1502 J Phys. Chem. 1990, 94, 1502-1508

Adsorptlon of Hydrogen, Carbon Monoxide, and Nitrogen on Rhenium(0001) and Copper Overlayers on Rhenium(0001)

J.-W. He and D. W. Goodman* Department of Chemistry, Texas A&M University, College Station, Texas 77843 (Received: April I I , 1989; In Final Form: July 25, 1989)

The interaction of hydrogen, carbon monoxide, and nitrogen with Re(0001) and with Re(0001) covered with submonolayer to multilayer Cu has been studied by thermal desorption spectroscopy (TDS) and low-energy electron diffraction (LEED). The adsorption of H2 on clean Re(0001) induces a 2 X 2 superstructure that can be produced by H2 exposure over a wide temperature range (1 10-300 K). Cu deposition on Re(0001) induces a new hydrogen desorption state with a higher bonding energy than that from bulk Cu. The population of this new hydrogen state is significantly reduced on the annealed Cu-Re(OOO1) surface and is believed to correlate with hydrogen adsorbed on the Cu overlayer perturbed by the Re substrate. Cu overlayers result in the attentuation of the H2 desorption peak associated with the Re substrate. On a 4.6 ML (ML = monolayer) Cu-covered Re(0001) surface, a new desorption state at -337 K is evident and is interpreted to arise from hydrogen desorption from 3-D Cu clusters. The TD spectrum of CO adsorbed on clean Re(0001) reveals four states with peak desorption temperatures at 175, 390, 961, and 1123 K. The initial adsorption of CO and annealing result in dissociation and leads to surface C and 0. Subsequent CO adsorption on this C- and 0-covered surface gives rise to two additional TDS features at -280 and 358 K. Adsorption of CO at 11 5 K induces a 5 X 1 structure, which upon annealing to - 1000 K, produces another square structure with a lattice constant 7 times that of the Re(0001) surface. This structure is likely composed of a superlattice of C formed from the dissociation of CO upon the annealing. Adsorption of CO on a Re(0001) surface with varying amounts of Cu shows that at low Cu coverages (<2 ML), a new desorption state is present, with a corresponding binding energy higher than that for bulk Cu and lower than that for clean Re(OO1). This state is interpreted to correspond to CO adsorbed at Cu sites perturbed by the underlying Re. At higher Cu coverages ( - 5 ML), the CO TDS features are essentially the same as those found for bulk Cu. No new TDS states of nitrogen have been observed following nitrogen adsorption on Cu-covered Re. The TDS features from the Re substrate are greatly reduced by Cu deposition, consistent with Cu serving as a simple site-blocking agent to the adsorption and dissociation of nitrogen on Re.

1 . Introduction The structural, electronic, and chemisorptive properties of metal

overlayers grown on single-crystal substrates have been the focus of considerable recent work. These metal overlayer systems are particularly exciting because they frequently demonstrate unique chemisorptive properties compared with either of the bulk metal components.'-I0 For example, it has been found that CO on monolayer Cu-covered Ru(0001) is stabilized relative to CO on bulk Cu.' In contrast to this behavior, the bonding of CO on a monolayer Ni or Pd film on W( 100) and W(110) has been found to be significantly weakened compared to bulk Ni or Pd.6,8

Recently, we have carried out studies of Cu overlayers on Re(0001) using Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), thermal desorption spectroscopy (TDS), and electron energy loss spectroscopy (ELS). Cu was found to form stable, uniform layers at a Cu coverage of <2 ML (1 ML = 1 monolayer = 1.52 X 10l5 atoms cm-*, the surface density of Re(0001)) and then nucleate into 3-D clusters upon subsequent Cu deposition and annealing.]] LEED studies showed a faint 2 X 2 structure at a submonolayer coverage of Cu and a 10 X 1 structure at OCu = 2 ML. The 2 X 2 structure is believed to be a superstructure of Cu on Re(0001). A -3-ML Cu-covered Re(0001) surface, annealed to 600 K, showed a rectangular pattern that is incommensurate with respect to the substrate lattice. Annealing further leads to nucleation of the Cu atoms into 3-

( I ) Yates, J. T., Jr.; Peden, C. H. F.; Goodman, D. W. J . Catal. 1985, 94,

(2) Goodman, D. W.; Yates, J. T., Jr.; Peden, C. H. F. Surf. Sci. 1985,

(3) Goodman, D. W.; Peden, C. H. F. J . Card 1985, 95, 321. (4) Houston, J. E.; Peden, C. H. F.; Blair, D. S.; Goodman, D. W. Surf.

(5) Peden, C. H. F.; Goodman, D. W. Ind. Eng. Chem. 1986, 25, 58. (6) Berlowitz, P. J.; Goodman, D. W. Surf. Sci. 1987, 187, 463. (7) Goodman, D. W.; Peden, C. H. F. J . Chem. SOC., Faraday Trans. I

516.

164, 417.

Sci. 1986, 167, 427.

1987, 83, 1967. (8) Berlowitz, P. J.; Goodman, D. W. Langmuir 1988, 4, 1091. (9) Greenlief, C. M.; Berlowitz, P. J.; Goodman, D. W. J . Phvs. Chem.

