Theoretical Study on Adsorption and Dissociation of NO2 Molecule on

pubs.acs.org/Langmuir © 2010 American Chemical Society

Theoretical Study on Adsorption and Dissociation of NO2 Molecule on Fe(111) Surface Hui-Lung Chen,*,† Shiuan-Yau Wu,‡ Hsin-Tsung Chen,^ Jee-Gong Chang,^ Shin-Pon Ju,§ Chiitang Tsai,† and Ling-Chieh Hsu‡ †

Department of Chemistry and Institute of Applied Chemistry, Chinese Culture University, Taipei 111, Taiwan, ‡ Department of Chemistry, National Taiwan Normal University, 88, Section 4, Tingchow Road, Taipei 116, Taiwan, ^National Center for High-Performance Computing, No. 28, Nan-Ke Third Road, Hsin-Shi, Tainan 74147, Taiwan, and §Department of Mechanical and Electro-Mechanical Engineering, Center for Nanoscience and Nanotechnology, National Sun-Yat-Sen University, Kaohsiung 80424, Taiwan Received November 9, 2009. Revised Manuscript Received January 25, 2010

We applied periodic density-functional theory (DFT) to investigate the adsorption and dissociation of NO2 on a Fe(111) surface. The most favorable adsorption configuration of NO2/Fe(111) is the FeNO2(S-μ3-N,O,O0 ) configuration with NO2 at the 3-fold-shallow site of the surface, which has an adsorption energy -64.59 kcal/mol. Of two geometries of NO2/Fe(111) for the stepwise NO2 deoxygenation, one is the most stable structure, FeNO2(S-μ3-N,O,O0 ), with activation barriers 10.38 and 19.36 kcal/mol to break the first (ON-O bond activation) and second (N-O bond activation) nitrogen-oxygen bonds, respectively; another configuration FeNO2(B-μ2-N,O) has a smaller energy barrier (3.88 kcal/mol) to break the first ON-O bond. All these findings show that NO2 can readily decompose on the Fe(111) surface. The rate constants for the two aforementioned processes were also predicted by VTST and RRKM theory, and the predicted total rate constants, ktotal (in units of cm3 molecule-1 s-1), can be represented by the equations ktotal = 5.61  10-5T-2.060 exp(-0.639 kcal mol-1/RT) at T = 100-1000 K. To acquire insight into the great catalytic activity of the Fe(111) surface for the decomposition of NO2, the nature of the interaction between the adsorbate and the substrate is subjected to a detailed electronic analysis.

1. Introduction Atmospheric nitrogen oxides (NOx) play an important role in photochemical smog, the formation of acid-rain precursors, the destruction of ozone in the stratosphere, and possibly greenhouse warming.1 For this reason, the interaction of NOx with transitionmetal surfaces becomes an important topic, especially in catalytic conversion of exhaust gases from motor vehicles. Reactions of NO2 on metal surfaces have been the subject of numerous investigations.2 The adsorption and decomposition of NO2 on *Corresponding authors: e-mail [email protected]; Tel þ886-228610511 ext 25313; Fax þ886-2-28614212.

(1) (a) IPCC Intergovernmental Panel on Climate Change, Climate Change 2001: The Scientific Basis, Technical Summary. UNEP, WMO, 2001. (b) Inglezaks, V. J.; Poulopoulos, S. G. Adsorption, Ion Exchange and Catalysis Design of Operations and Environmental Applications, 1st ed.; Elsevier: Amsterdam, The Netherlands; 2006; Chapter 1, pp 1-30. (2) (a) Taylor, K. C. In Catalysis Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer: Berlin, 1984; Vol. 5, p 119. (b) Schmatloch, V.; Kruse, N. Surf. Sci. 1992, 270, 488. (c) Brown, W. A.; King, D. A. J. Phys. Chem. B 2000, 104, 2578. (d) Tang, H.; Trout, B. L. J. Phys. Chem. B 2005, 109, 17630. (e) Xu, S. C.; Irle, S.; Musaev, D. G.; Lin, M. C. J. Phys. Chem. B 2006, 110, 21135. (f) Hellman, A.; Panas, I.; Gr€onbeck, H. J. Chem. Phys. 2008, 128, 104704. (3) Segner, J.; Vielhaber, W.; Ertl, G. Isr. J. Chem. 1982, 22, 375. (4) Dahlgren, D.; Hemminger, J. C. Surf. Sci. 1982, 123, L739. (5) Bartram, M. E.; Windham, R. G.; Koel, B. E. Surf. Sci. 1987, 184, 57. (6) Bartram, M. E.; Windham, R. G.; Koel, B. E. Langmuir 1988, 4, 240. (7) Schwalke, U.; Parmeter, J. E.; Weinberg, W. H. J. Chem. Phys. 1986, 84, 4036. (8) Schwalke, U.; Parmeter, J. E.; Weinberg, W. H. Surf. Sci. 1986, 178, 625. (9) Jirsak, T.; Dvorak, J.; Rodriguez, J. A. Surf. Sci. 1999, 436, L683. (10) Bare, S. R.; Griffiths, K.; Lennard, W. N.; Tang, H. T. Surf. Sci. 1995, 342, 185. (11) (a) Polzonetti, G.; Alnot, P.; Brundle, C. R. Surf. Sci. 1990, 238, 226. (b) Polzonetti, G.; Alnot, P.; Brundle, C. R. Surf. Sci. 1990, 238, 237. (12) Brown, W. A.; Gardner, P.; King, D. A. Surf. Sci. 1995, 330, 41. (13) Banse, B. A.; Koel, B. E. Surf. Sci. 1990, 232, 275. (14) Wickham, D. T.; Banse, B. A.; Koel, B. E. Surf. Sci. 1991, 243, 83. (15) Bartram, M. E.; Koel, B. E. Surf. Sci. 1989, 213, 137.

