Bimetallic Mo-Co/ZSM-5 and Mo-Ni/ZSM-5 catalysts for methane dehydroaromatization: A study of the effect of pretreatment and metal loadings on the catalytic behavior

Article abstract of DOI:10.1016/j.apcata.2019.117247

Methane dehydroaromatization is studied using different loadings (0.2 wt%, 0.6 and 1 wt%) of Co and Ni additives on a 6 wt% Mo/ZSM-5 catalyst to evaluate the promoting effect of Ni and Co on reactivity and stability of the catalysts. A synergy between Mo and the additives is observed when combining a reduction/carburization pretreatment with an optimum additive loading: benzene yield and catalytic stability are both improved. The fresh and spent samples were characterized by X-Ray Diffraction, ²⁷Al Magic-Angle Spinning Solid State Nuclear Magnetic Resonance Spectroscopy, Temperature-Programmed Reduction and Thermogravimetric Analysis. The results suggest that catalytic enhancement in precarburized Mo-Co and Mo-Ni samples is due to a combination of effects: extraction of Al from the ZSM-5 framework to form inactive Al₂(MoO₄)₃ is reduced, interaction of Mo with the additives enhances the retention of Mo species within the zeolite channels, prevents Mo aggregation in reaction and modifies the reducibility of the Mo sites.

Full text of DOI:10.1016/j.apcata.2019.117247

Journal Pre-proof
Bimetallic Mo-Co/ZSM-5 and Mo-Ni/ZSM-5 Catalysts for Methane
Dehydroaromatization: A Study of the Effect of Pretreatment and Metal
Loadings on the Catalytic Behavior
Apoorva Sridhar, Mustafizur Rahman, Antonia Infantes-Molina,
Benjamin J. Wylie, Collin G. Borcik, Sheima J. Khatib
PII:
S0926-860X(19)30402-8
DOI:
https://doi.org/10.1016/j.apcata.2019.117247
APCATA 117247
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
11 June 2019
10 September 2019
12 September 2019
Please cite this article as: Sridhar A, Rahman M, Infantes-Molina A, Wylie BJ, Borcik CG,
Khatib SJ, Bimetallic Mo-Co/ZSM-5 and Mo-Ni/ZSM-5 Catalysts for Methane
Dehydroaromatization: A Study of the Effect of Pretreatment and Metal Loadings on the
Catalytic Behavior, Applied Catalysis A, General (2019),
doi: https://doi.org/10.1016/j.apcata.2019.117247
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
©
2019 Published by Elsevier.

Title: Bimetallic Mo-Co/ZSM-5 and Mo-Ni/ZSM-5 Catalysts for Methane
Dehydroaromatization: A Study of the Effect of Pretreatment and Metal Loadings on the
Catalytic Behavior
a
a
Author Names and Affiliations: Apoorva Sridhar , Mustafizur Rahman , Antonia Infantes-
b
c
c
a
Molina , Benjamin J. Wylie , Collin G. Borcik , Sheima J. Khatib
a
Department of Chemical Engineering, Texas Tech University, Lubbock, Texas 79409, United
States
b
Department of Inorganic Chemistry, Crystallography and Mineralogy, Faculty of Science,
Campus de Teatinos, 29071 Malaga, Spain.
c
Department of Chemistry, Texas Tech University, Lubbock, Texas 79409, United States
Corresponding Author: Sheima J. Khatib, sheima.khatib@ttu.edu
Graphical Abstract
1

Highlights






Zeolite-supported Mo carbide catalysts have been modified with Co and Ni additives
resulting in a more stable performance in methane dehydroaromatization.
A synergy is established between Mo and each additive, with increased benzene yield and
stability over time on stream compared to the single metal, unmodified Mo catalyst.
Each additive presents an optimum loading for 6wt% Mo, which results in the most stable
catalyst (0.6 wt% Co and 0.2 wt% Ni).
It is found that the effect of the presence of the additives is only beneficial when the Mo
carbide species are formed ex situ by previous carburization of the catalysts.
The additives interact with the Mo, decreasing its mobility and changing its reducibility,
thus affecting the reactivity of the Mo species present in reaction.
The presence of the additives improves the catalysts’ stability after various regeneration
cycles.
Abstract:
Methane dehydroaromatization is studied using different loadings (0.2 wt%, 0.6 and 1 wt%) of Co
and Ni additives on a 6 wt% Mo/ZSM-5 catalyst in to evaluate the promoting effect of Ni and Co
on reactivity and stability of the catalysts. A synergy between Mo and the additives is observed
when combining a reduction/carburization pretreatment with an optimum additive loading:
benzene yield and catalytic stability are both improved. The fresh and spent samples were
27
characterized by X-Ray Diffraction, Al Magic-Angle Spinning Solid State Nuclear Magnetic
Resonance Spectroscopy, Temperature-Programmed Reduction and Thermogravimetric Analysis.
The results suggest that catalytic enhancement in precarburized Mo-Co and Mo-Ni samples is due
to a combination of effects: extraction of Al from the ZSM-5 framework to form inactive
Al
2
(MoO
4
3
) is reduced, interaction of Mo with the additives enhances the retention of Mo species
within the zeolite channels, prevents Mo aggregation in reaction and modifies the reducibility of
the Mo sites.
Keywords: methane dehydroaromatization; benzene; 6wt% Mo/HZSM-5; Co-modified Mo
catalysts; Co-modified Mo catalysts; regeneration.
2

1
. Introduction
Natural gas is an abundant fossil fuel, mainly composed of methane, which is used as feedstock in
the chemicals manufacturing industry. The high transportation costs of a gas however, are causing
much of the newly discovered natural gas to be flared, resulting in the underutilization of an
otherwise valuable resource, as well as emission of CO into the atmosphere. It is therefore
2
desirable to develop processes to upgrade natural gas to added-value products in one step.
The routes for upgrading methane can be classified in two major groups: (i) indirect conversion
processes and (ii) direct conversion processes. Indirect conversion involves two steps, where
methane is first converted to syngas by partial oxidation or by reforming and in the second step,
syngas is converted to other valuable products such as alkenes, paraffins, higher alcohols,
methanol, dimethyl ether, by Fischer-Tropsch process or by methanol to hydrocarbon catalysis [1].
In the direct conversion routes, the intermediate syngas step is skipped, and methane is directly
converted to added-value products in a single step, thus reducing processing costs. Examples of
direct conversion include oxidative coupling of methane (OCM), direct conversion to oxygenates,
for example methanol [2, 3] and dehydroaromatization reaction to aromatics [4]. Since Wang et
al. [5] reported the first active catalyst in 1993 for non-oxidative methane dehydroaromatization
(
MDA), which converts methane to aromatic compounds and hydrogen, using MoO supported on
x
ZSM-5, extensive investigation has been carried out on zeolite-supported molybdenum oxide
catalysts [6-23]. However, detrimental coke formation which causes relatively fast deactivation of
the catalysts is a major hindrance towards commercialization of the MDA process. The commonly
accepted reaction pathway indicates that during the reaction induction period, the supported Mo
oxide species are carburized by methane to Mo oxycarbide and then carbide species which are then
responsible of activating the C-H bond of methane, which in turn leads to the formation of ethylene
that undergoes oligomerization on the zeolite Brønsted acid sites, eventually forming aromatics
[
24-26]. Recently it was demonstrated that benzene is only formed when the Mo oxycarbide
species are fully carburized to form unanchored Mo carbide clusters within the channels [27]. Over
time, the larger aromatic molecules that cannot leave the zeolite channels form carbon deposits
which, along with those formed due to methane cracking on the external surface sites, deactivate
the catalyst.
3