1987, 91, 6669. (IO) Campbell, C. T.; Goodman, D. W. J . Phys. Chem. 1988, 92, 2569. (11) He, J.-W.; Goodman, D. W. J . Phys. Chem., previous paper in this

issue.

dimensional clusters with the (1 11) plane parallel to the substrate surface. The details of the Cu overlayer growth on Re(0001) have been published elsewhere.Il

In this paper, we report the interaction of hydrogen, carbon monoxide, and nitrogen with clean Re(0001) and with the Cu- Re(0001) system studied by TDS and LEED. The emphasis of this work has been to address the unique properties of the Cu- Re(0001) system toward the adsorption of key probe molecules. Also we present new relevant data for the adsorption of hydrogen and carbon monoxide on clean Re(0001).

2. Experimental Section The experiments were carried out in a conventional UHV

chamber equipped with LEED, double-pass cylindrical mirror analyzer for AES, and a mass spectrometer for TDS. This system has been described in detail previously." The sample was spot-welded to two 0.5-mm Ta wires that allow resistive heating to 1500 K and sample cooling to 1 10 K. A W/5% Re-W/26% Re thermocouple was spot-welded to the sample edge. The Re- (0001) surface was cleaned by heating to 1200 K in O2 (lo-' Torr) for several hours and then flashing under vacuum to 1900 K. After this cleaning process, AES indicated an impurity free surface (< 1 atom % C, S), and LEED showed a sharp hexagonal pattern with a low background intensity.

Cu deposition was performed by resistively heating a filament wrapped with a high-purity Cu wire. Prior to each Cu deposition, the source was degassed extensively. AES detected no impurity accumulation during metal deposition.

High-purity N,, H,, and CO (99.97%) were used without further purification. A portion of the gas exposures was carried out with a gas doser consisting of a stainless steel nozzle positioned directly in front of the crystal face. This configuration resulted in a local gas pressure enhancement of approximately 20 compared to the background pressure.

A heating ramp of - 10 K s-I was used in every case to record TD spectra.

3. Results and Discussion 3.1. Hydrogen Adsorption on Re(0001). Over the past decade,

there have been a number of studies on the interaction of hydrogen

0022-3654/90/2094- 1502%02.50/0 0 1990 American Chemical Society

Adsorption of H2, CO, and N 2 on Re(0001)

t t ln z W I- z N ln

P 2

‘ 00 300 500 100 900 I IMPERITURE I & )

100 300 500 700 900

TEMPERATURE (K)

Figure 1. TD spectra of hydrogen from Re(0001) following hydrogen exposures of (a) 0.05, (b) 0.10, (c) 0.30, (d) 0.60, and (e) 2.0 langmuirs. (0 TD spectra of H2, D2 and HD from Re(0001) after exposure to -0.3 langmuir of H2 and then - 1 langmuir of D2 at 1 IO K. The heating rate for all TDS was - 10 K/s .

with Re(0001) using surface analytical techniques.I2-I6 These studies are particularly noteworthy in view of recent reports of the catalytic activity of this surface for ammonia ~ y n t h e s i s . ’ ~ - ~ ~ Both Ducros et al.I2 and Dooley and Hass13 have reported that upon adsorption of hydrogen at room temperature, no ordered surface structure is induced. The H2 TD spectrum12 measured subsequently indicated only a single peak. Similar desorption spectra for hydrogen from Re(0001) have also been reported by Kelly et al.I4 and Zaera and S0m0rjai.l~

Figure 1 shows the mass 2 TD spectra from a Re(0001) surface following various exposures of H2 at 110 K. The TD spectra show peak maxima that vary from -575 K for a 0.05-langmuir (1 langmuir = lod Torr-s) H2 exposure (Figure la) to 490 K for a 2.0-langmuir exposure (Figure le). The 2.0-langmuir exposure results in saturation of the H2 TD spectra.

The T D spectrum of Figure l e is in general agreement with that reported by Ducros et al.I2 However, in addition to the primary feature at -490 K, a small shoulder a t -420 K is also evident, which appears only at an intermediate hydrogen coverage (Figure lc,d). This shoulder is not apparent in the spectra obtained subsequent to either high exposure (Figure le) or low exposure (Figure la) of H2. A shoulder found by Kelly et al. at -240 K14 was not observed in this work. Figure l a shows that the peak maximum shifts toward lower temperatures as the adsorbed hy- drogen (Hads) coverage is increased. This is consistent with second-order kinetics arising from a recombination of atomic hydrogen during the desorption.

Deuterium isotope exchange experiments to verify atomic hy- drogen recombination were carried out, and the results of one of these experiments (Figure I f ) are inset in Figure 1. The Re(0001) surface at 110 K was first exposed to -0.3 langmuir of H2 and then to - 1 langmuir of Dz. Figure If indicates clearly that the desorption features in Figure la-e correspond to atomic hydrogen recombination. If we assume that the desorbed molecules of H2, D2, and H D are accurately represented by the corresponding areas in Figure If, an equilibrium constant of 4.5 can be derived for the isotope exchange reaction, consistent with complete exchange. The desorption feature around 100 K likely arises from desorption of H2 from the sample supports.

