Radiation (non-contact) thermometry is a crucial technique widely employed in various industries and scientific fields to measure temperatures of objects without the need for direct contact. The accurate determination of temperature is vital for ensuring product quality, safety, and process optimization in diverse applications ranging from manufacturing to space exploration. However, the accurate calibration of radiation thermometers poses significant challenges, particularly at high temperatures.
In this paper, we present practical examples of the use of high-temperature fixed points (HTFPs) as a reference for calibration in applied radiation thermometry above the freezing temperature of Cu (1084,62 °C).
HTFPs in general
Since their introduction in 1999 HTFPs have been advanced and taken up by NMIs to become valuable for the realization and dissemination of temperatures above the freezing temperature of Cu [1]. This was achieved by a coordinated, worldwide research over a period of more than 20 years in part under the auspices of the Comité Consultatif de Thermométrie at the BIPM [2], and in part by a number of European lead and funded research programs [3, 4, 5, 6].
As a result of this research, robust manufacturing procedures have been established [7, 8], different fixed point cell designs with applications in radiation thermometry, contact thermometry and radiometry have been developed [9] and for a number HTFPs the thermodynamic temperature during the melt (defined as the point-of-inflection (POI) during the melting plateau) and their liquidus temperature was determined. Development with respect to thermodynamic temperature determination initially focused on three different metal-carbon alloys Co-C, Pt-C, Re-C [10]. For these alloys high purity materials were easy to source for cell manufacture and HTFPs out of these materials have proven to be robust. Their point-of-inflection values are reported in [10] and their liquidus values in [11]. During the last 15 years, additional eutectic and peritectic metal-carbon alloys were proposed. Within the framework of the European Metrology Programme for Innovation and Research (EMPIR) project Realising the Redefined Kelvin “Real-K” [6] four different materials were selected to close the temperature gaps between the so far developed HTFPs. These metals formed alloys that were metal-carbon eutectics, Fe-C (1153 °C), Pd-C (1492 °C), Ru-C (1953 °C) and a metal-carbide-carbon peritectic of WC-C (2748 °C). These four HTFPs were manufactured by different NMIs and compared to each other. The main selection criteria were the melting temperature (with higher melting temperature corresponding to a better cell), the plateau shape, the melting range (a smaller melting range is considered better) and the repeatability [12]. The best HTFP cells were selected for a campaign within Real-K to determine the thermodynamic temperature of both the POI during the melt and the liquidus temperature. Cells of slightly poorer quality were available within the project for a dissemination trial of thermodynamic temperature, with their thermodynamic temperature easily derived from the a priori known difference in temperature from the best cells.
Additionally, project partner CEM manufactured an Fe-C cell with 6mm cavity (compared to 3mm typically) to account for the larger measurement spot size of radiation thermometers often used in industrial applications. The temperature of this cell has been compared to the smaller aperture cells CEM provided in this work.
An overview of HTFP cells used here within the frame of this work is presented in Table 1. This table presents outer and cavity dimensions, manufacturer, and the observed differences to the HTFP cells investigated in [12].
Table 1: Overview of Fe-C and Pd-C high-temperature fixed point cells used for the dissemination trials
Material
|
Identifier
|
Manu-
facturer
|
Cell outer dimensions
|
Cavity inner dimensions
|
Observed difference to best cell
|
|
|
|
Diameter
/ mm
|
Length / mm
|
Diameter / mm
|
Length / mm
|
|
Fe-C
|
1FE-C3
|
CEM
|
24
|
45
|
6
|
35
|
-160 mK when measured in three-zone furnace [13]; and -200 mK below best HTFP cell in [12]
|
Fe-C
|
7Fe-C2
|
LNE-Cnam
|
24
|
45
|
3
|
35
|
0 mK
|
Fe-C
|
7Fe-C1
|
LNE-Cnam
|
24
|
45
|
3
|
35
|
-25 mK
|
Pd-C
|
1Pd-C3
|
CEM
|
24
|
45
|
3
|
35
|
-145 mK
|
Using HTFPs to disseminate the thermodynamic temperature scale
The mise-en-pratique for the definition of the kelvin (MeP-K) above the freezing temperature of silver (961,78 °C) facilitates the dissemination of thermodynamic temperature in two ways:
a) directly through a radiometer traceable to the radiant watt or
b) indirectly through HTFPs whose thermodynamic temperatures have been assigned either a priori or through calibration.