In an attempt to improve the catalytic performance, multiple studies have been performed by
adding other metal promoters to the Mo catalysts, with additives such as Cu [28][29], Cr [30], Zn
[
31,32], P [33], Ga [34,35] and Ag [31,36]. Although results are not conclusive and the
enhancement is short-lived, specifically Fe [37], Co [38-41] and Ni [42] are promising additives
in Mo-based catalysts employed in MDA. Aboul-Gheit et al. [37] tested 6Mo/ZSM-5, 3Mo-
3
Fe/ZSM-5, 3Mo-3Co/ZSM-5, and 3Mo-3Ni/ZSM-5 (numbers indicate wt %) with N
2
pretreatment for MDA and observed that adding Fe, Co, and Ni decreased the benzene yield
compared to the unmodified 6Mo/ZSM-5 catalyst. The decrease was in the order Fe > Co > Ni and
they explained it based on the deficiency in electrons in these elements, which causes them to form
+
a strong interaction with the H sites of the zeolite. In other words, strong metal-support
interactions cause the Brønsted acid sites, responsible for aromatization of the C species, to be
2
occupied and thus selectivity to benzene is inhibited. Thus, the results in this work suggest that
relatively high additive loadings do not have a promoting effect on the Mo catalysts since they
deplete the catalyst of the acid sites. Other groups found that presence of optimum additive
loadings, restricted to very low weight percent can be beneficial to the reaction. Vosmerikov et al.
[
42] tested catalysts consisting of 4wt% Mo supported on HZSM-5, modified with 0.1-2 wt% Ni
and found that indeed the catalyst with lower amount of Ni additive (0.1 wt%) exhibited the best
catalytic properties, while higher amounts of Ni resulted in fast catalytic deactivation caused by a
higher rate of formation of filamentous carbon deposits. The enhancing effect of Ni was explained
due to its higher hydrogenation character. Burns et al. [38] tested Co-modified catalysts containing
3
wt% Mo on HZSM-5 modified with Co using a Co/Mo molar ratio of 0.25. They observed an
initial boost in catalytic activity with Co however the effect was transitory. In separate papers, Liu
et al. [39] and Tang et al. [40] modified a 3 wt% Mo catalyst with 1 wt% Co and reported an
increase in benzene yield from 3.2% to 4.7% and 3.8% to 7.6% respectively, however no stability
data was reported. More recently, Fila et al. [41] tested Co-Mo/ZSM5 catalysts in MDA at high
pressures with Co loadings ranging from 0.4 to 1.6 wt% and found that the optimum loading of
Co for a 4wt% Mo/ZSM-5 catalyst was 0.8 wt%, rendering the highest and most stable total
benzene production. While all these studies show enhancement of catalytic benzene yield with
addition of specific amounts of additives, the improvements are temporary, and the catalysts still
deactivate after relatively short periods of reaction time.
4

Recently, our group discovered that HZSM-5 supported Mo carbides synthesized ex situ exhibit
better catalytic properties in MDA compared to supported Mo oxides, where the active carbide
species are formed in situ during the reaction induction period [43,44]. When Mo carbides are
formed in situ, they are formed by preheating the Mo oxide species in He up to reaction
temperature and then exposing them to CH reactant, which acts as carburizing agent. During the
4
heating stage in He and in the reaction induction period, the Mo oxide species, which are volatile,
have the chance to migrate and block the zeolite pores, before converting to non-volatile reduced
Mo species, thus causing faster catalyst deactivation than in carbides formed ex situ, where the Mo
carbide species formed under reduction at lower temperatures [43,44] are incapable of migrating
within the zeolite channels. Some groups have found that an additional contributing factor to
catalyst deactivation at high Mo loadings is the formation of Al
2
(MoO
4
3
) species due to extraction
of Al from the ZSM-5 framework when Mo loadings are high, exceeding 6 wt % [45-47].
When studying the effect of adding Fe to a ZSM-5 supported Mo oxide catalyst [48] we found that
adding 0.2 wt% of Fe to the precarburized catalysts further enhanced the benzene yield and catalyst
stability. While no mixed Mo-Fe phases were detected, it is possible that the ability of Fe to form
carbidic phases makes it an effective additive for Mo-based catalysts employed in MDA. Based
on this assumption, Co and Ni, which are elements within the iron group (first row in Group VIII)
in the periodic table, could also interact with Mo carbide in a similar way, and potentially serve as
promoters in Mo/ZSM-5 catalysts.
As a natural extension of our previous work, here our aim is to investigate the effect of adding
specific amounts of Co and Ni on ZSM-5 supported Mo catalysts employing both He pretreatment
and precarburization. To remain consistent with our past work [48], we have prepared catalysts
using the same Mo loading, 6 wt% Mo, and low additive loadings, specifically, 0.2, 0.6 and 1 wt%
of Co and Ni.
2
. Experimental
2
.1. Catalyst Preparation
Commercial ZSM-5 (Si/Al=15, Zeolyst International) was used as the support for the catalysts.
The commercial ZSM-5 was first calcined at 500°C with ramp rate of 2°C/min for 6 hours to
convert the zeolite powder from its ammonium form to its protonated form. Molybdenum oxide
was supported on the zeolite using the conventional incipient wetness impregnation method.
5

Ammonium heptamolybdate tetrahydrate (Sigma Aldrich), ferric nitrate nonahydrate (Sigma
Aldrich), cobalt nitrate (Sigma Aldrich) and nickel nitrate (Sigma Aldrich) were used as the metal
precursors for Mo, Fe, Co, and Ni respectively. Sequential impregnation method was employed to
prepare the ZSM-5 supported mixed Mo-X (X = Fe, Ni, Co) catalysts using the above-mentioned
precursors. The resulting samples were then dried overnight and calcined in the furnace at 110°C
with a ramp rate of 1°C/min and held at this temperature for 2 hours, followed by calcination at
5
00°C at 1°C/min for 5 hours. The catalysts that were prepared contained 6 wt% Mo and different
amounts of metal additive (0, 0.2, 0.6, and 1 wt%) and are respectively designated as 6Mo, 6Mo-
0
.2X, and 6Mo-1X (X=Co or Ni).
2
.2. Catalyst Reactivity
The catalysts were tested in a fixed-bed tubular quartz reactor with 8 mm inner diameter and a
length of 300 mm. The catalyst bed consisted of 300 mg of sample with particle size within the
range of 125-250 microns, held in place between two quartz wool plugs. The samples were heated
to 700°C with a 5 or 10°C/min temperature ramp, under specific gas flow rates which varied based
on the pretreatment condition employed (described in the following paragraph); after temperature
equilibration at 700°C was achieved, reaction conditions were applied by flowing 91% CH
4
/N
2
into the reactor at atmospheric pressure and at a space velocity of 1500 ml/gh. With nitrogen as
an internal standard, the products leaving the reactor were analyzed using an online gas
chromatograph equipped with flame ionization detector and thermal conductivity detector
(Shimadzu).
Two pretreatment methods were tested: (a) heating the catalyst to 700 °C under He flow at a ramp
rate of 10 °C/min (b) reducing the catalyst with a CH /H (1:9, v/v) mixture to 700 °C at a ramp
rate of 5 °C/min, and then carburizing for 10 mins under pure CH at 700 °C. We will refer to this
4
2
4
second pretreatment method as “precarburization”. The samples that were pretreated in He have
been denoted as “sample name – He”.
2
.3. Catalyst Regeneration
For the regeneration study, the catalysts were exposed to reaction conditions for 10 hours and then
the gas flow was switched to 95 ml/min of H flow for 1 hr at atmospheric pressure. The catalysts
2
were exposed to reaction conditions for 430 minutes after the first regeneration.
6

2
.4. Catalyst Characterization
Both fresh and spent catalysts were characterized by the following characterization methods.
.5. N Adsorption-Desorption Isotherms
2
2
Textural characteristics of the samples was determined by adsorption of nitrogen at -196°C using
Quantachrome Autosorb IQ. The sample was outgassed at 350°C for a minimum of 2 hours and
then cooled to -196°C. Adsorption isotherms were then recorded as a function of relative pressure
and specific volume of nitrogen adsorbed at Standard Temperature and Pressure (STP). The
surface area of the samples was determined using the Brunauer-Emmett-Teller (BET) method.
2
.6.X-Ray Diffraction of the Catalysts (XRD)
X-Ray diffraction patterns of the catalyst were obtained using Rigaku Ultima III PXRD with CuKα
radiation at 40kV and 20mA. The data was recorded at 2θ between 5° to 60° at a scan speed of
0
.25°/min.
2
.7.Temperature Programmed Reduction (TPR)
TPR was carried out on fresh catalysts before pretreatment using Quantachrome-Autosorb IQ. In
order to maintain consistency, 0.0700 g of the sample was heated to 500°C at 20°C /min under He
flow, and held for 30 mins, after which the sample was brought back to room temperature. The gas
was switched from He to 5% H
2
/He, and the sample was heated to 1000°C at 10°C /min. The data
was recorded using an in-built TCD.
2
.8.Thermogravimetric Analysis (TGA)
TGA was performed on the spent catalysts using METTLER TGA/SDTA851e. 0.015 g of sample
was heated to 1000 °C at 10 °C /min under airflow of 50 ml/min. The amount of coke in the spent
catalyst was calculated as follows [49]:
푀1−푀2
Coke amount (mg/g. cat) =
× ꢀ000
푀2
Where M1 denotes the weight percent of the coked catalyst before coke burn-off, and M2 denotes
the weight percent of the coked catalyst after coke burn-off.
7

