Comparative Study of NiNiP Leadframes from Different Processes

Transcription

1 Comparative Study of NiNiP Leadframes from Different Processes Wu-Hu Li *1, Jeffrey Khai Huat Low 1, Harry Sax 2, Raymond Solis Cabral 1, Esperidion De Castro Salazar 1, Pauline Min Wee Low 1 1 Infineon Technologies Asia Pacific Pte Ltd, 168 Kallang Way, Singapore , Republic of Singapore 2 Infineon Technologies AG, Wernerwerkstrasse 2, Regensburg, Bavaria 93049, Germany * Corresponding author. address: wu-hu.li@infineon.com, Tel: ; Fax: Abstract In this study, Ag/NiNiP-plated Cu leadframes from two different processes are compared in terms of NiNiP surface morphology, composition of NiP layer, grain structures of Ni and NiP plated layers, and impurities in the interface of Ni/NiP plated layers. Their impacts were assessed after solder plating and baking. Scanning electron microscopy (SEM) shows that the grain size of Leadframe A (from Process A) is relatively larger than that of Leadframe B (from Process B). The Auger electron spectroscopy (AES) analysis shows both leadframes have comparable P-content (typically around 12.5% by atom or 6.99% by weight) in their NiP layer after 20 nm sputtering. The focused-ion-beam (FIB) cut shows that the Ni layer of Leadframe A shows both big and small grains; the Ni layer of Leadframe B shows more uniform intermediate grain sizes. However, the grain structures of their NiP layers are very different. The NiP layer of Leadframe A appears amorphous and is clearly distinguishable from the Ni layer, but the NiP layer of Leadframe B looks grainy and its interface with the Ni layer is not very distinguishable. The depth profile of the time-of-flight secondary ion mass spectroscopy (TOF-SIMS) shows that there are higher copper, potassium and chloride impurities in the Ni and NiP interface for Leadframe A while those impurities in Leadframe B are much lower. The different characteristics of plated layers observed in the two leadframes are related to the chemical systems and process parameters used by the two leadframe processes. The impacts of these differences were detected after solder plating and post-plating baking. Leadframe B is able to pass 100 hours of baking at 155 C without any observed abnormality. By contrast, Leadframe A is only able to sustain 30 hours of baking at 155 C. Further baking on Leadframe A causes the NiP layer to peel from the Ni layer of the leadframe. This study demonstrates that the quality of NiNiP plated layer is very critical to the adhesion of Ni and NiP interface which might affect the package performance in the later stage. Further and deep study is needed to understand the failure mechanism of NiP peeling from the Ni layer after the baking test. 1. Introduction There are a few common types of leadframes in the market; silver (Ag)-plated copper (Cu) leadframes are the most common leadframes used for gold (Au) wire bonding devices in IC packages [1]. Nickel palladium (NiPd) based pre-plated leadframes (PPF), such as NiPdAu, NiPdAu-Ag, and NiPdAu-Pd have been introduced as lead (Pb)-free alternatives with no tin (Sn)-whisker risk. Nickel, nickel phosphorus (NiNiP) leadframes are mainly used for aluminum /10/$ IEEE 255 wire bonding. Aluminum wire bonding is used extensively in the electronics industry as an interconnection in hybrid integrated circuits (IC) and automotive applications. The NiNiP/Cu leadframe is produced by first plating a Ni layer on a Cu substrate, followed by a NiP top layer which is in contact with the environment. The quality of the NiNiP surface is critical for overall package performance. For example, the surface roughness, hardness, surface cleanliness and phosphorus concentration of the NiNiP surface affects the aluminum wire bonding processability [2]. The shelflife of the NiNiP leadframe has been reported by our group recently as having an impact on its package reliability [3]. It is reported that the NiNiP layer also affects the package reliability after stress tests due to their influence on the adhesion of molding compounds, and the package solderability due to the intermetallic compound (IMC) formation between the NiNiP surface and solder after heating [4 8]. The mechanism of Ni diffusion from NiP layers after heating in solders is widely discussed in the literature [4 8]. The diffusion of Ni into solder forms Ni-Sn IMC, such as Ni 3 Sn 4 [4, 5], Ni 3 Sn [5] and Ni 3 Sn 2 [5]. A P-rich Ni layer [5, 6], such as Ni 3 P [4, 7, 8], is also formed as a by-product of the Ni-Sn reaction between Ni-Sn IMC and the NiP layer. The rate of Ni diffusion in the NiP layer depends on the solder system, time, temperature applied and the thickness of the NiP coating. In this study, two Ag/NiNiP-plated Cu leadframes from two different processes are compared in terms of their surface morphology, composition, grain structures of Ni and NiP plated layers, and impurities in the NiNiP plating layer. Their impact on the assembly process was evaluated. 2. Experimental Procedure 2.1. Leadframe Material Ag/NiNiP-plated leadframes used in this study were obtained from two leadframe processes. The base material of the leadframe was a piece of Cu sheet with a thickness of 0.25 mm. A first full layer of Ni with a thickness of ~ 2 µm was electroplated on the Cu surface and a second full layer of NiP with a thickness of ~ 0.2 µm was electroplated on the first Ni surface to foster aluminum wire bonding during IC assembly. A third layer of Ag with a thickness of ~ 6 µm was selectively electroplated at the top of the die attachment paddle area of the leadframe to cater to die attachment during IC assembly. As a result, the whole leadframe was NiNiP-plated except the top side of the die attachment paddle which is Ag-plated. For comparison, leadframes from Process A and Process B were used. Our focus for this study is on the NiNiP plated-surfaces and layers.

