Selective oxidation of CO under conditions of catalyst surface ignition

Article abstract of DOI:10.1134/S0023158407050151

The oxidation of CO in the presence of an excess of hydrogen and to 20% CO₂ and H₂O in the starting mixture was studied in flow reactors with high and low rates of heat removal. The ignition of the catalyst surface was observed in the reactor with a low rate of heat removal; catalyst surface ignition initially occurred at a hot spot (section) of the catalyst bed and gradually propagated along the bed. Experimental data on the relaxation dynamics of residual CO concentration and temperature in a catalyst bed under conditions of small heater temperature disturbances near and at the critical temperature of ignition and the effect of oxygen concentration in the starting mixture on this process are reported. It was found exprimentally that the ignition regime in the tested cases was more favorable for the selective oxidation of CO in an excess of hydrogen than the reaction in an isothermal reactor; this was likely due to the more favorable temperature distribution over the length of the catalyst bed.

Full text of DOI:10.1134/S0023158407050151

ISSN 0023-1584, Kinetics and Catalysis, 2007, Vol. 48, No. 5, pp. 701–710. © MAIK “Nauka /Interperiodica” (Russia), 2007.
Original Russian Text © A.Ya. Rozovskii, M.A. Kipnis, E.A. Volnina, G.I. Lin, P.V. Samokhin, 2007, published in Kinetika i Kataliz, 2007, Vol. 48, No. 5, pp. 750–760.
Selective Oxidation of CO under Conditions
of Catalyst Surface Ignition
A. Ya. Rozovskii, M. A. Kipnis, E. A. Volnina, G. I. Lin, and P. V. Samokhin
Topchiev Institute of Petrochemical Synthesis, Russian Academy of Sciences, Moscow, 119991 Russia
E-mail: kipnis@ips.ac.ru
Received December 7, 2005; in final form, March 20, 2007
Abstract—The oxidation of CO in the presence of an excess of hydrogen and to 20% CO and H O in the start-
2
2
ing mixture was studied in flow reactors with high and low rates of heat removal. The ignition of the catalyst
surface was observed in the reactor with a low rate of heat removal; catalyst surface ignition initially occurred
at a “hot” spot (section) of the catalyst bed and gradually propagated along the bed. Experimental data on the
relaxation dynamics of residual CO concentration and temperature in a catalyst bed under conditions of small
heater temperature disturbances near and at the critical temperature of ignition and the effect of oxygen con-
centration in the starting mixture on this process are reported. It was found experimentally that the ignition
regime in the tested cases was more favorable for the selective oxidation of CO in an excess of hydrogen than
the reaction in an isothermal reactor; this was likely due to the more favorable temperature distribution over the
length of the catalyst bed.
DOI: 10.1134/S0023158407050151
INTRODUCTION
to the occurrence of the reaction in the kinetic and
external diffusion regions, respectively.
The catalytic oxidation of CO in the presence of
hydrogen (selective oxidation of CO) has been widely
The experimental measurements of the temperature
used as a method for hydrogen purification. This reac- dependence of CO conversion in the course of oxida-
tion has been studied in the past decade in the context tion on various catalysts lead to problems. Thus, it was
of pure hydrogen production for fuel cells (e.g., see [1, found [7–9] that the curve of CO conversion plotted as
2
] and references therein). The oxidation of CO on sup- a function of temperature on heating a reactor was
ported Pt catalysts seems best studied. However, in shifted to the region of higher temperatures, as com-
spite of numerous publications, the nature of processes pared with an analogous curve obtained on cooling
that occur in the oxidation of CO (even in the absence (it described a hysteresis loop). According to Yakerson
of ç ) remains unclear.
et al. [7, 8], it is evident that the observed phenomena
are inconsistent with the hypothesis that multiple
steady states are responsible for hysteresis. Note that,
on the contrary, it is our opinion that the classical inter-
pretation of hysteresis as a consequence of catalyst sur-
face ignition (the term introduced by Frank-
Kamenetskii [10]) is adequately correct (for more
details, see below).
2
A high-rate (high temperature and/or high é /ëé
2
ratio) or low-rate reaction branch is observed on sup-
ported platinum catalysts, depending on temperature and
the partial pressures of the components (see, e.g., [2]).
It is believed that the surface was covered with CO or
oxygen at low or high temperatures, respectively. This
is consistent with the apparent orders of reaction: at
high and low temperatures, the orders of reaction with
Additional problems arise in the oxidation of CO in
respect to CO were equal to +1 and a negative value complex mixtures, in particular, in an excess of hydro-
(
to –1.5), respectively, and the orders of reaction with gen and, especially, in the presence of ëé and ç é,
2 2
respect to é were equal to 0 and +1, respectively.
which are gas mixture constituents after various types
of hydrocarbon steam reforming. In this case, selectiv-
ity is the most important requirement imposed on a cat-
alytic system; in turn, selectivity is a function of mac-
rokinetic reaction characteristics.
2
Indeed, CO is strongly adsorbed mainly in a linear
form on the surface of platinum metal. Oxygen is
bound to surface sites more strongly than CO; however,
it is not sorbed on the surface of Pt covered with CO.
Oxygen does not displace preadsorbed CO, although it
In a detailed study performed by Snytnikov et al.
[1, 11], a 1% Ru/C catalyst was found more active than
can react with CO to form ëé [2–5].
