9874
J . Phys. Chem. 1994, 98, 9874-9881
Particulate Cu on Ordered Al2O3: Reactions with Nitric Oxide and Carbon Monoxide Ming-Cheng Wu and D. Wayne Goodman* Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255 Received: May 16, 1994; In Final Form: July 18, I994@
The growth of small Cu particles, prepared by vacuum vapor deposition, on A1203 films on Mo(100) has been studied in the 80-800 K substrate temperature range with Auger electron spectroscopy (AES) and temperature-programmed desorption (TPD). A model based on Auger measurements has been utilized to examine the number density and the average size of the Cu particles on AlzO3. Nitric oxide decomposition on the Cu particles has been further investigated using TPD and high-resolution electron energy loss spectroscopy (HREELS). TPD and HREELS data clearly show that a fraction of the NO molecules react with the Cu particles to produce gaseous NZand N20 in the 90-250 K temperature range. The formation of NZat 770 K is also evident and is due to the recombination of adsorbed N atoms from NO dissociation. The effects of retained oxygen, adsorption temperature, and particle size on the production of high-temperature NZhave been examined. The reaction between nitric oxide and carbon monoxide has also been studied. The results show that, besides the NZand N20 products arising from NO decomposition, a small quantity of COZ is produced by the reaction between adsorbed CO and adsorbed oxygen.
Introduction Small supported metal particles have become an area of intense research interest in recent years due primarily to their importance in heterogenesis catalysis. Very small metal particles and clusters containing from a few to thousands of atoms often exhibit physical and chemical properties different from their bulk counterparts. Several previous have demonstrated that small supported metal particles can be reproducibly prepared in a controlled way by vapor deposition of the parent metal onto planar supports in an ultrahigh-vacuum environment. The catalytic properties of these particles can then be addressed systematically with respect to particle size, morphology, and the influence of supports. Recently, we have undertaken an initiative to study small metal particles supported on planar oxide substrates using surface-sensitive techniques. This article reports a study of the CO/NO reaction carried out on Al~O3-supportedCu particles. This work focuses on the effects of Cu particle size on the COI NO reaction chemistry. Copper is an active component in many catalytic reactions, such as CO oxidation and NO, reduction. There is a recent, growing interest in the catalytic decomposition of nitric oxide over Cu-containing zeolite due to environmental c~ncems.~-* Because many oxide supports used in catalysis are bulk insulators or wide gap semiconductors, problems with surface charging are encountered with these materials during charged particle measurements. Various approaches, such as the use of a neutralization electron g ~ n , ~have . ' ~ circumvented these difficulties. In particular, the successful synthesis of epitaxial ultrathin oxide films has offered new opportunities and avenues to study the surface chemistry of these materials using surface science techniques. - l 4 Our strategy to study supported metal particles involves the following procedure: (1) preparation of ultrathin oxide films on a single-crystal refractory metal surface [the use of a refractory metal with an appropriate lattice conformation provides a thermally stable substrate for heteroepitaxy of singlecrystal oxide films], (2) deposition of oxide precursor metals @
Abstract published in Advance ACS Abstracts, September 1, 1994.
0022-365419412098-9874$04.50/0
onto the refractory metal surface in a controlled oxygen environment, and (3) vapor deposition of metals of catalytic interest onto the films. This approach allows us to use an assortment of surface-sensitive electron spectroscopies without difficulties associated with surface charging and to prepare metal particles in a controlled, reproducible way. In this study, we exploit our early success in preparing highly ordered oxide films to study CO/NO surface chemistry on particulate Cu deposits synthesized on planar, epitaxial A1203films.
Experimental Section The experiments were carried out utilizing an ultrahighvacuum (W) system, described e l s e w h e ~ e , ~with ~ - ' ~capabilities for high-resolution electron energy loss spectroscopy (HREELS), Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), and temperature-programmed desorption (TPD) and facilities for sample heating and cooling. The refractory metal Mo( 110) was chosen as a substrate for preparing the thin A1203 films. The crystal cleaning and handling can be found elsewhere.l1J2 Sample temperatures were monitored with a pair of W-5% Re/W-26% Re thermocouple wires spot-welded to the edge of the rear surface. Thin A1203films were prepared by exposing Mo( 110) to A1 vapor in a controlled oxygen atmosphere. Aluminum deposition was performed via thermal evaporation of high-purity wires inserted into a small cylindrical alumina tube.15 The cylinder was then tightly fitted into a spiral tungsten filament whose ends were spot-welded to two support electrodes. The assembly was surrounded by a cylindrical shroud, open at one end and capped at the other. The source was initially thoroughly outgassed, and the aluminum wires were melted prior to use. The stable flux, obtained by applying constant power to the filament, was monitored by carrying out a series of TPD spectra after each deposition onto the Mo( 110) substrate. This homemade minicrucible source15 was capable of delivering a stable and reproducible flux over a period of several months under UHV operation conditions. Copper was evaporated from a source containing a high-purity wire tightly wrapped around a tungsten filament. The source, thoroughly outgassed before use, was operated at constant power for each deposition.
