Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
DOI 10.1007/s10765-011-0944-8
F. Jahan M. J. Ballico
Received: 6 April 2010 / Accepted: 3 February 2011 / Published online: 2 March 2011
Springer Science+Business Media, LLC 2011
123
362 Int J Thermophys (2011) 32:361371
1 Introduction
Traditionally, the fixed-point cells used to define the ITS-90 temperature scale were
regarded as primary reference standards, as long as material of sufficient purity was
used in their construction and the quality of the phase-transition plateaus was suffi-
ciently good. The typical calibration uncertainty for NMIAs primary radiation ther-
mometers is at the 0.1 C level, so uncertainties in the primary fixed points at the
50 mK level are adequate. As metal at the 1 ppm impurity level is readily available
for constructing fixed points, and this corresponds to uncertainties at the <1 mK level,
issues of source material purity may be neglected. However, fixed points for radiation
thermometry differ from those used to realize the ITS-90 using SPRTs in several ways,
giving rise to a number of furnace-effects as follows.
(i) The graphite well is usually much shorter than those used for SPRTs. In the
latter, the well must typically be immersed 170 mm to 220 mm into the metal, to
reduce conduction errors in the SPRT to a sufficiently small level. However, for
pyrometry fixed points, a depth of less than 100 mm can achieve a suitably high
cavity emissivity, and longer cavities lead to viewing an increasing proportion
of the cavity wall (due to the finite F-number of the pyrometer). A consequence
is that during melting/freezing, when the outer surface of the crucible is a few
degrees hotter/colder than the liquidsolid interface, a larger fraction of this
temperature difference may be conducted to the cavity base.
(ii) The crucible is usually mounted horizontally in a furnace rather than vertically,
and as a consequence, the heat transfer to the melting or freezing metal is far
from cylindrically symmetric. For example, the crucible has much better ther-
mal contact with the furnace on its underside. Consequently, the liquidsolid
interface may not properly surround the blackbody cavity.
(iii) The low F-number of the radiation thermometers used to view the blackbody
cavity does not allow the crucible to be placed deeply inside a long furnace.
Consequently, the axial temperature gradients are generally much larger, and
the liquidsolid interface often advances axially rather than radially as for SPRT
fixed points.
(iv) Unlike fixed points for SPRT use, no additional inner nucleation is performed
to provide a second solidliquid interface, over the inner well. Consequently,
any breaks in the solidliquid interface surrounding the thermowell will provide
a heat leakage from the well to the furnace.
At present, any systematic errors arising from furnace-effect factors are assessed in
two ways: (a) first, the use of redundant fixed points (e.g., fitting the SakumaHattori
interpolation equation [1] using Zn, Al, and Au points and checking the measured
temperature at the Ag point) and (b) secondly, examination of the meltfreeze dif-
ferences and the flatness of the freeze plateau. Our experience is that the uncertainty
we assign to the radiation thermometry fixed points is dominated by the measured
meltfreeze differences and plateau flatness. However, as part of ongoing work to
improve the quality assurance of NMIAs calibration services, we wish to develop a
formal traceability of radiation thermometry fixed points to NMIA primary standards
validated by participation in a CIPM key comparison. This would confirm that the
123
Int J Thermophys (2011) 32:361371 363
2 Experimental Details
123
364 Int J Thermophys (2011) 32:361371
85 mm
43 mm
30 mm
95 mm
point, Ga, Sn, Zn, Al, and Ag. A 7 mm OD sealed-end quartz tube is used to sepa-
rate the thermocouple from the fixed-point cells and prevent cross contamination. The
reference junction was immersed 180 mm into a crushed ice slurry in a 30 cm deep
Dewar. Electrical measurements were performed using an HP34420A nanovoltmeter.
The series of graphite rings of increasing diameter placed at the front of the cruci-
ble for radiation thermometry provide very poor thermal contact (a 10 mm to 15 mm
air gap) between the furnace tube and the thermocouple, and inadequate immersion
could be achieved with this arrangement. Instead, several thin carbon disks separated
by 10 mm long quartz tube spacers were used in the region from the front of the
crucible to the front of the furnace. This significantly increased the thermal contact
123
Int J Thermophys (2011) 32:361371 365
from the furnace to the guide tube, without significant disruption to the temperature
environment near the crucible. This is supported by the observation that the furnace
set points are unchanged, and the melt- and freeze-plateaus have the same duration.
