Computer-generated kinetics for coupled heterogeneous

Subscriber access provided by Columbia University Libraries

Kinetics, Catalysis, and Reaction Engineering

Computer-generated kinetics for coupled heterogeneous/homogeneous systems: a case study in catalytic combustion of methane on platinum Katrin Blondal, Jelena Jelic, Emily J Mazeau, Felix Studt, Richard Henry West, and C. Franklin Goldsmith Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.9b01464 • Publication Date (Web): 27 Aug 2019 Downloaded from pubs.acs.org on September 1, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Computer-generated kinetics for coupled heterogeneous/homogeneous systems: a case study in catalytic combustion of methane on platinum Katrin Blondal,†,§ Jelena Jelic,‡,§ Emily Mazeau,¶ Felix Studt,‡ Richard H. West,¶ and C. Franklin Goldsmith∗,† †Chemical Engineering Group, School of Engineering, Brown University, Providence, RI 02912, USA ‡Institute of Catalysis Research and Technology, Karlsruhe Institute of Technology, 76344 Eggenstein-Leopoldshafen, Germany ¶Department of Chemical Engineering, Northeastern University, Boston, MA 02115, USA §These authors contributed equally to this work E-mail: [email protected] Abstract The automatic microkinetic mechanism generator for heterogeneous catalysis, RMGCat, has been updated extensively. Density functional theory calculations were performed for 69 adsorbates on Pt(111), and the resulting thermodynamic properties were added to RMG-Cat. The thermo database is significantly more accurate; it includes nitrogen-containing adsorbates for the first time, as well as better capabilities for predicting the thermochemistry of novel adsorbates. Additionally, RMG-Cat can now

1

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

simultaneously pursue mechanism expansion on both the surface and in the gas-phase. This heterogeneous/homogeneously coupled capability is tested on the catalytic combustion of methane on platinum. The results confirm that under some conditions, the catalyst is capable of inducing thermal ignition in the gas-phase.

Introduction A key trend in chemical engineering is the shift towards a more fundamental understanding of reaction kinetics at the atomic scale. This emphasis on elementary reactions allows for more predictive models that can be used in material design and process optimization. One of the largest problems facing microkinetics is complexity. Not only does the microkinetic mechanism contain a large number of intermediate species and elementary reactions, each of which must be parameterized in some form, but the number of possible species and reactions that need to be considered typically is vastly larger. Not surprisingly, there has been an explosion of interest in using computers to automate large portions of the process of building microkinetic mechanisms. These efforts tackle the problem of complexity from different perspectives. Some methods focus on automating the search for transition states. 1,2 Other methods apply machine learning techniques to screen for materials. 3–6 A third group, which has its origins primarily in the gas-phase community, apply a set of graph-based rules. 7–14 Finally, other methods seek to reduce the complexity by automating the mechanism analysis process. 15 The present work is focused on RMG-Cat of Goldsmith and West, which is an offshoot of the gas-phase RMG of Green and West. 12 The details of RMG and RMG-Cat are presented in References 12 and 13, respectively, so only a cursory summary is provided here. Gasphase species and adsorbates on metals are represented as chemical graphs, with atoms as nodes and chemical bonds as edges. Chemical reactions are classified by reaction types, and specific reaction families are recipes to convert chemical graphs from reactants to products. RMG-Cat has a database of thermodynamic properties, as well as the ability to estimate 2


Page 2 of 32

Page 3 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the thermodynamic properties of novel species. Similarly, the code has a database of predetermined rate constants and the ability to estimate the rate constants for new reactions. Finally, RMG-Cat uses a flux-based approach for mechanism expansion. 16 Chemical species that are essential to the kinetics of the process are stored in the “core”; new species that can be formed from an elementary reaction are initially stored in the “edge”. When the rate of formation of an edge species crosses a threshold, that species is moved from the edge to the core, and the process begins again. This process continues until the code reaches a userdefined termination criterion without moving additional edge species to the core. RMG and RMG-Cat excel at generating accurate mechanisms for the transient behavior of complex systems, particularly for systems involving higher molecular-weight compounds, where the number of possible reactions to consider can be vast. 17 RMG-Cat has now been merged into the master branch of RMG (version 2.4), and all the features described herein are now available as part of the RMG software suite. 12,18 This paper presents two significant advancements of RMG-Cat’s capabilities for building microkinetic mechanisms in heterogeneous catalysis. First, the thermodynamic database is substantially expanded to be both larger and more accurate. Second, RMG-Cat now can explore kinetic pathways simultaneously on the surface and in the gas phase. This new capability is particularly important for chemical processes that occur at high temperatures, where catalytic surfaces can lead to gas-phase radical chemistry, such as catalytic partial oxidation, 19,20 oxidative 21,22 and non-oxidative 23 coupling of methane to ethene, the Ostwald process, 24 as well as other systems with substantial interaction between simultaneous homogeneous and heterogenous chemistry. 25 To test these new features, we consider the catalytic combustion of methane on platinum. The result of these simulations confirm that the rapid temperature rise associated with the surface oxidation chemistry can lead to thermal ignition.

3



Page 4 of 32

Methods Database Development The thermodynamic database in the original release of RMG-Cat consisted of 21 H/C/Ocontaining species, which were taken from a DFT study on methane steam reforming on Ni(111). 26 The new thermo database is a notable improvement in three key respects. First, nitrogen was added to the system. Second, the thermodynamic database was restructured to include all major adsorbates that consists of up to two heavy (i.e. non-hydrogen) atoms. Third, the electronic structure properties were computed using more accurate electronic structure methods. Finally, the inclusion of platinum in addition to nickel as a catalyst adds new versatility to RMG-Cat. carbon bonding C Pt

