Reactions of CO and NO on Mg promoted cobalt

Reactions of CO and NO on Mg promoted cobalt

y..: .‘..‘.‘:.z+:~.‘.‘.:.:.:.:.:+:.. ..,......., ,:.: :.:.~~ ,.:..:. ):.~::(:: . ...,:.:.: :.:,:,:, ,.,( L1~:; “_~:,:~ ,:::: :,:~,: I’ .Y. .. ...

898KB Sizes 0 Downloads 2 Views



..A... . . . . :..I.?,>


..,......., ,:.: :.:.~~



. ...,:.:.: :.:,:,:,

,.,( L1~:; “_~:,:~ ,:::: :,:~,:

I’ .Y. .. .‘.:.i...,.....:..... .. . . .,...... .. .. “-:‘..:.:::.:::::::::,:~:: . . ..i.../... ~.,./,






surface science ELSEVIER

Applied Surface Science 89 (199.5) 103-l 11

Reactions of CO and NO on Mg promoted cobalt T. Vaara, J. Lahtinen *, P. Hautoj$vi Laboratory of Physics, Helsinki University of Technology, Fin-02150 Espoo, Finland

Received 28 November 1994; accepted for publication 22 February 1995

Abstract Nitric oxide adsorption and coadsorption with carbon monoxide have been studied using XPS and TDS on clean and magnesium covered polycrystalline cobalt. At small NO exposures, the adsorption on clean cobalt at room temperature was dissociative but at exposures over 3 L molecular adsorption was also observed. Submonolayer coverage of Mg had a negligible effect on the adsorption of NO, but a full monolayer coverage decreased both the molecular and the dissociative adsorption of NO. In the NO-CO coadsorption the NO molecules block the adsorption of CO and induce removal of the preadsorbed CO molecules on the clean and Mg covered cobalt. The presence of Mg stabilizes the CO molecules against the attack by NO molecules.

1. Introduction Reduction of nitric oxide to nitrogen is an environmentally important and widely studied application of catalysis. The need for low cost catalyst materials has given a motivation for a number of surface science studies on NO adsorption and NO + CO reactions on metal surfaces. E.g., nickel surfaces have extensively been studied but only little attention has been given to cobalt. NO adsorbs mainly dissociatively on most of the group VIII metals at room temperature while the adsorption of CO is expected to be mainly molecular at room temperature [l]. On the Ni and Co surfaces molecular adsorption of NO takes place as the exposure is increased. On Co(OOO1) NO exposures above 9.5 L (L = Langmuir unit) have been reported to

* Corresponding author. Fax: +358 0 451 3116; E-mail: [email protected] 0169-4332/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0169-4332(95)00038-O

yield molecular adsorption [2] but on Ni(100) exposures from 0.8 to 3 L have been sufficient [3,4]. The change from dissociative to molecular adsorption has been assigned either to an increase in the activation energy for dissociation due to the preadsorbed and dissociated NO [5] or simply to a decrease in the amount of sites on which NO molecules could dissociate [6]. Chen et al. also observed significant N-O bond strengthening due to oxygen in a NO-O coadsorption system on Ni(100) [7]. The strengthening indicated reduced electron density in the 27r * antibonding orbital of NO caused by the interaction of NO with the coadsorbed electronegative oxygen atoms. Promoter effects on NO adsorption have not been widely studied in the literature. An enhancement in the dissociation of NO on Rh(100) in the presence of K promoter has been reported, however [81. On Mg surfaces the adsorption of NO is dissdciative already at 77 K [9]. Bridge and Lambert have found that CO adsorbed on Co(OOO1) was oxidized by NO, resulting in the