Langmuir 2010, 26(10), 7157–7164

Pt(111),3-6 Ru(001),7,8 Rh(111),9 Ag(111),10-12 Pd(111),13,14 Au(111),15-17 and polycrystalline Au18 have been extensively studied using various techniques such as TPD, LEED, EELS, HREELS, XPS, and UPS. This work has demonstrated that the NO2 adsorbs dissociatively on Rh(111), Pd(111), Pt(111), Ru(001), and Ag(111) surfaces at low temperature but molecularly on Au(111) and polycrystalline Au. According to these experiments, NO2 coordinates to metal centers in the surface through either one (monodentate) or both (bidentate in a chelating structure) oxygen atoms or through the central N atom or simultaneously with N and one O atoms. Most preceding work was performed on crystals of large surface area or body-centered-cubic (bcc) structure or single crystals.19 Bzo
nski et al. undertook calculations of properties of bcc iron surfaces, such as structural energetics, surface relaxation, and magnetic properties;20 the sequence of stability of the investigated surfaces was (110)>(100)>(111). The (111) facet on a small crystallite of Fe is, however, thought to have a large catalytic activity due to the open surface structure,21 and this unique crystal face possess the greatest rate of turnover.22 No calculation of the mechanism of decomposition of NO2 on the Fe(111) surface is reported. Here we report our findings with periodic densityfunctional theory (DFT) for the adsorption and dissociation behaviors of NO2 on the surface of Fe(111). This understanding (16) (a) Wang, J.; Koel, B. E. J. Phys. Chem. A 1998, 102, 8573. (b) Wang, J.; Voss, M. R.; Busse, H.; Koel, B. E. J. Phys. Chem. B 1998, 102, 4693. (17) Beckendorf, M.; Katter, U. J.; Schlienz, H.; Freund, H. J. J. Phys. Chem. 1993, 5, 5471. (18) Wickham, D. T.; Banse, B. A.; Koel, B. E. Catal. Lett. 1990, 6, 163. (19) Ertl, G. Angew. Chem., Int. Ed. Engl. 1986, 6, 558. (20) Bzo
nski, P.; Kiejna, A. Surf. Sci. 2007, 601, 123. (21) (a) Spence, N. D.; Schoonmaker, R. C.; Somorjai, G. A. J. Catal. 1982, 74, 129. (b) Chen, H.-L.; Chen, H.-T.; Ho, J.-J. Langmuir 2010, 26, 775. (c) Li, H.-J.; Ho, J.-J. J. Phys. Chem. C 2010, 114, 1194. (22) Strongin, D. R.; Somorjai, G. A. J. Catal. 1988, 109, 51.

Published on Web 02/10/2010

DOI: 10.1021/la904233b

7157

Article

is significant for the future rational design of a surface catalytic model for the decomposition of toxic gaseous NO2.

2. Computational Methods All present calculations are performed with the DFT planewave method utilizing the Vienna ab initio simulation package (VASP).23-27 We use the projector-augmented- wave method (PAW)28,29 in conjunction with revised Perdew-BurkeErnzerhof (rPBE)30,31 density functionals. The Brillouin zone is sampled with the Monkhorst-Pack grid.32 The calculations were performed with the (4  4  4) and (4  4  1) Monkhorst-Pack mesh k-points for bulk and surface calculations, respectively. A 400 eV cutoff energy, which allows convergence to 1  10-4 eV in total energy, is used. All calculations were performed using the spin-polarization method to describe properly the magnetic property of the Fe(111) surface model. The p(2  2) lateral cell of Fe(111) surface is modeled as periodically repeated slabs with six layers. The bottom three atomic layers were kept frozen and set to the estimated bulk parameters, whereas the remaining layers were fully relaxed during the calculations. To test the rationality of the simulation model, the convergence of the adsorption energy with respect to the number of the Fe(111) layers, cutoff energy, and k-point mesh were examined. The adsorption energies of NO2 on six layers (used in the present study) and larger nine layers (the top three layers were allowed to relax) of the Fe(111) surface are consistent (see Table S1, Supporting Information): the difference in the calculated NO2 adsorption energy is negligible (less than 1.0 kcal/mol). With increasing cutoff energy and k-point mesh, the NO2 adsorption energy altered little (see Tables S2 and S3). We thus used only the computationally less expensive model and conditions to simulate our gas-surface system. The lateral cell has dimensions a=b=8.02 A˚ and c=20.01 A˚, which includes a vacuum region of thickness greater than 15 A˚ and ensures no interaction between the slabs. We calculated adsorption energies according to ΔEads ¼ E½surface þ adsorbate -ðE½surface þ E½adsorbateÞ in which E[surface þ adsorbate], E[surface], and E[adsorbate] are the calculated electronic energies of adsorbed species on Fe(111) surface, a clean Fe(111) surface, and a gas-phase molecule, respectively. Vibrational frequencies of the adsorbed structures were analyzed on diagonalizing the Hessian matrix of selected atoms within the VASP approach. The nudged-elastic-band (NEB) method33-35 was applied to locate transition structures, and paths of minimum energy (MEP) were constructed accordingly. The rate coefficients for the reaction of NO2 on the Fe(111) surface was calculated with the VTST and RRKM theory36 as implemented in the Variflex program.37 (23) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558. (24) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251. (25) Kresse, G.; Hafner, J. J. Phys.: Condens. Matter 1994, 6, 8245. (26) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15. (27) Kresse, G.; Hafner, J. Phys. Rev. B 1996, 54, 11169. (28) Bl€ochl, P. E. Phys. Rev. B 1994, 50, 17953. (29) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (30) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (31) Zhang, Y.; Yang, W. Phys. Rev. Lett. 1998, 80, 890. (32) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188. (33) Ulitsky, A.; Elber, R. J. Chem. Phys. 1990, 92, 1510. (34) Mills, G.; J
onsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305. (35) Henkelman, G.; Uberuaga, B. P.; J
onsson, H. J. Chem. Phys. 2000, 113, 9901. (36) Baer, T.; Hase, W. L. Unimolecular Reaction Dynamics. Theory and Experiments; Oxford University Press: Oxford, 1996. (37) Klippenstein, S. J.; Wagner, A. F.; Dunbar, R. C.; Wardlaw, D. M.; Robertson, S. H. Variflex 1999.

