Integrated Microsystems

2012 ◽  
Vol 81 ◽  
pp. 55-64 ◽  
Author(s):  
Masayoshi Esashi ◽  
Shuji Tanaka

Technology called MEMS (Micro Electro Mechanical Systems) or microsystems are heterogeneous integration on silicon chips and play important roles as sensors. MEMS as switches and resonators fabricated on LSI are needed for future multi-band wireless systems. MEMS for safety systems as event driven tactile sensor network for nursing robot are developed. Wafer level packaging for MEMS and open collaboration to reduce the cost for the development are discussed.

Author(s):  
Amy Lujan

In recent years, there has been increased focus on fan-out wafer level packaging with the growing inclusion of a variety of fan-out wafer level packages in mobile products. While fan-out wafer level packaging may be the right solution for many designs, it is not always the lowest cost solution. The right packaging choice is the packaging technology that meets design requirements at the lowest cost. Flip chip packaging, a more mature technology, continues to be an alternative to fan-out wafer level packaging. It is important for many in the electronic packaging industry to be able to determine whether flip chip or fan-out wafer level packaging is the most cost-effective option. This paper will compare the cost of flip chip and fan-out wafer level packaging across a variety of designs. Additionally, the process flows for each technology will be introduced and the cost drivers highlighted. A variety of package sizes, die sizes, and design features will be covered by the cost comparison. Yield is a key component of cost and will also be considered in the analysis. Activity based cost modeling will be used for this analysis. With this type of cost modeling, a process flow is divided into a series of activities, and the total cost of each activity is accumulated. The cost of each activity is determined by analyzing the following attributes: time required, labor required, material required (consumable and permanent), capital required, and yield loss. The goal of this cost comparison is to determine which design features drive a design to be packaged more cost-effectively as a flip chip package, and which design features result in a lower cost fan-out wafer level package.


2007 ◽  
Vol 46 (4B) ◽  
pp. 2768-2770 ◽  
Author(s):  
Yoshiyuki Takegawa ◽  
Toru Baba ◽  
Takafumi Okudo ◽  
Yuji Suzuki

2016 ◽  
Vol 2016 (DPC) ◽  
pp. 000751-000773
Author(s):  
Craig Bishop ◽  
Suresh Jayaraman ◽  
Boyd Rogers ◽  
Chris Scanlan ◽  
Tim Olson

Fan-Out Wafer Level Packaging (FOWLP) holds immediate promise for packaging semiconductor chips with higher interconnect density than the incumbent Wafer Level Chip Scale Packaging (WLCSP). FOWLP enables size and performance capabilities similar to WLCSP, while extending capabilities to include multi-device system-in-packages. FOWLP can support applications that integrate multiple heterogeneously processed die at lower cost than 2.5D silicon interposer technologies. Current industry challenges with die position yield after die placement and molding result in low-density design rules and the high-cost of accurate die placement. Efficiently handling die shift is essential for making FOWLP cost-competitive with other technologies such as FCCSP and QFN. This presentation will provide an overview of Adaptive Patterning, a new technology for overcoming variability of die positions after placement and molding. In this process, an optical scanner is used to measure the true XY position and rotation of each die after panelization. The die measurements are then fed into a proprietary software engine that generates a unique pattern for each package. The resulting patterns are dispatched to a lithography system, which dynamically implements the unique patterns for all packages within a panel. For system-in-packages, this process offers a unique advantage over a fixed pattern: each die shift can be handled independently. With a fixed pattern, the design tolerances need to be large enough for all die to shift in opposing directions, otherwise yield loss in incurred. With Adaptive Patterning, vias and RDL features remain at minimum size and are matched to the measured die shift. The die-to-die interconnects are dynamically generated and account for the unique position of each die. Thus, Adaptive Patterning retains the same high-density design rules regardless of how many die are in a package. Adaptive Patterning provides the capability to use high-throughput die placement to drive down cost, while enabling higher-density system-in-package interconnect. With this technology the industry can finally realize the cost, flexibility, and form factor benefits of FOWLP.


