Peptide purification: preparative HPLC, ion exchange and the yield versus purity trade-off

Synthesis makes the peptide. Purification decides whether you can use it. A crude peptide coming off a resin is a mixture of the target and a population of near-identical failures: chains missing a residue, chains with an extra one, epimers, oxidised and alkylated side chains, and truncated fragments left by capping. None of them can be removed by crystallisation or extraction, because they are too similar to the product. Almost all of them have to be separated chromatographically, and some cannot be: the aspartimide-related by-products formed during Fmoc chemistry include species that co-elute with the target peptide [8].

This guide covers how that is done at preparative and manufacturing scale: reversed-phase chromatography and its additives, ion exchange, HILIC and mixed-mode columns, the yield-purity trade-off and the multicolumn processes that break it, how the counter-ion is set, and how the purified pool becomes a dry powder. For how the crude is made in the first place, see peptide synthesis; for how the finished material is tested, see peptide purity testing by HPLC and MS.

Why purification dominates peptide manufacturing

Purification is not a finishing touch. A 2024 analysis of 40 peptide manufacturing processes by fourteen pharmaceutical companies found that “the downstream processes account directly for approximately 50% of the Process Mass Intensity (PMI), due to the use of dilute product solutions and long chromatographic gradients” [2]. It also identified a second, compounding effect: because the fully synthesised product is what gets handled downstream, “losses in the downstream process need to be offset by increased amounts in the upstream process, thereby increasing overall PMI” [2]. A gain in purification yield therefore also reduces the material that has to be spent upstream.

The solvent profile reflects it. In the purification stage of the surveyed processes, water accounted for 77.3% and acetonitrile 18.4% of solvent use, with ethanol, methanol, isopropanol, methyl tert-butyl ether and acetic acid making up the rest [2]. The ACS Green Chemistry Institute Pharmaceutical Roundtable had already flagged purification alongside synthesis as a sustainability problem for peptides in 2019 [12].

Process stage	Share of material use	Dominant solvent
Synthesis (SPPS)	PMI for synthesis and purification each range from 30 to 70% of the total across development phases [2]	Dimethylformamide, 89.2% of synthesis-stage solvent [2]
Purification	About 50% of total PMI [2]	Water 77.3%, acetonitrile 18.4% [2]
Isolation	Limited: roughly one crude precipitation plus lyophilization [2]	Precipitation solvent, then none

What arrives at the column

Cleavage from the resin is done with an acid cocktail containing scavengers, which trap the reactive carbocations released from the side-chain protecting groups, and the crude product is then separated from the resin by filtration and taken to chromatography [15]. What the column receives is therefore the peptide plus the scavengers, the cleaved protecting groups and their adducts, and whatever salts came with them. Ion-exchange chromatography is sometimes used as a pre-purification step precisely to strip those non-peptide hydrophobic impurities, together with the residual protecting groups and scavengers from cleavage, before the main separation [2].

Preparative reversed-phase chromatography

Reversed-phase liquid chromatography is “the mode of chromatography most frequently encountered when it comes to the purification of peptide and protein mixtures” [1], and preparative chromatography is “the established technology for the purification of most peptide APIs as it is known to be the most efficient approach to separate chemical substances with similar chemical structures” [2].

Stationary phase

C18 ligands are the most widely used, but C8 or even C4 can retain very hydrophobic peptides better [1]. An important limit: these phases distinguish diastereomers, such as peptide epimers, but not enantiomers [1]. That is why a D-amino acid impurity at a single position is sometimes separable and sometimes not, and why diastereomeric impurities appear in specifications as a distinct class.

Gradient, not isocratic

Peptides are large enough that small changes in mobile-phase composition produce large changes in retention, so gradient elution is recommended for macromolecules; it also improves separation from product-related impurities, since structurally similar species can adsorb very differently at a given composition [1]. The standard preparative system is a water and acetonitrile gradient, the most common solvent mixture used for peptide purification [2].

Additives and ion pairing

An acidic ion-pairing agent is almost always present. Trifluoroacetic acid or formic acid pair with the basic residues, which are positively charged at acidic pH, and improve peak shape [1]. The common acidic additives and their acid strengths, from a review of peptide counter-ions, are trifluoroacetic acid, acetic acid, formic acid, phosphoric acid and hydrochloric acid, routinely used at 0.1% by volume, with pKa values at 25 degrees Celsius of 0.52, 4.76, 3.75, 2.16 and about minus 7 respectively [5].

