Structure of Monolayer and Multilayer Magnesium Chloride Films

Feb 15, 1997 - The structures of magnesium chloride thin films grown on a Pd(111) surface have been studied using low-energy electron diffraction (LEE...
0 downloads 0 Views 221KB Size
Download PDF
2090

Langmuir 1997, 13, 2090-2096

Structure of Monolayer and Multilayer Magnesium Chloride Films Grown on Pd(111) D. Howard Fairbrother, Joel G. Roberts, Simone Rizzi,† and Gabor A. Somorjai* Department of Chemistry, University of California, and Lawrence Berkeley National Laboratory, University of California, Berkeley, California 94720 Received July 10, 1996. In Final Form: January 2, 1997X The structures of magnesium chloride thin films grown on a Pd(111) surface have been studied using low-energy electron diffraction (LEED), temperature-programmed desorption, and Auger electron spectroscopy. At coverages slightly less than 1 monolayer a Pd(111)-(4×4)-MgCl2 LEED pattern is observed by LEED that correlates with a coincident lattice match between Pd(111) and the (0001) or (001) face of either R or β close-packed MgCl2, respectively. At full monolayer coverage, the existence of significant attractive adsorbate-substrate interactions is sufficient to induce a lattice compression of 0.36 Å in the MgCl2 ionic solid in order to maximize the surface coverage. This more compressed structure gives rise to a Pd(111)-(x13×x13)R13.9°-MgCl2 overlayer. Upon deposition of thicker films the bulk structure is recovered, producing a simple MgCl2-(1×1) LEED pattern characteristic of the (001) orientation of crystalline MgCl2. This transition is consistent with the greater strength of interactions between successive MgCl2 layers compared to those between the Pd(111) surface and the first MgCl2 monolayer.

(I) Introduction The study of thin films grown on dissimilar materials has been the subject of a large number of scientific investigations. In general these studies have focused on metallic overlayer structures.1 Compared with these systems, the growth of insulators, such as wide band gap ionic solids, has been rather little studied.2 Their mechanism of growth, however, represents an interesting problem because of the large cohesive energies present within these ionic lattices. Recently, molecular beam epitaxy (MBE) has been applied to the growth of alkali halides on dissimilar salt surfaces.3 Using reflection high-energy electron diffraction and X-ray diffraction, Yang and co-workers, studying the growth of alkali halides, have shown that epitaxial growth can occur even below room temperature.4 Additional studies by Koma et al.5 also show that on a KBr substrate KCl and LiF grow with their crystallographic axes parallel to that of the substrate. However, although KCl films are produced in a layer-by-layer fashion, LiF growth occurs via three-dimensional island formation. Magnesium chloride (MgCl2), an alkali earth metal halide, has three crystalline forms.6 The most common is the cubic close-packed R-MgCl2 form with double chlorine (Cl-Cl) layers (...ClMgCl-ClMgCl...) and interstitial Mg2+ ions in 6-fold coordination. A thermodynamically less stable, hexagonal close packed (hcp) β form and a rotationally disordered version of the hcp crystal, δ-MgCl2, also exist. All of these forms can be categorized as layered structures with strong ionic bonds within layers and weaker van der Waals’ forces between adjacent layers. In addition magnesium chloride is an integral part of the active support for the Ziegler-Natta catalyst system †

Present address: Edison-Ricerca e Sviluppo, Milano, Italy. Abstract published in Advance ACS Abstracts, February 15, 1997. X

(1) Falicov, L. M.; Pierce, D. T.; Bader, S. D.; Gronsky, R.; Hathway, B.; Hopster, H. J.; Lambeth, D. N.; Parkin, S. S. P.; Prinz, G.; Salamon, M.; Schuller, I. K.; Victoria, R. H. J. Mater. Res. 1990, 5, 1299. (2) Klauser, R.; Oshima, M.; Sugahara, H.; Murata, Y.; Kato, H. Phys. Rev. B 1991, 43, 4879. (3) Yang, M. H.; Flynn, C. P. Phys. Rev. B 1990, 41, 8500. (4) Yang, M. H.; Flynn, C. P. Phys. Rev. Lett. 1989, 62, 2476. (5) Nakamura, Y.; Saiki, K.; Koma, A. J. Vac. Sci. Technol. 1992, A10, 321. (6) Dusseault, J. J. A.; Hsu, C. C. JMSsRev. Macromol. Chem. Phys. 1993, C33, 103.