Author(s):  
John M. Heck ◽  
Leonel R. Arana ◽  
Bill Read ◽  
Thomas S. Dory

We will present a novel approach to wafer level packaging for micro-electro-mechanical systems. Like most common MEMS packaging methods today, our approach utilizes a wafer bonding process between a cap wafer and a MEMS device wafer. However, unlike the common methods that use a silicon or glass cap wafer, our approach uses a ceramic wafer with built-in metal-filled vias, that has the same size and shape as a standard 150 mm silicon wafer. This ceramic via wafer packaging method is much less complex than existing methods, since it provides hermetic encapsulation and electrical interconnection of the MEMS devices, as well as a solderable interface on the outside of the package for board-level interconnection. We have demonstrated successful ceramic via wafer-level packaging of MEMS switches using eutectic gold-tin solder as well as tin-silver-copper solder combined with gold thermo-compression bonding. In this paper, we will present the ceramic via MEMS package architecture and discuss the associated bonding and assembly processes.


2019 ◽  
Vol 2019 (1) ◽  
pp. 000203-000210
Author(s):  
Burhan Ali ◽  
Mike Marshall

Abstract As the final step of IC fabrication, packaging is the process to encapsulate the chip and provide the interconnections for the I/O of the final form factor. The demand for increasingly higher I/O density, shrinking device size and lower cost that drive wafer processing also apply to the packaging process. Various technologies have been developed in order to achieve these goals with most of them being wafer-level packaging (WLP). Unlike traditional packaging process, most I/O interconnections are done at the wafer-level with redistribution layers (RDLs). RDLs are the layer where copper lines and vias form the electrical connections. Depending on the applications' market such as mobile, memory or the Internet of Things (IoT), fan-out wafer level packaging (FOWLP) provides the most promising method to support the I/O density requirements and fine RDL line/space. Moreover, fan-out panel level packaging (FOPLP) was also developed in order to capitalize on economies of scale and optimize substrate utilization. In this technology, a rectangular substrate is used in the process instead of a round-shape substrate like a wafer. Processes and equipment have long been developed for the wafer substrate market, but the previous developments cannot be directly applied to panel substrates. For instance, in the wafer line, spin on processes are very prevalent but these are not at all practical for a panel line. Some capital equipment manufacturers have been reluctant to embrace panel-level manufacturing due to the uncertainty as to whether it will prevail. Struggles with yield have been very common; some of which are due to die placement and others due to the lack of process control capabilities. With the explosion and adoption of FOWLP to enhance package shrinkage and performance the panel market becomes more and more viable. The companies that have embraced panel level manufacturing from the beginning have a distinct advantage due to their intimate knowledge and experience with the substrates as well as the relationship developed with capital equipment suppliers to develop the necessary technology in order to process the panels. However, there is still a great need to ensure the product mix deployed in panel form can have an acceptable yield; automated optical inspection and die placement metrology bridge that gap. Automated optical inspection allows for defect detection with traditional bright field (BF) or dark field (DF) illumination and also a new novel illumination technique that enables the detection of organic particles and/or residues that are often used in panel-level packaging processes. A system capable of macro defect detection with sub-micron capabilities allows for multi-purpose panel inspections. The system is also equipped with metrology capabilities for critical dimension and die placement measurements which meet the process node dimensional requirements. These features allow for process control of pick and place, overlay as well as feed-forward capabilities for die placement corrections. In a FOWLP/FOPLP process, chip first and chip last can be concluded among all available methods in the market. Die placement either start from the initial phase of the process or in the final phase of the process. In the chip first scenario, the chips are placed on a carrier by a pick-and-place system and then followed by an encapsulating molding process to reconstitute a substrate (reconstituted wafer or reconstituted panel). At this point a semi-additive process (SAP) is typically followed which includes a photo resist layer being coated, exposed and developed following copper (Cu) plating in order to form the redistribution layer. In this workflow, the die position are dominated by the accuracy of the pick-and-place tool and coefficient of thermal expansion (CTE) mismatch of the molding material and carrier. The trade-off between throughputs, placement accuracy and a feedback mechanism is the main impact from the pick-and-place tool in this process step. This affects both the chip first and chip last scenarios. The thermal expansion of the molding process not only adds additional die shift but also causes warpage of the reconstituted substrate that becomes an issue for automated handling systems and local process variation. Therefore, to know the actual die position and orientation after the die placement and molding process is crucial for matching with the following redistribution layers development. In one scenario it is possible to utilize the lithography system to perform die position metrology, however, this is time consuming and impacts the cost of ownership and overall throughput for the lithography process. A solution to this problem is provided by implementation of an optical metrology system. Since this information needs to be passed to the lithography tool in a usable manner for variable exposure positioning, the alignment of the stage coordinate system between the die metrology tool and lithography tool is a key point to ensure the correctness of the feed forward loop. For RDL development overlay between die and RDL via directly impact yield and are just as critical to the process as defect inspection and critical dimension measurements. Based on the corrections for each die, a yield prediction can be made and provides different strategies for the lithography tool's exposure field in order to balance throughput and exposure yield rate. In this paper, we demonstrate a solution using an automatic optical inspection (AOI) system to perform the die metrology for chip placement and RDL development in FOPLP and FOWLP. This includes die shift, die rotation, RDL inspection as well as the overlap between a reconstituted substrate and RDLs. This solution provides comprehensive coverage for packaging process control and significantly impacts yield optimization and throughput enhancement. With a multifunctional AOI system, it also reduces the cost of ownership for packaging processes.


