ZIF Glasses: Structure, Properties & Applications

• The paper details how ZIF glasses form by thermal treatment, preserving tetrahedral Zn–imidazolate connectivity in a disordered, continuous random network akin to a‑SiO₂.
• It surveys data-driven modeling that integrates experimental scattering and deep-learning simulations to elucidate topological variations and transport mechanisms.
• Practical insights include tunable mechanical, energetic, and photonic properties achieved via linker functionalization, thermal annealing, and host–guest interactions.

Zeolitic imidazolate framework (ZIF) glasses represent an important class of amorphous metal–organic frameworks (MOFs), distinguished by their retained tetrahedral Zn–imidazolate connectivity, intrinsic nanoporous structure, and broad potential in gas separation, catalysis, energy absorption, and optoelectronics. Unlike conventional oxide glasses, ZIF glasses combine the chemical versatility of MOFs with the unique mechanical and dynamic features of the amorphous state, opening avenues for the systematic design of functional materials with tunable properties. This article surveys the key structural principles, formation mechanisms, transport phenomena, mechanical and photonic properties, topological modeling strategies, and the role of computational and experimental approaches in elucidating and exploiting the properties of ZIF glasses.

1. Structural Principles and Formation Mechanisms

The structural essence of ZIF glasses lies in the retention of local Zn–imidazolate tetrahedral coordination upon amorphization, with long‐range periodic order lost beyond ~25 Å (Bennett et al., 2010). The canonical route to ZIF glass formation is thermal treatment of crystalline ZIF-4:

$$\text{ZIF-4} \xrightarrow{\text{heat, }300 ^\circ\text{C}} \text{a-ZIF}$$

where the as-prepared solvated ZIF-4 is desolvated above 200 °C and then irreversibly transformed to the amorphous (a-ZIF) state at 300 °C. Further heating to 400 °C can induce recrystallization to ZIF-zni of identical Zn(im)₂ composition. PXRD, neutron and X-ray total scattering, and reverse Monte Carlo refinement demonstrate that a-ZIF adopts a continuous random network (CRN) topology analogous to a-SiO₂, with tetrahedral Zn–imidazolate units linked in a disordered network.

Deep-learning molecular dynamics simulations have elucidated the melting mechanism of ZIF-4, showing that bond-breaking/reformation events and undercoordination (ZnNₓ with x < 4) drive restructuring at both short- and medium-range scales during heating (Shi et al., 2023). Melting involves spatial heterogeneity, with low-density amorphous regions nucleating in zones where ligand detachment is prevalent, ultimately yielding a high-density amorphous phase upon further thermal treatment.

In mechanistic detail, the amorphization process includes transient tri-coordination events at Zn²⁺ centers and connectivity rearrangements via bond breaking and ligand exchange, thereby irreversibly modifying the network topology (Mendez et al., 2023). The transition itself can be anisotropic: amorphous domain growth may be directionally preferred due to local density and bond lability variations.

2. Topology, Network Modeling, and Data-Driven Methods

Early models assumed ZIF glasses to be topologically analogous to amorphous tetrahedral networks (e.g. a-SiO₂); recent work refines this perspective. Data-driven hybrid reverse Monte Carlo refinement, integrating machine-learning interatomic potentials (ACE-MLIP) with experimental X-ray/neutron scattering, yields atomistic models of a-ZIF that differ markedly from inorganic CRNs (Nicholas et al., 31 Mar 2025). A-ZIF exhibits a broad, unimodal distribution of ring sizes—3-membered to ~15-membered Zn–imidazolate rings—whereas crystalline ZIFs exhibit narrower, even-membered rings.

To codify topology in amorphous networks, the cumulative all-rings vertex symbol (CARVS) was introduced:

$\vec{r} = \frac{1}{\sum_i a_i}[a_1, a_2, ..., a_n]$

where $a_i$ is the average count of rings of size $i$ per node. A topological distance metric

$$d(\alpha, \beta) = \frac{1}{\sqrt{2}}|\vec{r}(\alpha) - \vec{r}(\beta)|$$

quantifies similarity, facilitating mapping across crystalline and amorphous framework families. This systematic notation provides a rigorous bridge between disorder–property relationships in amorphous MOFs and existing materials informatics paradigms for crystalline frameworks.

3. Transport, Diffusion, and Host–Guest Dynamics

Diffusion in ZIF glasses (and ZIF-8 crystals) is governed by discrete-cavity dynamics, with diffusion coefficients controlled by particle interactions within cages (Becker et al., 2013). Two key coefficients are:

$$D_t(\mu) = \frac{\lambda^2\langle k\rangle}{\langle n^2\rangle - \langle n \rangle^2}; \qquad D_s(\mu) = \frac{\lambda^2\langle k\rangle}{\langle n\rangle}$$

where $\lambda$ is center-to-center cage distance, $k$ is transition rate, and $n$ particle number per cavity. The ratio $D_t/D_s = \Gamma(\mu)$, the thermodynamic factor, reflects how self-diffusion can exceed transport diffusion if the interaction free energy $f''(n)$ is concave, implying clustering inside cages. This mechanism persists under correlations, as confirmed numerically and experimentally.

Confinement experiments with glass-forming liquids (e.g. glycerol in ZIF-8 and ZIF-11) reveal that the temperature-dependent cooperativity length of $\alpha$-relaxation is truncated by pore size (Uhl et al., 2018). For ZIF-11, the crossover from super-Arrhenius (VFT) to Arrhenius dynamics when $L_\text{corr}$ reaches the 1.46 nm pore diameter offers direct evidence for critical divergence of length scales in the glass transition.

