Search

A physically-based dynamic simulator has been constructed to investigate the time-dependent behavior of equipment process, sensor, and control system for rapid thermal chemical vapor deposition (RTCVD) of polysilicon from SiH4. The simulator captures the essential physics and chemistry of mass transport, heat transfer, and chemical kinetics of the RTCVD process as embodied in equipment. In order to complete the system-level description, reduced-order models are also employed to represent processes involving high complexity of physics. Integration of individual simulator elements for equipment, process, sensors, and control systems enables the evaluation of not only the deposition rate and film thickness, but also of a broad range of dynamic system properties such as equipment performance, gas flow conditions, wafer temperature variation, wafer optical properties (absorptivity/emissivity), gas composition in reactor, total process cycle time, consumables volume, and reactant utilization. This makes the simulator directly applicable to the optimization of process recipe and equipment design, to process control strategy, and to fault classification. This case study of polysilicon RTCVD demonstrates (1) that integrated dynamic simulation is a versatile tool for representing system-level dynamics, and (2) that such representation is pivotal in successful application of modeling and simulation for manufacturing optimization and control.

In the semiconductor manufacturing industry, optimization of advanced equipment and process designs must include both manufacturing metrics (such as cycle time, consumables cost, and product quality) and environmental consequences (such as reactant utilization and by-product emission). We have investigated the optimization of rapid thermal chemical vapor deposition (RTCVD) of polysilicon from SiH4 as a function of process parameters using a physically-based dynamic simulation approach. The simulator captures essential time-dependent behaviors of gas flow, heat transfer, reaction chemistry, and sensor and control systems, and is validated by our experimental data. Significant improvements in SiH4 utilization (up to 7 x) and process cycle time (up to 3 x) can be achieved by changes in (i) timing for initiating wafer heating relative to starting process gas flow; (ii) process temperature (650 - 750oC ) ; and (iii) gas flow rate (100 - 1000 sccm). Enhanced gas utilization efficiency and reduced process cycle time provide benefits for both environmental considerations and manufacturing productivity (throughput). Dynamic simulation proves to be a versatile and powerful technique for identifying optimal process parameters and for assessing tradeoffs between various manufacturing and environmental metrics.

Gas delivery from a liquid source, common in semiconductor manufacturing, raises contamination control concerns not only due to impurity levels in the source. In addition, the lower vapor pressure of impurity species compared to that of the host (source) species causes impurity concentrations in delivered gas to increase as the source is used up. A physics-based dynamic simulator to describe the time-dependent variation of impurity level in such a gas delivery system has been developed and applied to important case of CHCIF2 impurities in host CHF3 liquid, as routinely used for dry etching processes. For a cylinder of CHF3 liquid with 100 ppm of CHCIF2 at 21.1o C (70o F), the concentration of CHCIF2 in the delivered gas is initially ~ 21 ppm, and rises slowly to ~ 100 ppm with ~ 25% of the initial material remaining. With further usage, the CHCIF2 level increases quickly to ~ 350 ppm when ~ 15% of the initial source material is left; at this point, the source has reached the liquid-dry point, i.e., all the remaining source material is gaseous, and the impurity concentration in delivered gas remains constant at 350 ppm until all material is gone. The time- dependence of CHCIF2 impurity concentration is also dependent on the operating temperature of the liquid source: for higher temperatures, the fast rise in impurity concentration and the liquid-dry point occur earlier, while the final impurity level after this point is lower. The dynamic simulator represents a useful tool for avoiding contamination problems with liquid delivery systems and for optimizing materials usage (for cost and environmental benefits) by structuring source usage procedures consistent with contamination-sensitivity of the process. The results also suggest benefits in materials usage if specific source temperatures (different from room temperature) were imposed. The physical basis of the dynamic simulator allows more general application to other systems.