See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/271553283 Effects of soft solder materials and die attach process parameters on large power semiconductor dies joint reliability Conference Paper · December 2013 DOI: 10.1109/EPTC.2013.6745703 CITATIONS READS 5 4,626 3 authors: Vemal Raja Manikam Samsun Paing Bosch Infineon Technologies, Melaka, Malaysia 17 PUBLICATIONS 577 CITATIONS 3 PUBLICATIONS 8 CITATIONS SEE PROFILE Amy Ang Infineon Technologies 4 PUBLICATIONS 17 CITATIONS SEE PROFILE All content following this page was uploaded by Vemal Raja Manikam on 29 December 2015. The user has requested enhancement of the downloaded file. SEE PROFILE Effects of soft solder materials and die attach process parameters on large power semiconductor dies joint reliability * Vemal Raja Manikam, Samsun Paing and Amy Ang Power Segment Development, Infineon Technologies (M) Sdn Bhd FTZ Batu Berendam, 75350 Malacca, Malaysia * vemal.rajamanikam@infineon.com Abstract Die attach utilizing soft solders, primarily high lead (Pb) solders for power semiconductor devices are still widely used across the industry. These solders offer excellent mechanical, thermal and electrical properties. Soft solders provide good heat dissipation of typically 35 W/mK, is ductile and exhibits desirable thermal expansion properties which tolerate that of silicon (4.2 ppm/K) and copper (16.5 ppm/K), so as to avoid joint failure due to thermal cyclic stresses. Typically, for large dies, bond-line-thickness (BLT), die tilt and post-reflow void rates are important. These factors affect the reliability and performance of the power device. Large, localized voids create air pockets which are detrimental to the device’s reliability, both electrically and thermally. The experimental work conducted here is on a large power semiconductor die having an area of 19mm2 and Ag coated die back, using a PbSn2-Ag2.5 Type 3 soft solder on bare copper (Cu) substrates. For large dies having unequal width and length, the flow of solder when bond pressure is exerted and released becomes crucial as well as challenging. It directly affects the BLT. When reflowed, most solders undergo flux evaporation which leads to hot slump of the die attach solder, which in turn affects the die tilt and final outcome of void rates between the die and substrate. Here, surface energy concepts and how well the solder material flows and wets both the die back and substrate surfaces need to be understood. Die back and substrate surfaces exert varying stresses before and after reflow. In this literature work, we examine the fundamental concept of impact bonding utilized for soft solders on large power devices, factors which govern the formation of an acceptable bond line thickness, as well as influence of surface conditions for solder paste wetting and spreading during reflow in order to gain a reliable die-to-substrate interconnection. Introduction Pb soft solder materials have formed a stable base amongst die attach materials. For power device technology utilizing Sibased devices, the operation temperatures range between 200250˚C. One of the most used Pb soft solder materials is PbSn2-Ag2.5. Despite the high Pb content, it provides good thermal, electrical and mechanical properties to form a reliable joint between the die and substrate. Generally, with increasing Pb content in a solder material, the liquidus point increases proportionally [1]. Silver (Ag) is added to aid in surface wetting properties of the solder material [2]. The Pb-Sn2-Ag2.5 solder material has a solidus point of 299˚C, and liquidus point between 304309˚C [2]. Having a liquidus point which is higher than the operational temperature of a device is important for die attach, in particular power devices. Fig.1 lists the common die attach materials used in the industry and academia over the years by categories, i.e. their liquidus temperature. The Pb-Sn2-Ag2.5 solder falls into the medium range, which cover majority of the common power device applications. Impact bonding is a common method for soft solder die attach, whereby the solder material is dispensed onto a substrate, then die is then placed on the dispensed solder followed by some exertion of pressure to enable the solder material to come into contact with the die back. The die-die attach solder-substrate structure is then sent for reflow processing, whereby the solder material undergoes melting to create a joint between the die and substrate. During reflow, the solder material typically experiences phase changes, which in turn affects the outcome of the soldered part, and subsequently its reliability. Such material-process interactions are important, and become critical for larger dies, especially ones with irregular width and length. Concerns would be on die tilt, void rate and final BLT. In Fig. 2, it can be seen that for a die with unequal width and length, the material flow would be more critical as it would differ along the longer and shorter paths. Fig. 1: Various die attach materials and solders, their operation temperature ranges and application possibilities by category. The Pb-Sn2-Ag2.5 solder falls into the medium range [1]. x y x B A y A: x=y B: x ≠ y Fig. 2: Comparison of die size width and length; A for equal dimensions and B for unequal dimension. Die attach impact bonding process with regards to large dies Die attach is challenging for much larger dies, as several factors need to be considered simultaneously, during setup of the die bond equipment. These challenges become critical for large dies with irregular width and length. In our assessment, we used a power semiconductor die having an area of 19mm2. The die was run using a typical die bond