3D Sensing Package Solutions

By Chiung Lee, Weilung Lu, and Adrian Arcedera

3D sensing technology is rapidly being adopted in a variety of growing markets. End-product applications include smartphones, tablets, augmented/virtual reality products, robot vacuum cleaners, industry inspection machines, and automotive vehicles. As shown in figure 1, Yole Développement expects the 3D imaging and sensing market to expand from $6.8B in 2020 to $15.0B in 2026, at a 14.5% compound annual growth rate (CAGR).

Fig. 1: Source: 3D Imaging and Sensing – Technology and Market Trends 2021, Yole Développement, 2021.

Today’s mobile market accounts for the dominant shipment volume of 3D sensing in flagship smartphones to enable authentication, enhance image quality, and provide in-depth measurement functionalities. Automotive applications use a gesture recognition sensor for convenience and safety through an in-cabin camera to monitor the driver’s status. In addition, LiDAR detects objects and measures distance to enable land departure warning and enable lane keeping assistance and automatic emergency braking safety functions.

3D sensing functions likes the human eye. A light source, such as the sun, illuminates an object and reflects a full spectrum of light. Human eyes capture the reflected light from the object but only a limited portion of the spectrum. The brain processes the received data and forms an image of the object in its relative location. Similarly, a typical 3D sensor consists of a light emitter (a laser diode or infrared LED) for transmission (Tx), a CMOS image sensor/single-photon avalanche diode (CIS/SPAD) to receive (Rx) the reflected light (photons) and an image processing IC to execute sophisticated algorithms to decipher the received patterns and produce a depth image of the scene.

There are three major approaches for 3D sensing: stereo vision, structured light, and the time of flight (ToF) method. Stereo vision uses two cameras to simulate human eyes by identifying the differences between two images. It requires higher computing at the system level while providing slower response time as well as higher power consumption. Structured light mainly refers to pattern imaging. A diffractive optical element (DOE) projects the specific pattern to the objects. For high accuracy, the alignment between the light source during the assembly is quite critical. This technology has widely applied to object and facial recognition. However, based on the complexity of software and hardware integration, only a few suppliers have the capability to provide this approach. Finally, a ToF sensor measures the distance by detecting the phase shift of the light signal that hits the subject and returns to the sensor. In comparison to other two 3D sensing techniques, ToF is relatively cheap, has fast response time, and consumes less power.

For specific applications, different 3D approaches are chosen based on cost, system complexity, accuracy, power consumption, detection range, and other considerations as shown in figure 2. However, the technology trend is gradually moving from stereo vision and structured light to ToF. As explained previously, the design challenges for these three technologies are different, requiring different packaging solutions for each.

Fig. 2: Source: 3D Imaging and Sensing – Technology and Market Trends 2021, Yole Développement, 2021.

In general, an optical sensor packaging assembly requires more accurate alignment, uses specific optical materials, and needs higher environment cleanliness control to avoid particles. For mobile and consumer applications, the cost impact, time to market, and miniaturization must also be addressed.

There are 3 main challenges to 3D sensing assembly design:

1. Ambient or background illumination noise: ambient light (especially sunlight) may provide uncorrelated photons.
2. Optical crosstalk from cover glass/plastic: the device’s cover glass/plastic may lead to optical crosstalk (noise). This makes the sensor receive two targets – one reflected from the 3D object and one from the cover glass/plastic.
3. Optical performance and element accuracy: this is key item related to design to integrate the emitter and receiver. It also relies on the package manufacturer’s ability to provide heterogenous integration and accurate alignment capability.

Today’s mainstream packaging solutions re-use a camera module-like structure which leverages the capability and equipment of the module house. However, to reduce and minimize the module size to fit into a mobile phone and other small form factor portable devices, package-level assembly integration will be essential sooner or later. With their experience in System in Package (SiP) module and microelectromechanical systems (MEMS) sensor packaging, outsourced semiconductor assembly and test (OSAT) suppliers will play an important role in this package-level integration.

Figure 3 shows one example of 3D sensor integration. The sensor (receiver) can be integrated with a vertical-cavity surface-emitting laser (VCSEL) driver in one die and another stand-alone VCSEL die (transmitter) is attached to the same substrate. A lid or holder assembly for the substrate creates two isolated compartments for the two dies to avoid crosstalk. Lenses or filters can be attached to the lid to provide the designated optical properties in terms of specific wavelength or required optical signal pattern.

Fig. 3: Optical sensor integration reduces the number of components.

Current optical sensor chip designs require custom package solutions which have assembly packaging and test challenges. For automotive applications, stringent reliability requirements increase the development time to select the right materials, perform process evaluation, optimize process parameters, and complete reliability verification testing. Following this concept, the standardization of the packaging solution becomes feasible to deliver products to achieve acceptable time to market. Either molded cavity or attaching a lid to create a cavity can be supported by the OSAT’s toolbox and capabilities. Figure 4 provides a comparison of the two approaches.

Fig. 4: A comparison of two 3D packaging techniques.

Molded cavity packages utilize film assistant molding (FAM) technology. Combined with specific optical glass as a window or filter and transparent die attach film (DAF) to attach the sensor die, the small footprint compact package with its relatively lower cost can perfectly fit consumer and mobile applications. On the other hand, for devices that need specific distance between sensor/receiver die and lens/filter such as DOE of structured light sensors, cavity packages with a lid will be the preferred solution. Both the molded solution and lid cavity solution can integrate receivers and transmitters to provide more flexibility for specific customer designs.

Conclusions

There are many customized packaging solutions for 3D sensors today. Moving forward, a standardized platform following the MEMS sensors’ path will be the preferred approach. Meeting the expected optical performance in terms of assembly accuracy, material selection, cost control and reliability are all the keys to success.

Learn more about Amkor’s 3D Sensing packaging solutions:

• MEMS
• Optical
• System in Package

Weilung Lu is a senior director for MEMS and Sensor Products at Amkor Technology.

Adrian Arcedera is a senior vice president for MEMS and Sensor Products at Amkor Technology.