On Reaction Pathways and Intermediates During Catalytic Olefin

at temperatures higher than 600 K. Although the underlying elementary reactions ..... is why Scheme 2 shows the corresponding equilibrium constants...
0 downloads 0 Views 1MB Size
Subscriber access provided by Nottingham Trent University

Kinetics, Catalysis, and Reaction Engineering

On Reaction Pathways and Intermediates During Catalytic Olefin Cracking over ZSM-5 Sebastian Standl, Tobias Kühlewind, Markus Tonigold, and Olaf Hinrichsen Ind. Eng. Chem. Res., Just Accepted Manuscript • Publication Date (Web): 04 Sep 2019 Downloaded from pubs.acs.org on September 4, 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 50 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

On Reaction Pathways and Intermediates During Catalytic Olen Cracking over ZSM-5 †,‡

Sebastian Standl,

Tobias Kühlewind,

Hinrichsen

†,‡



Markus Tonigold,

and Olaf

∗,†,‡

†Department of Chemistry, Technical University of Munich, Lichtenbergstraÿe 4, 85748 Garching near Munich, Germany

‡Catalysis Research Center, Technical University of Munich, Ernst-Otto-Fischer-Straÿe 1, 85748 Garching near Munich, Germany

¶Clariant Produkte (Deutschland) GmbH, Waldheimer Straÿe 13, 83052 Bruckmühl, Germany E-mail: [email protected] Phone: +49 (89) 289 13232. Fax: +49 (89) 289 13513

Abstract The concept of catalytic olen cracking is an alternative method to produce ethene and propene. Undesired higher olens formed on site are converted using acid zeolites at temperatures higher than 600 K. Although the underlying elementary reactions can be explained by interconversion steps of carbenium ions, relatively little is known about the exact procedure of adsorption and the nature of intermediates. However, detailed knowledge about these topics is indispensable for a comprehensive theoretical description. In this work, a microkinetic single-event model for olen cracking over ZSM-5 is analyzed in terms of reaction pathways and intermediates. An evaluation of adsorption states underlines the importance of dierentiating between physisorption 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

Page 2 of 50

and π -complex formation because the latter leads to signicantly higher accuracy when describing olen cracking. A further investigation of protonation predicts the resulting intermediates to be of comparably low stability. Therefore, their total concentration is negligible, a conclusion that should nevertheless not be used for all approaches from literature. Finally, protonation enthalpies are estimated; the resulting values suggest carbenium ions as intermediates at least for tertiary species, which is in line with the stability order of the obtained activation energies. These ndings can help to understand the interaction between olens and acid zeolites, a topic of high importance in constructing exact and physically consistent theoretical descriptions.

1

Introduction

The catalytic cracking of higher olens over acid zeolites is an alternative approach to produce ethene and propene,

1

which are important building blocks for the chemical industry.

Compared to steam cracking, temperatures and CO2 emissions are lower.

3,4

2

Additionally, the

use of a catalyst allows for product adjustment and higher exibility in feed. Fluid catalytic cracking (FCC), though it is the other established industrial route towards lower olens, is designed for gasoline production.

5

In times of rising environmental concerns,

ernmental rules and higher demand, especially for propene, importance.

7

6

stricter gov-

alternative processes gain in

1

The conversion of higher olens to ethene and propene is a concept to exploit the undesired byproducts of rening processes directly on site. Although no broad commercialization has been achieved yet, several project and demo plants have been realized.

1

When using

a suitable catalyst like ZSM-5, the crucial advantage of this concept is the high propene to ethene ratio that is achieved by adjusting reaction conditions and reactor setups.

8,9

In

an earlier study, it was shown that the industrial olen cracking process, which consists of a recycle reactor and a complete separation unit, is viable, especially at higher propene prices.

10

2

ACS Paragon Plus Environment

Page 3 of 50 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

However, more exibility in operational mode and feed will be necessary in the future. Robust and reliable models are thus inevitable in order to have a realistic depiction of the overall reactivity. In literature, several lumped models exist; of these can be found in a recent review.

17

1116

an overview and comparison

Using these models, a useful description of product

distribution is obtained within their respective experimental range, but although some of the approaches underly a mechanistic scheme, their application area is limited because of the missing microkinetic character.

18,19

As a consequence, the estimated parameters are

of articial character without clear physical background; they might contain for example adsorption enthalpies or active site concentrations.

This impedes a transfer to dierent

catalysts or reaction setups and therefore optimization of both process and catalyst properties in which an extrapolation out of experimentally covered regimes is required.

18,20

In contrast to this, a microkinetic model oers vast extrapolation possibilities because each elementary reaction taking place on the catalytic surface is explicitly included. such a model, the optimization of both process and catalyst is possible. yields insight into reaction intermediates and preferred pathways.

21

20

18

Using

Additionally, it

On the other hand, reac-

tion networks are huge and complex when each step is considered, especially for hydrocarbon conversion on zeolites.

Usually, a high number of reactions causes a high number of esti-

mated parameters and therefore much computational eort as well as challenging numerics. For that reason, the single-event theory was developed almost 30 years ago by the Ghent group.

18,22

It allows for the description of thousands of elementary reactions with only a

handful of dierent parameters; for details, refer to the literature. Von Aretin et al.

24

18,19,23

created a microkinetic model based on the single-event concept in order

to describe the cracking of 1-pentene over ZSM-5. Here, only ve fundamental parameters, i.e. four activation energies and one pre-exponential factor, were sucient to describe 1585 pathways of kinetic relevance. In a subsequent study,

25

the microkinetic character of this

model was proven: although the parameter estimation relied on only 1-pentene as feed, the model correctly predicted all olens from propene to heptene as reactants. Furthermore, the

3

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

Page 4 of 50

microkinetic description could be used to describe olen mixtures as feed and even catalysts with dierent properties. In microkinetics, it is extremely important that each implemented step has a reasonable physical background; only then, the transferability to dierent systems is guaranteed. The single-event model

24,25

is used here to gain a deeper understanding of the steps ongoing

during olen cracking with special focus on adsorption. In order to elaborate a consistent description that has broad validity, concepts and correlations from literature for the adsorption of hydrocarbons on zeolites are applied.

These are combined with own ttings and

assessed in terms of both accuracy and signicance. From this, general conclusions on the prevailing intermediates are drawn. Thus, for the rst time, a deep insight into the interplay of microkinetics and adsorption during olen cracking is provided, the results of which are compared with previously published work of both experimental

26

and theoretical

27,28

nature.

These coherences are helpful also for other processes such as methanol-to-olens (MTO) or hydrocracking.

2

Theoretical Methods

2.1

Reaction Network

In the original study, scribed in literature.

24

the reaction network was mainly derived through observations de-

2931

A summary of the underlying assumptions as well as a short ex-

planation of the dierent types of elementary reactions can be found in Section 1 in the Supporting Information. The same reaction network is used here, but with slight adaptations that improve the tting procedure.

When using the original network, some of the

implemented adsorption models cause insignicant values for one of the estimated cracking activation energies. For that reason, a renement of the reaction network was performed in a preliminary study.

Indeed, a version showing improved agreement and parameter signicance can be

4

ACS Paragon Plus Environment

Page 5 of 50 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

found. On the other hand, the dierences of the tting results are too small to be statistically signicant. This is why the whole preliminary study analyzing six dierent reaction networks RN01RN06 is shown in Section 2 of the Supporting Information. Only the dierences between RN01 (original study be mentioned here.

24

) and RN06 (this study) should

An important assumption of the original work

formation of primary intermediates.

24

was the irreversible

Examples exist in which ethene especially is not a

nal product, since it can be a reactant in dimerization steps.

11,14,15,32

This is why ethene

formation steps are assumed to be reversible in RN06; in addition, the cracking to primary intermediates is only allowed when the step leads to ethene. Scheme 1 represents the main characteristics of the two reaction networks.

Scheme 1: Dierences between the two reaction networks RN01 and RN06 with respect to the role of primary intermediates.

Due to the restriction of the formation of primary intermediates, the number of cracking reactions is lower in RN06, see Table 1. Furthermore, the reversibility of ethene formation leads to an equal amount of cracking and dimerization reactions. At this point, it should be underlined that the network in the original work was not wrong and is not corrected here; in contrast, the focus of this study is dierent. Whereas a kinetic model as it was derived in the previous contribution

24

should have broad validity

which makes an inclusion of a variety of reactions advantageous, the detailed analysis of intermediates requires that only pathways playing a crucial role at the respective conditions are included.

Although the single-event model is of microkinetic character, the reaction

5

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

Page 6 of 50

Table 1: Type of elementary reactions being implemented in the single-event kinetic model for olen cracking, including the amount of dierent reactions and of kinetically relevant steps; furthermore, the number of dierent olens and protonated intermediates is shown. type

RN01

RN06

olen protonation

956

957

cracking

601

238

pathways cracking

981

511

dimerization

140

238

pathways dimerization PCP branching methyl shift

293

511

1530

1530

148

148

1004

957

olens

591

591

protonated intermediates

498

451

olen deprotonation

network has thus to be adapted before an adsorption study is possible.

2.2

Adsorption States

For an analysis of adsorption models, it is important to dierentiate between the characteristics of hydrocarbon interaction with zeolites. whether an acid site is available or not;

33

26

For the adsorption of alkanes, it is irrelevant

the contribution of these Brønsted centers is neg-

ligible. The interaction is almost completely restricted to non-directed van der Waals more precisely, to dispersion forces between the alkane and the pore wall. be measured using dierent methods.

35,36

26

34

or,

This eect can

In contrast to this, the occurrence of a double

bond within olens increases the complexity of the adsorption process. When no acid site is present, the interaction is again restricted to the zeolite wall and thus to physisorption,

26

see

Scheme 2 on the right side. It should be noted that at regular olen cracking conditions, temperatures are suciently high so that quasi-equilibrated adsorption is achieved, is why Scheme 2 shows the corresponding equilibrium constants. ysisorption of olens is comparable to that of alkanes.

37,38

24,37

which

Quantitatively, the ph-

However, the left side of Scheme 2

shows the crucial dierence when a free acid site is available. Here, the directed interaction between the double bond and the acid site leads to the formation of a

6

ACS Paragon Plus Environment

π -complex,

which

Page 7 of 50 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

causes a higher degree of stabilization compared to pure physisorption. erage distance between olen and zeolite wall might be higher in a forces still contribute.

26,34

Although the av-

π -complex, the dispersion

34

Scheme 2: Two dierent main pathways of adsorption: chemisorption on an acid site (left side,

π -complex

plus protonation) and physisorption without contribution of an acid site

(right side).

In a subsequent step, the proton of the acid site is transferred to the olen. It is still under debate whether this protonation leads to a carbenium ion

19,39

or to an alkoxide

34,40,41

as an intermediate. For a long time, even simple examples of carbenium ions could not be identied via NMR spectroscopy; aromatic species.

43,44

42

evidence for their existence was only found for cyclic or

Quite recently, the identication of the

ZSM-5 was successful by combining measurement and theory. theoretical evidence on MOR

46

and FER.

47

tert -butyl 45

carbenium ion on

This corresponds to earlier

However, this example is seen to be one of the

most stable carbenium ions because their stability decreases in the order tertiary > secondary > primary.

46

Theoretical studies suggest that their existence as intermediates is possible, but

only favored when the positive charge is delocalized

7

48

or not easily accessible;

ACS Paragon Plus Environment

49

otherwise,

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

Page 8 of 50

the corresponding alkoxide should evolve. This is why in most of the calculations, carbenium ions cannot be found as minima in the potential energy surface, thereby excluding them as possible intermediates. Instead, they are assumed as short-living transition states here.

40,49,50

On the other hand, carbenium ions might be favored at high temperatures due to the gain of entropy compared to an alkoxide.

28,47

This matches experimental results with carbenium

ions being absent during isomerization reactions of 1-butene over ZSM-5 at they can be detected over FER when temperatures are higher than

230 K, 51

whereas

290 K. 52

Alkoxides lead to typical minima found on potential energy surfaces and are therefore frequently seen as intermediates in theoretical studies.

47

However, steric constraints arise

here due to the close distance between the hydrocarbon and zeolite wall.