Exposure of H2 to Re(0001) a t 110 K induces a LEED pattern as shown in Figure 2. The half-order spots start to appear a t

(12) Ducros, R.; Housley, M.; Piquard, G.; Alnot, M. Surf: Sci. 1981,108,

(13) Dooley, G. J.; Haas, T. W. Surf: Sci. 1970, 19, 1. (14) Kelly, D. G.; Odriozola, J. A.; Somorjai, G. A. J . Phys. Chem. 1987,

(15) Zaera, F.; Somorjai, G. A. Surf. Sci. 1985, 154, 303. (16) Liu, R.; Ehrlich, G. Surf: Sci. 1982, 119, 207. (17) Spencer, N. D.; Somorjai, G. A. J . Phys. Chem. 1982, 86, L3493. (18) Spencer, N. D.; Somorjai, G. A. J . Catal. 1982, 78, 142. (19) Asscher, M.; Somorjai, G. A. Surf. Sci. 1984, 143, L389.

235.

91, 5695.

The Journal of Physical Chemistry, Vol. 94, No. 4, I990 1503

- 1 langmuir and appear to saturate at -2 langmuirs. Here, we will characterize this LEED pattern as a 2 X 2 structure, although the missing spots a t ( m + n + yield an incomplete 2 X 2 lattice. The fractional spots are sharp with weak intensities. Annealing the 2 X 2 phase to -300 K, then cooling to 110 K, does not change the pattern. Heating the 2 X 2 phase to 510 K results in its disappearance due to desorption of H2. The Hads coverage of the 2 X 2 structure corresponds to 0.25 monolayer (ML).

H2 exposures were also carried out a t substrate temperatures of 193, 260, and 300 K. The 2 X 2 pattern was observed with H2 adsorption at all three temperatures with subsequent cooling of the exposed surface to <200 K. The formation of the 2 X 2 phase of Hads on Re(0001) was not observed in previous studies of H2 adsorption at 300 K.l2,I3 This discrepancy regarding the observation of the 2 X 2 phase is believed to be caused by De- bye-Waller effects.

In a study of the 2 X 1 phase of Ha& on a Pd( 1 10) surface,20 the LEED intensity of the half-order beams decreased considerably upon heating the 2 X 1 phase from 130 to 210 K. At 230 K, the half-order spots were hardly discernible. In the present work, heating of the 2 X 2 phase of Ha& on Re(0001) to 300 K results in its disappearance. However, upon cooling to 110 K, the 2 X 2 pattern reappears. The disappearance of the 2 X 2 structure at 300 K then apparently arises from thermally induced disorder of the Hads with an increase in sample temperature rather than from a change in the surface coverage of Hads.

The disagreement of the present work with previous studies12J3 might also be caused by imperfections of the substrate surface. It has been proposed that Ha& first adsorbs on Re(0001) a t defect sites.16 The subsequent population of Ha& at other surface sites arises via migration of HA from these defect sites. However, there is no evidence that the sample used in the present studies had a higher surface concentration of defects than the Re samples em- ployed in the previous work.

3.2. Hydrogen Adsorption on Cu/Re(0001). TD spectra of hydrogen from a 1.1-ML Cu-covered Re(0001) surface are shown in Figure 3. The Cu coverage was determined by the Cu/Re Auger ratio, ZCu(60cV)/ZRe(176 eV), and the Cu coverage relationship established in our previous study of the Cu-Re system.” Figure 3a is the TD spectrum from an unannealed Cu-Re(0001) surface following a saturation exposure of H2 at 11 5 K. Figure 3b is a similar TD spectrum from Cu-Re(0001) annealed to 900 K. The H2 desorption spectrum from a clean Re(0001) is shown for reference in Figure 3c.

Three points are noteworthy in comparing Figure 3a-c. First, the coincidence of the desorption features in the three curves at 500 K suggests that there is a significant population of hydrogen at the Cu-Re interface where the bonding is essentially to Re. The mechanism for this adsorption likely involves H2 dissociation on free Re sites followed by atomic hydrogen diffusion through the copper overlayer to the Cu-Re interface. Hydrogen exhibits a similar adsorption behavior for the Ni/W( 100) system.6

The second point to note in Figure 3 is that there is an additional feature, labeled as yI , in Figure 3a at -390 K. The broad tail of curve b on the low-temperature side, compared to curve c, indicates that the y l state does not disappear completely for the annealed surface. The activation energy of desorption (E,) of the yl state from a Redhead analysis2’ is estimated to be 24.4 kcal mol-I, assuming first-order kinetics and a frequency factor of 10” s-I. (These assumptions will be used to estimate all subsequent activation energies for desorption.) The yl peak we believe is associated with the desorption of H2 from overlayer Cu. Since H2 does not dissociatively adsorb on Cu because of a high-energy barrier to dissociation,22 H2 in the y I state must first dissociate on free Re sites and then “spillover” onto the Cu overlayer. A peak desorption temperature of 390 K, which is considerably

(20) He, J.-W.; Harrington, D. A.; Griffiths, K.; Norton, P. R. Surf. Sci.

(21) Redhead, P. A. Vacuum 1962, 12, 203. (22) Greuter, F.; Plummer, E. W. Solid S?a?e Commun. 1983, 48.

1988, 198, 413.

1504 The Journal of Physical Chemistry, Vol. 94, No. 4, 1990 He and Goodman

b a

0 0 n

e o 0 vu :+... 0 . 0

Figure 2. (a) LEED pattern of the 2 X 2 phase of Hads on Re(0001) and (b) a schematic drawing of its structure. This pattern was induced by a 2-langmuirs exposure of H2 at 110 K. The primary energy of the incident electrons was 284 eV. In b, the filled circles correspond to integral spots, the open circles to H-induced spots, and the dashed circles to missing spots.

