Effects of noise level in fitting in situ optical reflectance spectroscopy

摘要

Curve fitting of simulated optical reflectance spectroscopy data is used to evaluate the accuracy of parameters derived from the fits of actual data. These simulations show that to determine the index of refraction n to an accuracy of 0.0015 (corresponding 0.044% for Al_(0.5)Ga_(0.5)As at growth temperature), a reflectance noise with standard deviation σ≤ 0.00005 is required if the absolute reflectance calibration is unknown. The simulations also show that when the absolute reflectance is known within +- 0.05%, a noise of up to σ = 0.0008 would result in the same desired accuracy for n. The factors contributing to the uncertainty of the reflectance scaling factor include the temperatures of the photodetectors, stability of light intensity, and deposits on the window of the growth chamber. These factors are investigated experimentally and possible solutions that will allow calibration within the goal range are discussed.