Search

High-throughput molecular simulations and machine learning (ML) have been implemented to adequately screen a large number of metal–organic frameworks (MOFs) for applications involving adsorption. Grand canonical Monte Carlo (GCMC) simulations have proven effective in calculating the adsorption capacity at given pressures and temperatures, but they can require expensive computational resources. While they can be resource-efficient, ML models can require large datasets, creating a need for algorithms that can efficiently characterize adsorption; active learning (AL) can play a very important role in this regard. In this work, we make use of Gaussian process regression (GPR) to model pure component adsorption of nitrogen at 77 K from 10–5to 1 bar, methane at 298 K from 10 –5to 100 bar, carbon dioxide at 298 K from 10–5to 100 bar, and hydrogen at 77 K from 10–5to 100 bar on PCN-61, MgMOF-74, DUT-32, DUT-49, MOF-177, NU-800, UiO-66, ZIF-8, IRMOF-1, IRMOF-10, and IRMOF-16. The GPR model requires an initial training of the model with an initial dataset, the prior one, and, in this study of evaluating AL, we make use of three different prior selection schemes. Each prior scheme is updated with a sampling point resulting from the GP model uncertainties. This protocol continues until a maximum GPR relative error of 2% is attained. We make a recommendation on the best prior selection scheme for the total 44 adsorbate–adsorbent pairs primarily making use of the mean absolute error and the total amount of points required for convergence of the model. To further evaluate the AL framework, we apply the BET consistency criteria on the simulated and GP nitrogen isotherms and compare the resulting surface areas.

We present the results of molecular dynamics simulations of the ionic liquid (IL) 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide [C2C1im][NTf2] in the presence of external electric fields (EEFs) of varying strengths to understand the effects of EEFs on the glass transition temperature Tg. We compute Tgwith an automated and objective method and observe a depression in Tgwhen cooling the IL within an EEF above a critical strength. The effect is reversible, and glasses prepared with EEFs recover their original zero-field Tgwhen heated. By examining the dynamics and structure of the liquid phase, we find that the EEF lowers the activation energy for diffusion, reducing the energetic barrier for movement and consequently Tg. We show that the effect can be leveraged to drive an electrified nonvapor compression refrigeration cycle.

The fabrication of polymer-MOF composite gels holds great potential to provide emergent properties for drug delivery, environmental remediation, and catalysis. To leverage the full potential of these composites, we investigated how the presence and chemistry of polymers impact MOF formation within the composites and, in turn, how MOFs impact polymer gelation. We show that polymers with a high density of strongly metal-binding carboxylic acids inhibit MOF formation; however, reducing the density of carboxylic acids or substituting them with weaker metal-binding hydroxyl groups permits both MOF formation and gelation within composites. Preparing composites with poly(ethylene glycol) (PEG), which does not bind MOF zirconium (Zr)-oxo clusters, and observing gelation suggests that MOFs can entrap polymer chains to create cross-links in addition to cross-linking them through polymer-Zr-oxo interactions. Both simulations and experiments show composite hydrogels formed with poly(vinyl alcohol) (PVA) to be more stable than those made with PEG, which can reptate through MOF pores upon heating. We demonstrate the generalizability of this composite formation process across different Zr-based MOFs (UiO-66, NU-901, UiO-67, and MOF-525) and by spin-coating gels into conformable films. PVA-UiO-66 composite hydrogels demonstrated high sorption and sustained release of methylene blue relative to the polymer alone (3× loading, 28× slower release), and PVA-MOF-525 composite hydrogels capably sorb the therapeutic peptide Angiotensin 1–7. By understanding the influence of polymer-MOF interactions on the structure and properties of composite gels, this work informs and expands the design space of this emerging class of materials.

Ionic liquids are the basis of a potential alternative to conventional processes based on aqueous amines currently used for carbon dioxide capture. Mixtures of ionic liquids offer an additional degree of tailoring over the intrinsic tunable properties of “single” ionic liquids, while totally maintaining the ionic liquid character. In this work, two mixtures of mutually miscible ionic liquids have been investigated, namely, 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide + 1-ethyl-3-methylimidazolium ethylsulfate and 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide + 1-butyl-3-ethylimidazolium ethylsulfate. Their ability to absorb carbon dioxide in the pressure range up to ca. 16 bar, at 298.2 K, was determined at different composition ratios of the mixtures; their thermal phase-transition temperatures and physical properties (density, viscosity, and surface tension) were determined over the entire compositional range, at atmospheric pressure and the same temperature. The absorption data were adequately correlated by the NRTL (Non-Random Two-Liquid) model, whereas different equations were successful in satisfactorily correlating the physical properties of the mixtures and in predicting them from knowledge of the properties of the pure ionic liquid components.

