THE EFFECT OF MEMBRANES IN METHANE DEHYDROAROMATIZATION
ON A BIFUNCTIONAL MO/H-ZSM-5 CATALYST IN A PACKED BED
REACTOR
by Benjamin Lyons Kee c Copyright by Benjamin Lyons Kee, 2016
All Rights Reserved A thesis submitted to the Faculty and the Board of Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Master of Science (Mechan- ical Engineering).
Golden, Colorado Date
Signed: Benjamin Lyons Kee
Signed: Dr. Steven DeCaluwe Thesis Advisor
Signed: Dr. Canan Karakaya Thesis Advisor
Golden, Colorado Date
Signed: Dr. Gregory Jackson Professor and Head Department of Mechanical Engineering
ii ABSTRACT
Methane is becoming an increasingly attractive resource. Conversion to liquids and higher-value fuels add to the opportunities available for natural gas processes. Conventional methods for converting methane to larger hydrocarbons are through Fischer-Tropsch (F-T), which require multiple unit steps that can add inefficiencies. Methane dehydroaromatization (MDA) is a direct catalytic route to from methane to benzene. Unfortunately, the short- comings of MDA include likelihood to coke and poor yield. Conversion can be increased by removing product hydrogen via a membrane to shift the thermodynamic equilibrium towards the products. A model is developed to study the coupling of hydrogen-selective membranes with packed- bed methane dehydroaromatization (MDA) reactors using bifunctional Mo/H-ZSM-5 cata- lysts. The computational predictions are supported by previously published literature. The effect of hydrogen removal is evaluated as a function of reactor temperature, gas-hourly space velocity (GHSV), and hydrogen removal rate. Consistent with Le Chˆatelier’s principle, the results reveal that membrane integration significantly increases methane conversion. How- ever, the desired benzene yield can be diminished by the yield of undesired naphthalene and higher hydrocarbons. The benzene-to-naphthalene ratio depends strongly and nonlinearly on the membrane hydrogen removal. The results suggest that hydrogen membranes are most beneficial when the GHSV is relatively high and the catalyst temperature is relatively low. Although the single-pass benzene yield remains below 10%, the hydrogen membrane can increase benzene production rates. Hydrogen removal was also explored for MDA with steam addition. Water can preferen- tially react with naphthalene to hinder the catalyst deactivation process. Naphthalene can crack to form lighter gases such as hydrogen, ethylene, and benzene. However, condensation reactions may occur to form larger polyaromatic hydrocarbons (PAH) such as anthracene.
iii Two percent steam premixed into the feed of an MDA reactor was simulated. Results suggest that water addition slightly decreases conversion, but instead improves the selectivity toward smaller hydrocarbons. Water reacts with the naphthalene to hinder PAH growth, which can lead to longer catalyst lifetime. High flow rates, low temperatures, and low hydrogen removal rates maximize non-coking benzene production.
iv TABLE OF CONTENTS
ABSTRACT ...... iii
LISTOFFIGURES ...... vii
LISTOFTABLES...... ix
LISTOFSYMBOLS...... x
ACKNOWLEDGMENTS ...... xii
DEDICATION ...... xiii
CHAPTER1 INTRODUCTION ...... 1
1.1 Background ...... 1
1.1.1 MDAcatalysts ...... 2
1.1.2 ChallengeswithMDA ...... 4
1.1.3 MDAwithhydrogenremoval ...... 6
1.1.4 WateradditioninMDA ...... 7
1.2 Prior work in hydrogen selective membranes incorporated into MDA reactors . . 7
1.3 Prior work in improving catalyst durability ...... 10
1.4 Overviewofpresentwork...... 12
CHAPTER2 COMPUTATIONALMODEL ...... 13
2.1 Conservationequations...... 14
2.2 Gas-phasetransport ...... 14
2.3 Hydrogen-permeablemembrane ...... 15
2.4 Computationalmethods ...... 16
v 2.5 One-dimensionalmodeljustification ...... 18
CHAPTER3 MDAREACTIONMECHANISM ...... 19
3.1 Microscopicreversibility ...... 20
3.2 Naphthalenesteamcrackingchemistry ...... 22
3.3 Effectofmicro-andmacroporosity ...... 26
3.4 Model comparison against experimental data ...... 29
3.5 Typicalsolution...... 30
CHAPTER4 DRYMDARESULTS ...... 33
4.1 Nohydrogenremoval ...... 33
4.2 Hydrogenmembrane ...... 37
CHAPTER5 WETMDARESULTS...... 42
5.1 Nohydrogenremoval ...... 42
5.2 HydrogenMembrane ...... 46
CHAPTER 6 SUMMARY AND CONCLUSIONS ...... 50
6.1 Summaryofefforts ...... 50
6.2 Keyfindings...... 51
6.3 FutureWork...... 52
6.4 Outlook ...... 52
REFERENCESCITED ...... 54
vi LIST OF FIGURES
Figure 1.1 Methane molecule inside H-ZSM-5 cage structure ...... 3
Figure1.2 Molybdenumcarbidestructure ...... 4
Figure1.3 TransientbehaviorofMDA ...... 5
Figure 1.4 Membrane coupled packed-bed reactor in an annulus design ...... 6
Figure 1.5 Protonic-ceramic membrane coupled packed-bed reactor in an annulus design ...... 8
Figure 2.1 Illustration of the computational domain, including labeling of fluxes andreactorgeometry...... 13
Figure 3.1 MDA reaction pathways proposed by Wong et al...... 20
Figure 3.2 Reaction scheme for naphthalene decomposition proposed by Devi et al. . 24
Figure 3.3 Possible reaction scheme for naphthalene hydrocracking proposed by Wangetal...... 25
Figure 3.4 Comparison of model predictions (lines) against experimental data (markers) of Cat 1 from Sun et al. . Effect of temperature on methane conversion,benzeneandaromaticyield...... 27
Figure 3.5 Comparison of model predictions with the experimental data of Korobitsyna et al. . The measured outlet mole fraction are shown as diamonds on the right-hand sides of the plots...... 31
Figure 4.1 Effect of GHSV and reactor temperature on methane conversion rate for anMDAreactorwithnomembrane...... 34
Figure 4.2 Effect of GHSV on benzene selectivity and methane conversion an MDA reactor with no membrane. Increase or decrease in methane conversion −1 −1 corresponds to specific GHSV values between 750-25000 mL gcat hr as itisshowninFigure4.1...... 35
vii Figure 4.3 Effect of GHSV and reactor temperature on benzene yield for an MDA reactor with no membrane. Each data point on benzene to naphthalene −1 −1 ratio corresponds to the specific GHSV between 750-7500 mL gcat hr asitisshowninFigure4.1...... 36
Figure 4.4 Effect of hydrogen removal on CH4 conversion and product yield. ◦ −1 −1 T = 700 C, GHSV = 1500 mL gcathr ...... 37
Figure 4.5 Effect of hydrogen removal on (a) Methane conversion, (b) Benzene and naphthalene yields at temperatures 675 - 750 ◦C. GHSV values for all −1 −1 temperature points are 1500 mL gcat hr ...... 38
Figure 4.6 Effect of hydrogen removal on benzene formation and product −1 −1 distribution at a GHSV of 1500 mL gcat hr ...... 39
Figure 4.7 Effect of GHSV on C6H6 yield and formation rates in a membrane −1 −1 −1 −1 integrated MDA reactor. (a) 4500 mL gcat hr and (b) 6000 mL gcat hr . 40
Figure 4.8 Comparison of effect of GHSV on C6H6 yield and formation rate in a membrane integrated MDA reactor. Darker region indicates the desired operatingrange...... 41
Figure 5.1 Effect of water addition on benzene selectivity for varying GHSV −1 −1 between 750 - 15000 mL gcat hr ...... 43
Figure 5.2 Effect of water addition on methane conversion for varying GHSV −1 −1 between 750 - 15000 mL gcat hr ...... 44
Figure 5.3 Effect of water addition on PAH selectivity for varying GHSV between −1 −1 750 - 15000 mL gcat hr ...... 44
Figure 5.4 Effect of water addition on benzene to PAH yield ratio for varying −1 −1 GHSV between 750 - 15000 mL gcat hr ...... 45
Figure 5.5 Effect of water addition on benzene formation rate ...... 46
Figure 5.6 Selectivity of both benzene and PAH species for wet and dry conditions −1 −1 at a GHSV of 1500 mL gcat hr and973K ...... 47
Figure 5.7 Effect of hydrogen removal on benzene formation rate in wet conditions −1 −1 for GHSV of 1500 mL gcat hr ...... 48
Figure 5.8 Effect of water addition and hydrogen removal on benzene to PAH −1 −1 species yield ratio at a GHSV of 1500 mL gcat hr ...... 48
viii LIST OF TABLES
Table 3.1 MDA reaction mechanism on the Mo/H-ZSM-5 catalyst...... 23
Table 3.2 Reactor and catalyst properties used in kinetic modeling of experiments reportedbySunetal...... 28
Table 3.3 Comparison of model predictions against experimental data of Sun et al. at 1098 K. Effect of catalyst micro- and macroporosity on methane conversion,benzene,andaromaticyield...... 29
Table 3.4 Reaction conditions reported by Korobitsyna et al ...... 31
Table 4.1 Major variables and parameters for dry MDA simulations ...... 33
Table 5.1 Major variables and parameters for wet MDA simulations ...... 42
ix LIST OF SYMBOLS
Apparentvelocity...... v
Axialcoordinate ...... z
Axial mass fluxes of species k ...... jk
Bedpermeability ...... Bg
Bedporosity ...... εg
Carbon selectivity of species k ...... Sk
Carbon yield of species k ...... Yk
Catalystspecificsurfacearea ...... As
Conversionofmethane...... XCH4
e Effective diffusion coefficient of species k ...... Dk
Gasconstant ...... R
Gas-phasedensity ...... ρg
Gas-phasetemperature...... T
Gas-phasetortuosity ...... τg
Heterogenouschemicalproductionrates ...... s ˙k
Hydrogen-selective membrane permeability ...... kPd
Hydrogen-selectivemembranethickness...... LPd
Innerradiusofannuluspackedbed ...... Ri
Knudsen diffusivity of species k ...... Dk,Kn
Mass fraction of gas species k ...... Yk
x Membranecross-sectionalflowarea ...... AB
Membranehydrodynamicperimeter...... PB
Membrane mass flux of species k ...... jk,M
Microporosity...... εm
Mixture averaged diffusivity of species k ...... Dk,m
Molar density of species k ...... [Xk]
Molecular weight of species k ...... Wk
Numberofgas-phasespecies...... Kg
Outerradiusofannuluspackedbed...... Ro
Particlediameter ...... dp
Permeatesidepartial pressureofhydrogen ...... pH2,Perm
Poreradius ...... rp
Pressure...... p
Reactantsidepartialpressureofhydrogen ...... pH2,React
Sitedensity ...... Γm
Surface coverage site fraction of species k ...... θk,m
Time...... t
Viscosity...... µg
xi ACKNOWLEDGMENTS
I would like to acknowledge the support of my advisors Dr. Steven DeCaluwe and Dr. Canan Karakaya for their assistance and advice on this project. I would also like to ac- knowledge the help of Dr. Robert Kee and Dr. Huayang Zhu for their involvement in this work and many useful discussions on the subject. Finally, I would like to acknowledge Dr. Gregory Jackson for his participation as a committee member.
xii Dedicated to DPL
xiii CHAPTER 1 INTRODUCTION
The primary objective of this thesis is to explore the effects of hydrogen removal through a membrane in the conversion of methane to benzene via methane dehydroaromatization (MDA). The results provide new quantitative insight about the interactions between catalyst and membrane performance. The technical approach is based upon computational modeling, with particular atten- tion to shell-and-tube packed-bed membrane reactors. An elementary-step chemical reac- tion mechanism is developed to represent Mo/zeolite catalysts. Transport fluxes within the packed bed involve convective and diffusive contributions. Effective diffusion coefficients consider both micro- and macro-porsity, incorporating the transport within and around the catalyst pellets. Two MDA reactor setups are considered. One considers membranes that are selective to hydrogen alone (e.g., palladium or palladium alloys). The other considers premixing water into the feed stream. Several important results emerge from the study. Membranes that enable large hydrogen fluxes increase methane conversion, but at the expense of benzene selectivity. Benzene selectivity competes with undesired naphthalene formation, with the tradeoff depending on membrane performance. Low catalyst-bed residence times and moderate hydrogen removal rates increase benzene yields, but remain relatively low for achieving commercial viability. Long residence times with low levels of steam addition can significantly increase benzene selectivity. High flow rates and low temperatures with steam addition reduces the likelihood of coking. Model-based results suggest alternatives for process design and reactor operation.
1.1 Background
The recent abundance and inexpensive supply of natural gas presents new opportunities for conversion into useful chemical products. However, transportation of natural gas is
1 impractical, which has sparked interest in gas-to-liquids technology. Natural gas from oil refining is usually flared to meet environmental restrictions. As the prominent component in natural gas, methane is the focus of catalytic processes for converting methane to higher- value hydrocarbons [1]. Fischer-Tropsch (F-T) is the commercial standard for forming larger hydrocarbons. The issue with F-T processes is that methane must first be converted to synthesis gas (CO + H2) before conversion to larger hydrocarbons products. The multiple unit processes involved in F-T generate inherent inefficiencies. Methane dehydroaromatization (MDA) is a potentially viable route for converting methane to benzene[1–6]. Benzene has many attractive uses. Benzene is a common precursor for ethylbenzene and cumene, both of which are produced on commercially large scales. Addi- tionally, it is a common additive to gasoline and jet fuels. The many applications of benzene has driven MDA research. MDA can be expressed globally as
6CH4 ⇋ C6H6 + 9H2. (1.1)
This reaction is a direct, non-oxidative path to form higher hydrocarbons. MDA is an endothermic reaction that requires high temperature (700 − 800 ◦C) and relatively low gas-
−1 −1 hourly space velocities (750 − 1500 mL gcat hr ). MDA was first reported by Wang et al. [7] in 1993 using molybdenum supported on H-ZSM-5 zeolite, claiming 7-8% methane conversion and 100% benzene selectivity. Despite various lab-scale research advances since its discovery, low benzene yield and catalyst stability currently limit scale-up of MDA.