equipment, using PbSn2-Ag2.5 solder dispense followed by die placement with pressure being exerted as the die was placed onto the dispensed solder area. Fig. 3 illustrates the impact bonding process on the Pb-Sn2-Ag2.5 solder material [3]. In Fig. 2a, the solder paste is dispensed, followed by die placement with a bond force exerted onto the die top. In Fig. 2b, the movement of the die creates a “volcano effect” which relates to the flow of the solder material out from the bulk solder volume which has been dispensed [3]. Subsequently, in Fig. 2c, the thin crevice created by the die and substrate clearance enables capillary forces to act, whereby the “volcanoes” collapse and the solder volume flows back between the die and substrate. This in turn forces the die upward, using what we can call a spring-back force, which forms the bond-line-thickness. For larger dies, the flow of die attach material is important, in particular at the edges. If the solder material’s flow back is poor with regards to bond force, incomplete coverage will take place. It would lead to corner delamination issues. Fig. 4 depicts such a scenario with incomplete solder coverage for a large die, followed by a through-scan image in Fig. 5. The mechanism presented in Fig. 3, followed by failures indicate how severe the die attach process itself is, as it promotes and controls the material flow back to achieve acceptable solder material coverage on the die back. In subsequent sections, the mechanism presented here for impact bonding will also be discussed with regards to surface energy of both the die back material and substrate. a b c d Fig. 3: Schematic diagram of impact die attach bonding for the Pb-Sn2-Ag2.5 soft solder. a. a. b. b. Fig. 4: Optical images: a.) Incomplete solder at die edge; b.) Poor solder volume flow at edges, unable to form fillet height. Fig. 5: Through-scan images for post-reflowed dies. The arrows indicate delamination occurrences at the die edges. (i) Assessment of die attach solder material during and after reflow Most die attach solder materials composition play an important role in obtaining good BLT and die tilt. These are inherent properties of the solder material which translate to performance indicators of the die attach process itself. Any solder material constitutes of solder particles and fluxes. The 2 components combine to yield an easily handled solder paste [4]. Solder deposition on a substrate affects bond-linethickness value. Solder volume fraction, V, is a representation of the solder volume being deposited, whereby in this experimental work would be via dispensing as shown in Eq. 1 [4]: In Eq. 1, x represents the metal content (%w/w), ds is the density of the solder alloy, and df is the density of the solder flux. By obtaining the solder alloy and flux densities, we are able to generate a relationship for solder volume fraction and metal weight percent content in the solder, as depicted in Fig. 6. In our experiment, the Pb-Sn2-Ag2.5 solder paste had 90% solder alloy weight content. When converted to volume fraction, this amounts to approximately 44.6%. This would mean, during solder paste dispensing for die attach, a solder alloy volume fraction of 44.6% only is present, and the balance 55.4% would constitute of flux material. This flux material would evaporate and burn off during reflow. Therefore, the bond-line-thickness theoretically would be reduced after reflow, and can affect the die tilt properties too. To demonstrate this concept, a simple experiment was conducted. The Pb-Sn2-Ag2.5 soft solder was dispensed on Cu substrates and 5 specific points were measured before and after reflow. As mentioned earlier, the maximum reflow temperature was 400˚C, with a ramp rate of 2˚C/sec. A total of 7 units were measured and compared. The statistical analysis is shown in Fig. 7. There was a significant collapse in material height after reflow, which was estimated at about 70% of the original thickness. This fits the earlier theoretical assumption presented in Fig. 6. The loss of volume after reflow is crucial in establishing good bond line thickness, especially for large dies which could yield poor die tilt results as well. Therefore, controlling the bond height and its impact during die attach of these solders is important, when the loss of solder materialpost reflow is taken into consideration. PbSn2Ag2.5 volume versus weight % for metal particles The relationship between these 3 interfacial tensions can be represented by Eq. 2, whereby cos θ is the contact angle of the molten solder droplet on the substrate [4,5]: PbSn2Ag2.5 volume % 100 80 60 44.6% 40 20 0 20 40 60 80 90% 100 PbSn2Ag2.5 weight % Fig. 6: Volume fraction percentage relationship with solder alloy weight percent content for Pb-Sn2-Ag2.5 solder paste. 1 4 5 3 1 4 5 2 3 Pattern height before reflow Pattern height after reflow 2 Fig. 7: Oneway analysis of dispensed pattern height (micron) by measured points for the Pb-Sn2-Ag2.5 solder paste before and after reflow. (ii) Influence of die back and substrate surfaces on solder spreading and wetting Molten solder spreading and wetting is an important criteria towards achieving joint reliability for die attach [5]. If the material is unable to interact with the contact surfaces, a strong bond cannot be achieved. During solder reflow, the interaction between die back and substrate surfaces can have an impact on bond-line-thickness formation. This relationship will be discussed with regards to surface energy as well for the contact surfaces. To begin with, the interaction for a molten solder droplet on a Cu substrate can be represented by Fig. 8. In Fig. 8, using Young’s equation, the solder