D;', c (x1/2) I \

100 200 300 400 500 600 700 800 900

TEMPERATURE (K)

Figure 3. (a) TDS of H2 from a 1.1-ML Cu-covered Re(0001) surface following a -20-langmuir H2 exposure. The Cu overlayer was unan- nealed and deposited at 115 K. (b) TDS of H2 from a 1.1-ML Cu- covered Re(0001) surface following a -20-langmuir H2 exposure. The Cu overlayer, annealed to 900 K, was deposited at 11 5 K. (c) TDS of H2 from a clean Re(0001) surface following a -20-langmuir H2 expo- sure at 115 K.

higher than the 300 K feature of H2 from C U , ~ suggests that the overlayer Cu to which the hydrogen of the y1 state is bonded is significantly altered by its interaction with the underlying Re. Hydrogen is therefore bonded to the overlayer Cu more strongly than to bulk copper but considerably weaker compared to hydrogen on clean Re(0001) ( E , for the 6 state of hydrogen on clean Re- (0001) is estimated to be -32 kcal mol-'). It is also possible that the hydrogen of the y1 state is associated with hydrogen adsorbed at mixed Cu-Re sites at the perimeter of 2-D Cu islands.

The last point to note in Figure 3 is that for the annealed Cu-Re surface, the population of hydrogen in the y1 state is significantly reduced. Annealing will undoubtedly lead to a decrease in the defect population and, perhaps, to the nucleation of 2-D islands with larger effective diameters than for the unannealed surface. Since the population of the y1 state in Figure 3 is determined by the kinetics of hydrogen diffusion from free Re sites to the Cu overlayer, larger Cu islands will lead to a significant reduction of the effective Cu perimeter and thus will inhibit the population of the yI state.

Figure 4 shows the desorption spectra of D2 from Re(0001) with varying amounts of Cu, unannealed, following -20 langmuirs of D2 exposure at 115 K. As the Cu coverage is raised from 0.2 to 1.8 ML, the y1 state at -390 K becomes increasingly popu- lated. At a high Cu coverage (-4.6 ML), a new state (y2) appears at -337 K with E, - 20 kcal mol-'. For a D2 exposure to a Cu-Re(0001) surface annealed to 900 K, the y states become much less discernible (Figure 5 ) . On the unannealed surface,

100 300 500 700 900

TEMPERATURE (K)

Figure 4. TDS of D2 from a Re(0001) surface with varying amounts of unannealed Cu. The Cu coverages are (a) 0.2, (b) 0.6, (c) 1.8, (d) 4.6, and (e) 0 ML. The H2 exposure in each case was -20 langmuirs and was carried out at 115 K.

100 300 500 700 900

TEMPERATURE (K) Figure 5. TDS of D, from a Re(0001) surface with varying amounts of Cu annealed to 900 K. The Cu coverages are (a) 0.2, (b) 0.6, (c) 1.8, and (d) 4.6 ML. The H, exposure in each case was -20 langmuirs and was carried out at 115 K.

the intensity of the desorption peak from clean Re(0001) decreases as the Cu coverage increases. The overlayer Cu therefore blocks the Re sites for D2 adsorption, as observed for hydrogen adsorption on Cu-covered Ru(OOO~) .~~ The y2 state in Figure 4e may be

(23) Yates, J. T., Jr.; Peden, C. H. F.; Goodman, D. W. J . Cutul. 1985, 94.

Adsorption of H2, CO, and N2 on Re(0001) - r I n e

100 300 500 700 900 1100 1300

TEMPERATURE (K)

Figure 6. TD spectra of CO from a clean Re(0001) surface following varying exposures at 115 K: (a) 0.3, (b) 0.6, (c) 0.8, (d) 1, and (e) 1.5 langmuirs.

from Dads on 3-D Cu clusters that are likely formed upon heating during the TDS. A similar state has been observed previously in the TDS of D2 from bulk C U . ~ ~ An alternative explanation is that the yI state is from H adsorbed on the Re surface where the H-Re bonds have been weakened by the presence of Cu. However, if H were to adsorb on Re sites, the desorption would likely be hindered in the Cu overlayer, causing an increase in the activation energy for desorption, rather than a decrease.

One significant point remains to be addressed, namely, the adsorption of hydrogen on Re sites after multilayer Cu desorption. Such adsorption may occur simply because of defects in the Cu overlayers. The AES results indicated that Cu forms 3-D clusters to some degree during multilayer desorption at 1 15 K.” Defects in the Cu overlayers could expose Re sites for H2 adsorption. Indeed, Figure 3 demonstrates that annealing 1.1 ML of Cu on Re(0001) considerably decreases the population of the state, compared to the unannealed Cu-Re(0001). A second possible explanation is that Re-Cu could alloy at the interface, providing Re sites for the dissociation of H2 However, the lack of miscibility between Re and Cu makes this mechanism unlikely. An alter- native explanation is that the ai state arises from H adsorbed a t the Cu-Re interface with H-Re bonds being weakened by the presence of Cu.

