Stability of Fe- and Zn-Promoted Mo/ZSM‑5 Catalysts for EthaneDehydroaromatization in Cyclic Operation ModeBrandon Robinson,† Xinwei Bai,† Anupam Samanta,† Victor Abdelsayed,‡,§ Dushyant Shekhawat,‡
and Jianli Hu*,†
†Department of Chemical and Biomedical Engineering, West Virginia University, 395 Evansdale Drive, Morgantown, West Virginia26505, United States‡National Energy Technology Laboratory, United States Department of Energy, 3610 Collins Ferry Road, Morgantown, WestVirginia 26505, United States§AECOM, 3610 Collins Ferry Road, Morgantown, West Virginia 26505, United States
*S Supporting Information
ABSTRACT: The stability of Fe- and Zn-promoted Mo/ZSM-5 catalysts was studied toward the ethane dehydroaromatizationreaction. To elucidate the catalyst deactivation and regeneration mechanism, the catalytic performance was evaluated during fiveconsecutive reaction−regeneration cycles. The addition of Zn to Mo/ZSM-5 resulted in an initial increase in aromaticselectivity; however, the loss of about 50% Zn resulted in a total decrease in aromatic selectivity over five reaction andregeneration cycles. The addition of Fe to Mo/ZSM-5 resulted in no decrease in aromatic selectivity or aromatic yieldthroughout these reaction cycles. The formation of carbon nanotubes was observed on the Fe-promoted Mo/ZSM-5 catalyst,which was believed to improve gas diffusion to micropores. The presence of Fe- and Mo-agglomerated particles on the catalystsurface was observed at low Fe/Mo atomic ratios, resulting in the formation of base-grown carbon nanotubes. When the Fe/Moatomic ratio increased, the agglomerated particles were able to leave the zeolite surface, resulting in the formation of tip-growthcarbon nanotubes. The temperature-programmed reduction profile of the Fe-promoted Mo/ZSM-5 catalyst also suggested theformation of a more stable Mo and/or Fe species at 615 °C, enhancing carbon nanotube formation. The addition of Fe in thebipromoted Mo/ZSM5 catalyst was found to stabilize the catalyst surface and reduce its Zn loss.
1. INTRODUCTION
Ethane is the second largest component of natural gas, and ithas become increasingly important to find more efficient waysto convert ethane to valuable aromatic hydrocarbons to beused as liquid fuels and chemical intermediates in the polymerindustry. Commercial natural gas conversion to chemicaltechnologies are based on an indirect route via syngasproduction. Indirect routes are generally energy-inefficientand capital-intensive. Typically, more than 50% of the capitalcost is incurred in syngas production.1 In contrast, direct non-oxidative natural gas conversion eliminates the syngasproduction step and high cost air separation unit requiredfor oxygen separation. However, this technologies have notbeen commercialized because of technical challenges, such aslow selectivity, coking, heat management in the reactor, andcatalyst deactivation and regeneration.2 To this date, the directaromatization of C1−C2 molecules is still in the early stages ofdevelopment. This study is focused on developing stablecatalyst systems and regeneration processes having thepotential for commercial applications.The transformation of alkanes into aromatic hydrocarbons
over zeolite-supported metal catalyst have been studied formany decades for the non-oxidative dehydroaromatizationreaction.3−5 The three-dimensional structure and pore size ofH-ZSM-5 have been found to allow for the formation andtransportation of lower aromatic hydrocarbons, while highlyalkylated aromatics are excluded from the pore channel system.
