7711
J. Phys. Chem. 1993,97, 7711-7718
An Infrared and Kinetic Study of CO Oxidation on Model Silica-Supported Palladium Catalysts from to 15 Torr Xueping Xu and D. Wayne Goodman* Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255 Received: February 25, 1993; In Final Form: April 28, 1993
The catalytic CO Oxidation on model silica-supported palladium catalysts has been investigateed with in situ infrared reflection-absorption spectroscopy (IRAS), temperature-programmed desorption/reaction, and kinetic measurements in the pressure range of lC9-15 Torr and the temperature range of 350-1000 K. The model catalysts were prepared by evaporating palladium onto a silica thin film supported on a Mo( 110) substrate, an arrangement that facilitates the use of electron spectroscopies and IRAS. The CO oxidation reaction was studied in three pressure ranges: (1) coadsorption and reaction of CO and 0 under UHV conditions, (2) CO oxidation between 10-8 and 1O-a Torr (350-1000 K), and (3) at 15 Torr (500-650 K). A generic rate law is derived that adequately describes the observed kinetic behavior for both low- and high-pressure conditions. At low temperatures, the oxidation rate is only proportional to the ratio of 02 and CO pressure and exhibits an activation energy of 27 kcal/mol. The rate increases with temperature to a maximum (500-600 K for a pressure of 1O-a Torr) and then decreases. The temperature of the rate maximum increases with CO pressure, decreases with oxygen pressure, and increases with the total pressure for a constant Po2f PCOratio. The reaction order with respect to the CO pressure also changes from negative to positive with an increase in temperature. This change in the kinetic behavior is attributed to a change in the rate-determining step with temperature. At temperatures above the rate maximum, the steady-state CO coverage is near zero while the oxygen coverage increases with temperature. Both the activation enthalpy and the entropy of the surface CO 0 C02 reaction are found to be dependent on the oxygen coverage. The transition state of the reaction of CO and 0 lies nearer the reactants along the reaction coordinate for low steady-state CO and 0 coverage than for high 0 coverages.
+
1. Introduction
There has been a lingering concern regarding the application of surface science to the study of heterogeneouscatalysis.’ Surface sciencestudies under well-defined conditions can provide insights into many fundamental processes of heterogeneous catalysis;2 however, due to a pressure difference of some 10 orders of magnitude between ultrahigh vacuum (UHV) and practical catalytic conditions,certain important surface processes that occur during catalytic reactions may not be accessiblein UHV s t u d i e ~ . ~ This potential discrepancy in the two domains of study has been referred to as the “pressure gap”.3 Furthermore, most surface science studies are performed on well-defined single crystal surfaces,whereas practical catalysts consist typically of dispersed metals on high surface area supports. It has been shown that the interaction between a metal and its support can alter its catalytic properties.4 In addition, the reactivity and selectivityfor catalytic reactionsoften vary with the particle size of the supported metal.5 This latter discrepancy between the nature of the catalysts in the two domains of study has been referred to as the “material gap”. In order to bridge the pressure gap, several research groups have designed high-pressure cells coupled to UHV surface analytical chambers.69’ In such designs, catalytic reactions at elevated pressurescan be studied, whereas the surfacecomposition and structure can be determined before and after reaction. Issues related to the material gap have been addressed by synthesizing model supported catalysts suitable for UHV characterization and high-pressure studies.* Recently, we have prepared silica-supported palladium catalysts by evaporating palladium onto a silica thin film at preparation conditions that yield varying particle dispersions.9 The silica thin film is, in turn, supported on a Mo( 110) substrate.10 This planar silica surface has been shown to be an excellent model for a high surface area silica. A silica thin film is particularly useful for surface science studies with respect to sample handling and sample charging, 0022-3654/93/2097-77 11$04.00/0
-
