Halogen adlayers on silver (111) - The Journal of Physical Chemistry

Robert W. Owens and D. Alastair Smith. Langmuir 2000 16 (2), 562- ... Keith J. Stevenson, David W. Hatchett, and Henry S. White. Langmuir 1996 12 (2),...
0 downloads 0 Views 3MB Size
J. Phys. Chem. 1994, 98, 291-296

291

Halogen Adlayers on Ag( 11 1) Joachim Hossick Schott Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455

Henry S. White' Department of Chemistry, University of Utah, Salt Lake City, Utah 84112 Received: August 24, 1993; In Final Form: October 10, 1993e Atomically smooth halogen adlayers on Ag( 111) (AgX, where X = I, Br, C1, and F) have been prepared by quenching of heated Ag samples in concentrated acid halide (HX) solutions. The AgX adlayers are chemically stable in air for several hours, allowing detailed characterization of the adlayer atomic structures by scanning tunneling microscopy (STM). STM images of I adlayers are consistent with a nearly perfect 4 3 X 4 3 / R 3 0 structure; F, C1, and Br adlayers, however, are characterized by a parallel double row structure of adsorbed halogen atoms, oriented along the [ 1101 direction on the Ag( 111) substrate. The row structure observed by STM for C1 and Br adlayers is shown to be consistent with low-energy electron diffraction (LEED) patterns, but inconsistent with the corresponding structures derived from LEED. The degree of symmetry in the various adlayers decreases with increasing adatom electronegativity, indicating that the local charge distribution around the adlayer atom and its size are the controlling factors in determining the adlayer structure.

Introduction The nucleation and growth of halogens on silver surfaces are of fundamental interest in surface science,l-9 interfacial electrochemistry,lOJl and photography.12 Due to the electronegativity of the halogens, significant charge transfer occurs between the Ag surface and adsorbed halogen atoms, resulting in a quasiionic metal halogen bond.13 Theoretical data, however, suggest that the physical nature of the halogen adatom cannot be understood in the realm of a simple point-chargemodel but rather represents what has been termed "a polarized negative ion" by Lang and Williams.*3 Existing structural models for halogen monolayers on Ag( 11 1 ) have been deduced from surface extended X-ray absorption fine structure (SEXAFS) and low-energy electron diffraction (LEED) measurement^.^-^ LEED analysis of C1, Br, and I monolayers on Ag( 111) suggests that these halogens form a d3 X .\/3/R30 s t r u ~ t u r e l , ~at~ submonolayer ~~~J coverages. The LEED pattern, however, is weak and diffuse3*637for Cl and Br, and toour knowledge, no LEED data are published for F adlayers on Ag(l11). SEXAFS data2 seem to support the d3 X d3/ R30 structure proposed for I adlayers.1 However, a later LEED investigationsrevealed that I adlayersare somewhat morecomplex than a simple 4 3 X .\/3/R30 structure. A 'honeycomb vacancy structure" has been proposed on the basis of SEXAFS data43to resolve uncertainties in the structure of the CI adlayer derived from LEED studies3~6in the submonolayer regime. In addition, epitaxial growth of a silver halide layer on top of the Ag( 1 11) substrate has been discussed for CI, Br, and I adlayer~.3.**~ In summary, there is still considerable uncertainty concerning the structures of adsorbed halogens on Ag( 11l), despite the fundamental and technological interest in these materials. In the present article, we report the first chemical synthesis of atomically smooth halogen adlayers on Ag. The adlayers are chemically stable in air over a period of several hours, allowing the systematic investigation of the structures of F, C1, Br, and I adlayers on Ag using a scanning tunneling microscope (STM). To our knowledge, atomically resolved, real-space images of adsorbate-coveraged Ag surfaces have not been previously obtained by STM or related scanned-probe. techniques. The structures of C1, Br, and I adlayers obtained by STM are shown .Abstract published in Advance ACS Abstracts, December 1, 1993.

0022-3654/94/2098-0291S04.50/0

to be consistent with previous LEED and SEXAFS data; however, for both Brand C1 adlayers on Ag, STM reveals new atomic-level structural features which are not included in previous structural models deduced from either LEED or SEXAFS.