T. Vaara et al. /Applied Surface Science 89 (1995) 103-l I1

desorption of CO, at room temperature [2]. On a Ni(100) surface NO displaced the preadsorbed CO molecules [lo] and on Ni(ll1) the NO molecules pushed CO molecules from bridge sites to weakly bound on-top sites at 85 K Ill]. CO and NO molecules were assumed to compete for the surface electron density available for the backdonation into their antibonding 27~* orbitals. On Pt and Rh surfaces the adsorbate induced reconstruction plays a significant role during the NO and CO adsorption and reactions [12-161. We have previously studied the effects of Mg and MgO on the CO adsorption on cobalt single crystals and foils [17-211. Magnesium was found to increase the adsorption capacity of CO on cobalt foils because of the strong bonds formed between oxygen and magnesium. The observed effects on CO adsorption suggested that metallic Mg could significantly alter the adsorption of NO and also the reaction between NO and CO on cobalt foils. However, it is known that too strong binding between oxygen and the substrate inhibits NO + CO reactions because in a good NO oxidation catalyst NO should dissociate but CO should remain molecular [22]. In this work we report results from the NO adsorption and NOCO coadsorption experiments on Mg promoted cobalt. The aim of this paper is to answer whether the addition of Mg could be useful in the NO reduction-CO oxidation catalysts.

sured using a K-type thermocouple spot-welded to the back of the sample. The temperature was controlled by a PID-algorithm running on a PC-AT computer. Mg deposition was carried out with a resistively heated effusion cell and a manual shutter. The submonolayer coverages used in the experiments were produced by controlling the evaporation time and keeping the sample temperature at 550 K. A detailed description of the growth method has been given earlier by Vaari et al. [19]. The amounts of desorbing molecules in the TD experiments and during the gas exposures were monitored by a quadrupole mass spectrometer equipped with an off axis secondary electron multiplier. The TDS data were analyzed with the Redhead’s peak maximum method [23] and the desorption orders were determined from the rate constant versus inverse temperature plots as proposed by Cabrera [24]. In the XPS measurements a hemispherical energy analyzer and AlKcv radiation were used throughout the experiments. The XPS data was analyzed using a peak deconvolution program, utilizing a non-linear background subtraction algorithm by Proctor and Sherwood [25] and Powell’s direction-set fitting method [26] for peaks of Gaussian-Lorentzian type. The equipment was enclosed in a custom designed UHV chamber with the base pressure in the lo-‘”

2. Experimental The sample was a 0.125 mm thick polycrystalline cobalt foil with a nominal purity of 99.99%. The only observed impurities after initial sputtering were carbon and oxygen segregating to the surface. Carbon was titrated off with NO exposures followed by heating to 1050 K resulting in CO, CO, and N, desorption. Oxygen impurities were removed by cycles of Mg evaporation to attract oxygen to the surface followed by sputtering cycles with 1 keV Ar+ ions and annealing at 800 K [17]. Several small areas giving LEED patterns with (0001) symmetry were detected after the cleaning-annealing cycles. The sample was spot-welded to 0.25 mm tungsten wires through which the current for sample heating was conducted. The sample temperature was mea-

e 05





b 400


J 3910






Fig. 1. XP spectra of (a) the N 1s and (b) the 0 1s lines from a cobalt foil exposed to various amounts of NO. The components corresponding to molecularly and dissociatively adsorbed NO are marked with m and d, respectively. The solid lines have been drawn to guide the eye. Spectra from the Mg covered Co after NO saturation are shown with dotted line.

T. Vaara et al. /Applied Surface Science 89 (1995) 103-111

Torr range. A detailed description of the system can be found elsewhere [27].




NO on


3. Results 3.1. NO adsorption on cobalt Nitric oxide adsorption was studied with XPS by recording the 0 Is, N 1s and Co2p spectra as a function of NO exposure. The N 1s photoelectron line from clean cobalt after various NO exposures is shown in Fig. la. At small exposures the Nls line exhibited one component at the binding energy (BE) of 397.4 eV, gradually shifting upwards as the NO exposure was increased. The BE reached a value of 397.6 eV above 3 L exposures. The N 1s shift could be related to the stronger binding of N to the substrate. An additional shoulder around 399.6 eV emerged after 3.0 L exposure. Because NO is known to adsorb mainly dissociatively on several transition metal surfaces at room temperature [2,6], but at large exposures molecular adsorption has been observed as well, we assigned the high BE peak to molecularly adsorbed NO and the low BE peak to adsorbed N atoms. In the exposure range between 8 and 50 L no further increase in the N 1s components was observed. The 8 L exposure can thus be considered as a saturation exposure. In the XP spectra of the 0 1s line in Fig. lb a similar two component feature can be seen as in the N 1s line. The binding energies of oxygen in molecularly and dissociatively adsorbed NO were 531.3 + 0.3 and 529.4 f 0.1 eV, respectively. In addition, there is a detectable amount of subsurface oxygen even immediately after the cleaning procedure appearing in the 0 1s spectra at 532.3 eV. The interpretation of the 0 1s data after large NO exposures is also affected by the bulk diffusion of oxygen. At 50 L NO exposure there is a considerable increase in the 0 1s component at 532.3 eV while no increase in the N 1s line is seen indicating that the excess amount of oxygen on the surface is transferred to the subsurface area. The intensity comparison of the high and low BE components in the N 1s data showed that after saturation exposure of NO approximately one fourth of the nitrogen on the surface was present as molecular



