An Innovative Hybrid Cleaning Approach For Contaminant Removal in Semiconductor Packaging

The growing demand for power semiconductors, which control large currents and voltages, has resulted in their widespread use in various electronic devices, including electric vehicles (EVs). The requirement for greater performance and smaller devices has led to increased operating temperatures for these power semiconductors. This rise in temperature poses a challenge as it often leads to delamination. Delamination occurs when there is a mismatch in coefficients of thermal expansion (CTE) between the encapsulate materials, such as the mold compound used to encapsulate the semiconductor chip and the lead frame. One conventional approach to address delamination, without modifying the design of encapsulate material, is to roughen the surface of lead frame. This technology aims to improve the interlocking and increase the contact area between the mold compound and the lead frame surface to ensure high level reliability. The roughened surface of the Power Quad Flat No-Lead (PQFN) package presents challenges in conventional cleaning methods due to the presence of crevices and irregularities that can trap contaminants.

In the semiconductor industry, various alternative cleaning methods have been developed to address the challenges in flux cleaning process. Conventional cleaning approaches predominantly involve physical agitation or chemical action, such as centrifugal cleaning, ultrasonic cleaning, or in-line spray cleaning. Agitation plays a crucial role in flux cleaning, and ultrasonic cleaning has emerged as the most recommended technique. This cleaning method provides higher levels of agitation compared to other conventional methods, making it particularly effective for addressing the challenges posed by roughened lead frames. However, it is important to note that ultrasonic cleaning requires the use of a suitable cleaning agent to achieve optimal results [1].

Roughened lead frames present unique challenges in cleaning as the intricate lead-frame structures tend to trap contaminants, making thorough cleaning difficult to achieve. Also, the design of power semiconductors often includes tight spacing and inaccessible areas where conventional cleaning methods struggle to reach, resulting in high levels of contaminants. Ultrasonic cleaning can effectively address most of these challenges with its key features of agitation and cavitation, as well as its ability to penetrate recessed areas and tight spaces. When combined with a suitable cleaning agent, ultrasonic cleaning process can be a highly effective method. This study uses immersion ultrasonic solvent cleaning methods as the conventional approach [2].

The benchmark cleaning process involved immersing the samples in a concentrated solvent and subjecting them to controlled agitation using ultrasonic cleaning. Recirculation and filtration units were also employed. The duration and specifics of the cleaning process were determined based on the suitability of the package design, with the aim of removing non-clean flux residue and other contaminants from the lead frame’s surface. After the cleaning process, the samples underwent thorough rinsing and drying. However, it was observed that nearly all packages still had contamination present, indicating the need for improvement in the cleaning process. The presence of contamination on the lead-frame surface adversely affects the wire bond bondability. Whitish contaminants are primarily responsible for non-stick on lead (NSOL) failures, while stains, smears and solder flakes predominantly contribute to non-stick on pad (NSOP) failures.

Table 1 provides insights into the different types of contaminants that can be found during the flux cleaning process. To gain a deeper understanding of these stubborn contaminants, their compositions were thoroughly investigated. Extensive testing and analysis, including Energy Dispersive X-ray (EDX) analysis and Fourier transform infrared (FTIR) spectroscopy analysis, were conducted to identify the source and nature of these contaminants. Among all the contaminants, whitish contaminants originating from flux residue or flux clean chemical exhibited adhesive properties, making them difficult to remove even with standard cleaning methods.

Table 1: Elemental analysis of APTES coated lead frame.

Type of Contamination	Appearance	Potential Composition
Stain		Flux Residue, Flux Clean Chemical
Smear		Flux Residue, Solder Paste (Pb, Sn), Lead Frame (Cu)
Solder Flake		Solder Paste (Pb), Flux Clean Chemical, Lead Frame (Cu)
Whitish Contaminant		Flux Residue, Flux Clean Chemical

To validate the effectiveness of the proposed (hybrid) solution, it underwent extensive testing, encompassing various washing times, temperatures, and cleaning scenarios, to evaluate its efficacy in reducing surface contamination compared to conventional methods. The comprehensive testing yielded significant results, demonstrating a notable reduction in surface contamination levels and surpassing the capabilities of traditional approaches.