2
7
27
2
.9. Al Magic-Angle Spinning Nuclear Magnetic Resonance ( Al MAS-NMR)
All samples were packed into standard wall 3.2 mm pencil rotors (Agilent Technologies, Santa
2
7
Clara, CA and Loveland, CO) to a standard volume of 22 L. All Al spectra were recorded on
an Agilent DD2 600 MHz (14.1 T) spectrometer (Agilent Technologies, Santa Clara, CA and
2
7
Loveland, CO) equipped with a BALUN 3.2 mm probe. At this field strength the Al resonance
2
7
frequency is 156.4 MHz and the Al signal of an aqueous 1 M Al(NO
universal chemical shift reference. The magic-angle spinning (MAS) rate was maintained at
5,000 ± 3 kHz. The sample temperature setpoint was 10 °C, corresponding to a sample
3
) solution was used as a
3
1
2
7
temperature of 16 ± 2 °C. All Al spectra were acquired with a /12 excitation pulse (1.2 s).
1
1
Spectra were acquired without high powered decoupling on H after it was confirmed H
decoupling minimally impacted the observed spectra. All spectra were processed using NMRPipe
by Fourier transformation after apodization with a sinebell window function. All peaks were
selected and integrated using nonlinear line shape fitting functionality within NMRPipe. Because
2
7
each rotor contained slightly different total moles Al, each with different relative coordination
environment populations, all spectra were normalized using the sum of all peak integrated
1
intensities. H spectra were also acquired to confirm all samples were hydrated after exposure to
high temperatures (data not shown).
2
.10. Scanning Electron Microscopy (SEM)
The surface morphology of the samples was assessed using Hitachi-H-800 scanning electron
microscope. Fresh catalyst samples were analyzed at 8-10kV, while the spent samples were
analyzed at 8kV.
2
.11. X-Ray Photoelectron Spectroscopy (XPS)
X-Ray Photoelectron spectra (XPS) were collected using a Physical Electronics PHI 5700
spectrometer with non-monochromatic Mg Kα radiation (300W, 15 kV, 1253.6 eV) for the
analysis of the core level signals of C 1s, O 1s, Ti 2p and Au 4f and with a multi-channel detector.
Binding energy (BE) values were referenced to the C 1s peak (284.8 eV) from the adventitious
contamination layer. The spectrometer energy scale was calibrated using Cu 2p3/2, Ag 3d5/2, and
Au 4f7/2 photoelectron lines at 932.7, 368.3, and 84.0 eV, respectively. The PHI ACCESS ESCA-
V6.0 F software package and Multipak v8.2b were used for acquisition and data analysis,
respectively. A Shirley-type background was subtracted from the signals. Recorded spectra were
8

always fitted using Gauss-Lorentz curves, in order to determine the binding energy of the different
element core levels more accurately. The error in BE was estimated to be ca. 0.1 eV.
3
. Results and Discussion
3
.1. Catalytic Results
Figures 1 (a) and (b) represent the benzene yield obtained over time on stream (TOS) for all the
precarburized Mo-based catalysts (red symbols) with 6 wt% Mo and different amounts of a second
metal additive, X (X= Co or Ni). To serve as reference, the data points corresponding to He
pretreated catalysts have also been included (blue symbols). Based on our past studies monitoring
evolution of Mo species from oxides to carbides, the induction period of Mo catalysts occurs within
the first 40 mins of reaction [43-48]. Similarly, the data points represented in Figure 1 after 40
mins correspond to the catalytic activity after the induction period.
It is clear that the reduction/carburization pretreatment enhances the benzene yield obtained with
each catalyst, with respect to its He-pretreated counterpart, regardless of the nature and amount of
the additive metal loading, much in line with what was observed when using Fe as additive [48].
As we have demonstrated in our past work, this enhancement is due to a higher dispersion of the
Mo species when the catalysts are precarburized [43,44].
The effect of the additive type and loading on benzene yield varies depending on the type of
pretreatment received. For the He-pretreated catalysts (Figure 1, blue symbols), adding 0.2. 0.6
and 1 wt% Co reduces the benzene yield; adding 0.6 and 1% Ni reduces it, and 0.2 wt% Ni hardly
affects the yield. In other words, no overall catalytic enhancement is observed by addition of Co
or Ni when the catalysts are pretreated in He. The behavior of the precarburized catalysts follows
a different trend: precarburization of 6Mo increases the benzene yield and its stability over time
compared to its He pretreated version. Adding Co or Ni enhances the benzene yield even further,
but the effect depends on the loading employed for each additive. In the case of Co, adding 0.2wt%
Co (6Mo-0.2Co catalyst) hardly changes the activity compared to 6Mo, however when its loading
is increased to 0.6 wt%, Co enhances benzene yield over the entire reaction period time. Upon
increasing Co loading to 1 wt% the overall enhancement is not significant compared to 6Mo-
0
.6Co; initially, and up to 150 mins into reaction, no enhancement in benzene yield is observed
compared to 6Mo, and after that time, the benzene yield remains stable while that of 6Mo drops
slightly, increasing the difference in benzene yield between the two catalysts over time. When Ni
9

is used as additive, 0.2 and 0.6 wt% Ni loading result in very similar overall benzene yield
enhancement. When Ni loading is increased to 1wt%, the effect is similar to the case of 1wt% Co:
no yield enhancement is observed during the first minutes of reaction after the induction period,
however over longer reaction times, the difference between 6Mo and 6Mo-1Co becomes larger
due to the higher stability of the latter.
These results indicate that, as in the case of Fe [48], there are optimum Co and Ni loadings that
provide higher and more stable benzene yields, and this amount varies depending on the nature of
the additive. In the case of Ni, a loading as small as 0.2 wt% benefits the Mo catalyst, whereas for
Co, a slightly higher amount, 0.6 wt% is required to optimize benzene yield. These numbers are
similar to the ones reported in the literature; Fila et al. [41] point to 0.8 wt% of Co as optimum
additive loading for a 4 wt% Mo/ZSM-5 catalyst, and Vosmerikov et al. [42] found that a lower
loading of Ni additive (0.1% in 4 wt% Mo catalyst) had a higher enhancing effect compared to
higher Ni loadings. These works however did not involve precarburization of the catalysts, thus
the enhancement observed was temporary and the catalyst deactivation problem was not overcome,
with benzene yields eventually decreasing after several hours.
Clearly catalyst stability is an important feature of the modified precarburized catalysts. To analyze
the deactivation rates, the first order deactivation constants, k , for all the catalysts were calculated
D
from the slopes of the semilogarithmic plot of the rate of formation of benzene as a function of
TOS (shown in Figure S1 in the Supplementary Information section) from the moment that the
rate of formation of benzene begins to drop. The k
summarized in Table 1. (The semilogarithmic plots and k
pretreated catalysts are shown in Figure S2 and Table S1 respectively.) As would be expected,
the k values are higher in the case of the less stable, He-pretreated catalysts. As for the
precarburized catalysts, the effect of the addition of both Co and Ni is evident on the k values,
which become smaller when the additives are present. Moreover, higher loadings of each additive
promote the catalytic stability to a larger extent as is evidenced by the smaller k values obtained
for 6Mo-0.6Co, 6Mo-1Co, 6Mo-0.6Ni and 6Mo-1Ni. This phenomenon could be explained by
secondary methane cracking reactions caused by the additives, Co and Ni, where H is formed and
D
values for the precarburized catalysts are
D
values obtained for selected He-
D
D
D
2
is capable of gasifying part of the carbon deposited in reaction in situ, thus increasing catalyst
stability.
1
0

