Single-Event Kinetic Modeling of Olefin Cracking on ZSM-5: Proof of

Jun 23, 2017 - However, no publication exists that proves the theoretically derived feature of single-event parameters being applicable to different o...
2 downloads 11 Views 1MB Size
Subscriber access provided by Warwick University Library

Article

Single-event Kinetic Modeling of Olefins Cracking on ZSM-5: Proof of Feed Independence Sebastian Standl, Markus Tonigold, and Olaf Hinrichsen Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.7b01344 • Publication Date (Web): 23 Jun 2017 Downloaded from http://pubs.acs.org on June 25, 2017

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 free 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 accessible to all readers and 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.

Industrial & Engineering Chemistry Research 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 45

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

Single-event Kinetic Modeling of Olefins Cracking on ZSM-5: Proof of Feed Independence Sebastian Standla,b, Markus Tonigoldc and Olaf Hinrichsena,b* a

Technische Universität München, Department of Chemistry, Lichtenbergstraße 4, D-85748

Garching b. München, Germany b

Technische Universität München, Catalysis Research Center, Ernst-Otto-Fischer-Straße 1, D-

85748 Garching b. München, Germany c

Clariant Produkte (Deutschland) GmbH, Waldheimer Straße 13, D-83052 Bruckmühl, Germany

Keywords: Single-event, Kinetic model, Olefins cracking, ZSM-5

One of the crucial advantages of single-event kinetic models is the possibility of extrapolating them to other reaction conditions which is highly interesting for catalyst design. However, no publication exists which proves the theoretically derived feature of single-event parameters being applicable to different olefins as feed though derived from kinetic experiments with only one certain feed olefin. Therefore, this work provides evidence that a single-event kinetic model for 1-pentene cracking on ZSM-5 is able to reproduce experimental results from literature with different olefins as feed as long as a consistent set of thermodynamic data is used. The model

ACS Paragon Plus Environment

1

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 45

predicts product distributions from two different kinetic studies of olefins cracking on ZSM-5 with high accuracy at all temperatures (400 – 490 °C), all feed partial pressures (47.6 – 131 mbar), for all olefins as reactant (C3 – C7) and for both water containing and water free feeds. The calculations for arbitrary olefin mixtures as feed also show excellent agreement. Consequently, the model describes intrinsic kinetics of olefin interconversion. The underlying kinetic parameters are independent of reaction conditions, feed and composition of ZSM-5 (powder or extrudate) and can be transferred to other systems without adjustment. Limitations in extrapolation emerge when the binder influences the product distribution to a significant extent, for example by altering diffusion characteristics. Finally, reproduction of literature results is also performed as function of contact time which requires an implementation of water adsorption for one of the two studies. The analysis of evolutions over contact time reveal both a catalyst and a carbon number dependence of the protonation enthalpy with the latter being independent of investigated ZSM-5 type and thus also transferable.

1. Introduction Ethene and propene are the most important building blocks for the polymer industry1. Their main production route is still via steam cracking of naphtha or other hydrocarbon feedstocks2,3 which is disadvantageous because of several reasons. First, it requires high temperatures of over 800 °C4 and therefore is the process with the highest energy demand in chemical industry1,2. Second, this energy input causes significant CO2 emissions5 which are undesired for environmental issues and governmental regulations. Centi et al.6 report that the production of 1 Mt of ethene releases 1.53 Mt of CO2 when starting from naphtha. Third, steam cracking offers almost no possibility of influencing the product distribution. This is particularly problematic since its propene to ethene ratio (P/E) is 0.4-0.6 whereas a significant increase in propene

ACS Paragon Plus Environment

2

Page 3 of 45

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

demand is observed7. The scenario of an upcoming propene supply gap favors processes with higher P/E values8. Fluid catalytic cracking (FCC) is not a suitable solution since the focus is on gasoline production with propene as byproduct2, causing P/E ratios around 1. Several alternative concepts for the production of light olefins are proposed: catalytic dehydrogenation of lower alkanes9, oxidative dehydrogenation of small alkanes10, methanol to hydrocarbons11 including methanol to olefins (MTO), methanol to propene (MTP) and methanol to gasoline (MTG) or modified Fischer-Tropsch6. Another promising approach is the cracking of higher hydrocarbons on shape-selective zeolites like ZSM-57. Here, temperatures are lower, P/E ratios are higher and the catalytic reaction allows more flexibility in feed which is important in times of fast changing crudes7. With the use of higher olefins as reactant, it is possible to exploit these usually undesired byproducts of FCC and steam cracking processes12. Since Buchanan et al.13 described the mechanistic background of acid-catalyzed olefins cracking over 20 years ago, several studies in literature have been published, mainly with butenes as feed12,14-18. A kinetic model describes the dependences of a reaction on conditions like temperature, pressure or concentration in a mathematical way19. The further advantages depend on the complexity i.e. whether the model is of lumped or microkinetic character. In lumped models, several compounds are grouped together, for example according to their carbon number or to other properties like functional groups. Another subdivision is possible here: power law models fit both the rate constant and mostly the partial reaction order. They are of pure empirical character so the estimated parameters do not have any physical meaning. For instance, the activation energy can be negative because of adsorption effects (apparent activation energy) and the partial reaction orders are not necessarily equivalent to stoichiometry20. Nevertheless, such a model might predict reliable results within the experimentally covered range of conditions. The

ACS Paragon Plus Environment

3

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 45

other class of lumped models comprises a differentiation between adsorption and kinetics. A well-known example is the Langmuir-Hinshelwood-Hougen-Watson (LHHW) formalism which contains a reaction term, a term for the driving force and an adsorption term in the denominator. When this methodology is chosen, several assumptions are usually made at the beginning, for example concerning the rate-determining step. The resulting model is able to depict more physical effects like saturation, but still, the resulting parameters might not have reasonable values20. Lumped models can be useful for reactor design whereas process optimization and catalyst design are hardly possible because of the missing ability to extrapolate out of the experimentally covered range21. For these purposes, a microkinetic model is the preferred solution. Here, each elementary reaction which is taking place on the catalytic surface is included in the network. No rate-determining step is defined; instead, some elementary reactions might be considered as quasi-equilibrated. For the reactive intermediates, the pseudo-steady-stateapproximation applies21. In a microkinetic model, the estimated parameters have a clear physical background making them independent of reaction conditions and feed22. Furthermore, the rate equations are based on actual elementary reactions which enables insight into mechanisms. Lumped models are common practice in cracking of hydrocarbons because of the complex reaction networks and the many different isomers. For olefins cracking, the most important literature studies are listed and explained in a different publication23. When microkinetic modeling is performed, the single-event approach allows keeping the number of estimated parameters in a reasonable range21,24. Examples in literature exist both for hydrocracking25-29 and for olefins cracking30. In the latter, von Aretin et al.30 create a kinetic model for 1-pentene cracking on ZSM-5. Because the single-event parameters only depend on the types of intermediates being involved, i.e. primary, secondary or tertiary, and not on the carbon number,

