- Email: [email protected]
Al2O3
Ouczi, L ef al. (Edttors), New Fronfiers in Caflrlysk Proceeding8 of the 10th International Congress on Catalysis, 19-24 July, 1992,Budapest, Hungary 0 1993 Elsevier Science Publishera B.V. All rights resewed
TRANSIENT RESPONSE STUDY OF THE OXIDATION AND HYDROGENATION OF CARBON MONOXIDE ADSORBED ON Pd/A1203 G. Kadinov, S.Todorova and A. Palazov Institute of Kinetics and Catalysis, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria
Abet ract IR epectroecopy was applied t o study carbon monoxide oxidation and hydrogenation over a 6% Pd/AlzOs catalyat. Upon oxidation at 338K the l’ieland’l mechanism operatee, however, the etructure of the CO adlayer in the “islandsttis not homogeneoue. In contraet, during hydrogenation at 433K a continuous rearrangement of the CO moleculee occure.
1. INTRODUCTION In our recent study of the oxidation of CO over supported palladium, the formation and interconversion of different adeor bate phases wae auppoeed to explain the obeerved oacillatione in the rate of the reaction and in the CO coverage [ l ] . It wae ehown further that the hydrogenation of CO on palladium takee place on a small fraction of the metal surface 121. IR spectroscopy applied t o a traneient response atudy, under condition8 when both the oxidation and hydrogenation proceed via the Langmuir-Hinechelwood mechanism, could supply additional information about the interaction between coadeorbed speciea and the atructure of adsorbed CO layer during the reaction.
2. EXPERIMENTAL One and the same sample of a 6% Pd/AlaOs catalyet wae ueed in both the oxidation and hydrogenation studies. Palladium dispersion was measured by CO chemisorption (applying the procedure described elaewhere 121) and varied between D*0,20 in the hy drogenation experiments and Dm0,18 in those of oxidation. Oxidation waa carried out at 338K with an initial feed mixture of CO:Oz:Ark2:2:238, whereae hydrogenation was performed at 433K with CO:Hz=l:240. The total flow rate was 240 nl/min in both
281 8 studies. "Matheson" gasflow meter was used for preparing the gas mixtures. Spectra were recorded on a "Specord 75 IR" Carl Zeiss instrument. More details on the IR cell-reactor, sample preparation, gas purification, reduction procedure and cleaning of the metal surface between the separate measurements are given in Ref. 2 and 3. IR spectra were recorded periodically after stopping the CO flow in the inlet mixture of CO-02-Ar or CO-Hz. The time dependence of the absorbance at definite wavenumbers, the band maxima, was also followed after stopping the CO flow. Since some desorption of CO and changes in the band positions and intensities were possible, both sets of experiments were also carried out in absence of oxygen or hydrogen by replacing the respective flow with argon.
3. RESULTS The CO bands, initially present in the spectrum, did not practically change their position during the oxidation process (Fig. 1 ) until the total amount of adsorbed CO was consumed. The band at 1930 cm-I decreased its intensity faster and broadened after stopping the CO flow. A shoulder at 1860 cm-' and a wide plateau appeared which developed into a broad band with maximum at about 1900 cm-1 when the amount of adsorbed CO was drastically decreased.
C
C
0 .c
.-0 c
n L
0 L
0
0
II)
Lo
* n
n
a
1800
2000
1800
c rn-1
Figure 1. IR spectra of 5% Pd/Alz03 at Xth minute after stopping the CO flow in the feed mixture of C0:Oz:Ar (2:1,4:236), T=338K
z
2000
cm-'
Figure 2. IR spectra of 6% Pd/AlzOs at Xth minute after stopping the CO flow in the feed mixture of CO:Hz (1:240), T=433K
281 9 Totally different behaviour of the bands was registered during the CO hydrogenation (Fig. 2). All the bandachaneed their position and intensity in a way which resembled that of CO desorption in vacuum. For lack of space those desorption spectra are not presented here, but the typical changes were presented in another paper of ours [41. Well known features of the desorp tion spectra are the continuous "red" shift of all the bands, the faster decrease in intensity of the band at 2070 cm-1 and the higher intensity of the band at about 1900 cm-1 till very low coverages. Upon oxidation the time dependence of the bands below 2000 cm-1 exhibited an "S" form with a sharp decrease in intensity (Fig. 3), whereas the band at 2077 cm-' decreased gradually. In contrast, some increase in absorbance of the 1860 cm-1 band was observed during the initial period. The time dependence of the band maxima was more difficult to follow during the CO hydrogenation because of the continuous red shift. In addition, desorp tion was more pronounced with the higher temperature. Nevertheless, the comparison between the decrease of absorbance at defi nite wave numbers in hydrogen and argon flow (Fig. 2 and 4 ) allowed some conlusions about the relative reactivity of certain CO species.
