<<

ETDE-DE—1117

DGMK Tagungsbericht 2001-4

Proceedings of the DGMK-Conference "Creating Value from Light Olefins - Production and Conversion" October 10-12, 2001, Hamburg, Germany

DGMK

German Society for and Coal Science and Technology

Petrochemistry Division Alle Rechte, auch die der Obersetzung, des auszugsweisen Nachdrucks, der Herstellung von Mikrofilmen und der fotomechanischen Wiedergabe, nur mit ausdrucklicher schriftlicher Genehmigung der DGMK.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of DGMK.

Als Manuskript gedruckt.

ISSN 1433-9013 ISBN 3-931850-84-6

Preis: EUR 75,- (DM146,69) zuzugl. ges. MwSt.

Verbreitung und Verkauf nur durch:

DGMK

Deutsche Wissenschaftliche Geseilschaft ftir Erdol, Erdgas und Kohle e.V.

Kapstadtring 2, 22297 Hamburg Postfach 60 05 49, 22205 Hamburg Telefon: (040) 63 90 04-11/33 Telefax: (040) 63 00 736 Bankverbindung: Dresdner Bank AG, Hamburg, Konto 9 085 166 00 (BLZ 200 800 00) Amtsgericht Hamburg 69 VR 6898 DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. DGMK XX

German Society for Petroleum and Coal Science and Technology

Tagungsbericht 2001-4

Proceedings of the DGMK-Conference “Creating Value from Light Olefins - Production and Conversion" October 10-12, 2001, Hamburg, Germany

(Authors' Manuscripts)

edited by G. Emig, H.-J. Kramer & J. Weitkamp

KS002755650 I: KS DE016792478 I

CONTENTS

Page

Recent Advances in the Manufacture of Light Olefins by 7 Steamcracking St. J. Stanley

Steamcracker Revamp Projects: Challenges and Technologies 9 R. Zeppenfeld, R. Walzl

Optimizing Pyrolysis Upgrading 21 V. Coupard, J. Cosyns, Q. Debuisschert, L. Quicke, P. Travers

Maximizing Propylene Yields in Fluid Catalytic and 31 Related Technologies - Solutions for Increasing Propylene Production D. Greer, M. Houdek, R. Pittman, J. Woodcock

Conversion of Problematic Petrochemical-Refinery Fractions to 45 High Value Steam Cracker Feed H. Bischof, W. Dohler, F. Fuder, J. Laege

Preparation of a Synthetic Steamcracker Feedstock from Pyrolysis 49 Gasoline on Zeolite Catalysts A. Raichle, Y. Traa, J. Weitkamp

Conversion of Naphthenes into a Valuable Steamcracker Feed 57 Using H-ZSM-5 Based Catalysts C. Ringelhan, V. Kurth, G. Burgfels, J. G. Neumayr, W. Seuffert, J. Klose

ARINO™ (Aromatic RINg Opening) 65 Technology for Upgrading of Low Value Aromatics to High Value Steamcracker Feedstock H. Fritz, H. Bolt, U. Wenning

Methanol to Olefins (MTO): State of the Art and Perspectives 73 S. Kvisle, H. Reier Nilsen, T. Fuglerud, A. Gr0nvold, B. V. Vora, P. R. Pujado, P. T. Barger, J. M. Andersen

MTP, Methanol to Propylene - Lurgi's Way 85 M. Rothaemel, H.-D. Holtmann

Selective of 1,3- over Supported Gold 97 Catalysts S. Schimpf, M. Lucas, P. Claus II

Valuable Products from Butadiene, Dioxide and 103 V. A. Brehme, A. Behr

Making Olefins from Light Paraffins by Catalytic Dehydrogenation 111 and Oxidative Dehydrogenation K. Harth

The MTBE Issue from the Viewpoint of an Environmental 113 Protection Agency A. Friedrich

Creating Value from Isobutene 115 St. Muller, A. Gammersbach, H.-J. Kramer, F. Kaledat

C4 Fraction - A Raw Material for the Production of C10 Plasticizer 119 Alcohols J. Kolena, P. Moravek, J. Lederer

New PO Processes 127 Th. Haas, W. Hofen, G. Thiele, P. Kampeis

Oxidative Dehydrogenation (ODH) of over Vanadia-Based 131 Catalysts: Probing Active Sites under Working Conditions A. Bruckner, P. Rybarczyk, J. Radnik, G.-U. Wolf, H. Kosslick, M. Baerns

Oxidative Dehydrogenation of over Novel Mixed Oxides 139 A. Hartung, S. Gaab, J. Find, A. Lemonidou, J. A. Lercher

Advances in the Selective Oxidation of C3 and C4 147 R. K. Grasselli

The Influence of the Gas Phase Composition on the Catalytic 159 Partial Oxidation of I. GriBtede, M. Kohler, H.-G. Lintz, H.-C. Schwarzer

Selective Oxidation of Propane on Basic Metal Oxide Catalysts 167 F. Klose, B. Ondruschka, P. Scholz, R. Bdining

Pilot Plant Processing of n- to Maleic Anhydride above the 175 Explosion Limit W. M. Brandstadter, B. Kraushaar-Czarnetzki POSTERSESSION

Olefins Production in Catalytic Pyrolysis of Gas Condensate 183 and Straight-Run Gasoline A. L. Lapidus, F. G. Jagfarov, I. F. Krylov, N. A. Grigor'eva, A. Yu. Krylova

The Mechanism of Olefin Formation from Light Paraffins over MFI 189 Zeolites Kh. M. Minachev, A. L. Lapidus, A. A. Dergachev

Investigations of Mixed MCM-41/MFI Catalyst Systems for the 195 Manufacture of Light Olefins by Cracking of Hydrocarbons A. N. Bhave, A. Klemt, S. R. Patwardhan, W. Reschetilowski

Non-Oxidative Propane Dehydrogenation over Supported Pt-Zn- 203 Catalysts T. Donauer, R. Glaser, J. Weitkamp

Heterogeneous and Homogeneous Processes in Oxidative 211 Dehydrogenation of Propane in a Fixed Bed Catalytic Reactor O. Hein, A. Jess

Selective Oxidative Catalytic Conversion of n-Butane into Olefins 219 C2-C3 at Moderate Temperatures S. B. Kogan, M. L. Kaliya, N. Froumin, M. Herskowitz

Catalyst Development for the Oxidative Dehydrogenation of Ethane 227 and Propane to Olefins Applying an Evolutionary Approach M. Langpape, G. Grubert, D. Wolf, M. Baerns

Optimizing Structural Configurations of Petrochemical 235 Manufacturing Sites J. Fabri

Ethylene Oligomerization over Ni- and Pd-Zeolites 245 A. Lapidus, A. Krylova

Selective Dimerisation of Light Olefins in Biphasic Mode Using 253 Ionic Liquid Solvents - Design and Application of a Continuous Loop Reactor P. Wasserscheid, A. Jess, M. Eichmann

Ethylene as a C2-Building Block for Catalytic Synthesis of Fine 257 Chemicals M. Solinas, G. Francio, W. Leitner

Supercritical Isomerisation of n-Butane over Sulfated Zirconia 263 B. Sander, M. Thelen, B. Kraushaar-Czarnetzki IV

Effect of Sodium Promotion on the Performance of Fe x Oy/Si02 271 Catalysts in the Gas Phase Epoxidation of Propene V. Duma, R. Fodisch, D. Honicke

Enhanced Aromatic Formation in the Methanol to 277 Reaction Using Composite Catalysts D. Freeman, R. P. K. Wells, G. J. Hutchings

Development of a Process for the Production of Glycerol in Tertiary 285 Butyl Ether as Booster on Isobutene Basis A. Behr, L. Obendorf, V. A. Brehme

The Catalytic Wall Reactor as Tool for Kinetic Investigations in the 287 Selective Oxidation of Propene to Acrolein H. Redlingshofer, G. Emig

Quantitative Analysis of and Reformates by GC/TOFMS 289 R. Loscher, R. Hirsch, E. de Armas, R. Parry DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

St. J. Stanley ABB Lummus Global Inc., Bloomfield, USA

Recent Advances in the Manufacture of Light Olefins by Steamcracking

Manuscript was not available by the time of printing

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 7 8 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion*, Hamburg, 2001

R. Zeppenfeld, R. Walzl Linde AG, Hollriegelskreuth, Germany

Steamcracker Revamp Projects: Challenges and Technologies

Role of Steamcracker Revamping in Europe The growing use of synthetics in the consumer goods industry fuels an ever-increasing demand for ethylene and propylene, the major raw materials for industrial chemicals. To meet this demand, the petrochemical industry has to find economical ways of continuously expanding olefins production capacity. Capacity for ethylene production in Western Europe has grown by more than 100% during the last 20 years as shown in Fig. 1. Since 1993, no grass root cracker was commissioned but all new capacity in Europe (about 500 kt ethylene/year) results from revamping projects, of which Linde has carried out a remarkable portion. A new ethylene plant represents a very significant investment. In addition to the large capital cost, there are numerous associated expenses. To be profitable in international market, the annual capacity of a new ethylene plant must be at least 600 to 800 kta ethylene production per year to keep the fixed costs down to an acceptable proportion of the total and so remain competitive. Several reasons contributed to the apparent economic advantage of capacity revamping in Europe: • The moderate market growth (few percent per year) made it difficult, if not impossible, to justify an increase in capacity of that magnitude in one step. Revamping allows incremental new capacity with the economics of a grass-root cracker. • A lot of the European steamcrackers are now more than 25 years old, far older as their original design life. Corrosion is the most important cost factor arising from this situation. Corroded equipment and piping can be repaired or replaced very specifically as part of a capacity expansion project. • Higher efficiencies of modern equipment cannot justify its replacement alone. However, in combination with a capacity increase, better performance improves the revamp project's economics. • Due to major environmental legislative requirements (i.e. BAT) whole furnace sections may become candidates for complete replacements. Increasing of furnace capacity can be easily included in such a project with few extra cost. • The distributed plant ownership and distributed production sites resulted in a situation where no large single investment in ethylene production of more than 1 billion Euro and the related risks could be justified by any player.

9 DGMK-Tagungsbericht2001~4, ISBN 3-931850-84-6 • To obtain the operating permit for a new plant is much more difficult than for a modified plant. The recent mergers in Europe could change this situation to some extent but the risk of potential imports and their influence on the olefin prices originating from the new huge middle-east capacities may leave room only for very few new crackers in Europe. Consequently, revamping will play a continuously important role in the steamcracker business in Europe.

Basic Considerations for a Capacity Revamp The potential for increasing the production capacity of an existing plant depends very much on the individual situation. The major basic considerations are the original design cases of a cracker compared to the anticipated operation (the inherent spare capacity), previously applied revamp measures consuming this, the turnaround schedule of the cracker, the capacity of the infrastructure, the measures required to extend the life of the existing cracker, the feed and product integration scenario and last, but not least the target capacity for the revamp project. From a technical point of view, it is possible to increase the plant capacity of a plant to any desired value. However, from a commercial point of view it makes sense to expand only by a limited amount. This point is determined by a combination of the incremental capacity needed to meet business plans and the optimum specific investment per ton of new capacity. Depending on the situation of the producer, three major capacity targeting scenarios can be found: • Predefined Target Capacity This scenario requires minimum investment for a given incremental capacity and is valid for integrated producers with tight feed and product channels. • • Predefined Investment Budget The maximum capacity for a given investment budget has to be found. Integrated producers with flexible feed/product channels prefer this strategy. • Optimum Economics The target capacity with maximum ROI is anticipated mainly by non-integrated producers with flexible feed/product markets. It is a characteristic feature of any revamp project that additional capacity is achieved by eliminating plant bottlenecks step by step. A number of small bottlenecks and a few large bottlenecks must be overcome to achieve the typical capacity revamp objectives. Fig. 2 shows a typical investment versus incremental capacity curve for a steamcracker revamp. Major investment becomes due by installation of additional furnace capacity and modification or replacement of compressors and distillation towers. Fig.3 translates this into specific investment. Economic target capacities can be found directly below equipment limitations requiring major investment steps. Below a certain critical capacity the optimum point is considerably lower than the cost per ton of capacity of a new train. In addition to the pure investment, the down time required for implementation and tie-in is considerably influencing the economics for larger revamps. Fig.4 is highlighting this effect. Whereas the ROI of a revamp project is high as long as the revamp down time is shorter than the normal turnaround, it drops remarkably if the revamp activities extent the down time and produce additional operating losses. Large revamp projects have been skipped in the past due to unattractive down time.

to Principle Revamp Strategies Principle revamping strategies can be summarized as follows: • Modify feed, recycles and/or furnace yields to unload the separation section. • Make maximum use of existing equipment. • Increase capacity of individual items by modification or by installing additional units. • Modify process, introduce a new process unit - so reducing the load of other sections of the plants. Depending on the incremental capacity and, by this, of the extent of application of the above technical strategies, three major revamp scenarios are typically distinguished: Capacity Creep This scenario utilizes the design margins of the existing plants and includes only minor upgrade of equipment. Up to 15% of incremental capacity can be achieved by this. Very good economics are achieved in this range as no major investment is required and the project is short and easy to manage. Step In - Step Out The incremental capacity ranging from 10 to 70 % (related to the original design capacity) is normally utilized by a so-called Step In - Step Out procedure. Additional equipment units, additional local parallel trains of few equipment and/or additional process units are introduced and the original plant is extended at various locations in the process. Numerous tie-in points are created, the plot requirement situation is complicated and the implementation of the project becomes complex and time-consuming. Down time is critical for large capacity increase with this strategy and the overall risk in cost and performance are high. However, for medium size revamps Step In - Step Out is the most attractive option. Side Cracker Beyond about 160 % of the original design capacity, a complete new parallel train is normally more suitable than a revamp of the old unit. Integration with the existing plant is mainly done for utilities and for few intermediate streams to balance the loads and fully utilize the capacity of the existing installations. Fig.5 compares the above mentioned principle revamp strategies with respect to specific invest, risk, lead and down time.

The Revamp Toolbox The various revamp measures applied can be principally summarized as follows:

Overall Material Balance A major part of a successful revamp is the optimization of the overall material balance of the cracker. Integrated producers and those with flexible feedstock supply may select feedstock with high yields for the desired products and avoid revamping of less essential product routes to keep the revamp investment low. Another option is the upgrade of feedstock by pre-separation or hydrogenation. Recent new processes such as MAXENE or ARINO offer these options.

11 The cracking severity is a further parameter in this respect. Selection of P/E ratio could consider the maximum capacity of one of the major product routes, i.e. C3. Reduction or elimination of internal recycles in the separation section could unload complete plant sections. Recycles of C4 and pyrolysis gasoline may be hydrogenated to increase cracking yields and reduce the plant load. Finally, the elimination of heavier furnace recycle streams by secondary olefin technologies like PROPYLUR or Metathesis can be a successful unloading strategy for the existing plant, thus boosting the overall propylene yield and giving room for further capacity increase in the existing plant. Cracking Furnaces The revamped capacity of the cracking unit is usually achieved by utilization of spare furnace capacity, modification of the existing furnaces or adding additional furnaces. Attention must be paid to the size and number of existing furnaces in order to fit the new furnaces into the decoking cycle of the overall plant. Very often, cracking furnaces that are candidates for revamping are already in need of major maintenance, such as replacement of coils or replacement of the transfer line exchangers. In this situation, retrofitting the existing older furnaces with a modern coil type with a shorter residence time should be considered. This will probably require modification of the fire box and suspension facilities of the furnace, but it does result in higher furnace capacity and higher olefins selectivity, especially for the heavier feeds. Retrofitting of ethane cracking furnaces to operate at higher conversion rates results in a substantial reduction of the cracked gas flow. This does however sacrifice some ethylene yield and requires specific optimization. Applying lower pressure-drop TLE's and extending the decoking cycle by coated coils are further furnace related revamp options with minor effect. Due to major legislative requirements for reducing NOx emissions, the complete replacement of whole furnace sections becomes economic and may be combined with capacity revamping. Standard Equipment Other, more standard equipment like columns, heat exchangers and machinery may be subject to one of the following measures to extend its capacity: • Utilizing of design margins • Shifting of load by change of process parameters or unloading of equipment by installation of other process functions • Replacement of internals/fillings or modification of equipment • Adding of shells/units • Replacement of equipment on same foundation (footprint) Principle Process Modifications Having economically exploited all the revamp techniques for overall material balance and locally to equipment, further capacity improvements can only be achieved by larger changes in the process. However, simple duplication of process units is limited by rapidly increasing costs and potential operating problems. In this situation, it is advantageous to introduce modifications to the original process scheme to allow further capacity increases at lower costs and without having to operate parallel units. Because of the major impact

12 on investment, operability and project execution such creative ideas require substantial experience with the specific type of plant. Some examples are given below

Revamp Case Studies Deethanizer Revamp (Fig. 6) Introduction of a C3 absorber upstream of a front-end deethanizer column combined with lowering pressure in the original tower exhibits several advantages. Not only the relative column load is reduced but the column bottom temperature is reduced and is no longer subject to fouling. The deethanizer reboiler capacity is restored by stretching the temperature difference. Condensing with -40°C propylene refrigerant saves energy compared to the original scheme and reduces load of the ethylene refrigerant cycle. The capacity of the unit is boosted by up to 35% without changing the column internals.

Elimination of Integrated Refrigerant Cycle (Fig.7) A recycle stream to the suction of the cracked gas compressor arises in plants that purify hydrogen cryogenically by vaporizing a low pressure stream. The recycle is integrated to recover the ethylene contained in the stream. By adding a separate compressor to inject this stream deep into the cold train, it bypasses the cracked gas machine and part of the cold train separation equipment. The entire cracked gas and cold train system can be unloaded by up to 15%.

Additional Separation Unit (Fig.8) The load on the central separation sections in the recovery part of the plant (such as the deethanizer, the demethanizer, and the low temperature section) can be substantially reduced by the installation of a complete new process unit. The introduction of a soft separation steps debottlenecks the low temperature section, the demethanizer, the deethanizer and the C2 splitter as well as the refrigerant cycles. The soft separation achieves this recovering a methane-free C2plus stream and a C3-free C2minus stream in the middle of the chilling train.

Capacity Revamp by Integrated Side Cracker (Fig.9) An interesting side cracker option economic already for medium range revamps (+15%) of large production sites (> 1000 kta ethylene) has been studied recently and is applied in one place. The idea is to no longer process the C2 and C3 recycle streams in the main plant. A separate new train is built in parallel to the existing plant to process these stream. By means of this procedure additional naphtha can be cracked in the old plant, keeping there the ethylene production constant. Only that plant sections processing C3 and heavier fractions will experience higher load and may be to subject to revamp measures. The new side cracker train is a simple gas separation unit, producing only ethylene. Heavy fractions are returned to the main plant. The side cracker can also be fed with additional C3 LPG, or refinery off gas streams, or other materials that do not contain the heavy fractions resulting from liquid feed cracking. Such additional streams are suitable to boost the capacity of the side cracker and to make this installation more economic.

13 Project Execution Revamp projects exhibit specific challenges not typical for grass-root projects. Some of them are listed below: • The technical concept must consider the principle project schedule. Only those solutions are suitable which can be implemented or at least tied-in in the available down time frame. In some cases, an early scheduled shutdown has to be considered for the major tie-in work. The long-lead new units are then completed afterwards. • A turn-key lump sum EPC bid is available only after a complete basic engineering design because of the high risk involved in piping revamp requirements. • As a revamp project is often combined with measures for life extension, energy savings, environmental improvements as well as improving the operability, the scope of the project can easily grow in an unforeseeable manner. Therefore, the project plan has to consider this and a tight control of the scope is required to keep the original planning for cost and schedule. • Matching the turnaround schedule is crucial for a revamp project. The project schedule is strictly dedicated to this target. Any delay in restarting the plant can torpedo the revamp project's economics as not only the incremental capacity is bothered but the whole production of the plant. • A huge number of activities (revamp and maintenance) have to be carried out during the short turnaround period involving thousands of workers. A tight and careful turnaround planning and organisation is absolutely required therefore.

Each revamp project has its own very specific steps and dynamics. However, two extreme project execution strategies may be considered to highlight the various aspects. On the one hand, the classical systematic approach may be called “Low Risk Execution" and is outlined in Fig. 10. Initially, an extensive study phase, typically carried out by more than one technology licensor, makes sure that a variety technical strategies are explored and compared and, as a consequence, the most economic target capacity is defined. This phase can take more than one year. A +- 20% investment estimate is available for the finally competing solutions. Even the basic design will be carried out twice to keep the competition for the EPC bid open. Up to 48 month are required to complete this approach. The other extreme is called “Short Track Execution ” and is shown in Fig.11. By an integrated conceptual study, immediately followed by a basic design and overlapping EPC activities, such as ordering of long-lead items, a medium size capacity revamp can be completed within 25 month. EPC is carried out in an open-book approach to compensate for the missing competition. A technology licensor well aquainted with the plant specifics, a high level of confidence between owner and contractor as well as a close co-operation of the parties is required to meet this schedule. A comparison of both approaches shows that the economic advantages of a much shorter project duration overrules in most cases the draw-backs of less-optimized technical solutions and a lack of contractor competition. In reality, the long time frame of the low risk approach often leads to various changes in the design basis. This moving target consumes a lot of the technology optimization potential of this strategy. Most of the real revamp projects use approaches somewhere between the above mentioned extremes. In addition, due to unfavourable study results or changed market

14 conditions, most of the launched revamp projects are never reaching the implementation phase.

Conclusions During the last decades in Europe revamping played a predominant role for new steamcracker capacity. This resulted from the moderate market growth, the distributed industry structure in combination with the advantages of integrated life extension measures for aged plants. Defining the optimum target capacity is the key for good economics and guides the principle technical strategy. In addition to a large toolbox of well proven and straight ­ forward revamp methods innovative ideas are required for larger capacity steps. Each revamp project is different and requires relevant experience. Professional planning and execution is even more important than for a grass root project. A short track execution is most attractive but requires a competent and reliable contractor. Linde has carried out a remarkable number of revamp projects in the last twenty years also for plants not originally designed by Linde. Some of these projects were turnkey contracts, others included Linde's participation only in the conceptual and basic engineering phase. In addition to these projects, Linde has executed dozens of revamping studies. Revamping of ethylene plants has become an important business branch of the company, as strong as the design and construction of new plants.

25 t Other Revamps 20 .1 Linde Revamps 3 Name-Plate o o o 15 X l Revamp Hu 10 10 30% a. to Revamp o 5 15% I G 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000

partially estimated values Figure 1 Revamp History Western Europe

15 Increase in Ethylene Production (kta)

Figure 2 Investment per Capacity Increment

\ New Train

E 0) n K

o Revamped Plant Target 0) Capacities Q. CO

Increase in Ethylene Production (kta)

Figure 3 Specific Investment per Capacity Increment

16 Target Revamped Plant Capacities

New Train

Increase in Ethylene Production (kta)

Figure 4 Influence of Down Time on ROI

Step-in Creep Side Step-out Cracker

A Capacity (%) < 15 10-70 > 60

Specific Invest < 500 400- 700- ; (Euro/t C2H4) 800 900 Risk Low High Low

Lead Time (month) 20-30 23-45 35-40 irsci. Project Development Down Time (weeks) 2-4 4 - 6 <2

Figure 5 Revamping Strategies

17 Figure 6 Deethanizer Revamp

Ram ...... i

*0. Cold Recycle

4 H2 fCH4

T„ Cold Train & Cracked Demethanizer r Gas I C2 T Splitter L-> C2H6

Figure 7 Elimination of Integrated Refrigerant Cycle

18 Crjutod Gns>

Hydro- Pre ­ low Temp -> Coohm XI

Deethanizer

Hydro Pro- Cl /C3 1 ow romp gc.ri.ihon Cooling bu|MMticn cooling

£XC2H4

Deethaniser C2 Splitter : 0i\ L* C3+: i ■;f::y,XX C2H6; ^

Figure 8 Additional Separation Unit

-12%

Cracking Fuel Gas Naphtha Recovery + 15"-- Ethylene Propylene +14 % C4, PyGas, PyOil + 15%

Fuel Gas + 8K Recovery Section Ethylene + 15 %

Refinery Off Gas

Figure 9 Side Cracker Revamp

19 j~]j Actual Plant Status Assessment Study

j 4~] Scouting Study (i- 30%)

Pre-Basic Study (+- 20 %) Target Cap acity

i 2~] Basic Design (+-10%) <^> Invest Deci sion EPC Bi Iding

EPC (TKLS) Turnaround Shutdown Q]

0 12 24 36 month 48 Figure 10 Low Risk Execution

Conceptual Study (+- 25%) Target Capacity

Basic Des ign

!. EPC Proposal Invest Decision EPC (Open Book) [] Turnaround Shutdown

0 12 24 36 month 48 Figure 11 Short Track Execution

20 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

V, Coupard*, J. Cosyns**, Q. Debuisschert**, L Quicke**, P. Travers** *Kinetics and Catalysis Division, IFF, France **Axens (IFF Group Technologies), France

Optimizing Pyrolysis Gasoline Upgrading

Pyrolysis Gasoline Upgrading

The trend to more stringent environmental regulations concerning gasoline has led to a reduction in the amount ofpygas in the gasoline pool; consequently, new pygas upgrading routes have been developed. This article illustrates several examples of attractive processing schemes and catalysts for pygas upgrading.

^Introduction In addition to producing the many basic building blocks for the polymer industry, naphtha steam­ cracking yields significant amounts of olefin and aromatics-rich gasoline (pygas). Table 1 shows a typical pygas yield and composition.

Table 1 Typical naphtha cracker pygas (C5-200°C) yield and composition Total yield, wt % of cracker feed 22 Composition, wt% Paraffins + naphthenes 11.8 Olefins 5.5 Diolefms 18.1 2: 13.9 7.2 3 C9+ Aromatics 12.5 Total Aromatics 64.6

Pygas has an excellent research octane number due to its high aromatics and olefins contents. Until recently, pygas was often used as a gasoline pool blending stock after undergoing selective hydrogenation to eliminate gum-forming compounds such as diolefins, styrenic compounds and indenes. The trend to more stringent environmental regulations concerning gasoline has led to a reduction in the amount of pygas that can be added to the gasoline pool. The European specifications planned for 2005 will limit this practice even further, due to: • high benzene content compared to the 1 % volume maximum in the gasoline pool • sulfur content is generally higher than the proposed 50 ppm limitation • Pygas is rich in light olefins having low motor octane numbers (MON).

IFP has developed pygas upgrading routes with the following goals: • Upgrading pygas while avoiding the production of negative value products, i.e., those with a lower value than that of the steam-cracker feed • offer a wide variety of processing options • offer the most profitable schemes for various site-specific constraints.

Several examples of attractive processing routes and new catalysts are presented for the following applications: upgrading C$ diolefins; complete C5 olefin conversion; benzene production; C6 to C9 aromatics production; increased performance and service life of existing applications.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 21 2.Processing Routes for Upgrading C5 Diolefins Steam-cracker C5 cuts contain unsaturated hydrocarbons that can be converted to highvalue-added derivatives. The composition given in Table 2 shows that two components, isoprene and cyclopentadiene are present in reasonably high concentrations and have highvalue-added potential.

Table 2 Boiling points, octane numbers, and weight per cent values for compounds found in steam-cracked C5 cuts , MON C Composition, °C wt%

3 -methyl-1 - 20.2 97.5 - 0.5 27.9 92.3 90.3 8.5 1- 30.0 90.9 77.1 2.7 2-methyl-1 -butene 31.2 102.5 81.9 6.0 Isoprene 34.1 99 81 15.4 n- 36.1 61.7 62.6 12.3 cis- trans 2-pentene 36.4/37.0 98 80.0 2.0 2-methyl-2-butene 38.6 97.3 84.7 4.0 Cyclopentadiene 41.0 103 86 20.0 1,3-pentadiene 42.1 - - 14.0 44.3 93 70 13.4 49.3 101 85 1.2

Isoprene and cyclopentadiene are not easily purified by distillation due to the small differences in boiling points; however, cyclopentadiene is easily dimerized to higher-boiling dicyclopentadiene (DCPD). DCPD can be separated from the other Cs components by simple distillation. Global consumption of DCPD is forecast to grow by around 8% per year until 2005 after previously growing at a rate of 10% per year. There are several DCPD derivatives, including hydrocarbon resins and unsaturated polyester resins. Most of these are marketed as two grades of purity, 92% and 83%. IFP's selective cyclopentadiene dimerization process for the raw C5 cut is shown in Figure 1. The raw pyrolysis gasoline (RPG) is first depentanized to produce a C5 cut free of heavier components that would adversely affect DCPD product purity. The C6+ cut is sent to a hydrogenation plant.

Crude DCPD

C5s to Isoprene Extraction

C5 Saturates to Steam Cracker Olefins and Sulfur-Free Aromatics

Figure 1 - Upgrading C5 Diolefins

22 After the CPD dimerization, the remaining isoprene-enriched C, cut can be sent either to an extraction plant for isoprene recovery or to a hydrogenation plant. The DCPD process has the flexibility to cope with varying product targets; for CPD conversions of 85 to 90%, DCPD purity is 83%. In the same unit, a DCPD of 91-92% is obtained by limiting the CPD conversion to 50-60%. A unit put on stream in 1991 produces both 83% and 92% purity DCPD depending on the application.

Upgrading the isoprene-enriched C5 cut After the CPD separation step, theremaining C5 cut is concentrated in isoprene and pentadiene (I-P enriched C5 cut) and can be sent to an extractive distillation unit similar to that used for 1-3 butadiene extraction. There, an isoprene-pentadiene stream is removed from the C5 olefins and paraffins. Isoprene and pentadiene can then be separated by superfractionation.

When it is not necessary to upgrade the I-P enriched C, cut, the most economical route is to send it to a hydrogenation plant for complete C, olefin saturation. The saturated hydrocarbons obtained can then be recycled as feed to the cracker furnaces.

3-Processing Route for Complete C5 Olefin Conversion Previously, the C5 cut was separated from the aromatic rich C6+ cut after hydrogenation and sent to the gasoline pool. This practice is now limited due to the poor motor octane quality of this olefinic stream, see Table 2. The high vapor pressure of this cut, as indicated by the boiling points of the components, is undesirable as well. Therefore, a lower value is attributed to the olefinic C5 cut compared to that of the steam-cracker naphthafeed. Consequently, the C5 stream is recycled to the steam-cracker furnaces.

Although recycling this C5 stream to the cracker is a convenient way to dispose of the cut, it comes at the expense of a lower overall ethylene yield. Ethylene yields from naphtha components generally decrease in the following order starting withthe highest yield: n-paraffins > iso-paraffins > naphthenes > olefins > diolefins > aromatics. There is therefore an incentive to saturate the C5 olefins, because it replaces naphtha feed without reducing ethylene yield. In the past, the most common method was to send the C5 cut, mixed with the aromatics stream, to a second-stage hydrogenation unit dedicated to olefins saturation and hydro-desulfurization. This processing scheme (Figure 2) results in complete C, olefin saturation.

Cyclopentane production Under certain circumstances, it may be advantageous to capitalize on the high cyclopentane content of the treated C5 pyrolysis cut to produce cyclopentane. This product can represent one-third of the total weight of the C, cut. Cyclopentane has a high market value as a chlorofluorocarbon (CFC) blowing agent replacement in the manufacturing of rigid polyurethane foams. Cyclopentane can be separated by distillation from the lower boiling hydrocarbons,i.e., isopentane and pentane). If necessary, benzene traces present in the cyclopentane can be removed by a simple hydrogenation to .

23 F.G. C5sto C5-C9 H2S Fuel Gas Fumaces^ j

RPG agMgg Hydro ”

LJ r C10+ (Optional) c6-c9 Aromatics

Figure 2 - Conventional RPG hydrogenation scheme with complete saturation of C5 olefins

An economically improved scheme that provides a C5 cut containing about 75% saturated hydrocarbons is now available, Figure 3. Thiscut is well suited for recycle to the cracker furnaces.

F.G. F.G. H2S C5s to 'l Steam Cracking RPG Furnaces @ 'iir Hydro |

Aromatics (Optional)

Figure 3 - Improved RPG hydrogenation scheme for the production of 75% saturated C5 cuts

The improved first-stage hydrogenation (GHU-1) process is very similar to that employed conventionally. The major difference is that the catalyst, LD 365, features much higher diolefins and hydrogenation rates as well as high activity for light olefins hydrogenation. The performance obtained for a typical C5 cut is reported in Table 3:

Table 3 - C5 product composition from an improved GHU-1 Compound wt% Isopentane 14 n- 22 5 Pentane 26

24 Cyclopentene 0 Cyclopentane 34 100.0 ~ Total saturated Hydrocarbons: 74 wt%

Shown in Table 4 are the results of an economic comparison between the two C5 processing schemes illustrated in Figures 2 and 3.

Table 4 Economic comparison between C5 processing in conventional and improved GHU-1 schemes Conventional Scheme Improved Scheme ISBL Investment cost, million Euros 17.4 15.9 Utility cost ratio, 100 65 Catalyst cost, Euros/ton of processed feed 0.256 0.264 Bases: 50 t/h RPG, 12 t/h C5 cut

4.Hydrogenation route for combined butadiene, butenes and C5 olefins cut In geographical areas where there is no market for butadiene or butenes, the debutanizer can be eliminated from the steam-cracker separation train and the combined C4 and C5 stream can be sent to a full hydrogenation unit. The product from this unit contains at least 85% saturated hydrocarbons and can be recycled to the cracking furnaces to produce more ethylene and propylene with less naphtha feed. IFF has licensed several units based on this approach, several of which are in commercial operation.

S.Pro cessing Routes for Benzene Production Raw pyrolysis gasoline contains over 60 wt% of C6 to C]0 aromatics, over half of which isbenzene. A typical C6 cut contains 75% benzene and - after removal of diolefins, olefins and sulfur - is an excellent feed for extractive distillation processes. The C,+ gasoline cut has an excellent octane rating; typical RONs and MONs are 102 and 88, respectively. Therefore, this cut could be advantageously sent to the gasoline pool after hydrodesulfurization. When demand for benzene is high, the C7 + cut can be sent to a hydrodealkylation process to convert the to benzene. Improved schemes and technologies are available for two major options: • high purity benzene/C,+ cut for thegasoline pool, Figure 4 • maximum benzene production, Figure 5.

Option 1 In this scheme, the C6+ product from the GHU-1 is re-run to remove the heavy aromatics. The re­ run step can be avoided by good control of thegasoline end point in the primary distillation section of thesteam-cracker. The second stage hydrogenation step (GHU-2) produces a sulfur and olefin-free cut, rich in benzene. This cut can be sent directly to a benzene extractive distillation unit. The C6 cut from the

25 GHU-2 typically contains less than 0.5 ppm thiophene, less than 0.2 ppm total sulfur and an acid wash color (AWC) of less than one. The extracted benzene requires no additional processing, such as clay treatment. The C7 + cut contains less than 5 ppm sulfur, and is an excellent blending stock for the ultra-low sulfur gasoline pool.

C6sto C5s to S. C. Furnaces Benzene Extraction F.G. C.-C, F.G. n RPG '1st Stage Hydro.;. Cr-C, Gasoline TJ Pool L Aromatics (Optional)

Figure 4 - High purity benzene / C7 + to gasoline pool

Benzene Product C5s to S. C. Furnaces

Recycle C10+ (Optional) Heavy Aromatics

Figure 5 - Maximum benzene production

Option 2 The block flow diagram in Figure 5 shows how maximum high purity benzene production can be achieved. The C6+ cut is sent to a hydrodealkylation (HDA) unit. This non-catalytic process shows distinct advantages over other HDA processes. The technology produces highmolar yields of high- purity benzene product. Unlike catalytic HDA processes, the IFF HDA process operates without removal of the H2S produced in the GHU-2. Therefore, the HDS reactor is integrated completely in the HDA unit; consequently the second stage unit is reduced to a simple reactor system integrated in the HDA unit's heating train and using the same H, circuit.

26 The benzene product from the HDA unit contains traces of diolefins, olefins and, occasionally, thiophene that are reformed in the HDA reactor by the addition of H2S and under high H2S partial pressures.

IFP's unique hydrofining technology enables complete hydrogenation of trace diolefins and olefins in the HDA product as well as conversion of thiophene to thiophane. The hydrorefining reactor is easily integrated in the HDA unit's cooling train. The reactor product is cooled further and sent to the separation train stabilizer and a benzene tower that produces high purity benzene having an acid wash color near 0 and a thiophene content less than 0.5 ppm. The thiophane produced in the hydrofining reactor is removed in the C7 + cut, because its boiling point is close to that of toluene, and recycled to the second stage hydrogenation unit where it is destroyed. This simple and efficient technology avoids the need for clay treatment and its associated cost and spent clay disposal problems.

6.Route for High Purity C6 to C,„ Aromatics Production The objective of the scheme, shown in Figure 6, is to maximize production of high purity C6 to C9 aromatics. The C6+ cut is sent directly to the GHU-2. The second stage is designed for complete olefin hydrogenation and desulfurization of the aromatic cut. Typical performance obtained is given in Table 5. The C6-C9 aromatic rich cut is suitable for production of high purity benzene, toluene, and C9 aromatics.

C5s to SC Furnaces F.G. F.G. H2S

High Purity C6- Cg Aromatics

Figure 6 Processing route for high purity C6 to C10 aromatics production

Table 5 C6 - C9 cut characteristics after second stage hydrogenation value 0 Bromine index, mg/100 g 100 Sulfur, ppm <1 Thiophene, ppm <02 Acid Wash Color (AWC) C6 to C9 aromatics recovery, 99.5 % Benzene recovery, % 99.7 C6 cut Bromine index 20 C6 cut AWC 1-

27 7.1ncreased Performance and Service Life in existing units IFF has been committed for over 30 years to the continuous improvement of its processes and catalysts as well as to meet the evolving needs of the petrochemical market. Our first pygas hydrogenation unit (GHU) was put on stream in early 1967. Some current market trends are given below: • revamping to increase steam-cracking capacity • increased hydrogenation unit capacity with minimum investment • feed diversification. This continuing trend results in new contaminants being introduced in steam-cracker effluents, whichmust be taken into account.

First stage hydrogenation The heart of the GHU is the reaction section. Besides making mechanical and physical modifications such as the installation of improved pumps, distillation towers, distribution trays and quench boxes, the replacement of existing catalyst by more active, stable or poison-resistant catalyst can offer significant economic benefits. To this end, a full line of hydrogenation catalysts has been commercialized recently. The previous generation palladium LD 265 and nickel LD 241 catalysts, which have been the industry's benchmark products, have undergone major improvements and led to new generation LD 365 and LD 341 products. Their major characteristics and performance are compared in Table 6.

Table 6 Relative catalyst performance for first stage pygas hydrogenation Trade name Metal Relative Activity Relative LHSV Cycle length, per volume years * LD 241 Ni 1 1 0.5 LD 265 Pd 2 1.5-2 0.8-15 LD 341 Ni 2 15-2 0.8-1.5 LD 365 Pd 3 2.5-3 1-2 * Values obtained on full range gasoline i.e. C5-200 °C.

These new generation catalysts show remarkable performance improvements over catalysts that have been the industry's standards. Table 7 shows how thenew generation catalysts provide revamping and grassroots solutions for all petrochemical applications.

Table 7 Choosing the most economically attractive catalysts for various applications

Case Previous Catalyst New Catalyst Unit revamp Up to 150% capacity LD 241 LD 341 New feed with contaminants LD 265 LD 341 Grassroots unit Uncontaminated feed LD 365/341 Contaminated feed LD 341

If thepygas contains temporary poisons such as As, P, or Hg or permanent poisons such as Si or Pb,

28 the nickel catalyst, LD 341, is the product of choice. Because of its high metal content compared to palladium catalysts, LD 341 has a cumulative tolerance to poisons that is higher by a factor of five to ten. Our accumulative experience in pygas hydrogenation enables us to propose the best alternatives for all site-specific cases.

Second stage hydrogenation Good operation of the GHU-2 for complete olefin saturation and hydrodesulfurization depends on the performance of the operation of the first stage. For several years IFF has commercialized dual catalyst systems which ensures long cycle lenghts. The first bed contains a formulation that is dedicated to olefin hydrogenation. This product significantly limits polymer formation. Polymer formation, which is the main source of plugging and pressure drop problems, is avoided when employing this catalyst. The second bed contains a selective HDS catalyst, which avoids aromatics hydrogenation. Thehigh stability and high poison tolerance of the dual catalyst system are features that enable the elimination of a spare reactor and thus reduced investment costs.

S.Conclusion Interest in pygas upgrading is strong due to changing gasoline pool requirements. With more than 30 years experience in pygas process development and application, and over 60% of the market share, IFF has acquired a vast expertise that has led to a complete portfolio of pygas hydrogenation technologies that cover many different applications. A full line of high performance catalysts that meet most site-specific profitability requirements accompanies.

29 30 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

D. Greer, M. Houdek, R. Pittman, J. Woodcock UOP LLC, Des Plaines, USA

Maximizing Propylene Yields in Fluid Catalytic Cracking and Related Technologies - Solutions for Increasing Propylene Production

Abstract

Potential new sources of propylene in Europe are becoming more important as the deficit of regional production capacity becomes more evident. There are several technology options available to regional producers to help solve this capacity deficit. The technology solution of choice depends on many factors. Some of the key producer factors include feedstock availability and cost, characteristics of existing assets, ability and willingness to participate in different product markets, proximity to assets thatcan be integrated, and ability to integrate across business lines and companies. It is clear that propylene demand js growing at a higher rate than ethylene demand, and propylene supplied from existing ethylene machines (steam crackers) will not be sufficient to meet future demand. A number of technologies, some new and some already proven, have been introduced to enable producers to address the propylene shortfall. These technologies enable construction of new grass-roots plants with higher propylene yields than conventional technology. The technologies also enable higher propylene yields to be achieved from existing plants. The attractiveness of any given technology will depend on how the characteristics of that technology line up with the particular situation of the producer, including thekey factors listed above. A good match between producer characteristics and technology characteristics can result in competitive advantage over other regional or global producers. This paper describes some of the key characteristics of propylene producing technologies and then focuses on increasing propylene yields from FCC units. A comparison of alternatives is provided along with an in-depth review of the UOP PetroFCC™ technology.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 31 Worldwide Market Overview

Worldwide polymer and chemical grade propylene demand is expected to grow by an average of 5 to 6% per year over thenext several years1. Propylene demand growth continues to be driven primarily by polypropylene demand. The recent slowdown in the global economy has had an impact on 2001 demand growth for propylene as well as other petrochemicals, but there continues to be good fundamentals behind the drivers for high long ­ term propylene growth. The major source of propylene production currently is steam cracking, with approximately 66% of worldwide chemical propylene supplied from steam crackers. Approximately 32% is supplied from fluidized catalytic cracking units, and the remaining propylene is supplied from propane dehydrogenation units. The primary driver for ethylene steam cracking plants is ethylene demand, and ethylene demand is growing at 4 to 5% per year. It is interesting to note that the demand growth for propylene is higher than that of its main co-products: ethylene, gasoline and distillates. Conventional steam crackers are not expected to meet the growing demand for propylene. An increasing percentage of steam crackers are being constructed that use lighter feedstocks that have lower propylene yields. In particular, most of the new steam crackers being planned in the Middle East are expected to use ethane feedstock 2. Ethane based steam crackers produce much less propylene than liquids based steam crackers, so this will exacerbate the imbalance between propylene demanded and that supplied by steam crackers. It is therefore necessary to investigate and implement alternative methods of producing propylene. This includes increasing propylene production from existing plants as well as construction of new grass-roots plants.

West European Market Overview

European olefin producers currently face a particularly challenging situation with respect to propylene. Global and regional market conditions are converging to create a regional shortage of propylene that will require innovative solutions, including implementation of new technology. Some of thekey market conditions to consider include thefollowing: • There is already a net propylene supply deficit in Europe. • A growing amount of polyethylene is projected to be imported from planned low- cost ethane cracker complexes in the Middle East. • It is not clear if or when a new naphthacracker will be constructed in Europe. • Propylene demand continues to grow at a faster rate than ethylene, gasoline, and distillates. • FCC units have already been tapped for petrochemical grade propylene where it makes the most logistical and economical sense.

There is no clear existing domestic source of incremental propylene to meet existing and projected domestic demand. Meeting domestic propylene demand with domestic sources will likely require implementation of new technology or technology that has not been widely adopted in Europe. Given the highcost of shipping propylene, it makes sense for European companies to evaluate alternative methods to produce the propylene domestically.

32 Propylene Technology Options

There are a number of technology solutions that may be implemented to solve the propylene capacity deficit in Europe. Key solutions to consider include: • Oleflex™ process (propane dehydrogenation) • UOP/Hydro MTO™ process (methanol to olefins)) • FCC • PetroFCC process

These technologies all have different characteristics thataffect their applicability for certain situations and producers. It is tempting to make generalizations about the best technology for producing propylene, but the reality is that thereis no single best solution for every producer. It is instructive, however, to compare the characteristics of thedifferent propylene technologies. One of the key characteristics is feedstock. The major propylene producing technologies to consider for West Europe and the feedstock type for each technology are included in Table 1.

Process: Oleflex MTO FCC/ Splitter PetroFCC

Feedstock Propane X Methanol X Vacuum and Atmospheric Gas Oil X X Naphtha X X Table 1: Propylene TechnologyFeedstocks Another key characteristic is the product slate produced. The key products that impact producer economics are included in Table 2.

33 Process: Oleflex MTO FCC/ Splitter PetroFCC Naphtha Cracking (Provided for Product Comparison) Refinery Propylene X X Chem/ Polymer Propylene X X X X X Ethylene X X X Butylene X X X X Aromatics X X Gasoline X X X Hydrogen X X Butadiene X Fuel Oil X X X Fuel Gas X X X Table 2: Propylene Technology Key Products

Understanding the key feedstocks and products helps us begin to sort out how a particular technology fits with a particular producer ’s situation. Some general attributes to consider are included in Table 3.

Process: Oleflex MTO FCC/ Splitter PetroFCC

Attribute

Impact of High Oil Price on Feedstock Price - - Increase Increase Impact of High Oil Price on Product Price Increase Increase Increase Increase

Fit with Revamp of Existing Refining Assets - - Good Good

Fits a Refinery/Petchem Integration Strategy - Yes Yes Transportation Fuel is Key Driver No No Yes No Number of Key Products Low Low High High Transport Cost of Feedstock High Low Low Low

Table 3: Technology Characteristics

Mapping the technology characteristics as in Table 3 provides a method to begin to evaluate which technologies fit best with a given producer and what must be accomplished to successfully implement a new technology. For example, there is a lower degree of product complexity associated with theOleflex process compared to the PetroFCC process. A polyolefins producer can easily back-integrate to an Oleflex unit, but they must choose the location carefully because of the high cost of shipping the feedstock. The PetroFCC process, on the other hand, fits very well witha strategy of revamping existing refining assets, but it requires handling a somewhat more complex slate of products. Thisis just the beginning of developing a decision process to choose which technology to implement. There will be other

34 technologies to consider as well, such as olefin conversion processes 3. A global producer might choose a different technology (and location) to implement than a local producer, for example. A refiner may find thatthey can utilize existing assets to implement a cost- effective, competitive FCC revamp to produce propylene. At the same time, a polypropylene producer who wishes to supply a specific region may find that they can competitively implement an Oleflex project because they have access to competitively priced feedstock. There is certainly room for a variety of propylene technologies in Europe. The choice for a given producer ultimately depends on how the characteristics of each technology line up with their particular situation. A good match between producer characteristics and technology characteristicscan result in competitive advantage over other regional or global producers.

FCC Propylene Technologies

As the 21st century begins, fluid catalytic cracking (FCC) is the dominant worldwide refinery conversion process and the principal producer of motor fuel. However, the FCC process is not expected to be limited to that role in the future. For instance, over the last 10 years FCC refiners have steadily increased their production and recovery of propylene to meet the world ’s increasing demand for propylene. As more and more splitters were added to recover chemical and polymer grade propylene, many refiners adopted the use of ZSM-5 additives to boost the yields of FCC propylene. Although some refiners have elected to increase riser temperature or change their FCC catalyst formulation to increase propylene yields, these options are generally less attractive. Increased riser temperatures promote increased gas make, and catalyst changes often produce only minor shifts in propylene. ZSM-5 based additives have become the dominant means for increasing light olefin yields in an FCC. Continued additive technology development has also helped spur the use of these additives. ZSM-5 based additives have more than doubled their activity since first becoming available 4. UOP has recently leveraged its FCC experience and know-how to develop technologies that significantly expand upon the FCC operator ’s propylene options. One option is UOP’s PetroFCC process, which is a new FCC process that targets the production of propylene and other petrochemical feedstocks rather than fuel products. The PetroFCC process produces very high yields of light olefins and aromatics. UOP has also expanded its RxCat™ technology to compliment the use of ZSM-5 additives for the incremental production of propylene from existing FCC units.

Incremental Propylene from RxCat

In a conventional FCC, regenerated catalyst flows from the regenerator to the wye assembly at the base of the riser. The catalyst is then lifted via fluidization to the feed nozzles where the hot regenerated catalyst, 1300-1350 °F, contacts fresh feed. In the RxCat concept, some of the spent catalyst is recycled to a MxR™ mixing chamber, which replaces the traditional wye assembly. Inside the MxR chamber the relatively cool spent catalyst is thoroughly blended with the hot regenerated catalyst. The catalyst blend (1100 -1200 °F) is transported up the riser where it contacts the atomized FCC feed at the feed nozzles. The catalyst is quickly separated from the product using UOP’s VSS riser termination technology. As part of the VSS design, the catalyst is pre-stripped to remove hydrocarbons before they

35 have a chance to further react to dry gas and coke. A portion of this pre-stripped catalyst travels through the spent catalyst stripper to the regenerator while the remaining portion is recycled back to the MxR chamber to continue the process. RxCat’s cooler catalyst blend temperature allows for a much higher riser cat/oil ratio for a given riser temperature compared to conventional operations. The increased cat/oil promotes additional riser conversion despite the use of spent catalyst. UOP studies have confirmed that today ’s spent FCC catalyst maintains significantly high levels of conversion. Two factors may help explain this unexpected spent catalyst activity. First, today ’s FCC catalysts have relatively high initial activities and lose less activity upon coking compared to earlier catalyst formulations. Second, advanced FCC hardware such as improved feed nozzles, termination devices, and stripping technology significantly reduces riser delta coke. Consequently, this spent catalyst from modem FCC unit is “less spent. ” UOP has been one of the leaders in developing such state-of-the-art technologies and currently offers elevated Optimix™ feed distributors, VSS™ or VDS™ vortex separation riser termination devices, and improved spent catalyst stripping technology which is especially effective at high catalyst flux rates. These technologies result in such improvements as enhanced catalytic cracking and reduced thermal cracking, nearly eliminating post-riser, non-selective, back-mixed cracking 3. Other benefits include improved selectivity to valuable liquid products, reduced dry gas yields, and reduced delta coke. (Delta coke is theamount of coke deposited per pass of catalyst.) In addition to increased conversion from higher riser cat/oil ratios, numerous other advantages in both new unit and revamp applications result from the RxCat technology. The lower catalyst contact temperatures associated with RxCat technology have been shown to substantially reduce dry gas yields and the regenerator burning kinetics are often improved while reducing inert carryover. Finally, combining RxCat with ZSM-5 additive has been shown to boost both FCC propylene and butylene yields by 0.5-2 wt-% and 0.5-1.0 wt-% respectively. This increase in light olefins is also associated with a substantial increase in gasoline and decrease in dry gas. As a result, RxCat represents an exciting option for refiners currently using ZSM-5 who are interested in further increasing light olefin yields. ZSM-5 based additives have proven to be very effective at increasing propylene yields within an FCC unit. The most significant benefit occurs, however, when using additive levels below about 5 wt% additive. Further increasing the additive level in an FCC unit yields only marginal increases in propylene yields. For example, a refiner utilizing about 5 wt% ZSM-5 additive could have to nearly double thatadditive level for a gain of only 1-2 wt% in propylene yield. Doubling the FCC unit additive level could prove too costly for a refiner. The feed conversion in the riser could also begin to be reduced when the FCC catalyst is over diluted. In addition, ZSM-5 produces propylene by converting naphtha range olefins, so increasing additive will begin to impact naphtha yields. RxCat successfully offsets the increased conversion of naphtha olefins since this technology increases conversion and decreases dry gas yields thereby boosting overall gasoline yields. So refiners currently using moderate levels of ZSM-5 additive might find that UOP’s RxCat technology with ZSM-5 is a more feasible alternative to increasing propylene yields rather than further increasing their ZSM-5 additive level.

36 PetroFCC Process

The PetroFCC process was developed by UOP to provide a FCC process that maximizes yields of propylene and other valuable petrochemical feedstocks from heavier oil feedstocks. This process further leverages RxCat technology, along withother key UOP technologies and operating conditions. The remainder of this paper describes in detail the PetroFCC process and how thePetroFCC process can be a key enabler for increased profits for refiners.

PetroFCC Process Yield Structure Table 4 illustrates the expected yield structure for a typical VGO feedstock for both a traditional FCC unit and a PetroFCC unit.

Component, wt-% Traditional FCC PetroFCC

H2S, H], C[, & C2 2.0 3.0 Ethylene TO 6.0 Propane 1.8 2.0

Propylene 4.7 22.0

Butanes 4.5 5.0

Butylenes 6.5 14.0 Naphtha 53.5 28.0

Distillate 14.0 9.5

Fuel Oil 7.0 5.0

Coke 5.0 5.5

Table 4: Yield Patterns

The total light gas (C2-) make for the PetroFCC process is three times that of a traditional FCC. However, two-thirds, or 6 wt-%, is high value ethylene. Not only is the PetroFCC process ’ total C3 yield increased nearly four times, but the olefmicity of the C3 is also increased, resulting in a nearly five-fold increase in the propylene yield. The same type of effect is seen on the C4 yield and butylene selectivity. Significant reductions in the yields of the naphtha, distillate, and fuel oil products are also achieved compared to a traditional FCC. These yield improvements are realized at a coke yield only 0.5 wt-% greater than the traditional FCC operation. Compared to a conventional FCC, the naphtha produced from a PetroFCC is highly aromatic. Large quantities of C6 to C9 aromatics are present. Further processing of these aromatics can result in a stream primarily composed of highly valued p-xylene and benzene, thereby minimizing the production of the naphtha material. If additional naphtha blendstock material is desired, the refiner could elect to convert the C4s produced into alkylate or MTBE product. This option, if selected, would contribute a high-octane naphtha-blending component, which is completely free of sulphur, olefin, and aromatic content.

European PetroFCC Case Study A European refiner that desires to become a major player in the petrochemical market approached UOP for assistance. As an organization that currently operated several refineries,

37 they desired to make use of their available resources where possible and profitable. Specifically, they looked to convert a feedstock that was essentially a heavy diesel/atmospheric gas oil blend. In addition to this feed component, they also wanted to convert a mid-boiling range naphtha from one of their operating FCC units, an unconverted oil from one of their hydrocrackers, and some cycle oil products from various sources that their refineries were long on. To meet this refiner’s objectives, UOP recommended the PetroFCC process. Figure 1 shows the proposed PetroFCC design for this refiner. The PetroFCC process employs a customized catalyst system, slightly higher reactor riser outlet temperature, lower partial pressure, plus RxCat technology. The elevated riser temperatures and high cat/oil ratios associated withRxCat makes the use of UOP’s VSS riser termination, Optimix feed nozzles, High Efficiency Regenerator, and high flux stripper technologies critical to the design as well.

Figure 1: The PetroFCC

The expected yields for UOP’s PetroFCC process with this refiner’s particular blended feedstock is shown in Table 5. This combination of feed and catalyst resulted in a

38 very low delta coke system that did not provide sufficient coke to satisfy the unit heat requirements. Additional innovative process features were included in the design of this unit to satisfy the incremental heat requirements.

Component Yield, wt.-% HAH], C,,&C, 4.3 Ethylene 8.6 Propane 2.6 Propylene 21.5 5.5 Butylenes 11.7

Naphtha 2P.3 Distillate 10.7 Fuel Oil 3.0 Coke 2.8

Table 5: European Refiner’s Expected PetroFCC Process Yields This refiner was not interested in increasing traditional fuels output, so the PetroFCC process was ideal in that it limited the production of naphtha, distillate, and fuel oil. As mentioned earlier, the refiner was looking to use the PetroFCC process to convert unsuitable fuel components from elsewhere in the refinery to the more highlyvalued petrochemical feed stock components.

Aromatics Processing Because the refiner did not desire to produce any additional gasoline, and recognizing that the naphtha stream from this operation was essentially aromatic material, UOP and the refiner investigated various options to process this stream. It would not be possible to blend this material into the refiner’s existing gasoline pool due to its aromaticity. The value of this stream was potentially much greater when recovered as pure aromatic components as opposed to being valued as high octane gasoline blend stock. The ability to process and market this material properly would have a pronounced effect on the economic viability of this project. Extensive studies were undertaken to determine which products were the most attractive in the current and forecasted future markets. It was ultimately decided that p- xylene and benzene were the most attractive products to recover. An aromatics complex was designed that would maximize the production of not only the light olefins, but also p-xylene and benzene. By using UOP’s Sulfolane™, Tatoray™, Isomar™, and Parex™ processes, the customer would be able to recover these desired products directly, as well as convert the other xylenes, toluene, and aromatics present into p- xylene and benzene. Approximately 40% of the naphtha was recovered as p-xylene and nearly 15% as benzene. The non-aromatic portion of the naphtha was recovered in the Sulfolane unit and recycled back to the PetroFCC riser to be converted into light olefins. As an additional effort to convert this raffinate stream further to light olefins, UOP ran this naphtha stream through its circulating pilot plant to simulate the operation of a second reactor riser. This was done to investigate whether additional severity could facilitate

39 an improved overall conversion. The result of these investigations led to the conclusion that a two-riser PetroFCC process design with RxCat technology would not be more successful with this particular feedstock than the single-riser PetroFCC operation with RxCat technology.

PetroFCC Process Results The PetroFCC process unit proposed for this refiner met their objectives. Specifically, the21.5 wt-% propylene yield resulted in significantly more than 250 KMT A of propylene. Additionally, the blended feedstock would result in an ethylene yield in excess of 8.5 wt-%. Based on the amount of feedstock they expected to have available, this would easily meet their objectives for 125 KMTA of ethylene. Because a naphtha stream from elsewhere in the refinery was being converted in the PetroFCC unit and the resulting naphtha stream was being further processed in an aromatics complex, there was no net contribution to the refinery’s naphtha pool. If desired, the refiner could elect to convert the C4s that were produced into an alkylate or MTBE product. Based on the market value for the petrochemical feedstocks, the return on investment for the entire project exceeded 15% and was considered attractive with afour-year payout.

PetroFCC Revamp Opportunity

A more common scenario to improve light olefins yield is the revamp of an existing FCC unit to move the yield pattern toward that of a PetroFCC process unit. For a refiner evaluating a FCC revamp, the objectives are relatively simple: improve the profitability of the existing complex by increasing the throughput to the FCC unit and/or improving the yield selectivities to the highervalued products. This refiner has an existing propylene recovery unit as well as ready access to an existing aromatics complex and wants to maximize the utilization of these resources. UOP initiated an extensive revamp feasibility study for a customer to determine how- best to increase the yields of propylene and aromatics while remaining within prescribed capital constraints. The revamp study focused on incorporating as much of the PetroFCC technology as possible to deliver a yield slate that would be more representative of a PetroFCC unit. When this refiner’s FCC unit was originally designed, the refiner had every intention of producing additional propylene and recovering it as polymer grade quality with a propylene recovery unit. The unit was designed to improve propylene selectivity through shape selective zeolitic catalyst additives and a high riser outlet temperature. The revamp study re-optimized process variables to improve upon the original design and current operating propylene yield of 7.5 wt-%. Application of RxCat technology can provide significantly improved product selectivities and operational flexibility at a low cost. The physical modifications to the unit are minimal. The new equipment required for a revamp includes a MxR chamber to replace the wye section and a new standpipe with expansion joint and slide valve (highlighted section in Figure 1).

RxCat Technology Shifts the Yield Curve The revamp feasibility study uncovered that the breaking point with regard to required capital was the replacement or modification of the reactor shell. The optimum revamp case entailed replacement of the cyclones in this vessel to the largest size that could fit. The amount of gas volume that had to be handled by the reactor cyclones then became the

40 limiting factor in the revamp. In addition to replacing the reactor cyclones, the refiner intended to undertake substantial modifications to the fractionation and recovery sections of the complex to accommodate thelarge increase in light olefins production. With the mechanical and equipment limitations understood, UOP focused on adding technology, modifying process variables and reaction severity, and adjusting the feed rate to determine which set of parameters resulted in the most attractive option to the refiner. Numerous revamp options were investigated to define the potential operating envelope for the revamped unit. These options are presented in Figure 2. The first option looked at operating the unit at a feedrate of 120% of the original design. This coincided with the current operation. For this option, UOP found that the addition of RxCat technology and operating conditions would increase the propylene yield from the current value of 7.5 wt-% to nearly 10 wt-%. This option was ultimately constrained by limitations in the revamped recovery section. The second option was to operate therevamped unit at a feed rate equal to the original design value. Backing out feed necessitated the re-optimization of the operating conditions to take full advantage of RxCat technology and process modifications. In this option, the propylene yield could be increased to 10.8 wt-% before being limited by equipment constraints. UOP also looked at the option of decreasing the feed rate to values less than the original design. At a feed rate of 75% of the original design, operation could be optimized to take even more advantage of the RxCat technology and process modifications to deliver a propylene yield of nearly 15 wt-%. As a point of reference, UOP estimated the ultimate propylene yield selectivity for this refiner’s feedstock. Had this unit originally been designed as a PetroFCC unit, this feedstock would be capable of producing a product slate containing more than 20 wt-% propylene. Thus, even though this unit could be revamped to essentially double the propylene yield, it was not possible to take this existing FCC unit “all the way” to the performance of a PetroFCC unit due to the constraints imposed by the reactor cyclones and thefractionation and recovery sections.

