Accepted answer
Two of your three assumptions are roughly right and the temperature one is wrong — and on tirzepatide specifically it is wrong in a way that can move the reported purity by more than a point. Working through the four choices in order of how much they matter for this molecule.
1. Temperature, which matters most and is treated as least important
Tirzepatide's C-terminal region is Gly-Ala-Pro-Pro-Pro-Ser-amide. Three consecutive prolines. Proline amide bonds interconvert between cis and trans configurations slowly — the barrier is around 20 kcal/mol, giving interconversion half-lives on the order of seconds to tens of seconds at room temperature. That is the same order as the width of a chromatographic peak.
When conformational interconversion is slow relative to the separation, the two conformers behave as two partly resolved species and you get a broadened, shouldered or frankly split peak. When it is fast, you get one averaged peak. Temperature controls which regime you are in.
Practical consequence, and this is the important bit:
- At 25 C the conformer shoulder is partly resolved. An analyst integrating it as a separate peak reports it as an impurity, and the purity figure drops by however much area it carries — commonly 0.5 to 1.5% on a tri-proline sequence.
- At 55 to 60 C interconversion is fast, the shoulder collapses into the main peak, and the purity figure rises.
Neither number measures an impurity, because the shoulder is the same molecule — a conformer, not a related substance. A low purity figure from a room-temperature method on tirzepatide may be reporting a molecule against itself. The correct method runs hot enough to average the conformers, and you demonstrate you are in that regime by running two temperatures and showing the shoulder collapse. Five minutes of work, and it belongs in any method development on a proline-rich peptide.
Heat also lowers eluent viscosity, giving narrower peaks and lower backpressure, speeds pore mass transfer, and removes laboratory ambient temperature as a variable. The cost is on-column degradation over long gradients, so above about 60 C check that a re-injected sample gives the same profile.
Specify 45 to 60 C and ask for the two-temperature comparison. Of everything in this answer, that is the request most likely to change your number.
2. Stationary phase and pore size
Pore size matters more than the ligand at 4.8 kDa. On 100 A silica a peptide this size diffuses into and out of the pores slowly relative to the separation, which broadens peaks — the restricted-diffusion penalty. A 300 A packing fixes it and noticeably narrows peaks on anything above about 3 kDa.
| Phase | Behaviour on a 4 to 5 kDa acylated peptide | Use when |
|---|
| C18, 100 A | Very retentive; needs high organic to elute the fatty side chain; recovery losses possible | Default for small peptides; adequate but not optimal here |
| C18, 300 A | Best general choice: good retention with faster mass transfer and narrower peaks | The sensible default for these molecules |
| C8 or C4, 300 A | Less retentive, elutes at lower organic, often better recovery for very hydrophobic species | When a C18 method shows poor recovery or a very late, broad main peak |
| Phenyl-hexyl or diphenyl | Different selectivity via aromatic interactions; reorders aromatic-containing impurities | As the second method of an orthogonal pair |
| Charged-surface hybrid C18 | Improves peak shape for basic peptides without TFA | When you need MS compatibility and cannot afford tailing |
The acylated analogues are unusually hydrophobic for their size because of the diacid side chain, so they elute at high organic fraction and a standard small-molecule C18 method has them coming off late and broad. That argues for a wide-pore C8, or wide-pore C18 with a gradient reaching 60 to 70% acetonitrile.
3. Mobile phase acid or buffer
Your assumption that TFA is standard because it gives sharp peaks is correct, and the mechanism and the trade-offs are set out in a separate answer below. The short version for specification purposes: TFA for the primary purity method, phosphate at pH 2.5 as a confirmatory method when the result matters commercially, formic acid only when you need the mass spectrometer on the same injection and are willing to accept worse peak shape and a slightly flattering purity figure.
What I would actually specify
- Column: C18 or C8, 300 A pore, 150 x 4.6 mm, 3.0 to 3.5 um, or the UHPLC equivalent
- Temperature: 50 C, with a documented comparison at 25 C to demonstrate conformer averaging
- Mobile phase: 0.1% TFA in water and 0.085% TFA in acetonitrile for the primary purity method; a phosphate pH 2.5 method as the confirmatory one if the result matters commercially
- Gradient: not steeper than 0.6 %B per minute through the region where the main peak elutes
- Detection: 214 nm primary, with 280 nm and 320 nm channels recorded from the diode array
- Reporting threshold: 0.05%, with the signal-to-noise at that level stated
- Load: chosen so the main peak stays inside the detector's linear range
- System suitability: resolution not less than 1.5 against the nearest impurity, tailing not more than 1.5, replicate area RSD not more than 1.0%, bracketing standard within 2%
That is a specification a competent contract lab will recognise and can quote against. It will also produce a lower purity number than their default, for all the reasons in the other threads in this tag — which is the point of specifying it.
edited 25 Jun 2026 by gunnar_isaksen — tightened the wording; no substantive change
The proline conformer point is the best thing I have read about tirzepatide chromatography. It explains a split peak I spent a week chasing. – stopper_core 5 months ago Asking for the two-temperature comparison as a deliverable is a great idea. Cheap and it settles the conformer question outright. – loss_on_drying 4 months ago 3Wide-pore for anything above 3 kDa is the rule I wish came printed on the column box. – Dr_Yusuf_Adeyemi 2 months ago add a comment