1.1.1 MDA catalysts
MDA requires a bifunctional catalyst, with a transition-metal (typically Mo, W, Cu, Re, V, or Ga) incorporated into a zeolite catalyst framework [8–11]. Molybdenum is found to deliver the best MDA performance in terms of benzene selectivity. Among the various Mo- doped zeolite structures, H-ZSM-5 [6, 12–17] and H-MCM-22 [18–21] are found to be the most selective catalysts for benzene formation (50%-80% selective to benzene).
2 Figure 1.1: Methane molecule inside H-ZSM-5 cage structure
Figure 1.1 shows the cage structure of an H-ZSM-5 zeolite (HnAlnSi96-nO192), where n=1 for the structure shown. The structure consists of silicon (colored white) pentasil units linked by oxygen (colored red) bridges. H-ZSM-5 has a high silicon to aluminum ratio. An aluminum ion (Al3+, colored blue) replaces a single silicon ion (Si4+) in the structure. The difference in charge is compensated by a proton (H+), which acts as the Brønsted acid site, that initiates the aromatization chemistry. MDA on bifunctional transition-metal incorporated zeolite catalysts depends both on high catalytic activity and facile transport to and from active catalyst sites through the zeolite channels. While the cage size typically varies between 3.5 and 8.5 A,˚ cage structures with dimensions close to the atomic diameter of benzene (approximately 6 A)˚ favor benzene production [2, 22]. There are several methods to produce the Mo/zeolite catalyst. They are typically fab- ricated by wet impregnation, followed by calcination. A Mo precursor (such as ammonium
3 Figure 1.2: Molybdenum carbide structure
molybdate) is added into zeolite powders and forms MoOx upon calcination. Another method
starts with MoOx and is physically mixed with the zeolite powders. During a carburiza-
tion process, which involves exposing the MoOx to a CH4 atmosphere, the MoOx becomes a molybdenum-carbide structure, forming the active Mo/zeolite catalyst. The molybde- num transforms into a molybdenum carbide when exposed to methane, but the exact car- bide structure is unknown. During carburization, the molybdenum is incorporated at the
2+ Brønsted acid sites. Figure 1.2 shows the molybdenum-carbide complex in the Mo2(CH2)5
2+ form [23]. Among the possible molybdenum carbide structures, Mo2(CH2)5 is the most stable structure for low Si/Al ratios. The Mo loading on zeolite can vary between 1-20 wt.% [24–26], but 4-6 wt.% loading typically delivers the best performance [26, 27]. The aromatic yield begins to diminish at loadings above 10 wt.%.
1.1.2 Challenges with MDA
Although MDA offers a direct path to higher hydrocarbons, there are major obstacles preventing commercial viability. For one, MDA is an endothermic reaction, that requires
4 high temperatures due to the stability and low reactivity of methane. Even with a very active catalyst, the methane conversion is equilibrium limited around to 14% at 700 ◦C. The global MDA reaction (Eq. 1.1) competes with other chemical pathways, including naphthalene
(C10H8) production, written globally as
10CH4 ⇋ C10H8 + 16H2. (1.2)
Significant naphthalene production decreases the benzene yield and is also the monomer precursor for polyaromatic hydrocarbons (PAH), which cause catalyst fouling.
Figure 1.3: Transient behavior of MDA
Due to catalyst deactivation, MDA is an inherently transient process. Even the most selective MDA catalysts produce undesired products. Condensed carbon and polyaromatic hydrocarbons are thermodynamically the more stable products, both of which deactivate the catalyst [28–32]. Figure 1.3 shows the typical transient performance of an MDA catalyst. The catalyst is first carburized in a methane rich environment to form the active molybdenum- carbide. After carburization, methane is converted to benzene during MDA. However, due to the formation of undesired products, the catalyst typically deactivates within 3 to 6 hours of time on stream. Catalyst deactivation decreases product yield and catalyst activity. An in-situ regeneration step is included to restore the catalyst. Typically, regeneration occurs by exposing the catalyst to an oxidant (O2,H2O, CO, CO2, NO) or reducing gas (H2). Oxidant gases oxidize the surface carbon in an exothermic reaction. Following regeneration, normal
5 MDA operation can continue. The MDA process and regeneration cycle repeats.
1.1.3 MDA with hydrogen removal
Consistent with Le Chˆatelier’s principle, implementing a hydrogen membrane to remove
H2 from an MDA reactor can increase methane conversion. Equation 1.1 shows that for every mole of benzene produced, 9 moles of hydrogen are also produced. By removing H2 via a hydrogen-selective membrane, the thermodynamic equilibrium shifts toward the products. Hydrogen-membrane integrated MDA reactors are a fairly new concept. Figure 1.4 shows a potential implementation for such a reactor. As illustrated, the catalytically active packed- bed is housed in the annular region of the reactor. Hydrogen produced within the catalyst bed is transported across the membrane into an inert sweep channel. The desired MDA products exit downstream.
Figure 1.4: Membrane coupled packed-bed reactor in an annulus design
There are a number of membranes that can effectively transport hydrogen. However, due to the high reaction temperatures of MDA, polymer membranes cannot be considered. Palladium is a well known and well studied hydrogen membrane [33]. Palladium membranes
6 are highly selective to hydrogen, and operate in the same temperature regimes as MDA. Proton conducting ceramic membranes (e.g., BZY10) also operate at high temperatures and are selective to hydrogen. The fluxes through ceramic membranes are significantly smaller than hydrogen fluxes through palladium membranes.
1.1.4 Water addition in MDA
Adding steam can have multiple effects on the MDA chemistry, including methane re-
forming, reacting with surface carbon, modifying the Mo2 surface structure, and cracking naphthalene. Surface carbon and PAH formation are the primary contributors to catalyst de- activation. Steam addition can potentially mediate both surface carbon and PAH formation, which would increase MDA operation time and improve the overall reaction efficiency. Mixed ion conducting protonic-ceramic membranes offer potentially viable alternatives to palladium-based membranes for hydrogen removal. The mixed-conducting ceramics trans- ports both protons and oxides ions [34, 35] and provide practical conductivities above 600 ◦C. The transport of oxide ions allows for water formation on the catalyst side while hydrogen is extracted into the sweep. Figure 1.5 shows a reactor that incorporates a protonic-ceramic membrane. The sweep gas now includes a small amount of water. Proton conduction across the membrane yields a net hydrogen flux out of the packed bed, into the sweep. A relatively
small counter flux of oxide ions forms O2 on the catalytic side, which rapidly reacts with the available hydrogen to form H2O.
1.2 Prior work in hydrogen selective membranes incorporated into MDA reac- tors
Recently, hydrogen membranes integrated into reactors, as illustrated in Figure. 1.4, have been proposed to increase the methane conversion in MDA. In principle, H2 removal via a hydrogen selective membrane will increase the desired benzene yield by shifting the thermo- dynamic equilibrium (Reaction 1.1) toward the products. While membrane integration to packed-bed reactors is a relatively new concept, several experimental and numerical studies
7 Figure 1.5: Protonic-ceramic membrane coupled packed-bed reactor in an annulus design have investigated the effect of palladium, palladium alloys, and dense-ceramic oxide (e.g., barium zirconates) hydrogen-separation membranes in MDA. Several prominent examples are summarized here. Larachi and co-workers studied membrane-coupled MDA using a palladium alloy [36, 37] and porous Pd-Ag/stainless steel membranes [38] over a 0.5 wt.% Ru-3 wt.% Mo/H-ZSM-5 catalyst in a shell-and-tube configuration at temperatures in the range 500 ≤ T ≤ 700 ◦C. They showed that membranes could approximately double the benzene yield, but led to severe catalyst-deactivating carbon formation [36], including both graphitic and polyaromatic forms [37]. The catalyst activity was recoverable by introducing excess hydrogen in the feed stream [37]. Wang et al. [39] studied MDA over a Re/H-ZSM-5 catalyst in a packed-bed reactor coupled with a Pd-Ag membrane. The temperature and gas-hourly space velocities (GHSV)
◦ −1 −1 were relatively low (T = 585 C, 120 ≤ GHSV ≤ 1440 mL gcat hr ). In this study, the
8 membrane increased methane conversion by 50%, relative to non-membrane tests, but the maximum methane conversion remained below 7.5%. Kinage et al. [40] coupled a palladium membrane disk with a packed-bed reactor loaded with 3 wt.% Mo/ZSM-5 catalyst operated at 610 ◦C. They concluded that hydrogen mem- brane incorporation increased the methane conversion, as well as the aromatic (benzene, toluene, and naphthalene) yield. They also indicated that the hydrogen flux through the palladium membrane was 40-50% less than expected, due to carbon deposition on the mem- brane surface. Natesakhawat et al. [41] performed membrane-coupled MDA experiments using a shell- and-tube reactor packed with a 4 wt.% Mo/H-ZSM-5 catalyst. A 125 µm thick palladium membrane was used for hydrogen removal at 700 ◦C. Severe catalyst deactivation rates occurred with and without the membrane. In the membrane-coupled reactor, coke deposition on the palladium membrane reduced the hydrogen permeability by 75%. Despite predictions that the membrane could increase the benzene yield by up to 360%, the observed yields were very low (≤ 2%). Membrane efficiency was shown to be strongly affected by GHSV. Iglesia and co-workers studied membrane-coupled MDA in a series of papers, both com- putationally [42, 43] and experimentally [44, 45]. Button-cell experiments by Liu, et al. [45] coupled a dense SrCe0.95Yb0.05O3−α proton-conducting membrane with a Mo/H-ZSM-5 cat- alyst. Results demonstrated that at low temperatures (T ≤ 677 ◦C), the membrane removed less than 7% of the hydrogen produced and did not significantly affect methane conversion. As the temperature increased to 720 ◦C, the hydrogen removal increased to ca. 16%. A moderate increase in methane conversion rate was observed, but was primarily associated with an increase in the C12+ hydrocarbon yield and correlated with a slight increase in the catalyst deactivation rate. Finally, the authors reported that co-feeding 2% carbon dioxide enhanced the catalytic stability and reduced the formation rate of the C12+ components. It therefore remains to be seen whether membrane-coupled MDA can significantly increase
C6H6 yields and thereby improve the overall process efficiency. Based on prior experimental
9 studies, the general conclusion is that membrane integration increases CH4 conversion rate but also increases catalyst deactivation rates. For successful membrane-coupled MDA, re- actor configurations and operating conditions must be identified that increase benzene yield while maintaining catalyst stability.
1.3 Prior work in improving catalyst durability
Several experimental studies have been published on the beneficial role of steam in MDA.
Experiments on additional oxidants such as CO, CO2, and O2 have shown to regenerate the catalyst. Ichikawa and coworkers [46–48] showed improved MDA stability, explained by PAH suppression at the zeolite Brønsted acid sites. However, adding more than 1.8% water also caused rapid catalyst deactivation, likely associated with Mo2C oxidation and dealumination of the zeolite. Yao et al. [49] claimed a synergistic effect between methane reforming and
MDA chemistry while premixing H2O. Stable catalyst performance was reported for up to 60 hours time-on-stream. Carbon monoxide formation assisted in removing the surface carbon. However, the excess hydrogen inhibited the dehydrogenation reactions. Liu et al. [50] reported temperature programmed oxidation (TPO), X-ray photoelec- tron spectroscopy (XPS), and high-resolution transmission electron microscopy (HRTEM) measurements that showed increased catalytic activity due to the removal of weakly bound surface carbon prior to its transformation to surface carbon or aromatic carbon. Additionally the molybdenum was in the form of Mo2C and Mo2OxCy after carburization. Adding more than 2% water caused the undesired formation of oxides, Mo2O-MoO3, reducing catalytic activity.
Skutil and Taniewski [51] studied the effects of premixing H2O, CO, and CO2 in methane feed for an MDA process using Mo/HZSM-5 catalysts at 725 ◦C. They showed that up to
2% H2O and 2.5% CO2 increased the catalytic activity, benzene yield, and improved the catalyst stability. However, further increases in H2O (9.5%) and CO2 (5 and 11%) caused rapid catalyst deactivation. Premixing 8.5% CO enhanced the catalytic activity and stability, but slightly decreased the benzene selectivity.
10 Li, et al. [52] reported increased catalyst stability when low levels of CO2 were mixed into methane feed streams. Ohnishi, et al. [53] reported higher benzene formation rates when low levels (≈ 1.8%) of CO or CO2 were incorporated into methane feed streams. However, higher CO and CO2 levels were found to decrease the aromatic yields because of competition with the dry-reforming pathway,
CH4 + CO2 ⇋ 2CO + 2H2. (1.3)
Low levels of steam could beneficially alter the Mo/ZSM-5 catalyst structure. Under non-oxidative conditions, molybdenum oxide in contact with methane converts first to an
oxycarbide (MoOxCy) and then further transforms to the less active molybdenum carbide
(Mo2C). The active structure for MDA processes is generally accepted to be Mo2C. When an oxygen source is present (e.g., CO, CO2, or H2O), the oxygen not only tends to regenerate the catalyst, but also changes the structure of Mo2C to an oxycarbide (MoOxCy) over long reaction periods [54]. Steam addition could also affect the Brønsted-acid site density [55–58]. Excess acid sites promote further aromatization, but steam addition reduces Brønsted acidity, leading to improved benzene selectivity and catalyst stability. According to Ding et al. [55] zeolite deactivation is caused by the PAH formation. The Brønsted acid sites promote cyclization and aromatization. Any excess acid sites tend to continue further aromatization from ben- zene to naphthalene, and higher condensible PAH. Steam promotes reducing the Brønsted acidity, leading to significant improvement in benzene selectivity and catalyst stability.