23

21 New PetroFCC Wt.-% 19 Propylene * 7 15 13 11

9 Technology Shift

7 Design Current 5 75 100 125 % of Design Figure 2: RxCat TechnologyShifts the Curve

41 The predicted yield slates for many options were investigated to determine which one provided the most favorable economics for both short-term plans and long-term expansion goals from this complex. Operating at a feed rate equal to the original design but with the propylene yield improved to 10.8 wt-% effectively strikes a balance between increased propylene yield and increased naphtha aromatics. The economics for this revamp operation are very favorable. The refiner stands to recognize an incremental product margin improvement of $34 million per year and an improvement in net cash flow of approximately $16 million per year. The revamp requires an investment of $32 million; thus a 24-month simple payback is anticipated.

Conclusions

The two customer projects presented here, one a grassroots PetroFCC process complex combined with aromatics recovery, the other a revamp of a more traditional FCC unit, clearly illustrate the benefits that UOP’s PetroFCC and RxCat technology can deliver. No longer does the FCC have to be limited to making only transportation fuels. The PetroFCC, MTO, and Oleflex process all have a potential role to play to address the propylene gap in West Europe.

42 REFERENCES

1. CMAI, 2000

2. A. Pettman, “European Competitiveness ” A Key Issue Facing European Olefms/Polyolefin Producers ”, CMAI, 3rd EPTC, June 20,2001

3. Chem Systems, Ethylene/Propylene PERP Report, June 2001

4. Davison Catalgram, W.R. Grace & Co, Number 86, 1998

5. L.L. Upson and E.C. Nelson, RxCat Technology for More FCC Naphtha, Akzo Nobel Catalysts Symposium, Noordwijk, the Netherlands, June 22-25, 1998

6. J.M. Houdek, C.L. Hemler, L.L. Upson, and R. Pittman, PetroFCC - New Paradigm, ERIC, 2000

43 44 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

H. Bischof*, W. Dohler**, F. Fuder***, J. Laege* *Veba Oel Verarbeitungs GmbH, Gelsenkirchen, Germany, **Veba Gel Refining and Petrochemicals, Gelsenkirchen, Germany, *'*ARAL-Forschung Downstream, Bochum, Germany

Conversion of Problematic Petrochemical-Refinery Fractions to High Value Steam Cracker Feed

I. European Gasoline Programm conditions Aromatics Reduction

Future fuel specification mandated by the European Auto Oil Programm (AOP II) causes a big challenge for European refineries. The driver for this new specification are the ambitious German environmental policy and an agreement between the automobile- and the petroleum-industry an the one and the government on the other side. Therefore stringent low sulfur- and aromatics fuels requirements will be introduced in Germany in the year 2003.

Today's sulfur values together with the aromatics concentration were decided in 1999 within the framework of the European AOP. For the beginning of 2000, both numbers were fixed for sulfur at 150 ppm and 42 Vol.% for aromatics-concentration in gasoline and to 50 ppm sulfur for the 2000 / 2001- periode. Furthermore the reduction down to 10 ppm sulfur resp. a minimum of 35 Vol % total aromatics in motor gasoline must be reached in the year 2005.

Being aware of this environment the refining industry will have to invest in order to comply with the new stringent fuel specifications. This also is expected to exert an essential influence on the decisions which have to be taken at that moment. Also remembering the fact, that the forecast of German's oil consumption shows a degrease in gasoline. Starting in the year 2000 with 29,9 Mill t gasoline-consumption and having a consumption from only 25,3 Mill, t in the year 2010. This will restrict the chance to minimize the aromatics-concentration in gasoline-pool by dilution and cause surplus which have to be handled in a benefitial way.

Refinery performance and flexibility contribute directly to refinery economics. Let me start by looking to the main items a refinery can be influenced in order to directly increase competitiveness and profits.

2. The Refinery at Gelsenkirchen-Scholven signals to get an Surplus in Aromatics

Since the beginning of the year 2001, Veba Oil Refining & Petrochemical GmbH (VORP) produces the gasoline-quality "Super Plus " solely with a maximum of 10 ppm of sulfur and 42 Vol% of aromatics. It intents to continue investments in the future retrofit of refinery-plants for the desulfurisation of gasoline and diesel to produce all of these products with 10 ppm sulfur; but what's about the aromatics? How to reduce their concentration down to 35 Vol % till the end of 2003?

VORP operates the biggest integrated refinery- and processing system for oil and petrochemical products in Germany. The mothercompany Veba Oel AG is responsible for financial strategical assetmanagement of her daughters Aral AG, Veba Oil & Gas GmbH and VORP.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 45 The plant configurations of the refining system shows high complexity and a high share of conversion capacity, which results in a very low make of heating fuel oil. This translates into the ability to generate considerable cash margins. An other significant part of the contribution to refinery benefits comes from the petrochemical business.

The company VORP operates three steam crackers to produce ethylene and propylene. In that case, VORP is the biggest independent ethylene and propylene producer in West Europa. The combined capacity of the three steam crackers are 1.3 million tpy of ethylene and 0.9 million tpy of propylene. Additional petrochemical production-plants comprises aromatics like ortho- and para- xylene, , benzene, , and methanol.

High-boiling aromatics-containing fractions being side streams of petrochemical- or refinery plants. They are the source of aromatics which have to be reduced to reach further gasoline specification. For example, the reformer-process, feeded with the C6+ cut, producing aromatics used as high-octane numbered components in the gasoline-pool or they are manufactured to produce ortho- and para-xylenes. The side-stream coming from this ISOMAR - xylene-process is a C9-cut, which is blended to the gasoline-pool.

An other high aromatics-concentrated byproduct is heavy pyrolysis gasoline, coming from the steamcracking process. This process is feeded mainly with naphtha. This steamcracking process yields heavy pygas with a weight-ratio of 0.3 to 0.4 per ton of ethylene. The high aromatic pygas implements a high octane number to use it actuell as blend component for motor-gasoline (mogas).

In the future, all these problematic products have to be used in another purpose, if the high aromatic concentration doesn't alowe to use these components as feed for the gasoline pool, but a beneficial upgrade to a high value product is possible.

3. Managing the Aromatics Restriction without loosing Benefits.

Refining is a mature industry and petroleum products are commodities. The technology to produce these products are widely known. There are, however, many areas where considerable developments has been achieved over the past few years.

VORP has found, that there are some areas where it is both possible and worth while to undertake developments of its own. But it is known, that areas where big investment is necessary to develop technology are better done together with joint venture companies who make this development to their own core business.

The major changes to be faces in the future are driven by the market. To meet the market, the driving forces to develop technologies for converting problematic high aromatics containing fractions to high value products are the following one:

• Disposal of problematic sidestreams in petrochemical refineries under economical viewpoints. This is to reduce aromatics and presumably

46 polyaromatics content in motor gasoline and to satisfy the new environmental aspects and regulations influenced by the Auto Oil Programm (AOP II), • The new abundant steamcracker-feed produced by upgrading low value side steams of olefin plants will be "monetised" by recycling. • Activating feed sources in remote areas for steamcracker plants • Minimizing the import-dependency of feed-supply for steamcracker-plants by internal recycling of low value side products.

4. Pushing an idea to economical progress - A multi-company coorporation

The future of the petrochemical industry will be significantly influenced by increasing the overall economics by reprocessing low value-side streams containing heavy aromatics.

Economical and environmental benefits from the upgrading and recycling of problematic by-products are the driving forces to support the continuous innovation process of this reported project.

During the following part of my presentation, I will explain, that the actual use of these products in highly integrated refinery - petrochemical complexes are high-octane number compounds for the gasoline pool, but have to be limited in the future. Therefore these products will lose economical value due to limited blending rates in the gasoline pool and because of the limitation of aromatics and the minimisation of the flash boiling point of mogas.

My presentation will continue and focus on the following five items:

1. Process of teamwork to invent a progressive conversion - process to produce a suitable steamcracker - feed by reprocessing aromatic side-products. • Looking for alternatives to handle the high-aromatic fractions. • Creating a catalytic process to convert heavy pygas to C2-C4 - components as steamcracker-feed.

2. Test of several existing feeds in comparison with defined model-compounds. • Using lab-proven experience from developments of catalytic ring ­ opening reactions, to process heavy pygas under total hydrogenation and ring-opening - conditions.

3. Calculations to prove the benefit of the conversion-process under different side conditions of the refinery and looking for strategical advantages. This will be demonstrated especially for the location of the Ruhr Oel refinery at Gelsenkirchen-Scholven. • The hydro-ringopening process is able, to convert the total amount of heavy pyrolysis gasoline to steamcracker-feed. Byproducts are small amounts of heating gas.

47 4. Selection to integrate the new process into the petrochemical part of the Ruhr Oel refinery and to improve overall economics and gross margins in comparison to a typical olefin-feed. • The hydro-ringopening process fullfills the strategical aspects to manage the expected decrease of motor gasoline and therefore will make the aromatic problem under the conditions of AOP less critical. • The high hydrogen consumption in combination with the hydrogen- price, is the main factor to make the process beneficial or not.

5. Integration of engineering-companies, universities, catalyst-manufactures and the refinery- and petrochemical competition to pursue common interests and form a stake holder-synergy-team to push the idea to an economic progress.

48 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Raichle, Y. Traa, J. Weitkamp Institute of Chemical Technology, University of Stuttgart, Germany

Preparation of a Synthetic Steamcracker Feedstock from Pyrolysis Gasoline on Zeolite Catalysts

ABSTRACT

With the advent of the European Auto Oil Programme, the aromatics content of gasoline has to be reduced significantly and, thus, an oversupply of aromatics is expected. Therefore, the hydroconversion of aromatics-rich pyrolysis gasoline from naphtha steamcrackers into a high-quality synthetic steamcracker feedstock (mainly composed of ethane, propane and n-butane) at around 400 °C and 6 MPa may become a useful process. On the one hand, the aromatics can directly be converted with hydrogen on shape-selective zeolites in their bifunctional form {e.g., Pd/H- ZSM-5). On the other hand, this can be performed via a two-stage route comprising ring hydrogenation on conventional metal catalysts and consecutive ring opening of the resulting cycloalkanes on shape-selective zeolites in their acidic form {e.g., H- ZSM-5).

INTRODUCTION

With the advent of the second stage of the Auto Oil Programme of the European Union in 2005, Western European refiners will be forced to decrease the content of aromatics in gasoline to below 35vol.-%1. Considering the average aromatics content in 1999 of 45 vol.-%1, this means that within 5 years, the aromatics content of gasoline has to be reduced by 10 %. Thus, the production of about 14 Mt/a of aromatics (10 vol.-% of the Western European gasoline demand of 120 Mt/a1; aromatics have a considerably higher density than the other gasoline components) has to be abandoned, or a new outlet for these aromatics has to be found. Considering that the chemical demand for BTX aromatics in Western Europe amounts to about 10 Mt/a (1998/ only, it is obvious that such a quantity of aromatics cannot be absorbed by the market and would cause a collapse of prices. Facing the high forecasted worldwide annual growth rates of the demand for ethene 2 and propene 3 in the next ten years of about 4.5 and 4.7%, respectively, the situation will even worsen. Therefore, in this paper we suggest a novel catalytic process for hydroconversion of pyrolysis gasoline into a high-quality steamcracker feedstock.

TWO PROCESS OPTIONS

Converting aromatics from pyrolysis gasoline into a synthetic steamcracker feed enables a direct recycling of steamcracker by-products (see Figure 1). Thus, about

49 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 one quarter of the naphtha 4, which has to be imported to Western Europe on a large scale and the worldwide availability of which is expected to decrease 1, could be saved. For the conversion in steamcrackers, n- with two or more carbon atoms (C2+-n-alkanes) are the most desired products. /so-Alkanes produce less ethene and propene, but more methane, and they are, therefore, less suitable as a steamcracker feed. Cycloalkanes give mainly aromatics; methane and aromatics are virtually inert in the steamcracker 5. Consequently, when preparing a synthetic steamcracker feed, the formation of cycloalkanes, aromatics and methane should be avoided.

- Ethane Propene: Ethene: - LPG (C,H„, C4Hx) 17 wt.-% 33 wt.-% - Naphtha (C5Hx ,C6Hx) (worldwide: 52 wt.-%) Methane: C4 cut: (Europe: 86 wt.-%) 13 wt.-% - VGO (vacuum gas oil) 11 wt.-% Oil: 3 wt.-%

Hydrogen: 3 wt.-% Selective Pyrolysis gasoline: hydrogenation Steam 21 wt.-% 62 wt.-% Aromatics Synthetic steam ­ 25 wt.-% , cracker feed (ethane, propane 10wt.-% Alkanes and n-butane) 3 wt.-% Cycloalkanes

Direct route: Ring opening on bifunctional catalysts Two-stage route: a. Ring hydrogenation b. Ring opening on after conventional acidic zeolites technology Gasoline ------Mainly cycloalkanes *=------

Figure 1: Typical product distribution of a naphtha steamcracker 5 and options for using the by-product pyrolysis gasoline.

Recently, two process options for converting such aromatics into a high-quality synthetic steamcracker feed have been disclosed 6. As sketched in Figure 1, pyrolysis gasoline can directly be converted with hydrogen on shape-selective zeolites in their bifunctional form7, e.g., on Pd/H-ZSM-5 (see below). An alternative is the two-stage route comprising ring hydrogenation of the aromatics to cycloalkanes on conventional hydrogenation catalysts and consecutive ring opening on shape-selective zeolites in their monofunctional acidic form8 , e.g., on H-ZSM-5 (see below). Obviously, the main advantage of the direct route is that a single reactor suffices to achieve the desired catalytic conversion. Additionally, higher yields of C2+-n-alkanes (e.g., of ethane) can be achieved, if aromatics dealkylation takes place (see below). Conversely, advantageous features of the two-stage route are (i) an easier removal of the heat

50 generated in the two exothermic steps and (ii) the possibility to optimize the reaction conditions of ring hydrogenation and ring opening, in particular the nature of the catalyst and the process temperature, independently from each other.

C2+-n-ALKANES FROM AROMATICS VIA THE DIRECT ROUTE

Direct conversion of aromatics can be performed in a fixed-bed reactor on, e.g., a 0.2Rd/H-ZSM-5 catalyst (zeolite ZSM-5 with an n s/n Ai ratio of 20 in its Bronsted acid form, loaded with 0.2 wt.-% palladium referenced to the mass of the dry catalyst). Depending on the hydrocarbon feed, yields of the desired C2+-n-alkanes (mainly ethane, propane and n-butane) between 73 (from toluene) and 90 % (from benzene and ) are achieved at total conversion at 400 °C. The major by-products are Zso-alkanes (especially /so-butane) with yields between 6 (from benzene) and 24 % (from toluene), which are still acceptable as steamcracker feed, while the yield of the undesired product methane can be kept very low (Yme ihane = 3 to 4 %). No cycloalkanes or aromatics are observed at all (see Figure 2). As a whole, this product spectrum represents an excellent synthetic steamcracker feed. Constant product yields are observed during more than 10 h time-on-stream 7,9 .

Feed: Benzene Feed: Toluene Feed: Ethylbenzene

Propane, 62.8

Methane, 2.5 % /so-Alkanes, iso- Alkanes, 7.3 % 23.9 % Propane, C4+-n-Alkanes, C.-n-Alkanes, 16.3% 10.6%

Feed: o-Xylene Feed: p-Xylene Feed: 1,2,4-Trimethylbenzene

Propane, 49.1 %

0 /so-Alkanes, Jf iso-Alkanes, 13.7% 16.1 % CL-n-Alkanes, 17.6 % CL-n-Alkanes, 17.9 %

Figure 2: Product yields during hydroconversion of various aromatics at 400 °C and 6 MPa on zeolite 0.2Pd/H-ZSM-57 .

Key parameters to ensure such desirable results are (i) the use of a bifunctional zeolite, such as 0.2Pd/H-ZSM-5 (a shape-selective, medium-pore zeolite, see below), (ii) the application of elevated hydrogen partial pressures around 6 MPa (hydrocarbon partial pressure ca. 65 kPa, WHSV = 0.7 h" 1) and (iii) relatively high

51 reaction temperatures in the vicinity of 400 °C. If the reaction is performed at lower temperatures, /so-alkanes (at 300 °C) and cycloalkanes (below 300 °C) are the dominant products.

A pronounced dependence of the yields of individual products on the nature of the feed hydrocarbon is observed during the conversion of a broad variety of aromatics (which all occur in pyrolysis gasoline) at 400 °C, i.e., under conditions which are favorable for the manufacture of a synthetic steamcracker feed (cf. Figure 2). For instance, much more propane (62.8 %) is formed from benzene than from any other feed hydrocarbon. By contrast, from toluene, the yield of /so-alkanes (mostly iso­ butane) is unusually high (23.9 %). These differences can mainly be accounted for by different possibilities of classical fS-scission of the cycloalkanes with different numbers of carbon atoms9,10 . Still another pattern results during the conversion of ethylbenzene: Here, an unusually high yield of ethane (30.6 %) is achieved. Because this yield cannot be observed during the conversion of ethylcyclohexane (see below), it is attributed to a fast aromatics reaction, i.e., an acid-catalyzed deethylation followed by hydrogenation of ethene 7,9 , which proceeds in parallel to the ring hydrogenation and skeletal isomerization reactions. It is also evident from Figure 2 that, among the aromatic hydrocarbons studied, toluene gives by far the lowest yields of C2+-n-alkanes. Hence, with pre-treated, technical pyrolysis gasoline, yields of 80 % or higher can be expected.

C2+-/>ALKANES BY RING OPENING OF CYCLOALKANES

In the hydroconversion of (M-CHx, a model component for hydrogenated pyrolysis gasoline) at 400 °C on acidic zeolite H-ZSM-5 (n s/n Ai = 20 unless otherwise specified), yields of C2+-n-alkanes comparable to those during the direct conversion of toluene on Pd/H-ZSM-5 are obtained (ca. 72 %8,11 ); no changes in the conversion or yields are observed within 10 h time-on-stream. Again, such good catalytic results can only be achieved, if suitable conditions are applied: (i) a high hydrogen partial pressure, e.g., 6 MPa, (ii) a relatively high temperature (e.g., 400 °C) and (iii) a shape-selective catalyst (e.g., ZSM-5, see below). Unless otherwise specified, the reactions are again performed at a hydrocarbon partial pressure of ca. 65 kPa and a WHSV of 0.7 h" 1. However, a variation of the WHSV in the range from 0.35 to 2.7 h" 1 has only a minor influence on the product composition: During the conversion of methylcyclohexane at 375 °C, the yield of C2+-n-alkanes decreases only from 71 to 65 %. At 400 °C, the effect is even smaller. If methylcyclohexane is converted at about 450 to 500 °C and 0.1 MPa in nitrogen, light n-alkanes and alkenes such as ethene, propene and butenes are obtained. Even on the shape-selective zeolite H-ZSM-5, considerable amounts of higher hydrocarbons (mainly aromatics) are formed, giving rise to fast deactivation 12,13. In this case, a continuous regeneration as it is employed in catalytic cracking (FCC, DCC) would be necessary. With decreasing temperature, again, a strongly decreasing yield of the desired C2+-n-alkanes is observed, while the yields of /so-alkanes and cycloalkanes pass through maxima at 350 and 300 °C, respectively (see Figure 3). However, with increasing temperature between 350 and 400 °C the yield of the very undesired methane increases strongly from 1.3 to 3.9 %, which restricts the applicable temperature window at the upper end. If ethylcyclohexane is fed instead of

52 methylcyclohexane, an enhanced activity and selectivity to C2+-n-alkanes is observed. At 400 °C on zeolite H-ZSM-5, the yield of C2+-n-alkanes increases from 72 to 77 %, which is a small increase compared to the difference in the yields of C2+- n-alkanes during conversion of toluene (73 %) and ethylbenzene (90 %). Obviously, the dealkylation of aromatics is as fast as the aromatics hydrogenation and skeletal isomerization. By contrast, during cycloalkane conversion, isomerization of the cyclohexane ring to is faster than the dealkylation reaction. Therefore, an increase of yields by dealkylation reactions is only possible during aromatics conversion, which is a big advantage of the direct route.

100 -

/soAikanes

Methane

Cycloalkanes

Aromatics

0 325 5 Temperature / °C

Figure 3: Hydroconversion of methylcyclohexane on H-ZSM-5 at 6 MPa and various temperatures 8 .

The most important property of the catalyst is its pore architecture: With increasing geometrical constraints, i.e., decreasing pore dimensions in the order H-Y, H-Beta, H-ZSM-11, H-ZSM-5, H-ZSM-22 and H-ZSM-35, three important effects occur during the reaction of methylcyclohexane: (i) On the large-pore zeolites H-Y and H-Beta, considerable deactivation and especially with time-on-stream strongly decreasing yields of the desired C2+-n-alkanes are observed over the whole temperature range. This is attributed to a spatially demanding formation of polynuclear aromatics11,12. In contrast, no significant decrease in the yields of C2+-n-alkanes is observed during 7 h time-on-stream on the medium-pore zeolites H-ZSM-11, H-ZSM-5, H-ZSM-22 and H- ZSM-3511. (ii) The conversion decreases drastically from, e.g., 100 % at 400 °C on H- ZSM-5 to 28.4 % on H-ZSM-22 and to as little as 13.3 % on H-ZSM-35. In-line with adsorption data, this is attributed to the hindered diffusion of methylcyclohexane in these zeolites 6,11, (iii) On the other hand, a strong increase in the selectivity to C2+-n- alkanes is achieved (the selectivities reported here are at a conversion of about 15 %) in the order H-Y and H-Beta (0 %), H-ZSM-11 (18 %), H-ZSM-5 (28 %), H-ZSM- 35 (41 %) and H-ZSM-22 (64 %). This has been observed earlier during the

53 conversion of methylcyclohexane in nitrogen at 0.1 MPa12,13 and can be accounted for by a higher contribution of non-classical Haag-Dessau cracking versus the spatially more demanding classical, bimolecular cracking 11. Due to the different dependencies of the conversion and the selectivities on the geometrical constraints, by far the highest yields of the desired C2+-n-alkanes with 71 to 72 % are observed on H-ZSM-11 and H-ZSM-5, two zeolites with intermediate pore size (see Figure 4).

Zeolite H-ZSM-11 Zeolite H-ZSM-22

C^-n-Alkanes, Propane, . Cycloalkanes, 1.4 % Propane, Aromatics, / /so-Alkanes, 5.5 % 13.9 % 24.7 % 46.5 Ethane, 1.9 % ^ C4+-n-Alkanes, 5.0 % Ethane, 7.8 % Propane, 10.1 % 3.8 % Methane, Ethane, 3.3 % Methane, 2.4 % Methane, 1.2 % 0.6 % Aromatics, iso Alkanes, /so-Alkanes, 15.9% 0 26.6 % C^-n-Alkanes, 16.7 % 39.3 % Cycloalkanes, 0.8 %

Zeolite H-Beta Zeolite H-ZSM-5 Zeolite H-ZSM-35

C^-n-Alkanes, Propane, 33.1 % Propane, ___ Ethane, Cycloalkanes, 3.6 % 20.0 % — 10.4% iso Alkanes, 1.8 % C4+-n-Alkanes, 0.8 Methane, Propane, 3.4 % 3.9% Ethane, 1.3 % Aromatics, Methane, 0.8 % 1.3% /so-Alkanes, K-n-Alkanes, /so-Alkanes, 30.2 % 22.3 % 13.4 %

Figure 4: Product yields during hydroconversion of methylcyclohexane at 400 °C and 6 MPa on various acidic zeolites after 30 min time-on-stream (the white slices represent unconverted methylcyclohexane) 11.

Upon decreasing the ns/nAi ratio of zeolite H-ZSM-5, i.e., increasing the concentration of Bronsted acid sites, a strong increase of the conversion of methylcyclohexane is observed. This is especially pronounced up to a medium n s/nAi ratio of about 35; a further decrease has only minor effects on the catalytic activities 14. However, the selectivity to C2+-n-alkanes is hardly affected: At 400 °C, only a slight decrease takes place with decreasing ns/n Ai ratio. Because opposite dependencies are observed for the conversion of methylcyclohexane and the selectivity to C2+-n-alkanes, the maximal yield of the desired C2+-n-alkanes of 78.0 % (besides 18.1 % /so-alkanes, 3.4 % methane and 0.5 % cycloalkanes) is achieved on zeolite H-ZSM-5 with a medium ng/n A, ratio of 3514.

During the conversion of methylcyclohexane, significantly higher yields of C2+-n- alkanes (more than 78 %) can be achieved on zeolite H-ZSM-5 containing very small amounts of palladium (e.g., 100 wt.-ppm) than on the metal-free catalyst (72 %). In contrast, on typically bifunctional catalysts such as 0.2Pd/H-ZSM-5 only 67 % of C2+-

54 n-alkanes are obtained (see Table 1)6,15. The effect of these very small amounts of palladium is attributed to a higher contribution of Haag-Dessau cracking on these catalysts while on acidic and especially on typically bifunctional catalysts, classical cracking dominates in its bimolecular or bifunctional form15. Similarly enhanced yields of C2+-/7-alkanes (e.g., 75 %, cf. Table 1) are also observed, if gallium is incorporated into zeolite H-ZSM-5 in a concentration typical for bifunctional catalysts (0.14 wt.-%). Thus, relatively high contents of metals with a low hydrogenation/dehydrogenation activity such as gallium have an effect similar to that of ultra-low contents of metals with a high hydrogenation/dehydrogenation activity such as palladium 16. However, a serious drawback of the gallium-containing catalysts cannot be overlooked, viz. the relatively large yield of methane (7.0 %, see Table 1)16.

Table 1: Hydroconversion of methylcyclohexane at 400 °C and 6 MPa on various H- ZSM-5 catalysts with different weight contents of palladium 15 or gallium 16.

Conversion or 0 wt.-ppm Pd 100 wt.-ppm 0.2 wt.-% Pd 0.14 wt.-% Ga yield / % Pd Xm-CHx 100 99.8 100 100 ^Methane 3.9 3.5 3.5 7.0 Y Ethane 10.4 12.3 6.1 14.3 Y Propane 48.6 51.5 47.8 51.7 Y/>Butane 12.4 13.7 12.6 8.8 Y n-Pentane 1.0 0.9 0.2 0.3 Yc2+-n-Alkanes 72.4 78.4 66.7 75.1 Y/soAikanes 22.4 17.7 29.8 17.3 Ycycloalkanes 0.0 0.1 0.0 0.0 YAromatics 1.3 0.3 0.0 0.6

CONCLUSIONS AND OUTLOOK

A high-quality synthetic steamcracker feedstock, mainly consisting of ethane, propane and n-butane, can be produced from aromatics via two alternative routes: (i) In the direct route, pyrolysis gasoline is converted on a shape-selective zeolite in its bifunctional form. An example for a good catalyst of this type is Pd/H-ZSM-5. (ii) In the two-stage route, the aromatics of pyrolysis gasoline are first hydrogenated by commercially available technology into the corresponding cycloalkanes. These can be hydroconverted on monofunctional, acidic zeolites into C2+-n-alkanes with high yields. Studies with model compounds indicate that, under appropriate reaction conditions (ca. 400 °C and 6 MPa on a shape-selective zeolite, such as ZSM-5), the yield of C2+-n-alkanes from either BTX aromatics or their cycloalkane derivatives are as high as 70 % to over 90 %. For pre-treated, technical pyrolysis gasoline, a yield around 80 % is hence realistic. The yield of the very undesired methane can be kept below 4 %. Expectedly, the n s/n Ai ratio of zeolite H-ZSM-5 affects its activity, but it has very little influence on the selectivity. Surprisingly selective catalysts result, if zeolite H-ZSM-5 is doped with ultra-low amounts of palladium in the range of ca. 10 to 100 wt.-ppm. The incorporation of gallium into H-ZSM-5 has a beneficial effect on the yield of C2+-n-alkanes. At the same time, however, gallium-doped H-ZSM-5 tends

55 to give too high yields of methane. Mechanistically, all results point to a significant contribution of non-classical Haag-Dessau cracking. The economic viability of the new process will depend, in the first place, on the price for pyrolysis gasoline once the Auto Oil Programme becomes fully effective, i.e., beyond 2004.

ACKNOWLEDGEMENTS

Financial support from VESA OEL AG, Deutsche Forschungsgemeinschaft, Fonds der Chemischen Industrie and Max Buchner-Forschungsstiftung is gratefully acknowledged. Yvonne Traa thanks the Ministerium fur Wissenschaft, Forschung und Kunst Baden-Wurttemberg for financial support through the Margarete von Wrangell-Habilitationsprogramm fur Frauen.

REFERENCES

1. Retzny, W.J., Halsig, C.-P., in: Emig, G., Rupp, M., Weitkamp, J. (eds.), Tagungsbericht 9903: Proceedings of the DGMK-Conference, The Future Role of Aromatics in Refining and Petrochemistry, October 13 - 15, 1999, Erlangen, Germany, Hamburg, DGMK, p. 7 (1999). 2. Chang, T„ Oil Gas J., 98 (14), 56 (2000). 3. Weirauch, W., Hydrocarbon Process., Int., Ed., 79 (6), 9 (2000). 4. Raichle, A., Traa, Y., Weitkamp, J., Chem.-Ing.-Tech., 73, 947 (2001). 5. Dembny, C., Reference 1, p. 115. 6. Weitkamp, J., Bischof, H., Dohler, W., Laege, J., Fuder, F., Raichle, A., Traa, Y., a. ) DE Patent Application 19949211 A1, assigned to Veba Oel AG (2001); b. ) WO Patent Application 01/27223 A1, assigned to Veba Oel AG (2001). 7. Weitkamp, J., Raichle, A., Traa, Y., Rupp, M., Fuder, F., Chem. Common. , 1133 (2000). 8. Weitkamp, J., Raichle, A., Traa, Y., Rupp, M., Fuder, F., Chem. Common., 403 (2000). 9 . Raichle, A., Traa, Y., Weitkamp, J., Catal. Today, submitted (2001). 10 . Weitkamp, J., Jacobs, P.A., Martens, J.A., Appl. Catal., 8,123 (1983). 11. Raichle, A., Scharl, H., Traa, Y., Weitkamp, J., in: Galarneau, A., Di Renzo, F., Fajula, F., Vedrine, J. (eds.), Zeolites and Mesoporous Materials at the Dawn of the 21st Century, Proceedings of the 13th International Zeolite Conference, Montpellier, France, July 8 - 13, 2001, Studies in Surface Science and Catalysis, Vol. 135, Amsterdam, Elsevier, p. 302 and full paper No. 26-P-10 on accompanying CD (2001). 12. Scofield, C.F., Benazzi, E., Cauffriez, H., Marcilly, C., Braz. J. Chem. Eng., 15 (2), 218(1998). 13. Cerqueira, H.S., Mihindou-Koumba, P.C., Magnoux, P., Guisnet, M., Ind. Eng. Chem. Res., 40, 1032 (2001). 14. Raichle, A., Ramin, M., Singer, D., Hunger, M., Traa, Y., Weitkamp, J., Catal. Commun., 2, 69 (2001). 15. Raichle, A., Traa, Y., Fuder, F., Rupp, M., Weitkamp, J., Angew. Chem., 113, 1268 (2001); Angew. Chem., Int. Ed. Engl., 40, 1243 (2001). 16. Raichle, A., Moser, S., Traa, Y., Hunger, M., Weitkamp, J., Catal. Commun., 2, 23 (2001).

56 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

C. Ringelhan, V. Kurth, G. Burgfels, J. G. Neumayr, W. Seuffert, J. Klose Sud-Chemie AG, Bruckmuhl-Heufeld, Germany

Conversion of Naphthenes into a Valuable Steamcracker Feed Using H-ZSM-5 Based Catalysts

Abstract

The conversion of heavy pyrolysis gasoline into valuable steam cracker feedstock with a maximum yield of C2-C4 n-alkanes is achieved via a two step process. The first step involves a hydrogenation of aromatics to naphthenes followed by the subsequent ring opening and cracking in the second step. The ring opening and cracking reaction of naphthenes was studied in a bench scale tubular reactor over extruded H-ZSM-5 based zeolite catalysts developed by SUD-CHEMIE. In a series of screening tests using a commercial, hydrogenated heavy pyrolysis gasoline, the influence of the preparation parameters such as zeolite acidity, palladium content and loading procedure, as well as the type of binder were investigated. Furthermore, the influence of the process conditions space velocity and temperature was studied.

High yields of C2-C4 n-alkanes at low formation of undesired methane and aromatics were achieved over an alumina bound zeolite with medium Bremsted acidity loaded with palladium. The palladium loading procedure was observed to have no significant influence on the activity and selectivity. The diminution of the space velocity showed an increase in the C2-C4 n- yield and lower formation of aromatics, but a simultaneous increase in the methane make. Raising the temperature from 280°C to 370°C significantly increased the catalysts activity. At 400°C a strong methane make was observed.

1. Introduction

In the Auto Oil Program the European Community agreed to reduce the aromatics content of gasoline from 42 vol % at present down to 35 vol % in 2005 [1], This will result in a surplus of aromatics [1], One main source of aromatics is the so-called heavy pyrolysis gasoline, a by-product of the production of ethylene and propylene by steam cracking of naphtha. Due to the predicted worldwide increasing demand of ethylene [2,3] and propylene [2,4] the surplus of heavy pyrolysis gasoline will increase further. One way to convert heavy pyrolysis gasoline is the conversion into a valuable steam cracker feed with high amounts of C2-C4 n-alkanes. This is achieved via a two step process. In the first step the aromatics are hydrogenated to naphthenes over a Ni or precious metal based

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 57 hydrogenation catalyst, followed by the subsequent ring opening and cracking in the second step [5­ 7]. A suitable catalyst for the second step is a H-ZSM-5 based catalyst loaded with palladium [5-7], which is in contrast to such typical bifunctional catalysts characterized by a palladium content of only max. 0.1 wt% [6], The present paper deals with the investigation of the second step in order to maximize the yield of

Cr-C4 n-alkanes at low formation of methane and aromatics. Therefore in a screening test program, the influence of the preparation parameters such as zeolite acidity, palladium content, loading procedure and the type of binder was studied. Furthermore, the influence of the process conditions weight hourly space velocity (WHSV) and temperature was investigated.

2. Experimental

The ring opening and cracking reaction was investigated with a commercial, hydrogenated pyrolysis gasoline consisting of naphthenes (predominantly C7-C9 naphthenes) as well as small amounts of poly-naphthenes, paraffins, aromatics and olefins. Screening tests were carried out over a catalyst as fixed bed in a bench scale tubular reactor (FIGURE la) at 60 bar(g), a gas inlet temperature of 370°C, a WHSV of 2 kg/(kg h) and at a hydrogen to oil ratio of 700 Nl/kg. During the tests the reference inlet temperature 30 mm in front of the catalyst bed was kept constant (see FIGURE lb).

—PWRpfWVr- pre-heater vaporizer

temperature

empty space —

catalyst bed - bed length 160 cooler/ condenser

liqiud purge sampling catch pot

FIGURE I: Test unit (a) and reactor loading scheme (b)

In each test a sample of 20 g catalyst (extrusions, particle size: 1/16“ diameter and 2-3 mm length) was diluted with SiC in the weight ratio 1:6, in order to minimize the temperature gradient caused by the strong exothermal reaction within the catalyst bed. The activation of the catalyst was carried out with hydrogen at 400°C and atmospheric pressure.

58 The gas phase was analyzed online by a 4-channel Micro-GC equipped with thermal conductivity detectors. The C—C2 hydrocarbons were separated by a PoraPlot column, the C:j-C4 hydrocarbons by a AluminaPlot column and the C5-C6 hydrocarbons by a OV-1 column. The content of aromatics in the liquid phase was determined offline by HPLC. The solved gaseous components and selected naphthenes were analyzed by GC equipped with a RTX-1 column and a flame ionization detector. Due to the complex composition of the commercial feedstock for the evaluation of the activity a simple liquid conversion rate was used. The liquid conversion was calculated from the liquid amount before and after the reaction at atmospheric conditions. The presented key naphthenes were , methylcyclohexane, ethylcyclohexane and iso-propylcyclohexane.

3. Results and Discussion 3.1 Investigation of the preparation parameters

The effect of the preparation parameters palladium content, Bransted acidity, binder and the palladium loading procedure in the ring opening and cracking reaction will be discussed in this chapter. The first three parameters were studied with catalysts loaded with palladium via loading procedure A. In order to ensure the comparability between the catalyst samples the presented results were picked up after 72 hours on stream. The palladium influence studied over an alumina bound medium acidic H-ZSM-5 is shown in FIGURE II.

Product distribution C2-C4 n-alkane distribution

FIGURE II: Influence of the palladium content on the liquid conversion and the product distribution over an alumina bound medium acidic H-ZSM-5 at 370°C, 60 bar(g), WHSV 2 kg/(kg h) and a hydrogen to oil ratio of 700 Nl/kg

59 Investigations of the palladium free zeolite (monofunctional acidic catalyst) indicated lower liquid conversion and poor C2-C4 n-alkane yield, which should be caused by missing of the palladium. During the test no signs of olefin formation were observed, what is attributed to the activation of the hydrogen by the acid sites [5,7]. Significantly higher C2-C4 n-alkane yield at low formation of methane and aromatics were attained over a zeolite with small palladium content. A further increase of the palladium content, even smaller than that of a typical bifunctional catalyst [6], led to an improved C2-C4 n-alkane yield and simultaneously to a decrease in aromatics. However, the higher palladium resulted in an increased methane formation.

The C2-C4 n-alkane fraction presented in FIGURE II consisted of predominantly propane as well as small amounts of ethane and butane. With rising the palladium content the formation of ethane and propane increased, while no significant change in n-butane formation was observed.

The variation of the Bransted acidity performed over an alumina bound H-ZSM-5 with small palladium content is shown in FIGURE III. The liquid conversion indicated best activity with the medium acidic zeolite compared to the low and strong acidic zeolite. Furthermore, excellent C2-C4 n- alkane yields as well as low formation of methane and aromatics were found.

FIGURE III: Influence of the zeolite acidity on the liquid conversion and the product distribution over an alumina bound H-ZSM-5 with small Pd content at 370”C, 60 bar(g), WHSV 2 kg/(kg h) and a hydrogen to oil ratio of 700 Nl/kg.

The C2-C4 n-alkane fraction indicated different yields of propane as well as ethane but no significant change in butane yield depending on the zeolite acidity. With the medium acidic zeolite high yield of propane were observed. However, the strong acidic zeolite showed a high formation of ethane.

The influence of the binder silica respectively alumina investigated over a medium acidic H-ZSM-5 loaded with small palladium content is presented in FIGURE IV. From the liquid conversion it can be

60 seen that the activity of the alumina and silica bound zeolite were about 82 %. However, a slight difference was observed in the product distribution. The alumina bound zeolite showed a slightly better C2-C4 n-alkane yield at simultaneously lower formation of methane and aromatics compared to the silica bound zeolite.

/// // /// Product distribution C2-C4 n-alkane distribution FIGURE IV: Influence of the binder on the liquid conversion and the product distribution over a medium acidic H-ZSM-5 with small Pd content at 370°C, 60 bar{g), WHSV 2kg/(kg h) and a hydrogen to oil ratio of 700 NI/ kg

The palladium loading procedure studied over an alumina bound medium acidic H-ZSM-5 loaded with small palladium content via different loading procedures was observed to have no significant influence on the activity (both catalysts indicated liquid-conversion of 92 %) and the selectivity (FIGURE V).

□ loading procedure A H loading procedure B

Product distribution C2-C4 n-alkane distribution

FIGURE V: Influence of the Pd loading procedure on the liquid conversion and the product distribution over an alumina bound medium acidic H-ZSM-5 with high Pd content at 370°C, 60 bar(g), WHSV 2 kg/kg h) and a hydrogen to oil ratio of 700 Nl/kg

61 3.2 Investigation of process conditions

The variations of temperature and space velocity were carried out over an alumina bound medium acidic H-ZSM-5 loaded with high palladium content via loading procedure A. The temperature influence was investigated in a temperature range from 280°C to 400°C. The test results are presented in FIGURE VI.

Inlet temperature p'C] aromatics C2-C4 n-alkane methane X liquid X i-propylcyclohexane X ethylcydohexane X methylcyclohexane FIGURE VI: Influence of the temperature on the naphthene conversions, liquid conversion, C2-C4 n-alkanes, methane and aromatics over an alumina bound medium acidic H-ZSM-5 with high palladium content at 60 bar(g), WHSV 2 (kg/kg h) and a hydrogen to oil ratio of 700 Nl/ kg

In the range from 280 to 370°C the liquid conversion increased rapidly from 20 % to 80 %. Above 370°C only a slight increase was observed. With regard to C2-C4 n-alkane yield as well as low formation of methane and aromatics the best performance could be reached at 370°C. At higher temperatures, the C2-C4 n-alkanes could be increased further at simultaneously lowering of the aromatics content. However, the methane make was significantly increased. Additionally the naphthene conversion of the key naphthenes was studied in dependence of the temperature. The conversion profiles of methylcyclohexane and ethylcydohexane were very similar. Above 33CTC conversions of higher than 99 % were observed. At lower temperature the conversions decreased rapidly. At 280°C the conversion of methylcyclohexane and ethylcydohexane were less than 30 %. The iso-propylcydohexane conversion was always lower compared to the conversion of methylcyclohexane and ethylcydohexane, e g. 92 % compared to > 99 % at 400°C. Reducing the temperature the iso-propylcydohexane conversion decreased rapidly and at 280°C no conversion was observed anymore.

62 The results of the variation of the weight hourly space velocity are presented in FIGURE VII. The decrease of the WHSV from 2 kg/(kg h) to 1 kg/(kg h) gave no visible change in cracking activity and selectivity.

FIGURE VII: Influence of the weight hourly space velocity on the liquid conversion and the product composition over an alumina bound medium acidic H-ZSM-5 with high palladium content at 60 bar(g), 370'C, and a hydrogen to oil ratio of 700 Nl/kg

Only a further decrease to 0.5 kg/(kg h) resulted in a significant increase of the liquid conversion and C2-C4 n-alkane yields coupled with a decrease of the aromatics. However, this distinct reduction of the space velocity led to a stronger formation of methane. The reason of this effect is supposed to be an increase of secondary cracking reactions to form short chain alkanes caused by the prolonged residence time.

4. Conclusions

Using a H-ZSM-5 based catalyst commercial, hydrogenated heavy pyrolysis gasoline can be converted into a valuable steam cracker feed. Excellent performance is reached over an alumina bound medium acidic H-ZSM-5 with high palladium content, whereby the loading procedures under investigation are of minor influence. Furthermore the influence of the temperature and the space velocity was studied. The best results with regard to high liquid conversion and C2-C4 n-alkane yields at a simultaneous low formation of methane and aromatics were observed at a temperature of 370°C and a space velocity of 2 kg/(kg h).

63 Overall, the feasibility of the technology could be successfully demonstrated and the new catalyst is ready to be used in the ARINO process, licensed by LINDE. Customer related application testing in bench scale can now be offered to the industry.

5. References

[1] W. J. Retzny, C.-P. Halsig, in DGMK Tagungsbericht 9903, Proc. DGMK Conference: The Future Role of Aromatics in Refining and Petrochemistry, October 13-15, 1999, Erlangen, Germany, ed G. Emig, M. Rupp, J. Weitkamp, DGMK, Hamburg, p. 7-20 1999 [2] W. Bonse-Geuking, ErdOl, Erdgas, Kohle 116, p. 407 (2000) [3] T. Chang, Oil Gas J. 98, No. 14, p. 56 (2000) [4] W. Weirauch, Hydrocarbon Process., 79, No. 6, p. 9 (Int. Ed. 2000) [5] J. Weitkamp, A. Raichle, Y. Traa, M. Rupp and F. Fuder, Chem. Commun. P. 403 (2000) [6] Patent DE 199 49 211 A 1 VEBA OEL AG [7] A. Raichle, Y. Traa, F. Fuder, M. Rupp, J. Weitkamp, Angew. Chem. 113 No. 7, p. 1268 (2001)

64 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

H. Fritz, H. Bolt, U. Wenning Linde AG Engineering and Contracting Division, Hollriegelskreuth, Germany

ARINO™ (Aromatic RINg Opening) Technology for Upgrading of Low Value Aromatics to High Value Steamcracker Feedstock

1 Background

1.1 Heavy Pyrolysis Gasoline from Steamcracking

Heavy Pyrolysis Gasoline (heavy pygas) is an unavoidable byproduct of hydrocarbon steamcracking for ethylene production. The quantities produced depend on feedstock type and quality and on the severity of the cracking process. Typical heavy pygas yields of naphthacracking are in the range of 0.3 to 0.4 tons per ton of ethylene produced.

Heavy pygas is a mixture of a huge variety of different components in small amounts. Boiling in a range between 100 to 210°C the majority of the fraction consists of C7 to C9 components. Characteristic for heavy pygas is a high content of aromatics. The concentration is increased with cracking severity and vary in a range of 68 to 87 wt%. Toluene and Ethylbenzene/Styrene are the predominant components, but their small total amount and costly separation makes a recovery economically unattractive.

Due to its high aromatics content heavy pygas has a high octane number making it a good blending stock for motor gasoline. This utilisation requires only little processing and therefore almost all heavy pygas produced by steamcracking is blended to motor gasoline. Usually the C9+ fraction is removed before and used as fuel or is recycled to the furnaces and cracked to total extinction.

1.2 European Gasoline Pool

The actual European gasoline production is about 110 million tons per year (tpy) with an average concentration of aromatics of about 39 vol. %. Announced legislation of the European Commission (Auto Oil Program AOP II) will restrict from 2005 on the total aromatics content in motor gasoline to a maximum of 35 vol. %.

This limitation combined with decreasing gasoline consumption will cause a surplus of C7/C8 aromatics in an order of magnitude of 4 to 5 million tpy. In consequence the demand for aromatic blending stocks will be reduced considerably and a loss of the heavy pygas value is very likely.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 65 Surplus of heavy pygas will face ethylene producers not only with reduced profits but also with negative impact on the ethylene production rate. Since heavy pygas yield is proportional to ethylene production, an aromatics oversupply may force a turn down of the whole steamcracker. This post AOP II scenario will be even worse if refiners follow the suggestion to meet the aromatic specification by dilution with light naphtha isomerisate. The removal of light naphtha from steamcracker feedstock pool further increases the heavy pygas to ethylene product ratio. Declining naphtha qualities and climbing prices respectively are already observed.

2 Heavy Pygas Processing

2.1 State of the Art Technologies

The lack of p-xylene and a high portion of ethylbenzene means that heavy pygas is not a prime feedstock for p-xylene production. Even the output of a large steam ­ cracker would not have the economy of scale to justify the investment of a new xylene plant. Hydrodealkylation (HDA) should be the most suitable state of the art technology for converting heavy pygas. However, it must compete with benzene extraction from steamcracker pyrolysis gasoline and this source shall be enough to cover future benzene demand.

Since state of the art technology and European aromatic markets can hardly cope with heavy pygas surplus, alternative utilisation for heavy pygas gain in importance. For some ethylene producers it might be inevitable to find an alternative processing route for upgrading their heavy pygas to become independent from gasoline market.

2.2 The New ARINO™-Technology

It has therefore been thought justified to investigate new concepts and as a result LINDE AG, VEBA OIL and SUDCHEMIE AG have jointly developed the

ARINO™-Technology.

ARINO™, the short form of Aromatic RINg Opening, marking a technology for upgrading heavy pygas to a high quality steamcracker feedstock by hydroconverting aromatic compounds to paraffins.

2.3 Process Description

The conversion is carried out in two reaction steps. First the aromatics are saturated to the corresponding naphthenes under medium pressure and at moderate temperature using a commercially available nickel catalyst. Secondly the naphthenic rings are opened and the molecules are cracked to smaller paraffins in the presence of hydrogen. This reaction step requires elevated pressure and temperature. The catalyst used is palladium on ZSM 5 zeolithe carrier and has been optimised for this application. Figure 5 illustrates the basic principle and process parameters.

66 The ARINO™-Process consists of three process groups, the two reaction steps and a cryogenic product recovery. Figure 6 shows a simplified process flow diagram.

Heavy pygas processed by the ARINO™-Process has to be selectively hydrogenated and desulfurized using state of the art technologies which are usually applied up stream in steamcrackers or aromatic processing units.

Heavy pygas is then routed to the first step of the ARINO™-Process for saturation of the aromatic compounds. This process step design is very similar to conventional gasoline hydrogenation. Heavy pygas is diluted with product, mixed with hydrogen, preheated in a feed/effluent heat exchanger and passed to the reactor. The reaction is performed in various fixed beds which are operated under trickle phase conditions. Temperature control is maintained by liquid product quench, heat of reaction is removed with cooling water. After phase separation, hydrogen and naphthenics are compressed, mixed and fed to the second process step for ringopening and cracking to convert naphthenes to paraffins. The feed mixture is preheated and vaporised in a series of feed/effluent heat exchangers and superheated to reaction temperature in a fired heater. A single pass operation is advantageous and the reaction is performed in a cooled reactor at nearly isothermal conditions. Heat of reaction is used for high pressure steam generation. After heat exchange with the feed, the reaction effluent is led to the separation section. The paraffins produced are separated from the hydrogen surplus by successive partial condensation. Since the light paraffins are the most valuable products the recovery requires very low temperature and C3- and C2-refrigerant has to be applied. A small portion of unconverted aromatics is recycled to the reactors and a bleeding purge prevents methane build up in the hydrogen loop.

2.4 Material Balance

A typical material balance for the ARINO™-Process is given in Figure 7. Hydrogen is the only feedstock in addition to heavy pygas that is required. By rule of thumb the hydrogen consumed is about 10% of the weight of the converted heavy pygas.

Apart from a small fuel gas fraction the ARINO™-Process produces a single product only, a paraffins fraction concentrated in normal paraffins. As shown in Figure 8 the ARINO™-product contains 80% of ethane and propane and is therefore an excellent feedstocks for ethylene plants. The table also indicate the yields achieved by steamcracking of the ARINO™-product. Their product value is about 28% higher than that of naphthacracking, the ethylene yield about 40%.

The ARINO™-Process can be added to a steamcracker easily. All utilities required can be supplied from the cracker and the generated steam can be introduced into the steamsystem. In most cases hydrogen demand has to be balanced by import, the required amount depend on the consumption of existing downstream processes. The main interfaces between the ARINO™-Process and a steamcracker are shown in Figure 9.

67 3 Economics

The ARINO™-Process offers a lot of advantages. Primarily the ARINO™-Process is in line with the requirements of AOP II regulations and is a appropriate answer to heavy pygas surplus and decreasing gasoline demand. Refineries may profit from reduced need for gasoline dilution and additives import. In addition to this the ARINO™-Process makes the steamcracker operation independent from gasoline market and also more flexible in processing heavier naphtha qualities. While the ARINO™ product substitutes feedstock, naphtha imports can be reduced and at the same time the production capacity can be increased since the ARINO™-Product yields more ethylene.

Economic evaluation has shown that the economics of the ARINO™-Process are most sensitive to the heavy pygas value. A drop of heavy pygas value is a necessary condition to make the ARINO™-Process economic. The economy is improved by moderate hydrogen cost and is also influenced by naphtha price and quality. The extent of a future value drop of heavy pygas is hardly predictable and the point of break even for the ARINO™-Process depend on the very specific conditions at the respective production site.

68 Almost all Heavy Pyrolysis Gasoline is Blended into Refineries Gasoline Pool Fig.2: Typical Heavy Pyrolysis Gasoline from Naphtha Steamcracking

Hv. pygas / Ethylene ratio : ;'.0,3.V 0:4 t/L AMMO Bolling range 2iO "C 68 - 87 % wt Aromatic RINg Opening Aromatics content bil :tolv6n»8rvdEl''yttienz'«r»': Octane number (RON)

Technology for Upgrading of Low Value Aromatics to High Value Steamcracker Feedstock Gasoline blending is the common utilization

ARINO

In the Future . Heavy Pyrolysis Gasoline Blending Will Be Limited Upgrading of Pyrolysis Gasoline is Required Fig.3: European Gasoline Pool Fig.4:

. Heavy Pygas will loose its value

Gasoline production : : tpy; ;i 7 : ./. ■ Alternative processing routes Aromatics content (today) •. ^-39:%;: average :• have to be Investigated : (future) ..iiS iyai'iAQPIliiSRecifjcatipn -

I UNDE AG, VEBfl OIL and SODCHEMIE AG Surplus of 4,4 x 106tpy of C7 /C6 aromatics have developed the

Reduced demand for aromatic blending stocks ARH40 - Technology The ARINOTechnology enables the Upgrading The ARINOProcess Uses Only of Heavy Pygas to Steamcracker Feedstock Conventional and Proven Technology 4»-

Fig.5: Basic Principle and Parameters Saturation of Aromatics Naphthenic RINgOpening Cryo.

+ -3H2 Aromatics > ^ * Naphthenes Paraffins

Catalyst Ni ' ZSM5 + Pd

p ±.as t»r ||j|j|§|§||

T ~ 120°C

AMMO-

The ARINOProcess The ARINOT- Product Is Free of Residual Byproducts is a High Quality Steamcracker Feedstock Fig.7 Material Balance Fig,8

Comp. (wt%) Steamcracker Yield * Ha (100%) f ' | Purge* h2 . . 0,2 H2 (85%) ... • 4:89;. | 0,14 t/hr 3,1 t/hr j: ■ . ch4 1,9 Fueigas ' . 21,77 I ARINO”' | Ethane ■ 14,0 Ethylene - 47,26:.- Propane . 66jd , Propyiene 14,12 :

Heavy Pygas ; \ C2-C5 Paraffins n - Butane 8 ,1. , C4 - cut : 5,81 .

30,34 t/hr i . i 33,3 t/hr i - Butane 8,0 .. Pygas 5,15

I------J 3,8 ; Pyoil 0,98 .

ARINO 'after hydrogen recovery with PSA ARihIO - ' maximized for Ethylene The ARINO ™ Process Converts The ARINOProcess Allows Independence Heavy Pygas to Total Extinction from Gasoline and Aromatics Markets Fig.9: Integration into a Steamcracker Fig.10:

Advantages . pm°< . To meet AOP II and decreasing gasoline consumption Separation .„.n . Reduced gasoline dilution requirements

, Independence from gasoline market -Steam - Refrigerant ■ Independence from naphtha qualities

» Reduction of naphtha import 1 ARINO™ j Process • Increased ethylene production

mmr ARINO*

Drop of Heavy Pygas Value Makes the ARINO - Technology Economically Fig.11:

Overall economics depend on specific conditions of the production site and are

■ most sensitive on heavy pygas value

■ improved at moderate hydrogen cost

• influenced by naphtha price and quality

A reduction of heavy pygas value makes the ARINO™ - Technology economically attractive

ARINO 72 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

S. Kvisle*, H. Reier Nilsen**, T. Fuglerud*, A. Gronvold*, B. V. Vora***, P. R. Pujado***, P. T. Barger***. J. M. Andersen*** ‘Norsk Hydro ASA, Porsgrunn, Norway, “Norsk Hydro ASA, Oslo, Norway, “*UOP LLC, Des Plaines, USA

Methanol to Olefins (MTO): State of the Art and Perspectives

ABSTRACT

Natural gas is an attractive feedstock on account of its vast reserves worldwide. World gas reserves have continued to increase as a result of the discovery of new gas fields and increasing conservation methods used in the recovery of associated gas from crude oil production. While processes are not yet available for the direct conversion of natural gas to olefins, its conversion to methanol is well established, with the trend being toward the construction of ever larger methanol plants, presently being built for up to 5,000 metric tons per day (MTD) and soon possibly for up to 10,000 MTD. The availability of low cost methanol provides a very attractive economic incentive to selectively convert it to ethene and propene and further to downstream derivatives such as polyethylene, polypropylene and PVC through the Gas-To-Olefins (GTO) and Gas-To-Polymers (GTP) concepts.

The UOP/Hydro MTO process has been jointly developed by UOP and Norsk Hydro for the selective production of ethene and propene from either crude or refined methanol. The catalyst used in the process is based on a silicoaluminophosphate, SAPO-34. The technology has been extensively demonstrated in a demo plant by Norsk Hydro, and is being considered for commercial use in various projects. The MTO process converts methanol to ethene and propene at close to 80% carbon selectivity in a fluidized bed reactor. The carbon selectivity approaches 90% if butenes are also accounted for as part of the product slate. There is considerable flexibility in the operation of an MTO unit. Typically, the C27C3= ratio can be modified within a range from about 0.75 to 1.5 by adjusting the operating severity, with higher temperatures leading to higher C27C3= ratios. The UOP/Hydro MTO process is a major step forward compared to earlier MTO technologies.

For gas monetization GTO is more economical than GTL (fuels) and LNG, and MTO competes well with naphtha crackers for light olefin production. GTO/MTO introduces through the use of cheap natural gas a new feedstock which is less sensitive to oil price and offers an ethene/propene flexibility that can meet the increased future need for propene.

Introduction

Natural gas reserves are increasing as the rate of new discovery is greater than the rate of consumption. Proven reserves amount to roughly 141 trillion cubic meters. However, almost 60% of these reserves can be categorized as remote or stranded (1). Monetization of these gas reserves is an important factor for nations and companies. An additional aspect is the considerable amounts of associated gas that are still flared or vented, estimated to approx. 120 billion cubic meters in 1996 (2). There is a clear goal to find solutions for reduced flaring, and gas conversion may play a key role in this respect.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 73 There are many options for utilisation of natural gas as illustrated in Figure 1.

Electricity (Ammoni; Petrochemicals

Natural or Olefins Associated (GTO) f

Methanol

Other Chemical Products

Figure 1. Natural gas utilization

Chemical conversion to ammonia, methanol and refinery hydrogen is widely practiced but, although there are large markets for these base chemicals, in terms of gas consumption it represents only approx. 5% of global annual gas consumption (3). Alternatives for use of large amounts of natural gas include liquefied natural gas (LNG), conversion to liquid fuel products (GTL), and conversion to olefins (GTO) and further to polymers (GTP).

MTO/GTO has attracted more and more industrial interest over recent years. In addition to a current strong interest for the UOP/Hydro MTO process, there are also other developments. Lurgi has recently announced their Methanol-To-Propylene (MTP) process (4), ExxonMobil is trying to establish a strong MTO patent position and the Casale Group has purchased the rights to a Romanian MTO catalyst (5). AtoFina recently gave a joint presentation with UOP in Qatar on GTO/GTP perspectives for the Middle East (6), and at the recent 6th Natural Gas Conversion Symposium in Alaska a considerable increase in the interest for MTO technology could be noticed, exemplified by a presentation by BP titled “Market Led GTL: The Oxygenate Strategy ” (3). This presentation discussed the promise of oxygenates in power generation, domestic fuel, transportation fuel and for olefin production.

As a background to MTO and GTO it is also necessary to review some development trends in light olefins production, starting with Table 1 that shows the feedstock distribution for light olefin production (7). In Europe naphtha is the dominant feedstock while in the US and on world basis lighter feedstocks like ethane and LPG play a much more important role. Important aspects of feedstock supply is future availability, future pricing and cracker product distribution, in particular the ethene/propene ratio. This picture is of course complex and varies from region to region, but some general trends are seen.

74 Table 1. Olefin feedstock distribution (7)

Feedstock World Europe USA Ethane 29 5 55 LPG 11 10 17 Naphtha 54 75 23 Gas oil 6 9 4 Others 1 1 1

For naphtha there will probably be no supply problem. There is global sourcing with relatively low transport cost. However, the price follows the crude price like a shadow and high crude prices lead to high olefin feedstock costs. Furthermore, the propene/ethene ratio of typically 0,48 - 0,60 for a naphtha cracker is not in accordance with the consumption in e.g. Europe which corresponds to a ratio of 0,68 - 0,78. Naphtha crackers are expensive to build and cracker byproduct prices show a tendency to erode, favoring lighter feedstock.

Ethane cracking is economically attractive in areas where there is low risk for exposure to high gas prices due to competition from alternative uses (e.g. the energy market in Europe and US). Ethane extraction is, however, expensive and ethane cracking does not meet the increased propene need. LPG cracking and propane dehydrogenation are options, but can suffer from high feedstock costs in some regions, e.g. Europe.

MTO Catalysis and Reactor Technology

The MTO reaction is shown schematically in Figure 2.