ACS Paragon Plus Environment

4

Page 5 of 45

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

these values should be also valid for other olefinic feeds. This feature is a prerequisite to perform reactor optimization23,31 and catalyst design since kinetic data is not available for each desired composition. Moreover, the feed independence of such a model would be advantageous for production plants where composition of the supply changes very fast. However, no publication exists where feed independence of single-event parameters for olefins cracking is proven. The objective of this work is to show that the values from von Aretin et al.30, though stemming from only 1-pentene, are applicable to other olefinic feeds. This is demonstrated by reproducing experimental data from two literature studies32,33 without any adjustment of the kinetics. The data is taken from Huang et al.32 and from Ying et al.33 who used all olefins from C3 to C7 as feed and subsequently created lumped kinetic models. Further goals of this contribution are an application of the single-event case to arbitrary olefin mixtures as feed and an incorporation of water adsorption. The overall motivation is to increase flexibility; feed independence means kinetic experiments with one key component are sufficient. Moreover, this enhances the single-event kinetic model to a flexible calculation tool on site. The implementation of water adsorption enables the transfer to MTO after accounting also for the methanol related pathways: water release is inevitable here and olefins interconversion is similar. Finally, a general assessment of the transferability to catalysts with different properties is of interest for further applications which requires a description of the experimental data as function of contact time. Summarized, this work can be seen as proof of concept whether the theoretically derived features for single-event modeling22,24,34 also pertain for olefins cracking.

2. Methods 2.1 Single-event Kinetic Model for 1-Pentene Cracking on ZSM-5

ACS Paragon Plus Environment

5

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 45

The single-event methodology is extensively described in literature; see publications from Froment’s group for fundamentals22,34-36 and Thybaut and Marin24 for an overview of applications. The derivation of the single-event kinetic model for 1-pentene cracking on ZSM-5 is also shown elsewhere30,37. Nevertheless, the most important equations of the latter case should be explained here. The number of single-events  characterizes the number of similar configurations during an elementary reaction because of intramolecular symmetries21. It is calculated via the ratio of 25 symmetry numbers of the reactant (σ ) and the transition state (σ  ) . Molecular symmetries

contribute to the rotational part of the total standard entropy36; the number of single-events resembles the change in 

. In Equation 1, this is shown for the specific case when an arbitrary

olefin O is the reacting species. 

 =  = exp   

 ∆ 





(1)

As a consequence, the remaining part of the standard entropy is free of any symmetry contributions like amount or position of side groups35,38. This step is crucial for parameter reduction: within one reaction family, the only remaining structural effects which have to be differentiated in terms of rate coefficients are the types of intermediates with their stability differences, i.e. primary, secondary or tertiary. The number of single-events is used to connect the regular rate coefficient  with the single-event rate coefficient  in Equation 235.  =  

(2)

For the specific case of 1-pentene cracking, a reaction network with olefins up to C12 was created30 using Boolean matrices22,24,34,36,37. Over 4000 different elementary reactions were considered, including the reaction families of protonation, PCP branching, methyl shift, cracking, dimerization and deprotonation30. Pretests showed that even for low conversions of 1-pentene,

ACS Paragon Plus Environment

6

Page 7 of 45

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

both double-bond and skeletal isomers of the same carbon number were quasi-equilibrated30. Thus, only the reaction families which alter the carbon number, i.e. cracking and dimerization, remain as kinetic steps, see Equations 3 and 4. With this assumption, 1585 reaction pathways have to be differentiated30. In the following, ! and  resemble the types of intermediates during the elementary step. "#$ %!; ' = #$ %!; ' ()* =  #$ %!; ' ()* 

(3)



"+, %!; ' = +, %!; ' ()* -. =  +, %!; ' ()* -. 



(4)

Both physisorption and protonation are assumed to be quasi-equilibrated. The corresponding equilibrium constants can be calculated using the statistical thermodynamics derived correlations from Nguyen et al.39,40 which are based on a DFT methodology (QM-Pot) proposed by the Sauer 20$34 lead to the concentration of protonated group41. Equations 5 and 6 show how /01  and / intermediates ()* and to the concentration of physisorbed species ( , respectively. 

 

( )* =  

20$34 6O $78 ; !9 / 2,3 6O ; O $78 9 ( / 

5* 

:;  % ' ?

( = @A ∑

(5)

(6)

C  6C 9 ?C

The calculation effort for protonation is decreased by using well-defined reference olefins for 2,3 for the each carbon number22. This requires an additional equilibrium constant / isomerization from olefin O to its reference olefin O $78 . The concentration of protonated intermediates is small compared to the total amount of acid sites so that the denominator of the Langmuir-approach can be ignored for ()* 26. Further parameter reduction is possible by 

applying thermodynamic reversibility: the kinetic parameters for dimerization are expressed via the values of the backward reaction, i.e. cracking, the adsorption equilibrium constants and a

ACS Paragon Plus Environment

7

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 45

27D . Both / 2,3 and / 27D are calculated with experimental thermodynamic equilibrium constant / thermodynamic data for species up to C642. For higher olefins, Benson’s group theory is used43,44. Finally, the kinetic model is applied to over 100 experimental data points30. The combination of ode15s to solve the one-dimensional, homogeneous plug flow reactor model with lsqnonlin to minimize the differences between simulated and experimental output in MATLAB leads to the parameters shown in the left column of Table 130. Four activation energies and one preexponential factor are sufficient to describe the microkinetics in the reaction network. Parity between model and experiment is high as the value of the sum of squared residuals (SSQ) for the mole fractions is comparably small.

Table 1. Single-event kinetic parameters for the cracking of 1-pentene on ZSM-5 resulting from the original kinetic model30 (left) and from a revised model where thermodynamic data calculated with Benson’s group theory43,44 is used for all species (right). Parameter

Original30

Revised

EF,#$ %s; p'

229.6 ± 0.9 kJ/mol

229.9 ± 0.9 kJ/mol

EF,#$ %s; s'

199.7 ± 0.9 kJ/mol

200.2 ± 0.9 kJ/mol

EF,#$ %t; s'

171.2 ± 0.8 kJ/mol

171.5 ± 0.8 kJ/mol

EF,#$ %t; p'

211.8 ± 1.4 kJ/mol

211.9 ± 1.5 kJ/mol

J#$

2.2 ± 0.3 × 1016 1/s

2.7 ± 0.4 × 1016 1/s

SSQ

0.0339

0.0350

2.2 Comparison of Kinetic Models for Olefins Cracking This work focuses on reproducing experimental results from two different literature studies for olefins cracking32,33 in order to prove feed independence of the single-event parameters shown in

ACS Paragon Plus Environment

8

Page 9 of 45

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

Table 130. The two cases from literature should be introduced here. Both Huang et al.32 and Ying et al.33 performed their kinetic studies in a plug flow reactor. Table 2 compares their experimental parameters with the ones from von Aretin et al.30. It can be seen that most of them are similar, only three differences are noteworthy: the literature studies used different feeds from C3 to C7 whereas in the single-event case, only 1-pentene was investigated. In contrast to the other two examples, Huang et al.32 performed experiments with water-containing feeds which stems from dehydration reactions of alcohols to the corresponding olefins. Water is assumed to not only dilute the feed, but also to competitively adsorb on the acid sites, thereby attenuating the overall reaction45. Consequently, it has to be considered in the kinetic equations. Finally, Ying et al.33 analyzed higher conversions than the other two studies.

