1,882 777 24MB
Pages 671 Page size 396.75 x 639 pts Year 2007
ELECTRONICS MANUFACTURING WITH LEAD-FREE, HALOGEN-FREE, AND CONDUCTIVE-ADHESIVE MATERIALS
This page intentionally left blank
ELECTRONICS MANUFACTURING WITH LEAD-FREE, HALOGEN-FREE, AND CONDUCTIVE-ADHESIVE MATERIALS
John H. Lau Agilent Technologies, Inc.
C. P. Wong Georgia Institute of Technology
Ning-Cheng Lee Indium Corporation of America
S. W. Ricky Lee Hong Kong University of Science and Technology
Copyright © 2003 by The McGraw-Hill Companies, Inc. All rights reserved. Manufactured in the United States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a database or retrieval system, without the prior written permission of the publisher. 0-07-150087-1 The material in this eBook also appears in the print version of this title: 0-07-138624-6. All trademarks are trademarks of their respective owners. Rather than put a trademark symbol after every occurrence of a trademarked name, we use names in an editorial fashion only, and to the benefit of the trademark owner, with no intention of infringement of the trademark. Where such designations appear in this book, they have been printed with initial caps. McGraw-Hill eBooks are available at special quantity discounts to use as premiums and sales promotions, or for use in corporate training programs. For more information, please contact George Hoare, Special Sales, at [email protected] or (212) 904-4069. TERMS OF USE This is a copyrighted work and The McGraw-Hill Companies, Inc. (“McGraw-Hill”) and its licensors reserve all rights in and to the work. Use of this work is subject to these terms. Except as permitted under the Copyright Act of 1976 and the right to store and retrieve one copy of the work, you may not decompile, disassemble, reverse engineer, reproduce, modify, create derivative works based upon, transmit, distribute, disseminate, sell, publish or sublicense the work or any part of it without McGraw-Hill’s prior consent. You may use the work for your own noncommercial and personal use; any other use of the work is strictly prohibited. Your right to use the work may be terminated if you fail to comply with these terms. THE WORK IS PROVIDED “AS IS.” McGRAW-HILL AND ITS LICENSORS MAKE NO GUARANTEES OR WARRANTIES AS TO THE ACCURACY, ADEQUACY OR COMPLETENESS OF OR RESULTS TO BE OBTAINED FROM USING THE WORK, INCLUDING ANY INFORMATION THAT CAN BE ACCESSED THROUGH THE WORK VIA HYPERLINK OR OTHERWISE, AND EXPRESSLY DISCLAIM ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. McGraw-Hill and its licensors do not warrant or guarantee that the functions contained in the work will meet your requirements or that its operation will be uninterrupted or error free. Neither McGraw-Hill nor its licensors shall be liable to you or anyone else for any inaccuracy, error or omission, regardless of cause, in the work or for any damages resulting therefrom. McGraw-Hill has no responsibility for the content of any information accessed through the work. Under no circumstances shall McGraw-Hill and/or its licensors be liable for any indirect, incidental, special, punitive, consequential or similar damages that result from the use of or inability to use the work, even if any of them has been advised of the possibility of such damages. This limitation of liability shall apply to any claim or cause whatsoever whether such claim or cause arises in contract, tort or otherwise. DOI: 10.1036/0071386246
Professional
Want to learn more? We hope you enjoy this McGraw-Hill eBook! If you’d like more information about this book, its author, or related books and websites, please click here.
For more information about this title, click here
CONTENTS
Chapter 1. Introduction to Environmentally Benign Electronics Manufacturing 1.1
Trends in Industry
1.1.1 1.1.2 1.2
1.3
1.1
Automobile Industry Electronics Industry
1.1 1.2
Trends in Worldwide Environmentally Benign Manufacturing
1.2.1 1.2.2 1.2.3 1.2.4 1.2.5
Government Activity Industry Activity R&D Activity Education Activity Worldwide Efforts on Environmentally Benign Electronics Manufacturing
Trends in Environmentally Benign Electronics Manufacturing
1.3.1 1.3.2 1.3.3 1.3.4 1.3.5 1.3.6 1.3.7
IC Fabrication IC Packaging PCBs Lead-Free Solders Halogen-Free Flame Retardants Conductive Adhesives End-Of-Life Management
Acknowledgments References
Introduction UBM
2.2.1 2.2.2 2.3
2.4 2.5
Electroless Ni-P-Immersion Au UBM Al-NiV-Cu UBM Overview of Microball Wafer Bumping Microball Preparation Microball Management Microball Wafer Bumping
Sn-Ag-Cu Solder Ball Mounting on Wafers
2.4.1 2.4.2
WLCSP WLCSP with Stress-Relaxation Layer
Stencil Printing on Sn-Ag Solder on Wafers with Ni-Au UBM
2.5.1 2.5.2 2.5.3
1.4 1.4 1.5 1.5 1.6 1.6
1.9 1.9 1.9 1.10 1.11 1.12 1.13
2.1
2.1 2.1
Microball Wafer Bumping with Lead-Free Solders
2.3.1 2.3.2 2.3.3 2.3.4
1.3
1.14 1.14
Chapter 2. Chip (Wafer)-Level Interconnects with Lead-Free Solder Bumps 2.1 2.2
1.1
The Interface Between Electroless Ni and Solders Growth of the IMC and P-Rich Ni Layer Bump Shear Fracture Surface
v
2.1 2.6 2.6
2.6 2.6 2.9 2.12 2.12
2.12 2.15 2.20
2.20 2.22 2.24
vi
CONTENTS
2.6
Stencil Printing of Sn-Cu, Sn-Ag-Bi, and Sn-Ag-Cu Solders on Wafers with Ni-Au UBM
2.6.1 2.6.2 2.6.3 2.7
Stencil Printing of Sn-Cu, Sn-Ag-Bi, and Sn-Ag-Cu Solders on Wafers with Ti-Cu UBM
2.7.1 2.7.2 2.8
Interface of Reflowed Solder Bumps Interface of Annealed Solder Bumps Shear Strength of Solder Bumps Interface of Reflowed Solder Bumps Interface of Annealed Solder Bumps
Paste Printing of Solders on Wafers with Al-NiV-Cu UBM Acknowledgments References
Chapter 3. WLCSP with Lead-Free Solder Bumps on PCB/Substrate 3.1 3.2
Introduction Solder Joint Reliability of SnAgCu WLCSP with a Stress-Relaxation Layer
3.2.1 3.2.2 3.2.3 3.3
Solder Joint Reliability of SnAg and SnAgCu WLCSPs with TiCu and NiAu UBMs
3.3.1 3.3.2 3.4
Finite Element Results Thermal Cycling Results Effects of the Stress-Relaxation Layer on Capacitance Isothermal Fatigue Test Results Thermal Cycling Fatigue Test Results
Solder Joint Reliability of SnAg, SnAgCu, SnAgCuSb, and SnAgInCu WLCSPs with AlNiVCu UBM
3.4.1 3.4.2 3.4.3 3.4.4
Thermal Fatigue of SnAg, SnAgCu, SnAgCuSb and SnAgInCu WLCSPs on Ceramic Substrate Thermal Fatigue of SnAgCu WLCSP on PCB High-Temperature Storage of SnAgCu WLCSP on PCB Shear Strength of SnAgCu WLCSP on PCB
Acknowledgments References
Chapter 4. Chip (Wafer)-Level Interconnects with Solderless Bumps 4.1 4.2 4.3
Introduction Wafers for Electroless Ni-Au, Electroplated Au, and Electroplated Cu Bumps Electroless Ni-P-Immersion Au Bumps
4.3.1 4.3.2 4.4
Electroplated Au Bumps
4.4.1 4.4.2 4.5
Materials and Process Passivation Cracking Materials and Process Bump Specifications and Measurement Methods
Electroplated Cu Bumps
4.5.1 4.5.2
Materials and Process Special Considerations
2.27
2.27 2.29 2.30 2.31
2.31 2.31 2.34 2.34 2.35
3.1
3.1 3.1
3.1 3.2 3.4 3.5
3.5 3.8 3.15
3.15 3.15 3.15 3.17 3.20 3.20
4.1
4.1 4.1 4.1
4.2 4.2 4.6
4.6 4.6 4.8
4.8 4.8
CONTENTS
4.6
Electroplated Copper Wires
4.6.1 4.6.2 4.7
Wire-Bonding Microsprings
4.7.1 4.7.2 4.8
4.12
Materials and Process Equipment
4.12 4.14
Wire-Bonding Cu Stud Bumps
4.17
Materials and Process Shear Strength
Chapter 5. WLCSP with Solderless Bumps on PCB/Substrate Introduction Design, Materials, Process, and Reliability of WLCSPs with Au Bumps, Cu Bumps, and Ni-Au Bumps on PCB with ACF
5.2.1 5.2.2 5.2.3 5.2.4 5.2.5 5.2.6
4.23 4.24
5.1
5.1 5.1
5.1 5.1 5.4 5.9 5.10 5.10 5.11 5.12 5.12
5.13 5.14 5.14
5.16 5.19
Au-Stud-Bumped WLCSP with ACP/ACF on PCB
5.22
ACF/ACP with Nonconductive Fillers DSC Measurement Results DMA Measurement Results TMA Measurement Results TGA Measurement Results 85°C/85% RH Test and Results Thermal Cycling Test and Results
Au-Stud-Bumped WLCSP Diffused on Au-Plated PCB with NCA
5.8.1 5.8.2 5.9
4.21 4.23
Materials and Process Qualification Tests and Results
5.7.1 5.7.2 5.7.3 5.7.4 5.7.5 5.7.6 5.7.7 5.8
Materials and Process Flow Equipment for SBB Technology
Au-Stud-Bumped WLCSP with ICA on Flex
5.6.1 5.6.2 5.7
PCB ACF FCOB Assemblies with ACF Thermal Cycling Test of FCOB Assemblies with ACF SIR Test Results of ACF FCOB Assemblies Summary
Copper Wired WLCSP with Solders or Adhesives on Substrates Microspring WLCSP with Solders or Adhesives on PCB/Substrate Au-Stud-Bumped WLCSP with ICA on PCB
5.5.1 5.5.2 5.6
4.10
Wire-Bonding Au Stud Bumps
Acknowledgments References
5.3 5.4 5.5
4.9 4.10 4.11 4.11
4.9.1 4.9.2
5.1 5.2
4.9
Materials and Process Special Considerations
4.8.1 4.8.2 4.9
Structure Fabrication Materials and Process
vii
Materials and Process Reliability
Au-Stud-Bumped WLCSP Diffused on Au-Plated Flex with NCA
5.9.1 5.9.2
Materials and Process Reliability
5.10 Cur-Stud-Bumped WLCSP with Lead-Free Solders on PCB
5.10.1 Materials and Process 5.10.2 Reliability Acknowledgments References
5.22 5.23 5.23 5.23 5.26 5.26 5.27 5.29
5.32 5.35 5.37
5.37 5.38 5.42
5.42 5.43 5.45 5.45
viii
CONTENTS
Chapter 6. Environmentally Benign Molding Compounds for IC Packages 6.1 6.2
Introduction Environmentally Benign Molding Compounds for PQFP Packages
6.2.1 6.2.2 6.2.3 6.2.4 6.3
Flame Resistance Systems: Addition-Type Retardants Flame Resistance Systems: Novel Resin Systems Effects of Raised Reflow Temperature on Molding Compounds Halogen-Free Molding Compounds for Lead-Free Soldering
Environmentally Benign Molding Compounds for PBGA Packages
6.3.1 Halogen-Free Flame-Retardant Agents 6.3.2 PBGA Package Warpage Controlled by Tg Dispersion 6.3.3 PBGA Package Warpage Controlled by Stress-Absorbing Agents 6.4
Environmentally Benign Molding Compounds for MAP-PBGA Packages
6.4.1 6.4.2 6.4.3 6.4.4
Halogen-Free Flame-Retardant Resins Sample Preparation Effects of Tg , Shrinkage, and Viscosity on Package Coplanarity Moisture Sensitivity Tests
Acknowledgments References
Chapter 7. Environmentally Benign Die Attach Films for IC Packaging 7.1 7.2
Introduction Environmentally Benign Die Attach Films
7.2.1 7.2.2 7.3
Silver-Filled Film DF-335-7 for Leadframe PQFP Packages Insulating Film DF-400 for BT-Substrate PBGA Packages
Environmentally Benign In-Sn Die Attach Bonding Technique
7.3.1 7.3.2 7.3.3
In-Sn Phase Diagram Design and Process of In-Sn Solder Joints Characterization of In-Sn Solder Joints
Acknowledgments References
Chapter 8. Environmental Issues for Conventional PCBs 8.1 8.2
Introduction Influence of Electronic Products
8.2.1 8.2.2 8.2.3 8.2.4 8.2.5 8.3
Major Environmental Concerns Energy Issues Chemical Issues Disposal and Recycling Design for Environment
Environmental Research for the PCB Industry
8.3.1 8.3.2 8.3.3
Energy and Solvent Reduction Renewable Resins for PCB Reworkable Encapsulants for Disassembly
6.1
6.1 6.1
6.2 6.4 6.4 6.8 6.10
6.12 6.16 6.19 6.22
6.22 6.22 6.25 6.27 6.29 6.29
7.1
7.1 7.1
7.1 7.6 7.10
7.11 7.12 7.14 7.18 7.18
8.1
8.1 8.2
8.2 8.5 8.7 8.14 8.16 8.18
8.19 8.21 8.22
CONTENTS
8.4
International Driving Forces for Halogen-free Alternatives
8.4.1 8.4.2 8.4.3 8.4.4
Background and Challenge Driving Forces Material Availability Design Measures and Performance
References
Chapter 9. Halogenated and Halogen-Free Materials for Flame Retardation 9.1 9.2
Introduction Brominated Flame Retardants
9.2.1 9.2.2 9.2.3 9.3
Toxicological Aspects of Halogen-Free Flame Retardants
9.3.1 9.3.2 9.3.3 9.4
Production Aspects Classification Risk Assessment Fundamentals Denitrification Bioassay Procedures
Environmentally Conscious Flame-Retarding Plastics
9.4.1
Flame-Retardant Polycarbonate Resin
References
Chapter 10. Fabrication of Environmentally Friendly PCB 10.1 10.2
Introduction PCB DfE
10.2.1 10.2.2 10.2.3 10.2.4 10.3
Process Modeling Health Hazard Assessment Board Optimization Life Cycle Analysis (LCA)
Implementing Green PCB Manufacturing
10.3.1 Basic Processes 10.3.2 Process Modifications 10.3.3 Environmental Impact 10.4
Conformal Coating with Environmental Safety
10.4.1 10.4.2 10.4.3 10.4.4 10.4.5
Fundamentals Coating Selection Curing Methods Dispensing Methods Process Issues
References
Chapter 11. Global Status of Lead-Free Soldering 11.1 11.2 11.3 11.4 11.5 11.6
Introduction Initial Activities Recent Activities Impact of Japanese Activities U.S. Reaction What Are Lead-Free Interconnects?
ix
8.23
8.23 8.24 8.25 8.25 8.25
9.1
9.1 9.2
9.2 9.4 9.5 9.7
9.7 9.9 9.10 9.11
9.12 9.19
10.1
10.1 10.1
10.1 10.2 10.7 10.11 10.11
10.12 10.13 10.16 10.17
10.17 10.17 10.18 10.19 10.20 10.22
11.1
11.1 11.1 11.2 11.5 11.5 11.7
x
CONTENTS
11.7 11.8
Criteria for Lead-Free Solder Viable Lead-Free Alloys
11.8.1 11.8.2 11.8.3 11.8.4 11.8.5 11.8.6 11.8.7 11.8.8 11.9 11.10 11.11 11.12 11.13 11.14 11.15 11.16 11.17 11.18 11.19 11.20
Sn96.5/Ag3.5 Sn99.3/Cu0.7 SnAgCu SnAgCuX SnAgBiX SnSb SnZnX SnBi
Cost PCB Finishes Components Thermal Damage Other Concerns Consortium Activity Opinions of Consortia What Are the Selections of Pioneers? Possible Path Is Pb-Free Safe? Summary Information Resources
11.20.1 Legislation 11.20.2 Initiatives from Independent Corporations and Electronics Industry Organizations 11.20.3 Viable Alloys under Consideration References
Chapter 12. Development of Lead-Free Solder Alloys 12.1 12.2 12.3 12.4
Criteria Toxicity Cost and Availability Development of Lead-Free Alloys
12.4.1 Existing Alloys 12.4.2 Modification 12.5 12.6
Lead-Free Alloys Investigated Favorite Pb-Free Alloys
12.6.1 12.6.2 12.6.3 12.6.4 12.7 12.8
Japan Europe North America Comparison of Regional Preferences
Patent Issues Conclusion References
Chapter 13. Prevailing Lead-Free Alloys 13.1
Eutectic Sn-Ag
13.1.1 13.1.2 13.1.3 13.1.4
Physical Properties Mechanical Properties Wetting Properties Reliability
11.8 11.8
11.8 11.8 11.9 11.9 11.9 11.10 11.10 11.11 11.11 11.11 11.12 11.12 11.13 11.13 11.13 11.14 11.14 11.15 11.15 11.16
11.16 11.16 11.17 11.17 12.1
12.1 12.1 12.4 12.4
12.4 12.5 12.13 12.13
12.13 12.13 12.33 12.33 12.36 12.37 12.37 13.1
13.1
13.1 13.1 13.6 13.10
CONTENTS
13.2
Eutectic Sn-Cu
13.2.1 13.2.2 13.2.3 13.2.4 13.3
Physical Properties Mechanical Properties Wetting Properties Reliability
Sn-Ag-Bi and Sn-Ag-Bi-In
13.3.1 Physical and Mechanical Properties 13.3.2 Wetting Properties 13.3.3 Reliability 13.4
Sn-Ag-Cu and Sn-Ag-Cu-X
13.4.1 13.4.2 13.4.3 13.4.4 13.5
Sn-Zn and Sn-Zn-Bi
13.5.1 13.5.2 13.5.3 13.5.4 13.6
Physical Properties Mechanical Properties Wetting Properties Reliability Physical Properties Mechanical Properties Wetting Properties Reliability
Summary References
Chapter 14. Lead-Free Surface Finishes 14.1 14.2 14.3
Introduction Options for PCB Lead-Free Surface Finishes OSP
14.3.1 14.3.2 14.3.3 14.3.4 14.4
Benzotriazole Imidazoles Benzimidazoles Preflux
NiAu
14.4.1 Electrolytic Ni-Au 14.4.2 Electroless Ni/Immersion Au 14.4.3 Electroless Ni/Electroless (Autocatalytic) Au 14.5 14.6 14.7
Immersion Ag Immersion Bi Pd
14.7.1 Electrolytic Pd with or Without Immersion Au 14.7.2 Electroless (Autocatalytic) Pd with or Without Immersion Au 14.8 14.9
Electroless NiPd(Au Flash) NiPd(X)
14.9.1 Electrolytic NiPdCoAu Flash 14.9.2 Electroless NiPdNiAu Flash 14.10 Sn
14.10.1 Electrolytic Sn 14.10.2 Immersion Sn 14.11 Electrolytic NiSn 14.12 Sn-Bi
14.12.1 Immersion Sn-Bi Alloy 14.12.2 Electrolytic Sn-Bi Alloy 14.13 Sn-Cu (HASL)
xi
13.14
13.14 13.14 13.14 13.17 13.23
13.23 13.24 13.26 13.31
13.31 13.34 13.42 13.45 13.54
13.54 13.55 13.55 13.56 13.59 13.59
14.1
14.1 14.1 14.1
14.2 14.7 14.8 14.14 14.14
14.15 14.18 14.25 14.26 14.36 14.38
14.38 14.42 14.43 14.45
14.45 14.45 14.46
14.47 14.50 14.55 14.59
14.59 14.59 14.60
xii
CONTENTS
14.14 Electrolytic SnNi 14.15 Solid Solder Deposition (SSD)
14.15.1 14.15.2 14.15.3 14.15.4 14.15.5 14.15.6 14.15.7 14.16 14.17 14.18 14.19 14.20 14.21 14.22 14.23 14.24 14.25 14.26
HASL Optipad Sipad PPT Solder Cladding Solder Jetting Super Solder
Summary for PCB Surface Finishes Options of Component Surface Finishes NiAu (ENIG) Electrolytic Pd Electroless NiPd Electrolytic PdNi Sn Electrolytic Sn-Ag Electrolytic Sn-Bi Sn-Cu Summary of Component Surface Finishes References
Chapter 15. Implementation of Lead-Free Soldering 15.1
Compatibility of Lead-Free Solders with SMT Reflow Process
15.1.1 15.1.2 15.1.3 15.1.4 15.2 15.3 15.4 15.5 15.6 15.7 15.8 15.9
Experimental Design for Compatibility Evaluation Results of Compatibility Study Additional Factors to Be Considered Compatibility Assessment
Implementing Lead-Free Wave Soldering Effect of Reflow Profile on Lead-Free Soldering Flux Desired For Lead-Free Paste Soldering Flux Desired For Lead-Free Paste Handling Cleaning Performance of Lead-Free Solder Paste Flux Desired For Lead-Free Residue Cleaning Cleaning Chemistry/Process Desired for Lead-Free Residue Cleaning Selection of Lead-Free Solder Paste References
Chapter 16. Challenges for Lead-Free Soldering 16.1
Challenges for Surface Finishes
16.1.1 16.1.2 16.1.3 16.1.4 16.2
Black Pad Extraneous/Skip Plating Tin Whisker Surface Finish Cleaning Resistance
Challenges for Soldering
16.2.1 16.2.2 16.2.3 16.2.4
Intermetallic Compounds Dross Wave Solder Composition Lead Contamination
14.61 14.62
14.62 14.64 14.65 14.66 14.67 14.67 14.67 14.68 14.70 14.70 14.71 14.71 14.72 14.72 14.72 14.74 14.75 14.76 14.76
15.1
15.1
15.1 15.7 15.16 15.19 15.19 15.21 15.26 15.29 15.29 15.31 15.31 15.36 15.36
16.1
16.1
16.1 16.4 16.5 16.10 16.10
16.10 16.11 16.13 16.14
CONTENTS
16.2.5 16.2.6 16.2.7 16.2.8 16.3
Challenges for Reliability
16.3.1 16.3.2 16.3.3 16.3.4 16.3.5 16.3.6 16.4
Fillet Lifting Poor Wetting Voiding Rough Joint Appearance Tin Pest Intermetallic Compound Platelet Stiff Joint Thermal Damage Flux Residue Cleaning Conductive Anodic Filament
Unanswered Challenges References
Chapter 17. Introduction to Conductive Adhesives 17.1 17.2
Electronics Packaging: A Brief Overview Overview of Conductive Adhesive Technology
17.2.1 ACAs 17.2.2 ICAs 17.3 17.4
Proposed Approaches for Fundamental Understanding of Conductive Adhesive Technology and Developing Conductive Adhesives for Solder Replacement Research Objectives/Goals
17.4.1 Fundamental Study of the Chemical Nature and Behavior of Organic Lubricants on Silver Flakes 17.4.2 Investigation of the Conductivity Mechanism of Conductive Adhesives 17.4.3 Identification of the Main Mechanisms Underlying the Unstable Contact Resistance of ECAs on Non-Noble Metals and Approaches to Stabilization of Contact Resistance 17.4.4 Development of Conductive Adhesives with Satisfactory Conductivity, Stable Contact Resistance and Desirable Impact Strength 17.5
Outline of Research Acknowledgments References
Chapter 18. Conductivity Establishment of Conductive Adhesives 18.1 18.2
Introduction Experiments
18.2.1 18.2.2 18.2.3 18.2.4 18.2.5
Materials Transmission Electron Microscopy (TEM) Study of ECAs Conductivity Establishment During Cure Measurements of Cure Shrinkage Conductivity Development of Ag Particles and ECA Pastes with External Pressures 18.2.6 Conductivity Establishment of a Conductive Adhesive and Lubricant Behavior of the Ag Flake 18.2.7 Measurements of Modulus Change During Cure
xiii
16.20 16.25 16.26 16.28 16.29
16.29 16.30 16.31 16.33 16.34 16.35 16.36 16.39
17.1
17.1 17.4
17.4 17.8 17.15 17.16
17.16 17.17
17.17 17.18 17.19 17.19 17.19
18.1
18.1 18.2
18.2 18.2 18.2 18.2 18.3 18.4 18.4
xiv
CONTENTS
18.2.8 Cure Study of Conductive Adhesives 18.2.9 Measurements of Cross-Linking Density 18.3
Results and Discussion
18.3.1 Observation of Interparticle Contact Between Silver Flakes 18.3.2 Conductivity Establishment of Conductive Adhesives During Cure 18.3.3 Study of the Relationship Between Silver Flake Lubricant Layer and Conductivity in ECAs 18.3.4 Study of the Relationship Between Cure Shrinkage and Conductivity Establishment 18.4
Conclusions References
Chapter 19. Mechanisms Underlying the Unstable Contact Resistance of ECAs 19.1 19.2
Introduction Experiments
19.2.1 19.2.2 19.2.3 19.2.4 19.3
Materials Study of Bulk Resistance Shifts Study of Contact Resistance Shifts Study of Oxide Formation
Results and Discussion
19.3.1 Contact Resistance Shift Phenomenon 19.3.2 Investigation of Mechanisms Underlying the Unstable Contact Resistance Phenomenon 19.3.3 Observation of Metal Oxide Formation 19.4
Conclusions References
Chapter 20. Stabilization of Contact Resistance of Conductive Adhesives 20.1
Introduction
20.1.1 Factors Affecting Galvanic Corrosion 20.1.2 Additives to Prevent Galvanic Corrosion 20.2
Experiments
20.2.1 20.2.2 20.2.3 20.2.4 20.2.5 20.2.6 20.3
Materials Contact Resistance Test Devices Study of Curing Behaviors of ECAs Study of Dynamic Properties of ECAs Measurement of Moisture Absorption Measurement of Adhesion Strength
Results and Discussion
18.4 18.4 18.5
18.5 18.5 18.8 18.11 18.18 18.18
19.1
19.1 19.3
19.3 19.3 19.4 19.5 19.5
19.5 19.7 19.12 19.13 19.15
20.1
20.1
20.1 20.1 20.3
20.3 20.3 20.3 20.4 20.4 20.4 20.5
20.3.1 Effects of Electrolytes on Contact Resistance Shifts 20.5 20.3.2 Effects of Moisture Absorption on Contact Resistance Shifts 20.5 20.3.3 Stabilization of Contact Resistance Using Additives 20.7 20.4 20.5
Conclusions Summary References Index About the Author
20.13 20.13 20.14 I.1 A.1
PREFACE
Why did we want to write this book? Was it because of fear: fear of (potential) legislation, fear of trade barriers, fear of competition? Absolutely not! We wrote this book for ourselves, our children, and their children, so that we will all have a greener environment to live in! Books of this type can be huge and contain many different viewpoints, e.g., political, economic, cultural, and infrastructural. However, emphasis in this book is placed on fundamental principles, engineering data, and manufacturing technologies. There are four major subjects in this book: integrated circuit (IC) packaging (Chaps. 1 through 7), printed circuit board (PCB)/substrates (Chaps. 8 through 10), PCB/substrate assembly of IC packages (Chaps. 11 through 16), and novel conductive adhesive materials (Chaps. 17 through 20). Chapter 1 briefly discusses the trends in worldwide environmentally benign manufacturing, and especially electronics manufacturing. Chapter 2 presents chip (wafer)-level interconnects with lead-free solder bumps. Emphasis is placed on the under-bump metallurgies (UBMs) and wafer bumping with microball mounting and paste-printing methods. Chapter 3 examines the lead-free solder joint reliability of wafer-level chip-scale packages (WLCSPs) on organic and ceramic substrates. Chapter 4 discusses chip (wafer)-level interconnects with solderless bumps constructed from Ni-Au, Au, and Cu, copper wires, gold wires, gold studs, and copper studs. The design, materials, process, and reliability of WLCSPs with these solderless interconnects on PCB/substrate are presented in Chap. 5. Halogen-free molding compounds for plastic quad flat pack (PQFP), plastic ball grid array (PBGA), and mold array (MAP-PBGA) packages are briefly discussed in Chap. 6. Environmentally benign die attach films for PQFP and PBGA packages and lead-free die attach bonding techniques for IC packaging are examined in Chap. 7. The environmental issues regarding conventional PCBs/substrates are discussed in Chap. 8. The influence of electronic products and the relevant environmental research are reviewed and the international driving forces for alternative materials are highlighted. In Chap. 9, halogenated and halogen-free materials are assessed in detail. Some environmentally conscious flame retardants are introduced. The emerging technologies for fabricating environmentally friendly PCBs are described in Chap. 10. The emphasis is placed on design for environment, green PCB manufacturing, and environmental safety. Chapter 11 reviews the global status of lead-free soldering activity, including legislation, consortia programs, and regional preference on lead-free solder alternatives. Chapter 12 discusses the criteria for lead-free solder, the approaches taken for development of lead-free solders, and the varieties of alloys and properties developed by the industry. Chapter 13 compares in detail the physical, mechanical, and soldering properties and the reliability of the prevailing lead-free solder options. Chapter 14 goes over the lead-free surface finishes for both PCBs and component applications. Both manufacturing process and performance are discussed for each type of surface finish. Implementation of lead-free soldering is analyzed in Chap. 15, with more empha-
xv Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
xvi
PREFACE
sis on the requirement for reflow process. While lead-free soldering is inevitable, the challenges of executing it definitely have to be addressed first. These are discussed in detail in Chap. 16. Chapter 17 presents an overview of conductive adhesive technology and proposes approaches for a fundamental understanding. Chapter 18 examines the conductivity mechanisms of isotropic conductive adhesives. Emphasis is placed on the relationship between lubricant removal and conductivity in electrically conductive adhesives (ECAs) and the relationship between cure shrinkage and conductivity establishment. Chapter 19 discusses the mechanisms underlying the contact resistance shifts of ECAs. Contact resistance stabilization of ECAs is presented in Chap. 20, with emphasis on determination of the effects of electrolytes and moisture absorption on contact resistance shifts and on the stabilization of contact resistance using various additives. For whom is this book intended? Undoubtedly it will be of interest to three groups of specialists: (1) those who are active or intend to become active in research and development in electronics and photonics manufacturing with lead-free, halogen-free, and conductive adhesive materials; (2) those who have encountered practical lead-free, halogen-free, and conductive adhesive problems and wish to understand and learn more methods for solving such problems; and (3) those who have to choose a reliable, creative, high-performance, robust, and cost-effective packaging technique for their green products.This book also can be used as a text for college and graduate students who have the potential to become our future leaders, scientists, and engineers in the electronics and photonics industry. We hope this book will serve as a valuable reference to all those faced with the challenging problems created by the ever increasing interest in lead-free, halogenfree, and conductive adhesive materials. We also hope it will aid in stimulating further research and development on environmental, electrical, and thermal designs; materials; processes; manufacturing; electrical, thermal, and end-of-life management; testing; reliability; and more sound applications of lead-free, halogen-free, and conductive adhesive technologies in electronic and photonic products. Organizations that learn how to design lead-free, halogen-free, and conductive adhesive technologies in their interconnect systems have the potential to make major advances in the electronics and photonics industry and to gain great benefits in cost, performance, density, quality, size, weight, and market share. It is our hope that the information presented in this book may assist in removing roadblocks; avoiding unnecessary false starts; and accelerating design, materials, and process development of lead-free, halogen-free, and conductive adhesive technologies. It is an exciting time for these technologies! John H. Lau, PhD, PE, ASME Fellow, IEEE Fellow Palo Alto, CA C. P. Wong, PhD, NAE, IEEE Fellow, AIC Fellow Duluth, Georgia Ning-Cheng Lee, PhD New Hartford, NY S.-W. Ricky Lee Kowloon, Hong Kong
ACKNOWLEDGMENTS
Development and preparation of this book was facilitated by the efforts of a number of dedicated people at McGraw-Hill and North Market Street Graphics. We would like to thank them all, with special mention to Stephanie Landis of North Market Street Graphics, and Thomas Kowalczyk and Jessica Hornick of McGrawHill for their unswerving support and advocacy. Special thanks to Steve Chapman, executive editor of electronics and optical engineering, who made our dream of this book come true by effectively sponsoring the project and solving many problems that arose during the book’s preparation. It has been a great pleasure and fruitful experience to work with these people in transforming our messy manuscripts into a very attractive printed book. The material in this book has clearly been derived from many sources, including individuals, companies, and organizations, and we have attempted to acknowledge in the appropriate parts of the book the assistance that we have been given. It would be quite impossible for us to express our thanks to everyone concerned for their cooperation in producing this book, but we would like to extend due gratitude. Also, we want to thank several professional societies and publishers for permitting us to reproduce some of their illustrations and information in this book. For example, the American Society of Mechanical Engineers (ASME) conference proceedings (e.g., International Intersociety Electronic Packaging Conference) and transactions (e.g., Journal of Electronic Packaging), the Institute of Electrical and Electronic Engineers (IEEE) conference proceedings (e.g., Electronic Components and Technology Conference) and transactions (e.g., Advanced Packaging and Manufacturing Technology), the International Microelectronics and Packaging Society (IMAPS) conference proceedings (e.g., International Symposium on Microelectronics) and transactions (e.g., International Journal of Microcircuits and Electronic Packaging), the Surface Mount Technology Association (SMTA) conference proceedings (e.g., Surface Mount International Conference and Exposition) and journals (e.g., Journal of Surface Mount Technology), the IBM Journal of Research and Development, Electronic Packaging and Production, Advanced Packaging, Circuits Assembly, Surface Mount Technology, Connection Technology, Solid State Technology, Circuit World, Microelectronics International, and Soldering and Surface Mount Technology. John Lau would like to thank his former employers, Hewlett-Packard Company and Express Packaging Systems, for providing him excellent working environments that have nurtured him as a human being, provided job satisfaction, and enhanced his professional reputation. He also would like to thank Steve Erasmus and Ted Lancaster for their trust, respect, and support of his work at Agilent Technologies. Furthermore, he would like to thank his eminent colleagues at Hewlett-Packard Company, Express Packaging Systems, Agilent Technologies, and throughout the electronics and optoelectronics industry for their useful help, strong support, and stimulating discussions.Working and socializing with them have been a privilege and an adventure. He learned a lot about life and packaging and interconnection technologies from them.
xvii Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
xviii
ACKNOWLEDGMENTS
John Lau also thanks his daughter, Judy, and his wife, Teresa, for their love, consideration, and patience in allowing him to work many weekends on this book. Their belief that he is making a small contribution to the electronics and optoelectronics industry was a strong motivation for him. Knowing that Judy is going to Princeton for her graduate studies in physics this fall, and that Teresa and he are in good health, he wants to thank God for His generous blessings. C. P. Wong wants to thank his former colleagues at AT&T Bell Labs and Georgia Tech—in particular, his current colleagues and good friend Rao Tummala at the Packaging Research Center at the Georgia Institute of Technology (GIT)—for all their support. Special thanks to his former PhD student, D. Lu, for his outstanding work on electrical conductive adhesives. C.P. also thanks his wife, Lorraine, and his children, Michelle and David, for their support, love, and understanding all these years at Bell and GIT. Ning-Cheng Lee would like to express gratitude to Indium Corporation of America for providing a highly inspiring work environment. He also thanks his colleagues at Indium for their encouragement and support of his pursuit of solutions for the never ending challenges of this rapidly evolving world. Ning-Cheng Lee wants to thank his mother, Shu-shuen Chang, for her encouragement and blessing, and his wife, Shen-chwen, for her full support and patience— particularly for her tolerance toward his irregular work hours. He would also like to thank his sister Yu-Hsuan for her selfless and dedicated effort in taking care of their aged mother so that he can focus on outside challenges. Ricky Lee wishes to express his gratitude to his colleagues at Hong Kong University of Science and Technology and his industrial partners in the Asia-Pacific region. Without their efforts to establish a pro-electronic packaging environment, he probably would not have begun his endeavor in this discipline. Special thanks are also due to the Industry Department and Research Grant Council (RGC) of Hong Kong for their financial support to part of his research activities in electronic packaging. Ricky Lee is also indebted to his family. During a certain period while working on this book, he averaged only four hours of sleep a night. Without the spiritual support from the family, he could never have struggled through that exhausting time!
CHAPTER 1
INTRODUCTION TO ENVIRONMENTALLY BENIGN ELECTRONICS MANUFACTURING 1.1
TRENDS IN INDUSTRY
Many thousands of industries have arisen during the four great Western transformations—the Renaissance, the Reformation, the Industrial Revolution, and the Computer Age. The automobile and electronics industries are the largest and most important of the current industries.They will be briefly discussed in the following text.
1.1.1
AUTOMOBILE INDUSTRY
Until 1996, the automobile industry was the largest industry in the world. Its early shift from workshop manufacturing to mass production made cars affordable for billions of people. However, emissions of unburned hydrocarbons, nitrogen oxides, and carbon monoxide spread over urban areas and into the countryside, which was increasingly buried under asphalt and concrete roads. And the synthesis of plastics (used in automobiles) grew into a large, highly energy-intensive industry generating a variety of toxic pollutants previously never present in the biosphere and introducing huge numbers of nondecaying wastes into the environment. Post-1945 developments amplified these trends. New environmental risks were introduced as the developed world entered the period of its most impressive economic growth, terminated only by the 1973 to 1974 quintupling of oil prices. In just 25 years, the consumption of primary commercial energy nearly tripled, electricity generation grew about 8-fold, car ownership increased 6-fold, and production of most kinds of synthetic materials grew more than 10-fold. In the summer of 1970, the Massachusetts Institute of Technology first attempted a systematic evaluation of global environmental issues. Their summary of the Study of Critical Environmental Problems indicated the following relative importance as perceived at that time: (1) emissions of carbon dioxide from fossil fuel combustion; (2) particulate matter in the atmosphere, cirrus clouds from jet aircraft, the effects of supersonic planes on stratospheric chemistry, the thermal pollution of waters, and the impact of pesticides; and (3) mercury and other toxic heavy metals, oil on the ocean, and the nutrient enrichment of coastal waters. Just a month later, U.S. president Richard Nixon sent Congress the first report of the President’s Council on Environmental Quality. Soon afterward, the Environmental Protection Agency (EPA) was born and the environment entered big-time politics. It should be emphasized that this was the first time in history that a nation had taken comprehensive stock of the quality of its surroundings. One of the EPA’s 1.1 Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
1.2
CHAPTER ONE
greatest achievements is banning lead additives in gasoline, thus reducing the concentration of lead in the air by 94 percent from 1980 to 1999.
1.1.2
ELECTRONICS INDUSTRY
Since 1996, the electronics industry has been the largest industry in the world (more than $1 trillion).1–116 It is the most dynamic, fascinating, and important area of manufacturing.There are many categories of electronic products, such as consumer, computer, and communication items. Today, however, computers and their peripheral products account for the greatest percentage of the total revenue for electronic products. In 1992, 11.5 million personal computers (PCs) were shipped in the U.S. According to the data from the National Safety Council (1999), the number is projected to be 55.8 million in 2005, as shown in Fig. 1.1. In the past few years the electronics industry has been facing an impending change in light of upcoming halogen-free and lead-free technology legislations. This is because the electronics industry has relied on halogenated flame retardants and tin-lead solders for its products. In 1998, the European Commission introduced two draft proposals called the Waste Electrical and Electronic Equipment (WEEE) and Reduction of Hazardous Substances (ROHS) directives. The primary objective of these complementary proposals is to minimize the risks and impacts that the production, use, treatment, and disposal of waste electrical and electronic equipment have on human health and the environment. Additionally, the directives are intended to prevent uncontrolled disposal of electrical and electronic equipment and to foster the development of reuse and recycling methods in order to reduce the amount of waste for disposal. In short, they aim for “green” products!
FIGURE 1.1 In 1992, the number of PCs shipped in the U.S. was 11.5 million. According to the National Safety Council, the projected number for 2005 is 55.8 million.
ENVIRONMENTALLY BENIGN ELECTRONICS MANUFACTURING
1.3
It is interesting to point out that green products sell! For example, Matsushita’s market share of its lead-free MiniDisc player jumped from 4.6 to 15 percent in 6 months (1999) in Japan. Toshiba’s bromine-free printed circuit boards (PCBs) help the company to sell its Libretto and Dynabook notebook computers in Europe and has earned Toshiba some romantic names such as Blue Angel (in Germany), White Swan (in Finland), and TCOGY (in Sweden). The European Commission, which revised the original draft in 2000, included January 2, 2008 as the implementation date for the lead ban. In further revisions, the European Parliament and the Council of Ministers proposed that the ban on lead take effect on January 1, 2006 and January 1, 2007, respectively. Also, WEEE and ROHS addressed several concerns about the use of halogened flame retardants, primarily (1) formation of dioxins and furans during incineration or recycling, and (2) persistence and bioaccumulation. Prior to the proposed ban on halogen, lead, and other toxic materials, there are other projected milestones of the WEEE and ROHS directives; for instance, producers are expected to establish systems for recovering electronic waste by the end of 2003. Japan has begun its version of take-back legislation effective in 2001 for a variety of its domestic products. The Electric Household Appliance Recycling Law passed the obligation for collection and recycling of waste appliances to the producers of those appliances. It should be emphasized that worldwide interest in halogen-free flame retardants and lead-free solders continues to grow, if not for environmental or regulatory reasons, then because of market differentiation. Many Japanese manufacturers are ahead of the proposed regulated ban on halogen and lead. Also, most Japanese system manufacturers want their products to be labeled green for market share opportunities, so they drive subassembly, component, and PCB manufacturers to make the change to halogen-free and lead-free materials prior to impending regulations becoming effective.
1.2 TRENDS IN WORLDWIDE ENVIRONMENTALLY BENIGN MANUFACTURING Worldwide trends in environmentally benign manufacturing (EBM), especially in Europe, Japan, and the U.S., have been studied by Murphy62 in four categories: government, industry, research and development (R&D), and education. The government activities include take-back legislation, landfill bans, material bans, life cycle assessment (LCA) tool and database development, recycling infrastructure, economic incentives, regulation by medium, cooperative/joint efforts with industry, and financial and legal liability. The industrial activities include International Standards Organization (ISO) 14000 certification, water conservation, energy conservation/CO2 emissions, decreased releases to air and water, decreased solid waste/postindustrial recycling, postconsumer recycling, material and energy inventories, alternative material development, supply chain involvement, EBM as a business strategy, and life cycle activities. The R&D activities include relevant basic research (>5 years out) and applied R&D (5 years out) Polymers
Automotive/transportation
**
*
***
Systems
**
*
*** **
Applied R&D ( Wb) (otherwise)
(10.11)
Orientation 2:
Nb = Int(Lp/Lb) Int(Wp/Wb) + Int(Wp/Lb) Nb = Int(Lp/Lb) Int(Wp/Wb)
(if GW > Wb) (otherwise)
(10.12)
We must calculate the number of boards per panel using these formulas and choose the orientation that gives the maximum number of boards per panel. In this way more of the material will go into the product and less into the manufacturing waste. Total Number of Signal Layers. The number of signal layers can be estimated given the routable board area Ar and the number of “reference components” Nref using the density approach.6 The number of reference components is equivalent to the number of components when all components are weighted with reference to a
10.8
CHAPTER TEN
FIGURE 10.4 Layout of two panels: (a) orientation 1, (b) orientation 2.
14-pin DIP component as one unit. The categorical functional dependence is given in Table 10.4. 10.2.3.3 Optimization. Once we derive the process models and the preceding relationships, we can follow a procedure like the one depicted in Fig. 10.5 to optimize the board design parameters. The results of such an optimization are shown in Fig. 10.6. The plot shows the weighted mass of the waste streams on a per-board basis as a function of the board dimensions. From this plot, we can choose the dimensions of the board corresponding to the minimum waste. We also can observe the effect of relaxing a constraint or imposing an additional constraint on the optimization problem. These constraints can easily be imposed or removed during waste stream calculation from process models. Several useful conclusions can be drawn from these plots: ●
●
●
The minimum waste does not necessarily occur for the smallest or thinnest (i.e., fewest layers) board size. A large variation (more than 100 percent in the plots shown) in the amount of waste per board occurs as the dimensions and the number of layers of the board are varied. Thus the scope for waste minimization through design optimization is tremendous. The choice of panels also makes a difference in waste generation. Thus, whenever a variety of panels is available, we must calculate which to choose.
TABLE 10.4 Signal Layer Estimation Ar /Nref (in2/14-pin DIP)
Ns-layers
Above 1.0
2
0.8–1.0
2
0.6–0.8
4
0.42–0.6
6
0.35–0.42
8
0.2–0.35
10
0.0–0.2
10+
10.9 FIGURE 10.5 Board design optimization procedure.
10.10 FIGURE 10.6 Waste mass as a function of board dimensions for two different panel sizes: (a) 18 × 24 in, (b) 21 × 27 in.
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.2.4
10.11
LIFE CYCLE ASSESSMENT (LCA)
A bare board is only one component of an overall PCB assembly. Thus, it is important to perform a similar optimization analysis on every component of the assembly. Also, design variation in one component sometimes necessitates design variation in another. For example, making the board smaller may allow the use of flip chip ball grid array components, which may not be as environmentally benign as quad flat packages because of the potential worker exposure hazard in the solder-bumping process. Therefore, the optimization procedure eventually can be extended to incorporate selection of component packaging. Electronic packaging presents discrete choices of package types. We can formulate the process models for each package type as a function of a functional parameter, such as the pin count. For semiconductor or die manufacturing, environmental issues are largely a function of waste mitigation in the manufacturing process; currently, limited opportunity is available to effect environmental decisions through design changes. However, a change in technology brings about major changes in process steps, chemicals used, and process mechanics and a resulting significant change in process waste. Therefore, process models for semiconductor manufacturing must predict the waste per die as a function of the yield (which is a function of die size and wafer size) and the technology used. So far, we have examined mainly design optimization of a product for minimum process waste. One aspect neglected in this analysis is the amount of material going into the product itself and its eventual fate. Although manufacturing wastes represent the dominant life cycle impact for PCBs, we cannot neglect the environmental impacts of use and disposition phases of the life cycle for other components of a circuit board or computer. LCA examines a product from the time its raw material is mined to the time the product is disposed back into the environment. It looks at material and energy flows in mining, material refinement, manufacturing, use, consumption, and disposal (which includes recycling, reuse, remanufacturing, incineration, deposit in a landfill, etc.) associated with a particular product. While LCA conceptually is straightforward, to implement such analyses requires a great deal of data that often are unavailable or of low quality. LCA is extremely data intensive, and while results can be found for some comparisons, often the data are so poor that little can be learned from them. Furthermore, few have agreed on how the multiple health and environmental impacts of a product’s life cycle can be compared to those of another product, making comparisons between products difficult. LCA is ultimately an attempt to draw a quantitative connection between the existence of a product (its manufacture, use, and disposition) and environmental impact. DfE is an attempt to incorporate this information (i.e., connections, analyses) to minimize environmental impacts of design decisions. Just as more traditional engineering models help designers predict the performance of their designs in terms of speed, weight, energy consumption, and other more standard measures of performance, LCA is a model that predicts for designers the environmental performance of their designs.
10.3
IMPLEMENTING GREEN PCB MANUFACTURING
The U.S. PCB industry has significantly improved its environmental performance over the past 20 years. This industry, which has the potential to be a major polluter,
10.12
CHAPTER TEN
has consciously altered its processes and practices to minimize its toxic output. While PCB manufacturers once viewed the effects of environmental regulations as a threat to their long-term growth, they have reversed this position. One of the best examples of this remarkable turnaround is how they actively became partners with regulatory agency programs such as the U.S. Environmental Protection Agency (EPA) Design for the Environment program. The industry continues to work with these agencies to assess the performance, cost, and environmental impact of alternatives to traditional PCB manufacturing methods.7 Despite the great advances within the PCB industry, environmentally conscious designers can further reduce the impact of the products they design by being aware of the consequences of design on processes and material. The concept of connecting design decisions to environmental consequences further down the supply chain is referred to as DfE. It means, for example, that requiring a solder surface on finished PCBs to be shipped to an assembler has more environmental consequences than allowing the PCB manufacturer to use an alternative finish such as Sn. The proposed concept and the procedure for making such circuit boards are further reviewed in the subsequent sections.
10.3.1
BASIC PROCESSES
There are several types of PCBs, depending on use, operational environment, and cost constraints. The simplest type of board is single-sided, with no holes for connections to other layers; these typically are made out of inexpensive laminate material and used for consumer products where there is little electronic sophistication and cost is the driver. A copper-clad dielectric is coated with a resist material, which is patterned to protect the areas where circuitry will be formed and the remaining copper etched away. This is referred to as the print and etch process. The chief waste products are etchant, consumed resist, and scrap or excess board material. The dielectric material can be either a rigid plastic (typically paper phenolic or glass/epoxy) or flexible plastic (e.g., certain nylons or polyesters). The second type of board is double sided, with plated through-holes. This board also is in common use. Because of the need to plate the dielectric on the walls of the holes connecting both sides, either electroless copper or, more recently, direct plating chemistries are used to make the surface conductive. Once the surface is conductive, the walls are typically plated using standard electroplating copper baths. This means that the surface of the board also is plated either as a sheet of copper (panel plating) or as resist-defined circuit areas (pattern plating). In either case, the copper between circuit lines must be removed by etching. In pattern plating, of course, the background copper is much thinner than the circuit lines. The dielectric again can be rigid or flexible. The most sophisticated boards, requiring complicated circuit routing to accommodate interconnecting integrated circuit chips, are usually multilayer boards. These boards traditionally have been manufactured by building innerlayers using the print and etch process. These innerlayers then are stacked in register and laminated together with partially polymerized dielectric between the layers along with outer layers of plain copper sheet. The outer layer is then typically drilled with throughholes to connect the various layers. The outer layers are either panel or pattern plated after the hole walls are coated using electroless or direct metallization. Again the dielectric can be rigid, flexible, or a combination of the two. The various processes used to make these types of boards have undergone changes during the past few years, the most significant of which are listed in the next
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.13
section. All these changes have had a significant effect in reducing the environmental impact of this industry.
