Professional Documents
Culture Documents
Thermal Hysteresis in Thin-Film Platinum Resistance Thermometers
Thermal Hysteresis in Thin-Film Platinum Resistance Thermometers
net/publication/226112701
CITATIONS READS
10 344
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Kee Sool Gam on 02 June 2014.
International Journal of
Thermophysics
Journal of Thermophysical Properties
and Thermophysics and Its Applications
ISSN 0195-928X
Volume 32
Combined 11-12
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media, LLC. This e-offprint
is for personal use only and shall not be self-
archived in electronic repositories. If you
wish to self-archive your work, please use the
accepted author’s version for posting to your
own website or your institution’s repository.
You may further deposit the accepted author’s
version on a funder’s repository at a funder’s
request, provided it is not made publicly
available until 12 months after publication.
1 23
Author's personal copy
Int J Thermophys (2011) 32:2388–2396
DOI 10.1007/s10765-011-1041-8
Received: 10 May 2010 / Accepted: 8 July 2011 / Published online: 30 July 2011
© Springer Science+Business Media, LLC 2011
1 Introduction
The use of platinum resistance thermometers (PRTs) for industrial applications has
increased markedly in the last 20 years. The main reasons for this have been the
extension of the applications to new fields and the need for improved temperature
123
Author's personal copy
Int J Thermophys (2011) 32:2388–2396 2389
measurement accuracy. Industrial PRTs (IPRTs) can be divided into three types:
ceramic, glass, and thin films. Ceramic and glass PRTs are made using high-purity
platinum coils, while thin-film PRTs are generally manufactured using the deposition
of a thin platinum film upon an alumina substrate and a laser-trimming method. Con-
sidering the strong adhesion between the thin platinum film and the alumina substrate,
thin-film PRTs inevitably have strain over the operating temperature range, which
causes anomalies and instabilities in the resistance versus temperature (R–T ) charac-
teristics. The most prominent and observable effect of thermally induced strain is the
thermal hysteresis in the R–T characteristics. Thermal hysteresis is one of the main
uncertainty factors in the calibration of industrial platinum resistance thermometers
in laboratories [1,2]. Thermal hysteresis for ceramic and thin-film PRTs was investi-
gated extensively by Curtis in 1982 [3], and the test method is well described in ASTM
standards [4]. Since that study, the manufacturing technology for thin-film PRTs has
improved greatly and there are very little data for the thermal hysteresis of modern
thin-film PRTs. In this study, the thermal hysteresis for 30 industrial PRTs supplied
from two companies was measured and the effect of cycling on the hysteresis was
studied.
2 Experimental
The thin-film PRT sensor elements were supplied by IST, Switzerland and Omega,
USA. The sensor specifications are listed in Table 1. The sensor named “F2020”
marked as 1/3B in Table 1 is known to have better tolerances than the class A, which
has interchangeability of 0.1 ◦ C at the ice point [5].
The thin-film PRT sensors were encapsulated in seamless stainless-steel sheaths
with a diameter of 6.4 mm and a length of 45 cm. In all cases, four lead wires were
welded with Ag–Pb eutectic solder. Nickel wire, which was 99.99 % pure with a diam-
eter of 0.4 mm, was used as the lead wires. The sheaths were filled with high-purity
Maker Sensor name Tolerance class Range (◦ C) Quantity Size (mm) (L × W × Th)
123
Author's personal copy
2390 Int J Thermophys (2011) 32:2388–2396
alumina powder to insulate electrically the four lead wires and the sheath. Three man-
ufacturers fabricated 10 PRT probes each using the thin-film PRT sensors given in
Table 1. In total, 30 PRT probes were prepared for the hysteresis measurements.
Liquid baths are generally used for hysteresis measurements; each liquid has a limited
temperature range. This means that it may be necessary to move the test PRTs from one
bath to another bath very quickly to preserve their thermal history. This transfer process
could also change the resistance of PRTs because of thermal and mechanical shock.
To minimize these problems, we used a movable furnace instead of liquid baths. The
three-zone furnace (Isotech 511 Medusa-3, England), which has a maximum operating
temperature of 700 ◦ C, was used in the hysteresis measurements. An Inconel block
with six thermometer holes was fitted inside the furnace to improve its stability. Insu-
lation between the Inconel block and the upper surface of the furnace was provided by
several Inconel plates and fiberfrax. The stability of the furnace was within 10 mK at
400 ◦ C for 2 h, and the temperature differences between thermometer holes were about
1 mK. This furnace was interfaced to a PC using RS-232C communication ports, and
the LabVIEW program controlled the temperature. A standard platinum resistance
thermometer (SPRT, Hart Scientific Model 5699, S/n 0246) was used as the reference
thermometer for measuring the furnace temperature. This SPRT was calibrated from
the triple point of water to the freezing point of aluminum by the calibration proce-
dures of the KRISS temperature laboratory. The resistance of the SPRT was measured
using an AC bridge (ASL F900), and the furnace temperature was calculated by the
reference function and the deviation function given by the ITS-90. The resistance of
the test PRTs was measured using a Super Thermometer (Hart Scientific 1575) and
switching unit that was interfaced to the computer. All measurement instruments and
the three-zone furnace were interfaced to the PC for data acquisition and temperature
control.
