Methane conversion to ethylene over GaN catalysts. Effect of catalyst nitridation

Article abstract of DOI:10.1016/j.apcata.2020.117430

Vast availability of natural and shale gases makes methane a reliable source for synthesizing valuable chemical building blocks such as ethylene. A new stable supported GaN/SBA15 catalyst from an emerging class of nitride catalysts was reported for the direct non-oxidative methane coupling to ethylene. The effect of nitridation on the catalyst properties and activity was investigated. The optimum nitridation temperatures were 700 °C and 750 °C for the GaN/SBA15 and the unsupported GaN catalyst, respectively. Supported catalysts were more stable and had 5–10 times higher product (ethylene) formation rates per gram of gallium than the unsupported catalysts due to the higher surface area (>320 vs. <20 m² g^?1) and Ga-dispersion inside the pores. Compared to the oxide precursors, the nitrides exhibited a higher atom conversion efficiency for the CH₄ carbon leading to higher ethylene selectivity (71 % for GaN/SBA15, <58 % for Ga₂O₃/SBA15) and lower coke selectivity (27 % for GaN/SBA15, 40 % for Ga₂O₃/SBA15).

Full text of DOI:10.1016/j.apcata.2020.117430

Journal Pre-proof
Methane conversion to ethylene over GaN catalysts. Effect of catalyst
nitridation
Kanchan Dutta, Vishnu Chaudhari, Chao-Jun Li, Jan Kopyscinski
PII:
S0926-860X(20)30023-5
DOI:
https://doi.org/10.1016/j.apcata.2020.117430
APCATA 117430
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
18 November 2019
11 January 2020
21 January 2020
Please cite this article as: { doi: https://doi.org/
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.
©
2020 Published by Elsevier.

Revised manuscript No.: APCATA-D-19-02293R1
Methane conversion to ethylene over GaN catalysts. Effect of
catalyst nitridation.
1
1
2
1
Kanchan Dutta , Vishnu Chaudhari , Chao-Jun Li , Jan Kopyscinski *
1
Department of Chemical Engineering, McGill University,
3
610 University Street, Montreal, Quebec H3A 0C5, Canada
2
Department of Chemistry, McGill University
8
01 Sherbrooke Street West, Montreal, QC H3A 0B8
*
To whom all correspondence should be addressed:
Tel.: +1 514 398 4276
jan.kopyscinski@mcgill.ca
Graphical Abstract

Highlights





Optimum nitridation temperature for the supported GaN/SBA15 catalyst is 700 °C.
Nitridation improved atomic economy of CH conversion.
GaN/SBA15 is more stable and has a lower coke formation than unsupported GaN.
4
C
2
H
4
selectivity over GaN/SBA15 was 70% and coke selectivity was <30%.
Ga
2
O
3
produced CO
x
and H
2
O reducing C
2
H
4
selectivity below 60% over Ga
2
3
O /SBA15.
Abstract
2

Vast availability of natural and shale gases makes methane a reliable source for synthesizing
valuable chemical building blocks such as ethylene. A new stable supported GaN/SBA15 catalyst
from an emerging class of nitride catalysts was reported for the direct non-oxidative methane
coupling to ethylene. The effect of nitridation on the catalyst properties and activity was
investigated. The optimum nitridation temperatures were 700 °C and 750 °C for the GaN/SBA15
and the unsupported GaN catalyst, respectively. Supported catalysts were more stable and had 5-
1
0 times higher product (ethylene) formation rates per gram of gallium than the unsupported
2
-1
catalysts due to the higher surface area (>320 vs. <20 m g ) and Ga-dispersion inside the pores.
Compared to the oxide precursors, the nitrides exhibited a higher atom conversion efficiency for
the CH carbon leading to higher ethylene selectivity (71% for GaN/SBA15, <58% for
4
Ga
2
O
3
/SBA15) and lower coke selectivity (27% for GaN/SBA15, 40% for Ga
2
3
O /SBA15).
Keywords: ethylene; direct nonoxidative methane conversion; gallium nitride catalyst;
nitridation; SBA-15; catalyst synthesis and characterization.
1
Introduction
The base chemicals for the petrochemical industry like olefins (ethylene, propylene, butadiene)
and aromatics (benzene, toluene, and xylenes) are currently derived from oil and natural gas.
Estimated to be the world's most widely used petrochemical in terms of production volume,
3

ethylene is the key building block for polyethylene (50%), ethylene oxide (10%) and its
derivatives such as ethylene glycol [1]. It is also used to produce vinyl acetate, polyvinyl chloride,
polyester fiber and film, and a range of alcohols and solvents. An estimated 60% of total US
ethylene production capacity uses liquefied natural gas, with a further 38% derived using naphtha.
In 2018, ethylene accounted for 59% of the petrochemical industry revenue in Canada [2], while
aromatic hydrocarbons accounted for 38% of revenues in the United States (2017) [1].
Methane (natural gas, shale gas) is considered to be a reliable source for the petrochemical
industry. Methane activation and conversion to value-added chemicals are of great interest not
only from the fundamental research perspective but also in terms of industrial application as it
offers a potential for an alternative route to produce chemicals. Currently, natural gas is converted
via the syngas route (indirect and oxidative) to methanol and its derivatives. This conventional
syngas route has low efficiency, high capital cost, and associated greenhouse gas emissions. Direct
oxidative coupling of methane is another way [3]. However, a significant amount of CO, CO2, and
2
H O are produced, which lowers the product selectivity.
Direct non-oxidative conversion of methane to olefins and aromatics is an alternative and
theoretically a very efficient conversion route [4]. However, methane has a very stable C-H bond,
which is very difficult to activate under non-oxidative conditions [5]. The review articles from Ma
et al.[4], Spivey and Hutchings [6], and Karakaya and Kee [3] summarize the research work that
4

