INTERFACIAL REACTIONS DURING SOLDERING OF Sn-Ag-Cu LEAD FREE SOLDERS ON IMMERSION SILVER AND ELECTROLESS NICKEL/ IMMERSION GOLD SURFACE FINISHES SITI RABIATULL AISHA BINTI IDRIS A thesis submitted in fulfilment of the requirements for the award of the degree of Master of Engineering (Mechanical) Faculty of Mechanical Engineering Universiti Teknologi Malaysia NOVEMBER 2008
228
Embed
INTERFACIAL REACTIONS DURING SOLDERING OF Sn-Ag-Cu ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
INTERFACIAL REACTIONS DURING SOLDERING OF Sn-Ag-Cu LEAD FREE
SOLDERS ON IMMERSION SILVER AND ELECTROLESS NICKEL/
IMMERSION GOLD SURFACE FINISHES
SITI RABIATULL AISHA BINTI IDRIS
A thesis submitted in fulfilment of the
requirements for the award of the degree of
Master of Engineering (Mechanical)
Faculty of Mechanical Engineering
Universiti Teknologi Malaysia
NOVEMBER 2008
iii
To my beloved parents, sisters and brothers,
for their endless love, support and tolerance.
iv
ACKNOWLEDGEMENTS
First and foremost, I would like to express my sincere appreciation to my
thesis supervisors, Associate Professor Dr. Ali Ourdjini and Dr. Astuty Amrin for the
encouragement, guidance and motivation. Throughout the years, they had given me
faith to up bring this project and to deliver it according to the expectations. Their
patience and advice has walked me through all the difficulties I have met.
My heartfelt thanks go to Dr Azmah Hanim who has provided the
unconditional support, guidance and advice throughout the project. I shall never
forget the endless encouragement and assistance she has provided. I also would like
to thank Intel Technology (Malaysia) for providing the resources and funding my
research.
My special thanks also go to all Materials Science Laboratory technicians
especially Mr. Jefri and Mr. Ayub for providing the hardware and technical support
needed to complete this work. It would not have been possible for me to complete
this project without their help. Also, I would like to express my deepest gratitude to
my family members especially mummy and papa, and friends especially Zulkhairry,
Ieyja and Lia for their care and encouragement which have kept me confident and
motivated. Last but not least, the successful completion of this project would have
been impossible without the contribution from the above individuals who have lent
their helping hands. Thus, I would like to express my gratitude to all the individuals
mentioned above for their support and continuous encouragements.
v
ABSTRACT
In the on-going trend towards miniaturization in the electronics packaging industry, the increasing popularity of ultra fine line technologies has brought into question the physical aspects of pad topography and metallization. As the solder joints shrink in size, the thickness of the pad metallization available can be very small, thus rendering close control of the soldering process and development of intermetallic compounds at the solder joint is essential. Interfacial reactions and the structure of intermetallics at the solder/substrate interface play an important role in solder joint reliability and the present study was undertaken to investigate these interfacial reactions in order to have a better understanding on the formation of reactions and their growth. In this study, interfacial reactions between Sn-4Ag-0.5Cu and Sn-3Ag-0.5Cu solders and immersion silver (ImAg) and electroless nickel/immersion gold (ENIG) surface finishes were investigated. Emphasis is made on the effect of solder size, subsequent ageing of solder joints on the interfacial microstructures. Several techniques of materials characterization including optical, image analysis, scanning electron microscopy and energy dispersive X-ray analysis were used to examine and quantify the intermetallics in terms of composition, thickness and morphology. It was found that after soldering on ImAg only scallop-type Cu6Sn5 layer was formed and that its thickness increases with decreasing solder size. Subsequent ageing produced a second layer of Cu3Sn that forms between the Cu substrate and Cu6Sn5 layer. Growth kinetics showed that the Cu3Sn layer grew at a faster rate than the Cu6Sn5 and that Kirkendall voids were also observed within this Cu3Sn indicating that Cu diffuses much faster in the Cu3Sn than Sn in the Cu6Sn5. When soldering on ENIG finish, the reaction layer was found to consist of only one layer of (Cu, Ni)6Sn5 in the larger solders, while in the smallest solder (200 µm) both (Ni, Cu)3Sn and (Cu, Ni)6Sn5 were formed. These results reconciled well with the current theory of a critical Cu concentration determining the type of intermetallic layer that would form. The Ag content in the solder also affected the nucleation and growth of Ag3Sn plates as well as Cu-Sn intermetallic. Higher Ag containing Sn-Ag-Cu solder promoted growth of Cu6Sn5 rods and large Ag3Sn plates. Subjecting the solder joint to isothermal ageing induced thickening and coarsening of the intermetallics as well as changed in their morphologies. The results showed that the thickness of intermetallics increases with increasing the duration of ageing for both solders investigated and for all solder sphere sizes.
vi
ABSTRAK
Dalam menuju ke arah pengecilan dalam industri pembungkusan elektronik,
peningkatan kepopularan teknololgi ultra halus telah menimbulkan tanda tanya terhadap aspek fizikal topografi pad dan perlogaman. Sebaik sahaja saiz sambungan pateri mengecut, ketebalan perlogaman pad yang ada boleh menjadi kecil, maka ini menjadikan pengawalan tertutup proses pematerian dan perkembangan sebatian antara logam pada sambungan pateri adalah penting. Tindakbalas antara muka dan struktur sebatian antara logam pada pateri/ antara muka substrat memainkan peranan penting dalam keboleharapan sambungan pateri dan dengan itu, kajian ini dijalankan untuk menyelidik tindakbalas antara muka ini untuk mendapatkan pemahaman yang lebih baik terhadap pembentukan tindakbalas dan penumbuhannya. Dalam kajian ini, tindakbalas antara muka antara pateri Sn-4Ag-0.5Cu dan Sn-3Ag-0.5Cu dan kemasan permukaan immersion silver (ImAg) dan electroless nickel/ immersion gold (ENIG) telah diselidik. Penekanan diberikan kepada kesan saiz pateri, diikuti dengan penuaan sambungan pateri ke atas mikrostruktur antara muka. Beberapa teknik pencirian bahan telah digunakan untuk memeriksa dan menjumlahkan sebatian antara logam yang berkaitan dengan komposisi, ketebalan dan morfologi iaitu kaedah optik, analisis imej, scanning electron microscopy dan energy dispersive X-ray analysis. Didapati bahawa selepas pematerian ke atas ImAg, hanya lapisan Cu6Sn5 jenis scallop dijumpai dan ketebalannya meningkat dengan peningkatan saiz pateri. Proses penuaan menghasilkan lapisan kedua iaitu Cu3Sn yang terbentuk di antara Cu di dalam substrat dan lapisan Cu6Sn5. Kinetik pertumbuhan menunjukkan lapisan Cu3Sn tumbuh dengan kadar yang lebih cepat berbanding Cu6Sn5 dan Kirkendall voids juga didapati terbentuk di dalam Cu3Sn menunjukkan Cu meresap lebih cepat di dalam Cu3Sn berbanding Sn di dalam Cu6Sn5. Apabila pematerian ke atas ENIG dilakukan lapisan tindakbalas didapati terdiri daripada hanya satu lapisan (Cu, Ni)6Sn5 di dalam pateri yang besar manakala di dalam pateri yang kecil (Ø200 µm) kedua-dua (Ni, Cu)3Sn4 dan (Cu, Ni)6Sn5 terbentuk. Keputusan ini bersesuaian dengan teori yang digunakan iaitu kepekatan Cu kritikal menentukan jenis lapisan sebatian antara logam yang akan terbentuk. Kandungan Ag di dalam pateri juga memberikan kesan ke atas penukleusan dan pertumbuhan kepingan Ag3Sn dan juga sebatian antara logam Cu-Sn. Pateri Sn-Ag-Cu yang mengandungi Ag yang tinggi menggalakkan pertumbuhan rod Cu6Sn5 dan kepingan Ag3Sn yang besar. Penuaan ke atas sambungan pateri menggalakkan peningkatan ketebalan dan pengasaran sebatian antara logam dan juga perubahan ke atas morfologinya. Keputusan menunjukkan ketebalan sebatian antara logam meningkat dengan peningkatan masa penuaan untuk kedua-dua jenis pateri dan kesemua saiz pateri logam yang dikaji.
5.6.1 Characterization of Specimens Cross Section 93
5.6.2 Characterization of Specimens Top Surface 94
6 RESULTS AND DISCUSSION 95
6.1 Introduction 95
6.2 Top Surface Metallurgy (TSM) Deposition 96
6.3 Identification of Intermetallics in Solder Joints 97
6.4 Composition and Surface Morphologies of IMC 100
6.4.1 Intermetallics between SAC and ImAg 100
6.4.1.1 Reflow Soldering 100
6.4.1.2 Isothermal Ageing 106
6.4.1.3 Formation of Kirkendall Voids 116
6.4.2 Intermetallics between SAC and ENIG 118
6.4.2.1 Reflow Soldering 118
6.4.2.2 Isothermal Ageing 133
6.5 Thickness of Intermetallic Compound 145
6.5.1 Effect of Solder Volume on IMC Thickness 145
6.5.2 Effect of Surface Finishes on IMC Thickness 149
6.5.3 Growth Kinetics of IMC on ImAg Finish 150
6.5.4 Effect of Ag Concentration on IMC Thickness 154
6.5.5 Effect of Ageing Duration on IMC Thickness 159
7 CONCLUSIONS AND FUTURE WORKS 161
7.1 Conclusion 161
7.2 Future Works 162
REFERENCES 164
APPENDIX 175
PUBLISHED PAPERS 191
xi
LIST OF TABLES
TABLE NO. TITLE PAGE
2.1 Levels of interconnection for general electronic system 10 2.2 The advantages and disadvantages of wire bonding
interconnection 13 2.3 The advantages and disadvantages of TAB over the wire
bonding technology 15 2.4 Comparison of Interconnection Implementation 19 3.1 Comparison between different surface finish 39 4.1 Summary of reflow profiling 46 4.2 Benefits and limitations for vary reflow method 48 4.3 Melting properties of some common solder alloys 51 4.4 Lead-Free Solders for CSP Applications 56 4.5 Properties of Hard and Soft Solder Alloys 60 4.6 Lead-free solders with liquidus (T1), solidus (T2) and
eutectic (Te) 61 4.7 Solderability of different base metal 72 4.8 Potential IMC formation and un-compatibility between
solder and common substrates 73 5.1 The Swan and Gostin bath 87 5.2 Immersion silver bath formulation 89 5.3 Chemical composition of Klemm Solution II 93
xii
5.4 Etching time for cross sectional deep etching 94 6.1 Atomic number of elements 99 6.2 Atomic percentage of predicted IMCs 99 6.3 Compositions of the interfacial reaction products after reflow
soldering and ageing for 2000 hours at 150 oC 134 6.4 Intermetallic Thickness (µm) on ImAg surface finish 146 6.5 Intermetallic Thickness (µm) on ENIG surface finish 146 6.6 Calculation of the growth rate coefficient (D) for
SAC405/ ImAg 152 6.7 Calculation of the growth rate coefficient (D) for
SAC305/ ImAg 152
xiii
LIST OF FIGURES
FIGURE NO. TITLE PAGE
1.1 Schematic showing the main parts of an electronic package 2 2.1 Electronic packaging hierarchy 9 2.2 The wire bonding assembly shows how a bare chip is
