Dec 24, 2024

Intense Pulsed Light for Next-Generation Soldering Technology

Energy efficiency is always a popular topic in engineering. Reducing power consumption in manufacturing not only reduces cost, but also reduces CO2 emission and other environmental impact.

Soldering is one of the key technology nowadays for semiconductor manufacturing. The soldering process, as known as surface mounting technology (SMT), is to use solder materials (e.g., Tin) to create electric interconnection and mechanical supports between printed circuit boards (PCBs) and electronic packages. The essential step for SMT is to heat and melt the solder material to about 200 degC. And in the past two decades, this step has been done by the convection oven, which is commonly referred to as the reflow process.

According to a recent research³, SMT can consume ~0.26 kWh to produce a single PCB, with 97.97% of energy used by reflow process. Based on these values, when producing an iPhone 16, the energy cost for its PCBs can power the device for 650 hours, or equivalent 27 days of continuous usage.

Starting in 2020s, several scientists in South Korea started to investigate using intense pulsed light (IPL) for soldering process. This method uses xenon flash lamp to generate high-power broadband radation, which has been previously used for cosmetology and thermal therapy. Compared to reflow soldering, IPL soldering significantly reduces the power consumption by 80%. Besides, the processing time of IPL is only 8-12 seconds, compared to 5-10 minutes for reflow. The ultra-short processing time not only benefits for mass production, but also creates a transient temperature profile, which can substantially reduce thermal damage and increase product reliability.

Given all these advantages, our work aims to deepen the understanding of IPL soldering and to hopefully optimize the pulse condition in order to maximize the product quality and accerlerate mass production.

To achieve this, we built our preliminary model for our simulation. This includes a multiplayer PCB (prepreg and copper), silicon package, and solder balls. We put this device into our IPL chamber, which has similar configurations as the experimental setup in Samsung. Here you can see a cylindrcal light source and a semi-cylindrical ceiling to provide vertical reflective light. Besides, the chamber is made of polished stainless steel and has an air flow of 3m/s. Furthermore, the ambient temperature is held at 30 ℃.

Now let’s go to the details of the IPL source. The following numbers have been validated in several experimental papers: The total power is 1.17Megawatt, which is driven by high voltage source (1500V). The IPL pulse duration is 10ms and the frequency is 2Hz. These values consider the operation constrains, such as additional time for charging and cooling. Finally, to reach the melting temperature, a total of 12 cycles is assumed.

Here is the simulation result showing the cross-section temperature profile during the IPL window. You can see the silicon package has significant temperature rise on its top surface. And spcifically, the corner region has the highest temperature because of the most surface exposure, and the central bottom region has the lowest temperature because of the least exposure.

Next, this is the temperature profiles during the relaxation. Note that passive heat diffusion happens in the silicon package, and as a result, the temperature difference across the package is decreasing. In fact, silicon has high thermal conducitivity of about 130W/mK, which reulsts in signfiicant conduction throughout the system.

This image shows the thermal profile over many cycles. And after 12 cycles, the package temperature reaches about 193degC.

Below is the image of the vertical temperature gradient as a function of time. During IPL window, you can see large temperature gradiaent happens at the package component. However, during relaxation, such gradient is diminished because a large amount of heat is being conducted through the bottom components. You may also find another vertical gradient existing inside the multilayer PCB close to the solder. This is because the first few multilayers inside the PCB tend to stop the heat flux from further penetrating through bottom.

This image shows the vertical temperature change over many cycles. Although the overall temperature is increasing, its gradient is quite stable and the PCB typically has 10-15 ℃ lower than the package temperature.

