Adsorption of Mercaptobenzothiazole and Similar Molecules on

May 17, 2001 - Corrosion of cemented carbide grades in petrochemical slurries. Part I - Electrochemical adsorption of CN¯, SCN¯ and MBT: A study bas...
0 downloads 0 Views 136KB Size
5440

J. Phys. Chem. B 2001, 105, 5440-5449

Adsorption of Mercaptobenzothiazole and Similar Molecules on Cadmiumsulfide: A Theoretical Study Beate Flemmig, Ru1 diger Szargan, and Joachim Reinhold* UniVersita¨ t Leipzig, Wilhelm-Ostwald-Institut fu¨ r Physikalische und Theoretische Chemie, Leipzig, Germany ReceiVed: NoVember 30, 2000; In Final Form: March 23, 2001

The structure of the adsorbate systems formed by mercaptobenzothiazole (MBT) and analogue molecules on the CdS(101h0) surface is studied quantum-chemically using density functional theory. Preliminary calculations of the free adsorptive molecules indicate an energetic preference of their thione form compared to the thiol form. For the anions of the adsorptive molecules, the role of the endocyclic nitrogen and the exocyclic sulfur as possible donor atoms is examined by means of known chelate complexes. Clusters with 24 and 28 atoms that are saturated by point charges have been developed as surface models. Geometry optimizations show that the structure of the adsorbate systems is dominated by the formation of two coordinative bonds from the donor atoms of the adsorptive anions to two adjacent cadmium atoms of the surface. It results that the molecular plane of the adsorptives is tilted with respect to the normal of the crystal face. The calculated tilt angle for the MBT adsorbate agrees with angle-dependent XANES measurements, the only structural information presently available from experiment. It is found that the tilt angle changes with the variation of the heteroatom in the five-membered ring of the adsorptives. The molecule-surface interactions leading to these structural differences are analyzed. Further, the relaxation of the surface is included in the investigation. It becomes obvious that the direction of the relaxation of the free surface is reversed by the formation of the adsorbate bonds.

I. Introduction Although chemisorption is nowadays an intensively studied phenomenon, both from the experimental and from the theoretical side, it seems that the majority of theoretical investigations considers the adsorption of atoms or small molecules.1,2 However, adsorbates involving more complex compounds, like for instance organic molecules of medium size, are also very interesting from the technological and experimental point of view especially in the direction of the modification of properties of surfaces.3 The present study contributes to the relatively rare theoretical investigations in that field of interest. The adsorption of mercaptobenzothiazole (MBT), mercaptobenzooxazole (MBO), and mercaptobenzoimidazole (MBI) at sulfide surfaces has already been the subject of several experimental investigations such as XPS4-6 and X-ray absorption near edge structure (XANES)7,8 measurements. The underlying interest in these molecules and their adsorbates on mineral surfaces is their high potential for the development of more effective froth flotation processes.9,10 The adsorptives under study are prototypes of a promising class of new collector molecules that allow the flotation process to be carried out at neutral pH values and that are capable of selecting ores with more complex compositions. Flotation experiments with galenite, sphalerite, and pyrite indicate a high yield of metal sulfide with high selectivity, which could still be improved by hydrophobic substituents at the benzene ring.11 The spectroscopic experiments were carried out to get an insight into the structure of the adsorbate complexes that the * To whom correspondence should be addressed. Address: Universita¨t Leipzig, Wilhelm-Ostwald-Institut, Linne´strasse 2, D-04103 Leipzig, Germany. E-mail: [email protected]. Fax: ++49341-9736399.

molecules in question form at the mineral surface. XPS investigations of MBT on pyrite suggest that the molecule binds with one of its sulfur atoms, presumably the exocyclic one, to the surface.4 Angle-dependent XANES measurements at the N K edge of MBT on a cadmiumsulfide wafer reveal that the molecular plane of the adsorptive is tilted by 27 ( 5° with respect to the normal of the crystal face.8 From analogue measurements at the C K edge, the molecular plane follows a tilt angle of 25 ( 5°.7 This means that the molecules in the adsorbate adopt an upright position, i.e., not a flat-lying orientation, which is rather often found in adsorbates involving aromatic molecules.12 The complete structure of the adsorbate complex, however, and the knowledge of the specific interatomic interactions determining the formation of the adsorbate have not been available up to now. A theoretical investigation of such adsorbate systems has to include aspects of the solid surface and of the adsorptive molecules. The well established approaches in quantum chemistry, the cluster or the supercell model, can only account for either of the two aspects, whereas the other one has to be suitably approximated in the framework of the chosen model. For this study, we decided to stay in the molecular picture and to derive clusters of finite size to model the surface. This strategy is common in quantum chemistry and proved to be successful in many studies. Problems arise from the limited cluster size and the necessity to account for the interaction of the cluster with the surrounding lattice. For ionic systems, for instance oxides, relatively small clusters and extended point-charge arrays are widely used.13,14 However, even for the rather ionic TiO2, the alternative approach, the termination with pseudo-hydrogens, proved to be an adequate modelisation too.15 For moderately ionic solids possessing also a significant covalent character, a

10.1021/jp004333u CCC: $20.00 © 2001 American Chemical Society Published on Web 05/17/2001

Adsorption of Mercaptobenzothiazole

J. Phys. Chem. B, Vol. 105, No. 23, 2001 5441

TABLE 1: Calculated and Experimental Structural Parameters28-30 (Picometers and Degrees) of the Adsorptive Molecules in Their Different Chemical Forms MBT (X ) Sendo) C1-Sexo C1-X C1-N1 N1-C2 X-C3 C2-C3 N1-C1-X Sexo-C1-N1 C1-N1-C2 C1-X-C3

MBO (X ) O)

