Dry ice and peptide samples: sublimation, carbon dioxide, and the pH shift nobody records

Cold is usually treated as the one storage variable that cannot hurt anything. Degradation kinetics are exponential in temperature, so colder is slower, and the only questions worth asking about a shipment are whether it stayed cold and for how long it did not. That framing is sound for the heat side of the problem, which is where mean kinetic temperature does its work. It is incomplete for the cold side, because the most common way to achieve deep cold in transit is not a refrigerant that sits inertly beside the sample. It is a reagent that turns into several kilograms of gas and surrounds the sample with it.

Solid carbon dioxide has no liquid phase at atmospheric pressure. It passes directly from solid to gas at approximately −78.5 °C, and that sublimation is the entire basis of both its usefulness and its regulatory status. A shipper packed with dry ice is classified as a dangerous good — UN 1845, proper shipping name carbon dioxide, solid, Class 9 under the UN Model Regulations and the IATA Dangerous Goods Regulations — and the single packaging rule that survives across every transport mode is that the package must not be hermetically sealed. Gas has to be able to leave. The consequence for anything inside is rarely stated as plainly: the interior of that shipper is not air at −78 °C. It is an atmosphere that is essentially pure carbon dioxide, continuously replenished, at a concentration measured in thousands of micrograms per millilitre, and the sample container is sitting in it for the length of the transit.

Carbon dioxide is not a bystander in aqueous samples

Carbon dioxide dissolves in water and hydrates to carbonic acid, which dissociates to bicarbonate and a proton. The first dissociation has a pKa near 6.35 at room temperature, which places it awkwardly close to the pH at which most peptide solutions are formulated. Raise the partial pressure of carbon dioxide above a solution and the equilibrium moves, protons are generated, and the pH falls. None of this is exotic chemistry; it is the same equilibrium that governs carbonated water and ocean chemistry. What makes it a storage problem rather than a textbook exercise is that a dry ice shipper raises that partial pressure by roughly three orders of magnitude over ambient air and holds it there for days.

The route into the sample is worth getting right, because it is not what intuition suggests. Carbon dioxide has negligible solubility in ice. A solution that is already frozen solid does not take up appreciable carbon dioxide through its bulk. What happens instead is that the gas accumulates in the container’s headspace, where it is unreactive and unremarkable while everything stays frozen, and then dissolves into the liquid phase as the sample thaws. The acidification is therefore a thaw event, not a transit event, which is precisely why it goes unrecorded: a temperature logger in the shipper reports a flawless cold chain, and it is telling the truth.

Published work on protein solutions transported or stored on dry ice reports thawed pH values between roughly 5.5 and 6.0 for material formulated at 7.2, and drops as large as two and a half pH units in standard centrifuge tubes. The magnitude scales with the things that determine how much acid a solution can absorb before its pH moves: buffer capacity above all, then the ratio of headspace to liquid volume, then the exposure time. Unbuffered solutions are the most exposed, and modelling of headspace carbon dioxide against final pH has been performed for unbuffered samples and for dilute Tris at various fill volumes, with the low-capacity cases moving furthest. A peptide dissolved in plain water or bacteriostatic water has effectively no buffering at all.

What the container decides

The container is doing more work here than the refrigerant. Polypropylene microcentrifuge and screw-cap tubes are not a carbon dioxide barrier; the polymer is permeable to the gas, and the cap thread is a path rather than a seal. This is the specific finding that makes the effect common in practice — the standard tube that everything gets aliquoted into is the container that does not protect against it, and it does not protect even over short-term dry ice storage. A crimped glass vial with a properly seated elastomeric closure is a far better barrier, which is the same conclusion that container closure integrity testing reaches from the opposite direction: the closure system either controls exchange with the outside atmosphere or it does not, and that property is measurable rather than assumed.

Fill volume is the other lever, and it is the one most often set for convenience. A small aliquot in a large tube presents a large headspace reservoir against a small buffering mass. The same total sample split into fewer, fuller containers absorbs proportionally less. This runs in the same direction as the freeze-thaw argument for single-use aliquots, and the two considerations have to be traded against each other rather than optimised separately.

