<<

THE UNIVERSITY OF NEW SOUTH WALES

SCHOOL OF CHEMICAL ENGINEERING & INDUSTRIAL CHEMISTRY

COKE FORMATION

IN

COUPLED OXIDATIVE COUPLING

AND

STEAM

A thesis submitted for the degree of

MASTER OF ENGINEERING

by

GANG LAI B.Sc.

AUGUST 1991 ACKNOWLEDGMENTS

I would like to thank Prof. D.L. Trimm for his supervision, help and friendship throughout this research project.

I would like to thank Prof. M.S. Wainwright for his help, suggestion and friendship.

I would also like to thank Dr. R.J. Tyler and Dr. J.H. Edwards of CSIRO, Division of Coal Technology, North Ryde, for their help, suggestions and supply of information and materials.

My thanks go to the officers, Mr. J. Starling, Mr. P. Mcauley, Mr. D. Benke, and fellow students, especially Mr. Y.F. Huang, in the catalysis group for their assistance and consistent encouragement.

My gratitude go to the many friends who have encouraged me to take advantage of opportunity.

Finally, my deepest acknowledge go to my family members for all their patience, love, companionship, and understanding during this period of hard work. CERTIFICATE

I hereby certify that the work presented has been done by the candidate except where otherwise acknowledged, and that this thesis has not been submitted for a higher degree at any other university or institution.

GANG LAI I ABSTRACT

Studies have been carried out of coke formation in the context of the coupled reactions of oxidative coupling of and the of .

Previous studies had shown that the oxidative coupling of methane could be carried out in a fluid bed of catalyst. In such beds, oxygen depletion was observed near to the distributor. It had been suggested that hydrocarbons could be injected into the back end of the fluidised bed such that heat generated by oxidation could be used to facilitate the thermal to produce olefins. In that coke is known to be produced during such thermal reactions, the present studies were carried out to investigate coke formation in a coupled oxidative coupling/ pyrolysis reaction.

Studies of the oxidative coupling reaction showed that conversions and selectivities similar to those previously observed could be obtained.

Addition of hydrocarbons ( , propylene, ) after the oxidative coupling reactor led to improved selectivity to .

Coupling the pyrolysis reaction to the oxidative coupling reactor led to no formation of coke. However, when a drier was inserted between the two reactors, coke formation was observed. Under these circumstances, coke was produced from the hydrocarbons in the order II

benzene > propylene > ethane > methane.

Coke formation involved an induction period before a steady state rate of coking was established. In the presence of steam, generated in the oxidative coupling reactor, the evidence indicated that coke gasification occurred.

Coupling methane oxidative coupling to pyrolysis appears to offer improved overall efficiency with respect to heat generation and olefin production. III TABLE OF CONTENTS

Page

ACKNOWLEDGMENT

CERTIFICATE

ABSTRACT I

TABLE OF CONTENTS III

LIST OF TABLES VI

LIST OF FIGURES VII

CHAPTER 1 INTRODUCTION 1

CHAPTER 2 LITERATURE REVIEW 7

2.1 OXIDATIVE COUPLING OF METHANE 7

2.1.1 General 7

2.1.2 Feasibility of Ethylene Synthesis Via

Oxidative Coupling of Methane (OCM) 9

2.2 PYROLYSIS 19

2.2.1 Introduction 19

2.2.2 Stearn Cracking 20

2.2.3 Pyrolysis of Methane 20

2.2.3.1 General 20

2.2.3.2 Reaction Mechanism of

Methane Pyrolysis 21

2.2.4 Pyrolysis of Ethane 31

2.2.4.1 General 31 IV

2.2.4.2 Reaction Mechanism of

Ethane Pyrolysis 33

2.2.5 Hydrocarbon Pyrolysis 41

2.2.5.1 Gas Phase Reactions 41

2.2.5.2 Classification of

Homogeneous Reactions 48

2.2.6 Pyrolysis Reaction with

Hydrocarbon Additions 53

2.3 FORMATION 55

2.3.1 Thermodynamics of Carbon Formation 55

2.3.2 Classification of Carbon 57

2.3.3 Formation of Carbonaceous Materials 58

2.3.3.1 Gas Phase Carbon 59

2.3.3.2 Tars 62

2.3.3.3 Pyrolytic Carbon 65

2.3.3.4 Catalytic Carbon Formation 65

2.3.4 Homogeneous-Heterogeneous Interaction 72

2.3.5 Carbon Formation During

Oxypyrol ysis Reaction 78

2.3.6 Summary of Coke Formation 79

2.3.7 Gasification of Coke 81

CHAPTER 3 EXPERIMENTAL 86

3.1 Introduction 86

3.2 Apparatus 86 V

3.2.1 Oxidative Coupling Reactor 86

3.2.2 Oxytrap 88

3.2.3 Drier 88

3.2.4 Microbalance Reactor 89

3.2.5 Hangdown Fibres 91

3.2.6 Sample Basket 92

3.2.7 Furnace 93

3.3 Analysis 93

3.4 Materials 96

3.5 Procedure 96

CHAPTER 4 RESULTS AND DISCUSSION 98

4.1 Oxidative Coupling of Methane 98

4.1.1 Initial Investigation of Oxidative

Coupling of Methane ( Blank Run ) 98

4.1.2 Oxidative Coupling of Methane 99

4.2 Oxypyrolysis 108

4.3 Oxypyrolysis with Additives 111

4.4 Carbon Formation 114

CHAPTER 5 CONCLUSION 127

REFERENCE 130 VI LIST OF TABLES

Page

Table 2.1 Pyrolysis products from various feedstocks 33

Table 2.2 Chemical reactions in decomposition 45

Table 2.3 Concentrations of free radicals at reactor exit 49

Table 4.1 Methane conversion of oxidative coupling reaction 106

Table 4.2 Product selectivities of oxidative coupling of methane 106

Table 4.3 Product yields of oxidative coupling of methane 107

Table 4.4 Oxypyrolysis of oxidative coupling of methane products 109

Table 4.5 Oxypyrolysis of oxidative coupling of methane

products with ethane addition 111

Table 4.6 Oxypyrolysis of oxidative coupling of methane

products with propylene addition 112

Table 4.7 Oxypyrolysis of oxidative coupling of methane

products with benzene addition 112

Table 4.8 Coke formation with ethane addition (no drier) 115

Table 4.9 Coke formation with ethane addition (drier used) 115

Table 4.10 Coke formation with propylene addition (no drier) 115

Table 4.11 Coke formation with propylene addition (drier used) 120

Table 4.12 Coke formation with benzene addition (no drier) 120

Table 4.13 Coke formation with benzene addition (drier used) 120 VII LIST OF FIGURES

Page

Fig.1.1. Feedstocks production v. Production demand. 2

Fig.1.2. Strategies for conversion of methane to higher

hydrocarbons. 4

Fig.2.1. Thermodynamic data for oxidative and non-oxidative

self-coupling of methane. 8

Fig.2.2. Yields of products in the early stages of the pyrolysis of

methane in a static system at 1038 K and 441 torr. 23

Fig.2.3. Percentage ethane conversion with steam diluent (50 mol %)

at 800°C and 1 atm, versus space time. 35

Fig.3.1. Block diagram of flow system. 87

Fig.3.2. Microbalance reactor. 90

Fig.4.1. Oxidation of methane and ethane in the absence of catalyst. 100

Fig.4.2. Methane conversion over the oxidative coupling catalyst. 102

Fig.4.3. Selectivities of oxidative coupling of methane products. 103

Fig.4.4. Yields of oxidative coupling of methane products. 104

Fig.4.5. Total C2 selectivity of oxidative coupling of methane

reaction. 105

Fig.4.6. Ratio of C2H4 /C2H 6 in the OCM and oxypyrolysis

reaction. 110

Fig.4.7. Coke formation with 5% additives at 850°C. 116

Fig.4.8. Coke formation with 5% additives at 880°C. 117 VIII

Fig.4.9. Coke formation with 10% additives at 850°C. 118

Fig.4.10. Coke formation with 10% additives at 880°C. 119

Fig.4.11. cox selectivity of oxypyrolysis with 5% additives. 123

Fig.4.12. COx selectivity of oxypyrolysis with 10% additives. 124

Fig.4.13. Characteristic curve of coke gasification of deposited

coke. 125 1 CHAPTER 1

INTRODUCTION

The desire to convert methane economically to higher hydrocarbons arises from a prospective shortage of . On a global basis, both petroleum and have limited reserves but, on current usage patterns, the reserves of petroleum have a considerably shorter lifetime than those of natural gas. In addition to this global scene, the self-sufficiency of

Australia in crude oil is declining at a rate shown in Fig.1.1. There are however abundant reserves of other energy producing materials ( coal, shale oil and natural gas ) resulting in a move towards conversion of these resources into liquid fuels. The installation of New Zealand's conversion plant based on Mobil's MTG technology is an example of one response to this sort of situation.

Synthetic replacements for petroleum-derived products can be made from carboniferous natural resources, using technologies that differ in style and cost ( depending on the nature of the raw material ) and giving products that vary in chemical composition and usage potential. It seems clear that if one were forced into a situation of total dependence on synthetic petroleum replacements, usage patterns would indicate the need for more than one process. 2

Feedstocks production v. product demand Base case forecasts

Thousands ol megalkr• 43671 ML

AIP lorecast ol products demand 30 -·· ·················

Feedstocks output (BHPP base case)

20 - ~~,..!._/

1987 OoPIE production forecast 10 -···········································································································"·"

0------1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998

Fig. 1.1. Feedstocks production v. Production demand Base case forecasts

BHPP = BHP Petroleum DoPIE = Domestic Petroleum Industry Estimate

Ref: Petroleum Gazetta, 3, 25(4) (1988/9) 3 When available in relatively abundant supply, methane, the dominant and often near-exclusive component of natural gas, has a number of advantages as a raw material for the generation of synthetic petroleum­ replacement products. It is often cheap, sometimes very cheap, by comparison with alternatives, and the technology for the purification of natural gas to produce methane is generally simple. Since methane is a well-defined chemical, the technology for conversion to other fuels is more or less source­ independent, and the technology is readily transferable. On the other hand, one can expect the conversion technology to be product-market dependent, and a portfolio of product-oriented technologies would certainly be a desirable goal.

Methane is thermodynamically the most stable hydrocarbon, so conversion to higher hydrocarbons --- even into higher --- requires that deficit in free energy must be met in some way. Secondly, methane is a molecule of relatively low chemical reactivity, and there is a general problem in obtaining adequate control over the chemical reactions required for its conversion. These requirements have been met in a number of ways.

There are several main strategies available for the conversion of methane to higher hydrocarbons, which are summarised in Fig. 1.2, and they can be divided into two types: 4

PARTIAL PARTIAL OEHYDROG. OXIO OXID.

NH3 °:,z\,~·~S: IO\o/ high " , C12 temp. / ., ~ temP. HCN CO~CH10H CHP CSz I I ?1

VIA HETERO·ATOM \ \ \ \ \ \ \ HYDRO ARSONS HYDROCARBONS ' I C2l

Fig. 1.2. Strategies for conversion of methane to higher hydrocarbons Heaviest lines: Established or quasi-established technologies Dotted lined: Marginal possibilities [Anderson, J.R., App. Catal. .1Z, 177-196, 1989] 5

1. Derivatization:

Methane is firstly converted into an identifiable and ( in principle

) separable derivative which is then converted in a second step to higher hydrocarbons. In practice, it may be desirable to merge these two steps together, but in principle, the distinction remains.

2. Self-Coupling:

The higher hydrocarbon products are formed in a single and direct carbon-carbon coupling, involving only transitory, non-separable, reaction intermediates.

The chemical nature of the primary products obtained from the various strategies outlined in Fig.1.2 is quite diverse. In some cases, such as the Fischer-Tropsch reaction based on synthesis gas, or the conversion of methanol to hydrocarbons over ZSMS-type catalysts ( Mobil technologies ), considerable control of the nature of the products is available by adjustment of the reaction conditions and the nature of the catalyst. In both these cases a great deal of technological development effort has already been invested.

The possibility exists of adjusting the processes to a considerable extent towards matching specific product requirements which range from various sorts of transport fuels to chemical feedstocks. Post-conversion finishing technologies may still be needed, of course, but important products can be produced to match demand. 6

On the other hand, reactions such as direct-coupling are still only at the research laboratory stage, and can currently generate hydrocarbon products which consist exclusively of low molecular weight products. Major investment in post-conversion technology to produce the range of products is then needed for specific uses.

The present studies were aimed at reviewing and investigating some aspects of methane coupling reactions in the context of:

1. Oxidative coupling of methane to C2 hydrocarbons.

2. Steam cracking of hydrocarbons in the industrial process of ethylene

production.

3. By-products of the reactions, particularly carbon formation. 7 CHAPTER 2

LITERATURE REVIEW

2.1 OXIDATIVE COUPLING OF METHANE

2.1.1 General

Direct self-coupling of methane-derived C1 species to yield C2 hydrocarbons falls into the following categories:

(1) Simple Dehydrogenative Coupling

2CH4--->C2Hx + (8-X)/2 H 2

(2) Oxidative Coupling

2 CH4 + (8-X)/4 0 2 --> C2Hx + (8-X)/2 H 20

(X=2,4,6)

Simple dehydrogenative coupling of methane requires temperatures in excess of 800°C for practical conversions along with a high endothermic heat of 53 kcal/mol -- quite difficult to generate at temperatures of> 800°C.

On the other hand, oxidative coupling of methane removes the thermodynamic barrier of straight dehydrogenative coupling and the heat-of­ reaction problem. This is illustrated by the thermodynamic data summarised in Fig.2.1. 8

z 0 . .z.C.H ... 10!,~ (IH. - HtO 0 -4 0 2[H E •O a c i H •2Hi 0 IQ u .JI:

•-I!] -BO ~ i(H,•t!J ~ C H . ·~.. 1r1aO

-120 - t.00 600 aoo 1OOO tem~. / K

Fig. 2.1. Thermodynamic data for oxidative and non-oxidative self-coupling of methane. [Anderson, J.R., App. Catal.1Z, 177-196, 1989] 9 2.1.2 Feasibility of Ethylene Synthesis Via Oxidative Coupling of

Methane ( OCM )

There are two ways of oxidative coupling of methane to generate

C2 hydrocarbons at comparatively low temperatures: non-catalytic mode and catalytic mode.

(1) Non-Catalytic OCM

A reaction operating in a non-catalytic mode has been demonstrated by Keller and Bhasin (1982) using a range of metal oxides as oxygen transfer agents. Metal oxides are first converted to a fully oxidised state by treatment with oxygen, and then contacted with methane ( in the absence of gaseous oxygen), resulting in oxidative conversion of methane and partial reduction of the metal oxide. This reduction strategy was designed to eliminate the extensive methane oxidation which tends to occur if methane and oxygen are present simultaneously in contact with an oxidation catalyst.

Materials showing activity for C2 generation in this way at S00°C-900°C include oxides of Sn, Pb, Sb, Bi, TI, Ce and Mn. This sort of reaction would need to be run in practice on a cyclic regenerative or moving bed basis. On the other hand, there is no indication that a highly selective reaction would necessarily result without the metal oxide acting at least in part as a catalyst.

(2) Catalyst-Mediated OCM 10 Ito and Lunsford (1985) first described catalytic oxidative coupling of methane using a Li20 /MgO catalyst. Since then a substantial number of metal oxides ( and metal oxide mixtures) have been shown to have activity, including some of those identified as active non-catalytic oxygen transfer agents. Bhasin (1988) reviewed some previous research achievements:

1. Early Union Carbide Work

The primary objective of the early Union Carbide work was to uncover selective catalysts for the oxidative coupling of methane to form C2s products and principally ethylene.

A major problem in studying catalytic oxidative coupling is the presence of competing gas-phase, non-catalysed or catalysed total combustion reactions involving products or the methane feed. To minimise such non­ selective reactions, the reactor was operated cyclically, i.e., methane and air were fed one-at-a-time across the catalyst with short purging flows of nitrogen in between. Feeding in this fashion does impose another criterion on the catalyst: it must have the ability to retain one reactant until the other reactant is fed. Measurement showed that the concurrent feeding experiments gave low selectivities of 0-20%, while feed programming experiments gave selectivities of -50%. 11 An early determination of the benefits of cyclic as compared to steady flow operation was made using a flow reactor containing a catalyst consisting of lead oxide supported on a-alumina. Flows were such that the same numbers of moles per hour of the three feeds were added in each experiment. Thus both the gas-phase residence times of methane in the reactor and the average moles of methane fed per unit time perawnit catalyst were the same in both experiments. The partial pressure of methane in the cyclic experiment was higher -- a situation which, if anything, would lead to lower C2 selectivity.

