Why Cross-Check These Properties?
A reported pKa, logD, and solubility value may each look reasonable on its own but still be incompatible with the other measurements. That matters in PBPK work because these properties describe related consequences of ionization: pKa helps determine the neutral fraction at a given pH, while logD and total solubility can change with that fraction. A mismatch is a reason to investigate the data and conditions, not automatically a reason to discard a measurement.
The checks here use ideal relations for a monoprotic compound. Earlier material on pKa, ionization fractions, logD, and pH-dependent solubility established their basic definitions. The focus here is to use those relations jointly as diagnostic constraints.
For a monoprotic acid, the neutral form is \(HA\), and for a monoprotic base, the neutral form is \(B\). The following ideal checks assume equilibrium, the same compound form and temperature across measurements, compatible pH and concentration definitions, and a solution in which the relevant species behave ideally enough for the standard relations to apply. In the simplest distribution model, ionized species contribute negligibly to octanol partitioning. In the simplest solubility model, a solid neutral form controls equilibrium, with no salt precipitation, additional solid forms, or other solubility-limiting processes.
Two Cross-Checks From Ionization
For an acid, the neutral fraction is \(1/(1+10^{\mathrm{pH}-\mathrm{p}K_a})\); for a base, it is \(1/(1+10^{\mathrm{p}K_a-\mathrm{pH}})\). Under the ideal distribution assumption, only that neutral fraction partitions appreciably into octanol. Therefore, \(D\) is \(P\) multiplied by the neutral fraction. Rearranging gives an implied \(\log P\) from each reported pH-specific \(\log D\):
If pKa and the conditions are consistent, \(\log P_{\mathrm{implied}}\) should be approximately constant across pH-specific measurements. “Approximately” is important: experimental error, rounding, and nonideal partitioning can cause modest differences. A large, systematic change is a warning that at least one reported value or assumption needs checking.
The corresponding ideal total-solubility relations are \(S=S_0\) multiplied by the reciprocal of the neutral fraction. Thus, each solubility measurement implies a value of \(S_0\). Measurements made at different pH values should imply similar intrinsic solubilities if the model and conditions apply.
These equations describe trends, not a guarantee that real measurements follow them exactly. For example, the ideal weak-acid model predicts increasing total solubility as pH rises above pKa; the ideal weak-base model predicts decreasing solubility as pH rises above pKa. Comparing the direction and approximate magnitude of the observed changes with these predictions is a useful first diagnostic.
Worked Example: Acid logD Values Imply Different logP Values
Suppose a hypothetical monoprotic acid has reported pKa \(=4.50\). Its reported \(\log D\) is 3.00 at pH 2.50 and 2.20 at pH 6.50. Assess whether the pair is consistent with the ideal model. Assume the measurements refer to the same compound form and temperature.
At pH 2.50, the acid correction factor is \(1+10^{2.50-4.50}=1+0.0100=1.010\). The implied neutral partition coefficient is therefore:
At pH 6.50, the correction factor is \(1+10^{6.50-4.50}=101\), so:
The implied values differ by 1.200 log units, or a factor of about \(10^{1.200}=15.8\) in \(P\). That is a substantial discrepancy under the stated ideal assumptions. If the lower-pH value were correct and \(\log P\) were about 3.004, the predicted \(\log D\) at pH 6.50 would be \(3.004-2.004=1.000\), not 2.20. The set merits verification; possible issues include a pH or sample mismatch, an incorrectly assigned pKa, or appreciable partitioning by ionized species.
Check Solubility Against pH and pKa
A solubility cross-check uses the same pKa and the pH at which solubility was measured. For each observation, calculate the predicted ionization multiplier and divide the reported total solubility by it to obtain \(S_{0,\mathrm{implied}}\). Compare those inferred values across measurements. Use consistent solubility units throughout, such as mg/mL or mol/L; a unit conversion error can resemble a large chemical inconsistency.
Worked Example: A Weak Base Has Incompatible Solubility Reports
A hypothetical weak base has pKa \(=8.00\). Reported equilibrium total solubilities are 0.200 mg/mL at pH 6.00 and 0.0200 mg/mL at pH 10.00. Suppose both measurements are claimed to represent the same temperature, compound form, and neutral solid phase. Calculate the intrinsic solubility implied by each.
At pH 6.00, the base multiplier is \(1+10^{8.00-6.00}=101\). Thus:
At pH 10.00, the multiplier is \(1+10^{8.00-10.00}=1.010\). Thus:
The inferred values differ by a factor of \(0.0198/0.00198=10.0\). The pH 6.00 result implies much lower intrinsic solubility than the pH 10.00 result. Under the stated assumptions, these reports do not support one common \(S_0\). Verify the units, pH values, equilibration and solid phase, and whether the reported quantity is total dissolved compound rather than a filtered or otherwise operationally defined fraction.