Molybdenum supported on H-ZSM-5 is a well-studied catalystsystem for the dehydroaromatization reaction. Upon calcina-tion, molybdenum can be dispersed into the microporestructure of the zeolite.6 During the aromatization reaction,molybdenum is reduced from MoO3/ZSM-5 to its most activephase Mo2C/ZSM-5, which catalyzes the dehydrogenationstep of ethane to ethylene. Ethylene molecules are thencoupled together and oligomerize on the acid sites on the H-ZSM-5 surfaces where the aromatization step takes place.7,8
Molybdenum-based zeolite catalysts have shown goodactivity toward the aromatization of methane and ethane;however, catalyst deactivation is still a major issue. Researchershave attempted to reduce the deactivation rate and improvethe stability of Mo/ZSM-5 by reducing the coke formation.9,10
In addition to the thermodynamic limitation of this reaction,the concentration of the Brønsted acid sties of the zeolitesupport could contribute to catalyst deactivation at highsurface densities. Thus, balancing between metal loading aswell as the Si/Al ratio plays an important role in catalyststability.11−13 Many studies reported the addition of oxidantsto the feed stream to prevent coke formation or the use of lowconcentrations of oxygen and lower temperatures duringregeneration to slowly remove the formed coke to increasecatalyst stability.14−17
Received: May 1, 2018Published: June 7, 2018
Article
pubs.acs.org/EFCite This: Energy Fuels 2018, 32, 7810−7819
Several promoters have been studied on the Mo/ZSM-5catalyst to improve its stability and reduce its tendency tocoke.18,19 The addition of Fe to Mo/ZSM-5 was found toimprove the activity and stability of Mo/ZSM-5.20 This wasattributed to carbon nanotube (CNT) formation, whichsuppresses the low-temperature surface coke formation as aresult of competition for coke precursors. In general, theaddition of Fe to Mo/ZSM-5 results in a high selectivity forcoke and methane formation.19,20 Several studies have beenattributed to Mo and Fe bimetallic catalysts, and it is prevalentthat the Fe/Mo ratio is taken into consideration. Byadjustment of the Fe/Mo ratio, the selectivity of the catalystcan be increased. The observed CNTs on the zeolite surfacesin the presence of the Fe catalyst suggested that the increasedand prolonged activity could be attributed to lower cokedeposition on the catalyst surface, leading to improved gasdiffusion into the pores of the Mo/ZSM-5 catalysts.18,23,24
The possibility of the addition of a third promoter on thecatalyst, such a Zn, could increase the aromatic yield, especiallyfor the ethane compared to methane dehydroaromatizationreaction, because it requires lower activation temperatures,which will increase the Zn stability, especially at lowerloadings. It was found that Mo/ZSM-5 showed higher benzeneselectivity and a lower total number of strong acid sites whenZn was added.25,26 However, a recent study has determinedthat the most active Zn metals were located inside themicropore but suffered from evaporation, which led to veryrapid deactivation during the dehydroaromatization reaction ofmethane.27 Furthermore, it was reported that there were twotypes of Zn species present during the dehydroaromatization ofmethane: one being a ZnO species that is loosely bound andeasily reduced, and the second is the Zn species that is boundto the acid sites of ZSM-5 supports and is a more reactive andstable [Zn(OH)x]. The amount of [Zn(OH)x] present wasdependent upon the Zn loading. At higher loading, most Znwas present in the ZnO form.28 The objectives of this researchwere to develop stable metal-promoted zeolite catalysts forethane dehydroaromatization and test its stability undermultiple reaction−regeneration cycles. The experiments werefocused on catalyst synthesis, performance test, and character-ization to elucidate the deactivation mechanism and the effectsof the regeneration protocol.
2. EXPERIMENTAL SECTION2.1. Catalyst Preparation. The NH4−ZSM-5 zeolite catalyst
with a silica/alumina ratio (SAR) of 23 was purchased from Zeolyst,Inc. Ammonium heptamolybdate tetrahydrate, zinc nitrate hexahy-drate, and iron(II) chloride, anhydrous, were purchased from AcronsOrganics. The zeolite catalyst was first calcined at 500 °C in air for 3 hto convert NH4−ZSM-5 to H−ZSM-5. The conventional incipientwetness technique was used to prepare the Mo/ZSM-5 catalyst, whileMoFe, MoZn, and MoFeZn on ZSM-5 were prepared by the co-impregnation method. After drying the catalysts at 105 °C for 5 h, the
powders were calcined in air at 550 °C for 10 h. The chemicalcompositions of the prepared catalysts are shown in Table 1.
2.2. Catalyst Characterization. 2.2.1. X-ray Diffraction (XRD).Powder XRD analysis was performed on a PANalytical X’Pert Pro(PW3040) set to 45 kV and 40 mA. Samples were scanned from 5° to35° (2θ) using Cu Kα radiation. Highscore Plus Analyses softwaresupplied by PANalyticial was used for data analysis.
2.2.2. Surface Area and Micropore Analysis. Surface area analysiswas carried out in a Micromeritics ASAP 2020 unit. The catalystsamples were degassed at 300 °C for 10 h under vacuum to removeany surface moisture and absorbed gases. Nitrogen was used as theadsorption gas. The surface area was calculated using the Brunauer−Emmett−Teller (BET) model, and the t-plot method was used tocalculate micropore volume and micropore area.
2.2.3. Ammonia Temperature-Programmed Desorption (NH3−TPD). NH3−TPD experiments were carried out using a MicromeriticsAutochem 2910 equipped with a thermal conductivity detector. Thesamples were heated to 300 °C at 5 °C/min for 60 min under an inertflow of He to remove moisture and then cooled to 150 °C. Premixed15% ammonia in helium then flowed over the catalyst for 30 min at30 mL/min. A baseline was determined by flowing helium over thesample for 30 min at 50 mL/min to remove weekly boundedammonia from the catalyst surface. The samples were then heated to700 °C at 5°/min.
