This invention relates to determination of the true temperature of a body and true radiative emissivity of the body at that temperature.

Noncontact measurement of the temperature (assumed to be substantially uniform) of a thermally radiating body at elevated temperature is difficult, for a number of reasons. First, the body is usually a "greybody", and thus has an associated emissivity parameter β (also known) that is less than 1.0. A thermal "blackbody" would have an emissivity parameter β that is = 1.0 for all wavelengths. Second, the emissivity varies with the wavelength at which the thermal radiation is measured. Third, the perceived temperature and emissivity will be affected by any intervening medium, such as air or other low pressure gas, that the body is positioned in. This third difficulty may be substantially avoided by conducting the measurements in a high vacuum environment.

Recently, some workers have published discussions of remote sensing of temperature, not necessarily uniform, of a radiating body and possible discrimination between radiation emitted by the body and radiation emitted by other objects adjacent to the body. For example, Cogan, in "Remote Sensing of Surface and Near Surface Temperature from Remotely Piloted Aircraft," Applied Optics 24 (1985) 1030-­1036, discusses a technique for obtaining the temperature of specific water and land surfaces that are adjacent to other radiating bodies. Cogan utilizes a normalized radiance, which is the product of target radiance and the response curve function for the measuring device, both wavelength dependent, integrated over a predetermined wavelength range and divided by the integral of the response curve function over the same wavelength range. Cogan also utilizes a mean wavelength, weighted by the response curve function, in the well known Planck formula, Equation (2) below, to determine a measured mean atmospheric temperature. However, Cogan does not take separate account of the wavelength-dependent emissivity or decompose the radiation into a sequence of wavelength ranges for purposes of separate determination of the true temperature of the radiating body and the true body emissivity in a given wavelength range.

Rockstroh and Mazumder, in "Infrared Thermographic Temperature Measurement During Laser Heat Treatment," Applied Optics, 24 (1985) 1343-1345, simultaneously use a thermocouple to measure ambient temperature and a thermographic system to measure the greybody temperature of an adjacent body; and they attempt to predict local surface temperature of the greybody, assuming a constant emissivity of 0.8. Wavelength dependence of emissivity is not addressed, and no separate account is taken of the wavelength dependence of the radiance itself.

The apparent temperatures of actively burning regions and burned over regions in a forest fire are assessed by Stearns et al. in "Airborne Infrared Observations and Analyses of a Large Forest Fire," Applied Optics (1986) 2554-2562. Thermal scanners are used with small fields of view to determine apparent temperature of different regions in and around an active forest fire. The spectral-radiance of a forest fire is obtained as a function of wave number (cm⁻¹) at a quoted spectral-resolution of 0.95 cm⁻¹. The method used, if any, for accounting for the wavelength dependence of emissivity is not disclosed in the Stearns et al. paper, although a method of indirectly determining apparent temperature of local areas adjacent to a forest fire is apparently employed.

In "Quantative Measurements of Ambient Radiation, Emissivity, and Truth Temperature of a Greybody: Methods and Experimental Results," Applied Optics (1986) 3683-3689, Zhang, Zhang and Klemas discuss the contribution to radiance from self-radiation of a target and from ambient radiation reflected by the target. This paper notes the central importance of the emissivity and the truth temperature of a greybody, and it formally utilizes the Planck formula given in Equation (2) below, to formally manipulate certain integral expressions for body radiance in the presence of reflected radiation that arises from adjacent objects. The paper discusses the difficulty of distinguishing between temperatures of two bodies with similar but not identical emissivities and presents a method for obtaining the approximate emissivity of a greybody, where the body temperature can be both measured accurately and increased or decreased by a controllable amount. However, the possible wavelength dependence of emissivity is not separately taken account of in this paper.

The present invention is concerned with the provision of a method and apparatus for noncontact determination of the temperature and the emissivity, within a predetermined narrow wavelength range, of a radiating body.

