Professional Documents
Culture Documents
The Removal of Carbon Monoxide by Iron Oxide Nanoparticles
The Removal of Carbon Monoxide by Iron Oxide Nanoparticles
Abstract
NANOCAT® Superfine Fe2 O3 nanoparticles were evaluated both as a catalyst and as an oxidant for carbon monoxide
oxidation. It was found that the nanoparticles are much more effective as carbon monoxide catalysts than the non-nano oxide
powder. For the Fe2 O3 nanoparticles, the reaction order is first-order with respect to the partial pressure of carbon monoxide,
and zero-order with respect to the partial pressure of oxygen. The apparent activation energy was 14.5 kcal mol−1 and the
normalized reaction rate was 19 s−1 m−2 at 300 ◦ C. In the absence of oxygen, Fe2 O3 nanoparticles oxidize carbon monoxide
directly as an oxidant. The resulting reduced forms of Fe2 O3 also catalyze a disproportionation reaction for a considerable
amount of carbon monoxide. The significant amount of carbon monoxide it can remove through the catalytic oxidation,
direct oxidation, and the disproportionation reaction make it a very promising material in certain special applications, such
as removing the carbon monoxide from a burning cigarette, where the potential toxicity of other, more conventional catalysts
would be undesirable. The higher activity of Fe2 O3 nanoparticles over non-nano Fe2 O3 powders was attributed to a small
particle size (3 nm), the presence of an hydroxylated phase of iron oxide (FeOOH), as revealed by both high resolution
transmission electron microscopy (HRTEM) and a comparable study of FeOOH (goethite) powder.
© 2002 Published by Elsevier Science B.V.
Keywords: Nanoparticles; Iron oxide; Iron hydroxide; Catalysis; Carbon monoxide; Oxidation; Disproportionation reaction; Kinetics
the dual functions of iron oxide as a catalyst of carbon flow tube was then placed inside a Thermcraft furnace
monoxide oxidation in the presence of oxygen, and, controlled by a temperature programmer. The sam-
in absence of oxygen, as a direct carbon monoxide ple temperature was monitored by an Omega K-type
oxidant by losing the lattice oxygen. Furthermore, the thermocouple inserted into the dusted quartz wool.
resulting reduced forms of iron oxide could catalyze Another thermocouple was placed in the middle of
the disproportionation reaction of carbon monoxide, the furnace, outside the flow tube, to monitor and
with the less harmful carbon and carbon dioxide as record the furnace temperature. The temperature data
products. were recorded by a Labview-based program. Since
In this work, NANOCAT® was characterized both the catalytic oxidation of CO is very exothermic, the
as a catalyst and as an oxidant of carbon monoxide catalyst’s temperature could be significantly higher
oxidation. In the catalytic oxidation, the reaction or- than the heating furnace’s temperature, depending on
ders and the activation energy were measured by flow CO concentration and the flow rate. In this paper,
tube methods. The results demonstrate NANOCAT® the catalyst’s temperatures are used in all plots and
is a considerably superior catalyst than the other forms kinetic calculations.
of iron oxide catalysts in terms of activation temper- The inlet gases were controlled by a Hastings digi-
ature and conversion efficiency. In direct oxidation, tal flow meter. The gases were mixed before entering
the stoichiometry of the reaction and the contribution the flow tube. The effluent gas was analyzed either by
of the disproportionation reaction to the total carbon an NLT2000 multi-gas analyzer (non-disperse near in-
monoxide removal were analyzed. The structures of frared detector for carbon monoxide and carbon diox-
NANOCAT® and its reduced forms after the direct ox- ide, paramagnetic detector for oxygen), or a Balzer
idation were investigated by the HRTEM. The chemi- Thermal Star quadrupole mass spectrometer through
cal and structural attributes that make NANOCAT® a a sampling capillary. When the mass spectrometer
more active CO catalyst than other forms of iron oxide was used as the monitor, a 15% contribution from
are also discussed. the fragmentation of carbon dioxide (m/e = 44) to
carbon monoxide (m/e = 28) had been accounted for.
The NANOCAT® was purchased from Mach I Inc.