10.3.2
PROCESS MODIFICATIONS
10.3.2.1 Switch from Chromic-Based Etchants. Beginning in the late 1970s and continuing through the early 1990s, alternatives were developed to replace chromicbased etchants. This happened because chromic-based etchants were not easy to regenerate, etched at a slow rate, and had a low limit of dissolved copper.8 They were also being regulated by environmental and health and safety agencies due to the hexavalent species, which was considered carcinogenic. The copper chloride and ammoniacal etchants, which have now replaced chromic etchants in the PCB industry, overcame all these disadvantages and were cheaper. The changes also significantly reduced the volume of etchant used by the industry and, therefore, its generation of spent etchant. Unlike chromic-based etchants, spent ammoniacal and cupric chloride etchants can be regenerated, reclaimed, or reused in other manufacturing operations.9 10.3.2.2 Elimination of Chlorinated Solvents. With the invention of solventdevelopable dry film photoresists in the late 1960s, the PCB industry used significant amounts of trichloroethane and methylene chloride to develop and strip them. In the late 1970s and early 1980s, aqueous and semiaqueous processable resists were developed, using either bicarbonate/hydroxide or butyl carbitol/cellosolve as developers or strippers. Trichloroethane continued to be used as a cleaning solvent until 1990, when it was found to contribute to stratospheric ozone depletion. During the ensuing decade, it was virtually eliminated from use when the EPA adopted a phaseout program for all ozone-depleting substances in Title VI of the Clean Air Act amendments.10 Other cleaning solvents categorized as ozone-depleting substances, for example HCFC-141(b), also were eliminated when the industry switched to alternatives, such as citrus-based terpenes and aqueous-based cleaners. 10.3.2.3 Improved Process Control. Some of the industry’s most successful environmental improvements include the widespread adoption of simple housekeeping measures that minimize waste generation and improve process control. Examples of such improvements include taking steps to reduce chemical loss, increase process bath life, and recover materials that otherwise would be disposed.11 These steps often included better bath control through improved use of sensors and control equipment. In addition, total quality management (TQM) systems can produce improved environmental as well as economic results. Adoption of a total quality management program can reduce facility scrap rates and result in process changes that reduce chemical losses or conserve process baths. 10.3.2.4 Reduced Use of Sn-Pb as Etch Resist. For many years, electroplated Sn-Pb was the most common PCB metal etch resist used in pattern plating. Its widespread use also is due to the requirement of many upstream customers to be able to use reflowed Sn or Pb as the PCB surface finish of choice. The industry began to switch to Pb-free etch metal resists due to technical constraints associated with Sn-Pb etch resists. In addition, the industry saw widespread use of hot-air solder leveling (HASL) as the predominant PCB surface finish prior to adding components. Typically, any Sn-Pb plating had to be stripped off prior to the HASL process,
10.14
CHAPTER TEN
thereby generating a Pb-containing hazardous waste. Facilities that switched to HASL began to look for an etch resist that, when stripped, would not be an environmental hazard. Sn was the most logical alternative. As an etch resist, Sn performs just as well as Sn-Pb; however, Sn is not considered a hazardous waste and does not represent the worker safety issues associated with Pb.12 10.3.2.5 Increased Use of an Alternative Metal-Bonding Surface Finish. To be effective, a PCB surface finish must prevent copper oxidization, facilitate solderability, and prevent defects during assembly. For many years, the reflowed Sn-Pb surface finish, which applies Sn-Pb solder to all exposed PCB traces, was the predominant PCB surface finish. Currently, HASL, which applies Sn-Pb solder only to PCB through holes and pads, has significantly reduced the industry’s use of Pb. Despite its advantages, HASL still poses several drawbacks; for instance, unless its waste solder dross is recycled, it must be managed as a hazardous waste. HASL also results in domed solder surfaces, whereas new and more complex packaging designs require flat surfaces.13 Among its other problems, HASL does not effectively cover the electroless nickel/immersion gold process, which has grown in use due to the ease of bonding wire semiconductors to it. Also, it has difficulty in adapting to the increased miniaturization of component attachment points. The Institute for Printed Circuits (IPC) is currently assessing Pb-free alternatives to HASL through the EPA DfE PCB Project. The project will assess the economic, environmental, and performance characteristics of the following HASL alternatives; organic solder protectorates, immersion Sn and Ag, electroless nickel/immersion gold, immersion palladium, and electroless nickel/immersion palladium. 10.3.2.6 Improved End-of-Pipe Pollution Control Practices. In the 1970s, conventional metal precipitation systems were the most common type of wastewater treatment utilized by the industry. Although precipitation remains very common, some facilities are supplementing or replacing such systems with ion exchange and electrowinnowing technologies. For most facilities, copper, Pb, and nickel, all of which are amenable to ion exchange and electrowinnowing, are the only metal ions present in significant concentrations. Furthermore, these techniques result in salable forms of metals, which can produce economic revenue for the company as well as a “cleaner” waste treatment sludge that may not be subject to Resource Conservation and Recovery Act (RCRA) hazardous waste regulations. The use of nonchelated process chemistries also has reduced the generation of wastewater treatment sludge. Reduced sludge generation means cost savings, since the avoided cost of managing this sludge, which the EPA considers a listed hazardous waste (e.g., F006) in most cases, can be significant. 10.3.2.7 Increased Reuse and Recycling of Manufacturing By-Products. The circuit formation processes noted already are subtractive and, in many cases, the most reliable and cost-effective ways to manufacture PCBs. The subtractive method generates large amounts of copper-bearing waste streams. Approximately 60 percent of the copper is removed in the typical etching process, resulting in a significant amount of copper leaving the facility as waste. In general, the industry recycles a majority of its manufacturing by-products (e.g., wastewater treatment sludge, etchant, off-specification boards, frames, and solder dross). Recycling extracts copper for reuse, reducing the need for virgin copper ore to be mined and reducing potential groundwater contamination (which could occur if copper-containing waste or incinerator ash is left in a landfill). PCB manufacturers can use on-site recycling methods, such as electrowinnowing, to remove metallic ions
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.15
from spent process solutions, ion exchange regenerant, and concentrated rinse waters; or they can send their manufacturing by-products off site to facilities where the valuable constituents are reused or reclaimed. Currently, the following materials are not subject to the RCRA hazardous waste restrictions and rules when they are reclaimed: scrap boards, scrap trim, router dust, solder baths and dumps, and solder dross. In addition, some states have ruled that spent etchant, when shipped to specific facilities that use the etchant as a direct feed stock in manufacturing operations, is not subject to RCRA. 10.3.2.8 Increased Use of Direct Metallization. As noted, the electroless copper process has been used to make drilled through-holes conductive. Electroless copper uses large quantities of water, formaldehyde, and chelators, such as ethylene diaminetetraacetic acid (EDTA). These chelators also chelate metal waste streams, complicating their treatment. The EPA and the IPC, under a DfE project, assessed a number of direct metallization alternatives—carbon, graphite, palladium, and conductive polymers—and found that all of them cost less, perform as well, and have less environmental impact (no formaldehyde or EDTA and less water use) than electroless copper. In addition, facilities that switched to alternatives have found that these alternatives often are less hazardous to use; increase production flow; decrease maintenance requirements; and reduce cycle time, operating costs, and water usage, increasing the facilities’ bottom lines. 10.3.2.9 Increased Water Reuse and Recycling. The PCB industry is dependent on the use of large quantities of high-quality water, which is used primarily to rinse circuit boards between process steps. PCB facilities now increasingly employ water reuse and recycling technologies to extend process bath life and decrease their reliance on municipal water, which may be costly and, in some cases, of poor quality. Facilities located in areas where water supplies are scarce may face flow restrictions from municipal water authorities. The installation of on-site water recycling systems and the use of simple water use reduction methods (e.g., flow restrictors, conductivity controls, flow meters, counterflow rinse tanks, and spray rinse systems) have resulted in large water use reductions for a number of PCB facilities. Fortunately, the adoption of water conservation practices often results in economic gain. 10.3.2.10 Future Opportunities for Pollution Prevention. There are additional opportunities for improved environmental impact. These might include: ● ●
● ●
● ● ● ●
One hundred percent beneficial reuse and recycling of all hazardous by-products Use of raw materials from sustainable sources (e.g., copper foil from recycled copper, laminate from bio-based or recycled plastic sources) Zero-water-discharge manufacturing processes Systematic management of environmental health safety (EHS) compliance and performance [ISO 1400 I and Environmental Management Systems (EMAS)] Development of technically acceptable Pb-free solders Standardization and implementation of design for reuse and disassembly practices Development and implementation of energy-efficient manufacturing operations Integration of environmental cost and activity-based cost accounting tools into traditional accounting methods
10.16
10.3.3
CHAPTER TEN
ENVIRONMENTAL IMPACT
10.3.3.1 Current Technology Trends. Since the industry is being driven to produce lighter, denser, cheaper circuits, designers must take advantage of these technologies. Many of the newer processes utilize less material, produce less waste, and are more energy efficient. For example, the two major approaches to microvias utilize either laser to make vias without drilling or photodielectric materials. While both these approaches produce vias that connect only layer to layer where physically needed, this immediately increases the circuit density and reduces material usage. Photodielectrics can also produce circuit “channels” as well as vias, which in combination with additive metallization would further significantly reduce waste. In addition to being used to make vias, lasers also are being utilized more seriously to create pattern resists used in making the circuit traces. Laser direct imaging allows the circuit design to go directly from the digital output of the design to the board itself without utilizing a phototool. This eliminates the phototool and its waste streams. In the case of a photodielectric, it may even be possible to image the vias and channels directly into the dielectric. 10.3.3.2 Electrical Design Effects. As can be seen from the improvements at both the chemical usage and the technology levels, it is possible to utilize materials and processes to significantly reduce the environmental impact of a circuit board. However, this can happen only if the PCB manufacturer is allowed to pick the best process and materials and not be hampered by outmoded specifications that might call for some specific chemical usage. For example, if electroless copper in the vias or holes is specified, then the manufacturer cannot utilize direct metallization processes to do the same process. Thinner copper and utilization of polymer thick film conductors for via metallization, shielding, and even conductor metallization can save much waste, since either less copper is used or, in the case of polymer thick films, the process is strictly additive. In addition, newer technology should be implemented where possible. For example, microvia technologies can be used to significantly reduce layer count or board size. This freedom to trade off layer count for board size can be significant, especially if the original board dimensions do not fit well into the standard panel sizes used in the industry. Unused panel areas that end up as edge trim scrap typically must go through all the same process steps as the final board. This means that all that processing and material is wasted and accounts for a significant portion of the scrap. Using mixed flex-rigid boards, thinner flexible materials can be used in place of rigid materials to not only carry circuitry but also function as an interboard connector. The key to utilizing the right materials and processes is to work closely with the PCB manufacturers. Decisions can be arrived at that both meet technical specifications and have the least environmental impact. Many people currently are working on methods to incorporate DfE in design algorithms, but direct discussions with suppliers will often serve the same function. Many PCB fabricators also now have inhouse design capabilities and can suggest alternatives to a given design. The PCB industry has made great strides in reducing its environmental impact by stepping up its pollution prevention efforts. Additional improvements may be made by optimizing the utilized processes and materials. Still other ongoing efforts may reduce the environmental burden even further. While the manufacturers themselves can implement much of this, cooperation along the electronic supply chain is critical. Those asking for specific materials and design considerations must be aware of the consequences of those design decisions. Despite the number of efforts to implement
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.17
DfE mechanisms to make this process easier, it is just as important to work directly with suppliers to understand the constraints that design decisions impose and the opportunities for improvement should other choices be made.14
10.4 10.4.1
CONFORMAL COATING WITH ENVIRONMENTAL SAFETY FUNDAMENTALS
A conformal coating is a thin layer of insulating material applied to the surface of a PCB to protect sensitive components from thermal shock, moisture, humidity, corrosion, dust, dirt, and other damaging elements. When properly applied, these coatings provide a high degree of insulative protection and are usually resistant to many types of solvents and harsh environments. Coatings also provide excellent dielectric resistance. Demanding applications where conformal coatings are critical include automotive products, consumer electronics and appliances, industrial controls, military/aerospace systems, and medical devices.15 In the past, because of the cost of conformal coating materials and application processes, only the most expensive boards or those with special reliability requirements were coated. Recent advances in application technology and process ability have improved the economics of conformal coating use. Additionally, as circuit size diminishes and components become more delicate, protective barrier coatings are growing in importance. While many available conformal coatings are still solvent based, the market for solvent-free coatings in North America and other parts of the world is growing rapidly. In 1970, the U.S. government passed the Clean Air Act, which gave the EPA the authority to set national air quality standards to protect against common pollutants, including materials that release volatile organic compounds and ozone-depleting chemicals. EPA and Occupational Health and Safety Administration standards, stringent state and local regulations, and emerging environmental awareness have combined to encourage coatings formulators and electronics manufacturers to use solvent-free and low–ozone-depleting chemical/volatile organic compound coatings wherever possible in their manufacturing operations. Although solvent-free coatings are more expensive on a per-unit basis than solvent-based materials, much less volume is used. Because solvent-free coatings are 100 percent solids, they do not evaporate as part of the curing process. Solvent-based materials are typically 60 to 70 percent solvents, all of which are wasted during evaporation.
10.4.2
COATING SELECTION
Conformal coatings are generally classified according to the molecular structure of their polymer backbone. There are five traditional conformal coating chemistries: acrylic, epoxy, urethane and parylene (commonly grouped together as organics), and silicone (an inorganic). All except parylene were solvent based until a decade ago, when increased environmental concerns and resulting government regulations dictated the reformulation of conformal coatings to solvent-free, low–ozone-depleting chemical/volatile organic compound materials and processes. Many environmentally acceptable coatings are hybrid formulations that combine two or more coating chemistries (e.g., urethane acrylate and acrylic functional silicones) to improve performance properties, wetting, adhesion, and cure requirements.
10.18
CHAPTER TEN
Solvent-free organic coatings are typically tough, abrasion-resistant materials that offer improved moisture and chemical resistance and operate at temperatures ranging from −40 to 125°C. Typical organic coating dielectric strength is 1000 V/mil. Organic coatings in general, and acrylics and urethanes in particular, are resistant to a broad range of solvents. However, acrylics and urethanes may not be the best coating chemistries for environments exposed to wide fluctuations in temperature over short periods of time, as they tend to crack under thermal stress. Rework on acrylics and urethanes can be handled using mechanical abrasion or microsand blasting. Epoxies tend to be the least popular conformal coatings because of rework issues. Because most board substrates are made of epoxy, the manufacturer may destroy the board by removing the coating. Silicone coatings are soft, flexible materials with a high coefficient of thermal expansion, which allows them to absorb expansion and contraction stress without harming protected components and to function well in environments with extreme temperature cycling from −40 to 204°C. Silicones are very forgiving materials in production because they coat and adhere to just about any surface found on a PCB and offer good resistance to polar solvents, an attribute that makes them ideal for automotive electronics applications. Silicone’s dielectric strength is typically 500 V/mil. Parylene coatings are deposited onto PCBs using gas-phase polymerization to provide a very thin uniform coating. Boards coated with parylene must be processed in a batch operation using special high-vacuum equipment. An adhesion promotion process using silane and isopropyl alcohol followed by a rinse and bake-out step is generally required as a pretreatment for most electronic components bound for parylene deposition. Because this coating can find its way into gaps as small as 0.001 in, airtight masking of interconnects is required to prevent leakage. Parylene is applied in the cured state during the chemical vapor deposition process once the raw dimer material is sublimated.
10.4.3
CURING METHODS
There are various methods available to achieve rapid cure or solidification of conformal coatings, including two-component mixing, heat, moisture, and ultraviolet (UV) light exposure. Each of these methods is appropriate for specific coating chemistries and has distinct advantages and disadvantages. Traditional acrylic, urethane, and epoxy coatings can cure or solidify in minutes using heat or two-component technology, which involves a room-temperature chemical reaction. Silicone coatings may be cured by exposure to heat, UV light, or ambient moisture. Hybrid coating formulations, which incorporate multiple coating chemistries, are designed either to be UV curable or to rely upon dual-cure mechanisms such as UV light, heat, or ambient catalyzation to enhance cure efficiency and increase in-line cure speeds. Every cure method has its own set of advantages and disadvantages. Twocomponent mixing offers wide latitude in adjusting a coating’s cure speed and pot life. However, this technology is often considered undesirable because it requires the user to inventory and mix, in the proper ratio, two different materials. Catalyzed coatings use two-component room-temperature cure. Properly formulated, these materials have a 1:1 mix ratio and a pot life of 8 to 10 h (one shift) or up to 5 days depending on the chemistry, which makes them good candidates for robotic applications. These coatings wet and adhere well in the no-clean process, and have a very effective shadow cure. Recent advances in static mixing and meter mix equipment
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.19
make it possible to mate two-component materials with both atomized spray and selective application equipment. While heat cure improves wetting and lowers viscosity, some heat-cure coatings, particularly platinum-catalyzed silicones, are subject to cure inhibition. This occurs when the coating comes into contact with various sulfur, amine, or organometallic compounds that are sometimes found on boards as residual contaminates from the integrated circuit chip demolding or solder flux process. Additionally, achieving very rapid cure (less than 2 min) requires a temperature in excess of 150°C, which may be high enough to damage some components. Moisture-cure coatings solidify rapidly on exposure to ambient or induced moisture. Extremely moisture-sensitive materials may cure inconveniently (e.g., in the feed line, on the surface of the supply reservoir, or at the dispense nozzle). To control a potential rise in viscosity, the coating should be exposed to as little moisture as possible prior to application. UV light cure is an efficient process. UV light cure materials contain a photoinitiator that cures the coating in seconds when exposed to the proper UV light wavelength. One major problem encountered with UV materials is their inability to cure in areas not exposed to UV light. To overcome this problem, UV coatings have been formulated with a secondary cure mechanism to ensure full cure in areas that are not directly exposed to UV energy. Full cure in shadowed areas is extremely important for board performance with all coating chemistries, as elevated operating or test temperatures may cause uncured material to expand, rupturing the coating fillet and cracking solder joints or integrated circuit packages. UV light curing materials also are subject to oxygen inhibition, a process that occurs at the coating-air interface when oxygen reacts competitively with free radicals generated during UV light exposure. Oxygen inhibition can be overcome by increasing UV light intensity or by reducing oxygen concentration with a nitrogen blanket. Prior to selecting a coating chemistry, the end-use environment of the PCB should be reviewed, assessing the potential for exposure to solvents, temperature extremes, dramatic temperature gradients, and physical stress. No one coating is right for all boards and conditions. Working closely with reputable coating material suppliers that team up with equipment suppliers is the best way to ensure the selection of an appropriate material and process.
10.4.4
DISPENSING METHODS
Normally the final manufacturing step on a PCB assembly line, conformal coatings can be applied manually or with semi- or fully automated techniques. Preparation for conformal coating is a four-step process. First, board cleanliness is established and, if necessary, the board is cleaned. Next, connectors on the PCB are masked off as necessary to protect interconnects from the coating process. Coatings are then applied to the board and cured. Finally, any protective masks on the board are removed. Conformal coating dispensing techniques can be selective or nonselective. Nonselective systems apply the coating uniformly at a very fast pace, but add substantial time and cost in material waste and manual masking/demasking operations. Examples of nonselective dispensing techniques are dip, atomized spray, brush, and wave or flow coating. ●
Dip. Boards are immersed into liquid conformal coatings and withdrawn. Because most dipped coatings will not penetrate very narrow gaps, the dipping process is decreasing in popularity.
10.20 ●
●
●
CHAPTER TEN
Atomized spray. Performed with a standard automotive paint spraying gun, atomized spray is the most popular method of manual coating as it provides fast, uniform coverage. Brush. Typically used for small boards or localized repair and touch-up coating application, brushing is the least used application method and is best suited for low-volume production. Uniform thickness and bubbling are the main problems with brush application. Wave or flow coating. Boards are indexed over a wave or pumped tide of coating material that is applied to one side at a time.
Computer-controlled selective coating systems apply coatings to designated areas of the PCB. Because coatings are precisely dispensed in defined areas, masking and demasking operations as well as other off-line batch activities are significantly reduced or eliminated. Selective coating systems offer substantial savings in resin conservation and off-line masking and demasking labor operations. Some popular selective coating systems are as follows: ●
●
●
●
An airless process using a nozzle to dispense a shaped curtain of coating material, designed for use with solvent-based materials. Evaporation helps control cured film thickness and migration of wet material. A Venturi-type air-assisted dispensing head designed to dispense high-viscosity, solvent-free materials.This device produces a wide (≥1 in) coating curtain from the head and stretches it over the board’s surface. Overspray is dependent on the viscosity of the material being applied and the on/off times programmed into the dispensing profile. A dispensing head shaped like a probe. During selective dispensing, an air assist “twists” the coating stream into a bead, corkscrew, or conical mist pattern on the substrate. Though designed to dispense low-viscosity coatings, this head is not as sensitive to viscosity or chemistry as other technologies, and offers three different dispensing patterns that can be programmed for specific board geometries. Selective atomized spray head systems. These have been found to work well with a wide range of coating viscosities. Multiple tools and spray heads can be incorporated to perform varied dispensing tasks.
All selective systems can be configured with indexers, board inverters, and various cure systems to create an automated process that can be added to a conventional assembly line.
10.4.5
PROCESS ISSUES
The degree of solder mask cure is important to conformal coating performance. If not cured completely, ingredients such as glycols, bromides, and ionic compounds can exit the film during subsequent solder excursions, leaving residues that may ultimately affect the wetting characteristics and adhesion of the conformal coating. Some of these residues can also contribute to electrochemical migration and subsequent dendrite formation. Also, if the solder mask materials are not applied to clean, uncontaminated substrates, many problems can occur during conformal coating. While contaminated boards fail less quickly with a conformal coating, they can still fail because of trapped ionic and corrosive species between the substrate and the coating.
FABRICATION OF ENVIRONMENTALLY FRIENDLY PCB
10.21
Not long ago, board substrates were G-10 or FR-4 materials using RMA flux chemistries and CFC-113 solvent-cleaning methods. These limited variables made tracking compatibility problems relatively simple. Today’s wide selection of solder masks, flux chemistries, solder processes, and alternative cleaning processes has greatly complicated compatibility issues. By performing cleanliness and adhesion testing, board manufacturers can assess the baseline contamination of the PCB and determine whether further cleaning is necessary to ensure conformal coating adhesion, integrity, and reliability. Two effective methods of testing boards are ionic cleanliness testing, which determines levels of chloride per square inch, and surface insulation, an electrical test of solder joints. Although conventional ionic cleanliness testing methods are not as accurate as ion chromatography, they are excellent cleanliness monitors for the production environment. If contamination levels are higher than recommended (5.7 µg/NaCl/cm2), board manufacturers have a number of choices for cleaning the assembly. However, there is still no single, easy, environmentally safe answer to board cleaning. When considering replacement cleaning chemistries, manufacturers should note the cleaning capability, evaporation times, residue, and odor of the process. With any of the available cleaning processes, a chemical reaction may take place that could affect board performance and reliability. No-clean/low-solids fluxes are not usually removed by cleaning prior to coating application. The type and level of remaining residue is dependent on the board design and the solder profile. Inadequate preheat temperatures and dwell time duration can affect the success of a no-clean process. If the profile is not done correctly, ionic and corrosive species from the flux can cause a variety of performance problems. The flux application method and soldering environment also have an impact on how much unvolatized flux species will remain on the assembly. Because a conformal coating can retard corrosion but not prevent it, manufacturers using these materials must benchmark and maintain proper processes to prevent corrosion and ensure reliability. Excessive residues can be coated, but entrapment of the ionic and corrosive species will cause a variety of problems over time. Conventional ionic testing is not routinely performed on low solids because evaporation of the isopropyl alcohol in the test solution causes a visible white residue to form around the solder fillets, creating aesthetic concerns. Water-soluble fluxes work well when used in conjunction with conformal coating operations. As these materials are very aggressive, they must be washed off within minutes of soldering operations. As a result, board surfaces are generally well cleaned and few coating problems are encountered. However, water-soluble fluxes can outgas residual absorbed contaminants and water during heat excursion, causing dendrite formation or coating blisters in areas of flux stains. The drying process must be as controlled as the actual cleaning process. Again, proper process controls will greatly reduce potential problems. For effective, long-term conformal coating solutions, board manufacturers should work closely with conformal coating formulators to determine effective board cleaning methods. Although this was once the sole responsibility of board manufacturers, coating suppliers will now assess and test process-ready assemblies for cleanliness and coat them with application-appropriate coatings to determine their effectiveness in the assembly process. Coating companies will also work closely with equipment manufacturers to prequalify adhesives and dispense equipment, ensuring that materials and the application systems will run smoothly on the customer’s production line.
10.22
CHAPTER TEN
REFERENCES 1. Siddhaye, S., and P. Sheng, “Design for Environment: A Printed Circuit Board Assembly Example,” in Green Electronics, Green Bottom Line, Goldberg, L. H., and W. Middleton, eds., Newnes, Boston, pp. 113–122, 2000. 2. Allen, D., “Life Cycle Assessment and Design for the Environment,” Tutorial Notes, IEEE Symposium on Electronics and the Environment, San Francisco, CA, May 1997. 3. Siddhaye, S., and P. Sheng, “Integration of Environmental Factors for Process Modeling of Printed Circuit Board Manufacturing—II. Fabrication,” Proceedings of the IEEE International Symposium on Electronics and the Environment, pp. 226–233, San Francisco, CA, May 1997. 4. Worhach, P., and P. Sheng, “Integration of Environmental Factors for Process Modeling of Printed Circuit Board Manufacturing—I. Assembly,” Proceedings of the IEEE International Symposium on Electronics and the Environment, pp. 218–225, San Francisco, CA, May 1997. 5. Srinivasan, M., T. Wu, and P. Sheng, “Development of a Scoring Index for the Evaluation of Environmental Factors in Machining Processes: Part I, Formulation,” Transactions of NAMR, 23:115–121, 1995. 6. Balakrishnan, S., and M. Pecht, Placement and Routing of Electronic Modules, pp. 59–96, Dekker, New York, 1993. 7. www.epa.gov/opptintr/dfe/pwb/pwb.html. 8. Coombs, C. F., Printed Circuits Handbook, McGraw-Hill, New York, 1996. 9. PWB Project Case Study 2, On-Site Etchant Regeneration, EPA744-F-95005, July 1995. 10. FR 42 USC Section 7671(c). 5. PWB Project Case Study 1, Pollution Prevention Work Practices, EPA 744F-95-004. 11. U.S. EPA, “Printed Wiring Board Industry and Use Cluster Profile,” EPA 744-R-95-005, pp. 2–38, September 1995. 12. IPC Surface Mount Council White Paper, “PWB Surface Finishes,” SMCWP-005, April 1997. 13. U.S. EPA, “Implementing Cleaner Technologies in the Printed Wiring Board Industry: Making Holes Conductive,” EPA 744-R-97-001, February 1997. 14. Evans, H., and J. W. Lott, “Implementing Green Printed Wiring Board Manufacturing,” in Green Electronics, Green Bottom Line, Goldberg, L. H., and W. Middleton, eds., Newnes, Boston, pp. 153–160, 2000. 15. Ritchie, B., and L. Bennington, “New Conformal Coatings Combine Protection with Environmental Safety,” SMT, 14(4):44–48, April 2000.
CHAPTER 11
GLOBAL STATUS OF LEAD-FREE SOLDERING 11.1
INTRODUCTION
Lead (Pb) has been widely used in the industry for a long time. Of the approximately 5 million tons of lead consumed globally every year, 81 percent is used in the storage batteries, with ammunition and lead oxides together accounting for about 10 percent, as shown in Table 11.1.1 However, despite the longtime acceptance of lead by human society, lead poisoning is now well recognized as a health threat. The common clinical types of lead poisoning may be classified according to their clinical picture as (a) alimentary, (b) neuromotor, and (c) encephalic.2 Lead poisoning commonly occurs following prolonged exposure to lead or lead compounds. The damage often is induced slowly, but definitely. Some historians even speculate that the fall of the Roman Empire could be related to the use of lead in the pipelines that carried drinking water to Roman cities. Due to the profound evidence of toxicity, the use of lead chemicals in paint and gasoline has been prohibited for several years. Storage batteries, due to their almost 100 percent recycling levels, do not contribute to pollution or contamination and thus pose no immediate issue. On the other hand, although solder is only a small percentage by weight of electronic products (TVs, refrigerators, PCs, phones, etc.), these devices often end up in landfills after being disposed, and the lead could leach out into the water supply. For instance, in Japan the lead elution environmental standard in landfills is 0.3 mg/l. In the toxic materials detection tests recently performed by the Japanese Environmental Agency, it was confirmed that the amount of lead leaching from the pulverized remains of TV tubes and printed substrates for PCs and pachinko machines far exceeds the environmental standard.3 In the USA, the regulatory limit for lead in drinking water is 0.015 mg/l per EPA40 CFR141. The limit is 5 mg/l if the test is done by Toxicity Characteristics Leaching Procedure per EPA40 CFR261. A recent study4 demonstrates that the lead leached out from solder can be several hundred times higher than the limit. This concern about lead is a natural result of the growing global concern about the environment. This environmental awareness can be demonstrated by the “German Blue Bird” system, which has been widely used in the European community for some time. Consumers in Germany call it the “Blue Eco Angel,” but its official name is Environmental Label. This is a special logo for products with positive environmental features on the German market and has been in use for two decades. As of today, over 4000 products bear the mark.5
11.2
INITIAL ACTIVITIES
The attempt to ban lead from electronic solder was initiated by the U.S. Congress. In 1990, Reid S2638—subsequently modified to S729—proposed banning of all lead11.1 Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
11.2
CHAPTER ELEVEN
TABLE 11.1 Lead Consumption by Product Product Storage batteries
Consumption (%) 80.81
Other oxides (paint, glass and ceramic products, pigments, and chemicals)
4.78
Ammunition
4.69
Sheet lead
1.79
Cable covering
1.40
Casting metals
1.13
Brass and bronze billets and ingots
0.72
Pipes, traps, and other extruded products
0.72
Solder (excluding electronic solder)
0.70
Electronic solder
0.49
Miscellaneous
2.77
bearing alloys, including electronic solders, and instituting a tax of $1.69 per kg on primary lead and $0.83 per kg on secondary lead used in the industry. However, lead solders were removed from the bills after intense lobbying by the U.S. electronics industry. In 1994, Denmark, Sweden, Norway, Finland and Iceland signed a statement to phase out Pb in long run. On June 16, 1997, a press release from the Swedish Government identified lead as one of the elements to be eliminated from products over the following 10 years.The Sweden Environmental Quality Objectives direct that any new products, including batteries, introduced in Sweden should be largely free from Pb by 2020. Swedish manufacturers are also under a voluntary ban effective in 2000.6 At about the same time frame, recycling laws were proposed in various Asian countries. In October 1996, the Discard Processing/Resource Reclamation Committee of the Industrial Structure Council in the Japanese Ministry of International Trade and Industry (MITI) announced goals for recycling discarded automobiles. Also in 1996, the Japanese Automobile Industrial Association set up a self-managed environmental program. The Pb used in new automobiles is to be cut in half by 2000 (excluding Pb used in batteries) and to one-third of 1996 values by 2005. Most of the Pb usage in Japanese vehicles now is in paint and radiators.
11.3
RECENT ACTIVITIES
There are pending producer responsibility laws for electronic and electrical equipment in a number of European countries. Laws were passed in Holland and Switzerland before 1999 involving producer responsibility. Norway followed in 1999 and Sweden in 2000. In some cases producer responsibility may involve the manufacturer, importer, or reseller taking responsibility for the take-back of products and proper end-of-life treatment. Threshold limits for recycling of specified types of materials may be included also. Denmark has proposed its own Pb ban, but cathode ray tubes and electronics are not included.7 In 1998 the European Union (EU) introduced a draft directive (law) called the Waste from Electrical and Electronic Equipment Directive (WEEE), which calls for
GLOBAL STATUS OF LEAD-FREE SOLDERING
11.3
a ban on lead in all electronics (except automotive) by January 1, 2004. WEEE, which intends to ban the selling and/or importing of electrical/electronic equipment containing lead interconnects, encountered objection from many European electronics trade bodies including EUROBIT (information technology), ECTEL (telecommunications), the Printed Circuit Industry Federation, the Federation of Electronics Industries, etc. On June 13, 2000, the EU Commission officially adopted the WEEE proposals as two separate but associated draft directives for submission to the European Parliament—WEEE and Reduction of Hazardous Substances (ROHS). The ROHS proposals required replacement of lead and various other heavy metals and brominated flame retardants beginning January 1, 2008. On April 24, 2001, the Environmental Committee of the European Parliament adopted a number of amendments to the two pending EU waste directives and advanced their progress through the EU legislative process.8 The Committee agreed to exempt electronic applications where high-temperature melting solder is used, thereby alleviating the pressure in identifying lead-free high-temperature solder alternatives. The deadline for the proposed chemical ban was advanced from 2008 to 2006. Other results from the vote included: (1) an amendment to include servers and storage equipment in the ROHS chemical ban exemption failed; (2) consumables (e.g., printer cartridges) are now covered by WEEE, with producers responsible for collection; (3) collection targets have also been increased from 4 to 6 kg/head/year; (4) recovery targets increased by 5 to 10 percent; (5) the deadline for implementing collection and recycling schemes advanced from 5 years to 30 months after implementation, and (6) the costs of historic waste may be financed collectively by producers through a visible fee. The European Parliament voted on May 15, 2001, to adopt proposals to amend the date for the hazardous material ban in the WEEE/ROHS draft to 2006. Material and components where substitution is “impossible” are exempt from the ban, including lead in server, storage, voice and data transmission, and networking equipment. The Council of Ministers from each European state government discussed the proposals on April 10, 2002. Their opinion aligns with the parliament proposals in setting a target date of January 2006 for a ban on hazardous materials including lead. This opinion of the council will be discussed and voted on by the European Parliament in the next few months in order to allow the directives to be finalized. The list of hazardous materials is due to be reviewed in 2003, with the possibility of extending the ban to, e.g., polyvinyl chloride and other halogenated flame retardants, etc.5 The move toward Pb-free processes raised the attention of some major manufacturers. Nortel Networks is one of the lead-free pioneers in Europe. They initiated a lead-free program in 1991, selected Sn99.3/Cu0.7 in 1994, built 500 lead-free phones in 1998, and targeted meeting the second WEEE in 2001.9 In Korea, Samsung Group declared its commitment to protecting the environment in 1994, and completed development of a “green” semiconductor product that uses no halogen compounds (which contain such toxic substances as lead, chlorine, and bromine). Company officials say that the new-concept device will go into mass production in the second half of 2001. The new concept is initially being applied to Samsung’s 128-Mb synchronous dynamic random access memory package and 256Mbyte module for PC-133 systems. Samsung has replaced the tin-lead compound used to plate the package terminals with a tin-bismuth compound. Conventional tinlead solder paste has also been replaced by a tin-silver-copper solder. Moreover, external packages and printed circuit boards no longer contain halogens such as chlorine, bromine, or antimony. Examples of these green products include conventional packages, ball grid array (BGA) packages, and memory modules.10
11.4
CHAPTER ELEVEN
In Japan, the only legislative activities deal with the reclamation and recycling of electronics. The Home Electronics Recycling Law came into force on April 1, 2001, and applies only to TVs, refrigerators, and similar items. On January 30, 1998, the Japanese Electronic Industry Development Association (JEIDA) and the Japanese Institute of Electronics Packaging (JIEP) presented a report entitled “Challenges and Efforts Toward Commercialization of Lead-Free Solder—Roadmap 2000 for Commercialization of Lead-Free Solder.” This report, which includes a survey of 132 companies, offers the Japanese perspective on leadfree electronics. In this report, JEIDA proposes the following roadmap for lead elimination: ● ● ● ● ● ● ● ●
First adoption of lead-free solders in mass-produced goods: 1999 Adoption of lead-free components: 2000 Adoption of lead-free solders in wave soldering: 2000 Expansion of use of lead-free components: 2001 Expansion of use of lead-free solders in new products: 2001 General use of lead-free solders in new products: 2002 Full use of lead-free solders in all new products: 2003 Lead-containing solders used only exceptionally: 2005
The roadmap presented by JIEP is fairly similar to that of JEIDA, as shown in the following list:1 ● ● ● ● ● ● ●
Mass production using Pb-free solder: 1999 to 2000 Adoption of Pb-free components: 1999 Increased adoption of Pb-free components: 2000 to 2001 Full-scale recycling of assembly boards: 2001 to 2002 Pb-free solder used for new products preferentially: 2003 Pb-containing solder used only exceptionally: 2005 to 2010 Elimination of Pb solders: 2010
Some major Japanese original equipment manufacturers (OEMs) have begun to jointly develop recycling processes for electronic products. A number of major Japanese companies, e.g., Sony, Toshiba, Matsushita, Hitachi, and NEC, have made commitments to go lead free by 2001. This is in advance of Japanese legislation on take-back due to come into force in April 2001. The Pb-free advancement roadmaps of those companies are detailed in the following list: ●
●
Matsushita (Panasonic) has been shipping 40,000 MiniDisc players per month with lead-free solder since October 1, 1998, and plans to eliminate all lead interconnects in four products by April 2001. Matsushita’s market share of its lead-free MiniDisc player jumped from 4.6 percent to 15 percent in 6 months in Japan. This lead-free product was reported to be introduced into Europe in March, 1999. Matsushita is using SnAgBiIn and SnCu.5 Currently only disc players and TVs are using lead-free solder. Matsushita has also indicated it will begin marketing leadfree products in the U.S. in 2000. Each division of Matsushita is charged with using a lead-free solder for at least one electronic product by March 2002. Sony reduced its lead usage in 1999 by half of that used in 1996, and it plans to completely eliminate lead from all products except high-density packaging by
GLOBAL STATUS OF LEAD-FREE SOLDERING
●
●
●
●
●
●
11.5
2001. Sony’s suppliers have been instructed to provide only lead-free materials and parts. Sony uses SnAgBiCu and possibly also Sn93.4/Ag2/Bi4/Cu0.5/Ge0.1 solder for assembly. By 2001, all lead will be eliminated except in high-density electronics packaging. Akikazu Shibata of Sony predicted in 1999 that the company would be 50 percent lead free in 1 to 2 years, and more than 75 percent lead free in 5 years. Sony aims to introduce lead-free solder for all models produced in Japan and overseas by March 2001 and March 2002, respectively. Fujitsu has announced the following lead-free roadmap: ● Complete lineup of lead-free LSI products to be available by October 2000. ● Half of all printed circuit boards used in Fujitsu products to be lead free by December 2001. ● Total elimination of lead from all Fujitsu products by December 2002. This initiative includes not only components internally produced at Fujitsu, but also parts supplied by outside vendors. Toshiba eliminated Pb from refrigerators, TVs, cleaners, PCs, and other major products by December 2000 and plans to eliminate Pb solder in mobile phones by 2003. The company possibly uses SnAgCu.5 Hitachi cut lead usage by 50 percent by March 2000 from 1997 levels. Half of its domestic products were lead free in 2000, with Pb having been eliminated from refrigerators, air conditioners, TVs, VCRs, and PCs since 1999. The company will eliminate inner Pb interconnects by March 2002 and Pb will be completely phased out by March 2004. Hitachi uses SnAgBi and SnAgCu and is currently investing 1.2 billion yen ($11.2 million) to expand production of lead-free solder.5 NEC launched the world’s first three notebook computers with lead-free motherboards, manufactured with SnZn. It plans to install lead-free motherboards in desktop PCs next. NEC will reduce lead use by 50 percent by fiscal 2003 (versus 1998) and is currently using lead-free semiconductors, which began shipping in January 2001. NEC is labeling lead-free products to differentiate them from those that contain lead. The company uses SnAgCu, SnZn, SnCu, and SnZnBi.5 Mitsubishi plans to cut Pb usage to 50 percent by 2004 and to eliminate it entirely by 2005 for four major products.5 NTT has announced it will use no Pb or Cd in newly purchased equipment.11
11.4
IMPACT OF JAPANESE ACTIVITIES
By 2001 the leading Japanese OEMs will have introduced products that contain no Pb in their interconnect systems. This will allow the Japanese to be positioned to exclude products from Japan that do not meet these environmental standards. Furthermore, existence of Japanese products will justify European legislation requiring Pb reduction and highly recyclable electronic products by 2007,12 thereby further increasing the pressure on the rest of the world to convert to Pb-free processes.
11.5
U.S. REACTION
Since the initial attempt in Congress in early 1990s, very little activity has been seen in the U.S. until recently. The automotive segment is probably the only one with sus-
11.6
CHAPTER ELEVEN
tained interest. There is no legislation pending. The Lead Industry Association, Electronic Industries Alliance (EIA), Institute for Printed Circuits (IPC), and National Electrical Manufacturers Association all have been active in lobbying against leadfree legislation. The only activities under way in the U.S. are at the state level and deal primarily with electronics recycling rather than reduction in the use of toxic elements. Obviously, the message from offshore is loud and clear: either work on lead-free soldering now or forget about doing business. Thus the National Electronics Manufacturing Initiative (NEMI) called for a “Lead-Free Initiative Meeting” in February 1999 to review the situation, and since then has rolled out a series of action items to establish a Pb-free direction for the U.S. electronics industry. The second NEMI meeting, held on May 26 to 27, 1999, effectively motivated many manufacturers to get involved in Pb-free development. In April 1999 the IPC board of directors announced the following position statement: The US electronics interconnection industry, represented by the IPC, uses less than 2 percent of the world’s annual lead consumption. Furthermore, all available scientific evidence and US government reports indicate that the lead used in US printed wiring board (PWB) manufacturing and electronic assembly produces no significant environmental or health hazards. Nonetheless, in the opinion of IPC, the pressure to eliminate lead in electronic interconnections will continue in the future from both the legislative and competitive sides. IPC encourages and supports research and development of lead-free materials and technologies. These new technologies should provide product integrity, performance and reliability equivalent to lead-containing products without introducing new environmental risks or health hazards. IPC prefers global rather than regional solutions to this issue, and is encouraging a coordinated approach to the voluntary reduction or elimination of lead by the electronics interconnection industry.
The IPC statement probably fairly truly reflects the opinion of most of U.S. industry: “Pb in electronics is not perceived as a health issue, but government and commercial drivers will push for its adoption anyway. Thus IPC will facilitate activities to enable it to happen.”13 To serve the industry by helping with lead-free initiatives, the IPC developed and maintained the IPC lead-free e-mail forum. In addition, the IPC also organized a conference, IPC Works ’99, held in October 1999, with major emphasis on the Pbfree issue. It was at this conference that the IPC presented the industry with a first draft of the IPC Roadmap for Electronics Assemblies. The impact of this conference is that it was industry talking, including customers, suppliers, and competitors. The HAL User Group, an organization composed of manufacturers of printed circuit boards (PCBs), original equipment manufacturers (OEMs), contract manufacturers (CMs), chemical suppliers, and equipment manufacturers, also met in August, 1999 to address Pb-free surface finishes as a response to the pressure for Pbfree soldering. On January 17, 2001, the Environmental Protection Agency (EPA) lowered the threshold reporting level for lead starting with calendar year 2001 in its Toxic Release Inventory Rule. Therefore, any company that manufactures, processes, or otherwise uses lead or lead-containing products in quantities of 100 lb or more must file a report. This change will require many U.S. manufacturers who had never met the previous requirement of 10,000/25,000 pounds to file reports. The first report for the new requirements will be due on July 1, 2002.14
GLOBAL STATUS OF LEAD-FREE SOLDERING
11.7
To meet this pressing lead-free challenge from offshore, a number of American companies also engaged in lead-free programs, as reported by the IPC:5 ● ●
● ● ●
●
● ●
● ● ● ●
Boeing. Currently evaluating and developing reliability data on Pb-free finishes ChipPAC. Has qualified Pb-free BGAs and is scheduled to go into high-volume production in the fourth quarter of 2000 Delphi Delco. Conducting developmental activities, 2- to 3-year window Lucent Technologies. Aligning with industry through consortium activities Motorola SPS. Currently evaluating and running pilot plant production with leadfree soldering Shipley Co. LLC. Currently offering/developing Pb-free finishes for components and connectors as well as PWB final finish applications Sun Microsystems. Participating in industry consortia and monitoring the situation Texas Instruments. Introduced NiPd finish for components in 1989; believes it is in a leadership position Hadco. Investigating processes to replace SnPb board finishes IBM. Interim strategy developed in 1999; plans to stay ahead of the industry Honeywell. Has formed a team to formulate activities on Pb-free processes Visasystems. Has a patent on an organic solderability preservative (OSP)
However, a second opinion also exists on the aggressive move toward lead-free solder. On April 10, 2001, the IPC and 35 other trade associations sued the EPA.This suit is directed against the EPA ruling that reduces reporting thresholds of lead and challenges the PBT label for metals. As reported by Harvey Miller at InfraFOCUS, “PBT stands for persistent, bioaccumulative, and toxic; that is the combination that characterizes another class of toxins, often confused with metals—persistent organic pollutants, such as polychlorinated biphenyls, DDT, and other synthetic creations of the last 50 years that are truly dangerous to life. These chemicals cause cancer, disrupt metabolism, and hormonal signals. It is very difficult to eliminate them or reverse their damage. Metals, even essential ones, can present toxic effects when excessive amounts are present, but on every other score, they are comparatively benign.” This suit is supported by the numerous scientific reports in the Expert Workshop that was cosponsored by the EPA and was held in January 2000.
11.6
WHAT ARE LEAD-FREE INTERCONNECTS?
Pb may be present in metals, such as tin, as an impurity at a level of 10%
Under uniaxial tension at room temperature; also important for fabricating solder wires and preforms
On Cu in dip test
12.3
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
TABLE 12.2 Toxicity Ranking of Elements Presented in Order of Increasing Occupational Chronic Toxicity
Element
LDLo (mg/kg body weight)
Bi
15
Moderate toxicity: heart, liver, lungs
None
None
None
In
10†
High toxicity: lungs, gastrointestinal tract
Irritation
None, related solely to indium metal
0.1 (any compound)
Zn
124‡
Minimal toxicity: skin
Irritation (oxide)
Metal fume fever (oxide)
5 (oxide fume)
Cu
0.12‡
High toxicity: reproductive organs, possible carcinogen and teratogen
Irritation (dust, mist)
Irritation, metal fume fever
1 (dust)
Ag
1§
High toxicity: skin
None
Permanent discoloration of skin, eyes, mucous membranes; irritation; metal fume fever
0.1 (dust and fume); 0.01 (Ag and soluble compounds)
Moderate to low toxicity: gastrointestinal tract
Irritation
Difficulty breathing
2 (inorganic); 0.1 (organic)
High toxicity (oral): heart, liver, lungs
Irritation
Emphysema, pulmonary edema
0.5 (Sb and compounds)
High toxicity: nervous system, carcinogen of lungs, suspected teratogen
None
Nervous system effects, anemia, kidney damage; reproductive and developmental effects
0.05 (inorganic)
Sn
No value found
Sb
15
Pb
450‡
Effects
Acute toxicity
Chronic toxicity
OSHA PEL and ACGIH TLV* (mg/m3)
* For a 5-day workweek with 8-h workdays. † LDLo based on animal studies; high toxicity when injected, low toxicity when inhaled. ‡ Toxic dose, lower limit. § Toxic concentration, lower limit (mg/m3). ACGIH, American Conference of Government Industrial Hygienists; LDLo, lethal dose, lower limit; OSHA, Occupational Safety and Health Administration; PEL, Permissible Exposure Limit; TLV, threshold limit value.
However, the selection or elimination of an element for solder should not only consider the toxicity of the element itself, but also the overall quantity of the element to be used in an electronic product. Figure 12.1 shows the chemical content of a printed circuit board for a mobile product. The estimated overall contribution of metallic elements on toxic potential can be roughly ranked in decreasing order as shown here: Pb > Cu > Ni > Ag > Al > Sn > Au
12.4
CHAPTER TWELVE
FIGURE 12.1 Chemical content of a printed circuit board of a typical mobile product.4 Light columns: material mass. Dark columns: material assessment by means of toxic potential indicator (TPI).
Many of these metallic elements, such as Cu and Ag, are not primarily used as solder. Cu is used as the circuitry conductor or ground plane, while Ag may be used as a thick film material or a surface finish.As long as Cu and Ag continue to be used for those applications, eliminating them from solder materials due to toxicity considerations is virtually meaningless.
12.3
COST AND AVAILABILITY
The cost and availability of potential elements to be used in solders are shown in Table 12.3.5 With Pb being the cheapest element, all Pb-free alternatives are destined to be more expensive than eutectic Sn-Pb solder. Zn, Cu, and Sb are relatively low in cost; however, it is questionable that they can serve as essential constituents. The high cost of Ag and In suggests that those elements should not compose more than a small percentage of the solder. The limited availability of Bi and In imparts constraints on using them as significant constituents.
12.4 12.4.1
DEVELOPMENT OF LEAD-FREE ALLOYS Existing Alloys
The first group of Pb-free solder candidates is the existing Pb-free alloys. This includes (1) binary eutectic Sn-containing alloys, such as Sn-Ag, Sn-Au, Sn-Cu, Sn-Bi, Sn-In, Sn-Sb, and Sn-Zn, (2) noneutectic binary Sn-containing alloys,6 such as 97.5Sn-2.5Ag (melting range 221 to 226°C), 95Sn-5Ag (221 to 240°C), 90Sn-10Ag (221 to 295°C), 97Sn-3Cu (227 to 300°C), and 60Sn-40Bi (138 to 170°C), (3) Pb-free
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
12.5
TABLE 12.3 Cost and Availability of Elements
Element
Cost ($/lb, as of Feb. 3, 1999)
Density (lb/in3)
Annual U.S. consumption (millions of lb)*
Availability
Pb
$0.45
0.41
7040
Available
Zn
$0.50
0.258
1560
Available
Cu
$0.65
0.324
4900
Available
Sb
$0.80
0.239
100
Available
Bi
$3.40
0.354
9
Sn
$3.50
0.264
180
Limited
Ag
$84.20
0.379
3.5
Limited
In
$125.00
0.264
0.2
Scarce
Available
* As defined by the U.S. Bureau of Mines.
alloys without tin,6 such as 97In-3Ag (143°C), 99.3In-0.7Ga (150°C), 95In-5Bi (125 to 150°C), 67Bi-33In (109°C), and 88Au-12Ge (356°C), and (4) other Pb-free alloys,6 such as 65Sn-25Ag-10Sb (233°C), etc. Many of the materials in this group, particularly the binary eutectic Sn-containing alloys, have been used by the electronics industry for many years, and their properties and performance are well understood.
12.4.2
MODIFICATION
The second group of candidates is newly modified preexisting alloys. Modification is often accomplished by addition of a small amount of additional elements, such as Ag, Cu, Bi, In, Sb, Ge, P, Ni, Fe, Au, Ga, or Co, with the goals of improving the wetability, bond strength, oxidation resistance, and impurity tolerance level; reducing the melting temperature; refining the grain structure, and so on. 12.4.2.1 Wetting. The wetting process is favored by a low-surface-energy solder, which often can be regulated by addition of impurities. The general rule is that a small amount of surface-active impurity, usually low in surface energy, can produce a marked decrease in surface energy, while similar amounts of a surface-inactive impurity do not produce more than a very small rise in surface energy. It follows that the effects of surface-inactive impurities on solder should be too small to have any significant effect on wetting behavior.7 Figure 12.2 shows the effect of impurities on surface energies for some relevant binary systems with tin. The effect of the surface energy of additives on the surface energy of alloys can be further illustrated by examining the 60Sn-40Pb system. Surface tension isothermals (250°C) for 60Sn-40Pb with 0 to 4 percent Bi or 0 to 5 percent Sb show a nonlinear fall with increasing ternary addition, which may be explained by the low surface tension of the third elements. Surface tension isotherms for 60Sn-40Pb with 0 to 2 percent Ag (215 and 250°C) or 0 to 0.6 percent Cu (250°C) indicate higher values with increasing ternary addition, which may be explained by the higher surface tension of the third elements. However, there are exceptions to this rule; for example, increasing ternary addition of low-surface-tension P at 0 to 0.013 percent results in a higher surface tension.8
12.6
CHAPTER TWELVE
FIGURE 12.2 Effect of impurities on surface energies.
Another driving force for wetting is the rate of formation of intermetallic compounds. Since formation of intermetallics involves reaction between solder and base metal, this inevitably results in more spreading or more wetting. Accordingly, a higher formation rate typically results in a better wetting. The ability of Sn to easily form intermetallics with many metals is the primary reason it is the essential constituent in many solder alloys. Addition of other elements may affect the formation rate of intermetallics, thus affecting the wetting. Table 12.4 shows the common intermetallic compounds encountered in the electronics industry. 12.4.2.2 Melting Temperature and Bond Strength. Since tin, which is often the preferred primary constituent, exhibits a melting point of 232°C, additives that can lower the alloy melting temperature are often desired in order to minimize thermal damage to both components and boards. Although several elements such as Hg and Cd are capable of lowering the melting temperature of alloys, Bi and In are the two most commonly utilized elements for Pb-free solder applications due to their benign nature. Furusawa et al.9 reported that the addition of small amounts of some additive elements would reduce the melting temperature and the bond strength, but would initially increase the wetting of solders, then reach a maximum, then decrease wetting, as shown in Fig. 12.3. The wetting phenomenon observed in this case suggests that the relationship between wetting and surface energy may only be a secondary effect. The effect of additives on melting temperature seems to be applicable to Bi and In as additives, and is supported by the studies on the effect of Bi addition on Sn-Zn,10 Sn-3.5Ag,11 and Sn-Ag-Cu10 systems, including Sn-3.5Ag-1Cu,12 as shown in Figs. 12.4 through 12.6. It is also supported by the effect of In addition on Sn-3.5Ag,12 as shown in Fig. 12.5. Besides Bi and In, a number of other elements, such as Mg, Ag, Cu, Al, Ga, and Zn, also exhibit a melting temperature depression effect, as shown in Table 12.5.13 The effect of additives on bond strength reported by Furusawa also appears to be applicable to Bi and In as additives. Therefore, the effect of Bi content on the pull strength of the Sn-3.5Ag-Bi system (see Fig. 12.6) and the effect of addition of In and Bi on the tensile strength of the Sn-3.5Ag-1Cu alloy12 (see Fig. 12.7) all exhibit a decrease in mechanical strength with increasing content of additives.
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
12.7
TABLE 12.4 Common Intermetallic Compounds Encountered in the Electronics Industry Solder
Substrate metallization
Intermetallic compound
Au-based
Cu-based
Au3Cu, AuCu, AuCu3
Au-based
Pb-based
AuPb2, Au2Pb Au2Bi
Bi-based
Au-based
Bi-based
In-based
BiIn, Bi3In5, BiIn2
In-based
Cu
Cu11In9, Cu4In, Cu2In
In-based
Au-based
Au7In, Au4In, Au3In, Au7In3, Au3In2, AuIn, AuIn2
In-based
Ag-based
Ag3In, Ag2In, AgIn2
In-based
Ni
Ni3In2, Ni3In, NiIn, Ni2In3
In-based
Sn solder coating
In3Sn, InSn4
Sb-based
Brass or Zn coating
ZnSb, Zn3Sb2
Sb-based
Cu-based
Cu3Sb, Cu5.5Sb, Cu4.5Sb, Cu3Sb, Cu3.3Sb, Cu2Sb
Sb-based
Ag-based
AgSb, Ag3Sb
Sb-based
Au-based
AuSb2
Sn-based
Cu
Cu6Sn5, Cu3Sn
Sn-based
Au-based
AuSn, AuSn2, AuSn4
Sn-based
Ag-based
Ag3Sn
Sn-based
Pd-based
PdSn4
Sn-based
Ni
Ni3Sn4
It should be pointed out that Furusawa’s observation on the relation between addition and wetting is not well supported by the work on Bi. Zhao et al.12 reported that for the Sn-3.5Ag-1Cu system, wetting improves with increasing addition of In but deteriorates with increasing addition of Bi, as shown in Fig. 12.8. The adverse effect of Bi on wetting cannot be explained by its reduced surface tension, as shown in Fig. 12.2. Presumably this can be attributed to the poor wetting ability of Bi itself.
(a)
(b)
FIGURE 12.3 Effect of additive amount on solder melting point, joint bond strength, and wetting (spread factor).
12.8
CHAPTER TWELVE
FIGURE 12.4 Effect of Bi content on the melting range of Sn-Ag-CuBi and Sn-Zn-Bi determined by differential scanning calorimeter.
It has been reported that, based on the wetting study on a series of eutectic binary solder alloys, the ability to promote spreading for several elements can be ranked as follows: Sn > Pb > Ag > In > Bi.14 Ackroyd et al.15 studied the effect of additives on the wetting of 60Sn-40Pb. Results indicate that Al, Sb, As, Cd, P, S, and Zn all cause a decrease in spread area. Bi shows adverse effects on steel and brass but no effect on Cu. On the other hand, Cu addition causes a decrease in spread area on steel but a slight increase on brass. 12.4.2.3 Oxidation Resistance. Elements such as P are sometimes used as deoxidants in the production of Cu and solders from secondary metals.15 At levels of 0.01 percent, P significantly reduced the oxidation of 60Sn-40Pb at all temperatures in the stirring tests.14
FIGURE 12.5 Effect of addition of In or Bi on the melting temperature of Sn-3.5Ag-1Cu alloy.
12.9
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
FIGURE 12.6 Effect of Bi content on Sn-3.5Ag-Bi alloys.
12.4.2.4 Grain Structure. Creep is deformation of materials with time under a given tension or shear load. Creep occurs via thermally activated processes. It is important when the service temperature exceeds half the melting temperature (in degrees kelvin) of solder. Creep is the most important deformation mechanism of solder.16 Depending on the stress level, the deformation mechanism can be divided into three phases.17 With increasing stress τ, the creep mechanisms shift from dislocation climb-controlled bulk creep to grain boundary slide-controlled intergranular creep to dislocation glide-controlled creep. At high stress, the deformation mechanism undergoes transition to tertiary creep and elongation to failure. The creep is sensitive to microstructure, and the mechanisms include (1) onset of cavitation damage at grain boundaries and (2) plastic instability leading to inhomogeneous deformation. Morris et al.18 reported that cavitation is responsible for tertiary creep in bulk solder samples tested in tension. CavTABLE 12.5 Effect of Additive Elements in Depressing Melting Temperature of Sn-Binary Alloys Melting temperature depression (°C/wt%) Additive element
160–183°C
184–199°C
200–230°C 1.8
In
2.3
2.1
Bi
1.7
1.7
1.7
Mg
—
—
16.0
Ag
—
—
3.1 (above 221°C)
Cu
—
—
7.1 (above 227°C)
Al
—
—
7.4 (above 228°C)
Ga
2.6
2.5
2.4
Zn
—
3.8 (above 198°C)
—
12.10
CHAPTER TWELVE
FIGURE 12.7 Effect of addition of In and Bi on the tensile strength of Sn-3.5Ag-1Cu alloy.
ities nucleate primarily at three- or four-grain junctions. They grow with strain and merge to form larger voids to cause failure.This process is aggravated by (1) increase in grain size, which enhances the stress concentration at grain junctions; (2) irregular grain shapes, which introduce sites of unusual stress concentration; and (3) (possibly) intergranular precipitates, which constrain deformation at grain boundaries, thus resulting in uneven stress distribution. Plastic instability mainly incurs at shear bands, which often follow planes of microstructural weakness, such as phase boundaries and colony boundaries in eutectic materials.19,20 The development of shear bands is particularly pronounced in solders exhibiting unstable, eutectic microstructures that are easily recrystallizable, such as eutectic Sn-Pb. In these solders, the incipient shear bands cause development of the well-defined recrystallized bands for joints that are crept or fatigued in shear. Such a localized recrystallized material, usually observed near an intermetallic layer, accelerates damage processes and shortens the fatigue life of solder joints. Since a larger grain size results in a higher creep and failure rate, a microstructure with a refined grain structure is typically desired. In general, this is considered one
FIGURE 12.8 Effect of addition of In or Bi on the wetting angle of Sn-3.5Ag-1Cu alloys.
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
12.11
of the most effective methods for improving the reliability of solder alloys, and is often accomplished through the addition of a small amount of elements such as Cu, Zn, As, Fe, or Ag into the alloy. These elements often precipitate at the grain boundary, thus retarding the further growth or recrystallization of the grains in the solder. For example, addition of 1 percent Cu dramatically slows coarsening of eutectic Sn-Bi.21–23 On the other hand, addition of insoluble dispersoid Fe particles using a magnetic distribution technique forms a three-dimensional network of finely dispersed iron particles in a Bi-43Sn eutectic solder. These iron particles reduce the coarsening and the onset of tertiary creep of 57Bi-43Sn, thus improving the microstructural stability, raising the service temperature, and resulting in a fivefold increase in creep resistance at 100°C.24–27 The addition of 1 percent Zn significantly improves the mechanical strength of 95.5Sn-3.5Ag alloy by as much as 48 percent while maintaining the same level of ductility. It also significantly improves creep resistance. The high-temperature creep resistance of Zn-containing alloy is improved more than an order of magnitude. Strengthening is attributed to a substantial refinement of and more spherical Ag3Sn precipitates in the solidification microstructure. In this case, Zn is incorporated in the more corrosion-resistant Ag3Sn precipitates. These precipitates suppress the formation of Sn dendrites and leave the Sn-rich matrix primarily free of Zn in solid form.28,29 The addition of a small amount ( 1 percent is not desirable as it causes precipitates of additional intermetallic compound phases that deplete the finely dispersed precipitates in the surrounding matrix and induce nonuniformities in the microstructure that consequently deteriorate the mechanical properties.30 AT&T reported that small alloying additions of Ag dramatically improve the mechanical properties of 87Sn-8Zn-5In alloy (melting point 188°C) due to elimination of the coarse and nonuniform distribution of platelike dendrites and refining effective grain size in the solidified microstructure.25 However, care should be taken in the selection of elements because some elements may cause grain coarsening. For instance, addition of 0.001 percent Co to eutectic Sn-Bi inhibits dissolution of Cu and coarsens the microstructure relative to the solidified structure of the alloy in pure form or with small additions of As, Fe, or Cu.21–23 12.4.2.5 Impurity Tolerance. Some impurity elements have significant adverse impact on soldering performance and physical properties. The sensitivity of solder toward impurities is a function of the impurity element.15 Table 12.6 shows the detrimental effect of some impurities on the properties of 60Sn-40Pb. The tolerance of solder alloys toward impurities may be increased by the addition of some other elements. For instance, Mei et al. studied the effect of Pb contamination on Sn-Bi eutectic and found that formation of an Sn-Pb-Bi ternary eutectic phase resulted in drastic failure of the solder. The Pb dissolves into molten Bi-Sn during the soldering process, resulting in the formation of a 52Bi-30Pb-18Sn (melting point 96°C) ternary eutectic structure in the solidified solder joint. The solder joints became mechanically weak when subjected to thermal cycling at temperatures exceeding 96°C because the low-melting-point ternary eutectic phase
12.12
CHAPTER TWELVE
TABLE 12.6 Lowest Impurity Levels Producing Detrimental Effect on a 60Sn-40Pb Solder Impurity element Ag
Al
As
Au Bi
Cd Cu
Impurity, % 2
0.0005
Effect ●
Increases spread and strength of solder; grittiness in excess of solubility.