Next, we further investigate the transient temperature at specific points. We use point 1-3 to denote different locations inside the package, 4-7 inside the solder balls, and 8-12 inside the PCB. According to the results, the package is a part very sensitive to IPL. More specifically, its temperature profile changes dramatically when the IPL is on and off. On the other hand, for the solder joints, their heating profile is less sensitive to IPL pulse. This tells us the fact that although IPL soldering is powered by radiation, its major heating mechanism for solder joints is heat conduction. Now if we take a look at multi-layer PCB, things can be more interesting — those points with direct light exposure will be radiation dominant, while those points shadowed by others will be conduction dominant.

We further validate our simulation results with a 1D analytical model. Assuming the silicon package is a 1D semi-finite solid, we can refer to a well-established solution under a radiation boundary. However, because IPL is an electromagnetic source instead of a perfect blackbody, an equivalent T_ambient is required, which is equal to about 4300K. With all the information gathered, now we can predict how much temperature incrase in the surface during a IPL pulse, which is approximately 33.5K. Also, we can estimate the diffusion length, which is about 860um, a number that is way less than the package thickness. This small penetraion provides great benefit because it avoids heating up the bottom components which can otherwise cause mismatch in thermal expansion coefficient. Over time, this can give less thermal stress and enhance the product reliability.

Now that we know about the basic mechansim of IPL heating, we want to optimize the pulse conditon to maximize the package reliability while offering the good heating rate. Rembmer we mentioned the maximum and minimum temperature across the chip. It turns out that this is an important parameter to influence the product reliability. Now let’s define the temperature non-uniformity as the difference of these two values, which is equal to 29.5K in this image.

Here, we investigate multiple parameters In order to reduce this number. We summarize the parametric studies in the below table, Without changing the total power, the first two settings change the frequency and single pulse power, the third and fourth settings change the freqnecy and pulse duration, and the last two settings change the chamber conditions.

Firtst let’s take a look at the reference setting. The left image shows the temperature non-uniformity as a function of time. It is found that the greatest non-uniformity occurs at the moment when IPL stops. And over many cycles, this peak value remains similar to each other, which has a maximum number of 31.7K.

Next, we compare this non-uniformity in different settings. It is observed that higher frequency and lower pulsed power can lead to smaller temperature non-uniformity. The best scenario is setting #2, which has a non-uniformity of 6.8K.

We further compare the heating profile of solder joints in these settings. As long as the chamber conditions remain the same, final temperature are similar. Furthermore, the lowest final temperature of solder joints happen in the setting #5, where the chamber emissivity is set at 0.9. This indicates the importance of surface polishing, as a rough surface will take away radiation energy.

As we reach the end, I’d like to summarize our key findings. Through this research, we have demonstrated that IPL soldering is a highly effective method for localized heating on silicon packages, which significantly reduces risk of thermal damage and warpage. Although IPL soldering is driven by radiation, its primary heating mechanism for solder joints is heat conduction. Additionally, our parametric study reveals that a combination of higher frequency and shorter pulse duration can lead to better temperature uniformity and improved product reliability.

Beyond today’s topic, I’d like to share a broader perspective on the future of thermal management. The future trend is clearly moving towards solutions that are more precise, localized, dynamic, and intelligent. For soldering, we are moving away from steady-state heating using massive ovens and embracing advanced methods like IPL. Similarly, in electronic cooling, we are replacing bulky heat sinks with localized solutions such as microchannels and thermoelectric cooling. As we advance, thermal management will not only focus on performance but also on environmental sustainability—an essential consideration for the future.

1. https://www.indicelectronics.com/es/services/manufacturing/assembly-lines/bga ↩︎
2. Micron, BGA Manufacturer’s User Guide for Micron BGA Parts ↩︎
3. Semsri, A, International Journal of Sustainable Energy 43, no. 1 (2024): 2372567. ↩︎
4. https://www.smthouse.com/reflow-oven-system/ ↩︎
5. Zain, S, International Conference on Innovative Technology, Engineering and Science. Cham: Springer International Publishing, 2020. ↩︎
6. Samsung electronic, South Korea. ↩︎

–Modified from my research talk in PRTEC, Dec 19th 2024

Go to Top