Table 2. Experimental parameters for olefins cracking in the single-event case30 and in the two lumped models from literature32,33. Parameter

von Aretin et al.30

Huang et al.32

Ying et al.33

Temperature

633 – 733 K

673 – 763 K

673 – 763 K

Feed olefin

C5

C3 – C7

C3 – C7

Partial pressure 42.7 / 70.3 mbar feed

47.6 (C7) – 83.2 (C3) 131 mbar mbar

Dilution

N2 and H2O

N2

120 – 560 NmL/min

260 – 350 NmL/min

0.46

0.98

N2

Total volumetric 300 / flow rate NmL/min Maximum conversion

0.55

400

ACS Paragon Plus Environment

9

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 10 of 45

Table 3 shows that in contrast to the experimental parameters, more variety is found when comparing catalyst properties. Both literature studies use commercial ZSM-5 samples which means extrudates instead of pure powder as in the single-event case. Furthermore, the total acidity and the distribution of acid sites are different for Huang et al.32 which might be due to the higher Si/Al ratio and the use of a water-containing feed. Thus, scope of this work is not only to prove feed independence, but also to investigate whether a transfer to catalysts with different composition and properties is possible. Particularly, the influence of acid strength on the protonation properties according to Thybaut et al.25 is analyzed. It should be noted that for the single-event kinetic model30, the concentration of strong Brønsted acid sites was used which is Ct = 0.135 mmol/gcat and not the total acidity from Table 3. The latter value is shown here to ensure comparability with the literature studies.

Table 3. Properties of the catalysts used for olefins cracking experiments in the single-event model30 and in the two lumped models from literature32,33. Property

von Aretin et al.30

Huang et al.32

Ying et al.33

Si/Al ratio

90

200

103

BET surface area

454 m²/gcat

301 m²/gcat

340 m²/gcat

Total acidity

0.174 mmolNH3/gcat

0.012 mmolNH3/gcat

0.21 mmolNH3/gcat

BAS/LAS

4.27

1.35

n.a.

Powder or extrudate

Powder

Extrudate

extrudate

2.3 Reproduction of Experimental Data from Literature Ying et al.33 provide a table with all measured data points in their supporting information which can be used for this work. For Huang et al.32, no such overview was available; the

ACS Paragon Plus Environment

10

Page 11 of 45

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

experimental results have to be extracted from the figures. Their results are given in mole fractions whereas Ying et al.33 calculate the weight-related yield. In this work, the respective values from the original publications are used in order to enable comparisons. Their calculation is defined in Equations 7 and 8; both water and nitrogen are explicitly excluded. O

N = ∑

(7)

C OC

PQ, = ∑

O R

C OC RC

=

O

R

   OSTTU RSTTU

(8)

At first, the comparison between model predictions and literature results is performed over conversion to suppress the influence of contact time dependent effects like water adsorption or catalyst properties; later on, these are considered, see Section 3.4. For Huang et al.32, conversion has to be calculated according to Equation 9 with (V being the carbon number of olefin O . W = 1 − ∑

:Z [

(9)

C :ZC [C

The reaction network and the kinetic parameters from Table 1 are implemented in the singleevent kinetic model. Furthermore, experimental parameters are defined according to Table 2. A slight overpressure is assumed for both publications (-4 = 1.1 bar) to account for pressure drops in the catalyst bed. Catalyst masses and volumetric flow rates are adjusted to reach sufficient conversion. The reactor model is solved by applying ode15s from MATLAB. Because all isomerization reactions are assumed to be quasi-equilibrated, integration has to be performed only for each carbon number and not for each species, see Equation 10. \O]^ \_

= `]^

(10)

Finally, the simulated results are compared with the corresponding literature data. In section 2,3 and / 27D are obtained with thermodynamic data, 2.1, it was already described that both /

ACS Paragon Plus Environment

11

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 12 of 45

partly from experiments42 and partly from calculations43,44. However, a consistent set of thermodynamic data improves the results for feed independence as shown in Section 3.1. The exclusive use of the results from Benson’s group theory43,44 slightly influences the kinetic parameters. The new values after removing all experimental thermodynamic data42 are shown in the right column of Table 1.

3. Results 3.1 Simulation of Different Olefinic Feeds with the Original Model In a first step, results from Huang et al.32 and Ying et al.33 were reproduced according to Section 2.3. Figure 1 shows the simulated results of the original single-event kinetic model30 for pentenes as feed (lines) and the corresponding experimental data points. For both author groups, only the minimum and maximum temperatures of 400 and 490 °C are shown. The remaining plots can be found in the Supporting Information as Figures S1 and S2.

ACS Paragon Plus Environment

12

Page 13 of 45

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

Figure 1. Simulation of experimental data points (symbols) from Huang et al.32 (top) and Ying et al.33 (bottom) with pentenes as feed using the original single-event kinetic model30 (lines). Figure 1 reveals high parity between simulated and measured results. Thus, a transfer of the model to other systems is possible, a prerequisite for upscaling and reactor design. The conversion range of Ying et al.33 is significantly broader compared to Huang et al.32. Nevertheless, no increasing deviation of the modeled results with higher conversion is observable. However, Figure 1 shows that the theoretical results are slightly more accurate for the higher temperature of 490 °C, especially for Ying et al.33. Subsequently, the kinetic model is transferred to feeds different from pentenes, starting with butenes in Figure 2. Again, the plots for the remaining temperatures can be found in the Supporting Information, see Figures S3 and S4.

ACS Paragon Plus Environment

13

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 45

Figure 2. Simulation of experimental data points (symbols) from Huang et al.32 (top) and Ying et al.33 (bottom) with butenes as feed using the original single-event kinetic model30 (lines). The graphs in Figure 2 verify that the simulated results using microkinetic parameters from pentene cracking describe the reactivity of butenes with high accuracy. Significant conversion is achieved faster at milder temperatures since dimerization is preferred here and monomolecular cracking is not possible for butenes23. Again, parity between calculations and experiments is slightly lower at these conditions than at 490 °C. Moreover, agreement in Figure 2 is higher for Huang et al.32 compared to Ying et al.33. In the latter case, the kinetic model predicts an intense increase in ethene formation from conversions of 0.62 (left) and 0.55 (right) on which is not confirmed by experimental data. This can be explained with the measurements for the single-

ACS Paragon Plus Environment

14

Page 15 of 45

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

event kinetic model where conversion never exceeded 0.5530. Such a constraint might impair the predictions when the main part of the feed is consumed. Furthermore, the theoretical reaction network of the single-event case30 allows ethene formation out of all species from C5 to C12 whereas the studies from literature reduce its production to C5 and C6 compounds32,33. A common assumption of all three models30,32,33 is the irreversible ethene formation. Besides the minor uncertainties discussed here, predictions from the theoretical model are reliable also for butenes as feed. Figure 3 proves that this is not observed when using hexenes as reactant; Figures S5 and S6 depict the remaining temperatures.

