Article pubs.acs.org/JPCC
L-Arabinose
Conformers Adsorption on Ruthenium Surfaces: A DFT
Study Remedios Cortese,† Dario Duca,*,† Victor Alberto Sifontes Herrera,‡ and Dmitry Yu. Murzin*,‡ †
Dipartimento di Chimica “Stanislao Cannizzaro” dell’Università di Palermo, viale delle Scienze Ed. 17, I-90128 Palermo Italy Laboratory of Industrial Chemistry, Process Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FIN-20500 Åbo/Turku, Finland
‡
ABSTRACT: Adsorption of five L-arabinose tautomers, one acyclic and four cyclic (α, β, pyranose, and furanose) species, on a ruthenium surface was studied as a precursor process of the, nowadays more and more, industrially important sugar catalytic hydrogenation on metal surfaces in water medium. The study was mostly referred to a 37 atom metal catalyst fragment even though border effects on the adsorption processes were also checked employing a 61 atom metal fragment. To figure out conformational effects on the title process, the tautomer flexibility was, at first, investigated by the genetic algorithm based code Balloon considering the conformational spaces of the different aquo tautomers. On the whole, 30 Larabinose conformers, representing the complete conformational set (of a realistic water solution), were isolated by the genetic algorithm based code Balloon. These were further refined at density functional theory (DFT) level and then were analyzed when interacting with a ruthenium surface, always at DFT level, by SIESTA. It was found that (1) cyclic L-arabinose tautomers give rise to less strong adsorption than the acyclic tautomeric form; (2) L-arabinose molecules preferentially adsorb perpendicularly to the metallic surface; (3) one among the α-pyranose and one among the β-furanose derivatives are largely the most abundant adsorbed species; (4) the dominant L-arabinopyranose and L-arabinofuranose surface configurations are clearly related to corresponding not-adsorbed species that preserve both conformations and intramolecular hydrogen bonds during their adsorption. The consideration of the points above allowed us to pick out significant properties characterizing L-arabinose adsorption on ruthenium.
■
INTRODUCTION Biomass conversion to fuels and chemicals has attracted great attention as one of the future technologies for inhibiting global warming being CO2 neutral. In addition in the very near future because of apparent difficulties with availability of fossil fuels, there is an urgent demand for using other carbon resources to synthesize chemicals without increasing the emission of CO2. Biomass is one such resource being an alternate to fossil fuels.1,2 Cellulose and hemicelluloses are typical nonfood biomasses and could serve as sources of fine chemicals. The structure of biomass-derived starting materials, for example, cellulose, is very different from that of crude oil, which is the basis of todayʼs fuels and chemicals. Biomass (1) gasification to syngas and (2) hydrolysis to sugars represent two main technological lines in biorefinery the latter being also called an entry point to biorefinery.3 Hence, separation as well as catalytic treatments of several hemicelluloses have nowadays important perspectives. An example of a chemical available on the consumer market is the anticaries and anti-inflammatory sugar-substitute xylitol stemming from glucuronoxylan and arabinoglucuronoxylan, which in turn are mainly obtained by birch trees. In this case, the hemicellulose is hydrolyzed to several monosaccharides from which xylose is separated and hydrogenated over a © 2012 American Chemical Society
heterogeneous catalyst to xylitol. The catalyst used for this purpose has up to now been Raney nickel, but supported Ru catalysts showed encouraging results.4,5 In particular, Sifontes Herrera et al.5 recently claimed that arabinose, galactose, mannose, rhamnose, and maltose could be hydrogenated to the corresponding sugar alcohols on Ru/C catalysts by a process similar to that occurring in the catalytic hydrogenation of glucose to sorbitol over nickel or nickel-promoted (by molybdenum or chromium) catalysts. This is actually a major process of the sugar-processing industry6 biased, however, by either Ni or metal-promoter (Mo, Cr) leaching. Therefore, Co, Pt, Pd, Rh, and Ru supported catalysts were tested showing that the latter catalyst is very promising7 because of the highest activity and overall because of the absence of metal leaching as shown under experimental conditions by, namely, Ru/C catalyst.1,8 Sugars to be hydrogenated mutarotate in aqueous solutions coexisting as acyclic aldoses and ketoses as well as cyclic pyranoses and furanoses. The final equilibria of these forms depend on the molecule characteristics and solvent media. Received: March 19, 2012 Revised: June 14, 2012 Published: June 18, 2012 14908
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
Figure 1. Representation of the tautomeric water equilibria and numbering of L-arabinose. Only labeling of carbon atoms is shown; oxygen labels (On) have the same number (n) characterizing the label of the carbon to which oxygen atoms are bonded.
