As we have shown, there is a clear correlation between the frequency shift of the stretching frequency of CO adsorbed at positively charged centers at the surfaces of non-transition metal oxides and halides and the electric field sensed by the molecule (Stark effect). This is the reason why CO is considered a specific and sensitive probe of the surface fields of ionic solids.
A similar frequency–field correlation is likely to exist for molecularly ad- sorbed N2, although the stretching frequency of this molecule is less sensitive to Stark effect perturbations (and far fewer data are available).
Even potentially more important is the observation that the stretching frequency of molecular hydrogen adsorbed at cationic sites is also strongly affected by positive polarizing fields. From the few low-temperature experi- ments reported (and those documented in this review), it is emerging clearly that ˜ν(H–H) is red shifted upon adsorption of H2at positive centers and that theν/˜ ν˜ratio is much larger than that found for CO and,a fortiori,for N2. This result suggests that H2may be used much more as an efficient probe of surface fields.
In this context, the most data are available for the molecular CO probe (15, 22, 29, 631). For a given surface cation withoutdelectrons of suitable energy, the strength of the electric fieldEdetected by CO is dependent on the coordination state and the charge/radius ratio of the cation on which adsorption occurs. Taken together, these two facts form the basis of the idea that the blue-shiftν˜ of the CO frequency can be used to determine the Lewis acid strength of positive surface centers. The electric field–freq- uency correlation is the basis of the spectroscopic method for the determ- ination of the strengths of Lewis acid sites. It has been shown that the me- thod has general validity (at least for cationic sites without d electrons) and that it can be extended to evaluate the strengths of Brứnsted sites (where the positive center is H+). The validity of the method for the determination of the Brứnsted acid strength has been illustrated for zeolites (632). Nevertheless, in this review the Brứnsted acidity of hy- droxyl groups on oxidic surfaces is not discussed because most of the re- ported experiments have been performed with highly dehydroxylated surfaces.
Theν˜–Ecorrelation cannot be extended straightforwardly to the inter- actions of CO with transition metal ions because, in addition tod–πoverlap, backdonation can complicate the situation.
As discussed previously, the interaction of CO with positive centers with Lewis acidic character not only alters the frequency of CO but also mod- ifies the static dipole and dynamic polarizability (αv) of the CO. This is clearly demonstrated in Fig. 40, in which the dynamic polarizability of CO
376 A. ZECCHINA et al.
FIG. 40. Dynamic polarizabilityαvas a function of ˜ν(CO) frequency of CO adsorbed on oxidic systems (transition metal oxides are included) [adapted from Zecchinaet al.(22) with permission of Elsevier Science Publishers].
adsorbed on various oxides is reported as a function of the ˜ν (CO) fre- quency (498, 571). The results summarized in Fig. 40 demonstrate the following:
1. When the surface cation–CO interaction is mainly electrostatic, the dynamic polarizability increases slightly, in a linear fashion, with increasing
frequency (the change of αv can be used to estimate the strength of the electric field and indirectly the Lewis acid strength as shown for the shift of the CO stretch).
2. When low-valent transition metal ions are involved (Ni2+, Co2+, and Cu+),d–πoverlap forces become important, and the dynamic polarizability is greatly enhanced. In this case, the CO cannot be considered as a good probe of Lewis acid strength (the stretching frequency is close to or even less than that of CO gas).
The explanation for the abrupt increase ofαvillustrated in Fig. 40 can be understood if we consider that the dynamic polarizability is proportional to the charge oscillation from the surface to the molecule (and vice versa) that occurs during the vibrational motion and that this charge oscillation is greatly favored by d–π overlap forces (via the so-called donation–backdonation mechanism). The position of the value for CO–Cr3+in the diagram is most striking; it confirms that electrostatic forces dominate, and it also suggests that a smalld–πcontribution is present as well. Furthermore, theαvof CO adsorbed on ZnO seems to be slightly higher than expected on the basis of a purely electrostatic model.