Ma et al. introduced another effect of water addition in MDA [57]. Instead of using H2O as a co-reactant in the reaction stream, they pre-treated the H-ZSM-5 support with steam, then loaded 4 wt% Mo on the zeolite support. Compared to a conventional Mo/H-ZSM-5 catalyst, steam treatment increased benzene yield and catalyst durability. The tetrahedral framework Al is removed from the zeolite lattice during steam treatment, which reduces both the concentration and strength of Brønsted acid sites. However, they believed sufficient acid sites survived for proper aromatization. Excess Brønsted acid sites favor polyaromatic
11 hydrocarbons (PAH) formation, therefore removal of the excess acid sites suppress coke formation.
1.4 Overview of present work
Although the previous experimental studies demonstrate the difficulty in maximizing both benzene yield and catalyst stability, the range of conditions tested was overall rather narrow and the complex chemistry makes it difficult to extrapolate these results to other op- erating conditions. While a detailed parameter study for each process variable is too expen- sive for lab-scale experiments, recent developments in MDA kinetic modeling enable detailed chemical analysis for membrane-coupled MDA reactor design and optimization [2, 5, 43]. The present study extends previous modeling results by investigating the role of hydrogen- selective membranes and steam addition in annular packed-bed reactors to optimize the benzene production rates. A detailed, microscopically reversible elementary-step reaction mechanism is proposed and validated against previous experimental results to predict the axial variation of the gas-phase and surface species over a bi-functional Mo/H-ZSM-5 cat- alysts. Computational simulations are designed to investigate the effects of membrane hy- drogen permeability, stream addition, and reactor operating conditions (i.e., temperature and GHSV) on the methane conversion, benzene yield, and propensity for catalyst degrada- tion (as indicated by PAH concentrations). The results identify favorable MDA operating conditions that area capable of high benzene yield and production rates while minimizing coking.
12 CHAPTER 2 COMPUTATIONAL MODEL
To explore the design of membrane-coupled MDA, a computational model was developed to simulate the detailed MDA kinetics in a packed-bed annular reactor. As illustrated in Fig. 1.4, the integrated packed-bed MDA reactor contains an annular porous packed bed, gaseous species-permeable selective membrane, which is normally supported on the porous structure, and the internal sweeping gas-flow channel. Fig. 2.1 shows the radially symmetric computation domain. The model assumes one-dimensional, isothermal, axial
transport within the the packed bed, represented as species mass fluxes jk. Species transport
through the membrane is expressed as a radial mass flux jk,M. The inner and outer radii
of the packed-bed annulus are denoted as Ri and Ro, respectively. The cross-sectional flow area of the packed bed and the hydrodynamic perimeter of the membrane exterior can be
2 2 expressed AB = π (Ro − Ri ) and PB =2πRi, respectively.
jk,in jk jk jk,out r Catalytic Packed Bed Reactor R o j L k,M Pd Hydrogen Permeable Pd Membrane R i Inert Sweep Gas z
Figure 2.1: Illustration of the computational domain, including labeling of fluxes and reactor geometry.
13 2.1 Conservation equations
The model is formulated as a series of continuum partial differential equations, which are integrated over a sufficiently long time span to simulate quasi-steady-state operation of the reactor. The transient species and overall mass balances are represented as
∂ (εgρgYk) PB + ∇· jk = Ass˙kWk + jk,M, (2.1) ∂t AB
K K K ∂ (ε ρ ) g g P g g g + ∇· j = A s˙ W + B j . (2.2) ∂t k s k k A k,M k k B k X=1 X=1 X=1 The independent variables are time t and the axial coordinate z. The dependent variables are the mass fractions Yk and the gas-phase density ρg. The porosity of the packed bed is
εg. Chemical production rates via heterogeneous reactions are represented bys ˙k. Surface reaction rates are evaluated at the isothermal surface temperature. The specific surface area
of the active catalysts (i.e., active surface area per unit packed-bed total volume) is As and
Wk are the species molecular weights. The gas-phase pressure is determined from the ideal- gas equation of state. For operating temperatures from 650 ◦C to 800 ◦C, the homogeneous gas-phase chemistry is assumed to be negligible in comparison to the heterogeneous surface chemistry [42].
The surface species coverages θk,m are solved from the heterogeneous reaction rates ∂θ s˙ k,m = k , (2.3) ∂t Γm
where Γm is the available site density for surface site m (m = Mo2C or H-ZSM-5).
2.2 Gas-phase transport
The gas-phase species mass fluxes due to convection and diffusion are represented as
e jk = ρgYkv − DkWk∇ [Xk] (2.4)
e where [Xk] are the species molar concentrations and Dk are the effective diffusion coefficients. The apparent velocity v due to the pressure gradient ∇p through the porous packed-bed
14 structure can be represented by Darcy flow approximation B v = − g ∇p. (2.5) µ
where µ is the viscosity. The bed permeability Bg can be determined by the Kozeny-Carman relationship
3 2 εgdp Bg = , (2.6) 72τg (1 − εg)
where dp is the mean particle diameter and τg is the tortuosity. Diffusion within the MDA packed bed primarily occurs by two methods: i.) Molecular diffusion - dominated by molecule-molecule collusions and ii.) Knudsen diffusion - domi- nated by molecule-wall collusions. The space between the catalyst particles is represented
as the macroporosity εg, and the space within the catalyst particles as molecules navigate through the catalyst pellets toward active catalyst sites is represented as the microporosity
εm. Effective diffusion coefficients are calculated from the random pore model [59, 60] to incorporate both types of porosity.
2 e 2 εm (1+3εg) Dk = εgDk,m + Dk,Kn, (2.7) 1 − εg
The Knudsen diffusivities Dk,Kn are defined as
2 8RT D = r , (2.8) k,Kn 3 p πW r k where rp is the average pore radius. The mixture-averaged diffusion coefficients Dk,m are determined from kinetic theory, retrieved from Cantera1.
2.3 Hydrogen-permeable membrane
Palladium membranes are well known for hydrogen separation. While palladium mem- branes may not be ideal for MDA applications, the hydrogen removal method used in this
1www.cantera.org
15 study follows a Sieverts’ law correlation that is common for palladium membranes n − n pH2,React pH2,Perm jH2,M = kPd WH2 , (2.9) LPd where partial pressures of hydrogen within packed bed (‘reactant’) and the sweep channel
(‘permeate’) are pH2,React and pH2,Perm, respectively. The palladium membrane thickness is
LPd. The exponent n (0.5 ≤ n ≤ 1.0) depends on the hydrogen partial pressure. For low pressures, which are common for most MDA applications, the diffusion of atomic hydrogen is the rate limiting step, then n =0.5. If hydrogen-hydrogen interactions within the palladium bulk become significant, n may increase. The hydrogen permeability of a palladium membrane can be expressed in the Arrhenius form,
E k = k0 exp − a , (2.10) Pd Pd RT where Ea is the apparent activation energy.
In this study, the H2 permeate flux is assumed to be much lower than the inert sweep gas
flow, and hence the sweep side hydrogen concentration is assumed negligible (pH2,Perm = 0). Additionally, the permeability and thickness were abstracted into a single variable A = kPd/LPd which represents the permeance, resulting in an updated Sieverts’ law correlation
j 0.5 H2,M = APdpH2,ReactWH2 . (2.11)
2.4 Computational methods
The computation model is written and executed within the Matlab2 environment. The data presented in this study was calculated using the Matlab 2015b release. Within Mat- lab, the ode15s solver is used. Most chemical reaction rate equation sets are inherently stiff. The rate of different reactions can vary greatly, which can cause efficiency problems with computational solvers. The ode15s solver specializes in stiff differential equations sets. Addi- tionally, a block tridiagonal jacobian sparsity pattern is supplied, to increase computational
2http://www.mathworks.com/products/matlab
16 efficiency and accuracy. Cantera is used to handle the thermodynamics and kinetics of the model (most notably, the net production rates). The gas-phase state is set by temperature, density, and mass fractions. The surface state is set by temperature and site coverages. After the state is set at each node, thermodynamics and kinetics values are calculated. Only the steady-state solution is relevant in the context of this problem, therefore the ode15s solver integrates over time until convergence criteria is met. The desorption of car- bonaceous species from the surface is the time-limiting step in the MDA process. At steady- state, the gas-phase atomic carbon count between the inlet and outlet should balance. If not, carbon species are still adsorbing or desorbing from the surface, indicating that steady-state has not been reached. The model is initialized with a stirred reactor initial guess. MDA processes start with fresh catalyst and inert gas-phase, however the stirred reactor initial guess provides a valid starting point that reduces computation time. The transient solver approaches the steady- state solution from the initial stirred reactor guess fairly fast, relative to a blank initial guess. The present model uses flux boundary conditions. The inlet boundary condition is a constant gas-phase flux. At the outlet, an exhaust pressure is supplied. The flux at the outlet boundary experiences no diffusive flux, and is purely convective. Jacobian structures proved to be extremely important to computational efficiency. Better formed Jacobians in the ordinary differential equation solver allowed for smarter time steps and reduced computational time. Calculation of analytical Jacobians determined that the numerical Jacobian were poorly formed. One major issue with the formation of incorrect Jacobian structures stemmed from the communication with Cantera. Species normaliza- tion of both surface and gas-phase species caused inaccurate elements in the Jacobian formed from finite differences. Adding new functionality to setting the states in Cantera and using partial densities as dependent variables mediated this problem.
17 2.5 One-dimensional model justification
The developed model ignores any radial gradients. A one-dimensional model was justified by comparing the time scale for axial and radial transport
R2 L ≪ , (2.12) D U where R is the radial length, D is the diffusion coefficient of hydrogen, L is the axial length of the bed, and U is the axial velocity. The time scale for diffusion in the radial direction was 2 orders of magnitude shorter than flow in the axial direction. The shorter time for radial transport indicates that radial gradients are resolved much faster than the axial counterparts. Therefore, the radial components can be ignored. When a membrane is implemented, the hydrogen removal rate has the potential to sur- pass the rate at which hydrogen can reach the membrane. The hydrogen removal rate was compared to the radial transport of hydrogen from the center of the packed bed to the membrane. For the practical hydrogen removal rates considered in this study, the diffusion of hydrogen was much faster than the rate of hydrogen removal at the membrane surface. For the limiting case where the flow rates are low and hydrogen removal is high, the radial transport of hydrogen is slightly larger than the hydrogen removal rate.
18 CHAPTER 3 MDA REACTION MECHANISM
Despite many efforts, the literature on MDA reaction mechanism is limited due to the complexity of the chemistry and the inherently transient behavior of the process. MDA reactors typically involve bifunctional catalysts, which incorporate a transition metal into the zeolite framework. Methane activation occurring on the metal sites to produce C2H4 and
H2, then further cyclization and aromatization occurring on the acidic zeolite sites. Moreover, because the catalyst deactivates within a matter of hours, determining the intrinsic kinetics is difficult. However, both global and detailed reaction kinetics have been proposed based on a quasi-steady state approximation, which focuses on the period of time after the catalyst is carburized but before significant catalyst deactivation occurs [2, 5, 42, 43]. A graphical representation of the MDA mechanism (Figure 3.1) was taken from Wong et al. [5]. The two sites that make up the catalyst are the molybdenum carbide site and the Brønsted acid site in the zeolite. On the molybdenum carbide site, the methane be- comes activated by first breaking the C-H bond. The primary intermediate created on the molybdenum carbide is ethylene [61]. On the Brønsted acid site, the ethylene cyclizes to form cyclohexane through oligomerizion, β-scission, alkylation, and de-alkylation reactions. Through protonation, dehydrogenation, and aromatization reactions, the aromatic hydrocar- bon benzene is formed. Aromatization continues on the Brønsted acid sites to form larger aromatic species such as naphthalene. The presented mechanism only proceeds to naphtha- lene, but even larger aromatics are expected to form. Methyl radicals may also react with benzene to form toluene, instead of aromatizing further. Karakaya et al. [2] presented a detailed reaction mechanism for MDA on a bifunctional Mo/H-ZSM-5 catalyst. The mechanism consists of 50 irreversible steps, which include 13 gas-phase and 18 surface species. In this reaction mechanism, methane is initially con-
19 Figure 3.1: MDA reaction pathways proposed by Wong et al. [5] verted to ethylene and hydrogen over molybdenum carbide, with subsequent aromatization chemistry on the zeolite Brønsted acid sites, mainly producing benzene, toluene, and naph- thalene. The chemistry was validated by previously published reports for a range of reaction temperatures, gas-hourly space velocities, and molybdenum/zeolite loading ratios. Because the reaction steps were described as irreversible pairs, forward and reversible rate constants (pre-exponential factors) were chosen independently and did not require the thermodynam- ics of surface species. Although the species distributions in these simulations did not exceed the thermodynamic limitations within the given reaction conditions, the overall microscopic reversibility was not guaranteed for all operating conditions.
3.1 Microscopic reversibility
Membrane-coupled MDA processes require more attention to microscopic reversibility since hydrogen removal shifts the thermodynamic equilibrium to drive additional methane
20 conversion. Establishing microscopic reversibility typically requires thermodynamic proper- ties for both gas-phase and surface species. Thermodynamic properties of most gas-phase species are well established and well documented [62–64]. However, in many cases the ther- modynamic properties for surface species are unknown. Theory-based calculations can pre- dict the species’ thermodynamic properties, but often requires a tremendous amount of work and time. On the other hand, Deutschmann, et al. [65–67] and Vlachos et al. [68, 69] have separately demonstrated algorithms to develop thermodynamics consistent kinetics without surface species thermodynamics, an approach that is followed in the current work.
