94 research outputs found
A Microkinetic Understanding of CO Electro-Oxidation on Bimetallic Catalysts
No full textIn this work, we analyze the fundamental microkinetics of electro-oxidation reactions, specifically focusing on the CO electro-oxidation reaction (COOR). We first study the interplay between bifunctional and electronic effects on theoretical dual-site electrocatalysts for the COOR. We identify the general conditions that are necessary to engender bifunctional activity for generalized, dual-site materials. We then perform a case study on PtRu and find that the model which best captures the known experimental trend is one where the COOR occurs on Pt sites that are electronically modified by Ru, as opposed to a bifunctional mechanism wherein Ru provides reactive OH*. Next, we derive general analytical expressions for several macrokinetic observables in electrocatalysis, namely, the apparent transfer coefficient (Tafel slope), the apparent activation energy, and the apparent reaction orders. Using Campbell’s generalized degree of rate control (DRC), these expressions can be written in simple terms of experimentally-accessible quantities such as the surface coverages of intermediates. We then apply these expressions in an experimental study of the COOR on a series of Pd/C and AgPd/C alloys under alkaline conditions, where we find that the model which best explains the experimental trends in apparent transfer coefficient and apparent CO reaction order is one where Ag provides OH* adspecies via a non-competitive bifunctional mechanism at low potential shifting to a competitive Langmuir-Hinshelwood mechanism at higher potential. Finally, we perform another generalized microkinetic study probing the extent to which an oscillation of potential can be used as a lever to induce dynamic catalytic rate enhancements. We find that for series, faradaic reactions, oscillation of potential cannot create TOFs greater than the static Sabatier maximum over a given potential range, unless faradaically-driven poisoning species are present. However, small enhancements in TOF relative to the average steady-state TOF can be achieved. We discuss some of the thermodynamic implications of utilizing potential to drive this enhancement.</p
A Microkinetic Understanding of CO Electro-Oxidation on Bimetallic Catalysts
No full textIn this work, we analyze the fundamental microkinetics of electro-oxidation reactions, specifically focusing on the CO electro-oxidation reaction (COOR). We first study the interplay between bifunctional and electronic effects on theoretical dual-site electrocatalysts for the COOR. We identify the general conditions that are necessary to engender bifunctional activity for generalized, dual-site materials. We then perform a case study on PtRu and find that the model which best captures the known experimental trend is one where the COOR occurs on Pt sites that are electronically modified by Ru, as opposed to a bifunctional mechanism wherein Ru provides reactive OH*. Next, we derive general analytical expressions for several macrokinetic observables in electrocatalysis, namely, the apparent transfer coefficient (Tafel slope), the apparent activation energy, and the apparent reaction orders. Using Campbell’s generalized degree of rate control (DRC), these expressions can be written in simple terms of experimentally-accessible quantities such as the surface coverages of intermediates. We then apply these expressions in an experimental study of the COOR on a series of Pd/C and AgPd/C alloys under alkaline conditions, where we find that the model which best explains the experimental trends in apparent transfer coefficient and apparent CO reaction order is one where Ag provides OH* adspecies via a non-competitive bifunctional mechanism at low potential shifting to a competitive Langmuir-Hinshelwood mechanism at higher potential. Finally, we perform another generalized microkinetic study probing the extent to which an oscillation of potential can be used as a lever to induce dynamic catalytic rate enhancements. We find that for series, faradaic reactions, oscillation of potential cannot create TOFs greater than the static Sabatier maximum over a given potential range, unless faradaically-driven poisoning species are present. However, small enhancements in TOF relative to the average steady-state TOF can be achieved. We discuss some of the thermodynamic implications of utilizing potential to drive this enhancement.</p
Application, Evaluation and Development of Density Functional Theory for Discrete Systems
No full textDensity functional theory (DFT) has been used extensively for first-principles quantum chemical modeling of the electronic structures of molecules and materials. This work involves application, evaluation and development of DFT for molecular systems. We first implement DFT for evaluating reducing characteristics of phosphorus-based hydrides. Then, we evaluate, benchmark and analyze a wide spectrum of DFT methods based on their eigenvalues and link that with the accuracy of their predictions. Lastly, we propose an efficient procedure for parameterizing hybrid DFT functionals so that systematic errors in predictions can be avoided.
