Numerical Modeling of Hydroperoxyl-Mediated Oxidative

Subscriber access provided by Kaohsiung Medical University

Kinetics, Catalysis, and Reaction Engineering

Numerical Modeling of Hydroperoxyl–Mediated Oxidative Dehydrogenation of Formic Acid under SCR–Relevant Conditions Manasa Sridhar, John Mantzaras, Jeroen Anton van Bokhoven, and Oliver Kröcher Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.8b01592 • Publication Date (Web): 09 Jul 2018 Downloaded from http://pubs.acs.org on July 12, 2018

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 27 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

Numerical Modeling of Hydroperoxyl–Mediated Oxidative Dehydrogenation of Formic Acid under SCR–Relevant Conditions Manasa Sridhara,b, John Mantzarasa*, Jeroen Anton van Bokhovena,b, Oliver Kröchera,c

a

Paul Scherrer Institut, 5232 Villigen, Switzerland

b c

ETH Zurich, Institute for Chemical and Bioengineering, 8093 Zurich, Switzerland

École Polytechnique Fédérale de Lausanne (EPFL), Institute of Chemical Sciences and

Engineering, 1015 Lausanne, Switzerland

Revised paper: ie-2018-01592y

*Corresponding author: John Mantzaras Email: [email protected] Tel.: +41 56 310 4046

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

Abstract Formic acid decomposition to carbon dioxide is an important step during the decomposition of alternative formate–based ammonia precursor compounds, such as ammonium formate, guanidinium formate and methanamide in the selective catalytic reduction (SCR) process. At oxygen and water concentrations prevalent in a typical SCR feed, an oxidative dehydrogenation (ODH) type pathway of formic acid decomposition proceeds on titania-supported gold catalysts. Using the surface perfectly stirred reactor (SPSR) model, the ODH mechanism is demonstrated to satisfactorily predict the experimentally observed conversions. The single–site mechanistic model captured the negative trend in conversion with increasing formic acid concentrations, stemming from an extensive coverage of the active sites by formates. This in turn rendered the active sites unavailable for the formation of the active hydroperoxyls required for the rate determining step of C-H bond scission of formates. The positive order dependency on oxygen concentration and the promotional effect of water are quantitatively and semi–qualitatively described by the model. Predicted trends in the relative surface coverages of different reaction intermediates are in agreement with previously reported kinetic and spectroscopic measurements. These results are important fundamentally for the understanding of formic acid decomposition chemistry and general catalysis by gold, as well as for the design of dedicated hydrolysis catalysts for applications in the SCR process.

Keywords: Oxidative dehydrogenation of formic acid, titania-supported gold catalyst, hydroperoxy species, surface perfectly stirred reactor, sensitivity analysis

2 ACS Paragon Plus Environment

Page 2 of 27

Page 3 of 27 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. Introduction Formate-based ammonia precursors are promising alternatives to urea in SCR applications owing to their favorable properties, namely, lower freezing point, higher ammonia storage capacity and higher thermal stability and durability [1–7]. Considering that many of these precursors release formic acid upon thermolysis in the hot diesel engine exhaust [8,9], their successful application in the SCR process requires rapid and selective decomposition of formic acid to carbon dioxide to prevent side–product formation and suppress formic acid emission. In this context, titania–supported gold catalysts show excellent activity and selectivity to decompose formic acid to carbon dioxide under the highly oxidizing conditions relevant to SCR [10]. Importantly, the catalysts did not catalyze ammonia oxidation, which is highly desirable since the latter is required for the SCR process. Under the investigated conditions in [10], which are similar to those prevalent in the engine exhaust, formic acid decomposes to carbon dioxide and carbon monoxide. The addition of lanthana to the catalyst formulation resulted in the selective enhancement in the carbon dioxide production rates, while carbon monoxide formation was suppressed [11]. It is known that, instead of direct formic acid oxidation to carbon dioxide, formic acid could alternatively first decompose to carbon monoxide as an intermediate that subsequently oxidizes to carbon dioxide [12,13]. Further experiments revealed that carbon monoxide oxidation was irrelevant for the formic acid decomposition mechanism on the unmodified and lanthana–modified gold catalysts and that the lanthana–induced rate enhancement was caused only by acceleration of the direct oxidative decomposition pathway to water and carbon dioxide [11]. Kinetic and mechanistic investigations revealed that carbon dioxide and carbon monoxide formation from formic acid decomposition proceed as two independent reactions occurring on two different active sites: the former associated with gold and the latter exclusively with the support. Excess oxygen and water, present in the exhaust feed, influenced the kinetics of the two reactions differently. On the one hand, oxygen was mandatory for carbon dioxide production on gold. A synergy between oxygen and water led to higher carbon dioxide production rates, similar to the promotional effect of water [14–16] in oxygen activation during carbon monoxide and alcohol oxidation on gold. On the other hand, carbon monoxide formation occurring on titania was unaffected by the presence or absence of oxygen. Water suppressed the carbon monoxide formation rates as opposed to the promotional effect on carbon dioxide production.