Molecular Ethylene

Methanol Propylene

Butenes

Other By-Products h 2o II ..CO x ( ,-< I'.ll.lllllls

Figure 2. MTO reaction scheme

75 Methanol first goes through a dimethylether (DME) intermediate, and the reaction proceeds further to yield ethene and propene. A limited amount of butenes and other higher olefins is produced as well. The MTO reaction is exothermic. Carbon or coke accumulates on the catalyst and must be removed to maintain catalyst activity. The coke is removed by combustion with air in a catalyst regenerator system. Other byproducts include very small amounts of C,-C4 paraffins, hydrogen and COx.

The conversion of methanol to olefins requires a selective catalyst. The UOP/Hydro MTO process utilizes the highly selective MTO-100 catalyst which is based on SAPO-34, a silicoaluminophosphate sieve with a pore size of about 3.8 A. An alternative MTO catalyst is the medium-pore zeolite ZSM-5. Figure 3 illustrates the structures of SAPO-34 and ZSM-5 relative to the dimensions of key MTO products as well as typical product yields for the two types of catalysts (8).

SAPO-34 ZSM-5

-< y#- -< y *. Ethylene

Propylene >■* rr y«r' XA > -< y x x y j. Isobutylene

X A X*- -* y y-r y-T v» Benzene

Paraffins COx

Figure 3. Structures of SAPO-34 and ZSM-5 and MTO product yields (6)

The mechanism of the MTO reaction over different catalytic systems has been widely studied and a recent review is found in (9). Mechanistic studies are continuously being published, but will not be reviewed here. However, some important MTO characteristics of different types of MTO catalysts should be mentioned.

Small-pore molecular sieves (e.g. SAPO-34): » Major olefin products are ethene and propene • Fast deactivation by aromatic coke • SAPO mol sieves are more stable than corresponding zeolite structures towards coking (10)

Medium-pore zeolites (e.g. ZSM-5): • Major olefin product is propene • Significant C5*/aromatics by-products • Slow deactivation by aromatic coke

A high yield of ethene and propene has been a critical goal in the UOP/Hydro MTO process, and this is achieved with the SAPO-34 based catalyst. Depending on design and operation of the MTO unit, the overall yields of ethene plus propene are close to 80%, based on the carbon content of the methanol feed.

76 The above characteristics of the different types of MTO catalysts are important for choice of reactor technology and process design. A fluidized bed reactor and regenerator system is ideally suited for the MTO process based on SAPO-34. Different types of fluidized bed reactors can be used for the methanol conversion. A dense phase reactor with a continuous or periodic removal of catalyst through a slip stream to a regenerator offers excellent performance characteristics. By its nature, however, dense phase fluidized bed reactors operate at fairly low superficial gas velocities in the order of 1 m/s, and low space velocities. Thus, it is advantageous to minimize catalyst inventory requirements by employing a higher level of fluidization that in commercial practice can range from a circulating fast fluidized bed reactor to a dilute phase riser type reactor. Broad commercial experience exists in the commercial design and utilization of the latter types of reactor in FCC (fluidized catalytic cracking) applications (cf. (11), fig. 2). Because of its flexibility, a fast fluidized bed reactor offers significant operating advantages while reducing the catalyst inventory. A fast fluidized bed reactor is therefore chosen for the UOP/Hydro MTO process (12).

In Lurgi’s MTP process propene yields of approx. 70 % are obtained using a ZSM-5 based catalyst in a fixed bed reactor (4), the remaining products being primarily higher hydrocarbons. Lurgi’s process is aimed at propene as dominant product while the UOP/Hydro process has both ethene and propene as main products. For an MTO process aimed at high combined yield of ethene and propene a fast fluidized bed reactor with a SAPO-34 based catalyst represents the state-of-the-art.

UOP/Hydro MTO Process Description

The process flow scheme for a UOP/Hydro MTO process unit has been described elsewhere (13,14) and will only briefly be reviewed here. Figure 4 illustrates a typical process flow scheme.

Quench Caustic C2 Reactor Regenerator Tower Wash' Pe-C2 De-Cj Splitter Pe-C3 Splitter Pe-C4

Tail Gas

Figure 4. MTO process scheme

77 Methanol feed can be used either totally or in part as scrubbing liquid to recover the small amounts of DME in the reactor effluent before being introduced in the reactor. In the reactor, the conversion of methanol and DME to light olefins proceeds to completion (> 99.5% conversion) in a very short residence time. The reactor operates in the vapor phase at temperatures between 350 and 600 °C, and pressures between 1 and 3 bar gauge.

In the process of converting methanol to olefins a certain amount of coke is laid on the catalyst. Coke accumulates on the catalyst and must be removed to maintain catalyst activity. The coke is removed by combustion with air in a catalyst regenerator system. A slip stream of catalyst is circulated to the fluidized bed regenerator to maintain required activity. The operation of the reactor system is characterized as stable steady-state.

After the oxygenate recovery section, the effluent is further processed in the fractionation and purification section to separate the key products from the byproduct components. Ethene and propene are produced as Polymer Grade products and sent to storage.

A MTO Demo plant is operating at Hydro’s research facilities in Porsgrunn, Norway (see below). Detailed analysis of the MTO reactor effluent from this demo unit has been conducted while operating on commercially produced methanol feedstock. The analyses have identified the composition of acetylenic, diolefinic, and oxygenated contaminants present in the effluent product. No unique or unusual contaminants have been found in the MTO effluent and the effluent only requires conventional processing to produce Polymer Grade ethene and propene products. In fact, the levels of acetylenic and diolefinic compounds in the effluent from an MTO unit are significantly lower than from a cracker, simplifying the product recovery section. Regarding oxygenates, a direct side-by-side comparison of the oxygenates present in the effluent from an MTO unit and from an LPG cracker has confirmed that the same oxygenates are present in both streams, but at higher levels in the MTO unit (13,14).

The UOP/HYDRO MTO process design includes equipment to recover the major oxygenates and recycle them for conversion to olefins. Testing has been done to verify the impact of recycling these oxygenates on the MTO process design and performance. The remaining minor oxygenated impurities are removed from the product streams using proven technology that has been in commercial operation for many years, including UOP’s ORU process (13,14). Tests have also been conducted to demonstrate the removal of contaminants from the MTO reactor effluent and the suitability of the product for polymerization applications.

Ethene/Propene Flexibility

The UOP/HYDRO MTO Process offers a wide range of flexibility for altering the relative amounts of ethene and propene products by adjusting the operating severity in the reactor. The process can be designed for an ethene/propene ratio between 0.75 and 1.5. The overall yield of light olefins (ethene plus propene) changes slightly over this range with the highest yields achieved with about equal amounts of ethene and propene, roughly in the 0.8 to 1.3 range. This envelope provides the lowest methanol requirements, but the ratio can be adjusted to reflect the relative market demand and pricing for ethene and propene.

78 An example of product distribution is shown in Figure 5 for the production of 800,000 MTA of light olefins with equal amounts of ethene and propene. Approximately 3 tons of methanol are required per ton of light olefin or approx. 7000 t/d. This represents a carbon-based yield of almost 80%.

H Ethene

■ Propene

■ Butenes

HC5+ hydrocarbons ■ Fuel gas

Figure 5. Hydrocarbon product distribution for 800.000 MTA MTO plant.

MTO Demo Plant

An MTO Demo plant is operating at Hydro’s research facilities in Porsgrunn, Norway. The plant, which is fully integrated with reactor and regenerator, has a design capacity of 0,5 MT of methanol per day, but the plant has been operated at twice this capacity. One of the major goals of the plant was to demonstrate the stability of the process using methanol from a commercial plant. As can be appreciated from Figure 6 excellent stability is demonstrated.

sS Conversion £. WO , | 80

t 60 Selectivity to C2 I 40

I 20 Selectivity to C3E U 0 0 10 20 30 40 50 60 70 80 90

Figure 6. Demo plant results

79 The MTO-100 catalyst has also been subjected to several hundred cycles of reaction and regeneration in lab scale fluidized bed reactors and also to steam treatment. The catalyst shows very good stability towards irreversible deactivation.

Based on the extensive tests in the demo plant, the UOP/Hydro MTO process can be scaled up to a reactor/regenerator single unit capacity of up to one million tons of light olefins per year.

MTO Capital Costs

MTO unit capital costs have been estimated by UOP and also by independent EPC contractors in connection with potential projects. Detailed cost estimates have been developed including quotes for key equipment items. Total capital cost for an 800.000 MTA light olefin plant is 388 mill. USD (1999 US Gulf Coast). ISBL erected costs is estimated to 214 mill. USD, as verified independently by several EPC contractors. Costs for offsites and utilities (OSBL) have also been estimated for MTO projects and amounts to approx. 75 mill. USD for an integrated 800.000 MTA plant. These costs can vary significantly between projects but have generally been equivalent to about 35% of the ISBL estimated erected costs for integrated MTO projects. OSBL costs would include single-day storage facilities for intermediate products. For stand-alone MTO plants OSBL would amount to approx. 50% of ISBL costs. Other project costs are estimated to 99 mill. USD and include catalyst and adsorbent loading, allowances for project development and management costs, technology license fees, training, and start-up costs.

An economical comparison of an MTO unit and a naphtha cracker has been presented recently (15), using an ethene capacity of 500 kMTA. The IRR for the cracker was calculated for the years 1995-1999 using historical contract prices for naphtha and olefins. This gives a picture of the historical relationship between IRR and naphtha and olefin prices for a naphtha cracker. In a similar way, assuming a delivered methanol price of $90/ton to the MTO plant, the IRR of the MTO unit was also calculated using the historical olefin prices in the same time span. It was concluded that with this methanol price the IRR break-even point between a cracker and an MTO unit is around a naphtha price of 150 $/ton, corresponding to a crude oil price of around 15$/bbl. This can be taken as an indication of the competitiveness of an MTO unit with a naphtha cracker with a delivered methanol price in the order of $90 /ton.

Economics - GTO and GTP vs. GTL (Fischer-Tropsch) and LNG

MTO represents a gas utilization technology through the GTO concept, and a comparison with other gas utilisation alternatives is therefore of interest. Such an economical comparison of GTO/GTP shows the strong competitiveness of GTO/GTP as a gas utilization option (6,14). GTP represents a fully integrated complex producing polyethene and polypropene pellets from natural gas. A possible location for such a complex is the Middle East. Table 2 and 3 show the basis for and result of this comparison.

80 Table 2 - Gas monetization: economic basis

ITEM COST

Crude oil $18 /barrel Natural gas feed $ 0.75 / mm Btu LNG product $ 4.00 / mm Btu GTL products $23/ barrel GTP polyethylene $ 800 / MT GTP polypropylene $ 705 / MT GTP investment + 15% over U.S. Gulf Coast T ransportation $40 / MT polyolefins $12 / MT syn fuels (GTL) $70 i MT LNG Fixed operating costs 5% of ISBL (erected)

Table 3- Economic comparison gas monetization alternatives

LNG GTL GTP

Capacity 4,75 mm MTA 50.000 bpd 400.000 MTA PE liquid fuel 400.000 MTA PP Ann. gas consump. (BCM/yr) 7,9 4,8 2,0 20 yrs gas consump. (BCM) 160 96 40

Gas cost, $ mm / year 213 129 54 Operating cost, $ mm / yr 93 64 105 Total cash cost, $ mm / yr 306 193 159

Product revenue, $ mm / yr 965 382 602 - Transportation costs 333 25 32 Net revenue, $ mm / yr 632 357 570

Gross profit, $ mm / yr | 326 | 164 | 411

Capex, $ mm (low) 2000 1250 1550

Simple ROI 16% 13 26

Table 3 shows the economic comparison as simple return on investment (ROI). It should be emphasized that this simple comparison is based on generic economics and not based on data for a specific site. Nevertheless, GTP offers significantly higher returns out of these gas monetization options. GTO offers similar returns to GTP provided that the light olefins produced in the GTO cases are transferred across the fence to nearby olefin derivative facilities.

81 LNG and GTL require lower gas prices to be economical compared to GTP, see Figure 7. As gas prices increase, LNG and GTL returns decline faster because the higher value added by the GTO and GTP processes provides much higher net revenues per unit of natural gas consumed.

Table 3 illustrates that LNG and GTL have the potential to consume more gas than GTO/GTP. On the other hand more gas fields can support GTO/GTP than LNG or GTL. For GTL and LNG to be economical, it requires a gas field with reserves of about 100 billion cubic meters or more, which limits the number of potential sites. GTO/GTP can be used on gas fields with capacities of about 40 billion cubic meters.

35

0 0,25 0,5 0,75 1 1,25 1,50 Gas price ($/mmBTU)

Figure 7. Gas monetization ROIs

Since the syn gas facilities is the major cost element in methanol and GTL production an interesting option is to integrate GTO/GTP and GTL facilities. This can increase revenues and enhance the overall economics of integrated plants. This is discussed in more detail in (16).

MTO and GTO perspectives

In general, commercialization of large scale gas conversion technologies are challenging. Investments are high, technologies are often not well proven at the actual scale, there is competition with crude oil based products and plant location is often in remote areas. This is evidenced by the slow commercialization of GTL syn fuel plants despite the fact that the technology has been available for several decades.

82 However, as mentioned in the introduction there are several strong drivers for increased gas utilization and also several challenges for olefin producers. As a response to this GTO/MTO has the following specifics:

• GTO/GTP is the most economical gas monetization alternative and plant capacity is flexible with regard to gas field size. • GTO/MTO introduces through the use of cheap (remote) gas a new feedstock which is less sensitive to oil price. • MTO is very flexible with regards to propene/ethene ratio and the increased future propene demand can be met without a large co-production of ethene. • The olefin market is big, has a healthy growth and can consume large amounts of gas without being disturbed. • GTO can be developed as a chain where methanol is produced where cheap gas is located, and shipped to an MTO plant close to the olefin market. • GTO can be constructed as a fully integrated GTP complex with pellets shipped to the market

A very interesting perspective for MTO/GTO is the current development of mega methanol plants. A number of plants with capacities of 5000-10000 t/d, compared to the present 2500-3000 t/d, are being developed, which can produce methanol at low cost. Factory gate cost of $80/ton or less is possible (17). This makes MTO very attractive. Furthermore, methanol shipping costs have been reduced significantly over the last few years through larger vessels and dedicated ships which means that methanol can be delivered Europe significantly below $100/ton.

Since the market for ethene and propene is large compared to that of methanol itself, the impact of GTO on methanol could be dramatic. If methanol were used to meet one year’ s growth in ethene demand (3,5 mill, tons/yr), methanol production would have to increase by 14 mill tons or almost 50% (3).

The major challenges for MTO/GTO are confidence in new technology, in particular since investments are high, competition at low naphtha prices and also the fact that a GTO chain requires complicated decision-making since the chain involves both E&P and chemical companies.

Conclusions

Several MTO projects are currently being studied, of which the EATCO project in Egypt has been announced (18,19). The abundance of natural gas and the economy of mega methanol plants have made the MTO process a potent future alternative to the naphtha cracker for the production of ethene and propene. The MTO process is therefore the new gateway for the gas producers to market their resources.

References

1. “Stranded Gas Utilization. Methane Refineries of the Future". Chem Systems 2001. 2. M-F. Chabrelie (Cedigaz). Gas-To-Liquids World Forum, London, November 12-13, 1998

83 3. T. H. Fleisch, R. Puri, R. A. Sills, A. Basu, M. Gradassi and G.R. Jones, Stud. Surf. Sci. Catal, 136,423 (2001). 4. H-D. Holtmann, Lurgi Oel Gas Chemie GmbH. ERTC Petrochemical Conference, Rome, Italy, February 21-23, 2001. 5. Nitrogen & Methanol, 252, 15 (2001). 6. F. Bouvart (AtoFina) and S. Gembicki (UOP). 4th Doha Conference on Natural Gas, Doha, Qatar. March 14, 2001. 7. Chem Systems. Report to Norsk Hydro. 2000. 8. P. Barger (UOP). IZA Pre-Conference School. Poitiers, France. July 5-7, 2001 9. Microporous and Mesoporous Materials, 29 (1999). Special Issue Catalysis: Methanol to Hydrocarbons. Eds. M. Stocker and J. Weitkamp. 10. Yuen et al. Microporous Materials, 2,105 (1994). 11. B. Miller, J. Warmann, A. Copeland, and T. Stewart, Fast-tracked FCC revamp, Oil & Gas J., 46-53 (5 June 2000). 12. L.W. Miller, Fast fluidized bed reactor for the MTO process, US Patent allowed. 13. B.V. Vora, P R. Pujado, L. W. Miller, P.T. Barger, H. Reier Nilsen, S. Kvisle and T. Fuglerud. Stud. Surf. Sci. Catal, 136, 537 (2001). 14. B.V. Vora, P R. Pujado, J. M. Andersen, P.T. Barger, H. Reier Nilsen, S. Kvisle and T. Fuglerud. 6,h World Congress on Chemical Engineering, Melbourne, Australia. September 23-27, 2001. 15. J. Heber (Norsk Hydro). 3rd Asia Olefins and Polyolefins Markets Conference 2000, Bangkok, Thailand. January 20-21, 2000. 16. S.T. Bakas, J.M. Anderson, B.V. Vora, P.R. Pujado, H. Reier Nilsen, S. Kvisle and T. Fuglerud, IBC Gas-to-Liquids Conference, Houston, USA, September 17-19, 2001. 17. J. Haid and U. Koss. Stud. Surf. Sci. Catal., 136, 399 (2001). 18. European Chemical News, 6-12 March, 37, 2000. 19. Nitrogen & Methanol, 252, 9, 2001.

84 DGMK-Confefence “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

M. Rothaemel, H.-D. Holtmann Lurgi Oel Gas Chemie GmbH, Frankfurt, Germany

MTP, Methanol to Propylene - Lurgi s Way

1 INTRODUCTION

What are the incentives to promote technology development in the conversion of natural gas to valuable products? • plentiful gas supply sources, • environmental aspects and regulations, and • “monetizing ” the abundant natural gas reserves in remote areas. Both economic and environmental benefits from the use of natural gas are the driving forces and will support the continuous innovation regarding gas-based technologies. Since Lurgi introduced its new groundbreaking Mega-Methanol process for plants with a production of 5,000 tons methanol per day, methanol will be available at a constant low price in the forseeable future. This development has an enormous im­ pact on downstream technologies for the conversion of methanol to more valuable products. In this field, the use of methanol as feedstock for the production of olefins is one of the most promising new applications. Lurgi’s new Methanol To Propylene (MTP) process presents a simple, cost-effective and highly selective technology, yielding an excellent value-added product for the utilization of natural gas reserves via syngas and Mega-Methanol.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 85 2 NATURAL GAS - A KEY FEEDSTOCK FOR THE 21st CENTURY

The total proven gas reserves amount to approx. 140 trillion cubic meters worldwide [EIA 2001] which translates into a gas reserve-to-production ratio, i.e. a gas reserve lifetime of 61 years (see Figure 1). Furthermore, estimated additional gas reserves will cover a lifetime of 65 years more. Compared with the reserve lifetime of 41 years for petroleum and 230 years for coal, there is no doubt that natural gas will be a key fuel component in the 21st century [Punt 2000],

Figure 1. Proven Natural Gas Reserves-To-Production Ratio.

Among the world regions (Figure 1), the Middle East and Eastern Europe including the Former Soviet Union (FSU) possess the largest fraction of global natural reserves with 35% and 38%, respectively [EIA 2001]. North America and Europe, on the other hand, have gas reserve lifetimes of about 10 years only (see Figure 1), so these regions soon will be customers for natural gas and gas-based products. Due to the long distance to the gas producing regions, lo­ gistics and technologies of natural gas transport present serious challenges. The situation is complicated because the cost of transporting gas is high compared to crude oil; hence it will become increasingly important to develop or improve tech ­ nologies to convert natural gas to valuable products which can be transported cost- effectively to the markets. Existing technologies for natural gas conversion are shown in Figure 2: via the con­ version to syngas, hydrogen and ammonia, Fischer-Tropsch products, methanol and DME are produced. Currently, the production of chemicals requires around 5% of world gas consumption [Quigley and Fleisch 2000].

86 Lurgi OelGasChemie GmbH focuses on a new route from Ci to a valuable product by combining a chain of proprietary Lurgi technologies that base on low cost natural gas supply, large scale methanol plants and an exciting new process for the highly- selective conversion of methanol to propylene.

Figure 2. Uses of Natural Gas.

3 PROPYLENE - AN ATTRACTIVE PRODUCT

The world wide propylene supply/demand amounts to 53.5 million tons for 2000 [Zinger 2001]. Polypropylene is by far the largest and fastest growing of the propyl­ ene derivatives, and requires the major fraction of 58 % of the total propylene. The increasing substitution of other basic materials such as paper, steel and wood by PP will induce a further growth in the demand for PP and hence propylene. Other im­ portant propylene derivatives are acrylonitrile (10 %), oxo-alcohols (8 %), propylene oxide (7 %) and cumene (6 %). The predicted growth rates of the capacities for the derivatives are very impressive, i.e. 5.8% for PP. The average growth rate for propyl­ ene itself is estimated to be 5.6 % per year by 2004. How to satisfy this increasing demand for propylene? Currently, steam crackers and FCC units supply 66 % and 32 %, respectively of pro ­ pylene fed to petrochemical processes. However, as FCC units primarily produce motor gasoline, and steam crackers mainly ethylene, propylene will always remain a by-product (e.g. 0.04-0.06 t/t of ethylene for steam crackers with ethane feedstock and 0.03-0.06 t/t, respectively of motor gasoline and distillates production for FCC

87 units). Current forecasts indicate an annual growth rate of 4.4 % for steam cracker production and 7.2 % for FCC units by 2004. This results in an increasing gap of pro­ pylene production that has to be filled by other propylene sources. Lurgi’s new MTP process directly aims to fill that gap.

4 LURGI’S MEGA-METHANOL TECHNOLOGY

The term Mega-Methanol refers to plants with a capacity of more than one million metric tons per year. To achieve such a large capacity in a single-train plant a special process design is required. For this reason Lurgi focused on the most efficient inte ­ gration of syngas generation and methanol synthesis into the most economical and reliable technology for the new generation of future methanol plants [Streb and Gohna 2000] (see Figure 3).

improved gasification Gas-Cooled Water-Cooled Reactor Reactor high energy efficiency for MeOH synthesis Boiler low investment costs Feed Water large single-train capacity [Syngas methanol price: 80 $/t

Crude Methanol

Figure 3. Simplified Diagram of Lurgi’s Mega-Methanol Technology.

Autothermal Reforming. Pure autothermal reforming can be applied for syngas production whenever light natural gases are used as feedstock to the process. The desulfurized (and optionally pre-reformed) feedstock is reformed to synthesis gas at about 40 bar using oxygen as the reforming agent. Even when using pure methane as feedstock to the autothermal reforming, it is necessary to condition the synthesis gas, as the stoichiometric number, defined as (H2 - C02) / (CO + C02) on mole ba­ sis, is below 2.0. The most economic way to achieve the required gas composition is a special operation mode of the methanol synthesis with a very high CO conversion and a suppressed C02 conversion. The optimum composition is achieved by recy ­ cling hydrogen that can be separated from the purge stream downstream of the methanol synthesis by a membrane unit or pressure swing adsorption (PSA) unit.

88 Combined Converter Methanol Synthesis (see inset in Figure 3). In this innovative concept the compressed syngas is first used as cooling agent on the tube side of the gas-cooled reactor. In the downstream water-cooled methanol reactor, the pre­ heated gas is converted under near-isothermal conditions while the heat of reaction is utilized for the production of saturated steam. The partly converted gas is then routed to the shell side of the gas-cooled reactor where it is converted to methanol in the catalyst bed. Due to the combination of heat exchange (to preheat the syngas) and reaction, a declining and therefore thermodynamically favourable temperature profile across the catalyst bed is established in the gas-cooled reactor, thus leading to very high per-pass-conversion. The product gas is then cooled, crude methanol is condensed, separated and sent to the distillation unit. The gaseous stream is recy­ cled to the reactor loop after separation of a purge gas stream which, in turn, is routed to the purge gas separation unit where H2 is separated and returned to the syngas, and thus to the synthesis loop to adjust the proper stoichiometric number (as mentioned above). The most important advantages of the water- and gas-cooled reactor concept are as follows: • high syngas conversion efficiency: at the same overall conversion, the recycle ratio is about half of the ratio in a single-stage, water-cooled reactor; • high energy efficiency: in addition to the high-pressure steam generated in the water-cooled reactor, a substantial part of the sensible heat can be recovered at the gas-cooled reactor outlet; • low investment costs: capital cost savings of about 40 % for the synthesis loop can be realized due to the omission of a large feedstock preheater, savings for other equipment due to the lower recycle ratio and by the 50 % reduction of cata­ lyst volume in the water-cooled reactor; • large single-train capacity: studies have confirmed that single-train plants with capacities of 5,000 mtpd and above can be built in one reactor train. To summarize: the unique advantages of the Mega-Methanol technology make this process ideally suited as part of Lurgi’s route from Ci via syngas and methanol to propylene. An environmental sidenote: 100 billion cubic meters of natural gas are flared or vented annually [EIA 2001]. That amount would be sufficient for feeding about 74 Mega-Methanol plants with a capacity of 130 million tons per year in total.

5 LURGI’S METHANOL TO PROPYLENE (MTP) TECHNOLOGY

Lurgi’s new MTP process is based on an efficient combination of the most suitable reactor system and a very selective and stable zeolite-based catalyst. Reactor Concepts. Basically, the methanol-to-olefins reaction can be carried out in both fixed-bed and fluidized-bed reactors.

89 While for this exothermic reaction the fluidized bed has inherent advantages in terms of excellent heat transfer and superior temperature control, there are some disad­ vantages of this reactor type, especially the difficult scale-up procedure and the sta­ bility of the catalyst (attrition, life time). Furthermore, the fluidized-bed catalyst deacti ­ vates rapidly which requires continuous catalyst regeneration. Typically, the regen­ eration temperatures are higher than at reaction conditions which results in frequent temperature shock and additional stress on the catalyst particles. Furthermore, the currently available fluid-bed catalysts are suitable for ethylene-propylene co-produc­ tion only but not for propylene alone, hence the separation train has to include the columns and the cold box required for the purification of ethylene. For catalytic fixed-bed reactors (preferred to operate adiabatically) the temperature control of the methanol conversion is not as straightforward as for fluid beds. Usually, the addition of steam and distribution of the feed into more than one reactor is nec ­ essary to limit the adiabatic temperature increase in the reactors. However, fixed-bed reactors can be easily scaled up, have significantly lower investment costs and pro­ vide a more uniform residence time of reactants so that product selectivities can be maximized. Furthermore, a very selective fixed-bed catalyst is commercially manu­ factured by Siid-Chemie AG and provides maximum propylene selectivity, has a low coking tendency, a very low propane yield and also limited by-product formation. This in turn leads to a simplified purification scheme that requires only a reduced cold box system in comparison with on-spec ethylene production. Based on the extensive expertise of Lurgi in the field of catalytic fixed-bed reactors and of Sud-Chemie in the field of high-performance stable zeolites, we opted for the fixed-bed reactor with Sud-Chemie ’s proprietary catalyst as the basis for Lurgi’s MTP process.

Light End- Crude Methanol

DME Product Propylene Pre-Reactor Regeneration Conditioning i_

MTP 1 MTP 2 MTP 3 ■

Reactor Stages

Water Recycle Gasoline ]

Olefin Recycle

Process • Water ; REACTOR SECTION PRODUCT RECOVERY

Figure 4. Simplified Diagram of Lurgi s MTP Technology.

90 Lurqi's MTP Process (see Figure 4). The methanol feed from the Mega-Methanol plant is sent to an adiabatically operated DME (dimethylether) pre-reactor where methanol is converted to DME and water using a high-activity high-selectivity catalyst achieving nearly thermodynamic equilibrium. The methanol/water/DME stream is routed to the first MTP reactor where also the steam is added. Methanol/DME are converted to more than 99%, with propylene as the predominant hydrocarbon prod­ uct. Additional reaction proceeds in the 2nd and 3rd MTP reactors. The process condi­ tions in the three MTP reactors are chosen to guarantee similar reaction conditions and maximum overall propylene yield. The product mixture is then cooled and the product gas, organic liquid and water are separated.

The product gas is compressed and traces of water, CO2 and DME are removed by standard techniques. The cleaned gas is then further processed yielding chemical- grade propylene with a typical purity of more then 97%. Different olefin-containing streams are sent back to the main synthesis loop as an additional propylene source. To avoid accumulation of inert materials in the loop, a small purge is required for light-ends and the C4/C5 cut. Gasoline is produced as a by-product. Water is recycled to the steam generation for the process; the excess water resulting from the methanol conversion is purged. Where suitable, this process water can be used for agricultural purposes after appropriate treatment. The MTP process operates at slightly elevated pressure (1.3-1.6 bara), moderate steam addition (0.5-1.0 kg per kg of methanol) and low reactor inlet temperatures (400-450 °C).

Figure 5. MTP Production Figures

91 After a cycle of approx. 400-700 hours of operation, the catalyst has to be regener ­ ated by burning the coke with a nitrogen/air mixture. The regeneration is carried out at similar temperatures as the reaction itself, hence the catalyst particles do not ex ­ perience any unusual temperature stress during the in-situ catalyst regeneration pro ­ cedure. Furthermore, by applying a nitrogen purge after the regeneration is fin ­ ished,there is no possibility of oxygen breakthrough into the MTP synthesis loop which facilitates reaching the required propylene specification later on in the purifica ­ tion unit. The simplified overall mass balance is depicted in Figure 5 based on a combined Mega-Methanol / MTP plant. For a feed rate of 5,000 metric tons of methanol per day (1.667 million metric tons annually), approx. 519,000 metric tons of propylene are produced per year. By-products include fuel gas and LPG as well as liquid gasoline and process water. A closer look at the overall hydrocarbon product slate of the MTP process is provided in Figure 6: as can be seen, propylene is the predominant product of the MTP plant with a carbon based yield of more than 70%, while all other components are formed to a much smaller degree. The amount of coke being produced during normal opera ­ tion is extremely low. This allows for long catalyst cycle times until a regeneration becomes necessary.

MTP, Methanol To Propylene - Lurgi's Way MTP Product Slate Carbon Product Yields, wt.%

C2- C2= C3= C3 C4/C5 C6+ coke

Figure 6. MTP Product Slate

92 6 MTP ECONOMICS

The economics have been estimated for a methanol to propylene plant fed by 1.667 million metric tons annually, yielding 519,000 metric tons of propylene. The feasibility study is based on the following main assumptions: • total investment cost (TIC) budget: US$ 185 million • equity: 20 % • depreciation per year: 10 % The calculated internal rates of return (IRR) on total capital employed depend on the methanol feed price and the propylene product price that can be obtained. Different propylene price levels have been considered (see Figure 7). Depending on the site conditions and the natural gas price, Lurgi’s Mega-Methanol technology can reduce the production cost to US$ 80 per metric ton of methanol. Assuming US$ 380 per ton of propylene and US$ 90 per ton of methanol, the IRR will be 15.6 % per year, equivalent to a payback period after start-up of 4.6 years. The internal rate will increase to 28.7 % and the payback period decrease to 2.9 years assuming a methanol price of only US$ 70 per ton. For higher propylene price levels, such as US$ 400 per ton, the IRR figures result very attractive with 32.3 % per year at US$ 70 per ton of methanol, i.e. 2.6 years payback period. Future integration and optimization of the total plant complex including syngas, methanol, propylene production and offsites facilities will further decrease the capital investment and production costs.

Figure 7. MTP Economics

93 7 MTP TECHNOLOGY STATUS

The technological status of MTP in the areas of process and catalyst can be summa­ rized as follows: The basic process design data were derived from more than 4000 operating hours of a pilot plant at Lurgi’s Research and Development Center. Besides the optimization of reaction conditions also several simulated recycles have been analyzed. The next step is a larger-scale demonstration unit in order to obtain more data on catalyst life and to demonstrate the process to potential customers. The demonstration unit will be started up at the end of 2001. Furthermore, there is an ongoing effort to optimize the flow sheet for the commercial-scale process. The catalyst development is completed and the catalyst has been commercially manufactured by the catalyst supplier.

8 CONCLUSIONS

There are abundant natural gas reserves providing low cost feedstock for methanol production and aiming at better use of natural resources especially in case of associ­ ated gases being flared. Propylene produced from methanol will increase the value of natural gas considerably and offers an exciting potential of growth and a high earn ­ ings level. Lurgi’s expertise is well known in designing and building plants to produce synthesis gas from all kind of fossil fuels and convert the synthesis gas into methanol. The plant record is very impressive. Lurgi’s Mega-Methanol technology allows to reduce the methanol production costs to below US$ 80 per metric ton, wherever low cost natural gas is available. This opens up a completely new field for downstream prod ­ ucts like propylene. Based on a simple fixed-bed reactor system, usual processing elements and operat ­ ing conditions including a commercially manufactured catalyst, Lurgi’s MTP technol ­ ogy will provide an attractive way to "monetize" natural gas. Driven by the excellent market prospects and additional environmental aspects, Lurgi Oel Gas Chemie has developed its own technology chain starting from natural gas via methanol to propylene, based on the combination of highly efficient concepts at low investment costs.

REFERENCES Energy Information Administration (EIA): “International Natural Gas Informa ­ tion”, 14 Feb 2001, National Energy Information Center (http://www.eia.doe.gov/emeu/international/gas.html ) Ad. R. Punt: “Shell ’s Perspective on the GTM options", EFl - Gas to Market Conference, San Francisco, October 11 - 13, 2000. Th. M. Quigley and Th. H. Fleisch: “Technologies for the Gas Economy", EFl - Gas to Market Conference, San Francisco, October 11-13, 2000.

94 S. Streb and H. Gohna: „Mega-Methanol™ - paving the way for new down­ stream industries", World Methanol Conference, Copenhagen (Denmark), Novem­ ber 8 - 10, 2000

St. J. Zinger, “Propylene: A Valuation Evaluation ”, 2001 World Petrochemical Conference, Houston, March 28-29, 2001

95 96 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

S. Schimpf, M. Lucas, P. Claus Institute of Chemical Technology II, Technical University of Darmstadt, Germany

Selective Hydrogenation of 1,3-Butadiene over Supported Gold Catalysts

Abstract

We demonstrate the practicability of both supported gold catalysts and A1 microstructures

coated with Au/MOx (MO, : Si02, A1203, Ti02, Zr02) layers containing very low amounts of gold (typically < 0.1 wt. %) for the regioselective hydrogenation of 1,3-butadiene in the gas phase. At temperatures between 493...553 K and butadiene conversions between 3...23 % the selectivity to butenes were > 98% with 1-butene as main product in all cases. The gold catalysts showed deactivation behavior. Based on the gold content of the catalysts, the initial rates obtained with the microchannel reactor were considerably higher compared to the results of the fixed-bed reactor.

Introduction

Palladium based catalysts are widely used in selective hydrogenation of and. 1,3-

butadiene in C4 cuts because these reactions are of vital importance in manufacturing high purity olefine streams. Catalytic activity and selectivity can be changed due to, for example, modification of Pd sites by hydrocarbonaceous deposits or hydrogen transfer from oligomers. The addition of gold gave rise to an enhanced selectivity towards 1-butene whereas the

formation of 2-butenes was not significantly affected [1], On the other hand, during a long time only very limited attention was paid in realizing catalysis on the basis of gold because of its electronic structure, namely the completely filled

d band ([Xe] 4f145d 106s'), which is usually accompanied by very low activities [2], In the

absence of a support material, on pure gold surfaces in general neither H2, CO nor 02 are adsorbed, and hydrocarbons interact only weakly with gold surfaces. The situation has been changed since Haruta and coworkers reported on CO oxidation at room temperature, feasible only on very small gold nanoparticles on suitable supports [3], This observation was followed by enhanced search for other possible applications in catalysis [2], Unfortunately, focussing on oxidation reactions masked the capabilities of gold in hydrogenation reactions, even

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 97 though there are some very promising examples of possible applications [2], For example, we showed that titania and zirconia supported, nanosized gold particles were able to control the intramolecular selectivity in the hydrogenation of multiple unsaturated organic compounds, i.e. chemoselective , for example acrolein [4], It is the aim of this contribution to describe the hydrogenation properties of gold catalysts in a regioselective hydrogenation, namely 1,3-butadiene. Therefore, the present contribution deals, on the one hand, with catalyst systems based on supported (silica, titania and zirconia) gold nanoparticles fc?Au = 1 to 6 nm) which were applied in the selective hydrogenation of 1,3- butadiene in the gas phase using a micro fixed-bed reactor. On the other hand, a special micro reactor was used which contained A1 wafer having microstructures (300 pm x 700 pm) coated with gold on oxide composites.

Experimental

Supported gold catalysts (Table 1) were prepared by the following methods: deposition- precipitation, noted by DP, precipitation (P), incipient wetness (IW) and sol-gel technique (SG). Details of the preparation procedures and catalyst characterization were described elsewhere [4,5].

Table 1: Overview about supported gold catalysts.

Catalyst code Method of preparation Gold content Specific surface Gold particle

[wt.%] area [m2/g] size [nm] Au/Ti02-SG Sol-gel technique 4.8 117 1.1 Au/Ti02-DP Deposition-precipitation 1.7 42 5.3 Au/SiG2-IW Incipient wetness 1.6 171 3.9 Au/Zr02-P Precipitation 1.0 151 1.4

In the case of the microchannel reactor coating of the microstructure with oxide layers was achieved by the following methods: (a) deposition of a pseudobohemite primer prepared by dispersing a commercial aluminum hydroxide powder (Disperal,® Condea Chemie) in HNO3 aqueous solution followed by dipping the microstructured wafers into this dispersion, then drying at room temperature and, finally, by impregnation with an aqueous solution of HAuCLt; (b) spin-coating to obtain the oxide layers followed by dipping the wafer into an aqueous solution of HAuCL, or And;. The obtained materials denoted as Au/oxide/Al catalysts (Table 1) were dried at 393 K (in air) and then at 573 K (in Ar) and finally reduced at 573 K in flowing H2. ICP-OES analyses showed that the gold content was in the range of 0.005 to 0.11 wt.% related to the amount of the deposited oxide layer. Estimation of the core

98 level binding energy of An 4f7 a by using X-ray Photoelectron Spectroscopy (XPS) revealed a value of 84.2 eV which corresponds to Au° [6].

Table 2: Overview about the microstructures coated with Au/oxide layers.

Catalyst code Oxide layer Method of Gold Method of Au Au [mg] oxide coating precursor immobilization [wt.%]

A1VAI2O3-MSR 2 139.0 AI2O3 Dip-Coating HAUCI4 Impregnation 0.005

0.011 A11/AI2O3-MSR 3 196.3 AI2O3 Spin-Coating HAuCU Dip-Coating Au/SiCh-MSR 4 174.3 SiQz Spin-Coating HAuCU Dip-Coating 0.084

Au/TiOz-MSR 5 146.8 TiG2 Spin-Coating AuCl3 Dip-Coating 0.027

0.110 Au/ZrOz-MSR 6 108.7 ZrOz Spin-Coating AuCl3 Dip-Coating

Hydrogenation of 1,3-butadiene (BD) was carried out in the gas phase at temperatures between 513 K and 553 K in (a) an automated micro-fixed bed reactor system and (b) in the microchannel reactor shown in Fig. 1. Product analyses were conducted by gas chromato ­ graphy equipped on-line with the reactor systems.

Fig. 1: Microstructured wafer (size: 10 mm x 50 mm x 1 mm; channel size: 300 pm x 700 ]xm) for coating with gold/oxide catalysts (left part) and photograph of the microstructure reactor system (rightpart) [7],

99 Results and Discussion

Hydrogenation of 1,3-butadiene is highly selective for butene formation (S > 98 %) not only in the case of powdered gold-on-support catalysts used for fixed-bed hydrogenation but also by applying microstructured A1 wafers coated with gold/oxide composites (Table 3).

Table 3: Selectivity of products of 1,3-butadiene hydrogenation over gold based catalysts.

Type of gold catalyst 1-butene trans- cis- n-butane

2-butene 2-butene

supported gold catalysts 52 ±1 28 ±1 18 + 1 2 ±0.3 microstructures coated with gold/oxide 55 ±2 25 ±2 18+1 2 + 0.4

The selectivity pattern shown in Table 3 was obtained with all catalysts used for 1,3- butadiene hydrogenation with the exception of catalyst A11/AI2O3-MSR 2 where the observed selectivity to 1-butene and trans-2-butene was 50 % and 30 %, respectively. Conversion of 1,3-butadiene was always in the range of 10...23 % for supported gold catalysts and 3...12 % for the microstructures coated with gold/oxides. Applying the microstructured wafer without catalyst coating in hydrogenation (blind reaction) gave a conversion of 1 % at 240 °C and 3 % at 280 °C. Fig. 2 represents the results of a typical hydrogenation experiment showing, beside the selectivity behavior mentioned above, that the catalyst exhibited deactivation. This was observed for both types of catalysts. Deactivation did not change theselectivity (Fig. 2).

240 °C

20 40 60 80 100 120 140 160 180 200 220 Time on stream [min]

O conversion v trans-2-butene ■ n-butane • 1-butene a cis-2-butene

Fig. 2: Conversion and selectivity vx time on stream in the hydrogenation of 1,3-butadiene over catalyst A u/A1 20}-MSR 3 (H2/BD = 10, Wo^JFbd = 7.3 gh/mol).

100 Therefore, to compare the activity behavior of the catalysts, initial rates were estimated related to the amount of gold because of the different gold contents of the catalysts (Fig. 3 and

4) and corrected by the contribution of the above mentioned blind reaction.

10 9 «■ 8

zz, 4 h

1

0

Fig. 3: Catalyst activities (expressed as initial rates) of supported gold catalysts in the hydrogenation of 1,3-butadiene (T = 240 °C, H2/BD = 10, Wca! /FsD ~ 2.6 g h/mol).

5

AU/AI2O3/AI MSR2 4

3-1

=L AU/BO2/AI xj- MSR5 O 2 AU/AI2O3/AI MSR3 > AuZ 1 1 Au/Zr02/Ai, MSR6 A1VS1O2/AI MSR4 0 I— ----""---- Fig. 4: Catalyst activities (expressed as initial rates) of Al microstructures coated with gold/oxide catalysts in the hydrogenation of 1,3-butadiene (T = 240 °C, H2/BD = 10,

101 The results showed that the activity of gold catalysts in 1,3-butadiene hydrogenation (a) depends on the support material for both types of catalysts, (b) is up to two order of magnitudes higher in the case of the microstructured A1 wafers coated with Au/oxide catalysts.

Acknowledgement This work has been supported by the Max-Buchner-Forschungsstiftung under Grant 2139. The authors are grateful to Dr. H. Richter of Hermsdorfer Institut fur Technische Keramik (hitk) for applying spin-coating technique to microstructured wafers.

References

[1] H. Miura, M. Terasaka, K. Oki, T. Matsuda, in: New Frontiers in Catalysis (L. Guczi, F. Solymosi, P. Tetenyi, eds.), Elsevier, Amsterdam, 2379 (1993). [2] G. C. Bond, D. T. Thompson Catal. Rev.-Sci. & Eng. 41, 319 (1999). [3] M. Haruta, Catal. Today 36, 153 (1997). [4] P. Claus, A. Bruckner, H. Hofmeister, M. Lucas, C. Mohr, J. Amer. Chem. Soc. 122, 11430(2000). [5] C. Mohr, H. Hofmeister, M. Lucas, P. Claus, Chem. Eng. Technol. 23, 324 (2000).

[6] Handbook of X-ray Photonelectron Spectroscopy (J. Chastain, R. J. King, eds.), Phys. Electr., Inc., 1995. " [7] P. Claus, D. Honicke, T. Zech, Catal. Today, 67, 319 (2001).

102 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

V.A. Brehme, A. Behr Department of Chemical Engineering, University of Dortmund, Germany

Valuable Products from Butadiene, Carbon Dioxide and Hydrogen

1. Introduction

Carbon dioxide is supposed to be the most important gas responsible for the green-house effect causing a global warning [1], The mainpart is produced by combustion of fossil fuels and by cement industry. Additionally carbon dioxide is a waste gas which means it is relatively cheap. Furthermore a possible CCT-tax could prompt the use of carbon dioxide as carbon source in the chemical industry in the future.

Butadiene is produced in large amounts by steam-cracking. Even though the total number of processes based on butadiene is on the increase in the last years, the total number is still inferior compared to those using the monoenes butene or propene.

We focussed our work on the development of new processes based on butadiene, carbon dioxide and hydrogen, which permit the synthesis of multi-functional, industrielly interesting products withonly a few number of reaction steps.

The palladium-catalyzed co-oligomerization of 1,3-butadiene and carbon dioxide is one of the few successful examples of C,C-bond forming reactions of CO2 [2], The selective formation of 2-ethylidene-6--olide (1) in high yields is possible and is investigated in a continuous mini-plant by our group [3], In a consecutive reaction the y-lactone (2) is available [4],

[Pd/PCyJ [Pd/PCyJ

(D (2) Scheme 1. Formation of lactones from butadiene and carbon dioxide

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 103 The 5-lactone (1) itself is an interesting intermediate for the production of branched

C, acids (3), alcohols (4), amides, diols (5) + (6) and esters (see Scheme 2). These products are easily available by hydrogenation of (1) depending on the reaction conditions and the choice of the catalyst.

2-ethylheptanoic acid (3)

5-lactone (1)

2-ethylheptane-1,5-diol (6) 2-ethylheptanol (4)

Scheme 2. Products after hydrogenation of 5-lactone (1)

There are many applications for these products: 2-ethylheptanoic acid (3) can be used e.g. in alkyds, lubricants or stabilizers for pvc. The alcohol, 2-ethylheptanol (4) is a solvent and could play a major role as a plasticizer raw material for the production of di(ethylheptyl)phthalate. The compounds (5) and (6) are interesting monomers for the production of new polyesters, that could be useful! powder coatings for steel plates.

2. Reactions 2.1 Synthesis of 2-ethylheptanoic acid from the S-lactone (1) Heterogeneous hydrogenation In the heterogeneous hydrogenation of the 5-lactone (1) among the desired product 2-ethylheptanoic acid (3) the formation of saturated, isomeric 5-lactones (7) is observed (see scheme 3).

OH + [Pd/C]

Scheme 3. Heterogeneous hydrogenation

104 Variation of the solvent, the concentration of the catalyst, the temperature and the hydrogen pressure have an influence on the selectivity of this reaction. The results of the heterogeneous hydrogenations are summarized in table 1.

Table 1 Results of the heterogeneous hydrogenation of 5-lactone (1)

Products

No. Catalyst Pressure Solvent 5-lactone (1) (3) (7) Pd/C in g (Hz) in bar

1 0,25 10 methanol 0 28 72

2 0,25 10 THF 0 16 84

3 0,25 10 n- 0 17 83

4 0,05 10 THF 0 5 95

(Conditions: m(8-lactone)=10g, V(solvent)=90ml, t=60min, T=60°C, catalyst: Pd/C (5-10%Pd), stirrer velocity=700 rpm) The highest yield of 2-ethylheptanoic acid (3) in heterogeneous hydrogenation is achieved using methanol as solvent. Nevertheless a yield of 28% is rather low. Using non-polar solvents like n-heptane the formation of the isomeric, saturated 5-lactones (7) is enhanced. Selectivities up to 95% are possible. If an isomeric mixture of (7) is hydrogenated under the same conditions no more cleavage of the lactone ring is observed. Therefore a quantitative production of 2-ethylheptanoic acid via heterogeneous catalysis was not achieved.

Homogeneous hydrogenation Applying standard monophasic homogeneous hydrogenation catalysts in this reaction e. g.

Wilkinson’s catalyst ClRh(PPh3)3, the same product distribution as for the heterogeneously catalyzed hydrogenation is observed.

Homogeneous biphasic hydrogenation When a biphasic hydrogenation is carried out in presence of a water soluble in-situ rhodium- TPPTS catalyst an exothermic reaction takes place and 2-ethylheptanoic acid (3) is formed as the main product [9], After a reaction time of 4h the yield of 2-ethylheptanoic acid is about 63

%. The intermediates in this reaction are exclusively isomeric Cg-carboxylic acids (8 ) - (13) differing only in the number and position of the remaining double bonds (see Scheme 3). Neither any saturated nor partially hydrogenated lactones are observed.

105 (11) (12) l13) (2)

Scheme 3. Homogeneous biphasic hydrogenation

Two test series were carried out to recycle the homogeneous water-soluble catalyst. The recycling was realized by phase separation in a separating funnel. Afterwards the catalyst containing aqueous phase was retransfered in the autoclave and reused for thenext reaction.

Recycling experiments under adiabatic conditions The experiments were started at a reaction temperature of 90 °C and the catalyst was used five times. The results are summarized in table 2. Tmax indicates the highest temperature achieved in each experiment. After phase separation the water content as well as the rhodium content of theorganic phase were measured.

Table 2 Recycling of the homogeneous catalyst under adiabatic conditions

No. Tmax Conversion Yield Yield %-Water in Weight Rh in TOP in h'! TON

in °C (7) - (12) (2)in% organic phase organic phase 5-lactone (after 5 in% in% inpg/g min)

1 96 92 92 0 0,78 7,6 3286 623

2 122 100 90 10 0,62 5,6 9931 1300

3 131 100 84 16 0,76 9,2 9858 1978

4 122 100 87 13 0,79 10,4 6042 2655

5 119 100 90 10 1,00 10,4 5551 3332

(Conditions: p = 10 bar H2, stirrer velocity = 1000 rpm, m(S-lactone) = 75 g, m(Rh) = 1000 ppm, t = 30 min, n(P) / n(Rh) =10:1) In the initial run the conversion of the 5-lactone (1) is incomplete. In the first recycling run

thereaction rate increases effecting a complete conversion and even the formation of 10 % of

106 the completely hydrogenated product 2-ethylheptanoic acid (2) is observed. In the second recycle run the reaction temperature exceeds to 130°C causing the catalyst to decay partially. The solubility of rhodium in the organic phase can be related to its water content.

Recycling experiments under isothermal conditions

In a second test series the temperature in the vessel was kept constant at 110°C. Since thefirst recycling run the catalyst performs well and in the following five runs complete conversions are obtained without any tendency of catalyst deactivation. Overall 435g 5-lactone (1) have been converted selectivily with only 38mg rhodium to a mixture of the isomeric, unsaturated

2-ethylidene-heptenoic acids (8 ) - (13).

Heterogeneous hydrogenation of the isomeric mixture (8) - (13)

The homogeneous, biphasic hydrogenation of 5-lactone (1) leads to the corresponding isomeric mixture of the aliphatic carboxylic acids. Although the desired product, 2-ethylheptanoic acid (3), is already obtained in small amounts by homogeneous biphasic hydrogenation, a quantitative yield is only achieved applying a second heterogeneous hydrogenation step. Heterogeneous hydrogenation catalysts show higher activities in double bond hydrogenation compared to the homogeneous two-phase system. For this reason the cleavage of the

8 -lactone ring is done homogeneously in a two phase system. In the following double bond hydrogenation a commercial available heterogeneous palladium/charcoal catalyst was applied. This catalyst is very active due to its high surface area and in addition it is resistent to carboxylic acids. The heterogeneous hydrogenation was investigated in methanol and n-heptane. Methanol is a well known solvent for hydrogenations; n-heptane was chosen because of its complete miscibility gap with water. It could therefore be used as solvent in the two phase hydrogenation. The results are shown in figure 1.

107 Methanol 60°C

Methanol 80°C

4 Heptane 60°C

. Heptane 80°C

—O- Methanol 60CC without removal of water

time t in min

(Conditions: p(H2) = 10 bar, stirrer velocity = 700 rpm. m(ethylheptanoic acids) = 10 g, m(Pd/C) = 0,25g)

Fig. 1. Comparison of the heterogeneous hydrogenation of the isomeric 2-ethyIidene- heptenoic acids in different solvents Quantitative yields of 2-ethylheptanoic acids are achieved in the case of methanol after 3-5 minutes, in the case of n-heptane as solvent after 30 minutes. The reaction rates of the hydrogenations in methanol compared to n-heptane are obviously much faster. One reason can be seen in the higher solubility of the ethylidene-heptenoic acids in methanol. Before the hydrogenations the reaction mixtures were distilled to remove traces of water which are dissolved in the organic phase due to the previous two phase reaction. In contrast to the water free hydrogenation the presence of only 1% of water in the reaction mixture diminishes the reaction rate considerably showing the inhibition of the catalyst by the water content.

2.2 Synthesis of alcohols and diols from acids or lactones

Reduction of carboxylic acids toalcohols Hydrogenation of fatty acids and its derivatives to fatty alcohols is an industrially important process. Unfortunately the reaction takes place only at severe conditions (300 bar, 250°C) with heterogeneous copper chromite or copper zinc oxide catalysts affording expensive high pressure reactors. We focussed our work on homogeneous systems which show higher activity at milder reaction conditions allowing the synthesis of more sensitive products.

Different bimetallic homogeneous catalysts consisting of a Group 8 or 9 late transition-metal and a second Group 6 or 7 transition-metal carbonyl showed a synergistic effect [5] allowing the conversion in good yields under moderate conditions. An equimolar mixture of

108 Rh(acac)(CO)2 and Mo(CO )6 showed the highest activity and was therefore applied to the reduction of carboxylic acids to alcohols and of lactones to diols. In detail the reduction of 2-ethylheptanoic acid (3) to 2-ethylheptanol (4) was studied (see scheme 4) [10].

IRh/Mo]

(3) Scheme 4 Reduction from 2-ethylheptanoic acid to 2-ethylheptanol 2-etyhlheptanol (3) is an interesting softener alcohol for the production of di(2-ethylheptyl) phthalate which could be used as plasticizer [6], Because the commercially available product di(2-ethylhexyl) phthalate (DEHP) is found to be one of the more toxic phthalates [7] the interest in alternative compounds is increased. Compared to DEHP, di(isononyl) phthalate

(DINP) poses no risk to human reproduction or development [8 ]. Assumedly di(2-ethylheptyl) phthalate is in thesame way harmless due to the same number of carbon atoms.

Bimetallic catalysts consisting of Rh(acac)(CO)2 and Mo(CO )6 were applied in this reaction giving nearly quantitative yields of 2-ethylheptanol in four hours at 190°C and 150 bar hydrogen pressure withdioxane as solvent.

Reduction of lactones to diols Catalysts which are capable to reduce carboxylic acids to alcohols are also active in the reduction of lactones. Respectively the products are diols. This reaction is applied to the most

active bimetallic catalyst system, a mixture of Rh(acac)CO)2 and Mo(CO)« [11]. The reactants are 2-ethylheptane-5-olid (7) and 2-ethylheptane-4-olid (15) which are produced by

hydrogenation of thecorresponding unsaturated lactones (1) or (2).

Scheme 5 Reduction from 2-ethylheptane-5-olid to 2-ethyI-l,5-heptanediol

The highest yield of (6) (97%) is formed at 190°C and 150 bar hydrogen pressure after 2h. Higher temperatures and longer reaction times lead to product decomposition and hydrocarbon formation.

The reduction of 2-ethylheptane-4-olid (15) to 2-ethyl-1,4-heptanediol (5) (see scheme 6) is performed analoguous to the reduction of the six-membered ring lactone.

109 Scheme 6 Reduction from 2-ethylheptane-4-olid to 2-ethyl-l,4-heptanediol Product is 2-ethylheptane-l,4-diol (5). The yields in this reaction are rather low. At longer reaction times the conversion increases causing the formation of by-products like alcohols and hydrocarbons. An increase in product formation is hardly achieved. The reason is a higher thermodynamical stability of five membered lactone rings compared to six membered rings. Hence, more thermal energy is necessary to open the lactone ring, which causes simultaneously decomposition of the product. Additionally the commercially available lactones y-butyrolactone, 8-valerolactone and e-caprolactone are converted with theRh/Mo catalyst, giving thecorresponding diols.

3. Conclusion

Based on the three simple, readily available reactants butadiene, carbon dioxide and hydrogen new catalytic processes were developed. The advantages are the few number of reaction steps and thehigh selectivities which are achieved. The great number of different functional groups of the 8-lactone (1) will allow further reactions, which makes the molecule itself to an interesting intermediate in researchand industry. 4. Literature

[1] Daten zur Umwelt, Umweltbundesamt, Germany, 1997. [2] Y. Inoue, Y. Sasaki, H. Hashimoto, Bull. Chem. Soc. Jpn., 51 (1978) 2375. [3] A. Behr,M. Heite, Chem. Ing. Tech. 72 (2000) 58. [4] A. Behr,K. D. Juszak, J. Organomet. Chem., 255 (1983) 263. [5] D. He, N. Wakasa, T. Fuchikami, Tetrahedron Lett. 36 (1995) 1059. [6] A. Musco, R. Santi, G. P. Chiusoli, Get. Patent Application, No. 28 38 610, 5.9.1978. [7] B. Hileman, Chem. Eng. News, August, 7th (2000) 52. [8] B. Hileman, Chem. Eng. News, January, 3th (2000) 8. [9] A. Behr, V. Brehme, Ger. Patent Application, Appl. No. DE 10041571.7, submitted. [10] A. Behr, V. Brehme, Ger. Patent Application, Appl. No. DE 10124390.1 , submitted. [11] A. Behr, V. Brehme, Ger. Patent Application, Appl. No. DE 10124389.8, submitted.

110 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

K. Harth BASF AG, Ludwigshafen, Germany

Making Olefins from Light Paraffins by Catalytic Dehydrogenation and Oxidative Dehydrogenation

Light olefins such as ethylene, propylene or butenes are basic educts for the added value chains of the chemical industry and belong to the top ranked chemicals by production. Therefore, the development of efficient production methods for these olefins is of vital economic and scientific importance.

Until now, light olefins have been largely produced either by non-catalytic steam cracking of suitable hydrocarbon feedstocks such as naphtha, LPG or ethane or by catalytic cracking of heavier hydrocarbons from the vacuum distillation of refineries. These production methods are very capital intensive and in most cases imply the coproduction of a whole range of olefins and other products. On the other hand, catalytic dehydrogenation (oxidative or non-oxidative) of light alkanes in principle provides the opportunity to access the corresponding olefins with high selectivity and to utilize well abundant raw materials of natural gas reserves.

In case of C3- and C4-olefins, non-oxidative catalytic dehydrogenation methods made their way into industrial applications. Several different processes with specific catalysts and process conditions have been technically proven in production or pilot scale. However, these technologies still face a series of fundamental obstacles, like

- low alkane conversion due to thermodynamic limitations - quick catalyst deactivation with the necessity of complex regeneration procedures - expensive process technology for high-temperature dehydrogenation, for catalyst regeneration and for downstream separation of olefins, alkanes and by-products.

111 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 Therefore, high capital costs still narrow a broad application of alkane dehydrogenation methods in the chemical industry. Highly active and selective catalysts with low deactivation rates and extraordinary thermal and mechanical stability will be the key factors for improving the process performance and enabling wider applications.

The lecture will provide an overview on the current status of alkane dehydrogenation technology and highlight some of the most promising future developments.

112 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Friedrich Umweltbundesamt Berlin, Germany

The MTBE Issue from the Viewpoint of an Environmental Protection Agency

There is no manuscript available

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 113 114 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

St. Muller, A. Gammersbach, H.-J. Kramer, F. Kaledat EC Erdolchemie GmbH, Koln, Germany

Creating Value from Isobutene

INTRODUCTION EC Erdolchemie GmbH was founded as a petrochemical joint venture between Bayer AG and Deutsche BP AG in 1957. Since May 1, 2001, EC is a 100 percent subsidiary of Deutsche BP AG and is therefore part of the BP Group. The products of EC Erdolchemie with the corresponding annual production are shown in Figure 1.

3 Mio to Naphtha 240 Mio Nm3

Cracker Natural gas

540 kt 250 kt 280 kt 200 kt 250 kt

Nitric Acid 150 kt 355 kt

Figure 1. Product Spectrum of Erdolchemie

EC Erdolchemie GmbH (4,5 Mio tons annual capacity) is located in Cologne, Germany and accounts for approximately 18 percent of BP's total worldwide petrochemical capacity.

115 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 SUMMARY The oligomerisation unit is one of Erdolchemie's production plants. Raffinate-1, which consists mainly of Isobutene (approx. 45%) and linear Butenes (approx. 40%), is used as feedstock for the oligomerisation of Isobutene. Raffinate-1 is obtained after extracting Butadiene from the crude C-4 stream of the EC steam crackers.

The olefinic products are Diisobutylene (C-8 ), Codibutylene (C-8 ), Triisobutylene (C- 12), Tetraisobutylene (C-16) and Isoeicosene (C-20). Depending on the market situation and plant capacity these olefins are hydrogenated to their corresponding

saturated hydrocarbons (C-8 , C-12, C-16 and C-20).

Naphtha

r»H2 * C8 Steam- C2 Hydro- » C12 cracker >C3 genation ■*■ C16 U- C5+ > C20

C4

► Raff. -2 Butadiene Raff.-1 Hydro- Oligo- » C8= Plant isomerisation merisation ► C12= ► C16=

i *■ C20= Butadiene

Figure 2. Current EC Oligomerisation Process

The core element of this oligomerisation process, established in 1961, is the usage of slurry reactors with an acid ion exchange resin as a catalyst. The used catalyst is separated from the product mixture by centrifuges. A mixture of fresh and used catalyst is constantly added to the oligomerisation reactors. This process leads to a broad product spectrum, in which the ratio of the various components can only be influenced within certain ranges. Among the numerous applications are plasticizers, tackifiers, aerosols and solvents.