This can lead

to a destabilization of especially tertiary alkoxides. Although similar stability for all types of alkoxides is observed in some work, to carbenium ions, i.e. > tertiary. TON

50,55

41

most publications suggest an ordering opposite

the alkoxide stability decreases in the order primary > secondary

This holds for dierent zeolites like CHA,

or MFI.

58

5355

FER,

47,56

MOR,

46,55

FAU,

57

It has to be underlined that these theoretical results highly depend

on the applied methodology, the cluster model and a realistic description of the acid site including its surroundings.

27,50,59

For example, all alkoxide types show the same stability

when a small and exible cluster is used, but they reveal signicant dierences for a more realistic description of the surroundings of the acid site. observed as having a much higher stability than a

49

For the same reason, alkoxides are

π -complex in some studies, 34,56,57

whereas

values for the protonation enthalpy, which are close to or even above zero, are found in other work.

27,46,47,50,55,60

This contradiction is even seen within the same study when dierent

cluster models are used. another.

50

50

Because of this, no conclusions can be drawn from one zeolite to

Another reason for discrepancies in literature is the use of pure density functionals

or free clusters versus hybrid approaches.

56

The adsorption properties of olens on acid zeolites are not easily accessible through experiments due to their high reactivity.

34

Therefore, the theoretical insights discussed above

8

ACS Paragon Plus Environment

Page 9 of 50 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

are required to understand the underlying chemistry.

However, the results of quantum

chemical calculations depend on several factors and, in most cases, depict one specic case that might not represent a catalyst applied at industrial conditions.

Here, microkinetic

models can be a step linking realistic conditions with theoretical calculations, which is why the single-event model for 1-pentene cracking

24

should be further evaluated. Besides their

industrial relevance for selective propene production, they oer two main pathways:

8

9

pentenes are interesting reactants, as

monomolecular cracking and dimerization with subsequent

cracking. In literature, two recent studies tackling pentene adsorption on ZSM-5 are available. Schallmoser et al.

26

found experimental evidence for fast

brated double-bond isomerization at a low temperature of

π -complex 323 K.

evolution and equili-

Moreover, they observed

dimerization steps taking place with a lower rate. They compared the energetics of pentene interaction over silicalite as well as over FER: Whereas only physisorption took place in the former example due to the lack of acid sites, evolution of the

π -complex

could be observed

in the latter case without subsequent dimerization reactions, which were suppressed due to steric constraints. By comparing the resulting enthalpy values as well as the measured overall reaction enthaly of dimerization, they determined the protonation enthalpy to an alkoxide to be

−41 kJ mol−1 .

The other recent work is of theoretical nature and authored by Hajek et al.

27

They

applied static calculations, molecular dynamics and metadynamic methods to investigate pentene adsorption.

Their main nding is that

π -complex

and alkoxide showed almost

equal stability with the exception of 3-pentoxide being the least stable species because of steric repulsions.

Positive protonation enthalpies were obtained for all types of alkoxides

in the static calculations, but this was rather caused by an overestimation of

π -complex

stability. The inclusion of nite temperature eects yielded a more realistic description of the adsorption process; here, protonation enthalpies in the range of were found.

−20 kJ mol−1

to

10 kJ mol−1

The application of the metadynamic simulations showed that the formation

9

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

of an alkoxide out of the

π -complex

Page 10 of 50

is an activated step with carbenium ions as transition

states. However, all these results were restricted to a temperature of study,

28

the similar methodology was extended to temperatures of

323 K.

573 K

the static approach was found to be inappropriate, as it predicted the

In a subsequent

and

773 K.

π -complex

Again,

to be the

most stable intermediate for all examples by far, although tertiary carbenium ions were found to be favored compared to tertiary alkoxides and might occur as intermediates at high temperatures. Because the mobility of the chemisorbed species and therefore entropy was underestimated with the static approach, molecular dynamics had to be applied. Here, the

π -complex

was still detected as a stable intermediate, but with increasing temperatures,

a higher amount of olens was stabilized only via van der Waals forces, thus loosing the xation at the double bond. For linear species, both alkoxides and carbenium ions were not stable at typical cracking conditions: The former quickly transformed into the corresponding carbenium ion as a metastable intermediate that underwent rapid deprotonation into the

π-

complex then. Branched compounds were observed to be much more mobile than their linear counterparts, both in a neutral and in protonated state. Tertiary carbenium ions were found to be a dominating intermediate at high temperatures, whereas alkoxides might only exist in primary form for branched species.

2.3

Mathematical Description of Adsorption

A robust mathematical formulation of adsorption phenomena is necessary for implementation into a model.

The thermodynamic equilibrium constant for

protonation, physisorption and chemisorption of compound

K

ads

i

= π,

prot, phys or chem.

10

ACS Paragon Plus Environment

formation,

is dened in Eq. 1,

  ∆ads Hi◦ − T ∆ads Si◦ , (i) = exp − RT

with ads

π -complex

(1)

Page 11 of 50 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

It should be explicitly stated that within this work, physisorption solely means the interaction between an olen and the zeolite wall without the contribution of an acid site. This is strictly dierentiated from the

π -complex

that can only evolve at Brønsted acid sites, leading to an

additional interaction between the double bond of the olen and the proton. Protonation means the transformation of a

π -complex

to the protonated intermediate, the latter being

either a carbenium ion or an alkoxide. Finally, chemisorption describes the whole step from a gas phase olen towards the protonated intermediate. For simplicity, the change in enthalpy or entropy due to adsorption is denoted as "adsorption enthalpy" or "adsorption entropy" in this work.

As mentioned above, these adsorption steps are usually assumed as quasi-

equilibrated at typical cracking conditions. A common approach to obtain the concentration of adsorbed intermediates is the use of a Langmuir isotherm according to Eq. 2 ,

C ads (i) =

The Langmuir adsorption constant temperature.

Cmax KLads (i) p (i) . 1 + KLads (i) p (i)

KLads (i)

The relative coverage

θi

Whereas

Cmax

(2)

and therefore the concentration depend on the

of species

maximum concentration of adsorption sites

θi =

33

Cmax ,

i

is obtained by dividing Eq. 2 by the

see Eq. 3:

KLads (i) p (i) . 1 + KLads (i) p (i)

equals the concentration of acid sites

nation, the saturation concentration of compound

i

Ct

(3)

for the

π -complex

on the zeolite surface

and the proto-

C sat (i)

has to be

used for physisorption. The coherence between the Langmuir adsorption constant and the thermodynamic equilibrium constant of adsorption is shown in Eq. 4,

K ads (i) = KLads (i) p◦ ,

11

ACS Paragon Plus Environment

(4)

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

where

p◦

should be chosen as

105 Pa

according to IUPAC.

61

Page 12 of 50

For the protonation, the ther-

modynamic equilibrium constant equals the Langmuir constant as the transfer of a proton is part of the whole chemisorption step where the standard pressure is already considered during calculation of the

π -complex.

Both for adsorption enthalpy and for entropy, linear correlations with the carbon number

CNi

as variable can be found,

The values of

α

to

δ

33,34,62

see Eqs. 5 and 6,

∆ads Hi◦ = α CNi + β,

(5)

∆ads Si◦ = γ CNi + δ.

(6)

are usually tted to experimental

33,62

or theoretical

34

adsorption

data. Depending on whether olens or alkanes are investigated, Eqs. 5 and 6 describe the formation of a

π -complex

provided by Nguyen et al. al.

62

or the physisorption as rst adsorption step. The former case is

34

whereas both the studies by de Moor et al.

33

and Denayer et

characterize the physisorption of alkanes. In all three literature concepts, both enthalpy

and entropy are assumed to be temperature independent within the respective investigated range. The inuence of temperature is nevertheless considered when calculating the thermodynamic equilibrium constant. The contribution of Nguyen et al.

34

additionally contains

correlations in form of Eqs. 5 and 6 to calculate the formation of secondary alkoxides out of a

π -complex;

these protonation values contain a minor thermal correction. From the same

authors, a conference paper exists

58

comprising an alternative set of correlations; with these,

the formation of not only secondary, but also of tertiary alkoxides can be calculated. Both approaches

34,58

dierentiate between 1-alkenes and 2-/3-/4-alkenes as well as the special case

of isobutene adsorption.

63

More details about all three literature adsorption studies

are explained in Section 3 in the Supporting Information.

12

ACS Paragon Plus Environment

33,34,62

Page 13 of 50 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

2.4

Mathematical Formulation of Overall Reactivity

2.4.1 Von Aretin et al. 24 - Use of Nguyen Correlations In the original cracking model by von Aretin et al.,

24

Eq. 7 was used for describing the

cracking reaction rate of olen Oi from an intermediate of type

rcr (m; n) = k cr (m; n) C

+ Ri



the single-event rate coecient



to type

n,

Ct KLπ (Oi ) p (Oi ) P . 1 + j KLπ (Oj ) p (Oj )

= ne k˜cr (m; n) K prot (Oπi ; m)

The coherence between regular rate coecient

m

(7)

k and the number of single events ne as well as

is not discussed here and can be found in literature.

18,20,23

The structure of the rate equation for dimerization was found to be similar to Eq. 7, but multiplied by the partial pressure of the second reactant olen.

24

It follows that no adsorption

eects were considered to dene this second reactant. In contrast, the regular correlations by Nguyen et al.

34

were used to describe both

π -complex and alkoxide formation of the reactant

olen in Eq. 7. For the special case of isobutene protonation, the values leading to a tertiary alkoxide were applied.

63

An underlying assumption of Eq. 7 was the negligible concentration of protonated intermediates on the acid sites. In the work shown here, an alternative approach according to Eq. 8 is tested where this concentration is explicitly included,

rcr (m; n) = ne k˜cr (m; n) K prot (Oπi ; m)

1+

P

j

Ct KLπ (Oi ) p (Oi ) P . KLπ (Oj ) p (Oj ) + j KLchem (Oj ) p (Oj ) (8)

It should be noted that the product of model by von Aretin et al.,

24

KLπ (Oi )

and

K prot (Oπi ; m)

the protonation constant

equals

K prot (Oπi ; m)

KLchem (Oi ).

consisted of symmetry

contributions, an isomerization constant to a well-dened reference olen protonation constant.

18

In the

23

and the actual

In this work, reference olens are applied only in these cases in which

a carbon number dependence of protonation is not guaranteed.

13

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

Page 14 of 50

2.4.2 Martens et al. 64 - Use of Denayer Correlations Whereas both the evolution of the

π -complex as well as chemisorption take place on the same

acid sites, two dierent Langmuir approaches are necessary for the equations typically found in hydrocracking:

23,39,64

one for the saturation eect on the catalyst surface due to physisorp-

tion and another one for the balance of acid sites. The latter, however, is often neglected because of a low concentration of protonated intermediates. increased by implementing this additional balance.

39

23

Nevertheless, accuracy can be

After adapting the equation to olen

cracking by leaving out all hydrogenation/dehydrogenation steps, the approach neglecting the concentration of protonated intermediates can be found in Eq. 9,

rcr (m; n) = ne k˜cr (m; n) Ct K prot



phys

Oi

64

 C sat (i) K phys (O ) p (O ) i i L ;m . P phys 1 + j KL (Oj ) p (Oj )

(9)

It follows that besides proper physisorption values, this approach requires data for which can be calculated using Eq. 10,

64

C sat (i) =

with

V po

pound

i

V po , Vml (i)

as the total pore volume of the catalyst and

in liquid state. In this work,

C sat (i),

−1 2 × 10−4 m3 kgcat

(10)

Vml (i)

as the molar volume of com-

is chosen as the specic pore volume

of ZSM-5, whereas the molar volumes in liquid state can be calculated Hankinson-Brobst-Thomson (HBT) method

66

65

using either the

or the approach by Elbro et al.

mer uses empiric equations to express the coherence between

Vml (i)

67

The for-

and the critical volume,

reduced temperature as well as acentric factor; the equations and constants are shown elsewhere.

65

In contrast, the concept by Elbro et al.

organic groups are tabulated;

65

67

is a group contribution method. Dierent

for each group, three values are available that allow for a

consideration of temperature eects. For the protonation entropy, Martens et al.

14

ACS Paragon Plus Environment

64

suggested

Page 15 of 50 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

a calculation routine according to Eq. 11,

∆prot Si◦ = −Si◦,trans − ∆phys Si◦ .

