The copper wiring technique is a vital manufacturing technology for the current ultra-large-scale integrated (ULSI) circuits. In general, these Cu wiring are fabricated by the dual damascene process. The damascene process is mainly composed of chemical mechanical polishing (CMP) and electrical Cu plating techniques. Although these processes are well-established, these processes are fundamentally demonstrated in a wet environment and require expensive consumables and toxic chemicals. Furthermore, since CMP requires the adequate processing load for the efficient machining, the usage of a low-k material as an interlayer insulator, which has low mechanical strength, requires a prolonged processing time. From above background, it is longed-for that the dry etching process for Cu can replace the conventional wet machining process. In the previous study on the conventional dry etching for Cu, chlorine is often used as the typical etchant gas. The chlorine gas is harmful for humans and aggressive toward many of the metals constructing the etching apparatus. Furthermore, because the chlorinated Cu products exhibit a low vapor pressure around room temperature, substrate heating is generally required.

On the other hand, it has been revealed that a hydrogen-based plasma can be applied for Cu dry etching at low temperature. This hydrogen dry etching seems attractive in terms of preserving the environment and suppressing the use of toxic chemicals. If a higher etching rate greater than 100 nm/min can be obtained by pure hydrogen plasma etching, it is expected that a hydrogen dry etching method with no toxic chemicals could be substituted for the CMP process. In this study, therefore, we applied the high-pressure hydrogen glow plasma (>100 Torr) to Cu etching in an attempt to achieve a higher Cu etching rate.

In the experiment, a localized hydrogen-based plasma is generated at 100 Torr around the tip of a syringe needle electrode (outer diameter of 1mm). The very-high-frequency (VHF; 150 MHz) power source was used for the plasma generation. The samples used in this study were a 1 mm-thick Cu plate for investigating the Cu etching property, a 0.7 mm-thick Si(001) wafer and 0.7 mm-thick SiO₂ glass to compare the etching behavior of Cu with other ULSI construction materials. The sample stage temperature was controlled from −20 to 330°C. The H₂ gas was supplied to the plasma directly from the needle bore via a mass flow controller during the etching. In this study, to confirm the role of hydrogen in Cu etching, He, Ar and N₂ gases were used as the process gas instead of H₂.

As a result, the Cu etching occurs only when the process atmosphere contains H₂ gas, and the etching rates decrease with diluting the hydrogen concentration by He gas. The impact of plasma heating on the Cu etching is negligible, so the etching mechanism of the high pressure H₂ plasma differs from that of the electrospark machining. When the sample stage temperature is varied from −20 to 330°C, the Cu etching rate has no obvious dependence on the stage temperature. On the other hand, the Cu etching rate increases linearly with increasing the input power from 30 W to 100 W. The emission intensity of Balmer α line of H atom observed in optical emission spectrum of the plasma also increases linearly with the input power. The Cu etching rate of 500 nm/min can be achieved around the stage temperature of 0°C. Meanwhile, the etching rates for Si and SiO₂ around the stage temperature of 0°C are 100 μm/min and 50 nm/min, respectively. The selectivity of the etching rates for Cu and SiO₂ suggests that this etching technique could be applied to Cu wiring on a trenched SiO₂ layer. The etched Cu surface morphology is rough and exhibits many round pits and bumps. This surface roughening might be due to excessive hydrogen incorporation in the Cu bulk.