Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

Figures for FT by liaoqinmei


									Draft:                       FT-2; 16th July 2010.

Paper for submission to:     Journal of Catalysis ?

Towards an improved understanding of the multi-variate phase
composition of an iron catalyst active for the hydrogenation of carbon

Neil G. Hamilton 1, Ian P. Silverwood 1, Paul Webb 2, Robert P. Tooze 2, Stewart F.

Parker 3, Christopher D. Frost 3 and David Lennon 1*.

   1. Department of Chemistry, Joseph Black Building, University of Glasgow,
      Glasgow, G12 8QQ, Scotland, UK.
   2. Sasol Technology UK Ltd, Purdie Building, North Haugh, St Andrews, Fife,
      KY16 9ST, UK.
   3. ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot, Oxon OX11
      0QX, UK.

Proofs to:–

Dr David Lennon MRSC CChem,
Department of Chemistry,
Joseph Black Building,
The University of Glasgow,
Glasgow, G12 8QQ.
Scotland, U.K.

Fax:         (+44)-(0)-141-330-4888.


A prototype iron Fischer-Tropsch catalyst has been evaluated using the hydrogenation of
CO as a test reaction at 623 and 723 K with regard to the effectiveness of a pre-reduction
stage. Conventional micro-reactor studies define the reaction profile over a 6 hour
period. The reaction was then scaled up to accommodate analysis by inelastic neutron
scattering (INS). The resulting INS spectrum of the catalyst provides access to the
vibrational spectrum in the range 200 - 4500 cm-1, which is characterised by a
combination of magnetite phonon modes and hydrocarbon features. Although specific
molecular speciation is not possible, the latter entity for the 723 K sample is attributed to
the presence of a hydrocarbonaceous overlayer that is primarily comprised of polycyclic
aromatic compounds.      Post-reaction powder-XRD and TEM analysis of the catalyst
establish the presence of magnetite and iron carbides, with Hägg carbide playing a
dominant role. The pre-reduction stage is judged to be unhelpful in increasing methane
yield. Interestingly, the INS spectrum of the hydrocarbonaceous overlayer that forms
during the course of the reaction is seen to be sensitive to sample history. Trends
observed are discussed with respect to a recently proposed model for how iron based
catalysts evolve during reactions with synthesis gas at elevated temperatures.


1.          Introduction
Currently the majority of hydrocarbons required by our energy intensive world are
derived from oil-based feedstocks [ref]. Due to its significant role in providing access to
hydrocarbons from a range of sources, there is an increasing interest in Fischer-Tropsch
(F-T) catalysis and associated technologies [ref]. One such variant of the process that is
receiving much attention is the use of iron based catalysts, where recent studies have
established the inherent complexity of the actual working catalyst. For example, Schultz
has described the evolutionary behaviour of F-T catalysts and has highlighted the need to
further understand the transformations that take place as a catalyst passes through a
conditioning period on the way to steady-state production. In a variety of studies, Dry
has indicated the dynamic nature of iron based catalysts and goes on to highlight the
multivariate nature of the substrate responsible for optimum F-T activity [ref]. Work
from Shroff and co-workers have used a combination of electron microscopy and
diffraction techniques to establish the importance of iron carbides to the overall process
[ref]. More recently, Fierro and co-workers have examined the genesis of these iron
carbides and evaluated their role in the synthesis of hydrocarbons from synthesis gas
[ref]. Nevertheless, despite all this progress, a comprehensive understanding of what
constitutes an optimised F-T catalyst remains elusive at this time. Whereas all of these
studies illustrate the inherent complexity of a working F-T catalyst, they also demonstrate
that pre-treatments can influence the ultimate composition and performance of the actual

The work here builds on a previous study from this group that used the CO
hydrogenation reaction at ambient pressure as a test reaction to evaluate changes that
occur as an unsupported iron catalyst, representative of iron F-T catalysts [ref], is
exposed to CO/H2 mixtures at different temperatures. These conditions favour methane
formation, therefore one is not actually exploring F-T chemistry as no polymerisation
reaction takes place. However, as demonstrated by Fierro et al. [ref], this reaction regime
does address aspects of F-T related chemistry. The resulting knowledge base can then be
applied to the more complex F-T scenario in due course.

The previous work used a combination of conventional reaction testing, inelastic neutron
scattering (INS), temperature programme oxidation (TPO), powder X-ray diffraction
(XRD), Raman scattering and transmission electron microscopy (TEM) [ref].              The


application of INS to this important catalytic system is novel and provided new
information on the presence of a hydrocarbonaceous overlayer present at the catalyst
surface after a fixed period of reaction. Further, by making use of recently developed
quantitative procedures applied to INS investigations on heterogeneous catalysts [ref], the
extent of hydrogen retention by the catalyst was determined.           Correlation of this
information with structural information gleaned from XRD and TEM measurements, plus
the use of TPO to quantify retained carbonaceous species, enabled a model for the
composition of the active phase of the iron catalyst to be proposed. This correlated well
with previous descriptions of active F-T catalysts [ref].

This article builds on the previous study and goes on to consider whether a pre-reaction
reduction stage can affect catalytic performance. Schult and co-workers have previously
considered the effects of hydrogen activation of iron-based samples [ref10, fierro 1999].
Fierro et al. have considered a variety of pre-treatments, which are seen to affect the
reactive phase of the catalyst [ref]. Our previous study on an iron oxide (hematite)
catalyst involved no specific pre-treatment; instead the catalyst was activated/conditioned
in the reaction gas stream comprising CO/H2/He. This work uses a similar combination
of experimental probes to investigate the influence of a specific reduction step prior to
exposing the catalyst to the synthesis gas feed-stream. The pre-treatment is seen to
influence the methanation activity and the degree of carbon laydown. Interestingly, the
nature of the INS spectrum that correlates with the catalyst in an active phase is different
to that seen in the previous study. Thus, it appears that INS can be used to obtain the
vibrational fingerprint for the active phase of the iron catalyst under investigation. This
is an important result that establishes the usefulness of applying this technique to
catalytic systems which, due to strong substrate absorptions [ref], are not amenable to
investigation by infrared spectroscopy. Moreover, TEM measurements are extended to
include elemental mapping of the metal crystallites. The various strands of information
are then pooled together and discussed with respect to a recently proposed model for the
phase composition of the active catalyst [ref].


