The purity of a synthetic intermediate governs the yield that can be claimed for a step and the stoichiometry of the step that follows, yet it is often estimated from a chromatogram, which reports an area percent that need not equal a mass percent because different compounds absorb or ionise differently. Quantitative proton nuclear magnetic resonance, abbreviated qNMR, offers a more direct route. The area of a resonance is proportional to the number of protons giving rise to it, with a proportionality constant that is the same for every proton in a spectrum acquired correctly, so the ratio of two areas is the ratio of two molar quantities of protons without the need for a compound-specific response factor (Malz and Jancke, 2005; Pauli et al., 2005). The principle is old: a proton spectrum was used to quantify the components of an analgesic mixture by Hollis (1963).

The customary way to turn that ratio into an absolute amount is to add an internal standard, a pure compound of known mass whose protons are counted against the analyte's. Benzoic acid, dimethyl terephthalate, maleic acid and 1,3,5-trimethoxybenzene are common choices, the properties that make a good internal standard having been surveyed by Rundlof et al. (2010), and certified reference materials for the purpose are available (Weber et al., 2015). The internal standard is accurate and simple, but it must dissolve in the same solvent, stay chemically inert toward the analyte, and show at least one signal clear of every analyte resonance. For a precious intermediate the method also consumes sample and contaminates the recovered material.

These constraints have encouraged references that are not mixed with the sample. This article proposes a stepwise procedure for assigning the purity of synthetic intermediates by proton qNMR without an internal standard. It gathers the external-reference options into one workflow, states the assumptions each makes, and describes the acquisition and processing needed for areas to mean what the calculation requires. The procedure is a proposal built from theory and published method; it reports no measurements of its own and has not been validated here.

Two bodies of knowledge underlie the method: the conditions that make a proton spectrum quantitative, and the ways a reference can be supplied from outside the sample. The pharmaceutical applications of the technique are reviewed by Holzgrabe (2010).

The procedure has six steps, listed in Table 1 with the output of each and shown in sequence in Figure 1. The steps are described in turn below. Throughout, the number of protons per molecule for the integrated multiplet must be known from the structure, which for a synthetic intermediate of defined constitution is the usual case.

Step one is the choice of route. Where the sample can spare a few milligrams and no contamination matters, an internal standard remains the most accurate option and should be preferred; the external routes are for when it cannot. A coaxial insert suits a stable reference and a sample whose spectrum is crowded, since the reference peak can be placed in a clear region by choosing the reference compound. The electronic route suits the smallest samples and automated runs but depends wholly on the calibration holding between that calibration and the measurement.

The attraction of an external reference in a synthesis laboratory is that the sample is recovered unchanged and uncontaminated, which matters when the material is scarce or is to be carried forward. The electronic route further removes the need to keep and weigh a standard, which suits routine checks on many samples. These gains are bought with a stronger dependence on instrument stability and on careful matching of conditions, so the laboratory that adopts the method should calibrate on a schedule, keep tube and volume constant, and verify the calibration against a weighed internal standard from time to time on a representative sample. Used this way the method gives an absolute purity with a clear audit trail rather than an area percent.

This article proposes a procedure and does not test it. Several things would need to be established before it could be recommended without reservation. One is agreement: whether, on representative intermediates, the external, coaxial and electronic routes return the same mass fraction as a well-executed internal-standard determination, within the stated uncertainty. A second is robustness: how much the result drifts with the time between calibration and measurement, with small changes in sample volume, and with routine retuning and temperature variation. A third is reach: how the method performs on intermediates with few protons, with very long relaxation times, or with signals that overlap the usual reference regions. Each can be examined with samples a laboratory already holds, and each is a natural subject for a validation study.

Two cautions apply whatever the route. Quantitative NMR measures the protons it can integrate, so an impurity that is invisible by NMR, an inorganic salt or a proton-free species, is not counted and must be determined separately before a mass balance can close. And the number of protons assumed per molecule must be correct; an error in the assigned structure propagates directly into the purity.

Purity can in principle be assigned to a synthetic intermediate by proton qNMR without mixing a standard into the sample, by referencing an external measurement, a sealed capillary or a calibrated electronic signal. The six-step procedure set out here keeps the acquisition quantitative, makes the reference and its assumptions explicit, and reports the result as a mass balance with an uncertainty that includes the extra component the external routes introduce. Whether the routes agree with the internal standard closely enough to replace it for a given class of intermediate is an empirical question that this article leaves open and that practitioners are well placed to answer.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.