J . Phys. Chem. 1987, 91, 5931-5937
5931
Infrared and X-ray Photoelectron Spectroscopy Study of CO and CO, on PtKeO, T. Jin, Y. Zhou, G. J. Mains,? and J. M. White* Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712-1167 (Received: December 29, 1986)
The adsorptions of CO and C 0 2and Pt/Ce02 were studied by infrared and X-ray photoelectron spectroscopy. A heterogeneous distribution of linearly adsorbed CO on Pt at terrace, step, and corner sites was proposed based on changes in the absorption of the CO infrared stretching frequenciesfrom 2084 to 2060 cm-l upon heating under vacuum from 300 to 473 K. Pretreatment of the sample with 0, above 373 K lead to a blue-shifted peak at 2091 cm-I, interpreted as arising from CO adsorbed on Pt with an atom of oxygen at a neighboring site, and a new peak at 2131 cm-I, assigned to CO coadsorbed with an oxygen atom on the same Pt atom. Adsorption of CO, at room temperature resulted in the appearance of an IR band at 2065 cm-I, assigned to CO adsorbed on a Pt atom on a step or corner. The intensity of the CO band from C 0 2 adsorption was sensitive to pretreatment of the sample, being completely suppressed by preoxidation but enhanced by prereduction. Lattice oxygen vacancies in the support, CeO,, in the vicinity of the Pt particles are responsible for the formation of CO from CO,. XPS spectra show that Ce3+is formed by the prereduction treatment, supporting the proposal that lattice oxygen vacancies play an important role in the C 0 2 decomposition. Preoxidation is shown to lead to the formation of PtZ+XPS (40 spectra, which is removed by flash heating. Disproportionation of CO to carbon and C 0 2 was shown not to occur by pretreatment of the surface with CO at 573 K.
1. Introduction The rare earth (lanthanide) oxides exhibit a rich variety of chemical and physical characteristics which make them interesting subjects for catalytic studies.' Ce02was found to be the most active of the lanthanide oxides for the oxidation of b ~ t a n e ,a~ . ~ fact attributed to the low stability of the tetravalent ion and a low fourth ionization potential rather than to surface acid/base properties. Yao4 studied the oxygen storage capacity of CeO, and found that it was enhanced by the presence of either Pt or Pd. Summers and Ausen5 claimed that C e 0 2 donated oxygen to Pt in their study of the oxidation of CO on Pt/Ce02. s e c k and Bell6 reported that the addition of a rare earth oxide to Pd/Si02 increased the reactivity of C0-H2 mixtures and also promoted the disproportionation of CO to form carbon and C 0 2 . Lattice oxygen in C e 0 2 was proposed to play a role in the hydrogenation of C 0 . 7 Recent TPD results from this laboratory8 also implicate lattice oxygen at the P t / C e 0 2 interface in the oxidation/reduction of CO/CO2 on Pt/CeO,. In order to further understand the interaction between adsorbed CO and interface lattice oxygen on the Pt/Ce02, the following IR and XPS study was undertaken. There have been numerous IR studies of CO adsorbed on bulk Pte12 and some on supported Pt.13-15 On P t / C e 0 2 there is only one report, and it served as an important guide to the work reported here.I6 As the sample preparation procedures were somewhat different,17the similarities and differences in the results are noteworthy and are discussed below. A critical overview of CO adsorption by Ishi, Ohno, and Viswanathan'* has been published. Adsorbed CO has been studied recently by a variety of e ~ p e r i m e n t a l ' ~and - ~ ~theoretical techn i q u e ~ , ~and - ~ ~remains controversial. 2. Experimental Section Materials. CeO,, ceria (Alfa), was used as the support after washing with a 5% HC1 solution, rinsing with deionized water to remove cation impurities (Ca2+, etc.), and drying in an oven at 373 K overnight. Two platinized ceria samples, 2% Pt/CeO2(473) and 2% Pt/Ce02(673), were impregnated with a dilute solution of chloroplatinic acid, H2PtC16(Engelhard Ind.), dried at 373 K for a day, and calcined in air at 473 and 673 K for 1 day and 12 h, respectively. The surface areas are 5 m2/g for CeO, and, to within lo%, 3.6 m2/g for both platinized samples. The average size of the crystalline Pt particles seen in XRD was about 370 A for the 2% Pt/Ce0,(473) and 130 8,for the 2% Pt/Ce02(673) samples based on the half-width of the XRD peaks. +On sabbatical leave, Department of Chemistry, Oklahoma ersity, Stillwater, OK 74078.
