Why a freeze-dried cake looks the way it doesWhat the studies show

Water in a vial on a cold shelf does not freeze at 0 °C. It supercools, typically to 10 or 20 degrees below zero, and then nucleates all at once; the deeper the supercooling, the more nuclei form and the smaller the ice crystals that grow from them ³. The solute — the peptide, and any buffer or bulking sugar — is excluded from the growing ice and concentrates into the channels between crystals, where it ends up as a glass or, for something like mannitol, as a crystalline matrix. The ice crystals become the pores of the cake, so fast freezing with deep supercooling gives a fine-pored, dense-looking cake, and slow freezing or a deliberate annealing step gives larger crystals and a coarser, more open one ³. Larger pores let vapour escape faster in the next step, which is why manufacturers try to control nucleation and why two vials of the same solution can look quite different.

Primary drying removes the ice. The chamber is pumped down to a fraction of a millibar and the shelf is warmed just enough to supply the heat of sublimation while keeping the product below its collapse temperature — the point at which the freeze-concentrated glass between the pores softens and flows ⁴. Hold it below that and the pores keep their walls; let it rise above and the structure slumps, which is the collapse seen as a shrunken, glassy or partly melted-looking cake. Secondary drying follows: the ice is gone, but the solid still holds water bound to its surfaces, often several percent by weight, and the shelf is raised, sometimes well above room temperature, to drive it off over hours until residual moisture is typically below one or two percent ²⁵. Both steps are run against sensors and clocks, not by eye.

Shrinkage happens because the dried glass takes up less volume than the frozen matrix it replaced, and the cake pulls away from the vial wall as it loses water; cracking happens when that shrinkage is uneven, or when the top skin dries first and the material underneath contracts beneath it. A small amount of solid in a large volume — a few milligrams of peptide with no bulking agent — gives a sparse, fragile cake that may sit as a thin disc or a loose powder rather than a plug, because there was never enough material to build a wall ⁵. These are mechanical facts about the solid, not about the molecule inside it.

In 2017 a group of industry and academic scientists published a consensus review of cake appearance that asked exactly what such defects mean. Their conclusion was that cracking, shrinkage and a cake that has pulled from the wall are generally cosmetic and do not by themselves indicate a change in potency, moisture or how the solid dissolves, while gross collapse or melt-back can go with higher moisture and slower dissolution and is worth investigating — investigating, with an assay, not condemning on sight ⁵. The things that decide whether a freeze-dried peptide is what its label says — purity, identity, residual moisture, endotoxin — are measured, and none of them can be read from the shape of a white solid.