TEMPERATURE (K) Fig. 2. TD spectra of (a) m/e = 28 (N,) and (b) m/e = 30 (NO) from a clean cobalt foil after various NO exposures. Molecular NO desorption takes place around 390 K and nitrogen desorbs at 640 and 770 K. The 640 K peak shifts to 660 K as NO exposure is increased. A heating rate of 5 K/s was used.

NO. The 0 1s peak deconvolution corrected with the subsurface oxygen background gave essentially the same value indicating that the diffusion of oxygen into the bulk does not take place in the O-8 L exposure range. The total O/N intensity ratio of 2.1 showed that some nitrogen atoms desorbed immediately after adsorption. Similar finding on Ni(ll1) has been reported by Breitschafter et al. [6]. The TD spectra after NO adsorption are shown in Fig. 2. Desorption of N, at m/e ratios of 28 and 14 was detected at 640 and 770 K (with 1 K/s heating rate). Both the Nz desorption features were of the first order as confirmed by the desorption rate versus inverse temperature plots [24]. The intensity of the 770 K peak stayed constant in the exposure range from 0.2 to 40 L. The peak at 640 K began to develop after 0.7 L and its position shifted 20 K upwards with increasing NO exposure. At an m/e ratio of 30 a small first-order peak was detected at 390 K indicating molecular NO desorption with an activation energy of 100 kJ/mol assuming a pre-exponential factor of 1 X 1013 s-l. After saturation exposure of NO this signal represented approximately 5% of the total yield of the desorption products.


T. Vaara et al./Applied

Surface Science 89 (1995) 103-I I1

3.2. The efect of Mg on NO adsorption The N 1s and 0 1s spectra measured after saturation exposure of NO from 0.5 monolayer (ML) of Mg on Co are shown in Fig. 1. No changes due to Mg were found in the N 1s and 0 1s components corresponding to molecular NO but a 5% increase in the surface oxygen component was seen though no change in the corresponding N 1s spectrum was observed. The amount of nitrogen remained essentially constant at all Mg coverages up to 0.7 ML but the 01s peak increased slowly. At 1.0 ML a 15% decrease was observed in both components of the N 1s and the 0 1s peaks. Thermal desorption of N, (m/e = 28 and 14), NO (m/e = 301, and Mg (m/e = 24) were measured (with 5 K/s heating rate) after saturation exposure of NO. Because some background desorption of CO was detected, the m/e ratio of 14 was used in the analysis and it is shown in Fig. 3. According to the rate constant versus inverse temperature plots, all the desorption peaks were of the first order. The N, desorption peaks at 630 and 760 K stayed at constant temperatures at all Mg coverages and the total N2 peak area remained essentially




~g~co(Pow) NO 6.0 L

WwPolY) NO 8.0 L













Fig. 3. TD spectra of (a) m/e = 30 (NO) and (b) m/e = 14 (N2 fragments) at various Mg coverages after saturation exposures of NO. The NO desorption peak shifts with the Mg coverage and the peak area decreases. The total desorption yield of nitrogen decreases by 5% when the full Mg monolayer is reached. The N, desorption temperature remains constant. A heating rate of 1 K/s was used.