2.2.4. High-Resolution Transmission Electron Microscopy (HR-TEM). Transmission electron microscopy (TEM) micrographs andenergy-dispersive spectroscopy (EDS) analysis were obtained using aJEOL JEM-2100, equipped with Oxford EDS. The samples wereprepared by sonicating in isopropyl alcohol for 10 min and thenloaded onto a copper grid. The prepared TEM sample grids weredried in air for 8 h.
2.3. Catalyst Evaluation. 2.3.1. Reactor Configuration andReaction Conditions. The reaction was carried out in a MicromeriticsAutochem 2950 analyzer connected with a micro gas chromatograph(micro-GC) for gas analysis. For each experiment, 200 mg of catalystwas loaded into a quartz tube reactor. The reaction was carried outunder atmospheric pressure and continuous flow conditions. Thecatalyst was heated to 615 °C in argon at a flow rate of 50 mL/minwith the heating rate of 10 °C/min. Pure ethane was mixed withargon to create a 36% ethane mixture, which then flowed over thecatalyst at 50 mL/min. After 21 min of reaction, helium wasintroduced to purge ethane in the system. The catalyst remained at615 °C, where it was regenerated for 95 min in a flow of 2% oxygen inhelium. After catalyst regeneration, helium was introduced to purgeremaining oxygen in the system and the catalyst was ready for the next21 min reaction cycle. A total of five reaction cycles (1 fresh and 4regenerated) were performed for each catalyst. All reactant gases werepurchased from AirGas with ultrahigh-purity (UHP) grade.
2.3.2. Product Analysis. The reactant gases were analyzed by afour-channel Agilent 3000 micro-GC. The micrometrics unit internalvalves and line temperatures were maintained at 150 °C, and a 150−170 °C heated trace line attached to the inlet of the micro-GC wasused to maintain the products in gas form. The micro-GC wasequipped with four columns consisting of molecular sieve, PLOT U,aluminum, and OV-1, allowing for the analysis of hydrogen, methane,argon, ethane, ethylene, benzene, and toluene. Xylene was not traced,and naphthalene was separated out of the analysis stream; thus,concentrations were not reported. The ethane conversion and productselectivity were defined in eqs 1 and 2, respectively.
selectivity of aromatics (%)benzene (%) 6 toluene (%) 7ethane fed (%) ethane out (%) 2
100(2)
2.3.3. Temperature-Programmed Oxidation (TPO). TPO analyseswere carried out on the coked catalyst using a MicrometricsAutochem 2950 equipped with a thermal conductivity detector.About 100 mg of coked catalyst was first heated to 300 °C with aramp rate of 5 °C/min in an inert flow of argon at 30 mL/min for 60min. The sample was then cooled to ambient temperature, where itwas then heated in the presence of 2% oxygen at a ramp rate of 2 °C/min.2.3.4. Thermogravimetric Analysis (TGA). TGA of the coked
catalyst was carried out using a TA Instrument SDT 650. The samplewas heated to 150 °C for 60 min under an inert flow of helium to drythe sample. The sample was then heated from 150 to 700 °C in a flowof 2% oxygen. The temperature was then held constant for 30 min in2% oxygen to ensure complete removal of the coke in the temperaturerange.2.3.5. Temperature-Programmed Reduction (TPR). TPR was
carried out in a Micrometrics AutoChem 2950 equipped with athermal conductivity detector. The samples were heated to 300 °C at10 °C/min for 60 min under an inert flow of He to remove moistureand then cooled to 100 °C. Premixed 10% hydrogen in argon wasthen flowed for 20 min over the catalyst for 25 mL/min to achieve abaseline. Once a baseline was achieved, the samples were heated to900 °C at 5 °C/min.2.3.6. ICP−OES. All trace metals were analyzed using ICP−OES on
an Optima 7300 DV (PerkinElmer, Waltham, MA, U.S.A.), which is adual-view spectrometer with solid-state segmented-array charge-coupled device (SCD) detectors. Calibration standards werepurchased from Inorganic Ventures (Christiansburg, VA, U.S.A.)and are traceable to National Institute of Standards and Technology(NIST) standard reference materials. Fe, Mo, and Zn were measured
after digestion in aqua regia acid. A total of 12 mg of each sample wasdigested in 12 mL of aqua regia (9 mL of HCl and 3 mL of HNO3)acid and then further diluted 100-fold using a mixture of high-purity2% nitric acid prior to analysis.
3. RESULTS AND DISCUSSION3.1. Catalyst Characterization. 3.1.1. XRD Analysis. As
illustrated in Figure 1, the XRD patterns of the freshly
prepared catalysts are compared to the regenerated catalyst.Considering the diffraction angles of 22−25°, almost nochanges were observed in the fresh catalyst diffraction patterncompared to the regenerated catalyst, indicating that the ZSM-5 structure retained its crystallinity throughout the reaction
Figure 1. XRD patterns of fresh and spent catalysts after five cycles.