An increase in the C2 yield of about one-order-of-magnitude was observed with cyclic operation as compared to steady-flow operation. The major fraction of the C2s formed was ethylene, with small amounts of ethane

( and only traces of ) being formed.

Early tests, made generally in the temperature range of 600°C to

1000°C, showed that cobalt, manganese, cadmium oxides and possibly zinc

oxide possess some C2 forming activity. Iron, nickel and silver were essentially inactive, since their activity was comparable to that of a non­ catalysed reaction from several bare supports including Filtros, firebrick, and

Norton LA956 and LA4102 supports. Other catalysts which promoted more

C2 formation than the baseline included the oxides of tin, antimony, bismuth, 12 lead, thallium, and perhaps lithium and boron. All of these tests were made with a flow reactor with cyclic feeding.

The results show that the greatest region of activity corresponds to the so-called low-melting metals of Groups IIIA, IVA, and VA together with

Mn ( Group VIIB ) and Cd ( Group IIB ). The surprising fact is that some catalytic activity is possessed by quite a number of materials.

A further unusual fact is that the amount of oxygen transfered to the catalysts and subsequently to the coupling reaction during cyclic feeding is one-two orders of magnitude larger than can be accounted for by simple monolayer coverage of the catalyst by oxygen ( Keller et al., 1982 ).

The explanation for this situation seems to lie in hypothesis that oxygen can diffuse in and out of the bulk ( sub-surface atomic layers) of the catalyst, presumably oxidising the metal to a higher valance state during the air-flow part of the cycle. During the methane-flow part of the cycle, the oxygen diffuses out to the catalyst surface and reacts with methane. The stoichiometry of this latter step would be:

CH4 + Metal Oxide ---> 1 /2C2H4 + H 20 + Reduced Metal Oxide 13

It is clear that the catalysts tested here do not produce methane conversions high enough to be of commercial interest. A conversion of perhaps 25% or more to C2s would be required. Furthermore several of the metals or metal oxides have appreciable vapour pressures above 800°C.

Nevertheless, the abundance of materials which exhibit oxidative coupling activity suggests that further research be warranted.

Cyclic feeding techniques, besides producing a substantial efficiency increase to coupled products, provide a second important economic advantage. By properly diverting the reactor effluent stream, the air used to oxidise the catalyst can be vented, while the hydrocarbon effluent can be sent to a recovery unit. Such diversion eliminates most of the nitrogen from the recovery unit, thereby greatly reducing the recovery cost of ethylene.

2. Review And Highlights Of Other Recent Work

(i) The Early Exxon Work

Mitchell and Waghorne of Exxon (1979) converted methane over materials termed by them as "Catalysts-Reagents". By this term, and the manner of their experiments, catalytic compositions were used in reactions with methane at temperatures of 702°C. To activate methane, they employed a combination of Group VIII noble metals, Group VIB metal oxides ( alone or 14 as a mixtures and capable of being reduced to a lower valant oxides ) and

Group IIA metal ( such as barium) combined with a spinel coated inorganic oxide such as gamma-alumina. An example of such a catalyst-reagent would involve a MgA120 4 spinel blocked alumina containing 6.9% Ba, 5% Cr ( or V,U or W) and 0.1-0.3% Pt. Typical conversions were 40-50% with a broad carbon number products ranging from 4-5% C2H4, 0.1-2% C2H6 and much larger amounts of other aromatic hydrocarbons and coke. The residence times were long ( 30 minutes) and the catalyst-reagents were consumed and/ or poisoned by products. A variety of compositions and conditions were exemplified.

The use of Pt to activate methane showed that the hydrocarbon can be converted to a large number of aliphatic and aromatic hydrocarbons, a small fraction of the products being ethylene plus ethane.

(ii) Baerns, et al. ( Ruhr-Universitat Bochum, FRG )

Baems, et al. (1983-1986) studied several lead oxide catalysts supported on gamma-alumina, silica and silica-alumina supports --- some modified by alkali treatment. The coupling reaction was carried out in a continuous feed mode. The catalyst of choice, PbO, was indeed one of the several most selective catalysts found in the Union Carbide work ( Keller and

Bhasin, 1982 ). 15 It was clearly shown by Baerns, et al. (1983-1986) that low acidity supports are preferred for high selectivities to C2s. Silica and alkali treated silica-alumina were found to offer the best selectivity. Acidic supports were found to possess some methane conversion activity of their own but to give poor selectivities ( Bytyn and Baerns, 1986 ). Concentrations of PbO ( up to -

4 wt.% ) were thought to be covering the active surface acidity with a

consequent increase in C2-selectivity.

These results, in general, confirm the basis of selection of alpha­ alumina as the support for all metal oxides evaluated ( Keller and Bhasin,

1982 ). Alpha-aluminas are known to possess extremely low-to-nil acidity. A mechanism for the loss of selectivity was suggested to involve an attack on methyl carbonium ions by negatively charged surface oxygen species leading first to a methoxy surface species and then to methanol and/or formaldehyde which are finally oxidised to C02 + H20. Such non-selective combustion of

CH4 could be expected to occur on acidic sites. In a similar manner the coupling products C2H6 and C2H4 may be non-selectively oxidised.

(iii) ARCO Work (C. A. Jones, J. J. Leonard and J. A. Sofranko, 1984-87 ).

Jones, Leonard and Sofrako (1984, 1987) screened a large number of metal oxides supported on silica. Gamma-alumina supports were found to be undesirable as they reacted with metal oxides to form the metal 16 aluminates. Manganese, indium, germanium, antimony, tin, bismuth and lead oxides were found to be effective coupling agents, giving 10-50% selectivity to hydrocarbons.

Reactions were carried out in cyclic redox mode in which oxidised catalyst was reacted with methane in the absence of oxygen for several minutes to form coupling products and the reduced catalyst was oxidised with air in a separate step. In addition to C2 products, formation of C3+ hydrocarbons, benzene , and coke were observed. The most active coupling agents found were same as those found in the Union Carbide work

( Keller and Bhasin, 1982 ), except for indium and germanium which were not studied. However, indium and germanium belong to the same periodic table group as thallium ( Gp IIIA ) and tin/antimony ( Gp IVA ), respectively, which were found to be active coupling agents in the Union Carbide work (

Keller and Bhasin, 1982 ).

Recently the ARCO workers have evaluated alkali modified manganese/silica supported catalyst and found them to be more selective than the unmodified oxide catalysts ( Jones, et al.,1987 ). Thus, sodium pyrophosphate ( 5% ) doping of manganese on silica catalyst improved selectivity by 10-20% and provided the additional advantage of increasing catalyst life. XRD analyses of the catalysts revealed that the improved 17 selectivity catalysts were formed when the crystal phase braunite, Mn7Si012, was present.

The ARCO work has also investigated the continuous oxygen cofeed process approach. In comparison to the cyclic redox process, the continuous cofeed process has been shown to give similar conversion/ selectivity relationship and product distributions though the conversion/selectivity was somewhat lower. The ARCO workers suggest that methyl radicals formed by reaction with the solid metal oxide undergo hydrocarbon building ( via coupling and oligomerization) in the gas phase, while the dehydrogenation of paraffins to olefins and the destructive oxidation ( to CO/C02 ) occurs on the surface of catalyst. This mechanism is similar to that proposed by Lunsford (1986). Typically, at 25% methane conversion, C2+ selectivities of 70-75% have been obtained by the ARCO workers ( Jones, et al., 1984, 1987 ).

(iv) Otsuka and Coworkers ( Tokyo Institute of Technology, Japan, 1985-87)

Over 30 metal oxides were evaluated for methane coupling and these included several of the more active metal oxides discussed earlier (

Otsuka, et al., 1986, 1987 ). Rare earth metal oxides showed higher C2- selectivities than the other active metal oxides. Samarium oxide was shown to be the most active and selective catalyst for C2 formation. This early work 18 was carried out at fairly low space velocity of 60-30 hr-1 ( Otsuka, et al., 1986

). However, in a later study at higher space velocity, samarium oxides was found to be one-to-three orders of magnitude more reactive than most other metal oxides ( Otsuka, et al., 1987 ). Otsuka and coworkers have also shown the beneficial effects of alkali and alkali halide promoters with various metal oxide catalysts. In addition, combinations of a coupling catalyst with Ga-ZSM-

5 were shown to give higher selectivity /yield of aromatic hydrocarbons (

Otsuka, et al., 1986 ).

Several other laboratories in Japan are becoming increasingly involved in the direct methane conversion catalysis.

It is clear that oxidative coupling of methane is a reaction which is of considerable potential interest. This interest had been reflected in many other studies in addition to those reported above. Thus, for example, over

1000 papers on this subject --- or on areas related to oxidative coupling --­ have been identified. Research groups in the Netherlands ( Roos, et al. ),

Argentina ( Doval, et al. ), etc., have produced some excellent work but the results observed do not change the conclusions listed above.

In summary, although the catalytic oxidative coupling of methane to form ethylene/ethane has now been amply demonstrated in many research 19 laboratories worldwide, the commercial feasibility of such a process depends critically on:

(1) developing more active ( and selective ) catalysts that will permit

operation of a steady state ( or cofeed ) reaction at lower temperatures.

(2) long term catalyst stability, and/ or

(3) scaleability of unsteady state operations.

2.2 PYROLYSIS

2.2.1 Introduction

The reactions of hydrocarbons at high temperatures have long been of industrial interest. Although the original thermal cracking route to has largely been replaced by catalytic processes, reactions such as pyrolysis, steam cracking and hydrogasification continue to be of paramount importance. This is largely as a result of the fact that steam cracking, the thermal treatment of ethane, propane, or naphtha, is the major route to ethylene and other light olefins, which are the corner stones of modern industry ( La Cava, 1977 ). 20 Although these reactions are of considerable importance, there are difficulties in their operation. The pyrolysis reaction is an unselective process that can produce considerable quantities of carbon. This poses many problems, since it deposits on the internal surfaces of the reactor and peripheral equipment to cause decreased thermal conductivity and tube blocking. If this process is catalytic, carbon can also foul the catalyst surface.

2.2.2 Steam Cracking

Although the original role of thermal cracking is now better covered by more modern processes, the principles of thermal cracking still have a part to play in the chemical industry. The most important application is in steam cracking, a thermal cracking process conducted at low pressure and high temperatures, in a vapour phase and with dilution of the hydrocarbon with steam. This process is primarily used to produce light olefins ( ethylene, propylene, and ) which are the raw materials of the modern petrochemical industry.

2.2.3 Pyrolysis of Methane

2.2.3.1 General

Methane has not been generally used as a feedstock in simple thermal cracking processes because the decomposition temperature is too high 21 and the yield of useful products too low. Although it is the simplest hydrocarbon, the pyrolysis reaction is very complex, largely as a result of the free radical chains involved. To give some idea of the complexity, the SPYRO model for methane pyrolysis ( Dente & Ramzi, 1983) contains 385 reactions.

If these are reduced by merging models which include non-ring formation and ring formation reactions, the "simplified" model still contains 231 reactions.

The general features of methane pyrolysis follows a stepwise dehydrogenation sequence:

-H2 -H2 -H2 -H2 2 CH4 -----> C2H 6 -----> C2H4 -----> C2H2 ------> 2C

However, each of the four dehydrogenation steps is itself complex, involving a variety of free-radical and molecular intermediates. Because the C2 hydrocarbons are much less thermally stable than CH4 and are readily

attacked by CH3· radicals formed in the decomposition of methane, secondary reactions of these products set in at very early stages of the decomposition, resulting very low C2 hydrocarbon selectivity.

2.2.3.2 Reaction Mechanism of Methane Pyrolysis

Typical measurements of product formation in the early stages of the reaction in a static system at temperatures around 10oo·c are shown in 22 Fig.2.2. ( Chen et al., 1975 ). The course of the reaction may be resolved into three stages, all occurring within a total conversion of methane of< 2%.

A. Stage 1: Primary Formation of Ethane and Hydrogen

~H=+ 104 kcal/ mol (1)

~H=O kcal/mol (2)

~H=-88.2 kcal/mol (3)

~H=+15.8 kcal/mol

Reaction (1) is rate controlling and the only primary source of free radicals in the system. It is effectively always followed by reactions (2) and

(3), which have much lower activation energies. Reaction (3) is apparently the

only significant fate of CH3 • radicals at this stage of the decomposition. Note that stage 1 entails no pressure change, and only in stage 2 and subsequent stages does the pressure increase. In some static pyrolysis studies based on pressure change alone, short induction periods were noted that presumably correspond to stage 1. 23

20

H2 18

16

14

;:::- I 2 Q.. ~ ~ tO ...... 0 E 8 -"'O Q) >- 6

4

2

3 time(s x 103)

Fig. 2.2. Yields of products in the early stages of the pyrolysis of methane in a static system at 1038 Kand 441 torr. [Albright 1983] 24

B. Stage 2: Secondary Reactions of Ethane

(a) Radical-Chain Dehydrogenation:

~H=-6.0 kcal/mol (4)

~H=+38.9 kcal/mol (5)

~H=O kcal/mol (2)

~H=+32.9 kcal/mol

(b) Unimolecular Decomposition:

~H=+88.2 kcal/mol (3R)

Reactions (4), (5), and (2) constitute a chain reaction converting ethane to ethylene and hydrogen. Reaction (4) is rate controlling, and the reverse of reaction 4 (4R) is largely negligible at pressures around 1 atm. As

a result, almost every C2H 5 • radical formed in reaction (4) decomposes. At much higher pressures of CH4, (4R) may become significant and the dehydrogenation less efficient. Reaction (3R), the reverse of reaction (3), 25 becomes an important secondary source of CH3 • radicals at higher temperatures as C2H6 builds up.

C. Stage 3: Secondary Reactions of Ethylene, Acetylene, and

Propylene

Secondary reactions of ethylene

(a) Radical-Chain Dehydrogenation:

""1H=-0.4 kcal/mol (6)

""1H=+42.4 kcal/mol (7)

""1H=O kcal/mol (2)

""1H=+42.0 kcal/mol

(b) Radical-Chain Methylation:

""1H=-25.8 kcal/mol (8)

""1H=+36.2 kcal/mol (9) 26 tlH=O kcal/mol (2)

As ethylene formed from reaction (5) accumulates in the system, it begins to disappear in the reaction sequences (6), (7), and (2) and (8), (9), and (2) two parallel chain reactions that consume ethylene that are propagated by methyl radicals and by their abstraction and addition reactions, respectively. Reaction (6) is rate controlling in the first sequence, and its reverse is probably negligible at 1000 K and 1 atm of CH4. In the addition sequence, on the other hand, reaction (8R) is probably much faster than reaction (9), making the latter rate controlling at 1000 K. The ratio of propylene to acetylene in the products reflects the relative rates of the two chain sequences and will vary with temperature. At 1000 K, initial propylene yields were several times those of acetylene, but at 2000 K propylene was negligible compared to acetylene. Direct decomposition of ethylene to acetylene and rapid secondary loss of propylene are also factors at the higher tern pera tures.

Secondary reactions of acetylene

(a) Radical-Chain Dehydrogenation: 27

~H=+7.8 kcal/mol (10)

~H=+139 kcal/mol (11)

~H=O kcal/mol (2)

~H=+146.8 kcal/mol

This dehydrogenation sequence does not occur, because reaction

(11) is much too endothermic to be of any importance at 1000 K, and (lOR) is probably very fast. This sequence, therefore, cannot be a source of carbon in methane decomposition at these temperatures.

(b) Radical-Chain Methylation:

CH3CHCH· ---> CH3CCH + H· ~H=+44.7 kcal/mol (13)

H· + CH4 ---> CH3 • + H 2 ~H=O kcal/mol (2)

Net reaction: C2H 2 + CH4 ---> CH3CCH + H2 ~H=+l0.2 kcal/mol 28

As with ethylene, the reverse of addition reaction (12R) is probably much faster than reaction (13), so the latter is rate controlling.

Secondary reactions of propylene

(a) Radical-Chain Dehydrogenation:

f1H=-17.7 kcal/mol (14)

f1H=+58.6 kcal/mol (15)

f1H=O kcal/mol (2)

f1H=+40.9 kcal/mol

(b) Radical-Chain Methylation:

f1H=-26.2 kcal/mol (16)

f1H=O kcal/mol (2) 29

Net reaction: C3H 6 + CH4 ---> CH3CH2CH=CH2 + H 2 i3.H=+ 13.2 kcal/mol

(c) Unimolecular Decomposition:

i3.H=+87.9 kcal/mol (18)

or

i3.H=+93.2 kcal/mol

Both addition and abstraction sequences can occur with propylene.

Abstraction appears to predominate at 1000 K as yields of allene were much higher than those of , although rapid loss of the thermally less stable butene may also be a factor. Reaction (18) can be a significant source of radicals in the system when the propylene concentration becomes high enough.