2
On the contrary, Venderbosch et al. [6] believed that 1% Pt/C in the oxidation of CO in an excess of hydro-
external diffusion inhibition was responsible for special gen: the conversion of CO on the former catalyst was
features of the reaction, in particular, variable reaction equal to that on the latter at temperatures several tens of
orders. The low-rate and high-rate modes corresponded degrees lower. However, the 1% Ru/C catalyst
7
01

7
02
ROZOVSKII et al.
decreased the concentration of CO to 500 ppm at
In our preliminary communication [15], we demon-
10−120°ë (a mixture of 1.0 vol % CO, 1.5 vol % é2, strated that the reaction of selective CO oxidation per-
5 vol % H , 10 vol % H O, and 20 vol % CO ; a space formed under conditions of catalyst surface ignition
1
6
2
2
2
–
1
velocity of 7500 h ), whereas the 1% Pt/C catalyst considerably (by an order of magnitude) decreased the
decreased this concentration to ~20 ppm. On the other residual concentration of CO, as compared with an
hand, according to [12], the selectivity of ruthenium ordinary reaction in an isothermal reactor. The effect
catalysts in the oxidation of CO in an excess of hydro- was observed even at a relatively low O /CO ratio of 1.
2
gen was higher than the selectivity of platinum cata-
The aim of this work was to consider in more detail
lysts. It is likely that these contradictions are due to
macrokinetic parameters and experimental results for
macrokinetic reaction parameters.
the reaction of selective CO oxidation under conditions
of catalyst surface ignition. First, we shall dwell briefly
on the phenomenon of surface ignition.
The majority of special features in the oxidation of
CO depend on the high activity of catalysts used in
combination with the high exothermicity of the reac-
tion. In this case, a stepwise change of the reaction to
the external diffusion region is almost inevitable. In
actual practice, it has been repeatedly observed experi-
mentally as an anomalously high temperature depen-
dence of the rate of reaction. According to [2], this sug-
gests ignition. The temperature at which a gradient in
the curve of the temperature dependence of CO conver-
sion reaches a maximum is taken as the ignition tem-
perature. Such an approach can be considered a purely
formal interpretation of experimental observations. As
will be demonstrated below, this rapid growth is actu-
ally due to a spontaneous change of the reaction to the
external diffusion region with corresponding conver-
sion and temperature jumps in the catalyst bed. Of
course, the “ignition temperature” in [2] (and in corre-
sponding sources) is not related to the real surface igni-
tion temperature according to Frank-Kamenetskii [10]
and, generally, it has no direct physical meaning.
THEORETICAL FUNDAMENTALS
OF THE MACROKINETICS
OF HETEROGENEOUS CATALYTIC REACTIONS:
THERMAL REACTION CONDITIONS:
CATALYST SURFACE IGNITION
To avoid misunderstandings, note that, as a rule, the
ignition of a gas mixture cannot occur in heterogeneous
catalysis because of the high rates of heterogeneous
disappearance of radicals and atoms. At the same time,
a process that simulates ignition (the spontaneous tran-
sition of the reaction to the external diffusion region)
can be implemented. According to Frank-Kamenetskii
[
10], this transition can be observed in strongly exo-
thermic catalytic reactions in the case that the positive
+
heat flow q due to the reaction is equal to the negative
–
+
–
heat flow q related to heat removal, q = q , and the
growth of the positive heat flow with temperature is
+
higher than the growth of the negative heat flow, dq /dT >
Carlsson et al. [13] demonstrated (using Fourier
transform IR spectroscopy) a change in the composi-
tion of an adsorbed layer on Pt as the temperature of CO
oxidation was increased. Nevertheless, as found by
Carlsson et al. [14], catalyst surface ignition was the
origin of the temperature jump and the change of the
regime. Carlsson et al. [14] used an original technique
of the stepwise change of oxygen concentration in the
–
dq /dT. Then, as the temperature is increased, the heat
removal power becomes insufficient for the complete
removal of heat released in the reaction. The combina-
tion of conditions becomes crucial, and the reaction
abruptly changes to the external diffusion region; this is
accompanied by a spontaneous increase in the reaction
temperature. Frank-Kamenetskii designated this transi-
tion as catalyst surface ignition and the minimum sur-
face temperature at which the ignition is observed as
the critical ignition temperature.
starting mixture (0.5, 1, 2, 4, and 8 vol % O ) at a con-
2
stant CO concentration (1 vol %). As a result of this,
they managed to detect both ignition and extinction of
the surface of a honeycomb Pt catalyst in a quartz flow
Based on the heat explosion theory, Frank-
reactor. With a stepwise increase in the oxygen content Kamenetskii [10] developed a mathematical apparatus
of a mixture containing 1 vol % CO, ignition was for describing surface ignition; this mathematical appa-
observed at an oxygen concentration of 8 vol %. On the ratus allowed him to find critical conditions for ignition
subsequent stepwise decrease in the oxygen content, and definition parameters. This mathematical apparatus
surface extinction was observed at a much lower O2 was difficult to use because the catalyst surface temper-
concentration of 1 vol %, which corresponds to the ature, which usually cannot be measured experimen-
occurrence of a hysteresis loop. Carlsson et al. [14] tally (under conditions of a gas flow through a granular
simulated the dynamics of ignition; however, the con- catalyst bed, a sensor usually measures the temperature
stants of rate equations should be changed by more than of the surrounding gas mixture), was used as the main
one order of magnitude to describe surface extinction. variable. Rozovskii [16] used the approach of Frank-
It is likely that this partial failure was a consequence of Kamenetskii in order to describe the phenomenon of
incorrect concepts of the reaction mechanism (the surface ignition based on the temperature of the gas
Langmuir–Hinshelwood mechanism was accepted), mixture that contacts with a catalyst and to find a rela-
which affected the kinetic block of the corresponding tionship between the kinetic parameters of the process
program.
and the critical temperature of a gas.
