J . Phys. Chem. 1987, 91, 33 10-33 15
3310
dry hydrophobic core is 47rb3/3 = 8379 A3.Using the dry monomer tail volume V, = 611 - 65 = 546 A3 found from the contrast variation measurement one gets the aggregation number No = 83791546 = 15.3.
I"
0.0
0.5
1.o
Jii=ii&
1.5
2.0
x 10'
Figure 7. A plot of ii as a function of square root of monomer mole fraction that forms micelles. A seems to be linearly dependent on ( X Xc,,,c)1/2 as suggested by the ladder model. A linear extrapolation gives n, = 15 f 1 at the critical micelle concentration which, according to the u, and uH values determined here, corresponds to a compact spherical micelle. monomers which form micelles (X,,,= X,is the mole fraction of the free monomer). This plot is suggested by a ladder model.16 All the data points seem to fall into a straight line. We therefore linearly extrapolated the data points to obtain the minimum aggregation number No = 15 f 1 at the cmc. Due to low intensity the particle structure factor is generally very difficult to measure by SANS near the cmc. That a minimum micelles of an aggregation number 15 f 1 is quite reasonable can be seen by a simple geometrical consideration. The volume of the spherical (16) Missel, P. J.; Mazer, N. A.; Benedek, G. B.; Young, C. Y . ;Carrery, M. C. J . Phys. Chem. 1980, 84, 1044.
V. Summary and Conclusion The anionic surfactant AOT forms normal micelle in D 2 0 up to 1.1 g/dL concentration at room temperature. By using the SANS technique we have determined the micellar structure and its growth in dilute regime. The minimum micelle formed near the cmc was found to have a compact spherical packing containing 15 AOT monomers with a radius b = 12.57 %, which is equal to the fully stretched hydrocarbon tail length. The thickness of the hydrophilic layer is about 5 %, (Figure 2) and there are about 22 water molecules associated with each head group. Due to the strong interactions among micelles we used the contrast variation method at finite Q to determine two molecular parameters which characterize the monomer in a micelle. These two values were found to agree with an independent measurement in the case of the reverse m i ~ e l l e . ' The ~ AOT micelle grows moderately from spherical to oblate spheroidal confirming our expectation that a surfactant with double alkyl chain of comparable tail lengths tends to form micelles having an oblate spheroidal shape. Finally, the micellar growth was found to depend on the monomer concentration similar to a relation suggested by the ladder m0de1.l~
Acknowledgment. We are grateful to Dr. B. Schoenborn and Dr. D. Schneider for the use of the small angle neutron spectrometer in Brookhaven National Laboratory. We appreciate their generous help in the course of the experiment. We also thank Mr. J. Sung for purifying the AOT used in this study. The research is partially supported by a grant from the National Science Foundation administered through the Center for Material Science and Engineering of MIT and Exxon Research and Engineering Company. Registry No. AOT, 577-1 1-7.
Temperature-Programmed Desorption of CO and COPfrom Pt/CeO,. for Lattice Oxygen in CO Oxidation
An Important Role
T. Jin, T. Okuhara, G. J. Mains,? and J. M. White* Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712- 1I67 (Received: December 1 , 1986)
Platinum on ceria (Pt/CeO,) surfaces was studied by temperature-programmeddesorption (TPD) of CO and C 0 2 . Significant conversions of%O to CO, in the TPD of CO and of CO, to CO in the TPD of CO, were observed. The interconversion of CO and CO, was significantly affected by pretreatment with 0, or H, at elevated temperatures. Hydrogen treatment at 773 K completely suppressed CO, formation from CO but enhanced the formation of CO from COz. Oxidation of the sample increased the conversion of CO to C 0 2 to almost 80% but suppressed the conversion of CO, to CO. A number of mechanisms are considered to explain the interconversion. The reversible reaction of CO adsorbed on Pt with a lattice oxygen (or, conversely, CO,, with an oxygen vacancy) in the environs of the Pt crystals is proposed. The abundance of lattice oxygen and lattice oxygen vacancies at the Pt/CeO, interface determines the CO/C02 TPD profiles. The mechanism is consistent with two experiments in which isotopes were employed, a C'*O TPD experiment and an experiment in which the surface was oxidized with '*02.
Introduction ceria: c e o 2 , is an effectivecatalyst for the oxidation of butane] despite the fact that the lattice oxygen is less labile than those of other lanthanide oxides., It is also used for precious metal
supports. For example, Pt/Ce02 has been used for the catalytic oxidation of ethylene3 and for the methanation of C a 4 Furthermore, CeO, is effective as an additive to enhance the catalytic activity of Pt/SiO, and Pt/A1203.5.6 Yao et al.5 reported that ~~
On sabbatical leave; Department of Chemistry, Oklahoma State University, Stillwater, OK 74078. *Author to whom correspondence should be addressed.
