It depends on the phase diagram and the initial composition.

Dear Nolze.

Thank you for your answer.

In my case, the composition of Nd is very dulite (about 0.5at% or less), thus the initial product should be Mg12Nd according to phase diagram. The interesting thing is that, it forms Mg3Nd first rather than Mg12Nd, and this Mg3Nd will decompose to Mg12Nd during further heat treatment. Therefore, it "jump" on the phase diagram.

This means that Nd concentrates first, then it dissolves again. This precipitate process is generally accepted and not a single case. Anyway I think it's slightly violate the way nature does, where everything should follow the energy minimizing during every step. Certainly it must be the best way nature can work out even if I don't know why. Thus I wonder whether Mg alloy is a single case? Or others face the same condition?

 Can I ask how you characterize your samples (e.g. XRD, SAED)? And are you sure that Nd concentration is uniform in the entire solid solution before conducting any heat treatment?

Dear Zafari,

I am not good at characterization because my project is calculation. However, many people use TEM or HAADF STEM in their papers and find this process. Usually people heat the sample to solution temperature to make them "uniform", and then quench them and heat again to age harden it. I cannot sure whether they are uniform as I don't find they use any tech to confirm it, but it should be. Even though it does't uniform,you know it will be extremely hard to achieve 25at% (Mg3RE) in a 0.5at% solution in solution treatment.

Hi Yee Hsien, it's a good question. 

Sorry I don't get you. If your argument is merely based on calculation, how could you be that sure that Mg3Nd forms before Mg12Nd? It might simply because of some mistakes in calculations.

" it will be extremely hard to achieve 25at% (Mg3RE) in a 0.5at% solution in solution treatment."

I think it is possible. Sometimes elements can concentrate in one particular area and locally change the composition of an alloy to an entirely different thing. If 0.5at% is too small compared to 25at% in Mg3Nd, it is also too small compared to ~8at% in Mg12Al.     

the better preciseness and accuracy could be measured by more number of repetition of the same experiment. because, initially many of us doesn't find the exact thing what we need just because of our smaller mistakes.

and still if the things are done in a correct way then i up-vote your problem since your results are contradicting with phase diagrams.

Dear Zafari and Chuahan,

The work is not done by calculation, but by some other people who use TEM and tech alike. From their works,  Mg3RE is observed prior than Mg12RE. However, it is possible that some elements more like to concentration rather than pervading throughout the whole matric, and maybe people have similar mistakes, because the process are similar, and there is few experiments studying Mg-RE.

As far as I know this process is widely accepted in Mg-RE field, in spite of many unknown details.

Dear Liu

have you consider about the rate of cooling?

Phase diagram is the equilibrium condition produced by a very slow condition

where diffusion of atoms is possible to reach the balance.

However, when the cooling rate is quite high,

it can lead to the formation of "cored structure",

where the % of alloying element will be higher, compared to its surrounding,

since there is a little time for atoms to diffuse and reach the balance of the equilibrium condition.

Moreover, since Nd has higher melting temperature (compared to magnesium)

upon cooling, Nd solidified first, but many of them 'trapped', and forming Mg3Nd, (a condition that 'usually' formed in higher concentration), instead of Mg12Nd.

Thats why, as you said,

upon heat treatment, we give more heat and energy, ... and also time,

for Nd atoms to diffuse, .. decomposing Mg3Nd into Mg12Nd.

I hope this help.

From experience good points from Joy Rizki Pangestu Djuansjah- equilibrium concentration predictions are not necessarily the actual compositions found and you may need to take account of diffusivity of elements and time, and with other experiments to see if this is a reasonable assumption. 

As Yixian Liu says, be careful of results from characterisation, they may be presented as accurate but there are numerous reasons this may not be the case. Even repeated experiments will not solve all these problems.

Alloying element-rich to alloying element-poor is interesting as either a constant composition or an enrichment would as you suggest be expected. So worth checking with literature and your characterisation of this alloy to check this.

Only discuss this a bit more:

Everybody knows that simulated phase diagrams are based on apparently unrealistic conditions. On the other hand, when they are that unprecise why we are using them if they match? :-) Wishful thinking?

I am no expert in thermodynamics but I (personally) do not know any case where a phase diagram was that far away from the experiment. We are not talking about 2 or 5%.

And if you "freeze" a material as discussed above, the phase diagram does not change but in first approximation stores the chemical composition at this temperature. The temperature might be different than assumed, but this uncertainty can be discussed during error approximation. If phase diagrams would be that wrong our steel production (fast cooling, complex heat treatment) could never deliver a steel with the "same" specifications.

Dear Joy Rizki Pangestu Djuansjah, Thomas Simm and Gert Nolze,

Thank you for your answers, and I believe what Joy Rizki Pangestu Djuansjah said is very reasonable.

Sorry for suspending so long without replying to you, and it is quite busy recently.

The high temperature of Nd maybe a key point. If they are difficult to dissolve, it will certainly make a localised element-rich area, and leading to strange results. If so, then even repeating cannot solve this problem, provided the casting tech doesn't change. However, a very annoyed issue is that for Mg-Nd they do form a Nd-poor phase first, Mg7Nd, and then decompose to a element richer one, Mg3Nd. Very strange.

Characterisation is of vital importance in alloy research but, this is not my field and what I can do is merely finding what others have done and trust their work. 

The phase diagram is a very powerful tool and it has already proved itself outstanding functions as a guide in many alloy research and industry, which I never doubt. In fact if I don't trust phase diagram, I will not ask this question.

The relative reference are listed below, some of them have been updated but still work.

1. J.-F. Nie, "Precipitation and Hardening in Magnesium Alloys," Metallurgical and Materials Transactions A, vol. 43, pp. 3891-3939, 2012// 2012.

2. A. R. Natarajan, E. L. S. Solomon, B. Puchala, E. A. Marquis, and A. Van der Ven, "On the early stages of precipitation in dilute Mg–Nd alloys," Acta Materialia, vol. 108, pp. 367-379, 4/15/ 2016.

3.Z. Xu, M. Weyland, and J. F. Nie, "Shear transformation of coupled β1/β′ precipitates in Mg–RE alloys: A quantitative study by aberration corrected STEM," Acta Materialia, vol. 81, pp. 58-70, 12// 2014.