density functional theory (DFT) approach as implemented in the code. This method employs linear combination of pseudoatomic orbitals as basis set. The atomic core is replaced by a nonlocal norm-conserving, relativistic Troullier-Martins (TM) pseudopotential that is factorized in the KleinmannBylander form. Relaxation of all the structures involved in the present study was carried out together with the electronic calculation. The generalized gradient approximations (GGA) was chosen in order to describe the exchange and correlation potential, and the employed parametrization was that of Perdew, Burke, and Ernzerhofer (PBE). All the calculations were performed considering a 450 Ry energy cutoff to define the real space grid for numerical calculations. Geometry optimizations were performed by the conjugate gradient algorithm. In vacuo testing calculations, using Gaussian 03,14 were also carried out on both opened and closed L-arabinose species as explained in the following L-Arabinose Species subsection. Ruthenium Clusters. The relativistic pseudopotential for Ru, including nonlinear core correction terms able to account for the significant overlap of the core charges with the valence d orbitals, was generated using the atomic configurations 4d7, 5s1, 5p0, and 4f0. The cutoff radii of s, p, d, and f orbitals were 2.44, 2.73, 1.42, and 2.56 au. Nonlinear core corrections with a matching radius of 1.30 au for Ru were included. The pseudopotential well reproduced the eigenvalues obtained by all-electron calculations performed, as a comparison, on a ruthenium atom, including a few excited states. Ru valence states were described by using a double-polarized (DZP) basis set. The whole Ru model was benchmarked by a comparison between experimental and computational findings, which gave satisfying results. As an example, calculated and experimental bulk cohesive energies were 6.37 and 6.63 eV, respectively, whereas the experimental lattice parameters15 were reproduced pointing out a root-mean-square deviation equal to 4%, and finally, the relaxed structure of a Ru18 cluster gave the same geometry and binding energy parameters already obtained by Aguilera-Granja et al.16 To model the Ru(0001) surface, a cluster consisting of three layers, including 37 atoms, was employed. Such a model was large enough either to accommodate the L-arabinose molecules or to perform calculations in a reasonable time. A cluster
Besides sugar mutarotation and structure sensitivity, reaction kinetics involve side reactions, namely, isomerization, hydrolysis, and oxidative dehydrogenation. Moreover, catalysts deactivate and external and internal mass transfer limitations interfere with kinetics particularly under industrial conditions. Regarding the mechanism of the most commonly studied sugar catalytic hydrogenation (D-glucose over Ru/C catalysts), two main mechanistic hypotheses are currently accepted. One proposed by Maranhão et al.9 suggests the existence of two concurrent, acidic and basic, catalytic sites; the monosaccharide molecule adsorbs on acidic sites through the CO bond while the dissociative adsorption of hydrogen occurs on basic sites.9,10 The other proposed by Creeze et al.11 assumes the formation of an ionized β-pyranose species adsorbed on Ru through the O1, O5, and O6 (the atomic numbering is adapted from that used in the caption of Figure 1) centers with the latter being able to bond by its hydrogen the anomeric carbon.11,12 Computational analysis could be decisive to understand which kind of tautomers interacts more effectively with the catalyst surface and, hence, which reaction mechanism is more likely to occur along with sugar hydrogenation on the metal catalyst. At first, sugar adsorption on Ru surfaces was investigated taking into account both the presence of the five tautomers in aqueous solution (acyclic and cyclic−α, β, pyranose, and furanose−species) and their conformational flexibility. As a model reactant, the aldopentose L-arabinose, a monosaccharide containing five carbon atoms including one aldehyde (−CHO) functional group, was selected. This analysis was performed as a preliminary investigation to study the following hydrogenation mechanism involving adsorbed tautomer species because none of these can be a priori considered more or less active or even inactive. In the next section (Models and Computational Details), methodologies and model details concerning the conformational space search as well as the conformer and surface configuration optimizations are presented while in the Results and Discussion section, findings and inferences are discussed.
■
MODELS AND COMPUTATIONAL DETAILS Calculations on the L-arabinose/ruthenium adsorbed systems (see below) were performed by SIESTA (Spanish Initiative for Electronic Simulations with Thousands of Atoms)13 using the 14909
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
Figure 2. Surface adsorption of L-arabinose on Ru surfaces: (a) CBM1, (b) CBM2, and (c) CBM3 are adsorbed cyclic derivatives, whereas (d) ABM1 and (e) ABM2 are adsorbed acyclic derivatives.
