The chemisorption of CO and NO on Rh(110)

Surface Science 97 (1980) 346-362 0 North-Holland Publishing Company

THECHEMISORPTIONOFCOANDNOON

Rh(ll0)

R.J. BAIRD, R.C. KU and P. WYNBLATT Ford Motor Co., Engineering and Research Staff, P.O. Box 2053, Dearborn, Michigan 48121, USA

Received 18 January 1980; accepted for publication 3 April 1980

The room temperature chemisorption of CO and NO on Rh(ll0) has been studied by means of photoelectron spectroscopy, LEED, and temperature programmed desorption. It was found that CO chemisorbs and desorbs molecularly at all coverages. In contrast to CO, it was found that NO chemisorbs dissociatively at low coverages followed by nondissociative chemisorption at higher exposures. On flash desorption, the molecular NO decomposes and desorbs as Nz. The N 1s XPS line from chemisorbed molecular NO is found to be anomalously broad and suggests the presence of multiplet splitting from unpaired spin density on the chemisorbed molecular NO.

1. Introduction The reactions of carbon monoxide and nitric oxide over group VIII transition metal heterogeneous catalysts are of technological importance and have found applications in areas ranging from hydrocarbon synthesis to automotive emission control [l-3]. Of particular interest in recent emission control applications, are those metals such as Ru and Rh which show a high specificity for the reduction of NO to N2 rather than to NH3 [3,4]. While there already exists a considerable literature on the chemisorption of CO and NO on transition metal surfaces [5,6], it is only recently that detailed chemisorption studies on Rh have been reported [6-181. A number of workers have used the techniques of Auger electron spectroscopy (AES), low energy electron diffraction (LEED), and temperature programmed desorption (TPD) to investigate the chemisorption of small molecules onto well defined Rh single crystal surfaces [7-13,181. Also, Campbell and White [6,14] and Sexton and Sdmorjai [15] have used AES and TPD in extensive studies of the chemisorption and chemical reactions of CO and NO on polycrystalline Rh samples. To date though, few photoemission studies have been reported for chemisorption on single crystal Rh surfaces [17,18]. Therefore it was of interest to couple photoelectron spectroscopy measurements with the more commonly used surface analytical techniques such as AES, LEED, and TPD in an effort to gain a more detailed description of the molecular processes involved in CO and NO chemisorption on a Rh surface. 346

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

341

The general utility of photoemission measurements in studies of chemisorption and catalysis has been demonstrated in a number of prior studies [19-211. In some of these studies, it is apparent that the combined application of both X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) provides a particularly powerful tool for the study of surface-adsorbate interacttions. With the combined technique approach, UPS measurements can identify the molecular bonding in adsorbate overlayers, while XPS provides unambiguous elemental identification and some chemical state information through XPS line positions. In addition to qualitative results, XPS can also provide quantitative measurements with minimum perturbation of the adsorbate layers.

2. Experimental The experimental measurements were carried out in two separate ultra-high vacuum (UHV) systems on different single crystal specimens. Both specimens were spark cut from the same Rh crystal (obtained from Materials Research Corp., Orangeburg, NY) which had been oriented to +l” of the (110) direction. The oriented single crystal slices, which were -1 mm thick, were annealed in flowing hydrogen at 1000°C for 90 h to reduce the concentration of non-metallic impurities. The samples were then mechanically polished and chemically etched to produce a smooth single crystal surface. One UHV system which was used primarily for detailed TPD, AES, and LEED measurements is an ion pumped system fitted with a cylindrical mirror analyzer (CMA), hemispherical LEED optics, a quadrupole mass spectrometer and an ion gun. The sample in this system was heated by electron bombardment of a Ta backing plate which supported the crystal, The base pressure in this system was 2 X lo-” Torr. Full details of this system have been published previously [22]. The second UHV system which was used primarily for photoemission measurements was a commercial electron spectrometer (Physical Electronics 548) consisting of a high resolution double-pass CMA with a co-axial electron gun, a Mg-anode X-ray source, a differentially pumped noble gas discharge lamp and an ion gun. The system was also fitted with LEED optics and a quadrupole mass spectrometer (UTI1OOC) which allowed cross calibration of gas exposures in the two systems via the LEED and TPD results. The only significant modifications to the commercial system were the addition of a separately pumped gas admission manifold, and the construction of a sample holder which allowed the sample to be heated to over 1200°C by resistively heating a 0.001 inch Ta backing foil which supported the crystal. The sample temperature was measured by a Pt/Pt-lOwt%Rh thermocouple spot-welded to a recess in the crystal. Base pressure in this system was 7 X 10-r’ Torr. Photoelectron spectra were acquired digitally in a Nicolet 1072 multichannel analyzer (MCA) which was part of the PHI electron spectrometer system. The digital data were transferred from the MCA to a DEC-10 timeshare system for sub-