Figure 2. Nitrogen adsorption−desorption isotherms for the freshcatalyst.
and regeneration cycles. Furthermore, metal oxide peakscorresponding to Mo, Fe, or Zn could not be observed,indicating that the metal loadings were homogeneouslydispersed and their particle sizes were small below the XRDdetection limits.29
3.1.2. Analyses of the Surface Area and Acidity. Thenitrogen adsorption and desorption plot are shown in Figure 2for the fresh catalysts. The plot of H-ZSM-5 and all four metal-loaded ZSM-5 catalysts represent a type II isotherm. The largehysteresis loop observed in the range of P/Po = 0.40−1.0indicates that the catalyst have a mesoporous structure, wherenitrogen condenses on the external surfaces of the crystallitesand on the spaces in between.A summary of the surface area and micropore analysis is
shown in Table 2. The addition of Mo to the unpromptedzeolite resulted in a decrease in the total surface area, includingthe micropore area and the external surface area of the catalyst.As the total weight percent of the loaded metals increased, the
total surface area of catalysts decreased. The addition of Fe toMo/ZSM-5 resulted in a further decrease in the externalsurface area. However, their micropore area did not changecompared to Mo/ZSM-5, suggesting that the Fe metal particlesdid not diffuse into the micropore structure of the zeolite. Theaddition of Zn to the Mo/ZSM-5 catalyst showed similareffects compared to Fe on the micropore area and externalsurface area. This suggests that Zn occupies the external acidsites on the ZSM-5 structure with little to no diffusion into thepores of the ZSM-5 structure. For the MoFeZn/ZSM-5catalyst, there was a total decrease in both the micropore areaand external surface area compared to pure H-ZSM-5.Figure S1 of the Supporting Information represents NH3−
TPD profiles of H-ZSM-5 and all four fresh catalysts. Pure H-ZSM-5 exhibits two NH3 desorption peaks at 223 and 427 °C,representing weak acid sites (mostly Lewis acid) and strongacid sites (mostly Brønsted acid), respectively. The presence ofMo particles anchored to the Brønsted acid sites on theexternal surface and inside the pores is evident from thedecrease of the higher temperature peak shown in Figure S1 ofthe Supporting Information and confirmed by the BET analysisabove.30 The addition of Fe and Zn to Mo/ZSM-5 did notcause a further decrease in the Brønsted acid strength. On thebasis of the temperature shift and changes in peak shape,Figure S1the addition of Mo, Fe, and Zn has less impact onLewis acid sites compared to Brønsted acid sites.
3.2. Catalyst Performance in Ethane Dehydroaroma-tization. Figure 3 depicts the conversion and selectivity ofeach catalyst toward ethane dehydroaromatization. The secondcycle was chosen as a basis for comparison throughout this
Table 2. Surface Area and Micropore Results of the FreshCatalysts
study as a result of the presence of the induction period in thefresh catalyst, making it very difficult to compare the activity in
the first cycle. The promoting effect of metals on ZSM-5 wasranked on the basis of ethane conversion. The ethaneconversion level for the second cycle (Figure 3a) was foundin the order of MoFe > Mo> MoFeZn > MoZn, whereas forthe fifth cycle (Figure 3b), the conversion level followed theorder of Mo > MoZn ≈ MoFeZn > MoFe. The aromaticselectivity of these catalysts with respect to regeneration cyclesis shown in Figure 3c. Although the selectivity of the MoFecatalyst to aromatics was relatively lower than others, it showedstable aromatic selectivity with little to no decrease in theaverage selectivity from the second to fourth reaction cycle andeven a slight increase on the fifth cycle. The MoZn catalystshowed a slight increase in the overall average aromaticselectivity from the first to third reaction cycle. However, adecrease was observed from the third to fifth reaction cycle.The MoFeZn catalyst exhibited a trend similar to the MoZncatalyst; however, the slope of the deactivation of aromaticselectivity was less severe from the third to fifth cycle. Thissuggest that Fe had a stabilizing effect on the MoFeZn catalystas it did in the MoFe catalyst. It was reported in the literaturethat the presence of Zn can promote selectivity to aromatics inethane aromatization as a result of the type of Zn sitesformed.31 The increase in selectivity to aromatics on MoZnand MoFeZn catalysts over the first to third cycles could be
Figure 4. Comparison of time-on-stream between benzene concentrations for five reaction cycles.
Figure 5. Comparison of the total aromatic yields per reaction cyclein mole percent of benzene and toluene.
