229
Topics in Catalysis Vol. 40, Nos. 1–4, November 2006 ( 2006) DOI: 10.1007/s11244-006-0124-4
Titanium-monosubstituted polyoxometalates: relation between homogeneous and heterogeneous Ti-single-site-based catalysis Oxana A. Kholdeeva Boreskov Institute of Catalysis, Pr. Ac. Lavrentieva 5, Novosibirsk, 630090, Russia
The similarity in the catalytic behaviour of titanium-monosubstituted Keggin type polyoxometalates [PTi(L)W11O39]n- (TiPOMs) and heterogeneous titanium single-site catalysts in selective oxidations with H2O2 is demonstrated. Recent achievements in the application of Ti-POMs as soluble molecular models for studying mechanisms of oxidation catalysis are reviewed. KEY WORDS: titanium; single-site catalysts; polyoxometalates; hydrogen peroxide; peroxotitanium species; oxidation; mechanism.
Abbreviations: BPS: benzyl phenyl sulfide; BP: 2,2´,3,3´,5,5´-hexamethyl-4,4´-biphenol; CyH: cyclohexene; MPS: methyl phenyl sulphide; MPSO: methyl phenyl sulfoxide; MPSO2: methyl phenyl sulfone; POM: polyoxometalate; TMP: 2,3,6-trimethylphenol; TMBQ: 2,3,5-trimethyl-1,4-benzoquinone; TBA: tetra-nbutylammonium; Ti-POM or PW11TiL: [PTi(L)W11O39]n-; (PW11Ti)2OH: [Bu4N]7[{PTiW11O39}2OH]; (PW11Ti)2O: [Bu4N]8[{PTiW11O39}2O]; PW11TiOH: [Bu4N]4[PTi(L)W11O39]; PW11TiOMe: [Bu4N]4 [PTi(OMe)W11O39]; PW11TiOAr: [Bu4N]4[PTi(OAr)W11O39], ArOH = TMP; PW11Ti = O: [Bu4N]5 [PTi(O)W11O39]; I: [Bu4N]4[HPTi(O2)W11O39]; II: [Bu4N]5[PTi(O2)W11O39]; III: [H2PTi(O2)W11O39]31. Introduction The selective catalytic oxidation of organic compounds with environmentally benign, cheap and readily available oxidant, aqueous H2O2, is one of the main challenges of catalysis [1–22]. An increasing number of advanced microporous and mesoporous molecular sieves and amorphous mixed oxides, containing transition metal (TM) ions highly dispersed (isolated) in various inorganic matrices, appears to catalyze oxidations with hydroperoxides [6,9,12–19,23–42]. This stimulates a growing interest to the concept of single-site catalysts among people working in the field of selective oxidation. In the last years, more and more attention is given to the development of such catalysts using a rational, molecular engineering approach, which allows constructing the active sites, uniform in composition and distribution [43–52]. However, successful designing is impossible without fundamental knowledge about the structure of the active species and the nature of their reactivity. The establishment of so called ‘‘structure–reactivity relationships’’ is highly important to formulate requirements to the composition and structure of an optimal catalytic centre. Despite the immense achievements in surface science, investigation of these mechanistic issues * To whom correspondence should be addressed. E-mail:
[email protected]
remains complicated in heterogeneous catalysis, for which the catalytic species are normally difficult to characterise. That is why mechanistic studies based on model, soluble catalytic systems are becoming increasingly important in the catalyst designing area [53–61]. The remarkable properties of zeolites, silicates and aluminophosphates containing site-isolated titanium ions, both incorporated into the framework and grafted onto the surface using post-synthesis procedures, in selective oxidations with H2O2 are well-documented [9,12,13,17–19,23–26,28,29,31–37,40–42,62–64]. Developed by EniChem, microporous titanium-silicalite TS-1 is a powerful catalyst for a range of H2O2-based oxidation processes, including epoxidation, ammoximation and hydroxylation of rather small ( (PW11Ti)2O >> PW11TiOMe (figure 3(a)). In turn, the rates of the interaction of Ti-POMs with H2O2 followed the same order: PW11TiOH >> (PW11Ti)2OH > (PW11Ti)2O >> PW11 TiOMe (figure 3(b)). Studies by 31P NMR, IR, potentiometric titration and cyclic voltammetry revealed that all the Ti-POMs, except for PW11Ti = O, afford the same peroxo complex [Bu4N]4[HPTi(O2)W11O39] (I) upon interaction with aqueous H2O2 in MeCN. In sharp contrast to the other Ti-POMs, PW11Ti = O reacts with H2O2 very slowly and yields the well-known inactive peroxo complex [PTi(O2)W11O39]5) (II) [169,170]. Comparative characterization of I and II will be given in the next chapter. The catalytic activities of the Ti-POMs in TMP oxidation with aqueous H2O2 correlate with the rates of the formation of I and reduce in the same order (figure 3(c)). The PW11Ti = O species is catalytically inactive in both thioether [132,133] and alkylphenol [135] oxidation. These findings allowed us to suggest a two-step mechanism for the reaction of Ti-POMs with H2O2, which involves hydrolysis of the Ti–L bonds to yield Ti–OH species followed by its fast interaction with hydrogen peroxide producing titanium peroxo species I, active in oxidation of a few organic compounds (vide infra). ½Bu4 N4 ½PTiðOHÞW11 O39 þ H2 O2 ! ½Bu4 N4 ½HPTiðO2 ÞW11 O39 ðIÞ þ H2 O
