Chip Bond Pad Peeling

Phenomenon of Bond Pad Peeling
The phenomenon in which part of the surface of the pad (and sometimes part of the oxide layer under the pad) is peeled off from the pad along with the solder ball, which is called bond pad metal peeling off. Figure 1a illustrates the microscopic morphology of aluminum pad peeling after gold wire ball bonding. The pads shown in Figure 1b are, from top to bottom, the AI layer, the MoSi₂ layer, the borosisilicate glass (BPSG) layer, and the SiO₂ layer. It can be clearly seen that after the peeling occurs, the top 3 layers are peeled off the pad, revealing the SiO₂ layer at the bottom.

0010-13927 Pedestal Lift Assembly Preclean

Fig. 1 SEM diagram of aluminum pad shedding Fig. 2 shows the peeling phenomenon and microscopic morphology of the multi-layer composite pad composed of AI-Si-Cu pad and TiW layer.Fig.2. Peeling phenomenon and microscopic morphology of multilayer composite pads composed of AI-Si-Cu pads and TiW layers

Bonding pads are located on the surface of semiconductor devices and are subjected to chemical and mechanical loads. Chemical loads are triggered by front-end wafer fabrication processes such as passivation layer fenestration, dielectric layer fenestration, and surface cleaning operations; Mechanical loads are induced by electrical testing and packaging processes in the subsequent process. Therefore, the pads need to be strong enough to withstand these loads.

The core of the pad peeling problem is the competitive relationship between the adhesion between the solder ball and the aluminum layer on the pad surface, and the adhesion between the aluminum layer on the pad surface and its adhesion layer and the silicon matrix. In this competitive relationship, when the solder ball is subjected to external force, if the adhesion between the aluminum layer on the surface of the pad and its adhesion layer and the silicon matrix is strong enough, then it will be manifested as the solder ball and the pad are peeled off or the solder ball itself is broken, which is a normal situation. Conversely, when the adhesion between the aluminum layer on the pad surface and the silicon matrix is not large enough, the adhesion between the solder ball and the aluminum layer on the pad surface will be dominant. At this time, under the action of external force, the solder ball will peel off from the silicon matrix with the aluminum layer on the surface of the pad and its adhesion layer, resulting in the phenomenon of pad peeling.
In general, the bond between the solder ball and the aluminum layer on the pad surface is limited. When subjected to external loads, bond breakage and solder ball peeling should precede pad peeling. Therefore, if the pad is at a disadvantage in this competitive relationship, it means that the adhesion of the aluminum pad is weak and there is a quality risk.

The appearance of peeling phenomenon is often accompanied by internal injuries to the pads. These internal injuries are thought to be caused during package bonding or during electrical performance probe testing. Pad internal injury is an imperceptible quality hazard in the wire bonding process, and severe internal injury can lead to delamination or direct peeling of the pad. Integrated circuits with these quality hazards may be found and rejected in the electrical performance test. However, more internal injuries are in a critical state, and the initial electrical performance degradation is not obvious, and only in the subsequent screening tests, after temperature cycling, thermal shock, aging, mechanical vibration and other tests, will the problems be exposed, which is manifested as pad peeling, pitting, lead debonding, electrical performance open circuit, etc.
Although in most cases, the stress on the pads during the bonding process can be reduced by optimizing the ultrasonic parameters, cleaning the capillary, perfecting the bonding process, etc. However, in some cases, the measures taken in the post-packaging process cannot completely solve the problem of pad peeling. This is because in some cases, the pads themselves have quality hazards due to improper control in the chip manufacturing process, and this internal injury is congenital. In this case, the process parameters should not be blindly reduced in order to avoid pad peeling, as this will not only not compensate for internal injuries, but will also reduce the bond reliability between the bond wire and the pad. For such devices, it is reasonable to perform batch inspection or scrap them directly to prevent these weak links from causing problems in the subsequent screening or use process.

Ultrasonic parameters have been studied to point out that the process of pad peeling initially originates from cracks on the surface of aluminum pads and their internal metal layers, and the inappropriate combination of parameters such as bonding power, bonding force, time and temperature is the cause of this damage. Of these factors, ultrasonic power has the most significant effect, as the energy it provides drives a shear effect between the surface and inner layers of the pad. When the ultrasonic power is too high, it will cause damage to the metal layer of the pad, which in turn will cause the phenomenon of pad peeling. In the case of bonding force, the entire ball-pad bonding system requires more energy to slide because the pressure exerted by the capillary on the solder ball inhibits the tendency of shear motion. Therefore, in terms of causing pad peeling, increasing the pressure actually plays a role in inhibiting the influence of ultrasonic power, that is, increasing the pressure will reduce the occurrence of pad peeling.