State
Univ-
0022-3654/87/2091-5931$01.50/0
Prior to gas adsorption, the samples were further heated under evacuation ( lod Torr) at 800 K for 30 min and, when an oxidized or reduced surface was desired, treated with 1 Torr of 0, or H2 at the designated temperatures. CO, 0 2 , and H2 were purified by passage through a liquid N 2 trap to remove water. CO, was distilled at 195 K to remove water and was degassed at 77 K to remove air and any possible C O contamination. The gas purities were checked by mass spectrometry when they were dosed to the samples. Equipment. The infrared spectrometry was performed by means of a Nicolet 7199 FTIR spectrometer with a resolution of 2 cm-I. The scan number was 1000. Absorptions by the windows (CaF,) and by the blank powdered sample were subtracted from all the spectra. A stainless steel cell with CaF, windows was used. The sample disks, shown schematically in
(1) Rosynek, M. P. Catal. Reu.-Sci. Eng. 1977, 16, 111. (2) Hattori, T.; Inoko, J.-I.; Murakami, Y. J . Catal. 1976, 42, 60. (3) Yamaguchi, T.; Ikada, N.; Hattori, M.; Tanabe, K. J. Caral. 1981,67, 324. (4) Yao, M. C.; Yao, Y. E. J . Catal. 1984, 86, 254. (5) Summers, J. C.; Ausen, S.A. J . Catal. 1979, 58, 131. (6) Rieck, J. S.; Bell, A. T. J. Catal. 1986, 99, 278. (7) Mendelovici, L.; Steinberg, M. J. Catal. 1985, 98, 285. (8) Jin, T.; Okuhara, T.; Mains, G. J.; White, J. M., submitted for publication. (9) Engel, T.; Ertl, G. Adu. Card. 1979, 28, 1. (10) Tornquist, W. J.; Griffin, G. L. J . Vac. Sci. Technol., A 1986,4(3), 1437. (11) Hayden, B. E.; Bradshaw, A. M. Surf.Sci. 1983, 125, 787. (12) Greenler, R. G.; Burch, K. D.; Kretzschmar, K.; Klauser, R.; Bradshaw, A. M.; Hayden, B. E. Surf.Sci. 1985, 152/153, 338. (13) Primet, M. J . Catal. 1984, 88, 273. (14) Primet, M.; Basset, J. M.; Mathieu, M. V.; Prettre, M. J. Catal. 1973, 29, 213. (15) Tanaka, K.; White, J. M. J . Phys. Chem. 1982, 86, 3977. (16) Daniel, D. W. Ph.D. Thesis, University of Texas, 1984. (17) Daniel, D. W., private communication. (18) Ishi, %-I.; Ohno, Y.; Viswanathan, B. Surf. Sci. 1985, 161, 349. (19) Rogozik, J.; Dose, V. Surf.Sci. 1986, 176, L847. (20) Lang, J. F.; Masel, R. I. Surf.Sci. 1986, 167, 261. (21) Olgetree, D. F.; Van Hove, M. A,; Somorjai, G. A. Surf.Sci. 1986, 173, 351. (22) Freyer, N.; Kiskinova, M.; Pirug, G.; Bonzel, H. P. Appl. Phys. A 1986, 39, 209. (23) Mehandru, S. 9.;Anderson, A. B.; Ross, P. N. J . Catal. 1986, 100, 210. (24) Bagus, P. S.; Hermann, K. Phys. Reu. E Condens. Matter 1986, 33, 2987. (25) Korzeniewski, C.; Pons, S.; Schmidt, P. P.; Seversen, M. W. J . Chem. Phys. 1986, 85, 4153.
0 1987 American Chemical Society
5932
The Journal of Physical Chemistry, Vol. 91, No. 23, 1987
Jin et al.
Thermocouple
2084
I
I
,Ta wire
T a mesh Figure 1. Powder sample support.
Figure 1, were prepared by pressing the powdered samples onto a tantalum mesh which had been welded to two pieces of Ta wire, enabling the sample to be heated resistively. Cooling could be accomplished by contacting the tantalum leads with liquid N,. A thermocouple was spot-welded on the mesh such that it was pressed into the sample powder. The thicknesses of the disks were 0.20-0.22 mm for Ce0, and 0.08-0.10 mm for Pt/CeO,. XPS spectra were taken on a VG ESCALAB 5 spectrometer using an Mg Ka anode dissipating 270 W. The pass energy of the analyzer was 20 eV in all experiments. The XRD spectra were taken on a Philips PW 1729 XRD diffractometer.
3. Results IR Spectra of CO on PtlCeO,. Blank experiments proved the mesh and Ce0, did not absorb in the region between 1000 and 3000 cm-I. The samples were exposed to 10 Torr of C O for 10 min at 300 K and then evacuated at various temperatures for 5 min. Figure 2a-e shows the IR spectra of C O adsorbed on 2% Pt/CeO,(473). A single intense absorption was observed at 2084 cm-' (Figure 2a). Heating the sample to 373 K under vacuum led to a decrease in the peak intensity and a red shift of the absorption centroid to 2076 cm-' with a blue-side shoulder at 2081 cm-' (Figure 2b). Heating of the sample to 423 K under vacuum lowered the absorbed intensity, removed the shoulder at 2081 cm-', shifted the absorption maximum to 2072 cm-', and created a red-side shoulder a t 2063 cm-' (Figure 2c). Further heating to 473 K lowered the absorbed intensity and shifted the absorption maximum to 2062 cm-' with a shoulder at 2071 cm-' (Figure 2d). The C O peak(s) were removed completely by evacuation at 523 K (Figure 2e). Redosing the C O at 300 K after evacuating the sample at 600 K reproduced the original observation (Figure 2a). Figure 2f-j shows the IR spectra of C O adsorbed on 2% Pt/ Ce02(673). Intense absorptions were observed at 2082 and 2077 cm-', and a red shoulder was observed at 2066 cm-' (Figure 2 0 . Heating the sample to 373 K under vacuum led to a decrease in the absorbed intensities, the 2082-cm-I peak disappeared, the absorption maximum shifted to 2075 cm-', and the red-side shoulder at 2064 cm-' remained (Figure 2g). Heating of the sample to 423 K under vacuum lowered the absorbed intensity and shifted the absorption maximum to 2072 cm-I, and the redside shoulder at 2062 cm-' became the most intense (Figure 2h). Further heating to 473 K lowered the absorbed intensity, and the residual absorption maximum was at 2060 cm-' (Figure 2i). The CO peak(s) were removed completely by evacuation at 523 K (Figure 2j). Redosing the C O at 300 K after evacuating the sample at 600 K reproduced the original observation (Figure 2 0 . Figure 3a-e shows spectra of C O adsorbed on 2% Pt/Ce02(473) which had been pretreated by evacuation (800 K, 30 min), with H, (1 Torr, 673 K, 5 min), with 0, (1 Torr, 373 K, 5 min), with O2 (1 Torr, 373 K, 5 min) and then flash heated to 800 K, and with C O (1 Torr, 573 K, 5 min), respectively. When CO was dosed onto the H,-pretreated sample and evacuated at 300 K, except for peak maximum shifting from 2084 to 2082 cm-I, no significant differences were found between the treated and
e
i
20
21
22
19 2 2
20
21
19
Wave Number / l o 2 cm-' Figure 2. IR spectra of CO on evacuated (800 K, 30 min) Pt/CeO,: (a-e) calcined at 473 K, (f-j) calcined at 673 K; (a, f) CO adsorbed at 10 Torr for 10 min at room temperature and evacuated at room temperature for 5 min; (b, g) evacuation of (a) and (0 at 373 K for 5 min; (c, h) evacuation of (b) and (g) at 423 K for 5 min; (d, i) evacuation of (c) and (i) at 473 K for 5 min; (e, j) evacuation of (d) and (i) above 523 K for 5 min.
