J. CHEM. SOC. FARADAY TRANS., 1994, 90(9), 1285-1291 Modification of the Electronic Structure of Pd by U Films: Chemisorption of COT Thomas H.Goudert and Carlos A. Colmenares University of California , Lawrence Livermore National Laboratory, Livermore, CA 94550,USA X-Ray and ultraviolet photoelectron spectroscopies (XPS and UPS), Auger electron spectroscopy (AES) and thermal desorption spectroscopy (TDS) have been used to study the coadsorption of CO and U on polycrystalline Pd. We investigated the modification of the electronic structure of Pd at low U coverages and the mode of CO chemisorption (dissociative vs. associative) as a function of surface U concentration. U adsorption results in the narrowing of the Pd4d band. At low U coverages the density of states (DOS) at the Fermi level decreases and the U 5f electrons are localized. The CO saturation coverage at room temperature decreases with increasing U surface concentration, while CO adsorption at -165 "C is less affected. This is attributed mainly to a decrease of CO chemisorption energy but also to blocking of Pd adsorption sites by U.A decrease in the heat of chemisorp- tion for CO is explained by a change of the electronic structure of Pd by U. At low dosage U itself loses most of its reactivity, probably because of U-Pd solid-state bonding and below a critical U surface concentration CO dissociation becomes an activated process. High-temperature reaction between CO and Ucovered surfaces leads to the partial transformation of U into a surface oxycarbide.In the presence of this compound some CO is chemisorbed strongly on the surface and is stable even at 300°C,in contrast to CO on Pd metal. A similar effect has also been observed for classical promoters such as the alkali metals. One goal of fundamental catalysis research is to relate the that above monolayer coverage U overlayers interact catalytic properties of surfaces to their basic chemical proper- strongly with the Pd substrate and that surface alloying takes ties and ultimately their electronic structure. Surface reacti- place at relatively low temperatures ( <200 0C).9Therefore, it vity of catalysts is often discussed in terms of geometrical' is conceivable that the adsorption properties of Pd surface and electronic factors.* The latter relate catalytic properties atoms are also changed.to measurable electronic quantities (bandwidth, DOS at the Fermi level, orbital symmetry et~.).~In this context it has been argued that catalytic properties of the late transition Experimenta1 metals (TM) are due to an interplay of highly correlated d electrons and s-p electrons, which are found in broad bands. The measurements were performed using a double-pass cylin- This theory was introduced as the 'electron interplay drical mirror analyser (CMA). UPS measurements were made model'.4 Actinides may be interesting systems to test this using He I1 (40.81 eV) excitation radiation produced by a model because some of them have highly correlated f elec-windowless UV rare-gas discharge source.The total trons (U, Np, Pu) in metallic systems while others do not (Th, resolution in UPS was 0.2 eV. XP spectra were taken using Am).' Unfortunately, despite the similarities between actinide Mg-Ka (1253.6 eV) radiation with an approximate resolution 5f electrons and late TM d electrons the chemical properties of 1 eV. Auger spectra were measured by a lock-in technique of the elements are very different: actinides do not behave using 3 keV electrons. like late TMs but like early TMs owing to their descent from Thin layers of U have been prepared in situ by magnetron actinium which is a refractory metal. In earlier work we sputter deposition. This method was preferred to the conven- found the high reactivity of U manifested by the exclusive tional evaporation technique because of the difficulties associ- dissociative adsorption of CO on U metal even at -200 oC.6 ated with the latter: U has a low vapour pressure at the However we also found that U loses part of its reactivity melting point and has a strong tendency to alloy with most when diluted in an alloy such as UNi,.6 The solid-state filament or crucible materials.A UHV compatible, water- bonding between U and Ni seemed to decrease the affinity of cooled magnetron sputter head was operated at a target surface U for CO. Such decrease of surface reactivity has also current of 5 mA, a bias voltage of 300 V and a gas pressure of