| AVS 56th International Symposium & Exhibition | |
| Plasma Science and Technology | Tuesday Sessions |
| Session PS1-TuM |
| Session: | Advanced FEOL and BEOL Etch |
| Presenter: | L. Azarnouche, STMicroelectronics, France |
| Authors: | L. Azarnouche, STMicroelectronics, France E. Pargon, CNRS/LTM, France M. Martin, CNRS/LTM, France O. Luere, CNRS/LTM, France K. Menguelti, CNRS/LTM, France P. Gouraud, STMicroelectronics, France C. Verove, STMicroelectronics, France O. Joubert, CNRS/LTM, France |
| Correspondent: | Click to Email |
With the continuous scaling down of semiconductor device dimensions, the linewidth roughness (LWR) becomes a non negligible parameter that needs to be controlled in the nanometer range for the future technological nodes (1.7nm (3σ) for the 32 nm technological node). In previous studies, we demonstrated that the 193 nm photoresist mask LWR is the main contributor to the final gate LWR. We also observe that the LWR is mainly decreased during the plasma etching steps in which the resist mask is involved (BARC and hard mask etching). Preliminary conclusion is that the photoresist mask is the first vector of LWR decrease during plasma exposure. The resist LWR is therefore the key parameter to sucessfully control the final metal gate LWR in the nanometer range. In the present study, we first evaluate the impact of HBr cure plasma treatment and resist trimming processes on the resist LWR and second analyze how these plasma etching steps impact the final gate LWR. LWR measurements are performed using the CD-AFM technique much more powerful than commonly used CD-SEM. First results indicate that both resist trimming and plasma cure treatment processes improve the resist LWR and consequently the final gate LWR. We demonstrate that, during cure processes, plasma Vacuum UltraViolet (VUV) light is mainly responsible for the resist LWR decrease while during trim processes, the plasma VUV light combined to the lateral erosion of the resist induced by reactive radicals is identified as the main contributor to the resist LWR decrease.
However, we also show that the sequence of cure and trim processes has a less important impact on the gate LWR than a cure or a trim process only and that the position of the cure treatment in the sequence of plasma etching steps involved in the gate patterning process has some consequences in the final gate LWR. Finally, as the etching mask used to pattern the gate plays a primordial role in the final gate LWR, we compare the impact of two masking strategies: one using amorphous carbon layer as etching mask and the other one a spin on carbon hard mask.