Temperature Measurement of Si Substrate Using Optical-Fiber-Type Low-Coherence Interferometry Employing Supercontinuum Light

Abstract

We have measured the temperature of a Si substrate using an optical low-coherence interferometer employing supercontinuum light (SC). The accuracy of temperature measurement and the minimum measurable thickness of a layer are determined by the maximum resolving power of the optical path length of the medium in low-coherence interferometry, which depends on the coherent length defined by the spectrum profile and the wavelength of the light source. Low-noise, ultraflat, and highly coherent SC, generated using ultrashort laser pulses and optical fibers, was used as a light source. The wavelength dispersion of SC on the Si substrate was compensated by using a silicon mirror as a reference mirror, resulting in shaper interference waveforms of SC at the front and back surfaces of Si substrate than those of the superluminescent diode (SLD) light used as a conventional low-coherence light source. The measurement accuracy of the temperature using SC was improved to be ±0.4 °C from ±1.0 °C for the case of using the SLD. The temperatures of the Si substrate and SiO2 thin film were simultaneously measured using SC on an 800-µm-thick Si substrate with an 8.55-µm-thick SiO2 film. The temperature of the thin film, the thickness of which is several micrometers, was measured using SC and a compensation technique of wavelength dispersion using the silicon reference mirror.

Robust characteristics of semiconductor-substrate temperature measurement by autocorrelation-type frequency-domain low-coherence interferometry

“…where ¦k is the full width at half maximum of the light source spectrum and l c is the coherence length, which is dependent on measurement accuracy and the minimum measurable thickness of a substrate. 30) From these equations, DC, cross-correlation, and autocorrelation signals peak at z = 0, at the optical path length difference between the reference and sample arm signals, and at the optical path length difference between the front and back surfaces of a substrate, respectively. The optical path length of the substrate is equivalent to be the peak interval of crosscorrelation signals.…”

“…Because the measurement accuracy for substrate temperature is related to the full width at half maximum of the correlation peak, the measurement accuracy is improved by ACT-FD-LCI with a smaller full width at half maximum than TD-LIC and FD-LCI. 30) By varying the polarization of the signal arm using a polarization controller, we have investigated the effects of the polarization differences between FD-LCI using a dualpath interferometer and ACT-FD-LCI using a commonpath interferometer. In the case of FD-LCI, the peak intensity of cross-correlation decreased by 78% and the standard deviation of optical path length worsened from 15 to 46 nm.…”