D. Summary of the Mechanism

This rather simple mechanism just outlined depends on the following assumption:

1. The CH3 • radical is the only radical present in significant concentrations. 30 2. Hydrocarbon products disappear rapidly, largely through reaction with

methyl radicals in chain sequences with no net consumption of radicals.

Dissociation reactions such as (3R) and (18) are of minor importance.

3. Higher hydrocarbon radicals dissociate so rapidly that they take almost

no part in other reactions. An exception is C2H ·, which instead is

maintained at a very low concentration by reaction with methane (10R).

4. Hydrogen atoms always react with CH4 by reaction (2) to produce

hydrogen.

These assumptions appear to be largely true in methane decomposition at 1000 K. Their validity depends chiefly on the large excess of methane ( even at high conversions, hydrocarbon products amount to only a few percent of the methane) and on the high temperature relative to that of other hydrocarbon pyrolysis, which makes the higher radicals very short­ lived with respect to decomposition.

The several chain-reaction sequences propagated by methyl radicals follow a common pattern, with abstraction or addition followed by radical decomposition, but there are notable differences. Thus, with ethane, only abstraction is possible, but, with ethylene, both abstraction and addition sequences occur. With acetylene the abstraction sequence is blocked by the 31 stability of the C2H · radical, so only addition is important. With propylene, however, abstraction of the allylic hydrogen is apparently much faster than addition, and it predominates.

Finally, it should be noted that, in this mechanism, there is initially no radical-chain decomposition of methane. The secondary chain sequences propagated through abstraction also do not consume methane, and it is only with the methylation sequences propagated by addition reactions, beginning with ethylene, that methane is consumed. Even then, these chains are limited by the amount of allophane present, and probably at most might double the rate of disappearance of methane, that is, a chain length of two.

It is clear that such methyl radical-initiated reactions lead to increasing molecular weight and decreasing hydrogen content products and will lead ultimately to carbon formation. As a result, methane pyrolysis is expected to be and is very unselective.

2.2.4 Pyrolysis of Ethane

2.2.4.1 General

The pyrolysis of ethane is also the process of choice for production of ethylene, because of the following reasons: 32 1. There is abundant natural gas resources which contain reasonable

amount of ethane. At the same time, there is lack of a competitive market

for ethane in other commercial areas. These make ethane available to the

petrochemical industry and helps keeping the price attractive.

2. The pyrolysis reaction of choice when the primary goal is to produce

ethylene is the one that minimises the formation of methane, C4

hydrocarbons, and even propylene to some extent. The reaction

parameters that most affect this selectivity are temperature, hydrocarbon

partial pressure, residence time and feedstock composition. The most

efficient feedstock for ethylene production is ethane since it requires the

simplest processing unit with the lowest capital requirement.

3. Marketing of the products from ethane pyrolysis is greatly simplified by

the low yield of by-products.

4. Overall feedstock composition is one of the primary factors in

determining the quantity and composition of the by-products. As the

molecular weight of the feedstock increases, the amount of by-product

production also increases. This can be seen from the data shown in Table

2.1. 33

5. Pyrolysis of ethane produces the lowest yield of by-products, which in

turn minimises the size of downstream units such as the depropanizer,

debutanizer, and compressors.

Table 2.1 Pyrolysis products from various feedstocks [Albrigh, 1983)

Product (in pounds/100 pounds feedstock) Feedstock Ethylene Propylene Butadiene BTX Other

Ethane 82.3 1.8 2.6 0.7 12.6

Propane 43.7 21.2 4.1 4.8 26.2

Butane 42.2 14.6 3.9 4.8 34.5

Light naphtha 29.3 14.4 4.0 13.8 38.5 Full-range naphtha 27.2 12.8 4.5 11.3 44.2 Gas oil 25.0 12.4 4.8 11.2 46.6 Crude oil 25.2 8.3 3.5 15.3 47.7

2.2.4.2 Reaction Mechanism of Ethane Pyrolysis

The mechanism of ethane pyrolysis was discussed by Quinn (1963).

The following sequence of reactions was proposed to represent the reaction at low conversions: 34

Mathematical modelling of the sequence of free-radical reactions was attempted by Snow et al. (1959), Snow (1966) and Pacey et al. (1972). A simplified molecular model was proposed by Robertson et al. (1975) for the purposes of optimisation of ethane pyrolysis plants. In general, a major portion of ethane pyrolysis reactions can be described by a single stoichiometric equation:

Fig.2.3 shows that an approach to equilibrium does indeed occur.

At 1073 K, the equilibrium constant is almost 1.0 ( Kramer and Happel, 1955

). With 50 mol% steam, the above equation predicts conversion at equilibrium to be 78.1 %. 35

80

Z 60 0 l/) 0:: w > 40 z 0 u f 20

0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 SPACE TIME (s)

Fig. 2.3. Percentage ethane conversion with steam diluent (50 mol %) at 800°C and 1 atm, versus space time. Surfaces as follows: e, Vycor glass; •, stainless steel 410; +, stainless steel 304; •, Incoloy 800 steel; ", Hastelloy X. [from Thermal Hydrocarbon Chemistry, 1979) 36 A. Scission and Coupling

The most intensively studied and best-understood pair of pyrolysis reactions is ethane scission and methyl-radical coupling:

Researchers studied the scission reaction by extrapolating ethane pyrolysis data to zero reaction time. The coupling reaction was studied by photolysis and shock-tube experiments. Essentially no temperature dependence was observed ( zero activation energy) for the coupling reaction.

Thermochemists began constructing activated-complex models to unify the data, and today, much of our present understanding of unimolecular reaction theory can be attributed to the study of these two reactions.

Zaslonko and Smirnov (1979) reviewed the literature on methyl­ radical coupling. Data from sixteen recent studies show that the reaction rate decreases slightly with increasing temperature. Between 300 Kand 1750 K, the high pressure limiting rate constant is described by

k-,rev. = 1011.40 r-0.38 L mo1-1 sec-1

where, for the bimolecular reaction, d(CH3) / dt=-2 k-,revCCH 3)2. 37

At low total pressure, methyl coupling is pressure dependent. The activated complex, containing excess energy, must be deactivated by third body collision; otherwise the excited ethane molecule may redissociate. Fall­ off curves for the rate constant have been constructed by Luther and Troe

(1979).

The "high" value for methyl-radical heat of formation ( McCulloh and Dibeler, 1976) confirmed thermochemical consistency to the forward and reverse rate parameters. Zaslonko and Smirnov (1979) found that, for ethane scission, the following equation best fits the available high-pressure limiting data:

k-,for = 10 16 · 5 exp(-88, 160 cal mo1-1 /RT) sec-1

Recently measured values are given by Trenwith (1979), who used computer-assisted, zero-time extrapolation of ethane pyrolysis data:

16 72 0 17 1 1 k -,for = 10 · = • exp (88,850 ± 680 cal moi- /RT) sec-

Trenwith's curves indicate that roughly 30% fall-off occurs at 913

K and 1 atm. Fall-off becomes more significant at higher temperature and lower pressure ( Luther and Troe, 1979 ). Mechanistic models may have to incorporate corrections for fall-off behaviour. 38 B. Hydrogen Abstractions

Abstraction reactions are unique in that, over a wide temperature range, their Arrhenius plots show curvature. The reactions are relatively fast, as indicated by high frequency factors ( measured at high temperature) and moderate activation energies. Arrhenius curvature corresponds to decreasing activation energy with decreasing temperature. Pressure dependence, associated with a third-body requirement, is not observed.

Hydrogen abstraction from ethane is described as:

Here, there is little doubt of deviation from Arrhenius behaviour. A semi­ empirical equation was found by Clark and Dove (1973a):

ktor = 10-3 · 26 T4 • 0 exp(-8280 cal mol-1 /RT) L mo1- 1 sec-1

Pacey and Purnell (1972) used a different form for their nonlinear expression:

ktor = 10 1 1.ss exp (-23, 900 cal mol-1 /RT) + 10 7 · 4 exp (-9, 100 cal mol-1 /RT) L mo1- 1 sec-1 39 Both forms have appeal from a theoretical standpoint. Transition- state theory predicts a zero order temperature dependence as in Clark and

Dove's equation. The possibility of a change in mechanism ( e.g., the appearance of quantum tunnelling or a different transition-state configuration

) is in concord with Pacey and Pumell's asymptotic formula. Both equations are accurate to a large extent.

For the reverse reaction, abstraction from methane by the ethyl radical, Kaminski and Sobkowski (1979) quoted a linear equation.

krev = 108 · 36 exp(-11,400 cal mol-1 /RT) L mo1-1 sec-1

Because both the forward and reverse reactions share the same activated complex, both should exhibit similar Arrhenius deviation. The low preexponential factors quoted by Kaminski and Sobkowski indicate that the values were measured at low temperature. They are probably applicable below 500 K.

C. Decomposition and Addition

A primary elementary step for producing ethylene is the unimolecular decomposition of the ethyl radical. 40

It is a pressure-dependent reaction. The high-pressure limiting rate constant was recently determined by Pratt and Rogers (1979a):

k-,tor = 10 13 · 5 exp(-41,800 cal mo1-1 /RT) sec-1

However, values reported by value of Benson and O'Neal (1970) have found wide application:

13 5 1 1 k -,tor = 10 · exp(-40,700 cal mol- /RT) sec-

Pressure corrections are normally employed with both expressions.

The reverse reaction ( hydrogen-radical addition to ethylene ) has been studied extensively at room temperature. Michael et al. (1973) reviewed the literature. They extrapolated to a high-pressure limiting rate constant at

300 K:

k = 10 9 · 0 L mo1-1 sec-1 00,rev

However, there is some temperature dependence, and Penzhorn and Darwent (1971) found Ea=1500 cal/mol. This value can be combined with

the preceding room-temperature result: 41

10 1 1 1 1 k oo,rev = 10 • exp(-1500 cal mo1- /RT) L mo1- sec-

Although high-temperature data are scarce, pressure corrections may also be necessary for this reverse reaction under industrial pyrolysis conditions.

2.2.5 Hydrocarbon Pyrolysis

The pyrolysis of hydrocarbons in the industrial process of ethylene production has been reviewed by Zdonik (1970). Leathard and Purnell (1970) discussed paraffin pyrolysis while the pyrolysis of hydrocarbons has been studied by Purnell (1962) and Quinn (1963).

2.2.5.1 Gas Phase Reactions

Homogeneous gas phase reactions of stable molecules do not normally proceed at measurable rates unless the reactants are activated by some process of energy absorption. Homogeneous gas phase reactions are often initiated by the unimolecular decomposition of one or more of the reactants to form free radicals or atoms. Due to their high reactivity, these undergo a variety of reactions and a complex reaction system is established. 42 The pyrolysis of hydrocarbons, a thermal reaction, proceeds mainly via vibrationally excited molecules. At the relatively low temperatures required for this process ( up to about 1000°C ) the population of higher electronic energy is insignificant. Vibrationally excited molecules may be deactivated by collisions, or may dissociate to form atoms or free radicals. If radical-radical or radical-molecule reactions occur in the system, the excess energy may result in either vibrational or electronic excitation in the product.

The free radical mechanisms which occur in hydrocarbon pyrolysis consist of several related stages. The reaction begins as a result of the formation of free radicals or atoms via activation by energy absorption ( initiation ). The subsequent course of reaction may be considered in general terms. After initiation, usually the splitting of a carbon-carbon or a carbon­ hydrogen bond:

R ---> R· + H·

the reaction proceeds by propagation. This is a rapid process and many hydrocarbon species are formed: 43

etc.

These propagation reactions are all thermodynamically favourable.

During the reaction, inhibition or acceleration can occur as a result of the presence of compounds which combine with free radicals. The formation of a stable compound or a less reactive free radical leads to inhibition, while the formation of a reactive free radical will accelerate the course of the reaction.

A free radical reaction terminates by the combination of two free radicals to form a stable molecule:

2H· ---> H2

R 1 · + R 2 · ---> R 3 H 44

2R2 • ---> R.iH

etc.

The reactor wall can influence both initiation:

Wall H2 ------> 2H · and termination:

Wall H · ------> 1 /2 H2

The complexity of the overall reactions can be illustrated by considering a specific example, such as the free radical decomposition of propane. Reactions that could occur are listed in Table 2.2. These are deduced from the mechanism of the pyrolysis of hydrocarbons, a system that has been extensively studied.

The pyrolysis of many hydrocarbons and mixtures of hydrocarbons has been studied in detail. Mechanisms which explain the stages of initiation, propagation and termination have been proposed. Various gases have been added to the reaction mixtures to study the effect of inhibitors on the free 45

Table 2.2 Chemical reactions in propane decornpostion

Initiation

C 3118 --t CI-13. + C2H5 Propagation reactions

CIT3 • + C/18 ~ C3H7·+ CH4 C2H5 • + C/ls --!, C3H7·+ CZI-16

C2H5. ~ H• + C2H4 CII3 • + HZ ___., H• + CJ-14 H• CH CH• + + 4 --+ 3 Hz C 2I\ · + H2 H• + CZH6

C2115 • + C2114 -~ CH· + C3H6 3 H· + C2114 C2H5. C2H6 - ZCH • 3 CH·+ C 21\ - czHs· + CH4 3 CII • + -- 3 C2114 ---+ C/17 •

C3H7. + C2H6 ~ C2H5 . + C3H8 II • ~ C2116 C2115 . + 112 H• + - C3H7. C3l-\

C/17 · -~ CH· + C2H4 3 C3H7. --4 H· + C3H6 C3H7 • + H2 II • + C3H8

C3H7 • + CH ~- CH · + C3H8 4 3 H • + C/ls ~ C3H7 •+ Hz CH·+ 3 C3H6 ~ C4H9. C H • C H • 2 5 + C2114 4 9 _.,, H, C4H9. - + C4H8 C4H9. + Hz ~ H• + C4Hl0 CH, + C4H9. 3 C3H6 C4H9. - cz115 . + C2H4

C2H5. + -~ CH • + C4Hl0 4 9 CZH6 CH , C/l 10 3 + C3117. - 2C H • C/110 2 5

t -~ H· C4 11 10 C4Ir9·+ Hz CH 3 · t C4HIO ~ C 4119. + CH4 46

Table 2.2 (Continued)

Termination reactions

2H · ~ H2 H• CH• CH + 3 ---1' 4 H • + C2115. ~ C2H6 II • + C3ll7. - C/ls 2CH3 • ~ C2H6 CH•+ 3 C2If5. ~ C3H8 t 2C 2H 5 ·- C2H4 C2H6 2C2 H 5 ·~ C4Hl0 CH 3 • t C3H7. - Cil4 t C3116 47 radical reaction. Both product gases from the decomposition of the various hydrocarbons and non-product gases have been found to have an effect on reaction.

In addition to initiation by excitation, products may also lead to initiation. Thus, for example, product ethane was found to be the initiator in the thermal reaction of ethylene at 703 ·c-854 ·c ( Taiseki, 1969 ). The primary products are hydrogen, ethane, acetylene, propylene, 1-butene and 1,3- butadiene and the secondary products are methane, cyclo-olefins and aromatics. Product butadiene decreased the rates of formation of hydrogen, ethane and butadiene ( but not methane and propylene) by trapping active hydrogen atoms.

Information on the propagation of free radical reactions can be gained from studying the effects of inhibitors on free radical chain reactions.

The course of the propagation reactions can be influenced by some of the highly unsaturated products formed.

To summarise, in the pyrolysis of alkanes in the gas phase, molecular processes are generally negligible and chain radical mechanisms are the important reactions. Acceleration and inhibition of the propagation reactions in the pyrolysis of saturated hydrocarbons is complex. The addition of compounds such as hydrogen chloride, hydrogen sulphide, toluene, or 48 olefins can produce, at the same time, an inhibiting and an accelerating effect

( Niclause, 1968 ). The overall effect depends on numerous parameters and on the nature of the free radical chain reactions.

2.2.5.2 Classification of Homogeneous Reactions

The course of the homogeneous gas phase reactions occurring in the pyrolysis of hydrocarbons may be studied by developing suitable mathematical models. These theoretical considerations are based on the concept that it is possible to· suggest a series of chemical reactions and rate equations from which the steady-state concentrations of the gaseous species which participate in the reaction may be derived. Considerable improvements in such models may be obtained by incorporating experimental results which demonstrate which radicals predominate and the type of kinetics which are present.

An outstanding characteristic of free radicals is that their reactions are fast. For this reason, a pseudoequilibrium concentration of free radicals is obtained very quickly. The level is mostly dependent on temperature.

Free-radical pseudoequilibrium is dynamic, however, and shifts with changes in the concentrations of molecular species. During the early stages of pyrolysis, pseudoequilibrium is shifting rather quickly but, in the 49 late stages, concentrations of both molecular species and free radicals become relatively constant. Table 2.3 shows the reactor-exit radical concentrations for ethane pyrolysis. As a comparison, total molecular concentration is of the order of 10-2 mol/liter.

