Micro- and Opto-Electronic Materials, Structures, and Systems

Transcription

1 Micro- and Opto-Electronic Materials, Structures, and Systems Series Editor E. Suhir University of California, Santa Cruz, CA, USA For further volumes:

2 X.J. Fan E. Suhir Editors Moisture Sensitivity of Plastic Packages of IC Devices Foreword by C.P. Wong 123

3 Editors X.J. Fan Department of Mechanical Engineering Lamar University Beaumont,Texas USA E. Suhir ERS Co. Alvina Court Los Altos, California USA ISBN e-isbn DOI / Springer New York Dordrecht Heidelberg London Library of Congress Control Number: Springer Science+Business Media, LLC 2010 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper Springer is part of Springer Science+Business Media (

4 Foreword Moisture, according to the Merriam-Webster dictionary, is defined as liquid diffused or condensed in relatively small quantity. It is also defined, by Longman dictionary, as small amounts of water that are present in the air, in a substance, or on a surface. Both definitions emphasize on the small quantity, but presumably the state of moisture is in liquid form. In fact, moisture, a small quantity of H 2 O molecules, can be in vapor, liquid, or solid phase in air or in any substance. It is interesting to note that moisture sorption is different from water sorption. Moisture sorption process refers to a process in a humid air environment, while water sorption refers to a complete immersion into water. Hydrophobic or superhydrophobic (with water contact angle >150 ) materials can effectively prevent water liquid from penetrating through surface, but not for the transmission of moisture (water vapor) in a humid air environment. This interesting phenomenon is illustrated in the first chapter of the book, although the entire book is focused on moisture sensitivity of plastic packages of integrated circuit (IC) devices. Most of polymeric materials in IC packaging absorb moisture from an environment. The presence of moisture in plastic materials alters thermal stress through the alteration of thermo-mechanical properties, induces hygroscopic stress through differential swelling, induces vapor pressure that is responsible for the eventual popcorn cracking, reduces interfacial adhesion strength, induces electrical-chemical migration-induced corrosion, and finally alters dielectric properties of materials. Despite the pivotal role of moisture, research activities in moisture-induced failure remain relatively low, compared to the thermally induced failure in IC packaging. This is in part due to the lack of material data and aggravated by the lack of fundamental understanding of moisture transport, material characterization techniques, and procedures. These are reflected in the limited publications and the near absence of such properties from the material vendors. The development of three-dimensional (3D) microelectronics packaging with through-silicon via (TSV), ultrathin, and multi-die stacking technology has become essential to increase functionality with higher memory capacity in more complex and efficient architectures. Wafer thinning is required for such ultrathin die development from the original thickness of 750 μm down to as low as 30 μm. Consequently, new assembly and manufacturing processes must be invented to overcome thin wafer handling and cracking issues. Many new materials such as wafer-level coating v

5 vi Foreword films have emerged. As a result, cohesive film rupture may occur due to moisture during reflow. Therefore, one of the major challenges for a practical realization of 3D microelectronics packaging concept is to design materials and meet reliability requirements without cohesive failures subjected to moisture loads. This book provides information on the state-of-the-art technologies and methodologies related to moisture issues in plastic packages. The book covers the wide aspects including moisture diffusion and desorption, characterization and modeling, hygroscopic swelling, interfacial adhesion degradation, accelerated moisture sensitivity/reflow test, electrical-chemical migration, moisture-aging effect on longterm reliability, and several finely selected real-world case studies on various failure mechanisms due to moisture. This is the first book ever to cover the full spectrum of moisture-induced failure mechanisms in IC packages. It is a timely and important contribution to the technical literature for researchers, engineers, and practitioners both in academia and in electronics industry. The editors of the book, Dr. Fan and Dr. Suhir, have rich experience in both theoretical development and industrial practice. They have been offering the professional development courses at various IEEE (Institute of Electrical and Electronics Engineers) CPMT (components packaging and manufacturing technologies) Society conferences, and hundreds of participants have attended their lectures. They have succeeded in bringing together well-recognized experts in this field and present a fine collection of papers covering the full spectrum of the related topics. They are to be congratulated for bringing this very important topic forth in a timely manner. Atlanta, GA November, 2009 C.P. Wong

6 Preface Since moisture-sensitive plastic materials were introduced in integrated circuit (IC) device packaging several decades ago, moisture has been one of the major concerns for package designers and reliability engineers. With the recent development of the three-dimensional (3D) microelectronics packaging with through-silicon via (TSV) and multi-die stacking technologies, moisture-induced failures have become even more prominent due to the new materials employed and the overall reduction in package size and thickness. This book provides a comprehensive state-of-the-art and in-depth review of the fundamental knowledge and methodologies in the field of material and structural ( physical ) behavior and performance of various types of moisture-sensitive plastic packages of IC devices. The book consists of 21 chapters divided into six sections: (1) moisture diffusion, absorption and desorption, and adhesion degradation (Chapters 1, 2, 3, and 4); (2) hygroscopic swelling characterization and analysis (Chapters 5, 6, and 7); (3) integrated hydrothermal and thermal stress modeling (Chapters 8, 9, 10, 11, and 12); (4) case studies and applications (Chapters 13, 14, 15, 16, 17, 18, and 19); (5) electro-chemical migration (Chapter 20); and (6) molecular dynamics modeling and characterization (Chapter 21). Brief description of the chapter contents is set forth below. Chapter 1 presents an overview of moisture-induced failures in plastic packages of IC devices, and illustrates the fundamental characteristics of moisture diffusion, hygroscopic swelling, and adhesion degradation. Chapter 2 describes the latest investigations of anomalous moisture diffusion and the corresponding adhesion behaviors in epoxy molding compounds. Chapter 3 provides a method and detailed analysis for real-time moisture absorption and desorption in thin films. Chapter 4 reviews the existing methodologies of moisture diffusion modeling and whole-field vapor pressure analysis. Chapters 5, 6, and 7 describe several characterization methods and techniques for hygroscopic swelling, such as photomechanics measurement techniques and point measurement method using thermo-mechanical and thermo-gravimetric analyzers. Chapters 8, 9, 10, 11, and 12 provide a collection of the most advanced analysis and methods for integrated hydrothermal stress modeling. Chapter 8 describes recent progress in modeling of moisture diffusion and moisture-induced stresses in semiconductor and MEMS packages. Chapter 9 presents a novel methodology for integrated vapor pressure, hygro-swelling, and vii