2017 ◽  
Vol 2017 (1) ◽  
pp. 000721-000726 ◽  
Author(s):  
Chet Palesko ◽  
Amy Lujan

Abstract Fan-out wafer-level packaging (FOWLP) and embedded die packaging offer similar advantages over traditional packaging technologies. For example, both packages can be quite thin since the die is placed early in the manufacturing process and the package is fabricated around the die. This is in contrast to traditional packaging technologies, in which the package is fabricated first, and then the die is placed on top of the package. This results in a thicker package compared to fabricating the package around the die. Due to the ongoing miniaturization market requirements, thinner packages are becoming increasingly important. Both FOWLP and embedded die packaging also provide the capability of placing multiple die and passives in a single package. This capability can have both size and performance benefits since the interconnect distance between the embedded components is shorter. In this paper, the cost and cost drivers of FOWLP and embedded die packaging technologies will be compared. Activity based modeling will be used to characterize the cost of each activity in the two manufacturing flows.


2016 ◽  
Vol 2016 (1) ◽  
pp. 000180-000184 ◽  
Author(s):  
Chet Palesko ◽  
Amy Lujan

Abstract Fan-out wafer-level packaging (FOWLP) offers many significant benefits over other packaging technologies. It is one of the smallest packaging options, but unlike fan-in wafer-level packaging, the IO count of FOWLP is not limited to the area of the die. Given these advantages, FOWLP continues to grow in popularity. While the cost of FOWLP is usually reasonable, there are still opportunities for future cost reduction. Many FOWLP suppliers are exploring panel-based manufacturing instead of the current wafer-based approach. Since many more packages can fit on a large panel than on a wafer, the cost per package can be reduced. The surface area of a 370mm × 470mm panel is 1,739 sq.cm. compared to 706 sq.cm. for a 300mm wafer. This means more than twice as many packages can be manufactured on a single panel. However, this does not mean that the cost per package will be cut in half. Many of the costly manufacturing activities do not depend on the surface area of the panel or wafer and they will not be affected by a larger panel. This paper analyzes the current cost of FOWLP activities and highlights which activities will benefit from a move to panels. An analysis of each manufacturing activity is presented comparing the cost impact of panel versus wafer. The total potential cost savings is also presented.