The total translational entropy ing to Eq. 12,

Si◦,trans

(11)

is accessible via the Sackur-Tetrode equation accord-

68

Si◦,trans = kB NA

(

with the Boltzmann constant molecular mass

M (i)

[2 π M (i) kB T ]3/2 h3

ln

kB ,

!

the Planck constant

of compound

i

 + ln

h,

Vmg (i) NA



5 + 2

)

the Avogadro number

and its molar volume in the gas phase

using the adsorption values of Denayer et al.,

36,62

,

(12)

NA ,

Vmg (i).

the

When

no direct correlation for the physisorption

entropy exists. The approach in Eq. S6 in the Supporting Information, which is applied here, is slightly dierent from the one originally proposed by Martens et al.

64

where no standard

pressure is considered. In contrast, a direct possibility to calculate the physisorption entropy exists for the adsorption values by de Moor et al.

33

via Eq. 6.

2.4.3 Toch et al.: 69 Simplied Protonation Approach Toch et al.

69

suggested a simplied concept to calculate the changes in entropy. According

to them, physisorption leads to the loss of one degree of freedom of the translational entropy, see Eq. 13,

∆phys Si◦ =

−Si◦,trans . 3

This should resemble the xation of the hydrocarbon in to move in

x-

and

y -direction.

(13)

z -direction,

whereas it is still able

In this approach, electric contributions to the entropy are

negligible whereas rotatoric and vibrational entropies are assumed to remain constant. After the protonation step, the species is bound to the surface and should therefore lose the

15

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

Page 16 of 50

remaining two degrees of freedom of translational entropy according to Eq. 14,

∆prot Si◦ =

−2 Si◦,trans . 3

(14)

This concept allows for the use of physisorption enthalpy values both by de Moor et al. and by Denayer et al.

2.5

33

62

Parameter Estimation

All models in this work are tted to the same experimental data set.

24

For this, MATLAB

version R2018a is used. The objective function is dened by Eq. 15,

SSQ =

NExp NRes XX

yj ,k

being the experimental and

and each experimental data point mole fractions.

lsqnonlin

k.

yˆj ,k

(15)

j

k

with

(yj ,k − yˆj ,k )2 ,

the modeled mole fraction of each tting response

j

Nitrogen as diluent is explicitly considered within the

The unweighted sum of squared residuals

SSQ

is minimized by applying

with the Levenberg-Marquardt algorithm except for the cases in which constraints

are required. For these scenarios, an initial tting using the trust-region-reective algorithm is performed, whereas the Levenberg-Marquardt algorithm is applied in a second run using the previous results as initial values. optimality measure and to

Tolerances are set to

T olX = 10−8

dierential equations, the solver

ode15s

T olF un = 10−8

in terms of step size.

in terms of

In order to integrate the

is used which has proven to be advantageous for

partly sti systems. The integration tolerances are set to

AbsT ol = 10−11 and RelT ol = 10−8

for absolute and relative tolerances, respectively. The whole set of equations can be found in earlier contributions.

24,25

16

ACS Paragon Plus Environment

Page 17 of 50 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

3

Results and Discussion

3.1

Reaction Network

A detailed description of the performance of all reaction networks can be found in Section 2 in the Supporting Information.

Here, only the results of RN06 as reference case for the

adsorption analysis should be discussed as well as RN01 for comparison, see Table 2. Table 2: Estimated activation energies

Eacr

and pre-exponential factors

Acr ,

including 95%

condence intervals, and the sum of squared residuals SSQ; all activation energies are given −1 −1 in kJ mol , whereas the pre-exponential factor is shown in s . parameter

Eacr (s;p) Eacr (s;s) Eacr (t;s) Eacr (t;p) −16 ˜cr A × 10 SSQ

RN01

± 1.0 200.2 ± 0.9 171.5 ± 0.9 211.9 ± 1.5 2.73 ± 0.40 229.9

0.0350

RN06

± 1.0 199.8 ± 0.9 171.2 ± 0.9 210.7 ± 1.5 2.59 ± 0.38

229.6

0.0345

The slight deviation of the condence intervals in RN01 compared to a previous contribution

25

is ascribed to minor changes in the underlying numerics. As stated above, almost

no dierence can be observed between the results of both reaction networks. Nevertheless, RN06 leads to a higher agreement and, more importantly, to higher parameter signicance during tting of the adsorption models; the inclusion of reversible ethene formation avoids insignicant values for

Eacr (s;p)

and

Eacr (t;p).

Thus, without overinterpreting the results, it

can be stated that the backward reaction of ethene formation has to play a signicant role at certain reaction conditions covered by the experiments, although the protonation of ethene should be energetically unfavored. is highly ionic.

41,49,50

70

A possible explanation refers to the transition state that

Usually, these ionic transition states impede the formation of primary

products, not only in protonation,

46

but also in cracking and dimerization steps. For ethene

as product, however, the evolving primary intermediate can be easily stabilized by another oxygen in the zeolite lattice assuming a ring like transition state.

41

This step might be ster-

ically hindered for larger primary product intermediates, whereas it allows the formation of

17

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

ethene in cracking steps and also the protonation to ethyl intermediates as rst step of the backward reaction. Another possibility is the formation of a primary alkoxide out of ethene that might have higher stability compared to a primary carbenium ion. Nevertheless, also in that case, the following transition state is of ionic character. The agreement with experimental data resulting from an application of RN06 is found in the parity plots of Figure 1. (b) Ethene T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.08 0.06

Mole fraction experimental / -

0.1

0.04 0.02 0

(c) Propene T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.25

0.2

0.15

0.1

0.05

0.2

0.15

0.1

0.05

0 0

0.02

0.04 0.06 0.08 0.1 Mole fraction model / -

0.12

0 0

0.05

0.1 0.15 0.2 Mole fraction model / -

(d) 1

0.7

0

0.6 0.5 0.4

0.1 0.15 0.2 Mole fraction model / -

0.25

0.08

0.06

C=6 -C=12 T T T T T T T T

0.04

0.02

0 0.4

0.05

0.1 Mole fraction experimental / -

Mole fraction experimental / -

0.8

0.25

(e) Pentenes T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.9

Butenes T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.25 Mole fraction experimental / -

(a) 0.12 Mole fraction experimental / -

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

Page 18 of 50

0.5

0.6 0.7 0.8 Mole fraction model / -

0.9

1

0

0.02

0.04 0.06 0.08 Mole fraction model / -

= 633 K = 653 K = 673 K = 693 K = 703 K = 713 K = 723 K = 733 K 0.1

Figure 1: Parity plots for ethene, (a), propene, (b), butenes, (c), pentenes, (d), and C6 to C12 olens, (e), resulting from an application of reaction network RN06.

As preparation for the following adsorption studies, some minor modications of the microkinetic model are made. At rst, the reference olens

23

are removed. Theoretically, these

species are required to avoid a thermodynamic inconsistency in the single-event concept: The protonation should be independent from the carbon number, meaning that it only depends on the type of protonated intermediate.

23

If the same is assumed for the deprotonation, the

equilibrium constant of protonation will always be unity.

71

Hence, the reference olens are

introduced to maintain a carbon number dependence, but to avoid a separate tting of each isomer.

72

However, in the original single-event model, the equilibrium constants for protona-

18

ACS Paragon Plus Environment

Page 19 of 50

tion are not estimated, but calculated before tting.

24

The underlying correlations

34

ensure

a carbon number dependence of the resulting constants, which will be also shown in the next section. Consequently, the conversion into reference olens is dropped because it causes additional insecurities through calculating thermodynamic data. Another change is that the earlier simplication

24

of using only the equations for 1-alkenes to describe the

π -complex of

all isomers is removed. Depending on the olen structure, either the correlation for 1-alkenes or for 2/3/4-alkenes is used. These adaptations are already considered for the results shown in Table 2 and in Figure 1.

3.2

Adsorption States

As shown in Section 2.3, several dierent approaches exist to describe the adsorption of hydrocarbons on acid zeolites. At rst, the resulting values should be analyzed as a function of the carbon number. This is done in Figure 2 for enthalpy and entropy of physisorption or of

π -complex

formation. An evaluation is performed for linear 1-alkenes as well as linear

2-alkenes. A dierentiation between these can only be made when using the correlations by Nguyen et al.

34

(a)

(b) -40

-40

So / J/(mol K)

Nguyen 1-olefins Nguyen 2-olefins Denayer De Moor

-60 -80

So or

-100 -120

phys

Ho or

Ho / kJ/mol

-20

phys

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

-140

-60 -80 -100 -120 -140 -160

Nguyen 1-olefins Nguyen 2-olefins Denayer De Moor Toch

-180 -200

-160

-220 2

4

6

8

10

12

2

4

6

8

10

12

Carbon number / -

Carbon number / -

Figure 2: Dierences in enthalpy, (a), and entropy, (b), caused by

π -complex

formation or

physisorption, calculated for linear 1-alkenes and 2-alkenes; the dierent literature concepts cover the regular correlations by Nguyen et al. Moor et al.

33

34,63

(Nguyen), the physisorption data by de

(De Moor) as well as by Denayer et al.

by Toch et al.

69

62

and the simplied entropy approach

(Toch).

19

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

Page 20 of 50

Figure 2a shows increasing enthalpy release for higher carbon numbers, independent of the prevailing adsorption mechanism. This can be attributed to higher van der Waals forces caused by the growing chain length. As expected, the

π -complex between olen and acid site

leads to higher stabilities of the adsorbates compared to pure alkane physisorption eects. For the latter, no signicant dierence between the correlations by de Moor et al. et al.

62

can be observed.

33

and Denayer

Furthermore, as these two concepts rely on the measured data

of alkane adsorption, not all isomers can be described and thus, no dierentiation between 1-alkenes and 2-alkenes is possible. Nguyen et al.

34

In contrast, the theoretically derived correlations by

show that the stabilization through the

a fact that is conrmed by other studies. between physisorption and

27,28

π -complex

is higher for 2-alkenes,

With higher carbon numbers, the dierence

π -complex diminishes for 1-olens.

Due to the xation at one end

of the chain, more steric hindrance might arise for higher carbon numbers compared to the

π -complex direction.

of 2-olens, where at least one methyl group can be positioned towards another

27

In Figure 2b, similar trends are observed: the loss in entropy is higher for the evolution of the

π -complex

surprising, as the

and again, the 2-olens are subjected to a higher change.

π -complex

This is not

leads to a stronger xation, whereas physisorbed hydrocarbons

have higher mobility. However, similar to Figure 2a, the dierence between

π -complex

and

physisorption decreases for a longer hydrocarbon chain, where an eective xation is impeded by steric constraints.

Again, there is no signicant deviation between both physisorption

models. Finally, Figure 2b shows that assuming the physisorption entropy to correspond to one third of the translational entropy leads to almost carbon number independent values. This is caused by the small variations within molar masses and therefore in translational entropies of the dierent compounds.

Although this concept is an ecient solution that

works with networks where the involved species dier signicantly in molar masses, application to olen cracking might be problematic.

an

At least it is expected that the real

physisorption entropies show higher variation than the red data points in Figure 2b.

20

69

ACS Paragon Plus Environment

Page 21 of 50

The chemisorption step is analyzed in Figure 3.

Here, protonation enthalpies and en-

tropies for linear 1-alkenes as well as linear 2-alkenes are shown.

Figure 3b additionally

depicts the Boudart criteria for protonation entropies. Whereas the upper limit (black) is universal and thus valid for all species, the lower limit depends on the protonation enthalpy which is why it can be calculated only for the Nguyen correlations (blue and orange).

(a)

(b)

-30

20

Nguyen 1-olefins Nguyen 2-olefins Martens Toch

0 -20

So / J/(mol K)

Ho / kJ/mol

-40

-50

-60

-40 -60

prot

prot

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

-100

Nguyen 1-olefins Nguyen 1-olefins alt. Nguyen 2-olefins Nguyen 2-olefins alt.

-70

-80 2

4

6

8

10

-80

-120 -140

12

2

4

6

8

10

12

Carbon number / -

Carbon number / -

Figure 3: Dierences in enthalpy, (a), and entropy, (b), caused by protonation, calculated for linear 1-alkenes and 2-alkenes; the dierent literature concepts cover the regular correlations by Nguyen et al.

34,63

(Nguyen), the alternative correlations by Nguyen et al.

and the physisorption data by Denayer et al. by Martens et al. Toch et al.