H3C-R Pt

C

C

C

C

C

C

Pt

Pt

Pt

Pt

Pt

Pt Pt

carbon vdW

carbon single

carbon double

carbon triple

carbon quadruple

carbon bidentate

C

R C

R C

C

Pt

Pt

Pt

Pt Pt

H2C=R Pt

HC≣R Pt

H2C

R

Pt

HC

R

Pt

R

HC Pt

R

H2C

R

Pt Pt R→H

H3C-H Pt

R→O

H3C-OH Pt

R→N

Pt

Pt

CH

C

Pt

Pt

Pt

OH

Pt HC≣N Pt

H3C-CH3 H2C=CH2 HC≣CH Pt

CH2

H2C

Pt

H3C-NH2 H2C=NH Pt

R→C

H2C=O

CH3 Pt

Pt

H 2C

Pt

O Pt

NH2

Pt H2C

HC

CH3

OH HC Pt NH2

NH HC

HC

Pt

Pt CH3

CH2

O

HC

R

Pt Pt

R

R

C

H2C

R

Pt Pt R

Pt Pt

HC

R

Pt Pt

HC

R

Pt Pt

R

HC

Pt Pt

HC

R

Pt Pt

C

R

Pt Pt

OH

C

C

H2C O

Pt NH

Pt

Pt Pt

C

C

Pt CH2

Pt

HC

NH2 H2C

NH HC

Pt Pt

*1

Pt

CH3

HC

HC

C

C

Pt

Pt

Pt

Pt

H2C CH2 HC Pt Pt

O

Pt Pt

*2

N Pt

H2C

N

HC NH

Pt Pt

Pt Pt

*4 *3 CH H2C CH HC C

Pt Pt

Pt Pt

Pt Pt

HC

*1

N

Pt Pt

HC*3 CH2 HC*2 CH Pt Pt

Pt Pt

C

N

Pt Pt *4 C CH

C

Pt Pt

Pt Pt

C

Figure 1: Structure of the thermodynamic tree for an adsorbate binding through a carbon atom. Note that four adsorbates (those with an orange number) have two equivalent structures. RMG-Cat first looks to see what type of element is bonded to the platinum surface: hydrogen, carbon, nitrogen, oxygen, or some combination of two heavy atoms for bidentate 4


Page 5 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


nitrogen bonding

H2N-R Pt

oxygen bonding

N

O

Pt

Pt

N

N

N

N

N

R

O

O

O

O

Pt

Pt

Pt

Pt

Pt Pt

Pt

Pt

Pt

Pt Pt

nitrogen vdW

nitrogen single

nitrogen double

nitrogen triple

nitrogen bidentate

oxygen vdW

oxygen single

oxygen double

oxygen bidentate

R N

R N

O

Pt

Pt

Pt

Pt Pt

HN=R Pt

N≣R Pt

N

R N

R N

R N

N

Pt

Pt

Pt

Pt

Pt Pt

R

H

HN

R

Pt Pt

N

R

R

Pt Pt

N

HO-R

R

Pt Pt HN

R

Pt Pt

N

R

Pt Pt

N

R

R

R

Pt Pt

H R→H

H2N-H Pt

NH2

N

N

HO-H

OH

Pt

Pt

Pt

Pt

O

Pt OH

N

N

Pt NH

Pt

N

N

Pt

Pt

CH2

CH3

HN

N

N

O

Pt

Pt

Pt

Pt

OH R→O

H2N-OH Pt

HN=O

HN

Pt

Pt NH2

R→N

H2N-NH2 HN=NH Pt

Pt

N≣N Pt

HN Pt CH3

R→C

O

N O

HO-OH

O

Pt Pt

Pt Pt

Pt

Pt

HN

NH2 HN

NH

Pt Pt

*1 N N

HN N

Pt Pt

Pt Pt

N

*1

N

O

Pt Pt

Pt

OH

O

O O

Pt

Pt Pt

NH2

CH3

Figure 2: Structure of the thermodynamic tree for an adsorbate binding through either a nitrogen or an oxygen atom (bidentate duplicates with Figure 1 removed for clarity). species. The general structure of the thermodynamic tree is summarized for carbon in Figure 1. The top node is for an adsorbate bonded to platinum through a carbon atom. The next level down details the bond order; in addition to single, double, and triple bonds, the code also includes physisorbed carbon (“vdW”), a single carbon adatom (“quadruple”), and bidentate structures. The third layer considers the number and type of bonds to other atoms within the adsorbate. For example, a carbon atom with a double bond to the surface (approximately the middle column in the tree) can have either two single bonds to other ligands or one double bond. The wildcard R in the third row represents a possible bonding element, with current values of R = H, C, N, or O. The bottom four rows represent the specific species that can be formed with each of these four choices for the wildcard. Thus, beneath the “carbon double” branch, a total of seven specific adsorbates were computed and added to the database: *CH2 , *CHOH, *CHNH2 , *CHCH3 , *CO, *CNH, and *CCH2 . Note that several species that occupy two sites (“bidentate”) have two equivalent bond representations.

5



For example, the adsorbate *CH* N can be represented with two single surface bonds and a double bond between the carbon and nitrogen, or two double surface bonds and a single bond between the carbon and nitrogen; RMG-Cat correctly recognizes that both chemical graphs refer to the same species. The thermodynamic trees for adsorbates bonding through either nitrogen or oxygen are shown in Figure 2. Species that also appear in Figure 1 are removed for clarity. The thermodynamic tree for hydrogen is not shown, as it consists of only two species: *(H2 ) and *H. When RMG-Cat predicts a new species, it first looks to see if it can find this species in the database. If the species happens to be a specific match for one of the species in Figures 1 or 2, then it uses that species’ precompiled thermodynamic properties. If the species is not found, then RMG-Cat nonetheless can estimate that adsorbate’s properties. RMG-Cat first looks to see what the binding atom is, followed by the bond order. Next, it considers the H-atoms for the species in Figures 1 or 2 as wildcards, and it checks to see which of the species in the database is closest in structure to the new adsorbate. Each species in the database also includes an adsorption correction, which is the difference in enthalpy, entropy, and heat capacity between the adsorbate and its gas-phase precursor. The new adsorbate is removed from the surface, and the thermodynamic properties of the gas-phase precursor are obtained (either from an exact match in the gas-phase database or estimated via group additivity). The adsorption correction is then added to the gas-phase thermochemistry to obtain the thermochemistry of the adsorbate. 13 The current database now contains the vast majority of possible adsorbates that contain up to 2 heavy atoms, and so the possibility of a direct match is high for most applications that involve small (1-2 heavy atom) molecules that contain H/C/N/O on Pt. Moreover, by systematically populating the database, the accuracy of the adsorption corrections increases for larger molecules.

6


Page 6 of 32

Page 7 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Electronic Structure Theory Density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package (VASP), 27,28 interfaced through the Atomic Simulation Environment (ASE). 29 The BEEF-vdW functional was used for all calculations, 30 as it has been shown to describe both chemisorption energies and barriers on transition metal surfaces reasonably well. 31 The procedure used for computing the adsorbate geometries and energies is the same as in Ref. 32, so only a brief summary is provided here. The Pt(111) surface was represented by a 4-layer metal slab, which consists of a 3×3 unit cell and 17 Å vacuum between surfaces. The two uppermost metal layers were allowed to relax. A stepwise procedure was used to obtain the final energy. First, the initial geometry optimization was performed using planewave cutoff energy of 40 Rydberg and a (2 × 2 × 1) gamma-point centered Monkhorst-Pack kpoint grid. Next, the final geometry optimization used tighter convergence criteria of 50 Ryd and a (4 × 4 × 1) k-point grid. The final electronic energy was then obtained using 60 Ryd and a (6 × 6 × 1) k-point grid. The convergence criterion for geometry optimizations was set to a maximum force of 0.01 eV/Å. Gas-phase species were computed within a 10×10×10 Å box. Normal mode analysis was performed using a finite difference approximation. The resulting electronic energy, vibrational frequencies, and cartesian coordinates were used in the thermochemistry calculations. Based upon the system size, the adsorbates are at the 1/9th monolayer coverage. In the subsequent continuum-level calculations, the adsorbate properties are assumed to be in the low-coverage limit.

