Home

Search

Collections

Journals

About

Contact us

My IOPscience

Validation of the fixed-point realization of the thermocouple scale using a radiation

thermometer in the temperature range between the freezing points of Ag and Pd

This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2010 Metrologia 47 239
(http://iopscience.iop.org/0026-1394/47/3/014)
View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 200.27.124.194
The article was downloaded on 03/05/2010 at 15:16

Please note that terms and conditions apply.

IOP PUBLISHING

METROLOGIA

Metrologia 47 (2010) 239247

doi:10.1088/0026-1394/47/3/014

Validation of the fixed-point realization of

the thermocouple scale using a radiation
thermometer in the temperature range
between the freezing points of Ag and Pd
Yong-Gyoo Kim, Bong Hak Kim and Inseok Yang1
Division of Physical Metrology, Korea Research Institute of Standards and Science (KRISS), Daejeon,
305-340, Korea
E-mail: iyang@kriss.re.kr

Received 1 December 2009, in final form 19 February 2010

Published 12 April 2010
Online at stacks.iop.org/Met/47/239
Abstract
We have developed a blackbody comparison system to validate the fixed-point realization of
the thermocouple temperature scale using the radiation thermometry scale.
Platinum/palladium, type S and type B thermocouples were calibrated at the fixed points of
Ag, Cu, FeC, CoC and Pd (only for type B). The radiation thermometer used was an LP4
linear pyrometer operating at a central wavelength of 650 nm. To assign the FeC eutectic
melting temperature in our laboratory, a radiometric FeC cell was fabricated, and its melting
temperature was determined as (1154.0 0.1) C with k = 2. Two scales were compared
from 962 C to 1555 C in the blackbody comparison system. The two scales were consistent
within 0.5 C up to 1400 C, but the discrepancy for the type B thermocouple increased to
2.3 C at 1554.8 C. The thermocouple and radiometric scales realized at our laboratory were
in agreement up to the freezing temperature of Pd within the measurement uncertainty.
(Some figures in this article are in colour only in the electronic version)

1. Introduction
Thermocouples are widely used in industry to measure
temperatures above 1000 C. Dissemination of temperature
standards through thermocouples is achieved by calibrating
thermocouples at metal fixed points to obtain the best
uncertainty. The pure-metal fixed-point cells normally used
for such calibrations are Pd (1554.8 C) and Cu (1084.62 C).
Thermocouples can be calibrated using these fixed points,
and the deviation function can be interpolated. However, the
temperature difference between the freezing points of Cu and
Pd is about 470 C. To reduce the interpolation error in this
temperature range, many metalcarbon eutectic points have
been studied in several national metrological institutions [1].
The FeC, CoC (or NiC) and PdC eutectic fixed points
have been implemented to calibrate thermocouples up to the
freezing temperature of Pd [26]. By adding these eutectic
1

Author to whom any correspondence should be addressed.

0026-1394/10/030239+09$30.00

points between the two pure-metal fixed points of Cu and Pd,

the deviation function of the thermocouple calibrations can be
improved. However, there is a need to ensure the exact melting
temperature of any eutectic point.
The freezing temperature of Ag, Cu and Pd has been
defined in the International Temperature Scale of 1990
(ITS-90) and other measurements traceable to ITS-90. In
addition, the CoC eutectic temperature has been proposed as
being 1324.0 C with an uncertainty of 0.6 C [1]. However,
the melting temperature of FeC in the literature varies from
1153 C to 1154.17 C [711]. Furthermore, our experience
with a Pt/Pd thermocouple calibrated at the Ag and Cu points
and then extrapolated to higher temperature indicated that the
emf measured at the CoC eutectic point using the proposed
temperature value lay well on the extrapolated deviation
function, but the temperature value of the FeC needed to be
verified.
In this work, we first assigned the melting temperature
of an FeC cell used in our laboratory.
For this, a

2010 BIPM & IOP Publishing Ltd

Printed in the UK & the USA

239

Y-G Kim et al

35

Aperture/Cavity
2/4
60

Sleeves
Body

crucible were 60 mm and 35 mm, respectively. When the cell

was filled with the metal sample, the bottom lid was removed,
and an additional dummy cylinder with a length of 80 mm was
attached. After filling the powders into the cell, it was inserted
into an alumina tube, and melted in an atmosphere of pure
Ar using a vertical high-temperature furnace (designated as
KSF-V). Using this method, the cell could be fully filled with
a single filling.
2.2. Preparation of the thermocouples

Metal
Lid
Figure 1. Schematic diagram of the FeC eutectic black body cell
made of graphite for the radiation thermometer. Dimensions shown
are in millimetres.