0 1994 American Chemical Society
J. Phys. Chem., Vol. 98, No. 39, 1994 9875
Particulate Cu on Ordered A1203
t I
HREELS AI,02/Mo(l10)
A 875
0
' 00
200
300
400
500
600
KINETIC ENERGY (eV)
Figure 1. Auger spectra of A1203 films grown on Mo( 110): (a) dM203 = 4.4 8, and (b) dM203= 20.0 8,. The films were annealed to 1200 K in oxygen ambient to improve their crystalline quality. The spectra were acquired at Ep = 2.0 keV and at a sample current of 5 pA.
Spectroscopic grade I5NO (15N, 98%+) and CO (99.99%) were used without further purification in the adsorption studies. The reactants were introduced via a gas nozzle doser, which was directed to the crystal surface and at a distance of approximately 0.5 cm from the crystal face. The temperature-programmed description spectra were obtained at a linear heating rate of 5-10 IUSusing a quadrupole mass spectrometer. Auger measurements were carried out using an electron beam with an impact energy of 2 kV and at a sample current of typically 5 pA. The HREELS spectra were acquired in the scattering compartment of a two-tiered chamber.12-14 The HREELS data were collected in a specular scattering geometry with an incident angle of the electron beam of 60" about the surface normal.
500
0
500
1000
1500
2000
2500
WAVENUMBER (cm")
Figure 2. HREELS spectra of A1203 films grown on Mo(ll0): (a) ~AI~O =) 4.4 8, and (b) d ~ = 20.0 l ~8,. ~The ~films were annealed to 1200 K in oxygen ambient to improve their crystalline quality. The spectra were acquired at Ep = 4.0 eV and at the specularly reflected beam direction.
100-250 eV kinetic energy range of spectrum a arise from the Mo substrate. Structural studies of very thin ,41203 films (dA2035 8 A) using LEED showed a complex pattem in which most diffraction spots can be interpreted as arising from electron multiple scattering at the interface. With an increase in film thickness, the multiple diffraction effects were attenuated. A simple hexagonal structure was seen at dA120, 2 15 A although the diffraction spots were not as sharp as those from multiple diffraction, indicating some degree of disorder in the thick films. The lattice periodicity derived from the LEED photographs relates closely to the 0-0 distance in bulk Al2O3. The observed hexagonal pattern can therefore be interpreted as due to the presence of an ordered, close-packed oxygen anion layer associated with either the (001) orientation of a-Al2O3 or the Results and Discussion (111) face of y-AlzO3. A detailed structural analysis of the Preparation and Characterization of the A1203 Films. initial stage of epitaxy of A1203films on a similar refractory Thin A1203films were grown at room temperature on a freshly metal substrate, Ta( 1lo), has been given e1~ewhere.l~ cleaned Mo(ll0) surface. The growth of the films was The growth of A1203 films has been further examined using controlled at a rate of approximately one-half aluminum HREELS, as shown in Figure 2. The fundamental modes of equivalent monolayer per minute. The background pressure of the surface optical phonons (below 1000 cm-' in frequency) oxygen was 7 x Torr during film growth. The films were and their multiples and combinations are evident in the spectra. annealed to 1200 K in oxygen ambient to improve their The surface optical phonon losses of the thin film (spectrum a) crystalline quality. Shown in Figure 1 are Auger spectra are characterized by a two-mode pattern, whereas three modes acquired following film synthesis at two representative film of the phonon losses are typical for the thick film (spectrum thicknesses: (a) dAlzo, = 4.4 A and (b) dAlzo, = 20.0 A. The b). These results are in a complete agreement with previous thickness of the dA203= 4.48, film was calculated from the data in the l i t e r a t ~ r e . ~ OFrederick -~~ et al.24,25have suggested Auger intensity ratio of the A13+(LVV) and Mo(MNN) peaks that the appearance of the two phonon modes are characteristic using the following equation: of very thin A1203films. Growth of Particulate Cu on Al2O3. The growth of particulate Cu deposits on these well-defined A1203films has been studied in the 80-800 K substrate temperature range. The films utilized for supporting Cu particles were typically 20 8, thick and exhibited excellent thermal stability and chemical The value of &,+I&, was obtained from the standard Auger inertness toward adsorption. !t has been reported26 that very spectra of A1203 and Mo, respective1y.l6 The attenuation thin A1203 films (dNzo3< 8 A) grown on A1 substrates react lengths, &3+ and A M o , were taken from the literature to be 3.717 with Ni deposits at elevated temperatures, causing diffusion of and 6.7 A,18respectively. The thickness of the thick film ( d ~ ~ o ) Ni into the substrate through defects in the A1203film. = 20.0 A) was estimated by extrapolating the value of the thin It has long been that particulate metal deposits film based on evaporation time and constant aluminum flux. prepared at relative low substrate temperatures exhibit smaller It is seen in Figure 1 that there is no indication of the presence average particle size and higher particle density than those of a metallic A1° feature at 68 eV. The predominant spectral prepared at higher temperatures; however, the low-temperature features are the A13+(LVV) transitions at -54 eV and the preparations are unstable and undergo a major change in O(KLL) transition at -500 eV. The spectral features in the morphology and size upon annealing or chemisorption. To
