| Summary: | 博士 === 國立清華大學 === 工程與系統科學系 === 91 === The advanced semiconductor fabrication requires a much tighter process monitoring and control to improve production yield and reliability. Among the several hundreds of processing steps of modern ultralarge scale integrated circuits (ULSI) fabrication, plasma based processes play a crucial role in achieving the desired device performance. Plasma processes are primarily based on chemical and physical reactions of reactive species and charged particles with wafer surface. They also contribute to a great part of the problem associated with the fabrication yield because of their complex characteristics. Therefore, it is important to develop real-time control of plasma properties, such as ion density, ion energy, reactive species, etc., for plasma processing tools.
In this thesis, several non-invasive diagnostic systems of ion density and ion energy for real-time control system have been either developed or improved from existing tools. RF diagnostics, trace rare gases-optical emission spectroscopy (TRG-OES), and 36 GHz heterodyne interferometer were employed to measure positive ion density. On the other hand, the averaged ion energy was estimated by measuring peak RF voltage on the biased wafer electrode using a RF voltage probe.
In the second phase of this study, neural-network models for machine parameters—critical plasma parameters—etch rate were developed for polysilicon etch in an inductively-coupled chlorine plasma. Experimental results show that ion current density is proportional to the source power. While the RF bias voltage varies linearly with bias power. Increasing pressure will cause ion current density to decrease. Both RF bias voltage and chlorine atomic density increases with gas pressure, but the opposite was found for the ion current density. It was also demonstrated that plasma properties and etch rate depend weakly on the Cl2 flow rate. Consequently, the reproducibility of etch rate can be improved if the steady-state errors of both ion current density and rf bias voltage are reduced and a pressure controller is need to eliminate pressure variation.
For the proof of principle experiment, a real-time closed-loop control system of both ion density and ion energy in an inductively-coupled chlorine plasma etcher has been developed. The TRG-OES method was used to measure the chlorine positive ion density. An RF voltage probe is adopted to measure the rot-mean-square (RMS) RF voltage on the electrostatic chuck which varies linearly with the sheath voltage. One actuator is a 13.56 MHz RF generator to drive the inductively coil seated on a ceramic window. The second actuator is also a 13.56 MHz RF generator powering the electrostatic chuck. The two digital PI controllers were used to separately control the positive ion density and the RMS RF voltage. The plasma coupling effect and model uncertainty have been considered using the quantitative feedback theory (QFT) design technology such that the robust stability and performance can be achieved. The experimental results showed that the closed-loop control had a better repeatability of plasma parameters as compared with the open-loop control. The closed-loop control can eliminate the process the process disturbance resulting from reflected power. In addition, a better reproducibility in etch rate also obtained by the closed-loop control. Under a closed-loop control etch, the standard variation of etch rate was reduced by a factor of 2.5 —7.
|
|---|
