Perreault 2016
Perreault 2016
Perreault 2016
PII: S1385-8947(16)30269-8
DOI: http://dx.doi.org/10.1016/j.cej.2016.03.039
Reference: CEJ 14890
Please cite this article as: P. Perreault, Gregory-S. Patience, PyrolusiteCO reduction kinetics, Chemical
Engineering Journal (2016), doi: http://dx.doi.org/10.1016/j.cej.2016.03.039
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. 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.
PyrolusiteCO reduction kinetics
Abstract
Correspondingauthor
Email address: patrice.perreault@polymtl.ca (Patrice Perreault)
2
The prevalence of any specific phases depends on the oxygen partial pressure.
Kononov et al. [25] measured Mn2 O3 and Mn3 O4 when heating pyrolusite un-
der air at 1000 C, but only MnO2 and Mn3 O4 under argon. This last result is
surprising when considering that pyrolusite (MnO2 ) decomposes at 535 C into
a mixture of bixbyite, hausmannite, manganosite and molecular oxygen [26].
Pyrolusite is an ideal oxygen carrier. Indeed, various patents involving pyro-
lusite were filed [27, 28, 29], all of which were the result of the work initiated at
Polytechnique Montreal as part of a research contract with Total. Manganese
oxides are active catalysts for the combustion of organic compounds, including
methane, carbon monoxide, C3 s and VOCs [30, 31]. South African and Aus-
tralian manganese ore loss up to 20 % of their initial mass when reduced with
hydrogen at 1000 C: Pyrolusite forms MnO. The iron oxide, naturally present
with manganese minerals, converts to metallic iron [25]. Reducing the surface
of a manganese oxide increases its basicity, and decreases its surface site density
(less oxygen sites) [32]. In addition, the reduction creates new mineralogical
phases (e.g. reduction from MnO2 to Mn2 O3 ), which in turns swells the lat-
tice, by further lowering the number of oxide sites (increased spacing) [32]. The
increasing extent of reduction also decreases the Mn-O bond strength, due to
the gradual replacement of Mn4+ with the larger Mn3+ cation [32]. The sole
adsorption of CO also modifies the Mn-O bond energy [33]. The reduction of
manganese oxide can thus be regarded as autocatalytic. Manganese oxides in
the presence of excess oxygen completely oxidize propane. Adding alumina rich
manganese oxide is detrimental: propane now forms CO and propene at the
same conditions [30]. In addition, even in the absence of molecular gas-phase
oxygen, pure unsupported manganese oxide reduces methane without coking
[31], a highly desired property for chemical looping applications (no carbon
dioxide in the air reactor).
Aside from its redox properties, pyrolusite can also split CO2 into CO at
high temperature due to the naturally occurring alkali promoters present with
the manganese oxide phase. Imbihl and Demuth [34] postulated that CO2 is
activated into CO
2 over potassium promoters, then dissociates into CO and
3
atomic surface oxygen, which rapidly integrates into the lattice. Finocchio and
Busca [30] observed a strong IR peak at 2338 cm1 when heating Mn3 O4 in air
at 400 C, followed by vacuum. They associated this peak with trapped CO2
(in addition to the observed carbonate bands).
In this article, we characterize the kinetics of the reduction of pyrolusite by
CO, as well as how much oxygen transfers from the solid to the gas when both
CO and H2 are fed over the catalyst. We derive elementary steps based on the
gas profiles from syngas reduction experiments, as well as with a combination of
methane and carbon monoxide from 810 C to 960 C. We propose a method to
estimate two surface reaction characteristics: surface site density and amount of
adsorbed CO. Both parameters are obtained from the transient response with
a step change in the inlet concentration.
We sourced pyrolusite from Centra Africa (via Total SA France) and sieved it
to recover the 90 m to 106 m diameter particles. Beside MnO2 , pyrolusite con-
tains mostly iron oxides, alumina, silica and potassium, as measured by X-ray
fluorescence performed at the UQAMs Research Center NanoQAM (Table 1).
The bulk density was 1.78 g cm3 (Scott volumeter) and the particle density
was 4.28 g cm3 (Micromeritics Accupyc 1340 gas pycnometer). The fresh py-
rolusite, initially amorphous following XRD measurements, has a 7.3 m2 g1
specific surface area (Quantachrome Autosorb-1 gas sorption analyzer).