Thermochemistry A microkinetic mechanism requires more than just binding energies; it requires accurate parameterizations of the enthalpy and entropy of each adsorbate over a broad range of temperatures and pressures. The standard approach in the reactive flow community is to ◦ ◦ provide the enthalpy of formation, ∆f H298 , and entropy, S298 , at a standard state of 1 bar and

298.15 K and the temperature-dependent heat capacity in polynomial form. 33 Accordingly, 7



Page 8 of 32

one must first convert the electronic structure properties – i.e. electronic energy, vibrational frequencies, and rotational constants – into partition functions and from there derive the macroscopic thermodynamic properties. The molecular partition function for each adsorbate was computed using standard statistical methods 34 via in-house subroutines. The adsorbates are assumed to behave as harmonic oscillators. 17 species (those listed under the “vdW” column in Figures 1 and 2) are physisorbed on the surface. 14 of these 17 species have binding energies that are weaker than -100 kJ/mol and have two vibrational frequencies that are below 100 cm−1 . For these 14 species, the two harmonic modes that correspond to frustrated translation parallel to the surface were replaced with a 2D-gas partition function. The mass in the 2D-gas partition function is that of the gas-phase precursor, and the area over which the adsorbate can freely translate is the area of the unit cell, which is 6.9 Å2 . The method for computing the standard enthalpy of formation is analogous to the approach outlined for the compound method ANL0, 35 which relies on the Active Thermochemical Tables (ATcT) for reference values in a hypothetical reaction. 36,37 This approach begins with a fictitious reaction to form the gas-phase species Ca Ob Nc Hd (g) from four reference species: H2 (g), CH4 (g), NH3 (g), and H2 O(g), (for the elements H, C, N, and O, respectively):

aCH4 (g) + bH2 O(g) + cNH3 (g) +

3 d − 2a − b − c H2 (g) → Ca Ob Nc Hd (g) 2 2

(1)

The heat of reaction at 0 K for the working reaction in Equation (1) is computed using the electronic structure method of choice, in this case BEEF-vdW:

8


Page 9 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


C O Nc Hd (g)

a b ∆Hrxn,DFT (g) = EDFT

CH (g)

− aEDFT4

H O(g)

2 − bEDFT 3 d H2 (g) NH3 (g) − 2a − b − c EDFT − cEDFT − 2 2

(2)

X(g)

where EDFT is the zero-point corrected electronic energy of species X in the gas phase obtained from the BEEF-vdW calculations. The heat of formation for the gas-phase molecule is computed from the heats of formation for the reference species and the heat of reaction from Equation (2):

◦,C O Nc Hd (g)

a b ∆f H0,DFT

◦,CH (g)

◦,H O(g)

◦,NH (g)

4 2 3 = a∆f H0,ATcT + b∆f H0,ATcT + c∆f H0,ATcT d 3 ◦,H2 − 2a − b − c ∆f H0,ATcT + ∆Hrxn,DFT (g) + 2 2

(3)

◦ In the Supplemental Material, we compare the ∆f H0,DFT with reference values for the

same species in the Active Thermochemical Tables. In principle, we could go one step further and derive bond additivity corrections for the BEEF-vdW functional, 38 but such a process is beyond the scope of the present work. The reference values used in Equation (3) are: ◦ ◦ (H2 O) = -238.938 ± 0.027 kJ/mol, (CH4 ) = -66.556 ± 0.056 kJ/mol, ∆f H0,ATcT ∆f H0,ATcT ◦ ◦ (H2 ) = 0.0 ± 0.0. 37 (NH3 ) = -38.565 ± 0.030 kJ/mol, and ∆f H0,ATcT ∆f H0,ATcT

Next, the binding energy of the adsorbate, or equivalently the enthalpy of adsorption at 0 K, was computed from the difference in energy between the adsorbate and the gas-phase precursor and a vacant site: ∗

∗

C a O b Nc Hd ∆H0,ads = E Ca Ob Nc Hd − E Ca Ob Nc Hd (g) − E Pt

(4)

∗

where E Ca Ob Nc Hd , E Ca Ob Nc Hd (g) , and E Pt are the zero-point corrected electronic energies of the adsorbate Ca Ob Nc Hd on the platinum surface, the gas-phase precursor, and the vacant 9



Page 10 of 32

∗

Ca Ob Nc Hd is by definition a negative quantity. site, respectively. Note that ∆H0,ads

Finally, the requisite heat of formation of the adsorbate is the sum of the gas-phase heat of formation and the heat of adsorption:

∗

◦,C O Nc Hd (g)

◦,Ca Ob Nc Hd a b = ∆f H0,DFT ∆f H0,DFT

◦,Pt C a O b Nc Hd + ∆f H0,ref + ∆H0,ads

∗

(5)

where ∆H0,ads is the binding energy from Equation (4). Although platinum is not currently in the Active Thermochemical Tables, we assume that the heat of formation of bare platinum ◦ (Pt) = 0.0 ± 0.0. The heat of formation of the adsorbate at is by definition zero: ∆f H0,ref

298.15 K is then obtained by including the thermal correction going from 0 to 298.15 K for both the target adsorbate and the reference elements. For the gas-phase monatomic species H(g), N(g), and O(g), no DFT calculations were performed. Instead, the gas-phase energies for these species were obtained from direct dissociation of the elemental dimer, with the following heats of reaction taken from ATcT: H →2H

N →2N

O →2O

2 2 2 ∆Hrxn,ATcT (g) = 432.068 ± 0.000 kJ/mol; ∆Hrxn,ATcT (g) = 941.157 ± 0.047 kJ/mol; ∆Hrxn,ATcT (g)

= 493.688 ± 0.004 kJ/mol. 37 An important feature of this approach is that the final value for adsorbate heat of for∗

◦,Ca Ob Nc Hd mation, ∆f H0,DFT , is actually independent of the BEEF-vdW energy for the gas-phase

precursor, E Ca Ob Nc Hd (g) . Although the electronic energy of the gas-phase precursor is necessary for both the gas-phase heat of formation and for the binding energy, the contribution ∗

◦,Ca Ob Nc Hd of this term cancels out in the final determination of ∆f H0,DFT . This feature is im-

portant because GGA functionals may not perform as well for some gas-phase species. For instance, from inspection of the BEEF-vdW heats of formation at 0 K with the ATcT values (see Supplemental Material), it is clear that the GGA method performs worse for open-shell species than for closed-shell species. The end result of Equations (2-5) is that the heat of formation of each adsorbate is pegged to the heats of formation and heats of adsorption for