348

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

sequent analysis. The UPS data were smoothed using the Savitsky-Golay smoothing algorithm [23] and difference curves were obtained between spectra from clean and adsorbate covered surfaces. The difference curves were scaled to give a flat region in the vacinity of the Rh4d band in order to emphasize the spectral features due to the adsorbate. The XPS data were analyzed by least-squares fitting gaussian peaks to the experimental spectra. The fitting routine fit a sloping linear background as well as gaussian peaks which included an “inelastic tail” and X-ray satellite structure where appropriate. The details of the least squares fitting program have been described previously [24]. The XPS spectra were taken with an instrumental resolution of 0.8 eV and the UPS spectra with a resolution of 0.4 eV. All binding energies are referenced to the rhodium Fermi level (Er).

3. Results 3.1. Clean surface After insertion into the vacuum system, the samples were cleaned by alternate cycles of inert gas sputtering, heating to 4OO’C in 5 X 10m8 Torr of oxygen, and annealing at 1000°C in a vacuum of lo-r o Torr. A clean surface exhibited no detectable carbon or oxygen in the AES as shown in fig. 1. Also, no evidence for strong boron segregation to the surface was noted in these samples contrary to the experience of other workers with Rh [9 ,l 1 ,I 31. This may be due to the removal of boron as a volatile borane during hydrogen heat treatment. After annealing

at 8OO’C following

,

0

Oa exposure at 4OO”C, the clean Rh(ll0)

1 I L 1 * 11 100

200

300

ELECTRON Fig. 1. Representative

AES spectrum

18

400

11

500

11

600

L

700

ENERGY(eV) of clean Rh(l10)

surface.

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(Il0)

b*

f. Fig. 2. Photographs of LEED patterns observed from clean and adsorbate covered Rh(l10): (a) clean surface, 64 eV, 1 X 1; (b) clean surface, 116 eV, 1 X 1; (c) 2 L CO, 64 eV, (2 X 1)plgl note missing (0, l/2) spot; (d) 2 L CO, 116 eV, (2 X 1)plgl note missing (0, l/2) and (0,3/2) spots;(e) 0.5 L NO, 54 eV, 2 X 2; (f) 10 L 02, 54 eV, (2 X I)plgl.

R.J. Baird et al. / Chemisorption

350

1

of CO and NO on Rh(ll0)

I

,

Rh(llO)

XPS

I

Mg Ka 3d

z z

3

MVV ’

Auger

z Q

3P

? 5 z_

1000

800

600

BINDING Fig. 3. Representative

400

ENERGY

200

0

(eV)

XPS spectrum of clean Rh(l10).

exhibited a sharp 1 X 1 LEED pattern as shown in figs. 2a and 2b. This is consistent with the results reported by Frost and co-workers [16] and by Marbrow and Lambert [9]. However, heating the sample to temperatures greater than 1OOO’C and quickly quenching to room temperature gave a somewhat irreproduceable 5 X 1 LEED pattern on the (110) surface. This 5 X 1 pattern appeared to be associated with a small amount of carbon on the surface as indicated by AES. This 5 X 1 pattern would persist until the sample was again heated on Oz. In fig. 3 a representative XPS survey scan of the clean Rh(ll0) surface is shown indicating the major core lines and the lack of detectable carbon or oxygen. surface

3.2. CO chemisorption

Carbon monoxide exhibited a relatively simple chemisorption behavior on the Rh(ll0) surface. With the crystal at 70°C or with the LEED electron gun on at lower crystal temperatures, the LEED pattern remained 1 X 1 during CO uptake with a slight increase in the intensity of the diffuse background and a change in the relative intensity of the various beams with increasing CO exposure. However, exposure to CO with the electron beam off and the crystal at 30°C gave the (2 X 1) plgl LEED pattern reported by Marbrow and Lambert and shown in figs. 2c and 2d. On heating the crystal to desorb CO from the surface, the pattern reverted to the original sharp 1 X 1 pattern. In figs. 4a and 4b the C 1s and 0 1s regions of the XT’S spectrum are shown for clean and CO covered Rh(l10) surfaces along with the difference between the spectra of the clean and adsorbate covered surfaces. In both spectral regions the

R.J. Baird et al. / Chemisorption of CO and NO on Rh(ll0)

290

280

BINDING

530

520

IO

5

351

0

ENERGY(eV)

Fig. 4. Photoelectron spectra of clean and CO covered Rh(l10) and difference curves: (a) carbon 1s region showing interference of 0~56 and “7,~ satellites of Rh 3d lines; (b) oxygen 1s region showing interference of Rh 3p1/2 he; (c) He II excited valence band spectra showing prominent CO derived lines at 7.6 and 10.6 eV below Ef.

