Development of a kinetic model for light alkane aromatisation over zeolite catalysts

Development of a Kinetic Model for Light Alkane Aromatisation over Zeolite Catalysts Luong Huu Nguyen A thesis submitted for the degree of Doctor of Philosophy University of Bath Depa

Development of a Kinetic Model for Light Alkane Aromatisation over

Zeolite Catalysts

Luong Huu Nguyen

A thesis submitted for the degree of Doctor of Philosophy

University of Bath Department of Chemical Engineering

July 2005

COPYRIGHT Attention is drawn to the fact that copyright of this thesis rests with its author This copy of the thesis has been supplied on condition that anyone who consults

it is understood to recognise that its copyright rests with its authors and that no quotation from the thesis and no information derived from it may be published

without the prior written consent of the author

This thesis may be made available for consultation within the University Library and may be photocopied or lent to other libraries

for the purposes of consultation

Catalyst development can be merely a search for a catalyst, but a higher objective has always been to create a catalyst based on scientific principles

Bruce E Leach

Acknowledgements

First of all, I would like to express my deeply gratitude to my parents for their continual supports during my study in the United Kingdom

Also, I am truly thankful for all the help, support, and dedication I have received from

my supervisors, Dr Dmitry B Lukyanov and Professor Stan T Kolaczkowski They have given me good advices and brought me the confidence during my time doing the PhD Without their supports, this Thesis could not be completed

I also received great aids from other staffs of the Department of Chemical Engineering, such as Tatiana Vazhnova, Sally Barker, Mac Forsyth, Serpil Awdry, Sean Rigby, Mervyn Newnes, and Robert Brain I would like to send them my appreciation for their aids Especially, my heartfelt thanks would like to be sent to Tatiana Vazhnova, Julia Turner, Chris Gilbert, and Angel Vidal Sanchez, who provided the experimental data used to estimate kinetic parameters of the model developed in this Thesis

A special mention and deepest thank to the Ministry of Education and Training of Vietnam, who sponsored my PhD studies in the United Kingdom

Last but not least, I would like to thank all my friends, who always stand by my side

to share my sadness and happiness during my time here.

Summary

This thesis is concerned with the quantitative description of the aromatisation of light alkanes over bifunctional catalysts, including coke formation steps The approach followed is based on the development of a kinetic model describing the reaction steps for the transformation of propane and n-butane over metal-modified zeolite catalysts From a review

of the literature, it is concluded that ZSM-5 zeolites modified by gallium or zinc are the preferred catalysts for this reaction Also, from the review, it was clear that there was a lack

of information about the coke formation step and data on kinetic studies were scarce, in particular, for catalysts that undergo deactivation during reaction

The model developed in this thesis contains 50 reaction steps and 65 rate constants that describe the transformation of the 13 components involved in this reaction The results of

modelling studies show that the transformation of light alkanes over H-ZSM-5 occurs via two

routes: (i) protolytic cracking of C-C and C-H bonds in the alkane molecules; and (ii) hydrogen transfer between the feed alkane and alkenes adsorbed over acid sites The protolytic cracking steps are the rate-limiting steps in the process Also, it is shown that over Ga,H-ZSM-5, the contribution of hydrogen transfer steps in the alkane conversion process is reduced considerably, and that aromatics are mainly formed over metal sites

In agreement with available experimental data, the kinetic modelling shows that catalyst deactivation during the aromatisation of light alkanes over H-ZSM-5 at 500 0C is relatively slow However, over Ga,H-ZSM-5 the deactivation occurs much faster This work also shows that coke formation occurs over both acid and metal sites, and that the metal sites are deactivated faster than the acid sites Finally, it is shown that the extent of catalyst deactivation depends on the initial conversion of the feed alkane over the fresh catalyst

Chapter 2: Literature Review and Decisions Concerning Main Features 5

of the Kinetic Model

2.1 Aromatics production and aromatisation processes in the industry 6

2.1.1 Significance of aromatic hydrocarbons in the chemical industry 6

2.2 Introduction to zeolite and active sites for the aromatisation 13

of light alkanes over zeolite catalysts

2.5 Nature of active sites of the aromatisation catalyst 29

2.5.2 Metal sites 30

2.6 Reaction pathways of the aromatisation of light alkanes 33

2.8.2 Effects of coke formation on the aromatisation reaction 47

2.8.4 Some solutions for limiting coke formation 50

2.9 Kinetic studies and development of kinetic models 52

Chapter 3: Development of the Structure of the Kinetic Model 58

for the Aromatisation of Propane and n-Butane

3.2 Reaction scheme of propane and n-butane aromatisation 62 over zeolite acid sites and metal sites

3.3.2 Alkene oligomerisation-cracking steps (over acid sites) 67

3.4 Rate equations of reaction and adsorption steps 75

3.4.2 Alkene oligomerisation-cracking steps (over acid sites) 78

Chapter 4: Experimental Reactors Used for Kinetic Studies 99

and the Development of a Model to Interpret Experimental Data

4.1.1 Introduction to experimental reactors for gas-solid reactions 100

4.1.2 Experimental reactors and catalysts used in experimental studies 102

of the aromatisation of light alkanes

4.2 Definitions of space velocity and contact time 105

4.2.2 Contact time 106

4.3 Preliminary calculations for a plug flow reactor 107

Chapter 5: Acid Sites: Estimation of Kinetic Parameters in a Model 116

Representing the Aromatisation of a Feedstock of Propane

or n-Butane over H-ZSM-5

5.1 Step1: Conversion of alkane into alkenes (fresh catalyst) 122

5.1.1 Conversion of propane including the hydrogen transfer step 122

5.1.2 Conversion of propane without hydrogen transfer step considered 130

5.1.3 Comparing the cracking of n-butane with the cracking of propane 132

5.2 Step 2: Alkene interconversion (illustrated for propane feedstock) 134

5.3.1 Alkene aromatisation scheme illustrated for propane feedstock 141

5.3.2 Discussion of results for propane and n-butane feedstocks 145

5.4 Step 4: Alkane transformation over deactivated catalyst 150

5.4.1 Effects of coke formation on the reaction 1515.4.2 Effects of coke formation on the catalyst 154

Chapter 6: Acid and Metals Sites: Estimation of Kinetic Parameters in a Model 159