7158 DOI: 10.1021/la904233b

Chen et al. Table 1. Calculated and Experimental Lattice Parameters and Magnetic Moment of Bulk Fe methodsa

lattice parameters/A˚

magnetic moment/μB

PW91 (SP) 2.818 PBE (SP) 2.817 rPBE (SP) 2.836 experiment 2.866b a SP represents spin-polarized method. b See ref 38.

2.20 2.23 2.23 2.22b

Figure 1. Schematic presentation of Fe(111) surface used in the present studies: (a) side view and (b) top view. The T, D, and S represent top, deep, and shallow sites, while the middle of two top sites is considered as a bridge site and labeled as B.

3. Results and Discussion 3.1. Tests of Computational Conditions. In Table 1, we first present data for the lattice parameter and magnetic moment of bulk Fe calculated at various levels of theory, with pertinent experimental data from the literature. The lattice parameter of bulk Fe predicted with consideration of spin polarization at the rPBE level is 2.836 A˚, which approaches the experimental value 2.866 A˚38 more closely than for the PW91 and PBE levels, 2.818 and 2.817 A˚, respectively. We calculated, at the rPBE level, the magnetic moment of bulk Fe; the result 2.23 μB agrees satisfactorily with the experimental value, 2.22 μB.38 For the slab surface of Fe(111), the magnetism is reinforced; the magnetic moments of the first, second, and third layers are 2.90, 2.50, and 2.61 μB, respectively. The same tendencies were found for Fe(100) and Fe(110) surfaces (3.05, 2.52, and 2.65 μB vs 2.74, 2.53, and 2.54 μB).39 The predicted chemisorption energies of small molecules on metal surfaces at the rPBE functional agree better with experimental values than those calculated with PW91.40,41 For instance, the PW91 functional yields too large chemisorption energies numerically by about 14 kcal/mol, whereas the rPBE functional proves accurate (less than 5 kcal/mol divergence) for CO on Ni(111) and Pd(111) surfaces.40 The rPBE calculation yields not only reliable geometries but also adsorption energies; this behavior has been examined and confirmed for the analogous systems of H2 adsorption and CO and H2 coadsorption on the Fe(111) surface.42,43 Mortensen et al.44 found that the PW91 adsorption energies for N on both Fe(100) and Fe(111) surfaces agree with experiment but the rPBE energies are much too small, whereas for the computed results obtained for the heat of (38) Kittel, C. Introduction to Solid State Physics, 7th ed.; John Wiley & Sons: New York, 1996. (39) Bzo
nski, P.; Kiejna, A.; Hafner, J. Surf. Sci. 2005, 590, 88. (40) Hammer, B.; Hansen, L. B.; Nørskov, J. K. Phys. Rev. B 1999, 59, 7413. (41) Schreiner, P. R. Angew. Chem., Int. Ed. 2007, 46, 4217. (42) Huo, C.-F.; Li, Y.-W.; Wang, J.; Jiao, H. J. Phys. Chem. B 2005, 109, 14160. (43) (a) Ma, Z.-Y.; Huo, C.-F.; Liao, X.-Y.; Li, Y.-W.; Wang, J.; Jiao, H. J. Phys. Chem. C 2007, 111, 4305. (b) Huo, C.-F.; Ren, J.; Li, Y.-W.; Wang, J.; Jiao, H. J. Catal. 2007, 249, 174. (44) Mortensen, J. J.; Ganduglia-Pirovano, M. V.; Hansen, L. B.; Hammer, B.; Stoltze, P.; Nørskov, J. K. Surf. Sci. 1999, 422, 8.

Langmuir 2010, 26(10), 7157–7164

Chen et al.