10


Page 11 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the same four reference species. A second significant advantage of this approach is that it allows RMG-Cat to utilize the vast library of previously tabulated gas-phase thermodynamic properties that is built into RMG. 35,38,39 These databases used post-Hartree Fock wavefunction methods with better than chemical accuracy, and they include key anharmonic effects, such has hindered internal rotation. Consequently, the enthalpy, entropy, and heat capacity of the gas-phase species in RMG-Cat are significantly more accurate than if we had used BEEF-vdW electronic structure properties and rigid-rotor harmonic-oscillator models. The heat capacity for each adsorbate Ca Ob Nc Hd ∗ is computed directly from the partition functions (either purely harmonic, or the sum of 2D gas and harmonic for the weakly bound species). A fundamental assumption in the Langmuir model is that the thermodynamic properties of the catalyst are unperturbed by the presence of the adsorbate. 34 Consequently, the heat capacity of the vacant site is set to zero, and only the 3N normal modes are included in the heat capacity, where N = a + b + c + d is the number of atoms in the adsorbate.

Catalytic Combustion RMG-Cat can now utilize all the gas-phase functionality that is intrinsic to RMG. The rate-rule approach and flux-based algorithm simultaneously follow the kinetics not just for adsorption, surface bond fission, and surface abstraction, but also for the 40+ gas-phase reaction families as well. 12 Importantly, RMG includes precompiled libraries of accurate thermochemistry and literature mechanisms. The user can choose from a list of nearly 100 kinetic libraries that have been taken from the literature or pre-calculated. For the present work, gas-phase chemistry was taken first from the H2 /O2 mechanism of Burke, 40 followed by the small-molecule hydrocarbon oxidation mechanism “Foundational Fuel Chemistry Model Version 1.0 (FFCM-1)” of Stanford University. 41 To demonstrate RMG-Cat’s capabilities, we consider catalytic combustion of methane on platinum. Catalytic combustion of alkanes has received renewed attention for microreac11



tors, due to the ability to operate outside the flammability limits. 42,43 However, under some operating conditions, the temperatures that can result from catalytic combustion are sufficiently high to induce light-off or ignition in the gas phase. 44,45 Gas-phase light-off presents an operational and safety hazard, and the ability to predict when it could occur is essential for process optimization. RMG-Cat was initialized with six different initial conditions. Three of the reactors used an initial composition of 5% CH4 , 19.95% O2 , and 75.05% N2 , and the remaining three had a composition of 5% CH4 , 19.45% O2 , 74.55% N2 , and 1.0% NO. Each composition was run for temperatures of 800, 1500, and 1750 K, and an initial pressure of 1 bar. The final mechanism was valid for all six conditions. The microkinetic mechanism of Deutschmann and coworkers was used to provide kinetic parameters for a few key reactions. 46 The remaining reactions were estimated using the same BEP parameters 47 for the reaction families and rate rules as outlined in Reference 13. Two separate mechanisms were generated. In one mechanism, the exploration of gas-phase chemistry was turned off, and only surface reactions were considered. The second mechanism includes both gas-phase and surface kinetics. The methane catalytic combustion process was simulated using a catalytic reactor model in Cantera, 48 an open source library of subroutines for reactive flow simulations. A plug flow reactor (PFR) was represented by a linear chain of stirred constant volume zerodimensional reactors, where each one is integrated to steady state. Pressure drop along the length of the reactor was not considered in this simulation. The reactor was set to have a surface-area-to-volume ratio appropriate for a monolith reactor, where the middle section is coated with a catalytically active Pt surface. The model parameters and operating conditions are listed in Table 1. A common feature in high-temperature catalytic oxidation is thermal conduction through the monolith, and this second-order process necessarily cannot be captured by a PFR. 49–51 Accordingly, we wish to emphasize that the purpose of this simulation is not to reproduce

12


Page 12 of 32

Page 13 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


any one particular experiment with high-fidelity, nor is it to provide an accurate model of heat transport within the reactor. Rather, our intention is to demonstrate RMG-Cat’s capability to generate detailed, thermodynamically-consistent microkinetic mechanisms for homogeneous/heterogeneously coupled systems.

Table 1: The model parameters and reactor operating conditions used in the simulations. Inlet temperature

800 K

Reactor Pressure

1 atm

Reactor length

3 cm

Cross section area

5.3 cm2

Catalyst area to volume ratio 545 m−1 Inlet gas velocity

1.9 cm/s

Monolith channel porosity

0.567

Inlet gas molar composition

4% CH4 , 16% O2 , 80% N2

Initial surface coverage guess

100% vacant

Results Electronic Structure and Thermochemistry The BEEF-vdW binding energies, as well as the standard heats of formation and entropy, for the 69 adsorbates are listed in Tables 2 – 5. The corresponding vibrational frequencies are provided the the Supplemental Material. An optimized structure could not be obtained for two species in Figures 1. Specifically, the physisorbed ethene, *(C2 H4 ), always relaxed to the bidentate structure *CH2* CH2 , and the adsorbed vinyl, *CHCH2 , also relaxed to the bidentate structure *CH* CH2 . Similarly, three species in Figure 2 proved problematic. No optimized geometry could be 13



Page 14 of 32

found for the bidentate nitrogen dimer, *N* N; instead, only the physisorbed structure could be obtained *(N2 ). A bidentate structure for nitrosyl hydride, *NH* O, could not be located. An optimized geometry was obtained for the corresponding physisorbed species, *(HNO); however, this structure has a stronger binding energy than most other physisorbed species, -122.5 kJ/mol, which suggests that the optimized geometry is perhaps better thought of as a chemisorbed species with a charge separation, *N+ HO – . Finally, the geometry minimization for the bidentate nitric oxide, *N* O, contained one imaginary mode; this mode was replaced with a duplicate of the first real frequency, 19.4i → 68.0 cm−1 , in the calculation of the vibrational partition function. Table 2: The structures, names, binding energies, heats of formation at 0 K and standard state entropies of species adsorbed on the Pt surface through a hydrogen atom. Binding sites are represented by * and species in parentheses are physisorbed.

Structure

Name

Eb (kJ mol−1 )

∆f H0◦ (kJ mol−1 )

◦ S298 (J mol−1 K−1 )

H

*H Pt H

-239.2

-23.2

4.3

-5.8

-5.3

97.9

H

*(H2 ) Pt

Table 3: The structures, names, binding energies, heats of formation at 0K and standard state entropies of species adsorbed on the Pt surface through a carbon atom, as illustrated in Figure 1. Binding sites are represented by * and species in parentheses are physisorbed.