geometries and were always in agreement in pointing that all the geometries were related to minima. Moreover, the ZPE correction was not influential in stating the relative energetic order of the different isomers belonging either to closed or open species. Since the PBE/TM/DZP results very well agree with those obtained on the same systems by MP2 and B3LYP/ 6-311++G(d,p) approaches, we are confident on our choice of both exchange and correlation functional and basis sets here proposed. L-Arabinose Species/Ruthenium Cluster. Adsorption of all the L-arabinose conformers on the Ru(0001) surface was calculated using the cluster model approach. The adsorption energies were corrected for the basis set superposition error (BSSE) after the counterpoise calculation method summarized by the following equation:
consisting of 61 atoms arranged into three layers was also employed in order to check possible border effects that could affect the calculated energetic trends of the L-arabinose conformers on the smaller cluster. In both cases, the cluster upper-layer atoms, except the edge ones, were allowed to be fully optimized while the remaining atoms, including those of the upper-layer edges, were fixed at the crystallographic coordinate values. Spin-polarized calculations were also performed on the 37 atom fragment resulting in the magnetic moment per atom of 0.37 μ/atom. Considering this result, only spin-unpolarized calculations were performed on the title systems. L-Arabinose Species. The conformer selection employed in the present study was generated by Balloon, a genetic algorithm based code.17 The subsequent conformer optimizations were performed employing the TM pseudopotentials (for C, H, and O atoms) and the PBE functional of SIESTA.13 The valence states were described using standard DZP basis sets and an energy shift of 0.0005 Ry. The latter was chosen after a careful check on energy-shift values in the range 0.05−0.0001 Ry. However, literature on monosaccharides suggests the use of the B3LYP functional and diffuse basis sets18 for their calculations. Therefore, for testing the reliability of the PBE/ TM/DZP model proposed, the geometries of all the isolated Larabinose conformers were also relaxed using both MP2 and B3LYP models as implemented in the Gaussian 03 suite of programs14 with the 6-311++G(d,p) basis sets. In this case, vibration frequency analyses within the harmonic approximation were accomplished on the optimized geometries to check whether they actually described minima in the potential energy surface,19 while the zero point energy (ZPE) correction was applied in the energetic evaluations. Geometry and energy parameters obtained by B3LYP/6-311++G(d,p) and MP2/6311++G(d,p) models produced, irrespective of the L-arabinose species considered, very similar, if not coincident, molecular
AB AB AB AB ΔEads = [EAB (AB) − EAB (A) − EAB (B)]
+ [EAA B(A) − EAA (A)] + [EAB B(B) − E BB(B)] (1)
where A and B are the reactants and AB is the product. The subscripts refer to the geometry of the homonym fragment, the superscripts refer to the basis set employed, and finally, the terms in-between the round brackets refer to the chemical species (reactants or product) involved.
■
RESULTS AND DISCUSSION Figure 1 shows the L-arabinose tautomeric forms found in aqueous solution. All the tautomers are in equilibrium, and each of them presents conformational flexibility. The goal of this study was to individuate different conformers for each tautomers and to analyze their structural and energetic behavior in the adsorption processes occurring on a Ru(0001) surface model. We did this to understand if some tautomers are 14910
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
three-atom rings puckered either above or below the plane individuated by the four-atom ring. Of course, each three-atom ring has just one atom (puckered above or below the plane) not belonging to the four-atom ring. Following the usual literature notation employed to identify tautomer subfamilies,21 one atom puckered above the plane is, from now on, reported as a superscript number on the left of the class-identifying letters (C, B, S), while one atom puckered below the plane is reported as a subscript number on the right side of the same letters. When both the puckered atoms are below or above the fouratom ring plane, the corresponding label is placed as a subscript or as a superscript of the class-identifying letter, respectively. Using this notation, the α and β L-arabinopyranose conformers, both adsorbed (CBM1AP, CBM2AP, CBM3AP) and notadsorbed (AP), belonging to the different subfamilies are reported in Table 1. Certain surface interactions lead to conformational changes in the L-arabinopyranose derivatives.
disallowed or are conversely favored along adsorption processes. The present study is, in fact, preliminary to the investigation of the elementary steps involved in the hydrogenation of Larabinose for which adsorption is actually a prerequisite event. Because adsorption could be affected by the orientation of the adsorbate, among the possible interactions of the monosaccharide with the metallic surface, three different ways of binding per conformer in pyranose and furanose forms were investigated and two in the case of the acyclic form. The adsorption orientations above were chosen in order to include in the study just activated binding modes. With this expression, L-arabinose surface configurations giving rise to the activation either of the hemiacetalic bonds (in the cyclic tautomers) or of the double CO bond (in the aldehydic tautomers) are designated. The binding modes characterizing the adsorption of the cyclic (CBM1, CBM2, CBM3) and acyclic (ABM1, ABM2) tautomers are reported in Figure 2. In CBM1, the cyclic form of the L-arabinose is parallel to the ruthenium surface and the exposed face of the monosaccharide taking as reference the α pyranose, which contains the anomeric hydroxyl. In CBM2, L-arabinose is still parallel to the surface, but its exposed face is the opposite with respect to that of CBM1. At variance with the first two, in CBM3, Larabinose is perpendicular to the Ru surface. ABM1 and ABM2 systems are, finally, representative of mono- and dihapto interactions of acyclic L-arabinose with the ruthenium surface; these, respectively, occur through the oxygen atom and the unsaturated moiety of the aldehydic group. In the latter case, the L-arabinose interacting group and the metallic plane are parallel to each other. Taking into consideration the systems above, it was possible to gain information on: • the adsorption energy of the different conformers on the ruthenium surface, • the energetic order of the adsorbed tautomeric systems, • the effects of the tautomer binding modes on the energetic order above. To this end, because macroscopic properties of a given species populating conformational spaces can be evaluated as weighted averages of the corresponding atomistic properties of the single conformers,20 the Boltzmann population of the adsorbed tautomers, to be utilized as averaging kernel, was calculated employing the following equation: Ni e−Ei / RT = N −E / RT Ntotal ∑k =total1 e k
Table 1. Not-Adsorbed (AP) and Adsorbed (CBM1AP, CBM2AP, CBM3AP) L-Arabinopyranose Conformers and Corresponding Subfamilies to Which They Belong conformers L-A-α-p1 L-A-α-p2 L-A-α-p3 L-A-α-p4 L-A-β-p1 L-A-β-p2 L-A-β-p3 L-A-β-p4
AP 1
C4 C4 1 S3 B3,O 1 C4 1 C4 3,O B 1 S5 1
CBM1APa 1
C4 C4 1 S3 1 S3 1 C4 1 C4 3,O B 1 S5 1
CBM2APa 1
C4 C4 3 S5 5 S3 1 C4 1 C4 3,O B 1 S3 1
CBM3APa 1
C4 C4 1 S3 5 S3 1 C4 1 C4 3 S1 1 S5 1
a
CBM1, CBM2, and CBM3 correspond to the three different adsorption ways of L-arabinopyranose on the ruthenium surface.
