It is well recognized in the industry that thermocouples begin to decalibrate over time. The causes of the decalibration are multiple. However, there is also an insidious problem called virtual junction error that can affect mineral-insulated thermocouples and cause significant errors.

Failure of a sensor’s insulation at high temperature may result in virtual junctions that cause erroneous temperature readings of more than 300°F. To better understand the cause of virtual junction error, it is helpful to review some of the basics of thermocouple measurement principles and construction materials.

Fig. 1. The voltage is generated in the gradient region

Thermocouple Principles

What makes a thermocouple work is the Seebeck effect - a physical phenomenon causing a flow of electrons along dissimilar wires joined at one end, creating a voltage across the open end. The assumption of many is that the voltage generated by a thermocouple is created at the tip of the sensor - the place where you want to measure your process temperature. In reality, the voltage is created along the entire thermocouple by the difference in temperature between the hot end of the thermocouple and the cold end. Further, the voltage is created in the temperature gradient region and not where the thermocouple is isothermal (Fig. 1).

Table 1. Head-to-head comparison of MI-Dry IR compared to competitive MgO thermocouples

Mineral Insulation

Although there have been many improvements in thermocouple materials over time, the use of magnesium oxide (MgO) as the insulator has not changed for over 50 years. MgO is used as the insulant in over 95% of the thermocouples manufactured today. The effectiveness of an electrical insulating material is usually determined by measuring its resistance. In thermocouples, this is called Insulation Resistance (IR). It is not uncommon to see an MgO thermocouple with an IR value of 1 Gig-ohm at room temperature drop to readings of 100 ohms or less when operating at 1100°C (~2000°F). It is also well known that MgO absorbs moisture and can lose its insulation value as the temperature of the material increases.

Loss of IR can cause failure of the thermocouple. Since MgO absorbs moisture easily, care must be exercised in the thermocouple manufacturing process. Thermocouples are usually worked “hot” and “baked out” by the better manufacturers before sealing to eliminate as much moisture as possible from the inside of the sensor. Moisture trapped inside the sensor can react with the MgO, causing it to further lose its insulating properties. This moisture can contribute to a virtual junction.

A new dielectric mineral insulation, an extremely stable high-performance ceramic made specifically for use in mineral-insulated, metal-sheathed thermocouples and RTDs, is now being used to significantly improve thermocouple performance. This material (MI-Dry™) is much less hygroscopic than MgO and has superior electrical properties compared to thermocouples made with MgO. MI-Dry not only has superior electrical resistance, but it blocks the diffusion of trace elements to the thermocouple wires, resulting in extended life. The new ceramic is noncorrosive to metals up to 2000°C (3630°F), whereas MgO will react with most metals above 480°C (900°F). IR, measured by an ohmmeter from wire to sheath, is used to check thermocouple integrity. Higher IR is more desirable. Table 1 summarizes IR measurements from two experiments where 1/4” OD type-K thermocouples made with MI-Dry were compared to thermocouples insulated with MgO at 1200°C (2200°F).

Another test was conducted using 1/8” thermocouples. Comparison of IR measured at high temperatures is shown in Figure 2. Note that MI-Dry IR was approximately 100 times greater than MgO over the entire temperature range. Also note that the IR scale is a log scale. Insulation resistance of thermocouples is a function of geometry and temperature as well as the insulant. Geometrically similar thermocouples fabricated with MI-Dry consistently exhibit 50-100 times higher insulation resistance than those manufactured with MgO. This property reduces or prevents shunting and virtual junction errors.

Fig. 2. Insulation resistance (IR) vs. temperature

Virtual Junction Errors

Because of their ruggedness and wide temperature range, type-K thermocouples are by far the most common thermocouples used for high-temperature measurement in industry. The thermocouple elements have a measuring range from -400°F to as high as 2500°F. For severe service, these thermocouples are constructed with high-temperature protective sheath materials including stainless steel and Inconel. Magnesium oxide (MgO) is almost universally used as an electrical and chemical insulant, or dielectric, to separate the thermoelements from one another and from the sheath. As a thermocouple is brought to high temperature, the electrical resistance or IR of the dielectric, as with most materials, diminishes (Fig. 2).

Fig. 3. Typical application where virtual junction can occur

If the electrical resistance of the dielectric gets low enough, a conductive path can form allowing electrons to flow across the insulation. If the point at which the electrical path is formed is not at the measuring tip, a “virtual junction” is formed. Virtual junction error most commonly occurs in a thermocouple when, at some point along the thermocouple’s length (between the hot tip and the measuring end), the temperature exceeds the temperature at the hot tip and breakdown of the dielectric occurs (Fig. 3).