F 3




2 5


wCo(Pw em = 0.5



Mg 1s

Q z




- i,:

I( 0








NO EXPOSURE (L) Fig. 4. Mgls binding energy on cobalt foil as a function of NO exposure. The maximum value is obtained at 0.1-0.2 L NO exposure after which the BE returns to the level of metallic magnesium.

constant below one monolayer coverages but decreased by 5% at the full monolayer of Mg in accordance with the XPS data. The NO peak position shifted from 375 + 2 to 395 k 3 K as the Mg coverage increased giving an activation energy increase from 101 to 106 kJ/mol. Molecular NO desorption composed less than 5% of the total desorption yield. At low Mg coverages, the NO desorption peak area remained essentially constant but decreased by 10% at the Mg coverage of 1 ML in accordance with the XPS data. Mg desorption started near 1000 K and its desorption temperature shifted upwards as the NO/Mg ratio on the surface was increased, and finally no Mg could be desorbed from the surface below 1000 K. The effect of NO on the chemical state of Mg was studied by exposing the 0.5 ML Mg layer on Co to various amounts of NO. The change in the binding energy of Mgls as a function of NO exposure is shown in Fig. 4. Already 0.1 L was sufficient to increase the Mg 1s binding energy by 2.4 eV with a simultaneous increase in the peak intensity. As the NO exposure was further increased, the binding energy decreased and levelled off to 1303.1 eV after approximately 2 L exposure. The peak intensity of Mg 1s remained at the maximum level which was 1.3 times the metallic peak intensity. The atomic ratio of 0 to Mg was below 0.5 when the binding energy maximum was observed. Further increase in the NO exposures made the O/Mg ratio to approach unity.

T. Vaara et al. /Applied Surface Science 89 (1995) 103-I I1

3.3. Coadsorption of CO and NO on the clean cobalt The effect of preadsorbed NO on the adsorption of CO, and vice versa, were studied on the clean cobalt foil and on the cobalt foil with a 0.5 ML magnesium coverage. The C 1s spectra after CO saturation followed by various amounts of NO are shown in Fig. 5a. Two peak components are easily resolved. The high binding energy component has been assigned to molecular CO and the low binding energy component to dissociated species [21]. The C 1s peak corresponding to molecular CO shifted from 285.6 to 284.7 f 0.2 eV and its intensity decreased rapidly with the increasing NO exposure. The C 1s component assigned to surface carbon shifted from 282.9 to 283.8 f 0.2 eV and its intensity increased slightly with the increasing NO exposures. The intensity analysis of the carbon peaks revealed that on the clean cobalt 1% of the preadsorbed CO was dissociated but after a 0.2 L exposure of NO, removal of CO was observed and some surface carbon was detected. Less than 5% of the CO molecules was left on the surface after saturation exposure to NO.

c 16

4 L of CO+NO on





The 0 1s spectra had a similar two component structure as the C 1s. The integrated 0 1s peak area had a shallow minimum around 0.02 L exposure of NO after which the intensity increased indicating rapid oxygen removal at the onset of the NO adsorption most probably due to reaction between adsorbed CO and NO followed by NO adsorption at higher exposures. After CO saturation followed by saturation exposures of NO, the intensity of the 0 1s peak was the same as after saturation of NO on clean Co but a factor of two higher than after CO saturation on Co. The thermal desorption spectra of m/e = 28 after 1 L of CO followed by exposures of NO indicated that already 0.2 L of NO was able to extinguish the desorption of CO molecules. CO, formation was observed after 0.2 L NO exposures around 560 K and its maximum was reached at 1.0 L. The amount of CO, was approximately 15% of the total amount of carbon oxides. The effect of adsorbed NO on the adsorption of CO was studied by recording the XP spectra after various exposures of NO followed by 10 L of CO. When small exposures of NO were used, adsorption and dissociation of CO were observed but already a 0.05 L exposure of NO was able to suppress the molecular CO adsorption by a factor of seven. The amount of dissociated CO showed a maximum at the same NO exposure. Adsorption of CO after saturation exposure of NO had no effect on the 0 Is, N 1s or C 1s lines.