For a fixed temperature, the equilibrium constant Kc,i can describe the equilibrium com- position of reaction i as a function of temperature T
− ◦ K kf,i ∆iG ◦ νk,i Kc,i = = exp (ck) , (3.1) k ,i RT r k Y=1 where the kf,i and kr,i are the forward and reverse rates constants, respectively. The stoichio-
metric coefficients are written as νk,i. For gas-phase species, the reference concentrations at
◦ ◦ ◦ normal pressure P are written as ck = P /RT . Surface species reference concentration can ◦ be determined as ck =Γ/σk where Γ is the site density and σk is the the number of surface sites occupied by species k. The temperature dependence of reaction free energy change
◦ ∆iG is described as a function of the free energy of each species
K ◦ ◦ ∆iG = νk,iGkT. (3.2) k X=1 Combining Equations 3.1 and 3.2, yields
K K 1 ln k − ln k = − ν G◦T + ln (c◦)νk,i . (3.3) f,i r,i RT k,i k k k k ! X=1 Y=1 For a fixed temperature, Eq. 3.3 is a linear equation system. For a given forward and reverse rate, the free energy can be calculated through a weighted least-square fit method. The procedure is repeated for several temperatures to determine the remaining thermodynamic properties.
21 In this study, the detailed kinetic mechanism presented by Karakaya et al. [2] is extended to force microscopic reversibility on the molybdenum site. The activation barriers and pre- exponential factors are fitted to experimental data for a range of temperatures, such that the overall reaction enthalpy and entropy are consistent. Table 3.1 shows the updated reaction mechanism that is thermodynamically consistent on the molybdenum site. The presented site densities are for pure molybdenum and pure H-ZSM-5. The molybdenum loading may vary, generally between 1-10%. The molybdenum site density is constant, independent of loading, at 1.26 × 10−6 mol m−2, while the zeolite site density must be adjusted for the deactivation due to molybdenum loading. The pure site density for the zeolite is 1.66 × 10−5 mol m−2, and weighting by the loading (94% for 6% molybdenum loading) gives 1.56 × 10−5 mol m−2.
3.2 Naphthalene steam cracking chemistry
Detailed pathways for naphthalene/water reactions over Mo/Zeolite catalysts are not well established. Figure 3.2 shows a reaction pathway for naphthalene steam cracking reactions over olivine (Mg-Fe silicate) catalysts reported by Devi et al. [70]. The suggested path- way starts at naphthalene. PAH species such as fluoranthene and pyrene can form through methylnaphthalene. Benzene, toluene, ethylene, and acetylene are formed through the in- termediate indene. The ethylene produced from naphthalene cracking can react again on the zeolite site to form more benzene. The pathway also includes molecular-weight growth to PAH coke precursor species anthracene, phenantracene, crysene, and pyrene, all of which are known to be formed over Mo/zeolite catalysts. Figure 3.3 shows a reaction pathway for naphthalene steam cracking proposed by Wang et al. [71]. They investigated how water affected ring opening and contraction reactions
of naphthalene over Rh2O3/HY zeolite and Mo-Ni oxide catalysts. Water was proposed as an effective additive for naphthalene ring opening, which is favored over condensation reactions in the presence of water, ultimately resulting in smaller alkanes and aromatics. Naphthalene also undergoes hydrogenation reactions to reach tetralin and decalin. From
here, the rings can contract and open, then crack to form C4 hydrocarbons. However, there
22 Table 3.1: MDA reaction mechanism on the Mo/H-ZSM-5 catalyst.
A β E Reaction (cm, mol, s) − (kJ mol−1) × −6 −2 Molybdenum site chemistry, ΓMo2 =1.26 10 mol m → × +13 − 1. CH4 + (m) CH4(m) 1.49 10 0.223 94.5 20 θ(m) +14 2. CH4(m) → CH4 +(m) 5.72 × 10 -0.223 54.8 +12 3. CH4(m) → CH2(m)+H2 6.90 × 10 0.223 126.3 +13 4. CH2(m)+H2 → CH4(m) 1.31 × 10 -0.223 103.3 +12 5. CH2(m) + CH4 → C2H6(m) 2.14 × 10 0.223 110.4 +12 6. C2H6(m) → CH2(m)+CH4 2.42 × 10 -0.223 103.3 +12 7. C2H6(m) → C2H4 +H2 +(m) 6.28 × 10 0.223 146.1 → × +15 − 8. C2H4 +H2 + (m) C2H6(m) 1.31 10 -0.223 19.8 20 θ(m) −5 −2 Zeolite site chemistry, ΓH−ZSM−5 =1.66 × 10 mol m +13 9. C2H4 + H(z) → C2H5(z) 7.00 × 10 0.50 138.2 +13 10. C2H5(z) → C2H4 +H(z) 6.05 × 10 0.00 206.2 → × +13 − 11. C2H5(z)+C2H4 C4H9(z) 9.80 10 0.00 60.4 20 θC2H5(z) +13 12. C4H9(z) → C2H5(z)+C2H4 2.30 × 10 0.40 100.4 +13 13. C4H9(z)+C2H4 → C6H13(z) 9.80 × 10 0.20 30.4 +11 14. C6H13(z) → C4H9(z)+C2H4 1.00 × 10 0.00 109.4 +13 15. C6H13(z) → C6H12 +H(z) 9.50 × 10 0.00 127.7 +10 16. C6H12 + H(z) → C6H13(z) 7.00 × 10 0.00 59.3 +13 17. C6H12 + H(z) → C6H11(z)+H2 9.50 × 10 0.20 100.7 → × +11 − 18. C6H11(z)+H2 C6H12 +H(z) 1.00 10 0.00 99.4 35 θC6H11(z) +13 19. C6H12 +C2H5(z) → C6H11(z)+C2H6 9.80 × 10 0.00 68.0 +12 20. C6H11(z)+C2H6 → C6H12 +C2H5(z) 3.00 × 10 0.00 65.4 +13 21. C6H11(z) → cy-C6H11(z) 9.50 × 10 0.00 0.0 +12 22. cy-C6H11(z) → C6H11(z) 9.50 × 10 0.00 0.0 +13 23. cy-C6H11(z) → C6H10 +H(z) 9.50 × 10 0.00 178.0 +11 24. C6H10 + H(z) → cy-C6H11(z) 1.00 × 10 0.00 75.3 +13 25. C4H9(z) → C4H8 +H(z) 2.50 × 10 0.00 109.4 +13 26. C4H8 + H(z) → C4H9(z) 1.00 × 10 0.00 40.4 +13 27. C4H8 +C2H5(z) → C6H13(z) 1.50 × 10 0.00 40.4 +12 28. C6H13(z) → C4H8 +C2H5(z) 2.50 × 10 0.00 80.4 −01 29. C4H8 + H(z) → C4H7(z)+H2 1.50 × 10 0.35 88.3 → × +11 − 30. C4H7(z)+H2 C4H8 +H(z) 1.50 10 0.00 89.4 50 θC4H7(z) +13 31. C6H10 + H(z) → C6H9(z)+H2 9.00 × 10 0.00 129.3 +11 32. C6H9(z)+H2 → C6H10 +H(z) 3.00 × 10 0.00 80.4 +13 33. C6H9(z) → C6H7(z)+H2 9.70 × 10 0.00 122.3 +10 34. C6H7(z)+H2 → C6H9(z) 1.00 × 10 0.00 79.4 +13 35. C6H7(z) → C6H6 +H(z) 9.80 × 10 0.20 207.0 +11 36. C6H6 + H(z) → C6H7(z) 1.00 × 10 0.00 69.3 → × +12 − 37. C6H6 +C4H7(z) C10H13(z) 4.50 10 0.00 90.4 80 θC4H7(z) +11 38. C10H13(z) → C6H6 +C4H7(z) 8.50 × 10 0.00 69.4 +11 39. C10H13(z) → C10H12 +H(z) 1.50 × 10 0.00 0.0 +11 40. C10H12 + H(z) → C10H13(z) 1.50 × 10 0.00 0.0 +11 41. C10H12 → C10H11 +H(z) 1.50 × 10 0.20 112.3 +10 42. C10H11 + H(z) → C10H12 1.50 × 10 0.00 89.4 +10 43. C10H11 → C10H10 +H(z) 1.50 × 10 0.00 0.0 +12 44. C10H10 + H(z) → C10H11(z) 1.50 × 10 0.00 0.0 → × +12 − 45. C10H11(z) C10H9(z)+H2 5.50 10 0.00 112.3 90 θ(z) +11 46. C10H9(z)+H2 → C10H10 +H(z) 1.50 × 10 0.00 89.4 +11 47. C10H9(z) → C10H8 +H(z) 9.50 × 10 0.00 0.0 +11 48. C10H8 + H(z) → C10H9(z) 1.50 × 10 0.00 0.0 → × −05 − 49. CH4 + H(z) CH3(z)+H2 1.00 10 0.00 118.0 90 θH(z) → × +12 − 50. CH3(z)+H2 CH4 +H(z) 1.00 10 0.00 70.0 25 θCH3(z) +13 51. C6H6 + CH3(z) → C7H9(z) 1.00 × 10 0.00 150.0 +11 52. C7H9(z) → C6H6 + CH3(z) 1.00 × 10 0.00 90.1 +13 53. C7H9(z) → C7H8 +H(z) 3.00 × 10 0.00 122.3 +11 54. C7H8 + H(z) → C7H9(z) 1.00 × 10 0.00 89.4
23 Figure 3.2: Reaction scheme for naphthalene decomposition proposed by Devi et al. [70]
are still pathways to create larger hydrocarbons through condensation reactions. There has not been a definite reaction pathway proposed for naphthalene cracking over a Mo/Zeolite catalyst, but zeolites are shown to have naphthalene cracking activity. Buchireddy et al. [72] studied naphthalene steam cracking reactions over multiple types of zeolites (in- cluding H-ZSM-5), while varying acid strength and cage structure. A gas mixture containing
◦ CO, CO2,H2, CH4, C10H8, and H2O at 750 C with a steam to carbon ratio of 5.0, was in- troduced to show that naphthalene cracking is possible on zeolites. The proposed reaction path was
C10H8 → C(s) + gas + Lower Hydrocarbons. (3.4)
Buchireddy et al. [72] proposed that cracking reactions take place in the cage structure of zeolite over the Brønsted acid sites therefore higher acidity increases naphthalene cracking activity. The zeolite cage size also strongly influences the cracking. Naphthalene must be
24 Figure 3.3: Possible reaction scheme for naphthalene hydrocracking proposed by Wang et al. [71] transported inside the zeolite cages, therefore the zeolites with bigger cage size show higher activity. For example, the cage size of H-ZSM-5 is typically around 5.5 A,˚ which had 19% naphthalene conversion whereas the cage size of ZY-30 is 8 A˚ and increased the naphthalene conversion to 33%. Doping zeolites with the active metal Ni significantly increased the naphthalene conversion up to 98%. The steam and dry reforming activity of impregnated nickel allowed for naphthalene to form syn-gas [72],
C10H8 + 10H2O ⇋ 10CO + 14H2, (3.5)
C10H8 + 10CO2 ⇋ 20CO + 4H2. (3.6)
El-Rub et al. [73] studied the naphthalene steam cracking reactions over zeolite catalysts in a packed-bed reactor at the temperature range between 700-800 ◦C. Although species distributions were not measured, naphthalene reaction kinetics were derived in an Arrhenius
25 type of global rate expressions. This study uses the data given by El-Rub et al. [73]. The reaction rate is first order for naphthalene and zero order for water. The cracking products
include the gas-phase species CO, H2, CH4, C9H8, C7H8, C6H6, C2H4 and C14H10. These species are in accordance with the species described by Devi et al. [70] and gas-phase reaction
products described by Jess [74]. The major reaction products are CO, H2, and CH4. The naphthalene cracking reaction used in the current study is
C10H8 + 2H2O →
2CO+1.419H2 +0.5985CH4 +0.0535C9H8 +0.2C7H8+0.45C6H6 +0.01C2H4 +0.2C14H10. (3.7) The reaction rate for this equation is given as
ri = k [XC10H8 ] , (3.8) and the rate coefficient is defined as
−E k = AT β exp a . (3.9) RT −3 −1 −1 where A = 3.5×10 s , β=0, and Ea=55 kJ mol . This reaction assumes that the reaction rate is zero order in water and first order in naphthalene, for conditions when water is in excess [74–76]. It is evident that the reaction kinetic used in this study is preliminary and needs further validation. However, the kinetic proposed in this study captures experimental data [72–74]. Water has the potential to reform methane, but has been ruled unlikely in this context. The methane reforming would occur on the molybdenum site. However, the molybdenum loading on the zeolite is typically quite low (1-10%) therefore the naphthalene cracking chemistry is more likely. Therefore, the only water interaction taken into account in this model is the naphthalene steam cracking reaction.
3.3 Effect of micro- and macroporosity
As mentioned in Section 2.2, porosity in an MDA reactor exists on both the micro- and macroscale. During reactor operation, both quantities may change due to coke formation and
26 sintering. In particular, there is a direct relationship between the deactivation rate and the change in microporosity [77–80]. To test the model sensitivity to macro- and microporosity variations, simulations were compared to experimental results from Sun et al. [77], who ran
long-time stability tests for MDA while periodically switching between CH2 and H2 feeds to regenerate the catalyst, at temperatures ranging from 1033 - 1053 K. Both fresh and deactivated catalysts were characterized ex-situ to determine the changes in microporosity, macroporosity, and BET surface area of the catalysts. Because the physical properties of the catalyst had changed dramatically over operation time, the catalysts were labeled Cat 1 - Cat 5, corresponding to the time on stream (up to 1000 hr). For this study, simulations were run to predict the catalyst performance for Cat 1 and Cat 2. The catalyst properties and experimentally-determined reactor performance are given in Table 3.2.
Sun et al. [77] performed a MDA long time stability experiments with periodic CH4-H2 switching. They reported that the catalytic activity could be regained by regenerating the catalyst with H2. The fresh and deactivated catalyst were characterized to determine the change in microporosity, macroporosity, and BET surface areas of the catalysts. Because the physical properties of the catalyst had changed dramatically, they labelled those catalysts as Cat 1-Cat 5 for given time on stream (1000 h).
16
14 CH4 Conversion 12 Total aromatic yield 10 C yield 8 6 6 Conversion or Yield (%) 1030 1035 1040 1045 1050 1055 Temperature (K)
Figure 3.4: Comparison of model predictions (lines) against experimental data (markers) of Cat 1 from Sun et al. [77]. Effect of temperature on methane conversion, benzene and aromatic yield.
