New Methods for Measuring the hermal Emissivity of Semi-transparent and Opaque Materials By D. Demange, M. Bejet, and B. Dufour ONERA - DMSC - Fort de Palaiseau, Chemin de la Hunière 91761 Palaiseau, Frane CEDEX el: 33 () 1 69 93 61 23, Fax: 33 () 1 69 93 61 25, E-mail: demange@onera.fr Abstrat wo original thermal emissivity measurement devies were developed reently in the laboratory. One of these devies uses a soure as a means of heating a sample, and exhibits a ertain number of advantages, inluding the absene of parasiti radiation on the surfae studied, and the temperature homogeneity of the sample. his method was applied in measuring the emissivity of bulk eramis and erami deposits on metalli or erami substrates. he seond test faility, applied to opaque materials, measures the diretional spetral emissivity and makes it possible to work bak by alulation to the total hemispherial spetral emissivity. As an example, these two methods were used for haraterizing thermal barriers for turbine vanes and materials that ould be used in a High emperature Reator (HR). 1. Introdution In order to inlude internal or surfae thermal radiation in odes prediting heating in insulating strutures or in metalli systems where thermal radiation is preponderant, two original experimental devies was developed allowing the measurement of spetral thermal emissivity of semi-transparent materials and the diretional spetral thermal emissivity of opaque materials. As an example, these two methods were used for haraterizing thermal barriers for turbine vanes and materials that ould be used in a High emperature Reator (HR). o ensure orret measurements on semi-transparent materials, one must avoid all spurious radiation suh as blak body radiation from a furnae for instane. he use of a laser soure emitting at 1.6 µm allows one to heat the sample without any stray radiation. he method used is to rotate a ylindrial sample about its axis and to expose a setion of it to a high-power laser beam. he radiation emitted by the opposing setion is then analysed by an FIR spetrometer. o measure the diretional spetral emissivity of metals or more generally of opaque materials, a more standard apparatus has been set up. Heating of the speimen is arried out by a miro-furnae with eletrial resistane. he samples are put into ontat with one of the faes of a heating ylinder, and the whole apparatus is put into a testing enlosure under a seondary vauum (1-5 torr) to avoid all oxidation during the measurement. he nonpolluting miro-furnae is put into the enlosure and brings the sample to the measurement temperature. 2. Desription of experimental means he essential differene between the two testing means resides in the sample heating modes. For semi-transparent materials, the heating mode is a ontinuous power laser. his devie is, however, limited to measuring the normal spetral emissivity of the materials and thus annot be used to measure the diretional emissivity. he total hemispherial emissivity thus annot be measured. Another test faility was therefore developed to measure the diretional spetral emissivity diretly and to work bak Figure 1 Overall view of the test faility. to the total hemispherial spetral emissivity by alulation. Here, the sample is heated by ontat on one of the faes of a pivoting ylinder raised to the measurement temperature. Figure 1 shows the whole devie.

2.1. Measuring spetral thermal emissivity in semi-transparent materials Knowing the thermo-radiative properties of semi-transparent materials is primordial for studies aimed at improving the insulating properties of thermal barriers or protetions in the aerospae setor. he properties of these materials depend on the opti harateristis of the dense bulk material and on its porous or fiber texture. he important quantities are the normal or diretional spetral thermal emissivity and the absorption and diffusion oeffiients. Kirhhoff s seond law gives us the following relation: R and ε 1 R = (1),,, In this expression are the refletion and transmission oeffiients normal to the sample. here,, exists a "Christiansen" wavelength where the oeffiients R and are lose to and ε,,, is greater than.999 [4]. For wavelengths greater than the Christiansen wavelength, the transmission oeffiient is lose to (figure 6) and we have ε, 1 R. he emissivity is independent of the material thikness. For wavelengths, less than the Christiansen wavelength, the transmission oeffiient will depend on the material s absorption and diffusion oeffiient and on its texture. he emissivity may depend on the porosity and thikness of the sample being measured. For emissivity measurements on semi-transparent materials, all parasiti radiation from the heating soure (a furnae, for example) has to be avoided. A laser soure an be used for heating if the emission wavelength is Eletri motor owards FIR spetrometer Sample in rotation 6 mm diameter, 2 mm long Aperture Figure 2 View of the measurement ell Copper ylinder Aperture for introduing laser beam hermoouple within the material s highabsorptivity spetrum. CO 2 lasers emit at 1.6 µm, whih is within the wavelength domain ( ) lose to the Christiansen point. We developed a tehnique that onsists of rotating a ylindrial sample (figure 2) about its axis and exposing a portion of it to a high-power laser beam. he radiation emitted by the portion on the opposing side is then analyzed by spetrometer with Fourier transform, operating in the mean infrared (2 to 25 µm). he zone observed is thus free of parasiti radiation (no furnae), and the sample rotation ensures lateral homogenization of the sample (external heating). Using a multi-mode laser ensures a rather good longitudinal homogeneity of the sample. With this tehnique, we an measure the absolute temperature of the sample with a thermoouple implanted at the enter of the ylinder. A blakbody studied at Onera provides the spetral emission referene. Figure 3 Heating devie View of samples.