Figure 3. Simulation of experimental data points (symbols) from Huang et al.32 (left) and Ying et al.33 (right) with hexenes as feed using the original single-event kinetic model30 (lines). The significant deviation from experimental results in Figure 3 is obvious. The calculations overestimate monomolecular cracking of hexenes for both literature studies. Therefore, propene amounts are too high whereas fractions of secondary cracking products like butenes and pentenes are too low. The comparison between both plots underlines that this trend is observable from the beginning on and does not evolve from a specific conversion level. While the predictions for

ACS Paragon Plus Environment

15

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 45

hexenes would be still acceptable, the model fails when it should describe the results with propene as reactant, see Figure 4 and also Figures S7 and S8.

Figure 4. Simulation of experimental data points (symbols) from Huang et al.32 (left) and Ying et al.33 (right) with propene as feed using the original single-event kinetic model30 (lines). For this case, the calculations predict hexene to be the most abundant compound over a broad range of conversion while the measurements from both literature studies observe butenes as main product. Nevertheless, the remaining species are predicted in a right way. The results from Figures 3 and 4 can be seen as a hint that the equilibrium between propene and hexenes is described in an inaccurate way. This theory is further supported by the higher deviation for lower 27D is temperatures where influence of the dimerization and therefore of the equilibrium constant / dominant. Its calculation between 2 C3 ↔ C6 is exceptional compared to all other possibilities in the model, for example to 2 C4 ↔ C8 or C3 + C5 ↔ C8. The reason is that the propene/hexene equilibrium relies completely on experimental thermodynamic data taken from Alberty and Gehrig42 whereas all other equilibria include a mixture of experimental42 and calculated values, the latter resulting from Benson’s group theory43,44. Figure S9 in the supporting information shows that the differences in ∆8 a between both methodologies are sufficiently high to distort

ACS Paragon Plus Environment

16

Page 17 of 45

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 equilibrium constants, especially for the C6 reference olefin. As a consequence, the single-event kinetic model30 is recalculated without the experimental thermodynamic data from Alberty and Gehrig42 and with all species being calculated with Benson’s group theory43,44. As expected, this procedure does not have crucial influence on the estimated parameters, see Table 1 (right). Nevertheless, it improves transferability to other feeds as shown in Section 3.2. Inconsistent thermodynamic data from two different references as in the original version30 causes problems during extrapolation which is why only the revised version is used in the following.

3.2 Simulation of Different Olefinic Feeds with the Revised Model The experimental data from literature is reproduced with exactly the same methodology as before and the only difference being the implementation of the revised model from Table 1 (right). Figure 5 depicts the resulting plots for all feeds compared with Huang et al.’s32 measurements at 400 and 490 °C; see Figure S10 for the remaining temperatures. Heptenes as feed are shown here for the first time since their results are comparable when using the original model.

ACS Paragon Plus Environment

17

Industrial & Engineering Chemistry Research

Page 18 of 45

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 ACS Paragon Plus Environment

18

Page 19 of 45

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

Figure 5. Simulation of experimental data points (symbols) from Huang et al.32 with feeds ranging from C3 (top) to C7 (bottom) at the lowest (left) and highest (right) temperature investigated using the revised single-event kinetic model (lines). In Figure 5, the high parity for all feeds, all temperatures and all conversion levels is noteworthy. In particular, when considering that Huang et al.32 studied water-containing feeds in contrast to the single-event case30. The fact that the kinetic model predicts right results shows that water affects not the selectivity, but only the overall rate by competing adsorption45. Figure 5 proves that the single-event kinetic parameters which were estimated with 1-pentene30 are independent of feed, system and reaction conditions. Figure 6 shows similar plots for the results of Ying et al.33 in order to further confirm this statement; Figure S11 contains the remaining temperatures.

ACS Paragon Plus Environment

19

Industrial & Engineering Chemistry Research

Page 20 of 45

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 ACS Paragon Plus Environment

20

Page 21 of 45

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

Figure 6. Simulation of experimental data points (symbols) from Ying et al.33 with feeds ranging from C3 (top) to C7 (bottom) at the lowest (left) and highest (right) temperature investigated using the revised single-event kinetic model (lines). Again, reproduction of experimental data is successful for all feeds, but deviation is higher for Ying et al.33 compared to Figure 5. This might be caused by several reasons. In contrast to the single-event kinetic model30, Ying et al.33 consider paraffins and aromatics as side products both in their measurements and in their model. Whereas any information about this issue is missing in the study from Huang et al.32, a minor fraction of these byproducts was detected during the experiments for the single-event case31. Nevertheless, to reduce calculation effort for the complex reaction network, it was decided to ignore side reaction pathways and to add the amount

ACS Paragon Plus Environment

21

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 45

of byproducts to the olefins with the same carbon number30. It follows that the molar flow rate of pentenes and especially of hexenes contains a small additional fraction of side products. This might explain the model’s tendency of producing too many hexenes at 400 °C for propene as reactant (Figure 6) or the overestimation of pentenes for C4 feeds (Figures 5 and 6). On the other hand, also the kinetic model by Ying et al.33 overestimates hexene production and underestimates butene formation for C3 feeds which could be a hint for small offsets in the measurements. Moreover, the influence of missing side products should be more significant at higher conversions whereas the differences for 400 °C and propene as feed become smaller for increasing contact times. However, since the measured data in Figure 6 goes up to conversions of 0.95, an influence of the missing side products cannot be excluded. Another possible reason for deviations might be slightly different partial pressures of the feed olefin during the measurements because this parameter significantly influences the ratio between cracking and dimerization. The fact that simulations are of excellent accuracy for Huang et al.32 speak for an influence of experimental parameters or of catalyst related properties, see Section 3.4 and Chapter 4. Nevertheless, the predictions of the single-event kinetic model are right in their tendencies for all feeds of Ying et al.33 and beyond that very accurate for propene, butenes and pentenes as reactants. Thus, with these two case studies, the feed independence of the singleevent parameters derived from 1-pentene30 could be demonstrated.

3.3 Simulation of a Mixture of Olefins as Feed with the Revised Model It is still questionable whether the transfer to an arbitrary mixture of olefins is possible. This scenario is of interest because higher olefins are mostly undesired and recycled therefore23. A well-defined single feed setup like in the kinetic measurements is not the usual case for an

ACS Paragon Plus Environment

22

Page 23 of 45

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

industrial application. The single-event kinetic model is used to simulate the experimental results from Huang et al.32 who analyzed the cracking of an olefinic mixture ranging from C2 to C7 with a mole fraction of 0.07/0.235/0.22/0.235/0.12/0.12. Figure 7 shows the results when applying the revised kinetic model on this feed mixture and the comparison with experimental data32.

Figure 7. Simulation of experimental data points (symbols) from Huang et al.32 for a mixture of olefins as feed using the revised single-event kinetic model (lines). The plots show excellent agreement for all olefins and for both temperatures. The slight underprediction of butenes for the higher temperature of 480 °C is also observed in Huang et al.’s publication32. Besides this, no deviation is noticeable. This means although the parameters have been determined with only one key component, the resulting model is able to describe the characteristic reaction pathways for a mixture of different olefins. This clearly underlines the advantages of microkinetic modeling combined with the single-event approach. Instead of performing many experiments with different olefins, one detailed experimental schedule with one component is enough30. Of course, the number of experiments has to be comparably high here which might compensate the advantage in time. However, since the feeds in industry

ACS Paragon Plus Environment

23

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 24 of 45

become more and more complex, the additional flexibility and extrapolation possibility with the single-event model are crucial advantages.

