1 How Does Temperature Non-uniformity of an Annealing Furnace Affect SPRT Stability? Speaker: Mingjian Zhao Authors: Mingjian Zhao, Rose Heaton, Rong Ding, Deming Chen, and Matthew Newman Fluke Hart Scientific 799 E Utah Valley Drive American Fork, UT

2 Introduction (1) In recent years, more temperature calibration laboratories have begun using SPRTs as their reference standards to improve calibration uncertainties. To achieve the highest level of performance, a standard platinum resistance thermometer (SPRT) must be handled and maintained correctly. Proper annealing is necessary to maintain SPRT accuracy and stability.

3 Introduction (2) However, many laboratories cannot afford a proper annealing furnace to anneal their SPRTs. Some laboratories use alternative equipment such as drywell calibrators. Temperature uniformity of “annealing furnaces” sometimes does not meet the NMIs’ suggestion. Therefore, it is important to understand how temperature non-uniformity affects SPRT stability and what equipment can be used for SPRT annealing.

4 Reasons to anneal an SPRT (1) Vibration, shock or other form of acceleration during ship or routine use will produce strains in SPRT sensor. The strains change SPRT Rtpw value and t-R characteristics. A state with strains is an unstable one that trends returning to strain-free state. Proper annealing makes SPRTs returning to strain-free state, the most stable state of SPRTs.

5 Reasons to anneal an SPRT (2) Oxidation of platinum sensor wire could make the resistance increase, and proper annealing could reduce the affect of oxidation. The SPRT should be annealed after it is operated at high temperature. The SPRT should be checked in stability through short-term annealing before calibration. The SPRT will be calibrated as soon as the R(TPW) of the SPRT shows stable.

6 Control Chart Showing Annealing Correction SPRT annealed

7 General Request on Annealing Furnace To avoid the contamination, the SPRT and HTSPRT should not be annealed in a metal block furnace above 660°C The annealing furnace should have “good” vertical temperature gradient. The display accuracy should be checked and corrected every six months

8 SPRT Contaminations Introduced from the Annealing Furnace If an SPRT be contaminated, the R(tp) would increase, and the W(Ga) would decrease. W(Ga) might not meet the requirement of ITS-90 anymore

9 Remarks on SPRT Contamination (1) Metal Blocks with different materials have no effect on the stability of SPRTs up to 660°C. Base metals cannot diffuse through fused silica to contaminate platinum up to about 660°C. SPRTs can be annealed in any block up to 660°C. Pure graphite and alumina will not contaminate platinum up at least to 1100°C. It is suggested to used graphite, or alumina blocks to anneal SPRTs above 660°C.

10 Remarks on SPRT Contamination (2) The contamination of base metals to platinum through fused silica begins to happen at about 800°C. The drift rates of Rtp rise sharply above about 970°C. The drift rates of Rtp depend on the diameter of platinum wire. The 0.4 mm Pt wire sensor have drift rates about 60% of that for the 0.2 mm Pt wire sensor.

11 Experiment In order to determine how temperature non-uniformity of the annealing furnace affects SPRT stability, a series of investigations was recently performed at Fluke-Hart Scientific. Three-zone furnaces that allow the vertical temperature gradient to be adjusted were used as annealing furnaces. Three stable SPRTs were annealed in the furnaces with different temperature profiles. Resistances of the SPRTs at the triple point of water (Rtpw) were measured and compared before and after annealing.

12 Annealing Furnaces The three furnaces had the same design and construction. –99.8% pure alumina equilibrium block –Five cleaned fused-quartz re-entrant wells for SPRTs –High-purity, specially cleaned fused-quartz containers –Alumina disks and fiber ceramic disks were placed for insulation. Previous research confirms that a fused-quartz sheath SPRT is not contaminated by an alumina block. The three annealing furnaces were adjusted to have different vertical temperature profiles through three-zone control.

13 Temperature uniformity of the Annealing Furnaces ±8.1°C ±3.5°C ±0.9°C

14 SPRTs The SPRTs were specially designed and assembled for this study. Their sensing elements were longer than normal, 55 mm compared to the usual 45 mm. It was assumed that the longer element would be affected more by temperature non-uniformity. The three SPRTs were stabilized through an annealing process after manufacturing. The long-term stabilities of all three SPRT at the triple point of water were better than 1 mK after operation at 675°C for a few hundred hours.

15 Testing Procedure Anneal the SPRTs at 675°C for 15 hours in Furnace #1, cool down to 480°C at a scan rate of 1.9°C/min, remove from the furnace, and measure Rtpw. Repeat four times. Anneal the SPRTs at 675°C for 15 hours in Furnace #2,......, repeat four times. Anneal the SPRTs at 675°C for 15 hours in Furnace #1, ……, repeat four times. Anneal the SPRTs at 675°C for 15 hours in Furnace #3, ……, repeat four times. Anneal the SPRTs at 675°C for 15 hours in Furnace #1, ……, repeat four times.

16 Stability of SPRT S/N annealed in different furnaces

17 Stability of SPRT S/N annealed in different furnaces

18 Remarks Vibration, shock, and temperature cycling during handling and use cause strain in an SPRT, which changes its resistance. Proper annealing removes strain and oxidation effects and is required to maintain accuracy and stability. Having an annealing furnace is not an option but a necessity for temperature calibration laboratories that own SPRTs.

19 Conclusions Based on the investigation, temperature non-uniformity of the annealing furnace does not seem to affect SPRT stability significantly (around 1 mK). However, it is too early to conclude that an annealing furnace with large temperature non-uniformity is adequate for a primary standard laboratory or a national laboratory. Perhaps a lower cost furnace, even a dry-well calibrator or a Metrology Calibrator, is sufficient for a secondary temperature calibration laboratory. In any case, to avoid contamination, a fused-quartz sheath SPRT or HTSPRT should not be annealed in a metal block furnace above 660°C.

20 Further Research Needed The stability of SPRTs with extra long element should be improved Whether SPRTs that have drifted can be annealed just as well in a furnace that has large temperature non- uniformity.

21 Annealing Procedures for 25.5-ohm SPRTs Annealing temperature: 665°C or the maximum temperature of the SPRT Annealing time: 2 hours at 665°C, or 4 hours at 480°C. Ramp down to 480°C at a rate about 1.5° per min. Remove SPRTs out of furnace at 480 °C and cool them in the air. Measure Rtp before and after annealing. The changes of R(tpw) should be within 0.5 mK (or other special requirement). If not, repeat the process until the R(tpw) changes within the requirement.

22 Annealing Procedures for 0.25-ohm or 2.5-ohm SPRTs Annealing temperature: 965°C Annealing time: 2 hours Ramp down to 480°C at a rate about 1.5° per min. Remove SPRTs out of furnace at 480 °C and cool them in the air. Measure Rtp before and after annealing. The change of the R(tpw) should be within 0.7 mK (or other special requirement). If not, repeat the process until the R(tpw) changes within the requirement.

23 Precaution during the Annealing Process Annealing furnace used for high temperature should be clean without base metal in its space in order to protect platinum elements from metallic contamination at high temperature. Clean SPRT quartz glass sheath with pure ethanol and dry it with clean tissue in order to protect the sheath from devitrification.

24 Thank You !