Experimental Section Our method of preparing Ag surfaces is'adapted from the method developed by Zurawski et al.,14 who adsorbed iodine on Pt samples and demonstrated that the halogen adlayer effectively protects the sample from surface contaminants. Spherical Ag samples (1-2 mm in diameter) were prepared by melting one end of a -2-cm piece of Ag wire (99.9985% purity, 0.5-"diameter, Aesar/Johnson Mathey) in a Hz/02 flame. The spheres were further annealed in a H2 flame at -1000 K for -60 s, cooled for -20 s in air, and transferred into one of the following solutions: hydrofluoric acid (38%, Baker) to form F adlayers; hydrochloric acid (38%, Merck) to form C1adlayers; hydrobromic acid (48%, Mallinckrodt) to form Br adlayers; and hydroiodic acid (1-5%, Merck) or 0.01 M solutions of IZ in methanol to form I adlayers. After immersion of the Ag sample for 30-300 s, the samples were rinsed with triply distilled water, dried, and mounted in a home built specimen holder for STM studies. (Because of the high solubility of AgF in water, rinsing of these samples was usually omitted or very brief.) Surface coverages (0) of C1, Br, and I on Ag were estimated using Auger electron spectroscopy (PHI 595 SAM, cylindrical mirror analyzer). The halogen Auger peak heights were measured relative to the Ag Auger peak after correcting for the empirically derived Auger sensitivities. For CI adlayers, the Cl-to-Ag peak ratios varied from 0.3 to 0.6. Since the spot diameter of the electron beam impinging on the surface in the Auger experiments was about 50-100 pm, these ratios represent average coverages; the local coverageobtained from STM images varied considerably (see below). For Ag exposed to methanol/IZ or concentrated HI solutions, the surface coverage of I, 81, ranges from 0.3 to 1.0. Since a full first layer of I on Ag( 11 1) corresponds to 01 0.31, it appears that multilayers of I are readily formed, even for relatively short exposure times (60 s). Our measured coverages of adsorbed Br are less accurate, since the Auger sensitivity for Br is extremely low. However, we estimate OBr to be 0.5, independentof the immersion time for t > 30 s. F adlayers could not be detected by Auger spectroscopic analysis, most likely

-

Q 1994 American Chemical Society

292

The Journal of Physical Chemistry, Vol. 98, No. I , 1994

Hossick Schott and White

0

200

400

600

800

1000

800

1000

Energy I eV

Figure 1. SEM image of a chlorine-covered Ag sphere. Insert: a facet terminated by a -60' boundary.

because AgF decomposes into Ag and F upon exposure to ultraviolet light with photon energies in excess of -4 eV,15J6 which is generated when the Auger electron beam impinges on the sample surface. However, no discernible oxygen or carbon peaks were observed for samples immersed in the HF solutions, indirectly indicating the formation of a protective F adlayer. Our experiments were performed using a Nanoscope I1 STM," in which the tip is held at virtual ground and the bias voltage Vb is applied to the sample. Experiments were performed in air using mechanically cut Pt(70%)Ir(30%) tips. All images were recorded in the constant current mode using a scan rate of 8.6 Hz and consist of 400 X 400 data points. Atomically resolved STM images of pyrolytic graphite and reconstructed Au( 1 1 1) surfaces indicate that the lateral resolution of our instrument is typically better than h0.3 A. On the basis of reproducibility tests for the Ag structures reported below, we estimate that the interatomic distances reported here are accurate to within at least A0.3 A. Results and Discussion Chlorine/Ag( 11 1). Faceted areas are readily observed on all