116 Over the last 40 years the EC oligomerisation plant has steadily been expanded and improved. For example, hydrogenated C-16 (Isohexadecane) and C-20 (Isoeicosane) have been sold to the market since the mid 80s and a DIB (Diisobutylene) with high purity (> 95%) was launched in the mid 90s.

A further major improvement step that EC R&D has started to develop is a new oligomerisation concept. The main elements of this new concept are: 1. Continuous hydroisomerisation of Butene-1 to Butene-2 2. Separation of Isobutene from other butenes by distillation 3. Selective Dimerisation of Isobutene by reactive distillation

4. Optional oligomerisation of C-4 and / or C-8 to C-12 or C-16 in finishing reactors

Naphtha Oligomers Oligomerisation Isolation Unit Unit Steam- to Raff. -2 cracker Isobutene Crude-04 Butanes

Butadiene Raff.-1 Isom. / Purfication Hydrogenation Plant Unit

Butadiene to Raff. -2 Olefines Paraffines

Figure 3. Concept of a new Isobutene Process at EC

The advantages of the reactive distillation are high yields due to the removal of the products from the equilibrium, easy temperature control and longer catalyst lifetime. The main characteristic of this new concept compared to the current process is the high flexibility in terms of the product spectrum. Depending on the reaction conditions and set-up the process can vary the percentage of C-8 , C-12 or C-16 to meet changes in demand.

117 Further characteristics are the high purity of the DIB obtained after distillation of the products (e g. very low Codibutylene (Codi) formation), easy catalyst replacement (no continuous adding of fresh catalyst) and higher quality of Raffinate-2 because of the increased Butene-2 content. One possible application of the new process is to selectively produce Isooctane, which can be offered as a sulfur- and oxygene-free octane booster and could compete with MTBE.

Isobutene Butanes^

Raff. -2

Oligomers Raff.-1 Isooctane Raff.-2

Isom. / Purification Unit Oligom. / Isolation Unit Hydrogenation

Figure 4. Production of Isooctane by new Isobutene process

Further investigations and more detailed studies of this new oligomerisation concept are carried out to better understand the complete range of technical and commercial possibilities of this new process.

118 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

J. Kolena, P. Moravek, J. Lederer Unipetrol Central Research, Litvinov, Czech Republik

C4 Fraction - A Raw Material for the Production of Cm Plasticizer Alcohols

Introduction Steam cracking of hydrocarbons, though aimed at ethylene and propylene manufacture, produces also high quantities of C4 fraction as a byproduct besides other heavier fractions. A typical yield of C4 fraction ranges between 2,5 and 11 wt.% in relation to ethylene, depending on raw material and the pyrolysis severity. In average, the ratio of C4 fraction to ethylene represents about 4% 1. It has also been estimated, that world production of C4 fraction was as high as 21 mil. t in 1998 1. More then 80 % of C4 fraction from steam crackers, as well as a significant part of similar fraction from catalytic cracking units, is most frequently utilized as a raw material for butadiene extraction and subsequent MTBE production. The remainder, so called Raffinate 2, contains predominantly butenes and a smaller amount of butane. More complex system of C4 fraction utilization should therefore contain also Rafinate 2 processing.

Representing about 30 % of the amount of crude C4 fraction, Rafinate 2 is predominantly a mixture of 1-butene and 2-butene. It is a good material for subsequent chemical utilization to more valuable products. A typical composition (main components) of Rafinate 2 is presented in table I. Table I: Typical composition of crude C4 fraction and Rafinate 2.

Compound Crude C4 fr actio Rafinate 2 methane 0,1

C3 (and lighter) 0,3 n-butane 2,3 15 1,4 2,2 1 -butene 19,4 50,5 2-butene 8,3 30,4 isobutene 22,2 1,3- butadiene 46 0,2 ethyl- + vinyl- 0,9-1,3 acetylene 1,5

Rafinate 2 can be utilized as a source of additional propylene, produced via metathesis or catalytic cracking. Another variant can be isomerization of butenes to

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 119 isobutene, opening the way to diversity of downstream products (MTBE, TBA etc.). Las, but not least, Rafinate 2 can be utilized as a raw material for the productions of plasticizer alcohols. With the price usually close to LPG value, Rafinate 2 can be a successful competitor to propylene as the raw material for the alcohols production, even if lower content of active components is taken into account. A cost efficient technology is the most important precondition.

Plasticizer alcohols from C4 olefins Aliphatic alcohols with 7 to 13 carbon atoms rank among a group called “Plasticizer alcohols ”. Straight chain butenes can be converted into Cg and C10 alcohols, which are the raw materials for high quality plasticizers. The most common plasticizers are esters of said higher alcohols with phthalic acid. Though 2- ethylhexanol with its almost 70 % share in the market clearly represents the leading product in this area, Cg and C10 alcohols are considered more promising intermediates in the plasticizers manufacture in terms of future development. More dynamic grows of these higher alcohols is expected in foreseeing future. It is also expected that the consumption of these alcohols will be less affected by the environmental regulations imposed on phtalates. Some basic data about the worldwide production and consumption of main plasticizer alcohols are listed in table II 3 Table II: Production capacities and consumption of main plasticizer alcohols in recent years

Year 1997 2000

2-ethylhexanol capacity [mil. t/y] 2.9 3.7

2-ethylhexanol consumption [mil. t/y] 2.33 2.48

Share of 2-ethylhexanol on the market [%] approx. 68 approx. 66 Isononanol capacity [mil. t/y] 0.43 0.715

Mixtures of Cg alcohols are usually called isononanols, bearing also different trade names depending on the producer. They are produced by butenes (Rafinate 2) dimerization to a mixture of , which are hydroformylated to corresponding Cg aldehydes, the last being hydrogenated to alcohols. The mixture of octenes contains typically (Dimersol process) 7 wt. % of linear octenes, 58 wt. % of methylheptenes and 35 wt. % of dimethylhexenes. It is clear that isononanols that are produced by hydroformylation of this mixture and hydrogenation of resulting aldehydes contain diversity of isomeric Cg alcohols. This technology of Cg alcohols manufacture is rather complex, requiring considerable investment, mainly because of demanding dimerization step. The price of the product is therefore remarkably higher then this of 2-ethylhexanoi. In spite of this, isononanols are relatively widespread products, most common in the range of platicizer alcohols higher then 2-ethylhexanol. New capacities are under construction, mainly in East Asia. Decanols and isodecanols are the products for special purposes. Their production is not based on C4 olefins as raw materials. They are manufactured by hydroformylation of which are produced by trimerization of propylene. Both these products are the plasticizer intermediates with low share on the market.

120 Cio alcohols can also be produced similar way like 2- ethylhexanol, consisting in hydroformylation of mixed butenes to C5 aldehydes, which are converted to C10 alcohols by aldol condensation and hydrogenation (see Fig. 1).

2-propyl heptanol is the desirable final product, so it is very important to maximize the yield of 1-pentanal - a precursor of 2-propylheptanol - in the hydroformylation step or separate its branched isomer 2-methylbutanal before aldol condensation. 2-methylbutanal reacts a very complex way at the aldol condensation conditions, yielding several highly branched aldols or even glycols and their esters. This generally results in the final product quality decrease. As Rafinate 2 contains comparable amounts of 1-butene and 2- butene respectively, it is necessary to use a catalyst also accelerating double bond shift at hydroformylation conditions to convert 2-butene into linear aldehyde. It is well known from the literature 2 that the activities of common hydroformylation catalysts towards the double bond shift reaction decreases in following order:

Phosphine modifiedCo > unmodified Co carbonyl > modified Rh carbonyl.

ch3ch2ch =ch2 ch3(ch2)3 ch = o (1) > +CO + H ch3ch = ch2ch3 ■ ch3ch2ch ch3 (2) CH = O

2 CH3(CH2)3 CH = O ------► CH3(CH2)2CH = CHCHCH = 0 (3)

ch2ch2ch3

ch3(ch2)2ch = chchch = o + h2------► CH3(CH2)4CHCH2OH (4)

ch2ch2ch3 ch2ch2ch3

Figure 1: Preparation of 2-propylheptanol from mixed butenes - reaction scheme.

The activity of the rhodium catalyst, modified by triphenylphosphine, which is the most usual type, widely used in propylene hydroformylation, is close to zero in relation to double bond shift. For this reason it cannot be utilized in the mixed butenes fraction hydroformylation. Several years ago, Union Carbide launched a new technology named UNOXOL 10, aimed at producing 2-propylheptanol from Rafinate 2. A new type of phosphite modified Rh based catalyst has been developed for this process. The phosphite ligand used in this catalyst features complex structure (Fig 2) giving the catalyst excellent stereoselectivity. The rhodium catalyst modified by this ligand seems to be the most effective in the internal olefins hydroformylation giving the highest yield of linear aldehyde.

121 Figure 2: A structure of a complex phosphite ligand giving the Rh based catalyst high selectivity towards linear aldehyde in the hydroformylation of internal olefins.

The selectivity of this catalyst is high, so the crude hydroformylation product can be directly aldolized without separation of isomers. A mixture of C10 alcohols, containing about 87 wt. % of 2-propylheptanol is thus obtained. 2-propylheptanol is relatively new product on the plasticizers market. First commercial unit has been built up in Yeochun (South Korea) by LG Chemicals, based on Union Carbide licence . 2-propylheptanol seems to be quite promising plasticizers intermediate, featuring many advantages over 2-ethyihexanol. Its future position on the marked will depend first of all on the price. It’s been discovered long ago that also the classical unmodified cobalt catalyst, utilized in the high pressure hydroformylation process, catalyzes the double bond shift reaction in some extend 2. Unlike rhodium catalyst, this catalytic system is relatively cheep and the high pressure technology has had long industrial tradition, being utilized in propylene hydroformylation for many years. As the old propylene hydroformylation units are gradually closed down or reconstructed into the rhodium based technology, the idea of possible utilization of the high pressure cobalt based process for hydroformylation of mixed butenes attracts interest.

Kinetics of 1-butene and 2-butene hydroformylation, catalyzed by the classical Co carbonyl catalyst To assess the suitability of the high-pressure process, utilizing unmodified Co carbonyl catalyst for the hydroformylation of 1- and 2- butene, an experimental study has been performed in laboratory scale. A mixed, temperature controlled, batch stainless steel autoclave was used in the experiments. The reaction conditions were as follows: overall pressure 26 Mpa, CO:H2 molar ratio 1:1, temperature in the range of 120 to 140 °C, Co concentration approx. 2 g/l. Ethylbenzene was used as solvent. The catalyst — Co carbonyl - was prepared in situ by the reaction of Co 2- ethylhexanoate with synthesis gas (an equimolar mixture of CO and Hs) at hydroformylation conditions. Then olefin was added to start up hydroformylation. It is well known from the literature that hydroformylation is the first order reaction with regard to olefin and, under constant cobalt concentration and constant CO:H2

ratio, can be described by following equation: r = - dC0ief/ d t = k * c0ier. Our

122 experimental results proved validity of the first order equation. The rate constant of the first order reaction was taken as the reactivity criterion. The constants of both parallel reactions (1) and(2) (see Fig. 1) were evaluated separately. The reaction constants of 1 -butene hydroformylation in ethylbenzene as solvent under different temperatures are summarized in table III. Figure 3 presents responding Arrhenius plot. Table III: Reaction rate constants for 1-butene hydroformylation at the overall pressure of 26.0 Mpa, in ethylbenzene as solvent and Co carbonyl as catalyst

TjK] 393.15 403 15 413,15 Co [g/l] 1,99 1,985 2,03 1/T 0,002544 0,00248 0,00242

k1 0,0055 0,013 0,031

k2 0,0013 0,0032 0,009

k1/k2 4,23 4,06 3,44

As apparent from the data above (Table III., Figure 3), the high pressure hydroformylation of 1 -butene with unmodified cobalt catalyst yields predominantly 1- pentanal, the ratio of which towards its branched isomer slightly decreases with increasing temperature. At common conditions, the ratio around 4 can be expected.

Figure 3: Arrhenius plot (ln(k) vs. reverse temperature) for the reaction rate constants of both parallel reactions. Hydroformylation of 1-butene at 26.0 Mpa, ethylbenzene as solvent, Co concentration 2 g/l.

123 Similar data as these for 1-butene have been measured also for 2-butene hydroformylation. The reaction constants of 2-butene hydroformylation in ethylbenzene as solvent under different temperatures are summarized in table IV. The Arrhenius plot is presented by figure 4.

Table IV: Reaction rate constants for 2-butene hydroformylation at the overall pressure of 26.0 Mpa, in ethylbenzene as solvent and Co carbonyl as catalyst.

T[Kl 393,15 403.15 413,15 Co [g/ll 1,985 1,985 2,0208 1/T 0,002544 0,00248 0,00242

k1 0,0015 0,0047 0,009

k2 0,00051 0,0016 0,003

k1/k2 2,94 2,94 3.0

Figure 4: Arrhenius plot (ln(k) vs. reverse temperature) for the reaction rate constants of both parallel reactions. Hydroformylation of 2-butene at 26.0 Mpa, ethylbenzene as solvent, Co concentration 2 g/l.

Hydroformylation of 2-butene also gives predominantly 1-pentanal, its ratio to 2- methyl-1 -butanal being around 3. This ratio can be hardly controlled by temperature, as the activation energies of both parallel reactions are almost identical in the case of 2-butene. Relatively high selectivity to linear aldehyde is caused by double bond shift reaction, catalyzed by cobalt hydrocarbonyls, which are the catalyticaly active components.

124 Some experiments of propylene hydroformylation in the same solvent (ethylbenzene), at the same conditions, have been performed to compare butenes reactivity with propylene. These data are useful for the assessment of possibility to process mixed butenes in the hydroformylation reactors originally designed for propylene. Reaction rate constants at 130 “C for all olefins tested, as well as the activation energies of both parallel reactions (1) and (2), are summarized in table V. Table V: Reaction rate constants and activation energies of butenes and propylene hydroformylation at 130 °C and 26.0 Mpa. Ethylbenzene used as solvent, Co concentration 2 g/l.

Aiken 1-butene 2-butene propyien e ki 0,0130 0,0047 0,0092

k2 0,0032 0,0016 0,0023

ki/k2 4,06 2,94 4,00 Ei [ kJ/mol] 116,7 121,2 116,5

E2 [ kJ/mol] 125,5 114.8 139,2

The hydroformylation rates of 1-butene and propylene are almost the same, so are the selectivity values towards linear aldehydes. The reaction of 2-butene is somewhat slower, the selectivity towards linear aldehyde being also lower. Nevertheless, 1-pentanal is the main product when hydroformylating mixed butenes. Its ratio to 2-methylbutanal around 3,5 at maximum can be expected if a typical industrial mixture like Rafinate 2 is processed. This ratio is not very temperature sensitive.

Possibilities of producing 2-propylheptanol rich C10 plasticizer alcohols by the classical oxo synthesis process As discussed above, the hydroformylation rates of butenes and propylene are comparable. Same reactor and cooling systems can be used for both raw materials processing. Butenes yield in higher boiling products; therefore the solvents must be selected with this respect. Some solvents, e.g. toluene, frequently used in the propylene hydroformylation process, cannot be used because of proximity of boiling points with Cs aldehydes. Boiling points of the aldehydes derived from butenes or propylene respectively are compared in table VI. Table VI: Normal boiling points of C4 and C5 aldehydes - the products of propylene and butenes hydroformylation.

Starting olefin j Product I The product boiling ;point [ C] Propylene 2-methylpropanal 64 Propylene butanal 75,7 Butene 2-methylbutanal 92 Butene pentanal 103

125 The differences between boiling points of Cs and C< isomeric aldehydes are similar. If a suitable higher boiling solvent is chosen, the primary product distillation can be done by means of the same equipment in cases of butenes and propylene hydroformylation. No substantial differences can be expected in subsequent operations i.e. aldol condensation, hydrogenation and distillation. Revamping of the high-pressure oxo synthesis process to the mixed butenes feed processing seems to be viable.

Conclusion

The remainder from the standard C4 fraction processing, so-called Rafinate 2, is a promising raw material for the production of higher plasticizer alcohols. Besides the production of traditional Cg alcohols, i.e. nonanols and isononanols via dimerization and hydroformylation, Rafinate 2 can be utilized in the manufacture of 2- propylheptanol or 2-propylheptanol rich plasticizer alcohol mixtures. The key operation in this manufacture is the hydroformylation step. The most selective technology of mixed butenes hydroformylation utilizes the phosphite modified rhodium catalyst (UNOXOL10 - Union Carbide). The high-pressure hydroformylation process, based on the cobalt carbonyl catalyst, can also be used for this purpose quite effectively. Nevertheless, its selectivity to straight chain aldehyde, which is the precursor of 2-propylheptanol, can hardly exceed 70 %. Co-production of 2- methylbutanal or its derivatives is therefore inevitable. If there is a market for 2- methylbutanal derivatives, adapting older units, originally designed for propylene hydroformylation, is a realistic alternative.

References 1. Eur. Chem News, Sep. 7-14, (1998). 2. Falbe J.: Carbon Monoxide in Organic Synthesis, New York (1970). 3. Eur. Chem News 69, 1813,11 (1998). 4. Hydrocarbon Process. 73, (9), 36 (1994).

126 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

Th. Haas*, W. Hofen*, G. Thiele*, P. Kampeis** *Degussa AG, Hanau, Germany, **Krupp Uhde GmbH, Dortmund, Germany

New PO Processes

Nearly 5 million metric tons of propylene oxide are used world wide. It is mainly used to produce polyether polyols, which are applied in polyurethane foams. The second significant outlet is the production of propylene glycol, which is applied mainly in unsatured polyester resins. The world growth rate of propylene oxide is about 5 % per year. The world scale plant capacity is about 250.000 metric tons a year. Therefore every year a new PO plant has to be build to compensate for the market demand.

Table 1: Todays PO Technologies

R.iw mute ri. Is Co-products i Chlniuhyilrm , propylene, chlorine CaCI2(2.1 t/t PO) process I lime + caustic soda from chlorine production Cliloruhydrin propylene, chlorine, NaCI =▻ Integrated process by Dow caustic soda electrolysisplant . _ PO IBA propylene, t-butanol (2.4 t/t PO) process isobutane, oxygen =s> MTBE PO/SM propylene, ethyl ­ methylphenylcarbinol proruoo benzene, oxygen =*> styrene (2.3 t/t PO)

Until the seventies, the chlorohydrin route to propylene oxide was dominant. Today the PO production is dominated by organic hydroperoxide based routes. Oxidants are either, tertiary butyl hydroperoxide or ethylbenzene hydroperoxide. Both organic peroxide processes avoid any chlorine involvement. On the other hand they produce coproducts. These are tertiary butyl alcohol,methyl tertiary-butyl ether or styrene. Latest processes going on stream are all using the styrene co-product route, the POSM process.

127 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 Table 2: New PO Technologies

stage it murks Hyilrngi n low invest, high PE- yield Pl-l.iXIII.- pilot dependent on H202 price lab very early stage W vlllltl 1 |

Cumene 1 pilot high invest | prod 2002 high PE-yield

Pirui t oxidation lai' low PE-conversion low yield

All new PO-technologies focus on co-product free routes. In that case producers would be more independent of the styrene monomer and MTBE market. In case of MTBE there are also uncertainties concerning its use as gasoline additive. All new PO-technologies under development use an oxygen route.

A co-product free propylene oxide route is the Cumene process. It is very similar to the POSM process. The main difference is the recycle of the organic peroxide forming product. This recycle step is an hydrogenolysis reaction of a tertiariy alcohol.

0—OH

0—OH

OH +

Fig. 2: Reaction Scheme of the Cumene based Propylene Epoxidation.

128 Recently a second process has been published using a tungsten cluster catalyst to epoxidise propylene with hydrogen peroxide. The reaction takes place under homogeneous conditions whereas after the reaction the catalyst precipitates, can be filtered off and easily recycled. This process is not yet developed and technically proven.

The third important new technology is the direct oxidation of propylene to propylene oxide using oxygen. This reaction still suffers from very low propylene conversions and the formation of high amount of by products, like acrolein.

The fourth and most important new PO process is the conversion of propylene and hydrogen peroxide using an Titanium-silicate catalyst (1). This process has been invented by Enichem in the eighties. Many companies have further developed this process in the meantime. These companies are for example Lyondel, BASF, Enichem, Bayer, Degussa, etc.

The oxidation of propylene with hydrogen peroxide using a titanium-silicate catalyst takes only place selectively in an organic solvent. Preferably alcohols are used as such solvents. The excess of solvent has to be high to obtain high activities and selectivities. Therefore solvent recycle is essential for any continues process. The TS-1 catalyst shows very strong deactivation. Therefore overcoming this deactivation is the crucial point to obtain an feasible fixbed process. Another route is to run the process in a slurry mode including an integrated regeneration.

Several side reactions take place during this epoxidation, most are consecutive reactions. The most important one is the nucleophilic addition of methanol and water to propylene oxide.

main reaction MeOH

MeOH side reactions

+ isomers

Fig. 1: Reaction Scheme of the TS-1 catalysed Propylene Epoxidation.

129 In the consecutive reaction these by-products are further converted for instance to formaldehyde and acetic aldehyde. These aldehyds form hemi acetals and acetals down stream the process. Propylene oxide has to be purified from all these compounds.

The advances the two technically developed routes, the TS-1 route and the Cumene route are higher propylene oxide selctivites with respect to propylene, beside avoiding co-products.

1. Clerici, M.G., Bellusi, G., Romano, U., J. Catal. 129, 159 (1991).

130 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Bruckner, P. Rybarczyk, J. Radnik, G.-U. Wolf, H. Kosslick, M. Baerns Institut fur Angewandte Chemie Berlin Adlershof e.V., Berlin, Germany

Oxidative Dehydrogenation (ODH) of Propane over Vanadia-Based Catalysts: Probing Active Sites under Working Conditions

Introduction Vanadia-based catalysts are widely used in a number of industrial oxidation processes. Particularly, oxidative dehydrogenation (ODH) of light alkanes to the corresponding alkenes would be an attractive subject since cheap and environmentally friendly starting materials can be converted to valuable olefins used as feedstock for other processes. The ODH of propane has been extensively studied since the 1970 ’s and discussed in several review papers [1-3]. Unfortunately, maximum propene yields obtained so far hardly exceed 20 % since total com ­ bustion of both propane and propene leads to low selectivities, in particular at higher degrees of conversion. Due to these limitations, the ODH of propane is still far from being attractive for industrial application and a major goal of research is to develop highly selective catalysts for this process. In this sense, a better understanding of the nature of the active sites and their role in the catalytic cycle is needed. However, the current knowledge on these subjects is still incomplete and frequently contradictory. It is generally agreed that the ODH of propane over vanadia-based catalysts proceeds via a Mars-van Krevelen redox cycle with VOx species being the active sites. However, it is still uncertain whether sites of high catalytic performance should be tetrahedrally or higher coordinated, isolated, connected in two-dimensional surface layers by V-O-V bridges or even

form bulk vanadate phases. In crystalline MgsfVO^ and Mg 2V207, V043" units separated by

the Mg ions were found to be more selective than V2Oy4" dimers that contain easily removable

oxygen in V-O-V bridges [4, 5] while the opposite has also been claimed [6]. For vanadia supported on different oxides various oxygen species have been claimed to participate in the catalytic cycle; those in V-O-support metal bridges of isolated VOx species [7], those in

V-O-V bridges of low polymerized VOx units [8 ] and terminal V=0 bonds as well [9]. The discrepancies in the results obtained by different authors from investigations of similar catalysts might arise from different preparation procedures and conditions of catalytic tests, in particular, different oxygen partial pressures of the feed. Moreover, most of these conclusions were derived by characterizing the catalysts only before and after usage in the catalytic process under ambient conditions; an approach by which sufficient information on how a catalyst works during reaction is hardly obtained. In thiswork we have studied the be­ haviour of active VOx sites in unsupported VMgO materials and supported on mesoporous

Al2Os and Si02 by different spectroscopic in situ- and quasi-in .sv'/u-techniques such as EPR, UV-vis-DRS and XPS. For the first time, a simultaneous coupling of in .sz/u-EPR/on line- GC/UV-vis-DRS has been set up by implementing a UV-vis fibre optic sensor directly into the EPR flow reactor. Based on this knowledge and additional results from ex situ-

131 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 characterization (XRD,5IV-NMR, potentiometric titration, FTIR) some general structure- reactivity relationships for vanadia-based catalysts in the ODH of propane are derived.

Experimental Unsupported VMgO catalysts with different Mg/V atomic ratios (denoted as MggVi,

Mg 4Vi, MgiVi and MgiV 4) were obtained by adding freshly prepared Mg(OH )2 powder in portions to an aqueous solution of NH4VO3 and 1 wt.-% of NH3. After evaporation in a rotary evaporator, theresulting powder was dried at 110 °C for 18 h and calcined in air at 600 °C for 3 h [10]. Supported catalysts with a vanadium loading of 2.8 wt.-% were prepared by impreg ­ nating mesoporous AI2O3 [11], MCM-48, and SBA materials of different pore diameters with aqueous NH4VO3 followed by 20 h drying at 130 °C and 3 h calcination at 600 °C in air. EPR spectra were recorded by the c.w. spectrometer ELEXSYS 500-10/12 (Broker) in X-band. For in .sz/u-studies a homemade flow reactor equipped with a temperature program ­ mer and connected to a gas dosing apparatus was used [12]. For on-line product analysis the reactor outlet was connected to a GC 17AAF capillary gas chromatograph (Shimadzu) equipped with a 30 m x 0.32 mm Silicaplot column (Chrompack) and a FID. UV-vis-DRS measurements were performed using a Cary 400 UV- vis spectrometer (Varian) equipped with a diffuse reflectance accessory (praying mantis, Harrick) and a heat­ educt gas" able reaction chamber for in situ- mixture studies. To reduce light absorption, the PC plug-in spectrometer catalysts were diluted with a-Al203 EPR cavity (calcined at 1473 K for 4h). For simul­ taneous EPR/on line-GC/UV-vis cou ­ pling, a fibre optic quartz sensor (Op- catalyst bed ■ quartz wool tran WF, 200 x 1.5 mm) was directly implemented in the EPR flow reactor through a Teflon gasket (Fig. 1). The sensor is connected to an AVS-PC- products connection to 2000 plug-in spectrometer (Avantes) on Fine-GC by fibre optic cables (2000 x 0.4 mm). thermocouple XPS spectra were recorded using a VG ESCALAB 220 iXL spectrometer (VG Figure 1. Experimental set-up for simultaneous Instruments) at room temperature with in sz'to-EPR/on line-GC/UV-vis coupling a monochromated A1 Ka-source. Charging effects were reduced by a flood gun. The spectra were corrected with respect to the Cls signal at 284.5 eV. Signal intensities were normalized using the sensitivity factors of Scofield [13]. Quasi-in situ spectra were recorded after treating thecatalysts under ODH con ­ ditions using a reaction cell installed in the lock to the analysis chamber. After cooling, the samples were transferred into the spectrometer without contact to ambient atmosphere. XRD powder patterns were recorded using a STADI P transmission diffractometer (STOE) using Cu Ka radiation at room temperature.

132 Acidic surface sites were determined by pyridine adsorption using a FTTR spectrome ­ ter (Broker IFS 66) equipped witha heatable adsorption cell. Self-supporting wafers were pre­ treated in vacuum at 400 °C. FTIR spectra were recorded after pyridine adsorption at room temperature and subsequent evacuation.

BET surface areas were measured by N2 adsorption at -197 °C using a Gemini III 2375 surface area analyzer (Micromeritics). The mean vanadium valence state was deter ­ mined by potentiometric titration using a variant of the method developed by Niwa and Mu­ rakami [14]. Catalytic tests were performed in a fixed-bed U-type quartz reactor at 500 °C (educt mixture: 40% C3H8 , 20% 02, balance N2; W/F = 1.7-3.3 gca, h mof ’propane for unsupported VMgO catalysts and 0.6-0.9 gcat h mof ’propane for supported catalysts)

Results and discussion

Unsupported VMgO catalysts Within the group of VMgO catalysts, crystalline MgO is the dominating phase at low V contents (Table 1). As indicated by UV-vis-DRS and EPR measurements, sample MggV, contains, besides some tetrahedral VO4 units in traces of Mg 3(V04)2, a certain amount of VOx species in octahedral and/or square-pyramidal coordination that form most likely an amor ­ phous overlayer on the MgO phase.

Table 1. Structural and catalytic properties of VMgO catalysts sample crystalline phases Sbet Mean V V content S(C3H6) XfCA) R(C3H«/ detected by XRD [nf/g] valence [at.-%] [%] [%]

MggV, MgO, Mg3(V04)2a 87.4 4.68 4.7 57.5 b 24.0" 1.15 1021 Mg,V, MgO, Mg3(V04)2 46.9 4.86 8.7 55.2b 23.5b 0.16 102' Mg,V, Mg2V207 7.9 4.91 18.3 46.0° 18.9° 1.03 1021 Mg,V4 Mg(V03)2, V2O5 1.9 4.92 25.3 32.8° 16.3° 3.00 1021

“traces; b 500 °C, W/F = 3.7 g h mol"1;c 500 °C, W/F = 1.2ghmoVl;d molecules m"2 s"1 moly"1 (X=2.5-4.3 %)

With increasing V content, VOx species are almost exclusively tetrahedrally coordinated and form crystalline ortho- and pyrovanadate phases. Higher V coordination numbers dominate again when vanadium is in excess, due to the formation of V2O5 being the main component in sample MgiV 4 besides Mg(V0 3)2. The mean vanadium valence state in the fresh catalysts increases as the V content rises (Table 1). To check whether structure and valence state of the active VOx sites change during the catalytic reaction, samples MgiV 4 and Mg,V, have been studied under ODH conditions. From in situ-EYRlon line-GC measurements it can be seen that the V2Os phase in sample MgiV 4 is markedly reduced with increasing time on stream as indicated by the growing singlet of inter­ acting V02+ ions (Fig. 2a). In ,s/ta-reduction during reaction is also confirmed by quasi-in site-XPS results after pretreatment under ODH conditions which revealed that 55 % of the surface V5+ sites are reduced to V4+ (Fig. 3). Moreover, crystalline V02 has been detected in the used catalyst by XRD and the mean V valence decreased from 4.92 (Table 1) to 4.60. In-

133 terestingly, this in .s/Yu-reduction goes along with an improvement in the catalytic perform ­ ance (Fig. 2) suggesting that probably due to its lower redox potential, is more selective than V5+.

In situ-EPR / on line-GC In situ - UV-vis-DRS

Nanometers

Figure 2. In shtf-EPR/on line-GC and in sifu-UV-vis measurements of samples MgiV 4 (a)

and MgjV] (b) at 500 °C in a flow of 13.5 % C3Hg, 6.8 % O2/N2.

sample MgiV< sample M^V, V /

514 516 518 520 522 524 Binding energy /eV Binding energy /eV

Figure 3. Experimental (thick) and deconvoluted (thin) quasi-in situ- XPS peaks (V 2p3/i)

The typical charge-transfer (CT) bands of V2O5 in the UV-vis spectra around 400 and

480 nm vanish already partly during heating in N2 flow to 500°C and almost completely un ­ der reaction conditions (Fig. 2a). Simultaneously, absorption increases below 250 nm in the range of CT bands of V4* while the respective d-d bands are usually very broad and weak and, therefore, hardly visible. With increasing time on stream, the band intensity increases in the region of 300-350 nm which is probably due to the deposition of weakly condensed polyenyl species being coke precursors [15]. Such carbonaceous residues have been detected, too, on the used catalysts by FTIR microscopy. Partial reduction of V5+ to V4+ does also occur in sample MgiVi, although to a much lower extent. The mean V valence was found to drop only from 4.91 (Table 1) in the fresh to 4.87 in the used catalyst and almost no changes were observed in the in situ-UV-vis and -EPR spectra (Fig. 2b). Probably, in s/to-reduction in this catalyst is mainly restricted to the near­

134 surface layers. This can be concluded from quasi-in situ-X PS results which indicate that 43 % of the surface V5+ sites are reduced to V4+ (Fig. 3). Bulk techniques such as UV-vis and EPR might be not sensitive enough to detect such small amounts of V4+. Moreover, tetrahedral V4+ that is likely to be formed from tetrahedral V5+ in MgzVzO? is EPR-silent at ambient and ele­ vated temperatures [10]. By comparing the catalytic results (Table 1) with the structural properties of the VMgO catalysts described above it turns out that activity and selectivity decrease with grow ­ ing content. This is first of all due the dramatic loss in the BET surface areas caused by the formation of bulk vanadates and V205. As a consequence, vanadium dispersions and, thus, the total number of VOx species accessible for reactants decreases, too. However, when normal ­ ized on the BET surface area and the overall vanadium content, rates of propane conversion and propene formation, being a measure of the intrinsic activity of the VOx sites, are highest for samples Mg,V, and MgiV 4. In these catalysts, the active V species are five- and/or sixfold coordinated by oxygen. In contrast, tetrahedral VOx species in Mg 2V207 and, even more pro ­ nounced, in Mg 3(V04)2 appear to be less active (R values of samples Mg 4V| and MgiVi, Ta­ ble 1) but more selective. This is explained by the lower number of lattice oxide ions sur­ rounding tetrahedral V species that can be supplied for the attack of the hydrocarbon mole ­ cules. In sample Mg,V 4 the redox cycle implies repeated steps of lattice oxygen supply and gas-phase oxygen uptake. Both are facilitated by highly polymerized octahedral VOx species in V2O5 crystallites leading to the highest intrinsic activity but lowest selectivity. From the results obtained with VMgO catalysts it appeared promising to use supported vanadia catalysts with highly dispersed, preferably tetrahedrally coordinated VOx species to achieve high selectivities. The inherently lower activities should be compensated for by using high surface area supports of low surface acidity. Therefore, we have used mesoporous alu­ mina and silica as supports for highly dispersed VOx species.

Supported VOx catalysts

Recent in .v/Vu-UV-vis, 51V-NMR and EPR measurements of V/MCM-41 have shown that V sites in the as-synthesized sample are mainly pentavalent and in octahedral and/or square-pyramidal symmetry but loose water ligands at temperatures higher than 100 °C [16]. Very similar results have been obtained withVO x/MCM-48 and VOx/SBA catalysts studied in this work. This indicates clearly that catalytically active VOx species supported on mesopor ­ ous silicas are in tetrahedral coordination. A weak low-energy CT band occurs in the in situ- UV-vis spectra of all samples between 380 and 400 nm suggesting that the VOx species on the surface of MCM-48 and SBA supports are not only isolated but also partly connected via V- O-V bonds. However, almost all VOx species are able to change their coordination symmetry reversibly between tetrahedral and/or square-pyramidal upon repeated dehydra ­ tion/rehydration cycles. This suggests that they are very well dispersed on the support surface and completely accessible by reactants. Thus, it is justified to calculate VOx surface densities and turnover frequencies (TOF) assuming that all V sites are exposed (Table 2). The intrinsic activity of the VOx sites reflected by TOF values as well as the propene selectivities do not differ much for the three silica-supported VOx catalysts. This agrees well with the fact that their local structure and valence state under reaction conditions is also very similar. Moreover, the different pore diameters seem to be of minor influence (Table 2). Due

135 to the much higher surface area of sample VOx/MCM-48, the maximum propene yield achieved with this catalyst is higher in comparison to the VOx/SBA samples.

Table 2. Structural properties and catalytic results at 500 °C of catalysts with 2.8 wt.-% of V

Sample Surface Sbet Mean pore Mean V TOFa>b S(C3H4)» YmaxB^Ha)

density 8 [m2/g] diameter valence [s'] [%] [%] [V/nm 2] [A]

1.0 VOVAI2O3 273 48.2 4.81 0.44 73.3 12.3 VOX/SBA50 0.43 645 52.6 4.81 0.18 83.3 14.5 VOX/SBA200 0.7 421 190.5 4.83 0.26 82.3 12.4 VOx/MCM48 0.37 889 26.2 4.86 0.21 80.1 18.0

* apparent values, calculated assuming exposure of all V sites, b for Xp,op „c= 2.5 - 3.9 %

In as-synthesized VOVAI2O3, the majority of V sites is tetrahedrally coordinated and slightly higher polymerised than on VOx/SBA and VO,/MCM. In contrast to the latter, the V coordination in VOx/Al2C>3 does almost not change upon heating. Moreover, a certain amount of five- and/or sixfold coordinated vanadium sites is also present. These species and the higher number of V-O-V bonds in VCVAI2O5 might be the reason for the higher intrinsic activity of the VOx sites in comparison to VOx/SBA and VOx/MCM (TOP values, Table 2). However, propene selectivities over VCL/AI2O3 are lower than over the silica-supported sam­ ples. This might be due to the higher concentration and strength of Lewis acidic sites that have been detected on VOVAI2O3 by FTIR spectroscopy of pyridine adsorption. In .s/Zu-EPRj'on line-GC/UV-vis experiments have shown that V5+ species on the sur­ face of AI2O3 are partly reduced to V4+ already at temperatures lower than the onset of pro ­ pene formation as evidenced by the EPR signal of interacting and isolated V02+ species (Fig. 4). However, it is probable that a marked amount of pentavalent VO4 units still persists under reaction conditions since no decrease of the CT band around 370 nm is observed in the UV- vis spectra that have been recorded simultaneously by the fibre optic sensor (Fig. 4). The gradual increase of absorbance at wavelength higher than 400 nm is most probably due to the formation of coke species (also detected by FTIR) rather than to d-d transitions of V4+which are usually very weak. The absorbance in this region of the UV-vis spectrum increases even more dramatically when the composition of the reactant gas is changed from 28 % C3II;, 14

% O2/N2 (Fig. 4) to 60 % C3Hg, 30 % O2/N2. These spectral changes are reversible to a certain extent by flowing H2 through the catalyst bed at 500 °C which partly removes the coke de- posites while keeping the V02+ EPR signal almost constant. In the catalytic tests, no marked deactivation of the catalysts has been observed. This suggests that carbonaceous residues are mainly deposited on the support material and do not cover the active VOx sites. In contrast to the VCVAI2O3 catalyst, the intensity of the V02+ signal observed in the in rifw-EPR spectra of VOx/MCM and VOx/SBA catalysts is negligible. As shown by UV-vis measurements, the VOx species in these materials are essentially in tetrahedral coordination. When this coordination symmetry persists during reduction, the respective V4+ species remain EPR-silent at ambient and elevated temperatures. However, when a flow of wet nitrogen is passed through the catalyst bed after cooling to room temperature, the typical EPR signal of

136 V02+ species in octahedral and/or square-pyramidal coordination appears since the tetrahedral V4+ species formed under reaction conditions adsorb additional water ligands. In agreement with the mean V valence state of the used catalysts determined by potentiometric titration this indicates clearly, that VOx species on silica supports are also reduced to a certain degree un ­ der ODH conditions. However, in contrast to VCVAI2O3, coke formation seems to be much less pronounced on thesematerials due to the lower surface acidity.

g 7 2

490

4 5 0

400 350

250

400 600 800 10 20 30 40

v Z n m x. s z %

Figure 4. Simultaneous in situ- EPR/on line-GC/UV-vis measurements during heating of

sample VCVAI2O3 in a flow of 28 % C3H;, 14 % O2/N2 (W/F = 2.7 g h mol" 1) us­ ing the set-up in Fig. 1.

Conclusions By comparing the results of catalyst characterization withthose of the catalytic tests it is evident that structure-reactivity relationships in vanadia-based catalysts are rather complex and require to consider the combined action of different properties (e. g. surface acidity, dis ­ persion, coordination and valence state of the active V sites) which may even change under reaction conditions. Although there is no simple relation between these properties and the catalytic performance, some key features can be derived: • Both V5+ and V4+ catalyze the ODH of propane, however, V4+ seems to be more selective though less active than V5+. • V sites in octahedral and/or square pyramidal coordination are more active but less selec­

tive than VO4 tetrahedra. • Highly dispersed VO* species are more selective but less active than polymeric VOx spe­ cies in amorphous clusters or even in crystalline chain- or layer-like structures. • Strong surface acidity gives rise to low selectivities by favouring the adsorption of pro- pene and its conversion to CO* and coke (preferably under oxygen-limited conditions). These results suggest that improved propene yields can be obtained with catalysts that contain highly dispersed, preferably tetrahedrally coordinated VOx species on non- or low-acidic sup­ port surfaces. The catalytic data obtained in this work with VOx supported on mesoporous

137 AI2O3 and Si02 promise that further improvement could still be achieved by optimizing these materials on the basis of the knowledge described above.

Acknowledgement The authors thank Dr. U. Bentrup, Dr. D. Muller and Mrs. R. Jentzsch for experimental sup­ port and the German Federal Ministry of Education and Research for financial support (grant no. 03C0280).

References 1. Blasco, T. and Lopez-Nieto, J. MApplied Catalysis A: General, 157, 117 (1997). 2. Banares, M. A., Catalysis Today, 51, 319 (1999). 3. Mamedov, E. A. and Cortes Coberan, V., Applied Catalysis A: General, 127, 1 (1995). 4. Chaar, M., Patel, D. and Rung, H. H., J. Catal, 109,463 (1988). 5. Michalakos, P. M., Rung, M. C., Jahan, I. and Rung, H. H., J. Catal., 140, 226 (1993). 6. Siew Hew Sam, D., Soenen, V. and Volta, J. C., J. Catal., 123,417 (1990). 7. Banares, M. A., Martinez-Huerta, M. V., Gao, X., Fierro, J. L. G., and Wachs, I., Catal. Today, 61, 295, (2000). 8. Rhodakov, A., Olthof, B., Bell, A. and Iglesia, E., J. Catal., 181,205 (1999). 9. S. T. Oyama, J. Catal., 128,210 (1991). 10. P. Rybarczyk, H. Bemdt, J. Radnik, M.-M. Pohl, O. Buyevskaya, M. Baems, A. Bruckner, J. Catal, in press. 11. Rosslick, H., Eckelt, R., Muller, D., Pohl, M.-M., Richter, M., Fricke, R., Proceedings of the International Conference on Advanced Materials „Materials Week”, Munchen, September 25th-28 th, 2000. 12. Bruckner, A., Rubias, B., Lticke B., and R. StoBer, Colloids and Surfaces, 115, 179 (1996). 13. Scofield, J. H., J. Electron Spectrosc., 8, 129 (1976). 14. Niwa, M. and Murakami, Y., J. Catal., 76, 9 (1982). 15. Rarge, H. G., Laniecki, M., Ziolek, M., Onyestyak, G., Riss, A., Rleinschmit, P. and Siray, M., Zeolites: Facts, Figures, Future, 1327 (1989). 16. Bemdt, H., Martin, A., Bruckner, A., Schreier, E., Muller, D., Rosslick, H., Wolf, G.-U. and Lticke, B„ J. Catal., 191, 384 (2000).

138 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Hartung*, S. Gaab*, J. Find*, A. Lemonidou**, J. A. Lercher* *Lehrstuhl fur Technische Chemie II, Technische Universitat Munchen, Garching, Germany, **The Chemical Engineering Department, Aristotle University of Thessaloniki, Greece

Oxidative Dehydrogenation of Ethane over Novel Mixed Oxides

Introduction Ethene and propene are among the most important basic chemicals in modern petrochemical industry. Naphtha steam cracking and fluid catalytic cracking of vacuum gas oil are well proven industrial processes for their production, but increasing the capacity of these processes is only possible to some extent, as the changing regulation limits the utilization of by-products (notably aromatic molecules) in fuels. Catalytic dehydrogenation of alkanes over supported metal catalysts, as an alternative route leading to light olefins, suffers from major disadvantages, i.e., the high tendency to form coke and the short cycle time of the catalysts. Conceptually, catalytic oxidative dehydrogenation is an interesting alternative route as thermodynamic limitations due to the chemical equilibrium are removed by coupling the dehydrogenation with the hydrogen oxidation reaction. Moreover, the presence of oxygen limits coking and extends so catalyst use between regenerations [1-6]. Li20/Dy203/Mg0 [7-11] mixed oxides have been found to be excellently suited to catalyse the oxidative dehydrogenation and cracking of alkanes. If these catalysts are promoted with halogens (Cl") remarkable catalytic activity has been observed. Adsorption experiments with appropriate probe molecules (NH3, CC2) revealed that with increasing concentration in chloride the concentration and strength of Lewis acid sites increased. In parallel, the observed formation of surface carbonates indicates that the Li* cations were preferentially located on the surface of the mixed oxide catalysts [8], The rate of reaction increased markedly at higher concentrations of chloride, which is attributed to the higher acidity of these mixed oxides. The selectivity to ethene varies between 75 and 90% showing a maximum at approximately 3 wt% chloride in the mixed oxide [8], However, depending upon the preparation marked differences in the catalytic properties have been observed. Therefore, the present contribution addresses the role of the different phases for the oxidative dehydrogenation of ethane.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 139 Experimental Catalytic materials A series of aqueous suspensions of the metal oxides and salts were prepared and stirred for 5 hours at 350 K. While the metal sources were kept constant, the chloride containing species were varied using mixtures of NH4CI and HOI (25%, 50% and 75%). The chloride concentration used was adjusted to 6 wt% Cl in the catalysts. In the following the samples are named according to the percentage of HOI used in the chloride containing solution, i.e., S75 is the mixture of 75% HCI and 25% NH4CI. The pH of the aqueous chloride containing solutions was 6.5 for those prepared with NH4CI and 1.5 for all other. All aqueous slurries had a final pH of about 10. Water was evaporated under reduced pressure and the remainder was dried and calcined in air at 973 K for 12 hours. Starting chemicals were MgO (Merck), Dy203 (Aldrich), LiN03 (Merck), smoking HCI (Merck) and NH4CI (Merck) in the highest purity grade commercially available and used without further purification.

Physicochemical characterization The chemical composition was determined by atomic absorption spectroscopy (Li, Mg, Dy) and ion chromatography (Cl). The BET specific surface was determined by N2 adsorption to be about 5.5 m2/g in all samples. XRD phase analysis was done on a Siemens D5000 instrument in Bragg-Brentano geometry equipped with a secondary graphite monochromator from 10 to 80 2 Theta in steps of 0.04/2s. Thermogravimetric analyses (TG) of the calcined catalysts were performed in a vacuum thermobalance (Setaram G11) equipped with a mass spectrometer (Balzers QMS 125). The catalysts were heated with a constant heating rate of 5 K/min from room temperature to 973 K and then cooled down to room temperature.

Catalytic experiments Kinetic measurements were performed in a fixed bed quartz tube micro reactor. Typically, approximately 300 mg of catalyst were mixed with the same mass of inert material (quartz or SiC), the rest of the reaction zone was filled with pure inerts. Prior to each run, the sample was heated to 973 K for two hours in pure He to remove adsorbed water and to decompose carbonates. The reaction temperatures varied from 723 K to 923 K. Under the operating conditions used, the empty reactor did not lead to measurable conversion of ethane. The ethane/oxygen molar-ratio was 1:1.2 with 1 bar total pressure (partial pressures ethane 70 mbar, oxygen 85 mbar). The oxygen was introduced as a mixture of 10% oxygen in helium. If not otherwise noted the weight hourly space velocity was 0.8 h' 1. A quartz bar was inserted downstream to reduce the post catalytic volume. The products were analyzed

140 with a Hewlett Packard 6890 Series gaschromatograph equipped with an FID and a TCD detector operated in parallel. A SUPELCO Carboxen 1010 plot column was used for product separation.

Results and Discussion Physicochemical analysis Elemental analyses showed that calcinations of the samples led to an overall loss of 50 wt% Li and chloride. No loss in mass was found for MgO and Dy203. The chemical composition was identically for all five catalysts prepared. On average, the materials consisted of 86 wt% MgO, 7.5 wt% Dy203, 3.5 wt% Li20 and 3 wt% Cl. The XRD patterns revealed MgO as main phase. While all samples contained additionally Li2C03 and Dy203, Mg(OH)3Cl*4 H20 was detected only in S00. The patterns of S00 and S50 are shown in Figure 1. Please note that a logarithmic scale has been chosen for the intensity in order to highlight the phases of low concentrations. Crystalline chloride phases were not detected in the samples S25 to S100. The loss of mass during thermogravimetric analysis increased with decreasing amount of HCI used in the chloride solution. The maximum loss (9 wt%) was found for S00, while the lowest loss was found for S100 (5 wt%). For S00 the weight loss occurred in three steps. At 370 K the smallest step (1 wt%) was identified to be due to the loss of adsorbed water. It was only observed with S00. At 520 K a second step of about 3 wt% was observed, which is attributed mainly to the loss of water, but also to some traces of CC2. For S25 to S100 this step was less distinct than for S00. The third step of 5 wt% at 900 K is attributed to the desorption of CC2. It is the most important step for the samples S25 to S100. In the upper part of Fig. 2 the first derivative of the mass losses of S00 and of S50 are compared as typical examples. Additionally, the rate of desorption of H2C and C02 are displayed in the lower part. For all other samples the decreases resemble with the S50 sample. At 973 K, the decomposition of the catalyst was not finished, but the procedure had to be stopped to prohibit the sublimation of Li-compounds into the equipment.

Kinetic measurements The initial catalytic activity was investigated testing the prepared catalysts at a WHSV of 0.8 h" 1. The conversion of ethane versus composition (3a) and temperature (3b) as well as the selectivity to ethene (3c) and the yield in ethene (3d) are compiled in Fig. 3.

141 MgO: #

1e+4

2 Theta [°] Figure 1: XRD patterns of calcined SOO and S50; MgO (#), Li2C03 (o), 0y2O3 (*) and

Mg(OH)3CI.4 H20 (a) are marked in the diffractograms.

300 400 500 600 700 800 900

------SOO '-0,02 ■ 30- - - S50

--0,06

Temperature [K]

Figure 2: First derivatives of the mass lost detected for SOO and S50 are given in the upper part of the plot. In the lower part the desorbed gas species, H20 in grey and C02 in black, are shown in correlation to the DTG signals.

142 100

80 —1----

40------4------—-i------4—T

20 —

S00 S25 S50 S75 S100 Sample Temperature [K]

S25 80 -=^-=-$50 — •--850 ------1—a

60----- I___ J___ ]_

40------40 ------

20----- ,— -i— I—W-f-

Temperature [K] Temperature [Kj

Figure 3 a - d: The conversions of ethane and oxygen and the ethene yield and selectivity under typical operating conditions as described in the Experimental section.

At 923 K for S50 a maximum conversion in ethane was observed of 94% with a yield in ethene of 69%. The samples prepared with a starting pH of 1.5 showed all much higher activities than S00, which was prepared with NH4CI only. Conversions and yields were up to three times higher than with S00. For most samples maximum in ethene selectivity was

143 observed at approximately 850 K. Other main products detected were H20 (not quantified) and COx. Methane was detected in traces (0.5 to 1.5 %) only with S25 to S100 at 923 K.

In order to test the influence of the inert material in the reactor bed, time on stream experiments were performed one with quartz and one with SiC using catalyst S25. Figure 4 compares the conversion of ethane in both runs, which were performed for the duration of 18 hours. The SiC diluted catalyst showed a more pronounced deactivation than the catalyst diluted with quartz.

Quartz SiC

400 600 800 1000 Time [min]

Figure 4: Time on stream experiments of S25 using quartz (black circles) and SiC (grey triangles) as inert materials.

Conclusions Li/Dy/Mg/CI mixed oxides are effective catalysts for the oxidative dehydrogenation of ethane to ethene. The modification of the preparation led to subtle, but important variations in catalytic activity. The conversion and yield to ethene normalized to the specific surface area were up to three times higher with the best catalyst compared to the worst. With increasing temperature the selectivity to ethene increased up to 90% with a moderately expressed maximum at 850 K. These variations in activity are concluded to be caused by variations in the phase composition and the surface defect structure of the complex mixed oxides and the mixtures thereof. At room temperature all catalysts consisted of MgO, Dy203 and Li2C03. The preparation in pure NH4CI solution (S00) led additionally to Mg(OH)3CI*4 H20 (as observed by XRD). This result is supported by thermogravimetric experiments, which revealed that the largest weight loss occurs in this catalyst and originates from the desorption of water. Three

144 steps are observed for S00, which are attributed to the loss of adsorbed surface water (step 1), the simultaneous release of crystal water and the decomposition of carbonates (step 2) and finally the sole decomposition of carbonates of high stability (step 3). This last step is found to be most distinct in the samples S25 to S100. Step 2 can be assigned to the decomposition of Mg carbonate, which starts to decompose at 723 K (under atmospheric pressure). Because the experiments were performed in vacuum, we conclude that the decomposition temperature was somewhat lowered. Step 3 is attributed to the decomposition of Li carbonate species and other residual basic carbonates, which we were not able to identify. These results suggest that the different chloride solutions during preparation strongly influenced the physical state of the final catalysts. The variations are explained by taking considering the equilibrium state between MgO and Mg(OH)2 in aqueous solution. Under standard conditions the phase equilibrium boundary line between both phases is in the range of pH~9. The more acidic HCI seems to stabilize MgO on the side of the pure oxide, while NH4CI shifted the equilibrium into the direction of Mg(OH)2. This hypothesis is supported by the existence of crystalline Mg(OH)3CI*4 H20, which has been found only in S00. Thus, S00 possesses, therefore, a high concentration in OH groups. In S25 to S100 this phase was not detected. Therefore, mainly surface near OH-groups should be present. Under the catalytic conditions applied water is one of the main products. The obtained results are, therefore, a strong indication that a gas-solid equilibrium between H20/MgO and Mg(OH)x exists as shown below:

MgO(s, + H20(g) 4 * Mg(OH)2

Throughout the reaction this equilibrium is shifted to the right side the more water is formed. Because there is more Mg(OH)2-like species present in S00 from the beginning, these sites are blocked already and not available in the reaction. In this sense S100 should be the most active catalyst. The maximum in the activity with the mixed NH4CI/HCI system suggests, however, a more subtle surface chemistry. More experiments are needed to explore this. Time-on-stream experiments were conducted to analyse the influence of quartz and SiC on the catalytic performance. The pronounced deactivation of the SiC filled reactor led us to conclude that here the loss in Li* and Cl under the applied reaction and activation conditions is mainly responsible for the deactivation (see also ref. [9]). The lower deactivation rate observed in the quartz filled reactor suggests that Li is trapped by quartz forming Li- silicates.

145 Acknowledgements The financial support from the IB of the BMBF (Project Nr GRC99/019), which facilitated the cooperation between Aristotle University Thessaloniki and TU Miinchen, is grateful acknowledged. We are grateful to Dr. Lori Nalbantian, Thessaloniki, for the XRD results.

References [1] H. H. Kung, M. C. Kung, Appl. Cat. A 157, 105, (1997). [2] M. Xu, J. H. Lunsford, Reaction Kin. & Cat. Lett. 57, 3, (1996). [3] K. Ruth, R. Burch, R. Kieffer, J. Catai. 175, 27, (1998). [4] E. A. Mamedov, V. Cortes-Corberan, Appl. Cat. A 127, 1, (1995). [5] T. Blasco, J. M. Lopez Nieto, Appl. Cat. A 157, 117, (1997). [6] M. Banares, Catai. Today 51, 319, (1999). [7] M. V. Landau, M. L. Kaiiya, M. Herskowitz, P. F. Van den Oosterkamp, P. S. G. Bocque, ChemTech. 26, 24, (1996). [8] S. Fuchs, L. Leveies, K. Seshan, L. Lefferts, A. Lemonidou, J. A. Lercher, Topics in Catalysis 15, 169, (2001). [9] M. V. Landau, A. Gutman, M. Herskowitz, submitted to Molecular Catalysis. [10] S. B. Wang, K. Murata, T. Hayakawa, S. Hamakawa, K. Suzuki, Cat. Lett. 62, 191,(1999). [11] E. Morales, J. H. Lunsford, J. Catai, 118, 255, (1989).

146 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

R. K. Grasselli University of Delaware, Newark, USA and University of Munich, Germany

Advances in the Selective Oxidation of C3 and C4 Hydrocarbons

1. Introduction

As is well known, selective heterogeneous oxidation catalysis is of vital importance to the well-being of mankind, producing about twenty five percent of the most important industrial organic chemicals and intermediates used in the manufacture of industrial products and consumer goods. Within this group, the selective oxidation of C3 and C4 hydrocarbons commands an important place, since products derived from them include such strategic intermediates as acrolein, acrylic acid, acrylonitrile, methacrylic acid, MTBE, maleic anhydride, and propylene oxide, to mention just a few.

Over the past fifty years great efforts have been expended, particularly by industrial researchers, to make the selective oxidation processes and catalysts ever more efficient and environmentally friendlier. The very term “selective oxidation catalysis” implies efficiency, preservation of matter, and thereby also environmental responsibility. The recently coined term “green chemistry ” has been practiced already for the past fifty years by researchers active in the area of selective oxidation catalysis, and with ever- greater prowess as time went on and the fundamental understanding of catalyst behavior on an atomic and molecular level improved. For example the highly inefficient and expensive process of IG Fatten (HCN + acetylene) to produce acrylonitrile was totally replaced in the 1960 ’s by the highly efficient and environmentally friendly SOHIO process (propylene + ammonia + air). And the yield of acrylonitrile was raised over the past forty years from 50% to over 80%, through the discovery and development of several generations of improved catalysts. Another example is the Chevron discovered process in the early 1980 ’s for the selective oxidation of n-butane to maleic anhydride, replacing the carbon inefficient process of oxidizing benzene to maleic anhydride.

It is the objective of this paper to give a brief overview of the status and advances achieved in the selective oxidation of Cs and C4 hydrocarbons particularly from the standpoint of the catalysts effective in this area, and to point out some possible improvements for the future.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 147 2. Discussion

2.1. Ammoxidation of Propylene

CH2 = CHCHs + NH3 + 3/2 02 (air) —catalyst—> CH2 = CHCN + 3H20

This process was invented by SOHIO (now BP America) and was commercialized in the early 1960's [1,2]. It completely displaced the earlier and inefficient IG Farben process, which used the expensive and environmentally unfriendly starting materials HCN and acetylene.

The process is a 6 electron oxidation and uses as catalysts mixed metal oxides selected from either molybdates or antimonates. The process is carried out in fluid bed reactors >10 m in diameter, at 400 to 460 "C, atmospheric pressure and contact times between 3-12 seconds. The world production stands currently at approximately 10 billion ponds per year. The main uses of this key petrochemical intermediate are in acrylic fibers, resins, rubbers and many specialty products.

Over the past forty years, the in tank yield of acrylonitrile has been raised from the low 50’s to now exceeding 80% [3], The improvement stems mostly because of the discovery of ever more efficient catalysts (Table 1). All of these catalysts have been discovered, developed and commercialized by SOHIO. Figure 1 shows the increase in the world production over time and the dramatic increase in production with the introduction of the SOHIO process in 1960; each new generation of catalysts further accelerates the production.

The molybdate catalysts all contain bismuth, from the first to the latest generation. Bismuth happens to be the best a-hydrogen abstracting element in a molybdate structure, second best is tellurium, but the latter suffers from volatility problems under redox conditions of the ammoxidation process. The other worthy comment to make about the more advanced multicomponent molybdates (MCM) is that they all are at least biphasic, with two or more phases cooperating with each other. For example the system KaNibCocFedBiMo^O* [4] derives its superior catalytic properties through the cooperation of the Scheelite-based a-Bi-molybdate phase (the actual catalytic phase), which contains also some Fe, and the p-Fe-molybdate phase, (the co-catalytic reoxidation phase). The latter phase contains, and is structurally stabilized by, Ni and Co to maintain the bulk of the iron in the Fe 2+ oxidation state, comprising centers for easy dioxygen dissociation and its re-incorporation into the catalytic cycle. For phase cooperation to be effective, it is imperative that the two phases be in intimate contact with each other, in order to facilitate their cooperation on an atomic scale. This condition is readily met when the two phases contain at least one lattice plane each that is structurally closely matched, thereby facilitating the formation of coherent interfaces (epitaxy) between the two phases. In the above example, the two phases are in registry, with less than a 2% mismatch at their respective {010} lattice face.

148 The antimonate catalysts (Table 2) all contain at least one element, which has a reduction potential above that of antimony [2,3]. These elements are U, Fe, Mn, Cr, Ce, V, to mention a few. Their function is to keep the antimony in its highest oxidation state, 5+, preventing it to permanently slip to 3+, which would make the catalyst ineffective (some Sb 3+ is however needed for the a-hydrogen abstraction to proceed, thus a balance must be provided through compositional choices).

In commercial operation, antimonates notoriously drift to lower oxidation states with catalyst deactivation. This problem can be overcome by an engineering expedient of incorporating into the fluid reactor an autoregenration zone at the bottom of the reactor (propylene and ammonia spargers are placed well above the air sparger). A more elegant way is to introduce into the antimonate a reoxidation co-catalyst, making this system also at least biphasic. For example among the more advanced multicomponent antimonates (MCA) the composition Nao-3(Cu,Mg,Zn,Ni)(M(V,W)o.o 5-iMoo.i- 2.5Te o.2-s Fei 0Sbi 3-2oOx, the two phases that cooperate with each other are the actual catalytic phase FeSb04 and the reoxidation co-catalyst Fe xMoOy. The latter prevents the antimonite from deep reduction during the catalyic process.