3.4. The effect of Mg on the coadsorption of CO and 0.06




1 .oo I












BINDING ENERGY (ev) Fig. 5. Cls spectra after CO saturation followed by subsequent NO exposures on clean cobalt. The high BE component of the Cls line shifts from 285.6 to 284.7kO.2 eV and the low energy component from 282.9 to 283.8 f 0.2 eV. The components corresponding to molecularly and dissociatively adsorbed CO are marked with m and d, respectively. The solid lines have been drawn to guide the eye.

A magnesium coverage of 0.5 ML was used to study the effect of Mg on the coadsorption of CO and NO. The same XP and TD spectra were recorded as in the case of coadsorption on the clean surface. The C 1s component corresponding to molecular CO and the total amount of adsorbed carbon decreased with the NO exposure given after CO saturation but even large NO exposures were not able to remove all the molecularly adsorbed CO. The amount of remaining CO molecules was approximately 25% of the original molecular CO content. The amount of dissociated carbon increased slightly due to small


T. Vaara et al. /Applied



8.0 L co

Surface Science 89 (1995) 103-111






TEMPERATURE (K) Fig. 6. m/e = 28 desorption signals from the Mg covered cobalt foil after exposure of CO followed by various exposures of NO. The molecular CO desorption is extinguished by the NO doses. The CO desorption at 630 K is not affected by NO. A broad nitrogen desorption feature below 780 K dominates the spectrum at 1.0 L NO exposure. The desorption spectra of CO, shown with dotted line indicate a desorption temperature of 560 K. A heating rate of 5 K/s was used.

NO coverages. The binding energy of the C 1s peak of molecular CO remained constant at 285.2 f 0.2 eV, in contrast to the case of coadsorption on clean cobalt, but the C 1s peak of dissociated CO shifted from 283.2 to 283.7 eV due to NO exposure. The thermal desorption data after CO saturation followed by NO exposures are shown in Fig. 6. The desorption of CO at 360 K was extinguished, the intensity of the 600 K peak remained essentially constant and the desorption of N, below 780 K became gradually visible as the NO exposure increased, The total amount of desorption products decreased up to 0.2 L of NO after which an increase due to nitrogen desorption was observed. Part of the dissociated carbon was removed from the surface as CO at a temperature of approximately 1000 K simultaneously with the desorption of metallic magnesium. Exposures of NO exceeding 1 L lead to oxygen and Mg residuals at the surface after heating to 1050 K. The TDS data of CO desorption shown in Fig. 6 indicated the formation of carbon dioxide at 560 K in a first-order desorption process. The amount of desorbing CO, increased with the exposure of NO up to 1.0 L. A 15% fraction of the surface carbon was removed as CO,.

The intensities of the C 1s components corresponding to both molecular CO and carbidic C decreased linearly with NO pre-exposures followed by 10 L CO exposure suggesting direct site blocking by NO. The binding energy of the molecular C 1s peak remained constant around 285.1 _+0.2 eV and that of the dissociated peak at 283.1 f 0.2 eV. The pre-exposures of NO above 0.2 L lead to total suppression of the CO.

4. Discussion 4.1. NO adsorption The XPS data shown in Fig. 1 implied that NO molecules can adsorb on a clean cobalt surface both dissociatively and molecularly with a sticking coefficient near unity in agreement with previous studies on cobalt and other transition metals, especially Ni [6,28-331. The change from dissociative to molecular adsorption at large exposures of NO could have been due to an increase in the activation energy for dissociation prompted by e.g. the interaction of NO with the preadsorbed oxygen from the dissociation of NO, or to the decrease in the amount of free sites to which NO molecules could dissociate. The XPS binding energies of N 1s and 0 1s of both the molecular and dissociated components were the same as those measured on nickel surfaces, within the experimental error [6,29-311. The N 1s and 0 1s binding energy values of 399.6 and 531.3 eV, corresponding to NO on cobalt foil, were 1 eV lower than the results reported by Fukuda [28] but the binding energy differences between the molecular and the dissociated N 1s and 0 1s components were the same. It is probable that the difference in the absolute values was due to different calibration of the instruments. The cobalt surface was observed to adsorb more NO than CO, which was most likely due to the preference towards dissociative adsorption, because the adsorption capacity of the molecular adsorbants is restricted by the long range molecular repulsions. Assuming that the substrate atom density for the polycrystalline foil equals that of the Co(OOO1)surface and scaling with the experimental CO coverage of l/3 [2], the maximum NO adsorption capacity of