been observed for early transition metals (Zr)7 and the rare- 0.5-1.5 Pa. As sputter gas we used ultrahigh-purity Ar earth metals.' It provides the necessary condition for study- (99.9999%),which was further cleaned over hot Ca at 130°C.ing the catalytic properties of initially reactive elements. Deposition times varied from 1 to 10 s. The target and sub- In this paper we will discuss whether further dilution of U strate were kept at room temperature. A shield was installed in a TM matrix will suppress its reactivity sufficiently to to expose only the sample to the Ar-U plasma and keep U make it part of a low-reactivity catalyst. We studied the CO contamination of the chamber as low as possible. The target adsorption on a polycrystalline Pd surface doped with U. In was a high-purity U disc (30 mm radius and 3 mm thick), such a system we have to consider two independent effects: which was cleaned before introduction into the vacuum the reactivity of U surface atoms themselves and the changed system by mechanical polishing and nitric acid etching. The reactivity of Pd surface atoms.We showed in an earlier paper purity of uranium overlayers was checked by XPS and AES. Oxygen was tO.l atom% as determined by AES, while carbon was not detectable by either technique. t This work was performed under the auspices of the US Depart-The substre.te was an ultrahigh-purity, polycrystalline Pdment of Energy by Lawrence Livermore National Laboratory under foil. It was cleaned by sputtering (1 h at 1 keV, 5 x Pacontract No. W-7405-Eng-48. $ Present address : Commission of the European Communities, Ar and 10 mA current) at 600 K and flashing to 1100 K. This European Institute for Transuranium Elements, Postfach 2340, procedure was repeated several times until no impurities were D-7500 Karlsruhe, Germany. detected on the Pd surface by AES and UPS.1286 Results Clean U Overlayers on Pd We first determined how U deposits on the Pd surface in the coverage range below one monolayer (ML). In a previous study we found that at the multilayer coverage U accumu-lates on the surface, partially diffuses into the near-surface region and forms a near-surface alloy with Pd.9 Fig. 1 com-pares the U 4f spectra for several U coverages. The surface coverage was determined by the ratio of the U,,, :PdMNN Auger emissions. Interpretation of these data, however, depends on the way U deposits on the surface. If U sponta- neously diffuses into the near-surface region to form a near- surface alloy the dosage should be described in terms of U concentration. If U stays at the top surface forming an over- layer, U dosage should be described by partial coverage and overlayer thickness.The U 4f data (Fig. 1)indicate that at the lowest coverage (U : PdAEs= 0.23) U indeed stays on the top surface: the inelastic background is strongly suppressed as seen by comparing the intensity at the high binding energy (Eb) side with the intensity at the low E, side, drawn in Fig. 1 as a thin horizontal line below the high E, side. This points to the absence of inelastic scattering of the photoelectrons before escaping from the surface.Because in this paper we are mainly discussing systems with low U content we will describe the U dosage in terms of fractional coverage. We estimate it, using Seah's formula" and the low coverage data of Fig. 1 with an AES ratio of 0.23, to correspond to ca. 0.4 ML coverage. We compared these data to the XPS- derived surface coverage, determined by the ratio of the U 4f7,, : Pd 3d,,, areas,l0," and obtained also a surface coverage of about 0.4 ML. However, precise determination of the surface coverage is impeded by the uncertainty of the information depth, which for the UoVv transition lies between 0.6 and 4 ML. Therefore, we will not discuss data in terms of a precise coverage but will rather follow their evolution with coverage.Amore quantitative analysis, involving ISS data, will be reserved for a future publication. Further information on the surface U is obtained from linewidth, shape and binding energy of the U 4f lines (Fig. 1). With increasing U coverage the binding energy decreases from a value found for U 4f,,, in UPd, (390 eV)" to 388.5 eV, which is typical for U metal. We conclude that at low coverage, U-Pd interactions predominate with U atoms dis- persed on the Pd surface, while at high coverage U forms a thick overlayer of U metal. Further indications for atomic dispersion at low coverage are obtained from the linewidth of )I#, I,,, I,,, I,,, I,,, ,,,I ,,,, ,,,, r c.