The invention thus provides for measuring total energy Δ E₁₂ and Δ E₃₄ radiated by a body of volume V, of unknown temperature T and of unknown radiative emissivity β in each of two predetermined, adjacent wavelength ranges λ₁≦λ≦λ ₂ and λ₃≦λ≦λ ₄, respectively; forming the differentials R₁₂ = ΔE₁₂/(λ₂-λ₁)V and R₃₄ = ΔE₃₄/(λ₄-λ₃)V; expressing the logarithms of these differentials as linear functions of the variables log_e(β) and ¹/_T in the form

log_e(R₁₂) = a₁₂ + b₁₂log_e(β) + c₁₂ ($\frac{\text{1}}{\text{T}}$),

log_e(R₃₄) = a₃₄ + b₃₄log_e(β) + c₃₄ ($\frac{\text{1}}{\text{T}}$),

where a₁₂, b₁₂, c₁₂, a₃₄, b₃₄ and c₃₄ are known parameters that are substantially independent of the variables β and T and each such parameter may be determined from a knowledge of the wavelength values λ₁ and λ₂ or of the wavelength values λ₃ and λ₄; and determining the quantities T and β substantially according to the relations

This invention thus provides a method of and an apparatus for determining simultaneously temperature and the radiative emissivity, in one or more narrow predetermined wavelength ranges, of a thermally radiating body that is at a substantially uniform temperature. Also, the invention provides a method of and an apparatus for producing electromagnetic radiation of wavelengths in a narrow predetermined wavelength range that may be used in determining the temperature and radiative emissivity of a body as indicated above.

The invention is further described below by way of example with reference to the accompanying drawings, in which:

• Figures 1 and 2 schematically show respective different forms of apparatus suitable for producing, from a beam of electromagnetic radiation containing a band of wavelengths, radiation in two or more predetermined adjacent narrow wavelength ranges, the apparatus of Figure 1 being a discrete apparatus suitable for producing and analyzing four wavelength ranges, and that of Figure 2 being a monolithic embodiment suitable for producing and analyzing a range of wavelengths of arbitrary width.
• Figure 3 is a schematic view of a modification of the embodiment of Figure 2, used to produce and analyze a plurality of wavelength ranges.
• Figures 4 and 5 are schematic views of electrical configurations that may be used in determining the temperature and thermal emissivity of a radiating body using two options according to the invention.

This invention is intended to allow the measurement of the temperature of an object without physical contact with the object, independent of the composition, surface characteristics or transparency of the object. The nominal temperature measuring range is from about 350° C to about 1500° C, but this range can be extended by the use of optical components and sensors that are compatible with the appropriate wavelengths λ of light (λ>3µm) for lower temperatures, λ<0.5µm for higher temperatures).

Current temperature measurement devices using optical techniques rely on measurement of the total amount of light emitted by an object between two widely spaced wavelength (usually in the infrared). This measurement is related to the temperature of the object by the relation

T = (R_T/βk)^¼

where R_T is the total radiation measured, T is the object temperature in °K, k is the Stefan-Boltzmann constant and β is interpreted as an emissivity parameter (0≦β≦1).

One problem with this method is that the value of emissivity β varies with the object being measured. For scientific purposes, such as the measurement of plasma temperatures and other well characterized objects, emissivity β can be determined by careful measurement and the measuring device can be calibrated for that object. However, in industrial uses the variations in β for the objects being measured is often caused by variations in the manufacturing process that produced the objects, making the above equation almost useless.

A good example of this problem is in the measurement of the temperature of a silicon wafer, used in the manufacture of integrated circuits, in a rapid thermal annealer or in a selective CVD system. The emissivity variation from wafer to wafer may be as much as 50 percent, which can produce a temperature uncertainty of 20 percent or more; this is far too large for good process control.

The radiating body whose temperature is being measured is assumed to have a substantially uniform temperature. Such a body will radiate electromagnetic radiation ("light") at all wavelengths according to the Planck analysis of blackbody radiation. Planck's formula for the spectral distribution of blackbody radiation is derived and discussed in L. D. Landau and E. M. Lifshitz, Statistical Physics, Pergamon Press, 1958, pp. 172-179 and is given by

dE_ω = (Vh/π²c³) ω³dω/[exp(hω/kT)-1],      (1)

in terms of angular frequency ω, and

dE_λ = 16π²chVdλ/λ⁵[e^hc/λkT-1]      (2)

in terms of wavelength λ = ^2πc/ω = c/ν. Here h (= 6.62 x 10⁻²⁷erg-sec) is Planck's constant, V is the volume of the body, k(= 1.38 x 10⁻¹⁶erg/°K) is the Stefan-Boltzmann constant and c (= 3x10¹⁰cm/sec.) is the velocity of light in a vacuum.

If one integrates dE_ω above all angular frequencies ω one obtains as usual the total energy radiated by the body

E_T = (4σV/3c)T⁴ (ergs),

σ=π²k⁴/60h³c² = 5.67x10⁻⁵gm/sec³(°K)⁴.