Figure 6. TEM images of spent catalysts before regeneration: (a) Mo and (b) MoFe and EDS data. a: N/F stands for not found.
due to the formation of zinc hydroxyl species. However, afterthe third reaction cycle, a decline in aromatic selectivity wasobserved possibly as a result of the loss of the Zn species.27
Meanwhile, the presence of Zn causes the loss of Mo, whichwill be discussed later in this paper. Table 1 showed the ICP
analysis of spent catalysts after five reaction/regenerationcycles; the Zn content decreased by nearly 50%. This could beexplained by the loss of the volatile unbound ZnO species, thusleaving the more stable and active [Zn(OH)]+ species boundto the acid sites of the zeolite surface.28
Figure 7. (a) and (b) are both TEM images of the spent MoZn catalyst and EDS analysis. a: N/F stands for not found.
Figure 8. (a) and (b) are both TEM images of the spent MoFeZn catalyst and EDS analysis. a: N/F stands for not found.
Figure 4 shows the benzene concentration as a function oftime-on-stream (TOS) for all of the catalysts studied. Overall,the Mo/ZSM-5 catalyst showed the highest benzeneconcentration, followed by MoFe, MoFeZn, and MoZncatalysts. To study the commercial viability of cyclicregeneration, for each catalyst, the extent of deactivationbetween each regeneration cycle was analyzed. Ideally, a stablecatalyst should exhibit the characteristics that the time-on-
stream benzene concentration in all five cycles falls on thesame line. This indicates that the catalyst is regenerable andcan be used repeatedly. In contrast, if a divergence of thebenzene concentration between cycles is observed, that is anindication of either irreversible deactivation or an inefficientregeneration process. Understanding the cause of the“divergence” between cycles is thus important in developinga commercially viable catalyst and regeneration process. Thenumerical data in Figure 4 are quantitatively broken down togive the percent change in the benzene concentration fromcycle to cycle for each catalyst and then an overall change fromthe second to fifth reaction cycle. Results are summarized inTable S1 of the Supporting Information where it is shown thatthe MoFe catalyst exhibits little to no decrease in the benzeneconcentration from cycle to cycle. The other three catalystssuffer from the loss of activity and irreversibility betweencycles, with the Mo catalyst being the most significant.Similarly, MoFe exhibits the same behavior toward the
toluene formation during these 5 reaction cycles (Figure S2 ofthe Supporting Information). The variation in the tolueneconcentration was the minimum among other catalysts studiedduring these five reaction cycles, yet they still show sometoluene loss from cycle to cycle. This could be attributed to thebuildup of coke on the interior surface of the pores, resulting ina decrease in the diameter of the pore. Furthermore, thedecrease in the pore diameter could lead to the toluenedisproportionation reaction to form benzene and xylene.32
This may be an explanation that, for the MoFe catalyst, little tono decrease in the benzene concentration was observedbetween cycles, whereas a slight decrease in the tolueneconcentration continued.Figure 5 depicts the distribution of total aromatic yields of
all four catalysts. The total aromatic (benzene and toluene)yield follows the order of Mo > MoFeZn > MoFe > MoZnsupported on ZSM-5. For each catalyst, starting from thesecond cycle, a trend of decrease in the total aromatic yield isobserved on Mo, MoZn, and MoFeZn catalysts. However, the
Figure 9. TPO (solid) and TGA (dotted) profiles of the spentcatalyst after 5 reaction cycles.
Table 3. TGA Results of the Spent Catalyst Obtained at theEnd of the Fifth Reaction Cycle
catalyst total coke (mg) carbon formation (g g−1 of catalyst min−1)
Mo 2.49 0.118MoFe 3.10 0.147MoZn 2.38 0.113MoFeZn 3.08 0.146
Figure 10. TPR profiles of the fresh versus regenerated fifth cycle catalyst.
