Catalytic and Electrocatalytic Oxidation of Methane on Palladium. Electrodes in a Solid Electrolyte Cell. C. Athanasiou, G. Marnellos, P. Tsiakaras and M.
Ionics 2 (1996)
353
Catalytic and Electrocatalytic Oxidation of Methane on Palladium Electrodes in a Solid Electrolyte Cell C. Athanasiou, G. Marnellos, P. Tsiakaras and M. Stoukides Chemical Engineering Dept. and Chemical Process Engineering Res. Inst., Thessaloniki 54006, Greece
Abstract. The catalytic oxidation of methane on polycrystalline palladium films was studied at 550-750~ and atmosheric total pressure. The reaction was studied under both open and closedcircuit. Under open circuit, and when yttria-stabilized zirconia (YSZ) was used as solid electrolyte, the technique of Solid Electrolyte Potentiometry (SEP) was used to monitor the thermodynamic activity of oxygen adsorbed on the Pd electrode during reaction. The main products were those of complete oxidation, i.e. CO2 and H20. Under closed-circuit, the effect of electrochemical oxygen "pumping" to or from the catalyst was studied. Non-faradaic (NEMCA) phenomena were observed but the reaction rate enhancement factors (A) were not as large as with previously studied catalytic systems.
1. Introduction Due to its potential industrial interest, the oxidation of methane has received considerable attention in the last ten years [1, 2]. Methane is known for its abundance, but also for its chemical stability. When oxidized, it produces mainly carbon dioxide and water. Palladium catalysts have been reported to give relatively good results concerning selectivities to oxygenated products [3, 4]. In the present work, kinetics are combined with potentiometric data in order to study the mechanism of methane oxidation on reduced and oxidized palladium surfaces. The technique of solid electrolyte potentiometry (SEP) is used to continuously monitor the thermodynamic activity of oxygen adsorbed on the catalyst surface. To this end, the reaction is studied in a solid electrolyte cell in which one of the electrodes serves at the same time as the reaction catalyst. This technique, first used to study catalytic oxidations on noble metals [5], has been used in conjunction with kinetic measurements in a large number of metal-catalyzed oxidation reactions including methane oxidation on Ag and on Pd [6, 7]. More recently, Hildenbrand and Lintz elegantly showed that SEP can be applied successfully when copper oxides are used as electrodes [8]. The above authors studied the partial and complete oxidation of propene to acrolein and
reported on the relationship between reaction rates and phase transformations of the catalyst surface [8, 9]. In a similar manner, Balian et al, used SEP to study the oxidation of hydrogen on copper and on copper oxide surfaces [10]. In addition to SEP measurements, the present study includes results of closed-circuit operation in which, oxygen is electrochemically "pumped" to or from the catalyst surface. The idea of electrochemically enhancing reaction rates using solid electrolytes, has gained particular interest since in a lot of catalytic systems it has been shown that both, reaction conversions and product selectivities can be significantly altered [11, 12]. Nonfaradaic electrochemical modification of catalytic activity (NEMCA) has been observed in a large number of reactions and the origin and characteristics of these phenomena have been recently reviewed [12]. Hence, in the present study, the effect of electrochemical oxygen pumping on the rate of methane oxidation is examined and compared to other catalytic reactions that have been studied using the same catalytic and electrochemical techniques [5, 12].