relatively weak adsorbate lines are superimposed on features associated with strong substrate lines. In the 0 1s region, the inelastic tail of the Rh 3~,,~ peak at 521 .O eV contributes a large steeply sloping background under the 0 1s peak. Also, there is a broad “hump” in the background which may be either a loss feature associated with the Rh 3p,,, peak or a contribution from oxygen dissolved in the Rh lattice. In the C 1s region, the clean surface spectrum shows a number of features which have the proper spacing and relative intensity to be assigned to the ffs,6 and 07,s satellites of the very intense Rh 3d 3,2,s,2 doublet at 307.2 and 311.9 eV 125,261. Difference spectra remove the interference of the substrate features as shown in the uppermost curves in figs. 4a and 4b. The 0 1s region shows a single peak at 531.3 eV with a full width at half maximum (FWHM) of 2.4 eV while the C 1s region exhibits 2 peaks at 285.4 and 283.4 eV each having a FWHM of 2.2 eV. The higher binding energy carbon peak at 285.4 eV is consistent with binding energies for molecular CO observed on other transition metals [27,28]. The lower binding energy peak may be due either to a residue of the background subtraction or to carbon present in a second chemical state. This. second chemical state c.ould be CO in a different adsorption site or carbon from CO dissociation as suggested by Grant

352

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

and Hass [lo]. The C 1s (285.4 eV) to 0 1s area ratio, when corrected for cross section [29] and analyzer transmission (assumed to be proportional to (l/Ekin)“’ 30), is found to b e 0.9 * 0.2 while the total corrected C Is/O 1s area ratio is found to be 1 .O +_0.2. Fig. 4c shows representative He II excited photoelectron spectra for a clean Rh(ll0) surface, for the same surface exposed to 10 L of CO, and the adjusted difference between the two spectra. In the spectrum for the CO covered surface two strong peaks 7.6 and 10.6 eV below the Fermi level are noted. These peaks are present at all CO coverages and increase in intensity with increasing CO exposure. Comparison of the 7.6 and 10.6 eV peaks to gas phase spectra of CO [31] and to spectra of CO adsorbed on other transition metal surfaces [5,32,33] shows a good correlation between these peaks and the 4a and 17r-5a molecular orbitals of CO respectively. The UPS and XPS results which suggest molecular CO chemisorption on Rh(ll0) at room temperature are supported by the temperature programmed desorption results. Fig. 5 shows the TPD curves obtained by measuring the 28 amu partial pressure during flash heating of the Rh crystal after various CO exposures. In these experiments the initial CO exposure was carried out at 7O’C. The heating rate was lO’C/s and the pumping speed was 86 l/s for CO. The major CO desorption peak occurs at 240°C with a slight shift to lower temperature with increasing coverage. At high coverages there is a suggestion of two minor peaks at 145 and 185°C.

I

TPD

’

(28amu)

I

I

CO on Rh(llO)

TEMPERATURE (93 Fig. 5. 28 amu TPD curves for various CO exposures.

R.J. Baird et al. / Chemisorption

ofCO and NO on Rh(ll0)

353

If the main peak at 24O’C is taken to result from a first order desorption process, then the enthalpy of desorption, AHd can be calculated from eq. (1) [34] : AHd/RTff = v(dT/dt)-’

exp(-AHdRT,)

,

(1)

where TP = peak temperature (K), v = frequency factor, R = gas constant, and dT/dt = heating rate (K s-l). If the frequency factor v is assumed to be 1013 Hz, then AHd for CO on Rh (110) is found to be 29.7 kcal/mole in good agreement with the results on polycrystalline Rh [14,1.5] and other Rh crystal faces [9,13]. 3.3. NO chemisorption The chemisorption of NO on the Rh(ll0) surface is considerably more complex than the chemisorption of CO on the same surface. At 70°C the only change in the LEED pattern with NO exposure is an increase in the diffuse background which eventually obscures the substrate pattern. However, exposure to NO with the sample at 0°C initially gives a streaked 2 X2 pattern which evolves into the true 2 X 1 pattern shown in fig. 2e. This 2 X 1 pattern reaches maximum intensity at -0.5 L, and continued exposure leads only to an increase in diffuse background. Heating the sample to 300°C after saturation NO exposure results in the formation of a (2 X 1) plgl pattern which then reverts to a sharp 1 X 1 pattern on heating above 800°C. The photoelectron spectra for various room temperature NO exposures are shown in fig. 6. After 0.5 L of exposure, one principal peak at 397.4 eV is observed in the N 1s XPS spectrum and a broad peak centered at 530.5 eV appears in the 0 1s region. At this coverage, the He11 UPS spectrum shows an additional feature centered 6 eV below Ef and some increase in emission at Ef. However, with increasing NO exposure beyond 0.5 L a second broader peak at higher binding energy emerges in the N 1s region, and the higher binding energy side of the 0 Is peak increases in intensity. In the He II UPS spectrum, new features appear at 14.5,9.2, and 2.5 eV below Ef. These new features in the XPS and UPS spectra increase in intensity with increasing NO exposure up to saturation at between 5 and 10 L of exposure. The TPD results for NO chemisorption also exhibit considerable complexity as shown in figs. 7 and 8. In fig. 7 the 28 amu (N&O), 30 amu (NO) and 32 amu (0,) partial pressures are plotted as a function of sample temperature during sample heating. It can be seen that even after saturation NO exposures very little molecular NO desorbs from the surface. The primary desorption products are Nz at 195’C and 2.5O”C and O2 at -8OO’C. Fig. 8 shows the 28 amu desorption peak as a function of NO exposure. These results show a unique phenomenon in that a high temperature desorption peak disappears with increasing gas exposure. Below 0.5 L of exposure no molecular NO is detected leaving the surface and the Nz desorption peak occurs at 340°C. With increasing exposure beyond 0.5 L, the N, desorption is