-u)C C .-J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 the U 4f. It is narrow at very low and very high coverage showing that in both cases U is found in one well defined chemical environment with U-Pd and U-U bonding, respectively. Agglomeration of U in clusters or islands would have produced non-equivalent U atoms with slightly shifted E, . This actually happens at intermediate coverage (for U: Pd AES ratio of 1.05 and 1.92) and results in the broadening of the U 4f emission. The fairly symmetric shape at low coverage indicates a low local DOS at the Fermi level (EF)for the U atoms as discussed by Doniach and Sunjic,13 which we attributed previously to the localization of the 5f electrons. With increasing U coverage the U 4f emission becomes asymmetric pointing to the delocalization of the 5f electrons at higher U concentrations.Fig. 2 shows valence band spectra with increasing U dosages. The 4d valence band of pure Pd has a maximum value at the E, . At low coverages, U deposition results in the decrease of this maximum. The Pd band narrows, the Pd 4d intensity close to E, is suppressed and the d-band centroid is shifted to higher E,. Such findings can be understood in terms of filling of the Pd 4d band and of diluting Pd in a surface U matrix. The driving force for this is provided by the filling of the Pd 4d band in Pd metal.I4 At high U coverage the intensity at the EF increases again. The emission at the E, is now mainly due to the 5f states of the electropositive U which, as in U metal and most of its alloys, are delocalized. At very high U dosage the Pd 4d peak becomes narrow and symmetrical, and the U 5f emission at the EF further increases.At this multilayer stage the surface consists of a concentrated U phase with Pd impurities atomically dis- persed in it.9 UPS Study of CO Adsorption at Room Temperature Fig. 3 shows UPS He I1 spectra of U/Pd surfaces on which 10 Lt CO was adsorbed at room temperature. Fig. 3 (a)-@) correspond to Fig. 2 (a)-@). On pure Pd CO is only adsorbed molecularly, as shown by the lines at ca. 7-7.5 and 11 eV, which are attributed to the 17c/5a and 4a molecular orbitals. CO is not chemisorbed dissociatively on Pd." With increas- ing U concentration the CO signal decreases showing that less CO is chemisorbed on the surface.This may be due to three factors. First, part of the surface may be covered by U, '1'1' 1 ' 1 I Pd4d U 5f ..A Pd4d:la) I.l.fIl I 10 8 6 4 2 0 E,IeV UPS He I1 spectra of U overlayers on Pd. U: Pd(AES) =Fig. 2 (a)l~~,~l~~~~l,,,,l,~,,l,,,,l,,',i,,,,0.00,(b)0.20, (c)0.30, (d)0.50, (e) 1.86 and (f)3.58.U coverage: (a)420 410 400 390 380 370 360 0.00,(b)0.37, (c) 0.56 and (d)0.93 ML. EbW Fig. 1 U 4f spectra of U deposited on Pd at room temperature. U : Pd (AES) = (a)0.23, (b)1.05, (c) 1.92 and (d)5.96. 1 L (Langmuir)= Torr s-l. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Ratio of the CO 40 areas at high (10 L) and low (2 L) exposure, as a function of U coverage partial U coverage co 40 (10 L)/CO 40 (2 L) 0.00 2.00 0.54 1.oo 0.81 1.20 1.35 1.63 ~~ A decreased ratio indicates an increase in sticking probability of CO.that we are comparing values in the steeper initial part of the h = 2.94eV (a-*.) L111''1111111'IIII'IIIIIII'IIII II'I adsorption curve (Fig. 5). The experimental results show the opposite to be true. The ratio decreases, which indicates an 141210 8 6 4 2 0 E,IeV Fig. 3 UPS He I1 spectra after adsorption of 10 L CO on U/Pd at room temperature. U coverage: (a) 0.00, (b)0.37, (c) 0.56 and (6)0.93 ML. on which CO is not chemisorbed, and which thus blocks Pd adsorption sites. Second, the heat of chemisorption of CO on Pd may be changed by U so that at room temperature CO is more weakly adsorbed.Third, U may decrease the sticking probability of CO on Pd such that 10 L of CO would no longer be sufficient to saturate the surface. Let us discuss this in more detail. Fig. 4 shows the evolution of the CO 40 peak area with increasing U coverage. A steep initial decrease is followed by a more gradual one with increasing U dosage. The non-linearity observed can be explained either by block- ing of Pd adsorption sites consisting of several Pd atoms (ensemble effect) or by an additional change of the chemi- sorption properties of Pd by U. For a surface consisting of non-equivalent Pd adsorption sites, an overall decrease in chemisorption energy would leave fewer chemisorption sites capable of binding CO at room temperature. We will address this issue in more detail when discussing TD spectra.So far the experimental findings point to the first and second cases discussed above. To test the third case we compared the CO chemisorption signals at two different CO exposures (Table 1). A decrease in sticking probability would result in an increase in the ratio of the intensity of the CO signals (high- exposure signalflow-exposure signal) because the effective dosage (total dosage x sticking probability) decreased such 1.o h cn c.