TLC and HPLC See Different Impurities: Why It Matters
Share
A certificate that states "no impurity detected, thin-layer chromatography" is making a precise claim, and it is almost always read as a broader one. The claim is that nothing appeared above the method's threshold. The reading is usually that nothing is there. Between those two sentences sits everything that makes analytical chemistry commercially interesting.
None of this is an argument against thin-layer chromatography. It is an argument for knowing which question a method was built to answer before treating its answer as the answer to a different one.
Two methods, two purposes
A thin-layer limit test applies the sample and a dilute reference solution to a plate, develops it, visualises the spots and compares. If the sample's impurity spots are not more intense than the reference, the batch passes. The output is a comparison, not a concentration.
An HPLC related-substances method separates the components on a column and reports each as an area percentage, or as a mass percentage against a standard. The output is a set of numbers with individual limits and a total.
The first is a decision. The second is a measurement. Both are legitimate, and they are not interchangeable.
What the threshold really is
| Method | Typical detection capability | What the report means |
|---|---|---|
| TLC limit test | 0.1–0.5 per cent of the main spot | Above the reference spot or not |
| HPLC, area normalisation | 0.02–0.05 per cent, method dependent | Percentage of total peak area |
| HPLC, external standard | 0.02–0.05 per cent, with a validated limit of quantitation | Mass percentage against a standard |
The gap between the first row and the second is the source of most misunderstandings. A batch with 0.25 per cent of a related substance will pass a TLC limit test set at 0.3 per cent and will show 0.25 per cent on an HPLC profile. Both reports are accurate. Only one of them will satisfy a customer who asked for an impurity profile.
The second consequence is cumulative. Three impurities at 0.08 per cent each are invisible on a TLC plate and total 0.24 per cent on a chromatogram. A limit test cannot detect that sum, because it never sees the individual spots.
Where TLC is genuinely better
Three situations. Identification, where a comparison of retention and visualisation against a reference is robust and needs no assigned potency. Screening of raw materials, where the question is whether something unexpected is present and the answer needs to be cheap. And field or incoming testing at a site without a chromatography data system.
There is also a fourth, less discussed: TLC is less sensitive to the sample matrix. A heavily coloured or salty sample that ruins an HPLC column can be run on a plate without pretreatment. For an early-stage material that is a real advantage.
Where TLC quietly fails
Resolution of closely related structures. Bile acids differ by the position and orientation of hydroxyl groups, and those differences produce small changes in retention factor on a plate. A system that cannot separate chenodeoxycholic acid from ursodeoxycholic acid will report one spot where a chromatogram would report two, and the fact that the plate looks clean is not evidence of purity.
This is where the 7-keto related substances case is instructive. The compounds in a bile-acid family are structurally close, and a method that resolves them has to be chosen for that job rather than inherited.
How a limit test should be described
Precisely. "No impurity spot more intense than the reference solution (0.3 per cent)" is a complete statement. "No impurities detected" is not, because it omits the threshold and invites the reader to assume zero.
The same discipline applies to the reference solution's preparation and to the amount applied. A limit test is only as good as its stated limit, and a certificate that omits the number has removed the only part of the statement a buyer can use.
Bridging the two
Where a customer's specification is chromatographic and the supplier's is a limit test, the bridge is a comparison on the same lots, chosen to include material near the limit and material known to contain the impurity. The question is not whether the numbers match, because they never will. It is whether the two methods ever disagree about acceptability, and where.
If they never disagree on twenty lots, the limit test is adequate for that specification and the conversation can move on. If they disagree on one lot in twenty, that lot is the most valuable sample in the study and the reason should be found before the next shipment.
The comparison is also the only defensible basis for substituting one method for the other, which is the situation most customers actually face when they change supply. The limits themselves are the other half of that conversation, because a limit set for one method's capability may be unachievable by another's.
Where this leaves a supply-chain partner
We state which method produced each impurity line and, where the method is a limit test, we state the limit. Where a customer needs a chromatographic profile because their file requires one, we say whether that method exists and has been validated, rather than relabelling a plate result as a profile.
A single question resolves most of these cases: what was the reference spot, and at what percentage? A supplier who can answer has understood their own method. A supplier who cannot has been reporting a clean plate as a purity claim, and the difference between purity and assay is the next distinction that conversation will need.