We loaded 0.06 g to 0.1 g to a 7 mmID quartz reactor with a quartz frit gas
distributor. An electrical furnace heated the reactor (800 C to 950 C) under
an upward steady flow of gas consisting of 21 vol. % O2 in Ar. Upon reaching
the desired temperature, various reductants (CO, H2 and CH4 in Ar) and com-
binations reduced the sample and oxygen in Ar reoxidized it. We sequentially
reduced and oxidized the sample with H2 and O2 15 times at 800 C before
using the data for kinetic modeling, followed by kinetic runs with H2 , CO, CH4 ,
combined reduction (CO/H2 ,CH4 /H2 , CH4 /CO), as well as CO2 and H2 O split-
4
Table 1: Elemental composition of fresh pyrolusite as measured by X-ray fluorescence. The
sample also contains Ca, P, Zn, Cu, Mg, Na, S, Cr and S levels below 0.1 %.
Mn 62.40
O 26.46
Fe 4.06
Al 3.32
Si 1.62
K 0.81
Ba 0.48
Ni 0.17
Co 0.13
Ti 0.10
5
V1 MS
Thermocouple
Vent
Quartz frit 1 mm
distributor Fixed bed
Furnace
Ar
O2
CO, CO+H2, etc.
Vent
Ar
Figure 1: Micro-fixed bed setup to sequentially reduce and oxidize calcined pyrolusite. Argon
purges the reactor after each oxidation and reduction step.
6
We estimated the dispersion coefficient of the inlet lines from the 4-way valve to
the reactor by introducing a step change in the inlet gas composition from Ar to
CO + He and modeled the response of the MS signal at the outlet. The fitted
axial dispersion coefficient was 1.9 103 m2 s1 , i.e. equivalent to 36 CSTRs
in series, which is essentially plug flow.
We calculated the experimental oxygen transfer capacity by considering the
total CO2 emitted during the reduction by CO (mole balance on transferred
oxygen, i.e. Eq. 1).
P V tredf
Z
RO,exp = yCO2 dt (1)
RT tred0
7
mole balance), we measured a 50 % increase at 810 C when compared to that of
pyrolusite reduced by CO. At 960 C, the oxygen transfer increases by as much
as 86 % when reducing with 20 vol. % CO, and doubles when reducing with
25 vol. % CO. We see an almost linear trend between the 5 and 10 vol. %
reduction with CO only, and the higher CO concentrations with combined CO
and H2 . Assuming that the reduction by CO is limited by steric overcrowding
and so limited by oxygen diffusion from the bulk to the surface fueling the sur-
face reaction, reduction by hydrogen enhances the oxygen diffusion. Diffusion
jumps are mediated by lattice defects such as vacancies, following the random
walk theory [37]. Hydrogen facilitates diffusion by increasing the number of
vacancies, but not only in absolute number, but by increasing it further into
the solid crystal structure. Hydrogen permeates through various materials. We
suppose that hydrogen removes oxygen both from the surface (O) and the crys-
tal lattice (OO ), i.e. via Eq. 2 and 3. The oxygen transfer capacity approaches
an asymptote with increasing CO concentration and temperature in combined
CO/H2 reductions, as was the case of reduction by CO.
H2 + O
H2 O + (2)
H2 + O O
H2 O + V O (3)
8
Figure 2: Comparison of the oxygen transfer due to CO when reducing pyrolusite with a
stream of CO in Ar, and a combination of CO and H2 . The reported oxygen transfer is
calculated from a mole balance on the CO2 measured (coming from the CO only).
9
Figure 3: Reduction of pyrolusite with 25 % carbon monoxide at 957 C during alternating
CO2 splitting and reduction by CO.
10
Figure 4: Splitting of a 25 % CO2 stream with reduced pyrolusite at 957 C.
then dosed with CO2 (still MnO), and finally reoxidized by molecular oxygen
(Mn3 O4 ) did not match the characteristic peaks of manganese carbonates. Fre-
und and Roberts [38] reviewed the interactions of CO2 and various metal oxides
surfaces, and concluded on the formation of carbonate only at low temperature.