Representing the Aromatisation of a Feedstock of Propane or

n-Butane over Me,H-ZSM-5

6.1 Step1: Alkane transformation over fresh catalyst 161

6.1.1 Formulation of the model considering reactions over metal sites 161

6.1.2 Discussion of results for fresh catalyst: Effects of the metal introduction 165

on the reaction and catalyst activity

6.2 Step 2: Alkane transformation over deactivated catalyst 171

6.2.1 Effects of coke formation on the reaction: the contribution 173

of acid sites and metal sites

6.2.2 Effects of coke formation on the catalyst: the contribution 175

of acid sites and metal sites

Appendix 2: Theoretical Yield of Aromatics in Alkane Aromatisation 197

Appendix 3: Useful Unit Conversions and Relationships 198

Appendix 4: Flowchart Describing the Programme Used for Kinetic Modelling 199

Appendix 7: Experimental Data on Propane Aromatisation over H-ZSM-5

List of Figures

Figure 2.1 Some applications of aromatics (BTX) in the chemical industry (Emmrich et al.,

1999)

Figure 2.2 World aromatics supply by source (Emmrich et al., 1999)

Figure 2.3 The flow scheme of the Cyclar Process (UOP LLC, 2001)

Figure 2.4 The structure of channels in H-ZSM-5 zeolite

Figure 2.5 Pore diameters of zeolites and sizes of reactant molecules (Jentys and Lercher,

2001)

Figure 2.6 Active sites as Brönsted acid sites in hydrogen zeolites (HZ)

Figure 2.7 Representation of alkylcarbenium and alkylcarbonium ions (Martens and Jacobs,

2001)

Figure 2.8 Roles of active sites over (a) acid catalysts and (b) bifunctional catalysts for the

aromatisation of light alkanes

Figure 2.9 The postulated bifunctional mechanism of the role of Ga species in the propane

activation

Figure 2.10 Pathway of propane aromatisation reaction over Ga,H-ZSM-5 (Lukyanov et al.,

1995)

Figure 2.11 Mechanism of formation of coke species: (a) formation of

dimethylnaphthalene from butene and (b) formation of 1-methylanthracene from dimethylnaphthalene (Jana and Rao, 1994)

1,4-Figure 2.12 Mechanism of the coke formation through 5-membered naphthenes as coke

precursors

Figure 3.1 Reaction scheme of alkane aromatisation over Me,H-ZSM-5 catalysts

Figure 4.1 Velocity pattern in a plug flow reactor The velocity is uniform across the

diameter Concentration and temperature are also assumed to be constants across the diameter

(Hayes, 2001)

Figure 4.2 Schematic of a fixed-bed flow tubular reactor used in experimental studies

Figure 4.3 A volume element balance for a PFR

Figure 5.1 Systematic procedure used to estimate the rate constants in a model representing

the aromatisation of a light alkane over H-ZSM-5.

Figure 5.2 Propane activation over fresh H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the concentrations (mass fractions) of C1 (♦),C2 (▼), C2= (●), and C3=

(■) as functions of WHSV-1 Experimental data produced by Sanchez (2005) were used in this Figure

Figure 5.3 Propane conversion over H-ZSM-5 at 500 0C Experimental data points and calculated curves for propane conversion as a function of WHSV-1 Experimental data produced by Sanchez (2005) were used in this Figure

Figure 5.4 Propane conversion over fresh H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the concentrations (mass fractions) of C2= (●),C3= (♦), C4= (Î), C5

(■), and C6+ (▲) as functions of WHSV-1 Experimental data produced by Sanchez (2005) were used in this Figure

Figure 5.5 (a) Propane and (b) n-Butane conversion over fresh H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the concentrations (mass fractions) of aromatics (■), alkanes (◆), alkenes (z) and hydrocarbons C5+ (▲) as functions of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 5.6 (a) Propane and (b) n-Butane conversion over fresh H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the mass fractional conversion of propane and n-butane as a function of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 5.7 Propane and n-butane aromatisation over fresh H-ZSM-5 at 500 oC Calculated curves for the conversion of propane and n-butane as a function of contact time Experimental

data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 5.8 (a) Propane and (b) n-Butane conversion over fresh H-ZSM-5 at 500 oC Calculated curves for the mass fractional conversion of propane and n-butane as a function of catalyst-bed length

Figure 5.9 (a) Propane and (b) n-Butane aromatisation over fresh H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the propane and n-butane conversion as a function of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 5.10 Propane and n-butane aromatisation over fresh H-ZSM-5 at 500 oC Calculated curves for the concentration of aromatics as a function of contact time Experimental data

produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 5.11 Propane and n-butane conversions over fresh H-ZSM-5 at 500 0C Calculated curves for the concentration of grouped alkenes C2+= as a function of WHSV-1

Figure 5.12 Propane (a) and n-butane (b) conversion over H-ZSM-5 at 5000C The dependence of propane (or n-butane) convesion upon time-on-stream at different initial conversions The curves were calculated using the deactivation rate constant of 0.1 h-1 Experimental data produced by Gilbert (2004), Vazhnova (2004), and Sanchez (2005) were used in this Figure

Figure 5.13 n-Butane conversion over H-ZSM-5 at 500 0C The dependence of concentration

of C2+= alkenes upon time-on-stream at different initial mass fractional conversions (blank

bars for calculations at X’ = 0.13 and solid bars for calculations at X’ = 0.70)

Figure 5.14 Propane (a) and n-butane (b) conversion over H-ZSM-5 at 500 0C The dependence of alpha upon WHSV-1 at different time-on-stream (TOS) The curves were calculated using the deactivation rate constant of 0.1 h-1