The thermal hysteresis of 30 industrial PRT probes fabricated by three Korean compa-
nies was measured. The measurements were performed with five PRT probes in each
hysteresis measurement. The SPRT and PRT probes were installed in the holes of the
Inconel block of the furnace. Then the furnace temperature was increased from room
temperature to 500 ◦ C in 100 ◦ C steps and then decreased using the same steps. Resis-
tance data for the SPRT and the thin-film PRTs were gathered automatically after
the furnace temperature was confirmed stable within 5 mK at each step. Figure 1a
and b shows the hysteresis measurement step and typical hysteresis data for a set of
five PRTs measured at 400 ◦ C. Hysteresis data were calculated from the resistance
difference between the resistance values of the PRTs measured as the temperature
increased and those measured as the temperature decreased. The temperature differ-
ences were corrected using temperatures calculated using the calibration report of the
123
Author's personal copy
Int J Thermophys (2011) 32:2388–2396 2391
(a) 300
280 500 oC
260
400 oC
IPRT Resistance, 240
220
300 oC
200
180
200 oC
160
140 100 oC
120
0 500 1000 1500 2000 2500
Time, min
(b) 252
251
IPRT Resistance,
250
249
248
247
246
80 100 120 140 160 180 200
Time, min
Fig. 1 (a) Hysteresis measurement steps for IPRTs and (b) typical hysteresis data for 5 IRPTs measured
at 400 ◦ C
SPRT. Hysteresis measurement of PRTs was completed over about 2 days and the
stabilizing time at each step was 2 h to 5 h.
3 Results
Figure 2 shows a typical hysteresis curve for the thin-film PRT sensor F2020-2.
The magnitude of the hysteresis was 0.14 ◦ C for a 100 ◦ C span and decreased to
0.03 ◦ C for a 400 ◦ C span. We obtained a different hysteresis curve for the thin-film
PRT sensor F3105. The hysteresis curves show that the magnitude of the hystere-
sis was not proportional to the temperature spans, and the largest differences in the
hysteresis curves were independent of the temperature spans. This result was a little
123
Author's personal copy
2392 Int J Thermophys (2011) 32:2388–2396
0.3
0.2
0.1
Δt, oC decrease
0.0
-0.1
increase
-0.2
-0.3
0 100 200 300 400 500
Temperature, oC
Fig. 2 Typical hysteresis curve of the thin-film PRT sensor F2020-2
different from Curtis’ results, in which the magnitude of hysteresis was proportional
to the temperature span and the thermal hysteresis was very reproducible [1]. Cur-
tis tested many PRT element designs, including wire-wound industrial grade PRTs
in which platinum wire was encapsulated in ceramic or glass and thick-film PRTs.
The hysteresis of the platinum wire of the wire-wound PRT sensor was caused by the
differences in the thermal expansion coefficient between the platinum wire and the
ceramic or glass supports. Thermal expansion and contraction are reversible processes
in the wire-wound PRTs. However, the hysteresis mechanism of the thin-film PRTs is
difficult to understand and is independent of the thermal span because of its structure
and the coatings that protect the platinum film.
Hysteresis measurements were repeated three times with the temperature cycled
in the same way. The hysteresis data were calculated from the average of the mea-
surement results executed continuously three times. The hysteresis results for the 30
thin-film PRTs are summarized in Table 2. The thermal hysteresis for the 30 sensors
was distributed from 16 mK to 156 mK. The highest hysteresis was 77 mK to 156 mK
for the sensor F3105 (probe made by company B) and the lowest was 8 mK to 16 mK
for the sensor F3102 (probe made by company C) in the temperature range tested. The
hysteresis results of sensors from the three companies showed similar results: the hys-
teresis was distributed from 31 mK to 87 mK for company A, from 25 mK to 156 mK
for company B, and from 16 mK to 97 mK for company C. In particular, all probes
made by the three companies using the F3102 sensors showed very good hysteresis
characteristics: the values were distributed from 16 mK to 31 mK in the temperature
range tested.
