Phantom Pores: How Nitrogen Gas Tricks Scientists Measuring Carbon Materials

Porous carbon materials underpin some of the most important technologies in the transition to clean energy, from battery anodes and supercapacitors to precious-metal-free catalysts. For decades, researchers have relied on gas sorption analysis to map the invisible architecture of these materials, firing nitrogen, carbon dioxide or argon molecules at carbon surfaces and interpreting how much gas is taken up at different pressures. The resulting pore size distributions guide the design of better materials, so scientists generally trust what the instruments tell them. A new study published in Advanced Science now reveals that one of the field’s standard practices may be systematically misleading them, showing that ultranarrow features in pore size distributions of non-graphitic carbons can be artefacts of specific gas adsorption rather than genuine physical pores.

The research team, led by Tim-Patrick Fellinger with contributions from Simon W. J. Dietzmann, Asad Mehmood, Jian Liang Low and colleagues, set out to understand a puzzling observation: commercial activated carbons and advanced nitrogen-doped carbons frequently show extremely sharp peaks in their calculated ultramicropore distributions, with full widths at half maximum far below what physical considerations should allow. For pores differing in size by even a single carbon layer, theory suggests a peak width of at least 0.125 nanometres, yet some measurements report peaks as narrow as 0.04 nanometres. Such precision would imply pores defined to a fraction of an atom’s diameter, which is physically implausible in the disordered, amorphous world of non-graphitic carbon.

To resolve the mystery, the researchers prepared a pair of elegantly matched model materials. Starting from a zinc-containing zeolitic imidazolate framework, they synthesised a zinc-nitrogen-carbon material containing tetrapyrrolic ZnN4 sites, atomically dispersed metal centres coordinated by four nitrogen atoms embedded in the carbon plane. Acid extraction of the zinc then produced a second material in which the metal ions were removed but the carbon framework remained essentially unchanged, leaving behind tetrapyrrolic H2N4 cavities with nitrogen-nitrogen distances of roughly 0.4 nanometres. Spectroscopic analysis using extended X-ray absorption fine structure confirmed the tetrapyrrolic coordination in both materials, while X-ray photoelectron spectroscopy tracked the transformation from ZnN4 to H2N4 sites. Powder X-ray diffraction confirmed that no change in graphitisation occurred, meaning only the zinc was removed and the carbon skeleton was preserved.

With these precisely defined materials in hand, the team measured gas sorption using three different probe gases: nitrogen at 77.4 kelvin, carbon dioxide at 273 kelvin, and argon at 87.3 kelvin. Both materials displayed type I isotherms characteristic of microporous solids, and zinc extraction increased the apparent specific surface area by about 23 percent. But then came a surprise. At the lowest accessible relative pressures, around ten to the minus seven, the zinc-containing material actually adsorbed more nitrogen and more carbon dioxide than the supposedly more porous metal-free version. This trend reversal was counterintuitive: if the removal of zinc created new ultramicropores, the leached material should have taken up more gas, not less.

The key clue came from argon. Unlike nitrogen and carbon dioxide, argon is essentially non-polar and possesses no quadrupole moment, the asymmetry in its charge distribution that allows it to interact with electrically polarised surfaces. Argon adsorption followed the expected trend, with the metal-free material consistently taking up more gas across the entire pressure range, and no reversal appeared at low pressures. Because the anomalous behaviour depended on the quadrupole moment of the gas, with carbon dioxide showing the strongest effect, followed by nitrogen and effectively none for argon, the researchers concluded that the origin was microscopic and chemical rather than macroscopic and structural. The specific interactions between polar gas molecules and the embedded nitrogen-metal sites were distorting the measurements.

The pore size distributions told the same story. Analysis of nitrogen and carbon dioxide isotherms produced razor-thin apparent ultramicropore peaks, with full widths at half maximum of roughly 0.09 and 0.04 nanometres respectively, at apparent pore diameters between 0.53 and 0.82 nanometres depending on the gas used. Argon, in contrast, revealed only broader supermicropores centred around 1.3 nanometres with a full width of about 0.6 nanometres, features the team assigns to genuine physical porosity. The apparent surface areas themselves also diverged depending on the gas: for the zinc-containing material, nitrogen analysis suggested 750 square metres per gram while argon gave 566, with nitrogen consistently overestimating the accessible surface.