T. Vaara et al. /Applied

Surface Science 89 (I 995) 103-l 1 I

approximately 0.7 NO molecules per Co atom is obtained. On Ni(ll1) the saturation coverage of NO was approximately 0.5 per Ni atom [6]. The NO adsorption capacity of the cobalt surface was not affected by the presence of submonolayer coverages of Mg but a full monolayer of Mg decreased the total adsorption capacity by approximately 15%. This decrease could be due to the formation of a dense Mg layer close to 1 ML (in analogy with Mg on Ru(0001) [34]) on which the NO adsorption capacity was smaller than on the loosely packed layer. The amount of oxygen from NO adsorption on 1 ML Mg on Co is approximately the same as the amount of oxygen from CO on the same surface [21]. A full monolayer of Mg prevented the surface oxygen from diffusing into the subsurface area in contrast to clean cobalt. This was probably due to the larger reactivity of the oxygen to magnesium than to cobalt 1171. Only 5% of the NO molecules desorbed as NO from clean cobalt though the XPS measurements indicated that 20-25% of the NO adsorption was molecular. The difference between the XPS and the TDS results suggested a temperature driven dissociation of NO during TDS. Thermally induced dissociation of CO has been observed in the same system [21] and the probability of a thermally driven dissociation of NO should indeed be larger than that of CO simply due to the smaller dissociation barrier of NO. Removal of nitrogen occurred after adsorption, and because of the low reactivity of nitrogen, some of the N atoms were most likely desorbed from the surface rather than diffused into the Co bulk. The resulting unbalance between the amounts of N and 0 is similar to the one that has been reported by Breitschafter et al. on Ni(ll1). TDS results showed that the binding of nitrogen to the surface was increased as the NO coverage was increased. This was probably due to the interaction between N and the surface oxygen. The 0.2 eV binding energy shift observed in the N 1s line also showed a change in the interaction between N and the other surface compounds. In the 0 1s line such a shift could not be resolved, however. 4.2. Coadsorption

of NO and CO

The amount of molecular CO on the clean cobalt surface decreased as a function of NO exposure as


can be seen from the C 1s data in Fig. 5. According to both the XPS and the TDS data, there was a minimum in the oxygen content on the surface around 0.2 L NO exposure. The observed minimum can be assigned to the removal of CO as CO, from the surface during the adsorption of NO at room temperature. However, the NO exposures were not sufficient to remove all of the molecular CO. The remaining fraction of CO can be explained by the inevitable roughness of the polycrystalline foil which provides multiply coordinated sites with stronger CO-M bonds as seen in our earlier studies [21]. The replacement of CO by NO is in agreement with the results obtained on Co, Ni and Pd on which CO was replaced from the surface by the incoming NO [2,10,35]. According to XPS there was a large shift in the C 1s binding energy of the CO molecules as the NO coverage was increased. Charge transfer between the CO and NO molecules (or NO dissociation products) could produce the observed effect, but also the effect of oxygen on the charge transfer between CO and the metal surface could explain the shift. This is in qualitative agreement with the earlier studies of NO on Ni surfaces and coadsorbed oxygen on Rh surfaces where the presence of oxygen reduces the backdonation to the 27r * orbital of CO because of the larger electronegativity of free oxygen [11,36]. The blocking of the adsorption of CO by NO seems to contradict the findings of Bridge and Lambert on NO + CO reactions on the Co(OOO1)surface [2]. Their AES and TDS analysis showed that low ( < 3 L) pre-exposures of NO followed by greater CO exposures resulted in the desorption of CO and N, during heating and in the complete oxygen removal from the surface. Large NO pre-exposures (> 10 L), however, prevented the CO induced oxygen removal. The amount of oxygen was measured after heating the sample to 900 K, while our measurements were performed at room temperature. The removal of oxygen by bulk diffusion during heating could partially explain the differences between their observations and the current data. Molecular CO was observed with XPS on Mg covered cobalt even after large NO exposures, but according to TDS, CO desorption could not be detected after 0.2 L of NO. The amount of desorbed CO, scaled well with the XPS intensity of the CO


