J. Phys. Chem. 1982, 86, 3977-3980
is negative, but not excessive, with cation 21 being questionable. If prepared it would slightly lower the present lower bound on RE for known ions. Finally, ions 22 and 23 are not likely to become available, a t least not under the prevailing conditions found sufficient for other conjugated ions. Concluding Remarks We have shown that graph theoretical invariantsconjugated circuits within individual Kekule valence structures-provide a useful concept for analysis and classification of conjugated ions. Besides its qualitative features, various R,Q,X,Y systems and information on the difference in RE for an anion and ita cation, the approach offers a frame for quantitative comparisons which is more durable than tentatively adopted parametrization. The quality of comparisons bawd on our approach will crucially depend on the reliability of the pertinent information on adopted standards used in calibration. Although one cannot defend the currently adopted assumptions beyond a point, it is significant for the approach that much of the conclusions will follow from less-detailed information, such as the knowledge of the relative magnitudes for the introduced parameters R,, X,,etc. and their trends. Graph theory, as such, normally does not provide numerical novel
3977
data of its own. It is concerned with sorting, ordering, and comparing available data. The presented discussion of the resonance energies of known cations and anions gives support to the hypothesis that the essential component for determining molecular stability and rr-electron delocalization is the presence of conjugated circuits. The outcome depends on the balance between the opposing contributions from conjugated circuits of size 4n + 2 and 4n. The presence of the formal charge in cations and anions is only reflected in proliferation of Kekul6 valence structures that one has to analyze. The prime purpose of this paper was to demonstrate the applicability of this particular graph theoretical approach previously found useful for discussions of neutral conjugated hydrocarbons to ions. Much of the available experimental data are in agreement with the analysis. Occasional discrepancies remain to be clarified, and it is hoped that the outlined theoretical approach will provide the frame for such clarifications as well as assist future predictions of stable ions and anticipate their properties.
Acknowledgment. I thank Professor Hafner (Darmstadt) and Professor Murata (Osaka) for their interest in the work and correspondence and suggestions which resulted in an improved manuscript.
Dissociative Adsorption of COP on Oxidized and Reduced Pt/TiO, Katunnl Tanaka and J. M. Whlte' Department of Chemistry, Unhrwsity of Texas, Austin, Texas 78712 (Received: April 75, 1982)
On both oxidized and reduced Pt/Ti02, carbon dioxide decomposes at room temperature to form chemisorbed oxygen atoms and carbon monoxide. In the absence of Pt no decompositionwas found. On reduced samples, the oxygen atoms were stabilized on the titania support and reacted further with C02to form adsorbed bidentate carbonate. On reduced Pt/Ti02, the adsorbed CO produced in the decomposition reaction produced a linearly bound species at step sites in the initial stages of the reaction. As the reaction proceeded, CO adsorbed linearly on terrace sites was found. On oxidized Pt/Ti02,the decomposition rate was much slower and only linear step site species were observed. From the results we conclude that Pt step sites are the active sites for COPdissociation.
Introduction Since the discovery of Primet et al.' of frequencies associated to adsorbed CO after exposure of Rh/Al2O3 catalysts to C02, many papers have been published dealing with the interaction of C02with Rh-based catalysts. Somorjai et d." studied C 0 2adsorption on several Rh single crystal and polycrystalline samples and found evidence for dissociation of C02using electron energy loss vibrational spectra. However, several other groups find no evidence for dissociated C 0 2 on Rh."' Recently, Solymosi et a1.* reported that C 0 2 interacting with Rh/A1203dissociates in the presence, but not in the absence, of H2. There is disagreement on this point as Iizuka and Tanakas find on (1)M. Primet, J. Chem. SOC.,Faraday Trans. I , 74, 2570 (1978). (2) D.G.Castner, B. A. Sexton, and G. A. Somorjai,Surf. Sci., 71,519 (1978). (3)D.G. Castner and G. A. Somorjai, Surf. Sci., 83, 60 (1979). (4)L. H.Dubois and G. A. Somorjai, Surf. Sci., 88, L13 (1979). (5)A. C. Yang and C. W. Garland, J. Phys. Chem., 61,1504 (1957). (6)A. C. Collings and B. M. Trapnell, Trans. Faraday Soc., 53,1436 (1957). (7)C. T. Campbell and J. M. White, J. Catal., 54,289 (1978). (8)F.Solymosi, A. Erdohelyi, and M. Kocsis, J.Catal. 65,428 (1980). 0022-3654/82/ 2086-3977$0 1.2510
a similar substrate that C 0 2is dissociated at 300 K and moderate C 0 2 pressures even in the absence of H2. In this paper, we report the dissociation of C02on both oxidized and reduced samples of Pt/Ti02. A discussion of the sites involved in the dissociation of C02as well as those involved in the retention of the reaction products is presented. Previously we have reported that C02 is not adsorbed and no carbonate species are formed on reduced TiO2.lo Upon exposure of Pt/Ti02 to CO(g), we have found two kinds of adsorbed linear CO that are attributed to step and terrace sites." Here we find that C02dissociation leads to several kinds of carbonate species on Ti02 and both step and terrace CO species on Pt. Experimental Section All the procedures used here were identical with those reported in another paper from this laboratory." Infrared spectra were taken on a Nicolet FT-IR spectrometer and were recorded in absorbance with 2-cm-l resolution. (9)T. Iiuka and Y. Tanaka, J. Catal., 70, 449 (1981). (10)K.Tanaka and J. M. White, J.Phys. Chem., in press. (11)K. Tanaka and J. M. White, J. Catal., in press.