0022-3654/87/2091-3310$01,50/0
(1) Hattori, T.; Inoko, J.; Murakami, Y. J . Cutul. 1976, 42, 60 (2) Rosynek, M. P. Cutul. Reo.-Sci. Eng. 1977, 16, 111. (3) Mendelevici, L.;Steinberg, M. J. Curd. 1985, 93, 353. (4) Mendelevici, L.; Steinberg, M. J . Cutul. 1985, 96, 285.
0 1987 American Chemical Society
8
CeO, in Pt/Ce02-A1203 is able to store oxygen and that these materials are useful as a three-way catalyst to treat automotive exhaust.' Summers and Ausen6 attribute the increased catalytic activity of Pt/CeO, to the oxygen-donating property of CeO,. Rieck and Bells found the catalytic activity of Pd/Si02 for the The mechanism reaction of CO with of H2 oxygen to be storage, enhancedbelieved by the addition to occurof at CeOz. the
Yao, J. C.; Yao, Y . F. J . Cafal. 1984, 86, 254. Summers, J. C.; Ausen, S. J . Catal. 1979, 58, 131. Cho, B. K.;West, L. A. Ind. Eng. Chem. Fundam. 1986, 25, 158. Rieck, T. S.; Bell, A. T. J. Cafal. 1986, 99, 278. Metcalf, I. S.; Sundaresan, S . Chem. Eng. Sci. 1986, 41, 1109. (IO) Altman, E. I.; Gorte, R. J. Surf. Sci. 1986, 172, 71. (11) Zhu, Y.; Schmidt, L. D. Surf.Sei. 1983, 129, 107. (12) Wang, T.; Lee, C.; Schmidt, L. D. Surf. Sci. 1985, 163, 181. (13) Bain, F. T.; Jackson, S. D.; Thomson, S. T.; Webb, G.; Willocks, E. J . Chem. Soc., Faraday Trans. 1 1976, 72, 2516. (14) Foger, K.; Anderson, J. R. Appl. Surf.Sei. 1979, 2, 335. (15) Sahital, S . R.;Cohen, J. B.; Bumell, R.L.;Butt, J. B. J . Coral. 1977, (5) (6) (7) (8) (9)
50, 479. (16) Beck, D. D.; White, J. M. J . Phys. Chem. 1984, 88, 2764. (17) Kakuta, N.; White, J. M. J . Cafal. 1986, 97, 150.
A
6 .
I
,
5 ,:!.
+.
.T
4
e
.
2
h I
0
,--. \
5
\
,I'
-lit ,:
c 3
gas/noble metal/ceria interface, has been studied recently9 by using electrochemical methods. Despite these studies, the reactions of Pt on a ceria surface and the roles played by CeO, in the catalytic chemistry are not well understood. In the present study, TPD (temperature-programmed desorption) was used to elucidate the role of CeO, in the Pt/Ce02 system. There have been many studies of the TPD of C O from
Five samples were used in the present study as follows: CeO, (Alfa Products), 2.0% Pt/CeO2(473), 2.0% Pt/Ce0,(673), 10.0% Pt/Ce0,(673), and 2.0%Pt/Si02(673). (The number in parentheses is the temperature of calcination in air.) The CeO, sample was washed with 5% HC1 to remove Ca2+ and calcined in air at 800 K for 5 h. The platinized oxides were prepared by impregnating C e 0 2 and Si02(Cab-o-Sil) with an aqueous solution of chloroplatinic acid (H2PtC16, Engelhard Ind.) The resultant slurries were dried at 373 K overnight and calcined in air at 473 K for 1 day and at 673 K for 12 h, respectively. The surface areas, measured by the BET method using N2, were found to be 5 m2 g-' for CeO, and 3.6 m2 g-' for 2.0% Pt/Ceo2(473). The samples were heated under a vacuum for 1 h at 800 K before all measurements. The estimated average crystal size of Pt on C e 0 2 was 370 8, for 2.0% Pt/Ce0,(473) and 130 8, for 2.0% Pt/Ceo2(673) from the line width of the Pt X-ray diffraction peak. Measurements were not performed on the 2% Pt/Si02(673) sample; however, samples prepared in this manner typically have particle sizes near 100 A.I5 The dried powder was pressed into a 0.5 X 0.5 X 0.010 in. piece of Ta mesh, to which a thermocouple was attached, and set into an ultrahigh-vacuum system described previously.I6 The weights of the powder samples used were about 35 mg for the Pt/Ce02 and 7 mg for the Pt/Si02, respectively. The experiments (including pretreatment) were performed in a two-chamber (preparation and analysis) apparatus. The TPD procedure was the same as that described previo~s1y.l~ Prior to TPD, the sample was evacuated at 800 K until water desorption could not be detected by mass spectrometry. During this heating HCl gas evolved from the sample calcined at 473 K but not from those calcined at 673 K. When desired, the samples were further prereduced in lo4 Torr of Hz at 773 K for 5 min or preoxidized in lo4 Torr of O2 at 373 K for 5 min. The preoxidized samples were flash-heated to 800 K (2-4 K s-I) to remove chemisorbed oxygen prior to use. After the pretreatments,
l
'.,'.,I
I: -:i
\
I \
,: 1;
\
-
_----
_ _..._. .-..... ' . , . . . . . . . . . . . . . . ....,.-,,... \
8:
r
,
';j 4 0 1
b
c m
r
l
Temperature
I K
Figure 1. First TPD for C O adsorbed on Pt/CeO,. C O exposure was Torr at 200 K for 1 min. The sample was preevacuated 800 K for 1 h. (a) 2% Pt/CeO2(473), (b) 2% Pt/CeO2(673): CO, (---); CO,, (-); C O on Ce02, (-a).
LA
20
L I d _ . _ A i
0
'0
2
4
6
8
10
12
0
14
R u n Number o f C O TPD Figure 2. Changes in desorption amounts with run number for successive C O T P D experiments on 2% Pt/Ce02(673). The sample was preevacuated at 800 K for 1 h. Legend for left ordinate: open circles, CO; filled circles, CO,; open squares, CO COz. Legend for right ordinate: open triangles represent percent conversion of C O to C 0 2 .
+
C O or COz was introduced into the chamber at 200 K. Both oxygen and hydrogen were purified by passing through a liquid nitrogen trap. The TPD spectra were obtained by resistively heating the Ta mesh at a linear rate of 2 K s-l with a temperature programmer. AES spectra were taken at 673 K after the TPD. The C/Ce and Pt/Ce ratios were obtained from the AES peakto-peak heights at 272 eV (carbon), 661 eV (cerium), and 237 eV (platinum).
Results TPD of CO on Evacuated PtlCeO,. A blank (no dosing) TPD experiment using 2% Pt/Ce02(473) showed no desorption peaks up to 800 K. When CO ( lo4 Torr, 1 min) was dosed onto CeO,, 2.0% Pt/CeOZ(473),and 2.0% Pt/Ce0,(673) the TPD profiles shown in Figure 1 were obtained. The 2% Pt/Ce0,(473) gave a broad peak of C o at 415 K and a sharp peak of co2 at 490 K (Figure la). The peak area of C O was approximately 3 times larger than the CO, peak area for the sample calcined at 473 K. H ~ for the~ sample ~calcined~ at 673~K ( ~ ~ ib),, the i situation was reversed. The C O peak maximum was 410 K and a 3 times larger peak for C 0 2 desorbed at 490 K. For pure CeOz
~
~
3312 The Journal of Physical Chemistry, Vol. 91, No, 12, 1987
Jin et al.
TABLE I: Effects of Pretreatments on Desorption Amounts in CO or CO, TPD on Pt/Ce02"
TCb
2
473
3.5
2
673
6.6
10
673
4.5
2f
673
41.6
2
473
2
613
0
evacuationC CO2
co
% Pt
1.2 (26%) 11.0 (63%) 10.8 (71%) 0.0
0.07 (8%) 0.23 (1 5%) 0.0
(773)
oxidationd total
co
4.7
(a) CO TPD 3.2
17.6
3.1
co
COZ
total
1.4 (30%) 12.6 (75%)
4.6
8.8
16.7
25.0
0.0
35.5
reductione C02 0.5 (5%) 1.5 (6%)
total 9.3 26.5
15.3 41.6
0.83
0.90
1.30
1.53
33.5
(b) COZTPD 0.0 (0%) 0.0
0.50
0.50
0.86
0.86
(0%)
0.80
0.40 (33%) 0.97 (49%)
0.80
1.20
1.02
1.99
0.80
"The desorption amounts shown are in arbitrary units per unit weight sample. All data are those of the first TPD run. Adsorption conditions: CO, loJ Torr, 1 min at 200 K; CO,, 5 X 10" Torr, 1 min at 200 K. bCalcinationtemperature, K. cEvacuation at 800 K for 1 h. dOxidationby exposure to O2at 373 K for 5 min at lo4 Torr. FReductionby exposure to H2 at 773 K for 5 min at Torr. fPt/SiO,.
