|M.Sc Student||Pasternak Limor|
|Subject||Low-Temperature Direct Bonding of Silicon Nitride to|
|Department||Department of Chemical Engineering||Supervisor||Professor Yaron Paz|
|Full Thesis text|
Recent years have shown an increasing interest in the study of direct bonding of surfaces, mostly due to its prospects in the areas of integrated circuits and micro-electro mechanical systems. These applications include SOI (silicon on insulator) wafers, SOG (silicon on glass), SOS (silicon on sapphire) and encapsulation of MEMS devices (3D-integration). In a direct bonding process, two mirror-polished flat, smooth and surface-activated substrates adhere to each other at room temperature without adhesives and upon applying minimal external forces.
Direct bonding may provide cheap and reliable alternatives for the use of adhesives in the photovoltaic (PV) cells industry. Silicon-based solar cells contain several layers and the adhesion between them is an important factor, which affects the fabrication and the performance of the cells. Silicon nitride is extensively used as an anti-reflective coating in the PV industry and glass is used to externally cover the devices.
A series of bonding experiments between glass and SiN were performed with samples prepared under varied preparation parameters. The bonded pairs were evaluated optically and mechanically to study the effect of the preparation parameters on the bond quality. The bonded pairs that were activated by nitrogen for 80 sec and were thermally annealed post-contact at 400 °C for 2 hr resulted in the highest bonding energy, showed a cohesive failure during shear tests and exhibited the lowest void quantity and void area. This kind of plasma assisted direct bonding between glass and SiN is reported here for the first time.
HRTEM imaging performed on thin films prepared by Focused Ion Beam (FIB) revealed a clear defect-free interface of about 4 nm between silicon nitride and glass, which indicated a non-penetrating surface interaction. EELS and HRTEM-EDS analyses revealed that the interface contained silicon, oxygen and nitrogen forming silicon oxy-nitride during the bonding process.
The XPS and ATR FT-IR spectra of the glass and the silicon nitride surfaces showed a clear response to activation under both oxygen and nitrogen plasma. The ATR FT-IR measurements revealed the formation of silanol groups on the activated glass surface. This result suits the increase in the O:Si atomic ratios observed by XPS measurements. Similarly, the plasma activated silicon nitride showed an increase in the O:Si atomic ratios. Here, no silanol groups were formed on the surface as deduced by ATR-FTIR measurements. This fact, along with a characteristic shift in the Si2p peak, implies that an oxide layer is formed on top of the silicon nitride surface upon activation.
A physical modification was observed following the plasma activation. The bearing ratio of the glass surface was significantly increased, reflecting a surface modification which enables to achieve a better bond.
Based on the results from the different techniques, a mechanism for bonding silicon nitride and glass was suggested. The mechanism is based on generation of silanol groups on the surface of the glass and on oxidation of the silicon nitride surface, leading to siloxane covalent bonding.