Comparison of Laser and Reflow Soldering in Sn-3.0Ag-0.5Cu/ENEPIG Joints
Article information
Abstract
Abstract
In this study, the laser soldering process was compared with the conventional reflow soldering process using a Sn-3.0Ag-0.5Cu solder and an ENEPIG (Electroless Nickel-Electroless Palladium-Immersion Gold) surface-finished substrate. Laser soldering enables localized heating and shorter bonding time compared to reflow soldering, which reduces thermal damage to both the chip and the substrate. To evaluate the long-term thermal reliability, aging tests were conducted at 150 °C for up to 2000 h. Immediately after reflow soldering, (Cu, Ni)6Sn5 and (Ni, Cu)3Sn intermetallic compounds (IMCs) were formed at the joint interface, and the IMC layer thickness increased with aging time. In the case of laser-soldered joints, a very thin (Cu, Ni)6Sn5 layer was observed at the interface, and although its thickness increased with aging, (Ni, Cu)3Sn was not detected even after 2000 h of aging. After 2000 h, IMC spalling was observed at the interface. For both soldering processes, the shear strength decreased with increasing aging time, while the laser-soldered joints consistently showed higher values than those obtained by reflow soldering. All samples exhibited ductile fracture behavior.
1. Introduction
The increasing level of integration and the continued miniaturization of semiconductor packages have recently accelerated the reduction of interconnection pitch. Accordingly, interest and development in micro-joining technologies have been increasing, and there is a need to develop new soldering processes that can enhance precision and reduce thermal damage. The reflow soldering process, which has been widely used, requires a preheating step and has a relatively long processing time. This prolonged thermal exposure can induce thermal damage to chips and substrates and may lead to PCB warpage and solder bridging issues1-5).
As an alternative to address these issues, laser soldering processes have been proposed. Laser soldering offers advantages such as localized heating and short bonding times, which minimize thermal exposure during the process6-10). Therefore, it is gaining attention as a soldering process for fine-pitch and heat-sensitive components11-15).
Various surface-finished substrates are being used to ensure the reliability of joints. In particular, the Electroless Nickel-Electroless Palladium-Immersion Gold (ENEPIG) surface finish is widely used in the semiconductor packaging field due to its excellent corrosion resistance and bonding reliability. ENEPIG is a multilayer structure in which Pd and Au layers are sequentially deposited on a Ni(P) layer. ENEPIG provides higher resistance to black pad defects and superior solder wettability compared with the Electroless NickelImmersion Gold (ENIG) surface finish of electroless Ni/Au structures, making it advantageous for forming high-reliability joints. In the ENEPIG structure, the Ni layer suppresses Cu diffusion, the Pd layer prevents Ni oxidation, and the Au layer enhances solder wettability, enabling stable bonding16,17).
In this study, laser and reflow soldering processes were compared using Sn-3.0Ag-0.5Cu solder and ENEPIG-finished substrates. To evaluate long-term reliability, isothermal aging was conducted at 150 °C for up to 2000 h after each soldering process. The interfacial reaction and intermetallic compound (IMC) growth behavior were examined as a function of aging time. Subsequently, the mechanical properties were evaluated through shear testing under each condition.
2. Experiment Method
A solder paste with a composition of Sn-3.0Ag-0.5Cu (wt%, SAC305) was applied to an ENEPIG surface-finished substrate through a stencil mask, and then SAC305 solder balls with a diameter of 450 μm were aligned, as shown in Fig. 1. The size of the substrate was 17 × 17 mm, and the diameter of the Cu pad was 350 μm. The ENEPIG substrate consisted of a 5 μm Ni(P) layer, a 0.05 μm Pd layer, and a 0.075 μm Au layer. The solder balls were joined to the substrate via reflow soldering and laser soldering. Reflow soldering was conducted at a peak temperature of 250 °C for 90 s. For laser soldering, an area laser method was applied, in which a 200 W laser irradiated at a peak temperature of 240 °C for 2 s. The laser wavelength was 1070 nm, and the laser irradiation area was adjusted to match the substrate size, resulting in an energy density of 0.692 W/mm2. To evaluate long-term high-temperature reliability, the samples were subjected to isothermal aging at 150 °C for 2000 h. Subsequently, shear tests were conducted on 28 solder balls under conditions of a shear speed of 200 μm/s and a shear height of 50 μm to assess the mechanical properties of the joints.