3.3. CO Adsorption on Re(0001). The interaction of CO with Re(0001) surface has been extensively It has been reported that C O adsorbs nondissociatively on Re(0001) a t 100 K24,25 and at 300 Ka2’ Both chemisorbed and physisorbed species of C O have been identified by photoemission spectroscopy (XPS and UPS).24 Heating a CO-covered Re(0001) surface to 400 K leads to dissociation of the C0.24*2s TDS following exposure to CO at 300 K reveals three states: the a1 and a2 between 300 and 500 K and a /3 state between 700 and 900 K.242758 The al appears only at high CO coverages following a -300 K e x p o s ~ r e . ~ ~ . ~ ~ The fl state arises from the recombination of surface carbon and oxygen. A Re(0001) surface exposed to CO at 100 K, however, exhibits an additional state a t -340 K. This extra feature is interpreted to correspond to the conversion of a physisorbed state into a chemisorbed state.23

An ordered structure with faint extra spots has been reported to be induced by CO adsorption at 100 K.26 This LEED pattern is interpreted to result from three domains with a rectangle unit cell of 4 5 X 4. Exposure of CO to Re(0001) at 550 K was found to induce a pattern, identified to be (2 X d/5)R30° domains, oriented 1 20° with respect to one another.28

(24) Tatarenko, S.; Alnot, M.; Ehrhartt, J. J.; Ducrcq R. Surf. Sci. 1985,

(25) Ducros, R.; Tardy, B.; Bertolini, J. C. Surf. Sci. 1983, 128, L219. (26) Tatarenko, S.; Ducros, R.; Alnot, M. Surf. Sci. 1983, 126, 422. (27) Braun, W.; Meyer-Ehmsenm, G.; Neumann, M.; Schwarz, E. Surf.

(28) Housley, M.; Ducros, R.; Piquard, G.; Cassuto, A. Surf. Sci. 1977,

152, 471.

Sci. 1979, 89, 354.

68, 271.

The Journal of Physical Chemistry, Vol. 94, No. 4, 1990 1505

h v, t z 3

00 N

100 300 500 700 900 1100 1300 TEMPERATURE (K)

Figure 7. TD spectra of CO from Re(O001) on a C- and 0-covered surface following an -2-langmuir CO exposure at 1 15 K. The C and 0 were created (a) by exposure to 2 langmuirs of CO at 115 K and then annealing the surface to IO00 K and (b) by exposure to 2 langmuirs of CO at 606 K.

Figure 6 shows the CO TD spectra following varying exposures of CO at 110 K. The spectra show four peaks located a t 175, 390,961, and 1123 K and labeled as al, a4, fll, and f12, respectively. Each TD spectrum was collected following exposure to CO after first flashing the Re(0001) to 1900 K to produce a Re(0001) surface free of C and 0. Figure 6a indicates that, a t low coverage, CO preferably dissociates above room temperature and recombines as C O at T > 900 K. Figure 6 also shows that the dissociated CO state saturates with a -0.6-langmuir C O exposure at 115 K. The E, for aI , a4, pl, and & are estimated to be 10.1, 23.2, 58.9, and 69.2 kcal mol-l, respectively.

The present work differs in several respects to the previous s t ~ d i e s : ~ ~ . ~ ~ ~ ~ (1) The desorption spectra near room temperature reveals only one state (the a4). (2) Two fl states are observed at temperatures >900 K. (3) One TDS feature (aI) at temperature - 160 K has been detected. The a1 likely corresponds to phy- sisorbed C O that has been identified previously by XPS and UPS.” In contrast, the multiple states near 300 K, we will show, correlate with the presence of C and 0 surface impurities.

Figure 7 shows the desorption spectra of CO from Re(0001) with surface C and 0 impurities. The surface C and 0 in Figure 7a was produced by first exposing the Re(0001) surface to 2 langmuirs of CO at 110 K and followed by an anneal to 1000 K. The annealed surface (with -8% atomic C, estimated from the AES spectrum) was then cooled to 110 K and reexposed to 2 langmuirs of CO. The TD spectrum then acquired (Figure 7a) exhibits two additional peaks (a2 and a3) a t -280 and 358 K, compared to CO desorption from clean Re(0001) (Figure 6). These two C O desorption states have been previously ob- served.24*2628 Heating a CO-covered Re(0001) surface to 1000 K leads to dissociation of CO and the formation of surface C and 0, as shown by previous XPS and UPS24 and HREELS studies.2s The AES line shape of C on the annealed surface in the present study also indicates that the carbon is of the “carbidic” type. The above experiment clearly shows that the a2 and a3 states of Figure 7 are associated with a presence of surface impurities of C and 0. The E, for the a2 and a3 states are estimated to be 16.5 and 21.2 kcal mol-’, respectively.

To further demonstrate this, the following experiment was carried out: First, the surface was exposed to 2 langmuirs of CO at 610 K where CO is known to adsorb on Re(0001) dissocia- tively;2s the surface was then cooled to 110 K and reexposed to 2 langmuirs of CO, and the TD spectrum measured (Figure 7b). As in Figure 7a, three desorption states near room temperature are evident. This experiment also concludes that the a2 and a3 are related to the presence of surface C and 0. Furthermore, the bonding of CO to the C- and 0-contaminated Re(0001) in the a2 and ag states apparently is weakened relative to CO on the clean Re(0001) surface (the a4 state).