354

R.J. Baird et al. / Chemisorption I-~-~,

Nls

region MgK. NO on Rh(llO)

of CO and NO on Rh(ll0)

,~“‘I~~~

Ok

region MgK. NO on Rh(llO)

He II diff Y&F---NO on Rh(llO)

BINDING ENERGY(eV) Fig. 6. Photoelectron spectra of NO covered Rh(l10) with increasing exposure: (a) N 1s region showing presence of two distinct chemical states of N; (b) 0 1s region also indicating two states of 0; (c) difference curves of He II excited VB spectra. Note absence of NO derived features at 0.5 L exposure.

seen to shift to two peaks at lower temperature until no remnant of the original 340°C peak remains at saturation NO coverage. In contrast to the results reported on stepped crystal surfaces [ 121, no NzO was detected during flash desorption. Fig. 9 shows the results of a photoemission experiment coupled with the flash desorption. The sample was exposed to 10 L of NO and the XPS and UPS spectra were recorded. These spectra, shown in the upper set of curves in fig. 9, are identical to the 40 L exposure curves in fig. 6 and indicate that the surface is essentially saturated. The sample was then heated to 225°C and then cooled to room temperature. The middle spectra show that the high binding energy N 1s and 0 1s lines have disappeared and that the spectra are similar to the low NO coverage spectra. Finally, heating the sample to 9OO’C gives spectra that are again identical to the clean surface. 3.4. Oxygen chemisorption

Several O2 chemisorption experiments were performed in an attempt to clarify the origin of the 6eV peak in the He11 UPS spectra of chemisorbed NO, and the

R.J. Baird et al. / Chemisorption

‘,‘I’

TPD

fPd

NO on’Rh(llO)

355

of CO and NO on Rh(ll0)

(2&m:)

I;0

“b I c de f -

J

1

1

100

I

200

TEMPERATURE

I 300

L

I”0

(“C)

Fig. ‘7. TPD results for saturation desorption.

NO coverage of Rh(ll0)

200

TEMPERATURE

RhillO)

0:

0”:;:: 0.8L

I.IL

2.OL 3.3L

400

300

CC)

showing very little molecular NO

Fig. 8. 28 amu TPD results for various NO exposures showing disappearance of 340°C peak and growth of 195 and 250°C peaks with increasing exposure.

(2 X 1)plgl

LEED pattern that appeared in heating an NO covered surface over 300°C. It was found that O2 exposure produced a (2 X 1)plgl LEED pattern at 10 L exposure as shown in fig. 2f and gave the XI’S and UF’S spectra shown in fig. 10. The 0 1s XPS spectrum shows a main peak at 529.0 eV and a smaller shoulder at 530.9 eV. In the UPS difference spectrum a peak 6 eV below the Fermi level is evident.

4. Discussion The behavior of CO and NO chemisorbed on the Rh(ll0) surface is consistent with the position of Rh in the periodic table and the known chemisorptive behavior of CO and NO on other transition metal surfaces. Broden et al. have summarized the chemisorptive properties of transition metal surfaces and have noted that the tendency for dissociative CO and NO chemisorption increases the farther that an element is above and to the left of Pt in the periodic table, i.e., the more electro-

356

R.J. Baird et al. / Chemisorption

m

I

131s

of CO and NO on Rh(I IO)

z 1 region

1

’

’

’

MgK.