-5 0.8 G v 0.6 2 2 0.4 0 0 0.2 0'"""'"""""''"""""0 0.5 1.o 1.5 U coverage/ML Fig. 4 Evolution of the CO 4a/VB area ratio with U coverage after adsorption of 10 L CO at room temperature increasing sticking probability.Such effect has also been invoked as one factor for the promoting effect of alkali metalsI6 on the catalytic properties of metal surfaces. When the surface is covered by ca. 1 ML U the CO chemi-sorption signal is very weak and two new peaks appear at Eb = 4 and 6 eV (Fig. 3). We assign them to the 0 2p emis- sion from chemisorbed atomic oxygen and their presence shows that CO is adsorbed dissociatively at high U coverage. To enhance the weak oxygen features we subtracted the sub- strate background in the spectra of Fig. 6 until the emission at the E, disappeared. This procedure is somewhat delicate because, as we will discuss below, the valence band (VB) of U-Pd is modified by the presence of CO. For the two lower U coverages this procedure leaves a broad structure between 1 and 5 eV, part of which is due to changes in the shape of the VB after CO adsorption. The shoulder at 4 eV, which is also observed at high U coverage, might be due to chemi- sorbed oxygen because the 0 2p emisson of 0 on Pd appears at Eb = 4 eV." In this case some of the CO would dissociate even at low U coverage, but the resulting atomic oxygen would mainly interact with Pd and not with U because for 0 chemisorbed on U E, = 6 eV.6*'* This points to a strongly suppressed reactivity of U, which initially has a much higher affinity for 0 than Pd [AformH"(Pd0) = -20.4 kcal mol- '; A,,,,H"(UO,) = -257 kcal mol-1].'9 At a nominal U coverage of 1.3 ML (Fig.6), the 6 eV emission actually appears now as 0 bonded to U.It would have been inter- esting to study the evolution of the 0 1s emission as a func- tion of U coverage, because this would have allowed us to distinguish dissociated from molecular CO unambiguously. CO effective dosage Fig. 5 Comparison of two intensity ratios (high CO dosageflow CO dosage) for low (lz/li) and high (h,/h,) sticking probabilities. A decreased sticking probability results in an increase of the ratio (h,/h, -= 4/11). Fig. 6 UPS He I1 spectra after adsorption of 10 L CO on U/Pd at room temperature. The metallic background has been subtracted. U coverage: (a)0.5, (b) 0.8 and (c) 1.3 ML. However, the Pd 3p3,2 core-level line superimposes on the 0 1s line and, because it undergoes core-level shifts after U deposition and reaction, cannot be easily removed by a sub- traction or a curve-fitting procedure.CO adsorption on pure Pd leads to a strong suppression of the emission at the EF [Fig. 2(a) and 3(a)].This has been attributed to the fact that the electrons in the upper Pd 4d band (the antibonding part) participate in the CO-Pd bonding.20 Consequently a high DOS at the EF favours strong CO-Pd bonding. With increasing U concentration the VB becomes less affected by CO adsorption as seen by comparing the corresponding spectra in Fig. 2 and 3(a)-(d). This is partially due to a decreased CO concentration with higher U coverage. However, a more quantitative analysis of the VB shows that the decrease in concentration alone is not sufficient to explain this evolution.We measured the varia- tion of the intensity at the Fermi level normalized to the VB area between 0 and 6 eV (Table 2). A loss in intensity after CO adsorption is only observed for pure Pd. At low U cover- age there is no change and at high U coverage the intensity actually increases. This latter effect may be understood in terms of breaking U-Pd bonds by CO. As we discussed above, U-Pd bonding itself results in a decrease of the emis- sion at EF. Direct U-(C, 0)bonding, which exists at high U coverage, results in a breaking of the U-Pd bonds and the local DOS at EF on the Pd atoms increases again; at