MSUCOF-4-FeCp: Ferrocene COF for H2 Storage

• The paper demonstrates that MSUCOF-4-FeCp, a ferrocene-functionalized COF, achieves a simulated hydrogen uptake of 18.0 wt% at 700 bar, surpassing DOE targets.
• The study outlines a novel structural design where IRCOF-102 is modified with ferrocene moieties to form cooperative binding pockets that enhance hydrogen adsorption.
• The research employs a multiscale computational approach—combining fragment DFT, periodic DFT, and GCMC simulations—to validate both performance and cost-effective iron incorporation over precious metals.

MSUCOF-4-FeCp is a ferrocene-functionalized covalent-organic framework proposed for room-temperature hydrogen storage within the Multi-binding Sites United in Covalent-Organic Framework (MSUCOF) design paradigm. In the reported first-principles multiscale computational study, it is constructed by incorporating ferrocene ($\mathrm{FeCp}_2$) moieties into IRCOF-102 and is predicted to exceed the U.S. Department of Energy ultimate light-duty vehicle hydrogen-storage targets, with total uptake values of $18.0$ wt % and $72.6\ \mathrm{g\ H_2\ L^{-1}}$ at $298\ \mathrm{K}$ and $700\ \mathrm{bar}$, and deliverable capacities of $12.2$ wt % and $52.2\ \mathrm{g\ L^{-1}}$ between $700$ and $5\ \mathrm{bar}$ (Djokic et al., 16 Feb 2026).

1. Position within the MSUCOF literature

MSUCOF-4-FeCp occupies a distinct place in the MSUCOF series because it is not part of the original 2023 MSUCOF study. That earlier work investigated only three covalent-organic framework families—MSUCOF-1, MSUCOF-2, and MSUCOF-3—and seven transition-metal chelation variants, namely Co, Cu, Fe, Mn, Ni, Pd, and Pt. It did not report any material labeled “MSUCOF-4,” and it did not include any FeCp or ferrocenyl derivative (Djokic et al., 2023).

This distinction matters because structural, synthetic, force-field, and adsorption data for MSUCOF-4-FeCp do not exist in the 2023 paper. No FeCp coordination environment, bond lengths, bond angles, fitted isotherm parameters, or uptake numbers for such a material are given there. The 2026 work therefore represents a separate computational development that follows the MSUCOF approach rather than a continuation of the exact materials set previously tabulated (Djokic et al., 16 Feb 2026).

A common misconception is to treat MSUCOF-4-FeCp as though it were one of the transition-metalated frameworks in the original MSUCOF manuscript. That is incorrect: the earlier study’s GCMC protocol and QM-based force-field philosophy are relevant as methodological antecedents, but the actual ferrocene-functionalized framework is introduced only in the later study (Djokic et al., 2023).

2. Structural design and framework chemistry

The parent framework is IRCOF-102, described as a three-dimensional boroxine-linked COF isoreticular to COF-102 and built from tetrahedral tetrakis(4′-borono-[1,1′-biphenyl]-4-yl)methane nodes connected via boroxine rings. Its topology crystallizes in space group $I\bar{4}3d$ (No. 220) with cubic cell parameters $18.0$0 and $18.0$1. The framework contains ca. $18.0$2-diameter pores arranged in a three-fold-rotational coordination environment around each boroxine linkage, yielding a surface area of $18.0$3, pore volume of $18.0$4, void fraction of $18.0$5, and framework density of $18.0$6 (Djokic et al., 16 Feb 2026).

To generate the MSUCOF-4 scaffold, the biphenyl linkers are replaced by indene-derived tetrakis(4-(4-boronoinden-7-yl)phenyl)methane, designated Linker 2. This linker embeds a cyclopentadienyl ring at each node. Condensation of Linker 2 produces MSUCOF-4, an isoreticular variant of IRCOF-102 with one Cp ring per linker but without Fe. Post-synthetic metallation with $18.0$7 then yields MSUCOF-4-FeCp, in which each Cp ring within a tritopic pore region is deprotonated and coordinates one $18.0$8 center together with a second $18.0$9 anion from NaCp (Djokic et al., 16 Feb 2026).

