Chromatography Scale-Up: What to Keep Constant and What to Recheck

1. Define the Separation Before the Scale Factor

A scale-up plan begins with the function of the step: capture product, resolve a critical impurity, remove contaminants in flow-through, or exchange buffer. Write measurable acceptance criteria for recovery, impurity content, activity, pool volume, and processing time. A method that captures more material but misses the required impurity limit has not scaled successfully.

Separate established observations from assumptions. Record the actual resin, feed composition, target concentration, load per packed resin volume, bed height, and operating temperature behind the development result. Identify whether capacity, resolution, pressure, or product stability limits throughput. Different limitations require different scaling decisions; matching every historical instrument setting can preserve the wrong constraint.

Choose a meaningful reference run

Use a representative run with complete fraction analytics and mass balance. A clean chromatogram obtained with unusually dilute or highly purified feed may not represent manufacturing conditions. Include late load fractions, wash losses, and rejected elution fractions so the reference describes both accepted product and the routes by which material was lost.

2. Translate Bed Geometry into Flow and Volume

For a cylindrical packed bed, cross-sectional area A = πD²/4 and packed bed volume V = AL, where D is internal diameter and L is bed height. Superficial velocity u equals volumetric flow Q divided by A. Nominal empty-bed residence time is V/Q, or L/u. Use consistent units throughout the calculation.

Increasing diameter by a factor of ten increases area, bed volume, and required flow by a factor of one hundred when height and velocity remain constant. Sample mass and step volumes also increase one hundredfold if their normalized values remain unchanged. Pump capacity and buffer inventory must therefore be evaluated before ordering the larger column.

Do not simultaneously specify a different height, unchanged velocity, and unchanged residence time: those requirements are incompatible. If available equipment forces a height change, decide which variable controls the separation and conduct a bridging study. The discussion of residence time and productivity explains this tradeoff in greater detail.

3. Preserve the Variables That Control Separation

As an initial scale rule, retain the resin chemistry and particle properties, bed height, normalized target load, relevant feed composition, temperature, and linear velocity. Specify buffer composition and actual pH as well as conductivity. Conductivity alone does not uniquely define salt identity, buffer capacity, or the strength of interactions with the stationary phase.

Express equilibration, wash, elution, and regeneration volumes in column volumes, using the measured packed bed volume. For a gradient, preserve the composition change per column volume at the column inlet. At unchanged residence time, the same gradient expressed in column volumes also has the same duration; changing height or velocity can break that relationship.

Normalization depends on mechanism. For size exclusion, sample volume relative to bed volume and resolution requirements are central. For flow-through polishing, impurity load and breakthrough can be more relevant than target binding capacity. For bind-and-elute capture, target dynamic capacity and impurity competition influence acceptable loading. One universal loading rule cannot cover all three situations.

4. Recheck the System Around the Column

Measure delay volume between buffer mixing and the column inlet, and hold-up between the column outlet, detector, and collection valve. Express these volumes relative to bed volume and translate them into time at operating flow. A gradient programmed identically on two skids can reach the resin at different times and with different dispersion.

Inspect minimum and maximum flow, pump accuracy, gradient mixing, detector path length, conductivity response, and valve switching. Compare actual delivered composition profiles using a suitable nonbinding conductivity or other tracer step. Verify detector linearity at the intended product concentration; a clipped absorbance peak can distort both pool boundaries and apparent recovery.

Scale sample preparation separately

Feed conditioning introduces mixing and holding operations that are not described by column geometry. Determine addition order, mixing time, local pH or salt exposure, temperature, and maximum hold duration. Large vessels can expose part of the feed to transient extremes even when the final bulk sample meets its specification. Sample the conditioned feed before committing it to the column.

5. Confirm Packing and Hydraulic Performance

For comparable packing and fluid properties, packed-bed pressure drop generally increases with viscosity, bed height, and velocity. Increasing diameter alone does not imply the same proportional rise in bed pressure drop, because velocity can remain unchanged. Total measured system pressure also includes tubing, valves, filters, and distributors, whose contributions may differ substantially between scales.

Establish a pressure–flow curve using the relevant buffer and temperature, including the highest expected viscosity. Evaluate the packed bed and hardware against their separate limits. Soft matrices can compress nonlinearly, so a low-flow extrapolation may underestimate the risk at the intended operating point. Confirm performance before valuable feed is loaded.

Use an appropriate packing test and a consistent tracer method. Record tracer volume, velocity, injection path, and extra-column contribution when interpreting plate height and peak asymmetry. A satisfactory tracer result supports packing consistency but does not prove biochemical selectivity or impurity clearance. Larger distributors can introduce radial maldistribution that was negligible in a narrow development column.

6. Work Through an Illustrative Diameter Scale-Up

The example exposes two independent checks. The skid must deliver the required flow with adequate mixing and pressure margin, and the facility must prepare and hold the larger liquid volumes. Reconfirm that target concentration, load duration, and impurity burden remain representative rather than matching target mass alone.

