to estimate a load distribution. The load distribution thus estimated can be used as an input to further virtual qualification efforts. For example, the estimated load distribution from this analysis was used with other manufacturing parameters and FEA models to estimate the IC socket contact compression. The results showed that the minimum compression required for a good electrical contact was achieved over 99.9% of the time. Therefore, information that is obtained from virtual qualification was shown to be of great value in the determination of proper test conditions and design decision making.
Abstract: Land grid array (LGA) sockets provide a solderless printed circuit board (PCB) attachment method for microprocessors that require high interconnect density. Stamped metal LGA sockets consist of metal contacts embedded in a plastic housing, that form the electrical and mechanical connection between the component and PCB by means of compression. To achieve a stable contact interface between the socket and component/PCB, it is necessary to maintain a minimum normal force. Stress relaxation results in a loss of normal force, and could cause an increase in contact resistance, potentially leading to a failure. Contact alloy manufacturers report stress relaxation data for the alloy in strip form. This data may not be an appropriate measure of the stress relaxation in an LGA socket because it does not take into account the geometry of the increasingly miniaturized contacts or the effect of the surrounding polymer. This paper therefore aims to study the stress relaxation of stamped metal LGA sockets, by testing the actual socket in its final processed form and capturing the effect of the polymer on stress relaxation. Furthermore, by testing small pieces of the socket, independent of the loading assembly, a method of studying the temperature and load dependence of stress relaxation and the effect on the socket's electrical behavior was demonstrated, without the need for system level testing.
Abstract: Typical reliability evaluations of integrated circuit (IC) sockets are performed with stress loads and load levels that are defined by industry standards, without consideration to component properties, reliability requirements, or target operating environments. The problem with this approach is evident when evaluations are completed. If the standard requirements are met, then the test results provide no assurance of adequate product life for the customer. If the standard requirements are not met, then the test results provide no clear indication as to which failures are relevant and which are not. An alternative approach for the estimation of product reliability is found in health monitoring and electronic prognostics, where the reliability of an IC socket can be estimated from operating condition monitors, and physics of failure analysis and modeling. The results obtained in this manner can provide greater assurance that reliability requirements will be met and that the incipience of failures will be detected, with corresponding root causes identified. In this paper the operating environment of an IC socket inside an enterprise server was experimentally assessed by means of a sensor network, providing vast information for the definition of reliability models and future implementations of health monitoring and electronic prognostic methods.
Abstract: The evaluation of integrated circuit sockets for quality, reliability, and/or prognostic applications is performed with accelerated tests and stress conditions that are defined by industry standards. These methods typically use threshold values to detect degradation and failures of the electrical interconnect. However, this approach does not consider the physics of failure of the component under test and is not sensitive enough to detect small changes in behavior. This paper presents a physics of failure based methodology that, combined with a sequential probability ratio test, overcomes the issues of the traditional approach. This new approach, called Maxima-SPRT, enables early detection of degradation and provides the information needed for the successful implementation of prognostic tools. The method is validated using an elastomer socket, a daisy chain package, and a test board assembly.
Abstract: Some high performance computing systems interconnect the microprocessor and the printed circuit board (PCB) using an integrated circuit (IC) socket. In these systems the socket provides a low resistance electrical connection when its contacts are compressed between the IC and the PCB pads. Obtaining and maintaining an acceptable contact resistance is essential for the seamless operation of these interconnects. The electrical resistance of an IC socket contact is a function of applied load, contact compression, and contact surface quality. To apply load to a socket (and compress the contact) a mechanical system is utilized which may consist of springs, screws, bolster plate, heat sink, and other components. The analysis of contact compression and contact resistance in an IC socket assembly is not a simple task. It requires extensive material characterization and testing. Due to rapid design, vendor selection cycles, sample size requirements, and cost, the physical testing to estimate the variability of contact resistance in socket assemblies is not feasible. Probabilistic methods such as finite element analysis (FEA) are used to overcome this obstacle. FEA provides the means to model the response of an assembly to compressive loads and variation in the mechanical load system components. An issue with this approach is that each finite element analysis run takes hours to execute, making it an impractical approach for evaluations that consider numerous parameters and parameter values. This paper describes how the response surface methodology (RSM) can be utilized to reduce the number of FEA runs that are required for the analysis of contact compression and contact resistance. Statistical distributions of the parameters of interest are constructed and used with RSM and Monte Carlo simulations to generate a transfer function. This function is evaluated thousands of times by means of Monte Carlo simulations to quantify the probability that the contact compression (and cont- - act resistance) requirements may not be met, while considering manufacturing variability. The approach allows the virtual qualification of socket assemblies without resorting to expensive testing and prohibitive FEA computation cycles.