high concentration U indeed reacts with CO. We conclude that the decreased sensitivity of the Pd 4d band to CO adsorption after U coverage points to a reduced CO-Pd interaction, which would be in good agreement with a decreased amount of CO stable on the surface at room temperature.Table 2 Variation of the DOS at E, with CO adsorption, as a func- tion of U coverage normalized co partial U coverage coverage (CO/CO,,) i, i, (il -iz)/(il + i,) 0.00 1.00 0.032 0.0206 0.76 0.54 0.35 0.0107 0.0107 0.00 0.81 0.25 0.00874 0.00847 0.06 1.35 0.1 1 0.00544 0.00606 -0.22 The intensity at E, normalized to the VB area, is measured before and after 10 L CO adsorption (il and i,, respectively). The normal- ized difference (il -iz)/(il + i,)decreases with increasing U coverage. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 UPS Study of CO Adsorption at Low Temperature In the previous section we saw that U deposition on Pd inhi- bits the CO adsorption, either by simply blocking the Pd adsorption sites or by changing the chemisorption properties of Pd.We performed low-temperature adsorption of CO on U/Pd surfaces, followed by gradual heating of the surface to obtain further information on this phenomenon. We studied CO adsorption for U coverages below and above 1 ML. U coverage was calculated assuming that all U atoms stay on the surface even for full coverage which, as discussed above, is not entirely true above 1 ML because some of the U diffuses into the near-surface region. Thus the U coverage of 1.5 describes a near-surface interdiffusion layer which still con- tains some Pd atoms. The UPS study shows that even for such high coverages CO is chemisorbed on the surface at low temperatures (Fig.7 and 8), but with increasing U coverage the CO signal becomes attenuated as shown in Fig. 7, where all three curves have been normalized to the U-Pd VB. This normalization procedure actually tends to suppress the CO emissions because, with increasing U concentration, the cross-section of the VB increases because it becomes more like the U VB cross-section [5f3 (6d7~)~], which is larger than that of the Pd 4d.21 Thus the amount of CO chemisorbed on U/Pd is slightly higher than suggested by Fig. 7. It is remarkable that CO is chemisorbed associatively on a surface that is nominally completely covered by U. It may be that at these temperatures low-reactivity U atoms may act as adsorption sites, but we have also to consider that this effect may be produced by some Pd adsorption sites left in the surface layer or by an overestimation of the U coverage (because of uncertainties as to the information depth).Notice that the 40 for 1.24 ML U is enhanced when compared to the ln/5a emission. This is due to some physisorbed CO whose 1n/5a emission superimposes on the 4a line of CO chemi-sorbed on U. Physisorbed CO was not observed at lower U dosages and under the experimental conditions used. The heat of chemisorption of CO decreases with increasing U concentration. This is shown by the disappearance of the CO signal when the surface is warmed up: for a nominal U coverage of 0.95 ML heating to 0°C still leaves a CO signal [Fig.8(a)], while for a U coverage of 1.5 ML a complete disappearance results at this temperature [Fig. 8(b)]. Further- more, the difference spectra show that for a U coverage of 0.95 ML CO is desorbed from the surface with annealing, while for a coverage of 1.5 ML the increase of the 0 2p signal with annealing shows that at least some of the CO disso-'Is 1lI~1~~~1~~~1~~~1~~~l~~,l,,,l 16 12 a 4 0 E,IeV Fig. 7 UPS He I1 spectra after adsorption of 10 L CO on U/Pd at -165"C. U coverage: (a)0.00, (b) 0.66 and (c) 1.24 ML. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 -8 "1,',1'''1'''"''1''' ,,, ,,T A I"'I"'I"'I"'I"'I"'I"' B Fig, 8 UPS He I1 spectra for adsorption of 10 L CO on U/Pd at -165 "C and subsequent annealing to room temperature.A, U coverage <1.0;(a) -150, (b) -100, (c) -50, (d)0 and (e)50 "C. B, U coverage > 1.0; (a) -165, (b) -150, (c) -100, (d) -50 and (e) 0 "C. The metallic background has been subtracted. ciates. This indicates that the reactivity of U increases with coverage. TD Study and CO Adsorption at High Temperature From the preceding discussion it is clear that CO chemisorp-tion on Pd is affected by the presence of U, which does more than simply block Pd adsorption sites. To get information on the heat of adsorption of CO on U/Pd we performed TDS