2 2.2 Surface Analysis: To better understand the Ni and NiP plated layers, a few analytical techniques, such as scanning electron microscopy (SEM), Auger electron spectroscopy (AES), focused-ionbeam (FIB) cut, time-of-flight secondary ion mass spectroscopy (TOF-SIMS) and energy dispersive X-ray (EDX), were employed to analyze the plated layers and interfaces. leadframes are comparable, namely 12.4% and 12.7% for Leadframes A and B, respectively. 3. Results and Discussion 3.1 Surface Morphology of NiNiP Layer SEM on Surface The pictures in Fig. 1 show the surface morphology of NiNiP surface for both leadframes. In the top two pictures with lower magnification ( 2k), it shows that the surface morphology of NiNiP layer follows the texture direction of the Cu substrate in a horizontal direction. This is because the total NiNiP layer is very thin (about 2.2 µm), the morphology of NiNiP-plated layer follows the morphology of its Cu substrate. There are some horizontal lines observed in both pictures. The horizontal lines for Leadframe A were fine and uniform while the horizontal lines for Leadframe B are deeper. This is mainly due to the Cu substrate used before NiNiP plating, which might be different from one leadframe supplier to another leadframe supplier. The bottom two pictures with higher magnification ( 10k) show that the grain size of the NiNiP layer of Leadframe A is larger than the grain size of the NiNiP layer of Leadframe B. The boundaries of the grains in Leadframe A are also deeper and wider compared with Leadframe B. This is consistent with what was observed in a cross-section of FIB cut in Section 3.3. Leadframe A Leadframe B Fig. 2. AES spectra of NiNiP surface at 20nm depth: Leadframe A and Leadframe B 3.3. Grain Structure of NiNiP Layer FIB cut Fig. 3 shows the cross-section of FIB cut of the two leadframes at their external leads which is NiNiP plated. The top two pictures were obtained with lower magnification (about 13.5k). It shows clearly that there is an Ni layer on the Cu substrate. Leadframe A has both big and small grains in the Ni layer. Leadframe B has intermediate grain sizes in its Ni layer. Its grain size is much more uniform compared with Leadframe A. However, the NiP layer cannot be seen clearly in these two pictures since their magnification is too low. The bottom two pictures were obtained with higher magnification (about 67k). It shows that, for Leadframe A, its NiP layer appears amorphous and clearly distinguishable from its Ni layer. There are some pitting observed in the top of the NiP layer. The NiP layer in Leadframe B looks grainy in the FIB cut and is barely distinguishable from its Ni layer. Leadframe A Leadframe B Fig. 1. SEM of NiNiP surface: Leadframe A and Leadframe B 3.2 Composition of NiP Layer AES at 20nm Depth The AES of both NiNiP surfaces shows C, O, Ni and P elements at their surfaces. In order to measure P% more precisely, the samples were sputtered until 20nm depth before AES measurement. The spectra in Fig. 2 show typical AES spectra of NiNiP leadframe surfaces after 20nm sputtering for both leadframes. Both spectra show there are only Ni and P elements detected at 20nm depth. The P% (by atom) in both 256 Leadframe A Leadframe B Fig. 3. Cross-section of FIB cut of NiNiP plated layer on leadframe: Leadframe A and Leadframe B. The crystal structure resulting from an electrodeposition process is strongly dependent on the relative rates of formation of crystal nuclei and the growth of existing crystals. Generally, a decreasing crystal size is the results of factors which increase the cathode polarization. Plating processes have various variables that influence structure, e.g., chemical system used, metal ion concentration, addition agents, current density, temperature, agitation, and polarization. The difference observed here could be attributed to the different