The calculated energetic trend for the eight conformers is in agreement with the stability order usually found for the three conformer classes: chair > skew-boat > boat.21 Figure 3 shows the adsorption effects on the conformer energetic distribution. It is evident that either in the CBM1AP or in the CBM2AP sets the interactions with the ruthenium surface restrain the energy gaps characterizing the conformers of the same sets. On the contrary, in the CBM3AP set, there is a large energy gap with
(2)
where Ni is the number of surface configurations of kind i at the temperature T, N total is the total number of surface configurations characterizing the surface conformer ensemble (at the same temperature), and finally, Ei and Ek are the energies of the ith and kth adsorbed conformer. RT for sake of simplicity was set equal to 1. Boltzmann population was used to calculate the weighted adsorption energies and the geometrical descriptors of the five adsorbed tautomers. Pyranose Conformers. Balloon17 generated four conformers both for the α (L-A-α-p 1, 2, 3, 4) and β (L-A-β-p 1, 2, 3, 4) pyranose form of L-arabinose. The geometries of the conformers were afterward optimized as described in the Models and Computational Details section. Depending on the shape of their rings, they can be classified into three main families (classes): chair (C), boat (B), and skew-boat (S). Every class can be visualized as having one four-atom ring and two
Figure 3. Stability trends for the not-adsorbed (AP) and adsorbed (CBM1AP, CBM2AP, CBM3AP) L-arabinopyranose conformers. 14911
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
values and (2) L-arabinopyranose species have larger probability to interact via the CBM3 binding mode. Table 3 reports the Ru−O5 and C1−O5 averaged distances in the following used as geometrical descriptors, which
one conformer being much more stable than the others. This effect is outlined by Figure 4 where the statistical weights of the L-arabinopyranose conformers are reported.
Table 3. Averaged Distances Employed as Geometrical Descriptor: Ru−O5 and C1−O5 Characterizing the Adsorbed L-Arabinopyranose C-Sets distance/(pm) C-sets
Ru−O5
C1−O5
APa CBM1AP CBM2AP CBM3AP
281 422 252
141 144 143 146
a
The emiacetalyc averaged bond of the not-adsorbed L-arabinopyranose species (AP) is reported as reference.
characterize the different C-sets. The former distance, in fact, displays the interaction ability shown by a given conformer with the ruthenium surface. In respect to this, in principle, the shorter is the bond the more effective is the interaction. The latter is related to the hemiacetal bond activation of the conformer needing eventually a further hydrogenation on the same position. Moreover, both the Ru−O5 and C1−O5 averaged distances are straightforwardly related to the corresponding ⟨ΔEads⟩ values (see Table 2). On the whole, the results reported in Table 2 and Table 3 are coherent showing larger adsorption energy for shorter Ru−O5 distances. In detail, the Ru−O5 distances of CBM1AP and CBM3AP are quite similar and shorter than that of the CBM2AP set. In this, although an interaction occurs between one hydroxyl group of the L-arabinopyranose and the ruthenium particle, the hemiacetalic oxygen is pushed away from the metal surface because of an unfavorable orientation of the same oxygen. Analyzing the last column of Table 3, it is possible to figure out, irrespective of the considered C-sets, a slight increase of the C1−O5 distance. This proves the activation of the hemiacetal moiety when the L-arabinopyranose conformers interact with the ruthenium surface. From the energetic and structural results summarized in Table 2 and Table 3, the most pronounced effects (i.e., the largest hemiacetal bond activation) occur for the CBM3 adsorption mode. Furanose Conformers. For the α and β L-arabinofuranose tautomers, Balloon isolated six and five conformers, respectively. As for the pyranose species, all the generated structures were afterward optimized originating two main conformer families distinguished for their shapes, namely, envelope (E) and twist (T). Four atoms, among the five defining the furanose structure, are coplanar in the envelope conformation with the remaining one out of the plane. In the twist conformation, conversely, three of the five ring atoms define a plane, while of the remaining two, one is puckered above and the other is puckered below the plane. Table 4 shows the not-adsorbed and adsorbed L-arabinofuranose conformers as well as the corresponding subfamilies to which they belong. With respect to the subfamily notation, for the envelope structures, the superscript is referred to the atom out of the plane, whereas for the twist ones, superscript and subscript refer to the atom above and below the plane, respectively. As shown in Table 4, also for the L arabinofuranose tautomers, conformational changes can occur along with surface adsorption. From Table 4, it results that the
Figure 4. Abundance of the L-arabinopyranose tautomers calculated according to eq 2: not-adsorbed (AP) and adsorbed (CBM1AP, CBM2AP, CBM3AP) sets.