69

64

62

58

(Nguyen alt.)

in combination with the protonation approach

(Martens) as well as in combination with the protonation approach by

(Toch); for the protonation entropies, the upper (black) and the lower Boudart

criterion (blue and orange) is shown.

73

Figure 3a yields insight into the dierent types of Nguyen correlations. The regular set of equations leads to almost constant protonation enthalpies with the values for 1-olens being signicantly lower.

Contrary trends are obtained for the alternative correlations:

carbon number dependent values are obtained.

Here,

These slightly decrease for the 2-olens,

whereas a sharp increase is observed for 1-olens.

This increase for linear olens is not

expected from theory and can impair the results when using these equations in a kinetic model. Especially the value of ethene that shows the highest protonation stabilization by far is extremely unrealistic. The change in entropy due to protonation decreases for higher carbon numbers of 1olens according to Figure 3b. In contrast, the formation of alkoxides out of 2-olens causes

21

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

Page 22 of 50

increasing protonation entropies, which again is caused by steric constraints. trend results from the approach by Martens et al. protonated intermediates.

64

27

The same

where carbenium ions are assumed as

However, an entropy gain is predicted for a carbon number of

twelve, which is not reasonable. Furthermore, these values are not in accordance with the upper Boudart criterion al.

69

73

from a carbon number of eight on.

The approach by Toch et

also leads to results outside the Boudart criteria; here, the values are too low. As the

correlations of Nguyen et al.

34

were created for maximum carbon numbers of eight, their

limited validity for higher compounds can be seen in Figure 3b: the change in entropy for C12 is slightly above the upper Boudart criterion.

Thus, these correlations should not be

used for chain lengths higher than C12. As described in Section 2.4.2, dierent possibilities exist to describe the saturation concentration

C sat (i)

of hydrocarbons on the catalytic surface. For Figures 2 and 3, the group

contribution method by Elbro et al.

67

is used, which is also applied for most of the investi-

gations shown in this work if not stated otherwise. This is done for several reasons. Firstly, this method allows for the calculation of individual values for all species present in the huge reaction network, whereas for the HBT concept, tabulated values are required that cannot be found for hundreds of dierent compounds. Secondly, the HBT method should be applied below the critical temperature,

65

whereas for Elbro et al.,

than the normal boiling point are required.

65

67

temperatures not higher

This is because both concepts were created to

calculate molar volumes of the condensed phase; an application to the gas phase can only be a prediction based on similarities. The temperature limits of both methods are not possible at typical cracking conditions. However, extrapolation to temperatures around

700 K can be

performed for Elbro in contrast to the HBT method, even though the results might not be physically reasonable. Preliminary studies (not shown) revealed a better agreement for Elbro evaluated at the mean experimental temperature of

683 K

compared to the HBT method

applied at the respective boiling temperatures. On the other hand, the same studies yielded no signicant dierences from the results in Figures 2 and 3 when values stemming from the

22

ACS Paragon Plus Environment

Page 23 of 50

HBT method were used. For this procedure, it was irrelevant whether the tabulated values of alkanes or of olens were chosen. Figure 4 shows a plot similar to Figure 2 for 1-olens as well as 2-olens with a methyl side group at the second carbon atom. (b)

(a) -40

So / J/(mol K)

Ho / kJ/mol

-40

Nguyen 1-olefins Nguyen 2-olefins Denayer De Moor

-60 -80

So or

-120

phys

Ho or

-100

phys

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

-140

-60 -80 -100 -120 -140 -160

Nguyen 1-olefins Nguyen 2-olefins Denayer De Moor Toch

-180 -200

-160

-220 4

6

8

10

12

4

6

Carbon number / -

8

10

12

Carbon number / -

Figure 4: Dierences in enthalpy, (a), and entropy, (b), caused by

π -complex

formation or

physisorption, calculated for 2-methyl-1-alkenes and 2-methyl-2-alkenes; the dierent literature concepts cover the regular correlations by Nguyen et al. data by Denayer et al.

62

34,63

(Denayer) as well as by de Moor et al.

entropy approach by Toch et al.

69

33

(Nguyen), the physisorption

(De Moor) and the simplied

(Toch).

In Figure 4, the special role of isobutene is obvious because the results deviate in comparison to the other carbon numbers. No branched olens are covered by the regular Nguyen correlations which is why the results are similar to Figure 2; also for the remaining data sets, no signicant dierences are found. Again, the problematic description of olen physisorption using the approach by Toch et al.

69

can be observed. The protonation step for branched

olens is analyzed in Figure 5.

23

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

(a) Nguyen 1-olefins Nguyen 1-olefins alt. Nguyen 2-olefins Nguyen 2-olefins alt. Nguyen 2-olefins alt. sec.

60

(b) 20

20

Ho / kJ/mol

Nguyen 1-olefins Nguyen 2-olefins Martens Toch

0 -20

So / J/(mol K)

40

0 -20

-40 -60 -80

prot

prot

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

Page 24 of 50

-40

-100

-60

-120

-80

-140 4

6

8

10

12

4

6

Carbon number / -

8

10

12

Carbon number / -

Figure 5: Dierences in enthalpy, (a), and entropy, (b), caused by protonation, calculated for 2-methyl-1-alkenes and 2-methyl-2-alkenes; the dierent literature concepts cover the regular correlations by Nguyen et al. et al.

58

34,63

(Nguyen), the alternative correlations by Nguyen

(Nguyen alt. and Nguyen alt. sec.) and the physisorption data by Denayer et al.

in combination with the protonation approach by Martens et al. combination with the protonation approach by Toch et al.

69

64

62

(Martens) as well as in

(Toch); for the protonation

entropies, the upper (black) and the lower Boudart criterion (blue and orange) is shown.

73

Figure 5a shows totally dierent trends for the protonation enthalpies when a tertiary alkoxide is calculated using the alternative set of equations by Nguyen et al. olens alt.

and Nguyen 2-olens alt.).

63

(Nguyen 1-

Here, steric hindrance destabilizes the protonated

intermediate, which is even more pronounced for branched 2-olens. Although a lower stability is expected for tertiary alkoxides, the linear increase of these two data sets is again questionable.

The results of the alternative approach describing protonation of branched

2-olens to secondary alkoxides (Nguyen 2-olens alt.

sec.)

are more reasonable as these

are comparable to the regular Nguyen correlations. Also here, the special role of isobutene protonation becomes obvious. No additional equations for the entropy change are available in the alternative Nguyen concept which is why the results in Figure 5b are similar to Figure 3b. As stated above, using the concepts of Martens et al.

64

as well as of Toch et al.

69

is problematic for this reaction

network of olen cracking as the resulting values for protonation entropy are outside the Boudart criteria. This also holds for the Nguyen values for isobutene, but here, the lower criterion is above the upper one due to the positive protonation enthalpy.

24

ACS Paragon Plus Environment

Page 25 of 50 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

In summary, a variety of concepts and approaches to describe adsorption phenomena of hydrocarbons on acid zeolites exists, all of them leading to dierent results. Thus, in the following, which adsorption model suits the reactivity of olen cracking data best will be evaluated.

3.3

Mathematical Formulation of Overall Reactivity

In order to have a better overview of the dierent models and the corresponding results, these are divided into three parts and introduced with Tables 3 to 5. Here, it is shown whether the rst adsorption step consists of physisorption or the formation of a

π -complex.

Another

column reveals the implementation of protonation as second adsorption step, thereby leading to intrinsic parameters. If protonation is not explicitly considered, the estimated activation energies are of composite character and contain the respective protonation enthalpies. For both adsorption steps, the sources for the adsorption data are shown.

Furthermore, the

tables show the type of saturation value used for the Langmuir approach as well as the intermediate concentrations considered for the denominator.

Finally, the underlying rate

equation and specic remarks are presented. The rst section comprises the description of overall reactivity according to von Aretin et al.

24

combined with dierent implementations of the Nguyen correlations, see Table 3.

Table 3: First set of microkinetic models analyzed in this work, specied by type and data source of one or two adsorption steps, by the saturation value and considered concentrations of the Langmuir term and by the implemented rate equations. no.

ads. 1

01 02

π π

03

π

data 1

a Nguyen a

Nguyen

a

Nguyen

ads. 2

data 2

Langmuir

rate

prot.

a Nguyen

-

-

Ct : π & prot. Ct : π

Eq. 8

-

Eq. 16

comp. values ;

Eq. 8

alt. approach

prot.

Nguyen alt.

c

remarks

b

incl. Oref

Ct : π

& prot.

for s and t

a cf. references 34,63 b Composite values; c cf. references 58

Whereas the Nguyen correlations in Table 3 describe the evolution of a

25

ACS Paragon Plus Environment

π -complex as rst

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

Page 26 of 50

adsorption step, alkane physisorption approaches are analyzed in the second part. include the equations for overall reactivity according to Martens et al.,

64

These

combined with

either the Denayer or the de Moor adsorption correlations, see Table 4. Table 4: Second set of microkinetic models analyzed in this work, specied by type and data source of one or two adsorption steps, by the saturation value and considered concentrations of the Langmuir term and by the implemented rate equations. no.

ads. 1

04

phys.

data 1

a Denayer

ads. 2

data 2

-

-

Langmuir

C

sat

(i):

rate

phys.

remarks

Eq. 17

comp. values; w/

nd

05

phys.

c de Moor

-

C

-

sat

Ct ;

2

(i):

phys.

Eq. 17

incl. Oref ;

b

dim. : phys.

comp. values; w/Ct ; incl. Oref ;

06

phys.

c de Moor

prot.

- (H ); Martens

07

phys.

de Moor

c

-

d (S )

C

sat

2

(i):

Eq. 17

2 phys. &

π

dim.: phys.

comp. values; w/

C sat (i):

-

phys.

nd

Eq. 18

Ct ;

nd

incl. Oref ;

dim.: phys.

comp. values; w/o 2

nd

Ct ;

incl. Oref ;

dim.: phys.

a cf. references 62 b Relates to the second olen during dimerization; c cf. references 33 d cf. 64 references

Finally, as the Nguyen correlations lead to results of higher accuracy, these are further rened in the third part of the analysis. This includes the estimation of protonation enthalpies, see Table 5. Table 5: Third set of microkinetic models analyzed in this work, specied by type and data source of one or two adsorption steps, by the saturation value and considered concentrations of the Langmuir term and by the implemented rate equations. no.

ads. 1

08

π

a Nguyen

09

π

Nguyen

π

a Nguyen

10 11

π

data 1

a

a

Nguyen

ads. 2 prot. prot.

data 2

b est. (H ); a Nguyen (S ) est. (H );

a (S )

Langmuir

rate

remarks est. of s and t

Ct : π

& prot.

Eq. 8

Ct : π

& prot.

Eq. 8

s =

Nguyen prot.

pre-set (H );

a (S )

est. of t

Ct : π

& prot.

Eq. 8

Ct : π

& prot.

Eq. 19

Nguyen prot.

pre-set (H );

a (S )

0 kJ mol−1 ;

Nguyen

0 kJ mol−1 ; −1 t = −30 kJ mol −1 s = 0 kJ mol ; −1 t = −30 kJ mol ; s =

2

nd

c

dim. : phys.

a cf. references 34,63 b Estimation; c Relates to the second olen during dimerization;

26

ACS Paragon Plus Environment

Page 27 of 50 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

3.3.1 Von Aretin et al. 24 - Use of Nguyen Correlations Figures 3a and 5a show high protonation enthalpies when using the regular equations of Nguyen et al.

34

Thus, it is expected that a signicant amount of protonated intermediates,

i.e., alkoxides according to the Nguyen equations, exists on the acid sites and that the introduction of Eq. 8 leads to improved results. However, the inclusion of the concentration term of protonation to the Langmuir approach leads to a poorer description of experimental data compared to the original model

24

ignoring the concentration of alkoxides. In Table 6, the

results including the concentration term of protonated intermediates can be found as model no. 1. In addition, the parity plots are shown in Figure S1 in the Supporting Information. Table 6: Estimated activation energies

Eacr

and pre-exponential factors

condence intervals, and the sum of squared residuals

SSQ

Acr ,

including 95%

for the rst set of models using

dierent implementations of the Nguyen correlations; all activation energies are given in kJ mol−1 , whereas the pre-exponential factor is shown in s−1 . parameter

Eacr (s;p) Eacr (s;s) Eacr (t;s) Eacr (t;p) −16 ˜cr A × 10 SSQ

no. 1

no. 2

209.8

a 189.0

192.2

158.2

± 1.2 ± 1.2 174.8 ± 1.1 210.8 ± 5.6 0.95 ± 0.18 0.0480

± 2.8 ×106 a 0.0036 ± 0.0006

± 1.4 ± 1.2 91.9 ± 1.5 5 216.2 ± 6.1 ×10 0.32 ± 0.07

0.0373

0.0669

a a

129.0

a 256.7

± ± ±

no. 3 1.0 1.0 0.9

a Composite value.