difficulties frequently encountered with bulk oxide materials. In addition, a silica film on Mo( 110) is superior to thermally grown Si02 on a silicon substrate because Si02 on Mo( 110) is stable up to 1500 K; Si02 on Si decomposes at much lower temperatures. 1 1912 This paper addresses both pressure and material issues of catalysis by studying CO oxidation on model silica-supported palladium catalysts at UHV and elevated pressure conditions. The catalytic CO oxidation reaction has been studied on palladium catalysts at a variety of pressure conditions by several research groups, yet many questions regarding the reaction remain.13-21 Although CO oxidation is a relatively simple reaction (CO 1/202+ COz), the kinetics of the reaction is rather complicated; for example, the reaction order and the apparent activation energy vary with the reaction temperature and pressure. In this study, we have measured the reaction kinetics over a wide pressure (1p9-15 Torr) and temperature range (400-1000 K). A simple generic model is developed to explain all the kinetic observations; previous models apply only to a limited set of conditions.16J8
+
2. Experimental Section
All experiments were carried out in an UHV surface analysis chamber coupled to a high-pressure reaction cell. The chamber is equipped with an Auger electron spectrometer (AES) and a quadrupole mass spectrometer. The high-pressure reaction cell is directly coupled to the surface analysis chamber via a slidingseal interface.22 The high-pressure cell is equipped with flangemounted CaF2 windows for in situ Fourier transform infrared reflection-absorption spectroscopy (IT-IRAS). The IRAS spectra were acquired using a Mattson Cygnus 100spectrometer, with single beam optics adjusted for an 85’ incident angle. The pressure in the UHV chamber was measured with a GranvillePhillips ionizationgauge without any correction,and the pressure in the high-pressure cell was measured with a MKS Baratron 0 1993 American Chemical Society
Xu and Goodman
7712 The Journal of Physical Chemistry, Vol. 97, No. 29, 1993
capacitance manometer. The details of the apparatus have been described elsewhere.22 The preparation and characterization of the silica thin films on Mo( 110) have been described in refs 10, 11, and 23. The palladium evaporation source consists of a 0.25-mm Pd wire (99.997%, Johnson Matthey Chemical Limited) wrapped around a tungsten filament; a line-of-sight mass spectrometer was used to monitor the Pd flux. Palladium was deposited onto the silica thin film (- 100A) at either 100K or room temperature, followed by an anneal to 1000 K. The palladium coverage, dispersion, and structure were determined with thermal desorption spectroscopy, IRAS of CO, and atomic force microscopy; results of these studies are reported in more detail el~ewhere.~ However, in brief, the palladium particle size depends on coverage, and the particles are composed primarily of (1 11) and (100) facets. Carbon monoxide (99.99%) and 02 (99.998%) were obtained from Matheson, transferred to glass bulbs, and used without further purification. No metal carbonyl contamination was detected following a high-pressure reaction. Temperatureprogrammed desorption/reaction spectra (TPD) were collected with a computer-interfaced mass spectrometer. The heating rate was 5 K/s for CO TPD and 10K/s for 0 2 TPD. The temperature was measured with a W5%Re-W26%Re thermocouple spotwelded to the back of the crystal. The steady-state rate of CO oxidation at low pressure ( 1e - 1 W Torr) was measured by monitoring the C02 intensity with a mass spectrometer. Carbon monoxide and 0 2 were admitted to the UHV chamber through twoseparateleakvalves. Since thesystem is continuously pumped, the C02 formation rate was proportional to the increase in the C02 pressure above background. The rate of CO oxidation in the high-pressure (- 10 Torr) batch reactor was measured via gas-phase C 0 2 IR absorption and the change in the reactor cell pressure. The rate of CO oxidation was proportional to the rate of the total pressure decrease. The two methods gave comparable results; however, the results obtained using pressure measurements were found to be more convenient and accurate. The backof the crystal was coated with a 100-A silica film to minimize background reactions. CO oxidation was also carried out on samples without palladium at identical conditions, and the reaction rate was shown to be more than 1 order of magnitude smaller, demonstrating that contributions to the reaction by the heating leads and the sides of crystal were minimal.
CO/Pd/SIOz (1W A)
460 K
0
380
460
540
700
620
Temperature (K) Figure 1. Temperature-programmeddesorption of CO from Pd/silica/ Mo( 110) as function of CO exposure at 300 K. The silica film (- 100 A) fully coveredthe Mo( 110)surface.ll The Pd film was 6 X 1015atoms/ cm2( 5 ML) and annealed to 900 K yielding particles of 50 nm. The heating rate was 5 K/s. CO was adsorbed at -300 K with coverages of (a), 0.1; (b), 0.2; (c), 0.5; (d), 0.8; (e), 0.9; and (0, 1, relative to saturation at 300 K.
