Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Catalytic hydroconversion of diphenylmethane with unsupported MoS2 Kukard, Ross S. 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_november_kukard_ross.pdf [ 13.32MB ]
Metadata
JSON: 24-1.0166978.json
JSON-LD: 24-1.0166978-ld.json
RDF/XML (Pretty): 24-1.0166978-rdf.xml
RDF/JSON: 24-1.0166978-rdf.json
Turtle: 24-1.0166978-turtle.txt
N-Triples: 24-1.0166978-rdf-ntriples.txt
Original Record: 24-1.0166978-source.json
Full Text
24-1.0166978-fulltext.txt
Citation
24-1.0166978.ris

Full Text

Catalytic Hydroconversion of Diphenylmethane withUnsupported MoS2byRoss S. KukardB.Sc. Eng. Chem. (hons), The University of Cape Town, 2006M.Sc. Eng. Chem., The University of Cape Town, 2009A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Chemical and Biological Engineering)The University of British Columbia(Vancouver)September 2014© Ross S. Kukard, 2014AbstractThe mechanism by which hydroconversion catalysts promote residue conversionand coke suppression is unclear. Several theories are proposed in the literature butthese have all been opposed, usually due to their lack of controlled mechanisticstudies. A promising catalyst for residue hydroconversion is unsupported MoS2.This catalyst is effective but expensive and deactivates during the reaction. Modelcompound studies were needed to elucidate the mechanism of MoS2 catalysis inhydroconversion reactions, how this relates to residue hydroconversion and hencepropose deactivation mechanisms and regeneration methodologies.Model compound screening in a commercially available stirred slurry-phasebatch reactor identified diphenylmethane (DPM) as a suitable model reagent. Ex-periments were conducted at industrially applicable conditions of 445◦C, 13.8 MPaH2 and catalyst loadings of 0 - 1800 ppm Mo (introduced as Mo octoate whichformed the MoS2 active phase in-situ). Slow heat-up rates and wall catalysis, how-ever, made this reactor unsuitable for detailed mechanistic studies. A novel mixedslurry-phase micro-reactor system was designed using externally applied vortexmixing and removable glass-inserts to allow for greater analytical resolution anddetermination of the thermocatalytic mechanism. Deactivated MoS2 catalysts, ascoke-catalyst agglomerates recovered from residue hydroconversion studies [1],were evaluated using the DPM testing methodology and a deactivation mechanismproposed.It was determined that the unsupported MoS2 crystallites hydrogenate the DPMfeed to cyclohexylmethylbenzene (CHMB) which undergoes thermolysis to shortchain hydrocarbon radicals. These short chain radicals stabilise, by radical addi-tion or radical disproportionation, other radicals in the system by a chain stabili-iisation reaction, itself promoted by catalytic hydrogenation (for instance of olefinsformed during disproportionation). Deactivation of unsupported MoS2 in residuehydroconversion was proposed to be due to the formation of an unreactive, porouscarbonaceous structure upon which the otherwise unaltered catalyst particles be-come supported. The pores physically exclude larger species, such as asphaltenes,from reaching the active sites.Inter-recycle solvent extraction to remove coke precursors was proposed to in-hibit deactivation in residue hydroconversion whilst mechanical and chemical sizereduction were suggested for breaking the porous structure and re-exposing theMoS2 crystallites.iiiPrefaceAll of the work presented henceforth was conducted in the Department of Chemi-cal and Biological Engineering at the University of British Columbia, Point Greycampus.I, Ross S. Kukard, was the lead investigator of this work, responsible for allmajor areas of concept formation, micro-reactor design, construction and commis-sioning and all experimentation, data collection and analysis as well as preparationof this thesis. Kevin J. Smith was the supervisor of this research, involved through-out the project in concept formation and thesis edits. Hooman Rezaei commis-sioned the stirred batch reactor described in Section B.2.1 which I used to conductthe experiments and collect the data in Section 4.1. He also conducted the residuehydroconversion experiments to generate deactivated catalyst samples and compar-ison data (Section 4.2.3). The mechanical workshop in the Department of Chemi-cal and Biological Engineering, led by Doug Yuen, constructed the micro-reactorenclosures described in Section B.3.1 using designs that I prepared.I presented data from Section 4.1 at the 62nd Canadian Chemical Engineer-ing Conference (Vancouver, BC, Canada), 2012 and data from Section 4.2 at the23rd Canadian Symposium on Catalysis (Edmonton, AB, Canada), 2014. Kevin J.Smith was involved in the preparation of these presentations.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiNomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxixDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1 Current Hydroconversion Technology . . . . . . . . . . . . . . . 52.2 Residue Processing Technologies . . . . . . . . . . . . . . . . . . 62.2.1 Carbon Rejection . . . . . . . . . . . . . . . . . . . . . . 72.2.2 Hydroconversion . . . . . . . . . . . . . . . . . . . . . . 82.2.3 Slurry-Phase Catalytic Hydroconversion . . . . . . . . . . 102.3 Catalyst Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.3.1 Heavy Oil and Residue Oil Studies . . . . . . . . . . . . 12v2.3.2 Model Compound Studies . . . . . . . . . . . . . . . . . 132.4 Micro-Reactors for Catalyst Testing . . . . . . . . . . . . . . . . 162.4.1 Advantages and Disadvantages of Micro-Reactors . . . . 162.4.2 Micro-Reactors in Hydroconversion Studies . . . . . . . . 172.5 Catalyst Activity and Deactivation . . . . . . . . . . . . . . . . . 182.5.1 Catalyst Selection . . . . . . . . . . . . . . . . . . . . . 192.5.2 Molybdenum Disulphide . . . . . . . . . . . . . . . . . . 202.5.3 Processes of Deactivation . . . . . . . . . . . . . . . . . 262.5.4 Catalyst Regeneration Methodologies . . . . . . . . . . . 282.6 Summary of Findings from the Literature . . . . . . . . . . . . . 293 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.1 Experimental Objectives and Programme . . . . . . . . . . . . . 303.1.1 Interpretation of Questions to Objectives . . . . . . . . . 313.1.2 Experimental Programme . . . . . . . . . . . . . . . . . 353.2 Experimental Apparatus and Supplies . . . . . . . . . . . . . . . 393.2.1 Reaction and Analytical Supplies . . . . . . . . . . . . . 393.2.2 Reactors and Conditions . . . . . . . . . . . . . . . . . . 403.3 Analytical Equipment and Data Analysis . . . . . . . . . . . . . . 533.3.1 Gas Product Analysis . . . . . . . . . . . . . . . . . . . . 533.3.2 Liquid Product Analysis . . . . . . . . . . . . . . . . . . 543.3.3 Solid Product Analysis . . . . . . . . . . . . . . . . . . . 544 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 554.1 Stirred Batch Reactor . . . . . . . . . . . . . . . . . . . . . . . . 554.1.1 Model Compound Screening . . . . . . . . . . . . . . . . 554.1.2 Benzene, Toluene and Decalin Blanks . . . . . . . . . . . 594.1.3 Diphenylmethane Studies . . . . . . . . . . . . . . . . . 604.2 Batch Micro-reactor . . . . . . . . . . . . . . . . . . . . . . . . . 834.2.1 Inclined Stainless Steel Micro-Reactor . . . . . . . . . . . 834.2.2 Vertical Stainless Steel Micro-Reactor . . . . . . . . . . . 884.2.3 Glass Insert Micro-Reactor . . . . . . . . . . . . . . . . . 101vi5 Discussion of Experimental Results . . . . . . . . . . . . . . . . . . 1405.1 Model Compound Evaluation . . . . . . . . . . . . . . . . . . . . 1405.1.1 Model Compound Screening . . . . . . . . . . . . . . . . 1405.1.2 Diphenylmethane Studies . . . . . . . . . . . . . . . . . 1425.1.3 Summary of Model Compound Evaluation . . . . . . . . 1555.2 Novel Reactor System Design and Testing . . . . . . . . . . . . . 1575.2.1 Inclined Stainless Steel Micro-Reactor . . . . . . . . . . . 1585.2.2 Vertical Stainless Steel Micro-Reactor . . . . . . . . . . . 1625.2.3 Unmixed Glass Insert Micro-Reactor . . . . . . . . . . . 1695.2.4 Mixed Glass Insert Micro-Reactor . . . . . . . . . . . . . 1715.2.5 Summary of Micro-Reactor System Design and Testing . 1835.3 Catalyst Study and Deactivation Investigation . . . . . . . . . . . 1855.3.1 Active MoS2 . . . . . . . . . . . . . . . . . . . . . . . . 1855.3.2 Deactivated Coke-MoS2 Agglomerate . . . . . . . . . . . 1875.3.3 Mechanism of MoS2 Deactivation in Residue Hydrocon-version . . . . . . . . . . . . . . . . . . . . . . . . . . . 1925.3.4 Summary of MoS2 Activity and Deactivation . . . . . . . 1956 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218A Catalyst Deactivation and Regeneration . . . . . . . . . . . . . . . . 219A.1 Processes of Catalyst Deactivation . . . . . . . . . . . . . . . . . 219A.1.1 Fouling . . . . . . . . . . . . . . . . . . . . . . . . . . . 219A.1.2 Poisoning . . . . . . . . . . . . . . . . . . . . . . . . . . 222A.1.3 Others . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227A.2 Catalyst Regeneration Processes . . . . . . . . . . . . . . . . . . 230B Experimental Apparatus and Procedures . . . . . . . . . . . . . . . 233B.1 Detailed Experimental Programme . . . . . . . . . . . . . . . . . 233B.2 Batch Reactor Specifications and Operation . . . . . . . . . . . . 234viiB.2.1 Description and Specifications . . . . . . . . . . . . . . . 241B.2.2 Operating Procedure . . . . . . . . . . . . . . . . . . . . 244B.2.3 Safety Considerations . . . . . . . . . . . . . . . . . . . 255B.3 Micro-Reactor Design, Development and Operation . . . . . . . . 257B.3.1 Design and Development . . . . . . . . . . . . . . . . . . 257B.3.2 Operating Procedure . . . . . . . . . . . . . . . . . . . . 276B.3.3 Safety Considerations . . . . . . . . . . . . . . . . . . . 287C Analytical Apparatus, Procedures and Data Analysis . . . . . . . . . 289C.1 Gas Product Analysis . . . . . . . . . . . . . . . . . . . . . . . . 289C.1.1 Analytical Equipment and Procedures . . . . . . . . . . . 289C.1.2 Calibration, Data Acquisition, Analysis and Interpretation 292C.2 Liquid Product Analysis . . . . . . . . . . . . . . . . . . . . . . 296C.2.1 Analytical Equipment and Procedures . . . . . . . . . . . 296C.2.2 Data Acquisition, Analysis and Interpretation . . . . . . . 301C.2.3 Calibration, Analysis and Experimental Uncertainty . . . 307C.3 Solid Product Analysis . . . . . . . . . . . . . . . . . . . . . . . 318C.3.1 Analytical Equipment and Procedures . . . . . . . . . . . 318C.3.2 Calibration, Data Acquisition, Analysis and Interpretation 320D Calibration Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333D.1 Gas Chromatograph . . . . . . . . . . . . . . . . . . . . . . . . . 333D.1.1 HP5980A Calibration Results . . . . . . . . . . . . . . . 333D.1.2 Shimadzu GC-14B Calibration Results . . . . . . . . . . 333D.2 GCMS-QP2010 Gas Chromatograph - Mass Spectrometer LiquidCalibration Results . . . . . . . . . . . . . . . . . . . . . . . . . 335D.2.1 Benzene . . . . . . . . . . . . . . . . . . . . . . . . . . . 335D.2.2 Toluene . . . . . . . . . . . . . . . . . . . . . . . . . . . 337D.2.3 Diphenylmethane . . . . . . . . . . . . . . . . . . . . . . 339D.2.4 Diphenylethane . . . . . . . . . . . . . . . . . . . . . . . 341D.2.5 Diphenylpropane . . . . . . . . . . . . . . . . . . . . . . 343viiiE Detailed Experimental Results . . . . . . . . . . . . . . . . . . . . . 345E.1 Stirred Batch Reactor . . . . . . . . . . . . . . . . . . . . . . . . 345E.1.1 Model Compound Screening . . . . . . . . . . . . . . . . 345E.1.2 Diluted Diphenylmethane . . . . . . . . . . . . . . . . . 346E.1.3 Undiluted Diphenylmethane . . . . . . . . . . . . . . . . 347E.2 Stainless Steel Batch Micro-reactors . . . . . . . . . . . . . . . . 356E.2.1 Inclined Stainless Steel Micro-Reactor . . . . . . . . . . . 356E.2.2 Vertical Stainless Steel Micro-Reactor . . . . . . . . . . . 356E.3 Glass Insert Batch Micro-reactor . . . . . . . . . . . . . . . . . . 356E.3.1 Comparison with Stainless Steel Micro-Reactor . . . . . . 356E.3.2 Visual Mixing Studies . . . . . . . . . . . . . . . . . . . 356E.3.3 Comparison of Liquid Loading Volumes . . . . . . . . . . 356E.3.4 Thermocouple Wall Activity . . . . . . . . . . . . . . . . 356E.3.5 Effect of Mixing Speed . . . . . . . . . . . . . . . . . . . 375E.3.6 Optimum Mixing Speed Evaluation . . . . . . . . . . . . 375E.3.7 Spent Residue Hydroconversion Catalyst Evaluation . . . 376F Data Processing and Analysis . . . . . . . . . . . . . . . . . . . . . . 385F.1 Data Acquisition and Analysis . . . . . . . . . . . . . . . . . . . 385F.2 Gas and Liquid Mass Balances . . . . . . . . . . . . . . . . . . . 397F.2.1 Liquid Product . . . . . . . . . . . . . . . . . . . . . . . 397F.2.2 Gas Product . . . . . . . . . . . . . . . . . . . . . . . . . 398F.3 Kinetic Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . 402F.3.1 First Order . . . . . . . . . . . . . . . . . . . . . . . . . 402F.3.2 Second Order . . . . . . . . . . . . . . . . . . . . . . . . 403F.3.3 The Arrhenius Law . . . . . . . . . . . . . . . . . . . . . 404F.4 Thermodynamic Simulations . . . . . . . . . . . . . . . . . . . . 405F.5 Phase Density and Composition Simulations . . . . . . . . . . . . 407F.6 Physical Property Simulations . . . . . . . . . . . . . . . . . . . 408F.7 Hydrogen Solubility and Diffusivity . . . . . . . . . . . . . . . . 408F.7.1 Hydrogen Solubility Simulations . . . . . . . . . . . . . . 408F.7.2 Hydrogen Dissolution Rate . . . . . . . . . . . . . . . . . 426F.7.3 Hydrogen Diffusion . . . . . . . . . . . . . . . . . . . . 427ixF.8 Area:Volume Ratios . . . . . . . . . . . . . . . . . . . . . . . . . 431F.8.1 Gas-Liquid Interfacial Area . . . . . . . . . . . . . . . . 435F.9 Coke Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . 436xList of TablesTable 2.1 Comparison of physical properties of various crude oils andresidua. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Table 2.2 Comparison of fluidised bed and slurry-phase hydroconversionreactor performance. . . . . . . . . . . . . . . . . . . . . . . . 11Table 3.1 Summary of experimental programme. . . . . . . . . . . . . . 36Table 3.2 Stirred batch reactor operating conditions . . . . . . . . . . . . 43Table 3.3 Micro-reactor operating conditions . . . . . . . . . . . . . . . 49Table 4.1 Conversion results for model compound screening . . . . . . . 56Table 4.2 Major products for model compound hydroconversion . . . . . 58Table 4.3 Gaseous products for diphenylmethane hydroconversion . . . . 59Table 4.4 Benzene and toluene blank test conversions . . . . . . . . . . . 60Table 4.5 Major products for benzene blank tests . . . . . . . . . . . . . 60Table 4.6 Major products for toluene blank tests . . . . . . . . . . . . . 61Table 4.7 Major products for decalin blank test . . . . . . . . . . . . . . 61Table 4.8 Major products for undiluted diphenylmethane hydroconversion 69Table 4.9 Structures of hydroconversion side-products . . . . . . . . . . 70Table 4.10 Comparison of heating rate effect on diphenylmethane hydro-conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86Table 4.11 Major products for diphenylmethane hydroconversion in inclinedmicro-reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . 92Table 4.12 Coefficients for diphenylmethane hydroconversion kinetic mod-els in vertical micro-reactor . . . . . . . . . . . . . . . . . . . 94xiTable 4.13 Major gaseous products for diphenylmethane hydroconversionin vertical micro-reactor . . . . . . . . . . . . . . . . . . . . . 96Table 4.14 Coefficients for diphenylmethane hydroconversion kinetic mod-els in glass insert micro-reactor . . . . . . . . . . . . . . . . . 101Table 4.15 Comparison of liquid volume effect on diphenylmethane hydro-conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111Table 4.16 Scanning electron microscopy with energy dispersive X-ray quan-tification of solids from diphenylmethane hydroconversion after4 h . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119Table 4.17 Major constituents of isom./cond. lump for diphenylmethanehydroconversion in glass insert micro-reactor at 150 µL . . . . 134Table 4.18 Major gaseous products for diphenylmethane hydroconversionin glass insert micro-reactor . . . . . . . . . . . . . . . . . . . 135Table 4.19 Results from coke-catalyst agglomerate evaluation by diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . . 138Table 4.20 Major gaseous products for coke-catalyst agglomerate evalua-tion by diphenylmethane hydroconversion . . . . . . . . . . . 139Table 5.1 Physical property simulations for diphenylmethane hydrocon-version at various conditions . . . . . . . . . . . . . . . . . . 177Table 5.2 Solubility of coke-catalyst agglomerates in diphenylmethane . 192xiiList of FiguresFigure 2.1 Thermal cracking of diphenylmethane . . . . . . . . . . . . . 16Figure 2.2 Rendering of arbitrary 5-layer stack of MoS2 . . . . . . . . . 21Figure 2.3 Thermal decomposition of diphenylmethane . . . . . . . . . . 23Figure 2.4 Thermocatalytic decomposition of diphenylmethane . . . . . 25Figure 2.5 Surface reactions for hydrogen and diphenylmethane . . . . . 26Figure 2.6 Summary of main thermocatalytic decomposition mechanismsof diphenylmethane . . . . . . . . . . . . . . . . . . . . . . . 27Figure 3.1 Thermal cracking of DPM, DPE and DPP . . . . . . . . . . . 33Figure 3.2 Decahydronaphthalene structure. . . . . . . . . . . . . . . . . 33Figure 3.3 Reactor heating profile comparison . . . . . . . . . . . . . . 34Figure 3.4 Process flow diagram of stirred batch reactor . . . . . . . . . 44Figure 3.5 Laboratory implementation of stirred batch reactor . . . . . . 45Figure 3.6 Process flow diagram of micro-reactor . . . . . . . . . . . . . 50Figure 3.7 Laboratory implementation of micro-reactor . . . . . . . . . . 51Figure 4.1 Conversion results for diphenylpropane hydroconversion . . . 57Figure 4.2 Conversion results for diphenylmethane hydroconversion withsigmoidal trends . . . . . . . . . . . . . . . . . . . . . . . . 62Figure 4.3 Benzene molar yield for diphenylmethane hydroconversion . . 64Figure 4.4 Toluene molar yield for diphenylmethane hydroconversion . . 64Figure 4.5 Benzene:toluene molar ratio for diphenylmethane hydrocon-version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65xiiiFigure 4.6 Conversion results for undiluted diphenylmethane hydrocon-version with sigmoidal trends . . . . . . . . . . . . . . . . . 67Figure 4.7 Conversion with reaction temperature results for undiluted di-phenylmethane hydroconversion . . . . . . . . . . . . . . . . 67Figure 4.8 Logarithmic conversion with inverse reaction temperature forundiluted diphenylmethane hydroconversion . . . . . . . . . 68Figure 4.9 Benzene molar yield for undiluted diphenylmethane hydrocon-version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Figure 4.10 Toluene molar yield for undiluted diphenylmethane hydrocon-version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Figure 4.11 Cyclohexylmethylbenzene molar yield for undiluted diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 73Figure 4.12 Mass yield of other cracking products for undiluted diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 73Figure 4.13 Mass yield of isomerisation and condensation products for undi-luted diphenylmethane hydroconversion . . . . . . . . . . . . 74Figure 4.14 Benzene:toluene molar ratio for undiluted diphenylmethane hy-droconversion . . . . . . . . . . . . . . . . . . . . . . . . . . 74Figure 4.15 Benzene:toluene molar ratio against catalyst loading for undi-luted diphenylmethane hydroconversion . . . . . . . . . . . . 75Figure 4.16 Pressure change for undiluted diphenylmethane hydroconversion 76Figure 4.17 X-ray diffractogram for 600 ppm Mo in undiluted diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 78Figure 4.18 X-ray diffractogram for 1800 ppm Mo in undiluted diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 79Figure 4.19 Transmission electron microscopy image for 600 ppm Mo inundiluted diphenylmethane hydroconversion . . . . . . . . . 80Figure 4.20 Transmission electron microscopy image for 1800 ppm Mo inundiluted diphenylmethane hydroconversion . . . . . . . . . 81Figure 4.21 MoS2 sheet size distribution for 600 ppm Mo in undiluted di-phenylmethane hydroconversion . . . . . . . . . . . . . . . . 82Figure 4.22 MoS2 stack height distribution for 600 ppm Mo in undiluteddiphenylmethane hydroconversion . . . . . . . . . . . . . . . 82xivFigure 4.23 Conversion results to study wall activation during diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 84Figure 4.24 Conversion results comparing wall activity during diphenyl-methane hydroconversion . . . . . . . . . . . . . . . . . . . 85Figure 4.25 Conversion results for diphenylmethane hydroconversion in in-clined micro-reactor . . . . . . . . . . . . . . . . . . . . . . 87Figure 4.26 Benzene molar yield for diphenylmethane hydroconversion ininclined micro-reactor . . . . . . . . . . . . . . . . . . . . . 88Figure 4.27 Toluene molar yield for diphenylmethane hydroconversion ininclined micro-reactor . . . . . . . . . . . . . . . . . . . . . 89Figure 4.28 Cyclohexylmethylbenzene molar yield for diphenylmethane hy-droconversion in inclined micro-reactor . . . . . . . . . . . . 89Figure 4.29 Mass yield of other cracking products for diphenylmethane hy-droconversion in inclined micro-reactor . . . . . . . . . . . . 90Figure 4.30 Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion in inclined micro-reactor . . 90Figure 4.31 Benzene:toluene molar ratio for diphenylmethane hydrocon-version in inclined micro-reactor . . . . . . . . . . . . . . . . 91Figure 4.32 Conversion results for diphenylmethane hydroconversion in ver-tical micro-reactor . . . . . . . . . . . . . . . . . . . . . . . 93Figure 4.33 Benzene molar yield for diphenylmethane hydroconversion invertical micro-reactor . . . . . . . . . . . . . . . . . . . . . . 95Figure 4.34 Toluene molar yield for diphenylmethane hydroconversion invertical micro-reactor . . . . . . . . . . . . . . . . . . . . . . 96Figure 4.35 Cyclohexylmethylbenzene molar yield for diphenylmethane hy-droconversion in vertical micro-reactor . . . . . . . . . . . . 97Figure 4.36 Fluorene molar yield for diphenylmethane hydroconversion invertical micro-reactor . . . . . . . . . . . . . . . . . . . . . . 97Figure 4.37 Benzene:toluene molar ratio for diphenylmethane hydrocon-version in vertical micro-reactor . . . . . . . . . . . . . . . . 98Figure 4.38 Gas chromatograms for diphenylmethane hydroconversion invertical micro-reactor with 0 ppm Mo . . . . . . . . . . . . . 99xvFigure 4.39 Gas chromatograms for diphenylmethane hydroconversion invertical micro-reactor with 1800 ppm Mo . . . . . . . . . . . 100Figure 4.40 Comparison of conversion results for diphenylmethane hydro-conversion in stainless steel and glass insert micro-reactors . . 102Figure 4.41 Benzene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor . . . . . . . . . . . . . . . . . . . . 103Figure 4.42 Toluene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor . . . . . . . . . . . . . . . . . . . . 104Figure 4.43 Cyclohexylmethylbenzene molar yield for diphenylmethane hy-droconversion in glass insert micro-reactor . . . . . . . . . . 104Figure 4.44 Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion in glass insert micro-reactor . 105Figure 4.45 Benzene:toluene molar ratio for diphenylmethane hydrocon-version in glass insert micro-reactor . . . . . . . . . . . . . . 105Figure 4.46 Mixing of 400 µL diphenylmethane hydroconversion reactionproduct in glass mock-up . . . . . . . . . . . . . . . . . . . . 106Figure 4.47 Effect of mixing speed on “vortex” height in glass mock-up fordiphenylmethane hydroconversion reaction product . . . . . . 107Figure 4.48 Mixing of 150 µL diphenylmethane hydroconversion reactionproduct in glass mock-up . . . . . . . . . . . . . . . . . . . . 108Figure 4.49 Mixing of 400 µL diphenylmethane hydroconversion reactionproduct in glass mock-up with thermocouple . . . . . . . . . 109Figure 4.50 Mixing of 150 µL diphenylmethane hydroconversion reactionproduct in glass mock-up with thermocouple . . . . . . . . . 110Figure 4.51 Conversion results to study thermocouple wall activation dur-ing diphenylmethane hydroconversion . . . . . . . . . . . . . 112Figure 4.52 Benzene and toluene molar yields for thermocouple wall acti-vation during diphenylmethane hydroconversion . . . . . . . 113Figure 4.53 Benzene:toluene molar ratio for thermocouple wall activationduring diphenylmethane hydroconversion . . . . . . . . . . . 113Figure 4.54 Cyclohexylmethylbenzene molar yield for thermocouple wallactivation during diphenylmethane hydroconversion . . . . . . 114xviFigure 4.55 Mass yield of isomerisation and condensation products for ther-mocouple wall activation during diphenylmethane hydrocon-version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Figure 4.56 Influence of mixing speed on conversion results for diphenyl-methane hydroconversion in glass insert micro-reactors . . . . 116Figure 4.57 Benzene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor . . . . . . . . . . . . . . . . . . . . 116Figure 4.58 Toluene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor . . . . . . . . . . . . . . . . . . . . 117Figure 4.59 Benzene:toluene molar ratio for diphenylmethane hydrocon-version in glass insert micro-reactor . . . . . . . . . . . . . . 117Figure 4.60 Cyclohexylmethylbenzene molar yield for diphenylmethane hy-droconversion in glass insert micro-reactor . . . . . . . . . . 118Figure 4.61 Transmission electron microscopy of solids from diphenylmeth-ane hydroconversion at 0 RPM . . . . . . . . . . . . . . . . . 121Figure 4.62 Transmission electron microscopy of solids from diphenylmeth-ane hydroconversion at 2000 RPM . . . . . . . . . . . . . . . 122Figure 4.63 Scanning electron microscopy with energy dispersive X-rayimage of solids from diphenylmethane hydroconversion after 4 h123Figure 4.64 Field emission scanning electron microscopy images of solidsfrom diphenylmethane hydroconversion at 0 RPM . . . . . . 124Figure 4.65 Field emission scanning electron microscopy images of solidsfrom diphenylmethane hydroconversion at 2000 RPM . . . . 125Figure 4.66 Field emission scanning electron microscopy images of solidsfrom diphenylmethane hydroconversion at 2250 RPM with usualheat-up mixing . . . . . . . . . . . . . . . . . . . . . . . . . 126Figure 4.67 Field emission scanning electron microscopy images of solidsfrom diphenylmethane hydroconversion at 2250 RPM withoutheat-up mixing . . . . . . . . . . . . . . . . . . . . . . . . . 127Figure 4.68 Angled field emission scanning electron microscopy images ofsolids from diphenylmethane hydroconversion at 2250 RPM . 128Figure 4.69 Conversion results for diphenylmethane hydroconversion in glassinsert micro-reactor at 2000 RPM . . . . . . . . . . . . . . . 129xviiFigure 4.70 Benzene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor at 150 µL . . . . . . . . . . . . . . 130Figure 4.71 Toluene molar yield for diphenylmethane hydroconversion inglass insert micro-reactor at 150 µL . . . . . . . . . . . . . . 131Figure 4.72 Benzene:toluene molar ratio for diphenylmethane hydrocon-version in glass insert micro-reactor at 150 µL . . . . . . . . 131Figure 4.73 Cyclohexylmethylbenzene molar yield for diphenylmethane hy-droconversion in glass insert micro-reactor at 150 µL . . . . . 132Figure 4.74 Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion in glass insert micro-reactorat 150 µL . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133Figure 5.1 Thermocatalytic decomposition products of benzene and toluene144Figure 5.2 Thermodynamic simulations of benzene and toluene decom-position to methane . . . . . . . . . . . . . . . . . . . . . . . 144Figure 5.3 Thermodynamic simulations of phenyl and benzyl radical de-composition . . . . . . . . . . . . . . . . . . . . . . . . . . . 147Figure 5.4 Phase simulations of reaction mixture . . . . . . . . . . . . . 153Figure 5.5 Formation of fluorene and hexahydrofluorene from diphenyl-methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154Figure 5.6 Proposed thermocatalytic decomposition mechanism of diphenyl-methane in stirred batch reactor . . . . . . . . . . . . . . . . 156Figure 5.7 Comparison of thermolysis mechanisms of cyclohexylmethyl-benzene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160Figure 5.8 Proposed thermocatalytic decomposition mechanism of diphenyl-methane in inclined micro-reactor . . . . . . . . . . . . . . . 163Figure 5.9 Pressure drop to study H2 dissolution in diphenylmethane . . 165Figure 5.10 Modeled concentration profiles to study H2 diffusion throughdiphenylmethane . . . . . . . . . . . . . . . . . . . . . . . . 166Figure 5.11 Comparison of H* abstraction mechanisms from diphenylmeth-ane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167Figure 5.12 Comparison of phenyl radical stabilisation mechanisms . . . . 169xviiiFigure 5.13 MoS2 crystallite formation and movement in glass insert micro-reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182Figure 5.14 Proposed thermocatalytic decomposition mechanism of diphenyl-methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188Figure 5.15 Asphaltene model compound bibenzyl-cholestane . . . . . . . 190Figure 5.16 Proposed mechanism for unsupported catalyst deactivation inresidue hydroconversion . . . . . . . . . . . . . . . . . . . . 195xixNomenclatureRoman Symbols∆GTr Gibbs free energy of reaction at temperature, T , in kJ/mol∆HTr Enthalpy of reaction at temperature, T , in Ha, kcal/mol or kJ/molA : V Ratio of reactor and internals surface area to the volume of liquid re-action mixture, cm−1aext Linear fit parameters for response factor calibrations, dimensionlessaint Linear fit parameters for internal standards, dimensionlessAgc Peak area from gas chromatography analysis, counts.timeb Peak broadening factor for the Scherrer equation using the FWHMmethod, radiansbint Linear fit parameters for internal standards, dimensionlessBT Benzene:toluene molar ratio, dimensionlessC Mass concentration, wt%c Mass composition, fractionalC′ Molar concentration, mol%Convx.y Conversion factor from x to y units, (kcal/mol)/Ha or kJ/kcalxxD Mass diffusivity in calculation of the Sherwood number (see Section 5.2.4),m2/sDH Vessel diameter in calculation of the Reynolds number (see Section 5.2.4),mEx.Total Total energy of species x from Accelrys Materials Studio simulations,HaEa Activation energy, J/molF Spherical packing factor, dimensionlessh Height, mHx Enthalpy of species x, Ha, kcal/mol or kJ/molHcx Enthalpy correction to total energy for species x from Accelrys Mate-rials Studio simulations, HaID Inner diameterIQ Interquartile rangeK Scherrer constant, 0.76 used for MoS2, dimensionlessk Reaction rate coefficient, h−1k′ Reaction rate coefficient modified to account for catalyst loading, h−1.wt%−1catk′′ Reaction rate coefficient for a constant time experiment, h−1KSh Mass transfer coefficient in calculation of the Sherwood number (seeSection 5.2.4), m/sL Length, mLSh Characteristic length in calculation of the Sherwood number (see Section 5.2.4),mm Mass, gxxiMr Molar mass, g/moln Number of moles, molOD Outer diameterP Pressure, Pap Crystallite size from the Scherrer equation, nmQ Molar ratio, dimensionlessQ′ Volumetric ratio, dimensionlessQx Q1 and Q3 being the first and third quartiles respectivelyR Ideal gas constant, 8.3144621 J/mol.Kr′ Reaction rate on a mass basis, mol/h.wt%catRe Reynolds number (see Section 5.2.4), dimensionlessRgc Response factor for gas chromatography analysis, countss Standard deviation, dimensionlessSc Schmidt number (see Section 5.2.4), dimensionlessSh Sherwood number (see Section 5.2.4), dimensionlessSS SSresid is the sum of squares of the residuals and SStotal is the total sumof squares (the difference between the experimental and the average)T Temperature, Kt Time, sV Volume, m3v Mean fluid velocity in calculation of the Reynolds number (see Section 5.2.4),m/sxxiiVBP.i Molar volume of species i at its normal boiling point, m3/molX Mass conversion, fractionalX ′ Mass conversion, wt%Y Molar yield, molproduct formed/molreagent consumedY ′ Mass yield, gproduct formed/greagent consumedGreek Symbolsβ Integral breadth for use in the Scherrer equation (see Section C.3.2),radiansη Wilke-Chang parameter (see Section F.7)λ Wavelength in X-ray diffraction analysis, nmµ Dynamic viscosity, kg/m.sν Kinematic viscosity, defined as µ/ρ , m2/sφ Mass concentration ratio, dimensionlessφ ′ GCMS area ratio, dimensionlessφ ′′ Volumetric ratio, dimensionlessφ ∗ Solvent association parameter for Wilke-Chang correlation (see Section F.7)ρ Mass density, kg/m3θ Bragg angle of basal peak for use in the Scherrer equation (see Section C.3.2),radiansSubscripts.in At the beginning.out At the endxxiiiadded Mixed into solution when diluting samples for analysisatm AtmosphereBal BalanceBead Relating to the glass beads used as spacers in the glass insert micro-reactorBenz Benzenecal From or relating to calibration, e.g. φcal is the calibrated φ valuecat CatalystDec Decalindil Desired dilution for loadingDPE DiphenylethaneDPM Diphenylmethaneest Estimated (used for estimating DPM conversion for sample dilution)ext Not using and internal standard, only a response factorflow Flow rateG.L Gas to liquid (ratio)GLTotal Total gas and liquidGTotal Total gasH2 Hydrogeni One in a series of samplesinsert Relating to the glass insertint Internal standardxxivLTotal Total liquidMeth MethaneMo Molebdynummodel Model compoundmol.CS2.Mo Moles of CS2 with respect to MoO2 Oxygenppm Desired loadingpre Precursor concentrationProd Total productProp Propanepur Purity of stock speciesresid ResidueRTotal Total reactorrxn Reactionsamp From sampleshell Relating to the stainless steel shell of the glass insert micro-reactorTarg Target species (the species of interest when performing calculations)Tol TolueneDefinitionsAgilent Agilent Technologies Inc.Brooks Brooks InstrumentxxvCoke Solid deposits, both carbonaceous and metallic (metal crystallites forexample) in nature formed during residue hydroconversion reactionswhich are, by definition insoluble in toluene at 20◦C and 1 atmConversion The percentage of model compound consumed during the reactionwith respect to that initially loaded into the reaction (see Section C.2.3)Inner fence The inner interquartile boundaries, defined as Q1− 1.5× IQ < x <Q3+1.5× IQ, used for outlier identification (see Section C.2.3)OMEGA OMEGA Engineering Inc.Outer fence The outer interquartile boundaries, defined as Q1−3×IQ< x<Q3+3× IQ, used for outlier identification (see Section C.2.3)Parr Parr Instrument CompanyResidue Specifically vacuum residue, fraction of oil-derived hydrocarbons boil-ing above 525◦CShimadzu Shimadzu Scientific Instruments Inc., a division of Shimadzu Corpo-rationTrace Species in the GCMS chromatogram representing less than 0.25 area%on a reagent- and diluent-free basisYield The ratio, molar-based or mass-based, of a particular product or groupof products with respect to the model compound consumed during thereaction (see Section C.2.3)AcronymsA:V Ratio of reactor and internals surface area to the volume of liquidreaction mixtureAPI American Petroleum Institute gravity is a comparison of the density ofan oil with respect to water, being defined as: ◦API = 141.5SGat15.6◦C −131.5xxviB:T Benzene:Toluene molar ratioBBCh Bibenzyl-cholestane. An asphaltene model compound proposed byAlshareef et al. [2] and discussed in Section 5.3.2.BBP 4-Benzylbiphenyl, see Table 4.9BPR Back-Pressure RegulatorCHMB Cyclohexylmethylbenzene, see Table 4.9DNM Di(1-naphthyl)methaneDPE DiphenylethaneDPM DiphenylmethaneDPP DiphenylpropaneEPB 1-Ethyl-2-(1-phenylethyl)-benzene, see Table 4.9ETB 1,1’,1”-(1-Ethanyl-2-ylidene)tris-benzene, see Table 4.9FID Flame Ionisation DetectorFWHM Full width at half maximum height. A technique used to determine thepeak broadening factor for use in the Scherrer equation (see Section C.3.2)GC Gas ChromatographGCMS Gas Chromatography - Mass SpectroscopyHexF 1,2,3,4,4a,9a-Hexahydrofluorene, see Table 4.9HP Hewlett PackardIUPAC International Union of Pure and Applied ChemistsLEL Lower Explosive LimitLHHW Langmuir-Hinshelwood-Hougen-Watson kineticsxxviiMBP One of several methyl-substituted biphenyl species as shown in Table 4.9MFC Mass Flow ControllerMFM Mass Flow MeterMS Mass SpectroscopyMTP 2-Methyl-1,1,1-triphenyl-propane, see Table 4.9STP Standard Temperature and Pressure, defined as 273.15 K 100 kPaTC ThermocoupleTCD Thermal Conductivity DetectorXRD X-ray DiffractionxxviiiAcknowledgementsThe author would like