We recently showed that the CO2 splitting reaction was mediated by potassium
promoters, activating the inert molecule into a CO
2 activated state, which fur-
ther dissociates into CO and atomic oxygen migrating in the bulk of the oxide
[39].
11
values for most of the reduction. However, after 90 s of reduction, the CO2 value
drops below 2.9 % (at 810 C), i.e. lower than the apparent equilibrium CO2
values for reduction by CO. At 960 C, both the surface reactions and diffusion
are faster (Fig. 6). In this case, the initial CO2 concentration is 2.4 that for
CO only, and the CO2 mole fraction stays at higher value compared to when
reducing with pure CO (apparent equilibrium at 1.2 %). When comparing the
gas profiles at 817 C, and 960 C, we observe a greater global extent of reduction
(due to CO and H2 ) and the MS records a second peak for CO2 . From 817 C to
960 C, the initial /ceCO2 peak slightly decrease from 15 % to 13 %, but the H2
conversion is higher (higher hydrogen consumption highlighted by the gentler
slope for the rising hydrogen concentration). There is also a change of slope for
all compounds at 65 s. We assign the second CO2 peak to diffusion, i.e. a rapid
renewal of the surface oxygen. After CO reduced the surface of the oxide, it is
refueled. In that case, the diffusion rate is higher than the surface reaction rate,
at least for 15 s. Thereafter, the diffusion rate slows: The remaining oxygen lies
deeper in the oxide crystal structure. The change in the slope of CO, H2 , and
CO2 may be due to some critical depth upon which diffusion further slows down.
This diffusion depth might coincide with a localized change in stoichiometry
(appearance of the sufficiently large product grain that induces a discontinuity in
the diffusivity). Following this assumption, the faster initial diffusion rate could
only be attributed to deviations from stoichiometry, or equivalently, oxygen
diffusing from a crystal lattice of distinct nature (lattice sites of different oxide
phases, c.f. Eq. 4 and Eq. 5).
OO,site1 +
VO,site1 + O (4)
OO,site2 +
VO,site2 + O (5)
12
Figure 5: Reduction of pyrolusite with 25 % carbon monoxide and 20 % hydrogen at 817 C.
ide to produce carbon dioxide (Eq. 6), but methane concentration increases
to its inlet value unchanged. Carbon monoxide and methane molecules may
compete for the same oxidized sites, with methane being disadvantaged due
to steric factors. For the first 5 s of reduction, the methane concentration re-
mains at 15 %, then decreases while the CO gradually increases. The decrease of
the methane concentration coincides with an increase in hydrogen, suggesting
coking by thermal decomposition of methane (Eq. 7). Some of the hydrogen
generated reduces the solid producing water (not shown and not quantified due
to its distinct hydrodynamic behavior in the analysis section). Methane cokes
at a constant rate (almost constant methane concentration after the first 40 s
of the reduction). We confirmed coking of the oxide during the reoxidation of
the oxide with molecular oxygen (emission of CO and CO2 , then O2 breaks
through). The mass spectrometer detected CO and CO2 , with the total carbon
corresponding to 0.4 % of the total carbon fed to the reactor. In addition, when
13
Figure 6: Reduction of pyrolusite with 25 % carbon monoxide and 20 % hydrogen at 960 C.
plotting the total carbon concentration (sum of CO, CO2 and CH4 ) against
reduction time, the mole balance on carbon diverges with time-on-stream.
CO + O
CO2 + (6)
CH4 +
C +2H2 (7)
14
Figure 7: Reduction of pyrolusite with 25 % carbon monoxide and 15 % methane at 814 C.
would also explain the sharp increase in CO concentration (Eq. 8), as well as
the slight increase in hydrogen concentration (change in slope at around 30 s for
both the hydrogen and carbon monoxide). The location of the pit between the
two CO2 peaks probably indicates the point at which methane converts to CO,
but also the initiation of the thermal decomposition of methane. The change in
the slope for hydrogen concentration would then be attributed to the addition
of the hydrogen resulting from the partial combustion of methane to that due
to methane thermal decomposition via coking.
CH4 + O
CO +2H2 (8)
15
Figure 8: Reduction of pyrolusite with 25 % carbon monoxide and 15 % methane at 952 C.
overlaps with the first one. This second peak is now higher and wider than the
first one. Methane also partially converts to CO and H2 at higher rate than
at 814 C (second CO2 peak due to the CO that further reduces the solid),
in agreement with thermodynamic equilibrium composition calculations. At
952 C, methane cokes 5 times more than it did at 814 C. However, thermody-
namics predicts that coking decreases with temperature.