MBI (X ) N2)

exptl

thione

thiol

anion

exptl

thione

thiol

anion

exptl

thione

thiol

anion

166.2 173.2 135.3 138.0 174.0 138.7 109.2 127.4 116.4 92.3

165.3 178.3 136.8 139.0 176.6 140.7 107.8 126.4 118.1 92.2

176.8 178.1 129.2 139.0 175.9 141.6 116.6 124.8 110.7 87.9

169.5 187.1 131.7 137.1 175.0 142.9 111.0 130.3 113.8 89.6

164.3 136.9 135.3 138.0 138.3 139.0 107.5 129.7 110.5 108.6

164.5 137.6 137.1 139.2 137.9 139.7 106.6 128.8 111.0 108.7

175.7 137.1 129.3 140.0 138.1 140.1 116.7 128.9 103.6 103.4

169.1 143.9 132.8 137.7 135.6 141.3 111.3 131.5 106.3 105.2

167.1 136.2 136.2 138.2 138.2 140.0 106.4 126.8 110.4 110.4

166.6 137.7 137.7 139.1 139.1 141.0 104.5 127.7 111.8 111.8

176.9 137.8 130.8 139.2 138.9 141.6 113.9 125.9 104.5 106.5

171.5 140.9 134.2 137.5 137.5 143.0 109.8 130.2 106.3 109.2

somewhat larger cluster than for strongly ionic systems is usually required. Despite the specific problems concerning the cluster termination we regard the cluster approach appropriate especially for our investigations, where the local character of the adsobate is very pronounced because of the distinct molecular character of the adsoptives. We present in this paper a detailed density functional theory (DFT) study of the adsorbate systems formed by MBT, MBO, and MBI on the CdS(101h0) surface. This involves the free adsorptive molecules and, as a step toward the adsorbate, coordination compounds formed by the MBT anion with cadmium ions. The structures of the latter, which are also known experimentally, can be related to the calculated structures of the adsorbates. Suitable cluster models for the CdS surface are developed by a special point-charge saturation. Geometry optimizations yield the molecular and electronic structure of the adsorbate complex, considering the ideal as well as the relaxing surface. A detailed analysis of the reasons for the tilting of the adsorbed molecules is presented. II. Computational Details We employed DFT with the hybride functional B3LYP16,17 in this study. The 6-31G* basis set18 was used for the main group elements H, C, N, O, and S. For Cd, we applied the effective core potential ecp-28-mwb19 along with the corresponding valence basis set, thus, considering the 4s, 4p, 4d, and 5s electrons explicitly in the calculations. Test calculations were carried out for the free adsorptive molecules involving various methods (Hartree-Fock, MP2, DFT-B3LYP, and DFT-B3PW91) and basis sets (6-31G* and 6-31G+*). Diffuse functions have only negligible influence on the total energies and the structural parameters of the molecules, even in their anionic form. The optimized bond lengths for the MBT anion differ at most by 0.2 pm. All of the methods tested yield results of comparable quality in relation to the experimental data of the free adsorptive molecules. The DFT method has been selected with regard to a reliable and feasible treatment of the adsorbate complexes. The explicit form of the hybride functional seems to be of minor importance, as indicated by the very similar results we obtained with B3LYP and B3PW91. The calculations were performed with the Turbomole package by Ahlrichs and co-workers,20,21 which allows parallel treatment of the geometry optimization at the DFT level. The latter fact was vital for the feasibility of calculations of that type for the relatively large adsorbate systems with up to 1100 basis functions. They were performed on a parallel machine of the X-class, the HP SPP2000/X-48 at the Computing Centre of the University of Leipzig. For the graphic representations of the obtained structures, the Schakal 97 program22 was used. Frequency analyses were carried out for the completely optimized structures to characterize the stationary points. The

SCHEME 1

SCHEME 2

planar structures with Cs symmetry found for the free adsorptive molecules are all minima of the potential surface. The two mononuclear complexes are of C3 and C2 symmetry, respectively. The bonding between the MBT anion and a diatomic CdS unit is formed in the molecular plane. For the partially optimized adsorbate complexes, however, vibrational analyses are unreasonable because of the rigid surface cluster. Nevertheless, the adsorbate structures found can be regarded as minima on the corresponding potential surface because of their C1 symmetry. III. Results and Discussion 1. The Adsorptive Molecules. Three different chemical forms have to be considered for the adsorptives. First, they can undergo a tautomeric transformation from their generally preferred thione structure into the thiol structure (Scheme 1). The latter is often also referred to as the mercaptane structure, which is the origin of the above-mentioned widely used names and abbreviations for these molecules. To be consistent with the general usage, we keep these names in this study although terms referring to the other tautomeric form, like for instance benzothiazolethione for MBT, would better account for the spectroscopically confirmed preference23-25 of the thione form in the tautomeric equilibrium. As indicated by the pKa values for MBT (7.5),26 MBO (6.7),27 and MBI (4.3),28 the molecules can easily be deprotonated to their anionic form (Scheme 2). Subsequently, in the range of pH values around neutrality, there is always a certain amount of anionic species available, which is at least sufficient to cover the surface of the solid. Table 1 shows the results of the geometry optimizations for the considered adsorptive molecules in their three different chemical forms. We present those structural parameters that reveal information on the bonding properties of the systems. For the numbering of the atoms, see Scheme 3. The calculated values are compared with crystal structure data of the molecules MBT,29 MBO,30 and MBI,31 which crystallize all in the thione

