Modeling of CO2 adsorption capacity by porous metal organic frameworks using advanced decision tree-based models

Predictive modeling of CO2 solubility in piperazine aqueous solutions using boosting algorithms for carbon capture goals

Carbon dioxide (CO2) is the main greenhouse gas that drives global warming, climate change, and other environmental issues. CO2 absorption using amine solvents stands out as one of the most well-known industrial technologies of CO2 capture. However, accurate prediction of CO2 absorption in aqueous amine solutions under different operating conditions is crucial for designing an efficient amine scrubbing system in power plants. In this work, CO2 solubility in aqueous piperazine (PZ) solutions was modeled using 517 experimental data points covering a temperature range of 298 to 373 K, PZ concentration of 0.1 to 6.2 mol/L (M), and CO2 partial pressure of 0.03 to 7399 kPa. To this end, four robust machine learning algorithms, including gradient boosting with categorical features support (CatBoost), light gradient boosting machine (LightGBM), extreme gradient boosting (XGBoost), and adaptive boosting decision trees (AdaBoost-DT) were utilized. Among the developed models, the CatBoost model presented the highest accuracy with an overall determination coefficient (R2) of 0.9953 and an average absolute relative error of 2.36%. Sensitivity analysis revealed that CO2 partial pressure had the greatest influence on CO2 absorption in aqueous PZ solutions, followed by PZ concentration and temperature. Moreover, CO2 partial pressure positively influenced CO2 absorption in aqueous PZ solutions, while PZ concentration and temperature exhibited negative effects. Finally, the leverage technique indicated that both the experimental data bank used for modeling and the model's estimates were statistically acceptable and valid showing only 8 points (∼1.5% of total data) as possible suspected data.

Computational and Machine Learning Methods for CO2 Capture Using Metal-Organic Frameworks

Machine learning (ML) using data sets of atomic and molecular force fields (FFs) has made significant progress and provided benefits in the fields of chemistry and material science. This work examines the interactions between chemistry and materials computational science at the atomic and molecular scales for metal-organic framework (MOF) adsorbent development toward carbon dioxide (CO2) capture. Herein, a connection will be drawn between atomic forces predicted by ML algorithms and the structures of MOFs for CO2 adsorption. Our study also takes into account the successes of atomic computational screening in the field of materials science, especially quantum ML, and its relationship to ML algorithms that clarify advancements in the area of CO2 adsorption by MOFs. Additionally, we reviewed the processes for supplying data to ML algorithms for algorithm training, including text mining from scientific articles, and MOF's formula processing linked to the chemical properties of MOFs. To create ML algorithms for future research, we recommend that the digitization of scientific records can help efficiently synthesize advanced MOFs. Finally, a future vision for developing pioneer MOF synthesis routes for CO2 capture is presented in this review article.

Deep learning and big data mining for Metal-Organic frameworks with high performance for simultaneous desulfurization and carbon capture

Carbon capture and desulfurization of flue gases are crucial for the achievement of carbon neutrality and sustainable development. In this work, the "one-step" adsorption technology with high-performance metal-organic frameworks (MOFs) was proposed to simultaneously capture the SO2 and CO2. Four machine learning algorithms were used to predict the performance indicators (NCO2+SO2, SCO2+SO2/N2, and TSN) of MOFs, with Multi-Layer Perceptron Regression (MLPR) showing better performance (R2 = 0.93). To address sparse data of MOF chemical descriptors, we introduced the Deep Factorization Machines (DeepFM) model, outperforming MLPR with a higher R2 of 0.95. Then, sensitivity analysis was employed to find that the adsorption heat and porosity were the key factors for SO2 and CO2 capture performance of MOF, while the influence of open alkali metal sites also stood out. Furthermore, we established a kinetic model to batch simulate the breakthrough curves of TOP 1000 MOFs to investigate their dynamic adsorption separation performance for SO2/CO2/N2. The TOP 20 MOFs screened by the dynamic performance highly overlap with those screened by the static performance, with 76 % containing open alkali metal sites. This integrated approach of computational screening, machine learning, and dynamic analysis significantly advances the development of efficient MOF adsorbents for flue gas treatment.