Abstract: VCSEL arrays are being considered for use in interconnect applications that require high speed, high bandwidth, high
density, and high reliability. In order to better understand the reliability of VCSEL arrays, we initiated an internal project
at SUN Microsystems, Inc. In this paper, we present preliminary results of an ongoing accelerated temperature-humiditybias
stress test on VCSEL arrays from several manufacturers. This test revealed no significant differences between the
reliability of AlGaAs, oxide confined VCSEL arrays constructed with a trench oxide and mesa for isolation. This test did
find that the reliability of arrays needs to be measured on arrays and not be estimated with the data from singulated
VCSELs as is a common practice.
Abstract: In this paper, we describe the experiment and review our results to date. Degraded VCSEL arrays have been censored and are currently being subjected to a detailed physical characterization. This analysis will provide a glimpse into the manifestation of VCSEL failures, the mechanisms and physics of emission degradation. Optical interconnects will be required for the bandwidth needs for computing systems in the near future. We have investigated the specific parameters that produce the most reliable VCSELs - the emitter engines for optical interconnects.
Abstract: The need for high bandwidth, high speed interconnects with optimum routing through computer backplanes has led to the use of optical interconnects in multiprocessor computing systems. Most of the current commercially available optical interfaces are based upon 850nm vertical-cavity surface-emitting lasers (VCSELs). Extensive studies conducted by the VCSEL manufacturers show that the reliability of these devices continues to improve. In order to understand the risks and implications of using VCSEL-based modules in computer systems, we have conducted an experiment designed to provide insight into the emission degradation and failure of VCSEL devices. In this paper we briefly describe the experiment and review the results of the subsequent failure analysis on degraded VCSEL arrays. [Material example: AlGaAs].
Abstract: 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 developed for modeling and monitoring contact resistance of electrical interconnects. The technique combined 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 qualitative reliability model was developed for the mean contact resistance of an elastomer socket, using fundamental material properties and user defined failure criteria. 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. It was shown that elastomer sockets, in spite of experiencing stress relaxation at the macroscale (elastomer), can exhibit decreases in contact resistance, a result of stress redistribution at the microscale (Ag particles), which increases Ag-Ag particle stress and the effective contact area.
Abstract: A test system including a package with interconnect paths. The package may have electrical paths that are electrically connected by the interconnect paths. The electrically connected electrical paths may yield increased data without significantly increasing the required testing hardware.
Abstract: One embodiment of the present invention provides a system that performs a real-time root-cause-analysis for a degradation event associated with a component under test. During operation, the system monitors a telemetry signal collected from the component, and while doing so, attempts to detect an anomaly in the telemetry signal. If an anomaly is detected in the telemetry signal, the system performs a failure analysis on the telemetry signal in real-time while the telemetry signal is degrading. Next, the system identifies a failure mechanism for the component based on the failure analysis.
Abstract: A system that detects the onset of degradation for interconnections in a component within a computer system. During operation, the system monitors inferential variables associated with the interconnections during operation of the computer system. Next, the system determines a present state of the component from the monitored inferential variables. The system then compares the present state of the component with an initial state of the component. If the comparison indicates that the interconnections in the component have reached or will reach a limited operating state (LOS), the system performs a remedial action.
Abstract: A method for testing a substrate by using a photoemission microscope is provided which includes providing the substrate and applying a reverse bias voltage to the substrate. The method further includes detecting light emissions from a top surface of the substrate and characterizing viability of the substrate from the detection of the light emissions.
Abstract: A system that characterizes degradation of a component in a system. During operation, the system monitors inferential variables associated with a specimen of the component. Next, the system determines a time for the onset of degradation for the specimen and determines a time for the completion of degradation for the specimen. The system then computes a time interval between the onset of degradation and the completion of degradation, and uses the time interval to look up an entry in a defect library to obtain information which characterizes the degradation of the specimen of the component.
Abstract: One embodiment of the present invention provides a system that tests the quality and/or the reliability of a component. During operation, the system applies test conditions to a plurality of specimens of the component. While applying the test conditions, the system measures the same variable from each of the plurality of specimens. Next, the system computes a running average of the measured variable across the plurality of specimens. The system then computes residuals between the measured variable for each specimen and the running average. The system next determines from the residuals whether the associated specimens are degraded.
Abstract: A method for assessing a reliability of a complex component having a plurality of similar subcomponents is described. The complex component is divided into a plurality of component parts, each component part including at least one of the subcomponents. Specification ranges are identified for measured variables within which the component parts are considered to be operating within specification. The complex component is subjected to a reliability test and reliability data is obtained for the component parts by comparing the measured variable for each of the component parts with the specification ranges to determine failure times. A system and machine readable medium embodying program instructions is also provided.
Abstract: One embodiment of the present invention provides a system that determines the reliability of a component in a system. During operation, the system monitors inferential variables associated with a number of specimens of the component. The system then collects degradation data by first computing a likelihood value that indicates whether an inferential variable associated with a specimen of the component is behaving normally or abnormally. Next, the system determines whether the specimen of the component has degraded based on the likelihood value. If the specimen of the component is determined to have degraded, the system records the time when the specimen of the component was determined to have degraded. The system also uses the degradation data to determine the reliability of the component in the system.
