Dispensing Advantages for MEMS Wafer Capping

by Akira Morita, [Nordson ASYMTEK ]

Microelectromechanical systems (MEMS) integrate mechanical and electrical elements on a Si chip using semiconductor microfabrication. There are many applications for fluid dispensing in MEMS device manufacturing. In MEMS packaging, one area where dispensing plays a significant role is in wafer capping MEMS before dicing.

MEMS are extremely fragile. Imagine how small and sensitive they must be as they are used for applications as intricate as measuring capacitance by detecting nanometer distance change. The slightest touch, vibration, or contamination can damage them. The dicing process is full of obstacles because the flow of water used in dicing, and the particulates discharged in the process, can destroy or contaminate the MEMS. Therefore, MEMS devices need a protective covering to shield them during the dicing process. Thus, wafer capping is done before dicing.

MEMS wafer capping is not a new process; it's been around for about 10 years. However, back then, most of the MEMS devices were used in applications where the package had to be hermetically sealed and extremely reliable. The process was quite costly, so wafer capping wasn't a common practice. Today, MEMS are used in many more applications, especially in consumer products like microphones in mobile phones and the mouse for PCs. Since these new applications do not require that MEMS devices be hermetically sealed, new manufacturing methods and challenges have arisen. One of these new methods is dispensing the sealant and adhesives in the wafer capping process. Jet dispensing systems have been designed to specifically handle MEMS wafer capping.

The Wafer Capping Process

The package structure for MEMS wafer capping consists of a substrate, a wafer with MEMS, and a glass wafer (or another silicon wafer) containing the cavities that cap the MEMS. Die-attach adhesive, such as silver epoxy, is dispensed in thin horizontal and vertical lines onto a film-frame wafer to attach the MEMS devices. A layer of sealant is jetted around the rim of each cavity on the glass wafer. The sealant can be a UV cure adhesive, low-temperature solder, covalent, or a glass frit. The MEMS structure is then capped with the glass wafer. A UV light penetrates the glass wafer to cure the adhesive. The wafer is then diced. The cap keeps the MEMS protected during the wafer dicing process (Figure 1).

Figure 1: Schematic of the MEMS wafer capping process flow. (courtesy of Prismark Partners)

The equipment used in MEMS wafer capping is a dispensing work cell that consists of an automated fluid dispensing system integrated with a same-side film-frame loader/unloader. This single work cell minimizes the footprint and maximizes throughput. The system is fully enclosed with interlocked doors and windows (Figure 2).

Figure 2: Integrated workcell with a loader and dispenser for MEMS wafer capping.

MEMS are 1 to 100µm in size, and MEMS devices generally range in size from 20µm to 1mm. The volumes of fluid dispensed must be at the micron level and dispensed precisely in very thin lines. The corners must be extremely defined so each cavity section covers the MEMS, ensures the MEMS device is well seated, and the glass connects properly to the wafer. Volumetric control must be maintained to ensure fluid deposition is accurate and within tight tolerances. Many of the end products that contain these devices are manufactured in large volumes, so wafer handlers, automated fluid dispensers, and the software that drives them must automatically monitor and control fluid viscosity, dispense weight, pressure, line widths, and placement at the speed required to achieve desired throughput while maintaining high yields. Reduction of manufacturing costs is also important because packaging is a determining factor for the production cost of MEMS (Figure 3).

Figure 3. Dispensing onto a wafer

Jet Dispensing for Wafer Capping

In addition to all the advantages inherent in jet dispensing technology, there are three distinct advantages for MEMS wafer capping:

• speed
• the ability to dispense fine lines and sharp corners
• elimination of the knit line or "dog-bone" effect

The precision, accuracy, and speed that jetting provides meshes well with MEMS requirements. Jetting is the process where fluid is rapidly ejected through a nozzle, using the fluid momentum to break free from the nozzle. A discrete, controlled volume of material is ejected with each jetting cycle. Jetting is fast because when moving from one dispense location to the next, it is not necessary to move up and down in the Z-axis. Dispense speeds ranging from 20mm-70mm/sec for MEMS applications depending on the fluid and application requirements, offer higher throughput than needle dispensing. For example, in a typical application, a jet can dispense at 80mm/sec as compared to 30mm/sec with a needle. Because jetting is a non-contact process, it is less sensitive to the dispense gap for "potato chip" wafer topology of wafers that aren't perfectly flat.

Jet valves can dispense dots to form line widths as narrow as 300 or 400µm (figure 4). Innovations like calibrated process jetting (CPJ) significantly reduce process variability. Pressure is adjusted to maintain constant mass per dot and proprietary software and hardware automatically compensate the dispensing process for both fluid viscosity changes over time and batch-to-batch variations. These factors are especially important for working with MEMS as CPJ increases dispense accuracy, delivers higher yields, and ensures consistent Takt time (pace set for industrial manufacturing lines) for improved process capability, resulting in a highly repeatable process. Set-up to set-up and line-to-line variations are minimized and the need for operator interaction eliminated. Jetting is faster and makes better lines and sharper corners than other technologies.

Figure 4: DispenseJet® valve jets dots to form 400µm lines.

A significant concern in dispensing is the knit line — the beginning and end-point of the dispensed line. There needs to be a seamless connection and the line width must not vary. For example, if too much material is dispensed at the beginning or end of the line, or the end doesn't break cleanly, the result is referred to as the "dog-bone effect," because the extra fluid at the ends of the line resemble a dog bone. This often happens with needle dispensing due to excess fluid left when the needle is retracted. In contrast, a jet dispenser shears the material, eliminating the dog-bone effect and providing a seamless connection.

Other MEMS Dispensing and Jetting Applications

In addition to wafer capping, jetting is used in many other MEMS manufacturing applications. For example,jetting silicone isolates the MEMS chip from the environment by encapsulating the wire bonds. MEMS packages on flex printed circuit board require underfill for the CMOS image processor of a camera module assembly. Jet dispensing is an ideal way to deliver that underfill. Finally, the SmartPhone alone has at least 15 parts that utilize MEMS where automated fluid dispensing is required.

Akira Morita, Business Development Manager, Nordson ASYMTEK, may be contacted at akira.morita@nordsonasymtek.com.