Schematic of solder paste printing and solder ball attachment
For cross-sectional analysis of the joints, the samples were cold-mounted in epoxy resin and sequentially polished using abrasive paper from #100 to #2000, followed by polishing with 1 μm and 0.3 μm alumina suspensions. Unreacted residual Sn was removed using an etching solution composed of ethanol (C2H5OH), nitric acid (HNO3), and hydrochloric acid (HCl) in a weight ratio of 95:4:1. The cross-section and fracture surface of the joint were analyzed using scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS).
3. Results
Fig. 2(a) and (c) show the cross-sectional and top-view observation results of reflow soldering, while Figs. 2(b) and 2(d) show the cross-sectional and top-view observation results of laser soldering. The top-view observations were conducted after completely removing the solder using an etchant. Cross-sectional analysis confirmed that laser soldering formed a very thin IMC layer due to its short process time. In the top-view images, laser soldering also exhibited much finer IMC grains compared to reflow soldering. Reflow soldering formed relatively coarse scallop-type IMCs, whereas laser soldering resulted in plate-like or rod-like IMC morphologies.
Cross-sectional and top-view SEM micrographs of the interfacial IMCs after (a, c) reflow soldering and (b, d) laser soldering
Fig. 3 shows the IMC growth behavior of the reflow-soldered joints as a function of aging time. In the initial stage of aging, (Cu, Ni)6Sn5, (Ni, Cu)3Sn, and Ni3P were observed. As the aging time increased, the IMC layer gradually thickened, and after 2000 h, IMC spalling occurred, as shown in Fig. 3(g). This interfacial spalling is considered to be caused by the low adhesion of Ni3P and the thermal stress accumulated during prolonged aging18).
Cross-sectional SEM micrographs of the reflow-soldered SAC305/ENEPIG joints aged at 150 °C for (a) 0 h, (b) 24 h, (c) 250 h, (d) 500 h, (e) 1000 h, (f) 1500 h, and (g) 2000 h
Fig. 4 shows the EDS point analysis results of the reflow-soldered joints, with (a) showing the initial state before aging and (b) showing the state after 2000 h of aging. The analysis results confirmed the presence of (Cu, Ni)6Sn5, (Ni, Cu)3Sn, and Ni3P under both conditions. The (Ni, Cu) 3Sn phase corresponds to Cu3Sn lattice in which Ni partially replaces Cu, and it is believed to form due to Ni diffusion from the ENEPIG substrate8).
Interfacial IMC compositions of SAC305/ENEPIG joints after reflow soldering under different aging times, (a) 0 h and (b) 2000 h
Fig. 5 shows the IMC growth behavior of the laser-soldered joints as a function of aging time. In the initial stage of aging, (Cu, Ni)6Sn5 and Ni3P were observed. As aging time increased, the thickness of the IMC layer gradually increased, and partial island-like growth appeared, as shown in Fig. 5(b).
Cross-sectional SEM micrographs of the laser-soldered SAC305/ENEPIG joints aged at 150 °C for (a) 0 h, (b) 24 h, (c) 250 h, (d) 500 h, (e) 1000 h, (f) 1500 h, and (g) 2000 h
Fig. 6 shows the EDS point analysis results for the laser-soldered joints, with (a) representing the initial state before aging and (b) representing the state after 2000 h of aging. The presence of both (Cu, Ni)6Sn5 and Ni3P was confirmed in both conditions. However, after 2000 h of aging, IMC spalling similar to that observed under reflow conditions was observed at point 3 in Fig. 6(b).
