Document downloaded from:
This paper must be cited as:
The final publication is available at
Copyright
Additional Information
http://doi.org/10.1126/science.aag0274
http://hdl.handle.net/10251/81703
American Association for the Advancement of Science
Hernández Morejudo, S.; Zanón González, R.; Escolástico Rozalén, S.; Yuste Tirados, I.;
Malerod Fjeld, H.; Vestre, PK.; Coors, WG.... (2016). Direct conversion of methane to
aromatics in a catalytic co-ionic membrane reactor. Science. 353(6299):563-566.
doi:10.1126/science.aag0274.
Direct conversion of methane to aromatics in a catalytic co-ionic membrane
reactor
S.H. Morejudo, R. Zanón
2
, S. Escolástico
2
, I. Yuste
1
, H. Malerød-Fjeld
1
, P. K. Vestre
1
,
W. G. Coors
1
, A. Martínez
2
, T. Norby
3
, J. M. Serra
2*
, C. Kjølseth
1*
1
Coorstek Membrane Sciences, Forskningsparken, Gaustadalléen 21, NO-0349 Oslo, Norway.
2
Instituto de Tecnología Química (Universitat Politècnica de València - Consejo Superior de
Investigaciones Científicas) Av. de Naranjos s/n, 46022 Valencia, Spain.
3
Department of Chemistry, University of Oslo, FERMiO, Gaustadalléen 21, NO-0349 Oslo,
Norway.
*corresponding authors (email): jmserra@itq.upv.es, ckjolseth@coorstek.com
Abstract:
Non-oxidative methane dehydroaromatization (MDA:6CH
4
↔ C
6
H
6
+ 9H
2
) using shape-
selective Mo/zeolite catalysts is a technology to exploit stranded natural gas reserves by direct
conversion into transportable liquids. The reaction, however, faces two major issues: the one-
pass conversion/yield is limited by thermodynamics, and the catalyst deactivates fast due to
the kinetically-favored formation of coke. Here we show that integration of an
electrochemical BaZrO
3
-based membrane exhibiting both proton and oxide ion conductivity
into an MDA reactor enables high aromatic yields and outstanding catalyst stability. These
effects originate from the simultaneous extraction of hydrogen and distributed injection of
oxide ions along the reactor length. Further, we demonstrate that the electrochemical co-ionic
membrane reactor enables high carbon efficiencies (up to 80%) significantly improving the
techno-economic process viability, and sets the ground for its commercial deployment.
One Sentence Summary: The integration of a co-ionic membrane in a MDA reactor
remarkably enhances aromatics yield and catalyst lifetime.
Main text:
Natural gas constitutes a large and relatively clean fraction of the fossil hydrocarbon
resources, but high capital cost of multi-stage industrial conversion via syngas leaves much of
it stranded. Non-oxidative methane dehydroaromatization (MDA) is a promising catalytic
route allowing direct conversion of natural gas into valued petrochemicals such as benzene.
The MDA reaction is conventionally run at around 700ºC in presence of bifunctional catalysts
comprising carbided molybdenum nanoclusters dispersed in acidic shape-selective zeolites
such as ZSM-5 and MCM-22 (1). The process suffers from two major hurdles that challenge
its further development and industrial implementation: The per-pass conversion to aromatics
is limited by thermodynamics, and the catalyst activity rapidly drops with time on stream
owing to the accumulation of polyaromatic-type coke on the external zeolite surface that
impedes the access to internal active sites (2, 3). Attempts to overcome thermodynamic
limitations by selective removal of the co-product hydrogen from the MDA reactor using, for
instance, Pd-type (4) or ceramic (La
5.5
W
0.6
Mo
0.4
O
11.25-δ
) (5) membranes were not satisfying
due to enhanced coke formation that accelerated catalyst decay. Strategies based on fine-
tuning the zeolite acidity and porosity and co-feeding small amounts of CO
2
, CO, H
2
, and
H
2
O with methane were applied to stabilize the catalyst by restraining coking, but with
limited success (2, 6, 7). Recently, a direct non-oxidative methane conversion path on single-
iron sites embedded in a silica matrix (Fe@SiO
2
) with almost no coke formation and high
stability has been reported (8). This however requires very harsh conditions (950ºC) and
produces ethylene (rather than liquids) as the major product with selectivity of ca. 55%.
