6.1 Molecular Models
To understand the behavior of the heterogeneous catalyst, a molecular model must
be first built to mimic the active sites anchored on the support. Regarding the
Phillips catalyst, the hexavalent chromate species on a silica surface is believed to
be reduced to lower valence states, usually Cr(II), resulting in a mononuclear or
dinuclear Cr(II) site, which is bound to the silica surface through two oxygen
linkages. During the last decade, various molecular models have been built for
the active sites of the Phillips catalyst, as graphically shown in Fig. 23.
As early as 2004, we employed hexavalent chromic acid (1g) as a simple
molecular model for simulating the coordination of ethylene on the pre-reduced
monochromate site of the Phillips catalyst [142]. Soon after that, a more realistic
silsesquioxane-supported cluster model 2g was built for theoretical investigation in
order to elucidate the effects of silica gel and its surface fluorination on the
properties of the Phillips catalyst [143, 144]. Meanwhile, Hanmura et al. [122,
123] found that two simple chromium cations [Cr(II)OH
+ and Cr(III)O
+
, as shown
in 1f and 2f] could possibly dimerize ethylene into 1-butene without using any
organometallic cocatalyst. Because the chromium centers in these two kinds of
cations were directly bonded to an oxygen atom, and the Phillips catalyst was
composed of chromium supported on silica gel through oxygen linkages, the
authors claimed that 1f and 2f could be treated as homogenous cluster models for
the Phillips catalyst.
A group of cluster models 3g–10g created by Espelid and Børve in a series of
systematic DFT investigations on the active sites of the Phillips catalyst are shown
in Fig. 23 [120, 145–149]. Clusters 3g–6g were four kinds of mononuclear Cr(II)
sites varying in –O–Cr–O– angles. 4g was a pseudo-tetrahedral cluster. 6g was a
pseudo-octahedral cluster, and the other two clusters were built with different bond
angles to represent the heterogeneity of the silica surface. Cluster 3g was a fourmembered chromasiloxane ring with a much higher ring strain and thermodynamically unfavorable formation requiring a heat of 24.6 kcal mol
À1 . For clusters 4g and
5g, the heats of Cr anchoring reaction decreased with the increasing ring size.
Compared to the experimental frequencies of 986 Æ 46 cm
À1 for a dehydrated
silica-supported chromium oxide catalyst [83], the two computed harmonic Cr¼O
stretching frequencies were 1,016 and 1,054 cm
À1 for the cluster 4g. Furthermore,
the computed d–d transition of
5 A
0 -
5 A
00 at a vertical transition energy of
10,400 cm
À1 also agreed with the experimental observation of d–d transition in
Cr
2+ ions conducted by Weckhuysen and Wachs [150]. Therefore, the
six-membered chromasiloxane ring 4g was chosen by Espelid and Børve as a key
model for a series of DFT studies on the Phillips catalyst. The cluster 6g with
geometry constraint to reserve D 3h symmetry was only used in chromium d–d
transition study for comparison with clusters 4g and 5g [145]. The hydrogen
transfer was also evaluated by means of DFT studies using a large cluster 7g
[147]. Two dinuclear clusters, 8g and 9g, represented the silica-supported dichromate species sited on narrow and wide sites [146]. The cluster 10g was a trivalent
178
R. Cheng et al.
To understand the behavior of the heterogeneous catalyst, a molecular model must
be first built to mimic the active sites anchored on the support. Regarding the
Phillips catalyst, the hexavalent chromate species on a silica surface is believed to
be reduced to lower valence states, usually Cr(II), resulting in a mononuclear or
dinuclear Cr(II) site, which is bound to the silica surface through two oxygen
linkages. During the last decade, various molecular models have been built for
the active sites of the Phillips catalyst, as graphically shown in Fig. 23.
As early as 2004, we employed hexavalent chromic acid (1g) as a simple
molecular model for simulating the coordination of ethylene on the pre-reduced
monochromate site of the Phillips catalyst [142]. Soon after that, a more realistic
silsesquioxane-supported cluster model 2g was built for theoretical investigation in
order to elucidate the effects of silica gel and its surface fluorination on the
properties of the Phillips catalyst [143, 144]. Meanwhile, Hanmura et al. [122,
123] found that two simple chromium cations [Cr(II)OH
+ and Cr(III)O
+
, as shown
in 1f and 2f] could possibly dimerize ethylene into 1-butene without using any
organometallic cocatalyst. Because the chromium centers in these two kinds of
cations were directly bonded to an oxygen atom, and the Phillips catalyst was
composed of chromium supported on silica gel through oxygen linkages, the
authors claimed that 1f and 2f could be treated as homogenous cluster models for
the Phillips catalyst.
A group of cluster models 3g–10g created by Espelid and Børve in a series of
systematic DFT investigations on the active sites of the Phillips catalyst are shown
in Fig. 23 [120, 145–149]. Clusters 3g–6g were four kinds of mononuclear Cr(II)
sites varying in –O–Cr–O– angles. 4g was a pseudo-tetrahedral cluster. 6g was a
pseudo-octahedral cluster, and the other two clusters were built with different bond
angles to represent the heterogeneity of the silica surface. Cluster 3g was a fourmembered chromasiloxane ring with a much higher ring strain and thermodynamically unfavorable formation requiring a heat of 24.6 kcal mol
À1 . For clusters 4g and
5g, the heats of Cr anchoring reaction decreased with the increasing ring size.
Compared to the experimental frequencies of 986 Æ 46 cm
À1 for a dehydrated
silica-supported chromium oxide catalyst [83], the two computed harmonic Cr¼O
stretching frequencies were 1,016 and 1,054 cm
À1 for the cluster 4g. Furthermore,
the computed d–d transition of
5 A
0 -
5 A
00 at a vertical transition energy of
10,400 cm
À1 also agreed with the experimental observation of d–d transition in
Cr
2+ ions conducted by Weckhuysen and Wachs [150]. Therefore, the
six-membered chromasiloxane ring 4g was chosen by Espelid and Børve as a key
model for a series of DFT studies on the Phillips catalyst. The cluster 6g with
geometry constraint to reserve D 3h symmetry was only used in chromium d–d
transition study for comparison with clusters 4g and 5g [145]. The hydrogen
transfer was also evaluated by means of DFT studies using a large cluster 7g
[147]. Two dinuclear clusters, 8g and 9g, represented the silica-supported dichromate species sited on narrow and wide sites [146]. The cluster 10g was a trivalent
178
R. Cheng et al.