New method of high-precision thermometry

Abstract

Measurements of temperature using infrared radiation have become common in industry. They are used in process control for materials ranging from steel to glass to silicon, in machinery to monitor component temperature and performance, and to predict component failure. We developed a rapid, noncontact method of accurately measuring the temperature and the spectral emissivity of a surface. Inclusion of the emissivity in the measurement makes it possible to achieve high-precision results. In our method, the power spectrum emitted by a surface is measured first by a multichannel spectrometer; the surface is then illuminated by a full-spectrum light source with a known spectrum, and the sum of the emitted and reflected power from the surface is measured. The data from the two measurements and the governing relations for the radiation and reflection processes are then combined to generate a set of curves that intersects near a solution point, which yields the temperature and the emissivity in the range of wavelengths of one of the channels. The emissivity in the other channels is readily calculated and yields the wavelength-dependent emissivity of the surface. The accuracy of the method for very rapid (submicrosecond) measurements is approximately 1% or less. We named this method the spectrum method because it uses the spectrum of both the emitted and reflected light to make the calculation. The method is more accurate than a grey body calculation because the wavelength-dependent emissivity is an integral part of the calculation of the surface temperature. The accuracy of the temperature and emissivity measurements can be calculated for a system containing five or more channels. We discuss the sources of errors and quantitatively assess their effect. Relative as well as absolute values of the channel emissivities have a significant effect on errors in the emissivity and temperature measurements. This temperature diagnostic has been tested on a tabletop device that allows us to make quick and reproducible measurements.

Characterizing the emissivity of materials under dynamic compression (final report for LDRD project 79877).

“…Attaching an integrating sphere to a sample provides a measure of total hemispherical reflectivity during compression [6], but the method is potentially difficult and expensive proposition. Other reflectance approaches [7] work well in some circumstances, but require assumptions about the surface under which may not be valid in general.…”

Nevada National Security Site-Directed Research and Development FY 2011 Annual Report

“…Machined surfaces would present a problem because they can behave like diffraction gratings. Poulsen (2006) derives an equation for the emissivity in band n as a function of measured reflection and radiance, incident irradiance, and ideal blackbody emission in bands n and i. The blackbody emission F is calculated over a range of temperatures.…”

“…The proposed system was based on the system described by Poulsen (2006), who presents a method based on normalizing the radiance and reflectance measurements at several wavelengths to those at one wavelength and solving a set of simultaneous equations to obtain the temperature and emissivity at the reference wavelength. Normalization eliminates the dependence on absolute geometric factors in the measurement; this condition is particularly attractive for measurements on shocked surfaces because it allows measurement of emissivity without the need to collect light over 2π, as with an integrating sphere, which is experimentally difficult.…”

“…The multiple estimates of emissivity in band n must agree for measurements in different wavebands, i, so the correct temperature and emissivity are found at the intersection of the curves by where ε n = Emissivity in band n, C i = Calibration coefficient for band i, A potential problem with this method, which hurt us in this work, is that when the emissivity depends only slightly on wavelength, the uncertainty in ε n becomes large. Poulsen (2006) used a ~1 µs surface discharge light source. We proposed to make a similar measurement with a 10 ps, 60 MHz repetition-rate, supercontinuum laser source that emits light from 450 to 1800 nm.…”

“…One is the selection of nickel as the test piece; it is not very far from a graybody, and, as noted by Poulsen (2006), this method is "problematic when the channel emissivities are nearly equal or are very small." The oxidized nickel surface reflectivity values at 635 nm and 685 nm are very similar.…”