Metal–organic frameworks (MOFs) are a class of materials promising for gas adsorption due to their highly tunable nanoporous structures and host–guest interactions. While machine learning (ML) has been leveraged to aid the design or screen of MOFs for different purposes, the needs of big data are not always met, limiting the applicability of ML models trained against small data sets. In this work, we introduce an inductive transfer learning technique to improve the accuracy and applicability of ML models trained with a small amount of MOF adsorption data. This technique leverages potentially shareable knowledge from a source task to improve the models on the target tasks. As demonstrations, a deep neural networks (DNNs) trained on H2adsorption data with 13 506 MOF structures at 100 bar and 243 K is used as the source task. When transferring knowledge from the source task to H2adsorption at 100 bar and 130 K (one target task), the average predictive accuracy on target tasks was improved from 0.960 (direct training) to 0.991 (transfer learning), and transfer learning works in 89.3% of the cases. We also tested transfer learning across different gas species (i.e., from H2to CH4), with an average predictive accuracy of CH4adsorption being improved from 0.935 (direct training) to 0.980 (transfer learning), and transfer learning works in 82.0% of the cases. More importantly, transfer learning is shown to effectively improve the models on the target tasks with low accuracy from direct training. However, when transferring the knowledge from the source task to Xe/Kr adsorption, the transfer learning does not improve the predictive accuracy significantly and even makes it worse in ∼50.0% of the cases, which is attributed to the lack of common descriptors that is key to the underlying knowledge.

There is considerable interest in understanding ion transport in charged polymeric materials for applications in energy storage. In this work, all-atom molecular dynamics simulations and experimental measurements are combined to study ion mobility in poly(2-vinylpyridine) (P2VP) functionalized with N-methylpyridinium iodide (NMP+I–)—a system of particular interest due to its amenability in applications involving directed self-assembly. Experimental measurements demonstrate that increasing water content tends to decrease the glass transition temperature and enhance ion conductivity. While the simulations demonstrate similar trends, further analysis reveals that increasing water content also promotes ion–water interactions, which leads to overall shorter ion–polymer association times and thus enhanced ion mobility. Meanwhile, both the experimental measurements and the simulations demonstrate that decreasing the degree of NMP functionalization also enhances polymer-chain relaxation; however, the faster polymer dynamics negligibly affect I–mobility, at least in dry systems where the overall rates of ion transport are very low. Overall, these results highlight the importance of water in facilitating I–diffusion in these systems and provide insights into the impact of water content and NMP functionalization on polymer relaxation behavior. These results should be useful for identifying more effective polymer architectures and processing criteria in future charged polymeric materials.

Two types of electron acceptors were synthesized by coupling two kinds of electron-rich cores with four equivalent perylene diimides (PDIs) at the α-position. With fully aromatic cores, TPB and TPSe have π-orbitals spread continuously over the whole aromatic conjugated backbone, unlike TPC and TPSi, which contain isolated PDI units due to the use of a tetrahedron carbon or silicon linker. Density functional theory calculations of the projected density of states showed that the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) for TPB are localized in separate regions of space. Further, the LUMO of TPB shows a greater contribution from the orbitals belonging to the connective core of the molecules than that of TPC. Overall, the properties of the HOMO and LUMO point at increased intra-molecular delocalization of negative charge carriers for TPB and TPSe than for TPC and TPSi and hence at a more facile intra-molecular charge transfer for the former. The film absorption and emission spectra showed evidences for the inter-molecular electron delocalization in TPB and TPSe, which is consistent with the network structure revealed by X-ray diffraction studies on single crystals of TPB. These features benefit the formation of charge transfer states and/or facilitate charge transport. Thus, higher electron mobility and higher charge dissociation probabilities under Jsccondition were observed in blend films of TPB:PTB7-Th and TPSe:PTB7-Th than those in TPC:PTB7-Th and TPSi:PTB7-Th blend films. As a result, the Jscand fill factor values of 15.02 mA/cm2, 0.58 and 14.36 mA/cm2, 0.55 for TPB- and TPSe-based solar cell are observed, whereas those for TPC and TPSi are 11.55 mA/cm2, 0.47 and 10.35 mA/cm2, 0.42, respectively.

Sigma profiles are quantum-chemistry-derived molecular descriptors that encode the polarity of molecules. They have shown great performance when used as a feature in machine learning applications. To accelerate the development of these models and the construction of large sigma profile databases, this work proposes a graph convolutional network (GCN) architecture to predict sigma profiles from molecule structures. To do so, the usage of molecular mechanics (force field atom types) is explored as a computationally inexpensive node-level featurization technique to encode the local and global chemical environments of atoms in molecules. The GCN models developed in this work accurately predict the sigma profiles of assorted organic and inorganic compounds. The best GCN model here reported, obtained using Merck molecular force field (MMFF) atom types, displayed training and testing set coefficients of determination of 0.98 and 0.96, respectively, which are superior to previous methodologies reported in the literature. This performance boost is shown to be due to both the usage of a convolutional architecture and node-level features based on force field atom types. Finally, to demonstrate their practical applicability, we used GCN-predicted sigma profiles as the input to machine learning models previously developed in the literature that predict boiling temperatures and aqueous solubilities. Using the predicted sigma profiles as input, these models were able to compute both physicochemical properties using significantly less computational resources and displayed only a slight decrease in performance when compared with sigma profiles obtained from quantum chemistry methods.