2.         Experimental
The general experimental arrangements are described elsewhere [ref]. Details specific to
this work are outlined below.

2.1        Catalyst Preparation
The hematite (Fe2O3) catalyst was prepared by the precipitation technique [Fierro
microemulsion paper] where solutions of iron nitrate (Sigma-Aldrich 98%, 1 mol/L) and
ammonium hydroxide (Sigma-Aldrich 28% in double-distilled H2O, 5.6 mol/L) were co-
added to a precipitating bath containing 1L of deionised water at a rate that ensured the
pH of the liquor was maintained at pH 8. A temperature of 343 K was maintained using
a stirrer hot plate fitted with an electronic thermostat. After precipitation was complete
the mixture was allowed to age at temperature for 12 h. The resultant slurry was then
filtered, washed with deionised water and allowed to dry in air before being calcined at
673 K for a period of 8 hours. The resulting orange coloured solid material was then
ground and sieved to a grain size fraction of 250-500 m.

2.2        Microreactor Reaction Testing
A U-shaped tubular quartz microreactor (1/4” diameter) was housed within a temperature
programmable furnace (Neytech 25PAF) that was connected to a stainless steel gas
manifold equipped to supply reactant (CO 99.5 %, CK gas, H2 99.999 %, BOC) and
carrier (He 99.999 %, BOC) gases fed from mass flow controllers (Brooks 5850
operating via a Brooks 5878 control). A stainless steel mixing vessel (150 cm 3, charged
with glass Ballotini balls) in advance of the reactor ensured efficient mixing of the
incident gases. The helium line was fitted with an in-line scrubber (Varian). The eluting
gas stream was monitored by an in-line quadrupole mass spectrometer (Leda Mass,
Residual Gas Analyser, LM22).

For each reaction testing experiment ca. 50 mg of the precipitated Fe2O3 precursor was
used. The pre-reaction reduction stage comprised passing diluted H2 (30% H2 in He at a
total flow rate of 30 mL/min) prior to syngas exposure while a heating rate of 10 K/min
was applied until the temperature of the sample reached 773 K, at which point the
temperature was maintained for a period of 60 mins. The reduced sample was then
allowed to cool to room temperature in flowing helium then the incident gas feed was


switched to a 2:1 H2/CO mixture diluted in helium (CO, 3.35 mL/min; H2, 6.75 mL/min;
He, 21.25 mL/min) at ambient pressure. These conditions provide a total weight hourly
space velocity (WHSV) [Farrauto & Bartholomew] of 10.4 h-1. A linear heating
program of 5 K min-1 was applied up to the stated reaction temperature, where the
temperature of the system was maintained for a period of 6 h. The heating program was
then terminated and the catalyst sample was allowed to cool to room temperature under a
flow of the reactant gases. The mass spectrometer was calibrated for reactants and
products; the resulting values were then used to determine conversion, selectivity and
yield values [WA].

2.3        Inelastic neutron scattering measurements:
2.3.1      Sample preparation
A high capacity quartz tubular reactor (3/4” diameter) capable of handling catalyst
charges up to 50 g was used to prepare samples for the INS experiments. Carbon
monoxide (CK gases, research grade 99.97%) and hydrogen (CK gases, research grade
99.999 %) reactants and helium (Air Products, UHP grade 99.9992 %) carrier gas were
supplied via a series of digital mass flow controllers (Hastings model HFC302) powered
by a Teledyne THPS-400 controller. The gases were mixed in a mixing volume packed
with glass Ballatini balls located before the reactor to minimise laminar flow. The
stainless steel manifold was equipped with pressure relief valves (Parker HPRV S4A-
BN-K1-319) and a back pressure regulator (Parker Veriflo ABP-1ST-43-PP-X4) in order
to maintain the pressure of the system between 1-1.5 bar (absolute). The eluting gases
were continuously sampled via a differentially pumped quadropole mass spectrometer
(Spectra Microvision Plus).

The reactor containing approximately 25 g of hematite (Fe2O3) was contained within a
bucket furnace (Instron SFL, model no. TF105/3/12/F) that was linked to a PID
temperature controller (Eurotherm model 3508).        The control thermocouple was
embedded in the furnace wall. The furnace was packed with quartz wool to minimize
thermal gradients. The sample was reduced in a flow of diluted H2 (25% H2/He, 1L/min)
until H2O evolution, as measured by the on-line mass spectrometer, subsided (duration
ca. 2 hours) before being allowed to cool to room temperature. Reactions were then
carried out using a 2:1 H2/CO mixture diluted in helium (CO, 75 mL min-1; H2, 150 mL
min-1; He, 775 mL min-1; total WHSV = 0.98 h-1) and subjected to a linear heating


program of 5 K min-1 up to the stated reaction temperature. It is readily acknowledged
that the large volume reactor was operated at a much lower space velocity than that
attainable with the micro-reactor. This was a consequence of limitations within the scale-
up procedure, not least the volumes and flows of carbon monoxide and hydrogen that
could be accommodated within the experimental hall of the neutron facility.
Nevertheless, despite these differences, comparable methane yields were achievable from
the micro-reactor and the INS reactor, indicating that comparable chemistry was being
explored in both cases.