Wu and Goodman
9876 J. Phys. Chem., Vol. 98, No. 39, 1994
1
TPD
5i + z ? m
a
5
z v)
z
Z J E C U V A L E N T VONOLAVER
W
+ z
4
. W
ar
E
830
700
900
1000
1200
1100
1300
TEMPERATURE (K)
a 80 900
950
1000
$050
1100
1150
1200
TEMPERATURE \ K ]
Figure 3. TPD spectra of Cu following Cu deposition on A1203 films 1.0) for at various substrate temperatures. The Cu coverages (&, each spectrum were adjusted to be approximately the same. A linear
heating rate of 10 Us was used. prepare thermally and chemically stable particles of catalytic interest, it is imperative to carry out the metal deposition at elevated substrate temperatures. Figure 3 shows a series of TPD spectra of Cu deposited onto an A1203 film at various substrate temperatures, T,. The Cu coverages for each spectrum in Figure 3 were adjusted to be approximately the same. Noteworthy features in Figure 3 are the shift of the peak maxima to higher temperatures and the narrowing of the TPD peak as Ts is increased. The full width at half-maximum (fwhm) of the TPD spectra, for example, decreases from -70 K at T, = 80 K to -50 K at Ts = 600 K. Since the peak position and the shape of the desorption spectra determine the heat of sublimation, it is conceivable that at low substrate temperatures the broader TPD peak contains more components and thus reflects a broader size distribution of the CJ particles. Our Auger measurements further showed that Cu particles prepared at T, = 600 K were thermally stable up to an annealing temperature of 850 K, whereas a decrease in Cu Auger intensity was observed upon annealing for those prepared at T, = 80 K. Shown in Figure 4 is a family of TPD spectra of Cu deposited at T, = 600 K as a function of the Cu coverage in equivalent monolayers, Bcu. Bcu = 1 is defined as one monolayer (ML) of Cu on Mo( 1lo), the coverage that can be derived from the integrated TPD area of the well-defined desorption feature of the Cu monolayer on Mo( 110). Since the first monolayer of Cu grown on Mo(ll0) assumes the lattice structure of the ~ubstrate,~' the surface density of one equivalent monolayer corresponds to 1.43 x 1015 cm-2. The TPD spectra of Figure 4 exhibit an interesting trend in that the leading edge of the peak shifts continuously toward higher temperatures as Bcu is increased. It is also noteworthy that the TPD peak width, comparable to the width of the TPD peaks of Cu multilayers, remains approximately unchanged as Bcu is varied. The insert in Figure 4 shows the heat of sublimation, derived from the leading edge analysisz8 of the thermal desorption spectra, as a function of Cu equivalent monolayer. The heat of sublimation of Cu particles is seen to decrease rapidly from its bulk value of 80 & 3 kcaYmol at BC,, = 1.2 to 49 kcaYmol at 8cu = 0.2. Similar behavior has been observed previously for the Cu/SiOz system.z9 The decrease in the heat of sublimation may be
Figure 4. A family of TPD spectra of Cu deposited at Z', = 600 K as a function of equivalent monolayers, OCu: (a) 0.16, (b) 0.33, (c) 0.50, (d) 0.67, (e) 0.98, (f) 1.25, (g) 1.55, and (h) 2.09. The insert shows the heat of sublimation, derived from the leading edge analysis of the spectra, as a function of Cu coverage in equivalent monolayers. A
linear heating rate of 10 Us was used. 0.20
,
1
5i t
Cu/AI,O,/Mo(IIO)
Ts = 600K
3
100 il
m 1
a
,
D
0
O o04 0 6
/
t k"/
0021
0 00 00
05
10
15
20
25
30
35
Cu EQUIVALENT MONOLAYER
Figure 5. Auger intensity ratio of the Cu(LMM) peak to the O(KLL)
peak versus the Cu coverage in equivalent monolayers. The solid line is a theoretical simulation based on the isotropic growth model, as discussed in the text. The number density of Cu particles, derived from a curve fitting procedure, is 1.4 x 10" cm-2. The average sizes of the Cu particles are inserted as vertical bars. attributed to a decrease in the number of neighboring Cu atoms as the Cu particles become smaller. The average size of the Cu particles can be estimated from Auger measurements. Figure 5 shows Auger intensity ratios of the Cu(LMM) peak to the O(KLL) peak versus Cu equivalent monolayers. Each data point was collected after a fresh deposition of Cu onto the clean A1203 surface at T, = 600 K. The solid curve through the Auger data in Figure 5 represents a theoretical simulation based on a simple model described as follows: (1) Nucleation takes place homogeneously. (2) After an initial nucleation period, the number of particles remains constant until the stage where coalescence sets in. (3) The particle shape remains constant during growth. Using the algorithm formulated in the Appendix, the density of Cu particles, N , can be computed from a curve-fitting procedure. The calculated number of N = 1.4 x 10" cm-' compares favorably with those obtained by direct measurements with transmission electron microscopy (TEM): for the gold mica system, N = (1.0-1.5) x 10" cm-2 at Ts = 623 K;30 for
J. Phys. Chem., Vol. 98, No. 39, 1994 9877
Particulate Cu on Ordered A1203
_
-ii"
_
_
~
n
1060
TPD
TPD
15NO/Cu/A1,03/Mo(l 10)
15
NO/Cu/AI,031Mo( 110)
d
m/e=46, "N,O
m/e=32, 0
100
200
300
400
500
TEMPERATURE
600
700
800
200
Figure 6. Thermal desorption signals for d e = 30,31, and 46 from ISNOadsorbed on the Cu particles at 80 K versus 15N0exposure time, t: (a) t = 20 s, (b) t = 60 s, (c) t = 120 s, and (d) t = 300 s. The exposure at t = 640 s corresponds to saturation. O h was 0.67, corresponding to a cluster size of -65 A. A linear heating rate of 5 Ws was used.