3.4 Simulation of Literature Data as Function of Contact Time So far, it is shown that a single-event kinetic model derived from 1-pentene cracking experiments yields kinetic parameters which predict the reactivity of different pure or mixed olefins in a right way. This is analyzed as function of conversion on two different catalysts32,33, meaning the model calculates the right selectivities for a certain reaction progress. However, the two catalysts vary from each other and also from the one for which the kinetic model was developed30. They differ in number of acid sites and acid strength. This should influence activity whereas selectivity remains unchanged as shown above. The latter fact proves that all nonequilibrated steps from the reaction network are influenced by the change of acid strength to the same extent25. Consequently, both activation energies and preexponential factor are independent of this catalyst property. These are the kinetic descriptors of olefins cracking which have to be clearly separated from the catalyst descriptors according to the Ghent group24,46-48. Whereas kinetic descriptors are constant for each of the three investigated ZSM-5 samples, catalyst descriptors like number of acid sites, adsorption values and shape selectivity have to be adapted individually. The number of acid sites is adjusted to the corresponding values in Table 3 in order to reproduce the literature results as function of contact time; the kinetic descriptors from Table 1 (right) remain unchanged. However, neither for Huang et al.32 nor for Ying et al.33, simulations yield satisfying results. This can be attributed to the different acid strength of the sites. For example, the catalyst of Huang et al.32 had only one tenth of acid sites compared to the single-

ACS Paragon Plus Environment

24

Page 25 of 45

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

event case30. On the other hand, these remaining sites have higher acid strength which again increases activity. In the following, the methodology of Thybaut et al.25 is applied to account for this effect. Here, the change in activity between two catalysts because of different acid strength is ascribed to a change in protonation enthalpy ∆b73 0$34 c. Thus, the single-event kinetic model is applied to the experimental results from both author groups32,33 with the kinetic descriptors held constant and with ∆b73 0$34 c being the only estimated variable. This is performed only for the data with C5 as feed to avoid any influences of the carbon number, see below. For Huang et al.’s32 data, water adsorption has to be taken into account which causes two additional fitting parameters, see Equation (11). $78 /F\,de = /F\,de exp f

g∆hU ijk @ 

@

l − l mTn o

(11)

The reference temperature was set to 718 K as medium temperature of Huang et al.’s32 $78 experiments. This leads to a reference adsorption constant /F\,de of 2.9 x 104 1/bar whereas

the adsorption enthalpy of water ∆F\ cde is determined to -56.1 kJ/mol. Furthermore, estimations yield a ∆b73 0$34 c of -59.2 kJ/mol. Figure 8 (left) depicts the results of the single-event kinetic model after including ∆b73 0$34 c and water adsorption, here for a pentene feed and 400 °C as example. Excluding water adsorption yields lower parity.

ACS Paragon Plus Environment

25

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 45

Figure 8. Reproduction of literature data as function of contact time including water adsorption and the change in protonation enthalpy caused by different acid strength of the catalyst for Huang et al.32 (left) and with the adjusted protonation enthalpy for Ying et al.33 (right). For the data of Ying et al.33, only ∆b73 0$34 c has to be estimated since water was absent during their measurements; a value of 6.6 kJ/mol is obtained. Figure 8 (right) underlines that simulations of pentene cracking at 400 °C are close to the measurements when accounting for the change in acid strength. 32 The low value of ∆b73 0$34 c for Huang et al. is noticeable. This resembles a significant increase

in acid strength which is a common phenomenon for zeolites with high Si/Al values and therefore small number of acid sites25. In contrast to this, the catalyst used by Ying et al.33 is similar to the one in the single-event case30 leading to a ∆b73 0$34 c close to zero. Nevertheless, the acid strength of the latter one is slightly higher because of the lower number of sites. Figure 9 (left) illustrates these effects.

Figure 9. Change in protonation enthalpy when switching to experimental data of pentene cracking on different zeolites (left) and when changing the feed olefin with the zeolite effect already considered (right).

ACS Paragon Plus Environment

26

Page 27 of 45

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

Although the different acid strength of the catalysts is considered, simulations for the other feed olefins as function of contact time are not successful. Again, this can be attributed to the protonation enthalpy which shows a carbon number dependence according to Thybaut et al.25. In the single-event kinetic model, this property is calculated for each olefin. However, the correlations given by Nguyen et al.39,40 yield constant protonation enthalpy values for each carbon number. Thus, the single-event case is fitted separately for each feed and catalyst to the literature data32,33 with constant kinetic descriptors, constant water adsorption (if applicable) and a constant ∆b73 0$34 c value. This leads to the change in protonation enthalpy for each carbon number ∆:Z 0$34 c as estimated value. The results can be seen in Figure 9 (right). Thybaut et al.25 investigated different Pt/H-(US)Y-zeolites for hydrocracking and observed that ∆:Z 0$34 c is independent of the zeolite type. Figure 9 (right) shows that this can be confirmed for ZSM-5 and olefins cracking. Consequently, when transferring the single-event kinetic model :Z to another ZSM-5 type, only the catalyst descriptor ∆b73 0$34 c has to be determined; ∆0$34 c can be

extracted from Figure 9 (right). The trend of the carbon number dependence is different from Thybaut et al.25 who received the highest increase for the lower olefins whereas the effect was almost zero for carbon numbers higher than eight. However, they analyzed hydrocracking experiments where no dimerization occurs, meaning the reactivity is restricted to the feed olefin. In contrast to this, also olefins with other carbon numbers than the feed component undergo kinetic steps in the single-event case30. So ∆:Z 0$34 c is more an averaged value which characterizes the change in protonation enthalpy for the whole reactive mixture and not only for the feed olefin.

4. Discussion

ACS Paragon Plus Environment

27

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 28 of 45

The results from the previous chapter prove that single-event parameters which are derived from the cracking of one key olefin can be used to describe reactivity of all other olefins and even of their mixtures with high accuracy. These kinetic descriptors are independent of temperature, partial pressures, feed, system or catalyst. Figure 10 shows parity plots for Huang et al.32 (left) and Ying et al.33 (right) when analyzed over conversion.

