Analysis of SiP (System in Package)

SiP can not only assemble multiple chips but also serve as a dedicated processor, DRAM, flash memory, and passive components combined with resistors and capacitors, connectors, antennas, etc., on the same substrate. This means that a complete functional unit can be built in a multi-chip package so that a small number of external components need to be added to make it work.

system in package

Huawei, Xiaomi, OPPO, VIVO, and Samsung have released 5G phones one after another, and the sales of 5G phones have exceeded expectations. Millimeter wave 5G phones will increase the demand for SiP. Apple AirPods have added noise-canceling function, following Apple watch, which also uses SiP technology.

In general, SoC only integrates AP-based logic systems, while SiP integrates AP+mobileDDR. In a way, SiP = SoC+DDR. As the integration level gets higher in the future, eMMC is likely to be integrated into SiP as well. Chip development has shifted from the pursuit of power consumption and performance improvement (Moore's Law) to a more pragmatic approach to meet the needs of the market (beyond Moore's Law).

Ⅰ New technology beyond Moore's Law era, SiP package technology

The demand for thin and light phones and high performance is driving system-level integration. Cell phone users need both continued improvement in performance and increased functionality, as well as portability. These two mutually constraining factors have shaped the smartphone replacement process over the past decade.

(1) Thin and light. Take the iPhone for example, from the earliest body thickness of about 12mm to the iPhone XS 7.5mm, however, the iPhone11 thickness increased to 8.5mm.

2) Increased functionality and performance. Cell phones have gradually added new functions such as multi-camera, NFC mobile payment, dual card slot, fingerprint recognition, multi-cell, face unlock, ToF, etc. The performance of each component has also continued to improve, and the expansion of these functions and performance improvement has led to an increasing number of components, taking up more space inside the phone and also requiring more power consumption. However, the energy density of Li-ion batteries in cell phones is increasing slowly. Therefore, space-saving modularity and system-level integration are becoming the trend.

Some cell phone manufacturers have already released finished models, but the implementation of 5G features brings obvious challenges to the "thin" appearance of the phone, and even power consumption should not be underestimated. Lenovo released its 5G phone MOTO Z3 as early as August 2018, but its 5G functionality relies on a 5G module mounted on the back of the phone with its own 2000mAh battery.

Samsung officially released the 5G version of the S10 at the end of February 2019, and Huawei also officially released the folding screen 5G phone Mate X in March, of which the Huawei Mate X, with a thickness of only 5.4mm, ended up with a Leica triple camera, 5G baseband and four sets of 5G antennas placed on the side bulge. From the above phones, the implementation of 5G features still poses a clear challenge to the "thin and light" appearance of the phone.

Functional integration is the mainstay of both System-on-Chip SoC and System-in-Package SiP. The goal of both is to achieve a high degree of integration of multiple system functions in the same product, where SoC is designed and manufactured from the perspective of traditional Moore's Law-driven semiconductor chip process to integrate the functional components required for a system into a single chip, while SiP is packaged and assembled from the perspective of advanced back-end packaging and high-precision SMT process to integrate several bare chips and micro-passive devices manufactured by different integrated circuit processes. SiP, on the other hand, integrates several bare chips and miniature passive devices made by different IC processes into the same small substrate and forms a high-performance miniature component with system functions from a packaging and assembly perspective.

The number of components that can be integrated per unit area is getting closer to the physical limitations due to the limit of Moore's Law. SiP packaging technology enables higher integration and better performance of the combined system, which is the inevitable path to surpass Moore's Law.

Compared to SoC, SiP has two advantages.

(1) SiP technology is more integrated but has a shorter R&D cycle. SiP technology can reduce the repetitive packaging of chips, reduce layout and alignment difficulties, and shorten the R&D cycle. The 3D SiP package with chip stacking can reduce the amount of PCB board used and save internal space. For example, about 15 different types of SiP processes are used in iPhone 7 Plus to save space inside the phone. SiP process is suitable for communication and consumer product markets with short update cycles.