Which peptide chemistry the shift actually touches

An honest account has to note that a downward pH excursion is not uniformly damaging, and the pathways it accelerates are not the ones it suppresses.

Several of the classic solution-phase degradation routes are base-catalysed and are slowed, not accelerated, by acidification. Asparagine deamidation proceeds through a succinimide intermediate whose formation requires deprotonation of the backbone amide nitrogen, so its rate falls steeply below neutrality. Thiol–disulfide exchange, the scrambling pathway in cysteine-containing peptides, requires the thiolate rather than the thiol, and acid suppresses it — which is why disulfide-bonded material is conventionally handled at mildly acidic pH in the first place.

Other routes move the other way. Aspartate isomerisation and the hydrolytic cleavage of aspartyl bonds, particularly at Asp-Pro, are acid-favoured. So is the direct hydrolysis pathway that competes with succinimide formation at asparagine. A shift from 7.2 to 5.5 is not a move from a bad regime to a good one; it is a move to a different regime, with a different impurity profile.

The physical consequence may matter more than either. Peptide solubility passes through a minimum at the isoelectric point, where net charge is zero and the electrostatic repulsion that keeps molecules apart disappears. If a peptide’s pI lies between the formulated pH and the depressed pH, the thawing sample travels through its own solubility minimum while it is warming. What comes out of that is not a chemical impurity that an assay will report as a new peak — it is aggregation and particulate formation, a loss of material from solution that presents as low recovery rather than as a degradant, and that a purity method run on the supernatant will not see at all.

There is also a measurement consequence that is independent of whether any damage occurred. Carbon dioxide loading is reversible; the gas comes back out once the sample is no longer under a carbon dioxide atmosphere. A pH measured immediately after thaw can therefore be substantially lower than the same sample’s pH an hour later on the bench, which means the number recorded is a property of the shipping method rather than of the formulation. The standard mitigations follow directly from the reversibility: vent the container before thawing so the headspace is air rather than carbon dioxide, or allow the sealed sample to normalise in a conventional freezer — on the order of four days has been used — so the accumulated gas dissipates before anything is thawed or assayed.

The lyophilised vial is a different case entirely

None of the above applies with any force to a properly sealed lyophilised vial, and it is worth saying so explicitly rather than letting a solution-phase concern generalise. There is no bulk liquid water for carbonic acid to form in, the residual moisture that remains is bound within an amorphous solid, and a crimped closure is a real barrier. At −78 °C the cake is far below its glass transition temperature, molecular mobility is essentially arrested, and chemical change in the solid is about as slow as it can be made.

The risks for a lyophilised vial on dry ice are mechanical and procedural instead. A sealed secondary container that traps sublimating gas can pressurise, which is the reason for the venting rule. And a vial brought out of deep cold and opened before it has equilibrated to room temperature will condense atmospheric moisture onto and into a hygroscopic cake — water that plasticises the amorphous solid, depresses its glass transition temperature, and undoes in one careless minute what the cold chain spent the transit protecting. Letting the vial reach room temperature sealed, before breaking the closure, is the whole of the mitigation.

Reading the shipping method as a variable

The useful reframing is that the refrigerant is part of the formulation history, not part of the packaging. A frozen solution that travelled on dry ice in a permeable tube has been through a chemical treatment — a period under high carbon dioxide partial pressure, followed by an uncontrolled acid excursion during thaw whose magnitude was set by buffer capacity and headspace volume and recorded nowhere. Whether that treatment mattered depends on the peptide’s pI, its susceptible residues, and what was done with it afterwards, and for a great many samples the answer will be that it did not. But it is the kind of variable that turns into an unexplained result later: a recovery that came in low, a particle count that appeared between timepoints, an impurity ratio that shifted between two shipments of nominally identical material. The conditions that produce it are all knowable in advance, and all of them are decisions — what the sample is dissolved in, what it is sealed in, how full the container is, and whether it is thawed before or after the carbon dioxide has been allowed to leave.

Carbon dioxide is not a bystander in aqueous samples

What the container decides

Which peptide chemistry the shift actually touches

The lyophilised vial is a different case entirely

Reading the shipping method as a variable

Further reading