Figure 6.1 (a) Propane and (b) n-Butane conversion over fresh Ga,H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the concentrations (mass fractions) of aromatics (■), alkanes (◆), alkenes (●) and hydrocarbons C5+ (▲) as functions of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 6.2 (a) Propane and (b) n-Butane conversion over Ga,H-ZSM-5 at 500 0C Experimental data (points) and calculated curves for the mass fractional conversion of propane and n-butane as a function of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 6.3 (a) Propane and (b) n-Butane conversion over fresh H-ZSM-5 and Ga,H-ZSM-5

at 500 0C Calculated curves for the conversion of propane and n-butane as a function of WHSV-1

Figure 6.4 (a) Propane and (b) n-Butane aromatisation over fresh Ga,H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the propane, or n-butane, conversion as a function of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 6.5 (a) Propane and (b) n-Butane aromatisation over fresh Ga,H-ZSM-5 at 500 oC Experimental data (points) and calculated curves for the propane and n-butane conversion as a function of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were

used in this Figure

Figure 6.6 (a) Propane and (b) n-Butane aromatisation over fresh H-ZSM-5 and

Ga,H-ZSM-5 at Ga,H-ZSM-500 oC Calculated curves for C6+= concentration as a function of WHSV-1

Figure 6.7 (a) Propane and (b) n-Butane aromatisation over fresh Ga,H-ZSM-5 at 500 oC Experimental data (points) and calculated curves (solid curve for the reaction over the real catalyst with aromatisation activity over acid sites and dash curve for the reaction over the pseudo-catalyst without aromatisation activity over acid sites) for the aromatics concentrations as functions of WHSV-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 6.8 Propane (a) and n-butane (b) conversion over Ga,H-ZSM-5 at 5000C The dependence of propane and n-butane convesions upon time-on-stream at different initial conversions The curves were calculated using the deactivation rate constant over acid sites of 0.1 h-1 and that over metal sites of 0.6 h-1 Experimental data produced by Gilbert (2004) and Sanchez (2005) were used in this Figure

Figure 6.9 n-Butane conversion over Ga,H-ZSM-5 at 500 0C The dependence of concentration of C2+= alkenes upon time-on-stream at different initial mass fractional

conversions (blank bars for calculations at X’ = 0.08 and solid bars for calculations at X’ =

List of Tables

Table 2.1 The typical reaction conditions of Cyclar process (Maxwell and Stork, 2001)

Table 2.2 A summary of some zeolite-catalysed processes (Blauwhoff et al., 1999)

Table 2.3 Pore-size classification of zeolites (Fricke et al., 2000)

Table 2.4 Catalysts used in some stidies on the aromatisation of light alkanes

Table 2.5 Roles of active sites in the aromatisation of light alkanes

Table 2.6 The values of activation energy for the whole aromatisation reaction

Table 3.1 Lumped components of the kinetic model for the aromatisation of light alkanes Table 3.2 Literature data of the values of rate constants in the aromatisation reaction of light

alkanes

Table 3.3 Reaction scheme for aromatisation of a feedstock of propane or n-butane

Table 4.1 Several types of gas-solid reactors which can be used to carry out the kinetic

experiments (adapted from Berger et al 2001)

Table 4.2 Experimental fixed-bed reactors used in the aromatisation of light alkanes

Table 4.3 Results of thermodynamic calculations of the aromatisation of propane and

n-butane (based on calculations in Appendix 1)

Table 4.4 Examples of key parameters which feature in the literature used to investigate the

reaction

Table 5.1 Reaction scheme for aromatisation of propane over fresh H-ZSM-5

Table 5.2 Reaction scheme for aromatisation of n-butane over fresh H-ZSM-5

Table 5.3 The values of adsorption constants for aromatics and alkenes over acid sites used

in this model

Table 5.4 Range of values of the rate constants tested in the estimation

Table 5.5 Mass fraction of reaction products from modelling results (reaction over a real

catalyst including HT activity) and experimental data for propane activation over H-ZSM-5 at

500 0C

Table 5.6 Values of the rate constants obtained from the estimation for the activation

propane over fresh H-ZSM-5 (α = 1)

Table 5.7 Mass fraction of reaction products from modelling results (reaction over a

pseudo-catalyst without hydrogen transfer activity) and experimental data for propane activation over H-ZSM-5 at 500 0C

Table 5.8 Range of values of the rate constants tested in the estimation

Table 5.9 Values of rate constants obtained from the estimation for the activation n-butane

over fresh H-ZSM-5 (α = 1)

Table 5.10 Relative values of rate constants of protolytic cracking steps of propane and

n-butane over H-ZSM-5 at 500 0C

Table 5.11 Range of values of the rate constants tested in the estimation

Table 5.12 Mass fraction of reaction products from modelling results and experimental data

for propane transformation over H-ZSM-5 at 500 0C

Table 5.13 Values of the rate constants of propane cracking, hydrogen transfer between

propane and adsorbed alkenes, alkene oligomerisation steps and alkene cracking steps obtained from the estimation for the activation propane and alkene oligomerisation - cracking steps over fresh H-ZSM-5 (α = 1)

Table 5.14 Range of values of the rate constants tested in the estimation

Table 5.15 Mass fractional conversion and mass fraction of reaction products obtained from

modelling results for propane aromatisation (with N2 av. = 13.605 wt%) over fresh H-ZSM-5

at 500 0C

Table 5.16 Values of the rate constants obtained from the estimation for the whole process of

propane (or n-butane) aromatisation over fresh H-ZSM-5 (α = 1)

Table 5.17 Calculated selectivity of aromatics for propane and n-butane aromatisation over

H-ZSM-5 at a mass fractional conversion X’ = 0.30

Table 5.18 Illustration of how α (alpha), alkane conversion and mass fractions of reaction products vary with time-on-stream when propane aromatisation is simulated over H-ZSM-5

at 500 0C with k 41 = 0.1 h-1

Table 6.1 The values of adsorption constants for alkenes and aromatics over acid sites and

metal sites used in this model

Table 6.2 Range of values of the rate constants tested in the estimation

Table 6.3 The values of rate constants estimated for the alkane transformation steps over

fresh metal sites (α = 1)