●
Ag3Sn intermetallic compound is soft, ductile, and nonembrittling.
●
Oxide-promoting element; causes a lack of adhesion, grittiness, and dull solder surface.
●
No dewetting on Cu or brass; 0.001% showed onset of dewetting on steel and nickel.
●
Sb eliminates Al by promoting rapid drossing out of AlSb compound.
0.2
●
25% decrease in area of spread.
0.005
●
Dewetting and grittiness on brass, probably due to formation of As-Zn intermetallic compound.
0.1
●
Gritty joints and surfaces.
●
Weakens solder dramatically at 4%.
●
Discoloration and oxidation of solder coating.
●
Very slightly reduces the area of spread.
●
Increases the rate of spread.
●
25% decrease in area of spread.
●
Dull surface due to oxide film.
●
Grittiness due to Cu-Sn intermetallic compound.
●
Excessive solder increases the liquidus temperature of the solder, making it more viscous or sluggish.
●
Negligible effect on wetting.
0.5
0.15 0.29
Fe
0.02
●
Grittiness of solder coating.
Ni
0.05
●
Grittiness at over 0.02%.
P
0.01
●
Deoxidant.
●
Dewetting at 0.012% on Cu and steel.
●
Grittiness at 0.1% on Cu.
●
Additions up to 0.25% produce no dewetting effects, but give a severe gritty appearance of the solder coating due to the presence of discrete intermetallic compound particles of SnS and PbS.
●
Powerful grain refiner.
●
Area of spread decreases slightly with increase in Sb content.
●
Prevents transformation of beta Sn to alpha Sn at subzero temperatures.
●
Drosses out An, Al, and Cd from solder.
●
Oxide-forming element.
●
Dewetting at 0.001%.
●
Loss of solder brightness at 0.005%.
S
Sb
Zn
0.0015
1
0.003
DEVELOPMENT OF LEAD-FREE SOLDER ALLOYS
12.13
accelerated grain growth and phase agglomeration. The addition of small amounts of indium (2 to 3 percent) into 58 Bi-42Sn solder may eliminate the formation of the ternary eutectic phase, as indicated by the disappearance of the ternary phase peak in differential scanning calorimeter measurements.31
12.5
LEAD-FREE ALLOYS INVESTIGATED
The Pb-free solder alloys investigated are summarized in Table 12.7.32 Also listed are the two controls, 63Sn-37Pb and 62Sn-36Pb-2Ag. The nomenclature of each alloy category is based on the elemental composition percentage, with the composition with a higher percentage listed first.
12.6
FAVORITE Pb-FREE ALLOYS
The favorite choice of Pb-free alloys differs from region to region, as discussed in the following text.
12.6.1
JAPAN
As discussed in Chap. 11, Japan is leading in terms of implementing Pb-free soldering processes. Although the selection of alloys in Japan is not standardized yet, the Japanese Electronic Industry Development Association (JEIDA) does provide recommendations about alloys based on applications, as shown in Table 12.8.106 The JEIDA recommendations can also be presented according to the melting temperature for mid–melting range applications, as shown in Fig. 12.9.105 Figure 12.10 shows the survey results on lead-free soldering implementation status in Japan conducted by Senju in April 2001.10 It is interesting to note the high overlap between JEIDA recommendations and industry implementation status, suggesting an active participation of Japanese industrial Pb-free soldering manufacturers in the JEIDA program. Due to the wide range of products in the electronics industry and the highly diversified coverage for major electronic manufacturers, the selection of Pb-free alloys may be a multiple choice. This phenomenon is well exemplified by Fig. 12.11, which shows the road map of Panasonic on Pb-free alloy development.11
12.6.2
EUROPE
The European consortium, BRITE-EURAM, recommends that 95.5Sn-3.8Ag0.7Cu be considered as an all-purpose alloy. Other alloys with potential are 99.3Sn0.7Cu, 96.5Sn-3.5Ag, and Sn-Ag-Bi. In the UK, the Department of Trade and Industry (DTI) provided its perception on Pb-free alloy choices as shown in the following list. ● ● ●
High professional group (automotive, military): Sn-Ag-Cu(Sb) Medium professional group (industrial, telecommunications): Sn-Ag-Cu, Sn-Ag General consumer and low professional group (TV, audio-video, office equipment): Sn-Ag-Cu(Sb), Sn-Ag, Sn-Cu, Sn-Ag-Bi
TABLE 12.7 Pb-Free Solders Investigated
Alloy category
Composition
Solidus/ liquidus (°C) 183
Advantages
Structural coarsening; prone to creep.
Manufacturer or investigator NCMS (control)1
12.14
Sn-Pb
63Sn-37Pb
Sn-Pb-Ag
62Sn-36Pb-2Ag
Au-Sn
80Au-20Sn
280
Bi-Cd
60Bi-40Cd
144
Toxic.
Indium
Bi-In
67Bi-33In
109
Poor wetting on Cu.
Indium
179/180
Overall good properties, low cost. UTS 4442 psi. YS 3950 psi. Elongation 48%, YM 15.7 GPa.33 σ 464 dyn/cm,34 380 dyn/cm.35 UTS 27 MPa,36 SS 39 MPa.37
Disadvantages
NCMS (control)1
UTS 6904 psi, elongation 31%, YS 6287 psi, YM 18.0 GPa.33 Creep and corrosion resistant.
79
Hard and brittle; melting point too high; expensive.
Bi-In-Sn
57Bi-26In-17Sn
Bi-Sn
50Bi-50Sn
138/152
YS 8263 psi, UTS 8965 psi, elongation 21% and 53%.33
Wide pasty range.
NCMS33
52Bi-48Sn
138/151
YS 6414 psi, UTS 8834 psi, elongation 57%.33
Wide pasty range.
NCMS33
57Bi-43Sn
Bi-Sn-Ag
Melting point too low.
YS 7972 psi, UTS 8540 psi, elongation 77%.33 Good fluidity. UTS 8766 psi.1 Elongation 46%.33 Low σ, 349 dyn/cm,34 300 dyn/cm.35
NCMS
58Bi-42Sn
138
Strain rate sensitivity; poor wetting. Concerns: (1) eutectic 52Bi-32Pb16Sn (96°C); (2) Bi is byproduct of Pb mining. YM 11.9 GPa.33
95Bi-5Sn
134/251
Indium
56Bi-43.5Sn-0.5Ag
Ternary eutectic
NCMS33
57Bi-42.9Sn-0.1Ag
NCMS33
138/140
57Bi-42Sn-1Ag
BGA bend strength using 95.8Sn-3.5Ag-0.7Cu ball and 57Bi-452Sn-1Ag paste is 65% of that of Sn63 paste.38
HP39
57Bi-41Sn-2Ag
140/14733
YS 9487 psi, UTS 10,390 psi.33 Thermal fatigue life > Sn63 > 58Bi-42Sn.40
Elongation 31%.33
NCMS
56Bi-40.5Sn-2Ag1.5Sb
137/14533
YS 9063 psi, UTS 9946 psi.33
Low elongation, 27%.33
NCMS
55.5Bi-40Sn-3Ag1.5Sb
137/14733
YS 8665 psi, UTS 9379 psi.33
Elongation 45%.33 Wide pasty range.
NCMS
55Bi-40Sn-3Ag-2Sb
138/15033
YS 8984 psi, UTS 9807 psi.33
Elongation 44%.33 Wide pasty range.
NCMS
54Bi-39Sn-3Ag-2Sb
138/15433
Low elongation, 3.7%.33 Wide pasty range.
NCMS
Bi-Sn-Ag-Sb-In
54Bi-39Sn-3Ag-2Sb2In
99/13833
YS 5055 psi, UTS 11,640 psi.33
Low elongation, 13%.33 Very wide pasty range.
NCMS
Bi-Sn-Ag-Sb-Cu
54Bi-39Sn-3Ag-2Sb2Cu
YS 11,440 psi UTS 12,280 psi.33
Low elongation, 4%.33
NCMS
Bi-Sn-Cu
55Bi-42Sn-3Cu
>400
High Cu, wide pasty range, high liquidus, low elongation.33
NCMS
55Bi-43Sn-2Cu 48Bi-48Sn-4Cu
138/14033 >400
Elongation 41%.33 High Cu, wide pasty range, high liquidus, low elongation.33
NCMS NCMS
Bi-Sn-Ag-Sb
12.15
Bi-Sn-Fe
54.5Bi-43Sn-2.5Fe
Bi-Sn-In
56Bi-42Sn-2In
137 126/14033
YS 8985 psi, UTS 9478 psi.33
Creep and fatigue resistance.
Developmental stage.
AT&T
YS 7224 psi, UTS 8429 psi, elongation 116%.33
Quenched alloy shows ternary melting (99°C), 116% total elongation.33
IBM, NCMS
(Continues)
TABLE 12.7 Pb-Free Solders Investigated (Continued)
Alloy category Bi-Sn-In (cont.)
Composition
Solidus/ liquidus (°C)
57Bi-42Sn-1In
132/13833
57Bi-41Sn-2In
33
127/140
Advantages
Disadvantages Poor wetting.33
YS 7304 psi, UTS 8436 psi, elongation 72%.33
Manufacturer or investigator IBM NCMS Ford41
(37-57)Bi-(37-53)Sn(6-10)In
12.16
Bi-Sn-In-Cu
56.7Bi-42Sn-1In0.3Cu
132/13833
YS 8359 psi, UTS 8985 psi.33
Low elongation, 38%.33
NCMS, IBM
Bi-Sn-Sb
57Bi-41Sn-2Sb
141/15033
YS 8521 psi, UTS 9586 psi.33
Elongation 47%.33 Wide pasty range.
NCMS
57Bi-42Sn-1Sb
138/14933
YS 8285 psi, UTS 8944 psi, elongation 60%.33
Bi-Sn-Zn
55Bi-43Sn-2Zn
Bi-Sb
95Bi-5Sb
In-Ag In-Bi-Sn In-Sn
NCMS NCMS33
Ternary eutectic ∼275/ ∼308
97In-3Ag
143
90In-10Ag
141/237
48.8In-31.6Bi-19.6Sn
59
51.0In-32.5Bi-16.5Sn
60
60In-40Sn
118/127
52In-48Sn
118
50In-50Sn
118/125
Ford Poor wetting; expensive.
Indium Indium Indium
Au soldering.
Melting point too low; poor fatigue and mechanical properties; expensive.
Indium
Indium
Sn
100Sn
Sn-Ag
95Sn-5Ag 96.5Sn-3.5Ag
98Sn-2Ag
232
Wetting. UTS 21 MPa vs. 17.5 MPa for Pb,36 SS 26 MPa vs. 13 MPa for Pb.37
221/24542
No coarsening. UTS 10,100 psi, SS 8,400 psi.42
221
Good strength; creep resistance, Fatigue life 1.1 times that of Sn63,13 better than 95.5Sn-3.8Ag-0.7Cu, comparable with 99Sn-1Cu.43 Shear strength not affected by baking time and better than for Sn62.44 YM 26.2 GPa,33 56 GPa.45 UTS 55 MPa vs. 31–46 MPa for Sn6345 and 21 MPa for Sn.36 UTS 8900 psi, SS 4600 psi.42 SS 61.2 MPa46 vs. 26 MPa for Sn.37
Whisker and tin pest growth.
Indium
Welco Castings42 Poor isothermal fatigue at low strain; melting point slightly too high. Pad trace may crack due to high rigidity of solder.44 UTS 3873 psi, YS 3256 psi, elongation 24%,33 35% vs. 35–176% for Sn63.45 σ 493 dyn/cm.34
Indium
NCMS33
221/226
12.17
Motorola47
Sn-Ag-Au
Balance Sn-(12.2)Ag-(1-2.2)Au
Sn-Ag-Bi
93.5Sn-3.5Ag-3Bi
200/217 or 208/21733
SMT defect rate 5
Mitsui50
Sn-Ag-Bi-In-Zn
Balance Sn-(1-6)Ag(0.2-0.6)Bi-(0.20.6)In-(0.2-0.6)Zn
Lucent51
Sn-Ag-Bi-Sb
94Sn-2.5Ag-2Bi1.5Sb
219/22633
YS 7070 psi, UTS 8117 psi.33
Low elongation, 21%.33
NCMS
93Sn-3Ag-2Bi-2Sb
219/22633
YS 6918 psi, UTS 9212 psi.33
Low elongation, 36%.33
NCMS
12.18
Sn-Ag-Bi-In
(90.3-99.2)Sn-(0.53.5)Ag-(0.1-2.8)Cu(0.2-2)Sb
AIM52
Sn-Ag-Bi-Zn-Cu
(93.5-94)Sn-(2.53)Ag-(1-20bi-(12)Zn-1Cu
IBM53
Sn-Ag-Cd-Sb
95Sn-3.5Ag-1Cd0.5Sb
221/223
YS 7545 psi.33
93.6Sn-4.7Ag-1.7Cu 95Sn-4Ag-1Cu 95.5Sn-4Ag-0.5Cu
95.5Sn-3.9Ag-0.6Cu
NCMS,33 Alpha IBM54
(89.4-95.1)Sn-(33.8)Ag-(0.7-1.3)Cd(0.2-0.5)Sb Sn-Ag-Cu
Low elongation, 15%.33 Low usage, contains Cd.
217/24433
Sandia, Iowa State University
217/220 217/22533 or 221/23055 217
Published 50 years ago. Creep slower than for Sn63.56
Heraeus
Recommended by NEMI.
NEMI
95.5Sn-3.8Ag-0.7Cu
217/220
95.4Sn-3.6Ag-1Cu
217/ 217.946
Nokia and Multicore—yield and reliability equal or better than for Sn63. BRITE-EURAM project reports better reliability and solderability than for SnAg and SnCu, recommends this alloy for general-purpose use.
Nokia and Multicore, BRITEEURAM
SS 67 MPa.46
95.2Sn-3.5Ag-1.3Cu
NIST alloy, NCMS33
95.6Sn-3.5Ag-0.9Cu
217
Eutectic.
95.75Sn-3.5Ag0.75Cu
218
Creep rate much lower than for Sn63.57 Impact strength 2.5 times that of Sn63.49
96.1Sn-3.2Ag-0.7Cu 12.19
217/218
95.4Sn-3.1Ag-1.5Cu
216/217
97.25Sn-2Ag-0.75Cu Sn-Ag-Cu-Bi-Zn
Balance Sn-(3.57.7)Ag-(1-4)Cu(0-10)Bi-(0-1)Zn(Si, Sb, Mg, Ca, rare earth, miscellaneous metal 3)Bi(3.2-4.83)Ag 91.7Sn-4.8Bi-3.5Ag
211/215
Low cycle fatigue comparable to that of Sn63. YS 6,712 psi, UTS 10,349 psi.33
96.5Sn-3Bi-0.5Ag
223/226
Low cycle fatigue better than Sn63.
90.5Sn-6Bi-3.5Ag
220
Low cycle fatigue life 0.9 times that of Sn63.13 Low elongation, 16%.33
H-Technol13
H-Technol13 Matsushita
90.8Sn-6.1Bi-3.1Ag
137/21533
Low solidus, very wide pasty range.
NCMS
86.5Sn-10Bi-3.5Ag
137/20833
Very wide pasty range.
Matsushita, NCMS
81.7Sn-15Bi-3.3Ag
137/20033
Very wide pasty range.
Matsushita, NCMS
78Sn-19.5Bi-2.5Ag
138/19633
Elongation 17%, wide pasty range.33
NCMS, Kester
63.2Sn-30Bi-6.8Ag
137/28233
Very wide pasty range.
NCMS
56Sn-41Bi-3Ag
138/16633
Elongation 39%, wide pasty range.33
NCMS, IBM, Endicott
(40-60)Sn-(>40)Bi(0.05-1)Ag
YS 12,070 psi, UTS 13,450 psi.33
YS 9287 psi, UTS 10,130 psi.33
Lucent71
Sn-Bi-Ag-Cu
86.6Sn-10Bi-2.8Ag0.6Cu
Ono
90Sn-7.5Bi-2Ag0.5Cu
193/21333
90.8Sn-5Bi-3.5Ag0.7Cu
198/213
91.0Sn-4.5Bi-3.5Ag1.0Cu
210
93.3Sn-3.1Bi-3.1Ag0.5Cu
209/212
YS 12,370 psi, UTS 13,440 psi.33
Low elongation, 12%.33
Alloy H, Alpha Metals, developed at ITRI; NCMS33 NCMS33 Senju
Low cycle fatigue life 1.8 times that of Sn63.13
H-Technol
12.23
94.25Sn-3Bi-2Ag0.75Cu
205/21733
NCMS
Sn-Bi-Ag-Cu-Ge
93.4Sn-4Bi-2Ag0.5Cu-0.1Ge
202/217
NCMS33
Sn-Bi-Ag-In
Balance Sn-(6-14)Bi(3-4)Ag-(2-5)In
Samsung72
Sn-Bi-Ag-In-Cu
(83-92)Sn-(5-18)Bi(2.5-4)Ag-(0-1.5)In(0-0.7)Cu
Matsushita59
Sn-Bi-Au
Balance Sn-(3070)Bi-Au bump
Motorola73
Sn-Bi-Cu
50Sn-48Bi-2Cu
Sn-Bi-Cu-Ag
48Sn-46Bi-4Cu-2Ag 90Sn-7.5Bi-2Ag0.5Cu
138/15333 137/146
YS 8899 psi, UTS 9495 psi.33
Low elongation, 19%,33 wide pasty range.
NCMS
YS 9806 psi, UTS 10,070 psi.33
Low elongation, 3%.33 Poor ductility.
IBM, NCMS NCMS
(Continues)
TABLE 12.7 Pb-Free Solders Investigated (Continued)
Alloy category
Composition
Solidus/ liquidus (°C)
Advantages
Disadvantages
Manufacturer or investigator Cookson74
12.24
Sn-Bi-Cu-Ag-P
(0.08-20)Bi-(0.021.5)Cu-(0.01-1.5)Ag(0-0.10)P-(0-0.2) rare earth mixture— balance Sn
Sn-Bi-In-Ag
80Sn-11.2Bi-5.5In3.3Ag
170/22133
Wide pasty range.
NCMS
80.8Sn-11.2Bi-5.5In2.5Ag
169/20033
Wide pasty range.
NCMS
Sn-Bi-Zn
65.5Sn-31.5Bi-3Zn
133/17133
Wide pasty range.
NCMS, Alpha
Sn-Bi-Zn-Sb-MgAl-Te
Balance Sn-(5-15)Bi(0.01-3)Zn-(0.013)Sb-(0.01-3)Mg(0.01-3)Al-(0.013)Te
Sn-Cd
67.8Sn-32.2Cd
Sn-Cu
Sn-Cu-Ag
97Sn-3Cu
Korea Institute of Machinery and Metals39
177
Toxic. 42
227/335
99Sn-1Cu
227
99.3Sn-0.7Cu
227
95.5Sn-4Cu-0.5Ag
Elongation 53%, UTS 11,210 psi, YS 10,500 psi.33
218/22633
UTS 6420 psi.
SS 29.8 MPa.46 σ 461 dyn/cm.34
Indium Ford, Welco Castings42
Fatigue resistance 0.3 times that of Sn63.13 UTS 0.5 times that of Sn63.75 Melting range too wide and too high. YS 3724 psi, UTS 4312 psi. Low elongation, 27%.33
Engelhard (Silvabrite 100), NCMS
93Sn-4Cu-3Ag
221/>300
YS 6276 psi, UTS 7006 psi.33
Low elongation, 22%. 95°C pasty range, liquidus >300°C.33
NCMS
12.25
(92-99)Sn-(0.7-6)Cu(0.05-3)Ag
Engelhard76
Sn-Cu-Ag-Bi-Se
(79-97)Sn-(3015)Cu(0-4)Ag-(0-1)Bi-(01)Se
Touchstone77
Sn-Cu-Ag-Ni
(92.5-96.9)Sn-(35)Cu-(0-5)Ag-(0.12)Ni
Harris78
Sn-Cu-Bi-Ag
(88-99.35)Sn-(0.56)Cu-(0.1-3)Bi-(0.053)Ag
Oaley79
Sn-Cu-In-Ag
(80-81)Sn-(1012)Cu-(5-6)In-(24)Ag
IBM80
Sn-Cu-Se-Te
Balance Sn-(3-6)Cu(0.1-1)Se-(0.1-1)Te
Tanacorp81
Sn-Cu-Sb-Ag
95.5Sn-3Cu-1Sb0.5Ag
256
Melting point too high.
Motorola
97Sn-2Cu-0.8Sb0.2Ag
219/23033
YS 3758 psi, UTS 4323 psi. Low elongation, 27%.33
Kester SAFA-LLOY, NCMS
Sn-Cu-Zn-Ag-Ni
Kale Sadashiv S82
95.68Sn-(2.8-3.5)Cu(0.2-0.5)Zn-(0.080.16)Ag-(0.080.16)Ni
Sn-In
58Sn-42In
118/145
Wide pasty range, high In.
70Sn-30In
120/∼175
Poor creep.
Indium
(Continues)
TABLE 12.7 Pb-Free Solders Investigated (Continued)
Alloy category Sn-In-Ag
Composition 77.2Sn-20.0In-2.8Ag
Solidus/ liquidus (°C)
Creep resistant; virtually drop-in replacement.
Disadvantages
Indium, NCMS
113/24233
Low solidus, very wide pasty range.
NCMS
80Sn-14.4In-5.6Ag
189/19933
High In content.
96.9Sn-10In-3.1Ag
204/205
82Sn-15In-3Ag
NCMS No Sn-In eutectic problem, potential use for flip chip applications.83
NCMS33 Indium
12.26
86.4Sn-8.6In-5Ag
NCMS
(70-92)Sn-(4-35)In(1-6)Ag
Indium83
(71.5-91.9)Sn-(4.825.9)In-(2.6-3.3)Ag
Indium83
(70.5-73.5)Snbalance In-(6.57.5)Ag
IBM84
Sn-In-Ag-Bi
91.5Sn-4In-3.5Ag1Bi
Sn-In-Ag-Cu
85.9Sn-10In-3.1Ag1Cu 88.5Sn-8In-3Ag0.5Cu
Sn-In-Ag-Sb
Manufacturer or investigator
Slightly expensive. 114°C small peak due to eutectic Sn-In. σ 390 dyn/cm.35
73.2Sn-20In-6.8Ag
175/186
Advantages
85.7Sn-10.9In-3Ag0.4Sb 88.5Sn-10.0In1.0Ag-0.5Sb
208/213
196/202 201/ 217.6 211
Low cycle fatigue life 3.3 times that of Sn63.13
Low cycle fatigue life 5.3 times that of Sn63.13
H-Technol Joints may deform due to phase change at temperature cycling.
Delphi Delco85,86
Low yield strength.
H-Technol87 Qualitek, NCMS Qualitek
86.4Sn-8.6In-5Ag2Sb Sn-In-Bi
200/20533 UTS 6938 psi.33
70Sn-20In-10Bi
Meets NCMS acceptance criteria.
NCMS33
Low elongation, 4%. High In content.33
NCMS NIST33
82Sn-15In-3Bi
113
80Sn-10In-10Bi
153/199 or 170/20033
Wide pasty range.
NCMS33
85Sn-10In-5Bi 90Sn-8In-2Bi
33
206/215
High strength. YS 7160 psi, UTS 7970 psi.33
Melting point too high. Low elongation, 25%; high In content.33
80Sn-10In-9.5Bi0.5Ag
179/201
Creep and fatigue resistant.
Slightly expensive.
YS 14,560 psi, UTS 15,380 psi.33
Low elongation, 7%.33 32°C pasty range, high In content.
Ford NCMS33
12.27
82Sn-10In-5Bi-3Ag 78.4Sn-9.8In-9.8Bi2Ag
IBM,88 NCMS
IBM89
(70-90)Sn-(8-20)In(2-10)Bi Sn-In-Bi-Ag
IBM, NCMS
163/19533
IBM,90 NCMS
80Sn-(5-14.5)In-(4.514.5)Bi-0.5Ag
Ford91
Sn-In-Bi-Ag-Cu
(86-97)Sn-(0-9.3)In(0-4.8)Bi-(0.34.5)Ag-(0-0.5)Cuintermetallic filler
U.S. Army92
Sn-In-Cu
93.3Sn-6In-0.7Cu
213/217
Low cycle fatigue life 2.1 times that of Sn63. YM 1.5 times that of Sn63. UTS 1.2 times that of Sn63.75
H-Technol75
94.3Sn-5In-0.7Cu
213/217
Low cycle fatigue life 1.4 times that of Sn63. YM 2.1 times that of Sn63. UTS 1.1 times that of Sn63.75
H-Technol75
(Continues)
TABLE 12.7 Pb-Free Solders Investigated (Continued)
Alloy category Sn-In-Cu-Ga
12.28 Sn-In-Zn
Sn-Sb
Sn-Sb-Ag-Bi
Composition
Solidus/ liquidus (°C)
Advantages
Disadvantages Low cycle fatigue life 0.8 times that of Sn63.
Manufacturer or investigator H-Technol75
92.5Sn-6In-1Cu0.5Ga
213/217
YM 1.7 times that of Sn63. UTS 1.2 times that of Sn63.75
92.8Sn-6In-0.7Cu0.5Ga
210/215
Low cycle fatigue life 3 times that of Sn63. YM 1.7 times that of Sn63. UTS 1.2 times that of Sn63.75
H-Technol13,75
93Sn-6In-0.5Cu0.5Ga
209/214
Low cycle fatigue life 1.7 times that of Sn63. YM 1.6 times that of Sn63. UTS 1.3 times that of Sn63.75
H-Technol75
94.5Sn-4In-1Cu0.5Ga
215/218
Low cycle fatigue life 1.2 times that of Sn63. YM 1.6 times that of Sn63.
UTS 0.9 times that of Sn63.75
H-Technol75
94.8Sn-4In-0.7Cu0.5Ga
215/218
Low cycle fatigue life 1.8 times that of Sn63. YM 1.4 times that of Sn63.
UTS 0.85 times that of Sn63.75
H-Technol75
95Sn-4In-0.5Cu0.5Ga
215/219
Low cycle fatigue life comparable with that of Sn63. YM 1.4 times that of Sn63.
UTS 0.96 times that of Sn63.75
H-Technol75
77.2Sn-20In-2.8Zn
106/18033
YS 5095 psi, UTS 5381 psi.33
Low elongation, 31%.33 Wide pasty range.
NCMS
83.6Sn-8.8In-7.6Zn
178/19533
YS 6033 psi, UTS 6445 psi.33
Low elongation, 14%.33
NCMS
Creep resistant; good hightemperature shear; mechanically strong. UTS 5110 psi, YM 44.5 GPa.33 UTS 5900 psi, SS 6200 psi.42
Melting point too high, poor wetting. YS 3720 psi. Low elongation, 22%.33
Motorola
95Sn-5Sb
232/240
97Sn-3Sb
232/238
99Sn-1Sb
232/235
90Sn-7.5Sb-2Ag-
229/23833
0.5Bi
Indium Indium YS 8230 psi, UTS 8773 PSI.33
Low elongation, 19%.33
Alpha
Sn-Sb-Ag-Cu Sn-Sb-Bi-Ag
Sn-Sb-Bi-Cu
88.9Sn-5Sb-4.5Ag-
Sandia alloy,
1.6Cu
NCMS33
(90-95)Sn-(3-5)Sb(1-4.5)Bi-(0.1-0.5)Ag
Willard Industries93
95Sn-3.9Sb-1Bi0.1Ag
Plumbing solder.
NCMS33
95Sn-3Sb-1.5Bi0.5Ag
Plumbing solder.
NCMS33
Low elongation, 28%.33
NCMS
93.5Sn-3Sb-2Bi1.5Cu
225/23133
YS 7343 psi, UTS 9350 psi.33
IBM94
(93-94)Sn-(2.53.5)Sb-(1.5-2.5)Bi(1-2)Cu
12.29
Sn-Sb-Cu
95Sn-3Sb-2Cu
227/234
Low cycle fatigue life 1.9 times that of Sn63.13
H-Technol
Sn-Sb-Cu-Ag
94.5Sn-3Sb-2Cu0.5Ag
218/232
Low cycle fatigue life 2 times that of Sn63.
H-Technol13
Sn-Sb-Cu-Ag-Ni
(87-92.9)Sn-(4-6)Sb(3-5)Cu-(0-0.5)Ag(0-2)Ni
Harris78
Balance Sn-(0.752)Sb-(0.05-0.6)Ag(0.05-0.6)Cu-(0.050.6)Ni
Johnson95
Sn-Sb-Zn-Ag
(90-98.5)Sn-(0.54)Sb-(0.5-4)Zn-(0.52)Ag
Harris96
Sn-Sb-Zn-Ag-Cu
(86.8-98.8)Sn-(0.54)Sb-(0.5-4)Zn-(0.13)Ag-(0.1-2)Cu
Harris60
(Continues)
TABLE 12.7 Pb-Free Solders Investigated (Continued)
Alloy category Sn-Zn
Composition 91Sn-9Zn
Solidus/ liquidus (°C) 199
Advantages Good strength; abundant. YS 7478 psi, UTS 7708 psi.33
Disadvantages
12.30
Poor corrosion resistance and wetting; high drossing. Low elongation, 27%.33 σ 487 dyn/cm.34
Manufacturer or investigator Indium, NCMS
Balance Sn-(4-12)Zn
Motorola47,97
Sn-Zn-Ag
(59-82)Sn-(16-30)Zn(2-11)Ag
Lucent98
Sn-Zn-Bi
89Sn-8Zn-3Bi
192/19733
88Sn-7Zn-5Bi
185/194
Matsushita, Senju, Showa Denko99 Zn drossing.
NCMS, Alpha
Sn-Zn-Bi-Cu
Balance Sn-(7-9)Zn(70)Sn-(6-10)Zn-(310)In-( 63Sn-37Pb > 58Bi-42Sn > 60Sn-40Pb > 70Sn-30In > 60In-40Sn at room temperature. 96.5Sn-3.5Ag absorbed considerably more strain before failure than 63Sn-37Pb. The acceleration factor in a thermal cycling test versus field service will be greater for 96.5Sn-3.5Ag than for 63Sn-37Pb.22 96.5Sn-3.5Ag has far superior roomtemperature isothermal fatigue behavior to 63Sn-37Pb at high shear strain amplitudes, due to the resistance of 96.5Sn-3.5Ag to fatigue crack initiation, but is far inferior to 63Sn-37Pb at low strain amplitudes.22,26 The creep activation energy of the 96.5Sn-3.5Ag is higher for equivalent stresses and temperatures than the value of the eutectic tin-lead alloy, as reported by Villain.27
FIGURE 13.1 Comparison of elongation of 96.5Sn-3.5Ag and 63Sn37Pb as functions of strain rate at room temperature.
PREVAILING LEAD-FREE ALLOYS
FIGURE 13.2 Tensile test results of several solder alloys (96.5Sn-3.5Ag, 99.3Sn-0.7Cu, 93.3Sn-3.1Ag-3.1Bi-0.5Cu, and 63Sn-37Cu). The test condition is 6.56 × 10−4/s and 300 K.
FIGURE 13.3 Creep behavior of several eutectic or near-eutectic solder alloys.
13.5
13.6
CHAPTER THIRTEEN
TABLE 13.3 Creep Strength* of Several Lead-Free Solders
Temperature 20°C 100°C
Sn
96.5Sn-3.5Ag
99.3Sn-0.7Cu
95.8Sn-3.5Ag0.7Cu
95.5Sn-3.8Ag0.7Cu
3.3
13.7
8.6
13
13
5
2.1
5
5
1 2
* Creep strength (N/mm ) determined at 0.1 mm/min.
13.1.3
Wetting Properties
The wetting times determined with the use of wetting balance for several solder alloys are shown in Fig. 13.6, using an unactivated flux and copper coupons.28 The wetting ability descends in the following order: eutectic SnPb > SnAgCu > SnAg > SnCu under both air and nitrogen when tested at the same temperature, with nitrogen atmosphere always yielding a shorter wetting time than air atmosphere.28 On the other hand, Glazer reported that the wetting ability in three out of four cases decreases in the following order: 60Sn-40Pb > 100Sn > 95.5Sn-4Ag-0.5Cu > 95Sn5Sb > 96.5Sn-3.5Ag when tested at 260 to 280°C, as shown in Fig. 13.7 in a contact angle study and in Fig. 13.8 in a wetting time study.22,28 The difference in wetting time as a function of superheat temperature due to a variation in alloys or atmosphere diminishes if an activated flux is used on copper sheet, as shown in Fig. 13.9.28 However, it should be noted that while this may be true under certain conditions for wetting balance test results, considerable difference among those alloys is reported for solder paste reflow performance by Huang and Lee.29 On a scale of 0.0 to 10 for full spreading, the wettability of alloys at reflow increases in the following order: 89Sn-8Zn-3Bi (0.5) < 96.5Sn-3.5Ag (4.6) < 95Sn5Sb (4.7) < 99.3Sn-0.7Cu (5.2), 96.2Sn-2.5Ag-0.8Cu-0.5Sb (5.2) < 93.6Sn-4.7Ag1.7Cu (5.3) < 95.5Sn-3.8Ag-0.7Cu (5.4) < 58Bi-42Sn (6.0) < 91.7Sn-3.5Ag-4.8Bi (6.8) < 90.5Sn-7.5Bi-2Ag (7.0) < 63Sn-37Pb (9.8). The wetting behavior of alloys is affected by surface finishes as well.Therefore, for easily wettable surfaces, such as Sn-Pb surface finish on a small-outline integrated circuit (SOIC), both eutectic Sn-Ag and eutectic Sn-Pb exhibit virtually identical wet-
FIGURE 13.4 Creep-rupture data for several candidate lead-free alloys compared to 60Sn-40Pb at 25°C.
PREVAILING LEAD-FREE ALLOYS
13.7
FIGURE 13.5 Creep-rupture data for several candidate lead-free alloys compared to 60Sn-40Pb at 100°C.
ting time under air at the same superheat using an activated flux. However, if the surface is less wettable, such as a Pd-Ni surface finish, the wetting time for eutectic SnAg solder becomes considerably longer even if an activated flux is used, while that of Sn-Pb solder still remains about the same as that of the Sn-Pb surface finish.28 The wetting time of eutectic Sn-Ag may also be shorter than eutectic Sn-Pb, depending on the test condition, as shown in Fig. 13.10.30 At 260°C, the wetting time descends in the following order: 96Sn-2.5Ag-1Bi-0.5Cu > 96.2Sn-2.5Ag-0.5Sb0.8Cu > 63Sn-37Pb > 99.3Sn-0.7Cu > 96.5Sn-3.5Ag > 95.5Sn-4Ag-0.5Cu. Melton also reported that 96.5Sn-3.5Ag wets better than 63Sn-37Pb and is less sensitive to reflow atmosphere than 63Sn-37Pb.31
FIGURE 13.6 Wetting times as a function of temperature using copper coupons with a range of solder alloys and unactivated flux. (a) Air and (b) nitrogen.28
13.8
CHAPTER THIRTEEN
FIGURE 13.7 Contact angle of 60Sn-40Pb, 100Sn, 95Sn-5Sb, 96.5Sn-3.5Ag, and 95.5Sn-4Ag-0.5Cu at 260 to 280°C using four different fluxes.
Loomans et al. studied the contact angle of multicomponent lead-free solders and reported that for binary eutectic solders, the contact angles using rosin-IPA flux are: Sn-Bi, 40° (166°C); Sn-Zn, 60° (225°C); Sn-Ag, 45° (250°C).32 Vianco et al. reported the following contact angle values: 96.5Sn-3.5Ag, 60 to 75°; 95Sn-5Sb and 95.5Sn-4Cu-0.5Ag, 35 to 55°; 60Sn-40Pb, 20 to 35°. The high contact angle of 96.5Sn3.5Ag is probably related to the high surface tension of Ag,33 as well as the inability of flux to significantly lower the solder-flux interfacial tension.34 Here, 96.5Sn-3.5Ag was also noted to have a slower wetting than the rest of the alloys. The poor wetting of 96.5Sn-3.5Ag is consistent with the observation of Melton et al.35 In that study, compared with 63Sn-37Pb, the wetting of Sn on Cu is superior, 95.5Sn-4Cu-0.5Ag and eutectic Sn-Bi is acceptable, while 96.5Sn-3.5Ag is quite poor. Wetting of 96.5Sn-3.5Ag did not improve significantly in inert atmosphere, perhaps because Ag is not readily oxidized. On a Ni-Au-plated substrate, 96.5Sn3.5Ag and 58Bi-42Sn are acceptable on wetting, but poorer than 63Sn-37Pb.22 On thick film Au76-Pt21-Pd3 over Al2O3, eutectic Sn-Ag also wets more poorly than eutectic Sn-Pb.36
FIGURE 13.8 Wetting time of 60Sn-40Pb, 100Sn, 95Sn-5Sb, 96.5Sn-3.5Ag, and 95.5Sn-4Ag-0.5Cu at 260 to 280°C using four different fluxes.
13.9
PREVAILING LEAD-FREE ALLOYS
(a)
(b)
FIGURE 13.9 Wetting times as a function of superheat using copper coupons with a range of solder alloys and 0.5 percent activated flux. (a) Air and (b) nitrogen.
FIGURE 13.10 Effect of solder alloy and solder temperature on wetting time on oxidized copper.
13.10
CHAPTER THIRTEEN
Suganuma reported the wetting area decreases in the following order: 63Sn37Pb > 96.5Sn-3.5Ag > 75Sn-25Bi > 100Sn > 91Sn-9Zn. Wetting area of Sn increases with increasing doping level (up to 4 percent) of Ag, but decreases with increasing doping level (up to 9 percent) of Zn.37 The relative wetting performance of 96.5Sn3.5Ag and 100Sn contradicts the observation of Melton,35 as described earlier. The relative poorer wetting performance of eutectic Sn-Ag versus Sn-Pb is also reflected in the capillary flow test, as reported by Vianco et al.38 Here the capillary rise of 96.5Sn-3.5Ag (2.0 cm) is lower than that of 60Sn-40Pb (2.8 cm) at the 0.025cm gap, and is 1.8 cm versus infinity at the 0.008-cm gap. The rise rate (dyne/s) is 29 versus 32 and is consistent with the rise data. However, the void area (percent) of 96.5Sn-3.5Ag appears to be equal or smaller than 60Sn-40Pb, as indicated by 3.7 versus 3.8 at the 0.025-cm gap, and 11.9 versus 14.9 at the 0.008-cm gap. For 96.5Sn-3.5Ag and 63Sn-37Pb, at a lower soldering temperature Cu6Sn5 formation dominates, while at a higher temperature the Cu3Sn layer is much thicker. Activation energy for Cu3Sn growth is 58 kJ mol−1, and for the total compound layer it is 21 kJ mol−1.22 Au dissolves more rapidly into eutectic Sn-Ag and Sn than into eutectic Sn-Pb solder for the same amount of superheating.22,39–42
13.1.4
RELIABILITY
Eutectic Sn-Ag is more tolerant of Au than eutectic Sn-Pb. 96.5Sn-3.5Ag containing 5 percent Au is ductile and elongation deteriorated only very little due to much smaller AuSn4 intermetallic compound (IMC) grain size. 63Sn-37Pb containing 5% Au is brittle and the elongation decreased dramatically.22,39,40 96.5Sn-3.5Ag on Cu with thin IMC layers fractured at or near the solder-Cu6Sn5 layer. For joints with thicker IMC layers, fracture occurred at the Cu6Sn5-Cu3Sn interface.43 Glazer reported that for 95Sn-5Ag, no microstructural coarsening occurred and only the IMC layer thickness increased. Cracks propagated at the solder-intermetallic interface and through the solder. However, no complete failures were observed after 70 cycles.22 The fatigue test results indicate that the fatigue resistance of alloys can be ranked in increasing order: 63Sn-37Pb < 64Sn-36In < 58Bi-42Sn < 50Sn-50In < 99.25Sn-0.75Cu < 100Sn < 96Sn-4Ag.22 The microstructures of various Pb-free solder-Cu interfaces has been examined primarily by Suganuma.37 Most Sn alloys, including pure Sn, Sn-Ag, Sn-Bi, or their ternary alloys, form two IMCs at the interfaces with Cu [i.e., Cu6Sn5 (15 µm) and Cu3Sn (5 µm)]. The former is much thicker than the latter, and the interface integrity is strongly influenced by the presence of the Cu6Sn5 layer.37 A study of Siow et al. showed that 63Sn-37Pb solder joint has a higher toughness than that of 96.5Sn-3.5Ag. This trend could be attributed to the sharp Ag-rich phases present in the latter. Both types of solder joints failed by fracture through the solder instead of yielding, though there are signs of local plasticity. The failure mode was primarily microvoid nucleation and coalescence. Equiaxed dimples were observed in the fracture surface of the mode I loaded sample while elongated dimples were observed in the fracture surface of the mixed-mode loaded sample. Both solders preferred to fail in the shear mode.44 The temperature cycling performance of Pb-free solder joints has been reported in a number of works. Table 13.4 shows the brief summary on results of those work. Therefore, depending on the applications, the reliability of eutectic Sn-Ag may range from the best to the worst when compared with other alloys studied in those works. The performance of eutectic Sn-Ag appears to be comparable with eutectic Sn-Pb, with three cases being better46,48 and four cases being poorer14,50–52 than Sn-Pb.
TABLE 13.4 Temperature Cycling Performance of Sn-Ag-Bi, Sn-Ag-Bi-In, 96.5Sn-3.5Ag, 99.3Sn-0.7Cu, and SnPb—The Performance May Also Be Ranked in Descending Order for Some Works, with 1st Being the Best
Test condition
Application
SnPb
13.11
Pull strength, −40/+85°C
QFP with 90Sn10 Pb lead finish
Pull strength (Kgf), −40/+85°C
QFP
2nd; 3.1/0 cycle, 2.1/200 cycle, 1.9/500 cycle
−40/+80°C, crack
SMD
3rd
0/+100°C 10,000 cycles, shear strength
SMD
2nd
−50/+150°C
PBGA
1×
0/100°C, 30-min cycle
CBGA
Poorest
96.5Sn-3.5Ag
1st
99.3Sn-0.7Cu
Sn-Ag-Bi
Sn-Ag-Bi-In
Notes/ References
2nd (Sn-Ag-3Bi), ∼1st; 3rd (Sn-Ag-6Bi, or -10Bi, or -15Bi)
Ref. 45
1st (93.5Sn-3.5Ag3Bi) 2/0 cycle, 2.1/200 cycles, 2.2/500 cycles on Pd-Ni
Ref. 45
2nd
Ref. 46 1st (91.84Sn-3.33Ag4.83Bi), no electrical failure, shear strength higher than fresh Sn63
>2× Sn-Pb (but high modulus causes pad trace fracture)
Ref. 47
Ref. 48
96.5Sn-3.5Ag-3Bi ∼ 95.5Sn-3.8Ag-0.7Cu > 63Sn-37Pb
Ref. 49
TABLE 13.4 Temperature Cycling Performance of Sn-Ag-Bi, Sn-Ag-Bi-In, 96.5Sn-3.5Ag, 99.3Sn-0.7Cu, and SnPb—The Performance May Also Be Ranked in Descending Order for Some Works, with 1st Being the Best (Continued)
Test condition
Application
SnPb
96.5Sn-3.5Ag
99.3Sn-0.7Cu
Sn-Ag-Bi
Sn-Ag-Bi-In
Notes/ References
13.12
0/100°C, 60-min cycle
CBGA
Poorest
96.5Sn-3.5Ag-3Bi ∼ 95.5Sn-3.8Ag-0.7Cu > 63Sn-37Pb
Ref. 49
0/100°C, 240min cycle
CBGA
Poorest
96.5Sn-3.5Ag-3Bi > 95.5Sn-3.8Ag-0.7Cu ≥ 63Sn-37Pb
Ref. 49
−40/+125°C, 42min cycle, fatigue life
CBGA
1×
96.5Sn-3.5Ag-3Bi ≥ 95.5Sn-3.8Ag-0.7Cu ≥ 63Sn-37Pb
Ref. 49
−40/+125°C, 240-min cycle, fatigue life
CBGA
1×, poorer than 95.5Sn3.8Ag0.7Cu
96.5Sn-3.5Ag-3Bi slightly poorer than 63Sn-37Pb
Ref. 49
−40/+60°C, 30min cycle, fatigue life
CBGA
Poorest
95.5Sn-3.8Ag-0.7Cu ≥ 96.5Sn-3.5Ag-3Bi > 63Sn-37Pb
Ref. 49
−50/+150°C, fatigue life
PBGA
Sn-Ag ball/Sn-Pb paste > Sn-Ag ball/SAC paste Ⰷ Sn62 ball/SAC paste > Sn62 ball/Sn63 paste
Ref. 48
−40/+125°C, fatigue life
PBGA
Sn-Ag ball/Sn63 paste > Sn-Ag ball/SAC paste Ⰷ Sn62 ball/SAC paste, Sn62 ball/Sn63 paste
Ref. 48
−40/+125°C, fatigue life
FlexBGA Flip chip
6th–7th
8th
2nd
Ref. 50* Best on cracking
13.13
Fatigue life
Flip chip, unfilled
Fatigue life
Flip chip
0/+100°C
General
−40/+85°C, tensile test
General
1st
−40/+85°C, shear test
General
2nd ∼1st
Fatigue cycle life
General
1st
Fatigue cycle life
General
3rd
Ref. 14
30–60% of Sn63
2nd 2nd (on NiP-Au)
Ref. 51
3rd
1st
Ref. 52
3rd
1st
Ref. 14 2nd (Sn-Ag-3Bi), ∼1st
Ref. 45
1st (Sn-Ag-3Bi)
Ref. 45
2nd, close to 1st
Ref. 53 4th (3Bi), 7th (4.8Bi), 8th (7.5Bi)
2nd (2.5In2.5Bi), 6th (3In0.5Bi)
Refs. 54 and 55, with Sn-Ag-Cu best in cycle life
* CBGA, ceramic ball grid array; PBGA, plastic ball grid array; QFP, quad flat pack; SAC, Sn-Ag-Cu; SMD, surface-mount device. † Fatigue life: 1st (3.5Ag-1.5In), 2nd (Sn-2.5Ag-0.8Cu-0.5Sb), 3rd (Sn-4Ag-1Cu); 4th (Sn-4.6Ag-1.6Cu-1Sb-1Bi), 5th (Sn-4Ag-0.5Cu), 6th (Sn-3.4Ag-1Cu-3.3Bi) ∼ 7th.
13.14
CHAPTER THIRTEEN
Plastic ball grid arrays (PBGAs) with eutectic Sn-Ag spheres performed a minimum of two times better than Sn-Pb-Ag spheres in two automotive thermal cycling conditions, −50 to 150°C and −40 to 125°C,48 as shown in Figs. 13.11 and 13.12. The higher modulus of the Sn-Ag solder balls appeared to put more stress on the traces connected to non-solder-mask-defined (NSMD) motherboard pads resulting in fractures and opens, but only in the most severe cycling condition of −50 to 150°C. For Pb-free spheres, the Sn-Ni IMC appeared to grow at the same rate in both 125 and 150°C. whereas the growth rate in the Pb-containing PBGAs baked at 150°C was 2.6 times of that at 125°C. Shangguan reported that 96.5Sn-3.5Ag eutectic solder has superior overall properties and is suitable for solder interconnects in thick-film automotive electronics packages when used with a mixed bonded Ag conductor.56 For 0.4-mm-pitch surface-mount technology (SMT) assembly applications,Artaki et al. concluded that eutectic Sn-Ag, together with eutectic Sn-Bi and 91.8Sn-4.8Bi-3.4Ag, 77.2Sn-20In2.8Ag, and 96.2Sn-2.5Ag-0.8Cu-0.5Sb, are all feasible, although with narrower processing windows.57 This narrow processing window is concurred by Yang et al.,58 who studied the effect of processing conditions on joint quality, and concluded that low soldering temperatures, fast cooling rates, and short renew times are suggested for producing joints with the best shear strength, ductility, and creep resistance.
13.2 13.2.1
EUTECTIC Sn-Cu Physical Properties
Some physical properties of 99.3Sn-0.7Cu can be found in Table 13.1. The melting temperature of eutectic Sn-Cu is the highest among the prevailing Pb-free solders, suggesting a greater difficulty in adopting this alloy. Its surface tension, electrical resistivity, and density are comparable with eutectic Sn-Ag, presumably due to the high content of Sn in both alloys.
13.2.2
MECHANICAL PROPERTIES
The tensile and shear properties of 99.3Sn-0.7Cu are shown in Table 13.2 and Fig. 13.2. Eutectic Sn-Cu is lower in tensile strength but higher in elongation than both eutectic Sn-Ag and Sn-Pb, reflecting the softness and ductility of Sn-Cu. On the other hand, shear strength of Sn-Cu appears to be comparable with Sn-Pb, but lower than Sn-Ag. The creep strength of eutectic Sn-Cu is higher than 100Sn, but lower than eutectic Sn-Ag and Sn-Ag-Cu at both 20 and 100°C, as shown in Table 13.3. The data are consistent with the creep-rupture data shown in Figs. 13.4 and 13.5, where the time to rupture increases in the following order: eutectic Sn-Ag, Sn-Ag-Cu < eutectic SnCu < 60Sn-40Pb at 25 and 100°C.
13.2.3
WETTING PROPERTIES
The wetting properties of eutectic Sn-Cu and eutectic Sn-Ag are considered by Vincent et al. as having a great potential as replacements for Sn-Pb in wave and reflow processes.59 Wetting balance test results by Hunt et al., as shown in Fig. 13.6, indicate
13.15 FIGURE 13.11 Two-parameter Weibull plot as of 5619 thermal cycles of −50 to 150°C. No solder joint failures had been recorded on the configuration with SnAg solder balls assembled with Sn-Pb paste.
13.16 FIGURE 13.12 Two-parameter Weibull plot as of 9187 thermal cycles of −40 to 125°C. Only one failure had been recorded on the configuration with Sn-Ag solder balls assembled with Sn-Pb paste, and this occurred at 7555 cycles.
PREVAILING LEAD-FREE ALLOYS
13.17
that the wetting ability decreases in the following order: eutectic Sn-Pb > Sn-AgCu > Sn-Ag > Sn-Cu when an unactivated flux is used.28 The difference in wetting vanishes when an activated flux is used and when the wetting time is plotted against superheating, as shown in Fig. 13.9.28 However, depending on the test conditions of the wetting balance test, wetting time of eutectic Sn-Cu may also be shorter than eutectic Sn-Pb, as demonstrated in Fig. 13.10.30 In the Prismark report, Nortel found soldering quality equal to eutectic Sn-Pb in Meridian desktop telephone manufacturing. However, in air reflow the wettability was reduced, the fillet exhibited a rough and textured appearance, and the flux residue was dark brown.60 Preferably the use of eutectic Sn-Cu should be confined to wave soldering because low solder cost and inerting of waves is not costly. At reflow, on a full scale of 0 to 10, the reflow spreading of eutectic Sn-Cu (5.2/10) is better than eutectic Sn-Ag (4.6/10), but it is considerably poorer than eutectic Sn-Pb (9.8/10), as reported by Huang and Lee.29 Toyoda also studied spreading performance of several alloys, and observed the following spreading behavior in decreasing order: 63Sn-37Pb > Sn-Ag-Cu-4.5Bi, Sn-Ag-Cu-7.5Bi > Sn-3.5Ag0.75Cu > 99.25Sn-0.75Cu > 89Sn-8Zn-3Bi, as shown in Fig. 13.13.46
13.2.4
RELIABILITY
Although the tensile strength of eutectic Sn-Cu is fairly poor, the fatigue resistance is fairly good. In Glazer’s study, the fatigue resistance increases in the following order: 63Sn-37Pb < 64Sn-36In < 58Bi-42Sn < 50Sn-50In < 99.25Sn-0.75Cu < 100Sn < 96Sn-4Cu.22 However, the low-cycle isothermal fatigue (strain 0.2 percent, 0.1 Hz, R = 0.8, 300 K) performance shows a different trend, as shown in Table 13.5.4 Here the number of cycles to failure for eutectic Sn-Cu is less than one-third of that for eutectic Sn-Pb. Table 13.4 shows that for the two cases involving comparison of Sn-Cu with Sn-Pb, the former is consistently better than the latter.14,52 In addition, Syed studied the thermal cycling reliability of Pb-free solder joints for several package assemblies.50
FIGURE 13.13 Spreading performance of several Pb-free solders and eutectic Sn-Pb.
13.18
CHAPTER THIRTEEN
TABLE 13.5 Relative Performance in Fatigue Resistance of Lead-Free Solders in Low-Cycle Isothermal Fatigue Test* Melting temperature (°C)
Nf†
88.5Sn-3Ag-0.5Cu-8In
195–201
19,501
91.5Sn-3.5Ag-1Bi-4In
208–213
12,172
92.8Sn-0.7Cu-0.5Ga-6In
210–215
10,800
95.4Sn-3.1Ag-1.5Cu
216–217
8,936
96.2Sn-2.5Ag-0.8Cu-0.5Sb
216–219
8,751
95.5Sn-3.5Ag-1Bi
219–220
8,129
94.5Sn-0.5Ag-2Cu-3Sb
218–232
7,120
95Sn-2Cu-3Sb
227–234
6,821
93.3Sn-3.1Ag-3.1Bi-0.5Cu
209–212
6,522
96.5Sn-0.5Ag-3Bi
223–226
4,283
221
4,186
Alloy
96.5Sn-3.5Ag 92Sn-3.3Ag-4.7Bi 63Sn-37Pb 91.7Sn-3.5Ag-4.8Bi 99.3Sn-0.7Cu
210–215
3,850
183
3,650
211–215
3,179
227
1,125
* Strain, 0.2 percent; 0.1 Hz; R = 0.8; 300 K. † Nf: number of cycles to failure at 300 K (50 percent load drop, 0.2 percent strain range).