-
-
Annealed to 9 0 0 K Pco -1 O'Tori
_. I
11
1894
-
I
3. Results 3.1. Carbon Monoxide on Pd/SiOt. Figure 1 shows temperature-programmed desorption spectra for CO as a function of exposure on a model silica-supported palladium catalyst. For these data the nominal palladium coverage corresponded to - 5 monolayer (ML) and the palladium particle size was -500 A in diameter. The CO desorption spectra for other palladium coverages were very similar to those in Figure 1, varying only slightly in intensity. At low coverages, carbon monoxide desorbs in a peak centered at 490 K. An increase in COcoverage broadens the desorption peak toward low temperature and gradually shifts the peak desorption temperature to 460 K. Following CO adsorption a t 100 K, additional desorption features are observed between 200 and 360 K. The spectra for CO desorption from Pd/SiO2 can be accurately described by a convolution of CO desorption spectra from Pd single crystal surface^.^.^^*^^ The CO desorption feature a t 460 K has been shown to correspond to CO desorption from bridging and 3-fold hollow sites. Figure 2 shows a series of infrared reflection-absorption spectra for CO adsorbed onto supported palladium particles. Above 350 K, the absorption bands are characteristic of 2-fold and 3-fold CO adsorption (1830-2000 cm-1).3~26,2~ The surface CO molecules were in equilibrium with 10-6 Torr of CO during the IRAS data acquisition. Upon cooling the surface to 100 K,
2200
I
2120 2040
I
I
1960 1880
18W
Frequency (cm-' ) Figure 2. IRAS spectra of CO adsorbed on Pd/silica. The spectra were collected at a 10-6 Torr of CO background and at the indicated surface
temperatures. three well-resolved bands are evident. These bands at 21 10 and 1894 cm-1 are identical to those corresponding to a compressed CO adsorption structure (2x2) observed on Pd(ll1) at similar temperature and pressure conditions.3 These two bands are assigned to CO adsorption onto the atop and 3-fold hollow sites, respectively. The band at 2004 cm-1 is assigned to the 2-fold bridging site on Pd( 100) or Pd( 110).27928The blue shift of the bands with an increase in the CO coverage is due to dipoldipole coupling. 3.2. Oxygen on Pd/SiOs. Figure 3 shows the temperatureprogrammed desorption spectra for oxygen from Pd(5 ML)/ silica as a function of 02exposure. At low coverages, 02desorbs at -850 K. Increasing the 02 exposure gradually shifts the desorption peak to 800 K. A small amount of oxygen dissolves into the bulk palladium, indicated by a 02 desorption peak a t 1200K, approximately the palladium desorption temperature.9 0 2 desorbs in a single peak a t 800-900 K from Pd( 111) and
-
CO Oxidation on Silica-Supported Palladium Catalysts
The Journal of Physical Chemistry, Vol. 97, No. 29, 1993 7713
OplPdlSiOz 800 K I ?
?
s
PI I
200 K
yq 1835
500
600
700
800
900
1000
Temperature (K) Figure 3. Temperature-programmeddesorption of 02 from Pd/silica/ Mo( 110) as a function of oxygen exposure at 300 K. The palladium catalyst was the same as that of Figure 1. The heating rate was 10 K/s. Theoxygencoveragewas(a)0.1,(b)0.2,(c)0.3,(d)0.4,(e)0.6,(f)0.8, and (g) 1, relative to saturation.
T--T--l CO/O/Pd/SiOz (100
A)
420K
1
11 100 200
300 400 500
600 700
Temperature (K) Figure 4. Temperature-programmedreaction spectra of coadsorbed CO and 0 on Pd/silica. The Pd catalyst, prepared as that in Figure 1, was exposed to 02 to saturation, followed by CO adsorption to saturation at 100 K.
polycrystalline palladium but in two peaks centered at -700 and 800 K from Pd(100) and Pd(llO).l4J2333 The TPD spectra of oxygen and carbon monoxide and the IRAS of CO suggest that the majority surface facets are Pd( 111) and Pd( 100).9 3.3. Coadsorption of Carbon Monoxide and Oxygen on Pd/ SiOz. Figure 4 depicts the temperature-programmed reaction spectra of coadsorbed carbon monoxide and oxygen. Oxygen was adsorbed first tesaturation, followed by a saturation exposure of CO at 100 K. C02 is evolved mainly in a peak centered at 420 K, in a small peak at 510 K, and in a broad feature over the temperature range 200-400 K. At these saturation exposuresof 0 2 and CO, all surface oxygen was consumed and the unreacted CO desorbed at 490 K. At low CO exposures and saturation oxygen conditions,C02 production increased with CO exposure, with a concomitant decrease in 0 2 desorption at 800-900 K. The activation energies for the reaction of adsorbed carbon monoxide and oxygen are estimated to be 9-24 kcal/mol for the C02 product below 400 K, -25 kcal/mol for the 420 K peak, and -3 1 kcal/
2200 2100 2000 1900 1800 1700 1600
Frequency (cm-l) Figure 5. IRAS spectra of coadsorbed CO and 0 on Pd/silica/Mo( 110). The adsorption sequence was the same as that for Figure 4. The surface was annealed to the indicated temperatures and then cooled to 100 K before acquisition of the IRAS data.