to gratefully acknowledge the efforts and support of allof the individuals and organisations involved in the development of this researchendeavour, specifically:Prof. Kevin J. Smith for the opportunity to conduct this research, support in de-termining the direction and focus for this endeavour, and guidance in thedevelopment of this report.Prof. Naoko Ellis, Prof. Marek Pawlik and Prof. Paul Watkinson for their roleas supervisory committee and the invaluable information, assistance and di-rection they have provided.Hooman Rezaei for assistance operating the stirred batch reactor, assistance withanalyses and preparation of spent coke-catalyst samples.Sharhzad Jooya Ardakani for assistance with experimental analyses.Victoria Whiffen, Shahin Goodarznia, Farnaz Sotoodeh and Mina Alyani for gen-eral assistance in the lab.Richard Ryoo, Doug Yuen and all of the other “behind-the-scenes” administra-tive staff without whose tireless efforts research could not be conducted.The author would also like to thank UOP LLC, Alberta Energy Research Insti-tute (AERI) and National Science and Engineering Research Council (NSERC) ofCanada for financial support for this project.xxixDedicationThis thesis is dedicated to all those who helped me reach this point in my life:family, friends and colleagues. In particular those who weren’t even around to seeit begin ... I’m looking at you, Charlotte!xxxChapter 1IntroductionAs the supply of lighter, easier to process crude oil declines, oil refineries are shift-ing to heavier, more contaminated crude oils. These heavier oils pose numerouschallenges to the refineries. Foremost are the elevated yields of atmospheric andvacuum distillation residua obtained from their processing. This residue is valuableas it may be converted to liquid fuel products. Unfortunately, heavier oils containhigher levels of contaminants such as S, N and metals which are concentrated in theresidue. These contaminants complicate residue processing and pose an environ-mental hazard if not recovered and disposed of appropriately. This research aimsto contribute toward an effective residue hydroconversion technology by studyingthe activity and deactivation of an unsupported molybdenum sulphide catalyst in aslurry-phase hydroconversion reactor.Slurry-phase residue hydroconversion reactors are favoured for their high lev-els of residue conversion and high yields of valuable liquid products. Such systemsalso exhibit reduced “coke” yields (solid carbonaceous and metallic deposits) andincreased contaminant removal compared to alternatives such as fixed-bed reactors.These latter two points result in less environmentally damaging solid waste beinggenerated. MoS2-based catalysts offer improved residue hydroconversion perfor-mance over, for example, FeS-based alternatives. Unfortunately, such MoS2 cata-lysts are also more expensive with Mo metal being several hundred times more ex-pensive than Fe. To be economically feasible, deactivated MoS2 catalysts must beregenerated and recycled. A thorough understanding of the mechanism by which1such a catalyst functions and the process by which it deactivates is needed so that,in future, an efficient regeneration methodology can be developed.BackgroundOil refineries have two main options for the processing of residue oil feedstocks:carbon rejection and hydrogen addition. Hydrogen addition, performed in the pres-ence of a suitable catalyst, is generally considered to be more efficient, affordingimproved conversion, yield of the desired high value liquid products and contam-inant removal. Hydrogen addition may be performed in a fixed bed, fluidised bedor slurry-phase reactor. Slurry-phase reactors, whilst not as widely implemented inindustrial hydroconversion applications as the other two systems, have been shownto offer various benefits over competing technologies. Many of these slurry opera-tions use Fe-based catalysts on a “once-through” basis due to the low cost and easeof supply of such material. Research into alternative unsupported metal catalystshas led to the development of better performing Mo-based options. Such catalysts,however, demand prolonged operation through recycle and reuse to be economi-cally feasible, a process that requires a thorough understanding of the deactivationof the catalyst together with an efficient regeneration methodology.Unfortunately, whilst the deactivation of supported metal catalysts (such asthose utilised in fixed and fluidised bed hydroconversion applications) has beenextensively studied and effective regeneration techniques developed, there remainsrelatively limited research regarding the deactivation of the unsupported metal cat-alysts of interest for slurry-phase hydroconversion. It is believed that the deactiva-tion of unsupported Mo-based catalysts is the result of some interaction betweenthe active MoS2 phase and coke formed during reaction. Unfortunately, neither thenature of this interaction nor the mechanism by which the active catalyst functions,is well understood with research on the topic being limited to conceptual models.Understanding these processes is key to regeneration for, if the deactivation couldsomehow be inhibited, reversed or otherwise “reset” to regain the functionality ofthe active catalyst between recycles, the catalyst may be used repeatedly with areduced degradation in performance.2HypothesesThe hypotheses of this research project may be stated as follows:1. A mechanistic understanding of hydroconversion affected by active and de-activated unsupported MoS2 catalysts will allow for the deduction of the pro-cess of deactivation of such catalysts in residue hydroprocessing reactions.2. The deactivation is due to the morphology of the coke, which precipitatesduring the reaction and agglomerates with the catalyst particles, changingwith continued recycling of the coke-catalyst solid material recovered afterthe reaction. The hardening of the coke results in the formation of a solid,unreactive “support” for the catalyst and subsequent physical deactivation.Key QuestionsThese hypotheses lead to four major questions to be addressed in this investigationas presented below, together with minor questions where necessary to clarify thespecific concepts of interest.1. Can model compound experiments provide sufficient information to under-stand the hydrocarbon reaction mechanism over the MoS2 catalyst?(a) What model compounds suitably represent heavy oil residua feedstocks?(b) Are these model compounds simple enough to provide the analyticalresolution for mechanistic studies?2. Does the reaction environment (condition response time and stability, reactormaterial, mixing regime, etc.) affect catalyst performance?(a) What effect does each factor have and to what degree?(b) Can such effects be overcome through creative reactor engineering?3. What are the hydroconversion reaction mechanisms associated with freshand deactivated MoS2 catalysts?3(a) Do the hydroconversion mechanisms, proposed and opposed in the lit-erature, accurately represent the reaction as determined through modelcompound studies?(b) How does the process of deactivation affect the catalytic mechanism?(c) Can the mechanism of catalyst deactivation be determined and preven-tion or regeneration methodologies proposed?Scope of InvestigationThe primary aims of this study were to:• Develop a model compound testing methodology to accurately and effi-ciently test unsupported catalyst performance.• Determine the mechanism whereby MoS2 affects a catalytic hydroconver-sion reaction under appropriate conditions.• Compare the mechanism of active and deactivated MoS2 catalysts and henceelucidate the mechanism of deactivation.• Propose feasible preventative or regenerative techniques.4Chapter 2Literature ReviewTo understand the advancements in knowledge which this investigation fulfills, itis necessary to examine the technology and knowledge currently available in theassociated fields, the research surrounding such technologies and those aspectswhich are perceived to require further study. To this end, the current “state of theart” is presented as per published literature.2.1 Current Hydroconversion TechnologyThe oil processing industry has an inevitably finite feedstock. This limitation hasmanifested in a decline in the quality of the crude oils being processed. Highervalue feedstocks such as lighter “sweet” crude oils are steadily being replaced byheavier oils [3–7] or even alternative types of feeds altogether (such as bitumenderived from oilsands) [5]. Lighter feedstocks are desirable for they contain higherconcentrations of species in the gasoline and distillate fuels range and reducedcontaminant levels, factors which make them easier to process and generally resultin higher yields of more valuable products and reduced waste. Heavier feedstocks,characterised by higher average boiling points, are often referred to as “sour” asthey contain high concentrations of S, N and O together with metals such as V andNi [3]. Not only does the processing of these heavier feeds usually result in loweryields of the desired lighter products, but the higher concentrations of the variouscontaminants makes this processing more difficult and hence more expensive.5The use of these less-than-ideal feedstocks have their preliminary influences inthe atmospheric and vacuum distillation systems, usually the first processing stepsapplied to a feedstock during oil refining for, with their higher average boilingpoints, such feedstocks result in far higher yields of distillation bottoms or residueas shown in Table 2.1. Not only are these residua more difficult to process thanlighter streams due to their inherent physical properties (such as their vastly higherdensities [presented in Table 2.1 as ◦API, an inverse of density] and their tendencyto precipitate solids for instance) but the contaminants are often associated with theheavier species in these bottoms streams and are hence concentrated in the residueoil (as may be seen from the sulphur and metals contents in Table 2.1). It is thusessential from both an economic perspective, due to the higher proportion of theproduct such residua represent when processing heavier feedstocks, and from anenvironmental perspective, due to the increased amounts of contaminant materialspresent both in these streams and in the overall process, that technologies for theefficient refining of such residua be developed and implemented. In the context ofthis study, the focus is not on the processing of a heavy oil feedstock but ratherthe troublesome vacuum residue resulting from its distillation or from the use ofbituminous oilsands feedstocks.Table 2.1: Comparison of physical properties of various crude oils andresidua.Property Oils ResiduaLight crude Cold Lake Athabasca Cold Lake AthabascaAPI gravity (◦API) 38 10 9 2.1 2.1Viscosity (m2/s at 40◦C) 5 5000 7000 - -Sulphur (wt%) 0.5 4.4 4.9 6.15 6.18Metals1 (ppm) 22 220 280 470 490Vacuum residue2 (vol%) 38 10 9 100 3 100 3Adapted from Gray [3]. 1 - Parts per million by mass. 2 - Liquid volume percent boilingabove 525◦C. 3 - By definition, these residua are vacuum residue.2.2 Residue Processing TechnologiesSimply put, it is desired to convert such vacuum residue or bitumen feeds into morevaluable lighter products whilst simultaneously removing the contaminant species[3, 7], a procedure complicated by the physical properties of the feed and the ten-6dency of some of their constituent compounds (namely asphaltenes) to precipitateas difficult to handle sediment or solid deposits [3] (referred to as “coke”).The properties that make these streams difficult to process in terms of theirphysical properties arise from elevated carbon:hydrogen (C:H) ratios which inturn are due to the high concentrations of various aromatic and other unsaturatedspecies. Two primary routes exist for the conversion of such feeds, both serving toreduce the C:H ratio, hence resulting in a decline in the viscosity, boiling point andsolid formation tendencies of the feed. These routes involve either reducing theamount of carbon or increasing the hydrogen, termed “carbon rejection” and “hy-droconversion” respectively [3, 4, 8, 9]. Both processes are carried out at elevatedtemperatures which results in the thermal (radical) cracking of larger hydrocarbonmolecules in the heavy feed, further reducing the viscosity and boiling points, butpotentially forming hard carbonaceous deposits (as part of the coke) due to radicalcondensation reactions [2, 9–16].2.2.1 Carbon RejectionThe carbon rejection or coking process, operated at elevated temperature and pres-sure (between approximately 450 to 565◦C and 1 to 20 bar, for processes such asvisbreaking or fluid coking [8]) relies solely on thermally initiated radical reactionsto both crack larger, higher boiling molecules into lighter species and to condensecarbon-rich radical fragments into coke [3]. The removal of carbon as coke resultsin an overall reduction in the C:H ratio for the liquid species remaining, manifest-ing as a decline in the viscosity and average boiling point temperature [17, 18].Unfortunately, the radical fragments which condense to form coke are oftenasphaltene molecules [2, 14, 19, 20]. These asphaltenes, defined as those speciesbeing insoluble in n-pentane or n-heptane [21], are known to contain disproportion-ately large amounts of the heteroatomic and metallic contaminant species [19–22],which are thus concentrated in the resulting coke. Whilst this is beneficial in thatthese species are being removed from the liquid product, it does however meanthat the low value coke by-product, which may present up to 20 wt% of the finalproduct [17], is heavily contaminated and thus represents a significant environmen-tal hazard [23–25], with disposal costs thus associated with it. A greater financial7impact comes from the loss of valuable liquid products to such solid species, withcoke formation thus representing a significant negative economic impact on theprocess. This mechanism of contaminant removal is also not the most efficient asit does not directly remove unwanted species. For a contaminant atom (such asa metal, sulphur or nitrogen) to be removed, it must be contained within a radicalfragment which must then participate in a condensation reaction. Only this way aresuch contaminants captured in the solid coke which may be removed. Furthermore,as there is little control as to which of the fragments react (limited to temperatureand residence time control), over-cracking to form lower value gaseous productsis often a problem in the carbon rejection process [3, 8]. Despite these drawbacks,the low cost and simple operation of such processes means that they are profitableand hence common in commercial oil processing operations, manifesting as thevisbreaking, delayed-coking, fluid-coking and flexi-coking systems [4, 17].2.2.2 HydroconversionHydroconversion operating conditions vary greatly, with temperatures ranging from370 to 450◦C and pressures from 7 to 27 bar [8, 17, 26], depending on the reactortype (fixed bed, fluidised bed or slurry-phase), catalyst type and feed. This pro-cess is often conducted in the presence of either a supported metal catalyst, such asNiMo/Al2O3, or an unsupported metal catalyst, Fe or Mo for instance. The activemetal phase is the metal sulphide, with the catalyst either being introduced as sucha sulphide, or as a metal which is rapidly sulphided in-situ.Similarly to the carbon rejection process, cracking within a hydroconversionreactor occurs by radical reactions initiated by the elevated temperatures, with cokebeing formed by condensation reactions between radicals. The purpose of the cat-alyst in this system is to stabilise, “cap” or “quench” excess radicals, reducing bothcondensation reactions and over-cracking, resulting in lower yields of coke andgases. This is widely held to occur with the catalyst in some manner “activating”hydrogen dissolved in the residue oil to form free hydrogen radicals which thenstabilise hydrocarbon radicals [3, 8, 9], inhibiting continued reaction and addinghydrogen to the molecules (resulting in an overall decrease in the C:H ratio).At present there are three main methodologies for the implementation of hy-8droconversion: fixed bed reactors, fluidised (or ebullated) bed reactors and slurry-phase reactors. For the processing of contaminated heavy feedstocks, fixed bedsystems, despite being easy to implement and operate, are generally a poor choiceas rapid coking and metals deposition at the entry to the catalyst bed results in asharp increase in the pressure drop across the reactor, a decline in performance andfrequent process shutdowns [4, 8, 18]. This may be avoided through the use ofguard beds or guard reactors or by occasionally agitating the catalyst bed in somemanner to break up deposits and agglomerates [4]. Furthermore, it is necessaryto shut the reactor down in order to change the catalyst, although having multiplereactors running in parallel makes the impact of this action on overall plant opera-tion less significant [4, 8]. Despite these factors, most commercial hydroconversionsystems operate with fixed bed reactors, processing lighter, less contaminated feedsat milder conditions [18].Fluidised bed reactors are a significant improvement on fixed bed systemsspecifically in terms of being able to remove and replace a portion of the catalystcharge as it becomes deactivated without interrupting the operation of the unit [4].Such reactors, for instance the LC-Fining [3] and other fluidised-catalytic-cracking(or FCC) [4] reactors, are also able to handle far heavier and more contaminatedfeeds than fixed bed units as solids formed during the reaction, or even entrained inthe feed itself, do not deposit on an immobile bed, increasing pressure drop and in-hibiting performance as they would in a fixed-bed system [4]. Smaller catalyst par-ticles may also be used, with the reduction in diffusion length increasing the overallreaction rate without adversely affecting the pressure drop [4]. Unfortunately, suchcatalysts must also be tailored to operation in a mechanically demanding environ-ment, increasing their expense which, together with the additional complexity ofoperating a fluidised bed system, has limited the commercial applications of thesereactors in residue hydroconversion processes [8, 18], with only LC-Fining andH-Oil being operated commercially [5].Originally developed in Germany in the 1920’s, slurry-phase reactors have re-cently seen increased application for the hydroconversion of heavy feedstocks [8].Combining the advantages of the fluidised bed systems (reduced pressure drop dueto solid deposits and catalyst replacement without interruption) but allowing forsmaller catalyst particles to be utilised (millimeter range of fluidised bed supported9catalysts down to nanometer sized unsupported catalyst particles), slurry-phase re-actors are seen to offer high reaction rates and conversion together with reducedcoke yields [26].2.2.3 Slurry-Phase Catalytic HydroconversionNumerous reviews have been published detailing the catalytic and operational fac-tors of slurry-phase hydroconversion systems ([4, 5, 8, 26] for instance), and onlya brief overview of those processes of commercial interest are provided here. Theoriginal implementation of slurry-phase hydroconversion technology makes use oflow cost, single-use unsupported metal catalysts such as Fe, which may be intro-duced as iron sulphate (as per the CANMET process) [4, 8] or as an unrefined FeSx-containing mineral, usually pyrrhotite [5, 27], (as per the VEBA Combi Cracking(VCC) and HDH processes) [4, 5, 26]. These systems are operated at slightly moresevere conditions than the conventional fixed and fluidised bed units, allowing forhigher levels of conversion and contaminant removal to be achieved at the expenseof elevated coke generation and increased catalyst deactivation [4, 8] with typicaloperating conditions and product specifications presented in Table 2.2. These neg-ative effects are, however, not an issue as the unsupported Fe catalysts are disposedof after deactivation unlike the more expensive supported metal catalysts used infixed and fluidised bed systems [5]. Of the many slurry-phase processes developedsince the 1980’s, with the Micrometallic-coke (M-coke), ENI Slurry Technology(EST), Super Oil Cracking (SOC), Intevep HDH and (HC)3 being added to thosepresented above [4, 8, 17, 18, 26, 28, 29], none have seen implementation beyondpilot plant scale [4, 17] due to lower profitability as compared to thermal pro-cesses. VCC technology saw pilot scale operation of 4000 bbl/d, CANMET wasrun at some 5000 bbl/d and SOC was reported at 3500 bbl/d [4, 30]. Unfortu-nately, development of these technologies slowed by around 2000 with only ENItechnology still being operated on pilot scale [30]. There are several reasons forthis [17, 18, 31]: severe conditions, high hydrogen costs (due to hydrogen con-sumption and high operating pressures), longer residence times and, in particular,high catalyst costs due to high initial expense and rapid catalyst deactivation whichusually necessitates disposal and hence monetary losses.10To improve profitability and reduce waste and negative environmental impact,unsupported metal catalysts which can be used in such a manner as to achieve thesame improved performance as the once-through disposable versions, but with-out having to dispose of the deactivated catalyst (contaminated with coke and itsassociated heteroatomic and metallic species), have been developed [30]. This im-provement is only possible by recycling and reusing the catalyst multiple times,often requiring regeneration to regain lost activity and/or selectivity. One such cat-alyst, which has received a large amount of attention recently is Mo [1, 5, 18, 32].The MoS2 active phase has been shown to be more active overall (for example incoke inhibition and hydrodesulphurisation, HDS) [33], more selective toward thedesired middle distillate fuels range and produce less coke than the conventionalFe-based alternatives as shown in Table 2.2. MoS2 may be introduced as dispersedparticles produced ex-situ (in the form of micelles for instance) [26, 32] or formedin-situ by the addition of an oil- or water-soluble Mo compound (such as Monaphthenate or ammonium heptamolybdate respectively) to the reaction medium[26, 32].Table 2.2: Comparison of fluidised bed and slurry-phase hydroconversion re-actor performance.Property ProcessLC-Fining 1,2 CANMET 2 M-coke 3Reactor type Fluidised bed Slurry-phase Slurry-phaseFeed type Athabasca bitumen CLVR 4 Vacuum residueOperating temperature (◦C) 425 - 450 440 - 460 5 400 - 454Operating pressure (bar) 100 - 150 100 - 150 5 69 - 172Catalyst type NiMo/Al2O3 Fe sulphate Mo naphthanateCatalyst loading (wt%) 2 - 13 6 1 - 2 0.01 - 0.02 7Conversion (wt% 8) 66 86 > 95Demetallation (%) 60 - 70 88 9 > 90Desulphurisation (%) 65 80 9 Not reportedLiquid product 10 (vol%) 65.2 88.7 99.5Residue and solid product (vol%) 34.8 11.3 0.51- Gray [3]. 2 - Nalithem et al. [34]. 3 - Speight [9], Bearden and Aldridge [33]. 4 - ColdLake vacuum residue. 5 - Rana et al. [8]. 6 - Baussell et al. [35]. Loadings for variousresidue feeds, converted from 0.05 - 0.31 lb/bbl of feed (feed density of 1019 kg/m3). 7 -Converted from 100 - 200 ppm by mass. 8 - Material boiling above 524◦C. 9 - Furimsky[4]. 10 - Combined light and heavy naphtha and light and heavy gas oil.11Mo-based catalysts, prepared from refined molybdenum metal or molybde-num oxide for water-soluble salts or oil-soluble complexes or mechanically groundMoS2 (perhaps molybdenite ore), are, however, significantly more expensive thaniron-based alternatives. With the price of molybdenum oxide (roasted molybdeniteconcentrate) at the time of writing being around USD 21,000 per tonne [36, 37] andiron ore (fines taken as a slight overestimate of the price of red clay or similar op-tions) being only USD 130 per tonne [36]. Thus, despite improving performanceand reducing waste, it is vital that Mo-based catalysts be efficiently recovered,regenerated and recycled if the processes in which they are utilised are to be eco-nomically viable and competitive [4, 5, 8, 25, 26, 38].Recent research into the use of Mo-based catalysts has indicated that, depend-ing on the precursor used, the active MoS2 phase may be recovered from a batchreactor system, together with the solid coke formed during the reaction, and thiscoke-catalyst agglomerate reused [1, 32, 38]. The performance observed from suchrecycled catalyst is, however, seen to decline after several recycles.This decline in observed activity is due to the deactivation of the MoS2 in thereactor. Whilst a significant amount of research has been conducted to study themechanisms of deactivation in supported metal catalysts (such as those utilised infixed and fluidised hydroconversion reactors) [39–41], the deactivation of unsup-ported catalysts, particularly those utilised in hydroconversion systems, is not wellunderstood [1]. Fundamental to the regeneration and recycling of Mo-based un-supported metal catalysts is an understanding of such deactivation mechanisms,which itself depends on an understanding of the mechanism of the active catalyst.2.3 Catalyst TestingThe development of an understanding of the catalytic mechanism for active hy-droprocessing catalysts, and for that matter, the deactivation mechanism, is severelyhindered by the complexity of the feedstock in which these catalysts operate [28].2.3.1 Heavy Oil and Residue Oil StudiesCrude oil is comprised of many thousands of different organic, inorganic, organo-metallic, aqueous and solid species making identification and quantification of in-12dividual species in the feed or product virtually impossible [10, 28]. This inabil-ity to determine the exact chemistry of the reaction has led to various techniqueswhereby the reactions (and the associated understanding and modeling activities)are described in terms of observed parameters (such as hydrogen consumption dur-ing reaction) or grouped measurements such as simulated distillation (SIMDIS)for boiling point distribution or elemental analysis of a recovered fraction, suchas CHNS (carbon, hydrogen, nitrogen and sulphur) analysis. While useful for de-termining the effectiveness of a given catalyst in a reaction (in terms of coke sup-pression for instance), these grouped or lumped measurements do not provide thechemical species information required to reliably determine reaction mechanismaffected by the catalyst [6, 42, 43].To obtain the information required to determine how and why different cata-lysts work, it is necessary to greatly simplify the system through the use of modelcompounds [6, 42].2.3.2 Model Compound StudiesStudying catalyst activity in the presence of residue oil has both benefits and draw-backs. It is the actual feed in which the catalyst is designed to operate and, as such,studies using residue oil are extremely useful for tailoring catalysts, optimisingprocess conditions and developing reactor systems. As indicated in Section 2.3.1,however, the complexity of such a feedstock makes it almost impossible to identifyor quantify the myriad of species present. Furthermore, the physical and chemicalproperties of such heavy feeds (presented for various oils and residua in Table 2.1),and how they change during the reaction, can complicate studies in which they areused. One example of such a complication is that the metal species present in manyresidue oil feeds may themselves be catalytically active in the hydroconversion re-action [28].An alternative to using residue oil is to substitute it with a suitable model com-pound. The use of model compounds (which may be introduced as a single speciesor as a mixtures of several chemicals) greatly simplifies the reaction system as allspecies of which the simulated feedstock is comprised are known and quantified.Unfortunately these model compounds may not actually be present in the original13feed and yet they are expected to provide accurate and applicable information re-garding the reaction mechanism. Furthermore, whilst a model compound is idealfor studying the activity of an active catalyst (given that the clean feedstock is un-likely to cause the same degree of deactivation as residue oil), deactivation studiesbecome more complicated. For such an experiment a catalyst must be deactivatedin residue oil then its activity tested in the model compound. This means thatthe deactivated catalyst is being evaluated in an environment different from that inwhich it originally lost its activity, an environment which may affect how it func-tions (deactivation caused by coke deposition may be reversed, for instance, if thecoke dissolves in the model compound). The selection of the model compoundsand the reactors in which they are tested is thus not a trivial endeavor [42, 44].SelectionModel compound selection is often an optimization problem whereby the complex-ity of the feed is juxtaposed with its analytical simplicity. A more complex feed(not necessarily a mixture of more species but perhaps simply a larger molecule)may offer better representation of the feedstock than does a simpler molecule, butas the complexity of the feed, and hence its similarity to the residue oil, increases,so does the difficulty and complexity of chemical analyses and hence mechanisticevaluations. A large, polynuclear aromatic molecule with multiple alkyl branchesand heteroatomic constituents may be an accurate representation of the asphaltenicfraction of residue oil [15, 45], but the range of products collected from the reactionof such a molecule (gases, liquids and even solids from condensation and precip-itation reactions) would require the use of multiple analytical procedures for fullidentification and quantification. A smaller model compound may make analysisand interpretation easier, but such a species may not offer an accurate representa-tion of the residue oil feedstock.The use of model compounds to simplify reactions to the point of mechanis-tic understanding is by no means a novel concept and has been conducted in theoil processing and coal liquefaction fields for many years and continues to see agreat deal of active research [15, 27, 42, 44–56]. Some studies focused on specificreactions, such as the hydrodesulphurisation (HDS) of benzothiophene [50] or the14hydrodeoxygenation (HDO) of benzyl phenyl ether and bibenzyl ether [53, 56].Some aimed to determine the mechanisms associated with specific structural com-ponents such as the aromatic rings of naphthalene, fluorene and pyrene [42]. Oth-ers sought to develop novel model compounds which accurately represent specificspecies or groups of species in a feed, such as 1-dodecylpyrene [44], cholestane-benzoquinoline [2], 1,3,6,8-tetrahexylpyrene [15] or larger “archipelago” com-pounds [14] for representing the asphaltenic fraction of residue oil. Yet more stud-ied the interactions between different fractions, with both represented by modelcompounds, such as the hydroconversion of aromatic species in the presence of ahydrogen donor solvent (diphenylpropane in tetralin for instance) [11].Of the many species proposed and tested over the years, diphenylmethane (orDPM, shown in Figure 2.1) has been widely used and accepted [27, 45, 46, 52–55] as a model compound in both residue hydroconversion and coal liquefactionstudies. There are two main reasons for this: thermal stability and molecular andreaction simplicity.Diphenylmethane may be viewed, in a simplified manner, as having only twotypes of functional groups: an aromatic ring and a proximal alkyl-aryl C-C bondwhere the methyl bridge connects to the ring. This means that a thermally initiatedreaction would have to sever the proximal Calkyl-Caryl bond to form a benzyl and aphenyl radical (as illustrated in Figure 2.1), both unstable, high energy species [46].This Calkyl-Caryl bond is very difficult to break and the resultant thermal reactivityof DPM is low [45, 46, 48, 49]. Appreciably rapid thermal decomposition of DPMwould thus require initiation by hydrogenation of or, more commonly believed,radical addition to one of the stable phenyl rings, destabilising the molecule andallowing the thermal reactions to proceed [15, 21, 42, 44, 46, 49]. The radicalattack is thought to occur at the ipso position of the ring by either radical hydrogenaddition or hydrocarbon radical addition. Due to this stability, DPM is consideredto be a good kinetic representative of the much larger, refractory aromatic species(such as the asphaltenes) present in residue oil.15+ThermalDiphenylmethane BenzylradicalPhenylradical∆Gr718K = 218 kJ/molFigure 2.1: Thermal cracking of diphenylmethaneReactors and ConditionsWith the extent of literature published, one can find examples of virtually any reac-tor system being successfully used for heavy oil, residue oil and model compoundstudies. These range from the magnetically stirred 230 mL stainless steel systemby Matsumura et al. [55] which was operated in batch and semi-batch (for hy-drogen gas) mode to the 150 mL stainless steel system of Liu et al. [48] whichwas operated in fully continuous mode. Far smaller systems, often termed “micro-reactors”, are also quite popular, usually being operated in batch mode. Examplesof such units include the 50 mL reactor by Sato et al. [54] which was rocked to mixand the unagitated system developed by Savage et al. [44] which required only 45mg of reactant. Both of these reactors were constructed of stainless steel.2.4 Micro-Reactors for Catalyst TestingThe reactor systems indicated in Section 2.3.2 are not, however, all the same.