-
.c
?
1
40 I 100
101
a
P,
.--.
80 8
al
\
N
c
& A - +
1 60
8
40
,z c L
m
20
z
0
'm
40r
0
w U
0.0' 0
'
'
2
'
'
4
'
, '
j
6
I
8
Run Number o f CO T P D
Figure 3. (a) Relative amounts of desorption with CO TPD run number.
200
400
600
800
Temperature I K
Legend for left ordinate: open circles, CO; filled circles, CO,; open squares, CO + CO,. Legend for right ordinate: open triangles represent percent conversion of CO to CO,. (b) Changes of Pt/Ce and C/Pt Auger peak ratios with CO TPD run number on 10% Pt/Ce02(673): half-filled squares, Pt/Ce; half-filled circles, C/Ce.
Figure 4. First TPD for CO following pretreatment of 2% Pt/CeO2(673): CO, (---); CO, (-). (a) H2 treatment (lod Torr at 773 K for 5 min). (b) O2 treatment (lo4 Torr at 373 K for 5 min). The sample was flash-heated to 800 K just before TPD. CO exposure was lod Torr, 1 min, at 200 K.
only a CO desorption peak a t 250 K was found (Figure la,b, dotted lines); C 0 2was not detected. [The C O TPD from the 10% Pt/Ce02(673) sample, not shown, was very similar to that found for the 2% Pt/Ce0,(673) sample.] Hz, H 2 0 , and O2were not detected among the desorption products up to 800 K. Figure 2 shows the changes in the desorption yields with successive C O TPD (run number) for the 2% Pt/CeoZ(673) sample. The amount of C O desorbed increased whereas the amount of C 0 2 decreased with the TPD run number. The sum, C O + COz, increased with successive runs until run 9, where it appeared to level off. For the 2% Pt/Ce02(473) sample the changes in desorption yields upon successive runs (not shown) are correspondingly smaller but in the same direction. Figure 3 shows the effect of successive C O TPD treatments (run number) on the relative amount of C O and COz desorbed (Figure 3a) and on the AES peak ratios for Pt/Ce and C/Ce (Figure 3b) for the 10% Pt/Ce0,(673) sample. The changes in the desorption amounts with TPD run number are similar to those in Figure 2. However, the ratio of Pt/Ce and C/Ce remained basically unchanged with successive C O TPD treatments, indicating that carbon is not
deposited on the surface by this treatment. The C/Ce AES ratio for 10%Pt/CeO,(673) was very close to that found for pure CeOz, 0.2 to 0.25. CO TPD from Prereduced, Preoxidized, and Prehydrated P t / C e 0 2 . Parts a and b of Figure 4 show the TPD spectra observed for prereduced and preoxidized 2%Pt/Ce02(673), respectively. When the sample was prereduced in Torr of H2 at 773 K for 5 min, the amount of C 0 2 desorbed was effectively reduced to zero (Figure 4a). However, the CO, yield was recovered completely by the subsequent oxygen treatment at 373 K for 5 min, followed by a flash heating to 800 K (Figure 4b). The relationship between the amounts of desorbed gases and the number of successive TPD runs for the oxidized 2% Pt/Ceo2(673) sample is very close to that for a freshly evacuated sample (Figure 2). The desorption amounts for the C O TPD for Pt/Ce02 are summarized in Table I. Note that both the total amount of COz and C O and the conversion of CO to COz increased for the 2% Pt/Ce02 with an increase in the calcination temperature from 473 to 673 K. Note also that the prereduction of the 2% Pt/ CeO2(473) sample led to an increase in the total amount of
The Journal of Physical Chemistry, Vol. 91, No. 12, 1987 3313
TPD of C O and C 0 2 from Pt/CeOz
40 I
-
._ m
n
4 30
. %
I
.
10
c
Q
.i20
t
I
I
I
I 400 600 Temperature I K
Figure 5. TPD for CO: CO, (-- -); C 0 2 , (-). exposure lo4 Torr, 160 K, 1 min. 21
800
200
u)
600
800
Temperature / K
2% Pt/Si02(673),CO I
Figure 7. TPD for CO on 2% Pt/Ce02(673) after preoxidation with I8O2 (lo4 Torr, 373 K, 5 min): CI6O, (---); P O 2 , (-); C1802, (-);
C'60180, (-.-).