The resulting ferrocene moieties are positioned at junctions of three struts, forming cooperative binding pockets. Upon FeCp incorporation, the cell expands slightly to $72.6\ \mathrm{g\ H_2\ L^{-1}}$0 while retaining the same $72.6\ \mathrm{g\ H_2\ L^{-1}}$1 space group. The study also reports for MSUCOF-4-FeCp a porosity of $72.6\ \mathrm{g\ H_2\ L^{-1}}$2 and $72.6\ \mathrm{g\ H_2\ L^{-1}}$3. This suggests a design tradeoff in which some pore volume is sacrificed while introducing a denser distribution of targeted adsorption sites (Djokic et al., 16 Feb 2026).

3. Electronic-structure and simulation methodology

The computational treatment combines fragment DFT, periodic DFT, force-field fitting, and grand-canonical Monte Carlo. For fragment DFT, eclipsed and staggered ferrocene fragments plus single $72.6\ \mathrm{g\ H_2\ L^{-1}}$4 molecules were modeled in $72.6\ \mathrm{g\ H_2\ L^{-1}}$5 supercells using the M06 functional with the pob-TZVP-rev2 Gaussian basis set in CRYSTAL23. Geometry optimizations used tight SCF convergence of $72.6\ \mathrm{g\ H_2\ L^{-1}}$6, with forces converged to $72.6\ \mathrm{g\ H_2\ L^{-1}}$7; spin polarization and an extra-large integration grid were employed to resolve non-covalent energetics accurately (Djokic et al., 16 Feb 2026).

For the periodic solid, the full MSUCOF-4-FeCp and IRCOF-102 unit cells were optimized using hybrid HSE06 + D3 dispersion corrections in CRYSTAL23 with the same basis-set quality and convergence criteria. These calculations confirmed the eclipsed Cp conformation in the solid state, the cell expansion associated with metallation, and a band-gap narrowing from $72.6\ \mathrm{g\ H_2\ L^{-1}}$8 in IRCOF-102 to $72.6\ \mathrm{g\ H_2\ L^{-1}}$9 in MSUCOF-4-FeCp (Djokic et al., 16 Feb 2026).

Hydrogen–framework interactions were represented with a Morse potential,

$298\ \mathrm{K}$0

with parameters for $298\ \mathrm{K}$1–Fe, $298\ \mathrm{K}$2–C, and $298\ \mathrm{K}$3–H in ferrocene fitted in GULP to fragment DFT energy curves. The reported mean absolute errors are $298\ \mathrm{K}$4 (Djokic et al., 16 Feb 2026).

GCMC simulations were carried out in Materials Studio at $298\ \mathrm{K}$5 and $298\ \mathrm{K}$6–$298\ \mathrm{K}$7 using the Metropolis algorithm with move ratios translation : rotation : insertion : deletion : regrowth $298\ \mathrm{K}$8. Each pressure point used $298\ \mathrm{K}$9 equilibration and $700\ \mathrm{bar}$0 production steps. Bulk $700\ \mathrm{bar}$1 fugacities were obtained via the van der Waals equation,

$700\ \mathrm{bar}$2

with $700\ \mathrm{bar}$3 and $700\ \mathrm{bar}$4, giving $700\ \mathrm{bar}$5 deviation from Peng–Robinson at $700\ \mathrm{bar}$6 (Djokic et al., 16 Feb 2026).

4. Hydrogen binding and adsorption-site hierarchy

The zero-load binding energy is defined in the study as

$700\ \mathrm{bar}$7

including electronic plus zero-point and vibrational enthalpy corrections. For MSUCOF-4-FeCp, the reported primary $700\ \mathrm{bar}$8 binding energies lie in the $700\ \mathrm{bar}$9–$12.2$0 range, described as an ideal physisorption window for balancing high-pressure uptake with low-pressure release (Djokic et al., 16 Feb 2026).

The pressure-dependent isosteric heat $12.2$1 remains entirely within $12.2$2–$12.2$3. The study interprets this range as strong enough to enhance adsorption relative to weakly interacting porous frameworks, while still permitting facile desorption under practical delivery conditions. A plausible implication is that the ferrocene installation changes the adsorption problem from one dominated by very high porosity alone to one organized around a controlled spectrum of moderate adsorption energies (Djokic et al., 16 Feb 2026).

