ARTICLE pubs.acs.org/JPCC
Factors Controlling the Energy of Nitrogen Monolayer Coverage on High Surface Area Catalyst Oxide Carriers Francesco Arena,*,†,‡ Francesco Ferrante,§ Lorenzo Spadaro,†,‡ Antonio Prestianni,§ Antonino Raneri,† and Dario Duca§ †
Dipartimento di Chimica Industriale e Ingegneria dei Materiali, Universita degli Studi di Messina, Viale F. Stagno D’Alcontres 31, I-98166 Messina, Italy ‡ Istituto CNR-ITAE “Nicola Giordano”, Salita S. Lucia 5, I-98126 Messina, Italy § Dipartimento di Chimica Stanislao Cannizzaro, Universita degli Studi di Palermo, Viale delle Scienze Ed. 17, I-90128 Palermo, Italy ABSTRACT: Factors affecting the strength of nitrogen physisorption at monolayer coverage on different catalytic oxide carriers (e.g., ZnO, MgO, Al2O3, ZrO2, TiO2, and SiO2) have been addressed. Isotherm elaboration by the two-parameter BET equation provides C-constant values (80200) inversely related to the polarizing power (PP) of the oxide adsorbent irrespective of the surface area exposure. The energy of monolayer formation depends on the extent of charge-delocalization characterizing the surface cation-oxygen bond, which determines the acidbase character of the oxide and strength of van der Waals interactions with adsorbate molecules. Density functional theory (DFT) calculations on MgO and TiO2 systems support experimental BET findings, showing enhanced electron-density gradients and adsorption energy of the N2 molecule on the former system owing to a stronger polarity of the MgO bond.
’ INTRODUCTION Historically, the adsorption theory of Langmuir (1915) and its further application to the kinetic analysis of Hinshelwood (1927) sets the scientific basis of modern catalysis opening the way to a rational knowledge of heterogeneous reaction mechanisms and kinetics.1 Indeed, the successive implementation of the Langmuir adsorption theory by the multilayer adsorption model of Brunauer EmmettTeller (BET) in 1938 provided a method for the determination of the monolayer volume from the elaboration of physical adsorption isotherms,2 allowing one an easy calculation of the surface area of many solid adsorbents. Disclosing the fundamental aspects of the intrinsic structure of porous materials, the BET method prompted further decisive advances of the catalysis science.2 The elaboration of the physical adsorption isotherms via the so-called two-parameter BET equation p p0
1 p V 1 0 p
!¼
C1 p 1 þ V m C p0 Vm C
ð1Þ
where p/p0, Vm, and C are the relative pressure, the monolayer volume, and the adsorption constant, respectively, allowing for an easy determination of the monolayer volume as the reciprocal of the sum of the intercept (1/VmC) and slope ((C1)/VmC) values of the straight-line of the left-term are plotted as a function of p/p0.13 For its accessibility and reliability, it is designated as a standard method by the AST committee for the surface area measurement of solid materials.3 While the main scope of the r 2011 American Chemical Society
BET method rests the determination of the surface area (SA) of solids, the value of the C-constant has not been questioned because of its irrelevant influence on the accuracy of Vm calculation.1 Then, only few studies focused on size and meaning of the C-constant, which46 according to theory depends on the energy of the adsorbateadsorbent interaction.15 Therefore, this study is aimed at probing the factors affecting the energy of nitrogen physical adsorption at monolayer coverage on some common catalytic oxide carriers by a statistic evaluation of countless BET data. Experimental data are supported by density functional theory (DFT) calculations, already used to probe the adsorption pattern of small molecules on porous materials7,8 and metal-oxide surfaces.915 In particular, the latter were modeled using both cluster approaches9 and periodic DFT, with plane-wave10,11 or Gaussian12,13 basis sets. Nitrogen adsorption on MgO and TiO2 model systems was evaluated by the SIESTA method, which by using numerical basis in a periodic DFT paradigm framework was previously applied to investigate the adsorption of H2O and O2 on Al2O3,13 CO on Au/Al2O3,14 and NOx on SnO2.15
’ EXPERIMENTAL SECTION The list of the studied oxide samples with their main physicochemical properties is given in Table 1. Nitrogen adsorption Received: June 8, 2011 Revised: October 5, 2011 Published: November 04, 2011 24728
dx.doi.org/10.1021/jp205370a | J. Phys. Chem. C 2011, 115, 24728–24733
The Journal of Physical Chemistry C
ARTICLE
Table 1. List of Samples and Relative BET Data sample SiO2
code/supplier
Tcalc (K)/time (h)
Vm (cm3/g)
ZrO2
MgO
ZnO
87
0.113
377
81
91
0.119
396
75
45
0.130
198
71
DEGUSSA
41
0.111
178
90
CS 1020/PQ CORP.