0 1982 American Chemical Society
Tanaka and White
3Q78 The Journal of Physical Chemistry, Vol. 86, No. 20, 7982 2354 I
1245 I
2077
w
0
z a
m
2092
E
b
0 (I)
cud
t
J-
15 9 8 I 1 4
2 3,4 5
16173,1617
1
,e c
2500
2400 2300
2200
2100 2000
1900
1800 1700
W A V E N U M B E R S ( c m-
1600 1 5 0 0 1 4 0 0
1300 1200
'
Figure 1. COPabsorption at 25 "C on a R/Ti02sample reduced at 200 "C (400-200-400): (a)-(d) 20 torr of C02, exposure for 5 min, 15 min, 1 h, and 23 h, respectlvely; (e)evacuation of 25 "C for 30 min after (d).
Spectra reported here have been corrected by subtraction for absorption of the gas phase, the solid absorbent before adsorption, and the CaFz windows. All spectra were recorded at room temperature. Anatase (MCB) was reduced with H2 at 800 "C and soaked in dilute chloroplatinic acid solution to get 2 w t % Pt/TiOz. The samples was dried at 100 "C and then washed with distilled water until no chloride could be detected in the wash water. The same procedure has been followed to obtain active catalysts for photoassisted reactions. 12-16 Pellets for IR analysis were pressed between two pieces of paraffin paper; the advantages and disadvantages of this procedure have been discussed previously." Each pellet was oxidized at 400 "C in the IR cell in order to remove paraffin fragments. Three kinds of Pt/TiOz, each with a different reduction temperature, were used in this work. For simplicity each sample is denoted by three temperatures, the oxidation, reduction, and evacuation temperatures. For example, the notation 400-200-400 signifies oxidation at 400 "C, reduction a t 200 "C, and evacuation at 400 "C. Reactant COz was degassed and purified by pumping through two cold traps maintained at 77 and 195 K. These procedures were performed with great care to eliminate any residual CO that might be present in the reactant gas. All interactions of COPwith the substrates were carried out at 300 K. Results and Discussion Carbon dioxide adsorption on a 400-200-400 substrate is shown in Figure 1. When 20 torr of COzwas introduced, Figure la, bands at 2354,2283,2077,1431, and 1245 cm-l were observed. After 15 min, Figure lb, a new band was (12) S. Sato and J. M. White, Chem. Phys. Lett., 72, 83 (1980). (13)S.Sato and J. M. White, Znd. Eng. Chem.Prod. Res. Deu., 19,542 (1980). (14) S. Sato and J. M. White, J. Am. Chem. SOC.,102,7206.(1980). (15) S. Sato and J. M. white, J. Phys. Chem., 86, 336 (1981). (16) S. Sato and J. M. White, J. Catal., 69, 128 (1981).