T. Vaara et al. /Applied

Swface Science 89 (1995) 103-l I1

molecules left on the surface after the saturation exposure of NO. It is probable that thermally driven formation of CO, from the remaining CO molecules with the excess oxygen from NO dissociation occurred. The observed C 1s binding energy shifts of molecular CO also suggested a reaction between the adsorbed species, which could either indicate a formation of a reaction precursor to CO, desorption or decrease the charge (backldonation between CO and the metal, which again could result in an increased reactivity between CO and adsorbed oxygen. The binding energy shift of the CO molecule as a function of NO dose was not observed in XI’S when Mg was present. This difference could arise from the fact that the oxygen preferentially reacted with the Mg layer and was not affecting the backdonation to CO molecules. We thus propose that the excess surface oxygen interacting with the remaining CO molecules form CO, during the TDS on clean and Mg promoted cobalt. The amount of adsorbed CO molecules, which were neither dissociated nor desorbed from the surface because of the NO exposures, was a factor of two higher on the Mg covered surface than on the clean surface. The submonolayer coverages of Mg thus resulted in the increased stabilization of the CO molecules against the NO attack. The amount of tightly bound CO molecules is comparable to the area of the 600 K TDS peak on the Mg/Co surface in Ref. [21] and was interpreted to originate from the interaction of adsorbed CO and oxygen reacting with Mg. Accordingly, the Mg induced stabilization seems to originate both from the Mg providing more tightly bound multiply coordinated CO adsorption sites by the edges of the Mg islands and from the Mg reacting with the free oxygen which otherwise would decrease the 27~* backdonation between CO and the metal surface. The binding energy of the Mg 1s peak was observed to shift as a function of the NO exposure as shown in Fig. 4. Similar binding energy shift of Mg 1s has been observed due to the 0, adsorption on the same Mg/Co system [19]. The amount of oxygen required to achieve the Mg 1s binding energy maximum was 0.2 L at an Mg coverage of 0.5 ML which is approximately the same as the required exposure of NO to achieve the same binding energy value. CO has also been found to increase the Mg 1s

binding energy on cobalt but the BE value does not return to a lower level at large exposures [21]. The apparent downward shift of the Mgls peak as the oxidation proceeds is probably due to the shadowing effects brought by the coadsorbed oxygen atoms [20]. The observed BE shift supports the assumption of a strong interplay between the Mg and 0 atoms and the Mg induced stabilization of CO molecules. In a reaction with NO, the magnesium overlayer becomes completely oxidized to MgO which is known to block the adsorption of CO [17,18]. Mg does not seem to be a good promoter in the reaction between CO and NO on cobalt in a sense that the metal-oxygen bonds tend to increase due to Mg and this is not compensated by other effects. On the other hand, the Mg layer stabilizes CO against NO attack which again reduces the reaction probability.

5. Conclusions Nitric oxide adsorbed dissociatively on polycrystalline cobalt at room temperature but at large exposures molecular adsorption was also observed. Atomic nitrogen was more strongly bound to the surface as the NO exposure increased. Submonolayer coverages of magnesium on cobalt increased the uptake of NO. However, a full monolayer of Mg adsorbed less NO than the clean cobalt. Coadsorption of CO and NO at room temperature lead to the blocking of CO adsorption and to the partial removal of CO molecules as CO,. The presence of Mg on the surface during coadsorption stabilized the CO molecules against the attack by NO molecules at room temperature. Formation of CO, at elevated temperatures in the presence of NO and CO molecules was observed on both clean and Mg covered cobalt.

Acknowledgements Discussions with Mr. J. Vaari and Ms. A. Talo are greatly appreciated. Financial support from Technology Development Center of Finland (TEKES) is acknowledged. T.V. acknowledges financial support from the Alfred Kordelin foundation.