We here present a novel approach to circumvent the current limitations of MDA
reaction by integrating an ion-conducting membrane in the catalytic reactor. We report an
innovative catalytic membrane reactor (CMR) for intensification of the MDA process,
resulting in high and prolonged aromatic yields. The CMR is driven by a tailored co-ionic
membrane that enables fast and accurate simultaneous control of hydrogen extraction and
injection of oxygen species along the catalyst bed (Fig. 1A). The concerted action of both
functions leads to unprecedented gains in terms of aromatics yield and catalyst stability, and
consequently enabling the MDA technology.
The electrolyte of the membrane is based on acceptor doped BaZrO
3
which takes up
protons from steam and exhibits high proton (H
+
) and minor oxide ion (O
2-
) conductivity at
elevated temperatures (9). Applications using its protonic conductivity have shown promising
results (10-12), but, as shown here, it is in fact the co-ionic transport property of the material,
more specifically the conduction ratio of protons and oxide ions, that allows the successful
implementation into the MDA process. The tubular membrane consists of a dense 25µm thick
BaZr
0.7
Ce
0.2
Y
0.1
O
3-δ
(BZCY72) electrolyte film on a porous BZCY72-Ni support which also
works as the cathode (11). The metallic Ni has sufficient catalytic activity for the hydrogen
evolution and reduction of steam (Fig. 1B). A Cu-based anode is applied on the electrolyte
film so as to face the catalyst. It activates the electrochemical oxidation of H
2
into protons
while preventing secondary conversion of hydrocarbons into coke as typically reported for
Ni- or Pt-based electrodes (13). As the current density is increased, both hydrogen extraction
and oxygen injection increase proportionally, where the amount of oxygen injected is about
0.3% that of extracted hydrogen (Fig. 1C).
Figure 2 shows results of MDA experiments, comparing our CMR with a fixed bed
reactor (FBR) under otherwise similar conditions utilizing 6Mo/MCM-22 as catalyst. The
catalyst behaviour in the FBR is fully representative of the state-of-the-art at standard MDA
conditions: the aromatics yield initially increases during the induction period, reaches a
maximum of ca. 10%, and rapidly falls as the reaction progresses. In contrast, by applying an
electrical current to the CMR (ON mode) the aromatics yield continues to increase beyond the
induction period and attains a maximum of ca. 12% after which the catalyst activity starts to
decline (Fig. 2A). Worth to note is the almost instant catalytic response (also for conversion,
see Fig. S2) to ON-OFF switching as well as to changes in the intensity of the imposed
electrical current, which empowers our CMR system with the ability to accurately tune the
catalytic performance. Interestingly, the enhancement in conversion/yield observed upon
current application occurs while maintaining the characteristic high selectivity to aromatics,
particularly to benzene (>85% on a coke-free basis, Fig. 2B) of the shape-selective
6Mo/MCM-22 catalyst. Note, however, that operation in the CMR produces some CO, albeit
in relatively low selectivity (vide infra). The most striking result in Fig. 2A is, certainly, the
excellent stability of the catalyst in the CMR, with an average decay rate about one order of
magnitude lower than that observed in the conventional FBR. In consequence, while the
aromatics yield lowers to only ∼1.5% in the FBR after 45 h of reaction, it remains as high as
∼9% in the CMR, translating into a two-fold increase in the cumulative yield (Fig. 2C). The
remarkable stability exhibited by the catalyst in the CMR arises from a decreased coke-
forming tendency, which becomes more evident at increasing reaction times (Fig. 2C).
Thermodynamic calculations predict that in situ H
2
extraction increases methane
conversion and shifts selectivity towards heavier aromatics (and ultimately coke) at the cost
of benzene and C
2
hydrocarbons (Fig. S3), as experimentally proved using H
2
permselective
membranes (14, 15). While thermodynamics thus account for the increase in methane
conversion, the high benzene selectivity and improved catalyst stability during the galvanic
operation in our CMR cannot be anticipated by considering merely effects related to the in
situ H
2
extraction.
The BZCY72 membrane enables the concomitant transport of oxide ions towards the
catalytic reaction medium where they rapidly oxidize the produced H
2
to steam at the
electrode (16). We therefore investigated the isolated effect of steam on the performance of
6Mo/MCM-22 catalyst in the FBR by co-feeding 0.25-0.9 mol% steam together with
methane, corresponding to steam concentrations within the range achieved by the oxygen
injection in our CMR. Whereas the observed decrease in both conversion and aromatics
selectivity (Fig. S4) is thermodynamically consistent (17), post-reaction characterization of
the spent catalysts by TGA and TPO analyses shows that the improved stability achieved in
the CMR is ascribed to the inhibition of coke formation by the in situ generated steam (Fig.
