---
title: 'MIL-120(Al): Ultra-small Pore MOF & CO2 Adsorption'
url: https://www.emergentmind.com/topics/mil-120-al
type: topic
---

# MIL-120(Al): Ultra-small Pore MOF & CO2 Adsorption

MIL-120(Al) is a prototypical ultra-small pore metal-organic framework (MOF) reported as an attractive CO\(_2\) sorbent, and its behavior is strongly shaped by local framework dynamics associated with bridging hydroxyl, \(\mu_2\)-OH, groups rather than by the average crystallographic skeleton alone. In "Decoding local framework dynamics in the ultra-small pore MOF MIL-120(Al) CO2 sorbent with Machine Learned Potentials" [2508.20608], the material is analyzed by combining density-functional theory (DFT) with a purpose-trained machine-learned potential (MLP), establishing that subtle \(\mu_2\)-OH reorientation/flipping is a key determinant of pore accessibility, host–guest geometry, and adsorption energetics.

## 1. Structural framework and ultra-small pore character

MIL-120(Al) features edge-sharing AlO\(_6\) octahedra, connected via bridging \(\mu_2\)-OH groups, forming one-dimensional channels with ultra-small pores of approximately \(5.4 \times 4.7\) Å. Within this architecture, the \(\mu_2\)-OH groups are not merely structural linkages: their orientation modulates both pore dimension and chemistry at the local, angstrom/sub-angstrom scale [2508.20608].

A central point is that the orientation of the \(\mu_2\)-OH groups constitutes a "hidden" degree of freedom. Because X-ray diffraction cannot resolve hydrogen positions effectively, these orientations and their dynamics are invisible to standard crystallographic methods. Yet they directly influence hydrogen bonding between inorganic Al(OH)\(_4\)O\(_2\) chains and potential host–guest interactions via directional H-bonding inside ultra-small pores.

This places MIL-120(Al) in a class of sorbents for which an average structural model is insufficient. A plausible implication is that a crystallographically identical framework can host multiple locally distinct adsorption environments if hydrogen-bearing groups sample several orientations under operating conditions.

## 2. Configurational landscape of the \(\mu_2\)-OH groups

Six representative configurations were constructed—MIL-120(Al)-A, -B, -C, -D, -E, and -F—differing only in the spatial orientation of the four \(\mu_2\)-OH groups per unit cell. These models have identical average framework skeletons and simulated X-ray patterns, but their pore size distributions show minor but consequential differences, with main peak ranges of approximately 3.8 to 4.4 Å depending on the OH group orientation toward the channel [2508.20608].

The DFT optimization results identify MIL-120(Al)-A as the most stable configuration and MIL-120(Al)-F as the highest-energy variant, even though F had historically been used as the reference structure in the literature. The reported energy difference is
\[
\Delta E_{\text{A-F}} = 0.59~\text{eV per unit cell}.
\]

The structural basis for this ordering is also specified. MIL-120(Al)-A is stabilized by a cooperative, interlocking H-bond network between \(\mu_2\)-OH and adjacent chains, whereas the other variants, B-F, lack directional H-bonds and therefore lie at higher energies.

The significance of this result is not limited to energetic ranking. Since the six structures share the same average skeleton and simulated diffraction signature, the configurational state of the \(\mu_2\)-OH sublattice cannot be inferred from standard crystallographic observables alone. This suggests that experimentally observed disorder can coexist with a well-defined local energetic hierarchy.

## 3. Interconversion barriers and local framework dynamics

The configurational manifold is dynamically accessible. Using both DFT and MLP-based climbing-image nudged elastic band calculations, the reported interconversion energy barriers are
\[
0.07\text{–}0.19~\text{eV per unit cell}.
\]
These relatively low barriers suggest that all six states can be observed experimentally at room temperature, and the paper explicitly interprets them as evidence that \(\mu_2\)-OH groups can dynamically reorient under ambient conditions [2508.20608].

One specific example is given for the A \(\rightarrow\) B transition, with a barrier of \(0.073~\text{eV per unit cell}\) from DFT and \(0.099~\text{eV per unit cell}\) from the MLP. More broadly, the complete overlap of the DFT and MLP energy profiles is presented as evidence that the learned potential captures the transition-state landscape accurately.

The presence of CO\(_2\) modifies this landscape slightly. The guest lowers certain transition barriers by 3–8%, stabilizing specific transition states via direct host–guest interactions. This indicates that the framework dynamics are not independent of adsorption but are coupled to it. A plausible implication is that adsorption in MIL-120(Al) should be viewed as occurring in a fluctuating host whose local degrees of freedom are themselves perturbed by the guest.

