Heterogeneous integration is the solution for advanced packaging by packing more dies or components into smaller footprints. This entire packaging evolution requires multiple technology breakthroughs in different aspects such as substrate design, interconnect methods, and materials. This article will outline the trends in various soldering materials technology which cater to heterogeneous integration, such as solder pastes, flip-chip fluxes, and ball-attach fluxes.
As packages get smaller and thinner, chips and components are attached in higher densities. This results in various new challenges in the packaging level, such as heat dissipation, higher warpages, and miniaturization. In addition, pressure to reduce costs pushes the industry for more cost-effective methods. Solder interconnections remain one of the key areas in supporting heterogeneous integration development. Solder interconnections are required to be smaller and tighter. This is driving current soldering materials technology to a different level in order to handle various challenges such as fine aperture printing and small dot dispensing, residue cleaning, and solder joint integrity.
Authored by: Kenneth Thum and Sze Pei Lim, Indium Corporation.
Solder Deposition Methods
Solder interconnects can be formed with different forms of solder such as solder paste, solder preforms, and pre-bump solder with fluxes. Each form of solder has its advantages and suitability for different process or components. Solder paste remains the primary material to form solder joints for die-attach and passive components. Solder pastes can be deposited through printing, dispensing, jetting, or a dipping process. The preferred method is printing as it has the best throughput and consistency. However, there are some situations where printing is not possible, such as in cavities or lid/frame attach solder deposition. Dispensing or jetting is the solution for these kinds of situations.
Solder paste is the frequently used form of solder in forming solder interconnections. Solder paste consists of fluxes and solder powders. Both components contribute to the overall performance including printing and dispensing consistency.
The main function of flux is to remove oxides so that solder can coalesce well during reflow. It also contains an important component—the thixotropic agent—which determines the rheology of the paste. Solder paste behaves as a viscoelastic material; the viscosity changes in accordance to the shear force applied to the material. This is very important for solder deposition such as printing and the dispensing process. Solder paste needs to be in a higher viscosity state after deposition so that it doesn’t slump and cause bridging. However, at the same time, the viscosity should remain low enough during the deposition process so that it is easier to print or dispense. Different deposition methods have different rheology requirements.
Solder powders are another area to look into when selecting solder paste. Solder powders come in different alloys and sizes. Different alloys have different specific gravity and hardness; hence, different ratios of flux and powder might be required when a different alloy is used. The powder size needs to suit the process requirement as well. One general guideline is to select the powder size based on the smallest dimension of the stencil’s aperture or inner diameter of the needle gauge if it is a dispensing process. The nominal largest particle size of powder size chosen should be able to fit in at least five spheres of the aperture or needle to avoid clogging. Table 1 below shows the powder size distribution of the common powder sizes in the industry.
Finer powder would ease printing and the dispensing process; however, there is a tradeoff when powder sizes get finer. Firstly, when powders get finer, cold welding becomes much easier. Cold welding happens when the powder surface is too “clean” as flux will continuously activate even if the paste is stored in cold temperatures. This eventually causes clumps of solder in the solder paste and can cause bad printing and dispensing. Secondly, the surface area of a given mass is larger for finer powders compared to coarser powders, so there are more surfaces prone to oxidation in finer powders. The flux system needs to work harder if paired with finer powders. Therefore, finer powder pastes require special care during reflow to prevent defects such as graping, voiding, and poor wetting. This is especially challenging if the fine powder paste is reflowed in a non-inert atmosphere. Figure 1 compares the surface area of powder sizes from T3 to T6. Taking T3 and T6 as an example, the surface area is 0.173sqm/g and 0.662sqm/g, respectively. This is nearly four times the difference in surface area, and therefore, four times more oxides could be expected when using T6 solder paste.
Good quality powders are essential when using fine powder pastes. If the powder surface is rough, more surface area is added to the system and does extra harm to the fluxing capacity. The powders need to have good uniformity, roundness, and smooth surfaces for optimum paste performance. The powder size distribution is also important, ensuring minimal excessive size particles that cause clogging. Figure 2 below shows a normal T6 powder under a scanning electron microscope (SEM).
Irregular shaped particles can be observed and many satellites could be found within the powder. This level of quality might not necessarily be bad for non-critical applications, but a higher-grade powder would be required when attempting fine aperture printing or dispensing for better consistency. Special controls and techniques are required to produce semiconductor-grade powders, as shown in Figure 3.