A 3 mm thick carbon sleeve was used in the Sn and Zn cells to improve the thermal
contact between the 7 mm quartz thermocouple tube and the 14 mm blackbody cavity.
For each of the NMIAs radiation thermometry fixed-point cells, several calibrated
Au/Pt thermocouples were used to realize several meltfreeze plateaus. Typical results
in each of the fixed point cells are shown in Figs. 2, 3, 4, 5, and 6. The average results
for each thermocouple are presented in Table 1 and compared to the values obtained in
NMIAs reference cells (only freeze values are considered). There is excellent agree-
ment between the cell temperature difference results obtained by the two (three for
Al) thermocouples.
While the measured values of the freezing plateaus of the radiation thermometry
cells adequately serve the purpose of providing formal traceability of these cells to
the NMIA reference cells, a more detailed examination of the curves provided some
important additional information about the freezing and melting processes.
When the radiation thermometry Zn cell is used to calibrate radiation thermometers,
a meltfreeze difference of about 20 mK is observed, and prior to this measurement, it
4952
4945
0.12 C
30 mK
1 cm out
4950 4944
4943
4948
EMF , V
4942
0 15 30 45 60
4946
Reference Zn cell
4944
4942
0 50 100 150 200 250 300
Time , min
Fig. 2 Freezing plateau of a radiation-thermometry Zn point realized by Au/Pt thermocouple. A conduction
error test is shown in the inset. The EMF value obtained in the reference Zn cell is also shown
123
366 Int J Thermophys (2011) 32:361371
2.569
60 mK
full immersion
2.568
Reference Zn cell
Resistance ratio
2.567
10 mm out
20 mm out
30 mm out
2.566
2.565
60 80 100 120 140
Time , min
Fig. 3 Freezing plateau of a radiation thermometry Zn point realized by an SPRT, showing a conduction
error test. The resistance ratio obtained in the reference Zn cell is also shown
was unclear how much systematic error in the actual Zn-point temperature this implied.
The Au/Pt thermocouple data in Fig. 2 also show a meltfreeze difference (0.1 C),
although larger than that obtained with radiation thermometry. The conduction-error
test, in which the thermocouple is withdrawn by 10 mm, shows a change of only about
15 mK, confirming that the meltfreeze difference is a real effect. The thermocou-
ple data show that the actual cell temperature is within the range of the meltfreeze
data, confirming that this effect arises from the conduction effects within the cell, and
that the meltfreeze range provides a good estimate of the associated uncertainty. For
comparison, a freezing plateau of the same Zn cell realized by an SPRT in this same
assembly is shown in Fig. 3. With the SPRT withdrawn by 20 mm, the conduction
errors are significant, about 0.12 C, consistent with the measured freeze value at full
immersion being apparently 60 mK lower than the reference cell; as expected, the
40 mm long SPRT sensor is more sensitive to conduction from the furnace.
The thermocouple results obtained using the Sn cell (Fig. 4) show a nearly 0.1 C
difference between the melt and the freeze curves, and indicate that both the melt
and the freeze are biased towards the furnace temperature outside the cell (the melt is
higher and the freeze is lower). When the cell is viewed by a radiation thermom-
eter a similar, but slightly smaller difference of 40 mK to 70 mK is observed. These
thermocouple results support the use of the meltfreeze difference as an uncertainty
estimate.
123
Int J Thermophys (2011) 32:361371 367
2242
2240
150 mK
2238
EMF , V
2236
Reference Sn cell
2234
1 cm out
2232
2230
0 50 100 150 200 250 300 350
Time, min
Fig. 4 Typical meltfreeze curve of a radiation-thermometry Sn point realized by a Au/Pt thermocouple.
The EMF value obtained in the reference cell is also shown
123
368 Int J Thermophys (2011) 32:361371
9323
50 mK
9322
9321
EMF , V
9320
9318
9317
0 50 100 150 200 280 300 320
Time , min
Fig. 5 Typical meltfreeze curve of radiation thermometry Al point realized by a Au/Pt thermocouple
showing a conduction error test. The EMF value obtained in the reference Al cell is shown
determined for each cell by withdrawing the thermocouple by 10 mm, and noting the
change in the measured EMF. For these thermocouples we have determined [8] that
the conduction error roughly doubles for each 8 mm to 10 mm of immersion in a close
fitting tube, so we use this value as the uncertainty contribution for this term. The
conduction error in the Sn and Zn cells is significantly higher than for the Al and Ag
cells, and this is attributed to the larger well diameter for these cells, and the poorer
radiative heat transfer at low temperatures.