I
2084
7
20017 b
.. ...
:e 2 0 8 5
. ..
............'
2200
'... ...._...,.....,..
2000
1800
Wave Number I cm-l Figure 3. IR spectra of CO on pretreated 2% Pt/Ce0,(473): (a) same as Figure 2a; (b) pretreated with 1 Torr of H2 at 673 K for 5 min, cooled to room temperature, and exposed to CO as in Figure 2a; (c) pretreated with 10 Torr of 0,at 373 K for 5 min, cooled to r w m temperature, and exposed to CO as in Figure 2a; (d) repeat of (c) flash heated to 800 K before cooling and exposing to CO as in Figure 2a; ( e ) pretreated with 1 Torr of CO at 573 K for 5 min, cooled to 300 K and exposed to CO as in Figure 2a.
untreated surfaces (Figure 3a,b). When the C O was admitted to the oxidized Pt/CeO, at 373 K and evacuated, the CO ab-
The Journal of Physical Chemistry, Vol. 91, No. 23, 1987 5933
Adsorptions of C O and C 0 2 on P t / C e 0 2
1eo0
1
2200 2100 2000 1900
Wave Number / cm-' Figure 4. IR spectra of CO on 2% Pt/Ce0,(473) preoxidized at 473 K and cooled to 300 K: (a) CO adsorbed at 10 Torr, 300 K for 10 min and evacuated at 300 K for 5 min; (b) evacuation of (a) at 373 K for 5 min; (c) evacuation of (b) at 423 K for 5 min; (d) evacuation of (c) above 473 K for 5 min; (e) evacuation of (d) above 523 K for 5 min. 207 1
(1)
10.2 1530
I
2065
I
2064
(11)
T
I
h
I\
2100
1900
1700
1500
850
Wave Number / cm-' Figure 6. IR spectra of CO, on pretreated CeOz: (a) COz adsorbed at 10 Torr, 300 K for 10 min on freshly evacuated C e 0 2 and degassed at 300 K for 5 min; (b) pretreated with 1 Torr of H, at 523 K for 5 min, cooled to 300 K, and exposed to CO, as in (a); (c) pretreated with 1 Torr of 0, at 373 K for 5 min, flash heated to 800 K, cooled to 300 K, and exposed to CO, as in (a); (d) pretreated with 2 Torr of H20a t 300 K for 5 min and exposed to COz as in (a). The absorption due to water has been subtracted.
1380
b C
1850 1700 1550 1400 1250 1100
o.2
1300
W a v e Number / cm-' Figure 5. IR spectra of CO,on Pt/Ce02 (panel I is for 473 K calcination while panel I1 is for 673 K calcination): (a, e) C 0 2 adsorbed at 10 Torr, 300 K for 10 min on freshly evacuated samples degassed at 300 K for 5 min; (b, f) pretreated with 1 Torr of O2 at 373 K for 5 min, cooled to 300 K, and exposed to CO, as in (a, e); (c, g) pretreated with 0, as in (b, f), flash heated to 800 K, cooled to 300 K, and exposed to CO, as in (a, e); (d, e) treated with 1 Torr of H2 at 523 K for 5 min, cooled to 300 K, and exposed to C 0 2 as in (a, e).