S0743-7463(96)00680-4 CCC: $14.00

responsible for the polymerization of ethylene and propylene to their respective homo- and copolymers.7,8 In this paper we present results of a study on the growth and characterization of MgCl2 thin films grown on Pd(111). Instead of utilizing MBE an evaporation source has been employed to produce gas phase MgCl2. This approach is shown to be an effective means to produce monolayer and multilayer MgCl2 films on Pd(111) whose structures can be probed using low-energy electron diffraction (LEED). The resulting surface structures at monolayer and multilayer (thin film) coverages of MgCl2 have been interpreted on the basis of the interplay between adsorbate-substrate and adsorbate-adsorbate interactions. (II) Experimental Section Experiments were performed in an ultrahigh vacuum (UHV) chamber which, after bakeout, routinely attains a base pressure of 1000 K by electron bombardment from a rear-mounted W filament. Temperature measurements were achieved by means of a chromel-alumel thermocouple attached to the rear face of the crystal. The Pd(111) surface was cleaned by annealing at ≈950 K in 1 × 10-7 Torr of O2 followed by repeated cycles of Ar+ sputtering (0.5-3 keV, PAr ) (2-3) × 10-5 Torr) until the surface was judged clean by AES. The sample was then annealed at ≈1000 K for periods of several hours until there was no observable change in the quality of the Pd(111)-(1×1) LEED pattern. (7) Barbe, P. C.; Ceechin, G.; Noristi, L. Adv. Polym. Sci. 1986, 81, 1. (8) Encyclopedia of Polymer Science and Engineering; Wiley: New York, 1988; Vol. 13. (9) Weiss, W.; Somorjai, G. A. J. Vac. Sci. Technol. 1993, A11, 2138.

© 1997 American Chemical Society

MgCl2 Films Grown on Pd(111)

Langmuir, Vol. 13, No. 7, 1997 2091

Figure 1. (a) Schematic of the magnesium chloride deposition source. (b) Mass spectrum of the source effluent, operating at 810 K, recorded using an impact energy of 70 eV. Magnesium Chloride Source. Figure 1a shows a schematic of the magnesium chloride evaporation source used for deposition. It consists of a graphite crucible containing MgCl2 powder. Temperature control of the crucible is effected by resistive heating in conjunction with a Eurotherm temperature controller, using a chromel-alumel thermocouple attached directly to the MgCl2 crucible. To prevent desorption from the surrounding surfaces during deposition, liquid N2 is passed through the source, cooling a copper shroud which encompasses the entire assembly (see Figure 1a). Figure 1b shows a mass spectrum (30-120 amu) of the source effluent, operating at ≈810 K, recorded using an impact energy of 70 eV. This was obtained by subtracting the mass spectrum with the source at 720 K from one obtained with a source temperature at 810 K, the former being at a temperature where MgCl2 sublimation is insignificant. The resultant spectrum shows three distinct sets of peaks which can be correlated with 35/37Cl+, Mg35/37Cl+(amu 59 and 61), and Mg35/37Cl + (amu 942 96).10 It should be noted that the peaks associated with Cl+ and MgCl+ fragments can be attributed to cracking fragments from MgCl2 rather than from atomic chlorine or molecular MgCl produced by the source. The small peak at mass 44 can be ascribed to CO2 which has not been completely removed through background subtraction. Figure 1b clearly illustrates that, using an appropriate temperature, the source can be utilized to provide a flux of essentially pure MgCl2, consistent with thermodynamic equilibria.11 Deposition was typically accomplished by positioning the Pd(111) surface 1-2 cm in front of the heated source for a controlled exposure time. AES measurements carried out at various positions on the surface revealed that careful control of the surface’s spatial orientation with respect to the source was necessary to ensure that a uniform flux of MgCl2 at the surface is sustained. Unless noted, all depositions were carried out with a source temperature of 810 K.