Article

Figure 2. Located isomers of adsorbed NO2 on Fe(111) surface and their important geometry parameters calculated at the rPBE level of theory. The bond lengths are given in angstroms. Table 2. Calculated Adsorption Energy, Relaxation Energy, Distortion Energy, Interaction Energy (kcal/mol), Geometrical Parameters (A˚), and Predicted Vibrational Frequencies (cm-1) of Adsorbed NO2 Species on Fe(111) Surface adsorption site

adsorption energy

relaxation energy

distortion energy

interaction energy

d(Fe-NO2)a

d(N-O)b

νasym(NO)

νsym(NO)

νbend(ONO)

top (T) -42.20 0.69 -1.84 -40.82 2.013 1.249 (1.257) 1385 1278 763 T-η1-N -41.68 1.15 18.45 -61.34 1.861 1.209 (1.399) 1569 795 683 T-η1-O -44.29 1.15 11.99 -57.19 2.166 1.280 (1.280) 1227 1163 820 T-η2-O,O0 -41.92 0.69 -0.92 -41.74 1.951 1.289 (1.228) 1497 1144 749 T-η2-N,O bridge (B) -57.49 1.38 6.00 -64.80 1.971 1.286 (1.282) 1131 1102 731 B-μ2-O,O0 -51.14 3.00 8.76 -62.96 1.983 1.228 (1.350) 1153 889 710 B-μ2-N,O 3-fold-shallow (S) -64.59 3.92 26.98 -95.47 1.921 1.361 (1.361) 916 872 645 S-μ3-N,O,O0 3-fold-deep (D) -47.10 6.46 35.28 -89.02 1.982 1.444 (1.343) 965 709 564 D-μ3-N,O,O0 a The shortest distance between the adsorbed atom (N or O) and the corresponding adsorption site of surface. b The values in parentheses are the second N-O bond length of NO2 molecule.

formation of Fe4N and for the N/Fe(100) systems, they argued that the rPBE approximation is more reliable. 3.2. Adsorption of NO2, NO, N, and O on the Fe(111) Surface. To locate possible stable intermediates, such as NO2/ Fe(111), NO/Fe(111), N/Fe(111), and O/Fe(111), we placed NO2, NO, N, and O species at various sites on the Fe(111) surface as shown in Figure 1. The four adsorption sites of Fe(111) considered are described as top (T), bridge (B), 3-fold-shallow (S), and 3-fold-deep (D). For the top site (T), the molecule is adsorbed on top of the first-layer Fe atom of Fe(111); at the (B) site, above the center of the Fe-Fe bond of the two first-layer Fe sites; at the (S) site, above the second-layer Fe atom; and at the (D) site, above the third-layer Fe atom. NO2 in its electronic ground state has an angular structure of symmetry class 2A1 with the unpaired electron residing on the N atom. As mentioned in the Introduction, NO2 can adsorb on a Fe(111) surface in several isomeric forms. The resulting NO2/ Fe(111) structures (shown in Figure 2) are written as follows: top site, FeNO2(T-η1-N), FeNO2(T-η1-O), FeNO2(T-η2-O,O0 ), and FeNO2(T-η2-N,O); bridge site, FeNO2(B-μ2-O,O0 ) and FeNO2(B-μ2-N,O); 3-fold-shallow, FeNO2(S-μ3-N,O,O0 ); and 3-folddeep, FeNO2(D-μ3-N,O,O0 ). As seen from Table 2, the isomer FeNO2(S-μ3-N,O,O0 ) is energetically the most stable among all calculated NO2/Fe(111) structures with an adsorption Langmuir 2010, 26(10), 7157–7164

energy -64.59 kcal/mol. The two N-O bonds of isomer FeNO2(S-μ3-N,O,O0 ) are evidently extended, indicating a strong interaction between the NO2 and the catalyst. Isomer FeNO2(D-μ3-N, O,O0 ) has the longest N-O bond (1.444 A˚), but its corresponding adsorption energy, -47.10 kcal/mol, remains within the pertinent range. As the reason involves the extraordinarily unstable FeNO2(D-μ3-N,O,O0 ), we performed a detailed analysis by decomposing the adsorption energy into individual parts. Delbecq et al.45 mentioned that the adsorption energy is decomposable into three main components: the relaxation energy of the Fe(111) surface, the distortion energy of NO2, and the energy of interaction between NO2 and the Fe(111) surface. The results of these calculations are included in Table 2. In comparing the analysis results of FeNO2(S-μ3-N,O,O0 ) and FeNO2(D-μ3-N,O,O0 ), it is found that both relaxation energies and interaction energies are moderately approximate as compared to the data contributed by distortional part. The distortion energy of FeNO2(D-μ3N,O,O0 ) is 8.30 kcal/mol larger than that of FeNO2(S-μ3-N,O,O0 ), indicating that the iron surface of FeNO2(D-μ3-N,O,O0 ) configuration strongly distorts the NO2 molecule and destabilizes the entire structure. Consequently, for the gas-surface restructuring for FeNO2(S-μ3-N,O,O0 ), the contribution with the larger (45) Delbecq, F.; Zaera, F. J. Am. Chem. Soc. 2008, 130, 14924.

DOI: 10.1021/la904233b

7159

Article

Chen et al.