7 viii Preface thermo-mechanical stress modeling of IC packages. Chapter 10 describes a failure criterion for moisture sensitivity of plastic packages based on the theory of thin flexible plates of large deflections. Chapter 11 develops a continuum theory and describes its application to moisture-induced failures in IC packages. Chapter 12 reviews recent efforts to develop micromechanics-based failure theories/models and computational tools for material and process selection in the design and fabrication of plastic IC packages and provides recommendations for the improvement of their reliability under the anticipated service conditions. Chapters 13, 14, 15, 16, 17, 18, and 19 are dedicated to several case studies on moisture-induced failures in a wide range of package types, such as QFP (quat flat package), QFN (quat flat no-lead), and D 2 Pak (Chapter 14), QFN package (Chapter 15), system-inpackages and BGA (ball grid array) packages (Chapter 16), flip chip BGA packages (Chapter 17), and ultrathin 3D stacking die packages (Chapter 18). From material perspectives, epoxy molding compounds, die attach adhesives, and underfill materials are all covered in these case studies. Chapter 13 describes a new methodology for an equivalent acceleration of the IPC/JEDEC moisture sensitivity levels. Chapter 19 reviews an automated simulation system to perform moisture-related modeling for various package types. Chapter 20 describes the fundamentals of the phenomenon of the electrochemical migration (ECM), primarily manifested as bridging metallic dendrites. Chapter 21 shows how molecular dynamics simulation and the nano-scale characterization methods could be used to obtain an insight into the moisture-induced failure modes and mechanisms at the atomistic level. The original scope of the book was based on the professional development course notes of one of the editors, Dr. Fan, on moisture-related reliability in electronic packaging, which has been presented at the IEEE (Institute of Electrical and Electronics Engineers) CPMT (components packaging and manufacturing technologies) Society-sponsored conferences. To present a complete coverage on the latest development and the most recent advances in this field, we have invited experts in this field to bring together a full spectrum of moisture-induced failure mechanisms in IC packages. We are grateful to all the authors from the industry and the academia for their in-depth contributions and their efforts to bring this book to the readers. The first editor, Dr. Fan, would like to express his gratitude to many of his ex-colleagues at Intel, Philips Research, and the Institute of Microelectronics in Singapore. Many of the book chapters reflect the results of numerous collaborative efforts and extensive team work. Without this work our book would never be possible. The second editor, Dr. Suhir, expresses his deep appreciation to his friends and former colleagues at Bell Laboratories, Physical Sciences and Engineering Research Division, at Murray Hill, NJ, and Allentown, PA, for introducing him about 25 years ago to the subject of, and the challenges in, the exciting field of polymeric materials in general and plastic packages of IC devices in particular. Dr. Suhir would like to take this opportunity to acknowledge, with thanks, his collaborations, for almost 20 years, during the golden age of Bell Labs, with Shiro Matsuoka, Phil Hubbauer, Lloyd Shepherd, Louis Manzione, Don Dahringer, Harvey Bair, C.P. Wong, Arturo

8 Preface ix Hale, John Segelken, Alan Lyons, Bonnie Bachman, Charles Cohn, Quazi Ilyas, and many other top-notch materials scientists, physicists, chemists, and chemical engineers. Beaumont, TX Santa Cruz, CA November 2009 X.J. Fan E. Suhir

9 Series Preface This title is the second book in the series. The series encompasses a broad area of micro-, opto-electronic, and photonic engineering, with particular emphasis on materials, physics, mechanics, design, reliability, and packaging. The titles in the series feature eminent engineers and scientists as authors and/or editors focused on addressing major issues in the above areas of engineering. Our objective is to have a comprehensive series on the materials, mechanics, physics, packaging, functional performance, and reliability as they pertain to micro- and opto-electronics. The audience for these volumes are those who work in micro- and optoelectronics and photonics, as well as those in many related areas of applied science and engineering. The expected readers are practitioners and professionals, scientists and researchers, along with senior-level undergraduate and graduate students. These volumes can serve as expanded encyclopedias in the field of the mechanics of microand opto-electronic materials and structures. Selected titles could also serve as textbooks, reference works, and as general guidance works for those interested in these subjects. The series contains both descriptions of the state-of-the-art developments in particular fields, as well as new results obtained by authors, editors, and their colleagues. The authors also identify and address crucial, but still unresolved, issues that come up when discussing new developments and issues within the discussed topics. I am thankful to Dr. Fan, the editor of this title, who did the major work by bringing together an excellent team of experts and by putting together many outstanding chapters in this title. It has been a pleasure working with him. I would also like to take this opportunity to thank the authors and editors of the books that are now in the process of being written as well as those authors who have already completed their volume for this series. Potential authors, editors, and those specialists interested in making contributions to the current state of knowledge in a particular field of engineering or applied science within the scope of this book series are invited to send their book proposals to me. Santa Cruz, CA E. Suhir, Ph.D. Series Editor xi