Diazaphospholenes have emerged as a promising class of metal-free hydride donors and have been implemented as molecular catalysts in several reduction reactions. Recent studies have also verified their radical reactivity as hydrogen atom donors. Experimental quantification of the hydricities and electrochemical properties of this unique class of hydrides has been limited by their sensitivity towards oxidation in open air and moist environments. Here, we implement quantum chemical density functional theory calculations to analyze the electrochemical catalytic cycle of diazaphospholenes in acetonitrile. We report computed hydricities, reduction potentials, pKa values, and bond dissociation free energies (BDFEs) for 64 P-based hydridic catalysts generated by functionalizing 8 main structures with 8 different electron donating/withdrawing groups. Our results demonstrate that a wide range of hydricities (29-66 kcal/mol) and BDFEs (58-81 kcal/mol) are attainable by functionalizing diazaphospholenes. Compared to the more common carbon-based hydrides, diazaphospholenes are predicted to require less negative reduction potentials to electrochemically regenerate hydrides with an equivalent hydridic strength, indicating their higher energy efficiency in the tradeoff between thermodynamic ability and reduction potential.
Then, we benchmark 274 DFT functionals for how closely their computed HOMO and LUMO eigenvalues predict experimental negative ionization potentials (-IPs) and electron affinities (-EAs). The importance of predicting accurate HOMO energies is demonstrated by showing that it correlates with the accuracy of predicting other molecular properties, such as activation barriers, atomization energies, proton affinities and total atom energies where the accuracy of these properties computed by DFT functionals is directly proportional to the accuracy of their HOMO eigenvalues. The mean absolute deviations (MADs) of calculated HOMO energies from the experimental -IPs of 78 molecules with a wide range of molecular sizes are compared using a broad set of pure, hybrid and long-range corrected (LC) functionals. The functionals produced MADIPs of 2.5-5.0, 0.4-3.3 and 0.5-1.0 eV for pure, hybrid and LC functionals, respectively. LUMO energies and HOMO-LUMO gaps are benchmarked against EAs and first excitation energies (τ’s), respectively. LUMOs of LC and hybrid functionals correlate with EAs, while those of pure functionals correspond better to τ.
Finally, we develop a reparametrized version of B3LYP method based on a simple optimization algorithm that targets to adjust both eigenvalues and total atomic energies (TAEs). Since these are the most basic energies predicted by DFT methods, their accuracy is expected to improve any further energetic predictions. The new functional exhibits MADs of 0.8 and 0.3 eV in predicting HOMO eigenvalues and TAEs compared to 3.1 and 0.5 eV in B3LYP, respectively. This significant improvement, especially in eigenvalues, resulted in fixing some of the documented shortcomings of the original B3LYP, such as the underestimation of transition state barrier heights and the inaccurate description of large molecular systems. For example, the new method predicts barrier heights that are 2 kcal/mol more accurate than those predicted by B3LYP. Moreover, it shows a remarkable enhancement of energetic predictions associated with multiple difficult cases for DFT. The improvement achieved by the new method in energetic predictions is coupled with predicting more accurate electron densities which makes it an advancement in the correct direction of developing more accurate DFT functionals that predict accurate energetics based on accurate electron densities. The simple parameterization algorithm presented in this work can be utilized in future development of DFT methods in order to avoid overfitting that might result from extensive empiricism.</p
Application, Evaluation and Development of Density Functional Theory for Discrete Systems
No full textDensity functional theory (DFT) has been used extensively for first-principles quantum chemical modeling of the electronic structures of molecules and materials. This work involves application, evaluation and development of DFT for molecular systems. We first implement DFT for evaluating reducing characteristics of phosphorus-based hydrides. Then, we evaluate, benchmark and analyze a wide spectrum of DFT methods based on their eigenvalues and link that with the accuracy of their predictions. Lastly, we propose an efficient procedure for parameterizing hybrid DFT functionals so that systematic errors in predictions can be avoided.
