Surface Science 471 (2001) 151±162 www.elsevier.nl/locate/susc Theoretical study of the adsorption of urea related species on Pt(1 0 0) electrodes Maite Garcõa-Hern andez a,b, Uwe Birkenheuer a,1, Anguang Hu a,2, sch a Francesc Illas b,*, Notker Ro a b Institut f ur Physikalische und Theoretische Chemie, Technische Universit at M unchen, 85747 Garching, Germany Departament de Quõmica Fõsica & Centre Especial de Recerca en Quõmica Te orica, Universitat de Barcelona, C/Martõ i Franqu es 1, 08028 Barcelona, Spain Received 14 August 2000; accepted for publication 10 October 2000 Abstract We present a cluster model study on the two most likely urea adsorption complexes at Pt(1 0 0) electrodes. A goal of the investigation is to determine whether urea or ureylene, a HNCONH biradical species, forms the most stable adsorption complex. Geometry optimisations of both urea species adsorbed at various surface sites have been carried out with the parallel P A R A G A U S S program package using gradient-corrected density functionals. Scalar-relativistic allelectron calculations as well as pseudopotential calculations have been performed. For each optimised structure, vibrational frequencies have been calculated. From energy considerations, ureylene adsorption is found in both types of calculations to be preferred over urea adsorption, consistent with the interpretation of electrochemical measurements, even though eective core potential results tend to underestimate the binding energies. This assignment is supported by the calculated CO vibrational stretching frequency of chemisorbed ureylene which is closer to the value obtained from ``in situ'' Fourier transform infrared spectroscopy in an electrochemical environment than the CO frequency calculated for adsorbed urea. Ó 2001 Elsevier Science B.V. All rights reserved. Keywords: Platinum; Density functional calculations; Chemisorption 1. Introduction * Corresponding author. Tel.: +34-93-402-1229; fax: +34-93402-1231. E-mail address: [email protected] (F. Illas). 1 Present address: Max-Planck-Institut f ur Physik komplexer Systeme, N othnitzer Strasse 38, 01187 Dresden, Germany. 2 Present address: Department of Chemistry, University of Minnesota, Minneapolis, MN 55455 0431, USA. There is little doubt that understanding adsorption processes at a molecular level requires full comprehension of the interactions between molecules and solid surfaces [1±3]. This claim for understanding at a microscopic level is even more stringent for interactions occurring at electrochemical interfaces since it is impossible to control all parameters at a level similar to that which can be achieved in experiments carried out under 0039-6028/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 6 0 2 8 ( 0 0 ) 0 0 9 0 1 - 8 152 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 ultra-high vacuum (UHV), conditions. A case in point is ``in situ'' Fourier transform infrared spectroscopy (FTIR), applied in an electrochemical environment. While this technique enables to obtain precise experimental data about the species that are adsorbed at the electrode, their interpretation is not simple at all and thus in situ FTIR does not permit an univocal assignment because of the diculty, if not the impossibility, to compare to spectra of known species. Therefore, it is not surprising that the interpretation of FTIR in situ spectra is most often based on comparison to existing inorganic complexes. However, this complex-to-surface analogy does not guarantee a correct assignment [4]. As an alternative, quantum chemistry investigations applied to realistic surface models can provide an adequate, yet ¯exible, computational approach to the chemistry of such complex systems. Computational experiments permit to verify or discard a given hypothesis and suggest new interpretations of experimental data. The present work is devoted to such a theoretical study of the adsorption of urea and related species on a single crystal Pt(1 0 0) electrode. Several experimental studies [5±10] addressed structural aspects of urea adsorption at platinum electrodes and reactions of such adsorption systems. These investigations were carried out using dierent well-de®ned single-crystal surfaces and used a variety of experimental techniques including voltammetry [5,6,8], radiochemistry [8], ``ex situ'' low electron energy diraction [6,8] (LEED), and Auger electron spectroscopy [6,8]. These data permitted to derive the dependence of the potential on urea coverage, the potential saturation value, and the saturated urea adlayer surface structure. One particular feature of the adsorption of urea in the Pt(1 0 0) surface is the very large excess charge compared to that measured in other crystallographic surfaces. This charge excess due to the presence of urea in the media indicates that the adsorbate undergoes a charge-transfer process at the electrode. The adsorption and desorption processes take place with a net charge transfer of two electrons per adsorbed molecule [5,6,8]. This indicates that urea adsorption may produce adsorbed ureylene according to the following electrochemical process: Hydrogen adsorption at the Pt(1 0 0) surface may be an important intermediate