After a period of 6h at the stated reaction temperature the catalyst was allowed to cool to
room temperature under a flow of the reactant gases. Really? [Neil/Ian: please amend
accordingly] The reactor was then purged with helium to remove residual gaseous
species before being isolated and transferred to an argon filled glovebox (MBraun
UniLab MB-20-G, [H2O] < 1 ppm, [O2] < 2 ppm). Within the inert atmosphere of the
glove box, the sample was loaded into aluminium foil sachets that were then secured via
indium seals within a standard aluminium INS cell [Neutron book]. Inelastic neutron
scattering (INS) spectra of each of the samples were recorded at 20 K using the MAPS
direct geometry neutron spectrometer located at the ISIS Facility (RAL etc). INS spectra
were recorded with incident neutron energies of 4840 cm-1 and 2017 cm-1. The INS
spectra were processed with in-house software.

2.3.2      Post-reaction analysis:
Samples extracted from the INS cells post-acquisition were stored in a steel cabinet for
several days to enable the induced radioactivity to reduce to background levels.
Thereafter in an ambient atmosphere the samples were transferred to sealed glass vials
for storage. This process inevitably involved a degree of oxidation of the samples. No
exotherm was observed on the opening of the INS cells, where it was assumed that a slow
ingress of air in to the cell over an extended period of time had resulted in a mild
passivation of the samples. These samples were subsequently analysed by temperature
programmed oxidation, powder X-ray diffraction and transmission electron microscopy.        Temperature programmed oxidation (TPO)
Temperature programmed oxidation (TPO) measurements were used to detect and
quantify the carbonaceous deposits present in the catalyst sample post-reaction in the


large volume reactor. For each of the samples a small quantity (ca. 50 mg) was weighed
into a quartz microreactor and subjected to a flow of diluted oxygen (5% O 2/He, BOC, 50
mL/min) while a linear heating ramp (10 K/min) up to 1200 K was applied. The eluting
gases were sampled via in-line quadrupole mass spectrometry.        Powder X-ray diffraction (XRD)
Samples (ca. 1 g) of the iron-based catalysts after exposure to syngas at various
temperatures were ground in a mortar and pestle and then pressed into a silicon sample
holder. XRD patterns were acquired using a Siemens D5000 powder diffractometer
using Cu K radiation in Bragg-Brentano geometry in the 2 range 5-110° (step size
0.02°, 10s per step).        Transmission electron microscopy (TEM)
The microstructure of the post-reaction catalyst samples was studied by transmission
electron microscopy (TEM) using a Tecnai T20 microscope with an accelerating voltage
of 200 keV. Following the procedure outlined by Fryer [F], samples were dispersed in
methanol before being deposited on holey carbon film (300 m mesh grid, Agar
scientific) prior to inspection.    In addition to conventional electron transmission
measurements, energy filtered mapping [ref] was also undertaken. Here, the mapping for
iron and carbon containing components were respectively collected at energies of 700 eV
(L edge) and 284 eV (K edge) with exposure times of 5 and 8 seconds. A slit width of 35
keV was used throughout. The maps are a composite image of one measurement taken at
the ionisation edge and two others taken at energies just preceding the edge.


3.         Results and Discussion
3.1.       Catalyst characterisation
The temperature programmed reduction profile of the Fe2O3 (hematite) precursor in
flowing H2 is presented in Figure 1. The broad water feature located at 398 K and with
no counterpart in the hydrogen consumption plot is attributed to desorption of absorbed
water from the catalyst.     In agreement with Khadkhodayan & Brenner [ref], water
evolution and hydrogen consumption features about 650, 900 and 985 K are respectively
assigned to the progressive reduction of Fe2O3 towards Fe3O4, FeO and Fe0. FeO is
known to be a metastable phase that rapidly disproportionates into Fe0 and Fe3O4
(magnetite) [ref]. Thus, at a temperature of 773 K the reduction of Fe2O3 is expected to
yield Fe3O4 with a small proportion of Fe0.

The XRD pattern of the reduced sample presented in Figure 2 confirms the pre-reduced
catalyst sample to be composed mainly of Fe3O4 (major reflections at 35.1, 62.2, 29.7,
56.6 and 42.7), with a small relative proportion of Fe0 (reflections at 44.4, 82.2 and
64.7, indicated by stars in Figure 2). It is evident from this pattern that the reduction
stage places the iron oxide at a different starting position for the subsequent CO
hydrogenation reaction than when simply starting from hematite (-Fe2O3), as considered
in the previous study [ref]. No features due to goethite [ref] are discernible in Figure 2.

3.2        Micro-reactor studies.
Micro-reactor studies were performed at 623 and 723 K. The low temperature studies
(not shown) displayed minimal reactivity. Respective CO and H2 conversions of ca. 12
and 19 % were evident on commencement of exposure to the CO/H2 feedstream but these
diminished to negligible levels after approximately 150 minutes on-stream. No methane
was produced throughout the full reaction sequence. Apart from a momentary production
of water at 623 K, no other products were detected in the gas phase. These observations
indicate that a temperature of 623 K is insufficient to promote the hydrogenation of CO to
CH4 over either Fe3O4 or Fe0 species.

The 723 K dataset was more productive, with the reaction profile presented in Figure 3.
The reaction profile is shown in 3(a), the associated conversion and selectivity trends are
shown in 3(b) and 3(c) shows the methane yield as a function of time on stream.


Concentrating on Figure 3(a), evolution of CH4 is observed from ca. 125 mins, when the
temperature reaches 723 K. Methane evolution is accompanied by the production of CO2
and H2O. These products progressively approach steady-state operation within the 6h
time period. Interestingly, the CO, CH4 and H2O traces all appear to exhibit the same
evolution profile; this was not the case when the same catalyst was only reacted in a
syngas environment [ref]. In that case, CO2 evolution appeared to reach steady-state
when the temperature reached 723 K. This observation is consistent with the reduction of
iron oxide species in the untreated catalyst to Fe0 by CO, resulting in an initial increase in
CO2 partial pressure.