Table 6.4 Calculated conversions of propane and n-butane over Ga,H-ZSM-5 catalysts at

500 0C and a contact time of 0.0206 h

Table 6.5 Calculated yield and selectivity of aromatics, yields of alkanes, alkenes, and

hydrogen for propane and n-butane conversion over H-ZSM-5 and Ga,H-ZSM-5 catalysts at

500 0C and a mass fractional conversion of 0.30

Table 6.6 Alkane fractional conversion of propane and n-butane over a real catalyst

Ga,H-ZSM-5 with coking activity over acid sites and over a pseudo-catalyst Ga,H-Ga,H-ZSM-5 without

coking activity over acid sites after 40 h on-stream

C5+ Pentanes and higher alkanes

C6+ Hexanes and higher alkanes

C8+ Hydrocarbons with number of carbon atoms of 8 and higher

c p constant pressure heat capacity on a molar basis (J molF-1 K-1)

c’ p constant pressure heat capacity on a mass basis (J gF-1 K-1)

D Dienes

D r Reactor diameter = catalyst-bed diameter (mR)

F i 0 Molar flowrate of component i in the feed (moli h-1)

F i Molar flowrate of component i (moli h-1)

F total Totalmolar flowrate of the fluid (molF h-1)

g Gaseous state

ΔG T Free energy of the reaction at temperature T (J moli-1)

ΔH T Enthalpy of the reaction at temperature T (J moli-1)

ΔH R Heat formation of the reaction on a molar basis (J moli-1)

ΔH’ R Heat formation of the reaction on a mass basis (J gi-1)

h Contact time step (h)

H-A Acid

H-ZSM-5 ZSM-5 zeolite in the hydrogen form

k i Rate constant of reaction step i (moli gCAT-1 atm-1 h-1)

Rate constant of the coke formation step (h-1)

k -i Rate constant of the alkene cracking step (moli gCAT-1 h-1)

Rate constant of the alkene hydrogenation step (moli gCAT-1atm-1 h-1)

K i Adsorption equilibrium constant of component i over active sites

(nOi nV-1 atm-1)

K T Equilibrium constant of the reversible reaction of alkene oligomerisation-

cracking at temperature T (atm-1)

LHSV Liquid hourly space velocity (h-1)

M i Molecular weight of component i (gi moli-1)

M Molecular weight of the mixture (gF molF-1)

Me,H-ZSM-5 Metal-modified ZSM-5 zeolite

n Valence of an atom

n Oi Number of acid/metal active sites occupied by component i

n V Number of vacant acid/metal active sites

n TOTAL Total number of available acid/metal active sites

n 0 Total number of acid/metal active sites of the catalyst

N i Mole fraction of component i (moli molF-1)

N total Total mole number of the reaction mixture (moli gF-1)

O Butenes

P i Partial pressure of component i (atm)

P total Total pressure of the reaction (atm)

Q R Amount of energy required to maintain the reaction temperature at 500 0C

(watt)

Q F Volumetric flowrate of the fluid (m3

F h-1)

Q 1 Volumetric flowrate of the feed (m3F h-1)

Q 2 Volumetric flowrate of the products (m3

F h-1)

r i Rate of reaction step i (moli gCAT-1 h-1)

Rate of the coke formation step (h-1)

R i Rate of the transformation of component i (moli gCAT-1 h-1)

R Universal gas constant

R = 8.3144 J molF-1 K-1

R = 0.082 l atm molF-1K-1

s d Sum of the squares of the differences between the values of mass fractions of

components obtained from the model and from the experimental data (gi2 gF-2)

V R Reactor volume = catalyst-bed volume (mR3)

V CAT Total volume of catalyst particles (mCAT3)

W i Mass flowrate of component I (gi h-1)

W i 0 Mass flowrate of component i in the feed (gi h-1)

W total Total mass flowrate of the fluid (gF h-1)

w i 0 Mass fraction of component i in the feed flow (gi gF-1)

w i Mass fraction of component i (gi gF-1)

w ij model Mass fraction of component i at contact time j obtained from the model

(gi gF-1)

w ij experiment Mass fraction of component i at contact time j obtained from the experimental

data (gi gF-1)

W Catalyst weight (gCAT)

WHSV Weight hourly space velocity (gF gCAT-1 h-1; h-1)

X i Mole fractional conversion of component i

X’ i Mass fractional conversion of component i

X Alkylcyclohexenes

X 1 ’ Mass fractional conversion over a real catalyst with HT activity

X 2 ’ Mass fractional conversion over a pseudo-catalyst without HT activity

y i Mole fraction of component i in the feed (moli molF-1)

Y Alkylcyclohexadienes

Z Zeolite acid site

z Reactor length = Catalyst-bed length (mR)

[Z] Fraction of vacant zeolite acid sites (nV nTOTAL-1)

[iZ] Fraction of zeolite acid sites occupied by component i (nOi nTOTAL-1)

Z m Metal site

[Zm] Fraction of vacant metal sites (nV nTOTAL-1)

[iZm] Fraction of metal sites occupied by component i (nOi nTOTAL-1)

Greek symbols

α Fraction of non-deactivated zeolite acid sites (nTOTAL n0-1)

αm Fraction of non-deactivated metal sites (nTOTAL n0-1)

νij Stoichiometric coefficient of component i in reaction step j

η Effectiveness factor

φ Bed voidage (mAIR3 mR-3)

ρB B Density of the catalyst bed (gCAT mR-3)

ρC Density of the catalyst particle (gCAT mCAT-3)

ρF Density of the feed (gF mF-3)

τ Mass-based space time or contact time (= WHSV -1) (h gCAT gF-1; h)

τf Final contact time of the reaction (h)

