Performance Of Metal-organic Frameworks For Gas Adsorption And Adsorption-driven Heat Pumps By Machine Learning

Based on metal-organic frameworks（MOFs）as an adsorbent,high-throughtput computation（HTC）and machine learning（ML）were employed to study the performance of MOF-three systems（CO₂adsorption separation,atmospheric water capture and adsorption heat pump）,respectively.HTC include grand canonical Monte Carlo（GCMC）and molecular dynamics（MD）,and ML is dominated by nonlinear algorithms such as back propagation neural network and random forest.Two MOF databases were applied in this work,they were derived from computation-ready experimental MOFs（Co RE-MOFs）and hypothetical MOFs（h MOFs）.This study is divided into four chapters,the first part is CO₂ adsorption separation,the second part is atmospheric water capture,the last two parts are the AHP/AC performance and cost analysis,the specific study is as follows:1.The technique of capturing CO₂ from high CO₂ concentrations in recent years,such as power plants,has been widely studied,but the capture of low concentrations of CO₂ directly from the air still is a large challenge.GCMC,MD and ML were used to study 6013 Co RE-MOFs for CO₂ adsorption and diffffusion properties in air with very low concentrations of CO₂.First,influence of CO₂ adsorption and diffusion in air is obtained as the structure-performance relationships,and then the performance of CO₂adsorption and diffusion in air is further explored by four ML algorithms.Random forest（RF）was considered the optimal algorithm for prediction of CO₂selectivity,with an R value of 0.981,and this algorithm was further applied to quantitatively analyze the relative importance of each MOF descriptor.Finally,14 MOFs with the best properties were successfully screened out,and it was found that a key to capture a low concentration CO₂ from the air was the diffffusion performance of CO₂ in MOFs.When the pore-limiting diameter of a MOF was close to the CO₂ dynamic diameter,this MOF could possess high CO₂ diffffusion separation selectivity.2.The atmospheric water harvesting by strong adsorbents is a feasible method to solve the shortage of water resources,especially for arid regions.GCMC is employed to calculate the capture of H₂O from air for the 6013 Co RE-MOFs and 137,953hypothetical MOFs（h MOFs）.Through the univariate analysis of MOF structure-performance relationships,Q_st seems to be a key descriptor.Moreover,four ML algorithms are applied to predict the complicated interrelation between six descriptors and performance,and gradient boosting regression tree（GBRT）with R² of 0.9974possesses the best predictive ability.In addition,the trained GBRT model can predict the selectivity of h MOFs with an R² of 0.88.Besides,based on the relative importance of descriptors by ML,it can be quantitatively concluded that the Q_st is dominant governing the capture of H₂O.Finally,10 optimal Co RE-MOF and 10 h MOF with high performance are identified.3.With the increasingly serious energy-consumption,the adsorption-driven heat pumps/chillers（AHPs/ACs）were paid more attention.To select the promising adsorbents and suitable working flfluids in this technology,a high-throughput computational screening of 6013 computation-ready experimental metal organic frameworks（MOFs）is employed to identity the best candidates for methanol-MOF pairs in AHPs/ACs.The highest working capacity（（35）W）and coeffificient of performance（COP）under operating condition of heat pump are 512.86 mg/g and 1.83,respectively.The structure-performance relationships are derived for（35）W and COP with6 MOF descriptors,the saturated（35）W is,for the first time,defined for the improvement of heat pump effificiency,and 10 best MOFs are identified.Then four machine learning algorithms are applied to predict the relative importance of each MOF descriptor.The microscopic insights obtained from our bottom-up approach provide the guidelines for the design of new MOFs in adsorption-driven heat pumps and chillers.4.The key to achieving high efficiencies,high performance,and low costs of AHPs/ACs is to choose a suitable adsorbent.A computational screening of 6,013Co RE-MOFs is performed for methanol-MOF pairs in AHPs/ACs,and 137,953 h MOFs are further directionally screened based on the range of optimal Co RE-MOF descriptors.Comparing with the blind screening of Co RE-MOFs（～13%）,～80%of h MOFs by the directional screening exhibited high performance in AHPs/ACs,proving the effectiveness of directional screening.Then,the techno-economic analysis(C_total)of AHPs/ACs for each MOF was performed based on the equipment cost(C_equipment),cycle cost(C_cycle),and material cost(C_MOF).The C_equipment accounted for the greatest proportion of C_total,but the proportion of C_equipment decreased and the proportion of C_cyclegradually increased as the MOF data set tended to be the materials with better performance.This confirmed the reduction of C_total by an effective enhancement of C_cycle proportion,and that the high manufacturing cost of MOFs would not restrict their widespread use in AHPs/ACs.Among 6 ML algorithms,the random forest could yield the best prediction effect with little effect by the metal type of MOFs,whereas the heat of adsorption and MOF density were two key descriptors to determine the C_total.12lowest-cost Co RE-MOFs and h MOFs were identified for each application,and their average costs at state 2 were only～1 USD/k J in AHPs/ACs.Finally,a variant of the well-known MOF（Cu₃BTC₂）was predicted to possess superior techno-economy,which was confirmed by the parallel methanol adsorption experiment.These comprehensive insights from directional computational screening,ML algorithms to experiment can guide the development of low-cost MOFs for AHPs/ACs in a variety of energy conservation and industrial applications.Through the combination of molecular simulation（MS）and ML,the relationships between MOF-guest molecules in gas adsorption and AHP applications are explored in depth.The results show that ML-assisted MS can not only speed up HTC,but also excavate deep hidden information between properties and structures of materials to guide the design and synthesis of new MOFs.