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

SELECTIVE OXIDATION OF CO UNDER CONDITIONS OF CATALYST SURFACE IGNITION
703
Physically, the effect of ignition is due to the fact extinction: the upper steady-state mode, which corre-
that the constants responsible for the rate of reaction sponds to the external diffusion region of the reaction,
and the heat flow, which is proportional to this rate, was retained. Nevertheless, a further decrease in the
exponentially increase with temperature, whereas the temperature finally causes the next disruption (surface
heat removal is approximately proportional to the dif- extinction), at which the reaction abruptly changed to
ference between the temperature of the reaction gas the kinetic region once again. Correspondingly, the
mixture supplied to the system and the temperature in occurrence of hysteresis is characteristic in the situa-
the reaction zone. For strongly exothermic reactions, tion of catalyst surface ignition–extinction (for more
when the equality of positive and negative heat flows is details, see [10, 16]).
reached, the above derivative inequality holds automat-
Yakerson et al. [7, 8] proposed an alternative expla-
ically; this leads to a disruption of macrokinetic reac-
nation for the hysteresis phenomena observed: a local
tion conditions (ignition) as the temperature is
overheating of active sites as a consequence of hindered
increased: an increase in the temperature causes an
reaction-heat removal.At the same time, the probability
exponential increase in the rate of reaction, which
of the local overheating of active sites is lower than the
implies a further increase in the temperature. This unre-
probability of catalyst surface ignition by several orders
strained increase continues until the supply of reacting
of magnitude. This directly follows from ratios between
molecules to the catalyst surface becomes a rate-limit-
heat-conductivity factors and heat-transfer lengths for
ing step. Under new steady-state conditions, the reac-
the solid–gas-flow core system on the one hand and the
tion occurs in the external diffusion region.
active site set–support system on the other hand. Sub-
Although the major portion of heat released in the
botin et al. [7], who criticized the hypothesis that mul-
course of the reaction is consumed for heating the gas
tiple steady states are responsible for hysteresis, noted
mixture, a portion of heat is removed through the reac-
that, according to these concepts, a qualitative change
tor walls. In this case, the heater (furnace) actually
(
leap) of the system from one state to another will occur
becomes a cooler. Macrokinetic conditions can be con-
trolled within a limited range by varying the intensity of
heat removal through walls and heat supply due to the
reaction (by changing the space velocity and composi-
tion of the mixture).
at a point in time in the course of a gradual temperature
change and these transitions will take place instanta-
neously. This argumentation does not stand up under
scrutiny because a leap of this kind can occur only at a
spot (in a particular section of the bed), whereas a new
A change from the kinetic region to the external dif- regime can propagate to the catalysts bed (non-steady-
fusion region can occur either gradually or abruptly state process) for a rather long time depending on
depending on the catalyst activity, heat of reaction, and experimental conditions. At the same time, a leap in the
other parameters.According to Rozovskii [16], the con- catalyst bed is easy to detect by measuring the steady-
dition for the stability of a process on a catalyst grain state values of the rate of reaction, conversion, etc.,
can be written in the form
Qk exp(–E/RT ) RT 1 + b + b
⎞
under changes in experimental conditions.
2
0
2
0
0
⎛
-
---------------------------------------- -- < ------- -- --------------------- -- ,
(1)
⎝
⎠
EXPERIMENTAL
α
E
e
where Q and E are the heat and activation energy of the
Here, we report the results obtained with a 0.1%
reaction, respectively; k is a preexponential factor (Pt, Ru)/ceramic support catalyst provided by Haldor
0
(
which includes a function of concentration in the gen- Topsøe.
eral case); α is a heat-transfer factor; b = RT /E, T is
0
0
To characterize macrokinetically the reaction sys-
tem in more detail, we studied the selective oxidation of
CO in an excess of hydrogen in flow setups with dra-
matically different conditions of heat removal from the
reaction zone. In one of these setups, a nearly isother-
mal reactor (henceforth referred to as the isothermal
reactor) was used, whereas a reactor with a low inten-
sity of heat removal (henceforth tentatively referred to
as the adiabatic reactor) was used in the other setup. In
order to simulate practically important cases, the oxida-
the gas mixture temperature; and e is the natural loga-
rithmic base.
The invalidity of this inequality automatically leads
to surface ignition.
A more rigid situation is characterized by the fol-
lowing inequality [16]:
2
0
2
QβCE/α RT < 1 + 2b + 5b
(2)
(
where β is a mass-transfer factor, and C is the concen-
tion was performed in the presence of CO and water
tration of a key reactant in a gas-flow core). If this ine-
quality is valid, critical phenomena are impossible and
only a smooth change from the kinetic region to the
transition and external diffusion regions can occur.
2
vapor at concentrations to 20 vol %.
The starting gas mixtures (H , CO, O , CO , and N )
2
2
2
2
were prepared in a special gas cylinder and supplied to
After the surface ignition took place, a decrease in a setup with the use of special devices (see below);
the heater temperature below the critical ignition tem- water was supplied independently with a pump through
perature within the temperature range did not cause an evaporator.
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

7
04
ROZOVSKII et al.
tube near the heating coil. This processor also detected
8
Gas
the temperatures of thermocouples 14b and 14c.
1
8
7
6
4
The residual concentration of CO in the dried gas
mixture was measured (and recorded) in an on-line
mode with the use of BINOS 100 IR analyzer 15 (error
of <1 ppm). Drying was performed successively in
coolers 9 and 10 and silica gel trap 17. The flow rate of
a gas mixture was specified with flow controller 18 and
measured with digital meter 7 (IRG-1000). The gas
1
1
4b, 14c
U
2
4‡
1
9
composition at the reactor outlet (CO, O ) was moni-
5
1
2
3
tored by chromatography (zeolite 13X). As a rule, the
reactor pressure was no higher than 0.12 MPa. Usually,
U
1
5
the catalyst was heated to 200°ë in a flow of H ; after a
2
1
7
0
.5-h exposure, the temperature was decreased to 110–
IR
9
1
20°ë and water was supplied; next, H was replaced
2
with a gas mixture.
