A case study of ion implant in-line statistical process control

Ion implantation is one of the most critical processes in the front-end-of-line for ULSI manufacturing. With more complexity in device layout, the fab cycle time can only be expected to be longer. To ensure yield and consistent device performance it is very beneficial to have a Statistical Process Control (SPC) practice that can detect tool issues to prevent excursions. Also, implanters may abort a process due to run-time issues. It requires human intervention to dispose of the lot. Since device wafers have a fixed flow plan and can only do anneal at certain points in the manufacturing process, it is not practical to use four-point probe to check such implants. Pattern recognition option on some of the metrology tools, such as ThermaWave (TWave), allows user to check an open area on device wafers for implant information. The two cited reasons prompted this work to look into the sensitivity of TWave with different implant processes and the possibility of setting up an SPC practice in a high-volume manufacturing fab. In this work, the authors compare the test wafer result with that of device wafers with variations in implant conditions such as dose, implant angle, energy, etc. The intention of this work is to correlate analytical measurement such as sheet resistance (Rs) and Secondary Ion Mass Spectrometry (SIMS) with device data such as electrical testing and sort yield. For a 1.5% TWave control limit with the tested implant processes in this work, this translates to about 0.5 degrees in implant angle control or 2% to 8% dose change, respectively. It is understood that the dose sensitivity is not good since the tested processes are deep layer implants. Based on the statistics calculation, we assess the experimental error bar is within 1% of the measured values.