In The Spotlight: What Is Responsible For The Surging Demand For CIS?

After TSMC announced plans to construct a new fab in Arizona, the Taiwan-based company disclosed that they are considering building new fabs in Japan and Germany. While the Arizona fab will focus on producing 5nm nodes using extreme ultraviolet lithography (EUV) technology, the new plant in Japan reportedly would focus on the 28nm node. This 28nm fab in Japan would be in addition to a 28nm fab expansion in China.

Given that the latter node was introduced in 2008 and is not regularly used today to build central processing units (CPU) and graphics processing units (GPU), the question arises, why is TSMC building not one, but three new fabs centered around 28nm node production? The answer is simple: customer demand.

And in this case, that demand is powered by devices and applications that use augmented reality (AR) and virtual reality (VR). Most of the devices, including those that use CMOS image sensors (CIS), are manufactured on 28nm to 80nm node technology. This is why major foundries, including TSMC and Samsung, are preparing to ramp up their volume production for these more mature nodes.

As a further illustration of the demand for the 28nm node, consider this: Apple is planning on manufacturing compact, high-resolution, micro organic light-emitting diode (OLED) display devices on silicon wafers, and Sony is planning on building image signal processing (ISP) devices; both companies will be using 28nm node technology.

AR overlays digital content and information onto images of the physical world captured by camera, and it is one of the biggest technology trends now. Apps like Snapchat and games like Pokémon GO first popularized AR, but the technology is predicted to become a part of our daily lives, influencing how we shop at brick-and-motor stores or drive (or not drive in the case of autonomous vehicles) our cars.

VR, meanwhile, is already widely used at work and home. While some gamers have embraced VR with a passion, advanced manufacturers regularly use VR to train employees. As for my company, Onto Innovation, we adopted and started using VR technology — in this case, Oculus from Facebook — for training and field support during the pandemic since some people could not meet face-to-face due to travel restrictions.

These incredible advancements in the use of AR and VR wouldn’t be possible without CIS. And with both AR and VR growing in popularity, there is no doubt that CIS applications will increase in the future.

So far the demand for smartphone cameras has been primarily responsible for the growth of CIS, but future growth will be led by applications using AR and VR technologies, such as wearable devices like Google Glass and eventually autonomous vehicles.

Like other semiconductors, CIS is manufactured on silicon wafers and follows the same back-end packaging processes of grinding, sawing and electrical testing. A typical CIS device has an active pixel sensor (APS) region in the center of the die with electrical I/Os (bond pads) on the periphery. Deionized water is often used to clean up the mobile contamination left behind during the wafer thinning or die singulation process; this has an inherent risk of staining or leaving a residue on the APS that affects the quantization of light and is considered a killer or yield-limiting defect. And this is where we get into one of the key challenges of CIS inspection.

Fig. 1: CIS images show water stains.

CIS is different from other semiconductors because it is designed to absorb light, which it converts to an electrical signal. The light-absorbing nature of the APS makes visual inspection very difficult due to the fact that visual inspection relies on reflected light to produce an image. Under normal bright field illumination, an APS looks extremely dark. As a result, defects such as film variation or so-called watermarks are not visible.

Fig. 2: Low-contrast water stain is not visible on raw image (left); low-contrast water stain is visible after image processing (right).

Even if these defects were visible under bright field illumination, APS defects like watermarks have a low-contrast from the background, and the traditional image comparison detection method cannot find such low-contrast defects. Fortunately, there are products that are capable of detecting defects like watermarks already on the market.

With the growth rate of CIS predicted to increase from 6% to 9% over the next several years, adopting systems capable of finding low-contrast defects on light-absorbing semiconductors like CIS are more important than ever.