droplet undergoes 3 different interfacial tension; ΦLG or liquid-gas interfacial tension, ΦSG or solid-gas interfacial tension and finally ΦSL or solid-liquid interfacial tension. ΦLG ΦSG θ ΦSL Contact surface Fig. 8: Common analogy of differing forces acting for a liquid droplet on a contact surface [4]. To have good wetting and spreading, the value of θ has to be as small as possible. Most leaded molten solders, like the Pb-Sn2-Ag2.5 soft solder have low contact angles on bare Cu substrates. The wetting and spreadability is almost always much more superior compared to Pb-free soft solder solutions [2]. When surface energy comes into play, it is seen that there are 2 interacting surfaces here; the Ag die back and the Cu substrate. Both these surfaces display varying degrees of surface energy with regards to the surface tension of the solder paste before reflow and during reflow, i.e. creation of a molten state. Before reflow, the solder paste molecules would be attracted to the surface of the Ag die back as the die bond force in Fig. 3d is released. This is because the Ag die back surface is more energetically favorable, or has a higher surface energy than the solder paste which makes it suitable for wetting as compared to the Cu substrate. The Cu substrate prior to reflow would tend to have a much more oxidized surface than that of the Ag die back, which makes it less energetically favorable to be wet. Therefore, coupled with the solder paste’s interfacial attraction to Ag surfaces and the material spring-back as the bond force is released, the bondline-thickness obtained would be rather high. Here, it would depend also on the amount of bond force being used during impact bonding. When reflow starts, and the flux in the solder paste removes the oxides from the surface, rendering it more suitable for wetting, slumping would occur as the solder paste material spreads across the substrate. The volume loss as depicted in Fig. 6 together with the spread of the solder material would cause the bond-line-thickness to reduce significantly. As it can be seen from this discussion, there are several factors which affect the creation of a good BLT. If a good BLT can be obtained, it would also translate to lower die tilt, and controlled or acceptable rates of voiding after reflow. In essence, the reliability of the joint from controlled process parameters is improved. (iii) Solder paste reflow profile considerations The solder paste reflow profile ultimately plays an important role too with regards to obtaining a reliable solder joint for large dies. With respect to the irregular die width and length used in this study, the solder flux material will evaporate along the shortest path possible. This loss of flux from the solder paste after reflow was proven earlier with measured dispensed solder paste material height. The concern however would be on the longer path. In this study, the peak reflow temperature was kept at 400˚C under N2/forming gas conditions, which illustrates a fully molten or liquidus stage exists for the solder material, bearing in mind the liquidus point of the material is between 304-309˚C [2]. Post-reflow examinations were able to showcase good coverage of PbSn2-Ag2.5 solder material under the die. However, for such large dies in mass manufacturing operations, the concern lies in how the BLT is controlled as the flux volatiles evaporate. It could cause uneven BLT, in particular at the corners or worse still severe voiding when the flux material expands without proper escape paths. The key would be to select a suitable flux material with acceptable activation energy which reflects the ramp rate of the solder profile. Too fast a ramp rate coupled with high flux activation energy can lead to trapped flux volatiles and subsequently voiding [4-5]. These factors need to be balanced. Conclusions To obtain a reliable solder die attach joint using Pb-Sn2Ag2.5 soft solder on large power dies, we have studied varying material and process factors. These factors are interrelated and cannot act alone. Die attach is a critical process in electronic packaging, as it relates to BLT, voiding and die tilt which are all possible failure mechanisms. For high power applications, the severity of failure can be catastrophic due to the high-end systems which utilize it for automotive, aerospace and military applications. Fig. 9 summarizes these interacting factors for soft solder impact bonding. Incomplete coverage under die Tilted die Large dies after reflow Hot slump and solder material collapse due to flux evaporation Cold slumping of material Higher BLT Large dies before reflow Spring-back of solder to die back due to surface energy interactions and bond force Surface energy interaction of solder with post-fluxed substrate Lower BLT Voiding Fig. 9: Interaction between pre and post-reflow soft solder die attach factors which lead to obtaining a reliable solder joint. Acknowledgments The authors would like to acknowledge the support given by the Infineon Failure Analysis and Power Segment Engineering Development teams in Malacca, Malaysia, and Neubiberg, Germany, respectively. View publication stats References 1. V.R. Manikam and K.Y. Cheong: IEEE Transac. Comp. Packag. Manuf. Vol. 1 (2011) pp. 457-478. 2. K. Suganuma: Lead Free Soldering in Electronics: Science, Technology, and Environmental Impact. Marcel Dekker (2004). 3. F. Joho: Semicon Singapore (2006) pp. 1-3. 4. N.C. Lee: Reflow Soldering Processes and Troubleshooting SMT, BGA, CSP and Flip Chip Technologies. Butterworth-Heinemann (2002). 5. F.G. Yost, F. M. Hosking and D. R. Frear: The Mechanics of solder alloy wetting and spreading. Van Nostrand Reinhold (1993).