MoFe catalyst shows no decrease in the benzene yield but aslight decrease in the total aromatic yield as a result of thetoluene disproportionation reaction.Illustrated in Figure S3 of the Supporting Information is the
comparison of the hydrogen production rates between allcatalysts studied. The initial activity of hydrogen productionrates is in the following order: MoFeZn > Mo ≈ MoFe >MoZn, where Mo and MoFe catalysts exhibited comparablehydrogen production rates and the MoZn catalyst was thelowest. MoFeZn showed the highest initial production rate,probably attributed to the higher metal loading. The MoFecatalyst shows the least amount of reduction in the hydrogenconcentration per cycle (Table S3 of the SupportingInformation). The result is consistent with the time-on-streamchanges of benzene and toluene concentrations, as illustratedin Figure 4 and Figure S2 of the Supporting Information.Figure S4 of the Supporting Information depicts the time-
on-steam methane formation rate for all catalysts studied. Theactivity of methane formation is in the order of MoFeZn > Mo≈ MoFe > MoZn supported on ZSM-5. The changes in themethane concentration between cycles are summarized inTable S4 of the Supporting Information. From the second tofifth cycle, Mo and MoFeZn exhibited comparable changes inthe methane formation rate, whereas the MoFe catalystresulted in little to no change in the methane concentration.The result is consistent with the findings from theconcentration changes for benzene, toluene, and hydrogenshown in Figure 4 and Figures S2 and S3 of the SupportingInformation.3.3. Mechanistic Study of Catalyst Deactivation and
Regeneration. To elucidate the catalyst deactivationmechanism and the effectiveness of regeneration, TEM/EDSanalysis was carried out. The TEM images of all four catalystsare shown in Figures 6−8. Figure 6a shows TEM of the spentMo catalyst after five reaction cycles. Agglomeration of Moparticles was observed, as confirmed using EDS analysis (spot1).The spent MoFe catalyst is shown in Figure 6b along with
the EDS analysis for spots 1 and 2. The formation of CNTswas observed. Spot 1 contains a high concentration of carbonalong with agglomerated metal particles that have a Fe/Moatomic ratio of 12.65. Spot 1 does not show any supportstructure (Si and Al), indicating that the metal agglomerateshave leached out the surface of the catalyst during the reaction.Spot 2 also contains a large amount of carbon; however, theagglomerated particle shows a Fe/Mo ratio of 1.2 and islocated on the surface of the zeolite. This would suggest that,at a larger Fe/Mo ratio, the agglomerated MoFe particles tendto leave the surface of the catalyst. In addition, both tip-growthand base-grown CNTs can be observed in TEM analysis.The TEM images of the spent MoZn catalyst are shown in
panels a and b of Figure 7. Spot 1 contains a large atomicpercentage of Mo and Zn, with a Zn/Mo atomic ratio of 1.78,located on a support structure with some carbon deposited. Onspot 2, the presence Mo and Zn is not observed, indicating thatthe metal particles are not located in this area or that thedensity of metal particles is too small as a result of evendispersion. Spot 3 contains a high concentration of Mo and Zn,with a Zn/Mo atomic ratio of 0.97. On spot 4, Zn or Mo is notfound, suggesting the uniform dispersion in that area or themigration of metals to the spot 3 area.The TEM image of the spent MoFeZn catalyst is shown in
panels a and b of Figure 8. On spot 1, the TEM image shows
agglomerated Mo and Fe particles that are not located on thezeolite support. On spot 1, the Fe/Mo atomic ratio is 1.63.This is representative of tip-growth CNT, where the Mo−Feagglomerates pushed out to leave the surface of the zeolitesupport. Spot 2 contains agglomerated Fe and Mo, with amuch lower Fe/Mo ratio of 0.47. Spot 3 is another spot wherea lower Fe/Mo ratio of 0.40 is observed. The lower atomicratios could be explained by Masiero et al.,21 where Feinteracted structurally with Mo to form a new binary phase ofFe2(MoO4)3. In calculating the atomic ratio of Fe and Mo inFe2(MoO4)3, we find a Fe/Mo atomic ratio of about 0.41.Furthermore, the Fe/Mo ratio in Fe2(MoO4)3 coincides withwhat was found in the EDS analysis on spots 2 and 3. It may bepossible that Fe2(MoO4)3 is formed on the MoFeZn catalyst.Further investigation will be carried out and reported in ourfuture publications.It appears that, at lower Fe/Mo ratios, the Fe and Mo
agglomerates stay attached to the zeolite support structure,whereas at higher Fe/Mo ratios, the excess amounts of Feallow for agglomerated particles to detach from the surface.ICP results shown in Table 1 indicate a small loss in Feparticles in the spent catalysts containing Fe, which suggestthat most Fe particles are deposited back onto the surface ofthe catalyst during regeneration. However, some Fe particlesare speculated to be deposited onto the reactor walls. ICPanalysis also suggests that the MoFeZn catalyst lost less Mothan the MoZn catalyst, indicating that Fe is a more stablepromoter for Mo than Zn. The loss of Zn and Fe particles fromthe catalyst surface follow different mechanisms. To elaborate,the loss of Zn is due to the presence of the volatile state of zincoxide.28 In contrast, the loss of Fe occurs in a more randomway, which is mainly due to the formation of tip-growth CNTsextending randomly away from the catalyst surface.