Two ordered structures have been observed on Re(0001) fol- lowing a 2-langmuir CO exposure and a subsequent anneal. In Figure 8 is shown the 5 X 1 LEED structure, together with a

1506 The Journal of Physical Chemistry, Vol. 94, No. 4, 1990 He and Goodman

a b

Figure 8. (a) 5 X 1 LEED pattern of CO on Re(0001) and (b) its schematic drawing. The Re(0001) surface was first exposed to 3 langmuirs of CO at 1 15 K and then annealed to 340 K. The primary energy of the electron beam was 225 eV. In b, the filled circles correspond to integral spots and the open circles to CO-induced spots.

$07 Lt'

a b

Figure 9. (a) LEED pattern from dissociated CO and (b) its schematic drawing. The Re(0001) was exposed to 3 langmuirs of CO at 115 K and then annealed to 1000 K. E, = 235 eV. In b, the filled circles correspond to integral spots and the open circles to fractional order spots.

schematic drawing of the pattern, induced by CO adsorption a t 110 K and a subsequent anneal to 300 K. The fractional order spots are weak and diffuse at 110 K, probably due to a low mobility of CO at this temperature. Heating the 5 X 1 phase to room temperature sharpens the fractional spots. When the 5 X 1 structure is annealed further to 400 K, the fractional spots dis- appear, leaving lines along the ( 1010) direction. The lines are an indication of one-dimensional disorder of adsorbates along this direction. Annealing to - 1000 K produces a pattern shown in Figure 9. This pattern appears to be square with the lattice constant 8a where a is the lattice constant of the Re(0001) surface. Oxygen is known to form a 2 X 2 structure on Re(0001);I3 thus, the square pattern likely is associated with surface carbon rather than oxygen, although it is possible that oxygen may form a different surface structure in the presence of surface C. This pattern disappears upon heating the surface to 1200 K, the tem- perature at which recombination of C and 0 occurs as the p CO states. The LEED patterns observed in this work differ from the previously reported 4 7 X 4 structure found following CO ad- sorption at 110 KZ6 and the ( 4 5 X 3)R3Oo structure observed following CO exposure at 550 K.28 The reason for this dis- agreement is not clear.

3.4. CO Adsorption on CulRe(000Z). The TD spectra of CO from Re(0001) with varying Cu coverages, deposited at 11 5 K, are shown in Figures 10 and 1 1. CO exposure was carried out at a substrate temperature of 115 K following Cu deposition (without a subsequent anneal). For Cu coverages between 0.3

0 , a

100 300 500 700 900 1100 1300

TEMPERATURE (K) Figure 10. TD spectra of CO from Cu-Re(0001) surfaces following a -20-langmuir CO exposure at 115 K. Cu coverages are (a) 0.3, (b) 0.5, and (c) 0.8 ML.

and 0.8 ML, the intensity of the cy4 state from Re(0001) is gradually attenuated, a new state (y2) is - 15.9 kcal mol-' at a Cu coverage of 0.8 ML. A new peak, labeled as yl, begins to appear at -220 K. As the Cu coverage increases above 0.8 ML, the y1 state gradually becomes dominant and its peak maximum shifts to a lower temperature.

The TDS of CO from Cu( 1 lo), reported previo~sly,2~*~~ consists

Adsorption of H2, CO, and N2 on Re(0001)

1: \ I: !

100 300 500 700 900 1100 1300

TEMPERATURE (K)

Figure 11. TD spectra of CO from Cu-Re(0001) surfaces following a -20-langmuir CO exposure at 115 K. Cu coverages are (a) 1.5, (b) 2.9, and ( e ) 5.2 ML.

m a (r

v

c m Z W I-

t

m m

z v

100 150 200 250 300

TEMPERATURE (K) Figure 12. TD spectra of N2 between 100 and 300 K from unannealed Cu-Re(0001) surfaces following a -20-langmuir N2 exposure at 115 K. Cu coverages are (a) 0, (b) 0.3, (e ) 0.5, (d) 1.3, and (e) 2.7 ML.

of a single peak located at 200-220 K. The activation energy of the desorption was found to be coverage-dependent ( E , = 12.9 kcal mol-' for -0.3 ML of CO), and the sticking coefficient was constant for CO coverages between 0 and 0.5 ML.29-30 The y1 state then, found in the present study for the 5.2-ML Cu-covered Re(0001), is consistent with the previous work for bulk Cu. The activation energy for the yI is estimated to be 13.0 kcal mol-' for the peak at 222 K and 9.1 kcal mol-' for the peak at 157 K. The decrease in the activation energy with Cu coverage may either be due to a reduction in the substrate perturbation of the overlayer Cu or to the overall increase in C O surface coverage.

The y2 desorption state at -277 K is evident in Figures 11 and 12 at low Cu coverages. This state may arise from CO adsorption on a Cu overlayer that binds CO more weakly compared to the clean Re(0001) surface (Figure 6) and more tightly than to a surface of bulk Cu.29,30 As the Cu coverage increases, the ad- sorption sites on the mixed perimeter sites become less prevalent and the population of the y2 state is correspondingly reduced through a kinetic effect as discussed in the desorption of hydrogen from Cu-Re(0001) in section 3.2.