“Micro-reactor” is used in the literature to refer to a variety of different reactorswith varying designs and operating conditions. These are generally divided intotwo categories: flow and batch, each having distinct advantages and disadvantages.This study is concerned solely with batch micro-reactors.2.4.1 Advantages and Disadvantages of Micro-ReactorsMicro-reactors benefit from requiring less reagent and catalyst charge and produc-ing less waste. This makes them ideal for rapid, cost-effective and environmentallyconscious screening of catalysts and reaction conditions [57–60]. Reduced dimen-sions improve heat and mass transfer and allow for more accurate control of pro-cess conditions due to faster responses [57–61]. This improved condition controloften results in improved conversion and selectivity [57, 62]. As the dimensionsof the reaction chamber decrease, so the wall surface area to volume ratio (A:V)16increases, affording the opportunity for solid catalysts to be securely adhered tothis surface instead of being separately supported on particles [60, 61, 63].There are, unfortunately, also numerous disadvantages associated with scalingdown a reaction system. The lower product volumes necessitate care in recovery,work-up and analysis and may make the recovery of certain materials (such as spentcatalyst) difficult or impossible [60–62]. Control of the process conditions can bequite sensitive and even small fluctuations can propagate rapidly through a micro-reactor system [57–60]. The increased wall area to volume ratio may negativelyinfluence the system by inhibiting mixing through frictional effects [60] and exert-ing a noticeable, and often unwanted, catalytic influence [59]. One of the biggestchallenges associated with such systems in solid-catalysed reactions, however, isthat most implementations can not handle either solid particles or the precipitationof material. Due to the small dimensions, solids rapidly foul the reaction systemand necessitate unacceptably frequent shut down and cleaning [59, 61, 64].Mixing in micro-reactors is another complicated matter [60] and whilst manystudies have been conducted in the field, these have focused almost exclusively onmicro-flow units operating gas-gas, gas-liquid or liquid-liquid phase systems.2.4.2 Micro-Reactors in Hydroconversion StudiesDespite the complications associated with mixing and solid material (introducedsolid catalyst particles or precipitating products), micro-reactors see extensive usein applications where these factors may be considered to be quite extreme, specif-ically in the slurry-phase hydroconversion of vacuum residue. In such an applica-tion, for instance in catalyst screening or model compound testing [14], the reactionmixture is a gas-liquid-solid system with micrometer-scale unsupported catalystparticles suspended in the reaction liquid, as discussed in Section 2.2.3. As such,not only is thorough mixing necessary to ensure adequate contact between the gasand liquid-solid slurry, but also to ensure that the solid particles remain suspendedin the liquid and that the resulting slurry is well-mixed. Many of these systemsexperience coking, and hence the formation of additional solid material, during re-action which may precipitate and remain suspended or deposit on the reactor walls[14, 15]. The high temperature, S-rich environment also results in the sulphidation17of the metal reactor itself, potentially turning the surface of exposed walls into cat-alytically active centers [21, 59, 60], obfuscating the results of thermal experimentsor the effect of the catalyst added for study.Numerous researchers have developed ways to overcome these difficulties.Foremost is the almost exclusive use of batch [2, 14, 15, 44, 46, 54, 56, 65], ratherthan flow [53], systems. In this manner the fouling and plugging of flow paths bycatalyst or coke is avoided. Mixing is somewhat more complicated by the reducedsize of these units. Very few examples exist of internal agitation [63] (due partlyto fabrication and operational complexities of size and partly to the increased wallfriction effects indicated in Section 2.4.1) with most relying on an externally ap-plied mixing regime. Examples of such engineering include spinning [65], rocking[54] or vertical shaking [2, 15] of the reactor, whilst some researchers are contentwith no mixing [44], a potentially feasible approach in very small systems due toreduced diffusion distances [59] and the slow settling of ultra-fine particles. Unlikethe mixing studies conducted for micro-flow reactors, no literature could be foundexamining the effectiveness of various mixing techniques in micro-batch systems.The phenomenon of wall activation is, for the most part, overlooked in theslurry hydroconversion field of research. This factor, greatly exacerbated as thewall area:reaction volume (A:V) ratio increases with reduced reactor size, is onlybriefly indicated by a few researchers [21, 59, 66] and little is done to prevent ormitigate these effects. Some reactors are designed to include inert layers (such asthe inner glass tube in the flow systems by Matsuhashi et al. [53] or Khorasheh andGray [67], or the batch system by Alshareef et al. [2]) to separate the reaction liquidfrom the metal walls whilst others have resorted to extensive cleaning betweenreactions (such as the multi-stage acid etching by Savage et al. [44]) to help reduceinter-reaction effects or build-up. Despite knowledge of these wall effects, manyresearchers do not mention them and no studies are known which indicate the rateat which this catalytic influence develops or its extent.2.5 Catalyst Activity and DeactivationA discussion of catalyst activity and deactivation requires a knowledge of the activephase and how it is presented in the reaction mixture. Below is presented such a18discussion, with a particular focus on the MoS2 of interest in this research, togetherwith an overview of catalyst deactivation and regeneration processes.2.5.1 Catalyst SelectionCatalyst selection for a given hydroconversion system is a complex endeavourwhich examines all aspects of the system such as the properties of the feedstock,the reactor type and the operating conditions. A significant amount of literaturehas been published on this topic, with the work below being but a brief overview.Supported and Unsupported CatalystsSupported catalysts are common in oil processing. Such catalysts usually take theform of metal sulphide (Mo or W promoted by Ni or Co) crystallites on a porous γ-alumina, zeolite, silica, silica-alumina or carbon support material [4, 28, 41]. Suchsupported metal catalysts ease handling (due to larger particle sizes) whilst main-taining a high level of metal dispersion (as small crystallites over the large surfacearea of the porous support material) [4, 28]. The support itself may also play arole in the chemistry of the reaction (acid catalysed reactions by zeolite supportsfor example). Such supported catalysts are, however, prone to rapid deactivationby both physical blockage of the pores by coke and contaminant metal crystallitesand direct poisoning of the active sites themselves [4, 8, 28, 41].Unsupported catalysts, formed in-situ by the decomposition of oil- or water-soluble precursors added to the feed or introduced as finely divided solid parti-cles [5, 17, 32, 68, 69], overcome some of these difficulties but pose their ownchallenges. The micron- to nanometer-sized particles from the decomposition ofoil-soluble precursors [31], shown to be more active than water-soluble precursorsor mechanically ground particles [5, 18, 31, 68], offer extremely high dispersion[5, 31] but are also extremely difficult to separate from the solid coke precipitateswith which they are recovered from hydroconversion reactions. Recent studieshave thus focused on recycling the coke-catalyst agglomerates formed during sucha process in their entirety [1, 18, 32]. Initially, the lack of a porous support inhibitsthe deactivation of these catalysts as there are no inactive surfaces upon which cokecan amass, and potentially occlude the active metal crystallites, and there are no19pores susceptible to blockage [4, 8, 28, 39, 70]. Once sufficient coke has formedand agglomerates with the catalyst particles, however, this coke-catalyst agglom-erate is, for all intents and purposes, a supported catalyst. Whether agglomeratedor not, dispersed catalysts do still undergo poisoning and/or fouling due to the for-mation of contaminant metal crystallites on the active phase Bartholomew [71].Catalyst Active PhaseIn the slurry-phase systems, unsupported metal sulphides are the most commonlyused catalysts [5, 17]. Of these, MoS2 is the most effective, being introduced to thesystem as an oil-soluble precursor (such as Mo naphthanate or octoate) [5, 31]. Anoil-soluble precursor, dissolved in the feedstock, decomposes at elevated tempera-tures, allowing the metallic species to react with H2S or other sulphur-containingspecies in the oil, to form active metal sulphide particles [5, 33, 72]. Metallicsalts introduced to the system (ammonium heptamolybdate for instance) also de-compose to an active metal sulphide, with these reactions being shown to proceedthrough various oxy-sulphide salts [5]. As such, given the severe conditions andhigh sulphur concentrations, metal sulphides are the obvious choice for a stableactive phase in such reactions.2.5.2 Molybdenum DisulphideThis work focused on the use of MoS2 as an unsupported hydroconversion catalyst.With the proven effectiveness of this catalyst, numerous works have been publishedreporting the properties, structure, active sites and theorised catalytic mechanismsof this material [5, 21, 29, 31, 38, 68, 69, 73–78]. Below is presented a briefoverview of this literature.Structure and Active SitesThe structure of MoS2 is similar to that of other transition metal sulphides, par-ticularly WS2, and presents as a layered S-Mo-S crystal as depicted in Figure 2.2[28, 29, 69, 74, 75]. It may be seen that the MoS2 consists of trigonal prismsof S coordinated to Mo, forming large sheet-like structures. These sheets asso-ciate with one another by weak van der Waal’s forces to form stacked crystallites20[28, 29, 69, 74, 75]. The size of the sheets and the height of the stacks is influencedby the conditions under which the MoS2 is synthesized [28]. Occurring naturallyin large, ordered sheets as the mineral molybdenite [29, 31, 79], synthesized MoS2often presents as highly bent, disordered sheets termed a “rag” structure [29].Figure 2.2: Rendering of arbitrary 5-layer stack of MoS2 (created in AccelrysMaterials Studio 4.4 with published unit cell data [80])With the homogeneous sulphur basal planes of the MoS2 sheets consideredchemically inert, it is the rim and edge atoms where coordinatively unsaturatedsites and/or S anion vacancies exist which are considered to be the catalyticallyactive sites [5, 28, 38, 75–77, 81] (although the role played by the former is apoint of debate [29]). Other defects, such as distortions or inhomogeneities in thesulphur basal plane, are also active [29, 75, 76]. The sulphur anion vacancies atsuch active sites afford them Lewis acid character, allowing for the adsorption ofmolecules with unpaired electrons [40], and given the high density of such sites atthe rim/edge or along basal plane features such as folds, double or higher vacancypoints may occur [40]. It has also been proposed that -SH groups on Mo catalystsexhibit Brønsted acid characteristics and associated cracking activity [82]. The ac-tivity of an MoS2 catalyst may thus be improved by increasing the proportion ofrim/edge sites and basal plane defects. This is achieved by reducing the both size21of the sheets and the stack height or by intentionally distorting the MoS2 sheets,for instance by controlling the synthesis conditions or through chemical exfolia-tion [29, 38]. The promotional effects of defects and distortions may explain whymolybdenite, even when milled to the same particle size, exhibits reduced hydro-conversion activity as compared to synthetic MoS2 [31, 68].Catalytic MechanismBefore determining the influence which a catalyst may have in a hydroconversionreaction, it is necessary to understand what thermal processes (both radical chainreactions and thermal hydrogenolysis) may occur [83]. Depicted in Figure 2.3a, thethermolysis of DPM initiates with the homolytic cleavage of the Calkyl-Caryl bondto form benzyl and phenyl radicals. These radicals may propagate the reactionby either radical addition to or H* abstraction from other DPM molecules. Asan example of this propagation, a benzyl radical may abstract hydrogen from thealkyl carbon of DPM (shown in Figure 2.3b) to form benzene. The DPM radicalthus formed may continue to crack or the reaction may terminate, for instanceby radical addition with a benzyl radical to form 1,1,1-(1-Ethanyl-2-ylidene)tris-benzene (ETB) as shown in Figure 2.3c. With excess hydrogen, the reaction mayproceed via abstraction of H* from dissolved H2 as shown in Figure 2.3d for aphenyl radical. The resulting H* radical may then propagate the reaction by eitherabstracting H* from other species, similarly to phenyl in Figure 2.3b, to form H2,or through radical addition and subsequent C-C homolysis as shown in Figure 2.3e[83]. Thermal hydrogenolysis [83] occurs by direct C-C cleavage and hydrogeninsertion as shown in Figure 2.3f.Despite the extensive research published regarding the use of MoS2 as a hydro-conversion catalyst, the mechanism is still relatively poorly understood, with manyof those proposed, and indeed widely held and perpetuated, being considered atbest conceptual understandings [29, 75] or openly challenged as partially or totallyincorrect [21]. The role of the catalyst is generally presented as performing theheterolytic (to form Mo-H and S-H moieties) or homolytic (to form two S-H moi-eties) dissociation of hydrogen. These species either remain on the catalyst surfaceor desorb into the liquid phase [5, 21, 28, 29, 46, 75–77].22+Thermal(a)+H+(b)+(c)+ H H + H(d)+ H +(e)+H H+(f)Figure 2.3: Mechanisms for the thermal decomposition of diphenylmethane(DPM). (a) Initiation by Calkyl-Caryl thermolysis of DPM. (b) Propaga-tion by hydrogen abstraction from DPM by a phenyl radical. (c) Termi-nation by radical addition between a DPM radical and a benzyl radical.(d) Propagation by hydrogen abstraction from dissolved H2 by a phenylradical. (e) Propagation by radical addition of H* to DPM. (f) Thermalhydrogenolysis to form benzene and toluene by H2 insertion to DPM.23Since the 1960’s, a mechanism proposed by Curran et al. [78] that this pro-cess forms a reactive hydrogen species, usually referred to as “H*” as shown inFigure 2.4a, has been widely held and supported [5, 29, 31, 46, 68, 75–77]. Theseactivated hydrogen species are theorised to spill-over across the surface of the cata-lyst [5], and its support where applicable, and even back into the liquid phase [29],“capping” or “quenching” hydrocarbon free radicals formed through the thermaldecomposition of the feed (for instance by radical addition as shown for a phenylradical in Figure 2.4b). The overall mechanism, in the context of DPM hydrocon-version, is illustrated in Figure 2.6a. This theory has proved popular as it explainsmany of the trends observed in residue hydroconversion reactions upon the addi-tion of a catalyst: increased hydrogen consumption, reduced overcracking to gasand reduced condensation to solid coke. Additionally, such H* species are thoughtto promote conversion of hydrocarbon feedstocks. This is theorised to occur bytwo mechanisms. Firstly, the rapid stabilisation of thermolysis radicals inhibitscondensation reactions, promoting the decomposition of larger hydrocarbons. Sec-ondly, much like hydrocarbon free radicals through the process of radical hydrogentransfer (RHT) [15, 21, 44, 49, 84], activated hydrogen may serve to promote thedecomposition of otherwise thermally stable aromatic hydrocarbons through theiraddition to an aromatic ring and subsequent cracking [44, 49, 84–87], as shown forDPM in Figure 2.6b.Recent research [21, 29] has raised doubts as to the validity of the theory pro-posed by [78]. These studies, based mostly around the work of LaMarca et al. [88],propose that following an initial high activation energy thermal cracking step, de-composition of the feed occurs predominantly by radical chain reactions through aseries of radical hydrogen transfer (addition or abstraction) and scission steps be-fore terminating, by radical recombination or radical-to-olefin addition, as stablehydrocarbon products [21, 88]. In the context of DPM hydroconversion, initiationwould occur as per Figure 2.3a. Hydrogenation by the catalyst and continued ther-molysis breaks these species into shorter hydrocarbon radicals, for instance phenylradicals into C3 radicals per Figure 2.4c. Radical addition of these short chain rad-icals to the DPM followed by β -scission, per Figure 2.4d, decomposes the DPMfeed into benzyl radicals and alkyl-benzene species. These benzyl radicals maycontinue to react or terminate as per one of the mechanism presented above. A24generalised mechanism for this reaction is presented in Figure 2.6c. The radicalchain reaction theory of LaMarca et al. [88] suggests that the capping or quench-ing of radicals would be detrimental to the overall performance of the system asit would inhibit radical interaction with the feed and subsequent cracking. Unfor-tunately, many of the studies aimed at elucidating the mechanism of liquid-phasehydroconversion (often in the presence of a hydrogen donor or shuttle solvent) havemet with mixed results [21], some supporting the mechanisms proposed by Curranet al. [78] and others those of LaMarca et al. [88].2HH H Catalytic(a)+ H(b)+ 4.5 H2 Catalytic2(c)+ +(d)Figure 2.4: Mechanisms for the thermocatalytic decomposition of diphenyl-methane. (a) Initiation by formation of activated hydrogen species oncatalyst surface. (b) Termination by stabilisation of a phenyl radical byhydrogen radical addition. (c) Propagation by thermocatalytic crack-ing of a phenyl radical to propane radicals. (d) Propagation by radicaladdition of a propane radical to DPM followed by β -scission.A final mechanism of interested in such a catalytic system is catalytic hy-drogenolysis [89–91]. By this mechanism, dissolved hydrogen would dissocia-tively adsorb, heterolytically or homolytically, on the surface of the catalyst asdictated by a Langmuir adsorption isotherm [92]. A DPM aromatic ring adsorbedin the vicinity of these hydrogen species is likely to undergo hydrogenation, asshown in Figure 2.5a, to produce saturated products such as the 2-Benzyl-1,3-25cyclohexadiene illustrated, or hydrogenolysis, as shown in Figure 2.5b, to pro-duce a mixture of benzene and toluene. Given the adsorption, reaction and des-orption steps required for these reactions, it is possible that such a system would begoverned by Langmuir-Hinshelwood-Hougen-Watson (LHHW) kinetics [92–94]whereby one step would be rate limiting whilst the others could be considered inquasi-equilibrium.H HCatalyst surface(a)H H+Catalyst surface(b)Figure 2.5: Reactions occuring on the catalyst surface between adsorbed hy-drogen and diphenylmethane. (a) Hydrogenolysis of adsorbed phenylring by dissociatively adsorbed hydrogen to form partially hydro. (b)label 2.With this in mind, the state-of-the-art knowledge regarding the role of the cat-alyst in residue hydroconversion is that, using H2 dissolved in the liquid, the cata-lyst hydrogenates olefins (which are reactive and themselves promote overcrackingand coke formation reactions) to more stable saturates, hydrogenates poly-aromaticspecies to form hydrogen donor or shuttling compounds and exacts catalytic hy-drogenolysis [21].2.5.3 Processes of DeactivationWhilst understanding the mechanism of catalytic activity is important, develop-ment of a meaningful regeneration and recycle regime requires an understandingof how the catalyst deactivates, the chemical and morphological changes which re-26+ +H2 2HThermalCatalyticCapping orquenchingContinued cracking orcondensation reactions(a)H2 2HCatalyticH+ H +Continued cracking orcondensation reactions(b)+ +Continued cracking orcondensation reactionsCxHyCxHyCxHy(c)Figure 2.6: Simplified literature mechanisms for the thermocatalytic decom-position of diphenylmethane. (a) Capping of thermal radicals (adaptedfrom Curran et al. [78]). (b) DPM destabilisation by active hydrogenradicals (adapted from Wei et al. [46]). (c) DPM destabilisation by rad-ical hydrogen transfer (adapted from Gray and McCaffrey [21]).27sult in the loss of catalytic activity [7]. The deactivation of heterogeneous catalystshas been a topic of research for many years and the varied processes accounting forthis phenomenon are, generally, very well understood. The interpretation of theseprocesses to a specific application, however, is more difficult. In few systems isthis more the case than residue hydroconversion wherein the myriad of species inthe feed and the complexity of the reaction networks makes determination of thephysio-chemical deactivation mechanisms extremely complicated [7].The deactivation of catalysts, supported or unsupported, in heavy oil or residuehydroconversion is due to the formation and deposition of coke and metals and theinteraction of organometallic and heteroatomic species with the active sites [1, 4–7, 10, 18, 28, 39–41, 43, 70, 71, 95–104]. The exact nature of the deposits and theinfluence of these deposits and other species on the system are dependent on manyfactors: the composition of the feed, operating conditions, type of catalyst supportand type of active phase to name a few [41].For supported catalysts, it is generally held that initial deactivation occurs byrapid coke deposition on the surface and in the pores of the support. This depo-sition reaches a pseudo-steady-state at which point more gradual metal depositioncontinues to steadily deactivate the catalyst by both plugging pores and physicallyoccluding and/or poisoning active sites [10, 28, 95, 98]. Final and total deactiva-tion occurs rapidly when coke and metal deposits constrict and block the catalystpores, eliminating all contact between the active sites and the reaction mixture[28, 95, 97]. Few theories have been proposed for unsupported catalyst deacti-vation, but the most plausible explanation is that the precipitation of unreactivegraphitic coke envelopes and encapsulates active metal particles, preventing theirparticipation in the reaction [1, 18].Extensive descriptions of the various deactivation mechanisms including foul-ing, poisoning, thermal degradation, solid-state reactions and vapour-phase degra-dation and mechanical degradation are provided in Section A.1.2.5.4 Catalyst Regeneration MethodologiesCatalyst deactivation may be controlled in three ways: prevention, mitigation andregeneration. Prevention often involves a physical change to the reaction system28(such as removal of feed contaminants, the installation of guard beds or changingprocess conditions to avoid deactivation altogether [40, 71, 96]). Mitigating deac-tivation involves temporarily changing operating parameters to overcome the ef-fects of deactivation whilst not interrupting the system (for instance by continuallyadding fresh catalyst to compensate for deactivation [1] or steadily increasing thetemperature of the reaction [40, 71, 103] as is common in residue hydroprocess-ing). Despite these options, one of the most direct and commonly implementedmethodologies for extending catalyst life is the use of some form of regenerationstep prior to reintroduction of a recovered catalyst to the reactor [40, 105]. Suchtechniques, generally incorporating some form of thermal and/or chemical treat-ment, are many and are presented in the literature with varying success in differentapplications. Descriptions of such treatments are provided in Section A.2.2.6 Summary of Findings from the LiteratureAs oil refineries shift to heavier feedstocks, the efficient processing of vacuumresidue into valuable liquid fuels becomes ever more important. Whilst carbonrejection is a cheap and well understood technique, catalytic hydroconversion (withunsupported catalysts in slurry-phase reactors) is more effective. MoS2 is one ofthe best catalysts available for this application but it is expensive. Recycling thesolid coke-catalyst agglomerate has been shown to prolong its use but the extendedtime-on-stream results in deactivation. A regeneration methodology is required forcatalytic hydroconversion with unsupported MoS2 to be economically viable.The complexity of the residue oil feedstock presents a challenge in understand-ing the mechanism of this catalytic hydroconversion reaction. This mechanism ismostly unknown with published theories being disputed due to the lack of exper-imental evidence. Mechanisms for deactivation are limited to conceptual models.Simplifying the system through model compound studies could allow for deduc-tion of the reaction and deactivation mechanisms and aid in the development ofeffective regeneration methodologies.29Chapter 3ExperimentalFundamental to any meaningful experimental investigation is reproducibility, witha significant part of this being a thorough understanding of the experimental ap-paratus, procedures and calculations conducted. As such, descriptions of theseaspects must allow for other researchers to examine, critique and, if desired, accu-rately recreate the experiments reported.This section presents the experimental objectives of this investigation, the ap-paratus and analytical equipment used to pursue those objectives and the associatedprocedures and calculations.3.1 Experimental Objectives and ProgrammeThe objectives of this study are the practical means whereby the key questionsposed in Chapter 1 may be addressed and answered, a conversion of the more the-oretical questions into specific, experimental goals. With these goals in mind, adetailed experimental programme may be established with clear expectations re-garding which factors are to be examined in each experimental series, what datais to be obtained and the reasoning behind these decisions. As each of the keyquestions of this study yields multiple objectives, each of which may require somedegree of explanation, they are discussed separately in Section 3.1.1 below, withthe experiments resulting from these objectives being programmatically presentedin Section 3.1.2.303.1.1 Interpretation of Questions to ObjectivesCan model compound experiments provide sufficient information to understand themechanism of the MoS2 catalyst? Presented as “Phase 1” of the experimentalprogramme (Section 3.1.2), answering this question requires multiple model com-pounds to be selected and subjected to a series of screening experiments. These ex-periments will determine both the applicability of each model compound in termsof representing a residue hydroconversion reaction (by comparing observed con-version with published residue hydroconversion experiments under the same con-ditions) and the degree to which the analytical results from each reaction allow fora mechanistic understanding to be developed (that is, if the chemical analysis ofthe products is simple enough to allow for meaningful, contextual interpretation).As discussed in Section 2.3.2, numerous model compounds have been used inpublished hydroconversion studies, with diphenylmethane (DPM) being selectedas the focus for this work. The reason for this choice is that it affords both theresistance to thermal cracking desired in the representation of heavy species inresidue hydroconversion reactions (largely due to it only possessing stable prox-imal Calkyl-Caryl bonds and aromatic rings) and due to its small and predictablecracking mechanism (illustrated in Figure 3.1a). The simplicity of this speciesdoes, however, have a negative aspect in that regardless of which Calkyl-Caryl bondbreaks, the products will be identical (as illustrated in Figure 3.1a). Furthermore,secondary cracking of these products, for instance of toluene (after stabilisation ofa benzyl radical) to a methyl and phenyl radical, would be unlikely (in this casedue to the instability of the C1 radical), meaning that an accurate determination ofthe rate of stabilisation of the primary cracking radicals may be difficult.Two other model compounds were thus included in this study. Diphenylethane(DPE) and diphenylpropane (DPP), shown in Figures 3.1b and 3.1c respectively,are structurally similar to DPM but with two- and three-carbon n-alkyl linkagesrespectively. It was thought that increasing the alkyl bridge length would be atrade-off between increased reactivity (the Calkyl-Calkyl bonds being more suscep-tible to thermolysis than the Calkyl-Caryl bonds), and hence reduced applicability toresidue hydroconversion, and increased analytical insight. The latter point arises asthere exists a complex series of subsequent cracking, isomerisation and stabilisa-31tion reactions for the primary radicals of DPE or DPP thermolysis with the relativerates of these pathways providing valuable information into the reaction system.Additionally, to simulate the model compound representing only a low con-centration of difficult to crack species in a residue feed, dilution experiments wereincluded in the screening studies. Decahydronaphthalene (decalin), illustrated inFigure 3.2, was selected as the solvent for these studies as it and similar species(such as tetralin) are commonly used for this purpose in published work. The sat-urated nature of decalin makes it an effective hydrogen donor and/or hydrogenshuttle in this system, simulating species in a residue feed which play similar roles.“Phase 1” allowed a model compound to be selected (from the DPM, DPE andDPP tested) and a decision made as to whether decalin dilution should be used.Does the reaction environment affect catalytic performance? “Phase 2” of theexperimental programme, following selection of a model compound and a deci-sion regarding the use of a diluent, is to determine the effect of various operationalaspects of the reaction environment on the observed reaction. In residue hydropro-cessing, the reaction is often quantified in terms of broad observed properties (suchas hydrogen uptake, changes in viscosity or simulated distillation curves) ratherthan detailed mechanistic parameters. Given the focus of this study on this latterpoint, however, such detail is of utmost importance in terms of both the modelcompound reaction and catalyst morphology. As such, many operational factorsapart from the standard reaction temperature, reaction pressure, catalyst loadingand so forth must be examined, with their impact on the reaction being quantifiedand, to as great a degree as possible, mitigated. One such factor is the control of thereaction temperature and the heat-up rate. With the thermal cracking range of 400to 445◦C exemplified in Figure 3.3, it may be seen that a slow heat-up rate may un-necessarily expose the reaction mixture to elevated temperatures thereby resultingin unwanted and unquantified reactions which may be avoided by a system offer-ing a faster, better controlled heat-up. Similarly, it is desirable to cool the systemrapidly following the reaction to minimise continued reaction beyond the desiredreaction time. Other equally important factors include the extent and impact of wallactivation (whereby a stainless steel reactor wall may become catalytically active),the influence of mixing on unsupported catalyst morphology in a “clean” system32+Either or(a)(1)(2)(1)(2)2 x+(b)(1)(2)++(1)(2)(c)Figure 3.1: Thermal cracking of diphenylmethane, diphenylethane anddiphenylpropane to their primary products to illustrate potential for sub-sequent reactions. (a) DPM cracking to phenyl and benzyl radicals. (b)DPE cracking to phenyl, benzyl and ethylbenzene radicals. (c) DPPcracking to phenyl, benzyl, ethylbenzene and propylbenzene radicals.Figure 3.2: Decahydronaphthalene structure.and the impact of hydrogen:reactant ratio in such a simplified reaction mixture.Given the complexities associated with the management of many of these as-pects, Phase 2 commenced with the design of a novel micro-reactor system, basedon the observations and findings from Phase 1, to allow for the quantification of,improved control over and selective elimination of unwanted influences.3300:00 00:15 00:30 00:45 01:00 01:150100200300400445Reactor temperature (°C)Heating time (hh:mm)~ 20 min~ 8 minStirred reactor heat-upDesired heat-upFigure 3.3: Comparison of a heating profile typical of the stirred batch reactorused in this study and a theoretical desired heating profile.What are the hydroconversion reaction mechanisms associated with fresh and de-activated MoS2 catalysts? With a model compound reaction system selected andundesirable effects quantified or eliminated, “Phase 3” of the experimental pro-gramme could commence wherein detailed reaction mechanisms could be pro-posed based on the experimental results from this model compound testing regime.These results and proposed mechanisms could be contrasted and compared withpublished hypotheses regarding the mechanism of hydroconversion and the role ofthe active catalyst in such a system (in particular, the widely held hydrogen radicalcapping theories depicted in Figure 2.6).With both the model compound testing regime and proposed active catalystmechanism in place, MoS2 catalyst samples, deactivated in residue hydroconver-sion reactions, could be evaluated and the results compared to both the fresh MoS2results and the residue hydroconversion results obtained for the same deactivated34catalyst samples (the residue deactivated samples and associated reaction data be-ing obtained from Rezaei and Smith [1]). Furthermore, testing and residue resultcomparison of thermally treated deactivated MoS2 (prepared as part of a publishedstudy by Rezaei and Smith [1]) would afford additional insight into the processof unsupported MoS2 deactivation and allow for an informed discussion regardingpossible mitigation and regeneration methodologies for such catalysts.3.1.2 Experimental ProgrammeAs discussed in Section 3.1.1, the experimental programme of this study may be di-vided into three phases. The first phase, model compound screening and selection,was performed in a commercially available stirred batch reactor with the resultsbeing used for the second phase, the design and development of a micro-reactor tooffer a more accurate and controlled testing platform. The third phase, evaluationof active and deactivated catalysts and the deduction of reaction and deactivationmechanisms, was performed in this novel micro-reactor. Table 3.1 provides a briefoverview of the experiments performed during this study, with a full listing of allexperiments in Section B.1.Phase 1 - Model Compound EvaluationSelect model compounds and diluentDiphenylmethane, diphenylethane and diphenylpropane were selected as themodel compounds for this study. Decahydronaphthalene was chosen as thediluent. A dilution of 3 wt% model compound in decalin, together withundiluted model compound, were selected for evaluation.Select catalyst and loadingThe oil-soluble precursor molybdenum octoate was selected for use in thisstudy, forming the active MoS2 catalyst in-situ through thermal decomposi-tion with CS2 (added as the sulphur source at three times the stoichiometri-cally required amount). Catalyst loadings of 0 ppm, 600 ppm and 1800 ppmMo were chosen for evaluation (following the works by Rezaei et al. [32]and Rezaei et al. [18]).35Table 3.1: Summary of experimental programme.Phase1Reaction temperature 1 Reaction time Catalyst loading Model compound Dilution 2(◦C) (h) (ppm Mo) (wt%)445 1 0 - 600 DPM, DPE, DPP 3420 - 435 1 0 DPP445 0 - 8 0 - 600 DPM 3 - 100415 - 445 1 1800 DPM 100Phases2and3Reactor 3 Mixing speed Reaction time Catalyst loading Total feed loading 4(RPM) (h) (ppm Mo) (µL)Inclined SS 5 0 1 1800 400 - 5001 - 4 0 - 1800 400Vertical SS 5 0 0 - 4 0 - 1800 400Glass insert 0 0 - 4 0 - 1800 150 - 4000 - 2250 1 0 - 18001502000 1 - 4 0 - 18001 1800 61- All 250 cm3 stirred batch reactor experiments were conducted at an initial reaction pressure of 13.79 MPa H2 and a mixer speed of700 RPM. 2 - Decalin used as diluent for normal experiments. 3 - All experiments were conducted using undiluted DPM at a reactiontemperature of 445◦C and an initial reaction pressure of 13.79 MPa H2. 4 - The volume of mixed feed (model compound, catalyst andCS2) pipetted into the reactor or insert. 5 - 316 stainless steel. 