When switching back to argon after the reduction, carbon monoxide does
not fall back to zero instantaneously, or as fast as the other gases, suggesting the
presence of adsorbed CO species. Since the direct CO combustion is fast, these
species can be attributed to the partial methane conversion reaction (Eq. 8),
and the desorption that follows (Eq. 9).
CO
CO + (9)
16
3.5. Kinetic modeling of calcined pyrolusite reduced with carbon monoxide
We assumed a fixed bed with dispersion (Eq. 12) based on the results of
an inert tracer fed via a step change in the inlet concentration. The axial
dispersion coefficient was calculated by curve fitting of the tracer experiment.
We tested helium co-fed with CO as a representative hydrodynamic tracer.
However, He passes through the whole reactor, feed lines and analytical sec-
tion at more rapidly (6 s earlier at 808 C), and with 10 times less dispersion
than CO. Instead, and by ensuring that coking was negligible during the reduc-
tion with CO, we selected the total carbon fed to the reactor (sum of CO and
CO2 ) as the tracer. The dispersion coefficient averaged 1 102 m2 s1 , versus
1 103 m2 s1 for helium. The numerical value for the dispersion was the only
parameter when modeling the response of the system to a step change in the
inlet concentration.
17
stability, the time was set at 0.003 s (Equation 13).
t
1 (13)
z/uz
To ensure that the kinetic data was obtained in kinetic controlled regime,
we evaluated the Mears (Equation 14) and Weisz-Prater (Equation 15) criteria
for external and internal mass transfer limitations, respectively. Mass transfer
dominates when the reaction is fast, i.e. when its rate surpasses the rate at
which the reactant molecules travel from the bulk to the surface of the solid
(external mass transfer) or the transport rates of gas phase species from the
surface to the reacting sites located in the pores of the solid.
We considered the largest particle diameter (212 m) at the highest reaction
rate (0.12 mol m3 s1 ). At this condition, according to the Mears criterion
(Eq. 14) kinetics control because the ratio of the kinetic to mass transfer rate
is 1.3 105 3276832768 , which is several orders of magnitude lower than the
threshold criterion.
To calculate the Weisz-Prater number, we assumed that the concentration at
the surface of the solid was identical to that of the bulk, and assumed that the
internal diffusion was essentially dominated by the Knudsen diffusion. The mean
free path of the CO molecules at 952 C is 367 nm, i.e. greater than the pore
diameter (around 4 nm obtained with the BJH method). The resulting Weisz-
Prater number was 4.0 103 3276832768 , which is much less than threshold
value of 0.3. Thus the reaction was controlled by kinetics not external nor
internal mass transfer.
rA p Rn
0.15 (14)
kc CAb
rA R2
0.3 (15)
De CAs
Modeling the kinetics of the reduction by CO with an irreversible CO reac-
tion (Eq. 10), along with the reoxidation of the surface from the bulk (Eq. 11)
accounts for more than 95 % of the variance in the data (Fig. 9). However, the
18
value of the kinetic constant for the direct CO oxidation reaction decreased with
temperature (negative activation energy), as expected: The oxygen transfer ca-
pacity reached an asymptote with increasing temperature. The elementary step
describing the reoxidation of the surface sites with oxygen from the lattice has
a positive 55 kJ mol1 activation energy.
Helfferich [40] established the conditions for apparent negative activation
energy: A minimum of one reversible elementary step for which the reverse
reaction rate increases more rapidly than the forward one (or non-Arrhenius
reactions, e.g. involving radicals or decomposition of a solid). Possible equi-
librium elementary steps include adsorption and desorption of CO and/or CO2
molecules. The negative apparent activation energy would then be explained by
the decreasing heat of adsorption of CO with increasing temperature. Carbon
monoxide reacts to carbon dioxide at the surface of the oxide (conversion of an
oxidized site to a reduced one, where CO2 is adsorbed). However, since the mass
spectrometer detects CO2 as soon as CO is fed to the reactor, we considered
the desorption to be instantaneous. Another possibility for reversibility is the
reverse reduction reaction (Eq. 16). Indeed, the reduced form of this calcined
mineral reduces CO2 to CO (i.e. the reverse reaction).