carbon dioxide, respectively, on copper oxides and
description of the experimental apparatus has been pro-
2. Experimental
2.1. Apparatus and Catalyst Preparation. A detailed
354 vided in previous communications [7, 10]. The reactor basically consisted of an yttria-stabilized zirconia (YSZ) tube of 16mm IDI 19mm OD and closed at one end. The Pd catalyst is placed at the inside bottom of the tube. Two Ag electrodes (counter and reference) were placed at the outside bottom of the YSZ tube. The analysis of reactants and products was done by on-line gas chromatography using a Hewlett Packard 5890 Series II with two thermal conductivity detectors (TCD). A molecular sieve 5-A column was used to separate hydrogen, oxygen, nitrogen, methane and carbon monoxide and a porapak-Q column was used to separate carbon dioxide, ethane, ethylene and water. Furthermore, a paramagnetic oxygen analyzer (Oxynos 100 by Rosemount Analytical), was used for continuous monitoring of oxygen in the inlet and outlet streams. Constant currents or voltages were applied between the working and the counter electrodes with the help of a EG&G model 363 potentiostat-galvanostat. Currents and voltages are measured by means of Bar Graph HC-737 digital multimeters. The polycrystalline porous palladium film was prepared from an A-2985 palladium resinate obtained from Engelhard Corporation. A few drops of the resinate were deposited on the bottom of the YSZ tube followed by drying at 200~ for 2 hours. Then the catalyst was calcined at 500~ for 2 hours and at 900~ for another 2 hours. Scanning electron micrographs of the Pd electrode prepared with the above described procedure showed that the average diameter of the crystallites of a fresh catalyst was 2-3 microns. The total mass of metallic palladium used was about 100 mg. By using the SEM data and comparing rates with those reported previously [7], the total catalytic surface area was calculated to be about 200 cm 2. Silver, instead of palladium, was used for the preparation of the counter and reference electrodes mainly because silver adheres to the zirconia outside surface much more strongly than palladium. The preparation and characterization of the silver electrode has been described in detail in previous communications [6]. 2.2. Solid Electrolyte Potentiometry. The basic principle and applicability of SEP have also been explained previously [5, 10]. SEP utilizes a stabilized zirconia solid electrolyte cell with one of the electrodes exposed to the reacting mixture and thus serves as a catalyst for the reaction under study. The other electrode is exposed to the air and serves as a reference electrode. The ther-
Ionics 2 (1996) modynamic activity of atomic oxygen adsorbed on the catalyst surface is given by the Nernst equation : a o = (0.21) 1/2 exp (2FE/RT)
(1)
where F is the Faraday constant, R is the ideal gas constant, T is the absolute temperature, E is the electromotive force (emf) of the cell and a o is the activity of atomically adsorbed oxygen [5]. The validity of equation (1) is based on several assumptions [5], among which the most questionable for the present system, is that atomically adsorbed oxygen is the only species to equilibrate rapidly with oxygen ions at the gas-electrode-electrolyte boundary. Although certainly valid for the reference electrode, this assumption may not hold for the catalyst-electrode. If, in addition to the O- - equilibration with adsorbed oxygen: O'" (
) O+2e
(2)
a charge transfer reaction with adsorbed methane also takes place at comparable rate: CH4+40""
(
)
H20+2e
(3)
then a mixed potential is established and the measured emf provides a qualitative and not quantitative measure of surface activities [5]. In any case, if we assume that equation (1) holds, the thermodynamic activity of adsorbed atomic oxygen can be continuously monitored during reaction [5]. At the same time the gas-phase oxygen partial pressure above the catalyst surface can be measured independently and the two values can be compared. If thermodynamic equilibrium is established between adsorbed and gaseous oxygen, then Po2 1/2 = ao [5]. If, on the other hand, the steady state rates of oxygen adsorption and reaction on the catalyst surface become comparable, then the surface reaction can pull down the surface oxygen activity a o to become several orders of magnitude lower than Po21/2. 3. Results and Discussion The reaction rate and the surface oxygen activity behavior was studied at 550-850 ~ and 1 atm. The methane and oxygen partial pressures varied from 0.005 to 0.1 bar and nitrogen was used as diluent. As shown in
Ionics 2 (1996)
355
10 8
PCH4 = 5 kPa
1~ t
PCH4= 5 kPa
PO2 = 10 kPa
8-~
PO2 = 10 kPa
T = 600 ~
1
T = 650 ~
6d
"6
4
4~
~4
9
prereduced m , ~ A
~
t
_.
a
A
preoxidized prereduced 0
9
;
.
9
9
i
,
,
,
!
9
9
.
i
9
9
.
9
.
u
.
.
.
,
9
'
9
,
9
9
PCH4 = 5 kPa
0.1
; ~ :
,
,
9
.
PCH4= 5 kPa
PO 2 = 10 kPa
preoxidized
,
9
PdO
T = 600 ~
PO 2 = 10 kPa
preoxidized
g
: :--- -'prereAuce'd . . . . . . . . .
T = 650 ~
prereduced
0.1
PdO Pd Pd 0.01
.
0
.
.
,
200
,
400
,
600
,
-
800
.
1000
0.01
.
0
.
,
200
.
400
9
.
.
.
.
600
.
800
i000
Time on stream, min
Time on stream, rain
Fig. 1. Reaction rate (top) and cto (bottom) dependence vs time on stream. Triangles and circles correspond to "prereduced" and "preoxidized" treatment respectively. T=600~ PO2= 0.10 bar, PCH 4 = 0.05 bar.
