{"id":4011,"date":"2026-10-06T09:00:00","date_gmt":"2026-10-06T07:00:00","guid":{"rendered":"https:\/\/modafinil.pl\/hplc-purity-determination-what-the-method-sees-and-what-it-misses\/"},"modified":"2026-10-08T10:29:59","modified_gmt":"2026-10-08T08:29:59","slug":"hplc-purity-determination-what-the-method-sees-and-what-it-misses","status":"publish","type":"post","link":"https:\/\/modafinil.pl\/en\/hplc-purity-determination-what-the-method-sees-and-what-it-misses\/","title":{"rendered":"HPLC Purity Testing: What the Method Sees and Misses"},"content":{"rendered":"

Liquid chromatography is the default method for purity determination, and that is precisely why its assumptions tend to go unnoticed.<\/strong> A result of “99.3% by HPLC” looks like an absolute measurement, yet it is the sum of several silent assumptions: that everything in the sample came off the column; that it came off separately; and that the detector saw it in proportion to its amount. Any of these assumptions can fail to hold, and each fails in the same direction \u2014 inflating the result.<\/p>\n

This text describes exactly where the limits of the method lie, when an independent technique is needed, and how to read a chromatographic result without attributing to it a meaning it does not have.<\/p>\n

The detector sees a chromophore, not a substance<\/h2>\n

The most common setup is ultraviolet detection. A signal arises when a molecule absorbs radiation at the selected wavelength \u2014 that is, when it has a suitable system of conjugated bonds. A substance without a chromophore at that wavelength is transparent to the detector: it will pass through the column, elute from it, and leave no trace on the chromatogram.<\/p>\n

Hence the first consequence: area normalization<\/em>, i.e., calculating the share of the main peak in the sum of all peaks, tacitly assumes that all components of the sample produce a signal. A single impurity without a chromophore is enough to inflate the result \u2014 and without any visible symptom on the chromatogram.<\/p>\n

The second consequence is more subtle: even when all components absorb, they do so with different intensities. The same mass fraction gives a different peak area for different substances. This is why serious assay determinations are performed against a reference standard rather than by area normalization alone \u2014 and why purity determination methods are being developed in variants with internal standard correction.1<\/a><\/sup><\/p>\n

There are also detectors that answer different questions. Polarimetric detection, described in the chromatographic literature as early as the late 1980s, responds to optical activity and makes it possible to see what an absorbance detector cannot distinguish.2<\/a><\/sup> The choice of detector is therefore a choice of which impurities stand a chance of being noticed.<\/p>\n

Coelution: two compounds, one peak<\/h2>\n

If an impurity leaves the column at the same time as the main substance, its area is added to the main peak. The purity result goes up, and the chromatogram looks exemplary \u2014 a single tall, symmetrical peak.<\/p>\n

This is exactly why dedicated procedures for testing peak homogeneity were developed. Principal component analysis of diode array detector data makes it possible to check whether the spectrum recorded on the leading edge, at the apex, and on the trailing edge is the same spectrum.3<\/a><\/sup> If it is not \u2014 two substances are sitting under a single apex.<\/p>\n

This verification is not a formality. Chromatographic methods for specific substances are sometimes refined over years precisely in order to separate what previously eluted together \u2014 this applies both to pharmacopeial methods4<\/a><\/sup> and to determinations developed from scratch for difficult systems.5<\/a><\/sup> Method development is largely a history of successive separations that had not been seen before.<\/p>\n

What does not come off the column<\/h2>\n

The third silent assumption states that everything that went in came out. That is not always the case. Strongly retained substances may leave the column long after data acquisition has ended \u2014 and if the run time has been shortened, they simply will not appear in the result. Others bind irreversibly to the packing and do not elute at all.<\/p>\n

The symptom may be baseline drift or ghost peaks in the next injection. The symptom may also be silence \u2014 and that is the most difficult case, because it looks identical to a clean sample. Work devoted to the purification of compounds with difficult chromatographic behavior shows how much effort must go into simply establishing what elutes from the column and what does not.6<\/a><\/sup><\/p>\n

When chromatography alone is not enough<\/h2>\n

The conclusion from the above is not that the method is unreliable \u2014 it is that the method answers a narrower question than is usually attributed to it. This is why the certification of materials relies on orthogonal techniques based on a different physical principle.<\/p>\n

Quantitative nuclear magnetic resonance<\/strong> measures the number of nuclei, not absorbance, so an impurity without a chromophore is not invisible to it. The use of proton spectra for determining the purity of pharmaceutical reference materials is described in the metrological literature.7<\/a><\/sup><\/p>\n

Mass balance<\/strong> approaches the problem from the side of the impurities: water, residual solvents, and residue on ignition are determined, and the content of the substance is calculated by difference. This is how, among others, amino acid reference materials were certified.8<\/a><\/sup><\/p>\n

Combining two independent methods<\/strong> is the strongest solution. In the certification of a caffeine<\/a> reference material, the spectrophotometric result was verified by quantitative NMR \u2014 each method has a different blind spot, so agreement between the results carries information that neither of them would provide on its own.9<\/a><\/sup><\/p>\n

How to read a chromatographic result<\/h2>\n

A few questions worth asking before accepting the number.<\/p>\n