Loss on drying (USP <731>)Loss on Drying

USP <731> Loss on Drying determines the loss of mass from a sample under conditions specified in the individual monograph or the material's own specification: a set temperature, a set apparatus (oven, vacuum oven, or desiccator over a defined desiccant), and a set time or a 'dry to constant weight' endpoint. The result is reported as percent loss on drying, calculated from the sample's initial and final mass. It is one of the most widely used tests in a pharmaceutical, supplement, or food QC laboratory because it requires no specialized reagent chemistry, runs on inexpensive equipment, and gives a fast answer that correlates well with the property manufacturers actually care about: how much of the sample is not the intended dry substance.

The single fact that governs everything downstream is that LOD is a mass-loss test, not a chemical-identity test. Whatever volatilizes under the specified conditions — water, residual processing solvent, some essential-oil constituents in botanical material, or even a thermally labile portion of the active itself — is counted as 'loss on drying' indiscriminately. USP <731> does not distinguish the chemical nature of what left the sample; it only weighs what remains.

USP <731> describes the procedure generically (analytical balance, drying apparatus, tared container, weigh before and after drying to constant weight or for a specified time) and leaves the specific apparatus to the monograph or in-house specification. In practice, four apparatus choices dominate regulated labs.

The IR/halogen moisture balance deserves separate treatment because it is not itself a USP <731> method — it is a rapid instrumental technique that a laboratory validates against the compendial procedure and then uses as a routine, faster substitute. An IR balance heats the sample directly on the weighing pan with a halogen or infrared emitter and continuously tracks mass loss, typically ending on a programmed criterion (e.g., less than 1 mg change per 90 seconds, or a fixed time). Because the heating profile, sample geometry, and airflow of an IR balance differ materially from a convection oven, a bare transfer of the monograph's oven conditions onto the IR balance's temperature setting does not guarantee an equivalent result — bridging studies are required (see below).

USP <921> Karl Fischer titration is the specific, stoichiometric method for water content: an iodine-based redox reaction consumes water 1:1 (via the Bunsen reaction with SO2 and a base), and the titrant delivered is directly proportional to water mass. Method I (azeotropic distillation), Method Ia (direct titration), and Method Ic (coulometric) are the three sub-methods in <921>; coulometric KF is the standard choice for low-water-content samples (sub-1%) because it does not require standardizing a volumetric titrant.

The decision of which method governs the specification is a method-selection judgment that belongs in validation, not at the bench. If a raw material's monograph specifies LOD, use LOD as written — but if in-house data show the LOD result runs materially higher than a confirmatory KF result on the same lots, that gap is residual solvent or another volatile, and it should be characterized (typically by GC headspace) rather than silently absorbed into the moisture number.

Three material classes drive most of the method-selection decisions a lab actually faces.

1. Crystalline hydrates (e.g., citrate, sulfate, or hydrochloride salts with defined waters of crystallization) — LOD is often appropriate because the water is the only volatile species present in a well-characterized crystal form, but the drying temperature must be set below the hydrate's dehydration/desolvation transition or the LOD result will overshoot the true bound-water content and can even trigger a polymorph change during the test itself.
2. Hygroscopic actives and excipients — materials that pick up ambient moisture quickly need LOD (or KF) run under controlled humidity handling (glovebox or rapid weighing) because the act of opening the container and weighing the sample can materially change the result before drying even starts. These materials also need tighter re-test-interval controls because their moisture — and therefore their as-is potency — drifts in storage.
3. Residual process solvents (ethanol, isopropanol, acetone, methylene chloride, botanical extraction solvents) — LOD cannot distinguish these from water. ICH Q3C-listed solvents at meaningful levels require GC headspace quantitation as the specific method; KF gives the true water figure so the solvent contribution can be inferred as (LOD − KF) if a direct GC method is not yet validated, but that inference should be a bridging exercise, not a permanent substitute for direct solvent quantitation.

An infrared or halogen moisture balance is an instrument, and like any instrument used to generate a GMP result it needs installation/operational qualification (IQ/OQ) plus a bridging (equivalence) study before it can replace the compendial oven method for release testing.

1. IQ/OQ — balance calibration (linearity, repeatability) against NIST-traceable weights; heater temperature verification with a calibrated reference thermocouple at the pan surface, not just the setpoint display.
2. Method development on the IR balance — determine the temperature and endpoint criterion (time-based or rate-of-change-based) that reproduces the compendial oven result for the specific material, not a generic manufacturer default.
3. Bridging study — run a statistically justified number of lots (typically ≥6, spanning the specification range) by both the compendial oven/KF method and the candidate IR method; demonstrate equivalence with a predefined acceptance criterion (e.g., mean difference within ±0.2% and no systematic bias by regression).
4. Ongoing verification — periodic (e.g., quarterly) confirmation that the IR method still agrees with the reference method, since heating-element ageing and pan geometry wear can drift the correlation over time.
5. Change control — any change to sample mass, pan type, or endpoint criterion on the IR method requires re-bridging, because IR mass-loss kinetics are sensitive to sample bed depth and surface area in ways the oven method is not.

An LOD specification is not simply 'NMT X%' copied from a monograph without local verification. Setting a defensible LOD limit requires reconciling three inputs: the compendial or supplier-typical range, the material's demonstrated behavior across representative lots (process capability), and the potency-correction sensitivity — how much the finished-batch potency moves per percentage point of LOD change, given the formulation's tolerance window.

A common and defensible pattern for hygroscopic actives is a tiered spec: an incoming-acceptance LOD limit (wider, because supplier control is imperfect) and a tighter dispense-eligibility check that flags a lot for re-test if its LOD, measured at time of use, has drifted meaningfully from the receipt value — since a lot can be received compliant and become non-compliant on the shelf.

Once an LOD (or KF) result is available, the arithmetic that turns it into a dispense correction is mechanical but must be applied in the right order relative to potency factor and any salt-to-base conversion.

1. Determine the basis the assay/potency value is reported on. If the CoA potency is reported on an anhydrous basis but the material as received still contains its native moisture, the as-is mass must be inflated by (1 − LOD%/100)⁻¹ to deliver the anhydrous-equivalent target.
2. Example: target anhydrous-active mass is 500 g; lot LOD is 3.8%. As-is dispense mass = 500 / (1 − 0.038) = 519.8 g. Dispensing 500 g as-is would under-deliver active by roughly the LOD fraction.
3. If a salt-to-base conversion also applies, normalize salt-to-base first, then apply the LOD/anhydrous correction, then apply assay potency factor last, so rounding at each step does not compound unpredictably.
4. If the assay value on the CoA is already reported as-is (not anhydrous), do not apply the LOD correction a second time — this double-counts moisture and produces a significant over-dispense. Confirming the CoA's declared basis before computing anything is the single highest-value check in this whole workflow.

• LOD run at a temperature/time not matching the validated specification, with no record of why the deviation was acceptable.
• 'Dry to constant weight' claimed without the intermediate weighings that demonstrate the endpoint was actually reached.
• IR moisture balance used for release testing with no bridging study against the compendial oven/KF method, or a bridging study that has never been re-verified.
• LOD result treated as pure water content and fed directly into a Karl Fischer-based specification limit, or vice versa, without acknowledging the methods measure different things.
• Desiccator/oven not included in the calibration and preventive-maintenance program, so temperature drift over months goes undetected.
• Moisture result used to compute an anhydrous-basis dispense correction with no record of which basis the CoA potency value itself was reported on.
• Hygroscopic raw material's LOD not re-checked at or near the point of use despite a known moisture-uptake profile documented in its own stability data.