Crx+/SiO2
As stated previously, amorphous silica consists of a disordered array of SiO4 tetrahedra connected by Si–O–Si bridges. Unlike the other oxides
368 A. ZECCHINA et al.
reviewed previously, SiO2is considered to be fully covalent, and this is the principal reason why we consider it here, even though it is amorphous.
The Si–O–Si angle of the group bridging the corner-sharing tetrahedra can vary one a broad interval (120–160◦) without a substantial energy change, and this result explains the high thermal stability of the amorphous phase (46).
The surface of virgin silica is fully covered by hydroxyl groups, which can be only partially eliminated by outgassingin vacuoat high temperatures.
The IR spectroscopy of silanols in the absence and presence of adsorbates has been investigated intensively (46, 614–616) and is not reviewed here.
Briefly, the surface silanols are weakly acidic and can react with supported transition metal oxides, thus acting as anchoring sites for low-valent tran- sition metal ions. The structure, reactivity, and catalytic properties of these grafted ions (and, generally, of oxides supported in low concentrations on silica) represent the main subject of this case study. This system was chosen not only because of the outstanding catalytic properties of some of these material, but also because of the need for deeper understanding of the fac- tors influencing the coordination states of surface ions. Silica appears to be a good candidate because of its covalent bonding and tetrahedral structure. A detailed comparison is given of the properties of low-valent transition metal ions supported on SiO2, MgO, and ZnO.
The structure and reactivity of Fe2+, Mn2+, Co2+, Ni2+, and Cr2+anchored to silica surfaces (or, alternatively, of FeO, MnO, CoO, NiO, and CrO sup- ported on silica surfaces) have been investigated extensively with CO and NO as probe molecules (617–619). The most important results can be sum- marized as follows:
1. In the presence of low equilibrium pressures of CO, Mx+–CO (M =Fe, Co, Ni, Mn, and Cr) complexes are formed that are characterized byν(CO) stretching frequencies higher than that of CO gas. This means that we are dealing with nonclassical metal carbonyls and that the Mx+–CO bond has predominantlyσdonor character.
2. At the highest CO pressures (and the lowest temperatures), the forma- tion of multicarbonyl structures Mx+-(CO)nis sometimes observed (Ni2+and Cr2+) (506, 618–621). Ni2+/SiO2and Cr2+/SiO2are the only silica-supported catalysts active for alkene oligomerization at ambient temperature.
Although the first result is in line with those obtained for MgO–NiO, MgO–CoO, and ZnO–CoO solid solution, the second suggests that the chemistry of silica-supported ions has some unique characteristics that merit further consideration. The ability to form multicarbonyl species suggests a state of coordinative unsaturation higher than that observed for the same or similar ions in solid solutions and represents a strong indication that high
degrees of coordinative unsaturation and catalytic activity are related to each other.
Here, we focus on Cr2+/SiO2, a catalyst that has been extensively inves- tigated because of its applications in ethene polymerization (Phillips and Union Carbide processes) (469, 501, 503, 622, 623). The Phillips catalyst is prepared by reacting chronic acid with silanols to give surface chromates, which are then reduced by CO, leading to the formation of a surface sili- cate with divalent chromium. In the practical catalyst, the reduction of the surface chromate is carried out with ethene. The reaction between chromic acid and surface hydroxyl groups is likely not the only one occurring on the surface. For example, in the presence of water vapor, which is an oxidant for Cr2+, interaction with hydroxyl clusters may lead to triply anchored species and ultimately to anchored Cr3+. The Union Carbide catalyst is prepared by the reaction of chromocene with surface silanols, leading to the formation of cyclopentadiene (624). The activity of each catalyst is attributed primarily to divalent, low-coordinated species (506, 624, 625).
On the surface of amorphous silica supports, there are numerous conceiv- able locations of the Cr2+ions. These differ from each other in the number of ligands [two oxygen atoms of the≡≡SiO−groups and a variable number of oxygen atoms of the (≡≡SiOδ−Si≡≡) bridges in adjacent positions, as il- lustrated schematically in Fig. 37]. This structural variability is favored by the amorphous nature of the support and can be influenced by the thermal treatments (high-temperature treatments favor a penetration of Cr2+into the flexible silica framework) (620). The ligands surrounding the Cr2+center are not equivalent; two of them are negatively charged (SiO−) and hence strongly interacting, whereas the others are nearly neutral and consequently classified as weakly interacting.