Diazaphospholenes have emerged as a promising class of metal-free hydride donors and have been implemented as molecular catalysts in several reduction reactions. Recent studies have also verified their radical reactivity as hydrogen atom donors. Experimental quantification of the hydricities and electrochemical properties of this unique class of hydrides has been limited by their sensitivity towards oxidation in open air and moist environments. Here, we implement quantum chemical density functional theory calculations to analyze the electrochemical catalytic cycle of diazaphospholenes in acetonitrile. We report computed hydricities, reduction potentials, pKa values, and bond dissociation free energies (BDFEs) for 64 P-based hydridic catalysts generated by functionalizing 8 main structures with 8 different electron donating/withdrawing groups. Our results demonstrate that a wide range of hydricities (29-66 kcal/mol) and BDFEs (58-81 kcal/mol) are attainable by functionalizing diazaphospholenes. Compared to the more common carbon-based hydrides, diazaphospholenes are predicted to require less negative reduction potentials to electrochemically regenerate hydrides with an equivalent hydridic strength, indicating their higher energy efficiency in the tradeoff between thermodynamic ability and reduction potential.
Then, we benchmark 274 DFT functionals for how closely their computed HOMO and LUMO eigenvalues predict experimental negative ionization potentials (-IPs) and electron affinities (-EAs). The importance of predicting accurate HOMO energies is demonstrated by showing that it correlates with the accuracy of predicting other molecular properties, such as activation barriers, atomization energies, proton affinities and total atom energies where the accuracy of these properties computed by DFT functionals is directly proportional to the accuracy of their HOMO eigenvalues. The mean absolute deviations (MADs) of calculated HOMO energies from the experimental -IPs of 78 molecules with a wide range of molecular sizes are compared using a broad set of pure, hybrid and long-range corrected (LC) functionals. The functionals produced MADIPs of 2.5-5.0, 0.4-3.3 and 0.5-1.0 eV for pure, hybrid and LC functionals, respectively. LUMO energies and HOMO-LUMO gaps are benchmarked against EAs and first excitation energies (τ’s), respectively. LUMOs of LC and hybrid functionals correlate with EAs, while those of pure functionals correspond better to τ.
Finally, we develop a reparametrized version of B3LYP method based on a simple optimization algorithm that targets to adjust both eigenvalues and total atomic energies (TAEs). Since these are the most basic energies predicted by DFT methods, their accuracy is expected to improve any further energetic predictions. The new functional exhibits MADs of 0.8 and 0.3 eV in predicting HOMO eigenvalues and TAEs compared to 3.1 and 0.5 eV in B3LYP, respectively. This significant improvement, especially in eigenvalues, resulted in fixing some of the documented shortcomings of the original B3LYP, such as the underestimation of transition state barrier heights and the inaccurate description of large molecular systems. For example, the new method predicts barrier heights that are 2 kcal/mol more accurate than those predicted by B3LYP. Moreover, it shows a remarkable enhancement of energetic predictions associated with multiple difficult cases for DFT. The improvement achieved by the new method in energetic predictions is coupled with predicting more accurate electron densities which makes it an advancement in the correct direction of developing more accurate DFT functionals that predict accurate energetics based on accurate electron densities. The simple parameterization algorithm presented in this work can be utilized in future development of DFT methods in order to avoid overfitting that might result from extensive empiricism.</p
Predicting macro-kinetic observables in electrocatalysis using the generalized degree of rate control
No full textApplication of Grand Canonical Density Functional Theory to Electrocatalytic Interfaces
No full textGrand canonical DFT (GC-DFT) provides a fundamentally correct and accurate description of electrified interfaces and electrocatalysis. GC-DFT calculates the grand free energy at an arbitrary potential by optimizing the grand free energy while self-consistently solving for the number of electrons that matches the applied potential rather than calculating the electronic energy of the system with a fixed number of electrons. Therefore, this work implements GC-DFT to model electrocatalytic interfaces. We first study the effect of the electrochemical environment on the pure metal surfaces. Then, we utilize CG-DFT to model the electrochemical CO2 reduction to CO on selected metal surfaces and compare the results with the Computational Hydrogen Electrode (CHE) approach which is commonly used to model electrochemical reactions. Lastly, we model the full reaction pathway for CO2 reduction to CO over TiN4C site including transition states and kinetic barriers.