Author(s):  
Yi Tao ◽  
Ajay P. Malshe ◽  
W. D. Brown

In this work, low temperature selective solder (Pb37/Sn63) bonding of silicon chips or wafers for MEMS applications using a continuous wave (CW) carbon dioxide (CO2) laser at a wavelength of 10.6μm was examined. The low reflectivity, fair transmittance, and high absorptivity of silicon at the 10.6μm wavelength led to selective heating of the silicon and reflow of an electroplated or screen printed intermediate solder layer which produced silicon-solder-silicon joints. Finite element simulations were carried out to optimize the process parameters in order to achieve uniform heating and minimum induced thermal stress. The bonding process was performed on the fixtures in a vacuum chamber at an air pressure of one milliTorr to achieve fluxless soldering and vacuum encapsulation of silicon dies. The bonding temperature at the sealing ring was close to the reflow temperature of the eutectic lead tin solder, 183°C. Pull test results showed that the joint was sufficiently strong and could not be separated before the silicon die broke. Helium leak testing showed that the leak rate of the package was below 10−8 atm · cc/sec under optimized bonding conditions. The results of the Design of Experiment (DOE) method indicated that both laser incident power and scribe velocity significantly influenced bonding results. This novel method is especially suitable for vacuum bonding wafers containing MEMS and other micro devices with low temperature budgets where managing stress distribution is important. Further, sealed encapsulated and released wafers can be diced without damaging the MEMS devices at wafer scale.


Author(s):  
Luke Prenger ◽  
Xiao Liu ◽  
Qi Wu ◽  
Rama Puligadda

Multifunctional materials are a relatively new topic in the semiconductor industry for wafer-level packaging (WLP). With the increase in processing steps and the emergence of more advanced technologies, the use of multifunctional materials will become a more integral part in the future of temporary bonding and debonding (TB/DB) as well as other advanced packaging applications. One approach to multifunctional material design incorporates adhesive and laser release attributes in one material layer. Although this is similar to a thermal release material, it has greater thermal capabilities due to its ability to be cured and undergo laser debond. Many advantages may be obtained by combining a curable adhesive and laser release layer into one material. One of the greatest advantages is the reduction in overall processing time and steps required to bond wafer pairs as well as the reduction of chemical waste, due to the use of one material compared to two or more materials which significantly reduces the cost of ownership. Curable adhesive single layer systems offer access to higher temperatures with less material flow from the curable layer, strong adhesion for high stress applications where wafers can delaminate or spontaneously debond when using multilayer mechanically debonding systems such as Fan-Out Wafer Level Packaging (FOWLP), and offer lower wafer stress and warpage due to fewer material interfaces within the bonded wafer pairs causing less potential mismatch of materials coefficient of thermal expansion(CTE). Some challenges with this concept stem from the concern of the cleanability of a curable layer and potential laser damage to the device. In order to wet clean a curable layer, which is usually very solvent resistant due to the crosslinked nature, requires harsh solvent based solutions (that may contain either strong acid or base, require long cleaning time, and high temperature). This study will address all of the aforementioned challenges and includes the developmental advancements in material designs that resulted in the creation of new multifunctional materials. These multifunctional materials have been designed to be thermally curable, prevent material reflow of the bonding layer at higher temperatures, while still remaining wet cleanable without the use of harsh chemicals and long times. As with any material that utilize laser release methods there are concerns about device damage from laser energy penetrating to the device but multifunctional materials address this in two ways: they offer high absorbance of the laser energy at all commercially available laser tool wavelengths and they can be utilized as a thicker film as they act as the bonding layer as well. By overcoming their challenges, they will minimize the cost of ownership while driving advancement in future materials and processing.


Sign in / Sign up

Export Citation Format

Share Document