Because of the fickle nature of the antimonate catalysts, the molybdates are commercially the preferred ones. The last two generations of molybdate catalysts (Table 1) are designed to be practically indestructible and well able to withstand serious abuse in plant upsets [2,3]. These systems have been known to operate continuously for 10 years and more in the plant, with but only periodic addition of small amounts of M0O3, which slowly oozes out of the catalyst because during the redox process of the catalytic reaction, the volatile compound MoO(OH)2 is formed, which leaves the reactor. A great deal of the inherent M0O3 loss is however already foiled by steam coil rotation.

What improvements can still be made in the ammoxidation of propylene? First of all, there is no a priory reason why a 100% acrylonitrile yield (or nearly so) should not be attainable. There is no thermodynamic barrier. Probably the first thing to try to accomplish is to devise catalysts, which would be capable of lower temperature operation. One can take a hint from the selective oxidation of propylene, which proceeds with a better than 90%yield to acrolein, and the catalysts for the two reactions are related. Of course the difference is that the oxidation to acrolein is carried out 100°C lower, at about 320°C. Therefore, one of the approaches to take is to find a way to activate the ammonia at a lower temperature than what is possible with the current catalytic compositions.

Another improvement, which might be possible with the compositions as they stand, is to optimize the texture of the active phase. Not much effort has been expanded thus far in this direction. Modification of the Si02 support, or a totally different support, could also bring a few points. Compositional changes of the catalyst to reduce the M0O3 loss would also be of some value. Finally, engineering innovations might enhance acrylonitrile yields, such as the operation at lower pressures or by staging the air supply; even the use of oxygen and steam/C 02, instead of air might be of some benefit.

149 2.2. Oxidation of Propylene to Acrolein and Acrylic Acid

CH2 = CHCHs + Oz (air) — catalyst A — > CH2 = CHCHO + H20

CH2 = CHCHO + % 02 (air) — catalyst B —> CH2 = CHCOOH

The oxidation of propylene to acrolein is a 2 electron oxidation. The reaction is carried out commercially in shell and tube fixed bed reactors, with >10,000 tubes/reactor and tubes of approximately 2 to 2.5 cm in diameter and 3-5 meters in length. The reaction is carried out at 330-370°C, atmospheric pressure, and 1-3 sec contact time. The world production is modest, estimated at about 200 million pounds per year. The product is used as an intermediate for specialty products. The bulk of the acrolein (about 5 billion pounds/year) is not isolated, but passes uncondensed on to a second stage reactor where it is converted to acrylic acid.

The catalysts, which are employed for the selective oxidation of propylene to acrolein are very similar to those used in the production of acrylonitrile from propylene (Section 1.1). All of them are based on Bi-molybdates, the majority having been developed by SOHIO, Nippon Kayaku, Nippon Shokubay and BASF. Examples are: Ka(Ni, CojgFesBiPMo^ Ox-Si02 (85+% acrolein yield), (Na, K)a(Ni, Mg, Zn) b Fe cBidWeMoi 2Ox-Si02 (-93% acrolein yield).

The oxidation of acrolein to acrylic acid is also carried out in shell and tube fixed bed reactors of the same type as those used for the acrolein production. The reaction is carried out at 230-300°C, atmospheric pressure and 1-3 sec contact time. The world production exceeds 5 Billion pounds per year and is rapidly growing. The major uses of acrylic acid are in the production of polyesters, coatings, paints, adhesives, and super­ absorbents (disposable diapers).

The catalysts are all molybdates, derivatives of heteropoly acids containing phosphorous or vanadium as key elements. The advanced multicomponent catalysts (e g., CuaVb(Sn,Sb) cWdMoi2Ox-Si02) (-95% acrylic acid yield), contain also Cu as a redox element, allowing for higher activity of the catalyst and Sb, which enhances the selectivity of the catalyst. The acrylic acid catalysts have been developed mainly by the same major players who developed the acrolein catalysts (see above).

As can be seen from the impressive yields, about 95 % in both stages, respectively, for an overall acrylic acid yield of about 90 %, there is hardly much room for catalyst innovation. Selective doping might squeeze out another point or so, as might a careful optimization of texture. An improvement of M0O3 loss in both catalyst stages would be desirable. This might be attainable by selective doping of the existing catalysts, whereby a combinatorial chemistry approach might be useful. The rest must come from engineering innovations, such as oxygen staging, catalyst gradation along the bed, and improvement of the hotspot distribution.

150 2.3. Epoxidation of Propylene to Propylene Oxide

CH2 = CHCH3 + % 02 —catalyst—> CH2OCHCH3

The epoxidation of propylene to propylene oxide (PO) is a 2 electron oxidation. PO is manufactured by two basic processes: the traditional chlorohydrin (Dow) process, with CaCI2 as a stoichiometric byproduct, and the hydroperoxide (Arco) process, where t- butanol or styrene are coproducts [5], Both of these processes have obvious drawbacks because of their byproduct formation and for being environmentally hostile. For these reasons, much research has been expanded towards a direct selective oxidation process.

Propylene oxide is an important organic intermediate for the production of propylene glycol, polyether polyols, glycol ethers, alkanolamines and many specialty products. The world production of PO is in the range of 12 billion pounds per year.

Direct selective oxidation of propylene with dioxygen or air has met with little success thus far. The use of H202 as the oxidizing agent, or in situ produced peroxydic species from cofed H2 and 02 has shown some promise. ENI has succeeded to epoxidize propylene in good yields using H202 as the oxidizing agent and TS-1 as catalyst. More recently there has been a flurry of activity surrounding the use of Au based catalysts (particularly those supported on Ti02) and H2+02 cofed with propylene [6,7). Although the yields of PO are still modest (about 3%), the selectivities are very high, in the range of about 95%. One of the key factors appears to be the dispersion of Au and the stability of the dispersion. Surely improvements will be achieved in this area by doping the Au with appropriate elements to achieve increased yields and catalyst stability.

It is apparent from the research conducted thus far, that the catalysts active for the epoxidation of propylene are vastly different from those effective in the selective allylic oxidation. The latter use lattice oxygen as oxidizing moieties, the former poroxy species.

2.4. Ammoxidation of Propane

CH3CH2CH3 + NH3 + 402 (air) —catalyst—> CH2 = CHCN + 4H20

The ammoxidation of propane is an 8 electron oxidation. The process has not as yet been commercialized, although it has been pilot planted by BP America, Mitsubishi and Asahi. It is only a matter of time until it becomes commercial. The reaction will most probably be run in fluid bed reactors, when commercialized, and in a temperature range between 350 and 520°C, atmospheric pressure and 3-12 sec contact time.

While an array of catalysts has been investigated for this reaction, all of the best candidates thus far contain vanadium, whether they are molybdate or antimony based systems. The most promising thus far are the SOHIO VSb5Wo.5Teo.5Sn o.sO* [3], the ENI

151 (Cr,Sn,Ti,Ni,Mn,Fe) xVSbyO z [8], and the Mitsubishi Vo.sTeoaNbtmMoOx [9] systems (the Asahi catalyst is a variation of the Mitsubishi one, primarily replacing the Te with Sb). The highest yield of acrylonitrile is produced by the Mitsubishi system. One drawback of this system is its difficult synthesis (although it can be successfully mastered with a little extra effort), but most of all its Te content, which will be difficult to maintain during the operation of the process. In the redox reaction of the ammoxidation process, the Te gets steadily reduced to the 4+ oxidation state, which is volatile and will ultimately be lost out of the reactor.

While the yields of acrylonitrile produced from propane are becoming of commercial interest, there is still ample room for improvement. All three of the above contenders can be further improved, each in its own unique way. Some of the systems lack components to more effectively convert the intermediate propylene product to acrylonitrile, others operate at too high a temperature and need redox components or ammonia activators, and still others need to be made more redox stable.

Although the acrylonitrile yields from propane, using the above catalysts are nearly sufficient to commercialize the process; a true breakthrough in the catalyst would be welcome, which could challenge the well established propylene ammoxidation process.

2.5. Oxidation of Propane to Acrylic Acid

CH3CH2CH3 + 3 02 (air) —catalyst—> CH2 = CHCOOH + 2H20

The process is a 6 electron process. Thus far it has not been commercialized, because of the lack of appropriately effective catalysts. Once the catalyst is found, the process could be carried out either in fixed bed, transfer-line or fluid beds. Since grass root plants would probably be built for this purpose, fluid bed technology might hold an advantage.

Many catalysts have been studied for this reaction, by far the best among them is the Mitsubishi system, which has about the same composition as that for the ammoxidation of propane, V0.3TeozsNbo. 12MoO% [9]. As already mentioned in the above section 2.4., the Te poses a certain problem which needs to be addressed. One intriguing aspect of this catalyst system is its structure and how it is formed. At this juncture it is not certain whether two and possibly three phases cooperate with each other to give the rather respectable acrylic acid yields obtained (~ 40%). It might be, that a single phase is operative in this system. Currently, our evidence is mounting that it is a single phase. If this postulate ultimately prevails, the structure will lend itself ideally for alteration through molecular design manipulation. This should ultimately lead to higher acrylic acid yields, and also to higher acrylonitrile yields under ammoxidation conditions.

This area is also ripe for innovation and for a major breakthrough.

152 2.6. Ammoxidation of i-Butylene to Methacrylonitrile

CH2 = CH(CH3)2+ NH3 + 3/2 02 (air) --catalyst—> CH2 = CH(CH3)CN + 3H20

The ammoxidation of i-butylene is a 6 electron oxidation. The process has never been commercialized, although SOHIO ran a commercial propylene ammoxidation plant in the 1970 ’s for about a week on i-butylene, ammonia and air and USb 4.60x/Si02 as the catalyst. The run was successful, the in-tank yield of methacrylonitrile about 60%. Since the demand for methacrylonitrile is modest, the approach was not further pursued on a commercial scale.

Since the above plant test, research developed much more efficient catalysts, one of the best being Cso.5Ni4.5Co4.5Fe 3BiSbMoi 2Ox/Si02 [10], giving methacrylonitrile yields of about 85% with better than 90% selectivity (microreactor scale). This catalyst is already an excellent one, but could most probably be further optimized, if the demand for methacrylonitrile were to significantly increase.

2.7. Oxidation of i-Butylene to Methacrolein and Methacrylic Acid

CH2 = CH(CH3)2 + 02 (air) —catalyst A—> CH2 = CH(CH3)CHO + 3H20

CH2 = CH(CH3)CHO + % 02 (air) — catalyst B —> CH2 = CH(CH3)COOH

The oxidation of i-butylene to methacrolein is a 2 electron oxidation as is the oxidation of methacrolein to methacrylic acid. As in the oxidation of propylene, the oxidation of i- butylene to methacrylic acid would be produced in two consecutive fixed bed reactors, without interastage condensation of products. Current demand for methacrolein is too low to be of commercial interest. Methacrylic acid is currently produced by the well established acetone-cyanhydride (ACH) route which uses the toxic HCN as one of the starting materials and produces one mole of NH4HS04 for each mole of methacrylic acid produced; an environmental nightmare. An improvement of this process has recently been claimed by Mitsubishi, which proceeds via the methylester of 2-hydroxyisobutyric acid and dehydration via a Na-Y zeolite, thus circumventing the production of sulfate. Nonetheless, the incentive is high to replace the ACH process and the selective oxidation of i-butylene stands high on that list. Methacrylic acid is produced world wide on a 2 Billion pound per year scale, with its major uses being acrylic sheet (Plexiglas/Lucite), surface coating resins, water based paints, molding and extrusion compounds, biomedical appliances and optical products.

The catalysts for the first stage oxidation of the i-butylene to methacrolein are essentially the same as those for the production of methacrylonitrile, namely Cso.sNi4.5Co4.5Fe 3BiSbMo 12Ox/Si02 [10] and are cousins of the propylene selective oxidation catalysts.

153 The second stage catalysts for the oxidation of the methacrolein are substantially different from those used for the second stage acrolein oxidation; they are derived from heteropolyacids with partial replacement of the acidic hydrogen by cesium. Some of the best catalysts have the empirical formula of Csz5Ho.5[PMo1204o] and may be doped with Cu, V, Fe, As and Sb [11].

What further improvements can be done? The first stage catalyst is quite good and optimization studies would likely improve the yields of methacrolein further. The second stage catalyst could use further improvement. Here, fundamental studies, including kinetics, to better understand the individual mechanistic steps on a molecular level would be welcome. Also, the catalyst must be improved in its long-term stability. All of these shortcomings could be also addressed by computational chemistry approaches, which might also lead to entirely new catalytic systems. Further improvements could also be achieved by innovative engineering approaches.

2.8. Oxidation and Ammoxidation of i-butane to Methacrylic Acid and Methacrylonitrile

There is an incentive to directly oxidize i-butane to methacrylic acid, and less so to methacrylonitrile. At this juncture there exists no commercial process for these reactions.

Research has been expanded in the selective oxidation of i-butane to methacrylic acid, with the majority of the catalysts being of the heteropoly acid type, with intact Keggin

structures: HmXo.s_i sY0.2-15Z0-3P1-12Moi2 On , where X=V,As,Cu Y=alkali, Z=Sb,Sn, Group VIII elements [12]. The P is the central heteropoly atom and needed for the overall stability of the catalytic structure, the V, presumably to activate the paraffin, as is the case in most paraffin selective oxidation catalysts and the role of the Sb is to reduce some of the Mo®* to Mo 5*. Higher activities are presumably obtained with partially reduced catalysts and the Cu helps bring down the reaction temperature through its favorable redox potential.

Nonetheless, the yields of methacrylonitrile are meager and stand below 10%, and the catalysts lack long-term stability under reaction conditions. They reduce too readily, too far. For these reasons, additional research is needed to enhance the effectiveness of the catalysts. The area is wide open.

2.9. Oxidation of n-Butane to Maleic Anhydride

CH3CH2CH2CH3 + 7/2 02 (air) — catalyst —> C2H2(C0)20 + 4H20

The oxidation of n-butane to maleic anhydride is a 14 electron oxidation. It is the most commercially most successful light paraffin selective oxidation to date. The unique, and most used catalyst, which is essentially (VO)2P2C>7 was first invented by Chevron Corporation [Pat]. Since then commercial processes have come on stream using the

154 VPO catalyst, that include BP-UCB, ALMA, DuPont-Monsanto, and Mitsubishi Denka- Scientitle Design. The reaction is carried out at 320 - 360°C, atmospheric pressure and a contact time of between 1-10 sec. The reactors are either fluid bed or riser reactors. World production of maleic anhydride amounts to about 2 billion pounds/year. Major uses are for alkyd resins, laquers, plasticizers, lubricants, and intermediates for THF and butanediol.

The (VOjzPaOy catalyst is one of the most studied systems in the catalytic community [13,14], Recently, the Lonza group published an improvement of their ALMAX fluid bed catalyst [15], claiming that better yields are obtained when the catalyst was calcined in a fluidized bed calciner in a mixture of air and steam. The morphopogy is changed, from the usual N2 treated samples, leading to a lesser exposure of the basal {100}crystal faces, and allegedly higher selectivity at comparable conversion. The results and the interpretation of them are unfortunately not very clear. Optimization of preparation methods could lead to higher yields. Selective doping might also help bring down the waste products. It is indeed amazing, that thus far only the VPO system has been found effective in this area of catalysis. One would think, that there must be other systems. Perhaps here an intelligent combinatorial chemistry approach might be rewarding.

3. Summary

Selective oxidation and ammoxidation of 0$ and C4 hydrocarbons is an important endeavor for the petrochemical industry and of great benefit to society through the products which it produces for personal use, whether it be for building materials, carpeting, motor vehicles, tires, consumer products or luxury items.

The major processes and their catalysts for the selective oxidation of C3 and C4 hydrocarbons were briefly reviewed, their current status pointed out and some suggestions were made, where and how improvements may be forthcoming.

The key to the success of selective oxidation is, the various catalysts that have been developed over the years, their continual improvement and potential for still newer systems to be discovered. The deep understanding on an atomic level of the structural, surface and dynamic behavior of mixed metal oxides which comprise the bulk of the most effective catalytic systems, the application of modern and classical bulk and surface spectroscopic techniques, as well as the use of molecular probes and well placed combinatorial chemistry, will lead to the design and discovery of many new and more efficient, as well as environmentally friendlier selective oxidation catalysts. The time is nearing, when we will be able to introduce all of the desired catalytic functions into selected crystalline phases and space them in an optimum way for optimum product yield. The field is progressing rapidly.

In addition to developing new and more effective catalysts, engineering innovations are also going to help improve the desired product yields and selectivities, and help make the processes more efficient and environmentally friendlier.

155 References

[1] R.K. Grasselli, J. Chem. Ed. 63, 216 (1986). [2] R.K. Grasselli in: G.Ertl,, H. Knoezinger, J. Weitkamp (Eds ), Handbook of Heterogeneous Catalysis, 4.6.6. Ammoxidation, Wiley-VCH, 5, 2302 (1997). [3] R.K. Grasselli, Catal. Today, 49,141 (1999). [4] R K Grasselli, Topics in Catalysis, 15, 93 (2001). [5] Kirk-Othmer, Encyclopedia of Chem. Techn., (4thEd.) Wiley-NY, 20, 271 (1996). [6] B.S. Uphade, M. Okumura, N. Yamada, S. Tsubota and M. Haruta, Stud. Surf. Sci. Catal. 130,833(2000). [7] E.E. Sirangland, K.B. Stavens, R.P. Anders and W.N. Delgass, J. Catal, 191, 332 (2000). [8] N. Ballarini, R. Catani, F. Cavani, U. Cornaro, D. Ghisletti, R. Millini, B. Stocci and F. Trifiro, Stud. Surf. Sci. Catal. 135,135 (2001). [9] T. Ushikubo, Catal. Today, 57, 331 (2000). [10] R.K. Grasselli and H.F. Hardman, US Patent 4,503,001 (1985). [11] N. Mizuno and M.Misono, Chem. Rev., 98,199 (1998). [12] S. Yamamatsu and T. Yamaguchi, EP 425,666 (1998). [13] G. Cent!, Catal. Today, 16,1 (1993). [14] B. Kubias, F. Richter, H. Papp, A. Krepel and A. Kretschmer, Stud. Surf. Sci. Catal. 110, 461 (1997). [15] S. Abonettti, F. Budi, F. Cavani, S. Ligi, G. Mazzoni, F. Pierelli and F. Trifiro, Stud. Surf. Sci. Catal. 135,141 (2001).

156 Table 1: Propylene Ammoxidation Catalysts (Molybdates)

Acrylonitrile 1. Eariv Catalysts In-Tank Yields

Bi»PMoi2052 —SiOa -55%

Fe 4.s BU.5 P Mo 12O52 —S1O2 -65%

2. Advanced Multicomponent Catalysts (MCM)

Ka (Ni,Co)9Fe3BiPMoOi 2 -Si02 -75%

(K,Cs)3(Ni,Co,Mn)9.5(Fe,Cr)2. 5BiMoi20x —Si02 -78-80%

(K,Cs)a(Ni,Mg,Mn) 7 .s(Fe,Cr) 2.3Bio.5Moi20,-Si02 >80%

Table 2: Propylene Ammoxidation Catalysts (Antimonates)

Acrylonitrile 1. Early Catalysts In-Tank Yields

FeSbg.gOx —Si02 -65%

USb4.eOx —S1O2 -70%

2. Advanced Multicomponent Catalysts (MCA)

Nao .3 (Cu,Mg,Zn,Ni)o-4(V,W)o.o5-i Mo0.1-2.5Te 0.2-5Fe 10Sb13.20Ox —Si02 -75%

157 Production (M illions of Pounds) 4000.: 9ooo.: 2000 5000.: 7ooo.: 8000.: looo.: .: Figure

1.

World

Production Year |

of »

«

«

Acrylonitrile ■

j

■ DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

I. GriRtede, M. Kohler, H.-G. Lintz, H.-C. Schwarzer Institute of Chemical Reaction Engineering, University of Karlsruhe, Germany

The Influence of the Gas Phase Composition on the Catalytic Partial Oxidation of Propene

The partial oxidation of propene to acrolein was investigated in an integrally and isothermally operated fixed bed reactor. The influence of the gas phase composition has been studied in detail on a typical catalyst. The evaluation of the concentration profile along the reactor can be modelled by the use of a relatively simple reaction network. The results clearly indicate that the network is sufficient to describe the behaviour of the reaction system and allows the simulation of the reaction under typical industrial conditions. In the following the results of the experiments and the modelling are presented and discussed.

1. Introduction

Multitubular fixed bed reactors are normally used in the acrolein synthesis by partial oxidation of propene. Typical reaction conditions are 330-370°C, 1-2bar and feed compositions in the range of 4-8% propene, 6-15% oxygen, 5-30% water in nitrogen [1]. Oxidic multicomponent catalysts are the most active and selective catalysts in the process. Propene reacts mainly to acrolein and in parallel reactions to by-products such as CO and C02. Further reactions of acrolein lead to the formation of acrylic acid and also to by-products. In former works [2] at the Institute for Chemical Reaction Engineering, kinetic measurements resulted in the development of a relatively simple network to describe the reaction kinetics and allowed the modelling and the simulation of the concentration profiles along the fixed bed reactor. It was the aim of the present work to determine the influence of the gas phase composition on the reaction conditions. For that purpose the inlet concentrations of both, oxygen and water, were varied. Furthermore, the impact of the inert gas was investigated and experiments with nitrogen and propane were carried out.

2. Experimental

The measurements were carried out in a fixed bed reactor corresponding to one tube in an industrial multitubular reactor (length 1500 mm, diameter 15 mm). The reaction mixture is admitted through mass flow controllers. A saturation-condensation system admitted the water vapour. The reactor is divided in nine separately working heating

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 159 zones to assure isothermal conditions. There are eight sampling ports along the reactor for monitoring the gas phase composition. A side stream can be directed to the analysis section from each sampling port by the use of a multiposition valve. The organic products are analysed by GC, COx by an infrared photometer and oxygen by a magnetomechanic device. The experimental set-up is described in detail elsewhere [3], All experiments were carried out at a pressure of 1.5bar and a temperature of 360 °C. For the experiments an oxidic egg-shell catalyst with 20 weight -% active mass was used. The preparation is described elsewhere [4], The catalyst bed was diluted in a mass ratio 1:1 by steatit particles (diameter 0.5- 1 mm) to approximate plug flow.

The residence time up to sample port z is related to the active mass of the catalyst:

The concentrations of the carbon containing species at port z are normalised with respect to the incoming flow of propene, o:

X-. (2) yi propene, 0 3 with : Number of carbon atoms in molecule i n,: Molar flow rate; [hj = mol s'1

The amount of propane is not taken into account. In the case of reaction products the normalised concentration is identical with the yield YiiZ of such species

Yi,z = yu (3)

The conversion of propene is given by:

Hpropene, z= f~ypropene,z (4)

160 3. Results and Discussion

3.1 The reaction network

To evaluate the influence of the gas phase composition, all data are compared with the results of a standard measurement using 6mole% propene, 11mole% oxygen, 12mole% water in propane at the reactor inlet. The experimental values are shown in fig. 1, where the solid lines represent model calculations.

0.9 -

0.8 -

0.6- • propene A acrolein ■ acrylic acid ▼ by-products

0.3-

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

residence time /g.s/cm3

Fig.1: concentrations against residence time

As stated before the kinetics of the oxidation of propene can be described by a simple reaction network shown in fig. 2, where CO, CO2, acetone, acetaldeyhde, and acetic acid are lumped together in a pseudo species by-products.

C3H6 (1) - C2H3CHO (2)-^>C2H3COOH (3)

Fig. 2: reaction network

The rates of the individual reactions in this network are represented by following rate expressions:

161 ‘propene, 0 = ^111,12 ' Yl / [1+b 1 yl] (5.1)

'propene,0 = km,l4 ' Y) / [1+b * yi] (5.2)

'propene, 0 — km,23 * Y2 (5.3)

'propene.O — km,24 " Y2 (5.4)

'propene, 0 — km,34 " Y3 (5.5) with cprnpene,0: concentration of propene at the inlet of the reactor

The differential equation system for the compounds is obtained by combining the rate expressions with the conservation equations for a plug flow reactor. It is solved numerically by the Runge-Kutta method. An estimation of the mass transport and the temperature increase at the catalyst shows that these influences on the kinetic are negligible. Values of the kinetic coefficients obtained under standard conditions are summarised in table 1. They have been used to calculate the solid lines in fig. 1. Such values are not valid but for a defined feed composition as the concentrations of oxygen and water are not taken into account explicitly in the different rate expressions. They have to be refitted as soon as the concentrations at reactor inlet are changed. This will be shown below.

Table 1: kinetic parameters for xpr0 pene= 0.06, x0Xygen= 0.11,xwater= 0.12, rest Propane

k,, k...... cm [g s] cm' [g si cm [g s] cin ’,[(j-sJ - cm3/[g-s]

16.1 0.17 4.5 0.22 0.11 0.07

The overall activity of the catalyst may be defined by use of a lumped constant of pseudo-first order [5]:

. kip,!;+k n , h) + k„ (6) l-bX'/ln(l-X") the value of the activity parameter kmJ being valid for a given conversion of propene. In the following kmJ' will always be given at x‘ = 0.99.

162 3.2 The influence of the oxygen concentration

As one of the reactants oxygen is expected to modify the reaction rates significantly. Its content may be characterised by the ratio of the concentrations of oxygen and propene. At low values the complete conversion of oxygen should be limiting, if more oxygen is available the consecutive reactions of acrolein should be enhanced. Anyhow, due to the network of reactions involving oxygen it is not possible to define a stoichiometrically composed gas phase. A stoichiometric ratio should be between the value given by the partial oxidation of propene to acrolein

c3hg + o2 ------»- c2h3cho + h2o (R1) and its oxidation to C02:

C3H6 + 4.5 02 ^ 3 C02 +3 H20 (R2)

During the measurements the oxygen content was varied at a constant molar fraction of propene. The results are summarised in table 2. It is clearly seen that at low values of the oxygen content, an increase of its concentration enhanced the catalytic activity significantly. Above a value of the oxygen-propene ratio between 1.2 and 1.5 a further increase of the oxygen concentration has hardly any effect. Anyhow, the maximum yield of acrolein remained constant at about 87%.

Table 2: influence of the oxygen molar fraction on the kinetics, Xpropens — 0.06, Xwater = 0.12, Xoxygen and Xpropane Varied

cm '[cj s’

0.05 0.8 : 1 1.9 0.2 - 0.07 1.2 : 1 6.0 2.6 0.87 0.09 1.5 : 1 8.2 4.6 0.87 0.11 1.8 : 1 8.2 4.5 0.87 0.13 2.2 : 1 8.4 3.7 0.86 0.15 2.5 : 1 8.6 2.2 0.87

3.3 The influence of water on the reaction kinetics

For the determination of the influence of water on the kinetics and the maximum acrolein yield, its mole fraction was varied. As the results in table 3 show, the water content has only little influence on the activity of the catalyst.

163 The activity parameter km/ decreases slightly with an increasing water mole fraction. More distinctive is the influence on the acrolein yield. The highest maximum yield is detected without water in the feed gas.

Table 3. influence of water, Xpj-opene — 0.06, Xoxygen — 0.11 and x water) ^propane varied

k,, - ■' cm' [g s]

0.00 8.45 5.2 0.89 0.06 8.49 4.5 0.87 0.12 8.21 4.5 0.87 0.18 8.33 4.3 0.87

3.4 The influence of the nature of the inert gas on the reaction kinetics

In all the presented experiments, propane was used as inert gas. Commonly nitrogen is used for this purpose. The advantage of propane is a higher explosion limit for the oxygen-propene mixture. Further, as the heat capacity of propane is higher than that of nitrogen, propane has a positive effect on the heat dissipation at the catalyst com. In comparison to propane, for nitrogen the maximum yield is reached at a shorter residence time. But summarising, fig. 3 shows the influence of the inert gas on the reaction kinetics is negligible.

e propene A acrolein ■ acrylic acid ▼ by-products

— nitrogen — propane

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.3 1 Modifizierte Verweiizeit in g-s/cm3

Fig. 3: concentration profiles with nitrogen and propane as inert gas

164 3.5 The reaction of propane on the used catalyst

Another important point in this investigation was the remaining question if propane is really an inert compound. To obtain information about this, a reaction mixture of 11 mole % oxygen, 12 mole% water and 77 mole% propane was fed to the fixed bed reactor. Fig. 3 shows the consumed propane over the residence time and the measured mole fractions of propene, acrolein and CO, C02. At a residence time of 1 gs/cm3, only 0.26% of the inlaid propane was used up. As seen in fig.2, at this value of the residence time, the conversion of propene amount nearly 100%. Thus the propane conversion can be neglected against the conversion of propene and it can be regarded as an inert gas. Furthermore it was possible to describe the reaction kinetics of the propane reaction with following network:

Q)H8 (1) CsHs (2) -^>C2H3CHO (3)

rm,i4 ^ | rm,24 by-products (4j

Fig. 4: reaction network for the partial oxidation of propane

All the reactions in the network could be described by first-order reactions. The solid lines in fig.5 represent model calculations.

♦ consumed propane

• propene

▲ acrolein ▼ by-products

0.0012 -

0.0004 - •

residence time / g.s/crrf

Fig. 5: Propane consumption over residence time xpr0 pane= 0.77, x0Xygen= 0.11 and X water = 0.12

165 4. Conclusions

The results of the study show that optimal conditions are obtained at low water concentrations and a medium oxygen content in the feed. The substitution of nitrogen by propane which acts as an inert does not influence the kinetics of the propene oxidation. Consequently it is possible to use propane with its advantages against nitrogen as inert gas. Furthermore, the study shows that a relatively simple network can be used to simulate the partial oxidation of propene on an oxidic multicomponent catalyst. It is sufficient to use rate equations which do not take into account but the concentrations of the carbon containing compounds even though the concentrations of oxygen and water vary considerably from reactor inlet to outlet. On the other hand it is necessary to refit the kinetic coefficients as soon as the composition at reactor inlet is changed. Nevertheless, the good description of the experimental data enables to quantify the influence of the gas phase composition on the reaction kinetics of the partial oxidation of propene to acrolein.

Notation:

kinetic parameter, - r: reaction rate, mol/ (g s)

concentration, mol/cm3 V: volumetric flow rate, m3/s kinetic parameter, cm3/ (g s) x: mole fraction, - activity parameter, cm3/ (gs) X: propene conversion, -

mass, g yr- normalised concentration, -

molar flow rate, mol/s Y: Yield, -

References:

[1] K. Weissermehl, H.-J. Arpe; Industrielle organische Chemie, 5th edition, Wiley-VCH, Weinheim (1998) [2] D. Becker; Optimierung von Schalenkatalysatoren fur die partielle Oxidation von Propen zu Acrolein, Thesis, University of Karlsruhe (1989) [3] M. Kohler; Synergismus oxidischer Katalysatorkomponenten bei der Acrylsauresynthese, Thesis, University of Karlsruhe (2001) [4] European Patent 017000; Krabetz, Ferrmann, Engelbach, Palm, Sommer, Spahn (1980), [5] S. Breiter; H.-G. Lintz; Oxidation of Isobutene to Methacrolein on BiW/FeCoMoK Mixed Oxide Catalysts, Chem. Eng. Sci., Vol. 50(5), 785-791 (1995)

166 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

F. Klose*, B. Ondruschka**, P. Scholz**, R. Bruning** *Max-Planck-lnstitut fur Dynamik komplexer technischer Systeme, Magdeburg, Germany, “Friedrich-Schiller-Universitat Jena, InstitutfurTechnische Chemie und Umweltchemie, Jena, Germany

Selective Oxidation of Propane on Basic Metal Oxide Catalysts

Abstract

The catalytic behaviour of different basic metal oxide catalysts in conversion of propane is reported. The results obtained indicate that product selectivity between partial and deep oxidation can be influenced by varying GHSV, educt concentration, specific surface and catalyst porosity. It was found, that the formation of partially oxidized products increases significantly with the increase of GHSV and the decrease of catalyst specific surface.

I. Introduction

The oxidation of hydrocarbons belongs to the most important catalytic processes in chemical synthesis and environmental technology. The manufacturing of acrolein from propylene, and of maleic anhydride from butane on basic transition metal oxides, are well established [1,2,3], Deep oxidation of hydrocarbons to carbon dioxide and water supported by noble metal or copper, cobalt, manganese, or chromium oxide catalysts is known as a standard technique in exhaust gas cleaning. In the current literature on partial oxidation, a two-way-network is intensively discussed. The first reaction path yields partially oxidized products, which can then be further oxidized to carbon oxides. On the second reaction path, hydrocarbons are completely oxidized, immediately. Both reaction paths should occur on different sites of the catalyst surface [4,5,6], From the same literature, it may be concluded that oxygen on the catalytic surface plays an important role in determining the selectivity of oxidation. Thus, in recent years, some investigations with membranes were undertaken and were reported as successful by the authors [7,8]. In this presentation, we will show that there are some significant dependencies from the reaction conditions as well as from some catalysts properties such as surface and porosity.

II. Experimental a) Catalyst Preparation

The primary aim of catalyst preparation was to manufacture samples with different porosity and specific surface area. In order to achieve this, different procedures were used. Some commercially available samples were included in the investigations, as well. The commercially available catalysts represent the first group of samples, listed in Table 1. Samples 1-1 through 1-3 are designed for deep oxidation of hydrocarbons in exhaust gas cleaning. Catalyst 1-4 was prepared by a sequence of glowing, impregnating with aqueous transition metal solutions, and renewably glowing a copper wire net, as reported for example 3 in [9], This sample has only a very small specific surface of less than 1 m2/g. The second group comprise samples manufactured by impregnation of different supports (HITK Hermsdorf, Germany) with equal amounts of acetylacetonates of copper and chromium, followed by a calcination step. 20.0 g of every support treated in a small roll mixer for 2 hours with 4.1 g pre-milled Cu(acac)2 and 6.7 g pre-milled Cr(acac)3. After mixing, the samples were calcined at temperatures starting from room temperature up to 720 °C, keeping this value for 4 hours. This way, the acetylacetonates vaporize first and then diffuse into the pores of the support, before they are decomposed into oxides. A similar impregnation method, via vapour phase, was published by Babich etal. [10].

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 167 No.

- - - 8 8 P-1 P-2 P-3 P-4 C-3 C-l P-5 T-3 T-4 C-4 [181 Sample

2.1 n.a. 10.8 36.6 31.8 54.8 57.2 78.8 25.4 32.2 21.2 47.3 24.6 <0.1 123.2 143.0 [mVg} after

catalytic

test Surface 2.3 0.7 0.1 36.1 10.6 30.5 31.1 28.4 59.5 72.6 40.1 21.6 99.4 47.6 < 160.0 130.7 test BET before catalytic - - - - 550 550 550 720 720 500 975 720 720 720 720 450 fc/ Calcination Temperature. 91 (>4 3 Co

solution) 1 2 a. 10

n.a. 189 n waste

Mn0 Components

containing

3 * 4 30 29 lg/kg] 79 0 0.2 5.7 100 100 2 55.1 55.1 55.1 55.1 55.1 < <0.2 (Cr Active

Cr smoothing

water Doted

26 28 25 8.7 8.7 in

680 64.3 64.3 64.3 64.3 64.3 CuO leather

3 (2-2) solution) 15 analysis) 59 product

0

waste calcination 148 from 513 240 513 2

Fe and

crude

dyeing dyeing salts

the containing containing

smoothing *

3 mixing textile

elementary

(Cr

of

from leather

dissolving dissolving chromium

and NaOH

from and mechanical

2 (results with by values

(2-1) (2-1)

217 343 334 217 Si0 aqueous NaOH,

porosity metal metal extracted

3 sludge with

48 48

0

130 340 processes 2

materials molten

Al waste 3-2)

dying

acetylacetonates

with different

elementary

waste

Support 70 70 as

104 197 of CaO with waste

textile galvanization

(sample

smoothing given

containing solution, containing chromium chromium of

12 30 30 29 from and

and

MgO support carriers leather [9]

smoothing sludge

treatment

6 5 to no

0 chromium 54 54 preparation

sludges acetylacetonates 2

copper

dried

K leather

SrO dying Catalysts

wire, waste

with , and

impregnation

3 3 thermal milled aqueous 0 42 56

according 2 122 122 0

O 2 sludges an by 2 by

extracting treating

with Na

AI La Metal Alumosilicatic TiO, Textile by by calcined

with

waste

reported

Reference

and

support are

impregnated prepared prepared

of prepared prepared

composition calcined

was was impregnated

and

Procedure analysed

Manifacturing Catalysts Catalysts Catalysts Catalysts Commercial

IV-a IV-b III-a II III-a Ill-b III-b I 1 I 1 I Catalyst not Coarse-grounded - - - -

Impregnation Support Milled Solution Solution

1 4 3 2 1: - -

a b

- - -

-

- Concentrations

-

1 1-1 1-2 1-3 IV-b 1-4' 4-1 n.a. I II Ill-a III- IV- 3-7 Group 4-2 Group 3-2 3-3 3-5 2-4 2-6 Group 2-1 2-2 2-3 2-5 Group Sample No. Table

168 The third group of catalysts under investigation was prepared by grounding and thermally treating transition metal containing sludges, as reported in [11]. Constitution and properties are given in Table 1. These samples can be described as basic materials, due to their contents of alkali metal and alkaline earth metal oxides. Additionally, they contain 20-80 g/kg elementary carbon, which remains in the catalyst grains during calcination. This carbon is responsible for the porous framework in the "sludge catalysts". Samples of group 3 and 4 were calcined at 500-550 °C for 4 hours. Subsequently, according to these four different manufacturing procedures, two series of catalysts with decreasing surface areas were produced. One series was produced on the base of high active copper and chromium oxides, and the other, with less active iron oxide as the main active component. Samples 1-2, 1-4, 2-1 through 2-6, 3-3, belong to the first series. The samples 1-1, 3-2 through 3-7 represent the second series. b) Catalyst Characterization

All catalysts were characterised in their activity and selectivity behaviour, chemical composition, and specific surface. The composition was analysed with an ICP-OES "Trace Scan" from Thermo-Jarell- Ash. Preceding analysis, samples were dissolved in concentrated HNQ3. BET-surface was measured with an "Autosorb 1" from Quantachrome, after evacuation, at 300 °C, for 3 hours. The central part of our work investigates activity and selectivity of the catalyst samples. Measurements were carried out in a quartz tube reactor (d = 19 mm). All flow parameters and temperature values were registered and monitored with an self-written software package based on the HP VEE programming tool. Streams of gaseous components of the reaction mixture were monitored by means of mass flow controllers FC-260 from Tylan. The MFCs had different flow ranges, in order to mix a broad spectra of gas compositions. After passing the catalyst, a small amount of product stream was lead, via a heated transfer line, to a 250 pi sampling valve of a GC-MSD (GCD plus from Hewlett Packard). The GC operates with a PoraPLOT column (30 m x 0,32 mm). A temperature programme starting from 60 °C up to 230 °C was used to detect air, C02, H20, hydrocarbons, and partially oxidized products. From obtained GC data, conversion and selectivity rates were calculated.

III. Results

3.1. Propane Conversion

Propane is often employed as a model hydrocarbon in deep oxidation, due to its high thermal stability. Furthermore, the product stream of propane combustion can be easily analysed by GC with good separation of all components. In recent literature it is reported that product spectra may contain propylene, aldehydes, acetone, and carboxylic acids [12,13], Our aim was to clarify the selectivity behaviour and, furthermore, to get a better understanding of the possible reaction mechanisms. The measurements in our experimental set-up show, that for the non-catalytic total oxidation of propane, a temperature of approximately 680 °C is necessary. In the catalytic region, ODH to olefins becomes the most important side reaction besides deep oxidation. Partially oxidized products, containing oxygen play only a minor role in the product spectra. Dependency of the selectivity behaviour on reaction conditions and catalyst properties in this system, is examined in the following paragraphs.

3.1.1. Selectivity and Reaction Conditions

Dependency of selectivity on the reaction conditions was primarily investigated on mesoporous deep oxidation catalysts with large surface areas. Here, ODH or partially oxidized products are undesired, therefore, these side reactions must be suppressed. All tested samples decompose at temperatures below 450 °C to form propylene. At higher temperatures, however, a new significant propylene formation can be observed. Only at temperatures above 500 °C, the propylene destruction is dominant again.

169 a) Catalyst 2-2 (0,75 g, 5500 ppmv propane in air)

catalyst temperature f'C) catalyst temperature l°C] 10 l/h (GHSV = 5000) -*-20 l/h (GHSV = 10000) -+-30I/H (GHSV = 15000) | [-»-10 l/h (GHSV = 5000) 20im (GHSV = 10000) -*-30Lti (GHSV = 15000) |

b) Catalyst 2-4 (0,75 g, 5500 ppmv propane in air)

N: (I:

200 250 300 350 400 450 500 550 600 catalyst temperature !°C] catalyst temperature (°C) 10 l/h (GHSV = 5000) -*-20 LTi (GHSV = 10000)-*-30 IA (GHSV = 16000) | 10 l/h (GHSV = 5000) -*-20l/h (GHSV = 1 COOP) -*-30 l/h (GHSV = 15000)]

c) Catalyst 1-3 (0,75 g, 5500 ppmv propane in air)

catalyst temperature l‘C) -6-2,5 Q/20 l/h (GHSV = 6000) | catalyst temperature l°C] -»-0,759^0 l/h (GHSV =35000) -<-2.5^30 Th (GHSV = 10000) | •0,75 9120 l/h (QHSV = 24000) -6-2,5 g/20 l/h (GHSV = 6000) -Q.759'30 Kh (GHSV = 3S0Q0) -0-2,5^30 l/h (GHSV = 10000)

Fig. 1: influence of residence time on selectivity.

3.1.2. Contribution of the Catalyst

Rather than being influenced by the reaction conditions, the selectivity behaviour is influenced by porosity of the catalyst. For these investigations, some catalyst samples with different surface areas were prepared: deep oxidation catalysts with a mesoporous surface, and granulates from sludges with bigger pores and smaller surface areas between 3 and 30 m2/g. The extreme in reducing the catalyst surface is marked by metal wire catalyst 1-4, which can be described as compact material with no measurable inner porosity. All samples were tested under same standard reaction conditions: 2.5 g catalyst weight, 30 l/h , 5,500 ppmv propane in air. The results of measurements are reported in Fig. 2 and 3.

170 Propane Conversion Carbon Dioxide Selectivity

300250300350#0450530550600650700 cataMfermennre PO

Olefin Formation

»♦ ♦ ♦

catalyst temperature l°C] catalyst temperature [°C]

Fig. 2: Propane conversion and its selectivity in dependence on surface of catalysts - Cu- and Cr-oxides -

171 Propane Conversion Carbon Dioxide Selectivity

2D 250 309 350 400 450 500 600 630 TO cadvl tenreratue m catalyst temperature (°Q |-*-1-1 ~^^3-2 -B-3-3 -*-3-5 -0-3-7

Olefin Formation

200 250 300 350 400 450 500 550 600 660 700 catalyst temperature l‘CJ catalyst temperature pC] |-+-M -6-3-2-*-3-3 -*-3-5-0-3-71

Fig. 3: Propane conversion and its selectivity in dependence on surface of catalysts - Fe-oxides -

Two trends can be clearly observed. As expected, conversion rates of propane decrease significantly with lower surface of macroporous samples 3-4 to 3-7. In addition, selectivity of deep oxidation decreases significantly. In contrast the selectivities of propylene and ethylene formation increase rapidly. For the low-surface-area samples, up to 90 % of the reacting propane undergoes conversion to olefins. Comparing the amounts of propylene formed as a function of catalyst surface, it can be stated that an optimum occurs between increasing selectivity and decreasing hydrocarbon activation (samples 3-5 and 1-4).

172 The formation of ethylene is preferred at higher temperatures. Both selectivity and yield, increase with increasing temperature. Also, in the case of ethylene, the yields increase with decreasing catalyst surface and activity. The amounts of ethylene formed are significantly lower than those of propylene formed. Furthermore, it can be observed that, on the iron oxide catalysts, more ethylene is produced than on the copper and chromium containing samples The most remarkable behaviour was observed at metal wire catalyst 1-4. Below 610 °C this sample shows a very poor activity, less than 10 %. Besides deep oxidation, propylene formation with approximately 40 % selectivity can be measured. At temperatures above 620 °C, activity rapidly increases. In spite of a stochiometric propane-air ratio in the inlet stream of more than 1:8, first primary oxidation does not occur between 620 and 670 °C. There exists a window in which most of the propane is converted to propylene and ethylene. At 660 °C (propylene) and 670 °C (ethylene), both products show a maximum in yield curves. At 675 °C, the selectivity behaviour changes again, completely. Within these 5 -10 K the (non-catalytic) deep oxidation becomes the exclusive reaction. The measured data are reported in Fig. 4.

Fig. 4: the behaviour of metal wire catalyst 1-4 between 610 and 690°C

Literature [14,15] recently reported of similar metal gauze catalysts made from Pt/Rh. However, these catalysts showed activity only at temperatures above 800 °C and reaction proceeds .

IV. Conclusions

Our results clearly show that the selectivity of propane dehydrogenation is influenced by space velocity, since the amount of olefin formed is directly proportional to space velocity. Thereby, the proportion of olefins can increase by a factor of 2.5. The selectivity dependency on residence time in the catalyst, is a phenomenon that was proven in [16] for the partial oxidation of paraffin hydrocarbons to form maleic anhydride. The concentration of propylene is also directly proportional to the feed concentration. This dependency is not so significant. The surface area of the catalyst also plays an important role for selectivity of formation, according to our results. In this case, relations are somewhat more complicated. A decrease in catalyst surface area leads to an increasing ODH selectivity. The deep oxidation of the intermediate products is faster than the H-activation step. However, this can be prevented limiting oxygen availability (small specific surfaces and porosity of the catalysts, residence time, oxygen partial pressure). The formation of products with a different C-number (ethylene, acetaldehyde, methyl ethyl ketone) leads to the conclusion that in the mechanism an intermediate stage of activated hydrocarbon molecules must exist, where the break (preferred) and the recombination of C-C bonds is possible.

V. Acknowledgements

The authors would like to thank industrial enterprises for supplying catalyst materials.

173 VI. Literature

[1] Weissermel, K, Arpe, H.-J., Industrielle Organische Chemie, 5. komplett uberarbeitete Auflage, Wiley-VCH Weinheim, New York, Chichester, Brisbane, Singapore, Toronto 1998, 314; 406.

[2] Onken, U., Behr, A, Chemische ProzeBkunde, Lehrbuch derTechnischen Chemie Band 3, 1. Ausgabe, Thieme Verlag Stuttgart, New York 1996, 319.

[3] Wittcoff, H. A., Reuben, B. G., Industrial Organical Chemicals, I. ed„ Wiley New York 1996, 338-346.

[4] Chen, K., Khodakov, A., Yang, J., Bell, A. T., Iglesia, E., J. Catal. 186 (1999), 325-333.

[5] Zanthoff, H. W., Buchholz, S. A., Pantazidis, A., Mirodatos, C„ Chem. Eng. Sci. 54 (1999), 4397-4405.

[6] Dejoz, A., Lopez Nieto, J. M., Marquez, F., Vazquez, M, I., Appl. Catal.: A 180(1999), 83-94.

[7] Alfonso, M. J., Julbe, A., Farrusseng, D., Menendez, M., Santamaria, J., Chem. Eng. Sci. 54 (1999), 1265-1272.

[8] Mallada, R., Menendez, M„ Santamaria, J., Catal. Today, 56 (2000), 191-197.

[9] DE 196 11 395 C 1.

[10] Babich, I. V., Plyuto, Y. V., Van Der Voort, P, Vasant, E. F., J. Coll. Interf. Sc. 189 (1997), 144-150.

[11] Klose, F., Scholz, P., Kreisel G., Ondruschka, B., Kneise, R., Knopf, LI., Appl. Catal. B : Env. 28 (2000), 209-221.

[12] Busca, G., Daturi M., Finocchio, E., Lorenzelli, V., Ramis, G., Willey, R. J., Catal. Today 33 (1997), 239-249.

[13] Finocchio, E., Willey, R. J., Busca, G., Lorenzelli, V., J. Chem. Soc., Faraday Trans. 93 (1997), 175-180.

[14] lordanoglou, D. I., Bodke, A. S., Schmidt, L. D., J. Catal. 187 (1999), 400-409.

[15] Lodeng, R., Lindvag, O.A., Kvisle, S., Reier-Nielsen, H., Appl. Catal. A: Gen. 187, (1999), 25-31

[16] Sookraj, S.H., Engelbrecht D., CataLToday 49 (1999), 161 -169.

174 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

W. M. Brandstadter, B. Kraushaar-Czarnetzki Institute of Chemical Process Engineering CVT, University of Karlsruhe, Germany

Pilot Plant Processing of n-Butane to Maleic Anhydride above the Explosion Limit

Abstract

The influence of the n-butane feed concentration on the partial selective oxidation to maleic anhydride has been investigated using a pilot plant with a multi-sampling fixed-bed reactor. Butane concentrations were varied between 1 % and 2.5% v/v in air at a total pressure of 1.3 bar. It was observed that the catalyst appears to become less active and, accordingly, requires longer residence times to achieve the same conversion levels at increasing n-butane concentrations. However, the maximum yield is independent of the feed concentrations because the selectivities to maleic anhydride at the respective conversion levels are not affected. A simple kinetic model has been applied to describe the experimental data.

1. Introduction

The partial selective oxidation of n-butane to maleic anhydride (MA) has been the subject of numerous studies during the past decades. Typically, the public literature reports on experiments in a small range of n-butane concentrations up to 1.5 % because of the explosion limit, being at 1.7% v/v n-butane in air at atmospheric pressure. However, higher concentrations of n-butane in the feed could be commercially more attractive, if the higher throughputs achievable do also result in higher production capacities of MA. Safe processing in the explosion limit is possible with the multi-tubular fixed bed reactors typically used, provided that dimensions and packing of both, reactor and feed tubes enable the rapid quenching of radicals. In addition, the reactor cooling system must be sufficiently efficient to cope with the enhanced adiabatic temperature rise.

Our work was aimed at the evaluation of the potential benefits of processing at enhanced throughputs. For this purpose, we investigated the process kinetics in the relevant range of n-butane concentrations between 1% and 2.5% v/v in air.

2. Experimental

Experiments were run in a pilot plant equipped with a multi-sampling fixed-bed reactor. Isothermal operation was monitored or ensured, respectively, by means of

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 175 13 axially distributed thermocouples and 6 independent heating zones. The reactor is also equipped with 7 sampling ports along the catalyst bed which allow for the monitoring of the concentration profiles in the reactor without changing the flow rate.

Organic compounds were analysed on-line by means of a GLC with FID, whereas CO, C02, and 02 were continuously monitored via infrared absorption and magnetic susceptibility, respectively. In addition, the total off-gas is oxidised in a catalytic afterburner with CO/C02-analysers downflow in order to continuously check the carbon mass balance. A detailed description of the unit and the experimental procedures has been given elsewhere [1],

An industrial VPO egg-shell catalyst was used containing 50% m/m of active material. The catalyst particles have the shape of hollow cylinders with an outer diameter of 5 mm and a length of 4 mm. In all experiments, the catalyst was diluted with spheres of inert material (2 mm diameter) to ensure plug flow and an efficient heat transfer. In total, the catalyst bed had a length of 1.5 m and a diameter of 15 mm, and contained 100 g of VPO active material. Experiments were carried out at temperatures between 360°C and 420°C and at a total pressure of 1.3 bar.

The following quantities are determined at the location of each of the seven ports distributed along the catalyst bed. The modified residence time (Wu) is related to the mass of the catalytically active material and the volumetric flow rate at reactor inlet at reaction conditions. tmod.2 — rHcat.z / Fv

The gas composition at port z is expressed in terms of normalised dimensionless concentrations of species i (YilZ) according to

Yi,z= S/4 CilZ/ Cn-butane,0 — s/4 Fn.i.z/ FN,n-butane,0 where the molar flow of species i is related to the molar flow of n-butane at reactor inlet, and s, is the number of carbon atoms in molecule i. The sum of all normalised dimensionless concentrations Yi,z equals unity. In the case of reaction products Y,.z represents the yield. The conversion of n-butane (X„_buiane) can be expressed as

Xn-butane — 1 Y n-butane.

Finally, the reactor selectivity to product MA at port z (Sma.z) is obtained from

SmA,z = YMA,z / Xn-butane-

3. Results and Discussion

Typical results obtained with two different feed concentrations at 400°C reaction temperature are depicted in Figure 1. At a given residence time, the conversion level decreases with increasing butane concentration in the feed. This can be attributed to

176 a reaction inhibition by butane. It is also observed that the modified residence time, at which the MA yield is maximum, shifts to higher values when the butane concentration in the feed is increased. The plots on the right side show that the MA selectivity is not dependent on the feed concentrations. Consequently, the maximum yields of MA are practically identical for both butane concentrations.

n-butane, 1.0% O 2.5% • MA. 1.0% A

0.2 0.4 0.6 0.8 / (g*s)/ml ^n-butane

Fig. 1: concentrations versus residence time (left) and reactor selectivity to MA as a function of the conversion (right) at 400°C. Plotted are results obtained with 1.0% (hollow symbols) and 2.5% (filled symbols) n-butane in the feed.

Since no products other than MA and carbon oxides (COx) are formed in significant amounts, a lumped kinetic model can be proposed consisting of three reactions:

r =k *c MA^ r=k/c^ 1 1 n-butane n-butane ->co r =k *c 3 3 n-butane

Fig. 2: kinetic model used to fit the experimental data

As a first attempt, the rate equations are assumed to be of first order although the observed reaction inhibition by butane could suggest a different approach. To demonstrate the quality of this kinetic description, two sets of experiments are depicted together with the calculated values (solid lines) in Figures 3 and 4. It should be noticed that the deviations between measured and calculated values are most pronounced at low butane feed concentration and high temperature. The fit shown in Figure 4, which is still quite satisfactory, represents almost the worst case.

177 n-butane + MA X CO* *

0.2 0.4 0.6 0.8 / (g*s)/ml

Fig. 3: concentrations versus residence time (left) and reactor selectivity to MA versus conversion (right) at 380°C; 2.5% n-butane in the feed. The symbols represent the experimental data and the solid lines represent the kinetic model.

n-butane + MA X CO* x.

0.2 0.4 0.6 0.8

Fig. 4: concentrations versus residence time (left) and reactor selectivity to MA versus conversion (right) at 400°C; 1.0% n-butane in the feed. The symbols represent the experimental data and the solid lines represent the kinetic model.

The simple first order model, albeit not perfect, is charming because it allows for the straightforward definition of characteristic parameters which are useful to describe the catalyst performance.

First, we define an activity parameter (ka) which reflects the catalyst activity and can be obtained from the negative slope of the butane concentration (Y„_bu tane) at zero residence time (tm„d). ka=ki + k3

178 A pellet selectivity to MA (p Sma) can be defined which is independent of the reactor type used. It is specific only for the catalyst pellet type. p Sma= ki / (ki + k3) = ki / ka

The pellet selectivity corresponds to the differential selectivity at zero conversion. For any reactor type, it can be obtained by extrapolation of the integral (reactor) selectivity to zero conversion.

A stability parameter of MA (p Xma) is useful to characterise the stability of MA as an intermediate in the total oxidation of butane. In a plot of the reactor selectivity to MA as a function of the butane conversion, the stability parameter corresponds to the curvature of the selectivity. Its limits are zero and infinity. A more detailed description of these characteristic parameters has been given elsewhere [2], p Xma= ki / kz

In Figures 5, 6 and 7, we used these parameters to summarise and discuss our results.

1 1.5 2 2.5

n-butane.O

Figs. 5-7: Activity parameters (top left), pellet selectivities (top right) and stability parameters (bottom left) as a function of the mole fraction of n- butane in the feed for different temperatures. For better visualisation, solid lines are connecting the symbols belonging to the same temperature.

1 1.5 2 2.5 xn-butane,0

179 Figure 5 shows that the activity parameter is concentration dependent. It decreases with increasing amount of n-butane in the feed. This plot shows more clearly than Figure 1 that butane is inhibiting and that the first order approach is a simplification. The effect of the feed composition on the activity parameter is much smaller than the effect of the temperature. The pellet selectivity, in contrast, is almost unaffected by the n-butane feed concentration (Fig. 6). It decreases with increasing reaction temperature. No straightforward relation between the stability parameter (Fig. 7) and the temperature or the feed composition is obtained. However, a sensitivity analysis has shown that the stability parameter could be set to a value of 14 for the whole range of temperatures and feed compositions without significant effects on the quality of the data fitting.

We have also inspected more closely the magnitude of the yield maximum and its position with respect to the modified residence time. As a consequence of the inhibition by butane, the residence time required to achieve the maximum yield of MA must be prolonged when the concentration of butane in the feed is raised. As shown in Fig. 8, a higher temperature can compensate for this effect. The magnitude of the maximum yield is almost unaffected by the butane concentration (Fig. 9). Also, the temperature has a minor effect, only. Above 400°C, however, the yield maximum decreases significantly because the total oxidation is promoted.

380°C —0-

■n-butane. 0 "n-butane. 0

Fig. 8 (left): modified residence time required to achieve maximum yield to MA as a function of the mole fraction of n-butane in the feed. Fig. 9 (right): maximum yield to MA as a function of mole fraction of n-butane in the feed. For better visualisation, solid lines are connecting the symbols belonging to the same temperature.

180 The space-time-yield is an important magnitude to characterise the efficiency of the process. Since there are no organic byproducts in significant amounts, it could be possible to operate the reactor at maximum yield of MA without additional separation of organic byproducts. Assuming a stable catalyst performance, the production capacities plotted in Figure 10 were calculated according to:

PC = Fm.ma / mcat-pel. = P*Mma /(2*R*T) * YivlA.max Xn-butane,0 / tmod

Raising the temperature has a strong effect on the production capacity whereas there is no clear trend recognisable when the feed composition is altered. Since the fitted residence times of maximum yields were used to calculate the production capacity, which show a significant deviation from the experimental data in some cases, we expect that there could be a more clear trend visible, if a more sophisticated kinetic model is developed.

Conclusions

Our study showed that butane has an inhibiting effect in the partial oxidation to MA while the magnitude of the maximum yield remains almost unaffected. The question, however, whether processing in the explosion limit is an attractive option, can not be answered on the basis of our analysis. The calculated production capacities show no clear relation with the feed composition which can possibly be ascribed to the simplifying first order approach implemented in the kinetic model. Therefore, future work will be focused on the improvement of the kinetic description.

F 60 Fig. 10: Production capacity of MA a. 50 (g MA produced per kg catalyst per hour) as a function of the mole fraction of n-butane in the feed. For better visualisation, solid lines are connecting the symbols belonging to a 20 the same temperature.

"n-butsne.O

181 Notation

Ci concentration of species i Fm,MA mass flow of maleic anhydride Fv volumetric flow rate nicat.z catalytic active mass in reactor from inlet up to drawing port z meat pel. mass of whole catalyst bed (active and inert mass) Mma molar mass of maleic anhydride (98 g/mol) Fnj mole flow of species i p reaction pressure R gas constant (8.314J/(mol ‘K)) pellet selectivity to maleic anhydride Sma reactor selectivity to maleic anhydride PC production capacity modified residence time tmod.max modified residence time at which maximum yield of MA is reached T reaction temperature Xn-butane.O mole fraction of n-butane at reactor inlet Xn-butane conversion of n-butane Yi normalised concentration, yield of species i z index of sampling port £i number of carbon atoms in molecule i PXma stability parameter of maleic anhydride

References

[1] H.-G. Lintz, A. Quast; Chem. Eng. Technol. 22, pp. 817-822 (1999) [2] L. Riekert; Appl. Catal. 15, pp. 89-102 (1985)

182 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. L. Lapidus, F. G. Jagfarov, I. F. Krylov, N. A. Grigor'eva, A. Yu. Krylova I.M. Gubkin Russian State University of Oil and Gas, Moscow, Russia

Olefins Production in Catalytic Pyrolysis of Gas Condensate and Straight-Run Gasoline

At present the main method of lower olefines production is the process of liquid hydrocarbon pyrolysis. The major disadvantages of this process are: • The use of high temperatures (850 °C and higher) • The low selectivity regarding the main products (ethylene and propylene) formation • The low process control The application of catalysts is one of the most promising directions of the pyrolysis process advancement [1-7]. We have investigated the pyrolysis of gas condensate and straight nafta in the presence of steam over the high silica zeolite (HZSM, the molar ratio

Si02 : AI2O3 = 35, modified by the elements of the II and III groups of periodic system). The stable gas condensate (i.b.p. = 29 °C, f.b.p. = 390 °C, p = 0,757 g/cm 3, m.w. = 121) as well as straight nafta (i.b.p. = 37 °C, f.b.p. = 178 °C, p = 0,702 g/cm 3, m.w. = 100) were used as the stock. First of all, the investigation of the gas condensate pyrolysis over the initial pentasil-type HZSM was carried out (Table 1). From the Table 1 it is evident that the yields of the main pyrolysis products (ethylene, propylene, butenes and butadiene) are temperature dependent. With the increase of temperature the yield of ethylene also increases, and the yields of propylene and butenes at 640 °C pass through a maximum, the yield of butadiene constantly decreases. Catalytic action of zeolite has the essential impact on the yield of one of the main products - ethylene. Even at the temperature 680 °C its yield comes to 21,9%, and at 760 °C it is as much as 30,7 %, that far exceeds the

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 183 yield of ethylene obtained in the process of thermal pyrolysis with the same stock ’s characteristics carried out at higher temperatures.