Structure

Name

Eb (kJ mol−1 )

∆f H0◦ (kJ mol−1 )

◦ S298 (J mol−1 K−1 )

C

*C

-652.3

51.0

6.7

*CH

-602.1

-24.1

10.0

*COH

-411.0

-248.4

42.1

Pt CH Pt COH Pt

14


Page 15 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


CNH2 Pt CCH3

*CNH2

-392.7

-73.2

47.7

*CCH3

-539.4

-81.6

44.7

*CO

-142.8

-284.4

38.6

*CNH

-167.9

-11.3

59.5

*CCH2

-384.0

-12.6

27.7

*CH2

-351.2

5.2

19.8

*CHOH

-285.6

-206.3

59.1

*CHNH2

-257.6

-45.8

69.9

*CHCH3

-345.4

-30.5

51.5

*CHO

-213.2

-224.4

70.3

*CHNH

-214.2

11.1

69.1

*CH3

-170.8

-40.1

57.8

*CH2 OH

-182.4

-227.1

68.8

Pt O C Pt NH C Pt CH2 C Pt H

H C Pt OH

H C Pt

NH2

H C Pt

CH3

H C Pt H

O C Pt

H

NH C Pt

CH3 Pt H OH

H C Pt

15



Page 16 of 32

H NH2

H C

*CH2 NH2

-191.0

-63.5

80.0

*CH2 CH3

-168.9

-65.8

81.8

*CH2* O

-23.2

-165.4

51.0

*CH* O

-183.3

-194.2

32.6

-73.3

-5.8

28.8

*CH2* N

-165.0

28.5

32.9

*CH* NH

-240.3

-15.5

35.6

-62.7

36.5

36.5

-322.3

64.6

56.4

*CH2* CH2

-91.7

-52.9

45.7

*CH* CH2

-269.2

-15.2

29.3

Pt H CH3

H C Pt H H C

O

Pt

Pt

C

O

Pt

Pt

H

H

H

H C

N

Pt

Pt

C

N

Pt

Pt

*CH2* NH

H H

H

H C

N

Pt

Pt

C

N

Pt

Pt

H

C

N

Pt

Pt

*CH* N

*C * N H

H

H C

C

Pt

Pt

H

H

H C

C

Pt

Pt

H

16


Page 17 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


H

H C

C

Pt

*CH* CH

-193.9

7.7

47.7

*C* CH

-395.6

121.3

39.6

*C * C

-570.2

200.8

24.3

*(CH4 )

-11.6

-78.3

157.2

*(CH3 OH)

-30.9

-234.8

118.4

*(CH2 O)

-17.4

-160.3

137.1

*(CH3 NH2 )

-84.9

-100.8

99.9

*(CH2 NH)

-22.2

45.1

88.6

-1.0

98.3

90.2

*(CH3 CH3 )

-21.2

-96.4

164.9

*(CHCH)

-19.3

182.4

135.7

Pt H

C

C

Pt

Pt

C

C

Pt

Pt

H3C

H Pt

H3C

OH Pt

H2C

O Pt

H3C

NH2 Pt

H2C

HC

NH Pt N

*(HCN) Pt H3C

CH3

Pt HC CH Pt

Table 4: The structures, names, binding energies, heats of formation at 0K and standard state entropies of species adsorbed on the Pt surface through a nitrogen atom, as illustrated in the left half of Figure 2. Binding sites are represented by * and species in parentheses are physisorbed.

Structure

Name

Eb (kJ mol−1 )

∆f H0◦ (kJ mol−1 )

◦ S298 (J mol−1 K−1 )

N

*N

-419.9

Pt

17


41.0

9.8


Page 18 of 32

NH

*NH

-331.9

-5.2

11.0

*NOH

-314.5

-76.7

43.1

*NNH2

-196.8

59.5

43.3

*NCH3

-294.3

-27.8

42.6

*NO

-153.4

-106.0

35.7

*NNH

-102.3

95.5

42.6

*NCH2

-160.2

33.0

45.3

*NH2

-195.9

-28.9

20.4

*NHOH

-132.2

-80.9

43.5

*NHNH2

-122.5

66.2

47.4

*NHCH3

-178.5

-25.7

50.8

*NH* NH

-94.6

95.1

40.3

-123.5

74.0

32.6

Pt NOH Pt NNH2 Pt NCH3 Pt O N Pt NH N Pt CH2 N Pt NH2 Pt H

OH N Pt

H

NH2 N Pt

H

CH3 N Pt H

H N

N

Pt

Pt

N

N

Pt

Pt

H

*NH* N

18


Page 19 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


N

O

Pt

Pt

-134.1

-87.2

83.0

*(NH3 )

-64.6

-103.5

58.6

*(NH2 OH)

-62.7

-105.7

103.0

-122.5

-55.0

62.60

*(NH2 NH2 )

-94.6

1.4

97.3

*(NHNH)

-65.6

106.5

42.3

*(N2 )

-10.6

-29.9

127.9

*N * O H2N

H Pt

H2N

OH

Pt HN O

*(HNO) Pt H2N NH2 Pt NH

HN

Pt N

N

Pt

Table 5: The structures, names, binding energies, heats of formation at 0K and standard state entropies of species adsorbed on the Pt surface through an oxygen atom, as illustrated in the right half of Figure 2. Binding sites are represented by * and species in parentheses are physisorbed.

Structure

Name

Eb (kJ mol−1 )

∆f H0◦ (kJ mol−1 )

◦ S298 (J mol−1 K−1 )

O

*O

-346.0

-138.5

13.9

*OH

-190.1

-152.3

36.7

*OOH

-71.4

-121.3

77.2

*ONH2

-67.5

-55.6

85.9

*OCH3

-132.2

-151.4

67.9

-33.8

-110.2

36.0

Pt OH Pt OOH Pt ONH2 Pt OCH3 Pt O

O

Pt

Pt

*O * O

19



HO

Page 20 of 32

H

*(H2 O)

-18.3

-257.2

94.6

*(OHOH)

-28.0

-190.4

117.9

Pt HO

OH Pt

Mechanism Generation and Catalytic Combustion The final mechanism generated by RMG-Cat for the catalytic combustion of methane on platinum includes 21 adsorbates and 55 surface reactions with 38 gas-phase intermediates and 340 gas-phase reactions. The model “edge” contained an additional 196 species and 298 reactions that were considered and rejected. The entire mechanism generation process took less than 15 minutes on a single core of a 2018-era laptop with a 1.46 GHz processor. The results of the PFR simulations are presented in Figure 3. The solid lines are computed using the RMG-Cat mechanism that includes both surface and gas-phase reactions, whereas the dashed lines are from the mechanism that only contained surface chemistry. It is clear that the contribution of gas-phase chemistry to the total fuel conversion is negligible during the 10 mm section that includes the catalyst (i.e., the solid and dashed lines are identical for z ≤ 20 mm). However, once the catalyst zone ends, the concentration of fuel and oxidizer is still sufficiently high that thermal ignition occurs. The gases continue to react for another ∼ 5 mm before igniting. Within the catalyst zone, the surface is predominantly covered by oxygen adatoms, *O. The effect of gas-phase chemistry can be seen from the heat release rates, Figure 4. The heat release due to surface chemistry is predominantly at the end of the catalyst zone; although still sharp, it is broader than the contribution from the gas-phase. This sharp heat release rate at 25 mm is effectively a standing flame within the reactor. Such a temperature spike likely would pose a severe operational hazard. As the temperature increases along the length of the catalytic reactor, the change in the 20