found at 2092 cm-'. After 1 h, Figure IC,the 2354-cm-' band intensity decreased about 10% and the 2077-cm-' band increased by about a factor of 2. A 23-h exposure, Figure Id, led to an order of magnitude decrease in the intensity of the 2354-cm-l band and a shift to 2345 cm-'. In addition, new bands appeared at 1673 and 1617 cm-' while the intensity of the 2077-cm-' band increased by a factor of 3. Subsequent evacuation at 20 "C for 30 min, Figure le, left only the bands at 2092,2077, and 1617 cm-'. The bands of Figure 1are assigned as follows. In accord with the work of Morterra et and our previous work,1° the 2354- and 2283-cm-l peaks are assigned to coordinated C02 and 13C02(natural abundance), the 1245- and 1673cm-' bands to bidentate carbonate, and the 1598- and 1431-cm-l bands to bicarbonate, respectively. No monodentate carbonate was detected in these experiments. The bands at 2077 and 2092 cm-l are assigned to linear CO species on step and terrace sites, respectively. (The step-terrace language is used here but the results may be discussed equally well in terms of open and close-packed microcrystallite faces.) These band positions are in excellent agreement with those observed on both reduced and oxidized Pt/Ti02." The appearance of these two bands and the time dependence of their development indicates clearly that COz is dissociating on Pt/TiOz and that the product CO occupies terrace sites only after the step sites are filled. It is significant that the absorbance of the 2077-cm-' peak, Figure le, is 0.05 which is about 25% of that observed at 2094 cm-l when this substrate is exposed to CO." This indicates that the number of adsorbed CO molecules is much larger than could be accounted for on the basis of impurities in the C02. Additional evidence that the observed CO bands are not due to artifacts, such as CO desorption from the walls, comes from water adsorption experiments where no CO peak was observed. When this substrate was exposed to CO, a bridged species at 1854 cm-l was found.'l The fact that it is not (17) C. Morterra, A. Chiorino, and F. Boccuzzi, 2. Phys. Chem. (Frankfurt am Main), 124 211 (1981).
Dlssociathre Adsorption of COP on R/Ti02
The Journal of Physical Chemlstty, Vol. 86, No. 20, 1982 3979
w
0 Z 4
I I
2344
m a
0 (0
m
4
2343
1628
2056
I
A
1
2056
2
IO
2400
2300
2200
2000
2100
1900
1800
1700
1600
1500
1400
1300
I
1200
WAVEMUMBERS (cm-')
Flgure 2. CO, adsorption at 25
O C
on a Pt/TiO, sample reduced at 400
OC
(400-400-400):
(a)+)
exposure time of 5 min, 20 min, and 2 h,
respectively.
found here is ascribed to the relatively low coverage of CO. Hopster and lbach,l8using high-resolution electron energy loss spectroscopy, find that on Pt(ll1) bridged species appear only at high coverages of CO. The behavior of the carbonate bands provides insight into the surface processes involving the oxygen atoms formed in the decomposition of carbon dioxide. In the early stages of the reaction, Figure la-, coordinated COP dominates the spectrum. Some bicarbonate is found along with a hydrogen-bonded species involving two adjacent OH groups of bicarbonate species, 3615 cm-' (not shown here).1° In another papertlowe show that two interacting bicarbonate species form water and coordinatively unsaturated Ti sites which are active for the formation of bidentate carbonate species in a reaction with C02. This is shown schematically in eq 1. Here 0 represents a vacant 0"
II
.' T I ' . bidentate carbonate OH
f co2
I
bidentate carbonate or whether the reduced sample possesses these sites intrinsically. To address this question, we exposed a 400-400-400 sample to 20 torr of COPat 300 K. Bidentate carbonate species (1674 cm-'), bicarbonate (1435 cm-l), and coordinated COP(2347 cm-') were found after 5 min, Figure 2a, emphasizing that reduced samples do possess sites for bicarbonate formation. This contrasts to interactions in the absence of Pt where no carbonate species are found.1° With time, 20 min and 2 h as shown in parts b and c of Figure 2, there were slow changes. After 2 h, Figure 2c, the intensity of the coordinated COz species dropped by a factor of 3 and a small amount of linear CO appears at 2056 cm-l indicating dissociation of COP. It is also of interest that, during COPexposure, the intensity of a broad band in the 1000-1200-~m~~ range increased. This is due to T i 4 lattice vibrations and suggests that some of the oxygen atoms derived from the C02decomposition oxidize the previously reduced Ti02 surface. Many papers have dealt with the structure of Ti02 following reduction at 500 O C and it is thought to be best or TieOlPn In addition to oxidizing described as Ti40;4-21 the surface, oxygen atoms can change the surface coordination to lead to bidentate carbonate as shown in eq 2. Picking up two electrons corresponds to oxidation of the Ti center.
bicarbonate 0 'Ti/
0'-
0
t 2CQ
t H20 (1)
coordination site on Ti. These processes are also thought to take place here. While the formation of bidentate carbonate is a convenient way to describe one role played by oxygen atoms formed in the decomposition of COP,the situation is complicated on 400-200-400 materials since it is not clear whether the oxygen atoms produced in the decomposition process change the surface enough to form (18)H.Hopster and H.Ibach, Surf. Sci., 77,109 (1978).
