Somebody hands you a new compound and asks for its physicochemical properties. You come back with a number. Then a second person asks which number it is, and whether it was measured at pH 7.4, and the conversation stops.
These three get used as though they are interchangeable. They answer different questions, and choosing the wrong one gives you a value that is correct and useless.
The three, in one line each
pKa is the pH at which half of an ionizable group is charged and half is not. It belongs to the group, not to the whole molecule, so a molecule with two ionizable groups has two of them.
logP is the partition coefficient. It describes how the neutral form of the molecule distributes itself between octanol and water. It has no pH attached to it, because the form it describes does not change with pH.
logD is the distribution coefficient at a stated pH. It counts everything present at that pH, ionized and neutral together. A logD quoted without a pH is not a measurement.
How they are related
For a molecule with no ionizable group, logD and logP are the same number at every pH. Nothing is changing, so there is nothing for the pH to do.
For an ionizable molecule, logD equals logP while the molecule is neutral, and falls away as the pH crosses the pKa and the charged form starts to dominate. Charged molecules prefer the water.
For a monoprotic acid:
logD = logP - log(1 + 10^(pH - pKa))For a monoprotic base:
logD = logP - log(1 + 10^(pKa - pH))Two things follow from those expressions, and they are the practical points.
The first is that logP and pKa together tell you logD at any pH you like. If you have measured those two, you do not need to measure logD separately at every pH.
The second is that logD is most sensitive to pH exactly where the pH is near the pKa. Around that point a small error in the pH of your buffer produces a large error in the number you report. Away from it, the curve flattens and small pH errors stop mattering.
Which one your job needs
If you are developing a chromatographic method, you need the pKa. You are choosing a mobile phase pH, and the thing that decides whether your retention is reproducible is how far that pH sits from the pKa of your analyte. Sitting close to it means the ionization state, and therefore the retention, moves with every small variation in buffer preparation, temperature or column age. Two pH units clear of every pKa is the usual target, and you decide it at the design stage.
The pKa you need for that is not the one in the reference table. The literature value is aqueous, and your mobile phase is not. I have written separately about measuring the apparent pKa in your own mobile phase, which is the version that governs your separation.
If you are profiling a molecule for discovery, you need logD at 7.4. That is the pH the molecule meets in blood and in most of the places its behaviour matters. logP would tell you about a form that may barely exist there. For a basic compound with a pKa of 9, logP and logD at 7.4 are about 1.6 log units apart, and it is logD that tracks what the molecule does.
I have written up how to measure logD on an HPLC instead of by shake-flask. That is how I ran it at a CRO.
If the question is about solubility or formulation, you usually need logP and the pKa together, because you need to know both how lipophilic the neutral form is and at what pH it stops being neutral.
The three ways this goes wrong
A logD with no pH on it. I see it in reports and in supplier data sheets. If the pH is not written down, the number cannot be compared with anything.
A literature pKa used inside a hydro-organic mobile phase. Organic modifier moves the apparent pKa, sometimes by more than a unit. A method designed to sit two units clear of a literature pKa may be sitting on top of the real one.
logP quoted for a compound that is charged at every pH you care about. The number is real and it describes a species that is not present in your assay.
The habit underneath all three
The value that governs your experiment is the one measured under your conditions.
That is the same argument I have made about dwell volume, which is a completely different number measured a completely different way, and it holds for the same reason. A published value tells you what to expect. It does not tell you what you have.
Common questions
- What is the difference between logP and logD?
- logP is the octanol-water partition coefficient of the neutral form of a molecule, so it does not depend on pH. logD is the distribution coefficient of all forms present, ionized and neutral, at a stated pH. For a molecule with no ionizable group they are the same number.
- How do you calculate logD from logP and pKa?
- For a monoprotic acid, logD = logP − log(1 + 10^(pH − pKa)). For a monoprotic base, logD = logP − log(1 + 10^(pKa − pH)). logD falls away from logP as the pH crosses the pKa and the charged form dominates.
- Which number do I need for HPLC method development?
- The pKa. Retention of an ionizable analyte is reproducible when the mobile-phase pH is about 2 units away from every pKa. Use the apparent pKa in your hydro-organic mobile phase, because organic modifier can move it by more than a unit from the aqueous literature value.
- Why is logD usually quoted at pH 7.4?
- pH 7.4 is the pH of blood, where the molecule's behaviour matters most in discovery. A basic compound with a pKa of 9 has a logD at 7.4 about 1.6 log units lower than its logP.
- Is a logD value without a pH meaningful?
- No. logD changes with pH, most steeply near the pKa, so a logD with no pH cannot be compared with anything.