19
Figure 9: Kinetic modeling of the reduction of calcined pyrolusite at 808 C with 10 % CO.
The model (solid lines) accounts for 96 % of the variation in the data (red circles and green
lower triangles for CO2 and CO, respectively).
20
Figure 10: Comparison of the CO washout curves when switching to inert purge in the 814 C
to 951 C temperature range. The helium washout curve established the baseline from which
we calculate the amount of adsorbed CO species desorbing as soon as CO is no longer being
fed to the reactor.
21
helium at the same temperature. We obtained the number of moles of adsorbed
molecules per unit surface area (NCO ) by multiplying the area difference with
the total number of moles normalized over the surface area of the bed (BET
specific surface times the mass of oxide, i.e. Eq. 17). The numerical values
ranged from 9 mol m2 to 61 mol m2 (Fig 11), in agreement with the esti-
mated surface site density of 41 mol m2 . During reduction, the outermost layer
of the oxide converts to manganosite (MnO), corresponding to a maximum sur-
face CO concentration of 82 mol m2 , i.e. twice the value for the oxygen surface
site density (one unchanged manganese site and one oxygen-depleted site). The
calculated CO surface sites densities are consistent with this estimate.
Z twash
PV
NCO = (yCO yHe )dt (17)
RT SBET m tswitch
Ea
S = f () exp (18)
RT
P
F = (19)
(2mkT )1/2
22
Figure 11: Maximum adsorbed CO desorbing during the CO washout at the end of the
reduction cycle.
23
all conditions indicating that a CO adsorption-desorption equilibrium has been
established. The total amount of adsorbed CO almost linearly increases with
temperature, and reaches a constant value for temperatures higher or equal to
910 C. The adsorbed CO peaked at 7 mol m2 when reduced with 15 vol. %
CO, 20 mol m2 at 20 vol. % and 55 mol m2 at 25 vol. %.
We estimated the surface site density by calculating the area under the
curve of the initial response (initial CO2 peak) of the transient reduction of
pyrolusite with CO. This estimation relies on two assumptions: i) the surface
reaction (Eq. 6) is faster than the reaction of oxygen replenishment from the bulk
(Eq. 4) which is rate limiting; and ii) the solid is not reduced via a shrinking
core scheme where the gaseous reactants and products diffuse into a reacted
outermost porous solid matrix. Solid-state diffusion governs the reduction of
some metal oxides, e.g. wustite reduced by hydrogen [44, 45, 46] and ilmenite
reduced by carbon monoxide [47]. In addition, the surface reaction rate constant
is 2 orders of magnitudes greater than surface-to-bulk solid-state diffusivities
during oxygen uptake of perovskites [48]. The reduction of wustite by hydrogen
leads to a nonporous outer shell, where even hydrogen cannot permeate. This
phenomenon excludes the shrinking core as the possible reaction mechanism.
However, where solid-state reactions are rate limiting, the diffusion species is an
iron cation from the outer surface towards the center of the lattice [44, 45, 46].
To test these assumptions, we used the kinetic model for the reduction of
ilmenite with CO [47]. We characterized the reduction of ilmenite with carbon
monoxide, with a 2 steps mechanism: i) reaction of CO with surface oxygen
and, ii) replenishment of the surface reduced sites by oxygen from the bulk. We
assessed the response of ilmenite with a step change in CO due to the surface
reaction only, by extinguishing the reaction as soon as the surface oxidized sites
were depleted (Fig. 12). In this case, the reaction is very fast and limited by
external mass transfer, which explains the tailing of the CO2 concentration.
Eliminating this resistance lead to a pure step change with the dispersion. The
24
Figure 12: Kinetic modeling of the reduction of preoxidized ilmenite with and without the
contribution of the bulk. The solid lines represent the model with the contribution of the
oxygen from the bulk of the lattice. The dashed lines represent the model that considers the
surface reaction only. The tailing of the modeled gas profiles is due to the external mass
transfer resistance for such fast reactions.
main result from this modeling experiment is that the surface reaction for the
whole bed corresponds to the increasing CO2 concentration, up to where it
reaches its maximum value. This peak can thus be associated with the total
available surface in the reactor bed. We can calculate the total number of moles
of transferred oxygen during this initial peak to estimate the surface site density
(Eq. 20). For the reduction of ilmenite, the ratio of the surface and reoxidation
reaction rates is greater than 100 for the duration of the reduction.