The preheating temperature can soften the pads, and under the same conditions, increasing the preheating temperature can help reduce the failure rate of the pads. In summary, the selection of appropriate ultrasonic parameters is a key prerequisite to avoid internal damage to the pads due to bonding. Some researchers have carried out finite element simulation analysis on the influence of ultrasonic amplitude on the stress distribution of pads at 138kHz ultrasonic frequency, and the results are shown in Figure 3. As you can see from the diagram, during the bonding process, the stress changes with the movement of the capillary, and only when the capillary moves to the central area does the stress distribution be symmetrical. Further simulation analysis shows that the bonding stress in the pad increases with the increase of ultrasonic amplitude, as shown in Figure 4. These results show that the ultrasonic amplitude has a significant effect on the stress and deformation of the wire during the bonding process.Fig.3. Stress distribution at different timesFig.4. Effect of ultrasonic amplitude on pad stress

Chopper
Capillary blades play a crucial role in the bonding process of aluminum process chips. As a key carrier for the precise application of ultrasonic parameters to the pads, it is an indispensable part of the entire energy transfer process. If the capillary is abnormal, it is difficult for the ultrasonic power and pressure to act uniformly and stably on the pad, which will seriously interfere with the normal propagation of ultrasonic energy and negatively affect the bonding quality. On the other hand, according to the researchers' analysis, when the capillary undergoes a large number of bonding operations, the contaminated capillary head undergoes some changes. Due to the contamination, the contact surface area between the capillary head and the solder ball increases, which leads to increased adhesion between the capillary head and the solder ball, and the magnitude of the vertical tension load will also increase accordingly. During the lifting of the capillary, the vertical load is transferred sequentially from the capillary to the deformed ball and then to the pad. This vertical load is the direct source of power that triggers the skinning of the pads. When a vertical load is applied to the pad, it causes cracks to occur in the pad. These cracks first germinate in region A and then extend along the interface to the oxide layer below the pad, eventually leading to the occurrence of pad peeling, a process clearly illustrated in Figure 5. In addition, the statistics reflect a clear trend, with the use of capillary blades used more than 200,000 times in 87% of cases of pad peeling, further indicating that there is a strong link between the use of capillary and pad peeling. The research results also show that a reasonable selection of capillary model is a key solution to improve the problem of pad peeling of aluminum process chips. When we choose a capillary with a cone angle of CA=70° and a small end diameter CD, and increase the bonding pressure moderately, it can have a positive effect on the shaping of the solder joint. This results in a good, uniform and sufficient contact between the solder joint and the pad, effectively avoiding the problem of local stress concentration caused by the protrusion of the solder joint. In addition, the solder ball formed by the CA=70° capillary has a relatively thin extrusion bevel, which makes it less dissipating when conducting ultrasonic energy. In addition, compared with CA=120°, CA=70° capillary has a weaker degree of energy accumulation in the center region of the solder ball, which can significantly reduce the damage of bonding power to the aluminum layer in the center area of the pad, and the specific differences and advantages are shown in Figure 6.

0010-13774 Assy, Wafer Lift Cooldown/passthru

图5焊盘起皮原因示意图Fig.6. Comparison of the effects of different taper angles on bonding

Slippage
In some cases, the aluminum disc shedding (ABPO) rate could not be reduced to zero, even after optimizing the ultrasonic power, pressure, and preheating temperature, suggesting that there were at least other factors contributing to the ABPO phenomenon in the batch. The researchers found that by optimizing the software system to reduce the slippage during the bonding process, the slippage between the capillary and the solder ball can be significantly reduced. As a result, the shear stress on the inside of the pad is greatly reduced, thus effectively avoiding internal damage to the pad and thus eliminating the ABPO phenomenon. Figure 7 shows the slippage of the solder ball, and the traces left by the slippage on the gold wire ball can be clearly seen in the diagram. Figure 8 shows a comparison of the action of the capillary before and after the software optimization.Figure 7 Solder ball slippage

Fig.8. Comparison of the action of the capriillary before and after software optimization

The researchers observed with the help of laser confocal technology and found that the average slip depth of the cells with pad peeling was 9.6 μm, while the average slip depth was reduced to 7.44 μm after the software optimized to reduce the slip. The Vickers hardness of gold is calculated, and the average slip force of the element with pad peeling is 48.7 gf, and the average slip force is reduced to 29.2 gf after the software optimizes the slip reduction. The finite element simulation results show that due to the slip force, the shear strength of BPSG in the ABPO element is 1.74 GPa, while the shear strength of the element without ABPO is 1.29 GPa.

Process parameters
Some researchers believe that factors such as preheating temperature, bonding power, and bonding force have an impact on pad peeling. The specific impacts are shown in Table 1.Table 1 Effect of preheating temperature, bonding power, and bonding force on pad peeling

Wafer fabrication
During wafer fabrication, residues of halogens can have a corrosive effect on aluminum pads and their oxide films. At the same time, the metal layer in the pad will vaporize and expand after heating due to moisture absorption, which will lead to delamination, which will have an effect on the decline of the adhesion of the metal layer inside the pad. According to the SEM/EDX analysis of electroless nickel-palladium immersion gold (ENEPIG) pads after bonding and peeling, oxidation is the main cause of Pd and Ni delamination. Figure 9 shows the microscopic topography of the pad and leads after the ENEPIG pad is debonded, and the Focused Ion Beam (FIB) cross-section shown in Figure 10 shows three different regions: the area through which the lead is detached, the area that is slightly away from the area, and the conventional reference region.Fig.9. Micromorphology of the pads and leads after the ENEPIG pad is debondedFig.10. FIB cross-section

It has been pointed out that for the multilayer pad structure, five surface aluminum layers (M2) with different thicknesses of 250nm, 330nm, 450nm, 550nm and 650nm are set. After 200°C and 3 h aging, the results show that the thinner M2 layer is more prone to ball neck failure, cushioning and pilling failure, as shown in Figures 11 - 13.

Figure 11 Aluminum pads on the chip

Fig. 12 Typical failure modes in a tensile test

Fig.13 The proportion of peeling and solder ball debonding failure modes to the total failure modes in the bond tensile test varies with M2 thickness