interconnected to a substrate or another chip using a wire conductor 13
2.3 Schematic of TAB 14 2.4 Example of TAB devices 15 2.5 (a) Standard flip chip array with solder bumps.
(b) Cross-section of flip chip bonding 17 2.6 Solder Bump Structure 22 3.1 Dissolution rates of a few typical base metals in tin 27 3.2 Schematic diagram of the HASL technique 31 4.1 Typical solder reflow profile for eutectic Pb-Sn solder 45 4.2 The main heating options in reflow soldering 47 4.3 Typical wave soldering machine 49 4.4 The principle of hand soldering 50 4.5 Phase Diagram of Pb-Sn Alloy 54 4.6 The wetting angle 71
xiv
4.7 Cross section through a soldered joint, made with eutectic solder 74
4.8 Types of intermetallics formed between Cu and Sn 74 4.9 Needle-like Cu6Sn5 intermetallics 74 4.10 Schematic diagrams of the layers before and after isothermal
ageing 79
5.1 (a) Plan view and (b) Side view of copper substrate 81
5.2 Process flowchart of reflow soldering and specimen analysis 82 5.3 (a) Experimental set-up in the plating bath, (b) The plating
bath and (c) Schematic set-up of electroless nickel and immersion gold plating process 85
5.4 Schematic set-up of immersion silver plating 85 5.5 Commercial medium phosphorous concentration
electroless nickel plating solution: NIMUDEN 5X 86 5.6 Schematic process of electroless nickel plating 86 5.7 Schematic process of immersion gold plating on nickel 87 5.8 Immersion silver plating steps 88 5.9 Schematic process of immersion silver plating 89 5.10 Process flowchart for Immersion Silver 89 5.11 Solder joint formations for ENIG surface finish 91 5.12 Reflow profile fro Sn-Ag-Cu 92 5.13 IMCs formed from the top surface view 94 6.1 Copper substrate before plating (after pretreatment process) 96 6.2 Copper substrate plated with silver coating 97 6.3 FESEM-EDX results of IAg on Cu 97 6.4 Example of weight percentage calculation 98 6.5 Cross-sectional optical images after reflow: a) SAC405/ ImAg,
b) SAC305/ImAg 101
xv
6.6 Cross-sectional SEM images after reflow: a) SAC405/ImAg, b) SAC305/ImAg 101
6.7 Top view micrographs formed during reflow between SAC405
solder and ImAg. (a) 200µm, (b) 300µm, (c) 500µm and (d) 700µm 102
6.8 Top view micrographs formed during reflow between SAC305 solder and ImAg. (a) 200µm, (b) 300µm, (c) 500µm and (d) 700µm 103
6.9 Top view SEM images showing formation of large Ag3Sn
plates and Cu6Sn5 rods in SAC405/ ImAg (a, b) and Cu6Sn5 rods on SAC305/ ImAg (c) 104
6.10 Formation of Ag3Sn during reflow between SAC405
solder and ImAg:(a, b, c) Top surface morphology of the solder joint and (d) Cross section (x500) 105
6.11 Optical micrographs of cross-sectional views of SAC405/ ImAg (a-c) and SAC305/ImAg (d-f). (a, d): after reflow and (b, e): after ageing at 150oC for 250 hours and (c,f) after ageing at 150oC for 2000 hours 107 6.12 SEM images of cross-sectional views showing the effect of
ageing on the interfacial morphology. (a) SAC405/ImAg and (b) SAC305/ ImAg 109
6.13 Morphology of Cu6Sn5 on ImAg for 200µm solder bump of
6.17 Schematic of Ag3Sn particles embedded during intermetallic
growth 113 6.18 Ag3Sn on ImAg using 700µm solder (a) After reflow and
(b) After ageing for 500 hours 115
xvi
6.19 Schematic diagram of IMCs growth in Cu/Au specimen: (a) dissolution of Ag layer into molten solder, (b) formation of Cu6Sn5 during reflow soldering and (c) Conversion of Cu3Sn and Ag3Sn after isothermal ageing 116
6.20 SEM image of cross-sectional view of 500µm SAC405
solder/ ImAg after ageing for 500 hours 117 6.21 SEM image of cross-sectional view of 700µm SAC405
solder/ ImAg after ageing for 1000 hours 117 6.22 The mechanism of Kirkendall Voids formation 118 6.23 Cross-section views of the intermetallics formed between
ENIG and SAC405 (a-c) and SAC305 (d-f) solders (X500) 119 6.24 Cross section and top views of (Cu, Ni)6Sn5 IMC formed
during reflow between ENIG and SAC405 solder. (a, d) 300 µm, (b, e) 500 µm and (c, f) 700 µm 122
6.25 Cross section and top views of (Cu, Ni)6Sn5 IMC formed
during reflow between ENIG and SAC305 solder. (a, d) 300 µm, (b, e) 500 µm and (c, f) 700 µm 123
6.26 Top view of intermetallics formed between ENIG and
200 µm SAC405 (top) and SAC305 (bottom) solders 126 6.27 SEM images of cross sections of intermetallic formed between
ENIG and SAC405 for (a) 500 µm and (b) 700 µm solders 127 6.28 SEM images of cross sections of intermetallic formed between
ENIG and SAC305 for (a) 300 µm and (b) 500 µm solders 128 6.29a EDX results of interface intermetallic formed between ENIG
and 500 µm SAC405 solder during reflow 129 6.29b EDX results of interface intermetallic formed between ENIG
and 700 µm SAC405 solder during reflow 130 6.30a EDX results of interface intermetallic formed between ENIG
and 300 µm SAC305 solder during reflow 131 6.30b EDX results of interface intermetallic formed between ENIG
and 500 µm SAC305 solder during reflow 132 6.31 Cross sections of IMCs formed between ENIG and SAC405
solder. (a) reflow (500 µm) and (b) after 2000 hrs ageing (500 µm),(c) reflow (700 µm) and (d) after 2000 hrs ageing (700 µm) 134
xvii
6.32 SAC305 (a) after reflow and (b) ageing (2000 hrs) 135 6.33 Effect of ageing time on intermetallics formed between ENIG
and SAC405 solder with different solder sizes. 300 µm solder: a: reflow, b: ageing for 250 hours, g: ageing for 500 hours and h: ageing for 2000 hours. 500 µm solder: c: reflow, d: ageing for 250 hours, i: ageing for 500 hours and j: ageing for 2000 hours. 700 µm solder: e: reflow, f: ageing for 250 hours, 136
6.34 Top view of intermetallics formed between ENIG and 200 µm
SAC405: a: reflow, b: ageing for 250 hours, and c: ageing for 1000 hours 138
6.35 Formation of Ag3Sn in the 700 microns solder bump after reflow
soldering (SAC405). (a, b) Top surface morphology of the solder joint and (c, d) Cross section 141
6.36 SEM image showing different morphology of intermetallics
between center and periphery of solder joint 142 6.37 Different morphologies of IMC which form circular boundary
regions in Sn-Ag-Cu solder joint, (a) Sn-3Ag-0.5Cu solder and (b) Sn-4Ag-0.5Cu solder 143
6.38 Effect of Ageing on the morphology of Ag3Sn intermetallic.
(a) After reflow soldering and (b) After 500 hours ageing 144 6.39 Intermetallic thickness versus solder bump size for ImAg
surface finish as function of ageing time. (a) Sn-4Ag-0.5Cu and (b) Sn-3Ag-0.5Cu 147
6.40 Intermetallic thickness versus solder bump size for ENIG surface
finish as function of ageing time. (a) SAC405 and (b) SAC305 148 6.41 Intermetallic thickness versus ageing time between SAC405
and ImAg surface finish: (top) Cu3Sn and (bottom) Cu6Sn5 layer 153
6.42 Intermetallic thickness versus ageing time between SAC305
and ImAg surface finish: (top) Cu3Sn and (bottom) Cu6Sn5 layer 154
6.43 SEM top views of Ag3Sn intermetallic for SAC405/ ImAg
(a, b) and SAC305/ ImAg (c) 157 6.44 Ag3Sn IMC formation in ImAg surface finish 159
The Ni/Pd finish has also received considerable interest as a surface finish
for the Au features on printed circuit board. This surface finish system is being
35
targeted to replace the traditional HASL coating because the former systems leave a
very flat layer over Cu that will not interfere with the placement and integrity of
fine-pitch and ultra-fine-pitch, peripheral leaded packages and chip components.
Ni/P/Au surface finish process is similar to the ENIG process, except that a
Pd layer is deposited after the Ni layer and before the Au layer. The Pd layer can be
deposited by an electroless (autocatalytic) process so that unlimited thicknesses are
theoretically possible. Typically, the deposit also contains a 6-7 percent cent of
phosphorous (P), the source of which is the reducing agent required during the
plating process. The thickness can range from as little as 0.15 µm to as high as 1.5
µm, but the typical value is between 0.25-0.51 µm. Within this thickness range,
circuit board features show excellent solderability. Since Pd is harder than Au, the
strength of the surface finish is increased when the Pd layer is added oxidation
resistance of the Ni layer is improved.
Palladium coatings alone do not appear to offer the level of solderability
protection required for circuit boards that are exposed to multiple reflow cycles
(two-sided and/or mixed technology). Potential degradation mechanisms include Cu
diffusion from the underlying conductor and arriving at the surface of the Pd layer to
be subsequently oxidized, resulting in the degradation to solderability. As a second
mechanism, Pd can develop an oxide layer that potentially inhibits wetting and
spreading by the molten solder. In answer to this second scenario, a thin layer of
immersion-deposited Au (0.05-0.2 µm) is deposited over the Pd to maintain
adequate solderability when storage and processing conditions are too severe for the
exposed Pd surface. This is the basis of the Ni/Pd/Au finish (Vianco, 1998).
36
3.3.5 Immersion Silver (ImAg)
Immersion silver is deposited directly on the copper surface by a chemical
displacement reaction in which silver ions are exchanged for copper ions and
deposited on the exposed surface. The deposits provide a thin silver layer of 0.1 to
0.4 µm, which may be deposited with an organic material to reduce tarnish of the
immersion silver. This helps seal the surface and allow for extended shelf life.
Silver offers a flat and extremely solderable surface compared to HASL finish, Pb-
free inspection at assembly, lack of solder mask attack and surface contact
functionality (Cullen and Milad, 2004). The surface is also bondable for both
aluminum and gold wire (Gilleo, 2002; Clyde, 2001 and Milad, 2007). Unlike
ENIG, immersion silver is also a low temperature process. During soldering, the
molten solder wets and spreads on the surface of silver coating. The silver then
dissolves into the molten solder allowing the formation of a copper-tin intermetallic
solder joint, similar to HASL and OSPs. The silver from the coating forms silver-tin
(Ag3Sn) intermetallics in the solder.
One of the disadvantages with immersion silver has always been silver
migration in electronic environments (Vianco, 1998). This is due to the property of
silver to form water-soluble salts when exposed to moisture and electrical bias. The
incorporation of organics into the immersion silver minimizes this phenomenon.
Aside from that, occasional voiding in the solder joint was also reported. Studies are
being done to determine whether this problem comes from the excessive silver
thickness. If that is so, then an upper thickness limitation has to be set (Clyde, 2001
and Milad, 2007). The immersion silver also is an active surface and readily
combines with sulfur from the environment. Silver sulfide tarnishes the surface and
creates doubt about the integrity of the finish at inspection. Proper packaging of
immersion silver finished boards are critical to control sulfurization. The key in
packaging is to minimize contact of the surface with the environment and to ensure
all materials used in packaging and during storage are sulfur free (Milad, 2007).
37
Immersion plating or galvanic displacement, are terms that are usually
restricted to processes that are based on chemical replacement of the base metal by
the more noble metal coating. In this case the reaction will, in principle, either
completely cease or else proceed at an immeasurably slow rate once the base metal
surface is effectively masked by a thin coating. Solutions in this particular category
are only capable of yielding deposits of limited thickness as compared to electroless
plating. Although the theory of immersion plating is quite simple and the general
effects of all operating parameters are well known, at least quantitatively, the narrow
range of conditions under which useful coatings may be produced requires that any
new possibility must be tested by trial and error to determine how critical each of the
above factors are, since the end effect is mainly decorative, there has been little
stimulus to develop new processes for mass production.
The primary advantage of employing the immersion method of metal
deposition is that the bath can avoid the use of chemical reducers because the
chemical reduction of the metal from solution is driven by the oxidation of the
underlying metal. Once the metal from solution deposits to such a thickness that the
surface metal cannot be efficiently dissolved, the reaction slows. Therefore, the
immersion metal coatings are very thin. Silver is generally deposited to a maximum
thickness of about 0.5µm (Cullen, 2003). Immersion silver is deposited from a
dilute solution containing silver salts, metal complexors and organic surface
modifiers. The electromotive potential between silver and copper is 0.456 volts, so
the reaction is straightforward.
2Ag+ + Cu0 → 2Ag0 + Cu++ (3.9)
Organic compounds are added to the formula to inhibit tarnish and to prevent
electromigration. In some cases, an organic may be added to prevent the
precipitation of silver from the chemical bath. Precipitation would ordinarily occur
due to the interaction of silver ions and light.
38
3.3.6 Immersion Tin (ImSn)
The immersion tin process also uses the displacement reaction that
exchanges tin ions for copper ions to directly deposit a dense layer on the exposed
copper surface. It is one of a logical replacement for HASL for two reasons; first, it
is flat and co-planar, and second, it is lead-free. However, tin readily forms a
copper-tin intermetallic, namely Cu3Sn and Cu6Sn5, whose growth may affect the
soldering performance. A thicker Cu-Sn intermetallic is known to occur when
soldering with lead-free alloys on immersion tin finish. Thus, the tin thickness is
directly related to this intermetallic formation. A thick deposit of 1.0 µm (Clyde,
2001) or in a range between 0.76 to 1.27µm (30 to 50µinch) (Milad, 2007) is
feasible, thus ensuring a copper-free tin surface.
Immersion tin provides highly solderable surface and a dense uniform
coating with superior hole-wall lubricity. This characteristic makes it the choice for
backplane panels that are assembled by pin insertion. Immersion tin is a viable lead-
free finish option for some applications, but, how this finish will survive high
temperature assembly associated with lead free Sn-Ag-Cu solder alloy remains to be
seen. The solder joint intermetallic should not be a problem; however, the higher
temperature profile could accelerate the intermetallic formation compromising the
solderability of the surface (Clyde, 2001 and Milad, 2007).
Immersion tin also offers other disadvantages. The bath makeup entails the
use of thiourea, which is banned in certain geographical locations for environmental
reasons. During processing, the primary byproduct in the bath is copper thiourea.
Waste treatment allowance must be made for the containment of the thiourea and its
copper salt by-product. The shelf life of the surface is, to some extent, limited (less
than a year). This is due to the progression of the copper-tin intermetallic until it
reaches the surface and renders the product non-solderable. This could be
accelerated under excessive temperature and humidity conditions (Clyde, 2001 and
Milad, 2007).
39
Another issue with immersion tin is its propensity to form whiskers at room
temperature. Immersion tin whiskers do not grow as a result of exposure to heat,
vacuum, pressure, humidity or bias voltage. They grow naturally over time, which
would seem to indicate, that the primary source is Cu6Sn5 migration stress (Milad,
2007).
3.3.7 Summary
A good surface finish, however, is not just one that provides a flat and
solderable surface. It must also provide: (i) compatibility with other metals such
nickel, gold, and tin, (ii) strong consistent solder joint strength, (iii) consistent
solderability and (iv) long term electrical reliability (Stafstrom, 2000). Some of the
characteristics of the surface finishes are shown in Table 3.1 (Barbeta, 2004).