has been conducted with emphasis on metal-modified zeolites. Metals such as Mo, Zn, W, Re,
Cu, Mn, Ni, and Cr have been investigated. Molybdenum-containing zeolites exhibited the highest
activity in terms of methane conversion (3-16%) with benzene (50-75%) as the main product [5].
It is hypothesized that an active species on the catalyst surface (i.e., CH
dimerizes to produce ethylene. Finally, ethylene is oligomerized and cyclized on the zeolite to
form aromatics [4]. A combination of benzene and the active species like CH and CH can form
x
) is formed, which further
2
3
toluene. Two phenyl species (active aryl rings) can form naphthalene [7,8]. Further combination
of phenyl species can lead to the formation of polynuclear aromatic compounds (i.e., coke) that
deactivates the catalysts, which is the main drawback of this reaction. Adding promoters such as
transition metals in Period 4 (Fe, Co, Ni, and Cr) and metals in Group 13 (Ga and Al) show the
best improvement in terms of methane conversion, product selectivity, and catalyst stability [4].
It has been shown that Ga containing zeolites are highly active for the dehydrogenation of light
alkane [9,10]. But the nature of the active gallium species as well as the gallium alkane
intermediates is still under debate. Recently, Dumesic’s group developed PtSn-zeolite catalysts
that achieved high ethylene selectivities (70-90% at 700 °C, coke not included); yet coking was
still a huge problem [11]. They obtained methane conversions (in terms of fraction converted to
the products) less than 0.5%, typically 0.25% at 700 °C. Xiao and Varma [12] used bimetallic
PtBi-zeolite catalysts and reported ethane (C
2
6
H ) selectivity of 90% (coke not included) at 700 °C,
with methane conversion between 1-5%. However, they had 40-50% coke selectivity with up to
5

1
0% ethylene (C
2
H
4
) selectivity on 1% Pt-0.2% Bi catalyst at 650 °C and 0.1 atm CH partial
4
pressure. They suggested that carbon deposition occurs during the initial activation period. Bajec
et al. [13] used Fe/HZSM‐ 5, Mo/HZSM‐ 5, and Fe–Mo/HZSM‐ 5 catalysts for methane
activation and coupling to ethane and ethylene. They obtained 1-6% methane conversion, up to
5
0% ethylene selectivity and 11-35 wt% coke. Their reaction conditions were T = 700 °C, WHSV
−
1
=
2 h , ptot = 1.5 bar. Sheng et al. [14] used a less acidic boron-based [B]ZSM-5 catalyst for
methane activation and obtained 90% ethylene selectivity (gas phase) with less than 1% methane
conversion at 700 °C with an average rate of 0.3 mmolCH4 h gcat-1. Low acidity of the support
produced more ethylene than benzene and lower coke. Guo et al. [15] synthesized single iron sites
embedded in a silica matrix. They reported minimum carbon deposition, maximum ethylene
selectivity of 48%, and maximum methane conversion of 48% at 1090 °C.
-1
Almost all the catalysts studied so far were bifunctional precious metals and metal oxides. Our
manuscript focuses on metal nitrides, which are gaining interests as heterogeneous catalysts [16].
Nitride catalysts have been used for ammonia synthesis (Co, Mo, Ru nitrides), ammonia
decomposition to obtain CO
2
free H (Ru nitride), and for hydrotreating process (Mo nitride) [16].
2
Nitrides of Si, B have also been used as catalysts support. Compared to the conventional alumina
and silica supports, the nitrides can have greater thermal conductivity, increased inertness,
modified basicity, and enhanced hydrophobicity [16]. Gallium nitride, a material used in the
semiconductor industry, was tested for the first time for methane activation in a batch reactor [17].
6

The tests were carried out under Ultraviolet (UV) irradiation at room temperature, and also under
thermocatalytic conditions at 450 °C and 5-8 h residence time [18,19]. They demonstrated a high
selectivity for benzene (SC6H6 = 97% under UV and SC6H6 = 89.8%, at 450 °C, respectively).
However, due to low temperature and thermodynamic limitations, the methane conversion was
less than 0.5%. Higher temperatures (>650 °C) and continuous operations are necessary for
achieving larger methane conversion [20]. In the optoelectronics and semiconductor industry,
GaN is often synthesized in the form of nanostructures (nanotubes, nanowires, and nanorods)
through various procedures such as arc discharge, laser ablation, chemical vapor deposition,
plasma-assisted molecular beam epitaxy, and metal-organic vapor phase epitaxy. In most of these
procedures, NH at high temperature (nitridation) is used to produce GaN nanostructures from
3
gallium precursors. Some of these methods might not be applicable for the preparation of the
catalysts that require a large surface area and well-distributed active sites.
In this work, unsupported catalysts were synthesized, and for the first time, supported gallium
nitride catalysts were synthesized, and the effect of the nitridation conditions on the direct non-
oxidative methane conversion to value-added chemicals were investigated under flow conditions.
7

2
Experimental Section
2
.1 Catalyst preparation
Two types of catalysts were synthesized; one group was unsupported GaN, the second one was
GaN supported on SBA-15 with Ga target loading of 16 wt%. Unsupported catalysts were
prepared by the Evaporation Induced Self Assembly (EISA) technique adapted from Chaudhari
et al. [21]. Around 1 g of triblock copolymer (Pluronic P-123, Sigma Aldrich) was dissolved in
1
6 mL of pure anhydrous ethanol (Greenfield Global Inc.) until a homogeneous solution was
obtained. This was followed by the addition of 1.7 mL of nitric acid (67-70 wt%, Fisher Scientific)
and stirring for complete mixing with the acid. Another solution was prepared in 5 mL DI-water
and 5 mL of ethanol with 4.5 g of dry gallium (III) nitrate hydrate powder (99.9998 wt% trace
metal basis, ACROS Organic, Thermo Fisher Scientific). DI-water was used to get a homogeneous
solution as the nitrate is only slightly soluble in ethanol. The hygroscopic nitrate was always stored
under dry inert gas (Ar with < 5 ppm moisture, Praxair) desiccator to prevent moisture absorption.
The nitrate bottle was also flushed with Ar after every use. This ensured that the weight of the
nitrate was consistent and not influenced by absorbed moisture. The copolymer solution was then
added to the gallium precursor solution and stirred for 5 h at 600 rpm at room temperature. The
homogeneous, colorless mixture was kept inside a drying oven at 60 °C for 48 h. Thereafter, the
–
1
dried copolymer-gallium mixture was calcined at 650 °C for 6 h at a heating rate of 1 °C min in
which the copolymer was burned, and the gallium-precursor was converted to Ga
2
3
O . Before
8