10 Contents 1 Fundamental Characteristics of Moisture Transport, Diffusion, and the Moisture-Induced Damages in Polymeric Materials in Electronic Packaging... 1 X.J. Fan and S.W.R. Lee 2 Mechanism of Moisture Diffusion, Hygroscopic Swelling, and Adhesion Degradation in Epoxy Molding Compounds M.H. Shirangi and B. Michel 3 Real-Time Characterization of Moisture Absorption and Desorption Y. He and X.J. Fan 4 Modeling of Moisture Diffusion and Whole-Field Vapor Pressure in Plastic Packages of IC Devices X.J. Fan, T.Y. Tee, X.Q. Shi, and B. Xie 5 Characterization of Hygroscopic Deformations by Moiré Interferometry E. Stellrecht, B. Han, and M. Pecht 6 Characterization of Interfacial Hydrothermal Strength of Sandwiched Assembly Using Photomechanics Measurement Techniques X.Q. Shi, X.J. Fan, Y.L. Zhang, and W. Zhou 7 Hygroscopic Swelling of Polymeric Materials in Electronic Packaging: Characterization and Analysis J. Zhou, T.Y. Tee, and X.J. Fan 8 Modeling of Moisture Diffusion and Moisture-Induced Stresses in Semiconductor and MEMS Packages C. Jang and B. Han 9 Methodology for Integrated Vapor Pressure, Hygroswelling, and Thermo-mechanical Stress Modeling of IC Packages T.Y. Tee xiii

11 xiv Contents 10 Failure Criterion for Moisture-Sensitive Plastic Packages of Integrated Circuit (IC) Devices: Application and Extension of the Theory of Thin Plates of Large Deflections E. Suhir and X.J. Fan 11 Continuum Theory in Moisture-Induced Failures of Encapsulated IC Devices X.J. Fan, J. Zhou, G.Q. Zhang, and A. Chandra 12 Mechanism-Based Modeling of Thermaland Moisture-Induced Failure of IC Devices H.B. Chew, T.F. Guo, and L. Cheng 13 New Method for Equivalent Acceleration of IPC/JEDEC Moisture Sensitivity Levels B. Xie, X.J. Fan, and X.Q. Shi 14 Moisture Sensitivity Level (MSL) Capability of Plastic-Encapsulated Packages J.K. Fauty 15 Hygrothermal Delamination Analysis of Quad Flat No-Lead (QFN) Packages M.S. Zhang, S.W.R. Lee, and X.J. Fan 16 Industrial Applications of Moisture-Related Reliability Problems W.D. van Driel, D.G. Yang, C.A. Yuan, and G.Q. Zhang 17 Underfill Selection Against Moisture in Flip Chip BGA Packages X.J. Fan, T.Y. Tee, C.Q. Cui, and G.Q. Zhang 18 Moisture Sensitivity Investigations of 3D Stacked-Die Chip-Scale Packages (SCSPs) X.Q. Shi, X.J. Fan, and B. Xie 19 Automated Simulation System of Moisture Diffusion and Hygrothermal Stress for Microelectronic Packaging Y. Liu 20 Moisture-Driven Electromigrative Degradation in Microelectronic Packages L.F. Siah 21 Interfacial Moisture Diffusion: Molecular Dynamics Simulation and Experimental Evaluation H. Fan, E.K.L. Chan, and M.M.F. Yuen About the Editors Subject Index

12 Contributors E.K.L. Chan Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, A. Chandra Department of Mechanical Engineering, Iowa State University, Ames, IA 50011, USA, L. Cheng Department of Mechanical Engineering, National University of Singapore, Singapore, Singapore , H.B. Chew Department of Mechanical Engineering, National University of Singapore, Singapore, Singapore , C.Q. Cui Compass Technology Co., Ltd., Shatin, Hong Kong, People s Republic of China, cqcui@cgth.com H. Fan Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, mehaibo@ust.hk X.J. Fan Department of Mechanical Engineering, Lamar University, PO Box 10028, Beaumont, TX 77710, USA, xuejun.fan@lamar.edu J.K. Fauty East Decatur Street, Mesa, AZ 85205, USA, jkfauty@msn.com T.F. Guo Department of Mechanical Engineering, National University of Singapore, Singapore, Singapore , mpev9@nus.edu.sg B. Han Department of Mechanical Engineering, CALCE Electronics Products and Systems Center, University of Maryland, College Park, MD 20742, USA, bthan@umd.eduu Y. He Intel Corporation, Assembly Test & Technology Development, 5000 W. Chandler Blvd., Chandler, AZ 85226, USA, yi.he@intel.com C. Jang Department of Mechanical Engineering, University of Maryland, Glen Martin Hall, College Park, MD 20742, USA, csjang@umd.edu xv

13 xvi Contributors S.W.R. Lee Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, Y. Liu Fairchild Semiconductor Corp., 82 Running Hill Rd, South Portland, ME 04106, USA, B. Michel Department of Micro Materials Center Berlin (MMCB), Fraunhofer Institute for Reliability and Microintegration (IZM), Volmerstraße 9B, Berlin, Germany, M. Pecht Department of Mechanical Engineering, CALCE Electronics Products and Systems Center, University of Maryland, College Park, MD 20742, USA, X.Q. Shi Applied Science & Technologies Research Institute (ASTRI) 5/F, Photonics Center, Science Park, Shatin, Hong Kong, People s Republic of China, danielshi@astri.org M.H. Shirangi Department of Micro Materials Center Berlin (MMCB), Fraunhofer Institute for Reliability and Microintegration (IZM), Volmerstraße 9B, Berlin, Germany; Robert Bosch GmbH, Automotive Electronics, Development ASIC & Power Packages, Reutlingen, Germany, hossein.shirangi@izm.fraunhofer.de L.F. Siah Assembly Technology Development Q&R Malaysia (ATD Q&R-M), Intel Technology (M) Sdn. Bhd., Bayan Lepas FTZ Box 121, Pulau Pinang, Malaysia, lay.foong.siah@intel.com E. Stellrecht Department of Mechanical Engineering, CALCE Electronics Products and Systems Center, University of Maryland, College Park, MD 20742, USA, estell-recht@drs-ewns.com E. Suhir Department of Electrical Engineering, University of California, 1156 High Street, SOE2, Santa Cruz, CA ; Department of Mechanical Engineering, University of Maryland, College Park, MD, ERS Co. LLC, Los altos, CA, USA, suhire@aol.com T.Y. Tee Blk 408, Hougang Ave 10, # , Singapore, Singapore , tongyan.tee@gmail.com W.D. van Driel Philips Lighting, Mathildelaan 1,5611 BD Eindhoven, The Netherlands; Delft University of Technology, Mekelweg 2,2628 CD Delft, The Netherlands, willem.van.driel@philips.com C.P. Wong Materials Science and Engineering, Georgia Institute of Technology, 771 Ferst Drive, N.W. Atlanta, GA , USA, cp.wong@mse.gatech.edu B. Xie Applied Science & Technologies Research Institute (ASTRI) 5/F, Photonics Center, Science Park, Shatin, Hong Kong, People s Republic of China, bxie@astri.org