Z
PV
Ns = yCO2 ,peak dt (20)
RT SBET mcat
Further refining the analysis by neglecting external mass transfer limitations,
we confirmed that the surface concentration of oxygen atoms can be estimated
by considering the initial peak of the product formed in this simple equimolar
reaction. For example, when reducing calcined pyrolusite with carbon monox-
25
Figure 13: Extraction of the initial peak from the entire carbon dioxide response during
reduction of pyrolusite with carbon monoxide. We estimate the surface site density from the
area under the curve of the initial CO2 peak.
ide, the CO2 reaches its maximum value after 21 s, then gradually decreases.
Knowing that the surface reaction is fast, and that the global rate of reaction
slows when diffusion is predominant, we can assume that this shift between
surface and diffusion regimes corresponds to tmax . Then, we can artificially
extinguish the surface reaction from that point and plot the response of the sys-
tem by using the axial dispersion coefficient calculated from the hydrodynamic
response (Fig. 13). This method is equivalent to a simulated CO titration.
The height of the initial peak is a function of the reductant concentration:
The higher the CO fed to the reactor, the higher the CO2 produced. We com-
pared the CO2 gas profiles with increasing CO concentration. We reduced cal-
cined pyrolusite with carbon monoxide for 2 min at concentrations of 5, 10, 20
and 25 vol. %. The initial CO2 profile reached a maximum value, highlighting
the surface reaction saturation (Fig. 14). Reducing calcined pyrolusite with a
26
Figure 14: Carbon dioxide production when reducing pyrolusite with 5, 10, 20 and 25 %
carbon monoxide at 810 C.
27
Grignard reagent, in the context of their usage in ion-exchange reactions in wa-
ter. Briand et al. [50] and Wachs and Routray [51] chemisorbed methanol over
bulk oxides and monitored the IR spectra and mass variations in a thermogravi-
metric analyzer at 100 C. Indeed, the proposed usage for the oxides was the
selective oxidation of methanol at 300 C. Malloy and Donne [32] measured the
evolution of surface oxide and hydroxyl sites during the reduction of -MnO2
by chemical titration for a fuel cell application. They partially reduced the
manganese oxide with 2-propanol at room temperature and measured the pH
variations. The numerical values obtained by those researchers coincide, except
in the case of Wachs and Routray [51], an order of magnitude lower. Malloy
and Donne [32] measured surface site densities of 11 mol m2 to 10 mol m2
for MnO1.97 and Mn2 O3 , respectively. Tamura et al. [49] measured surface site
densities of 15 mol m2 to 23 mol m2 for bulk synthetic MnO2 , while Wachs
and Routray [51] obtained a 1.6 mol m2 value for MnO.
4. Conclusion
The recent plunge of the CO2 trading price constraints further the cost of
CLC technology, and as a consequence the cost and performance of the oxygen
carrier: mechanical stability, oxygen transfer capacity and reaction kinetics.
Pyrolusite is an alternative to ilmenite, because of its superior oxygen transfer
capacity, and excellent attrition resistance. We characterized the kinetics of the
reduction of calcined pyrolusite with carbon monoxide. The model consists of a
surface reaction between CO and an oxidized site, and the replenishment of the
reduced sites by oxygen from the lattice. The model accounts for 95 % of the
variance in the data. We estimated the surface site density based on the initial
CO2 peak during the reduction by CO (the transient response of the metal ox-
ide undergoing reduction at gradually increasing reductant concentration). The
CO2 peak reached 3 % mole fraction with increasing CO concentrations, which
relates to the saturation of the surface reaction with oxygen. We simulated the
suppression of the surface reaction using the kinetic and hydrodynamic models.
28
We estimated the surface site density as 41 mol m2 by calculating the area
under the CO2 initial peak, normalized over the total surface of the bed. Simu-
lated in situ CO titration has never been reported and usually relies on chemical
titration. This method allows to estimate the surface site density without hav-
ing to rely on a separate experimental setup and using online gas measurements.