54
0.100
233
94
1223/16
117
0.100
508
93
CS 2050/PQ CORP.
800/16
107
0.140
467
70
Teos/OMNISIL OS120 Teos/OMNISIL OS120
673/6 873/6
11 9
0.117 0.132
49 39
79 72
000-1,5E/AKZO NOBEL
64
0.100
261
109
000-1,5E/AKZO NOBEL
773/8
63
0.103
275
98
000-1,5E/AKZO NOBEL
823/8
61
0.101
265
108
000-1,5E/AKZO NOBEL
873/8
57
0.079
248
110
lab-made
400/6
10.2
0.101
45
93
lab-made
400/6
8.2
0.120
36
88
P25/DEGUSSA lab-made
500/6
11.9 3.3
0.100 0.105
52 14
94 96
lab-made
600/6
7.2
0.099
31
98
VP/DEGUSSA
9.6
0.100
42
97
#500/UBE Ltd.
9.5
0.103
42
207
#100/UBE Ltd.
400/16
25.2
0.115
110
199
#100/UBE Ltd.
800/16
7.2
0.099
31
199
#1000/UBE Ltd.
800/16
3.5
0.095
14
139
2.2 9.3
0.097 0.102
10 42
191 197
3.6
0.097
16
222
lab-made G72D/SUD CHEMIE G72D/SUD CHEMIE
a
CBET
Si4-5P/AKZO NOBEL
CS 2050/PQ CORP.
TiO2
SABET (m2/g)
lab-made Si4-5P/AKZO NOBEL
Al2O3
p/p0monolayera
973
p/p0 value of monolayer formation.
isotherms at 77 K were obtained using a fully automated ASAP 2010 (Micromeritics Instruments) static gas-adsorption device and elaborated according to the BET method in the 0.0 < p/p0 < 0.20 range for the calculation of the monolayer volume (Vm) from which surface area (SA) values were obtained (ΦN2, 16.2 Å2). Additional measurements using oxide samples corresponding to a nitrogen monolayer volume Vm > 2.0 NTP mL were carried out to enhance the accuracy of BET data for oxide samples with lower surface area ( 0) from the monolayer through the following equation: a1 b2 ðE1 EL Þ exp C¼ ð2Þ RT a2 b1 24729
dx.doi.org/10.1021/jp205370a |J. Phys. Chem. C 2011, 115, 24728–24733
The Journal of Physical Chemistry C
ARTICLE
Table 2. Polarizing Power (PP) and C-Constant Value of the Various Classes of Oxides oxide
PPa
PPb
CBETc
CScatchardd
SiO2
9.8
>8
81
87
Al2O3 TiO2
5.6 6.5
7.7 6.6
109 94
112 85
ZrO2
4.8
∼5
97
99
MgO
2.8
3.7
199
151
ZnO
2.7
3.3
222
184
a
Calculated from the ionic charge-to-radius ratio. b Taken from ref 25. c From the BET equation (eq 1). d From the Scatchard-type equation (ref 5).
Figure 1. (A) BET C-constant vs surface area (SA) of various oxide samples (Table 1); (B) Scatchard-type elaboration of the BET equation in the p/p0 range 0.050.2 for different classes of oxide; (C) Relationship between C-values from BET plot (eq 1) and Scatchard elaboration of the BET equation.