By implementing the proposed solution, a breakthrough in cleaning efficiency was achieved, addressing the challenge of surface contamination and mitigating the risk of failures. The improved cleaning efficiency contributes to enhanced product reliability, ensuring optimal performance and minimizing potential issues arising from residual flux.

To optimize the hybrid cleaning approach and identify critical process parameters, a structural design of experiments (DOE) methodology was employed. This systematic variation of parameters allowed for the identification of key factors significantly impacting cleaning efficiency. By focusing on these areas and implementing the hybrid cleaning approach, this study aims to enhance the overall cleaning process for roughened lead frames, improve wire bonding quality, and mitigate the risk of NSOL or NSOP failures. Figure 1 provides a comprehensive overview of the methodology employed in the hybrid cleaning process, illustrating the sequential steps involved in achieving optimal cleaning efficiency and minimizing contamination.

Fig. 1: Overall process of hybrid flux cleaning approach.

The hybrid cleaning approach combines immersion-type and spray-in-air type flux cleaning methods. The immersion-type cleaning involves submerging the roughened lead frames in a cleaning solvent, which allows the solvent to penetrate the crevices and irregularities on the roughened surface, reaching and loosening trapped contaminants, ensuring that the cleaning solvent thoroughly contacts all areas of the lead frames, promoting effective cleaning. Following the immersion stage of chemical wash and deionized (DI) rinsing, the lead frames undergo a spray-in-air cleaning process. This technique involves the use of pressurized DI water, which is directed at the lead frames’ surfaces. The high-velocity DI water effectively dislodges and removes residual contaminants from the roughened surface. The spray-in-air method reaches areas that may be difficult to clean through immersion alone, such as tight crevices or undercuts. Several hypotheses were formulated and thoroughly discussed in relation to the results obtained from the testing and evaluation of the proposed solution. A comprehensive analysis and interpretation of the data were conducted to validate or refute these hypotheses.

Hypotheses

H1: Longer chemical soak time will result in more contaminants being trapped on the roughened surface.

H2: Contamination rate can be minimized with a shorter rinsing bath duration.

H3: Contamination rate can be minimized by applying pressure during cleaning process.

Feasibility study of different cleaning scenarios

To validate the hypothesis regarding the impact of chemical soak time on the roughened lead-frame surface, a series of controlled experiments were executed. Different immersion times were tested, ranging from short durations to extended periods, while keeping other parameters constant. The lead frames were then thoroughly examined using advanced surface analysis techniques.

Cleaning Scenario 1: Different chemical soak time and temperature

One of the important parameters to be considered for optimal cleaning is the chemical soak time and temperature, while keeping other factors constant, such as DI rinsing time, ultrasonic power, ultrasonic frequency, air blow, and drying time. The results obtained, shown in figure 2(a), illustrate that the percentage of stain reaches its lowest point within the range of 100 to 200 seconds of washing time. However, beyond this range, the percentage of stain shows a nonlinear increase, peaking at 300 seconds of washing time. Figure 2(b) shows no significant impact on solder flake formation. The data in figure 2(c) indicates that a chemical temperature of 50°C results in lower percentage of smear compared to 40°C. However, longer soak times at 50°C increased the presence of whitish contaminants, according to figure 2(d).

Regarding failure rates, it is important to mention that no NSOP failures were observed. However, as for NSOL failures, samples cleaned with longer chemical time at 50°C exhibited a higher failure rate ranging from 4% to 7%, compared to the 2% to 3% failure rate observed at 40°C. These findings support the hypothesis stated in H1 by revealing that excessively long chemical soak times led to higher percentages of contaminants on roughened surface. This highlights the effectiveness of shorter soak times for better cleaning. Additionally, the results emphasize the importance of temperature optimization in reducing NSOL failures.