(III) Results (a) Adsorption/Desorption Characteristics. Figure 2 shows the TPD of MgCl2 films produced by exposing the Pd(111) surface, held at room temperature, to the heated source for various exposure times. Due to the difficulty in reproducing exact MgCl2 coverages from similar source exposure, the relative coverage in each case is determined by postexposure AES. The desorption profile monitored at m/e ) 94-96, encompassing the Mg35Cl2+ and Mg35Cl37Cl2+ parent peaks, displays two distinct desorption features. The lower peak exhibits an (10) The Mg37/37Cl2+ isotope (amu 98) is expected to only be ≈11% as intense as the 35/35MgCl2+ isotope (amu 94). Consequently it cannot be resolved above the background noise (see Figure 1b). (11) Pankratz, L. B. Thermodynamic Properties of Halides; United States Department of the Interior, Bureau of Mines: Washington, DC, 1984.

Figure 2. Desorption profile of magnesium chloride deposited on Pd(111) monitored at mass m/e ) 95-97 corresponding to the parent MgCl2 mass. These profiles are expressed as a function of increasing MgCl2 coverage as measured by AES.

exponential rise in the initial desorption rate accompanied by a sharp falling edge. This profile is consistent with multilayer desorption, as has been reported previously for MgCl2 on polycrystalline gold.12 In contrast, the higher temperature desorption peak displays characteristic firstorder kinetics with a roughly symmetric peak profile. Although the nonlinearity of the temperature ramp does not allow for a detailed kinetic analysis, the desorption behavior shown in Figure 2 indicates the presence of both multilayer and monolayer states. It should also be noted that there is some contribution to the multilayer peak from MgCl2 adsorbed on the manipulator support. Figure 3a shows the AES of the surface, held at room temperature, after exposure to the MgCl2 source for 30 min. The two peaks clearly visible at 181 and 1160 eV can be correlated with chlorine and magnesium, respectively. The lack of any Pd substrate peaks is consistent with MgCl2 multilayer formation. The existence of MgCl2 multilayers can be verified by utilizing the measured AES peak-to-peak ratios (ICl and IMg). This measured signal is influenced by the inherent stoichiometry (Mg/Cl)stoich., the different sensitivities at 3 keV (sCl and sMg),13 and the mean free paths (λCl and λMg)14 associated with the Auger secondary electrons of these two chemical elements. Since Figure 3a was recorded at normal incidence, this relationship can be expressed in the form

( ) ( ) ( ) IMg Mg ) ICl Cl

stoich.

sCl 1 - (exp(-1/λMg)) sMg 1 - (exp(-1/λCl))

(1)

Using eq 1 the (Mg/Cl)stoich. is determined to be 1:2.05, which is in excellent agreement with the expected 1:2 chemical stoichiometry. This result supports the idea that the source effluent is composed of only MgCl2. (12) Magni, E.; Somorjai, G. A. Appl. Surf. Sci. 1995, 89, 187. (13) ESCA Operator’s Reference Manual; ESCA Version 4.0 and MultiTechnique Version 2.0; Perkin-Elmer Corporation, Physical Electronics Division: Eden Prairie, MN, 1998. (14) Riviere, J. C. Surface Analytical Techniques; Oxford University Press: Oxford, 1990.

2092 Langmuir, Vol. 13, No. 7, 1997

Fairbrother et al.

cases, the thickness of the MgCl2 overlayer was estimated by monitoring the attenuation of the Pd peak at 333 eV upon deposition (PdMgCl2/PdClean.) from the equation

(PdMgCl2/PdClean.) ) exp(-d/λPd)

Figure 3. AES of the Pd(111) surface held at (a) room temperature and (b) 650 K and exposed to the source for 30 min. Both spectra were recorded using 2 mA of emission current and a 3 keV primary beam energy.