Table 2.3 Concentrations of free radicals at reactor exit [Albright,1983)

Ethane pyrolysisa Radical (concentration, mol C 1) H 10-8,3 CH3 10-8,3 C2H3 10-66 C2Hs 10-6,7 C3Hs 10-7,9 1-C3H7 N.S.b 2-C3H7 N.S. C4H7 10-6,6 1-C4H9 10-10.3 2-C4H9 N.S. aExperimental conditions: 1123 K, 50 mol % steam. bN.S., not shown.

There are three general classifications of homogeneous reactions involving the various free radicals. Each group consists of forward and reverse reactions. 50

(1) Scission and Coupling

Scission of a molecule introduces a pair of free radicals. High activation energy reactions are characteristic, coupled to the pyrolysis requirement of high temperature. Radical Coupling is the reverse reaction and zero activation energy reactions dictate the total pseudoequilibrium radical concentration as a function of temperature.

Scission reactions are certainly occurring during pyrolysis. C-C bond rupture is more likely than C-H bond rupture ( Bradley, 1974 ) based on the high activation energy for methane dissociation ( -104 Kcal/ mol ) .

Large paraffin molecules tend to be more susceptible than the smaller, more refractory molecules. Single bonds in higher olefins may go through scission, and some molecules may dissociate in two or more ways.

Coupling reactions are also abundant. If ten free radical species are assumed present, then ( theoretically ) fifty-five unique coupling reactions could be occurring. The literature indicates that rate constants are consistently high, and no evidence has been found for significant temperature dependence

( Walker, 1976 ). Because of their speed, these reactions strictly regulate the total radical concentration. 51 The scissions of ethane and propane are the important initiation steps occurring during light paraffin pyrolysis. They are net producers of free radicals. Likewise, coupling reactions ( the termination steps ) are net consumers of free radicals. The other major groups of homogeneous pyrolysis reactions ( hydrogen abstraction and decomposition/ addition ) always start and finish with a free radical.

(2) Hydrogen Abstraction

A free radical can abstract a hydrogen from a molecular species.

The result is that an original free radical becomes a molecule, and an original molecule becomes a free radical. The reverse reaction is similar. Low, but not zero, activation energies are characteristic. These reactions adjust and maintain the pseudoequilibrium distribution of the various free radicals.

There are many possible hydrogen abstractions that could occur during light hydrocarbon pyrolysis. If ten free-radical species and ten molecular species are assumed present, then forty-five reaction pairs are conceivable. Throughout the literature, the forward abstractions, which produce large free radicals, are found to be dominant over the reverse reactions. The key role of abstraction reactions is to quickly convert smaller free radicals ( the products of scission and radical decomposition) into larger free radicals, the precursors to olefins. 52

(3) Decomposition and Addition

Unimolecular decomposition of a free radical results in an allophane and a smaller free radical, often a hydrogen radical. Activation energy is normally 30-40 kcal/mol. Addition of a smaller radical to an allophane results in a single larger free radical. Typical activation energy is

1-2 kcal/mol for hydrogen radical addition, and 7-9 kcal/mol for methyl radical addition. The forward decomposition reaction is responsible for conversion to allophane and the reverse addition reaction is responsible for inhibition.

Starting with a one component feed, the above set of reactions leads to a broad range of products. Coupling reactions produce larger molecules. Free-radical analogs to these larger molecules are formed via hydrogen abstractions. The larger free radicals may decompose to form higher olefins, or couple to form even larger molecules. An array of reactions involving the various paraffins, free radicals, and olefins is necessary for a detailed description of the pyrolysis. Furthermore, heterogeneous wall reactions and gas-phase complicate the picture.

Light paraffin pyrolysis is very complex with seemingly endless permutations of elementary steps based on three general groups of forward and reverse reactions. The important point is that the maintenance of 53 pseudoequilibrium in free radical concentrations is fast and dynamic. Fast because several types of reactions occur at rates comparable to gas-collision frequency. Dynamic, not only because the free-radical distribution changes with conversion, but also because this fleeting population churns, converting paraffin to allophane, and constantly readjusting.

The complexity does not stop here. Smaller yields of by-products, such as tars and cokes, are formed.

It can be concluded that the pyrolysis reaction is an unselective process. However, the selectivities of the desired products can still be achieved higher than that of the yields derived products under certain reaction conditions.

2.2.6 Pyrolysis Reaction with Hydrocarbon Additions

Mimoun at el. (1990) investigated the oxypyrolysis of natural gas with the addition of ethane. It was reported that in the absence of added ethane, oxidative coupling of methane led to 80.8% C2 selectivity at 15.8%

methane conversion. This afforded a 12.8% C2 yield at gas compositions containing 9% of oxygen. With ethane injected after the catalyst bed, the conversion of added ethane was 76.8% ( 84% selectivity to ethylene, 13% selectivity to methane ). 54 Edwards and Tyler (1989) have extended this concept to consider injecting hydrocarbon into a fluidised catalyst bed. As was observed by the above-mentioned researchers, oxygen was rapidly consumed up in the OCM catalyst bed. Well within 1 cm of the fluid bed distributor, oxygen was totally consumed, and subsequent reactions in the bed did not involve oxygen.

In these terms, the combined bed can be envisaged as operating in several different ways. In the first section of the bed, oxidation reactions occurred. These could include oxidative coupling of methane, the total oxidation of methane or oxidation of carbon deposited on catalyst. All these reactions liberate heat. In the second part of the bed, injected ethane or another hydrocarbon undergoes pyrolysis as well as any thermal reactions with the catalyst. Endothermic pyrolysis produces ethylene and consumes heat ( Albright et al., 1983 ). It also deposits carbon. The catalyst acts as a heat transfer agent, as a collector for carbon, and possibly, is involved in reactions with the hydrocarbon.

Although this proposal has much to recommend it, many of the suggestions are unproven. This section of the work was carried out to address some of the unknowns. 55

2.3 CARBON FORMATION

Cracking plants are expected to operate for a long time before cleaning is necessary. Plugging, high pressure drops and low heat transfer efficiency are produced by the deposition of carbonaceous materials on the reactor walls. Eventually, the operation of the plant has to be discontinued and the system cleaned. The expense of a shutdown is so great that a minor, slow reaction of carbon deposition is a very detrimental factor in the whole economy of the plant ( Nelson, 1958 ).

Direct observations from an industrial plant ( Trimm, 1974 ) indicates that carbon does not deposit uniformly inside the furnace tubes.

Very slow coke formation was noticed in the evaporation section of the apparatus. In the reaction zone, coke deposited steadily but became appreciable in the last sections of the tube. At the furnace outlet, in the quenching system ( a transfer line heat exchanger ), coke was found to deposit on the exchanger walls and tubes.

2.3.1 Thermodynamics of Carbon Formation

Thermodynamic data show that decomposition of hydrocarbons to give carbon and hydrogen may occur at any temperature above room 56 temperature, with the exception of methane, the inversion temperature of which occurs at 570°C ( Parks and Huffman, 1932 ). More detailed information may be found in a paper by Egloff et al. (1930) and in the well­ known tables by Rossini et al. ( 1947 to date). A brief but very clear summary of the subject is included in a book by Germain (1969). There is also a recent paper on the equilibrium in C-H systems by Lersmacher et al. (1967).

Although carbon formation does occur in many systems, direct equilibrium between hydrocarbons and carbon is not experimentally feasible, with the exception of methane. For that reason not only the thermodynamics of the overall process but also those of the intermediate steps involved should be taken into account. Dehydrogenation is very likely to occur in many carbon formation processes. The thermodynamics of dehydrogenation has been studied by several authors and reviewed by Kearby (1955). A paper by

Frey and Huppke (1933) is of particular interest.

Although carbon may be the ultimate product in the process, other solid phases may be involved. In catalytic carbon formation on metals, carbides might be expected to play an important role. Lecoanet (1968), in his recent study on methane decomposition and synthesis on nickel, and Bromley and Strickland-Constable (1960) on carbon formation from carbon monoxide on nickel, have considered both the equilibrium with carbon and with nickel carbide. 57 The formation of one or more solid phases obviously complicates the study of chemical equilibria, since not only the exact kinetics but also the phases involved and the number of degrees of freedom may not be known.

"Failure to realise that an unsuspected equilibrium has been established in a system sometimes leads one investigator to rediscover, the hard way, the second law of thermodynamics" ( Castellan, 1964).

2.3.2 Classification of Carbon

1. Ideal graphite has an hexagonal structure with cell dimensions a=2.456A

and c=6.696A. The stacking of the layers is termed ABAB. .. and the

interlayer distance is 3.353A at room temperature. Because of the

magnitude of this distance, the interlayer bonding is of the weak Van der

Waal's type. Another less common type of graphite has also been

observed, with an arrangement of layers of the type ABCABC. .. and

called rhombohedral graphite. This is less stable and reverts to the

hexagonal type under mechanical or chemical treatment. Both stacking

modifications coexist in natural graphite.

2. Natural graphite is not ideal, and good three dimensional crystals are

uncommon. Synthetic graphite have an even shorter range of crystalline

order. In addition, the density of these graphites, as evaluated from liquid

or gas displacements, is much lower than the X-ray density because of 58

pore defect structures and voids in the bulk. It is also found that the

densities measured by flotation techniques are different from those

measured by mercury or helium displacement due to the existence of

pores of small diameters.

Crystalline defects also occur, including point defects and dislocations.

They are especially important in the context of the chemistry of graphite

because they possess enhanced reactivities. Defects occur naturally or they

can be induced by high energy particle bombardment.

3. Amorphous carbon was found from x-ray studies to consist of small

crystallites embedded in a mass of amorphous material. Within a

crystallite, successive layers are in a state of complete disorder with

respect to each other although each one consists of a perfect array of

hexagonally arranged atoms. This condition was termed "turbostatic" by

Biscoe and Warren (1942). As a consequence of this structure x-ray

powder photographs show diffuse bands.

2.3.3 The Formation of Carbonaceous Materials

The gas phase decomposition of hydrocarbons is generally considered to proceed via free radical mechanisms. However, these reactions are carried out in a reactor and the interaction of saturated hydrocarbons with 59 metal or silica surfaces may also occur, The initial process, both in the absence and presence of hydrogen, is the loss of hydrogen atoms and the formation of radicals. In the presence of hydrogen, fission of the carbon-carbon bonds occurs and lower molecular weight hydrocarbons are formed. The olefins produced by the free radical process polymerise to form aromatic hydrocarbons and polynuclear aromatic hydrocarbons. These condense or adsorb on the reactor wall to form high molecular weight carbonaceous deposits.

When the reactor wall exerts a heterogeneous effect on the carbon formation, hydrocarbons absorb and reactions occur to form a carbon deposit.

This may cause a depletion on some gas phase hydrocarbons and may influence homogeneous reaction.

Carbon deposits can present a variety of structures ranging from the near amorphous to a highly crystalline graphitic state, depending on the mechanisms by which they are formed.

2.3.3.1 Gas Phase Carbon

The gas phase carbon is a very light fluffy mass of carbon which has a very large surface area, porosity, greater interlayer spacings and smaller crystallite dimensions than surface carbon. The temperature of formation has 60 little effect on the interlayer spacings and crystallite dimensions of the gas phase carbon. The individual constituent particles of gas phase carbon are roughly spherical in shape and are linked together into networks of chain-like structures. The chemical structure of the pyrolyzed hydrocarbon has little effect on the structure of the gas phase carbon and formed have similar properties, which suggest formation via a common mechanism.

The concentration and size of the gas phase carbon particles is controlled by the processes of nucleation, coagulation and growth. Several theories of carbon formation, which are actually qualitative chemical models, have been developed, both for nucleation and for growth. An explanation of nucleation is that the nuclei are formed by condensation of drops of polynuclear aromatic hydrocarbons produced in the reaction. The initial hydrocarbon reacts by a gas phase reaction to produce polynuclear aromatic hydrocarbons, the partial pressure of which increases with reaction time until supersaturation is high enough to induce condensation droplet formation. The formation of liquid nuclei eliminates the supersaturation, and the formation of additional liquid nuclei is no longer possible. The polynuclear aromatic hydrocarbons which continue to be formed maintain the growth of the nuclei and the liquid droplets pyrolyse into gas phase carbon ( Lahaye,1974 ).

Studies of the pyrolysis of aromatic hydrocarbon and argon mixtures in a shock tube has thrown light on the initial formation of gas 61 phase polynuclear aromatic species from which soot is ultimately formed (

Graham, 1975 ). Initially, soot is formed directly from gas phase intermediates and its formation is not preceded by the sudden nucleation and coagulation of droplets of liquid polynuclear aromatic species.

Following nucleation, Lahaye (1974) suggests that growth of the droplets occur by condensation of polynuclear aromatics followed by pyrolysis to give soot. Some growth may occur from the polynuclear aromatics in the gas phase and further growth may occur by the aggregation of soot particles.

Nuclei grow very rapidly once they are formed, and it is likely that a species more complicated than the inlet gas molecule is required to produce carbon growth. The molecule responsible may well be polyunsaturated hydrocarbons formed by polymerisation and dehydrogenation, growth occurring by decomposition on the nuclei surface. Ring formation within the particles, accompanied by loss of hydrogen, leads to the formation of carbon.

It has been suggested that gas phase carbon is an aggregate of large polybenzenoid free radicals and that the intermediates and nuclei are highly conjugated free radicals that undergo Diels-Alder additions with smaller unsaturated species. There is no fundamental conflict between this theory and the polyunsaturated-polyacetylenic theory. 62

2.3.3.2 Tars

Tar formation during the pyrolysis of acetylene was studied by

Bethelot (1965). Benzene and aromatic products were observed and a mechanism of synthesis was suggested: hydrocarbon to acetylene to aromatics via a process of acetylene polymerisation. However, acetylene is only a minor product in hydrocarbon pyrolysis and for it to play a major role in tar production, conversion from acetylene to tars of nearly 100% is necessary.

As a result of this, a second mechanism for tar formation was proposed in which the hydrocarbon could react to a butadiene intermediate and on to aromatics. Aromatics and polynuclear hydrocarbons were suggested to be formed via a series of Diels-Alder reactions involving butadiene and or butadiene and aromatics. Although butadiene is formed from many hydrocarbons under conditions of pyrolysis, it is not likely to be an important intermediate in the synthesis of polynuclear hydrocarbons at high temperatures. The relative energies for the addition and for radical addition suggest that polynuclear hydrocarbon production proceed via radical and not via molecular addition. However, the predominating reactions in the formation of coke from aromatic systems are dehydrogenative demerisation, trimerisation, etc. 63 A study of the basic kinetics and mechanism for tar formation during the reaction of hydrocarbons with hydrogen was made by Brooks

(1973). The simple stoichiometric hydrogasification reaction:

actually involves a complex free radical reaction mechanism. When operating close to the stoichiometric ratio ( to maximise the methane content of the product gas ) conditions arise where the hydrogen concentration is low.

Condensation reactions of hydrocarbons can occur concurrently with breakdown reactions and lead to the formation of aromatic hydrocarbons and carbon. Any attempt to produce pure methane by utilising the stoichiometry of this reaction will thus result in the formation of some aromatic hydrocarbons. The yield of aromatics increases as the hydrocarbon to hydrogen ratio increases, particularly for cyclic hydrocarbons. The tendency for a hydrocarbon to form aromatics is approximately proportional to the readiness with which olefins are formed initially from pyrolysis. Aromatics are formed by reaction of these unsaturated compounds, the concentrations of which are greatly reduced by the addition of hydrogen.

Aromatic hydrocarbons which are formed under certain reaction conditions can react in several ways: 64 a) Breakdown of benzene ring to aliphatic hydrocarbons, hydrogen and

carbon.

b) Dealkylation of the side chain present on benzene rings, e.g.

This is potentially reversible.

c) Dimerisation of the aromatic system, with the elimination of hydrogen,

e.g.

This is potentially reversible. In the presence of a large excess of

hydrogen, the equilibrium will be well over to the left-hand side.

At temperatures above 780°C fission of the benzene ring can occur and large amounts of carbon can be formed. At these temperatures, Brooks

(1973) observed higher molecular weight compounds from the pyrolysis, including diphenyl, terphenyl, , and . Ring fission resulted in the production of the active ring building unit ( suggested 65 to be the C4H3 • radical ) which gave rise to the formation of polynuclear hydrocarbons.

2.3.3.3 Pyrolytic Carbon

Pyrolytic carbons are formed from carbon containing gases at high temperatures, either homogeneously or heterogeneously, giving two distinct type of carbon ( gas phase and surface carbon ). Gas phase carbon is a microcrystalline form of carbon, produced in spherical particles with crystallites oriented approximately parallel to the surface. Surface carbons are obtained as films with well ordered crystallites deposited on solid substrates.