Table 3.1: Comparison between different surface finish
Surface finish Advantages Disadvantages
HASL Nothing solders like solder. Easily applied. Lots of industry experience. Easily rework. Good bong strength. Withstands multiple thermal cycles.
Huge co-planarity differences. Contain lead. Not suitable for high aspect ratios. Not suited for <0.5mm pitch. PWB dimensional stability issues. Bridging problems on fine pitch assemblies. Inconsistent coating thickness.
OSP Flat, coplanar pads. Reworkable. Doesn’t affect final whole size. Short, easy process. Cu-Sn intermetallic formed has been reported to be stronger and more robust than Ni-Sn intermetallic from Ni-Au.
Require changes in the assembly line. Question remains over the reliability of exposed Copper after assembly. Limited thermal cycles. Cannot be reworked by the assembler. Limited shelf life. Test pins cut coating, leaving exposed copper. Limited in circuit testability. Not inspectable at assembly.
ENIG Planar surface. Consistent thickness. Withstands multiple thermal cycles. Long shelf life. Good for line pitch product.
Not Au wire bondable. Expensive. Should not be used on < 1mm pitch; black pad issues. Waste treatment of nickel. Cannot be reworked at PWB fabricator. Nickel is a suspected carcinogen. Not optimal for higher speed signals.
40
Table 3.1 (continue): Comparison between difference surface finish
Surface finish Advantages Disadvantages
ENEPIG Pd keeps Ni from passivating in presence of “porous” gold coating. Al and Au wire bondable. Planar surface. Good for fine pitch product.
Additional processing step. Adds cost. Dip tank process. Evidence that Pd poisons the solder paste after reflow. Waste treatment. Very complex processing steps.
Immersion
Silver
Good for fine pitch product. Planar surface. No black pad concerns. Short, easy process cycle. Eliminates nickel. Doesn’t affect final whole size. Long shelf life. Can be reworked/ reapplied by the fabricator. Inexpensive. Drop-in process for the assembler. Good for ultra-high speed signals.
Friction coefficient; may not be suited for compliant pin insertion. Some systems cannot throw into blind vias with aspect ratios > 1:1. Tarnishing must be controlled.
Immersion Tin Good for fine pitch product. Planar surface. Eliminates nickel. Can substitute for reflowed solder in selective strip. Inexpensive.
Form whiskers at room temperature. Short shelf life. Problem with soldermask compatibility.
CHAPTER 4
SOLDERING
4.1 Introduction
Soldering is a well known and widely used process where two or more metal
items are joined together using a fusible alloy with a melting temperature that is
lower than their own. The most commonly used solder is a fusible alloy consisting
essentially of a tin and lead mixture. It is the solvent action (the solder actually
dissolves a small amount of the metals surface, at a temperature that’s well below its
melting point and joins with it) of the solder alloy that causes it to fuse with and
attach to the surface of the metal items being joined. The solvent action that takes
place, between the solder and the metal, makes the joint chemical (not just physical)
in nature and causes the properties of the joint to differ from the original solders
properties and from those of the surface of the metal items being joined. When metal
parts are joined by solder, a metallic continuity is established as a result of the
interfaces where the solder is bonded to the metallic surfaces.
42
The metal joining process that is generally referred to as soldering (or soft
soldering) requires temperatures between 183oC and 450oC. The joining of metals at
temperatures above 450oC (and below the melting point of the metals being joined) is
more commonly referred to as brazing (or hard soldering). The actual melting and
fusing of the metal items that are being joined together is considered welding. There
are, of coarse overlapping situations that may occur when classifying a process. The
actual joining characteristics that take place are physically different in each of these
processes. Soft solders attach to metals by what is referred to as a solvent action that
takes place at relatively low temperatures. Hard solders or brazing alloys contain
metals that require higher temperatures to cause the solvent action to take place and
fuse the alloy with the metal being joined. Because welding involves melting and
fusing the surface of the metals that are being joined together, a filler or fusible
material is not always used.
Soldering is used primarily when the expected operating temperature of a
joint will not exceed around 149oC and thermal or electrical continuity can not be
adequately achieved, or maintained, by the use of a mechanical joint. It is one of the
most ideal methods available for the creation of a physical, electrical, or hermetically
sealed bond between various metal items that are being joined together. Soldering is
quite often used, in addition to other mechanical methods (twisting, crimping, etc.) to
improve electrical continuity, to help protect the joint from the effects of vibration, or
to encapsulate the joined metals preventing oxidation. Although soldering may be
used to provide some minor support to an assembly, the solder should not (excluding
sheet metal applications) be used as the primary mechanical support of a finished
joint.
43
4.2 Materials
The soldering process may be accomplished in a wide variety of ways, but
the four primary ingredients required will remain the same. They are;
1. The base metal (or metal items being joined)
The base metal is the metal that is in contact with the solder and forms an
intermediate alloy. There are many metals that will react willingly with
solders to form a strong chemical and physical bond, while others can be very
difficult, or even impossible to solder.
2. A type of flux (or a method of cleaning and maintaining the surface to be
soldered)
Flux is used to eliminate minor surface oxidation and to prevent further
oxidation of the base metals surface during the heating process. Although
there are many types of flux, each will include two basic parts, chemicals and
solvents. The chemical includes the active portion, while the solvent is
actually the carrying agent. It is the solvent that determines the cleaning
method required to remove the remaining residue after soldering.
3. The solder
Solder is the alloy used to create the solvent action, which generates the bond
between the base metals. The type and form of the solder is very important
and must be determined by the individual application being performed, as
well as the base metals and soldering method being employed.
4. Heat
When an alloy is heated it typically goes thorough multiple phases. It goes
from a solid state to what is known as a pasty stage, sort of halfway between
a liquid and a solid, and then to a liquid state. In soldering it is difficult to
work with a substance that goes through a pasty stage. Eutectic solder is
often used for this reason. A eutectic alloy is one that goes directly from a
solid state to a liquid state without a pasty stage. The eutectic tin-lead alloy is
44
made up of 63% tin and 37% lead. Eutectic tin-lead solder can be applied as
a liquid just above the melting point, and then as it cools it will transform
directly into a solid. This makes it possible to form solid solder joints very
quickly. Sometimes a 60% tin and 40% lead alloy is used. This alloy
exhibits a nearly eutectic change from solid state to a liquid state and can be
produced at a lower cost (Barker, 1993).
It is important to match the soldering method and the equipment that will be used, to
the soldering application that is being considered.
4.3 Soldering Techniques
There are three main methods used in soldering process: The difference is the
sequence in which solder, flux and heat are brought to the joint, and in the way in
which the soldering heat is brought to the joint or joints:
1. Reflow Soldering
2. Wave Soldering
3. Hand Soldering
4.3.1 Reflow Soldering
Reflow soldering is a metallurgical joining method and is a much older
process than wave soldering, going far back into antiquity; under the name of ‘sweat
soldering’ it is used in plumbing to this day. With the advent of hybrid technology
45
more than thirty years ago, sweat soldering was recognized as the logical way of
joining surface mount devices (SMD’s), which were specifically developed for
hybrids, to the metallic conductor pattern of the ceramic substrate.
The soldering process involves four basic elements: the base metal (in our
case the substrate surface finish metals), soldering flux, solder alloy and heat
(temperature). To begin with, solder and flux are placed on one or both joint
surfaces, either together in the form of a solder paste or separately, first the solder in
the form of a metallic coating and then the flux at a later stage. Subsequently the
joints are put together. The important point is that all this happens at room
temperature though with some procedures, the solder may have been pre-deposited
on one or both joint surfaces by a hot-tinning method. With all reflow strategies, the
assembled joints are finally heated to a temperature high enough to melt the solder,
and for long enough to let it tin the joint surfaces and fill all the joint gaps. Then,
heating is discontinued and the solder is allowed to solidify, the faster the better
(Strauss, 1994). Figure 4.1 and Table 4.1 shows the sequence steps and summary of
Figure 6.17: Schematic of Ag3Sn particles embedded during intermetallic growth
(Huang et al. 2006)
114
From Figure 6.13 to Figure 6.16 it is also very clear that as the ageing time is
increased the intermetallic layer has coarsened for a given solder bump size. This
indicates that the Cu6Sn5 intermetallic layer has indeed increased in thickness. Also
it can be seen that the grain size for smaller solder bump size is bigger compared to
the larger solder bump size for both SAC405 and SAC305 solders particularly after
reflow and ageing for shorter time (250 hours) indicating that thicker intermetallics
grow in smaller solders. For example the measured grain size for SAC405 solder
size of 200 µm and 700 µm after reflow were 5.075 µm and 2.034 µm respectively.
These values have increased after ageing for 2000 hours to 10.80 µm and 6.433 µm
respectively.
The present study also showed that more Ag3Sn of the block-type was formed
ahead of the interface when soldering with larger solder balls as shown in Figure
6.18a. The Ag3Sn also transforms to a more round and spherical shape during aging
instead of the faceted and blocky-shaped phase observed after reflow soldering
(Figure 6.18b). According to Wenger and Furrow (2000), Ag layer will dissolve
rapidly into the molten solder during reflow soldering because of the high solubility
of Ag in Sn. However, in the solid-solder condition, Ag atoms must come out of the
solder because the solubility of Ag in Sn is nearly zero (Jeon et al. (2003)). Then,
Ag atoms slowly accumulate on top of the Cu6Sn5 intermetallic. Figure 6.19
schematically illustrates the IMCs growth on Cu/Ag surface finish in as-reflow
(Figure 6. 19a and Figure 6.19b) and after aging process (Figure 6.19c)
115
(a) (a)
Ag3Sn
(b)
Figure 6.18: Ag3Sn on ImAg using 700µm solder (a) After reflow and (b) After
ageing for 500 hours
116
Substrate
Solder
Ag
(a)
Cu6Sn5
Cu
Sn
(b)
Cu3Sn Cu
(c)
Ag3SnAg
Cu6Sn5
Figure 6.19: Schematic diagram of IMCs growth in Cu/Au specimen: (a) dissolution
of Ag layer into molten solder, (b) formation of Cu6Sn5 during reflow soldering and
(c) Conversion of Cu3Sn and Ag3Sn after isothermal ageing
6.4.1.3 Formation of Kirkendall Voids
Figure 6.20 and Figure 6.21 show that Kirkendall voids formed inside the
Cu3Sn layer at the SAC solder/ ImAg interface. The Kirkendall voids start to form in
the Cu3Sn layer after ageing for 500 hours, probably attributed to the fast diffusion
rate of Cu and the fast reaction to form IMC at the interface. Research done by Xiao
et al. (2001) also showed that these Kirkendall voids formed in the same layer during
long-time ageing. It is important first to review some pointers on the Kirkendall
voids formation in solder joints. As mentioned before, the main diffusion element in
Cu6Sn5 is Sn and the main diffusion element in Cu3Sn is Cu. It seems that the
diffusion of Sn in Cu6Sn5 is slower, leading to a shortage of Sn to react with Cu in
the Cu3Sn layer. Thus, the lacking Sn in the lattice spaces in Cu3Sn can therefore
results in the formation of Kirkendall voids. This is also in agreement with the work
done by Laurila et al. (2005) where they mentioned that during the formation of
Cu3Sn IMC, both components, (Cu and Sn) diffuse into the Cu3Sn-phase but the
diffusion of Cu has been measured to be ~3 times faster, so the voids in the Sn-Ag-
Cu/Cu reaction couple are at the right location and thus could constitute a Kirkendall
plane. The growth rate of these voids is exponential with temperature, therefore
117
increasing significantly at higher temperature (particularly 125oC and above) during
ageing (Raiyo, 2005) as shown in Figure 6.22.
Figure 6.20: SEM image of cross-sectional view of 500µm SAC405 solder/ ImAg
after ageing for 500 hours.
Kirkendall voids
Kirkendall voids
Cu6Sn5
Cu3Sn
Cu6Sn5
Cu3Sn
Ag3Sn
Ag3Sn
Figure 6.21: SEM image of cross-sectional view of 700µm SAC405 solder/ ImAg
after ageing for 1000 hours.
118
Figure 6.22: The mechanism of Kirkendall Voids formation (Raiyo, 2005).
6.4.2 Intermetallics between Sn-Ag-Cu solders and ENIG Surface Finish
6.4.2.1 Interfacial Reactions during Reflow Soldering
Over the past several years there has been consistent growth in the use of
electroless nickel/immersion gold (ENIG) as a final finish. This finish offers
diffusion barrier in printed circuit board for the electronic packages. Since the
reaction rate of Ni with liquid Sn solder is slower than that of Cu, it acts to prevent
the rapid interfacial reaction between solder and Cu conductor in electronic devices.
This in turn, results in thinner intermetallic compounds. During reflow soldering, the
molten lead-free Sn-Ag-Cu solder alloy dissolves the entire Au layer into the liquid
solder, allowing Sn from the molten solder to react with the Ni layer to form Ni-Sn
intermetallics. In this research, the electroless nickel layer in the ENIG finish has
been deposited with a medium content of phosphorous of around 7 wt%.
In the present study the morphology and evolution of interfacial reactions
between ENIG and SAC405 and SAC305 lead-free solders were investigated using
cross-sectional and top surface views by the aid of optical and electron microscopy.