ð1Þ
The equilibrium constant for Eq. 1, estimated using 31P NMR, is 10. The interaction of Ti-POMs with ArOH f ollows the trends similar to their interaction with H2O2 and requires preliminary hydrolysis of the Ti–L bonds to produce Ti–OH, which reacts fast with ArOH [136]. Hence, the model study performed on Ti-POMs allowed to assess the reactivity of different Ti–L bonds, including Ti–OH, Ti = O, Ti–OMe, Ti–OAr, Ti–O(H)–
Ti and Ti–O–Ti, towards H2O and H2O2, and to establish the relationship between this reactivity and catalytic activity of the Ti-POMs in TMP oxidation with aqueous H2O2. Using the values of equilibrium constants reported in [137] one can roughly estimate the ratios between different Ti–L species in real reaction mixtures, containing Ti-catalyst, H2O, H2O2, MeOH and phenolic substrate.
4. Monoprotonated titanium peroxo complex 4.1. Structure Different structures of the active peroxotitanium species proposed for titanium–silicate catalysts are shown in figure 4. Both end-on (g1) and side-on (g2) structures were suggested, but all the structures necessarily included an activating proton and thus corresponded to hydroperoxotitanium (protonated) rather than peroxotitanium (non-protonated) species. At least two experimental observations can be mentioned as arguments for that. First, it has been found that basic additives deactivate both heterogeneous and homogeneous Ti-catalysts, while acid additives produce the opposite effect [9,67–70,176]. Second, all isolated and well characterized titanium peroxo complexes, including well-known peroxo complex II [169,170], appeared to be R
η1 or
end-on:
OSi Ti O H O SiO
O
SiO
I
Ti O
O H H O H
II
η2 or
side-on:
O
O H
III
O
H OSi SiO Ti O O SiO
Ti
Ti OH V
IV
Figure 4. The structures proposed for active peroxotitanium species in Ti, Si-catalysts [66,68].
O.A. Kholdeeva/Ti-single-site-based catalysis
inactive toward oxidation of organic substrates in stoichiometric reactions [133,156,177–180]. At the same time, no hydroperoxotitanium complex was known. Recently, we managed to prepare the first protonated peroxotitanium complex, [Bu4N]4[HPTi(O2)W11O39] or I, via interaction of the l-oxo dimeric heteropolytungstate [Bu4N]8[(PTiW11O39)2O] with a 15-fold excess of 35% aqueous H2O2 in MeCN [135]. This compound has been isolated and characterized by elemental analysis, X-ray, IR, resonance Raman (RR), 31P and 183W NMR, potentiometric titration and cyclic voltammetry. The 183W NMR spectrum of I consisted of six lines with an intensity ratio of 2:2:1:2:2:2, pointing to Cs symmetry of the anion or a fast proton exchange on the NMR time scale. Potentiometric titration with methanolic TBAOH confirmed the presence of one acid proton in the molecule of I. The addition of 1 equiv. of OH- to I (31P NMR: d -12.4 ppm) resulted in the formation of II (d -13.0 ppm). Iodometric titration indicated the presence of one peroxo group per molecule of I. Cyclic voltammetry study (figure 5) revealed that I has significantly higher redox potential (E1/2 = 1.25 V) compared to non-protonated peroxo complex II (E1/ 2 = 0.88 V). Therefore, protonation results in an increase of the oxidizing ability of the peroxotitanium group. In contrast to the NMR spectra of I and II, which are very sensitive to protonation, both UV–vis and vibrational spectra of these peroxo complexes appeared to be very similar [135]. The UV–vis spectra show a strong absorption with a maximum at 395 and 390 nm, respectively, which is attributed to the O2 fi Ti ligandto-metal charge transfer band [169,170]. Note that an absorption bond with a maximum at 361 nm was found upon dosing H2O/H2O2 solution to TS-1 [80,87]. Characteristic features of the IR spectra of I and II are two bands located at 630 and 690 cm)1 and 620 and 714 cm)1, respectively, which are absent in the IR
Figure 5. Cyclic voltammetric curves for I (curve 1) and II (curve 2) [134]. Peroxo complex, 0.005 M in MeCN; TBAClO4, 0.1 M. Potentials are reported relative to a standard Ag/AgCl electrode. Scan rate, 100 mV/s.