IOL NO on Rh(ll0)

OL NO on Rh(llO)

,!,I,I,‘,,“l’,‘~

IB

He II diff

IL NO on Rh(llO)

5 I

:

--4&r-% 5

35

BINDING

ENERGYtell)

Fig. 9. Photoelectron

spectra taken before and after interrupted NO flash experiment. Uppermost curves are NO saturated Rh(llO), center curves same regions after heating to 225”C, bottom curves after heating to 900°C; (a) N Is region; (b) 0 1s region; (c) difference curves from He II excited valence band spectra.

I

’

’

’

’

01s region

1

’

’

’

Mg K.

BINDING

‘~I”‘~l”“l”“l’~

He II diff.

V.B.

ENERGY

(eV)

Fig. 10. Photoelectron spectra of O2 covered Rh( 110) after 10 L exposure: difference curves from He II excited valence band spectra.

(a) 0 1 s region;

(b)

R.J. Baird et al. / Chemisorption

of COand NO on Rh(ll0)

357

positive the metal [5]. This behavior can be rationalized in terms of increased back bonding from d-like states in the metal into 2n* anti-bonding orbitals of the rr-acceptor molecules thus destabilizing the C-O or N-O molecular bond and favoring dissociation. Further, Broden et al. have proposed that the energy separation between the 40 and lrr molecular orbitals (A(4o-ln)) of chemisorbed diatomic molecules is a measure of the amount of back donation. This follows since the back donation should tend to weaken the C-O or N-O bond thus tending to destabilize the In molecular level while having little effect on the non-bonding 4a level thus increasing A (40-lrr). 4. I. CO chemisorption Based on the above scheme, the chemisorption of CO onto the (110) surface of Rh is expected to be molecular since CO chemisorbs molecularly on both Ru and Pd /IS]. The observed energies of 40 (10.6 eV) and overlapping In-5u peaks (7.6 eV) in the UPS spectrum are in agreement with the results of Braun et al. [17] if the work function of the CO covered Rh(ll0) is taken into account (this work function is found to be 6.0 eV as determined from the low energy cut off in the He1 UPS spectrum). The separation between the 4u and overlapping Irr-SO peaks is found to be 3.0 eV which is just midway between the values of A(4u-ln) observed for CO on Ru (A = 3.15 eV) and Pd (A = 2.90 eV) [S]. The 1s binding energies of carbon and oxygen in the chemisorbed CO are found to be 285.3 and 531.1 eV respectively. In table 1 the carbon and oxygen Is binding energies for CO chemisorbed on group VIII transition metals are tabulated along with the value of A(4u-ln) from ref. [5]. It is of interest to note that smaller values of core level binding energies tend to be associated with larger values of A(4u--171). This observed correlation is consistent with the interpretation that a larger value of A(4u-ln) reflects a greater tendency toward back-donation from

Table 1 Cls and 01s binding energies and A(& - In) CO chemisorbed on group VIII transition metals; values of A(40 - In) taken from ref. [5] and this work, core level binding energies from references indicated

RU Ni Rb Pd Ir Pt

A(40 - In) (ev)

C 1sBE (eV)

0 1sBE (eV)

3.15 3.08 3.0 2.90 2.75 2.6

_ 285.6 [27] 285.3 285.8 [46] 286.6 [27]

531.7 531.4 531.1 532.1 532.7

[45] [27]

[20] [27]

358

R.J. Baird et al. / Chemisorption

of CO and NO on RhfllO)