3 plating processes and controls in both chemicals and plating parameters Impurities in NiNiP Layers TOF-SIMS Depth Profile To better understand the difference observed in the crosssection of FIB cut, the TOF-SIMS depth profile was employed to analyze the chemical composition of the NiP layer and the Ni/NiP interface as a function of depth (Fig. 4). A few species, such as Na, Mg, Al, Si, K, Ca, PO, Fe, Ni and Cu in positiveion mode, and H, C, N, O, P, S, Cl, Ni, NiH and NiP in negative-ion mode were detected in the depth profile. This shows that the thickness of the NiP layer (around 350s sputter time) in Leadframe A is thinner than that of the NiP layer (around 850s sputter time) in Leadframe B in both negativeion and positive-ion modes. The depth profile of negative-ion mode of TOF-SIMS shows that there are two major peaks for S and Cl at the Ni/NiP interface for Leadframe A, while one smaller peak for S was observed at the Ni/NiP interface for Leadframe B. The depth profile of positive-ion mode of TOF- SIMS shows a decreasing trend for all species observed for Leadframe A as the sputtering went deeper, while the trends for Leadframe B were different. In order to have a better comparison between Leadframe A and Leadframe B, the relative intensity of depth profile of species was normalized to Ni intensity in respective spectroscopy with positive-ion mode. It was found that, as shown in Fig. 5, the relative intensities of Cu and K depth profile in Leadframe A were much higher than the relative intensities of Cu and K depth profile in Leadframe B. This implies that there are higher Cu and K impurities in Leadframe A compared with Leadframe B. As a summary, the depth profile of TOF-SIMS shows that there are higher Cu, K and chloride impurities detected in the Ni/NiP interface for Leadframe A while those impurities in Leadframe B are much lower. Leadframe A Leadframe B Fig. 5. Depth profile of relative intensity of Cu and K over Ni intensity in respective spectroscopy with positiveion mode (Leadframe A vs. B) 3.5. The Impact of NiNiP Plating Chemistry and Process Parameters It is known that there are a few types of Ni electro-plating bath chemistry in the market, such as sulphate nickel, chloride nickel, Watts nickel and sulphamate nickel etc. [9]. The impacts of different plating chemistries and plating currents on surface morphology, crystal size, and texture have been reported in the literature [9]. It was also reported that a standard Watts type nickel solution can be modified to produce NiP by introduction of an additive to provide the phosphorus for co-deposition. [10]. The impact of the concentration of the plating solution, current density, cathode efficiency, and temperature on the phosphorus content in the deposit, elongation, hardness, internal stress, resistance to Cu diffusion, magnetic properties, corrosion resistance and deposition structure have been studied in the literature [10]. However, we are not able to provide more details on the chemistry of Ni and NiP plating baths, and their key parameters used at these two processes since these are suppliers confidential information. Their plating chemistry and key parameters have contributed to the differences in terms of grain structures and impurities in plating layers discussed in previous sections. Fig. 4. Depth profile of TOF-SIMS of NiNiP plated layer on Leadframe A and Leadframe B: Negative-ion mode and Positive-ion mode The Impact of NiNiP Layer on Package Performance Bake Tape Test after Solder Plating In order to check the impact of the NiNiP layer on package performance, wire-bonding data were collected and compared. There is no significant difference observed for these two types of leadframe in terms of wire-bonding performance. To check the adhesion of the NiNiP layer to a solder layer, a bake and tape test was developed. The leadframe was plated with Sn first. After Sn plating, the leadframe went through a post-plating baking heat treatment at 155+/-5 C for a certain number of hours. The leadframe was then formed at external leads per normal assembly process. After that, a strip of plastic tape was placed across the bent Sn/NiNiP-plated area and then the tape was peeled quickly from the plated surface. The results of this bake and tape test for the two types of leadframes are summarized in Table 1. To our surprise, peeling of plated layers was detected for Leadframe A after 50 hours baking, while there was no abnormality observed for Leadframe B even after baking for 100 hours. Fig. 6 show a typical example of both good and bad samples. For the good sample (Fig. 6), there is no residue left on the tape side after

