Hollows, Hillocks - Copper Wafer

Hollows, Hillocks - Copper Wafer

Material Name: Copper
Record No.: 53
Primary Chemical Element in Material: Cu
Sample Type: Wafer
Uses: Cleaning
Etchant Name: None
Etching Method: Cleaning
Etchant (Electrolyte) Composition: No data
Procedure (Condition): No data
Note: To develop a more reliable plasma cleaning process, the H2 /N2 mixture plasma is suggested for Cu dual damascene process. First, the Cu wafer was exposed to N2 plasma with the SWP source. Compared with the control sample, the Cu surface showed no obvious change as observed by the SEM technique. However, when the Cu wafer was exposed to H2 plasma at 250°C, many holes formed on the Cu surface, as shown in Fig. 5. For the control sample, native Cu oxides were formed on the surface when the Cu wafer was exposed to air.

When the Cu wafer was exposed to H2 /N2 (3.75-7.5%) plasma with the SWP source at 250°C, hillocks formed obviously on the Cu surface, as shown in Fig. 8. Seen in Fig. 9, the top view of the sample undergoing longer plasma cleaning inspected by the SEM technique reveals the formation of hillocks and cracks along the grain boundary. The mechanism of this hillock formation merits further investigation.
The Cu wafer was transferred from the left load-lock chamber (wafer chuck was set at 80°C) to the reaction chamber for 5-10 min (the wafer chuck was set at 250-350°C), and then transferred back. In the reaction chamber, when pure H2 or pure N2 or H2 /N2 mixed gases was used without plasma cleaning, no obvious hillocks were formed on the Cu surface, even when the temperature of the chuck was increased up to 350°C. Some research has reported that metal ~example, aluminum! has a higher coefficient of thermal expansion than silicon or silicon dioxide. When metal films on the silicon substrate or silicon oxide are heated (for example to 450°C sintering) during device manufacturing process, they undergo compressive stress, which can be relieved by hillock formation. On the contrary, when metal films on silicon substrate or silicon oxide are cooled during the device manufacturing process, they undergo tensile stress, which results in crack formation. However, no obvious change on the Cu surface was observed by the SEM technique, compared with the control sample. The experimental results mentioned above revealed that thermal stress from the chuck temperature is not the dominant mechanism for Cu hillock formation. In this case, the range of chuck temperature variation from 80 to 350°C was the largest. It is believed that the range of temperature variation was still too small to form Cu hillocks or cracks. According to the data mentioned above, hillocks occur on the Cu surface mainly after plasma cleaning. Therefore, it is believed that plasma cleaning plays a dominant role in hillock formation. Previous studies have reported that aluminum hillock growth is enhanced during the early stage of insulator deposition. The enhanced formation of hillocks is attributed to the increase in substrate temperature with the application of radio frequency (rf) power.
Reference: Tsung-Kuei Kang, and Wei-Yang Chou, Avoiding Cu Hillocks during the Plasma Process, Journal of The Electrochemical Society, 151 (6) G391-G395, 2004.

Figure 5: SEM micrograph of Cu sample plasma-induced by pure H2 plasma (SWP source).

Figure 8: SEM micrograph of Cu sample plasma-induced by H2 /N2 plasma (SWP source).

Figure 9: Top-view SEM micrograph of Cu sample plasma-induced by H2 /N2 plasma for a long time (SWP source).