| Summary: | 碩士 === 逢甲大學 === 材料科學所 === 91 === This work employed conventional (diode) and inductively coupled plasma (ICP) magnetron sputtering to deposit thin films of tungsten having thickness 40 (or 20) nm. The effect of varying the deposition parameters (background pressure, substrate bias, and sputtering pressure) on the phase distribution, electrical resistivity, and microstrucure of the deposited films was investigated by using glazing-angle X-ray diffractometry (XRD), scanning electron microscopy (SEM), resistivity plots, and transmission electron microscopy (TEM). Meanwhile, the effectiveness of passivating the tungsten films in terms of acting as diffusion barriers for copper was evaluated.
There are two types of phases for the sputter deposited tungsten. One (a-W) is low-resistivity (~25 mW-cm) equilibrium phase of body centered cubic structure. Another (b-W) is a high-resistivity (~170 mW-cm) metastable phase exhibiting A15 (cubic) structure. Varying background pressure of the chamber by one order from 5 ´ 10-6 to 5 ´ 10-5 mb impacts markedly the phase of the films: reducing background pressure tends to favor the formation of a-W. However, tungsten films deposited by applying any values of substrate bias (up to -125 VDC) at the deposition window employed are dominated by b-W. Raising the sputtering pressure tends to accelerate the formation of b-W, ultimately forming films comprising only b-phase at 2 ´ 10-2 mb.
However, being able to control independently the flux density and bombarding energy, the ICP sputtering is capable of depositing tungsten films containing single a-phase over a broad range of sputtering pressure and substrate bias. Nonetheless, ICP-deposited tungsten films are also converted into b-phase upon increasing the sputtering pressure to 2 ´ 10-2 mb. Additionally, bending-beam measurement reveals that raising sputtering pressure indeed has an impact on the intrinsic stress of the films, and the trend of stress change is explained in terms of transportation/deposition of the emitted particles and the zone structure model proposed before.
Finally, N2/H2 plasma treatment was performed to passivate the a- and b-W thin barrier films [denoted as W(N)]. Then, the effect of the passivation parameters (temperature and duration) on increasing the barrier capability against copper diffusion was evaluated by comparing the threshold temperatures for the Si/barrier (40 or 20 nm)/Cu stacked samples to lose electrical integrity. b-W(N) outperforms b-W and a-W(N) outperforms a-W, whereas b-W is better than a-W. A preliminary result based on TEM analysis indicates that the difference in barrier property between b-W and a-W is attributed to differences in barriers’ microstructure and crystallinity.
|
|---|