14 Contributors xvii D.G. Yang Guilin University of Electronic Technology, Jinji Road 1, Guilin, China, C.A. Yuan TNO IenT, Eindhoven, The Netherlands, M.M.F. Yuen Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, G.Q. Zhang Philips LightLabs, Mathildelaan 1, 5611 BD Eindhoven, The Netherlands; Delft University of Technology, Mekelweg 2, 2628 CD Delft, The Netherlands, M.S. Zhang Department of Mechanical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, Y.L. Zhang School of Mechanical & Production Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore, Singapore , J. Zhou Department of Mechanical Engineering, Lamar University, PO Box 10028, Beaumont, TX 77710, USA, W. Zhou School of Mechanical & Production Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore, Singapore ,

15 Chapter 1 Fundamental Characteristics of Moisture Transport, Diffusion, and the Moisture-Induced Damages in Polymeric Materials in Electronic Packaging X.J. Fan and S.W.R. Lee 1.1 Introduction Many failures in microelectronic packages can be traced back to moisture [1, 2]. The mechanical behavior of polymer systems is affected significantly by the absorption of atmospheric moisture [3 8]. In general, there are three types of failure mechanisms when an electronic package is exposed to humidity conditions [8]. The first type of failure is often referred to as popcorn failure [4 7]. Packaged microelectronic devices are exposed to factory environmental conditions after the opening of protective dry-bags for surface mounting to printed circuit board (PCB). When moisture, present in the factory environment, is absorbed by the polymeric packaging materials, it condenses in free volumes or micro-/nanopores in the bulk materials and along interfaces. The condensed moisture will vaporize and produce high pressure during the surface mount process. The board and devices are entirely placed in a reflow oven, in which the peak temperature ranges typically from 220 to 260 C. The reflow process is completed within a few minutes. Polymer materials, such as dielectric films, adhesives, encapsulants, and plastic printed circuit boards, become extremely compliant when the temperature exceeds their glass transition temperatures. In addition, the interfacial adhesion strength drops substantially. As a result, delamination may occur due to the combined effects of thermo-mechanical stresses, hygroscopic stresses, vapor pressure, material softening, and adhesion degradation. An audible sound may be produced if water vapor is suddenly released due to package cracking. The second type of moisture-induced failure mechanism concerns the package reliability during a lifelong service period under various field environmental conditions. Polymeric materials in packages will swell upon moisture absorption. Hygroscopic swelling creates dimensional changes of materials. Similar to thermal stresses due to the mismatch of thermal expansion, hygroscopic stresses are induced. Moisture will also substantially affect the interfacial adhesion. In addition, aging effect becomes a concern for a long-period exposure to humidity conditions. Consequently, delamination may take place along weaker interfaces X.J. Fan (B) xuejun.fan@lamar.edu X.J. Fan, E. Suhir (eds.), Moisture Sensitivity of Plastic Packages of IC Devices, Micro- and Opto-Electronic Materials, Structures, and Systems, DOI / _1, C Springer Science+Business Media, LLC

16 2 X.J. Fan and S.W.R. Lee to cause electrical failures of devices. The third type of moisture-assisted failure is due to electrochemical migration (corrosion) in the presence of both electrical bias and moisture (see Chapter 20). There are two kinds of corrosion in electronic packages. The first kind is metal dendritic growth at the cathode side on the substrate surface (e.g., copper trace). Electrolytic dissolution of copper at the anode creates metal ions. Metal ions then migrate to the cathode side as moisture provides transport path (e.g., moisture absorbed by the solder resist). When metal ions reach the cathode, dendritic growth occurs. Continuous reduction of dissolved metal ions leads to the nucleation and growth of metal dendrites, and eventually the formation of anode cathode short failure. The second kind of corrosion, the so-called conductive anodic filament (CAF) growth, occurs at the sub-surface associated with glass fibers/epoxy resin interface. The CAF grows from anode to cathode along delaminated fiber/epoxy interfaces when moisture is present. A CAF is made from soluble copper salt at the anode and built at the anode by turning to insoluble salt due to ph effect. Dendrites occur as a result of solution at the anode and plating at the cathode. In order to address those moisture effects, three types of accelerated moisture sensitivity/reliability tests are often applied for package reliability qualifications [2]. Moisture/reflow sensitivity test is required prior to environmental stresses for all devices that are surface mounted to PCB. Moisture/reflow sensitivity test has been documented in the joint JEDEC/IPC industry standard J-STD-020D [9]. This test specification has established exposure conditions of temperature, humidity, and time for which the moisture sensitivity ranking of surface mount devices are classified and referenced. Moisture/reflow sensitivity test insures that the temperature, humidity, and/or the shipping requirements are met before assessing the reliability under operation conditions. The so-called highly accelerated stress test (HAST) (without electrical bias) is the second type of accelerated test to evaluate non-hermetic packaged devices in humid environments. The test employs temperature and humidity to accelerate the penetration of moisture through external protective material or along the internal build-up or joined interfaces. Electrical bias is not applied in this test in order to ensure that the failure mechanisms, which may be overshadowed by the bias, can be detected. This test is used to identify failure mechanisms internal to package and is destructive. An autoclave test or pressure cooker test (PCT) is similar to HAST. PCT has a fixed temperature (121 C) and fixed RH (100%). During HAST and PCT, the packages are placed in environmental chambers with humidity and temperature controls for a certain period of time. Both HAST and PCT are industrial standards for qualification requirements. Lastly, the biased temperature/humidity (TH) test is a moisture-related test used to evaluate the reliability of a powered device at an elevated temperature and in a high-humidity environment subject to the static bias of the nominal operation state (ranging from 0.1 to 7 V). There are two standard forms of biased moisture test that can be employed. The temperature, humidity, bias (THB) test is performed under the atmospheric pressure, at a temperature between 30 and 85 C, and with relative humidity (RH) from 50 to 85%. Highly accelerated stress test with bias (BiHAST) is performed at less than 3 atmospheres of pressure and at a temperature between 110 and 160 C.