radiometric cell using the same materials as our FeC cell

for thermocouples was fabricated, and its melting temperature
was measured using a radiation thermometer. Then, we
built a blackbody block system to validate the thermcouple
temperature scale by comparing the thermocouple temperature
scale calibrated using fixed points and a radiation thermometer
in the temperature range between the freezing points of Ag and
Pd. We verified the accuracy of our temperature scale and the
thermocouple reference functions. A radiation thermometer
is the interpolating thermometer in ITS-90 above the fixed
point of Ag. Therefore, the thermocouple temperature scale
is traceable to ITS-90 through a comparison with a standard
radiation thermometer. Such a comparison was carried out
using type S and type B thermocouples, which are widely
used in accurate temperature measurements, and two Pt/Pd
thermocouples, which are gaining importance in research
in high-temperature thermometry. The deviation functions
of the thermocouples were calculated and the temperature
scales were compared using a comparison blackbody block.
The behaviour of these deviation functions can vary for
different thermocouples as well. Our aim was to improve
temperature standards using thermocouple thermometry based
on the results of our comparison up to the freezing temperature
of Pd.

2. Experiments
2.1. Radiometric FeC cell
A graphite crucible cell was designed, as shown in figure 1,
and this was used to realize the eutectic points. The cell
had two sleeves (thickness = 2 mm) to prevent breakage
during the heating/cooling cycles. High-purity iron powder
(nominal purity = 99.998%) and carbon powder (nominal
purity = 99.9995%) obtained from Johnson Matthey Co. (UK)
was used in the cell. The graphite crucible with 99.999% purity
was obtained from the Carbone Lorraine Co. (Model 2320,
France, ash content <5 ppm). The total mass of the eutectic
ingot was about 49.3 g. The aperture diameter was 2 mm and
the cavity diameter was 4 mm. The depth of the blackbody
cavity was 45 mm. The overall length and outer diameter of the
240

One type S thermocouple (designated as HSTC), one type

B thermocouple (designated as HBTC) and two Pt/Pd
thermocouples (designated as HPtPd 1 and HPtPd 2) were
prepared for this work. The thermocouple wires had a
diameter of 0.5 mm, and were purchased from the Heraeus
Co. (Germany). The Pt and Pd wire thermoelements were
electrically annealed at 1450 C and 1300 C, respectively, for
a period of 1 h, and then at approximately 400 C overnight.
The type B and type S thermoelements were both annealed
electrically at 1700 C and 1450 C, respectively. After
annealing, the thermoelements were inserted into high-purity
twin-bore alumina insulators (length = 600 mm, diameter =
3 mm and bore diameter = 0.8 mm), which were pre-baked
at a high temperature. The measuring junction of the type S
and Pt/Pd thermocouples was made by connecting the legs
using a 0.1 mm diameter Pt wire to reduce the strain caused
by the different degrees of thermal expansion. The type B
thermocouple had a normal welded junction. After assembly,
the thermocouples were annealed at 1000 C for a period of
20 h in a vertical electric furnace to remove any deformation
formed during assembly.
2.3. Fixed-point cells
The thermocouples were calibrated using the freezing points
of Ag, Cu and Pd, and the melting points of FeC and CoC
eutectic alloys. The sealed-type Ag and Cu cells used in
this work were made at KRISS (Korea Research Institute
of Standards and Science, Korea) to disseminate temperature
standards. The depth from the bottom of the thermometer well
to the top of the metal was 22 cm. The FeC and CoC cells
were made in 2007, and the design and materials were the same
as those fabricated and reported in an earlier publication [2].
The Pd point was realized in a pure Ar atmosphere using a
method described elsewhere [12].
2.4. High-temperature furnaces
Three high-temperature furnaces were used in this work. A
vertical sodium heat-pipe furnace (designated as FPF-V) was
used to realize the Ag and Cu fixed points. Kanthal super heater
furnaces were used above 1100 C. Furnace KSF-V was used to
realize the FeC, CoC and Pd points, and a horizontal furnace
(designated as KSF-H) was used for the comparison of the
radiation thermometer and thermocouples. The temperature
of this furnace could be increased to 1544 C in our setup
conditions. The temperature gradient of the three furnaces
was measured at several temperatures.
Metrologia, 47 (2010) 239247

Validation of the fixed-point realization of the thermocouple scale

Table 1. Temperature gradient analysis of the furnaces for each

temperature using a thermocouple.

1400

Temperature / C

1200
1000
800
KSF-H
KSF-V
FPF-V
KSF-H(t95%)

600
400

KSF-V(t95%)

200
0

FPF-V(t95%)

10

20

30

40

50

60

Depth from the cell bottom /cm

Figure 2. Temperature profiles of three furnaces of FPF-V (used to
realize the Ag and Cu fixed points, profile measured at 960 C),
KSF-V (used to realize the FeC, CoC and Pd points, profile
measured at 1320 C) and KSF-H (used for the comparison of the
radiation thermometers and thermocouples, profile measured at
1400 C) as a function of the distance from the bottom of the cell.
The dashed lines represent the temperatures below which 95% of
the emf was produced.

