Report 163 · Lab Science
Absorbance equals molar absorptivity times concentration times path length. Every chemistry student learns A = εcl, and most labs quantify with it daily. A 2020 review by three spectroscopists in Jena argues that it is not a law of nature at all, but a limiting law in the same family as the ideal gas law, and it maps exactly where it fails.
The usual story of why Beer's law fails has two villains. One is chemistry: molecules that associate, dissociate or interact with the solvent at high concentration. The other is the instrument: stray light, a detector that goes nonlinear, a bandwidth too wide for a sharp peak. Both are real, and both are in every analytical textbook.
The review by Thomas Mayerhöfer, Susanne Pahlow and Jürgen Popp in ChemPhysChem (2020) sets both of those aside on purpose. It asks a narrower question: with no chemical interactions and a perfect instrument, is the law still right? Their answer, built on Maxwell's equations and dispersion theory, is: only approximately, and in some common measurement setups, not even that.
The law treats light as something that simply fades as it passes through a sample. But light is a wave, and waves interfere. The same group's 2019 paper puts the core problem in one line of its abstract: even without chemical interactions or instrumental errors, "absorbance should be only approximately proportional to concentration."
The reason, in plain terms: what scales cleanly with concentration is a property of the molecules (the imaginary part of their polarizability). Absorbance is tied to that through the sample's refractive index, and the refractive index itself changes with concentration and changes sharply around every absorption band. At low concentration the refractive index stays close enough to constant that absorbance is very nearly linear. As concentration rises, it is not.
Thin films and layers. This is where the review is most striking. For a film on a reflective substrate (the "transflection" setup common in infrared imaging of tissue on metal-coated slides), the authors write that absorbance does not increase linearly with thickness and "can even decrease." Peaks can shift in either direction and satellite peaks can appear, none of which reflects any change in the chemistry. The cause is standing waves: light reflected inside the layer interferes with itself, so the electric-field intensity varies through the film, and absorption follows the field.
Switching to an index-matched substrate such as calcium fluoride helps less than it seems. The review reports deviations from the law of about 30 percent for layers about 1 micrometre thick, falling below 5 percent for layers thicker than about 4 micrometres.
ATR. Attenuated total reflection is the default sampling method in many infrared labs because it needs no sample prep. The review says the standard ATR absorbance is usable, to a good degree, only for weak absorptions: an index of absorption below about 0.1. The amide I band of proteins, the workhorse of biological infrared work, peaks at about 2.5 times that limit. ATR bands are also shifted toward the maxima of the refractive index and can change shape, and with a diamond or zinc selenide crystal at 45 degrees, a sample with a refractive index above about 1.7 is no longer being measured by ATR at all.
Concentration. For ordinary transmission through a thick cuvette, the setting where Beer's law works best, the review's rule of thumb is that concentrations should be less than about one part per thousand of the neat substance, "higher for weaker absorptions – lower for stronger ones." It adds two conditions that are easy to overlook: the sample must be ratioed against the pure solvent, not just air, and the solution's refractive index must not drift far from the solvent's. At high concentration the model predicts asymmetric bands whose peaks shift; local-field effects in a solvent shift them further (an effect the review traces back to Kundt in 1878).
Mixed and heterogeneous samples. When a sample is a patchwork of domains large enough to see under a microscope, the measured spectrum is an average of transmittances, not of absorbances, and the law cannot hold. Bands flatten, and the review notes that under these mixing rules even a genuine two-component system may no longer show an isosbestic point, the crossing point analysts often use as proof that exactly two species are present.
The review is not a counsel of despair. One result holds up better than the peak-height version of the law: the integrated area of an absorption band is proportional to concentration, a result the authors derive from the sum rules that follow from the Kramers-Kronig relations. The caveat is that local-field effects, which matter in condensed samples, can break even that. The authors also point out that a band's peak position is generally not the true transition energy of the molecule, because optical effects alone can move it.
And in the everyday case, a dilute solution in a centimetre cuvette measured against a solvent blank, the law is a good approximation. That is exactly the case it was built on. The trouble starts when the same equation is carried into thin films, micro-spectroscopy, ATR and concentrated samples without anyone noticing that the assumptions left the building.
The authors' conclusion is worth quoting because it is the whole report in a sentence. The ideal gas law, they write, has one advantage over Beer's law thanks to its name: "it is never mistaken as being correct in reality." Their suggestion is to rename the Bouguer-Beer-Lambert law the "ideal absorption law."
Three habits follow directly from the review, and none of them requires Maxwell's equations. First, keep quantitative absorbance work in the regime the law was built for: dilute, thick path, ratioed to the solvent. Second, when you calibrate, do not take a straight line on faith; the review itself notes a nonlinear regression can be used for quantification, and checking the residuals is how you find out whether you need one (see Report 157 on why a high R² will not tell you). Third, treat band shifts and intensity changes in thin-film, transflection or ATR spectra as possibly optical before calling them chemical.
My own published research is in microwave spectroscopy. Nothing in this report comes from that work, and no measurement was made for it; the evidence is entirely the two papers below. I care about it because it is the purest version of a pattern this beat keeps returning to: a tidy equation that is correct inside its assumptions and quietly wrong outside them. I made a related point about what Raman spectra cannot tell you in Report 013.
This is one group's review. The thresholds above (the 0.1 ATR limit, the 30 and 5 percent film figures, the one-per-thousand concentration rule) are the review's rules of thumb, several drawn from the authors' own earlier papers and simulations. I did not find an independent study testing all of them, and some depend on how strong the band is, which the authors say explicitly.
The authors have a methodological stake. The correction methods they recommend are largely their own, based on electromagnetic theory. That is normal in a review and they declare no conflict of interest, but it is worth knowing.
The 2019 companion paper is paywalled. I read its abstract only (via the publisher's deposit with Crossref and Europe PMC) and quote only the abstract.
Scope. The review is strongest on infrared spectroscopy. It says little quantitatively about routine UV-visible assays beyond the thick-cuvette case, and it deliberately excludes the instrumental and chemical causes of deviation, which still apply on top of everything here.
Beer's law is a limiting law. It is excellent for dilute solutions in thick cuvettes measured against a solvent blank, and progressively worse for thin films, reflective substrates, ATR, concentrated samples and heterogeneous ones, even with a perfect instrument and no chemistry going on. When a spectrum changes, the first question should be whether the sample changed or the optics did.
Scope note: this report summarizes a published review of the limits of the Bouguer-Beer-Lambert law. No measurement was performed and no instrument or product is evaluated. The practical recommendations in "What to do with this in a lab" are the author's reading of the review.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.