Page 21 of 32

0.200

catalyst

O2

b)

catalyst

1600

0.150 0.125 0.100

H2 O

0.075 0.050

CO2

CH4

Temperature (K)

Gas-phase mole fractions

1800

a)

0.175

1400 1200 1000 800

0.025

CO

0.000 0

5

10

15

20

Distance (mm)

25

surface + gas reactions surface reactions only

600 30

0

5

10

15

20

Distance (mm)

25

30

Figure 3: PFR results for catalytic combustion of natural gas over Pt: (a) the major gasphase species, and (b) the temperature. The catalyst is located in the region between 10–20 mm from the inlet. The solid lines include both surface and gas-phase reactions; the dashed lines do not include gas-phase chemistry. 52

70

a)

catalyst

Contribution to total heat release (%)

2.5 2.0

Heat release (W)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1.5

gas reactions surface reactions

1.0 0.5 0.0 0

5

10

15

20

Distance (mm)

25

30

b)

catalyst

60 50 40

gas reactions surface reactions

30 20 10 0

0

5

10

15

20

Distance (mm)

25

30

Figure 4: Heat release rates for catalytic combustion in a PFR: (a) net rate of heat release, (b) contribution of each reactive phase towards the total heat release. 52

21



equilibrium constant allows for radical desorption to become more competitive. By the end of the reactor, the rate of desorption of hydroxyl radicals is within two orders of magnitude of the rate of oxygen associative desorption, thereby seeding the product gases with small concentrations of OH(g). However, the standing flame in Figure 4 is due predominantly to purely thermal chemistry in the gas phase, not radical desorption. Finally, we tested the ability of RMG-Cat to discover the relevant reactions for nitrogen chemistry, particularly for NOx . Although RMG-Cat now includes nitrogen chemistry, the rate of NOx formation in the simulations was negligible; this result is likely due to the fact that the peak temperature is still below the thermal NOx limit. 53 Indeed, RMG-Cat found three reactions that are critical for NOx formation: N + O2 * ) NO + O, HCN + OH * ) CN + H2 O, and CN + O2 * ) NO + CO. However, the flux through these three channels was insufficiently large to move these reactions from the edge to the core. Additionally, the reaction CH + N2 * ) HCN + N was not found in the edge, because the species CH itself was never present in the core. Additional simulations were performed with 0.1% NO in the inlet feed stream. RMGCat correctly found the surface reactions leading to the formation and desorption of NO2 , and it included these reactions in the core. More importantly, because of the strong sensitizing effects of NO on CH4 oxidation, 54,55 the concentration profiles were different: the methane is completely consumed within catalyst zone, through a combined process of gasphase activation via NO and surface oxidation. The same temperature spike that is present in Figure 3 was observed, but the location of the spike was shifted forward by ∼ 6 mm. These finding confirm that RMG-Cat now handles nitrogen-containing adsorbates. Future work will be done to expand and improve the accuracy of nitrogen chemistry in RMG and RMG-Cat.

22


Page 22 of 32

Page 23 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusions Two new features were added to the open-source software RMG-Cat. First, BEEF-vdW calculations were performed for 69 adsorbates on Pt(111). These calculations were used to create a new, more richly structured thermodynamics database. This database now includes nitrogen-containing species for the first time. Second, RMG-Cat can now simultaneously include gas-phase chemistry in the mechanism expansion process. These features are tested on the catalytic combustion of methane over a platinum monolith. The monolith is simulated as a plug flow reactor. The simulations confirm that high surface temperatures can lead to gasphase ignition. With this new functionality, RMG-Cat is ideally suited to provide accurate, thermodynamically consistent microkinetic mechanisms for homogeneous/heterogeneously coupled systems.

Acknowledgement The authors gratefully acknowledge support from the U.S. Department of Energy, through the Computational Chemical Sciences program within the Basic Energy Science division, with Dr. Mark Pedersen as the program manager. JJ and FS acknowledge support from the state of Baden-Wüttemberg through bwHPC (bwunicluster and JUSTUS, RV bw17D011).

Supporting Information Available • Filename: chem.cti, the microkinetic mechanism, in Cantera format • Filename: Supplemental_Material.pdf, tables of the vibrational frequencies for each species, as well as a table of heats of formation for gas phase species This material is available free of charge via the Internet at http://pubs.acs.org/.

23



Bio

Franklin Goldsmith was born in Asheville, North Carolina. He received a B.A. in Philosophy from the University of North Carolina, followed by a B.S. in both Chemical Engineering and Applied Mathematics from North Carolina State University, and then was a Fulbright Scholar in Mathematics in Freiburg, Germany. He obtained his Ph.D. in Chemical Engineering from the Massachusetts Institute of Technology. Goldsmith spent two years in Berlin as an Alexander von Humboldt Scholar in Inorganic Chemistry at the Fritz-Haber Institute

24


Page 24 of 32

Page 25 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


of the Max Planck Society, followed by one year as an Argonne Director’s Fellow in Theoretical Chemistry at Argonne National Laboratory. His research focuses on theoretical, computational, and experimental quantification of reaction rate constants, with a particular focus on combustion chemistry, heterogeneous catalysis, energetic materials, propellants, and atmospheric chemistry. He joined the School of Engineering at Brown University in 2014.