Subscripts

0 Relating to the values of the parameters before the reaction

B Relating catalyst-bed

C Relating to catalyst particle

CAT Relating to catalyst

d Relating to the difference between two values

f Relating to the final value

F Relating to fluid

i, j Relating to component i or j or reaction step i or j

m Relating to metal sites

O Relating to occupied active sites

R Relating to reactor size

Value at temperature T

total; TOTAL Relating to the total value

V Relating to vacant active sites

Supercripts

0 Relating to the values of the parameters before the reaction

Abbreviations

AAS Atomic absorption spectroscopy

ABS Acrylonitrile-butadiene-styrene resins

AES Atomic emission spectroscopy

BTX Benzene, toluene, and xylenes

CCR Continuous catalyst regeneration

CONCEN Subfunctions to describe the transformation rates of hydrocarbons and

hydrogen involved in the reaction CVD Chemical vapour deposition

DSC Differential scanning calorimetry

EPR Electron paramagnetic resonance

FTIR Fourier transform infrared

IR Infrared

LAB Linear alkyl benzene

LAS Linear alkyl benzene sulfonate

LPG Liquefied petroleum gas

MTA Metric tonnes per annum

MTBE Methyl tert-butyl ether

NGL Natural gas liquids

ODE Ordinal differential equation

PET Poly(ethyleneterephthalate)

PFR Plug flow reactor

SEM Scanning electron microscopy

SBR Styrene-butadiene rubber

STP Standard temperature and pressure

TEM Transmission electron microscopy

TOS Time-on-stream

TPD Temperature-programmed desorption

TPO Temperature-programmed oxidation

TPR Temperature-programmed reduction

XAS X-ray absorption spectroscopy

XPS X-ray photoelectron spectroscopy

XRD X-ray diffraction

CHAPTER 1

Introduction

Light alkanes, such as propane and n-butane, are mainly obtained from natural gas liquids (NGL); and additionally from by-products of different refinery operations (Matar and Hatch, 1994) They can be used either as a source of fuel for industrial and domestic

applications; or as a feedstock for methyl tert-butyl ether (MTBE) production (Satterfield,

1991) Since large amounts of light alkanes are readily available from various sources, conversion of these light hydrocarbons into more valuable chemicals such as aromatic hydrocarbons and branched-chain compounds is an important reaction from both the academic and industrial points of view In fact, the aromatisation of n-butane or propane expands the source of hydrocarbons and upgrades the value of the feed itself because the main reaction products, mainly benzene, toluene and xylenes (BTX), are a valuable source

of intermediates in the chemical and petrochemical industry (Matar and Hatch, 1994) Therefore, it is not surprising that a huge number of papers have concentrated on this reaction during the last two decades

Many studies on the aromatisation of light alkanes have been performed over zeolite catalysts Ga-, Zn-, and Pt-modified zeolite catalysts are the subjects of intensive studies for this transformation due to their high effectiveness in aromatisation These studies have been reviewed by Ono (1992), Guisnet and Gnep (1996), Meriaudeau and Naccache (1997), and

Fricke et al (2000) It has so far been found that the medium-pore zeolite H-ZSM-5

modified by metals is the most effective for aromatisation (Meriaudeau and Naccache, 1997) The activity and selectivity of ZSM-5 zeolite in the alkane aromatisation has been attributed mainly to its (i) high acidity and (ii) specific pore geometry (Bhattacharya and Sivasanker, 1995)

Although H-ZSM-5 zeolite shows a remarkably low rate of coke formation during aromatisation (Kanai and Kawata, 1990), the addition of metal species into the zeolite enhances not only its aromatisation activity but also its coking rate (Bayense and van

Hooff, 1991; Meriaudeau et al., 1993a; Kwak and Sachtler, 1994) There is always a

demand, especially in the industry, on producing an effective aromatisation catalyst with a low rate of coke formation Therefore, it is essential to understand the mode of coke formation as well as the mode of catalyst deactivation during the reaction, and to describe and predict the deactivation rate under different reaction conditions However, only a few studies on coke formation during the light alkane aromatisation were carried out, mainly

over gallosilicate catalysts, which have similar structures to ZSM-5 zeolites and are obtained by the isomorphous substitution of gallium for framework aluminium of the

zeolite during the synthesis (Choudhary et al., 1997a, b) Unfortunately, no kinetic

description has been reported for the deactivation of these catalysts

The bifunctional mechanism for the aromatisation of light alkanes has been accepted broadly (Meriaudeau and Naccache, 1997) Because the acid function supplied by zeolites and the dehydrogenating function supplied by metals influence the reaction intensively, a quantitative description of these effects is of key importance in understanding the behaviours of the reaction and its catalyst, especially under deactivating conditions Such information is vital for the design of catalysts with optimal selectivity, activity, and lifetime, and of the reactor as well However, analysis of the literature shows that this quantitative information is not available

The aim of this project was to develop a kinetic model to describe the major features of the aromatisation reaction, including coke formation and catalyst deactivation (no such model has been described in the literature for light alkane aromatisation), and to use this model for generation of kinetic modelling results of propane and n-butane aromatisation In future, it is planned to use the kinetic model for reactor modelling and process optimisation The overall aim of this research is to obtain further understanding of the factors behind the aromatisation reaction and their intrinsic effects on the basis of the quantitative data generated by kinetic modelling results Consequently, the project objectives were following:

• To develop the structure of a kinetic model for the aromatisation of light alkanes over a bifunctional catalyst

• To apply the model for the reaction of propane and n-butane over H-ZSM-5 and Ga,H-ZSM-5 catalysts

• To estimate kinetic parameters under reaction conditions investigated on the basis of the comparison of the modelling results with the experimental data, which have been obtained by Gilbert (2004), Vazhnova (2004), and Sanchez,

2005

• To obtain the quantitative information about the reaction and its catalyst on the basis of modelling results

This thesis is divided into seven chapters In Chapter 2, a literature review of studies

on the aromatisation of light alkanes is presented General information is also provided on the aromatics industry, zeolites, and active sites On the basis of the literature review, decisions are made on features to be included in the kinetic model developed in this work This then leads on to Chapter 3 where the development of the structure of the kinetic model for this reaction over metal-modified zeolite catalysts is described This investigation leads

to the application of assumptions that reduces the complexity of the model but still preserve the overall mechanistic description of the reaction

