PROCEEDINGS
OF THE TOMSK ORDER OF THE RED BANNER OF LABOR POLYTECHNIC
INSTITUTE named after S. M. KIROV

Vol. 70　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　1950

PROTECTIVE COATINGS FOR THERMOCOUPLES

B. V. VOLKENAU

Iron–constantan thermocouples are used for long-term measurements up to 600°C and for short-term measurements up to 800°C.

For modern high-temperature technology such limits are clearly insufficient and are explained only by the instability of the thermocouples themselves. If one seeks the upper limit of measurements with an iron–constantan thermocouple, then this limit should be taken as that point (temperature) at which the proportionality of the readings is violated,

\[ \frac{de}{dt} \cong \mathrm{const}; \]

an infinitely small change in thermoelectromotive force for an infinitely small change in temperature is constant. For iron such a point will be \(Ac_1 = 910^\circ\mathrm{C}\). On the other hand, a frequently encountered technological process may be taken as the upper limit. I took as such a process the hardening of carbon steel with a low carbon content. Such steel is hardened at temperatures up to 850°C. This temperature was therefore chosen as the upper limit for long-term measurements with an iron–constantan thermocouple. At this temperature a layer of iron oxide grows vigorously on the iron electrode. According to Arkharov [1], vigorous growth of the oxide layer begins at 570°C, and as the temperature rises the oxidation rate increases.

There are methods of protecting iron from oxidation by coating it with various metals: chromium and aluminum, and also with silicon. The electrodes of thermocouples are coated with the more stable metal [4; 5].

Aluminum coatings have long been used in technology and are better developed than the others. My own observations confirm the effectiveness of coating iron with aluminum in those cases where iron articles operate at high temperature.

Iron pots for annealing ball-bearing rings operate precisely at a temperature somewhat below that indicated, for 30–40 annealing operations, and then fail. An experimental pot coated with aluminum (the experiments were carried out by engineer P. M. Potapov) withstood 100 operations and remained intact.

Coating the iron electrode with aluminum should cause a change in the thermoelectromotive force, which is undesirable. Therefore, for the experiments a long-known and long-rejected pencil thermocouple was chosen, shown in Fig. 1. In such a thermocouple the iron electrode is an iron sheath and a “boss” (below, at left). The constantan electrode, as usual, is constantan wire welded to the boss. The boss, in turn, is welded to an iron tube forming the sheath. The thermocouple head consists of two iron rings with connecting bolts and a rubber stopper, through which the constantan electrode passes. The lower ring is welded to the iron tube—the thermocouple sheath; the upper ring is fastened to the lower bolts, pressing the rubber stopper.

Such a head design, with the entire sheath hermetically sealed, prevents the access of oxygen to the constantan electrode and to the inner surface of the iron electrode—the sheath—and protects these parts from the formation of oxides.

The outer part of the tube and of the plug is coated with aluminum. The tube is coated with aluminum over \(3/4\) of its height. Two methods were used for coating with aluminum: 1) heating the sheath for 18 hours at \(900^\circ\mathrm{C}\) in a mixture of 50% aluminum powder, 50% quartz sand, and 2% ammonium chloride, and 2) immersing the sheath in an aluminum melt with a small addition of iron (5%). The first method gives cleaner coatings, but requires more time than the second.

In making the thermocouple, difficulties were encountered in producing the joint between the plug and the constantan electrode. From a series of experiments, the best method proved to be the one proposed by Assoc. Prof., Candidate of Technical Sciences N. A. Balakin, which gives a clean joint without solder.