where a1, a2, b1, and b2 are constants depending on entropy changes of adsorbate molecules in the evaporationcondensation process in the first and successive layers, while E1 and EL are the heat of desorption from the monolayer and the latent heat of evaporation (5.6 kJ mol1) of the adsorbate, respectively.2,4 On the assumption that the evaporationcondensation pattern of adsorbed molecules in the first, second, or higher layers are the same as those of the liquid state,2 the pre-exponential factor is equal to one, and eq 2 reduces to ðE1 EL Þ RT
C ¼ e½
ð3Þ
However, in this respect, Kemball and Schreiner stressed that the entropy variations during adsorption could be quite different from that of liquefaction, resulting in orders of magnitude variations of the pre-exponential term for different adsorbateadsorbent
systems.4 Whereas, using a Scatchard-type elaboration of the BET equation, Pomonis et al. argued that the C-constant varies with surface coverage resulting in unprecedented C-values of 10002000 at very low nitrogen relative pressure (p/p0 < 0.05).5 Nevertheless, in the p/p0 range of 0.050.3, they found that the value of the C-constant keeps almost constant, resulting that it is equal to the one provided by the BET equation.5 From a theoretic point of view, this finding is not unexpected since the minimization of the surface free energy is the driving force behind many surface processes and phenomena.1 Thus, nitrogen adsorption must primarily occur on high-energy surface sites, implying that the adsorption energy at very low coverage depends on texture and morphology besides the chemical properties of the adsorbent.5,6,22 However, considering that the experimental BET data in Table 1 refer to a p/p0 range of 0.050.2, the C-values show no meaningful relationship with the surface area of the various systems (Figure 1A). However, the Scatchard-type elaboration of the BET equation in the p/p0 range of 0.050.2 results in straight-line relationships (Figure 1B), the slope of which provides a constant (average) value of the C-constant for the various classes of oxides (Table 2).5 Then, despite some small difference in the absolute value (Table 2), the comparison of the latter C-values with the ones obtained from the linear plot of the BET equation (Figure 1C) shows a straight-line correlation confirming the match of the C-constant provided by the two elaborations.5 Considering that the similar properties of the studied adsorbateadsorbent systems likely involve minor changes of the pre-exponential factor, different C-constant values (Table 2) should mirror a different energy of nitrogen adsorption on the various oxides.4,5 However, the C-constant for each class of oxide fluctuates in a relatively narrow range irrespective of the preparation and thermal treatment (Table 1). That is because the BET elaboration of isotherms in the p/p0 range of 0.050.2 do not take into account the adsorption on more coordinatively unsaturated surface sites (corner, edge, step, kink, etc.), which could be responsible of unusually high C values at very small coverage (Figure 1A).5 In addition, Ostrovskii found that for a number of adsorbentadsorbate systems, including powder and film adsorbents, the differential heat of adsorption under vacuum conditions might be constant or almost constant in a wide coverage range, since the collective characteristics of a crystal as a whole might prevail over those of the single surface atom.23 Thus, the experimental C values of the BET equation (Table 1) provide a measure of the net energy of nitrogen adsorption (E1 EL) at quasimonolayer coverage on the various oxides, corresponding to the adsorption on low-energy sites, mirroring mostly the chemical characteristics of the adsorbent. Since the induced charge-distributions 24730
dx.doi.org/10.1021/jp205370a |J. Phys. Chem. C 2011, 115, 24728–24733
The Journal of Physical Chemistry C
Figure 2. Influence of the polarizing power (PP) on the BET C-constant (A) and net energy of monolayer coverage (B).
(dipole moment) of adsorbed nitrogen molecules, responsible for weak attractive interactions involved by physical adsorption, depend on charge-localizations across the adsorbing surface,1 it is likely that the nature and polarity of the oxide surface bonds determine the strength of the induced-dipole interaction in the monolayer. In this respect, although there is a variety of criteria for materials classification, the chemical characteristics of oxides are mostly linked to the polar/covalent character of the MeO bond that, in turn, determine their surface acidbase behavior and reactivity pattern.1,2427 In particular, the oxides of nonmetals, characterized by a higher covalent character of the MeO bond, are classified as acidic, while at the opposite, those of metals featuring a prevalently ionic MeO bond are classified as basic oxides. In between lies the most densely crowded class of amphoteric oxides with mid features.1,2427 Further on this account, a quantitative ranking of inorganic oxides can be done on the basis of the physicochemical characteristics of the relative cation and, in particular, of the polarizing power (PP) that is the ratio between ionic charge and radius.2527 The data in Table 2 show for the studied oxides PP values ranging from less than 4 of prevalently ionic materials, like divalent oxides (i.e., MgO and ZnO),26 to more than 8.0 of typical covalent solids like silica,1,24,25 passing