stage of this reaction. The adsorption and desorption voltammogram peaks recorded on Pt(1 0 0) are very sharp and indicate that urea adsorption is reversible [5,6,8]. On the other hand, the stable c 5 20 LEED pattern observed on a clean Pt(1 0 0) surface changes with urea adsorption to a c 2 4 LEED pattern [6,8]. Radiochemistry measurements [8] gave a coverage of 0:26 0:04 ML (number of molecules per Pt(1 0 0) unit cell) which is fully consistent with the value of 0:24 0:03 ML obtained from the Auger electron spectroscopy carbon signal [8]. Climent et al. reported complementary studies of the adsorption of urea on Pt(1 0 0) [11] and Pt(1 1 1) [12] electrodes. The voltammetric and charge displacement analysis on Pt(1 0 0) are similar to those reported by Wieckowski et al. [5±10] under analogous conditions. Moreover, the integrated charge was found to be larger than the one corresponding to adsorption and desorption of a single hydrogen atom by 50%, thus indicating the existence of additional urea related surface processes. In situ FTIR spectra reported by Feliu et al. [11] provide additional information on the nature and structure of the urea species adsorbed on Pt(1 0 0). A bipolar band, typical of potential dependent frequencies of the vibrational modes of adsorbed species, appears centred at 1705 cm±1 . Based on the similarity to the urea carbonyl group in coordination compounds [13,14] and on the fact that this peak does not appear at potentials at which there is no adsorbed urea, the peak at 1705 cm±1 was assigned to urea bonded at the Pt(1 0 0) surface through the nitrogen atoms, in a bridge con®guration and with the carbonyl group normal to the electrode surface. Likewise, once the electrode has been in contact with the urea solution, the appearance of this peak M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 neither depends on the electrode potential nor on the presence or absence of urea in the solution where the Pt electrode is re-immersed. The ®rst theoretical study of urea adsorbed on Pt(1 0 0) was reported by Rikvold and Wieckowski [7]. In this work urea adsorption on Pt(1 0 0) was investigated within the gas-lattice model employing a Monte Carlo simulation. This simple model suggests that urea is adsorbed along Pt±Pt bonds occupying two adsorption sites and co-exists with hydrogen atoms adsorbed on top of the platinum atoms. Based on the fact that most of the surface hydrogen atoms desorb at a potential close to that corresponding to urea adsorption, these authors suggest that urea adsorbs on sites formerly occupied by hydrogen atoms. Further Monte Carlo simulations within the lattice gas-model [8±10] considered urea coordination at the platinum surface through the nitrogen atoms. Clearly, a non-empirical approach appears desirable. The present work is devoted to the analysis of the adsorption properties of urea and related species on Pt(100) electrodes. Binding energies, geometrical parameters, and vibrational frequencies of urea ± (H2 N)CO(NH2 ) ± and ureylene ± (HN)CO(NH) ± the two most likely adsorbed species, are studied using a cluster model approach to represent the Pt(1 0 0) surface and gradientcorrected density functionals (DFs) to account for correlation. Relativistic eects due to the presence of platinum atoms are also taken into account (for details see below). 2. Surface cluster models The cluster model approach to chemisorption assumes that the properties of interest are of local nature and, hence, restriction to a ®nite fraction of the extended surface does not introduce serious artefacts. This approach has long been used to model surfaces and chemisorption phenomena and its success and limitations are well documented [1± 3]. Reasonably large cluster models have been employed in the present study to simulate the various adsorption sites of urea and ureylene on the Pt(1 0 0) surface. Adopting the Ptn m1 ; m2 ; 153 Fig. 1. Schematic view of the cluster models used to represent dierent adsorption sites of the Pt(1 0 0) surface: (a) Pt14 (8,6), (b) Pt9 (4,5), and (c) Pt9 (5,4). m3 ; . . . notation where n denotes the total number of platinum atoms in the cluster and mi the number of atoms in the ith crystal layer (starting with the top-most one at the surface), the cluster models chosen for the various adsorption sites of Pt(1 0 0) are Pt14 (8,6), Pt9 (5,4), and Pt9 (4,5), respectively (see Fig. 1). These cluster models were considered as rigid fractions of a Pt bulk crystal with the nearest-neighbour Pt distance set to the experi [15]. The geometry optimental value of 2.77 A misation for urea and ureylene on Pt(1 0 0) was carried out using analytical energy gradients. C2 point group symmetry was imposed during geometry optimisation with the only additional constraint ± beside the ®xed Pt±Pt distances ± that the orientation of the molecular N±CO±N plane is preserved. In this way one ensures that the adsorbate will not change site during the geometry optimisation. In all cases the calculated gas phase geometry of urea was taken as starting 154 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 con®guration for both urea and, after abstracting two hydrogen atoms, for ureylene. Four dierent adsorption sites have been considered, the corresponding positions of the urea skeleton are indicated schematically in Fig. 2. These sites can be characterised as follows: (i) on top site ± each nitrogen atom on top of a Pt atom (Fig. 2a), (ii) hollow site ± each nitrogen atom above a four-fold hollow site (Fig. 2b), (iii) bridgeontop site ± each nitrogen atom in a bridging position above two surface Pt atoms such that there is one Pt atom just below the C atom (Fig. 2c), and (iv) bridge-hollow site ± each nitrogen atom again in a bridging position, but with a four-fold hollow below the C atom (Fig. 2d). 