Concentration matters more than it looks. One study found that at trifluoroacetic acid concentrations below 0.05%, “strong ion-exchange interactions can establish between analytes and hydrolyzed silanols, thus causing peak broadening” [1]. Going the other way, 0.2 to 0.25% can give better resolution and less peak tailing, particularly for peptides with several positive charges, but raises the risk of damaging the column, so a balance has to be struck [5]. Some authors suggest matching the additive concentration to the peptide concentration multiplied by the number of positive charges on the molecule, although the same review notes this cannot be applied to peptides carrying more than three positive charges [5].

Except for phosphoric acid, all of these acids can be removed by lyophilization after purification, so no separate desalting step is needed [5]. That convenience is the main reason trifluoroacetic acid became the default, and the reason most synthetic cationic peptides end up as trifluoroacetate salts [5].

Loading

Published preparative protocols on analytical columns and instrumentation have handled sample loads up to 200 mg for peptides and recombinant proteins [4]. A one-step preparative method in the chromatography review purified samples from 145 to 900 mg of peptide mixture, including peptides carrying non-proteinogenic substituents such as biotin and carboxyfluorescein, with purities “almost always above 95%” [1]. At manufacturing scale, both high-pressure and low-pressure chromatographic units are used, and both scale from laboratory to production [2].

Orthogonal steps: ion exchange, HILIC and mixed mode

One column rarely does the whole job. “Most of the manufacturing processes are based on an orthogonal strategy for purification (depending on the impurity profile of the crude and its batch-to-batch variability): 2 to 3 chromatographic steps are combined (RP and/or IEC)”, and purification of peptide active substances typically uses reversed-phase, ion-exchange and frequently size-exclusion chromatography [2]. The point of orthogonality is complementarity: the steps should exploit different properties, such as isoelectric point and hydrophobicity [2].

Mode	Separates on	Where it fits	Evidence
Reversed-phase (RP)	Hydrophobicity	Main polishing step; couples to MS	[1] [2]
Ion exchange (IEX)	Net charge at the mobile-phase pH	Orthogonal polishing; pre-purification to strip hydrophobic non-peptide material, protecting groups and scavengers; also sets the counter-ion	[1] [2]
Size exclusion (SEC)	Hydrodynamic size	Frequently used; separates high molecular weight species	[2]
HILIC	Hydrophilicity, in reverse order to RP	Polar peptides and modified residues	[1]
Mixed mode (RP or HILIC plus IEX)	Two properties at once	Impurities matching the product on one axis	[1]

Ion exchange suits peptides because most have a net charge that moves with pH; at acidic pH, carboxylates are protonated and the basic residues arginine, histidine and lysine are charged [1]. It also distinguishes species of similar hydrophobicity: with a strong cation-exchange column and a salt gradient, bradykinin variants differing only slightly in hydrophobicity or by a single charge could be separated [1]. In one study of a synthetic mixture of peptides carrying charges from plus one to plus seven at pH 3, elution order followed charge monotonically [1]. The interaction can also be made repulsive rather than attractive, depending on the nature of the ion exchanger, the mobile-phase pH and the peptide’s isoelectric point [1].

Mixed-mode phases combine two mechanisms on one particle. In one preparative evaluation of C8 columns functionalised with quaternary amines, selectivity and productivity rose up to twofold and yield improved by around 20% [1]. A strong cation-exchange phase with hydrophilic character separated a 21-residue amphipathic helical peptide from impurities acetylated on three different serine residues, species with identical charge and extremely similar hydrophilicity [1].

The yield versus purity trade-off

This is the central constraint of preparative peptide chromatography, and it is worth stating precisely. The usual situation is a ternary separation: the product elutes between a group of weakly retained impurities and a group of strongly retained ones [1] [6]. Because related impurities resemble the product, their peaks overlap it on both the front and the tail, “and the situation becomes even worse for higher loadings” [1].

That leaves two bad options: “the collection window can be narrowed, in order to obtain a higher purity, to the detriment of recovery, or it can be widened, and this leads to higher yields but lower purity” [1]. The review’s conclusion is unambiguous: “in batch chromatography it is practically impossible to reach both a high purity and high yield at the same time” [1]. Preparative and industrial stationary phases make it worse, since their efficiency is limited by strong axial dispersion and mass transfer resistance, so a satisfactory purification often cannot be achieved in a single chromatographic step at all [6].