2. Experimental (King of Prussia, PA). The average particle size is
3 nm, according to the manufacturer. The sample was
Experiments were carried out by using a quartz flow used without further treatment. The FeOOH (goethite)
tube reactor (length 50 cm, i.d. 0.9 cm). A schematic was purchased from Aldrich, the average particle size
diagram is shown in Fig. 1. A piece of quartz wool after dry grinding was 4.5 m. The ␣-Fe2 O3 was also
dusted with known amount of NANOCAT® was from Aldrich with an average particle size of 5 m.
placed in the middle of the flow tube, sandwiched by It was used without further treatment. The carbon
the other two clean pieces of quartz wool. The quartz monoxide (4%), and oxygen (21%) gases, all balanced
Fig. 3. Comparison of the carbon monoxide conversion efficiencies between NANOCAT® and ␣-Fe2 O3 powder. Flow rate: 1000 ml min−1 ;
CO: 3.44 vol.%; O2 : 20.6 vol.%; balanced with helium. Heating rate: 12 ◦ C min−1 ; mass of catalyst: 50 mg.
still higher than 90% in the second and the third run. NANOCAT® is first-order with respect to carbon
The upward shifting of the conversion curve is related monoxide.
to the decomposition of the FeOOH component in the The reaction order of oxygen was measured in a
NANOCAT® . More detailed discussion in this aspect similar fashion and also shown in Fig. 4. Care was
will be presented in a later section. taken to make sure that oxygen concentration was not
To further test the thermal stability of the lower than 0.5 of the carbon monoxide inlet concen-
NANOCAT® , another run was performed by increas- tration, based on the stoichiometry of the reaction.
ing and then holding the catalyst’s temperature at The purpose was to prevent any direct oxidation of
340 ◦ C with the same gas inlet as in Fig. 3. The car- carbon monoxide by NANOCAT® because of insuffi-
bon monoxide conversion was stable at above 90% cient oxygen. The temperature was fixed at 244 ◦ C but
for more than four hours. The furnace’s temperature the total flow rate was increased to 1000 ml min−1 in
was 230 ◦ C during the run. order to maintain the carbon monoxide conversion at
about 50%. The increase of the oxygen concentration
3.2.1. Reaction order of carbon monoxide and oxygen had very little effect on the carbon dioxide production
The reaction order of carbon monoxide was mea- in the effluent gas. Therefore, it can be concluded that
sured isothermally at 244 ◦ C. At this temperature, the the reaction order of oxygen is approximately zero.
carbon monoxide to carbon dioxide conversion rate Walker et al. [4] also observed first-order reaction for
is about 50%. With a total flow rate of 400 ml min−1 , the overall catalytic carbon monoxide reaction on 100
the inlet oxygen was kept constant at 11% while mesh Fe2 O3 /TiO2 .
the inlet carbon monoxide concentration was varied
from 0.5 to 2.1%. The corresponding carbon diox- 3.2.2. Apparent activation energy and
ide concentration in the outlet was recorded and the pre-exponential factor of catalytic reaction
data are shown in Fig. 4. The linear relationship be- Since the reaction is first-order for carbon monoxide
tween the effluent carbon dioxide concentration and and zero-order for oxygen, the overall reaction is a
the inlet carbon monoxide concentration indicated first-order reaction. In the plug-flow tubular reactor,
that the catalytic oxidation of carbon monoxide on the reaction rate constant, k (s−1 ), can be expressed
P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12 5
Fig. 4. Measurement of the reaction order for carbon monoxide and oxygen. For reaction order of carbon monoxide (open circle). Flow rate:
400 ml min−1 ; CO: 0.5–2.1 vol.%; O2 : 11 vol.%; balanced with helium. Temperature: 244 ◦ C; mass of NANOCAT® : 50 mg. For reaction
order of oxygen (solid circle). Flow rate: 400 ml min−1 ; CO: 0.79 vol.%; O2 : 0.4–1.6 vol.%; balanced with helium. Temperature: 244 ◦ C;
mass of NANOCAT® : 50 mg.