In Chapter 4, a mathematical model, which is then applied in subsequent chapters to simulate the performance of an experimental reactor, is developed A chemical kinetic model for the aromatisation of light alkanes over a bifunctional catalyst is then developed

in two phases, which are presented in Chapters 5 and 6, treating the acidic and metallic functions separately, making use of experimental data for the aromatisation of propane and n-butane over H-ZSM-5 and Ga,H-ZSM-5 catalysts at 500 0C, which have been obtained

by Gilbert (2004); Vazhnova (2004); and Sanchez (2005) In Chapter 5, the reaction steps over the acid sites are considered, and then their kinetic parameters are estimated In Chapter 6, the reaction steps over the metal sites are included in the model Kinetic parameters for the full model are then estimated utilising the values of kinetic parameters obtained in Chapter 5 On the basis of the modelling results obtained, the nature of the reactions and function of the catalyst is discussed Finally, in Chapter 7, the key conclusions are formulated and recommendations are made for further work

CHAPTER 2

Kinetic Model

This chapter reviews studies on the aromatisation of light alkanes described in the

literature Subjects discussed in this chapter include: aromatisation processes in the

industry; introduction to zeolite and active sites for the aromatisation of light akanes over zeolite catalysts; aromatisation catalysts; roles and nature of active sites, namely acid and metal sites, in the aromatisation; reaction pathways over acid zeolite and metal-modified zeolite catalysts; factors affecting the aromatisation reaction; coke formation in the aromatisation process; and kinetic studies and development of kinetic models for this reaction Based on these subjects, concluding remarks on this reaction are performed Also,

some decisions on the features of the kinetic model developed in this work are made on the basis of the literature review in this chapter

2.1 AROMATISATION PROCESSES IN THE INDUSTRY

2.1.1 Significance of aromatic hydrocarbons in the chemical industry

Aromatic hydrocarbons such as benzene, toluene and xylenes (BTX) are among the most important intermediate products for the chemical industry with a broad range of applications For example, benzene is an important feedstock for the production of polystyrene (PS); toluene is used as feedstock for polyurethane (PU) production; and p-xylene is employed in poly(ethyleneterephthalate) (PET) production Some applications of aromatic hydrocarbons in the chemical industry are shown in Figure 2.1 It has been reported that the total aromatics market is currently nearly 60 million tons per year

(Emmrich et al., 1999) In addition, aromatic hydrocarbons are also used as the components

to be blended in the gasoline to improve the octane number of gasoline Nowadays, although the content of aromatic hydrocarbons, especially benzene, in gasoline is limited due to strict environmental criteria, aromatisation is still an important process because of the great demand of BTX in the chemical and petrochemical industry As a result, the aromatisation processes with the high aromatics selectivity are required The forecast is that the world demand for mixed xylenes grows at average annual rates of over 5 % through

2020 (CMAI News, 2001)

Figure 2.1. Some applications of aromatics (BTX) in the chemical industry (Emmrich et al., 1999)

Ethylbenzene Styrene Polystyrene, ABS Resins, SBR

Cumene

Phenol

α-Methylstyrene

Aniline, Phenolic Resins, Surfactants

Adhesives, ABS Resins, Waxes

Phthalic Anhyride Alkyd Resins, Methylacrylate

Isophthalic Acid Polyesters, Alkyd Resins

Terephthalic Acid/

Dimethyltheraphthalate

Polyesters Benzene

LAS

2.1.2 Production of aromatic hydrocarbons

The major sources of aromatic hydrocarbons (BTX) include reformate from catalytic reforming, pyrolysis gasoline from steamcrackers, and coke oven light oil from

coke oven plants (Emmrich et al., 1999) Reformate from catalytic reforming provides the

main supply of benzene, toluene, xylenes and heavier aromatics (see Figure 2.2) However, the majority of toluene and heavier aromatics from the reformate is converted to benzene and xylenes, mainly for the purpose of p-xylene production In fact, for the initial purpose

to produce gasoline, the catalytic reforming process was used to improve the octane number

of straight-run naphtha Aromatic hydrocarbons, along with other high octane number components, are the main products of this process Because the aromatic content in the gasoline is now restricted, a trend to obtain the maximum aromatics yields for the aim of chemical production arises In the case that reaction conditions, feeds, catalysts, etc are chosen to maximise the aromatics yields, the process is called aromatisation This process has so far become the most effective to produce aromatics

Figure 2.2. World aromatics supply by source (Emmrich et al., 1999)

Naphtha feedstock represents about 80 % of the cost of producing BTX in the

aromatisation processes (Mowry et al., 1985) However, naphtha is in great demand for

gasoline and petrochemical production, and the value of naphtha is expected to rise as supplies become tighter Since 1996 the petroleum and petrochemical industry has performed technological alterations in the trend of using light naphtha and liquefied petroleum gas (LPG) as feedstock to produce gasoline and aromatic hydrocarbons This brings more profits for producers due to the low cost of feedstock, especially in the areas with the available sources of light alkanes At the present time, due to the influences of political instability in the Middle East and the ongoing military action in Iraq on the supply and price of crude oil, the introduction of LPG as feedstock for the aromatisation extends

Coke Oven Light Oil 4%

Reformate 72%

Pyrolysis Gasoline 24%

the supply of the feedstock and contributes to stabilise the industry of aromatics production Furthermore, it could be considered that this process is the link between the petroleum refining and the petrochemical industry: LPG is a by-product from some processes of the petroleum refining industry and aromatics are an important source of feedstock for the petrochemical industry

Already, a few commercial processes, based on ZSM-5 zeolite catalysts, have been announced, such as M-2 forming (Mobil), Cyclar (BP/UOP) and Aroforming (IFP/SALUTEC) (Bhattacharya and Sivasanker, 1995; Meriaudeau and Naccache, 1997)