References (1) Bhoorasingh, P. L.; Slakman, B. L.; Seyedzadeh Khanshan, F.; Cain, J. Y.; West, R. H. Automated Transition State Theory Calculations for High-Throughput Kinetics. The Journal of Physical Chemistry A 2017, 121, 6896–6904. (2) Jafari, M.; Zimmerman, P. M. Uncovering reaction sequences on surfaces through graphical methods. Physical Chemistry Chemical Physics 2018, 20, 7721–7729. (3) Ulissi, Z. W.; Medford, A. J.; Bligaard, T.; Nørskov, J. K. To address surface reaction network complexity using scaling relations machine learning and DFT calculations. Nature Communications 2017, 8 . (4) Medford, A. J.; Kunz, M. R.; Ewing, S. M.; Borders, T.; Fushimi, R. Extracting Knowledge from Data through Catalysis Informatics. ACS Catalysis 2018, 8, 7403–7429. (5) Goldsmith, B. R.; Esterhuizen, J.; Liu, J.-X.; Bartel, C. J.; Sutton, C. Machine learning for heterogeneous catalyst design and discovery. AIChE Journal 2018, 64, 2311–2323. (6) Tran, K.; Palizhati, A.; Back, S.; Ulissi, Z. W. Dynamic Workflows for Routine Materials Discovery in Surface Science. Journal of Chemical Information and Modeling 2018, 58, 2392–2400. (7) Broadbelt, L. J.; Pfaendtner, J. Lexicography of kinetic modeling of complex reaction networks. AIChE Journal 2005, 51, 2112–2121. 25



(8) Battin-Leclerc, F.; Blurock, E.; Bounaceur, R.; Fournet, R.; Glaude, P.-A.; Herbinet, O.; Sirjean, B.; Warth, V. Towards cleaner combustion engines through groundbreaking detailed chemical kinetic models. Chemical Society reviews 2011, 40, 4762–82. (9) Vandewiele, N. M.; Geem, K. M. V.; Reyniers, M.-F.; Marin, G. B. Genesys: Kinetic model construction using chemo-informatics. Chemical Engineering Journal 2012, 207208, 526 – 538. (10) Rangarajan, S.; Kaminski, T.; Wyk, E. V.; Bhan, A.; Daoutidis, P. Language-oriented rule-based reaction network generation and analysis: Algorithms of RING. Computers & Chemical Engineering 2014, 64, 124 – 137. (11) Van de Vijver, R.; Vandewiele, N. M.; Bhoorasingh, P. L.; Slakman, B. L.; Seyedzadeh Khanshan, F.; Carstensen, H.-H.; Reyniers, M.-F.; Marin, G. B.; West, R. H.; Van Geem, K. M. Automatic Mechanism and Kinetic Model Generation for Gas- and Solution-Phase Processes: A Perspective on Best Practices, Recent Advances, and Future Challenges. International Journal of Chemical Kinetics 2015, 47, 199–231. (12) Gao, C. W.; Allen, J. W.; Green, W. H.; West, R. H. Reaction Mechanism Generator: automatic construction of chemical kinetic mechanisms. Computer Physics Communications 2016, 203, 212–225. (13) Goldsmith, C. F.; West, R. Automatic Generation of Microkinetic Mechanisms in Heterogeneous Catalysis. Journal of Physical Chemistry C 2017, 121, 9970–9981. (14) Margraf, J. T.; Reuter, K. Systematic Enumeration of Elementary Reaction Steps in Surface Catalysis. ACS Omega 2019, 4, 3370–3379. (15) Gossler, H.; Maier, L.; Angeli, S.; Tischer, S.; Deutschmann, O. CaRMeN: a tool for analysing and deriving kinetics in the real world. Physical Chemistry Chemical Physics 2018, 20, 10857–10876. 26


Page 26 of 32

Page 27 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(16) Susnow, R. G.; Dean, A. M.; Green, W. H.; Peczak, P.; Broadbelt, L. J. Rate-based construction of kinetic models for complex systems. Journal of Physical Chemistry A 1997, 101, 3731–3740. (17) Magoon, G. R.; Aguilera-Iparraguirre, J.; Green, W. H.; Lutz, J. J.; Piecuch, P.; Wong, H. W.; Oluwole, O. O. Detailed chemical kinetic modeling of JP-10 (exotetrahydrodicyclopentadiene) high-temperature oxidation: Exploring the role of biradical species in initial decomposition steps. International Journal of Chemical Kinetics 2012, 44, 179–193. (18) Green, W. H.; West, R. H. Reaction Mechanism Generator. https://rmg.mit.edu/. (19) Hickman, D. A.; Schmidt, L. D. Production of Syngas by Direct Catalytic Oxidation of Methane. Science 1993, 259, 343–346. (20) Korup, O.; Goldsmith, C. F.; Weinberg, G.; Geske, M.; Kandemir, T.; Schlogl, R.; Horn, R. Catalytic partial oxidation of methane on platinum investigated by spatial reactor profiles, spatially resolved spectroscopy, and microkinetic modeling. Journal of Catalysis 2013, 297, 1–16. (21) Zavyalova, U.; Holena, M.; Schl ogl, R.; Baerns, M. Statistical Analysis of Past Catalytic Data on Oxidative Methane Coupling for New Insights into the Composition of HighPerformance Catalysts. ChemCatChem 2011, 3, 1935–1947. (22) Karakaya, C.; Kee, R. J. Progress in the direct catalytic conversion of methane to fuels and chemicals. Progress in Energy and Combustion Science 2016, 55, 60 – 97. (23) Guo, X.; Fang, G.; Li, G.; Ma, H.; Fan, H.; Yu, L.; Ma, C.; Wu, X.; Deng, D.; Wei, M.; Tan, D.; Si, R.; Zhang, S.; Li, J.; Sun, L.; Tang, Z.; Pan, X.; Bao, X. Direct, Nonoxidative Conversion of Methane to Ethylene, Aromatics, and Hydrogen. Science 2014, 344, 616–619.

27



(24) Imbihl, R.; Scheibe, A.; Zeng, Y. F.; GÃĳnther, S.; Kraehnert, R.; Kondratenko, V. A.; Baerns, M.; Offermans, W. K.; Jansen, A. P. J.; van Santen, R. A. Catalytic ammonia oxidation on platinum: mechanism and catalyst restructuring at high and low pressure. Physical Chemistry Chemical Physics 2007, 9, 3522–3540. (25) Haynes, B. S. Combustion research for chemical processing. Proceedings of the Combustion Institute 2019, 37, 1 – 32. (26) Blaylock, D. W.; Ogura, T.; Green, W. H.; Beran, G. J. O. Computational Investigation of Thermochemistry and Kinetics of Steam Methane Reforming on Ni(111) under Realistic Conditions. The Journal of Physical Chemistry C 2009, 113, 4898–4908. (27) Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 1996, 54, 11169–11186. (28) Kresse, G.; Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Computational Materials Science 1996, 6, 15 – 50. (29) Jacobsen, S. R. B.; K, W. An object-oriented scripting interface to a legacy electronic structure code. Comput. Sci. Eng. 2002, 4, 56–66. (30) Wellendorff, J.; Lundgaard, K. T.; Mogelhoj, A.; Petzold, V.; Landis, D. D.; Nørskov, J. K.; Bligaard, T.; Jacobsen, K. W. Density functionals for surface science: Exchange-correlation model development with Bayesian error estimation. Physical Review B 2012, 85 . (31) Mallikarjun Sharada, S.; Bligaard, T.; Luntz, A. C.; Kroes, G.-J.; Nørskov, J. K. SBH10: A Benchmark Database of Barrier Heights on Transition Metal Surfaces. The Journal of Physical Chemistry C 2017, 121, 19807–19815.