Fig. 5. Measurement of the apparent activation energy Ea and the pre-exponential factor A. Flow rate: 300 ml min−1 ; CO: 1.1 vol.%; O2 :
14 vol.%; balanced with helium. Heating rate: 12 ◦ C min−1 ; mass of NANOCAT® : 50 mg.
cannot explained why, for the first run, the onset tem- ing of the conversion curve to the higher temperature
perature of the NANOCAT® is at least 100 ◦ C lower in the second run, as shown in Fig. 3, indicates that
than the onset temperatures of iron oxide powder, sup- FeOOH component present in NANOCAT® could
ported or unsupported. The lower onset temperature is be the active component responsible for the lower
also reflected in the lower activation energy. The shift- onset temperature. FeOOH loses water to form Fe2 O3
Table 1
Summary of the activation energies and pre-exponential factors
Flow rate (ml min−1 ) Carbon monoxide (%) Oxygen (%) A0 (s−1 ) Ea (kcal mol−1 )
Fig. 6. Direct oxidation of carbon monoxide by NANOCAT® (solid line), FeOOH (broken line), and ␣-Fe2 O3 (dashed line). Flow rate:
200 ml min−1 ; CO: 4 vol.% (vol.); balanced with helium. Heating rate: 6 ◦ C min−1 ; NANOCAT® : 50 mg; FeOOH: 100 mg; ␣-Fe2 O3 : 100 mg.
during the first run. It is no longer available for the onset temperatures for NANOCAT® and ␣-Fe2 O3 al-
second run hence the shifting of the conversion curves. most mimic those in Fig. 3. The reason that FeOOH
Generally, the catalytic oxidation of carbon monoxide is easier than ␣-Fe2 O3 to lose oxygen is probably be-
by metal oxides can be broken down into two steps: cause FeOOH has a slightly longer Fe-O bondlength
and thus a slightly weaker Fe–O bond. The Fe–O
MO + CO = M + CO2 (5)
bondlength in FeOOH is 2.103 Å [9]. In ␣-Fe2 O3 , it
M + 21 O2 = MO (6) is 2.08 Å [10]. Therefore, it is reasonable to conclude
that the FeOOH component present in NANOCAT®
where M is the metal. provides the catalytic activity at low temperatures and
The first step of the catalytic reaction is for MO the small particle size provides the high conversion
to lose one oxygen atom to carbon monoxide to form efficiency.
carbon dioxide. The tendency of MO to lose one oxy-
gen atom to carbon monoxide should be an impor-
tant factor in determining how good a catalyst MO 3.3. NANOCAT® as direct carbon
is. The MO reaction with carbon monoxide in the ab- monoxide oxidant
sence of oxygen can be checked experimentally. Fig. 6
shows the temperature scans of the Eq. (5) reaction In the absence of oxygen, Fe2 O3 can also behave as
for NANOCAT® , ␣-Fe2 O3 , and a pure FeOOH pow- a reagent to oxidize carbon monoxide to carbon diox-
der from Aldrich, which has a particle size of 4.5 m ide with sequential reduction of the Fe2 O3 to produce
after dry grinding. In Fig. 8, carbon dioxide produc- reduced phases such as Fe3 O4 , FeO and Fe [11]. This
tion is used as the indicator of the reaction progress. property is important in certain potential applications,
It can be seen that despite the vastly different particle such as a burning cigarette, where the amount of oxy-
sizes (3 nm versus 4.5 m) the traces of NANOCAT® gen present is insufficient to oxidize all the carbon
and FeOOH are remarkably similar and the onset tem- monoxide present. In such instances, the Fe2 O3 can be
peratures are very close. On the other hand, the onset used as a catalyst first, then used again as an oxidant,
temperature for ␣-Fe2 O3 is significantly higher and and then destroyed. In this way, a maximum amount
the general curve shape is quite different as well. The of carbon monoxide can be converted to carbon
8 P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12
dioxide with only a minimal amount of Fe2 O3 added The net equation is
to the system.