Figure 6.23 shows the optical micrographs of cross-sections of solder joint made
from SAC405 and SAC305 solders with three different solder bump sizes, 300, 500
119
and 700 µm after reflow soldering. What can be observed from these micrographs is
that the intermetallics formed are thinner compared to those formed on ImAg surface
finish as described in section 6.4.1.2. As will be discussed later the thickness of
these intermetallics is also smaller in the bigger solder and vice versa.
0
0
0 0
F
S
c
a
b
0
igure 6.23: Cross-section views of the inte
AC405 (a-c) and SAC305 (d-f) solders (X50
f
d
e
0
rmetallics formed between EN
0).
X50
X50
X50
X50
X50
X50
IG and
120
It has also been established (Laurila et al. 2005) that there are three main
IMCs in the Ni-Sn system, which are stable at temperature of interest, i.e. below
260oC: Ni3Sn, Ni3Sn2 and Ni3Sn4. However, since ternary reactions among Ni, Cu
and Sn at the interface are involved and complicated, several different results in
terms of IMC phases and compositions on ENIG surface finish were reported. In
previous works by Li et al. (2005), Sharif et al. (2005) and Jeon et al. (2003) it was
reported that only the (Cu, Ni)6Sn5 phase was detected at the interface after reflow
soldering at temperatures below 300°C. Huang et al. (2006) studied the interfacial
reactions between Sn-0.7Cu and Sn-3.5Ag-0.7Cu solders on ENIG and reported that
after reflow soldering at 250oC for 1minute only needle-shaped (Cu, Ni)6Sn5
intermetallic formed at the interface.
There has been several publications detailing the formation of intermetallics
between Cu-containing lead-free solders and Ni-based surface metallization (Chen et
al. (2002), Ho et al. (2002), Tsai et al. (2003), Ho et al. (2006) based on the Sn-Cu-
Ni phase diagram proposed by Lin et al. (Lin et al. (2002), while other researchers
(Hsu et al. (2003) and Wang et al. (2003) used an earlier version of the phase
diagram. According to Vuorinen et al. (2008) none of these authors took into
account the effect of super-saturation of the molten solder with either Ni or Cu, or
indeed the change in temperature. When soldering Cu-bearing Pb-free solders such
as Sn-Ag-Cu solders and Ni-based metallization, the interfacial reactions formed
consist of ternary intermetallics, (Cu, Ni)6Sn5, (Ni, Cu)3Sn4 or both (Cu, Ni)6Sn5 and
(Ni, Cu)3Sn4 depending on the content of Cu in the solder (Ho et al. (2002), Jeon et
al. (2003))
Ho et al. (2002) studied the reactions between Ni and Sn-Ag-Cu solders at
250 oC. In the solder composition they kept the Ag concentration constant at 3.9
wt% but the Cu content was varied between 0 wt% and 3.0 wt%. They observed that
when the Cu content in Sn-Ag solders is below 0.5 wt%, a continuous layer of (Ni,
Cu)3Sn4 is formed above which a small amount of discontinuous (Cu, Ni)6Sn5
particles is formed between the Ni layer and solder. When the Cu content increased
to 0.5 wt%, the (Cu, Ni)6Sn5 becomes the continuous layer over the (Ni, Cu)3Sn4
121
layer. At higher concentrations of Cu (higher than 0.5 wt%) only a continuous layer
of (Cu, Ni)6Sn5 is formed. Several other studies also reported similar findings albeit
with a different threshold of Cu concentration that separates between whether Ni or
Cu based intermetallics are formed. Quite recently, however, Vuorinen et al. (2008)
reported almost the same results as Ho et al. (2002). Based on both thermodynamics
and kinetics they reported that if the Cu content of the solder is less than 0.4 wt%,
only (Ni, Cu)3Sn4 nucleates at the Ni/solder interface at 250 oC soldering temperature
and is more stable than (Cu, Ni)6Sn5. They also reported that the critical Cu
concentration is temperature dependent.
In order to reveal the morphology and type of intermetallics formed during
reflow selective etching of the solders was performed on the samples. Figure 6.24
and Figure 6.25 show the cross sections and top views of the interface intermetallics
in the SAC405/ ENIG and SAC305/ ENIG for the same solder joints shown in
Figure 6.23 after reflow soldering. Chemical analysis by energy dispersive X-ray
analysis (EDX) revealed that the intermetallics formed during reflow soldering the
SAC405 joints using 300, 500 and 700 µm solders were of the type (Cu, Ni)6Sn5
with needle-shape as shown in Figure 6.24 (d-f). The initial chemical composition of
the (Cu, Ni)6Sn5 intermetallics for both SAC405 and SAC305 solders was with the
range: Cu (24-31 at%), Ni (12-17 at%) and Sn (30-47 at%). Selected EDX analysis
results are given in Appendix A. Thus, the results of the present study are in good
agreement with the results reported by Ho et al. (2002), Young et al. (2003), Zeng et
al. (2004), Huang et al. (2006) and Vuorinen et al. (2008). This indeed is clear from
the cross-sectional as well as top view morphologies that only a continuous layer of
(Cu, Ni)6Sn5 has formed during reflow soldering with either SAC405 or SAC305
solders with the Cu concentration being at 0.5 wt%. Soldering of the solder joint in
the present study was performed at a temperature below 250 oC and based on the
experimental results of Vuorinen et al. (2008) (Cu, Ni)6Sn5 is expected to be more
stable in Sn-Ag-Cu solders containing around 0.5 wt% (like in the present study).
This is exactly what was observed here in all solder bump sizes investigated except
for the smallest solder bump size of 200 µm.
122
(Cu,Ni)6Sn5
(Cu,Ni)6Sn5
(Cu,Ni)6Sn5
e
d
f
b
a
c
X500
X500
X500
Figure 6.24: Cross section and top views of (Cu, Ni)6Sn5 intermetallic formed during
reflow between ENIG and SAC405 solder. (a, d) 300 µm, (b, e) 500 µm and (c, f)
700 µm.
123
(Cu,Ni)6Sn5
(Cu,Ni)6Sn5
(Cu,Ni)6Sn5fc
d
e
a
b
X500
X500
X500
Figure 6.25: Cross section and top views of (Cu, Ni)6Sn5 intermetallic formed during
reflow between ENIG and SAC305 solder. (a, d) 300 µm, (b, e) 500 µm and (c, f)
700 µm.
124
When soldering with the smaller solder balls of 200 µm, however, the results
obtained showed that (Ni, Cu)3Sn4 intermetallic formed at the interface instead of
(Cu, Ni)6Sn5 for both SAC405 and SAC305 solders (Figure 6.26). The composition
of this (Ni, Cu)3Sn4 intermetallic was: Ni (30-37 at%), Cu (5-9 at%) and Sn (43-47
at%). This observation is supported by the explanation proposed by Vuorinen et al.
(2008) based on super-saturation of the molten solder at the interface near the solid
base metal with both Ni and Cu during soldering.
As the solder reaches saturation the dissolution rate decreases and from this
stage on the dissolution of the metallization contributes only for the growth of the
intermetallic layer at the solder joint interface. This super-saturation, they claimed,
would also lead to the formation of a thick two phase solidification structure namely
(Ni, Cu)3Sn4 and (Cu, Ni)6Sn5. Since a smaller solder volume (smaller solder bump)
will be supersaturated with both Ni and Cu very quickly during soldering, then both
ternary Cu-based and Ni-based intermetallics would form as observed in the present
study. (Cu, Ni)6Sn5 intermetallic is a stable phase and the presence of Ni will
stabilize it further very strongly (Laurila et al. (2005)), and it was reported by Huang
et al. (2006) that as much as 25 at% of Ni may replace Cu atoms in the (Cu, Ni)6Sn5.
To provide further evidence for the results observed, cross-sections of solder
joints were selectively deep etched to reveal the morphology and analyse the
chemical composition of the interfacial reactions formed during soldering using all
solder ball sizes for both solders investigated. EDX analysis confirmed that when
soldering with 300, 500 and 700 µm solder ball only one intermetallic was formed,
which is (Cu, Ni)6Sn5 while in the 200 µm solder ball both (Ni, Cu)3Sn4 and (Cu,
Ni)6Sn5 formed at the interface as shown in Figure 6.27 and Figure 6.28 for SAC405
and SAC305 solders respectively. The EDX analysis results for the respective
micrographs are given in Figure 6.29 and Figure 6.30.
The presence of only one intermetallic phase, (Cu, Ni)6Sn5, after soldering
with larger SAC405 and SAC305 solders (300, 500 and 700 µm) can be attributed to
125
what is called by Huang et a. (2006) as the “Cu concentration effect”. As discussed
above the a value of 0.5 wt% of Cu was found to be the critical value that determine
the transition between (Cu, Ni)6Sn5 to (Ni, Cu)3Sn4. Since the Cu content in both
solders used in the present study is fixed at 0.5 wt%, it was expected that both (Cu,
Ni)6Sn5 to (Ni, Cu)3Sn4 would form in all solder bump sizes used. The Cu content is
solder joints is limited and during soldering, this Cu is involved in the interfacial
reaction. As this interfacial reaction or intermetallic grows in thickness Cu in the
solder is consumed and thus its concentration might change compared to the initial
solder composition. The fact that for small solder joints made with 200 µm solder
bump (Ni, Cu)3Sn4 formed at the interface is an indication that the Cu content at the
interface has decreased since saturation with both Ni and Cu occurs rapidly after
melting making dissolution of more Cu from the substrate almost impossible.
However, in larger solder joints more Cu can be dissolved in the solder before
saturation is achieved. In this situation the Cu concentration at the interface may not
decrease below the critical value resulting in the formation of only (Cu, Ni)6Sn5
intermetallic.
126
(Ni, Cu)3Sn4
(Ni, Cu)3Sn4
Ni3Sn2
Figure 6.26: Top view of intermetallics formed between ENIG and 200 µm SAC405
(top) and SAC305 (bottom) solders.
127
(Cu, Ni)6Sn5
Ag3Sn
(Cu, Ni)6Sn5
b
a
Figure 6.27: SEM images of cross sections of intermetallic formed between ENIG
and SAC405 for (a) 500 µm and (b) 700 µm solders.
128
(Cu, Ni)6Sn5
(Cu, Ni)6Sn5
b
a
Figure 6.28: SEM images of cross sections of intermetallic formed between ENIG
and SAC305 for (a) 300 µm and (b) 500 µm solders.
129
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.45 0.7713 12.84 1.40 19.43 Cu L 2.39 0.5637 28.88 1.26 40.38 Sn L 6.22 0.9055 46.79 1.44 35.02 Au M 1.32 0.7856 11.48 1.20 5.18 Totals 100.00
Figure 6.29a: EDX results of interface intermetallic formed between ENIG and 500
µm SAC405 solder during reflow (EDX for Figure 6.27a).
130
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.01 0.7742 9.28 1.36 14.71 Cu L 2.35 0.5827 28.47 1.24 41.71 Sn L 5.84 0.9083 45.43 1.40 35.63 Au M 1.90 0.7987 16.83 1.23 7.96 Totals 100.00
Figure 6.29b: EDX results of interface intermetallic formed between ENIG and 700
µm SAC405 solder during reflow (EDX for Figure 6.27b).
131
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.64 0.7733 15.19 1.67 22.89 Cu L 2.08 0.5535 27.01 1.52 37.62 Sn L 5.74 0.9031 45.65 1.69 34.03 Au M 1.33 0.7850 12.15 1.38 5.46 Totals 100.00
Figure 6.30a: EDX results of interface intermetallic formed between ENIG and 300
µm SAC305 solder during reflow (EDX for Figure 6.28a).
132
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.98 0.7769 19.32 4.24 28.75 Cu L 1.72 0.5359 24.28 4.02 33.38 Sn L 5.21 0.8986 43.91 4.42 32.32 Au M 1.29 0.7820 12.49 3.73 5.54 Totals 100.00
Figure 6.30b: EDX results of interface intermetallic formed between ENIG and 500
µm SAC305 solder during reflow (EDX for Figure 6.28b).
133
6.4.2.2 Effect of Isothermal Aging on Intermetallics
Intermetallics will continue to grow when subjected to heating or ageing.
Figure 6.31 shows typical cross sectional of the solder joints after reflow and after
ageing up to 2000 hours at 150oC for both SAC405 and SAC305 solders with the
solder bump sizes being 500 and 700 µm. In contrast to the intermetallics observed
after solder reflowing, ageing of the solder joints led to the intermetallics gradually
transforming to a layered structure as shown clearly by the SEM images of the cross-
sectional solder joints in Figure 6.32b and also the top views of the intermetallics
morphologies after reflow and ageing for up to 2000 hours for the SAC405 solder
(Figure 6.33). There is also an indication that the intermetallic layer grew in
thickness with increasing ageing time (detailed discussion on intermetallics thickness
is presented later in this chapter), which is consistent with previous research work.
Figure 6.33 and Figure 6.34 clearly show that due to ageing coarsening of the
intermetallics has occurred.
The present results also clearly showed that there was no change in the type
of intermetallics formed after the joints were subjected to ageing. That is the same
type of intermetallics observed after reflow were detected after ageing in the
respective solder bump sizes. Only (Cu, Ni)6Sn5 intermetallic layer was observed in
the 300, 500 and 700 µm solders whereas two layers: (Ni, Cu)3Sn4 and (Cu, Ni)6Sn5
intermetallics were formed in the smaller solder bump size of 200 µm as confirmed
by EDX analysis. The quantitative results of EDX analyses for the intermetallics
after reflow and ageing for both solders and for all solder bump sizes are shown in
Table 6.3.