Figure 10. Parity plots for the simulation of the results from Huang et al.32 (left) and Ying et al.33 (right); the symbols show the respective compounds whereas the color marks the feed olefin (blue = propene, orange = butenes, green = pentenes, dark blue = hexenes, red = heptenes). For both cases, the high agreement is obvious. The SSQ-values based on mole fractions are 0.0335 on the left and 0.7698 on the right side, respectively. For Huang et al.32, no systematic deviation is found. On the other hand, Figure 10 (right) reveals much more scatter for Ying et al.33. Even when butenes are used as feed where parity is extremely high for Huang et al.32, a slight deviation is observable in Figure 6, especially at 400 °C. The most significant differences between predictions and measurements are obtained for hexenes and heptenes as feed. In both cases, monomolecular cracking to propene or propene and butenes, respectively, is overestimated by the kinetic model as it can be seen from Figure 10 (right). The consequence is an

ACS Paragon Plus Environment

28

Page 29 of 45

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

underestimation of secondary cracking products. Interestingly, this deteriorates at higher temperatures for hexenes as feed, but it improves at 490 °C when using heptenes. The reason for these effects cannot be resolved. The higher conversions during the measurements of C6 and C7 might be a possible explanation. However, the data points of Huang et al.32 at the highest conversions almost reach the minimum values of Ying et al.33. When comparing these points, no deviation is found for Huang et al.32. An abrupt change of the kinetic model to less exact results is not realistic. Other possible reasons like small offsets during the measurements or slightly different partial pressures are already discussed in Section 3.2. Finally, a different shape selectivity of the catalyst is also possible. In the single-event case, this factor is incorporated by excluding sterically demanding species from the reaction network30 which is transferred to the two literature studies without adjustments. This matches Huang et al.32 although they used extrudates instead of the pure ZSM-5 powder whereas for Ying et al.33, it is possible that the binder affects diffusion properties and therefore selectivity. As Ying et al.33 detect more side products compared to the other two studies and the selectivity differences deteriorate for higher feed olefins, an influence of this catalyst descriptor cannot be excluded. Nevertheless, the trends of the kinetic model are still right and exact for a broad range of conditions. For the implementation of water adsorption, the partial pressure of water is assumed to be equal to the olefin partial pressure. This implies complete dehydration which might not be realistic for all measurements. Furthermore, the adsorption parameters are estimated with a comparably small number of data points. Consequently, they should not be overinterpreted with regard to their absolute values. Nevertheless, the model behind Equation 11 is suitable since it increases parity in comparison to a version without water adsorption. Moreover, the resulting

ACS Paragon Plus Environment

29

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 45

values have a physical meaning: the adsorption enthalpy is negative and the adsorption constant at the reference temperature reveals a reasonable magnitude when assuming that physisorption reduces the translational entropy around one third46. The fact that water does not influence selectivity and therefore the kinetic descriptors allows their transfer to processes where water contents are inevitable such as MTO. Whereas the water amount might affect the ratio between methanol and dimethyl ether, the kinetic model shown here should describe the olefins interconversion, i.e. cracking and dimerization, in MTO. It is already mentioned that Thybaut et al.25 come to a similar conclusion, but observe different trends for the change in standard protonation enthalpy through carbon number. Besides the main reason for this which is the different process of paraffin hydrocracking without dimerization reactions, it should be underlined that their physisorption model is different. Besides, the results in Section 3.4 are obtained with the total number of acid sites for both literature models32,33 because of missing catalyst data whereas only the strong Brønsted acid sites are used in the single-event case30. However, a slight uncertainty in the number of acid sites has only minor effects on the results. The calculation routine for thermodynamic data according to Benson44 is a comparably old, but reliable method. Several revised versions49-51 exist in literature. The results in this work underline the importance of thermodynamic data when transferring a kinetic model to other systems. Agreement might be improved by using one of the more recent group additivity methods. Nevertheless, even when the data calculated from Benson’s approach44 causes some uncertainties, this mistake affects all species since the methodology is consistently used for all reactions.

ACS Paragon Plus Environment

30

Page 31 of 45

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 precondition of this work is that the underlying elementary reactions between the different studies are the same. An additional reason for the higher deviation of Ying et al.’s data33 might be the side product formation. In the experiments for the single-event kinetic model, formation of byproducts like paraffins and aromatics was small30,31. This is why their production pathways are not included in the model. Huang et al.32 followed a similar methodology, no side products are reported here. Ying et al.33 observe up to 5% byproducts which are missing in the reproduction of their measurements. Although the mistake is comparably small, further accuracy can definitely be achieved by implementing side product formation in the single-event kinetic model.

5. Conclusions This work proves feed independence of kinetic parameters derived from experimental data of 1-pentene cracking. The underlying single-event kinetic model can describe the cracking reactivity of all olefins and of their mixtures with high accuracy so that its use is not restricted to C5 feeds. This makes it highly suitable for extrapolation, reactor design and on site calculations where a maximum of flexibility is required. Furthermore, a transfer to other processes with similar reactivity like MTO is possible after implementing the methanol related pathways. The single-event kinetic model predicts right selectivities during reproduction of experimental data from two literature studies32,33 which reveals that the underlying kinetic descriptors, i.e. activation energies and preexponential factor, are independent of reaction conditions, feed olefin and catalyst composition. A minor constraint arises through the slightly different product selectivity for the catalyst of Ying et al.33 when using higher feed olefins which could be caused by a different diffusivity of the binder.

ACS Paragon Plus Environment

31

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 45

In contrast to this, catalyst descriptors have to be adapted for simulations depending on contact time. This work shows that adjusting the number of acid sites is not sufficient since the acid strength also changes. The latter effect can be implemented by varying the protonation enthalpy. This value does not depend only on the catalyst, but also on carbon number. Nevertheless, the change in protonation enthalpy through different feed olefins is independent of the ZSM-5 type. In summary, when the product selectivity of a new catalyst is not significantly different, only the change in protonation enthalpy related to the zeolite has to be estimated whereas all other descriptors remain unchanged. This underlines the high flexibility of the single-event kinetic model for olefins cracking.

Supporting Information The following files are available free of charge: Plots with comparisons between theoretical predictions and experimental data for the remaining temperatures and analysis of the two thermodynamic data sets. (PDF) Corresponding Author *E-mail: [email protected] Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes The authors declare no competing financial interest.

ACS Paragon Plus Environment

32

Page 33 of 45

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

Acknowledgments The authors acknowledge the financial support from Clariant Produkte (Deutschland) GmbH and fruitful discussion within the framework of MuniCat. S. Standl is thankful for the support from TUM Graduate School. Nomenclature J#$

(

( )* 

(4 (V

EF,#$ %!; '

xy77\

x]^

x

Preexponential factor for rate coefficient of cracking Concentration of physisorbed olefin O on the zeolite surface Concentration of protonated intermediate RA stemming from olefin O on the zeolite surface Total concentration of acid sites on the zeolite Carbon number of olefin O Activation energy for cracking from reactant intermediate of type ! to product intermediate of type  Molar flow rate of the feed olefin at the reactor inlet Molar flow rate of all olefin isomers O:Z with carbon number (V in the reactor Molar flow rate of olefin O in the reactor

s g@

mol kg uF4 g@

mol kg uF4 g@

mol kg uF4 g@ -

J molg@

mol s g@

mol s g@

mol s g@

ACS Paragon Plus Environment

33

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

x4

xz.# |

∆8 a}~7$4

∆8 a€73

∆F\ cde

∆:Z 0$34 c

∆b73 0$34 c

/F\.de

$78 /F\,de

/01  %O '  #$ %!; '

Total molar flow rate at the reactor inlet (including water and nitrogen) Mass flow rate of pentenes at the reactor inlet Standard Gibbs free energy of formation extracted from data set of Alberty and Gehrig42 Standard Gibbs free energy of formation calculated according to Benson

43,44

Adsorption enthalpy of water Change in standard protonation enthalpy because of different feed olefin Change in standard protonation enthalpy because of different zeolite Equilibrium constant for water adsorption Equilibrium constant for water adsorption at the reference temperature Equilibrium constant for physisorption of olefin O Regular rate coefficient