Mid-temperature CO2 Adsorption over Different Alkaline Sorbents Dispersed over Mesoporous Al2O3

CO2 adsorbents comprising various alkaline sorption active phases supported on mesoporous Al2O3 were prepared. The materials were tested regarding their CO2 adsorption behavior in the mid-temperature range, i.e., around 300 °C, as well as characterized via XRD, N2 physisorption, CO2-TPD and TEM. It was found that the Na2O sorption active phase supported on Al2O3 (originated following NaNO3 impregnation) led to the highest CO2 adsorption capacity due to the presence of CO2-philic interfacial Al-O--Na+ sites, and the optimum active phase load was shown to be 12 wt % (0.22 Na/Al molar ratio). Additional adsorbents were prepared by dispersing Na2O over different metal oxide supports (ZrO2, TiO2, CeO2 and SiO2), showing an inferior performance than that of Na2O/Al2O3. The kinetics and thermodynamics of CO2 adsorption were also investigated at various temperatures, showing that CO2 adsorption over the best-performing Na2O/Al2O3 material is exothermic and follows the Avrami model, while tests under varying CO2 partial pressures revealed that the Langmuir isotherm best fits the adsorption data. Lastly, Na2O/Al2O3 was tested under multiple CO2 adsorption-desorption cycles at 300 and 500 °C, respectively. The material was found to maintain its CO2 adsorption capacity with no detrimental effects on its nanostructure, porosity and surface basic sites, thereby rendering it suitable as a reversible CO2 chemisorbent or as a support for the preparation of dual-function materials.

Data Driven Discovery of MOFs for Hydrogen Gas Adsorption

Hydrogen gas (H2) is a clean and renewable energy source, but the lack of efficient and cost-effective storage materials is a challenge to its widespread use. Metal-organic frameworks (MOFs), a class of porous materials, have been extensively studied for H2 storage due to their tunable structural and chemical features. However, the large design space offered by MOFs makes it challenging to select or design appropriate MOFs with a high H2 storage capacity. To overcome these challenges, we present a data-driven computational approach that systematically designs new functionalized MOFs for H2 storage. In particular, we showcase the framework of a hybrid particle swarm optimization integrated genetic algorithm, grand canonical Monte Carlo (GCMC) simulations, and our in-house MOF structure generation code to design new MOFs with excellent H2 uptake. This automated, data driven framework adds appropriate functional groups to IRMOF-10 to improve its H2 adsorption capacity. A detailed analysis of the top selected MOFs, their adsorption isotherms, and MOF design rules to enhance H2 adsorption are presented. We found a functionalized IRMOF-10 with an enhanced H2 adsorption increased by ∼6 times compared to that of pure IRMOF-10 at 1 bar and 77 K. Furthermore, this study also utilizes machine learning and deep learning techniques to analyze a large data set of MOF structures and properties, in order to identify the key factors that influence hydrogen adsorption. The proof-of-concept that uses a machine learning/deep learning approach to predict hydrogen adsorption based on the identified structural and chemical properties of the MOF is demonstrated.