For the 300, 500 and 700 µm SAC405 and SAC305 solders, the thickening
and coarsening of intermetallics appears to be slow than that in the 200 µm solders.
For example, after ageing for only 250 hours at 150oC the (Ni, Cu)3Sn4 showed quite
a significant coarsening compared to the 700 µm solder as shown in Figure 6.33. As
134
the ageing time increased, both (Cu, Ni)6Sn5 and (Ni, Cu)3Sn4 intermetallics
coarsened but it is quite evident that the (Ni, Cu)3Sn4 grains coarsened at a faster rate
as shown by the larger grains compared to the (Cu, Ni)6Sn5 (compare Figure 6.33
with Figure 6.34). These observations are consistent with previous studies (Huang
et al. (2006).
Table 6.3: Compositions of the interfacial reaction products after reflow soldering and ageing for 2000 hours at 150oC. Solder
Reaction Product after reflow Reaction product after ageing
for 2000 hours at 150oC 200 µm (Ni80, Cu20)3Sn4/ (Cu, Ni)6Sn5 (Ni85, Cu15)3Sn4/ (Cu60, Ni40)6Sn5
Yee, S. and Ladhar, H. (1998). Reliability Comparison of Different Surface Finishes
on Copper. Circuit World, 25(1): 25-29.
Yoon, J. W., Kim, S. W. and Jung, S. B. (2004). IMC Morphology, Interfacial
Reaction and Joint Reliability of Pb-free Sn-Ag-Cu Solder on Electrolytic Ni
BGA Substrate. Journal of Alloys and Compounds, 392, 247-252
Yu, D. Q., Wang, L., Wu, C. M. L. and Law, C. M. T. (2005). The Formation of
Nano-Ag3Sn Particles on the Intermetallic Compounds During Wetting Reaction.
Alloys and Compounds, 389: 153-158.
Yu, D. Q., Wu, C. M. L., Law, C. M. T., Wang, L. and Lai, J. K. L. (2004).
Intermetallic Compounds Growth between Sn-3.5Ag Lead-free Solder and Cu
Substrate by Dipping Method. Journal of Alloys and Compounds, 392, 192-199
174
Zeng, K. and Tu, K. N. (2002). Six Cases of Reliability Study of Pb-free Solder
Joints in Electronic Packaging Technology. Material Science and Engineering R.,
38: 55-105.
APPENDIX A
FESEM EDX RESULTS (SELECTED SAMPLES ONLY)
176
EDX result of 700 µm solder SAC405/ ENIG - Reflow (Spot 1)
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.14 0.7822 10.90 1.44 17.24 Cu L 2.14 0.5793 27.62 1.29 40.36 Sn L 5.22 0.9049 43.19 1.43 33.78 Au M 1.95 0.7986 18.28 1.26 8.62 Totals 100.00
177
EDX result of 200 µm solder SAC405/ ENIG_1000 hours ageing (Spot 2)
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni K 4.48 1.1337 6.22 0.68 10.04 Cu L 7.03 0.4106 26.92 1.10 40.20 Sn L 31.77 0.9047 55.22 1.13 44.15 Au M 5.68 0.7681 11.64 0.90 5.61 Totals 100.00
178
EDX result of 500 µm solder SAC405/ ENIG_500 hours ageing (Spot 1)
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 0.90 0.7530 7.98 1.33 12.96 Cu L 2.26 0.5761 26.23 1.20 39.37 Sn L 6.77 0.9155 49.51 1.44 39.78 Au M 1.96 0.8048 16.29 1.22 7.89 Totals 100.00
179
EDX result of 500 µm solder SAC405/ ENIG_500 hours ageing (Spot 2)
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 0.71 0.7440 6.71 1.41 11.10 Cu L 2.07 0.5772 25.30 1.24 38.68 Sn L 6.68 0.9196 51.24 1.52 41.95 Au M 1.93 0.8098 16.75 1.27 8.26 Totals 100.00
180
EDX result of specimen (200 µm solder) SAC405/ ENIG- Reflow
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.64 0.6788 17.28 1.29 30.52 Cu L 0.30 0.4892 4.34 1.05 7.08 Sn L 7.93 0.9324 60.92 1.61 53.21 Au M 2.02 0.8274 17.46 1.31 9.19 Totals 100.00
181
EDX result of 300 µm SAC305/ ENIG- Reflow
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 0.64 0.7680 6.71 1.21 14.36 Cu L 0.64 0.6031 8.47 1.02 16.76 Sn L 4.03 0.9211 35.11 1.40 37.17 Au M 5.51 0.8884 49.71 1.56 31.71 Totals 100.00
182
EDX result of 200 µm SAC305/ ENIG – 2000 hours ageing
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Ni L 1.81 0.7725 16.66 1.85 24.99 Cu L 1.99 0.5459 25.91 1.61 35.92 Sn L 5.78 0.9020 45.48 1.80 33.75 Au M 1.32 0.7838 11.95 1.35 5.34 Totals 100.00
183
EDX result of 200 µm SAC305/ ImAg – 1000 hours aging
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Cu L 9.61 0.4394 38.52 0.90 53.92 Sn L 32.52 0.9312 61.48 0.90 46.08 Totals 100.00
184
EDX result of 200 µm SAC305/ ImAg - Reflow
Element App Intensity Weight% Weight% Atomic% Conc. Corrn. Sigma Cu L 2.19 0.6911 27.60 1.05 48.49 Sn L 2.92 0.9070 28.03 1.22 26.37 Au M 4.38 0.8574 44.36 1.38 25.14 Totals 100.00
Abstract— This study investigates the interfacial reactions occurring during reflow soldering between Sn-Ag-Cu lead-free solder and two surface finishes: electroless nickel/ immersion gold (ENIG) and immersion silver (IAg). The study focuses on interfacial reactions evolution and growth kinetics of intermetallic compounds (IMC) formed during soldering and isothermal ageing at 150 oC for up to 2000 hours. Optical and scanning electron microscopy were used to measure IMC thickness and examine the morphology of IMC respectively, whereas the IMC phases were identified by energy dispersive X-ray analysis (EDX). The results showed that the IMC formed on ENIG finish is thinner compared to that formed on IAg finish. For IAg surface finish, Cu
6Sn
5 IMCs with scallop morphology
are formed at the solder/ surface finish interface after reflow while a second IMC, Cu
3Sn was formed
between the copper and Cu6Sn
5 IMC after the
isothermal ageing treatment. For ENIG surface finish both (Cu, Ni)
6Sn
5 and (Ni, Cu)
3Sn
4 are formed after
soldering. Isothermal aging of the solder joints formed on ENIG finish was found to have a significant effect on the morphology of the intermetallics by transforming to more spherical and denser morphology in addition to an increase in their thickness with increased ageing time. Keywords: lead-free; immersion silver; electroless nickel; ageing; intermetallic 1. Introduction
The harmful effects of lead (Pb) on the environment and human health have stimulated substantial research and development efforts to discover lead-free solder alloys for electronic application. The most promising solder alloy recommended by NEMI 2000 (National Electronic Manufacturing Initiative) is Sn-Ag-Cu ternary alloy which has advantages of good wetting property, superior interfacial properties, high creep resistance and low coarsening rate [1,2]. Apart from that, the other important issue for the development of Pb-free is to find a suitable surface finish on a printed circuit board (PCB), where it act as a diffusion barrier layer, wettable layer and a corrosion resistant layer. The surface finishes used in this study include immersion
silver (IAg), electroless nickel/immersion gold (ENIG) and copper as a reference.
The main reason of using IAg is because of the
higher co-planarity requirement for the fine pitch surface mount assemblies. Immersion finishes produce single element coating which result in relatively thin layers (typically less than 1 µm) because the deposition process halts when the substrate surface is completely covered with the coated material [3]. During soldering, the Ag coating does not melt. Instead, it dissolves into the molten solder, which may decrease the speed of the wetting. Some recent electromigration test results indicate that the migration of Ag is not a concern [4].The advantages of this surface finish are: lower material cost, wire-bondable, simpler operation, planar surface, long shelf life and also good for ultra-high speed signals.
Another type of surface finish studied in this
research is ENIG. Electroless nickel (Ni) layer is deposited on the Cu substrate as a diffusion barrier between Cu and solder materials [8]. While immersion gold (Au) layer is deposited onto the nickel layers in order to prevent oxidation and act as a solderable finishes in reflow soldering.
In the development of package materials system,
joint reliability should be considered as one of the most critical criteria [5,6]. This joint is produced during soldering where copper dissolve into the molten solder and react with Sn forming intermetallic compound (IMC) at the solder/copper interface. Although the formation of the IMC layer is desirable for good wetting and bonding, excessively thick layer is harmful because of its brittle nature that makes it prone to mechanical failures even at low loads [7].
In this study, we investigate the intermetallic growth and thickness between Sn-4Ag-0.5Cu solder alloy with different surface finishes such as immersion silver (IAg), and electroless nickel/ immersion gold (ENIG) during reflow soldering and after ageing treatment. The solder joint is evaluated in terms of IMC thickness.
2. Experimental details
Copper samples of size 45 mm × 50 mm × 1 mm were cut from sandwich copper sheet. The sample
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
substrates were lightly grinded and then polished to remove any oxide of copper and to increase the wettability. The samples then were subjected to pretreatment process in order to remove the dirt, grease, oxide layer and also to activate the Cu surfaces before plating process. There are two types of surface finish used in this study, namely, IAg and ENIG. Table 1 shows the thickness for each surface finish. For ENIG samples, the electroless Ni layer contains ~13wt% phosphorous. After ENIG plating process, the samples were then being deposited with gold layer through immersion plating without any pretreatment except rinsing in running tap water. While for IAg surface finish, a few combinations of immersion solution have been tried before we had finally arrived at the most stable combination which is: 50g/L of ethylene diamine tetra-acetic acid (EDTA), 20.4g/L of sodium hydroxide and 1g/L of silver nitrate. Figure 1 shows the steps followed in the present study to achieve the suitable solution with the desired silver thickness. The operating temperature of the plating bath was set up to 45 ºC. After that, all the samples were laminated with a layer of solder mask to restrict the molten solder from flat spreading during reflow. Then, the solder mask together with the patterned film was cured by ultraviolet (UV) light in order to produce small openings of 0.38 mm in diameter. After curing the samples, a thin layer of no clean flux is applied onto the substrate to remove the oxide layer and also to improve the wetting of molten solder during reflow soldering. Then, the substrate was manually populated by the solder balls with 700 µm of diameter in an array of 9 x 5 on each substrate. The metallurgical bonding between the solder balls and substrate surface was formed by reflow soldering at temperature 250 ºC.
After reflow soldering, the samples were subjected to ageing treatment at 150 ºC for 500 hours, 1000 hours and 2000 hours. Then, all the samples were characterized both at the top surface and cross section. From top surface, 3-D morphology of these intermetallics will be examined by etching away the solder (deep etching). Both scanning electron microscope (SEM) and Energy dispersive X-ray (EDX) are used to study the intermetallics at the interface. Cross sections of the specimens are prepared using standard metallographic steps and examined using Nikon optical microscope, image analyzer, SEM-EDX and FESEM (Field emission scanning electron microscope).
Table 1: Thickness of layer of each surface finish
SURFACE FINISH THICKNESS DEPOSITED
(µm) IAg (Ag) ~0.3 ENIG (Ni/Au) ~4 – 5 / ~0.3
Figure 1: Process flowchart for Immersion Silver
3. Results and discussion To achieve one of the objectives of the current
study, an attempt was made to deposit the silver layer with very thin thickness. Several plating solutions were tried to plate a silver layer with the desired thickness. The solutions range from acidic to alkaline with different process parameters. After many attempts, we successfully identified one neutral solution capable of depositing a layer of silver with the thickness of 0.2744µm for thin coating and 0.7854µm for thick coating. Figure 2 shows the results of copper substrate before plating (after pretreatment process) while Figure 3 shows the results of immersion silver plating. From the EDX analysis, the results clearly shows that the silver layer deposited using the neutral solution, which operates at a pH value of 6.46, fully covered the copper substrate (Figure 3).
193
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
Figure 2: Copper substrate before plating (after
pretreatment process)
Figure 3: FESEM-EDX results of IAg on Cu
3.1. Bare Cu
After reflow soldering, a uniform spherical shape of copper-tin intermetallic formed in the solder joint. As identified by EDX analysis, the reaction layer on Cu substrate is the Cu6Sn5 phase. This result similar to the previous worked done by Blaine Partee [8] and Amagai et al [9]. In addition to the Cu6Sn5 intermetallics, Ag3Sn IMCs also formed at the interface and in the bulk solder. Its morphology is that of large dendrite-like and plates as shown in Figure 6. These large Ag3Sn intermetallics are quite brittle, which may lead to serious problems under stressed conditions in the actual service for PCB. Zeng and Tu [1] mentioned that these Ag3Sn crystals
will easily lead to failure if formed in a high stress area, such as corner region between solder bump and top surface metallurgy (TSM), cracks can be initiated and propagated along the interface between Ag3Sn and solder.