Table 1. First order deactivation rate constants (k ) for precarburized catalysts.
D
Deactivation rate constant (k )  10-4
Catalyst
Mo
D
6
3.0
2.0
0.1
0.2
2.0
0.7
0.9
6
6
Mo-0.2Co
Mo-0.6Co
6
Mo-1Co
Mo-0.2Ni
Mo-0.6Ni
6
6
6
Mo-1Ni
The evolution of CH
4
conversion and the selectivity to different products (other than benzene)
with TOS for the precarburized and He-pretreated catalysts are represented respectively in Figures
S3 and S4. Apart from benzene, formation of ethylene, toluene and naphthalene are also observed,
although the selectivities to these products are low compared to benzene. It is worth noting that
the catalysts presenting lower selectivity to aromatics show higher ethylene selectivity, indicating
that although they present a lower overall selectivity to aromatics, they still possess active Mo-C
sites capable of activating CH and leading to C-C coupling.
4
Analyzing the activity of the single-metal additive catalysts, Co/ZSM-5 and Ni/ZSM-5 (Tables S2
and S3 respectively), can help to understand how the additives contribute to the overall catalytic
activity of the Mo-Co and Mo-Ni catalysts. It is observed that all the single metal Co catalysts,
0
.2Co, 0.6Co and 1Co are capable of activating CH
measured (2% CH conversion); a study of the yield to other products however shows absence
of products formed from C-C coupling: ethylene or aromatics, therefore Co is only capable of
cracking CH to H and coke in the absence of Mo. This phenomenon can explain why the CH
conversion values obtained after the induction period for the Mo-Co catalysts, within the range of
2-16%, are higher than thermodynamic equilibrium conversion (12%) calculated theoretically for
4
since CH
4
conversion values > 0% are
4
4
2
4
1
conversion of methane to benzene [50]. Despite the incapability of the Co species to produce
benzene in the single-metal Co/ZSM-5 catalysts, when they are added to 6Mo in the Mo-Co
catalysts (red symbols in Figure 1 (a)), their presence results in an enhancement in benzene yield,
1
1

therefore, the combination of Co with Mo has a synergistic effect on the production of benzene
and other aromatics at 0.2 and 0.6 wt% of Co. At 1 wt% loading the enhancement is simply due to
the higher stability conferred to the catalyst and is mostly evident after four hours of reaction. A
similar analysis can be performed with the Ni catalysts. For the single metal Ni catalysts, 0.2Ni
does not produce any ethylene or aromatics (Table S3), yet the addition of 0.2 wt% of Ni to 6Mo
(6Mo-0.2Ni) produces an enhancement in benzene yield. For the 0.6Ni catalyst benzene and some
naphthalene begin to form after around 470 mins of time on stream, yet the benzene yield obtained
with the 6Mo-0.6Ni catalyst is higher than that of 6Mo from the early stages of reaction after the
induction period, and gradually the difference in benzene yield increases, partly due to an additive
contribution of benzene formed by Ni species. The fact that the benzene yield in the Mo modified
catalysts, 6Mo-0.2Ni and 6Mo-0.6Ni, is higher than the sum of the benzene yields obtained
separately with 6Mo and 0.2Ni and 0.6Ni respectively during the first minutes of reaction after the
induction period, is indicative that here also, a synergy exists between Mo and Ni in the Mo-Ni
catalysts, that results in an enhancement in the benzene yield.
3
3
.2. Structural Characterization of the As-synthesized, Pretreated and Spent Catalysts
.2.1. X-Ray Diffraction (XRD)
To identify the structure of the crystalline phases in the catalysts, XRD patterns of the as-prepared,
pretreated and spent catalysts were acquired and are presented in Figure 2 (a)-(e). (The XRD
patterns within the range 2θ= 5-50° are represented in Figure S5.)
Two important aspects can be highlighted generally: (1) ZSM-5 diffraction peaks appear in all the
XRD patterns of the fresh and spent catalysts, however, in all the cases, the intensity of the ZSM-
5
peaks within the 2θ= 5-10° range is slightly weaker compared to the pure ZSM-5 reference due
to incorporation of the metal species in the micropore channel structure of the zeolite [51]; (2) all
figures show that the addition of a second metal does not result in the detection of any new mixed
Mo-Co or Mo-Ni crystalline phases regardless of the type of pretreatment employed. The absence
of peaks corresponding to mixed Mo-Co and Mo-Ni phases does not prove their non-existence,
however, if they are present, they are probably in the form of small clusters that are not detectable
by XRD. It is important to note that since Mo species inside the zeolite channels are too small to
be detected by XRD, it is reasonable to assume that any Mo phases detected here reside on the
external surface of the zeolite.
1
2

The XRD patterns of the fresh as-synthesized catalysts are shown in Figure 2 (a). The peak at 2θ=
7.3° corresponds to crystalline MoO (JCPDS File No. 05-05058). This peak partially overlaps
with the weak ZSM-5 peak at 27.5°, however its relative high intensity with respect to the ZSM-5
peak appearing at 26.9° indicates the additive effect of the ZSM-5 peak with that of MoO crystals
when both phases coexist. The intensity of the MoO peak decreases generally with the addition
2
3
3
3
of a second metal (Co or Ni), and the effect is bigger as the loadings of the additives become larger.
This phenomenon probably occurs due to an interaction of the second metal with the Mo oxide,
although it is not possible to distinguish whether the effect is caused by a decrease in the
crystallinity of the Mo oxide by distortion caused by interaction of the additive with the MoO , or
3
a reduction of the quantity of Mo oxide species residing on the external surface of the zeolite, or
both TPR measurements of the as-synthesized catalysts before pretreatment (discussed in section
3
.2.2. Temperature Programmed Reduction (TPR) below) provide more structural information that
can further explain these patterns.
The XRD patterns of the pretreated catalysts are represented in Figures 2 (b) and (c). The patterns
of the He-pretreated catalysts are similar to those of the as-synthesized catalysts although the
intensity of the peak corresponding to MoO
higher dispersion of the Mo species in the 6Mo catalyst. No peaks corresponding to
extraframework Al (MoO species are observed in the XRD patterns [47], indicating that if
zeolite dealumination has occurred, it has not happened to a great extent. Notwithstanding, we
3
decreases relative to the ZSM-5 peak, pointing to a
2
)
4 3
2
7
have performed Al-MAS-ss NMR measurements to detect formation of extraframework Al
2
7
species (see section 3.2.3. Al-MAS-NMR). Most of the precarburized catalysts exhibit a broad
peak around 2θ= 39.5°, attributed to Mo C (JCPDS File No. 35-0787), while the MoO peak is
absent in all of them, indicating that during precarburization, the crystalline Mo oxide phases
present on the external surface of the catalyst are reduced and carburized. The absence of Mo
2
3
2
C
peaks in 6Mo-0.6Co and 6Mo-0.6Ni is probably due to better dispersion of the carburized Mo
species in these samples. No other mixed Mo-Co or Mo-Ni phases are detected. Comparing the
XRD patterns of the spent He pretreated and precarburized catalysts, in Figures 2 (d) and (e), it is
observed that for the spent He pretreated catalysts the MoO
3
peak completely disappears and
instead a broad Mo C peak appears. This is expected since it has long been agreed among
2
researchers that during the induction period of the reaction the Mo oxide species convert to Mo
carbide species that are responsible for aromatization to occur [13,24-26]. In the case of the
1
3

precarburized catalysts, Mo
the Mo C peaks become more intense when the additive is at its highest loading, i.e. 1 wt%,
indicating sintering of the Mo C phases located on the external surface of the zeolite.
2
C peak was already present in the fresh catalysts, and after reaction,
2
2
3
.2.2. Temperature Programmed Reduction (TPR)
TPR was performed on the fresh as-synthesized oxide catalysts to verify the effect of the presence
of a second metal, Co or Ni, on the reduction profile of the catalysts before receiving any type of
pretreatment. The profiles present different reduction features in the 300–900 °C range depending
on both additive type and loading (Figure 3).
All the catalysts exhibit three distinct peaks in the range of 400 °C-810 °C (Figure 3 (a) and (b)).
Moreover, the weak reduction feature between 300-400 ºC is related to the partial reduction of
amorphous polymeric molybdate species [52]. This contribution is similar for all samples.
Previous research has shown that bulk MoO
3
reduction happens in a two-step process where MoO
3
6
+
4+
4+
0
gets reduced to MoO
2
(Mo Mo ) and MoO
or Al
2
is further reduced to Mo (Mo Mo ) [53].
When Mo oxides are supported on SiO
2
2
O
3
the TPR profiles change however [54, 55], due
to the interactions of the Mo oxide species with the support. Based on previous TPR studies of
supported Mo oxides [54-59], for the catalysts employed in the current work, the TPR peak
appearing at a lower temperature (510 °C) can be assigned to the complete reduction of amorphous,
poly-molybdate species that are non-uniformly distributed on the external surface of the zeolite
6
+
0
6+
(
Mo Mo ), whereas the peak at 580 °C can be assigned to the partial reduction of Mo oxide
4
+
species associated to Brønsted acid sites, located inside the zeolite channels, to Mo oxide species
6
+
4+
4+
0
(
Mo Mo ) [59]. The peak at 710°C is attributed to the reduction of Mo Mo . The fact that
the area underneath the 710 °C peak is not double that of the area under the lower temperature
peaks indicates that not all the species are fully reduced by the end of the TPR experiment and that
a sequential reduction process is not occurring.
Addition of promoters to 6Mo/HZSM-5 changes the position of the reduction peaks of the
catalysts, as observed in Figure 3. To analyze the effect of the additives on Mo, the TPR profiles
of monometallic Co and Ni samples were also measured. The Co catalysts did not consume
hydrogen (profiles not shown), which could be ascribed to highly dispersed Co particles of small
size as observed for other metals [60]. On the other hand, Ni samples show a broad reduction
1
4