Figure 2 shows the temperature profile as a function of

the distance from the bottom of the cell at 960 C for FPFV (triangles), 1320 C for KSF-V (rectangles) and 1400 C
for KSF-H (solid circles). The monitoring thermocouple was
located at its maximum immersion position in the cell, and
the temperature profile was measured by withdrawing the
thermocouple. The temperature profiles at other temperatures
were nearly the same for the same furnaces, as shown in
figure 2.
The temperature uniformity of furnace FPF-V was as high
as 1.0 C up to a distance of 23 cm from the bottom of the
furnace. This length covers the entire length of the Ag and Cu
freezing cells. The temperature profile was also the same as
that near the Cu freezing temperature. As the distance from
the bottom of the furnace increased, the temperature gradually
decreased. A large gradient formed from a distance of 38 cm
to the neck of the furnace. In the case of furnace KSF-V, the
temperature was uniform within 1.0 C up to a distance of
10 cm from the bottom of the furnace. Because the length of
the eutectic cell was 10 cm, it was thought that this furnace
would be adequate for realizing the eutectic fixed points [13].
However, above a distance of 22 cm from the bottom of the
furnace, a strong gradient developed, and generated most of
the thermal emf of the thermocouples. Therefore, most of
the uncertainty due to the thermocouple inhomogeneity was
generated from the portion of the thermocouple that lay in
this range. The temperature profile of furnace KSF-H was
measured from the location of the rear side of the comparison
block, because a thermocouple was placed in the rear side. The
temperature profile of furnace KSF-H was measured during
the comparison tests of the radiation thermometer and the
thermocouples. A temperature drop of only 2 C was found
up to a distance of 10 cm, and thus, the temperature uniformity
inside the comparison block was 1.0 C. The temperature
dropped steeply at a distance of 19 cm from the end of the
thermometer well in the block.
Metrologia, 47 (2010) 239247

Furnace

Test
temperature/ C

Uniform
zone/cm

Gradient
zone/cm

FPF-V
KSF-V
KSF-H

960
1320
1400

038
022
019

3850
2258
1940

Since the thermal emf is generated in a region of the

temperature gradient, we defined a uniform zone and a gradient
zone in the furnaces. In this work, a gradient zone was
defined as the region generating 95% of the thermal emf, E95% ,
from the set temperature to 50 C, which was approximately
the temperature of the thermometer neck of a furnace. The
temperature t95% , corresponding to E95% , is indicated by the
dashed lines in figure 2. The data given in figure 2 show
that the temperature fell markedly below t95% . Therefore, it
was suggested that the temperature gradient zone could be
reasonably represented by defining t95% . Table 1 summarizes
the results of the temperature gradient analysis of the furnaces
used in this work. This gradient zone was considered in the
calculations of the thermoelectric inhomogeneity as a large
uncertainty factor.
2.5. Thermoelectric voltage measurements
The reference junction used was a normal icewater mixture,
and pure copper wires were connected to the thermoelements.
The thermal emf was measured using a nanovoltmeter
(Keithley, Model 2182, USA) in all the thermocouple
measurements in this work. The deviation emf was calculated
by subtracting the reference emf from the measured emf at
each fixed point. The reference functions used were obtained
from ASTM standards [14, 15].
2.6. Radiation thermometer and comparison block
The radiation thermometer used in this work was an LP4
linear pyrometer (IKE, Germany) calibrated according to
ITS-90 using the freezing point of Cu employing a 650 mm
optical filter. The uncertainty budget of the calibration is
listed in table 2. The uncertainty of the calibration was
0.08 C (k = 2) at the Cu point, and it increased to 0.21 C
at 1544 C. A direct comparison of the thermocouples with
the radiation thermometer was carried out using a comparison
block up to 1544 C, which was the maximum temperature of
furnace KSF-H.
Figure 3 shows a schematic diagram of the intercomparison blackbody block made of high-purity graphite. The lefthand side of figure 3 shows the blackbody with a diameter =
15 mm and an aperture = 1.5 mm. The right-hand side of
figure 3 shows a thermocouple well designed for a 7 mm OD
thermocouple protecting tube. The thermocouple well was
located below the tilted face of the blackbody, off-centre of the
block. Originally, the block was designed for a 5 mm outer
diameter protecting tube made of platinum. However, in preliminary experiments, the platinum became contaminated and
melted because of unknown reactions that appeared to occur
241

Y-G Kim et al

Table 2. Uncertainty budget of the LP4 radiation thermometer.

Standard uncertainty/ C

Factors related to fixed-point cell

(1) Impurity
0.007
(2) Emissivity
0.006
(3) Cu freezing plateau
0.005
determination
(4) Repeatability
0.021
Factors related to spectral responsivity
(1) Wavelength
0
(2) Wavelength repeatability
0
(3) Wavelength drift
0
(4) Out-of-band transmission
0
Factors related to out-of-signal effect
(1) Size-of-source effect
0.004
(2) Non-linearity
0.017
(3) Drift
0.025
Expanded uncertainty (k = 2) 0.08

1544 C
0.013
0.011
0.009
Figure 4. Schematic diagram of the blackbody comparison system
of the radiation thermometer (left side) and the thermocouples
(right side).