During isothermal aging, the IMCs continue to grow by inter-diffusion between the Sn in the solder and Cu in the substrate. This solid state aging resulted in a significant increase in the thickness of the Cu6Sn5 IMCs (Figure 4). Figure 4b, shows that the IMC has coarsened and became more uniform instead of the scallop-type observed after reflow soldering. Moreover, after heating the solder joint, a new IMC phase has formed between the Cu and Cu6Sn5 IMC as shown by the thin grey layer in Figure 5 which is known as Cu3Sn. The possible mechanism for Cu3Sn can be proposed as follows: in solid state, Sn diffuses more slowly than Cu inside Cu6Sn5 IMC, so Cu accumulated at the interface between Cu and Cu6Sn5 resulting in the formation of Cu3Sn, which can consume some of Cu6Sn5 IMC at the beginning of the solid reactions. During ageing, the Cu diffuses towards the solder, so does Sn towards the Cu layer, resulting in the growth of both Cu3Sn and Cu6Sn5 IMC layers [10]
(a)
(a)Cu6Sn5
(b)
Ag = 75.13%
Cu = 24.87%
(b)Cu6Sn5
Ag3Sn
Figure 4: Morphology of Cu6Sn5 (a) After reflow, (b)
After ageing 1000 hours
194
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
Figure 5: Formation of Cu3Sn at the solder interface after isothermal aging
Figure 6: Formation of Ag3Sn after reflow soldering 3.2. Immersion Silver
In the solder joint between Sn/4Ag/0.5Cu and IAg
surface finish, the same type of IMC can be observed in Figure 7 which is Cu6Sn5. The morphology is uniformly round and smooth spherical shape. However, the size of the IMC is smaller compared to bare copper. Apart from that, the small particles and large plate-like Ag3Sn IMC was also seen in these solder joints.
Moreover, similar to bare copper, after heating the solder joint, a new IMC phase has formed between the Cu and Cu6Sn5 IMC as shown by the thin grey layer in Figure 5 which is known as Cu3Sn when identified by EDX analyses. This layer exactly followed the morphology of the Cu substrate. However, there were numerous Kirkendall voids formed in the Cu3Sn layer (Figure 8). Xiao et al [11] showed that Kirkendall voids formed during the long-time ageing. The formation mechanism of Kirkendall voids in Cu3Sn layer appeared to be different compared to the voids in Ni3P. In such cases, the main diffusion element is Sn in Cu6Sn5 but Cu in Cu3Sn [12]. Diffusion of Sn in Cu6Sn5 is very slow, which determines the entire growth of the IMCs,
leading to shortage of Sn to react with Cu in Cu3Sn layer. The lacking Sn in the lattice spaces in Cu3Sn can therefore result in the formation of Kirkendall voids. However, in our study, no Kirkendall voids were found in the Cu3Sn layer for bare Cu, indeed, Kirkendall voids were observed if IAg was used as surface finish. This indicates an interconnection between Kirkendall voids and immersion silver, but the details of the reason is still unclear.
Cu6Sn5
Cu3Sn
500 X
Cu6Sn5
Figure 7: Morphology of Cu6Sn5 after reflow.
Ag3Sn
1
Figure 8: Kirkendall voids found in Cu3Sn IMC
during long time ageing treatment. 3.3. Electroless Nickel/ Immersion Gold
Different types of IMCs found when using this surface finish. It is believed that the reasons of this formation are due to wetting properties of the lead-free solders composition and inconsistently distributed composition on the solder itself [13]. Figure 9 shows the circular boundary regions which form at the solder interface when using this type of surface finish. During reflow, the Au layer dissolves into the solder, exposing the Ni layer to react with the solder to form the solder joint. Diffusion of the Ni layer into the solder will results in the formation of IMCs, consuming the Ni layer.
2 3 4
195
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
Figure 9: Circular boundary regions which form in
the IMC layer of Sn/4Ag/0.5Cu.
Figure 10a shows the interface morphology at the solder joint which clearly shows that mainly needle-shaped (Ni,Cu)3Sn4 IMCs were observed after reflow. While (Cu, Ni)6Sn5 was observed to form between (Ni,Cu)3Sn4 and solder (Figure 11a). This results are in agreement with previous study [12] that the first phase to form and grow in the Ni-Sn systems is (Ni,Cu)3Sn4 or Ni3Sn4. Ni3Sn2 was believed to grow very slowly and probably requires a larger undercooling. Cheu Li [14] agree that this is the reason why normally Ni3Sn2 will be observed only near the edges of the solder joints. However, in this study, there was no Ni3Sn2 found.
Figure 10: Top views of ENIG finish (a) After reflow (b) After ageing 1000 hours
During ageing, IMC’s will continue to grow. Figure 10b and Figure 11b shows the top view and cross sectional images of solder bumps after aging process. In contrast to IMCs after reflow, all IMC’s were formed with a layered structure. There is also an indication that (Ni, Cu)3Sn4 grew thicker with isothermal ageing
(a)
(Ni,Cu)3Sn4
(Cu,Ni)6Sn5
(b)
(Ni,Cu)3Sn4
(a) (Cu,Ni)6Sn5(Ni,Cu)3Sn4
Figure 11: Cross section of Sn-Ag-Cu /ENIG finish (a) After reflow, (b) After ageing 2000
hours
3.4. Determination of IMC thickness
Thickening of intermetallics is a major concern in reliability issue of electronics packaging. This is because it will lead to the failure of the joint with fracture in the IMC itself or along the interface between the solder and IMC layer when expose to any mechanical forces, such as vibration, expansion and contraction caused by variation in temperature.
(b)(Ni,Cu)3Sn4
Figure 12 shows the effect of isothermal ageing on
the intermetallic thickness for all type of surface finishes studied. Generally, bare Cu substrate gave the greatest IMC thickness in both reflow and aging condition. This is probably due to the fast interaction between Cu and Sn. While IAg surface finish shows slower growth rate than bare copper, but higher than ENIG surface finish. However, up to certain extends,
196
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
the thickness of IMCs decrease for IAg and bare copper finish. This is due to the existence of the IMC layer slows down the diffusion of the Cu at the interface. For ENIG surface finish, the growth of the IMCs keep increasing until ageing 2000 hours. ENIG produced the thinnest IMCs because there is Ni layer that act as a diffusion barrier layer between the solder and Cu. Thus, the diffusion will be slower.
IMC Thickness VS Surface Finish
0.000
2.0004.000
6.0008.000
10.000
IAg ENIG COPPER
Surface Finish
Thic
knes
s (µ
)
As-reflow 500 hours 1000 hours 2000 hours
Figure 12: IMC thickness after reflow and ageing for
all surface finishes studied 4. Conclusion
In this study, the microstructure of the IMCs has been studied when lead-free solders react with different surface finishes. On bare Cu and IAg, the Cu6Sn5 and Cu3Sn IMCs formed at the interface. The Cu3Sn grew due to the slow diffusion rate of Sn in Cu6Sn5 IMC after ageing treatment. Apart from that, the intermetallics are also thickened and coarsen after ageing. While on ENIG surface finish, the IMCs formed were (Cu, Ni)6Sn5 and (Ni, Cu)3Sn4 phases at the interface after reflow. These IMCs also become thicker due to the persistent consumption of Ni during aging. However, ENIG produced the thinnest IMCs as compared to IAg and bare Cu.
Acknowledgement The authors would like to acknowledge the financial support provided by INTEL (Malaysia) Research Student Fellowship (vot 73724) and UTM for providing the research facilities. References 1. K. Zeng, K.N. Tu, Six cases of reliability study
of Pb-free solder joints in electronic packaging technology, Mater. Sci. & Eng. R 38 (2002) 55-105.
2. http://www.nemi.org 3. P. T. Vianco, An Overview of Surface Finishes
and Their Role in Printed Circuit Board
S. Chada, Investigation of Immersion Silver PCB Finishes for Portable Product Applica
Proceedings of SMTA International, Chicago, IL, October 2001, pp. 604-611.
J. W. Yoon, S. W. Kim, S. B. Jung, Mater. Trans. 45 (2004) 727-733.
6. W. M. Chen, M. McCloskey, S. C. O’Mathuna, Microelectron. Relia
7. K. H. Prakash and T. Sritharan, Acta Mater, 49 (2001), 2481 – 2489.
8. B. Partee, Intermetallics In Electronic Soldering, National Electronics Excellence, June 2004. M. Amagai, M. Watanabe, M. Omiya, K. Kiahimoto, T. Shibuya, Mechanicacharacterization of Sn-Ag-based lead-free solders, Microelectron. Reliab. 42 (2002) 951-966. T.Y.Lee et al., Morphology, kinetics and thermodySnPb and Pb-free solders (Sn-3.5Ag, Sn-3.8Ag-0.7Cu and Sn-0.7Cu) on Cu Journal of Materials Research (2002), 291-301. G. W. Xiao et al., Effect of Cu stud microstruture and electrointermetallics compound growth and reliability of flip chip solder bump, IEEE Transactions on Components and Packaging Technologies, (2001), 682-690. T. Laurila, V. Vourine and J.K. Kivilathi, Interfacial reactioand common base materials, Materials Science & Engineering R-Reports (2005), 1-60. Seeling K. and Suraski D., (2001), Materials and Process Considerations for LElectronics Assembly, Circuits Assembly, 12, No. 12. T. Cheu Li, Interfacial Reaction Between Sn-Ag-Cu LFinishes (2006), 140-141. Huang, Y. W., Collier, P., Teo, K. et al. Wafer Bumping by Stencil PrintinPan Pacific Microelectronics Symposium. February 10-13, 1998. Kauai, HI, 455-460.
2Intel Technology (M) Sdn. Bhd. Bayan Lepas, Penang
Abstract— Flip chip technology provides the ultimate in high I/O-density and count with superior electrical performance for interconnecting electronic components. Therefore, the study of the intermetallic compounds was conducted to investigate the effect of solder bumps sizes on several surface finishes which are copper and Electroless Nickel / Electroless Palladium / Immersion Gold (ENEPIG) which is widely used in electronics packaging as surface finish for flip-chip application nowadays. In this research, field emission scanning electron microscopy (FESEM) analysis was conducted to analyze the morphology and composition of intermetallic compounds (IMCs) formed at the interface between the solder and UBM. The IMCs between the SAC lead-free solder with Cu surface finish after reflow were mainly (Cu, Ni)6Sn5 and Cu6Sn5. While the main IMC’s formed between lead-free solder on ENEPIG surface finish are (Ni, Cu)3Sn4 dan Ni3Sn4. The results from FESEM with energy dispersive x-ray (EDX) have revealed that isothermal aging at 150ºC has caused the thickening and coarsening of IMCs as well as changing them into more spherical shape. The thickness of the intermetallic compounds in both finishes investigated was found to be higher in solders with smaller bump size. From the experimental results, it also appears that the growth rate of IMC’s is higher when soldering on copper compared to ENEPIG finish. Besides that, the results also showed that the thickness of intermetallic compounds was found to be proportional to isothermal aging duration. 1. Introduction The requirements of electronic packaging is toward higher I/O, greater performance, higher density, and lighter weight, the use of area array packaging technology is expected to increase. The type of packaging, Flip Chip, provides the ultimate in high I/O-density and count with superior electrical performance, and very small size [1]. It is well known that soldering involves a reaction between molten solder and substrate, which dissolves some of
the substrate and which forms some sort of intermetallic layer [2]. The interfacial chemical reactions enhance the wettability between the solder and the substrate. The intermetallic compound (IMC), grows at the solder and at the under bump metallurgy (UBM) interface at the practical operating temperature during the reflow process and eventually forms the solder bump. Due to its excellent conductivity and surface for soldering, copper has been widely used as the substrate materials. Though, surface finishes are still needed to be deposited onto the substrate surface as copper may oxidize easily. For electroless nickel/ electroless palladium/ immersion gold (ENEPIG) surface finish, nickel functions as the diffusion barrier with its low dissolution rate into tin. Meanwhile the palladium and gold layers can protect the underlying metals from oxidation. ENEPIG is formed by the deposition of electroless nickel (120 – 240 micro inches), followed by 5 to 15 micro inches of electroless palladium with an immersion gold flash (1 – 2 micro inches). The electroless palladium layer prevents any probability of nickel corrosion that may caused by the immersion gold deposition reaction. This layer is much harder than gold, providing added strength to the surface finish for wire bonding and connector attachment, while protecting the underlying nickel from oxidation [3].