Fig. 2: Graph of contaminant percentage variations to chemical soak time and temperature.

Cleaning Scenario 2: Different DI rinsing time, dripping time, and drying time

In addition to investigating the chemical washing conditions, the study also examined the influence of DI rinsing time, dripping time, and drying time on the cleaning process while keeping the chemical wash conditions constant. The effect of different rinsing and drying parameters on surface cleanliness was evaluated, and the results are tabulated in figure 3. The findings indicate that shorter rinsing and dripping times resulted in improved cleanliness levels, as evidenced by reduced levels of stains, smears, and whitish contaminants compared to longer rinsing or dripping times. These results align with hypothesis H2, which suggests that reducing the duration of the rinsing bath leads to a decrease in the rate of contaminants. The findings emphasize the importance of optimizing rinsing parameters to achieve better surface cleanliness.

Fig. 3: Responses of the contaminant percentage on different DI rinsing time, dripping time, and drying time.

Cleaning Scenario 3: Different DI rinsing method

A simulation was conducted to assess the effectiveness of different DI rinsing approaches, specifically evaluating the mechanical forces of ultrasonic cavitation and spray-in-air to remove the whitish contaminants. The chemical immersion wash time, temperature, and ultrasonic power remained the same for this simulation. The results obtained from Table 2 indicate that the hybrid rinsing method significantly reduced the percentage of whitish contaminants from 26% to 4.7%. Furthermore, the NSOL failure rate decreased from 12% to 0% when utilizing the hybrid rinsing method. However, according to the data presented in figure 4, there were no significant differences observed in the stitch pull results between the immersion and hybrid rinsing methods. Additionally, the stitch pull strength values in Table 3 also met the specified requirement of ≥ 4g.  The predominant failure mode in stitch pull tests was identified as breakage at the stitch neck. These results support hypothesis H3, which proposes that the contamination rate can be minimized by applying pressure specifically during the rinsing process of the cleaning procedure.

Table 2: Evaluation results on different rinsing methods.

Fig. 4: Stitch pull comparisons of different rinsing methods.

Table 3: Stitch pull strength on different rinsing methods.

To further investigate the influence of surface characteristics on the wetting behavior during the rinsing process, the study specifically examined the effects of wetting conditions and chemical composition, as outlined in Table 4. Wetting behavior refers to how effectively droplets spread and interact with the surface being rinsed. When using a solvent or rinsing agent with a high chemical-to-water ratio, the droplets exhibited good wetting behavior by spreading out and covering the surface to create a hydrophilic condition. This indicates that the surface was easily wetted by the rinsing agent, promoting effective cleaning. Conversely, when the droplets tended to bead up on the surface, it indicated a hydrophobic condition. In such cases, the rinsing agent may struggle to fully penetrate and clean the surface.

The wetting performance of the rinsing agent plays a crucial role in achieving cleanliness, particularly on roughened surfaces. Cleaning solvents typically have a lower surface tension of around 20 dynes/cm, enabling them to penetrate into the surface’s irregularities and effectively dissolve contaminants. However, this low surface tension can also cause the solvent to be drawn back into tight spaces, hindering its complete removal from those areas and potentially leaving behind residue or contaminants [3].

Figure 5 provides a visual representation of how surface tension influences droplet adhesion on a roughened surface.  To overcome this limitation, spray rinsing is necessary to remove any remaining solvent that may have become trapped in the crevices of the lead frame. Spray rinsing involves applying a forceful spray of water or rinsing agent onto the surface, which helps dislodge and remove the solvent from the irregularities. It is important to perform the spray rinsing while the lead frame is not completely dry, preferably in humid conditions. The presence of moisture can significantly boost the effectiveness of spray cleaning in removing contaminants that remain on roughened surfaces, thereby enhancing the overall cleaning effectiveness.