This type of carbon is nearer to graphite than gas phase carbon, having larger crystallites, higher density and lower interlayer spacings. The formation of both types of carbon is largely determined by the conditions of pyrolysis rather than by the structure of the starting gas, and both can be produced concurrently under certain conditions of temperature and pressure.

2.3.3.4 Catalytic Carbon Formation

In heterogeneous pyrolysis, all substrates can be considered to have some catalytic activity, since the surface process appears to be the controlling factor in producing laminar carbons. However, deposits formed on some metallic substrates at much lower temperatures leads to these substrates being 66 considered as catalysts, in comparison with "inert" surfaces such as graphite, silica and porcelain. Nickel, cobalt and iron have been found to be most effective substrates and have been the subject of many studies ( Trimm, 1977

). They not only accelerate the rate of carbon formation but also influence the structure of the carbons formed. The catalytic activity of the substrate has been shown to depend on the crystal faces presented to the gas phase (

Gwathmey, 1958 ). Different faces show different activities and only certain faces may be stable in a particular reaction environment, in that the surface will re-arrange under operating conditions. It may well be that reaction will preferentially take place at those places where the crystal lattice is rearranging or growing. In the study of carbon deposition from ethylene over single crystal nickel, Gwathmey (1958) noted that visible deposits were not formed until the surface had begun to rearrange, and he suggested that the carbon deposition was controlled by the rearrangement pattern of the crystal faces.

A summary of the literature on catalytic carbon formation reveals several common features in a large number of different systems:

a) Rates of deposition remain constant for extended periods of time.

b) The inclusion of metal in carbon deposits has been reported for

deposition from hydrocarbons. 67 c) The presence of carbides has been reported under similar circumstances.

d) A maximum rate of carbon formation has been observed at temperatures

in the region 550-600°C, and approximately zero order kinetics have been

determined for the low temperature region.

e) Hydrogen has been generally found to increase the rate of deposition.

f) The rate determining step has been associated with a solid-state diffusion

mechanism by many researchers, both at low temperatures ( < 350°C )

where carbides are the final solid products and, at higher temperatures

( 350-550 ·c ) where carbon is formed.

These observations suggest that a general mechanism for carbon formation must apply under a variety of circumstances.

Baker (1972) studied the growth of fibres from the pyrolysis of acetylene on sintered nickel films and proposed a mechanism by which mobile nickel particles act as growth centres. Carbon is taken into solution and diffuses along a thermal gradient to be precipitated at the rear of the particle, which thus continually moves forward at the head of the growing fibre. The growth curves showed a falling rate, indicative of gradual poisoning. After 10-15 seconds, the nickel particles lost their mobility and 68 growth ceased, when the particles were found to be encapsulated within a layer of carbon. Growth could be regenerated by heating in hydrogen or oxygen.

The non-fibrous deposits, other than laminar graphite, found in these systems have been described as

"polycrystalline", "nodular" and "Flocculent amorphous". Their common feature is their lack of orientation with the substrate and usually their crystallite size is small. Non-oriented graphite is formed together with laminar graphite at high rates of carbon deposition. It has been suggested that the carbon originates from gas phase polymerisation, or that the initial laminar layer is unable to cover the substrate completely before less ordered deposits begin to grow. This indicates a condition where the rate of deposition is greater than the rate of the surface rearrangement process.

Tamai (1968) studied the pyrolysis of several hydrocarbon gases over iron and nickel substrates at surface temperatures of 870-1030°C.

Deposition rates on iron from methane and ethane were temperature and pressure dependent and were attributed to some degree of gas phase reaction.

The deposition rates from ethylene over iron were practically independent of temperature and pressure. In all cases the reaction rate decreased with time and, with ethylene, this was a very pronounced effect. The results over nickel were similar to the iron-ethylene system and it was suggested that a surface 69 process was the rate controlling step. Generally, deposition rates on iron were about five times faster than on nickel.

Lafitau (1968) studied the pyrolysis of hydrogen-methane mixtures over nickel foils at about l000°C and found that the kinetic data could be explained in terms of a diffusion limited reaction in either the gaseous or solid phase, depending on conditions. The reaction rate decreased with increasing deposit weight, which suggests a fall in catalytic activity as the substrate becomes coated with the deposit.

A mechanism for carbon formation on nickel foils was proposed by

Trimm (1977). The adsorption of olefins on the surface is followed by dehydrogenation and hydrogenolysis reactions to produce carbon atoms: these then migrate through the nickel to active growth regions. Disruption of the nickel takes place and crystallites, detached from the surface, are carried with the growing carbon and catalyse further production of carbon.

Induction periods, observed under certain conditions, can be accounted for in terms of the nucleation of a new solid phase presenting an initial energy barrier. The rate determining step at low temperatures is the diffusion of carbon through the nickel.

The effects of diluent gases on carbon deposition from hydrocarbons on metal surfaces has been the subject of several studies. The 70 diluent gases may influence either the metal surface, or the gas phase reaction, or both. Hydrogen was considered to act on the gas phase reactions, and accelerated the deposition on iron. Helium affected the rates of deposition on nickel. The dilution of propane with nitrogen increased the rate of formation of carbon, at 1040°C in a quartz tube, the addition of 7.4% nitrogen to the propane increasing the rate by 34%. Larger amounts of nitrogen caused much smaller relative increase of the carbon formation rate.

A study by Anisonyan (1971) on carbon formation observed during the pyrolysis of methane in a quartz reactor, indicated a gas phase influence of hydrogen. When conversion was low, or when hydrogen was used as a diluent, only a trace of aromatic hydrocarbons was observed. At high conversions, the primary products were aromatic hydrocarbons, which though slowly formed, underwent rapid dehydropolycondensation on the reactor surface to yield carbon.

The observed features of the process of carbon formation on nickel foils can again be explained with the following model:

(i) Adsorption of the reactants on the surface.

(ii) Surface reaction, leading to adsorbed carbon species. 71 (iii) Diffusion of carbon through the nickel crystallites and precipitation

of graphite at the grain boundaries, lifting up the crystallites,

which will subsequently be carried on top of the growing carbon

thus ensuring constant rates of deposition.

At low temperatures (below 550°C) diffusion of carbon in nickel is assumed to be rate determining, while in the range from 550-650°C surface reaction seems to be controlling. At temperatures above 650°C carbon formation in the gas phase becomes important.

Nucleation of the carbon may occur on the external surface or at grain boundaries, the latter being energetically more favourable. Growth of surface nuclei cannot lift metal particles out of the substrate; instead it would result in the encapsulation of the active surface, inhibiting further deposition.

The presence of hydrogen may prevent this type of nucleation by keeping the surface clean of such encapsulating species. Nucleation at grain boundaries would then be enhanced and the growth of carbon there could push metal grain out of the substrate. Hydrogen enhances grain boundary grooving, which may account for the elimination of induction periods. 72

2.3.4 Homogeneous-Heterogeneous Interaction

In many gas phase reactions which are considered to be essentially homogeneous, interference of the reactor wall cannot be disregarded. One of the most important industrial examples is the thermal cracking of hydrocarbons. Several techniques have been reported in the literature to elucidate and eliminate the specific effects of the reactor wall, e.g. changing the surface to volume ratio of the reactor, pre-treatment of the reactor wall by oxygen, acids and sulphur compounds, or changing the wall material. Gold reactors, which are expected to have no or a very low activity on thermal cracking, have been used ( Slotboom 1974 ).

During investigations on the thermal hydrocracking of polyaromatic compounds, Penninger (1973) observed several effects of the reactor age on conversion and selectivity, which were attributed to a participation of the reactor wall. Experiments with a stainless steel reactor revealed considerable changes in conversion and product distribution during the first period of operation.

Subsequently, a steady-state system was established. The rate constants of the reactions were approximately half of the original values, while changes in the cracking selectivity were not observed. After the steady- 73 state in conversion and selectivity had been established ( after the primary ageing phase), participation of the reactor wall in the hydrocracking reactions remained. A reactor "temperature history" effect was identified. A steady conversion level at a new temperature was only established after a certain operation time at that temperature. The conversion increased if the previous temperature was lower than the new temperature; the conversion decreased if the previous temperature was higher. These results demonstrate that the reactor wall remains active in the thermal hydrocracking after the primary ageing stage of the wall has passed. The steady conversion at a particular temperature is independent of the previous temperature cycle; hence, the adaptation of the surface condition is apparently a reversible process. The presence of a hydrocarbon is essential in that hydrogen alone does not cause surface adaptation.

Slotboom (1974) demonstrated the continuous participation of the reactor wall in thermal hydrocracking, with the ageing effects of a gold-plated reactor. A completely different tendency from that of the stainless steel reactor was observed. With an increase in ageing time an increase in conversion occurred. The gold reactor experienced a slow process of activation which tended to the steady-state obtained with the stainless steel reactor. The conversion and product selectivity were essentially the same in the aged gold reactor and in the aged stainless steel reactor. 74

Thermal hydrocracking was found to proceed only in the presence of an active reactor wall; the absence of such a surface resulted in zero or very low reaction rates. With the gold-plated reactor, the formation of some products was highly sensitive to the ageing conditions, whilst other products were independent of this factor. The formation of the products which were independent of the ageing was considered as predominantly homogeneous.

The formation of the products which were highly sensitive to the ageing conditions of the gold reactor had been shown to proceed via a chain reaction.

This reaction did not proceed at all, or only at a very low rate, in the unaged gold reactor. Thus the chain reactions are developed only in the presence of an "active" reactor wall, and the surface area is not critical in the range of surface to volume ratios examined.

Thus, the initiation and termination reactions, i.e. the formation and removal of hydrogen atoms, are heterogeneous, and are both dependent on the size and nature of the surface to the same extent. Thus, the surface of the reactor has no effect on the steady-state concentration of the hydrogen atoms but remains essential for their formation.

The pyrolysis of propane was investigated in stainless steel, low carbon steel and nickel tubular reactors, which were treated with sulphur containing compounds and oxygen and hydrogen, to elucidate the role of the surface. Metal oxide surfaces in the reactor were effective in promoting 75 secondary reactions of propane and ethylene at 700°C with propane conversion increasing. The metal oxide on the reactor surface was reduced by hydrogen contact and the oxides were converted to a relatively durable and passive metal sulphide layer by contact with sulphur containing compounds.

Heterogeneous effects can paly a major role in determining the products and rates of reaction in the pyrolysis of some paraffins. These effects depend on the condition and nature of the reactor surface. A comparative critical survey of the literature has been published by Leathard and Purnell (1970), which shows that in no instance is there complete agreement on the data. The evidence indicates that, in the absence of oxygen, and with reaction vessel surfaces that are stable and not coated with carbonaceous material, the pyrolysis of propane, n-butane, and is. essentially homogeneous at the normal temperatures and pressures of paraffin studies. If the surface to volume ratio was increased sufficiently, then some heterogeneous contribution to the overall rate would probably be introduced.

There is general agreement that the pyrolysis of ethane and is markedly heterogeneous in almost all circumstances. However,

Quinn (1963) found that the initial rate of methane formation from ethane was independent of the surface to volume ratio for a variety of surface 76 treatments. As primary methane arises from the chain initiation step, this work showed that carbon-carbon bond scission in ethane is homogeneous above 550°C.

Leathard and Purnell (1970) considered that carbon conditioning of the reactor was necessary to achieve reproducible heterogeneous behaviour.

This was supported by the observation that results were irreproducible when a quartz vessel was first put into the furnace, but that reproducibility improved considerably after continued use.

A correlation was noted by Leathard and Purnell (1970) between the extent of hydrogen production and the degree of heterogeneity observed.

Hydrogen is a dominant product of both ethane and isobutane pyrolysis, but is a comparatively minor product of n-butane and isopentane pyrolysis and a trivial product of neopentane pyrolysis. The marked heterogeneity associated with ethane and isobutane pyrolysis is a result of the important role played by hydrogen atoms in chain propagation and the relative ease of their diffusion to and removal at the wall. The presence of wall effects in homogeneous reactions have been identified by changing the surface to volume ratio in the reacting system. If no change in the reaction rates were observed, it was assumed that the reaction was homogenous. However, this over-simplified criteria has been shown to fail under certain conditions. 77

The influence of the contact time and the surface to volume ratio in the reaction zone on pyrolytic carbon formation, was investigated during methane pyrolysis in a flow system ( Makarov, 1969 ) between 815-1100 °C. An assumption of the possibility of transition of the reaction centres from the gas phase to the surface was made to interpret the dependence of the experimental rates of pyrolytic carbon formation on the surface to volume ratio. The kinetic model proposed, postulated that the reaction proceeded on the surface and in the gas phase. The reaction was initiated by the active intermediates ( active centres ) such as radicals. Growth of carbon occurred by the addition reaction of hydrocarbon molecules to the active centre. The addition reactions were accompanied by dehydrogenation, ring formation and other reactions associated with changes in the electronic structure of molecules. The active centres generated in the gas phase were partly captured by the surface. Conditions which allowed the simultaneous conversion both in the gas phase and on the surface were named as the conditions of homo­ heterogeneous pyrolysis. The surface reaction was initiated by the centres both generated by methane decomposition and captured by the wall. The radical termination stages were the formation of highly carbonised products: soot in the gas phase and pyrolytic carbon on the surface.

During heterogeneous conversion, all stages of the reaction proceed on the surface. The reaction rate is proportional to the surface area, and pyrolytic carbon is the only carbonaceous product of the reaction. During 78 homo-heterogeneous conversion, individual or total stages of the reaction proceed simultaneously, both according to homogeneous and heterogeneous mechanisms.

The rate of pyrolytic carbon formation increased with an increase of the contact time. The lower the surface to volume ratio, the higher the increase of rate. Confirmation of the radical transition from the gas phase to the surface was demonstrated by an increase of the rate of growth of carbon layers on the surface when the surface to volume ratio was diminished. The rate of formation of active centres in the volume unit was approximately 104 larger than that on the surface unit. Therefore, the surface will influence the reaction rate when the surface to volume ratio is not less than 103 cm-1. At surface to volume ratios greater than 104 cm-1 the reaction is heterogeneous during all stages.

2.3.5 Carbon Formation During Oxypyrolysis

It is well known that pyrolysis of hydrocarbon at high temperature produces considerable amounts of carbon. Edwards and Tyler et al. (1989) noted that, after the oxygen was rapidly consumed in a fluid bed, there was

a loss of carbon in the form of C2 /C3 hydrocarbons as the gas passed through the oxygen-free zone of the catalyst bed. This was suggested to result from the loss of carbon from the gas phase due to cracking of hydrocarbons to 79 form carbon. The fact that carbon loss ceased once the gas left the bed suggested that cracking was catalysed by the coupling catalyst. This mechanism explains why the hydrocarbon selectivity declined as both the contact time and temperature were increased, since increasing either of these parameters should favour carbon formation.

2.3.6 Summary of Coke Formation

Coke formation in a pyrolysis system is obviously very complex and it is convenient at this point in the discussion to summarize the overall picture.

The main reaction that occur in a pyrolysis system are free-radical chain reactions in the gas phase. Coke formation is a minor reaction that is important because the coke accumulates in the system.

A reactant entering a pyrolysis reactor may undergo a series of free-radical reactions leading to gaseous, liquid, or solid products. Theses reactions occur in the gas phase.

The reactants and products of reaction may be affected by surfaces in the following ways. 80 Noncatalytic surfaces may: (a) act to collect condensed tar materials; (b) act to collect soot; (c) affect heat and mass distribution in the system; and (d) concentrate tars or soots, thereby allowing further noncatalytic reactions to occur.

Catalytic surfaces may, in addition, (a) promote the formation of carbon via a dissolution-precipitation mechanism; (b) alter the nature of the gases present in the reactor by catalysing gas-phase reactions; (c) alter, in turn, the nature and amount of tars or carbon formed in the gas phase or on a downstream surface; (d) become coated by inactive layers either deliberately

( alumina, silica ) or accidentally ( encapsulating coke ).

If a catalytic surface becomes covered in encapsulating coke ( as indeed it does ), then the limits of any catalytic action would be expected to be defined by the time needed to encapsulate the surface. Why, then, do catalytic effects play such an important role in surface reactions during pyrolysis?

There are two reasons for this: The first is that during the pyrolysis of hydrocarbons there is a distinctive induction period for the production of tars and carbon. For a flow reactor this may be translated in terms of the fact that the early part of the reactor never becomes covered in carbon or tar. As mentioned before, however, this tube material will influence downstream 81 coking. Catalytic effects, under these circumstances, will continue unless the early part of the tube is coated.

Second, the process of carburisation of the metal wall can be expected to continue, no matter whether coke is deposited catalytically or noncatalytically. Carburisation, and the subsequent disruption of the tube, will be affected by the metals present and will appear to be a "catalytic" effect.