5442 J. Phys. Chem. B, Vol. 105, No. 23, 2001 SCHEME 3

form. The theoretical values obtained from the optimizations agree well with the experimental data, i.e., the deviations are in the range of the experimental uncertainty. The only exception is the calculated C-Sendo bond length of the MBT molecule, which deviates around 5 pm from the experimental value. The analysis of the structures of the thione form of the molecules shows, first, that the bond lengths of the exocyclic bond have values around 165 pm, which are rather typical for a C-S double bond. Second, all bond lengths in the fivemembered ring of the thiones are fairly between the typical lengths of the corresponding single and double bonds. That means, the electronic delocalization is not confined to the benzene ring but is expanded over the whole molecule. Therefore, various mesomeric forms that consider each bond of the system as a single or double bond are relevant for the thiones. The transformation of the thione form into the thiol or mercaptane form has structural consequences, which are also reflected by the data given in Table 1. As one would expect from a glance at the valence bond formulas in Scheme 1, the C1-N1 bond in the heterocycle transforms into a bond with a length of approximately 130 pm, which is typical for a C-N double bond. Simultaneously, the C1-Sexo bond loses its doublebond character. Its length of more than 175 pm in the thiol form indicates that the exocyclic bond length is between a single and a double bond. Because of the distinct double-bond character of the C1-N1 bond in the thiol form, the delocalization of the π electrons is obviously in this tautomeric form not evenly expanded over the whole heterocycle. This is probably one reason for the preference of the tautomeric thione rather than the thiol forms of the molecules. In the calculations, this effect becomes apparent in the energetic stabilization of the thione in comparison to the thiol forms, which amounts to around 40 kJ/ mol for all three systems. Another effect contributing to the preference of the thione over the thiol forms follows from the spatial arrangement of the lone electron pair of the nitrogen atom N1. In the thione form, the lone-pair orbital is in the valence-bond picture an outof-plane π orbital, whereas it is an in-plane sp2 orbital in the thiol form. According to the valence shell electron pair repulsion (VSEPR) model,32 a smaller C1-N1-C2 angle should be expected for the latter. Indeed, in our calculations, this angle is reduced by about 7° in comparison with that of the thione form. We can conclude from this observation that the ring stress in the heterocycle of the thiol form is higher relative to that of the thione form. The two mesomeric structures for the anionic form depicted in Scheme 2 indicate that the negative charge of the system should be mainly distributed among the exocyclic sulfur and the nitrogen. The calculated atomic charges for the anions (e.g., MBT-: Sexo, -0.48; N, -0.42) are consistent with this expectation, which implies that these two atoms are both potential donor atoms. Concerning the influence of the variation X ) S, O, and NH on the structure of the heterocycle, we can infer from Table 1

Flemmig et al. that there is no remarkable difference between MBO and MBI but a rather strong one if we turn to MBT. This is independent of the respective chemical forms of the three systems. For MBO and MBI, with only first-row elements in the heterocycle, a nearly regular pentagon arises for the five-membered ring. For MBT, however, where the heterocycle contains the second-row element sulfur, both the C1-Sendo and the Sendo-C3 bonds are considerably longer than the other bonds in the ring. Consequently, the five-membered ring in MBT is a strongly irregular pentagon. This fact has a slight but remarkable influence on the structure of the adsorbate systems, which will be discussed in detail in section III.5. 2. Coordination of the Adsorptive Anions. We have pointed out in the previous section that the molecules in their anionic form possess two donor atoms for coordination, the exocyclic sulfur and the nitrogen. This permits various modes of coordination to metal centers. The X-ray structure analyses of corresponding complexes provide experimental evidence for this trend. One illustrative example is the polymeric complex [Cd(MBT)2]n, where the MBT anion acts partially as a chelating and partially as a bridging ligand.33 Other examples are the complex anion [Cd(MBT)3]- 34 and the similar neutral complex with two additional pyridine ligands (py), Cd(MBT)2(py)2,35 which are monomeric species with three and two, respectively, chelating MBT ligands coordinated to one metal center. We calculated the two latter species, which can be treated without the need of periodic boundary conditions or the search for an appropriate model of finite size as it would be the case for the polymeric compound. The monomeric complexes may serve as small analogues for the adsorbate complex where the direct interaction of the adsorptive anions with the cadmium ions can be studied. Figure 1 shows the calculated structures of the two complexes, and Table 2 provides selected structural parameters characterizing the coordination in the systems. There is a sufficient agreement between calculated and experimental values. In addition to the data presented in the table, we like to mention the angle describing the distortion of the trigonal prismatic coordination toward the octahedral coordination in the [Cd(MBT)3]- system. A value of 26° has been calculated in accordance with the experimental result (25.8°).34 The structural data show that the coordinative Cd-Sexo and Cd-N bonds have typical bond lengths around 270 and 245 pm, respectively. As a consequence of the bidentate coordination of the MBT ligand, a remarkably small bite angle Sexo-Cd-N of only 62° arises. This indicates a stress in the resulting chelate ring Cd-SexoC1-N. This conclusion is furthermore supported by the decrease of the Sexo-C1-N angle, which is about 6° smaller than in the free MBT anion (compare Table 1). The other internal structural parameters of the ligand are not or only slightly affected by the coordination. As a further approach toward the adsorbate complex, we consider the interaction of the MBT anion with the smallest imaginable representation of cadmiumsulfide, a two-atomic CdS moiety. Geometry optimizations for planar model systems with two different orientations of CdS relative to the MBT anion lead to the structures displayed in Figure 2. In orientation A, with CdS on the side of the nitrogen atom, cadmium interacts both with the exocyclic sulfur and the nitrogen. The respective bond lengths are 273 pm for Cd-Sexo and 254 pm for Cd-N, and the bond angle N-Cd-Sexo, the analogue to the bite angle, is 60.3° (see above). In orientation B, with CdS on the side of the endocyclic sulfur, cadmium interacts only with the exocyclic sulfur, forming a Cd-Sexo bond with the length 257 pm. For

Adsorption of Mercaptobenzothiazole

J. Phys. Chem. B, Vol. 105, No. 23, 2001 5443

Figure 2. Calculated structures of the model complex of MBT with a diatomic CdS unit: (a) orientation A, CdS on the side of the nitrogen and (b) orientation B, CdS on the side of the endocyclic sulfur.

Figure 1. Calculated structures of the two chelate complexes of cadmium with the MBT-anion as chelating ligand: (a) [Cd(MBT)3]and (b) Cd(MBT)2(py)2.