For a 27-mm, 256-PBGA assembly, at 0 to 100°C cycling, no failures in any alloys were observed after 9730 cycles.At −55 to 125°C cycling, after 6830 cycles, more than 50 percent failure rate occurred in eutectic Sn-Pb, Sn-Ag, and Sn-Cu. Eight failures occurred in Sn-3.4Ag-0.7Cu (30 percent higher life performance than Sn-Pb), and one failure at 6288 cycles occurred in Sn-4Ag-0.5Cu. At −40 to 125°C cycling, after 5080 cycles, Sn-Pb and Sn-Cu just started to fail, while no failures were observed in Sn-Ag, Sn-4Ag-0.5Cu, and Sn-3.4Ag-0.7Cu. For a 12-mm, 144-flexible ball grid array (fleXBGA) assembly, at 0 to 100°C cycling, eutectic Sn-Cu is 1.4 times better in performance than Sn-Pb, as shown in Fig. 13.14.50 Sn-4Ag-0.5Cu and Sn-3.4Ag-0.7Cu are similar in performance, and both are 1.6 to 1.7 times better in performance than eutectic Sn-Pb. Sn-Ag is the best, with no failure recorded after 10,740 cycles.At −55 to 125°C cycling, eutectic Sn-Cu is better in performance than Sn-Pb, as shown in Fig. 13.15.50 Sn-4Ag-0.5Cu and Sn-3.4Ag0.7Cu are comparable, both being 20 percent better in performance than Sn-Pb. Sn-Ag again is the best, with only three failures observed, and is 1.4 times better in performance than Sn-Pb. At −40 to 125°C cycling, Sn-Cu is similar to Sn-Pb in performance, as shown in Fig. 13.16.50 Not much improvement is observed for Sn-Ag-Cu over Sn-Pb, with Sn-4Ag-0.5Cu being slightly better than Sn-3.4Ag-0.7Cu. Eutectic Sn-Ag again comes to the top and is 1.4 times better in performance than Sn-Pb. For ball grid array (BGA) assembly, eutectic Sn-Cu may seem to be inferior to eutectic Sn-Ag in temperature cycling performance; however, the opposite trend is observed for flip chip assembly, as shown in Table 13.4. In Maestrelli’s study, eutectic Sn-Cu is observed to be the best in temperature cycling (0 to 100°C) performance, with Sn-4Ag-0.5Cu and eutectic Sn-Pb being the next, while eutectic Sn-Ag turns out to be the poorest in performance, as shown in Fig. 13.17.14 The higher reli-
PREVAILING LEAD-FREE ALLOYS
FIGURE 13.14 Temperature cycling (0 to 100°C) performance for 12-mm 144fleXBGA assembly.
13.19
13.20
CHAPTER THIRTEEN
FIGURE 13.15 Temperature cycling (−55 to 125°C) performance for 12-mm 144-fleXBGA assembly.
PREVAILING LEAD-FREE ALLOYS
13.21
FIGURE 13.16 Temperature cycling (−40 to 125°C) performance for 12-mm 144-fleXBGA assembly.
13.22
CHAPTER THIRTEEN
FIGURE 13.17 Plot of fatigue life as a function of thermal strain for a variety of solders over a temperature range of 0 to 100°C.
ability of eutectic Sn-Cu in flip chip applications is attributed to its compliant nature. Figure 13.18 shows the cross-section of flip chip solder joints after the same number of thermal cycles.14 Both Sn-Pb and Sn-Ag-Cu display a fracture without solder joint deformation. Sn-Cu solder joints, on the other hand, exhibits a deformed solder interconnect.This deformation of solder material due to the compliant nature of sol-
FIGURE 13.18 Flip chip solder interconnects after the same number of thermal cycles for eutectic Sn-Pb, Sn-Cu, and Sn-Ag-Cu.
PREVAILING LEAD-FREE ALLOYS
13.23
der is considered helpful in offsetting the impact of mismatch in the coefficient of thermal expansion (CTE), and hence is the main reason for Sn-Cu to be superior in fatigue life performance under an application with a large mismatch in CTE. Frear et al. also reported that for flip chip assembly, the thermal fatigue life descends in the following order: eutectic Sn-Cu > Sn-3.8Ag-0.7Cu, eutectic Sn-Pb > eutectic Sn-Ag.52
13.3
Sn-Ag-Bi AND Sn-Ag-Bi-In
As discussed in an earlier chapter, Sn-Ag-Bi is favored by Brite Euram, Department of Trade Industry of UK, NEMI of USA, and JEIDA. Sn-Ag-Bi-In is not investigated as extensively as Sn-Ag-Bi system. However, it is recommended by JEIDA, and is one of the major Pb-free alloys used for reflow soldering in Japan, as shown in Fig. 12.10.
13.3.1
PHYSICAL AND MECHANICAL PROPERTIES
The equilibrium phase diagram of the Sn-Ag-Bi system is shown in Fig. 13.19.61 The ternary eutectic point exhibits a melting temperature around 138°C, which is fairly comparable with the binary eutectic Sn-Bi alloy. For the ternary Sn-Ag-Bi composition with a narrow pasty range and a melting temperature close to 63Sn-37Pb, the
FIGURE 13.19 Equilibrium phase diagram for Sn-Ag-Bi system.
13.24
CHAPTER THIRTEEN
desirable composition can be prescribed by the shaded area in the lower left corner. Unlike the ternary eutectic point where the composition can be narrowed down easily, the desirable Sn-Ag-Bi composition for the higher melting temperature range cannot be identified easily. The most favorable composition appears to contain 1 to 5 at % Bi and 1 to 4 at% Ag, with the balance as Sn, as shown in Table 13.6. Also shown in Table 13.6 are the physical and mechanical properties of Sn-Ag-Bi and SnAg-Bi-In alloys, with eutectic Sn-Pb included for comparison. In most instances, the data listed from the same source were determined under the same test conditions. Most of the Sn-Ag-Bi alloys exhibit a pasty range from 210 to 220°C. Few alloys have a lower solidus temperature, but a wider pasty range, such as 90.5Sn-2Ag-7.5Bi. Additional Sn-Ag-Bi melting temperature information can be found in Fig. 12.6. In all instances except for 96.5Sn-0.5Ag-3Bi, the melting temperature is higher than that of 63Sn-37Pb, but lower than eutectic Sn-Cu (227°C) or Sn-Ag (221°C), suggesting a slight advantage on soldering temperature. On the other hand, the surface tension of Sn-3.5Ag-4.8Bi is higher than 63Sn-37Pb, implying a disadvantage in solder spreading. The density of Sn-3.4Ag-4.8Bi (7.53 g/cm3) is close to that of pure Sn (7.3 gm/cm3), due to the high content of Sn. The lower density of Sn-Ag-Bi system than eutectic Sn-Pb (8.4 g/cm3) indicates that a reduced solder weight of the joints can be expected. The hardness of Sn-Ag-Bi, as demonstrated by Sn-3Ag-5Bi (29.9 HV), is considerably higher than both eutectic Sn-Pb (12.9 HV) and Sn-Ag (16.5 HV) (see Table 13.1), and may pose a greater resistance against solder deformation during temperature cycling of assemblies. Tensile strength, yield strength, and shear strength of Sn-Ag-Bi systems are significantly higher than eutectic Sn-Pb, while the elongation or plasticity is much lower than Sn-Pb. This is consistent with the high hardness of Sn-Ag-Bi, mentioned earlier. Baggio studied the Sn-Ag-Bi system for Panasonic Mini Disk Player applications.45 By replacing some Sn of 96.5Sn-3.5Ag with Bi, the pull strength decreases linearly with increasing content of Bi, while the melting point decreases rapidly first, then decreases at a slower linear rate at Bi content above 6 percent, as shown in Fig. 12.6. Figs. 13.20 and 13.21 show examples of tensile stress-strain behavior for Sn-Ag-Bi and Sn-Ag-Bi-In systems, respectively, with 63Sn-37Pb and 96.5Sn-3.5Ag included for comparison.4 Addition of In to Sn-Ag-Bi enhances either the strength or the plasticity and is considered the optimal lead-free solder by Hwang4 in the absence of Cu, if aiming for a melting temperature of less than 215°C. In the study by Tanaka et al. on lead-free solders for mobile equipment, the impact resistance of solders descends in the following order: Sn-3.5Ag-0.75Cu > Sn-3Ag-1In-0.7Cu > Sn-3.5Ag0.5Bi-3In > 63Sn-37Pb > Sn-3.2Ag-3Bi-1.1Cu-Ge, Sn-Ag-X, as shown in Fig. 13.22.5 The superior impact resistance of Sn-Ag-Bi-In is attributed to the high plasticity of solder due to the presence of In. The creep rate of eutectic Sn-Ag-Bi is slightly lower than that of Sn-Ag-Cu and of Sn-Ag, and it is much lower than that of 60Sn-40Pb and eutectic Sn-Bi at low stress, presumably due to the high hardness and high strength of the Sn-Ag-Bi system, as indicated by Fig. 13.3. In the creep rupture tests conducted by Tanaka et al., the creep rupture performance of Sn-3.5Ag-0.5Bi-3In, as well as Sn-3.5Ag-0.75Cu, Sn-3Ag1In-0.7Cu, Sn-3.2Ag-3Bi-1.1Cu-Ge, and Sn-Ag-X, are all better than 63Sn-37Pb.5
13.3.2
WETTING PROPERTIES
Presence of Bi significantly improves the solder spreading properties of lead-free solders. In the reflow soldering compatibility study done by Huang and Lee, the solder spreading performance at solder paste reflow increases in the following order: 89Sn-8Zn-3Bi (0.5) < 96.5Sn-3.5Ag (4.6) < 95Sn-5Sb (4.7) < 99.3Sn-0.7Cu (5.2),
13.25
PREVAILING LEAD-FREE ALLOYS
TABLE 13.6 Physical and Mechanical Properties of Sn-Ag-Bi, Sn-Ag-Bi-In, and 63Sn-37Pb Systems Notes/ References
Property
Sn-Ag-Bi
Sn-Ag-Bi-In
63Sn-37Pb
Melting temperature (°C)
208–217 (93.5Sn3.5Ag-3Bi) 219–220 (95.5Sn3.5Ag-1Bi) 216–220 (95Sn3Ag-2Bi) 215–221 (95.5Sn2.5Ag-2Bi) 223–226 (96.5Sn0.5Ag-3Bi) 219–220 (95.5Sn3.5Ag-1Bi) 210 (92Sn-3Ag5Bi) 202–215 (91.8Sn3.5Ag-4.8Bi) 191–215 (90.5Sn2Ag-7.5Bi) 213 (94Sn-3Ag3Bi)
—
183
Ref. 13
208–213 (91.5Sn3.5Ag-1Bi-4In)— —
—
Ref. 4
—
Ref. 13
—
—
Ref. 13
—
—
Ref. 4
—
—
Ref. 4
—
—
—
—
Ref. 29
—
—
Ref. 29
—
—
—
—
Surface tension (dyne/cm)
420 (Sn-3.5Ag4.8Bi)
—
380 (260°C)
Ref. 62
Density (g/cm3)
7.53 (Sn-3.4Ag4.8Bi)
—
8.4
Ref. 29
Electrical resistivity (µΩ-cm)
11.6 (Sn-3Ag5Bi)
17
Ref. 63
Hardness [Vickers hardness (HV), kg/mm2]
29.9 (Sn-3Ag5Bi)
—
12.9
Ref. 63
Ultimate tensile strength (MPa)
82.5 (92Sn-3.3Ag4.7Bi) 43 (95.5Sn-3.5Ag1Bi) 54.7 (Sn-3Ag-2Bi)
—
46
—
—
52.2 (Sn-2.5Ag2Bi) 92.7 (Sn-2.5Ag19.5Bi) 71.4 (Sn-3.4Ag4.8Bi) Yield strength, 0.2% (MPa)
37.7 (Sn-3Ag-2Bi) 45.5 (Sn-2.5Ag2Bi) 83.2 (Sn-2.5Ag19.5Bi) 46.3 (Sn-3.4Ag4.8Bi)
12.1 (Sn-3Ag-5Bi5In)
106.0 (Sn-2Ag9.8Bi-9.8In) —
30.6
6.56 × 10−4/s, 300 K, Ref. 4 6.56 × 10−4/s, 300 K, Ref. 4 Ref. 13
—
Ref. 13
—
—
Ref. 13
—
—
Ref. 13
27.2
Ref. 13
—
Ref. 13
—
—
Ref. 13
—
—
Ref. 13
100.4 (Sn-2Ag9.8Bi-9.8In) —
13.26
CHAPTER THIRTEEN
TABLE 13.6 Physical and Mechanical Properties of Sn-Ag-Bi, Sn-Ag-Bi-In, and 63Sn-37Pb Systems (Continued)
Property Elongation (%)
Sn-Ag-Bi 10 (92Sn-3.3Ag4.7Bi) 31 (95.5Sn-3.5Ag1Bi) 30 (Sn-3Ag-2Bi) 26 (Sn-2.5Ag-2Bi) 17 (Sn-2.5Ag19.5Bi) 16 (Sn-3.4Ag4.8Bi)
Shear strength (MPa)
81.36 (Sn-3.33Ag4.83Bi)
Impact strength (J/cm2)
—
Sn-Ag-Bi-In
63Sn-37Pb
Notes/ References
48
6.56 × 10−4/s, 300 K, Ref. 4 6.56 × 10−4/s, 300 K, Ref. 4 Ref. 13
— —
Ref. 13 Ref. 13
—
—
Ref. 13
—
40.27
Ref. 36
31
Ref. 5
13 (90Sn-3.3Ag3Bi-3.7In) 30 (91.5Sn-3.5Ag1Bi-4In) 7 (Sn-2Ag-9.8Bi9.8In) — —
48 (Sn-3.5Ag0.5Bi-3In)
31 —
96.2Sn-2.5Ag-0.8Cu-0.5Sb (5.2) < 93.6Sn-4.7Ag-1.7Cu (5.3) < 95.5Sn-3.8Ag-0.7Cu (5.4) < 58Bi-42Sn (6.0) < 91.7Sn-3.5Ag-4.8Bi (6.8) < 90.5Sn-7.5Bi-2Ag (7.0) < 63Sn37Pb (9.8).29 The figures in parentheses represent the scores of spreading, with a score of 10 representing full spreading. 63Sn-37Pb exhibits the best spreading performance. Among the Pb-free alloys, except for Zn-containing solder, all Bicontaining solders, including 58Bi-42Sn, 91.7Sn-3.5Ag-4.8Bi, and 90.5Sn-7.5Bi-2Ag, display a better wetting than non-Bi-containing alloys. Among the three Bicontaining alloys, eutectic Sn-Bi is too low in melting temperature as a drop-in replacement for 63Sn-37Pb, and Sn-Ag-Bi systems are the most attractive due to their higher melting temperature and superior wettability.
13.3.3
RELIABILITY
Bi-containing Pb-free alloys are sensitive to the presence of Pb, due to the formation of a 52Bi-30Pb-18Sn ternary eutectic structure with a melting temperature of 96°C in the solidified solder joint. The solder joints become weak in mechanical strength when subjected to thermal cycling with the temperature exceeding 96°C because the low-melting ternary eutectic phase accelerates grain growth and phase agglomeration.64 Bi-containing solders also tend to have a fillet-lifting phenomenon, particularly at the wave-soldering stage, therefore posing reliability concerns.65 Both failure mechanisms will be discussed in detail in Chap. 16. Hwang studied the isothermal low-cycle fatigue performance of a series of Pb-free solders, as shown in Fig. 13.4.4 Among the alloys investigated, Sn-Ag-Bi systems are comparable or better than 96.5Sn-3.5Ag and 63Sn-37Pb, and are considerably better than 99.3Cu-0.7Sn. The low-cycle fatigue life performance (numbers in parentheses) can be ranked as follows: 95.5Sn-3.5Ag-1Bi (8129) > 96.5Sn-0.5Ag-3Bi (4283) > 96.5Sn-3.5Ag (4186) > 92Sn-3.3Ag-4.7Bi (3850) > 63Sn-37Pb (3650) > 91.7Sn-3.5Ag4.8Bi (3179) > 99.3Sn-0.7Cu (1125).
PREVAILING LEAD-FREE ALLOYS
13.27
FIGURE 13.20 Tensile stress-strain properties of 92Sn-3.3Ag-4.7Bi, 95.5Sn-3.5Ag-1Bi, 96.5Sn3.5Ag, and 63Sn-37Pb at 300 K and 6.56 × 10−4/s.
FIGURE 13.21 Tensile stress-strain properties of 90Sn-3.3Ag-3Bi-3.7In, 91.5Sn-3.5Ag1Bi-4In, 96.5Sn-3.5Ag, and 63Sn-37Pb at 300 K and 6.56 × 10−4/s.
13.28
CHAPTER THIRTEEN
FIGURE 13.22 Charpy impact test results for several lead-free alloys and 63Sn-37Pb.
It is interesting to note that all three In-containing alloys, including Sn-Ag-Bi-In, are ranked at the top of the list, with Sn-Ag-Cu or Sn-Ag-Cu-X being equal or better than Sn-Ag-Bi systems. Presumably, the superior low-cycle fatigue performance of In-containing alloys might be attributed to the enhanced plasticity introduced by In. The temperature cycling fatigue performance of Sn-Ag-Bi and Sn-Ag-Bi-In systems are shown in Table 13.4. Again, the performance is highly dependent on applications. Therefore, in four cases,45,49,54,55 Sn-Ag-Bi systems are inferior to Sn-Pb or 62Sn-36Pb-2Ag, while in eight other cases, the trend is opposite.45,47,49 Bradley studied the temperature cycling reliability of Sn-Ag-X systems, with results shown in Fig. 13.23.54,55 In his results, all three Sn-Ag-Bi alloys are more inferior than 62Sn-36Pb-2Ag, and the performance descends with increasing Bi content: SnAg-3Bi > Sn-Ag-4.8Bi > Sn-Ag-7.5Bi. The favorable performance of lower Bi content is also observed by Baggio, as shown in Fig. 13.24.45 Here the pull strength of solder joints is plotted as a function of number of cycles. Sn-Ag-Bi systems with 3 percent Bi exhibit the best performance, and are comparable with Sn-Pb. For Sn-Ag-Bi alloys with 6 to 15 percent Bi content, the pull strength dropped considerably, although it remains stable with an increasing cycle number. The work of Vianco et al. indicates that Sn-Ag-Bi is far more superior than eutectic Sn-Pb in a thermal fatigue test.47 In their study, an evaluation was performed that
FIGURE 13.23 Reliability performance of Sn-Ag-X alternatives.
PREVAILING LEAD-FREE ALLOYS
13.29
FIGURE 13.24 Reliability of Sn-Ag-Bi and Sn-Pb systems.
examined the reliability of surface-mount solder joints made with a 91.84Sn-3.33Ag4.83Bi alloy. A 1206-chip capacitor, an SOIC gull-wing, and plastic leaded chip carrier (PLCC) J-lead solder joints were thermally cycled between 0 and 100°C for 1,000, 2,500, 5,000, or 10,000 cycles. Continuous dc signal monitoring did not reveal any open events through the 10,000-cycle mark, and there was no evidence of solder degradation for any of the evaluated solder joint geometries through 2,500 cycles. Thermal cycles of 5,000 and 10,000 caused some surface cracks and isolated throughcracks in the thinner reaches of the solder fillets, as shown in Fig. 13.25.47 However, it should be pointed out that regardless of the development of cracks, the Sn-Ag-Bi still exhibits a shear strength greater than the as-fabricated 63Sn-37Pb solder joints after 10,000 thermal cycles, as shown in Fig. 13.26.47 Bartelo reported that at the 0/100°C temperature cycling condition with cycle times from 30 to 240 min, the joint reliability of 96.5Sn-3.5Ag-3Bi is comparable or better than 95.5Sn-3.8Ag-0.7Cu, which in turn is better than 63Sn-37Pb when assembling 625 I/O CBGA with 32 × 32 × 0.8 mm dimension.49 For temperature range −40/+125°C thermal cycling test, 96.5Sn-3.5Ag-3Bi was better than 63Sn-37Pb at 42 min cycle time, but slightly poorer than Sn-Pb at a 240-min cycle time.49 For −40/60°C temperature cycling condition with 30-min cycle time, Sn-Ag-Cu is slightly better than Sn-Ag-Bi, and both are better than 63Sn-37Pb. In Bartelo’s study, the alloys used for solder ball and solder paste are the same. The superior thermal cycling fatigue performance of the Sn-Ag-Bi system perhaps can be partly attributed to its high rigidity and thus low creep rate. This high rigidity is mainly caused by two solder-strengthening mechanisms. The first is solution strengthening, achieved by dissolving 4 to 5 percent Bi into Sn. The second is precipitation strengthening, achieved by Bi and Ag3Sn particles, as shown in Fig. 13.27.47 The high concentration of Ag3Sn and Bi precipitates is also responsible for the rough and grainy solder joint appearance, as shown in Fig. 13.28.47 Addition of In to the Sn-Ag-Bi systems significantly improves the fatigue resistance, as shown in Table 13.5 for isothermal low-cycle fatigue and in Fig. 13.21 for temperature cycling performance. In the former case, the Sn-Ag-Bi-In is far superior than eutectic Sn-Pb, while in the latter case, both systems are comparable in performance.
13.30
CHAPTER THIRTEEN
FIGURE 13.25 Maximum shear load (N) versus fillet-top cracking of the 1206-chip capacitor 91.84Sn-3.33Ag-4.83Bi solder joints as a function of number of thermal cycles at condition 0/+100°C. Note that the cracks developed with an increasing number of cycles.
FIGURE 13.26 Maximum shear load (N) versus fillet-top cracking of the 1206-chip capacitor 91.84Sn-3.33Ag-4.83Bi solder solder joints as a function of number of thermal cycles at condition 0/+100°C. Note that the shear load of Sn-Ag-Bi is higher than or equal to as-fabricated 63Sn-37Pb joints, even up to 10,000 thermal cycles.
13.31
PREVAILING LEAD-FREE ALLOYS
(a)
(b)
FIGURE 13.27 Microstructure of 91.84Sn-3.33Ag-4.83Bi. (a) Scanning electron microscope (SEM) photograph; (b) thermal expansion mismatch (TEM) photograph. The solder is strengthened by (1) solution strengthening (Sn with 4 to 5 percent Bi) and (2) precipitation strengthening (Bi and Ag3Sn particles).
13.4
Sn-Ag-Cu AND Sn-Ag-Cu-X
Sn-Ag-Cu is the most prevailing choice of Pb-free alternative globalwise. Besides that, the Sn-Ag-Cu-X family (including Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb— particularly the latter two) also received quite a lot of attention and will be introduced in more detail later.
13.4.1
PHYSICAL PROPERTIES
The physical properties of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb are shown in Table 13.7. Figure 13.29 shows the Sn-Ag-Cu phase diagram, as reported by Handwerker.61 NIST experimental work showed that the ternary eutectic composition is approxi-
(a)
(b)
FIGURE 13.28 The as-fabricated solder joints made with the 91.84Sn-3.33Ag-4.83Bi solder. (a) SEM micrograph; (b) optical micrograph.
13.32
CHAPTER THIRTEEN
TABLE 13.7 Physical Properties of Sn-Ag-Cu, Sn-Ag-Cu-X, and 63Sn-37Pb
Properties Melting temperature (°C)
63Sn37Pb 183
Sn-Ag-Cu 217 (95.5Sn3.8Ag0.7Cu) 216–217 (95.4Sn3.1Ag1.5Cu)
Density (g/cm3)
8.36
8.4
Electrical resistivity (µΩ-cm)
17
Sn-AgCu-In
Sn-AgCu-Bi
Sn-AgCu-Sb
207–216 (Sn-3.3 Ag-3Bi1.1Cu) 195–201 (88.5Sn3Ag0.5Cu8In)
209–212 (93.3Sn3.1Ag3.1Bi0.5Cu)
Notes/ References Ref. 66
216–219 (Sn-2.5Ag0.8Cu0.5Sb)
Ref. 4
218–232 (94.5Sn0.5Ag-2Cu3Sb)
Ref. 4
217–244 (93.6Sn4.7Ag1.7Cu)
Ref. 67
217 (95.6Sn3.5Ag0.9Cu)
Ref. 61
217–225 (95.5Sn4Ag0.5Cu)
Ref. 68
218 (95.75Sn3.5Ag0.75Cu)
Ref. 69
220 (96.5Sn3Ag0.5Cu)
Ref. 70
7.44 (Sn4Ag0.5Cu)
7.56 (Sn2Ag0.5Cu7.5Bi)
7.39 (Sn2.5Ag0.8Cu0.5Sb)
Ref. 71
7.39 (Sn4Ag0.5Cu)
Ref. 72
7.5 (Sn3.8Ag0.7Cu)
Ref. 8
10.6 (Sn3Ag-3Cu2Bi)
12.1 (Sn2.5Ag0.8Cu0.5Sb)
Ref. 63
13.33
PREVAILING LEAD-FREE ALLOYS
TABLE 13.7 Physical Properties of Sn-Ag-Cu, Sn-Ag-Cu-X, and 63Sn-37Pb (Continued) 63Sn37Pb
Properties
14.5
Hardness
Sn-Ag-Cu
Sn-AgCu-In
Sn-AgCu-Bi
Sn-AgCu-Sb
10–15 (Sn4Cu0.5Ag)
Ref. 3
13 (Sn3.8Ag0.7Cu)
Ref. 8
12.8
[Vickers hardness (HV), kg/mm2], Ref. 6 (HV), kg/mm2), Ref. 7
10.25 (as drawn), 12.45 (annealed) Sn-4.7 Ag-1.7Cu 28.6 (Sn3Ag-2Cu2Sb)
(HV, kg/mm2), Ref. 63
10.08
18.28 (Sn2.5Ag0.8Cu0.5Sb)
Rockwell hardness, 15-W scale hardness, Ref. 11
12.2
13.5 (Sn2.5Ag0.8Cu0.5Sb)
Rockwell hardness, 16-W scale hardness, Ref. 11 Brinell hardness, Ref. 8
12.9
34.5 (Sn3Ag-3Cu2Bi)
15 (Sn3.8Ag0.7Cu) Surface tension (mN m−1)
380 at 260°C
510 (Sn2.5Ag0.8Cu0.5Sb)
417 (air), 464 (N2) at 233°C CTE (ppm)
Notes/ References
18.74
Ref. 62
Ref. 3
14.83 (Sn3Ag-4Cu)
Ref. 8
25
Ref. 2
21
Ref. 16
24 (−70°C, 20°C, 140°C)
Ref. 15
13.34
CHAPTER THIRTEEN
FIGURE 13.29 Sn-Ag-Cu phase diagram. Alloys in shaded area have a freezing range of less than 10°C.
mately 95.6Sn-3.5Ag-0.9Cu (±0.1 percent), with a melting point of 217°C. Around the ternary eutectic point, as shown by the shaded area in Fig. 13.29, there are a number of Sn-Ag-Cu compositions developed, with melting temperatures all within 10°C of the ternary eutectic melting temperature. Figure 13.30 shows several examples of differential scanning calorimeter (DSC) thermograms of such alloys.73 Although the exact ternary eutectic composition is to be determined, generally it is considered that the physical, mechanical, and soldering properties, and the reliability of Sn-AgCu compositions near this ternary eutectic point should be fairly comparable. The density and electrical resistivity of Sn-Ag-Cu and Sn-Ag-Cu-X are all comparable with each other, and are similar to other high-Sn alloys (see also Tables 13.1 and 13.6), due to the dominant presence of Sn. The hardness, however, does differ from system to system. Therefore, Sn-Ag-Cu is comparable with Sn-Pb, while Bicontaining alloys exhibit a considerably higher hardness than Sn-Pb. The high hardness can be explained by the precipitation and Bi-dissolution strengthening mechanisms, discussed in Sec. 13.3.3.
13.4.2
MECHANICAL PROPERTIES
The mechanical properties of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb are shown in Table 13.8. Examples of tensile stress-strain relation at 300 K and 6.56 × 10−4/s for Sn-Ag-Cu, Sn-Ag-Cu-In, and Sn-Ag-Cu-Bi alloys, together with 63Sn-37Pb for comparison, are shown in Figs. 13.31, 13.32, and 13.33, respectively.4 Some additional information on the time to break in creep tests for several other alloys is shown in Fig. 13.34.46 The tensile strength of Sn-Ag-Cu is slightly higher [e.g., Sn-3.8Ag-0.7Cu or Sn3.5Ag-0.7Cu (Ref. 8)] or higher [e.g., Sn-4.7Ag-1.7Cu (Ref. 4)] than eutectic Sn-Pb. In the case of composition near the ternary eutectic point, such as Sn-3.8Ag-0.7Cu,
PREVAILING LEAD-FREE ALLOYS
13.35
FIGURE 13.30 Differential scanning calorimeter thermograms of four Sn-Ag-Cu solder alloys: (a) Sn-4Ag-0.5Cu, (b) Sn-3.6Ag-1Cu, (c) Sn-3.7Ag0.9Cu, and (d) Sn-3Ag-0.5Cu.
its yield strength,14 shear strength,8,18 impact strength,5 and creep resistance8 are all higher than Sn-Pb. Figure 13.34 shows that Sn-3.5Ag-0.75Cu exhibits the longest time to break in creep tests among all of the alloys tested, including both Pb-free and Sn-Pb alloys. For Sn-Ag-Cu alloys further away from ternary eutectic composition, such as 93.6Sn-4.7Ag-1.7Cu, not only does the melting temperature (217 to 244°C67) increase, but also the tensile4,7 and shear18 strengths increase, at the expense of reduction in elongation.4 The data on the mechanical properties of Sn-Ag-Cu-In system is fairly scarce. Hwang studied the yield strength σy, tensile strength σT, and plastic strain εp at fracture versus Ag contents for the Sn-Ag-Cu-In system at both 0.5Cu-8In and 0.5Cu4In, with results shown in Fig. 13.35.4 Results indicate that alloys with 8 percent In exhibit a higher yield strength and tensile strength but a lower plastic strain at fracture than alloys with 4 percent In. For an 8 percent In system, the optimal Ag content for tensile properties is 3 percent within the range of 0 to 4.1 percent Ag. The difference between 3 and 4.1 percent Ag alloys is fairly small, and all exhibit a higher tensile strength but a lower elongation than 63Sn-37Pb, as shown in Fig. 13.32. For a 4 percent In system, increasing Ag content from 3 to 4.1 percent results in a linear increase in tensile strength and yield strength, but a decrease in plastic strain.Tanaka et al.5 reported that Sn-3Ag-0.7Cu-1In exhibits a tensile strength comparable with that of 63Sn-37Pb. However, the former displays a considerably higher impact strength and an exceptionally lower creep rate than eutectic Sn-Pb (see Table 13.8 and Fig. 13.34).
13.36
CHAPTER THIRTEEN
TABLE 13.8 Mechanical Properties of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb Properties Ultimate tensile strength (MPa)
63Sn-37Pb Sn-Ag-Cu 46
46
Sn-Ag-Cu-In
48.5 (95.4Sn3.1Ag1.5Cu); 75 (93.6Sn4.7Ag1.7Cu)
64
42 (Sn3.5Ag0.75Cu)
43 (Sn3Ag-1In0.7Cu)
(87.4Sn4.1Ag0.5Cu-8In); 63 (88.5Sn3Ag0.5Cu-8In)
Sn-Ag-Cu-Bi 78 (93.3Sn3.1Ag-3.1Bi0.5Cu)
Sn-Ag-Cu-Sb
Notes/References
36.5 (Sn2.5Ag0.8Cu0.5Sb)
6.56 × 10−4/s, 300 K, Ref. 4
Ref. 5
45.1
Ref. 6 44 (as drawn), 53 (annealed) 93.6Sn4.7Ag1.7Cu
Ref. 7
48 (for both Sn3.8Ag0.7Cu and Sn-3.5Ag0.7Cu)
Ref. 8
49.2
Ref. 9
40.7
38.3 (Sn2.5Ag0.8Cu0.5Sb)
Ref. 11
33.92
39.5 (Sn2.5Ag0.8Cu0.5Sb)
Ref. 11
52.8 (Sn3Ag-2Cu2Sb); 25.8 (Sn-2.6Ag0.8Cu0.5Sb); 29.8 (Sn0.2Ag-2Cu0.8Sb)
Ref. 13
31–46 30.6
Ref. 12 48.3 (Sn3Ag-4Cu); 29.7 (Sn0.5Ag4Cu)
92.7 (Sn-2Ag7.5Bi-0.5Cu)
26.7 Yield strength (MPa)
37
28.1, 30.2
Ref. 2 45 (Sn3.8Ag0.7Cu)
Ref. 14
27.8 and 33.5 (Sn-2.5Ag0.8Cu0.5Sb)
Ref. 11
13.37
PREVAILING LEAD-FREE ALLOYS
TABLE 13.8 Mechanical Properties of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb (Continued) Properties
63Sn-37Pb Sn-Ag-Cu 27.2
Shear strength (MPa)
43.3 (Sn3Ag-4Cu); 25.7 (Sn0.5Ag4Cu)
Sn-Ag-Cu-In Sn-Ag-Cu-Bi 85.3 (Sn-2Ag7.5Bi-0.5Cu)
Sn-Ag-Cu-Sb
Notes/References
46.1 (Sn3Ag-2Cu2Sb); 22.8 (Sn-2.6Ag0.8Cu0.5Sb); 25.9 (Sn0.2Ag-2Cu0.8Sb)
Ref. 13
23
27 (Sn3.8Ag0.7Cu)
At 0.1 mm/min, 20°C, Ref. 8
14
17 (Sn3.8Ag0.7Cu)
At 0.1 mm/min, 100°C, Ref. 8
36.5 (Sn40Pb)
67 (Sn3.6Ag1.0Cu); 58 (Sn-4.7Ag1.7Cu); 63.8 (Sn3.8Ag0.7Cu)
64.1 (Sn3.8Ag0.7Cu0.5Sb)
At 0.1 mm/min; gap thickness: 76.2 µm; cooling rate = 10°/s, tested at 22°C, Ref. 18
4.5 (Sn40Pb)
24.4 (Sn3.6Ag1.0Cu); 21.6 (Sn4.7Ag1.7Cu); 25.1 (Sn3.8Ag0.7Cu)
28.9 (Sn3.8Ag0.7Cu0.5Sb)
At 0.1 mm/min; gap thickness: 76.2 µm; cooling rate = 10°/s; tested at 170°C, Ref. 18
34.5 (Sn40Pb)
At 1 mm/min at reflow temperature, Ref. 19
21.6 (Sn40Pb)
At 1 mm/min at 100°C, Ref. 19
40.27
Ring-and-plug test, Ref. 20
41.8
Ref. 21
28.4
Ref. 12
23.8
Ref. 2
48.4 Elongation (%)
31
Ref. 6 36.5 (95.4Sn3.1Ag1.5Cu); 20 (93.6Sn4.7Ag1.7Cu)
21.5 19 (93.3Sn(87.4Sn3.1Ag-3.1Bi4.1Ag0.5Cu) 0.5Cu-8In); 22.5 (88.5Sn3Ag0.5Cu-8In)
38.5 (Sn2.5Ag0.8Cu0.5Sb)
6.56 × 10−4/s, 300 K, Ref. 4
13.38
CHAPTER THIRTEEN
TABLE 13.8 Mechanical Properties of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb (Continued) Properties
63Sn-37Pb Sn-Ag-Cu
Sn-Ag-Cu-In Sn-Ag-Cu-Bi
Sn-Ag-Cu-Sb
35.5
Ref. 6
43.66; 52.87
50 (Sn2.5Ag0.8Cu0.5Sb)
35–176 48
31
Young’s modulus (Gpa)
33.58
Creep
6
Ref. 11
Ref. 12 22 (Sn3Ag-4Cu); 27 (Sn0.5Ag4Cu)
12 (Sn-2Ag7.5Bi-0.5Cu)
32 (Sn3Ag-2Cu2Sb); 9 (Sn-2.6Ag0.8Cu0.5Sb); 27 (Sn-0.2Ag2Cu-0.8Sb)
35 Impact strength (J/cm2)
Notes/References
Ref. 68
Ref. 2 77 (Sn3.5Ag0.75Cu)
75 (Sn3Ag-1In0.7Cu)
Ref. 5
51.16 (Sn2.5Ag0.8Cu0.5Sb)
Ref. 11
(60Sn40Pb)
27 (Sn4Ag0.5Cu)
25°C, 100 h to failure, MPa, Ref. 24
2.8 (60Sn40Pb)
7.5 (Sn4Ag0.5Cu)
25°C, 1000 h to failure, MPa, Ref. 24
323 (Sn1Ag0.5Cu); 3849 (Sn3.5Ag0.75Cu)
1007 (Sn3Ag0.7Cu1In)
218 (Sn-Ag-Cu7.5Bi); 1747 (Sn-Ag-Cu4.5Bi); 2203 (Sn-Ag-Cu-2Bi)
Time to break (h), Ref. 46
13 for both Sn-3.5Ag0.7Cu and Sn-3.8Ag0.7Cu
Creep strength, N/mm2 at 0.1 mm/min, 20°C, Ref. 8
5.0 for both Sn3.5Ag0.7Cu and Sn-3.8Ag0.7Cu
Creep strength, N/mm2 at 0.1 mm/min, 100°C, Ref. 8
PREVAILING LEAD-FREE ALLOYS
13.39
FIGURE 13.31 Tensile stress-strain relationships of 93.6Sn-4.7Ag-1.7Cu, 95.4Sn-3.1Ag-1.5Cu, and eutectic Sn-Ag, Sn-Cu, and Sn-Pb alloys at 6.56 × 10−4/s, 300 K.
FIGURE 13.32 Tensile stress-strain relationships at 300 K and 6.56 × 10−4/s for 88.5Sn3Ag-0.5Cu-8In, 87.4Sn-4.1Ag-0.5Cu-8In, 95.4Sn-3.1Ag-1.5Cu, and 63Sn-37Pb.
13.40
CHAPTER THIRTEEN
FIGURE 13.33 Tensile stress-strain relationships at 300 K and 6.56 × 10−4/s for 93.3Sn-3.1Ag3.1Bi-0.5Cu, 96.5Sn-3.5Ag, 99.3Sn-0.7Cu, and 63Sn-37Pb.
FIGURE 13.34 Time to break for solder joints in creep tests for several solder alloys.
PREVAILING LEAD-FREE ALLOYS
13.41
FIGURE 13.35 Yield strength σy, tensile strength σT, and plastic strain εp at fracture versus Ag contents for Sn-Ag-Cu-In system at both 0.5Cu-8In and 0.5Cu-4In.
Sn-Ag-Cu-Bi alloys exhibit a higher tensile strength4,13 and yield strength,13 a lower elongation,4,68 and a slower creep rate46 than eutectic Sn-Pb. Figure 13.33 exemplifies the tensile stress-strain behavior of the Sn-Ag-Cu-Bi alloy versus eutectic Sn-Pb, Sn-Ag, and Sn-Cu systems. Toyoda observed that for the Sn-Ag-Cu system, the creep resistance, or time to break in creep test, increases considerably with addition of Bi, as shown in Fig. 13.34.46 Addition of 2 percent Bi results in a considerable increase in time to break (2203 h). However, further increase in Bi content results in a rapid drop in benefit in creep resistance, and the time to break is reduced down to 1747 h for 4.5 percent Bi and 218 h for 7.5 percent Bi, compared with 1 h for eutectic Sn-Pb. Figure 13.36 shows that the tensile strength of the Sn-Ag-Cu-Bi system increases with increasing Bi content, then levels off at around 10 percent Bi.46 On the other hand, the elongation of this system drops rapidly with increasing Bi content until it reaches the 3 percent level, then it decreases slowly and later levels off with further increase in Bi content. The Sn-Ag-Cu-Sb system exhibits a more diverse variation in mechanical properties as a function of composition. The most widely studied composition is 96.2Sn2.5Ag-0.8Cu-0.5Sb (CASTIN). CASTIN was reported to be comparable11 or lower4,13 in tensile strength, comparable11 or lower13 in yield strength, lower,68 comparable,4 or higher11 in elongation than eutectic SnPb. Other compositions of interest that were studied include Sn-3Ag-2Cu-2Sb,13 Sn-3.8Ag-0.7Cu-0.5Sb,18 and Sn-0.2Ag-2Cu-0.8Sb (SAF-A-LLOY).13 Sn-3Ag-2Cu-2Sb is higher in tensile strength and yield strength, but lower in elongation than eutectic Sn-Pb. Sn-3.8Ag0.7Cu-0.5Sb is higher in shear strength, while Sn-0.2Ag-2Cu-0.8Sb is comparable in tensile strength and yield strength, but lower in elongation than 63Sn-37Pb.
13.42
CHAPTER THIRTEEN
FIGURE 13.36 Effect of Bi content on the tensile strength and elongation of Sn-Ag-Cu-Bi and Sn-Zn-Bi alloys.
13.4.3
WETTING PROPERTIES
The wetting properties of Sn-Ag-Cu and Sn-Ag-Cu-X alloys are shown in Table 13.9. The contact angle results of Table 15.9, together with Fig. 13.7, indicate that alloy wetting decreases in the following order: 60Sn-40Pb > Sn-Ag-Cu > Sn-Ag-Cu-Sb.18,22 Spread test results of Baggio show 62Sn-36Pb-2Ag > Sn-Ag-Cu-Bi > Sn-Ag-Cu when tested with a profile with a peak temperature of 240°C, a dwell time above liquidus of 60 s for Pb-free alloys, and a peak temperature of 215°C, 60-s dwell for SnPb-Ag. Both forced-air convection, air reflow atmosphere, and vapor phase reflow were used in his study.45 The spread performance is rated with a scale of 1 to 5, with 5 being the best. Toyoda also studied spreading performance of several alloys, and observed the following spreading behavior in decreasing order: 63Sn-37Pb > Sn-AgCu-4.5Bi, Sn-Ag-Cu-7.5Bi > Sn-3.5Ag-0.75Cu > 99.25Sn-0.75Cu > 89Sn-8Zn-3Bi, as shown in Fig. 13.13.46 A wetting time study done by Baggio showed that there was no significant difference between 62Sn-36Pb-2Ag (at 235°C), Sn-3.8Ag-0.7Cu, and Sn-3.3Ag-3Bi1.1Cu (at 260°C) for metallized PCB surface finishes, including immersion Pd, immersion Sn, immersion Ag, and Ni/Au.45 However, 62Sn-36Pb-2Ag wetted slightly faster on organic solderability preservative (OSP) than Sn-Ag-Cu and SnAg-Cu-Bi.45 Lotosky’s results show that at 260°C when tested on oxidized copper, the wetting time increases in the following order: Sn-4Ag-0.5Cu (1.1 s) < 63Sn-37Pb (1.85 s) < Sn-2.5Ag-0.8Cu-0.5Sb (2.05 s) < 96.5Sn-2.5Ag-1Bi-0.5Cu (2.7 s).30 Toyoda studied the wetting time of meniscograph for Sn-Pb, Sn-Cu, Sn-Ag-Cu, and Sn-AgCu-Bi alloys, and observed that the wetting time increases in the following order: 63Sn-37Pb < Sn-Ag-Cu-2Bi ∼ Sn-Ag-Cu-1Bi < Sn-3.5Ag-0.75Cu < Sn-1Ag-0.5Cu < Sn-0.7Cu-0.3Ag < Sn-0.75Cu. The wetting time decreases with increasing temperature at a slightly different rate, as shown in Fig. 13.37.45 In Toyoda’s work, the Bi content between 1 and 2 percent seems to have no effect on the wetting performance, including spreading and wetting time, of Sn-Ag-Cu-Bi system. Both Sn-AgCu-1Bi and Sn-Ag-Cu-2Bi display a wetting behavior that is fairly comparable with 63Sn-37Pb.
13.43
PREVAILING LEAD-FREE ALLOYS
TABLE 13.9 Wetting Properties of 63Sn-37Pb, Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, and Sn-Ag-Cu-Sb
Properties Contact angle (degree)
Wetting time (s)
63Sn37Pb 17 (Sn40Pb)
Sn-Ag-Cu
47
32
45
31
35
0.27 s at 235°C (Sn62) 0.20 s at 235°C (Sn62) 0.32 s at 235°C (Sn62) 0.20 s at 235°C (Sn62) 0.21 s at 235°C (Sn62) 0.24 s at 235°C (Sn62)
1.85 s Spread 4.55 (Sn62) 4.7 (Sn62) 4.4 (Sn62)
Sn-AgCu-Bi
21 (Sn4.70Ag1.70Cu)
22
0.36 s at 235°C (Sn62)
Sn-AgCu-In
0.28 s (Sn-3.8Ag0.7Cu) at 260°C 0.23 s (Sn-3.8Ag0.7Cu) at 260°C 0.25 s (Sn-3.8Ag0.7Cu) at 260°C 0.42 s (Sn-3.8Ag0.7Cu) at 260°C 0.26 s (Sn-3.8Ag0.7Cu) at 260°C 0.23 s (Sn-3.8Ag0.7Cu) at 260°C 0.27 s (Sn-3.8Ag0.7Cu) at 260°C
Sn-AgCu-Sb 44 (Sn3.8Ag0.7Cu0.5Sb)
Notes/References Ref. 18
Flux A611, 260– 280°C, Ref. 22 Flux A260HF, 260–280°C, Ref. 22 Flux B2508, 260–280°C, Ref. 22
1.1 s (Sn4Ag0.5Cu)
0.24 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.26 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.19 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.44 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.26 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.25 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 0.27 s (Sn3.3Ag-3Bi1.1Cu) at 260°C 2.75 s (96Sn2.5Ag-1Bi0.5Cu)
4.2 (Sn3.8Ag0.7Cu) 4.55 (Sn3.8Ag0.7Cu) 3.9 (Sn3.8Ag0.7Cu)
4 (Sn3.3Ag-3Bi1.1Cu) 4.6 (Sn3.3Ag-3Bi1.1Cu) 4.4 (Sn3.3Ag-3Bi1.1Cu)
Immersion Pd PCB, s, Ref. 45
Immersion Sn PCB, s, Ref. 45
Immersion Ag, PCB, s, Ref. 45
NiAu, PCB, s, Ref. 45
OSP 1, s, Ref. 45
OSP 2, s, Ref. 45
OSP 3, s, Ref. 45 2.05 s (Sn2.5Ag0.8Cu0.5Sb)
At 260°C, RF 800, oxidized Cu, Ref. 30
OSP 3*, Ref. 45 Immersion Ag*, Ref. 45
Immersion Pd*, Ref. 45
13.44
CHAPTER THIRTEEN
TABLE 13.9 Wetting Properties of 63Sn-37Pb, Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, and Sn-Ag-Cu-Sb (Continued)
Properties
63Sn37Pb
5 (Sn62)
5 (Sn62) 4.7 (Sn62) 4.7 (Sn62)
5 (Sn62)
Sn-Ag-Cu 4.4 (Sn3.8Ag0.7Cu) 4.35 (Sn3.8Ag0.7Cu) 4.8 (Sn3.8Ag0.7Cu) 3.9 (Sn3.8Ag0.7Cu) 5 (Sn3.8Ag0.7Cu)
Sn-AgCu-In
Sn-AgCu-Bi 4.7 (Sn3.3Ag-3Bi1.1Cu) 4.45 (Sn3.3Ag-3Bi1.1Cu) 4.95 (Sn3.3Ag-3Bi1.1Cu) 4.65 (Sn3.3Ag-3Bi1.1Cu) 5 (Sn3.3Ag-3Bi1.1Cu)
Sn-AgCu-Sb
Notes/References
NiAu*, Ref. 45
OSP 3,† Ref. 45 Immersion Ag,† Ref. 45
Immersion Pd,† Ref. 45
NiAu,† Ref. 45
* Peak 240°C, dwell 60 s for Pb-free, 215°C, 60-s dwell for Sn-Pb-Ag, scale 1 to 5 (best), forced-air convection, air. † Peak 240°C, dwell 60 s for Pb-free, 215°C, 60-s dwell for Sn-Pb-Ag, scale 1 to 5 (best), 230°C bp VPR.
For Sn-Ag-Cu-In, no wetting information is available. For the Sn-Ag-Cu-Sb system, Seelig et al.11 studied the effect of Ag content on wetting performance when using rosin-based, mildly activated (RMA), no-clean, and organic acid (OA) fluxes. Results indicate that at around 2.5 percent Ag content, the Sn-Ag-Cu-Sb system exhibits the shortest wetting time for two out of three fluxes, as shown in Fig. 13.38.11
FIGURE 13.37 Wetting time of meniscograph for Sn-Pb, Sn-Cu, Sn-Ag-Cu, and Sn-Ag-CuBi alloys.
PREVAILING LEAD-FREE ALLOYS
13.45
FIGURE 13.38 Effect of Ag content on the wetting time of Sn-Ag-0.8Cu-0.5Sb system using RMA, no-clean, and OA fluxes.
13.4.4
RELIABILITY
The isothermal low-cycle fatigue (strain 0.2 percent, 0.1 Hz, R = 0.8, 300 K) performance of several Sn-Ag-Cu and Sn-Ag-Cu-X alloys was studied by Hwang,4 with results shown in Table 13.5. The number of cycles to failure at 300 K (50 percent load drop, 0.2 percent strain range) decreases in the following order: 88.5Sn-3Ag-0.5Cu-8In (19,501) > 95.4Sn-3.1Ag-1.5Cu (8,936) > 96.2Sn-2.5Ag-0.8Cu-0.5Sb (8,751) > 94.5Sn0.5Ag-2Cu-3Sb (7,120) > 93.3Sn-3.1Ag-3.1Bi-0.5Cu (6,522) > 63Sn-37Pb (3,650). The data above suggest that Sn-Ag-Cu-X is a very viable family as lead-free alternatives. Some temperature cycling and heat treatment reliability data for Sn-Ag-Cu, SnAg-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and 63Sn-37Pb or 62Sn-36Pb-2Ag are shown in Table 13.10. Since all of the viable Pb-free alternatives comprise a high Sn content, the possibility of a high IMC formation rate and its impact on reliability are a concern. Grusd reported that the Cu dissolution rate during soldering increases in the following order: 60Sn-40Pb < Sn-Ag-1Cu < Sn-2.5Ag-0.8Cu-0.5Sb < Sn-Ag-Cu-Bi < Sn-Ag0.5Cu, and thus validates this concern.24 Figure 13.39 shows examples of intermetallic formations between Sn-Ag-Cu and copper substrate.73 Although the intermetallics thickness may be thicker than the Sn-Pb system, it does not adversely affect the solder joint reliability. Feldmann and Reichenberger studied the effect of 160°C storage aging time on the shear strength of chip resistor 1206.66 Results indicate that while 62Sn-36Pb-2Ag decreased in shear strength for about 30 percent after 1000 h of aging at 160°C, the Pb-free alternatives Sn-3.8Ag-0.7Cu and Sn-3.3Ag-3Bi-1.1Cu reduced only 12 and 4 percent, respectively, under the same test condition. In all of the results reported in Table 13.10, the Sn-Ag-Cu system is always equal or better than 63Sn-37Pb. This widely acceptable reliability performance of Sn-AgCu system, regardless of applications, strongly validates the acceptability of this system as a substitute for eutectic Sn-Pb. Figures 13.14 through 13.17 show the Weibull plots for Sn-Ag-Cu versus Sn-Pb in the BGA applications over several temperature cycling test conditions, thus demonstrating the superiority of the Sn-Ag-Cu system.
TABLE 13.10 Accelerated Life Testing Performance of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and Sn-Pb—The Performance May Also Be Ranked in Descending Order for Some Works, with 1st Being the Best Test condition
Applications
Sn-Pb (or Sn62)
Sn-Ag-Cu
Sn-Ag-Cu-In
Sn-Ag-Cu-Bi
Sn-Ag-Cu-Sb
Notes/ References
13.46
0–100°C
General
Good (on NiP/Au)
Comparable with Sn63 (Sn-4Ag0.5Cu on TiW/Cu UBM)
Ref. 14
−40°C, 15 min/+125°C, 15 min, shear strength after 2000 cycles
TQFP64, CR1206
Good
Comparable with Sn62
Comparable with Sn62
Ref. 66
160°C heat treatment, shear strength (N): 0/500/1000 h
CR1206, 100Sn component finish to OSP board finish
68/55/48 (Sn62)
68/62/60 (Sn-3.8Ag0.7Cu)
80/83/77 (Sn-3.3Ag3Bi-1.1Cu)
Ref. 66
Single-lead shear strength (N)
TQFP64, Sn-Pb vs. Ni-Pd component finish to OSP board finish
7/6.9 (Sn62)
8.8/8 (Sn3.8Ag0.7Cu)
6.5/10.3 (Sn-3.3Ag3Bi-1.1Cu)
Ref. 66
−55/+125°C
Cylindrical resistor, crack vs. cycles
1st (Sn3.5Ag0.75Cu)
3rd (Sn-Ag-Cu7.5Bi)
Ref. 46
−40/80°C, crack
3rd
1st (Sn3.5Ag0.75Cu)
1st (Sn-Ag-Cu1In)
1st (Sn-Ag-Cu7.5Bi)
Ref. 46
13.47
−40/125°C, 42-min cycle time
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
3rd, close to 2nd
2nd (Sn3.8Ag0.7Cu), close to Sn-Ag-Bi
Ref. 49
−40/125°C, 240-min cycle time
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
2nd
1st (Sn3.8Ag0.7Cu)
Ref. 49
0/100°C, 30min cycle time
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
2nd
1st (Sn3.8Ag0.7Cu), same as Sn-Ag-Bi, better than Sn63
Ref. 49
0/100°C, 60min cycle time
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
2nd
1st (Sn3.8Ag0.7Cu), same as Sn-Ag-Bi, better than Sn63
Ref. 49
0/100°C, 240-min cycle time
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
2nd
1st (Sn3.8Ag0.7Cu), better than Sn63
Ref. 49
TABLE 13.10 Accelerated Life Testing Performance of Sn-Ag-Cu, Sn-Ag-Cu-In, Sn-Ag-Cu-Bi, Sn-Ag-Cu-Sb, and Sn-Pb—The Performance May Also Be Ranked in Descending Order for Some Works, with 1st Being the Best (Continued) Test condition
Applications
Sn-Pb (or Sn62)
Sn-Ag-Cu
13.48
−40/60°C, 30-min cycle
625 I/O CBGA, 32 × 32 × 0.8 mm, same alloy for ball and paste
2nd, close to Sn-Ag-Cu
1st
−40/+125°C
fleXBGA
7th ∼ 6th
3rd (Sn4Ag-1Cu); 5th (Sn4Ag-0.5Cu)
Flip chip
Good
Comparable with Sn63
Flip chip of nonunderfilled systems
Flip chip
Sn-Ag-Cu-In
Sn-Ag-Cu-Sb
6th (Sn3.4Ag-1Cu3.3Bi) ∼ 7th
2nd (Sn2.5Ag0.8Cu0.5Sb)
Comparable with Sn63 (Sn-3.8Ag0.7Cu) Good (Sn3.5Ag0.75Cu)
General
Best
Ref. 51
Ref. 14 2nd, 96.2Sn2.5Ag0.8Cu-1Sb is about 80% of Sn63-Pb37
General
Notes/ References Ref. 49
1st
Good
Sn-Ag-Cu-Bi
Ref. 51
Ref. 52
Good (Sn3Ag-1In0.7Cu)
Ref. 5
Refs. 54, 55
FIGURE 13.39 SEM migrographs (2500×) using backscattered electron imaging to show the intermetallic layer between copper and Sn-3.6Ag-1Cu (top), Sn-3Ag-0.5Cu (middle), and Sn-4Ag0.5Cu (bottom).