; 1'
f'
400
1 7 - 7
20
1
200
400
600
800
Temperature / K
Figure 8. TPD for CI8O on 2% Pt/Ce0,(673) after preoxidation with I6O2 Torr, 373 K, 5 min): CI8O, (---); P O , 0s.); P O 2 , (-);
C'60'80, (-.-); C'802, (-..-).
2
d '
I.\
IR absorption.I8 For 2%Pt/CeO2(673) which had been evacuated (Figure 6b), a small C02peak was observed at 540 K in addition to that at 250 K. Moreover, a desorption peak of CO appeared 200
400
600
800
Temperature / K
Figure 6. TPD for C 0 2 : CO, (---); C02(-). (a) Ce02,C02 exposure lo4 Torr, 200 K, 1 min. (b) 2% Pt/Ce02(673)preevacuated at 800 K for 1 h. (c) 2% Pt/Ce02(673) O2as in Figure 4. (d) 2% Pt/Ce0,(673) H2 as in Figure 4.
desorption but preoxidation did not. The effect of coadsorption of H20on the C O TPD profile was examined by predosing water (10" Torr, 300 K, 1 min) prior to the introduction of CO onto a 2% Pt/Ce02(473) sample. Although a water peak appeared at 580 K, with shoulders a t 532 and 555 K, no changes in the TPD profiles of C O and C 0 2 were observed. The water desorption profile was similar to that observed for water on pure Ce0,. CO TPD on 2% P t / S i 0 2 (673). Figure 5 shows the TPD spectra of CO on 2% Pt/Si02(673). A broad C O peak centered at 560 K was observed. However, no CO, desorption was detected. Preoxidation of 2% Pt/Si02(673) did not change the C O TPD profile from that shown in Figure 5 . The desorption yields in the C O TPD for P t / S i 0 2 are compared with P t / C e 0 2 in Table Ia. TPD of COzfrom Ce02and PtlCeO,. Figure 6 shows the C 0 2 TPD spectra for C e 0 2 and for 2% Pt/CeOz(673). Note that the ordinate of Figure 6 is much smaller than those used in Figure 1. For CeO, (Figure 6a), CO, desorption peaks were observed at 250 K and around 700 K but no CO desorption was detected. The broad desorption peak centered near 700 K is due to a slightly larger surface area for this sample (compared with the sample containing Pt) and to decomposition of carbonate, identified by
at this temperature. As shown in Figure 6cd, the C O peak from the C 0 2 TPD was suppressed when the sample was preoxidized with O2 but enhanced drastically when prereduced with H2. A similar behavior was observed for the 2% Pt/Ce02(473) sample. The desorption amounts for the CO, TPD experiments are summarized in Table Ib. The total desorption amounts in the CO, TPD were much smaller than those observed for the C O TPD experiments (Table Ia), irrespective of pretreatment. CO TPD with IsO2 and ClSO. two isotopic experiments were performed: (1) TPD of CI6O from Pt/CeO2(673) which had been prereduced and reoxidized in I8O2and (2) TPD of Cl80 (93.1% lsO) over a similar sample which had been prereduced and reoxidized in I6O2. In both cases, the sample was flash-heated to 800 K prior to TPD. The results are given in Figures 7 and 8, respectively. In the first experiment (Figure 7) more than 95% of the carbon dioxide desorbed was CI6O2. Less than 5% C'60180 and no CI8O2were found over the whole temperature range. In the second experiment (Figure 8) the height of isotopically substituted carbon dioxides is in the order CI6O2> C'60180> C'*Oz. Note that the onset of additional CI6O corresponds with the production of C 0 2 , showing a desorption maximum at 494 K with the CO,. The distribution of isotopically substituted products in the CI8O TPD experiments is given in Table 11. The isotopic C 0 2 equilibrium values, shown in square brackets in Table 11, were calculated by using the observed abundances of I6O and lSO in the carbon dioxides. The calculated equilibrium percentages of C 0 2 isotopes are very close to that observed throughout the TPD, (18) Jin, T.; Zhou, Y.; Mains, G. J.; White, J. M., submitted for publication in J . Phys. Ckem.