halogen-covered Ag samples when viewed under a low magnification optical microscope. Scanning electron micrographs of a chlorinated Ag sample are shown in Figure 1. The boundaries of the faceted regions often form -60° angles, as shown in the insert of Figure 1, indicating that the individual facets expose a crystalline Ag( 111) surface. X-ray diffraction measurements are inconclusive because the focus diameter of the X-ray beam is of the order of the sample diameter. However, the spot arrangement of the Laue backscattering pattern does indicate that the surface orientation is predominantly (1 11). The formation of halogen adlayers, as well as surface contamination, was monitored by Auger electron spectroscopy. As shown in Figure 2a, no measurable oxygen or carbon signal is observed in the Auger electron spectrum of a CI-covered Ag surface. Similar spectroscopic analysis of Ag samples following immersion in HI, HCI, or HBr solution also did not yield any detectable oxygen or carbon signal, even after exposure of the prepared sample to ambient air for 2 h. Ag samples which were not immersed in a halogen solution but which were otherwise prepared in an identical fashion and allowed to cool in air yielded a significant oxygen Auger signal (see Figure 2b). These results indicate that the halogen adlayer effectively prevents the adsorption of expected contaminantssuch as oxygen. Our results are consistent with UHV studies of Kiskinova et a1.I8 and theoretical arguments of Lang et al.19 that indicate that electronegative adsorbates, such as C1, prevent the coadsorption of oxygen in UHV environments. However, after -3 h exposure to air, we were able to detect an oxygen signal on some halogencoated Ag samples. Therefore, all STM studies were performed within 2 h of sample preparation.

0

200

400

600

Energy I eV Figure 2. Auger electron spectra of Ag samples: (a) CI-coated Ag; (b) Ag sample, cooled in air. Note the absence of the oxygen peak at -550 eV for the CI-coated sample.

Bao et al. have recently described a method of preparing Ag spheresusing electron-beamheating in vacuum.2o Similar to our results obtained using a H2-rich flame-annealingtreatment, the Ag surfaces prepared by Bao et al. also have a predominantly (1 1 1) surface orientation. This latter study also demonstrated that prolonged exposure of the Ag spheres to pure oxygen at high temperature results in the formationof a Ag oxide surface phase. As evident from the Auger electron spectroscopic results, Figure 2, flame annealing of a Ag sphere in a reducing atmosphere, followed by immersing the sphere in a concentrated acid halide solution, prevents the adsorption of the oxygen. STM images taken within the faceted regions of all halogen coated Ag samples show atomically flat terraces. The image in Figure 3a, obtained in a faceted area, shows a terrace of -50 nm width, but larger terraces (up to 100-nm width) are found. The terraces are separated by 3.0-&high monoatomic steps which contain a high density of kinks (Figure 3b). The kinks seem to be related to a row structure of adatoms on the terrace (see below). These features are readily apparent in the images of C1-coated samples shown in Figure 3a,b. Atomically resolved STM images of the C1adlayer, as well as for all other halogens (see below), can be reproducibly obtained from sample to sample using different PtIr tips and are representative of large regions of the faceted regions of the Ag surfaces. However, atomically resolved images of Ag spheres cooled in air cannot be obtained,presumably due to the formation of a disordered oxide phase or the presence of adsorbed contaminants. The STM image shown in Figure 4a was obtained in the center of one of the terraces of a CI-covered surface. The surface is characterized by a double row structure of C1 atoms. The double-row arrangement, however, is not perfect: closer inspection of the image in Figure 4a reveals that half of the atomic rows are not perfectly straight but rather appear to have a quasisinusoidal shape with a periodicity of 3-4 nm. Consequently, regions in which the atoms are arranged in a straight double-row arrangement are separated by regions which appear to have a hexagonal packing pattern, hereafter referred to as transition regions. Detailed structures of the double row and transition regions are apparent in higher resolution STM images. For

-

Halogen Adlayers on Ag( 11 1)

The Journal of Physical Chemistry, Vol. 98, No. I , 1994 293

a) 10 x IO nm2

a) SO--x SO nm2

b) 4 x 4 nm2

b) 10 x 10 run2

Figure 3. (a) Unfiltered STM image of a C1-coated sample, showing terraces separated by monoatomic steps. (vb = -850 mV, it = 2.3 nA). (b) Loss-pass filtered STM imageof kinksat thestepsseparatingdifferent terraces ( v b = -1005 mV, it= 3 nA). Maximum contrast corresponds to a height difference of 0.6 nm. The average height difference at the step is -0.3 nm.