0.038
0.055
0.011
0.055
0.011

1160
1158

0.007
0.030
0.045
0.21

1084.62 C

Temperature / C

Uncertainty
factors

100

1156
1154
+ 3 oC
+ 5 oC
+ 8 oC

1152

44.3

35

1.5

15

1150
1148
0

20

7.0

75

Figure 3. A schematic diagram of the intercomparison blackbody

block made of graphite. All dimensions are shown in millimetres.
The left-hand side is the blackbody aperture and the right-hand side
is the thermocouple well.

in the furnace. Therefore, a protecting tube made of alumina

for the thermocouples was employed instead. The blackbody
comparator was installed into the comparison system, as shown
in figure 4. The LP4 was placed on the left side of the furnace
and a thermocouple was inserted from the right side. To prevent radiation heat loss, disc-type radiation shields made of
alumina were installed as shown in figure 4. At the right side
of the block, an alumina insulator was also installed. LP4
was set to collect radiation using an objective lens with a focal
length of 600 mm to form a 1 : 1 image on a field stop with a
diameter of 0.5 mm. A small volume of high-purity argon gas
containing 0.5% of hydrogen by volume was used from the
right side during the experiments to prevent oxidation of the
graphite.

3. Results
3.1. Determination of the FeC melting temperature using
the radiation thermometer
To determine the melting temperature of the FeC eutectic
point in our laboratory, three melting and freezing cycles were
performed at several offset temperatures using our furnace, as
shown in figure 4, replacing the comparison block with the
radiometric FeC cell.
242

40

60

80

100

120

Time /min

1.0

Figure 5. Typical melting curves of the FeC radiometric cell at the

offset temperatures of 3 C, 5 C and 8 C above the melting
temperature. As the offset temperature increased, the plateau length
decreased without significantly affecting the melting temperature.
Table 3. Measured FeC eutectic melting temperature using the
LP-4 radiation thermometer.
Melting offset temperature
+3 C

+5 C

+8 C

Melting
temperature/ C

1153.92
1154.01
1154.00
1154.01
1154.02
1154.01
1154.02
1154.01
1154.02
(Mean 1 )/ C 1153.98 0.06 1154.01 0.01 1154.01 0.01

Figure 5 shows typical melting curves of the FeC

radiometric cell at each offset temperature.
Freezing
was always carried out at the same offset temperature
of 6.5 C below the melting temperature. As shown in
figure 5, the melting plateau was well developed at all offset
temperatures. As the offset temperature increased, the plateau
length decreased without significantly affecting the melting
temperature.
The melting temperature was assigned as the inflection
point of a melting curve. This point was calculated by
fitting the measured temperature as a function of time using
a third-order polynomial equation. Under this condition,
the calculated melting temperature of the inflection points
varied in most cases by not more than 0.01 C. The calculated
temperature for each melting reaction is shown in table 3. The
first melting was carried out at an offset temperature of +3 C,
and the calculated melting temperature was lower by 0.1 C
Metrologia, 47 (2010) 239247

Validation of the fixed-point realization of the thermocouple scale

Table 4. Thermal emf measured at various fixed points, before and after the comparison.
Measured emf /V
Before comparison

After comparison

Fixed
point

HPtPd 1

HPtPd 2

HSTC

HBTC

HPtPd 1

HPtPd 2

HSTC

HBTC

Ag
Cu
FeC
CoC
Pd

10 783.2
13 242.9
14 717.2
18 574.1

10 780.9
13 240.3
14 714.4
18 571.9

9 154.5
10 581.0
11 401.3
13 455.4

4 479.3
5 619.0
6 303.3
8 100.3
10 735.0

10 783.7
13 243.3
14 718.7
18 575.3

10 781.5
13 240.7
14 717.2
18 572.9

9 152.9
10 579.1
11 402.3
13 455.5

4 479.2
5 620.6
6 303.4
8 099.5
10 738.0

Table 5. Thermocouple emf with temperature measured using the radiation thermometer in the blackbody comparison block.
(Measured emf 1 )/V

Temperature/ C

HPtPd 1

HPtPd 2

HSTC

HBTC

961.78
1084.62
1154.0
1250
1324
1400
1450
1544

10 780.9 1.0
13 240.6 0.8
14 717.7 0.7
16 854.7 0.2
18 570.7 0.2
20 391.6 0.2

10 779.7 1.5
13 239.6 1.3
14 716.3 0.8
16 853.5 0.5
18 569.3 0.5
20 389.9 0.3

9 152.5 0.9
10 579.5 0.4
11 402.5 0.1
12 557.2 0.1
13 454.5 0.1
14 377.1 1.5

4 475.1 0.1
5 614.8 0.2
6 301.6 0.2
7 296.7 0.5
8 095.1 0.3

9 503.7 0.6
10 584.3 0.9

than the other melting temperatures. After the first melting

experiment, the eight other melting temperatures coincided
within a maximum deviation of 0.01 C, showing a very high
reproducibility.
The standard deviation of nine melting points was
calculated to 0.03 C. The calibration uncertainty of LP4 at
1154 C was 0.09 C for k = 2, and the short-term stability
of LP4 was found to be 0.025 C. An uncertainty factor
of 0.01 C was assigned for an alignment error in LP4.
The effect of impurities on the melting curve was assumed
to be 0.01 C. The combined uncertainty of the melting
temperature of the radiometric FeC cell was calculated to
0.05 C. The melting temperature of FeC was determined to
be 1154.0 C with an expanded uncertainty of 0.1 C (k = 2).