nitridation, the catalysts were dried for 4.5 h in Ar (99.999%, MEGS Specialty Gases, 15-20 mL
N
–1
–1
min ) at 300 °C with a heating rate of 1 °C min (subscript N denotes normal condition with T
=
0 °C and 1 bar). The nitridation was carried out in a fixed bed reactor with anhydrous ammonia
–
1
(
NH
3
, 99.99%, MEGS Specialty Gases, 20 mL min ) flowing top-down through the catalyst bed
N
(around 1 g). A quartz tube (9 mm OD × 7 mm ID × 275 mm long) with a 15-40 µ frit at one end
was used as the reactor inside a vertical tube furnace (Mellen Company). At the reactor outlet, the
gas was mixed and diluted with Ar. A slipstream was analyzed via mass spectrometry (Pfeiffer
Omnistar GSD 301 O1), while the rest was sent through a 10 L hydrochloric acid bubbler (36.5–
3
8 wt%, Fisher Scientific) to absorb unreacted ammonia. Nitridation was carried out at five
–
1
nitridation temperatures (600, 650, 700, 750 and 800 °C at a heating rate of 1.5 °C min and 24 h
NH exposure from the start of heating to cool down), and ammonia exposure time (3-9 h at 750
C at a heating rate of 1.5 °C min ). After cooling to room temperature under the NH
3
–1
°
3
atmosphere,
the reactor was purged with Ar, and the catalyst sample was stored for further use under inert gas
in a desiccator.
Supported gallium catalysts were prepared as follows. Ordered mesoporous silica (SBA-15) was
synthesized as described in Zhao et al. [22]. Around 4 g of triblock copolymer (Pluronic P-123,
Sigma Aldrich) was dissolved by stirring (600 rpm for 5 h) in 95 mL of deionized water until a
homogeneous solution was obtained. To this solution, 4 mL of hydrochloric acid was added and
stirred for about 15 min at a temperature of 38.5 °C. Then 9.5 mL of tetraethoxysilane (TEOS,
9

9
9.9%, Alfa Aesar) was added. The resulting mixture was stirred at 600 rpm for an additional 24
h at 38.5 °C for completing TEOS hydrolysis and precipitation of silica. The above mixture was
placed inside a closed polypropylene digestion DigiTUBEs (SCP science) and kept inside an oven
at 100 °C for 48 h for hydrothermal treatment. After cooling down, the white powder was
separated by filtration and washed with deionized water several times, followed by overnight
drying at 60 °C. Finally, SBA-15 was obtained by calcining the dried solid at 550 °C for 5 h at a
–1
heating rate of 1 °C min .
Supported catalysts were synthesized by incipient wetness impregnation of SBA-15 with gallium
containing precursor. Typically, 1 g of SBA-15 was impregnated with 1 g of gallium (III) nitrate
hydrate in 4.5 mL aqueous solution. The almost wet solid was left for 24 h at room temperature
followed by drying overnight at 70 °C. The dried solid was then calcined at 550 °C for 5 h at a
–
1
heating rate of 1 °C min to produce gallium oxide inside SBA-15. Nitridation of supported
catalysts were carried out in the same manner as described above for the unsupported catalysts.
Due to the small bulk density, around 200 mg supported gallium oxide (Ga
2
3
O /SBA15) was
nitridated at temperatures of 650, 700, 750, and 800 °C. Finally, the supported catalysts were
stored in Ar filled vials inside the desiccator.
The produced unsupported catalysts were named Ga
2
O
3
, GaN-750, or GaN-750-03, referring to
the calcined and nitridated samples, respectively. In detail, for GaN-750, the 750 denotes the
1
0

nitridation temperature in degree Celsius (°C) with a total NH
50-03, the 03 refers to the nitridation time in hours (e.g., 3, 6 or 9 h). The produced supported
catalysts were named Ga /SBA15 and GaN/SBA15-750, referring to the calcined and nitridated
samples, respectively. In detail, for GaN/SBA15-750, the 750 denotes the nitridation temperature
3
exposure of 24 h, while for GaN-
7
2
O
3
in degree Celsius (°C). N-SBA15-800 refers to a nitridated SBA-15 (without Ga
2
3
O ) with a
nitridation temperature of 800 °C.
2
.2 Catalyst characterization
The gallium content in the supported catalysts was determined using Inductively Coupled Plasma
Optical Emission Spectroscopy (ICP-OES, iCAP 6500 dual view Thermo Scientific). The
-
supported catalyst was fused with twice the amount (by weight) of lithium tetraborate (SCP
science) in a 9 mL graphite crucible (SCP science). The fusion was carried out in a muffle furnace
ramping to 950 °C, where it was held for 30 min. A glassy bead was obtained after cooling down,
and the fusion flux was carried out in triplicate. Two different solutions were prepared from the
bead; one in 10% HNO
3
(trace metal grade) and another one in 3:1 HCl to HNO concentrated
3
solution (aqua regia). These solutions were digested in polypropylene DigiTUBEs (SCP science)
at 95 °C for 2 h until the bead was completely dissolved. After cooling down, the solutions were
filtered through 45 µ filter and diluted appropriately for ICP analysis. Unsupported catalysts were
either GaN or Ga
2
3
O or a mix of both, and no other metals were present in them. Gallium content
in unsupported catalysts can be calculated directly without ICP.
1
1