S5). The steam-promoted coke suppression during MDA has also been reported for an
oxygen-permeable membrane reactor (18) and likely occurs by a mechanism involving
scavenging of reactive carbon from the catalyst surface via steam reforming (19), which
accounts for the observed formation of CO (Fig. 2b). It is worth noticing the superior stability
achieved in our CMR as compared to the FBR experiment with an equivalent steam
concentration (0.25 mol%). This indicates that the controlled and distributed oxygen injection
is more effective in improving the catalyst stability than the continuous external addition of
steam. Additionally, the analysis of the XANES spectra at the Mo K-edge and XPS Mo3d
spectral signals (Fig. S5-Table S1) did not reveal appreciable changes in Mo speciation during
CMR operation with respect to the FBR fed with pure methane. Conversely, a higher average
Mo oxidation state is inferred for the catalyst used in the FBR experiment co-fed with 0.9
mol% steam, which might imply a certain loss of active molybdenum carbide species by re-
oxidation (19). We highlight that the crystalline structure of the zeolite host remained almost
intact upon contact with the in situ generated steam under MDA conditions (Fig. S5).
Therefore, the distributed O
2
injection allowed by the BZCY72 membrane effectively reduces
the coking rate while preserving the structural integrity of the zeolite and active Mo-carbide
sites.
A key hypothesis motivating the CMR is that in situ extraction of H
2
will shift the
equilibrium of the formation of aromatics and this will have major consequences in the
process industrialization. In Fig. 3A the experimentally obtained yield of aromatics is plotted
as a function of the magnitude of both H
2
extracted and O
2
injected. High H
2
extraction rates
(>60%) with respect to the H
2
produced in the MDA reaction can be achieved by using the
electrochemical cell reactor. By increasing the magnitude of the imposed co-ionic current, the
aromatic yield raises and surpasses the theoretical equilibrium yield (12.3%) at H
2
extraction
rates above 50%. As expected from the coke-suppression mechanism operating in the CMR,
CO formation is negligible when no current is imposed and raises parallel to the aromatics
yield with increasing co-ionic currents (Fig. 3A). These results unambiguously prove the
galvanic-driven solid-state injection of atomic oxygen ions. As a consequence of the oxide ion
supply and the resulting reduced coking, the catalyst degradation rate drastically drops by a
factor of 6 with respect to the FBR at low extraction rates and then continues decreasing
smoother at increasing currents (Fig. 3B).
To assess the practical implications of the described CMR we have performed process
simulations using Aspen tools. Fig. 4A schematizes a complete gas-to-liquid process based on
our MDA reactor architecture and includes recycling of the reactant methane stream. In this
process, the critical parameter for maximizing the per-pass conversion is the hydrogen
concentration in the recycle loop. By including a methanation stage, CO is converted to
methane and steam giving a typical H
2
concentration of 5% at the reactor inlet (Fig. S7).
Experimental CMR results under recycle operating conditions (5% H
2
co-feed) gives
aromatics yields ca. 6.5% with near-zero degradation rate (Fig. 4B). The process performance
metrics (Fig. 4C) for different extraction rates (60- 80%) are compared with (i) a plant based
on a conventional MDA reactor, implementing downstream gas fractioning using polymeric
membranes (FBR-PolyM) (20); and (ii) a plant based on a CMR employing Pd-membranes
(Pd-CMR) (4). Carbon efficiency is superior for our CMR system, improving steeply with
increasing extraction rates. At rates above 80%, the carbon efficiency achieved is similar to
that exhibited by large and optimized Fischer-Tropsch (FT) plants. The difference between
both processes relies on plant size and complexity. While traditional FT process requires
multiple steps including syngas production, the MDA co-ionic CMR produces aromatics
directly. This feature allows for modularity and flexibility to adapt to the size of the natural
gas field in contrast to FT plants that become uneconomic at small/medium scale (1 to 10
metric tons hour
-1
).
Fig. 1. Current controlled co-ionic membrane reactor. (A) Methane is converted to benzene
and hydrogen over Mo/zeolite. Hydrogen is transported as protons to the sweep side. Oxide
ions are transported to the reaction mediumto react with H
2
and form steam as intermediate
before to react with coke to form CO and H
2
. (B) SEM images of themembrane electrode
assembly (FIB section). Cathode porosity formed upon reduction of NiO can be observed
beneath the dense electrolyte. (C) H
2
extracted and O
2
injected (%) versus current density at
700°C.Cathode is swept with H
2
/CH
4
(10/90) and anode with H
2
O/H
2
/Ar (3/5/92).