Of course, in the CBM1 AP and CBM2 AP sets, the macroscopic properties are dependent on a larger number of conformers with respect to that characterizing the CBM1AP set. Therefore, as shown by Table 2, it is not a sound Table 2. L-Arabinopyranose/Ru: BSSE Corrected Adsorption Energy, ΔEads, Values Characterizing the Different Conformers in the Different Sets and Weighted Adsorption Energy, ⟨ΔEads⟩, Values Calculated in the Different C-Sets ΔEads/(kJ/mol) conformers
CBM1AP
L-A-α-p1
−17.4 −2.1 −19.9 −25.2 −16.3 −16.1 −13.5 −26.5
L-A-α-p2 L-A-α-p3 L-A-α-p4 L-A-β-p1 L-A-β-p2 L-A-β-p3 L-A-β-p4
N 4 ∑i = 1 N i (ΔEads)L ‐ A ‐ (α ∨ β) ‐ p a,b i total
CBM2AP
CBM3AP
−4.0 −21.9 −2.5 −30.1 −9.1 −8.0 −17.5 −21.6 −4.3 −19.0 −32.4 −18.9 −29.1 −0.4 13.3 −20.1 ⟨ΔEads⟩/(kJ/mol) CBM1AP
CBM2AP
CBM3AP
−19.0
−11.7
−30.1
a
The averaging kernel is a function of the calculated conformer energies, which are not reported in the table. bSubscript notation L-A(α ∨ β)-pi highlights that the conformer ΔEads summation is extended either on α or β L-arabinopyranose ith species.
approximation to take into account only the most stable conformer in the first two cases to evaluate the macroscopic properties. In this table, the BSSE corrected adsorption energies, ΔEads, of the different conformers and the weighted adsorption energies, ⟨Eads⟩, of the different conformer sets, Csets, are reported. Looking at Table 2, it can be observed that (1) energies range in-between physisorption and chemisorption 14912
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
Table 4. Not-Adsorbed (AF) and Adsorbed (CBM1AF, CBM2AF, CBM3AF) L-Arabinofuranose Conformers and Corresponding Subfamilies to Which They Belong conformers
CBM1AFa
AF
CBM2AFa
CBM3AFa
L-A-α-f1
4
4
4
4
3
E E 3 E
4
L-A-α-f2
3
4
L-A-α-f3
E E 3 E
E E
2 3T
E E 3 E
3
E
4
4
2 3T
3
3
4 5T
2
3
E
3
E
L-A-β-f1
2
E
3
3 4T
2
E
L-A-β-f2
2
2
2
L-A-β-f3
E E 4 E 4 E
4
4
4
3
4
L-A-α-f4
4
L-A-α-f5
3
L-A-α-f6
L-A-β-f4 L-A-β-f5
E E
E E
E E
E E 2 E
E E 2 E 4 E
2 3T
E E
E E 3 E 4 E
Figure 5. Stability trends for the not-adsorbed (AF) and adsorbed (CBM1AF, CBM2AF, CBM3AF) L-arabinofuranose conformers.
a CBM1, CBM2, and CBM3 correspond to the three different adsorption ways of L-arabinofuranose on the ruthenium surface.
envelope conformation ensemble is more populated than the twist ensemble suggesting a larger presence of the correlated set conformers. Table 5 shows that the 11 furanose conformers present different adsorption energies. The averaged adsorption Table 5. L-Arabinofuranose/Ru: BSSE Corrected Adsorption Energy, ΔEads, Values Characterizing the Different Conformers in the Different Sets and Weighted Adsorption Energy, ⟨ΔEads⟩, Values Calculated in the Different C-Sets ΔEads/(kJ/mol) conformers
CBM1AF
L-A-α-f1
−15.7 −27.2 −23.7 −14.4 −40.4 2.5 −7.3 −19.5 −18.2 −32.4 −24.2
L-A-α-f2 L-A-α-f3 L-A-α-f4 L-A-α-f5 L-A-α-f6 L-A-β-f1 L-A-β-f2 L-A-β-f3 L-A-β-f4 L-A-β-f5
N n ∑i = 1 N i (ΔEads)L ‐ A ‐ (α ∨ β) ‐ fi a total
CBM2AF
CBM3AF
−11.0 −23.3 −2.9 −12.1 −12.6 −13.9 −5.8 −10.8 −22.7 −16.3 −24.6 −11.2 −9.1 −7.2 −5.9 −23.3 −19.4 −26.0 −24.6 −33.9 −16.4 −22.5 ⟨ΔEads⟩/(kJ/mol) CBM1AF
CBM2AF
CBM3AF
−24.2
−24.6
−22.5
Figure 6. Abundance of the L-arabinofuranose tautomers calculated according to eq 2: not-adsorbed (AF) and adsorbed (CBM1AF, CBM2AF, CBM3AF) sets.
ruthenium surface should induce conformational changes, which perturb the conformational space populations. Table 6 reports the Ru−O5 and C1−O5 averaged distances, which characterize not-adsorbed (AF) and adsorbed (CBM1AF, CBM2AF, CBM3AF) C-sets. Both in the CBM1 and CBM2 adsorption modes, the Ru−O5 distance is quite large and, as observed for the CBM2AP systems, this is due to an interaction of one L-arabinose hydroxyl group with the ruthenium surface and to an unfavorable placement of the hemiacetal oxygen.