To explain these observations at the molecular level, the team turned to electronic density functional theory calculations of adsorption energies for each gas on three sites: plain graphitic carbon, the tetrapyrrolic H2N4 cavity, and the ZnN4 centre. For all three gases, adsorption at ZnN4 was stronger than on graphitic carbon, and the deviation grew with increasing quadrupole moment of the adsorbate. The zinc atom carries a substantial partial charge of about plus 1.6 electrons due to its ionic bonding character, creating a strongly polarised site that attracts quadrupolar molecules like carbon dioxide and nitrogen through quadrupole-ion interactions. Binding energies at these sites fall squarely in the range normally associated with ultramicropore filling, so the porosity analysis software interprets them as pores smaller than 0.7 nanometres, even though no pore physically exists there. Carbon dioxide also binds preferentially to the H2N4 cavity through quadrupole-dipole interactions aligned with the nitrogen-hydrogen bonds, though more weakly than at ZnN4.

The most striking validation came from a new analytical approach. Because standard pore size distribution fitting cannot directly yield adsorption energies, the researchers applied a multi-Langmuir analysis to the carbon dioxide isotherms, deconvoluting the data into contributions from three classes of binding sites: the carbon support, real micropores, and the strongest specific sites. For the strongest site class, the fitted adsorption free energies differed by 0.036 electron volts between the zinc-containing and metal-free materials, a value that matches almost exactly the 0.035 electron volt difference predicted by the density functional theory calculations for carbon dioxide binding at ZnN4 versus H2N4 sites. The internal consistency between experiment and simulation is compelling evidence that the phantom pores are indeed the chemical fingerprints of atomically dispersed metal-nitrogen sites. Encouragingly, the extracted maximum capacities at these sites, between 0.08 and 0.13 millimoles per gram, are comparable to active site densities independently reported for iron-nitrogen-carbon catalysts, hinting that simple gas sorption could one day serve as a quantitative probe of catalytically active sites, a quantity that has remained notoriously difficult to measure.

The practical implications for the field are substantial. The authors recommend that narrow ultramicropore peaks arising from in-plane functionalities be flagged as artefacts, for example with an asterisk, and that corresponding pore volumes be reported as apparent rather than true micropore volumes. For atomically dispersed metal-nitrogen-carbon materials and related nitrogen-doped carbons, surface areas and total pore volumes are best determined from argon sorption, which avoids the overestimation caused by specific adsorption. Beyond correcting current practice, the findings open a genuine opportunity: gas sorption, one of the most widely accessible characterisation techniques in materials chemistry, could be transformed into a tool for detecting and quantifying specific surface sites and in-plane functionalities, provided new analysis kernels are developed. That would help researchers disentangle morphological properties such as porosity and surface area from chemical properties such as active site geometry and concentration, both crucial for optimising carbon-based materials in energy storage, conversion and catalysis. What looked like a flaw in the measurement may turn out to be one of its most useful features.

Subject of Research: Pore size distribution analysis artefacts in non-graphitic carbons and atomically dispersed M-N-C materials caused by specific gas adsorption

Article Title: A Pore or not a Pore? Understanding Pore Size Distributions of Non‐Graphitic Carbon and Atomically‐Dispersed M‐N‐C Materials

Article References: Dietzmann, S. W. J., Mehmood, A., Low, J. L., Wu, S.-H., Prinz, C., Buzanich, A. G., Radnik, J., Appel, P. A., Emmerling, F., & Fellinger, T.-P. (2026). A Pore or not a Pore? Understanding Pore Size Distributions of Non‐Graphitic Carbon and Atomically‐Dispersed M‐N‐C Materials. Advanced Science, 13(50), Article e76048. https://doi.org/10.1002/advs.76048

Image Credits: AI Generated

DOI: 10.1002/advs.76048

Keywords: pore size distribution, gas sorption, M-N-C materials, nitrogen-doped carbon, ultramicropores, ZnN4 sites, density functional theory, activated carbon, BET surface area, argon sorption, Langmuir analysis, electrocatalysis