Use of this relation is subject to the infirmities discussed above.

The invention approaches this differently, by decomposing the wavelength range or (angular) frequency range of radiation emitted by the body into two or more predetermined adjacent sub-ranges, in each of which the emissivity β of the body at that temperature is substantially constant (dependence of β upon λ or ω is assumed to be weak to that β is approximately constant in each sub-range. The portion of energy ΔE_λ in each of the sub-ranges of interest is then determined. This yields a sequence of n(≧2) or more simultaneous equations in n+1 unknowns, and by adopting one additional assumption this sequence can be solved for the unknowns.

If one integrates Eq. (2) for a greybody with emissivity β_λ<1 and wavelength range λ₁≦λ≦λ₂ one obtains The temperature range of particular interest here is 350° C ≦ T ≦ 1500° C or 0.054 eV≦ kT≦ 0.153 eV, and the wavelength range of particular interest is 0.5 µm(2.48 eV) ≦λ≦3µm(0.413eV). Thus, hνkT=hc/λkT»1 for the wavelength and temperature ranges of interest so that the square bracketed term in the denominator in Eq. (3) may be replaced by the term exp(^hc/λKT). Further, using the mean value theorem from calculus the term β in the integrand may be replaced by a constant term β_c, and the relation in Eq. (3) becomes

ΔE₁₂=8πchVβ F(λ₁,λ₂;T)      (4) where λ₁₂ is an appropriate wavelength in the interval

(λ₁,λ₂): λ₁₂⁵ ≅ 4λ₁⁴λ₂⁴/(λ₁ + λ₂)(λ₁² + λ₂²)

Equation (3) may be rewritten as

where the quantity R₁₂ can be experimentally determined.

Equation (6) can be restated in terms of logarithms as

log_e {8πch/λ₁₂⁵R₁₂} = hc/λ₁₂kT - log_aβ.      (7)

An analogous relation can be written for an adjacent wavelength interval λ₃≦ λ ≦ λ₄ ; this second interval may, but need not, partially overlap the first wavelength interval λ₁≦ λ ≦ λ₂ . If the two wavelength intervals ((λ₁, λ₂) and (λ₃, λ₄) are sufficiently small and sufficiently close to one another, a common value of emissivity β may be used and Eq. (7) for each wavelength interval becomes

log_e{c₁/λ₁₂⁵R₁₂} = c₂/λ₁₂T - log_eβ,      (8)

log_e{c₁/λ₃₄⁵R₃₄} = c₂/λ₃₄T - log_eβ,      (9)

c₁ = 8πhc,

c₂=₅hc/k,

λ₁₂ = 4λ₁⁴λ₂⁴/(λ₁+λ₂)(λ₁²+λ₂²),

λ₃₄⁵ = 4λ₃⁴λ₄⁴/(λ₃+λ₄)(λ₃²+λ₄²).

for which the formal solutions are

$\frac{\text{1}}{\text{T}}$ = log_e^λ34^⁵R34/^λ12^⁵R12}^λ12^λ34/^c2{^λ34 ^-λ12},      (10)

log_eβ = |λ₁₂log_e{c₁/λ₁₂^⁵R₁₂} -λ₃₄ log_e{c₁/λ₃₄^⁵R₃₄}| / {λ₃₄-λ₁₂}.      (11)

Also, it may be important to measure four or five separate wavebands due to non-thermal emission effects, such as optical resonances and fluorescence, that may produce false measurements for one or two of the wavebands. A microprocessor can delete these false readings and ignore them if four or five bands are measured instead of only two.

The apparatus to measure the lefthand quantities in Equations (8) and (9) above, by isolation of two or more wavelength or frequency bands, can take one of at least three forms, as illustrated in Figures 1, 2 and 3. The approach shown in Figure 1 uses beam splitters, band filters and discrete photodetectors to produce and analyze the wavelength bands. The radiating object 1 produces infrared radiation 2, part of which is collected and collimated by a telescope 3 or similar means and is directed toward a first infrared beam splitter 4A. The beam splitter 4A produces two partial beams 2B and 2C of infrared radiation having the same frequency composition as the collimated beam 2A received by the beam splitter 4A from the telescope 3. The partial beam 2B is received by a second beam splitter 4B that further splits the partial beam 2B into two partial beams 2B1 and 2B2, each having substantially the same wavelength composition as the original beam 2A. The partial beam 2B1 is received by an optical beam pass filter 5B1 that passes only a first narrow band of wavelengths such as λ₁ ≦ λ ≦ λ₂; and this narrow band radiation is received by an infrared photodetector 7B1 that is optimized for sensitivity to this narrow band of wavelengths. The photodetector would produce a signal representing a quantity such as ΔE₁₂ shown in Equation (4). In a similar manner, three other narrow bands of wavelengths may be produced by passing the associated partial beams of radiation 2B2, 2C1 and 2C2 through narrow band pass filters 5B2, 5C1 and 5C2, respectively, and passing the resulting narrow band radiation to three more infrared photodetectors 7B2, 7C1 and 7C2, respectively, that are also optimized for sensitivity to the narrow band of wavelengths received by each photodetector. This approach is, of course, not limited to production and analysis of four narrow bands of wavelengths; any number of narrow bands of wavelengths, overlapping or nonoverlapping, may be produced and analyzed by the apparatus illustrated in Figure 1.