235
Figure 6. Resonance Raman spectra of solid 1 taken at different excitation k [134].
spectra of peroxo-free Ti-POMs and thus can be assigned to the symmetric and asymmetric metal-peroxide stretches [135]. The O–O stretching band, which is known to manifest around 900 cm)1 for monoperoxo complexes [181–184] is not seen in the IR spectra of I and II due to overlap with the strong W–O–W asymmetric stretch (895 cm)1) of the Keggin unit. The RR spectra of I and II taken with excitation at k = 488 nm display an intense band at 630 cm)1, which is lacking in the RR spectrum of peroxo-free, colourless Ti-POMs. As it can be judged from figure 6, this band shows strong resonance behaviour, the intensity being gradually decreased with increasing excitation k or upon slow decomposition of I [135]. Importantly, a strong band at 618 cm)1 has been revealed in the RR spectrum of TS-1 after treatment with H2O2/H2O [88]. According to the literature this band can be assigned to the symmetric Ti–O2 stretching vibration of the peroxide group [88,185,186]. Therefore, the RR study strongly supports a side-on (g2) structure of both I and the peroxotitanium species formed in TS-1. It is noteworthy that the EXAFS study performed recently on the TS-1/H2O2 system also evidenced in favour of the side-on structure of the peroxotitanium species [80]. The question about the site of the activating proton in I has been addressed [135]. No IR bands were observed for solid I in the 1800–1600 cm)1 region indicating that no H3O+ was present and that the proton is directly attached to the POM surface. The proton (OH) manifests in the IR spectrum of solid I at 3510 cm)1 [135]. The frequency and width (dg = 30 cm)1) of the OH stretching band indicate the presence of hydrogen bonding [187]. Both a peroxo oxygen atom and a Ti–O– W bridging oxygen were regarded as possible protonation sites [133]. The W = O and W–O–W oxygen atoms are less likely as possible sites for protonation since metals with higher charge are more electron withdrawing, thereby decreasing the nucleophilicity of the oxygen atom at a given pH [188].
236
O.A. Kholdeeva/Ti-single-site-based catalysis
Figure 7. Molecular electrostatic potential function plotted over an isodensity surface for the eclipsed peroxo form of [PTi(O2)W11O39]5-. Red regions represent proton-attractive sites [134].