Molar extinction coefficients of adsorbed CO (εCO) have been carefully determined for Mn+ã ã ãCO adducts on several oxides. A moderate mono- tonic increase of εCO with increasing ˜ν(CO) was found in the 2250- to 2150-cm−1 interval. Only for high ˜ν(CO) values (greater than about 2240 cm−1) was a more significant increase ofεCOreported (633). The mod- erate and monotonic increase in the 2250- to 2150-cm−1range is similar to that found independently forαv. This is expected, sinceαvandεare related through the equation
εCO=4π3αvν˜2 (14) The linear relationships betweenν˜ andEand betweenαvandEillus- trated so far represent useful and fast, but nevertheless indirect, methods for establishing a scale of the Lewis acid strengths of oxide and halide surfaces.
These correlations need to be substantiated by a direct estimation of the cation–CO interaction energy. This has been done by Boliset al.(17, 69) by studying the relationship between the adsorption enthalpy of CO on metal oxides and oxidic systems and the CO stretching frequency shifts (Table II).
The data reported in Table II and Fig. 41 show a good linear correlation, with the corresponding best fit giving the following relationship (whereν(CO)˜ is in cm−1andHads◦ is in kJ mol−1):
ν(CO)˜ = −1.015Hads◦ −3.7 (15) The good linearity of the data plotted in Fig. 41 (R = 0.977) and the fact that the intercept, at−Hads◦ =0 kJ mol−1, is nearly 0 cm−1is remarkable.
378 A. ZECCHINA et al.
TABLE II
Molar Standard Enthalpies of Adsorption (Hads◦ ) and Stretching Frequency Shifts [ν˜(CO)]
of CO Adsorbed on Various d, d0, and d10Metal Ions
Sample −Hads◦ (kJ mol−1) ν˜(CO) (cm−1) Reference
CO gas 0 0 Ewing (634)
NaCl 20 16 Boliset al.(635)
KZSM-5 zeolite 28 23 Boliset al.(636)
NaY zeolite 24 26 Egerton and Stone (637, 638)
NaY zeolite 30 26 Boliset al.(635)
BaY zeolite 35 24 Egerton and Stone (637, 638)
NaZSM-5 zeolite 38 35 Boliset al.(636)
CaY zeolite 45 42 Egerton and Stone (637, 638)
Ce2+-doped Al2O3 43 41 Boliset al.(17)
Ca2+-doped Al2O3 45 42 Boliset al.(69)
ZnO 46 39 Boliset al.(635)
CaY zeolite 51 51 Boliset al.(635)
TiO2(site)a 53 44 Boliset al.(635)
Al2O3(site A)a 58 62 Boliset al.(69)
ZnY zeolite 59 64 Boliset al.(635)
TiO2(siteã)a 61 52 Boliset al.(635)
ZnY zeolite 71 64 Egerton and Stone (637, 638)
TiO2(site) 71 65 Boliset al.(635)
Al2O3(site B)a 73 72 Boliset al.(69)
Al2O3(site C)a 83 87 Boliset al.(69)
aFor the definition of the surface site, see the corresponding reference.
It must be considered that (i) the IR spectra reported in the cited references have typical resolutions of about 2 cm−1; (ii) calorimetric measurements are made at a well-defined temperature (typically, 303 K), whereas IR spectra are only nominally collected at room temperature since the heating effect of the IR beam on the sample is not precisely known; and (iii) in the presence of different adspecies, the estimated heat values are not likely to correspond to theθ =0 limit. These results clearly demonstrate that the spectroscopic method for the establishment of reasonable Lewis acid strengths of oxides is based on solid ground. We emphasize here that the electric field sensed by the probe molecule at a given site is always the result of the sum of contributions from the cation and from the surrounding anions; consequently, what is probed is never a property of a single ion. The same consideration holds for calorimetric measurements.
The situation regarding a base strength scale (Lewis and Brứnsted) is far more complex and less well developed than that regarding Lewis acid strengths. A spectroscopically simple, versatile, and universal probe has not yet been found (although many probe molecules, including CO and methane,
FIG. 41. CO stretching frequency shift vs molar standard enthalpies of adsorption for CO adsorbed on various d, d0, and d10metal ions (from data reported in refs.17, 68, 635, 636,and 638).
have been found to respond selectively in some cases to the perturbation caused by negatively charged centers).