3. Computational details Recall that relativistic eects are important for systems containing heavy atoms; hence they must be included to describe the electronic structure of the Pt surface clusters used in the present study. Two dierent approaches have been used for that purpose. In a ®rst series of calculations all electrons (AE), are treated explicitly within a scalar-relativistic variant [16] of the linear combination of Gaussian-type orbitals DF method [17] (LCGTO-DF) as implemented in the program P A R A G A U S S for parallel computers [18,19]. The same computer code was used for a second series of calculations (the ECP approach) where scalarrelativistic eects are introduced indirectly through the usage of pseudopotentials [20]; in the present study these are relativistic pseudopotentials of the Stuttgart type [21]. In both cases, a large and ¯exible basis was employed to expand the Kohn± Sham orbitals. For the AE Pt atoms the basis set is (21s,17p, 12d,7f) contracted to [9s,8p,5d,2f]. The primitive basis set is derived from a (19s,14p,10d, 5f) GTO basis [22] by adding two s, three p, two d, and two f exponents as proposed by Ferrari et al. [23] in their study of Pt clusters supported on zeolites. For the other atoms ¯exible GTO basis sets were employed [24,25]. For C, N, and O, these are (9s,5p,1d) basis sets contracted to [5s,4p,1d], and for H the basis is (6s,1p) contracted to [4s,1p]. In the pseudopotential calculations a small core containing 60 electrons was used for Pt; the 5s, 5p, 5d, and 6s electrons are included in the valence shell together with a (7s,6p,5d) primitive basis set contracted to [6s,3p,2d] [26]. The gradientcorrected BP exchange-correlation functional, a combination of BeckeÕs exchange functional [27] and PerdewÕs correlation functional, [28] was employed throughout. All calculations were carried out in spin-restricted fashion, since no open-shell con®gurations were expected among the potential adsorption species. 4. Structure and stability of urea species on Pt(1 0 0) First, we brie¯y describe the results obtained for gas phase urea and compare them with available theoretical and experimental data. Both microwave spectroscopy [29] and matrix isolation studies [30] have shown that, contrarily to organic chemistry textbook arguments, urea is not a planar molecule. More recently Godfrey et al. [31] combined MP2/6-311++G(d,p) calculations with all previous experimental data to fully characterise the structure of the urea molecule. Later, Rousseau et al. [32,33] reported a detailed assignment of the vibrational frequencies of urea and deuterated urea. The geometric parameters of free gas phase urea predicted by the present DF, calculations (see Table 1) are in excellent agreement with the MP2 results just mentioned as well as with experiment [31]. The accuracy of the DF calculations is about for distances, 1±2° for bond angles, and 5° 0.01 A for the pyramidalisation angle as de®ned in Ref. [29]. We now turn to the adsorption process. Within the ECP approach, the equilibrium geometry of both adsorption species, urea and ureylene, have been determined all for adsorption sites of Pt(1 0 0) described above. For ureylene, which is the most interesting species from the electrochemical point of view, the adsorption geometries were also determined by scalar-relativistic AE calculations. A comparison of these two approaches will be presented later. The structural parameters of the adsorbed species are reported in Table 1 (for urea) and Table 2 (for ureylene). The ®rst surprising result, even without any reference to gas-phase M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 155 Fig. 2. Schematic top and side view of the adsorption complexes considered (only the NCON skeleton of urea is shown): (a) ontop, (b) hollow, (c) bridge-ontop, and (d) bridge-hollow. 