T. Vaara et al. /Applied

Surface Science 89 (1995) 103-l 11

References [l] G. Brodtn, T.N. Rhodin, C. Brucker, R. Benbow and Z. Hurych, Surf. Sci. 59 (1976) 593. [2] M.E. Bridge and R.M. Lambert, Surf. Sci. 94 (1980) 469. [3] Y. Sakisaka, M. Miyamura, J. Tamaki, M. Nishijima and M. Onchi, Surf. Sci. 93 (1980) 327. [4] G.L. Price and B.G. Baker, Surf. Sci. 91 (1980) 571. [5] H. Conrad, G. Ertl, J. Kippers and E.E. Latta, Surf. Sci. 50 (1975) 296. [6] M.J. Breitschafter, E. Umbach and D. Menzel, Surf. Sci. 109 (19811493. [7] J.G. Chen, W. Erley and H. Ibach, Surf. Sci. 224 (1989) 21.5. [8] L.J. Whitman and W. Ho, Surf. Sci. 204 (1988) L725. [9] R.G. Copperthwaite, A.F. Carley and M.W. Roberts, Surf. Sci. 165 (1986) Ll. [lo] A.V. Hamza, P.M. Ferm, F. Budde and G. Ertl, Surf. Sci. 199 (1988) 13. [ll] J.G. Chen, W. Erley and H. Ibach, Surf. Sci. 227 (1990) 79. 1121 C.H.F. Peden, D.W. Goodman, D.S. Blair, P.J. Berlowitz, G.B. Fisher and S.H. Oh, J. Phys. Chem. 92 (1988) 1563. [13] V. Schmatloch and N. Kruse, Surf. Sci. 269/270 (1992) 488. [14] J. Hall, I. ZoriC and B. Kasemo, Surf. Sci. 269/270 (1992) 460. [15] Y. Uchida, R. Imbihl and G. Lehmpfuhl, Surf. Sci. 275 (1992) 253. [16] Th. Fink, J.-P. Dath, M.R. Basset, R. Imbihl and G. Ertl, Surf. Sci. 245 (1991) 96. [17] J. Lahtinen, J. Vaari, A. Talo, A. Vehanen and P. Hautojlrvi, Surf. Sci. 245 (1991) 244. [18] J. Vaari, J. Lahtinen, A. Talo and P. Hautojami, Surf. Sci. 251/252 (1991) 1096.

[19] J. Vaari, J. Lahtinen

and P. Hautojlrvi,


Surf. Sci. 277 (1992)

253. [20] J. Vaari, .I. Lahtinen and P. Hautojarvi, Appl. Surf. Sci. 78 (1994) 255. [21] J. Vaari, T. Vaara, J. Lahtinen and P. Hautojarvi, Appl. Surf. Sci. 81 (1994) 289. [22] B.E. Nieuwenhuys, Surf. Sci. 126 (1983) 307. [23] P.A. Redhead, Vacuum 12 (1962) 202. [24] A.L. Cabrera, J. Chem. Phys. 93 (1990) 2854. [25] P.M.A. Sherwood, in: Practical Surface Analysis, 1st ed., Eds. D. Briggs and M.P. Seah (Wiley, New York, 19831. [26] Numerical Recipes: The Art of Scientific Computing (Cambridge University Press, Cambridge, 1986). [27] J. Lahtinen, A. Talo, J. Vaari, A. Vehanen and P. Hautojlrvi, Acta Polytech. Stand. Phys. 173 (1991). 1281 Y. Fukuda, Surf. Sci. 104 (1981) L234. [29] A. Sandell, A. Nilsson and N. Mittensson, Surf. Sci. 251/252 (1991) 971. [30] E.L. Hardegree and J.M. White, Surf. Sci. 175 (1986) 78. [31] A.F. Carley, S. Rassias, M.W. Roberts and W. Tang-Han, Surf. Sci. 84 (1979) L227. [32] E. Umbach, S. Kulkarni, P. Feulner and D. Menzel, Surf. Sci. 88 (1979) 65. [33] C. Egawa, S. Naito and K. Tamaru, Surf. Sci. 138 (1984) 279. [34] H. Over, T. Hertel, H. Bludau. S. Pflanz and G. Ertl, Phys. Rev. B 48 (1993) 5572. [35] P.W. Davies and R.M. Lambert, Surf. Sci. 110 (1981) 227. [36] T.W. Root, G.B. Fisher and L.D. Schmidt, J. Chem. Phys. 85 (1986) 4687.