TABLE 2: Calculated and Experimental Structural Parameters32,33 (Picometers and Degrees) of the Chelate Complexes of MBT with Cadmium [Cd(MBT)3]Cd-Sexo Cd-N Sexo-Cd-N N-Cd-N′ Sexo-Cd-Sexo′ C1-Sexo C1-N C1-Sendo Sexo-C1-N Cd-Npy Npy-Cd-Npy′

Cd(MBT)2(py)2

exptl

calc

exptl

calc

266.7 247.4 62.0 95.8 102.4 169.3 130.3 174.6 124.0

270.6 254.6 61.3 90.7 108.7 171.5 131.9 179.4 123.5

270.6 242.2 62.1 85.5 162.9 170.3 131.7 174.6 123.3 233.4 95.1

271.6 244.9 62.2 95.2 165.2 172.7 131.9 177.8 122.5 246.2 89.1

the latter orientation, the C1-Sexo-Cd is in approximate accordance with the sp2 hybridization of the exocyclic sulfur. In orientation A, cadmium is obviously moved off this optimal position for an interaction with the exocyclic sulfur in order to achieve an additional interaction with nitrogen. The calculated total energies reveal that the structure with the CdS oriented on the side of the nitrogen atom is stabilized by 14 kJ/mol. We infer from these calculations that it will most likely be the cadmium atoms of the substrate that interact with the adsorptives in the adsorbate and that there is a certain preference to a bidentate coordination mode.

Figure 3. Two clusters Cd12S12 and Cd14S14 selected as surface models.

3. The Cluster Models. In general, cluster models for a surface that consist of a finite number of atoms should be large enough to comprise some of the essential characteristics of the solid, but should be small enough to make the calculations feasible. In our case, we have to consider additionally the spatial extension of the adsorptive molecules, which requires a sufficiently large surface layer of the cluster models. After series of test calculations for clusters, which were cut out of the (101h0) crystal surface of the wurtzite structure, consisting of a growing number of CdS units, we constructed the two cluster models Cd12S12 and Cd14S14 (see Figure 3). These clusters are uncharged because they consist of the same number of cadmium and sulfur ions. They contain a symmetry plane, are of complementary shape, and, in principle, are of the same size. These two surface models were chosen to have all imaginable adsorption sites represented with comparable accuracy. A representative feature of an ideal surface is an equal distribution of the atomic charges within the respective atomic

5444 J. Phys. Chem. B, Vol. 105, No. 23, 2001

Flemmig et al.

TABLE 3: Mulliken Charges for the Cluster Models Saturated with Point Charges (q ) (0.20) for Cd12S12 Compared to the Unsaturated Cluster Cd12S12 Cd1/2 Cd3/4 S1/2 S3/4 Cd5 Cd6/7 S5 S6/7 Cd8 Cd9/10 S8 S9/10 Cd11/12 S11/12

Cd14S14

unsat.

sat.

+0.51 +0.33 -0.46 -0.35 +0.56 +0.36 -0.48 -0.43 +0.35 +0.35 -0.42 -0.38 +0.39 -0.36

+0.50 +0.46 -0.47 -0.47 +0.57 +0.46 -0.48 -0.51 +0.44 +0.46 -0.51 -0.49 +0.50 -0.46

sat. Cd1/2 Cd3/4 S1/2 S3/4 Cd5 Cd6/7 S5 S6/7 Cd8/9 Cd10/11 S8/9 S10/11 Cd12 Cd13/14 S12 S13/14

+0.55 +0.53 -0.43 -0.45 +0.55 +0.50 -0.55 -0.55 +0.35 +0.41 -0.49 -0.55 +0.69 +0.63 -0.58 -0.56

SCHEME 4

layers. To achieve this, we saturated the clusters with positive and negative point charges. The positions of these point charges correspond to the positions of the neighboring atoms in the crystalline structure. We remark that in the cluster Cd12S12 not every neighboring atomic position is equipped with a point charge. The atoms at the border of this cluster have not more than one of their next atoms represented by a point charge (this results in 18 and 26 point charges for the cluster models Cd12S12 and Cd14S14, respectively). This pattern of saturation and a value of 0.20 for the point charges proved to be most suitable in extensive test calculations. They are not in every respect consistent with chemical intuition, this has to be attributed to the finite size of our cluster models. Table 3 shows the influence of the point charges on the atomic charge distribution (see Scheme 4 for the numbering of the atoms in the clusters). The use of point charges leads to a more uniform distribution of the atomic charges, especially in the central region of the clusters Cd12S12 and Cd14S14, i.e., for the atoms Cd1-Cd4 and S5-S7. We remark that test calculations adopting a cluster termination by pseudo-hydrogens did not lead to a uniform charge distribution in the central region of the clusters. Larger point charge arrays as well as further CdS layers seem to be not necessary in our case. This was checked by calculations with a cluster model including a second layer of point charges and a second layer of CdS units, respectively, which did not lead to a different charge distribution in the top layer (deviation < 0.01). In other studies dealing with sulfides, which are indeed rather rare, even completely unsaturated clusters are used. These clusters are of approximately the same cluster size as our models, for instance, in a study on PbS (two layers, 32 atoms)36 or in a study on MoS2 (two layers, about 80 atoms, but on a semiempirical level of theory).37

