Every quality piece on this site so far has treated the vial as a neutral object. Purity is a property of the powder. Mass is a property of the molecule. Endotoxin is a property of the process. The glass and rubber around all of it were scenery.
That framing is wrong, and the packaging literature has known it is wrong for decades. A sealed vial is a small closed system in which several materials are in prolonged contact. Glass has a surface chemistry. Rubber holds water and releases it. Metal ions migrate. The seal either holds or does not, and whether it holds is a measurable property somebody either measured or did not.
None of this appears on a typical research-peptide certificate of analysis. That absence is why it is worth explaining.
The container closure system is a defined object
In regulatory language a container closure system (CCS) is the sum of the packaging components that contact or protect the product: for a lyophilized vial, the glass, the elastomeric stopper, the aluminium crimp seal, and the gas in the headspace above the cake.
This vocabulary is being rewritten right now. On 14 August 2026 the FDA published a draft guidance, Container Closure Systems for Human Drugs and Biological Products, with comments open until 13 October 2026. The 37-page draft supersedes the May 1999 guidance of nearly the same name and its May 2002 questions-and-answers companion. It sets out a risk-based framework in which selection, qualification and control of a container closure system must be scientifically justified and linked to the product's specific quality risks, and states that further guidance on extractables and leachables assessment will follow.
Two qualifications. This is a draft in a comment period, not a settled requirement, and it governs approved drugs and biologics rather than research compounds. It is the framework the field uses to think about the problem, not a standard anything on a research shelf is held to.
Glass is a chemical participant, not a wall
USP General Chapter <660> Containers—Glass classifies glass by hydrolytic resistance rather than by recipe. Type I is borosilicate, high resistance from the composition itself, suitable for all products. Type III is soda-lime-silica with moderate resistance. Type II is Type III whose inner surface has been treated to raise that resistance to a high level.
The failure mode that made this public is delamination — thin flakes of glass, called lamellae, shedding from the inner surface into the contents. In March 2011 the FDA issued an advisory to drug manufacturers on the formation of glass lamellae in certain injectable drugs, following a series of recalls. Documented risk factors include vials made by the tubing process at higher forming temperatures, formulations at high pH, and terminal sterilization. USP later added General Chapter <1660>, Evaluation of the Inner Surface Durability of Glass Containers, with approaches for predicting whether a glass and product combination is likely to shed particles.
There is a second, quieter glass effect specific to this shelf. The inner surface carries anionic silanol groups, and several catalog peptides are strongly net cationic in solution. The charge and solubility primer covered the measured consequence: for cationic membrane-active peptides, a large fraction of material can be lost from solution to container walls. Neither delamination risk nor adsorption behaviour is answerable from a purity percentage. Both are properties of a pairing, not of a powder.
The stopper is the largest hidden variable
Elastomeric closures have their own chapters — USP <381> for physical and chemical properties, USP <382> for functional suitability. The reason they matter for a lyophilized peptide has nothing to do with either, and everything to do with water.
Stoppers are steam sterilized. Steam drives moisture into the rubber. If the subsequent drying step is inadequate, that water leaves the stopper afterwards, into whatever is sealed underneath it.
Donovan and colleagues at Lilly Research Laboratories quantified this (PDA Journal of Pharmaceutical Science and Technology 2007;61(1):51-58). Comparing two commercial 13 mm lyophilization stoppers designated low-moisture and high-moisture uptake, they found both absorbed significant water during sterilization, the high-uptake stopper substantially more. Reaching not more than 0.5 mg per stopper required roughly 2 hours of drying at 105 °C for the low-uptake stopper and roughly 8 hours for the high-uptake one. After drying, stopper moisture re-equilibrated rapidly with ambient conditions, and the apparent equilibrium level was about seven times higher in the high-uptake stopper. Transfer into a model lactose cake depended on the stopper's initial water content and on storage temperature. The authors projected shelf-life uptake from product-contact surface area and concluded that stopper handling matters most for low-fill-mass lyophilized products.
Research peptide vials are, characteristically, very low fill mass. A thin lyophilized film under a full-size stopper is close to the worst case in that paper.
The consequence for documentation is direct. Residual moisture measured by Karl Fischer at release is a measurement taken before the stopper has finished handing over its water. Later in shelf life a second mechanism takes over: ambient moisture permeating through the rubber, at a rate set by the elastomer's moisture vapour transmission characteristics. Water is both a stability driver, per the degradation pathways piece, and a mass-accounting input, per net peptide content — and it arrives on a schedule nobody prints.
Leachables: the metal-ion case
Extractables are what a packaging material can release under forcing conditions. Leachables are what it does release into a given product over time. Only the second is about the actual pairing.
The most directly relevant published example involves a compound class on this shelf. Christensen and colleagues at Novo Nordisk (European Journal of Pharmaceutics and Biopharmaceutics 2007;66(3):366-371) studied the physical stability and secondary structure of a glucagon-like peptide-1 derivative in the presence of Al³⁺, Zn²⁺, Mg²⁺ and K⁺, named in the paper as possible leachables from container closure systems, across 0-50 ppm. Magnesium and potassium left the stability profile unchanged. Aluminium and zinc produced a concentration-dependent rise in thioflavin T fluorescence, and infrared spectroscopy confirmed the destabilisation as a shift in secondary structure from predominantly alpha-helix toward higher beta-sheet content.