During the hysteresis measurements, we found that the hysteresis values decreased
drastically as the measurement cycles preceded. Figures 4 and 5 are hysteresis curves
123
Author's personal copy
Int J Thermophys (2011) 32:2388–2396 2393
Company A
F2020-1 74 87 30 25
F2020-2 48 81 34 28
F4050 69 70 60 45
F3102 16 31 13 20
F3105 76 15 10 73
8W 82 82 74 13
7W 69 69 51 10
6W-1 74 74 40 9
6W-2 43 43 50 9
6W-3 65 65 50 12
Company B
F2020-1 25 25 24 14
F2020-2 32 32 62 63
F4050 55 55 49 69
F3102 23 23 30 18
F3105 102 102 156 77
8W 117 85 29 36
7W 44 78 20 10
6W-1 13 70 72 60
6W-2 60 74 69 39
6W-3 102 87 46 26
Company C
F2020-1 40 32 30 54
F2020-2 91 48 34 40
F4050 79 50 46 29
F3102 13 16 8 10
F3105 37 54 53 62
8W 34 65 41 37
7W 28 41 33 17
6W-1 52 16 97 84
6W-2 13 69 55 35
6W-3 22 56 39 26
for the sensor F2020-2 and the sensor F3105 obtained after the hysteresis test had been
executed for six cycles. The hysteresis of the sensor F2020-2 decreased from 81 mK to
11 mK and that of the sensor F3105 decreased from 76 mK to 38 mK. Figure 6 shows
the variation of the hysteresis values of the sensor F2020 measured after the hyster-
esis test had cycled six times. The hysteresis values reached the same values within
123
Author's personal copy
2394 Int J Thermophys (2011) 32:2388–2396
0.5
0.4
0.3
0.2
Δt, C 0.1
decrease
o
0.0
-0.1 increase
-0.2
-0.3
-0.4
-0.5
0 100 200 300 400 500
Temperature, oC
Fig. 3 Typical hysteresis curve of the thin-film PRT sensor F3105
0.020
0.015
0.010
0.005 decrease
Δt, oC
0.000
-0.005
-0.010
increase
-0.015
-0.020
0 100 200 300 400 500
o
Temperature, C
Fig. 4 Hysteresis curve in the sixth measurement cycle for the sensor F2020-2
123
Author's personal copy
Int J Thermophys (2011) 32:2388–2396 2395
0.06
0.04
0.02
decrease
Δt, oC
0.00
increase
-0.02
-0.04
-0.06
0 100 200 300 400 500
o
Temperature, C
Fig. 5 Hysteresis curve in the sixth measurement cycle for the sensor F3105
0.20
o
100 C
o
200 C
0.15 o
300 C
o
400 C
Δt, oC
0.10
0.05
0.00
1 2 3 4 5 6
Run number
Fig. 6 Variation of the hysteresis values of the sensor F2020 measured over six cycles
the uncertainty arising from the thermometer and measurement instruments. The SPRT
calibration uncertainty was 2 mK with k = 2, and the uncertainty of an AC bridge
(ASL F900) was calculated as 0.2 mK. The uncertainty of the Super Thermometer
(Hart Scientific 1575) used to measure the resistances of IPRTs was 2 mK. The third
factor, obtained from the measurement scatter of the resistance of IPRTs, was evaluated
as 5 mK. The combined uncertainty of the hysteresis measurements was 16 mK.
4 Conclusions
The thermal hysteresis of thin-film PRTs was measured using a movable three-zone
furnace in the range of 0 ◦ C to 500 ◦ C in 100 ◦ C steps. The thermal hysteresis was
distributed from 16 mK to 156 mK for all sensors. We found that the hysteresis values
123
Author's personal copy
2396 Int J Thermophys (2011) 32:2388–2396
Table 3 Hysteresis for thin-film PRT sensors measured after six cycles
F2020-1 8 2 11 16
F2020-2 11 3 4 3
F2020-1 (Company B) 53 34 19 3
F3105 10 38 7 1
8W 13 29 20 23
7W 12 17 13 6
6W-1 11 5 16 4
6W-2 3 1 5 11
6W-3 18 9 18 44
decreased drastically with repeated measurements. The cycling effect was also inves-
tigated for several sensors. After six cycles, the hysteresis values of the test sensors
decreased below 53 mK and the lowest value was 11 mK.
References
1. H.M. Hashemian, K.M. Petersen, in Temperature, Its Measurement and Control in Science and Industry,
vol. 6, ed. by J.F. Schooley (AIP, New York, 1992), pp. 427–432
2. B.W. Mangum, G.A. Evans, Jr., in Temperature, Its Measurement and Control in Science and Industry,
vol. 5, ed. by J.F. Schooley (AIP, New York, 1982), pp. 795–801
3. D.J. Curtis, in Temperature, Its Measurement and Control in Science and Industry, vol. 5, ed. by J.F.
Schooley (AIP, New York, 1982), pp. 803–812
4. Historical Standard, ASTM E644-02 Standard Test Methods for Testing Industrial Resistance Ther-
mometers (ASTM International, West Conshohocken, PA, 2002), pp. 10–11
5. http://www.omega.com/ppt/pptsc.asp?ref=F1500_F2000_F4000&Nav=temc13, Omega Engineering,
Inc., Stamford, CT (2010)
123
View publication stats