The total nitrogen content of unsupported gallium nitride catalysts was determined by Shimadzu
TOC-Vcph with TNM–1 Total Nitrogen Module. Prior to the analysis, around 25-50 mg of the
sample was dissolved in 6M hydrochloric acid (HCl) overnight at 80 °C. The gallium nitride was
converted to gallium chloride and ammonium chloride, while gallium oxide was converted to
gallium chloride and water, as shown in Eqs. (1) and (2), respectively.
GaN (s) + 4 HCl (aq.)  GaCl
3
(aq.) + NH
4
Cl (aq.)
(1)
(2)
Ga
2
O
3
(s) + 6 HCl (aq.)  2 GaCl
3
(aq.) + 3 H O
2
The homogenous and clear solution was cooled down, diluted with deionized water to adjust the
pH between 2–3. The diluted solution was analyzed for the total nitrogen content. The instrument
was always calibrated before the analysis using ammonium standard (Sigma Aldrich) containing
–
1
+
1
3
000 mg L of N as NH in water. The pH of the calibration solutions was adjusted between 2–
4
by using HCl. Ga
2
O
3
conversion (X
) was calculated based on equation 3 below.
Ga O
3
2
0
푛
−푛
ꢀꢁ2ꢂ3
ꢀ
ꢁ ꢂ
2
3
푋
=
3
(3)
퐺푎 푂
0
2
푛
ꢀ
ꢁ ꢂ
2
3
(
1 − 푛푁 ∙ 푀ꢀꢁ푁
)
푛푁
(1 − 푛푁 ∙ 푀ꢀꢁ푁)
Where, ꢃꢄ
=
3
+ ꢅ and ꢃ퐺푎2푂3
=
, are mol of Ga
O
3
before and
2
퐺푎 푂
2
푀ꢀꢁ
ꢂ
푀
ꢀꢁ ꢂ
2
3
2 3
after nitridation (unconverted), respectively. ꢃ denotes the mol of nitrogen in 1 g of unsupported
ꢆ
catalyst determined by total nitrogen analysis, ꢇ퐺푎ꢆ and ꢇ퐺푎2푂3 denote the molar mass of GaN
1
2

ꢆ
-1
-1
(
83.73 g mol ) and Ga
2
O
3
(187.44 g mol ), respectively. ꢃ = , with ꢈ as the total nitrogen
ꢆ
푀푁
-1
-1
content in g
N
g
cat and ꢇ as the molecular weight of nitrogen (14.01 g mol ).
ꢆ
This method, however, could not be applied for the supported catalysts because HCl could not
leach out gallium salt from the SBA-15 support. Using fusion flux, the silica structure was
completely disintegrated, and gallium was completely extracted, but the nitrogen was lost as
ammonia gas.
X-Ray Diffraction (XRD) was used to identify the crystallinity of the catalyst. Analyses were
conducted on a Bruker D8 Discovery X-Ray Diffractometer with a two-dimensional VANTEC-
5
2
00 detector and CuK
α
(λ = 1.5406 Å) radiation source. The tube voltage was 40 kV, tube current
–1
0 mA, and scan rate 5° min .
Nitrogen adsorption/desorption measurements (–196°C) were conducted using Micromeritics
Tristar 3000 BET analyzer to determine total surface area, pore size distribution, and pore volume.
Before the analysis, the samples were degassed under vacuum for 12 h at 250 °C. Surface areas
of the catalyst were calculated following the BET method.
High-resolution transmission electron microscopy (HRTEM) of fresh and spent catalysts was
carried out on FEI Tecnai G2 F20 with 200 kV. Prior to the analysis, a catalyst suspension in
ethanol was prepared by sonicating 1–2 mg of the catalyst in 2 mL anhydrous ethanol. Energy-
1
3

dispersive X-ray spectroscopy (EDS) was also carried out for detecting the elements present in
the catalyst (qualitative).
Scanning Electron Microscopy (SEM) analysis for the unsupported catalyst (Ga
2
3
O ) was
performed using a FEI Inspect F-50 field emission scanning electron microscope. In addition,
Energy-Dispersive X-ray Spectroscopy (EDS) experiments were performed for elemental analysis
(Ga, N, O, and C).
Temperature programmed oxidation coupled with mass spectrometry (TPO-MS) was carried out
for the fresh and spent catalysts to estimate the carbon either deposited during methane activation
or catalyst synthesis. Around 50 mg of supported catalyst or 250 mg of unsupported catalyst was
placed in a quartz tube (9 mm OD × 7 mm ID × 275 mm long and 15-40 µ frit) and heated in the
–1
presence of air (Ultra Zero Air, MEGS Specialty Gases, 22 mL
N
min ) to 950 °C with a heating
–1
rate of 5 °C min . The product gas at the reactor outlet was electrically heated to 200 °C, mixed
–
1
with Ar (MEGS Specialty Gases, 10 mL
N
min ), and then subsequently analyzed by a mass
spectrometer (Pfeiffer Omnistar GSD 301) calibrated for CO
2
(m/z = 44). The gas flow rates were
controlled by respective calibrated Vögtlin red-y smart controllers (± 0.3 % accuracy,
Switzerland).
In addition, temperature-programmed oxidation with thermogravimetric analysis (TPO-TGA) was
carried out for both fresh and spent catalysts (Q500 TA Instrument). 20–25 mg of catalyst was
1
4

first purged for 30 min with N
2
at room temperature to remove air from the system and then heated
–
1
at the rate of 5 °C min to 120 °C to remove the moisture. After a holding for 30 min at 120 °C,
–
1
the sample was then heated to 950 °C at 5 °C min in the presence of air and held at 950 °C for
3
0 min, while the weight change was recorded.
Catalyst synthesis (both unsupported and supported) and the subsequent analyses (BET, XRD,
ICP) were repeated for confirming repeatability and reproducibility.
2
.3 Catalytic activity measurements
The catalyst performance was measured in a vertical packed bed reactor at 700 °C and 1 barabs. A
quartz tube (9 mm OD × 7 mm ID × 335 mm long) with a 15-40 µ frit at one end was used as the
reactor inside a vertical tube furnace (Mellen Company). Around 500 mg of unsupported or 100
mg of supported catalysts were loaded in order to achieve the same bed height of 11  1 mm and
the same residence times of 1.3 s (superficial gas velocity at 700 °C) for both catalysts. The bulk
–
3
–3
densities of the GaN and GaN/SBA15 were 1300 kg m and 260 kg m , respectively. The inlet
gas contained 80 vol% CH (99.999%, Praxair) in Ar (99.999%, MEGS Specialty Gases), which
were mass flow controlled using calibrated Vögtlin red-y smart controller (± 0.3% accuracy,
Switzerland). The resulting CH -based gas hourly space velocities (GHSV) were the same for
both GaN and GaN/SBA15 with a value of 567 h . The reactor outlet and transfer lines were
electrically heated at 200 °C to avoid condensation of the organic components (i.e., C , C
4
4
–
1
6
H
6
7
8
H ,
1
5