Fig. 1

Between the copper electrodes of a butt-welding machine an iron plug is clamped, drilled with a drill of a somewhat larger diameter than the constantan electrode. Calcined borax is poured into the hole so that a small mound is formed on top. When the current is closed, in less than 1 minute the plug is heated to white heat, and the borax melts. At this time the constantan wire is introduced into the hole and the current is switched off; the constantan, having a lower melting temperature than white heat, melts and gives a very clean weld with the iron of the plug. The constantan wire should then be straightened and the plug welded to the tube. The straightening should be carried out carefully, so as to make it possible not to put insulating beads on the constantan. In our first experiments the beads were used; later they proved unnecessary. A constantan wire 2 mm in diameter in a \(1/2''\) gas pipe, 750 mm and 1 m long, never made contact with the pipe.

A total of 9 thermocouples were made. Four of them proved unsuccessful either because of poor aluminum coating or because of poor welding. The iron of the tubes and plug had the composition of ordinary commercial material (Table 1).

Table 1

The constantan that could be obtained was far from standard constantan (Table 2).

Table 2.

Many unsuccessful calibrations had to be made before it proved possible to obtain a satisfactory one, with little scatter in the readings.

Calibrations with a bare (without a sheath) chromel–alumel thermocouple, end-to-end, when the experimental thermocouple and the reference thermocouple were inserted from opposite ends of the furnace, gave too large a scatter of readings. The same result was obtained when the thermocouples were inserted from one side of the furnace and tied together. The results were somewhat better with a platinum-rhodium–platinum thermocouple. In this case the difference in readings between different calibrations averaged about 15°C, and in individual cases reached 25°C.

Better results were obtained when the bobbin was drilled to a depth of 3 mm and the hot junction of a platinum-rhodium–platinum thermocouple was placed in this hole. From the mean (true) data on the thermoelectromotive forces subsequently obtained, a diagram was constructed (Fig. 2) of the maximum deviations of these calibrations (solid broken lines). With these calibrations, 3 thermocouples were placed on long-term test in factory furnaces at temperatures of 800–900°C. The times when the furnaces cooled, were unloaded, and were heated were not counted. After 600 hours of testing, the calibration lay within the limits obtained before testing.

Fig. 2

Notations in the figure: vertical axis, “Maximum deviations”; horizontal axis, “Deviations from the mean value in degrees”; label at right, “Mean value”; legend: solid line — “3 thermocouples with 3-mm drilling, 16 calibrations”; dashed line — “thermocouple with 8-mm drilling, 3 calibrations.”

The not entirely satisfactory calibration obtained with the solid broken line led us to seek better results. Drilling to a depth of 8 mm gave a completely satisfactory agreement of the calibrations (Fig. 2,

dashed line). Two calibrations were made in succession, and a third after holding in the furnace for 200 hours. The deviation from the mean calibration is at most 0.8°C, and the scatter is somewhat more than 1°C. For technical thermocouples this is quite satisfactory accuracy.

Because the constantan used in the experiments was nonstandard, the thermoelectromotive forces obtained at different temperatures differ from calibration 538 (Table 3).

Comparative calibration in millivolts

Table 3

Comparing the thermoelectromotive force per 100°C in different intervals of the scale, we obtain Table 4.

Table 4

Thermoelectromotive force per 100°C in millivolts

In an electrically heated laboratory muffle furnace \(75 \times 240 \times 375\) mm, automatically controlled, a check was made of the inertia of the experimental thermocouples in comparison with a factory-made thermocouple. The factory-made thermocouple consisted of two electrodes with porcelain insulation, inserted

Fig. 3

in an iron sheath. Both thermocouples, tied together, were inserted into the furnace close to the platinum-rhodium–platinum thermocouple by which the furnace temperature was measured. The furnace was first adjusted to a specified temperature, and the pair of two thermocouples was inserted into the already heated furnace.

Fig. 4

In the three appended diagrams (Figs. 3, 4, and 5), I denotes the temperature according to the readings of the platinum-rhodium–platinum thermocouple,
II denotes the temperature according to the readings of the experimental thermocouple,
III denotes the temperature according to the readings of the factory-made thermocouple.