13.49
13.50
CHAPTER THIRTEEN
Among the viable Sn-Ag-Cu compositions, Sn-3.8Ag-0.7Cu and Sn-4Ag-0.5Cu appear to be more commonly accepted. Sn-Ag-Cu-In is much less studied than Sn-Ag-Cu. Tanaka et al. studied Pb-free technology solder for mobile equipment at NEC, and concluded that Sn-3.1Ag-1In0.7Cu, which is similar to Sn-3.5Ag-0.75Cu and Sn-3.5Ag-0.5Bi-3In, is equal to 63Sn37Pb in the fatigue test, but is inferior to Sn-Ag-Bi-1.1Cu-Ge and Sn-Ag-X.5 However, the latter two alloys are unacceptable for the Charpy Impact Test. All five Pb-free alloys are excellent in the creep rupture test when compared with 63Sn37Pb. Overall,Tanaka et al. concluded that Sn-3Ag-1In-0.7Cu is equal to eutectic SnPb, is poorer than both Sn-3.5Ag-0.75Cu and Sn-3.5Ag-0.5Bi-3In, but is better than Sn-Ag-Bi-1.1Cu-Ge and Sn-Ag-X. Toyoda investigated the temperature cycling performance of several Pb-free solders.46 The cycling condition is −40 to +80°C, with a 30-min temperature cycle, using a wave-soldered nylon connector in a paper phenol substrate. The failure is monitored by observing externally the crack development. Results indicate that the fatigue resistance increases in the following order: 63Sn-37Pb < Sn-3.5Ag < Sn-AgCu-7.5Bi, Sn-Ag-Cu-1In, Sn-3.5Ag-0.75Cu up to 1000 cycles, as shown in Fig. 13.40.46 Hwang also studied the isothermal low-cycle fatigue behavior of Pb-free alloys at 300 K, 0.2 percent strain, 0.1 Hz, and R = 0.8.4 For the Sn-Ag-Cu-In system, she reported that the optimum In content for low-cycle fatigue life performance is 8 percent for both Sn-In-4.1Ag-0.5Cu and Sn-In-3.1Ag-0.5Cu systems, as shown in Fig. 13.41.4 Hwang also studied the optimum Ag content for Sn-Ag-In-0.5Cu, with In maintained at 4 and 8 percent. Results shown in Fig. 13.42 indicate that the Ag content may be better to be higher than 3 percent.4 In the fatigue life test, 88.5Sn-3Ag-0.5Cu-8In (19,501 in Nf value) is the best among all the rest alloys, as shown in Table 13.5. Sn-Ag-Cu-Bi is outstanding in creep resistance and wetting, as indicated in Figs. 13.34 and 13.37. However, the presence of the Bi ingredient brings up concerns on
FIGURE 13.40 Temperature cycling performance of 63Sn-37Pb, Sn-Ag-Cu-7.5Bi, Sn-Ag-Cu-1In, Sn-3.5Ag-0.75Cu, and Sn-3.5Ag (test conditions: −40°/+80°C at 30-min cycle time).
PREVAILING LEAD-FREE ALLOYS
13.51
FIGURE 13.41 Fatigue life Nf versus In content for the Sn-Ag-Cu-In system at both Sn-In-4.1Ag-0.5Cu and Sn-In-3.1Ag-0.5Cu.
FIGURE 13.42 Fatigue life Nf as a function of Ag content for Sn-Ag-Cu-In system at both Sn-Ag0.5Cu-8In and Sn-Ag-0.5Cu-4In.
13.52
CHAPTER THIRTEEN
the sensitivity toward lead contamination. Feldmann and Reichenberger evaluated the solder joint shear strength of TQFP64 with Sn-Pb or Ni-Pd component finishes soldered onto PCB with OSP board finish using 62Sn-36Pb-2Ag, Sn-3.8Ag-0.7Cu, and Sn-3.3Ag-3Bi-0.7Cu, with the results shown in Fig. 13.43.66 The joint strength of 62Sn-36Pb-2Ag and Sn-3.8Ag-0.7Cu is insensitive to the component finish type. However, in the case of Sn-3.3Ag-3Bi-1.1Cu, solder joints with the Sn-Pb component lead surface finish are much lower than joints with Ni-Pd finish due to the formation of a ternary eutectic low melting phase, 52Bi-30Pb-18Sn, as will be discussed in Chap. 16. The temperature cycling performance of Sn-Ag-Cu-Bi was studied by Toyoda.46 At test conditions of −40°/+80°C at a 30-min cycle time, Sn-Ag-Cu-7.5Bi outperforms 63Sn-37Pb and Sn-3.5Ag and, similar to Sn-Ag-Cu-1In and Sn-3.5Ag-0.75Cu, exhibits no failure up to at least 1000 cycles, as shown in Fig. 13.40. However, at test conditions of −55/+125°C and a cycle time of 30 min, the temperature cycling performance of the solder joints of a cylindrical chip resistor using Sn-Ag-Cu-7.5Bi is slightly poorer than 62Sn-36Pb-2Ag and is considerably poorer than Sn-3.5Ag0.75Cu, as shown in Fig. 13.44.46 In Syed’s work as part of the National Center for Manufacturing Sciences (NCMS) project, fatigue life of the fleXBGA package [144 input/output (I/O), 18mil ball size, 0.8-mm ball pitch) was determined at the following conditions: −40/+125°C, 15-min ramp and dwell, 1 cycle/h, single-zone cycling chamber for a series of Pb-free solders. The results are shown in Table 13.11, indicating that Sn-AgCu-Bi is comparable with or slightly better than control eutectic Sn-Pb solder, poorer than Sn-Ag-Cu, Sn-Ag-Cu-Sb, Sn-Ag-Cu-Sb-Bi, Sn-Ag-In, but better than Sn-Ag.50 The results are roughly consistent with the findings of Toyoda46 and Feldmann and Reichenberger.66 Sn-Ag-Cu-Sb is one of the alloy systems that received a relatively good evaluation in the early stage of Pb-free soldering development. In Syed’s results, Sn-2.5Ag0.8Cu-0.5Sb is leading in fatigue life evaluation (see Table 13.11). In Elenius and Yeh’s work, 96.2Sn-2.5Ag-0.8Cu-1Sb is about 80 percent of Sn63-Pb37 for unfilled
FIGURE 13.43 Single lead shear strength of TQFP64 with Sn-Pb or NiPd surface finishes. The component is soldered onto PCB with an OSP surface finish with 62Sn-36Pb-2Ag, Sn-3.8Ag-0.7Cu, and Sn-3.3Ag-3Bi-0.7Cu.
PREVAILING LEAD-FREE ALLOYS
13.53
FIGURE 13.44 Temperature cycling performance of 62Sn-36Pb-2Ag, Sn-Ag-Cu-7.5Bi, and Sn3.5Ag0.75-Cu (test conditions: −55°C/+125°C and cycle time 30 min).
flip chip assembly.51 Seelig and Suraski reported that the fatigue life of 96.1Sn2.6Ag-0.8Cu-0.5Sb is comparable with or better than 96.5Sn-3.5Ag, as shown in Table 13.12,11 according to ASTME 606, 1-Hz triangular waveform oscillated between 0.15 and −0.15 percent strain. The passing mark was set at 10,000 cycles. The reliability potential of Sn-Ag-Cu-Sb was also investigated by determining the intermetallics growth rate at 125°C on several commonly used base metals, including copper, brass, nickel, and alloy 42. Results indicate that 96.1Sn-2.6Ag0.8Cu-0.5Sb (CASTIN) is most stable when compared with 60Sn-40Pb, 96.5Sn3.5Ag, and 99.3Sn-0.7Cu, as shown in Fig. 13.45.11 The advantage in slow intermetallics growth rate of Sn-Ag-Cu-Sb is particularly profound on copper. TABLE 13.11 Relative Comparison of Fatigue Life of Pb-Free Alloys* Relative fatigue life By first failure
By mean life
A1 (Sn-Ag)
Alloy
0.69
0.94
A11 (Sn-Ag-Cu)
1.27
1.28
A14 (Sn-Ag-Cu)
1.14
1.26
A21 (Sn-Ag-Cu-Sb)
1.29
1.33
A32 (Sn-Ag-Cu-Sb-Bi)
1.17
1.31
A62 (Sn-Ag-Cu-Bi)
1.01
1.12
A66 (Sn-Ag-In)
1.29
1.25
B63 (Sn-Pb)
1.00
1.00
* At the following conditions: −40/+125°C, 15-min ramp and dwell, 1 cycle/h, single-zone cycling chamber, fleXBGA package, 144 I/O, 18-mil ball size, 0.8-mm ball pitch.
13.54
CHAPTER THIRTEEN
TABLE 13.12 Fatigue Life of 96.1Sn-2.6Ag-0.8Cu-0.5Sb and 96.5Sn-3.5Ag* 96.1Sn-2.6Ag-0.8Cu-0.5Sb
96.5Sn-3.5Ag
11,194
10,003
26,921
6,267
24,527
11,329
* Test conditions: ASTME 606, 1-Hz triangular waveform oscillated between 0.15 and −0.15 percent strain. The passing mark was set at 10,000 cycles.
13.5 13.5.1
Sn-Zn AND Sn-Zn-Bi PHYSICAL PROPERTIES
91Sn-9Zn is attractive mainly due to its relatively low melting temperature. However, the high activity of Zn results in a great challenge in solder paste reflow applications. Bi is added accordingly in order to reduce the corrosivity of Zn under humid conditions and to reduce the melting temperature further. Table 13.13 shows the physical properties of 91Sn-9Zn, 89Sn-8Zn-3Bi, and eutectic Sn-Pb, although information on Sn-Zn-Bi is very scarce. In general, the physical properties of the Sn-Zn alloy are dictated by the property of Sn due to high Sn content. Sn-Zn exhibits a higher surface tension than eutectic Sn-Pb, as predicted by Fig. 12.2, hence inferring a poorer wetting behavior. Addition of low-surface-energy Bi is expected to lower the surface tension of the Sn-Zn system (see Fig. 12.2), thus promising a better wetting.
FIGURE 13.45 Rate of intermetallic growth of 60Sn-40Pb, 96.5Sn-3.5Ag, 99.3Sn-0.7Cu, and 96.1Sn-2.6Ag-0.8Cu-0.5Sb (CASTIN) at 125°C on copper, brass, nickel, and alloy 42.
13.55
PREVAILING LEAD-FREE ALLOYS
TABLE 13.13 Physical and Mechanical Properties of 91Sn-9Zn and 89Sn-8An-3Bi
Property Melting temperature (°C) Density (g/cm3) Electrical resistivity (µΩ-cm)
63Sn-37Pb
91Sn-9Zn
183
199
8.4
7.27
17 15 10–15
14.5 14.6 Thermal conductivity [W/(m·K)] Specific heat [J/(g·K)] CTE (ppm/K)
Hardness
Surface tension (mN m−1)
13.5.2
50.9 61 0.167
0.239
16.8 24 (−70°C, 20°C, 140°C) 21 25 18.74
23.9
12.8
21.3
17
21.5
417 (air), 464 (N2) at 233°C
89Sn-8Zn-3Bi 187–197
Notes/ References Ref. 6 Ref. 1 Ref. 63 Ref. 1 Ref. 3 Ref. 2 Ref. 2 Ref. 1 Ref. 1 Ref. 6 Ref. 15 Ref. 16 Ref. 2 Ref. 8
518 (air), 487 (N2) at 249°C
HV; kg/mm2, Ref. 6 Brinel hardness, Ref. 1 Ref. 3
MECHANICAL PROPERTIES
The mechanical properties of 63Sn-37Pb, 91Sn-9Zn, and 89Sn-8Zn-3Bi are shown in Table 13.14. In general, the Sn-Zn alloy is higher in yield strength, equal or higher in tensile strength, equal in shear strength, and equal or lower in elongation than eutectic Sn-Pb. On the other hand, Sn-Zn-Bi is higher in creep resistance than Sn-Pb. The effect of Bi addition on tensile properties of Sn-Zn results in embrittlement of the alloy, and is fairly similar to that of Sn-Ag-Cu. Figure 13.36 shows that at 8 percent Bi content, the tensile strength of the Sn-Zn-Bi system is at the maximum and is about twice that of Sn-Zn, while the elongation is at the minimum and is about one-sixth of that of the Sn-Zn.
13.5.3
WETTING PROPERTIES
Zn-containing alloys are notorious in terms of wetting performance, mostly due to the susceptibility of Zn to oxidation and the high surface energy of the systems. The latter results in difficulty in wave soldering bath maintenance, as reported by Baggio.45 Toyoda studied the solder spread of Pb-free alloys, and concluded that 89Sn-8Zn-3Bi is the lowest in spread when compared with Sn-Pb, Sn-Cu, Sn-Ag-Cu, and Sn-Ag-Cu-Bi,
13.56
CHAPTER THIRTEEN
TABLE 13.14 Mechanical Properties of 63Sn-37Pb, 91Sn-9Zn, and 89Sn-8Zn-3Bi Properties
63Sn-37Pb
91Sn-9Zn
Ultimate tensile strength (MPa)
45.1 30.6 51.7
45.4 53.1 54.7
89Sn-8Zn-3Bi
Ref. 6 Ref. 13 Ref. 1
Yield strength, A (0.2%, MPa)
27.2
51.6
Ref. 13
Shear strength (MPa)
48.4
48.8
Ref. 6
Elongation (%)
35.5 48 35
40 27
Ref. 6 Ref. 13 Ref. 2 Ref. 1
Creep (time to break, h)
1
33 94
Notes/References
Ref. 46
as shown in Fig. 13.13.46 In his results, it is also interesting to note that, opposite to the behavior of other alloys, the spread of Sn-Zn-Bi decreases with increasing temperature. Presumably this can be attributed to the oxidation of Zn during testing. Huang and Lee studied the reflow spreading performance of Pb-free solders, with results shown in Fig. 13.46.29 In this work, each alloy was evaluated in solder paste form using 10 different flux chemistries, with the average performance expressed as a wetting index, where 10 and 0 represent full wetting and no wetting, respectively. The reflow was conducted at two different reflow profiles, one with a peak temperature of 15°C above the liquidus, and another one with a peak temperature of 30°C above liquidus temperature. The spreading of Sn-Zn-Bi was extremely poor, with the solder pastes barely reflowed in many cases. The poor wetting performance of Sn-Zn-Bi is presumably attributable to the highly oxidative nature of Zn. Loomans et al. studied the contact angle of binary eutectic alloys using rosin-IPA flux, and reported the following results: Sn-Bi, 40° (166°C); Sn-Zn, 60° (225°C); and Sn-Ag, 45° (250°C).32 These angles were little affected by a number of 1 percent ternary additions to the solders. The significantly larger contact angle of Sn-Zn than that of Sn-Bi and Sn-Ag reflects the greater difficulty in wetting, and is consistent with results discussed earlier. Suganuma studied the microstructure of lead-free solders and of their interfaces with copper. He reported that the Sn-Zn alloys only form different Cu-Zn IMCs (beta-Cu-Zn and gamma-Cu5Zn8) at the interface. Even with the small amount of Zn added to Sn, Zn segregates to the interface to form the Cu-Zn IMC with Sn. However, most other Sn alloys, including pure Sn, Sn-Ag, Sn-Bi, or their ternary alloys, form two IMCs at the interfaces with Cu [i.e., Cu6Sn5 (15 µm) and Cu3Sn (5 µm).The former is much thicker than the latter and the interface integrity is strongly influenced by the presence of the Cu6Sn5 layer.37 Sn-Zn-(Bi) is not stable with most fluxes when used as a solder paste, mainly due to the high reactivity of Zn. However, few solder paste products have been developed, including no-clean applications, for this alloy system.6
13.5.4
RELIABILITY
The reliability of Sn-Zn-Bi system was studied by determining the shear strength of solder joints as a function of storage time at 125°C, as shown in Fig. 13.47.6 The sur-
PREVAILING LEAD-FREE ALLOYS
13.57
FIGURE 13.46 Solder paste wetting (spreading) performance for Pb-free and 63Sn37Pb solder pastes when reflowed at a peak temperature 15 and 30°C above the liquidus temperature of alloys. An index value of 10 represents full wetting, and 0 represents no wetting.
face finishes for PCBs tested included Au-Ni and preflux. Results indicate that both Sn-Zn-Bi and Sn-Pb maintained a steady shear strength with storage time up to at least 3000 h. Ni-Au resulted in a slightly higher shear strength than preflux, and SnZn-Bi joints appear to be slightly stronger than those of Sn-Pb. However, in another test where the solder joints were pretreated with thermal cycling, the shear strength started to decline after 2000 cycles for Sn-Zn-Bi, and from the very beginning of the Sn-Pb system, as shown in Fig. 13.48.6 In the case of Sn-Zn-Bi, no difference in strength and reliability can be discerned between Ni-Au and preflux finishes. While Showa Denko’s results indicate a lack of sensitivity of Sn-Zn-Bi toward heat treatment, Suganuma reported that the tensile strength (MPa) of Cu/solder/Cu joints for Sn-Zn did deteriorate with heat treatment. Thus, the tensile strength for as-soldered/after-exposure at 150°C for 100 h can be shown as follows: Sn-8Zn (Au-Pd-Ni), 73/54; Sn-8Zn, 53/15; Sn-3.5Ag, 52/37; Sn-7.5Bi-2Ag-0.5Cu, 44/35; and Sn-37Pb, 28/18.37 The deterioration extent for Sn-Zn on Cu without Au-Pd-Ni surface finish is greater than the rest systems studied, suggesting a change in Sn-Zn solder joint microstructure may have occurred in the absence of Au-Pd-Ni. Indeed, this may very well be the case. In the absence of a Ni diffusion barrier for copper, Zn and Cu form Cu-Zn IMCs (beta-Cu-Zn and gamma-Cu5Zn8) at the interface, as reported by Suganuma for the Sn-Zn system.37 A similar Cu-Zn IMC is also observed in the Sn-Zn-Bi system on Cu, as shown in Fig. 13.49.6 The development of Cu-Zn IMC can be prevented by predepositing a Ni barrier on top of Cu, as shown in Fig. 13.50.6
FIGURE 13.47 Shear strength of 89Sn-8Zn-3Bi and Sn-Pb soldered onto Au-Ni or preflux-coated copper at 125°C storage temperature.
FIGURE 13.48 Shear strength of 89Sn-8Zn-3Bi and Sn-Pb soldered onto Au-Ni or preflux-coated copper as a function of thermal cycle number.
FIGURE 13.49 Effect of thermal cycle test on microstructure of 89Sn-8Zn-3Bi (electron probe microanalysis view).
13.58
PREVAILING LEAD-FREE ALLOYS
13.59
FIGURE 13.50 Effect of thermal cycle test on microstructure of 89Sn-8Zn-3Bi (electron probe microanalysis view) in the presence of a Ni barrier layer.
13.6
SUMMARY
Sn-Ag-Cu is the most attractive system, mainly due to overall superior reliability and acceptable soldering property and cost. Addition of In, Bi, or Sb introduces improved performance with a trade-off. Sn-Ag-Bi-X is outstanding on soldering, but sensitive to Pb-contamination and fillet lift. Reliability of Sn-Ag can range from being poor to good, and is highly dependent on applications. Its soldering performance is marginally acceptable. Sn-Cu is poor in mechanical strength, and its reliability can range from poor to good, too, depending on the applications. Sn-Zn-Bi is attractive mainly due to the low melting temperature. The high activity of Zn prohibits this system to be used in high-end applications.
REFERENCES 1. Indium Corporation of America, data sheet. 2. Lee, N.-C., J. A. Slattery, J. R. Sovinsky, I. Artaki, and P. T. Vianco, “A Drop-in Lead-Free Solder Replacement,” Proceedings of the NEPCON West Conference, Anaheim, CA, February 28–March 2, 1995. 3. Glazer, J., “Microstructure and Mechanical Properties of Pb-free Solder Alloys for LowCost Electronic Assembly: A Review,” J. Electronic Materials, 23(8):693, 1994. 4. Hwang, J. S., “Solder Materials,” SMT, 15(7), July 2001. 5. Tanaka, Y., J. Takahashi, and K. Kawashima, “Lead Free Soldering Technology for Mobile Equipment,” Proceedings of IMAPS, pp. 336–341, Boston, September 20–22, 2000. 6. Denko, S., “Development of Sn-Zn Solder Paste of High Reliability,” Proceedings of IPCWorks ’99, Minneapolis, MN, October 27, 1999. 7. Anderson, I. E., Ö. Ünal, T. E. Bloomer, and J. C. Foley, “Effects of Transition Metal Alloying on Microstructural Stability and Mechanical Properties of Tin-Silver-Copper Solder Alloys,” Proc. Third Pacific Rim International Conference on Advanced Materials and Processing (PRICM 3), The Minerals, Metals, and Materials Society, Honolulu, HI, July 12–16, 1998. 8. IPC Leadfree website: www.leadfree.org, “NIST Database for Solder Properties with Emphasis on New Lead-free Solders.” 9. Solder Data Sheet, Welco Castings, 2 Hillyard Street, Hamilton, ON, Canada. 10. Gray, D. E., ed., American Institute of Physics Handbook, pp. 2–61 ff., McGraw-Hill, New York, 1957. (Note: Original units were in dyn/cm2; 10 dyn/cm2 = 1 N/m2 = 1 Pa.)
13.60
CHAPTER THIRTEEN
11. Seelig, K., and D. Suraski, “The Status of Lead-Free Solder Alloys,” Proc. 50th IEEE 2000 Electronic Components and Technology Conference, Las Vegas, NV, May 21–24, 2000. 12. Sigelko, J. D., and K. N. Subramanian, “Overview of Lead-Free Solders,” Adv. Mat. & Proc., 47–48, March 2000. 13. Technical Reports for the Lead-Free Solder Project, Properties Reports, “Room Temperature Tensile Properties of Lead-Free Solder Alloys,” Lead-Free Solder Project CD-ROM, National Center for Manufacturing Sciences (NCMS), Ann Arbor, MI, 1998. 14. Maestrelli, L. M., and R. C. Pfahl, “Technology for Environmentally Preferred Products,” IMAPS-Brazil, São Paulo, Brazil, August 1–3, 2001. 15. Wong, T., and A. H. Matsunaga, “Ceramic Ball Grid Array Solder Joint Thermal Fatigue Life Enhancement,” Proceedings: NEPCON West Conference, Anaheim, CA, February 28–March 2, 1995. 16. Lau, J., C. Chang, R. Lee, T.-Y. Chen, D. Cheng, T. J. Tseng, and D. Lin, “Design and Manufacturing of Micro Via-in-Pad Substrates for Solder Bumped Flip Chip Applications,” Journal of Electronics Manufacturing, 10(1):79–87, 2000. 17. McCabe, R. J., and M. E. Fine, “Athermal and Thermally Activated Plastic Flow in Low Melting Temperature Solders at Small Stresses,” Scripta Materialia, 39(2):189, 1998. 18. Anderson, I. E., T. E. Bloomer, R. L. Terpstra, J. C. Foley, B. A. Cook, and J. Harringa, “Development of Eutectic and Near-Eutectic Tin-Silver-Copper Solder Alloys for LeadFree Electronic Assemblies,” IPCWorks ’99: An International Summit on Lead-Free Electronics Assemblies, Minneapolis, MN, October 25–28, 1999. 19. International Tin Research Institute, publ. no. 656, through Hampshire, W. B., “The Search for Lead-Free Solders,” Proc. Surface Mount International Conference, p. 729, San Jose, CA, September 1992. 20. Hernandez, C. L., P. T. Vianco, and J. A. Rejent, “Effect of Interface Microstructure on the Mechanical Properties of Pb-Free Hybrid Microcircuit Solder Joints,” IPC/SMTA Electronics Assembly Expo, p. S19-2-1, 1998. 21. Solder Data Sheet, Welco Castings, 2 Hillyard Street, Hamilton, ON, Canada. 22. Glazer, J., “Metallurgy of Low Temperature Pb-Free Solders for Electronic Assembly,” International Materials Reviews, 40(2):65–93, 1995. 23. Frear, D., and E. Bradley, Motorola, cited by A. Woosley, G. Swan, T. S. Chong, L. Matsushita, T. Koschmieder, and K. Simmons, “Development of Lead (Pb) and Halogen Free Peripheral Leaded and PBGA Components to Meet MSL3 260C Peak Reflow,” Electronics Goes Green, Fraunhofer Institute, Berlin, September 13, 2000. 24. Grusd, A., “Connecting to Lead-Free Solders,” Circuit Assembly, 10(8):32–38, August 1999. 25. Hwang, J. S., and R. M. Vargas, “Soldering and Surf,” Mount Technology, 5:38–45, 1990. 26. Guo, Z.,A. F. Sprecher, Jr., H. Conrad, and M. Kim, Proceedings of Materials Developments in Microelectronic Packaging: Performance and Reliability Conference, pp. 155–162, ASM International, Montreal, PQ, Canada, 19–22 August 1991. 27. Villain, J., O. Bruller, and T. Qasim, “Creep Behavior of the Lead-Free Solder Alloy Sn3.5Ag at High Homologues Temperatures using Laser Extensometry with Miniprobes,” Proceedings of SMT/ES&S/Hybrid 2000, Nuremberg, Germany, June 27–29, 2000. 28. Hunt, C., and D. Lea, “Solderability of Lead-Free Alloys,” Proceedings of Apex 2000, Long Beach, CA, March 2000. 29. Huang, B., and N.-C. Lee, “Prospect of Lead Free Alternatives for Reflow Soldering,” IMAPS, Chicago, October 1999. 30. Lotosky, P., “Lead-Free Update,” Tutorial at IMAPS-Brazil, São Paulo, Brazil, August 1–3, 2001. 31. Melton, C., “Reflow Soldering Evaluation of Lead Free Solder Alloys,” Proc. of IEEE 43rd Electronic Components and Technology Conference (ECTC’93), pp. 1008–1011, Orlando, FL, June 1993.
PREVAILING LEAD-FREE ALLOYS
13.61
32. Loomans, M. E., S. Vaynman, G. Ghosh, and M. E. Fine, “Investigation of Multi-component Lead-Free Solders,” J. Electronic Materials, 23(8):741–746, August 1994. 33. Vianco, P. T., F. M. Hosking, and J. A. Rejent, Proceedings: NEPCON West Conference, pp. 1730–1738, Anaheim, CA, Cahners Exposition Group, Des Plaines, IL, 1992. 34. Vianco, P. T., F. M. Hosking, and D. R. Frear, “Lead-Free Solders for Electronics Applications—Wetting Analysis,” Conference: Materials Developments in Microelectronic Packaging: Performance and Reliability, pp. 373–380, Montreal, PQ, Canada, August 1991. 35. Melton, C., A. Skipor, and J. Thome, Proceedings: NEPCON West Conference, pp. 1489–1494, Anaheim, CA, Cahners Exposition Group, Des Plaines, IL, February 1993. 36. Hernandez, C. L., P. T. Vianco, and J. A. Rejent, “Effect of Interface Microstructure on the Mechanical Properties of Pb-Free Hybrid Microcircuit Solder Joints,” Proc. of SMTA/IPC Electronics Assembly Expo, p. S19-2, Providence, RI, October 24–29, 1998. 37. Suganuma, K., “Microstructures of Lead-Free Solders and of Their Interfaces with Cu,” Proc. of the Third International Symposium of Electronic Packaging Technology, pp. 198–203, Beijing, China, August 17–21, 1998. 38. Vianco, P. T., and D. R. Frearr, “Issues in the Replacement of Lead-Bearing Solders,” JOM, 14–19, July 1993. 39. Humpston, G., and D. M. Jacobson, Principles of Soldering and Brazing, ASM International, Materials Park, OH, 1993. 40. Satoh, R., in Thermal Stress and Strain in Microelectronics Packaging, Lau, J. H., ed., Van Norstand Reinhold, New York, pp. 500–531, 1993. 41. Bader, W. G., Weld. Res. Suppl., 48:551s–557s, 1969. 42. Heinzel, H., and K. E. Saeger, Gold Bull., 9(1):7–11, 1976. 43. London, J., and D. W. Ashall, Brazing Soldering, 11:49–55, Autumn 1986. 44. Siow, K. S., and M. Manoharan, “Combined Tensile-Shear Fracture Toughness of a LeadTin and a Tin-Silver Solder,” Proc. of SMTA/IPC Electronics Assembly Expo, p. S19-3, Providence, RI, October 24–29, 1998. 45. Baggio, T., “The Panasonic Mini Disk Player—Turning a New Leaf in a Lead-Free Market,” Proceedings of IPCWorks ’99, Minneapolis, MN, October 27, 1999. 46. Toyoda, Y., “The Latest Trends in Lead-Free Soldering,” Proc. of International Symposium on Electronic Packaging Technology, pp. 434–438, Beijing, China, August 8–11, 2001. 47. Vianco, P.T., J.A. Rejent, I.Artaki, and U. Ray,“An Evaluation of Prototype Circuit Boards Assembled with a Sn-Ag-Bi solder,” Proceedings of IPCWorks ’99, Minneapolis, MN, October 22, 1999. 48. Mawer, A., and K. Levis, “Automotive PBGA Assembly and Board-Level Reliability with Lead-Free Versus Lead-Tin Interconnect,” SMTA International, Chicago, IL, September 24–28, 2000. 49. Bartelo, J. C., “The Effect of Temperature Range During Thermal Cycling on the Thermomechanical Fatigue Behavior of Selected Pb-Free Solders,” APEX, Long Beach, CA, 2001. 50. Syed, A., “Reliability of Pb Free Solder Joints for Area Array Packages,” APEX, San Diego, CA, January 18, 2001. 51. Elenius, P., and S. Yeh, “Lead Free Solder for Flip Chip and Chip Scale Packaging (CSP) Applications,” Proceedings of IPCWorks ’99, pp. S-03-2-1–S-03-2-6, Minneapolis, MN, October 23–28, 1999. 52. Frear, D. R., J. W. Jan, J. K. Lin, and C. Zhang, “Pb-Free Solders for Flip-Chip Interconnects,” JOM, 53(6):28–32, 2001. 53. Lauer, T., and S. Wege, “Behaviour of Lead-Free Solder Joints under Thermal and Mechanical,” Proceedings of SMT/ES&S/Hybrid 2000, Nuremberg, Germany, June 27–29, 2000. 54. Bradley, E., “Update on the State of Pb-free Solder Assembly Inside and Outside Motorola,” Hermes Symposium, Dublin, 2000.
13.62
CHAPTER THIRTEEN
55. Fu, H., and J. Liu, “The Development of Lead-Free Soldering,” Proc. of ISEPT, pp. 449–454, Beijing, China, August 8–11, 2001. 56. Shangguan, D., et al., “Application of Lead-Free Eutectic Sn-Ag Solder in No-Clean Thick Film Electronic Modules,” IEEE Trans. on Components, Packaging & Manufacturing Technology, Part B: Advanced Packaging 17(4):603–611, November 1994. 57. Artaki, I., A. M. Jackson, and P. T. Vianco, “Evaluation of Lead-Free Solder Joints in Electronic Assemblies,” J. Electronic Materials, 23(8):757–764, August 1994. 58. Yang, W,. L. E. Felton, and R. W. Messler, Jr., “The Effect of Soldering Process Variables on the Microstructure and Mechanical Properties of Eutectic Sn-Ag/Cu Solder Joints,” Journal of Electronic Materials, 24(10):1465–1472, October 1995. 59. Vincent, J. H., and G. Humpston, “Lead-Free Solders for Electronic Assembly,” Circuits Assembly, 38–41, July 1994. 60. Prismark Partners LLC,“Lead Free Electronic Products—The Sky is Clearing,” November 1999. 61. Handwerker, C., “NCMS Lead Free Solder Project: A National Program,” Proceedings of IPCWorks ’99, Minneapolis, MN, October 27, 1999. 62. Artaki, I., D. W. Finley, A. M. Jackson, U. Ray, and P. T. Vianco, “Wave Soldering with PbFree Solders,” Proc. Surface Mount International Conference, p. 495, San Jose, CA, August 27–31, 1995. 63. Kang, S. K., et al., “Pb-Free Solder Alloys for Flip Chip Applications,” 49th Electronic Components Technology Conference, San Diego, CA, June 1–4, 1999. 64. Mei, Z., F. Hua, and J. Glazer, “Sn-Bi-X Solders,” SMTA International, San Jose, CA, September 13–17, 1999. 65. “Lead-Free Solder Project Final Report,” NCMS Report 0401RE96, August 1997. 66. Feldmann, K., and M. Reichenberger, “Assessment of Lead-Free Solders for SMT,” Apex 2000, Long Beach, CA, March 2000. 67. Miller, C. M., I. E. Anderson, and J. F. Smith, “A Viable Tin-Lead Substitute: Sn-Ag-Cu,” J. Electronic Materials, 23(7):595–601, July 1994. 68. Technical Reports for the Lead-Free Solder Project, Properties Reports, “Room Temperature Tensile Properties of Lead-Free Solder Alloys,” Lead Free Solder Project CD-ROM, National Center for Manufacturing Sciences (NCMS), Ann Arbor, MI, 1998. 69. Senju, patent JP5050286, covers 3–5% Ag, 0.5–3% Cu, 0–5% Sb, balance Sn. 70. U.S. Patent 4,695,428, JW Harris Company patent. 71. Herbert, R., “Lead-Free Alloy Trends for the Assembly of Mixed Technology PWBs,” Proceedings: NEPCON West Conference, Anaheim, CA, February 27–March 2, 2000. 72. Adapted from Rae, A., and R. C. Lasky, “Economics and Implications of Moving to LeadFree Assembly,” Proceedings: NEPCON West 2000 Conference, Anaheim, CA, February 27–March 2, 2000. 73. Anderson, I. E., K. Kirkland, and W. Willenburg, “Implementing Pb-Free Soldering,” SMT Guide, June 2001.
CHAPTER 14
LEAD-FREE SURFACE FINISHES 14.1
INTRODUCTION
To accomplish a lead-free soldering environment, not only the solders used for forming solder joints but also the surface finishes of pads on printed circuit boards (PCBs) and leads of components have to be lead-free. This is based on the considerations of both environmental safety and joint reliability issues. At this stage, an SnPb surface finish is widely used in the electronics industry. Figure 14.1 shows the PCB surface finishes technology adoption status in the United States, published by IPC Technology Marketing Research Council.1 Although Sn-Pb hot-air solder leveling (HASL) still remains the dominant choice since 1998, its usage has definitely been declining with time even before the heat on lead-free soldering was felt by the industry. In Fig. 14.1, several lead-free surface finishes are cited, including organic solderability preservative (OSP), Ni-Au, and Pd. Because the pressure on lead-free soldering is increasing at a tremendous rate, many new lead-free surface finish technologies have emerged since the late 1990s. In this chapter, the options on lead-free surface finishes will be introduced, with the chemistry, process, and performance of the most promising choices discussed. Since the same surface finish may be used on products varying considerably in design and application, the considerations for selecting a surface finish need to address a wide range of possible applications, including solderability, compatibility with solder alloys, solder joint reliability, wire bondability, connector abrasion resistance, electrical contact resistance, shelf life, and contrast in automated optical inspection or registration system.
14.2 OPTIONS FOR PCB LEAD-FREE SURFACE FINISHES Table 14.1 lists the options of lead-free surface finishes for PCBs. The system is categorized by the key element used. Each category is further classified by the type of process and chemistry. Examples are given for certain groups.
14.3
OSP
Organic solderability preservative (OSP) refers to organic coating that is applied to PCB pads as a preservative for solderability, and is also referred to as antitarnish. It includes rosins, resins, and azole chemicals.2–17 Organic solderability preservative is widely used in Japan. Depending on the PCB type, single-sided PCBs use OSP as virtually the only means for preserving solderability. For double-sided, multilayer boards, 40 percent of the surface finishes are OSP types, as shown in Fig. 14.2.18 14.1 Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
14.2
CHAPTER FOURTEEN
FIGURE 14.1 Surface finish technology market study. (Source: IPC Technology Marketing Research Council, 1999.)
Within the azole group, benzotriazole, imidazoles, and substituted benzimidazoles are the most popular group, and will be discussed individually. Regarding the rosin or resin, it is often referred to as preflux, and will be discussed briefly. In general, the primary benefits of OSP surface finishes are low-cost and flat-surface.
14.3.1
BENZOTRIAZOLE
Use of benzotriazole (BTA) can be dated back to at least the early 1960s. Instead of relying on a physical layer that is mechanically applied onto the top of base-metal copper to provide protection against corrosion and oxidation, benzotriazole chemically reacts with cuprous oxide and forms polymeric copper salt, as shown in Fig. 14.3.19 These polymeric salt molecules align with each other and form a protective film composed of a semipermeable, colorless, three-dimensional polymeric layer on the surface of copper. Cuprous oxide is much faster in reacting with BTA than cupric oxides. Although this film thickness was considered a nominal 50 to 100 angstroms (Å), with an average of about 80 Å, the polymer thickness may continue to grow (within limits) if the metal is continually exposed to oxidant and triazole, especially at low (acidic) pH. This multilayer can be up to 5000 Å thick, depending on many factors, such as temperature, time, pH, corrodent, oxidation reduction environment, and so on.20 This continual growth can also repair defects in an existing film by reacting with entrapped BTA molecules.21 Additional work that was later done by Tornkvist et al. on the surface orientation of BTA suggests a mean orientation of the first BTA layer on a cuprous oxide surface. As shown in Fig. 14.4, the upper nitrogen allows the adsorbed BTA molecule to coordinate to another Cu(I) ion and thereby form a multilayer protective film, consisting of [-Cu(I)-BTA-]n chains.22 Some other surface absorption mechanism has also been postulated.23 However, the latter mechanism appears to allow monolayer BTA protection only.
14.3
LEAD-FREE SURFACE FINISHES
TABLE 14.1 List of Lead-Free Surface Finishes* Surface finish system OSP
Ni-Au
Finish process and chemistry
Example
Benzotriazole
COBRATEC, ENTEK CU-56
Imidazole
AT&T, Protecto 5630 (Kester)
Benzimidazole (substituted)
ENTEK PLUS CU-106A (EnthoneOMI), Glicoat (Shikoku)
Preflux (rosin/resin)
Sealbrite, Solderite RT-05R
Electrolytic Ni-Au, or EG Electroless Ni/electroless (immersion) Au, or ENIG Electroless Ni/electroless (autocatalytic) Au Electroless Ni/electroless (substrate-catalyzed) Au
Ag Bi
Electroless (immersion, or galvanic) Ag
Alpha Level (Alpha Metals); Sterling Silver (MacDermid)
Electroless (immersion) Bi Electrolytic Pd or Pd alloys
Pd
Electroless (autocatalytic) Pd Electroless (autocatalytic) Pd/electroless (immersion) Au
Ni-Pd
Electroless Ni/electroless (immersion) Pd Electroless Ni/electroless (autocatalytic) Pd Electroless Ni/electroless (autocatalytic) Pd/electroless (immersion) Au
Ni-Pd (X)
Electrolytic Ni/PdCo/Au flash (Electroless) Ni/(electroless) Pd-Ni/electroless (immersion) Au
Sn
Electrolytic Sn
Matte—Lucent (large, polygonized)
Electrolytic Sn
Florida CirTech
Electrolytic Sn Electroless (immersion) Sn
White (new)
Electroless (immersion) Sn
Gray (old)
Electroless (modified immersion + autocatalytic) Sn
Flat Solderable Tin (FST)—Dexter
Ni-Sn
Electrolytic Ni/electrolytic Sn
Satin bright Sn—Lucent ECS
Sn-Ag
Electrolytic Sn-Ag
96.5Sn-3.5Ag
Electrolytic Sn-Bi
90Sn-10Bi
Sn-Bi
Electroless (immersion) Sn-Bi
Motorola
Sn-Cu
Electrolytic Sn-Cu
99Sn-1Cu
Sn-Ni
Electrolytic Sn-Ni
* For multilayer designs, the sequence of materials starts from the bottom layer.
14.4
CHAPTER FOURTEEN
(a)
(b)
FIGURE 14.2 Organic solderability preservative in Japan. (a) Surface finishes for double-sided, multilayer boards, and (b) surface finishes for single-sided PCBs.
Substitution in the triazole ring (positions 1 and 2) results in a considerable decrease in inhibiting efficiency versus BTA. A methyl group in the benzene ring (positions 4 and 5) gives, on the other hand, a value of the inhibiting efficiency, which is even higher than that obtained for BTA. These results are closely related to the ability of the molecules to chemisorb and form stable film on the surface. Therefore, the inhibiting efficiency is 4Me- (30 Å) and 5Me-BTA (70 Å) > BTA ( 1Me- and 2Me-BTA.22 Fabrication Process. The application process of BTA is detailed in Fig. 14.5.20 After the initial acid clean, the copper surface is microetched to produce a subtle roughness in order to enhance subsequent solder bonding, as well as possible in-
LEAD-FREE SURFACE FINISHES
14.5
FIGURE 14.3 Scheme of triazole inhibitor mechanism and chemistry.
circuit test probe contact. The etchant is then removed by deionized (DI) rinse, followed by acid rinse, DI rinse, then BTA coating, and completed with DI rinsing and drying. This OSP application process is slightly simpler than the HASL application process: (1) acid clean, (2) water rinse, (3) etch, (4) water rinse, (5) flux application, (6) preheat to 105 to 150°C, (7) solder coat (2 to 10 s/250 to 260°C), (8) excess blowoff, (9) water rinse, and (10) dry.23 Following is a list that compares the application processes for HASL and OSP:
FIGURE 14.4 Surface orientation of BTA on copper.
14.6
CHAPTER FOURTEEN
FIGURE 14.5 Deposition process for BTA on copper.
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
HASL Acid clean Water rinse Etch Water rinse Flux application Preheat 105–150°C Solder coat (2–10 s/250–260°C) Excess blowoff Water rinse Dry
1. 2. 3. 4. 5. 6. 7. 8. 9.
OSP Acid clean Water rinse Microetch Water rinse Acid rinse Water rinse OSP coat DIW rinse Dry
For BTA coating process, 2.5 percent concentration is satisfactory. The shelf life of BTA-coated PCB is about 2 years at normal storage condition. Generally, the CuBTA complex thickness is 40 to 140 Å.24 However, under high humidity, the shelf life may be as short as 3 to 6 months only.23
FIGURE 14.6 Effect of temperature on the oxidation of BTA-treated and -nontreated copper.
LEAD-FREE SURFACE FINISHES
14.7
FIGURE 14.7 Solderability for various treatments of copper foil.
Performance. Walker studied the effect of temperature on the oxidation extent of BTA-treated copper and nontreated copper by monitoring the weight gain of copper after thermal aging. Results indicate that BTA is much more effective in reducing oxidation at lower temperature than higher temperature, as shown in Fig. 14.6.25 Therefore, aging at 200°C for 400 min, the treated copper had no weight gain, whereas the nontreated sample exhibited a weight gain of about 2 mg/80 cm2. The difference between treated and nontreated copper diminished with increasing temperature. After aging at 400°C for 400 min, the BTA-treated copper only oxidized slightly less than the plain copper, and both exhibited a weight gain of around 50 mg/80 cm2. Obviously, increasing oxidation will deteriorate the solderability. Thwaites investigated the effect of aging on solder spread and wetting time of copper foil with various treatments, with results shown in Fig. 14.7.26 Benzotriazole, tin-lead coating, and preflux (resin lacquer) all preserved solderability very well in terms of area of spread after 6 months of normal storage. The uncoated copper showed no spread at all after the same period of storage. However, the difference between various treatments becomes more pronounced in terms of wetting time, with tin-lead being the most effective in retaining the wetting time, followed in order by resin lacquer, BTA, and uncoated copper. After 6 months’ storage, tin-lead coating increased only slightly in wetting time, while BTA increased from about 0.2 to around 5 s. For uncoated copper, no wetting can be observed at all.
14.3.2
IMIDAZOLES
The chemical structures and oxidation inhibition mechanism of alkylimidazoles is shown in Fig. 14.8. Similar to BTA, alkylimidazole reacts with copper oxide and forms polymeric alkylimidazolium copper film on the surface of copper. The
14.8
CHAPTER FOURTEEN
FIGURE 14.8 Scheme of chemical structure and inhibition mechanism of alkylimidazole.
thickness of this compound is approximately 100 Å.27 The reaction takes place almost immediately, on the order of seconds. Once the surface reaction has been completed, extended dwells in the bath do not advance the reaction; the process passivates. Consequently, the process is easily controlled. The chemistry is aqueous based and can be stored for long periods of time. The most critical part of the process is rinsing between the conditioner and OSP components of the process.27 The efficiency of oxidation inhibition of imidazole is temperature-sensitive. Ray et al. studied the effect of heat treatment on oxide thickness for imidazolecoated copper, with results shown in Fig. 14.9.5 At a low temperature (e.g., 100°C), the oxidation inhibition efficiency is very high, and no oxidation can be discerned after 4 h. The efficiency declines with increasing temperature. At 200°C, the oxide thickness increases about six times after 4 h of aging. The sensitivity toward aging temperature can result in poor wetting due to storage and due to multiple soldering processes, as demonstrated by Fig. 14.10.12 In reflow applications, poor wetting can cause opens for joints with the use of a zipper stencil aperture design.
14.3.3
BENZIMIDAZOLES
Use of benzimidazoles as an OSP for the electronics industry evolved from imidazole chemistry. In the 1970s, Sanwa of Shikoku Chemical Co. patented alkylimidazole and benzylimidazole coating. These were adopted in the United States in 1985. The coating thickness is typically around 0.3 µm, and may range from 0.2 to 0.5 µm. Product suppliers include MacDermid (M-Coat), Chemcut (Schercoat), Kester (Protecto), and Enthone-OMI (Entek).23,28 The protection mechanism is similar to that of imidazole, with a polymeric imidazolium copper film, as shown in Fig. 14.11.20
LEAD-FREE SURFACE FINISHES
14.9
FIGURE 14.9 Effect of heat treatment on oxide thickness for imidazole-coated Cu.
FIGURE 14.10 Examples of good wetting and poor wetting for PCBs with imidazole coating. (a) Incomplete wetting of pads, (b) open caused by a zipper stencil aperture pattern, (c) completely filled plated-through-hole (PTH) via, and (d) partially filled PTH via.
14.10
CHAPTER FOURTEEN
FIGURE 14.11 Inhibition mechanism of substituted benzimidazole on copper surface.
In 1997, Sirtori et al. from IBM-Semea studied the stability of a 2-butyl-5-chlorobenzimidazole film (BCB) before and after a thermal treatment, which simulated the annealing of a typical soldering process. The chemical bonds inside the BCB film and between this film and the copper substrate were investigated by x-ray photoelectron spectroscopy (XPS). The thickness of that film was measured by optical ellipsometry. Protection of the film versus oxidation and its physical modification during the thermal treatment was also investigated. The XPS valence data showed that a part of the chlorine was linked to the benzene ring, and the other part formed either copper chloride or another chloride linked to the nitrogen of the BCB molecule.The passivation film hardened and became an electric insulator during the thermal treatment, so that any successive electrical test was impaired.29 Introduction of substituted benzimidazoles (BAs) results in an improvement in soldering defect rate. In a study conducted by Siemens Information and Communication Networks, Inc., the board quality was the leading cause of wave-soldering defects in their experiment. When Entek 106A, a substituted BA, is used as an OSP, the bridging defect rate was slightly higher than HASL, and increased slightly with increasing residual oxygen level in the soldering atmosphere, as shown in Fig. 14.12.30 The solder mask quality is a more dominant factor than the oxygen level, with an improperly cured solder mask causing much more crossing and bridging than the oxygen does. Fabrication Process. The fabrication process of Entek 106A is shown in Fig. 14.13.20 The process is similar to that of other azole coating processes. Typically, BA chemistry deposits 0.1 to 0.4 µm and relies on ionic copper in the working solution to develop the protective coating. However, there is one limitation about this process: the mixed-metal surfaces can become stained or discolored when processed through the OSP. The discoloration is due to nonuniform deposition of the OSP on the different metals due to the presence of the copper ion.31 This problem can be corrected by utilizing a copper-free formula. By pretreating the copper surface with an alkaline precoat solution, the copper on the printed wiring board (PWB) becomes available to build a consistent OSP coating. The precoat is mildly alkaline and will not discolor active metal surfaces such as gold or aluminum, even if contaminated with copper. The ionic copper can be eliminated from the OSP solution since the copper on the PWB can react with OSP to form the desired coat-
LEAD-FREE SURFACE FINISHES
14.11
FIGURE 14.12 Effect of nitrogen purity on the wave-soldering defect rate. Most of the defects are bridging.
ing.20 This new process produces a slightly thinner OSP layer, about 0.2 to 0.25 µm thinner than the standard process. However, the resistance against being burned out by heat treatment under reflow atmosphere is still comparable with the standard OSP, as shown in Fig. 14.14. The new process produces a very thin, stain-free OSP coating on an active metal surface. The thickness is less than 0.05 µm and is virtually absent on the metal surface, as shown in Fig. 14.15 when tested on a Ni-Au or matte Ni surface.20 Like the original, it provides similar protection against incidental handling contact, 12 months’ shelf life, and similar solderability even with mixed technology in a no-clean air assembly environment. Because of this new mechanism, a thinner but equally robust OSP coating is possible while being selective to copper only. The OSP fabricated with a new process is more heat-resistant than the standard OSP, as shown in Table 14.2.32 Among the OSPs tested, BTA is the least heatresistant. Azoles react with metallic as well as with cuprous or cupric oxides. Ray et al. reported that films grown on the native oxide are more compact and polymerized to provide greater corrosion protection. Films grown from acidic media are thicker, but more permeable to oxygen.
FIGURE 14.13 Typical substituted BA application process.
14.12
CHAPTER FOURTEEN
FIGURE 14.14 Organic solderability preservative coating thickness on copper using the standard process and the new process.
Imidazole film thickness is typically 50 Å. Benzimidazole or alkylated BAs could range from 100 to 10,000 Å in film thickness, depending on solution concentration, immersion time, and other factors. The copper-BA film is polymerized and posesses a high degree of submicroscopic physical porosity. The alkylated BA is more nonporous and less polymerized. Unheated thick films may or may not be solderable, depending on azole chemistry. The thin films (10 percent) content. Midphosphorus is the most commonly used for microelectronics applications.52 Atotech investigations have shown that the phosphorus content in the upper range of 8 to 10 percent gives many of the desired properties that are needed for the ENIG finish.The phosphorus content in an EN deposit normally increases 1 to 2 percent as the nickel metal turnovers (MTOs) increase, as shown in Fig. 14.23.49 Cullen noted that, in theory, 1 µm of electroless NiP will prevent any migration under soldering conditions. In addition to the migration of copper through nickel, the solder will also dissolve the nickel during reflow. The dissolution also occurs to a much lesser extent after soldering. Based on the dissolution rate of nickel into solder to form the Ni3Sn4 intermetallic, 0.5 µm is considered more than enough nickel thickness to perform this function at typical soldering conditions.50 As mentioned earlier, the commonly employed EN thickness is 2.5 to 5 µm. Immersion Gold Process. An immersion reaction is an oxidation/reduction system in which a metal ion in solution is reduced to the metal at the expense of the surface metal, which is oxidized to an ion. The exchange occurs in one direction only and is determined by the relative positions of the interacting metals in the electromotive force series. In principle, any metal ions in solution higher in the electromo-
FIGURE 14.23 Phosphorus content of EN as a function of nickel MTO.
14.24
CHAPTER FOURTEEN
tive series will displace any base metal below it in the series. Since the base metal is oxidized, it is also considered a corrosion process for the base metal. In immersion plating of gold over nickel, gold ions from solution are reduced to gold metal. The electrons needed for this reduction are supplied by the nickel substrate itself. Immersion deposition or displacement plating will cease as soon as the substrate is completely covered by the immersion coating52: Νi0 + 2Αu+ → Νi2+ + 2Αu0 or Νi + 2Αu(CΝ)2− → Νi2+ + 2Αu + 4CΝ− The typical thickness of gold flash layer is less than 5.0 µin, or 0.125 µm.53 The immersion gold thickness used in the electronics industry ranges from 2 to 8 µin (0.05 to 0.2 µm). The studies carried out by Atotech indicated that an ideal gold thickness is 3 to 4 µin, with an operating window of 2.5 to 4.5 µin. Too low gold thickness will result in oxidation of the nickel and consequently poor wetting of the solder during assembly.Too high gold thickness will result in high levels of attack on the nickel surface, resulting in the possibility of interfacial fracture leading to poor solderability.49 At 150 µin (3.75 µm) of nickel and 3 to 5 µin (0.075 to 0.125 µm) of gold, the copper circuits are completely encapsulated. The solderability of the nickel is preserved by the gold to at least a 1-year shelf life. Performance. Similar to EG, ENIG offers advantages such as coplanarity, Al-wire bondability, and the ability to survive multiple soldering cycles (up to three reflows). Again, the nickel layer allows multiple hand reworks without copper dissolution being a factor. As described in the EG section, the nickel also acts like a rivet to improve through-hole thermal integrity.49 The gold finish on the nickel has good reflectivity. This allows this finish to be suitable for automated optical inspection (AOI). The difference in color between the gold and the solder after assembly makes for ease of visual inspection at that stage. The electrical conductivity of the finish does not interfere with electrical testing before or after assembly.53 Electroless nickel/immersion gold is the more common Ni/Au surface finish specification. However, if not properly fabricated, its thin or porous gold can allow nickel to migrate to the surface and oxidize, causing a nonsolderable surface mount pad. The nickel thickness and phosphorus content in the nickel also play an important role in obtaining reliable solder connections.11 Other problems encountered by ENIG include black pad, skip plating, extraneous plating, and embrittlement of the solder mask. Black pad is a symptom that is associated with some weak solder joints formed on ENIG surface finishes. After the weak joint is ruptured, the exposed nickel pad is black. The black-pad defect was found to be due to a hyperactive corrosive immersion gold (IG) process that changes the near-surface microstructure of P-Ni into one with a marginal to total nonwetting state. Figure 14.24 shows a high-magnification cross-sectional view of a solder joint formed on black pad.54 The black-pad defect shall be classified in terms of hyperactive corrosive activity.55 Charge buildup near the module boundaries triggers hypercorrosion. Long traces (with several ohms of resistance) with differential pad plating surface area may induce the condition of charge buildup by the galvanic reaction. The minor difference in electrical potential among the packages and leads caused different types or different degrees of chemical reactions between the ENIG bath, resulting in the preferential occurrence of black pads. Johal reported that a rapid buildup of an immersion gold layer encourages a higher attack on the E-Ni. This rapid attack occurs along the E-Ni grain boundaries, resulting in a possible black-pad defect.49
LEAD-FREE SURFACE FINISHES
14.25
FIGURE 14.24 A 2000× cross-sectional view of a gullwing solder joint formed on a black pad. Severe corrosion can be noted at the side of the pad surface.
Skip plating is seen on some boards as areas with missing nickel on copper pads, as shown in Fig. 14.25.49 Extraneous plating is an excessive buildup of nickel on copper, as shown in Fig. 14.26.49 Both symptoms can be attributed to an activation process of the copper surface. As discussed earlier, activation of the copper surface for EN plating can be achieved in several ways, and is normally done by seeding a noble metal such as palladium from an acidic solution of palladium sulfate/chloride. However, there is a balance of the amount of palladium seeding in combination to the activity of the nickel bath that must be maintained; otherwise, the skip plating or extraneous plating will occur.49 Besides the unbalanced activation process as a cause of skip plating, it was also hypothesized that a static is created in the solder mask operation on certain capacitive areas of the circuitry and that this static attracts volatiles during the mask cure operation and is the underlying cause of skip plating.56 Embrittlement of the solder mask is mainly caused by the EN bath. This bath, as well as the immersion gold bath, is operated at around 82°C with a 20-min soak time for nickel and 10 min for gold. Besides being affected by the elevated temperature, the strong reducing agent may also be absorbed into the soft porous mask, making it brittle and, consequently, may lead to peeling at the mask surface junction. Baking the boards after the ENIG process may minimize this effect.56
14.4.3
ELECTROLESS Ni/ELECTROLESS (AUTOCATALYTIC) Au
Most wire-bonded COB assemblies include SMT components on the same substrate. Thick gold is necessary for the wire-bonded devices, but is unacceptable for the SMT devices. The electroless gold process can be selectively plated over the
14.26
CHAPTER FOURTEEN
FIGURE 14.25 Example of skip plating in the ENIG process.
immersion gold. Following the immersion gold, a plating mask is applied to the panel exposing only the area requiring the thicker gold. The electroless gold is then deposited onto the unmasked areas. The finished gold thickness is 20 to 60 µin over nickel. The plating chemistry of electroless gold (cyanide-based) can be expressed by the following: 3 − − R2ΝΗ ⋅ ΒΗ3 + 4ΟΗ− + 3Αu(CΝ)2− → R2ΝΗ + ᎏ 2 Η2 + ΒΟ2 + 2Η2Ο + 3Αu + 6CΝ
14.5
IMMERSION Ag
Immersion silver is another lead-free surface finish formed by galvanic reaction. Silver is selected due to the following four considerations: (1) electromotive potential of Ag (+0.80 V) relative to copper (+0.344 V) allows use of immersion deposition process; (2) high electrical conductivity of Ag is compatible with touchpad applications, in-circuit probe test processes, and signal transmission requirement; (3) Ag is a noble metal and thus promises good stability; (4) Ag dissolves into solder quickly and thus promises good solderability.1 Fabrication Process. The immersion silver coating process consists of four baths, a precleaner, a microetch followed by the conditioner, and finally the plating bath. Table 14.6 shows the detailed process conditions for organic modified silver (Sterling™ Silver).1 The Ag coating is applied by an immersion process that exchanges
14.27
LEAD-FREE SURFACE FINISHES
FIGURE 14.26 Example of extraneous plating in the ENIG process caused by overactivation of copper.
copper from the base metal for silver in the silver nitrate bath, as illustrated by Fig. 14.27.1 As discussed in the section on the immersion Au process, a particular benefit of immersion processes is the self-terminating feature; that is, the process terminates when the coating completely covers the base material. As a result, the plating thickness is consistent and easily controlled.54 The process can be horizontal or vertical, with 8 min of cycle time for a conveyorized process. Typically, 120 panels/h can be treated in 8-m equipment at 50°C process temperature. Figure 14.28 shows an example of PCB with an immersion Ag surface finish.1 Silver finish is applied after the solder mask, with the silver deposited on the exposed copper surface. Since silver surfaces are readily tarnished, an organic inhibitor is included in the plating bath to protect the surface. The thickness of TABLE 14.6 Immersion Process Conditions for Sterling™ Silver Process
Temperature (°C)
Conveyorized
Immersion
Sterling™ acid cleaner
50
30 s
5 min
Sterling™ surface prep
40
60 s
60 s
Sterling™ predip
30
30 s
30 s
Sterling™ Silver
50
60 s
60 s
14.28
CHAPTER FOURTEEN
organic modified silver is typically 0.2 to 0.3 µm, although other thickness ranges (e.g., 0.08 to 0.16 µm) have also been used. Obviously, the coplanarity will be excellent for such a thin coating. Figure 14.29 shows the topology of the organic modified silver.1 Inclusion of an organic inhibitor is fairly evenly distributed in the Ag layer down to a depth of about 3 µin (0.075 µm), as indicated by the FIGURE 14.27 Copper displaced by silver in Auger depth profiling for organic modithe immersion silver process. fied silver finish (see Fig. 14.30).1 For Alpha-Level immersion silver, the coating consists of a layer of silver approximately 4 to 5 µin (0.1 to 0.125 µm) thick with a thin inhibitor layer superimposed, as shown in Fig. 14.31.57 The inhibitor layer is essentially an OSP, which is approximately 5 Å thick.58 Microetch. The immersion Ag finish can be produced as a shiny or a matte finish. This shiny or matte surface characteristics are due to the surface roughness, and are controlled by the etching rate of copper. This microetch step provides a secondary cleaning of the copper surface. In addition, it also microroughens the surface and increases the surface area. As a result, after the immersion Ag coating process, the
FIGURE 14.28 Picture of a PCB with an immersion Ag surface finish.