3314
The Journal of Physical Chemistry, Vol. 91, No. 12, 1987
TABLE II: Isotopic Distribution of Products in C'*O TPD over 2% Pt/Ce0,(673) monoxides dioxides temp, K CI60 CI80 CI6O, C160180 C ' * 0 2 380 0.45" 5.35 (7.8)b (92.2) 410 0.61 5.98 1.71 1.30 0.32 (9.3) (90.7) (53.3) (40.5) (6.2) [54.0]' [39.0] [7.0] 470 1.40 3.13 5.30 3.83 0.58 (30.9) (69.1) (54.6) (39.4) (5.9) [55.3] [38.1 [6.6] 500 1.08 0.74 9.90 4.35 0.48 (40.6) (59.4) (67.2) (29.5) (3.3) [67.2] [29.5] [3.3] 560 0.17 0.00 4.40 0.75 0.00 ( 100) (0.0) (85.4) (14.6) (0.0) I85.91 [13.6] [0.5]
"The naked numbers are the relative mass intensities. Parenthetic numbers are percentages of the oxide or dioxide. 'Bracketed numbers are calculated equilibrium compositions for carbon dioxide; see text. suggesting that isotopic equilibration, not unexpectedly, is rapid compared with desorption.
Discussion Mechanism of C02Formation. The broad C O desorption peak with a maximum near 410 K, shown in Figure 1, suggests that C O is bound to this surface at a rich variety of sites, reflecting both an inhomogeneity of particle sizes and adsorption sites. The strength of chemisorbed CO on Pt is known to vary with the nature of crystal siteIe2' and depends upon the extent to which the surface has been reconstructed.22 The TPD desorption profiles of CO, as broad as that in Figure 1, have been reported for supported Pt catalysts, Pt/AI2O3and Pt/SiO2.I4 The increase in the amount of CO adsorbed when the calcination temperature is increased from 473 to 673 K (Figure l a ) is consistent with an increase in the dispersion of metallic platinum produced. The increase in the C O adsorption upon successive C O TPD (Figure 2) strongly suggests that these treatments are reconstructing the Pt surface to produce more sites capable of adsorbing CO. Schmidt and co-workers have shown that supported Pt changes crystal orientations when heated in H2, An interesting feature of these experiments is the formation of a large amount of C 0 2 in the TPD of C O (Figure 1). A number of mechanisms can be proposed, e.g. CO(a)
+ H20(a)
-
CO(a)
+ OH
C02
- + - + + + - +
2CO(a) CO(a) CO(a)
CO,
I/,H2
C02
O(a)
OL
+ H2
(la) (lb)
C
(2)
CO,
(3)
CO,
V
(4)
where H,O(a), O(a), OL, and V represent adsorbed water, adsorbed oxygen atoms on the Pt, lattice oxygen, and a lattice oxygen vacancy on the CeO, surface. Since it is always difficult to completely dehydrate metal oxides, reactions l a and l b must be considered. Indeed, for Pt/SiO, preadsorbed water reacts with the C O to produce CO2.I4 However, in the present study, preadsorbed water on P t / C e 0 2 did not influence the formation of CO,. This indicates that reaction 1 (19) Ogletree, D. F.; VanHove, M. A,; Somorjai, G. A. Surf. Sci. 1986, 173, 351. (20) Lang, J. F.; Masel, R. J. SurJ Sci. 1986, 167, 261. (21) Freyer, N.; Kriskinova, M.; Pirug, G.; Bonzel, H. P. Appl. Phys. A 1986, 39, 209. (22) Moller, P.; Wetzel, K.; Eiswirth, M.; Ertl, G. J . Chem. Phys. 1986, 85, 5328. (23) Wang, T.; Schmidt, L. D. J . Catal. 1981, 70, 187.