Other minor contributions to the uncertainty arise from: the choice of position on
the freezing plateau, which we take as the width of the central half of the freezing
plateau; the estimated 5 mK uncertainty in the crushed-ice-point reference junction;
the reproducibility of the plateau between realizations; and the calibration and drift of
the 34420A nanovoltmeter used to measure the thermocouple EMF.
Considering the estimated measurement uncertainties, the measured temperatures
of the freezing plateaus of the radiation thermometry cells can be considered to be
consistent with those of the NMIA reference cells [9].
4 Conclusions
We have demonstrated that the temperatures realized by the small fixed-point cell
assemblies used for radiation thermometry can be measured traceably to ITS-90 fixed
123
Int J Thermophys (2011) 32:361371 369
16121.5
20 mK
16121.0
16120.5
Reference Ag cell
EMF , V
16120.0
16119.5
1 cm withdrawn
16119.0
16118.5
0 30 60 90 120 150
Time , min
Fig. 6 Typical meltfreeze curve of radiation thermometry Ag point realized by a Au/Pt thermocouple
shows conduction error test during freeze. The EMF value obtained in the reference Ag cell is also shown
Table 1 Summary of the average EMF for several Au/Pt thermocouples in the freezing plateaus of the
NMIA reference ITS-90 fixed points and the radiation thermometry fixed points
Reference cell Radiation TC serial Average EMF (V) measured TDUT TITS90
thermometry cell number at the fixed points (mK)
NMI Radiation
reference thermometry
cell cell
points using simple and robust Au/Pt thermocouples. The method allows the in situ
measurement of the meltfreeze plateaus of the radiation thermometry fixed-point
cells, using the same furnaces with which they are normally used. The uncertainty
123
370 Int J Thermophys (2011) 32:361371
-0.2
-0.4
16 mK
-0.6
-0.8
E , V
-1.0
-1.2
-1.4
TC:PtAu-0805
-1.6 TC:PtAu-0203
TC:PtAu-0804
-1.8
100 200 300 400 500 600
Immersion , mm
Fig. 7 Thermoelectric scans of three Au/Pt thermocouples at 200 C indicating their level of homogeneity
Table 2 Typical values of the uncertainty components for the calibration of radiation thermometry fixed
points using an Au/Pt thermocouple
Ag Al Zn Sn
123
Int J Thermophys (2011) 32:361371 371
Acknowledgment We would like to acknowledge Mr. Steve Meszaros for constructing the parts of the
assembly and general assistance during measurement.
References
1. F. Sakuma, S. Hattori, in Temperature, Its Measurement and Control in Science and Industry, vol. 5,
part 1, ed. by J.F. Schooley (AIP, New York, 1982), pp. 421427
2. T.P. Jones, J. Tapping, Metrologia 18, 23 (1982)
3. M.J. Ballico, NMI Quality System PMEADA8.2.12 (National Measurement Institute of Australia,
Lindfield)
4. F. Jahan, M.J. Ballico, Int. J. Thermophys. 28, 1822 (2007)
5. F. Jahan, M.J. Ballico, in Proceedings of 6th Biennial Conference of MSA (Metrology Society of Aus-
tralia, Canberra, 2005), pp. 4853
6. F. Jahan, NMI Quality System PM-EADA-8.2.30 (National Measurement Institute of Australia, Lindfield)
7. F. Jahan, M.J. Ballico, in Temperature: Its Measurement and Control in Science and Industry, vol. 7,
part 1, ed. by D.C. Ripple (AIP, New York, 2002), pp. 469474
8. M.J. Ballico, in Proceedings of TEMPMEKO 2004, 9th International Symposium on Temperature and
Thermal Measurements in Industry and Science, ed. by D. Zvizdic, L.G. Bermanec, T. Veliki, T. Staic
(FSB/LPM, Zagreb, 2004), pp. 801806
9. M.J. Ballico, K. Nguyen , Int. J. Thermophys. 30, 284 (2007)
123