profile with a maximum centered at 410-420 ºC, which is associated to reduction of unbounded
NiO [61] (profile shown in Figure S6).
In the case of Co-modified Mo catalyst (Figure 3 (a)), the low and high temperature peaks shift
slightly to higher temperatures upon adding Co (from 510 ºC to 535ºC; and from 710 °C to around
8
10 °C, respectively), however the position of the peak at 580 ºC hardly changes, although its
intensity does increase. Based on the TPR peak assignment of Hongmei et al. [59], the results
presented here suggest that the presence of Co affects the reducibility of the Mo oxide species on
the external surface of the zeolite, while promoting the retention of Mo oxide species inside the
zeolite channels. Since the latter are inside a confined space, with channels of ZSM-5 being 5 Å
in diameter, they are not detectable by XRD, explaining thus the lower intensity of the MoO
observed in the XRD patterns of the Co-modified catalysts.
3
peak
In the case of Ni-modified Mo catalysts (Figure 3 (b)), at increasing Ni loadings the reduction
peaks at 510 and 580 ºC shift to lower temperature, thus the presence of Ni promotes the
reducibility of MoO
presents an important increase in the intensity of the band at 580 ºC, that now presents a shoulder
at ca. 535 ºC. Therefore, a greater proportion of MoO species inside the ZSM-5 channels seem to
be present in this catalyst. Further Ni addition shifts MoO reduction bands to lower temperatures
that overlap with that of NiO reduction (see Figure S6), indicating that MoO reduction is
improved under Ni presence. Moreover, the contribution due to MoO inside zeolite channels
3
species. It is that at 710 ºC that remains unaltered. 6Mo-0.2Ni sample
3
3
3
3
shows a higher peak intensity and therefore, as in the case of Co, indicates that Ni additive is
capable of improving Mo dispersion. These data are in accordance with those obtained from XRD
of the as-synthesized catalysts (Figure 2 (a)).
The TPR profiles therefore corroborate the fact that the Mo oxide species are electronically
modified by the presence of Co and Ni. While Co addition seems to improve MoO
3
dispersion
inside the zeolite channels, Ni presence improves MoO dispersion and affects its reducibility.
3
3
.2.3. N
2
Adsorption Isotherms and BET Surface Area
The N adsorption isotherms of the fresh and spent catalysts are represented in Figures S7 and S8.
2
All the catalysts exhibit Type IV isotherms, indicating presence of micropores with some
mesopores. Table 2 summarizes the BET surface area of the as-synthesized, pretreated and spent
catalysts. Although the pretreated catalysts do show slightly lower surface areas compared to their
1
5

as-synthesized counterparts, these differences fall within the error range of the instrument,
discarding any specific trend. Moreover, the type of pretreatment received for each catalyst loading
does not modify the surface area significantly. The addition of any amount of Ni and Co leaves
the surface area of the as-synthesized catalysts insignificantly changed. A clear trend however is
observed when comparing the % drop in surface area from pretreated to spent catalysts: the drop
is generally higher for the He-pretreated catalysts compared to the precarburized ones, for
example, 6Mo-0.2Co shows a drop in surface area of 36 % when it is He-pretreated, versus 9 %
when it is precarburized. The trends in the % drop in surface areas from pretreated to spent catalysts
correlate with the difference in catalytic stability of the two groups of catalysts, where
precarburized catalysts are more stable (as observed earlier when comparing the k
D
values in Table
1
and Table S1). It is noteworthy that for the He-pretreated samples, the addition of Co and Ni
result in a higher surface area of the spent catalysts compared to the unmodified 6Mo catalyst: the
latter undergoes a 61% reduction in surface area, versus 41% when 0.2 wt% Co is added (6Mo-
0
.2Co) or 26% when 0.2wt% Ni is added (6Mo-0.2Ni). The observed differences in the % drop in
surface area are in line with the stability trends observed in the activity data for these catalysts.
When 1 wt % of additive is present, in 6Mo-1Co and 6Mo-1Ni, the fall in surface area from fresh
to spent sample is almost the same for both kinds of pretreatment. There are various possible
scenarios that could explain the differences in drop in surface area and catalyst deactivation for the
two groups of catalysts: (1) the Mo could extract Al from the zeolite framework to form
Al
2
(MoO
4
3
) with a consequent collapse of the zeolite structure. The following sections address
these possibilities; (2) the zeolite pores could be blocked by formation of carbon deposits. The
following sections will address these two hypotheses by measuring the coordination of Al species
formed after pretreatment (by 27Al MAS-ssNMR in section 3.2.4.) and by quantifying the coke
formed in reaction (by TGA in section 3.2.5.) respectively.
Table 2. BET surface areas of the as-synthesized, pretreated and spent catalysts. Numbers
in parenthesis indicate the % drop in surface area from fresh to spent catalyst. (The spent
catalysts underwent reaction at 700°C under 91% CH
4
/N with a space velocity of
2
1
500ml/g·hr for 10 hrs.)
He-pretreated
As-synthesized
Catalysts
Spent He-pretreated
Precarburized
Spent precarburized
2
2
2
2
catalyst* (m /g)
2
catalyst (m /g)
catalyst (m /g)
catalyst (m /g)
catalyst (m /g)
1
6

6
Mo
386
357
333
350
350
347
362
362
328
318
329
353
366
370
150 (59%)
209 (36%)
281 (12 %)
295 (10%)
276 (22%)
279 (11 %)
272 (26%)
354
332
349
330
337
353
360
284 (20%)
303 (9%)
6
6
Mo-0.2Co
Mo-0.6Co
313 (10%)
298 (15%)
280 (17%)
298 (16%)
276 (23%)
6
Mo-1Co
Mo-0.2Ni
Mo-0.6Ni
6
6
6
Mo-1Ni
*
Before pretreatment
2
7
3
.2.4. Magic-Angle Spinning Solid-State NMR ( Al MAS-ssNMR)
When extraframework Al
2
(MoO species that are inactive in MDA are formed on Mo/ZSM-5
4 3
)
catalysts, the process is accompanied by a collapse of the ZSM-5 zeolite structure, evidenced by a
drop in the BET surface area, a decrease in the intensity of the XRD ZSM-5 peaks, and fast catalyst
deactivation. Our BET data (Table 2) do not show a difference in the surface area of the catalysts
undergoing the different types of pretreatment and furthermore, the intensity of the ZSM-5 peaks
do not decrease significantly upon pretreatment, discarding the possibility of a collapse of the
ZSM-5 framework in our deactivated catalysts. Yet, it is important to determine if Al
2
4 3
(MoO )
species are formed, even if the quantities are low, since they can still contribute to catalyst
deactivation by forming at the expense of the MDA-active Mo species that were originally
2
7
anchored to the ZSM-5 framework. Al MAS-ssNMR was conducted to investigate the changes
in coordination of the Al species present in the 6Mo catalyst upon calcination, He pretreatment
and precarburization and the spectra presented in Figure 4 show that the type of pretreatment
strongly modifies the MAS-NMR spectrum. Table 3 presents the chemical shift, relative
population based upon integrated intensity (volume), and proposed geometry of each peak. All
the samples present peaks near 0, 14 and 54 ppm. We attribute these peaks to octahedrally
coordinated extraframework aluminum, hydrated Al
2
(MoO
4
3
) and tetrahedrally coordinated
framework aluminum respectively [45-47,62]. A weak peak at 32 ppm also appears in all samples,
showing its highest intensity in the He-pretreated sample. This peak has been assigned to penta-
coordinated or highly distorted tetrahedrally coordinated extraframework aluminum [63, 64]. This
1
7

same sample also presents an additional peak at -14 ppm, which has also been assigned to
Al (MoO [47].
2
)
4 3
2
7
The relative populations of extraframework Al increase in both pretreated catalysts relative to
the aluminum population still observed within the zeolite framework. In the case of the
precarburized 6Mo catalyst, the relative population of Al
2
(MoO
4
3
) increases from 12% to 15%.
2
7
A greater diversity of extraframework Al is observed in the He-pretreated catalyst, including a
2
7
unique strongly upfield shifted population of octahedrally coordinated Al at -14 ppm, with the
population of Al (MoO increasing up to 24%. Thus the data suggest that with He-pretreatment
2
)
4 3
more Al sites are extracted from the zeolite framework and while the BET and XRD results suggest
that this extraction does not happen to a significant extent, the NMR spectra indicate a clear
difference in the degree of extraction for the different pretreatment conditions, which can explain
the differences observed in the catalytic stability.
Table 3. Chemical shift and relative population based upon integrated intensity (volume), of
2
7
each peak observed in the Al MAS-ssNMR spectra of 6Mo catalysts
Catalysts
Chemical shift Relative Population (%)
ppm)
(
As-synthesized catalyst
5
4.7
4.4
9.3
4.4
59.2
9.4
3
2
1
1.0
12.3
18.0
33.9
14.7
5.5
-0.2
He-pretreated catalyst
5
3
3
1
3.8
6.0
1.2
4.1
17.8
21.6
6.6
-1.1
-13.1
Precarburized catalyst
5
3.8
0.3
8.4
46.2
16.2
3.2
4
2
1
8