The production of CO2 in Figure 3a is attributed to a number of possible processes: the
disproportionation of CO (the Boudouard reaction); the water gas shift reaction and the
reduction of iron oxide by CO [240]. Alongside the CO dissociation reaction, the former
process is expected to produce surface carbon. Carbon dissolves rapidly into metallic
carbon [ref], so these reactions constitute a source of carbon from which iron carbide
species may be produced.

The associated conversion and selectivity profiles for the experiment are presented in
Figure 3(b). While the CO and H2 conversions as a function of time exhibit similar
trends, the reaction profile for CO conversion is ca. 2.5 times greater than that for H2.
This indicates that, in accordance with comparable measurements on the unreduced
Fe2O3 catalyst [ref], large quantities of CO are         being consumed in a competing
reaction(s) to methane formation. These observations are consistent with CO being
consumed via a combination of carbon monoxide dissociation and the Boudouard
reaction, providing surface carbon atoms that can proceed to form amorphous carbon,
iron carbides and a hydrocarbonaceous overlayer [ft-1].

The methane yield as a function of time on stream is presented in Figure 3(c), where it is
seen to initiate upon attainment of reaction temperature (723 K).              Thereafter, it
progressively increases from baseline levels up to a maximum of 1.2 % after 470 mins on
stream where it appears to be approaching steady-state production.             The onset of
production suggests that temperatures of ≥ 723 K are required to initiate and sustain
methane production, which is consistent with the negative outcome at 623 K. It is
assumed that the barrier to reaction is the dissociative adsorption of CO over this catalyst.


It is noted that the apparent maximum of 1.2%, is considerably lower than the 12% of a
non pre-reduced analogue previously reported [FT-1]. This is a clear illustration that
sample pre-treatments can directly affect catalytic performance, with the active phase
responsible for the 1.2 % methane yield seen in Figure 3(c) being different to the active
phase present if the pre-reduction step is excluded. In this case, pre-reduction is seen to
impede catalytic activity. The same behavior is also observed under true FTS conditions,
where catalysts pretreated in syngas are reportedly more active than catalysts that have
been pretreated in hydrogen alone [155].

3.2        Inelastic neutron scattering.
3.2.1      Qualitative analysis.
The INS spectra recorded at 4840 and 2017 cm-1 for the Fe2O3 sample pre-reduced in
hydrogen then reacted in syngas for 6 h are presented in Figures 6 and 7, with (a) and (b)
respectively representing the 623 and 723 K datasets. The neutron scattering intensity
evident in these spectra is indicative of a low concentration of hydrogen within both
catalyst matrices.

The spectrum in Figure 4(a) is surprisingly weak and comprises of only a single feature
centred at 3032 cm-1. This is readily assigned to a C-H stretching mode, (C-H), with the
peak position indicating predominantly olefinic and/or aromatic character. Similarly, the
higher temperature sample (Figure 4(b)) is also described by a single (C-H) feature but
this band exhibits significantly more intensity than that seen for the 623 K sample. This
peak is centred at 3091 cm-1, indicating distinct olefinic and/or aromatic character and
suggesting an increase in unsaturated carbon. In contrast to the INS spectra reported
previously [FT-1] corresponding to the iron catalyst that had experienced no reduction
stage, no spectral features due to hydroxyl groups are evident in Figure 4. This suggests
that the hydrogen pre-treatment reduces the hydroxyl groups, thought to be associated
with either residual goethite or hydroxyl groups at the surface termination of either
magnetite or amorphous carbon [FT-1]. It is possible that the water detected in the
elutant stream reported for the inactive sample (673 K, Section 3.2) could be the product
from this dehydroxylation step.

The spectra recorded at a primary energy of 2017 cm-1 are presented in Figure 5, with the
623 K sample (Figure 5(a)) presenting bands at 1470, 960 and 610 cm -1. The low


intensity 1470 cm-1 band is assigned to a CH deformation mode, (C-H), possibly
reflecting a CH2 scissors mode of a saturated hydrocarbon. The 960 cm-1 band, assigned
to an in-plane O-H deformation, (O-H) for the unreduced sample [ref], cannot arise
from a comparable source in this case (no (O-H) in Figure 4). Thus, this feature is
assigned to an out-of-plane C-H deformation of an alkenic hydrocarbon moiety. The
broad feature at 610 cm-1 is thought to be associated with a catalyst substrate mode.
Specifically, it is assigned to an A1g Fe-O phonon mode of magnetite (Fe3O4) [ref].

The high temperature sample spectrum is presented in Figure 5(b), where bands are seen
at 1420, 1190 and 870 cm-1. A weak band at 590 cm-1 is also discernible. The 1420 cm-1
band is assigned to a semi-circle ring deformation mode which is possibly coupled with a
(C-H) mode that is associated with perimeter carbons of a polyaromatic network [ref,
albers et al.]. The 1190 cm-1 band is assigned to a CCH in plane deformation mode of a
polyaromatic hydrocarbon. e.g. as seen for naphthalene [sellers, pulay and boggs, jacs,
1985, 107, 6487].    The band head at 870 cm-1 is assigned to an out-of-plane CH
deformation of either an olefinic or aromatic group which, presumably, is the counterpart
mode for the (C-H) feature seen in Figure 4(b). A shoulder is present on this band at
960 cm-1, assigned above to alkenic (C-H), indicating that a range of chemical species
are present. The intensity of the A1g Fe-O phonon mode of magnetite at 590 cm-1 is
significantly reduced in Figure 5(b) suggesting a loss of coherence within the magnetite
matrix. It is possible that the hydrogen pre-treatment has facilitated reduction of the
magnetite towards metallic iron (Fe0), although it is noted that the magnetite signal is
clearly evident in the lower temperature spectrum. The absence of a (O-H) feature in
Figure 4(a) and (b) indicates that the hydrogen pre-treatment has facilitated the reduction
of hydroxyl groups observed previously [FT-1]. The small signal about 600 cm-1 in
Figure 5(b) seems to indicate that the reduction plus high temperature treatment has
resulted in a reduced magnetite concentration. If magnetite is necessary for the formation
of the relevant iron carbide to support CO hydrogenation (Hägg carbide?), then it is
possible that the reduced concentration of the magnetite could be responsible for the low
methanation rate displayed in Figure 3. This issue will be expanded upon further once all
the experimental evidence has been presented.