Fe2 O3 + 3CO = 2Fe + 3CO2 (10)
The reaction of NANOCAT® with carbon monox-
ide in the absence of oxygen is quite complicated. The proportions of carbon monoxide consumed in
First, the Fe2 O3 will be reduced stepwise to Fe, as the these three steps described by Eqs. (7)–(9) are 1:2:6,
temperature increases: respectively. Also, the freshly formed Fe can catalyze
3Fe2 O3 + CO = 2Fe3 O4 + CO2 (7) the disproportionation reaction of carbon monoxide
[12–14]. The reaction produces carbon dioxide and a
2Fe3 O4 + 2CO = 6FeO + 2CO2 (8) carbon deposit:
Fig. 7. (A) Direct oxidation of carbon monoxide by NANOCAT® : the concentration of effluent carbon monoxide and carbon dioxide.
Flow rate: 200 ml min−1 ; CO: 4 vol.%; balanced with helium. Heating rate: 6 ◦ C min−1 ; mass of NANOCAT® : 59 mg. CO2 (dashed line),
CO (solid line). (B) Different plot of same data from (A) with CO consumption and CO2 production overlapped. CO2 (dashed line); CO
(solid line); CO − CO2 (broken line).
P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12 9
The carbon can also react with the Fe to form iron car- consumed. However, for the disproportionation re-
bides, such as Fe3 C or Fe3 C7 , and thus deactivate the action of carbon monoxide catalyzed by the reduced
Fe catalyst. Once the Fe is completely transformed to forms of iron oxide as illustrated in Eq. (11), the
iron carbide, or its surface is completely covered by carbon monoxide consumed would be more than the
iron carbide or carbon deposit, then the disproportion- carbon dioxide produced, and there should be carbon
ation reaction of carbon monoxide stops. deposited on the surface.
For the direct oxidation experiment, only 4% car- To confirm the existence of the carbon deposit, the
bon monoxide balanced by helium was used in the reactor was first cooled down from 800 ◦ C to room
gas inlet. The carbon monoxide and carbon dioxide temperature under the inert atmosphere of helium gas.
concentrations were monitored in the effluent gas Then the inlet gas was switched to 5% oxygen in he-
while the temperature was increased linearly from lium and the reactor temperature was again linearly
ambient to 800 ◦ C at the rate of 12 ◦ C min−1 . The ramped up to 800 ◦ C. The net loss of oxygen, the
production of carbon dioxide and the depletion of car- production of the carbon dioxide, and their differ-
bon monoxide were almost mirror images, as shown ence are shown in Fig. 8. The corresponding reactions
in Fig. 7(A). However, a more careful comparison in are
Fig. 7(B) shows that the depletion of carbon monoxide
C + O2 = CO2 (12)
and the production of carbon dioxide are not exactly
overlapped. There is more carbon monoxide depleted 4Fe + 3O2 = 2Fe2 O3 (13)
than carbon dioxide produced. The difference be-
tween the carbon monoxide depletion and the carbon and/or
dioxide production, as indicated by the dashed line in 4Fe3 C + 13O2 = 6Fe2 O3 + 4CO2 (14)
Fig. 7(B), starts to appear at 300 ◦ C and extends all
the way to 800 ◦ C. All of the carbon monoxide reac- The production of carbon dioxide confirms the exis-
tions with different forms of iron oxide, as illustrated tence of the carbon in the sample. The difference be-
by Eqs. (7)–(9), would produce the same amount of tween the net loss of oxygen and the production of
carbon dioxide as the amount of carbon monoxide carbon dioxide is the oxygen used to oxidize the Fe
Fig. 8. Oxidation of the reaction residues Fe, carbon deposits and carbide to Fe2 O3 and carbon dioxide. Flow rate: 180 ml min−1 ; O2 :
5 vol.%; balanced with helium. Heating rate: 6 ◦ C min−1 ; CO2 production (dashed line); O2 consumption (solid line); O2 − CO2 (broken
line).
10 P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12
Fig. 9. HRTEM of Fe2 O3 nanoparticle heated to 800 ◦ C in the presence of 4% carbon monoxide showing graphite surrounding iron carbide.