The presence of a distinct, thermodynamically more stable monolayer phase is confirmed by comparing AES uptake measurements recorded as a function of exposure time. In these experiments, the sample was held at two separate temperatures during deposition, 298 and 650 K, the latter falling within the multilayer desorption profile (see Figure 2). With the sample held at room temperature, AES results indicate a rapid uptake of MgCl2 leading to the production of a multilayer film in less than 5 min. In contrast, a 30 min deposition at 650 K results in the AES shown in Figure 3b which remains virtually unchanged for deposition times of up to 2 h. The same surface spectrum can also be obtained by thick film deposition at room temperature followed by flashing the surface to 600 K to effect multilayer desorption. Additional support for the idea that Figure 3b corresponds to a monolayer coverage of MgCl2 on Pd(111) can be obtained by application of eq 1 in the limit of single layer coverage. At monolayer coverage, in the absence of detailed structural information, the 1 - exp(-1/λ) term, which represents the relative contribution from species below the surface, can be neglected. As a result of the considerable difference in the mean free paths associated with chlorine and magnesium, this effect produces a measurable change in the Mg:Cl peak-to-peak ratio in moving between multilayer and monolayer MgCl2 films. Indeed, application of eq 1 in these two limits is predicted to change the measured Mg:Cl ratio from 1:20 to 1:8.6. This is in excellent agreement with the 1:20.2 Mg:Cl ratio, in Figure 3a for a multilayer film, and the 1:8.4 shown in Figure 3b. It should be noted that an electron-stimulated desorption process, analogous to that previously reported for MgCl2 films grown on a gold substrate,12 produces a reduction in the Cl concentration along with a concomitant increase in oxygen signal associated with MgO formation. So as to minimize this effect, all AESs used in this study are the result of a single scan. (b) Structural Analysis. Thick multilayer MgCl2 films grown at room temperature yielded no ordered structures. However, when deposition was carried out on a Pd(111) surface held at 643 K, three distinct LEED patterns were observed as a function of increasing coverage. In these

(2)

where d represents the overlayer thickness and λPd the mean free path associated with secondary electron production from Pd,14 both expressed in angstroms. Two distinct LEED patterns are observed for MgCl2 coverages ≈1 monolayer (ML). This coverage assignment is based upon the similarity of the measured AES with that shown in Figure 3b. Figure 4a shows a (4×4) LEED pattern observed at this coverage, recorded with an incident electron energy of 130 eV. Upon flashing the surface to ≈1000 K, the ordered overlayer desorbs intact leaving the 3-fold symmetric spots of the Pd(111) surface, as shown in Figure 4b. Since both LEED patterns were recorded at the same geometric location and incident energy, it is possible to identify the position of the firstorder Pd spots within the overlayer structure as shown in Figure 4a. Pd peak attenuation measurements based on eq 2 indicate that this layer is 5-6 Å thick. At approximately the same monolayer MgCl2 coverage, another LEED pattern, shown in Figure 5, is also observed, recorded at (a) 49 eV and (b) 104 eV. The positions of the underlying Pd(111) spots are also indicated by white circles. This pattern can be identified as a Pd(111)(x13×x13)R13.9°-MgCl2 structure comprised of two separate domains.15 It should also be noted that it is this pattern, rather than the (4×4) pattern (Figure 6b), which is observed after multilayer desorption from a >1 ML film, initially deposited at room temperature. Several attempts to grow multilayers by means of MgCl2 deposition for >1 h produced films of approximately 12 and 24 Å thicknesses. In contrast to the more complicated LEED patterns observed at lower coverages, these thicker MgCl2 films exhibit a simple MgCl2(001)-(1×1) surface structure, as shown in Figure 6a with no observed contribution to the LEED pattern from the underlying Pd(111) substrate spots. Also shown in Figure 6b is the LEED pattern of the clean Pd(111) following complete thermal desorption of the MgCl2 recorded at the same incident beam energy (100 eV). A visual comparison of parts a and b of Figure 6 reveals that the crystallographic orientation of the ordered overlayer film is parallel to that of the Pd(111) unit cell. Furthermore, by comparison of the LEED patterns shown in Figure 6 the relative ratio of the unit cells is determined to be

aPd ≈ 3/4aMgCl2

(3)

Intensity-voltage (I-V) analysis of this overlayer film reveals that despite the absence of the Pd(111) substrate spots the LEED pattern still exhibits 3-fold symmetry. It should also be noted that all of the LEED patterns observed in the present study were recorded at a sample temperature of ≈130 K and were sensitive to electron beam damage becoming visually more diffuse upon prolonged exposure. This can be ascribed to the same type of electron-stimulated desorption process observed in AES measurements. (IV) Discussion The results outlined in section III allow us to develop a detailed picture of the growth and structure of MgCl2 (15) Hubbard, A. T. Surface Imaging and Visualization; CRC Press: New York, 1995.