202.3 171.3

The decline in agreement is surprising with respect to the high alkoxide stability suggested by Figures 3a and 5a. The lower agreement with experimental data speaks for an incorrect description of the protonation, an error which then accumulates in the Langmuir sum. In the following, the total concentration of adsorbed intermediates at typical reaction conditions is calculated to further analyze the reason for this deviation. the total concentration of

π -complexes

the relative amount of free acid sites

Figure 6 shows

and of alkoxides (left) as well as the coverages and

Θ∗

(right) as a function of pentenes conversion.

In

addition, the same depiction over catalyst mass can be found in Figure S2 in the Supporting Information.

Dierent reaction conditions applied during the kinetic measurements

27

ACS Paragon Plus Environment

24

are

Industrial & Engineering Chemistry Research

covered. The results are obtained by evaluating the single-event model according to Eq. 8

10-3 ( a)

2.2

1

(b)

0.8

2

Coverage / -

Concentration / mol/kg cat

in combination with the regular Nguyen correlations.

C

1.8

Cchem

1.6

T = 733 K

*

0.6 chem

0.4 T = 733 K 0.2

1.4 1.2

0 0

0.1

0.2

0.3

0

0.1

-3

(c)

1

0.3

(d)

C 9

C

chem

0.8

Coverage / -

Concentration / mol/kg cat

10

0.2

Conversion of C =5 / -

Conversion of C =5 / -

T = 693 K 8

*

0.6 chem

0.4 T = 693 K 0.2

7

0 0

0.1

0.2

0.3

0.4

0

0.1

0.2

0.3

(e ) 1

0.04

(f ) T = 633 K

*

0.8

Coverage / -

C Cchem

0.03

0.4

Conversion of C =5 / -

Conversion of C =5 / Concentration / mol/kg cat

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

Page 28 of 50

T = 633 K 0.02

chem

0.6 0.4 0.2

0.01

0 0

0.1

0.2

0.3

0.4

0

0.1

Conversion of C =5 / -

0.2

0.3

0.4

Conversion of C =5 / -

Figure 6: Total concentration of adsorbed species, either

π -complex

or chemisorption, (a),

(c) and (e), and relative coverages or amount of free acid sites, (b), (d) and (f ), at three

733 K and inlet partial pressure of 1-pentene pin C5= = 42.7 mbar, in in (a) and (b), 693 K and pC5= = 70.3 mbar, (c) and (d), 633 K and pC5= = 42.7 mbar, (e) and (f ); the total pressure pt is set to 1.23 bar for all subgures. dierent reaction conditions:

As Figure 6a reveals, alkoxide concentrations are negligible at high temperatures.

28

ACS Paragon Plus Environment

In

Page 29 of 50

combination with Figure 6b, it can be seen that under these conditions, even the total concentration of

π -complexes

is extremely low and only a minor fraction of the acid sites is

occupied. However, for intermediate temperatures and increased feed partial pressures, the concentration of alkoxides is higher than the one of

π -complexes

at the inlet region of the

reactor as shown in Figure 6c. In addition, throughout the whole reactor length, the coverage of

π -complexes

and alkoxides is almost similar, see Figure 6d.

to neglect the chemisorbed species at these conditions.

Thus, it is not reasonable

Another decrease in temperature

aggravates this situation, which is depicted in Figures 6e and f.

Here, the concentration

of alkoxides is signicantly higher along the whole catalyst bed. Furthermore, the relative concentration of free acid sites is comparably low. It follows that the use of a site balance including both the chemisorbed species and the compounds interacting in a

π -complex

is

inevitable to maintain a broad application range of the model. The dependence of adsorbed intermediates on temperature, reaction progress and feed partial pressure is further analyzed in Figure 7.

(a)

0.5

0.8

0.15

0.6

/-

0.2

chem

/-

/*

(c)

(b)

1

0.1 0.05

4 10-4

2 0

600

650

700

750

0 6

800

4 10

-4

2 0

Temperature / K

Catalyst mass / kgcat

600

650

700

750

800

4 10-4

0

0

0.5

pin / bar C5

0

600

800 750

0.5

700 650

pin / bar C5

Temperature / K

Temperature / K

0.5

0 1

800 750

800

/chem

/0.1

1

750

1

0.2

0

700

(f)

0.3

0.5

600

650

Catalyst mass / kgcat

(e)

1

2

Temperature / K

Catalyst mass / kgcat

(d)

/-

0.4 0.2

0 6

0 6

*

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

700 0

650 600

Temperature / K

1

800 750

0.5 pin / bar C5

700 0

650 600

Figure 7: Surface plots for the relative amount of free acid sites, (a) and (d), the

Temperature / K

π -complex

coverage, (b) and (e), and the alkoxide coverage, (c) and (f ), which are shown as a function of temperature and catalyst mass, (a) to (c), and of temperature and inlet partial pressure of 1-pentene, (d) to (f );

pt

1.195 bar for all subgures, whereas pin C5 = 56.5 mbar for −4 mass Wmax = 2.5 × 10 kgcat for (d) to (f ), respectively.

is set to

(a) to (c) and maximum catalyst

29

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

Page 30 of 50

From Figure 7a, it is obvious that higher temperatures lead to a less occupied catalyst surface. The site balance is of less importance at temperatures higher than

700 K

because

here, the relative amount of free acid sites is always above 0.92. Interestingly, the trend for the two types of adsorbates is dierent, see Figures 7b and c. The lower the temperature, the more alkoxides can be found on the surface whereas less olens interact in a at

600 K.

π -complex

When increasing the temperature, the free sites emerging from alkoxide desorption

can be eectively used by is observed around

π -complexes due to the lower entropy loss, which is why a maximum

650 K.

Above this value, also the

π -complex

coverage decreases, but to

a lower extent compared to alkoxides. A trend already observed in Figure 6 is the almost linear increase of

π -complexes

along the reactor length, whereas the alkoxide concentration

shows only minor variation. This can be explained with the changing reactant composition since especially butenes contribute to the

π -complex

to mainly C2 to C4 olens along the reactor.

coverage and 1-pentene is converted

Similar coherences for the temperature are

found in Figures 7d to f, in which the inuence of feed partial pressure is additionally shown. Here, higher values lead to higher coverage of both the latter reveals a sharper increase.

π -complex

and alkoxide; again,

Compared to temperature, the inuence of the feed

partial pressure is less pronounced, but nevertheless, at temperatures lower than

700 K,

the concentration of chemisorbed intermediates is not negligible at almost all feed partial pressure values. Summarized, the analysis shows that for a physically consistent microkinetic model, the concentration of chemisorbed species cannot be ignored when using the regular correlations by Nguyen et al.

34

Next, it has to be investigated why the agreement is worse when correctly accounting for the concentrations of all adsorbed intermediates. In a rst step, an adapted version of Eq. 9 is used, cf. Eq. 16,

rcr (m; n) = ne k˜cr,co (m; n) with

Ct KLπ (Oi ) p (Oi ) P , 1 + j KLπ (Oj ) p (Oj )

k˜cr,co (m; n) = k˜cr (m; n) K prot (Oπi ; m) . 30

ACS Paragon Plus Environment

(16)

Page 31 of 50 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

This is the typical approach used by the Ghent group are available.

23,64

when no values for protonation

It is adapted to olen cacking, thus accounting for the

π -complex

forma-

tion whereas the hydrogenation/dehydrogenation expressions are removed. The composite rate coecient

k˜cr,co (m; n)

consists of both the intrinsic rate coecient and the protonation

equilibrium constant. This is why the estimated activation energies contain the protonation enthalpy, whereas the protonation entropy contributes to the composite pre-exponential factor

A˜cr,co .

The results of this approach are listed in Table 6 where it is denoted as model

no. 02; the corresponding parity plot is shown in Figure S3 in the Supporting Information. Throughout this section analyzing dierent application variations of the Nguyen correlations, model no. 02 is the only example where reference olens are used. As stated above, these are necessary if no carbon number dependence of protonation is implemented which is the case when tting composite values with one mutual pre-exponential factor. The estimated activation energies of model no. 02 are signicantly lower. There are two possible explanations: either the protonation enthalpies have values of around

−50 kJ mol−1

and therefore decrease

the composite activation energies or the protonation enthalpies calculated via the Nguyen correlations are too low. The latter scenario would decrease the energetic starting level of the model, thereby increasing the estimated intrinsic activation energies. The only exception is the cracking value (t;p), which is higher than before, but also shows poor numeric significance. Model no. 02 leads to lower agreement with the experimental data. In contrast to model no. 01, the carbon number dependence of the protonation step is lost since no entropies are calculated here. Instead, the protonation entropy is part of the estimated pre-exponential factor, which is indeed three orders of magnitude lower. However, one pre-exponential factor to describe the overall reactivity including protonation and surface reaction is obviously not sucient. In contrast to that, no dimerization occurs for hydrocracking, which reduces the amount of carbon numbers to be covered, thus enabling an application of Eq. 9.

23,64

The less

accurate description of model no. 02 in olen cracking underlines the importance of correctly describing the protonation step. Although the

31

π -complex

ACS Paragon Plus Environment

is of high stability,

28,34

it is not

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

Page 32 of 50

sucient to have it as only intermediate described in detail in a microkinetic model for olen cracking. Next, an improved protonation model should be found. Therefore, the alternative equations of Nguyen et al.

58

are implemented according to Eq. 8, thus accounting for the con-

centration of protonated intermediates. The resulting model is denoted as no. 03 in Table 6. Apparently, this approach leads to a poor description of the experimental data, see the parity plots in Figure S4 in the Supporting Information. Since also the regular Nguyen protonation correlations show inaccurate results when correctly considering the concentration of protonated intermediates, the evolution of alkoxides or at least their high stability have to be challenged; the description both without

34

and with stability dierence

58

between secondary

and tertiary alkoxides is not satisfying, see models no. 01 and 03 in Table 6.

3.3.2 Martens et al. 64 - Use of Alkane Physisorption Approaches Before further rening the description of protonation with the Nguyen correlations, the results of the alkane physisorption approaches should be discussed. For this, a reaction rate formulation similar to Eq. 16 is proposed which includes the description of physisorption according to Eq. 9, see Eq. 17,

C sat (i) KLphys (Oi ) p (Oi ) , P 1 + j KLphys (Oj ) p (Oj )   k˜cr,co (m; n) = k˜cr (m; n) K prot Ophys ; m . i rcr (m; n) = ne k˜cr,co (m; n) Ct

with

(17)

Model no. 4 in Table 7 represents an application of this approach when using physisorption data by Denayer et al. concentrations.

62

in combination with the HBT method

66

for the saturation

Due to the lack of protonation data, all results in Table 7 are composite

values. Furthermore, all models are evaluated using reference olens. Finally, they contain the assumption of negligible concentrations of protonated intermediates; no balance for the

32

ACS Paragon Plus Environment

Page 33 of 50 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

acid sites is thus contained. For some of these models, no reasonable results were achieved when the Eley-Rideal mechanism of the original model

24

was implemented where the second

reactant olen in the dimerization is characterized only by its partial pressure. Thus, another physisorption term is introduced for all models in Table 7 to account for connement eects for the second olen in the dimerization. Table 7: Estimated activation energies

Eacr

and pre-exponential factors

condence intervals, and the sum of squared residuals

Acr ,

including 95%

SSQ

for the second set of models −1 using dierent alkane physisorption approaches; all activation energies are given in kJ mol , −1 whereas the pre-exponential factor is shown in s . parameter

Eacr (s;p) Eacr (s;s) Eacr (t;s) Eacr (t;p) −12 ˜cr A × 10 SSQ

no. 4

± 2.4 113.4 ± 2.7 96.7 ± 2.0 133.0 ± 5.0 0.97 ± 0.35

no. 5

153.2

167.3 136.2 114.9

± ± ±

no. 6 1.0 1.0 0.9

± 7.9 ×103 7.66 ± 1.14

200.0

0.1051

0.0332

± 1.7 (1.42 ± 0.14) × 104

± 1.4 160.0 ± 1.4 136.9 ± 1.2 200.0 ± 196.2 73.28 ± 15.62

0.0971

0.0608

168.8 149.0

±

191.7

± ±

no. 7 0.8

193.8

0.6

1.2

×103

158.3

The agreement with experimental data is poor for model no. 04, see also Figure S5 in the Supporting Information. The results do not improve when changing the underlying calculation for the saturation concentrations. Consequently, using the Denayer correlations cannot be recommended at olen cracking conditions. In contrast, reasonable results are achieved for model no. 05 where physisorption data by de Moor et al.