A way to ascertain the coordination state of the anchored ions and, hence, their accessibility to adsorbing molecules is represented by the quantification of their capacity to adsorb molecular probes such as CO (and NO) and by the measurement of the IR spectrum of the resulting surface carbonyls (and nitrosyls). Thus, we obtain information about the propensity of the surface ions to insert ligands in their incomplete coordination spheres and, hence, indirect information about the location and structure of sites existing prior to adsorption. Of course, as mentioned previously, the probing with complexing molecules is always associated with a perturbation of the surface structures (this phenomenon is equivalent to surface relaxation). This unavoidable effect must be considered when structures of sites prior to adsorption are proposed.
The IR spectra recorded for CO on a Cr2+/SiO2as the coverage increased (506) are illustrated in Fig. 38. The IR spectra can be grouped into two series. The first, corresponding to low coverage (nCO/nCr2+≤2), is associated
370 A. ZECCHINA et al.
FIG. 37. (Top) Representation of silica surface obtained by cutting amorphous SiO2. O2−species exposed at the surface are saturated by hydrogen (hydroxyl groups). (Bottom) Representation of Cr-silica surface obtained by eliminating two hydroxyl groups for each an- chored chromium species (Crn+, black; surface hydroxyls, gray; bulk oxygens; white; Si4+, light gray).
with the progressive intensification of the stretching bands of adsorbed CO characterized by values of ˜ν(CO) exceeding that of CO gas. The second series (corresponding to high coverages) is characterized by the progressive erosion of the bands formed in the first phase and by the formation of new species with ˜ν(CO) bands with values less than that of CO gas.
The four IR bands (∼2200, 2191, 2184, and 2178 cm−1) characterizing the first phase are the ˜ν(CO) bands of CO ligands filling the coordination vacancies of anchored Cr2+and Cr3+. An analysis of their evolution with the CO equilibrium pressure led to the following conclusions: (i) The bands at 2184 and 2178 cm−1indicate dicarbonyl species formed at Cr2+sites with the lowest coordination; (ii) the band at 2191 cm−1belongs to Cr2+species bonded to only one CO molecule at room temperature (which are thus appar- ently less coordinatively unsaturated); and (iii) the band at about 2200 cm−1
FIG. 38. IR spectra of12CO adsorbed at 77 K on Cr/silica at coverages ranging fromθ=1 (5.33 kPa) toθ→0. The spectra were collected for this review following the procedure described by Zecchinaet al.(507).
is likely associated with CO interacting with Cr3+(debate continues about this assignment) (506, 621).
These results mean that the surface ions are located on the silica surface in at least three different local environments (in approximate agreement with the picture illustrated in Fig. 37). The frequencies of the ˜ν(CO) stretching modes (all higher than that of the CO gas) are similar to those observed for CO on ZnO; thus, electrostatic andσ-type overlap forces are mainly involved. Most likely, the electrostatic interaction prevails in the complexes absorbing at 2200 cm−1(as expected for an ion carrying a charge of+3).
In summary, as was found for other oxides, in the first phase of the adsorp- tion process CO behaves as a probe of the Lewis acidity associated with the coordination vacancies. Moreover, the most coordinatively unsaturated Cr2+ sites are able to bond to a second CO ligand, even at the lowest equilibrium
372 A. ZECCHINA et al.
pressures, according to a two-step process [the structure of the dicarbonyl species has been discussed (506, 621)].
When the CO equilibium pressure is gradually raised, the IR peaks at 2191, 2184, and 2198 cm−1, attributed to theσ-bonded complexes formed in the first stage, are gradually eroded with formation of new, intense bands in the range 2120–2000 cm−1; the band at 2200 cm−1is unaffected. These observations can be interpreted as follows: With the exception of the species represented by the band at 2200 cm−1, all other species are able to bind additional CO ligands, leading to the formation of new, labile Cr2+(CO)n
(2≤n≤4) complexes. This behavior suggests that the Cr2+species involved are highly coordinatively unsaturated.
Here, the evolution of the spectra is described in detail. The crucial ques- tion is the following: Why does the addition of further CO ligands cause such a dramatic shift toward lower frequency and an equally dramatic increase of the integrated intensity? The only possible answer is that both frequency shift and intensity enhancement are associated with an abrupt increase of the d–πinteraction. Metal carbonyls of the type M(CO)nX2(X =Br, Cl, etc.) in- deed have bands in the same frequency range with similar intensity patterns (453, 626). However, the high lability of the tri- and tetracarbonyl com- plexes (they decompose as a result of outgassing at 77 K) poses a problem.