I
I
\b
2
4
6
8
1012
10
min
Figure 3. Time dependence of the relative absorbance at definite wavenumbers after stopping the CO flow in the feed mixture of CO:Oz:Ar=2:2:236, T=338K. a. 2175 cm-' (gas phase CO) b. 2070 cm-' c. 1967 cm-1 d. 1930 cm-1 e. 1860 cm-1
20
10
20
m in
Figure 4. Time dependence of the relative absorbance at definite wavenumbers after stopping the CO flow in the feed mixture of CO:Hz=1:240 (solid line), T=453K or CO:Ar=1:240 (dashed line), T=458K. a. 2175 cm-1 (gas phase CO) b. 2050 c m - 1 c. 1950 cm-1 d. 1925 cm-1 e. 1860 cm-1
2820
4. DISCUSSION The spectral features observed during CO oxidation (Fig. l ) , conform entirely with the llisland"mechanism of CO oxidation on supported palladium [ l ] , when the reaction proceeds at the interphase boundaries. A new feature was the unexpected broadening of the band at 1930 cm-1 and the appearance of the shoulder at 1860 cm-'.A band in the region 1820-1860 cm-1 appeared after adsorption of CO on supported palladium catalysts of low diaper sion at coverages lower than 0.3 [1,3]. This band ha8 been assigned to three-fold coordinated species on the (111) planes. At the same coverage the band assigned to bridged species on the (100) planes was situated in the region 1900-1920 cm-1. The broad band below 1930 cm-1 observed in the present study could be explained as a result of overlapping of three bande: 1860, 1920 and 1930 cm-1. Since the latter is due to bridged species on the {ill) planes at high coverages 31, the appearance of the former two bands implies that during oxidation some rearrangement occurs in the adsorbed layer. We euppose that the an "islands" are not homogeneous and consist of two regions inner part o f high local CO coverage on the (111) and {loo) planes, giving rise to the bands 1930 cm-1 and 1962 cm-1, respectively, and an outer part (at the boundaries), where the species are arranged in a way similar to the came of low coverage, giving rise to the bands at 1860 cm-1 and 1920 cm-'. The time dependence of the absorbance at 1930 cm-1 and 1962 cm-1 revealed a limit in the CO coverage below which the oxidation proceeded at a higher rate (Fig. 3). Further, the changes in the initial period confirm the proposed peculiarity o f the linear species on supported palladium [ l ] . This species desorbs faster from the (111) planes when the partial pressure of CO in the system decreases and the reaction initially occurs on the same planes. The spectral changes during the CO hydrogenation correspond to a continuous and steady decrease in coverage (Fig. 2 and 4), however, a rearrangement in the adlayer occurs and the rolecules are more uniformly distributed on the entire palladium surface. A comparison between the time dependence of the abmorbance at certain wavenumbers (Fig. 2 and 4) implies that CO adsorbed on the (100) planes was probably hydrogenated at a higher rate.
-
5.