Figure 11 also shows that the desorption states (a4, PI, and &) from the Re(0001) substrate are still apparent even with Cu coverages as high as 5.2 ML. This is probably caused by the formation of 3-D clusters during the deposition and heating. If this is the case, the accessible surface sites of Re likely are formed below 220 K, the temperature at which CO has completely de- sorbed from Cu. This result is consistent with the AES results,

The Journal of Physical Chemistry, Vol. 94, No. 4, 1990 1507

h

I I L - t

v,

I

c

2 1 ob0 1100 1200 1300 1400

TEMPERATURE (K)

Figure 13. TD spectra of N2 between lo00 and 1400 K from unannealed Cu-Re(0001) surfaces following a -20-langmuir N2 exposure at 11 5 K. Cu coverages are (a) 0, (b) 0.3, (e) 0.5, (d) 1.3, and (e) 2.7 ML.

indicating 3-D nucleation of Cu at 115 K." 3.5. Nitrogen Adsorption on Cu-Re(0001). In a previous

study,31 we have shown that the TD spectrum of N2 from Re- (0001) consists of two peaks, one at - 160 K, corresponding to molecularly adsorbed N2, and a second between 900 and 1300 K, that relates to the recombination of dissociated nitrogen. In this work,31 two ordered surface structures were observed by LEED. The low-temperature state a t 160 K is associated with a 4 X 1 structure. Heating this 4 X 1 phase to >300 K dissociates N2 to form the 2 X 2 s t r ~ c t u r e . ~ '

It has been established that nitrogen does not dissociatively chemisorb on Cu surfaces [ref 32 and references cited therein]. However, atomic nitrogen, generated by e--beam dissociation of molecular nitrogen, is found to adsorb on Cu and form ordered structure^.^^^^^

In Figures 12 and 13 are shown the TD spectra of N2 from Re(0001) with varying coverages of Cu deposited at 115 K. The N2 exposure was carried out a t 115 K on the unannealed Cu overlayer. For reference, the TD spectrum for N2 adsorption on clean Re(0001) is presented in Figures 12a and 13a. The de- sorption spectra for the Cu-covered surfaces show two peaks at - 170 and 1220 K. (No desorption peaks are observed between 300 and 1000 K.) Compared to clean Re(0001), it is noteworthy that no new peak is induced by Cu deposition, indicating that nitrogen does not spill over from the Re to the Cu overlayers. Also, importantly, the population of both the a and /3 states are con- siderably reduced by the presence of Cu. At a Cu coverage of 2.7 ML, the p state is barely discernible from the background. Obviously, the Cu overlayers block the Re surface sites for nitrogen adsorption and dissociation. At a Cu coverage of 2.7 ML, the p state has essentially disappeared; however, the a-state population is still -20% that of clean Re(0001). These results suggest that the site requirements on Re(0001) for N2 dissociation are more demanding than the site requirements for molecular adsorption.

4. Summary ( I ) For H2 on clean Re(0001), the TD spectrum shows one peak

that follows second-order kinetics. Isotope exchange experiments confirm that this TDS feature arises from the recombination of dissociated hydrogen. Exposure of H2 at 110 K induces a LEED pattern, described as a 2 X 2 structure of H on Re(0001).

( 2 ) The TD spectrum of H2 from a Re(0001) surface covered by Cu reveals a new hydrogen desorption state with a higher binding energy than that from a surface of bulk Cu. This state is attenuated by annealing the Cu prior to hydrogen adsorption and likely corresponds to hydrogen desorption from overlayer Cu sites perturbed by the underlying Re substrate. For a Cu coverage of -4 ML, a new desorption state at -337 K is induced, very likely corresponding to desorption of hydrogen from 3-D Cu clusters.

( 2 9 ) Harendt, C.; Goschnik, J.; Hirschwald, W. Surf. Sci. 1985, 152, 453. (30) Woodruff, D. P.; Hayden, B. E.; Prince, K.; Bradshaw, A. M. Surf.

Sci. 1982, 123, 397.

(31) He, J.-W.; Goodman, D. W. Surf. Sci., in press. (32) Lee, R. N.; Farnsworth, H. E. Surf, Sci. 1965, 3, 461. (33) Mohanmed, M. H.; Kesmodel, L. L. Surf. Sci. 1987, 185, L467.

1508 J . Phys. Chem. 1990, 94, 1508-1516

(3) Four states at 175,390,961, and 1123 K have been observed in the TD spectrum following the adsorption of CO on clean Re(0001). It has been shown that the two additional desorption features observed in previous work between 280 and 358 K cor- relate with a presence of surface impurities of C and 0. Ad- sorption of CO on Re(0001) at 11 5 K induces a 5 X 1 structure, whereas annealing this 5 X 1 phase to - loo0 K produces a second square structure with a lattice constant 7 times that of the Re- (0001) surface. This structure is incommensurate with respect to the substrate lattice and likely arises from C produced via dissociation of CO by annealing. (4) At low Cu coverages, TDS of Cu/Re(0001) following CO

adsorption reveals a new desorption state that has a binding energy lying between that for CO on bulk Cu and that for CO on clean Re(0001). This state is presumed to arise from CO adsorbed on

a Cu overlayer perturbed by the underlying Re. At high Cu coverages (>5 ML), the TD spectra of CO are dominated by bulklike features of CO on Cu.

(5) No new desorption states of nitrogen from overlayer Cu on Re(0001) compared to clean Re(0001) are observed; however, Cu is found to block Re sites for the adsorption and dissociation of nitrogen.

Acknowledgment. We acknowledge with pleasure the support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, and the Robert A. Welch Foundation.