sorptions were at 2091 and 2131 cm-' (Figure 3c). The ratio of the peak absorption a t 2131 cm-' to that at 2091 cm-' was 1:7. When this sample was flash heated to 800 K prior t o adsorption of CO, the spectrum of Figure 3d was obtained, which was very
close to that obtained for the untreated sample (Figure 3a). CO pretreatment, found to have a significant effect on the CO to CO, conversion in the TPD experiments,s had no effect on the IR absorption spectrum in the C O region. (Compare spectra e and a of Figure 3.) When a sample of 2% Pt/Ce0,(473) was preoxidized at 100 deg higher in temperature, 473 K, the spectra, shown in Figure 4, were obtained. Figure 4a shows that the ratio of the peak intensity at 2131 cm-I to that at 2091 cm-I increased to 11:12 as a consequence of the higher temperature preoxidation treatment. Heating the sample under vacuum to 373 K reduced the 2091-cm-I peak by 67% and shifted it to 2083 cm-I whereas the peak at 2131 cm-I was unchanged in intensity and position (Figure 4b). When the temperature was increased to 423 K, the 2083-cm-I peak completely disappeared and the 21 3 1-cm-' peak decreased about 26% in intensity (Figure 4c). When the temperature was increased to 473 K, the 2131-cm-I peak reduced further (Figure 4d) and completely disappeared at higher temperature (Figure 4e). Thus, the C O giving rise to the IR absorption at 2131 cm-I is thermally more stable than that which created the IR absorption at 2091 cm-'. Figure 5 shows the I R spectra obtained when C 0 2 was dosed on the two 2% Pt/Ce02 samples which had been calcined in air at the two different temperatures employed in this study, 473 K (Figure 5a-d) and 673 K (Figure 5e-h). The samples were exposed to 10 Torr of C 0 2 for 10 min at 300 K and then evacuated for 5 min before spectra were obtained. IR absorptions in the region of the CO stretching frequency were observed for samples calcined at both temperatures (see spectra a and e of Figure 5, 2071 and 2064 cm-', respectively). For the samples that had been pretreated with 1 Torr of Hz at 673 K, the IR absorption peaks (Figure 5d,h) were found at 2065 and 2060 cm-I, and the peak intensities were enhanced. When the samples were pretreated with 1 Torr of O2at 373 K, the IR absorption of CO was no longer detectable (Figure 5b,f). Two broad absorption peaks a t longer wavelengths were found in the samples which gave absorptions in the CO region (Figure 5a,d,e,h). These peaks, ca. 1380-1390 and 1530-1550 cm-', are in the region commonly attributed to carbonate absorption. They were not found on the samples which
Jin et al.
5934 The Journal of Physical Chemistry, Vol. 91, No. 23, 1987 C'
C
1600
T
I A'
I
0,
0 C
m
e 0 v)
a
e
1750 1800 1450
1300 1150 1000
Wave Number I cm-' Figure 7. IR spectra of C 0 2 on freshly evacuated Ce02: (a) same as Figure 6a; (b) evacuation of (a) at 473 K for 5 min; (e) evacuation of (b) at 573 K for 5 min; (d) evacuation of (e) at 673 K for 5 min; (e) evacuation of (d) at 373 K for 5 min.
were preoxidized. Neither these long-wavelength peaks nor those in the C O absorption region could be recovered by simply flash heating the preoxidized sample prior to CO dosing (Figure 5c,g). IR Spectra of CO, on CeO,. Figure 6 shows the IR spectra produced when C 0 2 is adsorbed alone on C e 0 2 which had been subjected to the pretreatments described for Pt/Ce02 in the previous section. In Figure 6a, the sample had been evacuated at 800 K for 30 min prior to cooling and exposure to C 0 2 . As can been seen, five distinct bands at 1600, 1481, 1420, 1325, and 1040 cm-I are observed in the carbonate region. On the Hz-treated surface (Figure 6b), the IR band at 1600 cm-' shifted to about 1590 cm-l, the 1325- and 1040-cm-' bands remained unchanged, and the 148 1-cm-I band decreased in intensity as the 1420-cm-' band increased about 50%. Pretreatment with 1 Torr of O2 at 373 K, followed by flash heating, cooling, and exposure to C 0 2 , decreased all of the IR absorptions in this region (Figure 6c), and a new band appeared at 1628 cm-l. The central bands, 1481 and 1420 cm-', which were enhanced by H2 treatment (Figure 6b), were reduced more drastically by the O2 treatment than the other three bands at 1586, 1312, and 1040 cm-l. The latter did not change upon H2 treatment. Figure 6d shows the IR spectrum produced when COz was dosed onto a CeO, sample which was pretreated with H 2 0 (2 Torr, 300 K, 5 min). The water spectrum has been subtracted. As can be seen, the observed spectrum is very similar to that for the evacuated sample (Figure 6a), indicating that preadsorbed water has no effect on the spectra obtained. The thermal stability of adsorbed COz on C e 0 2 was tested through a stepwise evacuation at increasing temperatures (Figure 7). Figure 7a-e shows the IR spectra obtained after dosing with COz followed by evacuation at 300, 423, 573, 673, and 773 K, respectively. The absorption peaks at 1600, 1325, and 1040 cm-" reduced synchronously and disappeared altogether at 673 K. The other two peaks, 1481 and 1420 cm-I, also reduced synchronously, but more slowly, and disappeared at 773 K. XPS Spectra of Pt/Ce02. Figure 8 shows the XPS spectra of the 2% Pt/Ce02(673) samples in the Ce (3d) region, binding energy (BE) range 875.0-925.0 eV. In the sample freshly evacuated at 800 K eight transitions were observed at 882.6,889,5, 898.7,885.9 eV (labeled A to D i n Figure 8a) and at 901.1,908.1, 917.3, 904.4 eV (labeled A' to D' in Figure 8a). Prereduction of the sample at 773 K (Figure 8b) gave reduced peaks labeled B and B' and increased peaks labeled D and D'. Upon an oxidation
,982.8 I
..,' I
,
880
8SO
900
910
920
Binding Energy I eV Figure 8. XPS Ce(3d) spectra of Pt/CeO, (dashed line is for pure Ce02): (a) evacuation at 800 K for 1 h; (b) pretreated with 4 X Torr of H2 at 773 K for 30 min; (c) pretreated with 1.5 X low5Torr of O2 at 373 K for 30 min; (d) same as (e) but flash heated to 800 K.
I
L
70.9
74.3 I
l
,
65
1
70
,
'
1
75
1
.