Designing a bridging study that means something
Bridging two impurity methods is not a comparison of numbers, because the two methods do not produce the same kind of number. It is a comparison of conclusions. The design follows from that: choose lots that span the decision boundary, and ask whether the two methods ever disagree about acceptability.
A workable set has five members. A lot well inside the limit. A lot close to the limit from below. A lot that just fails. A lot known to contain the specific impurity of interest at a known level. And a lot with an unusual matrix, if one exists, such as material from a different synthetic route. Run both methods on all five and tabulate the conclusions rather than the values.
What the study is looking for is the direction of any disagreement. If the limit test passes a lot that the chromatographic method fails, the limit test is not sensitive enough at that boundary. If the reverse happens, the limit test is rejecting material that is acceptable, which is a commercial problem rather than a safety one but still a problem.
Writing a limit test down properly
A limit test is defined by four numbers: the amount of sample applied, the concentration of the reference solution, the detection method and the acceptance statement. Omit any one and the test becomes a habit.
The phrase "no impurity detected" omits all four. It also makes a claim the method cannot support, because absence below a threshold is not absence. The defensible phrasing names the threshold in the same sentence: no impurity spot more intense than the reference solution corresponding to 0.3 per cent.
The same discipline applies to identification, where the question is different and the reporting standard is looser. The identification tests can legitimately be reported as compliant, because the question is whether the material behaves like the reference. Impurity limits cannot, because the question is how much, and a comparison answers that only in brackets.
The method definitions themselves are published and stable: EDQM for the European monographs, USP for its own, and the ICH Q3A framework for how impurity limits are set in the first place. A supplier whose limit test threshold is looser than the limit in the specification has a method that cannot fail the specification, which is not a control.
The question to ask is one sentence long: what is the reference spot, and at what percentage? Everything else in a limit test follows from that answer.
An illustrative five-lot comparison
The table below shows the shape of a bridging study. The values are illustrative and describe the pattern rather than any particular material.
| Lot | HPLC, largest impurity | TLC result | Conclusion agrees? |
|---|---|---|---|
| A | 0.04 per cent | No spot above reference | Yes |
| B | 0.12 per cent | No spot above reference | Yes |
| C | 0.28 per cent | Faint, below reference | Yes, with margin |
| D | 0.45 per cent | Faint, at reference | Borderline |
| E | 0.10 per cent, two spots | No spot above reference | Yes, but sum is invisible |
Lots A through C support the limit test: the two methods agree about acceptability with margin. Lot D is the interesting one, because it sits exactly where the limit test runs out of resolution, and a laboratory repeating it could go either way on a different plate. Lot E is the second lesson: two impurities at 0.05 per cent each are invisible individually and total 0.10 per cent, which a limit test cannot see no matter how well it is run.
A study with this shape justifies keeping the limit test for routine release and using the chromatographic method at a defined frequency, or at defined triggers such as a process change. A study where lots C and D disagree about acceptability justifies the opposite arrangement.
Frequently asked questions
How sensitive is a thin-layer limit test?
Typically in the region of 0.1 to 0.5 per cent of the main spot, depending on the detection method and the amount applied. It is a limit test, so it is designed to answer whether a substance is present above that threshold, not to quantify what is present below it.
Why do monographs still include TLC identification?
Because it is cheap, robust, and does not require a reference standard with an assigned potency. For identification the question is whether the substance behaves like the expected one, and a comparison of retention and visualisation answers that question well enough.
Can TLC replace HPLC for related substances?
Only where the specification is a limit test and the threshold is adequate for the risk. It cannot produce an area-percentage profile, and it cannot resolve closely related substances that differ only in a hydroxyl position without a system that has been shown to separate them.
What does not detected mean on a TLC report?
That no spot was visible above the threshold of the method, usually the level of the reference spot applied alongside. It does not mean the substance is absent, and it does not mean the total impurity content is below the threshold, because several impurities each below the threshold can sum to more than it.
Is HPLC always the better choice?
No. HPLC costs more per test, requires a validated method and a system suitability check, and at 210 nm it is sensitive to mobile phase and column variability. For a screening limit test on a raw material, TLC is often the more appropriate tool and the more reproducible one across sites.
How do you bridge the two methods?
Run both on the same lots, including lots that sit near the limit and lots known to contain the impurity of interest, then compare the conclusions rather than the numbers. The question is whether the TLC limit test and the HPLC profile ever disagree about a batch's acceptability, and if so, why.