C
10
8
H ). A quadrupole mass spectrometer (Pfeiffer Omnistar GSD 301) was used for product gas
analysis; the instrument was calibrated for methane (CH
4
), ethylene (C
2
H
4
), propylene (C
3
6
H ),
benzene (C
6
H
6
), toluene (C
7
H
8
), naphthalene (C10
H
8
), carbon dioxide (CO
2
) and hydrogen (H )
2
with argon (Ar) as the internal standard (additional details are in the supplementary information).
The calibration of the mass spectrometer was verified by a calibrated gas chromatograph (SRI GC
8
610) for CH
4
, H2, and C
6
6
H (Fig. S1).
–
1
Prior to heating, the catalyst was purged with Ar (25 mL
N
min ) for 1 h, then heated to 700 °C at
–1
a rate of 5 °C min , and subsequently, CH
4
was added. The internal reactor temperature (i.e.,
within the catalyst bed) was measured by a K-type thermocouple (Omega), which was used to
control the furnace.
Based on a detailed mass spectrum analysis and calibration factors, the CH
4
conversion, the
product molar flow rates (µmol min gGa–1) as a function of time on stream were determined (Fig.
and 7). Besides ethylene (C ), other hydrocarbons such as propylene (C ), benzene (C ),
–
1
6
2
H
4
3
H
6
6
H
6
toluene (C
7
H
8
) and naphthalene (C10
8
H ; see Fig. S16) were detected and quantified. The CH
4
conversion was calculated based on the gaseous carbon products (equation 4).
∑
휈 ∙ 푛
̇
푖 푖
푥퐶퐻4
=
× ꢌꢍꢍ
(4)
푛
̇
ꢉꢊ ,푖ꢋ
4
–
1
Where ꢃ̇ꢎ denotes the molar flow rate in µmol min , while ꢏ denotes the number of carbon atoms
ꢎ
th
in the i species in the product stream (i.e., C
2
H
4
, C
3
H
6
, CO
2
, C
6
H
6
, C
7
H8, and C10
8
H ). The
1
6

–
1
denominator represents the molar flow of methane in the feed (µmol min ). The conversion
closely represents the percentage (%) of carbon (C) converted to hydrocarbons and not coke.
Results reported in this work for the methane activation were consistent within 5% of C and H
elemental balances.
Reactivity experiments were repeated with the catalyst prepared from the same as well as from
different batches. The unsupported catalysts showed good repeatability and reproducibility, as
illustrated in Fig. S2A and B. The supported catalysts exhibited a slightly larger deviation for
different catalyst batches (Fig. S2C and D) that could be attributed to the incipient wetness
impregnation step, which is limited by slower redistribution of Ga inside the pores.
3
Results and discussion
3
.1 Nitridation of unsupported catalysts
The nitridation study for unsupported catalysts was divided into two parts, to investigate (1) the
effect of nitridation temperature, and (2) the effect of NH exposure time at a fixed temperature
750 °C). In the first part, the temperature was increased under the NH atmosphere, while in the
was added after the target temperature of 750 °C was reached (under Ar
3
(
3
second study, NH
3
atmosphere). During the nitridation, water is produced (Ga
was recorded by mean of mass spectrometry (H O, m/z = 18). The first study was a combination
of a temperature-programmed reaction with an extended soak time at five different temperatures
2
O
3
+ 2 NH
3
⇌ 2 GaN + 3 H O), which
2
2
1
7

to determine the onset and peak temperature for the Ga
2
3
O nitridation and the optimum condition.
The results clearly indicate that the nitridation started above 600 °C and that the maximum H O
2
formation rate was close to the targeted nitridation temperature for TNit = 650-800 °C (Fig. 1A).
No maximum was observed for 600 °C, indicating a very slow nitridation rate. For nitridation
temperatures of 750 and 800 °C, the water signal dropped after 8-10 h to the same level resulting
in a very similar Ga
2
3
O conversion, which was confirmed by the total nitrogen analysis (~88  5%,
Fig. 1A and B).
For GaN-650 and GaN-700, the water production rate (i.e., Ga
2
3
O conversion rate) was
significantly slower, and even after a total NH
6  5% and 70  5%, respectively (Fig. 1B). XRD analyses confirmed the trend as a distinct
transition from β-Ga to GaN with increasing nitridation temperature was observed (Fig. 1C).
3
exposure time of 24 h, the Ga
2
O
3
conversions were
4
2
O
3
The catalyst nitridated at 600 °C exhibited predominately broad XRD pattern of β-Ga
1.7°, 35.2°, 38.4°, and 64.7°; PDF file #00-006-0523 from the International Centre for Diffraction
Data). The XRD pattern for the samples treated at 650 °C started to show clear diffraction pattern
of GaN (2θ = 32.5°, 36.9°, and 57.9°; PDF file # 04-013–1733) with only minor peaks for β-Ga
At nitridation temperatures of 700 °C and higher, the catalyst samples had only diffraction patterns
corresponding to GaN, even though the Ga conversion was only between 70-90%.
2
O
3
(2θ =
3
2
3
O .
2
O
3
1
8

