What’s the difference between the traditional phase diagram and MP phase diagram?
Conventional phase diagrams are composition-temperature, what MP offers is more correctly known as a “convex hull phase diagram.” The way we calculate our thermodynamic information is using a computational technique known as Density Functional Theory; this is intrinsically a 0 K technique. It gives us the relative stability of phases to each other at 0 K, which is why the phase diagram does not have a temperature axis.
There are techniques to computationally predict full composition-temperature phase diagram but these are not currently accessible to high-throughput methods, which is what we need, since we need to be able to apply the technique to all 130k+ materials in our database.
Another question is I found the Mg-Y Phase diagram in ASM database, there are only three intermetallic phases, namely MgY，Mg2Y，Mg24Y5. While in MP database, it gives 4 most stable intermetallics, YMg149, YMg3, YMg, Y2Mg. What cause this differences?
One of the benefits of Materials Project is that it can calculate energetics for many different phases, and incorporates experimental and predicted structures from several sources. This can give us a more complete picture of what structures are possible.
The relevant information to look at in MP is the “energy above hull,” this gives a measure of (meta)stability for a given polymorph, so we can see for Y-Mg we only have a few experimentally known phases (e.g. mp-615, mp-11499, mp-1988), but we have dozens of phases that are predicted to only have a small degree of metastability, so in principle might be synthesizable. Consult the search results here for an example (note the rows shaded in green are those experimentally known).
Hope this helps,