example, the image in Figure 4b was recorded in a double-row region of the area shown in Figure 4a, whereas the image in Figure 4c was recorded in a transition region. Note that three unit cellscoexist in the transition region (seecell outlines in Figure 4c). The ratio of the surface densityof C1 atoms to the atom density in a perfectly terminated Ag( 111)-surface yields an absolute coverage of & I 0.6, as can be derived from the images shown in Figure 4b,c. We assume that &I 0.6 represents the saturation coverage of C1 on Ag( 111) since no higher coverages were observed during numerous experiments. The three nearest-neighbor CI-C1 distances in the double-row region, measured from the STM image in Figure 5, are 4.1,3.3, and 3.8 A. Comparison of the measured CI-overlayer interatomic distances with the distances between Ag atoms on the ideal Ag(1 11) surface (2.88 A) indicates that the C1 adlayer is incommensurate with the underlying substrate lattice. The geometric relationship of the C1 adlayer to the underlying (1 1 1) substrate is apparent from images of incomplete adlayers, an example of

-

Figure 4. (a) Unfiltered atomically-resolved STM image of the CI monolayer on Ag( 1 1 1). The double-row arrangement is faulted by row slip defects occurring with a periodicity of -2-3 nm (vb = -1085 mv, it = 4.3 nA). (b) Bandpass-filtered defect-free portion of the area shown 0.6 (vb= -995 mV, in Figure 3a; the adlayer unit cell is outlined. it = 4.6 nA). (c) Bandpass-filtered portion of the surface area shown in Figure 3a, featuring the row slip defect. Note that only every second row slips, whereas the other rows are unaffected. Three possible unit cells of the C1 adlayer are outlined. 0.6 (-1005 mV, it = 4.6 nA).

-

-

294

Hossick Schott and White

The Journal of Physical Chemistry, Vol. 98, No. I , 1994

1 1'

L -*Figure 5. Enlarged section (2.5

-

2.5 nm2) of the CI adlayer shown in Figure 4b. Inset indicates the nearest-neighbor distances ( a = 4.1, b = 3.8, c = 3.3 A). X

3.4 x 3.4 nm2

IF" Y

-

Figure 6. (a) Low-pass filtered image of an incomplete CI adlayer (ea 0.45). Thedarkspots represent the underlying Ag( 1 1 1) surface atoms. The hexagonal substrate unit cell is outlined. The bright spots represent C1 atoms ( v b = -680 mV, it = 5.5 nA).

-

which is shown in Figure 6 for 0.45. From this image, we tentatively conclude that the C1 adlayer grows in rows along the [ 1101 direction: the dark circular regions are assigned to the underlying Ag( 111) lattice sincethey form a hexagonal structure (see outline in Figure 6) with a periodicity of 3.1 A, close to the 2.88 atom spacing in the topmost layer of an ideally terminated Ag( 1 11) surface. The difference of -0.2 A may be due to a slight expansion of the Ag( 1 1 1) lattice under the influence of C1 chemisorption, consistent with the fact that the surface atom density in AgCl( 1 11) is significantly lower than on Ag( 1 1 1) (7.3 atoms/100~inAgCl(l1l)vs13.8atoms/100A2inAg(l 11)21). Thermal desorption experiments of Bowker and Waugh3 demonstrate that C1 does not desorb from the Ag( 1 11) surface in the form of Cl2 but always as AgCl molecules. This result suggests that the C1 adlayer and the topmost Ag atom layer form strongly bound complexes which resemble more closely AgCl molecules rather than a simple overlayer. Because of the polar nature of AgCl molecules, strong Coulombic interactions are expected to occur in both the C1 and the Ag layer. Therefore, it is reasonable to assume an expansion in the topmost layer of the Ag( 111)

I L < i 10>

Figure 7. Ball models of the C1 double-row structure on an ideally terminated (1 1 1) surface, (a) double row region. (b) transition region. Bold open circles, CI atoms; plain open circles, Ag atoms in the topmost layer; shaded circles, Ag atoms in the second layer.