(2) SiP can solve the problem of heterogeneous (Si, GaAs) integration. Different components of cell phone RF systems often use different materials and processes, such as silicon, silicon-germanium (SiGe) and gallium arsenide (GaAs), and other passive components. Current technology is not yet able to these different process technologies to create parts in a silicon single crystal chip. However, the SiP process allows the application of surface mount technology SMT integration of silicon and GaAs bare chips, as well as the use of embedded passive components, making it very cost-effective to make high-performance RF systems. The miniaturization of optoelectronic devices, MEMS, and other special process devices will also be heavily used in the SiP process.

In the past few decades, the electronics manufacturing industry has formed three distinct segments: wafer manufacturing, packaging, and system assembly, represented by TSMC, NLM, and Hon Hai, whose manufacturing precision is nanometer, micron, and millimeter respectively. With the increase in integration of consumer electronics products, the precision requirements of some modules and even system assembly are approaching the micron level, resulting in process overlap and business competition or synergy with the packaging and testing segment.

Specifically, the SiP process integrates molding and singulation processes in traditional packaging and testing with SMT and system testing processes in traditional system assembly.

Ⅱ SiP becomes the standard technology for cell phones in the 5G era

1. 5G cell phone explosion period is coming

5G commercialization is getting closer and closer, and operators in mainstream countries around the world have defined the time point. As of October 16, 2019, Huawei has signed more than 60 5G commercial contracts with leading operators around the world, and more than 400,000 5G Massive MIMO AAUs have been shipped around the world. Worldwide, most telecom operators in mainstream countries plan to start deploying 5G networks and gradually launch commercial services during 2019-2020.

According to CCS Insight, 5G cell phone shipments will reach 10 million units in 2019, accounting for 0.6% of cell phone shipments, and will exceed 900 million units in 2023, accounting for half of cell phone shipments.

High growth of 5G mobile phones

2. SiP is increasingly used in 5G cell phones

For historical reasons, the low-frequency band below 3GHz for public mobile communication has been basically divided up by previous generations of communication networks, and the frequency band is scattered, unable to provide the continuous large bandwidth required for 5G, so 5G must extend to a higher working band. At present, there is a worldwide consensus on the spectrum of 5G, and the middle band of 3~6 GHz will become the core working band of 5G, mainly used to solve the problem of seamless coverage in a wide area, while the high band above 6 GHz is mainly used for local supplementation to provide ultra-high data transmission service for users in hot areas under better channel conditions. Applications are also gradually converging to a consensus that the frequency band of 5G is divided into two parts: Sub-6 and millimeter-wave.

5G requires SUB-6 and millimeter-wave two sets of radio frequency systems

5G cell phones need to integrate more RF devices. The cell phone RF module is mainly to achieve the reception, processing, and transmission of radio waves. The key components include antenna, RF front-end, and RF chip, etc... The RF front-end includes antenna switches, low-noise amplifiers LNA, filters, duplexers, power amplifiers, and many other devices. From a single communication system in the 2G era to the smartphone era, which is compatible with 2G, 3G, 4G, and other wireless communication systems, the number of devices in the RF front-end of cell phones is increasing, and the performance requirements are getting higher and higher.

The number of RF devices required for 5G cell phones will far exceed that of the previous generation, and the complexity of the structure will increase significantly. 5G cell phones need to be forward-compatible with 2/3/4G communication systems, and the number of RF front-end modules required for a single device will increase significantly. According to Qorvo forecast, 5G single handset RF semiconductor usage will reach $25, compared to 4G cell phones nearly doubled. Among them, the receiver/transmitter filters from 30 to 75, including power amplifiers, RF switches, frequency bands, etc. have at least doubled the number of growth. The significant increase in the number of devices will significantly increase the complexity of the structure and raise the level of package integration required.

