Basic Concepts in Thermal Analysis

If you are new to the concepts of thermal analysis here are a few answers to basic questions.

It has been my experience that many concepts of thermal analysis apply to or at least can be beneficial across different metal types. It dosn't hurt to look at different examples of a liquidus from iron or aluminum when trying to understand an exotic metal or even a differrent common one. Above are some of the most common phase diagrams we use for foundry alloys.

What does thermal analysis do? And how can it help my foundry?

TA gives you a quick and dirty answer to how your meltal is freezing. If you don't add anything to your sample cup, it will tell you about microstructure and voids. If your metal is iron and you have tellurium in your cup (very common) it will tell you your carbon content, and can estimate your silicon if your process is stable for manganese and chrome.

A sample will take 3 to 5 minutes to run and you do not need a specialized person to run the sample. A good furnace ooperator or metal pourer will do. In an iron foundry, the furnace operator will take a sample about 5 minutes before the metal reaches temperature. Inside of 3 minutes the computer program will be telling him if he needs to add more carbon or silicon to the heat. He can then do this while power is still on the furnace, and the added material can be sucked uunder the metal and quickly dissolve. By the time the furnace is ready to tap, the chemistry is right, and productivity is maximized.

How does it work?

When metal turns solid, it gives back the heat you put in it to melt it. This heat slows the cooling of the sample and the temperature doesn't fall as fast. When this happens we call it an arrest. The temperature these arrests happen at, how strong they are and their relationship to other arrests all tell us about what is going on in the metal. And then from this information, we can calculate different properties of the melt.

What does an arrest look like?

Liquidus Temperature curve Liquidus cooling rate curve They look like bumps or flat spots on the temperature curve. They are generally small and hard do see so we have introduced the cooling rate curve to allow people to better see what is happening. In the examples to the right, the red temperature curve has an arrest that is called the Liquidus (more on this latter). It is long and picking out the temperature for the arrest could be any where on the flat spot. The second example shows the cooling rate which the computer derives from the temperature curve and is about 100x more sensitive to changes. Here you can see why the liquids point was choosen. It is at the bottom of the green cooling curve where the blue curve (2nd derivative) crosses zero (the horizonal black line). By using science, we take the guess work out of the problem.

Are all Thermal Analysis Systems like that??

No they are not. TA systems differ by how precise they are, and what kind of technology they use to filter out the noise and identify the arressts.

The oldest versions used a paper chart, and the melter had to guess where the arrest was. Should the carbon be running a little low, he would move just the paper on the instrument until he could truthfully report an arrest temperature in range.
The next generation produced in the '70s used some simple math to pick the arrest point. These instruments are still quite common in the developing world, but their accuracy is poor and they are very suseptable to how full the cup is, the iron temperature, and the cooling rate of the sample.
The third generation put a small LCD screen on the front of the instrument, and used a built in computer, but the computer was not strong, the programming was not well done, and the results still were not very good. These instruments are still common with one major manufactor still selling them. They don't show the cooling rate because their their filtering is poor and the cooling rate curve is a mess. Their standard error on carbon was reported by R.Heine as 0.05% C. I.e. 95% of the reported balues will be within +/- 0.10% carbon.
The fourth generation used desktop computers and actually started looking at more than just the three main arrests (Liquidus, Eutectic, Solidus). This generation started looking at the length of arrests and the shapes of the curves, and the ratios of parts of the curve. Some of these instruments were very expensive for what they did. They took a long time to setup and configure (days to week).
Our product is now into the 5th generation where we are now looking at predicting microstructure. In iron we detect oxididation, shrinkage, gas, and carbides. We can predict ferrite levels, mode of freezing (hypo, eutectic or hyper), nodule count, and nodularity. In Aluminum we can test for inoculation level(either TiB n 300 series or CuP in 200 series), beta crystals, gas, shrinkage, modification, copper and magsilicide.
The next generation of TA equipment will actually be able to put percentages on carbides, beta crystals, and other microstructure imprefections.