6 - Three residue hydroconversion coke-catalyst samples.36Select appropriate reactor and reaction conditionsA 250 cm3 stirred batch reactor, described in Section 3.2.2, was used for thepreliminary screening experiments of this study. This reactor was operatedat industrially applicable conditions of 415 - 445◦C and 13.79 MPa with areaction time (at temperature) of 0 - 8 h and a stirrer speed of 700 RPM.Conduct model compound screening experimentsFollowing calibration of the analytical instruments (described in Section 3.3)used in this study, each model compound was evaluated at the aforemen-tioned dilutions, catalyst loadings and reaction conditions. Blank experi-ments using the solvent (decalin) and anticipated major products (benzeneand toluene) were also performed. To ensure an accurate interpretation ofthe data obtained, experiments were repeated at least three times (limitedreagent and catalyst quantities notwithstanding).Evaluate resultsThe results from the screening experiments were examined alone for ana-lytical simplicity (allowing for a detailed mechanistic understanding) andagainst comparable published residue hydroconversion data [18, 32] to de-termine the applicability of each model compound and dilution to such asystem. The most appropriate model compound and dilution was selectedand major sources of uncontrolled influence in this reaction system (such asheating rate or wall activity) were identified.Phase 2 - Novel Reactor System Design and TestingQuantify undesirable influencesIdentifying the slow heat-up rate and catalytic wall activity as the major fac-tors which contribute to the reaction (outside of the “normal” operating pa-rameters of temperature, pressure, etc.), quantification of their effects onDPM model compound conversion and product yield could be performed todetermine the extent of their impact and hence the priority of their control ormitigation.Develop novel reactor system37A novel micro-reactor system was developed with a specific focus on thoseundesirable and difficult to control/mitigate factors shown to impact the re-action. This design and commissioning proceeded through an evolution-like process and is detailed in Section B.3 with the final design presented inSection 3.2.2.Evaluate reactor performance and refine methodologyOperation of the micro-reactor was scrutinised by conducting a series ofmodel compound experiments to correlate with those performed in the 250cm3 stirred batch reactor. These experiments allowed for a more precisequantification of the various influential factors, including those unique to a“clean” system or a micro-reactor setup. These factors included: heat-up rateon catalyst active phase formation and reaction, rate of wall activation andcatalytic influence, hydrogen diffusion rates and limitations and the impactof mixing on this process and the effect of mixing on unsupported catalystmorphology in a clean system. The results from these experiments allowedfor the adjustment and improvement of the testing methodology, improvingaccuracy and reproducibility.Phase 3 - Catalyst Study and Deactivation InvestigationConduct active catalyst experimentsWith the development of an accurate and reproducible reaction system where-in the activity of the catalyst could be isolated and studied without additionalinfluences (the first of its kind for such hydroconversion reactions), modelcompound experiments were conducted to gather data relating to the reac-tion rate and product distribution for an active MoS2 catalyst.Deduce active catalyst mechanismUsing the rate data and product distribution for the active MoS2 catalyst, thehydroconversion mechanisms proposed in the literature were scrutinised andmodifications or changes made based on these experimental observations.To examine the wider applicability of this mechanism, it was applied to datafrom published residue hydroconversion studies and found to explain andpredict the trends observed.38Conduct deactivated and heat treated catalyst experimentsTwo spent MoS2 catalyst samples were then evaluated in the model com-pound system. Both samples were recovered in the form of a coke-catalystagglomerate from residue hydroconversion studies [1]. The first had beensubjected to repeated recycles through the residue hydroconversion systemwhilst the second had been heat treated in an inert atmosphere after only asingle reaction. Both samples were shown to be deactivated in residue hy-droconversion experiments [1], affording minimal benefit over catalyst-freecomparisons.Deduce mechanism of deactivationComparison of model compound conversion and product distribution, juxta-posed with the published residue hydroconversion data [1, 18, 32], for freshMoS2 and the two deactivated MoS2 samples allowed for the hypothesisa-tion of a mechanism for the deactivation of this unsupported catalyst, inde-pendent of additional effects associated with parameters such as wall activityor slow heat-up rates. This information allowed for an informed discussionas to possible prevention and regeneration methodologies.3.2 Experimental Apparatus and SuppliesA list of the chemical species, their specifications and suppliers, used in this in-vestigation follows together with a description of the experimental and analyticalapparatus, their operation and a summary of the reaction conditions used.3.2.1 Reaction and Analytical SuppliesModel Compounds and DiluentDiphenylmethane (DPM, (C6H5)2CH2, Acros Organics, 99%), diphenyleth-ane(DPE, (C6H5)2C2H4, Alfa Aesar, 98+%) and diphenylpropane (DPP, C6H5)2C3H6,Alfa Aesar, 98%) were used in this study, without further purification, togetherwith decahydronaphthalene (decalin, C10H18, Sigma-Aldrich, mixture of cis- andtrans-decalin, 98%) as a solvent.39Catalyst PrecursorThe active MoS2 phase was formed in-situ through the reaction of the oil-solubleprecursor molybdenum octoate (C16H30MoO4, The Sheperd Chemical Company,molybdenum 2-ethylhexanoate in 2-ethylhexanoic acid, 15.5wt% Mo) and carbondisulphide (CS2, Sigma-Aldrich, ≥99.9%). The reaction occured rapidly duringthe heat-up period at temperatures as low as 415◦C (the lowest reaction temperatureinvestigated in this study).Reaction GasesUltra-high purity nitrogen (N2, Praxair, PP 4.8 [99.998%]) and hydrogen (H2, Prax-air, UHP 5.0 [99.999%]) were used to purge air from and pressurise the reactionsrespectively.Analytical Standards and GasesCalibration of the gas chromatographs (GC, see Section 3.3) was performed usinga certified gas mixture (Praxair). Gas chromatography-mass spectroscopy (GCMS)calibrations for liquid analyses were performed using standards prepared fromthe aforementioned model compounds and diluent together with benzene (C6H6,OmniSolv®, 99.94%) and toluene (C7H8, Fisher Scientific, 99.8%) (see Appen-dices C.2 and C.1 for details). All GC and GCMS apparatus used ultra-high purityhelium (He, Praxair, UHP 5.0 [99.999%]) for operation.3.2.2 Reactors and ConditionsIndustrial slurry-phase reactors are implemented almost exclusively as continuoussystems. On a laboratory scale, such systems are challenging due to increasedreagent and catalyst consumption (which may be expensive or available in limitedquantities), the complexities of operation (continuous feed mixing, product treat-ment, purging and recycling), the difficulty of maintaining stable operating con-ditions in a relatively small reaction volume (fluctuations increasing experimentaluncertainty and hampering reproducibility) and safety concerns arising from thecontinuous supply of toxic and flammable species to a unit operating at high tem-peratures and pressures.40The alternatives are batch or semi-batch reactors. Batch units usually consumeless reagent and catalyst, allow for easier control of reaction conditions and sim-plify feed introduction and product recovery. One drawback to the use of batchreactors in hydroconversion is the potential for hydrogen starvation. Whilst con-tinuous and semi-batch systems supply a constant stream of hydrogen to replenishthat consumed by the reaction, the hydrogen available to the reaction in a batchunit is limited.As discussed in Section 3.1.1 and indicated in Phase 2 of Section 3.1.2, tworeaction systems were utilised during this investigation. Both of these were batchreactors, the first being a commercially available stirred batch reactor and the sec-ond a custom designed and built batch micro-reactor. These units are describedbelow with additional details (such as operating procedures and reactor develop-ment) provided in Appendices B.2 and B.3. The hydrogen:model compound ratiowas varied in the micro-reactor system to examine the possibility of hydrogen star-vation in batch operation. The ratios studied were compared to those calculated foran equivalent semi-batch system.Batch ReactorUsed for model compound screening in Phase 1, a typical reaction in this 250 cm3stirred batch reactor began with the loading of 80 g of feed (model compound,diluent, CS2 and Mo octoate catalyst precursor). The system was purged with N2(500 sccm) before being purged (900 sccm) and pressurised with H2 to 13.8 MPa.With mixing held at 700 RPM, the temperature was ramped to 445◦C and held forbetween 0 and 8 h reaction time. After cooling, gas, liquid and solid products wererecovered for off-line analyses.Description and Safety The system used in Phase 1 of the study was a stirredbatch reactor supplied by the Parr Instrument Company (Parr). A schematic of thissystem is presented in Figure 3.4 with a photograph of how it was implemented inthe laboratory in Figure 3.5 and additional details in Section B.2.This reactor system was designed to allow for operation in either semi-batch orbatch mode. Details of features for semi-batch operation are provided in Section B.241with only those components relating to batch operation, selected for this study asdiscussed above, being described here. The total internal volume of the 316 stain-less steel reactor was 250 cm3, with liquid reagent loading masses and volumes,together with other pertinent operating parameters, being presented in Table 3.2.Either nitrogen (for purging air prior to heating) or hydrogen (for purging nitrogenand for pressurisation as a reagent) could be fed to the reactor, with the flow ratescontrolled by a Brooks Instrument 5850S mass flow controller.The reactor was heated by six 200 W heating rods positioned within the re-actor walls, all operating in unison and monitored by three OMEGA EngineeringInc. (OMEGA) K-type thermocouples, two positioned inside the reactor (one as abackup to the other) and one measuring the wall temperature, to ensure accurateand even temperature distribution. Temperature monitoring and control was per-formed by the workstation. A tight-fitting ceramic fiber insulating jacket was se-cured over the reactor, itself covered by a layer of foil-backed fiberglass insulation,to further improve control and efficiency. A cooling water loop passed through thereactor to help speed cooling following a reaction.The gas outlet line was wrapped with OMEGA high temperature heating tapeand covered with braided glass insulation. A Superior Electric variable transformercontrolled the current going to the heating tape so as to maintain an exit line temper-ature of between 60 and 65◦C to minimise condensation during operation and shutdown. The pressure of the reactor was monitored by both an analogue Ashcroft(welded, AISI 316 tube & socket) pressure gauge and an Ashcroft (A1906EP50)pressure transducer (for improved accuracy and data logging).Mixing of the reactor was achieved using a 316 stainless steel stirrer bar turnedby a Parr magnetic drive monitored and controlled by the workstation.The workstation, running the CalGrafix software package (CAL Controls Ltd,v3.0.0), controlled the temperature and mixing speed whilst monitoring and record-ing the temperature, mixing speed and reactor pressure.The entire reactor system was housed in a vented plexiglass cage equipped withHoneywell gas detectors (for H2 and H2S) to ensure that any gas leaks were quicklynoted and safely contained and removed. Rubber sheeting covered the floor ofthe cage to contain spills and prevent slipping. All bulk chemicals, gas cylinders,control boxes and the operator workstation were located outside the cage. The42reactor was equipped with a pressure rupture disk (Fike Corporation) such that,in the event of over-pressurisation, excess gas would be safely discharged througha buffer vessel (to capture any entrained liquid) and vented. High temperaturealarms, a back-up reactor thermocouple and the internal fail-safe mechanisms ofthe temperature controller helped avoid runaway temperatures.Table 3.2: Operating conditions of the stirred batch reactor.Condition ValueTemperature (◦C) 415 - 445Pressure (at temperature) (MPa) 13.79Reaction time (h) 0 - 8Mixer speed (RPM) 700Catalyst loading 1 (ppm Mo) 0 600 18000 0.315 0.945(g Mo octoate)CS2 loading 2 (g) 0.344(cm3) 0.273Dilution ratios (wt%) 3 100Model compound loading (g) 2.7 80(cm3) 2.7 80Decalin loading (g) 77.7 -(cm3) 86.7 -H2:model compound ratio 3 (cm3/cm3) 70.7 2.1(g/g) 0.31 0.01(mol/mol) 26.4 0.791- Catalyst masses calculated per total mass rather than model compound mass. 2 -Constant (at 1800 ppm Mo level) regardless of catalyst loading to ensure consistentsulphur concentrations. 3 - Calculated based on reaction conditions with mass and molarratios assuming DPM as model compound.43Figure 3.4: Process flow diagram of stirred slurry-phase batch hydroconversion reactor.1 - Mass flow controller and mass flow meters connected to flow control box. 2 - Mixer connected via power control box toworkstation. 3 - All six heating rods connected in parallel via distribution box to power control box. 4 - Pressure transducer connectedto workstation. 5 - Double thermocouple connected via power control box (for high temperature auto-shutoff) to workstation. Wallthermocouple monitored separately. 6 - Heating tape connected to variable transformer.44Figure 3.5: Stirred slurry-phase batch hydroconversion reaction system asimplemented in laboratory (reactor unloaded and not in operating po-sition).45Operation The operating procedure given below is a brief outline of one experi-mental run with full details provided in Section B.2.2.Beginning with a clean, dry, open reactor, the procedure was as follows:1. Weigh model compound, decalin (diluent), CS2 and Mo octoate and loadinto reactor2. Seal reactor3. Purge with 500 sccm N2 for 1 min4. Purge with 900 sccm H2 for 1 min5. Pressurise to 13.8 MPa with H26. Heat system to desired reaction temperature7. Maintain temperature for the desired reaction time8. Shut off heating and allow system to cool9. Depressurise system, collecting a gas sample if desired10. Open the reactor and recover liquid and suspended solids11. Clean reactor and internals with acetone12. Analyse reaction products (see also Section 3.3)(a) Gas samples may be analysed by GC as collected(b) Suspended solids may be recovered by vacuum filtration, washed withacetone and dried before analysis(c) Liquid samples require dilution with decalin and internal standard ad-dition prior to GCMS analysis46Micro-ReactorThe micro-reactor used for Phase 3, reaction and deactivation mechanism studies,is described below. In a typical experiment, 150 - 400 µL of feed (DPM, CS2 andMo octoate) were added to the glass insert (inner diameter of 4 mm and length of250 mm) and lowered into the stainless steel reactor shell (inner diameter of 6 mmand length of 500 mm). The system was purged with N2 (pressurising to 700 kPaand venting, repeating this cycle three times) and H2 (same cycling) before pres-surising to 13.8 MPa. Vortex mixing was begun at 2000 RPM and the temperaturewas ramped to 445◦C and held for 1 h reaction time. Gas analysis was done byin-line GC with liquid and solid products being recovered for off-line analyses.Description and Safety The reactor system described below is the final productof a lengthy design and testing program fully detailed in Section B.3.1. Figure 3.6shows a schematic of the micro-reactor system which was designed and constructedbased on the results of experiments conducted in the batch reactor of Section 3.2.2with a specific focus on improving operating parameter control and response andquantifying and/or mitigating the various factors found to have unwanted influ-ences on the reaction (such as wall activity). Figure 3.7 shows how this design wasimplemented in the laboratory.The micro-reactor developed for this study comprised a removable glass insert(inner diameter of 4 mm, length of 250 mm, total volume of approximately 2.85cm3) housed within a 316 stainless steel shell (inner diameter of 6 mm, length of500 mm). The shell was positioned vertically, resting on a vortex mixer, within an800 W Lindberg 55031 tubular furnace such that the reaction mixture within theinsert was centered in the isothermal zone. The temperature of the reaction mixturewas measured by a 1/16” OMEGA K-type thermocouple extending directly intothe liquid and controlled by an OMEGA CN8201 temperature controller whichwas interfaced with a Dell Precision Workstation 690 PC (3.73 GHz, 36 GB RAM,Windows 8 64 bit) allowing control and logging through OMEGA CN8-SW Multi-Comm software package (v3.16.000). Note that the insert was positioned such thatthe mouth extended beyond the heated zone as this was essential to prevent loss ofvolatile or supercritical species from the insert during reaction.47The system was equipped to allow for the supply of either nitrogen (for purgingair) or hydrogen (for pressurising for reaction), with needle valves being used tocontrol the flows. The gas supply line attaching directly to the reactor head was“pig-tailed” to allow vibration from mixing to be absorbed without damage to thelines or fittings. Pressure in the system was continuously monitored by both ananalogue ENFM USA Inc. pressure gauge and an OMEGA PX409-3.5KGUSBpressure transducer, the latter connected to the aforementioned computer and datalogged using OMEGA TRH Control (v1.03.11.297). After a reaction, the systemwas depressurised through a series of valves and the sampling port of an in-lineShimadzu Scientific Instruments Inc. (Shimadzu) GC-14B gas chromatograph (seeSection 3.3). A bubbler was used to visualise the depressurisation flow and keep itsuitably low to prevent liquid carry-over from the reactor. Following depressurisa-tion, the glass insert could be removed from the shell and the liquid and suspendedsolids recovered for analysis.One of the greatest challenges for a system of such a small size was how toachieve effective mixing. This was overcome through the use of a vortex mixerfor an externally applied mixing effect. The reactor shell was attached to a custommixer cup affixed to a Talboys 9456TAHDUSA advanced heavy-duty vortex mixerwith the reactor head secured using spring restraints, which served to both absorbvibration and hold the reactor in position. In this manner the base of the steel shellassembly moved in a circular motion with the head as the pivot point. A seriesof experiments were conducted to verify the efficacy of this mixing technique asdiscussed in Section 4.2.3 using a Megaspeed MS70K high speed camera recordingat 20,000 frames per second. Reaction product from a catalytic experiment (toensure accurate fluid composition and particle sizes) was sealed within its glassinsert using a cork and positioned inside a glass shell. This shell was the samediameter and length as its stainless steel counterpart, held in position about thesame pivot point above the vortex mixer. Glass beads of the same diameter as inthe stainless steel shell were used to position the insert at the same height. Visualmixing evaluations were performed both with and without the centrally locatedthermocouple.Despite the small size of this reaction system, compared to the stirred batchreactor described in Section 3.2.2, and hence the reduced amounts of liquid and48gas available to leak during any given run, a multitude of precautions were imple-mented to ensure safe operation. The assembly was installed inside a protectiveenclosure constructed of aluminium sheeting with a polycarbonate door. To reducethe risk of hydrogen ignition, following pressurisation, the hydrogen supply wasshut off as part of the standard procedure, limiting the amount of this gas availableto the system in the event of a leak. Furthermore, the enclosure was sufficientlylarge that even in the event of extraction failure and release of the hydrogen fromthe pressurised system, the air-hydrogen mix would still be below the lower ex-plosive limit. The temperature control system was fail safe and equipped withtwo alarms with automatic shut-off. The pressure monitoring software also had analarm to alert of over-pressurisation and a pressure release valve was installed (setto open at 17.24 MPa). To ensure stability of the entire system to mixer vibration,accidental jostling or seismic events, the enclosure and all equipment was securedto one another and/or to the counter.Table 3.3: Operating conditions and loadings utilised in micro-reactor.Condition/Factor Value/RangeTemperature (◦C) 415 - 445Pressure (at temperature) (MPa) 13.79Reaction time (h) 0 - 4Mixer speed (RPM) 0 - 2500Model compound loading 1 (mg) 151 402(µL) 150 400Catalyst loading (ppm Mo) 0 1800 0 1800(mg Mo octoate) 0 1.77 0 4.73CS2 loading 2 (mg) 0.65 0.65 1.72 1.72(µL) 0.51 0.51 1.36 1.36H2:model compound ratio 3 (cm3/cm3) 19.9 6.9(g/g) 0.09 0.03(mol/mol) 7.70 2.651- All micro-reactor experiments conducted using DPM as model compound. 2 -Constant (at 1800 ppm Mo level) regardless of catalyst loading to ensure consistentsulphur concentrations. 3 - Calculated based on reaction conditions and volume of glassinsert only.49Figure 3.6: Process flow diagram of batch slurry-phase hydroconversionmicro-reactor.1 - Thermocouple connected via OMEGA CN8201 controller to workstation formonitoring and control. 2 - Pressure transducer connected to workstation for logging.50Figure 3.7: Batch slurry-phase hydroconversion micro-reactor system as im-plemented in laboratory.51Operation Inherently similar in operation to the stirred batch reactor, use of themicro-reactor system had several distinct differences. A brief description of theprocedure for a typical run is given below with full details provided in Section B.3.2.Beginning with a clean, dry reactor shell and glass insert, the procedure was asfollows:1. Weigh DPM, CS2 and Mo octoate and load into glass insert2. Position insert within shell and seal the reactor3. Position assembly on vortex mixer and secure using spring mounts4. Purge with N2 (three 700 kPa - vent cycles)5. Purge with H2 (three 700 kPa - vent cycles)6. Pressurise to 13.8 MPa with H27. Heat system to 445◦C8. Maintain temperature for the desired reaction time9. Shut off heating and allow system to cool10. Depressurise system, starting in-line GC analysis when system pressure dropsto 70 kPa11. Open the reactor, remove insert and liquid and suspended solids12. Clean insert and thermocouple with acetone13. Analyse reaction products (see also Section 3.3)(a) Suspended solids may be recovered by settling or centrifugation to re-move the liquid product followed by washing with acetone and drying(b) Liquid samples require dilution with decalin and internal standard ad-dition prior to GCMS analysis523.3 Analytical Equipment and Data AnalysisThe experimental results of this study were obtained from the analyses of the solid,liquid and gaseous reaction products and their comparison with the liquid feedmixture to determine such comparators as model compound conversion and prod-uct yields. Analysis of the solid phase (predominantly MoS2) was exclusivelyqualitative in this study (to identify which species were present, their relative con-centrations and structures) whilst both gas and liquid phase analyses (performedby GC and GCMS) were both qualitative and quantitative.A brief description of the instruments utilised in these analyses is presented be-low. Additional details, operating procedures, examples of the data obtained fromeach, its interpretation and limitations, calibrations and sample calculations areprovided in Appendix C. An full uncertainty and propagation analysis is providedin Section C.2.3.3.3.1 Gas Product AnalysisAlthough gas samples for both the stirred batch and micro-reactor systems wereanalysed by gas chromatography, the method of their collection and the instrumentsused differed due to the differences in the experimental setups.For the 250 cm3 stirred batch reactor, product gas samples were collected fromthe vent gas following a run using Alltech Tedlar® gas sampling bags. These sam-ples were introduced into a Hewlett Packard (HP) 5890A GC, equipped with aPorapak® Q 80/100 mesh packed column and flame ionisation detector (FID), con-nected to a Hewlett Packard 3396 Series II integrator.The micro-reactor system was equipped with an in-line GC, simplifying thegas analysis by allowing for direct sampling of the product gases during reactorshutdown. The GC used in this setup was a Shimadzu GC-14B, equipped with anAgilent Technoligies Inc. (Agilent) HP-PLOT U column (19095P-UO4, ID 0.530mm, length 30 m, film 20.00 µm) and an FID, connected to a Shimadzu C-R8AChromatopac integratorCalibration of both GCs was performed using a certified hydrocarbon gas mix-ture (see Section 3.2.1 for gas mixture information).533.3.2 Liquid Product AnalysisLiquid products from both reactor systems were analysed on a Shimadzu GCMS-QP2010 gas chromatograph - mass spectrometer (GCMS) equipped with a Shi-madzu SHRXI-5MS column (220-94764-02, ID 0.25 mm, length 30 m, film 0.25mm) and AOC-20i autosampler (10 µL syringe).Due to the sensitivity of the system, it was necessary to dilute all liquid sampleswith decalin prior to analysis (to prevent overloading the column or saturating thedetector). Each sample was prepared in two dilution ratios, a richer sample foranalysis of lower concentration species (such as minor products) and a leaner ratiofor higher concentration species (such as unreacted model compound). One of themodel compounds not used in that particular reaction was added to each dilutedsample to act as an internal standard.Numerous standard samples for each model compound, benzene and toluenewere prepared and analysed for GCMS calibration.3.3.3 Solid Product AnalysisGiven that no solid precipitation products were expected to form, the solid productanalyses of this study served to: confirm the formation of the MoS2 active phasefrom the liquid precursors, identify contaminants or unexpected solid products anddetermine the structure of the MoS2 and other solid particles. Solids recovery wasby solid-liquid separation followed by washing with acetone and drying. Due tothe nature of the formation of the solids (precipitating as nanometer-sized parti-cles), size reduction for analysis was not necessary, with each solid sample simplyneeding to be appropriately mounted for analysis in each instrument.Compositional analysis of solid samples was performed by X-ray diffraction(XRD) and scanning electron microscopy with energy dispersive X-ray spectros-copy (SEM/EDX). A Bruker D8 Focus Bragg-Brentano was used for XRD analy-ses and a Hitachi S-2600N for the SEM/EDX analyses.Structural information relating to the recovered solids was obtained by trans-mission electron microscopy (TEM) and field emission scanning electron micros-copy (FESEM). TEM analysis was performed on an FEI Tecnai G2. FESEM wasperformed on a Hitachi S-4700.54Chapter 4Experimental ResultsThis chapter presents the results of the experimental program shown in Table 3.1(the stirred batch reactor experiments followed by the batch micro-reactor) togetherwith brief descriptions of the results and indications of major trends and points ofinterest. All values reported are subject to an experimental uncertainty of ±4.7%(see Section C.2.3) unless otherwise indicated.4.1 Stirred Batch Reactor4.1.1 Model Compound ScreeningTo determine which of the three model compounds selected for evaluation in thisstudy was the most suitable for extensive testing, a series of screening experimentswere conducted under industrially applicable conditions of temperature and pres-sure. Each model compound was to represent a low concentration species in thetotal feed and was hence diluted in decalin for these tests. The conversion resultsare presented in Table 4.1. An additional test was conducted using DPM whereindilution was performed with benzene instead of decalin, the data obtained beingused to determine the consequences of high model compound conversions pro-ducing a reaction mixture which would be supercritical under reaction conditions.Definitions and sample calculations for the conversion and yield comparators dis-cussed are presented in Section C.2.3.55ConversionBoth DPE and DPP were found to exhibit complete conversion, making them un-suitable for use in this study. DPM, however, presented conversions on a moremediocre level, making it more suitable. Changing the DPM diluent to benzenehad only a marginal impact on the observed conversion.Table 4.1: Conversion results obtained for diphenylmethane, diphenyleth-ane and diphenylpropane screening experiments performed in the stirredbatch reactor at 445◦C, 13.8 MPa H2, 1 h, 600 ppm Mo, 700 RPM at 3wt% in decalin.Model compound Catalyst loading Conversion(ppm Mo) (wt%)DPM 0 32.6600 35.7DPM 1 600 39.0DPE 0 99.6600 99.6DPP 0 99.6600 98.51- Benzene used as the solvent to study if supercritical phase has an influence on thereaction.In an attempt to reduce DPP conversion to a level suitable for study, a series ofexperiments was conducted examining the thermal reaction at decreasing reactiontemperatures. The results are presented in Figure 4.1. It may be seen that whilstreducing the temperature below approximately 430◦C does reduce the observedconversion, even temperatures as low as 420◦C result in only a 3 wt% decline inthe conversion (within experimental uncertainty of the other values).Product DistributionA summary of the major products from each of the screening experiments is pro-vided in Table 4.2. The presence of the catalyst is seen to have a dramatic influ-ence on the observed product distribution, shifting the reaction toward more hydro-genated and cracked products. It is clear from the DPP data in Table 4.2 that eventhough a reduction in the temperature does not have a significant effect on the ob-56420 425 430 435 440 44596979899100DPP conversion (wt%)Reaction temperature (°C)Figure 4.1: Conversion results obtained for diphenylpropane hydroconver-sion experiments performed in the stirred batch reactor at 420 - 445◦C,13.8 MPa H2, 1 h, 600 ppm Mo, 700 RPM at 3 wt% in decalin. Curveis for illustration of trend only.served conversion, the catalyst appears to quickly lose activity as the temperaturedeclines. This is evidenced by the product distribution shifting from the catalyt-ically hydrogenated and cracked distribution at higher temperatures to consistingof only two species (from the primary cracking of DPP) at lower temperatures.Whilst this may suggest thermal cracking of DPP to be the rate-limiting step inthis reaction, the catalyst only modifying the primary products, the excessivelyhigh conversion makes this claim impossible to substantiate without further study.These results also indicate that the catalyst acts predominantly in a hydrogenationrole rather than to exact catalytic hydrogenolysis. The latter mechanism would al-low the catalytic reactions to rapidly produce stable aromatic products, making theproduct more selective rather than the less selective distribution of hydrogenated,isomerised and cracked species observed.GC analyses determined that all of the screening experiments produced onlyminor quantities of gaseous products. The products from the 3 wt% dilution exper-iments could not be quantified due to their low concentrations. Table 4.3 presentsthe gas analyses from undiluted DPM experiments for quantification and identifica-tion of the gases formed. As may be seen, even after 6 h, very low concentrations57Table 4.2: Major products observed during model compound hydroconver-sion screening experiments of diphenylmethane (DPM), diphenylethane(DPE) and diphenylpropane (DPP) for different catalyst loadings and re-action temperatures in the stirred batch reactor at 13.8 MPa H2, 1 h, 700RPM at 3 wt% in decalin.Reactant ProductSpecies Composition 1 Species Composition 1(area%) (area%)DPM0 ppm Mo, 445◦C 600 ppm Mo, 445◦CMethylcyclohexane 39 Methylcyclohexane 24Toluene 32 Toluene 191-Methylcyclohexene 29 Ethylcyclohexane 173-Methylheptane 16Benzene 13Cyclohexylmethylbenzene 10DPE0 ppm Mo, 445◦C 600 ppm Mo, 445◦CToluene 43 Toluene 33Methylcyclohexane 18 Methylcyclohexane 23Ethylbenzene 12 Ethylbenzene 171,1-Diphenylethane 9 Ethylcyclohexane 101-Methylcyclohexene 9 1,1-Diphenylethane 9Anthracene 9 Benzene 8DPP0 ppm Mo, 445◦C 600 ppm Mo, 445◦CEthylbenzene 49 Ethylbenzene 67Toluene 33 Toluene 18Methylcyclohexane 10 Cyclohexylethylbenzene 151-Methylcyclohexene 8600 ppm Mo, 430◦C 600 ppm Mo, 420◦CEthylbenzene 78 Ethylbenzene 79Toluene 22 Toluene 211- Composition indicated is the percentage area from the GCMS chromatogram on aDPM-free basis (i.e. percentage of products formed) and limited to those speciescomprising >5%.of gaseous products were detected. For thermal reactions, little change was ob-served in the gas product composition with time. For catalytic systems, however,the gaseous products were observed to initially be suppressed to undetectable lev-els whilst after longer reaction times, larger species and higher concentrations wereobserved than for the equivalent thermal experiments.To confirm that the majority of products remained in the liquid phase and58Table 4.3: Gaseous products observed during hydroconversion of undiluteddiphenylmethane in the stirred batch reactor at 445◦C, 13.8 MPa H2, 700RPM.Catalyst loading Reaction time Product composition (wt%)(ppm Mo) (h) Methane Ethane Propane0 1 1.3 0.0 0.06 1.2 0.1 0.0600 1 0.0 0.0 0.06 1.8 0.4 0.2establish a quantification for the mass balance, the loaded and recovered massesfor eleven undiluted DPM experiments were compared and the mass balance wasfound to close to 98.9 ± 0.9 % (Section F.2).4.1.2 Benzene, Toluene and Decalin BlanksTo determine the stability of the major DPM decomposition products (benzene andtoluene) and the diluent (decalin) under reaction conditions, blank tests of thesespecies were conducted. Table 4.4 presents the feed and product compositions ofthe benzene and toluene tests together with their associated conversions. Whilstonly trace amounts of gaseous products were detected for these tests, Tables 4.5and 4.6 show the major liquid products found.Decalin blanks were observed to produce only trace gaseous and liquid prod-ucts, with a conversion of approximately 3 wt% (within experimental uncertaintyof zero). Given this value and that the feed was pure decalin, it was concludedthat decalin is, alone, unreactive under the reaction conditions. The liquid productswhich were formed, quantified in Table 4.7, all eluted from the column in the 9.00- 14.00 minute period. This is the portion of the chromatogram removed whenanalysing other products so as not to expose the GCMS filament and detector tohigh solvent concentrations. As such, the likelihood of confusion between decalinand model compound decomposition products was minimised.Of interest from this data is that both benzene and toluene show approximatelythe same conversion under the same reaction conditions and that the liquid prod-ucts formed are not only the same species, but are produced in roughly the sameamounts. It should be noted that the major products from benzene reaction are C759and C8 species, as are those from toluene decomposition (with the exception ofsome benzene being formed). Products from decalin decomposition were seen tobe cracking and dehydrogenation products.Table 4.4: Benzene and toluene blank test conversions performed in thestirred batch reactor at 445◦C, 13.8 MPa H2, 1 h, 600 ppm Mo, 700RPM, 3 wt% in decalin.Blank species Composition ConversionFeed Product(wt%) (wt%) (wt%)Benzene 3.3 2.9 11.0Toluene 3.9 3.6 10.1Table 4.5: Major liquid products detected for benzene blank tests performedin the stirred batch reactor at 445◦C, 13.8 MPa H2, 1 h, 600 ppm Mo, 700RPM, 3 wt% in decalin.Species Composition 1(area%)Methylcyclohexane 483-Methylheptane 36Ethylcyclopentane 151- Composition indicated is the percentage area from the GCMS chromatogram on abenzene-free basis (i.e. percentage of products formed) and limited to those speciescomprising >5%.4.1.3 Diphenylmethane StudiesWith DPE and DPP found to be unsuitable for this study due to their excessivelyhigh conversions under reaction conditions, the remainder of this work focused onDPM.Diluted DiphenylmethaneContinuing from the DPM results obtained during the model compound screeningexperiments, with 0 and 600 ppm Mo and 3 wt% dilution in decalin, the effect ofreaction time on the DPM hydroconversion reaction was examined, operating thesystem for between 0 and 8 h. The results are presented in Figure 4.2 for 0 and60Table 4.6: Major liquid products detected for toluene blank tests performedin the stirred batch reactor at 445◦C, 13.8 MPa H2, 1 h, 600 ppm Mo, 700RPM, 3 wt% in decalin.Species Composition 1(area%)Methylcyclohexane 413-Methylheptane 31Benzene 19Ethylcyclopentane 91- Composition indicated is the percentage area from the GCMS chromatogram on atoluene-free basis (i.e. percentage of products formed) and limited to those speciescomprising >5%.Table 4.7: Major liquid products detected for decalin blank test performed inthe stirred batch reactor at 445◦C, 13.8 MPa H2, 1 h, 600 ppm Mo, 700RPM.Species Composition 1(area%)Butylcyclohexane 411,2,3,4-tetrahydronaphthalene 2 361-Butylcyclohexene 231- Composition indicated is the percentage area from the GCMS chromatogram on adecalin-free basis (i.e. percentage of products formed) and limited to those speciescomprising >5%. 