MgCl2 Films Grown on Pd(111)

Langmuir, Vol. 13, No. 7, 1997 2093

Figure 4. (a) Pd(111)-(4×4)-MgCl2 LEED pattern obtained by deposition ≈1 ML MgCl2, as judged by AES, (b) Pd(111)-(1×1) LEED pattern of the clean Pd(111) surface after thermal desorption of the MgCl2 overlayer recorded at the same position as (a). Both LEED patterns were recorded at an electron beam energy of 130 eV and a sample temperature of ≈120 K.

Figure 5. Pd(111)-(x13×x13)R13.9°-MgCl2 LEED pattern obtained at 1 ML MgCl2, as judged by AES, recorded at an electron beam energy of (a) 49 eV and (b) 104 eV with a sample temperature of ≈120 K.

thin films grown on Pd(111). As a first step it is necessary to understand the adsorption/desorption characteristics of this system represented by the AES and TPD measurements. These results support the idea that the use of the source leads to the production of MgCl2 thin films at both multilayer and monolayer coverages. This is confirmed by AES measurements which reveal that following MgCl2 thermal desorption the surface contains no trace of either Cl or Mg. TPD and AES results also reveal that some measurable interaction exists between the underlying Pd(111) substrate and the MgCl2 overlayer, as evidenced by the existence of a distinct, slightly more thermally stable, monolayer phase.

Control of the MgCl2 growth rate can be affected by adjusting either the source flux or the surface temperature. However, in practice it is more convenient to control the growth rate by employing a fixed source temperature with a variable surface temperature. With this in mind, a knowledge of the thermal chemistry is necessary to permit accurate control of the overall growth rate. Thus from Figure 3a, it is clear that MgCl2 deposition at room temperature provides no discrimination between the rate of adsorption in either monolayer or multilayer states resulting in the rapid production of thick MgCl2 films. However, control of multilayer thickness is possible by raising the surface temperature to ≈640 K. This is a result

2094 Langmuir, Vol. 13, No. 7, 1997

Fairbrother et al.

Figure 6. (1×1) LEED patterns recorded for (a) Pd(111)-a 12 Å multilayer MgCl2 film and (b) the underlying Pd(111) surface after thermal desorption of the MgCl2 overlayer recorded at the same position as (a). Both LEED patterns were recorded at an electron beam energy of 100 eV and a sample temperature of ≈120 K.

of the fact that at this elevated temperature there is a measurable difference between the sticking probability in monolayer and multilayer states allowing for the initial population of a full monolayer before subsequent multilayer formation. Additional studies indicate that monolayer deposition of MgCl2 at room temperature, as measured by AES, produces a disordered overlayer as evidenced by the lack of any ordered LEED pattern. However, after annealing this surface to ≈570 K the (x13×x13)R13.9° LEED pattern is observed upon recooling the surface to room temperature. This result clearly shows that in addition to the difficulty in controlling the deposition rate there is insufficient thermal energy at room temperature to order the MgCl2 layer by surface diffusion of the ionic species. The (4×4) surface structure shown in Figure 4a at ≈1 ML indicates that the MgCl2 overlayer is in registry with the Pd(111) surface such that there is a coincidence point between the two lattices equal to four times the Pd-Pd lattice spacing (aPd). Furthermore, from the LEED symmetry, the unit cell of the overlayer must lie parallel to that of the hexagonal Pd(111) surface. This leads us to propose the structure shown in Figure 7, shown in terms of both a top and side view, which is based upon the unit cell of either R or β bulk crystalline MgCl216,17 exposed along either the (0001) or (001) orientation respectively. Since

aPd ) 2.75 Å and aMgCl2 ) 3.64 Å in either R or β phases the structure shown in Figure 7 is consistent with the fact that

4aPd ) 3aMgCl2

(4)

to within 0.7%. This leads to an overlayer coincidence at (16) Ferrari, A.; Braibanti, A.; Bigliardi, G. Acta. Crystallogr. 1963, 16, 846. (17) Bassi, I. W.; Polato, F.; Calcaterra, J. C.; Bart, J. C. Z. Kristallog. 1982, 159, 297.