mol for the 510 K peak. These activation energies are based on the peak temperatures utilizing the Redhead approximation.34 The broad range of the reaction activation energies and the COz formation temperatures is attributed to the variety of configurations of chemisorbed CO and 0.14J9 The 510 K peak is not observed in previous studies of CO/O coadsorption on single crystal palladium surfaces, thus this feature could correspond to reaction at the boundary of palladium and silica. Figure 5 shows IRAS spectra for coadsorbed CO and 0 as a function of surface temperature. At 100 K, CO exhibits a main IR absorption band at 21 31 cm-1, correspondingto atop CO, and several small features at 1890-1980 cm-1, which are attributed to the adsorption at 3-fold and 2-fold hollow sites. Upon increasing the surface temperature, the intensity of the atop adsorption decreases and the bridging adsorption intensity increases. In addition, the 2131-cm-I band splits into two bands at 2134 and 2098 cm-l following an anneal to 200 K. Within this temperature range, C02 product is formed. The CO stretch frequencyat 2 131cm-l for CO/O coadsorption is not observed for CO on clean palladium surfaces. A similar band at 2125 cm-I has been observed for CO and 0 coadsorption on Pd( 100)by high-resolution electron energy loss spectroscopy.14 This band is attributed to CO atop adsorption on the palladium atom that is directly bonded to an oxygen atom(s). The reaction of this intermixed carbon monoxide and oxygen to produce C02 occurs below 300 K, in agreement with earlier studies. Upon heating to >300 K, CO and oxygen form separate domains and CO is adsorbed onto the hollow and bridging sites (1800-2000 cm-l). The reaction between the CO and oxygen domains occurs at -420 K, with an activation energy of -25 kcal/m01,~~ in excellent agreement with values reported in the literature.16 3.4. CO Oxidation on Pd/SiOz at Low Pressure (lP7-10-6 Torr). Figure 6 shows the steady-state rate of CO oxidation on a silica-supportedpalladium catalyst as a function of pressure at reaction temperatures from 350 to 1000K. For thedata of Figure 6,the gas-phase CO and O2ratio was 2/1 and the total pressures were 2.2 X 1F-1.6 X 10-6 Torr. Decreasing the total pressure from 10-6 to 1 F Torr decreases the rate above 400 K but does not alter the overall behavior of the reaction rate with temperature. The data were collected continuously while heating the sample at a rate of 1 K/s. The reactions are at steady state since the data collected during sample cooling and heating are identical. Clearly, there are three temperature regions in Figure 6 of the steady-state rate data of CO oxidation on Pd/SiO2. At low
Xu and Goodman
7714 The Journal of Physical Chemistry, Vol. 97, No. 29, I993 PdISiO,
h
I
pco lP0.2-2
'total
1
0'04%
l
l
1
l
l
r
I
L
I
300 400 500 600 700 800 900 lo00
silicacatalyst temperature(350-1000 K)andthe totalpreawe( 1P-10-6 Torr). The C0/02 pressure ratio was 2. The oxidation rate, measured by the increase in the C02 partial pressure, was monitored with a quadrupole mass spectrometer while heating linearly at 1 K/s. PdlSIOp
h
P0~-5xlO'~Torr
I
I
1
400
300
I
I
I
I
600
500
I
I
700
800
Temperature (K) Figure 7. Steady-state CO oxidation rate as a function of CO partial pressure and temperature at constant 02 partial pressure. .C r
z
i
PdlSiopl
PCO-6x10~7Torr
I
300
I
400
I
I
500
1
1
800
1
I
700
I
I
I
I
2200 2100 2000 1900 1800 1700 1600
Temperature (K) Figure 6. Steady-stateCO oxidation rate as a function of the Pd(5 ML)/
.E a
Tor
Fon
1.6~10~ 48x10 -7 1.6~10 -7 7.2~104 22x1 0 4
l
Pco-1 xl O-'
Pop- 5 ~ 1 0 ~ ~ T o r
l
l
800
Temperature (K) Figure 8. Steady-state CO oxidation rate as a function of 02 partial pressure and temperature at constant CO partial pressure.