the silverlamorphous carbon system, N = 1.0 x 10" cm-2 at Ts= 742 K.31 The average sizes of the Cu particles, derived from a curve-fitting procedure, are inserted as vertical bars in Figure 5 . Reaction of Nitric Oxide with Particulate Cu. The reaction of nitric oxide with well-characterized particulate Cu deposits grown on highly ordered A1203 films has been studied using TPD and HREELS. To differentiate N2 molecules formed during reaction from background CO residual in the TPD, isotopically labeled nitric oxide, 15N0, was used in this study. Thermal desorption products from "NO adsorbed on Cu particles at 80 K have been followed by monitoring several masses (mle = 30, 31, 32, 46, and 47). 8cu was 0.67, corresponding to the cluster size of -65 A. Shown in Figure 6 are desorption signals for mle = 30, 31, and 46 versus 15N0 exposure. No desorption signals for m/e = 32 and 47 were observed in the 80-850 K temperature range, indicating the absence of gaseous 0 2 and l5NO2products, respectively. TPD spectra (m/e = 30) are characterized by the presence of two desorption states, designated as a j31 state and a j32 state, as indicated in Figure 6. The j32 state at -770 K can be easily explained as arising from the recombination of adsorbed "N from dissociation of 15N0. The interpretation of the lowtemperature state @I state) is not so straightforward. While the #I1 peak tracks the d e = 46 TPD peak as the 15N0exposure is varied, the difference in peak maxima by 16 K between the j31 state and the desorption peak of the parent molecules (mle = 31) is evident. The m/e = 46 TPD peaks arise from the desorption of I5N20. Since the mle = 30 fragment in the cracking pattern of I5N2O molecules, detected mass spectrometrically,32 is only 10% in intensity of that of mle = 46, a portion of the j31 peak must arise from 15N2production. It follows then that adsorbed 15N0desorbs through two major pathways: a fraction of adsorbed ''NO molecules desorb associativelyin the 90-250 K temperature range; the remaining molecules undergo dissociation to form adsorbed nitrogen and oxygen. The recombination of adsorbed 15N0 and adsorbed
600
400
(K)
800
1000
1200
TEMPERATURE ( K )
Figure 7. TPD spectra of Cu ( d e = 63.5) and of 0 2 (mle = 32), with desorption signals for d e = 30, 31, and 46. Ocu is 0.67, corresponding to a cluster size of -65 A. Linear heating rates of 5 and 10 Ws were used for desorption of mle = 30, 31,46 and d e = 32, 64,respectively.
15N to form 15N20 occurs at temperatures as low as 110 K. HREELS results described in the following sections further show that a "N2O species is present even at 80 K. The 15N2 production is also evident in the 110-250 K temperature range. Finally, two adsorbed 15N atoms recombine and desorb at temperatures exceeding 700 K. The chemistry of nitric oxide decomposition on A1203supported Cu particles is summarized as follows:
-
15~0,,,"NO
"NO(,)
-
l5'(a)
+ O(a)
+ 15N(a)- 15N20 - 15N, "N(,) + 15N(a)
"NO,,,
where the subscript (a) denotes adsorbed species. It becomes obvious from the stoichiometry of the above reaction mechanism that the "NO decomposition produces oxygen retained on the Cu particles. Figure 7 shows TPD spectra of Cu (m/e = 63.5) and of 0 2 (mle = 32), with desorption signals arising from the products of 15N0 decomposition. 0 2 is seen to desorb in the same temperature range as the desorption of Cu, indicating that the retained oxygen may dissolve into the Cu particles. The influence of the retained oxygen on the production of 15N2@2 state) will be addressed in detail in the following paragraph. Figure 8 shows the 8 2 desorption peak of 15N2versus the number of adsorption cycles. The experiments were carried out in a successive fashion withouj removal of the retained oxygen; "NO exposures were the same for the three identical experimental runs. The integrated I5N2TPD area in the first experimental run is seen to be only one-third of that of the second. The 15N2yieId then remained nearly constant in the successive experimental runs. This feature shows that the initially retained surface oxygen promotes the 15N2production significantly. The nature of the retained oxygen species and the state of Cu, namely, Cu+ or Cu2+,are not clear at the present
9878 J. Phys. Chem., Vol. 98, No. 39, 1994
Wu and Goodman
AVERAGE PARTICLE SIZE (A) I
-
LlZ?