Three caveats stated plainly. This was a liquid formulation, not a dry cake. The concentrations tested run above what shelf-life leaching typically produces, so this is mechanism rather than a shelf-life prediction. And it is one peptide, in a preliminary study, from 2007.
What makes it worth citing is the specificity of the mechanism. A beta-sheet shift is the structural route to aggregation, which the charge primer identified as the one degradation pathway that is not a covalent reaction — nothing is hydrolysed, oxidised or deamidated, so synthesis-time purity does not anticipate it. A trace metal from the closure is a plausible input to that pathway for compounds like semaglutide and its relatives, and it is the class of question the FDA's promised supplementary guidance is aimed at.
Integrity is a measurement, and the methods changed
Whether the seal holds is covered by USP <1207> Package Integrity Evaluation—Sterile Products, with subchapters on method selection and validation (<1207.1>), leakage test technologies (<1207.2>), and package seal quality (<1207.3>).
The chapter's organising distinction is between probabilistic and deterministic methods. Dye ingress — immerse the sealed vial in dye under vacuum, look for colour inside — is the legacy probabilistic test, and <1207> characterises it as less sensitive and less reproducible, its performance depending on dye concentration, vacuum level and exposure time. The deterministic set includes vacuum decay, helium leak detection, high-voltage leak detection and laser-based headspace analysis, which <1207.2> places in its highest leak-detection sensitivity band. For fairness, a BioPhorum industry position paper has argued that framing depicts ingress methods unfairly and that they retain valid applications.
One subtlety worth carrying: <1207.3> seal quality tests are not leak tests. They characterise the seal, which is related to but not the same as showing the package does not leak.
The one measurement that does not destroy the sample
The lot and sampling piece made an uncomfortable point: the most definitive analytical methods consume the material. Amino acid analysis requires acid hydrolysis. Karl Fischer consumes sample. So does the endotoxin test, and sterility. You are guaranteed never to hold the vial that was tested.
Headspace analysis is the exception. Near-infrared laser light tuned to a water absorption frequency is passed through the glass, above the cake. Absorption is proportional to water vapour concentration in the headspace; the width of the absorption line relates to headspace pressure. Nothing is opened.
Cook and Ward (PDA Journal of Pharmaceutical Science and Technology 2011;65(1):2-11) compared frequency modulation spectroscopy against Karl Fischer coulometric titration. The operational payoff was that individual vials could be re-analysed at a series of time points as a long-term monitoring exercise. They found a linear relationship between headspace moisture and Karl Fischer values for freeze-dried sucrose, examined the effect of stopper treatment and storage conditions, and mapped moisture across every vial on one dryer shelf to characterise intra-batch variability — the same heterogeneity the lot piece raised via the edge vial effect.
More recently, Pu, Li and Xiang at Bristol-Myers Squibb (Journal of Pharmaceutical Sciences 2023;112(3):859-867, online November 2022) built both a near-infrared method calibrated against Karl Fischer by partial least squares regression and a laser headspace method converting measured water vapour pressure directly to water content through a dynamic vapour sorption isotherm, bypassing Karl Fischer calibration entirely. Both agreed with Karl Fischer to a root mean squared error of prediction below 0.15%. The authors suggest off-line Karl Fischer could potentially be replaced by an at-line laser headspace method, and flag that quantitation limit and generality across formulations need further work.
This does not dissolve the sampling problem. It changes its shape. Headspace water is not cake water; it is a proxy read through a relationship established for that formulation. But it is the one route by which a specific sealed container can be characterised, repeatedly, and still exist afterwards.
None of it is a test you should expect on a research-peptide certificate of analysis. Glass type, stopper formulation and drying history, seal integrity method, headspace gas and pressure are container specifications, and a certificate of analysis is a document about material. The transfer is the one this series keeps arriving at: a specification means something only when it names an attribute, a method and a condition. The container is a condition. When a purity figure and a moisture figure are reported with no statement of what they were measured inside, the numbers are still real and the system that produced them is undescribed.
Questions researchers ask
Should I expect container information on a COA? Almost never, and its absence is not evidence of a problem. It means the question was not asked. The useful habit is knowing which questions a certificate structurally cannot answer, alongside chirality, folding and metal stoichiometry from earlier pieces.
Is a larger vial better? No general answer, and the Lilly stopper work suggests the relevant relationship is fill mass against stopper contact area rather than vial size alone.
Does this matter more for some compounds than others? Predictably so. Cationic sequences carry the documented surface-adsorption problem. Compounds where folding carries the activity, like IGF-1 LR3, have more to lose from physical destabilisation than a short unstructured peptide such as BPC-157. As with everything on the quality side of this library, the container question is a property of a pairing rather than of a category.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.