Fig. 2. Reaction rate (top) and txo (bottom) dependence v s time on stream. Triangles and circles correspond to "prereduced" and "preoxidized" treatment respectively. T=650~ PO2= 0.10 bar, PCH, = 0.05 bar.
previous communications [6, 10], the reactor behavior was very close to that of a well-mixed reactor (CSTR)
oxygen ratio (PcH 4 = 5 kPa, PO2 = 10 kPa) at the reactor exit. The top part of the figure contains reaction rate data
for the range of volumetric flow rates employed in the
and the bottom part contains potentiometric measure-
present study. There were no products other than CO2
ments, both versus time on stream. Two types of curves
and H 2 0 detected in the outlet stream.
are prepared, named "preoxidized" and "prereduced", res-
The catalytic activity of a freshly made catalyst
pectively. The procedure for the aquisition of the data
reached a steady value after a relatively long induction
was the following: The catalyst was first exposed to air
period that lasted about 48 hours. It was assumed that a
at 600~
steady state was reached when there was no more than
dized. Then, the reacting mixture was introduced and the
for about 8 hours so that it gets totally oxi-
3% change in the reaction rate and a o values over 1 hour.
system was left to reach steady state. The "preoxidized"
Such long induction times have been reported in the
data were then obtained for about 1000 minutes (-17
literature for this particular reaction and catalyst and
hours). At the end of the experiment, the reacting mix-
several explanations have been proposed [7, 13, 14].
ture was cut off and air was introduced again in order to
Of particular interest are the long transient periods of
clean up the catalyst from possible carbon deposition.
time required for a new steady state to be reached after
Then, hydrogen was passed over the catalyst for about 8
the introduction of a new feed composition in the re-
hours at the same temperature (600 ~
The reacting
actor. Depending on the conditions, it may take 3 or
mixture was then introduced and the "prereduced" darn
more hours from the time a new composition is intro-
were obtained for another 1000 minutes. The oxygen
duced in the reactor until a steady state is attained.
activity measurements that were obtained simultaneous-
Figure 1 contains rate and oxygen activity data obtained at 600~
and for a stoichiometric methane/
ly are shown in the bottom part of Fig. 1 where ao is plotted versus time on stream. The continuous horizon-
356
Ionics 2 (1996) 10
10
i PcH4= 5 kPa i P~ = 10 kPa T = 750 ~
prereduced 8 O
-66
-6 preoxidized ~
d
2
,
..
,
o
*
~4
o
preoxidized
~176
a 9
a
~
_ --
v
ov A
~
v
v
o 9
9
po 2=lOkPa T = 700 ~
-
~,o -
A
.IL
prereduced
1. PCH4= 5 kP a
PdO
PdO
Po 2 = 10 kPa T = 700 ~
preoxidized
............ r
O 0.1"
.
.
.
preoxidized . . . . . . .
.
.
prereduced
.
.
.
prereduced
@
0.1"
Pd
Pd
PCH4= 5 kPa Po 2 = 10 kPa T = 750 *C 0.01
.
200
.
,
400
.
.
.
,
600
.
.
800
IO00
0"010
"
"
200
Time on stream, min
"
"
4~J0 "
600
"
800
"
1000
Time on stream, min
Fig. 3. Reaction rate (top) and (zo (bottom) dependence v s time on stream. Triangles and circles correspond to "prereduced" and "preoxidized" treatment, respectively. T=700~ PO2= 0.10 bar, P C H 4 = 0.05 bar.
Fig. 4. Reaction rate (top) and oto (bottom) dependence vs time on stream. Triangles and circles correspond to "prereduced" and "preoxidized" treatment respectively. T=750~ PO2= 0.10 bar, PCH 4 = 0.05 bar.
tal line corresponds to the thermodynamic stability limit
stability limit of PdO is now much closer to the ao data
of the Pd-PdO system at this temperature. Hence, in the
(Fig. 2, bottom).
area below that continuous line, only metallic palladium
In a similar manner, Figs. 3 and 4 contain rate and ao
is thermodynamically stable. Similarly, in the area
data obtained at 700 ~
above that line, only the PdO is stable. It can be clearly
700 ~
seen in Fig.
and 750 ~
respectively. At
the stability limit of PdO falls on the a o curves
1 (bottom) that all the experimentally
(Fig. 3, bottom) find the "prereduced" and "preoxidized"
obtained values of a o fall in the regime in which the
curves seem to stay quite apart from each other even
thermodynamically stable phase is PdO. Although it
after 17 hours on stream (top of Fig. 3). At 750 ~
takes about two hours or more, the reaction rates curves
Pd-PdO stability line has moved above the ao curves
seem to eventually fall on the same path regardless of the pretreatment.