Figure 3. Located isomers of adsorbed NO, N, and O on Fe(111) surface and their important geometry parameters calculated at the rPBE level of theory. The bond lengths are given in angstroms.

interaction energy (ca. -95.5 kcal/mol) explains its greatest stability of all calculated NO2/Fe(111) structures. The calculated frequencies of NO2 adsorbed on Fe(111) are reported in Table 2. The calculated frequencies for the asymmetric ν(NO) and symmetric ν(NO) stretching, and bending ν(ONO), modes are in the ranges 916-1569, 709-1278, and 564-820 cm-1, respectively. The large red shifts (relative to ν(NO) frequency of NO2 molecule)46 reflect the weakening of the N-O bond, indicating that FeNO2(S-μ3-N,O,O0 ), FeNO2(D-μ3-N,O,O0 ), and FeNO2(Bμ2-N,O) are the most likely precursors for the dissociation, whereas we ignore the path with FeNO2(D-μ3-N,O,O0 ) for its small adsorption energy. For comparison with previous calculations,47,48 the adsoption energies of NO2 on Fe(111) are generally larger than for Au(111) (Eads=-12.68 to -16.27 kcal/mol)47 and Pt(111) (Eads=-28.86 to -31.13 kcal/mol)48 surfaces, indicating that the Fe(111) surface might exhibit a larger catalytic activity for NO2. Understanding the nature of Fe-NO, Fe-N, and Fe-O interactions and structures and energetics of NO/Fe(111), N/Fe(111), and O/Fe(111) species are important for exploring the dissociation of NO2 on the Fe(111) surface. As expected, the coordination of NO on the Fe(111) surface yields a NO/Fe(111) intermediate that might have several isomers (shown in Figure 3). Our calculations (see Table 3) show that the isomer FeNO(T-η1N) with NO coordinated to Fe through N is more stable than isomer FeNO(T-η1-O) with NO coordinated to Fe through O, indicating that an initial approach of NO to the Fe(111) surface would prefer the (N...Fe) construction. The three isomers FeNO(T-η1-N), FeNO(T,S-μ2-N), and FeNO(S-η1-N) are energetically favored conformations among all calculated structures with adsorption energies -60.54, -63.92, and -57.29 kcal/mol, respectively. The NO bridge between the first and second Fe layers is more stable than on the 3-fold-shallow site. The calculated ν(NO) frequencies of NO/Fe(111) are within 1151-1753 cm-1, characteristic of NO-chemisorbed states on the Fe(111) surface. Frequencies of the FeNO(T,S-μ2-N) configuration located below 1200 cm-1 reflect the weak N-O bond, which is elongated to 1.300 A˚. The coordination of N and O atoms on Fe(111) leads to N/Fe(111) and O/Fe(111) intermediates, respectively; the result(46) (a) Herzberg, G. Electronic Spectra and Electronic Structure of Polyatomic Molecules; Van Nostrand: New York, 1966. (b) Shimanouchi, T. Tables of Molecular Vibrational Frequencies, Consolidated Vol. 1, NSRDS NBS-39. (47) Torres, D.; Gonz
alez, S.; Neyman, K. M.; Illas, F. Chem. Phys. Lett. 2006, 422, 412. (48) Getman, R. B.; Schneider, W. F. J. Phys. Chem. C 2007, 111, 389.

7160 DOI: 10.1021/la904233b

Table 3. Calculated Adsorption Energies (in kcal/mol), Geometrical Parameters (A˚), and Predicted Vibrational Frequencies (cm-1) of Adsorbed NO, N, and O Species on Fe(111) Surface adsorption site

adsorption energy

d(Fe-NO)a

d(N-O)

frequency

for NO v(NO) -60.54 1.661 1.197 1753 T-η1-N -25.62 1.818 1.213 1539 T-η1-O -63.92 1.832 1.300 1151 T,S-μ2-N -57.29 1.693 1.213 1634 S-η1-N for N v(FeN) -90.47 1.553 1064 T-η1-N -115.67 1.715 661 T,S-μ2-N -131.61 1.776 691 S,D-μ2-N for O v(FeO) -114.29 1.600 837 T-η1-O -133.43 1.795 596 T,S-μ2-O -136.76 1.855 509 S,D-μ2-O a The shortest distance between the adsorbed atom (N or O) and the corresponding adsorption site of surface.

ing species are shown in Figure 3. These species might exist in three isomers: FeX(T-η1-X), FeX(T,S-μ2-X), and FeX(S,Dμ2-X), in which X = N, O, respectively. As Table 3 indicates, the radical adsorbates of N and O atoms adsorb strongly to the Fe(111) surface. Among many adsorption sites, those between top (T) and shallow (S) as well as shallow (S) and deep (D) sites; FeX(T,S-μ2-X) and FeX(S,D-μ2-X);are favored (N, ca. -115.7 to -131.6 kcal/mol; O, ca. -133.4 to -136.8 kcal/mol) over the top site, FeX(T-η1-X) (N, ca. -90.5 kcal/mol; O, ca. -114.3 kcal/ mol). The frequencies of N/Fe(111) and O/Fe(111) systems are calculated to be within 661-1064 and 509-837 cm-1. Escott et al.49 provided experimental evidence for N-induced reconstruction of the top layer of the Fe(111) surface. For their TPD spectrum result for the N/Fe(111) system at 0.25 ML coverage, they found that the N-induced reconstruction does not cover the entire surface at this N coverage. We thus ignore this reconstruction effect in our system as the N coverage that we supposed here (0.25 ML) is within this tolerance. We compare the N/Fe(111) and O/Fe(111) structures with published work50,51 on the adsorption behavior of N2 and O2 on an iron surface. Mortensen et al.50 calculated the adsorption of four N2 molecular states (two perpendicular and two parallel (49) Escott, D. K.; Pratt, S. J.; King, D. A. Surf. Sci. 2004, 562, 226. (50) Mortensen, J. J.; Hansen, L. B.; Hammer, B.; Nørskov, J. K. J. Catal. 1999, 182, 479. (51) Bzo
nski, P.; Kiejna, A.; Hafner, J. Phys. Rev. B 2008, 77, 155424.