80
Binding Energy I eV Figure 9. XPS Pt(4f) spectra of Pt/Ce02: (a) evacuation at 800 K for 1 h; (b) pretreated with 4 X Torr of H2 at 773 K for 30 min; (e) pretreated with 1.5 X Torr of O2at 373 K for 30 min; (d) same as (c) but flash heated to 800 K.
treatment at 373 K (Figure 8c), the peaks at B and B' were greatly enhanced and those labeled D and D' completely disappeared. This latter spectrum (Figure 8c) is the same as that published in the handbook26for CeO,, the peaks labeled B and B' are the 3d5/2 and 3d,/, transitions, and the A, A' and C, C' are shakedown ( 2 6 ) Muilenberg, G.E., Ed. Handbook of X-ray Photoelectron Spectroscopy; Perkin-Elmer, Physical Electronics Division: Eden Prairie, MN, 1979.
Adsorptions of CO and C 0 2 on Pt/Ce02 and shake-up satellites of these transition^.^^ When the oxidized sample was flash heated to 800 K, the spectrum shown in Figure 8d was obtained, similar to that observed for the freshly evacuated sample (Figure 8a). The dotted line in Figure 8 shows the spectrum obtained from pure C e 0 2 after being evacuated at 823 K for 20 min. Figure 9 shows the variation of the XPS spectra in the region of the Pt(4f) bands (60-85 eV). Spectra a 4 of Figure 9 were obtained by the same sequence of evacuation, prereduction treatment, preoxidation, and flash heating used in Figure 8. As can be seen, spectra a, b, and d of Figure 9 are very similar and correspond to the 4f7p and 4fSp peaks reported for Pt metal, 70.9 and 74.3 eV, respectively, A higher BE peak is observed for the oxidized surface (Figure 8c), attributed to the 4fs/2 transition of Pt2+. It should be noted that flash heating removed this higher BE peak. No changes were detected in the O(1s) peak as a consequence of these pretreatments, and hence they are not presented. 4. Discussion
Chemical States of the Pt/Ce02 Surfaces. The observation of only the Pto peaks on the Pt/Ce02 surfaces upon evacuation (Figure sa), hydrogen pretreatment (Figure 9b), and oxygen pretreatment followed by flash heating (Figure 9d) indicates that the Pt is in a completely reduced state on these surfaces. That Pt is partially oxidized to PtO by O2treatment at 373 K is clear from Figure 8c. Oxidation of small Pt clusters has been reported at room temperature,28 so this result is not unexpected. However, the oxygen atoms bound to the Pt surface are readily removed by heating in a vacuum to 800 K. This observation supports the conclusion reached earlier in the TPD experimentss that oxidation of C O to C 0 2 on flash-heated surfaces did not involve oxygen atoms on the Pt. The excellent agreement in the 875-925-eV region between the XPS spectrum for the oxygen-pretreated surface (Figure 8c) and that reported in the handbook26 for C e 0 2 suggests that these surfaces are identical insofar as XPS is concerned. The peaks D and D' in Figure 8, which are present in the evacuated sample, are even more evident in the hydrogen-treated sample (Figure 8b), and return upon flash heating of the oxygen-treated sample, are assigned to Ce3+,consistent with the assignment of these peaks as the dominant peaks produced when C e foil is partially oxid i ~ e d . ~Since ~ ? ~these ~ peaks were also observed for pure CeOz (dotted line in Figure 8), they do not require the presence of Pt for their formation. Burroughs et aL30 reported that no significant changes occurred in the XPS spectrum of C e 0 2 powder upon evacuation or preheating in O2or H2. Since they reported eight peaks like those shown in Figure 8a and since the peaks assigned to Ce3+appear after mild heat treatment (Figure 8d), their sample may have contained.Ce3+ which was not completely removed by their oxidation treatment, especially ,if their degassing temperature was high. Assignment of CO Frequencies on Freshly Evacuated Pt/Ce02. The stretching frequency of CO adsorbed on metals has been found to be very sensitive to surface c o ~ e r a g e . ' ~ ,The ~ ' intrinsic shifts are mainly due to the effect of the dipole-dipole coupling of the linear C0,13,32-34which makes peak assignments difficult when several adsorbates are present. Recently, however, consistent results were obtained which included the effect of dipole-dipole coupling for CO on Pt/Ti02,34on Pt/Si02,12,3sand on Pt/Al2O3.l3 A heterogeneous distribution of linearly adsorbed C O molecules (27) Praline, G.; Koel, B. E.; Hance, R. L.; Lee, H.-I.; White, J. M. J . Electron Spectrosc. Relat. Phenom. 1980, 21, 17. (28) Fleisch, T.; Mains, G . J. J . Phys. Chem. 1986, 90, 5317. (29) Barr, T. L. Quantitative Surface Analysis of Materials; ASTM SP 643; McIntyre, N. S., Ed.;ASTM: Philadelphia, PA, 1978; pp 83-104. (30) Burroughs, P.; Hamnet, A,; Orchard, A. F.; Thornton, G . J . Chem. SOC.,Dalton Trans. 1976, 1689. (31) Eischens, R. P.; Pliskin, W. A. Adv. Catal. 1953, 10, 1 . (32) Hammaker, R.; Francis, S. A,; Eischens, R. P. Spectrochim. Acta 1955, 21, 1295. (33) Clossley, A.; King, D. A. Surf. Sci. 1977, 68, 528. (34) Tanaka, K.; White, J. M. J . Catal. 1983, 79, 81. (35) Bartok, M.; Sarkany, J.; Sitkel, A. J . Catal. 1981, 7 2 , 236.