33

is used, which is also obvious

from Figure S6 in the Supporting Information. This observation is surprising regarding Figures 2 and 4: here, the correlations of de Moor and Denayer lead to almost similar enthalpy and entropy values. Nevertheless, the small variations lead to signicantly dierent tting results. In general, the alkane physisorption data by de Moor et al. olen interaction. However, the accuracy is lower than for the best

33

is suitable to describe

π -complex

model in Ta-

ble 6. Moreover, numerical security is less for the composite values, which is revealed by the high condence interval of the cracking parameter (t;p). In model no. 05, this value cannot be determined to a reasonable order of magnitude; it always tends towards the maximum constraint, which is

200 kJ mol−1

in this case. Consequently, when applied in olen cracking,

33

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

Page 34 of 50

the approach via alkane physisorption leads to estimated parameters of articial character. These values cannot be interpreted in terms of intermediates and thus allow no mechanistic insight. In a subsequent step, the approaches according to Martens et al., Toch et al.,

69

64

cf. Section 2.4.2, and

cf. Section 2.4.3, are tested. Both make use of Eq. 17, but the description of

physisorption and protonation entropy is dierent, cf. Eq. S6 in the Supporting Information and Eq. 11 with Eqs. 13 and 14, respectively. The approach by Martens et al.

64

is denoted

as model no. 6 in Table 7. This implementation leads to poor agreement, see Figure S7 in the Supporting Information, and low numerical security. This deteriorates for the approach by Toch et al.

69

(not shown).

Because of the high agreement achieved with model no. 5,

physisorption data by de Moor et al. switching to Denayer et al.

62

33

is used, but the ndings remain unchanged when

From this, it follows that both concepts expressing the overall

reactivity, though successfully applied in other reaction systems, are not suitable to describe the whole bandwidth of olen interconversion during cracking. This matches the high deviations found for both protonation entropies in Figures 3b and 5b where Boudart's criteria

73

are violated. Finally, the combination of physisorption and

π -complex formation is tested.

This would

correspond to an olen being only physisorbed on the catalytic surface rst, so without the contribution of an acid site. Subsequently, this physisorbed compound might migrate to an acid center where the by Nguyen et al.

34

π -complex

is formed. For the theoretical description, the correlations

and de Moor et al.

33

are combined. In contrast to the methodology in

Table 6, saturation eects are correlated to the catalyst surface and not to the acid sites in accordance with Thybaut et al.,

with

39

cf. Eq. 18,

C sat (i) K π (Oi ) p (Oi ) , rcr (m; n) = ne k˜cr,co (m; n) Ct LDS X X phys LDS = 1 + C sat (j) K π (Oj ) p (Oj ) + KL (Oj ) p (Oj ) . j

j

34

ACS Paragon Plus Environment

(18)

Page 35 of 50 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

The resulting description can be found as model no. 7 in Table 7 and as Figure S8 in the Supporting Information. An acceptable agreement is observed, but the results are less accurate compared to model no. 5. Still, the highest agreement is achieved when assuming a

π -complex

as crucial intermediate before protonation. Composite models that include only

physisorption suer from the fact that the estimated parameters contain several adsorption steps, which might be dicult to express in a common parameter. Furthermore, only the models based on

π -complex

formation allow for insight into intermediate stabilities.

3.3.3 Own Enthalpy Fitting - Use of Nguyen Correlations Since the alkane physisorption approaches lead to results of comparably low accuracy and the description of protonation using the Nguyen correlations as shown above is still not satisfying, the latter should be rened. An application of Eq. 8 allows for an estimation of the protonation enthalpies to secondary and tertiary intermediates because the protonation equilibrium constant is also part of the denominator and therefore distinguishable from the activation energies. Table 8:

The results can be found as model no. 08 in Table 8.

Estimated activation energies

exponential factors uals

39

A

cr

Eacr ,

protonation enthalpies

∆prot H ◦

and pre-

, including 95% condence intervals, and the sum of squared resid-

SSQ

for the third set of models where the Nguyen correlations are rened with own −1 ttings; all activation energies and protonation enthalpies are given in kJ mol , whereas the −1 pre-exponential factor is shown in s . parameter

Eacr (s;p) Eacr (s;s) Eacr (t;s) Eacr (t;p) A˜cr × 10−16 ∆prot H ◦ (s) ∆prot H ◦ (t) SSQ

no. 8

no. 9

no. 10

no. 11

± 1.1 157.8 ± 1.1 155.4 ± 57.0 194.9 ± 57.5 4.58 ± 0.71 0.03 ± 0.14 -16.1 ± 57.2

± 0.9 158.5 ± 0.9 152.3 ± 53.2 193.3 ± 53.8 5.04 ± 0.67

± 0.9 158.3 ± 0.9 169.7 ± 0.8 210.6 ± 8.0 4.86 ± 0.64

± 0.9 158.3 ± 1.0 171.7 ± 0.8 204.4 ± 2.9 5.18 ± 0.72

-

-

-

-

-

0.0279

0.0278

0.0280

0.0282

176.6

177.2

-12.5

±

177.0

53.1

177.6

The agreement of this model with experimental data is extremely high, which follows from the parity plots in Figure 8.

35

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research

(b) Ethene T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.08 0.06

Mole fraction experimental / -

0.1

0.04 0.02 0

(c) Propene T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.25

0.2

0.15

0.1

0.05

0.2

0.15

0.1

0.05

0 0

0.02

0.04 0.06 0.08 0.1 Mole fraction model / -

0.12

0 0

0.05

0.1 0.15 0.2 Mole fraction model / -

(d) 1

0.7

0

0.6 0.5 0.4

0.1 0.15 0.2 Mole fraction model / -

0.25

0.08

0.06

C=6 -C=12 T T T T T T T T

0.04

0.02

0 0.4

0.05

0.1 Mole fraction experimental / -

Mole fraction experimental / -

0.8

0.25

(e) Pentenes T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.9

Butenes T = 633 K T = 653 K T = 673 K T = 693 K T = 703 K T = 713 K T = 723 K T = 733 K

0.25 Mole fraction experimental / -

(a) 0.12 Mole fraction experimental / -

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

Page 36 of 50

0.5

0.6 0.7 0.8 Mole fraction model / -

0.9

1

0

0.02

0.04 0.06 0.08 Mole fraction model / -

= 633 K = 653 K = 673 K = 693 K = 703 K = 713 K = 723 K = 733 K 0.1

Figure 8: Parity plots for ethene, (a), propene, (b), butenes, (c), pentenes, (d), and C6 to C12 olens, (e), resulting from an application of model no. 08, see Table 5.

By comparing Figure 8e with Figure 1e, it follows that the improved description can be ascribed to the higher olens. The former systematic deviation disappeared, which speaks for a more realistic implementation of their reactivity. For these frequently branched compounds, the parameters starting from a tertiary intermediate are of crucial importance and should thus be discussed here. In contrast to the original single-event model, parameter (t;p) shows the highest activation energy.

24

the cracking

This result is only reasonable when

assuming carbenium ions as reactive intermediates. In such a case, the cracking step (t;p) converts the most stable intermediate into the least stable one, which requires the highest activation energy. Similarly, the cracking step (t;s) should have a higher activation energy than (s;s), whereas Table 6 reveals that these two values are almost equal for model no. 8. However, a high numeric insecurity for all parameters starting from a tertiary intermediate can be observed. Indeed, the results shown in Table 6 represent a global optimization minimum, but similar to the recent study by Cnudde et al., dicult to obtain.

28

exact quantitative values are

Therefore, the results should not be overinterpreted in terms of their

36

ACS Paragon Plus Environment

Page 37 of 50 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

numeric value, but they are meaningful in relation to each other and in the right order of magnitude. The fact that the tertiary protonation enthalpy is lower than the secondary one conrms the assumption of carbenium ions being the protonated intermediates. As stated above, the exact value is not reliable, but in all simulations, the enthalpy towards secondary intermediates is around zero, whereas the one to tertiary intermediates is lower. These results are completely dierent from the protonation enthalpies calculated with the Nguyen correlations, see Figures 3a and 5a. In contrast to the latter, see Figures 6 and 7, the concentration of protonated intermediates is extremely low for the enthalpies obtained by model no. 08, hampering the estimation of exact protonation values. In addition, as shown by DFT studies,

27,28

protonation enthalpies show a variety of values, depending, among others, on

molecular structure as well as the surrounding and strength of the acid site, a fact which also aggravates obtaining an average value. In model no. 08, the carbon number dependence of protonation is preserved by calculating the entropy loss with the regular Nguyen correlations. Although these assume alkoxides as intermediates, a suitable approximation for the entropy loss due to protonation is also obtained for carbenium ions. The additional mobility of the latter is then considered by an increased pre-exponential factor. Due to the fact that all simulations lead to a protonation enthalpy to secondary intermediates of almost zero, model no. 08 is re-evaluated using this value and without estimating this parameter. The resulting model is shown as no. 09 in Table 8. Although there is one estimated parameter less, the agreement with experimental data is of similar quality. This is why no additional parity plot is shown in the Supporting Information as no dierence to Figure 8 can be observed. than zero, see Table 8.

Again, the tertiary protonation enthalpy is signicantly lower

On the other hand, the numeric insecurity for this parameter is

still observed. In model no. 10, the protonation enthalpy to tertiary intermediates is set to

−30 kJ mol−1

according to the stability dierence between secondary and tertiary carbenium

ions estimated by Thybaut et al.

39

It can be seen that the model yields maximum agreement

(no parity plot shown, results similar to Figure 8), although another tting parameter is

37

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

Page 38 of 50

removed. This clearly speaks for an optimized depiction of adsorption and therefore overall reactivity: all activation energies are in the right stability order and the numeric signicance is very high. The results conrm the ionic character of protonated tertiary intermediates. Because of the crucial improvement in describing the reactivity of especially higher olens, it is reasonable to assume that branched olens contribute to the overall reactivity to a signicant extent. Concerning secondary and primary (= ethyl) intermediates, it can be stated that no protonated intermediates of high stability are formed. but of signicantly less stability as proposed by Nguyen et al. this study, the idea of a metastable intermediate either a deprotonation to the

π -complex

27

These might be alkoxides,

34

Based on the ndings in

can be conrmed for these species, where

or a reaction via a stabilized transition state are

realistic scenarios. In contrast to earlier proposals, no decision between carbenium ions or alkoxides as protonated intermediates for all species is possible. Their evolution highly depends on the molecular structure and conditions. This explains why a recent experimental investigation

26

mentions only alkoxides as intermediates: linear pentenes were investigated

here and lower temperatures as well as small feed partial pressures excluded the contribution of cracking. In contrast, the entropic advantage of carbenium ions should be crucial at industrial reaction conditions. Finally, similar to Section 3.3.2, the description of dimerization is varied in model no. 11; a physisorption term for the second olen in the dimerization is introduced in Eq. 19,

rdim (m; n) = ne k˜dim (m; n) K prot (Oπi ; m)

C sat (Ov ) KLphys (Ov ) p (Ov ) Ct KLπ (Oi ) p (Oi ) P . P 1 + j KLπ (Oj ) p (Oj ) 1 + j KLphys (Oj ) p (Oj ) (19)

For

KLphys (Ov ),

the experimentally derived correlation by de Moor et al.

nation with the method by Elbro et al.

67

to obtain

C sat (Ov ).