which was the maximum temperature of continuous usage for

those thermocouples. In the case of the type B thermocouple,
the calibration temperature of 1400 C was replaced by a
temperature of 1450 C, and the highest temperature used was
1544 C. Because the furnace set temperature did not exactly
coincide with the fixed-point temperatures shown in table 5,
the value of the actual measured temperature was corrected
for comparison with the fixed-point measurement results.
The furnace was set to maintain the calibration points for a
period of 1 h while the readings of the radiation thermometer
and the thermocouple were obtained. The comparison tests
were performed three times. Table 5 shows the measured
thermocouple emf values at specified temperatures with a
standard deviation. The standard deviation varied from
0.1 V to 1.5 V.

3.2. Fixed-point measurements

The thermocouples were calibrated using various hightemperature fixed points from Ag to Pd before and after
comparison with the radiation thermometer. The calibrations
were carried out from high to low temperatures. The melting
emf of the eutectic point was obtained as the inflection point of
the melting curve, and the freezing emf was obtained from the
maximum emf after supercooling. Table 4 shows the measured
fixed-point emf values before and after comparison. The fixedpoint emf was taken from an average value.
3.3. Comparison of the thermocouple with a radiation
thermometer
After the fixed-point emf measurements, the thermocouples
were installed in our comparison system. Comparisons were
performed from 960 C to 1544 C, near to the temperatures
listed in table 5. For the Pt/Pd and type S thermocouples,
the highest comparison temperature was limited to 1400 C,
Metrologia, 47 (2010) 239247

3.4. Uncertainty evaluation

The measurement uncertainty of the fixed-point calibration of
the thermocouples was evaluated to assess the measurement
uncertainty of the two scales. The accuracy of the freezing
temperature of the Ag and Cu sealed cells was estimated to
be 0.01 C. The uncertainty of the FeC and CoC melting
points was assigned to be 0.1 C and 0.6 C [1], respectively, for
k = 2. The reproducibility of the fixed points was taken from
previously reported results [2, 3]. The digital voltmeter used
was calibrated at KRISS, and its relative expanded uncertainty
was 8 106 . The short-term stability of the voltmeter was
taken from the equipment manual. The temperature of the
reference ice-point junction was stable to within 0.01 C
during the measurements. The change in emf before and after
the comparison tests was used to assess the short-term stability
of the thermocouples.
It was essential to perform inhomogeneity tests on the
thermocouples to calculate their measurement uncertainty. In
243

Y-G Kim et al

Table 6. Uncertainty budget of the fixed point and comparison

calibration at the freezing temperature of Cu for the type S
thermocouple.

-2
-4

E - Eref /V

-6
-8
-10
HPtPd_1
HPdPd_2
HSTC
HBTC

-12
-14
0

100

200

300

400

500

Immersion depth /mm

Figure 6. Inhomogeneity data measured after comparison with the

radiation thermometer along with the immersion depth in the liquid
bath. The terms E and Eref denote the measured and reference emf
at 180 C for each thermocouple, respectively.

our work, inhomogeneity scanning tests were performed using

a liquid bath operating at 180 C [16]. The thermocouples
were scanned before the fixed-point calibration tests, and
then after comparison with the radiation thermometer.
The inhomogeneity scan curves from before and after
the calibration and comparison were nearly the same.
Figure 6 shows the inhomogeneity scan results after the
comparison tests.
The uncertainty factor due to the thermoelectric
inhomogeneity was calculated by considering the temperature
gradient of the furnace and the inhomogeneity scan as a
function of the immersion depth, as follows [17, 18]:
inhomogeneity = (Et E25 C )
Emax (within gradient zone)
1

,
(1)
E180 C E25 C
0.95
where Et is the measured emf at the calibration temperature,
Emax is the maximum change in emf during an
inhomogeneity scan within the gradient zone as tabulated in
table 1, and E180 C and E25 C are the emfs at 180 C and 25 C,
respectively. Because the gradient zone of the furnaces was
estimated to generate 95% of the total emf, the values were
multiplied by a factor of 1/0.95 to calculate the inhomogeneity.
In table 6, uncertainty factors of the fixed-point and
comparison calibration at the freeezing temperature of Cu
for the type S thermocouple are listed. For the uncertainty
of the fixed-point calibration, uncertainty factors due to the
cell temperature and plateau determination were considered.
The method of uncertainty analysis used in the calculation
of the comparison calibration with the radiation thermometer
data was similar to that used for the fixed-point calibration,
except for determination of the reference temperature.
The uncertainty factor due to calibration of the radiation
thermometer LP4, alignment errors and short-term stability
was counted instead of using the fixed-point temperature
determination, as described in section 3.1. The inhomogeneity
was calculated differently because the comparison furnace
was different from the fixed-point calibration furnace.
The interpolation error at the comparison temperature was
estimated to be 0.1 V. The effect of the temperature
244