27 Table 3.2: Reactor and catalyst properties used in kinetic modeling of experiments reported by Sun et al. [77]
Parameters Values Pressure 1atm Inletvelocity 1.84-1.87cms−1
Inlet CH4 90% Inlet N2 10% Catalystbedlength 21.2mm Reactordiameter 10mm Tortuosity 2.5 Particlediameter 300 µm Catalystmass 1000mg Catalyst 5.5%Mo/H-ZSM-5 Cat1 Cat2 Time(min) 180-450 550 Temperature(K) 1048 1033 Surface/volume ×106 (m−1) 9.8 4.9 Macroporosity 0.23 0.10 Microporosity 0.05 0.024 BET surface area (m2 g−1) 233 119
Catalyst performance data for Cat 1 and Cat 2 are reproduced. Table 3.2 shows the details of the catalyst and reactor performance. Figure 3.4 compares experimental and model-predicted methane conversion, benzene yield, and aromatic yield for the Cat 1 catalyst as a function of temperature. The Cat 1 data used for comparison was taken between 185 min and 450 min time on stream, directly after regeneration by H2. Because the reaction time is considerably long, Sun et al. measured and reported coke selectivity. For the given time scale, the coke selectivity is reported to be below 20%. The present model does not differentiate the coke and naphthalene. Studies show that most of the surface carbon are PAH species that forms from further cyclization of naphthalene. For comparison, the experimental coke yield is included in the aromatic yield. Nevertheless, the model confirms that methane conversion is strongly dependent on temperature and it can predict benzene and aromatic yield with good agreement.
28 Table 3.3 compares model predictions with experimental data from Sun et al. for reactor performance at 1048 K with varying catalyst macro- and microstructure (i.e. Cat 1 vs. Cat 2 catalysts, as described in Table 3.2). The experimental results demonstrate that catalyst deactivation has a strong influence on the catalyst pore volume (micro and macro), and therefore on the reactor performance. Model predictions and experimental results show close agreement, implying that, given accurate models for catalyst properties as a function of time on stream, the kinetic model can be used to predict the transient evolution of reactor performance with catalyst deactivation.
Table 3.3: Comparison of model predictions against experimental data of Sun et al. [77] at 1098 K. Effect of catalyst micro- and macroporosity on methane conversion, benzene, and aromatic yield.
CH4 Benzene Total Conversion yield aromatic yield Experiment 13.98 8.39 13.30 Catalyst 1 Model 14.54 8.69 13.24 Experiment 12.80 7.75 12.09 Catalyst 2 Model 12.91 7.82 11.42
3.4 Model comparison against experimental data
The detailed reaction mechanism and computational model are validated against pre- viously published experimental observations, which cover a range of operating conditions and catalyst structural properties [77, 81]. The presented model is developed for steady- state isothermal reactor conditions. Therefore, validation comparisons are focused on quasi- steady-state experimental data from the previous reports. In other words, the reference point for time is immediately after the catalyst is carburized and the deactivation rate is considerably slow (Figure 1.3). The experimental observations were generally reported in terms of outlet gas-phase com- positions, methane conversion, and selectivity to major products. Although toluene and ethane are known side products in trace amounts, most studies did not report their concen-
29 trations in detail [77–79]. The present study uses methane conversion, in addition to selec- tivities and yields for both benzene and naphthalene as basic comparison metrics. Methane conversion is evaluated as
JCH4,in − JCH4,out XCH4 = , (3.10) JCH4,in
where JCH4,in and JCH4,out are the inlet and outlet methane molar fluxes. The carbon selec- tivity of product species k is evaluated as
Cn,kJk Sk = , (3.11) JCH4,in − JCH4,out
where Cn,k is the carbon number of species k (e.g., Cn=6 for C6H6) and Jk is the species outlet molar flux. The product yields are evaluated as
Yk = 100XCH4 Sk. (3.12)
The membrane effectiveness is reported as the ratio of hydrogen removed through the mem- brane to the hydrogen exiting the end of the packed bed
m˙ H2,M %H2 removed = , (3.13) m˙ H2,M +m ˙ H2,out
wherem ˙ H2,M is the total mass flow rate of H2 through the membrane andm ˙ H2,M is the packed
bed outlet mass flow rate of H2. Another important metric is the ratio of the selectivity toward benzene to the selectivity toward PAH species S Benzene/PAH ratio = C6H6 . (3.14) SC10H8 + SC14H10 When water is not included in the chemistry, naphthalene is the only PAH species considered, since anthracene formation is not included in the mechanism.
3.5 Typical solution
Figure 3.5 shows predicted gas-phase axial profiles within a typical MDA packed-bed reactor compared with experimental data from Korobitsyna et al. [81]. The experimental conditions are summarized in Table 3.4. The 13 mm long reactor operates at 1023 K and
30 atmospheric pressure. The catalyst is 4 wt% Mo/H-ZSM-5 catalyst. The inlet stream is
−1 99.99% CH4 (balanced by N2) at a velocity of 3.5 mm s .
10 0 CH4 CH4 -1 10 H2
C6H6 -2 C6H6 10 C10H8 C10H8
-3 C2H4 C H C7H8 2 4 Mole fraction 10
C H C2H6 10-4 2 6 0 2 4 6 8 10 12 Position in bed (mm)
Figure 3.5: Comparison of model predictions with the experimental data of Korobitsyna et al. [81]. The measured outlet mole fraction are shown as diamonds on the right-hand sides of the plots.
Table 3.4: Reaction conditions reported by Korobitsyna et al [81].
Parameters Values Temperature 1023K Pressure 1atm Inletvelocity 3.54mms−1
Inlet CH4 99.99% Inlet N2 0.01% Catalystbedlength 13mm Reactordiameter 10mm Macroporosity,% 0.40 Microporosity,% 0.05 Tortuosity 2.5 Particlediameter 7.5 × 10−1 mm Surface area to volume ratio 1.8×107 m
31 Fig. 3.5 compares the axial species profiles, as predicted by the packed bed model devel- oped in this work, including the outlet species concentrations, as measured by Korobitsyna
et al. [81]. Results show good agreement for both major (CH4, C6H6, C10H8) and minor
(C2H4, C2H6) gas-phase species, including a CH4 conversion of ca. 13%. In addition to the listed species, the model predicts significant amounts of H2 and trace amounts of C7H8, both of which were not monitored by Korobitsyna et al. [81].
32 CHAPTER 4 DRY MDA RESULTS
The implementation of hydrogen-selective membranes in MDA processes is first explored without the effect of water addition. In these cases, naphthalene is the primary species contributing to catalyst deactivation and the only species under the PAH label. The major variables and parameters in the results shown are summarized in Table 4.1.
Table 4.1: Major variables and parameters for dry MDA simulations
Dimensions Length 4.0 cm AnnulusInnerDiameter 1.0 cm AnnulusOuterDiameter 1.6 cm Inlet Temperature 675 - 750 ◦C Pressure 1 atm 0.95 CH Mole Fractions 4 0.05 N2 Catalyst Macroporosity 0.30 Microporosity 0.05 MolybdenumLoading 6 % Specific Area 9.4×10+6 m−1 Pore Radius 2.82×10−6 m Particle Diameter 7×10−4 m
4.1 No hydrogen removal
The model was used to explore the effect of operating conditions (GHSV and reactor temperature) on the performance of an MDA packed-bed reactor without hydrogen mem- brane coupling. For theses cases, the gas-hourly space velocity was typically varied between
−1 −1 750 ≤ GHSV ≤ 25000 mL gcat hr .
33 Figures 4.1, 4.2, and 4.3 show the predicted effects of varying GHSV and reactor tem- perature on methane conversion, benzene yield, the C6H6/C10H8 yield ratio, and benzene selectivity. The individual symbols in Figures 4.2 and 4.3 correspond to the specific GHSV values from Figure 4.1. Taken together, the figures convey the great difficulty in optimizing reactor operating conditions. Large GHSV lead to low conversion and high benzene selec- tivity, but low GHSV leads to fast catalyst deactivation. Typical operating conditions for MDA include high temperatures near 700 ◦C and low gas hourly space velocities between
−1 −1 750 - 1500 mL gcat hr to assure high methane conversion, but the benzene yield at these conditions are fairly low.
16 14 750°C 12 725°C 10 8 700°C conversion (%) 4 6 675°C CH 4 5001500 2500 3500 4500 5500 6500 7500 -1 -1 GHSV (mL gcat hr )
Figure 4.1: Effect of GHSV and reactor temperature on methane conversion rate for an MDA reactor with no membrane.
Increasing the temperature increases the thermodynamic limit of the methane conver-
sion and enables faster kinetic rates toward this limit, resulting in higher CH4 conversion (Figures 4.1 and 4.2). There is a simultaneous decrease in the selectivity toward benzene, as seen in Figure 4.2. Additionally, increasing reactor temperatures increase the likelihood of catalyst fouling via coking. Although the current mechanism does not incorporate coking reactions, the two ring aromatic naphthalene is generally accepted as the monomer of carbon- deposit-forming PAH species. Hence, increased naphthalene formation rate is used in this study as an indication that the reactor is operating within a catalyst fouling regime. For long
34 and stable operation times, the naphthalene formation rate and the concentrations should be low, resulting in high C6H6 to C10H8 yield ratios. As shown in Figure 4.3, increasing temperature increases benzene yield, and the formation of undesired products naphthalene and PAH become more favored.
Increased GHSV 75 675°C 70 700°C 725°C 65 750°C Selectivity (%)
6 60 H 6
C 55 468 10 12 14 16 CH4 conversion (%)
Figure 4.2: Effect of GHSV on benzene selectivity and methane conversion an MDA reactor with no membrane. Increase or decrease in methane conversion corresponds to specific GHSV −1 −1 values between 750-25000 mL gcat hr as it is shown in Figure 4.1.
Although decreasing the GHSV increases the methane conversion (Figure 4.1), the im- pacts on benzene selectivity, yield, and propensity for catalyst coking (Figure 4.2 and 4.3) are more complex. As shown in Figure 4.2, for each given temperature there appears an optimum GHSV, corresponding to the maximum benzene selectivity. This is due to com- peting mass transport limitations on the formation of naphthalene and benzene. As GHSV increases from low to moderate values, the naphthalene formation from benzene becomes transport limited. Benzene diffusion becomes much faster than the naphthalene formation rate, and the ultimately reaction favors benzene formation, which appears as a local maxima in the benzene selectivity. Further increasing the GHSV decreases the methane conversion rate as well as the benzene selectivity, as the benzene formation rate also becomes transport limited.
35 Increased GHSV 9 8 750°C 7 725°C 6 700°C 5 4 675°C Benzene yield (%) 3 1234567
C6H6/C10H8 yield ratio
Figure 4.3: Effect of GHSV and reactor temperature on benzene yield for an MDA reactor with no membrane. Each data point on benzene to naphthalene ratio corresponds to the −1 −1 specific GHSV between 750-7500 mL gcat hr as it is shown in Figure 4.1.
The transitions between reaction rate controlled regimes and transport controlled regimes differ for each temperature range. At low temperatures, the transition is sharper. A slight increase in the GHSV values can easily transition between kinetically controlled and trans- port controlled regimes. Increasing temperature can offer a wider range of GHSV values in the non-transport controlled regime. Under the given catalyst properties and reaction con- ditions, the optimum sensitivity at 675 ◦C corresponds to GHSV in the range of 3000 - 4500
−1 −1 ◦ −1 −1 ◦ mL gcat hr . At 700 C, the range expands to 3000 - 6000 mL gcat hr , at 725 C, the
−1 −1 ◦ range is 3000 - 9000 mL gcat hr , and finally at 750 C, the optimum selectivity range is at
−1 −1 4500 - 12000 mL gcat hr . Figure 4.3 highlights the challenges involved in optimizing MDA reactor operating condi- tions. The conditions with the highest benzene yields coincide with high carbon deposition rates (indicated by C6H6/C10H8 ratios below ≈ 2.4). At high GHSV values, the C6H6/C10H8 ratio is high and the reactor operates at reduced coke production conditions, but the reaction becomes transport limited. Therefore, methane conversion and benzene yields are low. At high temperatures, methane conversion and hence benzene yields increase, but the reaction heavily favors naphthalene production, resulting in significant coking and catalyst deacti-
36 ◦ −1 −1 vation. Even at very high temperatures (750 C) and low GHSV (750 mL gcat hr ), the maximum benzene yield remains below 9%. Based on Figures 4.2 and 4.3, MDA reactors should be operated at low temperatures (i.e., below 725 ◦C for long time operating hours). However, at temperatures below 725 ◦C, the methane conversion and consequently the ben- zene yield is too low to meet the economical expectations. The compromise between high benzene yield and catalyst stability is therefore a major issue for scale up.
4.2 Hydrogen membrane
Figure 4.4 shows the effects of hydrogen removal on MDA reactor performance. Methane conversion and major species yields are plotted as a function of the percent of hydrogen
◦ −1 −1 removed, for a reactor operating at 700 C and 1500 mL gcat hr . Hydrogen removal increases the methane conversion as well as naphthalene and benzene yields. The increase in both naphthalene and benzene yields follow a linear trend below 60% of hydrogen removal, where the methane conversion is below 20%. Further removal of hydrogen significantly increases methane conversion, which causes the product distribution to significantly favor naphthalene. At very high hydrogen removal rates, benzene formation drops dramatically and the reaction produces only naphthalene.
80 70 60 50
40 CH4 conversion 30 C6H6 yield 20 C10H8 yield 10
Conversion or yield (%) 0 0 20 40 60 80 100 H2 removal (%)
◦ Figure 4.4: Effect of hydrogen removal on CH4 conversion and product yield. T = 700 C, −1 −1 GHSV = 1500 mL gcathr
37 Figures 4.5a and 4.5b show that trends in Figure 4.4 are reproduced for a wider temper-
◦ −1 −1 ature range (675-750 C) and a GHSV of 1500 mL gcat hr . High hydrogen removal rates favor naphthalene formation. Even though very high benzene yields (≈ 15%) are achieved at ≥ 90% hydrogen removal, the corresponding naphthalene yield reaches ≈ 40%. At high hydrogen removal rates, the catalyst deactivation via PAH formation is highly probable.