We use GC-DFT to predict the surface energies, Wulff shapes, charge distributions and catalytically active sites of different metal surfaces under electrochemical conditions. We propose a method for computing surface energies from GC-DFT calculations of periodic slab models and use it to compute the surface energies of facets of pure metallic crystals to predict their Wulff shapes under electrochemical conditions. GC-DFT predicts that, for the pure metals studied, solvation only slightly affects the Wulff shape while applied potentials considerably affect the surface energies and corresponding Wulff shapes. We study the effect of the applied potential on the distribution of electron density over the atoms of surfaces of pure metals and alloys. This analysis shows that, under an applied potential, the electron density is unevenly distributed over the surface atoms and that the charges of atoms more exposed to solvent are more sensitive to bias. Our results also show that the most sensitive atom to bias can be used to identify the most favorable adsorption site and thus, the active sites of electrochemical reactions.
Then, we report the results of modeling CO2 reduction (CO2R) to CO over Ag(110) and Cu(211) surfaces at different applied potentials using GC-DFT and we compare it with the CHE approach. We modeled the reaction on the most favorable adsorption sites as obtained from the charge sensitivity to bias. GC-DFT predicts that the geometries of theses reacting systems depend on the applied potential and the Helmholtz free energies vary with the applied potential, which contradicts a central assumption of the CHE approach. The CHE approach neglects the change in the number of electrons on the electrode surface at different applied potentials, which reduces its accuracy as the potential changes from the potential of zero charge (PZC). Our results further demonstrate that the grand free energies of the reaction intermediates not only depend on the value of the applied potential, but also on the metal surface type, adsorption site, and adsorbate. GC-DFT’s ability to predict the effect of the applied potential on adsorbate geometry enables it to evaluate different possible reaction mechanisms at different applied potentials. For instance, GC-DFT predicts that the first step of CO2R likely switches from proton coupled electron transfer to sequential electron transfer then proton transfer at more reducing potentials, a result that cannot be determined by CHE because it assumes all electron transfers are coupled to proton transfers and neglects the effect of the applied potential on the adsorbate geometry.
Finally, we use GC-DFT to model the electrocatalytic CO2R to CO by TiN4C, including the activation energies of the elementary steps at various applied potentials, and the thermodynamics of CO2R to CO catalyzed by TiNxC defects. Based on the thermodynamic barriers, TiN4C and all defect configurations are predicted to be promising catalysts for CO2R to CO at certain applied potential range. We propose a criterion to choose the optimum applied potential for CO2R to CO based on the PZC of the reaction intermediates and the contention that the optimum applied potential for CO2R to CO lies in the range PZCC*CO) < V < PZC*CO2 ) that can be generalized to other electrocatalytic systems. Solvating H2O molecules are predicted to form strong hydrogen bonds to *COOH, especially at more oxidizing potentials, which significantly influence the thermodynamic barrier for *COOH protonation. Grand canonical nudged elastic band (GC-NEB) was used to predict the rate determining step (RDS) of CO2R to CO catalyzed by TiN4C. GC-NEB predicts that the RDS is potential dependent where CO desorption and CO2 adsorption are RDS at highly reducing and highly oxidizing potentials, respectively.</p
Application of Grand Canonical Density Functional Theory to Electrocatalytic Interfaces
No full textGrand canonical DFT (GC-DFT) provides a fundamentally correct and accurate description of electrified interfaces and electrocatalysis. GC-DFT calculates the grand free energy at an arbitrary potential by optimizing the grand free energy while self-consistently solving for the number of electrons that matches the applied potential rather than calculating the electronic energy of the system with a fixed number of electrons. Therefore, this work implements GC-DFT to model electrocatalytic interfaces. We first study the effect of the electrochemical environment on the pure metal surfaces. Then, we utilize CG-DFT to model the electrochemical CO2 reduction to CO on selected metal surfaces and compare the results with the Computational Hydrogen Electrode (CHE) approach which is commonly used to model electrochemical reactions. Lastly, we model the full reaction pathway for CO2 reduction to CO over TiN4C site including transition states and kinetic barriers.