## 4. Machine-learned potential and near-DFT-level fidelity

The computational strategy combines DFT with a purpose-trained MLP developed via DeePMD-kit and trained on an extensive DFT dataset including various \(\mu_2\)-OH orientations, structures with and without CO\(_2\), CI-NEB paths, and AIMD snapshots. The stated objective is systematic investigation of local dynamics at a level of detail that would be difficult to access by brute-force DFT alone [2508.20608].

The reported accuracy metrics are highly specific. The root-mean-square errors are \(0.217~\text{meV/atom}\) for the empty framework and \(0.268~\text{meV/atom}\) for the CO\(_2\)-loaded system, described as significantly lower than typical MLPs. In addition, the MLP is reported to reproduce the energy barriers and phonon spectra of the empty MOF, while also replicating DFT trends for phonons, elastic constants, and energy-volume curves.

For adsorption, the MLP-predicted CO\(_2\) adsorption energies and geometries deviate by at most \(2.3~\text{kJ·mol}^{-1}\) from DFT values. The paper further states that the MLP perfectly replicates DFT trends and accurately predicts NEB barriers and geometries.

The broader methodological importance lies in the level of local structural resolution retained by the potential. In this system, near-DFT-level fidelity is not only a benchmark of numerical quality; it is necessary because the relevant energetic differences arise from subtle hydrogen-bond topologies and orientation-dependent steric effects inside ultra-small pores.

## 5. Coupling between \(\mu_2\)-OH orientation and CO\(_2\) adsorption geometry

The orientation of the \(\mu_2\)-OH groups dictates the preferred alignment of CO\(_2\) in the pore. When \(\mu_2\)-OH groups point toward the pore, CO\(_2\) aligns perpendicular to the channel, maximizing H-bonding. The reported O(CO\(_2\))\(\cdots\)H(\(\mu_2\)-OH) distances are \(2.17\text{–}2.74\) Å. When the \(\mu_2\)-OH groups are more axial, as in MIL-120(Al)-C and -D, CO\(_2\) aligns parallel to the channel because of steric and hydrogen-bond considerations [2508.20608].

Both DFT and the MLP agree on this correlation between local hydroxyl orientation and adsorption geometry. The paper therefore treats adsorption site geometry as a consequence of local framework state rather than as a fixed property of a rigid pore.

This orientation dependence governs adsorption energetics. More confined, ordered \(\mu_2\)-OH group orientations, exemplified by MIL-120(Al)-F, yield stronger host–guest interactions but are described as less realistic and overstated compared to experiment. By contrast, the local ensemble associated with lower-energy or dynamically accessible configurations produces adsorption behavior more consistent with measured thermodynamics.

The mechanistic conclusion is precise: local reorientation of bridging hydroxyl groups is a key feature for gaining an accurate description of guest locations and energetics in ultra-small pore MOFs.

## 6. Adsorption thermodynamics, modeling implications, and interpretive consequences

The calculated adsorption energetics are configuration dependent, and this dependence extends to isosteric heat. The reported \(Q_{st}\) calculations show that MIL-120(Al)-A, -B, and -C agree closely with experiment, whereas MIL-120(Al)-F overestimates \(Q_{st}\) by approximately 27% [2508.20608]. Because F is also the highest-energy empty-framework variant, the paper explicitly questions the adequacy of using that historical reference structure as the dominant model under ambient or operational conditions.

The study attributes the functional behavior of MIL-120(Al) to framework flexibility at the level of the \(\mu_2\)-OH groups. Their rapid interconversion modulates pore dimension and chemical environment dynamically, producing a fluctuating host for CO\(_2\) molecules. The paper states that this leads to enhanced selectivity and packing for CO\(_2\) over, for example, N\(_2\) or CH\(_4\). Since the details provided do not include direct comparative adsorption metrics for those gases, that statement is best read as a qualitative functional implication rather than a quantified separation benchmark.

Two modeling consequences are emphasized. First, reliable CO\(_2\) adsorption modeling requires explicit inclusion of \(\mu_2\)-OH orientation flexibility. Second, generic force fields with rigid frameworks miss key aspects of host–guest interaction, especially in ultra-small or ultra-microporous MOFs such as MIL-120(Al). The dominant configuration in use is therefore likely not a single static structure but a lower-energy, dynamically disordered ensemble driven by \(\mu_2\)-OH H-bonding networks.

In materials-design terms, the paper argues that subtle local framework features, especially hydrogen-containing polar groups in crowded pores, are crucial for maximizing CO\(_2\) uptake, selectivity, and practical application. A plausible implication is that descriptor sets for sorbent screening should incorporate local orientational degrees of freedom whenever adsorption occurs in confined, highly directional environments.

Source: https://www.emergentmind.com/topics/mil-120-al