1
TPD
1
15
? -
NO/C~/AI,O,/Mo(llO)
50
50
100
I
2a
4:
t
5
5c z m
0
a
'i 00
400
500
600
700
800
10
15
20
25
30
CU EQUIVALENT MONOLAYER
TEMPERATURE (K)
Figure 8. The
05
900
desorption peak of I5N2 versus the number of adso tion cycles. Bcu was 0.67, corresponding to a cluster size of -65T. The experiments were carried out successively without removal of the retained oxygen; the 15N0exposures were the same for the three identical experimental runs. A linear heating rate of 5 Ws was used. j32
Figure 10. The integrated 15N2TPD area
(B2 state) as a function of the Cu coverage in equivalent monolayers. The average sizes of the Cu particles also are indicated.
HREELS 15
500
0
503
1003
NO/Cu/AI2O,/Mo(l10j
1500
2000
2500
WAVENUMBER (cm-') 400
500
600
700
800
900
TEMPERATURE ( K j
Figure 9. The p2 desorption peak of I5N2 obtained following 15N0 exposure at three different adsorption temperatures: (a) T, = 80 K, (b) Ts = 150 K, and (c) Ts = 300 K. 8cuwas 0.67, corresponding to a cluster size of -65 A. A linear heating rate of 5 Ws was used.
time. Further study utilizing techniques that probe the electronic states of Cu and oxygen is necessary to clarify these issues. The influence of the substrate temperature, T,, during adsorption on 15N2 production @* state) has also been examined. Shown in Figure 9 are the p2 desorption peaks at three different substrate temperatures. The 15N2yield appears to maximize at T, = 150 K. A decrease in 15N2production at T, = 300 K can be attributed to a decrease in the sticking coefficient of 15N0 at this elevated temperature. Plotted in Figure 10 is the integrated 15N2TPD area ( p 2 state) as a function of the Cu coverage in equivalent monolayers, &,. The average particle sizes, derived from Auger measurements, are indicated on the upper axis. No 15N2 production was observed until 8cu > 0.13 (-35 8, particles). The 15N2yield then reaches a maximum at 8cu 1.3 (-80 8, particles). The adsorption of 15N0 at 80 K on well-characterized particulate Cu deposits has been further studied using HREELS. The use of HREELS to address the nature of adsorbed species on oxide surfaces, however, suffers a significant handicap in that the accompanying vibrational spectra are dominated by losses due to excitation of surface optical phonons. The intense multiple phonon losses extend over a wide vibrational frequency
Figure 11. HREELS spectra of (a) a clean A 1 2 0 3 film, dAI2O3 = 8 A, and (b) CulAl203, d ~ 1 ~= 0 ) 8 A, Bcu = 2.7, following a saturation exposure of 15N0at 80 K. The spectra were acquired using an electron beam with a primary energy of Ep = 42 eV and at the specularly reflected beam direction.
range of the HREELS spectra, as is illustrated in Figure 2. Recently, we have developed a new approach to the acquisition of HREELS data in order to circumvent the difficulties associated with these phonon losses.12 By utilizing a highenergy incident electron beam, this new approach enables the direct observation of relatively weak adsorbate features without serious interference from the intense multiple surface optimal phonon losses.12 The use of an electron beam with a high impact energy often produces a relatively low count rate in the elastic peak. Because of the low signal levels encountered in these HREELS measurements, an optimal Fourier filtering technique has been utilized to remove high-frequency random noise and spikes in the spectra.33 All HREELS spectra reported below have been smoothed using this Fourier filtering technique.33 Shown in Figure 11 are two HREELS spectra acquired using an electron beam with a primary energy ofoEp = 42 eV. Spectrum a of a clean A1203 film (&*03 = 8 A) exhibits two loss features at 615 and 885 cm-', assigned to the optical phonon modes of the A1203 film, and a nearly featureless base line at frequencies above 1000 cm-'. Adsorption of 15N0 on the Cu precovered A1203 surface (&, = 2.7) gives rise to a distinct feature at 1520 cm-' and a shoulder at -1200 cm-', as shown in spectrum b of Figure 11. To facilitate the assignments of
J. Phys. Chem., Vol. 98, No. 39, 1994 9879