In summarising the INS spectra, the 623 K sample appears to indicate the formation of a
hydrocarbonaceous overlayer comprising predominantly unsaturated carbon, with some
aliphatic character. The high temperature sample appears to possess more aromatic
character typical of a polyaromatic hydrocarbon entity, although some olefinic and
aliphatic species also seem to be present. It is unfortunately not possible to assign either
set of spectra to a specific molecular species due to the limited resolution of the
instrument and/or because of the range of moieties present on the catalyst surface.
Interestingly, although these deductions on the nature of the hydrocarbonaceous
overlayer are broadly comparable to that reported previously for non-hydrogen treated
iron samples [ref], the spectra for the high temperature samples that exhibit sustained
methane production are not the same. Thus, the INS spectrum is able to distinguish the
vibrational spectrum of hydrocarbonaceous overlayers associated with active phases of
the iron catalyst, with the surface composition of that overlayer seen to vary depending
on sample history, as does methanation activity.

3.2.       Quantification of retained hydrogeneous moieties.
Previously, it has been shown that INS spectra can be used to quantify the amount of
hydrogen retained on post-reaction heterogeneous catalysts [ref]. The integrated intensity
of the CH) band of each of the catalyst samples after syngas exposure was determined
by fitting the INS data to a single Gaussian function using the Origin graphical software
package (Origin 6.1, OriginLab corporation). The overall fit to the data is indicated by a
red line in Figure 4. Based on the response factor taken from a polystyrene calibration
curve determined on the same spectrometer operated in exactly the same manner used to
acquire the spectra present in Figure 4 [ref], the hydrogen content contained in the
hydrocarbonaceous component of the overlayer of the pre-reduced iron catalyst treated in
syngas at 623 K is of 245  15 moles H g(cat)-1 (0.0247 % w/w). The error represents 1
standard deviation in triplicate INS measurements for hydrogen content of a standard
sample [SHLSOFPL]. Increasing the reaction temperature significantly increases the
intensity of this band, yielding a hydrogen content of 1416  88 moles H g(cat)-1 (0.142
% w/w). These values are presented in Table 1.

3.3            Post-reaction analysis.
3.3.1          Temperature programmed oxidation (TPO)


Temperature programmed oxidation (TPO) measurements were used to detect and
quantify the carbonaceous deposits present in the iron samples after the INS
measurements. The m/z 44 traces from the TPO profiles are presented in Figure 6. The
CO2 evolved during the temperature ramp of each of the post-reaction samples is a result
of the oxidation of the various carbonaceous deposits in each sample including: the
hydrocarbonaceous overlayer; iron carbide species; and possibly „free‟ carbon, which has
been reported to contribute to deactivation in iron-based FTS catalysts [FT Technology

The profile for the low temperature sample (Figure 6(a) is characterised by a Tmax at 700
K that possess a low temperature shoulder at 670 K. There is also a much weaker feature
observable that is centred at 855 K. The high temperature sample (Figure 6(b)) is
dominated by a sharp feature at Tmax = 745 K that includes a low temperature feature at
870 K. Both profiles are asymmetric and are consistent with the presence of several
types of carbonaceous deposit. The TPO spectrum reported previously [FT-1] for an iron
sample only exposed to synthesis gas at 723 K (no hydrogen pre-treatment) showed
negligible intensity above 800 K. Figure 6(b) displays intensity up to 1000 K, which is
close to the Tmax for graphite (1060 K, separate experiments not presented here).

It is apparent from Figure 6 that treatment at 723 K leads to a far greater carbon retention
than that seen at 623 K, although it is the higher temperature sample that displays greater
activity, albeit to a rather modest extent (Section 3.2, methane yield = 1.2 %). The
absolute values for the 623 and 723 K runs are respectively 2050  159 moles C g(cat)-1
(2.46 % w/w) and 19375  1505 moles C g(cat)-1 (23.3 % w/w), see Table 1. The error
represents 1 standard deviation in triplicate measurements of CO2 calibration experiments
on the thermal decomposition of CaCO3. Table 1 also includes values for the counterpart
measurements that excluded a hydrogen pre-treatment [ref] and shows the hydrogen
treatment to lead to increased quantities of carbon. Carbon retention is greatest for the
hydrogen pre-treated sample reacted at 723 K which retains 9.5 more carbon than that of
the 623 K value. Although the methane production rate is low for the hydrogen treated
723 K sample, Figure 3 shows the methane rate to be slowly increasing during this period
of carbon laydown. This is suggestive that the carbon is not directly responsible for the
reduced activity. Almese et al. have shown that metallic iron readily facilitates carbide


formation [ref]. Thus, it seems reasonable to assume that a hydrogen pre-treatment
encourages Fe0 formation which in the presence of synthesis gas then carburizes to form
iron carbides. If the extra carbon evidenced in Figure 6 is due to iron carbide formation,
it seems that this component alone cannot support CO hydrogenation as the methane
yield at 723 K is lower when the catalyst has experienced a hydrogen pre-treatment. If
the Fe0 has also resulted in increased “free carbon” [FT book] deposition however, this
may explain the catalysts decreased performance compared with the unreduced catalyst
[FT-1].      Unfortunately, it is not possible to discern from Figure 6 the relative
contributions from iron carbides, amorphous carbon and the hydrocarbonaceous