Figure 7. Schematic representation of the structure corresponding to the (4×4) LEED pattern seen from (a) top and (b) side views. The Pd(111) surface atoms, first layer Cl- and Mg2+ ions are represented by hollow, filled, and gray circles, respectively. In the top view only the registry of the first layer chloride and magnesium ions with respect to the substrate are shown. The structure has been chosen to maximize the number of Pd-Cl ontop interactions. Also shown in the top view is the base of the hexagonal MgCl2 unit cell, which is fully represented in the side view. In this structure first and second layer chloride ions are shown as solid and lightly hatched circles and Mg2+ is represented by a gray circle. Note that both the Pd-Pd and Cl-Cl unit cell dimensions are also shown.

every fourth palladium atom, giving rise to the observed (4 × 4) structure. It should be noted that in this and all subsequent LEED patterns the exact position of the overlayer with respect to the surface atoms has been chosen arbitrarily in the absence of detailed I-V analysis. The proposed structure is also consistent with the estimated overlayer thickness of ≈6 Å, which is in good qualitative agreement with that of the MgCl2 unit cell (≈5.9 Å).16 AES results indicate that the (x13×x13)R13.9° full monolayer has a similar overall MgCl2 coverage to that of the (4×4) structure. Taken in conjunction with the required substrate registry (Figure 8a), this result leads us to propose the structure shown in Figure 8b, which can be derived by a 13.9° rotation of the (4×4) structure with a contraction of 0.36 Å in the MgCl2 unit cell. On the basis of the relative sizes of the unit cells shown in Figures 7 and 8b and

MgCl2 Films Grown on Pd(111)

Langmuir, Vol. 13, No. 7, 1997 2095

Figure 8. (a) Unit cell corresponding to the (x13×x13)R13.9° LEED pattern. (b) Proposed structure corresponding to the Pd(111)-(x13×x13)R13.9°-MgCl2 LEED pattern showing the Pd(111) surface atoms along with the first layer Cl- ions. Note that both the Pd-Pd and Cl-Cl unit cell dimensions are also shown.

[MgCl2](4×4) ≈ 0.8[MgCl2](x13×x13)R13.9°

(5)

which explains the observed similarity in the relative MgCl2 coverages. The idea that the monolayer structure corresponding to the Pd(111)-(x13×x13)R13.9°-MgCl2 pattern is more densely packed than the (4×4) structure is also supported by recent experiments of MgCl2 deposition on Pt(111).18 On this surface an identical set of LEED patterns to those found on Pd(111) are observed. In this case careful annealing of the Pt(111)-(x13×x13)R13.9°MgCl2 pattern, at a temperature sufficient to induce MgCl2 desorption, is found to produce the Pt(111)-(4×4)-MgCl2 pattern, indicating that the (x13×x13) structure is more compressed than that associated with the (4×4) LEED pattern. Further MgCl2 deposition, as judged by AES, results in a simple (1 × 1) LEED pattern. On the basis of a comparison of the relative size of the unit cell to that of the underlying Pd(111) surface, this corresponds to a hexagonal unit cell with a lattice parameter of ≈3.67 Å. This is consistent with an MgCl2 film (aMgCl2 ) 3.64 Å)16 whose structure is identical to that of the (4×4). The simplification in the LEED pattern results from the increase in film thickness. This removes any contribution from the underlying surface lattice because of the extreme surface sensitivity of LEED coupled with the considerable thickness of each individual MgCl2 layer (≈5.9 Å).17 Indeed, using the interlayer spacing, the estimated thickness of the films grown in this study (12 and 24 Å) can be seen to correspond to two and four MgCl2 triplelayers (chlorine/magnesium/chlorine), respectively. The observation of ordered commensurate LEED patterns in contrast to incommensurate overlayers provides structural evidence for the presence of significant substrate-adsorbate interactions leading to a preferred registry. Similar results have been obtained for the orientation of thin oxide films grown on single crystal metal substrates, such as MgO grown on both Mo(100)19 and Fe(001).20 The existence of a preferred morphology in this present study is also consistent with TPD/AES data presented in section III, which indicates the existence of a distinct monolayer. In addition the observed registry implies that the substrate-adsorbate interactions, presumably dominated by that between Pd-Cl (see Figure 7), are sensitive to the position of the overlayer with respect to that of the substrate. For example, Figure 7 represents (18) Fairbrother, D. H.; Roberts, J. G.; Somorjai, G. A. Manuscript in preparation. (19) Wu, M.-C.; Corneille, J. A.; Estrada, C. A.; He, J.-W.; Goodman, D. W. Chem. Phys. Lett. 1991, 182, 472. (20) Park, Y.; Fullerton, E. E.; Bader, S. D. J. Vac. Sci. Technol. 1995, A13, 301.