2.2. Measuring diretional spetral thermal emissivity in opaque materials he priniple onsists in omparing the radiation emitted by the material s surfae to that emitted by a Vauum hamber blakbody. wo samples are plaed in a test hamber held at seondary vauum (<1-5 torr) in order to avoid any oxidation during the measurement. A massive ylinder of non-polluting Furnae Inonel is raised to temperature by a hermooax eletri resistor. he two samples are plated on one of the sides of the ylinder and are raised to the ZnSE window Sample measurement temperature (2 C to 85 C). ZnSe lenses he sample is oriented at an inident angle θ that an be varied from to 85 with respet to the normal. It is raised to temperatures between 25 C and 85 C. he diretional spetral luminane is measured with the same Fourier transform spetrometer as in the previous method. A Pyrox Mirror blakbody provides the spetral emission referene. he surfae temperature is measured by two N type Blakbody thermoouples implanted in the enter of the sample, and a orretion is made for the material s internal thermal gradient. Pinpoint Figure 4 summarizes the priniple of the whole measurement devie. A ooled ylindrial vauum hamber is made in Parabolique mirror stainless steel. It is equipped with a pumping devie to ahieve a vauum of better than 1-5 orr. A step o spetrometer motor is mounted on the top of the hamber to rotate the furnae during the tests and thus allow measurements for inidene angles between and 85. he sample is a parallelepiped 8 mm wide, 16 mm long and 2 mm thik. wo holes of.6 mm are Figure4 Overall view of the diretional spetral thermal provided in the diretion parallel to the sample plane emissivity measurement rig. and.5 mm from the observed fae, for the installation of two N type thermoouples (figure 3). 2.3. Determination of emissivity For a given system, the spetrometer response (s) depends only on the temperature (following a non-linear law) and the emissivity of the surfae, whih gives us: ε ε s = n s matériau For a referene blakbody, we have ε 1. référene Atually, the blakbody emissivity is always less than unity, and there may exist a temperature differene, even very small, between the CN and the sample, induing a non-negligible error in the emissivity measurement. So a sample of pure alumina provides a seond referene around (Christiansen wavelength) where we have ε 1. s matériau If R = is the ratio of the signals (s) generated by the IR spetrometer, and if we onsider a s n global orretion oeffiient k lose to unity and independent of the wavelength, we then have: ε = k R (3) n (2)

Figure 5 Solid angle diagram On sapphire, for = we get: ε k = R (4) ε he hemispherial spetral emissivity is then obtained by the relation ( 1 R If we allow 1 we will get k. he total diretional emissivity is alulated by the following relation: max ε L d, θ,, min ε, (5) θ max L d, min dω = sin θ dθ dϕ ): ε = 1 ε θ L dω π os (6),, θ, Ω As well as the total hemispherial emissivity, whih is equal to: max ε L d, min ε = (7) max L d, 3. DAA ANALYSIS min 3.1. Measurement of alumina emissivity, omparison of data obtained by the two methods Emissivity 1..8.6.4.2. Spetral emissivity of pure alumina (> 99.7%) Comparison with the two measurement methods Sample heated by ontat with Inonel furnae Sample heated by laser impat 2 4 6 8 1 12 14 16 Figure 6 - Spetral emissivity of pure alumina (> 99.7%) Comparison with two different methods Data obtained on dense highpurity alumina (>99,7%Al 2O 3) show the effet of the sample heating mode (method 1 or 2). Figure 6 shows three distint regions. he first is a zone between wavelengths of 2 and 4.5 µm where the emissivity is very different between the two methods and is muh higher when the sample of 2 mm thikness is raised to temperature by ontat with a heating ylinder. Here, the radiation emitted by the metalli surfae is transmitted through the alumina and radiates outward, whih gives us a high apparent emissivity. When heated by laser, only the material radiates and the emissivity is thus very weak despite a Wavelenght (mirons) ylindrial sample 6 mm in diameter. In the seond region between 4.5 and 11 µm, the emissivity is muh higher, the material s absorption oeffiient ε (Christiansen point). For wavelengths beyond, the inreases, and for = = 9.5 µm we have 1, material is opaque and we have ε, 1 R, whih explains why we get the same emissivity for both methods.,