Up to now, the direct realization of thermodynamic temperature at high temperatures is usually only practicable, due to the complexity and the expense of the equipment required, for National Metrology Institutes (NMIs). This means that calibration laboratories aiming at obtaining temperature traceability, have to rely on a scale calibrated against transfer standards. Up until now the calibration artefact has almost universally been radiation thermometers calibrated against a source of known radiance. However, it is also possible to use HTFPs to provide traceability and potentially more reliably than radiation thermometers. Here, HTFPs with the known relative difference to cells of known thermodynamic temperature have been used both for the dissemination of thermodynamic temperature and for comparison to the local realization of ITS-90.
In the following sections, three dissemination trials are described. All three are limited to the temperature range below 1500 °C due to the available furnaces, which are based on ceramic heaters and work tubes.
Dissemination trial between CEM and SGS Tecnos (Spain)
To open a path for the dissemination of thermodynamic temperatures towards industry, SGS Tecnos S.A. in Spain was identified as an accredited laboratory for radiation thermometry with capabilities of installing, operating and measuring a Fe-C HTFP cell (the Fe-C cell labelled 1Fe-C3).
SGS Tecnos S.A. is a temperature laboratory whose accreditation scope includes a procedure for the calibration of infrared radiation thermometers from -30 °C to 1550 °C. SGS Tecnos S.A.U declares an expanded uncertainty at the Fe-C eutectic fixed point (1153 °C) of 4.3 K.
In this temperature range above 800 °C the laboratory has a Ø 48 mm black body cavity housed inside a LAND Instruments three-zone furnace as a thermal radiation source and a MIKRON M190 infrared thermometer with a resolution of 0.1 °C as the traceable reference standard. The Spanish National Metrology Institute CEM provides ITS-90 traceability to SGS through calibration of the MIKRON M190 with an expanded uncertainty of 2.5 °C from 1000 °C to 1600 °C.
Along with the Fe-C cell, CEM provided a quartz holding tube, a series of graphite and ceramic insulators housed inside the tube and an Ar purge system (Figure 1). The quartz tube holding, the Fe-C HTFP, insulators and Ar purge were fitted inside the black body cavity of the LAND Instruments three- zone furnace.
The calibrated MIKRON infrared thermometer was aligned and focused on the aperture of the Fe-C HTFP at a distance of 600 mm (a distance for which the target spot size for this radiation thermometer is minimum and has a diameter of 3.3 mm).
Six melting/freezing plateaus were recorded. Steps replicated values used at CEM for comparison purposes: ± 20 °C, ± 20 °C, -20 °C /+10 °C, -15 °C /+15 °C, -10 °C /+15 °C, -25 °C /+25 °C.
The average ITS-90 POI value obtained from the six cycles was 1153.5 °C with a standard deviation of 0.2 °C. This standard deviation is included to the uncertainty budget for temperature traceability labelled as “POI determination”.
Considering the systematic difference observed by CEM when comparing the large aperture cell to the small aperture Fe-C HTFPs cells in [13] (where a difference of -160 mK ± 250 mK was observed when using the cell 1Fe-C3 in a three-zone furnace) and in [12] (for this cell -200 mK difference at POI during the melt to best HTFP cell) and taking account a total correction of 360 mK is applied when comparing to results of the thermodynamic temperature for the transition temperature in [13] for this cell. Additional corrections, e.g. for emissivity ( estimated to be < 0.03 K at 650 nm) and temperature drop ( < 10 mK for the cell dimension in in [13]) were not considered.
The uncertainty budget for realising the HTFP using 1Fe-C3 at SGS Tecnos is shown in Table 2.
In summary the observed temperature t90= 1153.86 °C ± 0,47 K agrees well with the value derived for this fixed-point cell to the thermodynamic temperatures reported in [12] t= 1153.76 °C.
Table 2: Uncertainty budget for the temperature measurement with MIKRON M190
Quantity
|
Type
|
Uncertainty contribution, ºC
|
Sens. Coef.