Registry No. Cu, 7440-50-8; Re, 7440-15-5; H, 1333-74-0; CO, 630-08-0 N, 7727-37-9.

Prediction of Low Occupamy Sorption of Alkanes in SiHcalite

R. Larry June, Alexis T. Bell, and Doros N. Theodorou*

Department of Chemical Engineering, University of California, Berkeley, Berkeley, California 94720 (Received: April 12, 1989)

Alkane sorption in a pentasil zeolite has been studied through molecular simulations. A series of alkanes, including methane, n-butane, and three hexane isomers, were studied in silicalite by using a detailed atomistic representation that allows for torsion around skeletal bonds. Statistical mechanical principles have been employed to predict sorption equilibria at low occupancy. Henry’s constants and ismteric heats of sorption were calculated through the evaluation of configurational integrals with a Monte Carlo integration scheme. Results are in good agreement with experiment. The spatial distribution of sorbate molecules within the pore network, as well as perturbations to their conformation due to confinement in the pores, were determined via Metropolis Monte Carlo algorithm. Simulations of sorbate spatial distributions show that linear alkanes, such as n-butane and n-hexane, prefer to reside in the channels and avoid the channel intersections; on the contrary, bulky side groups in branched alkanes, such as 2- and 3-methylpentane, force these molecules toward the more spacious channel intersections. An analysis of sorbate conformations indicates that molecules are perturbed from the ideal gas dihedral angle distribution toward the more linear trans states when confined in silicalite.

Introduction Zeolites are crystalline aluminosilicates consisting of covalently

bonded networks of Si04 and A10, tetrahedra arranged in such a way as to form a network of connected cavities and channels with dimensions commensurate with molecular diameters (3-10 A). Because of their unique structure, zeolites have found wide application in processes such as adsorption, ion exchange, and cata1ysis.l” For the optimization of many of these applications, knowledge of the effects of zeolite structure and composition on adsorptive properties is essential. While such properties can be established experimentally, it would be highly desirable to develop theoretical methods for predicting zeolite adsorption behavior, so as to facilitate selection and optimization of zeolites for a given purpose. An attractive approach to this goal is the use of computer simulation based on an atomistic description of sorbate-zeolite interactions.

Previous atomistic models of molecular sorption into narrow pores have been of two types: those dealing with idealized pore systems and those dealing with zeolites or similar cagelike structures. Simulation techniques, including molecular dynamics, grand canonical ensemble Monte Carlo, and Gibbs ensemble Monte Carlo among others, have been employed to study phe- nomena such as fluid structure, phase transitions, hysteresis effects, and wetting7-I4 for simple Lennard-Jones fluids in idealized pores. Theoretical studies15J6 involving the application of mean field approximations have also been employed to examine the phase behavior of fluids in micropores.

* Author to whom correspondence should be addressed.

0022-3654/90/2094-1508$02.50/0

Simulation methods have also been used to investigate sorption in zeolites. Grand canonical ensemble Monte Carlo simulations have been employed to determine sorption equilibria over a full range of occupancies in model zeolite structures by Soto and

( I ) Riekert, L. In Advances in Catalysis; Eley, D. D., Pines, H., Weisz,

(2) Breck, D. W. Zeolite Molecular Sieves: Structure, Chemistry and

(3) Zeolite Chemistry and Catalysis; Rabo, J. A., Ed.; ACS Monograph

(4) Barrer, R. M. Zeolites and Clay Minerals as Sorbents and Molecular

(5) Ruthven, D. M. Principles of Adsorption and Adsorption Processes;

(6) Newsam, J. M. Science 1986, 231, 1093. (7) Snook, I. K.; van Megen, W. J . Chem. Phys. 1980, 72, 2907-2913. (8) Peterson, B. K.; Gubbins, K. E.; Heffelfinger, G. S.; Marconi, U. M.

(9) Panagiotopoulos, A. Z. Mol. Phys. 1987, 62, 701-719. (IO) Peterson, B. K.; Gubbins, K. E. Mol. Phys. 1987, 62, 215-226. (11) Tan, Z.; Van Swol, F.; Gubbins, K. E. Mol. Phys. 1987, 62,

121 3-1224. (12) Schoen, M.; Cushman, J. H.; Diestler, D. J.; Rhykerd, C. L. J. Chem.

P h y ~ . 1988,88, 1394-1406. (13) MacElroy, J. M. D.; Suh, S.-H. Presented at the 1987 Spring Na-

tional Meeting of the A.I.Ch.E., Houston, TX, March 29 to April 2 1987. (14) Maada, J. J.; Tirrell, M.; Davis, H. T. J. Chem. Phys. 1985, 83,

P. B., Eds.; Academic Press: New York, 1970; Vol. 21; p 281.

Use; Academic Press: New York, 1978.

171; American Chemical Society: Washington, DC, 1976.

Sieues; Academic Press: New York, 1978.

Wiley Interscience: New York, 1984.

B.; van Swol, F. J . Chem. Phys. 1988.88, 6487-6500.

1888-1901 .- (15) Peterson, B. K.; Walton, J. P. R. B.; Gubbins, K. E. J . Chem. SO~.,

Faradav Trans. 2 1986.82, 1789-1800. (16fEvans, R.; Marconi, U. M. B.; Tarazona, P. J. Chem. SOC., Faraday

Trans. 2 1986, 82, 1763-1787.

0 1990 American Chemical Societv