the structure in which the number of Pd-Cl on-top interactions are maximized. At slightly higher MgCl2 coverages, the strength of these adsorbate-substrate interactions are responsible for the formation of the (x13×x13)R13.9° structure, thus maintaining a clear registry with the surface atoms, despite the necessary contraction of 0.36 Å in the MgCl2 unit cell. This lowering of the total free energy is consistent with the fact that the (x13×x13)R13.9° pattern represents the full monolayer structure. Upon moving from monolayer to multilayer coverages, analysis of the LEED pattern reveals that the surface structure reverts back to that of the ionic solid in which the crystallographic orientation remains parallel to that of the Pd(111) substrate (see Figure 6b). LEED measurements carried out for the ≈12 Å thick MgCl2 overlayer (2 unit cells high) indicate that the MgCl2(001)-(1×1) surface structure remains invariant for incident electron energies as high as 250 eV. Since the inelastic mean free path at 250 eV is ≈8 Å, the lack of any observable contribution from the (x13×x13)R13.9° implies that completion of the second layer lifts the monolayer structure and replaces it with a bilayer version of Figure 7. Another possibility that must be considered is that Figure 6a represents a simple extension of the (x13×x13)R13.9° monolayer minus any contribution to the observed LEED pattern from the substrate. However, this structure would result in an observable 13.9° rotation of the lattice vector with respect to that of the surface. This proposed structure is clearly in contrast to Figure 6a, which shows that the overlayer lattice vector remains parallel to that of the Pd(111) substrate. In contrast to the monolayer, the second layer is not expected to experience any significant interaction with the Pd(111) surface. Consequently this layer reverts to that of the crystalline unit cell,16,17 as shown in Figure 7. Furthermore, the strength of the interlayer Cl-Cl interactions appears to be greater than the energetic difference between the (x13×x13)R13.9° and (4×4) structures. This causes the monolayer structure to be lifted and replaced by the (4×4) structure (Figure 7), thus allowing interlayer interactions and packing to be maximized according to the bulk MgCl2 structure. This explains why the multilayer lattice vector remains parallel to that of the surface, because in this ordered multilayer phase, the underlying template is the (4×4) structure. The identical overlayer patterns observed for films of 12 and 24 Å thicknesses clearly indicate that once established, this growth mode extends beyond the second monolayer. The growth characteristics of this system are consistent with layer-by-layer rather than three-dimensional island growth. This is readily explained on the basis of the strong ionic bonding parallel to the surface plane and the much weaker dispersive forces between adjacent MgCl2 layers along the surface normal. The observation of 3-fold symmetry is indicative of packing found in either cubic R or hexagonal β phases of bulk MgCl2 along the (0001) or (001) orientations, respectively.6,16,17 Indeed a detailed structural analysis will be the subject of a future publication.21 Furthermore, the observed symmetry provides additional support for the idea that, in contrast to the monolayer, the stable multilayer phase possesses a structure analogous to that of bulk MgCl2. (V) Conclusions The structure of MgCl2 overlayers, at both monolayer and multilayer (thin film) coverages, grown on Pd(111) (21) Roberts, J. G.; Gireir, M.; Fairbrother, D. H.; VanHove, M.; Somorjai, G. A.; To be submitted for publication in Surf. Sci.

2096 Langmuir, Vol. 13, No. 7, 1997

illustrates the interplay between adsorbate-substrate and adsorbate-adsorbate interactions. The existence of significant adsorbate-substrate interactions at low coverages (