AVS 51st International Symposium
    Manufacturing Science and Technology Thursday Sessions
       Session MS-ThM

Paper MS-ThM6
Gate Sidewall Profile Control for Plasma Etch Tool

Thursday, November 18, 2004, 10:00 am, Room 303B

Session: Advanced Process Control
Presenter: J. Tanaka, Hitachi Ltd., Japan
Authors: J. Tanaka, Hitachi Ltd., Japan
A. Kagoshima, Hitachi High Technologies Corporation, Japan
D. Shiraishi, Hitachi High Technologies Corporation, Japan
H. Yamamoto, Hitachi High Technologies Corporation, Japan
S. Ikuhara, Hitachi High Technologies Corporation, Japan
M. Yoshigai, Hitachi High Technologies Corporation, Japan
Correspondent: Click to Email
The scale of semiconductor devices has been shrinking year by year. Plasma etch tools have to meet demands for the tight control of gate critical dimensions (CD) for complementary metal oxide silicon (CMOS) devices. Although the hardware of plasma etch tools has been modified to stabilize the gate CD after the etch process, there remains slight wafer-to-wafer drift of the gate CD caused by the changing wall surface conditions. Run-to-run feedback control of the etch process is beginning to be used to eliminate the residual gate CD drift. To eliminate the gate CD drift by run-to-run control, we controlled the oxygen flow rate. Then we found the gate CD could not be controlled without changing the gate sidewall profile. To monitor the gate sidewall profile we used a CD scanning electron microscope (CDSEM). The CDSEM can measure a line width from the intensity profile of secondary electron. Thus it is natural that we should expect the intensity profile measured by the CDSEM to have a strong correlation with gate profile. We found the change in the gate sidewall profile can be monitored using the intensity profile of CDSEM. The etch process used for the experiment consists of four steps: a breakthrough step, a main etch step 1 (ME1), a main etch step 2 (ME2), and an over etch step. In our first trial experiment we controlled oxygen flow rate in ME2. Although the gate CD changed as expected, the gate sidewall was tapered to degrade its vertical shape. Thus we moved the control step to ME1. When we changed the oxygen flow rate in ME1, we could control the gate CD without changing the gate sidewall profile. Finally a gate CD drift in a lot was suppressed keeping the same gate sidewall profile.