It is well accepted that the coke formation rates are related to the H2 concentration [41]. Graphitic coke as well as PAH, often form at low hydrogen concentration ranges.
a) 100 80 750°C 60 725°C 40 700°C
conversion (%) 675°C 4 20 CH 0 0 20 40 60 80 100 b) H2 removal (%) 20 80 750°C 725°C 60 15 700°C 675°C 10 40 750°C 725°C 5 20 675°C 700°C
Benzene yield (%) 0 0 Napthalene yield, % 0 20 40 60 80 100 H2 removal (%)
Figure 4.5: Effect of hydrogen removal on (a) Methane conversion, (b) Benzene and naph- thalene yields at temperatures 675 - 750 ◦C. GHSV values for all temperature points are −1 −1 1500 mL gcat hr .
Overall reactor performance is typically based on desired product benzene yield. Yield is an important comparison parameter, but does not solely determine the reactor performance. Another important parameter is the product selectivity. In Figure 4.5b, although the benzene yield increases, the selectivity is decreases below 10%, and the methane’s value is wasted to
38 form the unwanted product naphthalene. For the case in which the benzene yield is low but selectivity is high, the methane conserves its value as a fuel. Although the single-pass yield is low, multi-pass stage reactors can be designed to further convert methane to benzene. Another important parameter is the benzene formation rate. Typically, due to low inlet GHSV values, the benzene formation rates are low (≤ 1 µmol s−1). Figure 4.6 shows the
−1 −1 predicted benzene formation rates at 1500 mL gcat hr for temperatures between 675 and 750 ◦C, while varying hydrogen removal ratio. At each hydrogen removal rate, the benzene
formation rates is plotted against the C6H6/C10H8 yield ratio. Figure 4.6 shows that the benzene formation rate can increase up to 4 times with membrane integration. However, in
the regimes with the highest benzene formation rates, C6H6/C10H8 yield ratios are below 0.5 where primarily naphthalene is produced. Coke formation is inevitable in this range. The trend is similar for all temperatures studied, indicating that membrane-coupled MDA can not run efficiently at low GHSV values.
Increased H2 removal ratio )
-1 0.40 0.35 675°C
mol s 700°C
μ 0.30
( 725°C 0.25 750°C 0.20 0.15 PAH formation
formation region 6 0.10 H 6 0.05
C 012345 C6H6/C10H8 yield ratio
Figure 4.6: Effect of hydrogen removal on benzene formation and product distribution at a −1 −1 GHSV of 1500 mL gcat hr
−1 −1 Figure 4.7 displays the effect of increasing GHSV at (a) 4500 mL gcat hr and (b) 6000
−1 −1 mL gcat hr . The effect of membrane integration on benzene yield and formation rate is more pronounced at lower GHSV values. Increasing gas-hourly space velocity decreases the
benzene yield, but increases the maximum C6H6/C10H8 yield ratios as well as increasing the
39 benzene formation rates. Figure 4.6, 4.7a, and 4.7b show that, independently of temperature
and GHSV, membrane integration decreases the C6H6/C10H8 yield ratios, due to the decrease in benzene selectivity with hydrogen removal. Figure 4.7a and 4.7b also show that reactor efficiency is higher at temperatures ≤ 700 ◦C. At lower temperatures, the benzene yield decreases, but the reactor can run for much longer time periods due to the reduced coking potential while maintaining reasonable benzene production rates.
-1 -1 a) 4500 mL gcat hr Increased H2 removal ratio )
-1 1.4 14 Dashed lines: Yield 1.2 12
mol s Solid lines: Rate
μ 1.0 10 ( 675°C 0.8 700°C 8
725°C yield (%) 0.6 6 6 H
750°C 6
4 C
formation 0.4 6 H
6 0.2
C 012 3 4 5 6 7 C6H6/C10H8 yield ratio -1 -1 b) 6000 mL gcat hr )
-1 1.4 14 1.2 12 mol s
μ 1.0 ( Non-coking regime 10 0.8 8 yield (%) 0.6 6 6 H 6 C
formation 0.4 4 6 H 6 0.2 2
C 012 3 4 5 6 7 C6H6/C10H8 yield ratio
Figure 4.7: Effect of GHSV on C6H6 yield and formation rates in a membrane integrated −1 −1 −1 −1 MDA reactor. (a) 4500 mL gcat hr and (b) 6000 mL gcat hr
Taking benzene yield, formation rate, and selectivity as decision criteria, this study sug- gests that, for a membrane-coupled MDA reactor with the given reactor and catalyst prop-
◦ −1 −1 erties, a suitable operating case is at 700 C and 6000 mL gcat hr . Figure 4.8 provides a comparison between common GHSV and the suggested operating conditions by this study.
−1 −1 Typical gas-hourly space velocities for lab-scale experiments are near 750 mL gcat hr . The low GHSV leads to high benzene yields but suffers from poor selectivity to benzene, causing
40 −1 −1 faster catalyst deactivation. Increasing the GHSV to 6000 mL gcat hr results in higher ben- zene yields within non-coking regimes and more than double the benzene formation rates.
At the safer operating range, where the C6H6/C10H8 yield ratio ≥ 2.5, the benzene yield
◦ −1 −1 is around 4%. At 700 C and 6000 mL gcat hr , where the C6H6/C10H8 ratio is 3, the methane conversion is around 9%, benzene selectivity is 65%, and naphthalene selectivity is 20%. Hydrogen removal from the reactor is 50% with an average hydrogen removal rate of 1.6 µmol s−1.
) Increased H2 removal ratio -1 1.0 15 -1 -1 750 mL gcat hr -1 -1
mol s 0.8 6000 mL gcat hr μ ( 10 0.6 6000 mL g -1 hr-1 cat yield (%) 0.4 6
5 H 750 mL g -1 hr-1 6
cat C formation 0.2 6 Non-coking regime H 6 0 0 C 012345 C6H6/C10H8 yield ratio
Figure 4.8: Comparison of effect of GHSV on C6H6 yield and formation rate in a membrane integrated MDA reactor. Darker region indicates the desired operating range.
Designing a reactor for MDA would require high flow rates and low hydrogen removal rates. At theses conditions, conversion would be low, but potential to coke is significantly reduced, which would increase operation time. Multi-pass stage reactors could overcome some of the conversion issues. However, the benzene output from each individual reactor would be small.
41 CHAPTER 5 WET MDA RESULTS
To explore how steam addition and hydrogen removal affect MDA process, 2 percent water was added to the inlet stream. The variables and parameters in Table 4.1 still apply, but the nitrogen mole fraction was decreased to 3% to compensate for the addition of steam. Table 5.1 shows the updated variables and parameters.
Table 5.1: Major variables and parameters for wet MDA simulations
Dimensions Length 4.0 cm AnnulusInnerDiameter 1.0 cm AnnulusOuterDiameter 1.6 cm Inlet Temperature 700 - 750 ◦C Pressure 1 atm
0.95 CH4 Mole Fractions 0.03 N2 0.02 H2O Catalyst Macroporosity 0.30 Microporosity 0.05 MolybdenumLoading 6 % Specific Area 9.4×10+6 m−1 Pore Radius 2.82×10−6 m Particle Diameter 7×10−4 m
5.1 No hydrogen removal
Steam was added to the catalytic side of the reactor to mediate the PAH formation. To satisfy the zero-order naphthalene cracking kinetic, steam is premixed with the inlet CH4/N2 stream, such that the water is in excess relative to the naphthalene. Additionally, the gas-
42 hourly space velocities are sufficiently fast such that the water is never depleted. The water concentration never drops below stoichiometric values for the naphthalene cracking reaction.
75
70
65
60 700°C 725°C 750°C No H2O H2O
Benzene selectivity (%) 55 0 2 4 6 8 10 12 14 16 -1 -1 GHSV (L gcat hr )
Figure 5.1: Effect of water addition on benzene selectivity for varying GHSV between 750 - −1 −1 15000 mL gcat hr
Figure 5.1 shows the effect of water addition on the benzene selectivity, while changing the gas-hourly space velocity. At large GHSV, the naphthalene concentration is small, which limits the naphthalene cracking reaction rate since the reaction rate is heavily dependent on naphthalene concentration. As the GHSV decreases, naphthalene production increases, and likewise the cracking reaction rate. The effect of water is more evident at low GHSV. At the higher temperatures, the benzene selectivity increases more than 10 percentile. Increasing the GHSV weakens the effect of water, and asymptotically approaches the case that doesn’t include water. Steam addition has a noticeable effect on benzene selectivity, but weakly affects the methane conversion. Figure 5.2 shows that across several GHSV, the conversion is nearly identical to the conditions without water. At low GHSV, the methane conversion decreases when water is added. The naphthalene cracking reaction produces a significant amount H2 for every mole of reactant. Once again considering Le Chˆatelier’s principle, the increase of product hydrogen, pushes the MDA kinetics back toward the reactants. Additionally, reactant methane is also produced, which decreases the conversion. Small changes in conver-
43 15 700°C 725°C 750°C No H2O H2O 10 Conversion (%)
50 2 4 6 8 10 12 14 16 -1 -1 GHSV (L gcat hr )
Figure 5.2: Effect of water addition on methane conversion for varying GHSV between 750 −1 −1 - 15000 mL gcat hr sion signify that the effect of the water is primarily associated with adjusting the aromatic selectivity more than the overall conversion.
Increased GHSV 40 O 35 2 O 2 H 30 No H 700°C 25 725°C 750°C 20 15 PAH Selectivity (%) 10 5 10 15 Conversion (%)
Figure 5.3: Effect of water addition on PAH selectivity for varying GHSV between 750 - −1 −1 15000 mL gcat hr
With the addition of steam, higher-order aromatics have been included into the gas-phase species set. A prominent PAH that can form from the naphthalene cracking reaction is an- thracene (C14H10). Both naphthalene and anthracene are indicative of PAH species that would deactivate the catalyst. Figure 5.3 shows how the selectivity toward PAH species changes with the methane conversion. Without water, the PAH production increases al-
44 most linearly with conversion. When water is included, the selectivity toward PAH species matches the no-water case at low conversions. However, at high conversion, the cases that include water reach a maximum and actually begin to decrease. At the maximum, the naphthalene cracking chemistry is equally effective as the MDA chemistry in terms of PAH formation. Increasing the conversion shifts an increasing amount of the naphthalene into carbon monoxide, methane, and benzene, which ultimately reduces the selectivity toward the PAH species. Limiting PAH formation is critical to extending operation time. Wa- ter addition allows for MDA processes to reach higher conversions without worrying about excessive PAH formation.
Increased GHSV 6 5 O
4 2 O 2 H 3 No H 700°C 2 725°C 750°C 1 Benzene/PAH yield ratio 5 10 15 Conversion (%)
Figure 5.4: Effect of water addition on benzene to PAH yield ratio for varying GHSV between −1 −1 750 - 15000 mL gcat hr
The ratio between yield of benzene to PAH species has a local minimum when water is added. In Figure 5.4, the benzene to PAH species ratio initially decreases with increased conversion, due to the increased naphthalene formation from MDA. As the GHSV decreases or conversion increases, naphthalene formation becomes more favored. When the benzene to PAH species yield ratio reaches a minimum, the naphthalene formation rate from MDA begins to balance with the naphthalene steam cracking rate. The increased naphthalene con- centration accelerates the cracking reaction and converts some of the PAH species to lighter carbonaceous species, which decreases the benzene to PAH species yield ratio. Operation in
45 non-coking regimes is more likely with smaller benzene to PAH species yield ratios.
) Increased GHSV -1 1.5 O 2 mol s O 2 μ H 1.0 700°C No H 725°C 750°C 0.5
0 5 10 15 Benzene formation ( Conversion (%)
Figure 5.5: Effect of water addition on benzene formation rate
The molar production of benzene has a very slight increase with water addition. Fig- ure 5.5 shows that benzene production with water addition matches the dry case almost exactly for varying GHSV. Increasing the conversion through GHSV increases naphthalene formation. The increased concentration of naphthalene drives the naphthalene steam crack- ing reaction to decrease the PAH selectivity. However, benzene is one of the minor produc- tions from the cracking reaction. Therefore, there are no major benefits in terms benzene formation rates with water addition. The major benefit of water addition comes from the decreased PAH selectivity.
5.2 Hydrogen Membrane
The combination of water addition and hydrogen removal in MDA processes can increase both conversion and selectivity toward benzene. The permeance of the membrane was ad- justed between 1×10−7 and 1×10−4 mol m−2 s−1 Pa−0.5 to vary the flux of hydrogen through the membrane. Permeances greater than 1×10−4 mol m−2 s−1 Pa−0.5 were not explored in order to keep water concentration in excess relative to naphthalene. A moderate GHSV of
−1 −1 1500 mL gcat hr was used for the following results.
46 Increased H2 Removal 80 70 Benzene Selectivity 60 50 No H2O 40 H2O 30 Selectivity (%) 20 PAH Selectivity 10 5 10 15 20 25 30 35 40 Conversion (%)
Figure 5.6: Selectivity of both benzene and PAH species for wet and dry conditions at a −1 −1 GHSV of 1500 mL gcat hr and 973 K
Similar to dry MDA, hydrogen removal also has a detrimental effect on the benzene selec- tivity. Figure 5.6 shows the effect of steam addition on the selectivity of benzene and PAH species as membrane performance increases. Increasing the conversion through hydrogen removal decreases benzene selectivity and increases selectivity toward catalyst-deactivating PAH species. For dry MDA, the selectivity toward benzene and PAH species are near 28% hydrogen removal. The addition of water pushes the crossover between benzene and PAH selectivity to near 40%. Thus, wet MDA processes can achieve better performance than dry MDA processes, and the advantage grows as more hydrogen is extracted. However, extracting too much hydrogen can push the process into coking regimes. Figure 5.7 illustrates the advantage of combining both water addition and hydrogen removal. As shown previously, the benzene production rates are weakly affected by water addition. However, the major benefit comes from the improved benzene to PAH species yield ratio. Increasing hydrogen removal for the water addition case increases the benzene production rates while still operating in a non-coking regime. Too much hydrogen removal increases the benzene formation rates nearly 5 times, but stretches into a highly coking regime, which is not ideal for sustained operation.
