Why Use an Annealing Furnace?

by
Rose Heaton, Electrical Engineer
Xumo Li, Director of Metrology
Hart Scientific, Inc.

The International Temperature Scale of 1990 (ITS-90) expanded the range of the Standard Platinum Resistance Thermometer (SPRT). This expansion presented new challenges to the thermometer industry. Among these challenges are the care and maintenance of the SPRT and the High Temperature Standard Platinum Resistance Thermometer (HTSPRT). With correct care and maintenance, an SPRT can provide years of accurate, stable measurements. Without proper care and maintenance, an SPRT can be damaged beyond repair or recalibration in a very short period of time. In fact, careless day-to-day handling of an SPRT over a one-year period has been observed to increase its resistance at the triple point of water (RTPW) by an amount equivalent to 0.1 K. This could result in temperature errors as high as 10 mK without recalibration.¹

In the past, many engineers have not understood the need for an annealing furnace. Today more people are coming to understand the importance of the annealing furnace in the care and maintenance of their SPRTs as they become more comfortable with the ITS-90. Most national laboratories have at least one annealing furnace and many have several. In the early 1980s, the Chinese developed HTSPRTs, which were recognized throughout the world for their excellent stability. A carefully designed, non-contaminating annealing furnace was an important factor contributing to the remarkable stability of these HTSPRTs. Today Hart has incorporated similar technology in our 9117 Annealing Furnace. (For the purpose of this paper, both SPRTs and HTSPRTs will be referred to as SPRTs.)

The first question that comes to mind is: "How does annealing aid in the care and maintenance of an SPRT?" Several answers can be provided for this question.

First, during the course of normal use, an SPRT is subjected to mechanical shock, which induces strain in the sensor wire resulting in a change in resistance. Mechanical shock can be incurred by the slightest tap to the SPRT sensor while inserting or removing it from an instrument. Vibration during transport can also be a cause of mechanical shock. The SPRT is an extremely delicate instrument. Even with great care, mechanical shock can be introduced causing a significant change in resistance. Annealing the SPRT at 660°C for one hour can eliminate most of the strains caused by minor shocks and restore the resistance close to its original value. Annealing is always advisable after any transport of an SPRT. ²

Second, all solids inherently contain defects. A "crystalline defect" is defined as a lattice irregularity having one or more of it dimensions on the order of an Angstrom. Point defects, a type of crystalline defect, are associated with one or two atomic positions in the crystalline structure. The simplest and most common point defect is a vacancy or vacant lattice site, one normally occupied from which an atom is missing. Vacancies are formed during solidification and as a result of atomic vibrations. The concentration of the point defects is dependent upon the temperature. The equilibrium concentration of vacancies (Nv ) in the pure platinum wire of an SPRT increases exponentially to an increase in temperature as shown by

where N is the total number of atomic sites, Qv is the vibration energy required for the formation of a vacancy, T is the absolute temperature in Kelvins, and k is Boltzmann’s constant.

For most metals, the fraction of vacancies (Nv /N) just below the melting point is on the order of 10^-4 or one lattice site out of every 10,000.³ Removing an SPRT from a high temperature and rapidly cooling it to room temperature traps this high concentration of point defects in the crystalline structure causing an increase in resistance. This increase can be as high as 30 mK. Annealing the SPRT at 700°C for two hours can significantly reduce the increase due to the trapped point defects. The SPRT should be cooled to at least 500°C at a rate of roughly 100°C per hour. Once the SPRT has reached 500°C, it may be removed immediately to room temperature without harm. Although the removal of the SPRT to room temperature from high temperatures should be avoided as much as possible, it is an inescapable part of fixed-point calibration.

During the calibration process at high temperatures, it is recommended that the SPRT be preheated and kept in an annealing furnace at a temperature close to the fixed point to be calibrated. Once the SPRT has been calibrated, it can be removed quickly from the fixed-point cell and returned to the annealing furnace. Annealing the SPRT for two hours after calibration and slowly lowering the temperature to 500°C prevents the quenching in of lattice defects found in the platinum wire.

