What is the current state of predicting properties of metallic alloys?

[Stefan Udrea]

I've read an article written in the 60's by the russian physicist Pavel Kapitza where he was saying that one of the field where physics was less successful was in finding laws to predict the properties of the metalic alloys,and thus the compositions of various alloys is determined by the engineers' intuition and by trial and error.
I wonder, now that we are in 2006, has the situation changed ?

 

[nbo10]

Materials Science is still an experimentalist playground. The computing power & theory is still lacking for theorist to make serious predictions.

 

[PerennialII]

The word "predicting" is always a difficult one, and typically a huge challenge, but over different modeling size scales use of for example first principles, molecular dynamics, various continuum theories (like damage mechanics, crystalline plasticity, "continuum" version of dislocation dynamics) can yield computationally estimates of various material properties and do so quite successfully. As 'nbo10' said would call it an emerging field in comparison, since there is a LOT of theoretical work to be done and the computational cost for such analyses is often immense (for example, see who at university computing centers (or related) consumes the most resources). But again, things have also changed a lot since the 60s (for the better and quite fundamentally for many particular related fields of physics working on these matters), surely there is always quite a bit of experimental background, but wouldn't call it a totally empirical playground. Anything specific you're interested in?

 

[ZapperZ]

I've read an article written in the 60's by the russian physicist Pavel Kapitza where he was saying that one of the field where physics was less successful was in finding laws to predict the properties of the metalic alloys,and thus the compositions of various alloys is determined by the engineers' intuition and by trial and error.
I wonder, now that we are in 2006, has the situation changed ?

It is always a good idea (in fact, I'm trying to make it a requirement on here), that when you are talking about something you read, that you make an exact citation of the article. This will enable some of us to possibly find it, double-check if you read it correctly, and then make comments. Without that, it is almost impossible to know what exactly is being talked about.

Having said that, let's first of all established that band structure calculations have been quite sucessfull in "predicting" behavior of a lot of solids. The problem comes in, as has already been exhibited by some alloys, and even some transition metal oxide, when band structure theory fails. You then have materials that exhibit Mott-Hubbard band, Brinkman-Rice model, etc. These are more difficult, and in many instances, it is hard to predict when something will be described by which model. Most are done a priori, i.e. an experimental discovery is made first, then after many studies, we then realize that it is a Mott-Hubbard compound.

Take note that there is almost an unlimited cocktail of metal alloys that can be produced. So with such complexities, you can't really just punch in the ingredients and expect to be able to make an exact prediction on which one is going to be a metal, etc.

However, once we know what it is, i.e. if it is a Mott-Hubbard metal, a band metal, etc., then the rest of its properties is well-known and well-described.

Zz.

 

[asvani]

The problem with alloys is that it is primarily their disordered structure. Your run-of-mill electronic calculations will not work because they are geared towards nicely ordered repeating structures. As far as I know, for ternary alloys, ordinary DFT is hopeless and there still remains much work to be done to model ternary alloys and to understand their thermodynamics.

 

[Astronuc]

I've read an article written in the 60's by the russian physicist Pavel Kapitza where he was saying that one of the field where physics was less successful was in finding laws to predict the properties of the metalic alloys,and thus the compositions of various alloys is determined by the engineers' intuition and by trial and error.
I wonder, now that we are in 2006, has the situation changed ?

We now have experience with a plethora of alloys, but we still do not have a complete or intimate understanding of all properties of an alloy. For some simple alloys, e.g. ones which are solid solution binary alloys with a low or moderate portion of one constituent, may be more easily describe by the physics. However, one must also accurately understand the grain size and structure (crystal orientation).

Alloys with interstial elements and secondary/tertiary phases are much more complex, and add to that the grain microstructure, and we do not have the capability to provide intimate models of the physics. As PerennialII alluded, it is still largely semi-empirical, but there is usually an experiential database to reference.

Another thought - single crystals of binary alloys might be easy to model. Also, small crystals of relative uniformity are probably relatively easy to describe. The problem arises from non-uniformity in composition and microstructure/texture, particularly for polycrystalline/polyphase alloys.

This has nothing to do with alloys, but does cover some applications of models mentioned by Zz.
http://www.physik.uni-augsburg.de/theo3/Talks/I2CAM_Ringberg_2005.pdf