Page 4 of 27

Based on the in–depth kinetic and in situ DRIFTS analyses, a mechanism was proposed in which the formation of hydroperoxy species via a proton shift equilibrium between adsorbed oxygen and water bypassed the energetically intensive step of oxygen activation on gold [17– 24]. The rate-determining step involved the C–H bond scission of the abundantly present formate species with the aid of hydroperoxy species to form carbon dioxide and water. The proposed ODH mechanism induces reaction orders that are consistent with the experimentally

observed

conversions.

This

mechanism,

describing

an

‘oxidative

dehydrogenation’ (ODH) type pathway for formic acid conversion to carbon dioxide and water under highly oxidizing and wet conditions simulating the exhaust feed, is markedly different from the stoichiometric formic acid decomposition or formate dehydrogenation to carbon dioxide and hydrogen that are often studied in the context of hydrogen generation in fuel cells and the water gas shift reaction [25–28]. As a next step, it is essential to numerically model this mechanism to test its applicability in predicting the observed catalytic performance. The surface perfectly stirred reactor (SPSR) is a widely applied model to successfully simulate complex reaction networks involved in homogeneous fuel combustion as well as in catalytic emission control [29–39]. For instance, Chaniotis and Poulikakos used an SPSR to validate the kinetic model describing the catalytic partial oxidation of methane in a fuel cell [35]. Mantzaras and coworkers performed sensitivity analysis in an SPSR to investigate the influence of surface chemistry on the oxidation of carbon monoxide in carbon monoxide/hydrogen/air mixtures [31] and on hetero-/homogeneous combustion of methane/air [36] and ethane/air mixtures [33] over platinum, all under fuel-lean stoichiometries. In the SPSR, the contents of the reactor are assumed to be nearly spatially uniform owing to high diffusion rates or forced turbulent mixing. Thus, the rate of conversion of reactants to products is considered to be controlled by chemical reaction rates and not by mixing processes. The crucial element of the SPSR is its underlying assumption that sufficient mixing allows it to be described well by spatially averaged or bulk properties. The main advantage of such an approximation lies in the relatively small computational demand of the mathematical model. Further information of the SPSR model can be found in the reference [40]. In this work, the validity of the proposed ODH mechanism is tested by comparing SPSR model predictions with the experiments on gold supported on lanthana–modified titania. An estimate of the relative coverages of the different reaction intermediates is provided and sensitivity analyses are performed to determine the dominant elementary steps controlling the conversion of reactants. 4 ACS Paragon Plus Environment

Page 5 of 27 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


2. Experimental Catalyst preparation, testing and kinetic measurements. The preparation and testing of the titania-supported gold catalyst modified with lanthana can be found elsewhere [11,41,42]. In brief, titania anatase modified with 10 wt % lanthana (DT57, Crystal Global) was impregnated with a 0.032 M solution of HAuCl4 (Sigma–Aldrich). Aging and washing with aqueous ammonia solution and distilled water, in that order, was followed by calcination at 400 °C for 5 h in air. The base–modified catalyst was designated as Au/La–TiO2. The catalyst was coated on cordierite monoliths at a desired washcoat loading of ~2.5 g L–1 and tested in a laboratory reactor dedicated to investigating the interaction of catalyst-coated monoliths with sprayed liquid solutions [41]. Gaseous products at the reactor outlet were quantified by FTIR spectroscopy (Antaris, Thermo Nicolet) [42]. The total flow rate was set at 750 L h–1 and the contact time was varied by changing the number of monoliths stacked along the length of the reactor, while keeping constant the total catalyst concentration per unit volume. Concentrations of formic acid, oxygen, and water in the feed were varied in the range between 227 and 1070 ppm (see Table 1). External and internal mass transfer limitations were ruled out [11] which, together with the isothermal operation, facilitated the applicability of the SPSR model. The SPSR tool was indispensable due to the large number (~5,000) of optimization simulations. The formulae used for the calculation of conversion, product selectivity and yield

are described in our previous work [11]. The C–balance was closed within an accuracy of ±3% by summing up carbon monoxide, carbon dioxide and formic acid. Temperature programmed oxidation and time-on-stream experiments revealed that no carbon deposition occurred on the catalysts. Table 1. Experimental conditionsa Case