156 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 Table 1 Equilibrium structure of urea on Pt(1 0 0) from a relativistic pseudopotential description of the Pt core electrons Site cluster Distances C±O C±N N±Hup N±Hdown N±N N±Pt N±surf a H±Pt H±surf a Ontop Pt14 (8,6) ± ± 1.207 1.459 1.040 1.030 2.463 2.205 2.200 Hollow Pt14 (8,6) 1.235 1.386 1.017 1.024 2.343 3.853 3.399 2.899 2.420 Bridge-ontop Pt9 (5,4) ± ± 1.212 1.457 1.029 1.029 2.423 2.946 2.685 Bridge-hollow Pt9 (5,4) Free urea 1.232 1.390 1.030 1.018 2.337 3.525 3.234 2.504 2.265 1.228 1.396 1.021 0.998 2.338 Angles O±C±N 122.4 122.3 123.7 122.8 C±N±Hup 106.8 114.1 107.5 118.9 C±N±Hdown 108.6 120.6 107.5 114.2 N±C±N 115.2 102.4 112.5 114.4 O±C±N±Hup ÿ73.1 14.2 56.6 ÿ157.3 O±C±N±Hdown 41.7 160.2 ÿ56.5 ÿ14.2 angles in degrees. A C2 symmetry constraint was imposed during the geometry optimisation. Distances in A, a Distance to the top ``crystal'' plane of the cluster model. ± ± ± ± 123.1 112.8 119.2 113.7 12.8 148.9 Table 2 Equilibrium structure of the ureylene radical on Pt(1 0 0) from a relativistic pseudopotential description of the Pt core electrons Site cluster C±O C±N N±H N±N N±Pt N±surf a Ontop Pt14 (8,6) Distances 1.244 1.375 1.021 2.338 1.993 1.981 Hollow Pt14 (8,6) 1.227 1.399 1.028 2.235 2.459 1.696 Bridge-ontop Pt9 (5,4) 1.224 1.404 1.031 2.311 2.208 1.882 Bridge-hollow Pt9 (5,4) 1.221 1.450 1.030 2.491 2.120 1.598 Angles O±C±N 121.8 127.0 124.6 120.7 C±N±H 113.6 111.6 109.2 106.5 N±C±N 116.4 106.0 110.8 118.5 O±C±N±H 0.0 0.1 50.5 0.0 angles in degrees. A C2 symmetry constraint was imposed during the geometry optimisation. Distances in A, a Distance to the top ``crystal'' plane of the cluster model. urea, is that the relative orientation of the hydrogen atoms of the amide groups with respect to the surface diers substantially among the four different sites, see Fig. 2. This is at variance with the case of ureylene precisely because of the lack of one hydrogen atom in each amide group (see below). For urea in the ontop and bridge-ontop sites, the lone pairs of the nitrogen atoms point towards Free urea ± ± 1.228 1.396 1.018 2.338 123.1 112.5 113.7 ÿ12.8 the surface, and the hydrogen atoms of each amide group point away from the surface (see middle columns of Fig. 2a and c), i.e. the surface bond is mediated through the lone-pairs on the amide nitrogen atoms. Compared to the gas phase, the C±O distance decreases concomitantly with an increase of the C±N and the two N±H distances. Note also that for the bridge-ontop adsorption site M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 the urea molecule becomes C2v symmetric, even though this is not enforced by symmetry constraints. At the hollow and bridge-hollow sites, apart from a slight elongation of the N±H and C±O bonds and a shortening of the C±N bonds, the most interesting feature is that the atoms closest to the surface are two hydrogen atoms (see middle columns of Fig. 2b and d). Hence, at these two sites urea interacts with the surface directly through one hydrogen atom per amide group rather than through the nitrogen lone pairs. Apparently a bond con®guration is preferred which may be regarded as a precursor for dehydrogenation of urea via hydrogen adsorption on Pt(1 0 0). In spite of having lost two hydrogen atoms, the geometry of adsorbed ureylene is rather reminiscent of that of gas-phase urea. In fact, in adsorbed ureylene the C±O distance remains almost unchanged with respect to free urea as do most of the bond angles (see Table 2). Much more interesting is the position of the remaining hydrogen centres relative to the urea skeleton, right column of Fig. 2. Except for the bridge-ontop site, the hydrogen atoms lay almost perfectly in the N±(CO)±N plane, indicating a symmetric two-fold coordination (including the lone-pair) of the nitrogen atoms toward the Pt surface. The resulting coordination of the ureylene nitrogen atoms can be rationalised in terms of an almost sp3 hybridisation. Two of the sp3 hybrid orbitals form the N±C and the N±H bonds while the two remaining hybrids establish the surface bonding. Hereby, the platinum surface is meant to provide the one extra electron necessary to form the two equivalent single bonds. Notice that the participation of the extra electron to the surface bond does not necessarily imply a charge donation to the adsorbate. The formation of such a two-fold coordination is obvious for the most stable adsorption site, the bridge-hollow site, which in turn rationalises the energetic preference of that site. For the hollow and ontop sites, this two-fold coordination is not apparent, indicating a more delocalised bonding mechanism. For the bridge-ontop site, on the other hand, the dihedral OCNH angle is about 50°, indicating a deviation form planarity, at variance from the value zero for the other cases discussed above. The corresponding structure is again compatible with a typical 157 sp3 -type hybridisation of the amide nitrogen atoms, with one lone pair of each nitrogen atom oriented almost perpendicularly to the molecular plane and the other one pointing directly towards the surface. Finally, it is worth mentioning that the surface bonding mechanism for ureylene to the bridge-hollow site is favoured with respect to the direct hydrogen-surface interaction encountered in urea adsorption at that site simply because of the need to saturate the free valences of the ureylene radical. Now we turn to a discussion of the relative stability of urea and ureylene adsorbed at the dierent surface sites. In order to compare the adsorption energies of the various species, a thermodynamic cycle has been used. In this way, one avoids any explicit reference to the energy of the gas-phase biradical and at the same time one does not introduce any bias on the relative stability. For adsorbed urea, the interaction energy is given by that of the simple reaction: CO NH2 2 Ptn m1 ; m2 ! CO NH2 2 ads 1 As usual, a negative value indicates that products are