We use GC-DFT to predict the surface energies, Wulff shapes, charge distributions and catalytically active sites of different metal surfaces under electrochemical conditions. We propose a method for computing surface energies from GC-DFT calculations of periodic slab models and use it to compute the surface energies of facets of pure metallic crystals to predict their Wulff shapes under electrochemical conditions. GC-DFT predicts that, for the pure metals studied, solvation only slightly affects the Wulff shape while applied potentials considerably affect the surface energies and corresponding Wulff shapes. We study the effect of the applied potential on the distribution of electron density over the atoms of surfaces of pure metals and alloys. This analysis shows that, under an applied potential, the electron density is unevenly distributed over the surface atoms and that the charges of atoms more exposed to solvent are more sensitive to bias. Our results also show that the most sensitive atom to bias can be used to identify the most favorable adsorption site and thus, the active sites of electrochemical reactions.
Then, we report the results of modeling CO2 reduction (CO2R) to CO over Ag(110) and Cu(211) surfaces at different applied potentials using GC-DFT and we compare it with the CHE approach. We modeled the reaction on the most favorable adsorption sites as obtained from the charge sensitivity to bias. GC-DFT predicts that the geometries of theses reacting systems depend on the applied potential and the Helmholtz free energies vary with the applied potential, which contradicts a central assumption of the CHE approach. The CHE approach neglects the change in the number of electrons on the electrode surface at different applied potentials, which reduces its accuracy as the potential changes from the potential of zero charge (PZC). Our results further demonstrate that the grand free energies of the reaction intermediates not only depend on the value of the applied potential, but also on the metal surface type, adsorption site, and adsorbate. GC-DFT’s ability to predict the effect of the applied potential on adsorbate geometry enables it to evaluate different possible reaction mechanisms at different applied potentials. For instance, GC-DFT predicts that the first step of CO2R likely switches from proton coupled electron transfer to sequential electron transfer then proton transfer at more reducing potentials, a result that cannot be determined by CHE because it assumes all electron transfers are coupled to proton transfers and neglects the effect of the applied potential on the adsorbate geometry.
Finally, we use GC-DFT to model the electrocatalytic CO2R to CO by TiN4C, including the activation energies of the elementary steps at various applied potentials, and the thermodynamics of CO2R to CO catalyzed by TiNxC defects. Based on the thermodynamic barriers, TiN4C and all defect configurations are predicted to be promising catalysts for CO2R to CO at certain applied potential range. We propose a criterion to choose the optimum applied potential for CO2R to CO based on the PZC of the reaction intermediates and the contention that the optimum applied potential for CO2R to CO lies in the range PZCC*CO) < V < PZC*CO2 ) that can be generalized to other electrocatalytic systems. Solvating H2O molecules are predicted to form strong hydrogen bonds to *COOH, especially at more oxidizing potentials, which significantly influence the thermodynamic barrier for *COOH protonation. Grand canonical nudged elastic band (GC-NEB) was used to predict the rate determining step (RDS) of CO2R to CO catalyzed by TiN4C. GC-NEB predicts that the RDS is potential dependent where CO desorption and CO2 adsorption are RDS at highly reducing and highly oxidizing potentials, respectively.</p
A Theoretical Study of the Electrochemical CO2 Reduction Over Single Atom Catalysts Using Grand-Canonical DFT
No full textElectrochemical CO2 reduction (CO2ER) is a rational method to mitigate climate change effects. The first step of this process is CO formation, which can be selectively achieved over gold or silver metals. However, their high overpotentials, high costs, and/or low atomic efficiencies limit their usage industrially. Metal/nitrogen-doped graphene (MNC) electrocatalysts, which are atomically efficient, have been found to electroreduce CO2 to CO selectively. Herein, we test CO2ER to CO over 3d pyridinic MN4Cs using grand-canonical DFT to elucidate the applied potential effects on reaction energetics and electronic structures as well as to predict promising CO-producing sites. Our results capture the applied potential effects and display nonlinear energy changes with the applied potential. We conclude that TiN4C is a promising CO-producing site based on a pure thermodynamic analysis. However, further kinetic investigations are required to assess CO2ER activity. At the end of this paper, we discuss study limitations and future outlook.</p
Kinetic Analysis of Electrochemical Oxygen Reduction and Development of Ag-alloy Catalysts for Low Temperature Fuel Cells.