The Journal of Physical Chemistry, Vol. 91, No. 23, 1987 5935
on Pt is proposed to explain these intrinsic shifts. Greenler et al. further proposed,'2 based on experiments using single-crystal Pt surfaces compared with those for Pt/Si02, that the shifts merely due to surface coverage variation were the same within 6 cm-'. We report similar observations for Pt/Ce02. The IR absorptions for linear C O in the regions 2080-2085, 2071-2077, and 2060-2066 c d showed obviously different intensity changes upon evacuation at different temperatures (Figure 2a-c). We interpret the origin of these IR absorptions as linearly adsorbed C O molecules in different local environments due to the size/shape distribution of Pt clusters in the system. The high-frequency absorption (2085-2080 cm-I) is assigned to C O linearly adsorbed on close-packed Pt surfaces, Le., terraces, and the middle-frequency absorption (2077-207 1 cm-I) and low-frequency absorption (2066-2060 cm-I) are assigned to defect sites, Le., to steps and corners, respectively. Such an interpretation is consistent with the different behavior found in the two samples shown in Figure 2, where the average Pt metal crystallite sizes differ by almost a factor of 3. For the 2% Pt/Ce0,(673) sample (average crystal diameter 130 A), the fractions of absorption by the defect sites, i.e., the middle and low frequencies, are larger than those for the 2% Pt/Ce02(473) sample (average crystal diameter 370 A) at each evacuation temperature. The smaller Pt crystallites probably have a larger fraction of defect sites. Greenler et al.I2 suggested that the stretching frequency was higher for the C O on the steps than on the corners. However, we cannot distinguish these two species since both types of sites increase with increasing Pt dispersion. Adsorption Sites of CO on Pretreated Pt/Ce02. Primet et al. investigated the effect of coadsorbed molecules on the C O stretching frequency on Pt/Al2o3.I39l4When a Pt/A1203 sample on which CO was preadsorbed was exposed to oxygen, the IR band at 2080 cm-' shifted with time to two bands, one at 2120 cm-I, assigned to C O bound to a platinum atom directly bonded to an oxygen atom, and the other at 2095 cm-', assigned to C O bound to a Pt having an oxygenated Pt atom as the nearest neighbor. We report similar observations in Figures 3c and 4. The 2091-cm-' band may be attributable to a C O island surrounded by oxygen (as Primet suggested) or to adsorbed CO uniformly mixed with oxygen on the Pt surface. We prefer the latter model since oxygen surrounding large CO islands would not be expected to shift the shows CO in the center of the island, and single-crystal Pt that the interaction between CO(a) and O(a) did not necessarily occur at the interface between these islands. The high-frequency band (2 122-2 131 cm-I) has been assigned to C O adsorbed on metal atoms either bound to oxygen or bound to partially oxidized metal clusters for Pt13J4337and for Pd.38939 The two models might not be different. The data reported here, however, suggest that the depth of Pt oxidation varies with the temperature of the oxidation treatment, since this band increased as the temperature of the oxidation increased from 373 K (Figure 3c) to 473 K (Figure 4a). [We assume that the oxygen on the surface was increased by the higher temperature oxidation pretreatment.] Our observation that hydrogen pretreatment did not produce a significant change in the IR spectrum of the adsorbed C O is consistent with the XPS determination that the Pt(4f) spectrum was unchanged by this treatment or by simple evacuation. Upon oxidation, the formation of PtO is confirmed, and its subsequent decomposition by flash heating is consistent with the IR spectra (Figure 3c,d). All the IR bands observed here are in reasonable agreement with those of Daniel,16who used a different preparation procedure involving lower temperature calcination and no treatment with HCl. As noted below, this procedure leads to loss of C O uptake activity presumably due to carbon deposition on the Pt. Since the decomposition of adsorbed C O to C 0 2 and carbon on Pt surfaces has been proposed on supported PtI7 and (36) (37) (38) (39) 5, 67.
Akhter, S.;White, J. M. Sur-. Sci. 1986, 171, 527. Heyne, H.; Tompkins, F.C. Proc. R . SOC.London, A 1966,292,480. Stuve, E. M.; Madix, R. J.; Brundle, C. R. Surf. Sci. 1984, 146, 155. Sheppard, N.; Nguyen, T. T. Adv. Infrared Raman Spectrosc. 1978,
Jin et al.