Identifying chemical compounds is essential in several areas of science and engineering. Laser-based techniques are promising for autonomous compound detection because the optical response of materials encodes enough electronic and vibrational information for remote chemical identification. This has been exploited using the fingerprint region of infrared absorption spectra, which involves a dense set of absorption peaks that are unique to individual molecules, thus facilitating chemical identification. However, optical identification using visible light has not been realized. Using decades of experimental refractive index data in the scientific literature of pure organic compounds and polymers over a broad range of frequencies from the ultraviolet to the far-infrared, we develop a machine learning classifier that can accurately identify organic species based on a single-wavelength dispersive measurement in the visible spectral region, away from absorption resonances. The optical classifier proposed here could be applied to autonomous material identification protocols and applications.

Spontaneous parametric down conversion (SPDC) is a quantum second-order nonlinear optical process where the photons generated are frequently used in quantum information processing. Materials with large second-order nonlinearities (χ(2)) can be used as entangled photon sources with a high brightness. The source brightness scales as the square of the effective nonlinearity (deff), which is an intrinsic property of the material. Understanding material factors that can significantly alter this intrinsic property is useful in developing new materials that are SPDC efficient. In our work we focus on understanding factors affecting the entangled photon pair properties, such as the arrangements of ligands within the Zn(3-ptz)2metal–organic framework (MOF) crystal and temperature. We find that the arrangement and alignment of the pyridine rings in the crystal structure significantly affect the deffand birefringence (Δn). Smaller pyridine ring alignments relative to the optic c-axis increase the Δn, which in turn leads to larger photon pair correlation times (τc) in coincidence measurements. Our work has significant implications in understanding the effect of ligand arrangement on deffand τcfor any MOF crystal structure, providing a tool to rationalize the optimization of MOF crystals for the development of efficient nonlinear optical devices.

Machine learning (ML) accelerates the exploration of material properties and their links to the structure of the underlying molecules. In previous work [Shi et al. ACS Applied Materials & Interfaces2022, 14, 37161−37169.], ML models were applied to predict the adhesive free energy of polymer–surface interactions with high accuracy from the knowledge of the sequence data, demonstrating successes in inverse-design of polymer sequence for known surface compositions. While the method was shown to be successful in designing polymers for a known surface, extensive data sets were needed for each specific surface in order to train the surrogate models. Ideally, one should be able to infer information about similar surfaces without having to regenerate a full complement of adhesion data for each new case. In the current work, we demonstrate a transfer learning (TL) technique using a deep neural network to improve the accuracy of ML models trained on small data sets by pretraining on a larger database from a related system and fine-tuning the weights of all layers with a small amount of additional data. The shared knowledge from the pretrained model facilitates the prediction accuracy significantly on small data sets. We also explore the limits of database size on accuracy and the optimal tuning of network architecture and parameters for our learning tasks. While applied to a relatively simple coarse-grained (CG) polymer model, the general lessons of this study apply to detailed modeling studies and the broader problems of inverse materials design.

The hydrogen storage capabilities of 18 383 porous crystalline structures possessing various degrees of Mg functionalization and diverse physical properties were assessed through combined grand canonical Monte Carlo (GCMC) and quantum mechanical approaches. GCMC simulations were performed for pressures of 2 and 100 bar at a temperature of 243 K. Absolute uptake at 100 bar and deliverable capacity between 100 and 2 bar were calculated. Maximum absolute and deliverable gravimetric capacities were 9.35 and 9.12 wt %, respectively. Volumetrically, absolute and deliverable capacities were 51 and 30 g/L, respectively. The results reveal relationships between hydrogen uptake and the physical properties of the materials. We show that the introduction of an optimum amount of magnesium alkoxide to increase the isosteric heat of adsorption is a promising strategy to improve hydrogen uptake and delivery near ambient temperature.

In this work, we perform high-throughput computational screening of MOFs based on their selectivity of gas mixtures formed by six gas molecules (H2O, CO2, N2, N2O, CH4, and O2) at dilute conditions using Henry’s constants. Our results show that pore size is the most influential geometric factor for high selectivity and therefore should be highly considered in gas sensor design. Top-performing MOFs were observed to have high variability for surface area and void fraction values, and thus, those properties were deemed less important. The metals present in nodes were found to be important for adsorption of certain gases and should be taken into account when designing a sensor for a specific use. A ranking system was applied to find the best MOFs for sensing multiple gases and then the top 10 MOF’s properties were examined. We concluded that small pore size and a smaller difference between largest and smallest pores were ideal for a highly selective gas sensor. Finally, we recommend some MOF candidates for selectively adsorbing select gases at dilute conditions.