The next area to look into is the ratio between flux and powders in the solder paste. This is known as metal load, where the percentage represents the amount of powder within the solder paste by weight. The ratio would need to be optimized for different applications such as printing or dispensing. Typically, the metal load for dispensing paste will be lower as dispensing requires lower viscosity. The metal load requirement could also be different depending on the type of dispensing equipment used, such as a time pressure pump or auger pump. Secondly, different alloys require different metal load configurations because of the difference in alloy densities. For example, the specific gravity of SAC305 and Sn63/Pb37 is 7.38g/cm3 and 8.42g/cm3, respectively. Therefore, Sn63/Pb37 would need to have a higher metal content by weight to compensate for the alloy and flux ratio by volume.
Lastly, the process of mixing the flux and powder together is equally important in order to achieve consistent ultrafine paste deposition. For example, Figure 4 shows some defects at the paste level which might not be a problem when dispensing large dot sizes. Defects such as air bubbles, flux gel balls, and excessive particle size can affect dispensing consistency and also cause missing solder due to needle clogging.
Solder Paste Printing
Solder paste printing is a matured process in the industry; however, challenges arise when printing gets very fine. The smallest common passive that the industry is assembling are 008004 chips. There is also developing work for 0050025 components. The apertures can be as small as 100 x 100µm. In order to achieve the required area ratio, stencil thickness often needs to be very thin, around 25–50µm thick. Since the deposit size is very small, the allowable tolerance is also small. For example, for a 100µm round deposit size, 20% tolerance is equivalent to ±20µm; hence, the process needs to be controlled very tightly in every aspect. The printer setup in this instance requires a very stable board/substrate support underneath while printing. Secondly, the flatness is also crucial to ensure the squeegee stroke is uniform across the surface. As for the solder paste, the paste selected requires the correct combination of powder size and flux system. The powder size is typically selected based on the „Five Ball Rule,“ where the aperture will need to fit five particles of the selected size.
Several combinations of pastes, tooling, and stencil aperture sizes were investigated. Three solder paste specimens with different formulations and powder sizes were also studied. Finally, laser cut and electroformed stencils were compared by printing different apertures sizes.
In this study, Paste C performed the best among the three paste specimens. Figure 5 shows the boxplot comparison of volume deposited with a 0.42 area ratio stencil design.
The correct rheology of solder paste will help in printing such challenging apertures, and also minimize slumping after printing. The pad-to-pad distance could be as tight as 50µm; therefore, slump performance is crucial as well for such applications. If a more robust printing process is required, T7 powder could be utilized as well. Figure 6 shows the comparison between T6 and T7 solder pastes. It is clear that T7 will have less variation in printing; however, the reflow process will require more attention to prevent wetting-related problems due to the increment in powder surface area.
Solder Paste Dispensing
Solder paste printing is the preferred method for depositing solder paste because of the throughput and consistency. However, there are situations where printing is not possible, such as for lid attach solder dispensing or solder deposition in cavities. In these cases, dispensing could be the solution, but dispensing in ultra-small volumes could be very challenging, especially when dispensing sizes are 100µm and below. This, too, requires a close collaboration between the soldering material and dispensing equipment to produce these ultrafine deposits with high accuracy and consistency. The common dispensing pumps in the industry are time pressure pumps and auger pumps. Although these pumps have their advantages and disadvantages, they typically have difficulty in dispensing fine lines and dots due to the design limitations of the pumps. Various machine makers have developed new techniques for these applications, and soldering materials would need to follow the advancement and requirements accordingly.
This study investigates the process capability of different dispensing pumps and paste configurations. Three pumps were compared: time pressure pump, auger pump, and micro-squeezing pump. For comparison purposes, the solder paste specimens had different rheology and metal loads. The target for the study was 80 and 100µm round dots and lines. Before any measurements were taken, the target was to successfully dispense 4,000 dots and 400 lines, respectively, without any skips. Missing solder was not acceptable, as repair would be almost impossible in real world applications. Time pressure and auger pumps had missing solder paste when attempting 100µm dots, as shown in Figure 7.
The micro-squeezing pump successfully dispensed 4,000 dots and 400 lines continuously without skips. Figures 8 and 9 show the capability study of 80µm dot and line dispensing by using the micro-squeezing pump and the optimum solder paste.
There are several techniques in assembling flip-chips. The mass reflow method is the most common method, which is also the most cost-effective. However, when flip-chips evolved to thinner and tighter pitches, new challenges such as warpage arose. New attachment methods were developed to cater to the challenges in assembling these flip-chips. Thermal compression bonding (TCB) and laser-assisted bonding (LAB) are two of the newly developed methods. In these new processes, the heating process is much shorter and rapid compared to the mass reflow method. As a result, the short heat exposure might not have sufficient energy to evaporate the flux volatiles; hence, a different formulated flux would be required in TCB and LAB processes. The suitable flux would need to be activated in a very short time and yet still able to clean off the oxides effectively for good wetting. Secondly, the flux should not cause spattering as these processes typically have a much faster heating ramp-up rate. Table 2 shows different types of fluxes and their properties.