more stable than reactants. However, for adsorbed ureylene it is convenient to refer to the adsorbed species plus adsorbed hydrogen. Therefore, the interaction energy is calculated as the energy of the following reaction: CO NH2 2 Ptn m1 ; m2 ! CO NH2 2 ads 2H ads 2 Here, H ads designates adsorbed atomic hydrogen although one must realise that in the electrochemical environment this adsorbed species becomes oxidised. In order to obtain a unique reference energy for H ads , the interaction energy of atomic hydrogen was computed for dierent high symmetry sites of Pt(1 0 0) using the various cluster models shown in Fig. 1. The largest interaction energy, 69.0 kcal molÿ1 for the ECP approach (72.6 kcal molÿ1 with the AE treatment), was found for adsorption of a single H at the bridge site of the Pt14 (8,6) cluster which falls well into the range of about 60±80 kcal molÿ1 for hydrogen-metal binding energies as typically observed in experiment [34,35]. This value was taken 158 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 to compute the interaction energies according to reaction (2). It is worth comparing the present cluster calculations to recent periodic DF calculations using the ADF-Band code [36]. These authors report adsorption energies for H on Pt(1 1 1) relative to gas phase H2 . For a 0.25 coverage the adsorption energy calculated obtained with the BP functional is about 11 kcal molÿ1 . Using the experimental binding energy of ÿ103.3 kcal molÿ1 for the hydrogen molecule and the results reported by Olsen et al. [36], it is straightforward to obtain the corresponding atomic hydrogen chemisorption energy. This leads to a value of 63 kcal molÿ1 that can be directly compared to the present cluster model results reported above. This result corroborates the cluster estimates discussed above. Note, however, that the slab model calculations refer to Pt(1 1 1), use a frozen core approximation, and a dierent basis set whereas the present calculations are for H adsorption on Pt(1 0 0). Therefore, only an approximate comparison is meaningful. Given the dierences between both theoretical studies, the agreement between cluster and periodic calculations is rather satisfactory. For adsorbed urea, both hollow and bridgehollow sites are most stable (ÿ5 kcal molÿ1 ) and energetically nearly equivalent, the former being lower in energy by 0.7 kcal molÿ1 only (Table 3). Adsorption in the ontop site is endothermic by 0.6 kcal molÿ1 and, ®nally, the interaction above the bridge-ontop site is highly disfavoured by more than 25 kcal molÿ1 . Moreover, given the wellknown limitations of cluster models when adsorption energies are to be determined [1±3], we consider these ®rst three cases as comparable in their binding energy (see below). A rather dierent situation occurs when we consider the relative stability of adsorbed ureylene. With an interaction energy of 27 kcal molÿ1 , the most stable adsorption complex is at the bridgehollow site (Table 3). The adsorption process described by reaction (2) is preferred over nondissociative adsorption of urea ± reaction (1) ± at the bridge hollow site by about 22 kcal molÿ1 . This is consistent with the hypothesis that upon adsorption urea looses two hydrogen atoms [11]. Interaction at the three other sites is far less favourable than in the case of urea. This result is of special relevance since it is in agreement with the experimental observation of a urea adlayer which is strongly bound to the Pt electrode [11]. In fact, this adlayer is so tightly bound to the electrode surface that it is detected even after the electrode is removed from the urea solution and an in situ FTIR measurement is performed after inserting the electrode into a urea-free solution [11]. Therefore, the bridge hollow adsorption of the urea molecule can be regarded as a ®rst step of a more complex adsorption mechanism in an electrochemical environment. Next, we will address the accuracy of the pseudopotential approach compared to scalarrelativistic AE calculations. This comparison has been carried out for ureylene on the four active sites described above. Overall, the adsorbate structures arising from the more accurate AE calculations are very similar to those obtained with the ECP approach. The changes in the geometry of adsorbed ureylene are essentially negligible. The only noticeable geometrical variation is found for shorter in the Pt±N distance; it is by 0.03±0.06 A the AE calculations. Quite noticeable dierences are obtained for the adsorption energies. The AE calculations result in an additional stabilisation of the adsorbed species by 18±29 kcal molÿ1 compared to pseudopotential results (Table 3). This energy dierence is larger than usually encoun- Table 3 Interaction energies (in kcal molÿ1 ) of urea and ureylene on a Pt(1 0 0) surface according to the reactions (1) and (2) (see text) as obtained from relativistic pseudo potential (ECP) and AE calculationsa Urea Ureylene a ECP ECP AE Ontop Hollow Bridge-ontop Bridge-hollow 0.6 ÿ10.1 ÿ30.1 ÿ5.2 54.3 41.6 26.6 38.0 20.1 ÿ4.6 ÿ26.5 ÿ55.4 Negative energies imply products being more stable than reactants. M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 tered in transition metal