Based on this result, a nitridation temperature of 750 °C was sufficient as complete bulk conversion
might not be necessary for the methane conversion. The nitridation reaction is thermodynamically
unfavorable and highly endothermic as indicated by the large positive standard Gibbs free energy
and heat of reaction (∆ꢐ푅,ꢅ98퐾 = ꢌꢑꢌ.ꢒ 푘퐽 푚표푙−1 and ∆ꢓ푅,ꢅ98퐾 = ꢔꢕꢒ.ꢔ 푘퐽 푚표푙 ) [23].
However, the significant excess of ammonia, the removal of gaseous products (water), and higher
−1
temperatures (750 and 800 °C) resulted in an 88% conversion of Ga
2
3
O to GaN (Fig. 1B). For
lower temperatures (with the same gas flow rates), the Ga
2
O
3
conversion was significantly smaller
(by around 50%), indicating that more time and/or higher temperature was needed. Moreover, not
all Ga was accessible for NH due to collapsing of the pores.
3
In the second study, the Ga
nitridation temperature of 750 °C was achieved. The results are illustrated in Fig. 1D, E, and F.
The H O signal increased instantaneously at the onset of the nitridation and then declined rather
2
O
3
samples were exposed to NH for 3 h, 6 h, or 9 h after the desired
3
2
fast with time. The longer the nitridation time, the smaller the H
2
O signal, and the higher the Ga
2
O
3
conversion (i.e., 64% at 3 h vs. 85% at 9 h, see Fig. 1D and E). XRD analyses of these three
samples showed only diffraction patterns corresponding to GaN (Fig. 1F). Even with a Ga
2
O
3
conversion of only 64% at 3 h NH
3
exposure time, the peaks corresponding to oxide (β-Ga
2
O
3
)
were not visible. It has been suggested that the conversion of β-Ga
2
3
O to GaN proceed either
1
9

through the formation of amorphous gallium oxynitride (GaO
x
N
y
) intermediates or via Ga O [24].
2
The latter Ga O as gas phase intermediate is assumed at temperatures higher than 900 °C.
2
Nitrogen adsorption/desorption analysis of all unsupported Ga
type-IV isotherms with H3 and H4 type hysteresis loops with bimodal pore size distribution (Fig.
S3). The disordered structure was also observed for Ga synthesized by the EISA method [25].
Table 1 summarizes BET surface areas and pore volumes, as well as two pore sizes with the highest
2
3
O and GaN catalysts exhibited
2
O
3
2
–1
3
–1
contributions. The Ga
2
O had a surface area of 19 m g and a pore volume of 0.08 cm g with
3
a bimodal pore size distribution of 7.5 and 31.5 nm. More than 85% of the surface area
corresponded to meso- and macropores, while the rest were micropores.
2
–1
Upon nitridation at 650 °C, the surface area decreased by ~20% down to 15 m g with a pore
3
–1
volume of 0.075 cm g . This was probably due to the change from the monoclinic crystal
structure of β-Ga to the wurtzite structure of GaN. A further decrease of the surface area to 11
2
O
3
2
–1
m g at TNitridation of 800 °C could be attributed to sintering and/or pore structure collapse due to
the crystallization of GaN at temperatures above the calcination temperatures [26]. Commercial
2
–1
GaN powder had a slightly smaller surface area with SBET = 8 m g than catalyst prepared in this
work [20].
TEM and EDS analysis for Ga
2
O
3
and GaN-650 are illustrated in Fig. 2. Both gallium catalysts
are an agglomeration of nanoparticles with polycrystalline structure, as depicted with the darkfield
2
0

in the top-left insets of Fig. 2A and B. The nanoparticle size ranges from 20 to 200 nm (Fig. S4).
Polycrystalline Ga has several characteristic d-spacings representing the monoclinic structure
cell dimensions of a = 12.23 ± 0.02, b = 3.04 ± 0.01, c = 5.80 ± 0.01 Å and β = 103.7 ± 0.3°)[27].
Fig. 2 and S5 visualizes d-spacings of 3.0 ± 0.1 and 6.0 ± 0.1 Å corresponding to the [001] plane
28] as well as 9.9 Å associated with the [010] plane [29].
2
O
3
(
[
The synthesized GaN-650 catalyst had a Ga
2
3
O conversion after the nitridation of around 46 %;
thus, as expected, both GaN and Ga
2
3
O phases were visible in the TEM, with a possible core-shell
structure of Ga
2
3
O in the center (Fig. 2B and S6). GaN has a regular wurtzite structure with both
m- and c-planes. The m-plane is non-polar and made of alternating Ga and N ions, while the c-
plane is polar containing either Ga or N ions. The lattice parameters of the m-plane with 5.2 Å, as
well as the hexagonal structure of the c-plane with √3a = 5.5 Å could be observed, as illustrated in
Fig. 2B and Fig. S7, respectively. For the amorphous Ga
2
3
O part, a d-spacing of 9.9 Å was
measured. An amorphous gallium oxynitride (GaO
x
N ) might exist at the boundary between GaN
y
and Ga
2
3
O .
Fig. 2C depicts the EDS for fresh Ga
2
3
O , and fresh and spent GaN-650 catalysts. As expected, the
pure Ga
2
3
O sample had no peaks corresponding to nitrogen and much higher oxygen counts than
both of the GaN-650 samples (fresh and spent) that contain GaN as well as Ga
2
3
O phases. The
spent GaN catalyst had a higher carbon peak and, consequently, smaller peaks associated with
2
1

nitrogen and oxygen, confirming the presence of carbonaceous material deposition. Some
contributions to the carbon signal and all the copper count were from the sample grid. The EDS
data shown were normalized based on the gallium count and semi-quantitative in nature.
3
.2 Nitridation of supported catalysts
Supported gallium oxide (Ga /SBA15) samples showed a slightly higher H
nitridation when compared with SBA-15 without gallium oxide (GaN/SBA15 vs. N-SBA15 in
Fig. 3A). But no maximum was observed, which was probably due to the lower Ga content
2
O
3
2
O signal upon
2
O
3
(around 35 mg compared to 1000 mg for the unsupported Ga
2
3
O loaded for nitridation). As
mentioned above, the degree of nitridation (i.e., Ga
the product (GaN/SBA15) could not be determined.
The nitridated gallium-containing supported catalysts showed the presence of crystalline GaN but
did not exhibit any peak corresponding to β-Ga (Fig. 3B). Even for the Ga /SBA15 samples,
2
O
3
conversion) by total nitrogen analyses of
2
O
3
2
O
3
no diffraction pattern for β-Ga
2
3
O was observed, indicating very small non-crystalline particles.
The XRD peak around 2θ = 44° (Fig. 3B) corresponds to the aluminum sample holder of XRD.
Although the total N could not be performed for the supported catalysts, the peaks corresponding
to the nitrides are sharper at all nitridation temperatures compared to the unsupported catalysts
(GaN-600 and 650). Since the total Ga content in the supported catalyst was much lower, it can
be assumed that the conversion of Ga to GaN was near completion for all samples.
2
O
3
2
2

