SPIE Startup Challenge 2015 Founding Partner - JENOPTIK Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:

SPIE Photonics West 2017 | Register Today

SPIE Defense + Commercial Sensing 2017 | Call for Papers

Get Down (loaded) - SPIE Journals OPEN ACCESS


Print PageEmail PageView PDF

Micro/Nano Lithography

Using contact imprinting for adhesive wafer-to-wafer bonding

Bonding wafers via indirect contact imprinting of an adhesive layer has possible application in microfluidics and wafer-level packaging of microdevices.
5 April 2007, SPIE Newsroom. DOI: 10.1117/2.1200702.0657

Fabrication of microfluidic devices often requires assembly of two patterned wafers with microchannels, inlet and outlet holes, and even electrodes defined on their surfaces. Because applications often call for optical detection, one of the wafers is usually made of glass. Wafer-to-wafer bonding in such cases is possible, but it must be performed at temperatures below 500°C, the annealing temperature of the glass.

There are few established processes for low-temperature wafer-to-wafer bonding. Anodic bonding1 is a well-known process in the MEMS (microelectromechanical systems) industry. The bonding temperature is in the range 300–450°C, but the process requires a good-quality surface. Eutectic bonding,2 another low-temperature process, has the disadvantage of reduced yield and high sensitivity to contamination. Glass frit bonding,3 which is gaining popularity in industry due to its low cost, is an alternative to anodic bonding. However, it is considered to be a ‘dirty’ process and is used more for inactive surfaces.

Adhesive bonding enables joining of silicon or glass wafers at even lower temperatures (usually below 200°C). The technique is also less dependent on substrate material, particles, surface roughness, and planarity of the bonding surfaces.4 The opportunity to combine standard microfabrication deposition and patterning approaches with low demands on the wafer surface make this a very attractive and cost-effective solution.

Figure 1. (a) Spinning of SU8-5 photoresist onto a dummy wafer. (b) Contact-imprinting SU8-5 photoresist on a Teflon cylinder. (c) Contact-imprinting SU8-5 from the Teflon cylinder onto the desired wafer surface. (d) Alignment and contact with the second desired wafer. (e) Wafer-to-wafer bonding.

Several lithographically patternable materials, such as benzocyclobutene (BCB) and positive and negative photoresists, have already been studied as intermediate layers for adhesive wafer bonding. Besides BCB, the epoxy-based negative photoresist SU-8 shows the best results in bonding experiments.5,6 The advantages of SU-8 are its flexibility in controlling layer thickness (up to several hundreds of microns), its high chemical and thermal stability, and its good mechanical properties. For this reason, a fairly nonstandard adhesive bonding method at the wafer level has been proposed, using SU-8 photoresist as an intermediate layer. Unlike direct spin coating, in this case an indirect coating method is employed. It is achieved by selectively depositing the polymer layer on one of the bonding surfaces by contact imprinting. The proposed technique is especially suitable in applications where the classical methods for coating (spinning) cannot be successfully applied, e.g., on wafers with nonplanar surfaces or with freestanding structures.

The main steps of the contact-imprint bonding process are presented in Figure 1. For demonstration purposes, a cover wafer is bonded over a wafer with membranes. First, a thin layer of SU8-5 photoresist is spun onto a dummy silicon wafer pretreated with HNO3 in two successive steps— 500rpm/15s and 3000rpm/60s—to obtain a uniformly thick 12µm hydrophobic layer: see Figure 1(a). Next, this adhesive layer is transferred onto the surface of a Teflon cylinder (diameter of 38mm, length of 120mm), as in Figure 2. The cylinder is rolled over the desired bonding surface, partially transferring the adhesive layer (SU8-5) onto it, as shown in Figure 1(c). Both structured wafers are aligned and brought into contact—see Figure 1(d)—using an EV Group (EVG) mask aligner. The final step, wafer-to-wafer bonding, is performed on an EVG wafer bonder between 100 and 200°C for 30min and an applied force of 1000N in vacuum at 10-3mbar, as shown in Figure 1(e).

Figure 2. Imprinting of the SU8 layer from a (6in.) dummy wafer to the Teflon cylinder.

Figure 3. Optical image of a cross-section of bonded wafers.