33

is used in combi-

The values in Table 6 show

similar results compared to model no. 10 (no parity plot shown, results similar to Figure 8). By considering physisorption eects for the second reactant olen, model no. 11 oers the version of maximum consistency and physical correctness. On the other hand, the imple-

38

ACS Paragon Plus Environment

Page 39 of 50 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

mentation of physisorption implies some uncertainty concerning adsorption data so that the more detailed description is not advantageous in terms of model accuracy. However, model no. 11 should be used for extrapolation purposes because its broad validity can only be guaranteed when all underlying steps are described in the physically correct way. In summary, several conclusions can be drawn from the evaluation with dierent versions of the Nguyen correlations.

34,58,63

Firstly, the predicted high stability of the

π -complex is suit-

able to describe the overall olen cracking reactivity. Such high stability values are conrmed by both experiment

26

and DFT.

the regular Nguyen correlations

27,28

34

Secondly, the protonation enthalpy values predicted by

are comparably low, which means their application leads

to a signicant, non-negligible concentration of protonated intermediates. Thirdly, a consideration of this concentration term leads to an incorrect description of overall kinetics. Fourthly, an exact determination of protonation enthalpy values is dicult, however, the results speak for protonated intermediates of comparably low stability, thereby causing their negligible concentration on the acid sites. Fifthly, a detailed implementation of the protonation step is required nevertheless.

Here, the stability order in the simulations suggests

carbenium ions to be the reactive intermediates during olen cracking at high temperatures at least for tertiary species. Sixtly, an implementation of the lower stability of protonated intermediates leads to an optimized cracking model of maximum accuracy, minimum numeric uncertainty and vast extrapolation possibilities. As these results are signicantly dierent from hydrocracking,

23,64

they are also helpful for other processes with olen interconversion

like MTO.

4

Conclusions

A microkinetic single-event model is analyzed in terms of detailed adsorption steps. Thereby, broad insight into the reactivity of olen cracking is obtained. This allows for an assessment of important reaction pathways and intermediates. The resulting optimized model depicts

39

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

Page 40 of 50

all elementary reactions in a physically consistent way and is therefore highly suitable for calculating industrial applications. A huge variety of approaches to describe the adsorption of hydrocarbons on zeolites is found in literature, comprising physisorption,

π -complex

formation and protonation.

An

analysis of the original single-event model yields considerable amounts of protonated intermediates on the catalytic surface.

However, the agreement with experimental data is

signicantly lower when accounting for the corresponding concentration term. This observation is caused by an overestimation of the stability of the protonated intermediates by the underlying equations. However, the assumption of the

π -complex

as crucial intermediate is

advantageous in these models. In contrast, less accuracy is obtained when implementing alkane physisorption approaches. At least, this method provides the opportunity to include experimental adsorption data. On the other hand, for complex reaction networks with many isomers, extrapolations and estimations are inevitable for species for which no measured data is available. In addition, the agreement with kinetic data is considerably higher by implementing the

π -complex,

which

implies this approach realistically depicts the ongoing elementary reactions. An exact calculation of the protonation enthalpies is dicult, especially for the tertiary intermediates. Nevertheless, the results speak for comparably low stability and thus a negligible concentration of the protonated intermediates. The protonation enthalpy to secondary species is close to zero; their distinct nature is thus hard to assess.

The value of tertiary

species is lower, which speaks for carbenium ions as reaction intermediates for these species. By introducing a realistic stability dierence between secondary and tertiary intermediates, which is found in literature and within the bandwidth of the tted results, a model of high accuracy and physical consistency is obtained. This can be used to describe the complex olen interconversion chemistry, even at conditions beyond the experimentally covered regimes.

40

ACS Paragon Plus Environment

Page 41 of 50 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

Acknowledgement Special thanks go to Jonas Letica for his help in evaluating microkinetic modeling approaches. This research was funded by the Bavarian Ministry of Economic Aairs, Energy and Technology grant number [47-3665g/1075/1-NW-1501-0003]. The authors gratefully acknowledge the fruitful environment within the framework of MuniCat. Helpful discussions with Professor Johannes A. Lercher, Dr. Maricruz Sanchez-Sanchez, Dr. Yue Liu and Felix M. Kirchberger are highly appreciated. S. Standl is thankful for the support from TUM Graduate School.

Supporting Information Available Description of cracking reactivity, dierent reaction networks and adsorption models as well as additional parity plots and an evaluation of adsorption intermediate concentrations as function of catalyst mass.

References (1) Blay, V.; Epelde, E.; Miravalles, R.; Perea, L. A. Converting Olens to Propene: Ethene to Propene and Olen Cracking.

Catal. Rev.: Sci. Eng.

2018, 60, 278335.

(2) Torres Galvis, H. M.; de Jong, K. P. Catalysts for Production of Lower Olens from Synthesis Gas: A Review.

ACS Catal.

2013, 3, 21302149.

(3) Ren, T.; Patel, M. K.; Blok, K. Olens from Conventional and Heavy Feedstocks: Energy Use in Steam Cracking and Alternative Processes.

Energy

2006, 31, 425451.

(4) Neelis, M.; Patel, M. K.; Blok, K.; Haije, W.; Bach, P. Approximation of Theoretical Energy-Saving Potentials for the Petrochemical Industry using Energy Balances for 68 Key Processes.

Energy

2007, 32, 11041123. 41

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

Page 42 of 50

(5) Mokrani, T.; Scurrell, M. Gas Conversion to Liquid Fuels and Chemicals: The Methanol RouteCatalysis and Processes Development.

Catal. Rev.: Sci. Eng.

2009, 51, 1145.

(6) Centi, G.; Iaquaniello, G.; Perathoner, S. Can We Aord to Waste Carbon Dioxide? Carbon Dioxide as a Valuable Source of Carbon for the Production of Light Olens.

ChemSusChem

2011, 4, 12651273.

(7) Nesterenko, N.; Aguilhon, J.; Bodart, P.; Minoux, D.; Dath, J.-P. In

Zeolite-Like Materials ;

Zeolites and

Sels, B. F., Kustov, L. M., Eds.; Elsevier: Amsterdam, NLD,

2016; pp 189263.

(8) von Aretin, T.; Standl, S.; Tonigold, M.; Hinrichsen, O. Optimization of the Product Spectrum for 1-Pentene Cracking on ZSM-5 Using Single-Event Methodology. Part 1: Two-Zone reactor.

Chem. Eng. J.

2017, 309, 886897.

(9) von Aretin, T.; Standl, S.; Tonigold, M.; Hinrichsen, O. Optimization of the Product Spectrum for 1-Pentene Cracking on ZSM-5 Using Single-Event Mmethodology. Part 2: Recycle Reactor.

Chem. Eng. J.

2017, 309, 873885.

(10) Sundberg, J.; Standl, S.; von Aretin, T.; Tonigold, M.; Rehfeldt, S.; Hinrichsen, O.; Klein, H. Optimal Process for Catalytic Cracking of Higher Olens on ZSM-5.

Eng. J.

Chem.

2018, 348, 8494.

(11) Huang, X.; Aihemaitijiang, D.; Xiao, W.-D. Reaction Pathway and Kinetics of C3 C7 Olen Transformation over High-silicon HZSM-5 Zeolite at 400490 °C.

Chem. Eng. J.

2015, 280, 222232. (12) Ying, L.; Zhu, J.; Cheng, Y.; Wang, L.; Li, X. Kinetic Modeling of C2 C7 Olens Interconversion over ZSM-5 Catalyst.

J. Ind. Eng. Chem.

2016, 33, 8090.

(13) Epelde, E.; Aguayo, A. T.; Olazar, M.; Bilbao, J.; Gayubo, A. G. Kinetic Model for

42

ACS Paragon Plus Environment

Page 43 of 50 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

the Transformation of 1-Butene on a K-Modied HZSM-5 Catalyst.

Res.

Ind. Eng. Chem.

2014, 53, 1059910607.

(14) Borges, P.; Ramos Pinto, R.; Lemos, A.; Lemos, F.; Védrine, J. C.; Derouane, E. G.; Ramôa Ribeiro, F. Light Olen Transformation over ZSM-5 Zeolites: A Kinetic Model for Olen Consumption.

Appl. Catal., A

2007, 324, 2029.

(15) Oliveira, P.; Borges, P.; Ramos Pinto, R.; Lemos, A.; Lemos, F.; Védrine, J. C.; Ramôa Ribeiro, F. Light Olen Transformation over ZSM-5 Zeolites with Dierent Acid Strengths - A Kinetic Model.

Appl. Catal., A

2010, 384, 177185.

(16) Zhou, H.; Wang, Y.; Wei, F.; Wang, D.; Wang, Z. Kinetics of the Reactions of the Light Alkenes over SAPO-34.

Appl. Catal., A

2008, 348, 135141.

(17) Standl, S.; Hinrichsen, O. Kinetic Modeling of Catalytic Olen Cracking and Methanolto-Olens (MTO) over Zeolites: A Review.

(18) Vynckier, E.; Froment, G. F. In

nent Mixtures ;

Catalysts

2018, 8, 626.

Kinetic and Thermodynamic Lumping of Multicompo-

Astarita, G., Sandler, S. I., Eds.; Elsevier: Amsterdam, NLD, 1991; pp

131161.

(19) Kumar, P.; Thybaut, J. W.; Svelle, S.; Olsbye, U.; Marin, G. B. Single-Event Microkinetics for Methanol to Olens on H-ZSM-5.

Ind. Eng. Chem. Res. 2013, 52, 14911507.

(20) Thybaut, J. W.; Marin, G. B. Single-Event MicroKinetics: Catalyst Design for Complex Reaction Networks.

J. Catal.

2013, 308, 352362.

(21) Dumesic, J. A.; Rudd, D. F.; Aparicio, L. M.; Rekoske, J. E.; Treviño, A. A.

Microkinetics of Heterogeneous Catalysis ;

The

ACS professional reference book; American

Chemical Society: Washington, D.C., USA, 1993.

(22) Feng, W.; Vynckier, E.; Froment, G. F. Single-Event Kinetics of Catalytic Cracking.

Ind. Eng. Chem. Res.

1993, 32, 29973005. 43

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

Page 44 of 50

(23) Thybaut, J. W.; Marin, G. B.; Baron, G. V.; Jacobs, P. A.; Martens, J. A. Alkene Protonation Enthalpy Determination from Fundamental Kinetic Modeling of Alkane Hydroconversion on Pt/H-(US)Y-Zeolite.

J. Catal.

2001, 202, 324339.

(24) von Aretin, T.; Schallmoser, S.; Standl, S.; Tonigold, M.; Lercher, J. A.; Hinrichsen, O. Single-Event Kinetic Model for 1-Pentene Cracking on ZSM-5.

Ind. Eng. Chem. Res.

2015, 54, 1179211803. (25) Standl, S.; Tonigold, M.; Hinrichsen, O. Single-Event Kinetic Modeling of Olen Cracking on ZSM-5: Proof of Feed Independence.

Ind. Eng. Chem. Res.

2017,

56,

13096

13108.

(26) Schallmoser, S.; Haller, G. L.; Sanchez-Sanchez, M.; Lercher, J. A. Role of Spatial Constraints of Brønsted Acid Sites for Adsorption and Surface Reactions of Linear Pentenes.

J. Am. Chem. Soc.

2017, 139, 86468652.

(27) Hajek, J.; van der Mynsbrugge, J.; de Wispelaere, K.; Cnudde, P.; Vanduyfhuys, L.; Waroquier, M.; van Speybroeck, V. On the Stability and Nature of Adsorbed Pentene in Brønsted Acid Zeolite H-ZSM-5 at 323 K.

J. Catal.

2016, 340, 227235.

(28) Cnudde, P.; de Wispelaere, K.; van der Mynsbrugge, J.; Waroquier, M.; van Speybroeck, V. Eect of Temperature and Branching on the Nature and Stability of Alkene Cracking Intermediates in H-ZSM-5.

J. Catal.

2017, 345, 5369.

(29) Abbot, J.; Wojciechowski, B. W. Catalytic Cracking and Skeletal Isomerization of Hexene on ZSM-5 Zeolite.

Can. J. Chem. Eng.

1985, 63, 451461.

(30) Abbot, J.; Wojciechowski, B. W. The Mechanism of Catalytic Cracking of on ZSM-5 Zeolite.