No full textThis dissertation applies insights from quantum chemical calculations and heterogeneous kinetic analysis to interpret macroscopic reactivity trends in electrochemical systems and design optimal electrocatalysts. Specifically we explore the mechanism of the electrochemical oxygen reduction reaction (ORR) on the surfaces of Pt (a near-optimal catalyst) and Ag electrodes. We have identified design criteria for improving the reaction rate in each case and developed cost-effective Ag-based alloy materials with activity approaching that of more costly Pt catalysts.
We first demonstrate, using microkinetic modeling and density functional theory calculations, that deviations from ideal electrode kinetics (a linear potential vs. log current relationship) are inherent to the ORR and any multi-step heterogeneous electrocatalytic reaction. Deviations result from simultaneous changes in the rate of the rate-limiting elementary step and the number of available active sites on the electrode surface as potential is shifted. We show the ORR kinetic variations on Pt electrodes are well-reproduced by a simple description of changes in OH and H2O surface intermediate coverages, and that weaker binding materials exhibit higher rates due to higher active-site availability. In contrast, on Ag a very weak relation is found between adsorbate coverage and changes in the apparent rate law. This points to a strong role of under-coordinated active sites, which become poisoned at low potentials while the majority of the surface is still clean. Moving toward stronger binding on Ag should yield higher ORR activity by increasing turnover rates on the more predominant surface facets.
Using the mechanistic insights mentioned, we illustrate the design of relatively inexpensive Ag-Co surface alloy nanoparticle electrocatalysts for ORR, with equivalent area-specific activity to commercial Pt-nanoparticles at realistic fuel cell operating conditions. The Ag-Co materials were identified with quantum chemical calculations and synthesized with a novel bimetallic precursor decomposition technique that generates a surface alloy, despite bulk immiscibility of the elements. Characterization studies show the origin of activity improvement comes from a ligand effect, in which Co perturbs Ag surface sites. We also explore bimetallic precursor decomposition to produce Ag-Ni and Ag-Fe alloys but find that the products exhibit substantial segregation and have ORR activities similar to monometallic Ag.PhDChemical EngineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/102423/1/ahole_1.pd
Phosphonic Acid SAM Modification of Metal Oxides Toward The Creation of Better Dehydration Catalysts
No full textIn this Thesis, the synthesis and performance of new catalysts tailored specifically for dehydration reactions is investigated. Due to their inexpensiveness and bifunctionality, metal oxides have been successfully used to catalyze elimination reactions. However, as is habitual over heterogeneous catalysts, selectivity toward desired products is low for the system at hand: materials that can perform dehydration reactions also catalyze dehydrogenation and condensation reactions. Self-assembling monolayers (SAMs) have been used in this contribution as an approach to improve performance of some already existent catalysts.
Previous results showed that over TiO2 Anatase, the dipole moment of phosphonic acid SAMs affected the near-surface electrostatics, enabling regulation of the dehydration activity of primary alcohols by changing the elongation of the Cβ - H bond, whose scission limits the reaction. Activities and selectivities of the coated catalysts exceeded those of the native, uncoated materials.
The same approach has been tested on several metal oxides. Out of the different oxides that have been screened, an overall beneficial effect on dehydration was only observed for a group of them. These materials (TiO2, CeO2, SnO2) have a common descriptor: their moderate metal – oxygen bond strength (~4 eV). These results indicate that the response of dehydration activity and functionalization with phosphonic acids appears to be correlated with the electron mobility and energetics of the material.
Similarly, the effects of PA deposition have been tested with different alcohols, over TiO2: 1-butanol, 2-butanol and tert-butanol. Due to their different inherent ability to stabilize charges, the inductive effects generated as a result of the Cβ - H bond elongation were more ineffective for the higher substituted reactants. Several additional effects caused by the presence of SAMs such as changes in water mobility as a surface species and steric impediments caused by the interruption of the catalytic surface were observed. These interconnected phenomena collectively could explain the observed rates and provided experimental insight on the mechanism behind PAs effects on metal oxide surfaces and dehydration reactions.</p
- …