(Fig. 4, bottom), which means that the reaction was
the
studied under conditions that the thermodynamically
In Fig. 2, the same experimental procedure was
stable phase is metallic Pd rather then PdO. The reaction
followed and the same exit gas composition was used
rate data at that temperature (Fig. 4, top) indicate that
(PCH4 = 5 kPa, PO2 = 10 kPa), only the temperature was increased to 650 ~ It can be seen that at this tem-
the "prereduced" and "preoxidized" curves tend to coincide again.
perature, it takes much longer for the "prereduced" and
In Fig. 5, the rate and oxygen activity data for a
"preoxidized" rate curves (top part of figure) to fall on
stoichiometric exit gas composition
each other. The horizontal line that corresponds to the
PO2 =10 kPa), are plotted vs temperature. It can be seen
(PCH 4 = 5 kPa,
Ionics 2 (1996)
357
15
20
PCH4= 5 kPa PO2 = 10 kPa
Pcri4= 1 kPa 16 P02= 2 kPa -T= 6500C "6 *-----"----
prereduced
~- 10
"6
_.,___--~ .-----o'--"
I~ 12
ds -
preoxidized
A t"q
4: O"
.
.
.
.
,
.
.
.
.
1500 ,
3rd round
[PdO
~
1000] POt4= 1 kPa P o -- 2 kPa
5OO T=650"C preterit/f
0.1
J
o
Pd
- 500 PcH4= 5 kPa P o 2 = 10 kPa
0.01" 600
1
650
7;0
7~0
- 1000
- 1500 -2
800
.
.
.
.
.
-1
T, ~
.
.
.
i
1
.
.
.
.
!
.
.
.
.
2
i
'.
-
9
-
3
4
I/2 F 10-8 g-atoms O / sec
Fig. 5. Reaction rate (top) and s o (bottom) dependence on temperature. Triangles and circles correspond to "prreduced" and "preoxidized" treatment respectively. PO2= 0.10 bar, PCH4 = 0.05 bar.
Fig. 6. Effect of oxygen "pumping" on the reaction rate (top) and cell voltage (bottom). T=650~ PO2= 0.02 bar, PCH4 = 0.01 bar.
again that the maximum deviation between the "prereduced" and "preoxidized" rate data occurs around the point where the at data hit the stability limit of PdO which corresponds to a temperature close to 710 ~ At higher temperatures, the picture is reversed, i.e., the "prereduced" catalyst is less active than the "preoxidized" (Fig. 5, top). Figures 6, 7 and 8 contain closed-circuit data obtained at 650, 700 and 750 ~ respectively. A stoichiometric exit gas composition (PCH4=I kPa, PO2=2 kPa) was used again in all these experiments. In Figs. 6-8, the reaction rate (top part) and the voltage difference between working and reference electrode (bottom part) are plotted versus the rate of oxygen transport through the YSZ electrolyte which is expressed in gram atoms of oxygen per second and is equal to I/2F where I is the imposed current and F is Faraday's constant. The current was arbitrarily called positive (+), when O- " was "pumped" to the catalyst (working eleclrode) and consequently negative (-), when O" - was "pumped" away
from the catalyst. A general observation is that a positive current results in an increase of the reaction rate while a negative current has the opposite effect (Figs. 6 -
8). The same data that were shown in Figs. 6, 7 and 8 are now presented in the form of rate increase vs I/2F in Figures 9, 10 and 11, respectively. To have a means of evaluating the effect of non-faradaic promotion (NEMCA), the dimensionless parameter A is also shown in these figures. The enhancement factor A is defined as: A = Ar/(I/2F)
(4)
where Ar is the difference between rate of consumption at current I and at zero current (open-circuit), both expressed in gram atoms of oxygen per second. It can be seen that the A values are higher at 650 ~ (Fig. 9) where according to the SEP measurements, the catalyst is oxidized. At 750 ~ where the metallic phase is ther-
358
Ionics 2 (1996) 10-
10
PCH4 = 1 kPa PO2 = 2 kPa T = 700 ~
~8
8
Pcrh = 1 kPa PO2 = 2 kPa T = 750 ~
~6
%
'9
G4-
4" o o
~2
2-
1600
1500
PcH4 = 1 kPa 800. Po 2 = 2 kPa T = 700 ~ >
PCH4 = 1 kPa Po2 = 2 kPa 950 T = 750 ~
f
f
100 9 -600 -
> 3o0
/ -350
-1300. -2000 . . . . -4
o
, ........ -2
, .... 2
, .... 4
-1000 . . . . -20
, .... 6
i ....
-15
I/2F 10-8 g-atom O / sec
~
-lo
l
....
i ....
, ....
-5 o 5 10 I/2F 10 -s g-atom O / sec
, ....
15
20
Fig. 7. Effect of oxygen "pumping" on the reaction rate (top) and cell voltage (bottom). T=700~ PO2= 0.02 bar, PCH 4 = 0.01 bar.
Fig. 8. Effect of oxygen "pumping" on the reaction rate (top) and cell voltage (bottom). T=750~ PO2= 0.02 bar, PCH 4 = 0.01 bar.
modynamically stable, in general A