Machine Learning-Enhanced Biomass Pressure Sensor with Embedded Wrinkle Structures Created by Surface Buckling

Flexible piezoresistive sensors are core components of many wearable devices to detect deformation and motion. However, it is still a challenge to conveniently prepare high-precision sensors using natural materials and identify similar short vibration signals. In this study, inspired by microstructures of human skins, biomass flexible piezoresistive sensors were prepared by assembling two wrinkled surfaces of konjac glucomannan and k-carrageenan composite hydrogel. The wrinkle structures were conveniently created by hardness gradient-induced surface buckling and coated with MXene sheets to capture weak pressure signals. The sensor was applied to detect various slight body movements, and a machine learning method was used to enhance the identification of similar and short throat vibration signals. The results showed that the sensor exhibited a high sensitivity of 5.1 kPa-1 under low pressure (50 Pa), a fast response time (104 ms), and high stability over 100 cycles. The XGBoost machine learning model accurately distinguished short voice vibrations similar to those of individual English letters. Moreover, experiments and numerical simulations were carried out to reveal the mechanism of the wrinkle structure preparation and the excellent sensing performance. This biomass sensor preparation and the machine learning method will promote the optimization and application of wearable devices.

Estimation of the flow rate of pyrolysis gasoline, ethylene, and propylene in an industrial olefin plant using machine learning approaches

Light olefins, as the backbone of the chemical and petrochemical industries, are produced mainly via steam cracking route. Prediction the of effects of operating variables on the product yield distribution through the mechanistic approaches is complex and requires long time. While increasing in the industrial automation and the availability of the high throughput data, the machine learning approaches have gained much attention due to the simplicity and less required computational efforts. In this study, the potential capability of four powerful machine learning models, i.e., Multilayer perceptron (MLP) neural network, adaptive boosting-support vector regression (AdaBoost-SVR), recurrent neural network (RNN), and deep belief network (DBN) was investigated to predict the product distribution of an olefin plant in industrial scale. In this regard, an extensive data set including 1184 actual data points were gathered during four successive years under various practical conditions. 24 varying independent parameters, including flow rates of different feedstock, numbers of active furnaces, and coil outlet temperatures, were chosen as the input variables of the models and the outputs were the flow rates of the main products, i.e., pyrolysis gasoline, ethylene, and propylene. The accuracy of the models was assessed by different statistical techniques. Based on the obtained results, the RNN model accurately predicted the main product flow rates with average absolute percent relative error (AAPRE) and determination coefficient (R2) values of 1.94% and 0.97, 1.29% and 0.99, 0.70% and 0.99 for pyrolysis gasoline, propylene, and ethylene, respectively. The influence of the various parameters on the products flow rate (estimated by the RNN model) was studied by the relevancy factor calculation. Accordingly, the number of furnaces in service and the flow rates of some feedstock had more positive impacts on the outputs. In addition, the effects of different operating conditions on the propylene/ethylene (P/E) ratio as a cracking severity factor were also discussed. This research proved that intelligent approaches, despite being simple and straightforward, can predict complex unit performance. Thus, they can be efficiently utilized to control and optimize different industrial-scale units.

Compositional modeling of gas-condensate viscosity using ensemble approach

In gas-condensate reservoirs, liquid dropout occurs by reducing the pressure below the dew point pressure in the area near the wellbore. Estimation of production rate in these reservoirs is important. This goal is possible if the amount of viscosity of the liquids released below the dew point is available. In this study, the most comprehensive database related to the viscosity of gas condensate, including 1370 laboratory data was used. Several intelligent techniques, including Ensemble methods, support vector regression (SVR), K-nearest neighbors (KNN), Radial basis function (RBF), and Multilayer Perceptron (MLP) optimized by Bayesian Regularization and Levenberg-Marquardt were applied for modeling. In models presented in the literature, one of the input parameters for the development of the models is solution gas oil ratio (Rs). Measuring Rs in wellhead requires special equipment and is somewhat difficult. Also, measuring this parameter in the laboratory requires spending time and money. According to the mentioned cases, in this research, unlike the research done in the literature, Rs parameter was not used to develop the models. The input parameters for the development of the models presented in this research were temperature, pressure and condensate composition. The data used includes a wide range of temperature and pressure, and the models presented in this research are the most accurate models to date for predicting the condensate viscosity. Using the mentioned intelligent approaches, precise compositional models were presented to predict the viscosity of gas/condensate at different temperatures and pressures for different gas components. Ensemble method with an average absolute percent relative error (AAPRE) of 4.83% was obtained as the most accurate model. Moreover, the AAPRE values for SVR, KNN, MLP-BR, MLP-LM, and RBF models developed in this study are 4.95%, 5.45%, 6.56%, 7.89%, and 10.9%, respectively. Then, the effect of input parameters on the viscosity of the condensate was determined by the relevancy factor using the results of the Ensemble methods. The most negative and positive effects of parameters on the gas condensate viscosity were related to the reservoir temperature and the mole fraction of C₁₁, respectively. Finally, suspicious laboratory data were determined and reported using the leverage technique.