Subscript notation L-A-(α ∨ β)-fi highlights that the conformer ΔEads summation is extended either on α or β L-arabinofuranose ith species, n = 6 ∨ 5 for (α ∨ β) tautomers, respectively. a
Table 6. Averaged Distances Employed as Geometrical Descriptor: Ru−O5 and C1−O5 Characterizing the Adsorbed L-Arabinofuranose C-Sets distance/(pm)
energies, ⟨ΔEads⟩, are however pretty similar suggesting that there is not a preferential binding mode. Conversely, Figure 5 and Figure 6 together point out one dominant surface configuration per adsorption mode. An analogous behavior was already observed in the CBM3 AP conformer set. Interestingly, the most abundant conformer species change when considering not-adsorbed and adsorbed L-arabinofuranose derivatives. Because of this, during adsorption, the
C-sets
Ru−O5
C1−O5
431 397 231
141 142 142 146
a
AF CBM1AF CBM2AF CBM3AF a
The emiacetalyc averaged bond of the not-adsorbed L-arabinofuranose species (AF) is reported as reference.
14913
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
Figure 7. Acyclic L-arabinose trihapto adsorption modes: (a) ABM3 and (b) ABM4 configurations.
Table 7. Significant Averaged Distances and Angles of Not-Adsorbed (AC) and Adsorbed (ABM1, ABM2, ABM3, ABM4) LArabinose Acyclic C-Sets distance/(pm) C-sets AC ABM1 ABM2 ABM3 ABM4
Ru−O1 317 207 212 210
Ru−C1
211 210 210
angle/(deg)
Ru−O2
C1−O1
232
124 124 139 143 141
Ru−O1−Ru
74.7
H1−C1−O1 121.7 121.6 111.2 108.9 111.7
example of the corresponding adsorption modes characterizing the ABM3 and ABM4 C-sets is shown in Figure 7. The ABM3 configurations show trihapto adsorptions in which the carbon of the CO moiety interacts with one ruthenium atom while the oxygen is bridged between two other ruthenium atoms. The trihapto interactions of the ABM4 configurations conversely involve O2 and both the atoms of the CO moiety. ABM3 and ABM4 ⟨ΔEads⟩ values were −35.0 and −27.1 kJ/mol, respectively. Table 7 shows that the Ru−O1 averaged bond distance characterizing the ABM1 adsorption is larger with respect to those of the other adsorption modes. Because of this, it is possible to hypothesize that the ABM1 configurations do not allow close interactions with the ruthenium surface. Hence, the activation of the C1−O1 double bond, necessary for hydrogenating the CO moiety, should not take place. This fact is confirmed both by the ABM1 C1−O1 distance and the H1− C1−O1 angle, which are coincident to the homonym geometrical descriptor of the not-adsorbed AC conformers. Regarding all the other adsorption modes, there are slight differences in the Ru−O1 and Ru−C1 distances. However, when analyzing the C1−O1 and H1−C1−O1 parameter behaviors, the ABM3 adsorption mode seems more effective in activating the CO double bond. In fact, the corresponding C1−O1 distance is very close to that of a single bond, the value of the H1−C1−O1 angle to that of an sp3 carbon atom, and finally, the averaged energy to that characterizing chemisorption.
Therefore, even if the corresponding conformers interact with the metal surface, the hemiacetalic bond activation does not take place. On the contrary, interaction points between the Larabinofuranose tautomers and the metal surface occur in the CBM3 adsorption mode giving rise to the activation of the hemiacetalic bond. This is confirmed by the increased C1−O5 distance of the CBM3AF C-sets. Also for the furanose derivatives, adsorption energies are inbetween those characterizing physisorption and chemisorption processes with that characterized by the CBM3 adsorption mode the most favored to activate the L-arabinofuranose hemiacetalic bond. Acyclic Conformers. Balloon isolated 20 acyclic Larabinose conformers. To have a sample that was both significant and manageable at ab initio level, 11 representative conformers were selected. The discrete presence of the CO moiety mostly ruled the interactions of the L-arabinose acyclic conformers with the ruthenium surface. Two adsorption C-sets, each formed by 11 members, were initially isolated: ABM1 and ABM2. These are characterized by monohapto and dihapto interactions, respectively. All the conformers of the ABM1 set showed a scarce interaction with the surface as confirmed by the ⟨ΔEads⟩ value calculated on the corresponding C-set (−5.6 kJ/mol). The average parameter above was on the contrary quite large for the ABM2 set (−39.8 kJ/mol). This value is obtained considering just three residual conformers because several ABM2 configurations, when relaxed, originated two trihapto surface derivative sets both formed by four conformers, which in the following are indicated as ABM3 and ABM4. An 14914
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
Figure 8. L-Arabinose most probable adsorbed configurations: (a) L-A-α-p2 and (b) L-A-β-f5 surface conformers.