Langmuir 2010, 26(10), 7157–7164

Chen et al.

Article

Figure 4. Calculated possible potential energy diagram for the dissociation of NO2(g) on the surface of Fe(111). For transition state, values of imaginary frequencies, vi, are also presented.

orientations); the calculated adsorption energy for the most favorable configuration (an asymmetric parallel state) is about 9.5 kcal/mol. We find the bridge site to be considerably more stable, such that an N atom adsorbed on the top site could slide into the adjacent bridge site. Bzo
nski et al.51 calculated that dissociation of O2 on both Fe(100) and Fe(110) surfaces is a nonactivated, strongly exothermic process. We thus expected a stronger interaction between O atom and Fe surface (see Table 3) to be the origin of a further increase in the reactivity of dioxygen on a Fe skin. 3.3. Dissociation of NO2 on a Fe(111) Surface; Calculation of the Rate Coefficient. We constructed the potential energy surface (PES) of NO2 þ Fe(111) using the NEB method; we discuss reaction paths separately for two initial intermediates, FeNO2(S-μ3-N,O,O0 ) and FeNO2(B-μ2-N,O). The constructed potential energy profiles are depicted in Figure 4; the important geometric parameters of intermediates, transition structures, and products of these reactions are presented in Figure 5. As Figure 4 shows, among all tested conformers of NO2/Fe(111) structure FeNO2(S-μ3-N,O,O0 ) is energetically the most stable configuration and FeNO2(B-μ2-N,O) has the smallest energy barrier for the first deoxygenation of NO2. The first deoxygenation of NO2 from FeNO2(S-μ3-N,O,O0 ) produces NO/O/Fe(111), LM1, via transition structure TS1, with an energy barrier 10.38 kcal/mol and exothermic by 10.97 kcal/mol. In LM1 (see Figure 5), the adsorbed O atom is on the top site position, which is less stable from our calculated result. The O atom can thus readily diffuse to the more stable bridge site forming another NO/O/Fe(111) configuration, LM2. This process is found to be 27.82 kcal/mol exothermic and occurs with a small energy barrier (0.88 kcal/ mol) at transition structure TS2. We calculated also the energy barrier (TS3, 14.15 kcal/mol) for FeNO2(S-μ3-N,O, O0 ) to form LM2, directly, but this process involves a greater energy barrier than for the preceding case. The second deoxygenation might occur from LM2 with a slightly larger barrier, 19.36 kcal/mol for TS4, and forming N/2O/Fe(111), LM3, in which O is located on the top site position. Similarly, O can then diffuse to the bridge site forming P1 by passing Langmuir 2010, 26(10), 7157–7164

a small energy barrier 2.32 kcal/mol at TS5. The computed results of large exothermicity (43.34 kcal/mol) and slight energy barrier (2.32 kcal/mol) indicate that the diffusion of adsorbed O atom appears expeditious. The overall reaction NO2(g) þ Fe(111) f FeNO2(S-μ3-N,O,O0 ) f LM1 f LM2 f LM3 f P1 is calculated to be exothermic by 146.52 kcal/mol; this reaction requires no net thermal activation energy as the potential energies of four involved transition structures; TS1, TS2, TS4, and TS5;are less than the initial reference point, NO2(g) þ Fe(111). The second path begins from FeNO2(B-μ2-N,O) passing through energy 3.88 kcal/mol at TS6, producing LM4 that includes adsorbed NO on the top site position; the process is exothermic by 45.14 kcal/mol. Before the second deoxygenation occurs, the adsorbed NO might diffuse to the bridge site position forming the LM5 intermediate with a slight energy barrier 1.88 kcal/mol at TS7. Finally, the LM5 intermediate overcomes an activation barrier 18.10 kcal/mol at TS8 to break the second N-O bond and produce P2, with an exothermicity 19.08 kcal/ mol. This proposed path of reaction is totally exothermic by 116.01 kcal/mol and also occurs without thermal activation energy. From experiments, Jirsak et al.9 have shown that the Rh(111) surface reacts directly with nitrogen dioxide that decomposes at temperature 150 K producing adsorbed NO and oxygen adatoms. Most adsorbed NO would further dissociate (NO(a) f N(a) þ O(a)) between 300 and 400 K. Banse et al.13 and Wickham et al.14 observed that NO2 chemisorbs molecularly on Pd(111) at 110 K and decomposes above 180 K. When the temperature increases beyond 530 K, NO2 can serve as an effective source of oxygen. On several other transition-metal surfaces that have been experimentally examined, including Pt(111),3-6 Ru(001),7,8 and Ag(111),10-12 NO2 is adsorbed molecularly at lower temperature and then decomposed into NO(a) þ O(a) at higher temperature. However, the observation of further dissociation of adsorbed NO(a) into N(a) and O(a) is not obvious as these surfaces become passivated with the accumulation of dissociation products. Our calculated results imply that, at low temperature, NO2 can be effectively decomposed into N(a) and 2O(a); this finding requires experimental confirmation. DOI: 10.1021/la904233b

7161

Article

Chen et al.

Figure 5. Geometrical illustration of intermediates, transition states, and products for the NO2-Fe(111) interactions using the rPBE level of theory.

