In this work, we introduce MOFTransformer, a multi-model Transformer encoder pre-trained with 1 million hypothetical MOFs. The multi-modal model uses an integrated atom-based graph and energy-grid embeddings to capture both the local and global features of the MOFs, respectively. By ne-tuning the pre-trained model with small datasets (from 5,000 to 20,000), our model outperforms all other machine learning models across various properties that include gas adsorption, diffusion, electronic properties, and even text mined data. Beyond its universal transfer learning capabilities, MOFTransformer generates chemical insight by analyzing feature importance from attention scores within the self-attention layers. As such, this model can serve as a bedrock platform for other MOF researchers that seek to develop new machine learning models for their work.

It is highly important to implement various semiconducting, such as n- or p-type, or conducting types of sensing behaviors to maximize the selectivity of gas sensors. To achieve this, researchers so far have utilized the n–p (or p–n) two-phase transition using doping techniques, where the addition of an extra transition phase provides the potential to greatly increase the sensing performance. Here, we report for the first time on an n–p-conductor three-phase transition of gas sensing behavior using Mo2CT x MXene, where the presence of organic intercalants and film thickness play a critical role. We found that 5-nm-thick Mo2CT x films with a tetramethylammonium hydroxide (TMAOH) intercalant displayed a p-type gas sensing response, while the films without the intercalant displayed a clear n-type response. Additionally, Mo2CT x films with thicknesses over 700 nm exhibited a conductor-type response, unlike the thinner films. It is expected that the three-phase transition was possible due to the unique and simultaneous presence of the intrinsic metallic conductivity and the high-density of surface functional groups of the MXene. We demonstrate that the gas response of Mo2CT x films containing tetramethylammonium (TMA) ions toward volatile organic compounds (VOCs), NH3, and NO2 is ∼30 times higher than that of deintercalated films, further showing the influence of intercalants on sensing performance. Also, DFT calculations show that the adsorption energy of NH3 and NO2 on Mo2CT x shifts from −0.973, −1.838 eV to −1.305, −2.750 eV, respectively, after TMA adsorption, demonstrating the influence of TMA in enhancing sensing performance.

Identifying optimal synthesis conditions for metal− organic frameworks (MOFs) is a major challenge that can serve as a bottleneck for new materials discovery and development. A trialand-error approach that relies on a chemist's intuition and knowledge has limitations in efficiency due to the large MOF synthesis space. To this end, 46,701 MOFs were data mined using our in-house developed code to extract their synthesis information from 28,565 MOF papers. The joint machine-learning/rule-based algorithm yields an average F1 score of 90.3% across different synthesis parameters (i.e., metal precursors, organic precursors, solvents, temperature, time, and composition). From this data set, a positive-unlabeled learning algorithm was developed to predict the synthesis of a given MOF material using synthesis conditions as inputs, and this algorithm successfully predicted successful synthesis in 83.1% of the synthesized data in the test set. Finally, our model correctly predicted three amorphous MOFs (with their representative experimental synthesis conditions) as having low synthesizability scores, while the counterpart crystalline MOFs showed high synthesizability scores. Our results show that big data extracted from the texts of MOF papers can be used to rationally predict synthesis conditions for these materials, which can accelerate the speed in which new MOFs are synthesized.

The rechargeable Li–CO2 battery has attracted considerable attention in recent years because of its carbon dioxide (CO2) utilization and because it represents a practical Li–air battery. As with other battery systems such as the Li-ion, Li–O2, and Li–S battery systems, understanding the reaction pathway is the first step to achieving high battery performance because the performance is strongly affected by reaction intermediates. Despite intensive efforts in this area, the effect of material parameters (e.g., the electrolyte, the cathode, and the catalyst) on the reaction pathway in Li–CO2 batteries is not yet fully understood. Here, we show for the first time that the discharge reaction pathway of a Li–CO2 battery composed of graphene nanoplatelets/beta phase of molybdenum carbide (GNPs/β-Mo2C) is strongly influenced by the dielectric constant of its electrolyte. Calculations using the continuum solvents model show that the energy of adsorption of oxalate (C2O4 2–) onto Mo2C under the low-dielectric electrolyte tetraethylene glycol dimethyl ether is lower than that under the high-dielectric electrolyte N,N-dimethylacetamide (DMA), indicating that the electrolyte plays a critical role in determining the reaction pathway. The experimental results show that under the high-dielectric DMA electrolyte, the formation of lithium carbonate (Li2CO3) as a discharge product is favorable because of the instability of the oxalate species, confirming that the dielectric properties of the electrolyte play an important role in the formation of the discharge product. The resulting Li–CO2 battery exhibits improved battery performance, including a reduced overpotential and a remarkable discharge capacity as high as 14,000 mA h g–1 because of its lower internal resistance. We believe that this work provides insights for the design of Li–CO2 batteries with enhanced performance for practical Li–air battery applications.

Covalent-organic frameworks (COFs) are regarded as promising candidates for many different energy/environmental applications, but these materials are more difficult to synthesize compared to other porous materials such as metal−organic frameworks (MOFs). Herein, we developed a computational screening algorithm that uses MOFs as substrates in order to theoretically allow heteroepitaxial growth of three-dimensional COFs (3D COFs). The algorithm details the interface of MOF@ COF at the atomic/molecular level in order to create 3D COFs using a bottom-up approach. Consequently, 19 pairs of MOF@ COF resulted from the algorithm are selected as candidates for heteroepitaxial growth of 3D COFs on the surface of MOFs.