1
6
After performing an experiment and cooling to
1
6
~
110°ë, the pump was turned off and the catalyst was
1
3
1
0
1
1
cooled in hydrogen.
Calibration gas
1
2
Condensate
Isothermal Reactor
1
2
The setup with the isothermal reactor was based on
a KL-3d catalytic setup (Special Design Bureau of the
Zelinskii Institute of Organic Chemistry, Russian
Academy of Sciences). In this setup, the reactor was
nearly isothermal (high rate of heat removal); it was
made of a coaxially arranged metal cylindrical tube and
thermocouple pocket with a gap of 3–3.5 mm between
them. A weighed portion of the catalyst was diluted
with an inert material in a ratio of 1 : 10 and loaded in
the annular gap. The bed height was about 15 mm. The
setup was equipped with a sampler for the on-line chro-
matographic analysis of the vapor–gas mixture (with-
out water condensation and separation). The following
chromatographs were used: 3700 (column with Pora-
pak T for the determination of water concentrations)
and Chrom 5 (columns with Polysorb for the determi-
Fig. 1. Schematic diagram of the setup with the adiabatic
reactor: (1) reactor, (2) evaporator, (3) furnace, (4) water
pump, (5) catalyst, (6) porous quartz, (7) gas flow meter,
(
(
(
8) pressure gage, (9) water cooler, (10) separator,
11) cooler (0°C), (12) stop valves, (13) chromatograph,
14a–14c) thermocouples, (15) IR analyzer, (16) control
valves, (17) trap with silica gel, (18) gas flow regulator, and
19) processor.
(
A catalyst fraction of 0.250–0.315 mm was used.
Reduction with hydrogen at 500°C was performed
immediately in the reactor.
Adiabatic Reactor
Figure 1 shows a schematic diagram of the setup
with the adiabatic reactor. Quartz reactor 1 was made of
a curved tube (5.5 mm in i.d.); porous quartz 6 was
placed at the reactor inlet (water evaporation zone 2
heated with an electric coil). The reactor was arranged
in cylindrical electric furnace 3 (~50 mm in i.d.). Cata-
lyst sample 5 (~0.2 g) was placed without dilution
between porous quartz plugs 6 in the reactor. Two mea-
suring Chromel–Copel thermocouples 14b and 14c
were introduced into the reactor through a ground-glass
joint (the stainless-steel thermocouple shield diameter
was 1 mm; the temperature of the gas mixture was mea-
sured to within 0.1 K). As a rule, the thermocouples
were arranged with respect to the catalyst bed in the fol-
lowing manner: the lower thermocouple was 1–2 mm
higher than the bed outlet, and the upper thermocouple
was approximately 1–2 mm higher than the bed inlet.
The catalyst bed height was 11–12 mm.
nation of CO and with zeolite 13X for the determina-
2
tion of O , N , and CO). In a number of experiments, a
2
2
BINOS 100 IR analyzer was used for the determination
of the residual concentrations of CO. As a rule, the
reactor pressure was no higher than 0.17 MPa.
Only CO and H O were detected as the products of
2
2
reactions occurring over the tested temperature range
(≤250°ë).
RESULTS AND DISCUSSION
Figure 2 shows a typical curve of the temperature
dependence of CO conversion in the isothermal reactor.
As can be seen, this is an ordinary shaped smooth curve
up to nearly complete degrees of conversion. Because
the reaction on fine grains in the isothermal reactor is
characterized by high transfer coefficients, stepwise
changes in the macrokinetic regime are unlikely in this
The furnace temperature was specified and mea- reactor. Even without special calculations, it is
sured using multichannel processor 19 with adjusting expected that, even through the reaction changes to the
thermocouple 14a, which was arranged in the furnace external diffusion region, this change occurs gradually
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

SELECTIVE OXIDATION OF CO UNDER CONDITIONS OF CATALYST SURFACE IGNITION
705
CO conversion, %
00
CO conversion, %
1
100
90
80
70
60
50
40
30
20
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
1
2
1
0
Ignition
145 150
0
0
1
25
130
135
140
155
Furnace temperature, °C
1
00
120
140
160
180
200
220
Temperature, °C
Fig. 2. Typical temperature dependence of CO conversion
Fig. 3. Typical temperature dependence of CO conversion
for CO oxidation in the isothermal reactor. Reaction mix-
for CO oxidation in the adiabatic reactor. Gas mixture com-
ture, vol %: CO, 0.87; O , 1.04; CO , 17; H , 34; H O, 17;
position, vol %: CO, 0.83; O , 1.33; H , 34; CO , 17; H O,
2
2
2
2
2
2
2
2
and the balance N . Flow rate: (1) 45 or (2) 42 l (g Cat)–1 h .
–1
17; and the balance N . Flow rate: 98 l (g Cat) h–1. Non-
–1
2
2
steady-state values measured in the course of a spontaneous
heating of the catalyst bed (surface ignition) are marked
with crosses.
without regime disruptions. Indeed, the effect of the lin-
ear velocity of a gas flow (and, correspondingly, the
effect of external diffusion inhibition) on the rate of CO
oxidation becomes noticeable only at low linear veloc-
ities, which are lower than those used in the experi-
ments.
As the critical ignition temperature (T ) was
0
reached, the initial segment of the curve for the gas
temperature at the reactor inlet retained the same shape.