substrate surface. The observation of thedouble row arrangement of C1atoms suggests that this expansion is likely to be anisotropic. We speculate that the numerous kinks in the steps separating the subsequent terraces (Figure 3b) are related to the lattice expansion. Modeling the observed adlayer structures on an ideal (1 1 1) substrate (Figure 7) shows that C1 atoms may occupy both quasi2-fold bridge and quasi-3-fold hollow sites in immediately adjacent rows (1 and 1' in Figure 7a). This assignment may explain why the adatoms in adjacent rows of the double-row region have a different brightness in STM images (cf. Figures 4 and 5 ) ; if the electron density varies at different sites of the metal surface,2* the degree of charge transfer and therefore the occupation of the halogen p-level may be expected to be different in different adsorption sites. In addition, the metal surface may become increasing depleted of electrons with increasing C1 coverage, resulting also in a lower occupation of the C1 p-level in some adsorptionsites. Further support for this speculationcomes from the thermal desorption (TD) experiments of Bowker and Waugh? in which two distinct desorption peaks can be discerned in their TD spectra at full monolayer coverage, indicating the presence of at least two different Cl-Ag complexes with different binding energies. Similar depolarization phenomena are observed in alkali-metal adsorption experiments on metal surfaces upon increasing the alkali-metal atom coverage (e.g., see Bonze123). However, as shown in the STM images at higher C1 coverage in Figures 4 and 5, the brightness of the C1 also varies within one particular row, with a periodicity of -2-3 nm. Therefore, we assume that theadsorption sites within one row changesgradually (i.e., with a periodicity of 2-3 nm) from 3-fold hollow to 2-fold bridge at full monolayer coverage. With this assumption, the observed row slip seen in Figure4c can be qualitatively understood (Figure 7b, rows 1 and 1'): whenever the adsorption site in row 1 changes, the adsorption site in the immediately adjacent row 1' also changes. As sketched in Figure 7b, the row slip may force one C1 atom into an electron-poor site. In STM, this atom may be expected to appear less bright than neighboring atoms. A comparison with the STM image in Figure 4c shows that the brightness of the atoms in a slipping row varies considerably. All STM images were obtained with a negative bias voltage applied to the sample, with electron tunneling occurring from the filled portion of the halogen plevel into empty tip-metal states. Therefore, both the occupation of the p-level and its

Halogen Adlayers on Ag( 1 1 1)

I

Unlt cells (real space) e,,

-

The Journal of Physical Chemistry, Vol. 98, No. 1, 1994 295 0.6

I

LEED

I

a) b) C) Figure 8. (a+) The three real space unit cells (solid lines) of the C1 adlayer that occur in the images shown in Figure 3 and (d) the corresponding LEED pattern constructed from geometric diffraction theory. The unit cell of the Ag( 111) is also shown (broken lines). The crosses represent real space atomic positions and diffraction spots from the Ag(l1 I ) substrate. The open circles represent real space positions and diffraction spots corresponding to the C1 adlayer. Note that the unit cell (c) represents the geometric average of the unit cells ( a s ) ; therefore, only the LEED pattern corresponding to cell (c) is constructed in Figure 5d (see text).