In 1999, the first Cyclar-unit-based aromatics complex, designed to process 1.3 million MTA of LPG to produce 350,000 MTA of benzene, 300,000 MTA of para-xylene and 80,000 MTA of ortho-xylene, started up (UOP LLC, 2001) After operating for one year, a capacity expansion to 345,000 MTA para-xylene was planned for this aromatics complex The Cyclar process is presented in detail below

2.1.3 Cyclar process (BP/UOP)

The Cyclar process, which is a joint development of BP and UOP, converts LPG into more valuable and easily transported products with a high concentration of aromatic hydrocarbons (Maxwell and Stork, 2001) The reaction is best described as dehydrocyclodimerisation of light alkanes In fact, this is a process of many reaction steps, including dehydrogenation, hydrogen transfer, cracking, oligomerisation, and cyclisation The mechanism involves the initial transformation (cracking and dehydrogenation) of light alkanes into alkenes, followed by alkene oligomerisation and cyclisation to produce aromatic hydrocarbons, in which hydrogen transfer steps play an important role (Maxwell and Stork, 2001) These steps are catalysed by acid or metal or both sites The catalyst of the Cyclar process was disclosed to be based on gallium-modified ZSM-5 zeolite catalyst, which provides acid sites by H-ZSM-5 and metal sites by gallium species (Meriaudeau and Naccache, 1997)

Although the reaction sequence involves exothemic steps, the preponderance of dehydrogenation steps results in a highly endothermic overall reaction Thermodynamic calculations of the aromatisation reaction of propane and n-butane at 298 K on the basis of the assumption that propane is directly dehydrocyclised into benzene according to equation

(2.1) and that n-butane is dehydrocyclised into xylenes according to equation (2.2) show that the overall reaction is endothermic and that the minimum temperature at which the reaction occurs is about 704 K (or 431 0C) for n-butane aromatisation and 761 K (or 488

0C) for propane aromatisation (see Appendix 1)

2 C3H8 ⇔ C6H6 + 5 H2 (2.1)

2 C4H10 ⇔ C8H10 + 5 H2 (2.2)

In reality, the process is operated at temperatures above 450 0C due to the thermodynamic limitation (UOP LLC, 2001) Reaction pressure also has a considerable impact on the process performance A low pressure is recommended in order to obtain a maximum aromatics yield However, the high-pressure design requires only half the catalyst and is attractive where minimum investment and operating costs are the overriding considerations The typical reaction conditions of Cyclar process are shown in Table 2.1

Table 2.1. The typical reaction conditions of Cyclar process for propane aromatisation (Maxwell and Stork, 2001)

Because of the rapid deactivation of the catalyst during reaction, the continuous catalyst regeneration technology, which is similar to that applied in the low-pressure Continuous Catalytic Reforming process, is also applied in the Cyclar process (UOP LLC, 2001) As a result, it operates in the moving-bed mode and the catalyst is regenerated continually The process is divided into three major sections The reactor section includes the radial flow-reactor stack, heaters and heat exchanger The regenerator section includes the regenerator stack and catalyst transfer system The product recovery section includes

the product separators, compressors, strippers and gas recovery equipments The flow scheme of the Cyclar process is shown in Figure 2.3

Figure 2.3. The flow scheme of the Cyclar Process (UOP LLC, 2001)

Fresh and recycled feeds are combined and heat-exchanged against the reactor effluent The combined feed is then heated to the reaction temperature in the heating system

3 and then sent to the reactor system 2 The reactor system includes four adiabatic, flow reactors which are arranged in one vertical stack Due to the deactivation of the catalyst over time-on-stream, the reactor size is increased from the first to the last reactor in order to decrease the flowrate of the fluid, leading to an increase in the catalyst contact time Because the overall reaction is endothermic, the reactor temperature decreases during the reaction Therefore, between each reactor, the charge is reheated to the reaction temperature in the heating system 3

radial-The effluent from the last reactor is then separated into vapour and liquid products

in the separator 5 The liquid is sent to a stripper where light alkanes are removed from the

C6+ aromatic products The vapour from the separator is compressed and sent to a gas recovery section for separation into a 95 % pure hydrogen product stream, a fuel gas stream

of light alkanes, and a recycled stream of unconverted LPG As coke builds up on the Cyclar catalyst over time-on-stream at reaction conditions, partially deactivated catalyst is

H 2

Fuel Gases

continually withdrawn from the bottom of the reactor stack for regeneration The catalyst flows down through regenerator 1 where the accumulated carbon is burnt off The regenerated catalyst is purged and then lifted with hydrogen to the top of the reactor stack

2

Typically, the Cyclar process converts LPG into about 65 wt% aromatics, 6 wt% hydrogen, and 29 wt% fuel gas with an aromatics distribution of 30 wt% benzene, 42 wt%

toluene, 20 wt% xylenes, and 8 wt% heavier aromatics (Price et al., 1998) It can be seen

that the aromatic product is very rich in benzene and toluene From the industrial viewpoint, the process is desired to raise the percentage of xylenes in the aromatic mixtures due to their values in the chemical and petrochemical industry In addition, because the dehydrocyclisation reaction has an optimum theoretical product distribution of 89 wt% benzene and 11 wt% hydrogen (from propane) (see Appendix 2), there is considerable space for improvement of the Cyclar process and its catalyst

Catalysts used in the transformation of hydrocarbons, e.g Cyclar process, are always deactivated over time-on-stream by coke formation during the reaction (Karge, 2001) For example, in fluid catalytic cracking, the catalyst is partially deactivated only after a few seconds in the reactor In alkane dehydrogenation and aromatisation reactions, the deactivation proceeds during hours and days, respectively One of solutions used widely in the industry to overcome this problem is to use the continuous catalyst regeneration (CCR) For instance, CCR regeneration technology has been applied in the catalytic reforming and Cyclar processes to remove the coke formed on the catalyst during the reaction continually