temperatures, the reaction rate increases as a function of temperature to a maximum at -500 K. The reaction rate is maximum and nearly constant between 500 and 600 K. At higher temperatures (>600 K), the reaction rate decreases with an increase in reaction temperature. The pattern of CO oxidation rateon Pd/Si02 at different temperaturesis in excellent agreement with that reported for palladium single crystal surfaces.35 Figures 7 and 8 show the steady-state rate of C02 production as a function of temperature at different CO and 02 partial pressures. At constant 0 2 pressure,the temperature of maximum
Frequency (cm -') Figure 9. IRAS spectra acquired during steady-stateCO oxidation on
Pd/silica/Mo( 110) as a function of temperature. The CO and 02partial pressures were 1 X 10-6 and 5 X lo-' Torr, respectively. rate (T-) increases with the carbon monoxide pressure (Figure 7). In addition, at constant temperatures below T,,, the CO oxidation rate is negative order with respect to the CO pressure. At temperatures above T-, the reaction rate is positive order with respect to the CO pressure. Within certain temperature and pressure ranges, the reaction order with respect to the CO pressure changes from negative to positive. In contrast, Tmudecreases with an increase in the 0 2 pressure at constant CO pressure (Figure 8). The CO oxidation reaction is first order in 0 2 pressure below TmX. At temperatures above T- and at a low Po,/P~oratio (I), the oxidation rate changes to negative order with respect to the 0 2 pressure. The CO coverageon the palladium surface is very small at and above T- based on in situ infrared reflection-absorption spectra (Figure 9). No detectable CO absorption bands were observed above 500 K during steady-state reaction at 1 X 10-6 Torr of CO and 0.5 X lod Torr of 0 2 . Below 500 K, CO adsorbs onto the hollow and bridged Pd sites, exhibiting stretching frequencies between 1830 and 1980 cm-1. At 350 K, CO exhibits bridgingadsorption bands at 1935 and 1980 cm-' and a small atopadsorption band at 2080 cm-1. At 450 K, only a small band at 1830 cm-' corresponding to adsorption onto 3-fold hollow sites is observed. Below 400 K, the I U S spectra for the steady-state CO oxidation were essentially identical to the CO adsorption spectra obtained at the same CO partial pressure (Figures 9 and 2). This suggeststhat the surface CO coverage below 400 K and for a particular temperature/pressure condition is the same with or without gas-phase 0 2 . However, above 450 K where the reaction rate is appreciable, the surface CO coverage is much smaller with gas-phase 0 2 than without. 3.5. CO Oxidation on Pd/SiO2 at High Pressure (15 Torr). Figure 10 shows the Arrhenius plot of the rate of CO oxidation on the model silica-supported palladium catalysts for three palladium coverages. The initial reaction pressures were 10.0 Torr for CO and 5.0 Torr for 0 2 ;the reaction temperatures varied between 540 and 625 K, and the conversion was less than 50%. The absolute rate was measured by the pressure decrease in the reactor of known volume (0.75 L). The number of reactive sites was determined by both carbon monoxide and oxygen temperature-programmed desorption.9 The reaction rate is expressed in terms of turnover frequency (TOF), i.e., product molecules per site per second.
CO Oxidation on Silica-Supported Palladium Catalysts Temperature (K)
600 I
.
I
550
500
450
I
I
1
The Journal of Physical Chemistry, Vol. 97, No. 29, 1993 7715 Pd (2 ML)/silica (100 A)
)
-
Pd/Si02 (100 A)/~o(ilO) Wd) 0.3ML
9.
8,
I
0"
0.1
0
1 1
1.50
b
Pd(ill)\Y . .
PJI 00) 5% Pd/SiO, .-.o I 1.70
1 1.90 l O O O j l (K-l)
I
1
2.10
2.30
Figure 10. Arrhenius plot of CO oxidation on palladium. The data for three palladium coverages, 0.3,0.5,and 2 ML are shown. The reaction rate (TOF) was measured at a total pressure of 15 Torr and a CO/Oz ratio of 2. The data for single crystal palladium and 5% Pd/silica were taken from refs 20 and 36.
I
5 cd 2.2 \
2s? 2.0 C
650 K
I
I
I
1
I
I
1
2200 2100 2000 1900 1800 1700 1600 Frequency (cm-')
Figure 12. IRAS spectra of adsorbed CO on Pd/silica/Mo( 110) as a function of the reaction temperature. The total pressure was 15 Torr, and the CO/Oz ratio was 2. The contribution from gas-phase CO has been subtracted.
I,
Pd(1 ML)/SI%(lW A)
h
.P
Temperature
IR bands have been observed for high surfacearea silica-supported palladium catalysts.18 Increasing the reaction temperature from 350 to 650 K decreases the intensity of the 1990-cm-1 band. The total absorbance of the CO bands decreases -40% from 350 to 650 K.
1.8 -
iij 1.6-
If 1.4 Y p, 1.2 *N
~~
0.4 0.6
0.8
1.0
Log (Po,For0
1.2
I ,
0.4
0.6
0.8
1.0
1.2
Log PcoFor0
Figure 11. CO oxidation rate as a function of 0 2 and CO partial pressure for Pd(1 ML)/silica/Mo(110) at 570 K.