3.2. Measurement of alumina emissivity, omparison of data obtained as a funtion of texture Emissivity 1..8.6.4.2 Spetral emissivity of Al2O3 Comparison of data aording to struture Pure alumina at 99.7% with 25% porosity Pure dense Alumina at 99.7% Sapphire. 2 4 6 8 1 12 14 16 Wavelenght (mirons) Figure 7 - Spetral emissivity of Al 2O 3 Comparison of data aording to struture he material s texture may have a nonnegligible effet on its emissivity. On aluminum oxide, Al 2O 3, figure 7 shows the low emissivity of sapphire (aluminum oxide single rystal) over a spetrum between 2 and 3 µm. High-purity polyrystalline alumina (>99.7% Al 2O 3) without porosity exhibits an emissivity of.8 over the same spetrum. A seond alumina sample with 3% porosity shows its emissivity inreased up to.2. 3.3. Measurement of diretional spetral emissivity for metals Figure 8 gives an example of data obtained on stainless steel at 55 C. Diretional spetral emissivity of a stainless steel at 23 C.5.45.4 Emissivity.35.3.25 Sighting angle = 8 Sighting angle = 6.2.15.1.5 Sighting angle = Sighting angle = 5 Sighting angle = 4. 4 6 8 1 12 14 16 wavelenght (mirons) Figure 8 Diretional spetral emissivity of a stainless steel at 55 C We observe a derease of the emissivity as a funtion of wavelength, and an inrease as a funtion of observation angle. 3.4. Example of appliation: Measurement of the speifi heat of eramis by flash method he priniple of the method is to measure the temperature rise of a ylindrial sample after exposure of one of its faes to an energeti Dira pulse. With the absolute rise in temperature and the known energy effetively absorbed by the sample, we an alulate the material s speifi heat simply, by theory. An insulated sample of thikness e, mass m, speifi heat Cp, at temperature θ, is exposed to a short pulse of energy density (Q). After homogenization of its temperature it will undergo a maximum heating equal to: θ lim -θ = Q lim = ρ.c P.e he speifi heat an then be expressed by a relation of the form: * U. a max las P max ( ε l p C mu U ) (9) pyr pyr (8)

Speifi heat (J/Kg C) 14 12 1 8 6 4 2 Figure 9 Measurement of the thermal diffusivity of an alumina and of the zironium with 4.5% Y 2 3 In this expression, U las and U pyr represent the signals output by the laser alorimeter and by the pyrometer. he two oeffiients a l et ε represent, respetively, the absorption oeffiient under the laser beam and the emissivity over the pyrometer s sensitivity spetrum. * is the maximum redued temperature max [1], [2]. his method thus requires the knowledge of the produt a ε. For eramis, these two oeffiients are lose to unity for a wavelength range that is generally between 6 and 11 µm (proximity of the Christiansen point). So this property allows us to do without these quantities as long as we use a laser soure emitting on a domain of wavelength where the material has a high absorption, and a detetor equipped with filters making it possible to obtain an adequate spetral response with the material s maximum emissivity range. he devie for measuring the spetral thermal emissivity of semi-transparent materials, desribed in setion 2.1, is very useful beause it an be used to determine the value of the produt a ε and use relation (9) if this is very different from unity. Examples of data obtained on two eramis are given in figure 9. 4. Conlusion wo original thermal emissivity measurement devies were developed reently in the laboratory. One of these devies uses a soure as a means of heating a sample, and exhibits a ertain number of advantages, inluding the absene of parasiti radiation on the surfae studied, and the temperature homogeneity of the sample. his method was applied in measuring the emissivity of bulk eramis and erami deposits on metalli or erami substrates. he seond test faility, applied to opaque materials, measures the diretional spetral emissivity and makes it possible to work bak by alulation to the total hemispherial spetral emissivity. he sample is heated by ontat on one of the faes of a pivoting ylinder raised to the measurement temperature. his method is used for haraterizing metalli or arbon-based materials and an measure the speifi heat of eramis by the flash method. 5. Référenes Speifi heat of pure alumina and of zironia with 4,5 % Y 2O 3 Alumina run 2 Alumina run 1 Zironia with 4,5% yttria 2 4 6 8 1 12 14 emperature ( C) [1] H. S. Carslaw, and J.C. Jaeger. Condution of heat in solids. Oxford University Press (1959). [2] D. Demange New appliation of the "flash" method. 15 th European on hermophysial Properties (ECP) High temperature high pressures 22, volume 34, pages 9-18 [3] R. Siegel and C. Spukler, Analysis of thermal radiation effets on temperature in turbine engine thermal barrier oating - Materials Siene Engineering A245 15-159 1998. [4] B. Rousseau, J.F. Brun, D. de Sousa Meneses, and P. Ehegut emperature measurement : Christiansen and blakbody referene. International journal of thermophysis, vol. 26, n o. 4, July 25 l p p l p