experiments. We followed the evolution of the chemisorbed CO (CO 40 + 5a/ln) signal area when heating the sample on which we previously had adsorbed 2 L CO at -165 "C (Fig.9). Heating results in the sudden decrease of the CO signal. As we discussed above, this could be attributed (z priori either to CO desorption, at low U coverage, or to CO dissociation, at high U coverage. CO dissociation becomes the most likely process with increasing U concentration, because CO does not dissociate on pure Pd while on pure U it spontaneously decomposes. However, the U coverage used in the desorption experiment was below the threshold for CO dissociation, which lies between 0.95 and 1.5 ML nominal coverage [Fig. 8(a) and (b)].Thus we expect CO desorption rather than CO dissociation. For pure Pd we found a CO desorption temperature of 165°C(Fig. 9), which is in agreement with the literature.15 ~~~~~~~~~~~~~~~~~~'~~~~~~~~~~ -100 0 100 200 300 T/T Fig.9 TD spectra after adsorption of 10 L CO on U/Pd at -165°C. U coverage: (a) 0.00, (b) 0.25, (c) 0.75 and (d) 1.20 ML. Initial coverage, 2 L. Below this temperature the CO signal stays constant with increasing temperature, which shows CO to be adsorbed with one well defined energy, at least for a dosage of 2 L. With increasing U concentration the desorption temperature decreases gradually, which may be explained either by a geo- metrical effect, i.e. blocking of stable adsorption sites by U, or by a change of the overall chemisorption properties of Pd by U, which would be an electronic effect. The geometrical model assumes that there are different chemisorption sites at the surface, higher-coordinated ones with high CO chemi-sorption energy and lower-coordinated ones, onto which CO bonds less strongly.Partial coverage of the surface with U would result in a faster decrease of higher-coordinated adsorption sites than of the lower ones, which remain acces- sible to CO. CO would therefore be chemisorbed more weakly and hence the desorption energy is decreased. This model has been used to explain the decreased desorption temperature of CO on Au/Pd and Cu/Pd.22 However, it was shown that even for blocking 90% of the surface by Au the desorption temperature decreases only by 40°C. In our case covering the surface with U results in a decrease of the desorption temperature by up to 180"C, while the strong initial CO signal shows that more than 10% of the surface is available for CO bonding.Such a large decrease in the desorption temperature cannot be explained by a purely geo- metrical model and, therefore, indicates a change in the adsorption properties, probably the heat of chemisorption, of Pd by U. U indeed changes the electronic structure of Pd and, as we will argue below, this could very well modify its chemisorption properties. While the major effect of U is to decrease the desorption energy of CO on Pd, we also observed a secondary, opposite effect. When exposing a Pd surface, which was covered by 1.0 ML of U, to CO at 100°C part of the CO dissociates, as shown by the increase of the symmetrical 0 2p signal at 6 eV, indicating the formation of UO.Dissociation takes place because the U surface concentration is above the critical threshold coverage of ca. 0.9, above which U reactivity is high enough to induce CO dissociation; in fact, even if CO dissociation is an activated process it still occurs when CO is adsorbed at elevated temperatures. However, part of the CO does not dissociate but is adsorbed associatively. It is strongly bonded and heating even to 300°C does not result in its desorption, while on pure Pd desorption takes place below 200°C. This adsorbed CO does not change the desorption temperature for the majority of the surface CO in low- temperature adsorption-heating cycles, as discussed above Fig. 10 UPS He I1 study of high-temperature adsorption of CO.The nominal U surface coverage was 1.0. Tds= (a)and (c)-(e), -165 and (b) 100°C. Annealing temperature = (a)-(c), 100;(d)200 and (e) 300 "C. (Fig. 9), but the CO signal no longer drops to zero (Fig. 10). The C Is Ebin XPS is 287 eV, which is typical for chemi- sorbed CO. The signal at 13 eV in UPS is unexplained at present. If it is not due to a surface impurity building up during the repeated adsorption experiments, it has to be a product of the reaction of CO and the surface. Discussion We will discuss the observed changes in the surface reactivity of Pd with U coverage. The