Standard uncertainty/ C

Uncertainty
factors

Fixed-point Comparison

Factors related to reference

temperature determination
(1) Cell temperature
(2) Plateau determination
(3) LP4 uncertainty
(4) LP4 alignment error
(5) LP4 stability
Factors related to emf measurement
(1) Inhomogeneity
(2) Reproducibility
(3) Ice-point realization
(4) DVM stability and uncertainty
(5) t/c short-term stability
Factors related to interpolation
(1) Interpolation
Expanded uncertainty (k = 2)

0.006

0.043

0.006
0.002

0.116
0.115
0.010
0.006
0.014
0.005
0

0.23

0.011
0.040
0.006
0.243
0.238
0.019
0.006
0.014

0.005
0.005
0.49

Table 7. Expanded uncertainty (in C) of the fixed-point calibration

with k = 2.
Fixed point temperatures/ C
Thermocouple

961.78

1084.62

1154.0

1324

1554.8

HPt/Pd
HSTC
HBTC

0.15
0.22
0.88

0.17
0.23
1.01

0.30
0.39
1.09

0.68
0.74
1.41

1.90

difference between the blackbody cavity and the thermocouple

well was negligible because the thickness of the wall that
separates the two was only 1 mm as shown in figure 3.
In table 7, the calculated expanded uncertainty for k = 2
is listed for the fixed-point calibrations. In the calculation of
the uncertainty, the majority of the uncertainty was caused by
the inhomogeneity of the thermocouple. The uncertainty of the
fixed-point calibration of the Pt/Pd thermocouple varied from
0.15 C to 0.68 C. In the case of the type S thermocouple,
the uncertainty values were slightly higher than those of the
Pt/Pd thermocouple. At the CoC melting temperature, the
uncertainty values were considerably increased because of the
large uncertainty of the CoC melting temperature. Since
the emf of the type B thermocouple at 180 C was small, the
small emf difference in the inhomogeneity scan curve induced
a large standard uncertainty if we assumed a linear temperature
dependence of the inhomogeneity [17]. This makes the
standard uncertainty of type B thermocouples much larger than
that of the other thermocouples because of the strong effect
of the inhomogeneity factor on the uncertainty. However, as
indicated in a previous report [18], some thermocouples, such
as types K and E, show a relative decrease in the inhomogeneity
with temperature. Therefore, the large uncertainty of the
type B thermocouple may be the result of an overestimated
inhomogeneity. To clarify this problem, more studies need to
be carried out on testing the inhomogeneity scans of type B
thermocouples at various temperatures.
Metrologia, 47 (2010) 239247

Validation of the fixed-point realization of the thermocouple scale

Table 8. Expanded uncertainty (in C) of the comparison calibration with k = 2.

Comparison temperature/ C
Thermocouple

961.78

1084.62

1154.0

1250

1324

1400

1450

1544

HPt/Pd
HSTC
HBTC

0.28
0.44
0.88

0.31
0.49
1.01

0.32
0.52
1.08

0.35
0.56
1.20

0.38
0.61
1.28

0.42
0.67

1.45

1.59

1100

1200

1300

1400

1500

1100

1200

1300

1400

1500

The expanded uncertainty of the comparison calibration

is listed in table 8 at each temperature. As the temperature
increased, the uncertainty increased gradually. Below the
freezing temperature of Cu, the uncertainty of the comparison
calibration was larger than that of the fixed-point calibration.
However, at a higher temperature of 1324 C, the uncertainty
was much smaller than that of the fixed-point calibration
because of the smaller calibration uncertainty of the reference
temperature determination.