The reduced frequency of the CO groups, which is supposed to indicate an increase ofd–πoverlap contributions and hence a strengthening of the Cr2+– CO bonds, should be accompanied by a stability enhancement (whereas the opposite is observed). This apparent contradiction can be explained if we assume that the addition of the third and fourth CO ligands is not simply a ligand insertion into preexisting coordinative vacancies but rather a ligand displacement reaction whereby the bonding of the CO ligands is accompa- nied by the simultaneous displacement of one or two weakly bonded surface ligands (presumably the bridging oxygen of the≡≡SiOδ−Si≡≡groups). The total enthalpy change involved in the process may be small (and hence the removal of CO can be facile) because the energy required for the creation of a vacancy at the Cr2+centers is counterbalanced by the energy release associated with the coordination of a surface ligand. In other words, the Cr2+ centers on silica show some mobility and under appropriate conditions, can change their positions on the surface under the influence of complexing agents (strong relaxation) and optimize their interactions with CO viad–π interaction. It is therefore concluded that some of the coordination vacancies measured by CO cannot be considered as truly existingin vacuoand that different probe molecules can give different answers. This conclusion is well documented by the results of experiments concerning the adsorption of NO (more strongly bonded than CO) and N2(more weakly bonded than CO) (506, 623, 627). In particular, it has been observed that the weakly bonded
N2ligand is not able to induce appreciable surface mobility or relaxation.
The tendency toward strong relaxation in the presence of adsorbates differ- entiates the chemistry of transition metal ions on silica from the chemistry of the same ions on crystalline oxides (on which relaxation and mobility are definitely smaller). This property is likely to play a fundamental role in de- termining the properties of Cr2+(Ni2+) on silica in catalytic processes (e.g., ethene polymerization) for which a large number of coordination vacancies are needed.
Cr2+/SiO2is an efficient catalyst for ethene polymerization, even at room temperature (in the industrial process the temperature is about 380 K). For this reason, it is an ideal system forin situspectroscopic investigations of a working catalyst.
Many questions remain about the initiation, propagation, and termination steps of the ethene polymerization mechanism. The most important models proposed to date are the Cossee model, which requires a vacant coordina- tion site on the metal center in the position adjacent to the growing alkyl chain, where ethene is coordinated before insertion into the chain (628), and the Green–Rooney model, which requires the presence of a metal–carbene species and a vacant site where ethene is coordinated prior to insertion (629).
A common feature of the two models is that the metal centers should have at least two coordination vacancies prior to the interaction with ethene—one for the alkyl or the carbene species and one for the coordination of ethylene.
On the basis of the results discussed so far (which have demonstrated that a significant fraction of Cr2+centers is highly coordinatively unsaturated), it can be understood why Cr2+/silica is such a good catalyst, whereas Cr3+ions on chromia/silica or exposed on extended faces ofα-Cr2O3are not.
One of the most debated problems of both mechanisms concerns the ini- tiation step, i.e., how the alkyl or the carbene species are created at the naked Cr2+centers by interaction with ethene. IR spectroscopy has been used in attempts to solve this problem. Time-resolved spectra taken at room temperature immediately after contacting of ethene (Fig. 39) with a model Cr2+/SiO2catalyst reduced in CO (506, 630) led to the following conclusions:
(i) The pair of bands at 2920 and 2851 cm−1, which grow with time in a paral- lel way at a nearly constant rate, are assigned to the stretching models of the CH2 groups of the living polymer chains: (ii) no absorptions of vibrations of CH3groups can be observed in the early stages of the polymerization, although the corresponding modes have extinction coefficient larger than that of CH2); and (iii) no absorptions associated with carbene species can be observed in the series of spectra shown in Fig. 39.
From these results, it is inferred that the initiation mechanism involves a metallacyclic intermediate. The observation by Ruddick and Baydal (625) that 1-hexene formed from a metallacyclic precursor is produced at the early
374 A. ZECCHINA et al.
FIG. 39. IR time-resolved spectra of ethene polymerization reaction on Cr/silica. Last spec- trum after 15 s [adapted from Zecchinaet al.(506) with permission from Elsevier Science Publishers].
stages of polymerization confirms this mechanism. An implication of this conclusion is that the Cr2+centers that are active at room temperature must have at least three “available” coordination vacancies, a result in agreement with the conclusions obtained from CO adsorption. Another conclusion is that only extremely accessible (and mobile) centers can be active in ethene polymerization, which explains the uniqueness of silica as a support. Sub- stitutional transition metal ions exposed on low-index faces of ionic oxides do not meet these requirements and thus are not active. Only ions in rare defect sites (such as those found on partially reduced chromia) are suffi- ciently coordinatively unsaturated to act as catalytic centers. Of course, all these considerations for the Cr2+/silica system are valid only for those sites that are active even at room temperature and cannot be extended to sites that work at higher temperatures or to sites obtained from the precursor by reduction by ethene. It is not excluded that on these sites a different mechanism may operate.