Regular rate coefficient for cracking reaction between

Page 34 of 45

mol4 s g@

kg s g@

J molg@

J molg@

J molg@

J molg@

J molg@

bar g@

bar g@

bar g@ s g@ / bar g@ s g@ s g@

ACS Paragon Plus Environment

34

Page 35 of 45

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

reactant intermediate of type ! to product intermediate of type  Regular rate coefficient for dimerization reaction between +, %!; '

reactant intermediate of type ! to product intermediate of

bar g@ s g@

type  27D /

2,3 %O ; O $78 ' /

Single-event equilibrium constant for ratio between reactants and product of dimerization

bar g@

Single-event equilibrium constant for isomerization of olefin O to its reference olefin O $78

-

Single-event equilibrium constant for protonation of 20$34 %O $78 ; !' /

reference olefin O $78 stemming from olefin O to

-

intermediate of type ! Single-event rate coefficient for cracking reaction #$ %!; '

between reactant intermediate of type ! to product

s g@

intermediate of type  

Single-event rate coefficient

s g@ / bar g@ s g@

Single-event rate coefficient for dimerization reaction +, %!; '

between reactant intermediate of type ! to product

bar g@ s g@

intermediate of type 

ACS Paragon Plus Environment

35

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

… …y77\

!



Page 36 of 45

Molar mass of olefin O

kg molg@

Molar mass of feed olefin

kg molg@

Type of protonated intermediate: tertiary, secondary or primary Type of protonated intermediate: tertiary, secondary or primary



Number of single events

-

Partial pressure of olefin O

bar

-.

Partial pressure of olefin O†

bar

-4

Total pressure in the reactor

bar

`

Universal gas constant

`]^

Net rate of production of all olefin isomers with the carbon number CN

-

J molg@ K g@

mol kg uF4 g@ s g@

Reaction rate for cracking as rate-determining step from "#$ %!; '

reactant intermediate of type ! to product intermediate of mol kg uF4 g@ s g@ type 

ACS Paragon Plus Environment

36

Page 37 of 45

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

Reaction rate for dimerization as rate-determining step "+, %!; '

from reactant intermediate of type ! to product mol kg uF4 g@ s g@ intermediate of type 

SSQ

∆ 

ˆ ˆ $78

Sum of squared residuals for mole fractions between theoretical predictions and experimental data Standard entropy difference between reactant and transition state for symmetry related part of entropy Temperature

K

Reference temperature

K

‰

Catalyst mass

W

Conversion of olefin O

Pz,

J molg@ K g@

Weight-related yield of olefin O (excluding water and

kg uF4 -

-

nitrogen)

N

Mole fraction of olefin O (excluding water and nitrogen)

-

Š

Symmetry number of olefin O

-

Š

Š)* 

Symmetry number of transition state with olefin O as

-

reactant Symmetry number of protonated intermediate RA

-

ACS Paragon Plus Environment

37

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 45

stemming from olefin O

References (1) Ren, T.; Patel, M.; Blok, K. Olefins from conventional and heavy feedstocks: Energy use in steam cracking and alternative processes. Energy 2006, 31, 425. (2) Mokrani, T.; Scurrell, M. Gas conversion to liquid fuels and chemicals: The methanol route Catalysis and processes development. Cat. Rev. - Sci. Eng. 2009, 51, 1. (3) Torres Galvis, H. M.; de Jong, K. P. Catalysts for production of lower olefins from synthesis gas: A review. ACS Catal. 2013, 3, 2130. (4) Wei, Y.; Liu, Z.; Wang, G.; Qi, Y.; Xu, L.; Xie, P.; He, Y. Production of light olefins and aromatic hydrocarbons through catalytic cracking of naphtha at lowered temperature. Stud. Surf. Sci. Catal. 2005, 158, 1223. (5) Neelis, M.; Patel, M.; Blok, K.; Haije, W.; Bach, P. Approximation of theoretical energysaving potentials for the petrochemical industry using energy balances for 68 key processes. Energy 2007, 32, 1104. (6) Centi, G.; Iaquaniello, G.; Perathoner, S. Can we afford to waste carbon dioxide? Carbon dioxide as a valuable source of carbon for the production of light olefins. ChemSusChem 2011, 4, 1265. (7) Rahimi, N.; Karimzadeh, R. Catalytic cracking of hydrocarbons over modified ZSM-5 zeolites to produce light olefins: A review. Appl. Catal. A: Gen. 2011, 398, 1.

ACS Paragon Plus Environment

38

Page 39 of 45

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

(8) Chen, J. Q.; Bozzano, A.; Glover, B.; Fuglerud, T.; Kvisle, S. Recent advancements in ethylene and propylene production using the UOP/Hydro MTO process. Catal. Today 2005, 106, 103. (9) Sattler, J. J. H. B.; Ruiz-Martinez, J.; Santillan-Jimenez, E.; Weckhuysen, B. M. Catalytic dehydrogenation of light alkanes on metals and metal oxides. Chem. Rev. 2014, 114, 10613. (10) Schlögl, R. Concepts in selective oxidation of small alkane molecules. In: Modern heterogeneous oxidation catalysis: Design, reactions and characterization; Mizuno, N., Ed.; Wiley-VCH: Weinheim, 2009. (11) Olsbye, U.; Svelle, S.; Bjørgen, M.; Beato, P.; Janssens, T.V.W.; Joensen, F.; Bordiga, S.; Lillerud, K. P. Conversion of methanol to hydrocarbons: How zeolite cavity and pore size controls product selectivity. Angew. Chem. Int. Ed. 2012, 51, 5810. (12) Arudra, P.; Bhuiyan, T. I.; Akhtar, M. N.; Aitani, A. M.; Al-Khattaf, S. S.; Hattori, H. Silicalite-1 as efficient catalyst for production of propene from 1-butene. ACS Catal. 2014, 4, 4205. (13) Buchanan, J. S.; Santiesteban, J. G.; Haag, W. O. Mechanistic considerations in acidcatalyzed cracking of olefins. J. Catal. 1996, 158, 279. (14) Epelde, E.; Gayubo, A. G.; Olazar, M.; Bilbao, J.; Aguayo, A. T. Intensifying propylene production by 1-butene transformation on a K modified HZSM-5 zeolite-catalyst. Ind. Eng. Chem. Res. 2014, 53, 4614.