This trend was also observed in the curve for gas tem-
perature at the reactor outlet in the first several minutes;
however, it rapidly disappeared and changed to an
accelerating increase to a steady-state temperature in
the ignition regime. The curve of CO concentration
changes has practically the shape of a mirror reflection
of the curve of gas temperature at the outlet: both of
Figure 3 shows a typical curve of the temperature
dependence of CO conversion for the reaction in the
adiabatic reactor, in which catalyst surface ignition was
observed. To obtain another point in the temperature
dependence after reaching about 10% conversion, the
furnace temperature was increased by 1 K, a period was
allowed to elapse in order to reach a steady-state rate of
oxidation, the temperature was increased by 1 K once these curves reflect a gradual propagation of the regime
again, etc. As can be seen in Fig. 3, a spontaneous
increase in the conversion of CO with a corresponding
increase in the catalyst bed temperature (shown with a
dashed line in Fig. 3) was observed at a certain temper-
ature (critical ignition temperature) after a regular
increase in the furnace temperature.As the furnace tem-
perature was further increased, the conversion of CO
changed only slightly. It can also be seen that ignition
was observed at conversions to 30%.
Temperature, °ë
225
215
CO, ppm
10000
4
3
8000
6000
2
1
1
1
1
1
1
1
05
95
85
75
65
55
45
T0
4
2
0
000
000
The dynamics of changes in the concentration of CO
at the reactor outlet and in the temperatures of the fur-
nace and the gas mixture at the inlet and outlet of the
catalyst bed can be seen in Fig. 4, which shows a frag-
ment of data recorded in the course of a particular
experiment. As noted above, the experiment was per-
formed at the stepwise increase of the temperature with
a step of 1 K. It can be seen in Fig. 4 that, at tempera-
tures lower than the critical temperature, an increase in
the furnace temperature by 1 K resulted in a small
increase in the gas mixture temperatures at the inlet and
outlet of the catalyst bed and a decrease in the concen-
tration of CO in the effluent gas. All of the segments
have the shapes of saturation curves; the time taken to
reach a steady-state regime was about 10 min.
2
1
35
1
60 180 200 220 240 260 280 300
Time, min
Fig. 4. Fragment of the recordings of the temperature and
residual concentration of CO under a stepwise increase in
the furnace temperature: (1) residual CO concentration;
(2−4) furnace, bed inlet, and bed outlet temperatures,
respectively. T0 is the critical ignition temperature. Gas
mixture composition, vol %: CO, 0.80; O , 0.80; H , 33;
2
2
CO , 17; H O, 17; and the balance N . Consumption:
2
2
2
91 l (g Cat)–1 h–1.
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

7
06
ROZOVSKII et al.
(T – Tinit), °C
T, °C
3
00
1
1
1
2
1
0
9
8
7
6
5
4
3
2
1
(‡)
250
3
2
1
1
00
50
00
2
1
CO, ppm
2000
1
0
5
10
15
20
Time, min
(b)
8
4
000
000
0
Fig. 5. Dynamics of gas heating at the outlet of the catalyst
bed under a stepwise increase in the furnace temperature
with a step of 1 K: (1) 145
49°C, and (3) 149
vol %: CO, 0.8; O , 0.8; H , 33; CO , 17; H O, 17; and the
146°C, (2) 148
1
150°C. Gas mixture composition,
110
120
130
140
150
160
2
2
2
2
Tfurnace, °ë
–
1 –1
balance N . Flow rate: 91 l (g Cat) h .
2
Fig. 6. Hysteresis in the dependence of (a) the gas tempera-
ture in the catalyst bed and (b) the residual concentration of
CO on the furnace temperature. Gas mixture composition,
along the catalyst bed. The final portion of the curve of
the gas mixture temperature at the inlet of the catalyst
bed also exhibited an additional growth due to longitu-
dinal heat exchange.
vol %: CO, 0.83 O , 1.33; H , 34; CO , 17; H O, 17; and
2
2
2
2
–
1 –1
the balance N . Flow rate: 98 l (g Cat) h .
2
A more detailed pattern of the initial stage of cata-
lyst surface ignition can be seen in Fig. 5, which shows
the initial portions of temperature buildup curves for
the gas mixtures at the outlet of the catalyst bed with a
stepwise increase in the furnace temperature by 1 K. It
can be seen that, as the temperature was increased by 1 K
from 145 or 148°C, heating somewhat increased
Note that experimental conditions (a slow increase in
the temperature with a step of 1 K) were oriented to
minimize differences between heat input and removal
on catalyst surface impregnation. In the case of large
steps, these differences can be great and the ignition
regime propagates very rapidly to the catalyst bed.
An increase in the concentration of oxygen in the
gas mixture gave an analogous effect with the specified
experimental procedure (a stepwise increase in the fur-
nace temperature by 1 K) because of an increase in the
growth of a positive heat flow at a single step. In turn,
this shortened the non-steady-state period under
changes of conditions, as observed experimentally
(by ~2 K) in accordance with an increase in conversion
and the gas temperature reached a steady-state value
after ~10 min. At 149°C, the situation changed dramat-
ically: progressive heating was observed, which ended
(
as seen in Fig. 4) in the establishment of a new steady-
state regime of catalyst surface ignition.
Figure 6 shows a full cycle: sample heating to cata-
lyst surface ignition (and above) followed by a decrease
in the temperature up to surface extinction. An analo-
(Fig. 7).
Let us consider the distribution of temperature along
the bed on slowly heating the reactor to the critical tem-
perature of ignition. For strongly exothermic reactions,
a change to the regime of catalyst surface ignition
occurred at low conversions, at least of no more than
gous behavior was observed at a lower O /CO ratio of
2
1
: 1 [15]. In both cases, hysteresis was clearly defined
in both the curve of gas temperature changes in the cat-
alyst bed and the curve of the residual CO concentra-
tion.
0
.3 [16]. At low conversions, if the reaction is not inhib-
ited by products, a hot spot occurs at the end of the cat-
alyst bed; that is, the reaction occurs at a temperature
that increases along the catalyst bed and, correspond-
ingly, at an increasing rate of reaction.
