372
W. Piskorz and F. Zasada
NO Activation
The activation of NO by Cu(I) sites in zeolites is a matter of long-lasting studies,
since its discovery by Iwamoto et al. [259, 260], both experimental and computational [274–281]. The GGA and hybrid functionals (PBE0, HSE03, HSE06, B3LYP)
are most abundantly used, and the hybrid ONIOM scheme is also found. The hybrid
functionals are slightly more accurate in predicting geometric parameters and excitation energies, and give higher density localisation than GGA. The binding energies
of cations to the framework, the band gap width for pure siliceous zeolite, and the
exchange splitting for extraframework cations is also higher for the hybrid functionals [281]. Adsorption energies for CO and NO in metal-exchanged chabazite are
linearly correlated with the stability of the cation location, and the values obtained
with hybrid functionals are closer to experiment than obtained with GGA [281]. The
same conclusion holds also for bond length and stretching frequency relationship. The
frequency values calculated with GGA for Cu(I)-chabazite are more accurate, but, on
the other hand, the red-shift values are better for hybrids. Contrary to GGA, hybrid
functionals correctly predict a blue-shift of the CO stretching mode upon adsorption,
due to the more accurate prediction of the HOMO-LUMO gap of the molecule and
the band gap of the zeolitic host [281]. The exaggeration of the exchange splitting
and hence to the too wide HOMO-LUMO gaps in hybrids lead, however, to the less
accurate description of the system with high spins. The use of hybrid functionals
also overestimates the wavenumbers which can be corrected, in pragmatic although
not elegant way, by scaling [281].
The Fe-ZSM-5 zeolite, exhibiting catalytic activity in NO decomposition, has been
also studied computationally, e.g., by Heyden et al. [282] (B3LYP functional, results
comparable to the cited calculations at the CASSCF-MCQDPT2 level of theory),
who studied as many as 46 different surface species and 63 elementary reactions.
Based on the Landau–Zener theory, they concluded that the correction stemming
from the spin inversion probability was smaller than the immanent DFT error.
The influence of two zeolite models of different size, the T1 ([Al(OH) 4 ]
− ) and
the M7 (six silica tetrahedra and one aluminium oxide tetrahedron) structures, on
the Cu
+/2+ cation properties in NO activation was studied in the article of Kozyra
and Piskorz [283]. There the NOCV analysis was used with the following fragment
selection: (T1/M7-Cu
+/2+ ) and (NO) to elucidate the influence of the zeolite-cation
electron donation/backdonation on the cation-NO charge flow. Such analysis has been
done before for the interaction of Ag and Cu sites with ethene, ethyne, formaldehyde
[284], benzene [285], and NO [286, 287] (Turbomole, B3LYP functional, def2-TZVP
basis set, QM/MM scheme in QM-Pot [288] code). The geometric, energetic, and
spectroscopic (EPR) studies of TMI at ZSM-5 were also performed by Pietrzyk et al.
[289] They used the VWN functional (with correction of BPW91) and M5 and Z6
cluster models of framework. The following TM ions were used: Mo
5+ (configuration
d
1 , term
2 D), Fe
3+ , Mn
2+ , Cr
+ (d
5 ,
6 S), Fe
2+ (d
6 ,
5 D), Co
2+ (d
7 ,
4 F), Ni
2+ (d
8 ,
3 F),
Ni
+ , Cu
2+ (d
9 ,
2 D), and Cu
+ , Zn
2+ (d
10 ,
1 S). The authors of [289] conclude that
essentially two mechanisms of N 2 O decomposition can be distinguished: the electron
transfer, when the transient N 2 O
− is formed and the N 2 −O
− bond is cleaved, and
W. Piskorz and F. Zasada
NO Activation
The activation of NO by Cu(I) sites in zeolites is a matter of long-lasting studies,
since its discovery by Iwamoto et al. [259, 260], both experimental and computational [274–281]. The GGA and hybrid functionals (PBE0, HSE03, HSE06, B3LYP)
are most abundantly used, and the hybrid ONIOM scheme is also found. The hybrid
functionals are slightly more accurate in predicting geometric parameters and excitation energies, and give higher density localisation than GGA. The binding energies
of cations to the framework, the band gap width for pure siliceous zeolite, and the
exchange splitting for extraframework cations is also higher for the hybrid functionals [281]. Adsorption energies for CO and NO in metal-exchanged chabazite are
linearly correlated with the stability of the cation location, and the values obtained
with hybrid functionals are closer to experiment than obtained with GGA [281]. The
same conclusion holds also for bond length and stretching frequency relationship. The
frequency values calculated with GGA for Cu(I)-chabazite are more accurate, but, on
the other hand, the red-shift values are better for hybrids. Contrary to GGA, hybrid
functionals correctly predict a blue-shift of the CO stretching mode upon adsorption,
due to the more accurate prediction of the HOMO-LUMO gap of the molecule and
the band gap of the zeolitic host [281]. The exaggeration of the exchange splitting
and hence to the too wide HOMO-LUMO gaps in hybrids lead, however, to the less
accurate description of the system with high spins. The use of hybrid functionals
also overestimates the wavenumbers which can be corrected, in pragmatic although
not elegant way, by scaling [281].
The Fe-ZSM-5 zeolite, exhibiting catalytic activity in NO decomposition, has been
also studied computationally, e.g., by Heyden et al. [282] (B3LYP functional, results
comparable to the cited calculations at the CASSCF-MCQDPT2 level of theory),
who studied as many as 46 different surface species and 63 elementary reactions.
Based on the Landau–Zener theory, they concluded that the correction stemming
from the spin inversion probability was smaller than the immanent DFT error.
The influence of two zeolite models of different size, the T1 ([Al(OH) 4 ]
− ) and
the M7 (six silica tetrahedra and one aluminium oxide tetrahedron) structures, on
the Cu
+/2+ cation properties in NO activation was studied in the article of Kozyra
and Piskorz [283]. There the NOCV analysis was used with the following fragment
selection: (T1/M7-Cu
+/2+ ) and (NO) to elucidate the influence of the zeolite-cation
electron donation/backdonation on the cation-NO charge flow. Such analysis has been
done before for the interaction of Ag and Cu sites with ethene, ethyne, formaldehyde
[284], benzene [285], and NO [286, 287] (Turbomole, B3LYP functional, def2-TZVP
basis set, QM/MM scheme in QM-Pot [288] code). The geometric, energetic, and
spectroscopic (EPR) studies of TMI at ZSM-5 were also performed by Pietrzyk et al.
[289] They used the VWN functional (with correction of BPW91) and M5 and Z6
cluster models of framework. The following TM ions were used: Mo
5+ (configuration
d
1 , term
2 D), Fe
3+ , Mn
2+ , Cr
+ (d
5 ,
6 S), Fe
2+ (d
6 ,
5 D), Co
2+ (d
7 ,
4 F), Ni
2+ (d
8 ,
3 F),
Ni
+ , Cu
2+ (d
9 ,
2 D), and Cu
+ , Zn
2+ (d
10 ,
1 S). The authors of [289] conclude that
essentially two mechanisms of N 2 O decomposition can be distinguished: the electron
transfer, when the transient N 2 O
− is formed and the N 2 −O
− bond is cleaved, and