Jin et al. is not the important contributor to C02formation in the Pt/CeO, system. The fact that H2 treatment completely suppressed the formation of C 0 2 (Figure 4a) supports this idea because surface reduction with H2 would be expected to increase the number of OH groups, the adsorbed water, and/or lattice vacancies on the surface. Reaction 2, the disproportionation of CO into C 0 2 and C, Le., the Boudouard reaction, was given special consideration since Rieck and Bell* concluded that C 0 2 produced on C e 0 2 promoted Pd/Si02 came from this source. The Boudouard reaction is expected to deposit an amount of carbon stoichiometrically equivalent to the C 0 2 produced and leads to poisoning of the Pt surface in successive CO TPD runs. Indeed, we observe a decrease in the C 0 2 yield with successive CO TPD treatments. However, the AES results (Figure 3b) show that no carbon is deposited on the Pt surface despite eight cycles of C O TPD. It is well-known that carbon on metal surfaces (e.g., Ni) reacts with H 2 to form hydrocarbon^.^^ Thus, if C on the Pt surface was responsible for the reduction in C 0 2 yield, one might expect H2 treatment to restore the activity. Quite the contrary is observed; H, treatment eliminated C 0 2formation (Figure 4a). Therefore, we can exclude reaction 2 as an important mechanism. This conclusion is in agreement with studies of the TPD of CO from bulk Pt, where no C 0 2 formation is o b ~ e r v e dand , ~ ~with ~ ~the ~ absence of C 0 2 in the TPD of C O from Pt/Si02 shown in Figure 5b. Caution must be exercised here; the absence of reaction 2 may depend strongly on the sample preparati~n.~' Reaction 3 requires the presence of atomic oxygen on the Pt. O(a) desorbs from Pt as 0, with a desorption peak maximum between 735 and 800 K.28,29 The presence of chemisorbed oxygen on these Pt surfaces is considered unlikely since the Pt/Ce02 surfaces had been evacuated at 800 K for 1 h. Suggestions that residual oxidized forms of Pt (PtO, PtO,, etc.) not completely removed by the evacuation treatment contributes to C 0 2formation may also be rejected as a comparable amount of C 0 2 would then be expected for Pt/Si02. The formation of C 0 2 was not observed for either the evacuated sample (Figure 6) or the oxidized Pt/SiO,. Actually, crystalline metallic Pt was detected in the XRD measurement of the Pt/CeO, evacuated at 800 K for l h and no evidence for PtO and PtO, was found. PtO formed during oxidation'O could play a role in determining the state of this active oxygen pool in these experiments. The decomposition of PtO during flash heating provides a mechanism of transferring lattice oxygen to the vapor phase. In addition, the total amount of CO, formed during the 14 TPD runs shown in Figure 2 is about 6 times the total amount of CO adsorbed in a single TPD run. (Actually, the surface was still capable of forming C 0 2 from CO after the 14th run.) If O(a) on Pt is reponsible for C 0 2 formation from CO, the fraction of surface covered with O(a) would have to be at least 6 times that covered with CO. Consequently, the total desorption (CO + CO,) would be expected to increase about sixfold (based on run 14) with successive TPD runs as O(a) is consumed and more Pt is available for CO. However, this is not observed. Finally, the results of the isotopic experiments also exclude a contribution from atomic oxygen on the surface. Oxygen isotope exchange is observed in these experiments. As shown in Table 11, within COz the exchange almost reached an equilibrium distribution. However, if O(a) on Pt is responsible for the observed oxidation of C O and if the isotope exchange between O(a) on Pt and OLof C e 0 2 is neglected, the same distribution of C 0 2 should have been observed because the fraction of lSOin the intermediate for the formation of CO, should be the same in both isotope experiments. But, the observed ratios of C160180/C1602 desorbed (24) Araki, M.; Ponec, V. J . Catal. 1976, 44, 439. (25) McClellan, M. R.; Gland, J. L.; McFeelay, F. R. Surf Sci. 1981, 112, 63. (26) Collins, D. M.; Spicer, W. E. Surf. Sci. 1977, 69, 85. (27) Daniel, D. S. Ph. Dissertation, University of Texas, 1984. (28) Steininger, H.; Lehwald, S.; Ibach, H. Surf. Sci. 1982, 123, 1 , (29) Gland, J. L.; Sexton, B. A.; Fisher, G. B. Surf. Sci. 1980, 95, 587. (30) Fleisch, T.; Mains, G. J. J . Phys. Chem. 1986, 90, 5317.
TPD of C O and C02 from Pt/CeO,
The Journal of Physical Chemistry, Vol. 91, No. 12, 1987 3315
TABLE 111: Desorption Amounts of CO and C02 and Dispersion of Pt sample particle size dispersion total C O adsorbedb perimeter C 0 2 formedb
2% Pt/ CeOA673)
2% Pt/ CeOd473)
130 %, 6.6% 16.7 2201c 1709d 11.0
370 A 2.2% 4.6 272' 21 3d 1.2
ratio' 3 .O 3.6 8.1 9.2
"Column 1 /column 2. *Relative amounts obtained from TPD peak areas. Assuming cubic particles, relative units. dAssuming hemispherical particles, relative units.