1
4.6
15.3
19.2
-0.9
Since the drop in surface area in the deactivated catalysts is not explained by the dealumination, it
is necessary to test the second hypothesis, that pore blocking is caused by formation of coke. The
latter is analyzed by thermogravimetric analysis.
3
.2.5. Thermogravimetric Analysis (TGA)
To analyze the amount of carbon deposits present in the spent catalysts, TGA measurements were
performed; the TGA profiles are shown in Figures S9-S11. All the profiles exhibit an initial weight
loss below 250 °C attributed to the loss of absorbed ambient humidity [65]. At temperatures above
3
70 °C, the profiles present a slight weight gain attributed to the oxidation of the Mo carbide
species in air to MoO , which has a higher molecular weight compared to Mo C [66]. In the
3
2
samples studied here, other metals (Co or Ni) coexist with Mo and it is likely that they are either
reduced or forming a mixed phase with the Mo carbide species, therefore the weight gain cannot
be attributed only to formation of MoO . Weight loss observed between 480 °C and 680 °C
3
corresponds to the burn-off of coke that has deposited during reaction [67]. The latter was
quantified according to reference [49] in the Experimental section and values are shown in Table
4
.
Table 4. Amount of coke burnt off from the spent precarburized and He-pretreated catalysts.
The spent catalysts underwent reaction at 700°C under 91% CH /N with a space velocity
of 1500ml/g·hr for 10 hrs.)
(
4
2
Amount of coke burnt Amount of C formed in
off in spent He- reaction in spent
pretreated catalysts precarburized catalysts
Catalyst
(mg/g catalyst)
(mg/g catalyst)
6
Mo
Mo-0.2Co
Mo-0.6Co
Mo-1Co
50.9
44.8
31.4
53.5
83.3
83.6
6
6
73.8
6
120.3
1
9

6
6
Mo-0.2Ni
Mo-0.6Ni
42.4
66.3
79.0
94.2
6
Mo-1Ni
107.1
121.2
When comparing catalysts with the same type of additive, the general trend observed is that, both
in He-pretreated and precarburized samples the amount of deposited carbon increases with additive
loading, i.e. from 0.2 wt% to 1wt%, which further corroborates that the additives probably
contribute to reaction by cracking of CH to form coke.
4
Comparing the amount of carbon deposited during reaction in the precarburized catalysts versus
the He-pretreated ones, it is evident that there is always more coke formation in the former ones,
which are interestingly the more stable catalysts. In other words, the quantity of carbonaceous
deposits formed in reaction does not correlate with the decrease in surface area of the spent
catalysts, refuting the hypothesis that pore blocking and consequent deactivation are caused by
formation of carbon deposits. It is possible that the changes in surface area after reaction are due
to a phenomenon that is unrelated to carbon. Our previous work comparing Mo carbide species
formed ex situ (precarburized samples) versus those formed in situ (He-pretreated samples) points
to more Mo aggregation in He-pretreated catalysts [43,44], however this work was carried out
using extremely high Mo loading (10 wt%). Here, the Mo loading is almost half, 6 wt%, and we
do not observe a significant change in the BET surface area after pretreatment. In fact the surface
area does not change much between treatment in He or in CH . But Mo aggregation during reaction
4
cannot be discarded, since in the He-pretreated catalyst the mobile Mo oxide species are present
at the beginning of reaction when the temperature is high (700 C), while the precarburized
catalysts only present less mobile Mo carbide species when exposed to pure methane.
Furthermore, this argument can be extended to the precarburized additive-modified catalysts in
this work, by speculating that the presence of Co and Ni additives can further reduce the mobility
of the Mo species while they are being precarburized. In order to test this hypothesis, TGA of the
fresh as-synthesized Mo-Co and Mo-Ni catalysts was performed to evaluate whether the presence
of the additives was impacting the volatility and mobility of the species at high temperatures. The
TGA profiles of the in Figure 5 show a drop in weight at temperatures above 700 C,
2
0

corresponding to volatilization of the metal species. The temperature at which this happens, shown
in Table 4, varies depending on the amount of additives employed. Clearly a trend is observed:
the presence of both additives increases the temperature, which becomes higher as additive loading
increases, suggesting that both Co and Ni interact with the Mo species, decreasing their mobility
and volatility, and that this interaction is stronger as additive loading increases. The effect is more
marked with Ni additive versus Co. This finding suggests that by adding Co and Ni, these species
are capable of retaining the Mo species at higher temperatures, thus reducing possible sintering
within the channels and their consequent blocking, contributing to a higher stability in the
formation of benzene over reaction time. This also explains why no big drop in surface area is
observed for the He-pretreated 6Mo-1Co and 6Mo-1Ni catalysts since the presence of larger
amounts of the additives are capable of reducing the mobility of the Mo oxide species even under
high temperatures, thus preventing their aggregation. Furthermore, this hypothesis is aligned with
the tendencies observed by the XRD patterns of the as-synthesized catalysts (i.e. before
pretreatment) (Figure 2 (a)) where presence of the additives results in a drop in the intensity of
the most intense peak corresponding to MoO crystals, indicating that the additive interacts with
3
Mo on the external surface, reducing its crystallinity and distorting it.
XPS spectra of the precarburized catalysts were also measured before and after reaction and are
presented in Figure S12 in the SI section. The Mo/(Si+Al) atomic ratios measured by XPS are
representative mostly of the external surface of the zeolite and provide an idea of the degree of
exposure of Mo on the surface. The data indicate that upon adding both the Co and Ni additive,
the Mo exposure reaches a maximum value with 0.6 wt% loading but then decreases when
increased to 1% both in the fresh and spent samples, once again in line with the activity data of the
precarburized catalysts where the highest stability is obtained with the catalysts containing 0.6
wt% of the additives
Table 4. High temperatures at which weight drop occurs for untreated fresh Mo-Co and Mo-
Ni catalysts
Catalyst
Mo
Weight drop temperature (C)
6
750
2
1

6
6
Mo-0.2Co
Mo-0.6Co
800
815
820
830
810
855
6
Mo-1Co
Mo-0.2Ni
Mo-0.6Ni
6
6
6
Mo-1Ni
TGA measurements were also performed on the spent precarburized single-metal additive
catalysts, Co/ZSM-5 and Ni/ZSM-5. The amount of carbon deposited in reaction calculated from
the TGA profiles (Figure S11) is shown in Table 5. The sum of the amount of carbon deposits
formed with 6Mo and each of the Co and Ni catalysts is higher than the amount of carbon deposits
formed in the Mo-Co and Mo-Ni catalysts (with the exception of 1 Co), showing that in the
presence of Mo, the cracking capacity of the additive metals, Co and Ni, is reduced, further
suggesting that a Mo-Co and Mo-Ni interaction affects the activity of both the Mo and the additive
species when they coexist on ZSM-5. Testing the blank ZSM-5 in reaction produces a small
amount of carbon deposits (first row in Table 5), discarding a significant contribution of the zeolite
support to the formation of carbon deposits in the single-metal additive samples.
Table 5. Amount of coke burnt off from the spent precarburized single metal Co/ZSM-5 and
Ni/ZSM-5 catalysts. The numbers in brackets correspond to the bimetallic catalysts from
Table 4. (The spent catalysts underwent reaction at 700°C under 91% CH
velocity of 1500ml/g·hr for 10 hrs.)
4
/N with a space
2
Amount of C formed in reaction in
spent precarburized catalysts (mg/g
catalyst)
Sum of C formed in reaction with 6Mo
Catalyst
and pure Ni/ZSM-5 or Co/ZSM-5 (C
formed in bimetallic counterpart)
ZSM-5
1.6
83.3
2.7
-
6
Mo
-
0
.2Co
86.0 (83.6)
2
2