For surface models of finite size, geometry optimizations with all internal coordinates free to vary are unreasonable. We generally used the experimental value of 252.1 pm for the Cd-S bond length and a bond angle of 109.5° corresponding to an ideal tetrahedral coordination. To get an idea of the relaxation of the uncovered surface, we performed geometry optimizations for the cluster model Cd12S12 allowing the four central atoms Cd1, Cd2, S1, and S2 to relax relative to the rigid rest of the cluster. The resulting relaxation will be discussed in section III.6 in connection with the adsorption on a relaxing surface. 4. Structure of the Adsorbate on the Ideal Surface. The existence of several chelate complexes of the MBT anion with cadmium suggests that the adsorptive molecules interact in their anionic form with the metal atoms of the cadmiumsulfide surface. IR-spectroscopic investigations of MBT on copper indicate in analogy to our adsorbate the presence of MBT in the adsorbate in its anionic form.38 This assumption is also supported by results of XANES investigations.7 In the NKXANES spectrum of the neutral MBT molecule, a characteristic signal appears that could be assigned to the excitation from the nitrogen 1s level into the N-H σ-antibonding orbital.39 This signal is absent for both the MBT anion and the adsorbate of MBT on a cadmiumsulfide wafer, thus, indicating that MBT is adsorbed in its deprotonated form. Taking these observations into account, we performed geometry optimizations for model adsorbate complexes consisting of the adsorptive anions and our cluster models. We have to point out that we had, at this stage of the investigation, no reliable hints on the spatial orientation of the adsorptives on the surface and on the underlying interatomic interactions. To avoid the exclusion of any possible adsorbate structure and binding site, we had to consider start geometries that comprised all imaginable atomic adsorptive-surface interactions (e.g., N-Cd, S-Cd, and S-S) and various different orientations of the adsorptives on the surface. In C1 symmetry, i.e., without symmetry constraints, the geometry optimizations of the adsorptive anions on the rigid surface models lead, for all start geometries, to one and the same adsorbate structure, which is depicted in Figure 4 for the case of the MBT anion. For the other systems containing MBO and MBI, we obtain, in principle, the same result. The main feature of the adsorbate structure is the upright orientation of the adsorptive anion with the characteristic coordination of the exocyclic sulfur and the nitrogen to two adjacent cadmium atoms of the surface. In accordance with our findings, an upright position was obtained also for MBO on Pt(111) by CK NEXAFS investigations.40 A similar bidentate mode of coordination to two adjacent metal atoms of the surface has already been found in adsorbates of deprotonated carboxylic acids on oxides.2,41 We are convinced that the introduced adsorbate structure is the only one that represents a minimum on the potential energy surface. On the basis of the different start geometries for the optimizations, we can definitely exclude a flat-lying orientation of the planar adsorptive on the surface and a coordination of the adsorptive either by the exocyclic sulfur alone or by both the exocyclic and the endocyclic sulfur. The latter would correspond to the mode of coordination displayed in Figure 2b. Furthermore, a coordination by both donor atoms, the exocyclic sulfur and the nitrogen, to a single cadmium atom, possibly in a plane containing the Cd-S bonds of the surface like in Figure 2a, is to be excluded. Selected structural parameters of the three adsorbate systems considered are collected in Table 4. The bond lengths for the two main adsorptive-surface interactions Cd-Sexo and Cd-N

Adsorption of Mercaptobenzothiazole

J. Phys. Chem. B, Vol. 105, No. 23, 2001 5445 TABLE 4: Calculated Structural Parameters (Picometers and Degrees) of the Adsorbate Complexes of the MBT, MBO, and MBI Anions on the Cluster Cd12S12, for MBT in Comparison to the Cluster Cd14S14 MBT-/ Cd12S12 X ) Sendo

MBT-/ Cd14S14 X ) Sendo

MBO-/ Cd12S12 X)O

MBI-/ Cd12S12 X ) N2

269.3 244.8 108.2 94.3 172.3 180.6 132.7 134.4

262.9 243.7 101.1 87.9 172.7 179.6 132.8 134.9

268.4 240.0 107.6 94.3 171.6 139.4 133.3 135.9

267.2 237.6 109.0 94.4 173.8 138.6 134.6 134.9

Cd1-Sexo Cd2-N1 S1-Cd1-Sexo S2-Cd2-N C1-Sexo C1-X C1-N1 Sexo-C1-N

TABLE 5: Mulliken Charges for the Adsorbate Complexes of the MBT, MBO, and MBI Anions on the Cluster Cd12S12 in Comparison with the Values for the Free Anions MBTCd12S12 X ) Sendo Sexo X N1 H1 Σmol Cd1 Cd2 S1 S2

Figure 4. Calculated structure of the adsorbate complex of the MBT anion on the Cd12S12 surface model in three different views: (a) along the line of cadmium atoms, (b) perpendicular to the line of cadmium atoms, and (c) perpendicular to the surface.

of around 265 and 240 pm, respectively, indicate that the bonds in the adsorbate complexes have the same character as the classical coordinative bonds that are found in the chelate complexes. The formation of the relatively small bite angle in the mononuclear complexes, which should cause a considerable ring stress, is avoided in the adsorbate structure by the simultaneous coordination of the two donor atoms to two adjacent surface atoms. According to this coordination mode, we obtain in relation to the free adsorptive anions a small increase of the N-C1-Sexo angle by about 4°, which enlarges the distance between the two donor atoms. This increase is even

-0.14 +0.21 -0.46 +0.20 -0.36 +0.49 +0.58 -0.47 -0.49

X ) Sendo -0.46 +0.07 -0.42 +0.10 -1

MBOCd12S12 X)O -0.15 -0.46 -0.53 +0.19 -0.36 +0.46 +0.58 -0.47 -0.49

X)O -0.50 -0.48 -0.46 +0.09 -1

MBICd12S12 X ) N2 -0.22 -0.55 -0.56 +0.18 -0.34 +0.49 +0.58 -0.47 -0.48

X ) N2 -0.56 -0.55 -0.48 +0.09 -1

10° in relation to the chelate complexes. We want to point out that it is essential for this specific type of bonding between the adsorptives in question and the CdS surface that the N-Sexo distances match the Cd-Cd distance of the substrate. The close analogy of the bonding situation in the adsorbate systems and in the classical mononuclear chelate complexes suggests that the adsorption process under study is a typical chemisorption. In this context, we understand the adsorption energies in the range of -210 to -260 kJ/mol, which were obtained as the differences between the total energies of the adsorbate complexes and the respective free systems. We remark that these energy values are only sightly influenced if the surface model is extended by more point charges or by more CdS units. This was investigated for the MBT adsorption energy (218 kJ/ mol) by single-point calculations. They yield 225 kJ/mol for Cd12S12 with a second point-charge layer and 249 kJ/mol for Cd21S21, which includes a second CdS layer. In Table 5, the atomic charges of the adsorbate complexes are compared to those of the free adsorptive anions. This comparison shows that the adsorption is, in all three cases, associated with a transfer of about 65% of the negative charge of the adsorptive anion onto the surface. The most significant contribution to this charge transfer originates from the exocyclic sulfur atom of the respective adsorptive anion. The positive charges of the two cadmium atoms that are directly bound to the donor atoms are, however, only slightly diminished. We observe that the transferred charge is rather distributed over the whole surface model. 5. Tilted Orientation of the Molecules on the Surface. A closer look at the calculated adsorbate structures reveals that the adsorbed molecules do not adopt an exactly upright orientation and are not aligned exactly along the line of cadmium atoms. The first of these two characteristics is described by the so-called tilt angle. It is measured between the normal of the crystal face and the molecular plane as depicted in Figure 5. For the quantification of the second characteristic, the turn of

5446 J. Phys. Chem. B, Vol. 105, No. 23, 2001

Flemmig et al.