Regulators accept one standard mitigation. EMA notes that because complete separation of the product from pre- and post-eluting related impurities “is not always achieved, it is common to collect fractions and pool the main fractions with highest purity for further processing”, with separate acceptance criteria normally set for pooling main fractions and side fractions [3]. It also states that “in order to improve the yield, it is acceptable to perform repeated purification steps of the side-fractions with lower purity before combining them with the main fraction”, provided the procedure and the criteria for deciding when to do it are described; routine repurification is part of the manufacturing process rather than reprocessing, and must be covered by process validation [3].

Multicolumn countercurrent purification (MCSGP)

Continuous chromatography attacks the trade-off rather than managing it. Multicolumn countercurrent solvent gradient purification uses two or more identical columns, switched in position against the flow direction, with some columns interconnected so that impure product-containing streams are recycled internally and countercurrently instead of being discarded [6] [1]. The distinguishing feature against other continuous methods such as simulated moving bed is that MCSGP can run a solvent gradient [1]. The process was introduced in 2007, and one of its first peptide applications used six columns; two-column systems are now established at laboratory and production scale [6] [2] [1].

The published results on peptides are specific enough to plan against:

Peptide	Length	Batch result	MCSGP result	Reference
Calcitonin, industrial crude at 46% purity	32 residues	Maximum yield 66% at the required purity	Yield close to 100%, purity 93%	[6] [1]
Glucagon, industrial synthetic crude	29 residues	Recovery 71%, purity around 89%	Recovery about 88 to 90%, purity about 89 to 90%	[7] [1]
Icatibant	10 residues	Recovery around 12% at purity above 99%	Recovery above 95% at purity above 99%	[1]

The glucagon study is the most directly useful, because it quantifies both sides of the ledger: transferring the method from batch to twin-column MCSGP gave a 23% increase in recovery of the target product at unchanged purity of around 89%, and “this improvement comes at the expenses of a reduction of about 38% in productivity” [7]. The icatibant case went the other way, with productivity improved more than fivefold and solvent consumption cut by 80% [1]. Which outcome you get depends on the separation.

Other routes to a lower downstream footprint include displacement chromatography, whose reduced solvent consumption makes it a green alternative, and supercritical fluid chromatography, which can be run with carbon dioxide recycling at pilot scale but whose high pressures and dedicated equipment limit larger preparative use [2]. Mechanistic process modelling is well established for designing these separations, and in silico models have been used to develop multistep production-scale purifications with significant yield gains [2].

Setting the counter-ion

Purification determines the salt form, whether or not anyone chooses it deliberately. Fmoc solid-phase synthesis followed by reversed-phase purification with trifluoroacetic acid “lead to peptides as TFA salts”, and trifluoroacetate may be present either bound to the basic residues or simply adsorbed in the lyophilizate as a contaminant [5]. Trifluoroacetate is not inert in biology: in fetal rat osteoblast cultures it reduced cell numbers at 10^-8 to 10^-7 M, and trifluoroacetate salts of amylin and calcitonin consistently gave less proliferation than the hydrochloride salts [14].

Three approaches are used to change it:

1. Change the modifier in the final preparative run. Running the last purification with a different acidic modifier, such as acetic acid, delivers the corresponding salt. The drawbacks are low efficiency and a limited choice of compatible counter-ions; hydrochloric acid is unsuitable because of corrosion risk to stainless steel pumps, detectors and valves [5]. An evaluation on the dicationic octapeptide lanreotide found that preparative reversed-phase achieved partial to almost complete exchange to an ion from a weaker acid than trifluoroacetic acid [9].
2. Pass the pool over an ion-exchange resin. Resins modified with quaternary ammonium groups can be loaded with the desired anion by flushing, then used either as a slurry in the peptide solution or as a packed column. The method is described as very universal with very high exchange rates [5]. A carbonate-loaded variant removes trifluoroacetate as carbon dioxide while the anion is captured on the resin [5]. In manufacturing, “a final ion exchange step is often required after the last purification step to ensure the peptide is associated with the right counterion before going to the final isolation step” [2].
3. Lyophilize repeatedly from a stronger acid. The classical route is repeated freeze-drying from dilute hydrochloric acid. It is time-consuming, needs several repetitions, does not remove trifluoroacetate completely, and reported efficiency reaches up to about 98% after several cycles; exchange is only possible towards acids stronger than trifluoroacetic acid, so acetic acid cannot be reached this way [5]. It can also degrade the peptide, mainly by peptide bond hydrolysis [5]. In one study of the tryptophan-rich antimicrobial peptide indolicidin, hydrochloric acid above 10 mM in unbuffered solutions modified the secondary structure and reduced thermal stability, and 2 to 10 mM was adequate to remove essentially all trifluoroacetate without structural change [10].