compounds and indicates some limitation of the Pt pseudopotential representation (likely due to the ECP basis set). Note, however, that from Table 3 one can see that these energy dierences are fairly uniform over the whole series of adsorption sites investigated and hence, the relative stability of the dierent adsorption complexes is not aected. Finally, we come to a discussion of the adsorption energetics as it emerges from the results obtained with ECP and AE models. Besides the increase of the binding energies in AE calculations, we need to discuss two further limitations of the computational methodology employed. Since, so far the basis set superposition error (BSSE) has been neglected, the calculated interaction energies may be somewhat too large. On the other hand, common exchange-correlation approximations (such as the BP functional used in the present work) are known to underestimate weak interactions of about 5 kcal molÿ1 or less as found here for some adsorption complexes (Table 3) [37]. To estimate the basis set superposition error, we applied the counterpoise correction [38] to urea adsorbed in the most stable hollow adsorption site. This correction decreases the ECP binding energy by about 3 kcal molÿ1 , from ÿ5.2 to ÿ2.4 kcal molÿ1 . On the other hand, binding energies obtained from the relativistic AE calculations are expected to exhibit a larger BSSE eect. Indeed, with the counterpoise correction, the interaction energy of ureylene at the bridge-hollow site is reduced by about 7 kcal molÿ1 , while the ECP result changes by about 2 kcal molÿ1 only. However, it is obvious that even with the BSSE correction taken into account, ECP binding energies are notably underestimated compared to the corresponding AE results, by about 10±25 kcal molÿ1 . With such increased interaction energies (as they are expected to derive from AE calculations even in cases where the ECP approach yields weak interactions only), the above caveat concerning weak interactions does not seem pertinent in the present context. In any case, the stability trend of the various sites is expected to be identical for relativistic AE and relativistic ECP results, but some caution is required when discussing and interpreting absolute values of the interaction energies. 159 5. Vibrational frequencies of urea and urea species on Pt(1 0 0) Despite its apparent simplicity and the considerable amount of theoretical and experimental work [31±33], the vibrational spectrum of gasphase urea is still far from being completely interpreted. Interestingly, the present DF calculations are in very good agreement with experimental data, especially for the frequencies in the range from 1000 to 3500 cmÿ1 , corroborating some previous assignments that had to be quali®ed as tentative [30]. The main interest of the present work lies in the mCO stretching frequency of the adsorbed species because this is the only frequency that is identi®ed in the in situ FTIR experiments. Nevertheless, the gas phase vibrational spectrum is of interest for comparison, too. The calculated vibrational frequencies together with other available data are reported in Table 4. For gas-phase urea, the accuracy of the present BP results is noteworthy, with a mean absolute deviation of 22 cmÿ1 and maximum deviations of at most 50 cmÿ1 in the less favourable case. It is also worth pointing out that this accuracy is signi®cantly better than that of the MP2 method for the soft modes of urea [31] and comparable to that of empirically corrected HF results [32]. For the adsorption complexes, full vibrational analyses have been carried out for both adsorbed urea and ureylene. However, since the experimental information is rather limited there is little value in reporting the full series of data here. Instead, selected vibrational frequencies, namely for the C±O, C±N and N±H symmetric stretching motions, are reported in Table 5. Unfortunately, several dierent adsorption complexes exhibit calculated C±O stretching frequencies near 1700 cmÿ1 . For urea, somewhat higher CO frequencies are calculated, from about 1700±1800 cmÿ1 , than for ureylene, where the CO frequencies of the various cluster models range from about 1600± 1700 cmÿ1 (Table 5). Therefore, from these computed vibrational data it is impossible to infer either the nature of the adsorbed species or to the active site. However, for the most stable species and adsorption site, ureylene at the bridge-hollow site, the calculated C±O stretching frequency 160 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 Table 4 Vibrational frequencies (in cmÿ1 ) of gas-phase urea as obtained from Becke±Perdew, BP, DF calculations compared to available experimental dataa Present work Assignment BP Experimental Ref. [33] ss (NH2 ) xa (NH2 ) d(CN) sa (NH2 ) xs (NH2 ) d(CO) x(CO) ms (CN) qa (NH2 ) qs (NH2 ) ma (CN) ds (NH2 ) da (NH2 ) m(CO) ms (NH2 )a ms (NH2 )s ma (NH2 )a ma (NH2 )s 426 (A) 458 (B) 475 (A) 537 (B) 557 (B) 585 (A) 747 (B) 919 (A) 1014 (B) 1148 (A) 1370 (B) 1584 (A) 1587 (B) 1748 (A) 3466 (B) 3473 (A) 3590 (B) 3591 (A) Ref. [32] Ref. [30] 775 1032 227 410? 542? 578? 618? 790? 