Formic acid

Oxygen

(mol/mol)

(mol/mol)

Water (mol/mol)

Temperature (°C)

I

0.00065

0.1

0.05

160–300

II III IV

0.000227–0.00107 0.00065 0.00065

0.1 0.0025–0.03 0.1

0.05 0.05 0.005–0.013

200, 300 200, 300 200, 300



a

Page 6 of 27

The feed was diluted with nitrogen to achieve a total flow of 750 L h–1, contact time (W/F)

= 2.4·10–5 g s cm–3. Mechanistic model. The proposed mechanism for the ODH of formic acid under SCR-relevant conditions encompasses 16 reactions out of which 14 represent the forward and backward direction of seven fast and quasi–equilibrated elementary reactions [43]. The remaining two steps involve the slow and irreversible steps of hydrogen abstraction from the adsorbed formate with the aid of activated oxygen species. Table 2 lists the 16 surface reactions and values for the Arrhenius pre–exponentials, temperature exponents and the activation energies. HCOO*, HOO*, H2O*, O2*, O*, HO* and H* describe the surface intermediates adsorbed on the active site ‘*’. Coverage–dependent activation energies (COV) were introduced for steps 9 and 10. This mechanistic scheme does not take into account the formic acid decomposition to carbon monoxide on titania and to carbon dioxide in the gas phase, because their contributions are very low [11]. Methodologies for microkinetic surface mechanism development with thermodynamic (enthalpic and entropic) consistency have been put forth (Mhadeshwar et al. [44]). Nonetheless, such a rigorous approach was not followed herein, as it is best suited when basic knowledge on kinetic parameters is available (e.g. estimates of activation energies via DFT calculations, estimates of pre-exponentials via transition state theory-TST, experimental input via TPD etc.). Thermodynamic consistency is less important in the present application. For example, enthalpic inconsistency leads to incorrect solutions of the energy equation and this is important in non-isothermal simulations. However, the present model is isothermal (as also a practical SCR system is) such that no energy equation is solved. Furthermore, conditions are away from equilibrium since the residence times in the simulations are quite short (~12 ms). Numerical model. The numerical platform for the kinetic studies is the steady-state SPSR model [40]. The salient features of this model include perfect mixing and spatially uniform temperatures and species compositions. In this zero–dimensional model, the inlet species undergo instantaneous homogeneous mixing such that only kinetic limitations are present [40]. Simulations are carried out at a fixed SPSR temperature (mimicking the present isothermal reactor experiments), thus negating solution of the energy equation. The gas-phase species governing equations for a steady state SPSR in the presence of catalytic and gas phase reactions are: , − , = + , k = 1, 2, .. Kg, 6 ACS Paragon Plus Environment

(Eq. 1)

Page 7 of 27 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


where is the mass flow rate in the reactor in g/s, Yk,IN and Yk,OUT the mass fractions of the k–th gaseous species at the inlet and outlet, respectively, the volumetric molar production rate of k–th species via gaseous reactions in mol·cm–3·s–1, the surface molar production rate of k–th species via catalytic reactions in mol·cm–2·s–1, S and V the SPSR surface area (cm2) and volume (cm3), respectively, the molecular weight of the k–th species in g·mol-1, and Kg the total number of gas phase species. In the present case Kg = 5, accounting for formic acid, oxygen, water, carbon dioxide and nitrogen. The gaseous reaction rates in Eq. 1 have quite small contributions at the present moderate temperatures and are thus neglected. The known variables are , Yk,IN, S, and , where are calculated from the forthcoming Eq. 3. Table 2. Sequence of steps describing ODH of formic acid on Au/La–TiO2 and their corresponding kinetic parametersa