17 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 3 In order to fully understand failure mechanisms associated with moisture, it becomes exceedingly important to characterize the moisture absorption behavior of polymeric materials involved in electronic packaging and to quantify polymer matrix damage and adhesion degradation induced by moisture transport. In this chapter, the fundamental characteristics of moisture behaviors in polymer systems and the interactions of moisture with polymer matrix and along interfaces are reviewed. Both Fickian and non-fickian kinetics of moisture absorption and desorption are investigated. Saturated moisture concentrations at different humidity levels as a function of temperature are discussed for different materials in a wide range of temperature. A specific experiment, which was designed to reveal the passage of moisture vapor through a hydrophobic membrane, is described. The difference of water sorption (immersion in water) and moisture sorption in humid air is discussed. Mercury intrusion method is introduced to measure the sizes of nanopores or free volumes for different types of packaging materials. An approximate method to estimate the free volume fraction is proposed based on moisture weight gain test data. Moiré interferometry technique is applied to study the aging effect of hygroscopic swelling of a polymeric material as a function of time. Adhesion and fracture toughness characterization methods are applied to investigate the influence of moisture on the interfacial fracture toughness or the adhesion strength. Finally, certain issues on the state of moisture in polymer materials, effect of hygroscopic swelling, effects of fillers, and the duration of moisture absorption at different conditions are discussed. 1.2 Fickian and Non-Fickian Moisture Diffusion Fickian transport of moisture in polymers or polymer composites is described by the following equation: C t ( 2 ) C = D x C y C z 2. (1.1) This equation is known as the general diffusion law or Fick s second law of diffusion (Fick s law, for short), C (kg/m 3 ) is the moisture concentration, t (s) is time, and D (m 2 /s) is the diffusion coefficient or diffusivity (diffusion constant) of the moisture in the material. The temperature dependence of the diffusion constant can be described by the Boltzmann Arrhenius type of equation: ( D = D 0 exp E ) d, (1.2) kt where D 0 is a pre-factor, E d is the activation energy, k = J/K is the Boltzmann s constant, and T is the absolute temperature. Note that the pre-factor D 0 and the activation energy E d are different below and above the glass transition

18 4 X.J. Fan and S.W.R. Lee temperature [1]. In other words, there exist two sets of constants D 0 and E d for a material to fully describe the moisture diffusivity in it as a function of temperature. Consider a one-dimensional case of an infinite plate of thickness h. The general analytical solution, giving the temporal and spatial moisture concentration, C, at time t and the distance x from the mid-plane, can be obtained C(x,t) C 0 C i C 0 = 1 4 π ( 1) n [ (2n + 1) exp (2n + 1)2 π 2 ] D h 2 t cos n=0 (2n + 1)π x (1.3) h for the initial and boundary conditions C = C 0, h/2 < x < h/2, t = 0, C = C i, x = h/2, x = h/2, t 0, (1.4) where C i is the moisture concentration on the boundary, C 0 is the initial moisture concentration, D is the Fickian diffusion coefficient, and h is the thickness of the plate. Assuming the plate is initially dry and then placed into a humid environment, C 0 = 0 and C i = C sat (the saturated moisture concentration), equation (1.3) yields [ C(x, t) = C sat 1 4 π n=0 ( 1) n [ (2n + 1) exp (2n + 1)2 π 2 ] D h 2 t cos ] (2n + 1)π x h (1.5) and describes moisture absorption process as a function of time. With the time progressing, the plate will be fully saturated with a uniformly distributed moisture with the concentration C sat. On the other hand, if C 0 = C sat and C i = 0, equation (1.3) yields C(x, t) = C sat 4 π n=0 ( 1) n [ (2n + 1) exp (2n + 1)2 π 2 ] D h 2 t cos (2n + 1)π x (1.6) h and describes a desorption process, in which an initially fully saturated plate releases moisture into the dry enviroment. The moisture in the plate will be driven out completely as time goes on. Since it is not possible to measure the moisture concentration at an arbitrary inner point experimentally, equation (1.3) should be integrated over the thickness of the bulk plate so that it would be possible to measure at the plate surfaces the ratio of the amount M t of moisture at a given moment of time t to the amount M of moisture at t. The total moisture mass in the specimen can be obtained as a function of time as follows:

19 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 5 M t M = 1 8 π 2 n=0 [ 1 (2n + 1) 2 exp (2n + 1)2 π 2 ] D h 2 t for moisture absorption, (1.7) M t M = 8 π 2 n=0 [ 1 (2n + 1) 2 exp (2n + 1)2 π 2 ] D h 2 t for moisture desorption, (1.8) where M t is the total mass of moisture at time t and M is the mass related to the saturated concentration C sat by the formula C sat = M V, (1.9) where V is the total volume of the test specimen and M is the saturated mass of moisture over the plate. When analyzing the experimental data, D and M are optimized so that the difference between the experimentally determined mass M t and its calculated value obtained on the basis of equation (1.3) is minimized [10]. The change in the sample volume caused by the moisture absorption desorption is small and may be ignored. The accurate evaluation of the parameters C sat and D is critical to understand the physics of the absorption/desorption process and the related reliability issues. C sat determines the moisture absorption capacity of the specimen and D determines how fast the moisture could diffuse into or out of the sample. In equation (1.1) the Fickian diffusion is assumed to be independent of the moisture concentration. While the Fickian diffusion model provides a reasonable approximation to the characteristics of moisture uptake, it rarely provides a full description of the uptake data and is unable to account for the changes in the polymer due to relaxation, material degradation, or cracking. Non-Fickian behavior takes place as the consequence of the relaxation process in polymer molecules and/or as a result of an irreversible reaction between polymer and moisture, such as formation of hydrogen bonds and chemical reactions [11, 12]. It has been found that non- Fickian diffusion can take diverse forms [5]. Two-stage or dual-uptake diffusion has been observed in some polymeric materials and is referred to in the literature [5] as an anomalous moisture uptake. Moisture weight gain measurement is a common experimental technique to evaluate the moisture uptake. In order to investigate the kinetics of the Fickian and non-fickian moisture absorption, moisture weight gain measurements were carried out for several typical packaging materials. The specimen geometries and the temperature/humidity conditions were as follows: 1. A conventional underfill material. The test sample was a 1-mm-thick disk-like specimen with a diameter of 50 mm. Moisture absorption under 85 C/85% RH was measured.