Machine learning implemented exploration of the adsorption mechanism of carbon dioxide onto porous carbons

Adsorption of CO₂ on porous carbons has been identified as one of the promising methods for carbon capture, which is essential for meeting the sustainable developmental goal (SDG) with respect to climate action, i.e., SDG 13. This research implemented six supervised machine learning (ML) models (gradient boosting decision tree (GBDT), extreme gradient boosting (XGB), light boost gradient machine (LBGM), random forest (RF), categorical boosting (Catboost), and adaptive boosting (Adaboost)) to understand and predict the CO₂ adsorption mechanism and adsorption uptake, respectively. The results recommended that the GBDT outperformed the remaining five ML models for CO₂ adsorption. However, XGB, LBGM, RF, and Catboost also represented the prediction in the acceptable range. The GBDT model indicated the accurate prediction of CO₂ uptake onto the porous carbons considering adsorbent properties and adsorption conditions as model input parameters. Next, two-factor partial dependence plots revealed a lucid explanation of how the combinations of two input features affect the model prediction. Furthermore, SHapley Additive exPlainations (SHAP), a novel explication approach based on ML models, were employed to understand and elucidate the CO₂ adsorption and model prediction. The SHAP explanations, implemented on the GBDT model, revealed the rigorous relationships among the input features and output variables based on the GBDT prediction. Additionally, SHAP provided clear-cut feature importance analysis and individual feature impact on the prediction. SHAP also explained two instances of CO₂ adsorption. Along with the data-driven insightful explanation of CO₂ adsorption onto porous carbons, this study also provides a promising method to predict the clear-cut performance of porous carbons for CO₂ adsorption without performing any experiments and open new avenues for researchers to implement this study in the field of adsorption because a lot of data is being generated.

An XGBoost Algorithm Based on Molecular Structure and Molecular Specificity Parameters for Predicting Gas Adsorption

In this paper, an improved Extreme Gradient Boosting (XGBoost) algorithm based on the Graph Isomorphic Network (GIN) for predicting the adsorption performance of metal-organic frameworks (MOFs) is developed. It is shown that the graph isomorphic layer of this algorithm can directly learn the feature representation of materials from the connection of atoms in MOFs. Then, XGBoost can be used to predict the adsorption performance of MOFs based on feature representation. In this sense, it is not only possible to achieve end-to-end prediction directly from the structure of MOFs to adsorption performance but also to ensure the accuracy of prediction. The comparison between Grand Canonical Monte Carlo (GCMC) simulation and prediction supports the performance and effectiveness of the proposed algorithm.

Modeling Viscosity of CO2-N2 Gaseous Mixtures Using Robust Tree-Based Techniques: Extra Tree, Random Forest, GBoost, and LightGBM

