Predictive Abilities of Chemistry

[Jim Lundquist]

Hypothetically, if a chemist/physicist knew all the properties of hydrogen and all the properties of oxygen but knew nothing of water or its properties, would he be able to predict that combining 2 atoms of hydrogen and one atom of oxygen would produce a substance called water and its properties?

 

[Vanadium 50]

He certainly wouldn't know it was called water.

Water has lots of properties. Which ones are you talking about?

 

[BvU]

Hypothetically, if a chemist/physicist knew all the properties of hydrogen and all the properties of oxygen

Hypothetically: chemist: no chance. Physicist: a little

The dipole moment in particular has enormous consequences.

Nowadays we more or less routinely predict a lot of properties of molecules from quantum mechanics (e.g. with Cosmo)

However, the abundance of water in our environment makes this a very hypothetical question .

 

[Jim Lundquist]

He certainly wouldn't know it was called water.

Water has lots of properties. Which ones are you talking about?

First, of course he wouldn't know it is called water. This is a simply worded hypothetical question. At STP, hydrogen and oxygen are gases while water is a liquid. Could that liquid property be predicted...prior to sophisticated software such as Cosmo? We know that mixing hydrogen and oxygen and adding a spark can create water, and we know that water can be broken down into hydrogen and oxygen through electrolysis, but these processes were discovered. Without the knowledge of these discoveries...just the knowledge of hydrogen and oxygen and the properties of each, could it be predicted that the combination of these two gaseous elements at STP could result in liquid water or hydrogen peroxide? And I don't see what this has to do with the abundance of water as BvU stated. BTW, I'm not some student trying to get an answer for some class. I am just a curious 72 year old.

 

[BillTre]

This is a knowing the properties of the components, what are the properties of a system assembled from them kind of question.
It has to do with not reduction, but synthesis (as in holism).

It would seem to require at least two things of the predictor:
1) an understanding of what similar components do when assembled into similar systems (component and system knowledge)
2) being quite smart about coming up with the (otherwise) unexpected consequences of details of the components.

This is somewhat similar to (but less complicated) predicting how a biological entity might evolved in an environment.
In neither case would I expect much success without #1, and due to the complexity of biology I would not expect much success in biological predictions beyond the trivial.

 

[Jim Lundquist]

This is a knowing the properties of the components, what are the properties of a system assembled from them kind of question.
It has to do with not reduction, but synthesis (as in holism).

It would seem to require at least two things of the predictor:
1) an understanding of what similar components do when assembled into similar systems (component and system knowledge)
2) being quite smart about coming up with the (otherwise) unexpected consequences of details of the components.

This is somewhat similar to (but less complicated) predicting how a biological entity might evolved in an environment.
In neither case would I expect much success without #1, and due to the complexity of biology I would not expect much success in biological predictions beyond the trivial.

Thank you for your reply. I agree with your biological analogy (much more complex), but this is a thought problem in which the predictor has absolute knowledge of the components (hydrogen and oxygen), and absolutely NO knowledge of the system (water). The missing component would be the variable of the introduction of an electric spark, but that may be considered when contemplating how to get hydrogen and oxygen to achieve the polar covalent bond. Once that bond is achieved (at least on paper), can the liquid property at STP of that molecule be predicted?

 

[BvU]

prior to

Oh sorry, I missed that in the original posting. Any further restrictions ?

 

[BvU]

And I don't see what this has to do with the abundance of water

Understandable. I had trouble mentally moving into the frame of the hypothesis. So many relevant properties, all so well investigated.
After all, there are more liquids with a high dipole moment (but slightly less healthy ).

 

[Vanadium 50]

The fact that at STP water is liquid is easy to predict. That is a consequence of the 109.5 degree angle between the hydrogens. The exact freezing and boiling points are harder, as is the density decrease in ice, because those are consequences of the fact that 109.5 degrees is actually 104.5 degrees. If you want a crisper answer, though, you need to ask a crisper question. Exactly what properties?

 

[TeethWhitener]

combining 2 atoms of hydrogen and one atom of oxygen would produce a substance called water and its properties

If you feed a quantum chemistry program two hydrogen atoms and an oxygen atom and tell it to optimize the geometry, it will quickly converge on a very close approximation to a water molecule. If you fix the internal geometries of two water molecules and map out a potential energy surface, you can get a decent enough force field to start doing molecular dynamics to nail down certain bulk properties.

The fact that at STP water is liquid is easy to predict. That is a consequence of the 109.5 degree angle between the hydrogens.