3.3.2           X-ray diffraction (XRD)
The XRD patterns for both of the post-reaction catalysts are presented in Figure 7. The
pattern for the pre-reduced Fe2O3 sample after exposure to syngas at 623 K is presented
in Figure 7(a) and reveals the sample to be composed primarily of magnetite (Fe 3O4), as
evidenced by intense reflections at 35.1°, 62.2°, 29.7°, 56.6° and 42.7° [ref]. In addition,
the low intensity relatively broad reflections located at Bragg angles of 44.4° and 82.2°
signify the presence of metallic iron (Fe0). Furthermore, a broad band located at ca. 40°,
can be assigned to Hägg carbide (-Fe5C2) [ref]. Comparing Figure 7(a) with Figure 2
shows the post-reaction sample to contain less Fe0. There are a number of reasons for
this. Firstly, having an increased Fe0 content from the reduction will facilitate carbide
formation (-Fe5C2?) under reaction conditions. Secondly, the low space velocity for the
INS reactor provides a greater opportunity for product water to interfere in the forward
chemistry. For instance, water produced from the CO hydrogenation process could
oxidize Fe0. It is anticipated this side reaction will play a lesser role at shorter residence

The XRD pattern for the sample reacted at 723 K after reduction at 773 K is presented in
Figure 7(b) and contains approximately equal contributions from Fe3O4 and iron carbide.
The presence of metallic iron is inferred by the presence of diffraction lines located at
44.4° and 82.2°. However, the 44.4° signal overlaps with the carbide manifold and the
82.2° signal is coincident with a weak reflection in the Hägg carbide pattern [ref] located
at ca. 82°. Assignment of iron carbide phases by XRD should be performed with care as


-Fe3C and -Fe5C2 have very similar diffraction patterns. Indeed, the sample shown in
Figure 7(b) contains the greatest proportion of carbide for both un-reduced [FT-1] and
pre-reduced samples and additionally corresponds to the largest carbon retention (Table
1). In order to improve the signal to noise ratio about the challenging carbide region, an
extended scan diffractogram was obtained for the pre-reduced high temperature sample
and is presented in Figure 8. This shows both experimental data and reference data [ref]
for Hägg carbide (-Fe5C2), with good correlation. However, the similarities in the
diffraction patterns for Hägg and cementite carbides and the limited resolution inherent in
the measurement prevent the unambiguous assignment of the              total iron carbide
population to Hägg carbide, although it clearly has a dominant role as a structurally
distinct phase.

Fierro et al. [155] report graphitic carbon deposits to exhibit a reflection located at ca.
20° in the XRD patterns of samples exposed to syngas at 723 K. No signal is apparent in
Figure 7. One extra species that ought to be considered is that of amorphous carbon. By
definition, amorphous materials do not feature definitively in powder diffractograms and
neither Figure 7 or 10 provide any evidence for its presence. With other evidence
therefore, it would not be unreasonable to suggest that a dispersed, amorphous carbon,
invisible to XRD, may also be present.

3.3.3      Transmission Electron Microscopy (TEM):
TEM images at low and high magnification for samples of the pre-reduced catalyst after
exposure to syngas at 623 and 723 K are presented in Figures 9 and 10 respectively. The
low temperature sample is characterised by a series of particles of apparently random size
and shape, with some particles possessing a distinct overlayer with a thickness of ca. 5
nm. Micrographs of the high temperature sample (Figure 10) show a series of large
overlapping particles. Similar images have previously been reported by several research
groups [refs]. Here, consistent with deductions made previously [FT-1], these features
are attributed to an iron carbide phase (predominantly Hägg) that are surrounded by a
distinct outer layer of ca. 5 nm thickness. A high magnification image of the sample
(Figure 10(b)) shows the outerlayer to be especially thick (ca. 35 nm) in some areas of
the sample. This thicker “shell” type layer correlates with this sample exhibiting a high
carbon content (Table 1) and a large iron carbide presence (Section 3.3.2).


In order to observe the two dimensional elemental distribution of the post-reaction
sample previously reduced in H2 followed by exposure to syngas at 723 K, energy
filtered mapping [ref] was carried out, with the images presented in Figure 11. The
standard TEM image, taken with zero energy loss, i.e. elastically scattered electrons, is
presented in Figure 11(a), while images filtered for iron and carbon are respectively
presented in Figures 11(b) and (c). The standard transmission image is characterised by
two structures, both possessing an outer layer. The particle on the left side of the
micrograph (Figure 11(a)) shows a slight taper, which terminates in a „hook‟ to the left of
the image. The particle on the right of the image shows a more regular rectangular shape.
At distant edges the outer layer is approximately 5 nm thick but at the top left edge of the
„hook‟ structure it extends to 12 nm. At the top of both structural lobes this layer forms
between the two central dense forms and extends inwards by about 25 nm and across a
distance of ca. 42 nm. From inspection of the contrasts apparent in Figure 11(b), where
increasing whiteness corresponds to increasing concentration, it can be seen that the
particles shown in this micrograph contain a high concentration of iron but; although it is
still discernible as a shadow, the outer layer does not contain iron. Further, the carbon
map (Figure 11(c) shows the centre of the particles not to contain much carbon, although
some is present at the outer part of the iron-containing domain. The outer layer has a
strong carbon signal. These micrographs are interpreted as indicating that the central
dense particles are comprised of iron but not carbon, with the particular structure
examined in Figure 11 assigned to magnetite. The outer shell seems to be predominantly
carbon with no iron content, so that structure is ascribed to amorphous carbon. There is
no obvious contribution of an iron carbide phase in Figure 11.