Perhaps, as discussed before, solid-solid reactions and dissolution­ precipitation-crystallisation concepts are better means to describe what actually occurs.

2.3.7 Gasification of Coke

There are four basic reactions involving coke gasification, although it is common practice to include the water-gas shift reaction in any considerations of gasification.

C + C02 = 2 CO 82

The reaction of coke with oxygen will not be considered here, except insofar as results pertinent to gasification with other reagents will be discussed.

There are three general points which should be made in considering the coke gasification. The first concerns the importance of metallic impurities. There is no doubt that metal catalysts accelerate gasification markedly and, in this connection, the formation and gasification of coke are related. Where carbon formation leads to metal particles such as at the tip of carbon fibres, gasification will be catalysed. Noncatalytic gasification will be important when encapsulating carbons or tars are present. Second, the oxidation state of the metallic impurities is important ( Baker and Sherwood,

1980 ). Mckee (1974) found that the gasification of graphite by hydrogen was catalysed by iron, cobalt, and nickel, was slightly catalysed by manganese, chromium, vanadium, and molybdenum, but was not catalysed by copper, silver, zinc, cadmium, or palladium. Gasification by water gave generally the same reactivity if hydrogen was present, but if no hydrogen was added, the metal could oxidise and the catalytic activity of all additives was small or zero. 83 Finally, it is important to recognise that various components of tars and carbons may have different reactivities. This may result from their structure or from their chemical composition. Thus, for example, a polycyclic would be expected to react with steam more easily than carbon, but if the carbon is porous and the tar is in the form of a solidified drop, the effective reactivity will be much less because the surface area is less.

Various forms of carbon also show a range of reacti vi ties with, as a general rule, reactivity decreasing as the degree of order increases. This finding has been extended in one important direction in articles from McCarty et al. (1977) and McCarty and Wise (1979). The basis of the studies was to investigate the nature of intermediates in the conversion of carbon monoxide to methane. Four types of carbon were observed on nickel catalysts.

Chemisorbed carbon atoms were reactive to hydrogen at 550 K, and methane production was fast. Bulk nickel carbide, amorphous carbon, and crystalline carbon were all much less reactive, and significant did not occur till higher temperatures were reached ( McCarty and Wise, 1979 ). Any discussion of gasification must recognise that differences in reactivity may be important and that what may be observed overall could well be a composite effect involving several types of carbon. For example, the importance of gasification has been suggested to be small in many systems ( Trimm, 1977

) as a result of measurements of kinetics of gasification of stable carbons. The 84 identification of a very reactive form of carbon by McCarty et al. (1977) and

McCarty and Wise (1979) opens this finding to considerable question.

Steam and carbon dioxide modify the porosity of carbons, thereby producing activated carbons of different adsorptive behaviour ( Walker et al.,

1959 ). Steam is often added partially as a diluent and partially to gasify deposited coke ( Trimm, 1977; Rostrup-Nielsen, 1975; Catalyst Handbook,

1970 ). Hydrogen and carbon dioxide, produced in these systems, can have the same effect, as in steam reforming and steam cracking ( Zdonik et al.,

1970 ). Thus in some industrial reactors, coke formation is a balance between coke production and coke gasification. As a result, there is a great deal of interest in gasification catalysed by metals ( Bernardo and Trimm, 1976 ), because catalysts and reactors are generally fabricated from metallic material.

In addition, however, recent years have seen an enormous increase in interest in the gasification of coals by oxygen, steam, or carbon dioxide in connection with the program to produce synthetic crude oil. Although some reference will be made to the role of alkalis or alkaline earths as gasifying catalysts, this will not be primarily with reference to coal or coal char. This is partially because of the extent of the literature and partially because the reactivities of different coals make coal gasification studies complex and relevant mainly only to individual systems. 85 It can be concluded from literature review that:

Oxidative coupling of methane reaction produces mainly C2

Hydrocarbons ( with majority of ethane) together with a considerable amount of heat.

Pyrolysis of ethane produces ethylene accompanied by the formation of carbon.

The heat generated by the exothermic oxidative coupling of methane reaction can be partially utilised for the benefit of endothermic pyrolysis reaction.

C2 hydrocarbon ( particularly ethylene ) selectivity increases and the products are upgraded if a hydrocarbon ( preferably ethane ) is added after the OCM catalyst-bed into the cracking zone.

Carbon formation increases in the rate of aromatics > olefins > paraffins.

Steam in the reaction system gasifies carbon produced in the cracking zone. 86 CHAPTER 3

EXPERIMENTAL

3.1 Introduction

The present experiments were carried out to investigate combined oxidative coupling of methane and steam cracking of hydrocarbons, with particular emphasis on coke formation. In order to investigate logically the reactions, the experimental equipment was arranged so that individual reactions could be studied in isolation or in combination. This involved a flow system fitted with an on line microbalance for measurements of coke formation.

3.2 Apparatus

A block diagram of the flow system used is shown in Fig.3.1. This equipment consists essentially of two halves that can be operated separately or together.

3.2.1 The Oxidative Coupling Reactor

This reactor consisted of a a-alumina tube of dimensions 20 cm long by 5 mm internal diameter. Catalyst was held in position by Kaowool 87

Hydrocarbon

Microbalanco & Methanol Catalytic Jlow Oxygen oxidative Oxytrap conJrol coupling Nllrogen reactor reactor

Analysis Analysis Analysis

Fig.3.1. Block diagram of flow system. 88 plugs. Because of the high space velocities considerable care was necessary in locating the catalyst.

The reactor was held in an electric furnace in a position such that the catalyst bed was maintained in the constant temperature zone of the furnace. Because the catalyst bed was only 2 cm in length, temperature variations across the active material were less than 2 ·c in the absence of reaction.

3.2.2 The Oxytrap

In order to ensure no residual oxygen traces passed to the steam cracker, an oxytrap was placed in position between reactors. This trap was a high capacity trap obtained from Alltech. Regeneration of the oxytrap was effected by passing hydrogen at 150-200°C for 4-6 hours.

3.2.3 The Drier

The drier consisted of a Vycor tube filled with silica gel. The dimensions of the tube were 20 cm long by 1.0 cm diameter. The drier was regenerated daily by passage of hot ( ea 150°C) nitrogen for more than four hours. 89

3.2.4 The Microbalance Reactor

A diagram of the microbalance reactor is shown in Fig.3.2. The microbalance unit used was a C.I. Electronics Mark 2B, enclosed in a glass vacuum head fitted with two B34/35 sockets, which enabled the attachment of tubes or flasks for the counterweight and the sample. The microbalance head was attached to a laboratory wall by a rigid bracket, with rubber washers to diminish vibration.

The balance head was connected by cable to the control unit. A linear chart recorder ( used at 1 m V full scale ) was connected to the microbalance control unit via a matching unit provided by C.I. Electronic. The matching unit provided damping and range extending facilities. The microbalance could measure weight changes from 0 to 100mg in lg weights.

The balance head was found to be very sensitive to light and temperature, resulting in unstable weight traces. In order to minimise these effects, the head was covered with aluminium foil, and then with a Kaowool blanket. Rising heat from the reactor furnace was the main source of temperature changes in the head, so the furnace itself was covered with a triplicate layer of Kaowool. The flask containing the counterbalance was also covered with aluminium foil to reflect heat. 90

- Inlet II) II) - Outlet aIll

Thermocouple HW- l=;~;:=ll=

834

a, u

Cl)

Catalyst basket

Fig.3.2. Microbalance reactor. 91

Purging the head with nitrogen, in addition to keeping out noxious products, also contributed to temperature stability.

3.2.5 Hangdown Fibres

The sample was suspended from the arm of the microbalance by

Vycor fibres. These were drawn to thin sections and joined by hooks. The fibres were ea 5-10 cm in length.

Some difficulties were experienced in preparing Vycor fibres:

(1) It was very difficult to make them in the precise length required for the

experiment.

(2) It was extremely hard to make the hooks at both ends of the fibre in the

right size and smoothness.

(3) The fibres were brittle, causing them to break easily when the basket was

hung.

(4) The central tube of the microbalance reactor was bit too long so that it

was difficult for the fibres to stay free of the tube wall. 92 (5) Tiny amounts of unconsumed oxygen would always cause abrupt

expansion at high temperature, resulting in unhooking of the fibres.

3.2.6 Sample Basket

The catalyst used in the experiment was powder contained in a basket. At the early stages, a stainless steel mesh basket was employed. It was found that the mesh was catalytically active for carbon formation. A Vycor basket was then tried but could not be fabricated owing to the manufacturing difficulty. An alumina basket was then fabricated using the following method:

Composition: Alumina powder (XA 16) 1000 g

Water 835 ml

Carboxy-Methyl-Cellulose (CMC) 3.5 g

The alumina powder was mixed with CMC and water and the mixture roll milled for one and a half hours. The resulting suspension was poured into a plaster mould and repeatedly topped up with water as the fluid was absorbed by the mould. After about 35-40 minutes, the resulting alumina basket shrank away from the mould and it was then a simple matter to extract the basket. The basket was calcined in an oven using a temperature program (150°C/hr to 1600°C) and held overnight at 1600°C. 93 The finished basket was tested and found to be non-active for carbon formation.

3.2.7 Furnace

Two different size hand-made wire wound furnaces, insulated with

MgO and both having an maximum operating temperature of 1200°C, were used to heat the reaction chambers.

The furnace temperature was controlled by a Ero PTD temperature controller connected to a type K ( Chromel / Alumel ) thermocouple placed in the middle of the reactor. Good temperature control was obtained when the furnace was heated to 300°C and then to the operating temperature.

3.3 Analysis

Gas analysis was carried out using two types of gas chromatograph. These were placed in series on the exit lines of both the first and second reactor.

A sample loop size of 0.4377 ml was used for the thermal conductivity detector gas chromatograph ( Gow Mac Model 69-522 ). Carbon monoxide, carbon dioxide, nitrogen and oxygen were separated on a CTR1 94 column (10') or a Carbosphere column (6') from Alltech Associates. The second column was preferred. The column temperature was held at 20°C.

A sample loop of 0.1105 ml was used for a Varian 3300 chromatograph fitted with a flame ionisation detector. The column of F-1

Alumina (5') was used to separate hydrocarbons at 50°C.

Appropriate valves were used to allow sampling from oxidative coupling reactor or microbalance reactor.

Carbon formation was recorded as a weight increase versus time.

However, obtaining a stable weight trace required correct operation of all the components of the system and considerable time was often required to locate the source of the instability.

(1) Static electricity was found to cause a lot of instability in the system, a

problem reported by other researchers who have worked with

microbalance ( Lobo, 1971, Stirling, 1988 ). Static electricity may be

caused by purging of the microbalance head, heat convection, and/ or

friction between the walls of the reactor and the gases and solids passing

through it. Static electricity problems were minimised by regular washing

of the reactor with acetone, and by spraying of the central tube of the

microbalance reactor with silicone Free Antistatic Record Cleaner. 95 (2) Trace instability was found to increase with the sample weight.

(3) Trace instability was caused by the changing of furnace temperature,

which was probably due to the associated changes in the microbalance

head temperature, caused by heat convection.

(4) Another cause of trace instability was related to the length of hangdown

fibre segments. The longer the segments is, the greater the trace

instability becomes. This was predominantly because of the oscillation

effect.

(5) The life of the battery also affected the trace stability. The battery being

used in the Mark 2B was a 4.Sv Mercury one. The operation indicated

that the battery needed changing in every three month time.

(6) The alignment and location of the catalyst basket and associated

glassware were very critical to trace stability.

(7) There was a relationship between the gas flow rate and the position of

the catalyst basket. Generally, the faster the flow rate, the longer distance

between the basket and the lower end of the central tube should be kept

so that the vibration of the basket could be minimised. 96

3.4 Materials

The catalyst was a promoted strontium-magnesia catalyst supplied by CSIRO Division of Coal and Energy Technology. The composition involved Mg:Sr:La=80:20:2. The particle size was about 150 µm.

Methane, ethane, propylene and other gases were obtained from

CIG in the purest form available ( generally > 99% ).

3.5 Procedure

Methane, oxygen and nitrogen were passed initially to a flow control system, consisting of needle valves and rotameters. Previous calibrations ensured accurate flows to within 1 %. From the flow controllers, the gases were mixed and taken through a flow reactor held at controlled temperatures in an electrically heated furnace. This reactor contained catalyst used to promote the oxidative coupling of methane. The gases leaving the reactor could be analysed, vented or passed to the second half of the system.

The exit gases could either be taken through a drier and oxytrap or taken directly to the second reactor. Metered flows of hydrocarbons could be added between reactors ( Fig.3.1 ). All transfer lines between reactors were maintained at temperatures in excess of 120°C, in order to avoid condensation of water. 97 The second reactor was equipped with a microbalance from one arm of which hung a basket containing the same oxidative coupling catalyst.

Gases were passed over this basket and any changes in weight could be recorded continuously. The reactor and hanging basket were maintained at a preset temperature by external electrical heating. Gases leaving the second reactor could be taken for analysis or vented. 98 CHAPTER 4

RESULTS AND DISCUSSION

4.1 Oxidative Coupling of Methane

Oxidative coupling of methane reactions were carried out at different temperatures to achieve the maximum productivity of ethane and ethylene.

4.1.1 Initial Investigation of Oxidative Coupling of Methane (

Blank Run)

A blank run was carried out under following experimental conditions:

Weight of a-alumina in reaction: 300 mg

Total Flow: 250 ml/min.

95/5

The mixture was passed through the empty silica reactor at temperatures in the range 500°C to 900°C. It was found that there was no detectable gas-phase reaction of methane and oxygen at any temperature in this range at the residence time used in the experiments. 99 Ethane and oxygen mixtures were also passed through the empty silica reactor under the same experimental conditions. Unlike methane, ethane started to dehydrogenate at about 650°C. The major product was ethylene with small amounts of carbon oxides and methane, which increased rapidly with increased temperature. When the temperature was over 700°C ( at which point it was estimated that the oxygen consumption was complete ) the major product was ethylene, suggesting that ethane pyrolysis plays an major role at high temperatures.

The experiment was then conducted with ethane only passing through the empty reactor. The results obtained showed that ethane started to decompose at temperatures above 700°C. Ethylene was the major product.

These results, shown in Fig.4.1, are similar to results obtained by

Burch et al. (1990), Martin et al. (1989), and Geerts et al. (1989). It was reported by Burch and Tsang (1990) that, in the absence of oxygen, ethylene was remarkably stable. This is the reason why high selectivity to ethylene may be achieved in methane coupling under conditions of oxygen starvation.

4.1.2 Oxidative Coupling of Methane

Initial attention was focused on oxidative coupling experiments 100

Conversion % BO~------...... ,

70

60

50

40

30

20 -

10 -

0 L------'lil':...__-----f":...__--lll'-=4i>------

Fig.4.1. Oxidation of methane and ethane in the absence of catalyst. 1. Total flow: 250 rnl/rnin.

2. Vol.% of CH4/02: 95/5.

C2H6/02: 95/5. 3. No catalyst present. 4. No pyrolysis reaction. 101 conducted at different temperatures in order to search for the most suitable temperature range for oxidative coupling of methane ( OCM ). The following conditions were used:

Weight of catalyst: 200 mg

Weight of alumina pellets

associated with catalyst: 300 mg size: 1 cm long 1.5 mm diameter.

Total flow: 250 ml/min

Flow of methane: 237.5 ml/min

Flow of oxygen: 12.5 ml/min

The mixture of methane and oxygen was passed through the oxidative coupling reactor with catalyst :placed between two beds of a­ alumina. Temperatures were varied from 600°C to 900°C. The relevant results are listed in Tables 4.1, 4.2 and 4.3 and shown in Fig.4.2, 4.3 and

4.4. The total hydrocarbon selectivity versus temperature is also plotted in

Fig.4.5. 102

CH4 Conversion (%)

8

7

6

5

4

3

2

1

0 500 600 700 800 900 1000 T ( °C)

Fig.4.2. Methane conversion over the oxidative coupling catalyst. 1. Total flow: 250 ml/min.