|
Standard Uncertainty, ºC
|
t1Fe-C3 (see table 3 in [13])
|
1
|
0.12
|
1
|
0.120
|
Structure effect
|
1
|
0.014
|
1
|
0.014
|
POI determination
|
1
|
0.2
|
1
|
0.20
|
Device Under Test (DUT) resolution
|
rectangular
|
0.1
|
1
|
0.029
|
|
|
|
u(t)=
|
0.24 ºC
|
|
|
|
U(t)=
|
0.47 ºC
|
Dissemination activities at CMI
The Czech metrology institute (CMI) realizes the temperature scale by radiation thermometry over a temperature range from -35 °C to 1800 °C. Below the freezing temperature of silver (961.78 °C) traceability to ITS-90 is achieved via standard platinum resistance thermometer (SPRTs)s and thermocouples. CMI has only recently started to realize the temperature scale above 962 °C via Planck’s law in ratio form and using a radiation thermometer LP5 (λcenter = 649,06 nm) (i.e. ITS-90 above the silver point), which has been characterized for relative spectral radiance responsivity. A radiance reference is established through a Cu fixed point realised in a sodium heat pipe furnace (both from Isotech) [14]. CMI for the purposes of this research is considered as a traceability receiver through receiving a calibrated HTFP of Fe-C from another project partner.
Services in the field of radiation thermometry are covered with CMCs in a lower temperature region up to 962 °C, and currently only via accreditation in a high temperature region from 962 °C to 1800 °C with uncertainties 1.0 to 1.8 °C (k=2)
For HTFP measurements a three-zone furnace from the local manufacturer Clasic was used. This furnace has a total ceramic tube length of 80 cm and operates in the temperature range between 1000 °C to 1800 °C.
During the melt/freeze cycling of the HTFP this furnace was operated with a heating/cooling ramp of 5 °C/min. When heating the furnace from room temperature to 1100 °C the heating rate was set to 10 °C/min, which is also the maximum heating rate for this furnace. Identical ramp rates were used for cooling, as the furnace operates without active cooling. Note though that the cooling process is slower than the ramp rate because of the high thermal inertia of the furnace.
To implement the dissemination trial, the Fe-C HTFP (7Fe-C2, supplied by LNE-Cnam) was installed in the furnace.
Initially, it was planned to position the HTFP cell in the middle of the tube furnace, which has the most homogeneous temperature field (Pos. A in Figure 3). However, during the first set of measurements it became apparent, that due to the depth of the cavity it was difficult to align the radiation thermometer visually into the fixed-point cavity onto the Fe-C aperture. As a result, the Fe-C HTFP was moved 7.5 cm towards the front opening of the furnace (Pos. B in Figure 3) and the measurements were repeated at this position in the furnace. It is not thought that this change in position would have an adverse effect on the performance of the HTFP as the furnace has a high temperature uniformity.
Measurement of the Fe-C HTFP was realized with different temperature settings, the first two cycles with temperature steps around the melt of ±20 °C, the third with (+10/ -20) °C and the last one with (+15/ -20) °C steps. Due to furnace stabilization after initial warm-up, one additional ±20 °C melt-freeze cycle was added at the beginning of the measurement cycle. A sample measurement cycle is presented in Fig. 4.
From all these data, the POI the recorded melting curves was determined together with the melting range for each curve. An ITS-90 temperature value was then calculated relative to the copper freezing point of CMI. Results are given in table 3 below.
Table 3: CMI temperature for the Fe-C cell (7Fe-C2)
HTFP
|
t90
|
Reproducibility
|
Melting range
|
Expanded Uncertainty k=2
|
7Fe-C2
|
1153.54 °C
|
0.04 °C
|
0.18 °C
|
0.5 °C
|
In the evaluation of the total expanded uncertainty the following uncertainty contributions were considered: calibration of LP5 in the Cu fixed point, the drift of the LP5 at Cu freezing point, plateau identification (melting range), repeatability of plateau realization, wavelength of linear pyrometer, size of source effect, non-linearity, ambient conditions and others including the furnace effect [15].
The observed ITS-90 temperature of t90= 1153.54 °C ± 0,5 K agrees well with the value derived for this fixed-point cell from the thermodynamic temperatures reported in [12] 1153.77 °C ± 0,15 K. No correction was considered, cell 7Fe-C2 was the best Fe-C cells investigated in [12] . Additional corrections, e.g. for emissivity ( estimated to be less than 0.03 K at 650 nm) and temperature drop ( less than 10 mK for the cell dimension in in [12]) were not considered.
Dissemination activities at Tubitak
TUBITAK-UME is the NMI of Turkey and already has some experience with the use of HTFPs [7]. Here both Fe-C and Pd-C HTFPs were realized inside a three-zone furnace, open-ended alumina tube furnace (SiC heater, temperature range up to 1700 °C, with an inner length of 600 mm and inner tube diameter of 30 mm). This three-zone furnace was developed jointly between UME and a local company to extend UMEs capabilities of high-uniformity furnaces up to 1700 °C. The furnace has seven molybdenum disilicate MoSi2 heaters, three in the middle zone and two in each end zone. The multi-zone nature of the furnace allowes a temperature homogeneity in the central area better than 1°C to be achieved.