28


Page 28 of 32

Page 29 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(32) Wellendorff, J.; Silbaugh, T. L.; Pintos, D. G.; Nørskov, J. K.; Bligaard, T.; Studt, F.; Campbell, C. T. A benchmark database for adsorption bond energies to transition metal surfaces and comparison to selected DFT functionals. Surface Science 2015, (33) Kee, R. J.; Coltrin, M. E.; Glarborg, P. Chemically Reacting Flow: Theory and Practice, 1st ed.; John Wiley & Sons: Hoboken, New Jersey, 2003. (34) Hill, T. L. An Introduction to Statistical Thermodynamics; Dover: New York, 1960. (35) Klippenstein, S. J.; Harding, L. B.; Ruscic, B. Ab initio Computations and Active Thermochemical Tables Hand in Hand: Heats of Formation of Core Combustion Species. The Journal of Physical Chemistry A 2017, 121, 6580–6602. (36) Ruscic, B.; Pinzon, R. E.; Morton, M. L.; von Laszevski, G.; Bittner, S. J.; Nijsure, S. G.; Amin, K. A.; Minkoff, M.; Wagner, A. F. Introduction to Active Thermochemical Tables: Several “Key” Enthalpies of Formation Revisited. The Journal of Physical Chemistry A 2004, 108, 9979–9997. (37) Ruscic, B.; Bross, D. H. Active Thermochemical Tables (ATcT) values based on ver. 1.122d of the Thermochemical Network. https://atct.anl.gov/. (38) Goldsmith, C. F.; Magoon, G. R.; Green, W. H. Database of Small Molecule Thermochemistry for Combustion. Journal of Physical Chemistry A 2012, 116, 9033–9057. (39) Hashemi, H.; Christensen, J. M.; Gersen, S.; Levinsky, H.; Klippenstein, S. J.; Glarborg, P. High-pressure oxidation of methane. Combustion and Flame 2016, 172, 349 – 364. (40) Burke, M. P.; Chaos, M.; Ju, Y.; Dryer, F. L.; Klippenstein, S. J. Comprehensive H2/O2 kinetic model for high-pressure combustion. International Journal of Chemical Kinetics 2012, 44, 444–474.

29



(41) Smith, G.; Tao, Y.; Wang, H. Foundational Fuel Chemistry Model Version 1.0 (FFCM1). 2016; http://nanoenergy.stanford.edu/ffcm1, Version 1.0. (42) Maruta, K., Takeda, K., Ahn, J., Borer, K., Sitzki, L., Ronney, P. D., Deutschmann, O., Extinction limits of catalytic combustion in microchannels. Proceedings of the Combustion Institute 2002, 29, 957–963. (43) Kaisare, N. S.; Vlachos, D. G. A review on microcombustion: Fundamentals, devices and applications. Progress in Energy and Combustion Science 2012, 38, 321 – 359. (44) Veser, G.; Schmidt, L. D. Ignition and extinction in the catalytic oxidation of hydrocarbons over platinum. AIChE Journal 1996, 42, 1077–1087. (45) Reinke, M., Mantzaras, J., Bombach, R., Schenker, S., Inauen, A., Gas phase chemistry in catalytic combustion of methane/air mixtures over platinum at pressures of 1 to 16 bar. Combustion and Flame 2005, 141, 448–468. (46) Koop, J.; Deutschmann, O. Detailed surface reaction mechanism for Pt-catalyzed abatement of automotive exhaust gases. Applied Catalysis B: Environmental 2009, 91, 47 – 58. (47) Wang, S.; Temel, B.; Shen, J.; Jones, G.; Grabow, L. C.; Studt, F.; Bligaard, T.; AbildPedersen, F.; Christensen, C. H.; Nørskov, J. K. Universal Brønsted-Evans-Polanyi Relations for C-C, C-O, C-N, N-O, N-N, and O-O Dissociation Reactions. Catalysis Letters 2011, 141, 370–373. (48) Goodwin, D. G.; Moffat, H. K.; Speth, R. L. Cantera: An Object-oriented Software Toolkit for Chemical Kinetics, Thermodynamics, and Transport Processes. 2017; http: //www.cantera.org, Version 2.3.0. (49) Deutschmann, O., Behrendt, F., Warnatz. J., Formal treatment of catalytic combustion and catalytic conversion of methane. Catalysis Today 1998, 46, 155–163. 30


Page 30 of 32

Page 31 of 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(50) Schwiedernoch, R., Tischer, S., Deutschmann, O., Warnaz, J., Experimental and numerical investigation of the ignition of methane combustion on a platinum-coated honeycomb monolith. Proceedings of the Combustion Institute 2002, 29, 1005–1011. (51) Schwarz, H.; Dong, Y.; Horn, R. Catalytic Methane Combustion on a Pt Gauze: LaserInduced Fluorescence Spectroscopy, Species Profiles, and Simulations. Chemical Engineering & Technology 2016, 39, 2011–2019. (52) Blondal, K.; West, R. H.; Goldsmith, C. F. Data, scripts, and figures for Computergenerated kinetics for coupled heterogeneous/homogeneous systems: a case study in catalytic combustion of methane on platinum. figshare. Dataset. 2019; https://doi. org/10.6084/m9.figshare.8226593. (53) Glarborg, P.; Miller, J. A.; Ruscic, B.; Klippenstein, S. J. Modeling nitrogen chemistry in combustion. Prog. Energy Combust. Sci. 2018, 67, 31–68. (54) Dagaut, P.; Nicolle, A. Experimental study and detailed kinetic modeling of the effect of exhaust gas on fuel combustion: mutual sensitization of the oxidation of nitric oxide and methane over extended temperature and pressure ranges. Combustion and Flame 2005, 140, 161 – 171. (55) Sivaramakrishnan, R.; Brezinsky, K.; Dayma, G.; Dagaut, P. High pressure effects on the mutual sensitization of the oxidation of NO and CH4 -C2 H6 blends. Physical Chemistry Chemical Physics 2007, 9, 4230–44.

31



Graphical TOC Entry 0.200

1800

a)

0.175

catalyst

O2

b)

catalyst

1600

0.150 0.125 0.100

H2 O

0.075 0.050

CO2

CH4

Temperature (K)

Gas-phase mole fractions

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

CO 0

5

10

15

20

Distance (mm)

25

1200 1000 800

0.025 0.000

1400

surface + gas reactions surface reactions only

600 30

32

0

5

10

15

20

Distance (mm)


25

30

Page 32 of 32

Computer-generated kinetics for coupled heterogeneous

Recommend Documents