LEAD-FREE SURFACE FINISHES
14.29
FIGURE 14.29 Topology of organic modified silver coating.
surface with increased surface roughness or greater surface area appears as a matte finish, and the surface with minimal surface roughness appears bright and shiny. Plating Chemistry. The reaction of the immersion Ag process can be expressed as follows: Cu + 2Αg+ → Cu2+ + 2Αg
FIGURE 14.30 Auger depth profiling for organic modified silver finish. The silver deposit contains 2 to 4 percent carbon (by wt%) or approximately 30 percent (by at%). (Source: Auger analysis data provided by Arch Chemicals, Inc.)
14.30
CHAPTER FOURTEEN
FIGURE 14.31 Schematic of alpha-level immersion Ag.
The chemicals used in the plating bath are listed in Table 14.7. Ag coating thickness is affected by time, temperature, pH, and Ag ion concentration of silver nitrate bath.54 Figure 14.32 shows the effect of immersion time on Ag coating thickness.59 Performance. Perhaps the first concern about immersion Ag is the potential for Ag migration. Cullen investigated the Bellcore TR-78 electromigration performance of immersion silver (Sterling™ Silver) together with copper and HASL.1 Results indicate that, at 85°C/85%RH and 10 VDC bias condition, immersion silver exhibits a resistance value higher than that of copper, and HASL finishes at both 96 and 596 h, as shown in Fig. 14.33, thus eliminating the concern on Ag migration. Figure 14.34 shows the posttest coupon from the electromigration test, displaying no sign of dendrite formation. Results on a surface insulation resistance (SIR) test also show a safe pass for Bellcore TR-78 specifications. Chada et al., from Motorola, also reported that the immersion Ag surface finish performs adequately in the SIR and electromigration (EM) tests and is not readily prone to dendritic growth in the presence of high humidity.60 However, ENIG and OSP are superior in the water droplet conditions simulating condensation and are less likely to electromigrate under those circumstances.
TABLE 14.7 Chemistry of Immersion Ag Plating Bath Chemicals
Functions
Ag
Metal source, 0.46 V relative to Cu
HNO3
Produce Ag anion, and accelerate reaction
Cu complexation
Prevent the copper in solution to affect reaction
Inhibitors
Prevent bath sensitivity to light, and assure deposit uniformity
Surfactants
Prevent electromigration and inhibit tarnish
Buffers
Control pH of bath
LEAD-FREE SURFACE FINISHES
14.31
FIGURE 14.32 Effect of immersion time on Ag coating thickness.
The solderability of immersion Ag finish has been tested by Beigle by evaluating the hole-filling performance at wave soldering. Results indicate that immersion Ag is almost as good as HASL Sn-Pb, and is considerably better than OSP for both steam-aged and unaged samples, as shown in Table 14.8.59 Beigle also studied the wetting force of immersion Ag, HASL, two OSPs, and electrolytic Ni-Au surface finishes in meniscograph test with five different aging treatments: (1) fresh, (2) three reflows, (3) 40°C/93% RH for 96 h, (4) 40°C/93% RH for 96 h, followed by three reflows, and (5) 150°C for 2 h. Results indicate that the wetting force for all systems is comparable for fresh samples. Aging treatment results in a declining wetting force for systems other than HASL. Immersion Ag is less sensitive to aging treatment than both electrolytic Ni-Au and OSPs. The wetting defect rate of reflow soldering is expected to be closely correlated with wetting force performance. Cullen studied the wetting defect rate (IPC-J-STD 003 Test F) of a convection reflow soldering system for OSP, ENIG, and immersion Ag with three different thicknesses. Results indicate that only the OSP finish exhibits wetting defects, as shown in Fig. 14.35. Gordon et al. also reported that the solderability of immersion silver is relatively insensitive to storage at 85°C/85% RH conditions. Depending on the type and thick-
FIGURE 14.33 Electromigration data of immersion Ag (Sterling), copper, and HASL at 96 and 596 h, per Bellcore TR-78 specifications: 85°C/85% RH, 10 VDC bias.
14.32
CHAPTER FOURTEEN
FIGURE 14.34 Posttest coupon from electromigration test, indicating no sign of dendrite formation from the silver-finished traces.
ness, however, the immersion silver is sensitive to assembly processes. Immersion silver has demonstrated superior moisture and insulation resistance behavior compared with HASL, and modules manufactured with an immersion silver finish have passed all appropriate module validation testing for automotive electronic applications.61 Chada and Bradley reported sensitivity of immersion Ag toward corrosive atmosphere storage condition.60 In their work, wetting and spreading of both Pbcontaining and Pb-free solder pastes over the immersion Ag surface are adequate even when the surface is mildly corroded. However, ENIG and OSP exhibit greater solder spread than immersion Ag finish for all testing conditions studied, although the silver was more consistent. Also, if subjected to a corrosive environment for extended periods of time (>96 h flowing mixed gas), wetting deteriorates drastically TABLE 14.8 Effect of Surface Finish Type and Aging Treatment on the Hole-Filling Performance Percentage of hole fill Coating
No aging
Steam-aged
Silver-plated
99.7
99.9
100
99.97
93.57
92.62
HASL Organic A
LEAD-FREE SURFACE FINISHES
14.33
FIGURE 14.35 Effect of PCB surface finish types on solderability defect rate per IPC-J-STD-003 Type F in convection reflow soldering process.
for immersion Ag and ENIG finishes. This sensitivity of immersion Ag toward corrosive storage environment is consistent with the findings of Reed, who reported that immersion Ag is sensitive to corrosive environments62: the solderability of immersion silver is severely impacted by exposure to Cl2, SO2, and NO2 gases in the presence of water vapor. It is recommended by Reed that boards be packaged in polyethylene bags no matter what the storage environment is. If the raw boards are stored in dry air, a 1-year shelf life can be expected. Reed also noted that exposed silver after assembly will corrode, although the products of this chemical attack appear to be benign for SIR and dendritic growth. Regardless, it may cause difficulties for field repairs. As to the unprotected immersion silver, the test points and etched-on symbols will tarnish, and the solder mask is considered an effective protectant for silver in corrosive environments in service. Since the amount of silver on the board termination is estimated to contribute less than 0.1 percent in a typical 20-mil pitch solder joint, no effect is expected on the solder joint life.62 Immersion Ag is also good for ultrasonic Al wire-bonding applications. Figure 14.36 shows comparison of clad Al pad and immersion Ag (Sterling Silver) finish on ultrasonic 10-mil Al wire bonding.1 Results indicate that immersion Ag is slightly better than clad Al, and thus it is adequate for wire-bonding applications. Cullen also reported the test results on touchpad applications. In this work, the contact resistance of four different surface finishes following 100,000 touchpad activations was compared, as shown in Table 14.9.1 Immersion Ag, electroless Pd, and ENIG remain very good electrical contact, while the resistance of conductive carbon increases to 0.28 mΩ. In-circuit test is another important criterion to be met by any PCB surface finishes. Gordon et al. studied the in-circuit test performance of immersion Ag.61 In their work, the test pad was probed with blade probes made of heat-treated steel, coated with gold on top of a hard-nickel finish. For each probe, a spring force of 8.1 oz was provided. Results indicate that percent of reseats and percent of failures
14.34
CHAPTER FOURTEEN
FIGURE 14.36 Effect of surface finish types on ultrasonic 10-mil Al wire-bonding performance.
decrease with increasing Ag thickness and increasing surface roughness (or etching rate).61 This increase in surface roughness appears to be crucial for enhancing the contact between pad and probe, at least in the case of blade probes. Chada et al. observed that immersion Ag/BGA solder joints appear to have lower load levels at failure than OSP and ENIG finishes in the as-reflowed condition. The effects of silver thickness and peak reflow temperature are insignificant on the reliability. However, solid-state aging and multiple reflows lead to a lowering of failure load.60 In Parker’s study at Viasystems, the rupture strength of the joints formed on immersion Ag finish consistently exceeds 1.0 lb, which is within the range generally attributed to a joint formed on an HASL surface. More important, however, is the fact that the rupture always occurs at the lead interface and never at the pad/board interface. This indicates that the bond at the pad is superior to that formed at the lead.54 Chase et al., at Raytheon and Nokia, compared the effect of surface finishes, including immersion Ag, Ni/Au (ENIG), and solder HASL on the second-level reliability of fine-pitch area array assemblies.63 A temperature cycling test with a temperature range of −40 to +125°C was used, with dwell times of at least 20 min at high temperature and 15 min at low temperature. Chamber temperature ramp rates were 8 to 10°C/min. The average time for 1 cycle was 75 min. Results indicate that, for 144 I/O and 0.8-mm-pitch BGAs, the reliability increases in the following order: Ni-Au < immersion Ag < HASL, as shown in Fig. 14.37.63 However, it should be noted that TABLE 14.9 Contact Resistance of Touchpads After 100,000 Activations Surface finishes Conductive carbon
Contact resistance (mΩ)
Note
0.2773
Heavy gloss on one pad, slight gloss on all others. Circuit board between one pad is worn to shine.
Immersion Ag (Sterling)
0
Pads slightly dulled.
Electroless Pd
0
Green circuit between is turning brown.
ENIG
0
Some nicks and cuts in gold pads.
14.35 FIGURE 14.37 Comparison of 144 I/O BGA test results by surface finish. The temperature cycling range is −40 to +125°C.
14.36
CHAPTER FOURTEEN
some premature failure points of the immersion Ag system started from 254 cycles were considered to fall in another failure mechanism and were not plotted. These premature failures were attributed to the presence of postassembly hairline cracks on the board side; the cause of formation of those hairline cracks is not understood yet. For 156 I/O with 1.0-mm-pitch BGA system, immersion Ag is comparable with HASL and is better than ENIG.
14.6
IMMERSION Bi
Immersion Bi was introduced in 1996 as a nonprecious metal surface finish intended to address the coplanarity problem experienced by HASL in fine-pitch applications.59 Fabrication Process. The immersion bismuth process features three steps: 1. An acidic cleaner that removes surface oils and solder mask residue. 2. Microetch step that prepares the copper substrate topography for the deposit of immersion bismuth. For the bismuth to deposit on copper, the bath is highly acidic. 3. Bi plating step. The immersion bismuth bath is operated at 50°C, and a typical contact time of 1 to 2 min to produce a deposit of pure metallic bismuth onto the copper surface. Similar to other immersion processes, the reaction is complete when the copper can no longer be released and the surface is completely covered. Performance. The plated Bi metal is uniform dark gray, and easily distinguished from the substrate copper. During aging or thermal excursions, the Bi finish becomes more copperlike in appearance. This is believed to be caused by diffusion of the bismuth into the underlying copper. Beigle considered this to be a cosmetic effect and has no major impact on solderability under normal conditions.59 The solderability of immersion Bi was found to be better than Pd in a throughhole filling test. For as-plated finishes, Bi exhibited 99 percent hole fill, while Pd only displayed 60 percent hole fill.59 In another experiment comparing Bi with OSP, PCBs with through-holes varying in hole size from 0.062 to 0.022 in were used to examine the hole fill with 11 different no-clean fluxes. A highly activated organic acid–based flux was used as a control. All boards were preconditioned with two air-atmosphere heating cycles with a profile having a peak temperature of 220°C and a dwell time of 40 s. At wave soldering, each board was applied with 600 to 800 µg/in2 of flux. The soldering performance was evaluated using complete topside pad coverage as the acceptance criteria. Results indicate that immersion Bi is less sensitive than OSP to flux selection, as shown in Fig. 14.38.59 The solderability of immersion Bi for reflow applications was also reported by Beigle.59 The spread performance of solder paste was determined on 20-mil-pitch QFP pads with 70-mil length, using the following procedure: 1. Steam-age half of the test boards. 2. Precondition test boards—one infrared reflow pass in air. 3. Screen solder paste.
LEAD-FREE SURFACE FINISHES
14.37
FIGURE 14.38 Effect of flux selection on the hole filling yield of immersion Bi and OSP finishes.
4. Reflow solder in the air. 5. Inspect and measure solder wetting distance. Results indicate that the reflow spread performance decreases in the following order: HASL > immersion Ag > OSP > immersion Bi, as shown in Table 14.10.59 All surface finishes are considered acceptable. The temperature cycling reliability of solder joints on immersion Bi has been studied by Beigle and Guy with the use of LCCC68 component.59,64 Both works indicate that immersion Bi provides comparable temperature cycling performance to HASL. For instance, Beigle used thermal cycle condition from −55 to 125°C with a 0.5-h ramp and an additional 0.5 h at temperature for 1000 cycles. The continuity of each component was monitored during each cycle, and resistance over 600 Ω was considered open and as a component failure. Results indicate that within statistical significance, immersion Bi is comparable with HASL in reliability, as shown in Fig. 14.39.59 The wire bondability of immersion Bi is very poor. No adhesion can be registered in a pull test of gold wire-bonding attempt. One of the concerns about TABLE 14.10 Reflow Spread immersion Bi finish is sensitivity toward Performance of Various Surface Finishes Pb. When used with Sn-Pb solder alloys, the ternary eutectic alloy 8Sn-52Pb-40Bi Mean solder wetting distance (melting point 95°C) formed may cause Coating Not aged Steam-aged early failure and extensive porosity durBismuth 65.85 64.72 ing temperature cycling or service. The failure mechanism induced by the formaSilver 68.46 68.24 tion of ternary eutectic alloy 8Sn-52PbHASL 69.28 68.39 40Bi was elucidated by Mei et al. on OSP 67.86 66.19 eutectic Sn-Bi system.65
14.38
CHAPTER FOURTEEN
FIGURE 14.39 Solder joint reliability in temperature cycling of −55 to 125°C test using LCCC68 component soldered on various surface finishes.
14.7
Pd
Palladium (Pd) is an attractive, cost-effective alternative finish for printed wiring boards (PWBs) that is both solderable and wire bondable. Similar to gold, Pd is a noble metal; therefore, it is stable against many chemical reactions, such as oxidation. In addition, Pd exhibits several advantages over Au as a potential surface finish constituent: 1. Pd is cheaper than Au. 2. Pd exhibits a density 38 percent lower than Au (12.02 versus 19.32 g/cm3), thus further reduces the cost of Pd needed for surface finish applications. 3. Pd displays a tensile strength about 35 percent higher than Au. 4. Pd exhibits a hardness at 250 to 290 Vickers, about twice that of copper and three times that of gold, thus making it more suitable for contact purposes. 5. Pd dissolves in molten 60Sn-40Pb at a much lower rate than Au (about 0.01 versus 5 µm/s), thus it is less prone to contaminate the solder pot.66,67
14.7.1
ELECTROLYTIC Pd WITH OR WITHOUT IMMERSION Au
Electrolytic Pd or electrolytic Pd with Au flash provides a thin deposit on top of copper. The Pd is less than 0.5 µm, and is typically 0.25 µm in thickness. The Au flash is 0.025 µm in thickness.The fabrication process is described in the following subsection. Fabrication Process. The electrolytic Pd plating process is integrated with the PCB patterning process as follows: 1. 2. 3. 4. 5.
Cu plate Pd plate Resist strip Cu etch Solder mask application
14.39
LEAD-FREE SURFACE FINISHES
During the plating cycle, Pd is applied immediately after acid copper plating. The dwell time in the Pd is 1 to 2 min. After Pd plating, the photoresist is stripped and the boards are processed through the copper etcher, and finally the solder mask is applied. The Pd etch resist can then be activated for further processing if required. For instance, electroless Pd and/or immersion Au can be selectively applied, depending on the specific solderability and bonding requirements.68 Performance. The major advantage of a Pd finish is its use as an etch resist replacement for Sn-Pb. It reduces the manufacturing steps and the cycle time by reducing the plating time, and it eliminates the Sn-Pb stripping step. Other advantages provided include: ● ● ●
It is wire bondable. It is solderable. It has uniform thickness and excellent coplanarity, thus making it suitable for highdensity interconnect applications.
The solderability stability of a Pd finish against storage conditions is evaluated with steam and thermal aging. Table 14.11 shows the solderability test results of Pd against Ni/Au and Ni/Ag finishes.68 Apparently, Pd exhibits a superior solderability and stability. This is attributed to the low porosity of Pd versus that of Ni/Au or Ni/Ag. Applying a layer of Au flash on top of the Pd layer further improves the storage stability. Kakija et al. also reported that Pd and Pd-alloy electrodeposits preserve the integrity of the surface finish and provide good solderability by limiting porosity, inhibiting thermal diffusion, and increasing wetting speeds.69 However, it also has been reported that Pd may not always be a good diffusion barrier.Wang and Tu, at UCLA, noted that an intermetallic compound (IMC), which grows at a rate greater than 1 µm/s, has been observed in the liquid/solid reaction at 250°C between molten eutectic Sn-Pb solder and solid Pd. The intermetallic PdSn4 that is formed does not serve as a diffusion barrier between the reactants. Instead, it grows as lamellae into the molten solder, with the growth direction being normal to the liquid/solid interface. The molten solder between the lamellae serves as fast diffusion channels during the reaction. On the other hand, molten Sn reacts with Pd at a rate that is slower by one order of magnitude than Sn-Pb. The IMCs formed here grow as a diffusion barrier between the Sn and Pd.70 The bond strength of solder joints on a Pd finish is considerably lower than several other finishes. Ray et al. examined the pull strength of 50-mil-pitch, 20 I/O gullTABLE 14.11 Solderability of Pd Versus Ni/Au and Ni/Ag After Steam Aging and Thermal Aging* After Cu etch
After Cu etch, 85°C/85% SA,† 16 h
After Cu etch, 150°C TA,† 16 h
0.25 µm Pd/Cu laminate
99
96
96
0.025 µm Au/0.25 µm Pd/Cu laminate
99
99
98
1.25 µm Ag/1.25 µm Ni/Cu laminate
96
55
70
0.375 µm Au/1.25 µm Ni/Cu laminate
99
60
85
* All values are expressed as percentages. † SA: steam aging; TA: thermal aging.
14.40
CHAPTER FOURTEEN
FIGURE 14.40 Mechanical pull test data for 50-mil-pitch SOICs before and after 5000 thermal cycling (0 to 100°C, 3 cycles/h) using alloy 96.2Sn-2.5Ag-0.8Cu-0.5Sb (CASTIN). The component is 20 I/O gullwing-leaded SOIC.
wing-leaded small-outline integrated circuits (SOICs) before and after thermal cycling (0 to 100°C, 3 cycles/h) using a Pb-free alloy, 96.2 Sn-2.5Ag-0.8Cu-0.5Sb (CASTIN).71 Four surface finishes were compared: (1) ENIG (150 µin Ni/5 to 10 µin Au), (2) EN (150 µin)/electroless Pd (5 to 10 µin), (3) Pd (20 µin, or 0.5 µm) on copper, and (4) imidazole. Results indicate that the bond strength of the Pd surface finish is the lowest one among the four surface finishes, as shown in Fig. 14.40.71 This can be attributed to the formation of a large quantity of PdSn4 intermetallic. Similar to Au, the volume of intermetallics formed between Pd and Sn is significantly higher than that formed between Sn and other metals such as Cu or Ag. This is because the amount of Sn consumed for formation of intermetallics is much higher for Au (AuSn4) and Pd (PdSn4) than for Cu (Cu6Sn5 or Cu3Sn) and Ag (Ag3Sn). Here the Pd
TABLE 14.12 Pull Test Results of 1-mil Au Wire on Au-Flashed Electrolytic Pd Before and After Aging at 150°C for 64 h Ball bond minimum setting Force (g) Power (mW)
50 3.0
Ball bond maximum setting 50 9.9
Temp (°C)
120
120
Time (µs)
10
10
Pull force,
5.35
6.53
as plated (g) Std. deviation
1.69
0.21
Pull force,
6.38
6.29
1.24
0.53
64 h, 150°C (g) Std. deviation
LEAD-FREE SURFACE FINISHES
14.41
FIGURE 14.41 Mechanical pull test data for 256 I/O, 4-mm-pitch gullwing-leaded PQFPs before and after thermal cycling using alloy 96.2Sn-2.5Ag-0.8Cu-).5Sb (CASTIN). The thermal cycling condition is 0 to 100°C, 3 cycles/h.
thickness (20 µin) in the Pd finish is thicker than the Au thickness (5 to 10 µin) in ENIG or the Pd thickness (5 to 10 µin) in Ni/Pd, thus causing a greater deterioration in bond strength. Additional cycling treatment does not cause further deterioration in bond strength for the Pd finish, suggesting that the Pd finish may be a viable option as a surface finish. Similar tests conducted using 256 I/O 0.4-mm-pitch plastic quad flat pack (PQFP) also shows that the Pd finish exhibits the lowest pull strength. However, treatment with 2500 thermal cycles results in a slight decrease in bond strength for both Pd and Ni/Pd systems, as shown in Fig. 14.41.71 Presence of a large quantity of PdSn4 intermetallics can be seen easily. The wire bondability of Pd with Au flash was studied with the use of a 1-mil Au wire-and-ball-bonding process. Results indicate that the wire bondability is maintained after thermal aging at 150°C for 64 h, as shown in Table 14.12.68
14.42
CHAPTER FOURTEEN
14.7.2 ELECTROLESS (AUTOCATALYTIC) Pd WITH OR WITHOUT IMMERSION Au An electroless Pd coating process is essentially an autocatalytic process with or without an immersion Au flash ( Ni-Au, Ni-PdCo-Au.
LEAD-FREE SURFACE FINISHES
14.12
14.59
Sn-Bi
Sn-Bi alloy is attractive as a surface finish because it is lead-free, has good solderability, is low-cost, and has stability against environment. It is one of the four leadfree alloys—Bi-Sn, In-Sn, Ag-Sn, and Ag-Sb—recommended by the Occupational Safety and Health Administration as a replacement for Sn-Pb, based on their relative safety in the manufacturing environment. It is also one of three solder alloys— Sn-Ag, Sn-Bi, and Sn-Ag-Bi—recommended by the National Center for Manufacturing Sciences as a lead-free solder alloy option. The tin-bismuth alloy has already been used in PCB manufacturing as an etch resist.88 However, Bi-containing finish is sensitive to lead contamination due to the formation of a low-melting (95°C) ternary eutectic alloy 52Bi-30Pb-18Sn, thus compromising the reliability of solder joints.65 Another concern is the potential of having fillet lifting. Both phenomena will be discussed in detail in Chap. 16.
14.12.1
IMMERSION Sn-Bi ALLOY
Motorola has developed an immersion plating process to deposit approximately a 1.0-µm thickness of 70Sn-30Bi alloy onto copper surface mount pads as a PCB surface finish.11,89,90 Fabrication Process. The immersion tin-bismuth process is relatively simple to perform. A mild etch should be given to the copper surfaces prior to plating with tin-bismuth, and a rinse should follow the actual tin-bismuth plating process. The plating reaction can be performed by dipping in a stationary tank, or spraying/ flooding in a horizontal conveyorized system. The chemistry is based on salts of methane sulfonic acid. The immersion plating process deposits approximately 1.0 µm of a 70/30 Sn-Bi alloy in 1 min at 30°C. As other immersion plating processes, this is self-limiting, and the plating reaction stops once the maximum thickness is reached. Performance. The finish has a matte gray appearance, and thus it can be easily distinguished from the rest of the PCB by automated optical assembly and inspection equipment. Melton, at Motorola, conducted various tests on PCBs coated with the immersion tin-bismuth surface finish. Overall, this surface finish passed all of the tests with similar results as tin-lead. Capital costs are expected to be low, since the process is immersion plating. Material costs are low due to the thin thickness to be formed. The cost per square foot of deposited Sn-Bi alloy is estimated to be slightly more than that charged for OSP surface finishes.
14.12.2
ELECTROLYTIC Sn-Bi ALLOY
The tin-bismuth alloy has already been used in PCB manufacturing as an etch resist.91 With the addition of Bi, the chance of tin whiskering is highly reduced. One of the features of this electroplated tin alloy is the prevention of tin whiskers. Therefore, M&T marketed one tin electroplating chemical as a whisker-free tin formulation, which contains a small amount of bismuth that is codeposited with the tin.92
14.60
CHAPTER FOURTEEN
TABLE 14.19 Properties of 63Sn-37Pb and 99Sn-1Cu95 Property
63Sn-37Pb
Melting point (°C) Density (g/ml) Thermal conductivity (W/m⋅K) Cost per kg (bar) Cost per kg (paste)
227
8.4
7.31
56.61
65.73
Electrical conductivity (M mho/cm)
14.13
99Sn-1Cu
183
8.73
9.52
$0.85
$1.03
$140.8
$163.2
Sn-Cu (HASL)
99Sn-1Cu is attractive as a board surface finish due to promising solder alloy performance. Table 14.19 shows comparison between eutectic Sn-Pb and eutectic SnCu. Through their 1991 to 1994 study, Notel concluded that Sn-Ag and Sn-Cu are the most promising solder alloys among 200 alloy candidates. In the 1994–1995 lab study, Sn-Cu HASL was selected as the most promising lead-free board finish. In 1997, vertical Sn-Cu HASL was tried, with encouraging results. In 1998, further development of the Sn-Cu HASL board finish was conducted, including a horizontal HASL process.93 Fabrication Process. The process condition for manufacturing Sn-Cu HASL is shown in Table 14.20.94 Also shown is the process for Sn-Pb HASL as a comparison. Performance. The performance of a 99Sn-1Cu HASL surface finish was evaluated by comparing the pull strengths of solder joints made on various surface finishes, as shown in Table 14.21.93 The components used were plastic leaded chip carrier with Sn-Pb surface finish. Results indicate that horizontal Sn-Cu HASL is comparable with horizontal Sn-Pb HASL, Entek 106, and immersion Ag (Alpha-Level), and is better than Ni-Au and vertical Sn-Pb HASL, thus is a very viable option as a board surface finish. It should be noted that the HASL process is pertinent for conventional SMT assembly, but inadequate for fine-pitch SMT applications, due to uneven
TABLE 14.20 Comparison of Process Conditions Between 63Sn37Pb and 99Sn-1Cu in HASL Process95 Parameter
63Sn-37Pb
99Sn-1Cu
Bath temperature (°C)
250
280
Air knife temperature (°C)
250
280
Oil temperature (°C)
230
255
Air heat exchanger
250
300
Lower
Higher
150
200
Air pressure PCB preheat (°C)
14.61
LEAD-FREE SURFACE FINISHES
TABLE 14.21 Comparison of Pull Strengths of Joints on Various Surface Finishes Using 63Sn-37Pb and 99Sn-1Cu for Solder Joints95 Pull strength (N) PCB finish Ni/Au
63Sn-37Pb
99Sn-1Cu
19.3
18.8
Alpha-Level
25.6
20.2
Entek 106
23.6
22.7
Vertical Sn-Pb HASL
15.4
17.9
Horizontal Sn-Pb HASL
21.2
23.5
Horizontal Sn-Cu HASL
22.5
21.5
thickness in solder coating for large and small pads. This constraint holds true, regardless of the alloys selected for pad coating.
14.14
ELECTROLYTIC Sn-Ni
Sn-Ni alloy electroplate has been used as a final finish for PCB applications.At a tinnickel ratio of 65 wt % tin and 35 wt % nickel, the alloy forms a metastable phase, Ni-Sn. Thickness of the plating for the higher process temperature environments is usually 5 µm. For the more benign, lower process temperature environments, onehalf of that thickness may be sufficient.78 Fabrication Process. Tin-nickel is typically plated from a chloride-fluoride plating solution containing ammonium ion.94 It can also be plated with the nickel chloride/tin chloride solution complexed with potassium pyrophosphate, mainly used in Asia.95 Organic additives are used with all of the solutions for grain refinement and leveling. The cost of tin-nickel electroplating is about the same as solder mask over bare copper, with one of the major costs being the plating anodes, which may be tin and nickel or tin-nickel. Performance. The solderability of the deposit, although not as good as a pure tin or solder coating, is good enough for surface mount land finish applications. It has a flat surface, good storage life, and is compatible with a variety of solders. The Sn-Ni finished board can be etched in normal etching chemistries and may have a number of other finishes (e.g., Au, Sn-Pb, or Sn) overplated on the Sn-Ni land. Being nonmelting in nature, the Sn-Ni finish accepts a solder mask easily and can be used to produce solder coating on the lands only by the HASL process. Sn-Ni final finish is most often used when a higher process temperature is required. Sn-Ni electroplate is much harder than copper, Sn, and 63Sn-37Pb, with hardness being 750, 30, 100, and 12.8 HV, respectively. It maintains a low electrical resistance and is very corrosion-resistant, thus it can be used for edge connector applications, although the contact resistance is not as low as a gold overplate. The Sn-Ni deposit is free from problems such as tin whiskers or tin pest. When aged for a long time at 150°C, it will form copper-tin intermetallics with the underlying copper plating in the holes and surface.78
14.62
14.15
CHAPTER FOURTEEN
SOLID SOLDER DEPOSITION (SSD)
Solid solder depositions (SSDs) are a method of depositing solid solder onto lands of PCB.96,97 This solid solder not only serves as the surface finish for board lands, but also provides solder needed for forming solder joints, thus eliminating the need for the solder paste stencil printing process. Since the solder paste printing step often contributes approximately two-thirds of the defect rate of the PCB assembly process, SSD represents a potential of improving the yields. This is particularly true for fine-pitch applications with certain SSD processes. Although SSDs are not confined to lead-free solders only, they definitely provide options for lead-free surface finishes to PCB. The types of SSD may include HASL, Optipad, Sipad, PPT, solder cladding, solder jetting, and Super Solder, and are discussed briefly.
14.15.1
HASL
Hot-air solder level (HASL) techniques were developed in the early 1980s in a vertical mode. In the mid-1980s, horizontal machines were designed and gradually became the choice of larger PWB manufacturers. As of today, about two-thirds of the PWBs are with HASL finish, and the majority of HASL finishes are processed in the horizontal mode.1,98 Fabrication Process. The typical fabrication process for HASL is as follows: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
Acid clean Water rinse Etch Water rinse Flux application Preheat (105 to 150°C) Molten solder coat (2 to 10 s at 250 to 260°C) Excess solder blowoff Water rinse Dry
In a vertical HASL process, after the solder mask is applied, the panels are dipped vertically into a molten solder bath, then drawn vertically out of the bath and past high-pressure hot-air knives, which drive the molten solder through the platedthrough-holes, and provide some leveling of the solder along the exposed copper surface. In a horizontal HASL process, after the solder mask is applied, the orientation of the panel is changed to the horizontal plane. The panel is then pulled through the molten solder bath and past high-pressure hot-air knives. The horizontal orientation of the panels allows a conveyorized process, reducing the effects of gravitational pull on the molten solder and thus minimizing the tendency for puddling. The angle at which a PWB is presented to the air knives is important; the best results are achieved on quad flat packs (QFPs) at 45° to the air knife. The PWB is usually in contact with solder for about 2 s, so the copper tin intermetallic
14.63
LEAD-FREE SURFACE FINISHES
formed in that time is typically 0.15 to 0.30 µm, although the thickness of IMC can be as high as 1.9 µm through the HASL process.98,99 Horizontal HASL is capable of providing a solder finish that meets assembly requirements. For 0.010- to 0.020in pitch, a mean solder coating thickness of 12.5 µm, with an LCL of 1.75 µm and a UCL of 25 µm, is achieved with no solderability problems on a large pad. Typically, the smaller the pad, the thicker the solder coating is.99,100 Table 14.22 shows the range of finished solder coating thicknesses of all pads for horizontal and vertical HASL processes. However, a coating thickness of up to 75 µm at the center of a via hole has been noted.101 Improper cleaning was a leading cause of exposed copper and dewetting in HASL process, and high viscosity with minimal thickness is desired.102 Performance. The advantages of the HASL process include: ● ● ● ● ● ● ● ● ● ●
Excellent solderability Long shelf life (12 months) Universal acceptance Multiple heat cycle capability Easy visual inspection Good mask integrity Fair electrical contact Fair microwave applications Compatible with solder mask on bare copper applications No solder reflow under solder mask The disadvantages of the HASL process include:
● ● ● ● ● ● ●
Difficult process Boards being thermally stressed Poor surface contrast between solder and pad Not compatible with wire-bonding process Inconsistent solder volume deposition from pad to pad Poor surface coplanarity Hole compensation needed (50 to 75 µm) due to the nonuniform HASL solder deposit
TABLE 14.22 Finished Solder Coating Thickness for Vertical and Horizontal Sn-Pb HASL Processes Process
Hole wall
Board surface
Note
Vertical HASL
12.5–25 µm
0.125–25 µm
Not recommended for 20 mil
Horizontal HASL
12.5–25 µm
0.25–15 µm
Recommended for fine pitch
pitch or less
14.64
14.15.2
CHAPTER FOURTEEN
Optipad
Optipad is a process that delivers flattened solder deposits on pads serving as both the surface finish of pads and solder source for joint formation. In this process, a temporary dry film (Optimask) is applied to the board, printed, and developed, forming photodefined wells.A liquid solder is forced into the wells.The board is kept flat as it moves through the machine and cools. The temporary dry film is then stripped from the board. The solder thickness is typically 50 to 200 µm, which is controlled by the temporary solder mask.96,97 Fabrication Process. Detailed fabrication process steps for Optipad follow and are schematically shown in Fig. 14.56.96 1. 2. 3. 4. 5. 6. 7.
Apply the temporary solder mask. Apply the molten solder. Flatten the solder. Remove the temporary solder mask. Apply the sticky flux. Place the components. Reflow the package.
FIGURE 14.56 Schematic of Optipad process. (a) Bare board, (b) application of temporary mask, (c) liquid solder application, (d) stripping of temporary mask.
LEAD-FREE SURFACE FINISHES
14.65
Performance. The primary benefit in the Optipad process is the consistent solder volume attained and the flat deposit. The negative aspects include ●
● ●
The cost associated with the application and removal of the temporary solder mask The initial cost of the equipment The inability of maintaining the planarity of the second side
14.15.3
Sipad
Similar to Optipad, Sipad also forms flattened solder deposit on pads serving as both surface finish and solder source for joint formation. Instead of using a temporary solder mask and molten solder for deposition, Sipad employs a permanent solder mask for solder volume control and solder paste as a solder source for deposition. The average thickness of solder deposits formed is 50 µm and can be as high as 130 µm. Fabrication Process. The detailed fabrication process for Sipad can be described as follows96,97,103: 1. Analyze and alter the computer-aided design (CAD) as necessary. 2. Apply a photoimagable dry-film solder mask (Simask) with a thickness of no less than 100 µm. 3. Laminate and develop the Simask to form a well around each pad. 4. Print solder paste into a well via the standard solder paste stencil printing process. 5. Reflow solder paste via the standard process for form bumps with meniscus above the plane of the solder mask. 6. Wash the board to remove flux residue and solder balls. 7. Place the PCB into a flattening system. 8. Reheat solder to the melting point. 9. Flatten the pads between the platens of a cold press to freeze the solder deposit into a planar SSD flush with the surface of the solder mask. 10. Print adhesive no-clean flux onto the planar SSD surface. 11. Dry the flux to a tacky finish. 12. Cover the tacky finish with a release paper. 13. Remove the release paper, and place the component. 14. Reflow to form solder joints. Performance. The benefits of Sipad include ● ● ● ●
The very flat deposit The ability to alter the volume of solder Excellent solderability Good compatibility with all other surface finishes (e.g., OSP or Ni/Au) for solder paste deposition process
14.66
CHAPTER FOURTEEN
The solder volume is accomplished by regulating the opening of the solder mask around the surface mount pad and filling the entire well with solder paste. The concerns associated with the Sipad process include ● ● ● ●
The potential for foil formation The potential for meniscus formation The cost associated with the use of a thick dry film mask The inability of maintaining the planarity of the second side
14.15.4
PPT
Precision pad technology (PPT) forms a flattened solder deposit with the use of a regular solder mask and solder paste. A vibrating stainless-steel mesh is seated on top of the PCB printed with paste. The paste is then reflowed with a traveling hot-air knife, and solidified afterward with a pass of a cool-air knife. The thickness of solder is 50 to 200 µm. Fabrication Process. The fabrication process of PPT can be described as follows: 1. Prepare the PCB with a solder mask no less than 25 µm in thickness. 2. Stencil-print the solder paste onto the pads within the aperture of the solder mask. 3. Load the board into the reflow/formation system. 4. Place a vibrating tensioned stainless-steel screen onto the surface of the board. 5. Reflow the paste with a hot-air knife traveling over the board. 6. Solidify the solder with a pass of a cool-air knife. 7. Board-exit the system. 8. Clean the board, if necessary. TABLE 14.23 Typical Reflow Profile for PPT Process Using 63Sn-37Pb105 Peak temperature
214.5°C
Time over 150°C
57.5 s
Time over 180°C
15.0 s
Time over 200°C
6.0 s
Time over 210°C
2.5 s
The dwell time above liquidus temperature is normally less than 20 s. Table 14.23 shows a typical PPT reflow profile for a 63Sn-37Pb system.104 The screen serves as a mold to flatten, shape, and remove excess solder during reflow. Excess solder wicks above the mesh during reflow in the form of solder balls. The reflowed solder deposits are macroplanar with an embossed surface topography that facilitates retaining tack flux at subsequent assembly. Shorts and solder balls are eliminated at assembly, and the copper lands are encapsulated in a thick solder deposit, which increases bare-board shelf life.
Performance. The advantages of PPT are simplicity and the low cost associated with it. Furthermore, the mesh impression left on the surface of the board provides a flat surface for component leads to rest on and an area for tacky flux to pool prior to reflow at assembly. It offers the planarity and uniformity that are needed for finepitch and BGA assembly by reassigning solder paste printing responsibilities to the supplier. The limitation is the inability to maintain the planarity of the second side.
LEAD-FREE SURFACE FINISHES
14.15.5
14.67
SOLDER CLADDING
Solder cladding forms solid solder deposits by reflowing solder paste, a method in use since the 1960s.97 Fabrication Process. The fabrication process of solder cladding is described as follows: 1. 2. 3. 4. 5.
Print the solder paste. Reflow the solder paste. Apply the tacky flux. Place the components. Reflow the package.
Performance. Simplicity and cost are the two advantages to solder cladding. The main drawback is the formation of the rounded meniscus, thus causing potential difficulty in placing fine-pitch components.
14.15.6
SOLDER JETTING
The solder jetting process deposits liquid solder droplets of a controlled size to the land areas of the circuit board. Pressure and vibration via a piezoelectric mechanism are applied to a liquid metal reservoir, which forces the solder through an orifice to form a liquid droplet that flies through a charge electrode, electrostatic deflection plate, and a catcher, and then onto the surface mount pads. The size of the balls is in the range of 0.004 to 0.012 in.97 Fabrication Process. The fabrication process of solder jetting is described as follows: 1. Apply the jetted solder droplet onto the fine-pitch pads. 2. Apply the tacky flux to the fine-pitch pads. 3. Place components onto the fine-pitch sites. 4. Reflow via a hot bar. 5. Print the paste onto coarse-pitch sites. 6. Place the coarse-pitch components. 7. Reflow the coarse-pitch components. Performance. Ideally, this method allows the deposition of solder onto fine-pitch pads automatically through programming. Unfortunately, this is at the expense of being able to process the entire board in one step. Furthermore, uniformity, coplanarity, jetting landing precision, and throughput are issues that still need to be resolved before this process becomes widely acceptable.
14.15.7
SUPER SOLDER
Super Solder is a chemical deposition process. Applied in paste form, material mainly contains (RCOO)2Pb (a lead salt of an organic acid) and tin powder. Upon heating, the two components react, with lead being reduced to metal and tin being
14.68
CHAPTER FOURTEEN
oxidized to tin salt. The lead metal forms alloys with tin powder to create particles; the particles then settle, forming a solder deposit. A minimum pitch of 0.1 to 0.15 mm can be sustained.106 The thickness of the solder deposit is between 40 and 70 µm. Fabrication Process. The fabrication process of Super Solder is described as follows: 1. Preclean the panels via the microetch process. 2. Apply Super Solder, a proprietary organic lead and tin powder paste. 3. Heat to bond Super Solder to fine-pitch sites to form solder bumps. 4. Clean the panels. 5. Apply tacky flux to the fine-pitch sites. 6. Place the fine-pitch components. 7. Reflow the fine-pitch sites via a hot bar. 8. Paste-print the coarse-pitch sites. 9. Place the components. 10. Reflow the coarse-pitch sites. Performance. This process is well suited for applications such as tape automated bonding, where the insertion of solder mask dams may be difficult but is not designed to predeposit solder over all of the component sites on a board.97 Super Solder is used in Toshiba’s notebook and Panasonic’s notebook.105 The solder deposit is dome-shaped. The cost of the complete system may make Super Solder prohibitive for some applications. Most important, the exchange chemistry deposition approach, adequate for a Sn-Pb alloy, is highly questionable to be applicable for Pb-free solder systems.
14.16
SUMMARY OF PCB SURFACE FINISHES
As discussed previously, there is a wide range of lead-free coating options available for PCB surface finish applications. However, due to the complicated functional requirement of various electronic products, it is virtually unlikely to identify a single surface finish that will satisfy all of the requirements. Following are the pros and cons of finish options for specific applications. ●
●
Solderability. Similar to Sn-Pb HASL, Pb-free HASL is also considered to be superior in solderability, particularly for a reflow temperature higher than the melting temperature of HASL materials. However, the primary limitation of HASL is the inability to provide even finish thickness and quality for fine-pitch applications. This is true whether the solder is Sn-Pb or Pb-free; this excludes HASL as a major candidate for PCB finish applications. Plated metallization surface finishes often exhibit better solderability, if the metal can be dissolved into solder rapidly during soldering.Therefore, options such as Au,Ag, and Sn are often better in solderability than others, including OSPs. Aging resistance. Metal surface finishes often exhibit better resistance than OSP against aging (e.g., thermal or steam). This is particularly true for noble metal fin-
LEAD-FREE SURFACE FINISHES
●
●
●
●
●
●
●
●
●
●
14.69
ishes. However, under corrosive gas storage conditions, OSPs turn out to be outstanding in retaining their solderability. Tolerance against Pb contamination. Bi-containing systems are sensitive to Pbcontamination, mainly due to the formation of low-melting ternary eutectic 52Bi30Pb-18Sn alloy. Thick-board through-holes. Due to the throw-power limitation, electrolytic platings typically fail to provide an even coating for thick-board through-holes. As a result, the immersion process becomes a favorite choice. Wire bonding. Only nonmelting metallic surface finishes can be considered for wire-bonding applications. This quickly rules out OSP, Sn-, or Bi-containing systems. Presence of nickel underlayer improves the wire bondability, although the solderability may be compromised, as demonstrated by Pd systems. In-circuit probe testing. For in-circuit probe testing purposes, finishes with either tarnish or too high a hardness will have greater difficulty in probe penetration. In general, OSPs are not the most promising in this regard. Metallic surface finishes with a matte appearance often are more promising, due to a greater contact area provided by the rough surface topology. Tolerance against cleaning. Organic solderability preservatives have a very low tolerance against PCB cleaning prior to paste reflow, because OSPs tend to be removed by the cleaners as well. Metallic surface finishes are all fairly robust against board cleaning. IMC formation. Both Au and Pd suffer excessive intermetallics formation issues when the thickness of those metals is greater than approximately 0.25 µm. This is particularly an issue for Au when the same finish is to be used for wire bonding as well. Connector applications. Usually surface finishes with high electrical conductivity, high abrasion resistance, high oxidation resistance, and high hardness are desired for connector applications. The OSPs are ruled out immediately due to poor electrical conductivity. Noble metallic finishes often shine in this category. Finish consistency/inspectability. Finishes with challenges in both quality consistency and inspection capability are the trickiest manufacturing issues. For instance, ENIG tends to have black-pad problems from time to time. And, unfortunately, the symptom is not easily inspectable prior to assembly, and it is not inspectable without destructive testing after assembly. Tin whisker/tin pest. All tin-based metals or alloys suffer from potential problems with tin whisker and tin pest. Although both symptoms can be prevented to a certain extent, such as nickel underlying or alloying tin with secondary elements, 100 percent symptom-proof is still unassured. These potential problems highly reduced the options of finding a low-cost lead-free surface finish, as will be discussed later. Cost. The OSP is probably the lowest in cost. Next to that may be nonpreciousmetal plating finishes, especially the immersion processes. Additional savings promised by the electroless process include elimination of tie bars from circuit board design, hence the additional size reduction that is achievable.
Overall, it can be seen that none of the surface finishes can be considered an ideal solution as the answer for all PCB finish requirements.Therefore, the finish selection has to be determined by the requirements of specific applications involved in the product design.
14.70
CHAPTER FOURTEEN
14.17 OPTIONS FOR COMPONENT SURFACE FINISHES The requirements for component surface finishes are similar to those for PCB surface finishes. However, some of the features may be more important to components than to PCBs, such as ductility, due to the lead trimming and forming requirements. Shipley uses the following criteria as their plating process requirements80: ● ● ● ● ● ●
Must be compatible with existing high-speed plating equipment Must have a minimum deposition rate of 7.5 µm/min Must have familiar chemistry, preferably an MSA electrolyte Must have all products in the process that are fully analyzable Must have deposits that possess good solderability (same as or better than Sn-Pb) Must have deposits that possess good ductility (same as or better than Sn-Pb)
Table 14.24 lists the options of lead-free surface finishes for components. The system is categorized by the key element that is used. Each category is further classified by the type of process and chemistry. Examples are given for certain groups.
14.18
Ni/Au (ENIG)
Fabrication Process. The characteristics and manufacturing process of ENIG are the same as those discussed in Sec. 14.4.2. Performance. Popelar et al., at IC Interconnect, have investigated using ENIG for flip chip under-bump metallurgy (UBM) applications.106 In their process, the exposed aluminum I/O pads are plated with an EN cap with a typical thickness of 5
TABLE 14.24 List of Lead-Free Surface Finishes Surface finish system
Finish process and chemistry
Ni/Au
Electroless Ni/electroless (immersion) Au, or ENIG
Pd
Electrolytic Pd or Pd alloys
Ni/Pd
Example
Electroless Ni/electroless (autocatalytic) Pd Electroless Ni/electroless (autocatalytic) Pd/electroless (immersion) Au
Pd-Ni
Electrolytic Pd-Ni
Sn
Electrolytic Sn
Sn-Ag
Electrolytic Sn-Ag
96.5Sn-3.5Ag
Sn-Bi
Electrolytic Sn-Bi
90Sn-10Bi
Sn-Cu
Electrolytic Sn-Cu
99Sn-1Cu
For multilayer design, the sequence of materials starts from the bottom layer.
LEAD-FREE SURFACE FINISHES
14.71
µm, followed by a flash of immersion Au. The appropriate solder paste alloy is deposited via stencil printing, and subsequently reflowed and cleaned. No degradation in shear load or failure mode occurred among the three alloys (63Sn-37Pb, 90Pb-10Sn, and 95.5Sn-3.8Ag-0.7Cu) that were tested, indicating no critical UBM consumption (i.e., no excessive intermetallics growth) during reflow. Additional tests were performed comparing nickel UBM thicknesses of 1, 2, and 5 µm.Again, all bumps exhibit comparable shear strength, indicating no critical UBM consumption. These data suggest that ENIG can be a viable candidate as UBM for flip chip solder paste bumping applications.
14.19
ELECTROLYTIC Pd
Fabrication Process. See Sec. 14.7.1. Performance. Fan et al., from Lucent Technologies, studied electrolytic Pd or electrolytic Pd with Au flash as a potential leadframe surface finish.107 The postetch solderability comparison was made on 34 µm of copper laminate of 0.25 µm of palladium, 0.025 µm of gold flash over 0.25 µm of palladium, 1.25 µm of silver over 1.25 µm of nickel, and 0.38 µm soft gold over 1.25 µm of nickel. Both gold-flashed palladium and palladium alone passed the steam and thermal aging with only a slight degradation. The gold-flashed palladium was better than all of the samples tested, probably due to the gold on the palladium performing two functions: (1) inhibiting the diffusion of copper into the palladium, and (2) acting as an oxidation barrier. The failure of both silver and gold over nickel after thermal and steam aging were probably due to migration of nickel through the silver or gold layer and forming an oxide on the surface. Palladium, as an excellent migration/diffusion barrier, unlike gold and silver, produces a more stable final finish, which does not require an active flux to be solderable.The results conclude that thin deposits (30 percent) Low toxicity Abundant world reserve Low cost
14.76
CHAPTER FOURTEEN
TABLE 14.27 Plating Parameters for Sn-Cu Alloy Electrolyte Parameter Sn content (g/L)
Setting
Fabrication Process. The reportedly highspeed, methane sulfonic acid–based electrolyte chemistry developed by Shipley can be described in Table 14.27.80
60
Performance. The topology (2000×) of SnCu, plated at 20 Å/dm2 current density, is shown Acid content (mL/L) 200 in Fig. 14.60.80 The Sn-Cu deposit composition Additive (mL/L) 105 is fairly insensitive to variation in temperature and current density, as shown in Fig. 14.61.80 Temperature (°C) 40 To evaluate the potential of tin whisker Current density (Å/dm2) 25 growth, two Sn-Cu samples deposited with 25 and 30 Å/dm2 current density, respectively, were subjected to 60°C/95% RH for 500 h. As a control, pure tin after 500 h of aging showed tin whisker formation, while Sn-Cu finishes exhibit no whiskers at all for both samples. The solderability of Sn-Cu was studied by determining the zero-force cross time (ZCT) in the wetting balance test.Also evaluated were 90Sn-10Pb, 100Sn, 90Sn-10Bi, and 97Sn-3Ag. Results shown in Table 14.28 indicate that the solderability decreases in the following order: 90Sn-10Pb > 90Sn-10Bi > 97Sn-3Ag > 100Sn and 99Sn-1Cu.80 All finishes are considered acceptable in solderability. Cu content (g/L)
1.1
14.26 SUMMARY OF COMPONENT SURFACE FINISHES Schetty has compared the pros and cons of several viable component surface finishes, with results summarized in Table 14.29.80 Ni/Pd is too expensive and marginal TABLE 14.28 Solderability Test Results for Various Component Surface Finishes Component finish
ZCT (s)
Coverage (%)
Sn-Pb 90-10
Aging condition As plated
0.29
>95
Sn-Pb 90-10
Steam aged
0.46
>95
Sn-Pb 90-10
Heat aged
0.32
>95
Sn
As plated
0.58
>95
Sn
Steam aged
0.83
>95
Sn
Heat aged
0.68
>95
Sn-Bi 90-10
As plated
0.33
>95
Sn-Bi 90-10
Steam aged
0.41
>95
Sn-Bi 90-10
Heat aged
0.37
>95
Sn-Ag 97-3
As plated
0.42
>95
Sn-Ag 97-3
Steam aged
0.54
>95
Sn-Ag 97-3
Heat aged
0.53
>95
Sn-Cu 99-1
As plated
0.39
>95
Sn-Cu 99-1
Steam aged
0.83
>95
Sn-Cu 99-1
Heat aged
0.79
>95
LEAD-FREE SURFACE FINISHES
14.77
FIGURE 14.60 Topology (2000×) of a Sn-Cu surface finish plated at 20 Å/dm2 current density.
FIGURE 14.61 Effect of current density and temperature on Sn-Cu deposit composition.
14.78
CHAPTER FOURTEEN
TABLE 14.29 Pb-Free Electronic Finish Comparison Summary Pb-free Solderability finish
Mechanical properties
Melting point
Whiskering
Compatibility
Plating feasibility
Relative cost
NiPd
∗
∗
G
G
G
G
X
Sn
G
G
∗
X
G
G
G G
SnBi
G
G
∗
G
X
G
SnAg
G
G
G
G
G
X
∗
SnCu
G
G
G
G
G
G
G
Note: G: good; ∗: marginal; X: unacceptable.
on soldering and mechanical properties. Sn still suffers concerns on tin whiskering and tin pest problems. Sn-Bi is sensitive to Pb contamination. Sn-Ag is very difficult to plate high Sn deposit because of unfavorable electrochemical potentials of Sn versus Ag. Overall, the most promising component finish appears to be Sn-Cu, due to the lack of any obvious weakness.80
REFERENCES 1. Cullen, D., “Immersion Silver Performance Results,” Proceedings: IPCWorks ’99 Conference, Minneapolis, MN, October 27, 1999. 2. Ray, U., I. Artaki, H. M. Gordon, and P. T. Vianco, “The Influence of Temperature and Humidity on Printed Wiring Board Surface Finishes: Immersion Tin vs. Organic Azoles,” J. Electronic Materials, 23(8):779–785, August 1994. 3. Parker, J. L., and J. S. Horton, “Assembly of Printed Wiring Boards Coated with an Organic Solderability Preservative,” Surface Mount International Conference, San Jose, CA, 1992. 4. Artaki, I., et al., “Corrosion Protection of Copper Using Organic Solderability Preservatives,” Surface Mount International Conference, San Jose, CA, 1992. 5. Ray, U., I. Artaki, H. M. Gordon, and R. L. Opila, “Solderability and Thermal Stability of Azole Corrosion Inhibitors,” Proc. of NEPCON West, pp. 423–435, Anaheim, CA, 1993. 6. Wenger, G. M., and D. A. Machusak, “Soldering Evaluation of Organic Solderability Preservatives,” Proc. of NEPCON West, pp. 436–451, Anaheim, CA, 1993. 7. Wengenroth, K., “OSPs: Guidelines for Successful Soldering,” Proc. of NEPCON West, Anaheim, CA, February 23–27, 1997. 8. Castaneda, C., K. Chandler, N. Nguyen, R. G. Robertson, and J. White, “Evaluating Alternative PCB Fabrication Processes,” Circuits Assembly, pp. 58–61, January 1994. 9. Carano, M., “OSP evolution [PCB surface finishes],” Printed Circuit Fabrication, 20(7):28–31, July 1997. 10. Verbockhaven, D., G. Conard, and K. McKean, “Using Nitrogen on OSP-Coated Printed Circuit Boards,” Electronic Packaging and Production, 37(5):59–60, 62, April 1997. 11. Melton, C., and H. Fuerhaupter, “Lead-Free Tin Surface Finish For PCB Assembly,” Proc. of NEPCON West, Anaheim, CA, February 25–29, 1996. 12. Charbonneau, R. A., “An Evaluation of Anti-oxidant Coatings as an Alternative to HASL,” Proc. of NEPCON West, Anaheim, CA, February 25–29, 1996. 13. Langston, K. F., and S. Gutierrez, “Evaluating OSPs Used in the Card Assembly Process,” Proc. of NEPCON West, Anaheim, CA, February 25–29, 1996.