The activation energies for CO oxidation on the model silicasupported palladium catalysts vary only slightly with the palladium coverage (i.e., particle size). For example, for a nominal palladium coverage greater than 2 monolayer (particle size >200 A), the activation energy is constant at 27 kcal/mol, whereas for 0.6 and 0.3 monolayers (-40 and 30 A), the activation energy is 25 and 23 kcal/mol, respectively. The reaction rate (TOF)does not vary significantly with palladium coverage for these conditions. For comparison, the activation energy for CO oxidation on Pd(loo), Pd(llO), and Pd(ll1) is 29.4, 30.7, and 28.1 kcal/mol, respectively, at similar reaction conditions.36 Figure 10 also shows the published results of CO oxidation on palladiumsingle crystals and supportedpalladium catalysts.13.20.36 The reaction rates on the model silica-supported palladium catalysts are in excellent agreement with the values found for Pd(100), Pd(llO), and Pd(ll1). The CO oxidation at these reaction conditions is first order in O2pressure and negative first order with respect to the CO pressure, in agreement with earlier studies on single crystal and supported palladium catalysts.13~14~18.*0.36 Figure 11 shows logarithmic plots of the initial reaction rate at 570 K versus (a) the O~pressure(2-10 Torr at a constant 10 Torr of CO) and (b) the CO pressure (5-20 Torr at a constant 10 Torr of 0 2 ) . Figure 12 shows in situ I U S spectra for 10.0 Torr of CO and 5.0 Torr of 02 at various reaction temperatures. The spectra were collected using 10.0 Torr of CO at a surface temperature of 1000 K as a reference spectrum. Data acquisition time was 15s to limit the reactionconversion. The spectral features indicate three bands at -1900, 1990, and 2117 cm-l, corresponding, respectively, to hollow, bridging, and atop CO adsorption. Similar
4. Discussion This is the first complete study of catalytic CO oxidation that covers both a wide temperature range (350-1000 K) and a wide pressure range ( W - 1 5 Torr) for supported palladium catalysts. Accordingly,important issues related to the so-called "pressure" and "material" discrepancies between model and "real-world" studies can be simultaneously investigated. Although the CO oxidation reaction is one of the simplest reactions catalyzed by a metal surface, the reaction kinetics still are rather complex. The CO oxidation rate increases with the surface temperature to a maximum and then gradually declines. Likewisethe reaction orders with respect to the CO and 0 2 pressuresvary with reaction conditions. In the followingsections, the kinetics and mechanism of CO oxidation will be discussed in detail. 4.1. CO Oxidation Kinetics. It is generally accepted that CO oxidation follows a Langmuir-Hinshelwood mechanism,l9 Le.,
O(a)
+ CO(a)
-
CO, k,
(4)
The surface reaction rate constant, k4, which depends strongly on the adsorbate coverage and the configuration of 0 and CO, is large for the configuration in which 0 and CO are adjacent to each other. The rate constant for CO desorption, k2, is independent of coverage at low CO coverages; however, for CO coverages larger than 0.5 monolayers, where the CO layer is compressed, kl increases dramatically. kl and k3 are the adsorption rate constants for CO and 02,respectively. The adsorption of 0 2 occurs via a precursor such that the adsorption rate is first order in vacant surface ~ i t e s . 3 ~The J ~ desorption of oxygen is negligible at temperatures below 800 K. The initial
Xu and Goodman
7716 The Journal of Physical Chemistry, Vol. 97, No. 29, 1993
sticking probability of O2 is strongly influenced by chemisorbed CO and is near zero on the fully covered CO surface.19 At steady state, the adsorbate coverage and the reaction rate are determined by the following:
e,
-con
C0+’/202 1.0,
Pqriiica
k2k3
=k4ff
0
rate = k4808co =
k2 k3P0z
k2
+ klPCO - k3P01
( 1 - 2 )
(7)
3
B: 0.9
where a is the difference in the adsorption rates of CO and 02. 0.0 500
550
800
650
700
I 750
800
Reaction Temperature (K) These equations are derived solely from the proposed reaction mechanism (1)-(4) and theassumption of steady-stateconditions. Clearly, the reaction rate depends on all of the kinetic parameters as well as the CO and 0 2 partial pressures. However, it is noteworthy that the rate constants are not only functions of temperature but also of adsorbate coverages as well. To simplify eq 7, the reaction conditions are taken into consideration and eq 7 is compared with the experimental data. At low-temperatureand high-pressureconditions, k2 > (klPco - k3P@),the steady-state CO coverage is essentially zero according to eq 6. This behavior is indeed observed experimentally. The reaction rate is determined by the following:
Therefore,thereaction rate at temperatures above T- but below the 02 desorption temperature (800 K) is determined by the oxygen coverage and the 0 2 pressure, whereas theoxygen coverage is determined by both the CO and the 0 2 pressures (eq 5). The decrease of the reaction rate above T-, Le., a negative apparent activation energy (Figures 6-8), is due to an accumulation of surface oxygen at the higher temperatures. The oxygen coverage increases because the CO residencetime is significantly shortened at the higher temperatures. Figure 7 shows that the reaction rate increases with CO pressure at constant 0 2 pressure in this elevated temperature regime. This is because a,the difference in the CO and 0 2 adsorption rates, increases with the CO pressure, resulting in a decrease in the oxygen coverage (4).