5936 The Journal of Physical Chemistry, Vol. 91, No. 23, I987
Pd,6 this process was given special consideration in this study. As shown in Figure 3e, however, the CO peak was reproduced in both position and intensity after prolonged exposure to CO at 573 K, indicating that no carbon was deposited on the surface by this treatment. This result is consistent with our TPD study8where the same conclusion was reached. The conclusions based on these observations and those reached by Danie116J7and by Bell6 regarding the mechanism of formation of C02differ and may be due to differences in the conditions used for preparing the surfaces. Thermal Stability of CO on Preoxidized Pt/Ce02. The observation (Figure 4) that adsorbed CO having a stretching frequency at 2091 cm-' was less stable to heating than that at the higher frequency, 2131 cm-', is especially noteworthy. A similar observation was reported by Primet13 for Pt/A1203,where the observation was made that the low-frequency CO peak disappeared faster, however, upon heating in oxygen gas. It is commonly assumed that the stronger the bond between the carbon and oxygen atom, the weaker the bond between the carbon and the metal atom. This interpretation has been supported by studies on oxygentreated Pd(100) surfaces38and for oxygen-treated supported R U , ~ i.e., that CO cobonded with oxygen to a metal atom was removed faster than CO bonded to a metal atom with an oxygen atom bonded to a neighboring metal atom. The opposite observation, reported here, can be explained in terms of a competition between the desorption process and the reaction between CO and O(a) on the surface. The reaction between O(a) and CO(a) was shown to occur below the desorption temperature for CO in Pt single-crystal s t ~ d i e s ~in~which , ~ ' O2 and CO were coadsorbed. The C02formed in these experiments desorbed at 370 K whereas the CO desorbed at 450 K. It is not possible to determine whether the removal of the higher frequency (2131-cm-') peak is due to reaction with O(a) or due to desorption. Our TPD experiments, in which O(a) was eliminated by flashheating pretreatments, found both desorption of CO and oxidation which consumed lattice oxygen. However, one can conclude that CO on oxidized Pt is oxidized less readily than CO on metallic Pt . Decomposition of C02on Pt/Ce02. The observation of CO IR absorption peaks after dosing with C02 for both untreated (evacuated) Pt/Ce02 and H2-reducedPt/Ce02 samples but not for oxidized Pt/Ce02 strongly supports the argument that the CO is produced chemically from the C02 and dues not arise from an impurity in the C02. This is in accord with the TPD measurements8 which found CO production on Pt/Ce02 surfaces exposed to C02and an earlier report on Pt/Ti02.34The formation of CO from C 0 2 is not related to the chemical state of the Pt since, as noted previously, flash heating removed oxygen from the surface (Figure 3) but did not recover the ability to form CO from C02 (Figure 5c,g). Thus, the decomposition is attributed to the chemical state of the support, Ce02. However, since no CO was formed when C02was exposed to pure Ce02 (Figures 6 and 7) and the CO which formed is bonded to the Pt exactly as if the CO were dosed directly (Figure 5), Pt is necessary for the reaction to occur. On the basis of these observations, we conclude that the active site is the interfacebetween the Pt and the CeOz, where lattice oxygen vacancies play an important role. A model for the surface reaction is
Ce02 ~~
'
(40) Okuhara, T.; Kimura, T.; Kobayashi, K.; Misono, M.; Yoneda, Y. Bull. Chem. Soe. Jpn. 1904,57,938. (41) Gland, J. L.; Kollin, E. B. Surf.Sei. 1985, 151, 260.
= Ce
=O
Figure 10. Pictorial representation of the interconversion of CO and C02 on Pt/Ce02 showing lattice vacancies and carbonate formation.
which is shown more pictorially in Figure 10. C 0 2 adsorbs at a lattice vacancy of Ce02and decomposes, provided a Pt surface is nearby to accept the CO, thereby filling the vacancy. The adsorbed CO can then migrate to a more stable step or corner site. On the other hand, CO dosed from the gas phase adsorbs on the Pt terrace, equilibrates among more stable corner or step sites, and migrates to the interface where it picks up a lattice oxygen and forms C02. Adsorption of C02on Pure Ce02. The carbonate/carboxylate structures on rare earth oxides are complex?2 Four bonding sites, labeled A through D, have been considered for COzon the Ce02
1670-1695 1310-1338 650-970
1590-1630 1260-1270 1020-1030
1560 1410
14.70-1530 1300-1370 1040- 1080
A
B
C
D
As seen in Figure 7, five IR absorption peaks, 1600, 1481, 1421, 1325, and 1040 cm-', are observed when C02 was adsorbed on Pt/Ce02 When the sample was heated under vacuum, three of the peaks at 1600,1325, and 1040 cm-' decreased synchronously and faster than the other two, 1481 and 1420 cm-l. Thus, on the basis of the heating experiments, we may divide the peaks into two groups: I and 11. We see that the group I1 peaks increased after prereduction and decreased drastically upon preoxidation of the sample. Group I peaks, however, were not changed by prereduction and reduced less upon preoxidation. These observations lead us to assign group I to the model B peaks and group I1 to the model C peaks, since the latter would be expected to be (42) Topchieva, K. V.; Spiridonov, S. E.; Loginov, A. Y. Russ. J. Phys. Chem. (Engl. Transl.) 1986,60,239. (43) Nakamoto, K. Infrared Spectra of Inorganic and Coordination Compounds; Wiley: New York, 1978; pp 116-169. (44) Miyata, H. Shokubaikoza;Kodansha Scientific: Tokyo, 1985; Vol. 3, p 238. . (45) Kung, M. C.; Kung, H. H. Catal. Rev.-Sei. Eng. 1985, 27, 425.
J . Phys. Chem. 1987, 91, 5937-5940 more sensitive to the existence of oxygen lattice vacancies and more stable to heating.& The weak peak at 1628 cm-’ observed when COz was adsorbed onto the preoxidized surface may be attributed to model A, which might be formed by the reaction of CO, with residual surface OH groups. The other bands assigned to model A would be hidden among those assigned to models B and C. N o evidence for the carbonate ion, model D, was found. It is interesting that the IR absorption in the carbonate region was stronger for pure CeOz (Figures 6 and 7) than for platinized ceria (Figure 5 ) . This is consistent with the COz TPD observation that more COz desorbed from pure CeO, than from platinized ceria at high temperature, 700 K. Although the pure CeOZhas a slightly larger surface area than the platinized ceria samples, the effect is too great for this explanation. One explanation, that the decomposition of COz,at the interface on the platinized samples fills the lattice vacancies where C02might have remained as carbonate had the Pt not been there, is consistent with the observations.
5937
5. Conclusions 1. C O is adsorbed on Pt supported by CeO, to form at least
three linear species interpreted as on terraces, steps, and kinks. 2. C O on metallic Pt is easier to remove (oxidize) than C O on oxidized Pt surfaces. 3. C02decomposes at the interface between Pt and the support, CeO,, to produce CO adsorbed on Pt and to fill a surface oxygen vacancy on the CeO,. The amount of decomposition depends upon the oxidation state of the local CeOZ interface. 4. COz is adsorbed on CeO, to form both carbonate and bidentate carbonate. The CO, adsorption is enhanced by prereduction, suppressed by preoxidation, and not affected by preadsorption of water.