Table 8. Summarizing Information on Relevant Features of the Most Probable L-Arabinose/Ruthenium Surface Configurations: L-A-α-p2@CBM3AP and L-A-β-f5@CBM3AF distance/(pm)
other information
conformersa
Ru−O5
C1−O5
O1−H1
ΔEads/(kJ/mol)
%b
L-A-α-p2
2.52 2.43
1.46∥1.41 1.46∥1.45
1.02∥0.98 1.01∥0.98
−30.1 −22.5
99.5∥98.9 0.5∥0.12
L-A-β-f5
subfamily C4∥1C4 4 E∥4E
1
a
When existing, the analogous parameter values of the corresponding not-adsorbed conformers (either belonging to the AP or to the AF C-sets) are reported at the right of the double vertibar. bFor the not-adsorbed L-arabinose species %, reported at the right of the double vertibar, the missing (complementary) 0.98% has to be related to other conformers not explicitly considered.
The Most Abundant L-Arabinose Species. Besides the analysis of the not-adsorbed and adsorbed tautomers potentially involved in aqueous equilibria, it is important to state which surface configurations are the most favored. Therefore, the pyranose, furanose, and acyclic adsorbed species (on the whole, 79 surface configurations) were weighted and were regarded as members of just one ensemble. Even if the acyclic conformers generally exhibit larger adsorption energies with respect to pyranose and furanose derivatives, the L-A-α-p2 and L-A-β-f5 CBM3 adsorbed species (see Figure 8) display abundance percentages of 99.5 and 0.5, respectively. Hence, the remaining surface configurations are present in negligible amounts. This allows us to infer that acyclic tautomers should not react on ruthenium surfaces. Adsorbed L-A-α-p2 and L-A-β-f5 CBM3 conformers exhibit stabilizing intramolecular hydrogen bonds, which were still present in the not-adsorbed conformers. Table 8 reports some information on the dominant surface configurations and, for comparison, analogous details on the corresponding notadsorbed conformers. Table 8 shows that in both the favored species the hemiacetalic oxygen is close to the ruthenium surface assuring strong L-arabinose−ruthenium interactions, while the elongated hemiacetalic bond agrees with the occurrence of a corresponding bond activation. The small difference found in the C1−O5 bond distance characterizing the not-adsorbed and adsorbed furanose L-A-β-f5 conformers can be easily explained by recalling the presence of the strong intramolecular hydrogen bond, by determining a six-term ring, and by activating the hemiacetalic bond even in the not-adsorbed conformer. The adsorption energies of both the surface conformers, reported in
Table 8, are typical of medium-strong interacting configurations. Noticeably, the subfamilies of the not-adsorbed and adsorbed homonym configurations are unchanged. The bond distance elongations corresponding to the anomeric O1−H1 bonds of both the pyranose and furanose adsorbed species is pretty interesting. In fact, these elongations are coherent with the presence of interactions occurring between the anomeric hydroxyl and the ruthenium surface. This finding matches reasonably well with the inferences of Makkee et al.12 and Creeze et al.11 that suggest, analyzing the D-glucose hydrogenation, that the adsorption of β-pyranose is more favorable through the coordination of O1, O5, and O6 centers and with the interpratation of Beenackers et al.22 that suggested that ionization of the anomeric hydroxyl group of D-glucose, as an example taking place via the formation of a glucose anion, is a basis step either for isomerization or for epimerization processes. Surface Border Effects on the Ruthenium Cluster Properties. To check the occurrence of border effects potentially influencing the energetic trends found in the present investigation, a 61-atom model for the ruthenium cluster was also considered. This basically included three crowns of ruthenium atoms maintaining the ruthenium crystallographic arrangement and surrounding the three planes of the 37-atom model. Calculations were carried out on interacting α-pyranose and β-furanose derivatives because they resulted in the most abundant adsorbed species (with a relative population of 99.5 and 0.5%) on the smaller system. The population analysis on these species, performed by the already described Boltzmann approach, showed that the α-pyranose and β-furanose derivatives on the larger cluster are charac14915
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916
The Journal of Physical Chemistry C
Article