4.2. DFT calculations The conclusions about the structure of I based on the experimental findings are in agreement with DFT calculations performed for I and II by J.-M. Poblet’s group [135]. Thus, [PW(O2)TiW10O39]5) was computed to be 13.7 kcal mol)1 above the most stable (eclipsed) conformation of [PTi(O2)W11O39]5) This important energy excludes the W(O2) peroxo complex as a result of the interaction of Ti-POM with H2O2. The electrostatic potential (EP) function [189,190] was used for discriminating the most from the least favourable protonation sites in II. The representation which is given in figure 7 implies that the most likely positions to accept an incoming proton are the four oxygen atoms bridging the Ti and W atoms as it is indicated by the intense red colour in these regions. The O–O group could compete in the protonation process because the EP shows similar values compared to Ti–O–W sites. Again, protonation of the terminal W = O bonds is very unlikely [191]. A more quantitative description was obtained through the DFT study of the protonated species [HPTi(O2)W11O39]4) [135]. In the most stable structure A (figure 8) the hydrogen is
(b)
(a)
2.44
oriented towards the centre of the nearest M4O4 ring, a region with a high proton affinity. In structure B, the proton is roughly equidistant from two bridging oxygens and from one of the oxygens of the peroxo group. This form is 3.7 kcal mol)1 above structure A in the gas phase. Protonation of the peroxo ligand was also thoroughly investigated. All attempts to obtain a TiOO–H side-on coordination structure were, however, unsuccessful because the optimization always evolved towards geometries with a g1-coordination. This result contrasts with the recent B3LYP study of Sever and Root on the Ti(OH)3OOH model clusters who found structures with a g2-coordination for the OOH group [92]. This dissimilarity may be due to the different coordination number of titanium in the Ti-POM and Ti(OH)3OOH as well as due to the different basicity of Ti–O–W and Ti–O–H bonds. The relative energy of g1-structure C in relation to g2structure A is +3.8 kcal mol)1 in the gas phase. Structure D is quite unstable, presumably due to the lack of any stabilizing OÆÆÆH interaction. Importantly, the difference in the energies between all the structures significantly decreases when the solvent is included via a continuum model [192]. In MeCN (dielectric constant = 37.0), the relative energies of structures B, C and D in relation to structure A are only +0.4, +2.8 and +2.4 kcal mol)1. Therefore, although the calculations point to Ti–O–W as the most basic site, both Ti–OH–W and TiOO–H protonated species may coexist in solution. The DFT calculations clearly showed that, in addition to the intrinsic basicity of an oxygen site, inter and intra OÆÆÆH interactions in POM clusters are of great importance in determining the protonation site. The results of the DFT calculations prompted us to perform a RR labelling experiment using D2O2 instead of H2O2 for the preparation of I [135]. If the proton were attached to the peroxo group, a characteristic H/D downshift of the RR band would be expected, as observed for iron hydroperoxo species [193,194]. Nevertheless, the 630 cm)1 feature appeared to be not sensitive to the replacement of H for D. Therefore, the RR labelling experiment is in agreement with the DFT
(c)
1.87
(d)
3.16
2.27
2.23
2.64 2.63
2.22
Gas phase
0
+3.7
+3.8
+14.3
MeCN
0
+0.4
+2.8
+2.4
Figure 8. Optimized structures, relative energies (in kcal mol)1), and HÆÆÆO distances (in a˚) for several isomers of [HPTi(O2)W11O39])4 [134]. Notice the existence of a correlation between the number of HÆÆÆO interactions and the stability.
237
O.A. Kholdeeva/Ti-single-site-based catalysis
calculations and confirms that in solid I the activating proton is most likely localised at the Ti–O–W bridge rather than at the peroxo group.
OH
[TMP]/[I] = 2
OH
4.3. Reactivity The best evidence for a particular step in a reaction mechanism derives from finding experimental conditions where that step can be directly observed and studied. Following this principle, we generated I (d -12.4 ppm) in situ by adding 1 equiv. of protons (in the form of triflic acid) to nonprotonated II (d -13.0 ppm) and studied its interaction with MPS using 31P NMR and GC. Figure 9 shows the generation of I and its gradual disappearance after addition of MPS. The disappearance of I directly parallels the formation of sulfoxide from the thioether, as detected by GC. To the best of our knowledge, this experiment was the first unequivocal demonstration of a direct stoichiometric reaction between a peroxotitanium complex and an organic substrate [132–134]. The kinetic study performed on the stoichiometric reaction revealed that the reaction is first order in both reagents [132]. The interaction of I with BPS afforded a mixture of products shown in scheme 2. An outer-sphere electron transfer
no reaction
+ TBA5[PW11O39TiO2]
HO
OH
90%
O
MeCN, + TBA4[HPW11O39TiO2] 40oC
95% [I]/[TMP] =2 [
O
Scheme 6.
mechanism involving the formation of a thioether radical cation has been proposed based on the product and kinetic studies. Recently, we found that I easily reacts with alkylphenols, specifically with TMP, to give products which are also consistent with a homolytic oxidation mechanism [135]. In particular, oxidation of TMP with I yields TMBQ and BP, the ratio between which depends on the TMP/I molar ratio (scheme 6). When a 2-fold excess of TMP was used, the main reaction product was BP (90%), while a 2-fold excess of I led presumably to TMBQ (95%). This agrees with the reaction stoichiometry 2:1 and 1:2 for the TMP oxidation to BP and TMBQ, respectively. Again, unprotonated II was found to be inert. The formation of BP, a typical one-electron oxidation product, implies a homolytic oxidation mechanism that implicates the formation of phenoxyl radicals ArO. The radical coupling gives BP, while further interaction with I leads to TMBQ (scheme 7). As it has been mentioned already, the same products and analogous effect of the TMP/Ti ratio on the product distribution were found in the TMP oxidation with H2O2 catalyzed by mesoporous Ti,Si-catalysts [73,163].
