Insights

From 10 Grams to 10 Kilograms: Mapping Solid Form Stability in HPAPI Scale-Up

An anonymized project shows why polymorph behavior validated at gram scale cannot be assumed to hold at kilogram scale.

23 October 2026 7 min read PCL Research & Manufacturing
From 10 Grams to 10 Kilograms: Mapping Solid Form Stability in HPAPI Scale-Up

A polymorph that behaves predictably in a small round-bottom flask can turn unstable the moment the same crystallization is run in a reactor ten or a hundred times larger. For high-potency APIs, where every transfer happens inside containment and every gram is both expensive and hazardous, discovering that instability during a GMP campaign is not an acceptable way to learn it.

The Situation

A development program for a small molecule classified as an OEB4/OEB5 compound had completed early solid form selection work at laboratory scale. The target polymorph had been identified, characterized, and judged suitable for formulation based on solubility, stability, and manufacturability criteria.

As the project moved toward scale-up, the sponsor needed to confirm that the same solid form — and the same crystallization process — would behave consistently as batch size increased by roughly three orders of magnitude. The compound’s potency meant that every experimental iteration had to be executed inside high-containment infrastructure, which constrained both the number of trials and the techniques available for in-process monitoring.

This is a common inflection point in API development: the solid form is ‘locked,’ but the process that produces it has not yet been proven across the scale range the program actually needs.

What Was at Stake

Polymorphic transitions discovered late in a development program are expensive in ways that go beyond the cost of repeating a batch. For a high-potency compound, an unplanned crystallization failure inside a containment suite also means re-qualifying operator exposure limits, re-scheduling limited high-containment capacity, and potentially reopening analytical method validation if the new form behaves differently under existing release methods.

There was also a regulatory dimension. Any shift in polymorphic form, solvate content, or crystallinity between the material used in early toxicology or formulation studies and the material intended for later clinical supply can trigger bridging questions from regulators. Demonstrating solid form consistency across scale is, in practice, part of the CMC narrative a sponsor must be able to defend.

The sponsor’s priority was not simply to produce a kilogram-scale batch, but to produce one with a documented, scientifically justified rationale for why the solid form would remain stable — and a contingency plan if it did not.

Why Crystals Behave Differently at Different Scales

Crystallization is governed by nucleation and growth kinetics, and both are sensitive to parameters that change systematically with vessel size. Cooling rates that are nearly instantaneous in a small flask become gradients in a larger reactor, where heat transfer through the vessel wall is slower relative to the volume being cooled.

Agitation behaves similarly. Impeller tip speed, shear distribution, and mixing time scale non-linearly with vessel geometry, which changes local supersaturation profiles and can shift the balance between primary nucleation, secondary nucleation, and crystal growth. A process that generates a narrow particle size distribution at small scale can produce agglomerates, different habit, or occluded solvent at larger scale simply because the local environment experienced by a growing crystal is no longer the same.

These effects are well documented in the crystallization engineering literature, but their magnitude is compound-specific and cannot be reliably predicted from first principles alone for most small molecules. This is precisely why empirical, scale-spanning stability mapping — rather than a single confirmatory batch — is the more defensible approach for complex or high-value compounds.

Building a Scale-Spanning Stability Map

Rather than running a single scale-up confirmation batch, the program was structured as a staged stability map across intermediate scales, with analytical checkpoints at each step. Each intermediate batch was evaluated against the same solid-state panel used in the original form selection work, so that any deviation would be caught against a consistent baseline rather than against a one-off specification.

The analytical panel combined complementary techniques deliberately chosen to detect different failure modes:

Thermal and spectroscopic data were reviewed together rather than in isolation, since a change that is ambiguous on one technique is often unambiguous when cross-checked against another. This redundancy is standard practice in solid-state work precisely because no single technique is sensitive to every possible transformation.

Stability screening under accelerated temperature, humidity, and mechanical stress conditions was run in parallel on material from each scale point, rather than only at the end of the program. This allowed the team to see whether sensitivity to stress conditions was itself changing with scale — an early warning signal that is easy to miss if stability testing is treated as a final release gate rather than a running diagnostic.

Containment Adds a Layer of Complexity

Working at OEB4/OEB5 containment levels changes how this kind of stability mapping can be executed in practice. Sample transfers between crystallization, filtration, drying, and analytical testing all require closed or semi-closed handling, which limits the frequency and speed of in-process sampling compared with a non-potent compound.

This constraint pushes toward analytical methods that work reliably on small sample masses and tolerate the realities of contained handling — without sacrificing the sensitivity needed to detect a polymorphic shift early. It also means that every sampling point in the scale-up plan has to be justified in advance, since each one carries operator exposure and logistical cost.

In practice, this favors a protocol designed around fewer, well-chosen checkpoints rather than dense, exploratory sampling — the opposite of how early-stage, non-potent screening is often run. The discipline this imposes is, in our experience, often beneficial: it forces the analytical plan to be built around specific, falsifiable hypotheses about what could go wrong, rather than open-ended curiosity.

The Outcome

The stability map confirmed that the target polymorph remained the thermodynamically favored form across the scale range tested, with only minor, well-characterized shifts in particle size distribution attributable to the expected changes in agitation and cooling profile. No unexpected polymorphic transition, solvate formation, or amorphous content was observed at any of the intermediate checkpoints.

Because the stability map had been built incrementally, the sponsor also walked away with a documented rationale explaining why the process was robust — not just a batch record showing that it had worked once. This distinction matters when the same process later needs to be transferred to a different site, a different reactor train, or a different scale again.

Equally important, the project avoided the scenario the sponsor had been most concerned about: discovering a solid form issue only after committing a full high-containment manufacturing slot to a kilogram-scale batch.

The Transferable Lesson

The underlying lesson is not specific to this one project. Solid form selection and process scale-up are too often treated as sequential, separable workstreams — screen the form, then scale the process, then confirm at the end that nothing changed. For complex small molecules, and especially for high-potency or controlled compounds where every batch is costly and tightly constrained, that sequencing creates risk that is avoidable.

A more resilient approach treats solid form stability as a variable to be actively monitored across the entire scale-up trajectory, with analytical checkpoints built into the plan from the start rather than added reactively when something looks wrong. The additional analytical effort is modest compared with the cost of a failed or delayed GMP batch, particularly under containment.

This is, in essence, the ‘first-time-right’ philosophy applied to scale-up specifically: the goal is not to react quickly when a problem appears, but to structure the program so that the problem is visible — and addressable — well before it reaches a scale where it becomes expensive.

Scale-up is one of the points in API development where solid-state science and process engineering have to be worked together rather than in sequence, and where the cost of a late-stage surprise is highest for high-potency and controlled compounds. If your program is approaching a scale transition and you want a second opinion on where the risk actually sits, we are glad to talk it through.

Share

More insights

All articles