Figure 7. Turn of the molecular long axis of the molecules out of the line of cadmium atoms in the adsorbates for the adsorptives (a) MBT, (b) MBO, and (c) the model systems (X ) S and O).

SCHEME 5

Figure 5. Tilt angles of the adsorptives: (a) MBT, (b) MBO, and (c) the model system in the adsorbates.

Figure 6. Calculated structures of MBT and MBO and definition of their molecular long axis.

the molecules out of the line of cadmium atoms, we define a specific molecular long axis to lie in the molecular plane and to form a right angle with the C2-C3 bond. For MBT and MBO this axis is depicted in Figure 6. The angle between the projection of the molecular long axis onto the surface and the line of cadmium atoms describes the slight turn of the molecules out of the line of cadmium atoms (see Figure 7). In Figures 5 and 7, we recognize gradual differences for the tilt and turn angles in the adsorbate structures of MBT and MBO. MBI adopts an orientation that shows no remarkable

difference to the MBO adsorbate. Therefore, for MBI applies the same discussion as for MBO. The calculated tilt angle for MBT of 19° can be compared to the experimental values of 27 ( 5° and 25 ( 5°, which were obtained by angle-dependent XANES measurements.8,7 The correspondence between the calculated and the experimental tilt angle is remarkably good if one takes the approximations in our calculations into account. To understand the origin of the tilt and turn in general and to find an explanation for the respective differences between MBT and MBO, we derived smaller model systems for the adsorptive anions. Substitution of the benzene ring in the MBT and the MBO anion by two hydrogen atoms leads to the fivemembered heterocycles thiazolthiolate and oxazolthiolate that are unchanged with respect to the adsorbate-determining polar part of the original adsorptive anions (see Scheme 5). Consequently, adsorbate structures that are similar to those of the original adsorbates follow for the model systems. However, contrary to the original systems, the model adsorptives are not turned out of the line of cadmium atoms (Figure 7c). The orientation of the model systems on the surface is independent of the endocyclic heteroatom sulfur or oxygen. This is also in contrast to the original systems. The tilt angle for the two model systems lies between the tilt angles for MBT and MBO (Figure 5c). The tilt of the molecules in the adsorbate complex can generally be understood as the result of the trend of the surface atoms to continue the coordination realized in the bulk by forming chemical bonds to the adsorptive molecules. Therefore, the angles S1-Cd1-Sexo and S2-Cd2-N should both be close to the ideal tetrahedral angle. The model systems follow this trend rather closely. In the model adsorbates, the angles S1Cd1-Sexo and S2-Cd2-N are very similar, 104° and 101°, respectively, and indicate a distorted tetrahedral surrounding for the cadmium atoms, which is mainly due to the smaller Sexo-N distance relative to the Cd-Cd distance. For the model systems, there is obviously no driving force to turn out of the line of cadmium atoms. Such a turn would, indeed, be disadvantageous for the systems because the donor atoms would be moved off their optimal positions for the

Adsorption of Mercaptobenzothiazole

J. Phys. Chem. B, Vol. 105, No. 23, 2001 5447

Figure 9. Relaxation of the surface model Cd12S12.

Figure 8. Specific position of the benzene ring of the molecules relative to the surface: (a) MBT and (b) MBO.

coordination. We conclude that the model systems constitute an ideal situation for the most effective adsorptive-surface bonding interactions. This ideal orientation is independent of the type of the endocyclic heteroatom sulfur or oxygen. Consequently, the slightly different orientations of MBT and MBO on the surface must be related to the presence of the benzene ring. We attribute the turn of the molecules out of the line of cadmium atoms to a repulsive agostic interaction between the hydrogen of the benzene ring that is nearest to the surface (H1) and a particular cadmium atom of the surface (Cd4; see Figure 4). As a result of this repulsive interaction, the molecules turn out of the line of cadmium atoms toward the line of sulfur atoms, i.e., into the direction of a possible attractive interaction. By this means, the adsorptive molecules leave the position of the most effective interaction between both donor atoms and the surface. Although the S1-Cd1-Sexo angle is near to the tetrahedral value, a significant decrease results for the S2Cd2-N angle, indicating a stronger distorted coordination for Cd2 (see Table 4). From Figure 7, we infer that the turn angle for MBT is remarkably larger than for MBO. Obviously, the driving force for the turn of MBT is stronger than for MBO. We trace this back to the different proportions of the heterocycle (compare Figure 6). In the case of MBT, where the fivemembered ring is a highly irregular pentagon because of the relatively long Sendo-C1 and Sendo-C3 bonds, the molecular long axis forms a relatively small angle with the line of cadmium atoms (Figure 8a). For MBO, with an almost regular pentagonal heterocycle, a steeper orientation of this axis arises (Figure 8b), leading to an a priori larger distance between H1 and Cd4. We remark that the different amount of turning for MBT and MBO leads indeed to equal H1-Cd4 distances (316 pm). The difference between the tilt angles of MBT and MBO is caused by the different turning, which itself is the result of the