Table 1 Pyrolysis of gas condensate over the pentasil-type zeolite (contact time - 0,15 sec., steam dilution - 70 %)

Products yield (% wt.) at temperature,v | Products 600 620 640 660 680 700 720 740 760 | 0,9 | 2 0,4 0,5 0,5 0,6 0,8 0,9 1,1 1,2 0,6 1,1 1,2 0,9 2,0 0,4 0,8 0,8 0,5

2 0,3 0,5 0,2 0,2 0,4 0,7 1,7 0,4 1,5 12,6 4 4,4 5,5 6,9 8,7 13,6 17,2 18,8 21,7

2 6 2,4 3,3 2,5 2,5 2,9 2,5 2,7 2,5 2,2 2 4 11,1 13,7 15,8 18,8 21,9 24,3 26,5 29,2 30,7 I 3 8 1,4 2,5 0,9 0,8 0,6 0,4 0,3 0,3 0,2 10,2 3 6 12,8 13,8 14,4 13,5 11,5 8,9 8,3 7,0 £ 4 10 1,1 1,5 0,8 1,2 0,8 0 0,1 0,1 0,1 2 4 s 2,8 3,4 3,4 2,8 2,3 1,5 1,2 1,0 0,9 2 6 0 0 0 0,3 0,2 0 0,3 0 0 Gas 41,1 48,7 48,9 52,2 57,9 56,5 62,4 64,4 67,5 2 2- 4 30,5 33,8 35,9 37,3 37,8 38,0 38,5 40,3 40,4 (unsat.) Tar 57,7 50,1 49,8 46,3 40,5 41,6 35,6 33,4 30,0 Coke+loss 1,2 1,2 1,3 1,5 1,6 1,9 2,0 2,2 2,5 2 4/36 1/1,5 1/1,0 1/0,9 1/0,85 1/0,5 1/0,4 1/0,3 1/0,3 1/0,2

The ratio of ethylene and propylene yields varying between 1 : 1,15 (at 600 °C) and 1 : 0,2 (at 720 °C) is of significant interest. Thus, this ratio at temperatures below 660 °C is higher than in the thermal pyrolysis (1 : 0,5). Due to the relatively low yield of gas in the condensate pyrolysis over the zeolite at temperatures below 700 °C the further investigations over modified zeolite were carried out at the temperature range of 700-780 °C. The results of these investigations are shown in Table 2. The comparison of these data demonstrates that zeolite modification by the elements of the II and III groups (samples Z-II and Z-III) results in significantly higher yield of gas. Thus, the yield of gaseous products over zeolite at 700 °C amounts to 56,5 %, and amounts to 66,03 and 64,7 % over modified zeolites Z-II and Z-III, respectively. This tendency can be also seen at higher temperatures. In this case the

higher yield of the most valuable C% -C4 unsaturated hydrocarbons is observed. The ethylene/propylene ratio exceeding 1:0,5 is retained up to 740°C for Z-II and up to 720°C for Z-III, that is 80°C and 60°C higher than the unmodified sample, respectively.

184 Table 2 Pyrolysis of gas condensate over the modified zeolite*-* (modifier content - 1 % wt., contact time -0,15 sec., steam dilution - 70 %)

Products yield (% wt.) at temperature," over the different catalysts Products 700 720 740 760 780 Z-II Z-III Z-II Z-III 1 Z-II1 Z-III Z-II Z-III Z-II Z-ffl 2 1,03 0,7 1,4 0,9 1,4 0,9 1,45 1,2 1,5 1,2 1,7 0,7 0,6 0,7 0,7 0,5 0,8 0,8 0,9 0,4 2 0,3 0,3 0,4 0,4 0,5 0,2 0,6 0,2 1,0 0,6 4 11,9 16,8 14,6 13,1 16,5 14,9 17,0 17,9 18,5 19,2 2 6 3,4 2,6 2,4 2,9 2,7 2,7 2,9 2,7 3,0 2,7 21,8 2 4 26,1 22,1 26,4 24,1 28,9 26,5 32,1 30,5 35,7 0,6 3 8 1,0 0,6 0,7 0,5 0,5 0,5 0,4 0,6 0,2 3 6 17,1 16,8 15,2 17,2 16,7 14,9 15,0 13,9 12,0 10,6 2 4 10 0,7 0,7 0,8 1,0 1,0 0,6 1,0 0,4 0,9 0,1 2 4 s 2,3 3,3 1,2 3,5 1,7 3,5 2,0 1,3 2,5 2,7

4 6 0,5 0,4 0,6 U 0,6 0,7 1,0 0,2 1,0 0,2 2 2- 4 46,0 42,3 39,1 48,2 43,1 48,0 44,5 47,5 46,0 49,2 (unsat.) Gas 66,03 64,7 60,0 67,8 66,4 68,3 68,75 71,1 72,4 73,6 Tar 32,84 34,3 38,8 31,0 32J 30,4 29,75 27,4 25,8 24,7 | Coke+loss 1,13 1,0 1,2 1,2 1,4 1,3 1,5 1,5 1,8 1,7 1 2 ^ 1/0,7 1/0,8 1/0,7 1/0,65 1/0,7 1/0,5 1/0,6 1/0,4 1/0,4 1/0,3 3 6

*) Z-II and Z-III - zeolite is modified by the elements of the II and III groups of periodic system

The modification of zeolite results in decrease of methane yield by 3­ 5% compare to the unmodified sample. If we compare the effectiveness of action of the studied zeolite modifiers it should be pointed out the following: the incorporation of the III group element increases the yield of ethylene by 4-5 % starting at the temperature of 720 °C; the yield of butadiene is slightly higher at all temperatures studied; the yield of C2 -C4 unsaturated hydrocarbons is higher by 4-7 %; the yield of gaseous products at the temperatures up to 780°C is also higher. At the same time the higher yield of propylene is also observed at the temperatures up to 760°C at the II group element incorporation. Hence, the use of zeolite-based catalysts will insure the performance of wide fractional range hydrocarbons ’ pyrolysis at the temperatures that are 100-150 °C lower than the thermal pyrolysis temperatures. Thus, the yield of

C2 C4 unsaturated hydrocarbons comes to 50 % per feed stock and the yield

185 of ethylene is as much as 30-35 % which is much higher than in the thermal pyrolysis of the similar stock. The ethylene/propylene ratio also increases. The results of straight - run gasoline pyrolysis over the zeolite- containing catalysts (initial pentasil type zeolite and zeolite modified by the element of the III group of periodic system (Z-III) differs from the similar sample used in gas condensate pyrolysis) are given in Table 3. For comparison the results obtained during the thermal pyrolysis of straight -run gasoline are also shown in Table 3.

Table 3 Straight -ran gasoline pyrolysis over the zeolite-containing catalysts (contact time - 0,15 sec., steam dilution - 70 %)

Thermal 1 Products yield (% wt.) over the different catalysts pyrolysis at at temperature,0 temperature,0 j Products 740 760 780 zeolite Z-III-I zeolite Z-III-I zeolite Z-III-I 800-810 840-850

2 0,9 1,0 0,9 1,1 1,0 1,2 U 0,9 - 0,5 - 0,6 0,6 0,9 0,14 0,16 I 2 1,0 0,3 1,8 0,8 0,3 0,3 0,01 0,02 I 4 13,1 14,6 15,8 19,7 18,8 21,4 15,5 20,4

2 6 2,6 3,1 2,5 3,1 2,7 2,6 3,7 3,0 I 2 4 25,8 27,4 30,2 31,7 35,5 35,6 25,1 28,7

3 8 0,3 0,7 0,4 0,4 0,3 0,6 1,3 0,7

3 6 14,6 18,7 14,7 15,1 13,6 12,9 14,8 11,7 S 4 10 1,9 1,0 1,5 1,1 0,9 0,5 1,0 0,8

2 4 8 2,9 3,6 3,8 2,9 2,9 2,5 3,1 3,6 4 6 3,5 1,7 1,3 1,9 2,9 2,2 3,6 4,2 46,8 51,4 50,0 51,6 54,9 53,2 46,6 48,2 1 2” 4 (unsat.) Gas 66,6 72,6 72^ 78,4 79,5 80,7 69,35 74,18 Tar 32,3 26,2 25,8 20,5 19,2 17,9 29,05 24,12 | Coke+loss U 1,2 1.3 U 1,3 1,4 1,6 1.7 2 4/ 1/0,6 1/0,7 1/0,5 1/0,5 1/0,4 1/0,4 1/0,5 1/0,5 1 3 6

The consideration of data given in Table 3 shows that the use of zeolite-containing catalysts allows to reduce the process temperature by 70 ­ 100 °C with the same ethylene yield compare to the thermal pyrolysis. Approximetly 25 % of ethylene yield was obtained over the zeolite at 740°C and in the process of the thermal pyrolysis the same yield of ethylene was obtained at 800-810 °C. The yield of ethylene of 24,7 % was obtained over the modified zeolite at the same temperature 740 °C, and in the process of thermal pyrolysis 28,7 % of ethylene was formed only at 840-850 °C. At the temperatures of 760 °C and especially of 780 °C the yield of ethylene over

186 the zeolite and Z-III-1 catalyst amounts to 30-32 % and 35-36 %, respectively. In the process of the thermal pyrolysis of straight-run gasoline such yield of ethylene can not be reached at 840-850 °C and higher temperatures. The use of zeolite-containing catalysts increases the yield of gaseous products by 4-5 % at temperatures that are 50-90 °C lower the thermal pyrolysis temperatures. Approximetly 50 % wt. of C2-C4 unsaturated hydrocarbons ’ yield was also obtained at lower temperatures. It should be noted the decrease in butadiene yield (especially over modified catalyst) compare to the thermal pyrolysis. The ethylene/propylene ratio higher than 1:0,5 is observed over the zeolite-containing catalysts only at temperature 740 °C. It decreases at higher temperatures and at temperature 780 °C it is getting even lower than the ratio in the process of thermal pyrolysis of straight nafta. The obtained data demonstrate that catalytic liquid hydrocarbon pyrolysis for the purpose of ethylene and propylene production has a number of advantages over the thermal pyrolysis.

References 1. T.N.Mukhina, S.P.Chemykh, S.V.Adelson and others. Hydrocarbon pyrolysis over the catalysts//TSNIITENeftekhim.: L., 1978 - 72 p. 2. Z.G.Zulfugarov, E.B.Sharifova, F.A.Zeinalova. The effect of mordenite- containing catalysts composition on the hydrocarbon stock pyrolysis/ZNeftekhimiya: M., 1983 - v.23 - 5 - p.p. 841-845. 3. Yu.G.Egiazarov, B.H.Cherches, N.P.Krut’ko. Oil fractions pyrolysis over the indium-oxide catalyst// Neftekhimiya: M., 1979 - v.19 - 4 - p.p. 542-600. 4. H.M.Minachev, D.B.Tagiev, Z.G.Zulfugarov. Nafta pyrolysis over the catalysts// Neftekhimiya: M., 1980 - v.20 - 3 - p.p. 408-411. 5. US Patent 4471151 (1976)

6. Patent of Japan 193834 (1984) 7. Russian Federation Patent 2141379 (1999)

187 188 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

Kh. M. Minachev, A. L. Lapidus, A. A. Dergachev N. D.Zelinsky Institute of , Russian Academy of Sciences, Moscow, Russia

The Mechanism of Olefin Formation from Light Paraffins over MFI Zeolites

1. INTRODUCTION.

Low hydrocarbons are readily available as a cheap raw material for the synthesis of polymeric materials, high-octane components of motor fuels as well as various intermediates in the fine organic synthesis. Among the principal

sources of light paraffins, is the casing-head gas (C2-C5 alkane content up to 70%) and associated gas (overall content of ethane, propane, and butanes up to 30%), the scale of the extraction of which runs into tens and hundreds of billions of cubic meters per year respectively. Until recently, a considerable fraction of these gases was used primarily as industrial fuel and the gases are not yet employed in the synthesis of chemical products.

An effective utilization of gaseous hydrocarbons has demanded the search for new catalysts, among which synthetic zeolites are of particular interest. Pentasil type zeolites (MFI) show the unique ability to convert light paraffins into

higher-molecular-weight products, mainly into aromatics [1].

The primary step of the light alkanes was identified as the dehydrogenation of saturated molecules into corresponding alkenes [2,3]. Catalytic activity and selectivity of pentasils in the transformation of light hydrocarbons are associated with their acidic properties and specifically with the ratio of Lewis (L) and Bronsted (B) acid sites and with acid site strength. As a result of the introduction of some metals (Zn, Cd, or Ga) into pentasils, strong electron-acceptor sites are formed [2,4].

189 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 Until recently there has been no adequate ways of optimizing the ratio L/B and obtaining the efficient catalyst for the transformation of the lower paraffins into olefines and aromatics. This ratio can be varied by changing the concentration of the modifying element, by the reversible blocking of some of the sites with ammonia, pyridine or other poisons, and also by irreversible destruction of some of the sites, for example by means of high-temperature treatment. The present study was aimed at investigating the catalytic properties of metal-containing pentasils and the effect of varying zeolites acidity on selectivity of transformations of light paraffins into dehydrogenation products. The nature of the promoting action of Zn, Cd, Ga, and Pt in the olefin formation is discussed.

2. EXPERIMENTAL

Modified pentasils were prepared by impregnating the initial H-MFI (Si/Al=20-110) with solutions of zinc, cadmium, or gallium nitrates. The sample of Pt-MFI was prepared by impregnating H-form of pentasil (Si/Al=20) with

[Pt(NH3)4]Cl2 solution.

The catalytic experiments were performed with a continuous flow reactor at atmospheric pressure and T= 673-823 K by passing the undiluted C3-C4 gases through the catalyst bed.

Acidic properties of the catalysts were studied by IR spectroscopy, using the pyridine as a probe [4], I.r. spectra were measured after the samples loaded with pyridine were evacuated to 10"4 Pa at 473 K.

3. RESULTS AND DISCUSSION

The nature of the acid sites of the modified pentasils

Modification with Zn, Cd, and Ga significantly alters the acidic properties of MFI zeolites. The influence of these promoting agents on the concentration, strength, and nature of the acid sites that are active in the dehydrogenation of light paraffins has been studied in detail. Specifically, IR spectroscopy was used to monitor the state of the hydroxyl coverage and investigate the interaction of the pyridine with L- and B-sites of H-MFI and modified pentasils with different Si/Al ratios. The effect of high-temperature treatment on acidity of decationated and Zn-MFI was also investigated.

190 Adsorption of pyridine on zeolite samples was carried out at 313 K. Acidic properties of pentasils were compared at 513-543 K to eliminate the effect of physical adsorption of pyridine. Absorption bands due to pyridinium ions (1550 cm"1, B-sites) and coordinately bound pyridine (1448-1462 cm"1, L-sites) have been observed in the range of 1300-1600 cm"1 in the IR spectra of pyridine adsorbed on H-, Zn-, Cd-, and Ga-MFI. The position of the band due to the coordinately bound pyridine is changed with the nature of the cation and the hypsochromic shift (from 1448 cm"1 for H-MFI to 1462 cm"1 for Ga-MFI) is induced by the increase in the electrostatic field strength and the strength of the L-sites (for Cd-MFI: 1453 cm"1; forZn-MFI: 1456 cm"1).

Figure la shows the ratio L/B as a function of [Al] in the framework of the H-MFI and Zn-MFI. A common feature is characteristic of both series of catalysts — the increase in the L/B ratio with decreasing number of B-sites, which is approximately proportional to the number of aluminium atoms in the unit cell of pentasils. As a result of the modification with zinc, the L/B ratio increases by a factor of- 1.5 indicating the appearance of a large number of new L-sites.

Figure lb shows how the ratio L/B varies with the temperature of preheating (Tp h). Treatment of the H-MFI and Zn-MFI samples in the range 873-1073 K has no appreciable effect on the L/B ratio, but for higher Tph there is a considerable increase in L/B, due mainly to a fall in the number B-sites. Note that even for Tph=1273 K the intensity of the band at 1456 cm"1 (L-sites of Zn-

MFI) remains practically unchanged, while the band at 1550 cm"1 appears in the spectrum as a shoulder, so that the ratio L/B cannot be calculated for this sample.

1073 1173

Figure 1. Dependence of L/B ratio on the Al concentration (a) and on the preheating temperature (b): 1 - H-MFI; 2 - Zn-MFI.

191 The spectroscopic data lead to the following conclusions: -strong L-sites are formed in pentasil type zeolites as a result of introduction of zinc, cadmium or gallium; -for TPh higher than 1073 K there is a considerable fall in the number of B-sites in H-MFI and Zn-MFI; -L-sites of Zn-MFI are stable up to the temperature of 1273 K; -destruction of the H-MFI framework can be observed at Tph>l 173 K.

Catalytic data

The relation between the product distribution and the conversion level of isobutane (Xcmio) at 823 K over H- and Zn-MFI is presented in Figure 2.

* 30 -

ti, 20

Isobutene conversion. Isobutane conversion,

Figure 2. The yield of (C3 + C4 ) (a) and yield of (C, -C3) (b) as a function of isobutane conversion: 1 -H-MFI; 2 - Zn-MFI (873); 3 - Zn-MFI (1273)

The predominant products of isobutane conversion over H-MFI at

X c4hio 60% are CrC3 paraffins, which can be formed by isobutane cracking.

This phenomenon is especially pronounced at Xomio 3 0%. The ratio C4H8

/C3H6 is close to 0.8 -0.9 (Figure 3).

The main direction of isobutane transformation at the conversion level 10-60% over Zn-MFI is the formation of propylene and butenes. A similar picture was observed also for Cd- and Ga-MFI. In contrast to the H-MFI, over the metal- containing pentasils the unsaturated products are represented mainly by butenes

- the ratio C4H8 /C3H6 = 3-7 (Figure 3).

192 Isobutane conversion,

Figure 3. The dependence of the ratio C4H8 / C3H6 on isobutane conversion: 1 - H-MFI; 2 - Zn-MFI (873); 3 - Zn-MFI (1273)

Comparison of the product distribution over Zn-MFI calcined at 873 and 1273 K (conventionally termed Zn-MFI (873) and Zn-MFI (1273)) shows that after pretreatment of the catalyst at 1273 K dehydrogenation predominates, with a comparatively slight fall in cracking yield. Over the Zn-MFI (1273) at Xcmio

~30% selectivity for the formation of C3-C4 olefins is nearly 80% (Figure 4).

Figure 4. Selectivity to cracking products (1) and to (C3 +C4 ) (2)at Xc4Hio ~ 30% for H-, Zn-, and Pt-MFI

193 Our spectroscopic and catalytic results clarify the role played by the L- and B-sites in accelerating the formation of olefines from light paraffins over H-MFI and metal-containing pentasils. Activity of H-MFI in the transformation of paraffins into olefines is due to the presence of strong B-sites that are responsible for protonation, hydride ion abstraction, and cracking reactions. In the case of Pt-pentasils the formation of dehydrogenated structures involves Pt + as the active site [5,6].

The dehydrogenation of the light paraffins over the modified pentasils occurs on the strong L-sites, which contain ions of metals. These L-sites have strong electron-acceptor properties and show specific activity in dehydrogenation reactions. This conclusion is confirmed by the experiments with Zn-MFI (1273) that indicates that the sample is active and exhibits reasonable dehydrogenation selectivity although it has a negligibly low number of B-sites.

Thus development of an effective catalyst for dehydrogenation of light paraffins demands optimization of the ratio of strong L- and B-sites. The fraction of L-sites is determined by the number of atoms of the promoter. The content of B-sites in above catalysts depends essentially on high-temperature treatment.

REFERENCES

1. M.S.Scurrell, Appl. Catal., 32, 1 (1987). 2. Kh.M.Minachev and A.A.Dergachev, Russ. Chem. Rev., 59, 1522 (1990) . 3. C.R. Bayense, A.J.H.P. van der Pol and J.H.C. van Hoff, Appl.Catal., 72, 81 (1991). 4. Kh.M.Minachev and A.A.Dergachev, Russ. Chem. Bull., 6, 1071 (1998). 5. E.S.Shpiro, O.P.Tkachenko and Kh.M.Minachev, Indian J. Techn. 30, 161 (1992) 6. A.A.Dergachev, Solid Fuel Chemistry (Russ.), 32, 1 (1998).

194 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. N. Bhave*, A. Klemt **, S. R. Patwardhan*, W. Reschetilowski*** ‘Department of Chemical Engineering, Indian Institute of Technology, Bombay, India, “CRI KataLeuna GmbH Catalysts, Germany, “‘Institute for Industrial Chemistry, Dresden University of Technology, Germany

Investigations of Mixed MCM-41/MFI Catalyst Systems for the Manufacture of Light Olefins by Cracking of Hydrocarbons

Abstract

Recently, the large pore MCM-41 material has gained a considerable amount of interest, as a solution to the diffusion limitation provided by the conventional microporous cracking catalysts for the cracking of long chain or bulky hydrocarbon feeds. The present work deals with the synthesis, and characterization of novel, mixed A1-MCM-41/MFI catalyst systems. Furthermore, the catalytic behaviour of the as synthesized AJ-MCM-41/MF1 mixed systems for cracking hydrocarbons by MAT (Micro Activity Test), standardized by ASTM D-3907, have been investigated in the present work. In situ formation of the meso- and microporous structures

Al-MCM-41 and ZSM-5 have been obtained using a two-template [Ci6H33(CH3)3NBr and

(CH3CH2CH2)4NBr] synthesis gel system. The mixed phases were obtained by optimizing template ratios and reaction temperature. The comparison with conventional zeolite Y and pure MCM-41 with regard to activity and selectivity showed a better activity and selectivity towards

C3 and C4 olefins. The obtained results reveal a potential of mixed Al-MCM-41/ZSM-5 catalysts systems as cracking catalysts and in turn demand for further optimization of physico-chemical properties in future in order to be exploited as commercial cracking catalysts.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 195 1. INTRODUCTION Tailoring catalysts is one of the major constituents of the refinery optimization process [1]. The increasing propensity of the modern refinery, towards cracking of the bulkier residual feeds (the bottoms of the barrel), has increased considerably, the demand for the usage of wide pore catalysts in the hydrocarbon industry. The state of the art cracking catalysts consist of the active compound zeolite Y (0.74 nm), and the conventional octane booster ZSM-5 (0.54 x 0.56 nm) additive. However these microporous zeolites offer considerable diffusion limitation to the long chain or bulky hydrocarbon feeds, due to their geometry. The advent of the family of mesoporous molecular materials M41S, and in particular of MCM-41 [2], a prominent member of this family, has led to a considerable development in the use of these materials in petroleum related applications. Corma et al. [3] investigated the catalytic behaviour of MCM-41 material and concluded its higher activity and good selectivity for cracking reaction, but also inferred a poor hydrothermal stability for the same. Recently, it has been confirmed that the large pore aluminosilicate Al-MCM-41 with pore sizes of ca. 2.5 nm has a promising potential in cracking long chain and bulky hydrocarbons, owing to its mesoporosity [4], Karlsson et al. [5] incorporated crystalline ZSM-5 in the mesoporous siliceous MCM-41 materia] and obtained improved hydrothermal stability for such mixed MCM-41/MFI catalyst systems. The present work deals with the investigation of such Al-MCM-41/ZSM-5 mixed systems. The Al incorporated in the purely siliceous MCM-41 is on the account of providing enough acidity and activity for cracking reaction. This work deals with the detail synthesis of such A1-MCM-41/ZSM-5 mixed systems, their characterization and finally, their catalytic behaviour for cracking the long chain hydrocarbon by MAT (Micro Activity Test), standardized by ASTM 3907. Furthermore, comparisons with a zeolite Y and pure Al-MCM-41 material with regard to the activity and selectivity were performed. 2

2. EXPERIMENTAL The A1-MCM-41/MFI mixed systems were synthesized using mixed surfactants, one gel

synthesis method, as described by Klemt and Reschetilowski [6], The material was converted from Na-form to the acidic H-form by ion exchange, using the following procedure: The Na-form

sample was treated three times in aqueous solution of NH4NO3 (0.1 N), with a liquid/solid ratio

196 of 10, at 78 °C, and continuous stirring for 3 h. Each treatment was followed by washing with distilled water until constant conductivity was achieved followed by the overnight drying of materials at 120 °C. Ammonia was removed by calcining the material with a temperature programme as follows: RT - 120 °C, 1 °C min" 1, in synthetic air, (10 1 h"1), 120 °C for 30 min, 120 °C - 450 °C, 1 °C min" 1, in synthetic air, (101 h"1), 450 °C for 2 h, cooling to RT in N2 flow, 10 1 h"1. Thus the sample obtained in H-form was further used for the characterization. The catalyst samples were analyzed by XRD (S 5000 diffractometer with a nickel filtered

Cukc radiation) and the scattering intensities collected from 0.5 0 to 130 ° (29), by scanning at 0.5 ° (28), steps with a counting time of 15 s at each step. Nitrogen adsorption-desoprtion tests were carried out using volumetric analyzer Sorptomatic 1990. The catalyst samples were first degassed and calcined to 350 °C, 12 h at a rate 1 K min" 1, and subsequently the measurements were carried out at 77 K. Atomic Absorption Spectroscopy (AAS) was used to calculate the composition of the catalyst systems. The detail parameters and sample nomenclature are given in Table 1.

Table 1. Composition and textural properties of the materials investigated sample Si/Al ratio of the templates BET before MAT, BET after MAT,

ratio C3/C16, mol mol" 1 m2 g" 1 m2 g" 1

M10Z90 22.0 75/25 420 391

M75Z25 20.0 75/25 670 570

M90 Z1O 22.4 10/90 765 650 *H-Y 2.6 - 985 854 *MCM-41-17 17.3 - 849 818 Data marked with *, procured from [7],

Catalytic investigations were performed in a microactivity (MAT) arrangement, using n-hexadecane as the model feed, according to the ASTM D-3907-92, at 482 °C, with 1.33 g of n-hexadecane for 75 s and with 4 g of catalyst in a fixed bed [8], After the feed injection the reactor was stripped with flowing N2 (30 ml min" 1, for 20 min). The gas fraction was analyzed by GC-FID (VEGA 6300-01, Carlo Erba), with hydrogen as the carrier on a PLOT-megabore

197 column (GS-A1 50 m x 0.53 id). The liquid products were analyzed on Hewlett Packard 3365 Chem Station instrument, equipped with a DHA (Detailed Hydrocarbon Analysis). The tests under standard conditions were repeated under constant catalyst/oil ratio ~ 3 and temperature of 482 °C, to account for the reproducibility and steady state activity of the catalyst. The sample nomenclature used for the mixed systems is on the basis of relative amounts of Al-MCM-41 to MFI material formed as observed from the intensities and heights of the XKD peaks. The sample

M10Z90 denotes the relative amounts of A1-MCM-41/ZSM-5 materials, i.e. -10/90.

3. RESULTS AND DISCUSSION Well formed Al-MCM-41 and ZSM-5 type materials were obtained, according to the X-ray diffractograms. As shown in Figure 1, for example, the mixed systems M90Zio exhibits only one, rather broad peak, in the characteristic 29 < 5 ° region of MCM-41, and the typical

ZSM-5 characteristic peaks in the region, 29 = 7-10 0 and 29 = 22-25 0 [9], The peak height, the intensity and the structure of the isotherms are function of the ratio of Cg/Cis surfactants under the chosen conditions.

14000-

12000-

10000-

8000-

6000-

4000-

2000-

2 0, Grad

Figure 1. X-ray diffractogram of the M90 Z10 material.

198 600-

500-

400-

300-

200-

0.0 0.2 0.4 0.6 0.8 1.0 X axis title

Figure 2. Nitrogen isotherm of theM 90 Z10 material.

All the three mixed systems subjected to Nitrogen adsorption-desorption (for example the mixed systems M90 Z10, see Figure 2), show a hysteresis loop in the region p/p0 = 0.5-1.0, normally interpreted in terms of textural mesoporosity. Samples M75Z25 and M90 Z10 reveal a sharp inflection in the curve at p/po = 0.33, due to pore condensation. The distribution of the products obtained from cracking of n-paraffins is of importance to the refiners. n-Hexadecane was used as the model feed for MAT cracking test, performed on the catalyst. For studying the activity of the mixed systems, series of MAT tests were performed. The samples achieve steady state conversion values, which account for the deactivation behaviour. The mixed systems show a relatively higher activity as compared to the pure MCM-41-17 material and H-Y zeolite. Conversion for all the three mixed systems was comparable and in the range of 60 - 65 % (see Table 2). In case of mixed systems, the yield of gaseous and liquid products is ~ 40 %, whereas, zeolite H-Y and pure MCM-41-17, offers more liquid products (~ 65 %) and less gaseous products (~ 20 %). The product ratios iso CVtotal C4, (C3+C4) paraffin/olefin and C3/C4, account for the selectivity behaviour of the mixed systems, in comparison with H-Y and MCM-41-17. The mixed systems offered more isomerisation capability in the C4 range (~ 0.5), as compared to the H-Y and MCM-41-17 materials. It can be observed, that the increase in ZSM-5 content in the mixed system drastically increased the paraffm/olefin ratio. Thus, Al-MCM-41 produces more unsaturated products than ZSM-5, and

199 the ratio is independent of conversion. The system M90 Z10 offered the most unsaturated products

(paraffin/olefin ~ 0.2). Behaviour of M75Z25 is comparable to that of zeolite H-Y

(paraffin/olefin ~ 1.0).

Table 2. Catalytic properties of selected samples

sample M10Z9 O M75Z25 M90 Z10 *H-Y *MCM-41-17

conversion rate, wt. % 60 68 63 45 45

gas product, wt. % 48 35 45 25 25

liquid product, wt. % 45 42 44 70 67

iso C4/total C4 0.50 048 048 0.44 036

(C3+C4) paraffin/olefin 1.9 1.0 0.2 1.1 0.9

Data marked with *, procured from [7],

Differentiating the C3 and C4 products and their variation with conversion results in the

behaviour as depicted in the Figure 3. It can be observed that the ratio C3/C4 is scattered around

the value 80% for the sample with high AJ-MCM-41 content. On further lowering the

Al-MCM-41 content, as in case of system M75Z25, the ratio is seen to be independent of

conversion and with further reduction in amount of Al-MCM-41, as in case of mixed system

M10Z90 , the C3/C4 ratio goes on increasing. In general, the mixed systems are observed to give

more C3/C4 ratio as compared to thatof H-Y and pure MCM-41-17 material. From the Figure 4, it

is clear that both, Al-MCM-41 as well as ZSM-5 contribute to the higher yield of C3 and C4

fractions. As Al-MCM-41 content in the mixed system increases, the yield of propylene

increases, as ZSM-5 content increased, is also observed to increase the n-propane, n-butane and

isobutane yields. Thus in accordance with results of Koch and Reschetilowski [7], Al-MCM-41

in the mixed systems produced more unsaturated products (propene, 1-butene, isobutene),

whereas, ZSM-5 being the more active component, enhances the hydrogen transfer rate and thus

causes the yield of n-propane, n-butane and isobutane. The yield of undesirable fractions C1+C2,

is low as compared to other gaseous fractions.

200 1.0-1

0.9- M„Z„

0.8 -

0.7- y ' o" 0.6 - MCM-41-17

0.5-

0.4- HY

0.3-■ 60 65 70 75 80 conversion rate, wt. %

Figure 3. C3/C4 ratio of the cracking n-hexadecane.

M10Z90 M75Z25

C,+C2 C,= n-C 3 i-C4 n-C 4 i-C = 1-C = i-C5 n-C,

Figure 4. Products of the cracking n-hexadecane.

4. CONCLUSIONS

The two-template synthesis gel system implemented, resulted into the in situ formation of mesoporous (Al-MCM-41) and microporous (ZSM-5) structures. The characterization techniques employed confirm the formation of both, the mesoporous and microporous phases.

With n-hexadecane as the feed, the activity obtained with the A1-MCM-41/MF1 mixed systems was greater than that obtained with conventional zeolite Y and pure MCM-41 material.

With respect to the selectivity, the mixed systems produced more gaseous products than liquid

201 products, along with a negligibly low yield of Ci and C2 fractions. The presence of Al-MCM-41 in the mixed system, assists the formation of more unsaturated products, whereas the ZSM-5 content of the mixed systems promotes the hydrogen transfer reactions and thus results in the improved yield of saturated products, including the saturated isomers. The yield of unsaturated propene and butene is boosted by the presence of Al-MCM-41 in the mixed system, which can be further used in the alkylation units or in the production of octane boosters like MTBE and

TAME. Finally it can be stated with certitude that the as synthesized mixed A1-MCM-41/MFI

meso-microporous catalyst systems possess a high potential in hydrocarbon cracking activity. In

order to explore these catalysts for the commercial FCC applications, additional investigations

related to hydrothermal stability and metal resistance should be tested in near future.

Acknowledgement

The authors thank the Deutscher Akademischer Austauschdienst (DAAD) for the

financial support, for the work carried out at the Institute for Industrial Chemistry of the Dresden

University of Technology.

References

[1] Sie, S. T, Awd Ca&z/., 95, 587 (1994). [2] Kresge, C. T., Leonowicz, M. E., Roth, W. 1, Vartuli, J. C., Beck, J. S , AWure, 359, 710 (1992). [3] Corma, A., Grande, M. S., Gonzalez-Alfaro, V., Orchilles, A. V., J. Catal, 159, 375 (1996). [4] Koch, H, Reschetilowski, W., Microporous andMesoporousMaterials, 25, 127 (1998) [5] Karlsson, A., Stocker, M., Schafer, K., Stud. Surf. Sci. Catal., 125, 61 (1999). [6] Klemt, A., Reschetilowski, W., Microporous and Mesoporous Materials, submitted. [7] Koch, H., Reschetilowski, W., Proceedings ofDGMK Conference C4chemistry -

Manufacture and use of C4 hydrocarbons, Aachen, p.197 (1997). [8] ASTM-D-3 907-92. [9] Karlsson, A., Stocker, M., Schmidt, IS.,Microporous and Mesoporous Materials, 27, 181 (1999).

202 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

T. Donauer, R. Glaser, J. Weitkamp Institute of Chemical Technology, University of Stuttgart, Germany

Non-Oxidative Propane Dehydrogenation over Supported Pt-Zn-Catalysts

ABSTRACT

The dehydrogenation of propane to propene has been studied in the gas phase at 555 °C over bimetallic Pt-Zn-catalysts. With 0.5 wt.-% Pt and 4.0 wt.-% Zn, the influence of the support material properties, i.e., acidity/basicity and redox activity was investigated. Silica gel was found to be the best support material providing highly selective catalysts with almost negligible deactivation. Whereas catalysts with more acidic carriers favored aromatics formation and propane cracking, those based on strongly basic or redox active supports were almost completely inactive. The amount of carbonaceous deposits formed during the reaction was more pronounced over a Zn-promoted catalyst than with a Sn-containing analogue and strongly depended on the catalyst support. A variation of the platinum and the zinc content for silica- supported catalysts had little effect on the propene yield (which was controlled thermodynamically) and affected the rates of by-product formation to a minor, but noticeable extent.

INTRODUCTION

In recent years, the demand for propene as a basic chemical has been continuously growing, mainly as a result of an increasing polypropylene manufacturing capacity . Conventionally, propene and ethylene are produced as the major products of steam cracking, in most cases with naphtha or alkanes from wet natural gas as the feed. Since the steam cracker supplies a multi-component product, catalytic dehydrogenation of low alkanes is a particularly attractive route to olefins, if these are to be produced with higher purity and if alkane-rich feeds are available for processing. For instance, a 350,000 t/a propane dehydrogenation plant has been announced by Lurgi to be built in Tarragona, Spain ,2 and two further plants for propane dehydrogenation are planned in Saudi Arabia and to be completed in the near future 3. In contrast to oxidative alkane dehydrogenation, the non-oxidative route is an endothermic, equilibrium-limited reaction. The advantages of non-oxidative catalytic dehydrogenation include recovery of the produced hydrogen, no loss of feed by partial or total oxidation to oxygenates or carbon oxides and easier process control due to the absence of oxygen, i.e., explosion risk, and the easier handling of the endothermic reaction heat. Currently available processes for non-oxidative lower alkane dehydrogenation rely predominantly on platinum-based catalysts, e.g., Pt-Sn/y-AbOs (UOP Oleflex process) or Pt-Sn supported on ZnAI 204 or MgAI204 (Phillips STAR process), or on

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 203 C12O3/Y-AI2O3, e.g., in the Lummus Catofin and the FBD process by Snamprogetti 4. Numerous investigations have been devoted to the improvement and understanding of supported platinum catalysts, especially with tin as the promoter component, e.g., refs. . Although zinc has been mentioned, among others, to be an effective promoter for platinum-based catalysts8 "10, only little attention has been paid to the bimetallic system of platinum and zinc as a catalyst for dehydrogenation of propane or other lower alkanes. Recently, zinc and platinum on zeolite NaX and NaY have been used for the selective hydrogenation of 2-butenal 11 or for the liquid-phase hydrogenation of cinnamaldehyde 12. It was the aim of the present study to further investigate the suitability of supported Pt-Zn-catalysts for propane dehydrogenation in the gas phase. First, catalysts based on support materials with different acid/base and redox properties were tested. Secondly, the influence of the content of the noble metal and the promoting metal on the catalyst with silica as the carrier material was examined.

EXPERIMENTAL SECTION

The catalysts were prepared by a two-step impregnation of the carrier materials, first with a solution of 1.4 g Zn(CH 3COO)2 • 2 H20 in 100 cm3 of water per 10 g of carrier. After calcination in a nitrogen atmosphere at 540 °C for 14 h and in air for 4 h, another impregnation step followed with an aqueous solution of 94 mg

[Pt(NH3)4]CI2 ■ H20. The support materials used were Si02 (silica gel 60, Fluka),

Si02-Al203 (n s/n Ai = 2.6, Akzo Research), y-AI203 (Merck), MgO (Merck), CaO, BaO, ZnO, Mn 203, V2Os, MoOs, W03 (Alfa Aesar), Zr02 (Merck) and Cr203 (Merck). For catalysts with different loadings of platinum and zinc, the amount of precursor applied in the impregnation steps was adjusted at approximately constant ratio of water and mass of the carrier material. The chemical composition and the weight during the activation of the catalysts were determined by atomic emission spectrometry with inductively coupled plasma (Plasma 400, Perkin Elmer) and by thermogravimetric analysis (Setsys 1600, SETARAM). The resulting catalysts are labeled according to their metal loading, e.g., 0.5Pt-3.8Zn/Si0 2 for a catalyst containing 0.5 wt.-% platinum and 3.8 wt.-% zinc supported on silica. Before the catalytic experiments, the powders were pressed without a binder, crushed and sieved to a particle size between 0.2 and 0.4 mm. Prior to use, the catalysts were activated in flowing air (110 cm3/min) at 400 °C for 16 h and, subsequently, in flowing hydrogen (110 cm3/min) at 555 °C for 4 h. The catalytic experiments were carried out at ambient pressure in a flow-type apparatus with a tubular fixed-bed reactor made of quartz glass. Pure propane gas (99.95 %, Linde) served as the feed. Samples were taken periodically from the reactor effluent and analyzed by gas chromatography using a flame ionization detector (FID). After the experiments, the mass of carbonaceous deposits on the catalysts was determined by thermogravimetric analysis (vide supra) and by elemental analysis for C and H (Vario EL, elementar). Yields are reported based on the number of carbon atoms in the product molecule zc,\ according to

^ i.out rlj in ) 3 Y,= ^propane.in ^c,|

204 RESULTS AND DISCUSSION

Influence of the zinc promoter The results of a typical conversion of propane over O.SPt-S.SZn/SiOz are shown in Figure 1 (left part). After ca. 2 h time-on-stream, the propane conversion remains virtually constant, and a high selectivity for propene (> 90 %) is achieved. Within ca. 5 d on-stream the propane conversion over the same catalyst decreases only slightly (not shown), viz., from initially 27 % to 22 %, without any loss of the selectivity for propene.

—i------1------1------1------1------1------1------O 58 oOOOOOOOOOOOOOO- 0.5Pt-3.8Zn/SiO, 3000000000 0^°! O Q z> 60 St * X (propane) CF 40 O S (propene) w o 20 8 « „

• methane a C4 - C6 A ethane o benzene A A A aa aaaa aaa ▼ ethene

O O o O O O O O 0

TIME-ON-STREAM / h

Figure 1: Conversion of propane, selectivity for propene (upper part) and yields of the major by-products (lower part; C4 - C6: aliphatic hydrocarbons) in the conversion of propane over O.SPt-S.SZn/SiOz (left part) and O.SPt/SiOa (right part) (TR = 555°C, p R = 1 bar, mcat. = 400 mg, W/FPr0p ane = 13 g-h/mol).

The propene yield corresponds to that expected for thermodynamic equilibrium at the reaction temperature applied (25 %). The major by-products are methane and ethane from hydrogenolysis of propane by hydrogen from the dehydrogenation reaction and benzene from dehydrocyclodimerization of propane. Ethene from catalytic propane cracking and C4- to C6-hydrocarbons (both olefinic and paraffinic) are formed in minor amounts only. It should be noted that a propane conversion of ca. 3 % with a propene selectivity of ca. 50 % and yields of methane and ethene of 0.5 % and 1.0 %, respectively, are observed for the thermal reaction, i.e., without a catalyst in the reactor. In the absence of zinc, the catalyst deactivates from the beginning of the

205 experiment on with a concomitant decrease of the selectivity for propene (Figure 1, right part). The drop of the ethane yield and the steadily increasing yield of ethene indicate that propane cracking initially occurs by hydrogenolysis, but, at higher times- on-stream, takes place with less involvement of hydrogen. This may be due to the decreasing activity of the catalyst for dehydrogenation and, thus, a reduced supply of hydrogen for the hydrogenolysis reaction. The comparison of the results obtained over the two catalysts indicates that the promoter function of zinc is similar to that of tin in that the deactivation of platinum sites is reduced and C-C bond cleavage reactions are moderated by less strongly adsorbed hydrocarbons on the catalyst surface. As in the case of tin, this may be explained either by an electronic or an ensemble effect 4.

Influence of the support material With y -AI203 as the support material, the catalyst is subject to continuous deactivation as shown in Figure 2 (left part). Since, after 2 h time-on-stream, the propene selectivity remains virtually constant and the propane conversion decreases steadily, this deactivation influences the dehydrogenation and the reactions leading to the by ­ products in a similar manner. As in the conversion over O.SPt/SIOg (Figure 1, right part), the deactivation is accompanied by an increase in the ethene yield at the

tr 100 T------1------1------1------1------1------1------1------1------oooooooooooooooooc 80 - O 0.5Pt-4.0Zn/Y-AI 203 0.5 Pt-4.0Zn/MgO x cn 3 - O, 60 -- °°oooooo Ob OOOOOOO £E F 40 • X (propane) LUO O S (propene) ly 20 8% 0

Figure 2: Conversion of propane, selectivity for propene (upper part) and yields of the major by-products (lower part; C4 - C6: aliphatic hydrocarbons) in the conversion of propane over O.SPHt.OZn/y-AlaOs and 0.5-4.0Zn/MgO (reaction conditions as in Figure 1).

206 expense of that of ethane. Over catalysts supported on strongly basic oxides such as MgO, CaO, BaO and also ZnO, the propane conversion drops within less than 2 h time-on-stream to 3 to 4 %, and the product distribution approaches that observed with an empty or sand-filled reactor (vide supra). Interestingly, a deactivation is also observed with a silica-based catalyst which has been treated with an aqueous solution of CsOH after activation of the metal precursors in air in order to remove acidic silanol groups 13. Over this catalyst, the rate of propane cracking as reflected by the yield of methane remains constant during the experiment, and the deactivation affects the dehydrogenation reaction and, consequently, the yield ratio of ethane and ethene only (Figure 3, left part). Due to the strongly reduced density of acid sites almost no benzene is found in the product and the formation of coke is slow (vide infra). With Si02-Al203 as a more acidic support, the propane conversion and the propene selectivity are comparable to those obtained with the silica-based catalyst in the beginning of the experiment. However, at longer times-on-stream both the conversion and the selectivity for propene decline significantly (Figure 3, right part). At the same time, the activity of the catalyst for propane cracking increases. This is in accordance with earlier observations that the dehydrogenation selectivity of platinum catalysts is lower, if the support materials exhibit too strongly acidic properties 4. In the conversion of propane over noble metal catalysts consecutive oligomerization,

too —i—'—i—i—i—i—i—i—r O 55 >oooooooooooooooooooooooj OOOOOOOOOOOOOOOOOOOOOl 80 oo x co 0.5Pt-3.8Zn/SiO 2, . 60 treated with CsOH - II • X (propane) 40 O S (propene)

20 OUJ O<0 0 3

• methane ■ c4 - c$ A ethane O benzene ▼ ethene AaAA. AAAAAAAAAAAAAAAAAAA 1

x 1 * VVVVVWTV

18 0 6 12 18 24 TIME-ON-STREAM / h

Figure 3: Conversion of propane, selectivity for propene (upper part) and yields of the major by-products (lower part; C4 - C6: aliphatic hydrocarbons) in the conversion of propane over 0.5Pt-3.8Zn/SiO 2 (treated with CsOH solution) and 0.5Pt-5.2Zn/SiO 2-AI2O3 (reaction conditions as in Figure 1).

207 ring closure and dehydrogenation reactions may lead to the build-up of coke on the catalyst surface, either on the metal or on the support. The amounts of carbonaceous deposits formed in the propane conversion over platinum-zinc catalysts with different acidic/basic supports are reported in Table 1. (Note that the data are given for different times-on-stream.) Generally, a higher loading of coke is obtained for catalysts with acidic than with basic supports. However, a clear correlation between the amount of coke deposited on the catalyst and the activity or the stability of the catalyst for dehydrogenation is not existing. This is particularly obvious from the lower coke loading, but better stability of 0.5Pt-3.8Zn/SiO 2 as compared to the Si02- Al203-based catalyst and by the similarly low coke loading on the strongly deactivating zinc-free catalyst 0.5Pt/SiO2 and the silica-based catalyst treated with CsOH solution which deactivated rather slowly (Figure 3, left part). Likewise, a silica- supported catalyst with tin as the promoter deactivated slowly, but did not accumulate as much coke as the zinc-promoted counterpart under the same conditions. In other words, the stability of the catalyst with platinum-zinc is superior to that with platinum-tin as the active metal phase.

Table 1: Amount of coke formed on different catalysts during the conversion of propane (reaction conditions as in Figure 1).

catalyst time-on-stream / h coke loading / wt.-%

O.SPt / Si02 14 2.0 0.5Pt—5.2 Zn / Si02-Al203 24 24.0

0.5Pt-3.8Zn/SiO 2 17 18.2

O.SPt—3.2 Sn / Si02 22 0.8

0.5Pt-3.8 Zn / Si02 (treated with CsOH solution) 63 1.7 O.SPt—4.0 Zn / y-AI203 47 13.5 O.SPt-4.0 Zn / MgO 18 6.6 sand 15 0.0

The observed influence of the acid/base properties of the support on the catalyst stability may be explained in terms of adsorption of coke and coke precursors. These compounds may be formed on the active metal sites and lead to their deactivation. If the support is capable of adsorbing the coke precursors, e.g., due to its acidic properties, they may be transferred from the metal sites to the surface of the support. Hence, the metal catalyst exhibits improved stability. Similar to catalysts with basic oxides as carrier materials, those based on redox active supports, e.g., W03, Mn 203, V205, M0O3, Cr2Os deactivated rapidly and were fully inactive towards propane dehydrogenation. Beside an unfavorable dispersion and an influence of the support on the electronic properties of the active metal phase, a lack of porosity of the support materials may be responsible for the low activity of these catalysts. An important role of the support material is to stabilize the platinum dispersion, especially at the occasionally harsh conditions during coke burn-off in oxygen or air. Although platinum is known to sinter during oxidation treatment on silica4, the activity

208 and selectivity of a 0.5Pt-3.8Zn/SiO 2 catalyst remained nearly unchanged when treated in air at 400 °C for 2 h (reduction of the coke loading to 1.0 wt.-%) after 17 h time-on-stream in the propane conversion at 555 °C.

Influence of the active metal phasecomposition In an attempt to reduce the rate of by-products formation, particularly that of coke, the zinc and platinum contents on the catalyst with silica as the carrier were varied. As shown in Figure 4, an increase of the zinc content above 0.5 wt.-% leaves the propane/propene ratio achieved in the stationary stage after 7 h time-on-stream almost unaffected, but brings about a slight decrease in the overall fraction of by-products formed. This corroborates the earlier conclusion that propane dehydrogenation is rapid and controlled by thermodynamic equilibrium. The zinc-free catalyst produces significantly more by-products, if compared at the same conversion (vide supra , Figure 1). The platinum-zinc composition on these catalysts is in a range that allows the existence of intermetallic phases, also at reaction temperature 14. It has, however, been reported that platinum-zinc alloys exhibit only low dehydrogenation activity9,15 . Neither the propane conversion nor the product distribution are strongly affected upon increasing the platinum content of a silica-supported catalyst with 4.0 wt.-% zinc from 0.5 wt.-% to 2 wt.-% (not shown). In contrast to these catalysts, however, a catalyst with only 0.1 wt.-% platinum is subject to slow deactivation indicating that a minimum platinum content is necessary to obtain a stable catalyst under the present reaction conditions. However, a platinum-free 3.9Zn/Si0 2 catalyst does not show any dehydrogenation activity.

left bar: ■■■ propane i i propene m by-products

right bar: ezzzz3 methane ethane ethene izzzza benzene others

Figure 4: Product distribution (left bar) and fraction of by-products (right bar) in propane conversion over catalysts containing 0.5 wt.-% Pt and different amounts of zinc supported on silica (T = 555 °C, WHSV = 3.5 h" 1). Data are given for 2 h time-on-stream for 0.5Pt/SiO2 and for 7 h time-on-stream for all other catalysts (cf. text).

209 CONCLUSIONS

Bimetallic platinum-zinc-catalysts supported on oxides with low acidity such as silica gel are attractive and stable catalysts for the non-oxidative dehydrogenation of propane to propene and, presumably, of other lower alkanes as well. Although coke formation is more severe than with tin-supported analogues, the platinum-zinc- catalysts exhibit improved stability at extended time-on-stream and may be regenerated without significant loss of catalytic activity. Zinc may, therefore, be a cost-effective alternative to tin as a promoter for platinum-based dehydrogenation catalysts. A further reduction of the consecutive reactions leading to coke may be achieved by adjusting the modified residence time, and an improvement of the propene yield may be obtained by shifting the thermodynamic equilibrium with the reaction temperature. An optimization of these and other reaction parameters is currently under way in our laboratory.

REFERENCES 1. Thayer, A.M., Chem. Eng. News, 78 (11), 19 (2000). 2. CIT plus, 3 (4), 18 (2000). 3. Tullo, A.H., Chem. Eng. News, 79 (12), 18 (2001). 4. Buonomo, F., Sanfilippo, D., Trifird, F., in: Ertl., G., Knotzinger, H., Weitkamp, J. (eds.), Handbook of Heterogeneous Catalysis, Vol. 4, Weinheim, VCH, p. 2140(1997). 5. Llorca, J., Homs, N., Leon, J., Sales, J„ Fierro, J.L.G., Ramirez de la Piscina, P., Appl. Catal. A: General, 189, 77 (1999). 6. Cortright, R.D., Levin, P.E., Dumesic, J.A., Ind. Eng. Chem. Res., 37, 1717 (1989). 7. Margitfalvi, J.L., Borbath, I., Tfirst, E., Tempos, A., Catal. Today, 43, 29 (1998). 8. European Patent 351067 B1, May 6, 1992, assigned to The British Petroleum Company, (Inv.: Barri, S.A.I., Tahir, R.). 9. Pakhomov, N.A., Buyanov, R.A, Moroz, E.M., Kotelnikov, G.R., Patanov, V.A, React. Kinet. Catal. Lett., 9, 257 (1978). 10. Chen, Z.X., Derking, A., Koot, W., van Dijk, M.P., J. Catal., 161, 730 (1996). 11. Silvestre-Albero, J., Coloma, F., Sepulveda-Escribano, A., Rodriguez-Reinoso, F., in: Galarneau, A., Di Renzo, F., Fajula, F., Vedrine, J. (eds.), Zeolites and Mesoporous Materials at the Dawn of the 21st Century, Proceedings of the 13th International Zeolite Conference, Montpellier, France, July 8 - 13, 2001, Studies in Surface Science and Catalysis, Vol. 135, Amsterdam, Elsevier, 2001, p. 314 and full paper No. 29-P-15 on accompanying CD-ROM. 12. Blackmond, D.G., Oukaci, R., Gallezot, P., J. Catal., 131,401, (1991). 13. Schenk, U., Ph. D. thesis, University of Stuttgart, 2001, in preparation. 14. Massalski, T.B., Murray, L., Bennet, L.H., Baker, H„ Kacprzak, L., Binary Alloy Phase Diagrams, Vol. 2, Metals Park, Ohio, American Society for Metals, p. 1923(1986). 15. de Miguel, S.R., Jablonski, E.L., Castro, A.A., Scelza, O.A., J. Chem. Technol. Biotechnol., 75, 596 (2000).

210 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

0. Hein*, A. Jess** *lnstitut fur Technische Chemie und Makromolekulare Chemie, RWTH Aachen, Germany, “Lehrstuhl fiir Chemische Verfahrenstechnik der Fakultat fur Angewandte Naturwissenschaften, Universitat Bayreuth, Germany

Heterogeneous and Homogeneous Processes in Oxidative Dehydrogenation of Propane in a Fixed Bed Catalytic Reactor

1. Introduction Olefines like ethylene and propylene are traditionally produced by (thermal) steam cracking or co-produced in Fluid Catalytic Cracking (FCC-process). To overcome the increasing pro ­ pane demand (compared to ethylene and other olefines), also non-oxidative catalytic proces ­ ses were developed, e. g. for propane dehydrogenation. Thus, the catalyst has to be regene ­ rated from time to time by burning off the coke deposits that cause a drop in activity of the catalysts. As a consequence, a frequent regeneration and a complicate reactor design are re ­ quired. In addition, the heat needed for the endothermic dehydrogenation has to be intro ­ duced externally either by external heating, e. g. in case of steamcracking, or by pre- or inter ­ stage heating of the reactants in case of catalytic dehydrogenation. An alternative concept is the oxidative dehydrogenation (ODH) of light alkanes [1 - 5], In case ot autothermal reactor operation, the heat required for the endothermic dehydroge ­ nation is produced internally by the oxidation of a small part of the alkane and/or - as it will be discussed in this paper - by the oxidation of components like carbon monoxide or hydrogen, which are added to the feed. According to literature, a wide variety of catalysts for OCD have been investigated, whereas the number of studies on reaction engineering aspects is comparatively small. In the present study the influence of the process conditions (above all temperature and feed gas composi ­ tion including the addition of CO and H2) on the ODH of propane was studied in a tubular fixed bed reactor (Pt/Ni-catalyst; Sudchemie). To determine the contribution of homogeneous reactions additional experiments were done in an empty tube reactor. The focus of the study was not to investigate detailed mechanisms of the numerous reactions, but to determine the main pathways and the optimal process conditions with respect to olefin yield and selectivity in a catalytic fixed bed reactor. The following main pathways of ODH have to be differentiated: 1. heterogeneous ignition, I. e. oxidation of C3H8 and/or CO and H2 (if added to the feed), 2. catalytic (and depending on temperature homogeneous) dehydrogenation and cracking, 3. syngas formation by reaction of hydrocarbons with H20 or C02 (formed during ignition), 4. coke formation in the gas phase as well as on the catalyst, and 5. side reactions like methanisation and water gas shift reaction. 2. Experimental For the experiments an electrically heated continuous flow reactor (quartz) was used, which was either empty (thermal experiments) or filled with the catalyst (fixed bed). The propane conversion and the product yields are calculated based on the product analysis (gaschroma- tograph, on-line analysers for CO, COa, 02, CH4, H2). Depending on reaction conditions, coke is formed. In case of the experiments on thermal conversion in an empty tube reactor, the coke, which is formed in the hot zone in the middle of the tube, deposits on the reactor wall

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 211 and to a larger extent on a bed of quartz sand in the lower and cooler part of the reactor. In case of the experiments with the catalyst, coke formation leads to a carbon load of the catalyst. In both cases, the coke was combusted by an oxygen/nitrogen-mixture at the end of most of the experiments; the amount of coke (and thus the yield obtained during the thermal or catatytic dehydrogenation) is calculated planimetrically from the gas analysis (CO, C02) and the volume rate of the off-gas. In general, a mass balance of 95 % or better was obtained. The total pressure was always 1.4 bar, and the catalyst (Sudchemie, Pt-Ni-Kat. Typ G-43; 190 m2/g) was used with a particle diameter of about 0.25 mm. In case of the experiments with the empty tube reactor, a pronounced temperature profile with a heating and cooling sec ­ tion can not be avoided. To determine the residence time, the effective reaction volume was calculated (for details see e. g. [6]), which is defined as the volume necessary to reach the same conversion at a constant reference temperature (e. g. maximum temperature in the middle of the reactor) as under the non-ideal isothermal conditions of the experiments.

3. Thermodynamic analysis of OCD of propane Regarding the (partial) equilibrium of oxidative propane dehydrogenation, two extreme cases have to be distinguished, at least theoretically: Case A (Fig. 1): In the most unfavourable case the oxygen is exclusively consumed by the partial oxidation of propane to CO (and hydrogen). Thereby, relative much oxygen is needed to reach the required temperature for the dehydrogenation with respect to thermodynamics (and also with respect to kinetics). In addition, the hydrogen produced by dehydrogenationis not consumed, which is also not favourable with respect to the equilibrium of propane dehy­ drogenation. As shown in Fig. 1, the maximum yield ofpropene is limited to only 30 %.

c I

0>I

<

Fig. 1: Partial thermodynamic equilibrium of OCD of propane in case of exclusive 02 consumption by oxidation to CO und H2 (no formation of C02 or H20) (pressure: 1 bar; temperature reactor inlet: 300 °C; calculated with the computer program Mintal, developed at the Engler-Bunte-lnstitute, University Karlsruhe)

Case B (Fig. 2): The optimal case is an oxygen consumption without any CO/CO2-formation, i. e. only steam is produced. In this case, the yield of propane is not limited (Fig. 2). Thereby must be stated that this case is unrealistic, as propane will not primarily be dehydrogenated before a selective oxidation of hydrogen takes place, at least not in case ofaH 2-free feed gas.

212 100 1900

0 4------1------1------300 0,1 0,3 0,5 0,7 Oz - Addition in mol/mol C3He Fig. 2: Partial thermodynamic equilibrium of OCD of propane in case of exclusive 02- consumption by H2-oxidation (no CO or C02) (conditions: see Fig. 1) in case of a real technical process, the propene yield will be in between these two cases, and will depend on the reaction kinetics and the design and operation of the reactor. Nevertheless, one important indication is provided by the thermodynamic analysis of the adiabatic OCD of propane: The addition of hydrogen and/or CO to the feed gas (e. g. by recycle of H2) could leed to suppression of the oxidation of propane und thus to an increase of the propene yield. 3. Results and discussion 3.1 Thermal reactions during OCD First of all, the thermal dehydrogenation of propane was investigated to determine the contri ­ bution of thermal reactions under the conditions of OCD. Compared to the subsequent cataly­ tic experiments, the residence time was quite large to reach a measurable propane conver ­ sion within a comparable temperature span. At a residence time of 0,7 s, the propene yield is limited to about 20 % (800 °C, 55 % conversion); coke formation can be neglected for T < 800 °C (Fig. 3).

-C3H8

C2H4

Coke C3H6 Y,

800 Temperature in °C Fig. 3: Thermal dehydrogenation/conversion of propane (1,4 bar; feed gas (vol.-%): 11 % propane, rest nitrogen; residence time (800 °C): 0,7 s)

213 A higher conversion (or temperature) leads to a decrease of the propene yield by formation of ethene and methane. Experiments with addition of H2 and H20 clearly indicate that the conversion and the product selectivities are practically not changed; only coke formation is slightly suppressed by addition of hydrogen. The sum of the yields of CO and C02 formed by reforming of the hydrocarbons with steam were max. 1 % (850 °C).

3.2 Catalytic dehydrogenation of propane To test the „pure “ dehydrogenation activity of the catalyst SODCHEMIE, Pt-Ni-cat. G-43 190 m2/g, particle diameter (if not stated differently): 0.25 mm), experiments with the catalyst were done without oxygen addition. For a residence time (empty reactor) of 0.08 s mainly propene is formed for T < 650 °C respectively a conversion <10% (Fig. 3). For higher tempe ­ ratures the selectivity decreases to about 80 %, which is the result of methane and ethene formation. So for T > 650 °C shorter residence times are needed to improve the propene selectivity.