Acknowledgment. This research was supported in part by the Office of Naval Research and by the Robert A. Welch Foundation. Registry No. Pt, 7440-06-4; CeOz, 1306-38-3; CO, 630-08-0; C02, 124- 3 8-9.
An Investigation of the Micropolarity of Several Aqueous Nematic Lyotropic Liquid Crystals V. Ramesh, Hai-Shan Chien, and M. M. Labes* Department of Chemistry, Temple University, Philadelphia, Pennsylvania 191 22 (Received: December 29, 1986)
The shift in the absorption maximum of methyl orange with solvent polarity is used as a probe to characterize the micropolarity of several aqueous nematic lyotropic liquid crystals: sodium decyl sulfate (SDecS), potassium laurate (KL), myristyltrimethylammonium bromide (MTAB), and disodium cromoglycate (DSCG). Lyophase aggregates of SDecS, KL, and MTAB appear to be less polar than the corresponding dilute micellar dilute solution aggregates. The DSCG lyophase is found to be a very high polarity microenvironment.
Introduction Sodium decyl sulfate (SDecS), potassium laurate (KL), and myristyltrimethylammonium bromide (MTAB) are among a group of surfactants that form nematic lyotropic liquid crystalline phases in specific concentration-temperature These lyomesophases are concentrated surfactant solutions composed of anisotropic micellar aggregates which are disklike (NL) or rodlike (Nc) in shape.4 In contrast to dilute micellar solutions in which a relatively porous cluster of surfactants are in equilibrium with dispersed monomers, the lyomesophase aggregates are both denser and larger in size.5 Recently, we have demonstrated that lyotropic liquid crystalline phases of SDecS, KL, and MTAB used as solvents for chemical reactions hold potential for reactivity control? Since lyomesophase aggregates are “extended micelles”, an understanding of their microenvironment with regard to both their micropolarity and microviscosity is of importance. The use of indicator dyes to probe the micropolarity of amphiphile aggregates is conventional in surfactant chemistry. Dodecylpyridinium iodide,’ bromophenol blue: chlorophenol red: (1) Yu, L. J.; Saupe, A. J. Am. Chem. SOC.1980,102,4879. (2) Yu, L. J.; Saupe, A. Phys. Rev. Lett. 1980,45, 1000.
(3) Boden, N.; Radley, K.;Holmes, M. C. Mol. Phys. 1981, 42, 493. (4) Forrest, B. J.; Reeves, L. W. Chem. Rev. 1981,81,1, and references therein. (5) (a) Fendler, F. M. Acc. Chem. Res. 1979,12,111. (b) Fendler, F. M. Acc. Chem. Res. 1980, 13,7. ( 6 ) (a) Ramesh, V.; Labes, M. M. J . Am. Chem. SOC.1986, 108,4643. (b) Ramesh, V.; Labes, M. M. Mol. Cryst. Liq. Cryst. 1987,144, 257. (c) Ramesh, V.; Labes, M. M. J. Am. Chem. SOC.1987, 109,3228. (7) (a) Mukerjee, P.; Ray, A. J . Phys. Chem. 1966,70, 2144. (b) Mukerjee, P.; Cardinal, J. R.; Desai, N. R. In Micellization, Solubilization and Microemulsions, Vol. 1, Mittal, K.L., Ed.; Plenum: New York, 1977; p 241.
phenolbetaine ET(30),10 merocyanine” and coumarin dyes,l2 2-(4-hydroxyphenylazo)benzoic acid,I3 and methyl orange14have been employed as polarity ,probes for micelles, inverted micelles, microemulsions, phospholipid bilayers, and polysoaps. In the present study, the sensitivity of the absorption maximum (kmax) of methyl orange (MO) to changes in solvent polarity has been used as a probe for the micropolarity of lyomesophase aggregates formed by SDecS, KL, MTAB, and DSCG in water. The structures of the surfactants, methyl orange, and reference compounds are given in Figure 1.
Methods and Materials M O (Aldrich) and DSCG (Fisons Inc.) were used as received. SDecS (Eastman Kodak) and MTAB (Aldrich) were recrystallized twice from 95% ethanollwater and dried under vacuum. KL was synthesized and purified by literature methods.” To ensure a constant pK, for MO, triply distilled water buffered to pH 7 (phosphate buffer) was used in the preparation of micellar and liquid crystalline samples. (8) (a) Funasaki, N. J . Colloid Interface Sci. 1977,60, 54. (b) Funasaki, N. J. Phys. Chem. 1979,83,1998. (9) Mackay, R. A.; Jacobson, K.; Tourian, J. J . Colloid Interface Sci. 1980,76,515: (10) Zachariasse, K. A.; Phuc, N. V.; Kozankiewicz, B. J . Phys. Chem. 1981,85, 2676. (11) deMayo, P.; Amiri, A. S.; Wong, S.K. Can. J . Chem. 1984.62,1001. (12) Fernandez, M. S.;Fromherz, P. J. Phys. Chem. 1977, 81, 1755. (13) Baxter, J. H. Arch. Biochem. Biophys. 1964,108,375. (14) (a) Kunitake, T.; Shinkai, S.;Hirotsu, S. J . Org. Chem. 1977.42,306. (b) Takagishi, T.; Nakata, Y.; Kuroki,N. J . Polym. Sci., Polym. Chem. Ed. 1974,12,807. (15) Saupe, A,; Boonbrahm, P.; Yu, L. J. J . Chim. Phys. 1983, 80, 7.
0022-3654/87/2091-5937$01SO/O 0 1987 American Chemical Society