the metal, since increased back-donation should permit increased screening of a core hole on the adsorbate thus lowering the observed binding energies of the adsorbate lines [35,36]. The origin of the small peak centered at 283.4 eV in the Cls difference spectrum is not known. The uncertainties in the corrected C Is/O 1s area ratios make it impossible to assign this peak unambiguously. However, it can be seen from the 0 Is and valence band spectra in fig. 4 that there is no evidence for atomic oxygen on the surface thus tending to eliminate the possibility that this peak is due to atomic carbon from dissociated CO. The possibilities remain that this peak is either an artifact of the background subtraction or CO adsorbed in a different adsorption site. In either case this peak accounts for only 10% of the total C 1s area. The TPD results for CO chemisorbed on Rh(l10) compare well with other reported results. The observed enthalpy of desorption of 29.7 kcal/mole (127 kJ/mole) falls just below the range of 31 to 32 kcal/mole reported for CO desorption from a variety of different Rh surfaces. However, the low temperature shoulder on the CO desorption peak from the saturated Rh(ll0) is less pronounced than the low temperature features seen in the results of Thiel et al., for CO desorption from a Rh( 111) surface [ 131. This may indicate that CO is adsorbed predominately in one type of site on the Rh(l10) surface as opposed to the multiple site adsorption proposed for CO on Rh(lll). Finally the near 1 : 1 ratio of carbon to oxygen found from the XPS measurements, couples with the UPS results which show the 10.6 and 7.6 eV peaks at all coverages and no evidence of a 6 eV peak, strongly support the conclusion that CO is adsorbed molecularly onto Rh(ll0) at room temperature. 4.2. NO chemisorption The more complex chemisorption behavior of NO/Rh(l 10) can be explained in terms of a mechanism which involves initial dissociation of NO on clean Rh followed by non-dissociative chemisorption of molecular NO onto the nitrogen and oxygen covered surface. This explanation is strongly supported by the XPS and UPS results. At NO exposures less than 0.5 L the N 1s spectrum exhibits one peak at a low binding energy which is similar to the binding energies observed in metal nitrides [37], and to the binding energies assigned to atomic nitrogen on the nickel [19,38] and tungsten [39] surfaces. Furthermore, at these low coverages the He11 UPS spectrum shows only a broad peak 6 eV below Ef and an increase in emission just below Ef. The 6 eV peak is observed in the UPS spectrum of the oxygen covered Rh surface (fig. 10) and a similar feature has been observed for dissociated NO on Ru [32] and W [39]. Also, the 2 X 1 LEFD pattern reaches maximum intensity and sharpness at -0.5 L of exposure. With increasing NO exposure beyond 0.5 L new features appear in the XPS and UPS spectra, while in the LEED experiment the diffuse background begins to increase. The new N 1s peak is centered at a binding energy of 400.0 eV and it is

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

359

considerably broader than the lower binding energy peak. The 0 Is spectra also show an increase in the intensity of the higher binding energy peak at 531 eV. The areas of these new XPS peaks increases with increasing NO exposure and reach maximum intensity by 10 L of exposure. In the UPS spectrum new peaks appear at 14.5,9.2, and 2.5 eV below Ef and these features can be assigned to the 40, In-5u, and 27r* molecular orbitals of non-dissociatively chemisorbed NO. These assignments are consistent with binding energies for molecular NO adsorbed onto other transition metals [19,32,38,39]. The unique TPD behavior of NO on Rh is also consistent with this interpretation of the electron spectroscopy results. At exposures of less than 0.5 L when only atomic nitrogen and oxygen are present on the surface, the formation of di-nitrogen is hindered and Nz is found to desorb at 34O’C. At exposures greater than 0.5 L both molecular NO and atomic oxygen and nitrogen are present on the surface. Under these conditions it appears that the molecular NO decomposes at 195°C giving rise to the first peak in the high exposure Nz desorption curves. This interpretation is supported by the results of the interrupted flash experiment in which the XPS and UPS (fig. 9) show no molecular nitric oxide on the surface after heating to 225’C. The second peak at 25O’f.Zin the high exposure Nz desorption curves is then the desorption of the remaining atomic nitrogen as Nz. Since no O2 or known oxygen containing species other than NO are detected leaving the surface below 800°C, it is plausible to suggest that oxygen left on the surface by the decomposition of molecular NO promotes in some way the desorption of the atomic nitrogen as Nz. This view is supported by the results of a co-adsorption experiment in which the clean Rh(ll0) surface was exposed to 0.3 L of NO followed by 0.5 L of Oz. Under these conditions di-nitrogen was found to

k

TPD 28amu

TEMPERATURE

(‘C)

Fig. Il. 28 amu TPD from 02 -NO co-adsorption experiment: (a) Rh(ll0) 0.3 L NO;(b) Rh(ll0) surface exposed to 0.3L NO followed by 0.5L Oz.

surface exposed to

360

K.J. Baird et al. / Chemisorption

ofC0

and NO on Kh(ll0)