Can. J. Chem. Eng.

n-

n -Alkenes

1985, 63, 462469.

(31) Tabak, S. A.; Krambeck, F. J.; Garwood, W. E. Conversion of Propylene and Butylene over ZSM-5 Catalyst.

AIChE J.

1986, 32, 15261531. 44

ACS Paragon Plus Environment

Page 45 of 50 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

(32) Garwood, W. E. In

Intrazeolite Chemistry ;

Stucky, G. D., Dwyer, F. G., Eds.; ACS

Symposium Series; American Chemical Society: Washington, D.C., USA, 1983; Vol. 218; pp 383396.

(33) de Moor, B. A.; Reyniers, M.-F.; Gobin, O. C.; Lercher, J. A.; Marin, G. B. Adsorption of C2-C8

(34) Nguyen, tion

and

n -Alkanes C.

M.;

in Zeolites.

de

Moor,

Chemisorption

of

2011, 115, 12041219.

J. Phys. Chem. C

B.

A.;

Linear

Pot(MP2//B3LYP:GULP)-Statistical

Reyniers, Alkenes

M.-F.; in

Marin,

Zeolites:

Thermodynamics

Study.

G.

A

B.

Physisorp-

Combined

QM-

J. Phys. Chem. C

2011, 115, 2383123847. (35) Eder, F.; Lercher, J. A. Alkane Sorption in Molecular Sieves:

The Contribution of

Ordering, Intermolecular Interactions, and Sorption on Brønsted Acid Sites.

Zeolites

1997, 18, 7581. (36) Denayer, J. F.; Baron, G. V.; Martens, J. A.; Jacobs, P. A. Chromatographic Study of Adsorption of

102,

n -Alkanes

on Zeolites at High Temperatures.

J. Phys. Chem. B

1998,

30773081.

(37) Martens, G. G.; Marin, G. B. Kinetics for Hydrocracking Based on Structural Classes: Model Development and Applications.

AIChE J.

2001, 47, 16071622.

(38) Narasimhan, C. S. L.; Thybaut, J. W.; Marin, G. B.; Jacobs, P. A.; Martens, J. A.; Denayer, J. F.; Baron, G. V. Kinetic Modeling of Pore Mouth Catalysis in the Hydroconversion of

n -Octane

on Pt-H-ZSM-22.

J. Catal.

2003, 220, 399413.

(39) Thybaut, J. W.; Narasimhan, C. S. L.; Marin, G. B.; Denayer, J. F.; Baron, G. V.; Jacobs, P. A.; Martens, J. A. Alkylcarbenium Ion Concentrations in Zeolite Pores During Octane Hydrocracking on Pt/H-USY Zeolite.

45

ACS Paragon Plus Environment

Catal. Lett.

2004, 94, 8188.

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

Page 46 of 50

(40) Kazansky, V. B.; Frash, M. V.; van Santen, R. A. Quantumchemical Study of the Isobutane Cracking on Zeolites.

Appl. Catal., A

1996, 146, 225247.

(41) Rigby, A. M.; Kramer, G. J.; van Santen, R. A. Mechanisms of Hydrocarbon Conversion in Zeolites: A Quantum Mechanical Study.

J. Catal.

1997, 170, 110.

(42) Haw, J. F.; Nicholas, J. B.; Xu, T.; Beck, L. W.; Ferguson, D. B. Physical Organic Chemistry of Solid Acids: Lessons from in Situ NMR and Theoretical Chemistry.

Chem. Res.

Acc.

1996, 29, 259267.

(43) Bjørgen, M.; Bonino, F.; Kolboe, S.; Lillerud, K. P.; Zecchina, A.; Bordiga, S. Spectroscopic Evidence for a Persistent Benzenium Cation in Zeolite H-Beta.

Soc.

J. Am. Chem.

2003, 125, 1586315868.

(44) Clark, L. A.; Sierka, M.; Sauer, J. Stable Mechanistically-Relevant Aromatic-Based Carbenium Ions in Zeolite Catalysts.

J. Am. Chem. Soc.

2003, 125, 21362141.

(45) Dai, W.; Wang, C.; Yi, X.; Zheng, A.; Li, L.; Wu, G.; Guan, N.; Xie, Z.; Dyballa, M.; Hunger, M. Identication of

tert -Butyl

Cations in Zeolite H-ZSM-5:

NMR Spectroscopy and DFT Calculations.

Angew. Chem., Int. Ed.

Evidence from

2015,

54,

8783

8786.

(46) Boronat, M.; Viruela, P.; Corma, A. Reaction Intermediates in Acid Catalysis by Zeolites: Prediction of the Relative Tendency to Form Alkoxides or Carbocations as a Function of Hydrocarbon Nature and Active Site Structure.

126,

J. Am. Chem. Soc.

2004,

33003309.

(47) Tuma, C.; Sauer, J. Protonated Isobutene in Zeolites:

Angew. Chem., Int. Ed.

tert -Butyl

Cation or Alkoxide?

2005, 117, 48474849.

(48) Haw, J. F. Zeolite Acid Strength and Reaction Mechanisms in Catalysis. 54315441.

46

ACS Paragon Plus Environment

2002,

4,

Page 47 of 50 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

(49) Boronat, M.; Corma, A. Are Carbenium and Carbonium Ions Reaction Intermediates in Zeolite-Catalyzed Reactions?

Appl. Catal., A

2008, 336, 210.

(50) Boronat, M.; Zicovich-Wilson, C. M.; Viruela, P.; Corma, A. Inuence of the Local Geometry of Zeolite Active Sites and Olen Size on the Stability of Alkoxide Intermediates.

J. Phys. Chem. B

2001, 105, 1116911177.

(51) Kondo, J. N.; Shao, L.; Wakabayashi, F.; Domen, K. DoubleBond Migration of an Olen without Protonated Species on H(D) Form Zeolites.

101,

J. Phys. Chem. B

1997,

93149320.

(52) Stepanov, A. G.; Arzumanov, S. S.; Luzgin, M. V.; Ernst, H.; Freude, D. In Situ Monitoring of

n -Butene Conversion on H-Ferrierite by 1 H, 2 H, and 13 C MAS NMR: Kinetics

of a Double-Bond-Shift Reaction, Hydrogen Exchange, and the

J. Catal.

13

C-Label Scrambling.

2005, 229, 243251.

(53) Sinclair, P. E.; de Vries, A.; Sherwood, P.; Catlow, R. A.; van Santen, R. A. QuantumChemical Studies of Alkene Chemisorption in Chabazite: A Comparison of Cluster and Embedded-Cluster Models.

1998, 94, 34013408.

(54) Rozanska, X.; Demuth, T.; Hutschka, F.; Hafner, J.; van Santen, R. A. A Periodic Structure Density Functional Theory Study of Propylene Chemisorption in Acidic Chabazite: Eect of Zeolite Structure Relaxation.

J. Phys. Chem. B

2002, 106, 32483254.

(55) Rozanska, X.; van Santen, R. A.; Demuth, T.; Hutschka, F.; Hafner, J. A Periodic DFT Study of Isobutene Chemisorption in Proton-Exchanged Zeolites: Dependence of Reactivity on the Zeolite Framework Structure.

J. Phys. Chem. B

2003,

107,

1309

1315.

(56) Nieminen, V.; Sierka, M.; Murzin, D. Y.; Sauer, J. Stabilities of C3 C5 Alkoxide Species Inside H-FER Zeolite: A Hybrid QM/MM Study.

47

J. Catal.

ACS Paragon Plus Environment

2005, 231, 393404.

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

(57) Clark, L. A.; Sierka, M.; Sauer, J. In

Page 48 of 50

Impact of Zeolites and other Porous Materials on

the new Technologies at the Beginning of the New Millennium ; Aiello, R., Giordano, G., Testa, F., Eds.; Studies in Surface Science and Catalysis; Elsevier: Amsterdam, NLD, 2002; Vol. 142; pp 643649.

(58) Nguyen, C. M.; de Moor, B. A.; Reyniers, M.-F.; Marin, G. B. Eect of Nanopore Dimension and Network Topology on Alkene Sorption Thermodynamics. 2nd International Workshop NAPEN, Rhodes, Greece, June 0911. 2011.

(59) Correa, R. J.; Mota, C. J. A. Theoretical Study of Protonation of Butene Isomers on Acidic Zeolite: The Relative Stability Among Primary, Secondary and Tertiary Alkoxy Intermediates.

Phys. Chem. Chem. Phys.

2002, 4, 375380.

(60) Bhan, A.; Joshi, Y. V.; Delgass, W. N.; Thomson, K. T. DFT Investigation of Alkoxide Formation from Olens in H-ZSM-5.

J. Phys. Chem. B

2003, 107, 1047610487.

(61) International Union of Pure and Applied Chemistry Ed.

Terminology: Gold Book,

Compendium of Chemical

2nd ed.; IUPAC, 2014.

(62) Denayer, J. F.; Souverijns, W.; Jacobs, P. A.; Martens, J. A.; Baron, G. V. HighTemperature Low-Pressure Adsorption of Branched C5 C8 Alkanes on Zeolite Beta, ZSM-5, ZSM-22, Zeolite Y, and Mordenite.

J. Phys. Chem. B

1998, 102, 45884597.

(63) Nguyen, C. M.; de Moor, B. A.; Reyniers, M.-F.; Marin, G. B. Isobutene Protonation in H-FAU, H-MOR, H-ZSM-5, and H-ZSM-22.

J. Phys. Chem. C

2012, 116, 1823618249.

(64) Martens, G. G.; Marin, G. B.; Martens, J. A.; Jacobs, P. A.; Baron, G. V. A Fundamental Kinetic Model for Hydrocracking of C8 to C12 Alkanes on Pt/US-Y Zeolites.

Catal.

J.

2000, 195, 253267.

(65) Poling, B. E.; Prausnitz, J. M.; O'Connell, J. P.

The Properties of Gases and Liquids,

5th ed.; McGraw-Hill: New York, NY, USA, 2001.

48

ACS Paragon Plus Environment

Page 49 of 50 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

(66) Hankinson, R. W.; Thomson, G. H. A New Correlation for Saturated Densities of Liquids and Their Mixtures.

AIChE J.

1979, 25, 653663.

(67) Elbro, H. S.; Fredenslund, A.; Rasmussen, P. Group Contribution Method for the Prediction of Liquid Densities as a Function of Temperature for Solvents, Oligomers, and Polymers.

Ind. Eng. Chem. Res.

1991, 30, 25762582.

(68) Yaluris, G.; Rekoske, J. E.; Aparicio, L. M.; Madon, R. J.; Dumesic, J. A. Isobutane Cracking over Y-Zeolites.

J. Catal.

1995, 153, 5464.

(69) Toch, K.; Thybaut, J. W.; Vandegehuchte, B. D.; Narasimhan, C. S. L.; Domokos, L.; Marin,

G.

B.

A

Single-Event

Microkinetic

Model

for

Ethylbenzene

tion/Xylene Isomerization on Pt/H-ZSM-5 Zeolite Catalyst.

425426,

Dealkyla-

Appl. Catal., A

2012,

130144.

(70) Buchanan, J. S.; Santiesteban, J. G.; Haag, W. O. Mechanistic Considerations in AcidCatalyzed Cracking of Olens.

J. Catal.

1996, 158, 279287.

(71) Baltanas, M. A.; van Raemdonck, K. K.; Froment, G. F.; Mohedas, S. R. Fundamental Kinetic Modeling of Hydroisomerization and Hydrocracking on Noble-Metal-Loaded Faujasites. 1. Rate Parameters for Hydroisomerization.

28,

Ind. Eng. Chem. Res.

1989,

899910.

(72) Svoboda, G. D.; Vynckier, E.; Debrabandere, B.; Froment, G. F. Single-Event Rate Parameters for Paran Hydrocracking on a Pt/US-Y Zeolite.

Ind. Eng. Chem. Res.

1995, 34, 37933800. (73) Boudart, M.; Mears, D. E.; Vannice, M. A. Kinetics of Heterogeneous Catalytic Reactions.

Ind. Chim. Belge

1967, 32, 281284.

49

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

Graphical TOC Entry

Microkinetics

Adsorption Intermediates ZSM-5

50

ACS Paragon Plus Environment

Page 50 of 50