40 - -

Temperature in °C Fig. 4: Influence of temperature on the catalytic dehydrogenation of propane (1,4 bar; 607 °C; reaction time: 50 min; feed (vol.-%): 10 % propane, 10 % H2, rest N2; residence time (600 °C): 0,08 s)

If a fresh catalyst charge is used, coke is formed at the beginning of the experiment, which leads to a decline of the reaction rate and propane conversion. Coke formation is suppressed by hydrogen addition to the feed, but has no effect on dehydrogenation, at least, if the distance to the equilibrium conversion is high enough. After a reaction time of about 1 h, coke formation comes to a stand still, at least for the period of the experiments, i. e. 10 h. During OCD, steam is formed by oxidation. Therefore, the influence of steam on dehydrogenation was also investigated. As shown in Fig. 6, steam has a strong effect: The conversion of propane increases, and mainly syngas is then formed.

214 tendency

100 200 300 400 500 600 Reaction time in min Fig. 5: Influence of reaction time on the catalytic dehydrogenation of propane (607 °C; residence time (600 °C): 0,16 s; other conditions as in Fig. 4)

Sqo+co2 (above all CO)

‘C3H8

’C2H4

'C3H6

Temperature in °C Fig. 6: Catalytic conversion of propane in the presence of H2D (1,4 bar; feed (vol.-%): 11 % propane, 10 % H2, 49 % HzO, rest N2; residence time (600 °C): 0,14 s)

3.3 Catalytic oxidation and ignition behaviour of the catalyst Ignition of the feed is important for a stable operation of an autothermal OCD-reactor. Tests with different gases (about 2 % in air) by the ignition point method [9] showed the following order of reactivity: methane « propane < CO < H2. Thereby must be stated that in case of CO/ H2-mixtures, CO is preferentially oxidized. Catalytic oxidation was also investigated under isothermal conditions (2 % 02, 0.11 s, 200 - 600 °C). The results pictured in Fig. 7 by the change of the carbon distribution with tempera ­ ture, can be summarized as follows: - Compared to H2, CO is preferentially oxidized for T < 200 °C; at higher temperatures, the oxidation rate for both components is almost the same (not shown in Fig. 7). - Propane alone is oxidized at temperatures of about 200 °C. Addition of CO leads to selec ­ tive CO-oxidation, at least for T < 500 °C.

215 - For temperatures between 300 and 520 °C, the methanisation reaction takes place, which was also proven by separate experiments with a mixture of CO and H2 as feed gas. - For T > 500 °C, propane dehydrogenation starts; parallel to this .target" reaction, also reforming of the hydrocarbons with steam to CO and H2 has to be taken into account, after all for relative long residence times of more than 0.1 s (Fig. 7).

100 - S? 90 - o 80 - c c 70 ■ O 60 ■ 3 C3H6' •a 50 - % 40 - *o c 30 - c,h8 1 20 - o 10 -

390 460 530 580 Temperature in °C Fig. 7: Oxidative catalytic conversion of propane (1,4 bar; feed (vol.-%): 10 % propane, 11 % CO; 10 % H2, 2,2 % 02; rest H2; residence time (500 °C): 0,11 s; relatively high coke formation because of the short reaction time (6 min) (cp. Fig. 4)) 3.4 Oxidative catalytic dehydrogenation Based on the results described before, the following conditions were chosen for the final experiments of OCD of propane under non-isothermal conditions (10 % 02): - H2 and after all CO should be added to the feed. - The residence time should be less than 0.1 s, i. e. reaction temperatures higher than about 700 °C should be adjusted to reach a propane conversion of technical relevance. So experiments on OCD were done with addition of CO and H2 at a residence time of about 0.04 s. As the result of a relative high heat loss of the small quartz reactor, adiabatic con ­ ditions were not reached. Therefore, in all experiments the temperature profile was attenuated compared to autothermal operation. Nevertheless, the exothermic oxidation zone (increase of temperature in the front section of the fixed bed) and the zone of the endothermic dehydro ­ genation (decreasing temperature in the rear section) were clearly observable. (Remark: For the given reaction conditions, i. e. 0.04 s and Tmax = 750 °C, the propane conversion by ther ­ mal reactions is only about 1 %, as calculated based on the results of the experiments on ther ­ mal conversion.) Tab. 1 shows that the olefine selectivity is improved by the addition of hl 2 and after all by CO. Full Byconversion was always reached. For an almost constant propane con ­ version of about 20 %, the propene selectivity is improved from 27 % (no addition of CO or H2) to 30 % (H2-addition) and 38 % (CO-addition). The overall selectivity to olefines (propene, ethene and butenes) increases from 38 % to 46 % (H2-add.) and 52 % (CO-add.), respective ­ ly. For comparison, the following border case was also investigated (Tab. 1, no. 4): Only CO is oxidized, and then the dehydrogenation takes place. This case was simulated by feeding a mixture of CO, C02 and propane (without Oz) to the reactor, thereby adjusting a composition equivalent to a reaction mixture just after the preliminary and selective CO-oxidation. In this case a propene selectivity of 45 % and an overall olefine selectivity of 82 % are reached.

216 Tab. 1: Influence of the addition of CO resp. H2 on the OCD of propane (for comparison: border case of selective CO oxid. with subsequent dehydrogenation, simulated by addition of CO and C02 (no 02); reaction conditions: 1,4 bar; 700 °C +/-50 K; feed (Vol.-%): 1) 50 % propane, 10 % 02, 40 % N2; 2) 50 % propane, 10 % 02,40 % H2; 3) 50 % propane, 10 % 02 and 40 CO; 4) 55 % propane, 22,5 % CO and 22,5 % C02; residence time (750 °C): 0,04 s; CD: catalytic dehydrogenation)

C-Selectivity (with respect to propane) in % Xprop. O-Select. in % in % Carbon- Experiment Olefines Paraffines (X02 = oxides 100) C3H6 C2H4 c4+1 C2H6 ch4 CO C02 CO C02 H20

1. OCD in N2 18 27,1 10,9 0,3 0,8 7,8 40,8 12,3 54 32 13

2. OCD with 18 29,5 16 1 1 12,5 35,5 4,5 46 11 43 H2-addition

3. OCD with 17 37,7 13,9 0,4 0,9 8.6 4,4 31,9 12 77 11 CO-addition

4. CD with CO/C02-addition (no 02- 23 45 33 4 1,3 17 0 0 (simulation of select. CO- addition) oxid. + consecutive dehydr.)

1 Incl. small amounts of butanes and higher hydrocarbons.

Only about 1 mol 02 is required to provide (by oxidation of CO) the heat needed for the endothermic dehydrogenation of 4 mol propane (calculated without heat loss and without the heat needed to heat the feed gases to the inlet temperature). Therefore experiments with increased propane content were also conducted (Tab. 2). By this means a propene selectivity of 41 % and an overall olefine selectivity of 58 % are reached (Tab. 2, no. 7).

Tab. 2: Influence of the propane feed ratio on the OCD of propane with and without CO-addition (conditions: 1,4 bar; 725 °C +/-50 K; residence time (700 °C): 0,04 s)

Experiment Xprop. C-Selectivity (referring to Propane) in % O-Selectivity in % (feed gas in vol.-%) in % c3h6 C2H4 c4+ c2h6 CH4 CO co2 CO C02 H20 6. OCD with CO-addition 17 37,7 13,9 0,4 0,9 8.6 4,4 31,9 12 77 11 50 % C3, 40 % CO, 10 % 02

7. OCD with CO-addition 19 40,6 17,8 0,7 1,0 11,1 15,5 13,3 32 55 13 70 % C3, 20 % CO, 10 % Oz

217 3. Conclusions and outlook The presented results on oxidative catalytic dehydrogenation of propane indicate that for the given catalyst (Pt-Ni, SLID CHEMIE) thermal reactions do not contribute significantely to the overall conversion of propane for T < 750 °C. Addition of CO improves the OCD of propane: CO is preferentially oxidized, which corres ­ ponds to higher selectivities of propene and ethene. The best result optained by this means is a propene selectivity of 41 % and an overall olefine selectivity of 58 % for a propane conversion of 20 %. By optimizing the process parameters (temperature, residence time etc.), a further improvement up to values of 45 % and 82 %, respectively, are visible. This was demonstrated by converting a mixture of CO, C02 and propane (without 02), and thus simulating the border case of a preliminary and selective CO­ oxidation, before dehydrogenation starts. Further investigations will concentrate on experiments under adiabatic conditions, including long running experiments (cat. deactivation). In addition the following reactor concept will be tested: The reactor consists of two concentric tubes, whereby CO and 02 are fed to the inner pipe for catalytic oxidation. Subsequentely, dehydrogenation takes place in the annular gap between the two pipes after addition of propane and flow reversal. Heat is transferred through the wall of the inner pipe to avoid excessive high temperatures.

Acknowledgements Financial support by the Deutsche Forschungsgemeinschaft (Je257/1 ) is gratefully acknow ­ ledged (0326563A). The authors also wish to thank SOD-CHEMIE for supplying the catalyst.

Literature [1] Wolf, D. etal., Chem. Eng. Sci 56, 713 (2001). [2] Baerns, M., Buyevskaya, Erdol Erdgas Kohle 116, 25 (2000). [3] Grasselli, R. K.; Stern, D. L.; Tsikoyiannis, J. G., Appl. Cat. A 189, 9 (1999). [4] Grasselli, R. K.; Stern, D. L; Tsikoyiannis, J. G., Appl. Cat. A 189,1 (1999). [5] Sadykov, V. A.; Pavlova, S. N.; Saputina, N. F., Cat. Today 61,93 (2000). [6] Jess, A., Fuel 75,1441 (1996). [7] Patterson, M. J.; Angove, D. E.; Cant, N. W., Appl. Cat. B 26, 47 (2000). [8] Gunther, R.: Verbrennung und Feuerung. Springer-Verlag, Heidelberg 1984, 62/63. [9] Hein, O.; Jess, A.: Erdol Erdgas Kohle 116,18 (2000).

218 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

S. B. Kogan, M. L. Kaliya, N. Froumin, M. Herskowitz Ben-Gurion University of the Negev, Beer-Sheva, Israel

Selective Oxidative Catalytic Conversion of n-Butane into Olefins C2-C3 at Moderate Temperatures

Abstract The oxidative dehydrogenation of n-butane on Co and V deposited on -alu mina and silica was studied in the range of 500-550°C. Unexpected high activity of Co/AI203 in oxidative cracking of n-butane yielded ethylene and propylene whereas butylenes prevailed on other catalysts. XRD and XPS methods indicated existence of Co spinel and increased concentration of mobile oxygen that probably explains cracking activity of the catalyst.

Introduction. The annual worldwide demand of ethylene and propylene increases by 5% [1], A widely studied alternative for the production of light olefins is the oxidative dehydrogenation of paraffins (ODH). Thermodynamic limitations and complicated technology included periodical catalyst regeneration that is typical for traditional dehydrogenation process induced significant efforts for developing the novel selective catalysts for oxidation of hydrocarbons over the last decades. However analysis of results published over decades of intensive experimental studies indicates the need for highly selective catalysts for conversion of paraffins in oxidative process to light olefins with yield over 30% [2,3]. Technological innovations such as drastically shorting of contact time at 800-900°C [4] or staging of oxygen [5] yielded higher performance in ODH. However those methods render technology of olefins production very difficult. Olefins are formed over ODH via two main routes as follows : a) oxidative dehydrogenation of paraffins to corresponded olefins; b) oxidative cracking . While V, V-Mg, Mo, Sn-P, Ce-F, based catalysts provided only oxidative dehydrogenation route [2] Mg-Dy-Li-CI-O and Zr-Dy-Li-CI-0 catalysts give mainly light olefins [5-6]. Comparison between two groups of catalysts shows that oxidative cracking catalysts are significantly more selective than catalyst providing oxidative dehydrogenation. It could be caused by lowering of oxygen required in oxidative cracking comparing with oxidative dehydrogenation that depresses reactions of combustion. Oxidative cracking on known catalytic systems takes place at high temperatures over 600°C, practically in the ranges of temperature where steam cracking occurs. Operation temperature is very important, affecting catalyst stability and significantly restricting application of this process. Therefore the search of new catalysts that could be active in oxidative cracking of paraffins is done at comparatively low temperatures.

DGMK-Tagungsbericht 2001 -4, ISBN 3-931850-84-6 219 Cobalt oxide is well known as a very active oxidation catalyst [7-9]. In a preliminary stage it was found that cobalt oxide deposited on alumina provides oxidative cracking of n-butane over the low range of temperature 500-550°C and produces mainly ethylene and propylene. The next scope of this study is to determine the specific role of cobalt oxide compared with vanadium oxide in the oxidative dehydrogenation of n-butane. Experimental. The experimental rig consisted of a stainless tubular preheater and reactor in series, 17 mm ID and 250 mm long with 5.5 mm thermowell. Electric tapes controlled by Eurotherm heated the reactor and preheater. 1g of catalyst (granules with D= 1.5 mm and 2-3 mm length) diluted with 4g of inert SIC pellets to keep the reactor isothermal, was located between two layers of SIC particles of 3-4 mm in diameter. n-Butane, oxygen and nitrogen (>99% purity) were fed from cylinders. Brooks mass-flow meters controlled the flow rate. Condensers were staged in series after the reactor, both cooled to 3°C. Condensable organic compounds and water were condensed while other gases flowed to GC for analysis.The analysis of volatile reaction products was performed on line with GC HP-5890 that contained four columns: 45/60 molecular sieve 13X, 10 ft x 1/8"; Al203,50 m x 0.53 mm; 80/100 Hysep Q, 4 ft x 1/8" and 1 ft x 1/8" with internal switching valves and two detectors TCD and FID controlled by ChemStation analytical software. Analysis of water collected in the condensers displayed only traces of oxygenates. Blank experiment without a catalyst indicated conversion less than 1% at 550°C. Cobalt and vanadium catalysts were prepared by impregnation of y-AI203 (Norton SA -6175, S=250 m2/g, cylinders, fraction 3 x 1.5 mm) and Si02 (PQ, S=300 m2/g, cylinders, fraction 2 x 1.5 mm) by aqueous solution of Co(N03)2 6H20 or NH4V03 (Aldrich) with citric acid followed by drying 2 hr at 110°C and calcination 5 hr at 550°C. Co and V oxides were prepared by calcination of the same compounds at 550°C. Catalyst composition was measured with energy-dispersive X-ray detector (EDAX - JEM-35, JEOL Co., link system AN-1000, Si-Li detector). The surface area was determined using BET method (N2 adsorption, NOVA-1000; Quantachrome, Version 5.01). PHI 549 SAM/AES/XPS apparatus with double CMA and Mg K X-ray source (1253.6 eV) has been used for X-ray Photoelectron Spectroscopy (XPS) measurements of the catalyst surface. Phase composition was measured by XRD in conventional, automated Philips PW 1050/70 diffractometer equipped with long, fine focus Cu anode tube, 40 kW, 28 mA, a scintillation detector and a diffracted beam monochromater. Results and Discussion. Comparison of experimental results obtained at the constant initial molar ratio of oxygen to n-butane indicates significant effect of the support on the catalysts performance (table 1). Unsupported samples display low selectivity and main products of oxidation were carbon mono- and dioxide. Using of support for cobalt and vanadium oxides increased olefins selectivity and drastically affected the distribution of olefins. Cobalt and vanadium Si02 supported catalysts provide mainly formation of olefins via route of oxidative dehydrogenation and specific role of vanadium oxide consists only in higher content of butadiene among fraction of C4 olefins. Deposition of cobalt and vanadium on the alumina results in significant differences of routes of olefins formation. While V/Al203 yields mainly C4 olefins, Co/AI203 catalyst provides predominant formation of light olefins and ratio of olefins

220 (C2+C3)/C4 were 0.8 and 3.3 correspondingly on 2.7%V/AI2C>3 and 2.3%Co/AI203 catalysts. It means that n-butane on the Co/AI203 predominantly undergoes oxidative cracking.

Table 1 Effect of metal content on catalyst performance

T = 550 °C, molar ratio C4: P2: N2 = T.T.0.5, WHSV=22 h '1 Catalyst Conver­ Total u Ethylene Propylene Butenes Ratio % wt. sion olefins selectivity selectivity selectivity of % selectivity % wt % wt % wt olefins % wt (C2+C3)/ / C4 CO2O3 21.4 23.9 4.6 5.3 14.0 0.7 0.6 v2o5 13.4 5.0 0.5 3.9 0.3 0.7Co/ 12.9 78.7 21.0 9.6 48.1 0.6 /AI2O3 2.3Co/ 14.1 66.5 31.1 19.9 15.5 3.3 /Al203

10Co/AI2O3 19.6 44.2 18.4 15.3 10.5 3.2 0.1 0.7V/AI2O3 13.5 69.6 2.7 2.9 64.0 68.0 0.8 2.7V/ AI2O3 15.4 17.4 12.7 37.94 0.8 10V/AI2O3 23.9 58.6 14.9 11.4 32.3 zj~ 8.1 8.0 3.2Co/Si02 19.3 70.2 54.1 0.3 6.2 3.4V/ Si02 29.2 61.3 10.3 44.8 “) 0.4 1) Other products - mainly carbon oxides. 2) 45-50% of formed C4 olefins - butadiene.

Distribution of olefins significantly depends on conversion of n-butane. Fig.1 and 2 demonstrate that at low conversions, n-butane is mainly oxidized to butylenes on both catalysts while formation of light olefins is very limited. Increasing of conversion results in lowering of butylenes. However on the vanadium catalyst the route of oxidative dehydrogenation dominates over oxidative cracking up to conversion about 40%. On the contrary, formation of light olefins on Co/AI203 is significant at > 20% conversion.

Kinetic study of dehydrogenation on Co/AI203 indicated close to 1-st order on oxygen that agrees with work [10] as well as 1-st order on n-butane. However their influence on the distribution of olefins is quite different. Increasing n-butane pressure in the ranges of 0.1-0.5 atm at constant oxygen pressure has no effect on the ratio of light olefins to butylenes (fig.3). On the contrary increasing of oxygen pressure yields drastically increase of light olefins on Co /Al203 and practically has no effect on Co/Si02, V/Al203 and V/Si02. (fig.4). Light olefins, ethylene and propylene could be formed from n-butane via two different routes. One of them is oxidative cracking of butane, mentioned in a number of publications [4-6]. The other route for light olefins consists in decomposition of butylenes formed at oxidative dehydrogenation of n-butane. In this case preferable formation of light olefins on the Co catalyst at high n-butane conversion have to be explained by differences in the rate of destruction of butylenes on the Co and V catalysts. However oxidative conversion of 1-butene on V as well as on Co catalyst

221 yielded mainly butadiene with a limited formation of light olefins (table 2). Moreover no essential difference in distribution of light olefins in 1-butene conversion on both catalysts were revealed : ratio (C2+C3)/C4 was 0.23-0.30. It means that mechanism of light olefins formation via butenes could be excluded and oxidative cracking of n-butane remains the most probable route for production of ethylene and propylene.

" 60 -

> 50 - % 40 - Ethylene

Propylene a

B 10 - Butylenes

n-Butane conversion, %

Fig.1. Distribution of olefins on the 0.7% Co/Al203 T = 550 °C

; 60 - Butylenes ^ 50-

« 30- Ethylene <2 20

HI 10 - Propylene

n-Butane conversion, % Fig.2. Distribution of olefins on the 0.7% V/Al203 T= 550 °C

222 0 0.1 0.2 0.3 0.4 0.5 0.6 Pn.butane i atm Fig. 3. Influence of n.butane partial pressure on olefins distribution on catalyst 2.3% Co/AI203 T = 550 °C , P oxygen = 0.4 atm

2.5 -

1.5 -

3.2% Co/Si02 2.7% V/AI2O3

3.4% V/SiO; 0.5 -

oxygen , Fig.4. Effect of oxygen on ratio of cracking and dehydrogenation activities for Co and V catalysts. T= 500°C, Rbutane = 0.4 atm

Table 2 Conversion of 1 -butene

T = 550°C, molar ratio C4:Q2:N2 = 1:1:0 5 Catalyst Conversion Selectivity to, wt.% % Olefins Ethylene Propylene Butadiene 12.1 2.3%Co/AI203 7.5 84.6 7.5 65.0 6.8 86.0 6.8 2.7%V/AI203 ___ 9-7 69.5

223 Usually reactions of oxidative cracking of light parrafins proceed through formation of special activated oxygen species reacted with molecule of gaseous paraffin. Such mechanism was proposed for oxidative cracking on the Zr-Dy-Li-CI-O [11] and VgOs/TiOz [12] .Since activation energy values of cracking reactions are usually significantly higher than such values of oxidative reactions (by 2 times for cracking and oxidation of butane on Zr-Dy-Li-CI-0 [5]), reactions of cracking become relevant only at high temperatures. Catalytic oxidative cracking occurs at temperatures more than 630°C and 800°C for Zr-Dy-Li-CI-0 [5] and PVAI2O3 [4], and non-catalytic commercial steam cracking requires more than 750°C. The same distribution of olefins, ethylene propylene, butylenes were formed on the C0/AI2O3 at only 500- 550°C. Moreover at optimal conditions performance of C0/AI2O3 was comparatively high, and total yield of olefins reached 28-30% that is very close to the best values mentioned for oxidative dehydrogenation of light parafins [2], Usually a key factor for processes of oxidative cracking is a special adsorption and activation of oxygen. The data indicate that only on the Co/AI203 oxygen has a crucial impact on the light olefins formation (fig.4). Since no cracking were occurred without oxygen, the species of oxygen caused by interaction of Co and alumina were considered as the main factor for oxidative cracking. XRD analysis of cobalt containing catalysts indicates phase of C03O4 (fig.5, peaks about 2 =2 0°, 31° and 37°) and probably - phase of spinel CoAI204 (two last peaks). Evaluation of crystallinity degree (the corresponding samples calcined at 800 °C were assumed as 100 %) yields 55 % of Co oxide crystallinity in the tested samples. Average size of crystals - 18 nm. Crystallization with similar size of crystals was observed in 3.2 % of Co/Si02 sample. XPS plot of Co 2p (fig.6) shows 2 peaks : with B.E.=779.6 eV (C03O4) and with B.E.=783.8 eV attributed as spinel C0AI2O4 [13]. This peak is absent in the samples based on silica (fig.6). According to XRD spectrum of the sample 10%V/AI2O3 no phase of vanadia is detected (fig.5), i.e. maximal particle size does not exceed 4 nm. Also no vanadia crystals is observed in the silica based sample. XPS of vanadium supported catalysts (fig.6) indicates the only peak with B.E.= 516.7 eV in alumina and silica supported catalysts that is identified as vanadium in V205 [13]. Hence, an evident difference of the cobalt-alumina catalyst from all other catalysts tested is 2 species of cations - included in oxide and in spinel. Apparently that Co spinel structure induces an oxygen species different from oxides structures on the surface of Co/Si02, V/Si02, and V/AI2O3. Deconvolution of Ols peaks in XPS spectrum indicates some species of oxygen in alumina supported catalysts (table 3). Cobalt on alumina creates increased share of low energy ( B E. ~ ~ 530 eV) and more mobile oxygen that is less present in V/Al203 and does not exist in silica based samples . Thereby, C0/AI2O3 catalyst really differs from other catalysts tested due to formation of an additional species of cobalt and increased amount of mobile oxygen. Both factors are probably associated. The mobile oxygen species can be responsible for strong interaction with reactants and consequently - for cracking activity [14].

224 Table 3

XPS data on oxygen species in Co and V catalysts

Catalyst Oxygen (01s) distribution (%) depending on B E. (eV) 530-530.2 531.1-531.4 532.1-532.4 532.8 2.3%Co/AI203 25 45 30 0 10%Co/AI2O3 31 44 25 0 10%V/AI2O3 14 66 20 0 3.2%Co/Si02 0 0 0 100 3.4%V/Si02 0 0 0 100

Co304/CoAl204

10% V

y-Alumina

Fig.5. XRD analysis of alumina based catalysts

References 1. J.Weitkamp, A.Raichle.Y.Traa, M.Rupp, F.Fuder, Chem.Commun., 403 (2000). 2. F.Cavani, F.Trifiro, Catal. Today, 24, 307 (1995). 3. F.Cavani, F.Trifiro, Catal. Today, 51, 561 (1999).

225 4. M.Huff, L.Schmidt, J. Catal., 149, 1,127 (1994). 5. M.L.Kaliya, O.V.Malinovskaya, M.V.Landau, M.Herskowitz, P.F.van den Oosterkamp, Stud. Surf. Sci. Catal., 133, 113 (2001). 6. M.V.Landau, M.L.Kaliya, M.Herskowitz, P.F.van den Oosterkamp, P.S.Bocque, CHEMTECH, 24-29 February (1996). 7. J. Horacek, J.Korbl, V.Pechanec, Microchim. Acta, 2, 294 (1960). 8. G.Boreskov, B.Popov, V.Bibin, A.Kozishnikova, Kinetika i Kataliz (russ.), 9, 768 (1968). 9. Y.Shuurman, V.Ducarme, T.Chen, W.Li, C.Mirodatos, G.Martin, Appl. Catal., A, 163, 227 (1997). 10. N.Kijima, K.Matano, M.Saito, T.Oikawa, T.Konishi, H.Yasuda, T.Sato, Y.Yoshimura, Appl. Catal., A, 206, 237 (2001). 11. M.V.Landau, A.Gutman,. M.L.Kaliya, L.O.Kogan, M.Herskowitz, Stud. Surf. Sci. Catal., 110, 315 (1997). 12. D.Wolf, N.Dropka, Q.Smeikal, O.Buyevskaya, Chem. Eng. Science, 56, 713 (2001). 13. NIST X- Ray Photoelectron Spectroscopy Database , version 2.0. US Department of Commerce , National Institute of Standard and Technology. 14. E.A.Mamedov, V.Cortes Corberan, Appl.Catal., A, 127,1 (1995).

Co 2p 779.6 eV 516.7 eV

783.8 eV

10%V/A1,0 10% Co/Al,0

3,2%Co/SiO 3.4%V/SiO

Binding Energy, eV Binding Energy, eV

Fig. 6 XPS data

226 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

M. Langpape, G, Grubert, D, Wolf, M. Baerns Institute for Applied Chemistry Berlin Adlershof, Berlin, Germany

Catalyst Development for the Oxidative Dehydrogenation of Ethane and Propane to Olefins Applying an Evolutionary Approach

Abstract Catalyst optimisations for the oxidative dehydrogenation of ethane and propane have been performed using high-throughput approaches. In both cases the use of evolu­ tionary strategies led to a significant improvement of the catalytic performance of the catalytic materials.

Introduction The application of high-throughput approaches for the development of new or im­ proved catalytic materials is a growing field in heterogeneous catalysis. Since cata­ lytic performance of new materials cannot be predicted from fundamental knowledge only, experimental work is still necessary for catalyst development. High-throughput approaches which include design of experiments as well as parallel techniques for catalyst preparation and testing will have a positive economic impact on process de­ velopment, particularly due to the reduction of time in preparation and testing of new catalytic materials. In the present contribution the method of optimisation of such catalytic materials is illustrated for the oxidative dehydrogenation of propane (ODP) and ethane (ODE). Oxidative dehydrogenation of propane and ethane are interesting reactions since they may present an alternative to the highly endothermic thermal pyrolysis (see [1] and references therein). In the ODP reaction the highest propene yields which were generally found amount to approximately 20 to 25%. For the most intensively studied system, the magnesium vanadates, a propene yield of 24% was reported [2]. Other groups of these so-called redox-type catalysts which attracted attention are phos­ phates and molybdates of transition metals, but yields are generally lower than 15%. Besides, boria-aluminia catalysts [3] and rare-earth-catalysts [4] have been studied which resulted in yields up to ca. 20%. For the ODE reaction, different reaction con­ ditions have been investigated comprising (i) easily reducible metal oxide catalysts within a low-temperature range between 573 and 773 K and contact times from 0.1 to 10 s, and (ii) platinum or rare-earth metal oxide catalysts within a high-temperature

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 227 range between 1023 and 1273 K at contact times in the millisecond range. For the low-temperature conditions ethylene yields do not exceed 40% (eg.

Mo0.73V0.i8 Nb 0.09 Ox, Y(C2H4) = 37.7% [5]). In the high-temperature range catalyst- assisted homogenous gas-phase reactions generally play an important role and yields up to 60% can be obtained. Higher yields up to 69% have been achieved on a halide-containing catalyst [6], which is, however, undesirable in practice because of HX release and formation of chlorinated compounds. For an industrial application, ethylene yields over 70% at high ethylene selectivities are required. In earlier work we reported results for the OOP reaction on various catalyst composi ­ tions based on a large variety of different elements [7,8] for illustrating the high- throughput approach using a genetic algorithm for optimising the catalytic material. In

[8 ] the primary pool of elements contained V, Mg, B, Mo, La, Mn, Fe and Ga. It has been shown that already in the fifth generation the composition of the best catalytic materials were rather similar and a maximum propene yield of about 10% has been achieved at standard conditions. The final catalyst composition consisted of the ele­ ments V, Mg, and Ga besides minor amounts of Mo and Mn. By modifying catalyst preparation using MCM materials as support and varying reaction conditions yields up to 18% were obtained. The aim of the present work was to find new materials with improved catalytic per­ formance in OOP and in ODE. Since in OOP the known best performing materials include all vanadium it was excluded and other elements were used in the primary pool of elements: Ag, Ca, Cr, Cu, Fe, Mo, Mn, Ni, P, Sb, W and Zr. Similarly, also in the case of the ODE reaction no vanadium was considered; the primary pool com­ prises the oxides of elements with good redox properties, i.e. Co, Cr, Cu, Ga, Mn, Mo, Sn and W, with high activity towards oxygen adsorption, i.e. Ca, La, and with a high Me-O binding energy, i.e. Zr; finally, two promoters (Au, P) were included.

Methodology Three features are important for a rapid discovery of new materials with an interest­ ing catalytic performance: a useful strategy for the design of experiments, an accel­ erated preparation of materials and their fast testing for catalytic performance. In the last years, ACA developed an experimental design based on an evolutionary strategy described hereafter. It has been used for all optimisation procedures described in this contribution. For automated lab-scale catalyst preparation a commercially synthesis robot was used. Since it had been originally developed for pharmaceutical applica ­ tions it had to be adapted to the needs of preparing solid materials. For catalytic testing an in-house built 64-channel reactor module was used which has been de ­ scribed elsewhere in detail [9],

228 Evolutionary strategy The high number of material compositions and preparation variables in the develop ­ ment of solid catalytic materials leads to a combinatorial explosion of the parameter space which is a serious problem if no empirical and/or theoretical information on the structure-performance relationships of the materials are available, which is often the case. A rigorous combinatorial approach considering all possible combinations of parameters as well as methods based on factorial designs of experiments would re ­ sult in a tremendous experimental effort. Then, even with high-throughput tech ­ niques, the required test capacities can hardly be achieved. Evolutionary algorithms characterised by a self-adaptive, autonomic search are, however, expected to dimin ­ ish the experimental effort [10-15], Evolutionary algorithms use the concepts of mu­ tation, recombination and selection among populations of individuals. With their sto­ chastic feature and their population concept evolutionary algorithms do not need in­ formation on the topology of the multi-dimensional parameter space but can accu­ mulate information on it. This, in turn, yields the basis of further data mining (contin- gence analysis, regression, cluster analysis etc.) for extraction of usable knowledge and rules from a data pool which might be otherwise difficult to assess. Despite of the general set-up of the evolutionary algorithm, there are several design parameters which were adapted to the specific problem in the development of the ODP catalysts. These were: - Mode of encoding of composition and preparation parameters for each catalyst in the population: Here, a modular concept of encoding was suggested which allows the user to set-up a large variety of problem structures to optimise both catalyst composition and preparation procedure; - Mode of mutation and crossover: Suitable mutation and cross-over operators have to be chosen in order to avoid genetic noise in the later state of conver ­ gence. For this purpose, an analysis of the relationship between the number of mutation and crossover points on one hand and rate and certainty of convergence on the other hand was performed. - Mode of selection and reproduction: The heterogeneity of catalyst properties in a population is one of the most important factors determining the certainty of con­ vergence. Therefore, selection modes favouring reproduction of active and selec ­ tive catalysts with different extent were analysed. The proof of principle of this strategy was recently successfully confirmed [16].

Experimental The catalytic materials were prepared by using a synthesis robot “SOPHAS-Kat” from Zinsser Analytic GmbH. This robot offers the possibility of solid (granulated

229 support) and liquid handling, mixing and heating which allows a completely auto­ mated catalyst preparation (see [9]). The materials were prepared by sequential im­ pregnation on an inert support (granulated

Results

Propane Dehydrogenation Four generations comprising each 60 catalytic materials were prepared and tested so far under different conditions. The propene yields of the best materials of each gen ­ eration are shown in Figure 1. The frequency of appearance of each element in the composition of the 10 best materials of each generation is given in Table 1. Finally, the catalytic performance and the composition of the best compounds of each gen ­ eration are given in Table 2. From the first to the second generation, an improvement of the propene yield for the best materials is observed and a maximum yield of about 7% is achieved. Generation 2.1 corresponds to the same chemical compositions than generation 2 but materials are (i) prepared on smaller support particles (0.25 to 0.35 mm instead of 1 mm) and (ii) tested under different reaction conditions (feed composition C3H8 :02:N2 = 40:20:40 instead of 30:10:60). A significant increase in propene yield is observed with a maximum of over 11%. This seems to be due to particle size effects and the higher oxygen to propane ratio. Applying the evolutionary strategy under the new conditions a third generation has been prepared and tested and an improvement of the maximum yield to about 12% is achieved, which is com­ parable to the yields obtained on the best V-containing catalytic compounds of the earlier studies [8).

230 Figure 1: Propene yields for the 10 best catalytic materials of each generation at 773 K with a feed composition of C3H8 :02:N2 = 30:10:60 (generation 1 + 2) or C3H8 :02:N2 = 40:20:40 (generation 2.1 + 3); see also text.

Table 1: Frequency of different elements in the 10 best catalytic materials of each generation.

Elements Gen. Ag Ca Cr Cu Fe Mo Mn Ni P Sb W Zr

1 3 2 4 0 4 2 10 1 5 3 5 1 2 2 0 6 0 3 0 8 6 5 5 6 3 2.1 2 2 6 0 2 1 9 3 4 5 6 4 3 2 1 8 0 3 0 9 3 1 8 6 4

By comparison of the composition of the 10 best catalytic materials of each genera­ tion it is obvious that manganese is the dominant element, present in 90% of these materials. Other elements that are often present are chromium, tungsten and anti­ mony, which is, however, mainly present in low concentration. This shows that a low number of generations tends already versus a Mn-Cr-Sb-W catalytic system with different additionally elements as active phase. This has to be confirmed in further studies already under investigation.

231 Table 2: Catalytic performance of the three best catalytic materials of each gen ­ eration (conditions see Figure 1).

Gen. Ranking Composition X(CsHa) X(0;) S (CsHg) Y (C,H,) /% /% 1% /%

1 1 Cfo.21 Mn 0.43S bo .04 Wo. 33 13,1 100 51,0 6,7 2 Mn 05P o_22Sbo. 2Wo.08 12,0 100 46.3 5,6 3 Mrio.4Nio.2P 0.09 W0.31 10,5 76 49,3 5,2

2 1 Cro.25Mrio51Sbo. 04Wo.28 11,0 91 56,7 6,2 2 Mno. 47Nio.23Po. n W0.1 11,9 100 52,3 6,2 3 Mno. 44Nio.22 Wo 34 11,5 100 51,7 6

2.1 1 Sro.25Mno. 5i Sbo 04W0.2 25,2 100 45,3 11,4 2 Mno. 47Nio.23Po.11Wo.iB 21,8 97 50,3 10,9 3 Cro.2Mno. 42Sbo. 04Wo.32Zro.02 22,3 100 46,8 10,4

3 1 Cro.34Mno. 35Sbo. 03Wo.27Zro.01 25,4 100 46,7 11,9 2 Cro13Mno. 47Sbo. 12Wo.23 21,4 98 48,4 10,4 3 Cro1aMno. 39 Nio.i6Sbo .27 26,7 100 33,5 8,9

Ethane Dehydrogenation Six generations of 60 compounds each have been prepared and tested. The ethyl­ ene yields for the best compositions of generations 1 to 6 are shown in Figure 2. From the first to the fourth generation an important improvement of the best com­ pound is observed whereas the following two generations do not lead to a further significant improvement. Nevertheless, convergence between the mean ethylene yield of all materials of one generation and the yield of the best compound of the same generation is not yet reached, as can be seen in Figure 3; further improvement of the ethylene yield is still to be expected. For a better comparison of the catalytic performance of the compounds, a compara­ ble oxygen conversion of about 45% has been established for all materials by varying the contact time between 0.1 and 0.8 g-s-cm"3. From this comparison it can be seen that an improvement of the selectivity from 35% to 70% is achieved during the opti­ misation process for the best compound of each generation. Although the conver­ gence of the evolutionary process has not yet been reached, the genetic algorithm has focussed on elements that may be considered as essential: Cr, Mo, Mn and Co. Flowever, the final material may still contain other elements such as P, Sn, W and La, whereas Cu, Zr and Ca seem to play only a minor role in the composition of the best materials.

232 Figure 2: ethylene yields for the 10 Figure 3: comparison of the ethyl ­ best catalytic materials after each gen ­ ene yields for the best compound af­ eration at 773 K with a feed composition ter each generation with the mean of C2H6:02:Ar = 20:10:70 and a contact ethylene yield of the whole genera ­ time of 0.4 g s-cm"3. tion.

The maximum yield of about 17% is comparable to results obtained for a catalyst containing 5 wt% V206 on y-AI203 [17]. A material containing 9 wt% V2Os on a-AI203, which has been used as reference material, showed also a yield of about 17% under the applied standard conditions. However, the best yields in literature for low- temperature catalysts of ca. 38% are still higher, but they are obtained at different reaction conditions (t = 10.6 s, T = 623 K, feed contains 9% ethane and 6% oxygen) [5],

Conclusions The use of high-throughput techniques and especially the application of an evolu ­ tionary strategy led in both OOP and ODE reaction to the discovery of new chemical compositions of catalytic materials with promising catalytic performances. In the pre­ sent study the combination Mn-Cr-Sb-W has been found as an active and selective material for the ODP reaction. The obtained propene yields are lower than those of the best catalysts found in literature. This might be explained by the fact that prepa­ ration and testing conditions are not optimised yet. Similar conclusions can be drawn for the results of the ODE reaction. Promising catalytic materials contain mixtures of oxides from Cr, Mo, Co and Mn. The optimisa­ tion is not finished and further improvement is expected as it has been mentioned before.

233 However, the optimisation using the genetic algorithm should be considered as one step in the development of new catalysts. As we have shown in earlier studies [8] the obtained fundamental information, i.e. new active phases as well as site isolation, can be used as a starting point for a more detailed investigation of the catalyst com­ positions discovered including catalyst characterisation.

Acknowledgement This work was supported by German Federal Ministry for Education and Research, BMBF contract no. 03D0068E/3 (OOP) and by the European Union, contract no. G5RD-CT-1999-00022 (COMBICAT) (ODE).

[1] a) M. Baerns, O. Buyevskaya, Catal. Today 45 (1998), 13. b) O. Buyevskaya, M. Baerns, Catalysis, Vol. 16 (editor J. Spivey), Royal Society of chemistry, in press. [2] D. Sam, V. Soenen, J.C:Volta, J. Catal. 123 (1990), 417. [3] O. V. Buyevskaya, D. Muller, I. Pitsch, M. Baerns, Stud. Surf. Sci. Catal., 119 (1998), 671. [4] S. A. R. Mulla, O. V. Buyevskaya, M. Baerns, J. Catal., 197 (2001), 43. [5] E.M. Thorsteinson, T.P. Wilson, F.G. Young, P.H. Kasai, J. Catal. 52 (1978), 116. [6] H.X. Dai, C.F. Ng, C.T. Au, J. Catal. 193 (2000), 65. [7] O.V. Buyevskaya, D. Wolf, M. Baerns, Catal. Today 62 (2000), 91. [8] O.V. Buyevskaya, A. Bruckner, E.V. Kondratenko, D. Wolf, M. Baerns, Catal. Today 67 (2001), 369. [9] I. Hahndorf, O. Buyevskaya, M. Langpape, G. Grubert, S. Kolf, E. Guillon, M. Baerns, Chem. Eng. J., submitted for publication. [10] J. H. Holland: Adaption in natural and artificial systems, Ann Arbor, The Uni ­ versity of Michigan Press 1975. [11] K. De Jong: An Analysis of the behaviour of a class of genetic adaptive sys­ tems, Dissertation, University Michigan Ann Arbor, 1975. [12] D. E. Goldberg: Genetic algorithms in Search, Optimization and Machine Learning, reading/MA, Addison Wesley, 1989. [13] H.-P. Schwefel: Numerical optimisation of Computer Models, Chicester; John Wiley & Sons 1981. [14] T. Back, H.-P. Schwefel, Evolutionary computation 1 (1993), 1. [15] D. B. Fogel, Dissertation, University of California, San Diego 1992. [16] D. Wolf, O. Buyevskaya, M. Baerns, Appl. Catal. A 200 (2000), 63. [17] J. Le Bars, A. Auroux, M. Forissier, J:C: Vedrine, J. Catal. 162 (1996), 250.

234 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

J. Fabri Droege & Comp. AG, Diisseldorf, Germany

Optimizing Structural Configurations of Petrochemical Manufacturing Sites

The petrochemical industry is characterized by a set of important technological and economical challenges. These are, for example:

- availability of an optimized, low cost feedstock portfolio - reduction of operating costs, e.g. by intensified process integration - changing demand pattterns of key products - applicability of new processes and catalysts - management of price and margin volatility - reduced spendings on transport and logistics - creating added value from by-products

The paper will analyse how different structural configurations of petrochemical manufacturing sites may be affected by a changing business climate. The analysis will incorporate trends within the petrochemical sector tike the necessity for portfolio restructuring but also trends in the petroleum and gas industry which act as raw material suppliers. Centrepiece of all structural investigations is the ethylene cracker and its optional integration with petroleum refineries or/ and with different types of petrochemical derivates, like polyethylene, styrene or PVC. Besides assessing the pro ’s and con's of vertical backward or forward integration qualitatively a more specific evaluation and comparison of different integrated configurations will be given as well. Finally it will be analysed which structural configurations show up the best fit to comply with future requirements and which fields of activity might be most relevant for research and development.

235 DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 ” sites

2001 ' ,

must -

10-12,

; stream,

’’

portfolio

manufacturing models '

each management October

sites

for

business

Conversion value '

in petrochemicals

simulation the 86731-0

and

in

8673-111 identified

211 and

configuration

be manufacturing 211

+49

to

site

IT-based dimension

IRBP-conference,

+49 [email protected] success

petrochemical

Production on

has

to -

with

synergies

of

key Fax: another e-mail:

messages models is

is

petrochemical

issues Olefins

in AG) Kev dependant ENERCLUB,

integration

Light drivers IP,

manufacturing

portfolio operation Comp.

the

AG

drivers &

from

operational value site

AFTP, substantial

Comp. Telephone:

value

and

& ..

5-6 Droege Value integrated business are

configurations

Dusseldorf

optimising Fabri

of

for complex' the

DGMK,

'

j advanced

of of Droege Jorg

by the benefits structural Creating

Poststrafte D-40213

Dr. (Partner c/o “ option changes-in

to between

of

best gap

structural bridged

given be Integration Management Management Impact Optimizing The The

• • • • • • '« Paper Optimizing “

236

to

rating/ Cash-Flow

order

­

in own

prospects

cycles investors

contribution

share company

development the (multiples)

holders petrochemicals and upside- through volatility value

expectations

better price

in

matching a

positive reduced growth-/ throughout gain sustainable stock financed

' » ■ •

and

success

optimized straternc operational to MB###?

key

excellence

models is

industry petiochemical financing

managerial catalysts products processes new leasing outsourcing drivers

value

sites drivers

petrochemical manufacturing complex

and

of industry

the :

in

processes challenges

sites customer

sites located

costs

distribution

key

...... infrastructure well stable portfolio effects for

overhead • • •

utility

integrated proprietary

• • integrated reliability low lean scale

• ■ • ' • Management

237 only!

illustration

for

+950 B

configuration

+250 plus

plus

+5% +300 bn

6 site -500 derivate portfolio

cracker refinery on

+900

B

+300 plus +9%

+350 bn

-300 -400 4 derivate portfolio crackei '

­ dependant +700

+200 port plus A

+5% +150 bn

-350 folio 3 cracker derivate drivers

m <350| plus

+8% +250 bn

value 3 cracker

refineiy in

+300 alone)

-200 bn -20%

1 -300 cracker changes (stand

rate rate

of

of

(+10%) (+10%) )

growth

price price

vs.

change cost (+5$/bbl.)

oil

(mn

50%) change 4%)

Impact versus value

value

(NPV) naphtha economic labour crude prices gasoline (a.- (up

Value drivers Business relative • • •$ • • of total value

238

of

invest­

and models

..)

maximisation opportunities

existing

strategy identification value from attractive ment

^structure - - closures

development ' ...) operational

ratios)

structural simulation P/E

e.g.

( simulate infrastructure

desinvestments, to

products utilities, order IT-based

operation in

(e.q.

maximisation

(LP) plant investments,

in

costs profit

model, new

through

g. architecture requires

site

e.

target:

variable petrochemical

flexibilities programming

all

all

optimisation IT-based linear modular/flexible optimisation includes reflects modifications, feedstocks

• • • • optimisation

analysis

operational costs

input vs.,

factors core and

logistic options supply-demand

contract

on

price

based includes considers spot

• • • Structural 239

ton)

/

(36

sites

/year /year

/ton

mn mn

75 benefit

integration -440 ~ ~500 manufacturing

sites

petrochemical manufacturing

in

upward-integration integrated

of

substancial

are

Benefits backward-integration benefits

-

|H

Chemicals Ludwigshafen

Exxon BASF ChemSite Integration

240

value/ cost

g.

e.

adds

example exceeds

derivate ROCE

capital,

16%, , \of and integrated margin conceptual

costs

integrated fixed Integrated combined

PE PVC monomer costs

plus

plus

%= mainly variable combined ethylene ethylene ethylene each

Z

(e.g ton) for 900

PE-price 1 2

Option Option identified be

V to gX

e.

14% has

/ROCE V route

costs

fixed costs

costs (stand-alone)

polyethylene variable mainly variable polyvinylchloride ethylene

integration /

price

(e.g = ton) 900 2 PE-price C Best

241 strong sites! position

own management

of

manufacturing

segments

value portfolio

in

target integrated

attractive

on

optimized,

­

dimension focussing

with

ness market attractive

; ! scale,

comply

another

not

is

strong positions,

may position

own market

portfolio

of portfolio

products leading

target

portfolio

rent

objective: problem: business cut

n|pfins-hnsprt EtOH major major the

• • benzene ­ ness market attractive Optimizing

242

® © © owner specific Acid portfolio I

E in

but.differen P outsourced) challenge:

Acrylic

derivates -I be

key site

players

the

model (could

is

business

leading

steamcracker of

margins benefits

manufacturing

advanced and

or

infrastructure group

a

development

by

benefits integration integrated

infrastructure refinery site

of common

one

.

; operated integration

common sharing common

segments - - - maintaining site

• •

optimizing single

of handle

to products

derivates

easy

single

between

of

manyfac model

rather

competitiveness infrastructure

gap

entire on

position steamcracker

the

the services

traditional

common .for

optimisation market sustainable and

refinery not owner

transparency

one however, integrated little often products

• • • Bridging

243 244 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Lapidus, A. Krylova N. D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Russia

Ethylene Oligomerization over Ni- and Pd-Zeolites

C4-C20 olefins are widely used to produce detergents, plasticizers, addition agents to lubricants, etc. In particular, C4 olefins are of special interest [1], They are important row materials for the synthesis of divinyl, acetaldehyde, amyl alcohol. Detergents obtained from Cg-C ,2 straight olefins are very prospective because they are biologically cleaved. Cg-Cj 2 branched olefins could be applied as a feed stock for high quality motor fuel production.

Different butane isomers can be formed as products of low temperature ethylene dimerization by the following way:

2C2H4 C4Hg -l+24 kcal/mol (1)

2C2H4 cis-C4Hg-2 + 27 kcal/mol (2)

2C2H4 trans-C 4Hs-2 + 27 kcal/mol (3)

2C2H4 iso-C 4H8 + 27 kcal/mol (4)

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 245 It should be noted that

Fig. 3 Butene-1 concentration in the equilibrium ethylene dimerization is mixture formed at ethylene oligomerization exothermic reaction [2],

And lower temperature

favors butene formation.

According to [3], its

concentration in equilib­

rium mixture formed at

atmospheric pressure could 300 400 500 600 700 800 900 1000 achieve 93-99% at Temperature, K temperature of 500-600 K that is ~230-330°C (see Fig. 1). But butene-1 is the less stable product. That is why its content in C^olefins formed by C2H4 dimerization is not high.

Complex compounds of VIII Group metals are used as homogeneous catalysts in olefin metathesis and oligomerization. But heterogeneous catalysts are more prospective due to higher productivity and stability. Moreover, stages of product separation and catalyst regeneration could be excluded if heterogeneous catalyst is applied.

In our work Ni and Pd supported on zeolites were applied to carry out ethylene oligomerization.

Ni-zeolite catalysts

Ethylene dimerization over Ni/zeolite catalysts has been studied in quartz fixed- bed reactor operating at atmospheric pressure. Each sample was previously

calcined in flowing air at 500°C to transform Ni into oxide form.

246 Type of zeolite was obtained to Fig.2 Influence of zeolite on ethylene dimerization affect properties of Ni/zeolite over 5%Ni/Zeolite catalyst at 250°C catalyst in ethylene oligo ­

merization (see Fig.2).

Ni/HY was the most active

catalyst in terms of C2H4

conversion. The index achieved 55%.

Selectivity to butenes was

estimated as total yield of

butenes related to ethylene

converted. Ni/CaX catalyst was the most selective in ethylene dimerization.

Nickel loading in Fig.3 Influence of Ni loading in Ni/CaY on ethylene dimerization at 250°C Ni/zeolite catalyst

affected catalytic

properties (see Fig.3).

The highest activity

(58%) and selectivity

(15%) were obtained at

2.5% of Ni. In this case

trans-butene -2 was the

main product (78% from Ni loading, wt.% total butenes). Decrease

of reduction temperature from 250 to 200°C led to lower activity (Xc2H4= 40%,

Sc4= 12%), but the concentration of trans-butene-2 in butanes achieves 90%.

247 As it follows from Table 1, synthesis temperature influenced on activity and selectivity of Ni catalyst. The optimum temperature, product yield and

Table 1 Influence of temperature on ethylene dimerization

C4 Yield, C4 Composition, % Zeolite t °r % C4-I trans-Q.^-2 czs-C4-2 CaX 200 9 20 49 31

250 17 23 44 34

300 18 28 43 29 335 21 33 43 25

CaY 150 5 7 76 18

175 7 6 78 16

200 7 10 80 11 250 5 8 81 11 composition were depended on the catalyst formulation. For example, Ni/CaX catalyst was the most active at 335°C. Total yield of butenes was 21% at C2H4

conversion of 44%. on 80% of butenes fell within cis- and trans-butene-2. But both isomers were reduced with temperature. Contrary to them C4-l rose from 20 to 33%.

Acidity of Ni-catalyst plays an Fig.4 Influence of AJ in zeolite on ethylene conversion on Ni/mordenite important role in ethylene

oligomerization. Number of

electron acceptor sites could be

20 - estimated, for example, on % 15 - amount of ion- -3-3 radicals adsorbed on 1 g of

catalyst. It is dependent on type

of zeolite and aluminum A1 content, %

248 content in its lattice. For example, the highest activity of 2%Ni/mordenite

catalyst was observed at A1 content of 1.6% (see Fig.4). The same sample

demonstrated also the highest acidity.

Pd-zeolite catalysts

Ethylene dimerization over 2%Pd/zeolite catalysts has been studied in fixed-bed reactors operating at atmospheric and high pressure. Each sample was previously calcined in flowing air at 500°C to transform Pd into oxide form.

Pd/zeolite catalysts are normally operate at lower temperature (100-200°C) in

comparison with Ni samples.

Fig.5 Ethylene oligomerisation over Activity and selectivity of 2%Pd/fceolite catalysts Pd catalysts were dependent

on zeolite type and

composition as well as

method of catalyst

preparation. For example, at

atmospheric pressure

XC2H4 Pd/CaY catalyst was more

active but less selective in

CaX comparison of PdX sample

(see Fig.5).

Contrary to Ni/CaX, Pd supported on CaX was very selective to trans-butene-2.

Yield of this isomer achieved 50%. And total yield of C4 hydrocarbons was

78 %.

249 Increase in pressure from 1 to 10 atm led to double ethylene conversion (see

Table 2). But dimerization products decreased from 67 to 27%.

Table 2 Influence of pressure on ethylene dimerization over 2%Pd/CaX at 100°C

Yield of C4, % from C2H4 converted P, atm Xc2H4> % C4H10 C4-I trans-C^-l cis-C 4-2

1 10 0 7 47 23

5 13 2 3 23 10

10 20 5 3 10 9

15 6 5 4 18 9

Active sites of ethylene oligomerization

Active center taking part in ethylene oligomerization is bifunctional. It contains

both metal in oxide form (Ni or Pd as oxides on the surface or ions inside of

zeolite structure) and acidic site. It should be notes that zeolites themselves are

inactive in olefin oligomerization. Moreover, reduction of the catalyst and

formation of Ni° or Pd° leads to lower catalytic activity.

Probably, active site of ethylene oligomerization is formed in places of metal

localization nearby with Bronsted acid Ni2+ + HOB NiOH H sites. The active sites of , for example, In

catalysts are formed from both NiO and NiO + H Ni2+ by interaction with water or proton NiO + H Ni-O...... H"1 (see a scheme to the left).

Formation of -complex between ethylene and Ni2+ has been assumed to be the

first stage of oligomerization. Reacting with proton of active site this complex is

transformed into carbonium ion (see the scheme below). The carbonium ion

250 reacts with Ni-ethylene complex forming butene. Ate the same time active site is regenerated.

CH2jCH2-<~^ CH2-CH3 +C2H^ CH2tCH2...9H2-CH3 * n> QHg + Ni-0"-H+ Ni-0--H+ Ni—O Ni-

The active site includes Al3+ transformed by water into acid Bronsted center.

Al3+ contributes significantly to activity of catalyst of ethylene oligomerization.

It should be also note that ethylene oligomerization is accompanied by reactions such as olefin disproportionation, formation of high molecular weight products etc.

References

1. A.Lapidus//Petrochemistry. 1998. 38. 6. P.458.

2. A.Takahasi, N.Nogi, H.Takahama, C.Mita // Kague Kadaku Zassi. 1963. 99. 973

3. A.A.Vvedensky "Thermodynamic calculations in petrochemistry".

Leningrad. Gostoptechizdat. 1960. P. 324.

251 252 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

P. Wasserscheid*, A. Jess*, M. Eichmann** *lnstitut ftir Technische Chemie und Makromolekulare Chemie der RWTH Aachen, Germany, **Bayer AG, Leverkusen, Germany

Selective Dimerisation of Light Olefins in Biphasic Mode Using Ionic Liquid Solvents - Design and Application of a Continuous Loop Reactor

Catalysis in liquid/liquid biphasic systems provides an attractive possibility to combine the advantages of homogeneous catalysed reactions - mild conditions and high selectivity - with the advantages of heterogeneous catalysis. The reaction mixture consists of two immiscible solvents. Only one phase contains the catalyst allowing easy product separation by simple decantation. The catalyst phase can be recycled without any further treatment. Another attractive feature of biphasic catalysis is the possibility to extract intermediate products during the catalytic reaction into the organic layer. This often offers new possibilities to control the product selectivity. However, the right combination of catalyst, catalyst solvent and product is crucial for the success of biphasic catalysis [1],

Recently, a new approach has been adopted for biphasic catalysis, involving the use of catalyst solvents known as ionic liquids, which are simply salt mixtures with low melting points (< 100°C). Ionic liquids form biphasic systems with many organic liquids such as e. g. C6- or C8-olefmes. Moreover, their non-volatile character allows distillative product separation from the catalyst without the formation of azeotrops and without any solvent contamination of the product [2, 3, 4]. The biphasic oligomerisation of olefins using chloroaluminate ionic liquids as catalyst solvent for Ni-complexes has been first described by Chauvin et al. [5, 6, 7, 8],

In our poster contribution, we describe the use of slightly acidic chloroaluminate ionic liquids, buffered with weak organic bases, as solvents for the selective dimerisation of propene and 1-butene. Using this concept it is possible to obtain similar selectivities to those observed in monophasic catalysis e. g. in toluene, but with significant enhanced catalytic activity [9]. As shown in scheme 1, the biphasic reaction mode offers facile catalyst separation and recycling of the ionic catalyst solution.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 253 Organic phase

+ Isomers

[cation]" 1", AlCV,

ai 2ci7- Ionic liquid phase

Scheme 1: Biphasic dimerisation of butene-1 using chloroaluminate ionic liquids as catalyst solvent

After catalyst screening and kinetic experiments in batch mode (350 ml autoclave) we decided to scale up into a continuous loop reactor [10]. Our aim was to prove the technical applicability of the biphasic catalytic reaction involving chloroaluminate ionic liquids and to study the stability of the ionic catalyst solution. With respect to the latter task we decided to immobilise the ionic catalyst solution over the whole reaction time in the reactor without any exchange or regeneration. Additionally, the biphasic, exothermic character of the dimerisation reaction required intense mixing (to maximise the exchange surface) and a high ratio of heat exchange surface/reactor volume (for efficient heat removal). A scheme of the loop reactor used in our continuous experiments is shown below (scheme 2).

Circulation-

Scheme 2: Scheme of the loop reactor used for continues dimerisation experiments

254 The loop reactor used for our experiments had a volume of 160 ml. The biphasic reaction mixture was circulated by a circulation pump with high flow rates (4 1/min). Under these conditions, two static mixers in the reactor loop provided an efficient dispersion of the ionic catalyst solution in the organic phase. Two cooler/heater ensured full temperature control. Prior to the catalytic reaction, the ionic catalyst solution was placed into the reactor loop. During the reaction, the separation of the product from theionic catalyst solution was realised with a gravity separator that was integrated in the reactor loop. Thus, in contrast to the use of an external separator, all catalyst was always present in the reactor. This allowed a direct observation of the catalyst’s reactivity by analysis of the isolated products.

In the continuous dimerisation of propene the reaction was found to be mass transfer limited down to very low catalyst concentrations (1 mmol/1). At a catalyst concentration of 0,13 mmol/1 a turnover-frequency of 400.000 mol propene/mol catalyst and h was determined (other conditions: ionic liquid share of the liquid reactor content: 20 vo!.-%; 30 °C; 30 bar; propene conversion: 80 %). Main products were - almost independent from the propene conversion - dimers () which were produced in over 95 % selectivity (other products: nonenes). With N-methylpyrrole being used as the organic buffering base, the catalytic system showed slow deactivation due to some leaching of base and A1C13. With aluminiumalkyles as buffer the reactivity of the catalytic systems remained constant over at least 60h. However, in the latter case the temperature had to be kept below 15 °C to avoid reduction of the Ni- complex [11].

The linear dimerisation of 1-butene was investigated as well in the continuous loop reactor. For these experiments [(HCOD)Nihfacac] dissolved in buffered chloroaluminate ionic liquids was used as ionic catalyst solution. For this reaction it was essential to use organic bases (e. g. N-methylpyrrole) as buffering agent since Al-alkyles led to very fast isomerization of the 1- butene feedstock. After 3h reaction time in the continuous loop reactor the catalytic system under investigation showed still technically very attractive activity (TOF= 2700 h"!) and selectivity (selectivity to Cg -product >98%, selectivity to linear Cg -product= 52%) [12].

In conclusion, it could be demonstrated that the Ni-catalysed, biphasic dimerisation of propene and 1-butene can be carried out in continuous mode using a loop reactor concept. Further investigations will include the use of industrial feedstocks and feedstock mixtures, and optimization of themass transport in the loop reactor.

255 1 B. DrieBen-Holscher, P. Wasserscheid, W. Keim, CATTECH, June (1998) 47. 2 P. Wasserscheid, W. Keim, Angew. Chem. 112 (2000) 3926. 3 T. Welton, CAem Jfev. 99 (1999) 2071. 4 J.D. Holbrey, K.R. Seddon, Clean Products and Processes, 1 (1999) 223. 5 Y. Chauvin, B. Gilbert, I. Guibard, J. Chem. Soc., Chem. Common. , (1990) 1715. 6 Y. Chauvin, S. Einloft, H. Olivier, Ind. Eng. Chem. Res., 34 (1995) 1149. 7 Y. Chauvin, H. Olivier, C. Wyrvalski, L. Simon, E. de Souza, J. Catal., 165 (1997) 275. 8 Y. Chauvin, S. Einloft, H. Olivier, US Patent 5,550,304, (1996). 9 a) B. Ellis, W. Keim, P. Wasserscheid, Chem. Commun, 1999, 337-338; b) B. Ellis, W. Keim, P. Wasserscheid, PCT/GB98/00992 1998. 10 M Eichmann, dissertation, RWTH Aachen, (1999) 11 P. Wasserscheid, A. Jess, M. Eichmann, Chem. Ing. Technik, 72 (2000) 991. 12 P. Wasserscheid, M. Eichmann, Catalysis Today, 66 (2001) 309.