This tells you there's a dipole moment, but not necessarily anything else. You really need to know that extensive hydrogen bonding exists. E.g., the HNH angle in ammonia is 107.8, and ammonia is a gas at STP. Even the dipole moment itself isn't necessarily indicative. E.g., HF has a dipole moment of 1.82 debye (cf. water @ 1.84 debye), and HF boils at just under room temperature (19 C or so), whereas water boils at 100C. Even the dipole moment and the geometry aren't necessarily indicative. E.g., sulfur dioxide, with an OSO angle of 119 and a dipole moment of 1.63 debye, with a boiling point of -10C.

 

[Borek]

In general, as Vanadium 50 wrote, it boils down to what you mean by "properties". Finding properties of a single molecule (angle, bond lengths, bond energies, orbitals, vibrational modes, excitation energies and so on) is reasonably easy. This is a simple molecule, as TeethWhitener wrote quantum chemistry programs do these calculations reasonably fast on a typical PC. Things get much more complicated when you are interested in properties of bulk water, as it requires doing calculations for systems consisting of many molecules*, and that's computationally much more intense.

*back in eighties a fellow student did his MSc trying to find out how many molecules can be considered "bulk". IIRC in his systems "bulk" started at around 100 items.

 

[BvU]

I remember Jurgen Rarey (Oldenburg, DDBST) stating: 'a million molecules doesn''t make a liquid yet'
This seems to agree.

 

[Borek]

I remember Jurgen Rarey (Oldenburg, DDBST) stating: 'a million molecules doesn''t make a liquid yet'

Chances are it again depends on what kind of properties one looks for. Some can be evident for smaller numbers, some may need much larger ones.

 

[Nik_2213]

IMHO, they'd need to know a lot about the modelling, as water is *weird*. The umpteen ice phases, dielectric properties, clathrate formation etc etc...

Oh, and ice-cubes float at STP.
D'uh, who ordered that ??

 

[BvU]

Hypothetically, if a chemist/physicist knew all the properties of hydrogen and all the properties of oxygen

It's a hypothetical knowall physicist doing the modeling ...

 

[DrDu]

This is a difficult question and there are people saying that this is not possible. The point is the following: Speaking of a liquid is already an idealisation which strictly speaking has only a meaning for an infinite amount of substance. However, if you take the limit of the amount of water molecules to go to infinity gravitation will take over and the water molecules will collapse into a star. So if you would have never heared of liquids, you would have a hard time to predict their properties from elementary equations. The same holds for things like molecular geometry. The geometry of a water molecule with 109.5 degree bond angles is not something you can derive from the equations of quantum mechanics without further input from quantum mechanics. It involves an approximation called the Born - Oppenheimer approximation which rests on the nuclei being much more heavy than the electrons and whose mathematical properties are rather awkward.
I once saw a report where the picture of a flame was projected onto the back of a blind woman using an array of fine needles. She was absolutely astonished that a flame actually has some shape. This shape is some idealisation and you have a hard time to derive it from the equations governing combustion.
The new concepts arising as idealisations which are obtained by taking some limit are called emergent properties.

 

[Vanadium 50]

in his systems "bulk" started at around 100 items.

I'm surprised that 100 worked that well. At 1000 units, about half are on the surface. At 100, they are essentially all surface. People who study "clusters" - aggregations of atoms or molecules with relatively small numbers of atoms (hundreds or thousands, not Avagadro's numbers) see very different properties until most items are not at or even near the surface. That's usually many thousands. (At about 1000, it's half and half)

In the case in question, it is not too difficult to show that liquid water has a higher density than hexagonal ice. It is probably much more difficult to show that hexagonal ice is the stable form at room-like temperatures and pressures. The latter property is not one of oxygen and hydrogen atoms but the properties of bulk ice. Worse, the energetics of the hexagonal structure are close to that of the diamond structure. That means a million units might not tell you what the structure of bulk ice is. (Then again, it may not matter since the densities are pretty close)

 

[Ygggdrasil]

If you feed a quantum chemistry program two hydrogen atoms and an oxygen atom and tell it to optimize the geometry, it will quickly converge on a very close approximation to a water molecule. If you fix the internal geometries of two water molecules and map out a potential energy surface, you can get a decent enough force field to start doing molecular dynamics to nail down certain bulk properties.

Are those force fields derived from empirical data on compounds like water or are they built solely from first principles?

 

[TeethWhitener]

Are those force fields derived from empirical data on compounds like water or are they built solely from first principles?

You can do either. The most accurate models do both. The TIPS models used in CHARMM combine the geometry and dipole moment derived from first principles with long-range Lennard-Jones parameters that are empirically derived.
https://en.m.wikipedia.org/wiki/Water_model
In principle, you could get approximate LJ parameters from first principles, but even ab initio quantum chemistry methods end up using empirical corrections to deal with dispersion forces and cut down on computational time.