2. Vol.% of CH4 /02: 95/5. 3. Catalyst weight: 200 mg. 4. No pyrolysis reaction. 103

Selectivity (%) 70,-----=--___:__:______---i

60

50

40

30

20

10

0 500 600 700 800 900 1000 T (oc)

---- C2 II 6 +- C2Il1 -',1(- co -El- co2

Fig.4.3. Selectivities of oxidative coupling of methane products. All conditions are listed in Fig.4.2. 104

6.0

5.0

'1.0

3.0

2.0

1.0

0.0 L_---=!=------'---______L______,._ ____~--~ 500 GOO 700 uoo \JOO 1000 T (°C)

Fig.4.4. Yields of oxidative coupling of methane products. All conditions are listed in Fig.4.2. 105

100 releclivity (%)

!JO

80

70

60

50

40

20

10

o~---~---~---~---~---~ 500 GOO 700 800 900 1000 T ( °C)

Fig.4.5. Total C2 selectivity of oxidative coupling of methane reaction. All conditions are listed in Fig.4.2. 106

Table 4.1 Methane conversion of oxidative coupling reaction

Products (v /v%) Methane T(°C) C2H6 C2H4 co C02 Conversion (%) 600 0.119 0.019 1.187 1.691 2.84 650 0.516 0.082 0.849 1.700 3.37 700 1.023 0.205 0.611 1.660 4.25 750 1.861 0.630 0.402 1.439 6.14 800 2.701 1.168 0.380 1.376 8.55 850 2.641 1.296 0.372 1.302 8.59 880 2.349 1.362 0.361 0.992 7.90

1. Catalyst weight: 200 mg. 2. Total flow : 250 ml/min. 3. Mixtures of CH4/02: 95/5 (v /v%). 4. Oxidative coupling reaction only.

Table 4.2 Product selectivities of oxidative coupling of methane

Products (v /v%) Selectivity (%) T(°C) C2H6 C2H4 co C02 C2H6 C2H4 co C02 600 0.119 0.019 1.187 1.691 7.6 1.2 37.6 53.6 650 0.516 0.082 0.849 1.700 27.6 4.4 22.7 45.4 700 1.023 0.205 0.611 1.660 43.3 8.7 12.9 35.1 750 1.861 0.630 0.402 1.439 54.6 18.5 5.9 21.1 800 2.701 1.168 0.380 1.376 56.9 24.6 4.0 14.5 850 2.641 1.296 0.372 1.302 55.3 27.2 3.9 13.6 880 2.349 1.362 0.361 0.992 53.5 31.0 4.1 11.3

Conditions are listed in Table 4.1.

As was reported by many researchers and verified by the 107

Table 4.3 Product yields of oxidative coupling of methane

Products (v /v%) Yield (%) T(°C) C2H6 C2H4 co C02 C2H6 C2H4 co C02 600 0.119 0.019 1.187 1.691 0.23 0.04 1.13 1.60 650 0.516 0.082 0.849 1.700 0.98 0.16 0.80 1.61 700 1.023 0.205 0.611 1.660 1.94 0.39 0.58 1.57 750 1.861 0.630 0.402 1.439 3.53 1.19 0.38 1.36 800 2.701 1.168 0.380 1.376 5.12 2.21 0.36 1.30 850 2.641 1.296 0.372 1.302 5.00 2.46 0.35 1.23 880 2.349 1.362 0.361 0.992 4.45 2.58 0.34 0.94

Conditions are listed in Table 4.1. present experiments, two major requirements must be reached in order to achieve higher selectivities and yields of C2 products:

(1) High linear space velocities, which limit product back-mixing.

(2) Elevated catalyst-bed hot-spot temperatures with mixtures of methane

and oxygen which results in a rapid consumption of oxygen, thereby

limiting product degradation.

From the initial experiments, which focused on the homogeneous gas-phase reactions, methane is found to be thermally stable at high temperatures ( Fig. 4.1 ). Ethane is not stable both in the presence and absence of oxygen ( Fig. 4.1 ). Ethylene is not stable in the presence of 108 oxygen but stable in the absence of oxygen ( Burch and Tsang, 1990 ). The rates of the oxidation reaction in the gas phase decrease in the order ethylene > ethane > methane.

The OCM results indicate that both methane conversion and C2 selectivities increase with increasing temperature. Methane conversion reaches a maximum of 8.59% at 850°C and slightly reduces to 7.90% at

880°C ( Fig.4.2 and Table 4.1 ). C2 selectivity from the oxidative coupling of methane goes up to 82.5% with 55.3% of ethane and 27.2% of ethylene at sso·c, and 84.5% with 53.5% of ethane and 31.0% of ethylene at 880°C (

Fig.4.3 and Table 4.2 ).

It can be seen that temperatures between 850 ·c and 880 ·c were the ideal range in order to reach high C2 selectivity while maintaining relatively high methane conversion ( Fig. 4.2 and 4.3, Tables 4.1 and 4.2 ).

4.2 Oxypyrolysis

Attention was then focused on the pyrolysis of oxidative coupling products, achieved by passing the OCM products only through the microbalance reactor. The outlet gases were separated using gas chromatography and the results analysed using a OAP A software computing system. The results are listed in Table 4.4. There was no catalyst used in the microbalance reactor since attention was focused on pyrolysis 109 at this stage.

Table 4.4 Oxypyrolysis of oxidative coupling of methane products

T(°C) Products (v /v%) Selectivity (%)

C2H6 C2H4 co C02 C2H6 C2H4 co C02

850 1.031 2.578 0.398 0.996 23.9 59.9 4.6 11.6

880 0.536 3.001 0.423 1.002 12.6 70.6 5.0 11.8

1. Both reactors maintain at the same temperature. 2. Inlet gases were the products from OCM reactor. 3. All the other conditions are listed in Table 4.1.

The results demonstrate that selectivity to ethylene increased from 27.2% to 59.9% at 850°C and 31.0% to 70.6% at 880°C ( Table 4.4 ).

The ratio of ethylene to ethane increased remarkably compared to that of the OCM results ( Fig.4.6 ), suggesting that ethylene was mainly from pyrolysis of ethane. The ratio of ethylene to ethane in the present experiments is somewhat lower than that reported by Mimoun et al. (1990), but this is mainly because of the different design of the two reactors. The two reactors in Mimoun's study were joined together and little heat generated by the exothermic OCM reaction was lost before the endothermic ethane pyrolysis. In the present study, there is some considerable distance between the two reactors, causing a considerable loss of the OCM reaction heat. 110

Ratio of CzII4/CzII5 7.0 ------

6.0

5.0 -

4.0

3.0

2.0

1.0 ____ ..

0.0 ~------~------~------~ OOO 850 900 950 T (°C)

-- OCM products --+- Oxypyrolysis product

Fig.4.6. Ratio of C2H4 /C2H6 in the OCM and oxypyrolysis reaction. 1. OCM reaction: Total flow: 250 ml/min.

Vol.% of CH4 /02: 95/5. Catalyst weight: 200 mg. 2. Oxypyrolysis: Both reactors remained at same temperature. Inlet gases were the products from OCM reactor. 111

4.3 Oxypyrolysis with Additives

Ethane, propylene and benzene were injected after the OCM reactor into the microbalance reactor at 5% and 10% volumetric percentage respectively based on the flow rate leaving the OCM reactor. The flow of methane into the OCM reactor was adjusted to keep the total gas flow unchanged. Product analysis are listed in Tables 4.5, 4.6 and 4.7.

Table 4.5 Oxypyrolysis of oxidative coupling of methane products with ethane addition

Vol % Products ( %) Mole%. T(°C) C2H6 Added C2H6 C2H4 co C02 C2H6 C2H4 co C02 850 5 1.326 3.580 0.398 0.999 23.7 63.9 3.6 8.9

880 5 0.669 3.927 0.421 1.008 12.6 74.0 4.0 9.5 850 10 3.251 9.139 0.470 1.012 24.8 69.6 1.8 3.9

880 10 1.553 9.218 0.499 1.043 13.5 79.9 2.2 4.5 l. Ethane was injected after OCM catalyst-bed into the cracking zone. 2. All the other conditions are listed in Table 4.4. ·on carbon basis.

It was found that ethane addition after the OCM catalyst-bed resulted in an increase in ethylene selectivity ( Table 4.5 ). With 5% ethane added, ethylene selectivity increased from 59.9% to 63.9% at 850°C, 70.6% to 74.0% at 880°C. With 10% ethane added, ethylene selectivity increased from 59.9% to 69.7% at 850°C, and 70.6% to 79.9%. These results support the suggestion that, in the high temperature oxygen-free zone, ethane 112

Table 4.6 Oxypyrolysis of oxidative coupling of methane products with propylene addition

Vol% Products ( %) Selectivity (%) T(°C) C3~ Added C2H6 C2H4 co C02 C2H6 C2H4 co C02 850 5 1.230 3.366 0.477 1.007 23.0 63.1 4.5 9.4

880 5 0.598 3.558 0.501 1.016 12.2 72.4 5.1 10.3 850 10 1.800 5.050 0.484 1.006 23.7 66.5 3.2 6.6

880 10 0.998 6.031 0.512 1.017 12.8 77.4 3.3 6.5

1. Propylene was injected after OCM catalyst-bed into the cracking zone. 2. The conversion of added propylene was 90%, the remaining feed passing through the reactor. 3. All the other conditions are listed in Table 4.4. Table 4.7 Oxypyrolysis of oxidative coupling of methane products with benzene addition

Vol % Products (%) Selectivity (%) T(°C) Benzene Added C2H6 C2H4 co C02 C2H6 C2H4 co C02 850 5 1.233 3.378 0.521 1.017 22.8 62.5 4.8 9.9

880 5 0.601 3.546 0.547 1.099 12.1 71.4 5.5 11.1 850 10 1.804 4.941 0.535 1.090 23.9 65.4 3.5 7.2

880 10 0.998 5.878 0.558 1.104 13.0 76.3 3.6 7.2

1. Benzene was injected after OCM catalyst-bed into the cracking zone. 2. The conversion of added benzene was 95%, the remaining feed passing through the reactor. 3. All the other conditions are listed in Table 4.4. produced ethylene through pyrolysis.

The replacement of ethane by propylene addition after the OCM catalyst-bed indicated similar results ( Table 4.6 ). Ethylene selectivity increased from 59.9% to 63.1 % at 850°C, 70.6% to 72.4% at 880°C with 5% 113 propylene addition; and 59.9% to 66.5% at sso·c, 70.6% to 77.4% at sso·c with 10% propylene addition. According to La Cava (1977), the main products of the thermal reaction of propylene are to be ethylene, methane, hydrogen, butenes and butadiene ( Kunugi, et al., 1970 ).

The results of benzene addition after the OCM catalyst-bed demonstrated that, with 5% benzene added to pyrolysis zone, ethylene selectivity increased from 59.9% to 62.5% at sso·c, 70.6% to 71.4% at sso·c.

When 10% benzene was added to the microbalance reactor, ethylene selectivity increased from 59.9% to 65.4% at sso·c, and 70.6% to 76.3% at sso·c c Table 4.7 ).

It can be seen that the increase of ethylene selectivity is in the order of:

ethane > propylene > benzene

when these three additives were injected after the catalyst bed into the cracking zone. This results from the fact that pyrolysis of benzene leads to carbon formation through polycyclic structures, which is less selective towards ethylene compared to propylene and ethane pyrolysis.

This is a very important factor in achieving high ethylene selectivity. It was unexpected to find that the selectivities of carbon oxides 114 were reduced, which was not indicated by the mass balance. However, this was probably caused by the structure of the microbalance reactor as will be discussed later.

4.4 Carbon Formation

Relevant experiments were then carried out to study carbon formation on 20 mg OCM catalyst placed in an alumina basket in the microbalance reactor, with the injection of ethane, propylene and benzene into the cracking zone.

It was somewhat surprising to find no observable trace of carbon on the catalysts on contrast to the results obtained by Edwards

et al. (1989). In order to study this in more detail, the possibility of carbon gasification as a result of reaction with steam was investigated by the introduction of a drier and an oxytrap between the OCR and the MBR

( Fig.3.1 ). It was found that the drier had significant effects on carbon formation, while the oxytrap had little effects on carbon formation, confirming that oxygen had been consumed in the OCM bed. Tables 4.8 to

4.13 contain the results obtained with and without the application of the drier.

It can be seen from the carbon formation results that an induction period for coke formation, lasting about 2-8 minutes, was 115

Table 4.8 Coke formation with ethane addition (no drier)

Vol% PRODUCTS(v /v%) Mole%* CARBON T("C) Ethane Added C2H6 C2H4 co C02 C2H6 C2H4 co C02 '4'g/ sec/ g.cat)

850 5 1.326 3.580 0.398 0.999 23.7 63.9 3.6 8.9 0 880 5 0.669 3.927 0.421 1.008 12.6 74.0 4.0 9.5 0

850 10 3.251 9.139 0.470 1.012 24.8 69.6 1.8 3.9 0 880 10 1.553 9.218 0.499 1.043 13.5 79.9 2.2 4.5 0

1. No drier used in the system. 2. All the other conditions are listed in Table 4.5. 'on carbon basis. Table 4.9 Coke formation with ethane addition (drier used)

Vol% PRODUCTS (v /v %) Mole%* CARBON T("C) Ethane added C2H6 C2H4 co C02 C2H6 C2H4 co C02 }Ilg/ sec/ g.cat)

850 5 1.324 3.580 0.349 0.985 23.8 64.3 3.1 8.8 90.4 880 5 0.667 3.926 0.382 0.991 12.6 74.4 3.6 9.4 116.7

850 10 3.250 9.137 0.427 1.000 24.8 69.8 1.6 3.8 101.9 880 10 1.554 9.219 0.446 1.031 13.5 80.1 1.9 4.5 130.6

1. Drier was linked between two reactors. 2. All the other conditions are listed in Table 4.5. "on carbon basis. Table 4.10 Coke formation with propylene addition (no drier)

Vo!% PRODUCTS (v /v %) SELECTIVITY(%) CARBON T("C) C3H6 Added C2H6 C2H4 co C02 C2H6 C2H4 co C02 pg/sec/g.cat)

850 5 1.230 3.366 0.477 1.007 23.0 63.1 4.5 9.4 0 880 5 0.598 3.558 0.501 1.016 12.2 72.4 5.1 10.3 0

850 10 1.800 5.050 0.484 1.006 23.7 66.5 3.2 6.6 0 880 10 0.998 6.031 0.512 1.017 12.8 77.4 3.3 6.5 0

1. No drier used in the sustem. 2. All the other conditions are listed in Table 4.6. associated with the initial stages of reaction. After this, coke formation increased, but settled down to a linear increase with time on line ( steady- state ) ( Fig.4.7, 4.8, 4.9, 4.10 ). 116

Carbon Formation (mg/g.cal.) 350 ~------i

300

250

200

150

0 5 10 15 20 25 30 35 Time (rnin)

'"· Elhan1! · I Pnipyknc

Fig.4.7. Coke formation with 5% additives at 850°C. 1. 20 mg of oxidation catalyst in microbalance basket. 2. Hydrocarbons were injected after the OCM catalyst-bed into cracking zone. 3. All the other conditions are listed in Fig.4.6. 117

Carbon Formalion (mg/g.cal.) 350 ~------,

300 /

250

~ 200 /

0 5 10 15 20 25 30 Tirne ( rni n)

----- Ethane --j-- Propylene _::,¥- Benzene

Fig.4.8. Coke formation with 5% additives at 880°C. All conditions are listed in Fig.4.7. 118

Carbon Formalion (rng/g.cal.) :350 ~------,

300

250

200 y -~­

() 5 10 15 20 25 :JO :35 Time (min)

------E lhane ---f- Propylene + Denzene

Fig.4.9. Coke formation with 10% additives at 8S0°C. All conditions are listed in Fig.4.7. 119

Carbon Formalion (mg/g.cat.) 400 ~------,

300

200

JOO

0 5 10 I 5 20 25 :10 Tirne (min)

----- Elhane + Propylene -~ Benzene

Fig.4.10. Coke formation with 10% additives at 880°C. All conditions are listed in Fig.4.7. 120

Table 4.11 Coke formation with propylene addition (drier used)

Vo!% PRODUCTS(v /v %) SELECTIVITY(%) CARBON T('C) C3H6 Ip{,! sec/ g.cat.) Added C2H6 C2H4 co C02 C2H6 C2H4 co Co2

850 5 1.232 3.365 0.411 0.993 23.2 63.5 3.9 9.4 127.6 880 5 0.598 3.559 0.428 1.001 12.3 73.0 4.4 10.3 139.2

850 10 1.803 5.049 0.414 0.991 23.9 66.8 2.7 6.6 135.8 880 10 0.997 6.032 0.431 1.000 12.9 77.8 2.8 6.5 151.7

1. Drier was linked between two reactors. 2. All the other conditions are listed in Table 4.6.

Table 4.12 Coke formation with benzene addition (no drier)

Vo!% PRODUCTS (v /v %) SELECTIVITY(%) CARBON T('C) Benzene added C2H6 C2H4 co C02 C2H6 C2H4 co C02 (pg/sec/g.cat)

850 5 1.233 3.378 0.521 1.071 22.8 62.5 4.8 9.9 0 880 5 0.601 3.546 0.547 1.099 12.1 71.4 5.5 11.1 0

850 10 1.804 4.941 0.535 1.090 23.9 65.4 3.5 7.2 0 880 10 0.998 5.878 0.558 1.104 13.0 76.3 3.6 7.2 0

1. No drier used in the system. 2. All the other conditions are listed in Table 4.7.

Table 4.13 Coke formation with benzene addition (drier used)

T('C) Vo!% PRODUCTS (v /v %) SELECTIVITY(%) CARBON Benzene added C2H6 C2H4 co C02 C2H6 C2H4 co C02 VJg/sec/g.cat)

850 5 1.235 3.348 0.413 0.994 23.4 63.3 3.9 9.4 191.6 880 5 0.596 3.457 0.431 1.003 12.5 72.5 4.5 10.5 207.6

850 10 1.801 4.917 0.416 0.995 24.3 66.2 2.8 6.7 216.9 880 10 0.992 5.755 0.434 1.005 13.3 77.1 2.9 6.7 231.5

1. Drier was linked between two reactors. 2. All the other conditions are listed in Table 4.7.

Induction periods are common during studies of carbon formation ( Turner, 1976 ). It has been suggested that particular crystal faces may favour carbon formation, and that the induction period is associated with the restructuring of the crystal to expose these faces. No 121 evidence for the truth or otherwise of this suggestion was observed in the present study.