LEAD-FREE SURFACE FINISHES
14.79
14. Parquet, D. T., and D. W. Boggs, “Alternatives to HASL: A User’s Guide for Surface Finishes,” Electronic Packaging and Production, pp. 38–42, August 1995. 15. Ivankovits, J. C., and B. M. Adams, “Effects of Controlled Atmospheres on Solder Wetting,” Surface Mount Technology, pp. 23–26, October 1993. 16. Ewell, G. J., “New Approaches to Preserving Solderability on PCBs,” Proc. of NEPCON West, Anaheim, CA, February 23–27, 1997. 17. Wetz, L. A., K. Kirschenbaum, and S. Kalisz, “OSP Characterization in Flip Chip Ball Grid Array Packaging,” APEX, San Diego, CA, January 14–18, 2001. 18. Glicoat presentation from Shikoku Chemicals Corporation, June 16, 1999. 19. Dugdale, I., and J. B. Cotton, Corrosion Science, 3:69–74, 1963. 20. DeBiase, J. D., “Organic Solderability Preservatives: Benzotriazoles and Substituted Benzimidazoles,” SMI 96, San Jose, CA, September 10–12, 1996. 21. “COBRATEC Corrosion Inhibitors: Copper and Brass Protection,” PMC Specialties Group, Inc., Data Sheet, Corr 3034. 22. Tornkvist, C., D. Thierry, J. Bergman, B. Liedberg, and C. Leygraf, “Methyl Substitution in Benzotriazole and Its Influence on Surface Structure and Corrosion Inhibition,” J. Electrochem. Soc., 136(1):58–64, January 1989. 23. DeBiase, J., “No More HASL,” Circuits Assembly, pp. 48–53, October 1993. 24. Ho, M. K., “Copper Surface Finish Promotes Solderability,” EP&P, pp. 39–41, October 1987. 25. Walker, R., Metal Finishing, 71(9):63–66, 1973. 26. Thwaites, C. J., Transactions of the Institute of Metal Finishing, 43:143–152, 1965. 27. Parker, J. L., Jr., “The Performance and Characteristics of the Imidazole Coating,” SMI 96, San Jose, CA, September 10–12, 1996. 28. Murray, J., “Beyond Anti-Tarnish: An SMT Revolution,” Printed Circuit Fabrication, 16(2):32–34, February 1993. 29. Sirtori, V., L. Lombardi, and G. Redaelli, “Chemical Composition and Thermal Stability of 2-Butyl, 5-Chloro, Benzimidazole Film,” Journal of Electronic Materials, 26(5):459–462, May 1997. 30. Shea, C., T. J. Chinnici, and K. Stillings, “Effects of Reduced Purity Nitrogen in the Inert Wave Soldering Environment,” Proc. of NEPCON West, Anaheim, CA, February 1999. 31. Wengenroth, K., and E. Stafstrom, “OSPs: The Next Generation,” Proc. of NEPCON West, Anaheim, CA, February 2000. 32. Wengenroth, K., E. Stafstrom, and J. Fudala, “OSPs: Addressing Future Surface Finishing Needs,” SMTA, Chicago, IL, 1999. 33. Boggs, D. W., “Anti-Tarnish: One Alternative to HASL,” EP&P, pp. 34–38, August 1993. 34. Shiau, L. C., C. E. Ho, and C. R. Kao, “Reactions Between SnAgCu Lead-Free Solders and Ni-Bearing Surface Finish in Advanced Electronic Packages,” in Proc. of APACK, pp. 99–104, Singapore, December 5–7, 2001. 35. Dalich, G. M., “A Comparison of Immersion and Electrolytic Gold as a Final Finish for PCBs,” SMTA International, Chicago, 2000. 36. Bader, W. G., Weld. Res. Suppl., 48:551s–557s, 1969. 37. Heinzel, H., and K. E. Saeger, Gold Bull., 9(1):7–11, 1976. 38. Humpston, G., and D. M. Jacobson, “Principles of Soldering and Brazing,” ASM International, Materials Park, OH, 1993. 39. Satoh, R., in Thermal Stress and Strain in Microelectronics Packaging, Lau, J. H., ed., Van Nostrand Reinhold, New York, pp. 500–531, 1993. 40. Glazer, J., “Metallurgy of Low Temperature Pb-Free Solders for Electronic Assembly,” International Materials Reviews, 40(2):65–93, 1995.
14.80
CHAPTER FOURTEEN
41. Nakajima, K., A. Lewis, and N. Brathwaite, “Implementation and Qualification of Chip Scale Package On-Board Assembly Process,” Proc. of NEPCON West, Anaheim, CA, February 21–25, 1999. 42. Lai, Z., and J. Liu, “Effect of the Microstructure of Ni/Au Metallization on Bondability of FR-4 Substrate,” Proc. of the Third International Symposium of Electronic Packaging Technology, pp. 168–172, Beijing, China, August 17–21, 1998. 43. Stafstrom, E., “Good as Gold,” Printed Circuit Fabrication, 20(4):26–30, 1997. 44. Glazer, J., P. Kramer, and J. W. Morris, “Effect of Au on the Reliability of Fine Pitch Surface Mount Solder Joints,” Circuit World, 18(4):41–46, 1992. 45. Darveaux, R., K. Banerji, A. Mawer, and G. Dody, “Reliability of Plastic Ball Grid Array Assembly,” Ball Grid Array Technology, Lau, J., ed., McGraw-Hill, New York, chap. 13, pp. 379–442, 1995. 46. Cinque, R. B., and J. W. Morris, Jr., “The Effect of Gold-Nickel Metallization Microstructure on Fluxless Soldering,” Journal of Electronic Materials, 23(6):533–539, June 1994. 47. Ludwig, R. T., N.-C. Lee, C. Fan, and Y. Zhang, “Evaluation of Two Novel Lead-Free Surface Finishes,” SMTA International, Chicago, October 2001. 48. Houghton, B., “Electroless Nickel/Immersion Gold Joint Cracking,” Future Circuits International, 5, 121–126, 2001. 49. Johal, K., “Are You in Control of Your Electroless Nickel/Immersion Gold Process?,” SMTA International, Chicago, October 2001. 50. Cullen, D. P., “Overview and Resolution of Electroless Nickel Immersion Gold Failure Modes,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 51. Mei, Z., S. K. Liem, and A. Shih, “A Failure Analysis and Rework Method of Electronic Assembly on Electroless Ni/Immersion Au Surface Finish,” SMTA International, Chicago, 1999. 52. Coderre, J., “Electroless Nickel/Gold and Process Control,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 53. Milad, G., “Surface Finishes: Metallic Coatings Over Nickel Over Copper,” SMI 96, San Jose, CA, September 10–12, 1996. 54. Parker, J. L., Jr., “The Performance and Attributes of the Immersion Silver Solderability Finish,” Proc. of NEPCON West, Anaheim, CA, February 1999. 55. Biunno, N., “A Root Cause Failure Mechanism for Solder Joint Integrity of Electroless Nickel/Immersion Gold Surface Finishes,” SMTA, International, Chicago, 1999. 56. Dalich, G. M., “A Comparison of Immersion and Electrolytic Gold as a Final Finish for PCBs,” SMTA International, Chicago, 2000. 57. Lotosky, P., “Lead-Free Update,” Tutorial at IMAPS-Brazil, São Paulo, Brazil, August 1–3, 2001. 58. Parker, L.,“Introduction of the Immersion Silver Solderability Coating into Production,” Proc. of NEPCON West, Anaheim, CA, February 2000. 59. Beigle, S., “Non-Precious Metal Coatings for Fine Pitch Assembly and Direct Chip Attachment,” SMI, San Jose, CA, September 10–12, 1996. 60. Chada, S., and E. Bradley, “Investigation of Immersion Silver PCB Finishes for Portable Product Applications,” SMTA International, Chicago, October 1–4, 2001. 61. Gordon, R., S. Marr, and D. Shangguan, “Evaluation of Immersion Silver Finish for Automotive Electronics,” SMTA International, Chicago, 2000. 62. Reed, J., “Immersion Silver as a Replacement for Solder Finish,” in Proc. of IPC/SMTA Electronics Assembly Expo 1998, p. S23-3-1, Providence, RI, October 1998. 63. Chase, L., P. Viswanadham, and S. Dunford, “Comparison of Ag, Ni/Au and Solder PWB Surface Finishes on the Second Level Reliability of Fine Pitch Area Array Assemblies,” SMTA International, Chicago, 2000.
LEAD-FREE SURFACE FINISHES
14.81
64. Guy, J., “Solder Joint Reliability Impact of Using Emersion Metallic Coatings,” Proc. of NEPCON West, Anaheim, CA, February 23–27, 1997. 65. Mei, Z., F. Hua, and J. Glazer,“Sn-Bi-X Solders,” SMTA International, San Jose, CA, September 13–17, 1999. 66. Humpston, G., and D. M. Jocobson, “Principles of Soldering and Brazing,” ASM International, Materials Park, OH, 1993. 67. Milad, G., and H. Roberts, “Electroless Palladium Coatings for Bonding Fine Pitch Devices,” Proc. of NEPCON West, Anaheim, CA, February 25–29, 1996. 68. Stacy, B. F., J. A. Abys, C. Fan, L. J. Mayer, I. Kadija, E. J. Kudrak, and H. K. Straschil, “Palladium for a PWB Surface Finish,” Proc. of NEPCON West, Anaheim, CA, February 23–27, 1997. 69. Kakija, I. V., J. A. Abys, J. J. Maisano, E. J. Kudrak, and S. Shjmada, “Thin Multilayer Coatings for Semiconductor Packaging Applications. Part I: Solderability,” Plating and Surface Finishing, pp. 56–62, February 1995. 70. Wang, Y., and K. N. Tu, “Ultrafast Intermetallic Compound Formation Between Eutectic SnPb and Pd Where the Intermetallic Is Not a Diffusion Barrier,” Applied Physics Letters, 67(8):1069–1071, August 21, 1995. 71. Ray, U., I. Artaki, D. W. Finley, G. M. Wenger, T. Pan, H. D. Blair, J. M. Nicholson, P. T. Vianco, “Assessment of Circuit Board Surface Finishes for Electronic Assembly with Lead-Free Solders,” SMI 96, San Jose, CA, September 10–12, 1996. 72. Seto, P., J. Evans, and S. Bishop, “Palladium Surface Finish for Soft Touch Switches and High Density SMT Assemblies,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 73. Toben, M., and M. Kanzler, “Palladium’s Niche [Surface Mount Applications],” Printed Circuit Fabrication, 19(7):24–27, July 1996. 74. Ludwig, R. T., N.-C. Lee, C. Fan, and Y. Zhang, “Evaluation of Two Novel Lead-Free Surface Finishes,” SMTA International, Chicago, September 30–October 4, 2001. 75. Yeung, D., and M. Nakamura, “Electroless Gold Plating Contact Pads for Memory Modules/Cards,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 76. Edgar, R., “Various Applications and Production Uses of Immersion White Tin in the Fabrication of Printed Circuit Wiring Boards,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 77. Zhang, Y., “Electroplated Pure Tin—A Lead Free Alternative,” Proceedings: IPCWorks ’99 Conference, Minneapolis, MN, October 27, 1999. 78. Hinton, P. E., “Tin-Plating, Tin-Nickel Electroplate and Tin-Plating Over Nickel as Final Finishes on Copper,” SMI 96, San Jose, CA, September 10–12, 1996. 79. Hirsch, S., and C. Rosenstein, “Tin and Tin-Lead Plating,” Metal Finishing Guidebook and Directory, 94(1A): 1996. 80. Schetty, R., “Lead Free Legislative Situation,” Proceedings: IPCWorks ’99 Conference, Minneapolis, MN, October 27, 1999. 81. Zhang, Y., G. Breck, F. Humiec, K. Murski, and J. A. Abys, “An Alternative Surface Finish for Tin/Lead Solders—Pure Tin,” SMI 96, San Jose, CA, September 10–12, 1996. 82. Warwick, and S. J. Muckett, Brazing and Soldering Journal, British Association for Brazing and Soldering, Autumn 1982. 83. Hunt, C., “A Model for Solderability Degradation,” SMI 96, San Jose, CA, September 10–12, 1996. 84. Ormerod, D., “The Development and Use of a Modified Immersion Tin as a High Performance Solderable Finish,” Proceedings of NEPCON West, Anaheim, CA, February 1998. 85. Ormerod, D. H., “Production Application of Flat Solderable Tin Finishes: Some Practical Considerations,” SMTA International, Chicago, 2000.
14.82
CHAPTER FOURTEEN
86. Ray, U., I. Artaki, and P. T. Vianco, “Influence of Temperature and Humidity on the Wettability of Immersion Tin Coated Printed Wiring Boards,” IEEE Transactions on Components, Packaging, and Manufacturing Technology, Part A, 18(1):153–162, March 1995. 87. Price, J. W., Tin and Tin-Alloy Plating, Electrochemical Publications, Ayr, Scotland, 1983. 88. Murphy, T., “Tin Bismuth Alloy Plating, a Fusible Low Temperature Etch Resist for High Aspect Ratio P.C. Boards,” Technical Paper presented at IPC Fall Meeting, IPC-TP-972, San Diego, CA, October 7–12, 1990. 89. Melton, C., A. Growney, and H. Fuerhaupter, “Immersion Plating of Tin-Bismuth Solder,” U.S. Patent 5,391,402, February 21, 1995. 90. Melton, C., A. Growney, and H. Fuerhaupter, “Immersion Plating of Tin-Bismuth Solder,” U.S. Patent 5,435,838, July 25, 1995. 91. Murphy, T., “Tin Bismuth Alloy Plating, a Fusible Low Temperature Etch Resist for High Aspect Ratio P.C. Boards,” Technical Paper presented at IPC Fall Meeting, IPC-TP-972, San Diego, CA, October 7–12, 1990. 92. “M&T Tin Sol B Tin-Bismuth Alloy Plating Process,” Technical Information Sheet No. PSn-TSB, M&T Chemicals, 1979. 93. Snowdon, K., “Lead-Free—The Notel Experience,” Proceedings: IPCWorks ’99 Conference, Minneapolis, MN, October 27, 1999. 94. E. Davis, Transactions Institute of Metal Finishing, 31, 401, (1954). 95. Izaki, M., H. Enomoto, and T. Omi, Plating and Surface Finishing, p. 84, June 1987. 96. Yee, S., “Tin/Lead Coating Directly on Copper,” SMI 96, San Jose, CA, September 10–12, 1996. 97. Holzmann, A., “An Overview of Solid Solder Deposits,” Proceedings of NEPCON West, Anaheim, CA, February 1998. 98. Marshall, H., “Hot Air Solder Leveling—The Lazarus Finish,” IPC Printed Circuits Expo, Long Beach, CA, April 26–30, 1998. 99. Goodell, S., “Fine Pitch Technology with Horizontal Hot Air Leveling,” in Proceedings of NEPCON West, pp. 1163–1164, Anaheim, CA, February 7–11, 1993. 100. Prasad, R., “Building Tomorrow’s PCBs,” Printed Circuit Fabrication Asia, 1(4):12–15, Winter 1993. 101. Klein-Wassink, R. J., Soldering in Electronics, 2nd ed., Electrochemical Publications, Ayr, Scotland, 1989. 102. Higson, J., “The Future of Hot-Air Leveling,” Printed Circuit Fabrication, p. 26, 2000. 103. Kehoe, M., “SIPAD SSD: Technology Update—Applications and Experiences in the Field,” Proceedings of NEPCON West, Anaheim, CA, February 2000. 104. DeBlis, J., “Implementing Solid Solder Deposits in a Manufacturing Environment,” SMTA International, Chicago, 2000. 105. Popelar, S., A. Strandjord, and B. Niemet, “A Compatibility Evaluation of Lead-Based and Lead-Free Solder Alloys in Conjunction with Electroless Nickel/Immersion Gold Flip Chip UBM,” IMAPS, Baltimore, MD, 2001. 106. Tuck, J., “Fine Pitch Japanese Style,” Circuits Assembly, pp. 22–25, February 1994. 107. Fan, C., J. A. Abys, and A. Blair, “Wirebonding to Palladium Surface Finishes,” Proceedings of NEPCON West, Anaheim, CA, February 23–27, 1997. 108. Kim, P. G., K. N. Tu, and D. C. Abbott, “Soldering Reaction Between Eutectic SnPb and Plated Pd/Ni Thin Films on Cu Leadframe,” Applied Physics Letters, 71(1):61–63, July 7, 1997. 109. Prasad, S., F. Carson, G. S. Kim, J. S. Lee, Y. C. Park, Y. S. Kim, K. S. Min, S. S. Lu, L. Hui, X. Hai, S. H. Khor, and C. L. Tan, “Plating Chemical Evaluations and Reliability Plating Chemical Evaluations and Reliability of Pb of Pb-Free Leadframe Packages Free Leadframe Packages,” Pan Pacific: February 13, 2001.
CHAPTER 15
IMPLEMENTATION OF LEAD-FREE SOLDERING 15.1 COMPATIBILITY OF LEAD-FREE SOLDERS WITH SMT REFLOW PROCESS Due to the toxicity of lead, there is a tremendous amount of effort to eliminate lead from the solders used in the electronics industry. The move toward lead-free solder alternatives in North America and Europe accelerated significantly1 since the Japanese industry announced its aggressive lead-free road map.2 For instance, Toshiba, Matsushita, and Hitachi have announced plans to eliminate all lead interconnects in their products by 2001, 2004, and 2004, respectively. However, the preferred solution for lead-free alternatives varies from region to region, and there are a number of alloys considered promising. The most favorable Pb-free solder systems identified by the industry3,4 comprise primarily alloys of Sn with Ag, Bi, Cu, Sb, or Zn, such as 99.3Sn0.7Cu, 96.5Sn3.5Ag, 95.5Sn3.8Ag0.7Cu, 93.6Sn4.7Ag1.7Cu, 96.2Sn2.5Ag0.8Cu0.5Sb, 91.7Sn3.5Ag4.8Bi, 90.5Sn7.5Bi2Ag, 89Sn8Zn3Bi, 95Sn5Sb, and 58Bi42Sn. Unfortunately, although some reliability data have been generated in the past,3 most of the promising alloys were evaluated under a single flux system. The compatibility between flux and alloy often dictates the performance of reflow soldering, such as solder balling, wetting, processing window, and stability. Since the flux chemistry varies from supplier to supplier, and since the use of more than one supplier is considered crucial for ensuring a steady process, an alloy compatible with a wider range of flux systems obviously will have greater prospects of being accepted by the surface-mount technology (SMT) industry. In this study, a group of the most promising Pb-free alloys reported are tested against a broad range of commonly used flux chemistries, such as water wash, no-clean, halide-containing, halide-free, nitrogen reflow systems, and air reflow systems, in the form of solder paste. The handling and reflow soldering performance of these pastes is evaluated and ranked in order to assess the prospect of the alloys being widely used for reflow soldering applications by the industry.
15.1.1
EXPERIMENTAL DESIGN FOR COMPATIBILITY EVALUATION
15.1.1.1 Materials Alloys. Among the most promising lead-free alloys, 10 representative alloys were chosen, including 99.3Sn0.7Cu, 96.5Sn3.5Ag, 95.5Sn3.8Ag0.7Cu, 93.6Sn4.7Ag1.7Cu, 96.2Sn2.5Ag0.8Cu0.5Sb, 91.7Sn3.5Ag4.8Bi, 90.5Sn7.5Bi2Ag, 89Sn8Zn3Bi, 95Sn5Sb, and 58Bi42Sn. The eutectic tin-lead 63Sn37Pb was used as a control. Fluxes and Solder Pastes. Ten fluxes varying widely in chemistry, as shown in Table 15.1, were used to make solder pastes in order to evaluate the compatibility of alloys with reflow soldering applications. The solder paste samples were made by mixing each flux with solder powder (−325/+500 mesh, 25 to 45 µm) for each alloy. 15.1 Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
15.2
CHAPTER FIFTEEN
The metal content of solder paste for each alloy system is shown in Table 15.2, and is set to provide approximately the same solder volume as that of eutectic SnPb solder pastes with 90 percent metal content when the flux density is 1 g/ml. For the purpose of calculation, densities of F1 to F9 can be approximated as 1.0 g/ml, while a density of F10 is 1.25 g/ml.The density of the alloys shown in Table 15.2 was determined with a pycnometer. 15.1.1.2 Tests. As stated earlier, the scope of this work involves assessing the compatibility of lead-free alloys with reflow soldering, with emphasis on the handling and soldering performance of solder pastes. In the case of handling, incompatibility of an alloy with a certain flux chemistry often results in excessive chemical
TABLE 15.1 Fluxes Used for Lead-Free Solder Pastes Flux
Description
F1
No-clean, halide-free, air reflow, probe-testable
F2
No-clean, halide-free, air reflow, probe-testable
F3
No-clean, halide-containing, air reflow
F4
No-clean, halide-containing, air reflow
F5
Rosin-based, mildly activated type, halide-containing, air reflow
F6
No-clean, halide-free, medium residue, nitrogen reflow
F7
No-clean, halide-free, low residue, nitrogen reflow
F8
No-clean, halide-free, ultra-low residue, nitrogen reflow
F9
Water-washable, halide-free, medium-temperature process, air reflow
F10
Water-washable, halide-containing, high-temperature process, air reflow
TABLE 15.2 Metal Content of Solder Paste Samples for Each Alloy System
Alloy
Density (g/ml)
Metal content for F1–F9 system (wt/wt %)
Metal content for F10 system (wt/wt %)
63Sn37Pb
8.40
90.0
88.0
96.5Sn3.5Ag
7.36
89.0
86.5
99.3Sn0.7Cu
7.34
89.0
86.7
95.5Sn3.8Ag0.7Cu
7.38
89.0
86.5
93.6Sn4.7Ag1.7Cu
7.43
89.0
86.6
96.2Sn2.5Ag0.8Cu0.5Sb
7.47
89.0
86.5
91.7Sn3.5Ag4.8Bi
7.57
89.0
86.8
90.5Sn7.5Bi2Ag
7.58
89.0
86.8
58Bi42Sn
8.56
90.0
88.2
95Sn5Sb
7.25
89.0
86.3
89Sn8Zn3Bi
7.39
89.0
86.5
15.3
IMPLEMENTATION OF LEAD-FREE SOLDERING
reaction between the alloy and flux either at storage temperature or on exposure to the ambient atmosphere. This in turn results in a thickened or crusted paste, and accordingly a poor shelf life and poor tack time. On the other hand, for certain alloys, the solder oxide may not be readily removable by certain flux chemistries. This would result in poor solder balling, poor wetting, and often a poor solder surface appearance. In certain other cases, some solder alloys may react with base metal very slowly and thus would exhibit fairly poor wetting when compared with eutectic SnPb systems. For situations like those, conventional flux systems may be inadequate, and a more aggressive flux may be needed in order to achieve wetting comparable with that of SnPb systems. In view of the aforementioned situations, it becomes clear that the tests required in order to assess the compatibility of solder alloys with reflow soldering should include (1) shelf life, (2) tack time, (3) solder balling, (4) wetting ability, and (5) solder joint surface appearance. However, before embarking on these performance evaluations, some other information may need to be generated, such as the melting temperature. Knowledge of melting behavior—particularly liquidus temperature— is required for setting up a reflow profile. Melting Temperature. For each alloy system, the melting temperature (see Table 15.3) was determined on a Seiko differential scanning calorimeter. The sample was preconditioned at 300°C, followed by cooling to 0°C at a cooling rate of 5°C/min, then reheated to 300°C at a heating rate of 5°C/min. The onset of the melting endotherm was recorded as solidus temperature, and the peak of the endotherm was recorded as liquidus temperature. The liquidus temperature data in Table 15.3 have not been corrected for thermal lag effect, which is 0.9°C. Wetting Ability. The wetting ability of solder pastes was tested by printing solder paste onto copper pads coated with organic solderability preservative (OSP) on a printed circuit board (PCB) and followed by reflow. The stencil thickness was 6 mil (150 µm), and the ratio of aperture opening versus pad dimension was 1 to 1. The registration of paste to pad was set 70 percent off so that only 30 percent of the paste was printed onto the pads and 70 percent was printed onto the solder mask. Upon reflow, the solder paste coalesced and pulled away from the solder mask and wetted to the pad side (see Fig. 15.1). The wetting ability of solders was determined by
TABLE 15.3 Melting Temperatures for Solder Alloys Solidus (°C)
Liquidus (°C)
Note
63Sn37Pb
Alloy
182.1
183.0
Eutectic
96.5Sn3.5Ag
219.7
220.8
Eutectic
99.3Sn0.7Cu
225.7
227.0
Eutectic
95.5Sn3.8Ag0.7Cu
216.3
217.5
Multicore
93.6Sn4.7Ag1.7Cu
215.9
217.1
Ames
96.2Sn2.5Ag0.8Cu0.5Sb
216.9
218.2
AIM
91.7Sn3.5Ag4.8Bi
202.1
215.1
Sandia
90.5Sn7.5Bi2Ag
190.6
214.7
Tamura
58Bi42Sn
136.3
138.5
Eutectic
95Sn5Sb
238.3
240.3
Indium
89Sn8Zn3Bi
190.6
195.4
Senju
15.4
CHAPTER FIFTEEN
FIGURE 15.1 Schematic of wetting test.
examining the extent of solder spreading on the pads, and the average of 10 pads was expressed as wetting index (WI), which is defined in Table 15.4. A higher WI value represents a better wetting ability. The reflow process was conducted with the use of a BTU VIP70 forced-air convection oven. Two tent-shaped reflow profiles were used for each alloy, with the peak temperature being a function of the liquidus temperature. The first profile (cool profile) exhibits a peak temperature 15°C above the liquidus temperature, with the second profile (warm profile) being 30°C above the liquidus. The ramp-up rate was about 0.7 to 0.8°C/s. For flux systems F6, F7, and F8, a nitrogen atmosphere was used for reflow, while for the remaining systems an air reflow atmosphere was used. The use of two profiles provided insights on (1) minimal temperature required and (2) potential for improving soldering performance with the use of a higher temperature.
TABLE 15.4 Definition of WI
WI
Spread area (% of pad)
0
0
1
10
2
20
3
30
4
40
5
50
6
60
7
70
8
80
9
90
10
100
IMPLEMENTATION OF LEAD-FREE SOLDERING
15.5
Solder Balling. Solder balling performance was evaluated by examining under a 20× optical microscope the average number of solder balls per pad for the reflow results just described. The average performance of 10 pads is expressed as solder balling index (SBI), as defined in Table 15.5. A higher SBI value represents a better solder balling performance.
TABLE 15.5 Definition of SBI SBI
Number of solder balls
0
No reflow
1
>501, with some reflow
2
401–500
3
301–400
4
201–300
5
151–200
6
101–150
7
51–100
8
21–50
9
11–20
10
0–10
Tack Time. The tack time of a solder paste was determined using the following procedure: (1) Print solder paste onto ceramic coupons, as prescribed by J-STD-006 procedure. (2) Condition the specimen under 76 percent relative humidity. (3) Measure the tack value, per J-STD-006 procedure, of the conditioned specimen at fresh, 8 h, 24 h, 48 h, and 72 h. The specimen was discarded after each tack measurement. The tack data are expressed as tack time index (TTI), which is defined in Table 15.6. A higher TTI value represents a longer tack time. Shelf Life. The shelf life of solder pastes was determined by monitoring their viscosity stability at 25°C over a period of 1 month. A changing viscosity, typically increasing with time, was considered undesirable. For each solder paste sample, the viscosity was determined at 1 day, 7 days, and 30 days after paste manufacturing. The
TABLE 15.6 Definition of TTI TTI
Description
0
Decreasing tack curve reaching a value at 25%
2
Overall instability = 20–25%
4
Overall instability = 15–20%
6
Overall instability = 10–15%
8
Overall instability = 5–10%
10
Overall instability = 0–5%
Solder Surface Appearance. The solder bump surface was examined under an optical microscope, and the appearance is expressed as solder appearance index (SAI), as defined in Table 15.8. A higher SAI value suggests a more desirable solder joint quality. Compatibility. The compatibility of an alloy with reflow soldering was determined by adding up the performance of all five categories. However, much more weight was assigned to solder balling and wetting performance. The solder appearance received a slightly higher weight than shelf life and tack time. Hence, the compatibility C was calculated according to the following formula: C = 1 × SBI + 1 × WI + 0.3 × SLI + 0.3 × TTI + 0.4 × SAI TABLE 15.8 Definition of SAI SAI
Description
0
Not reflowed, with “powder” appearance
1
Partially reflowed with some melted area
2
No less than one large cavities
3
Numerous small pinholes
4
Solder dewetted to form a few bumps
5
Rough area 70–100%
6
Rough area 50–70%
7
Rough area 30–50%
8
Rough area 10–30%
9
Smooth and dull, or shiny with slight roughness (rough area Ag > In > Bi, based on the test results of a series of binary solders. The inferior wetting ability of Bi versus Pb at high content level may be caused by a metallurgical factor that overrides the surface tension factor. These spreading data suggest that the flux desired for
TABLE 15.15 Summary of Compatibility of Flux Systems with a Variety of Alloys
Flux F1 (NC, air, no-X, probe)
SBI
WI
Sum of SBI and WI
SLI
TTI
Sum of SLI and TTI
7.4
5.2
12.6
8.4
7.6
16.0
F2 (NC, air, no-X, probe)
7.1
5.5
12.6
7.6
6.5
14.1
F3 (NC, air, X)
7.1
6.0
13.1
6.9
7.6
14.5
F4 (NC, air, X)
7.5
5.4
12.9
6.9
7.6
14.5
F5 (RMA, air, X)
7.6
6.4
14.0
5.3
1.1
6.4
F6 (NC, air, no-X)
8.5
6.1
14.6
4.5
4.4
8.9
F7 (NC, N2, no-X, low R)
9.0
5.0
14.0
8.5
0.0
8.5
F8 (NC, N2, no-X, ultra-low R)
8.9
5.9
14.8
8.5
6.2
14.7
F9 (WS, air, no-X, med. temp)
4.0
2.6
6.6
9.1
1.1
10.2
F10 (WS, air, X, high temp)
7.8
6.8
14.6
4.4
5.6
10.0
Courtesy of Benlih Huang and Ning-Cheng Lee, Indium Corporation of America, published in IMAPS’99— Chicago.
15.28
CHAPTER FIFTEEN
58Bi42Sn should exhibit not only a lower activation temperature, but also a higher flux activity to compensate for the difference caused by the alloy. For SnZnBi solder, the poor performance is caused by the high oxide content as well as the high oxidizing activity of solder at reflow. Flux needed for this alloy should have considerably high flux capacity and high oxygen barrier ability. The latter is required to minimize the formation of new oxide, as reported by Lee.14 In Table 15.9, the only flux that provided marginal performance for this alloy was F10. F10 was formulated to handle air reflow for temperature up to more than 300°C, hence satisfying the requirement of both high flux capacity and high oxygen barrier ability. Fluxes containing tin metalloorganic compounds may also help wetting by decomposing and plating a tin layer on top of copper during soldering.15 However, before adopting this flux chemistry, the potential for circuit shorts or leakage current should be closely examined. Other approaches may also be available. At least one flux has been reported to perform acceptably at SMT application.16 As for the rest of the lead-free alloys, such as SnAgCu or SnAgBi, the demand for flux ability is less than that for SnZnBi, but follows the same direction. In general, an increase in flux capacity or oxygen barrier ability on the top of existing flux systems will be needed. This may be achieved relatively easily by improving current RMA, high-solid-content no-clean, and of course water-washable flux systems, as demonstrated by fluxes F5, F6, and F10 in Table 15.15. All three fluxes are among the top performers of existing fluxes, and therefore are the most promising systems to be successfully upgraded. It should be noted that the oxygen barrier ability can be substituted by employing an inert reflow atmosphere, as confirmed by the high performance of fluxes F7 and F8, and was elucidated in theory by Lee.14 Figure 15.13 indicates that an improved soldering performance can be achieved by employing either a flux with a lower K value or a reflow atmosphere with a lower oxygen partial pressure. A lower K value means a higher rosin or resin content, which serves as an oxygen barrier.
FIGURE 15.13 Theoretical relation between soldering performance S and oxygen partial pressure P of reflow atmosphere. K = R0/R, where R0 is rosin or resin content (50 percent) of regular RMA flux and R is that of the flux under analysis. (Courtesy of N. C. Lee.14)
IMPLEMENTATION OF LEAD-FREE SOLDERING
15.29
Besides regulating the oxygen barrier content or oxygen partial pressure, the soldering performance can also be enhanced by using a more effective activator system. Among all of the activators, halides are considered the most effective in terms of fluxing performance per unit flux volume, and accordingly are the top candidates for improving fluxes. Other organic chemicals may also be considered. However, the high activator solid volume required tends to cause too high a viscosity for solder pastes. A self-evident requirement is the thermal stability. Thermally decomposed or oxidized organic chemicals will lose the properties exhibited at a lower temperature. Without pertinent thermal stability, none of the features mentioned can be realized. In general, all of the flux ingredients should be able to survive the whole reflow process.
15.5 FLUX DESIRED FOR LEAD-FREE PASTE HANDLING The handling performance is dictated by (1) reactivity of solder surface with fluxes and (2) solder surface texture and shape.A high reactivity will cause increasing paste viscosity and decreasing flux capacity. On the other hand, a rough surface texture or irregular powder shape may promote excessive surface adsorption of chemicals used in fluxes, hence altering the composition of flux medium in solder paste. Except for SnZnBi alloy, almost all of the lead-free solder alloys have comparable or slightly better handling performance than 63Sn37Pb. This suggests that there is virtually no need to further improve the flux chemistry for handling most of the lead-free solder pastes. As to SnZnBi solder system, obviously a highly stable flux system is needed in order to retard the reaction between Zn and fluxes. At least one flux has been reported to perform properly for handling.16
15.6 CLEANING PERFORMANCE OF LEAD-FREE SOLDER PASTE Lead-free solder pastes face challenges not only from soldering and handling, but also from a cleaning perspective. Similar to eutectic Sn-Pb solder systems, the requirement for cleaning may also be demanded for no-clean solder pastes for some applications. Examples include (1) recovering boards suffering solder balling defects, (2) a single paste for both no-clean and cleaning customers for contract manufacturers, (3) high-frequency radio frequency applications, (4) integrated circuit packages, (5) military applications, (6) automotive applications, (7) aerospace applications, and (8) medical applications. In order to understand the impact of lead-free flux technology on cleaning, 25 commercial lead-free solder pastes were tested with eight cleaner chemistries and several cleaning processes. The 25 pastes tested are shown in Table 15.16, and are categorized into three groups—no-clean hard residue, no-clean soft residue, and water washable. The soft residue group also includes probe-testable pastes. The alloys tested include SnAg, SnAgCu, and SnAgCuSb, with melting temperature estimated to be approximately 215 to 221°C.
15.30
CHAPTER FIFTEEN
TABLE 15.16 Solder Pastes Used in the Cleaning Study No-clean hard residue
Alloy
1
No-clean, amber hard residue
SnAgCuSb
2
No-clean
SnAgCu
3
No-clean, halide-free rosin, high-reliability electronics
SnAgCu
4
No-clean, same as 3 with slight halide content
SnAgCu
5
RMA, synthetic rosin type
SnAgCu
6
No-clean RMA highly active resin/rosin-based formulation
Sn96.5Ag3.5
7
No-clean mildly activated resin-based formulation
SnAgCuSb
8
No-clean mildly activated resin-based formulation
SnAgCuSb
9
No-clean
Sn95Ag5
10
No-clean, extended work life
Sn95Ag5
11
RMA, extended work life
Sn95Ag5
12
No-clean
Sn95.5Ag4.0Cu.5
13
RMA, mildly activated resin paste flux No-clean soft residue
Sn95.5Ag4.0Cu.5 Alloy
14
No-clean solder paste
Sn95.8Ag3.5Cu.7
15
No-clean, clear light soft residue
Sn95.5Ag3.9Cu.6
16
No-clean solder paste, soft residue
Sn95.5Ag4.0Cu.5
17
No-clean, clear light soft residue
Sn95.5Ag4.0Cu.5
18
No-clean, pin-penetrable low residue
SnAgCu
19
No-clean, pin-penetrable low residue
SnAgCuSb
20
No-clean, pin-penetrable low residue
SnAgCu
21
No-clean pin probe–testable mildly activated resin-based formulation
SnAgCuSb
Water soluble
Alloy
22
Water soluble
Sn95.5Ag3.9/Cu.6
23
Water soluble, amber semiliquid
Sn95.5Ag4.0Cu.5
24
Water soluble
Sn95Ag5
25
Water soluble, polymer/dendrimer activator system
Sn95.5Ag4.0Cu.5
15.6.1
CLEANING RESULTS
The cleaning results are shown in Table 15.17. Note the cleaning processes are categorized as (1) semiaqueous and/or aqueous-solvent sprayable cleaning, (2) saponified aqueous spray cleaning, and (3) solvent boil-rinse-vapor degrease. The cleaning performance is further displayed in Tables 15.18 and 15.19 to show the effect of flux chemistry or cleaning chemistry/process on cleaning.
IMPLEMENTATION OF LEAD-FREE SOLDERING
15.31
15.7 FLUX DESIRED FOR LEAD-FREE RESIDUE CLEANING Flux chemistry with a soft residue appears to exhibit a better cleanability than that with a hard residue, as shown in Table 15.18. This can be attributed to the lack of crystal formation in the soft residue, thus allowing it to be dissolved relatively easily into the cleaner. For hard residues, the crystal formation is fairly common and the dissolution process would have to overcome the crystallization energy before the residue molecules can be pulled away from the main residue body. Water-washable flux residues also display a relatively good cleanability, presumably due to the same factor. Although the solder alloy composition does vary from SnAg to SnAgCu to SnAgCuSb, its impact on flux cleanability is considered minimal due to the comparable process temperature and the dominant constituent being the same (Sn). Although not supported with experimental data, it has been noted that lead-free solder pastes pose a greater difficulty in cleaning than their eutectic SnPb counterparts.17 The authors concur with this statement and attribute this phenomenon to three factors. The first is a higher reflow temperature, which causes more side reactions within the flux such as oxidation or cross-linking reaction. The second is the greater flux activity needed to boost the wetting of lead-free solders. This higher flux activity may induce more side reactions, thus causing greater difficulty in cleaning, as exemplified by the slightly poorer cleanability of paste 4 than paste 3 in Table 15.17. According to the paste manufacturer, pastes 3 and 4 have virtually the same flux chemistry, except that paste 4 contains halide and exhibits a slightly higher activity than paste 3. The third factor is the greater amount of tin salt formation at reflow due to the use of high-tin solders. Tin salts seem to cause greater difficulty in cleaning than lead salts, and thus may result in more white residues.This stipulation is supported by the observation that the white residues encountered with 63Sn37Pb solder pastes often are not observed for high-lead solder pastes, even if the same flux is employed for both occasions. This reduced cleanability associated with lead-free solder pastes does not have to be an unsolvable problem. Although use of high reflow temperature and fluxes with higher flux activity is still inevitable, the thermal stability of flux ingredients still has room for improvement. Many organic chemicals exhibit thermal stability up to 300°C (or even up to 350°C), thus allowing plenty of area to be explored.
15.8 CLEANING CHEMISTRY/PROCESS DESIRED FOR LEAD-FREE RESIDUE CLEANING By reviewing Table 15.19, referring back to Table 15.17, it is clear that the most effective cleaning system is saponified aqueous with spray (average 3.49). The least effective system is the solvent boil-rinse-vapor degrease process (average 3.03). Semiaqueous and/or aqueous solvent sprayable cleaning processes fall in between (average 3.27). However, it is interesting to note that the best two cleaning systems (4.00 and 3.98) all turn out to be within this solvent and/or aqueous spray category, as shown in Table 15.17. The fact that both top performers in cleaning belong to the semiaqueous and/or aqueous solvent sprayable process demonstrates that the flux residues of lead-free solder pastes are still fairly soluble and do not have to rely on the saponification reaction in order to pull the residues into the cleaner. The cleanability is highly dependent on the solvency of cleaner used, with results ranging from
15.32
Semiaqueous #1 Semiaqueous #2
Spray under immersion inline Spray under immersion inline
Boil-rinse-vapor w/o ultrasonic Average
1 No-clean, amber hard residue 2.25 3.25 4 4 4 3.5 2.75 4 4 4 3 1 3.31
2 No-clean 2.5 3 1.25 4 4 4 2.5 4 4 3.5 2 1 2.98
3 No-clean, halide-free rosin, high-reliability electronics 3 2.25 1 4 4 4 3 4 4 4 3 1 3.1
4 No-clean, same as 3 with slight halide content 2.5 3.5 1 4 4 3 3.75 4 4 3.5 1.25 1 2.96
5 RMA, synthetic rosin type 3 2 2 4 4 3.75 4 4 4 4 3 2.25 3.33
6 No-clean RMA highly active resin/rosin-based formulation 3.5 4 4 4 4 3.5 3 4 4 4 3 1.25 3.52
Boil-rinse-vapor w/ ultrasonic
Boil-rinse-vapor w/o ultrasonic
Boil-rinse-vapor w/ ultrasonic
Boil-rinse-vapor w/o ultrasonic
Spray in air—inline
Spray in air—batch
Semiaqueous #2
Spray under immersion batch
Cleaner chemistry
Solvent vapor deg. #3
Solvent vapor deg. #3
Solvent vapor deg. #2
Solvent vapor deg. #2
Solvent vapor deg. #1
Saponifier
Saponifier
Aqueous solvent sprayable #2
Aqueous solvent sprayable #1
Spray in air—inline
Machine
Spray in air—inline
Paste
TABLE 15.17 Results of Lead-Free Solder Paste Flux Residue Cleaning Study
7
No-clean mildly activated resin-based formulation
2
2
1
4
4
4
2
4
4
4
2
1
2.83
8
No-clean mildly activated resin-based formulation
2
1.25
1.25
4
4
3
2
4
4
4
2
1
2.71
9
No-clean
3
3.25
1
4
4
4
3.5
4
4
4
3.4
2
3.35
15.33
10
No-clean, extended work life
1.75
1
1.25
4
4
3.75
3.5
4
4
4
2.25
1
2.88
11
RMA, extended work life
2.25
1
1.25
4
4
3.25
3.5
2.25
4
4
2.9
1
2.78
12
No-clean
3.5
4
4
4
4
4
3.5
3.9
4
4
3.5
2.25
3.72
13
RMA, mildly activated resin paste flux
2
1.75
1
4
4
4
2.25
4
4
4
1
1
2.75
14
No-clean solder paste
4
2.25
0.75
4
4
4
4
4
4
4
3
1.25
3.27
15
No-clean, clear light soft residue
2
2.5
1.75
4
4
2
3.25
4
4
4
2
1
2.88
16
No-clean solder paste, soft residue
2.75
4
4
4
4
2.5
3.5
3.5
3.9
3.25
2.25
1
3.22
17
No-clean, clear light soft residue
3.65
4
4
4
4
2.5
3.5
3
3
3
3.25
1.25
3.26
18
No-clean, pin-penetrable low residue
3.5
3
3
4
4
4
4
4
4
4
3.4
2
3.58
19
No-clean, pin-penetrable low residue
4
3.25
2.75
4
4
4
4
4
4
4
4
2
3.67
20
No-clean, pin-penetrable low residue
2.25
2
2
4
4
4
3
4
4
4
3.5
1.75
3.21
21
No-clean pin probe–testable mildly activated resin-based formulation
2.5
4
4
4
4
3
4
4
4
4
3.4
2.5
3.62
22
Water soluble
4
4
4
4
4
4
4
1
4
1.75
3
1.9
3.3
23
Water soluble, amber semiliquid
4
4
4
4
4
4
4
2
3.5
2.25
3
2.25
3.42
24
Water soluble
4
4
4
4
4
4
4
0.5
4
3
3.25
1
3.31
25
Water soluble, polymer/dendrimer activator system
4
4
4
3.5
4
4
4
2
3
1.5
2.25
1.9
3.18
Average
2.96
2.93
2.49
3.98
4
3.59
3.38
3.45
3.9
3.59
2.74
1.46
3.21
Grading scale: 0 = no cleaning, 1 = significant residue, 2 = medium residue, 3 = low residue, 4 = totally clean.
15.34
CHAPTER FIFTEEN
TABLE 15.18 Effect of Flux Chemistry on Cleaning Performance Flux chemistry
Cleaning performance
No-clean, hard residue
3.11
No-clean, soft residue
3.35
Water washable
3.30
TABLE 15.19 Effect of Cleaning Chemistry/Process on Cleaning Performance Cleaning chemistry/process
Cleaning performance
Semiaqueous and/or aqueous solvent sprayable cleaning
3.27
Saponified aqueous spray cleaning
3.49
Solvent boil-rinse-vapor degrease
3.03
fairly poor (2.49) to totally clean (4.00). The reason that this category falls below the saponified aqueous with spray category can be attributed to varying process parameters of time, temperature, solvency, and impingement energy. There is another factor working against the saponification aqueous spray system. The SMT trend is driving toward a no-clean process, with no-clean already playing a dominant role in the industry as of today. Since all no-clean flux residues are expected to be hydrophobic and hardly dissolve in water, the water-based saponification approach really is in a poor starting position if the goal is to clean all types of leadfree paste residues. For water-washable solder paste systems, the saponification aqueous spray system is still believed to be one of the top choices, as shown in Table 15.17, where all water-washable pastes received scores of 4.0 on cleaning. Perhaps it can be concluded that, with proper choice of solvents, the solvent and/or aqueous spray system is the most desirable cleaning approach for residues of lead-free solder pastes. Spray is critical for the success of cleaning. It is the primary difference between the semiaqueous and/or aqueous solvent sprayable system and the solvent boilrinse-vapor degrease system. The results appear to be insensitive to spray-in-air or spray-under-immersion, as long as spray is applied. Ultrasonic appears to be secondary in effectiveness. Table 15.20 shows the comparison of systems with and without ultrasonic. Systems with ultrasonic aid display a considerably better cleaning performance.
TABLE 15.20 Effect of Ultrasonic Agitation on Cleaning Performance Cleaning chemistry/process
Ultrasonic on/off
Cleaning performance
Solvent vapor degreaser
On
3.90
#2
Off
3.59
Solvent vapor degreaser
On
2.74
#3
Off
1.46
TABLE 15.21 Guidelines for Selecting Lead-Free Solder Paste
Property
Test method
Characteristics to be examined
Remark
15.35
Solderability
Reflow through a production reflow furnace
Spreading, solder balling, solder appearance of a fine dot print
Wide profile matrix and air reflow recommended. Peak temp. range 230– 260°C.
Corrosion
Cu corrosion test at 40°C/93% relative humidity
Discoloration of copper coupon
J-STD-004 Cu corrosion test.
SIR
SIR
SIR reading, dendritic formation
J-STD-004 SIR test.
Cleanability (for cleaning required applications)
Production cleaning process
Ionics, flux residue
Ionics measurement is not meaningful for no-clean flux or solder paste.
Printability
Print speed up to 200 mm/s, aperture down to 10 mil diameter
Print definition for fine dots
Type 4 powder may be needed for 10 mil diameter aperture.
Open time
Print, tack check, place component, reflow
Printability, tack, solderability with increasing open time
80% relative humidity recommended for exposure atmosphere.
Assembly yield
Actual production assembly process
Defects such as solder balling, bridging, tombstoning, nonwetting, voiding, etc.
Consult with paste supplier regarding handling and profile.
Reliability
Temp. cycling, vibration
Electrical continuity
Test condition application dependent.
Testing sequence starts from top.
15.36
15.9
CHAPTER FIFTEEN
SELECTION OF LEAD-FREE SOLDER PASTE
Implementing lead-free soldering requires a full scope of concerted upgrading efforts, including solder materials, reflow equipment, components, boards, and inspection equipment. In the case of solder paste materials, the primary challenge is delivering a good soldering performance, since the prevailing lead-free solders all perform more poorly than eutectic SnPb. To further aggravate the wetting difficulty, the most wettable surface finish, hot-air solder leveling, is gradually fading away from the industry due to the overall migration toward fine-pitch design. At this stage, none of the lead-free surface finishes can provide a wettability as good as that of hot-air solder leveling. Since the move toward use of lead-free solder alloys and lead-free finishes is inevitable, the industry can only hope that the intrinsic poor solderability associated with Pb-free solders and finishes can be compensated by a “better” flux, or a flux enabling the Pb-free soldering quality matching that of eutectic SnPb systems. Unfortunately, this is a wish that cannot be easily met. Although it is possible to develop a Pb-free solder paste with soldering performance matching that of eutectic SnPb, the natural trade-off is typically a high corrosivity and a poor cleanability. Other features such as rheology or solder paste handling often are less affected by the migration toward lead-free flux chemistry. Selection of lead-free solder paste has to be a very cautious process as regards not only testing the obvious requirement but also examining the most likely compromises in performance, which often are not obvious. The recommended guideline for selecting lead-free solder paste can be summarized in Table 15.21. For eutectic SnPb solder paste evaluation, the corrosion and surface insulation resistance (SIR) often are evaluated at a later stage. However, for lead-free solder paste evaluation, the two properties are recommended to be investigated once the paste has passed the solderability test. This is due to the high probability that many pastes may have overcompromised the noncorrosion feature in order to deliver the soldering performance. This selection plan should be regarded as a guideline only. Detailed testing items and test conditions can be modified according to the requirements of each assembly operation.
REFERENCES 1. Buetow, M., “The Latest on the Lead-Free Issue,” Technical Source, IPC 1999 Spring/ Summer Catalog. 2. Bradley, E., “Overview of No-Lead Solder Issue,” NEMI meeting, Anaheim, CA, February 23, 1999. 3. “Lead-Free Solder Project Final Report,” NCMS Report 0401RE96, Ann Arbor, MI, August 1997. 4. Richards, B. P., C. L. Levoguer, C. P. Hunt, K. Nimmo, S. Peters, and P. Cusack, “An Analysis of the Current Status of Lead-Free Soldering,” British Department of Trade and Industry Report, April 1999. 5. Lee, N. C., “Optimizing Reflow Profile via Defect Mechanisms Analysis,” Proceedings of IPC Printed Circuits Expo, 1998. 6. Hance, W. B., and N. C. Lee, “Voiding Mechanisms in SMT,” Proceedings of the 17th Annual Electronics Manufacturing Seminar, China Lake, CA, 1993. 7. Diepstraten, G., “Analyzing Lead-Free Wavesoldering Defects,” SMT’s Guide to Lead-Free Soldering, 2–5, June 2001.
IMPLEMENTATION OF LEAD-FREE SOLDERING
15.37
8. Barbini, D., “Wave Soldering with Lead-Free Alloys,” NEMI meeting, January 17, 2001. 9. Prasad, S., F. Carson, G. S. Kim, J. S. Lee, P. Roubaud, G. Henshall, S. Kamath, A. Garcia, R. Herber, and R. Bulwith, “Board Level Reliability of Lead-Free Packages,” SMTA International, Chicago, IL, September 24–28, 2000. 10. Butterfield, A., V. Visintainer, V. Goudarzi, “Lead Free Solder Flux Vehicle Selection Process,” SMTA International, Chicago, IL, September 20–24, 2000. 11. Shina, S., H. Belbase, K. Walters, T. Bresnan, P. Biocca, T. Skidmore, D. Pinsky, P. Provencal, and D. Abbott, “Design of Experiments for Lead Free Materials, Surface Finishes and Manufacturing Processes of Printed Wiring Boards,” SMTA International, Chicago, IL, September 20–24, 2000. 12. Carrol, M. A., and M. E. Warwick, “Surface Tension of Some Sn-Pb Alloys: Part 1—Effect of Bi, Sb, P, Ag, and Cu on 60Sn-40Pb Solder,” Materials Science and Technology, 3: 1040–1045, December 1987. 13. Humpston, G., and D. M. Jacobson, “Principles of Soldering and Brazing,” ASM International, Materials Park, OH, 1993. 14. Lee, Ning-Cheng, “A Model Study of Low Residue No-Clean Solder Paste,” Nepcon West, Anaheim, CA, 1992. 15. Vaynman, S., and M. E. Fine, “Fluxes for Lead-Free Solders Containing Zinc,” SMTA International, Chicago, IL, September 20–24, 2000. 16. Showa Denko, “Development of Sn-Zn Solder Paste of High Reliability,” IPCWorks’99, Minneapolis, MN, October 27, 1999. 17. Bivins, B. A., A. A. Juan, B. Starkweather, N. C. Lee, and S. Negi, “Post-Solder Cleaning of Lead-Free Solder Paste Residues,” SMT International 2000, Chicago, IL, September 2000.
This page intentionally left blank
CHAPTER 16
CHALLENGES FOR LEAD-FREE SOLDERING
The implementation of lead-free soldering is not a smooth ride. Challenges surface one by one in all aspects, including surface finishes, soldering processes, and reliability. In general, most of the challenges can be answered with certain approaches, although some of those approaches remain hypothetical or theoretical. However, some challenges still remain unanswered. In this chapter, most of the challenges encountered in lead-free soldering are listed and discussed.
16.1
CHALLENGES FOR SURFACE FINISHES
SnPb plating and hot-air solder leveling (HASL) have been used by the electronics industry for decades. Although technical challenges still exist, in general most of the bugs have been worked out, and applications and usages of SnPb surface finishes are regarded as a routine operation with the major emphasis on maintaining control. The same cannot be said for lead-free surface finishes. Due either to the specific chemistry utilized or to the short history of these finishes, virtually every lead-free surface finish exhibits some challenges. In this section, several of these challenges are listed and discussed.
16.1.1
BLACK PAD
Electroless nickel/immersion gold (ENIG) is one of the prevailing lead-free surface finishes. It has been used by the industry for years due to its excellent solderability for fine-pitch surface-mount technology and ball grid array (BGA) package devices. This is particularly true for thick boards, where the electroplating process experiences difficulty in providing even plating for through-holes. However, sporadic solder joint failure may occur due to joint weakness. A number of investigations have been conducted.1–6 Puttlitz1 first reported a phenomenon called black pad associated with defective joints, in which the pads exhibit a dark gray to black appearance that will only partially wet. This observation was confirmed by later investigators.2–6 An example of black pad is shown in Fig. 16.1. Biunno4 studied the types and formation mechanisms of black pads. He classified the black pad phenomenon into eight categories, listed by increasing extent of the syndrome: (1) minimal immersion gold (IG) spike penetration, (2) deep IG spike penetration, (3) shallow spreading IG penetration, (4) deep spreading IG penetration, (5) IG separation of electroless nickel nodules, (6) small section black band, (7) corner section black band, and (8) large section black band. Biunno concluded that the black pad defect was caused by a hyperactive “corrosive” IG process that changes the nearsurface microstructure of PNi into a black band, with a marginal to total nonwetting state. The black pad defect is classified in terms of hyperactive corrosive activity. 16.1 Copyright © 2003 by The McGraw-Hill Companies, Inc. Click here for terms of use.
16.2
CHAPTER SIXTEEN
FIGURE 16.1 Black pads produced by applying +1 V during the plating process. The four black pads on the right were connected to +1 V, and the lighter pads from the left to middle were connected to ground. Under this test condition, all pads connected to +1 V turned black and all pads connected to ground turned orange.