2
–1
The prepared support SBA-15 had a total BET surface area of 912 m g (~20% micro and ~80%
2
–1
mesopores), which decreased to 420 m g (~15% micropores) when nitridated at 800 °C (N-
SBA15-800). This might be due to partial pore collapse at higher temperatures, as shown by the
3
–1
reduced pore volume (i.e., from 0.89 to 0.61 cm g ). Adding Ga to the support (including
2
–1
impregnation and calcination at 550 °C) also reduced the total surface area to 426 m g and pore
3
–1
volume to 0.53 cm g but did not change the average pore size (Table 2). Moreover, the data
show that the share of the micropore surface area decreased from 20% to 13%.
The surface areas of Ga
2
3
O /SBA15 was still 20 times larger than for the unsupported catalyst
2
–1
samples with a value of 19 m g (see Ga
2
O
3
in Table 1). Thus, it was easier for NH to access
3
Ga
temperatures.
Upon nitridation up to 700 °C, the surface area did not change much, but the average pore size
decreased from 7.9 to 6.4 nm when compared to the Ga /SBA15 sample. An increase in the
nitridation temperature above 750 °C (i.e., GaN/SBA15-750 and 800) resulted in the reduction of
2
O
3
resulting in most likely in a higher conversion during the nitridation, especially at lower
2
O
3
2
–1
the total surface area down to 320 m g . The share of micropores decreased even further to 8%.
At higher nitridation temperatures, the surface area reduction was most likely due to a combination
of sintering of GaN particles and pore structure collapsed [26].
Compared to the unsupported catalyst (GaN), the supported catalysts (GaN/SBA15) had a
unimodal pore size distribution at around 6.4 nm (Table 2, Fig. S9).
2
3

Based on ICP analyses, Ga-loadings of 13 ± 1 wt% in Ga
1 wt% in GaN/SBA15 were determined. The loss of around 3 wt% from the target loading could
be attributed to the loss of gallium nitrate solution during catalyst synthesis.
HRTEM was carried out for Ga /SBA15, GaN/SBA15-650, and GaN/SBA15-800 catalysts,
2
O
3
/SBA15 (target 16 wt% Ga) and 11
±
2
O
3
both fresh and spent (Fig. 4 and Fig. S10). For the fresh Ga
2
3
O /SBA15, no gallium nanoparticles
were observed, while the ordered hexagonal pore structure of the SBA-15 was clearly visible with
a size of around 8 nm (same as determined by BET measurement see Table 2). In EDS, however,
the presence of Ga was confirmed (Fig. 4D). This suggests that the Ga
2
O nanoparticles were
3
highly amorphous as supported by XRD results for the Ga
2
3
O /SBA15 (Fig. 3B). GaN, on the other
hand, was clearly distinguishable from the SBA-15 in the TEM, as depicted in Fig. 4B and C for
GaN/SBA15-650 and 800, respectively. Moreover, Fig. 4B and C show clearly that the GaN
nanoparticles (3-5 nm, Fig. 4 B and C) were inside the 6-8 nm pores of the SBA-15 support. XRD
measurement confirmed that GaN had a crystalline structure (see Fig. 3B).
Although the spent GaN/SBA15-800 did not show much change in morphology, several
agglomerations of particles inside the hexagonal array of pores for the spent GaN/SBA15-650
were observed with the TEM (see Fig. S10B). These agglomerations were most likely carbon
deposition due to coking during the methane conversion. EDS analysis (Si normalized) confirmed
that the spent GaN/SBA15-650 catalyst had a much higher carbon count than the fresh catalyst
2
4

(Fig. 4D). Consequently, the peak corresponding to Ga was significantly smaller in the spent than
in the fresh catalyst.
3
.3 Activity measurements over unsupported catalysts
Mass spectrum analyses revealed that the Ga
2
O
3
sample produced amount of H O (m/z =18) and
2
CO (m/z =44) in the first hour during the methane activation (Fig. 5A to D), while the nitridated
2
catalysts did not exhibit any formation of H
2
O and CO (Fig. 5E to H and Fig. S11 for GaN-750).
2
The GaN-600 sample showed an insignificant amount of H
2
O and CO , (see Fig. S12), even
2
though this sample contained ~65% of Ga
2
3
O . This may indicate that during the nitridation, mostly
the outer surface of the low porosity Ga
2
O
3
was converted to GaN, while the bulk remained Ga
2
O
3
(
core-shell structure).
This also explains the different behavior of the H
2
formation (m/z = 2) for the Ga
2
O
3
(max. H
2
at
2
h) and GaN (max. H
2
at < 1 h) samples. During the induction period, CH
4
produced H O, CO,
2
and CO on Ga 3, creating an oxygen vacancy in the crystal lattice, also predicted by our DFT
2
2
O
work [21]. SEM-EDS analysis of fresh and used Ga
2
3
O catalysts (Fig. S13 and S14) showed only
a marginal loss of oxygen within the error of estimation. Also, the XRD for fresh and spent Ga
2
O
3
were practically indistinguishable, indicating no change in the crystal structure (Fig. S15).
Since CO and C
2
4
H have the same nominal mass (m/z =28), it was impossible to distinguish
between them via mass spectrometry. CO and CO might have been formed via gallium-methoxy
2
2
5