through intermediate PP values (47) of amphoteric systems.1,2527 Apart from some experimental data scattering, a plot of the C-constant as a function of the PP values of the various systems (Figure 2A) shows an exponential-decreasing trend that confirms the fundamental influence of the chemical characteristics of the adsorbent on the energy of nitrogen adsorption at monolayer coverage. In particular, the variations of the C-constant by ca. 120 units according to eq 2 account for a spanned energy range of only 0.60.7 kJ mol1, resulting in a decrease of the net desorption energy in the monolayer (E1 EL) from 3.4 kJ mol1 of polar MgO and ZnO basic oxides to 2.73.0 kJ mol1 of acidicamphoteric systems (Figure 2B). Notably, this energy range matches the E1 EL values found by BrunauerEmmettTeller for different catalytic materials (700800 cal/mol), on the basis of which they postulated an almost constant energy of monolayer coverage and a consequent invariance of the C-constant.2 At variance, Timmermann argued that a variation in the C-constant between 15 and 25 and 60 and 70 corresponds to an enhanced adsorbateadsorbent interaction strength.6,22 Therefore, the above findings disclose that the C-constant mirrors the polarity
ARTICLE
of the MeO bond, which in turn determines the surface acidic/ basic behavior of the oxide system.1,2427 In this respect, since a very weak basic character implies a simple polarization with no formation of a true coordinative σ bond with the cationic center, CO has been widely employed for the surface characterization of cationic centers at the surface of metal oxides, showing both higher stretching frequency and adsorption enthalpy with decreasing the PP of the oxide.26,27 For the sake of comparison, the electronic interactions of nitrogen at the surface of two of the above systems, namely, magnesium and titanium oxides, were also studied using the DFT approach. In particular, the [100] surface of MgO was chosen for DFT calculations since its highest stability due to the fact that equal amounts of Mg2+ and O2 ions imply a neutral or nonpolar character and a scarce reactivity of the densely packed [100] and [110] surfaces.1 Indeed, the considered C-constant values, referring to nitrogen adsorption at quasi-monolayer coverage, are likely to refer to the less active exposed crystal planes of the various oxides, well matching the assumption of adsorption at the poorly reactive [100] surface for DFT calculations that, for comparative purposes, was assumed for both MgO and TiO2. Then, DFT data were elaborated to obtain the N2 surface distance, the NtN bond length, the N2 surface angle, the N2O angle, and the adsorbate adsorbent interaction energy, summarized in Table 3. Along DFT calculations for magnesium oxide, the [100] slab was formed by 150 explicit atoms, periodically replicated along the x, y, and z directions with cell constants a = b = 15.018 Å and c = 3a. The large value of c ensures, as in the case of anatase, the presence of vacuum above the exposed surface, and the consideration of a 150 atoms slab largely attenuate the lateral interactions of the adsorbed N2 molecule with the periodic replica (see Figure 3). The anatase [100] slab consisted of 144 explicit atoms replicated with cell constants a = 18.907 Å, b = 11.340 Å, and c = 2.3 Å. The number of atoms in each slab has been chosen in order to give stoichiometric multimetal layers slabs (Mg75O75 and Ti48O96). The same models were used to investigate the adsorption process by adding a nitrogen molecule on the slab surface. The optimized geometry of the MgO [100] slab does not show large deviation from the bulk structure. A slight deformation of the slab in the z direction can be envisaged and quantified, in the average, as a drift of magnesium atoms toward the bulk (the negative z axis direction) of 0.032 Å plus a drift of the oxygen atoms toward the vacuum (the positive direction) of 0.009 Å, which resulted in a slightly wrinkled surface. Essentially no variations with respect to the bulk structure have been found in the geometrical parameters taken along the x and y directions. In the optimized geometry of the [100] slab + N2 system, the N2 molecule is bonded to the cation with the MgNtN distance equal to 2.363 Å, while the calculated NtN bond length (1.110 Å) results slightly greater, if any, than that of the isolated molecule (1.107 Å, optimized with the same protocol described above). Further, the N2 molecule appears slightly tilted with respect to the slab plane with an angle MgN2 equal to 176.5°, while the average NO distance, evaluated on the oxygen atoms neighboring the interacting Mg site, is equal to 3.13 Å. The interaction energy of the N2/magnesia system is calculated as ΔE ¼ Eðslab þ N2 Þ EðslabÞ EðN2 Þ and henceforth reported by its module (|ΔE|) for the sake of comparison with the energy values obtained by BET elaboration, results equal to 23.9 kJ mol1. 24731
dx.doi.org/10.1021/jp205370a |J. Phys. Chem. C 2011, 115, 24728–24733
The Journal of Physical Chemistry C
ARTICLE
Table 3. DFT Parameters of Nitrogen Physical Adsorption on Model Oxide Systems model system
Mn+N2 (Å)
NtN length (Å)
N2 surface angle (deg)
N2O distance (Å)
|ΔE| (kJ 3 mol1)
MgO
2.363
1.110
176.5
3.13
23.9
TiO2
2.466
1.107
177.7
2.97
20.5
Figure 3. Optimized structures of N2/MgO (A) and N2/TiO2 (B) systems. Oxygen, dark spheres; magnesium or titanium, gray spheres; nitrogen, paler gray spheres.