Cryo-EM imaging of ZIF-8 has identified atomic surface defects, guest-induced lattice expansion ($\epsilon \sim 0.03$; $d = d_0(1+\epsilon)$), and gate-opening linker motions, underscoring the direct impact of host–guest interactions on both network flexibility and separation properties (Li et al., 2019).

4. Mechanical, Elastic, and Energy Absorption Properties

Elastic properties of ZIF frameworks and their glasses are characterized by low Young’s and shear modulus, high anisotropy (ZIF-62: $A^U \sim 5$), and the occurrence of negative Poisson’s ratio (auxetic) and negative linear compressibility (Xiong et al., 2017). These unusual responses link to the flexibility of organic linkers and local topology. Mechanical characterization of monolithic ZIF glasses reveals grain boundary sliding at low stresses, transition to bond breaking and local densification under higher loads, and fracture toughness $K_{Ic}$ in the range $0.074$–$0.145$ MPa·m$^{1/2}$, implying ductile deformation and resistance to cracking (Tricarico et al., 2021).

Dynamic water intrusion experiments demonstrate that ZIF-8 and related cage-type glasses exhibit pronounced rate-dependent energy absorption, with intrusion limited by the nanosecond-scale nucleation of critical water clusters ($N_\text{crit} \sim 4$) inside cages, governed by a modified nucleation free energy:

$\Delta G = -N\Delta \mu + A\gamma$

where $A$ is interfacial area and $\gamma$ surface tension (Sun et al., 2021). Strain rates exceeding the molecular filling rate $$\dot{\epsilon} = a\Delta\epsilon/(80 r \tau_\text{filling})$$ induce enhanced hysteresis, key to high-rate impact absorption. These insights guide the design of ZIF glasses for reusable mechanical energy absorbers.

5. Chemical Tunability, Hydrophilicity, and Glass-Forming Ability

Geometry, topology, and linker functionalization enable precise control of hydrophilicity/hydrophobicity in ZIF glasses (Ortiz et al., 2019). Pore shape and linker polarity modulate water–framework interaction energies, transitioning from type V isotherms for hydrophobic systems (ZIF-8, $A_{Hads} \sim 20$–$30$ kJ/mol) to type I for hydrophilic systems (ZIF-90, $A_{Hads} \sim 80$ kJ/mol). Amphiphilic ZIFs (e.g. ZIF-65/RHO) host both hydrophilic sites and hydrophobic domains, enabling multiple adsorption regimes.

Glass-forming ability (GFA) is quantitatively linked to total bond-order density (TBOD) and largest cavity diameter (LCD), as revealed by high-throughput data mining of MOF databases and molecular dynamics simulations (Shi et al., 2023). High TBOD and small LCD favor melt stability and suppress bond cleavage at high temperatures. TBOD can be modulated externally by applying pressure or electric field (DFT-calculated increase for $E_\text{field}\sim0.1$ eV/Å). These descriptors provide a predictive “structural gene” for screening and engineering ZIF glasses.

Thermodynamic simulations clarify that ligand saturation and Zn coordination energetics dictate building block stability and growth kinetics, with the fourth imidazolate ligand addition less exergonic than the first three, influencing defect generation and glass formation (Méndez et al., 7 Nov 2024).

6. Photonic Properties and Functional Applications

Recent work has shown that ZIF glasses exhibit broadband white light emission upon melt-quenching, with strong enhancement and red shift when annealed above $T_a \geq 1.07 T_g$ ($T_g$ = glass transition temperature) (Li et al., 13 Aug 2025). The physical mechanism is a relaxation of intermediate-range structure and strengthening “weak” (pyrrole-like) Zn–N bonds, increasing conjugation between aromatic linkers and Zn, thereby narrowing the band gap and enhancing ligand-to-metal charge transfer (LMCT).

Optimized annealing at $T_a \approx 1.13 T_g$ yields absolute internal photoluminescence quantum yield (PLQY) of $12.2\%$ (excitation at 365 nm), with the glass remaining durable under continuous LED operation (luminous efficacy 4.2 lm/W, $74.1\%$ retention at 180 min). This control of photonic output via thermal processing makes ZIF glasses compelling for low-power, energy-efficient lighting and optoelectronic components.

7. Modeling Methodologies and Computational Approaches

Modeling ZIF glasses requires simulation tools that capture both chemical specificity and topological disorder. Atomic cluster expansion machine-learning potentials support large-scale sampling consistent with scattering experiments, while reverse Monte Carlo and hybrid RMC techniques remain key for structure refinement against experimental data (Nicholas et al., 31 Mar 2025). DeepMD-driven molecular dynamics advances the simulation of melting and glass formation mechanisms at atomistic resolution (Shi et al., 2023).

Coarse-grained Martini force fields offer a computationally efficient approach for mesoscale modeling of ZIF-8, reproducing lattice parameters and elastic constants reasonably well, but face challenges in capturing key phenomena such as the “swing effect.” Precise Lennard–Jones parametrization is crucial for balancing host–guest and host–host interactions, which is particularly significant in modeling the glass transition and mechanical response (Alvares et al., 2023).

Conclusion

ZIF glasses bridge the chemical versatility of MOFs and the physical signatures of glasses, offering a unique platform for structure–property tuning. Advances in structural characterization, topological modeling, and molecular simulation provide a rigorous framework for quantifying disorder, predicting mechanical and transport properties, and engineering their functional response for applications in separation, impact energy absorption, and optoelectronics. The field continues to expand with ongoing work on photonic performance, disorder–property relationships, and informatics-driven material design.