20 6 X.J. Fan and S.W.R. Lee 2. A thin bismaleimide-triazine (BT) 70-μm-thick core material with the dimensions 7 mm 7 mm. In this study, moisture absorption desorption experiments were performed at temperatures 30, 60, and 80 C, respectively. At each isothermal temperature, two relative humidity (RH) levels were chosen: 60 and 80%RH. In situ measurements were performed on this thin sample. 3. A 0.6-mm-thick bismaleimide-triazine (BT) core material. The sample size was 50 mm 50 mm. Moisture absorption was measured under 30 C/60% RH conditions. 4. A conventional molding compound material sample. Two different sample geometries from the same material in the form of molded disks were fabricated, one with a thickness of 1 mm (MC1) and a diameter of 50 mm and another with a thickness of 2 mm (MC2) and a diameter of 100 mm, respectively. The specimens were exposed to the 85 C/85% RH conditions in a humidity chamber. To determine the dry weights, the samples were initially dried ( baked ) at 125 C for 24 h. An electronic balance was used for the weight measurements, except for the thin BT core, in which an in situ measurement was performed [10]. The samples were periodically removed and weighed and then returned to the chamber for further soaking. Figure 1.1 is the plot of the Fickian fit for moisture uptake for an underfill sample compared to the experimental data. To determine the diffusivity D and the saturated weight gain M, the least-squares fitting technique was used. In this approach, the sum of the square of the differences between the experimental weight gain and the Mt,i cal ) 2, was calculated based on equation (1.7), using some initial estimated values of the parameters D and M. M exp t,i was calculated one, ( M) 2 = k i=1 ( M exp t,i Fig. 1.1 Fickian curve fit for a homogeneous underfill material under 85 C/85% RH (sample size 50 mm 50 mm 1 mm)

21 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 7 the ith point of the experimentally determined weight gain at the moment t of time, while Mt,i cal was calculated on the basis of equation (1.7). Here k is the total number of the experimental points used when calculating ( M) 2. Then, D and M were varied until ( M) 2 was minimized. The obtained D and M are taken as the diffusivity and the saturated moisture mass of the sample for that particular temperature/humidity conditions. As one could see from Fig. 1.1, the Fickian diffusion law predicts the experimental behavior of the test specimen within the test period very well. With the increasing time t, the absorption curve smoothly levels off to the saturation level M. The test results in a previous study on different types of underfill also exhibited Fickian behaviors [2]. An example of moisture absorption desorption experiment for a thin BT core is shown in Fig The temperature was kept at 60 C and the RH level was cycled between 0 and 60%. During the first 60 min, a 0% RH condition was imposed to drive residual moisture out of the sample. This was followed by a 200 min of moisture absorption at 60% RH, then a 200-min desorption at 0% RH. The following conclusions could be drawn from Fig. 1.2: (1) During the first 60 min, the moisture was not completely driven out of the sample. A longer time was needed to accomplish that. (2) The subsequent absorption desorption cycles were reciprocal: the sample reached during re-sorption approximately the same saturated moisture level that the weight was lost as the result of drying. This indicates that the process is completely mechanical ( physical ) and there was no chemical reaction between the water molecules and the material. Figure 1.3 indicates, however, that the Fickian fit is not satisfactory. This could be attributed to the fact that the BT core has a highly inhomogeneous structure involving glass fibers and resin matrix. Rapid moisture diffusion into or out of the substrate at the initial stages of the moisture diffusion might be due to the moisture diffusion through the polymer-based region located at the outer layers of the substrate at the Fig. 1.2 Moisture absorption desorption experiment conducted using a sorption TGA at 60 C with the RH level cycled between 0 and 60% [10] Weight (%) Weight RH C Time (min) Reference Sensor RH (%)

22 8 X.J. Fan and S.W.R. Lee Fig. 1.3 Desorption curves of a 70-μm BTcore.Solid line represents the fitted weight loss versus time curve using D = cm 2 /s by Fick s law [10] Weight Change (μg) Experimental Fitted (D = 1.65 µm 2 /s) Time (sec) beginning of the diffusion process. At the later stages, however, moisture diffusion was forced to progress through fibers which are moisture insensitive. Therefore, the inner components of the substrate acted as a physical barrier against the moisture diffusion, thereby reducing the rate of moisture diffusion or diffusivity. Figure 1.4 is a plot of the moisture weight gain for a 0.6-mm-thick BT core at 30 C/60% RH. After the extended hours of moisture absorption, the moisture uptake started to increase again. One could clearly see a two-stage moisture absorption. The material behaved initially as Fickian, and then exhibited a non-fickian diffusion behavior afterward. The moisture absorption for a molding compound is plotted in Fig. 1.5 for specimens of two different thicknesses. An initial moisture uptake to quasi-equilibrium (the first stage of virtual saturation) is followed by a slower linear moisture uptake. This suggests that at least two different mechanisms are present in the moisture absorption process. The first one is the absorption of the water molecules in free volumes or nanovoids. The second mechanism is the hydrogen bonding formation between the water molecules and the polymer chains due to their molecular polarity. The former mechanism reaches a saturation point and is a reversible phenomenon in nature. The latter seems to be linear without a clear saturation, at least Fig. 1.4 Moisture weight gain curve for a thick BT core sample with the thickness of 0.6 mm subjected to 30 C/60% RH Weight Gain (%) Time (hr)