3.4. TPO and TGA. The TPO profiles were only usedqualitatively to characterize types of coke formed on thecatalyst surfaces during reaction based on the peak shape andburn-off temperatures. Figure 9 shows the TPO and TGAprofiles of the spent catalysts obtained after the fifth reactioncycle. The peak shape of Mo and MoZn catalysts wouldsuggest the presence of a single temperature burningcarbonaceous species with resulting peak temperatures of445 °C. The twin-peak TPO spectra was observed on MoFeand MoFeZn catalysts. The second peak is associated with thepresence of higher burning temperature carbon species (516°C). The TPO analysis suggests that the addition of Fe resultsin the formation of a higher ordered coke species, e.g., CNTs,as observed from TEM analysis.The TGA profile of the spent Mo and MoZn catalysts results
in one continuous negative curvature, whereas for MoFe andMoFeZn catalysts, the TGA curvature has a positive rise inslope at the temperature of the lowest point between the twopeaks in the TPO profiles. The TGA results for all fourcatalysts are summarized in Table 3. As shown in Table 3,MoFe and MoFeZn catalysts are more selective toward cokeformation, which is consistent with literature reports.20−22 TheMo catalyst showed the highest conversion but still had about25% less coke than the MoFe catalyst.
3.5. TPR. Figure 10 shows the TPR profiles for the freshand regenerated catalysts (fifth cycle). For both Mo and MoZncatalysts, a temperature shift of 51 °C (first peak from 473 to422 °C) is observed between fresh and spent catalysts.However, for the catalysts containing Fe, only a 27 °C shift inthe temperature of the first peak is observed, which is evident
of the presence of a more stable metal state. For the spentcatalyst obtained after five cycles, the peak shape of MoFe isidentical to MoFeZn, whereas the peak shape of Mo is almostidentical to MoZn. This is mainly attributed to the loss of Znover the course of the five cycles, as shown in the ICP analysis;therefore, the impact of Zn on stabilizing Mo graduallydiminished as the reaction cycles proceed.
4. CONCLUSIONThe stability of Fe- and Zn-promoted Mo in ethanedehydroaromatization was evaluated over five cycles ofreaction and regeneration. All catalysts were found to maintaintheir zeolite crystalline structure. Mo was determined to be inthe pores and on the exterior surface with strong interactionswith Brønsted acid sites. On Fe- and Zn-promoted Mocatalysts, Fe and Zn were located mainly on the exteriorsurface. The Mo-containing ZSM-5 catalyst exhibited thehighest conversion and aromatic yield; however, suffered a lossin activity between reaction cycles. The addition of Zn resultedin an increase in aromatic selectivity during early reactioncycles but suffered a decrease in selectivity in the consecutivereaction cycles. The promoting effect of Zn was short-lived incyclic operation as a result of the loss of almost 50% Zn overthe course of the 5 reaction cycles. The initial increase inselectivity was attributed to the formation of zinc hydroxidespecies, [Zn(OH)]+.The addition of Fe resulted in great stability in aromatic
selectivity, total aromatic yield, and hydrogen formation rateover the course of five reaction cycles. The improved stabilitywas attributed to the formation of CNTs that allowed forimproved gas diffusion into the pores. At lower Fe/Mo atomicratios, the agglomerated particles were found on the surface ofthe catalyst, resulting in the favored base growth CNTs. Athigher Fe/Mo ratios, tip-growth CNTs were observed, whichwas believed to be the cause of the loss of Fe and Mo. MoZnlost more Mo than MoFeZn, suggesting that Mo was morestable in the presence of Fe than it was with Zn. Metal particleswith a lower Fe/Mo atomic ratio would result in even greaterstability.
■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.energy-fuels.8b01516.
Change of the benzene concentration between cycles(Table S1), change of the toluene concentrationbetween cycles (Table S2), change of the hydrogenconcentration between cycles (Table S3), change of theethane concentration between cycles (Table S4), NH3−TPD of the fresh catalysts (Figure S1), comparison oftime-on-stream toluene concentration changes for 5reaction cycles (Figure S2), comparison of time-on-stream hydrogen concentration changes for 5 reactioncycles (Figure S3), and comparison of time-on-streammethane concentration changes for 5 reaction cycles(Figure S4) (PDF)
ORCIDJianli Hu: 0000-0003-3857-861XNotesThe authors declare no competing financial interest.
■ ACKNOWLEDGMENTSThe authors acknowledge financial support from the WestVirginia Higher Education Policy Commission under GrantHEPC.dsr.18.7. The authors also acknowledge financialsupport from the National Energy Technology Laboratory,Morgantown, WV, U.S.A., and the Oak Ridge Institute forScience and Education, Oak Ridge, TN, U.S.A.