2 - Commonly known as tetralin.600 ppm Mo. There are two key features to note from this data. Firstly, even at0 h (when the reactor is immediately cooled upon reaching reaction temperature),a DPM conversion of approximately 30 wt% is obtained. This is indicative of thereaction proceeding at temperatures below 445◦C, a fact shown by the DPP tem-perature dependency in Figure 4.1 and discussed for DPM below in the context ofFigure 4.7. Secondly, there appears to be little difference between the conversionsobserved for 0 and 600 ppm Mo experiments upon reaching reaction temperature.After a short delay, however, the conversion for both Mo loadings begins the rise,the 600 ppm Mo system more rapidly than 0 ppm Mo. This results in a distinctlysigmoidal curve. It is the combination of this sigmoidal curvature and the non-zero conversion at 0 h which make first and second order kinetic fits unsuitable for61modeling these results. For completeness, these fits are show in Figure E.1 with thekinetic constants provided in Table E.2. Langmuir-Hinshelwood-Hougen-Watson(LHHW) kinetics would also not result in the different trends (one sigmoidal, onenot) observed. By such kinetics, the difference in the shape of the curves wouldindicate different kinetic expressions, indicative of different species adsorbing onthe active sites. Both systems begin with the same concentration of reactants andLangmuir adsorption theory would thus predict similar surface coverage for bothsystems. As both the 0 and 600 ppm Mo systems possess active sites but in dif-ferent numbers (for 0 ppm Mo only the FeS of the reactor walls and internals ispresent but with 600 ppm Mo that FeS is present in addition to the MoS2), wereadsorption of the same species governing the LHHW expression, the 0 ppm Mosystem would be expected to present with the same trend to that of the reactionwith 600 ppm Mo, simply lower. Were LHHW kinetics applicable to this system,these results would suggest different species adsorbing on the active sites of FeSand MoS2, a point to be considered in later analyses.0 2 4 6 80255075100DPM conversion (wt%)Reaction time (h)Figure 4.2: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the stirred batch reactor at 0 - 600 ppmMo, 445◦C, 13.8 MPa H2, 0 - 8 h, 700 RPM at 3 wt% in decalin togetherwith sigmoidal trend lines. Curves are for illustration of trends and arenot kinetic fits. ◦ 0 ppm Mo.  600 ppm Mo.As discussed in Section 2.5.2, the major products from the catalytic hydrocon-62version of DPM are expected to be benzene and toluene. This is particularly true ifthe primary role of the catalyst were to perform catalytic hydrogenolysis which, inDPM hydroconversion, has been shown [89] to result in very selective equimolaryields of benzene and toluene. As shown in Section 4.1.1, however, this was notalways the case for this series of experiments. Numerous other cracking, isomeri-sation and condensation species were observed to form during the reaction. Withthe gas selectivity shown to be low (see Table 4.3) and with the additional liq-uid product species changing from reaction to reaction, the mechanisms for theirformation being unclear, product analyses for these experiments were limited tobenzene and toluene yields, presented in Figures 4.3 and 4.4 respectively. As maybe seen, the 600 ppm Mo experiments begin with low yields of both benzene andtoluene with these increasing steadily with an increase in conversion, the tolueneyield exceeding that of benzene for all experiments, a trend contrary to both cat-alytic hydrogenolysis and the mechanism of Curran et al. [78]. The 0 ppm Moexperiments show higher levels of benzene and toluene than 600 ppm Mo, withthese levels also increasing with conversion (following what appears to be an ini-tial delay). Once more, toluene yield exceeds that of benzene. To better visualisethe benzene:toluene (B:T) molar ratio, Figure 4.5 is provided. It is clear that underall conditions evaluated in this series, the toluene yield exceeds the benzene yield,increasing rapidly upon reaching reaction temperature before appearing to levelout.Undiluted DiphenylmethaneIn an attempt to isolate the cause of the sigmoidal results and unexpected liquidproduct species, the diluent (which may have served a role as a hydrogen shuttle)was removed from the system and catalyst concentrations up to 1800 ppm Mo wereevaluated (in an attempt to overcome catalytic wall effects). The results from theseundiluted experiments, which included a temperature dependency study for DPMmimicking that of DPP from the screening studies, are presented below.The conversion of undiluted DPM with reaction time is shown in Figure 4.6,with this data showing a strongly sigmoidal shape. Whilst the conversion at 0h is lower than for the equivalent diluted experiments (approximately 3 wt% as6330 40 50 60 70 80 900.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.3: Benzene molar yield results obtained for diphenylmethane hydro-conversion experiments performed in the stirred batch reactor at 0 - 600ppm Mo, 445◦C, 13.8 MPa H2, 0 - 8 h, 700 RPM at 3 wt% in decalin.Curves are for illustration of trends only. ◦ 0 ppm Mo. 600 ppm Mo.30 40 50 60 70 80 900.00.20.40.60.81.0Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.4: Toluene molar yield results obtained for diphenylmethane hydro-conversion experiments performed in the stirred batch reactor at 0 - 600ppm Mo, 445◦C, 13.8 MPa H2, 0 - 8 h, 700 RPM at 3 wt% in decalin.Curves are for illustration of trends only. ◦ 0 ppm Mo. 600 ppm Mo.6430 40 50 60 70 80 900.00.20.40.60.81.0Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.5: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the stirred batch reactor at 0- 600 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 8 h, 700 RPM at 3 wt% indecalin. Curves are for illustration of trends only. ◦ 0 ppm Mo. 600 ppm Mo.65compared to 30 wt%), it is still not zero, indicating that reactions occur before thesystem reaches the reaction temperature. The conversion after extended reactiontimes is also lower in the undiluted system, but seemingly only for the 600 ppmMo experiments (approximately 40 wt% after 6 h with no dilution as compared toapproximately 80 wt% after 6 h for the diluted system). It should also be notedthat the 0 ppm Mo experiments, following an initial delay, show a rapid rise inDPM conversion to a level within experimental uncertainty of the 600 ppm Motests within the 6 h reaction time. As was the case for the diluted experiments,neither first nor second order kinetics provided satisfactory fits for the 600 ppmMo results, but a first order equation was found to be a reasonable approximationfor the 0 ppm Mo system (see Figure E.2 and Table E.7). Once more the change inthe shape of the curve indicates that, if LHHW kinetics were applied, the equationwould be different due to adsorption of different species on the FeS and MoS2active sites, making comparison using such kinetics of little value.In the undiluted systems the conversion for the 600 ppm Mo series is observedto initially exceed that of 0 ppm Mo with the latter attaining the same levels after6 h. Increasing catalyst concentration to 1800 ppm Mo was observed to increasethe conversion whilst reducing the temperature (shown in Figure 4.7) resulted in amore rapid decline in conversion than was found for DPP (see Figure 4.1). Figure 4.7may be interpreted in the context of the Arrhenius Law (as shown in Section F.3.3)to obtain Figure 4.8. From this plot the activation energy, Ea, for DPM hydrocon-version with 1800 ppm Mo is found to be 154±3 kJ/mol.Samples exemplifying the major liquid products observed in these reactionsare presented in Table 4.8 for the different reaction times and temperatures withthe structures of several of the species with less intuitive IUPAC names providedin Table 4.9 together with acronyms used in this study for clarity. Unlike themany products observed in the diluted model compound experiments (shown inTable 4.2), the undiluted experiments show fewer major species, many of whichcan be seen to contain either DPM or some hydrogenated form of it as a structuralcomponent. There remain, however, numerous additional species comprising frac-tions of a percent of the product which are not shown and yet which combine toform an appreciable portion of the total.Despite this progress in simplifying the reaction, with the major products in all660 2 4 601020304050DPM conversion (wt%)Reaction time (h)Figure 4.6: Conversion results obtained for undiluted diphenylmethane hy-droconversion experiments performed in the stirred batch reactor at 0 -1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 6 h, 700 RPM with sigmoidaltrend lines. Error bars indicate standard deviation. Curves are for illus-tration of trends and are not kinetic fits. ◦ 0 ppm Mo. 600 ppm Mo. △ 1800 ppm Mo.415 430 44501020304050 DPM conversion (wt%)Reaction temperature (°C)Figure 4.7: Conversion results obtained for undiluted diphenylmethane hy-droconversion experiments performed in the stirred batch reactor with1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 1 h, 700 RPM.671.38 1.40 1.42 1.44 1.46-2.2-2.0-1.8-1.6-1.4-1.2-1.0ln(-ln(1-X))1/T (K-1) x103Figure 4.8: Logarithmic conversion results obtained for undiluted diphenyl-methane hydroconversion experiments performed in the stirred batchreactor with 1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 1 h, 700 RPMagainst inverse reaction temperatures for determination of the activationenergy. Curve indicates fit by linear regression.68Table 4.8: Major products observed during undiluted diphenylmethane hydroconversion experiments performed in thestirred batch reactor with 0 - 1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 0 - 6 h, 700 RPM.Species Composition 1 Species Composition 1 Species Composition 1(area%) (area%) (area%)0ppmMo0 h, 445◦C 1 h, 445◦C 6 h, 445◦CToluene 51 Benzene 47 Benzene 43Benzene 49 Toluene 40 Toluene 35Fluorene 5 MBP1,2,3 11ETB 4 ETB 7MBP2 3 Fluorene 2o-Xylene 1600ppmMo0 h, 445◦C 1 h, 445◦C 6 h, 445◦CToluene 50 Benzene 24 Benzene 41Benzene 50 Toluene 19 Toluene 36MBP2,3 17 ETB 11HexF 16 MBP1,3 10Fluorene 13 Fluorene 21800ppmMo1 h, 415◦C 1 h, 430◦C 1 h, 445◦CToluene 41 Toluene 47 Toluene 49Benzene 37 Benzene 42 Benzene 45CHMB 16 CHMB 9 CHMB 43-Methylheptane 3 3-Methylheptane 2 HexF 1HexF 3 3-Methylheptane 11- Composition indicated is the percentage area from the GCMS chromatogram on a DPM-free basis (i.e. percentage of productsformed) and limited to those species comprising >1%.69Table 4.9: IUPAC names, structures and acronyms of several species ob-served in the diphenylmethane hydroconversion liquid products.IUPAC name Acronym Structure2-Methyl-1,1’-biphenyl MBP11-Methyl-3-(phenylmethyl)-benzene MBP21-Methyl-4-(phenylmethyl)-benzene MBP31-Ethyl-2-(1-phenylethyl)-benzene EPB1,1’,1”-(1-Ethanyl-2-ylidene)tris-benzene ETB4-Benzylbiphenyl BBP2-Methyl-1,1,1-triphenyl-propane MTP1,2,3,4,4a,9a-Hexahydrofluorene HexFCyclohexylmethylbenzene CHMB70tests being the benzene and toluene expected, some of the products appearing underone set of conditions are absent under another. To graphically examine the yieldsof these studies, it was thus necessary to group many of the lesser species to createlumps, one representing the cracking species (those with fewer carbons than DPM,excluding benzene and toluene) and another the isomerisation and condensation(isom. and cond.) species (those with carbon numbers equal to or higher thanDPM, excluding cyclohexylmethylbenzene). One species of particular interest, andits yield hence remaining outside of a lump, is cyclohexylmethylbenzene (CHMB)which was observed as a major product under 1800 ppm Mo but as only a minorproduct for the other catalyst loadings.The molar yields for benzene, toluene and CHMB are presented in Figures 4.9,4.10 and 4.11 respectively, with the mass yields for the lumped cracking and iso-merisation/condensation products in Figures 4.12 and 4.13 respectively. From thisdata it may be seen that the benzene and toluene yields initially increase with con-version, appear to pass through a maximum and then begin to decrease again. Thereis significant scatter in the results but it is noted that increasing the catalyst con-centration appears to reduce the benzene yield whilst increasing the toluene yield.This is better seen in Figures 4.14 and 4.15 wherein the trend of decreasing B:Tratio with increasing catalyst loading is clear. It is also evident that increasingthe catalyst loading increases the yield of CHMB, suggesting this to be a catalyticproduct. This trend, combined with the results in Table 4.8, indicate the primaryfunction of the catalyst to be hydrogenation rather than hydrogenolysis.Whilst benzene and toluene yields exhibit a maxima with increasing DPM con-version, the yields of the cracking and isom./cond. products steadily decline. Itappears that the yield of cracking species may increase with increasing catalystconcentration but the extremely low yield values make definitive trends difficult tosubstantiate. The formation of isom./cond. products, however, represents majorreaction pathways (which was clear from Table 4.8). For this lump it may be seenthat while 0 and 600 ppm Mo show roughly the same yields, increasing the catalystloading to 1800 ppm Mo clearly suppresses their formation.An additional benefit to undiluted model compound studies on this scale wasthat a measurable pressure change could be recorded during the reaction as shownin Figure 4.16. Such a significant change was not observed during diluted experi-710 10 20 30 40 500.00.20.40.60.81.0 Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.9: Benzene molar yield results obtained for undiluted diphenyl-methane hydroconversion experiments performed in the stirred batchreactor at 0 - 1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 0 - 6 h, 700RPM. Curve is for illustration of trend only. ◦ - 0 ppm Mo.  - 600ppm Mo. △ - 1800 ppm Mo.0 10 20 30 40 500.00.20.40.60.81.0 Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.10: Toluene molar yield results obtained for undiluted diphenyl-methane hydroconversion experiments performed in the stirred batchreactor at 0 - 1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 0 - 6 h, 700RPM. Curve is for illustration of trend only. - 0 ppm Mo.  - 600 ppmMo. △ - 1800 ppm Mo.720 10 20 30 40 500.00.10.2CHMB yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.11: Cyclohexylmethylbenzene molar yield results obtained forundiluted diphenylmethane hydroconversion experiments performedin the stirred batch reactor at 0 - 1800 ppm Mo, 415 - 445◦C, 13.8MPa H2, 0 - 6 h, 700 RPM. Curves are for illustration of trends only.◦ 0 ppm Mo.  600 ppm Mo. △ 1800 ppm Mo.0 10 20 30 40 500.000.020.040.06Other cracking products mass yield(g/gDPM reacted)DPM conversion (wt%)Figure 4.12: Mass yield of other cracking products (lumped) obtained forundiluted diphenylmethane hydroconversion experiments performedin the stirred batch reactor at 0 - 1800 ppm Mo, 415 - 445◦C, 13.8MPa H2, 0 - 6 h, 700 RPM. Curves are for illustration of trends only.◦ 0 ppm Mo.  600 ppm Mo. △ 1800 ppm Mo.730 10 20 30 40 500.00.10.20.30.4Other isom. and cond. productsmass yield (g/gDPM reacted)DPM conversion (wt%)Figure 4.13: Mass yield of isomerisation and condensation products(lumped) obtained for undiluted diphenylmethane hydroconversion ex-periments performed in the stirred batch reactor at 0 - 1800 ppm Mo,415 - 445◦C, 13.8 MPa H2, 0 - 6 h, 700 RPM. Curves are for illustra-tion of trends only. ◦ - 0 ppm Mo.  - 600 ppm Mo. Trend linefor both 0 and 600 ppm Mo. △ 1800 ppm Mo.0 10 20 30 40 500.000.250.500.751.001.25 Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.14: Benzene:toluene molar ratio obtained for undiluted diphenyl-methane hydroconversion experiments performed in the stirred batchreactor at 0 - 1800 ppm Mo, 415- 445◦C, 13.8 MPa H2, 0 - 6 h, 700RPM. Curve is for illustration of trend only. ◦ 0 ppm Mo. 600 ppm Mo. △ 1800 ppm Mo.740 600 1200 18000.00.51.01.5 Benzene:toluene molar ratio (mol:mol)Catalyst loading (ppm Mo)Figure 4.15: Benzene:toluene molar ratio obtained for undiluted diphenyl-methane hydroconversion experiments performed in the stirred batchreactor at 0 - 1800 ppm Mo, 415- 445◦C, 13.8 MPa H2, 0 - 6 h, 700RPM compared with catalyst loading. Error bars indicate standard de-viation. Curve is for illustration of trend only.75ments, presumably as a smaller quantity of model compound resulted in minimalgaseous exchange and undetectable pressure variations. From this data it may beseen that the pressure drop (a combination of the decline as hydrogen is consumedand the rise as gaseous products, shown to be minimal, are formed) has an inversesigmoidal trend. After an initial delay, both 0 and 600 ppm Mo experiments exhibita decline in the pressure (corresponding to a rapid uptake of hydrogen) followed bya leveling off. The 1800 ppm Mo data shows a greater pressure drop (indicative ofmore hydrogen being consumed and/or less gaseous products being formed) thanthe 0 and 600 ppm Mo experiments, but mimics the trend over the given range ofconversions.0 10 20 30 40 50-2.0-1.5-1.0-0.50.0 Pressure change (MPa)DPM conversion (wt%)Figure 4.16: Pressure change observed for undiluted diphenylmethane hy-droconversion experiments performed in the stirred batch reactor at 0- 1800 ppm Mo, 415- 445◦C, 13.8 MPa H2, 0 - 6 h, 700 RPM. Curvesare for illustration of trends only. ◦ - 0 ppm Mo.  - 600 ppm Mo.Trend line for both 0 and 600 ppm Mo. △ 1800 ppm Mo.XRD and TEM analyses were performed on the solid material recovered fromthe undiluted DPM experiments. The objectives of these tests were to confirm theformation of the desired MoS2 from the Mo octoate precursor and CS2, determinethe dimensions (sheet size and stack height) of the MoS2 crystallites and to identifyany other solid species formed during the reaction (such as FeS which would formon the walls of the reactor and slough off into the liquid).76Figures 4.17 and 4.18 show the XRD results for solids from 600 ppm and 1800ppm Mo experiments respectively. As may be seen, the major crystalline speciespresent in both samples is MoS2 with the 1800 ppm Mo sample showing a smallamount of graphite and the 600 ppm Mo sample showing trace FeS. The apparentdisappearance of FeS in the 1800 ppm Mo samples is believed to be due primarilyto dilution in the recovered solid. The solid samples analysed were recovered frommultiple experiments, conducted in a randomised order, and thoroughly mixed.It is proposed that the amount of FeS is the same in both samples but the loweramount of total MoS2 in the 600 ppm Mo experiment means the same mass ofFeS from the reactor walls represents a higher concentration in this sample thanthe 1800 ppm Mo one and thus presents more prominent peaks. Using the Scherrerequation (the equation and associated peak analysis may be found in Section C.3.2)with the integral breadth broadening factor of the 002 basal peak, the average stackheight of the 600 ppm Mo sample was calculated to be 2.31 nm, corresponding toapproximately 3.75 sheets. This indicates that the majority of the crystallites werecomprised of between three and four sheets. For the 1800 ppm Mo sample, thestack height was calculated to be 2.01 nm, approximately 3.26 sheets. Whilst this15% difference may indicate slightly higher stacks in the case of 600 ppm Mo, thismay simply be experimental variation given the numerous steps associated withthe recovery, preparation and analysis of solid samples by XRD.TEM analyses for the same 600 ppm Mo and 1800 ppm Mo samples were con-ducted (presented in Figures 4.19 and 4.20 respectively) to examine the structureof the MoS2 crystallites, obtain a comparative value for the stack height and de-termine the sheet width. Figure 4.19 further indicates the inter-plate and d-spacingof the crystallites at 0.62 and 0.27 nm respectively. These compare well with theliterature values for MoS2 of 0.6155 and 0.2738 nm respectively [106].Both solid samples assumed the same “rag” structure of sheets and showed al-most identical sheet width and stack height distributions (presented in Figures 4.21and 4.22, respectively, for the 600 ppm Mo sample). From these analyses, the mostcommon sheet width was observed to be 3 - 4 nm, with a most abundant stackheight of two sheets.7720 40 60 80010002000300040005000MoS2 {002}Counts (-)Angle, 2  (°)MoS2FeSFigure 4.17: X-ray diffractogram for solid material obtained from undiluted diphenylmethane hydroconversion exper-iments performed in the stirred batch reactor at 600 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 700 RPM.7820 40 60 8001000200030004000MoS2 {002}Counts (-)Angle, 2  (°)MoS2Carbon (C, graphite)Figure 4.18: X-ray diffractogram for solid material obtained from undiluted diphenylmethane hydroconversion exper-iments performed in the stirred batch reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 700 RPM.79Figure 4.19: Transmission electron microscopy image for solid material obtained from a diphenylmethane test in thestirred batch reactor at 600 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 700 RPM showing the both the inter-platespacing (d002 = 0.62 nm) and the d-spacing (d100 = 0.27 nm).80Figure 4.20: Transmission electron microscopy image for solid material ob-tained from a undiluted diphenylmethane test in the stirred batch reac-tor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 700 RPM.810 1 2 3 4 5 6 7 8 9 10 11 12 13 140255075100Number of sheets (counts)Sheet length (nm)Figure 4.21: Sheet size distribution of MoS2 crystallites using transmissionelectron microscopy data for solid material obtained from a undi-luted diphenylmethane test in the stirred batch reactor at 600 ppm Mo,445◦C, 13.8 MPa H2, 1 h, 700 RPM.1 2 3 4 5 6 7 8 90204060Number of stacks (counts)Number of sheets in stack (counts)Figure 4.22: Stack height distribution of MoS2 crystallites using transmis-sion electron microscopy data for solid material obtained from a undi-luted diphenylmethane test in the stirred batch reactor at 600 ppm Mo,445◦C, 13.8 MPa H2, 1 h, 700 RPM.824.2 Batch Micro-reactorWith the larger stirred batch reactor yielding unsatisfactory results (specifically interms of non-zero conversions at 0 h, appreciable experimental uncertainties andvariable product species), a micro-reactor system was designed to mitigate the ef-fects of wall catalysis and slow heat-up rates, the two factors thought to contributethe most to experimental variations. The development of the micro-reactor is pre-sented in detail in Section B.3.1 and progressed through three designs: an inclinedstainless steel, a vertical stainless steel and the final glass insert system. The firsttwo were constructed of seamless stainless steel tubing (3 mm inside diameter)with a central thermocouple and no mixing. The third used a glass insert (4 mm in-side diameter) with a central thermocouple, positioned within a stainless steel shell(6 mm inside diameter) and agitated with a vortex mixer. All three micro-reactorswere heated in a tubular furnace. DPM was used as the model compound withliquid loadings of 150 - 400 µL and catalyst concentrations of 0 - 1800 ppm Mo.The key comparators for these studies were the observed DPM conversion,benzene, toluene, CHMB, cracking and isom./cond. product yields (as discussedin Section 4.1.3 and defined in Section C.2.3) and the B:T molar ratio. Due to thesmall volume of model compound as compared to the total gas volume of the sys-tem, the pressure variations during reaction were not a reliable comparator. Whilstquantitative gas analysis by in-line GC showed only trace amount of gaseous hy-drocarbon products, the changing qualitative “fingerprint” of these products wasfound to provide some insight into the reaction (this was only performed for thevertical stainless steel and glass insert systems).4.2.1 Inclined Stainless Steel Micro-ReactorThe first functional incarnation of the micro-reactor was an inclined stainless steelunit. As shown in Table B.3, this system was used to study the effects of mul-tiple factors including: wall activation, hydrogen:DPM ratio (with and without acatalyst), heat-up rate, catalyst loading and reaction time.A series of 21 experiments conducted with 1800 ppm Mo to study the activa-tion of a fresh stainless steel wall. The results from these tests are presented inFigure 4.23. For multiple 1 h experiments, the DPM conversion was observed to83rise rapidly and then slow. To promote stabilisation of the reactor walls, experi-ments with longer reaction times were conducted. Wall activity was only observedto plateau after a cumulative reaction time of approximately 48 h. To ensure com-plete activation, the system was then left, loaded with liquid over a period of 72h (one weekend). Subsequent 1800 ppm Mo tests confirmed that DPM conver-sion had stabilised but at a far higher level as seen in Figure 4.24. The significantinfluence of the wall activity was concerning.0 5 10 15 2005101520DPM conversion (wt%)Sequential experiment number (-)1 h2 h3 h4 hFigure 4.23: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the inclined micro-reactor at 1800 ppmMo, 445◦C, 13.8 MPa H2, 1 - 4 h reaction time (as indicated), 0 RPMto study wall activation. N - 1800 ppm Mo activation study.Recalling that the hydrogen:DPM ratio was a point of concern in the stirredbatch system as per Table 3.2, a preliminary examination into this effect in an 1800ppm Mo catalytic reaction was examined by increasing the DPM volume from 400to 500 µL. All comparators were found to be within experimental uncertainty and,as such, the hydrogen:DPM ratio for this system is sufficient so as not to interferewith the reaction mechanism (for instance through hydrogen starvation). Moreintensive experiments were conducted when the glass insert was introduced.To determine the influence of the faster heat-up rate (the micro-reactor able toachieve reaction temperature in 20 min as compared to 80 min for the stirred batchsystem), the results from rapid heat-up experiments were compared, in Table 4.10,840 1 2 3 4010203040DPM conversion (wt%)Reaction time (h)Figure 4.24: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the inclined micro-reactor at 1800 ppmMo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM to compare the results obtainedfor an active and inactive wall. Error bars indicate standard deviation.△ - 1800 ppm Mo with active wall. N - 1800 ppm Mo activation study.with those where the rate was set to mimic that of the stirred batch reactor (note thatthese were obtained during the stabilising period of reactor wall activation). Whilstmany of the values lie within experimental uncertainty, the faster ramp rate doesappear to result in lower conversion, less benzene (and hence a reduced B:T ratio)and less CHMB. These results are believed to be a more accurate representationof the reaction occurring at 445◦C rather than in the tail end of the heat-up aswas observed in the stirred batch reactor. Further corroboration of this is seen inFigure 4.25 (and subsequent datasets) where DPM conversion at 0 h is negligible.So as to compare the performance of this micro-reactor with that of the stirredbatch reactor, to determine the influence of the catalyst and the kinetics of thereaction, a series of experiments was conducted varying the catalyst concentrationand the reaction time. The DPM conversions, products yields and B:T ratios arepresented below. It is noted that there is still a fairly large scatter associated withthese results, disappointing with improved reproducibility being one of the aims ofswitching to a micro-reactor.DPM conversion data in Figure 4.25 shows the conversion at 0 h to be the85Table 4.10: Comparison of results for diphenylmethane hydroconversion ob-tained in the inclined micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPaH2, 1 h, 0 RPM to examine the effect of heating rate.Ramp rate DPM conversion Product yield(◦C/min) (wt%) (mol/molDPM reacted)Benzene Toluene CHMB5.5 8.6 0.68 0.73 0.1921.5 7.5 ± 1.1 0.63 ± 0.10 0.74 ± 0.10 0.05 ± 0.01Product yield B:T ratio(g/gDPM reacted) (mol/mol)Cracking Isom./cond.0.05 0.05 0.940.05 ± 0.02 0.07 ± 0.02 0.85 ± 0.06desired 0 wt%, indicating that minimal reaction occurs during the rapid heat-upperiod. The curves shown are merely trend lines as first and second order kineticfitting was not possible giving the complex shape of the data. The close correlationsof these curves suggests that LHHW kinetics do not apply either for, unlike the dif-ference in the trends observed in the stirred batch reactor (Figures 4.2 and 4.6) andattributed to different kinetic expressions resulting from different species adsorbingon the FeS and MoS2 active sites, the similarity in the trends of Figure 4.25 sug-gest comparable mechanisms at play (i.e. the same species adsorbing on both FeSand MoS2 active sites). Following an initial, rapid rise in conversion, the catalyticsystems pass through a curious stabilisation period around 2 h before continuing torise. This is in contrast with the clearly sigmoidal shaped curves from the stirredbatch reactor. It is noted that whilst the conversion for 600 ppm Mo is comparableto that from the stirred batch reactor, neither 0 nor 1800 ppm Mo share this charac-teristic. 0 ppm Mo exhibits a smoothly increasing curve with a higher conversionthan previously seen whilst 1800 ppm Mo is observed to have a DPM conversionlower than that of 600 ppm Mo below approximately 3 h reaction time.To examine the changes in the reaction mechanism that result in these trends,the products yields must be studied. From the benzene, toluene, CHMB, crackingand isom./cond. product yields in Figures 4.26 through 4.30, it may clearly be seenthat the trends differ from those of the stirred batch reactor. Whilst all catalystloadings show roughly the same benzene yield, 1800 ppm Mo shows a slightlysuppressed toluene yield. The benzene:toluene ratio (shown in Figure 4.31) for860 1 2 3 4010203040DPM conversion (wt%)Reaction time (h)Figure 4.25: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the inclined micro-reactor at 0 - 1800ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Error bars indicatestandard deviation. Curve is for illustration of trend and is not a ki-netic fit. ◦ - 0 ppm Mo.  - 600 ppm Mo. △ - 1800 ppm Mo.all three remains, however, very close to unity. The CHMB yield may be seen toincrease with increasing catalyst loading, mimicking the trend seen in the stirredreactor but at higher levels. The yield of other isom./cond. products declines withincreasing catalyst concentration as was noted in the stirred reactor, shadowing thetrends over the conversion range although at elevated levels for 0 and 600 ppm Mo.Unlike the stirred system, however, the trends for cracking products show a distinctdivergence of the 1800 ppm Mo system, producing more cracked species at higherconversion than either the 0 or 600 ppm Mo experiments, the latter two showingsimilar results to one another.Given the similarities between the results observed in both the stirred batchreactor and inclined micro-reactor between 0 and 600 ppm Mo, the 600 ppm Mocatalyst concentration was omitted from subsequent experiments.To help determine what mechanistic changes result in the observed differencesin conversion and yield between the inclined micro-reactor and stirred batch sys-tems, Table 4.11 shows the composition of several key liquid products. These sam-ples cover both the catalyst loading and reaction time ranges. It is noted that the8710 20 30 400.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.26: Benzene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the inclined micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo.  600 ppmMo. △ 1800 ppm Mo.reaction product is now clean enough that all products above 0.25 area% may beshown without excessive complexity. It seems apparent that whilst the 0 ppm Moand, to a lesser extent, 600 ppm Mo systems exhibit some degree of random re-action resulting in the formation of numerous side products, the 1800 ppm Moreaction has access to a more streamlined pathway, producing only five products inroughly the same quantities regardless of DPM conversion.4.2.2 Vertical Stainless Steel Micro-ReactorIn preparation for the implementation of the glass insert micro-reactor, the inclinedstainless steel unit was oriented vertically and the catalyst concentration (limitedto 0 and 1800 ppm Mo) and reaction time experiments repeated. The change inorientation would reduce the gas-liquid interface area as well as the contact be-tween the liquid and the active reactor and thermocouple walls (the thermocouplesheath itself is 316 stainless steel and subject to the same activation as the reactor),affecting the observed conversion and product yields.Figure 4.32 shows the DPM conversion for the series. The curves shown are8810 20 30 400.00.20.40.60.81.0Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.27: Toluene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the inclined micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo.  600 ppmMo. △ 1800 ppm Mo.10 20 30 400.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.28: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the in-clined micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 -4 h, 0 RPM. Curves are for illustration of trends only. ◦ 0 ppmMo.  600 ppm Mo. △ 1800 ppm Mo.8910 20 30 400.000.020.040.06Other cracking products mass yield(g/gDPM reacted)DPM conversion (wt%)Figure 4.29: Mass yield of other cracking products (lumped) obtained fordiphenylmethane hydroconversion experiments performed in the in-clined micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4h, 0 RPM. Curves are for illustration of trends only. ◦ 0 ppmMo.  600 ppm Mo. △ 1800 ppm Mo.10 20 30 400.00.10.20.30.4Other isom. and cond. productsmass yield (g/gDPM reacted)DPM conversion (wt%)Figure 4.30: Mass yield of isomerisation and condensation products(lumped) obtained for diphenylmethane hydroconversion experimentsperformed in the inclined micro-reactor at 0 - 1800 ppm Mo, 445◦C,13.8 MPa H2, 0 - 4 h, 0 RPM. Curves are for illustration of trends only.◦ 0 ppm Mo.  600 ppm Mo. △ 1800 ppm Mo.9010 20 30 400.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.31: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the inclined micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo.  600 ppmMo. △ 1800 ppm Mo.91Table 4.11: Major products observed during diphenylmethane hydroconversion experiments performed in the inclinedmicro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 - 4 h, 0 RPM (using acronyms from Table 4.9).Species Composition 1 Species Composition 1 Species Composition 1(area%) (area%) (area%)0ppmMo1 h 2 h 4 hToluene 48 Toluene 46 Toluene 45Benzene 40 Benzene 41 Benzene 40CHMB 6 CHMB 6 CHMB 7HexF 4 HexF 3 MBP2,3 3EPB 2 EPB 2 HexF 2MTP 2 ETB 2Fluorene 1600ppmMo1 h 2 h 4 hToluene 46 Toluene 47 Toluene 48Benzene 41 Benzene 42 Benzene 41CHMB 6 CHMB 7 CHMB 7HexF 4 HexF 3 HexF 2ETB 2 ETB 1 Fluorene 1EPB 2 MBP3 11800ppmMo1 h 2 h 4 hToluene 41 Toluene 40 Toluene 42Benzene 38 Benzene 38 Benzene 39CHMB 17 CHMB 19 CHMB 16HexF 2 HexF 2 HexF 23-Methylheptane 1 3-Methylheptane 1 3-Methylheptane 21- Composition indicated is the percentage area from the GCMS chromatogram on a DPM-free basis (i.e. percentage of productsformed) and limited to those species comprising >0.25%.92first order fits with the kinetic coefficients presented in Table 4.12. The hard in-flection at 2 h observed in the inclined system has all but disappeared, with anyvariation remaining being encapsulated by experimental uncertainty. Curiously,the results of the 0 ppm Mo experiments now exceed those of the 1800 ppm Moones (although the kinetic coefficient of 0.082 h−1 for the 0 ppm Mo system is veryclose to the 0.088 h−1 observed in the stirred batch reactor). Once more, the appli-cation of LHHW kinetics would suggest similar kinetic expressions for both the 0and 1800 ppm Mo systems, i.e. the same species adsorbing on the active sites ofboth FeS and MoS2, a conclusions which is in contradiction with the observationin the stirred batch reactor. It is thus clear that LHHW kinetics are not applicableto this catalytic system.0 1 2 3 40102030 DPM conversion (wt%)Reaction time (h)Figure 4.32: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the vertical stainless steel micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Error barsindicate standard deviation. Curves shown are first order kinetic fits asper Table 4.12. ◦ 0 ppm Mo. △ 1800 ppm Mo.93Table 4.12: Coefficients for the kinetic models of diphenylmethane hydro-conversion for data obtained in the vertical micro-reactor at 0 - 1800ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM, depicted in Figure 4.32.Catalyst loading Kinetic coefficient 1(ppm Mo) (h−1)0 0.082±0.0021800 2 0.076±0.0011- rDPM 0ppm = k0ppm.CDPM , rDPM 1800ppm = k1800ppm.CMo.CDPM .2- Reporting k1800ppm.CMo for comparison in context of the experiments.The product yield data is presented in Figures 4.33 through 4.36. Interestinglyin this system, no cracking products were detected for any of the experiments andthe only isom./cond. product observed was fluorene, with an fluorene yield plotthus replacing the isom./cond. lump below.Despite the similarity in the DPM conversion data, the product compositionfor these experiments is surprising. The 1800 ppm Mo system now clearly ex-hibits lower benzene and toluene yields than its 0 ppm Mo counterpart and yet theB:T ratio (seen in Figure 4.37) remains almost the same for both and now slightlybelow unity (at approximately 0.95). The CHMB yield for 1800 ppm Mo, main-taining a similar range to the inclined reactor experiments, now appears to passthrough a minimum as conversion increases. 