Joint Constraint Linking logD and Solubility
The two ideal relations can be combined. At a fixed pH, \(D=P f_0\), where \(f_0\) is the neutral fraction, and \(S=S_0/f_0\). Multiplying eliminates the pH-dependent neutral fraction:
This derivation shows why pH-specific logD and solubility can provide a joint diagnostic: their product should be pH-independent under the ideal model, provided the same units and conditions are used for the solubility measurements and the same reference definitions apply. It does not mean that logD or solubility is individually constant with pH. Also, the numerical value of \(\log S\) depends on the concentration unit; compare products or log sums only when solubility units are the same.
Worked Example: A Coherent Acid Property Set
Consider a hypothetical monoprotic acid with pKa \(=5.00\), \(\log P=3.00\), and \(S_0=0.0100\) mg/mL. Use the ideal assumptions above to predict \(\log D\) and total solubility at pH 3.00 and pH 7.00, then check the joint constraint.
At pH 3.00, the acid multiplier is \(1+10^{3.00-5.00}=1.010\). Therefore, the neutral fraction is \(1/1.010\), and:
At pH 7.00, the multiplier is \(1+10^{7.00-5.00}=101\), giving:
Using the same units for both solubilities, the products are \(10^{2.996}\times0.0101\approx10.0\) and \(10^{0.9957}\times1.01\approx10.0\), in the corresponding \(D\)-times-mg/mL units. Equivalently, \(\log D+\log_{10}(S\text{ in mg/mL})\) is about 1.000 at both pH values. The property set is internally coherent under the ideal model.
Diagnosing Mismatches Without Overinterpreting Them
A failed check is a diagnostic signal, not proof that a particular value is wrong. Before attributing a discrepancy to chemistry, confirm that the measurements refer to the same chemical identity, salt or free form, temperature, and relevant pH. Check that pKa and pH use compatible conditions, that logD is reported at the stated pH, and that solubility units and definitions match. Confirm whether the solubility value is equilibrium total solubility and whether the sample was fully dissolved or affected by precipitation.
The ideal relations may also fail because their assumptions are too simple for the compound or assay. Ionized species may partition into octanol; multiple ionization steps may matter; measured apparent pKa can depend on conditions or chemical form; and solid-state changes, salt formation, aggregation, or supersaturation can alter observed solubility. A multiprotic compound should be assessed using its relevant charge-state distribution, not forced into a monoprotic equation. Differences of a few tenths of a log unit may be compatible with measurement uncertainty or model simplification, whereas a large disagreement should prompt targeted review.
This is a consistency screen, not a method for selecting a “correct” measurement from conflicting reports. Depending on the evidence, retain uncertainty, seek conditions-matched measurements, or document a provisional input and its rationale. In particular, do not infer that a single value is reliable merely because it agrees with one other measurement; check all available conditions and definitions.
A useful alternative to this calculation-based screen is to assess each property separately against its own assay and uncertainty, without imposing a joint ideal model. That approach can be appropriate when assay conditions differ or the compound violates the assumptions, but it may miss a clear internal contradiction. The cross-check is most informative when conditions are matched; separate assessment is more cautious when they are not. Neither approach removes the need to inspect the measurement definitions and chemical form.
Confirm chemical form, temperature, pH, units, and whether the reported values describe the same compound and measurement basis.
Use pKa and pH to infer \(\log P\) from each logD value and \(S_0\) from each total-solubility value.
Assess the size and pattern of differences against measurement uncertainty and model assumptions; document unresolved discrepancies rather than silently forcing agreement.
Check Your Understanding
Use the ideal monoprotic relations where applicable. Give your reasoning, and then compare with the solutions below.
- A weak acid has pKa 4.00. At pH 6.00, what is the ideal multiplier relating total solubility to \(S_0\)?
- A base has pKa 7.00 and \(\log D=1.00\) at pH 7.00. What \(\log P\) does the ideal model imply?
- Two solubility measurements imply \(S_0\) values of 0.010 and 0.011 mg/mL under matched conditions. Is this exact agreement? Is it plausibly close enough to be consistent, depending on measurement uncertainty?
- For the same compound, logD changes with pH while total solubility changes in the opposite direction. Does that alone establish an inconsistency? Explain what additional check is useful.
Solutions. (1) The acid multiplier is \(1+10^{6.00-4.00}=101\). Thus the ideal total solubility is 101 times \(S_0\). (2) For a base at pH equal to pKa, the correction is \(\log_{10}(1+10^0)=\log_{10}(2)=0.3010\). Hence \(\log P_{\mathrm{implied}}=1.00+0.3010=1.301\). (3) They are not identical, but they differ by about 10%: \(0.011/0.010=1.10\). Whether that is compatible depends on measurement uncertainty and the assumptions; it is not by itself compelling evidence of a major contradiction. (4) No. Opposing trends can be expected from the ideal ionization relations. Calculate implied \(\log P\) and \(S_0\) at each pH, and compare the values while checking that the measurements and conditions are comparable.