1014 1157 1394 1604 1749 1776 3434 3460 3533 3559 1394 1594 1594 1734 3440 3440 3448 3548 410 578 790 960 1014 1394 1594 1594 1734 3440 3440 3548 3548 a The question marks indicate tentative experimental assignments. The symmetry character of the calculated modes, A or B of the C2 point group, is also given. Table 5 Calculated pseudopotential BP values of the symmetric vibrational frequencies of urea and ureylene (in cmÿ1 ) at various sites of the Pt(1 0 0) surfacea Assignment Ontop Pt14 (8,6) Hollow Pt14 (8,6) Bridge-ontop Pt9 (5,4) Bridge-hollow Urea ms (CN) m(CO) ms (NH2 )s ma (NH2 )s 707 1772 3139 3362 930 1646 3391 3584 737 1736 3313 3347 933 1681 3264 3568 919 1748 3473 3591 ms (CN) m(CO) ms (NH2 )s ma (NH2 )s Ureylene ms (CN) m(CO) ms (NH)s 1046 1583 3491 772 1646 3410 860 1658 3280 764 (773) 1693 (1686) 3395 (3552) 919 1748 3473 ms (CN) m(CO) ms (NH2 )s Gas phase urea Pt9 (5,4) Assignment a Frequencies calculated at the all-electron level are given in parentheses. Calculated gas phase frequencies of urea and their assignment are given for comparison. amounts to 1693 cmÿ1 , which is very close to 1705 cmÿ1 , the frequency experimentally observed by Climent et al. [11] in in situ FTIR experiments. Remarkably, by comparison to dierent inorganic complexes, this experimental group deduced precisely the same assignment. Notice that, after urea adsorption as ureylene at the bridge-hollow site, the C±N stretching mode undergoes a shift to a lower frequency, consistent with the trend observed in coordination compounds having urea as an N-bonded ligand [14]. However, the fact that the surface-complex anal- M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 ogy permits, in this case, a correct assignment of the coordination mode and of the observed vibrational frequency does not validate this approach in general. It is important to re-iterate that an adsorbate frequency assignment by comparison to the vibrational frequencies of these adsorbates acting as ligands in complexes can be very misleading [4]. In the present case of urea adsorbed on Pt(1 0 0) electrodes, both frequencies obtained either from the in situ FTIR measurements or in Ptcomplexes are fortuitously close, but this is by no means a strict rule. Actually, inspection of Table 5 reveals that variations of the order of 100 cmÿ1 are typically encountered for the C±O stretching frequency of the two urea species as a function of coordination to the Pt atoms of the electrode. This clearly indicates the limitation of the surfacecomplex analogy. Rather, comparison to reasonable theoretical models of the type used in the present work provides a more adequate framework for understanding the adsorbate±substrate chemical bonding that occurs in these adsorption complexes. Finally, it is worth pointing out that the present analysis remains largely unchanged if the computationally more accurate AE approach is employed. In fact, in Table 5 we also present a comparison of the frequencies for the most stable bridge-hollow site obtained either by the ECP approach or by the scalar-relativistic AE approach. The vibrational frequencies of interest dier by less than 4%. Therefore, we conclude that as far as vibrational frequencies are concerned, the approximate treatment of the platinum core via a relativistic pseudopotential results in a considerable saving of computational time without any signi®cant loss of accuracy. 6. Summary and conclusions The adsorption of urea and ureylene on a Pt(1 0 0) single crystal electrode has been studied using two dierent DF based quantum chemistry approaches (AE scalar-relativistic and pseudopotential) as well as reasonably large cluster model representations of dierent active sites of Pt(1 0 0). 161 The choice for urea and ureylene is related to the fact that electrochemical experiments suggest urea to lose two H atoms in the adsorption process leading to adsorbed ureylene. Four dierent adsorption sites have been modelled and the geometry, the interaction energy and the vibrational frequencies of both adsorbed species have been calculated. In the present study, dissociative urea adsorption ± with two removed hydrogen atoms to adsorbed as atomic hydrogen on Pt(1 0 0) ± is found to be the most favourable adsorption process. Thus the present study corroborates previous interpretations derived from electrochemical experiments [11]. The bridge-hollow site, schematically shown in Fig. 2d, turns out to be the most stable adsorption complex of ureylene. Furthermore, calculated stretching frequencies of the C±O vibrations of adsorbed urea and ureylene, also at dierent sites, exhibit similar values, all of them rather close to the value obtained in the in situ FTIR experiment. Hence, the calculated vibrational frequencies alone do not provide enough information for deciding whether the adsorbed species is urea or ureylene and even less on the nature of the adsorption site. Yet, the cluster model calculations permit an assignment of the adsorbed species and the active site based on calculated interaction energies. Interestingly, for the most stable species, adsorbed ureylene, at the most favourable site, the bridge-hollow site, the calculated value for the C±O stretching frequency is closest to the experimental value. To summarise, the relative stability