energetic position relative to the tip Fermi-level determine the brightness of the halogen adatoms in STM. Positive sample biases can not be used at all for any halogen-silver adsorption system because current instabilities, which result from a strong nonlinear resonance tunneling effect, do not allow the feedback loop of the instrument to stabilize. These effects are discussed in detail in a separate paper describingtunneling spectroscopicmeasurements of the Ag halogen systems.24 To compare our (real space) STM results with (reciprocal space) LEED patterns, we constructed the reciprocal lattices from our STM images (see, for example, the procedure described in ref 25). It is clear that the three different adlayer unit cells outlined in the image presented in Figure 4c, and schematically depicted in Figure 8, coexist in surface areas as small as 15 nm2. Since the dimensions of this area are clearly smaller than the transfer width of the electron beam in a typical LEED experiment,25 LEED will sample all three cells coherently. Therefore, at saturation coverage the LEED pattern will result from an average of the three patterns, which is equivalent to the unit cell corresopnding to the transition region, Figure 8c. The resulting LEED pattern constructed from this pattern is depicted in Figure 8d. Apart from a slight mismatch (see below), the LEED pattern observed by Bowker and Waugh3 at saturation coverage is essentiallyidentical to that constructed from the STM images. The mismatch arises because the two unit cell vectors in Figure 8c are of slightly different length (4.1 and 3.8 A). Bowker and Waugh3 interpreted their LEED pattern assuming a complete CI monolayer with a C1 atom density equal to that in AgCl( 11 1) (Le., 4.1-A Cl-Cl nearest-neighbor distance). This discrepancy can be readily explained as follows. Since we observe a row structure of C1 atoms along one of the three principal symmetry axis of the hexagonal Ag( 111) surface, we conclude that three domains of C1 atom rows may exist on extended Ag( 111) surface areas. One domain, however, may prevail throughout one facet. Since STM samplesa small surface area, the probability of obtaining atomically resolved images of surface regions which show two or three domains is extremely small. Wedid not observeanyinour study. The physicaldiameter of the electron beam in a LEED experiment, on the other hand, is of the order of 1 mm. Therefore, the actual LEED pattern consists of superimposed diffraction spots resulting from different domains (see, e.g., ref 25) which, in case of a (1 11) surface, are rotated by 120° with respect to each other. Considering this effect, the LEED pattern constructed from the unit cell in Figure 8c has to be rotated by two consecutive 120° rotations and the three resulting patterns have to be added so as to simulate the LEED dxperiment. It is clear then, that the slight mismatch of 0.3 A in one lattice direction will not be apparent in the measured LEED pattern. We conclude that the STM images in Figure 4 and the LEED pattern of Bowker and Waugh3 and Goddard and Lambert6 at saturation coverage are in excellent agreement. In summary, we emphasize that all observed LEED patterns for the Cl/Ag( 11 1) system can be well understood on the basis of our

-

-

STM images.26 However, the actual physical structure deduced from LEED analysis is inconsistent with the real-space, atomically resolved STM images. As a final comment regarding the C1 adlayer structure, it is interesting to consider the "honeycomb vacancy structure" for the C1 adlayer on Ag( 11l), proposed by Lamble et al.4~5based on LEED and SEXAFS measurements. This unusual structure, 0.75, has nearestwhich requires an absolute coverage of neighbor CI-CI distances of 2.9 and 5.0 A. In addition to these values, the transformed SEXAFS spectrum, which is presented in refs 4 and 5, and which gives the distribution of interatomic distances in the surface layer, shows additional strong peaks occurring at -3.2, 3.9, and 4.2 A (the peaks at 2.9 and 5.0 A which were used in the original structure analysis appear only as weak shoulders). We therefore suggest that the SEXAFS data should be reinterpreted with the peaks at 3.2, 3.9, and 4.2 A assigned to interatomic distances within the C1 adlayer, in good agreement with values obtained from STM for coverages (&I) ranging from 0.45 to 0.6 (3.3, 3.8, 4.1, and 6.2 A), Fluorine/Ag( 111). The double-row structure observed for F adlayers on Ag(l1 l), shown in Figure 9a, is very similar to that for C1. However, the interatomic distances reflect a compression of the atoms within each double row; the three nearest-neighbor distances are 2.8, 3.0, and 4.2 A. The saturation coverage corresponding to a completed double row structure, e F 0.7, is therefore slightly higher than that for CI. Assuming that the rows are oriented along the [ 1101 direction, modeling of the F adlayer structure on Ag( 111) suggests that the adsorption sites in the double-row arrangement alternate between 3-fold hollow and 2-fold bridge sites in adjacent rows, similar to the structure for the C1 adlayer. However, due to the shorter nearest-neighbor distances of the F adlayer, the ball model (not shown) for the F adlayer suggests that a transition region where a row slip occurs (asin thecladlayer) isnotnecessarytoaccommodatetheadatoms in rows on the 3-fold hollow and 2-fold bridge sites. This finding is consistent with the more uniform brightness of the atoms in the F adlayer, relative to that of the atoms in the CI adlayer. Bromine/Ag(lll). The monolayer structure for Br (Figure 9b) is best characterized as intermediate between the structure of the I monolayer (see below) and the double-row structures of F and CI. The absolute coverage, (3Bn for the saturated monolayer shown in Figure 9b is -0.45, the three nearest neighbor interatomic distances are 5.3, 5.1, and 4.1 A. Modeling the Br adlayer on the Ag(ll1) surface suggests that Br atoms, as in the case of F and C1, adsorb on at least two different surface sites, most likely 2-fold bridge and 3-fold hollow sites. Iodioe/Ag(lll). The interatomic distances of 4.9 and 4.6 A in the I monolayer (Figure 9c) suggest that I adsorbs in a slightly distorted 4 3 X d 3 / R 3 0 structure on Ag( 11l), similar to the 5 X 4 3 iodine adsorption pattern observed by Gao et al.27 on Au(1 11). However, the geometric relation of the I adlayer to the underlying substrate is not known from our experiments. Attempts to remove the I adlayer by using extremely low tunneling resistances, as recently demonstrated by Haiss et a1.28 for I on Au( 11l), failed, most likely because the binding energy of I on Ag is higher than on Au. The apparent slight compression of the adlayer lattice in one crystallographic direction may be related to the row structure growth observed for the smaller halogens. This compression also indicates, in accordance with earlier LEED studies,* that I is not restricted to just one adsorption site on Ag( 11l), but exhibits some degree of flexibility. Considering again that three different domains may coexist on extended surface areas, the anisotropy in the interatomic distances is expected to average out in LEED and SEXAFS experiments. However, if one of the domains reaches macroscopic dimensions, slightly different LEED patterns may beobtained in different experiments. This may explain the apparently contradicting findings in refs 1 and 8. The structure observed in our STM images and those

