Peptide Separation and Purification: Loadability vs. Capacity
How much peptide can a chromatography column handle? The useful answer is not a capacity number. It is how much product that meets specification you can recover per run, and how quickly.
In our recent preparative work with crude liraglutide, the medium that gave the narrowest peak recovered the least peptide: 65.5%, compared with about 79% for our All-Carbon microbeads (NanoPak-C) and a second commercial C18. Our medium also had the shortest cycle.
This post explains why saturation capacity, binding capacity, and loadability measure different things, what our published and new GLP-1 data show, and what to ask of any loading claim.
Capacity is not one number
Chromatography columns contain a packed bed of microscopic particles. Molecules interact with particle surfaces and move through the column at different speeds, separating the product from impurities. As more material is loaded, peaks broaden, become asymmetric, or shift in retention time. Capacity measurements describe these changes, but what they mean depends on how they are obtained.
Four related terms are often used interchangeably, but they answer different questions:
Term | What it describes |
Equilibrium adsorption capacity | The amount adsorbed at equilibrium under specified solution conditions; an adsorption model may estimate a limiting saturation capacity |
Chromatographically estimated saturation capacity | A parameter inferred from how chromatographic behavior changes with load, using a particular calculation method |
Dynamic binding capacity | The amount loaded under flow before the target reaches a specified breakthrough level in the column effluent |
Practical preparative loadability | The amount loaded per run while still achieving the required purification outcome |
For preparative purification, loadability is the most directly relevant measure. Even so, a loadability number means little without the purity, recovery, and processing time it comes with.
Why a broader peak does not mean a failed purification
Analytical chromatography prioritizes sharp peaks and reliable measurement. Preparative chromatography prioritizes isolating usable product, so it deliberately tolerates some overloading and peak broadening. The practical limit is not the first sign that a peak widens. It is the point at which more loading stops the process from meeting specifications or lowers its productivity.
The distinction that matters is between peak width and impurity overlap:
Peak width describes how broadly the target emerges from the column.
Impurity overlap describes how much unwanted material emerges alongside it.
A narrow peak can be hard to purify if a closely related impurity sits beneath it. A broader peak can be more useful if that impurity is well separated.
Our GLP-1 data: The published baseline
In our ACS Omega paper, Liquid Chromatography Evaluation of a Novel Graphitic Carbon Stationary Phase Using Glucagon-Like Peptide-1 Analogs, we compared All-Carbon microbeads with commercial silica-C18 controls, using semaglutide and liraglutide as probe peptides on 150 × 4.6 mm analytical columns.
We estimated a saturation-capacity parameter from how peak width grew as more peptide was injected, at six loads from 10 to 1,000 µg. Each value below averages the estimates from those injections, which is why the spread is wide.
Medium | Particle diameter | Estimated saturation capacity (µg per column, mean ± SD) |
Silica-C18 control | 4 µm | 883 ± 702 |
Silica-C18 control | 9 µm | 1,232 ± 949 |
All-Carbon microbeads | 6 µm | 615 ± 214 |
The carbon material had a lower mean, but the difference was not statistically significant (one-way ANOVA, p = 0.33). That does not show equivalence, superior capacity, or a maximum practical load. It gave us a comparative starting point.
From peak behavior to recovered product
In more recent work, we injected about 30 mg of crude liraglutide onto 150 × 4.6 mm columns packed with about 10 µm All-Carbon microbeads or one of two commercial C18 packings, C18-A and C18-B. That load is about 30 times the highest injection in the published study and roughly 50 times the 615 µg capacity estimate, on columns of the same size. The estimated saturation capacity is clearly not a ceiling on practical loading.
Metric | All-Carbon | C18-A | C18-B |
Peak width | 5.6 min | 5.5 min | 4.5 min |
Product purity | 98.6% | 98.5% | 98.0% |
Recovery | 79.0% | 79.2% | 65.5% |
Cycle time | 12 min | 15 min | 15 min |
Source: single preparative runs per medium; manuscript in preparation.
C18-B produced the narrowest band, yet recovered the least liraglutide at the lowest purity. All-Carbon and C18-A gave similarly broad bands and similar purity and recovery, with All-Carbon completing its cycle in 12 minutes rather than 15 because liraglutide elutes earlier on the carbon medium. These are single runs, so small numerical differences, such as 98.6% versus 98.5%, should not be read as meaningful.
The narrowest peak was not the best predictor of performance. What mattered was how much liraglutide could go into an acceptable product pool.
The two studies answer complementary questions, and the capacity estimate was reported for the 6 µm grade while the preparative work used the 10 µm grade, so they are not a before-and-after capacity comparison. The published study characterizes chromatographic behavior; the recent work measures the outcomes that matter in manufacturing.
What might explain the difference?
One possible explanation is better separation of liraglutide from a leading impurity, one that elutes just before the product. Under overloaded conditions, components compete for the stationary phase, and a strongly retained product can push a less retained impurity ahead of it. This is known as the displacement effect, and it reduces overlap. The opposite can also happen: in the tag-along effect, a later-eluting impurity is dragged into the product band at high load.