0 ppm Mo shows a larger initial yieldof CHMB than was observed in the inclined system and this yield is seen do de-crease sharply with conversion. These trends (excluding the increase in the 1800ppm Mo data) are reminiscent of the stirred reactor results and again suggest thatcatalytic hydrogenolysis in not a significant mechanism in this reaction system. Ifthe catalyst were performing both hydrogenation and hydrogenolysis, one wouldanticipate saturated rings to be present in the product is appreciable quantities (theresult of CHMB hydrogenolysis) which is not observed. Instead, both catalyst con-centrations result in comparable yields of fluorene which decreases with increasingconversion.Figures 4.38 and 4.39 show typical gas product analyses for the 0 and 1800 ppmMo systems after reaction times of 1 and 4 h. Quantification of C1 to C4 speciesis provided in Table 4.13. As was seen in the stirred batch reactor, the gaseous940 10 20 300.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.33: Benzene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the vertical micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.products constitute a minimal mass toward the total products. The distributionof these species is, however, very interesting. The trend of the catalytic systemto produce more and larger fragments is apparent. For the 0 ppm Mo systems,increasing the reaction time from 1 to 4 h results in an increase in the gas productsformed, with the only apparent difference being a minor increase in C6 isomers.The 1800 ppm Mo systems, exhibit roughly the same species as the 0 ppm Mo ones,but the total amount and distribution differs. 1800 ppm Mo is noted to producemore of all gaseous species and that, with increasing reaction time, the amounts ofthese species increase. This includes notable increases in the C5 and C6 isomersnot observed to such an extent in the 0 ppm Mo experiments.Despite the magnitude of the benzene and toluene peaks in these chromatograms,their composition in the gas phase remains less than 1 wt%. The larger benzenepeak, compared to toluene, is due to its lower vapour pressure and more rapid rateof vaporisation. The low concentrations escaping to the gas phase were not thoughtto alter the liquid composition to any noticable extent.950 10 20 300.00.20.40.60.81.0Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.34: Toluene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the vertical micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.Table 4.13: Major gaseous products observed during diphenylmethane hy-droconversion experiments performed in the vertical micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 - 4 h, 0 RPM.Conditions Composition (wt%)Methane Ethane Propane iso-Butane Butane0 ppm Mo, 1 h 0.17 0.01 0.02 0.001 0.0020 ppm Mo, 4 h 0.27 0.03 0.03 0.002 0.011800 ppm Mo, 1 h 0.18 0.05 0.05 0.001 0.031800 ppm Mo, 4 h 0.37 0.07 0.08 0.004 0.04960 10 20 300.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.35: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the verti-cal micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0RPM. Curves are for illustration of trends only. ◦ 0 ppm Mo.△ 1800 ppm Mo.0 10 20 300.00.10.2Fluorene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.36: Fluorene molar yield for diphenylmethane hydroconversion ex-periments performed in the vertical micro-reactor at 0 - 1800 ppm Mo,445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves are for illustration oftrends only. ◦ 0 ppm Mo. △ 1800 ppm Mo.970 10 20 300.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.37: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the vertical micro-reactor at0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.9800:00 05:00 10:00 15:00 20:00 25:00 30:000.00.51.01.534Signal intensity (-) x104Retention time (mm:ss)MethaneEthanePropaneButaneBenzeneTolueneC6 isomersC5 isomersisoButane(a)00:00 05:00 10:00 15:00 20:00 25:00 30:000.00.51.01.55101520Signal intensity (-) x104Retention time (mm:ss)(b)Figure 4.38: Examples of gas chromatograms obtained for diphenylmethanehydroconversion experiments performed in the vertical micro-reactorat 0 ppm Mo, 445◦C, 13.8 MPa H2, 1 - 4 h, 0 RPM. (a) 1 h reactiontime. (b) 4 h reaction time.9900:00 05:00 10:00 15:00 20:00 25:00 30:000.00.51.01.5345Signal intensity (-) x104Retention time (mm:ss)(a)00:00 05:00 10:00 15:00 20:00 25:00 30:000.00.51.01.551015Signal intensity (-) x104Retention time (mm:ss)(b)Figure 4.39: Examples of gas chromatograms obtained for diphenylmethanehydroconversion experiments performed in the vertical micro-reactorat 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 - 4 h, 0 RPM. (a) 1 h reactiontime. (b) 4 h reaction time.1004.2.3 Glass Insert Micro-ReactorUse of the glass insert micro-reactor allowed for yet more control over the factorsinfluencing the reaction. Exposure to the catalytically active reactor wall was elim-inated (although the active thermocouple wall was still in contact with the liquid),the rapid heat-up and cool-down rates allowed for better control over the tempera-tures to which the reaction mixture was exposed and implementation of an externalvortex mixing system allowed for agitation.Unmixed Comparison with Stainless Steel Micro-ReactorThe first studies using this system were unmixed experiments for comparison withthe vertical stainless steel micro-reactor. Figure 4.40 shows a direct comparisonwherein it may be seen that the glass system exhibits a conversion roughly halfthat of the stainless steel reactor. It is noted that the experimental uncertainty isnow lower and that the 1800 ppm Mo experiments exceed their 0 ppm Mo coun-terparts for all reaction times studied. Once more the trends may be roughly ap-proximated by first order kinetic models, the coefficients presented in Table 4.14.The closeness of the 0 and 1800 ppm Mo results are suspected to be due to thecatalytic influence of the thermocouple walls (FeS) acting as additional catalyticcentres. There appears to be little indication that the FeS and MoS2 form anymanner of synergistic effect, an observation supported by published studies in thisregard [107–109].Table 4.14: Coefficients for the kinetic models of diphenylmethane hydro-conversion for data obtained in the glass insert micro-reactor at 0 - 1800ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM, depicted in Figure 4.32.Catalyst loading Kinetic coefficient 1(ppm Mo) (h−1)0 0.036±0.0011800 2 0.042±0.0011- rDPM 0ppm = k0ppm.CDPM , rDPM 1800ppm = k1800ppm.CMo.CDPM .2- Reporting k1800ppm.CMo for comparison in context of the experiments.Figures 4.41 through 4.44 illustrate the yields for benzene, toluene, CHMB andthe isom./cond. lump with the B:T ratio in Figure 4.45. Only trace amounts of other1010 1 2 3 40102030 DPM conversion (wt%)Reaction time (h)Figure 4.40: Comparison of conversion results obtained for diphenylmethanehydroconversion experiments performed in the vertical stainless steeland glass insert micro-reactors at 0 - 1800 ppm Mo, 445◦C, 13.8 MPaH2, 0 - 4 h, 0 RPM. Error bars indicate standard deviation. Curvesshown are first order kinetic fits as per Table 4.12 for the stainless steelreactor and Table 4.14 for the glass insert. • 0 ppm Mo, stainlesssteel. N 1800 ppm Mo, stainless steel. ◦ 0 ppm Mo,glass insert. △ 1800 ppm Mo, glass insert.cracking species were detected for either catalyst concentration for the reactiontimes investigated. This was also the case for CHMB in the 0ppm Mo experiments,wherein only trace concentrations were observed. Furthermore, fluorene was themajor constituent of the isom./cond. lump except below 2 h reaction time whenboth fluorene and 4-benzylbiphenyl (BBP) were observed.As per the vertical micro-reactor, the benzene yield for 1800 ppm Mo is notedto be below that of the 0 ppm Mo but in the glass insert, the toluene yields for bothcatalyst concentrations quickly reach the same level and follow the same trend.The benzene:toluene ratio also differs with 0 ppm Mo showing a high initial ratiorapidly declining with increasing conversion to join that of the roughly constant1800 ppm Mo at approximately 0.92.Figure 4.43 presents the CHMB yield for 1800 ppm Mo, only trace amountsdetected for 0 ppm Mo. It may be seen that the yield remains almost constant forall conversions at a level below those observed for the stainless steel micro-reactor.102The isom./cond. lump products are again seen to follow a decreasing trendwith increasing DPM conversion with 1800 ppm Mo below that of 0 ppm Mo.0 5 10 15 200.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.41: Benzene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.Visual Mixing StudiesWith no literature published on vortex mixing in micro-reactors, determination ofa suitable mixing speed was a two step process. The first determined, visually, atwhat mixing speed and/or liquid volume a vortex was formed, the impact of thecentrally-located thermocouple and whether or not particles were suspended bythe motion. The second was a study of the impact of different mixing speeds (bothwith and without a full vortex) on the system performance.The visual study was performed using a high speed camera to film a glassmock-up of the micro-reactor system. Reaction product from an 1800 ppm Moexperiment (445◦C, 13.8 MPa H2, 1 h, 0 RPM) was sealed in its insert and po-sitioned inside the glass shell as described in Section 3.2.2. These observationswere performed at ambient conditions (approximately 101.325 kPa and 20◦C) withSection 5.2.4 discussing how the fluid, and hence mixing, properties would change1030 5 10 15 200.00.20.40.60.81.0Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.42: Toluene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.0 5 10 15 200.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.43: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h,0 RPM. Curves are for illustration of trends only. ◦ 0 ppm Mo.△ 1800 ppm Mo.1040 5 10 15 200.00.10.2Other isom. and cond. productsmass yield (g/gDPM reacted)DPM conversion (wt%)Figure 4.44: Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h,0 RPM. Curves are for illustration of trends only. ◦ 0 ppm Mo.△ 1800 ppm Mo.0 5 10 15 200.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.45: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPM. Curves arefor illustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.105at reaction conditions.A problem quickly became evident as illustrated in Figure 4.46. With 400 µLof liquid, even the maximum mixing speed of 2500 RPM of the heavy duty vor-tex mixer was insufficient to establish a vortex and agitate the full liquid volume,with the solid particles remaining settled. Further studies determined the height ofthe vortex to be a function of the mixer speed (this relationship being illustratedin Figure 4.47) and independent of liquid volume. It was thus necessary, to en-sure agitation of all liquid in the insert and suspension of the solid particles, forthe volume of liquid to be reduced to only 150 µL. With this volume, shown inFigure 4.48, solid suspension occurs at 2250 RPM and above. It may be seen,however, that a true vortex is not established with fluid instead taking the form ofa rotating concave wave for all speeds tested.Figure 4.46: 2500 RPM mixing of 400 µL of reaction product obtained fordiphenylmethane hydroconversion performed in the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM as seen inglass mock-up.This data represents an ideal system where the liquid movement is unobstructed.1061500 2000 2250 25000102030"Vortex" height (mm)Mixing speed (RPM)Figure 4.47: Effect of mixing speed on “vortex” height as studied in glassmock-up using reaction product obtained for diphenylmethane hydro-conversion performed in the glass insert micro-reactor at 1800 ppmMo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM. Curve is for illustration oftrend only. Data for 150 µL and 400 µL were indistinguishable.When implemented in the micro-reactor, however, a thermocouple extends into theliquid of the insert. To determine what effect this obstruction would have on themixing, the central thermocouple was duplicated in the glass mock-up, the resultsbeing presented in Figures 4.49 and 4.50 for 400 µL and 150 µL liquid loadings,respectively. It is apparent that the smooth wave motion is no longer present withthe movement of the suspended thermocouple allowing it to act as a stirrer bar,agitating the liquid, suspending solids and entraining large amounts of gas. Thisis most effective for 150 µL where gas entrainment begins at approximately 2000RPM and increases through 2250 RPM before the liquid wave reforms on the insertwalls at 2500 RPM. In the 400 µL system even 2500 RPM is insufficient to allowfor the same extent of agitation to be achieved.One caveat to these visual mixing evaluations is that whilst they were per-formed using reaction product in an accurate mock-up, they were done at ambientconditions and not the 445◦C and 13.8 MPa of the reactor. As such, certain con-siderations must be made in their interpretation as discussed in Section 5.2.4.107(a) (b) (c) (d)Figure 4.48: 1500 - 2500 RPM mixing of 150 µL of reaction product obtained for diphenylmethane hydroconversionperformed in the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM as seen in glassmock-up. (a) 1500 RPM. (b) 2000 RPM. (c) 2250 RPM (lowest speed at which solid suspension was observed).(d) 2500 RPM.108(a) (b)Figure 4.49: 2000 - 2500 RPM mixing of 400 µL of reaction product ob-tained for diphenylmethane hydroconversion performed in the glassinsert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0RPM as seen in glass mock-up with central thermocouple. (a) 2000RPM. (b) 2500 RPM.109(a) (b) (c) (d)Figure 4.50: 1500 - 2500 RPM mixing of 150 µL of reaction product obtained for diphenylmethane hydroconversionperformed in the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM as seen in glassmock-up with central thermocouple. (a) 1500 RPM. (b) 2000 RPM. (c) 2250 RPM. (d) 2500 RPM.110Comparison of Liquid Loading VolumesThe necessity of using a reduced volume required a comparative study to be per-formed. A lower liquid loading would change the hydrogen:DPM ratio and po-tentially affect the reaction mechanism. Table 4.15 shows the results from theseexperiments. As may be seen, reducing the liquid volume from 400 to 150 µL re-sults in several key changes to the reaction, most notably increasing the DPM con-version (perhaps an indication of diffusional limitations in this unmixed system),increasing the yield of CHMB and decreasing the yield of isom./cond. products.Table 4.15: Comparison of results for diphenylmethane hydroconversion ob-tained in the glass insert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8MPa H2, 1 h, 0 RPM, 150 - 400 µL liquid to examine the effect of liquidloading volume.Conditions DPM conversion Product yield(wt%) (mol/molDPM reacted)Benzene Toluene CHMB0 ppm Mo, 150 µL 7.7 ± 0.5 0.92 ± 0.02 0.91 ± 0.02 0.03 ± 0.000 ppm Mo, 400 µL 3.3 ± 0.3 0.92 ± 0.03 0.78 ± 0.02 0.00 ± 0.001800 ppm Mo, 150 µL 11.1 ± 1.1 0.83 ± 0.01 0.86 ± 0.01 0.12 ± 0.051800 ppm Mo, 400 µL 5.7 ± 0.2 0.84 ± 0.02 0.89 ± 0.02 0.02 ± 0.03Product yield B:T ratio(g/gDPM reacted) (mol/mol)Cracking Isom./cond.0.00 0.04 ± 0.02 1.02 ± 0.000.00 0.15 ± 0.01 1.12 ± 0.060.00 0.00 0.97 ± 0.050.00 0.08 ± 0.02 0.94 ± 0.01Thermocouple Wall ActivityBefore beginning mixing experiments, a series of tests was conducted to examinethe wall activity of the central thermocouple. This series differed from the reactorwall activation study in Section 4.2.1 by conducting the tests without catalyst (soas to rule out activation by MoS2 deposition). As such, a series of 0 ppm Mo exper-iments were run after installation of a fresh thermocouple to determine the activa-tion of the stainless steel sheath and its influence on the reaction mechanism. TheDPM conversion for these sequential experiments is shown in Figure 4.51 whereinit may be seen that an initially low conversion rises rapidly with each subsequent111test, plateauing at a value comparable with that seen with the aged thermocouple(shown in Table 4.15).Figure 4.52 shows combined benzene and toluene yield data and Figure 4.53the corresponding B:T ratio. It may be noted that without the activity of the ther-mocouple, the system tends to the formation of excess toluene but that the yieldsquickly equalise with the B:T ratio climbing toward unity as the thermocouple wallbecomes more active. The CHMB yield in Figure 4.54 shows a roughly steadyformation of this species with it being noted that its presence (not seen in previ-ous glass insert experiments) being due to the reduced liquid loading as shown inTable 4.15. The isom./cond. lump product yield shown in Figure 4.55 is seen toincrease with wall activity. This fraction comprises only fluorene at lower con-versions but includes BBP as the conversion increases with wall activity (notablythese are the same products observed for the aged thermocouple). The speciespresent and the trends observed suggest that the catalytically active thermocouplewall (predominantly FeS) performs a role comparable to that of the MoS2 catalystbeing investigated, i.e. that it is foremost a hydrogenation catalyst.0 2 4 6 8 100246810DPM conversion (wt%)Sequential experiment number (-)Figure 4.51: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the glass insert micro-reactor at 0 ppmMo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM to study thermocouple wallactivation. Curve is for illustration of trend and is not a kinetic fit.1120 3 6 90.000.250.500.751.001.25Benzene and toluene molar yield(mol/molDPM reacted)DPM conversion (wt%)Figure 4.52: Benzene and toluene molar yield results obtained for diphenyl-methane hydroconversion experiments performed in the glass insertmicro-reactor at 0 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM to studythermocouple wall activation. Curves are for illustration of trends only.• - Toluene. ◦ - Benzene.0 3 6 90.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.53: Benzene: toluene molar ratio results obtained for diphenylmeth-ane hydroconversion experiments performed in the glass insert micro-reactor at 0 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM to study ther-mocouple wall activation. Curve is for illustration of trend only. ◦0 ppm Mo. △ 1800 ppm Mo.1130 3 6 90.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.54: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM tostudy thermocouple wall activation. Curve is for illustration of trendonly.0 3 6 90.00.10.2Other isom. and cond. productsmass yield (g/gDPM reacted)DPM conversion (wt%)Figure 4.55: Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM tostudy thermocouple wall activation. Curve is for illustration of trendonly.114Effect of Mixing SpeedWith the activity of the thermocouple wall stabilised, an investigation into the ef-fect of mixing speed on the reaction could be performed. Figure 4.56 shows theDPM conversion over the range of mixing speeds investigated wherein a clear andunexpected trend may be seen. Increasing the mixing speed beyond 1500 RPM re-sults in a rapid decline in the conversion for both the 0 and 1800 ppm Mo systems.The difference between these two datasets, representing the Mo catalyst activity,appears to show a maximum at 2000 RPM. Whilst the indicated decline at highermixing speeds may be an experimental artifact, for the purposes of this study, 2000RPM was identified as the optimum mixing speed and selected for more detailedstudy. Interestingly the DPM conversion observed for 0 ppm Mo at 2250 RPM isvery similar to those values obtained for the system with an inactive thermocouple(see Figure 4.51).Figures 4.57 and 4.58 show the benzene and toluene yields, respectively. Notethat the data is presented versus the DPM conversion rather than mixer speed asthe non-linearity of conversion with mixing speed would make meaningful com-parisons from such graphs difficult. Whilst the toluene yields appear to follow thesame trend for both 0 and 1800 ppm Mo, the 0 ppm Mo system is noted to exhibithigher benzene yields (and hence higher B:T ratios as seen in Figure 4.59) than its1800 ppm Mo counterpart. 1800 ppm Mo was seen to have higher CHMB yieldsthan 0 ppm Mo and whilst 0 ppm Mo showed no cracking products, 1800 ppm Mowas observed to produce 3-methylheptane. This cracking product only appearedat 1500 RPM with a yield of 2.1 ± 0.2 mol/molDPM reacted. The only isom./cond.product observed was fluorene for 0 ppm Mo at 0 RPM with a yield of 0.04 ± 0.02mol/molDPM reacted.Analysis of Recovered SolidsThe maximum in the observed catalytic activity at 2000 RPM, together with visualevaluation of the product samples changing from fine, suspended particles to large,glitter-like flakes, prompted an investigation into the structure of the recoveredsolids by TEM, SEM and SEM-EDX. Samples selected for analysis covered bothmixing speed and reaction time ranges but were limited to those experiments with1150 1500 2000 225004812DPM conversion (wt%)Mixing speed (RPM)Figure 4.56: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the glass insert micro-reactor at 0 - 1800ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 - 2250 RPM. Curves are for il-lustration of trends and are not kinetic fits. ◦ 0 ppm Mo.△ 1800 ppm Mo. X Difference between 0 and 1800 ppmMo conversion.0 4 8 120.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.57: Benzene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.1160 4 8 120.000.250.500.751.001.25Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.58: Toluene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.0 4 8 120.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.59: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0 RPM. Curves are forillustration of trends only. ◦ 0 ppm Mo. △ 1800 ppmMo.1170 4 8 120.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.60: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 0RPM. Curves are for illustration of trends only. ◦ 0 ppm Mo. △1800 ppm Mo.1181800 ppm Mo (there were no solids from 0 ppm Mo experiments).Figures 4.61 and 4.62 show the TEM results for 0 and 2000 RPM samples, re-spectively. Whilst inter-plate and d-spacing measurements (shown in Figure 4.61b)confirm that the solid material is MoS2, the shape and size of the crystallites is seento change dramatically with conditions. It is clear that the solid material from the0 RPM experiments is almost identical to that from the stirred batch reactor (com-pare with Figures 4.19 and 4.20) in terms of stack height, sheet width and “rag”arrangement, the material present after 2000 RPM has undergone extreme changes.The short stacks of small sheets, observed for all reaction times at 0 RPM and ini-tially at 2000 RPM, appear to agglomerate and fuse with prolonged reaction time at2000 RPM to form large sheets up to 100 nm wide and stacks more than 20 sheetsthick.Analysis by SEM-EDX is shown in Figure 4.63 with compositional results inTable 4.16. It may be seen that whilst the 0 RPM system is comprised almostentirely of S and Mo with a close correlation between these species (confirmedby their almost 1:1 quantification), at 2000 RPM an appreciable amount of Fe isfound in the sample. This Fe, likely present as FeS as seen in Figure 4.17 andby the close correlation in Figure 4.63b, is thought to form by sulphidation of thestainless steel thermocouple wall (as evidence by the Ni in the solid) and sloughoff into the liquid due to the vigorous agitation of the system. Note that as theseEDX analyses were performed without the use of an internal standard, it is difficultto obtain independent quantification results. The compositions are presented forcomparative purposes only.Table 4.16: Results for scanning electron microscopy with energy dispersiveX-ray quantification for solid material obtained from a diphenylmeth-ane hydroconversion performed in the glass insert micro-reactor at 1800ppm Mo, 445◦C, 13.8 MPa H2, 4 h, 0 - 2000 RPM.Mixing speed Composition (wt%)(RPM) S Mo Fe Ni0 51 48 1 02000 53 27 18 2To further examine the structure of the recovered solid material, FESEM wasperformed on solid samples representing a range of mixing speeds and reaction119times as shown in Figures 4.64 through 4.65. The solid material from the 0 RPMexperiment presents as small, divided platelets collected loosely into larger struc-tures. This material does not appear to change with increased reaction time. Mate-rial from reactions with 2000 RPM mixing, however, appear as larger plates fusinginto tight sheets with increased reaction time. These observations compare wellwith TEM data in Figures 4.61 and 4.62. Of interest is that this fusing is apparenteven when comparing the 0 h results shown in Figures 4.64e and 4.65e. This sup-ports the theory that in both mixed and unmixed systems small particles of MoS2precipitate separately during heat-up. In the mixed system, these particles then ag-glomerate and fuse. There is no evidence to support that MoS2 crystallite growthin the mixed system is due to initial MoS2 particles acting as nucleation centres.To study this crystal growth and simultaneously determine the extent to which itoccurs at higher mixing speeds, two 2250 RPM experiments were conducted, theresults presented in Figures 4.66 and 4.67. The mixing in one experiment beganat the start of heating (as per the standard procedure), the other only once reac-tion temperature was achieved. After 1 h reaction time, very little difference isdiscernible between the two samples. To clarify the presentation of the plates andfused structures, one of the 2250 RPM samples was angled during FESEM anal-ysis, the large, smooth sheets of agglomerated and fused crystallites being clearlyvisible in Figure 4.68.Optimum Mixing Speed EvaluationWith the optimum mixing speed selected (2000 RPM providing the maximumDPM conversion difference between 0 and 1800 ppm Mo), the reaction was studiedover a range of reaction times both with and without agitation. The DPM conver-sion results are presented in Figure 4.69 wherein it may be seen that the trends forboth the 0 and 2000 RPM are similar with 1800 ppm Mo showing a rapid increasefollowed by a more steady slope. For 0 ppm Mo, however, a clearly sigmoidalcurvature was apparent. At 1 h the 1800 ppm Mo conversion results for 0 and 2000RPM are almost identical but deviate with longer reaction times. This deviationis noted for all reaction times with 0 ppm Mo. The sigmoidal DPM conversionresults once again resulted in poor kinetic fits (see Figure E.3 and Table E.40).120(a)(b)Figure 4.61: Transmission electron microscopy images for solid material ob-tained from diphenylmethane hydroconversion performed in the glassinsert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0RPM. Images used to confirm MoS2 formation and examine changesin crystallite dimensions (stack height and sheet width) with time inthe absence of without mixing. (a) 0 h. (b) 4 h showing the both theinter-plate spacing (d002 = 0.62 nm) and the d-spacing (d100 = 0.27nm).121(a)(b)Figure 4.62: Transmission electron microscopy images for solid material ob-tained from diphenylmethane hydroconversion performed in the glassinsert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h,2000 RPM. Images used to examine changes in crystallite dimensions(stack height and sheet width) with time when the reactor was mixed.(a) 0 h. (b) 4 h.122(a)(b)Figure 4.63: Scanning electron microscopy with energy dispersive X-ray images for solid material obtained from di-phenylmethane hydroconversion performed in the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPaH2, 4 h, 0 - 2000 RPM. Elemental maps inverted for clarity with dark spots indicating positive detection. (a) 0RPM, 4 h. (b) 2000 RPM, 4 h.123(a) (b)(c) (d)(e) (f)Figure 4.64: Field emission scanning electron microscopy images for solidmaterial obtained from diphenylmethane hydroconversion performedin the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPaH2, 0 - 4 h, 0 RPM. Images used to examine particle agglomerationwith time in the absence of mixing. (a) 4 h. (b) 4 h. (c) 4 h. (d) 4 h.(e) 0 h. (f) 2 h.124(a) (b)(c) (d)(e) (f)Figure 4.65: Field emission scanning electron microscopy images for solidmaterial obtained from a diphenylmethane hydroconversion performedin the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2,0 - 4 h, 2000 RPM. Images used to examine particle agglomerationwith time when the reactor was mixed. (a) 4 h. (b) 4 h. (c) 4 h. (d) 4h. (e) 0 h. (f) 2 h.125(a) (b)(c)Figure 4.66: Field emission scanning electron microscopy images for solidmaterial obtained from diphenylmethane hydroconversion performedin the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2,1 h, 2250 RPM (including during heat-up as per normal procedure).Images used to examine particle formation and agglomeration whenthe reactor was mixed during heat-up and reaction.126(a) (b)(c) (d)(e)Figure 4.67: Field emission scanning electron microscopy images for solidmaterial obtained from diphenylmethane hydroconversion performedin the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2,1 h, 2250 RPM (excluding during heat-up). Images used to examineparticle formation and agglomeration when the reactor was mixed onlyduring reaction and not heat-up.127(a) (b)Figure 4.68: Field emission scanning electron microscopy images for solid material, the sample angled for better inter-pretation of the structure, obtained from diphenylmethane hydroconversion performed in the glass insert micro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 2250 RPM (excluding during heat-up). Images used toexamine the macro-structure of the particle agglomerates.1280 1 2 3 40102030 DPM conversion (wt%)Reaction time (h)0 RPM2000 RPMFigure 4.69: Conversion results obtained for diphenylmethane hydroconver-sion experiments performed in the glass insert micro-reactor at 0 - 1800ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 2000 RPM. Error bars indicatestandard deviation. Curves are for illustration of trends and are notkinetic fits. • 0 ppm Mo, 0 RPM. N 1800 ppm Mo, 0RPM. ◦ 0 ppm Mo, 2000 RPM. △ 1800 ppm Mo, 2000RPM.129The benzene and toluene yields are illustrated in Figures 4.70 and 4.71 withthe corresponding B:T ratio in Figure 4.72. For both 0 and 2000 RPM a trend oflesser benzene yields for 1800 than 0 ppm Mo is observed with the 2000 RPM dataperhaps declining slightly while the 0 RPM results do not. Both catalyst loadingsshow increasing toluene yields with increasing conversion, a trend mirrored in the1800 ppm Mo 2000 RPM system. The mixed 0 ppm Mo system, however, showsa steady decline in toluene yield with increasing DPM conversion. The B:T trendsappear similar over a given conversion range with 1800 ppm Mo falling below 0ppm Mo. It is also noted that the 2000 RPM samples exhibit lower B:T vratiosthan their unmixed counterparts.0 5 10 15 20 250.00.20.40.60.81.0Benzene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.70: Benzene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 - 2000 RPM,150 µL liquid load. Curves are for illustration of trends only. •0 ppm Mo, 0 RPM. N 1800 ppm Mo, 0 RPM. ◦ 0 ppmMo, 2000 RPM. △ 1800 ppm Mo, 2000 RPM.The yield of CHMB is shown in Figure 4.73 from which a decrease in the thisyield with increased mixing may be noted. Whilst mixing appears to suppressCHMB yield for 1800 ppm Mo, the effect is promotional for 0 ppm Mo. If thetrends for 400 and 150 µL comparisons are recalled (see Table 4.15), an increasein the hydrogen:DPM ratio resulted in an increase in CHMB production. It seems1300 5 10 15 20 250.000.250.500.751.001.25Toluene molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.71: Toluene molar yield results obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 - 2000 RPM,150 µL liquid load. Curves are for illustration of trends only. •0 ppm Mo, 0 RPM. N 1800 ppm Mo, 0 RPM. ◦ 0 ppmMo, 2000 RPM. △ 1800 ppm Mo, 2000 RPM.0 5 10 15 20 250.000.250.500.751.001.25Benzene:toluene molar ratio (mol:mol)DPM conversion (wt%)Figure 4.72: Benzene:toluene molar ratio obtained for diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 - 2000 RPM,150 µL liquid load. Curves are for illustration of trends only. •0 ppm Mo, 0 RPM. N 1800 ppm Mo, 0 RPM. ◦ 0 ppmMo, 2000 RPM. △ 1800 ppm Mo, 2000 RPM.131likely that mixing the 0 ppm Mo system allows for the dissolution of more hydro-gen and hence the formation of more CHMB.Other cracking and isom./cond. species were few and limited to specific con-ditions. Only trace cracking species were observed for the 0 ppm Mo experi-ments but 1800 ppm Mo presented a near-constant yield of 3-methylheptane ofapproximately 0.022 mol/molDPM reacted for all reaction times. The yield of otherisom./cond. species is presented in Figure 4.74 from which it may be seen thatwhilst such species appeared in both 0 RPM systems (1, 2 and 4 h for 0 ppm Mobut only 2 and 4 h for 1800 ppm Mo), only higher conversion 0 ppm Mo reactionsshowed any such species when stirred (only observed in 4 h experiments). Thecompositions of these isom./cond. lumps are shown in Table 4.17.0 5 10 15 20 250.00.10.2CHMB molar yield (mol/molDPM reacted)DPM conversion (wt%)Figure 4.73: Cyclohexylmethylbenzene molar yield results obtained for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h,0 - 2000 RPM, 150 µL liquid load. Curves are for illustration of trendsonly. • 0 ppm Mo, 0 RPM. N 1800 ppm Mo, 0 RPM. ◦0 ppm Mo, 2000 RPM. △ 1800 ppm Mo, 2000 RPM.Quantification of the C1 to C4 gaseous products are presented in Table 4.18.Several trends may be noted from this data:• Increasing reaction time increases the yield of all gaseous products, espe-cially the larger species,1320 5 10 15 20 250.000.020.040.06Other isom. and cond. productsmass yield (g/gDPM reacted)DPM conversion (wt%)Figure 4.74: Mass yield of isomerisation and condensation products for di-phenylmethane hydroconversion experiments performed in the glassinsert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4h, 0 - 2000 RPM, 150 µL liquid load. Curve is for illustration of trendonly. • 0 ppm Mo, 0 RPM. N - 1800 ppm Mo, 0 RPM. ◦ - 0ppm Mo, 2000 RPM.• Increasing the catalyst concentration increases the yield of all gaseous prod-ucts, especially the larger species,• Mixing decreases the yield of all gaseous products, especially the smallerspecies.Spent Residue Hydroconversion Catalyst EvaluationThe final series of experiments involved evaluation of deactivated catalysts fromresidue hydroprocessing reactions. Three samples were tested, all consisting ofcoke-catalyst agglomerate recovered from residue hydroprocessing reactions per-formed by Rezaei et al. [32] in the same stirred reactor as this study but operated insemi-batch mode at 445◦C, 13.8 MPa H2 (flowing at 900 sccm), 1 h, 700 RPM us-ing Cold Lake vacuum residue as the feed. In these reactions molybdenum chloridein reversed micelles was used as the MoS2 precursor, with the resulting catalystshown to possess near identical properties and functionality to that derived from133Table 4.17: Major constituents observed in the isom./cond. product lumpobtained during diphenylmethane hydroconversion experiments per-formed in the glass insert micro-reactor at 0 - 1800 ppm Mo, 445◦C,13.8 MPa H2, 0 - 4 h, 0 - 2000 RPM, 150 µL liquid load (usingacronyms from Table 4.9).Catalyst loading Mixing speed Reaction time Species Composition 1(ppm Mo) (RPM) (h) (area%)001 Fluorene 1002 Fluorene 51ETB 494 Fluorene 39HexF 39ETB 222000 4 HexF 69Fluorene 311800 02 HexF 59Fluorene 414 HexF 60Fluorene 401- Composition indicated is the relative contribution of each species to the isom./cond.lump based on the percentage area from the GCMS chromatogram on a DPM-free basis(i.e. percentage of products formed).134Table 4.18: Major gaseous products observed during diphenylmethane hydroconversion experiments performed in theglass insert micro-reactor at 0 - 1800 ppm Mo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 - 2000 RPM, 150 µL liquid load.Catalyst loading Mixing speed Reaction time Composition (wt%) ×103(ppm Mo) (RPM) (h) Methane Ethane Propane iso-Butane Butane00 1 23.5 2.3 1.4 0.0 0.74 37.2 3.5 5.0 0.4 2.52000 1 3.5 0.4 0.3 0.0 0.24 16.8 2.1 2.9 0.0 0.018000 1 37.9 7.3 5.9 0.5 4.84 47.6 9.8 12.8 1.3 9.42000 1 18.3 6.2 3.9 0.4 3.64 14.7 3.5 3.9 0.5 3.3135Mo octoate. The recovered coke-catalyst agglomerate was washed with toluene,ground and dried prior to use. The three samples tested in this investigation arepresented below, introduced to ensure 1800 ppm Mo was present in all reactions.Each was evaluated at 445◦C, 13.8 MPa H2, 2000 RPM for 1 h with a 150 µLliquid load of undiluted DPM, these conditions being selected to obtain the bestresolution of catalyst performance (the greatest difference between thermal andcatalytic systems).