-

-

296

Hossick Schott and White

The Journal of Physical Chemistry, Vol. 98, No. I , 1994

a) 2.5 x 2.5 nm2

Conclusions

The flame-annealing and immersion method that we have described for synthesizing adlayers of F, C1, Br, and I on Ag(1 1 1) provides a means of obtaining well-ordered and atomically smooth Ag surfaces. Because Ag is a relatively reactive metal, it may be possible to extend the annealing and immersion method we have developed to the synthesis of well-ordered layers of other adsorbates(e.g., sulfides and thiols) on Ag. We have alsorecently reported that the method described herein can be used tosynthesize mixed halogen layers (e.g., a well-ordered adlayer of F and C1

atoms) and tostudy thecompetitiveadsorption of different halogen atoms using STM.29 The images of C1, Br, and I monolayers obtained using STM are entirely consistent with previous LEED and SEXAFS data. However, the adlayer structures deduced from reciprocal-space LEED and SEXAFS data for the C1 and Br adlayers are inconsistent with the real-space STM structures. Thus, we suggest that the LEED and SEXAFS data require reevaluation. Our STM images show that the relative degree of symmetry within the halogen monolayer structure systematically increases with decreasingadatom electronegativityand increasing adatom size. We conclude that the actual adsorbate layer structure depends on the local charge distribution within the metal-halogen complex and on the size of the halogen atom. Thus the atomic arrangements of the F, CI, Br, and I monolayers on Ag( 1 1 1) may reflect nonspherical lateral Coulombic forces between localized charges around neighboring adatoms.

b) 2.5 x 2.5 nm2 - I r

Acknowledgment. The authors thank D. 2. Liu for performing the Auger spectroscopic analyses and Prof. M. D. Ward for helpful discussions. This work was supported by the Office of Naval Research. J.H.S. gratefully acknowledges partial support from the Deutsche Forschungsgemeinschaft, Grant No. SCH 428/ 1-1.