On the basis of the aforementioned PES for the dissociation of NO2 on Fe(111), we computed the rate coefficients for reaction path 1: NO2(g) þ Fe(111) f FeNO2(S-μ3-N,O,O0 ) f LM1 f LM2 f LM3 f P1 and path 2: NO2(g) þ Fe(111) f FeNO2(B-μ2-N,O) f LM4 f LM5 f P2. For these calculations, the stretching potential energy surface representing the barrierless adsorption NO2(g) þ Fe(111) f FeNO2(S-μ3-N,O, O0 ) and NO2(g) þ Fe(111) f FeNO2(B-μ2-N,O), which are the rate-determining paths, were calculated along the reaction coordinate Fe-N, which was varied from its equilibrium value to 5.0 A˚ with step size 0.20 A˚. At each fixed Fe-N distance, the geometries of the bottom three atomic layers of Fe(111) surface were fixed, whereas the atoms of the remaining layers and NO2 were fully optimized at the rPBE level. Based on of variational transition-state theory (VTST),52 the obtained stretching potential energy surface involving the Fe-N distance is approximated with a Morse potential, V(R) = De{1 - exp[-β(R - R0)]}2, in which R is the reaction coordinate, R0 is the equilibrium Fe-N bond distance, and De is the bond energy without zero-point energy. A more detailed description on this subject can also refer to relevant books.53 The parameters for this Morse potential are R0 = 2.025 A˚, β = 0.997 A˚-1, and De = 64.59 kcal/mol for reaction path 1 and R0 = 1.983 A˚, β = 1.273 A˚-1, and De = 51.14 kcal/ mol for reaction path 2, respectively. We performed these calculations for the temperature range 100-1000 K. The predicted rate coefficients (in units of cm3 molecule-1 s-1) (52) (a) Hase, W. L. J. Chem. Phys. 1972, 57, 730. (b) Wardlaw, D. M.; Marcus, R. A. Chem. Phys. Lett. 1984, 110, 230. (c) Wardlaw, D. M.; Marcus, R. A. J. Chem. Phys. 1985, 83, 3462. (d) Wardlaw, D. M.; Marcus, R. A. J. Phys. Chem. 1986, 90, 5383. (e) Wardlaw, D. M.; Marcus, R. A. Adv. Chem. Phys. 1988, 70, 231. (f) Song, K.; de Sainte Claire, P.; Hase, W. L.; Hass, K. C. Phys. Rev. B 1995, 52, 2949. (53) (a) Gilbert, R. G.; Smith, S. C. Theory of Unimolecular and Recombination Reactions; Blackwell Scientific: Carlton, Australia, 1990. (b) Holbrook, K. A.; Pilling, K. J.; Robertson, S. H. Unimolecular Reactions; Wiley: Chichester, UK, 1996.

7162 DOI: 10.1021/la904233b

in the broad temperature range are represented with kpath 1 ¼ 2:09  10 -4 T -2:304 expð -0:821 kcal mol -1 =RTÞ kpath 2 ¼ 1:36  10 -6 T -1:687 expð -0:371 kcal mol -1 =RTÞ ktotal ¼ 5:61  10 -5 T -2:060 expð -0:639 kcal mol -1 =RTÞ The rate coefficient for the adsorption and decomposition of NO2 on Fe(111) is defined in54 d½Xsurf =dt ¼ kðθ=As Þ½Xg which has the unit of a flux, molecules cm-2 s-1; here θ represents the fraction of available surface sites, As the surface area, and [X]g the concentration of gaseous NO2 in molecules/ cm3, respectively. At 298 K, the total value of the rate coefficient is represented as ktotal = 1.53  10-10 cm3 molecule-1 s-1. 3.4. Electronic Structures of Intermediates in NO2 Decomposition. Figure 6 shows plots of the contour surface of the electron-density difference, ΔFdiff = F[surface þ adsorbate] F[surface] - F[adsorbate], for each adsorbate/substrate system in the NO2 decomposition path: (a) before interaction, (b) FeNO2(S-μ3-N,O,O0 ), (c) FeNO2(B-μ2-N,O) and (d) LM4, respectively (see the inset in Figure 6 for atom labeling). We recognize from these drawings whether the interaction reflected in this polarization is chiefly physical, involving only electrostatic and dispersion forces, or contains significant chemical contributions. Through charge transfer from the surface Fe atoms to NO2 (π* states; see the changes in Figure 6b,c), the geometries of gaseous NO2 are evidently distorted in both bond angles and bond lengths. The effective charges calculated with Bader’s (54) Rettner, C. T.; Ashfold, M. N. R. Dynamics of Gas-Surface Interaction; Springer-Verlag: Berlin, Germany, 1991; Chapter 5.

Langmuir 2010, 26(10), 7157–7164

Chen et al.