• “Fresh” coke-catalyst recovered after only a single residue hydroconversionexperiment,• Deactivated coke-catalyst recovered after five residue hydroconversion re-covery and reuse cycles,• “Fresh” coke catalyst thermally aged under 100 sccm He at 700◦C for 15 h[1].Table 4.19 shows the results from these experiments together with coke yieldand hydrogen conversion data (used to evaluate catalyst performance) from theiruse in residue hydroconversion reactions for comparison. In the context of residuehydroconversion, wherein a 0 ppm Mo experiment exhibits a coke yield of 21 wt%[32], both the heat treated and fifth recycle catalysts are considered deactivateddue to their elevated coke yields as compared with the fresh specimen. Hydrogenconversion is also seen to decline, although the change is not as dramatic, with thefresh catalyst exhibiting a greater hydrogen conversion than either the heat treatedor recycled catalyst experiments (a 0 ppm Mo experiment yielding a hydrogenconversion of 13 % [32]). Evaluation in DPM hydroconversion, however, suggeststhe heat treated catalyst to be the most active (its DPM conversion and productyields on par with 1800 ppm Mo introduced as Mo octoate) with both the freshcoke-catalyst and the fifth recycled species showing poor performance in terms ofDPM conversion and a clearly change to the reaction mechanism per the productyields. Of interest was that only trace amounts of other cracking or isom./cond.species were present in the liquid products. For additional detail, the gas productanalyses for these DPM hydroconversion tests are provided in Table 4.20. It is clearthat whilst the fresh catalyst (both heat treated and not) show gas product yields136on par with previous 1800 ppm Mo experiments, the DPM hydroconversion withthe fifth recycle coke-catalyst agglomerate results in the formation of significantlymore gaseous products (although combined these still only account for less than 1wt% of the gas).137Table 4.19: Results observed during diphenylmethane hydroconversion experiments performed in the glass insertmicro-reactor at 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 2000 RPM, 150 µL liquid load using various coke-catalyst agglomerates from residue hydroconversion experiments.Coke-catalyst Residue hydroconversion DPM hydroconversiontype 1 Coke yield 2 H2 conversion DPM conversion Product yield (mol/molDPM reacted) B:T ratio(wt%) (%) (wt%) Benzene Toluene CHMB (mol/mol)Fresh 2.9 18 0.5 0.65 1.28 0.0 0.51Fresh, heat treated 3 11.5 15 11.1 0.74 0.81 0.21 0.92Recycled 11.5 14 1.9 0.19 1.36 0.10 0.141- Coke-catalyst agglomerate samples courtesy of Rezaei and Smith [1]. 2 - Coke yield expressed as wt% of total product. 3 - “Fresh”catalyst heat treated under 100 sccm He at 700◦C for 15 h.138Table 4.20: Major gaseous products observed during diphenylmethane hy-droconversion experiments performed in the glass insert micro-reactorat 1800 ppm Mo, 445◦C, 13.8 MPa H2, 1 h, 2000 RPM, 150 µL liquidload using various coke-catalyst agglomerates from residue hydrocon-version experiments.Catalyst type 1 Composition (wt%) ×103(ppm Mo) Methane Ethane Propane iso-Butane ButaneFresh 13.2 1.9 1.3 0.0 1.6Fresh, heat treated 3 15.7 1.5 1.1 0.0 1.1Recycled 279 106 80 6 511- Coke-catalyst agglomerate samples courtesy of Rezaei and Smith [1]. 2 - “Fresh”catalyst heat treated under 100 sccm He at 700◦C for 15 h.139Chapter 5Discussion of ExperimentalResultsPresented below is an in-depth discussion of the results obtained during this study,their interpretation and implications. The results of additional experimental tests(not included in the main program of Table 3.1), data processing, thermodynamicanalyses, etc. are presented, as needed, to provide additional detail. This discussionis organised according to the “Phases” of the work presented in Section 3.1.2.5.1 Model Compound Evaluation5.1.1 Model Compound ScreeningThe conversion and product distribution results for the screening of the three modelcompounds selected for evaluation in this study (diphenylmethane [DPM], diphenyl-ethane [DPE] and diphenylpropane [DPP]), diluted to 3 wt% in decahydronaphtha-lene (decalin), are presented in Section 4.1.1. It was determined that both DPE andDPP are extremely reactive under the operating conditions of 445◦C and 13.8 MPaH2, achieving near complete conversion at both 0 and 600 ppm Mo (see Table 4.1).DPM showed lower conversions in the range of 30 - 40 wt%, making it more suit-able for continued mechanistic studies in this work.Unfortunately, only DPE and DPP would allow for additional information to be140gathered by studying the secondary cracking of their alkyl branches. In an attemptto compromise between the desired operating temperature of 445◦C and a lowerconversion, DPP was studied at temperatures as low as 420◦C (Figure 4.1). Thistemperature reduction saw a conversion decrease of only approximately 3 wt%.The product distributions from all three model compounds (presented in Table 4.2)were greatly informative. Many of the expected species were present: toluenefrom DPM, and toluene and ethylbenzene from DPE and DPP. Of note is the vari-ety of species formed, with Table 4.2 showing only those present as >5 area% inthe GCMS analyses. This suggests a complex reaction network, with the primarycracking radicals shown in Figure 3.1 undergoing significant continued cracking,isomerisation and condensation reactions. As shown in Figure 2.6 for DPM, whetherby the mechanisms proposed by Curran et al. [78], Wei et al. [46] or LaMarca et al.[88], the addition of the MoS2 catalyst should serve to simplify the reaction productby providing lower energy pathways to stable species. What is observed, however,is that the addition of 600 ppm Mo increases product diversity and shifts the distri-bution toward more hydrogenated and cracked species (Table 4.2). Unfortunatelythe complexity of these product distributions makes clarification of the mechanismsalmost impossible.From the DPM product compositions in Table 4.2, it is clear that the additionof MoS2 results in the formation of benzene, C2 substituents on rings and cyclo-hexylmethylbenzene (CHMB, being hydrogenated DPM). The benzene appears tosupport the mechanism of Curran et al. [78] (see Figure 2.6a) whereby the primarythermolysis radicals are rapidly stabilised. This theory does not, however, explainthe increase in product diversity. The variety of product species and the C2 alkylbranches support the mechanism of LaMarca et al. [88] (see Figure 2.6c) wherebyhydrocarbon radicals initiate cracking but this does not support the formation ofbenzene. The presence of hydrogenated rings and CHMB suggest that the catalysthas a strong hydrogenation role in the reaction which may be key to understand-ing the system. Even with this simple model compound it is apparent why somany contrasting mechanisms have been proposed and opposed in the literature[21, 78, 88] (see Section 2.5.2 for details). Comparing the conversion and productdistribution results for DPM, DPE and DPP it is extremely difficult to identify therole of the catalyst.141The DPP product compositions in Table 4.2 also afford some understandingregarding the thermal requirements of the reaction. It appears that below 445◦C,certainly below 430◦C, the rate of thermal decomposition has slowed with only theprimary cracking products, ethylbenzene and toluene, being observed. At thesetemperatures it appears that the catalyst has reduced activity with none of the char-acteristic hydrogenation species being present. This seems counter-intuitive as thelower energy barrier for catalytic reactions should allow these to dominate at lowertemperatures. This trend is due to the susceptibility of the DPP alkyl bridge tothermolysis, achieving >96% conversion even at 420◦C.5.1.2 Diphenylmethane StudiesDiluted DiphenylmethaneWith DPM selected as the most appropriate of the three model compounds tested inthis study, given its mid-range conversion under reaction conditions and simplifiedproduct spectrum, a more in-depth analysis was performed. These experimentswere conducted at the same conditions as the screening experiments but with vary-ing reaction time, the DPM conversion and product yields being studied in detailto determine reaction mechanisms.One helpful piece of information when analysing the decomposition of DPM,particularly when using the mechanisms proposed in literature [21, 29, 46, 78, 88](see Figure 2.6) as a starting point, is the stability of the major species, namelythe products (benzene and toluene) and the diluent (decalin), under reaction condi-tions. This information is provided in Tables 4.4 through 4.7. It is clear that bothbenzene and toluene, once formed, react under the given conditions to form varioushydrogenation, cracking and isomerisation products as shown in Figure 5.1. Theconversions of benzene and toluene are comparable, as are the products formed,but both these and decalin are less reactive than DPM. Whilst toluene hydrogena-tion to methylcyclohexane, isomerisation to ethylcyclopentane or cracking to 3-methylheptane or benzene all seem to be trivial mechanisms, the presence of thesesame product species in almost the same ratios in the benzene blank unexpectedlysuggests a very similar mechanism (one which is able to produce C7 species from142a C6 reagent). The decalin blank shows different product species in minimal quan-tities and its decomposition is thus not responsible for the C7 species observed.It seems clear that this is a process of hydrogenation with minimal dehydro-genation (all products from benzene decomposition being saturated species) andthat cracking, recombination and isomerisation are involved (benzene forms C7species and toluene forms C6 species). It is theorised that toluene decomposes tobenzene which is subsequently hydrogenated to cyclohexane. This cyclohexanerapidly cracks and recombines to form C7 alkanes which isomerise and cycliciseto the observed products (cyclicisation perhaps occurring during the abstraction ofH*). The equivalent conversions of benzene and toluene decomposition and thepresence of benzene in the toluene product suggest that the removal of C1 fromtoluene occurs faster than the hydrogenation of benzene. Thermodynamic simu-lations, presented in Figure 5.2, show the Gibbs free energy of reaction (∆Gr) oftoluene to methane (an example for illustration) to be lower than that of benzene tomethane, the former reaction hence occurring more readily than the latter.An alternative explanation would be that both benzene and toluene undergohydrogenation, cracking, recombination and isomerisation with the formation ofbenzene from toluene being a separate process.Uncertainty in the exact mechanism aside, it appears that under reaction condi-tions, both benzene and toluene decompose at roughly equivalent rates but toluenedoes so to form benzene as a product. This would result in higher yields of benzene(i.e., higher benzene:toluene or B:T molar ratios) than may otherwise be expected.Returning to the DPM conversion data shown in Figure 4.2, it may be seenthat both 0 and 600 ppm Mo experiments reach reaction temperature with a DPMconversion of approximately 30 wt%. This non-zero conversion at the start ofthe experiment is due to the slow heat-up of the reactor (as shown in Figure 3.3)which affords the thermal reaction an opportunity to begin prior to reaching oper-ating temperature. The closeness in conversion between 0 and 600 ppm Mo uponreaching 445◦C suggests that minimal catalyst activity occurs during heat-up. Thistrend is supported by the DPP temperature data in Table 4.2 which showed a cleardecline in catalytic products when the reaction temperature was lowered.The conversion results show clear sigmoidal trends. Common in enzymaticreactions (in Michaelis-Menten kinetics for instance) [112], sigmoidal curvature is143methylcyclohexane48 area% 3-methylheptane36 area%ethylcyclopentane15 area%(a)methylcyclohexane41 area% 3-methylheptane31 area%benzene19 area%ethylcyclopentane9 area%(b)Figure 5.1: Major products from the thermocatalytic decomposition of ben-zene and toluene performed in the stirred batch reactor at 445◦C, 13.8MPa H2, 1 h, 600 ppm Mo, 700 RPM, 3 wt% in decalin. Compositionsindicate GCMS area percentage as shown in Tables 4.5 and 4.6. (a)Benzene. (b) Toluene.+9 H210 H2+ 6 CH47 CH4∆Gf718K = -338 kJ/molBenzene∆Gf718K = -405 kJ/molTolueneFigure 5.2: Thermodynamic simulations of benzene and toluene decompo-sition to methane simulated in Accelrys Materials Studio (v4.4) usingDMol3 [110, 111] geometry optimisation and frequency calculations.Simulation details and parameters are provided in Section F.4.often indicative of cooperative systems. This may occur on the catalyst, wherebyconversion at an active site is promoted by the adsorption of species on neighbour-ing sites, or in the reaction fluid, whereby the formation of specific species maypromote the conversion of others. Such sigmoidal trends do not correspond with“standard” first or second order kinetic models, and attempts to fit such models144were unsuccessful (see Section E.1.2). Similarly Langmuir-Hinshelwood-Hougen-Watson kinetics were found to be unsuitable as changes in the conversion trendswould suggest changes in the kinetic expression for the same catalyst, a conclusionwhich is contrary to the assumptions of this mechanism.Either or both of these cooperative effects may be at play in this system. Onthe catalyst surface, for example, dissolved H2 may need to adsorb and react in suf-ficient quantities before DPM conversion can proceed at an appreciable rate (thisapplies to both the 0 and 600 ppm Mo systems as even 0 ppm Mo has access to thecatalytic reactor walls). In the liquid phase, precursors may be necessary to initiatecracking reactions (as per the mechanisms for hydrogen and hydrocarbon radicalsproposed by Curran et al. [78] and LaMarca et al. [88] in Figures 2.6b and 2.6crespectively) for a similar effect. Simulations (shown in Section F.7) performedin AspenTech Aspen Plus (v7.3) indicated H2 solubility in decalin to be approxi-mately 25 mol% (0.4 wt%) and in DPM to be approximately 20 mol% (0.3 wt%)under reaction conditions (note that whilst the Peng-Robinson equation-of-stateproperty method used in this simulation is not always accurate for the estimationof H2 solubilities in hydrocarbons, the results correspond well with experimentalliterature values [113–116] and are thus used for illustrative purposes). Given thishigh concentration of dissolved H2, its adsorption and reaction on the catalyst isunlikely to be responsible for the initial lag observed in the DPM conversion. Itseems more probable that some hydrocarbon species must be responsible, the con-centration being required to increase either on the catalyst surface or in the liquidto attain the observed increase in reaction rates. Given the increase in conversionwith 600 ppm Mo, it is apparent that whatever the nature of this species and itsmechanism, the catalyst plays a role in its formation and/or subsequent reaction.An additional point of interest in the DPM conversion results of Figure 4.2 isthe plateauing of both the 0 and 600 ppm Mo conversion results after extendedreaction times. This may be due simply to reduced DPM concentrations for the600 ppm Mo, but a similar trend for 0 ppm Mo is observed at a lower conversion.With H2 in vast excess and thus its limitation unlikely, this leveling off may indicateconsumption of the rate-enhancing species either on the catalyst surface or in theliquid.The presence of decalin may also complicate the analysis. Hydrogenated species145have been shown to act as H shuttles (discussed in Section 2.3.2) and the decalinsolvent may thus aid in the hydrogenation (and subsequent cracking and/or iso-merisation) of the various species present. Following donation of H by decalin,re-hydrogenation may be promoted by the catalyst, hence the increase in rate with600 ppm Mo. With the great excess of decalin in these diluted experiments, thiscontribution does not help to explain the leveling off observed in the DPM conver-sion.To better understand the mechanism, product yields must be examined. Giventhe great variety in other cracking, isomerisation and condensation products ob-served, product analyses for these diluted DPM experiments was limited to ben-zene and toluene, the major anticipated products. Together with these yields, thebenzene:toluene (B:T) molar ratio served as a measure of continued reaction of thebenzyl and phenyl radicals as per the mechanism of Figure 2.6a.Despite DPM conversions for 0 and 600 ppm Mo being the same upon reachingreaction temperature, their products are not (as shown in the product yield and ratioresults in Figures 4.3 through 4.5). The 0 ppm Mo system exhibits greater yieldsof both benzene and toluene than does the 600 ppm Mo reaction with both serieshaving B:T molar ratios of <1:1.An effective catalyst should either produce sufficient H* to rapidly stabilisethe benzyl and phenyl radicals with an ideal ratio of 1:1 [5, 29, 31, 46, 68, 75–78] or perform catalytic hydrogenolysis to directly produce benzene and toluene inthis ratio. As mentioned above, however, toluene reacts slightly more readily thanbenzene (and even forms benzene as it decomposes), resulting in a higher thanexpected B:T ratio. If we examine the benzyl and phenyl radicals, however, it isnoted (as per the Gibbs free energy results shown in Figure 5.3) that phenyl radi-cals decompose to shorter radicals (methyl radicals in this example) more readilythan do benzyl radicals (despite neither being spontaneous). This helps explain thetrends observed. With 0 ppm Mo, the phenyl radicals decompose and the benzylradicals remain to be stabilised to toluene, resulting in a lower B:T ratio. Increas-ing the catalyst concentration increases the rate of stabilisation and hence reducesthe extent of phenyl radical decomposition, raising the B:T ratio.Whilst explaining the B:T ratios observed, this theory does not address somekey trends.146• The 0 ppm Mo B:T ratio increases with increasing DPM conversion to alevel comparable with the 600 ppm Mo reaction.• The benzene and toluene yields decrease with increasing catalyst concentra-tion (these should increase if the catalyst were promoting the stabilisation ofthe benzyl and phenyl radicals before they could decompose).• The cracking and isomerisation product yields increase with increasing cata-lyst concentration per Table 4.2 (these should decrease by the same argumentas above).Although these few experiments do not provide enough data for a definitivemechanism to be presented, several conclusions may be drawn. The catalyst servesa hydrogenation role and, in doing so, promotes the formation of various hydro-genation, cracking and isomerisation products, thereby reducing the yield of ben-zene and toluene. As the concentration of these products increases, DPM con-version increases (perhaps per the mechanism proposed by LaMarca et al. [88] asshown in Figure 2.6c) and the rate of stabilisation of benzyl and phenyl radicalsincreases (for instance by radical addition or radical hydrogen transfer), increasingthe yields of benzene and toluene and raising the B:T ratio.To clarify the DPM reaction mechanism, it is necessary to simplify the systemby eliminating the H shuttle, decalin.+6.5 H27 H2+ 6 CH37 CH3∆Gr718K = 548 kJ/molPhenyl∆Gr718K = 750 kJ/molBenzylFigure 5.3: Thermodynamic simulations of phenyl and benzyl radical de-composition to methyl radicals simulated in Accelrys Materials Studio(v4.4) using DMol3 [110, 111] geometry optimisation and frequencycalculations.Simulation details and parameters are provided in Section F.4.147Undiluted DiphenylmethaneFigure 4.6 presents the conversion data for the undiluted DPM experiments. Apartfrom omitting the decalin diluent, the maximum reaction time was reduced to 6h (a point where the plateau observed previously was well established) and 1800ppm Mo catalyst loadings were included. Whilst the DPM maximum conversionwas reduced to approximately 40 wt% at the reduced reaction time of 6 h, it shouldbe noted that the conversion range under reaction conditions is equivalent to thatof the diluted experiments. The undiluted DPM experiments reached reaction tem-perature with a conversion <5 wt% whilst the diluted DPM experiments, withmaximum DPM conversions of approximately 80 wt% after 8 h, reach reactiontemperature with approximately 30 wt% DPM conversion. The sigmoidal trendsobserved in the diluted system are still obvious but with a major change, starting atroughly the same DPM conversion upon reaching reaction temperature, the 0 and600 ppm Mo systems diverge with 600 ppm Mo exceeding 0 ppm Mo after 1 hbefore re-converging to present comparable conversions after 6 h. 1800 ppm Moexhibits the highest DPM conversion after 1 h. All 1800 ppm Mo experiments wereconducted for 1 h but at different reaction temperatures to determine the impact oftemperature on catalyst activity (as evidenced by changing product compositions,analogous to the DPP experiments discussed in Section 5.1.1). Unlike the dilutedexperiments, a first order kinetic fit now appears suitable for modeling of the 0ppm Mo data (suggesting fewer other species affecting the reaction and perhaps a“cleaner” thermal product) whilst neither first nor second order models were foundto accurately approximate the 600 ppm Mo results (the fits presented, for complete-ness, in Section E.1.3).XRD and TEM data (Figures 4.17 through 4.20) confirmed the formation ofthe desired MoS2 active phase. The average crystallite size was found to be lessthan 4 nm across (see the particle size distribution in Figure 4.21) and stackedto a thickness of less than three sheets (the stack height distribution provided inFigure 4.22). These narrow stacks would exhibit a high dispersion (the proportionof the active rim-edge atoms versus total atoms). Of interest in the XRD data arethe FeS and C peaks present in the 600 and 1800 ppm Mo samples respectively.The FeS, which is also catalytically active in hydroconversion reactions [4, 8],148likely forms on the reactor walls and internals in the high temperature, sulphur-rich environment of the reaction and sloughs off into the liquid over time. Thiscontaminant is certainly present in the 1800 ppm Mo sample too but given thatthe rate of FeS formation and sloughing is surely equivalent regardless of MoS2loading (a great excess of CS2 being added to each reaction), the relative amountof FeS in the 1800 ppm Mo XRD sample would be less significant. Another pointof interest are the graphite peaks in the 1800 ppm Mo sample. Although small,these suggest the condensation of large polycyclic aromatic species, the formationof which from DPM would require extensive radical recombination.Returning to the DPM conversion data (Figure 4.6), another explanation for thechange in the observed trend may be H starvation. In the diluted system, with 3wt% model compound, the H2:DPM molar ratio was approximately 26 mol/mol(Table 3.2). This, together with the H-donating decalin solvent, would make H2supply a minor factor in the overall reaction. In the undiluted system, however,the volume of the reactor simply limits the amount of H2 present at the start ofthe reaction with a H2:DPM molar ratio of only approximately 0.8 mol/mol. Thismakes 100% conversion of DPM to benzene and toluene impossible as insufficientH2 is present for the reaction. This limiting factor may be responsible for the 600ppm Mo DPM conversion leveling at the same point as the 0 ppm Mo.The lower conversions observed for 0 and 600 ppm Mo at 0 h in the undi-luted DPM reactions, as compared to their diluted counterparts, may be due tothe absence of decalin. From the diluted studies it was clear that DPM conver-sion resulted in high yields of hydrogenated products (which subsequently crackedand isomerised). Decalin, acting as a H shuttle [11, 21], would have promotedsuch hydrogenation reactions, allowing for thermolysis to occur readily even atlower temperatures. Without this H shuttle, the reaction is forced to hydrogenatespecies directly from dissolved H2 or await higher temperatures for thermolysis ofthe DPM and/or catalytic influence. To study this effect, the 1800 ppm Mo ex-periments were conducted at temperatures down to 415◦C as shown in Figure 4.7.While the true interest in this series was to compare product distributions withsimilar undiluted DPM experiments, an interesting segue is a comparison with theDPP temperature data. In such a comparison, the DPM conversion was observedto be far more temperature dependent with DPP conversion almost unchanged at149430◦C whilst DPM showed a loss of almost 30%. This is due to two effects: theDPP is more susceptible to thermolysis due to its three-member alkyl linkage, andthe decalin acted as a H shuttle, promoting hydrogenation and hence conversion.Examining the products formed during these undiluted DPM hydroconversionexperiments (Table 4.8) it is noted that the dominant products upon reaching reac-tion temperature are benzene and toluene. These species are present in roughly thesame proportions for 0 and 600 ppm Mo and would suggest minimal catalyst ac-tivity below 445◦C but for the 1800 ppm Mo data. The data for this higher catalystloading at lower temperatures shows clear evidence of hydrogenation and crackingproducts not seen in the other experiments and suggests that the catalyst, while pro-ceeding at a slower rate at the lower temperatures, is still active. These reactions arealso noted to be more selective with between two and six species present as >1%of the GCMS product area, likely due to decalin not participating as a H shuttlewith the feasible reaction mechanisms thus limited. Of interest is that fluorene ispresent in both the 0 and 600 ppm Mo samples and CHMB and 3-methylheptaneare observed in the 1800 ppm Mo samples (recalling that 3-methylheptane was aproduct of both benzene and toluene decomposition as shown in Figure 5.1).The product yields for the undiluted DPM experiments are presented in Fig-ures 4.9 through 4.15. For benzene and toluene the high yields (correspondingwith these being major products) are noted as is the maximum through which theypass at a DPM conversion of approximately 30 - 40 wt%. The benzene yield isseen to be suppressed at higher catalyst concentrations while the toluene yield ispromoted. The B:T molar ratio shows an apparent relationship with catalyst con-centration, declining as the amount of MoS2 increases, falling below 1:1 at 1800ppm Mo. These trends suggest the catalyst to be preferentially inhibiting or con-suming benzene and forming toluene. These observations are inconsistent withthe hydrogen activation mechanisms presented in Figures 2.6a and 2.6b [46, 78].By these mechanisms, an increase in the catalyst concentration should affect anincrease in the H* concentration. Per the mechanism of Curran et al. [78] ( 2.6a)this increased H* concentration would increase the rate of benzyl and phenyl radi-cal capping, reducing the disparity in benzene and toluene yield with the B:T ratiotending toward 1:1. By the mechanism of Wei et al. [46] ( 2.6b), a higher H* con-centration should promote benzene formation (the benzyl radical being susceptible150to continued cracking), resulting in an increase in the B:T ratio with increased cata-lyst concentration. These results do, however, appear to corroborate the mechanismof LaMarca et al. [88] (Figure 2.6c). With this mechanism requiring a supply ofhydrocarbon radicals and with the catalyst clearly promoting DPM consumption, itis logical that the catalyst promotes the formation of the required hydrocarbon rad-icals. Unfortunately, the product distributions do not support this. Both 0 and 600ppm Mo experiments show appreciable quantities of methyl- and ethyl-substitutedDPM (the MBP and ETB species), but none of the required ipso-substituted speciesrequired for separation of the two rings. Despite these ipso-substituted species be-ing formed as radicals (and hence susceptible to rapid cracking), some would beexpected to stabilise and be observed in the product. Whilst this mechanism mayplay a role in the 0 and 600 ppm Mo experiments, the lack of either methyl- orethyl-substituted species in any of the 1800 ppm Mo systems suggests that it isnot promoted by the catalyst. The mechanism further fails to explain the pres-ence of fluorene and/or hexahydroflourene (hexF) in many of the experiments. Thepresence of the numerous short-chain substituents does, however, suggest that hy-drocarbon radicals are formed during the reaction and are added to the DPM rings.The apparent change in the reaction mechanism at a DPM conversion of 30- 40 wt% is proposed to be due to the formation of a supercritical phase. Underreaction conditions, DPM is a liquid but benzene and toluene are not. Data fromAfeefy et al. [117] indicate the critical temperature and pressure of DPM to beapproximately 500◦C and 2.8 MPa, benzene to be 290◦C and 4.8 MPa and tolu-ene to be 320◦C and 4.1 MPa. Figure 5.4 shows the proportion of each species inthe vapour and the density of both the vapour and liquid phases with changes inDPM conversion. As DPM conversion increases, the density of the liquid phasedeclines (benzene and toluene being less dense than DPM) whilst that of the gasphase increases (benzene, toluene and DPM vapour being more dense than H2).Simulations indicate that at a DPM conversion of approximately 40 wt% the reac-tion mixture becomes a supercritical fluid. This promotes gas-liquid mass transfer(the H2 and hydrocarbons forming a single phase) but reduces the effectiveness ofthe catalyst. Formation of the supercritical phase increases the reaction volume andhence decreases the overall catalyst concentration. Furthermore, a reduction in thedensity of the fluid would result in less effective suspension of the solid catalyst151particles for a given mixer speed.The influence of these phase changes and their mass transfer effects on the re-action may be clearly seen in the benzene, toluene and isom./cond. product yieldsin Figures 4.9, 4.10 and 4.13 respectively. Both benzene and toluene yields declineupon formation of the supercritical phase whilst the yield of isom./cond. productsincreases, all three returning to levels similar to those observed at low DPM con-versions. This is theorised to be due to the inability of the catalyst to effectivelyperform its role when the supercritical phase is formed (due to reduced solid-fluidcontact). Under these conditions, stabilisation of the benzyl and phenyl radicals(to toluene and benzene) formed by DPM decomposition occurs more slowly, withthese radicals undergoing continued reaction to isom./cond. products as seen. For-mation of the supercritical phase is not believed to play a role in the plateauingof the DPM conversion results seen in Figure 4.6. The improved H2/hydrocarbonmixing is instead likely to promote DPM conversion as shown in the benzene dilu-ent experiments presented in Table 4.1 (where operation under supercritical condi-tions was seen to promote DPM conversion by approximately 10%).To understand the mechanism, the yields of the lumped cracking and isom./cond.products must be considered together with that of a species prominent in the 1800ppm Mo results, cyclohexylmethylbenzene (CHMB). One observation is the strik-ing similarity between the yields for 0 and 600 ppm Mo, a trend most likely dueto the activity of the reactor walls and internals. This complication aside, it isclear that whilst the catalyst suppresses the formation of isom./cond. products,consistent with the promotion of radical capping before they can react and recom-bine with one another, it promotes the formation of cracked products, inconsistentwith such a theory. The catalyst was also seen to strongly promote the formationof CHMB, with the yield for 1800 ppm Mo greatly exceeding that of both 0 and600 ppm Mo. Interestingly, the yields of cracking, isom./cond. and CHMB prod-ucts decline with increasing conversion (while remaining in the liquid phase) asthe yields of benzene and toluene rise. This suggests the initial reactions to besomewhat uncontrolled, proceeding by various slow mechanisms, and yet formingspecies which focus the mechanism to one which exacts the rapid production ofbenzene and toluene. This theory is supported by the sigmoidal conversion trends.An interesting theory to consider revolves around the identity of the isom./cond.1520 10 20 30 400.10.20.30255075Fluid density (g.cm-3)DPM conversion (wt%)BenzeneLiquidVapourSupercritical  boundaryProportion invapour phase (%)TolueneDPMFigure 5.4: Simulated separation of reaction species to the vapour phase (theproportion of the total species in the system reporting to the vapour)and corresponding liquid and vapour densities with changing diphenyl-methane conversion assuming equimolar benzene:toluene product. Per-formed in AspenTech Aspen Plus (v7.3).Simulation details and parameters are provided in Section F.5.species, namely fluorene and hexF. As shown in Figure 5.5, fluorene may be formedby the abstraction of H* from DPM, closing of the inter-ring bond (which sta-bilises the radical onto a tertiary carbon) and re-aromatisation by abstraction of asecond H*. HexF may be formed in a similar manner from CHMB with the ini-tial H* being abstracted from the saturated ring. It is also possible that fluorene,once formed, may be hydrogenated to hexF. Thus, while only small quantities ofCHMB were detected in the 600 ppm Mo experiments, the presence of apprecia-ble quantities of hexF betrays its participation in the reaction and suggests it to beextremely reactive. In the 0 and 600 ppm Mo systems, the radicals resulting fromthe abstraction of H* from DPM or CHMB may be stabilised by radical additionwith hydrocarbon radicals present in the reaction mixture resulting in the methyl-and ethyl-substituted species observed. The 1800 ppm Mo system proceeds via153an additional step for, with ample catalyst, the DPM is hydrogenated to CHMB.H* may be more easily abstracted from the saturated ring of CHMB than from theDPM, minimising H* abstraction from the DPM and hence reducing the forma-tion of fluorene. Rapid re-hydrogenation, by the catalyst, of the CHMB radicalformed by such a donation would also limit hexF formation (overall, hexF forma-tion requires two H* be abstracted and its formation may hence be interrupted byre-hydrogenation of the intermediates). It is unlikely in the 1800 ppm Mo systemsthat H* would be abstracted from the DPM before regaining it from such speciesformed by the catalyst by the following logic. To regain H* before stabilisationby a hydrocarbon radical (to form MBP for instance which was not observed inthe 1800 ppm Mo product) would suggest a very high concentration of such H*species on the catalyst surface or in the liquid. It would be easier for moleculesrequiring H* to react with those already formed by the catalyst rather than abstractthem from DPM. Stabilisation of radicals by H* was shown in Section 5.1.2 to beunlikely (through an examination of benzene and toluene yields and ratios). Theabove proposed reactions were supported by thermodynamic simulations determin-ing the Gibbs free energies of reaction as presented in Section F.4.DPMCHMBFluoreneHexF+ 3H2- H- HH- HH- H+ 3H2Figure 5.5: Proposed mechanism for the formation of fluorene and hexahy-drofluorene from diphenylmethane.One possible reason for the abstraction of H* from DPM and CHMB, and thesubsequent self-stabilisation of these molecules to fluorene and hexF respectively,may be the lack of H2 in the system. As discussed above, these undiluted DPM ex-periments could not achieve complete conversion due to H starvation. Examiningthe pressure measurement data, however, suggests that the 1800 ppm Mo systemconsumes more H more rapidly than either the 0 or 600 ppm Mo experiments. Thisnot only supports the observation of the CHMB hydrogenation product but may in-154dicate that the systems do not reach a point where they would be influenced by Hstarvation (the rate of H2 transfer in the 1800 ppm Mo case is not being matchedin the 0 and 600 ppm Mo reactions, indicating their rates of consumption in theliquid, and the subsequent concentration gradients for gas-liquid diffusion, are notas high and are not limiting the reaction).Based on these results, a new mechanism for the thermocatalytic hydroconver-sion of DPM may be proposed as shown in Figure 5.6. Here it is noted that thecatalyst, rather than activating hydrogen as per the mechanisms of Figures 2.6aand 2.6b [46, 78], appears to hydrogenate the DPM to CHMB. This CHMB seemsto serve the role of a H shuttle or perhaps as a precursor to the short chain hydro-carbon