The relatively long relaxation of the system upon
surface ignition was due to the fact that ignition condi-
tions initially occurred in a section of the bed (in the
neighborhood of a hot spot). An additional heat flow
due to an increase in the temperature in the ignition
zone was favorable for its propagation to adjacent
regions and the bed as a whole. Because the major por-
tion of heat was removed from the system by heating
If the temperature of the inlet gas mixture was much
higher than the critical temperature of ignition, the hot
spot approached the frontal section of the bed.
From the above, it follows that, for the reliable
the gas flow, the propagation of ignition along the direc- determination of the critical ignition temperature of a
tion opposite to gas motion (establishment of a steady- given mixture, the reaction temperature should be
state regime) occurred slowly for an outside observer. increased and the gas temperature should be measured
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

SELECTIVE OXIDATION OF CO UNDER CONDITIONS OF CATALYST SURFACE IGNITION
T – Tinit), °C CO, ppm
707
(
3
60
1
00
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
320
280
3
2
2
2
1
120
80
40
00
60
1
1
2
3
4
0
3
2
0
0
5
10 15 20 25
30
35
40
1
55
160
165
170
175
180
185
T, °C
Time, min
Fig. 7. Effect of the concentration of oxygen in the initial
Fig. 8. Residual concentrations of CO upon oxidation in the
(1) isothermal and (2, 3) adiabatic (ignition mode) reactors.
gas mixture on the relaxation curves of gas temperature at
the outlet of the catalyst bed after ignition at an O /CO ratio
Gas mixture composition, vol %: CO, 0.85; O , 0.85; H ,
2
2 2
of (1) 1.0, (2) 1.3, or (3) 1.6. Gas mixture composition,
33; CO , 17; H O, 17; and the balance N . Space velocity ×
2 2 2
vol %: CO, 0.80; O , (0.80–1.30); H , 33; CO , 17; H O,
3
–1
2
2
2
2
1
0 : (1) 13.5, (2) 12.8, or (3) 14.2 h .
1
7; and the balance N . Flow rate: (1) 91 or (2, 3)
2
9
6 l (g Cat)–1 h–1.
Nevertheless, with maximized heat removal, as was the
case in the isothermal reactor (fine catalyst grains in the
annular gap between metal tubes at a tenfold dilution
with an inert material), this portion can be sufficiently
high for affecting the macrokinetic situation in the reac-
tor. This is responsible for the occurrence of a strongly
at the end of the catalyst bed starting with not very high
conversions at a small step near the ignition tempera-
ture. The ignition temperature can be tentatively evalu-
ated in preliminary experiments performed with a great
temperature step.
exothermal reaction under nearly isothermal condi-
+
The residual concentration of CO in the gas mixture tions. The intensity of positive heat flow q depends on
is one of the most important characteristics of the selec- the rate of reaction. Decreasing the space velocity of a
tive oxidation of CO. It usually passes through a mini- gas mixture for reactions whose overall orders are
+
mum (so-called window) depending on the oxidation lower than unity, we can decrease q by somewhat
temperature: activity is insufficient at low temperatures, decreasing also the intensity of heat removal (because
whereas an excess of oxygen is consumed for the oxi- of a decrease in the heat transfer coefficient with
dation of hydrogen at high temperatures.
decreasing the linear velocity of a gas flow). In this
case, the fraction of heat transfer through reactor walls
increased; finally, this affected the macrokinetic char-
acteristics of the system.
Figure 8 compares the residual concentrations of
CO after oxidation in the isothermal and adiabatic
(ignition mode) reactors. As can be seen, the ignition
mode is much more favorable for the selective oxida-
tion of CO. In this case, the residual concentration of
CO in the adiabatic reactor decreased by approximately
one order of magnitude, as compared with that in the
isothermal reactor. The temperature in the isothermal
Determination of the Apparent Activation Energy
The calculation of kinetic parameters for the reac-
reactor was maintained close to the gas temperature at tion of CO oxidation from data obtained in experiments
the bed outlet in the adiabatic reactor. It can also be with ignition and a comparison of the resulting values
seen in Fig. 8 that, in both of the reactors, the residual with values found using a traditional procedure can
concentrations of CO occurred in a valley region.
serve as an indirect test for the correctness of the
approach used. Because the critical temperature of sur-
face ignition is related to the kinetic parameters of the
reaction, these kinetic parameters can be determined
using the experimental temperatures of ignition found
by varying the following conditions: gas flow rate, con-
centration, etc. For example, if the rate of the gas flow
was varied in two experiments in order to obtain the
ignition temperatures T0(1) and T0(2), the apparent activa-
tion energy can be determined from the equation [16]
It is well known that, in the general case, an increase
in the reaction temperature impairs the selectivity of
oxygen consumption: the fraction of reacted oxygen
consumed for the oxidation of CO decreases. Corre-
spondingly, it is believed that the above temperature
distribution along the catalyst bed is a factor favorable
for the selective oxidation of CO under conditions of
catalyst surface ignition resulting in low residual con-
centrations.
As mentioned above, the heat of reaction was
mainly consumed for heating the reaction gas mixture
and only a portion was removed through reactor walls.
RT
T
[ln(α /α ) + 2ln(T /T )]
0
(1) 0(2) 1 2 0(1) 0(2)
E =
----------------------------------------------------------------------------------------------------
. (3)
T
– T
0(2)
0(1)
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

7
08
ROZOVSKII et al.
Table 1. Determination of the activation energy from the critical ignition temperatures (adiabatic reactor)
Ignition temperature, °C
Experi-
ment no.
Activation energy,
Feed rate, l/h (NTP)
Mass rate v, g/h
kcal/mol
gas in the catalyst bed (T0)
furnace
First reduction*
166.1
1
19.9
19.1
19.2
19.4
35.4
17.9
17.3
17.3
17.5
32
146
146
147
146
154
3
164.3
4
Average
2
163.3
14.8
165.6
176.3
Second reduction*
170.3
5
18.7
18.6
18.5
18.6
60.9
16.9
16.8
16.7
16.8
54.9
144
145
148
146
170
6
167.5
8
Average
7
168.6
14.4
168.8
193.9
Note: The values of v and T set in bold were used in the calculations.