Figure 13. Steady-state surface oxygen coverage on Pd during CO
oxidation as a function of the surface temperature and the 02 partial pressure. Data were calculated from eq 10 and Figure 8. The kinetic behaviors in Figure 8 for constant CO pressure can be explained in a similar way. At low POJPCOratio, a is large and the oxygen coverage is low (Figure 8a and b), thus the reaction rate is nearly constant between 550 and 800 K. At the low oxygen coverage limit, the reaction rate is first order with respect to the 02 pressure. At high P@/Pcoratios, a is small and the surface oxygen coverage approaches saturation. The reaction rate, in turn, becomes negative order with respect to the 0 2 pressure (Figure 8d and e). This reaction order change from positive to negative has also been observed in the high-pressure CO oxidation.13 Above 800 K, the reaction rate is also influenced by 0 2 desorption such that the decrease in the oxidationrate within this temperature region is not as dramatic as that below 800 K. This is seen as a shoulder near 800 K in the plot of the rate versus temperature in Figure 6. 4.2. Steady-State CO and Oxygen Coveraga. The steadystate CO coverage is essentially zero at and above T, based on in situ infrared reflection-absorption spectral data and eq 6. Below T-, the CO coveragedepends on temperature and CO pressure and decreases with the reaction temperature. For the particular conditions, 10 Torr of CO and 5 Torr of 0 2 , the surface CO coverage is nearly constant below 450 K and decreases -40% from 450 to 650 K (Figure 12). The surface oxygen coverage at T, and above can be calculated from eq 10. The adsorption rate constant, k3, is measured by monitoring the reaction rate as a function of the 0 2 pressureat conditionswhere 80 -zero (low PoJPco ratio, Figure 8a). The 0 2 adsorptionis assumed to beconstant at temperatures above T, since, withinthis regime, the CO coverage is essentially zero. Figure 13 depicts the oxygen coverage as a function of the POJPCO ratio for the data of Figure 8. The oxygen coverage at TmXis approximately zero at PoJPm < 0.3 and near saturation at P0,lPco > 3. Theoxygen coverageincreaseswith temperature (below 800 K) for all cases. 4.3. The Transition State of CO Oxidation. Based on eq 5, a plot of ln(Oo) versus 1/ T should be linear and the slope and interceptyield thedifferencein the activation enthalpyand entropy of CO desorption (k2) and its reaction with oxygen (k4). This follows since the adsorption rate constants, kl and k3, are equal to the product of the flux and the initial sticking probability of CO and 02,respectively. The sticking probabilities are known to be 0.4 for oxygen on clean palladiuma and - 1 and -0.8 for CO on the clean and oxygen-covered palladium,’69@ respectively. Figure 14 shows the corresponding plots for two
CO Oxidation on Silica-Supported Palladium Catalysts
The Journal of Physical Chemistry, Vol. 97, No. 29, 1993 7717
Temperature (K) 700
750
-
650
600
O U
C
0
O C\ /O\
~~~’=12kcaVmol
B
A
0
I\
0
0
0
c m c X n 3 - m Adsorbates
Activated Complex
Figure 15. Schematicrepresentations for the proposed activated complex of CO oxidation at (A) low and (B) high oxygen coverage conditions.
-6 1.2
1
1.3
I
I
I
I
1.4
1.5
1.6
1.7
1.a
lOOO/T ( K - I ) Flgure 14. Steady-stateoxygencoverageversus the reciprocal temperature Torr and PO = 4.3 X for two pressure conditions: (A) PCO = 6 X Torr and (B) PCO = 6 X lo-’ Torr and PO, = 1.7 X lo-$Torr.
pressure conditions representing different surface oxygen coverages. Figure 14 clearly demonstrates that the activation enthalpy and entropy differences between CO desorption and CO reaction are dramatically influenced by the surface oxygen coverage. At very low oxygen coverages (00 I 0.02 of saturation), PAH* = AH2’ - AH4*= 17 kcal/mol and P A S = A&* - A&* = 26 cal/K.mol. In the oxygen coverage range of 00 = 0.02415, A M * = 12 kcal/mol and P A S = 18 cal/K.mol. In the medium oxygen coverage (00 = 0.15-0.6), A M * = 4 kcal/mol and P A S = 6 cal/K.mol. At high oxygen coverages (00 >0.6), the activation enthalpy and entropy for the reaction of CO with oxygen approach those for CO desorption. Similar activation enthalpies for CO desorption and reaction at high oxygen coverage are anticipated. The activation energies of both the CO desorption and reaction depend on the configuration of coadsorbed CO and 0, according to Figure 3 and previously reported work.14941 The activation energy for the reaction of the intermixed CO and 0 overlayer at high oxygen coverage is 10 kcal/mol. The TPD spectra of Figure 3 and the IRAS spectra of Figure 4 demonstrate that the atop-adsorbed CO on the oxygen-coveredsurface desorbs and reacts below 200 K. In addition, atop-adsorbed CO desorbs at -200 K from the clean Pd(ll1) ~ u r f a c esuggesting ,~ that the desorption energy of atop CO on the oxygen-covered Pd( 111) surface is less than 12 kcal/mol. Since oxygen is adsorbed at the bridging and 3-fold hollow sites, it is anticipated that CO adsorbs onto the atop palladium sites with adjacent oxygen atoms, Le., in a weakly bonded state, prior to reaction. The larger difference in activation energy for CO desorption and reaction at low oxygen