The model for rf bias effects that we have developed should make it easier for industry to select optimal operating frequencies and may stimulate the adoption of new methods for RF biasing, such as multiple-frequency bias and non-sinusoidal bias. NIST studies of electrical endpoint detection should enable semiconductor manufacturers to obtain better control of plasma etching processes and other processes that involve rapid changes in wafer surface conditions.
Provide advanced measurement techniques, data, and models needed to characterize plasma etching and deposition processes important to the semiconductor industry, enabling continued progress in model-based reactor design, process development, and process control.
Recent experimental efforts have focused on a new method for measuring electron number density in plasmas, the wave cut-off method. In collaboration with KRISS, the Korea Research Institute of Standards and Science, a wave cut-off probe has been designed and implemented in NIST laboratories. Unlike other techniques, cut-off probes do not suffer from problems with deposition of insulator layers and therefore they are more compatible with real plasma processing conditions.
The cut-off probe has been used to characterize the effect of radio-frequency (RF) bias on the plasma electron density in an inductively coupled reactor. In such reactors, RF bias is used to control ion energies, but it also has unintended effects on electron density and ion flux. Interest in these unintended effects has been stimulated by new methods for RF biasing, such as multiple-frequency bias and non-sinusoidal bias, which make use of frequencies higher than those previously used. We measured the effect of RF bias on electron density over the entire frequency range used in industrial plasma processes. These measurements validated a model that is general and fundamental enough to apply not only in our plasma reactor, but in all inductively coupled reactors. The model makes it easier for workers in industry to select optimal operating frequencies.
We have also performed a fundamental investigation of electrical detection of plasma etching endpoints. In plasma etching, one requires an endpoint signal that indicates the layer being etched is fully consumed, so that the etch process can be stopped before underlying layers are damaged. Electrical signals, such as the voltage or current being drawn by the plasma, are often used to detect endpoint. Unfortunately, the origin of the electrical changes that occur at endpoint are not well understood. Consequently, the electrical signals that are most commonly used, which are chosen for reasons of convenience or based on purely empirical work, are not necessarily the most reliable indicators of endpoint. We performed electrical measurements and simultaneous wave cut-off measurements of electron density during Ar/CF4 plasma etches of silicon dioxide films on silicon substrates. By comparing the electrical data to wave cut-off results and to models of plasma electrical behavior, we were able, for the first time ever, to fully determine the fundamental origin of electrical changes observed at and near endpoint. The work provides recommendations for the plasma etching industry that identify, based on fundamental reasons, which electrical signals are the most reliable for endpoint detection.