0
.6Co
3.1
21.4
13.5
39.5
107.2
86.4 (73.8)
104.7 (120.3)
96.8 (79.0)
1
Co
0
0
.2Ni
.6Ni
122.8 (94.2)
190.5 (121.2)
1
Ni
3
.2.6. Scanning Electron Microscopy (SEM)
Figures S13 and S14 represent the SEM images of the spent He-pretreated catalysts and spent
precarburized catalysts respectively: (a) 6Mo-0.2Co (b) 6Mo-0.6Co (c) 6Mo-1Co (d) 6Mo-0.2Ni
(e) 6Mo-0.6Ni (f) 6Mo-1Ni. No major morphological change was observed between the two types
of pretreated catalysts, but extensive carbon nanotube formations were observed in both spent
precarburized and He-pretreated catalysts with the higher, 1wt% loading of additive, 6Mo-
1
Co/ZSM-5 and 6Mo-1Ni/ZSM-5. This is in agreement with reports that Co and Ni are responsible
for the catalytic thermal decomposition of CH to nanostructured carbon materials [68-70].
4
However, in the case of the precarburized catalysts, the amount of carbon nanotubes formed in
reaction is clearly lower at lower additive loadings, therefore it can be claimed that for all three
additives, when their loadings are low, precarburization inhibits the formation of carbon
nanotubes. Small amount of carbon nanotubes was observed in spent oxide catalysts, 6Mo-
0
.2Co/ZSM-5 and 6Mo-0.2Ni/ZSM-5. On the other hand, in the spent precarburized catalysts,
carbon nanotubes were observed only in 6Mo-0.2Ni/ZSM-5, but not in 6Mo-0.2Co/ZSM-5
catalysts.
3
.3.Regeneration cycles on spent catalysts
The interaction between Mo and the additives, Co or Ni, in the precarburized catalysts can have
an additional practical benefit over time: it can facilitate the regeneration of the catalysts without
affecting their stability over several regeneration cycles. To test this, two regeneration cycles under
hydrogen flow were applied after being exposed to 10 hours of reaction and the evolution of the
benzene yield over TOS over the cycles is illustrated in Figure 6. The catalysts picked for this
2
3

study were the Co- and Ni-modified catalysts with optimum additive loadings, 6Mo-0.6Co and
Mo-0.2Ni, and each were compared to the unmodified 6Mo catalyst. We observe that both the
6
precarburized 6Mo-0.6Co/ZSM-5 and 6Mo-0.2Ni/ZSM-5 catalysts not only present a higher
benzene yield compared to unmodified 6Mo, but also higher stability over each regeneration cycle:
the 6Mo catalyst recovers part of the benzene yield after each regeneration, however over TOS, its
yield slightly drops; but in the case of both of the modified catalysts, 6Mo-0.6Co and 6Mo-0.2Ni,
not only is the benezene yield obtained after the regeneration slightly higher, but also remains
stable and even increases slightly over time. While not in the scope of the work presented here,
further structural characterization before and after each cycle would be necessary to further relate
the differences in behavior of these catalysts with the regeneration cycles.
4
. Conclusions
The effect of adding Co and Ni to Mo/ZSM-5 catalysts varies depending on the type of
pretreatment received. When Mo carbides are formed in situ, during the reaction induction period
(after He pretreatment) the presence of additives is detrimental to their catalytic activity and
benzene yield. However, when the catalysts are precarburized, a synergy is established in the
presence of an optimum loading of both Co and Ni rendering catalysts that produce higher and
more stable benzene yields. This behavior is due to the combination of various effects: 1)
precarburization diminishes formation of inactive Al
2
(MoO
4
3
) species as well as preventing
aggregation of Mo species during the reaction induction period; 2) the presence of additives further
decreases the volatility and mobility of the Mo species, possibly enhancing the retention of the Mo
species inside the zeolite channels during reaction and reducing catalyst deactivation; 3) when Mo
and the additive species coexist, no crystalline mixed phases are observed, however their
interaction modifies the activity of the individual metals, benefitting the selectivity towards
benzene. Both Co- and Ni-modified catalysts present higher catalytic stability after several
regeneration cycles compared to the unmodified Mo/ZSM-5 catalyst.
Acknowledgements
This research was supported by the U.S. Department of Energy, Office of Science, Office of Basic
Energy Sciences, under Award DE-SC0019074. AIM thanks the Ramon y Cajal Program RyC-
2
015-17870 (Spanish Ministry of Economy and Competitiveness, Spain) and RTI2018-099668-
B-C22 Project (Spanish Ministry of Science, Innovation and Universities) for XPS measurements.
2
4

The authors would like to thank Dr. Daniel Unruh from the Department of Chemistry and
Biochemistry at Texas Tech University for his help in acquiring the XRD patterns.
References
[
1] M. Ruitenbeek, B.M. Weckhuysen, Angew. Chem. Int. Ed. 53 (2014) 11137-11139.
[
2] M. A. Newton, A. J. Knorpp, A. B. Pinar, V L. Sushkevich, D. Palagin, J. A. van Bokhoven,
J. A. C. S. 140 (2018) 10090-10093.
[
[
[
[
[
[
[
3] T.Z.H. Gani, H.J. Kulik, ACS Catal. 8 (2018) 975-986.
4] S. Ma, X. Guo, L. Zhao, S. Scott, X. Bao, J. Energy Chem. 22 (2013) 1-20.
5] L. Wang, L. Tao, M. Xie, G. Xu, J. Huang, Y. Xu, Catal. Lett. 21 (1993) 35-41.
6] D. Wang, J.H. Lunsford, M.P. Rosynek, J. Catal. 169 (1997) 347-358.
7] L. Chen, L. Lin, Z.S. Xu, X.S. Li, T. Zhang, J. Catal. 157 (1995) 190-200.
8] F. Solymosi, A. Cserenyi, A. Szoke, T. Bansagi, A. Oszko, J. Catal. 165 (1997) 150-161.
9] W. Liu, Y. Xu, S.T. Wong, L. Wang, J. Qui, N. Yang, J. Molec.Catal. A: Chem. 120 (1997)
2
57-265.
[
[
[
10] B. M. Weckhuysen, D. Wang, M.P. Rosynek, J.H. Lunsford, J. Catal. 175 (1998) 338-346.
11] Y. Xu, S. Liu, X. Guo, L. Wang, M. Xie, Catal. Lett. 30 (1994) 135-149.
12] J. Gao, Y. Zheng, J.M. Jehng, Y. Tang, I.E. Wachs, S.G. Podkolzin, Science 348 (2015) 686-
6
90.
[
13] I. Lezcano-Gonzalez, R. Oord, M. Rovezzi, P. Glatzel, S.W. Botchway, B.M. Weckhuysen,
A. M. Beale, Angew. Chem.Int. Ed. 55 (2016) 5215-5219.
[
14] N. Kosinov, F. J. A. G. Coumans, G. Li, E. Uslamin, B. Mezari, A. S. G. Wijpkema, E.A.
Pidko, E. J. M. Hensen, J.Catal. 346 (2017) 125-133.
[
(
15] P. Zhu, G. Yang, J. Sun, R. Fan, P. Zhang, Y. Yoneyama, N. Tsubaki, J. Mat. Chem. A 5
2017) 8599-8607.
[
16] A. Martinez and E. Peris, E., Appl. Catal. A: Gen. 515 (2016) 32-44.
[
17] N. Kosinov, F. J. A. G. Coumans, G. Li, E. Uslamin, A. S. G. Wijpkema, B. Mezari, E. J. M.
Hensen, ACS Catal. 7 (2017) 520-529.
2
5

[
18] I. Vollmer, B. van der Linden, S. Ould-Chikh, A. Aguilar-Tapia, I. Yarulina, E. Abou-Hamad,
Y.G. Sneider, A.I.O. Suarez, J.-L. Hazemann, F. Kapteijn, J. Gascon, Chem. Sci., 9 (2018) 4801-
807.
4
[
19] N. Kosinov, A.S.G. Wijpkema, E. Uslamin, R. Rohling, F.J.A.G. Coumans, B. Mezari, A.
Parastaev, A.S. Poryvaev, M.V. Fedin, E.A. Pidko, E.J.M. Hensen, Angew. Chem. Int. Ed. 57
(
[
[
[
[
2018) 1016-1020.
20] J. J.Spivey, G. Hutchings, Chem. Soc. Rev. 43 (2014) 792-803.
21] Z.R. Ismagilov, E,V. Matus and L.T. Tsikoza, Energ. Environ. Sci. 1 (2008) 526-541.
22] F. Solymosi, A. Cserenyi, A. Szoke, T. Bansagi, A. Oszko, J.Catal. 165 (1997) 150-161.
23] W. Liu, Y. Xu, S.T. Wong, L. Wang, J. Qui, N. Yang, J. Molec. Catal. A: Chem. 120 (1997)
2
57-265.
[
(
24] R. W. Borry III, Y. H. Kim, A. Huffsmith, J. A. Reimer, E. Iglesia, J. Phys. Chem. B 193
1999) 5787-5796.
[
[
[
25] Y.H. Kim, R.W. Borry III, E. Iglesia, Micropor. Mesopor. Mat. 35-36 (2000) 495-509.
26] W. Ding, S. Li, G.D. Meitzner, E. Iglesia, J. Phys. Chem. B 105 (2001) 506-513.
27] M. Agote-Arán, A. B. Kroner, H.U. Islam, W.A. Sławiński, D.S. Wragg, I. Lezcano-
González, A.M. Beale, ChemChatChem, 11 (2019) 473-480.
[
[
[
[
28] S. Li, C. Zhang, Q. Kan, D. Wang, T. Wu, L. Lin, Appl. Catal. A: Gen. 187 (1999) 199-206.
29] S.T. Qi, B.L. Yang, Catal.Today, 98 (2004) 639-645.
30] Y.P. Zhang, D. J. Wang, J. F. Fei, X.M. Zheng, Aust. J. Chem. 55 (2002) 531-534.
31] Y.P. Zhang, D.J. Wang, J.H. Fei, X.M. Zheng, React. Kinet.Catal. Lett. 74 (2001) 151-161.
[
(
32] A.K. Aboul-Gheit, A.E. Awadallah, S.M. El-Kossy, A.L.H. Mahmoud, J.Nat. Gas Chem. 17
2008) 337-343.
[
(
33] S. Burns, J.S.J. Hargreaves, P. Pal, K.M. Parida, S. Parija, J. Molec. Catal. A: Chem. 245
2006) 141-146.
[
[
34] B.S. Lu, L. Jiang, H. Sun, C.T. Au, Appl. Surf. Sci. 253 (2007) 5092-5100.
35] B.S. Liu, Y. Yang, A. Sayari, Appl. Catal. A: Gen. 214 (2001) 95-102.
2
6