■ REFERENCES(1) de Klerk, A. Gas-to-liquid conversion. Proceedings of the ARPA-ENatural Gas Conversion Technologies Workshop; Houston TX, Jan 13,2012.(2) Skutil, K.; Taniewski, M. Fuel Process. Technol. 2006, 87 (6),511−521.(3) Olson, D. H.; Kokotailo, G. T.; Lawton, S. L.; Meier, W. M. J.Phys. Chem. 1981, 85 (15), 2238−2243.(4) Richardson, J. T. Principles of Catalyst Development; PlenumPress: New York, 1989.(5) Hagen, A.; Roessner, F. Catal. Rev.: Sci. Eng. 2000, 42 (4), 403−437.(6) Xu, Y.; Liu, S.; Guo, X.; Wang, L.; Xie, M. Catal. Lett. 1994, 30(1−4), 135−149.(7) Solymosi, F.; Szoke, A. Appl. Catal., A 1998, 166 (1), 225−235.(8) Guisnet, M.; Gnep, N. S. Appl. Catal., A 1996, 146 (1), 33−64.(9) Bai, J.; Liu, S.; Xie, S.; Xu, L.; Lin, L. Catal. Lett. 2003, 90 (3−4),123−130.(10) Bai, J.; Liu, S.; Xie, S.; Xu, L.; Lin, L. React. Kinet. Catal. Lett.2004, 82 (2), 279−286.(11) You, S. J.; Park, E. D.Microporous Mesoporous Mater. 2014, 186,121−129.(12) Yang, Y.; Sun, C.; Du, J.; Yue, Y.; Hua, W.; Zhang, C.; Shen,W.; Xu, H. Catal. Commun. 2012, 24, 44−47.(13) Ding, W.; Meitzner, G. D.; Iglesia, E. J. Catal. 2002, 206 (1),14−22.(14) Ma, H.; Kojima, R.; Ohnishi, R.; Ichikawa, M. Appl. Catal., A2004, 275 (1−2), 183−187.(15) Zamani, Y.; Yousefian, S. H.; Nakhaeipour, A.; Moshtari, B.;Bahadoran, F.; Taheri, S. Proceedings of the 13th InternationalConference on Process Integration, Modelling and Optimisation forEnergy Saving and Pollution Reduction; Prague, Czech Republic, Aug28−Sept 1, 2010; Vol. 21, pp 1045−1050.(16) Bai, X.; Samanta, A.; Robinson, B.; Li, L.; Hu, J. Ind. Eng. Chem.Res. 2018, 57 (13), 4505−4513.(17) Bedard, J.; Hong, D. Y.; Bhan, A. J. Catal. 2013, 306, 58−67.(18) Xu, Y.; Wang, J.; Suzuki, Y.; Zhang, Z. G. Appl. Catal., A 2011,409−410, 181−193.(19) Ma, S.; Guo, X.; Zhao, L.; Scott, S.; Bao, X. J. Energy Chem.2013, 22, 1−20.(20) Burns, S.; Hargreaves, J. S. J.; Pal, P.; Parida, K. M.; Parija, S.Catal. Today 2006, 114 (4), 383−387.(21) Masiero, S. S.; Marcilio, N. R.; Perez-Lopez, O. W. Catal. Lett.2009, 131 (1−2), 194−202.(22) Liu, S.; Dong, Q.; Ohnishi, R.; Ichikawa, M. Chem. Commun.1997, 1455−1456.(23) Xu, Y.; Wang, J.; Suzuki, Y.; Zhang, Z. G. Catal. Today 2012,185 (1), 41−46.(24) Xu, Y.; Suzuki, Y.; Zhang, Z. G. Appl. Catal., A 2013, 452, 105−116.(25) Zhang, Y. P.; Wang, D. J.; Fei, J. H.; Zheng, X. M. React. Kinet.Catal. Lett. 2001, 74 (1), 151−161.(26) Aboul-Gheit, A. K.; Awadallah, A. E.; Aboul-Enein, A. A.;Mahmoud, A. L. H. Fuel 2011, 90 (10), 3040−3046.
(27) Lai, Y.; Veser, G. Environ. Prog. Sustainable Energy 2016, 35 (2),334−344.(28) Abdelsayed, V.; Smith, M. W.; Shekhawat, D. Appl. Catal., A2015, 505, 365−374.(29) Lu, Y.; Wang, D.; Song, Y.; Yang, Q.; Fu, M.; Yu, D.; Fang, Y.Synthesis of hierarchical-structured Zn/Mo-HZSM-5 and itsapplication in dimethy ether aromatization. Proceedings of the 2015Asia-Pacific Energy Equipment Engineering Research Conference; Zhuhai,China, June 13−14, 2015.(30) Liu, H.; Shen, W.; Bao, X.; Xu, Y. Appl. Catal., A 2005, 295 (1),79−88.(31) Mehdad, A.; Lobo, R. F. Dehydroaromatization of ethane overzinc containing zeolites. Proceedings of the 2016 AIChE AnnualMeeting; San Francisco, CA, Nov 13−18, 2016.(32) Fang, L.-Y.; Liu, S.-B.; Wang, I. J. Catal. 1999, 185 (1), 33−42.