Third, oxidation impacts the purity of the element and therefore, the accuracy of the temperature readings. A surface film of platinum oxide (PtO2) forms in the range of –40 to 300°C. Initially, this can increase resistance at rates equivalent to 0.5 mK per hour. Oxidation forms in the body of the wire in the range of 300 to 500°C and can cause resistance to increase by a rate as high as several mK per hour.⁴ The platinum oxides can be dissociated by annealing the SPRT at 660°C for 1–2 hours.⁵

Fourth, according to supplemental information for the ITS-90, "after using a high-temperature resistance thermometer at temperatures above about 700°C, the thermometer should be annealed before making measurements at lower temperatures, in particular before making the measurements at RTPW."⁶

The National Institute of Standards and Technology (NIST) lists annealing as the first step in the calibration procedure for SPRTs. Experimentation specific to NIST’s PRT laboratory conducted by B.W. Mangum, E.R. Pfeiffer, and G.F. Strouse provides the "optimum" time for annealing an SPRT before calibration (listed in Table 1).⁷ Although Table 1 lists specific cool-down times for specific temperatures, Hart experts recommend the general 100°C-per-hour rule, which will satisfy the cool down requirements, equalize the point-defect concentration, and simplify the process.

The second question to be answered is: "What type of furnace is needed to anneal an SPRT?"

At high temperatures, the lattice structure of most metals becomes quite loose. This allows some metal ions to come off the surface of any metal used in the furnace at high temperatures. This is analogous to steam rising from hot water. Since molecular activity increases with temperature, so does the amount of ion loss and the risk of contamination. Ion transfer occurs at different temperatures for different metals. Contamination has been attributed to copper, nickel, iron, and manganese. In addition, super-cooled quartz becomes transparent to these metal ions permitting their transfer to the pure platinum wire of the SPRT sensor. The new alloy formed by this process has a different alpha (a ) curve than the pure platinum, which results in a loss of calibration.

Contamination from metal ions may come from a metal furnace block, heating elements, or any other source inside the furnace. Small amounts of metal used in the processing of ceramics may also cause contamination. Contamination is a significant problem encountered by many SPRT users. It is virtually impossible to recover a contaminated SPRT, therefore, contamination should be avoided if at all possible.

Consequently, it is extremely important that the annealing furnace be designed to eliminate any possibility of contamination to the SPRT. The Hart 9117 Annealing Furnace utilizes a specially designed quartz-encased graphite cell to prevent contamination to the thermometer. The block is assembled into a high-purity, specially cleaned quartz container which is then evacuated and charged with argon to prevent oxidation. This specially designed quartz/graphite cell provides uniformity and stability while preventing contamination of the thermometer.

The annealing furnace must also provide good stability and uniformity. The Hart 9117 provides a stability of ±0.5°C with a uniformity of 1°C over the first three inches of the thermometer. Finally, the furnace should provide a means of setting a ramp rate for the heating and cooling of the SPRT. The Hart 9117 controller has fully programmable ramp and soak rates specifically designed for annealing. The furnace operates in a range of 300°C to 1100°C providing a full temperature range for the annealing of SPRTs.

The annealing furnace is not an option in the modern temperature laboratory but a necessity for the proper care and maintenance of SPRTs. Whether the laboratory calibrates its own SPRTs using fixed-point cells or ships the SPRT to a competent laboratory for calibration, an annealing furnace is an invaluable tool for maintaining the SPRT calibration—from annealing after shipment to assisting in the fixed-point calibration process.

REFERENCES

1 Bureau International Des Poids et Mesures, Supplementary Information for the International Temperature Scale of 1990, p. 89 (1990).

2 Xumo Li, "Producing the Highest Accuracy from SPRTs," Measurements & Control, (June, 1996), pp. 118-120.

3 William D. Callister, Jr., Materials Science and Engineering: An Introduction, 2nd ed. (Wiley, New York, 1991), p. 72.

4 Bureau International Des Poids et Mesures, Supplementary Information for the International Temperature Scale of 1990, p. 97 (1990).

5 T. J. Quinn, Temperature, 2nd ed. (Academic Press, London, 1991) pp. 233–234.

6 Bureau International Des Poids et Mesures, Supplementary Information for the International Temperature Scale of 1990, p. 91 (1990).

7 G. F. Strouse, "NIST implementation and realization of the ITS-90 over the range 83K to 1235K: reproducibility, stability, and uncertainties," in TEMPERATURE: Its Measurement and Control in Science and Industry (6), James F. Schooley (Ed.)., (1992).

Table 1: SPRT Annealing Procedure Based on NIST Research