As described, the absence of an iron carbide feature in Figure 11 contradicts the
previously described model [FT-1] and is inconsistent with associating the dense
structures seen in Figure 10 to iron carbide. However, in recognising the fact that
samples that have not experienced a hydrogen pre-treatment stage exhibit greater activity
in the methanation reaction [FT-1], Figure 12 is proposed to help explain this dichotomy.
Figure 12(a) relates to the previously described suggestion for a catalyst that has
experienced syngas at 723 K [FT-1], where the observed activity is attributed to the
clusters surrounding the magnetite which support the methanation activity.           These
clusters are observable in Figure 10(a). Figure 12(b) is a suggestion for the hematite


catalyst at 723 K that has experienced a pre-reduction step. Here the number of „active
nodes‟ is reduced, hence the activity is diminished. It is assumed that the initial reduction
step has mitigated against formation of the necessary interfaces that support catalytic
activity. The black line in Figure 12(b) is intended to represent the cross section sampled
in the TEM micrograph of Figure 11.                    Namely, it is only sampling a
magnetite/amorphous carbon interface, i.e. an inactive phase of this heterogeneous

4.     Discussion.
Table 1 presented the quantities of hydrogen bound to carbon atoms and the overall
quantity of retained carbon atoms at the catalyst surface after the low and high
temperature treatments. Comparing these values leads to C:H ratios of 8.4:1 and 13.7:1
respectively.   The INS spectrum for the high temperature sample mildly active for
methane formation was characterised in term of a predominantly polyaromatic
hydrocarbonaceous overlayer (Section 3.2). Taking anthracene as a possible candidate
molecule, this equates to a C:H ratio of 1.4:1.           Thus the calculated C:H ratios
significantly exceed molecular species implicated by the INS.          This can be readily
rationalised by assuming that the INS spectrum is indicative of a hydrocarbonaceous
overlayer, which is only associated with a fraction of the retained carbon. With reference
to the XRD and TEM results, the remaining fraction of carbon can be attributed to iron
carbide and amorphous carbon. X and co-workers have previously consider a role for
“free carbon” in F-T chemistry [ref] and our previous work on syngas treated iron
catalysts interpreted a Raman spectrum as signifying the presence of amorphous carbon.
Similar processes are expected to prevail here for these hydrogen pre-treated catalysts, so
the large C:H ratios reported in Table 1 are thought to reflect contributions from
predominantly    iron   carbides    but   with    an    additional   contribution   from    a
hydrocarbonaceous overlayer, with amorphous carbon making up the difference.
Unfortunately, due to limited resolution in the TPO spectrum, the quantity of amorphous
carbon, at these levels thought to be „invisible‟ to XRD and INS and not readily definable
by TEM, remains unknown at this time.

Carbon to hydrogen ratios previously reported for the iron catalyst reacted in syngas, i.e.
no pre-reduction step [ref], are included in Table 1. Comparing the hydrogen-pre-treated
C:H values to those that did not experience the extra hydrogen treatment, it is apparent


that the additional reduction stage leads to a greater proportion of carbon to hydrogen.
As described previously, this is connected with the hydrogen reducing the hematite down
to metallic (Fe0), which then facilitates iron carbide formation.       For both treatment
regimes (syngas reaction with and without hydrogen reduction) only the 723 K data were
active for CO hydrogenation, with the simpler syngas treatment leading to a considerably
more active catalyst. Interestingly, XRD indicates both the highly and moderately active
catalysts reacted at 723 K to have similar structural motifs: magnetite, iron carbide
(predominantly Hägg) and Fe0. Previous work has established the importance of iron
carbides and magnetite for sustaining methanation and F-T activity [refs] yet these results
seem to be suggesting that their presence on its own isn‟t enough to affect favourable
methane yields. Is it possible that the interfacial regions are important?

Section 3.3.2 presents XRD showing similar characteristics to those previously reported
for a non-hydrogen treated high temperature sample. Probably the simplest explanation
for the reduced methanation activity reported in Section 3.2 is that the hydrogen
treatment reduces the number of active nodes that sustain methanation. However, it is
also possible that the active site for the combination of CO and H2 is in fact a discrete
interfacial region (a hydrocarbonaceous overlayer that forms at the interface between an
iron carbide phase and a layer of amorphous carbon?) and syngas treatment favours the
formation of this interface whereas a pre-reduction stage compromises the interface.
Clearly, more work is required to explore this hypothesis further.

Acknowledging that this scenario only relates to the less economically relevant CO
hydrogenation reaction, it is noted that INS appears to have some value in being able to
discriminate between different active phases. Whereas the C-H stretching region is
certainly informative (Figure 4), it is the C-H defomation region that probably provides
the most discerning perspective. Figure 13 presents the INS spectra in the 200-1700 cm-1
region for the 723 K samples corresponding to (a) no pre-treatment and (b) a hydrogen
pre-treatment.     Thus, Figure 13(a) represents the vibrational fingerprint of the
hydrocarbonaceous overlayer that forms over an iron catalyst that corresponds to
favourable methane yields. Interestingly, this spectrum is providing information on the
form of the catalyst particle as well as the nature of the overlayer residing thereon; the
active sample being characterised by a definitive magnetite phonon feature about 600 cm -
    plus specific hydrocarbon modes, whereas the low activity spectrum (Figure 13(b) has a


negligible magnetite contribution and a different profile of hydrocarbonaceous modes.
The acquisition of the vibrational fingerprint for the active phase of a catalyst surface
constitutes a useful component of a feedback mechanism for catalyst optimisation
strategies.   A target for future research is to more closely correlate the vibrational
fingerprint with optimised and sustainable methane yields.         Formulation of crude
structure/activity relations can then be used to guide further investigations when these
catalysts are operated within a Fisher-Tropsch regime.

5.       Conclusions and a postulated model.
A multi-technique investigation has been used to investigate the CO hydrogenation
reaction over a hematite catalyst that has experienced an initial treatment in hydrogen.
Below the main conclusions are listed.