0
*
Reduction with H at 500°C for 2.5 h.
2
Because the ratio between heat-transfer coefficients (α) temperature. In a real experiment, heat is removed from
can be replaced by the ratio between mass rates (v)
the reaction zone in several steps: from the gas flow
core to the inner reactor surface; from the latter to the
outer wall; and, finally, from the reactor wall through an
air layer to the inner furnace surface, which plays the
role of a cooler in the case of ignition. Because the
intensity of all of these processes is a linear function of
the temperature difference between heat-exchanging
bodies, it is clear that the final amount of removed heat
is also a linear function of the difference between tem-
peratures at the ends, that is, at the gas flow core and at
the furnace wall. Hence, it follows that the relations
0
.65
α /α = (v /v )
,
1
2
1
2
it is not difficult to calculate E.
The availability of reactors that are dramatically dif-
ferent in the intensities of heat removal allowed us to
determine the apparent activation energy both by a tra-
ditional procedure and under conditions of a disruption
of the regime in terms of critical ignition temperatures.
A comparison between the resulting values allowed us
to judge the correctness of the approach used.
[16] are also valid in the case under consideration.
It is easy to demonstrate that, with an unknown rate However, it should be taken into account that heat-
equation, the temperature dependence of contact times transfer coefficients for constituent steps are involved
required for reaching equal conversions at different in the overall coefficient with weight fractions corre-
temperatures can be used to determine the apparent sponding to the contributions of the given steps to the
activation energy in the isothermal flow reactor.
overall heat-transfer process. Therefore, the direct cal-
culation of transfer coefficients from experimental data
is impossible. Moreover, errors can occur because of
deviations from linearity in the distribution of tempera-
ture across the catalyst bed. However, the resulting dis-
tortions can be ignored because of the high intensities
of gas-flow stirring in the layer of a grain material and
the short length of cross-sectional transfer.
The determination of the activation energy from the
critical temperatures of ignition from experimental data
obtained in the adiabatic reactor, in which the reaction
mixture temperature changes along the reactor, is less
obvious.
As follows from the above discussion, the critical
temperature of ignition can be detected by a stepwise
change in the heater (furnace) temperature with the
Thus, Eq. (3) can be directly used for calculating the
continuous measurement of the gas temperature at the apparent activation energy.
end of the catalyst bed. In this case, the smaller the step,
the higher the accuracy of the determination of the crit-
ical ignition temperature and activation energy.
In this work, to determine activation energies from
critical ignition temperatures, these latter were deter-
mined in experiments with various feed rates of the ini-
The problem of the use of relations from [16], which tial gas mixture in terms of the above procedure (step-
are applicable, strictly speaking, to a spot or section of wise heating by 1 K). The gas temperature in the cata-
the catalyst bed in the adiabatic reactor, for this purpose lyst bed at which a disruption of the regime occurred
is open to question. In terms of the previous approach was taken as the ignition temperature. Table 1 summa-
[16], this problem is reduced to the choice of a cooler rizes the corresponding data.
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

SELECTIVE OXIDATION OF CO UNDER CONDITIONS OF CATALYST SURFACE IGNITION
709
Table 2. Determination of the activation energy by a traditional method (isothermal reactor)
Temperature, °C
Activation energy,
kcal/mol
Feed rate, l/h (NTP)
Conversion of CO, %
furnace
catalyst
9
.2
–
–
180
180
180
195
195
195
210
211
211
62
51
40
62
51
40
62
51
40
1
2
1
2
3
2
3
5
3.8
0.2
3.7
1.3
3.5
1.0
4.0
1.0
–
193
–
12.5
190
205
–
200
Table 2 summarizes data obtained using a traditional mixture. The found activation energies were consistent
method in the selective oxidation of CO in the isother- within the limits of experimental error.
mal reactor.
Experimental data are given to demonstrate the
The flow-rate ranges 19.1–60.9 and 9.2–51 l/h for dynamics of changes in the regimes: the relaxation of
the adiabatic and isothermal reactors correspond to the the catalyst bed temperature and the residual concentra-
linear-velocity ranges 22–70.5 and 2.6–14.4 cm/s, tion of CO under small disturbances of the heater tem-
respectively.
perature near (up to) the ignition temperature and at the
The above data indicate that the resulting values of critical ignition temperature, as well as the effect of the
E are consistent within the limits of experimental and concentration of oxygen in the starting mixture on the
calculation errors; this evidences the correct interpreta- dynamics of catalyst surface ignition. In the contest of
tion of the observed phenomena.
these data, the interpretation of nontrivial properties of
the selective oxidation of CO given in a number of pub-
lications seems insufficiently correct.
Phase transitions on the surface or in the bulk of a
catalyst, as well as surface ignition, can be responsible
for a stepwise change in macrokinetic reaction condi-
tions. However, the following differences in the behav-
iors of reaction systems in the cases of phase transitions
CONCLUSIONS
In this work, we studied the oxidation of CO in the
presence of an excess of hydrogen in a starting mixture
containing up to 20% CO and H O in reactors with
2
2
high (isothermal reactor) and low (tentatively adiabatic
reactor) rates of heat removal. In the reactor with a low
rate of heat removal, a disruption of the regime (catalyst
surface ignition), which is characteristic of strongly
exothermic reactions, was observed as the oxidation
temperature was increased. Surface extinction with
decreasing temperature also occurred stepwise. In this
case, the extinction temperature was lower than the
ignition temperature; that is, a characteristic hysteresis
was observed. In a flow reactor, the critical conditions
of ignition initially occurred at a hot spot (section) in
the catalyst bed; then, the ignition regime gradually
propagated along the bed. The time taken to reach
(
on the surface, in an adsorption layer, or in the bulk of
the catalyst) and stepwise changes in macrokinetic
reaction conditions can be noted:
•
The critical temperatures of surface ignition and
extinction depend on the rate of the gas flow: they
increase with the rate, and this increase is not observed
in the case of phase transitions. Moreover, a change
from low to high conversions (and vice versa) can occur
either with or without a disruption of the regime
depending on the gas flow rate.
