Sciencemadness Discussion Board

Eutectic determination

deltaH - 8-2-2015 at 11:55

As a rule of thumb, really low melting eutectics are often multicomponent beasts of three or sometimes even four components. It is a nightmare to experimentally hunt for a eutectic composition with so many components as the usual approach is to map the liquidus temperatures of multicomponent compositions incrementally over the range of compositions. So taking just 10% increments in composition for a three component system results in 1000 experiments already :o

I was wondering if there isn't a smarter way to do this, so I was thinking, if one just took an equimolar mixture of all the salts (say for example a mixture of nitrates of various metals), heated it so everything melts, then let it cool slowly while stirring. At some point you will hit the solidus temperature where solids begin to form, but you continue stirring and cooling slowly until the mixture consists mostly of solids and a very small amount of liquid melt as a thick slush. At that point, say when most has frozen, you sample the remaining liquid only with a hot glass syringe, maybe with a small ceramic frit on the tip to filter the crystals. Now you analyse this sample by ICP at your local university to determine the relative amounts of the metals and presto... you have your eutectic composition???

If necessary, you can take that new composition and repeat the experiment, but starting from there and keep repeating a few times to approach the global minimum.

If this could work, then you could determine the eutectic composition for any number of components, easy-peasy, so obviously it doesn't, but why not?

[Edited on 8-2-2015 by deltaH]

subsecret - 8-2-2015 at 14:59

Tell me if I'm wrong, but I don't think this would work. You're not dealing with the metals as elements, but rather as ions.

j_sum1 - 8-2-2015 at 15:42

As you rightly point out, ternary and quaternary systems are a pain, but the situation need not be quite as complex as that. Simple methods can be used to find the eutectic in two component systems -- tracing the liquidus as you suggest. And once you know its composition approximately, you can run some experiments to plot the curves more finely and determine it more accurately.

A third or fourth component is often added to a system to modify the structure of one of the phases and is generally in relatively low quantity. As a first approximation you generally assume that it doesn't affect the eutectic temperature or composition too much.

Eutectics generally have a very easily observed microstructure -- As the melt solidifies, one phase will crystallise first. Then the local concentration is tipped to the other side of the eutectic composition and the other phase solidifies. This pushes the local concentration back in the other direction and the first phase solidifies again. And so forth. Pearlite in steel is a classic, being alternate layers of cementite and ferrite. (Other structures than layering are possible.) So, if you want the exact composition of a eutectic then you can simply analyse the material produced. This of course does not tell you the exact composition and temperature of a ternary eutectic, but it does indicate the direction that the third component will introduce. Once you know that you can increase the amount of the third component and fine-tune the ratios of the first two until you find the point you need.

I would have to question why you would need to find the eutectic of a ternary system so accurately though. I guess it is useful (eg in solders) to have a well-defined melting point and it is useful to avoid local concentrations and associated brittle compounds and intermetallics when crystallising. However, I think that in most practical cases, close enough is good enough. It's an engineering problem and provided there are no deleterious effects you simply proceed with your project. In any case, metallurgy is your friend. Taking a quick look at the microstructure tells you a lot about what you have obtained.

blogfast25 - 8-2-2015 at 16:32

Quote: Originally posted by deltaH  
So taking just 10% increments in composition for a three component system results in 1000 experiments already :o



With 'as little' as 27 compositions tested you'd be able to approx. locate the eutectic with curve fitting.

See Factorial Experiment Designs.

j_sum1 - 8-2-2015 at 19:25

Quote: Originally posted by blogfast25  
Quote: Originally posted by deltaH  
So taking just 10% increments in composition for a three component system results in 1000 experiments already :o



With 'as little' as 27 compositions tested you'd be able to approx. locate the eutectic with curve fitting.

See Factorial Experiment Designs.
Good point. However, that presupposes a simple ternary system. If there are a bunch of intermetallics involved then there will be more than just three liquidus surfaces that need to be modelled. That is why I would begin with a known eutectic on a two species system and incrementally add the third component (adjusting the ratio of the first two as needed) to trace out the "valley" of the two liquidus surfaces until the ternary eutectic is reached.

It needn't be too difficult. Suppose you are investigating a system of alpha, beta and gamma.


[Spoken with all the authority of someone who has never had to do this himself. There are multiple things that could go wrong here and good lab practice would be paramount.]

deltaH - 8-2-2015 at 22:12

Thank you all for your points.

I wasn't thinking of metallic alloys though, rather eutectics from salts, for example nitrates and so-called deep eutectic solvents, but some of your points are universal, nevertheless.

Is there something inherently wrong with the approach I propose? Would it tend to the eutectic?



[Edited on 9-2-2015 by deltaH]

j_sum1 - 8-2-2015 at 22:26

I can't think of any particular reason why your approach wouldn't work if it was a simple system. I have no experience of salt systems of the type you describe however and I don't know what crazy things may be thrown up. Double salts perhaps? Water of crystallisation misbehaving? Unexpected reactions? Who knows?

I would think your whole process would be a lot easier if you could get close to the eutectic before you start. Otherwise you are going to have problems with collecting a purely liquid sample. And you really want your sample that you collect to be the final bit of liquid if possible. A kind of iterative approach may help. Begin with kg quantities. Let most of it solidify and collect the last 20 mL of liquid. Then re-melt this, allow it to solidify and collect your 1g sample.

And of course, depending on what your salts are you might not need to send the sample away for analysis. If you already know what the components are then you might be able to get away with a simple titration or something similar.