1
3
species (Ga–O–CH
10]. However, the number of methoxy groups are assumed to be much smaller than the gallium-
methyl species. The reaction pathway towards CO and CO has been reported by Chaudhari et
al.[21]. It has been shown that gallium oxide is active towards the reverse water-gas shift reaction
during the alkane dehydrogenation in the presence of CO [30]. Assuming the reverse water-gas
3
) that were experimentally observed via C NMR over Ga-modified zeolites
[
2
2
shift reaction at equilibrium, the maximum CO formation and hence the minimum C
2
4
H formation
rates could be determined. Since negligible amounts of H
2
O and CO were observed for the
2
nitridated samples, the mass-to-charge ratio m/z = 28 was associates with C
F).
2
4
H only (Fig. 5E to
All unsupported catalyst samples exhibited similar behavior in terms of CH
as product formation rate except ethylene. The ethylene rates were higher for the nitridated
catalysts when compared with the oxide precursor (Ga ), but the rates were similar within the
4
conversion as well
2
O
3
nitridated samples from 600-800 °C. At the onset of the reaction, CH
4
conversion values of 4%
were determined, which rapidly decreased to less than 1% after 3 h on stream. The steady-state
rate attained was up to 0.1 mmolCH4 h gGa–1.
–1
C
2
4
H had the highest hydrocarbon-based formation rate at the beginning of the reaction with up
–
1
–1
to 5 µmol min gGa that was 3 to 4 orders of magnitude larger than for the other hydrocarbons.
Benzene formation rates were about 100 times smaller, while the rates for propylene, toluene, and
2
6

naphthalene (Fig. S16) were more than 1000 times lower. The maximum hydrocarbon and H
formation rates for the nitridated catalysts occurred within the first hour, while the non-nitridated
catalyst (Ga ) had longer induction time with an observed the maxima between 1 and 2 h (Fig.
and 7).
2
2
O
3
6
Independent of the nitridation temperature and time, all unsupported catalysts deactivated faster
within 3-5 h due to coking (Fig. 6 and 7), which was evident from the color change from yellow
(
fresh GaN) or white (fresh Ga
programmed oxidation results (see below). The degree of nitridation influenced slightly more the
formation rates of aromatic compounds (C and C ) than of olefins (C and C ). The
2
O
3
) to black (spent, Fig. S17) as well as the temperature-
6
H
6
7
H
8
2
H
4
3
H
6
Ga
2
O
3
catalyst had the lowest C
6
H
6
and C
7
8
H formation rates, followed by the catalyst nitridated
6
00 °C (GaN-600). GaN-700 and GaN-750 exhibited the highest C
6
H
6
and C
7
8
H formation rates.
If C
2
H
4
would be the only hydrocarbon product, the ratio of the observed H
2
to C
2
4
H formation
rates should be 2, corresponding to the stoichiometric factor shown in equation 5.
2
CH
4
 C
2
H
4
+ 2H
2
(5)
For the unsupported catalysts, the ratio of hydrogen and ethylene formation rate changed with
time, as illustrated in Fig. S18A and B. At the onset, the ratio was close to two, while it increased
to a maximum of 7-12 depending on the catalyst after 1 to 2 h on stream. Afterward, the ratio
attained a value of 3 at the end of the run. The ratio was higher than 2; thus, more H was formed,
2
2
7

indicating that a considerable amount of hydrocarbons were adsorbed on the catalyst surface as
CH * or C * species with a yet unknown stoichiometry. Since the C formation rate was a
factor of 100 to 1000 larger than the other gaseous hydrocarbons, they would not influence the
observed H /C ratio significantly. No methane activation was observed when the reaction was
x
H
y z
H
2 4
2
2
H
4
carried out in an empty quartz reactor (blank experiments); therefore, homogeneous gas phase
reactions were negligible at 700 °C.
A single CO
programmed oxidation (TPO-MS), which shifted slightly to higher temperatures 480 ± 6 °C for
the GaN catalysts as illustrated in Fig. 8. A H O peak (m/z = 18) was observed at the same
2
peak at 467 ± 5 °C for the spent Ga
2
O
3
was observed during temperature-
2
temperature (Fig. S19A). Based on this temperature range, it can be assumed that the
carbonaceous material was rather amorphous instead of graphitic carbon, which requires higher
oxidation temperatures [31]. In addition, the single CO peak indicates a single carbon surface
2
species that behaved similarly on the GaN and Ga
2
3
O samples under an oxidizing atmosphere.
TPO-TGA measurements of the spent samples confirmed the results, as depicted in Fig. S20A and
B. Interestingly, at temperatures above 550 °C the weight of the used GaN samples increased due
to the oxidation to Ga
2
3
O , which was also observed in our previous study [20].
The quantitative analysis of the spent catalyst after 5 h yielded a relative carbon amount of 22 mg
–
1
–1
g
cat for Ga
2
O and 10-16 mg gcat for GaN catalysts (Table S1).
3
2
8

Based on the TPO analysis and product gas composition data, the overall selectivities for the
adsorbed carbon as well as specific selectivities and total C yield were determined and
2
H
4
summarized in Table 3. It is evident that the adsorbed carbon/coke, yet with an unknown C
x
H
y
stoichiometry, was the primary product, whereas the total hydrocarbon selectivity was between
3
4
0-50 mol%. The hydrocarbons itself were predominately C
2
H
4
with selectivity values of up to
. For the latter, a range was calculated
selectivity depending on the influence of
and C selectivities ranged
9 mol% for GaN-750-03 and 27 to 34 mol% for Ga
2
O
3
indicating the minimum and maximum value for C H
2
4
the reverse water-gas shift reaction and thus CO formation. C
from 0.5 to 0.9 mol% and 0.1 to 0.3 mol%, respectively.
6
H
6
H
3 8
To put these results in context, the equilibrium composition was calculated with the assumption
of CH as the only reactant and H2, as the only products. Aromatic compounds
4
C
2
H4, and C H
3 6
benzene, toluene, and naphthalene being less than 100 times than C
2
4
H had not been considered
in the calculations. Thermodynamically, the maximum CH
4
conversions and C
2
4
H yields (without
coke) are 4.1% and 3.5% at 700 °C, 9.1%, and 8.1% at 800 °C and 17.0% and 15.5% at 900 °C,
respectively (see Fig. S21). Carbon deposition was excluded from the conversion calculations
(Fig, 6) as the molar rates of surface intermediates were not measured as a function of time.
Methane conversion based on carbon incorporated in the products (excluding coke) can be
compared with the equilibrium calculations (Fig. S21). The equilibrium calculations did not
2
9