• The critical temperatures of ignition and extinc-
steady-state conditions depended on the disturbance tion depend on catalyst activity: the higher the activity,
caused in the regime of the reaction under near-critical the lower the values of both of the above temperatures.
conditions, which is responsible for the difference In a surface phase transition, the transition temperature
between heat input and heat removal on ignition and, is fixed at least over a limited range.
correspondingly, the rate of regime propagation along
the catalyst bed.
•
Phase transitions occur at a certain combination of
conditions (temperature, concentrations, etc.) and do
For indirectly testing the correctness of this not depend on the intensity of heat exchange. On the
approach, the apparent activation energies were deter- contrary, the intensity of heat removal can play a crucial
mined from data obtained in the isothermal reactor and role in surface ignition, so that the conditions of igni-
from the dependence of the critical ignition tempera- tion vary in various reactors, on the dilution of a cata-
ture in the adiabatic reactor on the feed rate of the gas lyst with an inert material, etc. Moreover, in principle,
KINETICS AND CATALYSIS Vol. 48 No. 5 2007

7
10
ROZOVSKII et al.
the same reaction with the same mixture can be per-
formed either with a disruption of the regime or without
it by varying the intensity of heat removal; however,
this is practically impossible in the case of phase tran-
sitions.
2. Kahlich, M.J., Gasteiger, H.A., and Behm, R.J.,
J. Catal., 1997, vol. 171, p. 93.
3. Bourane, A. and Bianchi, D., J. Catal., 2002, vol. 202,
p. 34.
4. Bourane, A. and Bianchi, D., J. Catal., 2002, vol. 209,
p. 114.
Correspondingly, in the general case, the choice
between the above alternative interpretations is unam-
biguous.
5
6
7
8
. Hoebink, J.H.B.J., Nievergeld, A.J.L., and Marin, G.B.,
Chem. Eng. Sci., 1999, vol. 54, no. 20, p. 4459.
. Venderbosch, R.H., Prins, W., and van Swaaij, W.P.M.,
Chem. Eng. Sci., 1998, vol. 53, no. 19, p. 3355.
In this work, we found experimentally that the igni-
tion mode in the test cases is much more favorable for
the selective oxidation of CO in an excess of hydrogen,
as compared with the reaction in the isothermal reactor.
This is likely due to a more favorable temperature dis-
tribution along the catalyst bed.
. Subbotin, A.N., Gudkov, B.S., and Yakerson, V.I., Izv.
Akad. Nauk, Ser. Khim., 2000, no. 8, p. 1379.
. Gudkov, B.S., Subbotin, A.N., Dykh, Zh.L., and Yaker-
son, V.I., Dokl. Akad. Nauk, 1997, vol. 353, no. 3, p. 347
[Dokl. Phys. Chem. (Engl. Transl.), vol. 353, no. 3,
p. 126].
The almost complete removal of CO from hydro-
9
. Arnby, K., Torncrona, A., Andersson, B., and Sko-
glundh, M., J. Catal., 2004, vol. 221, p. 252.
gen-containing mixtures (1% CO and O and to 20%
2
CO and H O) can be implemented by combining an
2
2
1
0. Frank-Kamenetskii, D.A., Diffuziya i teploperedacha v
khimicheskoi kinetike (Diffusion and Heat Transfer in
Chemical Kinetics), Moscow: Nauka, 1967.
adequately chosen catalyst with optimized macroki-
netic reaction conditions even at a consumption of
–
1
–1
about 100 l (g Cat) h . The corresponding results will
be reported elsewhere.
1
1
1
1
1
1. Snytnikov, P.V., Cand. Sci. (Chem.) Dissertation,
Novosibirsk: Boreskov Inst. of Catalysis, 2004.
2. Han, Y.-F., Kahlich, M.J., Kinne, M., and Behm, R.J.,
Phys. Chem. Chem. Phys., 2002, vol. 4, p. 389.
ACKNOWLEDGMENTS
3. Carlsson, P.-A., Skoglundh, M., Fridell, E., Jobson, E.,
This work was supported by Haldor Topsøe, the Pre-
sidium of the Russian Academy of Sciences (program
no. 26), and the Russian Foundation for Basic Research
and Andersson, B., Catal. Today, 2002, vol. 73, p. 307.
4. Carlsson, P.-A., Skoglundh, M., Thormahlen, P., and
Andersson, B., Top. Catal., 2004, vol. 30/31, p. 375.
(project no. 06-03-32848).
5. Rozovskii, A.Ya., Kipnis, M.A., Volnina, E.A., Lin, G.I.,
and Samokhin, P.V., Kinet. Katal., 2004, vol. 45, no. 4,
p. 654 [Kinet. Catal. (Engl. Transl.), vol. 45, no. 4,
p. 618].
REFERENCES
1
. Snytnikov, P.V., Sobyanin, V.A., Belyaev, V.D., Tsyrulni- 16. Rozovskii, A.Ya., Kinetika topokhimicheskikh reaktsii
kov, P.G., Shitova, N.B., and Shlyapin, D.A., Appl.
Catal., A, 2003, vol. 239, p. 149.
(Kinetics of Topochemical Reactions), Moscow: Nauka,
1980, p. 116.
KINETICS AND CATALYSIS Vol. 48 No. 5 2007