different proportions of the heterocycles. In the case of MBO, the loss of the ideal coordination position for the nitrogen donor atom by the slight turning may be partially compensated for by a conformational change around the exocyclic C-S bond, which provides a trigonal planar surrounding for the nitrogen and brings the molecule in a slightly more upright position (smaller tilt angle). This corresponds to the situation expressed by the mesomeric structure of the anion with the negative charge at the sulfur. Consequently, we see in Figure 5b the Cd2-N bond in the molecular plane and the Cd1-Sexo bond out of that plane. In the case of MBT, this way of compensation is obviously not favored. The Cd1-Sexo bond remains in the molecular plane. That causes the Cd2-N bond to lie relatively far outside the molecular plane because of the stronger turning (Figure 5a). This indicates a trend toward a change of the hybridization at the nitrogen atom from sp2 to sp3, which is described by the other mesomeric form of the anion with the negative charge at the nitrogen. The relatively large tilt angle found in the experiments for the MBT system is, after all, the result of the specific geometric proportions of the heterocycle and the presence of the benzene ring. If there is a forced turning, then both this turning and a stronger tilting will effectively reduce the repulsive agostic interaction between the benzene ring and the surface. We should stress at this point that the tilted orientation of the molecules under study on the surface does not follow from lateral interactions between the adsorbed species, unlike to other adsorbates.42,43 This becomes obvious from our calculations, where the tilted orientation is obtained for a model adsorbate of a single molecule on the CdS cluster. To verify this, we performed a calculation for an adsorbate complex with two MBT anions bound to two adjacent adsorption sites of the cluster Cd14S14. We obtained only a slight deviation of the adsorbate structure compared to the situation with a single admolecule. The absence of a significant intermolecular interaction is probably due to the relatively long distance of 672 pm between two adjacent lines of cadmium atoms, which is determined by the substrate structure. 6. Structure of the Adsorbate on the Relaxing Surface. Finally, we focus on the adsorbate structure that arises if the relaxation of the surface is included. We model the surface relaxation by a partial geometry optimization of the cluster Cd12S12 leaving its four central atoms free to relax. It results in a lowering of the surface cadmium atoms and a slight opposite movement of the sulfur atoms (Figure 9). In an analogue optimization for the considerably larger cluster model Cd21S21, we included, additionally to the four central atoms of the uppermost atomic layer, central atoms of the second and third atomic layer (Figure 10). The optimized structure reveals that the extend of the relaxation in the second and the third atomic layer is rather negligible. The principal feature of the surface relaxation, the lowering of the cadmium atoms, is obtained with both cluster models. It is in accordance with other calculations of the CdS(101h0) surface relaxation using the supercell approach44-46 as well as with the

5448 J. Phys. Chem. B, Vol. 105, No. 23, 2001

Flemmig et al. unrelaxed surface confirms that the orientation of the molecules is mainly caused by the influence of the surface atoms to which they directly bind. IV. Conclusions

Figure 10. Relaxation of the surface model Cd21S21.

Figure 11. Calculated structure of the adsorbate complex of the MBT anion on the relaxing surface model Cd12S12.

available experimental results for CdSe(101h0).47 This way of relaxation is consistent with the discussed general tendency of the electron-deficient three-fold coordinated surface atoms to establish a trigonal planar surrounding because of a preferred higher s character of their bonding orbitals. The opposite trend applies to the electron-rich three-fold coordinated atoms of the surface, which prefer a higher p character of their bonding hybride orbitals.48 In Figure 11, we show the structure of the adsorbate on the relaxing surface, which was again modeled by leaving the four central atoms of the surface model Cd12S12 free to relax (the cluster Cd21S21 is too large to perform the corresponding optimizations). To account for a situation where the adsorptive molecules approach the free relaxed surface, the outer atoms of the cluster were fixed to positions corresponding to those obtained for the relaxation of the free surface model. It turns out that the relaxation of the free surface is reversed under the influence of the adsorptive anion, i.e., the cadmium atoms are raised and the sulfur atoms are lowered. These findings confirm the relatively strong interaction between the adsorptive molecules and the surface, which is even sufficient to reverse the relaxation of the free surface. The orientation of MBT, MBO, and the model adsorptives on the relaxing surface is principally the same as that on the unrelaxed surface. Again, we obtain the largest tilt angle for MBT (21°), a small one for MBO (9°), and an intermediate value for the model systems (13°; compare Figure 5). For the other orientational angle, the turn of the molecules out of the line of cadmium atoms, we find, also like for the unrelaxed surface, the largest angle for MBT (14°), a smaller value for MBO (10°), and no turn for the model systems (compare Figure 7). These angles are measured with respect to the slanted plane of the four relaxing central atoms of the cluster. The close analogy of the orientational angles for the relaxed and the

It was the aim of this paper to achieve an understanding of the structure of a specific adsorbate complex of medium-sized organic molecules on the CdS(101h0) surface. It turns out that the respective anionic form of MBT and analogue molecules binds in the adsorbate with its two donor atoms to two adjacent cadmium centers of the solid. The molecular plane is characteristically tilted with respect to the normal of the crystal face. By means of this structural parameter, the calculated adsorbate structure can be related to the experimental results available for the adsorbate with MBT. We have shown that the tilt angle is mainly determined by the tendency of the surface cadmium atoms to continue the tetrahedral coordination from the bulk. However, the tilting is also sensitive to additional molecule-surface interactions between the benzene ring of the adsorptives and further atoms of the surface. The latter interaction can be modified by the heteroatom in the five-membered ring of the adsorptive molecules. It is obvious from the calculations that the tilt angle in the adsorbate is not caused by intermolecular interactions on the surface. This is confirmed by a calculation of two MBT anions bound to two neighboring adsorption sites on the cluster, where no significant difference to the adsorbate structure with a single admolecule was obtained. If the surface relaxation is included into the calculations of the adsorbate complex, the lowering of the surface cadmium atoms into the bulk, which was obtained for the free surface, is reversed. Acknowledgment. The authors thank the Fonds der Chemischen Industrie for financial support. The Deutsche Forschungsgemeinschaft is gratefully acknowledged for providing a scholarship for B.F. as a member of the Graduate College “Physical Chemistry of Interfaces”. References and Notes (1) Brivio, G. P.; Trioni, M. I. ReV. Mod. Phys. 1999, 71, 231. (2) Minot, C.; Markovits, A. J. Mol. Struct. (THEOCHEM) 1998, 424, 119. (3) Uchihara, T.; Oshiro, H.; Kinjo, A. J. Photochem. Photobiol. A: Chem. 1998, 114, 227. (4) Schaufuss, A.; Rossbach, P.; Uhlig, I.; Szargan, R. Fresenius’ J. Anal. Chem. 1997, 358, 262. (5) Szargan, R.; Schaufuss, A.; Rossbach, P. J. Electron Spectrosc. Relat. Phenom. 1999, 100, 357. (6) Contini, G.; Di Castro, V.; Palzonetti, G.; Comelli, G.; Brena, B.; Marabini, A. M. Surf. Sci. 1997, 391, 65. (7) Hallmeier, K. H.; Mayer, D.; Szargan, R. J. Electron Spectrosc. Relat. Phenom. 1998, 96, 245. (8) Mayer, D.; Hallmeier, K. H.; Zerulla, D.; Szargan, R. In SolidLiquid InterfacessMacroscopic Phenomena and Microscopic View; Wandelt, K., Thurgate, S., Eds.; Springer Series Surface Science, SpringerVerlag: Heidelberg, Germany, in press. (9) Cozza, C.; Di Castro, V.; Palzonetti, C.; Marabini, A. M. Int. J. Miner. Process. 1992, 34, 23. (10) Numata, Y.; Wakamatsu, T. In Metallurgical Processes for the Early Twenty-First Century; Sohn, H. Y., Ed.; The Minerals, Metals and Materials Society: Salt Lake City, UT, 1994. (11) Maier, G. S.; Dobias, B. Miner. Eng. 1997, 10, 1375. (12) Tysoe, W. T.; Ormerod, R. M.; Lambert, R. M.; Zgrablich, G.; Ramirezcuesta, A. J. Phys. Chem. 1993, 97, 3365. (13) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. 1994, 94, 2095. (14) Rittner, F.; Boddenberg, B.; Fink, R. F.; Staemmler, V. Langmuir 1999, 15, 1449. (15) Casarin, M.; Maccato, C.; Vittadini, A. J. Phys. Chem. B 1998, 102, 10745.