REFERENCES
1
A. Palazov, 0 . Kadinov, Ch. Bonev, R. Dimitrova, D. Shopov, Surf. Sci., 226 (1990) 21 0 . Kadinov, Ch. Bonev, S. Todorova, A. Palaaov, 0 . Liete, J. Volter, in press A . Palazov, 0 . Kadinov, Ch. Bonev, I). Shopov, J. Catal,, 74 (1982) 44 A. Palaeov, 0 . Kadinov, Ch. Bonev, D. Shopov, Commun. Dept. Chem. Bulg. Acad. Sci., 1 1 (1976) 786
2
3 4
TRANSIENT RESPONSE STUDY OF THE OXIDATION AND HYDROGENATION OF CARBON MONOXIDE ADSORBED ON Pd/A1203 G. Kadinov, S.Todorova and A. Palazov Institute of Kinetics and Catalysis, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria
Abet ract IR epectroecopy was applied t o study carbon monoxide oxidation and hydrogenation over a 6% Pd/AlzOs catalyat. Upon oxidation at 338K the l’ieland’l mechanism operatee, however, the etructure of the CO adlayer in the “islandsttis not homogeneoue. In contraet, during hydrogenation at 433K a continuous rearrangement of the CO moleculee occure.
1. INTRODUCTION In our recent study of the oxidation of CO over supported palladium, the formation and interconversion of different adeor bate phases wae auppoeed to explain the obeerved oacillatione in the rate of the reaction and in the CO coverage [ l ] . It wae ehown further that the hydrogenation of CO on palladium takee place on a small fraction of the metal surface 121. IR spectroscopy applied t o a traneient response atudy, under condition8 when both the oxidation and hydrogenation proceed via the Langmuir-Hinechelwood mechanism, could supply additional information about the interaction between coadeorbed speciea and the atructure of adsorbed CO layer during the reaction.
2. EXPERIMENTAL One and the same sample of a 6% Pd/AlaOs catalyet wae ueed in both the oxidation and hydrogenation studies. Palladium dispersion was measured by CO chemisorption (applying the procedure described elaewhere 121) and varied between D*0,20 in the hy drogenation experiments and Dm0,18 in those of oxidation. Oxidation waa carried out at 338K with an initial feed mixture of CO:Oz:Ark2:2:238, whereae hydrogenation was performed at 433K with CO:Hz=l:240. The total flow rate was 240 nl/min in both
281 8 studies. "Matheson" gasflow meter was used for preparing the gas mixtures. Spectra were recorded on a "Specord 75 IR" Carl Zeiss instrument. More details on the IR cell-reactor, sample preparation, gas purification, reduction procedure and cleaning of the metal surface between the separate measurements are given in Ref. 2 and 3. IR spectra were recorded periodically after stopping the CO flow in the inlet mixture of CO-02-Ar or CO-Hz. The time dependence of the absorbance at definite wavenumbers, the band maxima, was also followed after stopping the CO flow. Since some desorption of CO and changes in the band positions and intensities were possible, both sets of experiments were also carried out in absence of oxygen or hydrogen by replacing the respective flow with argon.
3. RESULTS The CO bands, initially present in the spectrum, did not practically change their position during the oxidation process (Fig. 1 ) until the total amount of adsorbed CO was consumed. The band at 1930 cm-I decreased its intensity faster and broadened after stopping the CO flow. A shoulder at 1860 cm-' and a wide plateau appeared which developed into a broad band with maximum at about 1900 cm-1 when the amount of adsorbed CO was drastically decreased.
C
C
0 .c
.-0 c
n L
0 L
0
0
II)
Lo
* n
n
a
1800
2000
1800
c rn-1
Figure 1. IR spectra of 5% Pd/Alz03 at Xth minute after stopping the CO flow in the feed mixture of C0:Oz:Ar (2:1,4:236), T=338K
z
2000
cm-'
Figure 2. IR spectra of 6% Pd/AlzOs at Xth minute after stopping the CO flow in the feed mixture of CO:Hz (1:240), T=433K
281 9 Totally different behaviour of the bands was registered during the CO hydrogenation (Fig. 2). All the bandachaneed their position and intensity in a way which resembled that of CO desorption in vacuum. For lack of space those desorption spectra are not presented here, but the typical changes were presented in another paper of ours [41. Well known features of the desorp tion spectra are the continuous "red" shift of all the bands, the faster decrease in intensity of the band at 2070 cm-1 and the higher intensity of the band at about 1900 cm-1 till very low coverages. Upon oxidation the time dependence of the bands below 2000 cm-1 exhibited an "S" form with a sharp decrease in intensity (Fig. 3), whereas the band at 2077 cm-' decreased gradually. In contrast, some increase in absorbance of the 1860 cm-1 band was observed during the initial period. The time dependence of the band maxima was more difficult to follow during the CO hydrogenation because of the continuous red shift. In addition, desorp tion was more pronounced with the higher temperature. Nevertheless, the comparison between the decrease of absorbance at defi nite wave numbers in hydrogen and argon flow (Fig. 2 and 4 ) allowed some conlusions about the relative reactivity of certain CO species.
