If anyone could clear up some of this conceptual confusion, that would be much appreciated.

  1. Is $\mathrm pK_\mathrm a$ an intrinsic property of a molecule? Or extrinsic? A property is extrinsic if it depends on the environment (e.g. weight), and intrinsic if it is independent of the environment (density). Source: Thoughtco: Intrinsic Property Def.

  2. Doesn't $\mathrm pK_\mathrm a$ change according to environment/conditions the molecule is in? According to the definition, $\mathrm K_a = [\ce{H^+}][\ce{A^-}]/[\ce{HA}]$. Then, given a $\mathrm pK_\mathrm a$ table, what are all the $\mathrm pK_\mathrm a$ values calculated with reference to? What are the standard conditions in which scientists calculate their $\mathrm pK_\mathrm a$ values.

  3. Is there a way to calculate $\mathrm pK_\mathrm a$ from known intrinsic properties of a molecule? E.g. if I give you one molecule of a compound, and only one molecule, is it possible to calculate $pK_a$ based off the intrinsic structure of that molecule, given no other information?

  • $\begingroup$ It's not a property of molecule at all. Just some random equilibrium constant of highly limited use. $\endgroup$ – Mithoron Dec 9 '18 at 1:40
  • 1
    $\begingroup$ Well, given that it's related to the equilibrium constant of a reaction the molecule takes part in, I would say it's as much of an 'intrinsic' property as equilibrium constants in general are (they depend on the temperature and a few other things). In this case, the other reactant is water, present in excess as the solvent; perhaps that's why we tend to think of $pKa's$ as properties 'of the molecule'. They are in fact properties of the acid+water system within the specified conditions. $\endgroup$ – user6376297 Dec 9 '18 at 15:13
  • $\begingroup$ The tables you find in the literature are just experimentally determined values of $pKa's$ in water; I don't see what 'reference' you mean. The temperature is usually mentioned. $\endgroup$ – user6376297 Dec 9 '18 at 15:14
  • $\begingroup$ As for the 'prediction' of $pKa's$, there are in silico tools that are able to calculate them, usually based on QSAR models. In essence, a lot of $pKa$ data were examined, and the structural features of the molecules were associated with their contribution to the $pKa$ value. This was the 'training' of the model. Then the model becomes like a function taking as input the structural features, and producing as output a prediction of the $pKa$, with various degrees of accuracy depending on how good the training set and the model itself were. So yes, it is possible. $\endgroup$ – user6376297 Dec 9 '18 at 15:20
  • 1
    $\begingroup$ BTW, I am fairly shocked by the above statement that $pKa's$ are of 'highly limited' use. In my day-to-day work (computational chemistry, drug discovery), $pKa's$, either predicted or experimental, are essential to many of our activities, and have an enormous impact on many properties of biopharmaceutical interest. Not to mention the application in synthetic chemistry, where the choice of a base is fundamentally down to its ability to deprotonate certain groups (sometimes selectively). $\endgroup$ – user6376297 Dec 9 '18 at 15:25