The black pad defect does not occur in a random or sporadic manner; rather, it usually shows a clean separation at the transformed Ni surface. Virtually no NiSn intermetallic is observed at the Ni surface or on the component lead, as shown in Fig. 16.2. The back surface has a “mud-cracked” appearance. In addition, the phosphorus concentration exceeds 10 wt%. Volume enrichment of phosphorus by defect activity occurs via the removal of Ni atoms into a depth of the near surface, and at the same time no gold is deposited during the defect corrosion process. There is a nearly complete absence of NiSn intermetallic. Obviously, the presence of either black band or phosphorus inhibits the formation of NiSn intermetallics, thus reducing the bond strength of solder joints. However, phosphorus enrichment also occurs through natural intermetallic formation. Here Ni is depleted from the near surface due to formation of intermetallic. Therefore, phosphorus enrichment should not be regarded as a direct cause of weakened solder joints. High-magnification scanning electron microscopy (SEM) suggests that black band is voided or less dense than the underlying bulk Ni layer. This is supported by the results from focused ion beam microprobe study. Figure 16.3 shows a micrograph of black band. The ion beam milling produced some damage and preferential etching. However, it can be seen that the near-surface structure of black band is voided and less dense, as indicated by the side wall structure of the grove produced by the gallium ion beam slicing. Advanced corrosion of the nickel surface by the IG chemistry may be induced by an electric charge imbalance between the printed circuit board (PCB) and ionic species within the IG plating bath. This stipulation is supported by experiments where black pads were produced by applying +1 V during the plating process, as shown in Fig. 16.1. Highly accelerated gold plating occurred on all the pads connected to ground, and large cubic gold crystals were observed throughout the orange pad surface. The average grain size for the orange pads was more than 1000 times that of normal IG. The orange color of these pads is due to the large grain size and surface roughness. High phosphorus content is not always associated with black pad, as reported by Mei et al.5 Instead, an extraordinarily high carbon content is observed. The cause of
CHALLENGES FOR LEAD-FREE SOLDERING
FIGURE 16.2 Plane view of the surface structure of the black pad after solder joint failure.4
FIGURE 16.3 Focused ion beam microprobe micrograph of a spike/spread defect region.4
16.3
16.4
CHAPTER SIXTEEN
this high carbon content is not understood yet. However, the small electrical potential bias as the proposed root cause of black pad by Biunno may still be applicable in Mei’s work. In Mei’s failed boards, the failed packages were only a few plastic quad flat pack packages, and the failure occurred more frequently on a particular PQFP as well as on certain lead locations. Mei et al. speculate that there were small differences in electrical potential among packages, and among all leads of the plastic quad flat pack, due to the complex circuitry design of the board. This small electric potential difference may have induced different kinds or different degrees of chemical reactions in the IG bath. Overall, the black pad phenomenon has been studied in depth regarding its physical structure. Although understanding of the formation mechanism in terms of chemistry is still lagging behind, black pad can be controlled through a tight control of plating conditions.
16.1.2
EXTRANEOUS/SKIP PLATING
For the ENIG system, critical properties of the surface on which plating will occur include contamination, organics, roughness, residual copper, residual solder mask, oxidation, and residual tin. Lack of control of these critical properties can cause either extraneous plating or skip plating. The example of extraneous plating shown in Fig. 16.4 was caused by residual copper between traces.7 Excessive extraneous plating may cause circuit shorts. Skip plating is seen on some boards as areas that do not plate nickel, and has been discussed by Young,8 who proposes that static is cre-
FIGURE 16.4 Example of extraneous plating for ENIG system.
CHALLENGES FOR LEAD-FREE SOLDERING
16.5
ated in the solder mask operation on certain capacitive areas of the circuitry. This static attracts volatiles during the mask cure operation and is the underlying cause of skip plating.
16.1.3
TIN WHISKER
In the autumn of 1998, the NASA Goddard Space Flight Center (GSFC) was informed of an on-orbit commercial satellite failure attributed to a tin whisker– induced short circuit. The source of the tin whiskers was a pure tin–plated relay.9 Tin whisker is a growth of tin crystal protrusion on the surface of a tin-based metal, including tin-based surface finish. The shape of a tin whisker may vary from fiberlike to highly irregular. Due to its threat of inducing circuit shorts, tin whisker has been the subject of numerous investigations since the early 1950s.10–58 Tin whisker can be formed not only on pure tin surfaces but also on some tin alloys. Figure 16.5 shows the tin whisker formed on 98Sn2Cu surface finish plated on copper. The laminated copper can be recognized by its grain orientation. The intermetallics are concentrated at the tin-copper interface and grain boundaries.59 In a study of ChipPAC, Prasad et al.60 also observed whisker formation on pure tin, SnCu, and SnBi surface finishes, as shown in Fig. 16.6. In order to understand and prevent the formation of tin whisker, Lee and Lee61 studied the spontaneous growth mechanism of tin whisker. The authors theorized that the generation of compressive stress in tin film is caused by the diffusion of copper from the substrate into the tin along its grain boundaries and the subsequent formation of Cu6Sn5. The whisker growth is attributed to the compressive stress. Diffusion of copper from the substrate into the tin as a cause of tin whisker formation is supported by the findings of Boguslavsky of Shipley.62 In that work, tin whisker development was studied for samples with eutectic SnCu surface finish with a thickness of 10 to 12.5 µm. In one set of samples, a Ni underlayer about 1.5 µm thick was applied between the SnCu surface finish and the copper base. In another
FIGURE 16.5 Focused ion beam images of tin whisker before and after gallium ion beam milling for 98Sn2Cu plating on copper base.
16.6
CHAPTER SIXTEEN
FIGURE 16.6 Whisker photos for pure tin, SnCu, and SnBi surface finishes.
set of samples, no Ni underlayer was applied. Both sets of samples were stored at 55°C in dry heat. Results indicate that the samples without the Ni underlayer displayed whiskers in a few days, while the samples with the Ni underlayer showed no whisker growth. SEM/electron diffraction analysis indicates development of semicontinuous copper diffusion from the Cu base toward the SnCu surface finish for samples without the Ni underlayer. On the other hand, no copper diffusion can be discerned at all for samples with a Ni underlayer, as shown in Fig. 16.7. Apparently a Ni underlayer effectively stops copper diffusion into nickel. Without the Ni underlayer, there is copper diffusion, presumably through the grain boundaries. The light gray spots dispersed from the interface toward the dark background are presumably CuSn intermetallics. The dark background color is the 99.3Sn0.7Cu surface finish. Lee and Lee’s theory that the whisker growth is caused by compressive stress was checked by Fan at Lucent EC&S.63 Fan studied the whisker growth of bright tin and
FIGURE 16.7 SEM/EDS for 99.3Sn0.7Cu surface finish with and without Ni underlayer on top of copper base.62
16.7
CHALLENGES FOR LEAD-FREE SOLDERING
satin bright tin finish by monitoring the stress of surface finishes by x-ray diffraction. Both finishes showed zero stress after plating. After 4 months of room temperature storage, a compressive stress of about −10 MPa for bright tin or −7 MPa for satin bright tin in the tin plated directly on copper was measured. On the other hand, a tensile stress of about 10 MPa for bright tin or 7 MPa for satin bright tin plated on a nickel underlayer over copper was registered. After 4 to 18 months, the stress levels did not show significant change. Whiskers were found on the tin without a nickel underlayer, while no whiskers were found for tin with a nickel layer. Similar results were seen for the tin finishes after 18 months of aging at 50°C.Again, no whisker was seen for the finish with the nickel underlayer. The findings of Fan indicate that a compressive stress aggravates while a tensile stress hinders the whisker growth, thus concurring with the theory of Lee and Lee. The use of a Ni underlayer generated tensile stress, thus depressing formation of whiskers. It should be pointed out that although Cu diffusion is believed to cause compressive stress and thus whisker growth, formation of intermetallic Cu6Sn5 may not contribute to whisker formation. Table 16.1 shows the volume of Cu, Sn, and intermetallic Cu6Sn5. The actual molar volume of Cu6Sn5 is smaller than the calculated molar volume if the volume is assumed to be additive. In other words, forming intermetallic Cu6Sn5 will result in a reduction in volume and thus presumably a tensile stress. Perhaps the compressive stress caused by the copper dissolved in solder overrides the tensile stress generated by Cu6Sn5 and consequently results in a net compressive stress. This net compressive stress is considered to be relievable by reflow with a reasonable cooling time. This allows the intermetallics to be formed more readily, and would explain a decrease in the tendency for whisker formation.64 Besides copper diffusion, increased carbon content may also cause whisker growth. Ohkawara and Muroi65 studied the whisker growth rate from zinc plating versus chemical species in baths. They observed that high concentrations of cyanide and alkali in baths lowered whisker growth in zinc plating, but the mechanism and suitable bath composition remain to be clarified. Their work concluded that the carbon content of zinc plating, and the amount of zinc cyanide complex and free cyanide ions in baths, are related to whisker growth. On the other hand, zincate ions are related to the electrodeposition rate but not to whisker growth. Ohkawara and Muroi66 continued the mechanism investigation work and studied the influence of internal stress and crystal structure on whisker growth from zinc plating. In a zinc cyanide system, internal stress, lattice distortion (strain), and the carbon content of plating are found to relate to one another. Figure 16.8 shows the effect of carbon content on lattice distortion and whisker growth. Figure 16.9 shows the effect of lattice distortion on internal stress and whisker growth.The value of these physical and met-
TABLE 16.1 Volume of Cu, Sn, and Intermetallic Cu6Sn5
Material
Atomic or formula weight (gm/mol)
Density (gm/cm3)
Molar volume (cm3/mol)
Notes
Cu
63.54
8.9
7.14
Sn
118.71
7.3
16.26
Measured Measured
Cu6Sn5
974.79
8.28
117.73
Measured
—
124.14
Calculated, assuming additive volume
16.8
CHAPTER SIXTEEN
FIGURE 16.8 Effect of carbon content on lattice distortion and whisker growth. M ratio represents NaCN/Zn of the zinc plating baths.66
FIGURE 16.9 Effect of lattice distortion on internal stress and whisker growth. M ratio represents NaCN/Zn of the zinc plating baths.66
CHALLENGES FOR LEAD-FREE SOLDERING
16.9
allurgical properties of plating decreased with an increase in the M ratio (NaCN/Zn) of the baths, and whiskers did not grow below a particular value (i.e., internal stress 55 MPa; lattice distortion (strain) 0.25 percent; carbon content 0.06 percent). It appears that a higher carbon content in the plated zinc layer results in a higher lattice distortion, which in turn results in a higher internal stress. With carbon serving as an impurity inclusion, this internal stress very likely is compressive in nature. The parallel relationship between zinc and tin systems regarding compressive stress versus whisker growth not only supports the compressive stress model, but also suggests that tin whisker might also be induced by high carbon content in the tin surface finishes. Tin whisker growth is also affected by the grain size of tin. Zhang of Lucent67 reported that a large-grained coating is favored for reducing whisker growth, since a large-grained coating exhibits fewer grain boundaries for copper to diffuse through. It also typically has zero or very low compressive stress to start with. When driving force (compressive stress) is present, the large grain also requires more energy to be squeezed out than fine grains do. The effect of aging conditions on whisker growth rate has been a controversial subject. The GSFC conducted a literature survey to determine optimal conditions for producing tin whiskers, and found that brass substrates with “bright” tin electroplate of approximately 200 µin were highly prone to whisker formation. In addition, storage at 50°C was also considered to be an accelerating factor.9 The GSFC’s results indicate that a higher density of whisker growth is observed on samples stored under room ambient conditions (∼22°C, 30 to 70 percent relative humidity) when compared to samples stored at 50°C. Furthermore, straight heat aging of the parts (without thermal cycling) at +90°C for 400 h generated no whiskers on parts from the same lot. Thermal cycling is very efficient in generating whisker. Kadesch and Leidecker9 observed that thermal cycling between −40 and +90°C for 331 cycles produced whisker, and that the whisker length increased when the number of cycles was increased to 500. Brusse of NASA68 also reported that 100-µm-long whiskers were seen on a Sn finish (6 µm thick) plated on a nickel underplate layer (6.5 µm thick) over a silver frit substrate after going through 100 thermal cycles of −40 to 90°C. More whiskers were found when the number of thermal cycles was increased. The same finish did not show whiskers after 90°C aging. The case just described indicates that the presence of a Ni underlayer does not guarantee freedom from whiskers. Motorola, Shipley, and FCI have also observed whiskers on Ni barrier samples in some of their experiments. It seems that if Ni is involved in the growth of whiskers, either the Ni plating process should be better understood or Ni may merely delay the onset of whisker growth. NASA has investigated the possibility of eliminating whisker growth with the use of conformal coating.9 Results indicate that conformal coating does not slow down the whisker growth rate. Instead, for the first year of the experiment, nodules formed more rapidly and in greater numbers under the conformally coated side than the nonconformally coated side. Later, the density of nodules was essentially equal on both sides. However, conformal coating does delay the dielectric breakdown, since the whisker would buckle before penetrating the coating on an adjacent surface. Whisker growth may be affected by introducing other elements into tin. Prasad et al.60 evaluated plating chemicals and the reliability of Pb-free leadframe packages. When compared with pure Sn and SnCu, SnBi systems exhibit minimum whiskers; however, no plating system is whisker-free. Whisker growth rate is found to be related to substrate materials, with brass Ⰷ C194/C151 > C7025. Alloy 42 leadframe
16.10
CHAPTER SIXTEEN
did not seem to experience whisker problems with Pb-free plating chemical systems within the time frame studied.
16.1.4
SURFACE FINISH CLEANING RESISTANCE
Although flux residue cleaning is known as a challenge in the Pb-free soldering process,69,70 cleaning unreflowed solder paste on PCB can also pose a challenge. For solder paste processes, the PCB often has to be cleaned if the printed paste misregisters, smears, or dries out. This is not a problem for any metallization, such as HASL or NiAu, since no solvent can remove any of those metal finishes. However, in the case of organic solderability preservatives, board cleaning can result in removal of organic surface finishes. For single-sided PCB, this is not an issue. However, for double-sided PCB, removal of surface finishes may cause wetting difficulty after one thermal excursion, whether the subsequent soldering is reflow, wave solder, or rework.
16.2
CHALLENGES FOR SOLDERING
Both reflow and wave soldering face many challenges. Examples include intermetallic compounds, dross, wave solder composition, lead contamination, fillet lifting, poor wetting, voiding, and rough joint appearance. These will be discussed in the following sections.
16.2.1
INTERMETALLIC COMPOUNDS
All of the common base materials form tin intermetallic compounds. The amount of intermetallic compound at an interface or within the solder joint is highly dependent on the base material. Other than tin itself, gold dissolves most rapidly into tin-based solders, due to the high solubility of gold in tin—approximately 15 wt% at 250°C.71 However, the solubility of gold in tin and lead in the solid state is very low and virtually all of the gold dissolved subsequently precipitates as AuSn4. Gold is well known as being in a class of its own in terms of its ability to embrittle joints. This is mainly due to the high dissolution rate of gold in solder. Another factor is the stoichiometry of the compounds formed. For example, 3 at% gold in the liquid phase gives rise to 15 at% AuSn4, while 3 at% silver or copper yields only 4 at% Ag3Sn and 5.5 at% Cu6Sn5, respectively.72 Silver also dissolves quite readily in solder. The solubility of silver at 250°C is about 6 wt% in pure tin73 and about 3.5 wt% in eutectic SnPb.74 Pd does not dissolve as rapidly as Au or Ag in Sn. However, similarly to Au, Pd forms intermetallic PdSn4, and thus a small amount of Pd can also rapidly generate large quantities of intermetallics due to the stoichiometry factor. This is particularly a concern for reflow soldering on a thick layer of Pd surface finish, where the solder volume in contact with the base’s metallization is very limited. Additional exposure to temperature cycling inevitably will aggravate the formation of intermetallic compounds. Figure 16.10 shows a secondary electron micrograph of the cross section of a Pb-free solder joint between a small-outline integrated circuit (SOIC) lead and a Pd-finished PCB after 2500 thermal cycles.75 The solder alloy used here is Sn2.5Ag0.8Cu0.5Sb (CASTIN). A large quantity of PdSn4 can be seen everywhere
CHALLENGES FOR LEAD-FREE SOLDERING
16.11
FIGURE 16.10 Secondary electron micrograph of cross section of a solder joint between a SOIC lead and a Pd-finished PCB after 2500 thermal cycles.75
within the solder joint. The presence of a large quantity of PdSn4 in solder joints after reflow, even prior to thermal cycling treatment, is responsible for weak joint strength in devices.75 Both Au and Pd are among the favorite Pb-free surface finishes. A thin layer of Au or Pd will not pose any concern in terms of intermetallic formation. However, since a relatively thick layer of Au or Pd may be needed for wire bondability applications, care should be taken in balancing the need for both wire bondability and solder joint reliability. Copper is the most commonly used base material. The solubility of copper in tin at 250°C is about 1.5 wt%.74 Although Cu dissolves in Sn faster than in 63Sn37Pb, the intermetallic compound formation rate in Sn or Pb-free highSn alloys often is about equal to or slower than that in eutectic SnPb, as shown in Table 16.2.72 The initial CuSn intermetallic formation is faster for Pb-free alloys doped with copper than for those without copper dopant. In all incidences, the intermetallic formation rate declines rapidly with increasing thickness of the intermetallic layer and thus does not pose process or reliability concerns.
16.2.2
DROSS
For SnAg, SnCu, and SnAgCu systems, dross formation is not a major concern.Table 16.3 compares the oxide thickness development of seven tin-based alloys at 140°C above the melting temperature of the alloys.75 Both SnCu and SnAg compare favor-
16.12
CHAPTER SIXTEEN
TABLE 16.2 Quantity of Intermetallic Formed at Copper Interfaces (µm) and Copper Dissolved (µm) During Soldering at Various Temperatures/Times with Three Different Alloys 5s
15 s
30 s
60 s
IMC
Dissolved
IMC
Dissolved
IMC
Dissolved
IMC
Dissolved
235°C
0.3
(5)
0.5
(6)
1.0
(9)
1.5
(11)
245°C
0.4
(5)
0.5
(6)
1.0
(9)
1.5
(12)
260°C
0.8
(5)
1.2
(6)
1.5
(9)
1.7
(10)
235°C
1.0
(2)
1.3
(3)
1.
(3)
2.0
(5)
245°C
1.0
(2)
1.0
(6)
1.8
(6)
2.5
(6)
260°C
1.0
(5)
1.0
(6)
2.1
(7)
2.5
(9)
235°C
0.5
(3)
1.0
(4)
1.4
(4)
1.8
(5)
260°C
0.5
(5)
1.5
(5)
1.5
(7)
1.8
(9)
95.5Sn3.5Ag
99.3Sn0.7Cu
63Sn37Pb
Figures in parentheses indicate the quantity of copper dissolved into the solder matrix.
ably with eutectic SnPb in terms of oxide formation rate. On the other hand, alloys containing Bi, Sb, Zn, or In all oxidize more rapidly than 63Sn37Pb in the molten state, suggesting a higher dross formation rate at wave soldering than that for eutectic SnPb. Indeed, Lotosky76 reported that SnAgBiCu alloy suffers a higher solder mass loss due to dross removal, as shown in Fig. 16.11. On the other hand, SnAg is comparable with eutectic SnPb while SnAgCu or SnCu are lower in solder mass loss rate than eutectic SnPb, consistent with the findings of Miric and Grusd.77 The high
TABLE 16.3 Oxide Thickness: Initial and After Oxidizing the Solder Preform in Air at 140°C Above the Melting Point of the Alloy Oxide thickness (Å)
Alloy
Oxidation temperature (°C)
Initial
After 10 min
Sn99.3Cu0.7
367
20
50
50
Sn oxide
Sn96.5Ag3.5
361
30
50
50
Sn oxide
Sn63Pb37
323
30
50
500
Sn oxide
Bi58Sn42
278
350
800
Sn oxide
Sn95Sb5
380
20
875
1425
Sn oxide
Sn91Zn9
339
70
200
325
Zn oxide
52In48Sn
257
20
175
600
In oxide
After 50 min
Dominant oxide type
Before oxidizing the preform in air, the initial oxides were removed by heating the preform in nitrogen to 500°C and then holding it for 10 min in a flow of hydrogen (hydrogen reduces oxides); afterward, the preform was cooled in nitrogen to a temperature 140°C above the solder’s melting point. Then a nitrogen flow was switched to air flow to start oxidation. Finally, the preform was cooled to room temperature in nitrogen and the oxide thickness was measured using auger electron spectroscopy.
CHALLENGES FOR LEAD-FREE SOLDERING
16.13
FIGURE 16.11 Loss of solder quantity due to dross removal at wave soldering.76 Grams per minute of mass removed during typical dross removal per minute of wave run time. Data based on minimum of four separate 2- to 4-h experiments.
dross formation rate associated with alloys containing Bi, Sb, Zn, and In can be addressed by inerting the wave zone, thus minimizing the cost of using those alloys.
16.2.3
WAVE SOLDER COMPOSITION
Unlike in reflow soldering, where the alloy composition of solder paste remains constant, wave solder composition is constantly changing due to the leaching process. Upon contact with the solder wave, the metallization of PCB dissolves into the molten solder.Where copper is the base material, copper in the molten solder is converted into solid intermetallic Cu6Sn5 once it reaches the limit of solubility.As shown in Table 16.4, this intermetallic compound can be removed easily by skimming the surface of the molten eutectic SnPb bath, due to the relatively lower density of
TABLE 16.4 Density of Metal Materials Related to Soldering Material
Density (gm/cm3)
Cu6Sn5
8.28
Cu3Sn
8.9
Ni3Sn4
8.65
63Sn37Pb
8.4
95.5Sn3.8Ag0.7Cu
7.5
96.5Ag3.5Ag
7.5
99.3Sn0.7Cu
7.3
Sn
7.3
Ni
8.9
Cu
8.9
16.14
CHAPTER SIXTEEN
Cu6Sn5 (8.28 gm/cm3) vs. eutectic SnPb (8.4 gm/cm3). As a result, the bath composition can be maintained easily by adding pure tin to compensate for the loss of Sn due to the removal of intermetallic Cu6Sn5. Unfortunately, this simple bath maintenance practice cannot be utilized when performing Pb-free soldering. As shown in Table 16.4, all of the major Pb-free solders, including SnAg, SnCu, and SnAgCu, exhibit a much lower density than the intermetallic Cu6Sn5. Consequently, the intermetallic Cu6Sn5 formed tends to precipitate. Although the bath composition is still maintained by addition of pure Sn, removal of the solid Cu6Sn5 from the bottom of the bath becomes a tedious task.
16.2.4
LEAD CONTAMINATION
Pb contamination is very likely, particularly at the early phase of transition to leadfree soldering. The presence of Pb may appear in the solder materials as impurity, in the surface finish of components or PCBs, or as solder deposits, such as solder balls on BGA. Table 16.5 shows examples of composition and impurity analysis data for some Pb-free solder alloys.78 For the 10 samples analyzed, the highest level of Pb impurity is 265 ppm. One of the primary sources of Pb contamination is tin. Table 16.6 shows impurity analysis data for two typical 99.9 percent Sn lots, with one lot exhibiting 150 ppm Pb impurity.78 Although solders with lower Pb impurity levels are achievable, the cost associated with the process is considered prohibitive. The presence of lead contamination often results in a drop in melting temperature. Bieler79 studied the effect of Pb contamination on the properties of eutectic SnAg. SnAg alloy doped with three levels of Pb was investigated, with composition shown in Table 16.7. The effect of Pb content on melting behavior was studied with differential scanning calorimetry (DSC). Addition of Pb introduces a small new melting peak with the onset of melting at 179°C, as shown in Fig. 16.12. An increase in Pb content not only increases the proportion of the low melting phase, but also shifts the high melting phase toward a lower temperature. The National Electronics Manufacturing Initiative (NEMI) also reported that 1 percent Pb contamination will lower solidus temperature by 40 to 50°C, as shown in Table 16.8.80 For Bi-containing alloys, the sensitivity of melting temperature toward Pb contamination increases with increasing Bi content. Toyoda81 studied the effect of Pb contamination on the melting behavior of a SnAgCuBi system and found that the impact of Pb contamination becomes most significant at Pb content greater than 0.5 percent. At Pb content higher than 2 percent, no additional drop in solidus can be discerned, as shown in Fig. 16.13. The impact of Pb contamination on Pb-free soldering is much more than reduction in melting temperature. While the solder wetting appears to be insensitive to Pb contamination, as reported by Vianco et al.,82 the mechanical strength and fatigue resistance of Pb-free solders turn out to be extremely sensitive to the presence of Pb. Baggio et al.83 investigated the effect of Pb contamination on the fracture strength of Sn3.5Ag3Bi alloys for reflow applications. Results indicate that at Pb content greater than 0.2 percent, the fracture strength drops drastically, with only 20 percent fracture strength remaining at 1 percent Pb contamination, as shown in Fig. 16.14. The failure mechanism induced by Pb contamination was first investigated by Mei et al.84 on a eutectic SnBi system. Figure 16.15(a) shows the as-solidified microstructure of 58Bi42Sn solder joints between a hot-air leveled 63Sn37Pb pad and an 80Sn20Pb-coated component lead. Figure 16.15(b) shows the microstructure after 400 cycles between −45 and +100°C over 16 days. The Pb dissolves into molten BiSn during the soldering process, resulting in the formation of a 52Bi30Pb18Sn ter-
TABLE 16.5 Composition and Impurity Analysis Data for Some Lead-Free Solder Alloys Alloy
Sn
Ag
Bi
Cd
Cu
Fe
In
Mg
Ni
Pb
Sb
Tl
16.15
58Bi42Sn
5946
41.80%
3 ppm
58.19%
1 ppm
5 ppm
10 ppm
5 ppm
0.3 ppm
15 ppm
50 ppm
40 ppm
58Bi42Sn
5939
42.80%
30 ppm
57.18%
2 ppm
5 ppm
15 ppm
2 ppm
0.0 ppm
10 ppm
75 ppm
30 ppm
93.5Sn3.5Ag3Bi
93.41%
3.50%
3.06%
3 ppm
5 ppm
30 ppm
10 ppm
0.2 ppm
8 ppm
150 ppm
75 ppm
1 ppm
90.5Sn7.5Bi2Ag
90.01%
2.13%
7.84%
2 ppm
5 ppm
30 ppm
5 ppm
0.0 ppm
40 ppm
50 ppm
75 ppm
1 ppm
91.8Sn3.4Ag4.8Bi
91.55%
3.47%
4.97%
2 ppm
3 ppm
20 ppm
5 ppm
0.0 ppm
10 ppm
50 ppm
40 ppm
1 ppm
93.6Sn4.7Ag1.7Cu
93.20%
5.07%
75 ppm
3 ppm
1.70%
20 ppm
50 ppm
N/A
5 ppm
60 ppm
40 ppm
1 ppm
95.5Sn3.8Ag0.7Cu
95.58%
3.78%
20 ppm
1 ppm
0.62%
30 ppm
5 ppm
N/A
50 ppm
40 ppm
75 ppm
1 ppm
99.3Sn0.7Cu
99.27%
0.5 ppm
30 ppm
1 ppm
0.7%
20 ppm
5 ppm
0.3 ppm
15 ppm
30 ppm
50 ppm
1 ppm
95.5Sn4.0Ag0.5Cu
95.45%
4.11%
30 ppm
3 ppm
0.4%
50 ppm
20 ppm
N/A
20 ppm
250 ppm
50 ppm
1 ppm
96.5Sn3.5Ag
96.28%
3.68%
55 ppm
5 ppm
12 ppm
24 ppm
10 ppm
N/A
N/A
265 ppm
25 ppm
N/A
16.16
CHAPTER SIXTEEN
TABLE 16.6 Impurity Analysis Data for Two Lots of 99.9 Percent Sn (ppm) Lot
Ag
Bi
Cd
Cu
Fe
In
Mg
Ni
Pb
Sb
Tl
A
2
50
2
1
20
5
0.3
1
150
50
1
B
N/A
30
1
10
50
20
0.3
50
40
75
1
TABLE 16.7 Composition (wt%) of Eutectic SnAg Solder and Three Ternary SnAgPb Alloys Alloys
Sn
Ag
Pb
Ternary alloy A
94.61
3.43
1.96
Ternary alloy B
91.91
3.33
4.76
Ternary alloy C
89.35
3.24
7.41
Eutectic alloy E
96.5
3.5
—
TABLE 16.8 Effect of 1 Percent Lead Contamination on the Solidus Temperature of Alloys
Alloy
Melting temperature (°C)
Solidus of alloy with 1% Pb contamination (°C)
99.3Sn0.7Cu
227
183
96.5Sn3.5Ag
221
179
58Bi42Sn
138
96
nary eutectic structure in the solidified solder joint. The solder joints became weak in mechanical strength when subjected to (1) thermal cycling at temperatures greater than 96°C, because of low melting ternary eutectic phase accelerated grain growth and phase agglomeration; or (2) long-time aging at 85°C, probably because of the eutectoid decomposition of the X phase in the ternary eutectic structure. Addition of small amounts of indium to 58Bi42Sn solder may eliminate the formation of the ternary eutectic phase, as indicated by DSC measurements. The deterioration mechanism of Pb-free solder joints by Pb contamination is also applicable to non-Bi-containing alloys. Seelig and Suraski85 reported that investigation of a field failure of SnAgCu solder joints showed that the failure is caused by intergranular separation driven by the lead in the solder. Figure 16.16 shows a distinct phase between the normal grains. Pb formed a ternary SnPbAg phase, presumably with a melting temperature of 179°C. This low-melting phase surrounds the Pb-free grains and exhibits poor adhesion to the Pb-free alloy, thus causing the grain separation. Besides confirming the applicability of a low-melting phase—ternary SnAgPb in this case—at intergranular space as failure mode, Seelig and Suraski also proposed a Pb concentration mechanism via the zone refining principle. They proposed that, upon cooling, the Pb gradually enriched at the last solidified spot due to the low-melting nature of the ternary eutectic SnPbAg alloy, as illustrated in Fig. 16.17. This enriched low-melting phase pocket serves as void, and very likely
FIGURE 16.12 DSC heating curves for three ternary SnAgPb alloys.79
FIGURE 16.13 Solidus line of SnAgCuBi + Pb.81
FIGURE 16.14 Effect of Pb contamination level on the fracture strength of Sn3.5Ag3Bi alloy.83
16.17
16.18
CHAPTER SIXTEEN
FIGURE 16.15 (a) As-solidified microstructure of 58Bi42Sn solder joints between a hot-air leveled 63Sn37Pb pad and an 80Sn20Pb-coated component lead; (b) the microstructure after 400 cycles between −45 and +100°C over 16 days.84
CHALLENGES FOR LEAD-FREE SOLDERING
16.19
FIGURE 16.16 The Pb-free materials comprise the lighter areas, with the darker SnPb area surrounding them (3500× micrograph).85
FIGURE 16.17 A lead-free solder joint may be contaminated by the SnPb surface finish of component leads. Pb will enrich and settle at the last area to cool—under the lead at the PCB interface.85
16.20
CHAPTER SIXTEEN
FIGURE 16.18 Micrograph of a lead-free solder joint fracture resulting from Pb contamination and displaying Pb pockets.85
may initiate joint failure. Figure 16.18 shows a micrograph of a lead-free solder joint fracture resulting from Pb contamination and displaying Pb pockets. Similar to the case for Bi-containing systems, the presence of Pb contamination also causes early failure in fatigue tests for SnAgCu systems, as shown in Table 16.9 for bulk solder testing.
TABLE 16.9 Effect of Pb Contamination on Low-Cycle Fatigue Testing ASTM E606 Performance of 95.5Sn4Ag0.5Cu Alloy in Bulk Solder Testing Sample 95.5Sn4Ag0.5Cu
16.2.5
Cycles to failure
Result
13,400
Pass
0.5% Pb contamination
6,320
Fail
1% Pb contamination
3,252
Fail
FILLET LIFTING
The National Center for Manufacturing Sciences (NCMS) has reported a phenomenon called fillet lifting associated with some Pb-free solder joints.86 Fillet lifting at wave soldering consists of solder fillet pulling away from the copper land on the board. The separation occurs mostly at the solder to intermetallics, with cracks stopping at the knee on the land side, as shown in Fig. 16.19. Although mostly observed at wave soldering, fillet lifting may also occur at reflow soldering. The direct driving force for fillet lifting is a mismatch in thermal coefficient of expansion (TCE) between solder and PCB, as illustrated in Fig. 16.20. Upon cooling, the solder shrinks more in the x-y direction, while the PCB shrinks faster in the z direction. As a result, a lifting force is generated in both the x-y plane and the z direction. However, this driving force is not sufficient to create fillet lifting, since virtually all solder joints experience a similar mismatch in TCE, but many alloys such
CHALLENGES FOR LEAD-FREE SOLDERING
16.21
FIGURE 16.19 Cross-sectional view of fillet lifting.86
as 63Sn37Pb do not have fillet lifting problems. Apparently, driving forces beyond a mere mismatch in TCE are required for fillet lifting to occur. In the NCMS report, fillet lifting occurred more frequently with pasty alloys, including Bi-containing alloys and Pb-contaminated high-tin alloys. Suganuma87 studied the mechanism of fillet lifting in lead-free soldering. In his analysis, solidification of a through-hole fillet propagates rapidly from the top surface to the inner region. Cu lead serves as a heat sink. The heat flow from the inside of the throughhole propagates through the Cu sleeve and land and keeps the narrow layer of solder facing the Cu land pad in a liquid state. Handwerker88 noted that for pasty materials, the liquid can no longer redistribute to accommodate the stresses when there is more than about 90 percent solid. Upon cooling, the maximum stress is generated at fillet tip due to mismatch in TCE. Since the solder at the interface with PCB cools more slowly than the rest of the fillet joint, hot tearing of the semiliquid solder then occurs at this interface and results in fillet lifting. Handwerker has done some in situ studies that have shown that fillet lifting occurs before the eutectic temperature of the system is reached. This is exactly the same as hot tearing during metal casting.88
FIGURE 16.20 The driving force for fillet lifting is a mismatch in TCE.
16.22
CHAPTER SIXTEEN
Suganuma’s work on Bi-bearing alloys pointed out two more possible factors contributing to fillet lifting. First is Bi enrichment. Suganuma reported that upon solidification, Bi is enriched into the liquid interface region between solder fillet and Cu land by the formation of tin-rich dendrites. The diffusion distance is only a few micrometers, and a substantial amount of Bi along the interface remains in a liquid state [see Fig. 16.21(a)]. Heat flow through the Cu land retards cooling of the interface region, as shown in Fig. 16.21(b), and the mismatch in TCE between solder and substrate results in fillet lifting. The second factor is the skeleton wicking effect. The dendrite formation not only promotes Bi segregation, but also provides a skeleton that may suck in residual liquid and aggravate the fillet lifting process. This skeleton wicking effect may also cause fillet surface cracking in other high-tin alloy systems. Harrison and Vincent89 studied SnAgCuSb and reported that 60 to 80 percent of dendrites formed during solidification, causing liquid wicking back and shrinkage cracking on the final surface to solidify, as shown in Fig. 16.22. Fillet lifting can also occur at reflow soldering. In this case, instead of partial fillet lifting, the whole solder joint is lifted, as demonstrated by the work of Nakatsuka et al.90 (See Fig. 16.23). By analyzing the solder joint microstructure of lifted or weakened joints, it was observed that, similar to the fillet lifting situation in wave solder-
FIGURE 16.21 Schematic of fillet lifting mechanism for Bi-bearing alloys. (a) Liquid Bi enrichment due to Sn-rich dendrite formation. (b) Mismatch in TCE and slow cooling at solder-pad interface result in hot tearing of liquid Bi phase and fillet lifting.87
CHALLENGES FOR LEAD-FREE SOLDERING
16.23
FIGURE 16.22 SEM of surface shrinkage cracking on fillet of SnAgCuSb joint caused by dendrite formation.89
ing, Bi enrichment also occurred at the interface between solder and Cu land, as shown in Fig. 16.24. In order to understand the driving force of Bi enrichment, a redistribution experiment was conducted on a SnAg32Bi alloy system. The high Bi content was expected to augment the potential Bi redistribution capability. First, the alloy was allowed to fill the gap between two copper plates. The two copper plates were then cooled down at different rates in order to generate temperature gradients. A band of redistributed Bi (probably Sn1Ag57Bi, ternary eutectic, melting point 137°C, the lowest-melting-point composition in the SnAgBiCu system) was noted on the inner higher-temperature side. The amount of redistribution decreases with decreasing temperature gradient, as shown in Fig. 16.25. The plates on the left side of Fig. 16.25(a) and (b) are higher in temperature, and the sample in (a) has a greater temperature gradient than the sample in (b). It was concluded by Nakatsuka et al. that the component joint lifting was caused by warping of PCB and a strength decrease resulting from the hardness and brittleness of a vicinity of eutectic composition with low melting point that was redistributed at the last cooled location within the solder. The Bi enrichment process can be minimized by applying a cooling process with minimal temperature gradient between components and PCB via equal heating from both the top and bottom.87 However, it should be noted that eutectic SnBi, although hard and brittle, has been successfully used in electronic assembly for many decades without problems. Therefore, it is the author’s opinion that the joint lifting may have actually occurred before the joint was fully solidified, and a hot tearing mechanism similar to that responsible for fillet lifting in wave soldering results in the component separation.
16.24
CHAPTER SIXTEEN
FIGURE 16.23 Solder joint lifting in the vicinity of SnAgBiCu solder/Cu pad interface.90
Fillet lifting was also observed on the top side only for SnPb finished components wave-soldered with 99.3Sn0.7Cu. Suganuma87 studied this phenomenon and observed that Pb enrichment occurred at the interface between solder and lead as well as between solder and copper pad, as shown in Fig. 16.26. The mechanism proposed by Suganuma postulates that SnCu solder touches the lead wire of the bottom of a PCB and flows up to the top side through the through-hole. The surface SnPb coating dissolved by the SnCu liquid flow is conveyed to the top side. SnCu with Pb exhibits a lower solidus temperature (see Fig. 16.27). Presumably the low-melting Pb enriched at the last cooling spot through a zone refining mechanism.85 In summary, fillet lifting can be minimized by the following approaches: 1. 2. 3. 4.
Avoiding Pb contamination Rapid cooling to depress dendrite growth Even cooling of both top side and bottom side Minimizing the pasty range of solders
CHALLENGES FOR LEAD-FREE SOLDERING
(a)
16.25
(b)
FIGURE 16.24 Cross-sectional view of weakened solder joints. Only (a) ruptured joints or (b) weak joints (6 N) showed Bi redistribution in the vicinity of the solder/Cu pad interface.90
16.2.6
FIGURE 16.25 Results of redistribution experiment with SnAg32Bi. TL, TH, and t are the temperatures of points L and H and the time, respectively. (a) A band of redistributed Bi was noted on the inner higher-temperature side. (b) The amount of redistribution decreases with decreasing temperature gradient.90
POOR WETTING
Perhaps poor wetting is the first drawback noticed by the industry when trying to implement lead-free soldering. This poor wetting can be caused by the component finish, the pad surface finish, or the solder itself. In the reflow case, the joint of TQFP64 with NiPd lead finish and Sn3.3Ag3Bi1.1Cu solder exhibits poor wetting at both toe and heel locations.91 The same component with SnPb component finish displays a very good wetting, with solder wicking up at both toe and heel locations. The poor wetting performance of Pb-free solders can be quantitatively reflected by the wetting time study of Toyoda,81 as shown in Fig. 16.28. Here the eutectic SnPb wets the best, followed by SnAgCuBi family and then by SnAgCu family, with eutectic SnCu exhibiting the poorest wetting. The poor wetting of Pb-free alloys may be an inherent characteristic, since the surface tension of high-tin alloys typically is higher than that of eutectic SnPb.
16.26
CHAPTER SIXTEEN
FIGURE 16.26 Cross-sectional view of solder joint with fillet lifting occurring on top side only for SnPb finished components wave-soldered with 99.3Sn0.7Cu. Pb enrichment is observed at the interface between solder and lead as well as between solder and copper pad.87
On the other hand, the wetability of eutectic SnPb HASL is insurmountable when compared with those of all lead-free finishes. This is attributed to the fact that only coalescence of molten solder is needed in order to wet to HASL surfaces. For all other surface finishes, such as organic solderability preservatives or NiAu finishes, metallurgical diffusion is required in order to achieve solder wetting.
16.2.7
VOIDING
Poor voiding is another shortcoming noticed by the industry when dealing with Pbfree soldering. Voiding is a phenomenon commonly associated with solder joints. Generally the voids are caused by the outgassing of flux entrapped in the sandwiched solder during reflow. This is especially true when reflowing a solder paste in surface-mount technology applications. The void content increases with decreasing solderability.92 With decreasing solderability, the substrate oxide can be cleaned less readily, thus allowing more opportunity for the flux to be entrapped to form voids. As discussed in the previous section, Pb-free soldering suffers from poor wetting, which inevitably results in poor voiding. Surface finish appears to be a more critical factor in affecting voiding than the solder alloy.76 During the transition stage of implementing Pb-free soldering, the use of Pb-free materials together with SnPb materials is very likely. Jessen93 studied the effect of
CHALLENGES FOR LEAD-FREE SOLDERING
FIGURE 16.27 SnCu solder touches the lead wire of the bottom of a PCB and flows up to the top side through the through-hole. The surface SnPb coating dissolved by the SnCu liquid flow is conveyed to the top side and redistributed at the interface.87
16.27
solder material on BGA voiding performance, with results shown in Fig. 16.29. His results indicate that void content in the BGA joints decreases in the following order: SnAgCu paste/SnPb solder ball > SnAgCu paste/SnAgCu solder ball > SnPb paste/SnPb solder ball. This relative voiding tendency can be explained by the model shown in Fig. 16.30. For the BGA attachment process, if the melting temperature of solder ball is higher than that of the solder paste, no flux fumes will be able to penetrate into the solder ball and form voids. However, if the ball exhibits a lower melting temperature than the paste, as shown in Fig. 16.30(b), voiding will be a big problem. As soon as the ball reaches melting temperature, a large quantity of the flux volatiles generated will enter the molten solder and form voids rigorously. This rigorous void-forming process will continue until the solder paste coalesces, which in turn will cause the flux to be expelled from the interior of the molten solder. The voiding action will then subside due to shortage in fresh supply of volatiles. It is interesting to note that this model also explains the voiding behavior involving Sn62 and Sn63. It has been noted that the BGA assembly with Sn62 ball and Sn63 paste yielded much more voiding than the system with Sn63 ball
FIGURE 16.28 Meniscograph of wetting time test results of solders.81
16.28
CHAPTER SIXTEEN
FIGURE 16.29 Voiding performance of CSP169 and CBGA256. TSSOP48 and R2512 are rated by insufficient solder volume instead of voiding.93
and Sn62 paste.94 Since Sn62 exhibits a solidus temperature of 179°C while Sn63 melts at 183°C, the inferior voiding performance of Sn62 ball/Sn63 paste becomes easily understandable. The model in Fig. 16.30 dictates that a mixed alloy system may be tolerable only if the ball does not have a lower melting point than the paste. Violating this rule will result in unacceptable voiding in the joints.
16.2.8
ROUGH JOINT APPEARANCE
The solder joints in the lead-free process typically exhibit a rough, grainy appearance. Figure 16.31 shows lead-free solder joints from a cellular phone using no-clean solder paste (with 95.5Sn3.8Ag0.7Cu, type 3, 89.3 percent). Compared with the typ-
(a)
(b)
FIGURE 16.30 Voiding mechanism of BGA assembly process with alloy A for solder ball and alloy B for solder paste. Voiding becomes significant if the melting point of alloy A is less than that of alloy B.
CHALLENGES FOR LEAD-FREE SOLDERING
16.29
FIGURE 16.31 Solder joints of cellular phone using no-clean solder paste (with 95.5Sn3.8Ag0.7Cu, type 3, 89.3 percent).
ical shiny, smooth eutectic SnPb solder joints, the joints’ appearance is fairly rugged. This is mainly attributed to the high tin composition of Pb-free alloys, which tend to develop dendrites easily, as shown in Fig. 16.22. The primary impact of a rough joint appearance is inspection. A new criterion has to be established in order to differentiate a good rough joint from a bad rough joint.
16.3
CHALLENGES FOR RELIABILITY
Although the dust gradually settles in terms of solder alloy selection, challenges for reliability still exist. This is true even for the prevailing Pb-free alloys. Examples include tin pest, intermetallic compound platelet, trace fracture, thermal damage, conductive anodic filament (CAF), and flux residue removal.
16.3.1
TIN PEST
Tin pest is growth of pestlike formations on the surface of tin, as demonstrated by Fig. 16.32. It is caused by transformation of β-tin to α-tin. Figure 16.33 shows SEM images of the microstructures of β-tin and α-tin. The loose structure of α-tin undoubtly poses a major threat to the reliability of solder joints. The tin pest phenomenon has occurred on solder 99.5Sn0.5Cu in as cast form, as reported by Karlya
16.30
CHAPTER SIXTEEN
FIGURE 16.32 Tin pest development progress with increasing aging on 99.5Sn0.5Cu.95
et al.95 Although no tin pest has been observed on solder joints in PCB thus far, the conditions for formation of tin pest should be assessed in order to assure the reliability of Pb-free solder joints.
16.3.2
INTERMETALLIC COMPOUND PLATELET
SnAgCu is the most attractive system, mainly due to its overall superior reliability and acceptable soldering property and cost. SnAg is also a favorite choice. Reliability of SnAg can range from poor to good, and is highly dependent on applications. Its soldering performance is marginally acceptable, as discussed in Chap. 13. Microstructure study of solder joints from both alloys shows presence of Ag3Sn and Cu6Sn5. However, large primary Ag3Sn precipitate was found to appear in Sn3.9Ag0.6Cu when the sample was cooled slowly, as shown in Fig. 16.34.96 This large, sharp Ag3Sn precipitate degrades the tensile property of this alloy, thus compromising the reliability of solder joints. Development of large Ag3Sn precipitate has also been observed in solder joints of 62Sn36Pb2Ag on Cu base.72 Here the Ag3Sn nucleated on the Cu6Sn5 layer and grew into large platelets.
FIGURE 16.33 SEM of grip end of 99.5Sn0.5Cu specimen aged at 255 K for 7 months.95
CHALLENGES FOR LEAD-FREE SOLDERING
16.31
FIGURE 16.34 SEM and electron probe microanalysis (EPMA) images of mildly cooled Sn3.9Ag0.6Cu. A large primary Ag3Sn precipitate is found in the Sn matrix.96
The large Ag3Sn platelets can also appear in Pb-free solder joints. Lee et al.97 studied the microstructural stability of flip chip solder bumps on Cu pads. Four alloys were investigated: eutectic SnAg, SnAgCu, SnCu, and SnPb. While a small or thin Cu6Sn5 phase is present in every alloy system, huge Ag3Sn platelets are found in eutectic SnAg and Sn3.8Ag0.7Cu systems, as shown in Fig. 16.35. For the Sn3.8Ag0.7Cu sample, the solder was etched away to expose the platelet structure. The presence of large Ag3Sn platelets can cause solder joints of an area array package to split and slide along the interface between solder and platelet, and consequently result in early failure. The formation of Ag3Sn platelets can be prevented by lowering the Ag content in solder. In Suganuma’s work, Sn3Ag0.5Cu is recommended as a reliable choice.
16.3.3
STIFF JOINT
Trace fracture has been observed for BGA with SnAg solder balls assembled with SnPb paste after 2877 −50 to 150°C cycles (see Fig. 16.3698). This is attributable to the
16.32
CHAPTER SIXTEEN
FIGURE 16.35 SEM micrographs showing different interfacial intermetallic formation as a function of alloy.97
FIGURE 16.36 Flat section of a trace failure on an NSMD test board pad at 2877 −50 to 150°C cycles. Part had SnAg solder balls and was assembled with SnPb paste.98
CHALLENGES FOR LEAD-FREE SOLDERING
16.33
high rigidity of SnAg solder ball. Upon thermal cycling, the stress caused by mismatch in TCE between BGA and PCB cannot be absorbed by the compliance of solder joints, and consequently results in trace fracture. For Pb-free soldering, this is a common issue, since many Pb-free alloys exhibit a high stiffness, such as SnAgBi, SnAgBiCu, SnAgCuSb, and SnSb systems. Although the trace fracture problem could be addressed with solder mask–defined pad design or a widened trace at junction point with pad, the loss in trace routine spacing will limit its acceptance in highdensity interconnect applications.
16.3.4
THERMAL DAMAGE
Pb-free alloys typically exhibit a higher melting temperature than eutectic SnPb. This consequent higher soldering temperature inevitably may induce some thermal damage to the components or boards. Figure 16.37 shows a cross section of 144LQFP package after level 2a/260°C, indicating a crack in the molding compound.99 The crack initiates from the leadframe paddle and propagates along a diagonal path through the molding compound toward the bottom of the package, although it does not extend to the external package boundary. Similar damage on plastic ball grid array (PBGA) caused by Pb-free soldering is exemplified by Fig. 16.38, where the cross-sectional view of the two-layer PBGA package after level 2a/260°C stressing indicates delamination within the die attach layer and internal substrate layers.99 Besides the mechanical cracks in components, the high-temperature soldering
FIGURE 16.37 Cross section of 144LQFP package after level 2a/260°C indicating crack in molding compound.99
16.34
CHAPTER SIXTEEN
FIGURE 16.38 Cross section of the two-layer PBGA package after level 2a/260°C stressing indicating delamination within the die attach layer and internal substrate layers.99
process can also cause damage to the board materials, as discussed in the following section.
16.3.5
FLUX RESIDUE CLEANING
Cleaning the flux residue of lead-free solder pastes is more challenging than cleaning that of SnPb systems.69,70 This is primarily due to (1) higher reflow temperature; (2) higher flux capacity, and therefore higher flux-induced side reactions; and (3) more tin salt formation. Bivins et al.69 studied lead-free flux residue cleanability. The results indicate that visual cleanability decreases from no-clean soft-residue fluxes to water-washable fluxes to no-clean hard residue fluxes. Improvement in visual cleanability may link to improvement in the thermal stability of fluxes. Semiaqueous and/or aqueous solvent sprayable cleaners yield the best visual cleaning efficiency, if the solvent is selected properly. Spray is the most critical mechanical agitation, regardless of whether it is spray in air or spray under immersion. Ultrasonic aid also imparted a significant positive effect. Saponified aqueous spray is one of the top choices for cleaning water-washable residues. Better solvency and spray are the directions for further improving cleaning. Improvement in flux thermal stability will help both soldering and cleaning. Another study on postsolder cleaning of lead-free solder paste residues used surface insulation resistance (SIR) and ionic contamination as criteria for cleanability.70 From these experiments it was concluded that: 1. Cleaning efficiency is highest for water-soluble paste and lowest for no-clean, halide-free, ultra-low-residue paste. Halide-containing, full residue no-clean paste is slightly better than low-residue paste.
CHALLENGES FOR LEAD-FREE SOLDERING
16.35
2. Reflow temperature does not significantly affect cleaning of flux residues from lead-free solders. 3. Cleaning chemistry has no effect on ionic contamination. For SIR performance, vapor degreasing cleaning chemistry shows the poorest performance. 4. Physical approaches, including cleaning time and ultrasonic agitation, have negligible effect on cleaning efficiency, while the chemical approach shows significant effect. The latter is demonstrated by the major improvement in efficiency when alkaline cleaner is added to water. This negligible effect of physical approaches reflects the greater difficulty in removing the residue of Pb-free solder pastes than eutectic Sn-Pb systems, presumably due to the higher reflow temperature used for Pb-free reflow processes. 5. Cleaning efficiency based on ionic contamination is virtually independent of cleaning efficiency based on SIR.
16.3.6
CONDUCTIVE ANODIC FILAMENT
Conductive anodic filament (CAF) formation is a failure mode for printed wiring boards in which a conductive filament forms along the epoxy-glass interface growing from anode to cathode. Figure 16.39 shows an example of CAF. The white region
FIGURE 16.39 Cross section of a printed wiring board showing CAF growing along the epoxy-glass interface.100
16.36
CHAPTER SIXTEEN
indicates a copper-containing filament growing along the epoxy-glass interface. Using backlighting, CAF appears as dark shadows coming from the copper anode to the cathode, as shown in Fig. 16.40. CAF is closely associated with reflow temperature. Turbini et al.100 studied CAF occurrence frequency and SIR value as a function of reflow temperature using 21 fluxes. The results indicate that a higher board-processing temperature results in increased numbers of CAFs for most of the fluxes tested, as shown in Table 16.10. The findings here pose a great concern for electrical reliability, since virtually all Pbfree soldering requires a high-temperature soldering process. Perhaps the problem can be resolved by using board materials with a higher thermal stability, although costs will be higher.
16.4
UNANSWERED CHALLENGES
Up to this point, many challenges associated with lead-free soldering have been discussed. Although with certain difficulty, in general those issues can be addressed one way or the other. However, there are some challenges for which the answers have
FIGURE 16.40 CAF appears as dark shadows coming from the copper anode to the cathode when a backlighting is applied.100
16.37
CHALLENGES FOR LEAD-FREE SOLDERING
TABLE 16.10 Comparison of SIR Levels and Number of CAFs Associated with Two Different Reflow Temperatures
Flux
SIR (Ω) 201°C reflow
SIR (Ω) 241°C reflow
Polyethylene glycol-600 (PEG)
10
None
379
PPG/HBr
>1010
>1010
1
423
High 109
High 109
1
406
PEPG 18/HCl
High 109
High 109
10
135
PEPG 18/HBr
1010
High 109
9
279
High 109
High 109
9
Polypropylene glycol 1200 (PPG)
Polyethylene propylene glycol 1800 (PEPG 18)
Polyethylene propylene glycol 2600 (PEPG 26)
10
10
#CAF at 201°C reflow
#CAF at 241°C reflow
None
91
PEPG 26/HCl
High 10
High 109
6
218
PEPG 26/HBr
1010
High 109
None
51
Glycerine (GLY)
>1010
High 109
None
56
GLY/HCl
>1010
High 109
None
583
GLY/HBr
>1010
High 109
3
104
Ocyl phenol ethoxylate (OPE)
9
9
Low 10
Low 10
None
83
OPE/HCl
Low 109
Low 109
14
62
OPE/HBr
>1010
High 109
2
599
Linear Aliphatic Polyether (LAP)
9
Low 10
Not tested
None
Not tested
LAP/HCl
Low 109
Low 109
15
203
LAP/HBr
Low 109
Low 109
None
272
still not been found. For instance, Pb-free alternatives to high-Pb solders, such as 97Pb3Sn or 90Pb10Sn, are still not available. The higher cost of implementing Pbfree soldering is another issue. This includes solder materials, component and board redesigns, and equipment upgrades. Toxicity and recycling are other unanswered challenges. Table 16.11 shows the environmental impact of lead-free solders compared to SnPb.101 Toxicity of Ag and recycling of Bi are examples for which the solution has not been identified yet. Overall, although quite some progress has been made in lead-free soldering, many issues still have to be addressed, and the battle is far from being over.
TABLE 16.11 Overview of the Environmental Impact of Lead-Free Solders Compared to SnPb TPI
16.38
Acute toxicity
Ecotoxicity
SnPb (SnPb37)
100%
Pb: Highly toxic; teratogenic; mutagenic ? cancerogenic ?
Pb: Accumulates; highly toxic to many organisms
SnAg (SnAg3.5)
29%
Ag: Argyria
Ag: Toxic to microorganisms but low bioavailability
SnAgCu (SnAg4Cu0.5)
32%
Ag: Argyria
SnCu (SnCu0.7)
14%
SnBi (SnBi58)
6%
SnAgBi (SnAg3.5Bi4.8) SnZn (SnZn9)
Metal products 100%
Manufacturing
Recycling
Disposal
Optimized process
SnPb solder retrieval at secondary Cu smelters
Pb leaching 40 ppm Pb in leachate
7%
High energy demand
Up to 10% Sn tolerated at precious metal refining