1

H2O + * → H2O*

A Eact b (kJ mol–1) (mol·cm·s) 1.20E+18 0.0 52.0

2 3 4 5 6

H2O* → H2O + * O2 + * → O2* O2* → O2 + * H2O* + O2* → HOO* + HO* HOO* + HO* → H2O* + O2*

8.50E+14 1.04E+17 1.00E+15 1.60E+21 1.20E+22

0.0 0.0 0.0 0.0 0.0

110.0 60.0 120.0 50.0 50.0

7

HCOOH + HO* → HCOO* + H2O

1.42E+26

0.0

38.0

8

HCOO* + H2O → HCOOH + HO*

5.25E+18

0.0

30.0

9

HCOO* + HOO* → CO2 + H2O + * + O* 1.70E+13

0.0

32.3+3.65*θHCOO*

10

HCOO* + O* → CO2 + HO*+ *

2.50E+20

0.0

25.8+1.46*θHCOO*

11 12

2HO* → H2O* + O* H2O* + O* → 2HO*

2.00E+20 2.00E+20

0.0 0.0

40.0 60.0

13

H2O* + * → HO* + H*

1.55E+22

0.0

40.0

14

HO*+ H* → H2O* + *

2.40E+21

0.0

30.0

15

H* + O* ⇄ HO* + *

2.30E+22

0.0

40.0

16

HO* +* → H* + O*

2.0E+22

0.0

40.0

Step Elementary steps

a

The rate constant, k, is given by = · ·

!"

, with b = 0. θHCOO* is the surface

coverage of adsorbed formate. The kinetic parameters of steps 1-4 have been adjusted, based on values from references [45–47]. The coverage equations for the surface species are given by: 7 ACS Paragon Plus Environment


%&' %(

= )

*' +

, m = 1, 2,… Ks.

Page 8 of 27

(Eq. 2)

The left–side in Eq. 2 is not a true transient term and is only introduced to facilitate numerical convergence to steady state; therein, limited time integrations may be performed to bring the solution within the convergence domain of the employed Newton solver [40]. In Eq. 2, ), is the site occupancy (number of sites occupied by the m–th surface species), Γ the surface site density, θ, the surface coverage of the m–th species and Ks the total number of surface species. In the present case Ks = 8, which accounts for HCOO*, HO2*, HOO*, O2*, O*, HO*, H* and * (the latter denotes the available free sites). The value of ), is unity since each surface species occupies only one site. Since the value of Γ for gold is not known, the corresponding value for platinum [48] (equal to 2.7×10-9 mol cm-2) is employed. The surface reaction rates are given by: Ns

s&m = ∑ν ml kfl l =1

Kg + Ks

∏ j =1

v´jl

Cj

,

(Eq. 3)

with Ns the number of surface reactions and ν,ℓ the stoichiometric coefficient of species m in surface reaction λ. For the gas phase species ( j = 1, 2, .. Kg) the concentrations in Eq. 3 are ./ = ρ/ // (mol/cm3), while for the surface species ( j = Kg+1, Kg+2, .. Kg+Ks) the corresponding surface concentrations are ./ = Γ 1/ /)/ (mol/cm2). The reaction rate coefficients 2ℓ in Eq. 3 are: kf l = Al T

βl

Ks

µ  ε θm   −E  exp  l  ∏ θi ml exp  ml  ,  RT  m =1  RT 

(Eq. 4)

with ℓ, βℓ and 3ℓ the pre–exponential, temperature exponent, and activation energy of surface reaction λ, respectively. The parameters ε,ℓ and µ,ℓ describe coverage dependencies of the reaction rate coefficient, which account for variation in the adsorption binding states due to changing surface coverage [49]. In the present mechanism µ,ℓ = 0 for all surface reactions, while ε,ℓ is non–zero only for reactions 9 and 10 for the surface formate HCOO*. Equations 1 and 2 constitute a system of non-linear algebraic equations, solved by a modified Newton method for the unknown mass fractions Yk,OUT of gas-phase species and surface coverages θ, [40]. The computed Yk,OUT in turn yield the % species conversions, 8 ACS Paragon Plus Environment

Page 9 of 27 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


Kk = 100(Yk,IN - Yk,OUT )/Yk,IN. In Newton’s method, a sequence of iterations is followed, each successively approaching the true solution. For the sake of brevity, the approximate solution vector at a given iteration is called 4 56. When 4 56 is inserted into the governing equations, the equations are satisfied within a residual vector 76 . The objective then is to find a solution 4 56 such that 876 94 56:8 ≤ ;, with a herein selected ε = 10-20. A more detailed overview of this approach can be found in reference [40]. The sensitivity coefficients (.?,@ℓ >?,@ℓ

D=ℓ

,

(Eq. 5)

where ℓ is the pre–exponential of the λ-th reaction, Eℓ the change in the pre–exponential (always 2% increase of ℓ ), F

Numerical Modeling of Hydroperoxyl-Mediated Oxidative

Recommend Documents