The tube’s internal diameter of 37 mm was designed specially to allow the realization of commonly used radiometric HTFPs cells with external diameters not exceeding 25 mm and with a cavity aperture of about 3 mm, as well as relatively large cells, designed for the calibration of thermocouples with an external diameter of 35 mm and thermo-well diameter about 8 mm.
For radiometric HTFP cells with an external cell diameter of about 25 mm, a special holder with an internal diameter of 26 mm and an outer diameter of 35 mm was constructed from high-purity and high-density graphite material. Several such holders with various lengths allow realization of most fixed points with different geometries without any challenge.
Within the work presented here, the following two cells were investigated: Fe-C (designated 7Fe-C1 and supplied by LNE-Cnam) and Pd-C (designated 1Pd-C3 and supplied by CEM). Different cell holders and insulation material sets were prepared and dedicated to each fixed point. For the measurement, three different temperature profiles were realized in the furnace dT=0 °C (no temperature gradient), dT=-10 °C (bottom of the cell at higher temperatures), dT=10 °C (front of the cell at higher temperatures). Here, only the results for dT=0 °C are presented, the other measurements allowed the furnace effect to be investigated. The results of that study are presented in [16]. The HTFPs were heated and cooled with a ramp of 8 °C per minute. All HTFP plateaus were obtained using an LP5 radiation thermometer with center wavelength at 650 nm. ITS-90 temperatures were determined by comparison to UME´s primary Cu freezing point blackbody. Table 4 lists the complete uncertainty budget for the determination of the ITS-90 temperatures of these two fixed-points.
Table 4: Uncertainty budget for the measurement for the ITS-90 temperature measurement of Fe-C and Pd-C fixed-points at TUBITAK-UME
Uncertainty Components
|
Standard Uncertainty/ °C
|
1084.62 °C
|
1154 °C
|
1493 °C
|
Components related to fixed-point cell
|
|
|
|
Impurity
|
0.012
|
0.020
|
0.022
|
Emissivity
|
0.042
|
0.046
|
0.070
|
POI determination
|
|
0.008
|
0.006
|
Structure effect
|
0.014
|
0.014
|
0.014
|
Cu freezing Plateau determination
|
0.009
|
0.014
|
0.022
|
Repeability
|
0.010
|
0.016
|
0.018
|
Factors related to spectral responsivity
|
|
|
Wavelength
|
0.000
|
0.003
|
0.024
|
Repeability
|
0.000
|
0.002
|
0.016
|
Drift
|
0.000
|
0.025
|
0.189
|
Out Of Band transmission
|
0.000
|
0.001
|
0.007
|
Factors related to out-of-signal effect
|
|
|
Furnace Effect
|
0.034
|
0.034
|
0.034
|
Stability
|
0.026
|
0.026
|
0.026
|
Size-of-source
|
0.024
|
0.027
|
0.041
|
Non-linearity
|
0.008
|
0.009
|
0.014
|
Drift
|
0.010
|
0.011
|
0.016
|
Expanded uncertainty (k=2)
|
0.14
|
0.16
|
0.43
|
For the HTFP of Fe-C the measured ITS-90 temperature was 1153.61 °C ± 0,16 K which agrees well with the value derived for this HTFP from the thermodynamic temperatures reported in [12] t= 1153.77 °C ± 0,15 K. A correction of -25 mK was performed as this was the observed difference of this HTFP relative to the best Fe-C HTFPs in [12] . Additional corrections, e.g. for emissivity ( estimated to be less than 0.03 K at 650 nm) and temperature drop ( less than 10 mK for the cell dimension in in [12]) were not considered.
For the HTFP of Pd-C the measured ITS-90 temperatures was 1491,94 °C ± 0,43 K which agrees well with the value derived for this HTFP from the thermodynamic temperatures reported in [12] t= 1491.75 °C ± 0.16 K. A correction of -145 mK was performed as this was the observed difference of this HTFP relative to the best Pd-C cells in [12] . Additional corrections, e.g. for emissivity (estimated to be less than 0.04 K at 650 nm) and temperature drop (estimated to be less than 0.03 K for the cell dimension in in [12]) were not considered.