Can the melting and boiling points of chemicals be deduced or are they strictly empirical?

With enough computing power could the melting and boiling points of substances be predicted or must they be derived empirically?

Could you synthesize/engineer a novel molecule with a particular desired melting or boiling point?


I am not a Chemist but a ChemE.

Not sure what is your definition of “predicted”. If the definition is like calculating the orbitals of higher atoms from ab-initio principles like that of hydrogen, then probably not.

For many organic chemicals, the group contribution method works well : Estimation of the normal boiling point of organic compounds via a new group contribution method - ScienceDirect

Within limits yes. You can pick the boiling point and then pick groups to get the boiling point (per the group contribution theory)

Yes we do this when adjusting for process changes with prills.

A DSC imo would be the easiest way for you to tune you melt. By using different conditioning times/temps you can seed crystals for more, or less efficient packing. These shifts are only a few degrees though.

Or you could simply ‘blow’ your rxn to adjust you iodine value if its unsaturated. it can certainly be done just depends on the complexity and purity of your compound.
I havent played around with quantum calculations in quite some time but GAMESS is freeware if you have the wherewithal to decipher a bunch of .txt files you could calculate these thermodynamic events

In principle, you could calculate all chemical properties of everything to any precision you’d like using nothing but Schrödinger’s Equation, Maxwell’s Equations, and the charges and masses of all of the nuclei.

In practice, this is usually such a difficult calculation that it no human or human-built computer knows how to do it.

Numerical molecular dynamics simulation to predict materials properties has been a thing for yonks. It’s not trivial, but people who know what they are doing (and there are several approaches to take) do know how to estimate and predict properties like the melting point of a system.

Now, it is not clear how empirical you want to count such simulations. There are definitely approximations involved.

example source code to play with: GitHub - lammps/lammps: Public development project of the LAMMPS MD software package

I appreciate all your replies and maybe appreciate even more the smarts you seem to be attributing to me. Truth is that other than teaching me a new word (mesoscopic) the stuff in the links is far, and I mean far, beyond me. Likewise the practical suggestions.

I guess my primary question was answered most closely by Chronos - it’s theoretically possible, but not gonna happen anytime soon. On the other hand, it sounds like for specific systems and molecules, you can get ‘close enough’ to the answer.

As an aside, am I right in supposing that quantum computers (which I believe are intrinsically parallel), will breeze through this stuff once their technology is mature?

Thanks again.

I am definitely not a chemist, but I have a vague memory that when chemists were filling in the blanks to the periodic tables, they could give estimates of the physical properties of an undiscovered element, depending on the properties of other elements in that particular column, the anticipated nucleus weight (# of proteins and estimates of # of neutrons) and so on. Am I misrembering?

Some properties are easier to estimate than others, and some estimates will be closer than others. For instance, if you have a high-molecular-weight nonpolar substance, it’s probably going to be a waxy solid, and you can take a stab at its melting point based on its molecular weight. But simplistic rules like that are bound to have exceptions, and figuring out what those exceptions are will be hard.

There are ways to compute melting point that involve approaches such as density functional theory and cluster expansion. Researchers have tried to design compounds with high melting point and a few years ago there were reports of some hafnium nitrogen carbide setting record for highest melting point of a substance, even though it was all theoretical.

Very interesting. I had never heard of density functional theory before. Although I understand none of the math and physics it uses, I am impressed that even with its complexity and what I presume are its huge computational demands, it’s still far from a perfect method. I now understand better how intractable the problem is.