Particle size effects in the zero coverage regime have been discussed in the previous section. It is now an interesting question how the difference between the adsorption energies on the different systems changes with increasing oxygen coverage. To examine this effect, the oxygen coverage for the different particle sizes and Pd(111) was normalized to the total number of adsorbed oxygen atoms, shown in Figure 8.3 (a). The normalized coverageΘN(NO,ads) for the different
Pd nanoparticle sizes has been determined in the following way:
ΘN(NO,ads) = NO,ads
NPd,sur f (8.2)
NO,ads is the number of adsorbed oxygen atoms, which is dependent on the pulse number and NPd,surfis the number of Pd surface atoms, which is shown in Fig. 8.3 for the different Pd cover-ages.
The oxygen adsorption energy versus this normalized coverage,ΘN(NO,ads), for the investigated Pd/Fe3O4systems and Pd(111) is shown in Figure 8.7. The difference in the adsorption energy
Abbildung 8.7: Oxygen adsorption energy versus the normalized coverage,ΘN(NO,ads), for Pd nanoparticles of different sizes and Pd(111)
between the different systems is large in the zero coverage regime (70±17 kJ/mol). At high oxy-gen coverages, the difference in the binding energies for the different Pd surfaces become less pronounced. At saturation of the surface Pd sites, this difference in the binding energies decays to≈35 kJ/mol.
It has been shown with IRAS experiments, that oxygen initially adsorbs on low coordinated sites such as edge and corner sites. Accordingly, the strong particle size dependence of the oxygen adsorption energy at low coverages results from the different binding energies of oxygen on these low coordinated adsorption sites. After saturation of the edge sites, oxygen occupies the (100) and (111) facets on large nanoparticles. Adsorption on sites with weaker oxygen binding energies and prominent adsorbate-adsorbate interactions at higher oxygen coverages could be responsible for the smaller difference in the adsorption energy between the different Pd/Fe3O4 systems.
A significant higher binding energy on edge sites of Pd particles has not been found in earlier TPD studies [45, 99, 272]. This disagreement can be, besides to the limitations of TPD to probe oxygen adsorption energies, attributed to the fact that these studies were performed mostly at the oxygen saturation coverage. The major contribution to the desorption peak at oxygen saturation arises due to desorption from ordered adsorbate structures from facet sites on large nanopartic-les or sites with a similar adsorption energy in comparison to these sites on smaller particnanopartic-les.
The oxygen desorption energy on these sites is similar to that on the Pd(111) facet, which could explain why such effects have not been observed in TPD studies so far.
8.7 Summary
The coverage dependent adsorption energy and sticking coefficient of O2 has been measured on Pd(111) and on Pd nanoparticles with the Pd deposition coverage 0.6 Å (2.3 nm), 1.5 Å (2.9 nm), 4 Å (3.4 nm) and 7 Å (6.4 nm). The sticking probability and the adsorption energy, measured as a function of the coverage, on Pd(111) are in good agreement with the literature results. The coverage dependence of the sticking coefficient on the Pd nanoparticles shows a complex behavior. Saturation of the surface Pd sites has been determined at an oxygen coverage of 0.38±0.04, which is in good agreement with the findings of Henry et al. [77]. The adsorption energy of O2on Pd nanoparticles decreases strictly monotonically until the saturation coverage.
The initial adsorption energy of 275±14 kJ/mol on large Pd nanoparticles is significantly higher compared to the initial adsorption energy of 206±7 kJ/mol on Pd(111). With complementary IRAS measurements, it was found that this high initial adsorption energy results from a change of the adsorption site from threefold hollow on Pd(111) to edge sites on Pd nanoparticles. By comparing the oxygen adsorption energy on Pd nanoparticles of different sizes, a second effect has been found to influence the oxygen binding energy. The decrease of the Pd nanoparticle size results in a decrease in the oxygen binding strength which has also been measured for CO adsorption on Pd/Fe3O4. This general phenomenon can be explained with two microscopic ef-fects: weakening of the chemisorption interaction and reduction of the VdW-interaction due to a contraction of small Pd nanoparticles.
It has been observed that the difference in the oxygen adsorption energy between Pd nanopar-ticles of different sizes strongly decreases with increasing coverage. Accordingly, differences in the oxygen binding energy between the different systems are less pronounced at high oxygen coverages.
9 Interaction of oxygen with Pd particles at high oxygen exposures
At the lowest temperatures, oxygen adsorbs molecularly on Pd surfaces. Between 85 K and 200 K, dissociation takes place [94, 166, 237, 239, 246, 252, 254] and atomic overlayer structures are formed, which have been discussed in the previous chapter. At even higher temperatures and pressures, oxygen atoms diffuse into the substrate. Subsurface diffusion and surface oxide for-mation may occur and oxygen may form bulk oxides at high oxygen chemical potentials. The catalytic properties generally change with the oxygen concentration in the catalyst [160, 167].
In this chapter, oxygen adsorption on Pd nanoparticles at 300 K will be discussed. It will be shown with combined microcalorimetry and sticking coefficient measurements, that a large num-ber of oxygen atoms occupy sites which do not correspond to adsorption sites on surface Pd atoms. Oxygen could be quantitatively converted to CO2upon titration with CO at 300 K.