Figure 6. Illustration of charge-density difference for NO2 dissociation on Fe(111) via the proposed potential energy pathway: (a) before interaction, (b) FeNO2(S-μ3-N,O,O0 ), (c) FeNO2(B-μ2N,O), and (d) LM4. ΔFdiff = F[surface þ adsorbate] - F[surface] F[adsorbate], and the isosurfaces were calculated at 0.02 e A˚3. The values are effective charges which are calculated by Bader analysis program.

program55 also demonstrate clearly a transfer of charge between the adsorbate and the substrate; the distortion energies of both adsorbates increased with increasing charge transfer from the substrate to the adsorbate; in Figure 6b,c 1.19 electron of FeNO2(S-μ3-N,O,O0 ) is transferred from Fe(111), but this value decreases to 0.72 electron for FeNO2(B-μ2-N,O). As a result, one can predict that the FeNO2(B-μ2-N,O) configuration possesses a smaller distortion and interaction energy than the FeNO2(S-μ3-N, O,O0 ) counterpart (see Table 2). The larger distortion energy for FeNO2(S-μ3-N,O,O0 ) did not, however, yield a smaller activation barrier for the first deoxygenation. Conversely, dissociating the first N-O bond for the isomer FeNO2(B-μ2-N,O) is much easier (see Figure 4). The difference in the charge distribution between parts c and d of Figure 6 explains this phenomenon: we observed that 0.72 |e| transferred from the Fe surface to NO2, and most electrons tend to locate on the Oa atom (-0.70 |e|, see Figure 6c). Besides, the charge -0.46 |e| on Ob atom is suggested to originate from N (0.44 |e|), and this obvious charge separation is similar to the adsorbed NO molecule in Figure 6d (N and O atoms have 0.32 and -0.49 |e|, respectively). On the basis of the above observations, we suggest that the FeNO2(B-μ2-N,O) configuration would have an “early transition state”56 to break its N-Oa bond. We plotted also the electronic local density of states (LDOS) of the system projected on orbitals for the adsorbed constructs of nitrogen and oxygen species and of the Fe(111) substrate (Figure 7). Figure 7a shows the LDOS before the NO2-Fe(111) interaction. Parts b and c of Figure 7 corresponding to the LDOS of FeNO2(S-μ3-N,O,O0 ) and FeNO2(B-μ2-N,O) configurations, respectively. In Figure 7a, the four maxima contributed by O (or N) atom (p orbital) near both sides of the Fermi level exhibit the nonbonding orbital of NO2. As the adsorption proceeds (see Figure 7b,c), they clearly show stronger hybridization (55) (a) Bader, R. F. W.; Beddall, P. M. J. Chem. Phys. 1972, 56, 3320. (b) Bader, R. F. W. Atoms in Molecules-A Quantum Theory: Oxford University Press: Oxford, UK, 1990. (56) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334.

Langmuir 2010, 26(10), 7157–7164

Article

Figure 7. Local density of states (LDOS) for NO2 dissociation on Fe(111) via proposed potential energy pathway: (a) before interaction, (b) FeNO2(S-μ3-N,O,O0 ), (c) FeNO2(B-μ2-N,O), and (d) LM4. The blue, black, and red lines represent Fe(d), N(p), and O(p), respectively. The dashed line represents the Fermi level.

(certain states vanish abruptly and become broad in a range 2.5 to -5.0 eV) between the O (or N) atom (p orbital) and the Fe(111) surface (d orbital). This result provides a rationalization of the larger molecular adsorption energy of NO2 on Fe(111). We found also that there is one more split maximum at the left side of Figure 7b (compare to Figure 7c); in this region the overlap between N(p), O(p), and Fe(d) is more significant. These observations account for the stronger interaction of structure FeNO2(S-μ3-N,O,O0 ). As displayed in Figure 7d, pronounced broadening occurs at the dissociation of an adsorbed NO2 species into the sites of the Fe substrate.

4. Summary Our calculations with spin-polarized density-functional theory indicate that the Fe(111) surface exhibits a large catalytic activity to decompose NO2. Our data show that isomer FeNO2(S-μ3N,O,O0 ) is energetically favorable among all calculated structures of NO2/Fe(111), and isomer FeNO2(B-μ2-N,O) has the smallest activation barrier to break the ON-O bond in the first step. The adsorption and dissociation of NO2 on Fe(111) surface occurs with no thermal barrier and is highly exothermic, ca. 116.0-146.5 kcal/mol. This reaction produces oxidation and nitridation of the Fe(111) surface. We thus predict that NO2 might attain a dissociative adsorption state on the Fe(111) surface. The rate coefficients for adsorption and decomposition of NO2 (through paths 1 and 2) are predicted with VTST and RRKM theory, and the nature of the interaction between adsorbate and substrate is also subjected to a detailed electronic analysis. This information about the reaction mechanism, the catalytic activity of various surface sites, and the relevance of the surface structure would be otherwise difficult to achieve with experimental measurements, indicating that periodic DFT calculations might play a vital role in the rational design of improved catalytic surfaces for the dissociation of NO2. Acknowledgment. H.-L. Chen acknowledges the (1) National Science Council, Republic of China, under Grant NSC 98-2113-M-034-002-MY2, for the financial support, (2) the financial support by Chinese Culture University, and (3) National Center for High-performance Computing, Taiwan, for the use of computer time. In addition, we are deeply DOI: 10.1021/la904233b

7163

Article

indebted to Professor M. C. Lin (from NCTU, Taiwan, and Emory University, USA) for persistent encouragement and instruction. Supporting Information Available: Convergence tests for adsorption energies with different number of layers

7164 DOI: 10.1021/la904233b

Chen et al.

(Table S1); convergence tests for adsorption energies with constant k-point and varied cutoff energies (Table S2); convergence tests for adsorption energies with constant cutoff energy and varied k-points (Table S3). This material is available free of charge via Internet at http:// pubs.acs.org.

Langmuir 2010, 26(10), 7157–7164