Figure 2 illustrates the use of a monolithic detector array for producing and analyzing narrow bands of wavelengths. A radiating object 1 produces infrared radiation 2, and a portion of this radiation is received by a telescope 3 and collimated into a beam 2′ that is directed toward an infrared diffraction grating 6 that is oriented with a non-zero incidence angle ϑ_inc relative to the incoming beam 2′. Diffraction of the incoming beam 2′ by the grating 6 produces many orders of diffraction, each of which will be re-radiated from the diffraction grating 6 at a somewhat different diffraction angle ϑ_dn (for the nth order diffraction). Each order of diffraction of interest may be received by a suitably positioned infrared photodetector that is part of a photodetector array 8, where each such photodetector is optimized for sensitivity to a particular interval of radiation wavelengths that are received by such photodetector array 8.

Figure 3 illustrates a monolithic detector array that uses discrete detectors. The radiating object 1 produces infrared radiation 2, part of which is collected an collimated by a telescope or similar means 3 and is directed to an infrared diffraction grating 6 as in Figure 2. Each predetermined wavelength band (nth order) is received and reflected by a cylindrical, spherical or other convex reflector 13 that directs each of the wavelength bands to a different infrared photodetector 7 that is optimized for sensitivity to that particular wavelength band.

Figure 4 illustrates an electronic control circuit suitable for use with the discrete detector approaches illustrated in Figures 1 and 3. Each infrared photodetector 7 in an array of such photodetectors receives a narrow wavelength band of radiation, which produces a photodetector output signal that varies with the intensity of the wavelength band of radiation received. The photodetector output signal is received by a low noise, high gain amplifier 9 whose amplified output is directed to a sample-­and-hold circuit 10 that is used to provide discrimination in time of the radiation received by the associated detector 7. Each sample-and-hold circuit 10 produces an output signal for one of the narrow bands of radiation wavelength of interest; and the sample-and-hold circuit 10 output signals are all directed to an analog multiplexer 11 that provides switching of analog signals received through a single output terminal to a single analog-to-digital (A/D) converter that provides an output signal for a data bus 14. The logarithms of the quantities of interest may also be formed within the multiplexer 11 so that the output signals from the multiplexer 11 are the quantities that appear in Equations (8) and (9) above.

Figure 5 illustrates a control circuit suitable for use with the monolithic detector array approach illustrated in Figure 2. An infrared photodetector from an array 8 of such detectors receives the nth order diffraction as an input signal and passes the photodetector output signal to a low noise, high gain amplifier 9 that passes the amplified signal directly to an A/D converter 12 whose digitized output is delivered directly to a data bus 14. Logarithms of the quantities shown in Equations (8) and (9) may be formed within the A/D module 12 or elsewhere as desired.

The sample-and-hold devices shown in Figure 4 are used to ensure that all the wave bands are measured at the same time or within the same designated time intervals. The electronics for the monolithic detector array of Figure 2, shown in Figure 5, would not require the sample-and-hold devices or the analog multiplexer and would need only one amplifier as these functions would effectively be performed by the monolithic detector array under control of additional electronics that would be directed by the control signals on a control bus. The design of the control electronics would depend on the choice of the particular detector to be used.

One additional requirement of the system is that all optical components of the system must be transparent (transmission ≧ 80%) from 500 nm to 3000 nm. This will produce accurate temperature readings down to 400° C. a requirement of this type means that the optics should be made of infrared transmitters such as sapphire or CaF₂.

Although the preferred embodiment of the invention has been shown and described herein, variation and modification may be made without deporting from the scope of the invention.