47 ai taGS f10 Lg be mL on 1500 removal of hydrogen GHSV and a addition at water ratio of Effect 5.8: Figure ra formation benzene on removal g mL hydrogen 1500 of of GHSV Effect 5.7: Figure ogroeaintmsfrbneeproduction. benzene re downtime for times decrease would operation longer range this regim in low-coking Operation in 20%. operate near still can aro reactor conversions the at addition, operate must P to addition benzene steam same without the case reach s To PAH to increases. removal benzene hydrogen the 5.8, as Figure In larger. is regime non-coking h ezn oPHseisrtoicesswt ta additio steam with increases ratio species PAH to benzene The
cat − -1 Benzene/PAH yield ratio 1 Benzene formation (μmol s ) 0.1 0.2 0.3 0.4 0.5 0.6 0 1 2 3 4 5 6 hr 02 04 060 50 40 30 20 10 0 − 6 5 4 3 2 1 0 1 cat − 1 hr Increased H Benzene/PAH yieldratio − Increased H 1 Conversion (%) 48 2 Removal 2 Removal No H No H H H 2 2 2 2 s u nrae h conversion the increases but es, O O O O
n 0 oe.Wt steam With lower. 10% und 700°C 700°C urdfrrgnrto and regeneration for quired zn oPHseisyield species PAH to nzene Hseisyedrto the ratio, yield species AH eisyedrtodecreases ratio yield pecies 725°C 725°C ,adidctsta the that indicates and n, 750°C for conditions wet in te 750°C Hydrogen membrane addition to wet MDA processes can increase the benzene yield and can increases benzene production rates while still operating in safe, non-coking regimes. Water addition primarily attacks the PAH formation, which translates into increased benzene selectivity and higher benzene to PAH species yield ratios. The hydrogen removal increases the conversion to reach higher production rates. Based on these results for water addition and hydrogen removal, the high flow rates, low hydrogen removal rates benefit non-coking benzene production.
49 CHAPTER 6 SUMMARY AND CONCLUSIONS
This study explored the effects of hydrogen-selective membranes for both wet and dry MDA processes. The modeling efforts in this study simulated the quasi-steady-state per- formance of MDA for varying permeability of hydrogen-selective membranes, and proposed operating conditions to improve operation.
6.1 Summary of efforts
A detailed kinetic model was developed to simulate methane dehydroaromatization in a packed-bed annular reactor with a bifunctional Mo/H-ZSM-5 catalyst. The model allowed for various dimensions, inlet conditions, membrane permeances, and catalyst properties. The model was optimized to increase computational efficiency through better jacobian formation and communication with external software. The detailed chemical mechanism presented by Karakaya et al. [2] was improved to include microscopic reversibility on the molybdenum site. Hydrogen removal can strongly affect the thermodynamic equilibrium so microscopic reversibility becomes more important. Since thermodynamic properties for most of the surface species are unknown, forward and reverse reaction rates for the heterogenous chemistry can be determined by a weighted least- squares fit. Model fidelity is demonstrated by close agreement with previously published experimental results. The transport mechanisms inside the packed-bed model were extended to include the microporosity of the catalyst pellets. Molecules must travel through the space between the catalyst pellets as well as through the catalyst pellets to reach the active catalyst sites. The random-pore model included both macroporosity and microporosity into an effective diffusion coefficient.
50 The effect of hydrogen removal on MDA reactor performance was determined by changing the permeances of a Sieverts-law membrane in the reactor. Membrane performance deter- mined hydrogen flux out of the packed bed. The effect of water addition on MDA reactor performance was determined by premixing 2 mole percent of water into the CH4/N2 feed. Conversion was changed by adjusting GHSV or membrane permeance.
6.2 Key findings
For dry MDA and no hydrogen removal, benzene production is fairly limited. Short residence times and low temperatures can increase selectivity to benzene, but the yield is still too low for commercialization. When a hydrogen-selective membrane is added, conversion is can be increased, but the benzene selectivity decreases. The benzene yield increases but at a much slower rate than the naphthalene yield. The molar production rate of benzene does reach a maximum, however it is in a heavily coking regime. Wet MDA improved the reaction selectivity toward benzene. Water addition had little effect on the conversion, but some naphthalene converted back into hydrogen, benzene, and ethylene. Benzene selectivity still decreases with flow rate, but at low flow rates, the water addition increases the selectivity to benzene nearly 10 percentile. Contrary to the dry case, in MDA with water addition, the selectivity to PAH species decreases with increasing conversion. The combination of water addition and hydrogen removal allows for increased conversion and selectivity. Water addition improves both the selectivity to benzene as well as the benzene to PAH species yield ratios. Selectivity to benzene still decreases as more hydrogen is removed, but the water addition decreases this effect. Hydrogen removal increases the conversion rates, and likewise the benzene production rates. Using both water addition and hydrogen removal allows for increased benzene production while still operating in non-coking regimes. Optimum operating conditions for membrane coupled MDA reactors are at temperature
◦ −1 −1 regimes ≤ 700 C and high flow rates 4500 ≤ GHSV ≤ 7500 mL gcat hr . Two percent water
51 should be added to reduce the likelihood of coking and increases selectivity to benzene. Small hydrogen removal rates can improve benzene yield by increasing conversion.
6.3 Future Work
The current naphthalene steam cracking reaction is limited to a first-order kinetic rate. A water transport membrane cannot be implemented yet because water may interact differ- ently at low concentrations. Additional reactions may occur when the water is not in excess. Further developing the MDA mechanism to include water interactions would open up op- portunities for transporting water across a membrane into the catalyst side. As mentioned before, a mixed ionic-electronic conducting ceramic could be implemented to allow for both hydrogen and water transport across the membrane, down the length of the packed-bed. However, current knowledge of the water chemistry limits the modeling advancements. The current model allows for changes in both microporosity and macroporosity, but how each changes over operation time isn’t well known. As the catalyst deactivates, both porosities decrease, but it is not clear the rate at which each changes. Knowing the how the catalyst deactivates in terms of porosity and available surface sites, would allow for full transient simulation on MDA. A transient simulation of MDA with deactivation would allow for optimization of operating conditions to increase operation time and decreases regeneration time.
6.4 Outlook
For MDA to be an commercially viable process, the sale of benzene must overcome many costs. MDA occurs at high temperatures, which requires specific material requirements and safety protocols. Additionally, the product stream is not pure benzene. Further purification of products would likely increase costs. Synthesis of the catalysts also accrue costs. The hydrogen also has potential to be sold, however the sweep stream must be carefully chosen to allow for easy purification. MDA is still tough to visualize as commercially viable process with the current low yields and production rates. Major improvements for MDA scale-up
52 will likely come from an improved catalyst. A new catalyst will need to better promote benzene production and avoid forming catalyst-deactivating species. The present study did not thoroughly explore MDA in terms of economy. Instead, based upon catalyst properties and operating conditions, model results help guide reactor design and operation.
53 REFERENCES CITED
[1] S. Ma, X. Guo, L. Zhao, S. Scott, and X. Bao. Recent progress in methane dehydroarom- atization: From laboratory curiosities to promising technology. J. Energy Chem., 22: 1–20, 2013.
[2] C. Karakaya, H. Zhu, and R.J. Kee. Kinetic modeling of methane dehydroaromatization chemistry on Mo/Zeolite catalysts in packed-bed reactors. Chem. Engr. Sci., 123:474– 486, 2015.
[3] X. Guo, G. Fang, G. Li, H. Ma, H. Fan, L. Yu, C. Ma, X. Wu, D. Deng, M. Wei, D. Tan, R. Si, S. Zhang, J. Li, L. Sun, Z. Tang, X. Pan, and X. Bao. Direct, nonoxidative conversion of methane to ethylene, aromatics, and hydrogen. Science, 344:616–619, 2014.
[4] S. Majhi, P. Mohanty, H. Wang, and K.K. Pant. Direct conversion of natural gas to higher hydrocarbons: A review. J. Energy Chem., 22:543–554, 2013.
[5] K.S. Wong, J.W. Thybaut, E.Tangstad, M.W. Stocker, and G.B. Marin. Methane aromatization based upon elementary steps: Kinetic and catalyst descriptors. Micropor. Mesopor. Mat., 164:302–312, 2012.
[6] J.-P. Tessonnier, B. Louis, S. Rigolet, M.J. Ledoux, and C. Pham-Huu. Methane dehydro-aromatization on Mo/ZSM-5: About the hidden role of Brønsted acid sites. Appl. Catal. A, 336:79–88, 2008.
[7] L. Wang, L. Tao, M. Xie, G. Xu, J. Huang, and Y. Xu. Dehydrogenation and aromati- zation of methane under non-oxidizing conditions. Catal. Lett., 21:35–41, 1993.
[8] J. Yang, F. Deng, M.J. Zhang, Q. Luo, and C.H. Ye. W/HZSM-5 catalyst for methane dehydroaromatization: a multinuclear MAS NMR study. J. Mol. Catal. A: Chem, 202: 239–246, 2003.
[9] L.-Q. Huang, Y.-Z. Yuan, H.-B. Zhang, Z.-T. Xiong, J.-L. Zeng, and G.-D. Lin. Dehydro-aromatization of CH4 over W-Mn(or Zn, Ga, Mo, Co)HZSM-5(or MCM-22) catalysts. Stud. Surf. Sci. Catal., 147:565–570, 2004.
[10] D.D. Anggoro and I. Istadi. Optimization of methane conversion to liquid fuels over W-Cu/ZSM-5 catalysts by response surface methodology. J. Nat. Gas Chem., 17:39–44, 2008.
54 [11] M.R. Toosi, B. Sabour, T. Hamuleh, and M.H. Peyrovi. Methane dehydroaromatization over Mo and W catalysts supported on HZSM-5: the effect of preactivation and use of the CH4/H2cycle. Reac. Kinet. Mech. Cat., 101:221–226, 2010.
[12] D. Wang, J.H. Lunsford, and M.P. Rosynek. Catalytic conversion of methane to benzene over Mo/ZSM-5. Top. Catal., 3:289–297, 1996.
[13] A.K. Aboul-Gheit and A.E. Awadallah. Effect of combining the metals of group VI supported on H-ZSM-5 zeolite as catalysts for non-oxidative conversion of natural gas to petrochemicals. J. Nat. Gas Chem., 18:71–77, 2009.
[14] H. Liu, S. Yang, J. Hu, F.P. Shang, Z.F. Li, C. Xu, J.Q. Guan, and Q.B. Kan. A comparison study of mesoporous Mo/H-ZSM-5 and catalysts in methane non-oxidative aromatization. Fuel Process. Technol., 96:195–202, 2012.
[15] Y.-H. Kim, R.W. Borry, and E. Iglesia. Genesis of methane activation sites in mo- exchanged H-ZSM-5 catalysts. Micropor. Mesopor. Mat., 35-36:495–509, 2000.
[16] A.V. Kucherov. Effect of the formation of secondary pores in zeolite ZSM-5 on the properties of molybdenum-zeolite catalysts for methane aromatization. Russ. J. Phys. Chem. A, 88:386–392, 2014.
[17] J. Hu, S.L. Wu, Y.Y. Ma, X.Y. Yang, Z.F. Li, H. Liu, Q.S. Huo, J.Q. Guan, and Q.B. Kan. Effect of the particle size of MoO3 on the catalytic acyivity of Mo/ZSM-5 in methane non-oxidative aromatization. New J. Chem., 39:5459–5469, 2015.
[18] D. Ma, Y.Y. Shu, X.W. Han, X.M. Liu, Y.D. Xu, and X.H. Bao. Mo/HMCM-22 catalysts for methane dehydroaromatization: A multinuclear MAS NMR study. J. Phys. Chem. B, 105:1784–1793, 2001.
[19] D. Ma, Q.J. Zhu, Z.L. Wu, D.H. Zhou, Y.Y. Shu, Q. Xin, Y.D. Xu, and X.H. Bao. The synergic effect between Mo species and acid sites in Mo/HMCM-22 catalysts for methane aromatization. J. Phys. Chem. Chem. Phys., 7:3102–3109, 2005.
[20] N.B. Chu, J.Q. Wang, Y. Zhang, J.H. Yang, J.M. Lu, and D.H. Yin. Nestlike hol- low hierarchical MCM-22 microspheres: Synthesis and exceptional catalytic properties. Chem. Mater., 22:2757–2763, 2010.
[21] X.Y. Yin, N.B. Chu, J.H. Yang, J.Q. Wang, and Z.F. Li. Synthesis of the nanosized MCM-22 zeolite and its catalytic performance in methane dehyro-aromatization reac- tion. Catal. Commun., 43:218–222, 2014.
55 [22] C.L. Zhang, S. Li, Y. Yuan, W.X. Zhang, T.H. Wu, and L.W. Lin. Aromatization of methane in the absence of oxygen over Mo-based catalysts supported on different types of zeolites. Catal. Lett., 56:207–213, 1998.
[23] D. Zhou, S. Zuo, and S. Xing. Methane dehydrogenation and coupling to ethylene over a Mo/HZSM-5 catalyst: A density functional theory study. J. Phys. Chem. C, 116: 4060–4070, 2012.
[24] L.Y. Chen, L.W. Lin, Z.S. Xu, X.S. Li, and T. Zhang. Dehydro-oligomerization of methane to ethylene and aromatics over Molybdenum/HZSM-5 catalyst. J. Catal., 157: 190–200, 1995.
[25] W. Liu, Y. Xu, S.-T. Wong, L. Wang, J. Qin, and N. Yang. Methane dehydroaroma- tization and aromatization in the absence of oxygen on Mo/HZSM-5: A study on the interation between Mo species and HZSM-5 by using 27Al and 29Si MAS NMR. J. Mol. Catal. A., 120:257–265, 1997.