23 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 9 Fig. 1.5 Normalized moisture uptake of a conventional mold compound with different thicknesses at 85 C/85% RH (x-axis is sqrt(time)/thickness) for the timescale of observation in this study. The thicker sample (MC2) was further exposed to a humid environment for 8 months. The experimental data showed that the moisture weight gain reached the level of about 0.3% after 8 months compared to the 0.2% at the Fickian saturation point. Our latest experimental data showed that about 40% of residual moisture content was kept for MC2 at 110 C after a long period of desorption [11]. Although the composition and chemistry are different for the underfill material, BT, and the molding compound samples, it is not clear what particular mechanisms caused the difference in the moisture uptake for those materials. The molding compound specimen showed a stronger non-fickian absorption kinetics than the underfill specimen. Vrentas and Duda [13] introduced a diffusion Deborah number (DEB) D to characterize the presence of the non-fickian effects during absorption experiments. There are a number of non-fickian diffusion kinetics [5], in which the two-stage non-fickian type of absorption is widely applied. The general manifestation of the two-stage sorption is characterized by a rapid initial uptake, then a quasi-equilibrium process, followed by a slow approach to the final, true equilibrium. Ultimately, complete saturation will be reached in all instances. A theory that satisfactorily describes the features of a two-stage sorption has been suggested by Hopfenberg [14]. In their diffusion relaxation model they considered the absorption process to be composed of two empirical independent contributions: a diffusion part, M F (t), that is governed by Fick s law, and a structural part, M R (t), resulting from polymer relaxations. The total weight gain at time t may be expressed as the linear superposition of these two contributions: M (t) = M F (t) + M R (t) (1.10)

24 10 X.J. Fan and S.W.R. Lee In this equation, M F (t) is given by the solutions to the diffusion equation (1.1). It is assumed that more than one independent relaxation process might be possible, so that the part M R (t) could be expressed by a series M (t) = i M,i (1 e k it ), (1.11) where M,i represents the equilibrium absorption due to the ith relaxation process and k i is the first-order relaxation constant of the ith relaxation process. It has been shown that such a model is able to describe the non-fickian behaviors [5]. Whether the non-fickian effects are considered in the field conditions or in the testing conditions depends on the duration of the time of exposure to moisture. If the exposure time does not exceed the time before polymer relaxation commences, then Fick s law can be accepted to describe the moisture diffusion. Otherwise non-fickian s behavior must be considered. IPC/JEDEC standards require that the duration of moisture exposure at different humidity levels is usually 7 9 days. To determine the exposure time when Fickian law is valid, modeling should be employed. Such modeling should consider material s chemistry, specimen s geometry, and humidity level. 1.3 Saturated Moisture Concentration Saturated moisture concentration C sat is a measure of the moisture absorption capacity under given humidity and temperature conditions. According to the definition of the relative humidity as RH = p actual partial water vapor pressure of the air = 100% (1.12) p sat saturated partial water vapor pressure of the air the partial water vapor pressure of the air for the given relative humidity and absolute temperature is p = RH p sat, (1.13) where p sat is the saturated water partial vapor pressure and is a function of temperature. Assuming that Henry s law applies, we have or p = C sat S (1.14) S = C sat p, (1.15)

25 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 11 where S is the solubility, a material property, which is independent of the ambient relative humidity. From equations (1.13) and (1.15), we obtain C sat = (S p sat )RH. (1.16) Equation (1.16) indicates that the solubility S can be obtained when C sat is measured at different temperatures. Since the solubility and the saturated water vapor pressure are functions of temperature, the saturated moisture concentration is proportional to the relative humidity, as long as Henry s law is fulfilled. Such a linear relationship has been confirmed by the moisture weight gain test obtained for the thin BT sample from the RH step scan experiment, as shown in Fig This figure shows that, at a fixed temperature (30 C in this case), C sat is linearly proportional to the RH level. The saturated moisture concentration in the BT sample is shown in Fig. 1.7 as a function of temperature for two different RH levels. The data were obtained from moisture absorption desorption experiments. Figure 1.7 shows that the parameter C sat in BT core is essentially temperature independent. For most polymer materials at the temperatures well below the glass transition temperature, C sat is temperature independent. This implies that the solubility decreases with temperature according to equation (1.16). The situation is different when the temperature exceeds the glass transition temperature, T g. The saturated moisture concentration for a die-attach film in the range from 30 to 80 C at 60% RH level is shown in Fig. 1.8 as a function of temperature. A strong dependence of temperature is evident. The tested film has a T g at around 35 C. For temperatures exceeding T g, the saturated moisture concentration depends strongly on the temperature. As a matter of fact, since the saturated moisture concentration has strong dependence on the free volume fraction, the saturated moisture concentration increases significantly when temperature exceeds the glass transition temperature. It is important to investigate the saturated moisture concentration C sat (mg/cm 3 ) Experimental Data Linear Fit Slope = 0.11 Fig. 1.6 C sat as a function of relative humidity at 30 C [10] Relative Humidity (%)

26 12 X.J. Fan and S.W.R. Lee Fig. 1.7 Saturated moisture content as a function of temperature for two different RH levels [10] C sat (mg/cm 3 ) μm BT core 60% R.H. 80% R.H Temperature ( C) Fig. 1.8 Saturated moisture concentration as a function of temperature for a die-attach film at 60% RH level (T g of the film is 35 C) Saturated Moisture Uptake (%) Measured Linear Fit Y = T Temperature ( C) at reflow soldering temperatures since an over-saturation phenomenon might take place at the reflow soldering conditions [15]. Over-saturation is a situation that the material might continue to absorb more moisture despite that desorption takes places during soldering reflow. When saturated moisture concentration is a function of temperature, the moisture diffusion modeling using normalization approach is no longer valid [2]. A direct concentration approach (DCA) has been proposed [15, 16] to conduct moisture diffusion modeling at reflow soldering conditions. 1.4 Water Sorption and Moisture Sorption When a polymer material is immersed into water, the sorption process is referred to as water sorption. For a material that is subjected to a humid air condition with a relative humidity less than 100%, it is referred to as moisture sorption process.