[
(
36] M.W. Ngobeni, A.F. Carley, M.S. Scurrell, C.R. Nocolaides, J. Molec. Catal. A: Chem. 305
2009) 40-46.
[
[
[
[
[
[
37] A. Aboul-Gheit, M. El-Masry, A. Awadallah, Fuel Process. Technol. 102 (2012) 24-29.
38] S. Burns, J.S.J. Hargreaves, P. Pal, K.M. Parida, S. Parija, Catal. Today 114 (2006) 383-387.
39] S. Liu, Q. Dong, R. Ohnishi, M. Ichikawa, Chem. Comm. 15 (1997) 1455-1456.
40] S. Tang, H. Chen, J. Lin, K.L. Tan, Catal. Comm. 2 (2001) 31-35.
41] V. Fila, M. Bernauer, B. Bernauer, Z. Sobalik, Catal. Today 256 (2015) 269-275.
42] V. Vosmerikov, V. Zaikovskii., L. Korobitsyna, G. Echevskii, V. Kozlov, Y. Barbashin, S.
Zhuravkov, Kinet.Catal. 50 (2009) 725.
[
43] M. Rahman, A. Sridhar, S. J. Khatib. Appl. Catal. A: Gen. 558 (2018) 67-80.
[
(
44] M. Rahman, A. Infantes-Molina, A. Boubnov, S. Bare, A. Sridhar, S.J. Khatib, J. Catal.
2019) (accepted).
[
45] J. P. Tessonnier, B. Louis, S. Walspurger, J. Sommer, M.J. Ledoux, C. Pham-Huu, J. Phys.
Chem. B, 110 (2006) 10390-10395.
[
(
46] J.P. Tessonnier, B. Louis, S. Rigolet, M.J. Ledoux, C. Pham-Huu, Appl. Catal. A: Gen., 336
2008) 79-88.
[
[
[
[
[
47] W. Zhang, D. Ma, X. Han, X. Liu, X. Bao, X. Guo, X. Wang, J. Catal. 188 (1999) 393-402.
48] A. Sridhar, M. Rahman, S. J. Khatib, ChemCatChem, 10 (2018) 2571-2583.
49] H. Liu, T. Li, B. Tian, Y. Xu, Appl. Catal. A: Gen. 213(1) (2001) 103-112.
50] P.M. Bijani, M. Sohrabi, S. Sahebdelfar, Chem. Eng. Tech. 35 (2012) 1825-1832.
51] B. Li, S. Li, N. Li, H. Chen, W. Zhang, X. Bao, B. Lin, Micropor. Mesopor. Mat. 88 (2006)
2
44-253.
[
[
[
[
52] A. Martínez, E. Peris, M. Derewinski, A. Burkat-Dulak, Catal. Today 169 (1) (2011) 75-84.
53] P. Arnoldy, J. De Jonge, J. Moulijn, J. Phys. Chem. 89 (1985) 4517-4526.
54] J. R. Regalbuto, J.-W. Ha, Catal. Lett. 29 (1994) 189-207.
55] S. Rajagopal, H.J. Marini, J. A. Marzari, R. Miranda, J. Catal. 147 (1994) 417-428.
2
7

[
56] R.W. Borry III, E. Iglesia 1997; Available from:
http://www.osti.gov/scitech/servlets/purl/620594.
[
57] T. E. Tshabalala, N. J. Coville, M. S. Scurrell, Appl. Catal. A: Gen. 485 (2014) 238-244.
[
(
58] K. Sun, W. Ging, K. Gasem, H. Adidharma, M. Fan, R. Chen, Ind. Eng. Chem. Res. 56 (40)
2017) 11398-11412.
[
59] L. Hongmei, X. Yide, Chin. J. Catal. 27 (4) (2006) 319-323.
[
60] D. Ballesteros-Plata, A. Infantes-Molina, M. Rodriguez-Cuadrado, E. Rodriguez-Aguado, P.
Braos-García, E. Rodriguez-Castellon, Appl. Catal. A: Gen. 547 (2017) 86-95.
[
[
61] J. Zieliński, Catal. Lett. 12 (1992) 389-393.
62] J. Pérez-Ramírez, J.C. Groen, A. Brückner, M.S. Kumar, U. Bentrup, M.N. Debbagh, L.A.
Villaecscusa, J. Catal. 232 (2005) 318-334.
[
5
63] P.B. Bodart, J.B. Nagy, G. Debras, Z. Gabelica, P.A. Jacobs, J. Phys. Chem. 90 (1986) 5183-
190.
[
1
64] A. Omegna, M. Haouas, A. Kogelbauer, R. Prins, Micropor. Mesopor. Mater. 46 (2001) 177-
84.
[
65] S. Jiang, H. Zhang, Y. Yan, X. Zhang, RSC Adv. 5 (2015) 41269-41277.
[
[
66] H. Liu, T. Li, B. Tian, Y. Xu, Appl. Catal. A: Gen. 213 (2001) 103-112.
67] B. Liu, L. France, C. Wu, Z. Jiang, V.L. Kuznetsov, H.A. Al-Megren, P.P. Edwards, Chem.
Sci. 6 (2015) 5152-5163.
[
(
68] T. Reshetenko, L. Avdeeva, Z. Ismagilov, A. Chuvilin, V. Ushakov, Appl. Catal. A: Gen. 247
2003) 51-63.
[
[
69] S. Takenaka, Y. Shigeta, E. Tanabe, K. Otsuka, J. Catal. 220 (2003) 468-477.
70] Y. Li, X. Zhang, X. Tao, J. Xu, W. Huang, J. Luo, Z. Luo, T. Li, F. Liu, Y. Bao, H. Geise,
Carbon 43 (2005) 295-301.
2
8

Figure Captions
Figure 1. Benzene yield versus time on stream (TOS) for (a) Co-modified and (b) Ni-modified
6
Mo/ZSM-5 catalysts. The shaded area corresponds to the induction period of the reaction. The
blue symbols correspond to He-pretreated catalysts and red symbols correspond to precarburized
catalysts. (The data points for unmodified 6Mo/ZSM-5, and single Co/ZSM-5 and Ni/ZSM-5 are
included in both graphs to serve as reference for comparative purposes.)
Figure 2. XRD patterns for (a) as-synthesized, (b) He-pretreated, (c) precarburized, (d) spent
He-pretreated and (e) spent precarburized 6Mo/ZSM-5 and Co- and Ni-modified 6Mo/ZSM-5
catalysts. (The XRD pattern for HZSM-5 is included in each figure as reference.)
Figure 3. TPR profiles of fresh (a) Co-modified and (b) Ni-modified 6Mo/ZSM-5 catalysts. (The
profile for unmodified 6Mo/ZSM-5 is included in each graph as reference.)
Figure 4. MAS-ssNMR spectra of as-synthesized (black), precarburized (blue), and He-pretreated
6
Mo/ZSM-5 catalyst.
Figure 5. TGA of all fresh as-synthesized catalysts before receiving any type pretreatment.
Figure 6. Regeneration of (a) 6Mo-0.6Co (b) 6Mo-0.6Ni.
2
9