two main issues we want to address are the evolution in reactivity of the U surface atoms and the changes in adsorption properties of Pd caused by U.Reactivity of the U Surface Atoms Bulk U metal reacts immediately with CO to form a surface oxycarbide,6 which is a solid solution of UO and UC, where the 6d7s electrons participate in U-0 and U-C bonding while the 5f electrons remain largely non-hybridized. When U is deposited on Pd it loses much of this reactivity. At a nominal U coverage of 0.95 ML annealing leads to CO desorption instead of CO dissociation [Fig. 8(a)], which shows that for this coverage the activation energy for CO dissociation is higher than the desorption energy. At higher U concentration the activation energy for dissociation decreases, becomes smaller than the desorption energy and CO dissociation predominates over CO desorption. The dis- sociation product seems to be a U-0 complex as shown by the Eb of the 0 2p which is the same as for 0 chemisorbed on U [6 eV in Fig.8(b)]and differs from 0 chemisorbed on Pd (4 eV).17 The decreased reactivity of U may be attributed to the interaction between U and Pd, which places the initially highly reactive 6d7s electrons in stable solid-state bonds. The strong bonding interaction between U and Pd actually results in the decrease of the DOS at EF for both of the U and Pd atoms. However, at higher U concentration the emission at EF comes back and is now due to U 5f states (Fig. 2). The electronic structure of the U surface atoms begins to resemble J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 that of U metal and thus the reactivity increases.This experi- ment does not allow us to decide whether the decreased reac- tivity at low U concentration is directly linked to the decreased DOS at the Fermi level of the U atoms or whether it is due solely to the dilution of U on Pd. However, the decreased reactivity of U in UNi, ,6 where the 5f electrons are still delocalized, points to the dilution and solid-state bonding of U as the main factor. In addition, the high DOS at EFis due to the 5f states, which participate only weakly in bonding. Therefore, we would expect that in this case the DOS at EF plays only a secondary role. The experiments performed yield conflicting evidence about the role of U atoms as adsorption sites for CO. At low temperature, even for almost full U coverage of the surface, chemisorption of CO was observed, even though with some- what decreased intensity.On the other hand, even at satura- tion coverage the TD spectra did not reveal the existence of several adsorption sites with differing adsorption energies, as one would expect when both Pd and U act as adsorption sites. Instead, the desorption energy seems to decrease contin- uously with increasing U concentration and there is only one desorption peak [shown as a step, because we measured it in integrated mode (Fig. 9)]. It could be that the U atoms reside immediately below the surface, allowing Pd sites to interact directly with CO, even at nominal full U coverage on the surface. Adsorptive Properties of Pd U strongly affects the chemisorption properties of the Pd surface.The magnitude of this effect cannot be explained by preferential blocking of chemisorption sites, as was shown by comparing the U/Pd to the Au/Pd and Cu/Pd systems. Instead, the heat of chemisorption is lowered by U. This may be an electronic effect because even at low concentration U adsorption results in a measurable decrease of the local DOS at E, on the Pd atoms. This quantity has been directly related to the heat of adsorption and presented in Table 2, where the intensity at EF changes less with CO adsorption when the surface is precovered by U. The driving force for the bonding between U and Pd is indeed the filling of the Pd 4d band, which becomes a closed shell showing less ten- dency for interaction with CO.On the other hand, it has been argued that the heat of chemisorption of CO on late transition metals, will decrease when alloying the metal with electropositive elements having no d electrons (Si), thus resulting in the decrease of the CO adsorption energy. Alloy- ing with an electropositive element containing d electrons (i.e. an early TM) leads, in addition, to the weakening of the C-0 bonding because of the interaction between the TM d states and the CO antibonding 2n*.23Thus U may have two effects: it may lower the heat of chemisorption through