The very weak intensity of linear CO bands in Figure 2, as compared to Figure 1,is related to the strong met(19)R.T.K.Baker, E. B. Prestridge, and R. L. Garten, J. Catal., 59,
_29._1 _(1979). ~
(20) R. T. K. Baker, J . Catal., 63, 523 (1980). (21)S. J. Tauster, S. C. Fung, R. T. K. Baker, and 3. A. Horsley, Science, 211,1121 (1981). (22) B.-H. Chen and J. M. White, J. Phys. Chem., in press.
3980
The Journal of Physical Chemktry, Vol, 86, No. 20, 1982
Tanaka and White
C
10
2400
2300
2200
2100
2000
1900
iaoo
1700
1600
1500
1400
\ 1
)O
W A V E NUMBERS (cm-')
Figure 3. CO, adsorptlon at 25
OC
on an oxidized Pt/li02sample (400-NO-400): (a)-@) exposure time of 10 min, 90 min, and 3 day, respectively.
al-support interaction (SMSI) effect widely discussed in the literature23and prevailing on strongly reduced Pt/ Ti02. In the early stages of the reaction, Figure 2a, the formation of bidentate carbonate species at 1674 cm-' is accompanied by the dissociation of C02. However, the SMSI effect lowers significantly the chemisorption capacity for CO and very little adsorbed CO accumulates. The decrease of the 1435-cm-l intensity and the appearance of the water bending mode at 1628 cm-' is readily explained by reaction 1. Figure 3 shows the results of C02 exposure on a 400NO-400 sample, Le., oxidized at 400 "C but not reduced. After a 10-min exposure, Figure 3a, to 20 torr of C02 at 25 "C, bands were observed at 2352,2284,1674,1628,1597, and 1434 cm-'. These are assigned to coordinated COPand 13C02,bidentate carbonate, water, and a pair of bands due to bicarbonate, respectively. After 90 min, Figure 3b, the coordinated COz intensity (2352 cm-') dropped by about an order of magnitude while bands due to water (1628 cm-') and bidentate carbonate (1674 cm-9 doubled. These increases were accompanied by a decrease of the bicarbonate (1434 cm-') intensity and the relation is understood in terms of reaction 1. After 3 day, Figure 3C, the coordinated C02 intensity dropped to a negligible level, a band due to linear CO on Pt appeared at 2070 cm-' (step sites), the water band at 1628 cm-' grew, and the bicarbonate bands at 1597 and 1434 cm-' decreased. The slope observed below 1700 cm-' is attributed to inadequate subtraction. The slow formation of adsorbed CO species is very interesting. In another paper1' we reported that CO inter(23) S. J. Tauster, S. C. Fung,and R.L. Garten, J. Am. Chem. SOC., 100, 170 (1978).
aded with this kind of support to form C02and carbonates but the oxygen atoms on step sites were more difficult to remove than those on terrace sites. The results obtained here suggest that some sites are still active for C02 dissociation when the surface is covered with oxygen atoms. However, the reactions are extremely slow. The observed CO frequency, 2070 cm-', indicates adsorption at step sites and is consistent with the results of Figure 1 where step sites are filled before terrace sites. These experimental results imply that C02dissociation takes place either on step sites or at the interface between Pt sites and the TiOz support. If the latter occurs then migration of CO to steps must occur subsequently. Such processes are well-known on single-crystal surfaces.24 Since adsorbed oxygen atoms cover step sites strongly and selectively" it is difficult to see how CO is selectively adsorbed on step sites. We ascribe C02 dissociation, under all conditions used in this work, to reaction at step sites. This readily accounts for the results of Figures 1 and 3 where the Pt morphology, non-SMSI, is characterized by rough hemispherical particles with high concentrations of steps and kinks rather than flat particles dominated by Pt(lll).19 As noted in the Introduction, there is disagreement about whether or not C02 dissociation occurs on Rh samples. Our results point to surface heterogeneity as one possible explanation for this disagreement. Step sites or open faces, present in significant concentrations, could account for the dissociation. Acknowledgment. This work was supported in part by the Office of Naval Research. (24) C. T. Campbell, G. Ertl, H. Kuipers, and J. Segner, Surf. Sci., 107
207 (1981).
