Humidity and the Lyophilised Cake: Collapse, Clumping and What a Wet Vial Looks Like

A lyophilised cake does not fail in a humid climate because it gets hot. It fails because it gets wet, and heat then finishes the job. The dried plug in a sealed vial is an amorphous glass, and water is the most effective plasticiser that glass will ever meet: every additional percent of residual moisture lowers the temperature at which the solid softens. Drop that softening point below the temperature of the room and the cake stops behaving as a solid. It retracts from the wall, slumps, and at the extreme becomes a dense bead at the bottom of the vial.

That is the whole mechanism, and it explains why ambient heat and ambient humidity are not two problems to be solved separately. What follows stays on the physical variable rather than on the full specification a lyophilised vial is meant to be stored to: what water does to the cake, what the compendial humidity limits actually say, where the water inside a sealed vial comes from, and how much of a cake's appearance is genuinely diagnostic — written for somebody holding laboratory material where the mains supply is intermittent and a moisture titrator is not in the building.

Water is the plasticiser, not the heat

Freeze-drying removes ice by sublimation and leaves a porous solid held together by an amorphous sugar or polyol matrix. That matrix has a glass transition temperature, Tg. Below it, molecular mobility is low enough that the structure holds and solid-state reactions crawl. Above it, the material is a very viscous liquid that flows on a timescale you can watch over weeks. Collapse is therefore not a chemical event in itself. It is the mechanical consequence of holding a glass above its own transition.

The numbers are stark. In a lyophilised monoclonal antibody formulation dried to residual moistures spanning roughly 1% to 8%, the measured glass transition fell from about 80 °C at the dry end to about 25 °C at the wet end, and cakes held above their own Tg aggregated faster than drier ones ^[3]. Twenty-five degrees is not a laboratory abstraction in Lagos, Mombasa or Douala. It is a mild morning. A cake carrying that much water sits above its transition for most of the year.

Two cautions attach to that figure. It belongs to one formulation, and the transition depends on what the cake is made of — a sucrose-rich matrix behaves differently from a mannitol-rich one. And a quoted collapse temperature is not a sharp material constant: it shifts with the measurement method and with the rate at which water leaves the glassy phase, so freeze-dry microscopy and thermal analysis need not agree ^[4]. Treat any single quoted temperature as an order of magnitude, not a threshold you can sit just underneath.

"Store in a dry place" has a number

The phrase on the label is not vague; it is defined. The United States Pharmacopeia chapter on packaging and storage requirements defines a dry place as one that does not exceed 40% average relative humidity at 20 °C, or the equivalent water vapour pressure at another temperature. The determination is meant to be made by direct measurement — not fewer than twelve equally spaced readings covering a season, a year, or the documented storage period — with individual readings up to 45% tolerated provided the average holds ^[1]. Almost no uncontrolled storage in a coastal or equatorial market meets that, and a cupboard is not made compliant by being called dry.

Set that against the conditions the World Health Organization uses for stability testing. Its guideline assigns markets to climatic zones and fixes the long-term test condition accordingly, with the hot-and-humid zones tested at 30 °C/65% RH and 30 °C/75% RH ^[2]. A product proven against a 75% relative humidity condition survived that environment in its final container — a statement about a sealed pack, not an opened one. Which zone a country designates is a decision of its own medicines regulator, so for a specific market the document to read is that agency's published stability guideline, not a supplier's summary of it.

Where the water in a sealed vial comes from

The intuition that a crimped vial is isolated from the room is the most expensive mistake in this area. A sealed vial has an internal water budget, and the largest reservoir in it is usually not the cake — it is the rubber closure. Work on moisture transfer from stopper to product measured the kinetics and the equilibrium of that transfer at 5 °C, 25 °C and 40 °C, and the conclusion was blunt: low residual moisture at manufacture does not guarantee low moisture across shelf life, because the product equilibrates with water held in the stopper ^[5]. The transfer is temperature-driven. A store running at 35 °C is not merely holding the vial near its transition; it is pumping the closure's water into the cake.

How much water is available to move depends on the closure formulation and how it was treated before sealing. A comparison of rubber lyophilisation closures found that different formulations absorb and desorb water to materially different extents, which is why moisture specifications and drying steps for stoppers belong to product design rather than manufacturing housekeeping ^[6]. None of that is visible downstream. It is, however, the reason a vial that has never been opened can hold more water after a hot year than it did on dispatch.

The second source is one you control completely, and it arrives the instant a cold vial meets a humid room. At 30 °C and 80% relative humidity the dew point is roughly 26 °C. Any surface colder than that — a vial straight from a refrigerator at around 4 °C — condenses liquid water onto itself immediately. With an intact seal that water sits on the outside of the glass and around the crimp, where it attacks labels and printed lot numbers. Where a closure has already been pierced, some of it has a route inward.