of the dierent adsorbed species and the vibrational frequency analysis support the hypotheses by Climent et al. [11] that urea adsorbs as ureylene, that this species is bonded to the surface through the two nitrogen atoms, and that adsorption occurs at the bridge-hollow site. Nevertheless, it is worth pointing out that the hypotheses suggested in the work of Climent et al. [11] were mostly based in chemical intuition and on invoking the so-called surface-complex analogy whose success is by far not guaranteed in general [4]. Here, it was demonstrated that ``®rst principles'' calculations lead to the same ®nal picture, advocating that theoretical investigations on models of adsorbates at surfaces provide an alternative and more secure 162 M. Garcõa-Hern andez et al. / Surface Science 471 (2001) 151±162 way to interpreting in situ electrochemical interfaces at a molecular level. Acknowledgements The authors are indebted to Prof. Juan Feliu and Dr. Victor Climent for bringing the problem of urea adsorption to their attention. M. GarcõaHernandez is grateful to the ``Generalitat de Catalunya'' for a predoctoral grant. This work has been supported by Deutsche Forschungsgemeinschaft, Fonds der Chemischen Industrie, Spanish ``Ministerio de Educaci on y Ciencia'', project CICyT PB98-1216-C02-01, and ``Generalitat de Catalunya'', project 1999-SGR-00040. References [1] J.L. Whitten, H. Yang, Surf. Sci. Rep. 24 (1996) 59. [2] P.S. Bagus, F. Illas, in: P.R. Schleyer, N.L. Allinger, T. Clark, J. Gasteiger, P.A. Kollman, H.F. Schaeer III, P.R. Schreiner (Eds.), Encyclopedia of Computational Chemistry, Encyclopedia of Computational Chemistry, vol. 4, Wiley, Chichester, 1998, p. 2870. [3] G. Pacchioni, Heterog. Chem. Rev. 2 (1996) 213. [4] A. Markovits, M. Garcõa-Hern andez, J.M. Ricart, F. Illas, J. Phys. Chem. B 103 (1999) 509. [5] M. Rubel, C.K. Rhee, A. Wieckowski, P.A. Rikvold, J. Electroanal. Chem. 315 (1991) 301. [6] C.K. Rhee, J. Electrochem. Soc. 139 (1992) 13C. [7] P.A. Rikvold, A. Wieckowski, Phys. Scripta T 44 (1992) 71. [8] M. Gamboa-Aldeco, P. Mrozek, C.K. Rhee, A. Wieckowski, P.A. Rikvold, Q. Wang, Surf. Sci. Lett. 297 (1993) L135. [9] P.A. Rikvold, M. Gamboa-Aldeco, J. Zhang, M. Han, Q. Wang, H.L. Richards, A. Wieckowski, Surf. Sci. 335 (1995) 389. [10] P.A. Rikvold, J. Zhang, Y.-E. Sung, A. Wieckowski, Electrochim. Acta 41 (1996) 2175. [11] V. Climent, A. Rodes, J.M. Orts, J.M. Feliu, J.M. Perez, A. Aldaz, Langmuir 13 (1997) 2380. [12] V. Climent, A. Rodes, J.M. Orts, A. Aldaz, J.M. Feliu, J. Electroanal. Chem. 461 (1999) 65. [13] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, Wiley, New York, 1986. [14] R.B. Penland, S. Mizushima, C. Curran, J.V. Quagliano, J. Am. Chem. Soc. 79 (1957) 1575. [15] R.W.G. Wycko, Crystal Structures, second edition, vol. 1, Interscience Publishers, New York, 1965. [16] N. R osch, K. Kr uger, M. Mayer, V.A. Nasluzov, in: J.M. Seminario (Ed.), Recent Development and Applications of Modern Density Functional Theory, vol. 4, Theoret. Comput. Chem. Elsevier, Amsterdam, 1996, p. 497. [17] B.I. Dunlap, N. R osch, Adv. Quant. Chem. 21 (1990) 317. [18] T. Belling, T. Grauschopf, S. Kr uger, F. N ortemann, M. Staufer, M. Mayer, V.A. Nasluzov, U. Birkenheuer, A. Hu, A.V. Matveev, N. R osch, Program ParaGauss 2.1, Technische Universit at M unchen, 1999. [19] T. Belling, T. Grauschopf, S. Kr uger, M. Mayer, F., N ortemann, M. Staufer, C. Zenger, N. R osch, in: High Performance Scienti®c and Engineering Computing, Proceedings of the First International FORTWIHR Conference, February 1998, in: H.-J. Bungartz, F. Durst, C. Zenger (Eds.), Lecture Notes in Computational Science and Engineering, vol. 8, Springer, Heidelberg, 1999, p. 439. [20] A. Hu, M. Staufer, U. Birkenheuer, V. Igoshine, N. R osch, Int. J. Quant. Chem. 79 (2000) 209. [21] D. Andrae, U. Haeussermann, M. Dolg, H. Stoll, H. Preuss, Theor. Chim. Acta 77 (1990) 123. [22] O. Gropen, J. Comput. Chem. 8 (1987) 982. [23] A.M. Ferrari, K.M. Neymann, T. Belling, M. Mayer, N. R osch, J. Phys. Chem. B 103 (1999) 216. [24] F.B. van Duijneveldt, IBM Res. Rep. RJ945, 1971. [25] S. Huzinaga (Ed.), Gaussian Basis Sets, Elsevier, Amsterdam, 1984. [26] M. Dolg, H. Stoll, H. Preuss, R.M. Pitzer, J. Phys. Chem. 97 (1993) 5852. [27] A.D. Becke, Phys. Rev. A 38 (1988) 3098. [28] J.P. Perdew, Phys. Rev. B 33 (1986) 8622; Erratum 34 (1986) 7406. [29] R.D. Brown, P.D. Godfrey, J. Storey, J. Mol. Spectrosc. 58 (1975) 445. [30] S.T. King, Spectrochim. Acta 28A (1972) 165. [31] P.D. Godfrey, R.D. Brown, A.N. Hunter, J. Mol. Struct. 413±414 (1997) 405. [32] B. Rousseau, C. Van Alsenoy, R. Keeulers, H.O. Desseyn, J. Phys. Chem. A 102 (1998) 6540. [33] R. Keeulers, H.O. Desseyn, B. Rousseau, C. Van Alsenoy, J. Phys. Chem. A 103 (1999) 4621. [34] K. Christmann, Prog. Surf. Sci. 48 (1995) 15. [35] K. Christmann, Rep. Surf. Sci. 9 (1998) 1. [36] R.A. Olsen, G.J. Kroes, E.J. Baerends, J. Chem. Phys. 111 (1999) 11155. [37] A. G orling, S.B. Trickey, P. Gisdakis, N. R osch, in: J. Brown, P. Hofmann (Eds.), Topics in Organometallic Chemistry, vol. 4, Springer, Heidelberg, 1999, p. 109. [38] F. Jensen, Introduction to Computational Chemistry, Wiley, Chichester, 1999.
© Copyright 2026 Paperzz