27 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 13 There are two distinct diffusion mechanisms involved in the transport of moisture: moisture transfer across the surface and through the bulk material, respectively. In order to study the difference between water sorption and moisture sorption, an experiment of moisture sensitivity/reflow test for two groups of quad flat no-lead (QFN) plastic packages was conducted [17]. One group of the packages was coated with a hydrophobic membrane fabricated by a grafting method [18] to prevent water uptake, while the other group was as received. Two types of moisture preconditioning were applied: immersion into water at a constant temperature of 60 C and placement in a humidity chamber at 60 C/60% RH. Both sorption durations were 192 h. Table 1.1 summarizes the weight gain data and the results after the reflow. The coated packages under water did not exhibit any weight change. However, the coated packages in humid air condition gained almost the same amount of moisture as the uncoated packages either in water or in humid air. Furthermore, the reflow test indicated that the coated packages at 60 C/60% RH preconditioning had the same failure rate as the uncoated packages in the same preconditioning conditions, while the coated packages immersed into water did not show any failures after the reflow. These results indicate that a hydrophobic polymer film is very effective in blocking liquid water from penetrating through the package surface, but not for water vapor transmission (see Fig. 1.9). Several other hydrophobic polymeric materials have been found to exhibit similar behavior, as far as water vapor passage in humid air conditions is concerned (Winkler, P., of Badger Meter Inc. 2009, personal communications). Table 1.1 Effect of hydrophobic membrane on moisture sorption versus water sorption Sorption condition Hydrophobic film coated Averaged weight gain (%) Reflow test failure rate Water immersion (60 C) Yes /12 Water immersion (60 C) No /12 Moisture sorption (60 C/60% RH) Yes /12 Moisture sorption (60 C/60% RH) No / Characterization of Pore Size, Porosity, and Free Volume Porosity of, or free volume fraction in, polymers is a critical material property related to the moisture absorption. Polymer volume is divided into three elements: occupied volume (the van der Waals volume), interstitial free volume, and holefree volume [19]. The hole-free volume is accessible for penetrant transport and may be altered by absorption and desorption of penetrants. Changes in the total polymer volume are largely governed by changes in the hole-free volume. Experimental technique known as mercury intrusion porosimetry characterizes material s porosity by applying various levels of pressure to a sample immersed in mercury [20]. The pressure required to intrude the mercury into sample s pores

28 14 X.J. Fan and S.W.R. Lee Water-proof coating film Water vapor molecules Water liquid molecules Fig. 1.9 Schematic diagram for a hydrophobic coating film, which can prevent liquid phase moisture, but not vapor phase moisture, from penetrating into the film is inversely proportional to the size of the pores. Mercury intrusion porosimetry is based on the capillary law governing liquid penetration into small pores. This law, in the case of a non-wetting liquid like mercury, is expressed by the Washburn s equation D = 4γ cos θ, (1.17) P where γ is the surface tension of mercury, θ is the contact angle between the mercury and the sample, P is the applied pressure, and D is the pore diameter, all in consistent units. The volume V of mercury penetrating the pores is measured directly as a function of the applied pressure. This P-V curve serves as a unique characterization of the pore structure. The Washburn equation assumes that all pores are cylindrical. Although in reality pores or free volumes are rarely cylindrical, this equation provides a practical representation of pore distributions, providing very useful results for most applications. As pressure increases during an analysis, pore size is calculated at each pressure point, and the corresponding volume of mercury required to fill in these pores is measured. These measurements taken over a range of pressures provide the pore volume versus pore size distribution for the sample material. Mercury porosimetry can determine the pore size distribution quite accurately. Comprehensive data provide extensive characterization of sample porosity and density. Available results include total pore volume, pore size distribution, and pore diameter. Mercury porosimetry is typically applied over a capillary diameter range from 50 nm to 360 μm.

29 1 Fundamental Characteristics in Polymeric Materials in Electronic Packaging 15 Several types of polymer materials for electronic packaging applications, such as die-attach films, solder mask, and underfills, were tested using the mercury intrusion technique [21]. Those materials were also examined for moisture weight gain test at 85 C/85% RH. The results showed that the relative weight gain ranged from 0.48 to 1.27% for those tested materials. However, no significant pore size down to 50 nm was observed for all materials under investigation. This implies that the free volumes, which are occupied by the absorbed moisture, are typically in nanometer range. The same materials were then subjected to a simulated reflow process. An example of microstructures of a die-attach film before and after the reflow is shown in Fig It can be seen that the significant voiding of the film due to the internal vapor pressure was developed after the reflow process [22]. Soles et al. [23] used positron annihilation lifetime spectroscopy (PALS) to quantify the polymer network topology, which produces a nanovoid volume fraction as a function of temperature. A strong correlation was observed between the absolute zero volume fraction and the ultimate moisture uptake. Although the correlation is clear, the absolute moisture zero hole volume fraction is not sufficient to predict the ultimate moisture uptake. An approximate estimate method to obtain the free volume fraction of polymers is proposed using moisture weight gain test. Consider a material with a saturated moisture concentration C sat at the given temperature and humidity. The initial free volume fraction f 0 can be found as f 0 = C sat ρ, (1.18) where ρ is the moisture density in the pores or free volumes. Since the density of the liquid water is 1.0 g/cm 3, the moisture density in free volume, i.e., the ρ value in equation (1.18), must be less than or equal to 1.0 g/cm 3 : f 0 C sat. (1.19) Here C sat has the unit of g/cm 3. In general, C sat depends on the relative humidity and temperature. The water liquid will fill in the free volume completely at 100% Fig Microstructures of a soft die-attach film (a) before and (b) after reflow with moisture absorption