4 the tape was peeled from the leadframe surface. For the bad sample (Fig. 6), some flakes were observed on the tape side after the tape was peeled from the leadframe surface. Table 1. Bake and Tape Test Results Baking at 155 C for Various Periods Bake Time (hours) Leadframe A Pass Pass Pass Fail Fail Leadframe B Pass Pass Pass Pass Pass Good unit Bad unit Leadframe Side Tape Side Fig. 6. Examples of bake and tape test, the left picture is of the leadframe side and the right picture is of the tape side: Good unit and Bad unit. To further confirm the peeling interface, EDX was performed to check the elements on the leadframe side and the flakes on the tape side for the peeled sample. It was found that there was only Ni element detected on the leadframe side (Fig. 7). While there were Ni and P elements detected at the flakes on the tape side (Fig. 7). This indicates that the peeling happened at the Ni and NiP interface instead of the NiP/solder interface. This failure at the interface is quite different from what is normally observed during Sn plating on NiNiP leadframes, in which Sn peeling is observed between the plated Sn layer and the leadframe top layer (NiP) interface. This failure is more related to the different NiP and Ni layers and their NiP/Ni interfaces discussed in previous sections. Fig. 7. Typical EDX spectra of peeled sample: at leadframe side and at the tape side. Conclusions The comparison of two leadframes presented throughout this work shows that the NiNiP layer of the leadframe is very critical for the overall package performance. The adhesion between the Ni and NiP layers might peel off after certain temperature and duration if its chosen plating chemistry and parameters are unsuitable. The observations made in this study raise a warning that an NiNiP leadframe surface might cause some potential problem in package reliability in board level. Extra measurements and controls should be in place for the leadframe, to reduce its potential risk. Further investigation is needed to understand the details of the failure mechanism of the NiP peeling after baking. Acknowledgments We would like to thank Dr. Jordan Steffen, Dr. Engl Reimund and Dr. Thomas Sven for helpful discussions. The support from the analysis laboratories and the two leadframe suppliers is also greatly appreciated. References 1. Tracy D.P. and Vardaman E.J., Global Semiconductor Packaging Materials Outlook, (Austin, TX: SEMI and TechSearch International, 2007). 2. Onuki J., Koizumi M. and Yoshioka O., Bonding Strength between Al-Wires and Ni-P Plated Lead Frames, Mater. Trans. JIM, 34 (1993), pp Li, W. H., Shelf-life study of Ag/NiNiP-plated Cu leadframes without anti-tarnish coating, IEEE Transactions on Electronics Packaging Manufacturing, accepted. 4. Jeon Y.D., Paik K.W., Bok K.S., Choi W.S. and Cho C.L., Studies on Ni-Sn intermetallic compound and P- rich Ni layer at the electroless nickel UBM solder interface and their effects on flip chip solder joint reliability, Electronic Components and Technology Conference, Proceedings. 51st, (2001), pp Islam M.N., Chan Y.C., Alam M.O. and Sharif A., Comparative Study of the Dissolution Kinetics of Electrolytic Ni and Electroless NiP Layers by Molten Sn3.5Ag Solder Alloy, J. Electr. Packag., 127 (2005), pp Alam M.O., Chan Y.C. and Hung K.C., Reaction kinetics of Pb-Sn and Sn-Ag solder balls with electroless Ni-P/Cu pad during reflow soldering in microelectronic packaging, Electronic Components and Technology Conference, Proceedings. 52nd, (2002), pp Goyal D., Lane T., Kinzie P., Panichas C., Chong K.M. and Villalobos O., Failure mechanism of brittle solder joint fracture in the presence of electroless nickel immersion gold (ENIG) interface, Electronic Components and Technology Conference, Proceedings. 52nd, (2002), pp

5 8. Alam M.O., Chan Y.C. and Hung K.C., Reliability study of the electroless Ni P layer against solder alloy, Microelectron. Reliab., 42 (2002), pp Watanabe T, Nano-plating: Microstructure Control Theory of Plated Film and Data Base of Plated Film Microstructure, (Elsevier, 2004), pp Luke, D.A., Nickel phosphorus electrodepositions, Trans IMF, 64 (1986), pp