(5) Sifontes Herrera, V. A.; Oladele, O.; Kordas, K.; Eränen, K.; Mikkola, J.-P.; Murzin, D. Yu.; Salmi, T. J. Chem. Tech. Biotechnol. 2011, 86, 658−668. (6) Turek, F.; Chakrabarti, R. K.; Lange, R.; Geike, R.; Flock, W. Chem. Eng. Sci. 1983, 38, 275−283. (7) Gallezot, P.; Nicolaus, N.; Flèche, G.; Fuertes, P.; Perrard, A. J. Catal. 1998, 180, 51−55. (8) van Gorpa, K.; Boermana, C. V.; Cavenaghib, E.; Berbena, P. H. Catal. Today 1999, 52, 349−361. (9) Maranhão, L. C. A.; Sales, F. G.; Pereira, J. A. F. R.; Abreu, C. A. M. React. Kinet. Catal. Lett. 2004, 81, 169−175. (10) Martins Castoldi, M. C.; Camara, L. D. T.; Monteiro, R. S.; Constantino, A. M.; Camacho, L.; Walkimar de, M.; Carneiro, J.; Aranda, D. A. G. React. Kinet. Catal. Lett. 2007, 91, 341−352. (11) Creeze, E.; Hoffer, B. W.; Berger, R. J.; Makkee, M.; Kapteijn, F.; Moulijn, J. Appl. Catal. A: Gen. 2003, 251, 1−17. (12) Makkee, M.; Kieboom, A. P. G.; Van Bekkum, H. Carbohydr. Res. 1985, 138, 225−236. (13) Soler, J.; Artacho, E.; Gale, J.; García, A.; Junquera, J.; Ordejón, P.; Sánchez-Portal, D. J. Phys.: Condens. Matter 2002, 14, 2745−2779. (14) Frisch, M. J. et al. Gaussian 03, revision D.02; Gaussian, Inc.: Wallingford, CT, 2005. (15) Chelikowsky, J. R.; Chan, C. T.; Louie, S. G. Phys. Rev. B 1986, 34, 6656−6661. (16) Aguilera-Granja, F.; Balbás, L. C.; Vega, A. J. Phys. Chem. A 2009, 113, 13483−13491. (17) Vainio, M. J.; Johnson, M. S. J. Chem. Inf. Model. 2007, 95, 2462−2474. (18) Guler, L. P.; Yu, Y.; Kenttämaa, H. I. J. Phys. Chem. A 2002, 106, 6754−6764. (19) Foresman, J. B.; Frisch, Æ. Exploring Chemistry with Electronic Structure Methods, 2nd ed.; Gaussian Inc.: Pittsburgh, PA, 1996; pp 61−90. (20) (a) Duca, D.; Bifulco, G.; Barone, G.; Casapullo, A.; Fontana, A. J. Chem. Inf. Comput. Sci. 2004, 44, 1024−1030. (b) Giuffrida, S.; Barone, G.; Duca, D. J. Chem. Inf. Model. 2009, 2, 1223−1233. (21) Rao, V. S. R.; Qasba, P. K.; Balaji, P. V.; Chandrasekaran, R. Conformations of Carbohydrates; Harwood Academic Publishers: Amsterdam, 1998. (22) Beenackers, J. A. W. M.; Kuster, B. F. M.; Van der Baan, H. S. Carbohydr. Res. 1985, 140, 169−183.
terized by a population equal to 98.0% and 2.0%, respectively. Considering the slight population changes corresponding to the almost doubling of the cluster size, it is possible to state that there are not important effects on the whole discussion performed in this paper on the adsorption properties of Larabinose on ruthenium clusters because of the increase of the size of the ruthenium fragment.
■
CONCLUSIONS Adsorption of the L-arabinose tautomers on ruthenium was modeled as a precursor process of the catalytic sugar hydrogenation in water. The main concluding points are reported in the following: • physisorption and chemisorption energy values characterize both cyclic and acyclic L-arabinose tautomer surface interactions; • L-arabinose molecules preferentially adsorb perpendicularly to the metallic surface showing stabilizing interactions between sugar anomeric hydroxyl moieties and ruthenium centers; • although dihapto acyclic conformers show the largest adsorption energies, one among the α-pyranose and one among the β-furanose derivatives are largely the most abundant adsorbed species; • the conformer family of a not-adsorbed species can change during adsorption; however, the dominant adsorbed L-arabinopyranose and L-arabinofuranose configurations above are related to the corresponding notadsorbed species belonging to 1C4 and 4E conformer families, which preserve both conformation and intramolecular hydrogen bond during adsorption. These points that should regard the first instants involved in the sugar hydrogenation processes, those in which hydrogen is still not interacting with the catalyst surface, allow us to infer that L-arabinose conformers are mostly adsorbed keeping the conformational properties and the intramolecular hydrogen bonds of the not-adsorbed species with the α-pyranose tautomer perpendicularly interacting with the metal surface being the one perhaps hydrogenated.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected] (D.D.); dmurzin@abo.fi (D.Yu.M). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS R.C. acknowledges the support of the Åbo Akademi University for providing the Johan Gadolin Scholarship that enabled her to work at the Åbo Akademi Process Chemistry Centre and the CSC − IT Center for Science Ltd. of Helsinki where the largest part of the calculations was carried out.
■
REFERENCES
(1) Gallezot, P. Catal. Today 2007, 121, 76−91. (2) Gallezot, P. Catal. Today 2011, 167, 31−36. (3) Rinaldi, R.; Schüth, F. ChemSusChem 2009, 49, 1096−1107. (4) Kuusisto, J.; Tokarev, A. V.; Murzina, E. V.; Roslund, M.; Mikkola, J.-P.; Murzin, D. Yu.; Salmi, T. Catal. Today 2007, 121, 92− 99. 14916
dx.doi.org/10.1021/jp3026336 | J. Phys. Chem. C 2012, 116, 14908−14916