In additional experiments, the catalyst on which carbon had already being deposited was also tested for coke formation. It was found that no induction period was observed. Carbon was deposited on the catalyst steadily from the very first stage.

The results also show that steam, produced in the oxidative coupling reaction, is present in sufficient quantity to gasify coke formed by pyrolysis of hydrocarbon. When the gases were dried, carbon formation was observed. When they were not, no carbon formation could be seen.

Comparison of the rates of carbon formation in Tables 4.8 and 4.9, Tables

4.10 and 4.11, Tables 4.12 and 4.13 show that carbon formation rates of up to 231.Sµg/sec/g.cat. can be gasified by steam generated in the oxidative coupling reaction. The amount of steam produced from OCM reaction is estimated to be just enough to gasify that much carbon. Although this implies a very fast rate of gasification, the oxidative coupling catalyst in the microbalance basket is basic and, as a result, would be expected to catalyse carbon gasification by steam (Catalyst Handbook, 1970).

At first sight, it is surprising that gasification of carbon is not connected with changes in the gas phase products. However the design of the microbalance reactor is such that sensitivity for gaseous products is not 122 central to the experiments. This arises from the fact that the microbalance sample must hang free in order to record weight changes and this, in itself, allows gas bypassing. The net result is that gas analysis is indicative but not definitive.

Despite this, however, there are some signs of changes in gas analysis when carbon formation and carbon gasification is greatest. Thus, for example, the addition of benzene produces the highest rates of carbon formation ( Tables 4.9, 4.11 and 4.13 ) and, in the presence of steam, the volume percentage of carbon oxides in the products is significantly higher (

Tables 4.8, 4.10 and 4.12 ). Since carbon formation follows the rate of benzene > propylene > ethane, the gasification of carbon in the system takes place in the same order. Figures 4.11 and 4.12 show this general feature.

An attempt was made to test the gasification rate of carbon which had already been deposited on the catalyst by the removal of the drier on line. It was found that the deposited carbon could not be gasified

( Fig. 4.13 ). This could have resulted from the fact that surface carbon had reorganised to form well ordered crystallites. This type of carbon is nearer to graphite than gas phase carbon, having larger crystallites, higher density and lower interlayer spacings. Secondly, the steam generated from oxidative coupling of methane reaction was just enough to gasify the coke produced from pyrolysis. Comparing these two different types of carbon, 123

COx Selectivity 20 ~------,

15 -

10 -

5 I I 800 850 900 950

0 T ( C)

--- Ethane --+- Propylene --¾--- Benzene

Fig.4.11. COx selectivity of oxypyrolysis with 5% additives. 1. No drier used in the system. 2. All the other conditions are listed in Fig.4.7. 124

CO x Selectivity (%) 20.0 ,------,

15.0

10.0

5.0

0.0 ~------~------~------~ uoo 850 900 950 T ( °C)

- Ethane -+- Propylene ~ Benzene

Fig.4.12. COx selectivity of oxypyrolysis with 10% additives. 1. No drier used in the system. 2. All the other conditions are listed in Fig.4.7. 125

Carbon Formation (mg/ g.cat.)

Time (min)

Fig.4.13. Characteristic curve of coke gasification of deposited coke. 1. The drier was removed from the system during reaction. 2. 20 mg of oxidation catalyst in microbalance basket. 3. Hydrocarbons were injected after the OCM catalyst-bed into cracking zone. 4. All the other conditions are listed in Fig.4.6. 126 steam would be expected to gasify the gas phase carbon more easily, because the reactivity of carbon is known to change with time on line, freshly deposited carbon being more easily gasified ( Gwathmey and Cunningham, 1958 ).

Experiments were also completed to compare the coke forming tendencies of oxidised and reduced catalysts. Regrettably the experimental apparatus takes considerable time to settle down, not least because of the induction period for coke formation. As a result, no significant differences in coke formation could be observed. This does not necessarily mean, however, that such differences did not exist. It infers only that coke formation on oxidised and reduced catalysts was similar after the period when the balance settled down ( ea 2 min. ) and the carbon induction period ( 2-8 min. ). There is a very good chance that during this period, the surface equilibrated to the same composition in both cases. 127 CHAPTER 5

CONCLUSION

1. The oxidative coupling of methane ( OCM ) reaction affords higher

yields of C2 products under conditions of high linear space velocity and

elevated catalyst-bed hot-spot temperature.

2. Oxidative coupling of methane reaction produces heat.

3. Pyrolysis of hydrocarbons requires heat.

4. The heat generated by the exothermic OCM reaction can be partially

utilised for the benefit of the endothermic conversion of hydrocarbon

pyrolysis to produce ethylene.

5. The C2 ( and especially ethylene ) selectivity increases and remains

highest if ethane is added to the post-catalyst bed region ( pyrolytic

zone ) after the oxygen has been consumed.

6. Injection of other hydrocarbons between reactors causes some increase

in the ethylene selectivity. 128 7. The increase of the ethylene selectivity after the injection of a

hydrocarbon into the cracking zone is in the order of

ethane > propylene > benzene.

8. The addition of hydrocarbons after the catalyst-bed, together with the

OCM products, into the microbalance reactor causes no observable

carbon formation when steam is present.

9. The total COx amount increases, after hydrocarbon addition into the

cracking zone, in the order of

benzene > propylene > ethane

when steam is present in the system.

10. When steam is removed from the system, carbon formation is observed

on the catalyst in the microbalance reactor.

11. Coke formation rate is in the order of

benzene> propylene > ethane. 129

12. Overall carbon removal is faster than carbon deposition in the oxidative

coupling/pyrolysis system.

13. To prevent heat loss from exothermic OCM reaction, the OCM reactor

and pyrolysis reactor should be joined together as reported by Mimoun

et al. (1990). This has several major advantages:

(a) The need for a separate ethane steam cracker is eliminated because

both oxidation and pyrolysis are carried out in the same reactor

assembly.

(b) Ethane can be used as a thermal quench, thus reducing the

temperature of the effluent.

(c) The hydrogen produced in the pyrolysis zone can be used to

hydrogenate all of the carbon monoxide ( and some of the carbon

dioxide ) resulting from non-selective oxidation, thereby

regenerating methane. This allows the recovery of carbon. 130 REFERENCE

"Catalyst Handbook", Wolf Sci. Texts, London, 1970.

Albright, L.F., "Pyrolysis: Theory and Industrial Practice", Academic Press, NY, 1983.

Anderson, L.R., App. Catal., 1Z, 177-196, 1989.

Anisonyan, A., Dokl. Akad. Nauk., Arrn. Z1:, 24038x, 1971.

Baker, R., Barber, M., Harris, P., Feates, F. and Waite, R., Carbon lQ, 93, 1972.

Baker, R.T.K. and Sherwood, R.D., J. Catal., fil., 378, 1980.

Benson, S.W. and O'Neal, H.E., " Kinetic Data on Gas Phase Unirnolecular Reactions ", NSRDS-NBS No.21, Nat. Bur. Stand., Washington, D.C., 1970.

Bernardo, C.A. and Trimm, D.L., Carbon, H, 225, 1976.

Bethelot, from Badger, E., Progress in Phys. Org. Chern., J, 1, 1965.

Bhasin, M.M., Methane Conversion, 343-357, 1988.

Biscoe, J. and Warren, B.E., J. App. Phys., .Ll, 364, 1942.

Bradley, J.N., Proc. Roy. Soc. London, Ser. A, 337, 199-216, 1974.

Bromley, J. and Strickland-Constable, R.F., Trans. Faraday Soc., .2Q., 1492, 1960.

Brooks, C.T. and Thompson, B.H., 165th Arn. Chern. Soc., National Meeting, Dallas, U.S.A., 1973.

Burch, R. and Tsang, S.C., App. Catal., .Q.2, 259-280, 1990.

Bytyn, W. and Baerns, M., Applied Catalysis, 28, 199-207, 1986.

Castellan, G.W., "Physical Chemistry", Addison-Wesley, 1964. 131 Chen, C.-J., Back, M.M. and Back, R.A., Can. J. Chem.,~ 3580, 1975.

Clark, T.C. and Dove, J.E., Can. J. Chem., .2.1 2147-2154, 1973a.

Dente, M.E. and Ranzi, E.M., Mathematical Modelling of Hydrocarbon Pyrolysis Reactions in " Pyrolysis: Theory and Industrial Practice " by L.F. Albright, Academic Press, 152, 1983.

Edwards, J.H., Do, KT. and Tyler, R.J., Chemeca'89, paper 21C, p 726-733, 1989.

Egloff, G., J. Phys. Chem., J1, 1671, 1930.

Frey, F.E. and Huppke, W.F., Ind. Eng. hem. g 54, 1933.

Geerts, J.W.M.H., Van Kasteren, J.M.N. and Van der Wiele, K., Catal. Today, i, 453, 1989.

Germain, J.E., "Catalytic Conversion of Hydrocarbons", Acad. Press, 1969.

Graham, S., Horner, J. and Rosenfeld, J., 2nd European Comb. Syrnp., Orleans, France, 1975.

Gwathrney, A.T. and Cunningham, R.E., Advances in Catalysis, IB 57, 1958.

Ito, T. and Lunsford, J.H., Nature, 314, 721, London, 1985.

Jones, C.A., Leonard, J.J., and Sofranko, J.A., J. Catal., 103, 311-319, 1987; Energy and Fuels, .112-16, 1987; U.S. Patents 4, 443, 644, 645, 647, 648 and 649, April 17, 1984.

Kaminski, A.M. and Sobkowski, J., Radiochern. Radionanal Lett. J2., 209-220, 1979.

Kearby, K.K., "Catalysis", Ed. Emmett. III 453, 1955.

Keller, G.E. and Bhasin, M.M., J. Catal., 73 9-19, 1982. 132

Kramer, L. and Happel, J., "The Chemistry of Petroleum Hydrocarbons" ( Brooks, B.T., Kurtz, S.S., Boord, C.E., and Schmerling, L. eds. ), Vol. 21, pp. 146-147. Reinhold, New York, 1955.

La. Cava, A.I., Pyrolysis and Thermal Hydrogasification of Hydrocarbons, PhD thesis, University of London, 1976.

Lafitau, H. and Jacque, L., Bull. Soc. hem., France, 4779, 1968.

Lahaye, J., Prado, G. and Donnet, J., Carbon n., 27, 1974.

Leathard, D.A. and Purnell, J.H., Ann. Rev. Phys. Chem., l1., 197, 1970.

Leathard, D.A. and Purnell, J.H., Ann. Rev. Phys. Chem., ll, 197, 1970.

Lecoanet, A., These de I'Universite de Paris, 1968.

Lersmacher, L.H.B. et. al., Carbon .:2., 205, 1967.

Lin, CH.H., Campbell, K.D. and Lunsford, J.H., J. Phys. Chem., 2Q, 534-537, 1986.

Lobo, L.S. and Trimm, D.L., J. Catal., ,f2, 15, 1973.

Luther, K. and Troe, J., Symp. ( Int. ) Combust. ( Proc. ), lZ, 535-542, 1979.

Martin, G.A., Bates, A., Ducarme, V. and Mirodatos, C., App. Catal., .±Z, 287, 1989.

McCarty, J.G. and Wise, H., J. Catal., ,22, 406, 1979.

McCarty, J.G., Wentreck, P.R. and Wise, H., Am. Chem. Soc., Div. Pet. Chem., Prepr., Chicago, III, p 1315, 1977.

McKee, D.W., Carbon n., 453, 1974.

MeCulloh, KE. and Dibeler, V .. , J. Chem. Phys., M, 4445-4450, 1976. 133

Michael, J.V., Osborne, D.T. and Suess, G.N., J. Chem. Phys., 2§., 2800-2806, 1973.

Mimoun, H., Robine, A., Bonnaudet, S. and Cameron, C.J., App. Catal., .:2§., 269-280, 1990.

Mitchell, H.L. and Waghorne, R.H., U.S. Patent 4, 172, 810, October 30,1979 and 4, 205, 194, May 27, 1980.

Nelson, W.L., "Petroleum Refinery Engineering", McGraw Hill Book o., London, 1958.

Niclause, M., Bull. Soc. Chem. France, 1, 1599, 1968.

Otsuka, K., Jinno, K. and Morikawa, A., J. Catal., 100, 353-359, 1986.

Otsuka, K. and Komatsu, T., Chem. Letters ( Chem. Soc. Japan ), 483-484, 1987; and Chem. Letters, 1955-1958, 1986.

Pacey, P.D. and Purnell, J.H., IEC Fundamentals, .11 233, 1972.

Pacey, P.D. and Purnell, J.H., J.C.S. Faraday I,~ 1462-1473, 1972.

Parks, G.S. and Huffman, H.M., "Free Energy of Some Organic Compounds", Reinhold, 1932.

Penninger, J.H.M. and Slotboom, H.W., Erdol Kohle, 12, 445, 1973; Chem. Abstr. Z2, 136271c, 1973.

Makarov, K. and Pechik, V.I., Carbon I, 279, 1969.

Penzhorn, R.D. and Darwent, B.B., J. Chem. Phys., 22, 1508-1511, 1971.

Pratt, G. and Rogers, D., J.C.S. Faraday I, Z2, 1089-1110, 1979a.

Purnell, J.H. and Quinn, C.P., Proc. Roy. Soc. London, Ser. A., 270, 267, 1962.

Quinn, C.P., Proc. Roy. Soc., London, Ser. A, 275, 190, 1963. 134 Quinn, C.P., Proc. Roy. Soc., London, Ser. A 275, 190, 1963.

Quinn, C.P., Trans. Faraday Soc.~ 2543, 1963.

Robertson,R.W.J. and Hanesian, D., Ind.Eng. Chem. Process Des. Dev., 1.1, 216, 1975.

Rossini, F.D. et. al., "Selected Values of Properties of Hydrocarbons", Am. Petroleum Indt., 1947 to date.

Rostrup-Nielsen, J.R., "Steam Reforming Catalysts", Teknisk Forlag A/S, Copenhagen, 1975.

Slotboom, H.W. and Penninger, J.H.M., Ind. Eng. Chem. Process Des. Dev., 1l, 296, 1974.

Snow, R.H., J. Physical Chem, ZQ, 2780, 1966.

Snow, R.H., Peck, R.E. and Von Fredersdorf, C.G., ALCHE Journal, .2., 304, 1959.

Sofranko, J.A., Leonard, J.J. and Jones, C.A., J. Catal., 103, 302-310, 1987; and U.S. Patents 4, 443, 644, 645, 647, 648 and 649, April 17, 1984.

Taiseki, K., Tomoya, S., Kazuhiko and Yoichi, S., Ind. Eng. Chem. Fund.,.§., 374, 1969.

Tamai, Y., Nishiyama, Y. and Takahashi, M., Carbon 2, 593, 1968.

Trenwith, A.B., J.C.S. Faraday I, Z2, 614-620, 1979.

Trimm, D.L., Catal. Rev. Sci. Eng., }2, 155, 1977.

Trimm, D.L., "Comments from A Visit to a Steam Cracking Industrial Plant", Private Communication, 1974.

Walker, P.L., Jr. Rusinko, F. and Austin, LG., Adv. Catal., 11133, 1959. 135 Walker, R.W., "Gas Kinetics and Energy Transfer" ( Ashmore, P.G. and Donavan, R.J., eds ), Vol.2, pp.296-334. Chern. Soc., London, 1976.

Zaslonko, LS. and Srnirnov, V.N., Kinet. Catal., N, 575-583, 1979.

Zdonik, S.B., Green, E.J. and Hallee, L.P., "Manufacturing Ethylene", The Petroleum Publishing Co., Yulsa Oklahoma, U.S.A., 1970.