A critical parameter in the wafer-to-wafer bonding process is the residual stress induced by the bonding. We ascertained the residual stress in the SU8-5 layer using a standard stress measurement system (KLA Tencor). A 12µm-thick SU8-5 layer was deposited by spinning. After curing at 65°C for 2min and at 90°C for 10min, the layer was exposed to UV at a dose of 100mJ/cm2. The results indicated tensile stress values in the range 20–40MPa. These values, as well as the good elastic properties of SU8-5 (Young's modulus in the range 2–4GPa) and its high chemical and thermal stability (glass-transition temperature >200°C), testify to the suitability of SU8-5 photoresist as a material for wafer-to-wafer adhesive bonding. For bonding strength tests the full bonded wafers were diced (using DISCO 3350 equipment) into 6×6mm pieces. We tested the dies on an Instron 5848 microtester with a special design tool for measuring shear strength. The results indicated values in the range 18–25MPa.

To prove the feasibility of the method, we used it to bond planar and nonplanar wafers (with 20µm-deep etched areas). One of the bonded wafers was glass, which allows the unbonded areas to be easily observed without any special equipment. For the nonplanar wafers, a 3×3mm2 pattern was etched at a depth of ~20µm on a 4×4mm2 chip using deep reactive ion etching (RIE, Adixen AMS 100) through a 2µm-thick aluminum mask. The results showed that a fully bonded area is achieved even with two planar wafers. Figure 3 shows the cross-sectional image of such bonded wafers taken after the wafers were diced. The dicing process itself amounts to a bonding test: the bonding must be strong enough to overcome the force generated during dicing. Figure 3 shows that the thickness of the SU-8 layer after bonding is around 2µm. Figure 4 presents a top-view optical image of fully and partially bonded areas. Note that the yield of this bonding process is high, in the range 95–100%, similar to the performance typically obtained with classical anodic bonding.

Our method has been successfully used to fabricate microfluidic devices for electrical characterization of cells, as shown in Figure 5. In this device, to limit the flow of excess SU-8 in the microfluidic channels, levees were constructed around the microfluidic channels to reduce the bonded area.

Figure 4. (a) Fully and (b) partially bonded areas.

Figure 5. Microchannel created using SU-8 photoresist adhesive bonding.

A new indirect-bonding technique has been applied using SU-8 negative photoresist as the adhesive layer and contact imprinting as the method for its transfer onto the desired surface(s). The main advantages of the method are low cost, high yield, low bonding temperature, and low stress induced by the bonding process. The method is especially suitable when surface topography or device functionality rule out traditional spinning.

This project is funded by the Institute of Bioengineering and Nanotechnology, Agency for Science, Technology, and Research (IBN/04-R44007-OOE).

Liming Yu
Department of Mechanical Engineering
National University of Singapore

Liming Yu is currently a research fellow in the Department of Mechanical Engineering, National University of Singapore. His research interests include MEMS technologies, bio-MEMS, and dielectrophoresis.

Frances Eng Hock Tay
Department of Mechanical Engineering
National University of Singapore
Medical Devices Group
Institute of Bioengineering and Nanotechnology

Francis E. H. Tay is an associate professor in the Mechanical Engineering Department at the National University of Singapore. He is an active researcher in microsystems, and some of his current research interests include biochips, wearable systems, and microfluidics. He is concurrently a group leader in the Medical Devices Group in the Institute of Bioengineering and Nanotechnology. He has also been a technical manager in the Micro and Nano Systems Cluster at the Institute of Materials Research.

Daniel P. Poenar
School of Electrical and Electronic Engineering
Nanyang Technological University

Daniel P. Poenar received a PhD degree from the Technical University of Delft in 1996. He is an assistant professor in the Electrical and Electronic Engineering School, Nanyang Technological University. His research interests are MEMS and micromachining, and their application for the development of sensors and actuators, especially for (bio)chemical devices. He is coauthor of more than 40 papers published in journals and conferences proceedings.

Ciprian Iliescu
Medical Devices
Institute of Bioengineering and Nanotechnology

Ciprian Iliescu is a senior research scientist at the Institute of Bioengineering and Nanotechnology. His current research projects are related to dielectrophoresis, electrical characterization of cells by impedance spectroscopy, transdermal drug delivery using microneedle arrays, and microfabricated dialysis systems. He is the author and coauthor of more then 120 papers published in journals and conferences proceedings.