Fig. 5

Temperatures are plotted along the ordinate axis; time in minutes is plotted along the abscissa axis on a logarithmic scale. The platinum-rhodium–platinum thermocouple shows a drop in temperature after the insertion of the pair of two ther-

thermocouples in sheaths. Heat is transferred from the furnace to the thermocouples in different ways. At first the experimental thermocouple is heated more intensely, while the factory thermocouple is heated more slowly; later both thermocouples indicate almost

Fig. 6

the same temperature and reach the furnace temperature almost simultaneously.

At a low furnace temperature (Fig. 5), the drop in the readings of the platinum–platinum thermocouple is negligible. The time required for the inserted thermocouples to reach the furnace temperature is considerable—2 hours.

Three thermocouples were cut along the plane passing through the axis of the tube in order to determine wear (Fig. 6). Almost no corrosion was found in any of them, except at the place where the knob was welded to the tube.

To determine the penetration of aluminum by diffusion to the junction of the knob with the constantan electrode, and to determine the change in

Fig. 7

the composition of the iron, a method of thermal analysis was used with an instrument designed by engineer V. K. Dobrer (Fig. 7). The figure gives a schematic diagram. The instrument consists of a step-down transformer (on the left), feeding the heater. The heater has good external thermal insulation, not shown in the diagram. The heater is constructed in the form of an electric soldering iron. When its tip touches the specimen (cross-hatching), lying on an iron support (hatching ...

oblique), a different thermoelectromotive force is obtained, depending on the material of the article. Before the test is begun the heater must be heated to a definite temperature. This temperature is determined from the thermoelectromotive force arising when the tip and the base are brought into contact.

The experiment for studying the change in the composition of the iron of the bobbin and of the tube (sheath) was set up as follows. The cut sheath (Fig. 6) was thoroughly cleaned along the plane of the cut and along part of the lateral surface of the cylinder. With the cleaned part of the cylinder, the cut sheath was placed on the base of the thermoanalyzer. When the tip touched various points of the cut sheath (Fig. 8), from the surface of the cylinder (point 7) and then toward the axis of the thermocouple (point 1), different thermoelectromotive forces were obtained. The values of the thermoelectromotive force at the various points, corresponding to the points in Fig. 8, are plotted in Fig. 9.

Fig. 8

Taking into account that in the experiments a \(1/2''\) tube was used and, consequently, the distance from the surface of the cylinder to the axis of the tube is \(11\ \text{mm}\), it is not difficult to see that the change in the composition of the iron extends only a very small distance from the surface, only \(2.5\ \text{mm}\). What these changes are the thermoanalyzer does not show, nor is this important for us. What is important is that they

Fig. 9

do not penetrate deeply. Obviously, the sheath and the entire thermocouple can serve, beyond the 600 hours already worked, for a considerable time longer.

Conclusions

1. The proposed thermocouple is simpler than the iron-constantan thermocouples now in use. It has one electrode and no porcelain insulation.

2. The proposed thermocouple can be calibrated with sufficient accuracy for technical purposes.

3. The service life of the proposed thermocouple is considerably more than 600 hours without a change in calibration.

4. The proposed thermocouple responds more rapidly to temperature changes than an ordinary one, but the time required for heating the proposed and the ordinary thermocouple to the furnace temperature is the same.

Literature

1. Arkharov V. N.—Oxidation of Metals, 1945.

2. Berkovskii I. Ya. and Kolokolova A. G.—Nickel Alloys, Metallurgizdat, 1941.

3. Instructions for the Verification of Technical Thermoelectric Pyrometers and for the Calibration of Working and Standard Thermocouples, 25—40, Moscow, 1941.

4. Kinkulkin B.—Steel, No. 8, 1933.

5. Held. S. S. Mesure des temperatures par les couples thermoelectriques. Chaleur et Industrie, 1939, No. 230.

6. Roeser and Wensel. Reference tables for iron constantan and copper constantan thermocouples. Journal of research of N. B. of Standards, 1938, vol. 20.