After showing the oxygen sticking probabilities on Pd/Fe3O4systems, the CO titration measu-rements on oxygen covered Pd will be discussed. Subsequently, the number of adsorbed oxygen atoms and evolved CO2molecules on Pd/Fe3O4catalysts are compared and discussed.
9.1 Oxygen sticking measurements on Pd particles
In this chapter, oxygen adsorption on Pd(111) and two kinds of Pd/Fe3O4systems will be discus-sed. Both Pd/Fe3O4systems have been prepared according to the procedure described in chapter 6. The first Pd/Fe3O4systems was directly transfered as prepared into the reaction chamber to perform the microcalorimetry experiments. This system will be referred to as [Pd/Fe3O4]1 in the following. The second substrate was exposed to≈1.1·1014O2molecules cm–2s–1for≈50 s and subsequently to a CO flux of∼1.5·1014CO molecules cm–2s–1for t=20 min at 490 K and annealed to T=580 K without CO exposure for t=5 min. The surfaces, which have been subject to this procedure once or twice are referred to as [Pd/Fe3O4]2 and [Pd/Fe3O4]3.
The sticking probability measurements for O2on Pd(111) and on Pd nanoparticles of different sizes, deposited on Fe3O4, have already been discussed in the previous chapter. Figure 9.1 shows the oxygen sticking probability as a function of the number of adsorbed oxygen atoms (NO). In these experiments, a chopper opening time of 266 ms, a pulse period of 2 s and a beam intensity of 1.1 1014molecules cm–2 s–1was used. The black squares, the red circles and the olive tri-angles show the sticking probability on the freshly prepared systems and after one or two cycles of O2and CO exposure at 300 K and cleaning in CO at 490 K before annealing to≈580 K.
Each plotted data set is the average of three to seven independent measurements. The error bars indicate the error of the mean.
As discussed above, S(0) on Pd(111) (black squares) is 0.47±0.03 and decays strictly mono-tonously with increasing coverage until the saturation coverage is reached. As the number of
Abbildung 9.1: Oxygen sticking probability plotted versus the number of adsorbed oxygen atoms on Pd(111) and on Pd/Fe3O4for three different Pd coverages. The black scatters show the sticking probabilities on the freshly prepared catalysts after annealing to 600 K. The red and green scatters show the oxygen sticking proba-bilities for the same catalysts, once or twice exposed to O2at 300 K and cleaning in CO at T=490 K before annealing to≈580 K.
adsorbed oxygen atoms reaches 0.45-0.5 1015cm–2, saturation of the Pd surface sites is reached on Pd(111).
Figure 9.1 also shows the coverage dependent sticking coefficient for O2after one (red circles) and two (green triangles) cycles of O2 and CO exposure at 300 K, extended CO exposure at T=490 K and annealing at T=580 K. It can be observed, that the coverage dependent sticking pro-bability is similar to that on [Pd/Fe3O4]1. This demonstrates the reproducibility of the sticking probability measurements.
On large Pd nanoparticles, the sticking probability decays gradually at low oxygen exposures. A more pronounced decay is observed as the number of adsorbed oxygen atoms increases. For the deposition thickness 1.5 Å, a pronounced decay of the sticking probability is already observed at the lowest oxygen exposures. The sticking data at a lower number of adsorbed atoms as the end of this prominent decay of the sticking probability is denoted as regime I. It has been argued in Chap. 8 that the number of adsorbed molecules in regime I corresponds to an average coverage
of 0.38±0.04 on the Pd nanoparticles. Similar values for the oxygen saturation coverage on Pd nanoparticles have been found in previous studies [77]. Accordingly, regime I can be assigned to the saturation of the surface Pd sites. In regime II, a slow decrease of the sticking coefficient with further exposure can be observed on [Pd/Fe3O4]1. For the deposition coverage 7 Å Pd, there is a clear separation between regime I and II on [Pd/Fe3O4]1. A more gradual transition between the two regimes is observed for the deposition coverage 4 Å Pd, and even more for 1.5 Å Pd. The total number of adsorbed oxygen atoms is determined to be 14.7 1014cm–2, 9.6 1014cm–2and 4 1014cm–2for the deposition coverages 7 Å (6.4 nm), 4 Å (3.4 nm) and 1.5 Å (2.9 nm), respectively.
The oxygen sticking probability is much different on [Pd/Fe3O4]2 and [Pd/Fe3O4]3 than on [Pd/Fe3O4]1 for all three particle sizes. The measured initial sticking probability is lower by 0.15, 0.1 and 0.12 for the Pd deposition coverages 7 Å, 4 Å and 1.5 Å. The determined total number of adsorbed oxygen atoms on [Pd/Fe3O4]2 and [Pd/Fe3O4]3 is significantly lower com-pared to the systems directly after preparation and amounts to 0.5 1014cm–2, 0.39 1014cm–2 and 0.16 1014cm–2for the Pd deposition coverages 7 Å (6.4 nm), 4 Å (3.4 nm) and 1.5 Å (2.9 nm).
The gradual decay of S(NO) in regime II cannot be observed on [Pd/Fe3O4]2 and [Pd/Fe3O4]3.
Therefore, the sites which are saturated in regime II on [Pd/Fe3O4]1 cannot be occupied on [Pd/Fe3O4]2 and on [Pd/Fe3O4]3.