        A hydrogen pre-treated hematite catalyst was inactive for methane formation at
         623 K and provided a methane yield of only 1.2% at 723 K.
        INS spectra reveal a low retention of hydrogen for both the low and high
         temperature catalysts.      The low temperature sample is characterised by a
         hydrocarbonaceous overlayer comprising both aliphatic and olefinic moieties.
         The high temperature sample has more aromatic character typical of a
         polyaromatic hydrocarbon entity, although some olefinic and aliphatic species are
         also apparent.
        Quantification of the INS signals shows the hydrogen content of C-H containing
         entities for the low and high temperature samples to be respectively 245  15 and
         1416  88 moles H g(cat)-1.
        Temperature programmed oxidation measurements shows the retained carbon for
         the low and high temperature samples to be 2050  159 and 19375  1505 moles
         C g(cat)-1 respectively .
        XRD data show the high temperature sample to be composed of magnetite, iron
         carbides (predominantly Hägg but the presence of cementite is also possible) and
         a small quantity of metallic iron.
        TEM measurements are interpreted as indicating the presence of iron carbide
         particles that are accompanied by a “shell” layer attributed to amorphous carbon.


         Energy    mapping      measurements      also    indicate    the   presence   of    a
         magnetite/amorphous carbon interface.
        A pre-reduction stage is unhelpful in increasing methane yield. As XRD and
         TEM data indicate that the sample composition after 6 h on stream is broadly
         similar to that seen in the absence of the additional treatment (no new features are
         evident), it is suggested that the pre-treatment perturbs the formation of important
         interfacial domains that support syngas conversion.
        The INS spectrum of the hydrocarbonaceous overlayer that forms during the
         course of the reaction is sensitive to the sample history.

6.       Acknowledgements.
Sasol Technology UK Ltd., STFC (CCLRC) and the University of Glasgow are thanked
for the provision of a studentship (NGH). The EPSRC are thanked for personnel (IPS)
and equipment funding through grant number EP/E028861/1. The Rutherford Appleton
Laboratory is thanked for access to neutron beam facilities. Technical assistance for the
electron microscopy was provided by Mr Jim Gallagher and Mr Colin How (University
of Glasgow).




                                   Figure Captions.
Figure 1:    Temperature programmed reduction profile for Fe2O3 precursor reduced in
             H2 at a heating rate of 3 K/min.

Figure 2:    XRD pattern for Fe2O3 precursor after reduction in H2 at 773 K for a
             period of ca. 2 h (reflections from metallic iron indicated by stars).

Figure 3:    (a) Reaction profile for CO hydrogenation over reduced Fe-based catalyst
             at 723 K. (b) Associated conversion and selectivity profile. (c) Methane

Figure 4:    INS spectra for post reaction samples of reduced Fe-based catalyst
             exposed to a H2/CO mixture at 623 K (a) and 723 K (b) recorded with
             incident neutron energy of 4840 cm-1. The red line shows a Gaussian
             curve fitted for quantification of CH moieties.

Figure 5:    INS spectra for post reaction samples of reduced Fe-based catalyst
             exposed to a H2/CO mixture at 623 K (a) and 723 K (b) recorded with
             incident neutron energy of 2017 cm-1.

Figure 6:    Comparison of post-reaction temperature programmed oxidation (TPO)
             profiles for Fe2O3 exposed to a H2/CO mixture at (a) 623 K, and (b) 723

Figure 7     XRD patterns for post reaction samples of reduced Fe-based catalyst
             exposed to a H2/CO mixture at 623 K (a) and 723 K (b).

Figure 8     Comparison of literature carbide XRD pattern [ref] with post reaction
             sample of reduced Fe-based catalyst exposed to a H2/CO mixture at 723

Figure 9:    Transmission electron microscope (TEM) images for a post-reaction
             sample of Fe2O3 after exposure to a H2/CO mixture at 623 K. Scale bar
             represents 20 nm in (a) and 10 nm in (b).

Figure 10:   Transmission electron microscope (TEM) images for a post-reaction
             sample of Fe2O3 after exposure to a H2/CO mixture at 723 K. Scale bar
             represents 20 nm in (a) and 5 nm in (b).

Figure 11:   Energy filtered TEM images of reduced Fe-based catalyst exposed to
             H2/CO mixture at 723 K. (a) Zero loss image, scale bar represents 20 nm,
             (b) Iron map, scale indicated, (c) Carbon map, scale indicated.

Figure 12:   Schematic of active working catalyst (a) treated solely ion syngas and (b)
             after reductive pre-treatment in hydrogen. The line shown across the top
             right in (b) corresponds to the position for the material seen in TEM
             (Figure 11).


Figure 13:   INS spectra for Fe2O3 catalyst after reaction in H2/CO mixture at 723 K
             without reduction [Ft-1] (a) and after pre-reduction in H2 (b) [i.e. Figure
             5(b)]. Spectra recorded with incident neutron energy of 2017 cm-1.


    Reaction                  Sample treatment           (CH)          Carbon      C:H
 Temperature (K)                                                       retention
                                                        mol H g-     mol C g-
                                                           1             1
                                                             (cat).        (cat).
           623            Hydrogen pre-treatment then    245  15     2050  159    8.4:1
                              syngas exposure. (a)

           723            Hydrogen pre-treatment then   1416  88      19375       13.7:1
                              syngas exposure. (a)                      1505

           623            No hydrogen pre-treatment,    296  18       158  12     0.53:1
                           only syngas exposure. (b)

           723            No hydrogen pre-treatment,    755  47      6167  479    8.17:1
                           only syngas exposure. (b)

Table 1.         Quantification of hydrocarbonaceous features retained by the Fe catalyst
after 6 h on-stream at 623 and 723 K. Hydrogen associated with carbon and oxygen
atoms has been determined from the integrated intensity of the (CH) feature in the INS
spectra (Figure 4). The carbon retention value is obtained from TPO experiments (Figure
7). The C:H ratio is derived from the carbon TPO values and the hydrocarbon INS
features ((CH)).
(a).         This work.
(b).         [Ft-1].


To top