7. Build a Scale Comparison Around Product Quality

Compare normalized chromatograms, but base acceptance on measured material quality. Overlay conductivity, pH where available, target-specific concentration, and impurity profiles against column volumes referenced to the inlet. Account for sampling and detector delay before interpreting a shifted peak as a change in resin behavior.

Analyze complete mass balance with uncertainties appropriate to the assay. Investigate discrepancies across feed, flow-through, wash, accepted pool, rejected fractions, and strip. If performance changes, vary one suspected equipment or handling difference deliberately before concluding that the chemistry requires redesign. Preserve samples where stability permits so the investigation can compare both scales using the same analytical sequence.

8. Recognize When Simple Geometric Scaling Is Insufficient

High-resolution separations, very large biomolecules, compressible beds, and strongly viscous feeds may need additional development. Diffusion and pore accessibility constrain mass transfer, while altered particle size changes both efficiency and pressure. Substituting a different resin grade at the same time as increasing scale creates a material change as well as a geometric change.

Cleaning and reuse also require scale-specific checks. Verify chemical contact time, concentration, delivery through all flow paths, flushing endpoints, and storage coverage. A larger bed can require longer preparation or drain periods even when the chromatographic portion of the cycle is unchanged. Include those intervals when checking product exposure and batch scheduling.

When the manufacturing target lies far beyond the development geometry, use an intermediate scale to test uncertain packing, mixing, or handling assumptions. The purpose is to reduce a named risk, not to apply a fixed sequence of column sizes. Document why the chosen intermediate experiment can detect the failure mechanism of concern.

9. Diagnose a Scale Difference Before Changing Chemistry

Use the shape and timing of a discrepancy to prioritize investigation. If conductivity transitions shift together while normalized product selectivity is unchanged, examine inlet delay and method alignment first. If the target broadens but a nonbinding tracer also broadens, investigate packing and extra-column dispersion. If only product behavior changes, examine feed condition, loading, stability, and resin interactions.

Differentiate a pressure increase during loading from one present during equilibration. A feed-specific increase suggests viscosity, particulates, precipitation, or fouling; a persistent increase may arise from packing, hardware, or a restricted flow path. Compare pressure at a common buffer, temperature, and velocity before attributing the change to resin deterioration.

Separate operational and biochemical recovery

At larger scale, product can remain in transfer lines, collection vessels, or filters even when the column elution itself is unchanged. Quantify these recoverable hold-up losses separately from breakthrough or irreversible adsorption. Rinsing a line can increase recovered mass but dilute the pool, so evaluate both recovery and downstream consequences before incorporating the rinse into the method.

When a gradient peak shifts, compare actual inlet composition at elution rather than elapsed time alone. An unchanged elution conductivity with a time shift suggests a delivery difference, while changed quality at the same inlet profile points toward a different mechanism. Neither observation is conclusive in isolation; combine chromatograms with fraction identities and equipment measurements.

Choose the smallest corrective experiment that tests the leading explanation. For example, correct collection delay without changing buffer chemistry, or test the feed at the original scale after its larger-vessel conditioning. Avoid changing resin, gradient, loading, and pooling simultaneously, because a successful repeat would then provide little information about the original cause.

Document the diagnosed cause and the evidence supporting the correction. If the cause remains unresolved, preserve that uncertainty in the handoff and define the additional check needed before broader operation. A reproducible corrected result is more informative than an unexplained acceptable pool.

10. Prepare a Scale-Up Decision Record

The final record should connect each preserved variable to its scientific rationale and each changed variable to supporting evidence. Include calculated and measured bed volume, diameter, height, flow, residence time, phase volumes, sample loading, pressure limits, and pool criteria. Define stop conditions for unusual pressure, unexpected breakthrough, or feed instability before the first larger run.

Retain raw chromatograms, instrument configuration, fraction identities, packing data, buffer preparation records, deviations, and the actual outcome against predefined criteria. Distinguish a successful engineering run from broader process qualification. Transfer the operating range and known limitations, not only the settings of the best experiment.

For candidate materials, review the available ion exchange chromatography media against the required separation and flow range. If equipment compatibility or column preparation needs development, discuss the scope through custom chromatography services. Define the experimental evidence required for your application before committing to a specific scale or material.

For the first larger run, calculate the expected timing of each phase and the expected quantity of each collected stream in advance. Compare those expectations with actual measurements during execution. An unexpected difference in delivered volume can indicate calibration, valve routing, or bed-volume errors before final analytics are available. Keep the comparison diagnostic rather than using an approximate expectation as a substitute for a validated acceptance limit. Resolve any discrepancy that could change loading or pool identity.

References

1. Carta G, Jungbauer A. Protein Chromatography: Process Development and Scale-Up. 2nd edition. Wiley-VCH; 2020. doi:10.1002/9783527824045
2. ICH. Q11: Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities). 2012.