Table 4: Surface wetting conditions on different composition.

Composition	Wetting Condition	Observation
100% chemical		Wetting
High chemical to water ratio		Wetting
Low chemical to water ratio		Dewetting

Fig. 5: Illustration of different surface tensions on a roughened surface.

Parameter Window: Spray-in-air rinsing time and belt speed

Based on the predicted profile obtained from figure 6, the higher belt speed during spray rinsing leads to improved cleaning outcomes. A higher belt speed facilitates faster movement of the contaminated surface through the spray zone, enabling more efficient rinsing and removal of contaminants. This increased speed helps minimize the contact time between contaminants and the surface, reducing the chances of re-deposition or incomplete removal. Additionally, coupling the higher belt speed with high pressure during the spray rinsing process has shown even better results. The high pressure enhances the mechanical action of the spray, effectively dislodging and removing contaminants from the surface. It aids in breaking the adhesion between contaminants and the surface, ensuring thorough cleaning.

Fig. 6: Predicted profile for spray-in-air cleaning process.

Immersion cleaning versus hybrid cleaning

By interpreting the data from different cleaning scenarios and simulation tests, the data analysis provides valuable insights into the optimal cleaning conditions. For immersion-type cleaning, the findings indicate the following optimal cases:

Chemical Soak Time and Temperature: Excessively long chemical soak times led to higher percentages of contaminants on the roughened surface. Shorter soak times at 50°C were found to be more effective for better cleaning outcomes.

Chemical Bath Life: The percentage of whitish contaminants remained consistent across different chemical bath lifetimes, suggesting that the duration of chemical bath exposure does not significantly influence their presence.

DI Rinsing Time, Dripping Time, & Drying Time: Shorter rinsing and dripping times resulted in improved cleanliness levels, with reduced levels of stains, smears, and whitish contaminants. Additionally, a higher drying time showed a slight reduction in the rate of contaminants.

Meanwhile, the hybrid cleaning simulations highlighted the following optimal cases:

DI Rinsing Method: The hybrid rinsing method significantly reduced the percentage of whitish contaminants and NSOL failure rate compared to immersion rinsing. The application of pressure during the rinsing process effectively minimized the contamination rate.

Rinsing Pressure and Speed: Higher belt speed and higher pressure during spray rinsing contribute to improved cleaning outcomes.

As a comparison, Table 5 presents a comparison of the optimal cleaning results for immersion-type flux cleaning and hybrid flux cleaning. The results show that for most contaminants such as stains, smears, and solder flakes, the occurrence becomes zero in both methods. However, the rate of whitish contaminants remains high at 17.10% for immersion cleaning, whereas the use of hybrid cleaning reduces the rate of whitish contaminants to 2.14%, resulting in a reduction of over 80%. Additionally, there is a significant improvement in the NSOL failure rate, which decreases from 5.4% to only 0.01%.

Table 5: Result of immersion versus hybrid flux cleaning.

Summary

In summary, the innovative hybrid cleaning approach combines the advantages of conventional solvent cleaning with advanced techniques to effectively address persistent contaminants. The results obtained from comparative studies demonstrate the superiority of this approach in achieving comprehensive contaminant removal, thereby enhancing the reliability and performance of semiconductor packaging. The successful implementation of the hybrid cleaning approach serves as a reference for future advancements in cleaning methodologies.

References

1. P. Jain, “Root Cause Analysis of Solder Flux Residue Incidence in the Manufacture of Electronic Power Modules,” India, 2011.
2. S. B. Awad and N. F. Awad, “High Intensity Ultrasonic Cleaning for Particle Removal,” in Particle Adhesion and Removal, R. Jaiswal and K.L. Mittal, Ed., USA, Scrivener Publishing LLC, 2015, pp. 203-239.
3. R. P. Prasad, Surface Mount Technology: Principles and Practice, Springer-Science + Business Media, B.V., 1997, pp. 619 – 622.