Table 2
The stoichiometry of the carbon monoxide + Fe2 O3 reaction (unit: mmol)
Species Measured Theoretical Description
Carbon monoxide + Fe2 O3 reaction
Iron oxide 0.344 59.0 mg of NANOCAT® with 7 wt.% of water, as
measured by TG
Carbon monoxidetotal 2.075 Total carbon monoxide consumption
Carbon dioxidetotal 1.551 Total carbon dioxide production
C = COtotal − carbon dioxidetotal 0.524 Total carbon in the residue
Carbon dioxidedisprop. = C 0.524 Carbon dioxide produced from the disproportional reaction
according to Eq. (9)
Carbon dioxideFe2 O3 = carbon dioxidetotal − 1.027 1.032 Carbon dioxide produced according to Eqs. (5)–(7)
carbon dioxidedisprop.
Oxygen + Fe, C reaction
Oxygentotal 1.060 Total oxygen consumption in the oxidation reaction
Carbon dioxide 0.564 Carbon dioxide production from the oxidation of
carbon deposit
C = carbon dioxide 0.564 Total carbon content in the residues
OxygenFe2 O3 = oxygentotal − C 0.496 0.516 The oxygen used to oxidize Fe to Fe2 O3
P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12 11
back to Fe2 O3 . This was also supported by the color and iron carbides according to Eq. (11). This is in
change of the sample from black to bright red. reasonable agreement with the 0.564 mmol deter-
As a further check, a sample heated to 800 ◦ C in the mined by the oxidation of the reaction residue. The
presence of carbon monoxide and He was quenched carbon dioxide produced from the reduction of Fe2 O3
and examined with high-resolution TEM with energy (carbon dioxideFe2 O3 ), is the difference between the
dispersive spectroscopy. Essentially, two phases were carbon dioxidetotal and the carbon dioxide produced
observed, an iron-rich phase and carbon. These phases from the carbon monoxide disproportionation reaction
are shown in Fig. 9. The iron-rich phase formed a nu- (carbon dioxidedisprop. ). The 1.027 mmol of carbon
cleus for the precipitation of the carbon. The lattice dioxideFe2 O3 agrees very well with the 1.032 mmol
fringes of the carbon have a 3.4 Å spacing, verifying calculated from the initial amount of Fe2 O3 , accord-
that the carbon is graphite. The iron-rich core produced ing to Eq. (10). In the O2 + (Fe, Fe3 C, and C) oxida-
EDS spectra indicating only the presence of iron and tion reactions, the oxygen spent on the oxidation of
carbon. The lattice fringes in Fig. 9 could be indexed the Fe species to Fe2 O3 also agrees very well with the
as the metastable iron carbide Fe7 C3 with Pnma sym- oxygen needed as calculated from the Eqs. (12)–(14).
metry. Lattices of other crystals match Fe3 C. A hard It is interesting to observe the extent to which the
mass was found on the bottom of the reactor tube. Ex- disproportionation reaction of carbon monoxide pro-
amination of this material in the TEM indicated that ceeded. The amount of total carbon monoxide con-
it consisted of a mixture of iron carbide, graphite, and sumed (COtotal ) of 2.075 mmol is more than double
essentially pure iron. that of the carbon monoxide consumed (1.027 mmol)
The carbon monoxide disproportionation reaction is by Eq. (10). Regarding the extra carbon monoxide
also effective in carbon monoxide removal. A detailed consumption, 50% became carbon deposits and car-
stoichiometric account of the reduction and oxidation bides, and the other 50% became carbon dioxide.
reactions is given in Table 2. In the CO + Fe2 O3 reac- Therefore, the contribution of the carbon monox-
tion, the difference between the total carbon monoxide ide disproportionation reaction to the total carbon
consumption (COtotal ) and the total carbon dioxide monoxide removal is significant.
production (carbon dioxidetotal ) of 0.524 mmol can The expected stepwise reduction of NANOCAT®
be attributed to the formation of the carbon deposits is illustrated in Fig. 10, which is a plot of the
Fig. 10. Direct oxidation of carbon monoxide by NANOCAT® : the stepwise reduction of Fe2 O3 (see text for explanation).
12 P. Li et al. / Applied Catalysis B: Environmental 1326 (2002) 1–12