In the exposed surface of the [100] anatase slab there are two types of oxygen atoms: one in the middle of a triangular arrangement of titanium atoms, OT, parallel to the plane; and one, OS, bonded to a titanium atom of the surface layer and to a titanium atom of the second layer. Some distortions of these surface arrangements can be devised after geometry optimization. In particular there is a drift of the titanium atoms toward the negative z axis direction, which is more marked than the one observed in magnesium oxide. To quantify this drift, it can be considered that the Ti3OT dihedral angle changes from 0° to 15° (the variation of the Mg3O dihedral angle in the MgO surface is only 2°) and the TiOT distances in this triangular arrangement stretches by 0.02 Å in the average. Conversely, the TiOSTi angle does not change sensibly only because a shortening of the TiOS distance, by 0.08 Å in the average, is balanced out by a corresponding shortening (0.2 Å) of the TiTi distances. These distortions are attributable to the nature of the [100] surface of anatase since the incomplete coordination around the titanium atoms (a distorted octahedral in the bulk system) causes surface effects more marked than those of a deficient coordination of magnesium atoms in MgO (perfect octahedral in the bulk). The adsorption site for the N2 molecule is a titanium cation bonded to three OT and one OS. No sensible distortions of the surface geometry
Figure 4. Map of the electronic density in the portion of the MgO (A) and TiO2 (B) surface in the neighboring of N2 adsorption centers. The isolines in the range 0/0.1|e| have been drawn, with a step of 1/150. (Oxygen, red spheres; magnesium or titanium, blue spheres; nitrogen, gray spheres. The program XCrySDen has been used to create this image.28)
of anatase have been revealed after optimization of the whole system (Figure 3B). The calculated distance TiNtN is 2.466 Å with a TiNtN angle of 177.7°, while the NtN bond length is the same of the isolated molecule (Table 3). The average NO distance evaluated on the oxygen atoms neighboring the interaction Ti site is equal to 2.97 Å, while the interaction energy of the N2/anatase system is equal to 20.5 kJ mol1. 24732
dx.doi.org/10.1021/jp205370a |J. Phys. Chem. C 2011, 115, 24728–24733
The Journal of Physical Chemistry C The electronic density maps of the N2/titania (Figure 4A) and N2/magnesia (Figure 4B) systems are shown in Figure 4. This summarizes some information already discussed but overall pictorially shows the metaloxygen interaction modes at the metal oxide surfaces. According to the above findings, MgO exhibits a higher ionic character than TiO2 along the metal oxygen bonds and (not shown) a more covalent interaction between the N2 molecule and Ti4+ rather than the Mg2+ center. Nevertheless, the absolute monolayer adsorption energies determined by BET, 8.5 and 9.0 kJ/mol for TiO2 and MgO, respectively, and DFT approaches are pretty different. Indeed, the latter results are higher by +12.2 and +14.9 kJ/mol than those calculated from the BET C-constant of TiO2 and MgO, respectively. These differences are ascribable to the fact that DFT calculations are essentially based on the assumption of an idealized interaction of the adsorbate with a perfectly clean and free oxide surface, exposing a regular alternation of cations and anions (see Figure 3). Whereas, fully dehydroxylated surfaces are rather exceptional, and in reality, oxide surfaces are variably covered by hydroxyl groups.1,26,27 This likely perturbs the ideal adsorbateadsorbent interaction probed by DFT calculations, resulting in a lower interaction energy. Then, a more realistic comparison of the adsorption energy must rather take into account the differential values that compensate both the computational errors, also related to adsorption energies instead of enthalpies, and the local effects produced by experimental convoluted phenomena, occurring either on the surface or into the bulk of the adsorbateadsorbent systems. On this account, the differences of interaction energy of nitrogen with titania and magnesia surfaces are comparable, resulting equal to ca. 3 and 1 kJ mol1 for DFT and BET approaches, respectively (see Table 3 and Figure 2B). Indeed, a greater MeNtN distance (with Me, Mg, or Ti) by ca. 0.1 Å despite an almost equivalent radius of Mg and Ti ions, along with a shorter ONtN distance, implying a stronger repulsion between nitrogen and oxygen atoms,28,29 substantiate a weaker interaction strength of N2 on TiO2. In conclusion, DFT calculations confirm that the energy of the N2 adsorption depends on the polar/covalent character of the MeO bond, which in turn determines the acidbase nature of the oxide system.