Particulate Cu on Ordered A1203 these loss features, the following paragraph summarizes the previous vibrational data of NO adsorbed on low index planes of single-crystal Cu. There have been a few studies of NO adsorbed on low index planes of single-crystal These studies have shown that NO adsorption is quite complicated and that its decomposition chemistry is dependent on the surface orientation, the substrate temperature, and the e x p ~ s u r e . ~ HFEELS ~-~~ studies,37 for example, have shown that, on Cu(llO), NO initially adsorbs associatively at low temperatures ( T < 110 K) in a skewed adsorption geometry (with the nitrogen end down). The loss features observed at -840 and -1560 cm-l have been attributed to the NO bend, &NO), and the NO stretch, v(NO), respectively, of a bent NO.37 In contrast, decomposition of a portion of the initially adsorbed NO occurs on Cu(100) even at T = 85 K, leading to the formation of adsorbed oxygen and nitrogen.37 The latter then combines with arriving NO molecules to form N20 at low exposures. The corresponding loss peaks observed at -800 and -1260 cm-l are evident in the HREELS spectra and can be assigned to the NO bend and stretch, respectively, of a N20 species. The NN stretch of adsorbed NzO should give rise to a peak at -2224 cm-l, which was not detectable in the spectra.37 With an increase in NO exposure, first the loss features at -770 and -1500 cm-l were observed followed by the loss peak at 1770 cm-'. The 770 and 1500 cm-l features were attributed to the &NO) mode and the v(N0) mode, respectively, of skewed NO, and the 1770 cm-l feature was ascribed to the v(N0) mode of linear-bonded NO.37 On Cu( l l l ) , NO adsorption was associative at T = 85 K.36 Loss features at -815 and -1510 cm-l were attributed to the vibrational modes of NO adsorbed on a bridging site of Cu(11 l).36 With an increase in NO exposure, first bridging and then atop NO species (at -670 and -1775 cm-l) were found, and both were believed to be in a skewed adsorption geometry.36 Recent theoretical studies38 suggest that bridging NO with a bent bonding configuration on Cu( 111) should exhibit a red shift in frequency by 713 cm-' compared with the gas phase value of 1876 cm-l. The bridging NO species was therefore reassigned to a 3-fold bonded NOS3* Considering the above discussion, we assign the loss feature at 1520 cm-' to the NO stretch of 15N0species and the shoulder at -1200 cm-' to the NO stretch of a 15N20 species. The bending modes of 15N0 and 15N20 were not observed in the present study because of the interference from the substrate phonon losses. The NN stretch of 15Nz0also was not apparent due to the small perpendicular component of the dipole moment; Le., the NN bond of adsorbed 15N20 probably lies parallel to the surface.
Reaction of Nitric Oxide and Carbon Monoxide with Particulate Cu. The reaction of nitric oxide and carbon monoxide with particulate Cu deposits supported on A1203 films has been studied using TPD and HREELS. 15N0 and CO reactants were fully mixed in the gas manifold with a 15NO: CO ratio of 1:1 prior to admission into the UHV chamber. The adsorption of the gas mixture was then carried out via a gas doser at T, = 80 K. Thermal desorption products from 15N0 CO adsorbed on the Cu particles have been followed by monitoring several masses (mle = 28, 30, 31, 32, 44, and 46). Shown in Figure 12 are desorption signals for mle = 28, 30, 31, 44, and 46 following two different exposures. 8cuin Figure 12 was 2.7, corresponding to a cluster size of -105 A. No desorption signal for mle = 32 was observed in the 80-850 K temperature range, indicating the absence of gaseous 0 2 product. The desorption of the parent molecules of 15N0 and CO and the gaseous products 15N2and 15N20,due to 15N0 decomposi-
+
t
TPD 15NOtCO/Cu/Al,O,/Mo(i
tA
10)
m i e 4 1 , 15N0
t A
100
I
200
300
400
500
600
700
800
TEMPERATURE (K)
Figure 12. Thermal desorption signals for mle = 28, 30, 31,44, and 46 from ISNO CO adsorbed on Cu particles at 80 K versus exposure time, I: (a) r = 60 s and (b) r = 120 s. The exposure at I 320 s corresponds to saturation. &, was 2.7, corresponding to a cluster size of -105 A. The I5NO and CO reactants were fully mixed in the gas manifold with a 15NO:C0ratio of 1:l prior to admission into the UHV chamber. A linear heating rate of 5 Ws was used.
-
+
HREELS
L
15
b
NO+CO/CU/A~~O,/M~( 1 10)
),{dL, 1
L
;
885
,
615
530
0
500
1000
1500
a
_ , _
2000
, 2500
WAVENUMBER (cm-')
Figure 13. HREELS spectra of (a) a clean A 1 2 0 3 film,dAizoj= 8 A; (b) CdAl203, d ~ z o=, 8 A, 8cu = 0.67, following a saturation exposure = 8 A, 8cu = 2.7, of "NO CO at 80 K;and (c) CulA1203, following a saturation exposure of l5NO CO at 80 K. The ISNO and CO reactants were fully mixed in the gas manifold with a I5NO: CO ratio of 1:l prior to admission into the UHV chamber. The spectra were acquired using an electron beam with a primary energy of Ep = 42 eV and at the specularly reflected beam direction.