Interaction-energy deconvolution further resolves two main adsorption populations at high pressure: $12.2$4, assigned to the Fe center, and $12.2$5, assigned to weaker Cp-ring sites. At $12.2$6, $12.2$7 of $12.2$8 is reported to occupy the Fe site. This site hierarchy is consistent with the design premise that cooperative pockets centered on ferrocene can create a dominant primary adsorption motif supplemented by weaker secondary filling sites (Djokic et al., 16 Feb 2026).

5. Storage performance and DOE benchmarks

The reported adsorption metrics place MSUCOF-4-FeCp above both the DOE 2025 targets of $12.2$9 wt %, $52.2\ \mathrm{g\ L^{-1}}$0 and the DOE ultimate targets of $52.2\ \mathrm{g\ L^{-1}}$1 wt %, $52.2\ \mathrm{g\ L^{-1}}$2 for automotive hydrogen storage (Djokic et al., 16 Feb 2026).

Quantity	Value	Context
Total gravimetric uptake	$52.2\ \mathrm{g\ L^{-1}}$3 wt %	$52.2\ \mathrm{g\ L^{-1}}$4
Total volumetric uptake	$52.2\ \mathrm{g\ L^{-1}}$5	$52.2\ \mathrm{g\ L^{-1}}$6
Working capacity, gravimetric	$52.2\ \mathrm{g\ L^{-1}}$7 wt %	$52.2\ \mathrm{g\ L^{-1}}$8
Working capacity, volumetric	$52.2\ \mathrm{g\ L^{-1}}$9	$700$0
IRCOF-102 total gravimetric uptake	$700$1 wt %	identical conditions
IRCOF-102 total volumetric uptake	$700$2	identical conditions

Deliverable, or working, capacity is defined between a charging pressure of $700$3 and a delivery pressure of $700$4. The net-uptake curves peak at $700$5, and the study reports broad positive net uptake across $700$6–$700$7. Excess isotherms are stated to show analogous trends when subtracting only pore-volume gas (Djokic et al., 16 Feb 2026).

The same study compares MSUCOF-4-FeCp with precious-metal-functionalized MSUCOF-1-PtCl$700$8, for which $700$9 and $5\ \mathrm{bar}$0 wt % are quoted. On that basis, the ferrocene-functionalized framework is presented as outperforming prior MSUCOF variants while using Earth-abundant iron rather than platinum (Djokic et al., 16 Feb 2026).

6. Economic interpretation, limitations, and proposed validation

A central practical claim of the 2026 work is that ferrocene functionalization provides a cost-effective alternative to precious-metal incorporation in COFs. The paper gives approximate raw-material figures of $5\ \mathrm{bar}$1 for Pt, and states that iron’s natural abundance exceeds $5\ \mathrm{bar}$2 ore. On this basis, it describes a $5\ \mathrm{bar}$3-fold economic advantage over platinum (Djokic et al., 16 Feb 2026).

At the same time, the status of MSUCOF-4-FeCp remains predictive rather than experimentally established in the provided record. The study recommends experimental synthesis of MSUCOF-4, followed by post-synthetic ferrocene installation and gas-adsorption measurements spanning $5\ \mathrm{bar}$4–$5\ \mathrm{bar}$5 at $5\ \mathrm{bar}$6. It further recommends XPS and Mössbauer spectroscopy to confirm FeCp incorporation, and in situ IR or Raman spectroscopy to probe $5\ \mathrm{bar}$7 binding (Djokic et al., 16 Feb 2026).

The proposed future computational directions include molecular dynamics, quantum corrections to $5\ \mathrm{bar}$8 binding, and extension to alternative metallocenes such as cobaltocene and ruthenocene. This suggests that MSUCOF-4-FeCp functions not only as a specific candidate material but also as a design template: high-porosity COF architectures can be reprogrammed by installing organometallic motifs that create cooperative adsorption pockets with binding enthalpies in the targeted $5\ \mathrm{bar}$9–$I\bar{4}3d$0 range (Djokic et al., 16 Feb 2026).