Interfacial IMC compositions of SAC305/ENEPIG joints after laser soldering under different aging times, (a) 0 h and (b) 2000 h
Fig. 7 shows the variations of IMC thickness formed according to each soldering process. The thicknesses of Cu6Sn5 and Cu3Sn were measured separately. The initial stage of aging, the IMC formed by laser soldering was significantly thinner than that formed by reflow soldering. However, the IMC thickness rapidly increased with aging time, and after 1500 h, a thicker IMC was formed than under reflow soldering conditions. This is because the rapid heating of laser soldering limits the initial diffusion reaction, resulting in the formation of a thin IMC. However, the thin IMC layer kept a wide reaction interface between Cu and Sn, promoting continuous solid-state diffusion during the subsequent aging and accelerating IMC growth. In both soldering conditions, Cu6Sn5 was significantly contributed to the overall IMC thickness increase, while the thickness change of Cu3Sn was relatively minor. For the laser-soldered joints, the IMC thickness is distinguished based on whether or not spalled IMC was included.
Variation of IMC thickness depends on different aging times
Fig. 8 shows the results of the shear strength variation under each soldering condition. Shear tests were conducted to evaluate the mechanical properties of the joints. The initial shear strength was greater in laser soldering than in reflow soldering because the IMC was thinner and more compact. As aging time increased, the shear strength decreased in both conditions due to solder softening and IMC growth. Laser soldering exhibited consistently higher shear strength compared to reflow soldering, indicating superior mechanical reliability.
Variation of shear strength depends on different aging times
Fig. 9 shows SEM images of the fracture surfaces after the shear test for each soldering condition. Ductile fracture behavior was observed under all conditions. In the case of laser soldering, localized voids were observed on the fracture surface after 1000 h of aging, which are indicated in yellow. The formation of these localized voids is believed to result from rapid heating and cooling during laser soldering, which caused coalesced micro-voids generated at the interface to become trapped within the joint rather than escaping to the outside. As shown in the schematic illustrations of Fig. 10, the fracture path of laser-soldered joints tended to remain within the solder bulk even after aging. However, stress concentration occurred around the voids after aging, resulting in the observation of voids on the fracture surface19).
Fracture surfaces at different aging times
Schematics illustration of fracture behavior after aging
4. Conclusion
This study analyzed the microstructure and mechanical properties of SAC305/ENEPIG joints formed via laser soldering. Due to the short processing time of laser soldering, the initial IMC formation was thin and dense. Fine IMC grains in plate-like or rod-like morphologies were observed unlike reflow soldering, which formed coarse scallop-type IMC. In the reflow soldering joint, (Cu, Ni)6Sn5, (Ni, Cu)3Sn, and Ni3P were identified, while in the laser soldering joint, (Cu, Ni)6Sn5 and Ni3P were observed, with some island-like IMC growth. After 2000 h of aging, IMC spalling occurred in both conditions due to the low adhesion of Ni3P and the accumulation of thermal stress. According to the IMC thickness measurement results, the laser-soldered joint maintained higher shear strength than the reflow-soldered joint, despite the rapid IMC growth during aging. Fracture surface analysis also revealed ductile fracture under all conditions. Although some localized voids were observed, the overall fracture path remained stable within the solder bulk. Consequently, laser soldering maintains high shear strength even after aging due to the initial formation of a thin and dense IMC, and it provides stable bonding characteristics. This demonstrates that long-term reliability can be achieved by applying laser soldering technology in next-generation high-density packages.
Acknowledgement
This research was supported by the Ministry of Trade, Industry and Energy’s Semiconductor Advanced Packaging Leading Technology Development Project (RS-2025-02220503) and the National Research Foundation of Korea (RS-2023-00247545) funded by the Ministry of Science and ICT.