Adsorption of Mercaptobenzothiazole (16) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (17) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (18) Hehre, W. J.; Radom, L.; Schleyer, P. v. R.; Pople, J. Ab Initio Molecular Orbital Theory; Wiley: New York, 1986. (19) Andrae, D.; Ha¨usermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123. (20) Ahlrichs, R.; Ba¨r, M.; Baron, H. P.; Bauernschmitt, R.; Bo¨cker, S.; Ehrig, M.; Eichkorn, K.; Elliott, S.; Haase, F.; Ha¨ser, M.; Horn, H.; Huber, C.; Ko¨lmel, C.; Kollwitz, M.; Ochsenfeld, C.; O ¨ hm, H.; Scha¨fer, A.; Schneider, U.; Treutler, O.; Armin, M. v.; Weigend, F.; Weis, P.; Weiss, H. Turbomole, version 4.0; Universita¨t Karlsruhe: Karlsruhe, Germany, 1996. (21) Arnim, M. v.; Ahlrichs, R. J. Comput. Chem. 1998, 19(15), 1746. (22) Keller, E. Schakal 97, Program for Graphic Representation for Crystal Structures; Universita¨t Freiburg, Freiburg, Germany, 1997. (23) Ellis, B.; Griffiths, P. J. F. Spectrochim. Acta 1966, 22, 2005. (24) Bo¨hlig, H.; Ackermann, M.; Billes, F.; Kudra, M. Spectrochim. Acta A 1999, 55, 2635. (25) Uher, M.; Kovac, S.; Martvon, A.; Jezek, M. Chem. ZVesti 1978, 32(4), 486. (26) Sartori, G.; Liberti, J. J. Electrochem. Soc. 1950, 97, 20. (27) Beilstein Handbook of Organic Chemistry Fifth Supplementary Series 27/10; Beilstein: Frankfurt a. Main, Germany, 1995; p 491. (28) Lochon, P.; Scho¨nleber, J. Tetrahedron 1976, 32, 3023. (29) Chesick, J. P.; Donohue, J. Acta Crystallogr. B 1971, 27, 1441. (30) Groth, P. Acta Chem. Scand. 1973, 27, 945. (31) Form, G. R.; Raper, E. S.; Downie, T. C. Acta Crystallogr. B 1976, 32, 345. (32) Gillespie, R. J. Molecular Geometry; van Nostrand Reinhold: London, 1972.

J. Phys. Chem. B, Vol. 105, No. 23, 2001 5449 (33) Hursthouse, M. B.; Khan, O. F. Z.; Mazid, M.; Montevalli, M.; O’Brien, P. Polyhedron 1990, 9, 541. (34) McCleverty, J. A.; Gill, S. R.; Kowalski, S. Z.; Bailey, N. A.; Adams, H.; Lumbard, K. W.; Murphy, M. A. J. Chem. Soc., Dalton Trans. 1982, 493. (35) Baggio, R.; Garland, M. T.; Perec, M. J. Chem. Soc., Dalton Trans. 1993, 3367. (36) Becker, U.; Hochella, M. F.; Vaughan, D. J. Geochim. Cosmochim. Acta 1997, 61, 3565. (37) Ma, X.; Schobert, H. H. J. Mol. Catal. A 2000, 160, 409. (38) Wilson, H. M. M. Vib. Spectrosc. 1994, 7, 287. (39) Flemmig, B. To be published. (40) Contini, G.; Carravetta, V.; Parent, Ph.; Laffon, C.; Polzonetti, G. Surf. Sci. 2000, 457, 109. (41) Szymanski, M. A.; Gillan, M. J. Surf. Sci. 1996, 367, 135. (42) Zerulla, D.; Mayer, D.; Hallmeier, K. H.; Chasse´, T. Chem. Phys. Lett. 1999, 311, 8. (43) Wong, Y. T.; Hoffmann, R. J. Phys. Chem. 1991, 95, 859. (44) Rantala, T. T.; Rantala, T. S.; Lantto, V.; Vaara, J. Surf. Sci. 1996, 352-354, 77. (45) Wang, Y. R.; Duke, C. B. Phys. ReV. B 1988, 37, 6417. (46) Schro¨er, P.; Kru¨ger, K. P.; Pollmann, J. Phys. ReV. B 1994, 49, 17092. (47) Horsky, T. N.; Brandes, G. R.; Canter, K. F.; Duke, C. B.; Paton, A.; Lessor, D. L.; Kahn, A.; Horng, S. F.; Stevens, K.; Stiles, K.; Mills, A. P. Phys. ReV. B 1992, 46, 7011. (48) Lanno, M.; Friedel, P. Atomic and Electronic Structure of Surfaces; Springer-Verlag: Berlin, Germany, 1991; p 110.