The accuracy and success of thermocouples is a major factor in the research program of a high-pressure laboratory. COMPRES is working on improving the situation with thermocouple use in multi-anvil assemblies, in order to raise sucess rates.

Type C thermocouples (tungsten-5%rhenium / tungsten-26%rhenium) are in widespread use for high-pressure research. Although other thermocouple types (especially Type S and Type K) are also commonly used, COMPRES will work on improvements in the success rate of Type C thermocouples first. This is because these are thought to have a smaller pressure effect on the thermocouple emf, leading to smaller systematic errors (because of their high stiffness, the band structure may not change as much with pressure - or so we hope!). Also, they survive to extremely high temperatures. Finally, they are OK for the reducing conditions suspected to exist in many high-pressure assemblies. In the COMPRES 8/3 assembly, temperatures of 2319 degrees C have been reached for 10 minutes, and measured. That was the limit of the thermocouple reader and the normal limit of the ambient pressure calibration of Type C thermocouples).

Because of the brittleness of tungsten, and to create the thermoelectric effect for measuring temperature, rhenium is added. The wire is manufactured by pulling through successively smaller sintered diamond dies, and annealing between steps. A final annealing is done after the target diameter is reached. However, for our purpose we still need to work on the common problem of thermocouple breakage during compression. The type C thermocouples break often, especially in the gasket area, where strain is highest. Here are some steps we are taking to reduce thermocouple breakage:

A. Wire Thickness

Jim Van Orman has tested both .010 and .005 wire thicknesses in the 8/3 assembly, and has suggested that a size somewhere between .010 and .005 would be ideal for this size assembly.

Overall, it seems that having a variety of wire sizes to test will be useful, since a different thickness might be ideal for each different assembly size.

The primary suppliers of thermocouples, Engelhard and Omega, only sell .010 and .005, and they outsource their wire (ie they don't make their own). However, a major manufacturer of type C wire, Rhenium Alloys, Inc., sells the wires to order.

COMPRES now has 6 different thicknesses of type C wire available for testing: .010, .009, .008, .007, .006 and .005 inches. Since each assembly might perform better with a different thickness of wire, and since 1/1000 inch is quite significant at these small sizes, we will keep small stocks of all 6 of those sizes for testing, until optimal sizes are known.

One note of caution: the wires are supplied without calibration. Does this matter? Should we calibrate them ourselves? How would this be done? These questions still need to be addressed by COMPRES.

B. Annealing

It may help to anneal the wire before use. Stony Brook does this during the preheating of assemblies in an argon oven; although, they bend the wire after annealing, which may cancel the annealing process; also, the thermocouple companies use a mix of argon and hydrogen for annealing because argon gas contains significant oxidizing impurities. COMPRES will work further on the proper annealing of the thermocouple wires to prevent breakage.

C. Welding

A good setup for welding the W/Re wire would help in ease of assembly-making and use. Currently the wires are crossed next to the sample. This makes the assembly of the 8/3 somewhat tricky, and also means that the thermocouple connection isn't made until the sample is partly compressed. Rhenium Alloys, Inc. suggests the following: use a tungsten electrode, and a welding box of some kind with either Argon or Argon/Hydrogen gas. We will develop good techniques for welding small W/Re thermocouples without excessive embrittlement (or with a post-annealing), and will test them in the cell assemblies.

D. Trenching of the furnaces and octahedra

The thermocouple is brought outside the assembly through trenches cut in the furnace, sleeves, and octahedron. These are usually cut by hand. With the live tooling on the CNC lathe, we are testing the idea of having the trenches pre-cut. This requires a numerical specification of the correct angles, widths and depths for the thermocouple trenches, which we will have to arrive at with successive testing. In the case of foil furnaces (like the Re furnace for the 8/3 assembly) the trenches have been made in the foil by computerized wire EDM cutting. The hope is that some of these techniques will improve the overall thermocouple success rates.


Back to COMPRES cell assembly page