Test of Virtual Junction Effect

To illustrate how virtual junction errors can create major headaches for those responsible for a company’s control and instrumentation systems, AccuTru conducted a series of tests that illustrate both the effect of virtual junction errors in measurement systems and the ability of MI-Dry mineral insulation to prevent virtual junction errors in your systems.

For one test, AccuTru’s research team used two type-K thermocouples, one with MI-Dry insulation and the other with conventional MgO insulation. Both thermocouples were 10 feet long and 1/8” diameter. The thermocouple sheath material used was Inconel 600.

Fig. 4. Diagram of test setup

Test Conditions

Two high-temperature ovens were used in the test. One contained the tip and the second the midsection of the thermocouples (Fig. 4). Reference sensors were placed in both the tip and midsection furnaces. The temperature in the tip furnace was raised to 1500°F, then the midsection furnace was raised to 2000°F. Type-K thermocouples were chosen because they are so important in industry and are by far the most common thermocouples used for high-temperature measurement.

Fig. 5. Results of 1/8” OD type-K virtual junction error test

The test was conducted over a period of 430 hours. Once the test equipment was set up, the controllers for the furnaces were set to ramp up and hold the set-point temperature for the duration of the test. Figure 5 clearly illustrates the virtual junction effect. With the midsection of the thermocouples at 2000°F and the thermocouple tips at 1500°F, the MgO-insulated sensor began to show signs of the formation of a serious virtual junction problem within about 25 hours. By 200 hours, the temperature deviation from the actual tip temperature was more than 250°F. For the duration of the test, the AccuTru sensor, insulated with MI-Dry remained at a constant 1500°F (Fig. 6).

The following test results are shown in Figure 5.
  • Virtual junction error was apparent in the MgO thermocouple within the first few hours.
  • After 100 hours, the MgO sensor was reading 10% high (170°F) while the MI-Dry sensor was reading correctly.
  • After 400 hours, the MgO sensor was reading 19% high (290°F) while the MI-Dry sensor continued to read correctly.

Fig. 6. Magnitude of virtual junction error

The sensors were allowed to cool to ambient temperature, and the test was repeated. This time the nearly 300°F error in the MgO sensor reappeared immediately, indicating that the effect on the sensor was permanent.

Figure 6 shows the magnitude of error in virtual junction conditions. The same results should be seen for other thermocouple types such as type-N and type-J or other wire pairs that can operate in this temperature range since it is the insulation that is compared in this experiment.

Virtual Junction Test Results - 1/4" Thermocouples

Comparable tests on 1/4” type-K thermocouples at the same temperatures show almost exactly the same results. This indicates that the virtual junction problems experienced in type-K thermocouples is a function of the MgO insulant, not the thermocouple size or insu-lation thickness.

Applications in Furnaces, Boilers, Heaters and Turbines

Virtual junction errors are common in applications such as brazing applications, heat treating, ethylene furnaces, chemical reactors, in-dustrial boilers, heaters and turbine engines due to flame impingement and concentrated radiation somewhere along the thermocouple’s length. In fact, the test conditions reported above simulate the conditions in a typical ethylene reactor where thermocouples are being used to measure tube temperatures but must pass through the firebox to reach the tubes. The increased temperature in the midsection of the thermocouple can accelerate the breakdown of the dielectric mineral insulation and allow electrical shunting between the thermoele-ments. This shunting causes incorrect temperature readouts because the thermocouple is effectively shortened and begins measuring the temperature gradients from a point different than the tip.


Large measurement errors due to virtual junctions in thermocouples made with MgO have been demonstrated in lab tests and observed in the manufacturing processes listed above. The observant operator might detect errors of this magnitude by comparison to other measure-ments or observations. Nonetheless, the temperature measurement is rendered useless. Serious problems can occur when these errors go unnoticed or are small enough not to be obvious, especially when the measurements are depended upon for process control. Costly and perhaps catastrophic results can ensue.

The new dielectric mineral insulation developed by AccuTru provides superior per-formance over MgO in thermocouple construction. In addition to reducing or eliminating virtual junction errors, it can reduce thermocou-ple decalibration, leading to longer sensor life and more stable signals. This dielectric, called MI-Dry, has several advantages over MgO:
  • Less hygroscopic
  • Increased electrical-insulation properties at high temperatures
  • Slows and/or prevents thermocouple decalibration

These advantages can reduce or eliminate the risk of temperature uncertainties caused by virtual junction error for most high-temperature applications. IH

For more information: Dan Barberree is Vice President for Research and Development at AccuTru Sensor Technologies, Kingwood, Texas 77339; tel: 1-800-594-5737;;

Additional related information may be found by searching for these (and other) key words/terms via BNP Media SEARCH thermocouple, virtual junction, insulation resistance, dielectric insulation, hygroscopic