References and Notes

C\

2.5 x 2.5 nm2

-

Figure 9. Atomically resolved STM images of halogen adlayers on Ag(1 11): (a) F adlayer, 0~ 0.7 ( v b = -1 135 mV, it= 2.2 nA); (b) Br adlayer, € 3 ~ 0.45 ~ (Vb = -550 mV, it = 5.7 nA); (c) 1 adlayer, €31 0.3 (Vb = -358 mV, it = 4.9 nA). All images were band-passed filtered

-

-

once.

obtained from earlier LEEDI**and SEXAFS2 measurementsare in good agreement. Wenote that the I-adlayer structureresulting from exposure of Ag to molecular iodine dissolved in methanol is identical to that obtained by exposure to concentrated .HI solutions.

(1) Forstman, F.; Berndt, W.; Biittner, P. Phys. Rev. Lett. 1973,30, 17-9. (2) Citrin, P. H.; Eisenberger, P.; Hewitt, R. C . Phys. Rev. Lett. 1978, 41,309-12. (3) Bowker, M.; Waugh, K. C. Surf. Sci. 1983,134,63964. (4) Lamble, G. M.; Brooks, R. S.; Ferrer, S.; King, D. A. Phys. Rev. B 1986,34,2975-8. (5) Lamble, G. M.;King, D. A. Phil. Trans. R. Soc. London A 1986, 31 8,203-7. (6) Goddard, P. J.; Lambert, R. M.Sur/. Sci. 1977,67, 18&94. (7) Goddard, P. J.; Schwaha, K.; Lambert, R. M. Surf. Sci. 1978,71, 351-63. (8) Maglietta, M.;Zanazzi, E.; Bardi, U.; Sondericker, D.; Jona, F.; Marcus, P. M.Surf. Sci. 1982,123, 1 41-5 1. (9) Bardi, U.; Rovida, G. Surf. Sci. 1983, 128,145-68. (IO) Valette,G.; Hame1in.A.; Parsons, P. 2. Phys. Chem. (Munich)1978, 113,71-89. (1 1) Bins, V. I.; Smith, C . K. Electrochim. Acta 1987, 32,25948. (12) Hamilton, J. F. Ado. Phys. 1988, 37, 359441. (13) Lang, N. D.;Williams, A. R. Phys. Rev. B 1978,18,61636. (14) Zurawski,D.;Rice, L.; Hourani, M.; Wieckowski,A. J. Electroanal. Chem. 1987,230,221-31. (15) Zmbov, K. F.; Margrave, J. L. J . Phys. Chem. 1967,71,4468. (16) Clements, R. M.;Barrow, R. F. Chem. Commun. 1968, I , 27-8. (17) Digital Instruments, Inc. (18) Kiskinova, M.;Goodman, D. W. Surf. Sci. 1981, 108, 6+76. (19) Lang, N. D.;Holloway, S.; Nsrskov, J. K. Surf. Sci. 1985, 150, 24-38. (20) Bao, X.;Barth, J. V.; Lehmpfuhl,G.;Schuster, R.;Uchida, Y .;Schlogl; Ertl, G. Surf. Sci. 1993,284, 14-22. (21) Wyckoff, R. W. G. In Crystal Structures, 2nd ed.; Interscience Publishers: New York, 1963;Vol. 1. (22) Smoluchowski, R. Phys. Rev. 1941,60,661-72. (23) Bonzel, H. P. Surf. Sci. Rep. 1987.8.43-125. (24) Hossick Schott, J.; White, H. S. J . Phys. Chem., following paper in this issue. (25) Ertl, G.; Kiippers, J. Low Energy Electrons andSurfuce Chemistry; VCH Verlagsgesellschaft mbH: Weinheim, Germany, 1985;pp 201-26. (26) Since STM is sensitive to the surface of a sample, the dissolution of C1 in the bulk of the Ag crystal, as suggested by Bowker and Waugh) cannot be ruled out on the basis of our results. (27) Gao, X.; Weaver, M. J. J. Am, Chem. Soc. 1992, 114,8544-51. (28) Haiss, W.; Sass, J. K.;Gao, X.; Weaver, M.J. Surf. Sci. Lett. 1992, 274, L593-8. (29) Hossick Schott, J.; White, H.S. Lungmuir, in press.