In addition to the attachment methods, there is a need to ensure flux residues are cleaned away effectively during the underfill and molding processes so as to not cause delamination or ECM-related problems. This is especially important when flip-chips have higher IO density with a narrower pitch. It could be difficult to clean away the residues even though water-soluble flux was used. As a result, ultra-low residue fluxes were introduced as an alternative for the industry. The residues left from these fluxes would be minimal to allow underfill material to flow well underneath the flip-chip. They would also be compatible with the molded underfill or capillary underfill material so that they would still have good adhesion to the substrate and would not cause delamination and solder extrusion during subsequent reflow.
Fan-out wafer level packaging or panel-level packaging is one of the focused areas in heterogeneous integration. Usually, the last process for these packages is the ball-attach or bumping process to prepare the package for the next solder interconnection assembly.
Although there are advantages gained from the wafer level packaging technique, different sets of challenges also arise in this process. One of the bigger problems is warpage due to CTE mismatch. As a result, this can cause bridging and missing ball in the ball-attach process. Secondly, many passivation materials are being tested; some could have a compatibility issue with the flux used.
If the flux is not compatible with the passivation, it may cause swelling and delamination on the passivation layer. The passivation layer also has an effect on the residue cleanability. If the passivation is not fully cured or has a rough texture on the surface, it could entrap flux residues and could be harder to clean as well.
A study was conducted in order to simulate the compatibility between the passivation material and flux. Selected ball-attach fluxes were printed on two different types of substrates; both had the same passivation layer but one had pad openings and the other one had none. The substrates were then reflowed three times using standard reflow conditions to simulate a worst case scenario. Incompatibility between fluxes and passivation are easier to detect with more flux and multiple reflow cycles.
Figure 12 shows an example of incompatibility between the flux and passivation layer. No abnormality was observed during the first reflow. However, after the second reflow onwards, a white ring was observed on the substrate after residue cleaning. The white ring was etched into the substrate; hence, it couldn’t be removed with the residue cleaning process.
Different chip makers have different approaches to heterogeneous integration. Often, it has very different sets of requirements for solder materials to suit each application and method. The common industry standard being used is the IPC standard; however, it might not be sufficient to cover every aspect and requirement in advanced packaging. A close collaboration is required between chip makers and material suppliers to develop an optimum suite of materials for better yield and process capability.
- Solder paste for heterogeneous integration has a higher requirement for flux rheology, flux compatibility with fine powders, powder quality, and paste mixing technology.
- Very fine aperture printings with area ratios down to 0.42 are possible with the correct combination of equipment setup and solder paste material. Finer powders will give a better printing process window, however, the reflow process window will be traded off.
- The dispensing method could be used when printing is not possible. Smallest dispense dots or line width could be as fine as 80µm.
- The flip-chip process has evolved with different attachment methods, tighter pitches, and lower standoffs. Different attachment methods have different sets of requirements for flip-chip fluxes, mainly due to the differences in heating methods. The ease of cleaning residues is also an area of focus, where ultra-low residue, no-clean fluxes could be an alternative for such processes.
- Ball-attach fluxes need to have sufficient tackiness and also good wetting power to minimize missing balls and bridging problems. The compatibility of flux with the passivation layer is also important as it could leave stubborn residues that couldn’t be cleaned off while also causing delamination.
1. Sze Pei Lim, K. Thum, and A. Mackie, “Meeting Solder Paste Printing Challenges for SiP in ‘Smart’ IoT Devices,” Chip Scale Review Magazine, July–August 2016.
2. K. Thum, Sze Pei Lim, and KC Tai, “Ultrafine Solder Dispensing for Heterogeneous Integration,” presented at the SMTAI, Chicago, October 2019.
3. John Lau, Ming Li, DeWen Tian, Nelson Fan, Eric Kuah, Wu Kai, Margie Li, J. Hao, Zhang Li, Kim Hwee Tan, Rozalia Beica, Cheng-Ta Ko, Yu-Hua Chen, Sze Pei Lim, Ning-Cheng Lee, Koh Sau Wee, Jiang Ran, and Cao Xi, “Warpage and Thermal Characterization of Fan-Out Wafer-Level Packaging,” IEEE/ECTC, Orlando, Florida, May 2017.
4. Lim et al., “Laser Assisted Bonding Technology Enabling Fine Bump Pitch in Flip-Chip Package Assembly,” IEEE 17th Electronics Packaging and Technology Conference (EPTC), Singapore, 2015.
5. Andy Mackie, Hyoryoon Jo, and Sze Pei Lim, “Flip-Chip Flux Evolution,” IMAPS, Boston, October 2019.
First presented at the 21st Electronics Packaging Technology Conference (EPTC), December 3–6, 2019, Singapore.