-10
-20

(a) HPtPd_1

-30
-40
-50
-60
-70
900
-10
-20

4. Discussion

1000

(b) HPtPd_2

-30

Metrologia, 47 (2010) 239247

Deviation emf /V

-40

Figure 7 shows the results of calibration compared with

the radiation thermometer. The results of the fixed-point
calibration are also shown. The closed circles show the
comparison results with the radiation thermometer. The
squares denote the measurement data at the various fixed
points. The solid lines denote the least square interpolation
fits for the comparison results, showing very smooth behaviour.
The fit function was a second-order polynomial for the type S
and Pt/Pd thermocouples and a third-order polynomial for the
type B thermocouple. The error bars denote the expanded
calibration uncertainty for k = 2. Only half of the error bars
are plotted in figure 7 for clarity. The upper- and lower-half
bars denote the uncertainty of the fixed-point and comparison
calibrations, respectively. It was found that the Pt/Pd and type S
thermocouples had very similar behaviour with temperature,
and that there was a small difference in the emf values between
the two calibration methods in the temperature range 962 C to
1400 C. In the case of the type B thermocouple, the two scales
were nearly the same up to 1320 C, but then the discrepancy
increased with increasing temperature.
The temperature difference between the two scales
was calculated and is tabulated in table 9, which includes
the combined uncertainties. At the comparison points,
which were not fixed-point temperatures, the differences in
temperature and uncertainty were calculated by interpolation
or extrapolation. For the Pt/Pd thermocouples, the differences
were small over the entire comparison range (<0.2 C).
The type S thermocouple also showed a small temperature
difference of <0.1 C. In the case of the type B thermocouple,
the two scales agreed within 0.5 C up to a temperature of
1324 C, but the discrepancy increased to 2.29 C at 1554.8 C.
However, the temperature scale of a thermocouple using the
fixed-point calibration and the comparison with the radiation
thermometer were within the uncertainty over the entire
temperature range.
As shown in figure 7(d), the deviation function of
the type B thermocouple using the fixed-point calibration

-50
-60
-70
900
20
15

1000

(c) HSTC

10
5
0
-5
-10
900
1000
1100
1200
1300
1400
30
20 (d) HBTC
10
0
-10
-20
-30
-40
-50
900
1000 1100 1200 1300 1400 1500

1500

1600

Temperature / C
Figure 7. Deviation emf (circles with lower half of the uncertainty
bars) from the reference function measured by comparing with the
radiation thermometer, and interpolation fits (lines) for (a), (b) Pt/Pd
thermocouples, (c) a type S thermocouple and (d) a type B
thermocouple. The results from the fixed-point calibrations at Ag,
Cu, FeC, CoC and Pd (for type B only) are also shown (squares
with upper half of the uncertainty bars). For clarity, only half of the
error bars are plotted.

increased, but the deviation using the comparison calibration

decreased. If our Pd cell had been contaminated with
oxygen, which is the main contaminating species of Pd in
air, then the thermocouple emf should decrease [19, 20].
However, our results showed a much higher emf value than
the reference value, indicating that the deviation was not from
oxygen contamination. This apparent discrepancy may be
245

Y-G Kim et al

Table 9. Temperature difference between the fixed-point and the comparison calibration of thermocouples, and the combined expanded
uncertainties with k = 2 at each temperature.
(Temperature difference expanded uncertainty)/ C
Temperature/ C
961.78
1084.62
1154.0
1250
1324
1400
1450
1544
1554.8

HPtPd 1

HPtPd 2

0.09 0.33
0.08 0.35
0.03 0.44
0.10 0.62
0.18 0.78
0.21 0.95

0.01 0.33
0.02 0.35
0.06 0.44
0.04 0.62
0.11 0.78
0.17 0.95

related to the possible discrepancy in the reference emf of

the thermocouple at the Pd freezing temperature or to other
contamination of our Pd cell from an unknown source. Further
studies on the freezing and melting temperatures of Pd using
thermocouples are needed to clarify this phenomenon.

5. Conclusions
Using a blackbody comparison block system, the temperature
scale of thermocouples realized by fixed-point calibration was
validated using a radiation thermometer scale realized by
calibration at the freezing temperature of Cu. Two Pt/Pd
and one type S thermocouples were calibrated at the Ag, Cu,
FeC and CoC fixed points. A type B thermocouple was
also calibrated at the above fixed points and the Pd point.
To assign the melting temperature of the FeC eutectic cell
fabricated in this work, a radiometric FeC cell was prepared,
and its melting temperature was measured using an LP4
radiation thermometer. The melting temperature of the FeC
eutectic point was determined to be (1154.0 0.1) C for
k = 2. The calibrated thermocouples were compared with
the LP4 radiation thermometer in a horizontal furnace in the
temperature range 962 C to 1544 C. From 962 C to 1400 C,
the two temperature scales agreed within 0.2 C for the Pt/Pd
thermocouples and 0.1 C for the type S thermocouple. The
type B thermocouple showed agreement within 0.5 C up
to 1324 C, but the temperature difference between the two
scales increased with increasing temperature. At 1554.8 C,
the extrapolated discrepancy of the type B thermocouple was
calculated to be as high as 2.3 C. However, the discrepancy in
temperature was less than the measurement uncertainty over
the entire temperature range measured. Therefore, we confirm
that the temperature scale realized using the thermocouples
used in this work from 961.78 C to 1554.8 C agreed well with
the ITS-90 temperature realized using a radiation thermometer
in our laboratory. As a method of validation of thermocouple
calibration capability using fixed points including metal
carbon eutectics without an international comparison, it is
reasonable to compare the thermocouple scale with a radiation
thermometer, which is a standard interpolating instrument
above the freezing temperature of Ag. At KRISS, we have
used the radiation thermometer scale calibrated at the freezing
point of Cu to validate the thermocouple temperature scale
calibrated between the freezing points of Ag and Pd.
246