Carbon dioxide (CO₂) has an essential role in most enhanced oil recovery (EOR) methods in the oil industry. Oil swelling and viscosity reduction are the dominant mechanisms in an immiscible CO₂-EOR process. Besides numerous CO₂ applications in EOR, most oil reservoirs do not have access to natural CO₂, and capturing it from flue gas and other sources is costly. Flue gases are available in huge quantities at a significantly lower price and can be considered economically viable agents for EOR operations. In this work, four powerful machine learning algorithms, namely, extra tree (ET), random forest (RF), gradient boosting (GBoost), and light gradient boosted machine (LightGBM) were utilized to accurately estimate the viscosity of CO₂-N₂ mixtures. To this aim, a databank was employed, containing 3036 data points over wide ranges of pressures and temperatures. Temperature, pressure, and CO₂ mole fraction were applied as input parameters, and the viscosity of the CO₂-N₂ mixture was the output. The RF smart model had the highest precision with the lowest average absolute percent relative error (AAPRE) of 1.58%, root mean square error (RMSE) of 2.221, and determination coefficient (R ²) of 0.9993. The trend analysis showed that the RF model could precisely predict the real physical behavior of the CO₂-N₂ viscosity variation. Finally, the outlier detection was performed using the leverage approach to demonstrate the validity of the utilized databank and the applicability area of the developed RF model. Accordingly, nearly 96% of the data points seemed to be dependable and valid, and the rest of them were located in the suspected and out-of-leverage data zones.

Machine learning approaches for predicting arsenic adsorption from water using porous metal-organic frameworks

Arsenic in drinking water is a serious threat for human health due to its toxic nature and therefore, its eliminating is highly necessary. In this study, the ability of different novel and robust machine learning (ML) approaches, including Light Gradient Boosting Machine (LightGBM), Extreme Gradient Boosting, Gradient Boosting Decision Tree, and Random Forest was implemented to predict the adsorptive removal of arsenate [As(V)] from wastewater over 13 different metal–organic frameworks (MOFs). A large experimental dataset was collected under various conditions. The adsorbent dosage, contact time, initial arsenic concentration, adsorbent surface area, temperature, solution pH, and the presence of anions were considered as input variables, and adsorptive removal of As(V) was selected as the output of the models. The developed models were evaluated using various statistical criteria. The obtained results indicated that the LightGBM model provided the most accurate and reliable response to predict As(V) adsorption by MOFs and possesses R², RMSE, STD, and AAPRE (%) of 0.9958, 2.0688, 0.0628, and 2.88, respectively. The expected trends of As(V) removal with increasing initial concentration, solution pH, temperature, and coexistence of anions were predicted reasonably by the LightGBM model. Sensitivity analysis revealed that the adsorption process adversely relates to the initial As(V) concentration and directly depends on the MOFs surface area and dosage. This study proves that ML approaches are capable to manage complicated problems with large datasets and can be affordable alternatives for expensive and time-consuming experimental wastewater treatment processes.

Prediction of the Ibuprofen Loading Capacity of MOFs by Machine Learning

Metal-organic frameworks (MOFs) have been widely researched as drug delivery systems due to their intrinsic porous structures. Herein, machine learning (ML) technologies were applied for the screening of MOFs with high drug loading capacity. To achieve this, first, a comprehensive dataset was gathered, including 40 data points from more than 100 different publications. The organic linkers, metal ions, and the functional groups, as well as the surface area and the pore volume of the investigated MOFs, were chosen as the model’s inputs, and the output was the ibuprofen (IBU) loading capacity. Thereafter, various advanced and powerful machine learning algorithms, such as support vector regression (SVR), random forest (RF), adaptive boosting (AdaBoost), and categorical boosting (CatBoost), were employed to predict the ibuprofen loading capacity of MOFs. The coefficient of determination (R²) of 0.70, 0.72, 0.66, and 0.76 were obtained for the SVR, RF, AdaBoost, and CatBoost approaches, respectively. Among all the algorithms, CatBoost was the most reliable, exhibiting superior performance regarding the sparse matrices and categorical features. Shapley additive explanations (SHAP) analysis was employed to explore the impact of the eigenvalues of the model’s outputs. Our initial results indicate that this methodology is a well generalized, straightforward, and cost-effective method that can be applied not only for the prediction of IBU loading capacity, but also in many other biomaterials projects.