Measuring water content by Karl Fischer

Water is the smallest thing on a peptide certificate and one of the most consequential. It dilutes the peptide mass you weigh, plasticises the cake, enables hydrolysis, and is the first number to look at when a lot behaves differently from the last one. It is also the number most often reported without a method, which makes it impossible to compare between suppliers.

Why the number matters three times over

Water enters three separate calculations, and confusing them is a common source of disagreement:

As weight. A vial labelled 10 mg contains 10 mg of powder, not 10 mg of peptide. Water is part of that powder, so it reduces the peptide mass available for an experiment.
As a stability variable. In the solid state, water is the medium in which hydrolysis and some aggregation pathways proceed. A cake above its intended moisture level ages faster, even at the same storage temperature.
As a release result. Water content is a lot-specific measurement that belongs on the certificate with a limit and a method — not a generic “complies”.

The arithmetic that combines water, counter-ion and purity into net peptide content is set out in net peptide content and cost per mg. This article is about how the water figure itself is obtained and how to read it.

What the methods actually measure

Method	Typical range	What it measures	Main pitfalls
Loss on drying (LOD)	0.1–100%	Anything volatile that leaves the sample under the specified conditions	Counts residual solvents and some volatile degradants as “water”; result depends on temperature and time
Karl Fischer, volumetric	~0.1–100%	Water by a stoichiometric reaction with iodine	Sample size must be large enough; side reactions with aldehydes and ketones; reagent drift
Karl Fischer, coulometric	~10 µg to a few %	Water by electrochemical generation of iodine, quantified by charge	Small samples are less representative; drift and side reactions matter more at low levels
Oven + Karl Fischer	Method-dependent	Water released from a solid on heating, carried into the titration cell	Oven temperature must release water without decomposing the peptide or driving off solvents
Near-infrared (NIR)	Calibration-dependent	Water by a multivariate model	Only as good as the calibration set; not a primary method

Karl Fischer is the reference method for water because the chemistry is specific: one mole of water reacts with one mole of iodine in the presence of sulfur dioxide and a base. Volumetric KF delivers iodine from a titrant of known concentration; coulometric KF generates iodine at the anode and calculates the amount from the current and time, using Faraday's law. For a lyophilised peptide at a few percent moisture, either can work; for low-level work, coulometric is the usual choice.

What “loss on drying” is not

LOD is a gravimetric method: weigh, heat under defined conditions, weigh again, call the difference water. It is cheap and it is not specific. A peptide that retains acetonitrile or TFA, or that loses a volatile protecting-group by-product, will report a loss that is not water. Conversely, a sample that takes up water during cooling before the second weighing can report too little.

This matters when the two numbers are compared across suppliers. “Water content ≤5% by LOD” and “water content 2.1% by coulometric KF” are not the same measurement, and a difference between them may reflect the methods rather than the material. When a certificate says “water” without naming the method, the defensible assumption is that the number cannot be compared with anything.

Sampling is the hard part

A peptide cake is hygroscopic. The moment a vial is opened in a humid room, the surface layer begins to take up water, and a sample taken from the top of the cake may not represent the vial. Three practical consequences:

Equilibrate sealed vials to room temperature before opening. Cold glass attracts condensation on the inside as well as the outside.
Work fast, and take the sample from a consistent place. If the method samples the whole vial, say so. If it samples a portion, record how it was taken.
Use a dry glove box or a low-humidity enclosure for hygroscopic material. For a 1–2 mg sample in a coulometric cell, a few micrograms of surface water is a large relative error.

The sample size also matters. Coulometric cells are sensitive at the microgram level, so a 1 mg sample of a 2% moisture powder contains about 20 µg of water — measurable, but only if the blank and the drift are under control. A 10 mg sample is more representative of the vial but may overload a small cell. The method should state the sample size, because the result depends on it.

Interferences and drift

Karl Fischer is specific for water, but not infinitely so. The classic interferences are:

Aldehydes and ketones can react with methanol in the reagent to form acetals or ketals, releasing water and producing a positive bias. Some are worse than others, and the effect is time-dependent.
Strong acids and bases shift the reagent pH away from its optimum and change the reaction stoichiometry.
Oxidising and reducing substances can react with iodine or with the iodide system directly.
Incomplete dissolution leaves water trapped in the solid, which is why oven methods exist for insoluble material.

Drift is the instrument's background signal: water entering the cell from the atmosphere, from the reagent or from a leak. A method that does not report a stable drift before injection and a blank value is not under control, particularly at low moisture levels. For a certificate, the useful fields are the result, the method (volumetric, coulometric or oven), the oven temperature if used, and the sample size.

How to read a water result on a COA

What the COA says	What it establishes	What it does not
“Water: 3.2% (Karl Fischer, coulometric)”	A lot-specific measurement by a named method	Whether the sample was representative, or whether the limit is appropriate for the product
“Water: ≤5%”	A specification limit exists	The measured value; a limit is not a result
“Water: complies”	Almost nothing	Any number, any method, any basis
“Loss on drying: 4.1%”	Volatile content under stated conditions	That the volatile material is water
No water field at all	—	Whether the material was dried, and how much peptide the vial actually contains

A single result is also less informative than a trend. Three lots at 1.8%, 2.0% and 2.1% describe a controlled process; three lots at 2%, 4% and 6% describe a drying step that is drifting, even if all three are within a 8% limit. The trend view is the same one used for purity in lot-to-lot consistency.

What to specify

For research-grade material, a workable specification has four parts:

A limit appropriate to the peptide and the container. Limits are product-specific; a common research convention is a few percent, but the number should be justified rather than copied.
A method — coulometric KF, volumetric KF or oven KF — named on the certificate.
A per-lot result, not a range and not “complies”.
A storage statement that matches the sensitivity: sealed, desiccated, and equilibrated before opening.

If the water result and the peptide content do not reconcile — for example, a 98% peptide content alongside 6% water and a counter-ion — the mass balance is the place to look. The components have to add up, and a certificate where they do not is telling you something about how the numbers were produced.

For laboratory research use only. This article describes analytical methods and specification practice; it makes no claim about the water content of any specific product and does not support administration to humans or animals.

Editor's take

Water is the quiet variable. It changes the weight you measure, the stability you observe and the price you effectively pay — and it is invisible in a purity percentage.

Sources & evidence

Method validation: ICH Q2(R1) — ICH quality guidelines.
Specifications: ICH Q6B — same ICH page.
Water / particulates / endotoxin: USP <731>, <788>, <85> — USP.
Chinese Pharmacopoeia (ChP 2020) Part IV: chapters 0512, 0832, 0903, 1143 — cited by chapter number.

No laboratory testing was performed by the author for this guide; acceptance limits must come from your own specification.

Limitations

This guide describes methods and interpretation, not acceptance limits. Limits must come from your own specification and, where applicable, the relevant pharmacopoeial monograph. No testing was performed by the author for this page.