HSTC
0.03 0.50
0.07 0.54
0.06 0.65
0.04 0.82
0.06 0.96
0.08 1.12

HBTC
0.47 1.24
0.49 1.43
0.15 1.54
0.21 1.75
0.41 1.91

1.17 2.22
2.13 2.46
2.29 2.49

References
[1] Woolliams E R, Machin G, Lowe D H and Winkler R 2006
Metal (carbide)carbon eutectics for thermometry and
radiometry: a review of the first seven years Metrologia
43 R1125
[2] Kim Y G, Yang I, Kwon S Y and Gam K S 2006 Features of
Co/C and Ni/C eutectic transitions for use in thermocouple
thermometry Metrologia 43 6770
[3] Yang I, Kim Y G and Gam K S 2008 The dependence of the
melting temperatures of the eutectics FeC and CoC,
designed for thermocouple thermometry, on the furnace
offset temperature during the preceding freeze Meas. Sci.
Technol. 19 015106
[4] Edler E, Ederer P, Baratto A C and Vieira H D 2007 Melting
temperature of eutectic fixed point cells usable for the
calibration of contact thermometers Int. J. Thermophys.
28 198392
[5] Ogura H, Izuchi M and Arai M 2008 Evaluation of
cobaltcarbon and palladiumcarbon eutectic point cells for
thermocouple calibration Int. J. Thermophys. 29 21021
[6] Morice R, Megharfi M, Favreau J O, Morel E, Didialaoui I and
Filtz J R 2005 Realization of metalcarbon fixed points for
calibration of contact thermometers at high temperatures
Proc. TEMPMEKO 2004, 9th Int. Symp. on Temperature
and Thermal Measurements in Industry and Science
(Cavtat-Dubrovnik, Croatia) ed D Zvizdic (Zagreb, Croatia:
FSB/LPM) pp 84752
[7] Pearce J V, Ogura H, Izuchi M and Machin G 2009 Evaluation
of the PdC eutectic fixed point and the Pt/Pd thermocouple
Metrologia 46 4739
[8] Edler F, Morice R, Ogura H and Pearce J V 2010 Investigation
of CoC cells for improved thermocouple calibration
Metrologia 47 905
[9] Yamada Y, Sakate H, Sakuma F and Ono A 2001 High
temperature fixed points in the range 1150 C to 2500 C
using metalcarbon eutectics Metrologia 38 2139
[10] Sadli M, Bourson F, Fanjeaux M, Briaudeau S, Rougie B and
Bonnier G 2005 Study of metalcarbon eutectic points:
from construction to temperature determination Proc.
TEMPMEKO 2004, 9th Int. Symp. on Temperature and
Thermal Measurements in Industry and Science
(Cavtat-Dubrovnik, Croatia) ed D Zvizdic (Zagreb, Croatia:
FSB/LPM) pp 6116
[11] Bloembergen P, Yamada Y, Sasajima N, Wang Y and Wang T
2007 The effect of the eutectic structure and the residual
effect of impurities on the uncertainty in the eutectic
temperatures of FeC and CoC Metrologia
44 27993
[12] Kim Y G, Gam K S and Kang K H 1999 Realization of the
palladium freezing point for thermocouple calibrations
Metrologia 36 46572
Metrologia, 47 (2010) 239247

Validation of the fixed-point realization of the thermocouple scale

[13] Pearce J V, Lowe D H, Head D I and Machin G 2008

Optimizing contact thermometry high-temperature fixed
point cells (>1100 C) using finite-element analysis Int. J.
Thermophys. 29 25060
[14] ASTM Standard E 230-03 2003 Standard Specification and
Temperature-Electromotive Force (EMF) Tables for
Standardized Thermocouples (West Conshohocken, PA:
ASTM International)
[15] ASTM Standard E 1751-00 2000 Standard Guide for
Temperature Electromotive Force (Emf) Tables for
Non-letter Designated Thermocouple Combinations (West
Conshohocken, PA: ASTM International)
[16] Kim Y G, Yang I and Gam K S 2008 Measurement of the
inhomogeneity of Pt/Pd thermocouples using an
immersion-into-liquid method Instrum. Sci. Technol.
36 25766

Metrologia, 47 (2010) 239247

[17] Jahan F and Ballico M 2003 A study on the temperature

dependence of inhomogeneity in platinum-based
thermocouple Temperature: Its Measurement and Control
in Science and Industry vol 7 ed D C Ripple (New York:
AIP) pp 46974
[18] Kim Y G, Song C H, Gam K S and Yang I 2009 Change in
inhomogeneity with temperature between 180 C to 950 C
in base-metal thermocouples Meas. Sci. Technol. 20 075102
[19] Jones T P and Hall K G 1979 The melting point of palladium
and its dependence on oxygen Metrologia 15 1613
[20] Jahan F and Ballico M 2003 The mini-coil method for
calibration of thermocouples at the palladium point
Temperature: Its Measurement and Control in Science and
Industry vol 7 ed D C Ripple (New York: AIP) pp 5238

247