Verification is straightforward: ion chromatography is the first-choice technique for counter-ion determination, with 19F NMR giving a characteristic trifluoroacetate singlet near minus 75 ppm and capillary electrophoresis offering a cheaper alternative with higher detection limits [5] [11]. The details are in our guide to peptide counter-ions.

From pooled fractions to dry powder

The pooled fractions are a dilute aqueous-organic solution. Industrial isolation concentrates them, removes the cosolvent, reduces bioburden by microfiltration and then freeze-dries the material. Cosolvent removal is done by tangential flow filtration with diafiltration or by evaporating the organic solvent, and large-scale tray lyophilizers are used for active substance batches below about 50 kg [2].

EMA states that “lyophilisation of synthetic peptides is considered common practice”, with precipitation, crystallisation and spray drying as alternative isolation methods whose relevant process parameters must be described [3]. The ACS analysis explains why freeze-drying wins despite its energy cost: it is “a beneficial process for the isolation of the final peptide product contributing to the PMI due to high yield, compared to alternative procedures such as precipitation, crystallization, and spray drying” [2]. Spray drying has higher throughput and lower energy consumption at large scale, but exposes the product to high temperatures, which may harm purity and yield [2]. Crystallisation “can be a valuable option but with limited applicability to peptide manufacturing” [2].

Cycle design is its own discipline. Freeze-drying processes are often set by trial and error, so commercial cycles are “often neither robust nor efficient”, when rational design around the collapse temperature and the glass transition of the freeze-concentrate is achievable for most products [13]. Our peptide lyophilization guide covers freezing, primary and secondary drying and excipient choice.

Two practical points carry over from purification. First, the volatile acid used as modifier leaves with the water, which is exactly why volatility was a selection criterion in the first place [5]. Second, residual water and counter-ion both count against net peptide content, so the purity figure from the final chromatogram is not the share of the powder that is peptide.

What the purification process has to deliver

The purified pool has to meet a specification, not just a chromatogram. EMA’s expected analytical content for a synthetic peptide active substance includes appearance, identification, purity as total and individual impurities, high molecular weight impurities, assay or content, counter-ion identity and content, residual trifluoroacetate, water content, residual solvents and elemental impurities [3]. High molecular weight impurities are defined as undesired by-products of greater mass than the target, typically formed during synthesis or storage by side reactions or incomplete purification [3].

Two details are easy to overlook. Limits for assays determined by liquid chromatography “are expressed in terms of the counter-ion free, anhydrous substance, unless otherwise justified” [3], so the basis of the number matters as much as the number. And purification columns are a cross-contamination risk: appropriate measures must be in place to prevent cross-contamination from the successive purification of different peptides on the same column [3].

Questions to ask about a supplier’s purification process

1. How many chromatographic steps, and on which properties? Two to three orthogonal steps are the manufacturing norm [2].
2. What is the crude purity on the load? It sets the achievable yield at your purity target [1].
3. Which additive, at what concentration? This determines the delivered salt form [5].
4. Is a dedicated counter-ion exchange step included? If not, expect trifluoroacetate [2] [5].
5. Are side fractions repurified routinely? If so, the procedure and criteria should be documented [3].
6. How is the pool isolated? Lyophilization, precipitation or spray drying, and what residual moisture results [3] [2].
7. What prevents carry-over between products on shared columns? [3]

How Winhycare Pharma purifies peptides

Winhycare Pharma supplies research peptides as research-use-only material, at purity of at least 99% as stated in our product brochure, with QC documentation, and offers custom peptide synthesis for sequences outside the catalogue. Semaglutide, tirzepatide, retatrutide, cagrilintide and BPC-157 are also supplied as API to licensed pharmaceutical manufacturers through our peptide API hub; US-DMF and CTD Module 3 support is available for semaglutide only. Process and analytical development, including purification method development and scale-up, is offered as peptide CRDMO services from Hyderabad, India, and purified peptide can be taken into vials through our lyophilisation service. To discuss a purification problem, use the custom synthesis enquiry form.

For laboratory research use only. Not for human or veterinary use.