Figure 1: The deposition layers of ENEPIG surface finish
The nickel/palladium/gold plated boards have a shelf life of up to two years or more. The process is almost similar to the nickel/gold process, except that it uses a palladium metal layer that is deposited after the
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
nickel layer, but prior to the final gold layer. ENEPIG is the finish that has the widest latitude for a variety of applications. It provides a flat co-planar surface. Sometimes referred to as the universal finish, it is a good soldering surface, a gold wire bondable surface, aluminium wire bondable surface, as well as a contacting surface. 2. Experimental Details In order to investigate the effect of solder bump size on intermetallic compound (IMC) formation between solder bump (SnAgCu) and substrate, three different sizes of solder balls, namely 300, 500, and 700 µm in diameter were soldered on two different finishes; Bare Copper surface finish and Electroless Nickel / Electroless Palladium / Immersion Gold (ENEPIG) finishes. The Copper substrate is the copper-polymer sandwiched substrates (FR4). The experiments must start with depositing the under bump metallurgy first. The ENEPIG is an under bump metallurgy which consists of layers of Nickel, Palladium and Gold towards copper plate. The first step in the ENEPIG process is to catalyze the copper surface for Ni deposition. The copper substrates were first subjected to grind mechanically to remove the oxide layer and rust followed by rinse with running tap water. Then, the copper substrates were subjected to a pretreatment cleaning process to remove dust, grease and oxide layers as well as to activate the copper surface for ENEPIG plating preparation. Since Ni is less noble than Cu, activation of the copper surface for Electroless Ni plating can be achieve by seeding a noble metal such as palladium chloride (or palladium sulfate [4]. The substrate was activated twice during the pretreatment before electroless nickel plating and before the electroless palladium plating was initiated with palladium solution. The immersion gold plating was performed right after electroless palladium without pretreatment except rinsing with water. The Nickel deposition is aided by a hyphophosphite (H2PO2) reducing agent that decomposes during this reaction and results in the decomposition of phosphorus in the electroless Ni layer [5]. In this research, medium phosphorus / medium phosphorus (7-10%) was used. Several combinations of electroless Nickel plating solutions have been tried out before arriving at one most stable plating succession. The components of optimum nickel plating solution are one part of NIMUDEN 5X and four parts of distilled water with 50 minutes plating times until reach the thickness of 5-6 µm at temperature of 950C and the range of pH is 4.3-4.5. For palladium plating the solution used consisted of 2g/L PdCl2, 200ml/L NH4OH (28% NH3), 3.5g/L Na2EDTA.2H2O, 20mg/L (8 drops) Thiodiglycollic
acid, and 8.5g/L NaH2PO2H2O where the operating temperature of the plating bath was set around 450C and a pH within the range of 8.0-9.0. The final step for ENEPIG plating is immersion gold where the gold layer acts as an oxidation barrier. The immersion gold plating is conducted immediately after electroless palladium with no pretreatment, except rinsing in running tap water. The combination for the immersion gold solution are 2g/L Potassium cyanoaurate (KAuCN2), 75g/L Ammonium chloride, 50g/L dehydrate Sodium citrate, and10g/L dehydrate Sodium hyphophosphite. The operating temperature of the plating bath was set around 93ºC, and a pH within the range of 7.0-7.5 After preparing the substrate and under bump metallurgy, the differences sizes of solder balls were placed onto the surface and then the whole structure was subjected to reflow soldering. This process was followed by an isothermal solid state aging at a temperature of 1500C for different aging times. Analysis of the results obtained focused mainly on characterization the intermetallic compounds formed between the Cu/Au under bump metallurgy and solders. This was made as function of process parameters such as thickness and morphology of IMC.
3. Results and Discussion 3.1. Effect of Solder Bump Size on Intermetallic Compound Thickness In order to study the effect to the intermetallics thickness, they were measured from the cross sections of different solder joints and it was shown in histograms in Figure 2. These histograms illustrate the effect of solder joint size on interfacial reactions occurring during soldering. It is clearly shown that the intermetallics grow faster when soldering with a smaller solder volume (small solder bump) compared with those formed in larger solder volume. The results are similar with all solder joints which is a clear trend that can be observed here, where the IMC layer thickness decreases with increasing solder size from 300µm to 700µm for ENEPIG and copper surface finishes. The dissolution phenomena of metals involved in the interfacial reactions are controlled by the ratio of solder volume to contact pad area [6]. With an increase in this V/A ratio the diffusion distance for Cu to saturate the liquid solder increases, thus resulting in slower interfacial reactions. This explanation could be applied in the present study. Even the metal pad used is different for all specimens and depends on the solder sizes, the ratio V/A still increases with increasing the solder ball diameter. This confirms the finding that with decreasing the
199
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
solder volume i.e; increasing the V/A ratio will lead to high intermetallics growth [7].
IMC Thickness VS Solder Bump Size
012345678
300(µm) 500(µm) 700(µm)Solder Bump Size (µm)
IMC
thic
knes
s (µm
)
0 (hr)500(hrs)1000(hrs)
Figure 2.a: IMC thickness for Pb-free solder on Cu surface finish
IMC Thickness VS Solder Bump Size
00.5
11.5
2
2.53
300(µm) 500(µm) 700(µm)
Solder Bump Size (µm)
IMC
thic
knes
s (µm
)
0 (hr)500(hrs)1000(hrs)
Figure 2.b:IMC thickness for Pb-free solder on ENEPIG surface finish
On the other hand, from the figures 2, it is found that the Cu substrates formed thicker intermetallics than ENEPIG. Figure 2.b shows that the intermetallics thickness has not shown significant increase after 1000hrs. This is attributed to the fact that after soldering the IMC on copper finish has scallop-type morphology with a rough interface. However, due to the effect of the ageing process the IMC has somewhat leveled off and became more continuous and exhibit a smoother interface [7]. IMCs produced in ENEPIG also decrease when the solder bump size increases. This is due to the existence of the Ni diffusion barrier layer between the solder and substrate. Since the reaction rate of Ni with liquid Sn solder is smaller than that of Cu, it acts to prevent the rapid interfacial reaction between solder and Cu conductor in electronic devices. This in turn, results in thinner intermetallic compounds. During reflow soldering, the molten lead-free Sn-4.0Ag-0.5Cu solder alloy dissolves the entire Au and Pd layer into the liquid solder, allowing Sn from the molten solder to react with the Ni layer to form Ni-Sn intermetallics. 3.2. Composition and Morphology of Intermetallic Compounds Analysis. During reflow soldering, the formation of the solder joint exists when there are interactions between liquid solder with the base metal, copper or nickel.
This interfacial reaction results in the formation of intermetallic compounds. Besides, separate samples of all categories were aged at 1500C for 500 hours and 1000 hours. In all these samples several phases were identified by optical and electron microscopy together with EDX. The solid-state reactions and growth of intermetallics during long term exposure to high temperatures are also important because the morphology, distribution and thickness of these intermetallics will affect the solder joint reliability. The morphologies of the various intermetallics formed in the solder joint during soldering and after solid state ageing were examined on all specimens using the deep etching of the solder. It is well known that soldering involves a reaction between molten solder and substrate, which dissolves some of the substrate and which forms some sort of intermetallic layer [2]. The interfacial chemical reactions enhance the wettability between the solder and the substrate. When tin-containing solder comes in contact with the copper pad surface, a layer of Cu–Sn IMC, consisting of the Cu6Sn5 phase adjacent to the solder and the Cu3Sn phase next to the copper land pad surface is formed in between and serve as the bonding material for solder joint. There is an extra IMC layer observed between the Cu6Sn5 and the Cu pad in all solder bump sizes studied after ageing for up to 500 hrs (Figure 3.b) and 1000hrs. This layer is confirmed by EDX analysis as Cu3Sn. No such Cu3Sn is observed at the interface after reflow in the cross sectional optical micrographs used in this study (Figure 3.a). In the case of smaller solder, the bulk solder is saturated with Cu quickly, Sn supply also is limited than Cu from the substrate, the Cu6Sn5 layer that formed first will transform into Cu3Sn. The same effect also happened for the other solder bump sizes which are the 500 µm and 700 µm.
(a)
(b)
Cu6Sn5
500 X Cu6Sn5
Figure 3; Cu3Sn and Cu6Sn5 IMCs formed between lead-free
solder and bare copper for 300 µm solder bump after (a) reflow soldering and (b) 500 hrs aging.
Cu3Sn 500 X
200
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
For the solder joint at Cu suface, there are Cu-Sn intermetallics compound at the interface as result of the diffusion of copper into the solder after reflow soldering. However, the dissolution rates of Cu along the Cu surface are not uniform. There are areas where all the Cu atoms have not been used in the formation of Cu6Sn5 IMC. This Cu surplus may lead to the formation of long tubes or fibres of Cu6Sn5. Figure 4 shows the IMCs at the top surface on Cu surface finish after reflow with different solder bump sizes. It is clearly shown that the same IMC morphology is observed for all solder bump sizes investigated. Thus, it can be concluded that difference in solder bump size does not have an effect on the type of intermetallic compound formed between copper surface finish and lead free solder.
(a)
(b)
(c)
Figure 4.5; Morphology of Cu6Sn5 formed between Sn-Ag-Cu solder and bare copper after reflow for difference solder sizes; (a)
300 µm, (b) 500 µm, and (c) 700 µm.
During isothermal aging, the IMCs continue to grow by inter-diffusion between the Sn in the solder and Cu in the substrate. This solid state aging resulted in a significant increase in the size of the Cu6Sn5 IMCs (Figure 5). Figure 5.b shows that after 500 hours aging, the IMC has coarsened and became more uniform and denser as well as the IMC morphology after 1000 hours instead of the scallop-type observed after reflow soldering. Moreover, after heating the solder joint, a new IMC phase has formed between the Cu substrate and Cu6Sn5 IMC as shown by the thin grey layer in Figure 3.b which is known as Cu3Sn. In the temperature range of interest (e.g. below 350 0C), the interfacial reaction with molten Sn-based solder results in the formation of Cu3Sn(ε) and Cu6Sn5(η) layers[8] (Figure 6). The same effect also happened with the solder bump sizes of 500 µm and 700 µm. In addition, there is the Ag3Sn IMCs which was observed in solder-Cu joint on the Cu6Sn5. The shapes formed vary but generally they were plate-like. After aging, Ag3Sn IMCs will become more spherical and smaller (Figure 5.b).
Cu6Sn5
(a)
Cu6Sn5
(b)
Ag3Sn
Cu6Sn5
Ag3Sn
201
Cu6Sn
4000 X
Cu6Sn
4000 X
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
(c)
Figure 5; Morphology on the top surface of IMCbetween Sn-Ag-Cu solder and bare copper for 30
bump with difference aging durations; (a) as reflowand (c) 1000 hrs
Figure 6: Cu–Sn binary phase diagram [
However, different phenomenon haENEPIG surface finish. In general, at tebelow 260 0C (reflow temperature is 217is the first phase to form at the liqconductor interface. The first stage of thethe dissolution of Ni in liquid solder, unsupersaturated with Ni. This is because thof gold and palladium dissolve into the bexposing the nickel layer to react with thform solder joint. In fact, there are mointermetallics were present at the inENEPIG Surface finish compared to thmaking it much more complicated to itypes of intermetallics that exist. This other than Sn, Ag and Cu from the soldpart in the formation of intermetallic after reflow. Lead free Sn-Ag-Cu solders are capabledifferent types of intermetallics, which arcircular boundary phases as shown in Ficontributing factor to the formation of thboundary regions is due to the differeproperties ranging from the edge to the c
solder bump. The different deep etching reaction at these regions also could be considered as one of the
Cu6Sn
s formed 0 µm solder , (b) 500 hrs,
9]
ppened to mperatures 0C) Ni3Sn4 uid Sn/Ni reaction is til solder is e thin layer ulk solder, e solder to re types of terface of
e Bare Cu, dentify the is because
er also take compounds
of forming e present in gure 7. The ese circular nt wetting
entre of the
contribution factor.
Inner Region
4000 X
n
Figure 7: Different morphologies of intermeta
circular boundary regions in lead free Sn-A
Several intermetallics were found in Ag-Cu solder joint which are (Ni, CCu6Sn5, and Ag3Sn. Neither goldcontaining intermetallics were founsolder joints. Although gold and padetected in EDX analysis, they arelow and are usually ignored in the dthe composition of intermetallic comp Generally, (Ni, Cu)3Sn4 and Ni3Sn4 of needle-like or rod shape afterblocky Ag3Sn intermetallics were above them at the centre region. The these intermetallics for the different sreflow are shown in Figure 8. It wAg3Sn were rarely found in the sol300 µm solder bumps. This is due smaller solder volume has relativelyin the solder and thus forming lessolder joint compared to that whisolder bump with higher solder volum
Generally, isothermal aging wmorphologies of intermetallic compoSn-Ag-Cu solder joint. Figure 9 (a)needle-like Ni3Sn4 were formedsoldering for sample using 300µmHowever, they changed to a morchunky shape after 500 hours aging t9 (b)). The intermetallics continuecoarsen when subjected to further ahours aging treatment. The same trend also applied to the so500µm and 700µm solder bumps whthe transformation of intermetallics fneedle-like shape to more compaspherical shape with aging duration.
Besides that, there is another intermformed at the outer region which is kThe same trend also applied at th
202
Outer Regio
llics which form the g-Cu solder joint.
the lead free Sn-u)3Sn4, Ni3Sn4 ,
nor palladium-d in lead free
lladium are also relatively very etermination of ounds.
, take the forms reflow. While usually formed morphologies of older sizes after
as observed that der joints using to the fact that
low Ag content s Ag3Sn in the ch uses bigger e.
ill affect the und in lead-free shows that the after reflow solder bumps. e spherical and reatment (Figure d to grow and ging from 1000
lder joints using ich experienced rom a loose and ct, coarse and
etallic that has nown as Ni3Sn.
is region where
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
from the needle shape of Ni3Sn after reflow to a more spherical and chunky shape after 500 hours aging treatment. The intermetallics continued to grow and coarsen when subjected to further aging from 1000 hours aging treatment (Figure 10). At circular boundary phases, it shows different intermetallic which is Cu6Sn5 (Figure 10.a).
(a)
(b)
(c)
Figure 8; Top surface morphology of intermetallics formed in centre region after reflow using (a) 300µm (b) 500µm (c) 700µm
solder bumps
(a)
(b)
(c)
Figure 9; Top surface micrographs showing morphologies of in Sn-Ag-Cu solder joint
using 300µm solder bumps at inner region
(a) after reflow (b) 500 hrs (c) 1000 hrs
(a)
Ag3Sn
Ag3Sn
203
Cu Sn
Ni3Sn4
6 5
Ni Sn
(Ni,Cu)3Sn4
(Ni,Cu)3Sn4
3 4
NMC 2007, Johor Bahru, Malaysia National Metallurgical Conference
[2] Wassink, R.J.K. (1997). Soldering in Electronics, Second ed., Scotland: Electrochemical Publications Ltd., p. 141.