+
+
tion, are evident in Figure 12. The observation of a small C02 desorption peak (mle = 44) at temperatures between approximately 150-250 K in the TPD spectra of Figure 12 is a clear indication for the reaction between CO and 15N0 on the supported Cu particles. At Bcu 5 0.67, however, the COZ desorption peak (not shown here) was not observed. Since the COz yield is very small, the effects of particle size on CO? production are still unclear. The adsorption of 15N0 CO on particulate Cu deposits has also been studied using HREELS. Shown in Figure 13 are HREELS spectra acquired following a saturation exposure of 15N0 CO (15NO:C0 = 1:l). For reference, the spectrum of the clean A1203 surface is also included. Exposure of the clean
+
+
Wu and Goodman
9880 J. Phys. Chem., Vol. 98, No. 39, 1994 surface to the 15NO/C0 gas mixture gives rise to several adsorbate loss features in the 1000-2500 cm-I frequency range. Very similar to I5NO adsorption, the 1255 and 1465 cm-l losses are attributed to the @NO) mode of adsorbed 15N0 and I5N2O, respectively. The loss peak at 2110 cm-I is due to excitation of the carbon-oxygen stretch, v(CO), of adsorbed CO. From the studies described above, the I5NO and CO reaction on particulate Cu can be summarized as follows: 15~0(,)
-
+ I5N(,) - "N20 15N(,)+ I5N(,)- 15N2
Conservation of mass leads immediately to
Oh = d
(1)
where d is the thickness of equivalent monolayers. In the VW mode, the Auger intensity associated with a particular transition of a substrate, Is, is given by
I, = Q i - e
+ eck/As)
(2)
e
where is the peak intensity of the bare substrate and As the attenuation length of the ejected Auger electrons of the substrate traveling through the adlayers. Likewise, the Auger intensity of the adlayers, I,, can be written as follows:
15NO,,,
CO,,)
- co
(3)
e
where and 1, are the peak intensity of an infinitely thick adlayer and the attenuation length of the adlayers, respectively. Substituting dl8 for h in eq 1, the ratio of I, to Is yields the following equation:
where the subscript (a) denotes an adsorbed species. (4)
Conclusions The reaction of nitric oxide and carbon monoxide on model Cu/A1203 catalysts has been studied using surface science techniques. The results can be summarized as follows: 1. Highly ordered, stoichiometric A1203 films can be successfully synthesized on a Mo( 110) surface. The films (-20 A thick) exhibit excellent thermal stability and chemical inertness toward adsorption. 2. The growth of particulate Cu deposits, prepared by vacuum vapor deposition, on the A1203 films has been studied in the 80-800 K substrate temperature range. The elevated substrate temperatures employed during Cu deposition significantly improved the crystalline quality and the thermal stability of the Cu deposits. A model based on Auger measurements has been utilized to examine the number density and the average size of the Cu deposits. The number density of the Cu particles is estimated to be 1.4 x 10" cm-2 at a substrate temperature of 600 K. 3. Nitric oxide decomposition on the A1203 supported Cu particles has also been investigated. TPD and HREELS data clearly show that a fraction of the I5NO molecules react with the Cu particles to produce gaseous I5N2and I5N2O in the 90250 K temperature range. The formation of 15N2at 770 K is also evident and is due to recombination of adsorbed I5N atoms from 15N0 dissociation. The effects of retained oxygen, the adsorption temperature, and the particle size on the production of high-temperature I5N2were also examined. 4. The nitric oxide and carbon monoxide reaction has also been studied. The results showed that, besides the I5N2 and 15N20products arising form NO decomposition, a small quantity of C02 was produced via the reaction between adsorbed CO and adsorbed oxygen.
Acknowledgment. We acknowledge with pleasure the support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences. Appendix As an approximation, the growth of three-dimensional particles in the Volmer-Weber (VW) mode can be simply described by two parameters: 8, the portion of the surface covered by the particles, and h, the mean height of the particles.
Assuming that each particle grows in a hemispherical shape and that its morphology remains constant during growth, the dependence of €J on d can be derived. The volume of a particle, V , is equal to V = 2 ~ 3 1 3= hs, where r is radius and s is the bottom area of the particle which is equal to d , 8 can therefore be written in terms of the number of the particles, N , and s: 8 = Ns = 97cNh2/4. The dependence of 8 on d, using eq 1 to eliminate h, yields
6 = ('/@4
'I3d2l3
(5)
The particular power in the above power law dependence does not require the assumption of the growth of hemisphericalshaped particles. It is a direct consequence of our simple growth model, and the power 213 is a fingerprint for the quasi-isotropic growth. The detailed discussion of the quasi-isotropic growth model in terms of thermodynamic considerations has been given by Zhu et al.39 The average particle size, 2r, is easily obtained from the above derivation:
2 r = ( 1 2 1 7 1/3 ~ d~I13
(6)
In the case of CdAl203, the attenuation lengths, 1, (Cu(LMM)) and As (O(KLL)), are taken from the literature to be is 15.040 and 10.1 Ao: respectively. The value of obtained from the standard Auger spectra of Cu and A1203.16 The density of Cu particles, N , can then be computed using a curve-fitting procedure.