This remains a hypothesis. Our liraglutide elution profiles do not show the leading impurity separately, so they cannot establish displacement. Differences in selectivity and in how fractions were pooled could also contribute. Because the product pools also differ slightly in purity, the most direct comparison would be recovery at the same purity and individual-impurity limits.
What counts as acceptable purity and recovery in peptide purification?
There is no universal pair of values. Purity must meet the product's requirements or the next manufacturing step's needs, and recovery must make the process technically and economically worthwhile.
For pharmaceutical peptides, “99% pure” is not enough on its own. Chromatographic purity differs from peptide content by mass, and each impurity needs its own controls. An intermediate purification step may have different requirements from final polishing.
Recovery measures how much target peptide reaches the accepted product pool:
Recovery (%) = (target-peptide mass in the accepted product pool ÷ target-peptide mass loaded) × 100
The denominator is the target-peptide mass, not the total mass of crude feed.
Purity and recovery are linked through fraction selection. As material leaves the column, it is collected in portions called fractions. Pooling only the cleanest fractions improves purity, but sacrifices peptide; including edge fractions improves recovery but admits more impurities. The fairest comparison between media is therefore the highest recovery achievable at the same purity and individual-impurity limits.
Why the inside of the bead matters
If calculated capacity and peak width do not predict purification performance, what should material development optimize? The answer lies in the combination of accessible surface, molecular transport, and selectivity. Three bead properties need to be considered together:
Property | Meaning |
Pore size | Dimensions of internal openings and channels |
Intraparticle porosity | Fraction of particle volume occupied by internal voids |
Specific surface area | Surface area per unit mass of material |
Smaller pores can provide more surface area for a given pore volume, but that surface offers little benefit if the peptide cannot reach it efficiently. Larger pores improve access and transport but can reduce surface area. Neither the highest surface area nor the largest pores automatically makes the best peptide medium.
Our published study illustrates this trade-off. The 6 µm All-Carbon material had a pore size of 255 Å and a specific surface area of 419 m²/g, and the 10 µm grade used in our preparative work had 163 Å pores and 389 m²/g. The silica-C18 controls had smaller pores (70 and 94 Å) and surface areas of 788 and 459 m²/g. The 4 µm control was measured by BET and the other media by water intrusion porosimetry, so the comparison needs some caution. These measurements describe the materials, but they don't isolate which feature drives a given purification outcome.
The goal is accessible surface area: enough interaction area, reached efficiently by the target peptide, with surface chemistry that supports separation and recovery.
How can useful capacity be improved?
Material design and method development have to work together:
Increase peptide-accessible surface area, not surface area in pores the target cannot use efficiently.
Match pore architecture to the molecule, balancing pore size, internal volume, and transport.
Optimize surface chemistry to balance retention with recoverability.
Improve selectivity. Better separation from a critical impurity can raise acceptable loading without raising intrinsic adsorption capacity.
Optimize loading and collection: solubility, injection volume, mobile phase, flow, gradient, and fraction selection all affect the result.
A controlled evaluation can start with a purified peptide to compare overload behavior and recovery, move to a defined target-and-impurity mixture, and then confirm with a representative process feed. Purified peptide alone cannot establish impurity clearance.
What to ask of a loading claim
A larger loading number alone does not describe a better process. For example, a hypothetical claim of 10 mg per gram of packing describes the mass loaded per run, not the purified product recovered. A loading claim becomes useful when it comes with:
Ask | Why it matters |
Is the load total feed or target peptide? | Crude feed can contain much less target than its total mass |
What was the feed’s impurity profile? | Clean feed is far easier to load heavily than crude feed |
What purity and individual-impurity levels were achieved? | Loading is only meaningful at a stated quality |
What was the target recovery at that purity? | Defines how much usable product each run delivers |
What was the full cycle time, including washing and re-equilibration? | Sets runs per day and throughput |
How much solvent was used per unit of acceptable product? | Drives cost and environmental footprint |
A useful claim reads like this: “The medium processed a defined feed at 10 mg/g of packing and delivered the required impurity profile at the measured target-peptide recovery.”
From capacity to manufacturing value
Saturation capacity remains useful, but it is one part of the picture. Pore architecture and surface chemistry create the opportunity; impurity separation, product recovery, and cycle efficiency determine its manufacturing value. Our GLP-1 work follows that progression, from characterizing NanoPak-C’s chromatographic behavior to measuring what it delivers under preparative loading.
The published study is available in ACS Omega (DOI: 10.1021/acsomega.5c02172). The preparative results discussed here are part of a separate manuscript in preparation.
If you are evaluating media for GLP-1 or other peptide purification and want to compare them on recovered product rather than capacity, we would be glad to talk. Contact Millennial Scientific at inquiry@millennialscientific.com or use our contact form.