(b)
4
(c)
Figure 10; Top surface micrographs showing morpSn-Ag-Cu solder joint
using 500µm solder bumps
(a) after reflow (b)500 hrs (c) 1000 hr 4. Conclusions In this study, the mean thickness of the compounds was found to be thicker in smvolume compared to larger solder volumother hand, ENEPIG surface finish prodintermetallic compounds compared to surface finish and the main type of compounds formed are Ni3Sn4 and (Ni,Cintermetallic compounds type and mormore or less the same for the diffevolumes investigated in this research.surface finishes, aging resulted in growthin terms of overall thickness, coaintermetallic compounds and also morphology of intermetallic compounspherical shape. Acknowledgments The authors would like to thank UniversMalaysia for providing research facilities References [1] Lau,S.T. (1994). Chip On Board Technologie
Modules.New York: Van Nostrand Reinhold.
Ni3Sn
[3] Dusek,M., Nottay,J., and Hunt,C. (2001).Compatibility of Lead-Free Solder with PCB Materials. UK: Centre National Physical Laboratory
[4] Johal,K. (2001) Are You in Control of your Electroless Ni/Immersion Gold Process?. Chicago: SMTA International
[5] Fassel, W. M. and friends (1966) Electroless Plating of
Metals (U.S. Patent 3264199)
[6] Schaefer,M., Laub,W., Fournelle, R.A., and Liang,J. (1997)
Proc. Design Reliability Solder Interconnects. p.247
2Intel Technology (M) Sdn. Bhd. Bayan Lepas Free Industrial Zone Phase 3, Halaman Kampung Jawa
1900, Penang, MALAYSIA
Abstract— Solder joint reliability is dependent on both thickness and morphology of the intermetallics that form and grow at the solder joint interface during soldering and subsequent thermal ageing and examining the morphology of these intermetallics is of great importance. The focus of this paper is to present experimental results of a comprehensive study of the interfacial reactions during soldering of Sn-Ag-Cu lead-free solders on copper (Cu), immersion silver (ImAg), electroless nickel/ immersion gold (ENIG) and electroless nickel/ electroless palladium/ immersion gold (ENEPIG) surface finishes. Using scanning electron microscopy detailed a study of the 3-D morphology and grain size of the intermetallics was conducted. The results showed that when soldering on ENIG and ENEPIG finishes several morphologies of intermetallics with different grain sizes form at the solder joint interface compared to a single intermetallic morphology that forms when soldering on copper and immersion silver. An attempt was made to discuss the effect of several factors that may have an influence on the type of morphology the intermetallics may grow into. The results obtained in the present investigation also revealed that the technique of removing the solder by deep etching to examine the morphology of intermetallics is a convenient and efficient method to investigate the intermetallics formed at the solder joints. Keywords: lead-free solder, surface finish, intermetallics, solder joint
1. Introduction
The demand has recently increased for new bump formation technologies which enable the simultaneous formation of large numbers of bumps with a narrow bump pitch at low cost and short tact
processing. However, some reliability issues may be arising from the utilization of smaller solder bump size. Due to its excellent conductivity and surface for soldering, copper has been widely used as the substrate materials. However, several types of metal coating must also be deposited on copper surfaces as board finishes for the purpose of providing wetting surfaces and protection against the environment. The selection of a metallurgical system (solder – top surface metallurgy) is very important because of its influence on the reliability of electronic assemblies. Typical surface finish metallurgy consists of two main layers: 1) a solderable layer in contact with the underlying copper and 2) a protective layer on top of the solderable layer. The purpose of the solderable layer is to provide the surface to which the liquid solder wets and then adheres upon solidification. This same solderable layer also acts as the diffusion barrier by preventing diffusion of the solder to the copper substrate. The protective layer serves to protect the solderability of the solderable layer from degradation due to exposure to ambient environment until reflow soldering occurs. During reflow the solder melts and the protective layer dissolves into the molten solder exposing in the process the solderable layer to the molten solder. The solderable layer now is also subjected to dissolution by the molten solder until solidification is complete. This results in the formation of an intermetallic layer between the solderable layer and solder. This intermetallic layer will grow in thickness during subsequent thermal ageing after assembly due to solid – solid reaction between the solderable layer and the solder by solid-state diffusion.
In this paper we present results of an experimental investigation to illustrate the importance of surface finish metallurgy on the type and morphology of
IEMT 2008, Penang, Malaysia International Electronic Manufacturing Technology
intermetallics formed during soldering with SAC solder.
2. Experimental Details Four different surface finish metallurgies were selected for this study: bare copper (Cu), immersion silver (ImAg), electroless nickel/ immersion gold (ENIG) and electroless nickel/ immersion gold (ENEPIG). The ImAg, ENIG and (ENEPIG) surface finishes were deposited on a free-oxygen pure copper substrate with the dimensions (width x length x thickness) of 45 x 50 x 1 mm. The copper substrate was first subjected to a pretreatment process to remove the oxide and activate the copper surface before the desired finish layers are deposited. Prior to soldering, a dry solder mask was laminated onto the plated substrates using a laminated machine. Then, the solder mask was exposed to the ultraviolet (UV) light through a patterned film. The exposure is to ensure an array of pads is made onto the solder mask upon the subsequent development stage in the developing solution. The substrates were then populated with Sn-4Ag-0.5Cu (SAC405) solder spheres. The solder spheres were arranged in several rows and bonding to form the solder joints was made by reflow in a furnace with the peak reflow temperature set at 250 oC. Before soldering, all substrates were treated with a no clean flux to remove surface oxide. In order to reveal the morphology of intermetallics formed during the soldering process a useful method of selective chemical etching of the top surface was employed and examination of the intermetallics was made by means of scanning electron microscopy. Energy dispersive x-ray (EDX) was used to identify the type and composition of intermetallics formed.
3. Results and Discussion
Thermodynamics is usually used to describe the intermetallics which form at the interface between metal pads in the surface finish and liquid Sn-based solders and thus phase diagrams are important to explain why such an intermetallic can form or not. The type of intermetallic formed depends on the surface metallurgy used. It is well established that when soldering on bare Cu and ImAg, Cu6Sn5 IMC is formed during reflow, while (Cu, Ni)6Sn5 is formed when soldering on ENIG and ENEPIG.
As the SAC solders/ immersion silver is reflowed the Ag layer will dissolve into the molten solder exposing the Cu substrate to the solder. Once the
solder is melted, the Sn in molten solder will quickly react with the solid Cu and develops small nuclei which quickly grow and form a scallop like islands of the IMC at the interface. Figure 1 and Figure 2 show the top surface morphologies of the Cu6Sn5 IMC formed on Cu and ImAg finishes respectively. It is quite clear when observing the grain size of the scallops that thinner IMC is formed on ImAg compared to when soldering on Cu. It is important to note that the morphology and indeed size of the scallop-like IMC observed in Figure 1 and Figure 2 was uniform over the entire surface of the pad area as shown in Figure 3.
Figure 1. Scallop-type Cu6Sn5 IMC formed on Cu after reflow
Figure 2. Scallop-type Cu6Sn5 IMC formed on ImAg
after reflow
206
IEMT 2008, Penang, Malaysia International Electronic Manufacturing Technology
Figure 3: Scallop-type Cu6Sn5 IMC formed on Cu
As observed above in both Cu and ImAg finishes, the interfacial reaction occurs between the liquid Sn solder and Cu in the substrate pad. However, an interesting observation made in the present study is the presence of a ring-shape of the intermetallic layer found during soldering between the SAC solder/ENIG and to a lesser degree between SAC solder/ENIG couples. Figure 4 shows the top view morphology of the intermetallic layer formed between SAC405 and ENIG finish. It is interesting to notice the presence of a ring-shape with varying sizes of intermetallics. EDX analysis confirmed that the intermetallic is still (Cu, Ni)6Sn5. Figure 4: Ring-shape of intermetallics near the edge of the ENIG pad. Figure 5 and Figure 6 show more examples of the various sizes of the (Cu, Ni)6Sn5 intermetallic on ENIG and ENEPIG surface finishes respectively.
Figure 5: Different grain sizes of (Cu, Ni)6Sn5 formed on ENIG finish Figure 6: Different grain sizes of (Cu, Ni)6Sn5 formed on ENEPIG finish Previous researchers [2, 3] also observed the same pattern when soldering with SAC solder on ENIG but no such ring pattern was observed when soldering with Sn-Pb solder alloy. The reasons for the formation of such a pattern of varying grain sizes of the intermetallic phase and its possible effect on the
207
IEMT 2008, Penang, Malaysia International Electronic Manufacturing Technology
solder joint reliability is not clear. Several factors may have an influence on such a scale of intermetallic morphology change. The elements involved in the interfacial reactions are the same: Sn, Ag, and Cu except that when soldering on ENIG and ENEPIG there is Ni involved. It is believed it is probably due to the wetting properties of the SAC lead-free solder composition ranging being different over the pad area. Another factor is the non-homogeneity of the solder alloy itself since unlike the Sn-Pb alloy, the SAC solders contain three main elements, Sn, Cu and Ag. Yoon et al. [4] claimed that this situation occurred probably due to discrepancy of Cu content in the solder matrix between the centre and edge during solidification. However, the authors believe that the formation of these different intermetallic grains with varying sizes may also be correlated with the solder sphere contact to the substrate metallization pad. In the Sn-Pb / ENIG system, for example, the interfacial reaction occurs mainly between liquid Sn and Ni whereas with SAC solder/ ENIG system, both Cu (from the solder) and Ni (in the surface finish) are involved in the reaction. Since the Ni-Sn reaction is slower than that between Cu-Sn, and coupled with varying wetting over the pad area, this may lead to faster growth of Cu-Sn based intermetallic in areas where there is more Cu supply. It has been reported by Laurila et al. [5] that the dissolution rates of Cu along the Cu pad surface are not necessarily uniform. The fact that even when using SAC solder on Cu and ImAg finishes (no Ni is involved in these systems) there was only one uniform intermetallic layer over the entire pad area, is evidence that competitive reactions between Cu and Ni with Sn may have an influence on the morphology and size of grains formed.
4. Conclusions
The results showed that the type of surface finish metallurgy used has a strong influence on the morphology and size of intermetallic grains formed during soldering with SAC-based solders. Having both Ni and Cu in the surface finish/ solder system may lead to competitive growth and hence results in varying grain sizes of the intermetallic layer. The results also revealed that the technique of removing the solder by selective etching to examine the morphology of intermetallics is a convenient and efficient method to investigate the intermetallics formed at the solder joints.
References
[1] M. Schaefer, W. Laub, R.A. Fournelle, and J. Liang, Proc. Design Reliability Solder Interconnects, eds, R.K. Mahidhara,
[2] Tan, C.L., M. Eng. Thesis, Universiti Teknologi Malaysia, 2006
[3] Azmah. H. A., PhD Thesis, Universiti Teknologi Malaysia, 2007
[4] Yoon, J. W., Kim, S. W. and Jung, S. B. Intermetallic Morphology, Interfacial Reaction and Joint Reliability of Pb-free Sn-Ag-Cu Solder on Electrolytic Ni BGA Substrate”. Journal of Alloys and Compounds, 2004, 392, 247-252
[5] Laurila, T., Vuorinen, V. and Kivilahti, J. K. Interfacial Reactions between Lead-free Solders and Common Base Materials. Materials Science and Engineering,, 2005, R49: 1-60.
208
IGCES 2008, Johor, Malaysia International Graduate Conference on Engineering and Science
EFFECT OF SOLDER VOLUME AND PAD AREA ON INTERMETALLIC
COMPOUNDS FORMATION DURING SOLDERING BETWEN Sn-4Ag-0.5Cu AND IMMERSION SILVER FINISH
I. Siti Rabiatull Aisha, A. Ourdjini and A. Astuty, O. Saliza Azlina
Faculty of Mechanical Engineering, University of Technology Malaysia
Abstract Recently, silver as an electrochemical deposit on copper substrate has attracted much attention in the microelectronics field. To deposit silver particles on copper, immersion plating is used and it was characterized by field emission microscope (FESEM), energy dispersive X-ray analysis (EDX) and also image analyzer to determine the thickness of the silver plating. Apart from that, this paper also examines various sizes of Sn-4Ag-0.5Cu lead free solders which are Ø200µm, Ø300µm, Ø500µm and Ø700µm. With different solder joint sizes, the dissolution rate of top surface metallurgy (TSM) and intermetallic compound (IMC) growth kinetics will be different. The effect of solder volume/ pad metallization area (V/A) ratio on IMC growth was investigated during reflow soldering and solid state ageing. Higher V/A ratio produced thinner and more fragmented IMC morphology while lower V/A ratio produced better defined IMC layer at the interface after reflow soldering. After ageing at 150oC for up to 2000 hours, the initial scallop morphology of the Cu6Sn5 IMC layer changed to that of a more planar type and also became thicker. Besides, another IMC layer formed after ageing which is Cu3Sn. Several techniques of materials characterization including optical microscope, image analysis, scanning electron microscopy and energy dispersive X-ray analysis were used to examine and quantify the intermetallics in terms of composition, thickness and morphology. Keywords: intermetallics, lead-free solder, surface finish, immersion silver, soldering