radicals, stabilising the benzyl and phenyl radicals to toluene and benzeneand/or promoting DPM conversion as in the mechanism of 2.6c [88]. The DPMitself may act as a H donor to stabilise benzyl and phenyl radicals. Unfortunately,due to the slow heat-up rate of the reactor, the influence of the catalytically activewall, the possibility of H starvation effects and the supercritical phase formation, amore precise mechanism could not be substantiated from the data obtained in the250 cm3 stirred batch reactor.5.1.3 Summary of Model Compound EvaluationThe data obtained from the stirred batch reactor was both extremely useful andenlightening.The DPM, DPE and DPP model compounds were evaluated, with DPE andDPP being found too reactive for use in this study. The decalin diluent, whilstnot decomposing during the reaction, was shown to influence the conversion ofthe model compounds and their product distributions through its role as a H shut-tle. Both benzene and toluene were seen to decompose, although to only a minordegree, and appeared to follow the same decomposition mechanism.In undiluted DPM, the catalyst precursors were confirmed to form the desiredMoS2 active phase with a suitable particle size and distribution. Catalytic wallactivity was confirmed both through comparison of 0 and 600 ppm Mo data andthe presence of FeS in the recovered solids. Mechanisms for the hydroconversionreaction currently proposed in the literature [21, 46, 78, 88] were seen to be inap-155Thermal++Catalytic+ 3H2+H abstraction++H abstractionContinued cracking, isom., cond. reactionsShort chain radicalsCxHyThermal+ zH2Radical addition and crackingRadical addition andstabilisation+ CxHy+Catalytic+ 3H2Figure 5.6: Proposed thermocatalytic decomposition mechanism of diphenylmethane from data gathered in the stirredbatch reactor with 0 - 1800 ppm Mo, 415 - 445◦C, 13.8 MPa H2, 0 - 6 h, 700 RPM.156propriate and the beginnings of a new mechanism were proposed.Full details of this mechanism could not be reliably deduced due to variouscomplicating factors of the reactor system including:• Non-zero conversion upon reaching the reaction temperature due to slowheat-up rates (due itself to the size of the reactor),• Complicated product compositions with various uncertain mechanistic routesdue to either H starvation or the influence of the catalytically active walls,• Obscuring of catalytic activity by the catalytic activity of the wall (specifi-cally 0 ppm Mo versus 600 ppm Mo),• An unacceptable level of experimental uncertainty (likely due to the abovereasons),• Achievement of a supercritical phase between 30 and 40 wt% conversion,beyond which the mechanism changes dramatically.To further study this system it was necessary to accomplish several goals:• Reduce reactor size for faster heat-up,• Increase the H2:DPM ratio,• Isolate the reaction from the reactor walls and internals,• Reduce uncertainty / improve reproducibility,• Avoid supercritical phase formation (most easily by remaining at lower con-versions rather than changing operating temperature or pressure).5.2 Novel Reactor System Design and TestingWith data from the stirred batch reactor allowing DPM to be selected as a suit-able model compound and for the development of an initial DPM hydroconversionmechanism, a novel micro-reactor system was designed to overcome some of thecomplicating factors and refine the results.1575.2.1 Inclined Stainless Steel Micro-ReactorThe inclined stainless steel micro-reactor allowed for data to be collected and anal-ysed for preliminary comparison with that from the stirred batch system. Thisreactor was unmixed but its small size allowed for rapid heat-up (20 min com-pared to 80 min for the stirred batch system) and an improved H2:DPM molar ratio(2.7 compared to 0.8). Cool-down times showed slight improvements, the micro-reactor dropping to less than 400◦C in 60 s compared to 90 s for the stirred batchsystem. One downside of size reduction was an increase in the A:V ratio which isinversely proportional to the reactor diameter. With the stainless steel walls of thestirred batch reactor being shown to have a noticeable catalytic influence, and thatwith an A:V ratio of 1.3 cm2/cm3 (see Section F.8 for A:V diagrams and calcula-tions), wall effects would be clear in the stainless steel micro-reactor with its A:Vof 20.2 cm2/cm3, an increase of approximately 15.6 times.The first series of experiments conducted in this system was thus a study ofthe activation of the fresh stainless steel walls (of both the reactor and the thermo-couple), the DPM conversion data for these tests being shown in Figures 4.23 and4.24. These results indicated that for an 1800 ppm Mo system, the wall activityaccounted for a conversion increase of 85±3% for all reaction times (except 0 hwhere both activating and stable systems showed negligible conversion). Usingthis result, a rough correlation could be determined whereby each unit of the A:Vratio increased the conversion of an 1800 ppm Mo system by approximately 4.2%.That is, each 1 cm2/cm3 corresponded to approximately 76 ppm Mo (an effectiveinfluence with the active phase present as FeS on the walls and internals of thereactor).The rapid heat-up rate of the micro-reactor was a success with all experimentsin these systems shown to reach reaction temperature having undergone minimalDPM conversion. Comparison of this faster heating rate with one set to mimic thatof the stirred batch reactor (DPM conversion and product yield results provided inTable 4.10) confirming this (although experimental uncertainty made only roughconclusions possible). In general it appeared that the slower heat-up rate did allowDPM to begin reacting before reaching reaction temperature. The products fromthis lower temperature reaction correspond well with those gathered in the stirred158batch reactor, with the results of the slower heat-up indicating a lower B:T ratio, ahigher yield of CHMB and less isom./cond. products.With the walls of the reactor and thermocouple activated and stable, the exper-iments conducted in the stirred batch system were repeated in the inclined micro-reactor. The maximum reaction times were limited to 4 h so as to avoid supercrit-ical phase change occurring at elevated DPM conversions. The DPM conversionresults are shown in Figure 4.25. Comparison with the stirred batch reactor con-version data in Figure 4.6 shows the conversion achieved in the micro-reactor tobe less than 10 wt% below that of the stirred batch reactor. Superficially this indi-cates a good correlation between the reaction systems, but when the far greater A:Vratio in the micro-reactor, and the greater conversion which should result, are con-sidered, these lower conversion results reveal a more complicated system. Whilstboth the 600 and 1800 ppm Mo systems show a rapid initial increase in DPMconversion with reaction time, exceeding that of the 0 ppm Mo reaction, all threecatalyst concentrations were observed to converge after approximately 2 h and risesteadily thereafter. Furthermore, the 600 and 1800 ppm Mo systems appear almostidentical with the 600 ppm Mo reaction possibly exceeding that of 1800 ppm Moat some points during the reaction.To understand these trends, the product distributions must be considered (Fig-ures 4.26 through 4.31 and Table 4.11). The first clear trend is that the results of 0and 600 ppm Mo are extremely similar in terms of benzene and toluene yields andthe associated B:T ratio. This is certainly due to the high amounts of wall catalysisoverwhelming the lower Mo loading. Many of the trends observed in the stirredbatch reactor are repeated, with increased catalyst loading promoting CHMB andcracked species formation while inhibiting isom./cond. products. Curiously thetoluene yield now declines with 1800 ppm Mo, a fact which pushes the B:T ratiofor these experiments above 1:1.Despite the greater influence of wall catalysis, these reactions are markedlymore selective, with even the most diverse mixture (0 ppm Mo after 4 h) showingonly seven different species above 0.25 area% on the GCMS chromatogram. Thesewell-defined product species, together with the increase in the yield of crackedproducts for 1800 ppm Mo above 20 wt% conversion may help explain the trendsobserved and add another piece to the mechanistic puzzle.159All of the reaction products now contain CHMB in appreciable quantities, con-firming that this is a catalytic product resulting from the hydrogenation of DPM.All experiments also show hexF, the result of H* abstraction from CHMB and itssubsequent stabilisation. Only small quantities of fluorene are detected and onlyfor 0 and 600 ppm Mo after longer reaction times suggesting that H* abstractionfrom DPM is an undesirable alternative. This is supported by the reduced quanti-ties of isom./cond. products in the 1800 ppm Mo experiments wherein sufficientCHMB is present to avert H* abstraction from DPM and its subsequent attack byhydrocarbon radicals to form MBP or ETB.The decline in the CHMB yield and the increase in the toluene yield with in-creasing conversion are related and suggest that CHMB decomposes during thereaction to form both toluene and a saturated hydrocarbon radical. Thermody-namic simulations, presented in Figure 5.7, support this mechanism with the ∆Grassociated with CHMB decomposition to benzyl radicals being favoured over thatto phenyl radicals.++∆Gr718K = 281 kJ/molPhenyl∆Gr718K = 163 kJ/molBenzylFigure 5.7: Comparison of Gibbs free energies of reaction for the thermoly-sis of cyclohexylmethylbenzene simulated in Accelrys Materials Studio(v4.4) using DMol3 [110, 111] geometry optimisation and frequencycalculations.Simulation details and parameters are provided in Section F.4.The fate of the cyclohexyl radical is unclear at this point. It does not appear inthe product as cyclohexane and must thus decompose to short chain hydrocarbonradicals. The increased concentration of these radicals would, however, force addi-tional H* abstraction from both DPM and CHMB (which may stabilise to fluoreneor hexF respectively) or radical addition to the DPM (to form MBP or ETB), re-sulting in an increase in the isom./cond. product yield. This is counter to the trendobserved. It thus seems that these hydrocarbon radicals instead interact preferen-tially with other species in the reaction mixture, a mechanism which requires more160data to clarify.With the highest yields of CHMB of the three series, 1800 ppm Mo would beexpected to show the highest toluene yield too. This is not the case with it in-stead showing the lowest yield. This trend is connected with the 3-methylheptaneobserved in all 1800 ppm Mo samples. At such elevated catalyst concentrations(1800 ppm Mo combined with the wall catalysis) and CHMB yields, two pathwaysare possible. Toluene may be hydrogenated and crack to 3-methylheptane but thiswould produce various other products not observed in these experiments (see thetoluene blank test product composition results in Table 4.6). Alternatively, the lessfavoured CHMB to phenyl radical pathway may represent a significant route ofdecomposition. Each mole of DPM decomposing via this pathway would not onlyproduce a methyl-cyclohexane species (which would rapidly undergo thermoly-sis to 3-methylheptane) but it would simultaneously deprive the system of a moleof toluene whilst maintaining the production of benzene (hence the relatively un-changed benzene yields for different catalyst concentrations).One final trend yet to be explained in the inclined micro-reactor is the conver-gence of the 0, 600 and 1800 ppm Mo DPM conversion results above 2 h reactiontime. This trend is believed to be linked to the cracking product yields which areseen to be irregular for 600 ppm Mo and increase sharply above 2 h for 1800 ppmMo. This micro-reactor was designed to allow volatile species (such as shortercracked products) to leave the liquid phase, condense in the cold zone and runback. At an angle of only 30◦, however, the droplets required to form before run-ning back would be quite large (note that the A:V ratio does not change as the bulkliquid volume declines as may be seen in the calculations provided in Section F.8).As suggested by the mechanism of LaMarca et al. [88] ( 2.6c), the cracking prod-ucts leaving the liquid phase may promote decomposition of the DPM. The con-version convergence and the fate of the cyclohexyl radical now become clear. In-creased catalyst loads promote the formation of CHMB and its decomposition tobenzyl and cyclohexyl radicals. These cyclohexyl radicals decompose into shortchain hydrocarbon radicals and themselves promote the decomposition of DPM.This explains the more rapid initial rise in conversion for 600 and 1800 ppm Moexperiments. The short chain hydrocarbons do, however, leave the liquid phaseand begin condensing elsewhere in the reactor, slowing the rate of consumption of161DPM and speeding the formation of more short chain radicals. When the dropletsdo eventually return to the bulk, the concentration of such cracking species is ob-served to rise again (most sharply in the case of 1800 ppm Mo samples as theseinclude the 3-methylheptane from benzyl radical hydrogenation and cracking).Figure 5.8 summarises the proposed mechanism thus far together with the re-sults from additional thermodynamic simulations. From the Gibbs free energies ofreaction shown it is apparent that catalytic hydrogenation of DPM to CHMB andits subsequent thermal cracking is comparable to the thermal cracking of DPM. Bythis mechanism, the catalyst serves to hydrogenate DPM to form CHMB whichundergoes thermolysis to either benzyl and cyclohexyl radicals or, less favourably,phenyl and methyl-cyclohexyl radicals (see Figure 5.7). In the former case, benzylradicals stabilise to toluene (for instance by H* abstraction) while the cyclohexylradicals crack to short chain hydrocarbons which promote the decomposition ofDPM (although whether by DPM attack or product stabilisation is unclear). In thelatter case, phenyl radicals stabilise to benzene while methyl-cyclohexyl radicalscrack to 3-methylheptane (and possibly further to short chain hydrocarbon radicalsas per cyclohexyl radicals).5.2.2 Vertical Stainless Steel Micro-ReactorWith the orientation of the inclined stainless steel micro-reactor appearing to affectoperation, and to gather data for comparison with the glass insert micro-reactor,the stainless steel reactor was operated in a vertical orientation. This system usedthe same reactor body and thermocouple (with the activated walls) as the inclinedunit. This change in orientation was shown to impact the gas-liquid transfer area(as seen in Section F.8.1, vertical orientation of the reactor halved the surface areafrom that of the 30◦ inclined orientation from 0.234 to 0.117 cm2) and condensedliquid run-back rate but not the A:V ratio (see calculations in Section F.8).From Figure 4.32 it is apparent that a change in the orientation affected achange in the DPM conversion. Beneficially, the DPM conversion after 4 h is nowbelow the supercritical level. Interestingly, the thermal and catalytic systems nowlie within experimental uncertainty of one another (the 1800 ppm Mo, surprisingly,being lower on average) and may be represented by first order reaction fits. The162Thermal++Catalytic+ 3H2+H abstraction++H abstractionContinued cracking,isom., cond. reactionsShort chain radicalsCxHyThermalRadical addition and crackingRadical addition andstabilisation+ CxHy++Thermal+ zH2∆Gr718K = 218 kJ/molDPM∆Gr718K = 56 kJ/molDPM∆Gr718K = 163 kJ/molCHMBCatalytic+ 3H2Figure 5.8: Proposed thermocatalytic decomposition mechanism of diphenylmethane from data gathered in the inclinedmicro-reactor with 0 - 1800 ppmMo, 445◦C, 13.8 MPa H2, 0 - 4 h, 0 RPMwith simulations performed in AccelrysMaterials Studio (v4.4) using DMol3 [110, 111] geometry optimisation and frequency calculations.Simulation details and parameters are provided in Section F.4.163coefficient of this fit for 0 ppm Mo was seen to be comparable to that of the 0 ppmMo experiments of the stirred batch reactor (0.082 h−1 and 0.088 h−1 respectively)and suggests the reason for the similarity of these trends to be H2 limitations. Thisreasoning fits well with the observed trends if rapid dissolution of H2 into but slowdiffusion through the liquid phase are assumed. The inclined system, with a largersurface area, reaches reaction temperature with more dissolved H2 than does thevertical system. This accounts for the more rapid initial increase in DPM conver-sion in the inclined system and, as dissolved H2 is consumed near the gas-liquidinterface, little can penetrate deeper into the liquid with the reaction being seen toslow at increased reaction times. The lesser gas-liquid surface area of the verticalsystem means less H2 is available upon reaching reaction temperature (this mimicsthe stirred reactor system where H2 was limited by bulk supply rather than surfacearea). The initial rate is thus lower than the inclined system and stays steady for allreaction times as H2 is consumed near the gas-liquid interface and is unable to dif-fuse deeper. In the vertical system with 1800 ppm Mo, additional consumption ofH2 by the catalyst to form hydrogen-rich products (such as CHMB) limits the sup-ply even further, resulting in the slight decrease observed when compared to the 0ppm Mo experiments. This argument relies on two assumptions: the dissolution ofH2 into the liquid is fast (resulting in the supply near the gas-liquid interface uponreaching reaction temperature) and the diffusion of H2 through the liquid is muchslower (resulting in the confinement of H2 close to the gas-liquid interface and therelationship of the hydrogen concentration with the gas-liquid surface area).The former assumption is often made in the literature when working with apure gas which is highly soluble in the liquid [116, 118–120], such as H2 has beenshown to be in DPM in this study (recall that under reaction conditions, H2 candissolve in DPM to approximately 20 mol%). In studies using batch systems, thisrapid dissolution often makes quantification of the gas-side mass transfer coeffi-cient (kGa) extremely difficult due to the complexities of measuring the rapid dropin gas pressure [119]. To support the use of this assumption in this work, ex-periments were conducted whereby the pressure of the micro-reactor, loaded withDPM and at 20◦C, was rapidly increased and monitored with time (the pressurebeing sampled to within 7 kPa [1 psi] every 1 s by the pressure transducer). Theresults are shown in Figure 5.9. Simulations into H2 solubility (simulation and164calculation details provided in Section F.7) showed an increase in solubility withtemperature (although not the expected trend, this is supported by published studies[113, 114, 121, 122]), so these tests at 20◦C could be considered under-estimationsof the rate of dissolution. The pressure test data showed that following pressuri-sation to 12.76 MPa, the gas pressure dropped by more than 82 kPa within 40 s.This pressure change, by ideal gas assumptions, would result in a H2 concentrationof approximately 14.0 mol% in the liquid (calculations detailed in Section F.7.2).Whilst additional factors may exaggerate this value (such as spring-back of thetransducer mechanism following the rapid pressurisation or adsorption of H2 onmetal surfaces), it is clear that the H2 does rapidly dissolve in the DPM (possiblysaturating the liquid near the gas-liquid interface) and the assumption is valid.0 20 40 6012.6812.7012.7212.7412.76Hydrogen pressure (MPa)Time from maximum pressure (s)Figure 5.9: Pressure drop after initial, rapid pressurisation with H2 in verticalmicro-reactor loaded with 150 µL diphenylmethane at 20◦C to studythe H2 dissolution rate.The latter assumption, that diffusion through the liquid is significantly slowerthan this dissolution, required that the H2 concentration along the vertical axisbe modeled as a function of both time and liquid depth. Described in detail inSection F.7, this required estimation of the liquid-side mass transfer coefficient(kLa) using the modified Wilke-Chang correlation [123] (1.09×10−9 m2/s in DPMunder reaction conditions), data from molecular simulations performed in AccelrysMaterials Studio (v4.4) and Fick’s second law of diffusion to produce the concen-165tration profiles presented in Figure 5.10. It may be seen that even after 60 min,without consumption due to reaction, H2 penetration by diffusion is limited, evenat a depth of only 1 cm (recall that the depth of 400 µL of liquid in the micro-reactor is 3.42 cm). This supports the assumption of liquid-side diffusion beingslower than dissolution from the gas, thereby justifying the DPM conversion inter-pretations above.0.00.51.01.5t = 15 mint = 30 mint = 45 mint = 60 min Hydrogen concentration in diphenylmethane (kmol/m3)Saturation concentration = 1.376 kmol/m3t = 15 mint = 30 mint = 45 mint = 60 min0.0 0.5 1.00.000.050.10Distance from gas-liquid interface (cm)Figure 5.10: Concentration profiles of H2 diffusing from saturated gas-liquidinterface into diphenylmethane at 445◦C and 13.8 MPa as per Fick’ssecond law of diffusion using data from AspenTech Aspen Plus (v7.3)solubility simulations1 and the modified Wilke-Chang correlation2[123] (with data from Accelrys Materials Studio simulations3).1,2- Simulation details, parameters and calculations provided in Section F.7. 3 - v4.4using DMol3 [110, 111] geometry optimisation and frequency calculations as shown inSection F.4.Whilst the DPM conversion data indicated the dependence of the system on H2diffusion rates through the liquid, the product distributions in the vertical stainlesssteel reactor (Figures 4.33 through 4.37) provided additional detail regarding themechanism of the reaction.Although the B:T molar ratio for both 0 and 1800 ppm Mo experiments re-mained roughly equivalent and constant (slightly below 1:1) for all DPM con-versions, the 1800 ppm Mo system exhibited lower yields of both benzene and166toluene than were observed for 0 ppm Mo. This suggests the primary route forbenzene and toluene formation to be the same for both systems. Both systemsshowed an increase in benzene and toluene yield with increasing DPM conversion.The reduced toluene yield with 1800 ppm Mo was previously attributed to catalytichydrogenation of DPM to CHMB and its subsequent decomposition to phenyl andmethyl-cyclohexyl radicals. This pathway was characterised by 3-methylheptanein the reaction product (the result of methyl-cyclohexyl thermolysis). In the verti-cal reactor, however, only trace amounts of other cracking products were detectedin either the 0 or 1800 ppm Mo systems, this despite the CHMB yields being inthe same order of magnitude as those of the inclined reactor. Furthermore, the onlyisom./cond. product appearing in measurable quantities was fluorene, with hexFbeing noticeably absent from the analyses. The fluorene yield trends for both 0 and1800 ppm Mo were seen to be equivalent, suggesting the formation of this speciesto be independent of catalytic activity. This is a plausible assumption given thatits formation likely occurs via the abstraction of H* from DPM by hydrocarbonradicals and that the ∆Gr for such single-step abstraction by benzyl and phenylradicals is only slightly higher than with a short chain radical (such as a propylradical) acting as an intermediate (as seen in Figure 5.11).∆Gr718K = 68 kJ/molDPM∆Gr718K = 56 kJ/molDPM+ ++ ++∆Gr718K = 12 kJ/molPropane+Figure 5.11: Comparison of Gibbs free energies of reaction for the abstrac-tion of H* from diphenylmethane by benzyl and propyl radicals sim-ulated in Accelrys Materials Studio (v4.4) using DMol3 [110, 111]geometry optimisation and frequency calculations.Simulation details and parameters are provided in Section F.4.Despite quantification of the gaseous products indicating their representation ofonly a minimal mass product stream (gas product compositions shown in Table 4.13),their composition is indicative of the liquid reaction occurring. As seen in the167stirred batch reactor, the addition of MoS2 catalyst not only increases the yield ofgaseous products, but also shifts the spectrum toward larger species (particularlythose in the C5 and C6 range as shown in Figures 4.38 and 4.39). This, togetherwith the higher yield of CHMB in the 1800 ppm Mo experiments, supports thetheory of CHMB decomposition to short chain hydrocarbon radicals.It is proposed that CHMB forms rapidly at low conversions (catalysed by boththe active walls and MoS2, hence the slight increase in the 1800 ppm Mo system)and begins to decompose as previously discussed. The limited H2 available in theliquid limits the amount of CHMB formed and as decomposition progresses, sothis initial concentration declines. The yields of CHMB eventually tend towardthe pseudo-stable levels observed in the inclined system as the reaction shifts torelying on H2 as it dissolves into the liquid rather than the initially higher amountaccumulated during heat-up. The short chain hydrocarbon radical products fromthis CHMB decomposition serve to stabilise the benzyl and phenyl radicals fromDPM thermolysis. Whether this occurs by radical addition (for instance a phenylradical and a methyl radical to form toluene), radical hydrogen transfer (such asfrom propane to a phenyl radical to form benzene and a propyl radical) or by radicaldisproportionation (a propyl radical and a phenyl radical reacting to form benzeneand propene, the latter being re-hydrogenated by the catalyst), examples depictedin Figure 5.12, is unclear. Several such mechanisms may be responsible and somemay even work in series (propane, for instance, proceeding through a chain reactionof the second, third and forth mechanisms before terminating by radical addition).As these radicals form and are stabilised, they may vaporise and leave the liquidmixture but, due to the orientation of the reactor, rapidly condense and return. Forthis reason, the concentration of short chain hydrocarbons in the liquid remainshigher than was observed in the inclined system (where larger droplets had to formand run back into the bulk reaction mixture).This consistently higher concentration of short chain hydrocarbons and theirradicals has three key effects. Firstly, the stabilisation of the benzyl and phenylradicals from DPM is promoted, hence the higher yields of benzene and toluenethan in the inclined reactor. Secondly, this supply of stabilising species reduces theextent to which H* abstraction from DPM and CHMB occurs, decreasing fluoreneyields and all but eliminating hexF formation (fluorene production having a lower168overall ∆G f than hexF), a trend also observed in the inclined system. Finally, thehigher concentration of these short chain species in the liquid promotes the reverserecombination reaction of their formation from CHMB. This limits the extent towhich CHMB decomposition occurs and favours that reaction with the lower ∆Gr,specifically the formation of benzyl and cyclohexyl radicals. This accounts forthe negligible yield of 3-methylheptane (the methyl-cyclohexyl radical is an un-favoured product), the B:T ratio below 1:1 (toluene is preferentially formed, albeitslightly) and the lower benzene and toluene yields in the 1800 ppm Mo experiments(a higher yield of CHMB means a lower yield of benzene and toluene).∆Gr718K = -311 kJ/molPhenyl∆Gr718K = -77 kJ/molPhenyl+++∆Gr718K = -274 kJ/molPhenyl++CH3+ H2∆Gr718K = -45 kJ/molPropeneFigure 5.12: Comparison of Gibbs free energies of reaction for variousphenyl radical stabilisation mechanisms as simulated in Accelrys Ma-terials Studio (v4.4) using DMol3 [110, 111] geometry optimisationand frequency calculations.Simulation details and parameters are provided in Section F.4.5.2.3 Unmixed Glass Insert Micro-ReactorProgressing from the vertical stainless steel micro-reactor experiments to the useof the glass insert micro-reactor allowed for a determination of the influence ofthe active reactor and thermocouple walls. The glass insert micro-reactor, usingthe same activated thermocouple as the vertical stainless steel unit, was shown tohave an A:V ratio of 4.0 cm2/cm3, down from the 20.2 cm2/cm3 of the stainlesssteel micro-reactor (see Section F.8) but still higher than the stirred batch reactorat 1.3 cm2/cm3. As is clear from the DPM conversion and product yield results inFigures 4.40 through 4.45, this decrease in the A:V ratio had a notable impact on169both the observed conversion and product distribution.Presented in Figure 4.40, the DPM conversion using the glass insert was ob-served to be adequately modeled by first order kinetics. A comparison of the ki-netic coefficients between the glass insert micro-reactor and the vertical stainlesssteel micro-reactor (Tables 4.14 and 4.12 respectively) indicated a rate decline of53±3% in the 0 ppm Mo reactions and 42±2% in the 1800 ppm Mo reactions. Thisdifference in changes between the two micro-reactors accounts for the 1800 ppmMo DPM conversion data now lying above that of the 0 ppm Mo system. Unlikethe stainless steel systems, these trends no longer lie within experimental uncer-tainty. The reason for the decline in conversion is due to the decline in the A:Vratio. Reduced contact with activated stainless steel surfaces reduces the effectivecatalyst concentration from these sources from approximately 1500 ppm Mo in thestainless steel micro-reactor to approximately 300 ppm Mo with a glass insert (at76 ppm Mo per cm2/cm3 as shown in Section 5.2.1). It should be noted that thechange in DPM conversion with changing A:V ratio is not linear and suggests thateven small quantities of catalyst are able to promote the decomposition of DPM.With this greater difference in the effective catalyst loading between the 0 and1800 ppm Mo experiments, the product yields begin to reveal a more accuratemechanism. Foremost is that only trace quantities of CHMB were detected in the0 ppm Mo products, a predictable result based on the deductions of this being acatalytic hydrogenation product. The benzene and toluene yields and the B:T ra-tio now all exhibit differences between 0 and 1800 ppm Mo experiments. At lowDPM conversions, the 1800 ppm Mo system rapidly hydrogenates DPM to CHMBwhich subsequently decomposes to benzyl and cyclohexyl radicals. This not onlypromotes the formation of toluene (as evidenced by an initially higher toluene yieldfor 1800 ppm Mo), but simultaneously reduces the formation of benzene. It is pro-posed that the benzene yield is suppressed due to a double effect. The first is areduction in the thermolysis of DPM to benzyl and phenyl radicals due to its hy-drogenation to CHMB. The second is the formation of short chain hydrocarbon rad-icals from this CHMB which may stabilise the phenyl radicals to alkyl-substitutedspecies which may then crack to benzyl radicals and stabilise to toluene. As the re-action progresses, such short chain hydrocarbon radicals are formed in sufficientlylarge quantities in the 0 ppm Mo system too and the toluene yield for that system170rises. The benzene yield for 1800 ppm Mo stays suppressed as the higher cata-lyst concentration continues to produce CHMB. This theory is supported by theisom./cond. product yield which indicates a greater suppression of these reactionsin the 1800 ppm Mo experiments, a trend explained in Section 5.2.2.The lack of significant cracking products in both the glass insert micro-reactorand the vertical stainless steel micro-reactor support the theory proposed surround-ing the mechanisms of Figure 5.12. By this theory, the short chain hydrocarbonradicals (formed predominantly from CHMB and hence promoted by the catalyst)perpetuate a radical chain reaction, stabilising other radicals in the system andhence limiting their own concentration.5.2.4 Mixed Glass Insert Micro-ReactorWith an understanding that the mechanism is limited by the liquid-side diffusionof H2 (the dissolution being comparatively fast), agitation of the system to increasegas-liquid contact and H2 distribution through the liquid was deemed necessaryto understand the mechanism without H starvation limitations and to ensure solidsuspension (a factor which would be crucial when working with coke-catalyst ag-glomerates).With the vortex mixing of a slurry-phase micro-reactor having no precedencein the literature, determination of the nature of the mixing and quantification of itseffectiveness were conducted in two ways. The first was by visual inspection of thereaction mixture at different mixing speeds. The second, by analysis of the impactof mixing on the DPM hydroconversion reaction.Visual Mixing StudiesA “conventional” stirred tank is designed with the ratio of liquid depth, H, to vesselwidth, D, of approximately 1.0 [119, 120]. Baffles may be used with the standardwidth of D/10. The impeller of such a system would be positioned H/3 from thebottom of the vessel [120]. By these standards, the glass insert of the micro-reactorused in this study would certainly be “unconventional” [119] with H/D = 62.5.Definitions aside, the complexity of designing and operating such a long, narrowimpeller were found prohibitive. Albal et al. [119] studied the use of various mix-171ers in conventional and unconventional vessels and determined that, with impellermixing, the mixing time for gas-liquid and liquid-liquid systems was reduced inthe unconventional arrangement, likely due surface breakage and “surface entrain-ment” occurring at lower mixing speeds [119, 120]. Albal et al. [119] also notedimproved homogeneity and dispersion for gas-liquid-solid systems. Whilst not di-rectly applicable to this study where no impeller was employed, these results wereencouraging. Of interest is that the “critical stirrer speed” for surface breakage[119] and solids suspension [119, 124] was determined by visual studies usingglass vessels. A similar methodology was employed in this study using a glassmock-up of the micro-reactor system and a sample of reaction product as describedin Section 4.2.3.Vortices in liquids are well known and understood [125] and may be viewedas rotating masses of fluid with minimal axial or radial mixing, properties whichusually make them indicative of poor mixing. This, however, is usually only truein larger vessels with powered mixing where the velocity of a particle increasesproportionally to its distance from the central axis, a rotational or rigid-body vortex.In a confined system, such as the glass insert, friction with the walls would inhibitparticle motion away from the central axis, resulting in a velocity gradient andformation of an irrotational vortex. The velocity gradients in such a vortex wouldcreate sheer forces between inner and outer layers of fluid, moving at differentrotational velocities about the central axis, and promote liquid-liquid and liquid-solid mixing. Formation of the vortex core would also increase gas-liquid exchangeby increasing the transfer area.Whilst this was the theory supporting the development of a vortex-mixed reac-tor, the analysis of mixing in true vortices in this study is moot. As may be seenfrom the high speed camera stills in Figures 4.46 through 4.50, vortex formationdoes not occur in this system, at least not under ambient conditions. Instead it ap-pears that the cohesive forces of the liquid mixture exceed the centrifugal forces ofthe vortex mixer and result in the formation of a rotating wave. Whilst the radialdepth of this wave was found to remain roughly constant at 1.7 mm, the height ofthe wave/“vortex” was almost linearly related to mixer speed (over the range in-vestigated). At the maximum sustainable mixer speed of 2500 RPM, the “vortex”height was approximately 28.5 mm, insufficient to agitate the liquid and suspend172the solid particles. Reducing the liquid volume to 150 µL allowed the entire liquidvolume to be drawn up into the wave at 2500 RPM but 2000 RPM was found to bethe critical speed at which particle suspension occurred.Implementation of the glass insert was not, however, simply liquid in a tube butrather included the central thermocouple. Studying the effect of this thermocoupleon mixing was extremely enlightening with it appearing to act as a stirrer bar. With400 µL of liquid a mixing speed of 2500 RPM was required for surface entrainmentwhereas with 150 µL, entrainment began at 2000 RPM, appeared to peak at 2250RPM but at 2500 RPM had been replaced by the familiar wave of liquid. Althoughsimply visual confirmation of mixing, these results suggested a significant degreeof gas-liquid-solid agitation.The use of actual reaction product for these visual mixing studies ensured