What a wet vial looks like

Visual inspection is the only assay most receiving sites will ever run, so it is worth being precise about what the eye can resolve. The vocabulary below describes physical states with their most likely causes. None of them is a measurement, and the third column matters as much as the first.

The cheapest control available is a photograph of each cake against a plain background on the day it arrives, filed with the consignment record. Appearance alone is weak evidence. Appearance against a dated baseline of the same vial is strong evidence, and it is the only way to separate a cake delivered collapsed from one that collapsed in your cupboard — a distinction that matters more than it sounds.

What appearance proves, and what it does not

Here the literature is less comfortable than vendor pages suggest. A systematic study that deliberately induced collapse during freeze-drying, across a monoclonal IgG1, a second pharmaceutically relevant protein and lactate dehydrogenase, found no negative impact on the properties of the cake or on protein stability, nor on reconstitution time or structural integrity, with residual moisture comparable to non-collapsed controls ^[7]. A cosmetically ugly cake is not, by itself, evidence of a degraded one.

The follow-up work is where the distinction earns its keep. Comparing cakes that collapsed during freeze-drying against cakes that collapsed later, during three months of storage at elevated temperatures, the same group found the protein significantly better stabilised in the material that collapsed during drying; the poorer outcome in storage-collapsed cakes correlated with crystallisation and hydrolysis of the stabiliser and with non-enzymatic browning ^[8]. Collapse as a manufacturing outcome and collapse as a storage outcome are different events, and only the second is a record of your conditions.

One further counter-intuitive result is worth carrying, because the instinct in a humid climate is to drive everything as dry as possible. Modelling water activity alongside glass transition across lyophilised antibody formulations held at 40 °C for nine months, one group reported the best monomer retention in a middle band of water activity rather than at the dry extreme, with a distinctly poor outcome at the lowest water activity studied ^[9]. Whatever residual moisture a manufacturer targeted was a target, not a floor.

Two limits on all of the above, stated plainly. Nearly every dataset here is monoclonal antibody or model-protein work, much of it accelerated rather than real-time, and a short synthetic peptide is a different physical object: no tertiary structure to lose, though solid-state hydrolysis and deamidation at asparagine and glutamine residues still track water content. And none of these studies ran on the formulations a research supplier ships. The mechanism transfers; the numbers do not. Anyone quoting a precise moisture threshold for a research peptide, in either direction, is extrapolating.

Measuring instead of guessing, with what is in the building

Residual moisture in a lyophilised solid is determined compendially by Karl Fischer titration or by loss on drying. Neither is available at a receiving site, both are destructive, and the sample they consume is the material you were trying to protect. So the practical question is which cheap, non-destructive records genuinely narrow the uncertainty.

• A logger with minimum and maximum memory, placed with the vials rather than on the wall, recording temperature and relative humidity together. Wall and shelf readings diverge more than people expect.
• A written mains-power record: date, time down, time restored. A refrigerator without power is a slowly warming insulated box, and that log is the only way to reconstruct what the vials experienced.
• Indicating desiccant inside a closed secondary container, replaced on colour change rather than on a calendar.
• A dated photograph of every cake at receipt, and again whenever the material is inspected.
• Transit duration and route kept with the consignment record — overland legs and airport apron holds are where the largest unlogged excursions happen.

Be honest about what a logger inside a storage box reports. It describes the box, not the vial interior, and while a seal is intact those two are only loosely coupled, since the cake's water comes largely from the closure. Its value is documenting the environment at the moment a seal is broken, and catching slow drift nobody notices in real time. Note too an asymmetry in the standards: temperature excursions are handled kinetically, through a mean kinetic temperature that weights hot periods more heavily than cool ones within a tolerated band of roughly 15 °C to 30 °C around a 25 °C mean, whereas the humidity limit in the same chapter is a flat arithmetic average ^[1]. There is no mean kinetic humidity, so swinging between dry and damp looks compliant on paper in a way that swinging between cool and hot does not.

Power cuts and transit: the shape of the excursion matters

A steady 30 °C and an average of 30 °C reached by swinging between 22 °C and 38 °C twice a day are not the same exposure. Every warm phase accelerates the stopper-to-cake transfer described earlier; every cool phase pulls the vial's outer surface toward the dew point of a room that is usually humid. Intermittent mains supply does not produce a mild average. It produces thermal cycling, and the cake integrates all of it rather than remembering the mean.

Transit compounds this, because the legs nobody logs are the hot ones: an apron hold, a customs shed under a metal roof, an overland stretch in an unventilated vehicle — precisely the gaps a receipt-side assessment of a temperature excursion has to work around. The mitigations are unglamorous and they work. Store and ship vials inside a closed secondary container so their immediate air is not the room's air. Keep that container off exterior walls and away from roof spaces. Bring it to room temperature before opening rather than opening a cold vial into humid air. And record what you can, rather than assuming a sealed pack is a sealed system. It is not, quite. It is a slow one.