are very different in Figure 7 (1:19) and in Figure 8 (1:2). Thus, reaction 3 is ruled out. The smaller incorporation of lSOinto C 0 2 in Figure 7 as compared with Figure 8 can be explained by our reaction 4. That is, the heavy oxygen, 180,is stored as lattice oxygen and is exchanged readily with the huge amount of I6OL during the flash heating, prior to the CI6O TPD. On the basis of the previous discussion, we conclude that reaction 4,the reaction of adsorbed C O with a lattice oxygen atom producing carbon dioxide and a surface lattice vacancy, is responsible for the observations. Formation of CO from CO,. As shown in Figure 6 and summarized in Table Ib, a significant amount of CO was desorbed in the CO, TPD. The amount of CO was strongly dependent upon the pretreatments, disappearing altogether when the surface was oxidized and appearing enhanced when the surface was reduced. In view of the fact that in some instances over 50% of desorbed C 0 2 appeared as C O while in others no C O desorbed, these observations are hardly explainable in terms of C O impurities in the C 0 2 . A conversion mechanism must be operative. It is claimed that clean noble metal surfaces do not adsorb C 0 2 dissociatively at this t e m p e r a t ~ r e however, ;~~ dissociative adsorption of C 0 2 on Pt/Ti02j2 and on R h / A 1 2 0 ~ 3 ~a3t 4room temperature has been confirmed by IR. CeO2 adsorbs C02to form carbonateI8 but does not form C O (Figure 6a). The increase of CO yield by hydrogen pretreatment supports the view that a lattice vacancy plays an important role in C 0 2 dissociation. The yield of CO depends upon both the presence of Pt and the lattice vacancy. The results of the experiments reported here can be understood in terms of the reverse of reaction 4. Active Sites for the Interconversion of CO and C 0 2 . The importance of lattice oxygen and/or lattice vacancy and the requirement of Pt for the interconversion of C O and C 0 2 (Figures 1 and 6) lead us to propose that the reaction occurs at the interface between Pt and CeO,. If the conversion occurs at this interface, as suggested, the amount of C O converted to C 0 2 is expected to be proportional to the circumference of the particles. The total C O adsorbed and the amount of COz formed for the 2%Pt/ Ce0,(673) and the 2%Pt/Ce02(473) samples are compared in Table 111. The circumference was estimated from particle diameters, based upon the observed X-ray crystal sizes, assuming (31) Campbell, C. T.; White, J. M. J. Cafal. 1978, 54, 289. (32) Tanaka, K.; White, J. M. J. Phys. Chem. 1982, 86, 3977. (33) Izuka, T.; Tanaka, Y. J . Caral. 1981, 70, 449. (34) Izuka, T.; Tanaka, Y.; Tanabe, K. J . Caral. 1982, 76, 1.
hemispherical particles. Although the model is crude and the particle size not uniform, Table 111 shows that the ratio of the average particle circumference and the ratio of CO, desorbed are in reasonable agreement, supporting this analysis. Yao et al.s suggested, based upon their temperature-programmed reduction (TPR) experiments, that there are two kinds of lattice oxygen which can be removed by reaction with Hz at 773 and 1000 K, respectively. The replenishment of surface lattice oxygen from the bulk is controlled by diffusion at 1000 K. In the present case there is a pool of active oxygen which is more than 6 times larger than the amount of C O adsorbed on the Pt surface. This pool of active oxygen can be depleted by H2 treatment at 773 K and replenished by O2 treatment at 373 K, reversibly. It is the condition (fullness) of this active oxygen pool which reflects the oxidation state of the cerium in the environs of the Pt particles and determines the extent of the interconversion of CO and COP Infrared measurements in this laboratory confirm absorption in the I R band of C O when C 0 2 is adsorbed on Pt/ CeO2.I8 The CO IR peak produced in the C 0 2 adsorption is attributed to C O on Pt and is small compared to that observed when C O is absorbed on this system. This is probably due to a limited number of oxygen vacancies at the interface which abstract oxygen from CO,. On the basis of these observations, we conclude the reaction between C O adsorbed on Pt and the nearby lattice oxygen on CeO, occurs readily at temperatures above 400 K. We propose the following model, which involves the movement of C O across the Pt to the CeO,. We propose that the interface takes part in the reaction by being reduced or oxidized during the TPD of C O or CO,: 0
111 U
0 II
0
I
u
II W
\;+;; ;/
II 0
Ill
111
/
U
Pt
II
0
-.i 111
\b-
Ce02 Conclusions 1. In the TPD of C O from Pt/CeO,, lattice oxygen at the interface between Pt and Ce0, plays an important role in the formation of COz. 2. Reduction with H2 and oxidation with 0, have profound effects on the formation of CO, in the TPD of CO from Pt/CeO2, indicating that the oxidation state of Ce at the interface is changed readily by these treatments. 3. The formation of C O in the TPD of CO, for Pt/CeO, appears to be the reverse process of C 0 2 formation in the TPD of CO. C 0 2 donates oxygen to a lattice vacancy at the interface to produce C O adsorbed on the Pt. Acknowledgment. This research was supported by the Office of Naval Research. Registry No. Pt, 7440-06-4; Ce02, 1306-38-3; CO, 630-08-0; C 0 2 , 124-38-9.