desorb at 210°C as shown in fig. 11. The only disconcerning note in this explanation is the fact that the 0 1s XPS spectra do not show an increase in the amount of oxygen at the surface in the partial flash experiment relative to the 0.5 L NO exposure. It is possible that the oxygen is present as some sub-surface species as is often postulated to exist in group VIII transition metals [13,20]. However there is no evidence for this and this point is under further investigation. The other open question raised by the XPS results is the source of broadening of the N 1s line from non-dissociatively chemisorbed NO. As is evident from fig. 6 the 400.0 eV N 1s line is asymmetric and it was found that by using least squares fitting procedures [24] this line could not be adequately fitted by a single gaussian peak. However, two gaussian peaks gave a good fit, and the two components then had the same width as the 397.4 eV N 1s peak. The separation of the two component peaks was found to be 1.4 eV and the area ratio of the lower to higher binding energy line varied from 2.5 to 3.2. Among the possible explanations for this double peak structure, three are most plausible: a chemical shift effect due to adsorption at inequivalent sites; a many electron effect in the photoemission process, i.e.. shakeup or shakedown; or a multiplet splitting effect due to unpaired spin density on the chemisorbed NO. Each of these will be considered in turn. Of the three explanations a chemical shift effect seems least likely. A shift of 1.4 eV would be a large chemical shift for different adsorption sites. For example, Norton and co-workers find a difference of only 1 .O eV between the Cls binding energies of linear and bridge bonded CO on Pt [40]. Also, the relative areas of the two peaks seem to be reasonably constant and independent of exposure. This would imply that if the two peak structure were a chemical state difference, then these two states of significantly different binding energy would have to populate simultaneously. The second possible source of the broadening, many electron effects, cannot be dismissed so easily. The higher binding energy component at 401.4 eV could be a shake-up satellite of the 400.0 eV peak brought about by promoting a 27r* electron to an empty state near the Fermi level. This would leave two holes localized on the adsorbate which has an unfilled 27r* level below Ef. Such a final state goes counter to current theories of relaxation about core holes localized on adsorbates [36,41, 421. According to these theories, it is possible that the 400.0 eV peak is the fully relaxed peak in which an electron from the metal has transferred into the partially empty 2n* orbital of the chemisorbed NO to screen the N 1s core hole. The peak at 401.4 eV would then correspond to the NO ionic state on the surface in which the core hole is not fully screened, The UPS results for NO on Rh(ll0) show that the ground state 2n* orbital is already 2.5 eV below Ef, and on ionization of the N Is level it is expected that the 27r* orbital should be pulled down even lower in energy. (For comparison, in gas phase spectra of 02, the molecule with equivalent cores to N 1s ionized NO, the 2n’ orbital of O2 is found to be 3 eV more tightly bound than the 2n* orbital of NO [31].) Therefore, since the splitting in the N 1s lines is only 1.4 eV it would seem that the electron transferred to the adsorbate in the screening

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

361

would have to originate near the bottom of the Rh d-band. The final possible explanation of the two peak structure, multiplet splitting, gives surprising good agreement between the adsorbate results and the gas phase results in which multiplet splitting is known to play a role [43,44]. In the gas phase the observed multiplet splitting and triplet to singlet area ratio of the N 1s line of NO are -1.5 eV and 3 : 1 [43,44] ; the 0 Is line of NO shows only some broadening and separate peaks are not resolved. This result is very similar to the adsorbate results in which the observed N Is splitting is 1.4 eV and the area ratios are -3 : 1. The 0 1s is already broad so that an additional small broadening would be undetctable. While this agreement with the gas phase data does not prove that unpaired spin density remains on the chemisorbed NO it does strongly support this argument. Further work involving examination of the photon induced Auger spectra of chemisorbed NO is underway in order to better understand the nature of the photoelectron final state and to resolve this question.

5. Conclusions

(1) CO is chemisorbed non-dissociatively on Rh(ll0) at all coverages and has a heat of desorption of -29.7 kcal/mole. (2) NO is chemisorbed dissociatively onto Rh(ll0) at exposures of less than 0.5 L giving a 2 X 1 LEED pattern. (3) NO is chemisorbed non-dissociatively on Rh(ll0) in a disordered arrangement at exposures greater than 0.5 L. (4) After NO exposures of less than 0.5 L, Nz desorbs from Rh(ll0) at 340°C. Following exposures of greater than 0.5 L, molecular NO on the surface decomposes at 195°C and the remaining atomic nitrogen desorbs at 25O’C. (5) The desorption of N2 from the atomic oxygen and nitrogen covered Rh(1 IO) surface can be promoted by the co-adsorption of additional oxygen. (6) The molecular NO present on the F&(1 10) surface appears to retain some unpaired spin density in the 27r* orbital as evidenced by multiplet splitting of the N 1s core line.

Acknowledgements

We would like to thank our colleagues Kress for critically reading this manuscript.

E.N. Sickafus,

J.T. Kummer

and J.W.

References [ I] R.B. Anderson, in: Catalysis, Vol. 4, Ed. P.H. Emmett (Reinhold, [2] J.C. S&latter and K.C. Taylor, J. Catalysis 49 (1977) 42.

New York,

1956).

362

R.J. Baird et al. / Chemisorption

of CO and NO on Rh(ll0)

[3] M. Shelef and H.S. Gandhi, Ind. Eng. Chem. Prod. Res. Develop. 11 (1972) 393.