bonding with Pd and weaken the intramolecular CO bonding by direct interaction. Decrease in the heat of chemisorption is not a local effect but it occurs through the changed electronic structure of the surface. Therefore, the presence of U does not create several different adsorption sites (there is one exception to this, as discussed below), but the adsorptive properties of the surface as a whole are changed.However, after high-temperature adsorption of CO we observe a secondary effect, which results in an increase in adsorption energy for some of the CO. Small amounts of CO stay on Pd at 300°C while on pure Pd it is desorbed below 200°C. Doping Pd with conventional promoters (K)also leads to strongly bonded C024 which is stable at high tem- peratures. Therefore, we think that by the high-temperature reaction between CO and the surface, some of the surface U is transformed into a promoter. Above a critical surface con- J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1291 centration U reacts with CO to UO (and UC), i.e. it is par- tially oxidized to U2+which could have similar properties to alkaline-earth-metal cations (K'). This would also explain why after high-temperature reaction we observe two chemi- cally different chemisorption sites, a majority of sites whose chemisorption energy decreases with increasing U concentra-tion and a minority population of adsorption sites with strongly increased chemisorption energy. 3 4 5 6 7 T. E. Madey, J. T. Yates, Jr., D. R. Sandstrom and R. J. H. Voorhoeve, in Treatise on Solid State Chemistry, ed. N. B. Hannay, Plenum Press, New York, 1976, vol. 6B, p. 1. Z. Knor, in Catalysis (Specialist Periodical Report), The Royal Society of Chemistry, London, 1985, vol. 7, p.148. T. Gouder, Ph.D.Thesis, Namur, Belgium, 1987. T. Gouder, C. A. Colmenares, J. R. Naegele, J. C. Spirlet and J. Verbist, Surf:Sci., 1992,264, 354. R. Hauert, P. Oelhafen, R. Schlogel and H-J. Guntherodt, Surf Sci., 1985, 160, L493. 8 J. N. Andersen, J. Onsgaard, E. Zdansk, A. Nilsson and N. Mir- Summary We studied the CO adsorption on Pd surfaces covered by U. 9 10 tensson, Surf Sci., 1989,217, 127. T. H. Gouder and C. A. Colmenares, Surf Sci., 1993,295,241. M. P. Seah, J. Catal., 1979,57,450. U itself loses its reactivity at low concentrations where CO dissociation becomes an activated process. This passivation, which is probably due to the involvement of the highly reac- tive 6d7s electrons in stable solid-state bonds, provides the necessary condition for investigating the catalytic properties of actinides.On the other hand, U modifies the chemisorp- tion properties of the Pd surface in two different ways. First a change of the electronic structure of Pd, in particular the filling of the 4d band, results in a decreased chemisorption energy for most of the CO. This is a global effect that affects all surface Pd atoms. Second, after high-temperature reaction, some of the surface U becomes oxidized and starts behaving 11 12 13 14 15 16 17 18 19 M. P. Seah, J. Vac. Sci. Technol., 1980 17, 16. Y. Baer, H. R. Ott and K. Andres, Solid State Commun., 1980, 36,387. S. Doniach and M. Sunjic, J. Phys. C, 1970,3,284. Y. Baer, in Handbook on the Physics and Chemistry of the Actin- ides, ed. A. J. Freeman and G. H. Lander, Elsevier, Amsterdam, 1984, p. 271. B. Oral, Y.C. Lee and R. W. Vook, Appl. Surf Sci., 1990,44,65. H. P. Bonzel, Surf Sci. Rep., 1987,8,43. H. Conrad, G. Ertl, J. Kuppers and E. E. Latta, Surf. Sci., 1977, 65, 245. T. Gouder, C. A. Colmenares, J. R. Naegele and J. Verbist, Surf Sci., 1990,235,280. Handbook of Chemistry and Physics, CRC Press, Roca Baton, like a classical promoter increasing the chemisorption energy of co. 20 FL, 54th edn., 1973. R. Hauert, P. Oelhafen, R. Schlogel and H-J. Guntherodt, Solid State Commun., 1985,55, 583. 21 J. J. Yeh, and I Lindau, At. Data Nucl. Data Tables, 1985,32, 1. References 22 G. A. Kok, A. Noordermeer and B. Nieuwenhuys, Su$ Sci., 1 2 W. M. H. Sachtler and R. A. van Santen, Adv. Catal., 1977, 26, 69. R. A. van Santen, in Fundamental Aspects of Heterogeneous Catalysis Studied by Particle Beams, ed. H. H. Brongersma and 23 24 1985,153,505. R. Hauert, P. Oelhafen and H-J. Guntherodt, Sure Sci., 1989, 220, 341. A. Berko and F. Solymosi, Surf Sci., 1986, 171, L498. R. A. van Santen, NATO AS1 Series, Series B: Physics, Plenum Press, New York, 1991, vol. 265, p. 83. Paper 3/06 1 19A ;Received 13th October, 1993