256 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion*, Hamburg, 2001

M. Solinas, G. Francio, W. Leitner Max-Planck-lnstitut fQr Kohlenforschung, Mulheim a. d. R., Germany

Ethylene as a C2-Building Block for Catalytic Synthesis of Fine Chemicals

1. Introduction

The development of highly stereoselective carbon-carbon bond forming reactions

continues to be one of the major challenges in fine chemical synthesis.1 The nickel-catalysed hydrovinylation is a metal-mediated coupling reaction with a remarkable potential for enantioselective synthesis.2 It comprises the formal addition of hydrogen and a vinyl group to a prochjral olefin and leads to high value products in a very elegant and atom efficient way.

Using ethene as the cheapest vinylic coupling partner, the transformation results effectively in

a chain elongation of two carbon atoms simultaneously creating a stereogenic centre in allylic position (Scheme 1). Moreover, the chiral product contains a double bond that can be further functionalised giving access to a broad variety of end products. 3 In particular, chiral olefins obtained from the hydrovinylation of styrene derivatives are key intermediates for synthesis of the largely diffused anti-inflammatory drugs 2-aryl-propionic acids.

The possible application of hydrovinylation to industrial fine chemicals synthesis is hampered by two major limitations.

The choice of catalyst for this reaction is still very limited. Ligand 4, developed by

Wilke,2b is currently the only one that led to high chemo- and enantio-selectivity. The major drawback of 4 is the rather complicated synthesis. Moreover, the chiral pool derived starting materials allow the synthesis of 4 in only one enantiomeric form.

On the other hand, like all the other homogeneous metal-catalysed reactions, the immobilisation/recycling of thecatalyst is still a critical issue.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 257 Our interest in the field of the hydrovinylation reaction is therefore directed to the synthesis of new effective and readily available catalysts combined with new concepts for their immobilisation.

Scheme 1. The hydrovinylation reaction.

[Ni(aflyl)CI]2; Ligand; Activator; C2H4 Solvent X X = H 1a X = H 2a X = H 3a X = ;-Bu 1b X = /-Bu 2b X = /-Bu 3b X = Br 1c X = Br 2c X = Br 3c

2. Catalyst development

In search for new hydrovinylation catalysts that allow for systematic modification of the ligand framework, we have turned our attention to chiral phosphoramidite 4 The selection of the ligand 5s was guided by the rather heuristic design principle that an efficient ligand system for asymmetric hydrovinylation should possess a P-N bond and contain more than one element of chirality, one of them preferentially being an atropoisomeric unit. The active catalyst was formed in situ by mixing from [Ni(allyl)Cl]2 and ligand 5 in the presence of

NaBARF (BARF = tetrakis-[3,5-bis(trifluormethyl)phenyl]borate) as activator.

Representative results are summarised in Figure 1.

258 Figure 1. Ni-catalyzed enantioselective hydrovinylation of styrenes withligand 5Ial

t b 1c Substrates w Reaction conditions: substrate to nickel ratio between 500 and 2000; t - lh; T= - 60°C; solvent CH2C12

The Ni-catalyst based on {R„^cSc)-5 gave quantitative conversion of la with 84% selectivity for the desired product 2a and an excellent enantioselectivity of 94% (S).6

Moreover, the catalyst system proved extremely efficient and remarkably robust for the hydrovinylation of la. Turnover numbers up to 8200 were reached at initial turnover frequency approaching 14 000 h"1.

In conclusion, our results show that chiral phosphoramidites are the first efficient and modular ligand system for highly enantioselective hydrovinylation. The large potential for structural variation and the straightforward synthesis of these ligands make them currently the best lead structure for catalyst development in this field.

3. Immobilisation

The quest for new strategies to immobilise organometallic catalyst is one of the major challenges in homogeneous catalysis. Supercritical carbon dioxide (scC02) is also finding increasing interest because it combines an environmentally benign character with favourable

259 physico-chemical properties for chemical synthesis.7 Catalyst separation schemes have been devised on basis of the tuneable phase behaviour of scCO; (CESS process). 8 This strategy was also tested for the hydrovinylation of la using a catalyst based on ligand 4.9 The reaction

proceeded smoothly in compressed CO2 if the activator was chosen carefully. The extractive

properties of CO; could be used to selectively removed product 2a from the catalyst and the

oligomeric compounds. Recycling of the catalyst proved difficult, however, owing to its

decomposition in theabsence of substrate.

Therefore, we designed a continuous flow process using ionic liquids rather then solid

salts as activators.This work was carried out in co-operation with Dr. Peter Wasserscheid at

the RWTH Aachen.

Figure 2. Schematic view of the continuous flow reaction apparatus.

C, Compressor; CT, Cold Trap; D, Dosimeter; DP, Depressuriser; F, Flowmeter; M, Mixer; MF, Metal Filter; P, HPLC Pump; p, pressure transducer; R, Reactor; T, Thermocouple.

Our continuous flow apparatus for homogeneous catalysis in IL/C02 systems consists of

a reactor bed partially filled with the ionic catalyst solution (Figure 2). A continues flow of a

mixture of styrene, ethene and compressed CO; was injected from the bottom of the reactor.

260 passing through the catalyst solution. 10 The product were recovered by a cold trap and analysed by GC-MS. Figure 3 shows the results of a lifetime study for pre-catalyst based on

[Ni(allyl)Cl]2 and the ligand 4 dissolved, activated and immobilised in the system

[EMlM][Tf2N]/C02 by this straightforward technique.

Figure 3. Hydrovinylation of styrene under continuous flow conditions.^

Conversion % -*-e.e. %

$ 60 40 O 55

Reaction Time; (h) w A pre-catalyst formed by mixing 0.19 mmol of [Ni(allyl)Cl]2 and an equimolecular amount of 4 was dissolved in 39 mL of [EMIM][Tf2N], filled into the reactor (R, figure 2) under argon atmosphere and cooled to 0 °C. The reactor pressure was maintained constant at 80 bar by the continues flow of compressed C02 (the exit flow was adjusted to about 30 Lh"').The reaction was then allowed to proceed with a constant styrene flow of 0.01 mL/min and 1 mL pulses of C2H4 (90 bar) every 0.5 min.

The catalyst shows a remarkable stable activity over 61 h and enantioselectivity is dropping only slightly over the long reaction period. The results clearly indicate - at least for the hydrovinylation of styrene with the catalytic system [Ni(allyl)Cl]2/4 - that an IL catalyst solution can show excellent catalytic performance under continuous product extraction with compressed C02.

As a general conclusion, continuous flow systems consisting of an ionic catalyst

solution and compressed CO2 offer a new intriguing immobilisation technique for

homogeneous catalysis. The combination of non-volatile ILs with non-hazardous CO2

261 represents a particularly attractive approach to environmentally benign processes. Thus, our new immobilisation technique provides an attractive approach to combine the molecular design of homogeneous catalysts with the advanced process design of heterogeneous catalysis.

4. Conclusion

Ethylene can be used as a C2 building bloch for highly selective Ni-catalysed C-C bond forming reaction. A fruitful interplay of catalyst development and process design is believed to be of key importance for future progress in this area.

[1] (a) Applied Homogeneous Catalysis with Organometallic Compounds', B. Comils, W. A. Herrmann, Eds.; VCH: New York, 1996; Vols. ] and 2. (b) Transition Metals for Organic Synthesis: Building Blocks and Fine Chemicals:; M. Seller, C. Bolm, Eds.; Wiley: Weinheim, 1998; Vols. 1 and 2. [2] (a) B. Bogdanovi , Adv. Organomet. Chem., 17, 105, (1979). (b) G. Wilke, Angew. Chem., Ini. Ed. Engl, 27, 185, (1988). (c) T. V. RajanBabu, N. Nomura, J. Jin, B. Radetich, H. Park, M. Nandi, Chem. Eur.J.,5, 1963,(1999). [3] Comprehensive Asymmetric Catalysis; E. N. Jacobsen, A. Pfaltz,, H. Yamamoto, Eds.; Springer: Berlin, 1999; Vols. 1-3. [4] For selected recent examples highlighting the potential of this class of ligands see: (a) G. Francid, F. Faraone, W. Leitner, Angew. Chem., Int. Ed. Engl., 39, 1428, (2000). (b) 0. Huttenloch, J. Spieler, H. Waldmann, Chem. Eur. J., 3 671, (2001). (c)B. L. Feringa, Acc. Chem. Res., 33, 346, (2000). [5] B. L. Feringa, M. Pineschi, L. A. Arnold, R. Imbos, A. H. M. de Vries, Angew. Chem., Int. Ed. Engl., 36, 2620,(1997). [6] G. Francid, W. Leitner, J. Am. Chem. Soc., submitted. [7] (a) Chemical Synthesis Using Supercritical Fluids, ed. P. G. Jessop, W. Leitner, Wiley-VCH, Weinheim, 1999. (b) P.G. Jessop, T. Ikariya, R. Noyori, Chem. Rev., 99, 475, (1999). [8] (a) G. Francid, K. Wittmann, W. Leitner, J. Organomet. Chem., 621, 130, (2001). (b) S. Kainz, A. Brinkmann, W. Leitner, A. Pfaltz, J. Am. Chem. Soc., 121, 6421, (1999). (c) D. Koch, W. Leitner, J. Am. Chem. Soc., 120, 13398, (1998). [9] A. Wegner, W. Leitner, Chem. Commun., 1583, (1999). [10] A. Bdsmann, G. Francid, E. Janssen, M. Solinas, W. Leitner, P. Wasserscheid, Angew. Chem., Int. Ed. Eng., 40, 2697, (2001).

262 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

B. Sander, M. Thelen, B. Kraushaar-Czarnetzki Institute of Chemical Process Engineering, University of Karlsruhe, Germany

Supercritical Isomerisation of n-Butane over Sulfated Zirconia

Abstract

The conversion of supercritical n-butane to isobutane over commercial catalyst extrudates comprising alumina-bound sulfated zirconia without promoters or platinum was studied. No hydrogen was co-fed. The experiments were carried out either with pure n-butane or with mixtures of n-butane and nonreactive propane at different total pressures to study the effect of butane concentration on the kinetics of the supercritical isomerisation. Within a certain temperature range, catalyst deactivation, be it caused by desulfurisation or by coke deposition, can be prevented. Experimental data obtained at 488 K, which temperature enables stable operation, were used for the kinetic analysis. It was observed that increasing n-butane concentrations in the feed result in lower conversion levels, indicating saturation behaviour. A simple scheme consisting of only two rate equations was found to be sufficient to provide a satisfactory formal description of the kinetics, indicating that two consecutive steps in the reaction network are rate determining.

Introduction

Isobutane is an important feedstock for the production of isobutene and alkylate. Present technologies for the manufacturing of isobutane are based upon the hydro­ isomerisation of n-butane over bifunctional catalysts comprising platinum supported on gamma- or eta-alumina [1, 2]. A small amount of organic chlorides and traces of water must continuously be co-fed to maintain a highly acidic aluminium chloride surface. The extreme sensitivity of these catalysts towards organic oxygenates, carbon dioxide and, in particular, towards water is considered as a disadvantage. However, the thermodynamics of the isomerisation demand for a low reaction temperature. Over chlorided Pt-alumina catalysts, butane can be processed at around 150°C whereas more robust zeolite-based catalysts are practically inactive at this temperature.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 263 Sulfated zirconia (SZ) could be the alternative catalyst for the isomerisation of n- butane. Sulfated zirconia by itself is less active than chlorided Pt-alumina, but the presence of transition metal oxides (e.g., of Fe, Mn or Cr) as promoters, or of supported platinum results in a strong enhancement of the activity.

Catalysts based on SZ are non-corrosive and much more stable against oxygen- containing contaminants than the chlorided isomerisation catalysts. The major problem here is the extremely low coke stability [3] which, under typical gas phase conditions, reduces the runtime to some minutes. The deposition of coke can largely be suppressed by co-feeding hydrogen and using at the same time a bifunctional Pt- SZ catalyst. However, these conditions favour a deactivation through reduction of the sulfate groups to which is released into the gas phase [4],

Recently, we reported on the possibility to prolong the lifetime of a SZ catalyst by converting n-butane in the supercritical state [5], Steady-state processing without activity decay is possible below a certain reaction temperature, and the production of isobutane per unit mass of catalyst and unit time can be strongly enhanced because of the high density of the feed. The temperature limit of stable operation probably depends on the individual catalyst sample. The catalyst used in our study, a commercial sample of alumina-bound SZ extrudates (MEL Chemicals), was found to be stable in the supercritical isomerisation up to a temperature of 500 K. At higher temperatures, sulfur was removed from the catalyst, and additional deactivation due to enhanced coke deposition cannot be excluded. In the same study, which was performed with pure n-butane as a feed, we found that the pressure has no effect on the deactivation rate, but rather on the steady-state conversion level.

A more detailed investigation of concentration effects is the subject of the work presented here. The concentration range of n-butane in the feed has been broadened by using mixtures of n-butane with propane. As previously, commercial extrudates of alumina-bound SZ without any promoters have been used as a catalyst. The effects of promoters on the catalytic performance in the supercritical isomerisation of n-butane is presently under investigation and will be reported elsewhere.

Experimental

The catalyst used was a sample of sulfated zirconium hydroxide (Zr(OH)4) bound to alumina in the form of cylindrical extrudates of 2.45 mm diameter and an average length of 3 mm (MEL Chemicals, MEL XZ0707/03). After calcination, the amount of alumina binder in the extrudates was 20% m/m, and the sulfur content was 2.6% m/m. Modified residence times reported here are always related to the total mass of the catalyst including alumina binder. Catalytic experiments were carried out with a continuous-flow unit equipped with a tubular fixed-bed reactor of 350 mm length and 15 mm internal diameter. In all experiments, the catalyst was diluted with SiC particles of 0.2 mm diameter in a volumetric ratio of 1:1. A detailed description of the catalyst pre-treatment, the unit, the start-up and processing procedure has been given elsewhere [5].

264 420- D □

400 - -3,5

380-

Mole Fraction n-Butane in Propane

Fig. 1: critical temperature (hollow symbols) and critical pressures (filled symbols) as a function of the mole fraction n-butane in propane; data are from ref. [6],

In the present study, the reaction temperature was fixed at 488 K. The modified residence time MRT, defined as the quotient of the total catalyst mass and the total volumetric flow into the reactor at reaction conditions, was varied between 5000 and 70000 kg s/m3. Pure n-butane or mixtures of n-butane with 52%, 70% and 90% v/v propane, respectively, were used as a feed. Additionally, the n-butane concentration in the feed can be adjusted through the total pressure which was varied between 4.6 and 8.1 Mpa.

Propane is a suitable diluent because it is completely miscible both with n-butane and the reaction products, and it is not reactive by itself at teperatures below about 523 K. Thermodynamic data of supercritical mixtures of butane and propane, available from the literature, enable the calculation and adjustment of densities and concentrations at given temperatures and pressures. The plot in Figure 1 shows that there exist an almost linear correlation between the molar fraction of n-butane in propane and the critical pressure.

Results and Discussion

In experiments with pure n-butane as a feed, it can be observed that the catalyst appears to be less active when the pressure and, consequently, the butane concentration is increased. As shown in Fig. 2, the conversions level at a given

265 o 4.6 MPa; reduced density = 0.41

□ 6.1 Mpa; reduced density = 0.65

Mpa; reduced density = 0.99

10000 20000 30000 40000 50000 Modified Residence Time Z kg- s/m3

Fig. 2: conversion of pure n-butane at 488 K and different pressures as a function of the modified residence time.

residence time drops with increasing reduced density, which is defined as the ratio of installed density at reaction conditions and the density at the critical point.

This negative pressure effect indicates that the catalyst surface becomes saturated, i.e., that at least one of the species involved in the reactions has an inhibiting effect on the rate. The selectivity to isobutane is not influenced by the pressure. Rather, it decreases from about 90% to 83% with increasing conversion at the conditions indicated in Figure 2.

With pure n-butane as a feed, it is impossible to vary the reactant concentration at constant temperature without changing the fluid density at the same time. To some extent, at least, this problem can be solved by using diluted n-butane, and the concentration range of n-butane in the feed can be broadened. As an example, a set of experiments carried out with different n-butane concentrations at a fixed reduced density of 0.42 is depicted in Figure 3. The results clearly show that higher conversion levels can be achieved at a given residence time when the feed concentration is reduced. The solid lines have been calculated on the basis of a kinetic model (vide infra). The data obtained with 10% n-butane (filled symbols) have not been fitted because both, control experiments with crushed extrudates and calculations of the Weisz modulus showed that pore diffusion is limiting the rate at these conditions.

266 To = 488 K

o 30%

♦ 10 Vol.% n-Butane

A 30 Vol.% n-Butane

O pure n-Butane

15000 30000 45000 60000 75000 Modified Residence Time / kg s/m

Fig. 3: conversion of n-butane at different concentrations but constant reduced density(0.42) as a function of the modified residence time.

It should be noted that the use of crushed catalyst particles is no general option in supercritical experiments because of pressure drop problems. Near the critical pressure, in particular, even a small pressure difference across the catalyst bed can result in a significant density gradient and, consequently, in local differences in reactant diffusivities and concentrations.

The study of mass transport effects deserves further investigation. Whereas the rate of pore diffusion is markedly reduced upon changing from gas phase to supercritical conditions, the external mass transfer could be improved because of higher Sherwood numbers (Sh) and a better solubility of reactants in the supercritical fluid. In particular, the low coking rate of catalysts in supercritical reaction media is ascribed to the improved solubility of coke precursors which enables a more efficient removal from the catalyst surface. At the first instance, however, it was our aim to analyse and describe the kinetics of the chemical reaction. For this purpose, we have omitted all experimental data which were affected by mass transfer limitations.

A close inspection of the distribution of the byproducts shows that C3 and C5 species, as well as C2 and Cg species are formed in approximately equimolar amounts, suggesting the existence of a Cg species as an intermediate.

However, the most simple and satisfactory kinetic description can be realised on the basis of only two reactions in series using model 1: n-butane -> isobutane -> byproducts

267 in which the byproducts propane, , ethane, and are lumped together to one pseudo-component.

Model 1: r — _k (-‘0,636 0,636 -k1,2 2.419 •nC4 - K1,1 VnC4 and =ku.c;nC4 • c isoC4 molV0 636r mol mol V2 419 T mol 1 ki,i = 6.18 • 10"5 and ki>2 = 3.20 • 10"9 m3 J (kg-s m3 J [kg-sj

In a second approach (model 2), which describes the experimental data equally well, the consumption rate of n-butane exhibits a negative reaction order in isobutane.

Model 2: r — _k p 0,661 p -0,165 l» p0,668 p -0,165 2.823 *nC4 — K2,1 ' UnC4 * L,isoC4 and HsoC4 - K2,1 ' UnC4 * ^isoC4 •C -2.823 -Hi = 4.29 - 10"10f^ mol k2li = 9.96 • 1CT5 r'l and k2,2 kg-sj m3 J |Jkg-sJ Vm3 ,

1750

1500 '4“i= = 8, n-Butane

|1250 ♦ Data points o -51000 - *- Model 1 c o 2 750 - Model 2 | 500 Isobutane -1> Model 3 o 250 ,—i — Byproducts 0 —» .. n . ■ i.; #.; .au : ^ ■. p ■. a. ,q_ 0 10000 20000 30000 40000 50000 60000 70000 80000 Modified Residence Time / kg s/m3

Fig. 4: reactant concentrations as a function of the modified residencetime in the isomerisation of pure n-butane at 488 K and a pressure of 4.6 MPa.

From the viewpoint of mechanistic catalysis and surface science, a third model (shown on the next page) is most appealing because the rate expressions have similarity with Langmuir-Hinshelwood kinetics, a widely used approach to describe catalyst saturation behaviour. Also, the assignment of odd reaction orders could be

268 avoided in model 3. However, the formal similarity with LH kinetics should not be over-interpreted. For instance, no proof can be given at present that the coefficients

Kn c4 and Kisoc4 are adsorption constants.

Model 3: and

with k31 = 1.14 • 10"7

K„c4 = 7.18- 10"

All three models are suitable to describe the experimental results. As compared to the other two models, model 3 has the disadvantage that the formation of isobutane is slightly overestimated at cost of the byproducts. This is shown in Figure 4, where data of the conversion of pure n-butane are depicted together with the results calculated on the basis of the three models. For other n-butane feed concentrations, as well, models 1 and 2 fit the product data more accurately than model 3.

All three models have been developed such that the respective rate coefficients ky as well as the coefficients K„c4 and KisoC4 are invariant in the reactant concentrations. Because the range of concentrations evaluated in this study is quite broad, the models can be considered as robust.

It is a remarkable feature of all models that only two rate equations are required. In all cases, the two equations must represent two consecutive rate determining steps in the reaction network.

Conclusions

The isomerisation of n-butane under supercritical conditions is an interesting option to prevent or delay coke deactivation, enabling the use of sulfated zircona which is less sensitive to water and oxygenates than conventional isomerisation catalysts. Because of the high density of the feed, production capacities of isobutane can be much higher than in the gas phase. However, with the nonpromoted SZ catalysts used in this study, the conversion level was found to decrease at high feed concentrations, whereas pore diffusion became rate limiting at low feed concentrations. Further research, for instance, involving SZ catalysts with different chemical and diffusional properties will be necessary to improve the process.

Acknowledgement

The authors thank the German Research Foundation (DFG) for financial support and MEL Chemicals for providing the catalyst.

269 References

[1] D. Rosati, in: R. Meyers (Ed.), Handbook of Petroleum Refining Processes, McGraw-Hill, New York, 1986. [2] P R Sarathy and G.S. Suffridge, Hydrocarbon Process. 2,43 (1993). [3] S.Y. Kim, J.G. Goodwin Jr. and D. Galloway, Catal. Today 63, 21 (2000). [4] B.-Q. Xu and W.M.H. Sachtler, J. Catal. 167, 224 (1997). [5] B. Sander, M. Thelen and B. Kraushaar-Czarnetzki, Ind. Eng. Chem. Res. 40, 2767 (2001). [6] A. Kreglewski and W.B. Kay, J. Phys. Chem. 73, 3359 (1969).

270 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

V. Duma, R. Fodisch, D. Honicke Chemnitz University of Technology, Chemnitz, Germany

Effect of Sodium Promotion on the Performance of Fe x Oy/Si02 Catalysts in the Gas Phase Epoxidation of Propene

Introduction

Propene oxide is one of the most important organic intermediates and is used for the production of many useful chemicals. Today propene oxide is mainly manufactured by one of two commercial processes viz. the organic-hydroperoxide process or the chlorohydrin process [1a, 2], both of them performed in the liquid phase. The compulsory formation of a coupling product in the former and the chlorine used in the latter process are substantial drawbacks of these industrial processes. Therefore, the development of a process for the direct gas phase epoxidation of propene which overcomes these disadvantages would be of great economical importance. Our approach to find an effective method for the gas phase epoxidation of propene was to use nitrous oxide as an oxidant and a suitable solid catalyst [3, 4], The catalyst for the epoxidation of propene must adsorb and activate the N20 molecule but without its decomposition. Furthermore, the catalyst must not have strong acidic and basic properties because with such properties propene oxide can be easily isomerized to propanal and acetone [1a, 2, 5]. Thus, silica gel was chosen as the catalyst support which offers a high surface area, low acidity, and no intrinsic activity in N20 decomposition. In order to improve the adsorption capacity of silica the use of a highly dispersed transition metal oxide as an active component should be effective. In a comparison [6] between the activity of oxides of several 3d transition metals in the NzO decomposition, it is shown that Fe 3+ and Fe 2* ions have the lowest level of activity. In another investigation [7] concerning the N20 decomposition activities of several catalysts, supported iron oxide also shows one of the lowest rates of activity. Drago et al. [8] found a very low activity level of silica-supported iron oxide as well in the decomposition of N20. Furthermore, alkali promotion and high temperature calcination are suitable measures for reducing the acidity of the catalyst. For these reasons, catalysts were prepared in the present study by doping various silica gels with small amounts of iron oxide, followed by sodium impregnation and calcination.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 271 Experimental

Catalyst preparation

The support used was a silica gel from Degussa (Aeroperl 380/50), produced from SiCI4 by the flame hydrolysis process. This support was impregnated by incipient wetness method with solutions of different concentrations of Fe(lll)-acetylacetonate in toluene. The so prepared catalysts were dried and then calcined at 873 K in air. To decrease the acidity of the catalysts a neutralization procedure was then carried out using an aqueous solution of sodium acetate (NaAc) followed by filtration, washing with distilled water, and calcination at a temperature of 973 K for 6 h in air. The iron loadings of the prepared catalysts were 100, 300 and 1000 ppm.

Oxidation Experiments

The propene oxidation experiments took place with propene of 99.5 vol% purity, N20 of 99.5 vol% purity, and helium of 99.996 vol%, all of them from Messer Griesheim. They were carried out in a fixed-bed quartz-tube reactor (10 mm i.d.) under an absolute pressure of 1200 mbar using a continuous-flow system. The reactor, placed vertically, was surrounded by an electrical heater. The reactor temperature was measured and controlled via a thermocouple placed within the catalyst bed. The gaseous reactants were fed using mass flow controllers from Brooks. The reaction products were analyzed on-line using an IR-photometer from Rosemount to detect C02 and a gas chromatograph 5890 II Plus from Hewlett-Packard for the organic and inorganic products. The gas chromatograph was equiped with two capillary columns viz. Molsieb 5A and FFAP connected with a TCD and a FID, respectively, which enabled the analysis of 02, N2, CO, and NzO by the former and organic products by the latter detector. The catalyst (usually 1 g) was placed into the reactor, near the bottom, on a porous quartz frit. Before each propene oxidation experiment, the catalyst was oxidized at 793 K in flowing air in order to remove remaining organic compounds from the previous experiment which were oxidized to C02 and to yield a high oxidation state of the active component of the catalyst. When no more C02 was detected in the outgas, the reactor temperature was adjusted to the desired oxidation temperature for the reaction experiment. Then, the air flow was replaced by a mixture of 15% N20 and 85% He. When the N2Q concentration at the outlet of the reactor was constant, 1 % propene was added to the mixture and the flow of He was correspondingly diminished to keep the total flow constant at a GHSV of 4 lh" 1g"1. This moment was considered as the inital

272 point of time (time-on-stream = 0) for the oxidation experiment. During the oxidation experiment the reaction parameters such as pressure, temperature, gas composition, and velocity were kept constant and the reaction products were continuously analyzed by the IR-photometer and periodically by the gas chromatograph. After a period of time the experiment was stopped; the flows of propene, nitrous oxide, and helium were interrupted and the catalyst was regenerated by oxidation at 793 K in a flow of air.

Results

In order to investigate the influence of the iron content on the catalytic performance, a series of catalytic runs were carried out using different catalysts with varying iron content. The degree of propene conversion as a function of the iron content after a time on stream of about 80 minutes is depicted in figure 1 for both the promoted

Sodium promoted catalyst Unpromoted catalyst

300 Iron content [ppm] Fig. 1: Degree of propene conversion as a function of the iron content of the catalyst 1% propene, 15% N20, 84% He, 648 K, GHSV = 4 lh"'g 1 catalyst (with 0.1 mol/l NaAc solution) and the unpromoted catalyst. As can be seen, the degree of propene conversion increases with increasing iron content of the catalyst. However, the effect of sodium promotion is only less pronounced. It led to a small increase in conversion degree when using the catalysts with 300 and 1000 ppm iron, respectively.

273 A completely different picture arises, when focusing on the selectivity to propene oxide as depicted in figure 2. Applying the unpromoted catalyst, the selectivity to

3 1 1

2 f

w

100 300 1000 Iron content [ppm] Fig. 2: Selectivity to propene oxide as a function of the iron content of the catalyst 1% propene, 15% N20, 84% He, 648 K, GHSV = 4 Ih V1

propene oxide is below 1.5% and decreases with increasing iron content of the catalyst. Contrary, when applying the sodium promoted catalyst, the selectivity to propene oxide increases by more than one order of magnitude and shows a very distinct maximum at an iron content of 300 ppm. As already mentioned, the impregnation with the sodium acetate solution was done in order to neutralize the acidity of the catalysts. In figure 3 the selectivity to propene oxide at varying temperatures is shown for a series of catalysts with an iron loading of 300 ppm but which have undergone different sodium impregnation treatments. The selectivity to propene oxide was very low over the unpromoted catalyst, increased with the sodium loading, reached a maximum for the impregnation in the intermediate concentration of 0.1 mol/I aqueous solution of sodium acetate, and decreased with further increase of the sodium acetate loading. Surprisingly, this result was independent of the chosen temperature. Furthermore, to investigate the effect of sodium promotion on the properties of the catalysts TPR (temperature programmed reduction) spectra of sodium promoted and unpromoted catalysts were recorded. The TPR spectra for catalysts with 300 and 1000 ppm iron content as well as with and without sodium are shown in figure 4. Despite the weak intensity of the signals due to the small amounts of iron oxide, one effect can be

274 ■623 K '• 648 K >673 K

Cone, of NaAc solution used for the promotion [mol/I]

Fig. 3: Selectivity to propene oxide as a function of temperature and sodium amount 1% propene, 15% N20, 84% He, GHSV = 4 lh" V1 seen, a shifting of the reduction temperature toward higher values for the sodium promoted catalysts than for the unpromoted catalysts. Obviously, this effect seems to be directly related to the observed propene oxide selectivity.

Temperature [K] 300 400 500 600 700 800 900 973 isotherm

Fe1000Na0.1

I’HWEWf»***!,Pel 000

Fe300Na0.1 Fe300

500 1000 1500 2000 2500 3000 3500 Time [s] Fig. 4: TPR spectra of sodium promoted and unpromoted silica supported Fe 203 catalysts

275 Summary

The present study describes a novel method of forming propene oxide by the heterogeneously catalyzed gas phase epoxidation of propene with nitrous oxide. The catalyst consisted of silica-supported, sodium-promoted iron oxide. The optimal iron loading was 300 ppm. The impregnation of the catalysts with alkali (sodium) was of great importance in order to minimize the side reactions and thereby to improve the selectivity toward propene oxide.

Acknowledgement

The authors express their gratitude to the Ponds der Chemischen Industrie and the Max-Buchner-Forschungsstiftung for financial support of this work.

References

1. „Ullmann ‘s Encyclopedia of Industrial Chemistry," 5th ed. VCH, Weinheim. (a) D. Kahlich, U. Wiechern, J. Lindner, Vol. A22, p. 239 (1993); (b) S. Rebsdat, D. Mayer, Vol. A10, p. 117 (1987) 2. D. L. Trent, in „Kirk-Othmer Encyclopedia of Chemical Technology," 4th ed., Vol. 20, p. 271. Wiley, New York, 1996 3. V. Duma, D. Honicke, J. Catal. 191, 93 (2000) 4. V. Duma, Heterogen katalysierte Gasphasen-Epoxidation von Propen an FeO x/Si02-Katalysatoren, Dissertation, TU Chemnitz, 2001 5. J. M. Coxon, R. G. A. R. Maclagan, A. Rack, A. J. Thorpe, D. Whalen, J. Am. Chem. Soc. 119, 4712 (1997) 6. .Handbook of Heterogeneous Catalysis" (G. Ertl, H. Knozinger, J. Weitkamp, Eds.). VCH, Weinheim. (a) A. Cimino, F. S. Stone, Vol. 2, p. 845 (1997); (b) G. Bergeret, P. Gallezot, Vol. 2, p. 439 (1997) 7. T. Yamashita, A. Vannice, J. Catal. 161, 254 (1996) 8. R. S. Drago, K. Jurczik, N. Kob, Appl. Catal. B 13, 69 (1997)

276 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

D. Freeman, R. P. K. Wells, G. J. Hutchings Department of Chemistry, Cardiff University, United Kindom

Enhanced Aromatic Formation in the Methanol to Hydrocarbon Reaction Using Composite Catalysts

Abstract The conversion of methanol to hydrocarbons using physical mixtures of group thirteen oxides (P- Ga203 and ln 203) and the zeolite H-ZSM-5 is reported and discussed. The addition of p-Ga203 to H- ZSM-5 markedly increases the selectivity to aromatic hydrocarbons. The effect is observed only at high conversion and at temperatures > 400°C. The maximum effect is observed for 1:1 mixtures of P-Ga20; and H-ZSM-5. A similar effect is observed for a physical mixture of ln 203 with H-ZSM-5 (1:1 by wt); however, the effect with ln 203 is observed at lower temperatures (300°C) since, at 400°C, the composite catalyst deactivates very rapidly. The results are explained in terms of a new active site tha is formed at the point of contact between theparticles of Ga203 or ln 203 and the zeolite crystallites.

Introduction Methanol conversion to hydrocarbons has been well studied since the landmark work of Chang anc Silvestri reported the use of H-ZSM-5 as an effective catalyst (1). Other oxides have been used a; catalysts, e.g. W03/y-Al203 (2) but, in general, zeolites give superior yields of hydrocarbons and longei effective lifetimes. Zeolites also have the added advantage that they can be readily regenerated by £ simple air oxidation treatment. The literature on methanol conversion over zeolite and oxide catalysts is extensive and this has recently been reviewed by Stocker (3) fort he formation of light alkenes. Earlier reviews by Chang (4), Yurchak (5) and Hutchings and Hunter (6) have discussed the conversion of methanol to gasoline. To date, a broad range of zeolites have been investigated as catalysts, including zeolite Y (7), zeolite p (8), mordenite (9), EU-2 (10) and clinoptilolite (11), as well as other

microporous studies, e.g. AIPO4-5, SAPO-5 and MeAPO-5 (12) and SAPO-34 (13). In recent research, considerable attention has been given to two topics in particular. First, the mechanism of formation of the initial carbon-carbon bond has remained a matter of considerable debate. Recently, this mechanism has been investigated using computational approaches, and this is giving new insights on this difficult topic (14,15). Second, the control of the product distribution to form light alkenes exclusively remains important (13). Although numerous zeolites and oxides have been investigated as catalysts, there has been surprisingly little attention given to the study of composite zeolite catalysts, in which a zeolite is physically mixed with a second component. The addition of Fe203 to a zeolite catalyst was described by Qinghua et al. (16) and this was found to affect both the conversion and selectivity. This lack of attention to composite catalysts is most surprising given the extensive literature concerning the use of Ga2G3/H- ZSM-5 catalysts for the aromatisation of propane (17-19) and Mo-H-ZSM-5 catalysts for the aromatisation of methane (20-22). In these reactions, the use of physical mixtures of the two components is found to give different catalyst performance when compared with the pure components alone. We have now studied the use of Ga203/H-ZSM-5 and In 203/H-ZSM-5 composite catalysts for the methanol conversion reaction and, surprisingly, we have found that addition of Ga203 or ln 203 to

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 277 H-ZSM-5 as a physical mixture can significantly affect the product distribution. In particular, the selectivity to aromatic hydrocarbons is enhanced with these catalysts. In this paper, we present our preliminary results using ln 203 as a catalyst component, and compare these results with Ga2G3 as a catalyst component.

Experimental Zeolite H-ZSM-5 (Si/Al ratio = 80) was obtained as a commercial sample. Gallium oxide ((B-GazOs)

and Indium oxide (ln 203) were obtained from Aldrich. These materials were used either as powders or as coarse particles which were prepared by pelleting and sieving to give the required size fraction of particles. Composite catalysts were prepared in two ways (a) as a simple physical mixture of the two components, either as powders or particles formed by shaking the two materials together, and (b) by grinding the two components together to ensure theyare thoroughly mixed. The composite catalysts were tested as catalysts for the methanol conversion reaction using a standard stainless steel laboratory microreactor. Methanol was fed to the reactor using a calibrated syringe pump and was mixed with helium diluent prior to vaporisation. Catalyst samples (typically 0.2-03 g) were contained in the heated zone of a tubular microreactor (i.d. = 9 mm) and all lines to and from the reactor were heated to avoid the condensation of reactants and/or products. Product analysis was carried out using gas chromatography using a flame ionisation detector and a poropac Q column (2 m x 3 mm). Catalysts were characterised using powder X-ray diffraction.

Results and Discussion

H-ZSM-5 and p-Ga203 were separately reacted with methanol at 400°C and WHSV = 0.7 g methanol/g catalyst/h. The results are shown in Table 1. H-ZSM-5 converts methanol to hydrocarbons, as expected, and two experiments were conducted, one using glass beads and the second using boron nitride as inert

diluent (1:1 by wt). In contrast, p-Ga203 was inactive under these conditions. Subsequently, the zeolite

(0.125 g) and (3-Ga203 (0.125 g) were mixed together as a simple physical mixture. This was achieved

by mixing together particles of the zeolite (0.6 -1.0 mm) and particles of p-Ga203 (0.25 - 0.3 mm) in a glass tube without crushing. This physical mixture gave a significantly higher yield of aromatic

compounds and lower yields of C4 and C5 hydrocarbons when compared with the same mass of H-

ZSM-5 alone. In a subsequent experiment, H-ZSM-5 and P-Ga203 were mixed more thoroughly by crushing the two powders together. Under the same reaction conditions, this physical mixture gave significantly higheryields of aromatic hydrocarbons and over 50% of the products now were observed

to be C6-C9 aromatics. C; and Cg aromatic compounds are particularly favoured. The effect is long lived and, even after 870 min time-on-line, significant yields of aromatic compounds are still observed (Table 1).

A series of experiments were conducted at 400°C in which the ratio of p-Gaz03/H-ZSM-5 was varied for physical mixtures of the two components prepared by crushing the two powders together. The results are shown in Table 2. The highest yields of aromatic compounds are obtained for the 1:1

physical mixture. Catalysts containing up to 25 wt% Ga203 are stable over the test period, ca. 900 min.

However, catalysts containing higher amounts of p-Ga203 gradually deactivate and the yield of

aromatic hydrocarbon products declines. It is clear that, even with 10 wt% p-Ga2G3, a significant

enhancement in the yield of aromatic products is observed. In general, the yield of C3-C5 hydrocarbons

declines as the wt% of p-Ga203 is increased, and the yield of xylenes and is significantly enhanced.

278

(%)

0 9.3 10.4 18.2 51.4 38.5 Aromatic yield hydrocarbon

12 H - 9 5.5 5.5 2.7 18.5 20.6 C crushed

- not

3.4 5.9 2.7 15.4 24.2 CgHio ratio

- 1.4 1.1 mol 5.1 2.2 3.3

C,Hs 50 0.6-1.0mm);

b > - 1.5 1.3 3.1 2.4 4.6 (%) CA

pellets,

0.25-0.3mm)

- 5.5 7.3 3.0 4.9 3.5 Cs+ alkene/alkane

(0.125g, Selectivity

5 pellets,

tr c catalysts' 2.5 4.1 12.7 10.4 12.6

typically Product

4 H-ZSM-5 tr (0.125g, c

6.0 9.3

(0.125g) 27.6 28.2 21.9 (0.125g)

.

with alkane, 0

3 . nitride beads tr

beads c

38,4 35.5 30.6 36.1 21 mm) and

min"' p-GaiOj/H-ZSM-S

glass glass boron

0.6-1 tr Cz

8.8 6.5 6.3 15.7 13.2 alkene 60ml

with 0.25-3.0mm) with with

as

He 4

, 1 pellets, H-ZSM-5, tr

0.2 0.2 0.3 0.8 0.9 CH mixed mixed mixed

pellets,

quoted over

(0.25g,

0mm) 1.0mm) (0.125g, 0.5 100 100 100 100 100 (%)

WHSV=0.7h"

0.6- 0.6-1. 30min, 03 0.25-0.3mm)

Methanol 2

Conversion conversion

hydrocarbons

-Ga 400°C, 6 11

pellets, pellets, f -C&+

pellets, 870min 2

of

C

Methanol 25g, time-on-line time-on-line 6

for mixed crushed crushed

3 3 3 (0.1 (0.125g, (0.125g,

0 0 0 conditions: mixture

2 2 2

beads

d time-on-line crushed, -Ga -Ga -Ga

Catalyst

f, g,

-GazOa Selectivities as as H-ZSM-5 Reaction H-ZSM-5 Physical 1

H-ZSM-5" p-GazOa/glass H-ZSM-5/ H-ZSM-5/ H-ZSM-5/BN Table H-ZSM-5/

279

(%) 0 0 8.1 15.5 37.2 51.4 39.6 32.3 38.5 30.8 Aromatic yield hydrocarbon 8 12 - - H 1.9 9 4.6 17.8 16.1 18.5 14.7 20.6 20.6 C catalysts

- - 7.2 2.1 10.7 14.2 17.8 15.4 12.3 24.2 CgHio /H-ZSM-5 3 - - 1.2 1.9 5.1 3.4 3.7 0 2.4 3.3 2.3 2 C?H,

b -Ga 0.25g

- -

1.8 1.9 1.5 (%) 1.6 1.3 2.9 3.4 2.4 =

CA over

alkane

- - 3.2 8.6 4.3 3.0 5.2 9.1 6.1 3.5 Ce+ catalyst

, and

1 Selectivity

5 - tr C 8.3 5.0 8.6 4.1 2.5 4.6 5.6 4.1 conversion min"

alkene

Product 4 - 60ml tr C

6.0 9.9 9.3 total 10.7 11.5 10.7 21.2 20.4

He methanol as

, 3 on tr c 17.5 21.0 43.7 32.5 30.6 28.5 30.5 41.3 24.8 0.7h"‘ quoted

2 loading

tr C 8.3 3 7.7 9.8 15.7 11.7 13.2 15.3 17.0 27.7 0 2 WHSV = WHSV

4 -Ga tr 1.8 0.2 0.5 0.4 0.6 0.9 0.4 0.7 ch 54.9

hydrocarbons

+ 30min of 4

870min 400°C,

at -C

at

2 C

Effect 100 100 100 0.7 0.5 100 100 100 for 74.3 90.3

Methanol Conversion conditions:

time-on-line

time-on-line

Reaction Select!vities Final Initial (wt%) 2 0 3

10 25 75° 50° 50" 75" 90° 90" 100 0 2 Ga Table

280 35-,

30-

25-

y 20-

c 15-

10-

5-

0 |#--- • | - 4—* —•---- •—...... * —•-----, 0 200 400 600 800 1000 1200 Time on line (min)

Figure 1 Methanol conversion (300°C, WHSV =1.3 h"1, He 60 ml/min) over the In 203/H-ZSM-5 composite catalyst compared with H-ZSM-5. Key ■ In 203/H-ZSM-5; • H-ZSM-5.

The effect of temperature on methanol conversion over (3-Ga2C>3/H-ZSM-5 (1:1 by wt) physical mixtures has also been investigated. The results shown in Table 3 show that the effect of enhanced aromatic product formation is observed only at high temperatures (> 400°C). Below 400°C, p-Ga203 addition did not affect the product distribution markedly. In a subsequent set of experiments, ln 203 was investigated as a catalyst component and physical mixtures of In 203/H-ZSM-5 (1:1 by wt) were investigated. The results are shown in Table 4 and Figure 1. At 400°C, the composite catalyst was inactive, giving a methanol conversion of <1%. However, at 300°C, a significant effect was observed and the conversion of methanol was significantly enhanced when ln 203 was present as a catalyst component, which was observed for ca. 1,000 min time-on-stream (Figure 1). These experiments indicate that composite catalysts, in which an oxide is mixed with the zeolite H-ZSM-5, may provide a relatively simple approach to enable control of the product composition in the methanol conversion reaction, and clearly this approach requires further experimentation. In particular, we consider that a new active site is formed in these composite catalysts at the interface between the zeolite crystallites and the p-Ga203 or ln 203. At these new sites, light alkenes are aromatised and experiments are now in hand to investigate the nature of the active sites in these composite catalysts.

Acknowledgement We thank the EPSRC for financial support.

281 ot , u 0 0 15.8 15.9 17.7 Ar 12 H - - 9 7.4 5.3 4.9 C 10 - - 8.4 5.8 5.5 CgH - - 1.0 3.6 2.2 C,H, (%)'

- - 1.7 0.8 2.8 QH,

/H-ZSM-5 . 3 44 - tr 0 C 2 54.5 32.1 22.0 3mm) f3-Ga - C, 17.6 48.7 41.1 40.6 0.25-0.

2 - c 11.8 18.5 10.9 48.0 pellets, 4

- 0.2 0.7 3.2 0.6 ch (0.125g)

(0.125g, 0

0.1 100 100 97.5 Conv ______beads

-GazO; ,„, __ d glass tr

7.5 9.7 15.8 Ar 1 with with 2

, H 0 9 1.4 7.3 2.4 C '

mixed mixed

selectivity

min -

tr 5.7 2.9 2.9 C#,o 60ml

b product physically physically h - 0

1.0 1.2 1.3 He C,

, on

-1 b - 0 3.1h 1.9 1.8 3.2 QH« = 0.6-1.0mm) 0.6-1.0mm)

+ 4 temperature temperature tr

C H-ZSM-5(%) 53.9 39.0 26.3 44.7 of WHSV

pellets, pellets,

- c, 18.5 31.5 40.7 46.4 Effect products

2 - (0.125g, (0.125g, c

conditions: 10.4 19.3 11.6 23.2

aromatic

tr 0.6 0.2 0.5 0.2 CH,

Reaction H-ZSM-5 H-ZSM-5 total 3

tr 0.6 100 100 94.3 Conv Table

______500 350 400 200 300 (°C) emperature

282

(%)

0 0 0 0 9.7 9.0 10.8 Aromatic yield hydrocarbon 12 - - - - H 9 4.5 4.2 2.0 C - - - -

3.7 3.9 2.7 C#,o 8 1.0mm) h - - - - 7 1.1 1.4 0.4 c (0.6-

6 h - - - - 6 ’ “ 0.5 0.3 3.1 c (%) particles

- - - - - Cg-t- 14.2 20.2 catalysts give

to

s - - - - Selectivity tr c 8.3 6.6 sieved

25g) 4 - - tr tr c Product 15.1 15.2

47.3' (0.1

25g) InaOg/H-ZSM-S

3 (0.1 pelleted, - - alkane tr tr

and c

beads

38.2 29.3 34.6

and 0)

and 2

beads ln glass 2

- - tr tr c 6.3 17.1 18.0 alkene glass with min. together

as

4 with

H-ZSM-5, 60ml

tr tr

tr tr mixed 0.1 0.4 0.6

ch crushed

He quoted

over mixed

0.125g)

1.3h'*,

(0.125g) =

conversion 6.1 100 30.8 <0.5 <0.5 <1.0 <0.5 1.0mm,

hydrocarbons

+ Methanol 6 WHSV Conversion

(0.6- H-ZSM-5 -C

min.

2 min. C

(0.6-1.0mm,0.125g) and Methanol

870

30

for

400 300 300 300 300 400 400 (°C) particles Temp.

conditions:

8 h ' particles (0.125g) r f f'

0; 03 2 2 0 Select!vities H-ZSM-5 C4+ ln In Time-on-lime Reaction Time-on-line 4 /H-ZSM-5 /H-ZSM-5 /H-ZSM-5 3 3 3 Catalyst 0 0 0 2 2 2 Table Table Tn in H-ZSM-5' bzO,' In InA' H-ZSM-5

283 References 1 C.D. Chang and A.J. Silvestri, J. Catal., 47, 249 (1977). 2 G.A. Olah, H. Doggweiler, J.D. Felberg, S. Frohlich, MJ. Grdina, K. Karpless, T. Keumi, S. Inaba, W.M. Ip, K. Lammertsma, G. Salem and D C. Tabar, J. Am. Chem. Soc., 106, 2143 (1984). 3 M. Stocker, Microporous Mesoporous Mater., 29, 3 (1999). 4 C.D. Chang, Stud. Surf. Sci. Catal., 61, 393 (1991). 5 S. Yurchak, Stud. Surf. Sci. Catal., 36, 257 (1988). 6 G.J. Flutchings and R. Hunter, Catal. Today, 6, 279 (1990). 7 P. Salvador and W. Kladnig, J. Chem. Soc., Faraday Trans. 1, 73, 1153 (1977). 8 G.J. Hutchings, P. Johnson, D.F. Lee, A. Warwick, C.D. Williams and M. Wilkinson, J. Catal., 147,177 (1994). 9 M. Sawa, M. Niwe and Y. Murakami, Chem. Lett., 8, 1637 (1987). 10 J.L. Casci, B.M. Lowe and T. Vincent, UK Patent 2077709 (1981). 11 G.J. Hutchings, T. Themistocleous and R.G. Copperthwaite, Appl. Catal., 43, 133 (1988). 12 O.V. Kikhtiyanin, K.M. Mastithin and K.G. lone, Appl. Catal., 42, 1 (1988). 13 B.V. Vora, T.L. Marker, P.T. Barger, H R. Nilsen, S. Krisle and T. Fuglerud, Stud. Surf. Sci. Catal., 107, 87(1997). 14 N. Tajima, T. Tsuneda, F. Toyama and K. Hirao, J. Am. Chem. Soc., 120, 8222 (1998). 15 S.R. Blaszkowski and R.A. van Santen, J. Am. Chem. Soc., 118, 5152 (1996). 16 X. Quinghua, C. Guoguan, W. Qingxia and W. Gougwei, Ranliao Huanxue Xuebao, 22, 103 (1994). 17 D. Seddon, Catal. Today, 6, 351 (1990). 18 G.D. Meitzner, E. Iglesia, J.E. Baumgartner and E.S. Huang, J. Catal., 140,209 (1993). 19 G. Buckles and G.J. Hutchings, J. Catal., 151, 33 (1995). 20 S.B. Derouane-Abd. Hamid, J.R. Anderson, I. Schmidt, C. Bouchy, C.J.H. Jacobsen and E.G. Derouane, Catal. Today, 63,461 (2000). 21 B.M. Weckhuysen, D. Wang, M.P. Rosynek and J.H. Lunsford, J. Catal., 175, 338 (1998). 22 J.-Z. Zhang, M.A. Long and R.F. Howe, Catal. Today, 44, 293 (1998).

284 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

A. Behr, L. Obendorf, V. A. Brehme University of Dortmund, Germany

Development of a Process for the Production of Glycerol in Tertiary Butyl Ether as Octane Booster on Isobutene Basis

The synthesis of glycerine tertiary butyl etherswas investigated. The „higher ethers" (diethers and triether) can be used as octane boosters. A screening of possible catalysts gave the result that homogeneous catalysts perform better than heterogeneous ones, p-toluene sulfonic acid was found to be the best catalyst. Thus, it was the only one considered in this investigations.

Concentration versus time profiles were experimentally determined. A simplified kinetic model of the reaction was developed fitting the experimental data. Good results were obtained modelling the reaction system in three consecutive equilibrium reactions, applying power laws for the reaction rates and Arrhenius equations for temperature dependence.

The liquid-liquid-equilibrium was investigated. The diethers and the triether were treated as one pseudo component „higher ethers". Thus, the equilibrium of the quaternary system glycerine-monoether-,higher ethers“-isobutene was measured under pressure. A complex miscibility gap was formed consisting of the two partially miscible binary systems glycerine- higher ethers" and glycerine-isobutene. NRTL parameters were fitted to the experimental data.

Based on the above investigations a continuous production process for higher ethers" was developed (Fig. 1).

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 285 glycerine higher ethers isobutene

Figure 1: Process flowsheet

The process consists of a reactor cascade (R-l, R-2, R-3) followed by an extraction (C-l) at reaction pressure. Isobuten is fed directly to the reactor while glycerine is fed to the extractor, extracting catalyst and monoether from the reaction product. The mixture is then passed back to the reaction. Unreacted isobutene in the raffinate phase of the extraction is stripped overhead in a stripping column (C-2) and passed back to the reaction. The bottom product of the flash is passed to a vacuum rectification column (C-3) obtaining pure „higher ethers11 overhead. The bottom product of the rectification is passed back to the reaction.

286 DGMK-Conference "Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

H. Redlingshofer, G. Emig Technical University of Erlangen-Nuremberg, Germany

The Catalytic Wall Reactor as Tool for Kinetic Investigations in the Selective Oxidation of Propene to Acrolein

The choice of an appropriate reactor for kinetic measurements is decisive. Difficulties occur if highly exothermic or endothermic reactions should be investigated. To ensure isothermal reaction conditions during the highly exothermic oxidation of propene to acrolein the catalytic wall reactor

(CWR) was used for collecting isothermal data.. Besides propene and acrolein the most important side products (acrylic acid, carbon oxides, acetaldehyde, formaldehyde, acetic acid) were monitored to be able to determine a detailed reaction scheme.

The experimental results in the CWR clearly showed a change in the rate determining step of the main reaction depending on temperature. At low temperatures catalyst reoxidation is rate determining with increasing oxygen content accelerating the formation of acrolein considerably. At higher temperatures catalyst reduction by propene is the rate determining step. The reaction temperature indicated a significant optimum for the selectivity to acrolein. This temperature approximately corresponds with the point where the rate determining step changes. Further, the special role of water in the reaction mixture could be described in detail: Water not only increases the selectivity to acrolein by suppressing the formation of carbon oxides, at low temperatures it further improves catalyst reoxidation remarkably.

Thus, the CWR showed to be a very useful tool for investigating highly exothermic reactions especially when changes of the reaction behaviour are closely linked to the temperature. These would not have been easily detectable in common fixed beds with nearly unavoidable temperature gradients.

Taking the experimental results of important side products into account a detailed reaction scheme consisting of consecutive and parallel reactions was developed. On the basis of an isothermal pseudo-homogeneous plug-flow model the experimental data were used to determine the kinetics of the oxidation of propene including 10 reactions and 30 significant and uncorrelated kinetic parameters. The main reaction of propene had to be divided into two sections to describe the sudden change of the rate determining step with reaction temperature. The resulting calculated data including all important side products were in very close accordance to the experimental values.

In parallel to the investigations in the laboratory CWR measurements in a pilot plant with a conventional fixed bed reactor under non-isotbermal conditions were performed. The independently in the CWR determined kinetics were used in a two dimensional dispersion model to describe the behaviour of the fixed bed reactor.

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 287 288 DGMK-Conference “Creating Value from Light Olefins - Production and Conversion", Hamburg, 2001

R. Loscher*, R. Hirsch*, E. de Armas**, R. Parry** *LECO Instrumente GmbH, Monchengladbach, Germany **LECO Corporation, St. Joseph, Ml, USA

Quantitative Analysis of Gasolines and Reformates by GC/TOFMS

Purpose of Gasoline Traditional Petroleum Analysts (Naphtha) Quantitation 1 hour 15 minutes

Knowledge of the composition of petroleum products is of ever increasing importance to the petroleum refining industry. Quantitation is invaluable for:

•Alcylation and Reforming Process Control •Product Quality Assessment •Regulatory Purposes •Environmental Concerns

DGMK-Tagungsbericht 2001-4, ISBN 3-931850-84-6 289 Fast Chromatographic Techniques

• Shorter Analytical Column e Increased Carrier Flow Rate e Increased Temperature

e Reduce Column Inner Diameter • Cryo Focusing Inlets e Multi-Dimensional Techniques

All Result in Reduced Chromatographic Peak Widths

Fast GC with MS Detection

# MS acquisition rate is the limiting factor for defining narrow chromatographic peaks 9 Time of Flight MS is the only commercially available instrument with acquisition rates sufficient for Fast GC Instrumentation TOF-MS

Pegasus II Diagram Push Pulse Plate Electron Focusing •Vial with Calibration Compound Optics Heated ? •Transfer line Q Ion Source Chamber Sampling

Z Steering and Deflection Plates Rough T’&.-J

Y Steering Plates It ;■ Turbomolecular| ff

Time-of-Flight -MS TOF Ion Ratios don’t change!

Seconds 43.6 43.8 44 44.2 44.4 4AS 44.8 45 1—91 I

In TOF/MS systems Ion Ratios remain constant across a Chromatographic Peak. This sets the path ASTM-P0030 PIANO Standard

• LECO® Pegasus* IITOFMS 25 spectra/second (45 to 450u) • 0.5pL injection, Split 400:1 • 30m x 0.25mm ID, 0.25pm film (SPB-1) • 40°C hold for 2.0 min., ramp to 250°C at 10°C/min^ hold 2.0 min.

i1 n t " mmi Secmfc 100 ZB XC ICC_____SB SB re so sb looo

ASTM-P0030 PIANO Standard Coelution Example 1

Setmds 310.5 311 311.5 312 3115 313 313.5 314 314.5 315 [——TlCxO.3 —^8lT

292 ASTM-P0030 PIANO Standard Coelution Example 1

3113 3123 3115 3145 '——TIC ■ TIC

With automated deconvolution the software is able to determine the amount of the TIC that belongs to each peak. In this way an accurate area % can be determined.

Area % Results for PIANO Standard 70 to 118 seconds

Analyte Type RT (sec) Area Area % Weight % 1-Butene, 3-methyl- Olefin 73,919 14285000 0,15204 0,3545 Butane, 2-methyl- (CAS) i Iso-Paraffin 75,519 8351200 0,088884 0.4143 1-Pentene Olefin 76,959 23348000 0,2485 0,7554 1-Butene, 2-methyl- Olefin 77,679 14241000 0,15157 0,264 Pentane (CAS) Paraffin 78,279 10627000 0,11311 1,7846 1,3-Butadiene, 2-methyl- Olefin 78,879 17310000 0,18424 0,4366 2-Pentene, (Z)- Olefin 79,159 13372000 0,14232 0,3617 2-Pentene, (E> Olefin 80,199 16757000 0,17836 0,3296 1-Pentene, 4-methyl- Olefin 87,279. 18750000 0,19956 0,6281 Cyclopentane Naphthene 88,999 35683000 0,37979 1,0285 Pentane, 2-methyl- (CAS) Iso-Paraffin 90,079 13858000 0,14749: 0,6199 Pentane, 3-methyl- Iso-Paraffin 93.599 58437000 0.62196 1.0176 1- Olefin 95,039 48970000 0,52121 1,2881 Paraffin 98,279 117060000 1,246 1,809 2-Hexene, (E)- Olefin 99,719 22573000 0,24026 0,3162 2-Pentene, 2-methyl- Olefin 100,44 34964000 0,37214 0,6195 2-Hexene, (Z)- Olefin 102,8 41736000 0,44422 0,7084 Pentane, 2,2-dimethyl- Iso-Paraffin 106,68 24553000: 0,26132 0,3353 Cyclopentane, methyl- Naphthene 107,68 50922000 0,54199 0,6871 Pentane, 2,4-dimethyl- Iso-Paraffin 109 35760000 0,3806 0,6988 Butane, 2,2,3-trimethyl- Iso-Paraffin 111,36 54777000 0,58301 0,7423 Benzene Aromatic : 117,68 131640000 1,4011 1,6446

293 TIC Overlay of PIANO Standard and Reformate Sample

Area % Results for Reformate Sample 70 to 118 seconds

Name Type R.T. Area Area % Butanai, 2-methyl- Iso-Paraffin 68,67 3144000 0,05374 Butane Paraffin 70,11 18643000 0,31865 143utene Olefin 71,39 664970 0,011366 1-Butene, 3-methyl- Olefin 73,75 636220 0,010875 Butane, 2-methyl* Iso-Paraffin 75,31 71167000- 1,2164 1- Pentene Olefin 77,51 3733600 0,063816 Pentane Paraffin 78,07 33395000 0,5708 2- Pentene Olefin 78,99 2152800: 0,036797 2-Pentene, (E)- Olefin 80,75 6049900 0,10341 Butane, 2,2 Olefin 86,55 528010 0,009025 i-Pentene, 4-methyl- Olefin 87,47 843710 0,014421 Pentane. 2-methyl- Iso-Paraffin 89,91 68798000 1,1759 Pentane, 3-methyl- Iso-Paraffin 93,43 118710000 2,0291 1- Pentene, 2-methyl- Olefin 94,71 3073000. 0,052525 Hexane Paraffin 98,15 124910000 2,1351 2- Pentene, 4-methyl- Olefin 100,31 2955600 0,050519 1.3- Butadiene, 2-ethyi- Olefin 101,27 729990 0,012477 2-Hexene Olefin 102,67 1056500 0,018058 2-Pentene, 3-methyi-, {Zy- Olefin 104,87 2381900 0,040713 Pentane, 2,2-dimethyl- Iso-Paraffin 106,59 17894000 0,30586 Cyciopentane, methyl- Naphthene 107,55 50775000 0,86788 Pentane, 2,4-dimethyl- Iso-Paraffin 108,91 28216000 0,48229 Butane, 2,2,3-trimethyi- Iso-Paraffin 111,31 2013200 0,034411 Benzene Aromatic 117.55 244610000 4,1811 Conclusions

9 PIANO Standard 139 Analytes in 16 minutes ® Reformate 242 Analytes in 16 minutes

GC/TOFMS Allows for Accelerated Analysis of Petroleum Products

Coeluting Peaks Can Be Used for Area % Calculations with Deconvoluted TIC

This presentation showed that by applying Fast GC/TOFMS to routine petroleum analysis, typically running 1-2 hours, the analysis can be dramatically reduced, 10-15 x faster and still retain analytical objectives while improving productivity.

295