’ CONCLUSIONS The factors affecting the strength of nitrogen physisorption at monolayer coverage on different catalytic oxide carriers have been addressed by a statistical evaluation of many BET data. Isotherm elaboration by the two-parameter BET equation provides C-constant values inversely related to the polarizing power (PP) of the oxide adsorbent, proving that the energy of monolayer formation depends on the extent of charge-delocalization of the surface cation-oxygen bonds. Density functional theory calculations support the reliability of BET findings, confirming that the interaction strength of nitrogen molecules with the oxide surface depends on the polarity of the MeO bonds, which in turn determines the acidbase character of the system.
ARTICLE
(4) Kemball, C.; Schreiner, G. D. L. J. Am. Chem. Soc. 1950, 72, 5605–5607. (5) Pomonis, P. J.; Petrakis, D. E.; Ladavos, A. K.; Kolonia, K. M.; Armatas, G. S.; Sklari, S. D.; Dragani, P. C.; Zarlaha, A.; Stathopulos, V. N.; Sdoukos, A. T. Microporous Mesoporous Mater. 2004, 69, 97–107. (6) Timmermann, E. O. Colloids Surf., A 2003, 220, 235–260. (7) Jagiello, J.; Ania, C. O.; Parra, J. B.; Jagiello, L.; Pis, J. J. Carbon 2007, 45, 1066–1071. (8) Jagiello, J.; Betz, W. Microporous Mesoporous Mater. 2008, 108, 117–122. (9) Staemmler, V. Top. Organomet. Chem. 2005, 12, 219–256. (10) Wang, C.; Siao, S. S.; Jiang, J. J. Phys. Chem. C 2010, 114, 18588–18593. (11) Schneider, W. F.; Hass, K. C.; Miletic, M.; Gland, J. L. J. Phys. Chem. B 2002, 106, 7405–7413. (12) Reimann, C.; Bredow, T. J. Mol. Struc. 2009, 903, 89–99. (13) Fernandez, E. M.; Eglitis, R. I.; Borstel, G.; Balbas, L. C. Comput. Mater. Sci. 2007, 39, 587–592. (14) Fernandez, E. M.; Balbas, L. C. Phys. Status Solidi 2006, 203, 1277–1283. (15) Prades, J. D.; Cirera, A.; Morante, J. R.; Pruneda, J. M.; Ordejon, P. Sens. Actuators, B 2007, 126, 62–67. (16) Ordej on, P.; Artacho, E.; Soler, J. M. Phys. Rev. B 1996, 53, 10441–10444. (17) Sanchez-Portal, D.; Ordejon, P.; Artacho, E.; Soler, J. M. Int. J. Quantum Chem. 1997, 65, 453–461. (18) Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Ordejon, P.; Sanchez-Portal, D. J. Phys.: Condens. Matter 2002, 14, 2745–2752. (19) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865–3868. (20) Troullier, N.; Martins, J. L. Phys. Rev. B 1991, 43, 1993–2006. (21) ABINIT: First-Principles Approach to Material and Nanosystem Properties. http://www.abinit.org/ (accessed Jan 2011). (22) Timmermann, E. O. J. Chem. Soc., Faraday Trans I 1989, 85, 1631–1645. (23) Ostrovskii, V. E. Catal. Today 2002, 77, 141–160. (24) Somorjai, G. Introduction to Surface Chemistry and Catalysis; John Wiley and Sons, Inc.: New York, 1994. (25) Busca, G. Phys. Chem. Chem. Phys. 1999, 1, 723–736. (26) Zecchina, A. In Materials Design for Catalytic Applications; Bellussi, G., Ed.; Milan, Italy, 1993, 205218. (27) Busca, G. In Materials Design for Catalytic Applications; Bellussi, G., Ed.; Milan, Italy, 1993, 234252. (28) Politzer, P. J. Am. Chem. Soc. 1969, 91, 6235–6239. (29) Politzer, P. Inorg. Chem. 1977, 16, 3350–3355.
’ REFERENCES (1) Chorkendorff, I.; Niemantsverdriet, J. W. Concepts of Modern Catalysis and Kinetics; Wiley-VCH GmbH & Co. KGaA: Weinheim, Germany, 2005. (2) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309–319. (3) ASTM. D-3663-84, STM for Surface Area of Catalysts; 1988; pp 36. 24733
dx.doi.org/10.1021/jp205370a |J. Phys. Chem. C 2011, 115, 24728–24733