References and Notes (1) Poppa, H. Catal. Rev.-Sci. Eng. 1993, 35, 359. (2) Poppa, H. Vacuum 1984, 34, 1081. (3) Poppa, H. Ultramicroscopy 1983, 11, 105. (4) Iwamoto, M. In Future Opportunities in Catalytic and Separation Technology; Misono, M., Moro-oka, Y., Kimura, S., Eds.; Elsevier: Amsterdam, 1990; Vol. 54, pp 121-143. (5) Iwamoto, M.; Hamada, H. Catal. Today 1991, 10, 57. (6) Shelef, M. Catal. Left. 1992, 15, 305. (7) Hall, W. K.; Valyon, J. Catal. Lett. 1992, 15, 311. (8) Cho, B. K. J . Catal. 1993, 142, 418. (9) Liehr, M.; Thiry, P. A,; Pireaux. J. J.; Caudano, R. Phys. Rev. B 1986, 33, 5682.
Particulate Cu on Ordered A1203 (10) Pireaux, J. J.; Vermeersch, M.; Caudano, R. J . Electron Spectrosc. Relat. Phenom. 1992, 59, 33. (11) Wu, M.-C.; Comeille, J. S.; Estrada, C. A,; He, J.-W.; Goodman, D. W. Chem. Phys. Lett. 1991, 182, 472. (12) Wu, M.-C.; Estrada, C. A.; Corneille, J. S.; Goodman, D. W. J . Chem. Phys. 1992, 96, 3892. (13) Truong, C. M.; Wu, M.-C.; Goodman, D. W. J . Chem. Phys. 1992, 97, 9447. (14) Wu, M.-C.; Truong, C. M.; Goodman, D. W. J . Phys. Chem. 1993, 97, 9425. (15) Wytenburg, W. J.; Lambert, R. M. J . Vac. Sci. Technol. A 1992, 10, 3597. (16) Davis, L. E.; MacDonald, N. C.; Palmberg, P. W.; Riach, G. E.; Weber, R. E. Handbook of Auger Electron Spectroscopy; Physical Electronics Division, Perkin Elmer: Eden Prairie, MN, 1976. (17) Chen, J. G.; Crowell, J. E.; Yates, J. T., Jr. Phys. Rev. B 1987, 35, 5299. (18) Battye, F. L.; Jenkin, J. G.; Liesegang, J.; Leckey, R. C. G. Phys. Rev. B 1974, 9, 2887. (19) Chen, P. J.; Goodman, D. W. Su$. Sci. 1994, 312, L767. (20) Erskme, J. L.; Strong, R. L. Phys. Rev. B 1982, 25, 5547. (21) Strong, R. L.; Firey, B.; de Wette, F. W.; Erskine, J. L. Phys. Rev. B 1982, 26, 3483. (22) Crowell, J. E.; Chen, J. G.; Yates, J. T., Jr. Su$. Sci. 1986, 165, 37. (23) Chen, P. J.; Colaianni, M. L.; Yates, J. T., Jr. Phys. Rev. B 1990, 41, 8025. (24) Frederick, B. G.; Apai, G.; Rhodin, T. N. J . Electron Spectrosc. Relat. Phenom. 1990, 54/55, 415.
J. Phys. Chem., Vol. 98,No. 39, 1994 9881 (25) Frederick, B. G.; Apai, G.; Rhodin, T. N. Phys. Rev. B 1991, 44, 1880. (26) Chen, J. G.; Crowell, J. E.; Yates, J. T., Jr. Surf. Sci. 1987, 185, 373. (27) Tikhov, M.; Stolzenberg, M.; Bauer, E. Phys. Rev. B 1987, 36, 8719. (28) de John, A. M.; Niemantsverdriet, J. W. Surf. Sci. 1990,233, 355. (29) Xu, X.; He, J.-W.; Goodman, D. W. Surf. Sci. 1993, 284, 103. (30) Lee, E. H.; Poppa, H.; Pound, G. M. Thin Solid Films 1976, 32, 229. (31) Poppa, H. J . Appl. Phys. 1967, 38, 3883. (32) McLafferty, F. W.; Stauffer, D. B. The WileylNBS Registty of Mass Spectral Data; J o h n Wiley & Sons: New York; Vol. 1, p 2. (33) Press, W. H.; Flannery, P. B.; Teukolsky, S . A,; Vetterling, W. T. Numerical Recipes in Pascal The Art of Scientific Computing: Cambridge University Press: New York, 1989; p 459. (34) J o h n s o n , D. W.; Matloob, M. H.; Roberts, M. W. J . Chem. Soc., Faraday Trans. 1 1979, 75, 2143. (35) Balkenende, A. R.; Gijzeman, 0. L. J.: Geus, J. W. Appl. Surf. Sci. 1989, 37, 189-200. (36) So, S. K.; Franchy, R.; Ho, W. J . Chem. Phys. 1991, 95, 1385. (37) Wendelken, J. F. Appl. Su$. Sci. 1982, 11/12, 172. (38) Femindez-Garcia, M.; Conesa, J. C.; Illas, F. Surf.Sci. 1993,280, 441. (39) Zhu, Q.-G.; Zhang, A.-D.; Williams, E. D.; Park, R. L. Su$. Sci. 1986, 172, 433. (40) Tanuma, S.; Powell, C. J.; Penn, D. R. Su$. Interface Anal. 1988, 11, 577.