ACS Paragon Plus Environment

39

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 45

(15) Lin, L.; Qiu, C.; Zhuo, Z.; Zhang, D.; Zhao, S.; Wu, H.; Liu, Y.; He, M. Acid strength controlled reaction pathways for the catalytic cracking of 1-butene to propene over ZSM-5. J. Catal. 2014, 309, 134. (16) Zhao, G.; Teng, J.; Xie, Z.; Jin, W.; Yang, W.; Chen, Q.; Tang, Y. Effect of phosphorus on HZSM-5 catalyst for C4-olefin cracking reactions to produce propylene. J. Catal. 2007, 248, 29. (17) Sazama, P.; Dĕdeček, J.; Gábová, V.; Wichterlová, B.; Spoto, G.; Bordiga, S. Effect of aluminium distribution in the framework of ZSM-5 on hydrocarbon transformation. Cracking of 1-butene. J. Catal. 2008, 254, 180. (18) Zhu, X.; Liu, S.; Song, Y.; Xu, L. Catalytic cracking of C4 alkenes to propene and ethene: Influences of zeolites pore structures and Si/Al2 ratios. Appl. Catal. A: Gen. 2005, 288, 134. (19) Marin, G. B.; Yablonsky, G. S. Kinetics of chemical reactions: Decoding Complexity. Wiley-VCH: Weinheim, 2011. (20) Dumesic, J. A.; Rudd, D. F.; Aparicio, L. M.; Rekoske, J. E.; Treviño, A. A. The microkinetics of heterogeneous catalysis. ACS: Washington, 1993. (21) Kumar, P.; Thybaut, J. W.; Svelle, S.; Olsbye, U.; Marin, G. B. Single-event microkinetics for methanol to olefins on H-ZSM-5. Ind. Eng. Chem. Res. 2013, 52, 1491. (22) Vynckier, E.; Froment, G. F. Kinetic modeling of complex catalytic processes based upon elementary steps. In: Kinetic and thermodynamic lumping of multicomponent mixtures; Astarita, G., Sandler, S. I., Eds.; Elsevier: Amsterdam, 1991.

ACS Paragon Plus Environment

40

Page 41 of 45

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

(23) 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 2: Recycle reactor. Chem. Eng. J. 2017, 309, 873. (24) Thybaut, J. W.; Marin, G. B. Single-event microkinetics: Catalyst design for complex reaction networks. J. Catal. 2013, 308, 352. (25) 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, 324. (26) 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. Catal. Lett. 2004, 94, 81. (27) 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. J. Catal. 2000, 195, 253. (28) Svoboda, G. D.; Vynckier, E.; Debrabandere, B.; Froment, G. F. Single-event rate parameters for paraffin hydrocracking on a Pt/US-Y-Zeolite. Ind. Eng. Chem. Res. 1995, 34, 3793. (29) Guillaume, D.; Valéry, E.; Verstraete, J. J.; Surla, K.; Galtier, P.; Schweich, D. Single event kinetic modelling without explicit generation of large networks: Application to hydrocracking of long paraffins. Oil Gas Sci. Techno. 2011, 66, 399.

ACS Paragon Plus Environment

41

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 45

(30) 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, 11792. (31) 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, 886. (32) Huang, X.; Aihemaitijiang, D.; Xiao, W.-D. Reaction pathway and kinetics of C3-C7 olefin transformation over high-silicon HZSM-5 zeolite at 400-490 °C. Chem. Eng. J. 2015, 280, 222. (33) Ying, L.; Zhu, J.; Cheng, Y.; Wang, L.; Li, X. Kinetic modeling of C2-C7 olefins interconversion over ZSM-5 catalyst. J. Ind. Eng. Chem. 2016, 33, 80. (34) Feng, W.; Vynckier, E.; Froment, G. F. Single-event kinetics of catalytic cracking. Ind. Eng. Chem. Res. 1993, 32, 2997. (35) 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. Ind. Eng. Chem. Res. 1989, 28, 899. (36) Froment, G. F. Single event kinetic modeling of complex catalytic processes. Cat. Rev. Sci. Eng. 2005, 47, 83. (37) von Aretin, T.; Hinrichsen, O. Single-event kinetic model for cracking and isomerization of 1-hexene on ZSM-5. Ind. Eng. Chem. Res. 2014, 53, 19460. (38) Martinis, J. M.; Froment, G. F. Alkylation on solid acids. Part 2. Single-event kinetic modeling. Ind. Eng. Chem. Res. 2006, 45, 954.

ACS Paragon Plus Environment

42

Page 43 of 45

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

(39) Nguyen, C. M.; De Moor, B. A.; Reyniers, M.-F.; Marin, G. B. Physisorption and chemisorption of linear alkenes in zeolites: A combined QM-Pot(MP2//B3LYP:GULP)statistical thermodynamics study. J. Phys. Chem. C 2011, 115, 23831. (40) Nguyen, C. M.; De Moor, B. A.; Reyniers, M.-F.; Marin, G. B. Isobutene protonation in HFAU, H-MOR, H-ZSM-5 and H-ZSM-22. J. Phys. Chem. C 2012, 116, 18236. (41) Sauer, J.; Sierka, M. Combining quantum mechanics and interatomic potential functions in ab initio studies of extended systems. J. Comp. Chem. 2000, 21, 1470. (42) Alberty, R. A.; Gehrig, C. A. Standard chemical thermodynamic properties of alkene isomer groups. J. Phys. Chem. Ref. Data 1985, 14, 803. (43) Benson, S. W.; Cruickshank, F. R.; Golden, D. M.; Haugen, G. R.; O'Neil, H. E.; Rodgers, A. S.; Shaw, R.; Walsh, R. Additivity rules for the estimation of thermochemical properties. Chem. Rev. 1969, 69, 279. (44) Benson, S. W. Thermochemical Kinetics John Wiley & Sons: New York, 1976. (45) Gayubo, A. G.; Aguayo, A. T.; Castilla, M.; Moran, A. L.; Bilbao, J. Role of water in the kinetic modeling of methanol transformation into hydrocarbons on HZSM-5 zeolite. Chem. Eng. Commun. 2004, 191, 944. (46) Toch, K.; Thybaut, J. W.; Vandegehuchte, B. D.; Narasimhan, C. S. L.; Domokos, L.; Marin, G. B. A Single-event microkinetic model for “ethylbenzene dealkylation/xylene isomerization” on Pt/H-ZSM-5 zeolite catalyst. Appl. Catal. A: Gen. 2012, 425-426, 130.

ACS Paragon Plus Environment

43

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 45

(47) Thybaut, J. W.; Choudhury, I. R.; Denayer, J. D.; Baron, G. V.; Jacobs, P. A.; Martens, J. A.; Marin, G. B. Design of optimum zeolite pore system for central hydrocracking of long-chain n-alkanes based on a single-event microkinetic model. Top. Catal. 2009, 52, 1251. (48) Choudhury, I. R.; Hayasaka, K.; Thybaut, J. W.; Laxmi Narasimhan, C. S.; Denayer, J. D.; Martens, J. A.; Marin, G. B. Pt/H-ZSM-22 hydroisomerization catalysts optimization guided by single-event microkinetic modeling. J. Catal. 2012, 290, 165. (49) Cohen, N. Revised group additivity values for enthalpies of formation (at 298 K) of carbonhydrogen and carbon-hydrogen-oxygen compounds. J. Phys. Chem. Ref. Data 1996, 25, 1411. (50) Domalski, E. S.; Hearing, E. D. Estimation of the thermodynamic properties of C-H-N-O-Shalogen compounds. J. Phys. Chem. Ref. Data 1993, 22, 805. (51) Sabbe, M. K.; Saeys, M.; Reyniers, M.-F.; Marin, G. B. Group additive values for the gas phase standard enthalpy of formation of hydrocarbons and hydrocarbon radicals. J. Phys. Chem. A 2005, 109, 7466.

ACS Paragon Plus Environment

44

Page 45 of 45

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

Table of Content  =   EF,#$ J#$

ACS Paragon Plus Environment

45