In addition, the surface site density is obtained at the reaction conditions, and
using the reactant as the probe molecule, two necessary conditions for reliable
measurements. Our estimation is consistent with measured values from Tamura
et al. [49], Briand et al. [50], Wachs and Routray [51]. Similarly, we estimated
the adsorbed CO at the end of the 2 min reduction cycle by comparing the
washout curves between helium and carbon monoxide when switching between
reducing and inert conditions. The adsorbed CO peaked at 55 mol m2 when
reduced with 25 vol. % CO for temperature greater than 910 C, corresponding
to a 70 % covered surface.
Nomenclature
29
Ci,b Concentration of species i in the bulk phase (mol m3 )
Ci,s Surface concentration of species i (mol m3 )
De Effective diffusivity (m2 s1 )
Dz Axial dispersion coefficient (m2 s1 )
Ea Activation energy (J mol1 )
Zw Flux of incident molecules (molec m1 s2 )
Ki Equilibrium constant for reaction i Mi Molar mass of species i (g mol1 )
NCO Adsorbed CO density (mol m2 )
Nr Initial concentration of oxygen in the reactor bed (mol m3 )
Ns Surface site density (mol m2 )
O Surface atomic oxygen
P Pressure (Pa)
R Universal gas constant (8.314 Pa m3 mol1 K1 )
Rp Particle radius (m)
RO,th Theoretical oxygen transfer capacity (g g1 )
S Sticking probability
SBET Specific BET surface per unit mass (m g2 )
T Temperature (K)
Vi Species i vacancy
V Volumetric flowrate (m3 s1 )
X Solid conversion
Zw Flux of incident molecules (molec m1 s2 )
Bed porosity
i Normalized concentration of species i
b Bulk density (kg m3 )
p Particle density (kg m3 )
i Surface coverage of species i
30
References
References
[1] M. Ishida, K. Kin, H. Jin, Chemical loop combustion system power gener-
ation plant systemcomprises 1st reactor where fuel is reacted with metal
oxide, 2nd reactor oxidising reduced oxide from 1st reactor etc, 1993.
31
[10] J. C. Abanades, R. Murillo, J. R. Fernandez, G. Grasa, I. Martnez, New
CO2 capture process for hydrogen production combining Ca and Cu chem-
ical loops, Environ. Sci. Technol. 44 (2010) 69016904.
32
(CLC) and chemical looping with oxygen uncoupling (CLOU), Chem. Eng.
J. 269 (2015) 67 81.
33
[27] S. Rifflart, H. Stainton, P. Perreault, G. Patience, Combusting solid, liq-
uid/gaseous hydrocarbons by chemical looping oxidation reduction using
natural manganese mineral as active mass, comprises activating manganese
mineral by gradually increasing temperature under oxidizing atmosphere,
2012.
[28] T. Sozinho, F. Guillou, H. Stainton, Active mass, used for chemical looping
oxidation-reduction process to combust solid, liquid/gaseous hydrocarbons,
comprises pyrolusite type natural manganese ore, where mass is enriched
with nickel oxide and is present in particle form, 2013.
[33] Y. Luo, Y.-Q. Deng, W. Mao, X.-J. Yang, K. Zhu, J. Xu, Y.-F. Han, Prob-
ing the surface structure of -Mn2 O3 nanocrystals during CO oxidation by
operando raman spectroscopy, J. Phys. Chem. C 116 (2012) 2097520981.
34
[35] Z. A. Feng, M. L. Machala, W. C. Chueh, Surface electrochemistry of CO2
reduction and CO oxidation on Sm-doped CeO2x : coupling between Ce3+
and carbonate adsorbates, Phys. Chem. Chem. Phys. 17 (2015) 12273
12281.
[37] H. Mehrer, Random Walk Theory and Atomic Jump Process, in: M.
Cardona, P. Fulde, K. von Klitzing, H.J. Queisser (Eds.), Diffusion in
Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes,
Springer, 2007.
[39] P. Perreault, G.-S. Patience, Chemical looping syngas from CO2 and H2 O
over manganese oxide minerals, Can. J. Chem. Eng. In press (2016).
35
[46] J. Szekely, J. W. Evans, H. Y. Sohn, Gas-solid reactons, Academic Press,
INC., New York, U.S.A., 1976.
36