I
I
\b
2
4
6
8
1012
10
min
Figure 3. Time dependence of the relative absorbance at definite wavenumbers after stopping the CO flow in the feed mixture of CO:Oz:Ar=2:2:236, T=338K. a. 2175 cm-' (gas phase CO) b. 2070 cm-' c. 1967 cm-1 d. 1930 cm-1 e. 1860 cm-1
20
10
20
m in
Figure 4. Time dependence of the relative absorbance at definite wavenumbers after stopping the CO flow in the feed mixture of CO:Hz=1:240 (solid line), T=453K or CO:Ar=1:240 (dashed line), T=458K. a. 2175 cm-1 (gas phase CO) b. 2050 c m - 1 c. 1950 cm-1 d. 1925 cm-1 e. 1860 cm-1
2820
4. DISCUSSION The spectral features observed during CO oxidation (Fig. l ) , conform entirely with the llisland"mechanism of CO oxidation on supported palladium [ l ] , when the reaction proceeds at the interphase boundaries. A new feature was the unexpected broadening of the band at 1930 cm-1 and the appearance of the shoulder at 1860 cm-'.A band in the region 1820-1860 cm-1 appeared after adsorption of CO on supported palladium catalysts of low diaper sion at coverages lower than 0.3 [1,3]. This band ha8 been assigned to three-fold coordinated species on the (111) planes. At the same coverage the band assigned to bridged species on the (100) planes was situated in the region 1900-1920 cm-1. The broad band below 1930 cm-1 observed in the present study could be explained as a result of overlapping of three bande: 1860, 1920 and 1930 cm-1. Since the latter is due to bridged species on the {ill) planes at high coverages 31, the appearance of the former two bands implies that during oxidation some rearrangement occurs in the adsorbed layer. We euppose that the an "islands" are not homogeneous and consist of two regions inner part o f high local CO coverage on the (111) and {loo) planes, giving rise to the bands 1930 cm-1 and 1962 cm-1, respectively, and an outer part (at the boundaries), where the species are arranged in a way similar to the came of low coverage, giving rise to the bands at 1860 cm-1 and 1920 cm-'. The time dependence of the absorbance at 1930 cm-1 and 1962 cm-1 revealed a limit in the CO coverage below which the oxidation proceeded at a higher rate (Fig. 3). Further, the changes in the initial period confirm the proposed peculiarity o f the linear species on supported palladium [ l ] . This species desorbs faster from the (111) planes when the partial pressure of CO in the system decreases and the reaction initially occurs on the same planes. The spectral changes during the CO hydrogenation correspond to a continuous and steady decrease in coverage (Fig. 2 and 4), however, a rearrangement in the adlayer occurs and the rolecules are more uniformly distributed on the entire palladium surface. A comparison between the time dependence of the abmorbance at certain wavenumbers (Fig. 2 and 4) implies that CO adsorbed on the (100) planes was probably hydrogenated at a higher rate.
-
5.
REFERENCES
1
A. Palazov, 0 . Kadinov, Ch. Bonev, R. Dimitrova, D. Shopov, Surf. Sci., 226 (1990) 21 0 . Kadinov, Ch. Bonev, S. Todorova, A. Palaaov, 0 . Liete, J. Volter, in press A . Palazov, 0 . Kadinov, Ch. Bonev, I). Shopov, J. Catal,, 74 (1982) 44 A. Palaeov, 0 . Kadinov, Ch. Bonev, D. Shopov, Commun. Dept. Chem. Bulg. Acad. Sci., 1 1 (1976) 786
2
3 4