coverages agrees with that of previous work, At these reaction conditions, the oxygen and CO coverages are low ( e l % ) . The CO desorption energy at the zero coverage limit is -36 kcal/mol on Pd( 11l)25340.42 and -39 kcal/mol on Pd( 100).24,43The reaction activation energy is -25 kcal/mol at low 8co and moderate 80.16 The difference is 11-14 kcal/mol. The entropy difference between CO desorption and reaction also demonstrates that the transition state of CO 0 C02 is strongly affected by the oxygen coverage. The approximately equal entropy for CO desorption and reaction strongly suggests that the activated complex for surface CO oxidation resembles more closely the product (gaseous C02) at high oxygen coverages. At low oxygen coverages, the relatively large difference in the activation entropy, Le., low activation entropy for CO 0
-
-
+
+
-
C02, suggests that the activated complex resembles more closely the reactants. We propose that the different adsorption configurations of CO and 0 at low and high oxygen coverages are responsible for the apparent differences in the transition states of CO oxidation at these two surface conditions (Figure 15). At low coverages, CO and 0 randomly occupy the bridging and/or hollow sites. Therefore, the transition state can be represented as shown in Figure 15a, where a CO molecule has diffused to the Pd atom to which the oxygen atom is adsorbed. The observed island formation of CO and 0 on palladium indicates a repulsive interaction between CO and 0 . 1 4 9 4 1 Subsequent to overcoming this repulsive interaction or barrier between CO and 0, the reaction proceeds exothermically. The entropy of this activated complex, however,is low since the complex resembles an adsorbed species (Figure 15a). On the other hand, at the high oxygen coveragelimit, CO and 0 are coadsorbed onto the same palladium atom and form a surface complex, e.g., O-Pd-COI4 or a weakly adsorbed CO species, a precursor to reaction. This adsorption configuration is characterized by a CO stretch frequency at 2130 cm-l. The transition state for the reaction of this species to form C02 is likely located along the reaction coordinate near the product such that the activated complex more closely resembles the C 0 2 molecule (Figure 15b). Furthermore, this adsorption configuration can also explain the fact that the C02 molecule is vibrationally hotter from higher oxygen coverage than from lower oxygen coverage, observed by Coulston, et a1.44 4.4. The’MaterialGap”. The kinetic behavior of COoxidation on Pd/SiO2 is parallel to that observed for Pd single crystals and Pd foils. The activation energy and turnover frequencies found for the model Pd/SiOz catalysts are also in the same range of thosereported for Pd singlecrystals (Figure 10). Inaddition, the turnover frequencieson the planar Pd/SiO2 catalysts of this study are comparable to those measured on high surface area catalysts at similar reaction conditions. Therefore, there is no apparent effect of support on CO oxidation catalyzed by palladium, Le., there is no significant material gap for this reaction system. 4.5. The “Pressure Gap”. The generic rate equation (7) adequately describesthe kinetic behavior of CO oxidation within the low- and high-pressure regimes. The empirical rate law of eq 9, where the rate of CO oxidation is proportional to P@/Pco, follows from the generic rate law (7). From this empirical rate law it is generally assumed that the surface is covered with CO and the rate-limiting step is the adsorption of 0 2 , 4 5 In order to obtain the empirical rate law (9), it is necessary to assume that the adsorption of 0 2 only requires one vacant surface site. This assumption has been shown to be valid in UHV experiments in that 0 2 is adsorbed via a precursor state.37~38 The empirical rate law (9) shows that the CO oxidation rate does not depend on the total pressure if the ratio of the CO and 0 2 partial pressures is constant. However, it is noteworthy that the rate law of eq 9 is only valid at those reaction conditions where the rate of CO desorption is much smaller than the rate of CO adsorption. Landry et ai. reported that the CO oxidation rate on palladium increases by a factor of 7.4 at 445 K for equimolar CO and O2 as the total pressure was increased from
7718 The Journal of Physical Chemistry, Vol. 97, No. 29, 1993 10-6 to 102 mbar.21 No explanation for this increase in rate was given.21 The data in Figure 6 also show that the oxidation rate at 445 K increases with an increase in the total pressure from 2.2 X lo” to 1.6 X 10-6Torr. This is simply because at low pressures, the CO desorption rate is competitive with the CO adsorption rate which, in turn, lowers the oxidation rate according to (7). Accordingly, the rate law and the apparent activation energies for the low-pressure or high-temperature regimes deviate from the empirical rate law of eq 9. There are several reports of CO oxidation at low temperatures (lo Torr) that show deviation from the empirical rate law of eq 9.15J8,20The reaction order with respect to CO gradually changes from -1 to -0.2 with a decrease in temperature ( 1 Torr). Thesetrends are not reproduced in the low-pressure (