[26] J. Lu, Y. Xu, Y. Suzuki, and Z.G. Zhang. A rapid and effective method for evaluating the initial activity of Mo/HZSM-5 catalyst in the methane dehydroaromatization reaction at severe conditions. Catal. Commun., 12:127–131, 2010.
[27] C. Karakaya and R.J. Kee. Direct catalytic conversion of natural gas to fuels and chemicals. Preprint submitted to Prog. Energy Combust. Sci, 2016.
[28] J. Lunsford. Catalytic conversion of methane to more useful chemicals and fuels: a challenge for the 21st century. Catal. Today, 63:165–174, 2000.
[29] H. Lacheen and E. Iglesia. Stability, structure, and oxidation state of Mo/H-ZSM-5 catalysts during reactions of CH4 and CH4-CO2 mixtures. J. Catal., 230:173–185, 2005.
[30] Y. Xu, X. Bao, and L. Lin. Direct conversion of methane under non-oxidative conditions. J. Catal., 216:386–395, 2003.
[31] Y. Shu, R. Ohnishi, and M. Ichikawa. Pressurized dehydrocondensation of methane toward benzene and naphthalene on Mo/HZSM-5 catalyst: Optimization of reaction parameters an promotion by CO2 addition. J. Catal., 206:134–142, 2002.
[32] E.V. Matus, I.Z. Ismagiloc, O.B. Sukhova, V.I. Zaikovskii, L.T. Tsikoza, Z.R. Ismagilov, and J.A. Moulijin. Study of methane dehydroaromatization on impregnated Mo/ZSM-5 catalysts and characterization of nanostructured molybdenum phases and carbonaceous deposits. Ind. Eng. Chem. Res., 46:4063–4074, 2007.
56 [33] Ø. Hatlevik, S.K. Gade, M.K. Keeling, P.M. Thoen, A.P. Davidson, and J.D. Way. Palladium and palladium alloy membranes for hydrogen separation and production: History, fabrication strategies, and current performance. Sep. Purif. Technol., 73:59– 64, 2010.
[34] R.J. Kee, H. Zhu, B.W. Hildenbrand, E. Vøllestad, M.D. Sanders, and R.P. O’Hayre. Modeling the steady-state and transient response of polarized and non-polarized proton- conducting doped-perovskite membranes. J. Electrochem. Soc., 160:F290–F300, 2013.
[35] E. Vøllestad, H. Zhu, and R.J. Kee. Interpretation of defect and gas-phase fluxes through mixed-conducting ceramics using Nernst–Planck–Poisson and integral formulations. J. Electrochem. Soc., 161:F114–F124, 2014.
[36] O. Rival, B.P.A. Grandjean, C. Guy, A. Abdelhamid, and F. Larachi. Oxygen-free methane aromatization in a catalytic membrane reactor. Ind. Eng. Chem. Res., 40: 2212–2219, 2001.
[37] F. Larachi, H. Oudghiri-Hassani, M.C. Iliuta, B.P.A. Grandjean, and P.H. McBreen. Ru-Mo/HZSM-5 catalyzed methane aromatization in membrane reactors. Catal. Lett., 84:183–192, 2002.
[38] M.C. Iliuta, B.P.A. Grandjean, and F. Larachi. Methane non-oxidative aromatization over Ru-Mo/HZSM-5 at temperatures up to 973 K in a palladium-silver/stainless steel membrane reactor. Ind. Eng. Chem. Res., 42:323–330, 2003.
[39] L. Wang, K. Murata, A. Sayari, B. Grandjean, and M. Inaba. Production of ultra highly pure H2 and higher hydrocarbons from methane in one step at mild temperatures and development of the catalyst under non-equilibrium reactions. Chem. Commun., 19: 1952–1953, 2001.
[40] A.K. Kinage, R. Ohnishi, and M. Ichikawa. Marked enhancement of the methane de- hydrocondensation toward benzene using effective Pd catalytic membrane reactor with Mo/ZSM-5. Catal. Lett., 88:199–202, 2003.
[41] S. Natesakhawat, N.C. Means, B.H. Howard, M. Smith, V. Abdelsayed, J.P. Baltrus, Y. Cheng, J.W. Lekse, D. Link, and B.D. Morreale. Improved benzene production from methane dehydroaromatization over Mo/HZSM-5 catalysts via hydrogen-permselective palladium membrane reactors. Catal. Sci. Technol, 5:5023–5036, 2015.
[42] L. Li, R. Borry, and E. Iglesia. Reaction-transport simulations of non-oxidative methane conversion with continuous hydrogen removal – homogeneous-heterogeneous reaction pathways. Chem. Engr. Sci., 56:1869–1881, 2001.
57 [43] L. Li, R.W. Borry, and E. Iglesia. Design and optimization of catalysts and membrane reactors for the non-oxidative conversion of methane. Chem. Eng. Sci., 57:4595–4604, 2002.
[44] R.W. Borry III, E.C. Lu, Y.H. Kim, and E. Iglesia. Studies in surface science and catalysis. Elsevier Sci., Amsterdam, Netherlands, 1998.
[45] Z. Liu, L. Li, and E. Iglesia. Catalytic pyrolysis of methane on Mo/H-ZSM-5 with continuous hydrogen removal by permeation through dense oxide films. Catal. Lett., 82: 175–180, 2002.
[46] S. Liu, R. Ohnishi, and M. Ichikawa. Promotional role of water added to methane feed on catalytic performance in the methane dehydroaromatization reaction on Mo/HZSM-5 catalyst. J. Catal., 220:57–65, 2003.
[47] H. Ma, R. Kojima, S. Kikuchi, and M. Ichikawa. Effective coke removal in methane to benzene (MTB) reaction on Mo/HZSM-5 catalyst by H2 and H2O co-addition to methane. Catal. Lett., 104:63–66, 2005.
[48] S. Liu, Q. Dong, R. Ohnishi, and M. Ichikawa. Unique promotion effect of CO and CO2 on the catalytic stability for benzene and naphthalene production from methane on Mo/HZSM-5 catalysts. Chem. Commun., pages 1217–1218, 1998.
[49] S. Yao, C. Sun, J. Li, L. Gu, and W. Shen. Reaction coupling of methane steam reforming and methane dehydroaromatization for improving durability of Mo/HMCM- 49 catalyst. Chin. J. Catal., 30:1022–1028, 2009.
[50] B.S. Liu, L. Tian, L. Li, C.T. Au, and A.S.-C. Cheung. Performance of 3% Mo/ZSM- 5 catalyst in the presence of water during methane aromatization in supersonic jet expansion. AIChE J., 57:1852–1859, 2011.
[51] K. Skutil and M. Taniewski. Some technological aspects of methane aromatization (direct and via oxidative coupling). Fuel. Process. Technol., 87:511–521, 2006.
[52] Y. Li, T. Wu, W. Shen, X. Bao, and Y. Xua. Combined single-pass conversion of methane via oxidative coupling and dehydro-aromatization. Catal. Lett., 105:77–82, 2005.
[53] R. Ohnishi, S. Liu, Q. Dong, L. Wang, and M. Ichikawa. Catalytic dehydrocondensation of methane with CO and CO2 toward benzene and naphthalene on Mo/HZSM-5 and Fe/Co-Modified Mo/HZSM-5. J. Catal., 182:92–103, 1999.
[54] S. Yuan, J. Li, Z. Hao, Z. Feng, Q. Xin, P. Ying, and C. Li. The effect of oxygen on the aromatization of methane over the Mo/HZSM-5 catalyst. Catal. Lett., 63:73–77, 1999.
58 [55] W.P. Ding, G.D. Meitzner, and E. Iglesia. The effects of silanation of external acid sites on the structure and catalytic behavior of Mo/H-ZSM-5. J. Catal., 206:14–22, 2002.
[56] C.H.L. Tempelman and E.J.M. Hensen. On the deactivation of Mo/HZSM-5 in the methane dehydrogenation reaction. Appl. Catal., B, 176:731–739, 2015.
[57] D. Ma, L. Su, Z. Xu, Z. Tian, Y. Xu, L. Lin, and X. Bao. Remarkable improvement on the methane aromatization reaction: A highly selective and coking-resistant catalyst. J. Phys. Chem. B, 106:8524–8530, 2002.
[58] Y. Lu, D. Ma, Z. Tian, X. Bao, and L. Lin. A high coking-resistance catalyst for methane aromatization. Chem. Commun, pages 2048–2049, 2001.
[59] N. Wakao and J.M. Smith. Diffusion in catalyst pellets. Chem. Eng. Sci, 17:825–834, 1962.
[60] A. Srinivasan and C. Depcik. One-dimensional pseudo-homogeneous packed-bed reactor modeling: I. chemical species equation and effective diffusivity. Chem. Eng. Technol., 36:22–32, 2013.
[61] P. M´eriaudeau, L.V. Tiep, V.T.T. Ha, C. Naccache, and G. Szabo. Aromatization of methane over Mo/H-ZSM-5 catalyst: on the possible reaction intermediates. J. Mol. Catal. A, 144:469–471, 1999.
[62] Alexander Burcat. Prof. Burcat’s thermodynamic data. http://garfield.chem.elte.hu/Burcat/burcat.html, Accessed: 2/15/2015.
[63] M.E. Coltrin, R.J. Kee, and F.M. Rupley. Surface Chemkin: A generalized formalism and interface for analyzing heterogeneous chemical kinetics at a gas-surface interface. Int. J. Chem. Kinet., 23:1111–1128, 1991.
[64] O. Deutschmann, S. Tisher, C. Correa, D. Chatterjee, S. Kleditzsch, V. Janardhanan, N. Mladenov, H.D. Minh, H. Karadeniz, and M. Hettel. DETCHEM user manual. www.detchem.com, 2014.
[65] L. Maier, N. Schaedel, K.H. Delgado, S. Tisher, and O. Deutschmann. Steam reforming of methane over nickel: Development of a multi-step surface reaction mechanism. Top. Catal., 54:845–858, 2011.
[66] O. Deutschmann. Modeling and Simulation of Heterogeneous Catalytic Reactions: From the Molecular Process to the Technical System. Wiley-VCH, Weinheim, Germany, 2012.
[67] C. Karakaya and O. Deutschmann. Kinetics of hydrogen oxidation on Rh/Al2O3 cata- lysts studied in a stagnation-flow reactor. Chem. Eng. Sci., 89:171–184, 2013.
59 [68] A.B. Mhadeshwar and D.G. Vlachos. Hierarchical, multiscale surface reaction mecha- nism development:CO and H2 oxidation, water-gas shift, and preferential oxidation of CO on Rh. J. Catal., 234:48–63, 2005.
[69] A.B. Mhadeshwar and D.G. Vlachos. A catalytic reaction mechanism for methane partial oxidation at short contact times, reforming, and combustion, and for oxygenate decomposition and oxidation on platinum. Ind. Eng. Chem. Res., 46:5310–5324, 2007.
[70] L. Devi, K.J. Ptasinski, and F.J.J.G. Janssen. Decomposition of naphthalene as a biomass tar over pretreated olivine: Effect of gas composition, kinetic approach, and reaction scheme. Ind. Eng. Chem. Res., 44:9096–9104, 2005.
[71] Q. Wang, H. Fan, S. Wu, Z. Zhang, P. Zhang, and B. Han. Water as an additive to enhance the ring opening of naphthalene. Green Chem., 14:1152–1158, 2012.
[72] P. R. Buchireddy, R. M. Bricka, J. Rodriguez, and W. Holmes. Biomass gassification: Catalytic removal of tars over zeolites and nickel supported zeolites. Energy Fuels, 24: 2707–2715, 2010.
[73] Z.A. El-Rub, E.A. Bramer, and G. Brem. Tar removal in an entrained flow cracker (EFC) with application to biomass gasification. In Pyrolysis and gasification of biomass and waste: Proceedings of an expert meeting, pages 337–346, Strasbourg, France, 30 September - 1 October 2002.
[74] A. Jess. Mechanisms and kinetics of thermal reactions of aromatic hydrocarbons from pyrolysis of solid fuels. Fuel, 75:1441–1448, 1996.
[75] X. Garcia and K. H¨uttinger. Steam gasification of naphthalene as a model reaction of homogeneous gas/gas reactions during coal gasification. Fuel, 68:1300–1310, 1989.
[76] I. Narv´aez, J. Corella, and A. Or´ıo. Fresh tar (from a biomass gasifier) elimination over a commercial steam-reforming catalyst. Kinetics and effect of different variables of operation. Ind. Eng. Chem. Res., 36:317–327, 1997.
[77] C. Sun, G. Fang, X. Guo, Y. Hu, S. Ma, T. Yang, J. Han, H. Ma, D. Tan, and X. Bao. Methane dehydroaromatization with periodic CH4-H2 switch: A promising process for aromatics and hydrogen. J. Energy Chem., 24:257–263, 2015.
[78] C.H.L. Tempelman, X. Zhu, and E.J.M. Hensen. Activation of Mo/HZSM-5 for methane aromatization. Chin. J. Catal., 36:829–837, 2015.
[79] Y. Xu, Y. Song, Y. Suzuki, and Z.-G. Zhang. Effect of superficial velocity on the coking behavior of non-oxidative CH4 dehyroaromatization at 1073 K. Catal. Sci. Technol., 3: 2769–2777, 2013.
60 [80] Y. Xu, Y. Suzuki, and Z.-G. Zhang. Comparison of the activity and stability of nanosized and microsized based Fe-Mo/HZSM-5 catalysts in the non-oxidative CH4 dehydroarom- atization under periodic CH4-H2 switching operation at 1073 K. Appl. Catal. A, 452: 105–116, 2013.
[81] L.L. Korobitsyna, N.V. Arbuzova, and A.V. Vosmerikov. Physicochemical properties and activity of Mo-containing zeolite catalyts of nonoxidative conversion of methane. Russ. J. Phys. Chem. A, 87:919–922, 2013.
61