ArO
ArOH
I
.
HO
OH O
.
ArO
I O
Scheme 7.
The kinetic study of the stoichiometric reaction between TMP and I revealed fractional (0–1) order in TMP and thus supported the formation of a g1-intermediate which most likely contains both peroxo moiety and phenol molecule (scheme 8). The lack of kinetic
I + ArOH
K
O OH Ti
k0 OAr H
Figure 9. 31P NMR spectra of [Bu4N]5[PTi(O2)W11O39] (II, 0.02 M) after addition of 1 equiv. of H+ followed by addition of MPS (0.1 M).
Scheme 8.
OAr• + TiO• + H2O
238
O.A. Kholdeeva/Ti-single-site-based catalysis
1. Hydrolysis
Ti-OL + H2O
TiOH + LOH
L = Ti, Si, Me or others
- POM or Silicate
O TiOH + H2O2
2. Hydroperoxo complex formation O 3. TMP binding
Ti H
O-OH
+ ArOH
Ti H
O
Ti
O
O Ar H O-OH
4. Inner sphere electron transfer
Ti
OAr•+
TiO•+ H2O
OAr H
5. Further transformations of ArO.
Scheme 9.
isotope effect (kArOD/kArOH = 1) indicated that the rate limiting step of the reaction is most likely an innersphere electron transfer within this intermediate. The Ti-POM-based model study allowed us to deduce a general mechanism for alkylphenol oxidation with H2O2 over titanium single-site catalysts that implicates the reaction steps shown in scheme 9. The fundamental knowledge about the reaction mechanism helped us appreciably in optimising the practically important process of TMBQ production over mesoporous Ti,Si-catalysts [195]. Keeping in mind the results on both the thioether and alkylphenol oxidation, we may propose I as a useful model compound for studying homolytic mechanisms of titanium-catalyzed oxidations. Again, the presence of a proton in the molecule of I is crucial for its reactivity. One of the possible explanations of this phenomenon is the increase of the redox potential of the peroxo group upon protonation (figure 5); however, we believe that proton may also facilitate the formation of the reactive g1-intermediate shown in scheme 8 [9,178]. The DFT
calculations (vide supra) support the latter hypothesis. Significantly, I is not reactive towards cyclohexene epoxidation under stoichiometric conditions. 5. Diprotonated titanium peroxo species Finally, we would like to address the question: what will happen if we increase the number of protons in the Ti-POM peroxo species? We had an observation that increasing proton amount in Ti-POM results in appearance of products typical for heterolytic oxygentransfer processes. Thus, when heteropolyacid H5PW11TiO40 was employed as catalyst, no one-electron oxidation products were present (scheme 10) [144]. This phenomenon has been carefully investigated using cyclohexene as substrate, and it was found that the increase in the number of protons in Ti-POM from 1 to 2 results in changing the reaction mechanism from a homolytic to a heterolytic one (table 2). This result prompted us to study a stoichiometric reaction between CyH and a diprotonated Ti-POM peroxo species (III)
Scheme 10.
239
O.A. Kholdeeva/Ti-single-site-based catalysis
Table 2 Clinical profile of study subjects Clinical profile of study subjects Clinical profile of study subjects Clinical profile of study subjects Clinical profile of study subjectsClinical profile of study subjects Clinical profiles
Normal controls
Asthmatic Atopic
Number of subjects Age [mean (range)] Sex (male/female) Current smoker (%) FVC1%, predicted FEV1%, predicted* PC20, methacholine (mg/ml)* Log (total IgE) (IU/ml)* Positive rate of specific IgE (Df, %)* Positive rate of specific IgE (Dp, %)* Positive rate of skin test (%)*
171 28.7 (7–75) 85/86 30.8 89.0±1.7 93.7±1.2 24.0±1.4 1.84±0.63 33.30 38.10 26.0
Nonatopic
380 31.5 (7–77) 193/187 19.1 87.4±6.9 83.4±0.9 2.6±4.3 2.49±0.51 65.04 71.44 69.40
170 43.6 (9–80) 55/115 21.0 88.8±7.2 82.0±1.0 3.1±5.5 1.69±0.48 0 0 0
*P value