[4] K.C. Taylor and R.L. Klimisch, J. Catalysis 30 (1973) 478. [S] G. Broden, T.N. Rhodin, C. Brucker, R. Benbow and Z. Hurych, Surface Sci. 59 (1976) 593, and references therein. [6] T. Campbell and J.M. White, Appl. Surface Sci. 1 (1978) 347, and references therein. [7] C.W. Tucker, J. Appl. Phys. 37 (1966) 3013. [8] C.W. Tucker, J. Appl. Phys. 37 (1966) 4147; corrections C.W. Tucker, J. Appl. Phys. 38 (1967) 2696. [9] R.A. Marbrow and R.M. Lambert, Surface Sci. 67 (1971) 489. [lo] J.T. Grant and T.W. Hass, Surface Sci. 21 (1970) 76. [ll] D.G. Castner, B.A. Sexton and G.A. Somorjai, Surface Sci. 71 (1978) 519. [12] D.G. Castner and G.A. Somorjai, Surface Sci. 83 (1979) 60. [13] P.A. Thiel, E.D. Williams, J.T. Yates and W.H. Weinberg, Surface Sci. 84 (1979) 54. [14] C.T. Campbell and J.M. White, J. Catalysis 54 (1978) 289. [15] B.A. Sexton and G.A. Somorjai, J. Catalysis 46 (1977) 167. [16] D.C. Frost, S. Hengrasmee, K.A.R. Mitchell, F.R. Shepard and P.P. Watson, Surface Sci. 76 (1978) L585. [17] W. Braun, M. Neumann, M. Iwan and E.E. Koch, Solid State Commun. 27 (1978) 155. [18] R.J. Baird, R.C. Ku and P. Wynblatt, J. Vacuum Sci. Technol. 16 (1979) 435. [19] C.R. Brundle, 3. Vacuum Sci. Technol. 13 (1976) 301. [20] P.A. Zhdan, G.K. Boreskov, AI. Boronin, A.P. Schepelin, W.F. Egelhoff and W.H. Weinberg, Appl. Surface Sci. 1 (1978) 25. [21] P.R. Norton, J. Catalysis 36 (1975) 211. [22] H.P. Bonzel, Surface Sci. 27 (1971) 387. (231 A. Savitzky and M. Golay, Anal. Chem. 36 (1964) 1627; and corrections J. Steiner, Y. Termonia and J. Deltour, Anal. Chem. 44 (1972) 1906; H.H. Madden, Anal. Chem. 50 (1978) 1383. [24] C.S. Fadley, Ph.D. Thesis, University of California at Berkeley (1970) (LBL Report UCRL-19535 (1970)). [25] P.M.Th.M. Van Attekum and J.M. Trooster, J. Electron Spectrosc. 11 (1977) 363. [26] M.O. Krause and J.G. Ferreira, J. Phys. B8 (1975) 2007. [27] P.R. Norton and R.L. Tapping, Chem. Phys. Letters 38 (1976) 207. [28] J.C. Fuggle, D. Menzel, Surface Sci. 53 (1975) 21. [29] J.H. Scofield, J. Electron Spectrosc. 8 (1976) 129. [30] P.W. Palmberg, J. Electron Spectrosc. 5 (1974) 691. [31] D.W. Turner, C. Baker, A.D. Baker and C.R. Brundle, Molecular Photoelectron Spectroscopy (Wiley, New York, 1970). [32] H.P. Bonzel and T.E. Fischer, Surface Sci. 51 (1975) 213. [33] D.E. Eastman and J.E. Demuth, Japan. J. Appl. Phys. Suppl. 2, Pt 2 (1975) 827. [34] P.A. Redhead, Vacuum 12 (1962) 203. [35] K. Hermann and P.S. Bagus, Phys. Rev. B16 (1977) 4195. [36] E.W. Plummer, W. Salanec, J.S. Miller, Phys. Rev. B18 (1978) 1673. [37] R.J. Colton and J.W. Rabalais, Inorg. Chem. 15 (1976) 237. [38] A.F. Carley, S. Rassais, M.W. Roberts and W. Tang-Han, Surface Sci. 84 (1979) L227. [39] R.I. Masel, E. Umbach, J.C. Fuggle and D. Menzel, Surface Sci. 79 (1979) 26. [40] P.R. Norton, J.W. Goodale and E.B. Selkirk, Surface Sci. 83 (1979) 189. [41] N.D. Lang and A.R. Williams, Phys. Rev. B16 (1977) 2408; N.D. Lang and A.R. Williams, Phys. Rev. B18 (1978) 616. [42] K. Schonhammer and 0. Gunnerson, Solid State Commun. 23 (1977) 691. [43] J. Hedman, P.-F. Heden, C. Nordling and K. Siegbahn, Phys. Letters 29A (1969) 178. [44] D.W. Davies, R.L. Martin, M.S. Bana and D.A. Shirley, J. Chem. Phys. 59 (1973) 4235. [45] J.C. Fuggle, T.E. Madey, M. Steinkilberg and D. Menzel, Surface Sci. 52 (1975) 521. 1461 R.J. Baird, unpublished results.