Why the Container Matters: Glass, Plastic and Sample Loss

A sample can lose material from solution without spilling a drop. Container surfaces, transfers and filters can all shape what an analytical instrument finally measures.

Consider a hypothetical laboratory dividing one research sample between two clean tubes. The liquid looks identical in both. Later, analysis reports less of the target compound in one tube. Nothing visibly leaked, and the instrument appears to be working. This hypothetical puzzle has a less obvious suspect: the surface surrounding the sample. Published studies show that recovery from a vial can influence both the measured signal and its repeatability.^[2]^[3]

The wall can become part of the sample

Adsorption means molecules accumulate on a surface.^[1] Material attached to the container wall is no longer represented in the same way by an aliquot of the liquid.^[1] In 2015, Kasper Kristensen, Jonas Henriksen and Thomas Andresen measured this effect for mastoparan X, melittin and magainin 2 using analytical HPLC.^[1]

For their 220-microlitre, one-micromolar samples incubated for an hour, only about 10–20% was recovered from the tested borosilicate glass and ordinary polypropylene containers.^[1] Recovery depended on concentration and the ratio of surface area to liquid volume; repeated transfers into fresh containers caused further losses.^[1] These were three cationic, membrane-active peptides under specified experimental conditions.^[1] Their measured losses cannot be assigned to GLOW, KLOW or an untested formulation.^[1]

The useful lesson is the existence of a measurable interaction. The visible liquid is therefore only part of the story: the experiment must also account for what remains associated with its surroundings.^[1]

Glass versus plastic is an incomplete comparison

A material name alone does not settle the choice. Miriam Goebel-Stengel and colleagues compared recovery of several iodine-125-labelled endocrine peptides from glass and plastic surfaces.^[2] Their abstract describes substantial differences among the tested combinations; siliconization reduced recovery in their experiments, whereas bovine serum albumin improved it.^[2] The investigators measured radioactivity, and the compounds, labels and conditions belong to that study.^[2]

That finding makes a broad instruction such as always use glass difficult to defend. It also shows why a surface treatment deserves evidence of its own. A modification intended to improve one interaction does not establish a benefit for every molecule. The appropriate comparison concerns the actual compound and experimental conditions, including what else is present in solution.^[2]

A low-binding label needs a test context

Eppendorf’s 2016 Application Note 382 offers a useful example of how to read a commercial comparison.^[5] Rafal Grzeskowiak, Sandrine Hamels and Eric Gancarek, all affiliated with Eppendorf organizations, compared tubes using fluorescently labelled bovine serum albumin in phosphate-buffered saline.^[5] Their experiments used a specified low protein concentration and a 24-hour room-temperature incubation; the company’s Protein LoBind tubes gave the highest recovery in that comparison.^[5]

This is product-specific manufacturer evidence, not an independent ranking of all laboratory containers.^[5] The relevant question is how closely the tested protein, liquid, concentration, contact time and tube match a proposed analysis. A percentage printed beside a product name belongs to its test conditions. Removing those conditions turns a useful measurement into an unsupported promise.^[5]

The liquid and the instrument must work together

Container recovery also belongs to the analytical method. In a 2007 paper, Paul van Midwoud and colleagues investigated a cytochrome c digest by nanoLC–MS.^[3] Their abstract links poor repeatability chiefly to recovery from the sample vial.^[3] Changing the liquid composition improved performance, but the authors also identified a tradeoff: a change that helps material stay in solution can reduce its affinity for the reversed-phase separation material.^[3]

Their study included dimethyl sulfoxide as an organic modifier.^[3] That is evidence about their analytical system, not a transferable solvent recipe.^[3] Reconstitution names a preparation operation; it does not identify the liquid used. Naming the operation therefore does not resolve whether the resulting solution is compatible with the container or the measurement.^[3]

Small losses can accumulate along a route

A numerical illustration makes the distinction between one container and a whole preparation route easier to see. Suppose a purely hypothetical sample begins with 100 units of measurable material. Assume that each transfer leaves 10% of the incoming material unavailable to the next step, that nothing is recovered later, and that no other change occurs. These are invented assumptions, not measured container-loss rates.

Hypothetical stage	Material entering the step	Material available afterward
First transfer	100 units	90 units
Second transfer	90 units	81 units
Third transfer	81 units	72.9 units

The three losses total 27.1 units, because each percentage applies to what remains. The calculation does not predict a real preparation; actual surfaces may behave differently at each stage. It explains why a test of the final vial alone would not describe everything that happened earlier.

There is also a useful distinction between repeatability and recovery. In this example, a perfectly consistent route would produce 72.9 units every time. Agreement among those repeated outcomes would be excellent, even though the route delivered less material than it received. The missing question would still be how the final amount relates to the starting amount.

Filters add another surface—and another kind of loss

A filter can affect both the concentration of an analyte and the volume of liquid collected.^[4] Sigma-Aldrich’s manufacturer study compared membrane filters using several pharmaceutical formulations.^[4] Acetaminophen and acetylsalicylic acid showed strong binding to the tested nylon filters, while caffeine did not show binding to the three membrane types in that comparison.^[4] The study separately measured liquid retained by filter devices.^[4]

These observations distinguish selective binding from liquid simply remaining in the device.^[4] They also explain why membrane name and pore size alone do not establish recovery. The report found that processing differences between samples and standards affected the calculated recovery.^[4] Its formulations and commercial filters provide examples, rather than proof that a particular membrane suits a different research material.^[4]

The opening puzzle therefore leads to a better way of reading an analytical result: consider the route the sample travelled. The container, liquid, transfers and filter help define that route. A convincing recovery claim describes enough of it to show what was actually tested—and enough of the comparison to show what the result means.^[4]

Research and attribution note

This article explains published research and manufacturer documentation; it reports no Chudik experiment. The Goebel-Stengel and van Midwoud papers were reviewed through their official PubMed abstracts. Other cited material was read in full text or its relevant body sections. The transfer table is hypothetical.

For the broader measurement discussion, read Inside the Vial.