The 5G frequency band is divided into two parts: Sub-6 and millimeter-wave. The performance of the Sub-6 part of the signal is more similar to that of LTE signals, and the difference in RF devices mainly lies in the increase in quantity, while the millimeter-wave part brings revolutionary changes in RF structure. SiP technology will be increasingly used in 5G cell phones and play an increasingly important role.

1) Step 1: 5G needs to be compatible with communication technologies such as LTE, which will require more RF front-end SiP modules.

2) Step 2: millimeter-wave antennas and RF front end to form AiP antenna modules.

3) Step 3: more parts such as baseband, digital, memory, etc. are integrated into larger SiP modules.

3. 5G millimeter-wave increase AiP demand

5G millimeter-wave band requires more RF front-end devices. The characteristics of the versatile consumption require the distance between the radio front end devices and the antenna to be shortened. The millimeter-wave antenna size can be reduced to 2.5 mm. At the same time, the effect of high-frequency radiation on the shield antenna on the peripheral circuit. The above needs need to set the antenna to the RF device into a module, and the antenna size is small, providing a guarantee for the feasibility of the module.

The millimeter wave phone requires more RF front ends and antennas. Millimeter wave high frequency communications will need to integrate more than 3 amplifiers and dozens of filters. In addition, millimeter wave communications requires a smaller size and quantity antenna. The general antenna length is 1/4 of the radio wavelength. Once the operating frequency band of 30 GHz is used, it means that the wavelength will be less than 10 mm, the corresponding antenna size is 2.5mm, and it is 1/10 of the 4G era. At the same time, due to the high frequency communication loss, the coverage ability is weak, thereby introducing more antennas, and forms an antenna array through MIMO technology to enhance coverage. According to QROVO forecast, the number of antennas of single 5G mobile phones is expected to reach 10-12.

Qualcomm has already commercialized the 5G millimeter wave antenna module AiP standard QTM052, and the Samsung Galaxy S10 5G millimeter wave version of the phone uses three of these antenna modules, placed on the top, left, and right inside the center frame. Multiple antenna modules can avoid signal interference from different user's hand grip positions.

Antenna performance varies greatly depending on the design of the phone, space constraints inside the phone, and the structure or substrate material next to the antenna. Standardized AiP antenna modules are difficult to meet the different needs of different handset manufacturers. Apple and other manufacturers are expected to develop their own customized AiP antenna modules based on their own phone designs. We estimate that Apple's AiP demand alone is expected to reach billions of dollars in 3 years. 3.

Ⅲ Apple wearable products actively use SiP technology

Wearable products are IoT products to which Apple attaches great importance, and Cook believes that the health business based on wearable products will become Apple's biggest contribution to human beings. Apple has developed three health-related platforms, ResearchKit, HealthKit, and CareKit, and is working with Stanford University School of Medicine and others to promote the integration of wearable products with healthcare.

The Apple Watch is a complex device with nearly 900 components in a small space and has been using the SiP process since its first generation in 2015. The SiP module of Apple Watch integrates most of the functional components of the Apple Watch, including CPU, storage, audio, touch, power management, WiFi, NFC, and other 30 separate functional components, more than 20 chips, more than 800 components, and only 1mm thick.

APPLE Watch uses shaped SIP technology

AirPods' ordinary version is relatively simple and did not use SiP technology in the early days, but the AirPods Pro released at the end of October has active noise cancellation, which requires the integration of more parts, and also uses SiP technology. We calculate that AirPods are expected to generate billions of dollars in SiP demand.

With the development of technology, global electronic products are gradually moving toward multi-functional integration and low-power design, thus making SiP technology, which can integrate multiple bare crystals in a single package, increasingly attract attention. Chip development has shifted from the pursuit of lower power consumption and higher performance (Moore's Law) to a more pragmatic approach to meet market demands (beyond Moore's Law). The demand for maximizing chip performance, minimizing size after packaging, and customizing tailoring is rapidly rising, and SiP integration technology has become one of the most important technologies in the semiconductor industry.