On the other hand, the application of milder operating conditions can also extend the catalyst life; however, the catalyst activity will be lowered Another solution is to produce the catalyst having a high coke-resistance In order to achieve this, it is necessary to understand the modes of coke formation as well as catalyst deactivation during the reaction However, the problem on coke formation in the aromatisation of light alkanes has so far been paid only a little attention (at least, in the open literature) In fact, it can be seen that in the reviews on the aromatisation of light alkanes, the subject of coke formation has not been mentioned (see Reviews by Ono (1992), Guisnet and Gnep (1996), Meriaudeau and

Naccache (1997), and Fricke et al (2000)) This may be due to the complication of the

coke formation process and the well-known coke-resistance of the ZSM-5 zeolite in the

aromatisation reaction of light alkanes Therefore, in order to have a detailed insight into the deactivation of the catalyst during the reaction, further studies on this subject would be significant and essential in both academic and industrial areas

2.1.4 Conclusions

In conclusion, aromatic hydrocarbons (mainly BTX) are an important feedstock in the chemical and petrochemical industry These aromatic hydrocarbons are mainly produced by the catalytic reforming of naphtha Aromatisation of light alkanes provides a cheaper way to produce BTX in the places where sources of light alkanes are available Some aromatisation processes in the industry have followed this trend For example, the Cyclar process, which uses a bifunctional catalyst based on gallium-modified ZSM-5 zeolite, converts LPG into BTX aromatics Due to coke formation during the reaction, the continuous catalyst regeneration technology is applied in the Cyclar process Although a good BTX yield is obtained, further studies to improve the BTX yields and to reduce coke formation during the reaction for this process are essential

2.2 INTRODUCTION TO ZEOLITE AND ACTIVE SITES FOR THE AROMATISATION OF LIGHT AKANES OVER ZEOLITE CATALYSTS

2.2.1 Zeolites

Zeolite was discovered by Cronstedt in 1756 (Bruno, 1995) However, the first use

of zeolites as catalysts occurred in 1959 when zeolite Y was used as an isomerisation catalyst by Union Carbide (Dyer, 1988) Since then, zeolites have widely been used as catalysts in the transformation of hydrocarbons because of their shape-selective catalytic properties Over 130 different framework structures are now known, and about one-tenth of these structures are found in catalysts of commercial interests (Haw, 2002) A summary of some zeolite-catalysed processes is shown in Table 2.2

Table 2.2. A summary of some zeolite-catalysed processes (Blauwhoff et al., 1999)

Fluid Catalytic Cracking

Conversion of Methanol into

Synfuels via Light Alkenes

Aromatics from Light Alkanes

Isomar (UOP); Octafining (Engelhard);

Mobil Vapor-Phase Isomerization (Mobil); Mobil High-Temperature Isomerisation (Mobil)

Zeolites have a rigid, 3-dimensional crystalline structure (similar to a honeycomb) consisting of a network of interconnected tunnels and cages Another special aspect of this structure is that the pore and channel sizes are nearly uniform, allowing the crystal to act as

a molecular sieve Zeolites are crystalline aluminosilicates with fully cross-linked open framework structures made up of corner-sharing SiO4 and AlO4 tetrahedra (Bruno, 1995)

A representative empirical formula of a zeolite is

M’2/nO . Al2O3. xSiO2. yH2O

where M’ represents the exchangeable cation of valence n M’ is generally a Group I or II ion, although other metal, non-metal and organic cations may also balance the negative charge created by the presence of Al in the structure

Zeolites can be classified into three basic classes by the size of their cavities: small, medium, and large (Bruno, 1995) (see Table 2.3) The medium pore zeolites, whose structure contains 10-ring oxygen systems with their elliptical diameter of 5.5 to 5.6 Å, are interesting in the transformation of hydrocarbons because they have channels uniform to the pore size, eliminating the “supercage” problem A very good example is the zeolite ZSM-5 This zeolite, developed by Mobil Oil Corporation, is an aluminosilicate with a high silica and low alumininum content Its structure is based on channels with intersecting tunnels The structure of channels in ZSM-5 zeolite is shown in Figure 2.4

Table 2.3. Pore-size classification of zeolites (Fricke et al., 2000)

Pore size (Å)

Large pore (e.g., zeolite Y)

Medium pore (e.g., zeolite ZSM-5)

Small pore (e.g., zeolite A)

7 to 10

5 to 6

<5

Figure 2.4. The structure of channels in H-ZSM-5 zeolite (Blauwhoff et al., 1999)

In practice, the substitution of Al3+ in place of the tetrahedral Si4+ in the framework structure of ZSM-5 zeolite requires the presence of an added positive charge When this is

H+, the acidity of the zeolite is very high Besides, H-ZSM-5 possesses a very narrow range

5.4x5.6 Å (near circular) 5.1x5.6 Å (elliptical)

of acid strengths which is an important factor to simplify kinetic analysis (Wang et al.,

2004) Due to their acidity and pore sizes, these zeolites are especially suitable for reactions selective to aromatics From Figure 2.5, it can be seen that the pore diameter of MFI zeolite (e.g H-ZSM-5) does favour the formation of BTX aromatics This is one of the key reasons why ZSM-5 zeolites have been used widely as catalysts for the aromatisation of light alkanes

Figure 2.5. Pore diameters of zeolites and sizes of reactant molecules (Jentys and Lercher, 2001)

2.2.2 Acid sites: Brönsted and Lewis sites

An acid is generally defined as a species that can donate a proton (Brönsted) or accept electrons (Lewis) In the beginning, this concept was applied in the solution chemistry The dissociation of an acid HA in a solvent S is described as an acid – base equilibrium as follows:

H-A + S ⇔ A- + HS+ (2.3)

where H-A is an acid, and S is a solvent, which is also considered to be a base

It has been found that some solids (e.g zeolites) also have this similar performance

in certain conditions, and then they are called solid acids Solid catalysts are more desirable than liquid catalysts because the separation of the solid catalyst from the product is simple, and because practical solid catalysts are stable at temperatures sufficiently high for